{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.25, "eval_steps": 500, "global_step": 25000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1e-05, "grad_norm": 1.0852325698269518, "learning_rate": 3e-06, "loss": 10.8502, "step": 1 }, { "epoch": 2e-05, "grad_norm": 1.0764721550065255, "learning_rate": 6e-06, "loss": 10.85, "step": 2 }, { "epoch": 3e-05, "grad_norm": 1.0926147944947429, "learning_rate": 9e-06, "loss": 10.8498, "step": 3 }, { "epoch": 4e-05, "grad_norm": 1.0859033473331443, "learning_rate": 1.2e-05, "loss": 10.8492, "step": 4 }, { "epoch": 5e-05, "grad_norm": 1.0906737956052586, "learning_rate": 1.5e-05, "loss": 10.8465, "step": 5 }, { "epoch": 6e-05, "grad_norm": 1.0895692490682567, "learning_rate": 1.8e-05, "loss": 10.8458, "step": 6 }, { "epoch": 7e-05, "grad_norm": 1.0913822945668856, "learning_rate": 2.1000000000000002e-05, "loss": 10.8367, "step": 7 }, { "epoch": 8e-05, "grad_norm": 1.076648393918225, "learning_rate": 2.4e-05, "loss": 10.8153, "step": 8 }, { "epoch": 9e-05, "grad_norm": 1.0616757084723456, "learning_rate": 2.7e-05, "loss": 10.8111, "step": 9 }, { "epoch": 0.0001, "grad_norm": 1.0802856236028353, "learning_rate": 3e-05, "loss": 10.7995, "step": 10 }, { "epoch": 0.00011, "grad_norm": 1.066268838859797, "learning_rate": 3.2999999999999996e-05, "loss": 10.7856, "step": 11 }, { "epoch": 0.00012, "grad_norm": 1.0695024318288597, "learning_rate": 3.6e-05, "loss": 10.7751, "step": 12 }, { "epoch": 0.00013, "grad_norm": 1.0509312651691944, "learning_rate": 3.9e-05, "loss": 10.7566, "step": 13 }, { "epoch": 0.00014, "grad_norm": 1.0428486200491074, "learning_rate": 4.2000000000000004e-05, "loss": 10.7464, "step": 14 }, { "epoch": 0.00015, "grad_norm": 1.0323891787095008, "learning_rate": 4.4999999999999996e-05, "loss": 10.7372, "step": 15 }, { "epoch": 0.00016, "grad_norm": 1.0039368052557194, "learning_rate": 4.8e-05, "loss": 10.7203, "step": 16 }, { "epoch": 0.00017, "grad_norm": 0.973581565324899, "learning_rate": 5.1000000000000006e-05, "loss": 10.7061, "step": 17 }, { "epoch": 0.00018, "grad_norm": 0.9665094864989575, "learning_rate": 5.4e-05, "loss": 10.6912, "step": 18 }, { "epoch": 0.00019, "grad_norm": 0.9428758378586757, "learning_rate": 5.7e-05, "loss": 10.6779, "step": 19 }, { "epoch": 0.0002, "grad_norm": 0.9431515825607029, "learning_rate": 6e-05, "loss": 10.6671, "step": 20 }, { "epoch": 0.00021, "grad_norm": 0.9246458493532533, "learning_rate": 6.3e-05, "loss": 10.6537, "step": 21 }, { "epoch": 0.00022, "grad_norm": 0.9206934968159175, "learning_rate": 6.599999999999999e-05, "loss": 10.639, "step": 22 }, { "epoch": 0.00023, "grad_norm": 0.9152225030599742, "learning_rate": 6.9e-05, "loss": 10.6272, "step": 23 }, { "epoch": 0.00024, "grad_norm": 0.9101679877876945, "learning_rate": 7.2e-05, "loss": 10.6142, "step": 24 }, { "epoch": 0.00025, "grad_norm": 0.9099590574781056, "learning_rate": 7.500000000000001e-05, "loss": 10.601, "step": 25 }, { "epoch": 0.00026, "grad_norm": 0.9093682715479869, "learning_rate": 7.8e-05, "loss": 10.5871, "step": 26 }, { "epoch": 0.00027, "grad_norm": 0.9036175189442642, "learning_rate": 8.1e-05, "loss": 10.5753, "step": 27 }, { "epoch": 0.00028, "grad_norm": 0.9033718484336727, "learning_rate": 8.400000000000001e-05, "loss": 10.5624, "step": 28 }, { "epoch": 0.00029, "grad_norm": 0.9067415030491869, "learning_rate": 8.7e-05, "loss": 10.5477, "step": 29 }, { "epoch": 0.0003, "grad_norm": 0.9152440544157157, "learning_rate": 8.999999999999999e-05, "loss": 10.5333, "step": 30 }, { "epoch": 0.00031, "grad_norm": 0.9122648708383427, "learning_rate": 9.3e-05, "loss": 10.5179, "step": 31 }, { "epoch": 0.00032, "grad_norm": 0.9091031697803933, "learning_rate": 9.6e-05, "loss": 10.5045, "step": 32 }, { "epoch": 0.00033, "grad_norm": 0.9095468136863347, "learning_rate": 9.900000000000001e-05, "loss": 10.4877, "step": 33 }, { "epoch": 0.00034, "grad_norm": 0.9108234370612854, "learning_rate": 0.00010200000000000001, "loss": 10.4718, "step": 34 }, { "epoch": 0.00035, "grad_norm": 0.9162402865630151, "learning_rate": 0.00010500000000000002, "loss": 10.4546, "step": 35 }, { "epoch": 0.00036, "grad_norm": 0.9114412465753974, "learning_rate": 0.000108, "loss": 10.4378, "step": 36 }, { "epoch": 0.00037, "grad_norm": 0.9012823116898077, "learning_rate": 0.000111, "loss": 10.4207, "step": 37 }, { "epoch": 0.00038, "grad_norm": 0.905600573119951, "learning_rate": 0.000114, "loss": 10.4015, "step": 38 }, { "epoch": 0.00039, "grad_norm": 0.9073115291741589, "learning_rate": 0.000117, "loss": 10.3827, "step": 39 }, { "epoch": 0.0004, "grad_norm": 0.9112034567745622, "learning_rate": 0.00012, "loss": 10.361, "step": 40 }, { "epoch": 0.00041, "grad_norm": 0.9001962110912249, "learning_rate": 0.000123, "loss": 10.3422, "step": 41 }, { "epoch": 0.00042, "grad_norm": 0.9057302480900619, "learning_rate": 0.000126, "loss": 10.3219, "step": 42 }, { "epoch": 0.00043, "grad_norm": 0.9101639037851382, "learning_rate": 0.000129, "loss": 10.299, "step": 43 }, { "epoch": 0.00044, "grad_norm": 0.9097658850257722, "learning_rate": 0.00013199999999999998, "loss": 10.2779, "step": 44 }, { "epoch": 0.00045, "grad_norm": 0.9112197620092461, "learning_rate": 0.000135, "loss": 10.255, "step": 45 }, { "epoch": 0.00046, "grad_norm": 0.9030509759254304, "learning_rate": 0.000138, "loss": 10.2333, "step": 46 }, { "epoch": 0.00047, "grad_norm": 0.909935878054922, "learning_rate": 0.000141, "loss": 10.2068, "step": 47 }, { "epoch": 0.00048, "grad_norm": 0.9080711409550536, "learning_rate": 0.000144, "loss": 10.1836, "step": 48 }, { "epoch": 0.00049, "grad_norm": 0.9075427394108988, "learning_rate": 0.000147, "loss": 10.1601, "step": 49 }, { "epoch": 0.0005, "grad_norm": 0.9115402833822758, "learning_rate": 0.00015000000000000001, "loss": 10.1348, "step": 50 }, { "epoch": 0.00051, "grad_norm": 0.9123864422518003, "learning_rate": 0.000153, "loss": 10.1081, "step": 51 }, { "epoch": 0.00052, "grad_norm": 0.9087603210312261, "learning_rate": 0.000156, "loss": 10.082, "step": 52 }, { "epoch": 0.00053, "grad_norm": 0.9004321599195088, "learning_rate": 0.000159, "loss": 10.0586, "step": 53 }, { "epoch": 0.00054, "grad_norm": 0.9194408947742885, "learning_rate": 0.000162, "loss": 10.0265, "step": 54 }, { "epoch": 0.00055, "grad_norm": 0.9109092479970925, "learning_rate": 0.000165, "loss": 10.0024, "step": 55 }, { "epoch": 0.00056, "grad_norm": 0.9034410075551015, "learning_rate": 0.00016800000000000002, "loss": 9.9748, "step": 56 }, { "epoch": 0.00057, "grad_norm": 0.9088310165095549, "learning_rate": 0.000171, "loss": 9.9476, "step": 57 }, { "epoch": 0.00058, "grad_norm": 0.9072222387188987, "learning_rate": 0.000174, "loss": 9.9231, "step": 58 }, { "epoch": 0.00059, "grad_norm": 0.9117524209160145, "learning_rate": 0.000177, "loss": 9.8912, "step": 59 }, { "epoch": 0.0006, "grad_norm": 0.8993095807646821, "learning_rate": 0.00017999999999999998, "loss": 9.865, "step": 60 }, { "epoch": 0.00061, "grad_norm": 0.8983569610899129, "learning_rate": 0.000183, "loss": 9.8375, "step": 61 }, { "epoch": 0.00062, "grad_norm": 0.9044118676660488, "learning_rate": 0.000186, "loss": 9.8067, "step": 62 }, { "epoch": 0.00063, "grad_norm": 0.8968556961344649, "learning_rate": 0.000189, "loss": 9.7793, "step": 63 }, { "epoch": 0.00064, "grad_norm": 0.8932322761861354, "learning_rate": 0.000192, "loss": 9.7493, "step": 64 }, { "epoch": 0.00065, "grad_norm": 0.9012805896747853, "learning_rate": 0.00019500000000000002, "loss": 9.7174, "step": 65 }, { "epoch": 0.00066, "grad_norm": 0.8945673092169469, "learning_rate": 0.00019800000000000002, "loss": 9.6935, "step": 66 }, { "epoch": 0.00067, "grad_norm": 0.8966961567883143, "learning_rate": 0.000201, "loss": 9.6641, "step": 67 }, { "epoch": 0.00068, "grad_norm": 0.9035315852540858, "learning_rate": 0.00020400000000000003, "loss": 9.632, "step": 68 }, { "epoch": 0.00069, "grad_norm": 0.8884844509246163, "learning_rate": 0.00020700000000000002, "loss": 9.6048, "step": 69 }, { "epoch": 0.0007, "grad_norm": 0.9014609193010997, "learning_rate": 0.00021000000000000004, "loss": 9.5684, "step": 70 }, { "epoch": 0.00071, "grad_norm": 0.8912146000887589, "learning_rate": 0.00021299999999999997, "loss": 9.5412, "step": 71 }, { "epoch": 0.00072, "grad_norm": 0.8946751138251678, "learning_rate": 0.000216, "loss": 9.5146, "step": 72 }, { "epoch": 0.00073, "grad_norm": 0.8936103707135835, "learning_rate": 0.00021899999999999998, "loss": 9.4811, "step": 73 }, { "epoch": 0.00074, "grad_norm": 0.8944690762762719, "learning_rate": 0.000222, "loss": 9.4547, "step": 74 }, { "epoch": 0.00075, "grad_norm": 0.8901019220356112, "learning_rate": 0.000225, "loss": 9.419, "step": 75 }, { "epoch": 0.00076, "grad_norm": 0.8989916490115168, "learning_rate": 0.000228, "loss": 9.3919, "step": 76 }, { "epoch": 0.00077, "grad_norm": 0.8950786665970848, "learning_rate": 0.000231, "loss": 9.3618, "step": 77 }, { "epoch": 0.00078, "grad_norm": 0.8906614388540859, "learning_rate": 0.000234, "loss": 9.3311, "step": 78 }, { "epoch": 0.00079, "grad_norm": 0.8922201842477633, "learning_rate": 0.00023700000000000001, "loss": 9.2961, "step": 79 }, { "epoch": 0.0008, "grad_norm": 0.8944676014783551, "learning_rate": 0.00024, "loss": 9.261, "step": 80 }, { "epoch": 0.00081, "grad_norm": 0.8994604493514015, "learning_rate": 0.00024300000000000002, "loss": 9.2301, "step": 81 }, { "epoch": 0.00082, "grad_norm": 0.9081887685982744, "learning_rate": 0.000246, "loss": 9.2056, "step": 82 }, { "epoch": 0.00083, "grad_norm": 0.8969756349033113, "learning_rate": 0.00024900000000000004, "loss": 9.1708, "step": 83 }, { "epoch": 0.00084, "grad_norm": 0.9002173540875773, "learning_rate": 0.000252, "loss": 9.1399, "step": 84 }, { "epoch": 0.00085, "grad_norm": 0.8962626923808309, "learning_rate": 0.000255, "loss": 9.1164, "step": 85 }, { "epoch": 0.00086, "grad_norm": 0.8974847972599506, "learning_rate": 0.000258, "loss": 9.0803, "step": 86 }, { "epoch": 0.00087, "grad_norm": 0.8897217554173651, "learning_rate": 0.000261, "loss": 9.0575, "step": 87 }, { "epoch": 0.00088, "grad_norm": 0.89571441212455, "learning_rate": 0.00026399999999999997, "loss": 9.0214, "step": 88 }, { "epoch": 0.00089, "grad_norm": 0.8904446195114993, "learning_rate": 0.000267, "loss": 8.9949, "step": 89 }, { "epoch": 0.0009, "grad_norm": 0.8898588618543369, "learning_rate": 0.00027, "loss": 8.9673, "step": 90 }, { "epoch": 0.00091, "grad_norm": 0.8846782407589194, "learning_rate": 0.000273, "loss": 8.9371, "step": 91 }, { "epoch": 0.00092, "grad_norm": 0.8861052380663628, "learning_rate": 0.000276, "loss": 8.9059, "step": 92 }, { "epoch": 0.00093, "grad_norm": 0.887769377915217, "learning_rate": 0.000279, "loss": 8.8757, "step": 93 }, { "epoch": 0.00094, "grad_norm": 0.8876037385438905, "learning_rate": 0.000282, "loss": 8.8493, "step": 94 }, { "epoch": 0.00095, "grad_norm": 0.8845306977273465, "learning_rate": 0.000285, "loss": 8.8182, "step": 95 }, { "epoch": 0.00096, "grad_norm": 0.8942873956095471, "learning_rate": 0.000288, "loss": 8.7858, "step": 96 }, { "epoch": 0.00097, "grad_norm": 0.885093824966415, "learning_rate": 0.000291, "loss": 8.7614, "step": 97 }, { "epoch": 0.00098, "grad_norm": 0.8868745046767721, "learning_rate": 0.000294, "loss": 8.7241, "step": 98 }, { "epoch": 0.00099, "grad_norm": 0.8870170622728157, "learning_rate": 0.000297, "loss": 8.7038, "step": 99 }, { "epoch": 0.001, "grad_norm": 0.8872899849688387, "learning_rate": 0.00030000000000000003, "loss": 8.6777, "step": 100 }, { "epoch": 0.00101, "grad_norm": 0.8900580366729636, "learning_rate": 0.00030300000000000005, "loss": 8.6439, "step": 101 }, { "epoch": 0.00102, "grad_norm": 0.8782792949253078, "learning_rate": 0.000306, "loss": 8.624, "step": 102 }, { "epoch": 0.00103, "grad_norm": 0.8831155019167556, "learning_rate": 0.000309, "loss": 8.5998, "step": 103 }, { "epoch": 0.00104, "grad_norm": 0.8733679539923869, "learning_rate": 0.000312, "loss": 8.5783, "step": 104 }, { "epoch": 0.00105, "grad_norm": 0.8744723269171101, "learning_rate": 0.000315, "loss": 8.5519, "step": 105 }, { "epoch": 0.00106, "grad_norm": 0.8685646437973352, "learning_rate": 0.000318, "loss": 8.5224, "step": 106 }, { "epoch": 0.00107, "grad_norm": 0.8618721810364073, "learning_rate": 0.000321, "loss": 8.498, "step": 107 }, { "epoch": 0.00108, "grad_norm": 0.8685042548991871, "learning_rate": 0.000324, "loss": 8.4695, "step": 108 }, { "epoch": 0.00109, "grad_norm": 0.8610554719114292, "learning_rate": 0.000327, "loss": 8.4497, "step": 109 }, { "epoch": 0.0011, "grad_norm": 0.8688551326351219, "learning_rate": 0.00033, "loss": 8.4258, "step": 110 }, { "epoch": 0.00111, "grad_norm": 0.8807293418345888, "learning_rate": 0.000333, "loss": 8.3951, "step": 111 }, { "epoch": 0.00112, "grad_norm": 0.9486358341091515, "learning_rate": 0.00033600000000000004, "loss": 8.3744, "step": 112 }, { "epoch": 0.00113, "grad_norm": 1.0731259256559935, "learning_rate": 0.000339, "loss": 8.3381, "step": 113 }, { "epoch": 0.00114, "grad_norm": 0.8765287858307944, "learning_rate": 0.000342, "loss": 8.3246, "step": 114 }, { "epoch": 0.00115, "grad_norm": 0.888519882671605, "learning_rate": 0.00034500000000000004, "loss": 8.3, "step": 115 }, { "epoch": 0.00116, "grad_norm": 0.8624494481396647, "learning_rate": 0.000348, "loss": 8.2774, "step": 116 }, { "epoch": 0.00117, "grad_norm": 0.8669997406867073, "learning_rate": 0.000351, "loss": 8.2472, "step": 117 }, { "epoch": 0.00118, "grad_norm": 0.8453477351571969, "learning_rate": 0.000354, "loss": 8.2308, "step": 118 }, { "epoch": 0.00119, "grad_norm": 0.8521071532477, "learning_rate": 0.000357, "loss": 8.2189, "step": 119 }, { "epoch": 0.0012, "grad_norm": 0.8273790560738514, "learning_rate": 0.00035999999999999997, "loss": 8.1923, "step": 120 }, { "epoch": 0.00121, "grad_norm": 0.8393654427306546, "learning_rate": 0.000363, "loss": 8.1655, "step": 121 }, { "epoch": 0.00122, "grad_norm": 0.8059519568643615, "learning_rate": 0.000366, "loss": 8.1479, "step": 122 }, { "epoch": 0.00123, "grad_norm": 0.8189584135948407, "learning_rate": 0.000369, "loss": 8.1261, "step": 123 }, { "epoch": 0.00124, "grad_norm": 0.8067888726335559, "learning_rate": 0.000372, "loss": 8.1038, "step": 124 }, { "epoch": 0.00125, "grad_norm": 0.8198363821900397, "learning_rate": 0.000375, "loss": 8.0789, "step": 125 }, { "epoch": 0.00126, "grad_norm": 0.7909108940386875, "learning_rate": 0.000378, "loss": 8.0658, "step": 126 }, { "epoch": 0.00127, "grad_norm": 0.7843545168994203, "learning_rate": 0.000381, "loss": 8.0421, "step": 127 }, { "epoch": 0.00128, "grad_norm": 0.8003774808810522, "learning_rate": 0.000384, "loss": 8.0238, "step": 128 }, { "epoch": 0.00129, "grad_norm": 0.7638903889263088, "learning_rate": 0.00038700000000000003, "loss": 8.0106, "step": 129 }, { "epoch": 0.0013, "grad_norm": 0.7931725212286161, "learning_rate": 0.00039000000000000005, "loss": 7.9852, "step": 130 }, { "epoch": 0.00131, "grad_norm": 0.851354650276413, "learning_rate": 0.000393, "loss": 7.955, "step": 131 }, { "epoch": 0.00132, "grad_norm": 0.9256168821467203, "learning_rate": 0.00039600000000000003, "loss": 7.9489, "step": 132 }, { "epoch": 0.00133, "grad_norm": 1.005976872013275, "learning_rate": 0.00039900000000000005, "loss": 7.9296, "step": 133 }, { "epoch": 0.00134, "grad_norm": 1.1153886851453207, "learning_rate": 0.000402, "loss": 7.9211, "step": 134 }, { "epoch": 0.00135, "grad_norm": 0.8471639827806281, "learning_rate": 0.00040500000000000003, "loss": 7.8895, "step": 135 }, { "epoch": 0.00136, "grad_norm": 0.8134678474512278, "learning_rate": 0.00040800000000000005, "loss": 7.8718, "step": 136 }, { "epoch": 0.00137, "grad_norm": 0.8615881294022464, "learning_rate": 0.000411, "loss": 7.8543, "step": 137 }, { "epoch": 0.00138, "grad_norm": 0.7278802715029148, "learning_rate": 0.00041400000000000003, "loss": 7.8276, "step": 138 }, { "epoch": 0.00139, "grad_norm": 0.7307783350096743, "learning_rate": 0.00041700000000000005, "loss": 7.8159, "step": 139 }, { "epoch": 0.0014, "grad_norm": 0.7140372083943641, "learning_rate": 0.00042000000000000007, "loss": 7.7973, "step": 140 }, { "epoch": 0.00141, "grad_norm": 0.7575231935901182, "learning_rate": 0.000423, "loss": 7.7858, "step": 141 }, { "epoch": 0.00142, "grad_norm": 0.701678588628802, "learning_rate": 0.00042599999999999995, "loss": 7.7727, "step": 142 }, { "epoch": 0.00143, "grad_norm": 0.6771555412467659, "learning_rate": 0.00042899999999999997, "loss": 7.7449, "step": 143 }, { "epoch": 0.00144, "grad_norm": 0.6615684105299617, "learning_rate": 0.000432, "loss": 7.7264, "step": 144 }, { "epoch": 0.00145, "grad_norm": 0.7168928847645009, "learning_rate": 0.000435, "loss": 7.7227, "step": 145 }, { "epoch": 0.00146, "grad_norm": 0.6785803038006687, "learning_rate": 0.00043799999999999997, "loss": 7.6843, "step": 146 }, { "epoch": 0.00147, "grad_norm": 0.674569045840065, "learning_rate": 0.000441, "loss": 7.6778, "step": 147 }, { "epoch": 0.00148, "grad_norm": 0.6220171351450662, "learning_rate": 0.000444, "loss": 7.6627, "step": 148 }, { "epoch": 0.00149, "grad_norm": 0.6302944022326216, "learning_rate": 0.00044699999999999997, "loss": 7.6572, "step": 149 }, { "epoch": 0.0015, "grad_norm": 0.7028824721139024, "learning_rate": 0.00045, "loss": 7.6276, "step": 150 }, { "epoch": 0.00151, "grad_norm": 0.8627581414583672, "learning_rate": 0.000453, "loss": 7.6119, "step": 151 }, { "epoch": 0.00152, "grad_norm": 0.9443366186732501, "learning_rate": 0.000456, "loss": 7.6082, "step": 152 }, { "epoch": 0.00153, "grad_norm": 0.8165953782619042, "learning_rate": 0.000459, "loss": 7.593, "step": 153 }, { "epoch": 0.00154, "grad_norm": 0.7443216104832789, "learning_rate": 0.000462, "loss": 7.5691, "step": 154 }, { "epoch": 0.00155, "grad_norm": 0.8234472492039366, "learning_rate": 0.000465, "loss": 7.5599, "step": 155 }, { "epoch": 0.00156, "grad_norm": 0.6944689992522087, "learning_rate": 0.000468, "loss": 7.5279, "step": 156 }, { "epoch": 0.00157, "grad_norm": 0.6018571838743536, "learning_rate": 0.000471, "loss": 7.5212, "step": 157 }, { "epoch": 0.00158, "grad_norm": 0.6619988530948462, "learning_rate": 0.00047400000000000003, "loss": 7.5046, "step": 158 }, { "epoch": 0.00159, "grad_norm": 0.6265648411112951, "learning_rate": 0.000477, "loss": 7.4897, "step": 159 }, { "epoch": 0.0016, "grad_norm": 0.6464511310681584, "learning_rate": 0.00048, "loss": 7.4689, "step": 160 }, { "epoch": 0.00161, "grad_norm": 0.6320665002117156, "learning_rate": 0.00048300000000000003, "loss": 7.4463, "step": 161 }, { "epoch": 0.00162, "grad_norm": 0.5803650259635545, "learning_rate": 0.00048600000000000005, "loss": 7.4486, "step": 162 }, { "epoch": 0.00163, "grad_norm": 0.5737128458576856, "learning_rate": 0.0004890000000000001, "loss": 7.4298, "step": 163 }, { "epoch": 0.00164, "grad_norm": 0.7222142092183056, "learning_rate": 0.000492, "loss": 7.4324, "step": 164 }, { "epoch": 0.00165, "grad_norm": 0.6646597416212581, "learning_rate": 0.000495, "loss": 7.4085, "step": 165 }, { "epoch": 0.00166, "grad_norm": 0.5400418940901194, "learning_rate": 0.0004980000000000001, "loss": 7.3779, "step": 166 }, { "epoch": 0.00167, "grad_norm": 0.509415377419389, "learning_rate": 0.000501, "loss": 7.3695, "step": 167 }, { "epoch": 0.00168, "grad_norm": 0.6649011711530104, "learning_rate": 0.000504, "loss": 7.3502, "step": 168 }, { "epoch": 0.00169, "grad_norm": 0.47037629648013785, "learning_rate": 0.0005070000000000001, "loss": 7.3396, "step": 169 }, { "epoch": 0.0017, "grad_norm": 0.5403588419672608, "learning_rate": 0.00051, "loss": 7.3246, "step": 170 }, { "epoch": 0.00171, "grad_norm": 0.5602918800765249, "learning_rate": 0.000513, "loss": 7.3054, "step": 171 }, { "epoch": 0.00172, "grad_norm": 0.673389772217806, "learning_rate": 0.000516, "loss": 7.3094, "step": 172 }, { "epoch": 0.00173, "grad_norm": 1.0978277339576752, "learning_rate": 0.0005189999999999999, "loss": 7.2985, "step": 173 }, { "epoch": 0.00174, "grad_norm": 1.0527490659306924, "learning_rate": 0.000522, "loss": 7.3309, "step": 174 }, { "epoch": 0.00175, "grad_norm": 0.447215175816087, "learning_rate": 0.000525, "loss": 7.2686, "step": 175 }, { "epoch": 0.00176, "grad_norm": 1.434718568428607, "learning_rate": 0.0005279999999999999, "loss": 7.2722, "step": 176 }, { "epoch": 0.00177, "grad_norm": 0.515979004078738, "learning_rate": 0.000531, "loss": 7.2504, "step": 177 }, { "epoch": 0.00178, "grad_norm": 0.6894449502470986, "learning_rate": 0.000534, "loss": 7.2483, "step": 178 }, { "epoch": 0.00179, "grad_norm": 0.564765632510306, "learning_rate": 0.000537, "loss": 7.2342, "step": 179 }, { "epoch": 0.0018, "grad_norm": 0.5678924126144868, "learning_rate": 0.00054, "loss": 7.2065, "step": 180 }, { "epoch": 0.00181, "grad_norm": 0.592977565407316, "learning_rate": 0.000543, "loss": 7.1937, "step": 181 }, { "epoch": 0.00182, "grad_norm": 0.562527508746574, "learning_rate": 0.000546, "loss": 7.1887, "step": 182 }, { "epoch": 0.00183, "grad_norm": 0.44528510710573116, "learning_rate": 0.000549, "loss": 7.1767, "step": 183 }, { "epoch": 0.00184, "grad_norm": 0.44473177727238294, "learning_rate": 0.000552, "loss": 7.1633, "step": 184 }, { "epoch": 0.00185, "grad_norm": 0.4257521538615004, "learning_rate": 0.000555, "loss": 7.1411, "step": 185 }, { "epoch": 0.00186, "grad_norm": 0.4409972547719895, "learning_rate": 0.000558, "loss": 7.1489, "step": 186 }, { "epoch": 0.00187, "grad_norm": 0.390495735764414, "learning_rate": 0.000561, "loss": 7.1234, "step": 187 }, { "epoch": 0.00188, "grad_norm": 0.3832212119451659, "learning_rate": 0.000564, "loss": 7.1014, "step": 188 }, { "epoch": 0.00189, "grad_norm": 0.41263318289872636, "learning_rate": 0.000567, "loss": 7.097, "step": 189 }, { "epoch": 0.0019, "grad_norm": 0.42879499878411836, "learning_rate": 0.00057, "loss": 7.0798, "step": 190 }, { "epoch": 0.00191, "grad_norm": 0.45074305144161203, "learning_rate": 0.000573, "loss": 7.0705, "step": 191 }, { "epoch": 0.00192, "grad_norm": 0.49519390713379596, "learning_rate": 0.000576, "loss": 7.0632, "step": 192 }, { "epoch": 0.00193, "grad_norm": 0.5696097064113711, "learning_rate": 0.000579, "loss": 7.0638, "step": 193 }, { "epoch": 0.00194, "grad_norm": 0.534428946566245, "learning_rate": 0.000582, "loss": 7.057, "step": 194 }, { "epoch": 0.00195, "grad_norm": 0.490207896666695, "learning_rate": 0.000585, "loss": 7.0356, "step": 195 }, { "epoch": 0.00196, "grad_norm": 0.4167189031811958, "learning_rate": 0.000588, "loss": 7.0223, "step": 196 }, { "epoch": 0.00197, "grad_norm": 0.48007746323055095, "learning_rate": 0.000591, "loss": 7.0154, "step": 197 }, { "epoch": 0.00198, "grad_norm": 0.876048914805006, "learning_rate": 0.000594, "loss": 7.0165, "step": 198 }, { "epoch": 0.00199, "grad_norm": 1.370346112091805, "learning_rate": 0.0005970000000000001, "loss": 7.0136, "step": 199 }, { "epoch": 0.002, "grad_norm": 0.6762012224362532, "learning_rate": 0.0006000000000000001, "loss": 6.9949, "step": 200 }, { "epoch": 0.00201, "grad_norm": 0.7937633271746596, "learning_rate": 0.000603, "loss": 6.9721, "step": 201 }, { "epoch": 0.00202, "grad_norm": 0.8964746740403945, "learning_rate": 0.0006060000000000001, "loss": 6.9666, "step": 202 }, { "epoch": 0.00203, "grad_norm": 0.7263793741406347, "learning_rate": 0.0006090000000000001, "loss": 6.9637, "step": 203 }, { "epoch": 0.00204, "grad_norm": 0.6263036991113456, "learning_rate": 0.000612, "loss": 6.9498, "step": 204 }, { "epoch": 0.00205, "grad_norm": 0.6750439819273765, "learning_rate": 0.000615, "loss": 6.936, "step": 205 }, { "epoch": 0.00206, "grad_norm": 0.5237479635289414, "learning_rate": 0.000618, "loss": 6.9276, "step": 206 }, { "epoch": 0.00207, "grad_norm": 0.44905870783635865, "learning_rate": 0.000621, "loss": 6.9149, "step": 207 }, { "epoch": 0.00208, "grad_norm": 0.5974162102879337, "learning_rate": 0.000624, "loss": 6.9145, "step": 208 }, { "epoch": 0.00209, "grad_norm": 0.5078613737809945, "learning_rate": 0.000627, "loss": 6.8932, "step": 209 }, { "epoch": 0.0021, "grad_norm": 0.6578576689890827, "learning_rate": 0.00063, "loss": 6.8833, "step": 210 }, { "epoch": 0.00211, "grad_norm": 0.7468480045410584, "learning_rate": 0.000633, "loss": 6.8851, "step": 211 }, { "epoch": 0.00212, "grad_norm": 0.768921505492079, "learning_rate": 0.000636, "loss": 6.8736, "step": 212 }, { "epoch": 0.00213, "grad_norm": 0.43674889526480726, "learning_rate": 0.000639, "loss": 6.8482, "step": 213 }, { "epoch": 0.00214, "grad_norm": 0.44770222580869473, "learning_rate": 0.000642, "loss": 6.8466, "step": 214 }, { "epoch": 0.00215, "grad_norm": 0.4987908268903726, "learning_rate": 0.000645, "loss": 6.8422, "step": 215 }, { "epoch": 0.00216, "grad_norm": 0.3778866757099475, "learning_rate": 0.000648, "loss": 6.8342, "step": 216 }, { "epoch": 0.00217, "grad_norm": 0.503647499102225, "learning_rate": 0.000651, "loss": 6.8089, "step": 217 }, { "epoch": 0.00218, "grad_norm": 0.5993267866741838, "learning_rate": 0.000654, "loss": 6.8135, "step": 218 }, { "epoch": 0.00219, "grad_norm": 0.8234186906536067, "learning_rate": 0.000657, "loss": 6.7909, "step": 219 }, { "epoch": 0.0022, "grad_norm": 0.8231498183396392, "learning_rate": 0.00066, "loss": 6.8151, "step": 220 }, { "epoch": 0.00221, "grad_norm": 0.6347034416977888, "learning_rate": 0.0006630000000000001, "loss": 6.782, "step": 221 }, { "epoch": 0.00222, "grad_norm": 0.6018163518897156, "learning_rate": 0.000666, "loss": 6.7718, "step": 222 }, { "epoch": 0.00223, "grad_norm": 0.6049102827685588, "learning_rate": 0.000669, "loss": 6.7674, "step": 223 }, { "epoch": 0.00224, "grad_norm": 0.4980503627910826, "learning_rate": 0.0006720000000000001, "loss": 6.7608, "step": 224 }, { "epoch": 0.00225, "grad_norm": 0.5566517086736245, "learning_rate": 0.000675, "loss": 6.7493, "step": 225 }, { "epoch": 0.00226, "grad_norm": 0.4371230113022049, "learning_rate": 0.000678, "loss": 6.7428, "step": 226 }, { "epoch": 0.00227, "grad_norm": 0.40182304387748197, "learning_rate": 0.0006810000000000001, "loss": 6.7143, "step": 227 }, { "epoch": 0.00228, "grad_norm": 0.5112677711899808, "learning_rate": 0.000684, "loss": 6.7211, "step": 228 }, { "epoch": 0.00229, "grad_norm": 0.4048973279712385, "learning_rate": 0.000687, "loss": 6.7135, "step": 229 }, { "epoch": 0.0023, "grad_norm": 0.42618577183903295, "learning_rate": 0.0006900000000000001, "loss": 6.691, "step": 230 }, { "epoch": 0.00231, "grad_norm": 0.4712439901030759, "learning_rate": 0.000693, "loss": 6.6777, "step": 231 }, { "epoch": 0.00232, "grad_norm": 0.5546996117138073, "learning_rate": 0.000696, "loss": 6.6902, "step": 232 }, { "epoch": 0.00233, "grad_norm": 0.6251585292690813, "learning_rate": 0.0006990000000000001, "loss": 6.6788, "step": 233 }, { "epoch": 0.00234, "grad_norm": 0.7044048270867854, "learning_rate": 0.000702, "loss": 6.6708, "step": 234 }, { "epoch": 0.00235, "grad_norm": 0.7650244494427701, "learning_rate": 0.000705, "loss": 6.6578, "step": 235 }, { "epoch": 0.00236, "grad_norm": 1.0242839403066175, "learning_rate": 0.000708, "loss": 6.6514, "step": 236 }, { "epoch": 0.00237, "grad_norm": 1.5412268350529807, "learning_rate": 0.0007109999999999999, "loss": 6.6806, "step": 237 }, { "epoch": 0.00238, "grad_norm": 0.7226146264721942, "learning_rate": 0.000714, "loss": 6.645, "step": 238 }, { "epoch": 0.00239, "grad_norm": 0.8688282422006292, "learning_rate": 0.000717, "loss": 6.6357, "step": 239 }, { "epoch": 0.0024, "grad_norm": 0.7827205547140218, "learning_rate": 0.0007199999999999999, "loss": 6.6472, "step": 240 }, { "epoch": 0.00241, "grad_norm": 0.9034551371971334, "learning_rate": 0.000723, "loss": 6.6355, "step": 241 }, { "epoch": 0.00242, "grad_norm": 0.9465232376141441, "learning_rate": 0.000726, "loss": 6.6133, "step": 242 }, { "epoch": 0.00243, "grad_norm": 0.8587844668110328, "learning_rate": 0.000729, "loss": 6.6044, "step": 243 }, { "epoch": 0.00244, "grad_norm": 0.7399571613579958, "learning_rate": 0.000732, "loss": 6.6023, "step": 244 }, { "epoch": 0.00245, "grad_norm": 0.574923768026061, "learning_rate": 0.000735, "loss": 6.5879, "step": 245 }, { "epoch": 0.00246, "grad_norm": 0.605974416339262, "learning_rate": 0.000738, "loss": 6.5742, "step": 246 }, { "epoch": 0.00247, "grad_norm": 0.5710304673194679, "learning_rate": 0.000741, "loss": 6.5636, "step": 247 }, { "epoch": 0.00248, "grad_norm": 0.6965257499480564, "learning_rate": 0.000744, "loss": 6.5776, "step": 248 }, { "epoch": 0.00249, "grad_norm": 0.6456846673011167, "learning_rate": 0.000747, "loss": 6.5561, "step": 249 }, { "epoch": 0.0025, "grad_norm": 0.5753461631243048, "learning_rate": 0.00075, "loss": 6.5548, "step": 250 }, { "epoch": 0.00251, "grad_norm": 0.42353871858505077, "learning_rate": 0.000753, "loss": 6.5158, "step": 251 }, { "epoch": 0.00252, "grad_norm": 0.49557105625334785, "learning_rate": 0.000756, "loss": 6.5269, "step": 252 }, { "epoch": 0.00253, "grad_norm": 0.5766748014233265, "learning_rate": 0.000759, "loss": 6.518, "step": 253 }, { "epoch": 0.00254, "grad_norm": 0.5067846710386406, "learning_rate": 0.000762, "loss": 6.5107, "step": 254 }, { "epoch": 0.00255, "grad_norm": 0.47970421756406123, "learning_rate": 0.0007650000000000001, "loss": 6.5096, "step": 255 }, { "epoch": 0.00256, "grad_norm": 0.5320243074578035, "learning_rate": 0.000768, "loss": 6.4888, "step": 256 }, { "epoch": 0.00257, "grad_norm": 0.6544133785890598, "learning_rate": 0.000771, "loss": 6.4948, "step": 257 }, { "epoch": 0.00258, "grad_norm": 0.7416037390762996, "learning_rate": 0.0007740000000000001, "loss": 6.4783, "step": 258 }, { "epoch": 0.00259, "grad_norm": 0.6373527730888897, "learning_rate": 0.000777, "loss": 6.4749, "step": 259 }, { "epoch": 0.0026, "grad_norm": 0.47810450763796425, "learning_rate": 0.0007800000000000001, "loss": 6.4546, "step": 260 }, { "epoch": 0.00261, "grad_norm": 0.5580595012782733, "learning_rate": 0.0007830000000000001, "loss": 6.4672, "step": 261 }, { "epoch": 0.00262, "grad_norm": 0.5815251459505872, "learning_rate": 0.000786, "loss": 6.4461, "step": 262 }, { "epoch": 0.00263, "grad_norm": 0.5190442800996695, "learning_rate": 0.0007890000000000001, "loss": 6.451, "step": 263 }, { "epoch": 0.00264, "grad_norm": 0.5014715985622594, "learning_rate": 0.0007920000000000001, "loss": 6.434, "step": 264 }, { "epoch": 0.00265, "grad_norm": 0.5767159920930507, "learning_rate": 0.000795, "loss": 6.4267, "step": 265 }, { "epoch": 0.00266, "grad_norm": 0.7264139293538269, "learning_rate": 0.0007980000000000001, "loss": 6.428, "step": 266 }, { "epoch": 0.00267, "grad_norm": 0.912670346256251, "learning_rate": 0.0008010000000000001, "loss": 6.4255, "step": 267 }, { "epoch": 0.00268, "grad_norm": 0.9513958897134662, "learning_rate": 0.000804, "loss": 6.4269, "step": 268 }, { "epoch": 0.00269, "grad_norm": 0.684045795635244, "learning_rate": 0.0008070000000000001, "loss": 6.4015, "step": 269 }, { "epoch": 0.0027, "grad_norm": 0.5702935812973139, "learning_rate": 0.0008100000000000001, "loss": 6.4003, "step": 270 }, { "epoch": 0.00271, "grad_norm": 0.6129714435298934, "learning_rate": 0.000813, "loss": 6.4061, "step": 271 }, { "epoch": 0.00272, "grad_norm": 0.7245489936319242, "learning_rate": 0.0008160000000000001, "loss": 6.393, "step": 272 }, { "epoch": 0.00273, "grad_norm": 0.7005289276676516, "learning_rate": 0.0008190000000000001, "loss": 6.383, "step": 273 }, { "epoch": 0.00274, "grad_norm": 0.8087304828857205, "learning_rate": 0.000822, "loss": 6.3662, "step": 274 }, { "epoch": 0.00275, "grad_norm": 1.1750908820172745, "learning_rate": 0.0008250000000000001, "loss": 6.3881, "step": 275 }, { "epoch": 0.00276, "grad_norm": 1.2316116625020541, "learning_rate": 0.0008280000000000001, "loss": 6.3829, "step": 276 }, { "epoch": 0.00277, "grad_norm": 0.8466007057008188, "learning_rate": 0.0008310000000000001, "loss": 6.3637, "step": 277 }, { "epoch": 0.00278, "grad_norm": 1.0699566045014268, "learning_rate": 0.0008340000000000001, "loss": 6.3585, "step": 278 }, { "epoch": 0.00279, "grad_norm": 0.9176192714012301, "learning_rate": 0.0008370000000000001, "loss": 6.3555, "step": 279 }, { "epoch": 0.0028, "grad_norm": 0.8370570497956159, "learning_rate": 0.0008400000000000001, "loss": 6.337, "step": 280 }, { "epoch": 0.00281, "grad_norm": 0.9837106590112089, "learning_rate": 0.0008430000000000001, "loss": 6.3349, "step": 281 }, { "epoch": 0.00282, "grad_norm": 0.9638325924361532, "learning_rate": 0.000846, "loss": 6.3337, "step": 282 }, { "epoch": 0.00283, "grad_norm": 0.8201342967629913, "learning_rate": 0.0008489999999999999, "loss": 6.3234, "step": 283 }, { "epoch": 0.00284, "grad_norm": 0.7494504986377379, "learning_rate": 0.0008519999999999999, "loss": 6.3111, "step": 284 }, { "epoch": 0.00285, "grad_norm": 0.6500575839227768, "learning_rate": 0.000855, "loss": 6.314, "step": 285 }, { "epoch": 0.00286, "grad_norm": 0.6158978014821535, "learning_rate": 0.0008579999999999999, "loss": 6.2896, "step": 286 }, { "epoch": 0.00287, "grad_norm": 0.5596137127282185, "learning_rate": 0.000861, "loss": 6.2911, "step": 287 }, { "epoch": 0.00288, "grad_norm": 0.5983803771698389, "learning_rate": 0.000864, "loss": 6.2729, "step": 288 }, { "epoch": 0.00289, "grad_norm": 0.6730668726749236, "learning_rate": 0.0008669999999999999, "loss": 6.2867, "step": 289 }, { "epoch": 0.0029, "grad_norm": 0.7493875593433929, "learning_rate": 0.00087, "loss": 6.2693, "step": 290 }, { "epoch": 0.00291, "grad_norm": 0.7983375848355683, "learning_rate": 0.000873, "loss": 6.2651, "step": 291 }, { "epoch": 0.00292, "grad_norm": 0.8346121383412798, "learning_rate": 0.0008759999999999999, "loss": 6.2655, "step": 292 }, { "epoch": 0.00293, "grad_norm": 0.915961517973993, "learning_rate": 0.000879, "loss": 6.2508, "step": 293 }, { "epoch": 0.00294, "grad_norm": 0.966838750452259, "learning_rate": 0.000882, "loss": 6.2734, "step": 294 }, { "epoch": 0.00295, "grad_norm": 0.479548574482138, "learning_rate": 0.0008849999999999999, "loss": 6.2216, "step": 295 }, { "epoch": 0.00296, "grad_norm": 0.583968812139162, "learning_rate": 0.000888, "loss": 6.2407, "step": 296 }, { "epoch": 0.00297, "grad_norm": 0.5310639812280135, "learning_rate": 0.000891, "loss": 6.2108, "step": 297 }, { "epoch": 0.00298, "grad_norm": 0.5175342241650946, "learning_rate": 0.0008939999999999999, "loss": 6.2236, "step": 298 }, { "epoch": 0.00299, "grad_norm": 0.4652973315008358, "learning_rate": 0.000897, "loss": 6.1974, "step": 299 }, { "epoch": 0.003, "grad_norm": 0.45180686655250246, "learning_rate": 0.0009, "loss": 6.1981, "step": 300 }, { "epoch": 0.00301, "grad_norm": 0.4267045703969487, "learning_rate": 0.0009029999999999999, "loss": 6.1896, "step": 301 }, { "epoch": 0.00302, "grad_norm": 0.42788415431197124, "learning_rate": 0.000906, "loss": 6.1747, "step": 302 }, { "epoch": 0.00303, "grad_norm": 0.44705407060167995, "learning_rate": 0.000909, "loss": 6.1846, "step": 303 }, { "epoch": 0.00304, "grad_norm": 0.6488030623423143, "learning_rate": 0.000912, "loss": 6.1818, "step": 304 }, { "epoch": 0.00305, "grad_norm": 1.1848923233227069, "learning_rate": 0.000915, "loss": 6.1971, "step": 305 }, { "epoch": 0.00306, "grad_norm": 1.146308347479262, "learning_rate": 0.000918, "loss": 6.1896, "step": 306 }, { "epoch": 0.00307, "grad_norm": 0.5610571056277671, "learning_rate": 0.000921, "loss": 6.1649, "step": 307 }, { "epoch": 0.00308, "grad_norm": 0.6387882167592376, "learning_rate": 0.000924, "loss": 6.1535, "step": 308 }, { "epoch": 0.00309, "grad_norm": 0.6644607985790572, "learning_rate": 0.000927, "loss": 6.1687, "step": 309 }, { "epoch": 0.0031, "grad_norm": 0.9254403317850548, "learning_rate": 0.00093, "loss": 6.1578, "step": 310 }, { "epoch": 0.00311, "grad_norm": 1.1729449300284347, "learning_rate": 0.000933, "loss": 6.1641, "step": 311 }, { "epoch": 0.00312, "grad_norm": 0.709455259431002, "learning_rate": 0.000936, "loss": 6.1296, "step": 312 }, { "epoch": 0.00313, "grad_norm": 0.9111247454263828, "learning_rate": 0.0009390000000000001, "loss": 6.1519, "step": 313 }, { "epoch": 0.00314, "grad_norm": 0.9354538212933341, "learning_rate": 0.000942, "loss": 6.1613, "step": 314 }, { "epoch": 0.00315, "grad_norm": 0.8799875640055459, "learning_rate": 0.000945, "loss": 6.1333, "step": 315 }, { "epoch": 0.00316, "grad_norm": 0.9808323579373607, "learning_rate": 0.0009480000000000001, "loss": 6.1147, "step": 316 }, { "epoch": 0.00317, "grad_norm": 0.9583160858065677, "learning_rate": 0.000951, "loss": 6.1333, "step": 317 }, { "epoch": 0.00318, "grad_norm": 1.0107022128845053, "learning_rate": 0.000954, "loss": 6.1332, "step": 318 }, { "epoch": 0.00319, "grad_norm": 0.9453172619742498, "learning_rate": 0.0009570000000000001, "loss": 6.1183, "step": 319 }, { "epoch": 0.0032, "grad_norm": 0.881992443544615, "learning_rate": 0.00096, "loss": 6.1175, "step": 320 }, { "epoch": 0.00321, "grad_norm": 0.8621522106832189, "learning_rate": 0.000963, "loss": 6.1031, "step": 321 }, { "epoch": 0.00322, "grad_norm": 0.8993520755674765, "learning_rate": 0.0009660000000000001, "loss": 6.0937, "step": 322 }, { "epoch": 0.00323, "grad_norm": 0.8168073718243659, "learning_rate": 0.000969, "loss": 6.0911, "step": 323 }, { "epoch": 0.00324, "grad_norm": 0.5742393221545353, "learning_rate": 0.0009720000000000001, "loss": 6.0859, "step": 324 }, { "epoch": 0.00325, "grad_norm": 0.6225863989817222, "learning_rate": 0.0009750000000000001, "loss": 6.0536, "step": 325 }, { "epoch": 0.00326, "grad_norm": 0.6133472134077819, "learning_rate": 0.0009780000000000001, "loss": 6.0689, "step": 326 }, { "epoch": 0.00327, "grad_norm": 0.5150229110459733, "learning_rate": 0.000981, "loss": 6.0685, "step": 327 }, { "epoch": 0.00328, "grad_norm": 0.4607860070893556, "learning_rate": 0.000984, "loss": 6.042, "step": 328 }, { "epoch": 0.00329, "grad_norm": 0.5192581817133665, "learning_rate": 0.000987, "loss": 6.0442, "step": 329 }, { "epoch": 0.0033, "grad_norm": 0.5292078450907967, "learning_rate": 0.00099, "loss": 6.0364, "step": 330 }, { "epoch": 0.00331, "grad_norm": 0.6947168538867383, "learning_rate": 0.0009930000000000002, "loss": 6.0435, "step": 331 }, { "epoch": 0.00332, "grad_norm": 0.8487757774754058, "learning_rate": 0.0009960000000000001, "loss": 6.0304, "step": 332 }, { "epoch": 0.00333, "grad_norm": 0.899631264631538, "learning_rate": 0.000999, "loss": 6.0174, "step": 333 }, { "epoch": 0.00334, "grad_norm": 0.7073321966133805, "learning_rate": 0.001002, "loss": 6.0239, "step": 334 }, { "epoch": 0.00335, "grad_norm": 0.7526163680323161, "learning_rate": 0.001005, "loss": 6.0225, "step": 335 }, { "epoch": 0.00336, "grad_norm": 1.0400532380171426, "learning_rate": 0.001008, "loss": 6.0251, "step": 336 }, { "epoch": 0.00337, "grad_norm": 0.8580613520432612, "learning_rate": 0.0010110000000000002, "loss": 5.9967, "step": 337 }, { "epoch": 0.00338, "grad_norm": 0.8200745373702635, "learning_rate": 0.0010140000000000001, "loss": 6.014, "step": 338 }, { "epoch": 0.00339, "grad_norm": 0.7727894868210932, "learning_rate": 0.0010170000000000001, "loss": 6.0161, "step": 339 }, { "epoch": 0.0034, "grad_norm": 0.8707811190707261, "learning_rate": 0.00102, "loss": 5.9943, "step": 340 }, { "epoch": 0.00341, "grad_norm": 0.7603523129117276, "learning_rate": 0.001023, "loss": 5.9713, "step": 341 }, { "epoch": 0.00342, "grad_norm": 0.726490122797017, "learning_rate": 0.001026, "loss": 5.9758, "step": 342 }, { "epoch": 0.00343, "grad_norm": 0.7698838300440034, "learning_rate": 0.0010290000000000002, "loss": 5.9813, "step": 343 }, { "epoch": 0.00344, "grad_norm": 0.810359133047605, "learning_rate": 0.001032, "loss": 5.9733, "step": 344 }, { "epoch": 0.00345, "grad_norm": 0.7483100579200008, "learning_rate": 0.001035, "loss": 5.9619, "step": 345 }, { "epoch": 0.00346, "grad_norm": 0.7290056537727018, "learning_rate": 0.0010379999999999999, "loss": 5.9677, "step": 346 }, { "epoch": 0.00347, "grad_norm": 0.879931767534797, "learning_rate": 0.001041, "loss": 5.9453, "step": 347 }, { "epoch": 0.00348, "grad_norm": 0.821531229371674, "learning_rate": 0.001044, "loss": 5.9515, "step": 348 }, { "epoch": 0.00349, "grad_norm": 0.7749600052843896, "learning_rate": 0.001047, "loss": 5.9348, "step": 349 }, { "epoch": 0.0035, "grad_norm": 0.8742079309636731, "learning_rate": 0.00105, "loss": 5.9574, "step": 350 }, { "epoch": 0.00351, "grad_norm": 0.75264610249205, "learning_rate": 0.001053, "loss": 5.9539, "step": 351 }, { "epoch": 0.00352, "grad_norm": 0.6495925446513019, "learning_rate": 0.0010559999999999999, "loss": 5.9277, "step": 352 }, { "epoch": 0.00353, "grad_norm": 0.7308566609637007, "learning_rate": 0.001059, "loss": 5.9295, "step": 353 }, { "epoch": 0.00354, "grad_norm": 0.7419374622409085, "learning_rate": 0.001062, "loss": 5.9081, "step": 354 }, { "epoch": 0.00355, "grad_norm": 0.6305963094195233, "learning_rate": 0.001065, "loss": 5.9163, "step": 355 }, { "epoch": 0.00356, "grad_norm": 0.7593726199074547, "learning_rate": 0.001068, "loss": 5.9116, "step": 356 }, { "epoch": 0.00357, "grad_norm": 0.7130446666608321, "learning_rate": 0.001071, "loss": 5.9022, "step": 357 }, { "epoch": 0.00358, "grad_norm": 0.705496115321125, "learning_rate": 0.001074, "loss": 5.8948, "step": 358 }, { "epoch": 0.00359, "grad_norm": 0.7437786860075282, "learning_rate": 0.001077, "loss": 5.8901, "step": 359 }, { "epoch": 0.0036, "grad_norm": 0.6350063029419483, "learning_rate": 0.00108, "loss": 5.8869, "step": 360 }, { "epoch": 0.00361, "grad_norm": 0.7400426345488472, "learning_rate": 0.001083, "loss": 5.886, "step": 361 }, { "epoch": 0.00362, "grad_norm": 0.6062975404814006, "learning_rate": 0.001086, "loss": 5.8687, "step": 362 }, { "epoch": 0.00363, "grad_norm": 0.7647489229089097, "learning_rate": 0.001089, "loss": 5.8616, "step": 363 }, { "epoch": 0.00364, "grad_norm": 1.0426193764053155, "learning_rate": 0.001092, "loss": 5.8896, "step": 364 }, { "epoch": 0.00365, "grad_norm": 0.9699640617637266, "learning_rate": 0.001095, "loss": 5.8858, "step": 365 }, { "epoch": 0.00366, "grad_norm": 0.574276128757519, "learning_rate": 0.001098, "loss": 5.8508, "step": 366 }, { "epoch": 0.00367, "grad_norm": 0.6185761162607817, "learning_rate": 0.001101, "loss": 5.8561, "step": 367 }, { "epoch": 0.00368, "grad_norm": 0.5694281157783647, "learning_rate": 0.001104, "loss": 5.8542, "step": 368 }, { "epoch": 0.00369, "grad_norm": 0.6437285541711164, "learning_rate": 0.001107, "loss": 5.8594, "step": 369 }, { "epoch": 0.0037, "grad_norm": 0.8677179765560395, "learning_rate": 0.00111, "loss": 5.8442, "step": 370 }, { "epoch": 0.00371, "grad_norm": 0.9363826836877185, "learning_rate": 0.001113, "loss": 5.823, "step": 371 }, { "epoch": 0.00372, "grad_norm": 0.6769054561744376, "learning_rate": 0.001116, "loss": 5.835, "step": 372 }, { "epoch": 0.00373, "grad_norm": 0.711623190683124, "learning_rate": 0.001119, "loss": 5.827, "step": 373 }, { "epoch": 0.00374, "grad_norm": 0.8787133685385395, "learning_rate": 0.001122, "loss": 5.8376, "step": 374 }, { "epoch": 0.00375, "grad_norm": 1.1697788362092756, "learning_rate": 0.0011250000000000001, "loss": 5.8511, "step": 375 }, { "epoch": 0.00376, "grad_norm": 0.9709193918687598, "learning_rate": 0.001128, "loss": 5.8427, "step": 376 }, { "epoch": 0.00377, "grad_norm": 1.0376916283065132, "learning_rate": 0.001131, "loss": 5.8397, "step": 377 }, { "epoch": 0.00378, "grad_norm": 1.2375411877265758, "learning_rate": 0.001134, "loss": 5.8481, "step": 378 }, { "epoch": 0.00379, "grad_norm": 0.9944842378779577, "learning_rate": 0.001137, "loss": 5.822, "step": 379 }, { "epoch": 0.0038, "grad_norm": 1.1307760763710821, "learning_rate": 0.00114, "loss": 5.859, "step": 380 }, { "epoch": 0.00381, "grad_norm": 0.8573478329639129, "learning_rate": 0.0011430000000000001, "loss": 5.8136, "step": 381 }, { "epoch": 0.00382, "grad_norm": 0.7268105691491051, "learning_rate": 0.001146, "loss": 5.7943, "step": 382 }, { "epoch": 0.00383, "grad_norm": 0.7419446712141832, "learning_rate": 0.001149, "loss": 5.8074, "step": 383 }, { "epoch": 0.00384, "grad_norm": 0.7923675534238995, "learning_rate": 0.001152, "loss": 5.81, "step": 384 }, { "epoch": 0.00385, "grad_norm": 0.9624110447648943, "learning_rate": 0.001155, "loss": 5.805, "step": 385 }, { "epoch": 0.00386, "grad_norm": 0.9319530344899567, "learning_rate": 0.001158, "loss": 5.8081, "step": 386 }, { "epoch": 0.00387, "grad_norm": 0.7643274958196878, "learning_rate": 0.0011610000000000001, "loss": 5.7788, "step": 387 }, { "epoch": 0.00388, "grad_norm": 0.7506191440573698, "learning_rate": 0.001164, "loss": 5.7762, "step": 388 }, { "epoch": 0.00389, "grad_norm": 0.7532595251955709, "learning_rate": 0.001167, "loss": 5.7827, "step": 389 }, { "epoch": 0.0039, "grad_norm": 0.73320332461548, "learning_rate": 0.00117, "loss": 5.7649, "step": 390 }, { "epoch": 0.00391, "grad_norm": 0.7031896008677531, "learning_rate": 0.001173, "loss": 5.7704, "step": 391 }, { "epoch": 0.00392, "grad_norm": 0.6785810028909003, "learning_rate": 0.001176, "loss": 5.7796, "step": 392 }, { "epoch": 0.00393, "grad_norm": 0.4867762853103691, "learning_rate": 0.0011790000000000001, "loss": 5.7469, "step": 393 }, { "epoch": 0.00394, "grad_norm": 0.4635737108120594, "learning_rate": 0.001182, "loss": 5.7346, "step": 394 }, { "epoch": 0.00395, "grad_norm": 0.46294401028763466, "learning_rate": 0.001185, "loss": 5.7175, "step": 395 }, { "epoch": 0.00396, "grad_norm": 0.4787180663927712, "learning_rate": 0.001188, "loss": 5.7326, "step": 396 }, { "epoch": 0.00397, "grad_norm": 0.473675499009933, "learning_rate": 0.001191, "loss": 5.7249, "step": 397 }, { "epoch": 0.00398, "grad_norm": 0.4875545781350187, "learning_rate": 0.0011940000000000002, "loss": 5.7309, "step": 398 }, { "epoch": 0.00399, "grad_norm": 0.562929492137865, "learning_rate": 0.0011970000000000001, "loss": 5.7326, "step": 399 }, { "epoch": 0.004, "grad_norm": 0.5814123728626063, "learning_rate": 0.0012000000000000001, "loss": 5.7207, "step": 400 }, { "epoch": 0.00401, "grad_norm": 0.5133861042849526, "learning_rate": 0.001203, "loss": 5.6995, "step": 401 }, { "epoch": 0.00402, "grad_norm": 0.42439274620769013, "learning_rate": 0.001206, "loss": 5.7051, "step": 402 }, { "epoch": 0.00403, "grad_norm": 0.46979039810569395, "learning_rate": 0.001209, "loss": 5.7011, "step": 403 }, { "epoch": 0.00404, "grad_norm": 0.6400564820167204, "learning_rate": 0.0012120000000000002, "loss": 5.6973, "step": 404 }, { "epoch": 0.00405, "grad_norm": 0.9145906738089299, "learning_rate": 0.0012150000000000002, "loss": 5.7031, "step": 405 }, { "epoch": 0.00406, "grad_norm": 1.3255500600189978, "learning_rate": 0.0012180000000000001, "loss": 5.7063, "step": 406 }, { "epoch": 0.00407, "grad_norm": 0.8028038937216726, "learning_rate": 0.0012209999999999999, "loss": 5.688, "step": 407 }, { "epoch": 0.00408, "grad_norm": 0.9847538347741682, "learning_rate": 0.001224, "loss": 5.7174, "step": 408 }, { "epoch": 0.00409, "grad_norm": 0.837601769367776, "learning_rate": 0.001227, "loss": 5.6896, "step": 409 }, { "epoch": 0.0041, "grad_norm": 1.2038243323400906, "learning_rate": 0.00123, "loss": 5.7085, "step": 410 }, { "epoch": 0.00411, "grad_norm": 0.7395928558220229, "learning_rate": 0.001233, "loss": 5.6997, "step": 411 }, { "epoch": 0.00412, "grad_norm": 0.8216091961802617, "learning_rate": 0.001236, "loss": 5.6935, "step": 412 }, { "epoch": 0.00413, "grad_norm": 1.140214049358121, "learning_rate": 0.0012389999999999999, "loss": 5.7021, "step": 413 }, { "epoch": 0.00414, "grad_norm": 1.039942035452628, "learning_rate": 0.001242, "loss": 5.6845, "step": 414 }, { "epoch": 0.00415, "grad_norm": 1.134607365917555, "learning_rate": 0.001245, "loss": 5.6908, "step": 415 }, { "epoch": 0.00416, "grad_norm": 0.9596978489821891, "learning_rate": 0.001248, "loss": 5.6703, "step": 416 }, { "epoch": 0.00417, "grad_norm": 1.043455051006676, "learning_rate": 0.001251, "loss": 5.7098, "step": 417 }, { "epoch": 0.00418, "grad_norm": 1.165018490109447, "learning_rate": 0.001254, "loss": 5.6861, "step": 418 }, { "epoch": 0.00419, "grad_norm": 1.007469512856913, "learning_rate": 0.0012569999999999999, "loss": 5.6646, "step": 419 }, { "epoch": 0.0042, "grad_norm": 1.410591782248098, "learning_rate": 0.00126, "loss": 5.7025, "step": 420 }, { "epoch": 0.00421, "grad_norm": 0.7733656450445685, "learning_rate": 0.001263, "loss": 5.6715, "step": 421 }, { "epoch": 0.00422, "grad_norm": 0.6699538596951103, "learning_rate": 0.001266, "loss": 5.6588, "step": 422 }, { "epoch": 0.00423, "grad_norm": 0.5873287754113306, "learning_rate": 0.001269, "loss": 5.6538, "step": 423 }, { "epoch": 0.00424, "grad_norm": 0.6444567501036798, "learning_rate": 0.001272, "loss": 5.6532, "step": 424 }, { "epoch": 0.00425, "grad_norm": 0.6334269686168538, "learning_rate": 0.001275, "loss": 5.6565, "step": 425 }, { "epoch": 0.00426, "grad_norm": 0.6731028337322849, "learning_rate": 0.001278, "loss": 5.655, "step": 426 }, { "epoch": 0.00427, "grad_norm": 0.5902581772724022, "learning_rate": 0.001281, "loss": 5.6354, "step": 427 }, { "epoch": 0.00428, "grad_norm": 0.5926402895935227, "learning_rate": 0.001284, "loss": 5.6139, "step": 428 }, { "epoch": 0.00429, "grad_norm": 0.6674533850736829, "learning_rate": 0.001287, "loss": 5.6273, "step": 429 }, { "epoch": 0.0043, "grad_norm": 0.7049075078295334, "learning_rate": 0.00129, "loss": 5.6258, "step": 430 }, { "epoch": 0.00431, "grad_norm": 0.7061497333512059, "learning_rate": 0.001293, "loss": 5.6127, "step": 431 }, { "epoch": 0.00432, "grad_norm": 0.7508446834433861, "learning_rate": 0.001296, "loss": 5.6216, "step": 432 }, { "epoch": 0.00433, "grad_norm": 0.9060630970118144, "learning_rate": 0.001299, "loss": 5.6177, "step": 433 }, { "epoch": 0.00434, "grad_norm": 0.9999640540361012, "learning_rate": 0.001302, "loss": 5.6265, "step": 434 }, { "epoch": 0.00435, "grad_norm": 0.7892162060492922, "learning_rate": 0.001305, "loss": 5.6142, "step": 435 }, { "epoch": 0.00436, "grad_norm": 0.7307564408784262, "learning_rate": 0.001308, "loss": 5.5916, "step": 436 }, { "epoch": 0.00437, "grad_norm": 0.6809142164228668, "learning_rate": 0.001311, "loss": 5.6072, "step": 437 }, { "epoch": 0.00438, "grad_norm": 0.7111840477444925, "learning_rate": 0.001314, "loss": 5.5903, "step": 438 }, { "epoch": 0.00439, "grad_norm": 0.6900645963261404, "learning_rate": 0.001317, "loss": 5.6171, "step": 439 }, { "epoch": 0.0044, "grad_norm": 0.819282796874481, "learning_rate": 0.00132, "loss": 5.5903, "step": 440 }, { "epoch": 0.00441, "grad_norm": 0.8210691697996518, "learning_rate": 0.001323, "loss": 5.5903, "step": 441 }, { "epoch": 0.00442, "grad_norm": 0.7312534209405126, "learning_rate": 0.0013260000000000001, "loss": 5.579, "step": 442 }, { "epoch": 0.00443, "grad_norm": 0.8543448836580776, "learning_rate": 0.001329, "loss": 5.5857, "step": 443 }, { "epoch": 0.00444, "grad_norm": 0.9642366068930326, "learning_rate": 0.001332, "loss": 5.595, "step": 444 }, { "epoch": 0.00445, "grad_norm": 1.0151719432819601, "learning_rate": 0.001335, "loss": 5.5844, "step": 445 }, { "epoch": 0.00446, "grad_norm": 0.9098827299445092, "learning_rate": 0.001338, "loss": 5.5855, "step": 446 }, { "epoch": 0.00447, "grad_norm": 0.7366716149436047, "learning_rate": 0.001341, "loss": 5.5785, "step": 447 }, { "epoch": 0.00448, "grad_norm": 0.688063707422099, "learning_rate": 0.0013440000000000001, "loss": 5.5612, "step": 448 }, { "epoch": 0.00449, "grad_norm": 0.6289607789766871, "learning_rate": 0.001347, "loss": 5.5606, "step": 449 }, { "epoch": 0.0045, "grad_norm": 0.8296220427029676, "learning_rate": 0.00135, "loss": 5.5654, "step": 450 }, { "epoch": 0.00451, "grad_norm": 0.8639016329270134, "learning_rate": 0.001353, "loss": 5.5574, "step": 451 }, { "epoch": 0.00452, "grad_norm": 0.8710379060383533, "learning_rate": 0.001356, "loss": 5.5436, "step": 452 }, { "epoch": 0.00453, "grad_norm": 0.9193516591602126, "learning_rate": 0.001359, "loss": 5.5351, "step": 453 }, { "epoch": 0.00454, "grad_norm": 1.1479753211147432, "learning_rate": 0.0013620000000000001, "loss": 5.5609, "step": 454 }, { "epoch": 0.00455, "grad_norm": 0.777429711985851, "learning_rate": 0.0013650000000000001, "loss": 5.5488, "step": 455 }, { "epoch": 0.00456, "grad_norm": 0.6628903616445764, "learning_rate": 0.001368, "loss": 5.5436, "step": 456 }, { "epoch": 0.00457, "grad_norm": 0.6885714365872945, "learning_rate": 0.001371, "loss": 5.5338, "step": 457 }, { "epoch": 0.00458, "grad_norm": 0.8476587819535301, "learning_rate": 0.001374, "loss": 5.5439, "step": 458 }, { "epoch": 0.00459, "grad_norm": 1.0268132771248308, "learning_rate": 0.0013770000000000002, "loss": 5.5398, "step": 459 }, { "epoch": 0.0046, "grad_norm": 0.829796705713153, "learning_rate": 0.0013800000000000002, "loss": 5.5243, "step": 460 }, { "epoch": 0.00461, "grad_norm": 0.682304970956673, "learning_rate": 0.0013830000000000001, "loss": 5.5134, "step": 461 }, { "epoch": 0.00462, "grad_norm": 0.7632352528809807, "learning_rate": 0.001386, "loss": 5.5282, "step": 462 }, { "epoch": 0.00463, "grad_norm": 0.7888238491454299, "learning_rate": 0.001389, "loss": 5.5068, "step": 463 }, { "epoch": 0.00464, "grad_norm": 0.6055698609216487, "learning_rate": 0.001392, "loss": 5.5197, "step": 464 }, { "epoch": 0.00465, "grad_norm": 0.534605737151693, "learning_rate": 0.0013950000000000002, "loss": 5.4889, "step": 465 }, { "epoch": 0.00466, "grad_norm": 0.5742494169981804, "learning_rate": 0.0013980000000000002, "loss": 5.5068, "step": 466 }, { "epoch": 0.00467, "grad_norm": 0.5366049876051217, "learning_rate": 0.0014010000000000001, "loss": 5.495, "step": 467 }, { "epoch": 0.00468, "grad_norm": 0.5022698964006017, "learning_rate": 0.001404, "loss": 5.4802, "step": 468 }, { "epoch": 0.00469, "grad_norm": 0.4740914476325971, "learning_rate": 0.001407, "loss": 5.4823, "step": 469 }, { "epoch": 0.0047, "grad_norm": 0.5019536528871698, "learning_rate": 0.00141, "loss": 5.4905, "step": 470 }, { "epoch": 0.00471, "grad_norm": 0.4682933268188089, "learning_rate": 0.001413, "loss": 5.4686, "step": 471 }, { "epoch": 0.00472, "grad_norm": 0.5247906848769013, "learning_rate": 0.001416, "loss": 5.4781, "step": 472 }, { "epoch": 0.00473, "grad_norm": 0.639980530606925, "learning_rate": 0.001419, "loss": 5.4907, "step": 473 }, { "epoch": 0.00474, "grad_norm": 0.7807932825015014, "learning_rate": 0.0014219999999999999, "loss": 5.4763, "step": 474 }, { "epoch": 0.00475, "grad_norm": 0.7383779646696264, "learning_rate": 0.001425, "loss": 5.469, "step": 475 }, { "epoch": 0.00476, "grad_norm": 0.5375784338255847, "learning_rate": 0.001428, "loss": 5.4946, "step": 476 }, { "epoch": 0.00477, "grad_norm": 0.7920239446472193, "learning_rate": 0.001431, "loss": 5.4563, "step": 477 }, { "epoch": 0.00478, "grad_norm": 0.9834272495974716, "learning_rate": 0.001434, "loss": 5.4917, "step": 478 }, { "epoch": 0.00479, "grad_norm": 1.1904363785289231, "learning_rate": 0.001437, "loss": 5.4771, "step": 479 }, { "epoch": 0.0048, "grad_norm": 0.8836750578846579, "learning_rate": 0.0014399999999999999, "loss": 5.462, "step": 480 }, { "epoch": 0.00481, "grad_norm": 1.0971029164726256, "learning_rate": 0.001443, "loss": 5.4854, "step": 481 }, { "epoch": 0.00482, "grad_norm": 1.0424527071772058, "learning_rate": 0.001446, "loss": 5.4551, "step": 482 }, { "epoch": 0.00483, "grad_norm": 1.1383385814972007, "learning_rate": 0.001449, "loss": 5.4737, "step": 483 }, { "epoch": 0.00484, "grad_norm": 1.1589895400739378, "learning_rate": 0.001452, "loss": 5.4781, "step": 484 }, { "epoch": 0.00485, "grad_norm": 1.223866148569803, "learning_rate": 0.001455, "loss": 5.4947, "step": 485 }, { "epoch": 0.00486, "grad_norm": 0.6598880348489398, "learning_rate": 0.001458, "loss": 5.4403, "step": 486 }, { "epoch": 0.00487, "grad_norm": 0.8338085747463339, "learning_rate": 0.001461, "loss": 5.4654, "step": 487 }, { "epoch": 0.00488, "grad_norm": 0.8132081616614524, "learning_rate": 0.001464, "loss": 5.4355, "step": 488 }, { "epoch": 0.00489, "grad_norm": 0.8613199238884757, "learning_rate": 0.001467, "loss": 5.4338, "step": 489 }, { "epoch": 0.0049, "grad_norm": 0.7539827955426024, "learning_rate": 0.00147, "loss": 5.4409, "step": 490 }, { "epoch": 0.00491, "grad_norm": 0.6915747459996715, "learning_rate": 0.001473, "loss": 5.4368, "step": 491 }, { "epoch": 0.00492, "grad_norm": 0.6709201812755653, "learning_rate": 0.001476, "loss": 5.431, "step": 492 }, { "epoch": 0.00493, "grad_norm": 0.6684348811489516, "learning_rate": 0.001479, "loss": 5.4267, "step": 493 }, { "epoch": 0.00494, "grad_norm": 0.5545769515776193, "learning_rate": 0.001482, "loss": 5.428, "step": 494 }, { "epoch": 0.00495, "grad_norm": 0.5667135849985778, "learning_rate": 0.001485, "loss": 5.4318, "step": 495 }, { "epoch": 0.00496, "grad_norm": 0.553251223973209, "learning_rate": 0.001488, "loss": 5.4133, "step": 496 }, { "epoch": 0.00497, "grad_norm": 0.5321464111975676, "learning_rate": 0.001491, "loss": 5.3957, "step": 497 }, { "epoch": 0.00498, "grad_norm": 0.49096949355178177, "learning_rate": 0.001494, "loss": 5.419, "step": 498 }, { "epoch": 0.00499, "grad_norm": 0.5792511038424676, "learning_rate": 0.001497, "loss": 5.3935, "step": 499 }, { "epoch": 0.005, "grad_norm": 0.7094946648983601, "learning_rate": 0.0015, "loss": 5.4021, "step": 500 }, { "epoch": 0.00501, "grad_norm": 0.7830780899072106, "learning_rate": 0.001503, "loss": 5.4012, "step": 501 }, { "epoch": 0.00502, "grad_norm": 0.8705740259484833, "learning_rate": 0.001506, "loss": 5.4132, "step": 502 }, { "epoch": 0.00503, "grad_norm": 0.8118458630499436, "learning_rate": 0.0015090000000000001, "loss": 5.4131, "step": 503 }, { "epoch": 0.00504, "grad_norm": 0.8780863177851627, "learning_rate": 0.001512, "loss": 5.4024, "step": 504 }, { "epoch": 0.00505, "grad_norm": 0.8916117611460668, "learning_rate": 0.001515, "loss": 5.3938, "step": 505 }, { "epoch": 0.00506, "grad_norm": 0.9362197750510605, "learning_rate": 0.001518, "loss": 5.3997, "step": 506 }, { "epoch": 0.00507, "grad_norm": 1.0130920325114388, "learning_rate": 0.001521, "loss": 5.4015, "step": 507 }, { "epoch": 0.00508, "grad_norm": 0.8372448631493079, "learning_rate": 0.001524, "loss": 5.3973, "step": 508 }, { "epoch": 0.00509, "grad_norm": 0.8595290882383629, "learning_rate": 0.0015270000000000001, "loss": 5.3824, "step": 509 }, { "epoch": 0.0051, "grad_norm": 0.8908771623663538, "learning_rate": 0.0015300000000000001, "loss": 5.4017, "step": 510 }, { "epoch": 0.00511, "grad_norm": 1.018640071103151, "learning_rate": 0.001533, "loss": 5.3841, "step": 511 }, { "epoch": 0.00512, "grad_norm": 1.1039344108877285, "learning_rate": 0.001536, "loss": 5.3839, "step": 512 }, { "epoch": 0.00513, "grad_norm": 1.018705128757106, "learning_rate": 0.001539, "loss": 5.3992, "step": 513 }, { "epoch": 0.00514, "grad_norm": 1.0856823606740633, "learning_rate": 0.001542, "loss": 5.3891, "step": 514 }, { "epoch": 0.00515, "grad_norm": 1.0775919048452265, "learning_rate": 0.0015450000000000001, "loss": 5.3996, "step": 515 }, { "epoch": 0.00516, "grad_norm": 1.0104511780701346, "learning_rate": 0.0015480000000000001, "loss": 5.3964, "step": 516 }, { "epoch": 0.00517, "grad_norm": 1.0413855474558291, "learning_rate": 0.001551, "loss": 5.3886, "step": 517 }, { "epoch": 0.00518, "grad_norm": 0.7612973684089466, "learning_rate": 0.001554, "loss": 5.3656, "step": 518 }, { "epoch": 0.00519, "grad_norm": 0.6774626111559078, "learning_rate": 0.001557, "loss": 5.3631, "step": 519 }, { "epoch": 0.0052, "grad_norm": 0.5696849549020969, "learning_rate": 0.0015600000000000002, "loss": 5.3692, "step": 520 }, { "epoch": 0.00521, "grad_norm": 0.5729212961920945, "learning_rate": 0.0015630000000000002, "loss": 5.3421, "step": 521 }, { "epoch": 0.00522, "grad_norm": 0.6524634891271429, "learning_rate": 0.0015660000000000001, "loss": 5.3482, "step": 522 }, { "epoch": 0.00523, "grad_norm": 0.6726078234829947, "learning_rate": 0.001569, "loss": 5.3604, "step": 523 }, { "epoch": 0.00524, "grad_norm": 0.6519432518685719, "learning_rate": 0.001572, "loss": 5.3646, "step": 524 }, { "epoch": 0.00525, "grad_norm": 0.6991224311762642, "learning_rate": 0.001575, "loss": 5.3575, "step": 525 }, { "epoch": 0.00526, "grad_norm": 0.6202961503223249, "learning_rate": 0.0015780000000000002, "loss": 5.3306, "step": 526 }, { "epoch": 0.00527, "grad_norm": 0.5608544792658494, "learning_rate": 0.0015810000000000002, "loss": 5.3379, "step": 527 }, { "epoch": 0.00528, "grad_norm": 0.6177007267681738, "learning_rate": 0.0015840000000000001, "loss": 5.3206, "step": 528 }, { "epoch": 0.00529, "grad_norm": 0.48495492558902425, "learning_rate": 0.001587, "loss": 5.3246, "step": 529 }, { "epoch": 0.0053, "grad_norm": 0.4709323576989762, "learning_rate": 0.00159, "loss": 5.3304, "step": 530 }, { "epoch": 0.00531, "grad_norm": 0.529532719880688, "learning_rate": 0.001593, "loss": 5.3166, "step": 531 }, { "epoch": 0.00532, "grad_norm": 0.5098383277074169, "learning_rate": 0.0015960000000000002, "loss": 5.3216, "step": 532 }, { "epoch": 0.00533, "grad_norm": 0.4454381993762929, "learning_rate": 0.0015990000000000002, "loss": 5.3134, "step": 533 }, { "epoch": 0.00534, "grad_norm": 0.4252458077548216, "learning_rate": 0.0016020000000000001, "loss": 5.3027, "step": 534 }, { "epoch": 0.00535, "grad_norm": 0.5072171102652949, "learning_rate": 0.001605, "loss": 5.3077, "step": 535 }, { "epoch": 0.00536, "grad_norm": 0.6459822593413646, "learning_rate": 0.001608, "loss": 5.3107, "step": 536 }, { "epoch": 0.00537, "grad_norm": 0.7666830941778624, "learning_rate": 0.0016110000000000002, "loss": 5.31, "step": 537 }, { "epoch": 0.00538, "grad_norm": 0.8221507236048067, "learning_rate": 0.0016140000000000002, "loss": 5.2923, "step": 538 }, { "epoch": 0.00539, "grad_norm": 0.831779757381357, "learning_rate": 0.0016170000000000002, "loss": 5.2913, "step": 539 }, { "epoch": 0.0054, "grad_norm": 0.7856275345794541, "learning_rate": 0.0016200000000000001, "loss": 5.3016, "step": 540 }, { "epoch": 0.00541, "grad_norm": 0.6804075992032543, "learning_rate": 0.001623, "loss": 5.2858, "step": 541 }, { "epoch": 0.00542, "grad_norm": 0.7690794494453981, "learning_rate": 0.001626, "loss": 5.3108, "step": 542 }, { "epoch": 0.00543, "grad_norm": 0.8644486526927273, "learning_rate": 0.0016290000000000002, "loss": 5.2796, "step": 543 }, { "epoch": 0.00544, "grad_norm": 0.8496144379530629, "learning_rate": 0.0016320000000000002, "loss": 5.3021, "step": 544 }, { "epoch": 0.00545, "grad_norm": 0.9475980596452462, "learning_rate": 0.0016350000000000002, "loss": 5.3177, "step": 545 }, { "epoch": 0.00546, "grad_norm": 0.9845718031318067, "learning_rate": 0.0016380000000000001, "loss": 5.295, "step": 546 }, { "epoch": 0.00547, "grad_norm": 0.9409915747667029, "learning_rate": 0.001641, "loss": 5.2939, "step": 547 }, { "epoch": 0.00548, "grad_norm": 0.9843592942492043, "learning_rate": 0.001644, "loss": 5.2968, "step": 548 }, { "epoch": 0.00549, "grad_norm": 0.9493094260301282, "learning_rate": 0.0016470000000000002, "loss": 5.2934, "step": 549 }, { "epoch": 0.0055, "grad_norm": 0.8611361277859524, "learning_rate": 0.0016500000000000002, "loss": 5.3029, "step": 550 }, { "epoch": 0.00551, "grad_norm": 0.9715881441414039, "learning_rate": 0.0016530000000000002, "loss": 5.3065, "step": 551 }, { "epoch": 0.00552, "grad_norm": 1.1094992375785262, "learning_rate": 0.0016560000000000001, "loss": 5.3056, "step": 552 }, { "epoch": 0.00553, "grad_norm": 1.0534001152383816, "learning_rate": 0.001659, "loss": 5.2999, "step": 553 }, { "epoch": 0.00554, "grad_norm": 0.9710868679896414, "learning_rate": 0.0016620000000000003, "loss": 5.3023, "step": 554 }, { "epoch": 0.00555, "grad_norm": 0.7839221021589211, "learning_rate": 0.0016650000000000002, "loss": 5.2983, "step": 555 }, { "epoch": 0.00556, "grad_norm": 0.7751353928928436, "learning_rate": 0.0016680000000000002, "loss": 5.2838, "step": 556 }, { "epoch": 0.00557, "grad_norm": 0.976465943327027, "learning_rate": 0.0016710000000000002, "loss": 5.2743, "step": 557 }, { "epoch": 0.00558, "grad_norm": 1.1542789282902877, "learning_rate": 0.0016740000000000001, "loss": 5.3159, "step": 558 }, { "epoch": 0.00559, "grad_norm": 0.925460422693844, "learning_rate": 0.001677, "loss": 5.2857, "step": 559 }, { "epoch": 0.0056, "grad_norm": 0.9401693950640773, "learning_rate": 0.0016800000000000003, "loss": 5.2634, "step": 560 }, { "epoch": 0.00561, "grad_norm": 0.9568676376365609, "learning_rate": 0.0016830000000000003, "loss": 5.2875, "step": 561 }, { "epoch": 0.00562, "grad_norm": 1.052222958057522, "learning_rate": 0.0016860000000000002, "loss": 5.2885, "step": 562 }, { "epoch": 0.00563, "grad_norm": 0.8981786385090996, "learning_rate": 0.001689, "loss": 5.2804, "step": 563 }, { "epoch": 0.00564, "grad_norm": 0.8729925726037424, "learning_rate": 0.001692, "loss": 5.2632, "step": 564 }, { "epoch": 0.00565, "grad_norm": 0.7853718417215511, "learning_rate": 0.001695, "loss": 5.2593, "step": 565 }, { "epoch": 0.00566, "grad_norm": 0.7408826324772542, "learning_rate": 0.0016979999999999999, "loss": 5.253, "step": 566 }, { "epoch": 0.00567, "grad_norm": 0.6229776590552076, "learning_rate": 0.0017009999999999998, "loss": 5.2429, "step": 567 }, { "epoch": 0.00568, "grad_norm": 0.5598722343375323, "learning_rate": 0.0017039999999999998, "loss": 5.2516, "step": 568 }, { "epoch": 0.00569, "grad_norm": 0.5222238686823768, "learning_rate": 0.001707, "loss": 5.2269, "step": 569 }, { "epoch": 0.0057, "grad_norm": 0.4848165136559503, "learning_rate": 0.00171, "loss": 5.2449, "step": 570 }, { "epoch": 0.00571, "grad_norm": 0.4212619410786628, "learning_rate": 0.001713, "loss": 5.2393, "step": 571 }, { "epoch": 0.00572, "grad_norm": 0.46300721556593205, "learning_rate": 0.0017159999999999999, "loss": 5.2148, "step": 572 }, { "epoch": 0.00573, "grad_norm": 0.46690826663873664, "learning_rate": 0.0017189999999999998, "loss": 5.2177, "step": 573 }, { "epoch": 0.00574, "grad_norm": 0.5544353822714976, "learning_rate": 0.001722, "loss": 5.2291, "step": 574 }, { "epoch": 0.00575, "grad_norm": 0.5414515614228337, "learning_rate": 0.001725, "loss": 5.2052, "step": 575 }, { "epoch": 0.00576, "grad_norm": 0.460862784072886, "learning_rate": 0.001728, "loss": 5.1957, "step": 576 }, { "epoch": 0.00577, "grad_norm": 0.37283009231118824, "learning_rate": 0.001731, "loss": 5.2169, "step": 577 }, { "epoch": 0.00578, "grad_norm": 0.394929191519935, "learning_rate": 0.0017339999999999999, "loss": 5.1956, "step": 578 }, { "epoch": 0.00579, "grad_norm": 0.46312560639878303, "learning_rate": 0.0017369999999999998, "loss": 5.1828, "step": 579 }, { "epoch": 0.0058, "grad_norm": 0.5018384741903004, "learning_rate": 0.00174, "loss": 5.1996, "step": 580 }, { "epoch": 0.00581, "grad_norm": 0.607886610189308, "learning_rate": 0.001743, "loss": 5.2004, "step": 581 }, { "epoch": 0.00582, "grad_norm": 0.6369652189686417, "learning_rate": 0.001746, "loss": 5.1762, "step": 582 }, { "epoch": 0.00583, "grad_norm": 0.6307825776815035, "learning_rate": 0.001749, "loss": 5.1999, "step": 583 }, { "epoch": 0.00584, "grad_norm": 0.6660509674445032, "learning_rate": 0.0017519999999999999, "loss": 5.1973, "step": 584 }, { "epoch": 0.00585, "grad_norm": 0.6419220280838183, "learning_rate": 0.0017549999999999998, "loss": 5.1782, "step": 585 }, { "epoch": 0.00586, "grad_norm": 0.6225781818783162, "learning_rate": 0.001758, "loss": 5.1816, "step": 586 }, { "epoch": 0.00587, "grad_norm": 0.7071422118089229, "learning_rate": 0.001761, "loss": 5.1689, "step": 587 }, { "epoch": 0.00588, "grad_norm": 0.7860915960717592, "learning_rate": 0.001764, "loss": 5.1776, "step": 588 }, { "epoch": 0.00589, "grad_norm": 0.7980122505569107, "learning_rate": 0.001767, "loss": 5.1811, "step": 589 }, { "epoch": 0.0059, "grad_norm": 0.7215274327961783, "learning_rate": 0.0017699999999999999, "loss": 5.1658, "step": 590 }, { "epoch": 0.00591, "grad_norm": 0.8634769330088186, "learning_rate": 0.001773, "loss": 5.1655, "step": 591 }, { "epoch": 0.00592, "grad_norm": 0.997893173060324, "learning_rate": 0.001776, "loss": 5.2022, "step": 592 }, { "epoch": 0.00593, "grad_norm": 1.03294909692111, "learning_rate": 0.001779, "loss": 5.2054, "step": 593 }, { "epoch": 0.00594, "grad_norm": 1.1929198243438337, "learning_rate": 0.001782, "loss": 5.2059, "step": 594 }, { "epoch": 0.00595, "grad_norm": 0.9565031631375653, "learning_rate": 0.001785, "loss": 5.1996, "step": 595 }, { "epoch": 0.00596, "grad_norm": 0.7489131564683893, "learning_rate": 0.0017879999999999999, "loss": 5.1856, "step": 596 }, { "epoch": 0.00597, "grad_norm": 0.8084904095480396, "learning_rate": 0.001791, "loss": 5.1654, "step": 597 }, { "epoch": 0.00598, "grad_norm": 0.8735012375123758, "learning_rate": 0.001794, "loss": 5.1801, "step": 598 }, { "epoch": 0.00599, "grad_norm": 0.8866107272305487, "learning_rate": 0.001797, "loss": 5.1794, "step": 599 }, { "epoch": 0.006, "grad_norm": 0.9944298023441023, "learning_rate": 0.0018, "loss": 5.186, "step": 600 }, { "epoch": 0.00601, "grad_norm": 1.2284675534425753, "learning_rate": 0.001803, "loss": 5.1964, "step": 601 }, { "epoch": 0.00602, "grad_norm": 0.9174538372133724, "learning_rate": 0.0018059999999999999, "loss": 5.167, "step": 602 }, { "epoch": 0.00603, "grad_norm": 1.0377055005145264, "learning_rate": 0.001809, "loss": 5.199, "step": 603 }, { "epoch": 0.00604, "grad_norm": 0.914710796715033, "learning_rate": 0.001812, "loss": 5.1662, "step": 604 }, { "epoch": 0.00605, "grad_norm": 1.2614192403498132, "learning_rate": 0.001815, "loss": 5.1973, "step": 605 }, { "epoch": 0.00606, "grad_norm": 0.7058550445524863, "learning_rate": 0.001818, "loss": 5.1744, "step": 606 }, { "epoch": 0.00607, "grad_norm": 0.7505416501870767, "learning_rate": 0.001821, "loss": 5.1771, "step": 607 }, { "epoch": 0.00608, "grad_norm": 0.8233533261017171, "learning_rate": 0.001824, "loss": 5.1679, "step": 608 }, { "epoch": 0.00609, "grad_norm": 0.7138670005342819, "learning_rate": 0.001827, "loss": 5.1654, "step": 609 }, { "epoch": 0.0061, "grad_norm": 0.6840037013923436, "learning_rate": 0.00183, "loss": 5.1301, "step": 610 }, { "epoch": 0.00611, "grad_norm": 0.9092201750767857, "learning_rate": 0.001833, "loss": 5.1441, "step": 611 }, { "epoch": 0.00612, "grad_norm": 0.9586638015997742, "learning_rate": 0.001836, "loss": 5.1502, "step": 612 }, { "epoch": 0.00613, "grad_norm": 0.9726859873682268, "learning_rate": 0.001839, "loss": 5.1479, "step": 613 }, { "epoch": 0.00614, "grad_norm": 0.8517287134785719, "learning_rate": 0.001842, "loss": 5.1632, "step": 614 }, { "epoch": 0.00615, "grad_norm": 0.808391187458492, "learning_rate": 0.001845, "loss": 5.1468, "step": 615 }, { "epoch": 0.00616, "grad_norm": 0.9667719890903622, "learning_rate": 0.001848, "loss": 5.139, "step": 616 }, { "epoch": 0.00617, "grad_norm": 0.9287751010032772, "learning_rate": 0.001851, "loss": 5.1507, "step": 617 }, { "epoch": 0.00618, "grad_norm": 0.8398949490968032, "learning_rate": 0.001854, "loss": 5.1338, "step": 618 }, { "epoch": 0.00619, "grad_norm": 0.8191529309136791, "learning_rate": 0.001857, "loss": 5.1377, "step": 619 }, { "epoch": 0.0062, "grad_norm": 0.6738755871000573, "learning_rate": 0.00186, "loss": 5.1151, "step": 620 }, { "epoch": 0.00621, "grad_norm": 0.542431752937381, "learning_rate": 0.001863, "loss": 5.1102, "step": 621 }, { "epoch": 0.00622, "grad_norm": 0.5826015665301125, "learning_rate": 0.001866, "loss": 5.1076, "step": 622 }, { "epoch": 0.00623, "grad_norm": 0.6277970980683506, "learning_rate": 0.001869, "loss": 5.0901, "step": 623 }, { "epoch": 0.00624, "grad_norm": 0.6204003612309649, "learning_rate": 0.001872, "loss": 5.0981, "step": 624 }, { "epoch": 0.00625, "grad_norm": 0.5202043778302353, "learning_rate": 0.001875, "loss": 5.1041, "step": 625 }, { "epoch": 0.00626, "grad_norm": 0.4583601362158537, "learning_rate": 0.0018780000000000001, "loss": 5.0782, "step": 626 }, { "epoch": 0.00627, "grad_norm": 0.49456742347801874, "learning_rate": 0.001881, "loss": 5.0657, "step": 627 }, { "epoch": 0.00628, "grad_norm": 0.4979469593628374, "learning_rate": 0.001884, "loss": 5.0939, "step": 628 }, { "epoch": 0.00629, "grad_norm": 0.4931618065441273, "learning_rate": 0.001887, "loss": 5.0612, "step": 629 }, { "epoch": 0.0063, "grad_norm": 0.5205717896967359, "learning_rate": 0.00189, "loss": 5.0588, "step": 630 }, { "epoch": 0.00631, "grad_norm": 0.46185074539676735, "learning_rate": 0.0018930000000000002, "loss": 5.0466, "step": 631 }, { "epoch": 0.00632, "grad_norm": 0.5076264188019036, "learning_rate": 0.0018960000000000001, "loss": 5.0644, "step": 632 }, { "epoch": 0.00633, "grad_norm": 0.5650163435420743, "learning_rate": 0.001899, "loss": 5.0598, "step": 633 }, { "epoch": 0.00634, "grad_norm": 0.6177308428090045, "learning_rate": 0.001902, "loss": 5.0443, "step": 634 }, { "epoch": 0.00635, "grad_norm": 0.5966699478227224, "learning_rate": 0.001905, "loss": 5.0351, "step": 635 }, { "epoch": 0.00636, "grad_norm": 0.46738522618380884, "learning_rate": 0.001908, "loss": 5.0351, "step": 636 }, { "epoch": 0.00637, "grad_norm": 0.5725125755370889, "learning_rate": 0.0019110000000000002, "loss": 5.0522, "step": 637 }, { "epoch": 0.00638, "grad_norm": 0.6188024234981583, "learning_rate": 0.0019140000000000001, "loss": 5.0398, "step": 638 }, { "epoch": 0.00639, "grad_norm": 0.612857971520305, "learning_rate": 0.001917, "loss": 5.0146, "step": 639 }, { "epoch": 0.0064, "grad_norm": 0.7145019607417652, "learning_rate": 0.00192, "loss": 5.021, "step": 640 }, { "epoch": 0.00641, "grad_norm": 0.8625442633088365, "learning_rate": 0.001923, "loss": 5.0332, "step": 641 }, { "epoch": 0.00642, "grad_norm": 0.9132924514062416, "learning_rate": 0.001926, "loss": 5.0263, "step": 642 }, { "epoch": 0.00643, "grad_norm": 0.9679240940884513, "learning_rate": 0.0019290000000000002, "loss": 5.0516, "step": 643 }, { "epoch": 0.00644, "grad_norm": 0.9705199088056594, "learning_rate": 0.0019320000000000001, "loss": 5.0234, "step": 644 }, { "epoch": 0.00645, "grad_norm": 0.8188630256116934, "learning_rate": 0.001935, "loss": 5.0249, "step": 645 }, { "epoch": 0.00646, "grad_norm": 0.8158823988142996, "learning_rate": 0.001938, "loss": 5.0348, "step": 646 }, { "epoch": 0.00647, "grad_norm": 0.7787935632242667, "learning_rate": 0.001941, "loss": 5.037, "step": 647 }, { "epoch": 0.00648, "grad_norm": 0.9237403874154956, "learning_rate": 0.0019440000000000002, "loss": 5.0316, "step": 648 }, { "epoch": 0.00649, "grad_norm": 1.036550642596614, "learning_rate": 0.0019470000000000002, "loss": 5.0289, "step": 649 }, { "epoch": 0.0065, "grad_norm": 0.8648530228627821, "learning_rate": 0.0019500000000000001, "loss": 5.0479, "step": 650 }, { "epoch": 0.00651, "grad_norm": 0.984999530178637, "learning_rate": 0.001953, "loss": 5.0283, "step": 651 }, { "epoch": 0.00652, "grad_norm": 1.1933341828071018, "learning_rate": 0.0019560000000000003, "loss": 5.0354, "step": 652 }, { "epoch": 0.00653, "grad_norm": 0.8822963296035995, "learning_rate": 0.0019590000000000002, "loss": 5.0195, "step": 653 }, { "epoch": 0.00654, "grad_norm": 0.9601302022272827, "learning_rate": 0.001962, "loss": 5.0019, "step": 654 }, { "epoch": 0.00655, "grad_norm": 1.1161400796235366, "learning_rate": 0.001965, "loss": 5.0334, "step": 655 }, { "epoch": 0.00656, "grad_norm": 1.0829794237799832, "learning_rate": 0.001968, "loss": 5.0307, "step": 656 }, { "epoch": 0.00657, "grad_norm": 0.9921530841514947, "learning_rate": 0.001971, "loss": 5.0086, "step": 657 }, { "epoch": 0.00658, "grad_norm": 1.028510611823797, "learning_rate": 0.001974, "loss": 5.0305, "step": 658 }, { "epoch": 0.00659, "grad_norm": 1.0763635896939565, "learning_rate": 0.001977, "loss": 5.0177, "step": 659 }, { "epoch": 0.0066, "grad_norm": 1.3314133581668428, "learning_rate": 0.00198, "loss": 5.0358, "step": 660 }, { "epoch": 0.00661, "grad_norm": 0.8366683487964699, "learning_rate": 0.001983, "loss": 4.987, "step": 661 }, { "epoch": 0.00662, "grad_norm": 0.9690057003124267, "learning_rate": 0.0019860000000000004, "loss": 4.9897, "step": 662 }, { "epoch": 0.00663, "grad_norm": 1.1210889842101375, "learning_rate": 0.0019890000000000003, "loss": 5.0167, "step": 663 }, { "epoch": 0.00664, "grad_norm": 0.9227592782421535, "learning_rate": 0.0019920000000000003, "loss": 4.9926, "step": 664 }, { "epoch": 0.00665, "grad_norm": 0.7430643643535455, "learning_rate": 0.0019950000000000002, "loss": 4.9603, "step": 665 }, { "epoch": 0.00666, "grad_norm": 0.730237759629391, "learning_rate": 0.001998, "loss": 4.9749, "step": 666 }, { "epoch": 0.00667, "grad_norm": 0.7298395123979716, "learning_rate": 0.002001, "loss": 4.9682, "step": 667 }, { "epoch": 0.00668, "grad_norm": 0.7359965443598192, "learning_rate": 0.002004, "loss": 4.9858, "step": 668 }, { "epoch": 0.00669, "grad_norm": 0.803685757229409, "learning_rate": 0.002007, "loss": 4.9957, "step": 669 }, { "epoch": 0.0067, "grad_norm": 0.7455537997836268, "learning_rate": 0.00201, "loss": 4.9782, "step": 670 }, { "epoch": 0.00671, "grad_norm": 0.7155119749171314, "learning_rate": 0.002013, "loss": 4.9551, "step": 671 }, { "epoch": 0.00672, "grad_norm": 0.7618179803756451, "learning_rate": 0.002016, "loss": 4.9472, "step": 672 }, { "epoch": 0.00673, "grad_norm": 0.927797245884825, "learning_rate": 0.002019, "loss": 4.953, "step": 673 }, { "epoch": 0.00674, "grad_norm": 1.0633432044595463, "learning_rate": 0.0020220000000000004, "loss": 4.9634, "step": 674 }, { "epoch": 0.00675, "grad_norm": 0.8942052250721098, "learning_rate": 0.0020250000000000003, "loss": 4.9712, "step": 675 }, { "epoch": 0.00676, "grad_norm": 0.8664052109661106, "learning_rate": 0.0020280000000000003, "loss": 4.9561, "step": 676 }, { "epoch": 0.00677, "grad_norm": 0.7768891936596246, "learning_rate": 0.0020310000000000003, "loss": 4.9501, "step": 677 }, { "epoch": 0.00678, "grad_norm": 0.8508591245716224, "learning_rate": 0.0020340000000000002, "loss": 4.9609, "step": 678 }, { "epoch": 0.00679, "grad_norm": 0.899178484097066, "learning_rate": 0.002037, "loss": 4.9511, "step": 679 }, { "epoch": 0.0068, "grad_norm": 0.9041496786679454, "learning_rate": 0.00204, "loss": 4.9463, "step": 680 }, { "epoch": 0.00681, "grad_norm": 1.1323904416700994, "learning_rate": 0.002043, "loss": 4.9558, "step": 681 }, { "epoch": 0.00682, "grad_norm": 0.9111266905834888, "learning_rate": 0.002046, "loss": 4.9418, "step": 682 }, { "epoch": 0.00683, "grad_norm": 0.8925821061333421, "learning_rate": 0.002049, "loss": 4.9271, "step": 683 }, { "epoch": 0.00684, "grad_norm": 0.9085464749855919, "learning_rate": 0.002052, "loss": 4.9489, "step": 684 }, { "epoch": 0.00685, "grad_norm": 0.7656795310652994, "learning_rate": 0.0020550000000000004, "loss": 4.9405, "step": 685 }, { "epoch": 0.00686, "grad_norm": 0.7811896035655447, "learning_rate": 0.0020580000000000004, "loss": 4.9138, "step": 686 }, { "epoch": 0.00687, "grad_norm": 0.6835895457049103, "learning_rate": 0.0020610000000000003, "loss": 4.9092, "step": 687 }, { "epoch": 0.00688, "grad_norm": 0.7478929092218646, "learning_rate": 0.002064, "loss": 4.9101, "step": 688 }, { "epoch": 0.00689, "grad_norm": 0.722678645543914, "learning_rate": 0.002067, "loss": 4.9105, "step": 689 }, { "epoch": 0.0069, "grad_norm": 0.777097923162003, "learning_rate": 0.00207, "loss": 4.9057, "step": 690 }, { "epoch": 0.00691, "grad_norm": 0.8568701965666653, "learning_rate": 0.0020729999999999998, "loss": 4.8774, "step": 691 }, { "epoch": 0.00692, "grad_norm": 0.7917857300738825, "learning_rate": 0.0020759999999999997, "loss": 4.8965, "step": 692 }, { "epoch": 0.00693, "grad_norm": 0.6390484895042627, "learning_rate": 0.0020789999999999997, "loss": 4.8963, "step": 693 }, { "epoch": 0.00694, "grad_norm": 0.5563965742668177, "learning_rate": 0.002082, "loss": 4.8677, "step": 694 }, { "epoch": 0.00695, "grad_norm": 0.6126256581377956, "learning_rate": 0.002085, "loss": 4.868, "step": 695 }, { "epoch": 0.00696, "grad_norm": 0.603510667767381, "learning_rate": 0.002088, "loss": 4.8758, "step": 696 }, { "epoch": 0.00697, "grad_norm": 0.6639527590552766, "learning_rate": 0.002091, "loss": 4.8784, "step": 697 }, { "epoch": 0.00698, "grad_norm": 0.8067091784202179, "learning_rate": 0.002094, "loss": 4.8757, "step": 698 }, { "epoch": 0.00699, "grad_norm": 0.9128354608980798, "learning_rate": 0.002097, "loss": 4.8621, "step": 699 }, { "epoch": 0.007, "grad_norm": 0.9298486189112464, "learning_rate": 0.0021, "loss": 4.8859, "step": 700 }, { "epoch": 0.00701, "grad_norm": 0.9079094485202986, "learning_rate": 0.002103, "loss": 4.8619, "step": 701 }, { "epoch": 0.00702, "grad_norm": 1.1115710589227372, "learning_rate": 0.002106, "loss": 4.8769, "step": 702 }, { "epoch": 0.00703, "grad_norm": 0.8563727136979473, "learning_rate": 0.0021089999999999998, "loss": 4.846, "step": 703 }, { "epoch": 0.00704, "grad_norm": 0.8507471900724869, "learning_rate": 0.0021119999999999997, "loss": 4.8622, "step": 704 }, { "epoch": 0.00705, "grad_norm": 0.769782464639154, "learning_rate": 0.002115, "loss": 4.8683, "step": 705 }, { "epoch": 0.00706, "grad_norm": 0.823170135002925, "learning_rate": 0.002118, "loss": 4.8626, "step": 706 }, { "epoch": 0.00707, "grad_norm": 0.8501476997865872, "learning_rate": 0.002121, "loss": 4.8385, "step": 707 }, { "epoch": 0.00708, "grad_norm": 0.9757796217120096, "learning_rate": 0.002124, "loss": 4.8255, "step": 708 }, { "epoch": 0.00709, "grad_norm": 1.136516664517519, "learning_rate": 0.002127, "loss": 4.875, "step": 709 }, { "epoch": 0.0071, "grad_norm": 0.8442578613736921, "learning_rate": 0.00213, "loss": 4.8495, "step": 710 }, { "epoch": 0.00711, "grad_norm": 0.8659317374368097, "learning_rate": 0.002133, "loss": 4.8259, "step": 711 }, { "epoch": 0.00712, "grad_norm": 0.8943878610557214, "learning_rate": 0.002136, "loss": 4.8499, "step": 712 }, { "epoch": 0.00713, "grad_norm": 1.005520267323647, "learning_rate": 0.002139, "loss": 4.8703, "step": 713 }, { "epoch": 0.00714, "grad_norm": 0.8575705768347217, "learning_rate": 0.002142, "loss": 4.8532, "step": 714 }, { "epoch": 0.00715, "grad_norm": 0.6910728311533939, "learning_rate": 0.0021449999999999998, "loss": 4.8149, "step": 715 }, { "epoch": 0.00716, "grad_norm": 0.523242752687879, "learning_rate": 0.002148, "loss": 4.8032, "step": 716 }, { "epoch": 0.00717, "grad_norm": 0.6187034661565012, "learning_rate": 0.002151, "loss": 4.8222, "step": 717 }, { "epoch": 0.00718, "grad_norm": 0.6631189634234681, "learning_rate": 0.002154, "loss": 4.8103, "step": 718 }, { "epoch": 0.00719, "grad_norm": 0.8615055808443407, "learning_rate": 0.002157, "loss": 4.8254, "step": 719 }, { "epoch": 0.0072, "grad_norm": 0.9623520860140555, "learning_rate": 0.00216, "loss": 4.8223, "step": 720 }, { "epoch": 0.00721, "grad_norm": 0.8352385840849053, "learning_rate": 0.002163, "loss": 4.8225, "step": 721 }, { "epoch": 0.00722, "grad_norm": 0.7006874812702103, "learning_rate": 0.002166, "loss": 4.7938, "step": 722 }, { "epoch": 0.00723, "grad_norm": 0.6468265093949516, "learning_rate": 0.002169, "loss": 4.8026, "step": 723 }, { "epoch": 0.00724, "grad_norm": 0.6351725533732875, "learning_rate": 0.002172, "loss": 4.7959, "step": 724 }, { "epoch": 0.00725, "grad_norm": 0.6951521484434561, "learning_rate": 0.002175, "loss": 4.7939, "step": 725 }, { "epoch": 0.00726, "grad_norm": 0.6685844017951591, "learning_rate": 0.002178, "loss": 4.8021, "step": 726 }, { "epoch": 0.00727, "grad_norm": 0.6591838242250027, "learning_rate": 0.0021809999999999998, "loss": 4.7801, "step": 727 }, { "epoch": 0.00728, "grad_norm": 0.6854740798872353, "learning_rate": 0.002184, "loss": 4.7935, "step": 728 }, { "epoch": 0.00729, "grad_norm": 0.8458670368113279, "learning_rate": 0.002187, "loss": 4.7998, "step": 729 }, { "epoch": 0.0073, "grad_norm": 0.8820596950270692, "learning_rate": 0.00219, "loss": 4.7957, "step": 730 }, { "epoch": 0.00731, "grad_norm": 0.7920873615429597, "learning_rate": 0.002193, "loss": 4.787, "step": 731 }, { "epoch": 0.00732, "grad_norm": 0.76241161465089, "learning_rate": 0.002196, "loss": 4.7953, "step": 732 }, { "epoch": 0.00733, "grad_norm": 0.7536070296480728, "learning_rate": 0.002199, "loss": 4.7713, "step": 733 }, { "epoch": 0.00734, "grad_norm": 0.7828888260304195, "learning_rate": 0.002202, "loss": 4.8005, "step": 734 }, { "epoch": 0.00735, "grad_norm": 0.789301121324275, "learning_rate": 0.002205, "loss": 4.8061, "step": 735 }, { "epoch": 0.00736, "grad_norm": 0.8665427368268374, "learning_rate": 0.002208, "loss": 4.7947, "step": 736 }, { "epoch": 0.00737, "grad_norm": 0.9213665258793059, "learning_rate": 0.002211, "loss": 4.7668, "step": 737 }, { "epoch": 0.00738, "grad_norm": 1.1051560061666965, "learning_rate": 0.002214, "loss": 4.7801, "step": 738 }, { "epoch": 0.00739, "grad_norm": 1.2853173109817828, "learning_rate": 0.0022170000000000002, "loss": 4.7871, "step": 739 }, { "epoch": 0.0074, "grad_norm": 0.8846868930049064, "learning_rate": 0.00222, "loss": 4.7826, "step": 740 }, { "epoch": 0.00741, "grad_norm": 0.9670459302610688, "learning_rate": 0.002223, "loss": 4.7564, "step": 741 }, { "epoch": 0.00742, "grad_norm": 1.0093936436375988, "learning_rate": 0.002226, "loss": 4.8054, "step": 742 }, { "epoch": 0.00743, "grad_norm": 0.9547940794526594, "learning_rate": 0.002229, "loss": 4.7628, "step": 743 }, { "epoch": 0.00744, "grad_norm": 1.0433514490805686, "learning_rate": 0.002232, "loss": 4.781, "step": 744 }, { "epoch": 0.00745, "grad_norm": 0.9325385949963747, "learning_rate": 0.002235, "loss": 4.7894, "step": 745 }, { "epoch": 0.00746, "grad_norm": 0.9019275826038629, "learning_rate": 0.002238, "loss": 4.7783, "step": 746 }, { "epoch": 0.00747, "grad_norm": 0.7966099158360889, "learning_rate": 0.002241, "loss": 4.7608, "step": 747 }, { "epoch": 0.00748, "grad_norm": 0.8565839398988285, "learning_rate": 0.002244, "loss": 4.7749, "step": 748 }, { "epoch": 0.00749, "grad_norm": 0.816231494499623, "learning_rate": 0.002247, "loss": 4.7509, "step": 749 }, { "epoch": 0.0075, "grad_norm": 0.7743660614307156, "learning_rate": 0.0022500000000000003, "loss": 4.7909, "step": 750 }, { "epoch": 0.00751, "grad_norm": 0.8113380512241561, "learning_rate": 0.0022530000000000002, "loss": 4.7775, "step": 751 }, { "epoch": 0.00752, "grad_norm": 0.9985073455505076, "learning_rate": 0.002256, "loss": 4.7451, "step": 752 }, { "epoch": 0.00753, "grad_norm": 0.9478375008624312, "learning_rate": 0.002259, "loss": 4.7859, "step": 753 }, { "epoch": 0.00754, "grad_norm": 0.7802553547569654, "learning_rate": 0.002262, "loss": 4.7862, "step": 754 }, { "epoch": 0.00755, "grad_norm": 0.9025323762262523, "learning_rate": 0.002265, "loss": 4.7581, "step": 755 }, { "epoch": 0.00756, "grad_norm": 0.8357144009512405, "learning_rate": 0.002268, "loss": 4.7438, "step": 756 }, { "epoch": 0.00757, "grad_norm": 0.8081799661033972, "learning_rate": 0.002271, "loss": 4.7545, "step": 757 }, { "epoch": 0.00758, "grad_norm": 0.7188920586125809, "learning_rate": 0.002274, "loss": 4.7374, "step": 758 }, { "epoch": 0.00759, "grad_norm": 0.6594203091253575, "learning_rate": 0.002277, "loss": 4.7482, "step": 759 }, { "epoch": 0.0076, "grad_norm": 0.5807043361033956, "learning_rate": 0.00228, "loss": 4.7315, "step": 760 }, { "epoch": 0.00761, "grad_norm": 0.5117441659228741, "learning_rate": 0.002283, "loss": 4.7, "step": 761 }, { "epoch": 0.00762, "grad_norm": 0.5351172229626283, "learning_rate": 0.0022860000000000003, "loss": 4.6819, "step": 762 }, { "epoch": 0.00763, "grad_norm": 0.47299369798782465, "learning_rate": 0.0022890000000000002, "loss": 4.7227, "step": 763 }, { "epoch": 0.00764, "grad_norm": 0.4213562585185439, "learning_rate": 0.002292, "loss": 4.7275, "step": 764 }, { "epoch": 0.00765, "grad_norm": 0.44406465565712894, "learning_rate": 0.002295, "loss": 4.7138, "step": 765 }, { "epoch": 0.00766, "grad_norm": 0.5032974849101272, "learning_rate": 0.002298, "loss": 4.7114, "step": 766 }, { "epoch": 0.00767, "grad_norm": 0.7140868069096875, "learning_rate": 0.002301, "loss": 4.7151, "step": 767 }, { "epoch": 0.00768, "grad_norm": 0.8614017286346677, "learning_rate": 0.002304, "loss": 4.7339, "step": 768 }, { "epoch": 0.00769, "grad_norm": 0.811936140634688, "learning_rate": 0.002307, "loss": 4.7346, "step": 769 }, { "epoch": 0.0077, "grad_norm": 0.6775529648178301, "learning_rate": 0.00231, "loss": 4.6843, "step": 770 }, { "epoch": 0.00771, "grad_norm": 0.7223031465513833, "learning_rate": 0.002313, "loss": 4.7255, "step": 771 }, { "epoch": 0.00772, "grad_norm": 0.6676019393281739, "learning_rate": 0.002316, "loss": 4.7053, "step": 772 }, { "epoch": 0.00773, "grad_norm": 0.6378542516153257, "learning_rate": 0.0023190000000000003, "loss": 4.7069, "step": 773 }, { "epoch": 0.00774, "grad_norm": 0.6207815175915733, "learning_rate": 0.0023220000000000003, "loss": 4.7187, "step": 774 }, { "epoch": 0.00775, "grad_norm": 0.5964347529354114, "learning_rate": 0.0023250000000000002, "loss": 4.7097, "step": 775 }, { "epoch": 0.00776, "grad_norm": 0.630013209430953, "learning_rate": 0.002328, "loss": 4.7096, "step": 776 }, { "epoch": 0.00777, "grad_norm": 0.6207512829319234, "learning_rate": 0.002331, "loss": 4.6848, "step": 777 }, { "epoch": 0.00778, "grad_norm": 0.6524761943334527, "learning_rate": 0.002334, "loss": 4.6998, "step": 778 }, { "epoch": 0.00779, "grad_norm": 0.8187914523175398, "learning_rate": 0.002337, "loss": 4.6652, "step": 779 }, { "epoch": 0.0078, "grad_norm": 0.9648917298565269, "learning_rate": 0.00234, "loss": 4.7012, "step": 780 }, { "epoch": 0.00781, "grad_norm": 1.0384259728763472, "learning_rate": 0.002343, "loss": 4.7097, "step": 781 }, { "epoch": 0.00782, "grad_norm": 0.9635439545076986, "learning_rate": 0.002346, "loss": 4.7024, "step": 782 }, { "epoch": 0.00783, "grad_norm": 0.7812806082921958, "learning_rate": 0.002349, "loss": 4.6794, "step": 783 }, { "epoch": 0.00784, "grad_norm": 0.781961859641362, "learning_rate": 0.002352, "loss": 4.6914, "step": 784 }, { "epoch": 0.00785, "grad_norm": 0.7694622494278262, "learning_rate": 0.0023550000000000003, "loss": 4.6875, "step": 785 }, { "epoch": 0.00786, "grad_norm": 0.8298262678177186, "learning_rate": 0.0023580000000000003, "loss": 4.6951, "step": 786 }, { "epoch": 0.00787, "grad_norm": 0.880296894881306, "learning_rate": 0.0023610000000000003, "loss": 4.6973, "step": 787 }, { "epoch": 0.00788, "grad_norm": 0.9208137721706963, "learning_rate": 0.002364, "loss": 4.6994, "step": 788 }, { "epoch": 0.00789, "grad_norm": 0.8821666937927062, "learning_rate": 0.002367, "loss": 4.7255, "step": 789 }, { "epoch": 0.0079, "grad_norm": 0.9083723262351499, "learning_rate": 0.00237, "loss": 4.7016, "step": 790 }, { "epoch": 0.00791, "grad_norm": 1.102156246123534, "learning_rate": 0.002373, "loss": 4.6636, "step": 791 }, { "epoch": 0.00792, "grad_norm": 1.0575279803993471, "learning_rate": 0.002376, "loss": 4.7024, "step": 792 }, { "epoch": 0.00793, "grad_norm": 0.8169656544039797, "learning_rate": 0.002379, "loss": 4.6881, "step": 793 }, { "epoch": 0.00794, "grad_norm": 0.678074438498389, "learning_rate": 0.002382, "loss": 4.6904, "step": 794 }, { "epoch": 0.00795, "grad_norm": 0.6991656768310356, "learning_rate": 0.002385, "loss": 4.6673, "step": 795 }, { "epoch": 0.00796, "grad_norm": 0.7517528447717965, "learning_rate": 0.0023880000000000004, "loss": 4.6604, "step": 796 }, { "epoch": 0.00797, "grad_norm": 0.6539739996063073, "learning_rate": 0.0023910000000000003, "loss": 4.701, "step": 797 }, { "epoch": 0.00798, "grad_norm": 0.7347989701629836, "learning_rate": 0.0023940000000000003, "loss": 4.6689, "step": 798 }, { "epoch": 0.00799, "grad_norm": 0.9872840389626579, "learning_rate": 0.0023970000000000003, "loss": 4.7125, "step": 799 }, { "epoch": 0.008, "grad_norm": 1.5507273997188689, "learning_rate": 0.0024000000000000002, "loss": 4.7008, "step": 800 }, { "epoch": 0.00801, "grad_norm": 0.8492322576000088, "learning_rate": 0.002403, "loss": 4.7238, "step": 801 }, { "epoch": 0.00802, "grad_norm": 0.814504548640702, "learning_rate": 0.002406, "loss": 4.7004, "step": 802 }, { "epoch": 0.00803, "grad_norm": 1.1123606888778432, "learning_rate": 0.002409, "loss": 4.6958, "step": 803 }, { "epoch": 0.00804, "grad_norm": 0.9176660269114695, "learning_rate": 0.002412, "loss": 4.6917, "step": 804 }, { "epoch": 0.00805, "grad_norm": 0.8611830884359055, "learning_rate": 0.002415, "loss": 4.683, "step": 805 }, { "epoch": 0.00806, "grad_norm": 0.8560370675329328, "learning_rate": 0.002418, "loss": 4.6967, "step": 806 }, { "epoch": 0.00807, "grad_norm": 0.9747797626725606, "learning_rate": 0.0024210000000000004, "loss": 4.6971, "step": 807 }, { "epoch": 0.00808, "grad_norm": 0.8688248558202334, "learning_rate": 0.0024240000000000004, "loss": 4.6774, "step": 808 }, { "epoch": 0.00809, "grad_norm": 0.7810743476258258, "learning_rate": 0.0024270000000000003, "loss": 4.6681, "step": 809 }, { "epoch": 0.0081, "grad_norm": 0.8213343057667987, "learning_rate": 0.0024300000000000003, "loss": 4.7024, "step": 810 }, { "epoch": 0.00811, "grad_norm": 0.7477361334276762, "learning_rate": 0.0024330000000000003, "loss": 4.6714, "step": 811 }, { "epoch": 0.00812, "grad_norm": 0.7141317849387557, "learning_rate": 0.0024360000000000002, "loss": 4.6487, "step": 812 }, { "epoch": 0.00813, "grad_norm": 0.6767630610959015, "learning_rate": 0.0024389999999999998, "loss": 4.6789, "step": 813 }, { "epoch": 0.00814, "grad_norm": 0.6590399241347584, "learning_rate": 0.0024419999999999997, "loss": 4.6805, "step": 814 }, { "epoch": 0.00815, "grad_norm": 0.6023723300104491, "learning_rate": 0.0024449999999999997, "loss": 4.6845, "step": 815 }, { "epoch": 0.00816, "grad_norm": 0.6246267032157732, "learning_rate": 0.002448, "loss": 4.6577, "step": 816 }, { "epoch": 0.00817, "grad_norm": 0.7633099639321208, "learning_rate": 0.002451, "loss": 4.6687, "step": 817 }, { "epoch": 0.00818, "grad_norm": 0.7806805982678847, "learning_rate": 0.002454, "loss": 4.6553, "step": 818 }, { "epoch": 0.00819, "grad_norm": 0.5501560610295794, "learning_rate": 0.002457, "loss": 4.6552, "step": 819 }, { "epoch": 0.0082, "grad_norm": 0.5777814189038608, "learning_rate": 0.00246, "loss": 4.6416, "step": 820 }, { "epoch": 0.00821, "grad_norm": 0.5160511618881874, "learning_rate": 0.002463, "loss": 4.6164, "step": 821 }, { "epoch": 0.00822, "grad_norm": 0.43680612738745367, "learning_rate": 0.002466, "loss": 4.6137, "step": 822 }, { "epoch": 0.00823, "grad_norm": 0.4259879914150642, "learning_rate": 0.002469, "loss": 4.6172, "step": 823 }, { "epoch": 0.00824, "grad_norm": 0.4527560352782629, "learning_rate": 0.002472, "loss": 4.6356, "step": 824 }, { "epoch": 0.00825, "grad_norm": 0.43208606152268814, "learning_rate": 0.0024749999999999998, "loss": 4.6397, "step": 825 }, { "epoch": 0.00826, "grad_norm": 0.5843185047977487, "learning_rate": 0.0024779999999999997, "loss": 4.6038, "step": 826 }, { "epoch": 0.00827, "grad_norm": 0.8997674361314323, "learning_rate": 0.002481, "loss": 4.6637, "step": 827 }, { "epoch": 0.00828, "grad_norm": 1.2604684983866747, "learning_rate": 0.002484, "loss": 4.671, "step": 828 }, { "epoch": 0.00829, "grad_norm": 0.8292065747963988, "learning_rate": 0.002487, "loss": 4.6469, "step": 829 }, { "epoch": 0.0083, "grad_norm": 0.8821037896462126, "learning_rate": 0.00249, "loss": 4.6692, "step": 830 }, { "epoch": 0.00831, "grad_norm": 0.7576971723716095, "learning_rate": 0.002493, "loss": 4.6376, "step": 831 }, { "epoch": 0.00832, "grad_norm": 0.759884157810248, "learning_rate": 0.002496, "loss": 4.6375, "step": 832 }, { "epoch": 0.00833, "grad_norm": 0.7334765635370318, "learning_rate": 0.002499, "loss": 4.632, "step": 833 }, { "epoch": 0.00834, "grad_norm": 0.743907849255925, "learning_rate": 0.002502, "loss": 4.6343, "step": 834 }, { "epoch": 0.00835, "grad_norm": 0.8768740048015519, "learning_rate": 0.002505, "loss": 4.633, "step": 835 }, { "epoch": 0.00836, "grad_norm": 0.8050749482625906, "learning_rate": 0.002508, "loss": 4.6653, "step": 836 }, { "epoch": 0.00837, "grad_norm": 0.6963558438544545, "learning_rate": 0.0025109999999999998, "loss": 4.6012, "step": 837 }, { "epoch": 0.00838, "grad_norm": 0.8008054438848442, "learning_rate": 0.0025139999999999997, "loss": 4.6308, "step": 838 }, { "epoch": 0.00839, "grad_norm": 1.048577251560161, "learning_rate": 0.002517, "loss": 4.6417, "step": 839 }, { "epoch": 0.0084, "grad_norm": 1.068829465042819, "learning_rate": 0.00252, "loss": 4.6393, "step": 840 }, { "epoch": 0.00841, "grad_norm": 0.9585677425278923, "learning_rate": 0.002523, "loss": 4.6304, "step": 841 }, { "epoch": 0.00842, "grad_norm": 0.9086027488597042, "learning_rate": 0.002526, "loss": 4.6299, "step": 842 }, { "epoch": 0.00843, "grad_norm": 0.9467491452172306, "learning_rate": 0.002529, "loss": 4.6351, "step": 843 }, { "epoch": 0.00844, "grad_norm": 0.9086629177949005, "learning_rate": 0.002532, "loss": 4.6535, "step": 844 }, { "epoch": 0.00845, "grad_norm": 1.0087851359380933, "learning_rate": 0.002535, "loss": 4.6308, "step": 845 }, { "epoch": 0.00846, "grad_norm": 1.0982346271909766, "learning_rate": 0.002538, "loss": 4.6654, "step": 846 }, { "epoch": 0.00847, "grad_norm": 0.8915036821171446, "learning_rate": 0.002541, "loss": 4.6379, "step": 847 }, { "epoch": 0.00848, "grad_norm": 1.0647705326129857, "learning_rate": 0.002544, "loss": 4.6485, "step": 848 }, { "epoch": 0.00849, "grad_norm": 1.068501032060827, "learning_rate": 0.002547, "loss": 4.6458, "step": 849 }, { "epoch": 0.0085, "grad_norm": 0.9913137717408333, "learning_rate": 0.00255, "loss": 4.6526, "step": 850 }, { "epoch": 0.00851, "grad_norm": 0.8594454569866243, "learning_rate": 0.002553, "loss": 4.6627, "step": 851 }, { "epoch": 0.00852, "grad_norm": 0.896044658032046, "learning_rate": 0.002556, "loss": 4.6576, "step": 852 }, { "epoch": 0.00853, "grad_norm": 0.8600636350866336, "learning_rate": 0.002559, "loss": 4.633, "step": 853 }, { "epoch": 0.00854, "grad_norm": 0.8249013935069577, "learning_rate": 0.002562, "loss": 4.6232, "step": 854 }, { "epoch": 0.00855, "grad_norm": 0.7799882774394947, "learning_rate": 0.002565, "loss": 4.6068, "step": 855 }, { "epoch": 0.00856, "grad_norm": 0.761486927017014, "learning_rate": 0.002568, "loss": 4.6334, "step": 856 }, { "epoch": 0.00857, "grad_norm": 0.7158635780925376, "learning_rate": 0.002571, "loss": 4.6223, "step": 857 }, { "epoch": 0.00858, "grad_norm": 0.774083453902133, "learning_rate": 0.002574, "loss": 4.6142, "step": 858 }, { "epoch": 0.00859, "grad_norm": 0.850850893304573, "learning_rate": 0.002577, "loss": 4.6099, "step": 859 }, { "epoch": 0.0086, "grad_norm": 0.7669731990407863, "learning_rate": 0.00258, "loss": 4.6341, "step": 860 }, { "epoch": 0.00861, "grad_norm": 0.7162038190133796, "learning_rate": 0.0025830000000000002, "loss": 4.6456, "step": 861 }, { "epoch": 0.00862, "grad_norm": 0.6479803425037955, "learning_rate": 0.002586, "loss": 4.5912, "step": 862 }, { "epoch": 0.00863, "grad_norm": 0.6368722094988631, "learning_rate": 0.002589, "loss": 4.6167, "step": 863 }, { "epoch": 0.00864, "grad_norm": 0.6118405956106333, "learning_rate": 0.002592, "loss": 4.5664, "step": 864 }, { "epoch": 0.00865, "grad_norm": 0.5650431634807904, "learning_rate": 0.002595, "loss": 4.5813, "step": 865 }, { "epoch": 0.00866, "grad_norm": 0.6733850251887965, "learning_rate": 0.002598, "loss": 4.6085, "step": 866 }, { "epoch": 0.00867, "grad_norm": 0.7146438239332881, "learning_rate": 0.002601, "loss": 4.5993, "step": 867 }, { "epoch": 0.00868, "grad_norm": 0.5365995387053527, "learning_rate": 0.002604, "loss": 4.5719, "step": 868 }, { "epoch": 0.00869, "grad_norm": 0.5667577739936428, "learning_rate": 0.002607, "loss": 4.5823, "step": 869 }, { "epoch": 0.0087, "grad_norm": 0.6451283729695402, "learning_rate": 0.00261, "loss": 4.5731, "step": 870 }, { "epoch": 0.00871, "grad_norm": 0.7555644992037868, "learning_rate": 0.002613, "loss": 4.5997, "step": 871 }, { "epoch": 0.00872, "grad_norm": 0.8189768002730787, "learning_rate": 0.002616, "loss": 4.5791, "step": 872 }, { "epoch": 0.00873, "grad_norm": 0.6554558150815033, "learning_rate": 0.0026190000000000002, "loss": 4.5881, "step": 873 }, { "epoch": 0.00874, "grad_norm": 0.44798120348082565, "learning_rate": 0.002622, "loss": 4.5966, "step": 874 }, { "epoch": 0.00875, "grad_norm": 0.5565097001333607, "learning_rate": 0.002625, "loss": 4.5495, "step": 875 }, { "epoch": 0.00876, "grad_norm": 0.47756263300713336, "learning_rate": 0.002628, "loss": 4.5755, "step": 876 }, { "epoch": 0.00877, "grad_norm": 0.5306407935778719, "learning_rate": 0.002631, "loss": 4.5933, "step": 877 }, { "epoch": 0.00878, "grad_norm": 0.4614494798142028, "learning_rate": 0.002634, "loss": 4.5228, "step": 878 }, { "epoch": 0.00879, "grad_norm": 0.4660093179336554, "learning_rate": 0.002637, "loss": 4.5537, "step": 879 }, { "epoch": 0.0088, "grad_norm": 0.5031002143838149, "learning_rate": 0.00264, "loss": 4.5622, "step": 880 }, { "epoch": 0.00881, "grad_norm": 0.5059040564958315, "learning_rate": 0.002643, "loss": 4.5642, "step": 881 }, { "epoch": 0.00882, "grad_norm": 0.5172200646638598, "learning_rate": 0.002646, "loss": 4.5628, "step": 882 }, { "epoch": 0.00883, "grad_norm": 0.5784463708218814, "learning_rate": 0.002649, "loss": 4.5589, "step": 883 }, { "epoch": 0.00884, "grad_norm": 0.6180073373235953, "learning_rate": 0.0026520000000000003, "loss": 4.5464, "step": 884 }, { "epoch": 0.00885, "grad_norm": 0.8787173694483682, "learning_rate": 0.0026550000000000002, "loss": 4.5525, "step": 885 }, { "epoch": 0.00886, "grad_norm": 1.0673888729842724, "learning_rate": 0.002658, "loss": 4.6158, "step": 886 }, { "epoch": 0.00887, "grad_norm": 0.6898543177879018, "learning_rate": 0.002661, "loss": 4.5725, "step": 887 }, { "epoch": 0.00888, "grad_norm": 0.853477571603772, "learning_rate": 0.002664, "loss": 4.5844, "step": 888 }, { "epoch": 0.00889, "grad_norm": 0.8678931299790948, "learning_rate": 0.002667, "loss": 4.5451, "step": 889 }, { "epoch": 0.0089, "grad_norm": 0.7924796160958208, "learning_rate": 0.00267, "loss": 4.5923, "step": 890 }, { "epoch": 0.00891, "grad_norm": 0.9978786126328875, "learning_rate": 0.002673, "loss": 4.5837, "step": 891 }, { "epoch": 0.00892, "grad_norm": 0.8940798431065213, "learning_rate": 0.002676, "loss": 4.5952, "step": 892 }, { "epoch": 0.00893, "grad_norm": 0.8958101523241404, "learning_rate": 0.002679, "loss": 4.5818, "step": 893 }, { "epoch": 0.00894, "grad_norm": 0.7899502816364142, "learning_rate": 0.002682, "loss": 4.5617, "step": 894 }, { "epoch": 0.00895, "grad_norm": 0.8690679111994202, "learning_rate": 0.0026850000000000003, "loss": 4.5764, "step": 895 }, { "epoch": 0.00896, "grad_norm": 0.7586764471402897, "learning_rate": 0.0026880000000000003, "loss": 4.5653, "step": 896 }, { "epoch": 0.00897, "grad_norm": 0.8398125438690992, "learning_rate": 0.0026910000000000002, "loss": 4.5844, "step": 897 }, { "epoch": 0.00898, "grad_norm": 0.916381884952665, "learning_rate": 0.002694, "loss": 4.5838, "step": 898 }, { "epoch": 0.00899, "grad_norm": 1.003578523934926, "learning_rate": 0.002697, "loss": 4.5931, "step": 899 }, { "epoch": 0.009, "grad_norm": 1.0819102819564956, "learning_rate": 0.0027, "loss": 4.6243, "step": 900 }, { "epoch": 0.00901, "grad_norm": 1.0380833806593022, "learning_rate": 0.002703, "loss": 4.585, "step": 901 }, { "epoch": 0.00902, "grad_norm": 0.9970218234631744, "learning_rate": 0.002706, "loss": 4.5964, "step": 902 }, { "epoch": 0.00903, "grad_norm": 0.9843987698563316, "learning_rate": 0.002709, "loss": 4.59, "step": 903 }, { "epoch": 0.00904, "grad_norm": 0.9436236396032095, "learning_rate": 0.002712, "loss": 4.5739, "step": 904 }, { "epoch": 0.00905, "grad_norm": 0.9617986484804574, "learning_rate": 0.002715, "loss": 4.5864, "step": 905 }, { "epoch": 0.00906, "grad_norm": 0.9793075174387083, "learning_rate": 0.002718, "loss": 4.6075, "step": 906 }, { "epoch": 0.00907, "grad_norm": 0.829821867155264, "learning_rate": 0.0027210000000000003, "loss": 4.6113, "step": 907 }, { "epoch": 0.00908, "grad_norm": 0.7029669246243915, "learning_rate": 0.0027240000000000003, "loss": 4.5776, "step": 908 }, { "epoch": 0.00909, "grad_norm": 0.6424625298559724, "learning_rate": 0.0027270000000000003, "loss": 4.5922, "step": 909 }, { "epoch": 0.0091, "grad_norm": 0.541466576098649, "learning_rate": 0.0027300000000000002, "loss": 4.591, "step": 910 }, { "epoch": 0.00911, "grad_norm": 0.5322572887883155, "learning_rate": 0.002733, "loss": 4.5472, "step": 911 }, { "epoch": 0.00912, "grad_norm": 0.5035937451256259, "learning_rate": 0.002736, "loss": 4.5482, "step": 912 }, { "epoch": 0.00913, "grad_norm": 0.5644505438751861, "learning_rate": 0.002739, "loss": 4.5413, "step": 913 }, { "epoch": 0.00914, "grad_norm": 0.5623403578732925, "learning_rate": 0.002742, "loss": 4.569, "step": 914 }, { "epoch": 0.00915, "grad_norm": 0.623567024043847, "learning_rate": 0.002745, "loss": 4.5303, "step": 915 }, { "epoch": 0.00916, "grad_norm": 0.7282453918583485, "learning_rate": 0.002748, "loss": 4.5134, "step": 916 }, { "epoch": 0.00917, "grad_norm": 0.849259998793401, "learning_rate": 0.002751, "loss": 4.53, "step": 917 }, { "epoch": 0.00918, "grad_norm": 0.8847496389011684, "learning_rate": 0.0027540000000000004, "loss": 4.5649, "step": 918 }, { "epoch": 0.00919, "grad_norm": 0.5812965220998871, "learning_rate": 0.0027570000000000003, "loss": 4.5257, "step": 919 }, { "epoch": 0.0092, "grad_norm": 0.6986684330141565, "learning_rate": 0.0027600000000000003, "loss": 4.5448, "step": 920 }, { "epoch": 0.00921, "grad_norm": 0.7458959997852286, "learning_rate": 0.0027630000000000003, "loss": 4.5421, "step": 921 }, { "epoch": 0.00922, "grad_norm": 0.7148661298609507, "learning_rate": 0.0027660000000000002, "loss": 4.5394, "step": 922 }, { "epoch": 0.00923, "grad_norm": 0.9687178948744826, "learning_rate": 0.002769, "loss": 4.53, "step": 923 }, { "epoch": 0.00924, "grad_norm": 0.8657140698466169, "learning_rate": 0.002772, "loss": 4.5532, "step": 924 }, { "epoch": 0.00925, "grad_norm": 0.7168831350145214, "learning_rate": 0.002775, "loss": 4.5283, "step": 925 }, { "epoch": 0.00926, "grad_norm": 0.6574829303703393, "learning_rate": 0.002778, "loss": 4.5395, "step": 926 }, { "epoch": 0.00927, "grad_norm": 0.6438234464350411, "learning_rate": 0.002781, "loss": 4.5034, "step": 927 }, { "epoch": 0.00928, "grad_norm": 0.5862425539112647, "learning_rate": 0.002784, "loss": 4.5433, "step": 928 }, { "epoch": 0.00929, "grad_norm": 0.5140610933427406, "learning_rate": 0.0027870000000000004, "loss": 4.5207, "step": 929 }, { "epoch": 0.0093, "grad_norm": 0.5616746929674309, "learning_rate": 0.0027900000000000004, "loss": 4.5043, "step": 930 }, { "epoch": 0.00931, "grad_norm": 0.6437618467570244, "learning_rate": 0.0027930000000000003, "loss": 4.5177, "step": 931 }, { "epoch": 0.00932, "grad_norm": 0.6931141138815896, "learning_rate": 0.0027960000000000003, "loss": 4.5276, "step": 932 }, { "epoch": 0.00933, "grad_norm": 0.5978591705981104, "learning_rate": 0.0027990000000000003, "loss": 4.4907, "step": 933 }, { "epoch": 0.00934, "grad_norm": 0.603121732270584, "learning_rate": 0.0028020000000000002, "loss": 4.5154, "step": 934 }, { "epoch": 0.00935, "grad_norm": 0.6873365438378514, "learning_rate": 0.002805, "loss": 4.528, "step": 935 }, { "epoch": 0.00936, "grad_norm": 0.8285273293509559, "learning_rate": 0.002808, "loss": 4.5353, "step": 936 }, { "epoch": 0.00937, "grad_norm": 0.8237643167810483, "learning_rate": 0.002811, "loss": 4.5562, "step": 937 }, { "epoch": 0.00938, "grad_norm": 0.7428902051912631, "learning_rate": 0.002814, "loss": 4.5253, "step": 938 }, { "epoch": 0.00939, "grad_norm": 0.7604729540920855, "learning_rate": 0.002817, "loss": 4.5245, "step": 939 }, { "epoch": 0.0094, "grad_norm": 0.7827067718534406, "learning_rate": 0.00282, "loss": 4.5235, "step": 940 }, { "epoch": 0.00941, "grad_norm": 0.8076797169385439, "learning_rate": 0.002823, "loss": 4.51, "step": 941 }, { "epoch": 0.00942, "grad_norm": 0.9871062678858393, "learning_rate": 0.002826, "loss": 4.5397, "step": 942 }, { "epoch": 0.00943, "grad_norm": 1.070901691951404, "learning_rate": 0.002829, "loss": 4.5537, "step": 943 }, { "epoch": 0.00944, "grad_norm": 1.1009445115031264, "learning_rate": 0.002832, "loss": 4.5564, "step": 944 }, { "epoch": 0.00945, "grad_norm": 0.9908698154571977, "learning_rate": 0.002835, "loss": 4.5644, "step": 945 }, { "epoch": 0.00946, "grad_norm": 0.8740733112948412, "learning_rate": 0.002838, "loss": 4.5681, "step": 946 }, { "epoch": 0.00947, "grad_norm": 0.8705398172356058, "learning_rate": 0.0028409999999999998, "loss": 4.5461, "step": 947 }, { "epoch": 0.00948, "grad_norm": 1.030159035482114, "learning_rate": 0.0028439999999999997, "loss": 4.5447, "step": 948 }, { "epoch": 0.00949, "grad_norm": 1.1928707625743042, "learning_rate": 0.002847, "loss": 4.5979, "step": 949 }, { "epoch": 0.0095, "grad_norm": 0.8189553326064546, "learning_rate": 0.00285, "loss": 4.5703, "step": 950 }, { "epoch": 0.00951, "grad_norm": 0.7516073328932897, "learning_rate": 0.002853, "loss": 4.5403, "step": 951 }, { "epoch": 0.00952, "grad_norm": 0.770673808631835, "learning_rate": 0.002856, "loss": 4.5655, "step": 952 }, { "epoch": 0.00953, "grad_norm": 0.7009141910952996, "learning_rate": 0.002859, "loss": 4.5043, "step": 953 }, { "epoch": 0.00954, "grad_norm": 0.5282029249680377, "learning_rate": 0.002862, "loss": 4.5519, "step": 954 }, { "epoch": 0.00955, "grad_norm": 0.5219597666883256, "learning_rate": 0.002865, "loss": 4.538, "step": 955 }, { "epoch": 0.00956, "grad_norm": 0.5409798826026391, "learning_rate": 0.002868, "loss": 4.5165, "step": 956 }, { "epoch": 0.00957, "grad_norm": 0.6331938100277174, "learning_rate": 0.002871, "loss": 4.5418, "step": 957 }, { "epoch": 0.00958, "grad_norm": 0.7239336518776873, "learning_rate": 0.002874, "loss": 4.5243, "step": 958 }, { "epoch": 0.00959, "grad_norm": 0.7150326004222285, "learning_rate": 0.002877, "loss": 4.522, "step": 959 }, { "epoch": 0.0096, "grad_norm": 0.6351033882583478, "learning_rate": 0.0028799999999999997, "loss": 4.5137, "step": 960 }, { "epoch": 0.00961, "grad_norm": 0.6041315929206551, "learning_rate": 0.002883, "loss": 4.4839, "step": 961 }, { "epoch": 0.00962, "grad_norm": 0.5086415396789048, "learning_rate": 0.002886, "loss": 4.5083, "step": 962 }, { "epoch": 0.00963, "grad_norm": 0.4921917843701592, "learning_rate": 0.002889, "loss": 4.5074, "step": 963 }, { "epoch": 0.00964, "grad_norm": 0.4478664141352426, "learning_rate": 0.002892, "loss": 4.495, "step": 964 }, { "epoch": 0.00965, "grad_norm": 0.48082018280974176, "learning_rate": 0.002895, "loss": 4.4696, "step": 965 }, { "epoch": 0.00966, "grad_norm": 0.4468338524710237, "learning_rate": 0.002898, "loss": 4.4813, "step": 966 }, { "epoch": 0.00967, "grad_norm": 0.48445225824086435, "learning_rate": 0.002901, "loss": 4.5202, "step": 967 }, { "epoch": 0.00968, "grad_norm": 0.48753092096488154, "learning_rate": 0.002904, "loss": 4.4828, "step": 968 }, { "epoch": 0.00969, "grad_norm": 0.6098233552287393, "learning_rate": 0.002907, "loss": 4.5184, "step": 969 }, { "epoch": 0.0097, "grad_norm": 0.6841177464426628, "learning_rate": 0.00291, "loss": 4.4838, "step": 970 }, { "epoch": 0.00971, "grad_norm": 0.6296009793604628, "learning_rate": 0.002913, "loss": 4.4828, "step": 971 }, { "epoch": 0.00972, "grad_norm": 0.5562757859803139, "learning_rate": 0.002916, "loss": 4.489, "step": 972 }, { "epoch": 0.00973, "grad_norm": 0.6219386413564563, "learning_rate": 0.002919, "loss": 4.4862, "step": 973 }, { "epoch": 0.00974, "grad_norm": 0.6213225377585438, "learning_rate": 0.002922, "loss": 4.4703, "step": 974 }, { "epoch": 0.00975, "grad_norm": 0.6189120844763403, "learning_rate": 0.002925, "loss": 4.495, "step": 975 }, { "epoch": 0.00976, "grad_norm": 0.6980011774079288, "learning_rate": 0.002928, "loss": 4.4782, "step": 976 }, { "epoch": 0.00977, "grad_norm": 0.8069639632421017, "learning_rate": 0.002931, "loss": 4.5242, "step": 977 }, { "epoch": 0.00978, "grad_norm": 0.8591505642618554, "learning_rate": 0.002934, "loss": 4.5028, "step": 978 }, { "epoch": 0.00979, "grad_norm": 1.0057608067163726, "learning_rate": 0.002937, "loss": 4.5085, "step": 979 }, { "epoch": 0.0098, "grad_norm": 0.9763019684979988, "learning_rate": 0.00294, "loss": 4.5104, "step": 980 }, { "epoch": 0.00981, "grad_norm": 1.1460525604086425, "learning_rate": 0.002943, "loss": 4.5283, "step": 981 }, { "epoch": 0.00982, "grad_norm": 0.8686566673097217, "learning_rate": 0.002946, "loss": 4.5182, "step": 982 }, { "epoch": 0.00983, "grad_norm": 0.7552343773130578, "learning_rate": 0.0029490000000000002, "loss": 4.509, "step": 983 }, { "epoch": 0.00984, "grad_norm": 0.7358884017265429, "learning_rate": 0.002952, "loss": 4.4971, "step": 984 }, { "epoch": 0.00985, "grad_norm": 0.6607503765583332, "learning_rate": 0.002955, "loss": 4.5113, "step": 985 }, { "epoch": 0.00986, "grad_norm": 0.7870472473630633, "learning_rate": 0.002958, "loss": 4.4806, "step": 986 }, { "epoch": 0.00987, "grad_norm": 0.8309954161006763, "learning_rate": 0.002961, "loss": 4.5007, "step": 987 }, { "epoch": 0.00988, "grad_norm": 1.0363998670740173, "learning_rate": 0.002964, "loss": 4.5401, "step": 988 }, { "epoch": 0.00989, "grad_norm": 0.9365239840713727, "learning_rate": 0.002967, "loss": 4.498, "step": 989 }, { "epoch": 0.0099, "grad_norm": 0.6864119295113198, "learning_rate": 0.00297, "loss": 4.5125, "step": 990 }, { "epoch": 0.00991, "grad_norm": 0.8171121562090214, "learning_rate": 0.002973, "loss": 4.5158, "step": 991 }, { "epoch": 0.00992, "grad_norm": 0.745069203907096, "learning_rate": 0.002976, "loss": 4.5062, "step": 992 }, { "epoch": 0.00993, "grad_norm": 0.7044611843609352, "learning_rate": 0.002979, "loss": 4.5043, "step": 993 }, { "epoch": 0.00994, "grad_norm": 0.7308973125697386, "learning_rate": 0.002982, "loss": 4.5052, "step": 994 }, { "epoch": 0.00995, "grad_norm": 0.8310300849484714, "learning_rate": 0.0029850000000000002, "loss": 4.5102, "step": 995 }, { "epoch": 0.00996, "grad_norm": 0.8154352480510918, "learning_rate": 0.002988, "loss": 4.5127, "step": 996 }, { "epoch": 0.00997, "grad_norm": 0.6863991992573223, "learning_rate": 0.002991, "loss": 4.4952, "step": 997 }, { "epoch": 0.00998, "grad_norm": 0.7462946911384132, "learning_rate": 0.002994, "loss": 4.5024, "step": 998 }, { "epoch": 0.00999, "grad_norm": 0.7146477573054392, "learning_rate": 0.002997, "loss": 4.483, "step": 999 }, { "epoch": 0.01, "grad_norm": 0.7363866046556987, "learning_rate": 0.003, "loss": 4.5015, "step": 1000 }, { "epoch": 0.01001, "grad_norm": 0.7767634479973017, "learning_rate": 0.003, "loss": 4.5068, "step": 1001 }, { "epoch": 0.01002, "grad_norm": 0.7384346352174465, "learning_rate": 0.003, "loss": 4.4747, "step": 1002 }, { "epoch": 0.01003, "grad_norm": 0.9090237711013536, "learning_rate": 0.003, "loss": 4.5011, "step": 1003 }, { "epoch": 0.01004, "grad_norm": 0.9963866692217829, "learning_rate": 0.003, "loss": 4.5176, "step": 1004 }, { "epoch": 0.01005, "grad_norm": 0.7385625753811653, "learning_rate": 0.003, "loss": 4.4726, "step": 1005 }, { "epoch": 0.01006, "grad_norm": 0.8633843540772702, "learning_rate": 0.003, "loss": 4.4763, "step": 1006 }, { "epoch": 0.01007, "grad_norm": 0.7060101206147821, "learning_rate": 0.003, "loss": 4.5223, "step": 1007 }, { "epoch": 0.01008, "grad_norm": 0.6108586434140804, "learning_rate": 0.003, "loss": 4.5217, "step": 1008 }, { "epoch": 0.01009, "grad_norm": 0.828764309531409, "learning_rate": 0.003, "loss": 4.4986, "step": 1009 }, { "epoch": 0.0101, "grad_norm": 0.8644895268754212, "learning_rate": 0.003, "loss": 4.5093, "step": 1010 }, { "epoch": 0.01011, "grad_norm": 0.848614232580047, "learning_rate": 0.003, "loss": 4.5177, "step": 1011 }, { "epoch": 0.01012, "grad_norm": 0.8343157703645258, "learning_rate": 0.003, "loss": 4.5029, "step": 1012 }, { "epoch": 0.01013, "grad_norm": 0.768035628231052, "learning_rate": 0.003, "loss": 4.5208, "step": 1013 }, { "epoch": 0.01014, "grad_norm": 0.7039487217317226, "learning_rate": 0.003, "loss": 4.4647, "step": 1014 }, { "epoch": 0.01015, "grad_norm": 0.6611578667499322, "learning_rate": 0.003, "loss": 4.5009, "step": 1015 }, { "epoch": 0.01016, "grad_norm": 0.7043885975091321, "learning_rate": 0.003, "loss": 4.5125, "step": 1016 }, { "epoch": 0.01017, "grad_norm": 0.756430652075804, "learning_rate": 0.003, "loss": 4.4573, "step": 1017 }, { "epoch": 0.01018, "grad_norm": 0.8264253518319646, "learning_rate": 0.003, "loss": 4.5131, "step": 1018 }, { "epoch": 0.01019, "grad_norm": 0.6230428354913109, "learning_rate": 0.003, "loss": 4.4842, "step": 1019 }, { "epoch": 0.0102, "grad_norm": 0.5985675812220549, "learning_rate": 0.003, "loss": 4.4799, "step": 1020 }, { "epoch": 0.01021, "grad_norm": 0.6443728807360013, "learning_rate": 0.003, "loss": 4.4622, "step": 1021 }, { "epoch": 0.01022, "grad_norm": 0.5977304558601406, "learning_rate": 0.003, "loss": 4.471, "step": 1022 }, { "epoch": 0.01023, "grad_norm": 0.6033117691674914, "learning_rate": 0.003, "loss": 4.4639, "step": 1023 }, { "epoch": 0.01024, "grad_norm": 0.5721338516828588, "learning_rate": 0.003, "loss": 4.4293, "step": 1024 }, { "epoch": 0.01025, "grad_norm": 0.519887079849401, "learning_rate": 0.003, "loss": 4.4534, "step": 1025 }, { "epoch": 0.01026, "grad_norm": 0.5755595398770954, "learning_rate": 0.003, "loss": 4.4517, "step": 1026 }, { "epoch": 0.01027, "grad_norm": 0.5583328190762978, "learning_rate": 0.003, "loss": 4.4696, "step": 1027 }, { "epoch": 0.01028, "grad_norm": 0.5058651878513157, "learning_rate": 0.003, "loss": 4.4465, "step": 1028 }, { "epoch": 0.01029, "grad_norm": 0.42380394436046726, "learning_rate": 0.003, "loss": 4.4233, "step": 1029 }, { "epoch": 0.0103, "grad_norm": 0.39795258529145094, "learning_rate": 0.003, "loss": 4.4496, "step": 1030 }, { "epoch": 0.01031, "grad_norm": 0.4022804170357935, "learning_rate": 0.003, "loss": 4.4638, "step": 1031 }, { "epoch": 0.01032, "grad_norm": 0.5582373276086869, "learning_rate": 0.003, "loss": 4.4376, "step": 1032 }, { "epoch": 0.01033, "grad_norm": 0.6139042225833007, "learning_rate": 0.003, "loss": 4.4788, "step": 1033 }, { "epoch": 0.01034, "grad_norm": 0.6557191072266806, "learning_rate": 0.003, "loss": 4.4457, "step": 1034 }, { "epoch": 0.01035, "grad_norm": 0.5676080638884442, "learning_rate": 0.003, "loss": 4.4414, "step": 1035 }, { "epoch": 0.01036, "grad_norm": 0.5050742194579916, "learning_rate": 0.003, "loss": 4.428, "step": 1036 }, { "epoch": 0.01037, "grad_norm": 0.5525989752332315, "learning_rate": 0.003, "loss": 4.4551, "step": 1037 }, { "epoch": 0.01038, "grad_norm": 0.5387172699734339, "learning_rate": 0.003, "loss": 4.4503, "step": 1038 }, { "epoch": 0.01039, "grad_norm": 0.6460000205561591, "learning_rate": 0.003, "loss": 4.4441, "step": 1039 }, { "epoch": 0.0104, "grad_norm": 0.6935756269119994, "learning_rate": 0.003, "loss": 4.4606, "step": 1040 }, { "epoch": 0.01041, "grad_norm": 0.5741917605191801, "learning_rate": 0.003, "loss": 4.4375, "step": 1041 }, { "epoch": 0.01042, "grad_norm": 0.6148258717657724, "learning_rate": 0.003, "loss": 4.4633, "step": 1042 }, { "epoch": 0.01043, "grad_norm": 0.6159104547012477, "learning_rate": 0.003, "loss": 4.4115, "step": 1043 }, { "epoch": 0.01044, "grad_norm": 0.6904933868900409, "learning_rate": 0.003, "loss": 4.4475, "step": 1044 }, { "epoch": 0.01045, "grad_norm": 0.8049701467278718, "learning_rate": 0.003, "loss": 4.4778, "step": 1045 }, { "epoch": 0.01046, "grad_norm": 0.9237805199835166, "learning_rate": 0.003, "loss": 4.4531, "step": 1046 }, { "epoch": 0.01047, "grad_norm": 0.9025221925048675, "learning_rate": 0.003, "loss": 4.4626, "step": 1047 }, { "epoch": 0.01048, "grad_norm": 0.8439861117867156, "learning_rate": 0.003, "loss": 4.4594, "step": 1048 }, { "epoch": 0.01049, "grad_norm": 0.7790844674480221, "learning_rate": 0.003, "loss": 4.4347, "step": 1049 }, { "epoch": 0.0105, "grad_norm": 0.776695316132939, "learning_rate": 0.003, "loss": 4.4785, "step": 1050 }, { "epoch": 0.01051, "grad_norm": 0.7863011853992664, "learning_rate": 0.003, "loss": 4.4553, "step": 1051 }, { "epoch": 0.01052, "grad_norm": 0.8021665228413719, "learning_rate": 0.003, "loss": 4.4676, "step": 1052 }, { "epoch": 0.01053, "grad_norm": 0.8250289833375999, "learning_rate": 0.003, "loss": 4.4839, "step": 1053 }, { "epoch": 0.01054, "grad_norm": 0.8127326662842188, "learning_rate": 0.003, "loss": 4.4756, "step": 1054 }, { "epoch": 0.01055, "grad_norm": 0.6832114273454296, "learning_rate": 0.003, "loss": 4.4478, "step": 1055 }, { "epoch": 0.01056, "grad_norm": 0.5228897971280976, "learning_rate": 0.003, "loss": 4.4394, "step": 1056 }, { "epoch": 0.01057, "grad_norm": 0.4987455249643701, "learning_rate": 0.003, "loss": 4.4465, "step": 1057 }, { "epoch": 0.01058, "grad_norm": 0.5069816603194066, "learning_rate": 0.003, "loss": 4.4371, "step": 1058 }, { "epoch": 0.01059, "grad_norm": 0.5329576461470134, "learning_rate": 0.003, "loss": 4.4393, "step": 1059 }, { "epoch": 0.0106, "grad_norm": 0.5913240098111924, "learning_rate": 0.003, "loss": 4.4328, "step": 1060 }, { "epoch": 0.01061, "grad_norm": 0.6641649573073009, "learning_rate": 0.003, "loss": 4.4323, "step": 1061 }, { "epoch": 0.01062, "grad_norm": 0.6948839344713771, "learning_rate": 0.003, "loss": 4.4277, "step": 1062 }, { "epoch": 0.01063, "grad_norm": 0.7496196359877443, "learning_rate": 0.003, "loss": 4.4299, "step": 1063 }, { "epoch": 0.01064, "grad_norm": 0.7407156531442353, "learning_rate": 0.003, "loss": 4.4428, "step": 1064 }, { "epoch": 0.01065, "grad_norm": 0.8146214155770399, "learning_rate": 0.003, "loss": 4.4515, "step": 1065 }, { "epoch": 0.01066, "grad_norm": 1.0565696790056225, "learning_rate": 0.003, "loss": 4.4664, "step": 1066 }, { "epoch": 0.01067, "grad_norm": 0.9516719454349619, "learning_rate": 0.003, "loss": 4.4775, "step": 1067 }, { "epoch": 0.01068, "grad_norm": 0.8836216542706563, "learning_rate": 0.003, "loss": 4.4595, "step": 1068 }, { "epoch": 0.01069, "grad_norm": 0.8856659067512235, "learning_rate": 0.003, "loss": 4.4705, "step": 1069 }, { "epoch": 0.0107, "grad_norm": 0.7369477178565681, "learning_rate": 0.003, "loss": 4.5149, "step": 1070 }, { "epoch": 0.01071, "grad_norm": 0.7091155745251179, "learning_rate": 0.003, "loss": 4.4353, "step": 1071 }, { "epoch": 0.01072, "grad_norm": 0.7806676177349076, "learning_rate": 0.003, "loss": 4.4735, "step": 1072 }, { "epoch": 0.01073, "grad_norm": 0.779498494599653, "learning_rate": 0.003, "loss": 4.4707, "step": 1073 }, { "epoch": 0.01074, "grad_norm": 0.7011503302080855, "learning_rate": 0.003, "loss": 4.4305, "step": 1074 }, { "epoch": 0.01075, "grad_norm": 0.7831586118222456, "learning_rate": 0.003, "loss": 4.4706, "step": 1075 }, { "epoch": 0.01076, "grad_norm": 0.6817777196152708, "learning_rate": 0.003, "loss": 4.4647, "step": 1076 }, { "epoch": 0.01077, "grad_norm": 0.6515982616563128, "learning_rate": 0.003, "loss": 4.4569, "step": 1077 }, { "epoch": 0.01078, "grad_norm": 0.6249083984468133, "learning_rate": 0.003, "loss": 4.4127, "step": 1078 }, { "epoch": 0.01079, "grad_norm": 0.4801837441136389, "learning_rate": 0.003, "loss": 4.4399, "step": 1079 }, { "epoch": 0.0108, "grad_norm": 0.4834910792844242, "learning_rate": 0.003, "loss": 4.414, "step": 1080 }, { "epoch": 0.01081, "grad_norm": 0.5002980543849196, "learning_rate": 0.003, "loss": 4.4211, "step": 1081 }, { "epoch": 0.01082, "grad_norm": 0.46027227348307553, "learning_rate": 0.003, "loss": 4.4371, "step": 1082 }, { "epoch": 0.01083, "grad_norm": 0.5053419259251781, "learning_rate": 0.003, "loss": 4.4248, "step": 1083 }, { "epoch": 0.01084, "grad_norm": 0.5478371338567822, "learning_rate": 0.003, "loss": 4.432, "step": 1084 }, { "epoch": 0.01085, "grad_norm": 0.638950358189342, "learning_rate": 0.003, "loss": 4.4218, "step": 1085 }, { "epoch": 0.01086, "grad_norm": 0.7298487594957375, "learning_rate": 0.003, "loss": 4.4062, "step": 1086 }, { "epoch": 0.01087, "grad_norm": 0.7670847873505783, "learning_rate": 0.003, "loss": 4.4209, "step": 1087 }, { "epoch": 0.01088, "grad_norm": 0.5867260721567662, "learning_rate": 0.003, "loss": 4.4531, "step": 1088 }, { "epoch": 0.01089, "grad_norm": 0.4907796734695404, "learning_rate": 0.003, "loss": 4.4179, "step": 1089 }, { "epoch": 0.0109, "grad_norm": 0.4903707687409907, "learning_rate": 0.003, "loss": 4.4411, "step": 1090 }, { "epoch": 0.01091, "grad_norm": 0.511946289807033, "learning_rate": 0.003, "loss": 4.4087, "step": 1091 }, { "epoch": 0.01092, "grad_norm": 0.5375667414298495, "learning_rate": 0.003, "loss": 4.4066, "step": 1092 }, { "epoch": 0.01093, "grad_norm": 0.5526188836109072, "learning_rate": 0.003, "loss": 4.4252, "step": 1093 }, { "epoch": 0.01094, "grad_norm": 0.6569826260877659, "learning_rate": 0.003, "loss": 4.4168, "step": 1094 }, { "epoch": 0.01095, "grad_norm": 0.6683609284273635, "learning_rate": 0.003, "loss": 4.4455, "step": 1095 }, { "epoch": 0.01096, "grad_norm": 0.6112917342063837, "learning_rate": 0.003, "loss": 4.4132, "step": 1096 }, { "epoch": 0.01097, "grad_norm": 0.5936520390981126, "learning_rate": 0.003, "loss": 4.4026, "step": 1097 }, { "epoch": 0.01098, "grad_norm": 0.5505899855201338, "learning_rate": 0.003, "loss": 4.4193, "step": 1098 }, { "epoch": 0.01099, "grad_norm": 0.5445174984633135, "learning_rate": 0.003, "loss": 4.3908, "step": 1099 }, { "epoch": 0.011, "grad_norm": 0.5222662353377032, "learning_rate": 0.003, "loss": 4.3945, "step": 1100 }, { "epoch": 0.01101, "grad_norm": 0.46913668980272694, "learning_rate": 0.003, "loss": 4.405, "step": 1101 }, { "epoch": 0.01102, "grad_norm": 0.5475692182539631, "learning_rate": 0.003, "loss": 4.4129, "step": 1102 }, { "epoch": 0.01103, "grad_norm": 0.74691541890188, "learning_rate": 0.003, "loss": 4.4147, "step": 1103 }, { "epoch": 0.01104, "grad_norm": 1.169818056602323, "learning_rate": 0.003, "loss": 4.4275, "step": 1104 }, { "epoch": 0.01105, "grad_norm": 0.9343656682141298, "learning_rate": 0.003, "loss": 4.4588, "step": 1105 }, { "epoch": 0.01106, "grad_norm": 0.8373607049055661, "learning_rate": 0.003, "loss": 4.427, "step": 1106 }, { "epoch": 0.01107, "grad_norm": 1.0060128392113217, "learning_rate": 0.003, "loss": 4.4596, "step": 1107 }, { "epoch": 0.01108, "grad_norm": 1.0553010173790884, "learning_rate": 0.003, "loss": 4.4404, "step": 1108 }, { "epoch": 0.01109, "grad_norm": 1.2691952138628209, "learning_rate": 0.003, "loss": 4.4668, "step": 1109 }, { "epoch": 0.0111, "grad_norm": 1.3693624188172686, "learning_rate": 0.003, "loss": 4.519, "step": 1110 }, { "epoch": 0.01111, "grad_norm": 0.8462220605493783, "learning_rate": 0.003, "loss": 4.4548, "step": 1111 }, { "epoch": 0.01112, "grad_norm": 0.9753802461608932, "learning_rate": 0.003, "loss": 4.4881, "step": 1112 }, { "epoch": 0.01113, "grad_norm": 1.109859012060144, "learning_rate": 0.003, "loss": 4.4607, "step": 1113 }, { "epoch": 0.01114, "grad_norm": 1.0190945675303893, "learning_rate": 0.003, "loss": 4.4798, "step": 1114 }, { "epoch": 0.01115, "grad_norm": 1.329753511251153, "learning_rate": 0.003, "loss": 4.5276, "step": 1115 }, { "epoch": 0.01116, "grad_norm": 1.1336923243330905, "learning_rate": 0.003, "loss": 4.5138, "step": 1116 }, { "epoch": 0.01117, "grad_norm": 0.9232555932261653, "learning_rate": 0.003, "loss": 4.4861, "step": 1117 }, { "epoch": 0.01118, "grad_norm": 0.9897100726190513, "learning_rate": 0.003, "loss": 4.4981, "step": 1118 }, { "epoch": 0.01119, "grad_norm": 0.8425069501586361, "learning_rate": 0.003, "loss": 4.496, "step": 1119 }, { "epoch": 0.0112, "grad_norm": 0.9470337823805681, "learning_rate": 0.003, "loss": 4.4807, "step": 1120 }, { "epoch": 0.01121, "grad_norm": 0.8493765745632654, "learning_rate": 0.003, "loss": 4.4938, "step": 1121 }, { "epoch": 0.01122, "grad_norm": 0.7060199222018124, "learning_rate": 0.003, "loss": 4.4573, "step": 1122 }, { "epoch": 0.01123, "grad_norm": 0.6253251988669283, "learning_rate": 0.003, "loss": 4.4403, "step": 1123 }, { "epoch": 0.01124, "grad_norm": 0.579388445721472, "learning_rate": 0.003, "loss": 4.4818, "step": 1124 }, { "epoch": 0.01125, "grad_norm": 0.5338341898835932, "learning_rate": 0.003, "loss": 4.445, "step": 1125 }, { "epoch": 0.01126, "grad_norm": 0.47354440883207194, "learning_rate": 0.003, "loss": 4.4467, "step": 1126 }, { "epoch": 0.01127, "grad_norm": 0.3984491115653698, "learning_rate": 0.003, "loss": 4.4717, "step": 1127 }, { "epoch": 0.01128, "grad_norm": 0.3566421914933017, "learning_rate": 0.003, "loss": 4.4458, "step": 1128 }, { "epoch": 0.01129, "grad_norm": 0.34656729580538187, "learning_rate": 0.003, "loss": 4.4347, "step": 1129 }, { "epoch": 0.0113, "grad_norm": 0.35275594407110167, "learning_rate": 0.003, "loss": 4.4105, "step": 1130 }, { "epoch": 0.01131, "grad_norm": 0.3969022192306343, "learning_rate": 0.003, "loss": 4.4228, "step": 1131 }, { "epoch": 0.01132, "grad_norm": 0.5205676939993591, "learning_rate": 0.003, "loss": 4.4476, "step": 1132 }, { "epoch": 0.01133, "grad_norm": 0.7603712505670571, "learning_rate": 0.003, "loss": 4.4182, "step": 1133 }, { "epoch": 0.01134, "grad_norm": 1.1337080760085034, "learning_rate": 0.003, "loss": 4.4452, "step": 1134 }, { "epoch": 0.01135, "grad_norm": 0.7787650219297657, "learning_rate": 0.003, "loss": 4.4689, "step": 1135 }, { "epoch": 0.01136, "grad_norm": 0.5572304947875402, "learning_rate": 0.003, "loss": 4.431, "step": 1136 }, { "epoch": 0.01137, "grad_norm": 0.8150586648529397, "learning_rate": 0.003, "loss": 4.4648, "step": 1137 }, { "epoch": 0.01138, "grad_norm": 0.6150246614041321, "learning_rate": 0.003, "loss": 4.4126, "step": 1138 }, { "epoch": 0.01139, "grad_norm": 0.5367844011428362, "learning_rate": 0.003, "loss": 4.4175, "step": 1139 }, { "epoch": 0.0114, "grad_norm": 0.6012525993749752, "learning_rate": 0.003, "loss": 4.4009, "step": 1140 }, { "epoch": 0.01141, "grad_norm": 0.5424141892266514, "learning_rate": 0.003, "loss": 4.4201, "step": 1141 }, { "epoch": 0.01142, "grad_norm": 0.5412453890044091, "learning_rate": 0.003, "loss": 4.3966, "step": 1142 }, { "epoch": 0.01143, "grad_norm": 0.5770302538150057, "learning_rate": 0.003, "loss": 4.4119, "step": 1143 }, { "epoch": 0.01144, "grad_norm": 0.5130699212907381, "learning_rate": 0.003, "loss": 4.3751, "step": 1144 }, { "epoch": 0.01145, "grad_norm": 0.5251054242059607, "learning_rate": 0.003, "loss": 4.4104, "step": 1145 }, { "epoch": 0.01146, "grad_norm": 0.49049755788491844, "learning_rate": 0.003, "loss": 4.3944, "step": 1146 }, { "epoch": 0.01147, "grad_norm": 0.5045052675153547, "learning_rate": 0.003, "loss": 4.3854, "step": 1147 }, { "epoch": 0.01148, "grad_norm": 0.4827133382574848, "learning_rate": 0.003, "loss": 4.3821, "step": 1148 }, { "epoch": 0.01149, "grad_norm": 0.5391199926065147, "learning_rate": 0.003, "loss": 4.4015, "step": 1149 }, { "epoch": 0.0115, "grad_norm": 0.556670900436413, "learning_rate": 0.003, "loss": 4.3937, "step": 1150 }, { "epoch": 0.01151, "grad_norm": 0.4138744913914359, "learning_rate": 0.003, "loss": 4.3843, "step": 1151 }, { "epoch": 0.01152, "grad_norm": 0.34659544473454046, "learning_rate": 0.003, "loss": 4.3933, "step": 1152 }, { "epoch": 0.01153, "grad_norm": 0.3606979565928329, "learning_rate": 0.003, "loss": 4.4077, "step": 1153 }, { "epoch": 0.01154, "grad_norm": 0.3838071333965551, "learning_rate": 0.003, "loss": 4.3817, "step": 1154 }, { "epoch": 0.01155, "grad_norm": 0.43382343220964015, "learning_rate": 0.003, "loss": 4.3724, "step": 1155 }, { "epoch": 0.01156, "grad_norm": 0.568032323517354, "learning_rate": 0.003, "loss": 4.3801, "step": 1156 }, { "epoch": 0.01157, "grad_norm": 0.7429160925487139, "learning_rate": 0.003, "loss": 4.4178, "step": 1157 }, { "epoch": 0.01158, "grad_norm": 0.8073054366394413, "learning_rate": 0.003, "loss": 4.387, "step": 1158 }, { "epoch": 0.01159, "grad_norm": 0.6292689134714561, "learning_rate": 0.003, "loss": 4.38, "step": 1159 }, { "epoch": 0.0116, "grad_norm": 0.4911473702594272, "learning_rate": 0.003, "loss": 4.3886, "step": 1160 }, { "epoch": 0.01161, "grad_norm": 0.6900507602397448, "learning_rate": 0.003, "loss": 4.3715, "step": 1161 }, { "epoch": 0.01162, "grad_norm": 0.6121668130868384, "learning_rate": 0.003, "loss": 4.383, "step": 1162 }, { "epoch": 0.01163, "grad_norm": 0.5188516342723972, "learning_rate": 0.003, "loss": 4.398, "step": 1163 }, { "epoch": 0.01164, "grad_norm": 0.47059613888734086, "learning_rate": 0.003, "loss": 4.3849, "step": 1164 }, { "epoch": 0.01165, "grad_norm": 0.4930235270437344, "learning_rate": 0.003, "loss": 4.379, "step": 1165 }, { "epoch": 0.01166, "grad_norm": 0.5090992397072657, "learning_rate": 0.003, "loss": 4.3819, "step": 1166 }, { "epoch": 0.01167, "grad_norm": 0.5100693561752512, "learning_rate": 0.003, "loss": 4.3474, "step": 1167 }, { "epoch": 0.01168, "grad_norm": 0.542972407415325, "learning_rate": 0.003, "loss": 4.3564, "step": 1168 }, { "epoch": 0.01169, "grad_norm": 0.6608507785020126, "learning_rate": 0.003, "loss": 4.4017, "step": 1169 }, { "epoch": 0.0117, "grad_norm": 0.7569575045672728, "learning_rate": 0.003, "loss": 4.3919, "step": 1170 }, { "epoch": 0.01171, "grad_norm": 0.7480827801441813, "learning_rate": 0.003, "loss": 4.3755, "step": 1171 }, { "epoch": 0.01172, "grad_norm": 0.6792739909760177, "learning_rate": 0.003, "loss": 4.3636, "step": 1172 }, { "epoch": 0.01173, "grad_norm": 0.6630549828260829, "learning_rate": 0.003, "loss": 4.4004, "step": 1173 }, { "epoch": 0.01174, "grad_norm": 0.7082139785795519, "learning_rate": 0.003, "loss": 4.3812, "step": 1174 }, { "epoch": 0.01175, "grad_norm": 0.7551341949449439, "learning_rate": 0.003, "loss": 4.3695, "step": 1175 }, { "epoch": 0.01176, "grad_norm": 0.7273809152124429, "learning_rate": 0.003, "loss": 4.4015, "step": 1176 }, { "epoch": 0.01177, "grad_norm": 0.7688229463519815, "learning_rate": 0.003, "loss": 4.3744, "step": 1177 }, { "epoch": 0.01178, "grad_norm": 0.8108730548430517, "learning_rate": 0.003, "loss": 4.3934, "step": 1178 }, { "epoch": 0.01179, "grad_norm": 0.9813575828904657, "learning_rate": 0.003, "loss": 4.3791, "step": 1179 }, { "epoch": 0.0118, "grad_norm": 1.0649482117107174, "learning_rate": 0.003, "loss": 4.4668, "step": 1180 }, { "epoch": 0.01181, "grad_norm": 0.8898608281741577, "learning_rate": 0.003, "loss": 4.4423, "step": 1181 }, { "epoch": 0.01182, "grad_norm": 0.8357467858332619, "learning_rate": 0.003, "loss": 4.4019, "step": 1182 }, { "epoch": 0.01183, "grad_norm": 0.836760730371029, "learning_rate": 0.003, "loss": 4.3978, "step": 1183 }, { "epoch": 0.01184, "grad_norm": 0.877827474309717, "learning_rate": 0.003, "loss": 4.4398, "step": 1184 }, { "epoch": 0.01185, "grad_norm": 0.9245006346426254, "learning_rate": 0.003, "loss": 4.4185, "step": 1185 }, { "epoch": 0.01186, "grad_norm": 0.8254790906319688, "learning_rate": 0.003, "loss": 4.4365, "step": 1186 }, { "epoch": 0.01187, "grad_norm": 1.0721875866066062, "learning_rate": 0.003, "loss": 4.4147, "step": 1187 }, { "epoch": 0.01188, "grad_norm": 0.9234750910217812, "learning_rate": 0.003, "loss": 4.4331, "step": 1188 }, { "epoch": 0.01189, "grad_norm": 0.9441254842548545, "learning_rate": 0.003, "loss": 4.4335, "step": 1189 }, { "epoch": 0.0119, "grad_norm": 0.8429630478570035, "learning_rate": 0.003, "loss": 4.4107, "step": 1190 }, { "epoch": 0.01191, "grad_norm": 0.8544918450444529, "learning_rate": 0.003, "loss": 4.3931, "step": 1191 }, { "epoch": 0.01192, "grad_norm": 0.8750082673567889, "learning_rate": 0.003, "loss": 4.427, "step": 1192 }, { "epoch": 0.01193, "grad_norm": 0.7782283225810788, "learning_rate": 0.003, "loss": 4.4195, "step": 1193 }, { "epoch": 0.01194, "grad_norm": 0.8365585080211665, "learning_rate": 0.003, "loss": 4.4265, "step": 1194 }, { "epoch": 0.01195, "grad_norm": 0.8044450759919293, "learning_rate": 0.003, "loss": 4.4051, "step": 1195 }, { "epoch": 0.01196, "grad_norm": 0.663410887761066, "learning_rate": 0.003, "loss": 4.4006, "step": 1196 }, { "epoch": 0.01197, "grad_norm": 0.7191563284623688, "learning_rate": 0.003, "loss": 4.3713, "step": 1197 }, { "epoch": 0.01198, "grad_norm": 0.6109157259818949, "learning_rate": 0.003, "loss": 4.4347, "step": 1198 }, { "epoch": 0.01199, "grad_norm": 0.49919171672958657, "learning_rate": 0.003, "loss": 4.4063, "step": 1199 }, { "epoch": 0.012, "grad_norm": 0.4537605024865045, "learning_rate": 0.003, "loss": 4.3913, "step": 1200 }, { "epoch": 0.01201, "grad_norm": 0.4780231672540523, "learning_rate": 0.003, "loss": 4.4204, "step": 1201 }, { "epoch": 0.01202, "grad_norm": 0.5600100605398236, "learning_rate": 0.003, "loss": 4.3868, "step": 1202 }, { "epoch": 0.01203, "grad_norm": 0.7311892376344462, "learning_rate": 0.003, "loss": 4.4051, "step": 1203 }, { "epoch": 0.01204, "grad_norm": 0.9179100255668401, "learning_rate": 0.003, "loss": 4.4073, "step": 1204 }, { "epoch": 0.01205, "grad_norm": 0.8405883995035961, "learning_rate": 0.003, "loss": 4.4071, "step": 1205 }, { "epoch": 0.01206, "grad_norm": 0.5805185590607895, "learning_rate": 0.003, "loss": 4.4348, "step": 1206 }, { "epoch": 0.01207, "grad_norm": 0.5705660963600058, "learning_rate": 0.003, "loss": 4.359, "step": 1207 }, { "epoch": 0.01208, "grad_norm": 0.5103594427986314, "learning_rate": 0.003, "loss": 4.4136, "step": 1208 }, { "epoch": 0.01209, "grad_norm": 0.4394123490710629, "learning_rate": 0.003, "loss": 4.3897, "step": 1209 }, { "epoch": 0.0121, "grad_norm": 0.5463405097177936, "learning_rate": 0.003, "loss": 4.3949, "step": 1210 }, { "epoch": 0.01211, "grad_norm": 0.5044353172078102, "learning_rate": 0.003, "loss": 4.3781, "step": 1211 }, { "epoch": 0.01212, "grad_norm": 0.4270131501697402, "learning_rate": 0.003, "loss": 4.3982, "step": 1212 }, { "epoch": 0.01213, "grad_norm": 0.3945884644715176, "learning_rate": 0.003, "loss": 4.3729, "step": 1213 }, { "epoch": 0.01214, "grad_norm": 0.40327036636948815, "learning_rate": 0.003, "loss": 4.3565, "step": 1214 }, { "epoch": 0.01215, "grad_norm": 0.41033770529140634, "learning_rate": 0.003, "loss": 4.3626, "step": 1215 }, { "epoch": 0.01216, "grad_norm": 0.39745070811400113, "learning_rate": 0.003, "loss": 4.3556, "step": 1216 }, { "epoch": 0.01217, "grad_norm": 0.3765107225543665, "learning_rate": 0.003, "loss": 4.3601, "step": 1217 }, { "epoch": 0.01218, "grad_norm": 0.3728731687229966, "learning_rate": 0.003, "loss": 4.3546, "step": 1218 }, { "epoch": 0.01219, "grad_norm": 0.39445630505711005, "learning_rate": 0.003, "loss": 4.3541, "step": 1219 }, { "epoch": 0.0122, "grad_norm": 0.4147897339756862, "learning_rate": 0.003, "loss": 4.3467, "step": 1220 }, { "epoch": 0.01221, "grad_norm": 0.4491697002019059, "learning_rate": 0.003, "loss": 4.3628, "step": 1221 }, { "epoch": 0.01222, "grad_norm": 0.480111589813077, "learning_rate": 0.003, "loss": 4.3459, "step": 1222 }, { "epoch": 0.01223, "grad_norm": 0.4554234447859426, "learning_rate": 0.003, "loss": 4.3426, "step": 1223 }, { "epoch": 0.01224, "grad_norm": 0.4400762187377611, "learning_rate": 0.003, "loss": 4.3264, "step": 1224 }, { "epoch": 0.01225, "grad_norm": 0.6794202546583998, "learning_rate": 0.003, "loss": 4.3569, "step": 1225 }, { "epoch": 0.01226, "grad_norm": 1.0814698752713194, "learning_rate": 0.003, "loss": 4.3701, "step": 1226 }, { "epoch": 0.01227, "grad_norm": 1.155555006554351, "learning_rate": 0.003, "loss": 4.4276, "step": 1227 }, { "epoch": 0.01228, "grad_norm": 0.6549739358010055, "learning_rate": 0.003, "loss": 4.3793, "step": 1228 }, { "epoch": 0.01229, "grad_norm": 0.8770247693165437, "learning_rate": 0.003, "loss": 4.3857, "step": 1229 }, { "epoch": 0.0123, "grad_norm": 0.8853901281486561, "learning_rate": 0.003, "loss": 4.4222, "step": 1230 }, { "epoch": 0.01231, "grad_norm": 0.8670800529033448, "learning_rate": 0.003, "loss": 4.3911, "step": 1231 }, { "epoch": 0.01232, "grad_norm": 0.9431543452250409, "learning_rate": 0.003, "loss": 4.367, "step": 1232 }, { "epoch": 0.01233, "grad_norm": 0.8277618322356394, "learning_rate": 0.003, "loss": 4.3795, "step": 1233 }, { "epoch": 0.01234, "grad_norm": 0.8336748449561182, "learning_rate": 0.003, "loss": 4.3975, "step": 1234 }, { "epoch": 0.01235, "grad_norm": 0.7544795324708781, "learning_rate": 0.003, "loss": 4.3863, "step": 1235 }, { "epoch": 0.01236, "grad_norm": 0.5802889369771423, "learning_rate": 0.003, "loss": 4.3798, "step": 1236 }, { "epoch": 0.01237, "grad_norm": 0.5720134479022692, "learning_rate": 0.003, "loss": 4.39, "step": 1237 }, { "epoch": 0.01238, "grad_norm": 0.5465298337070073, "learning_rate": 0.003, "loss": 4.3824, "step": 1238 }, { "epoch": 0.01239, "grad_norm": 0.4850058714656761, "learning_rate": 0.003, "loss": 4.3643, "step": 1239 }, { "epoch": 0.0124, "grad_norm": 0.4837743776235617, "learning_rate": 0.003, "loss": 4.3458, "step": 1240 }, { "epoch": 0.01241, "grad_norm": 0.5464699719483495, "learning_rate": 0.003, "loss": 4.3589, "step": 1241 }, { "epoch": 0.01242, "grad_norm": 0.5595595597987126, "learning_rate": 0.003, "loss": 4.3633, "step": 1242 }, { "epoch": 0.01243, "grad_norm": 0.5979309976720427, "learning_rate": 0.003, "loss": 4.3445, "step": 1243 }, { "epoch": 0.01244, "grad_norm": 0.6121994614553141, "learning_rate": 0.003, "loss": 4.3608, "step": 1244 }, { "epoch": 0.01245, "grad_norm": 0.5870074144222708, "learning_rate": 0.003, "loss": 4.3495, "step": 1245 }, { "epoch": 0.01246, "grad_norm": 0.5518561222562742, "learning_rate": 0.003, "loss": 4.3238, "step": 1246 }, { "epoch": 0.01247, "grad_norm": 0.4927404519806348, "learning_rate": 0.003, "loss": 4.3772, "step": 1247 }, { "epoch": 0.01248, "grad_norm": 0.5614227665664012, "learning_rate": 0.003, "loss": 4.3637, "step": 1248 }, { "epoch": 0.01249, "grad_norm": 0.6237525658945793, "learning_rate": 0.003, "loss": 4.3523, "step": 1249 }, { "epoch": 0.0125, "grad_norm": 0.6285540244025752, "learning_rate": 0.003, "loss": 4.3285, "step": 1250 }, { "epoch": 0.01251, "grad_norm": 0.6642330670109609, "learning_rate": 0.003, "loss": 4.3632, "step": 1251 }, { "epoch": 0.01252, "grad_norm": 0.7486788521183766, "learning_rate": 0.003, "loss": 4.3494, "step": 1252 }, { "epoch": 0.01253, "grad_norm": 0.8786228409254562, "learning_rate": 0.003, "loss": 4.3342, "step": 1253 }, { "epoch": 0.01254, "grad_norm": 0.890563511155565, "learning_rate": 0.003, "loss": 4.3886, "step": 1254 }, { "epoch": 0.01255, "grad_norm": 0.7085396917976445, "learning_rate": 0.003, "loss": 4.3296, "step": 1255 }, { "epoch": 0.01256, "grad_norm": 0.6070625004395708, "learning_rate": 0.003, "loss": 4.3521, "step": 1256 }, { "epoch": 0.01257, "grad_norm": 0.7279044200924197, "learning_rate": 0.003, "loss": 4.3471, "step": 1257 }, { "epoch": 0.01258, "grad_norm": 0.726475334235256, "learning_rate": 0.003, "loss": 4.353, "step": 1258 }, { "epoch": 0.01259, "grad_norm": 0.657918565683394, "learning_rate": 0.003, "loss": 4.3505, "step": 1259 }, { "epoch": 0.0126, "grad_norm": 0.6055846983934545, "learning_rate": 0.003, "loss": 4.3785, "step": 1260 }, { "epoch": 0.01261, "grad_norm": 0.5773200966833153, "learning_rate": 0.003, "loss": 4.3685, "step": 1261 }, { "epoch": 0.01262, "grad_norm": 0.6458766772660607, "learning_rate": 0.003, "loss": 4.3644, "step": 1262 }, { "epoch": 0.01263, "grad_norm": 0.6994622766289077, "learning_rate": 0.003, "loss": 4.3614, "step": 1263 }, { "epoch": 0.01264, "grad_norm": 0.6737363060454188, "learning_rate": 0.003, "loss": 4.3487, "step": 1264 }, { "epoch": 0.01265, "grad_norm": 0.6931523604065615, "learning_rate": 0.003, "loss": 4.362, "step": 1265 }, { "epoch": 0.01266, "grad_norm": 0.7992655892294886, "learning_rate": 0.003, "loss": 4.3669, "step": 1266 }, { "epoch": 0.01267, "grad_norm": 0.8435255340483996, "learning_rate": 0.003, "loss": 4.3876, "step": 1267 }, { "epoch": 0.01268, "grad_norm": 0.82699021549605, "learning_rate": 0.003, "loss": 4.3772, "step": 1268 }, { "epoch": 0.01269, "grad_norm": 0.8697659895900343, "learning_rate": 0.003, "loss": 4.3448, "step": 1269 }, { "epoch": 0.0127, "grad_norm": 0.8134112717759022, "learning_rate": 0.003, "loss": 4.3816, "step": 1270 }, { "epoch": 0.01271, "grad_norm": 0.6786099831237219, "learning_rate": 0.003, "loss": 4.3563, "step": 1271 }, { "epoch": 0.01272, "grad_norm": 0.5433088474514179, "learning_rate": 0.003, "loss": 4.3456, "step": 1272 }, { "epoch": 0.01273, "grad_norm": 0.6040582030046006, "learning_rate": 0.003, "loss": 4.3736, "step": 1273 }, { "epoch": 0.01274, "grad_norm": 0.6407861769281096, "learning_rate": 0.003, "loss": 4.3663, "step": 1274 }, { "epoch": 0.01275, "grad_norm": 0.6627818267548817, "learning_rate": 0.003, "loss": 4.3377, "step": 1275 }, { "epoch": 0.01276, "grad_norm": 0.6923007139718157, "learning_rate": 0.003, "loss": 4.3441, "step": 1276 }, { "epoch": 0.01277, "grad_norm": 0.5987383094358855, "learning_rate": 0.003, "loss": 4.3497, "step": 1277 }, { "epoch": 0.01278, "grad_norm": 0.48773862011134517, "learning_rate": 0.003, "loss": 4.3516, "step": 1278 }, { "epoch": 0.01279, "grad_norm": 0.48017539333754444, "learning_rate": 0.003, "loss": 4.3344, "step": 1279 }, { "epoch": 0.0128, "grad_norm": 0.3998167780826478, "learning_rate": 0.003, "loss": 4.3274, "step": 1280 }, { "epoch": 0.01281, "grad_norm": 0.48111796457160566, "learning_rate": 0.003, "loss": 4.3157, "step": 1281 }, { "epoch": 0.01282, "grad_norm": 0.48993303263312776, "learning_rate": 0.003, "loss": 4.3556, "step": 1282 }, { "epoch": 0.01283, "grad_norm": 0.6253877533100372, "learning_rate": 0.003, "loss": 4.3297, "step": 1283 }, { "epoch": 0.01284, "grad_norm": 0.5705490801287882, "learning_rate": 0.003, "loss": 4.3318, "step": 1284 }, { "epoch": 0.01285, "grad_norm": 0.5653707188447052, "learning_rate": 0.003, "loss": 4.3333, "step": 1285 }, { "epoch": 0.01286, "grad_norm": 0.6731371693634968, "learning_rate": 0.003, "loss": 4.3633, "step": 1286 }, { "epoch": 0.01287, "grad_norm": 0.8405912851979439, "learning_rate": 0.003, "loss": 4.3324, "step": 1287 }, { "epoch": 0.01288, "grad_norm": 0.8660374446207693, "learning_rate": 0.003, "loss": 4.3541, "step": 1288 }, { "epoch": 0.01289, "grad_norm": 0.7397369781522664, "learning_rate": 0.003, "loss": 4.3429, "step": 1289 }, { "epoch": 0.0129, "grad_norm": 0.757608447039217, "learning_rate": 0.003, "loss": 4.3696, "step": 1290 }, { "epoch": 0.01291, "grad_norm": 0.7650019196438267, "learning_rate": 0.003, "loss": 4.3518, "step": 1291 }, { "epoch": 0.01292, "grad_norm": 0.7951894033365094, "learning_rate": 0.003, "loss": 4.3829, "step": 1292 }, { "epoch": 0.01293, "grad_norm": 0.7837011153839035, "learning_rate": 0.003, "loss": 4.3714, "step": 1293 }, { "epoch": 0.01294, "grad_norm": 0.6526824466083696, "learning_rate": 0.003, "loss": 4.3313, "step": 1294 }, { "epoch": 0.01295, "grad_norm": 0.6353462230193634, "learning_rate": 0.003, "loss": 4.3599, "step": 1295 }, { "epoch": 0.01296, "grad_norm": 0.6488557896271576, "learning_rate": 0.003, "loss": 4.3294, "step": 1296 }, { "epoch": 0.01297, "grad_norm": 0.5423372327946681, "learning_rate": 0.003, "loss": 4.339, "step": 1297 }, { "epoch": 0.01298, "grad_norm": 0.5444047364511552, "learning_rate": 0.003, "loss": 4.3162, "step": 1298 }, { "epoch": 0.01299, "grad_norm": 0.562127858257584, "learning_rate": 0.003, "loss": 4.3274, "step": 1299 }, { "epoch": 0.013, "grad_norm": 0.6451595635625548, "learning_rate": 0.003, "loss": 4.3212, "step": 1300 }, { "epoch": 0.01301, "grad_norm": 0.7106141053798949, "learning_rate": 0.003, "loss": 4.338, "step": 1301 }, { "epoch": 0.01302, "grad_norm": 0.7037506677578025, "learning_rate": 0.003, "loss": 4.3858, "step": 1302 }, { "epoch": 0.01303, "grad_norm": 0.549949183276315, "learning_rate": 0.003, "loss": 4.3254, "step": 1303 }, { "epoch": 0.01304, "grad_norm": 0.5830602260260574, "learning_rate": 0.003, "loss": 4.3529, "step": 1304 }, { "epoch": 0.01305, "grad_norm": 0.6092381647637285, "learning_rate": 0.003, "loss": 4.3402, "step": 1305 }, { "epoch": 0.01306, "grad_norm": 0.9341457164339962, "learning_rate": 0.003, "loss": 4.343, "step": 1306 }, { "epoch": 0.01307, "grad_norm": 1.4224559913594186, "learning_rate": 0.003, "loss": 4.4238, "step": 1307 }, { "epoch": 0.01308, "grad_norm": 0.6683671842157866, "learning_rate": 0.003, "loss": 4.3493, "step": 1308 }, { "epoch": 0.01309, "grad_norm": 1.0351217140401547, "learning_rate": 0.003, "loss": 4.4146, "step": 1309 }, { "epoch": 0.0131, "grad_norm": 0.8334924149650484, "learning_rate": 0.003, "loss": 4.3523, "step": 1310 }, { "epoch": 0.01311, "grad_norm": 0.8015005705393878, "learning_rate": 0.003, "loss": 4.3919, "step": 1311 }, { "epoch": 0.01312, "grad_norm": 0.9176746638594273, "learning_rate": 0.003, "loss": 4.3611, "step": 1312 }, { "epoch": 0.01313, "grad_norm": 0.8139738106820199, "learning_rate": 0.003, "loss": 4.3563, "step": 1313 }, { "epoch": 0.01314, "grad_norm": 0.727251837713852, "learning_rate": 0.003, "loss": 4.3474, "step": 1314 }, { "epoch": 0.01315, "grad_norm": 0.6676632023954393, "learning_rate": 0.003, "loss": 4.3714, "step": 1315 }, { "epoch": 0.01316, "grad_norm": 0.6813048603278513, "learning_rate": 0.003, "loss": 4.3558, "step": 1316 }, { "epoch": 0.01317, "grad_norm": 0.6300708633746414, "learning_rate": 0.003, "loss": 4.378, "step": 1317 }, { "epoch": 0.01318, "grad_norm": 0.7076482507132812, "learning_rate": 0.003, "loss": 4.3605, "step": 1318 }, { "epoch": 0.01319, "grad_norm": 0.6736047843532478, "learning_rate": 0.003, "loss": 4.364, "step": 1319 }, { "epoch": 0.0132, "grad_norm": 0.5964453447416318, "learning_rate": 0.003, "loss": 4.3581, "step": 1320 }, { "epoch": 0.01321, "grad_norm": 0.47030220882918905, "learning_rate": 0.003, "loss": 4.3301, "step": 1321 }, { "epoch": 0.01322, "grad_norm": 0.4742080870560624, "learning_rate": 0.003, "loss": 4.3574, "step": 1322 }, { "epoch": 0.01323, "grad_norm": 0.48984225873706044, "learning_rate": 0.003, "loss": 4.3497, "step": 1323 }, { "epoch": 0.01324, "grad_norm": 0.5369510525758135, "learning_rate": 0.003, "loss": 4.3352, "step": 1324 }, { "epoch": 0.01325, "grad_norm": 0.5504701205401483, "learning_rate": 0.003, "loss": 4.3385, "step": 1325 }, { "epoch": 0.01326, "grad_norm": 0.5673099911125634, "learning_rate": 0.003, "loss": 4.3243, "step": 1326 }, { "epoch": 0.01327, "grad_norm": 0.4865587770330079, "learning_rate": 0.003, "loss": 4.3399, "step": 1327 }, { "epoch": 0.01328, "grad_norm": 0.4578059012385763, "learning_rate": 0.003, "loss": 4.3277, "step": 1328 }, { "epoch": 0.01329, "grad_norm": 0.4815378415098633, "learning_rate": 0.003, "loss": 4.2921, "step": 1329 }, { "epoch": 0.0133, "grad_norm": 0.5592011199263653, "learning_rate": 0.003, "loss": 4.3364, "step": 1330 }, { "epoch": 0.01331, "grad_norm": 0.6524837201480481, "learning_rate": 0.003, "loss": 4.3256, "step": 1331 }, { "epoch": 0.01332, "grad_norm": 0.8576005564857463, "learning_rate": 0.003, "loss": 4.3511, "step": 1332 }, { "epoch": 0.01333, "grad_norm": 0.9139584243573721, "learning_rate": 0.003, "loss": 4.3382, "step": 1333 }, { "epoch": 0.01334, "grad_norm": 0.7699767099990749, "learning_rate": 0.003, "loss": 4.3209, "step": 1334 }, { "epoch": 0.01335, "grad_norm": 0.9105795238149825, "learning_rate": 0.003, "loss": 4.3513, "step": 1335 }, { "epoch": 0.01336, "grad_norm": 0.9449124572578428, "learning_rate": 0.003, "loss": 4.3595, "step": 1336 }, { "epoch": 0.01337, "grad_norm": 0.8839799501947162, "learning_rate": 0.003, "loss": 4.3666, "step": 1337 }, { "epoch": 0.01338, "grad_norm": 0.8700783598647539, "learning_rate": 0.003, "loss": 4.3367, "step": 1338 }, { "epoch": 0.01339, "grad_norm": 0.8610679300232287, "learning_rate": 0.003, "loss": 4.3656, "step": 1339 }, { "epoch": 0.0134, "grad_norm": 0.9245398144375081, "learning_rate": 0.003, "loss": 4.3527, "step": 1340 }, { "epoch": 0.01341, "grad_norm": 0.7746264590506864, "learning_rate": 0.003, "loss": 4.3623, "step": 1341 }, { "epoch": 0.01342, "grad_norm": 0.7444570148279184, "learning_rate": 0.003, "loss": 4.3319, "step": 1342 }, { "epoch": 0.01343, "grad_norm": 0.8969785054543269, "learning_rate": 0.003, "loss": 4.3417, "step": 1343 }, { "epoch": 0.01344, "grad_norm": 0.8665723124328953, "learning_rate": 0.003, "loss": 4.3502, "step": 1344 }, { "epoch": 0.01345, "grad_norm": 0.6831543138899113, "learning_rate": 0.003, "loss": 4.3776, "step": 1345 }, { "epoch": 0.01346, "grad_norm": 0.625697906097118, "learning_rate": 0.003, "loss": 4.3469, "step": 1346 }, { "epoch": 0.01347, "grad_norm": 0.6576058750432187, "learning_rate": 0.003, "loss": 4.3425, "step": 1347 }, { "epoch": 0.01348, "grad_norm": 0.5243855341021313, "learning_rate": 0.003, "loss": 4.3344, "step": 1348 }, { "epoch": 0.01349, "grad_norm": 0.5645682386429235, "learning_rate": 0.003, "loss": 4.3534, "step": 1349 }, { "epoch": 0.0135, "grad_norm": 0.6464802745457856, "learning_rate": 0.003, "loss": 4.3487, "step": 1350 }, { "epoch": 0.01351, "grad_norm": 0.6140774871479286, "learning_rate": 0.003, "loss": 4.3731, "step": 1351 }, { "epoch": 0.01352, "grad_norm": 0.5910057101458851, "learning_rate": 0.003, "loss": 4.3486, "step": 1352 }, { "epoch": 0.01353, "grad_norm": 0.5189635003660854, "learning_rate": 0.003, "loss": 4.3285, "step": 1353 }, { "epoch": 0.01354, "grad_norm": 0.48063452026290554, "learning_rate": 0.003, "loss": 4.3159, "step": 1354 }, { "epoch": 0.01355, "grad_norm": 0.4467878184437842, "learning_rate": 0.003, "loss": 4.3256, "step": 1355 }, { "epoch": 0.01356, "grad_norm": 0.4118799509480309, "learning_rate": 0.003, "loss": 4.3178, "step": 1356 }, { "epoch": 0.01357, "grad_norm": 0.4100502484702344, "learning_rate": 0.003, "loss": 4.3041, "step": 1357 }, { "epoch": 0.01358, "grad_norm": 0.412455124213396, "learning_rate": 0.003, "loss": 4.3003, "step": 1358 }, { "epoch": 0.01359, "grad_norm": 0.46760647383491716, "learning_rate": 0.003, "loss": 4.3395, "step": 1359 }, { "epoch": 0.0136, "grad_norm": 0.5617691070267921, "learning_rate": 0.003, "loss": 4.3077, "step": 1360 }, { "epoch": 0.01361, "grad_norm": 0.7602325372403749, "learning_rate": 0.003, "loss": 4.3297, "step": 1361 }, { "epoch": 0.01362, "grad_norm": 0.9475459172774735, "learning_rate": 0.003, "loss": 4.3591, "step": 1362 }, { "epoch": 0.01363, "grad_norm": 0.8968742883061624, "learning_rate": 0.003, "loss": 4.3557, "step": 1363 }, { "epoch": 0.01364, "grad_norm": 0.6939217241789496, "learning_rate": 0.003, "loss": 4.3451, "step": 1364 }, { "epoch": 0.01365, "grad_norm": 0.7258884734824279, "learning_rate": 0.003, "loss": 4.3658, "step": 1365 }, { "epoch": 0.01366, "grad_norm": 0.6969466740776088, "learning_rate": 0.003, "loss": 4.3146, "step": 1366 }, { "epoch": 0.01367, "grad_norm": 0.5835124525702842, "learning_rate": 0.003, "loss": 4.3609, "step": 1367 }, { "epoch": 0.01368, "grad_norm": 0.6385510495556607, "learning_rate": 0.003, "loss": 4.3329, "step": 1368 }, { "epoch": 0.01369, "grad_norm": 0.6527049554291405, "learning_rate": 0.003, "loss": 4.3277, "step": 1369 }, { "epoch": 0.0137, "grad_norm": 0.5958773621355411, "learning_rate": 0.003, "loss": 4.3484, "step": 1370 }, { "epoch": 0.01371, "grad_norm": 0.5116264880954566, "learning_rate": 0.003, "loss": 4.3379, "step": 1371 }, { "epoch": 0.01372, "grad_norm": 0.47497442415709884, "learning_rate": 0.003, "loss": 4.285, "step": 1372 }, { "epoch": 0.01373, "grad_norm": 0.4919302188338485, "learning_rate": 0.003, "loss": 4.3263, "step": 1373 }, { "epoch": 0.01374, "grad_norm": 0.5098435569626014, "learning_rate": 0.003, "loss": 4.3313, "step": 1374 }, { "epoch": 0.01375, "grad_norm": 0.48516361465791263, "learning_rate": 0.003, "loss": 4.2907, "step": 1375 }, { "epoch": 0.01376, "grad_norm": 0.47820279723214054, "learning_rate": 0.003, "loss": 4.3156, "step": 1376 }, { "epoch": 0.01377, "grad_norm": 0.5605397917837966, "learning_rate": 0.003, "loss": 4.3156, "step": 1377 }, { "epoch": 0.01378, "grad_norm": 0.6074855045770513, "learning_rate": 0.003, "loss": 4.3477, "step": 1378 }, { "epoch": 0.01379, "grad_norm": 0.790948814593468, "learning_rate": 0.003, "loss": 4.3238, "step": 1379 }, { "epoch": 0.0138, "grad_norm": 0.9450973850406263, "learning_rate": 0.003, "loss": 4.3379, "step": 1380 }, { "epoch": 0.01381, "grad_norm": 0.8427221722996694, "learning_rate": 0.003, "loss": 4.3388, "step": 1381 }, { "epoch": 0.01382, "grad_norm": 0.6717154970979985, "learning_rate": 0.003, "loss": 4.3301, "step": 1382 }, { "epoch": 0.01383, "grad_norm": 0.7203290959554749, "learning_rate": 0.003, "loss": 4.3316, "step": 1383 }, { "epoch": 0.01384, "grad_norm": 0.8108684001139901, "learning_rate": 0.003, "loss": 4.3616, "step": 1384 }, { "epoch": 0.01385, "grad_norm": 0.7536569525897636, "learning_rate": 0.003, "loss": 4.3491, "step": 1385 }, { "epoch": 0.01386, "grad_norm": 0.7170208076766044, "learning_rate": 0.003, "loss": 4.3224, "step": 1386 }, { "epoch": 0.01387, "grad_norm": 0.7442453828077332, "learning_rate": 0.003, "loss": 4.3132, "step": 1387 }, { "epoch": 0.01388, "grad_norm": 0.6263710315963331, "learning_rate": 0.003, "loss": 4.3468, "step": 1388 }, { "epoch": 0.01389, "grad_norm": 0.6832177735784069, "learning_rate": 0.003, "loss": 4.3313, "step": 1389 }, { "epoch": 0.0139, "grad_norm": 0.6389171239669225, "learning_rate": 0.003, "loss": 4.3298, "step": 1390 }, { "epoch": 0.01391, "grad_norm": 0.6878908541552311, "learning_rate": 0.003, "loss": 4.3331, "step": 1391 }, { "epoch": 0.01392, "grad_norm": 0.7071113368075915, "learning_rate": 0.003, "loss": 4.3467, "step": 1392 }, { "epoch": 0.01393, "grad_norm": 0.7509678939394911, "learning_rate": 0.003, "loss": 4.3167, "step": 1393 }, { "epoch": 0.01394, "grad_norm": 0.6897663506233761, "learning_rate": 0.003, "loss": 4.3034, "step": 1394 }, { "epoch": 0.01395, "grad_norm": 0.6186721093392761, "learning_rate": 0.003, "loss": 4.3322, "step": 1395 }, { "epoch": 0.01396, "grad_norm": 0.6301020321478175, "learning_rate": 0.003, "loss": 4.3224, "step": 1396 }, { "epoch": 0.01397, "grad_norm": 0.6584085329483399, "learning_rate": 0.003, "loss": 4.3397, "step": 1397 }, { "epoch": 0.01398, "grad_norm": 0.6241984225824584, "learning_rate": 0.003, "loss": 4.3151, "step": 1398 }, { "epoch": 0.01399, "grad_norm": 0.6679710819080718, "learning_rate": 0.003, "loss": 4.3393, "step": 1399 }, { "epoch": 0.014, "grad_norm": 0.7222909611996308, "learning_rate": 0.003, "loss": 4.3009, "step": 1400 }, { "epoch": 0.01401, "grad_norm": 0.7571875348826584, "learning_rate": 0.003, "loss": 4.3281, "step": 1401 }, { "epoch": 0.01402, "grad_norm": 0.6970928306893117, "learning_rate": 0.003, "loss": 4.3342, "step": 1402 }, { "epoch": 0.01403, "grad_norm": 0.7040895953951464, "learning_rate": 0.003, "loss": 4.3087, "step": 1403 }, { "epoch": 0.01404, "grad_norm": 0.6766856807345807, "learning_rate": 0.003, "loss": 4.3304, "step": 1404 }, { "epoch": 0.01405, "grad_norm": 0.6656515972408905, "learning_rate": 0.003, "loss": 4.3452, "step": 1405 }, { "epoch": 0.01406, "grad_norm": 0.6427355374677336, "learning_rate": 0.003, "loss": 4.3293, "step": 1406 }, { "epoch": 0.01407, "grad_norm": 0.5569348725391735, "learning_rate": 0.003, "loss": 4.3382, "step": 1407 }, { "epoch": 0.01408, "grad_norm": 0.653968270617751, "learning_rate": 0.003, "loss": 4.2995, "step": 1408 }, { "epoch": 0.01409, "grad_norm": 0.7092043165323135, "learning_rate": 0.003, "loss": 4.3349, "step": 1409 }, { "epoch": 0.0141, "grad_norm": 0.6596834265722145, "learning_rate": 0.003, "loss": 4.3119, "step": 1410 }, { "epoch": 0.01411, "grad_norm": 0.668518681865148, "learning_rate": 0.003, "loss": 4.2919, "step": 1411 }, { "epoch": 0.01412, "grad_norm": 0.7357916681226567, "learning_rate": 0.003, "loss": 4.3321, "step": 1412 }, { "epoch": 0.01413, "grad_norm": 0.7903826580906821, "learning_rate": 0.003, "loss": 4.3294, "step": 1413 }, { "epoch": 0.01414, "grad_norm": 0.9662669804499839, "learning_rate": 0.003, "loss": 4.3198, "step": 1414 }, { "epoch": 0.01415, "grad_norm": 0.9750513967867162, "learning_rate": 0.003, "loss": 4.327, "step": 1415 }, { "epoch": 0.01416, "grad_norm": 0.766343796622766, "learning_rate": 0.003, "loss": 4.3409, "step": 1416 }, { "epoch": 0.01417, "grad_norm": 0.6901214923655755, "learning_rate": 0.003, "loss": 4.3293, "step": 1417 }, { "epoch": 0.01418, "grad_norm": 0.7874750009631106, "learning_rate": 0.003, "loss": 4.3546, "step": 1418 }, { "epoch": 0.01419, "grad_norm": 0.9550810426960523, "learning_rate": 0.003, "loss": 4.3477, "step": 1419 }, { "epoch": 0.0142, "grad_norm": 0.8975542415095807, "learning_rate": 0.003, "loss": 4.3589, "step": 1420 }, { "epoch": 0.01421, "grad_norm": 0.9518531410750217, "learning_rate": 0.003, "loss": 4.3287, "step": 1421 }, { "epoch": 0.01422, "grad_norm": 0.8839562802775344, "learning_rate": 0.003, "loss": 4.338, "step": 1422 }, { "epoch": 0.01423, "grad_norm": 0.8905632330465241, "learning_rate": 0.003, "loss": 4.3444, "step": 1423 }, { "epoch": 0.01424, "grad_norm": 0.8441033206602504, "learning_rate": 0.003, "loss": 4.3588, "step": 1424 }, { "epoch": 0.01425, "grad_norm": 0.8358861815023613, "learning_rate": 0.003, "loss": 4.3652, "step": 1425 }, { "epoch": 0.01426, "grad_norm": 0.8667170217049547, "learning_rate": 0.003, "loss": 4.345, "step": 1426 }, { "epoch": 0.01427, "grad_norm": 0.9004457242110626, "learning_rate": 0.003, "loss": 4.3475, "step": 1427 }, { "epoch": 0.01428, "grad_norm": 0.6649852293703221, "learning_rate": 0.003, "loss": 4.33, "step": 1428 }, { "epoch": 0.01429, "grad_norm": 0.7036433265664787, "learning_rate": 0.003, "loss": 4.3424, "step": 1429 }, { "epoch": 0.0143, "grad_norm": 0.6541401785894563, "learning_rate": 0.003, "loss": 4.3493, "step": 1430 }, { "epoch": 0.01431, "grad_norm": 0.7703756821195129, "learning_rate": 0.003, "loss": 4.3401, "step": 1431 }, { "epoch": 0.01432, "grad_norm": 0.7699810491963369, "learning_rate": 0.003, "loss": 4.3292, "step": 1432 }, { "epoch": 0.01433, "grad_norm": 0.8450381783395694, "learning_rate": 0.003, "loss": 4.3304, "step": 1433 }, { "epoch": 0.01434, "grad_norm": 0.9597314824876599, "learning_rate": 0.003, "loss": 4.3233, "step": 1434 }, { "epoch": 0.01435, "grad_norm": 0.9933571600055039, "learning_rate": 0.003, "loss": 4.3322, "step": 1435 }, { "epoch": 0.01436, "grad_norm": 0.8443498365288639, "learning_rate": 0.003, "loss": 4.3488, "step": 1436 }, { "epoch": 0.01437, "grad_norm": 0.7389871326992684, "learning_rate": 0.003, "loss": 4.3563, "step": 1437 }, { "epoch": 0.01438, "grad_norm": 0.5911590915667476, "learning_rate": 0.003, "loss": 4.3434, "step": 1438 }, { "epoch": 0.01439, "grad_norm": 0.4545412791958111, "learning_rate": 0.003, "loss": 4.3018, "step": 1439 }, { "epoch": 0.0144, "grad_norm": 0.4765435630344443, "learning_rate": 0.003, "loss": 4.3117, "step": 1440 }, { "epoch": 0.01441, "grad_norm": 0.45011781235472936, "learning_rate": 0.003, "loss": 4.3167, "step": 1441 }, { "epoch": 0.01442, "grad_norm": 0.43134401837040753, "learning_rate": 0.003, "loss": 4.2983, "step": 1442 }, { "epoch": 0.01443, "grad_norm": 0.443910914161992, "learning_rate": 0.003, "loss": 4.3596, "step": 1443 }, { "epoch": 0.01444, "grad_norm": 0.44181787364588626, "learning_rate": 0.003, "loss": 4.3224, "step": 1444 }, { "epoch": 0.01445, "grad_norm": 0.3727371105020461, "learning_rate": 0.003, "loss": 4.2792, "step": 1445 }, { "epoch": 0.01446, "grad_norm": 0.41251015974859473, "learning_rate": 0.003, "loss": 4.2943, "step": 1446 }, { "epoch": 0.01447, "grad_norm": 0.4261928826004591, "learning_rate": 0.003, "loss": 4.3018, "step": 1447 }, { "epoch": 0.01448, "grad_norm": 0.4376299402914724, "learning_rate": 0.003, "loss": 4.31, "step": 1448 }, { "epoch": 0.01449, "grad_norm": 0.4789249067229029, "learning_rate": 0.003, "loss": 4.3061, "step": 1449 }, { "epoch": 0.0145, "grad_norm": 0.5866918905830225, "learning_rate": 0.003, "loss": 4.3209, "step": 1450 }, { "epoch": 0.01451, "grad_norm": 0.6723277356365825, "learning_rate": 0.003, "loss": 4.305, "step": 1451 }, { "epoch": 0.01452, "grad_norm": 0.7335524034560207, "learning_rate": 0.003, "loss": 4.3067, "step": 1452 }, { "epoch": 0.01453, "grad_norm": 0.6471889029033779, "learning_rate": 0.003, "loss": 4.3216, "step": 1453 }, { "epoch": 0.01454, "grad_norm": 0.5685184481528504, "learning_rate": 0.003, "loss": 4.2828, "step": 1454 }, { "epoch": 0.01455, "grad_norm": 0.6421432623095609, "learning_rate": 0.003, "loss": 4.3177, "step": 1455 }, { "epoch": 0.01456, "grad_norm": 0.7565614399335465, "learning_rate": 0.003, "loss": 4.3006, "step": 1456 }, { "epoch": 0.01457, "grad_norm": 0.8829290957594472, "learning_rate": 0.003, "loss": 4.2963, "step": 1457 }, { "epoch": 0.01458, "grad_norm": 0.7109663363437955, "learning_rate": 0.003, "loss": 4.2893, "step": 1458 }, { "epoch": 0.01459, "grad_norm": 0.6673339135688805, "learning_rate": 0.003, "loss": 4.3158, "step": 1459 }, { "epoch": 0.0146, "grad_norm": 0.718687934953482, "learning_rate": 0.003, "loss": 4.3224, "step": 1460 }, { "epoch": 0.01461, "grad_norm": 0.9126976820757061, "learning_rate": 0.003, "loss": 4.3127, "step": 1461 }, { "epoch": 0.01462, "grad_norm": 0.9933957093431367, "learning_rate": 0.003, "loss": 4.3197, "step": 1462 }, { "epoch": 0.01463, "grad_norm": 1.005474136949317, "learning_rate": 0.003, "loss": 4.3328, "step": 1463 }, { "epoch": 0.01464, "grad_norm": 0.8550438674851449, "learning_rate": 0.003, "loss": 4.3041, "step": 1464 }, { "epoch": 0.01465, "grad_norm": 0.95918166934002, "learning_rate": 0.003, "loss": 4.3575, "step": 1465 }, { "epoch": 0.01466, "grad_norm": 0.942212627065398, "learning_rate": 0.003, "loss": 4.3242, "step": 1466 }, { "epoch": 0.01467, "grad_norm": 0.8654410144546052, "learning_rate": 0.003, "loss": 4.361, "step": 1467 }, { "epoch": 0.01468, "grad_norm": 0.9143233338435491, "learning_rate": 0.003, "loss": 4.3391, "step": 1468 }, { "epoch": 0.01469, "grad_norm": 0.9793313044459684, "learning_rate": 0.003, "loss": 4.3458, "step": 1469 }, { "epoch": 0.0147, "grad_norm": 0.9281471829982783, "learning_rate": 0.003, "loss": 4.3406, "step": 1470 }, { "epoch": 0.01471, "grad_norm": 0.8585363920781103, "learning_rate": 0.003, "loss": 4.351, "step": 1471 }, { "epoch": 0.01472, "grad_norm": 0.7828394264486358, "learning_rate": 0.003, "loss": 4.3304, "step": 1472 }, { "epoch": 0.01473, "grad_norm": 0.6693695526508024, "learning_rate": 0.003, "loss": 4.3418, "step": 1473 }, { "epoch": 0.01474, "grad_norm": 0.64488720414541, "learning_rate": 0.003, "loss": 4.3238, "step": 1474 }, { "epoch": 0.01475, "grad_norm": 0.6058521136594603, "learning_rate": 0.003, "loss": 4.3387, "step": 1475 }, { "epoch": 0.01476, "grad_norm": 0.6790050602776871, "learning_rate": 0.003, "loss": 4.3289, "step": 1476 }, { "epoch": 0.01477, "grad_norm": 0.7601717274529913, "learning_rate": 0.003, "loss": 4.3343, "step": 1477 }, { "epoch": 0.01478, "grad_norm": 0.8194656851972575, "learning_rate": 0.003, "loss": 4.321, "step": 1478 }, { "epoch": 0.01479, "grad_norm": 0.8305578095720372, "learning_rate": 0.003, "loss": 4.3542, "step": 1479 }, { "epoch": 0.0148, "grad_norm": 0.5999680255334954, "learning_rate": 0.003, "loss": 4.3147, "step": 1480 }, { "epoch": 0.01481, "grad_norm": 0.5003001300590547, "learning_rate": 0.003, "loss": 4.3103, "step": 1481 }, { "epoch": 0.01482, "grad_norm": 0.536436369218564, "learning_rate": 0.003, "loss": 4.2976, "step": 1482 }, { "epoch": 0.01483, "grad_norm": 0.5356648177629082, "learning_rate": 0.003, "loss": 4.2936, "step": 1483 }, { "epoch": 0.01484, "grad_norm": 0.527550162162803, "learning_rate": 0.003, "loss": 4.3417, "step": 1484 }, { "epoch": 0.01485, "grad_norm": 0.42071613110356837, "learning_rate": 0.003, "loss": 4.3019, "step": 1485 }, { "epoch": 0.01486, "grad_norm": 0.36653534804688837, "learning_rate": 0.003, "loss": 4.2968, "step": 1486 }, { "epoch": 0.01487, "grad_norm": 0.382280891744833, "learning_rate": 0.003, "loss": 4.2989, "step": 1487 }, { "epoch": 0.01488, "grad_norm": 0.33605679846225744, "learning_rate": 0.003, "loss": 4.3064, "step": 1488 }, { "epoch": 0.01489, "grad_norm": 0.3437240442761271, "learning_rate": 0.003, "loss": 4.2981, "step": 1489 }, { "epoch": 0.0149, "grad_norm": 0.39625220166661973, "learning_rate": 0.003, "loss": 4.2911, "step": 1490 }, { "epoch": 0.01491, "grad_norm": 0.46513148809594546, "learning_rate": 0.003, "loss": 4.3054, "step": 1491 }, { "epoch": 0.01492, "grad_norm": 0.6303349430841945, "learning_rate": 0.003, "loss": 4.2763, "step": 1492 }, { "epoch": 0.01493, "grad_norm": 1.0123021829900336, "learning_rate": 0.003, "loss": 4.3178, "step": 1493 }, { "epoch": 0.01494, "grad_norm": 0.9944229901028623, "learning_rate": 0.003, "loss": 4.3204, "step": 1494 }, { "epoch": 0.01495, "grad_norm": 0.6683641506440907, "learning_rate": 0.003, "loss": 4.3148, "step": 1495 }, { "epoch": 0.01496, "grad_norm": 0.7298167952745037, "learning_rate": 0.003, "loss": 4.2971, "step": 1496 }, { "epoch": 0.01497, "grad_norm": 0.6993756176552323, "learning_rate": 0.003, "loss": 4.269, "step": 1497 }, { "epoch": 0.01498, "grad_norm": 0.5745337821538584, "learning_rate": 0.003, "loss": 4.3187, "step": 1498 }, { "epoch": 0.01499, "grad_norm": 0.7283462126028899, "learning_rate": 0.003, "loss": 4.3429, "step": 1499 }, { "epoch": 0.015, "grad_norm": 0.6525566784858753, "learning_rate": 0.003, "loss": 4.2959, "step": 1500 }, { "epoch": 0.01501, "grad_norm": 0.5378940164040006, "learning_rate": 0.003, "loss": 4.2893, "step": 1501 }, { "epoch": 0.01502, "grad_norm": 0.6104287567817624, "learning_rate": 0.003, "loss": 4.3317, "step": 1502 }, { "epoch": 0.01503, "grad_norm": 0.5521816301476207, "learning_rate": 0.003, "loss": 4.3091, "step": 1503 }, { "epoch": 0.01504, "grad_norm": 0.5286969991166217, "learning_rate": 0.003, "loss": 4.307, "step": 1504 }, { "epoch": 0.01505, "grad_norm": 0.5474279287326073, "learning_rate": 0.003, "loss": 4.2885, "step": 1505 }, { "epoch": 0.01506, "grad_norm": 0.5984521889334727, "learning_rate": 0.003, "loss": 4.3094, "step": 1506 }, { "epoch": 0.01507, "grad_norm": 0.5232830237119964, "learning_rate": 0.003, "loss": 4.3199, "step": 1507 }, { "epoch": 0.01508, "grad_norm": 0.5466455223079885, "learning_rate": 0.003, "loss": 4.2974, "step": 1508 }, { "epoch": 0.01509, "grad_norm": 0.5109583189523649, "learning_rate": 0.003, "loss": 4.3142, "step": 1509 }, { "epoch": 0.0151, "grad_norm": 0.4768548441806941, "learning_rate": 0.003, "loss": 4.2831, "step": 1510 }, { "epoch": 0.01511, "grad_norm": 0.4945814249796924, "learning_rate": 0.003, "loss": 4.2714, "step": 1511 }, { "epoch": 0.01512, "grad_norm": 0.48932835669515945, "learning_rate": 0.003, "loss": 4.2822, "step": 1512 }, { "epoch": 0.01513, "grad_norm": 0.47324181197347215, "learning_rate": 0.003, "loss": 4.2875, "step": 1513 }, { "epoch": 0.01514, "grad_norm": 0.47033014673052803, "learning_rate": 0.003, "loss": 4.294, "step": 1514 }, { "epoch": 0.01515, "grad_norm": 0.5275023241779081, "learning_rate": 0.003, "loss": 4.2871, "step": 1515 }, { "epoch": 0.01516, "grad_norm": 0.581253055640685, "learning_rate": 0.003, "loss": 4.2949, "step": 1516 }, { "epoch": 0.01517, "grad_norm": 0.5995156865571368, "learning_rate": 0.003, "loss": 4.245, "step": 1517 }, { "epoch": 0.01518, "grad_norm": 0.6879717293853466, "learning_rate": 0.003, "loss": 4.2878, "step": 1518 }, { "epoch": 0.01519, "grad_norm": 0.8658495938690788, "learning_rate": 0.003, "loss": 4.3291, "step": 1519 }, { "epoch": 0.0152, "grad_norm": 1.0602978136723928, "learning_rate": 0.003, "loss": 4.3012, "step": 1520 }, { "epoch": 0.01521, "grad_norm": 1.0404355729302586, "learning_rate": 0.003, "loss": 4.315, "step": 1521 }, { "epoch": 0.01522, "grad_norm": 0.840417859609453, "learning_rate": 0.003, "loss": 4.3065, "step": 1522 }, { "epoch": 0.01523, "grad_norm": 0.8469845154820008, "learning_rate": 0.003, "loss": 4.305, "step": 1523 }, { "epoch": 0.01524, "grad_norm": 0.7929331263188221, "learning_rate": 0.003, "loss": 4.3335, "step": 1524 }, { "epoch": 0.01525, "grad_norm": 0.698848745465075, "learning_rate": 0.003, "loss": 4.3297, "step": 1525 }, { "epoch": 0.01526, "grad_norm": 0.7207511646846017, "learning_rate": 0.003, "loss": 4.3193, "step": 1526 }, { "epoch": 0.01527, "grad_norm": 0.763308654071157, "learning_rate": 0.003, "loss": 4.2936, "step": 1527 }, { "epoch": 0.01528, "grad_norm": 0.8559921469360918, "learning_rate": 0.003, "loss": 4.3141, "step": 1528 }, { "epoch": 0.01529, "grad_norm": 0.7339586416101129, "learning_rate": 0.003, "loss": 4.3059, "step": 1529 }, { "epoch": 0.0153, "grad_norm": 0.6111884320793721, "learning_rate": 0.003, "loss": 4.3081, "step": 1530 }, { "epoch": 0.01531, "grad_norm": 0.5381224123525938, "learning_rate": 0.003, "loss": 4.2987, "step": 1531 }, { "epoch": 0.01532, "grad_norm": 0.4961970461284698, "learning_rate": 0.003, "loss": 4.2768, "step": 1532 }, { "epoch": 0.01533, "grad_norm": 0.4947055691935514, "learning_rate": 0.003, "loss": 4.2996, "step": 1533 }, { "epoch": 0.01534, "grad_norm": 0.4918952201118249, "learning_rate": 0.003, "loss": 4.3, "step": 1534 }, { "epoch": 0.01535, "grad_norm": 0.5164537975132518, "learning_rate": 0.003, "loss": 4.3084, "step": 1535 }, { "epoch": 0.01536, "grad_norm": 0.5875973667797125, "learning_rate": 0.003, "loss": 4.2915, "step": 1536 }, { "epoch": 0.01537, "grad_norm": 0.6164689981624115, "learning_rate": 0.003, "loss": 4.2859, "step": 1537 }, { "epoch": 0.01538, "grad_norm": 0.5757348971929218, "learning_rate": 0.003, "loss": 4.2647, "step": 1538 }, { "epoch": 0.01539, "grad_norm": 0.47927981795762703, "learning_rate": 0.003, "loss": 4.2913, "step": 1539 }, { "epoch": 0.0154, "grad_norm": 0.471938076315617, "learning_rate": 0.003, "loss": 4.2958, "step": 1540 }, { "epoch": 0.01541, "grad_norm": 0.41121323274233956, "learning_rate": 0.003, "loss": 4.2738, "step": 1541 }, { "epoch": 0.01542, "grad_norm": 0.5568654804454234, "learning_rate": 0.003, "loss": 4.3052, "step": 1542 }, { "epoch": 0.01543, "grad_norm": 0.6966538302893103, "learning_rate": 0.003, "loss": 4.2931, "step": 1543 }, { "epoch": 0.01544, "grad_norm": 0.9220011119214641, "learning_rate": 0.003, "loss": 4.2897, "step": 1544 }, { "epoch": 0.01545, "grad_norm": 0.9623507193232845, "learning_rate": 0.003, "loss": 4.3672, "step": 1545 }, { "epoch": 0.01546, "grad_norm": 0.8825760974316569, "learning_rate": 0.003, "loss": 4.3354, "step": 1546 }, { "epoch": 0.01547, "grad_norm": 0.7561843927072682, "learning_rate": 0.003, "loss": 4.3343, "step": 1547 }, { "epoch": 0.01548, "grad_norm": 0.9189821005507283, "learning_rate": 0.003, "loss": 4.3071, "step": 1548 }, { "epoch": 0.01549, "grad_norm": 0.8566529575807743, "learning_rate": 0.003, "loss": 4.2971, "step": 1549 }, { "epoch": 0.0155, "grad_norm": 0.8661559591366523, "learning_rate": 0.003, "loss": 4.3171, "step": 1550 }, { "epoch": 0.01551, "grad_norm": 0.8122000693068794, "learning_rate": 0.003, "loss": 4.2987, "step": 1551 }, { "epoch": 0.01552, "grad_norm": 0.8086836292136306, "learning_rate": 0.003, "loss": 4.3198, "step": 1552 }, { "epoch": 0.01553, "grad_norm": 0.6333228013975016, "learning_rate": 0.003, "loss": 4.3325, "step": 1553 }, { "epoch": 0.01554, "grad_norm": 0.6383647650899242, "learning_rate": 0.003, "loss": 4.3203, "step": 1554 }, { "epoch": 0.01555, "grad_norm": 0.581782197952493, "learning_rate": 0.003, "loss": 4.2877, "step": 1555 }, { "epoch": 0.01556, "grad_norm": 0.534117297541484, "learning_rate": 0.003, "loss": 4.3229, "step": 1556 }, { "epoch": 0.01557, "grad_norm": 0.6143591785578596, "learning_rate": 0.003, "loss": 4.2988, "step": 1557 }, { "epoch": 0.01558, "grad_norm": 0.7112024057013003, "learning_rate": 0.003, "loss": 4.2905, "step": 1558 }, { "epoch": 0.01559, "grad_norm": 0.8091627873292117, "learning_rate": 0.003, "loss": 4.3321, "step": 1559 }, { "epoch": 0.0156, "grad_norm": 0.9168702262156472, "learning_rate": 0.003, "loss": 4.328, "step": 1560 }, { "epoch": 0.01561, "grad_norm": 0.8247101322453514, "learning_rate": 0.003, "loss": 4.3121, "step": 1561 }, { "epoch": 0.01562, "grad_norm": 0.7450493194783364, "learning_rate": 0.003, "loss": 4.2998, "step": 1562 }, { "epoch": 0.01563, "grad_norm": 0.7681213259871257, "learning_rate": 0.003, "loss": 4.3069, "step": 1563 }, { "epoch": 0.01564, "grad_norm": 0.661087083405721, "learning_rate": 0.003, "loss": 4.3306, "step": 1564 }, { "epoch": 0.01565, "grad_norm": 0.7106779315303835, "learning_rate": 0.003, "loss": 4.3058, "step": 1565 }, { "epoch": 0.01566, "grad_norm": 0.7769110124623916, "learning_rate": 0.003, "loss": 4.3279, "step": 1566 }, { "epoch": 0.01567, "grad_norm": 0.7669441510687823, "learning_rate": 0.003, "loss": 4.3233, "step": 1567 }, { "epoch": 0.01568, "grad_norm": 0.7337582454656044, "learning_rate": 0.003, "loss": 4.3021, "step": 1568 }, { "epoch": 0.01569, "grad_norm": 0.7439380247484012, "learning_rate": 0.003, "loss": 4.3114, "step": 1569 }, { "epoch": 0.0157, "grad_norm": 0.6078620360255516, "learning_rate": 0.003, "loss": 4.3206, "step": 1570 }, { "epoch": 0.01571, "grad_norm": 0.540728762720418, "learning_rate": 0.003, "loss": 4.3025, "step": 1571 }, { "epoch": 0.01572, "grad_norm": 0.46359920637885943, "learning_rate": 0.003, "loss": 4.3098, "step": 1572 }, { "epoch": 0.01573, "grad_norm": 0.44623862939090225, "learning_rate": 0.003, "loss": 4.2906, "step": 1573 }, { "epoch": 0.01574, "grad_norm": 0.40532340498334285, "learning_rate": 0.003, "loss": 4.3042, "step": 1574 }, { "epoch": 0.01575, "grad_norm": 0.4266009557557695, "learning_rate": 0.003, "loss": 4.2644, "step": 1575 }, { "epoch": 0.01576, "grad_norm": 0.44274797480320294, "learning_rate": 0.003, "loss": 4.2843, "step": 1576 }, { "epoch": 0.01577, "grad_norm": 0.5663565081777049, "learning_rate": 0.003, "loss": 4.2511, "step": 1577 }, { "epoch": 0.01578, "grad_norm": 0.7961886454684144, "learning_rate": 0.003, "loss": 4.2885, "step": 1578 }, { "epoch": 0.01579, "grad_norm": 1.0530430537405686, "learning_rate": 0.003, "loss": 4.2959, "step": 1579 }, { "epoch": 0.0158, "grad_norm": 0.8358119926881179, "learning_rate": 0.003, "loss": 4.3048, "step": 1580 }, { "epoch": 0.01581, "grad_norm": 0.6320448131483168, "learning_rate": 0.003, "loss": 4.3019, "step": 1581 }, { "epoch": 0.01582, "grad_norm": 0.8911221736338306, "learning_rate": 0.003, "loss": 4.2956, "step": 1582 }, { "epoch": 0.01583, "grad_norm": 0.8257028036510016, "learning_rate": 0.003, "loss": 4.2853, "step": 1583 }, { "epoch": 0.01584, "grad_norm": 0.6483281568426533, "learning_rate": 0.003, "loss": 4.3088, "step": 1584 }, { "epoch": 0.01585, "grad_norm": 0.8350847075329101, "learning_rate": 0.003, "loss": 4.3083, "step": 1585 }, { "epoch": 0.01586, "grad_norm": 0.7763840572673907, "learning_rate": 0.003, "loss": 4.2866, "step": 1586 }, { "epoch": 0.01587, "grad_norm": 0.6255279228525287, "learning_rate": 0.003, "loss": 4.2947, "step": 1587 }, { "epoch": 0.01588, "grad_norm": 0.610133020546242, "learning_rate": 0.003, "loss": 4.322, "step": 1588 }, { "epoch": 0.01589, "grad_norm": 0.5957874820344337, "learning_rate": 0.003, "loss": 4.3124, "step": 1589 }, { "epoch": 0.0159, "grad_norm": 0.5537392612219366, "learning_rate": 0.003, "loss": 4.3034, "step": 1590 }, { "epoch": 0.01591, "grad_norm": 0.661097427877216, "learning_rate": 0.003, "loss": 4.3134, "step": 1591 }, { "epoch": 0.01592, "grad_norm": 0.6367246417778138, "learning_rate": 0.003, "loss": 4.3034, "step": 1592 }, { "epoch": 0.01593, "grad_norm": 0.6726022188533456, "learning_rate": 0.003, "loss": 4.3067, "step": 1593 }, { "epoch": 0.01594, "grad_norm": 0.806293600982127, "learning_rate": 0.003, "loss": 4.302, "step": 1594 }, { "epoch": 0.01595, "grad_norm": 0.7676445449638729, "learning_rate": 0.003, "loss": 4.3248, "step": 1595 }, { "epoch": 0.01596, "grad_norm": 0.7697612568511226, "learning_rate": 0.003, "loss": 4.3187, "step": 1596 }, { "epoch": 0.01597, "grad_norm": 0.6464451152190547, "learning_rate": 0.003, "loss": 4.278, "step": 1597 }, { "epoch": 0.01598, "grad_norm": 0.5841344344075732, "learning_rate": 0.003, "loss": 4.2687, "step": 1598 }, { "epoch": 0.01599, "grad_norm": 0.6015534953604151, "learning_rate": 0.003, "loss": 4.2919, "step": 1599 }, { "epoch": 0.016, "grad_norm": 0.633770866019041, "learning_rate": 0.003, "loss": 4.2557, "step": 1600 }, { "epoch": 0.01601, "grad_norm": 0.7204786711919867, "learning_rate": 0.003, "loss": 4.3091, "step": 1601 }, { "epoch": 0.01602, "grad_norm": 0.8805640455608943, "learning_rate": 0.003, "loss": 4.2878, "step": 1602 }, { "epoch": 0.01603, "grad_norm": 0.9584887795577403, "learning_rate": 0.003, "loss": 4.3165, "step": 1603 }, { "epoch": 0.01604, "grad_norm": 0.7106738441517636, "learning_rate": 0.003, "loss": 4.3017, "step": 1604 }, { "epoch": 0.01605, "grad_norm": 0.5768021422231241, "learning_rate": 0.003, "loss": 4.27, "step": 1605 }, { "epoch": 0.01606, "grad_norm": 0.7201210994211575, "learning_rate": 0.003, "loss": 4.2807, "step": 1606 }, { "epoch": 0.01607, "grad_norm": 0.657321999239318, "learning_rate": 0.003, "loss": 4.3034, "step": 1607 }, { "epoch": 0.01608, "grad_norm": 0.6130445370824126, "learning_rate": 0.003, "loss": 4.2741, "step": 1608 }, { "epoch": 0.01609, "grad_norm": 0.5653798376158339, "learning_rate": 0.003, "loss": 4.283, "step": 1609 }, { "epoch": 0.0161, "grad_norm": 0.513983969651206, "learning_rate": 0.003, "loss": 4.2462, "step": 1610 }, { "epoch": 0.01611, "grad_norm": 0.46001761549467546, "learning_rate": 0.003, "loss": 4.2646, "step": 1611 }, { "epoch": 0.01612, "grad_norm": 0.4061198970721476, "learning_rate": 0.003, "loss": 4.2535, "step": 1612 }, { "epoch": 0.01613, "grad_norm": 0.38584265397880474, "learning_rate": 0.003, "loss": 4.298, "step": 1613 }, { "epoch": 0.01614, "grad_norm": 0.3634152641224797, "learning_rate": 0.003, "loss": 4.3203, "step": 1614 }, { "epoch": 0.01615, "grad_norm": 0.3355266833643695, "learning_rate": 0.003, "loss": 4.286, "step": 1615 }, { "epoch": 0.01616, "grad_norm": 0.37730904621972394, "learning_rate": 0.003, "loss": 4.267, "step": 1616 }, { "epoch": 0.01617, "grad_norm": 0.39995734331949706, "learning_rate": 0.003, "loss": 4.2576, "step": 1617 }, { "epoch": 0.01618, "grad_norm": 0.47091829338804436, "learning_rate": 0.003, "loss": 4.2623, "step": 1618 }, { "epoch": 0.01619, "grad_norm": 0.6125482990333543, "learning_rate": 0.003, "loss": 4.2565, "step": 1619 }, { "epoch": 0.0162, "grad_norm": 0.7863263974252909, "learning_rate": 0.003, "loss": 4.2735, "step": 1620 }, { "epoch": 0.01621, "grad_norm": 1.0441142509760066, "learning_rate": 0.003, "loss": 4.2794, "step": 1621 }, { "epoch": 0.01622, "grad_norm": 0.9658316314084775, "learning_rate": 0.003, "loss": 4.3224, "step": 1622 }, { "epoch": 0.01623, "grad_norm": 0.8426526797330657, "learning_rate": 0.003, "loss": 4.2964, "step": 1623 }, { "epoch": 0.01624, "grad_norm": 0.9277624680796056, "learning_rate": 0.003, "loss": 4.3239, "step": 1624 }, { "epoch": 0.01625, "grad_norm": 0.9569376041612643, "learning_rate": 0.003, "loss": 4.3005, "step": 1625 }, { "epoch": 0.01626, "grad_norm": 0.9128199086554656, "learning_rate": 0.003, "loss": 4.3221, "step": 1626 }, { "epoch": 0.01627, "grad_norm": 1.0524568862767718, "learning_rate": 0.003, "loss": 4.3041, "step": 1627 }, { "epoch": 0.01628, "grad_norm": 1.051492497501998, "learning_rate": 0.003, "loss": 4.3114, "step": 1628 }, { "epoch": 0.01629, "grad_norm": 0.9795755381617424, "learning_rate": 0.003, "loss": 4.3139, "step": 1629 }, { "epoch": 0.0163, "grad_norm": 0.8842687005527395, "learning_rate": 0.003, "loss": 4.301, "step": 1630 }, { "epoch": 0.01631, "grad_norm": 0.8178000647706527, "learning_rate": 0.003, "loss": 4.3269, "step": 1631 }, { "epoch": 0.01632, "grad_norm": 0.8099859797576497, "learning_rate": 0.003, "loss": 4.3331, "step": 1632 }, { "epoch": 0.01633, "grad_norm": 0.8783297751338949, "learning_rate": 0.003, "loss": 4.3124, "step": 1633 }, { "epoch": 0.01634, "grad_norm": 0.9683176525782718, "learning_rate": 0.003, "loss": 4.3368, "step": 1634 }, { "epoch": 0.01635, "grad_norm": 1.0378085972515856, "learning_rate": 0.003, "loss": 4.3267, "step": 1635 }, { "epoch": 0.01636, "grad_norm": 0.8713002838428562, "learning_rate": 0.003, "loss": 4.3348, "step": 1636 }, { "epoch": 0.01637, "grad_norm": 0.6789025774910509, "learning_rate": 0.003, "loss": 4.3135, "step": 1637 }, { "epoch": 0.01638, "grad_norm": 0.7190263237234923, "learning_rate": 0.003, "loss": 4.3194, "step": 1638 }, { "epoch": 0.01639, "grad_norm": 0.6786031217139576, "learning_rate": 0.003, "loss": 4.3383, "step": 1639 }, { "epoch": 0.0164, "grad_norm": 0.6789558457431205, "learning_rate": 0.003, "loss": 4.3166, "step": 1640 }, { "epoch": 0.01641, "grad_norm": 0.575948861365857, "learning_rate": 0.003, "loss": 4.2883, "step": 1641 }, { "epoch": 0.01642, "grad_norm": 0.5365877148633437, "learning_rate": 0.003, "loss": 4.3254, "step": 1642 }, { "epoch": 0.01643, "grad_norm": 0.5249085338304371, "learning_rate": 0.003, "loss": 4.3093, "step": 1643 }, { "epoch": 0.01644, "grad_norm": 0.5581803861532463, "learning_rate": 0.003, "loss": 4.3259, "step": 1644 }, { "epoch": 0.01645, "grad_norm": 0.5119961826198595, "learning_rate": 0.003, "loss": 4.2692, "step": 1645 }, { "epoch": 0.01646, "grad_norm": 0.5843459575894068, "learning_rate": 0.003, "loss": 4.3002, "step": 1646 }, { "epoch": 0.01647, "grad_norm": 0.6503610890489644, "learning_rate": 0.003, "loss": 4.299, "step": 1647 }, { "epoch": 0.01648, "grad_norm": 0.7233252288289422, "learning_rate": 0.003, "loss": 4.2792, "step": 1648 }, { "epoch": 0.01649, "grad_norm": 0.6990475841330179, "learning_rate": 0.003, "loss": 4.315, "step": 1649 }, { "epoch": 0.0165, "grad_norm": 0.6420391516572309, "learning_rate": 0.003, "loss": 4.2891, "step": 1650 }, { "epoch": 0.01651, "grad_norm": 0.5226368448156455, "learning_rate": 0.003, "loss": 4.2942, "step": 1651 }, { "epoch": 0.01652, "grad_norm": 0.610714083694528, "learning_rate": 0.003, "loss": 4.2513, "step": 1652 }, { "epoch": 0.01653, "grad_norm": 0.6780860125642644, "learning_rate": 0.003, "loss": 4.295, "step": 1653 }, { "epoch": 0.01654, "grad_norm": 0.6227039363886209, "learning_rate": 0.003, "loss": 4.284, "step": 1654 }, { "epoch": 0.01655, "grad_norm": 0.5328823365124649, "learning_rate": 0.003, "loss": 4.2906, "step": 1655 }, { "epoch": 0.01656, "grad_norm": 0.4763312033593008, "learning_rate": 0.003, "loss": 4.2565, "step": 1656 }, { "epoch": 0.01657, "grad_norm": 0.4375990417622915, "learning_rate": 0.003, "loss": 4.275, "step": 1657 }, { "epoch": 0.01658, "grad_norm": 0.41683501760814345, "learning_rate": 0.003, "loss": 4.2614, "step": 1658 }, { "epoch": 0.01659, "grad_norm": 0.36860502577039955, "learning_rate": 0.003, "loss": 4.2638, "step": 1659 }, { "epoch": 0.0166, "grad_norm": 0.3864652159021637, "learning_rate": 0.003, "loss": 4.2474, "step": 1660 }, { "epoch": 0.01661, "grad_norm": 0.41682106736234836, "learning_rate": 0.003, "loss": 4.2505, "step": 1661 }, { "epoch": 0.01662, "grad_norm": 0.476701756557898, "learning_rate": 0.003, "loss": 4.2548, "step": 1662 }, { "epoch": 0.01663, "grad_norm": 0.5969063400757824, "learning_rate": 0.003, "loss": 4.2805, "step": 1663 }, { "epoch": 0.01664, "grad_norm": 0.8680763918881209, "learning_rate": 0.003, "loss": 4.3046, "step": 1664 }, { "epoch": 0.01665, "grad_norm": 1.1237216483274195, "learning_rate": 0.003, "loss": 4.2931, "step": 1665 }, { "epoch": 0.01666, "grad_norm": 0.7417184604500126, "learning_rate": 0.003, "loss": 4.2771, "step": 1666 }, { "epoch": 0.01667, "grad_norm": 0.7184967197248885, "learning_rate": 0.003, "loss": 4.2946, "step": 1667 }, { "epoch": 0.01668, "grad_norm": 0.8198630062596927, "learning_rate": 0.003, "loss": 4.305, "step": 1668 }, { "epoch": 0.01669, "grad_norm": 0.7847871916676272, "learning_rate": 0.003, "loss": 4.2844, "step": 1669 }, { "epoch": 0.0167, "grad_norm": 0.9437529917495628, "learning_rate": 0.003, "loss": 4.2839, "step": 1670 }, { "epoch": 0.01671, "grad_norm": 0.9912342863425158, "learning_rate": 0.003, "loss": 4.2849, "step": 1671 }, { "epoch": 0.01672, "grad_norm": 0.881393206900321, "learning_rate": 0.003, "loss": 4.2921, "step": 1672 }, { "epoch": 0.01673, "grad_norm": 0.8454070541781228, "learning_rate": 0.003, "loss": 4.2895, "step": 1673 }, { "epoch": 0.01674, "grad_norm": 0.7698332601553451, "learning_rate": 0.003, "loss": 4.3011, "step": 1674 }, { "epoch": 0.01675, "grad_norm": 0.7526884572805107, "learning_rate": 0.003, "loss": 4.2859, "step": 1675 }, { "epoch": 0.01676, "grad_norm": 0.608849775196141, "learning_rate": 0.003, "loss": 4.288, "step": 1676 }, { "epoch": 0.01677, "grad_norm": 0.587083227500864, "learning_rate": 0.003, "loss": 4.2754, "step": 1677 }, { "epoch": 0.01678, "grad_norm": 0.5796575721567071, "learning_rate": 0.003, "loss": 4.2871, "step": 1678 }, { "epoch": 0.01679, "grad_norm": 0.5746576795495367, "learning_rate": 0.003, "loss": 4.2709, "step": 1679 }, { "epoch": 0.0168, "grad_norm": 0.6091387034729067, "learning_rate": 0.003, "loss": 4.2876, "step": 1680 }, { "epoch": 0.01681, "grad_norm": 0.5440787081039196, "learning_rate": 0.003, "loss": 4.2859, "step": 1681 }, { "epoch": 0.01682, "grad_norm": 0.5021874872953848, "learning_rate": 0.003, "loss": 4.2746, "step": 1682 }, { "epoch": 0.01683, "grad_norm": 0.5600641996520215, "learning_rate": 0.003, "loss": 4.2891, "step": 1683 }, { "epoch": 0.01684, "grad_norm": 0.6044969913973193, "learning_rate": 0.003, "loss": 4.2655, "step": 1684 }, { "epoch": 0.01685, "grad_norm": 0.7175607093807479, "learning_rate": 0.003, "loss": 4.2832, "step": 1685 }, { "epoch": 0.01686, "grad_norm": 0.7107262709635087, "learning_rate": 0.003, "loss": 4.278, "step": 1686 }, { "epoch": 0.01687, "grad_norm": 0.6547857303100532, "learning_rate": 0.003, "loss": 4.2707, "step": 1687 }, { "epoch": 0.01688, "grad_norm": 0.6516245322949453, "learning_rate": 0.003, "loss": 4.258, "step": 1688 }, { "epoch": 0.01689, "grad_norm": 0.6229389769593134, "learning_rate": 0.003, "loss": 4.2721, "step": 1689 }, { "epoch": 0.0169, "grad_norm": 0.6103687657490365, "learning_rate": 0.003, "loss": 4.2607, "step": 1690 }, { "epoch": 0.01691, "grad_norm": 0.5335862349810607, "learning_rate": 0.003, "loss": 4.27, "step": 1691 }, { "epoch": 0.01692, "grad_norm": 0.5575535626191942, "learning_rate": 0.003, "loss": 4.2695, "step": 1692 }, { "epoch": 0.01693, "grad_norm": 0.568239333319019, "learning_rate": 0.003, "loss": 4.2967, "step": 1693 }, { "epoch": 0.01694, "grad_norm": 0.6200186749942788, "learning_rate": 0.003, "loss": 4.2481, "step": 1694 }, { "epoch": 0.01695, "grad_norm": 0.7306002506956177, "learning_rate": 0.003, "loss": 4.2852, "step": 1695 }, { "epoch": 0.01696, "grad_norm": 0.9861942877771345, "learning_rate": 0.003, "loss": 4.2865, "step": 1696 }, { "epoch": 0.01697, "grad_norm": 0.9202516729690784, "learning_rate": 0.003, "loss": 4.2714, "step": 1697 }, { "epoch": 0.01698, "grad_norm": 0.5813652600697033, "learning_rate": 0.003, "loss": 4.2517, "step": 1698 }, { "epoch": 0.01699, "grad_norm": 0.7906905904924819, "learning_rate": 0.003, "loss": 4.2666, "step": 1699 }, { "epoch": 0.017, "grad_norm": 0.8251190957906807, "learning_rate": 0.003, "loss": 4.2912, "step": 1700 }, { "epoch": 0.01701, "grad_norm": 0.5760899682258609, "learning_rate": 0.003, "loss": 4.2636, "step": 1701 }, { "epoch": 0.01702, "grad_norm": 0.6197604398341542, "learning_rate": 0.003, "loss": 4.2892, "step": 1702 }, { "epoch": 0.01703, "grad_norm": 0.6600695849762079, "learning_rate": 0.003, "loss": 4.2772, "step": 1703 }, { "epoch": 0.01704, "grad_norm": 0.6083291374384225, "learning_rate": 0.003, "loss": 4.2615, "step": 1704 }, { "epoch": 0.01705, "grad_norm": 0.6628061782787544, "learning_rate": 0.003, "loss": 4.2656, "step": 1705 }, { "epoch": 0.01706, "grad_norm": 0.6779759701213098, "learning_rate": 0.003, "loss": 4.2652, "step": 1706 }, { "epoch": 0.01707, "grad_norm": 0.6565491562339272, "learning_rate": 0.003, "loss": 4.2864, "step": 1707 }, { "epoch": 0.01708, "grad_norm": 0.72356139072098, "learning_rate": 0.003, "loss": 4.2639, "step": 1708 }, { "epoch": 0.01709, "grad_norm": 0.7465033977044144, "learning_rate": 0.003, "loss": 4.2882, "step": 1709 }, { "epoch": 0.0171, "grad_norm": 0.8329207828708123, "learning_rate": 0.003, "loss": 4.2425, "step": 1710 }, { "epoch": 0.01711, "grad_norm": 0.7868917410962508, "learning_rate": 0.003, "loss": 4.2732, "step": 1711 }, { "epoch": 0.01712, "grad_norm": 0.7908188817256777, "learning_rate": 0.003, "loss": 4.283, "step": 1712 }, { "epoch": 0.01713, "grad_norm": 0.7158559020644585, "learning_rate": 0.003, "loss": 4.2586, "step": 1713 }, { "epoch": 0.01714, "grad_norm": 0.6677106459549558, "learning_rate": 0.003, "loss": 4.2919, "step": 1714 }, { "epoch": 0.01715, "grad_norm": 0.6431660408355871, "learning_rate": 0.003, "loss": 4.2988, "step": 1715 }, { "epoch": 0.01716, "grad_norm": 0.6261280035670769, "learning_rate": 0.003, "loss": 4.2745, "step": 1716 }, { "epoch": 0.01717, "grad_norm": 0.654451460271765, "learning_rate": 0.003, "loss": 4.308, "step": 1717 }, { "epoch": 0.01718, "grad_norm": 0.7724030260270882, "learning_rate": 0.003, "loss": 4.2766, "step": 1718 }, { "epoch": 0.01719, "grad_norm": 0.8282403870632441, "learning_rate": 0.003, "loss": 4.3039, "step": 1719 }, { "epoch": 0.0172, "grad_norm": 0.8311264118816778, "learning_rate": 0.003, "loss": 4.2657, "step": 1720 }, { "epoch": 0.01721, "grad_norm": 0.6805806640783407, "learning_rate": 0.003, "loss": 4.2779, "step": 1721 }, { "epoch": 0.01722, "grad_norm": 0.7016597203863224, "learning_rate": 0.003, "loss": 4.2711, "step": 1722 }, { "epoch": 0.01723, "grad_norm": 0.8810025499012973, "learning_rate": 0.003, "loss": 4.2844, "step": 1723 }, { "epoch": 0.01724, "grad_norm": 0.843829134491477, "learning_rate": 0.003, "loss": 4.2918, "step": 1724 }, { "epoch": 0.01725, "grad_norm": 0.7958111364701067, "learning_rate": 0.003, "loss": 4.2741, "step": 1725 }, { "epoch": 0.01726, "grad_norm": 0.7747340922535558, "learning_rate": 0.003, "loss": 4.2656, "step": 1726 }, { "epoch": 0.01727, "grad_norm": 0.7603713990411842, "learning_rate": 0.003, "loss": 4.2713, "step": 1727 }, { "epoch": 0.01728, "grad_norm": 0.6106099006000164, "learning_rate": 0.003, "loss": 4.2671, "step": 1728 }, { "epoch": 0.01729, "grad_norm": 0.6017782155965361, "learning_rate": 0.003, "loss": 4.2747, "step": 1729 }, { "epoch": 0.0173, "grad_norm": 0.48590601550516704, "learning_rate": 0.003, "loss": 4.2446, "step": 1730 }, { "epoch": 0.01731, "grad_norm": 0.44227547910603526, "learning_rate": 0.003, "loss": 4.2625, "step": 1731 }, { "epoch": 0.01732, "grad_norm": 0.38121233110398844, "learning_rate": 0.003, "loss": 4.2664, "step": 1732 }, { "epoch": 0.01733, "grad_norm": 0.36765597229957125, "learning_rate": 0.003, "loss": 4.2477, "step": 1733 }, { "epoch": 0.01734, "grad_norm": 0.3640345239833258, "learning_rate": 0.003, "loss": 4.2562, "step": 1734 }, { "epoch": 0.01735, "grad_norm": 0.41470571844372184, "learning_rate": 0.003, "loss": 4.2282, "step": 1735 }, { "epoch": 0.01736, "grad_norm": 0.4147613401672916, "learning_rate": 0.003, "loss": 4.2285, "step": 1736 }, { "epoch": 0.01737, "grad_norm": 0.38708456345044623, "learning_rate": 0.003, "loss": 4.2561, "step": 1737 }, { "epoch": 0.01738, "grad_norm": 0.40641591602818083, "learning_rate": 0.003, "loss": 4.2459, "step": 1738 }, { "epoch": 0.01739, "grad_norm": 0.4886770308383728, "learning_rate": 0.003, "loss": 4.2634, "step": 1739 }, { "epoch": 0.0174, "grad_norm": 0.5760188069320612, "learning_rate": 0.003, "loss": 4.2625, "step": 1740 }, { "epoch": 0.01741, "grad_norm": 0.7342918297652692, "learning_rate": 0.003, "loss": 4.2751, "step": 1741 }, { "epoch": 0.01742, "grad_norm": 0.9752937691555351, "learning_rate": 0.003, "loss": 4.2762, "step": 1742 }, { "epoch": 0.01743, "grad_norm": 1.253306135016296, "learning_rate": 0.003, "loss": 4.3051, "step": 1743 }, { "epoch": 0.01744, "grad_norm": 0.7172428934954133, "learning_rate": 0.003, "loss": 4.2682, "step": 1744 }, { "epoch": 0.01745, "grad_norm": 0.7979588075730017, "learning_rate": 0.003, "loss": 4.2486, "step": 1745 }, { "epoch": 0.01746, "grad_norm": 0.7015732940358177, "learning_rate": 0.003, "loss": 4.2632, "step": 1746 }, { "epoch": 0.01747, "grad_norm": 0.5874159188966563, "learning_rate": 0.003, "loss": 4.2593, "step": 1747 }, { "epoch": 0.01748, "grad_norm": 0.6008438864541681, "learning_rate": 0.003, "loss": 4.2481, "step": 1748 }, { "epoch": 0.01749, "grad_norm": 0.720103709774815, "learning_rate": 0.003, "loss": 4.2531, "step": 1749 }, { "epoch": 0.0175, "grad_norm": 0.7257456079108818, "learning_rate": 0.003, "loss": 4.2707, "step": 1750 }, { "epoch": 0.01751, "grad_norm": 0.6976887398900127, "learning_rate": 0.003, "loss": 4.2594, "step": 1751 }, { "epoch": 0.01752, "grad_norm": 0.7240389649357895, "learning_rate": 0.003, "loss": 4.3015, "step": 1752 }, { "epoch": 0.01753, "grad_norm": 0.7475175322353548, "learning_rate": 0.003, "loss": 4.2524, "step": 1753 }, { "epoch": 0.01754, "grad_norm": 0.7135161794547551, "learning_rate": 0.003, "loss": 4.283, "step": 1754 }, { "epoch": 0.01755, "grad_norm": 0.6772806399058181, "learning_rate": 0.003, "loss": 4.2234, "step": 1755 }, { "epoch": 0.01756, "grad_norm": 0.5967370744888761, "learning_rate": 0.003, "loss": 4.2978, "step": 1756 }, { "epoch": 0.01757, "grad_norm": 0.6301290522164216, "learning_rate": 0.003, "loss": 4.2774, "step": 1757 }, { "epoch": 0.01758, "grad_norm": 0.6353245600564023, "learning_rate": 0.003, "loss": 4.2822, "step": 1758 }, { "epoch": 0.01759, "grad_norm": 0.7049975443178684, "learning_rate": 0.003, "loss": 4.254, "step": 1759 }, { "epoch": 0.0176, "grad_norm": 0.8640938687611407, "learning_rate": 0.003, "loss": 4.2815, "step": 1760 }, { "epoch": 0.01761, "grad_norm": 0.9711991589720537, "learning_rate": 0.003, "loss": 4.2777, "step": 1761 }, { "epoch": 0.01762, "grad_norm": 1.005290948352538, "learning_rate": 0.003, "loss": 4.2996, "step": 1762 }, { "epoch": 0.01763, "grad_norm": 0.8102250953399133, "learning_rate": 0.003, "loss": 4.2897, "step": 1763 }, { "epoch": 0.01764, "grad_norm": 0.6930668584086521, "learning_rate": 0.003, "loss": 4.255, "step": 1764 }, { "epoch": 0.01765, "grad_norm": 0.7660300448887684, "learning_rate": 0.003, "loss": 4.2792, "step": 1765 }, { "epoch": 0.01766, "grad_norm": 0.7821227683200507, "learning_rate": 0.003, "loss": 4.2929, "step": 1766 }, { "epoch": 0.01767, "grad_norm": 0.7890818722807966, "learning_rate": 0.003, "loss": 4.2574, "step": 1767 }, { "epoch": 0.01768, "grad_norm": 0.6779554463169366, "learning_rate": 0.003, "loss": 4.2602, "step": 1768 }, { "epoch": 0.01769, "grad_norm": 0.5803675139310225, "learning_rate": 0.003, "loss": 4.2549, "step": 1769 }, { "epoch": 0.0177, "grad_norm": 0.4980219429490106, "learning_rate": 0.003, "loss": 4.2387, "step": 1770 }, { "epoch": 0.01771, "grad_norm": 0.5000355123582322, "learning_rate": 0.003, "loss": 4.2683, "step": 1771 }, { "epoch": 0.01772, "grad_norm": 0.5357025492673388, "learning_rate": 0.003, "loss": 4.2422, "step": 1772 }, { "epoch": 0.01773, "grad_norm": 0.6531498662416136, "learning_rate": 0.003, "loss": 4.238, "step": 1773 }, { "epoch": 0.01774, "grad_norm": 0.6353732229323361, "learning_rate": 0.003, "loss": 4.26, "step": 1774 }, { "epoch": 0.01775, "grad_norm": 0.5393954118207904, "learning_rate": 0.003, "loss": 4.2437, "step": 1775 }, { "epoch": 0.01776, "grad_norm": 0.5278117971354394, "learning_rate": 0.003, "loss": 4.234, "step": 1776 }, { "epoch": 0.01777, "grad_norm": 0.5939469450707184, "learning_rate": 0.003, "loss": 4.2579, "step": 1777 }, { "epoch": 0.01778, "grad_norm": 0.647641685256514, "learning_rate": 0.003, "loss": 4.2506, "step": 1778 }, { "epoch": 0.01779, "grad_norm": 0.8446707979262026, "learning_rate": 0.003, "loss": 4.2758, "step": 1779 }, { "epoch": 0.0178, "grad_norm": 1.0570404395411184, "learning_rate": 0.003, "loss": 4.2552, "step": 1780 }, { "epoch": 0.01781, "grad_norm": 0.9787736640469828, "learning_rate": 0.003, "loss": 4.2815, "step": 1781 }, { "epoch": 0.01782, "grad_norm": 0.847524414138198, "learning_rate": 0.003, "loss": 4.2954, "step": 1782 }, { "epoch": 0.01783, "grad_norm": 0.9639344454548374, "learning_rate": 0.003, "loss": 4.2851, "step": 1783 }, { "epoch": 0.01784, "grad_norm": 1.0002151519491012, "learning_rate": 0.003, "loss": 4.293, "step": 1784 }, { "epoch": 0.01785, "grad_norm": 0.8702738147646256, "learning_rate": 0.003, "loss": 4.2479, "step": 1785 }, { "epoch": 0.01786, "grad_norm": 0.9523430146284131, "learning_rate": 0.003, "loss": 4.2832, "step": 1786 }, { "epoch": 0.01787, "grad_norm": 0.9420856301640329, "learning_rate": 0.003, "loss": 4.298, "step": 1787 }, { "epoch": 0.01788, "grad_norm": 0.9591055635668716, "learning_rate": 0.003, "loss": 4.3053, "step": 1788 }, { "epoch": 0.01789, "grad_norm": 1.102256290411044, "learning_rate": 0.003, "loss": 4.2935, "step": 1789 }, { "epoch": 0.0179, "grad_norm": 0.7277732179516643, "learning_rate": 0.003, "loss": 4.2907, "step": 1790 }, { "epoch": 0.01791, "grad_norm": 0.6846616583850235, "learning_rate": 0.003, "loss": 4.3029, "step": 1791 }, { "epoch": 0.01792, "grad_norm": 0.6106358124886894, "learning_rate": 0.003, "loss": 4.2999, "step": 1792 }, { "epoch": 0.01793, "grad_norm": 0.576370585363149, "learning_rate": 0.003, "loss": 4.2594, "step": 1793 }, { "epoch": 0.01794, "grad_norm": 0.5513804110166618, "learning_rate": 0.003, "loss": 4.2731, "step": 1794 }, { "epoch": 0.01795, "grad_norm": 0.5576782681087106, "learning_rate": 0.003, "loss": 4.2693, "step": 1795 }, { "epoch": 0.01796, "grad_norm": 0.6017873787216107, "learning_rate": 0.003, "loss": 4.285, "step": 1796 }, { "epoch": 0.01797, "grad_norm": 0.5615285064368845, "learning_rate": 0.003, "loss": 4.245, "step": 1797 }, { "epoch": 0.01798, "grad_norm": 0.5947488985532953, "learning_rate": 0.003, "loss": 4.2588, "step": 1798 }, { "epoch": 0.01799, "grad_norm": 0.6522379824680304, "learning_rate": 0.003, "loss": 4.2758, "step": 1799 }, { "epoch": 0.018, "grad_norm": 0.6500825178193054, "learning_rate": 0.003, "loss": 4.2646, "step": 1800 }, { "epoch": 0.01801, "grad_norm": 0.5651900712619045, "learning_rate": 0.003, "loss": 4.2555, "step": 1801 }, { "epoch": 0.01802, "grad_norm": 0.519529158537289, "learning_rate": 0.003, "loss": 4.2696, "step": 1802 }, { "epoch": 0.01803, "grad_norm": 0.46754842901397486, "learning_rate": 0.003, "loss": 4.2723, "step": 1803 }, { "epoch": 0.01804, "grad_norm": 0.398959720777888, "learning_rate": 0.003, "loss": 4.2421, "step": 1804 }, { "epoch": 0.01805, "grad_norm": 0.3895139350920711, "learning_rate": 0.003, "loss": 4.2599, "step": 1805 }, { "epoch": 0.01806, "grad_norm": 0.4525813527222224, "learning_rate": 0.003, "loss": 4.2195, "step": 1806 }, { "epoch": 0.01807, "grad_norm": 0.5410650860568182, "learning_rate": 0.003, "loss": 4.2567, "step": 1807 }, { "epoch": 0.01808, "grad_norm": 0.7496417695037725, "learning_rate": 0.003, "loss": 4.2549, "step": 1808 }, { "epoch": 0.01809, "grad_norm": 1.0581291305731644, "learning_rate": 0.003, "loss": 4.2876, "step": 1809 }, { "epoch": 0.0181, "grad_norm": 0.9128142976079175, "learning_rate": 0.003, "loss": 4.2835, "step": 1810 }, { "epoch": 0.01811, "grad_norm": 0.5801170839897873, "learning_rate": 0.003, "loss": 4.255, "step": 1811 }, { "epoch": 0.01812, "grad_norm": 0.7987219111431881, "learning_rate": 0.003, "loss": 4.246, "step": 1812 }, { "epoch": 0.01813, "grad_norm": 0.7650732236875453, "learning_rate": 0.003, "loss": 4.2961, "step": 1813 }, { "epoch": 0.01814, "grad_norm": 0.54823874278624, "learning_rate": 0.003, "loss": 4.2509, "step": 1814 }, { "epoch": 0.01815, "grad_norm": 0.6377545297521496, "learning_rate": 0.003, "loss": 4.2406, "step": 1815 }, { "epoch": 0.01816, "grad_norm": 0.6497702590778472, "learning_rate": 0.003, "loss": 4.2694, "step": 1816 }, { "epoch": 0.01817, "grad_norm": 0.491025518180584, "learning_rate": 0.003, "loss": 4.252, "step": 1817 }, { "epoch": 0.01818, "grad_norm": 0.4902524212628076, "learning_rate": 0.003, "loss": 4.2511, "step": 1818 }, { "epoch": 0.01819, "grad_norm": 0.5084432143761608, "learning_rate": 0.003, "loss": 4.2348, "step": 1819 }, { "epoch": 0.0182, "grad_norm": 0.501595173107435, "learning_rate": 0.003, "loss": 4.2533, "step": 1820 }, { "epoch": 0.01821, "grad_norm": 0.46978349716766427, "learning_rate": 0.003, "loss": 4.2375, "step": 1821 }, { "epoch": 0.01822, "grad_norm": 0.48716840432846525, "learning_rate": 0.003, "loss": 4.2508, "step": 1822 }, { "epoch": 0.01823, "grad_norm": 0.5021205527055153, "learning_rate": 0.003, "loss": 4.2638, "step": 1823 }, { "epoch": 0.01824, "grad_norm": 0.46836488848103075, "learning_rate": 0.003, "loss": 4.2464, "step": 1824 }, { "epoch": 0.01825, "grad_norm": 0.47295661191938854, "learning_rate": 0.003, "loss": 4.2499, "step": 1825 }, { "epoch": 0.01826, "grad_norm": 0.49963653178248174, "learning_rate": 0.003, "loss": 4.2612, "step": 1826 }, { "epoch": 0.01827, "grad_norm": 0.5766246258857382, "learning_rate": 0.003, "loss": 4.2338, "step": 1827 }, { "epoch": 0.01828, "grad_norm": 0.6974117711131345, "learning_rate": 0.003, "loss": 4.2498, "step": 1828 }, { "epoch": 0.01829, "grad_norm": 0.920645288162959, "learning_rate": 0.003, "loss": 4.274, "step": 1829 }, { "epoch": 0.0183, "grad_norm": 1.0716665979253641, "learning_rate": 0.003, "loss": 4.2835, "step": 1830 }, { "epoch": 0.01831, "grad_norm": 0.7899385717194873, "learning_rate": 0.003, "loss": 4.263, "step": 1831 }, { "epoch": 0.01832, "grad_norm": 0.6909653405865042, "learning_rate": 0.003, "loss": 4.2522, "step": 1832 }, { "epoch": 0.01833, "grad_norm": 0.7631451682534196, "learning_rate": 0.003, "loss": 4.292, "step": 1833 }, { "epoch": 0.01834, "grad_norm": 0.8518021198547727, "learning_rate": 0.003, "loss": 4.2616, "step": 1834 }, { "epoch": 0.01835, "grad_norm": 0.9608680856911266, "learning_rate": 0.003, "loss": 4.2683, "step": 1835 }, { "epoch": 0.01836, "grad_norm": 0.9370565424064707, "learning_rate": 0.003, "loss": 4.2765, "step": 1836 }, { "epoch": 0.01837, "grad_norm": 0.8673926076902538, "learning_rate": 0.003, "loss": 4.2749, "step": 1837 }, { "epoch": 0.01838, "grad_norm": 0.8912993490476474, "learning_rate": 0.003, "loss": 4.2839, "step": 1838 }, { "epoch": 0.01839, "grad_norm": 0.8988310733902859, "learning_rate": 0.003, "loss": 4.3084, "step": 1839 }, { "epoch": 0.0184, "grad_norm": 0.7369259574653209, "learning_rate": 0.003, "loss": 4.2963, "step": 1840 }, { "epoch": 0.01841, "grad_norm": 0.8862375704705813, "learning_rate": 0.003, "loss": 4.2485, "step": 1841 }, { "epoch": 0.01842, "grad_norm": 0.887067457430843, "learning_rate": 0.003, "loss": 4.2772, "step": 1842 }, { "epoch": 0.01843, "grad_norm": 1.0320986494734232, "learning_rate": 0.003, "loss": 4.2654, "step": 1843 }, { "epoch": 0.01844, "grad_norm": 1.0361956597350352, "learning_rate": 0.003, "loss": 4.2646, "step": 1844 }, { "epoch": 0.01845, "grad_norm": 1.0400807778898855, "learning_rate": 0.003, "loss": 4.2766, "step": 1845 }, { "epoch": 0.01846, "grad_norm": 0.8263035202069795, "learning_rate": 0.003, "loss": 4.2922, "step": 1846 }, { "epoch": 0.01847, "grad_norm": 0.9419440433358707, "learning_rate": 0.003, "loss": 4.2908, "step": 1847 }, { "epoch": 0.01848, "grad_norm": 1.106935925830779, "learning_rate": 0.003, "loss": 4.3123, "step": 1848 }, { "epoch": 0.01849, "grad_norm": 0.9994394701361481, "learning_rate": 0.003, "loss": 4.3126, "step": 1849 }, { "epoch": 0.0185, "grad_norm": 1.0255088552432885, "learning_rate": 0.003, "loss": 4.3054, "step": 1850 }, { "epoch": 0.01851, "grad_norm": 0.7784174512435112, "learning_rate": 0.003, "loss": 4.2831, "step": 1851 }, { "epoch": 0.01852, "grad_norm": 0.6916717444333976, "learning_rate": 0.003, "loss": 4.278, "step": 1852 }, { "epoch": 0.01853, "grad_norm": 0.6721602041720395, "learning_rate": 0.003, "loss": 4.2766, "step": 1853 }, { "epoch": 0.01854, "grad_norm": 0.7396050760038246, "learning_rate": 0.003, "loss": 4.3052, "step": 1854 }, { "epoch": 0.01855, "grad_norm": 0.6847507864796065, "learning_rate": 0.003, "loss": 4.2912, "step": 1855 }, { "epoch": 0.01856, "grad_norm": 0.5367217650881859, "learning_rate": 0.003, "loss": 4.2681, "step": 1856 }, { "epoch": 0.01857, "grad_norm": 0.44508328180737716, "learning_rate": 0.003, "loss": 4.2464, "step": 1857 }, { "epoch": 0.01858, "grad_norm": 0.5010417067873371, "learning_rate": 0.003, "loss": 4.2561, "step": 1858 }, { "epoch": 0.01859, "grad_norm": 0.5235096490961759, "learning_rate": 0.003, "loss": 4.249, "step": 1859 }, { "epoch": 0.0186, "grad_norm": 0.5541997655191133, "learning_rate": 0.003, "loss": 4.2298, "step": 1860 }, { "epoch": 0.01861, "grad_norm": 0.5732056681109601, "learning_rate": 0.003, "loss": 4.2735, "step": 1861 }, { "epoch": 0.01862, "grad_norm": 0.5224594796450702, "learning_rate": 0.003, "loss": 4.2366, "step": 1862 }, { "epoch": 0.01863, "grad_norm": 0.5478228224050402, "learning_rate": 0.003, "loss": 4.2429, "step": 1863 }, { "epoch": 0.01864, "grad_norm": 0.5857768503913141, "learning_rate": 0.003, "loss": 4.2761, "step": 1864 }, { "epoch": 0.01865, "grad_norm": 0.6587143054588716, "learning_rate": 0.003, "loss": 4.2521, "step": 1865 }, { "epoch": 0.01866, "grad_norm": 0.7633854583380161, "learning_rate": 0.003, "loss": 4.2576, "step": 1866 }, { "epoch": 0.01867, "grad_norm": 0.8083803725891301, "learning_rate": 0.003, "loss": 4.2353, "step": 1867 }, { "epoch": 0.01868, "grad_norm": 0.6884248738266137, "learning_rate": 0.003, "loss": 4.25, "step": 1868 }, { "epoch": 0.01869, "grad_norm": 0.6947172674027756, "learning_rate": 0.003, "loss": 4.262, "step": 1869 }, { "epoch": 0.0187, "grad_norm": 0.6248124486931457, "learning_rate": 0.003, "loss": 4.2526, "step": 1870 }, { "epoch": 0.01871, "grad_norm": 0.5753685708476268, "learning_rate": 0.003, "loss": 4.2255, "step": 1871 }, { "epoch": 0.01872, "grad_norm": 0.549064563357354, "learning_rate": 0.003, "loss": 4.269, "step": 1872 }, { "epoch": 0.01873, "grad_norm": 0.5405289107554082, "learning_rate": 0.003, "loss": 4.2299, "step": 1873 }, { "epoch": 0.01874, "grad_norm": 0.5900459025774544, "learning_rate": 0.003, "loss": 4.2352, "step": 1874 }, { "epoch": 0.01875, "grad_norm": 0.6830694870998308, "learning_rate": 0.003, "loss": 4.2563, "step": 1875 }, { "epoch": 0.01876, "grad_norm": 0.7610298506439436, "learning_rate": 0.003, "loss": 4.265, "step": 1876 }, { "epoch": 0.01877, "grad_norm": 0.7821612711704626, "learning_rate": 0.003, "loss": 4.2702, "step": 1877 }, { "epoch": 0.01878, "grad_norm": 0.733027236852046, "learning_rate": 0.003, "loss": 4.2525, "step": 1878 }, { "epoch": 0.01879, "grad_norm": 0.7522348111794812, "learning_rate": 0.003, "loss": 4.2356, "step": 1879 }, { "epoch": 0.0188, "grad_norm": 0.8189114551871165, "learning_rate": 0.003, "loss": 4.252, "step": 1880 }, { "epoch": 0.01881, "grad_norm": 0.7856053328002562, "learning_rate": 0.003, "loss": 4.2461, "step": 1881 }, { "epoch": 0.01882, "grad_norm": 0.7957940195866625, "learning_rate": 0.003, "loss": 4.2764, "step": 1882 }, { "epoch": 0.01883, "grad_norm": 0.702292744282797, "learning_rate": 0.003, "loss": 4.2466, "step": 1883 }, { "epoch": 0.01884, "grad_norm": 0.7056698003000252, "learning_rate": 0.003, "loss": 4.2474, "step": 1884 }, { "epoch": 0.01885, "grad_norm": 0.7144737169743209, "learning_rate": 0.003, "loss": 4.2257, "step": 1885 }, { "epoch": 0.01886, "grad_norm": 0.8271097653458277, "learning_rate": 0.003, "loss": 4.2677, "step": 1886 }, { "epoch": 0.01887, "grad_norm": 0.8342710620866813, "learning_rate": 0.003, "loss": 4.2611, "step": 1887 }, { "epoch": 0.01888, "grad_norm": 0.734160166098297, "learning_rate": 0.003, "loss": 4.2419, "step": 1888 }, { "epoch": 0.01889, "grad_norm": 0.6217950481881471, "learning_rate": 0.003, "loss": 4.2534, "step": 1889 }, { "epoch": 0.0189, "grad_norm": 0.5252786175682846, "learning_rate": 0.003, "loss": 4.2284, "step": 1890 }, { "epoch": 0.01891, "grad_norm": 0.506921167946394, "learning_rate": 0.003, "loss": 4.2769, "step": 1891 }, { "epoch": 0.01892, "grad_norm": 0.5754872458989723, "learning_rate": 0.003, "loss": 4.2536, "step": 1892 }, { "epoch": 0.01893, "grad_norm": 0.6096816763681697, "learning_rate": 0.003, "loss": 4.2344, "step": 1893 }, { "epoch": 0.01894, "grad_norm": 0.5451335732215307, "learning_rate": 0.003, "loss": 4.239, "step": 1894 }, { "epoch": 0.01895, "grad_norm": 0.5874969024823826, "learning_rate": 0.003, "loss": 4.2256, "step": 1895 }, { "epoch": 0.01896, "grad_norm": 0.5969294645671555, "learning_rate": 0.003, "loss": 4.2484, "step": 1896 }, { "epoch": 0.01897, "grad_norm": 0.6886300551222337, "learning_rate": 0.003, "loss": 4.251, "step": 1897 }, { "epoch": 0.01898, "grad_norm": 0.7942601312190114, "learning_rate": 0.003, "loss": 4.225, "step": 1898 }, { "epoch": 0.01899, "grad_norm": 0.7734115484923887, "learning_rate": 0.003, "loss": 4.2391, "step": 1899 }, { "epoch": 0.019, "grad_norm": 0.6488046036081325, "learning_rate": 0.003, "loss": 4.2302, "step": 1900 }, { "epoch": 0.01901, "grad_norm": 0.5904392009872547, "learning_rate": 0.003, "loss": 4.274, "step": 1901 }, { "epoch": 0.01902, "grad_norm": 0.659409531187468, "learning_rate": 0.003, "loss": 4.2537, "step": 1902 }, { "epoch": 0.01903, "grad_norm": 0.6235497653402786, "learning_rate": 0.003, "loss": 4.2072, "step": 1903 }, { "epoch": 0.01904, "grad_norm": 0.5825245241241579, "learning_rate": 0.003, "loss": 4.262, "step": 1904 }, { "epoch": 0.01905, "grad_norm": 0.5746731920133754, "learning_rate": 0.003, "loss": 4.2376, "step": 1905 }, { "epoch": 0.01906, "grad_norm": 0.589593122182735, "learning_rate": 0.003, "loss": 4.2554, "step": 1906 }, { "epoch": 0.01907, "grad_norm": 0.5928254576730478, "learning_rate": 0.003, "loss": 4.2503, "step": 1907 }, { "epoch": 0.01908, "grad_norm": 0.6382953184353721, "learning_rate": 0.003, "loss": 4.2621, "step": 1908 }, { "epoch": 0.01909, "grad_norm": 0.6563665705772666, "learning_rate": 0.003, "loss": 4.2294, "step": 1909 }, { "epoch": 0.0191, "grad_norm": 0.6240866408195912, "learning_rate": 0.003, "loss": 4.2393, "step": 1910 }, { "epoch": 0.01911, "grad_norm": 0.7154059568811579, "learning_rate": 0.003, "loss": 4.2201, "step": 1911 }, { "epoch": 0.01912, "grad_norm": 0.7017374192608556, "learning_rate": 0.003, "loss": 4.229, "step": 1912 }, { "epoch": 0.01913, "grad_norm": 0.8305634249070047, "learning_rate": 0.003, "loss": 4.2505, "step": 1913 }, { "epoch": 0.01914, "grad_norm": 1.0869820826090637, "learning_rate": 0.003, "loss": 4.2792, "step": 1914 }, { "epoch": 0.01915, "grad_norm": 0.9088319848025848, "learning_rate": 0.003, "loss": 4.2668, "step": 1915 }, { "epoch": 0.01916, "grad_norm": 0.7254253941299005, "learning_rate": 0.003, "loss": 4.2403, "step": 1916 }, { "epoch": 0.01917, "grad_norm": 0.7115705570447282, "learning_rate": 0.003, "loss": 4.2636, "step": 1917 }, { "epoch": 0.01918, "grad_norm": 0.7162193279368824, "learning_rate": 0.003, "loss": 4.2488, "step": 1918 }, { "epoch": 0.01919, "grad_norm": 0.7467289060307204, "learning_rate": 0.003, "loss": 4.2482, "step": 1919 }, { "epoch": 0.0192, "grad_norm": 0.7103235241692503, "learning_rate": 0.003, "loss": 4.2521, "step": 1920 }, { "epoch": 0.01921, "grad_norm": 0.6854531851275022, "learning_rate": 0.003, "loss": 4.2411, "step": 1921 }, { "epoch": 0.01922, "grad_norm": 0.6161601419108617, "learning_rate": 0.003, "loss": 4.2512, "step": 1922 }, { "epoch": 0.01923, "grad_norm": 0.5835969123302268, "learning_rate": 0.003, "loss": 4.2556, "step": 1923 }, { "epoch": 0.01924, "grad_norm": 0.5301507132316713, "learning_rate": 0.003, "loss": 4.2545, "step": 1924 }, { "epoch": 0.01925, "grad_norm": 0.5123517463043641, "learning_rate": 0.003, "loss": 4.2425, "step": 1925 }, { "epoch": 0.01926, "grad_norm": 0.515823650813431, "learning_rate": 0.003, "loss": 4.2482, "step": 1926 }, { "epoch": 0.01927, "grad_norm": 0.5134711460859718, "learning_rate": 0.003, "loss": 4.2381, "step": 1927 }, { "epoch": 0.01928, "grad_norm": 0.5394990415668017, "learning_rate": 0.003, "loss": 4.2124, "step": 1928 }, { "epoch": 0.01929, "grad_norm": 0.600924196245884, "learning_rate": 0.003, "loss": 4.2341, "step": 1929 }, { "epoch": 0.0193, "grad_norm": 0.5841670599595433, "learning_rate": 0.003, "loss": 4.222, "step": 1930 }, { "epoch": 0.01931, "grad_norm": 0.6141351653971785, "learning_rate": 0.003, "loss": 4.2269, "step": 1931 }, { "epoch": 0.01932, "grad_norm": 0.630171190249486, "learning_rate": 0.003, "loss": 4.2357, "step": 1932 }, { "epoch": 0.01933, "grad_norm": 0.6348706668276403, "learning_rate": 0.003, "loss": 4.2241, "step": 1933 }, { "epoch": 0.01934, "grad_norm": 0.7486061864959305, "learning_rate": 0.003, "loss": 4.2368, "step": 1934 }, { "epoch": 0.01935, "grad_norm": 0.837656099844549, "learning_rate": 0.003, "loss": 4.2566, "step": 1935 }, { "epoch": 0.01936, "grad_norm": 1.235132295685672, "learning_rate": 0.003, "loss": 4.2553, "step": 1936 }, { "epoch": 0.01937, "grad_norm": 1.1952922931590582, "learning_rate": 0.003, "loss": 4.2793, "step": 1937 }, { "epoch": 0.01938, "grad_norm": 0.8545368063846817, "learning_rate": 0.003, "loss": 4.2672, "step": 1938 }, { "epoch": 0.01939, "grad_norm": 0.8693625520829354, "learning_rate": 0.003, "loss": 4.2945, "step": 1939 }, { "epoch": 0.0194, "grad_norm": 0.9374945609912207, "learning_rate": 0.003, "loss": 4.2792, "step": 1940 }, { "epoch": 0.01941, "grad_norm": 0.8417804839614496, "learning_rate": 0.003, "loss": 4.2519, "step": 1941 }, { "epoch": 0.01942, "grad_norm": 0.6835551217455754, "learning_rate": 0.003, "loss": 4.2765, "step": 1942 }, { "epoch": 0.01943, "grad_norm": 0.6326583755464011, "learning_rate": 0.003, "loss": 4.247, "step": 1943 }, { "epoch": 0.01944, "grad_norm": 0.6336766610195198, "learning_rate": 0.003, "loss": 4.2845, "step": 1944 }, { "epoch": 0.01945, "grad_norm": 0.7023249661919384, "learning_rate": 0.003, "loss": 4.2549, "step": 1945 }, { "epoch": 0.01946, "grad_norm": 0.7532289611006365, "learning_rate": 0.003, "loss": 4.2536, "step": 1946 }, { "epoch": 0.01947, "grad_norm": 0.7241899675160898, "learning_rate": 0.003, "loss": 4.262, "step": 1947 }, { "epoch": 0.01948, "grad_norm": 0.6630467127195124, "learning_rate": 0.003, "loss": 4.2785, "step": 1948 }, { "epoch": 0.01949, "grad_norm": 0.7074334475974129, "learning_rate": 0.003, "loss": 4.2694, "step": 1949 }, { "epoch": 0.0195, "grad_norm": 0.6466673056304594, "learning_rate": 0.003, "loss": 4.2571, "step": 1950 }, { "epoch": 0.01951, "grad_norm": 0.654263769555975, "learning_rate": 0.003, "loss": 4.2499, "step": 1951 }, { "epoch": 0.01952, "grad_norm": 0.5576275389008273, "learning_rate": 0.003, "loss": 4.2552, "step": 1952 }, { "epoch": 0.01953, "grad_norm": 0.5279513641785941, "learning_rate": 0.003, "loss": 4.2571, "step": 1953 }, { "epoch": 0.01954, "grad_norm": 0.6435233353308943, "learning_rate": 0.003, "loss": 4.2595, "step": 1954 }, { "epoch": 0.01955, "grad_norm": 0.9239174746697658, "learning_rate": 0.003, "loss": 4.2611, "step": 1955 }, { "epoch": 0.01956, "grad_norm": 1.1595233997823533, "learning_rate": 0.003, "loss": 4.2763, "step": 1956 }, { "epoch": 0.01957, "grad_norm": 0.6293751148089686, "learning_rate": 0.003, "loss": 4.2223, "step": 1957 }, { "epoch": 0.01958, "grad_norm": 0.638290484912288, "learning_rate": 0.003, "loss": 4.2174, "step": 1958 }, { "epoch": 0.01959, "grad_norm": 0.7260665225455007, "learning_rate": 0.003, "loss": 4.245, "step": 1959 }, { "epoch": 0.0196, "grad_norm": 0.6163944682591169, "learning_rate": 0.003, "loss": 4.232, "step": 1960 }, { "epoch": 0.01961, "grad_norm": 0.6978053518640758, "learning_rate": 0.003, "loss": 4.236, "step": 1961 }, { "epoch": 0.01962, "grad_norm": 0.6587488710703514, "learning_rate": 0.003, "loss": 4.2366, "step": 1962 }, { "epoch": 0.01963, "grad_norm": 0.5260594263866168, "learning_rate": 0.003, "loss": 4.2173, "step": 1963 }, { "epoch": 0.01964, "grad_norm": 0.490485720862937, "learning_rate": 0.003, "loss": 4.2458, "step": 1964 }, { "epoch": 0.01965, "grad_norm": 0.45667875143823883, "learning_rate": 0.003, "loss": 4.2346, "step": 1965 }, { "epoch": 0.01966, "grad_norm": 0.5549508938343672, "learning_rate": 0.003, "loss": 4.2386, "step": 1966 }, { "epoch": 0.01967, "grad_norm": 0.5947105420816633, "learning_rate": 0.003, "loss": 4.2168, "step": 1967 }, { "epoch": 0.01968, "grad_norm": 0.7273871579402625, "learning_rate": 0.003, "loss": 4.2427, "step": 1968 }, { "epoch": 0.01969, "grad_norm": 0.8131307161709899, "learning_rate": 0.003, "loss": 4.2447, "step": 1969 }, { "epoch": 0.0197, "grad_norm": 0.8153412941444168, "learning_rate": 0.003, "loss": 4.242, "step": 1970 }, { "epoch": 0.01971, "grad_norm": 0.7076673538361063, "learning_rate": 0.003, "loss": 4.2271, "step": 1971 }, { "epoch": 0.01972, "grad_norm": 0.6697097595169021, "learning_rate": 0.003, "loss": 4.2266, "step": 1972 }, { "epoch": 0.01973, "grad_norm": 0.6208505648120144, "learning_rate": 0.003, "loss": 4.2061, "step": 1973 }, { "epoch": 0.01974, "grad_norm": 0.6078249030800872, "learning_rate": 0.003, "loss": 4.2469, "step": 1974 }, { "epoch": 0.01975, "grad_norm": 0.6193855466160911, "learning_rate": 0.003, "loss": 4.2383, "step": 1975 }, { "epoch": 0.01976, "grad_norm": 0.5723863500156643, "learning_rate": 0.003, "loss": 4.228, "step": 1976 }, { "epoch": 0.01977, "grad_norm": 0.614684480005163, "learning_rate": 0.003, "loss": 4.2636, "step": 1977 }, { "epoch": 0.01978, "grad_norm": 0.8002502395832268, "learning_rate": 0.003, "loss": 4.269, "step": 1978 }, { "epoch": 0.01979, "grad_norm": 1.1734742193181502, "learning_rate": 0.003, "loss": 4.2712, "step": 1979 }, { "epoch": 0.0198, "grad_norm": 1.0542111183452296, "learning_rate": 0.003, "loss": 4.2573, "step": 1980 }, { "epoch": 0.01981, "grad_norm": 0.8526982225261958, "learning_rate": 0.003, "loss": 4.2367, "step": 1981 }, { "epoch": 0.01982, "grad_norm": 0.7282783707767795, "learning_rate": 0.003, "loss": 4.2575, "step": 1982 }, { "epoch": 0.01983, "grad_norm": 0.7406100600030446, "learning_rate": 0.003, "loss": 4.2502, "step": 1983 }, { "epoch": 0.01984, "grad_norm": 0.7725656708208015, "learning_rate": 0.003, "loss": 4.2593, "step": 1984 }, { "epoch": 0.01985, "grad_norm": 0.8754720019719067, "learning_rate": 0.003, "loss": 4.2582, "step": 1985 }, { "epoch": 0.01986, "grad_norm": 0.9542865913273004, "learning_rate": 0.003, "loss": 4.2928, "step": 1986 }, { "epoch": 0.01987, "grad_norm": 0.9042901186857757, "learning_rate": 0.003, "loss": 4.248, "step": 1987 }, { "epoch": 0.01988, "grad_norm": 0.9236965754299921, "learning_rate": 0.003, "loss": 4.2653, "step": 1988 }, { "epoch": 0.01989, "grad_norm": 0.8847485842619162, "learning_rate": 0.003, "loss": 4.2507, "step": 1989 }, { "epoch": 0.0199, "grad_norm": 0.7262740969822487, "learning_rate": 0.003, "loss": 4.2593, "step": 1990 }, { "epoch": 0.01991, "grad_norm": 0.6829209219384283, "learning_rate": 0.003, "loss": 4.2367, "step": 1991 }, { "epoch": 0.01992, "grad_norm": 0.6687499652079453, "learning_rate": 0.003, "loss": 4.2729, "step": 1992 }, { "epoch": 0.01993, "grad_norm": 0.6270120579020493, "learning_rate": 0.003, "loss": 4.2552, "step": 1993 }, { "epoch": 0.01994, "grad_norm": 0.695428830391491, "learning_rate": 0.003, "loss": 4.2763, "step": 1994 }, { "epoch": 0.01995, "grad_norm": 0.6511062075978684, "learning_rate": 0.003, "loss": 4.265, "step": 1995 }, { "epoch": 0.01996, "grad_norm": 0.7071538607579048, "learning_rate": 0.003, "loss": 4.2345, "step": 1996 }, { "epoch": 0.01997, "grad_norm": 0.837055308686826, "learning_rate": 0.003, "loss": 4.2713, "step": 1997 }, { "epoch": 0.01998, "grad_norm": 0.9265767295150127, "learning_rate": 0.003, "loss": 4.2575, "step": 1998 }, { "epoch": 0.01999, "grad_norm": 0.776786308598987, "learning_rate": 0.003, "loss": 4.2456, "step": 1999 }, { "epoch": 0.02, "grad_norm": 0.544115707262644, "learning_rate": 0.003, "loss": 4.223, "step": 2000 }, { "epoch": 0.02001, "grad_norm": 0.5968209168617383, "learning_rate": 0.003, "loss": 4.2587, "step": 2001 }, { "epoch": 0.02002, "grad_norm": 0.6230815008994128, "learning_rate": 0.003, "loss": 4.2573, "step": 2002 }, { "epoch": 0.02003, "grad_norm": 0.6144419617298068, "learning_rate": 0.003, "loss": 4.2517, "step": 2003 }, { "epoch": 0.02004, "grad_norm": 0.5244593563834906, "learning_rate": 0.003, "loss": 4.2348, "step": 2004 }, { "epoch": 0.02005, "grad_norm": 0.4555996828045135, "learning_rate": 0.003, "loss": 4.2382, "step": 2005 }, { "epoch": 0.02006, "grad_norm": 0.47265795693259455, "learning_rate": 0.003, "loss": 4.2356, "step": 2006 }, { "epoch": 0.02007, "grad_norm": 0.5068378835965273, "learning_rate": 0.003, "loss": 4.2513, "step": 2007 }, { "epoch": 0.02008, "grad_norm": 0.5368129535807056, "learning_rate": 0.003, "loss": 4.2399, "step": 2008 }, { "epoch": 0.02009, "grad_norm": 0.6542343628757692, "learning_rate": 0.003, "loss": 4.2391, "step": 2009 }, { "epoch": 0.0201, "grad_norm": 0.7550868705205384, "learning_rate": 0.003, "loss": 4.2546, "step": 2010 }, { "epoch": 0.02011, "grad_norm": 0.8523385764253874, "learning_rate": 0.003, "loss": 4.2454, "step": 2011 }, { "epoch": 0.02012, "grad_norm": 0.7590508053821384, "learning_rate": 0.003, "loss": 4.2437, "step": 2012 }, { "epoch": 0.02013, "grad_norm": 0.6112232485700291, "learning_rate": 0.003, "loss": 4.2454, "step": 2013 }, { "epoch": 0.02014, "grad_norm": 0.580521076473927, "learning_rate": 0.003, "loss": 4.2288, "step": 2014 }, { "epoch": 0.02015, "grad_norm": 0.5972048548713036, "learning_rate": 0.003, "loss": 4.2254, "step": 2015 }, { "epoch": 0.02016, "grad_norm": 0.502339786086644, "learning_rate": 0.003, "loss": 4.228, "step": 2016 }, { "epoch": 0.02017, "grad_norm": 0.5744684235266054, "learning_rate": 0.003, "loss": 4.224, "step": 2017 }, { "epoch": 0.02018, "grad_norm": 0.686457625935553, "learning_rate": 0.003, "loss": 4.2281, "step": 2018 }, { "epoch": 0.02019, "grad_norm": 0.8023792999105707, "learning_rate": 0.003, "loss": 4.2399, "step": 2019 }, { "epoch": 0.0202, "grad_norm": 0.779213027451944, "learning_rate": 0.003, "loss": 4.2469, "step": 2020 }, { "epoch": 0.02021, "grad_norm": 0.747271315164086, "learning_rate": 0.003, "loss": 4.2601, "step": 2021 }, { "epoch": 0.02022, "grad_norm": 0.7712990219385492, "learning_rate": 0.003, "loss": 4.2242, "step": 2022 }, { "epoch": 0.02023, "grad_norm": 0.678848353295381, "learning_rate": 0.003, "loss": 4.1929, "step": 2023 }, { "epoch": 0.02024, "grad_norm": 0.7080112136621078, "learning_rate": 0.003, "loss": 4.2211, "step": 2024 }, { "epoch": 0.02025, "grad_norm": 0.7015387197783911, "learning_rate": 0.003, "loss": 4.2713, "step": 2025 }, { "epoch": 0.02026, "grad_norm": 0.6618325774284712, "learning_rate": 0.003, "loss": 4.249, "step": 2026 }, { "epoch": 0.02027, "grad_norm": 0.6649314977064322, "learning_rate": 0.003, "loss": 4.2493, "step": 2027 }, { "epoch": 0.02028, "grad_norm": 0.7311305035019059, "learning_rate": 0.003, "loss": 4.2336, "step": 2028 }, { "epoch": 0.02029, "grad_norm": 0.829605271368658, "learning_rate": 0.003, "loss": 4.243, "step": 2029 }, { "epoch": 0.0203, "grad_norm": 0.8235570307782024, "learning_rate": 0.003, "loss": 4.2354, "step": 2030 }, { "epoch": 0.02031, "grad_norm": 0.8039697362095677, "learning_rate": 0.003, "loss": 4.2412, "step": 2031 }, { "epoch": 0.02032, "grad_norm": 0.7849326035149696, "learning_rate": 0.003, "loss": 4.2148, "step": 2032 }, { "epoch": 0.02033, "grad_norm": 0.7613884372790184, "learning_rate": 0.003, "loss": 4.2543, "step": 2033 }, { "epoch": 0.02034, "grad_norm": 0.7131747920559885, "learning_rate": 0.003, "loss": 4.2453, "step": 2034 }, { "epoch": 0.02035, "grad_norm": 0.7017452038094153, "learning_rate": 0.003, "loss": 4.2298, "step": 2035 }, { "epoch": 0.02036, "grad_norm": 0.6789439698745199, "learning_rate": 0.003, "loss": 4.232, "step": 2036 }, { "epoch": 0.02037, "grad_norm": 0.5966860547597203, "learning_rate": 0.003, "loss": 4.2423, "step": 2037 }, { "epoch": 0.02038, "grad_norm": 0.5724376516658686, "learning_rate": 0.003, "loss": 4.224, "step": 2038 }, { "epoch": 0.02039, "grad_norm": 0.5467166846835673, "learning_rate": 0.003, "loss": 4.2314, "step": 2039 }, { "epoch": 0.0204, "grad_norm": 0.5471552582468747, "learning_rate": 0.003, "loss": 4.2353, "step": 2040 }, { "epoch": 0.02041, "grad_norm": 0.5793884373224328, "learning_rate": 0.003, "loss": 4.1982, "step": 2041 }, { "epoch": 0.02042, "grad_norm": 0.5725852678050465, "learning_rate": 0.003, "loss": 4.197, "step": 2042 }, { "epoch": 0.02043, "grad_norm": 0.7022295528607846, "learning_rate": 0.003, "loss": 4.2474, "step": 2043 }, { "epoch": 0.02044, "grad_norm": 0.8601243032306309, "learning_rate": 0.003, "loss": 4.2273, "step": 2044 }, { "epoch": 0.02045, "grad_norm": 0.8046253539648219, "learning_rate": 0.003, "loss": 4.2424, "step": 2045 }, { "epoch": 0.02046, "grad_norm": 0.855965931301316, "learning_rate": 0.003, "loss": 4.2237, "step": 2046 }, { "epoch": 0.02047, "grad_norm": 1.0357234424752648, "learning_rate": 0.003, "loss": 4.2697, "step": 2047 }, { "epoch": 0.02048, "grad_norm": 1.0439384854619227, "learning_rate": 0.003, "loss": 4.266, "step": 2048 }, { "epoch": 0.02049, "grad_norm": 0.9412093564870287, "learning_rate": 0.003, "loss": 4.2808, "step": 2049 }, { "epoch": 0.0205, "grad_norm": 0.9243748599866667, "learning_rate": 0.003, "loss": 4.2289, "step": 2050 }, { "epoch": 0.02051, "grad_norm": 0.7552410516608989, "learning_rate": 0.003, "loss": 4.2287, "step": 2051 }, { "epoch": 0.02052, "grad_norm": 0.6501532577962565, "learning_rate": 0.003, "loss": 4.226, "step": 2052 }, { "epoch": 0.02053, "grad_norm": 0.7652062065498313, "learning_rate": 0.003, "loss": 4.2323, "step": 2053 }, { "epoch": 0.02054, "grad_norm": 0.8858538626101129, "learning_rate": 0.003, "loss": 4.2751, "step": 2054 }, { "epoch": 0.02055, "grad_norm": 0.7846832781207133, "learning_rate": 0.003, "loss": 4.231, "step": 2055 }, { "epoch": 0.02056, "grad_norm": 0.7220550070691274, "learning_rate": 0.003, "loss": 4.2248, "step": 2056 }, { "epoch": 0.02057, "grad_norm": 0.7368090739126946, "learning_rate": 0.003, "loss": 4.2673, "step": 2057 }, { "epoch": 0.02058, "grad_norm": 0.7702851748272688, "learning_rate": 0.003, "loss": 4.2441, "step": 2058 }, { "epoch": 0.02059, "grad_norm": 0.6929544007844861, "learning_rate": 0.003, "loss": 4.2219, "step": 2059 }, { "epoch": 0.0206, "grad_norm": 0.6471217940270918, "learning_rate": 0.003, "loss": 4.2645, "step": 2060 }, { "epoch": 0.02061, "grad_norm": 0.7586644703414864, "learning_rate": 0.003, "loss": 4.2388, "step": 2061 }, { "epoch": 0.02062, "grad_norm": 0.8861471216459856, "learning_rate": 0.003, "loss": 4.2462, "step": 2062 }, { "epoch": 0.02063, "grad_norm": 0.9668317520421095, "learning_rate": 0.003, "loss": 4.2463, "step": 2063 }, { "epoch": 0.02064, "grad_norm": 0.9399457261661122, "learning_rate": 0.003, "loss": 4.2579, "step": 2064 }, { "epoch": 0.02065, "grad_norm": 0.8512260541477842, "learning_rate": 0.003, "loss": 4.235, "step": 2065 }, { "epoch": 0.02066, "grad_norm": 0.7854637650895366, "learning_rate": 0.003, "loss": 4.2504, "step": 2066 }, { "epoch": 0.02067, "grad_norm": 0.7442570611327062, "learning_rate": 0.003, "loss": 4.248, "step": 2067 }, { "epoch": 0.02068, "grad_norm": 0.7336263498203625, "learning_rate": 0.003, "loss": 4.2275, "step": 2068 }, { "epoch": 0.02069, "grad_norm": 0.6599353273196071, "learning_rate": 0.003, "loss": 4.2493, "step": 2069 }, { "epoch": 0.0207, "grad_norm": 0.5850836932137713, "learning_rate": 0.003, "loss": 4.2188, "step": 2070 }, { "epoch": 0.02071, "grad_norm": 0.6074733078697232, "learning_rate": 0.003, "loss": 4.2476, "step": 2071 }, { "epoch": 0.02072, "grad_norm": 0.5920803977844451, "learning_rate": 0.003, "loss": 4.2537, "step": 2072 }, { "epoch": 0.02073, "grad_norm": 0.5924717703900239, "learning_rate": 0.003, "loss": 4.2322, "step": 2073 }, { "epoch": 0.02074, "grad_norm": 0.594618787583075, "learning_rate": 0.003, "loss": 4.2368, "step": 2074 }, { "epoch": 0.02075, "grad_norm": 0.5532830540657053, "learning_rate": 0.003, "loss": 4.2603, "step": 2075 }, { "epoch": 0.02076, "grad_norm": 0.6184715370341762, "learning_rate": 0.003, "loss": 4.2559, "step": 2076 }, { "epoch": 0.02077, "grad_norm": 0.6677430190730856, "learning_rate": 0.003, "loss": 4.2214, "step": 2077 }, { "epoch": 0.02078, "grad_norm": 0.8898743231750045, "learning_rate": 0.003, "loss": 4.2253, "step": 2078 }, { "epoch": 0.02079, "grad_norm": 1.1171860016868687, "learning_rate": 0.003, "loss": 4.2251, "step": 2079 }, { "epoch": 0.0208, "grad_norm": 1.0675912571031756, "learning_rate": 0.003, "loss": 4.2358, "step": 2080 }, { "epoch": 0.02081, "grad_norm": 1.02172099105771, "learning_rate": 0.003, "loss": 4.2752, "step": 2081 }, { "epoch": 0.02082, "grad_norm": 0.7358740084680295, "learning_rate": 0.003, "loss": 4.2588, "step": 2082 }, { "epoch": 0.02083, "grad_norm": 0.678950866367781, "learning_rate": 0.003, "loss": 4.2499, "step": 2083 }, { "epoch": 0.02084, "grad_norm": 0.6900443906154086, "learning_rate": 0.003, "loss": 4.2447, "step": 2084 }, { "epoch": 0.02085, "grad_norm": 0.709096344071076, "learning_rate": 0.003, "loss": 4.2198, "step": 2085 }, { "epoch": 0.02086, "grad_norm": 0.6415519097957504, "learning_rate": 0.003, "loss": 4.2227, "step": 2086 }, { "epoch": 0.02087, "grad_norm": 0.6111018129236451, "learning_rate": 0.003, "loss": 4.2218, "step": 2087 }, { "epoch": 0.02088, "grad_norm": 0.5614795365012508, "learning_rate": 0.003, "loss": 4.2531, "step": 2088 }, { "epoch": 0.02089, "grad_norm": 0.4937030612790938, "learning_rate": 0.003, "loss": 4.2175, "step": 2089 }, { "epoch": 0.0209, "grad_norm": 0.5203347838934052, "learning_rate": 0.003, "loss": 4.2372, "step": 2090 }, { "epoch": 0.02091, "grad_norm": 0.5562686236826955, "learning_rate": 0.003, "loss": 4.2263, "step": 2091 }, { "epoch": 0.02092, "grad_norm": 0.6151784723639154, "learning_rate": 0.003, "loss": 4.2246, "step": 2092 }, { "epoch": 0.02093, "grad_norm": 0.5966017113888583, "learning_rate": 0.003, "loss": 4.1906, "step": 2093 }, { "epoch": 0.02094, "grad_norm": 0.5339381970320102, "learning_rate": 0.003, "loss": 4.197, "step": 2094 }, { "epoch": 0.02095, "grad_norm": 0.6298496102722991, "learning_rate": 0.003, "loss": 4.2379, "step": 2095 }, { "epoch": 0.02096, "grad_norm": 0.7946260614902717, "learning_rate": 0.003, "loss": 4.2367, "step": 2096 }, { "epoch": 0.02097, "grad_norm": 0.9470719378143645, "learning_rate": 0.003, "loss": 4.222, "step": 2097 }, { "epoch": 0.02098, "grad_norm": 0.9116485878338217, "learning_rate": 0.003, "loss": 4.2439, "step": 2098 }, { "epoch": 0.02099, "grad_norm": 0.7133877388191158, "learning_rate": 0.003, "loss": 4.2442, "step": 2099 }, { "epoch": 0.021, "grad_norm": 0.6125463461181375, "learning_rate": 0.003, "loss": 4.2549, "step": 2100 }, { "epoch": 0.02101, "grad_norm": 0.6276708108187462, "learning_rate": 0.003, "loss": 4.2467, "step": 2101 }, { "epoch": 0.02102, "grad_norm": 0.76243038603311, "learning_rate": 0.003, "loss": 4.2509, "step": 2102 }, { "epoch": 0.02103, "grad_norm": 0.831814217647559, "learning_rate": 0.003, "loss": 4.2186, "step": 2103 }, { "epoch": 0.02104, "grad_norm": 0.8037622250501507, "learning_rate": 0.003, "loss": 4.223, "step": 2104 }, { "epoch": 0.02105, "grad_norm": 0.773242874233749, "learning_rate": 0.003, "loss": 4.2254, "step": 2105 }, { "epoch": 0.02106, "grad_norm": 0.7723032929932112, "learning_rate": 0.003, "loss": 4.2206, "step": 2106 }, { "epoch": 0.02107, "grad_norm": 0.6777196727394571, "learning_rate": 0.003, "loss": 4.2302, "step": 2107 }, { "epoch": 0.02108, "grad_norm": 0.7155266180317776, "learning_rate": 0.003, "loss": 4.2133, "step": 2108 }, { "epoch": 0.02109, "grad_norm": 0.7307044349882023, "learning_rate": 0.003, "loss": 4.2034, "step": 2109 }, { "epoch": 0.0211, "grad_norm": 0.7246010606158528, "learning_rate": 0.003, "loss": 4.2211, "step": 2110 }, { "epoch": 0.02111, "grad_norm": 0.7003859578773313, "learning_rate": 0.003, "loss": 4.2271, "step": 2111 }, { "epoch": 0.02112, "grad_norm": 0.6571289356027246, "learning_rate": 0.003, "loss": 4.2265, "step": 2112 }, { "epoch": 0.02113, "grad_norm": 0.7142951538736501, "learning_rate": 0.003, "loss": 4.2259, "step": 2113 }, { "epoch": 0.02114, "grad_norm": 0.642746334410043, "learning_rate": 0.003, "loss": 4.2136, "step": 2114 }, { "epoch": 0.02115, "grad_norm": 0.5561928823494658, "learning_rate": 0.003, "loss": 4.2025, "step": 2115 }, { "epoch": 0.02116, "grad_norm": 0.4854911541916316, "learning_rate": 0.003, "loss": 4.2271, "step": 2116 }, { "epoch": 0.02117, "grad_norm": 0.4406562024092029, "learning_rate": 0.003, "loss": 4.2222, "step": 2117 }, { "epoch": 0.02118, "grad_norm": 0.44267808922719293, "learning_rate": 0.003, "loss": 4.2156, "step": 2118 }, { "epoch": 0.02119, "grad_norm": 0.4799600611934466, "learning_rate": 0.003, "loss": 4.2269, "step": 2119 }, { "epoch": 0.0212, "grad_norm": 0.5994897503195796, "learning_rate": 0.003, "loss": 4.202, "step": 2120 }, { "epoch": 0.02121, "grad_norm": 0.8543941081816084, "learning_rate": 0.003, "loss": 4.2042, "step": 2121 }, { "epoch": 0.02122, "grad_norm": 1.1792972986526515, "learning_rate": 0.003, "loss": 4.2347, "step": 2122 }, { "epoch": 0.02123, "grad_norm": 0.6741324159024102, "learning_rate": 0.003, "loss": 4.1979, "step": 2123 }, { "epoch": 0.02124, "grad_norm": 0.5314863327354328, "learning_rate": 0.003, "loss": 4.2178, "step": 2124 }, { "epoch": 0.02125, "grad_norm": 0.7086470592360375, "learning_rate": 0.003, "loss": 4.2389, "step": 2125 }, { "epoch": 0.02126, "grad_norm": 0.7578318446121494, "learning_rate": 0.003, "loss": 4.2324, "step": 2126 }, { "epoch": 0.02127, "grad_norm": 0.8658999413430398, "learning_rate": 0.003, "loss": 4.253, "step": 2127 }, { "epoch": 0.02128, "grad_norm": 0.8986841353197695, "learning_rate": 0.003, "loss": 4.2443, "step": 2128 }, { "epoch": 0.02129, "grad_norm": 0.8385623851205096, "learning_rate": 0.003, "loss": 4.2447, "step": 2129 }, { "epoch": 0.0213, "grad_norm": 0.8355943131013069, "learning_rate": 0.003, "loss": 4.2211, "step": 2130 }, { "epoch": 0.02131, "grad_norm": 0.793529743088508, "learning_rate": 0.003, "loss": 4.2487, "step": 2131 }, { "epoch": 0.02132, "grad_norm": 0.7536314785478804, "learning_rate": 0.003, "loss": 4.2567, "step": 2132 }, { "epoch": 0.02133, "grad_norm": 0.7395002132947835, "learning_rate": 0.003, "loss": 4.2539, "step": 2133 }, { "epoch": 0.02134, "grad_norm": 0.7463724088809484, "learning_rate": 0.003, "loss": 4.2559, "step": 2134 }, { "epoch": 0.02135, "grad_norm": 0.8316464480461356, "learning_rate": 0.003, "loss": 4.2584, "step": 2135 }, { "epoch": 0.02136, "grad_norm": 0.8795824396478255, "learning_rate": 0.003, "loss": 4.2294, "step": 2136 }, { "epoch": 0.02137, "grad_norm": 0.9902402983774369, "learning_rate": 0.003, "loss": 4.2714, "step": 2137 }, { "epoch": 0.02138, "grad_norm": 1.1393734479896327, "learning_rate": 0.003, "loss": 4.2998, "step": 2138 }, { "epoch": 0.02139, "grad_norm": 0.880379491281086, "learning_rate": 0.003, "loss": 4.2799, "step": 2139 }, { "epoch": 0.0214, "grad_norm": 0.7579598710856811, "learning_rate": 0.003, "loss": 4.2687, "step": 2140 }, { "epoch": 0.02141, "grad_norm": 0.7913256313860993, "learning_rate": 0.003, "loss": 4.2459, "step": 2141 }, { "epoch": 0.02142, "grad_norm": 0.8825604009248581, "learning_rate": 0.003, "loss": 4.2418, "step": 2142 }, { "epoch": 0.02143, "grad_norm": 0.9777590043525748, "learning_rate": 0.003, "loss": 4.2642, "step": 2143 }, { "epoch": 0.02144, "grad_norm": 1.0275563743451126, "learning_rate": 0.003, "loss": 4.2678, "step": 2144 }, { "epoch": 0.02145, "grad_norm": 0.7699073494834429, "learning_rate": 0.003, "loss": 4.2427, "step": 2145 }, { "epoch": 0.02146, "grad_norm": 0.5960629563316048, "learning_rate": 0.003, "loss": 4.2482, "step": 2146 }, { "epoch": 0.02147, "grad_norm": 0.5490133168290037, "learning_rate": 0.003, "loss": 4.2644, "step": 2147 }, { "epoch": 0.02148, "grad_norm": 0.5767497961920597, "learning_rate": 0.003, "loss": 4.2739, "step": 2148 }, { "epoch": 0.02149, "grad_norm": 0.4897484785451818, "learning_rate": 0.003, "loss": 4.2537, "step": 2149 }, { "epoch": 0.0215, "grad_norm": 0.45116068678975524, "learning_rate": 0.003, "loss": 4.2384, "step": 2150 }, { "epoch": 0.02151, "grad_norm": 0.37866677504367807, "learning_rate": 0.003, "loss": 4.2406, "step": 2151 }, { "epoch": 0.02152, "grad_norm": 0.35025559255506084, "learning_rate": 0.003, "loss": 4.2149, "step": 2152 }, { "epoch": 0.02153, "grad_norm": 0.34582936725569746, "learning_rate": 0.003, "loss": 4.2226, "step": 2153 }, { "epoch": 0.02154, "grad_norm": 0.3203952559256608, "learning_rate": 0.003, "loss": 4.2119, "step": 2154 }, { "epoch": 0.02155, "grad_norm": 0.29250609362491636, "learning_rate": 0.003, "loss": 4.2391, "step": 2155 }, { "epoch": 0.02156, "grad_norm": 0.3438204060042884, "learning_rate": 0.003, "loss": 4.2258, "step": 2156 }, { "epoch": 0.02157, "grad_norm": 0.3595905001354517, "learning_rate": 0.003, "loss": 4.1905, "step": 2157 }, { "epoch": 0.02158, "grad_norm": 0.42595432724572757, "learning_rate": 0.003, "loss": 4.2116, "step": 2158 }, { "epoch": 0.02159, "grad_norm": 0.6633695244520449, "learning_rate": 0.003, "loss": 4.2281, "step": 2159 }, { "epoch": 0.0216, "grad_norm": 1.007521973270438, "learning_rate": 0.003, "loss": 4.2436, "step": 2160 }, { "epoch": 0.02161, "grad_norm": 1.2248634852939608, "learning_rate": 0.003, "loss": 4.2476, "step": 2161 }, { "epoch": 0.02162, "grad_norm": 0.44924134107129593, "learning_rate": 0.003, "loss": 4.2258, "step": 2162 }, { "epoch": 0.02163, "grad_norm": 0.708237336037978, "learning_rate": 0.003, "loss": 4.2521, "step": 2163 }, { "epoch": 0.02164, "grad_norm": 0.7651720506018448, "learning_rate": 0.003, "loss": 4.2406, "step": 2164 }, { "epoch": 0.02165, "grad_norm": 0.7208522677692027, "learning_rate": 0.003, "loss": 4.2056, "step": 2165 }, { "epoch": 0.02166, "grad_norm": 0.7338853238041586, "learning_rate": 0.003, "loss": 4.248, "step": 2166 }, { "epoch": 0.02167, "grad_norm": 0.7434196157225088, "learning_rate": 0.003, "loss": 4.223, "step": 2167 }, { "epoch": 0.02168, "grad_norm": 0.726421176827236, "learning_rate": 0.003, "loss": 4.2235, "step": 2168 }, { "epoch": 0.02169, "grad_norm": 0.6029539671407806, "learning_rate": 0.003, "loss": 4.2188, "step": 2169 }, { "epoch": 0.0217, "grad_norm": 0.6113346670733881, "learning_rate": 0.003, "loss": 4.2262, "step": 2170 }, { "epoch": 0.02171, "grad_norm": 0.5554372591572285, "learning_rate": 0.003, "loss": 4.2379, "step": 2171 }, { "epoch": 0.02172, "grad_norm": 0.5137481804853822, "learning_rate": 0.003, "loss": 4.2085, "step": 2172 }, { "epoch": 0.02173, "grad_norm": 0.43826435194750174, "learning_rate": 0.003, "loss": 4.2079, "step": 2173 }, { "epoch": 0.02174, "grad_norm": 0.4550674538213562, "learning_rate": 0.003, "loss": 4.2039, "step": 2174 }, { "epoch": 0.02175, "grad_norm": 0.4826492500585534, "learning_rate": 0.003, "loss": 4.2274, "step": 2175 }, { "epoch": 0.02176, "grad_norm": 0.5204683333159428, "learning_rate": 0.003, "loss": 4.2082, "step": 2176 }, { "epoch": 0.02177, "grad_norm": 0.5967890811282951, "learning_rate": 0.003, "loss": 4.2167, "step": 2177 }, { "epoch": 0.02178, "grad_norm": 0.6463619597188142, "learning_rate": 0.003, "loss": 4.1999, "step": 2178 }, { "epoch": 0.02179, "grad_norm": 0.6800531318971829, "learning_rate": 0.003, "loss": 4.2149, "step": 2179 }, { "epoch": 0.0218, "grad_norm": 0.6580527094136631, "learning_rate": 0.003, "loss": 4.229, "step": 2180 }, { "epoch": 0.02181, "grad_norm": 0.6208184897317314, "learning_rate": 0.003, "loss": 4.2289, "step": 2181 }, { "epoch": 0.02182, "grad_norm": 0.5846113298952428, "learning_rate": 0.003, "loss": 4.2148, "step": 2182 }, { "epoch": 0.02183, "grad_norm": 0.6556384450212658, "learning_rate": 0.003, "loss": 4.2271, "step": 2183 }, { "epoch": 0.02184, "grad_norm": 0.8938194731118257, "learning_rate": 0.003, "loss": 4.21, "step": 2184 }, { "epoch": 0.02185, "grad_norm": 1.4521805158873795, "learning_rate": 0.003, "loss": 4.2324, "step": 2185 }, { "epoch": 0.02186, "grad_norm": 0.7715834342950255, "learning_rate": 0.003, "loss": 4.231, "step": 2186 }, { "epoch": 0.02187, "grad_norm": 0.8581699640168193, "learning_rate": 0.003, "loss": 4.2381, "step": 2187 }, { "epoch": 0.02188, "grad_norm": 1.069027806910656, "learning_rate": 0.003, "loss": 4.2275, "step": 2188 }, { "epoch": 0.02189, "grad_norm": 1.0002813983051297, "learning_rate": 0.003, "loss": 4.2597, "step": 2189 }, { "epoch": 0.0219, "grad_norm": 0.9466471656385088, "learning_rate": 0.003, "loss": 4.2521, "step": 2190 }, { "epoch": 0.02191, "grad_norm": 0.8750563192902713, "learning_rate": 0.003, "loss": 4.236, "step": 2191 }, { "epoch": 0.02192, "grad_norm": 0.840341624246557, "learning_rate": 0.003, "loss": 4.2668, "step": 2192 }, { "epoch": 0.02193, "grad_norm": 0.7540813398590741, "learning_rate": 0.003, "loss": 4.2289, "step": 2193 }, { "epoch": 0.02194, "grad_norm": 0.6938742966655788, "learning_rate": 0.003, "loss": 4.2759, "step": 2194 }, { "epoch": 0.02195, "grad_norm": 0.6175768159775354, "learning_rate": 0.003, "loss": 4.2375, "step": 2195 }, { "epoch": 0.02196, "grad_norm": 0.525709002541586, "learning_rate": 0.003, "loss": 4.2424, "step": 2196 }, { "epoch": 0.02197, "grad_norm": 0.5074240255085518, "learning_rate": 0.003, "loss": 4.2329, "step": 2197 }, { "epoch": 0.02198, "grad_norm": 0.5397790588786494, "learning_rate": 0.003, "loss": 4.2239, "step": 2198 }, { "epoch": 0.02199, "grad_norm": 0.6107923796986595, "learning_rate": 0.003, "loss": 4.1928, "step": 2199 }, { "epoch": 0.022, "grad_norm": 0.843024485274174, "learning_rate": 0.003, "loss": 4.2284, "step": 2200 }, { "epoch": 0.02201, "grad_norm": 0.9840256721686866, "learning_rate": 0.003, "loss": 4.2546, "step": 2201 }, { "epoch": 0.02202, "grad_norm": 1.0784933389723226, "learning_rate": 0.003, "loss": 4.2428, "step": 2202 }, { "epoch": 0.02203, "grad_norm": 0.8532329582759304, "learning_rate": 0.003, "loss": 4.2706, "step": 2203 }, { "epoch": 0.02204, "grad_norm": 0.7139369856529667, "learning_rate": 0.003, "loss": 4.259, "step": 2204 }, { "epoch": 0.02205, "grad_norm": 0.7536669209652429, "learning_rate": 0.003, "loss": 4.258, "step": 2205 }, { "epoch": 0.02206, "grad_norm": 0.7857084413842441, "learning_rate": 0.003, "loss": 4.224, "step": 2206 }, { "epoch": 0.02207, "grad_norm": 0.7046511159473173, "learning_rate": 0.003, "loss": 4.2327, "step": 2207 }, { "epoch": 0.02208, "grad_norm": 0.6175016908695475, "learning_rate": 0.003, "loss": 4.2135, "step": 2208 }, { "epoch": 0.02209, "grad_norm": 0.6407707276531509, "learning_rate": 0.003, "loss": 4.2282, "step": 2209 }, { "epoch": 0.0221, "grad_norm": 0.5593920924645548, "learning_rate": 0.003, "loss": 4.2057, "step": 2210 }, { "epoch": 0.02211, "grad_norm": 0.5677819571205628, "learning_rate": 0.003, "loss": 4.216, "step": 2211 }, { "epoch": 0.02212, "grad_norm": 0.5260851177623344, "learning_rate": 0.003, "loss": 4.2125, "step": 2212 }, { "epoch": 0.02213, "grad_norm": 0.5193985345861573, "learning_rate": 0.003, "loss": 4.2187, "step": 2213 }, { "epoch": 0.02214, "grad_norm": 0.5150545706654319, "learning_rate": 0.003, "loss": 4.2111, "step": 2214 }, { "epoch": 0.02215, "grad_norm": 0.5981375357124284, "learning_rate": 0.003, "loss": 4.2178, "step": 2215 }, { "epoch": 0.02216, "grad_norm": 0.6928559733565658, "learning_rate": 0.003, "loss": 4.1968, "step": 2216 }, { "epoch": 0.02217, "grad_norm": 0.8190269184049962, "learning_rate": 0.003, "loss": 4.2012, "step": 2217 }, { "epoch": 0.02218, "grad_norm": 1.0426971364675302, "learning_rate": 0.003, "loss": 4.2407, "step": 2218 }, { "epoch": 0.02219, "grad_norm": 0.9675276960649843, "learning_rate": 0.003, "loss": 4.2359, "step": 2219 }, { "epoch": 0.0222, "grad_norm": 0.7683132622272745, "learning_rate": 0.003, "loss": 4.2262, "step": 2220 }, { "epoch": 0.02221, "grad_norm": 0.870281319346899, "learning_rate": 0.003, "loss": 4.2211, "step": 2221 }, { "epoch": 0.02222, "grad_norm": 0.9031315012246321, "learning_rate": 0.003, "loss": 4.2256, "step": 2222 }, { "epoch": 0.02223, "grad_norm": 0.981876469109676, "learning_rate": 0.003, "loss": 4.243, "step": 2223 }, { "epoch": 0.02224, "grad_norm": 1.0242324809795773, "learning_rate": 0.003, "loss": 4.2285, "step": 2224 }, { "epoch": 0.02225, "grad_norm": 0.9687768727218721, "learning_rate": 0.003, "loss": 4.2331, "step": 2225 }, { "epoch": 0.02226, "grad_norm": 0.8180232793014202, "learning_rate": 0.003, "loss": 4.2439, "step": 2226 }, { "epoch": 0.02227, "grad_norm": 0.8470443159450464, "learning_rate": 0.003, "loss": 4.244, "step": 2227 }, { "epoch": 0.02228, "grad_norm": 0.892588366281847, "learning_rate": 0.003, "loss": 4.2316, "step": 2228 }, { "epoch": 0.02229, "grad_norm": 0.9147609345948713, "learning_rate": 0.003, "loss": 4.2199, "step": 2229 }, { "epoch": 0.0223, "grad_norm": 0.7916782102081119, "learning_rate": 0.003, "loss": 4.2579, "step": 2230 }, { "epoch": 0.02231, "grad_norm": 0.8498212362921145, "learning_rate": 0.003, "loss": 4.2621, "step": 2231 }, { "epoch": 0.02232, "grad_norm": 0.8525982405062639, "learning_rate": 0.003, "loss": 4.268, "step": 2232 }, { "epoch": 0.02233, "grad_norm": 0.8718378998708317, "learning_rate": 0.003, "loss": 4.2384, "step": 2233 }, { "epoch": 0.02234, "grad_norm": 0.9791386495186906, "learning_rate": 0.003, "loss": 4.2628, "step": 2234 }, { "epoch": 0.02235, "grad_norm": 1.017250595852017, "learning_rate": 0.003, "loss": 4.2938, "step": 2235 }, { "epoch": 0.02236, "grad_norm": 0.963171893049642, "learning_rate": 0.003, "loss": 4.2483, "step": 2236 }, { "epoch": 0.02237, "grad_norm": 0.8231940931617269, "learning_rate": 0.003, "loss": 4.2712, "step": 2237 }, { "epoch": 0.02238, "grad_norm": 0.6960455304532995, "learning_rate": 0.003, "loss": 4.2462, "step": 2238 }, { "epoch": 0.02239, "grad_norm": 0.6934460578181165, "learning_rate": 0.003, "loss": 4.2313, "step": 2239 }, { "epoch": 0.0224, "grad_norm": 0.7144921574976273, "learning_rate": 0.003, "loss": 4.2144, "step": 2240 }, { "epoch": 0.02241, "grad_norm": 0.7809367350220192, "learning_rate": 0.003, "loss": 4.2455, "step": 2241 }, { "epoch": 0.02242, "grad_norm": 0.7589150655929028, "learning_rate": 0.003, "loss": 4.2271, "step": 2242 }, { "epoch": 0.02243, "grad_norm": 0.6383337337000002, "learning_rate": 0.003, "loss": 4.2512, "step": 2243 }, { "epoch": 0.02244, "grad_norm": 0.5988340655516636, "learning_rate": 0.003, "loss": 4.2259, "step": 2244 }, { "epoch": 0.02245, "grad_norm": 0.6045948555456119, "learning_rate": 0.003, "loss": 4.2118, "step": 2245 }, { "epoch": 0.02246, "grad_norm": 0.5516002496150979, "learning_rate": 0.003, "loss": 4.2377, "step": 2246 }, { "epoch": 0.02247, "grad_norm": 0.5370313797025699, "learning_rate": 0.003, "loss": 4.2109, "step": 2247 }, { "epoch": 0.02248, "grad_norm": 0.48933002653887103, "learning_rate": 0.003, "loss": 4.2154, "step": 2248 }, { "epoch": 0.02249, "grad_norm": 0.42750734194283135, "learning_rate": 0.003, "loss": 4.2034, "step": 2249 }, { "epoch": 0.0225, "grad_norm": 0.3448759025517304, "learning_rate": 0.003, "loss": 4.2156, "step": 2250 }, { "epoch": 0.02251, "grad_norm": 0.3579581291564086, "learning_rate": 0.003, "loss": 4.2368, "step": 2251 }, { "epoch": 0.02252, "grad_norm": 0.3338868919973551, "learning_rate": 0.003, "loss": 4.2252, "step": 2252 }, { "epoch": 0.02253, "grad_norm": 0.2833980373722387, "learning_rate": 0.003, "loss": 4.2038, "step": 2253 }, { "epoch": 0.02254, "grad_norm": 0.3354848889066889, "learning_rate": 0.003, "loss": 4.2108, "step": 2254 }, { "epoch": 0.02255, "grad_norm": 0.34008562867521774, "learning_rate": 0.003, "loss": 4.192, "step": 2255 }, { "epoch": 0.02256, "grad_norm": 0.40752535782237725, "learning_rate": 0.003, "loss": 4.1822, "step": 2256 }, { "epoch": 0.02257, "grad_norm": 0.7100736173105294, "learning_rate": 0.003, "loss": 4.204, "step": 2257 }, { "epoch": 0.02258, "grad_norm": 1.2598699348045919, "learning_rate": 0.003, "loss": 4.2395, "step": 2258 }, { "epoch": 0.02259, "grad_norm": 0.8257304825969743, "learning_rate": 0.003, "loss": 4.1899, "step": 2259 }, { "epoch": 0.0226, "grad_norm": 0.5276115234939553, "learning_rate": 0.003, "loss": 4.235, "step": 2260 }, { "epoch": 0.02261, "grad_norm": 0.7264131381401914, "learning_rate": 0.003, "loss": 4.2191, "step": 2261 }, { "epoch": 0.02262, "grad_norm": 0.694830058733267, "learning_rate": 0.003, "loss": 4.2151, "step": 2262 }, { "epoch": 0.02263, "grad_norm": 0.6143146874121539, "learning_rate": 0.003, "loss": 4.228, "step": 2263 }, { "epoch": 0.02264, "grad_norm": 0.6817476502572059, "learning_rate": 0.003, "loss": 4.2217, "step": 2264 }, { "epoch": 0.02265, "grad_norm": 0.7652944991612053, "learning_rate": 0.003, "loss": 4.1817, "step": 2265 }, { "epoch": 0.02266, "grad_norm": 0.756233264477269, "learning_rate": 0.003, "loss": 4.2012, "step": 2266 }, { "epoch": 0.02267, "grad_norm": 0.736689028154492, "learning_rate": 0.003, "loss": 4.2277, "step": 2267 }, { "epoch": 0.02268, "grad_norm": 0.7402792571581283, "learning_rate": 0.003, "loss": 4.2036, "step": 2268 }, { "epoch": 0.02269, "grad_norm": 0.6161983408604007, "learning_rate": 0.003, "loss": 4.2232, "step": 2269 }, { "epoch": 0.0227, "grad_norm": 0.4962119655121251, "learning_rate": 0.003, "loss": 4.2158, "step": 2270 }, { "epoch": 0.02271, "grad_norm": 0.5302680760721596, "learning_rate": 0.003, "loss": 4.2188, "step": 2271 }, { "epoch": 0.02272, "grad_norm": 0.6394586434048679, "learning_rate": 0.003, "loss": 4.221, "step": 2272 }, { "epoch": 0.02273, "grad_norm": 0.7045324906964335, "learning_rate": 0.003, "loss": 4.2241, "step": 2273 }, { "epoch": 0.02274, "grad_norm": 0.7390109699118109, "learning_rate": 0.003, "loss": 4.2137, "step": 2274 }, { "epoch": 0.02275, "grad_norm": 0.7499581715087588, "learning_rate": 0.003, "loss": 4.2081, "step": 2275 }, { "epoch": 0.02276, "grad_norm": 0.7669015119641691, "learning_rate": 0.003, "loss": 4.2021, "step": 2276 }, { "epoch": 0.02277, "grad_norm": 0.7961571446299023, "learning_rate": 0.003, "loss": 4.2142, "step": 2277 }, { "epoch": 0.02278, "grad_norm": 0.9864062453804502, "learning_rate": 0.003, "loss": 4.2276, "step": 2278 }, { "epoch": 0.02279, "grad_norm": 0.9358624340244462, "learning_rate": 0.003, "loss": 4.2442, "step": 2279 }, { "epoch": 0.0228, "grad_norm": 0.812834471353606, "learning_rate": 0.003, "loss": 4.2285, "step": 2280 }, { "epoch": 0.02281, "grad_norm": 0.6684943400610098, "learning_rate": 0.003, "loss": 4.2176, "step": 2281 }, { "epoch": 0.02282, "grad_norm": 0.669619706421883, "learning_rate": 0.003, "loss": 4.2295, "step": 2282 }, { "epoch": 0.02283, "grad_norm": 0.6330282624675099, "learning_rate": 0.003, "loss": 4.2424, "step": 2283 }, { "epoch": 0.02284, "grad_norm": 0.5990447110243563, "learning_rate": 0.003, "loss": 4.2354, "step": 2284 }, { "epoch": 0.02285, "grad_norm": 0.628614060536484, "learning_rate": 0.003, "loss": 4.1768, "step": 2285 }, { "epoch": 0.02286, "grad_norm": 0.6697445056976077, "learning_rate": 0.003, "loss": 4.2382, "step": 2286 }, { "epoch": 0.02287, "grad_norm": 0.7952589491731732, "learning_rate": 0.003, "loss": 4.2228, "step": 2287 }, { "epoch": 0.02288, "grad_norm": 1.0045332281072228, "learning_rate": 0.003, "loss": 4.2596, "step": 2288 }, { "epoch": 0.02289, "grad_norm": 1.1532936973113286, "learning_rate": 0.003, "loss": 4.2697, "step": 2289 }, { "epoch": 0.0229, "grad_norm": 0.819312692673835, "learning_rate": 0.003, "loss": 4.2317, "step": 2290 }, { "epoch": 0.02291, "grad_norm": 0.7813095877942711, "learning_rate": 0.003, "loss": 4.2134, "step": 2291 }, { "epoch": 0.02292, "grad_norm": 0.7936575982558383, "learning_rate": 0.003, "loss": 4.207, "step": 2292 }, { "epoch": 0.02293, "grad_norm": 1.0074422757908905, "learning_rate": 0.003, "loss": 4.2266, "step": 2293 }, { "epoch": 0.02294, "grad_norm": 1.40961843940546, "learning_rate": 0.003, "loss": 4.2493, "step": 2294 }, { "epoch": 0.02295, "grad_norm": 0.7960792761101736, "learning_rate": 0.003, "loss": 4.2214, "step": 2295 }, { "epoch": 0.02296, "grad_norm": 0.7966163433797685, "learning_rate": 0.003, "loss": 4.2574, "step": 2296 }, { "epoch": 0.02297, "grad_norm": 0.9441515456535015, "learning_rate": 0.003, "loss": 4.2281, "step": 2297 }, { "epoch": 0.02298, "grad_norm": 0.9385676483518116, "learning_rate": 0.003, "loss": 4.2472, "step": 2298 }, { "epoch": 0.02299, "grad_norm": 0.8766202093405785, "learning_rate": 0.003, "loss": 4.243, "step": 2299 }, { "epoch": 0.023, "grad_norm": 0.8448821550504723, "learning_rate": 0.003, "loss": 4.2644, "step": 2300 }, { "epoch": 0.02301, "grad_norm": 0.9129468837162014, "learning_rate": 0.003, "loss": 4.2345, "step": 2301 }, { "epoch": 0.02302, "grad_norm": 0.8382327875972556, "learning_rate": 0.003, "loss": 4.2375, "step": 2302 }, { "epoch": 0.02303, "grad_norm": 0.6647794411742001, "learning_rate": 0.003, "loss": 4.2192, "step": 2303 }, { "epoch": 0.02304, "grad_norm": 0.567815223248062, "learning_rate": 0.003, "loss": 4.2476, "step": 2304 }, { "epoch": 0.02305, "grad_norm": 0.5306257096149566, "learning_rate": 0.003, "loss": 4.2208, "step": 2305 }, { "epoch": 0.02306, "grad_norm": 0.4934389293531822, "learning_rate": 0.003, "loss": 4.1942, "step": 2306 }, { "epoch": 0.02307, "grad_norm": 0.4668795011764801, "learning_rate": 0.003, "loss": 4.2225, "step": 2307 }, { "epoch": 0.02308, "grad_norm": 0.4206347578571924, "learning_rate": 0.003, "loss": 4.2125, "step": 2308 }, { "epoch": 0.02309, "grad_norm": 0.4940433849553269, "learning_rate": 0.003, "loss": 4.2381, "step": 2309 }, { "epoch": 0.0231, "grad_norm": 0.5558762008475976, "learning_rate": 0.003, "loss": 4.2182, "step": 2310 }, { "epoch": 0.02311, "grad_norm": 0.675026786943639, "learning_rate": 0.003, "loss": 4.2246, "step": 2311 }, { "epoch": 0.02312, "grad_norm": 0.8262505859792513, "learning_rate": 0.003, "loss": 4.2059, "step": 2312 }, { "epoch": 0.02313, "grad_norm": 0.9646885017751712, "learning_rate": 0.003, "loss": 4.2035, "step": 2313 }, { "epoch": 0.02314, "grad_norm": 0.9620181533726643, "learning_rate": 0.003, "loss": 4.2279, "step": 2314 }, { "epoch": 0.02315, "grad_norm": 0.7208688026542501, "learning_rate": 0.003, "loss": 4.2198, "step": 2315 }, { "epoch": 0.02316, "grad_norm": 0.6871143951606865, "learning_rate": 0.003, "loss": 4.233, "step": 2316 }, { "epoch": 0.02317, "grad_norm": 0.7343972203270459, "learning_rate": 0.003, "loss": 4.2178, "step": 2317 }, { "epoch": 0.02318, "grad_norm": 0.672729666884604, "learning_rate": 0.003, "loss": 4.2019, "step": 2318 }, { "epoch": 0.02319, "grad_norm": 0.5294586305376263, "learning_rate": 0.003, "loss": 4.2232, "step": 2319 }, { "epoch": 0.0232, "grad_norm": 0.45021836108537966, "learning_rate": 0.003, "loss": 4.228, "step": 2320 }, { "epoch": 0.02321, "grad_norm": 0.4332276722950085, "learning_rate": 0.003, "loss": 4.2056, "step": 2321 }, { "epoch": 0.02322, "grad_norm": 0.415331953066457, "learning_rate": 0.003, "loss": 4.2122, "step": 2322 }, { "epoch": 0.02323, "grad_norm": 0.4782940324906927, "learning_rate": 0.003, "loss": 4.188, "step": 2323 }, { "epoch": 0.02324, "grad_norm": 0.44973622532022933, "learning_rate": 0.003, "loss": 4.1851, "step": 2324 }, { "epoch": 0.02325, "grad_norm": 0.5367241864779755, "learning_rate": 0.003, "loss": 4.2104, "step": 2325 }, { "epoch": 0.02326, "grad_norm": 0.6485437733030329, "learning_rate": 0.003, "loss": 4.2262, "step": 2326 }, { "epoch": 0.02327, "grad_norm": 0.8108781550029248, "learning_rate": 0.003, "loss": 4.2167, "step": 2327 }, { "epoch": 0.02328, "grad_norm": 0.8492355692805335, "learning_rate": 0.003, "loss": 4.2165, "step": 2328 }, { "epoch": 0.02329, "grad_norm": 0.8631583290092871, "learning_rate": 0.003, "loss": 4.2143, "step": 2329 }, { "epoch": 0.0233, "grad_norm": 0.7974678745136863, "learning_rate": 0.003, "loss": 4.208, "step": 2330 }, { "epoch": 0.02331, "grad_norm": 0.7747922073327778, "learning_rate": 0.003, "loss": 4.2484, "step": 2331 }, { "epoch": 0.02332, "grad_norm": 0.7695930959807518, "learning_rate": 0.003, "loss": 4.2133, "step": 2332 }, { "epoch": 0.02333, "grad_norm": 0.8630069509222793, "learning_rate": 0.003, "loss": 4.2258, "step": 2333 }, { "epoch": 0.02334, "grad_norm": 0.8750358925315561, "learning_rate": 0.003, "loss": 4.1963, "step": 2334 }, { "epoch": 0.02335, "grad_norm": 0.8509103316399725, "learning_rate": 0.003, "loss": 4.2374, "step": 2335 }, { "epoch": 0.02336, "grad_norm": 0.7392709118908145, "learning_rate": 0.003, "loss": 4.2257, "step": 2336 }, { "epoch": 0.02337, "grad_norm": 0.6895369704663263, "learning_rate": 0.003, "loss": 4.2245, "step": 2337 }, { "epoch": 0.02338, "grad_norm": 0.7845765410625875, "learning_rate": 0.003, "loss": 4.2198, "step": 2338 }, { "epoch": 0.02339, "grad_norm": 0.874209978127432, "learning_rate": 0.003, "loss": 4.2152, "step": 2339 }, { "epoch": 0.0234, "grad_norm": 1.0163700561103481, "learning_rate": 0.003, "loss": 4.2423, "step": 2340 }, { "epoch": 0.02341, "grad_norm": 0.8157418854610229, "learning_rate": 0.003, "loss": 4.2078, "step": 2341 }, { "epoch": 0.02342, "grad_norm": 0.74346999717558, "learning_rate": 0.003, "loss": 4.2154, "step": 2342 }, { "epoch": 0.02343, "grad_norm": 0.7104070086359986, "learning_rate": 0.003, "loss": 4.2179, "step": 2343 }, { "epoch": 0.02344, "grad_norm": 0.6775910262494933, "learning_rate": 0.003, "loss": 4.2138, "step": 2344 }, { "epoch": 0.02345, "grad_norm": 0.7196047539014085, "learning_rate": 0.003, "loss": 4.2099, "step": 2345 }, { "epoch": 0.02346, "grad_norm": 0.8287165981467728, "learning_rate": 0.003, "loss": 4.2085, "step": 2346 }, { "epoch": 0.02347, "grad_norm": 0.9371300769994676, "learning_rate": 0.003, "loss": 4.252, "step": 2347 }, { "epoch": 0.02348, "grad_norm": 1.0680546915325266, "learning_rate": 0.003, "loss": 4.2253, "step": 2348 }, { "epoch": 0.02349, "grad_norm": 0.8280068780663966, "learning_rate": 0.003, "loss": 4.2341, "step": 2349 }, { "epoch": 0.0235, "grad_norm": 0.7133386737402891, "learning_rate": 0.003, "loss": 4.2259, "step": 2350 }, { "epoch": 0.02351, "grad_norm": 0.7596846967896084, "learning_rate": 0.003, "loss": 4.2382, "step": 2351 }, { "epoch": 0.02352, "grad_norm": 0.7235609932262353, "learning_rate": 0.003, "loss": 4.1922, "step": 2352 }, { "epoch": 0.02353, "grad_norm": 0.6625191060428061, "learning_rate": 0.003, "loss": 4.2225, "step": 2353 }, { "epoch": 0.02354, "grad_norm": 0.7210256062729306, "learning_rate": 0.003, "loss": 4.2281, "step": 2354 }, { "epoch": 0.02355, "grad_norm": 0.6784409469826875, "learning_rate": 0.003, "loss": 4.2244, "step": 2355 }, { "epoch": 0.02356, "grad_norm": 0.6340113796322493, "learning_rate": 0.003, "loss": 4.2034, "step": 2356 }, { "epoch": 0.02357, "grad_norm": 0.5872909081662198, "learning_rate": 0.003, "loss": 4.2037, "step": 2357 }, { "epoch": 0.02358, "grad_norm": 0.6012777742156055, "learning_rate": 0.003, "loss": 4.2251, "step": 2358 }, { "epoch": 0.02359, "grad_norm": 0.71706729373374, "learning_rate": 0.003, "loss": 4.2046, "step": 2359 }, { "epoch": 0.0236, "grad_norm": 0.815136439263558, "learning_rate": 0.003, "loss": 4.2434, "step": 2360 }, { "epoch": 0.02361, "grad_norm": 0.8003325831199607, "learning_rate": 0.003, "loss": 4.2261, "step": 2361 }, { "epoch": 0.02362, "grad_norm": 0.8097756127870204, "learning_rate": 0.003, "loss": 4.2177, "step": 2362 }, { "epoch": 0.02363, "grad_norm": 0.7914189012727164, "learning_rate": 0.003, "loss": 4.1921, "step": 2363 }, { "epoch": 0.02364, "grad_norm": 0.7565060855875264, "learning_rate": 0.003, "loss": 4.2126, "step": 2364 }, { "epoch": 0.02365, "grad_norm": 0.7332274502623093, "learning_rate": 0.003, "loss": 4.2307, "step": 2365 }, { "epoch": 0.02366, "grad_norm": 0.7247749395727022, "learning_rate": 0.003, "loss": 4.2058, "step": 2366 }, { "epoch": 0.02367, "grad_norm": 0.6754475163243551, "learning_rate": 0.003, "loss": 4.2138, "step": 2367 }, { "epoch": 0.02368, "grad_norm": 0.6381389762534017, "learning_rate": 0.003, "loss": 4.1873, "step": 2368 }, { "epoch": 0.02369, "grad_norm": 0.6120733905364695, "learning_rate": 0.003, "loss": 4.1978, "step": 2369 }, { "epoch": 0.0237, "grad_norm": 0.7329047764963631, "learning_rate": 0.003, "loss": 4.2283, "step": 2370 }, { "epoch": 0.02371, "grad_norm": 0.7596120866036141, "learning_rate": 0.003, "loss": 4.2137, "step": 2371 }, { "epoch": 0.02372, "grad_norm": 0.7920778056584535, "learning_rate": 0.003, "loss": 4.202, "step": 2372 }, { "epoch": 0.02373, "grad_norm": 0.6541726044177485, "learning_rate": 0.003, "loss": 4.1876, "step": 2373 }, { "epoch": 0.02374, "grad_norm": 0.4980669374964598, "learning_rate": 0.003, "loss": 4.1906, "step": 2374 }, { "epoch": 0.02375, "grad_norm": 0.48620968233730555, "learning_rate": 0.003, "loss": 4.1923, "step": 2375 }, { "epoch": 0.02376, "grad_norm": 0.5343060557665553, "learning_rate": 0.003, "loss": 4.2157, "step": 2376 }, { "epoch": 0.02377, "grad_norm": 0.5346079035944942, "learning_rate": 0.003, "loss": 4.1829, "step": 2377 }, { "epoch": 0.02378, "grad_norm": 0.5345430148085093, "learning_rate": 0.003, "loss": 4.1926, "step": 2378 }, { "epoch": 0.02379, "grad_norm": 0.5981101437562691, "learning_rate": 0.003, "loss": 4.1826, "step": 2379 }, { "epoch": 0.0238, "grad_norm": 0.6559236976426223, "learning_rate": 0.003, "loss": 4.1575, "step": 2380 }, { "epoch": 0.02381, "grad_norm": 0.7048531875426997, "learning_rate": 0.003, "loss": 4.1695, "step": 2381 }, { "epoch": 0.02382, "grad_norm": 0.7200426361012294, "learning_rate": 0.003, "loss": 4.2311, "step": 2382 }, { "epoch": 0.02383, "grad_norm": 0.6892017198060378, "learning_rate": 0.003, "loss": 4.1909, "step": 2383 }, { "epoch": 0.02384, "grad_norm": 0.6858037418633958, "learning_rate": 0.003, "loss": 4.1796, "step": 2384 }, { "epoch": 0.02385, "grad_norm": 0.7487982436774401, "learning_rate": 0.003, "loss": 4.2087, "step": 2385 }, { "epoch": 0.02386, "grad_norm": 0.7730947217702177, "learning_rate": 0.003, "loss": 4.1913, "step": 2386 }, { "epoch": 0.02387, "grad_norm": 0.7349341583623172, "learning_rate": 0.003, "loss": 4.1849, "step": 2387 }, { "epoch": 0.02388, "grad_norm": 0.6224368008762636, "learning_rate": 0.003, "loss": 4.2345, "step": 2388 }, { "epoch": 0.02389, "grad_norm": 0.6226480564621836, "learning_rate": 0.003, "loss": 4.1982, "step": 2389 }, { "epoch": 0.0239, "grad_norm": 0.6254191342622962, "learning_rate": 0.003, "loss": 4.1914, "step": 2390 }, { "epoch": 0.02391, "grad_norm": 0.6631610099694426, "learning_rate": 0.003, "loss": 4.2141, "step": 2391 }, { "epoch": 0.02392, "grad_norm": 0.8204612150837974, "learning_rate": 0.003, "loss": 4.2245, "step": 2392 }, { "epoch": 0.02393, "grad_norm": 1.0323111703597159, "learning_rate": 0.003, "loss": 4.1943, "step": 2393 }, { "epoch": 0.02394, "grad_norm": 1.0006388257189816, "learning_rate": 0.003, "loss": 4.2112, "step": 2394 }, { "epoch": 0.02395, "grad_norm": 0.7572494256888903, "learning_rate": 0.003, "loss": 4.2171, "step": 2395 }, { "epoch": 0.02396, "grad_norm": 0.5951984266508508, "learning_rate": 0.003, "loss": 4.2255, "step": 2396 }, { "epoch": 0.02397, "grad_norm": 0.6650102744746705, "learning_rate": 0.003, "loss": 4.2116, "step": 2397 }, { "epoch": 0.02398, "grad_norm": 0.6825039278747758, "learning_rate": 0.003, "loss": 4.2155, "step": 2398 }, { "epoch": 0.02399, "grad_norm": 0.6712187274269288, "learning_rate": 0.003, "loss": 4.1963, "step": 2399 }, { "epoch": 0.024, "grad_norm": 0.9110222795383771, "learning_rate": 0.003, "loss": 4.2186, "step": 2400 }, { "epoch": 0.02401, "grad_norm": 1.0426771564263924, "learning_rate": 0.003, "loss": 4.2141, "step": 2401 }, { "epoch": 0.02402, "grad_norm": 0.8152831624353678, "learning_rate": 0.003, "loss": 4.2285, "step": 2402 }, { "epoch": 0.02403, "grad_norm": 0.8195630042569122, "learning_rate": 0.003, "loss": 4.2044, "step": 2403 }, { "epoch": 0.02404, "grad_norm": 0.8040024828088439, "learning_rate": 0.003, "loss": 4.243, "step": 2404 }, { "epoch": 0.02405, "grad_norm": 0.885388520382783, "learning_rate": 0.003, "loss": 4.2578, "step": 2405 }, { "epoch": 0.02406, "grad_norm": 1.0169970243763446, "learning_rate": 0.003, "loss": 4.2493, "step": 2406 }, { "epoch": 0.02407, "grad_norm": 1.1565060428680438, "learning_rate": 0.003, "loss": 4.2434, "step": 2407 }, { "epoch": 0.02408, "grad_norm": 1.1677513319022517, "learning_rate": 0.003, "loss": 4.2497, "step": 2408 }, { "epoch": 0.02409, "grad_norm": 1.151489835753338, "learning_rate": 0.003, "loss": 4.2634, "step": 2409 }, { "epoch": 0.0241, "grad_norm": 0.9730712891416885, "learning_rate": 0.003, "loss": 4.2617, "step": 2410 }, { "epoch": 0.02411, "grad_norm": 1.0934561640650202, "learning_rate": 0.003, "loss": 4.2362, "step": 2411 }, { "epoch": 0.02412, "grad_norm": 0.7564247171167698, "learning_rate": 0.003, "loss": 4.2331, "step": 2412 }, { "epoch": 0.02413, "grad_norm": 0.7364564581436038, "learning_rate": 0.003, "loss": 4.2269, "step": 2413 }, { "epoch": 0.02414, "grad_norm": 0.7499470770108977, "learning_rate": 0.003, "loss": 4.2434, "step": 2414 }, { "epoch": 0.02415, "grad_norm": 0.6688500224519887, "learning_rate": 0.003, "loss": 4.2361, "step": 2415 }, { "epoch": 0.02416, "grad_norm": 0.7749349047772426, "learning_rate": 0.003, "loss": 4.2233, "step": 2416 }, { "epoch": 0.02417, "grad_norm": 0.7661333853513042, "learning_rate": 0.003, "loss": 4.2471, "step": 2417 }, { "epoch": 0.02418, "grad_norm": 0.7328202199869238, "learning_rate": 0.003, "loss": 4.2352, "step": 2418 }, { "epoch": 0.02419, "grad_norm": 0.656398684318785, "learning_rate": 0.003, "loss": 4.2286, "step": 2419 }, { "epoch": 0.0242, "grad_norm": 0.7711466513132326, "learning_rate": 0.003, "loss": 4.2421, "step": 2420 }, { "epoch": 0.02421, "grad_norm": 0.8045742727715058, "learning_rate": 0.003, "loss": 4.2109, "step": 2421 }, { "epoch": 0.02422, "grad_norm": 0.7983836017582577, "learning_rate": 0.003, "loss": 4.2596, "step": 2422 }, { "epoch": 0.02423, "grad_norm": 0.6735385382658433, "learning_rate": 0.003, "loss": 4.2455, "step": 2423 }, { "epoch": 0.02424, "grad_norm": 0.5460827395253804, "learning_rate": 0.003, "loss": 4.2363, "step": 2424 }, { "epoch": 0.02425, "grad_norm": 0.5835576651817266, "learning_rate": 0.003, "loss": 4.2198, "step": 2425 }, { "epoch": 0.02426, "grad_norm": 0.6629194085179074, "learning_rate": 0.003, "loss": 4.227, "step": 2426 }, { "epoch": 0.02427, "grad_norm": 0.613590623947857, "learning_rate": 0.003, "loss": 4.2496, "step": 2427 }, { "epoch": 0.02428, "grad_norm": 0.5297080386636701, "learning_rate": 0.003, "loss": 4.2173, "step": 2428 }, { "epoch": 0.02429, "grad_norm": 0.6059447656989463, "learning_rate": 0.003, "loss": 4.2063, "step": 2429 }, { "epoch": 0.0243, "grad_norm": 0.4983588726462706, "learning_rate": 0.003, "loss": 4.2252, "step": 2430 }, { "epoch": 0.02431, "grad_norm": 0.4714791093056159, "learning_rate": 0.003, "loss": 4.1992, "step": 2431 }, { "epoch": 0.02432, "grad_norm": 0.462354333964094, "learning_rate": 0.003, "loss": 4.2211, "step": 2432 }, { "epoch": 0.02433, "grad_norm": 0.5377944411614727, "learning_rate": 0.003, "loss": 4.2352, "step": 2433 }, { "epoch": 0.02434, "grad_norm": 0.72053618445036, "learning_rate": 0.003, "loss": 4.205, "step": 2434 }, { "epoch": 0.02435, "grad_norm": 0.9717077197610761, "learning_rate": 0.003, "loss": 4.2126, "step": 2435 }, { "epoch": 0.02436, "grad_norm": 1.0867436803782795, "learning_rate": 0.003, "loss": 4.2337, "step": 2436 }, { "epoch": 0.02437, "grad_norm": 0.7354406575791109, "learning_rate": 0.003, "loss": 4.204, "step": 2437 }, { "epoch": 0.02438, "grad_norm": 0.5659849585693548, "learning_rate": 0.003, "loss": 4.2172, "step": 2438 }, { "epoch": 0.02439, "grad_norm": 0.7582631377117253, "learning_rate": 0.003, "loss": 4.1906, "step": 2439 }, { "epoch": 0.0244, "grad_norm": 0.7519666599937619, "learning_rate": 0.003, "loss": 4.2388, "step": 2440 }, { "epoch": 0.02441, "grad_norm": 0.7095321843053358, "learning_rate": 0.003, "loss": 4.2023, "step": 2441 }, { "epoch": 0.02442, "grad_norm": 0.7610624757093507, "learning_rate": 0.003, "loss": 4.2377, "step": 2442 }, { "epoch": 0.02443, "grad_norm": 0.7404568199239232, "learning_rate": 0.003, "loss": 4.1741, "step": 2443 }, { "epoch": 0.02444, "grad_norm": 0.737847324812997, "learning_rate": 0.003, "loss": 4.2035, "step": 2444 }, { "epoch": 0.02445, "grad_norm": 0.8038624975628923, "learning_rate": 0.003, "loss": 4.2104, "step": 2445 }, { "epoch": 0.02446, "grad_norm": 0.9990291440971695, "learning_rate": 0.003, "loss": 4.2202, "step": 2446 }, { "epoch": 0.02447, "grad_norm": 0.9412206600088139, "learning_rate": 0.003, "loss": 4.2318, "step": 2447 }, { "epoch": 0.02448, "grad_norm": 0.8294445152569498, "learning_rate": 0.003, "loss": 4.2285, "step": 2448 }, { "epoch": 0.02449, "grad_norm": 0.689492443833379, "learning_rate": 0.003, "loss": 4.2155, "step": 2449 }, { "epoch": 0.0245, "grad_norm": 0.7414638724886635, "learning_rate": 0.003, "loss": 4.222, "step": 2450 }, { "epoch": 0.02451, "grad_norm": 0.685024298981967, "learning_rate": 0.003, "loss": 4.198, "step": 2451 }, { "epoch": 0.02452, "grad_norm": 0.8346267890501767, "learning_rate": 0.003, "loss": 4.2105, "step": 2452 }, { "epoch": 0.02453, "grad_norm": 0.7934750617758409, "learning_rate": 0.003, "loss": 4.1906, "step": 2453 }, { "epoch": 0.02454, "grad_norm": 0.751875693379249, "learning_rate": 0.003, "loss": 4.1943, "step": 2454 }, { "epoch": 0.02455, "grad_norm": 0.756218693628031, "learning_rate": 0.003, "loss": 4.2083, "step": 2455 }, { "epoch": 0.02456, "grad_norm": 0.7329329951461823, "learning_rate": 0.003, "loss": 4.1738, "step": 2456 }, { "epoch": 0.02457, "grad_norm": 0.6861603609350935, "learning_rate": 0.003, "loss": 4.2258, "step": 2457 }, { "epoch": 0.02458, "grad_norm": 0.7668783056267265, "learning_rate": 0.003, "loss": 4.1965, "step": 2458 }, { "epoch": 0.02459, "grad_norm": 0.9611463702448455, "learning_rate": 0.003, "loss": 4.2358, "step": 2459 }, { "epoch": 0.0246, "grad_norm": 1.2716359645134816, "learning_rate": 0.003, "loss": 4.2528, "step": 2460 }, { "epoch": 0.02461, "grad_norm": 0.6580810366854414, "learning_rate": 0.003, "loss": 4.2104, "step": 2461 }, { "epoch": 0.02462, "grad_norm": 0.663557756308718, "learning_rate": 0.003, "loss": 4.2392, "step": 2462 }, { "epoch": 0.02463, "grad_norm": 0.7136733681949328, "learning_rate": 0.003, "loss": 4.2379, "step": 2463 }, { "epoch": 0.02464, "grad_norm": 0.733904485720918, "learning_rate": 0.003, "loss": 4.2153, "step": 2464 }, { "epoch": 0.02465, "grad_norm": 0.7484568404460167, "learning_rate": 0.003, "loss": 4.2316, "step": 2465 }, { "epoch": 0.02466, "grad_norm": 0.7165740498978449, "learning_rate": 0.003, "loss": 4.2108, "step": 2466 }, { "epoch": 0.02467, "grad_norm": 0.7346927598557101, "learning_rate": 0.003, "loss": 4.1872, "step": 2467 }, { "epoch": 0.02468, "grad_norm": 0.7134409778557855, "learning_rate": 0.003, "loss": 4.2303, "step": 2468 }, { "epoch": 0.02469, "grad_norm": 0.7525235307550779, "learning_rate": 0.003, "loss": 4.2116, "step": 2469 }, { "epoch": 0.0247, "grad_norm": 0.6233720556235373, "learning_rate": 0.003, "loss": 4.2038, "step": 2470 }, { "epoch": 0.02471, "grad_norm": 0.6188790379984411, "learning_rate": 0.003, "loss": 4.1898, "step": 2471 }, { "epoch": 0.02472, "grad_norm": 0.5576000965553038, "learning_rate": 0.003, "loss": 4.1904, "step": 2472 }, { "epoch": 0.02473, "grad_norm": 0.5733874773234577, "learning_rate": 0.003, "loss": 4.1763, "step": 2473 }, { "epoch": 0.02474, "grad_norm": 0.5058689455726723, "learning_rate": 0.003, "loss": 4.2046, "step": 2474 }, { "epoch": 0.02475, "grad_norm": 0.5073987946337717, "learning_rate": 0.003, "loss": 4.1805, "step": 2475 }, { "epoch": 0.02476, "grad_norm": 0.5627394935612221, "learning_rate": 0.003, "loss": 4.1881, "step": 2476 }, { "epoch": 0.02477, "grad_norm": 0.6236986968628518, "learning_rate": 0.003, "loss": 4.2085, "step": 2477 }, { "epoch": 0.02478, "grad_norm": 0.7245870854664911, "learning_rate": 0.003, "loss": 4.1902, "step": 2478 }, { "epoch": 0.02479, "grad_norm": 0.8546743805326398, "learning_rate": 0.003, "loss": 4.1806, "step": 2479 }, { "epoch": 0.0248, "grad_norm": 0.8522780412659844, "learning_rate": 0.003, "loss": 4.2257, "step": 2480 }, { "epoch": 0.02481, "grad_norm": 0.6535653800757177, "learning_rate": 0.003, "loss": 4.2011, "step": 2481 }, { "epoch": 0.02482, "grad_norm": 0.6344492058438189, "learning_rate": 0.003, "loss": 4.2089, "step": 2482 }, { "epoch": 0.02483, "grad_norm": 0.6268828941754008, "learning_rate": 0.003, "loss": 4.1621, "step": 2483 }, { "epoch": 0.02484, "grad_norm": 0.5957623078783284, "learning_rate": 0.003, "loss": 4.1893, "step": 2484 }, { "epoch": 0.02485, "grad_norm": 0.6986492947662325, "learning_rate": 0.003, "loss": 4.2156, "step": 2485 }, { "epoch": 0.02486, "grad_norm": 0.7414315414337274, "learning_rate": 0.003, "loss": 4.24, "step": 2486 }, { "epoch": 0.02487, "grad_norm": 0.6569802880365471, "learning_rate": 0.003, "loss": 4.1965, "step": 2487 }, { "epoch": 0.02488, "grad_norm": 0.5756977058711745, "learning_rate": 0.003, "loss": 4.1695, "step": 2488 }, { "epoch": 0.02489, "grad_norm": 0.5841485659003718, "learning_rate": 0.003, "loss": 4.1603, "step": 2489 }, { "epoch": 0.0249, "grad_norm": 0.6313109185297392, "learning_rate": 0.003, "loss": 4.1659, "step": 2490 }, { "epoch": 0.02491, "grad_norm": 0.587258429220503, "learning_rate": 0.003, "loss": 4.1919, "step": 2491 }, { "epoch": 0.02492, "grad_norm": 0.6796255021246645, "learning_rate": 0.003, "loss": 4.2045, "step": 2492 }, { "epoch": 0.02493, "grad_norm": 0.9386099062698943, "learning_rate": 0.003, "loss": 4.1849, "step": 2493 }, { "epoch": 0.02494, "grad_norm": 1.0325794653718277, "learning_rate": 0.003, "loss": 4.2058, "step": 2494 }, { "epoch": 0.02495, "grad_norm": 0.8034595706724524, "learning_rate": 0.003, "loss": 4.2081, "step": 2495 }, { "epoch": 0.02496, "grad_norm": 0.716975622530452, "learning_rate": 0.003, "loss": 4.1957, "step": 2496 }, { "epoch": 0.02497, "grad_norm": 0.7796292204460298, "learning_rate": 0.003, "loss": 4.2044, "step": 2497 }, { "epoch": 0.02498, "grad_norm": 0.8038852090110774, "learning_rate": 0.003, "loss": 4.1929, "step": 2498 }, { "epoch": 0.02499, "grad_norm": 0.8923238576494511, "learning_rate": 0.003, "loss": 4.2097, "step": 2499 }, { "epoch": 0.025, "grad_norm": 0.7727598755079679, "learning_rate": 0.003, "loss": 4.1936, "step": 2500 }, { "epoch": 0.02501, "grad_norm": 0.6495309431942861, "learning_rate": 0.003, "loss": 4.2218, "step": 2501 }, { "epoch": 0.02502, "grad_norm": 0.7137648751903825, "learning_rate": 0.003, "loss": 4.2041, "step": 2502 }, { "epoch": 0.02503, "grad_norm": 0.6940762540850597, "learning_rate": 0.003, "loss": 4.2094, "step": 2503 }, { "epoch": 0.02504, "grad_norm": 0.6648229735163252, "learning_rate": 0.003, "loss": 4.1887, "step": 2504 }, { "epoch": 0.02505, "grad_norm": 0.6560245286717432, "learning_rate": 0.003, "loss": 4.1892, "step": 2505 }, { "epoch": 0.02506, "grad_norm": 0.6908954566997763, "learning_rate": 0.003, "loss": 4.2059, "step": 2506 }, { "epoch": 0.02507, "grad_norm": 0.7070750009001837, "learning_rate": 0.003, "loss": 4.1925, "step": 2507 }, { "epoch": 0.02508, "grad_norm": 0.6441719185425088, "learning_rate": 0.003, "loss": 4.1881, "step": 2508 }, { "epoch": 0.02509, "grad_norm": 0.6590084093122774, "learning_rate": 0.003, "loss": 4.1949, "step": 2509 }, { "epoch": 0.0251, "grad_norm": 0.7745888760033045, "learning_rate": 0.003, "loss": 4.2104, "step": 2510 }, { "epoch": 0.02511, "grad_norm": 0.7154617490554088, "learning_rate": 0.003, "loss": 4.2169, "step": 2511 }, { "epoch": 0.02512, "grad_norm": 0.6776987606227394, "learning_rate": 0.003, "loss": 4.198, "step": 2512 }, { "epoch": 0.02513, "grad_norm": 0.766684849732399, "learning_rate": 0.003, "loss": 4.1988, "step": 2513 }, { "epoch": 0.02514, "grad_norm": 0.727274001755, "learning_rate": 0.003, "loss": 4.1879, "step": 2514 }, { "epoch": 0.02515, "grad_norm": 0.8761441531793727, "learning_rate": 0.003, "loss": 4.1813, "step": 2515 }, { "epoch": 0.02516, "grad_norm": 1.0755772178541487, "learning_rate": 0.003, "loss": 4.2163, "step": 2516 }, { "epoch": 0.02517, "grad_norm": 1.1923834520518084, "learning_rate": 0.003, "loss": 4.2272, "step": 2517 }, { "epoch": 0.02518, "grad_norm": 0.9485906454874968, "learning_rate": 0.003, "loss": 4.2292, "step": 2518 }, { "epoch": 0.02519, "grad_norm": 0.7415170618596796, "learning_rate": 0.003, "loss": 4.1985, "step": 2519 }, { "epoch": 0.0252, "grad_norm": 0.6944412745696904, "learning_rate": 0.003, "loss": 4.2134, "step": 2520 }, { "epoch": 0.02521, "grad_norm": 0.7068074135536067, "learning_rate": 0.003, "loss": 4.208, "step": 2521 }, { "epoch": 0.02522, "grad_norm": 0.7700461552439566, "learning_rate": 0.003, "loss": 4.2059, "step": 2522 }, { "epoch": 0.02523, "grad_norm": 0.9732286210252518, "learning_rate": 0.003, "loss": 4.2168, "step": 2523 }, { "epoch": 0.02524, "grad_norm": 1.2306281629844058, "learning_rate": 0.003, "loss": 4.2416, "step": 2524 }, { "epoch": 0.02525, "grad_norm": 0.8190686438601459, "learning_rate": 0.003, "loss": 4.241, "step": 2525 }, { "epoch": 0.02526, "grad_norm": 0.742262591344972, "learning_rate": 0.003, "loss": 4.182, "step": 2526 }, { "epoch": 0.02527, "grad_norm": 0.8569784978950316, "learning_rate": 0.003, "loss": 4.2213, "step": 2527 }, { "epoch": 0.02528, "grad_norm": 1.0273779919027248, "learning_rate": 0.003, "loss": 4.2493, "step": 2528 }, { "epoch": 0.02529, "grad_norm": 1.4013611330219795, "learning_rate": 0.003, "loss": 4.236, "step": 2529 }, { "epoch": 0.0253, "grad_norm": 0.6267544057110621, "learning_rate": 0.003, "loss": 4.234, "step": 2530 }, { "epoch": 0.02531, "grad_norm": 0.7194215796299043, "learning_rate": 0.003, "loss": 4.2346, "step": 2531 }, { "epoch": 0.02532, "grad_norm": 0.8042428452042162, "learning_rate": 0.003, "loss": 4.206, "step": 2532 }, { "epoch": 0.02533, "grad_norm": 0.7542292983077659, "learning_rate": 0.003, "loss": 4.242, "step": 2533 }, { "epoch": 0.02534, "grad_norm": 0.7181838268024839, "learning_rate": 0.003, "loss": 4.2168, "step": 2534 }, { "epoch": 0.02535, "grad_norm": 0.8381740222549681, "learning_rate": 0.003, "loss": 4.2139, "step": 2535 }, { "epoch": 0.02536, "grad_norm": 0.745200112168368, "learning_rate": 0.003, "loss": 4.2088, "step": 2536 }, { "epoch": 0.02537, "grad_norm": 0.6069281829403523, "learning_rate": 0.003, "loss": 4.1974, "step": 2537 }, { "epoch": 0.02538, "grad_norm": 0.590427378826761, "learning_rate": 0.003, "loss": 4.1708, "step": 2538 }, { "epoch": 0.02539, "grad_norm": 0.5837823282964978, "learning_rate": 0.003, "loss": 4.2201, "step": 2539 }, { "epoch": 0.0254, "grad_norm": 0.538959841561464, "learning_rate": 0.003, "loss": 4.2093, "step": 2540 }, { "epoch": 0.02541, "grad_norm": 0.42790014377314195, "learning_rate": 0.003, "loss": 4.1959, "step": 2541 }, { "epoch": 0.02542, "grad_norm": 0.3893070770616937, "learning_rate": 0.003, "loss": 4.1716, "step": 2542 }, { "epoch": 0.02543, "grad_norm": 0.43694196231407156, "learning_rate": 0.003, "loss": 4.2002, "step": 2543 }, { "epoch": 0.02544, "grad_norm": 0.5763133602608426, "learning_rate": 0.003, "loss": 4.1903, "step": 2544 }, { "epoch": 0.02545, "grad_norm": 0.8655900046851385, "learning_rate": 0.003, "loss": 4.179, "step": 2545 }, { "epoch": 0.02546, "grad_norm": 1.0957171422862184, "learning_rate": 0.003, "loss": 4.2013, "step": 2546 }, { "epoch": 0.02547, "grad_norm": 0.7852241671565661, "learning_rate": 0.003, "loss": 4.2326, "step": 2547 }, { "epoch": 0.02548, "grad_norm": 0.568158079879503, "learning_rate": 0.003, "loss": 4.2222, "step": 2548 }, { "epoch": 0.02549, "grad_norm": 0.6340487565296964, "learning_rate": 0.003, "loss": 4.217, "step": 2549 }, { "epoch": 0.0255, "grad_norm": 0.7406725193163024, "learning_rate": 0.003, "loss": 4.1793, "step": 2550 }, { "epoch": 0.02551, "grad_norm": 0.6828594076090021, "learning_rate": 0.003, "loss": 4.2087, "step": 2551 }, { "epoch": 0.02552, "grad_norm": 0.6588298930877766, "learning_rate": 0.003, "loss": 4.1785, "step": 2552 }, { "epoch": 0.02553, "grad_norm": 0.6711194686482366, "learning_rate": 0.003, "loss": 4.2417, "step": 2553 }, { "epoch": 0.02554, "grad_norm": 0.7183069611221199, "learning_rate": 0.003, "loss": 4.1948, "step": 2554 }, { "epoch": 0.02555, "grad_norm": 0.7594283334724947, "learning_rate": 0.003, "loss": 4.1875, "step": 2555 }, { "epoch": 0.02556, "grad_norm": 0.7544165526703748, "learning_rate": 0.003, "loss": 4.1963, "step": 2556 }, { "epoch": 0.02557, "grad_norm": 0.8031811764161461, "learning_rate": 0.003, "loss": 4.2063, "step": 2557 }, { "epoch": 0.02558, "grad_norm": 0.7417815254052628, "learning_rate": 0.003, "loss": 4.2195, "step": 2558 }, { "epoch": 0.02559, "grad_norm": 0.668258837116674, "learning_rate": 0.003, "loss": 4.216, "step": 2559 }, { "epoch": 0.0256, "grad_norm": 0.7338986949145504, "learning_rate": 0.003, "loss": 4.1905, "step": 2560 }, { "epoch": 0.02561, "grad_norm": 0.7321938936149771, "learning_rate": 0.003, "loss": 4.1926, "step": 2561 }, { "epoch": 0.02562, "grad_norm": 0.7675474512341959, "learning_rate": 0.003, "loss": 4.2107, "step": 2562 }, { "epoch": 0.02563, "grad_norm": 0.7554556387560584, "learning_rate": 0.003, "loss": 4.1935, "step": 2563 }, { "epoch": 0.02564, "grad_norm": 0.7775647520234856, "learning_rate": 0.003, "loss": 4.2162, "step": 2564 }, { "epoch": 0.02565, "grad_norm": 0.7534991095864281, "learning_rate": 0.003, "loss": 4.1929, "step": 2565 }, { "epoch": 0.02566, "grad_norm": 0.7288648663551685, "learning_rate": 0.003, "loss": 4.2173, "step": 2566 }, { "epoch": 0.02567, "grad_norm": 0.7215669034727294, "learning_rate": 0.003, "loss": 4.2131, "step": 2567 }, { "epoch": 0.02568, "grad_norm": 0.8563040223027246, "learning_rate": 0.003, "loss": 4.2138, "step": 2568 }, { "epoch": 0.02569, "grad_norm": 1.1080869622355456, "learning_rate": 0.003, "loss": 4.2025, "step": 2569 }, { "epoch": 0.0257, "grad_norm": 1.1075262903516023, "learning_rate": 0.003, "loss": 4.2183, "step": 2570 }, { "epoch": 0.02571, "grad_norm": 0.729899139052775, "learning_rate": 0.003, "loss": 4.2237, "step": 2571 }, { "epoch": 0.02572, "grad_norm": 0.6904609591483578, "learning_rate": 0.003, "loss": 4.2007, "step": 2572 }, { "epoch": 0.02573, "grad_norm": 0.8131518041712269, "learning_rate": 0.003, "loss": 4.2089, "step": 2573 }, { "epoch": 0.02574, "grad_norm": 0.7847619978392997, "learning_rate": 0.003, "loss": 4.2345, "step": 2574 }, { "epoch": 0.02575, "grad_norm": 0.778225241603638, "learning_rate": 0.003, "loss": 4.1711, "step": 2575 }, { "epoch": 0.02576, "grad_norm": 0.8302954615357605, "learning_rate": 0.003, "loss": 4.1776, "step": 2576 }, { "epoch": 0.02577, "grad_norm": 0.8152270887192647, "learning_rate": 0.003, "loss": 4.1841, "step": 2577 }, { "epoch": 0.02578, "grad_norm": 0.7255559351586255, "learning_rate": 0.003, "loss": 4.2031, "step": 2578 }, { "epoch": 0.02579, "grad_norm": 0.7449798683561115, "learning_rate": 0.003, "loss": 4.1513, "step": 2579 }, { "epoch": 0.0258, "grad_norm": 0.6502036682045843, "learning_rate": 0.003, "loss": 4.2051, "step": 2580 }, { "epoch": 0.02581, "grad_norm": 0.6737707023470867, "learning_rate": 0.003, "loss": 4.2262, "step": 2581 }, { "epoch": 0.02582, "grad_norm": 0.764427888615585, "learning_rate": 0.003, "loss": 4.224, "step": 2582 }, { "epoch": 0.02583, "grad_norm": 0.8504470874659034, "learning_rate": 0.003, "loss": 4.2026, "step": 2583 }, { "epoch": 0.02584, "grad_norm": 1.046258332322794, "learning_rate": 0.003, "loss": 4.221, "step": 2584 }, { "epoch": 0.02585, "grad_norm": 0.9031370563550914, "learning_rate": 0.003, "loss": 4.2014, "step": 2585 }, { "epoch": 0.02586, "grad_norm": 0.689996238011, "learning_rate": 0.003, "loss": 4.217, "step": 2586 }, { "epoch": 0.02587, "grad_norm": 0.7274353541891899, "learning_rate": 0.003, "loss": 4.2051, "step": 2587 }, { "epoch": 0.02588, "grad_norm": 0.7517253783207216, "learning_rate": 0.003, "loss": 4.2032, "step": 2588 }, { "epoch": 0.02589, "grad_norm": 0.8107477123925885, "learning_rate": 0.003, "loss": 4.2123, "step": 2589 }, { "epoch": 0.0259, "grad_norm": 0.7806508541147034, "learning_rate": 0.003, "loss": 4.2204, "step": 2590 }, { "epoch": 0.02591, "grad_norm": 0.6296252175534539, "learning_rate": 0.003, "loss": 4.1497, "step": 2591 }, { "epoch": 0.02592, "grad_norm": 0.5752391696870554, "learning_rate": 0.003, "loss": 4.1918, "step": 2592 }, { "epoch": 0.02593, "grad_norm": 0.517372047833294, "learning_rate": 0.003, "loss": 4.1949, "step": 2593 }, { "epoch": 0.02594, "grad_norm": 0.5906922889145111, "learning_rate": 0.003, "loss": 4.2115, "step": 2594 }, { "epoch": 0.02595, "grad_norm": 0.6262440905091147, "learning_rate": 0.003, "loss": 4.1658, "step": 2595 }, { "epoch": 0.02596, "grad_norm": 0.5949484998914474, "learning_rate": 0.003, "loss": 4.2045, "step": 2596 }, { "epoch": 0.02597, "grad_norm": 0.6006456318859384, "learning_rate": 0.003, "loss": 4.1727, "step": 2597 }, { "epoch": 0.02598, "grad_norm": 0.5609952650146841, "learning_rate": 0.003, "loss": 4.2084, "step": 2598 }, { "epoch": 0.02599, "grad_norm": 0.6441455777508067, "learning_rate": 0.003, "loss": 4.2067, "step": 2599 }, { "epoch": 0.026, "grad_norm": 0.7930588451098299, "learning_rate": 0.003, "loss": 4.2116, "step": 2600 }, { "epoch": 0.02601, "grad_norm": 0.8782723700452447, "learning_rate": 0.003, "loss": 4.2212, "step": 2601 }, { "epoch": 0.02602, "grad_norm": 1.1759564543141179, "learning_rate": 0.003, "loss": 4.2063, "step": 2602 }, { "epoch": 0.02603, "grad_norm": 0.8180315225343916, "learning_rate": 0.003, "loss": 4.19, "step": 2603 }, { "epoch": 0.02604, "grad_norm": 0.7244608750007608, "learning_rate": 0.003, "loss": 4.1809, "step": 2604 }, { "epoch": 0.02605, "grad_norm": 0.7440911950642829, "learning_rate": 0.003, "loss": 4.1962, "step": 2605 }, { "epoch": 0.02606, "grad_norm": 0.7799278089279328, "learning_rate": 0.003, "loss": 4.2191, "step": 2606 }, { "epoch": 0.02607, "grad_norm": 0.7713270098545678, "learning_rate": 0.003, "loss": 4.2101, "step": 2607 }, { "epoch": 0.02608, "grad_norm": 0.899578361209406, "learning_rate": 0.003, "loss": 4.1969, "step": 2608 }, { "epoch": 0.02609, "grad_norm": 0.8737732849824265, "learning_rate": 0.003, "loss": 4.24, "step": 2609 }, { "epoch": 0.0261, "grad_norm": 0.8947617166383927, "learning_rate": 0.003, "loss": 4.1906, "step": 2610 }, { "epoch": 0.02611, "grad_norm": 0.8562728807594239, "learning_rate": 0.003, "loss": 4.1837, "step": 2611 }, { "epoch": 0.02612, "grad_norm": 0.7163237085139472, "learning_rate": 0.003, "loss": 4.2141, "step": 2612 }, { "epoch": 0.02613, "grad_norm": 0.7068103471337065, "learning_rate": 0.003, "loss": 4.2043, "step": 2613 }, { "epoch": 0.02614, "grad_norm": 0.7050979670307006, "learning_rate": 0.003, "loss": 4.1848, "step": 2614 }, { "epoch": 0.02615, "grad_norm": 0.6688602437390169, "learning_rate": 0.003, "loss": 4.1711, "step": 2615 }, { "epoch": 0.02616, "grad_norm": 0.7303156076641851, "learning_rate": 0.003, "loss": 4.1875, "step": 2616 }, { "epoch": 0.02617, "grad_norm": 0.8575518403182025, "learning_rate": 0.003, "loss": 4.1807, "step": 2617 }, { "epoch": 0.02618, "grad_norm": 0.7963276560962136, "learning_rate": 0.003, "loss": 4.2077, "step": 2618 }, { "epoch": 0.02619, "grad_norm": 0.7084342029167996, "learning_rate": 0.003, "loss": 4.2176, "step": 2619 }, { "epoch": 0.0262, "grad_norm": 0.5889989743836745, "learning_rate": 0.003, "loss": 4.1739, "step": 2620 }, { "epoch": 0.02621, "grad_norm": 0.5658317825134358, "learning_rate": 0.003, "loss": 4.2007, "step": 2621 }, { "epoch": 0.02622, "grad_norm": 0.5963501897765274, "learning_rate": 0.003, "loss": 4.1995, "step": 2622 }, { "epoch": 0.02623, "grad_norm": 0.6088839510075245, "learning_rate": 0.003, "loss": 4.2019, "step": 2623 }, { "epoch": 0.02624, "grad_norm": 0.639758346515609, "learning_rate": 0.003, "loss": 4.2384, "step": 2624 }, { "epoch": 0.02625, "grad_norm": 0.6760695260129397, "learning_rate": 0.003, "loss": 4.2327, "step": 2625 }, { "epoch": 0.02626, "grad_norm": 0.7394049384951815, "learning_rate": 0.003, "loss": 4.1692, "step": 2626 }, { "epoch": 0.02627, "grad_norm": 0.771569025728823, "learning_rate": 0.003, "loss": 4.1881, "step": 2627 }, { "epoch": 0.02628, "grad_norm": 0.9437506721801893, "learning_rate": 0.003, "loss": 4.2288, "step": 2628 }, { "epoch": 0.02629, "grad_norm": 0.9552775699761499, "learning_rate": 0.003, "loss": 4.1691, "step": 2629 }, { "epoch": 0.0263, "grad_norm": 0.775368472978819, "learning_rate": 0.003, "loss": 4.1937, "step": 2630 }, { "epoch": 0.02631, "grad_norm": 0.7034974913313367, "learning_rate": 0.003, "loss": 4.2143, "step": 2631 }, { "epoch": 0.02632, "grad_norm": 0.84944572579913, "learning_rate": 0.003, "loss": 4.1797, "step": 2632 }, { "epoch": 0.02633, "grad_norm": 0.9315535752380871, "learning_rate": 0.003, "loss": 4.1973, "step": 2633 }, { "epoch": 0.02634, "grad_norm": 0.848318245690014, "learning_rate": 0.003, "loss": 4.2175, "step": 2634 }, { "epoch": 0.02635, "grad_norm": 0.8121240624947638, "learning_rate": 0.003, "loss": 4.2139, "step": 2635 }, { "epoch": 0.02636, "grad_norm": 0.750298491855964, "learning_rate": 0.003, "loss": 4.196, "step": 2636 }, { "epoch": 0.02637, "grad_norm": 0.7715588746790815, "learning_rate": 0.003, "loss": 4.2261, "step": 2637 }, { "epoch": 0.02638, "grad_norm": 0.8941355293873781, "learning_rate": 0.003, "loss": 4.1969, "step": 2638 }, { "epoch": 0.02639, "grad_norm": 1.1208036732650624, "learning_rate": 0.003, "loss": 4.1946, "step": 2639 }, { "epoch": 0.0264, "grad_norm": 0.9871221010165437, "learning_rate": 0.003, "loss": 4.1957, "step": 2640 }, { "epoch": 0.02641, "grad_norm": 0.8874989090023564, "learning_rate": 0.003, "loss": 4.2028, "step": 2641 }, { "epoch": 0.02642, "grad_norm": 0.7458704018113644, "learning_rate": 0.003, "loss": 4.2079, "step": 2642 }, { "epoch": 0.02643, "grad_norm": 0.6742933550823261, "learning_rate": 0.003, "loss": 4.1912, "step": 2643 }, { "epoch": 0.02644, "grad_norm": 0.6392354944191677, "learning_rate": 0.003, "loss": 4.2038, "step": 2644 }, { "epoch": 0.02645, "grad_norm": 0.628167080207442, "learning_rate": 0.003, "loss": 4.2151, "step": 2645 }, { "epoch": 0.02646, "grad_norm": 0.7307984815804015, "learning_rate": 0.003, "loss": 4.1822, "step": 2646 }, { "epoch": 0.02647, "grad_norm": 0.7359159985928708, "learning_rate": 0.003, "loss": 4.2047, "step": 2647 }, { "epoch": 0.02648, "grad_norm": 0.8954441506020118, "learning_rate": 0.003, "loss": 4.2303, "step": 2648 }, { "epoch": 0.02649, "grad_norm": 0.9330516996289556, "learning_rate": 0.003, "loss": 4.2172, "step": 2649 }, { "epoch": 0.0265, "grad_norm": 0.8975744896998445, "learning_rate": 0.003, "loss": 4.1824, "step": 2650 }, { "epoch": 0.02651, "grad_norm": 0.7356166899037149, "learning_rate": 0.003, "loss": 4.1944, "step": 2651 }, { "epoch": 0.02652, "grad_norm": 0.7477158925723587, "learning_rate": 0.003, "loss": 4.2246, "step": 2652 }, { "epoch": 0.02653, "grad_norm": 1.0339979416429652, "learning_rate": 0.003, "loss": 4.1888, "step": 2653 }, { "epoch": 0.02654, "grad_norm": 1.038966179449438, "learning_rate": 0.003, "loss": 4.2084, "step": 2654 }, { "epoch": 0.02655, "grad_norm": 0.9423283848127929, "learning_rate": 0.003, "loss": 4.2139, "step": 2655 }, { "epoch": 0.02656, "grad_norm": 0.8935307253871441, "learning_rate": 0.003, "loss": 4.1861, "step": 2656 }, { "epoch": 0.02657, "grad_norm": 0.9450781561652172, "learning_rate": 0.003, "loss": 4.2216, "step": 2657 }, { "epoch": 0.02658, "grad_norm": 0.9653780583023913, "learning_rate": 0.003, "loss": 4.2275, "step": 2658 }, { "epoch": 0.02659, "grad_norm": 0.9812881650344129, "learning_rate": 0.003, "loss": 4.2266, "step": 2659 }, { "epoch": 0.0266, "grad_norm": 0.7604935125019064, "learning_rate": 0.003, "loss": 4.199, "step": 2660 }, { "epoch": 0.02661, "grad_norm": 0.7898310738914669, "learning_rate": 0.003, "loss": 4.2224, "step": 2661 }, { "epoch": 0.02662, "grad_norm": 0.7821828341075345, "learning_rate": 0.003, "loss": 4.2109, "step": 2662 }, { "epoch": 0.02663, "grad_norm": 0.708129537974612, "learning_rate": 0.003, "loss": 4.191, "step": 2663 }, { "epoch": 0.02664, "grad_norm": 0.614176433467567, "learning_rate": 0.003, "loss": 4.1879, "step": 2664 }, { "epoch": 0.02665, "grad_norm": 0.5871099840572066, "learning_rate": 0.003, "loss": 4.2354, "step": 2665 }, { "epoch": 0.02666, "grad_norm": 0.5413499528000574, "learning_rate": 0.003, "loss": 4.2215, "step": 2666 }, { "epoch": 0.02667, "grad_norm": 0.5004044538450345, "learning_rate": 0.003, "loss": 4.2007, "step": 2667 }, { "epoch": 0.02668, "grad_norm": 0.40369112920994166, "learning_rate": 0.003, "loss": 4.1654, "step": 2668 }, { "epoch": 0.02669, "grad_norm": 0.4696848864082333, "learning_rate": 0.003, "loss": 4.1859, "step": 2669 }, { "epoch": 0.0267, "grad_norm": 0.5279108698543552, "learning_rate": 0.003, "loss": 4.1854, "step": 2670 }, { "epoch": 0.02671, "grad_norm": 0.5815932345300244, "learning_rate": 0.003, "loss": 4.2011, "step": 2671 }, { "epoch": 0.02672, "grad_norm": 0.6492435660477966, "learning_rate": 0.003, "loss": 4.195, "step": 2672 }, { "epoch": 0.02673, "grad_norm": 0.7437813435791281, "learning_rate": 0.003, "loss": 4.1949, "step": 2673 }, { "epoch": 0.02674, "grad_norm": 0.9399474520802658, "learning_rate": 0.003, "loss": 4.1688, "step": 2674 }, { "epoch": 0.02675, "grad_norm": 0.9600957867204232, "learning_rate": 0.003, "loss": 4.1739, "step": 2675 }, { "epoch": 0.02676, "grad_norm": 0.8275264852729368, "learning_rate": 0.003, "loss": 4.1903, "step": 2676 }, { "epoch": 0.02677, "grad_norm": 0.7656126092118313, "learning_rate": 0.003, "loss": 4.2176, "step": 2677 }, { "epoch": 0.02678, "grad_norm": 0.8861030153032051, "learning_rate": 0.003, "loss": 4.2184, "step": 2678 }, { "epoch": 0.02679, "grad_norm": 0.9307488076809898, "learning_rate": 0.003, "loss": 4.1901, "step": 2679 }, { "epoch": 0.0268, "grad_norm": 0.805330899134272, "learning_rate": 0.003, "loss": 4.204, "step": 2680 }, { "epoch": 0.02681, "grad_norm": 0.6764861017583965, "learning_rate": 0.003, "loss": 4.2192, "step": 2681 }, { "epoch": 0.02682, "grad_norm": 0.6705225418393664, "learning_rate": 0.003, "loss": 4.1964, "step": 2682 }, { "epoch": 0.02683, "grad_norm": 0.7147920779432392, "learning_rate": 0.003, "loss": 4.2187, "step": 2683 }, { "epoch": 0.02684, "grad_norm": 0.7227393027644701, "learning_rate": 0.003, "loss": 4.1944, "step": 2684 }, { "epoch": 0.02685, "grad_norm": 0.6704279875962865, "learning_rate": 0.003, "loss": 4.1953, "step": 2685 }, { "epoch": 0.02686, "grad_norm": 0.715031165504719, "learning_rate": 0.003, "loss": 4.1799, "step": 2686 }, { "epoch": 0.02687, "grad_norm": 0.7065638636939903, "learning_rate": 0.003, "loss": 4.1889, "step": 2687 }, { "epoch": 0.02688, "grad_norm": 0.6230800929041215, "learning_rate": 0.003, "loss": 4.1646, "step": 2688 }, { "epoch": 0.02689, "grad_norm": 0.6057809629157821, "learning_rate": 0.003, "loss": 4.1944, "step": 2689 }, { "epoch": 0.0269, "grad_norm": 0.6893536336683767, "learning_rate": 0.003, "loss": 4.2016, "step": 2690 }, { "epoch": 0.02691, "grad_norm": 0.889230963330981, "learning_rate": 0.003, "loss": 4.2018, "step": 2691 }, { "epoch": 0.02692, "grad_norm": 0.9254094540705526, "learning_rate": 0.003, "loss": 4.2128, "step": 2692 }, { "epoch": 0.02693, "grad_norm": 0.7506239085511861, "learning_rate": 0.003, "loss": 4.2167, "step": 2693 }, { "epoch": 0.02694, "grad_norm": 0.8733320575384562, "learning_rate": 0.003, "loss": 4.1861, "step": 2694 }, { "epoch": 0.02695, "grad_norm": 1.0144492752003067, "learning_rate": 0.003, "loss": 4.2059, "step": 2695 }, { "epoch": 0.02696, "grad_norm": 1.1997862653069007, "learning_rate": 0.003, "loss": 4.223, "step": 2696 }, { "epoch": 0.02697, "grad_norm": 0.766270544323272, "learning_rate": 0.003, "loss": 4.2162, "step": 2697 }, { "epoch": 0.02698, "grad_norm": 0.8571240526593925, "learning_rate": 0.003, "loss": 4.2045, "step": 2698 }, { "epoch": 0.02699, "grad_norm": 0.8669468835973597, "learning_rate": 0.003, "loss": 4.2202, "step": 2699 }, { "epoch": 0.027, "grad_norm": 0.8651896057336759, "learning_rate": 0.003, "loss": 4.1922, "step": 2700 }, { "epoch": 0.02701, "grad_norm": 0.9521341410803329, "learning_rate": 0.003, "loss": 4.1854, "step": 2701 }, { "epoch": 0.02702, "grad_norm": 0.8813922958551177, "learning_rate": 0.003, "loss": 4.2351, "step": 2702 }, { "epoch": 0.02703, "grad_norm": 0.7638729271713497, "learning_rate": 0.003, "loss": 4.2319, "step": 2703 }, { "epoch": 0.02704, "grad_norm": 0.6755135996338237, "learning_rate": 0.003, "loss": 4.1703, "step": 2704 }, { "epoch": 0.02705, "grad_norm": 0.7165610100052193, "learning_rate": 0.003, "loss": 4.2247, "step": 2705 }, { "epoch": 0.02706, "grad_norm": 0.7030374550316207, "learning_rate": 0.003, "loss": 4.1863, "step": 2706 }, { "epoch": 0.02707, "grad_norm": 0.7231373056232907, "learning_rate": 0.003, "loss": 4.1831, "step": 2707 }, { "epoch": 0.02708, "grad_norm": 0.7250863254981224, "learning_rate": 0.003, "loss": 4.2139, "step": 2708 }, { "epoch": 0.02709, "grad_norm": 0.7719809697889917, "learning_rate": 0.003, "loss": 4.1893, "step": 2709 }, { "epoch": 0.0271, "grad_norm": 0.7100576834257082, "learning_rate": 0.003, "loss": 4.1967, "step": 2710 }, { "epoch": 0.02711, "grad_norm": 0.7047852834402715, "learning_rate": 0.003, "loss": 4.1784, "step": 2711 }, { "epoch": 0.02712, "grad_norm": 0.6821379465786586, "learning_rate": 0.003, "loss": 4.19, "step": 2712 }, { "epoch": 0.02713, "grad_norm": 0.6215562238389798, "learning_rate": 0.003, "loss": 4.1809, "step": 2713 }, { "epoch": 0.02714, "grad_norm": 0.5936031532564715, "learning_rate": 0.003, "loss": 4.1774, "step": 2714 }, { "epoch": 0.02715, "grad_norm": 0.7208100744430564, "learning_rate": 0.003, "loss": 4.1746, "step": 2715 }, { "epoch": 0.02716, "grad_norm": 0.8524838885321797, "learning_rate": 0.003, "loss": 4.1901, "step": 2716 }, { "epoch": 0.02717, "grad_norm": 0.909492018841103, "learning_rate": 0.003, "loss": 4.1949, "step": 2717 }, { "epoch": 0.02718, "grad_norm": 0.7607068991983567, "learning_rate": 0.003, "loss": 4.2085, "step": 2718 }, { "epoch": 0.02719, "grad_norm": 0.680611263653079, "learning_rate": 0.003, "loss": 4.1801, "step": 2719 }, { "epoch": 0.0272, "grad_norm": 0.8513436164853936, "learning_rate": 0.003, "loss": 4.1774, "step": 2720 }, { "epoch": 0.02721, "grad_norm": 0.8639998466288582, "learning_rate": 0.003, "loss": 4.1904, "step": 2721 }, { "epoch": 0.02722, "grad_norm": 0.8772434078332314, "learning_rate": 0.003, "loss": 4.1929, "step": 2722 }, { "epoch": 0.02723, "grad_norm": 0.7454610624653967, "learning_rate": 0.003, "loss": 4.1955, "step": 2723 }, { "epoch": 0.02724, "grad_norm": 0.6102446662829699, "learning_rate": 0.003, "loss": 4.1921, "step": 2724 }, { "epoch": 0.02725, "grad_norm": 0.7200880629570631, "learning_rate": 0.003, "loss": 4.1879, "step": 2725 }, { "epoch": 0.02726, "grad_norm": 0.8294654813405002, "learning_rate": 0.003, "loss": 4.1624, "step": 2726 }, { "epoch": 0.02727, "grad_norm": 0.8088332759795207, "learning_rate": 0.003, "loss": 4.1952, "step": 2727 }, { "epoch": 0.02728, "grad_norm": 0.7352078940910927, "learning_rate": 0.003, "loss": 4.1816, "step": 2728 }, { "epoch": 0.02729, "grad_norm": 0.6377589758910231, "learning_rate": 0.003, "loss": 4.2114, "step": 2729 }, { "epoch": 0.0273, "grad_norm": 0.5626731480237941, "learning_rate": 0.003, "loss": 4.1797, "step": 2730 }, { "epoch": 0.02731, "grad_norm": 0.601770644577552, "learning_rate": 0.003, "loss": 4.1785, "step": 2731 }, { "epoch": 0.02732, "grad_norm": 0.6533204081603035, "learning_rate": 0.003, "loss": 4.1725, "step": 2732 }, { "epoch": 0.02733, "grad_norm": 0.7772077073193634, "learning_rate": 0.003, "loss": 4.2077, "step": 2733 }, { "epoch": 0.02734, "grad_norm": 0.792467609111156, "learning_rate": 0.003, "loss": 4.1587, "step": 2734 }, { "epoch": 0.02735, "grad_norm": 0.6727759433070883, "learning_rate": 0.003, "loss": 4.2012, "step": 2735 }, { "epoch": 0.02736, "grad_norm": 0.58224974581077, "learning_rate": 0.003, "loss": 4.1727, "step": 2736 }, { "epoch": 0.02737, "grad_norm": 0.6428828815739132, "learning_rate": 0.003, "loss": 4.1675, "step": 2737 }, { "epoch": 0.02738, "grad_norm": 0.6634618804729325, "learning_rate": 0.003, "loss": 4.1909, "step": 2738 }, { "epoch": 0.02739, "grad_norm": 0.7033876856529395, "learning_rate": 0.003, "loss": 4.185, "step": 2739 }, { "epoch": 0.0274, "grad_norm": 0.8647329601476124, "learning_rate": 0.003, "loss": 4.181, "step": 2740 }, { "epoch": 0.02741, "grad_norm": 0.8928671349531622, "learning_rate": 0.003, "loss": 4.1979, "step": 2741 }, { "epoch": 0.02742, "grad_norm": 0.8249061089459991, "learning_rate": 0.003, "loss": 4.1871, "step": 2742 }, { "epoch": 0.02743, "grad_norm": 0.9230002686874078, "learning_rate": 0.003, "loss": 4.2149, "step": 2743 }, { "epoch": 0.02744, "grad_norm": 0.7539080769994848, "learning_rate": 0.003, "loss": 4.1915, "step": 2744 }, { "epoch": 0.02745, "grad_norm": 0.7938564140089303, "learning_rate": 0.003, "loss": 4.2078, "step": 2745 }, { "epoch": 0.02746, "grad_norm": 0.6662256659438859, "learning_rate": 0.003, "loss": 4.1751, "step": 2746 }, { "epoch": 0.02747, "grad_norm": 0.6274900377940358, "learning_rate": 0.003, "loss": 4.211, "step": 2747 }, { "epoch": 0.02748, "grad_norm": 0.5811994917442085, "learning_rate": 0.003, "loss": 4.1596, "step": 2748 }, { "epoch": 0.02749, "grad_norm": 0.5961679125130114, "learning_rate": 0.003, "loss": 4.1658, "step": 2749 }, { "epoch": 0.0275, "grad_norm": 0.6582517345217201, "learning_rate": 0.003, "loss": 4.1997, "step": 2750 }, { "epoch": 0.02751, "grad_norm": 0.6778193569737023, "learning_rate": 0.003, "loss": 4.2149, "step": 2751 }, { "epoch": 0.02752, "grad_norm": 0.7403973900306051, "learning_rate": 0.003, "loss": 4.2023, "step": 2752 }, { "epoch": 0.02753, "grad_norm": 0.9384163074447154, "learning_rate": 0.003, "loss": 4.1905, "step": 2753 }, { "epoch": 0.02754, "grad_norm": 1.548459064020229, "learning_rate": 0.003, "loss": 4.1973, "step": 2754 }, { "epoch": 0.02755, "grad_norm": 0.6860176417927059, "learning_rate": 0.003, "loss": 4.1785, "step": 2755 }, { "epoch": 0.02756, "grad_norm": 0.6932160826643162, "learning_rate": 0.003, "loss": 4.207, "step": 2756 }, { "epoch": 0.02757, "grad_norm": 0.9436043093562364, "learning_rate": 0.003, "loss": 4.2258, "step": 2757 }, { "epoch": 0.02758, "grad_norm": 0.932694610265977, "learning_rate": 0.003, "loss": 4.1741, "step": 2758 }, { "epoch": 0.02759, "grad_norm": 1.0183137769286543, "learning_rate": 0.003, "loss": 4.209, "step": 2759 }, { "epoch": 0.0276, "grad_norm": 0.9568486135614797, "learning_rate": 0.003, "loss": 4.2119, "step": 2760 }, { "epoch": 0.02761, "grad_norm": 0.9337487058436278, "learning_rate": 0.003, "loss": 4.2307, "step": 2761 }, { "epoch": 0.02762, "grad_norm": 0.9026730273084307, "learning_rate": 0.003, "loss": 4.2002, "step": 2762 }, { "epoch": 0.02763, "grad_norm": 0.8685760175250772, "learning_rate": 0.003, "loss": 4.2102, "step": 2763 }, { "epoch": 0.02764, "grad_norm": 0.7723173760613741, "learning_rate": 0.003, "loss": 4.2012, "step": 2764 }, { "epoch": 0.02765, "grad_norm": 0.6561586138290864, "learning_rate": 0.003, "loss": 4.2025, "step": 2765 }, { "epoch": 0.02766, "grad_norm": 0.6637574717745266, "learning_rate": 0.003, "loss": 4.1694, "step": 2766 }, { "epoch": 0.02767, "grad_norm": 0.7045106510208712, "learning_rate": 0.003, "loss": 4.2241, "step": 2767 }, { "epoch": 0.02768, "grad_norm": 0.7181706443439274, "learning_rate": 0.003, "loss": 4.1577, "step": 2768 }, { "epoch": 0.02769, "grad_norm": 0.8144040817598821, "learning_rate": 0.003, "loss": 4.2199, "step": 2769 }, { "epoch": 0.0277, "grad_norm": 0.8055251249780363, "learning_rate": 0.003, "loss": 4.1893, "step": 2770 }, { "epoch": 0.02771, "grad_norm": 0.7962246824499264, "learning_rate": 0.003, "loss": 4.1781, "step": 2771 }, { "epoch": 0.02772, "grad_norm": 0.7090461373687846, "learning_rate": 0.003, "loss": 4.1657, "step": 2772 }, { "epoch": 0.02773, "grad_norm": 0.7104242382637275, "learning_rate": 0.003, "loss": 4.2, "step": 2773 }, { "epoch": 0.02774, "grad_norm": 0.7495245887650404, "learning_rate": 0.003, "loss": 4.2232, "step": 2774 }, { "epoch": 0.02775, "grad_norm": 0.7766387259726212, "learning_rate": 0.003, "loss": 4.1914, "step": 2775 }, { "epoch": 0.02776, "grad_norm": 0.8895259744115902, "learning_rate": 0.003, "loss": 4.1913, "step": 2776 }, { "epoch": 0.02777, "grad_norm": 1.0136447873992764, "learning_rate": 0.003, "loss": 4.1757, "step": 2777 }, { "epoch": 0.02778, "grad_norm": 0.9270708223991836, "learning_rate": 0.003, "loss": 4.2171, "step": 2778 }, { "epoch": 0.02779, "grad_norm": 0.7449945983496725, "learning_rate": 0.003, "loss": 4.1821, "step": 2779 }, { "epoch": 0.0278, "grad_norm": 0.6376494685394721, "learning_rate": 0.003, "loss": 4.1771, "step": 2780 }, { "epoch": 0.02781, "grad_norm": 0.6404601478719466, "learning_rate": 0.003, "loss": 4.1583, "step": 2781 }, { "epoch": 0.02782, "grad_norm": 0.6406219320739934, "learning_rate": 0.003, "loss": 4.1805, "step": 2782 }, { "epoch": 0.02783, "grad_norm": 0.6965574269000856, "learning_rate": 0.003, "loss": 4.1881, "step": 2783 }, { "epoch": 0.02784, "grad_norm": 0.6353952783759534, "learning_rate": 0.003, "loss": 4.18, "step": 2784 }, { "epoch": 0.02785, "grad_norm": 0.6672684060024227, "learning_rate": 0.003, "loss": 4.1968, "step": 2785 }, { "epoch": 0.02786, "grad_norm": 0.8009924222538798, "learning_rate": 0.003, "loss": 4.1936, "step": 2786 }, { "epoch": 0.02787, "grad_norm": 1.0512671252378267, "learning_rate": 0.003, "loss": 4.2056, "step": 2787 }, { "epoch": 0.02788, "grad_norm": 0.9344102358218531, "learning_rate": 0.003, "loss": 4.193, "step": 2788 }, { "epoch": 0.02789, "grad_norm": 0.6966687816191528, "learning_rate": 0.003, "loss": 4.1893, "step": 2789 }, { "epoch": 0.0279, "grad_norm": 0.5812915901595114, "learning_rate": 0.003, "loss": 4.1592, "step": 2790 }, { "epoch": 0.02791, "grad_norm": 0.6498216487712081, "learning_rate": 0.003, "loss": 4.1719, "step": 2791 }, { "epoch": 0.02792, "grad_norm": 0.7276557261785184, "learning_rate": 0.003, "loss": 4.1906, "step": 2792 }, { "epoch": 0.02793, "grad_norm": 0.8633571725468037, "learning_rate": 0.003, "loss": 4.192, "step": 2793 }, { "epoch": 0.02794, "grad_norm": 0.9525705083591802, "learning_rate": 0.003, "loss": 4.1932, "step": 2794 }, { "epoch": 0.02795, "grad_norm": 0.8234824382946722, "learning_rate": 0.003, "loss": 4.2045, "step": 2795 }, { "epoch": 0.02796, "grad_norm": 0.6758642343246372, "learning_rate": 0.003, "loss": 4.194, "step": 2796 }, { "epoch": 0.02797, "grad_norm": 0.5487569332106049, "learning_rate": 0.003, "loss": 4.2, "step": 2797 }, { "epoch": 0.02798, "grad_norm": 0.663303491252161, "learning_rate": 0.003, "loss": 4.2001, "step": 2798 }, { "epoch": 0.02799, "grad_norm": 0.718345645781191, "learning_rate": 0.003, "loss": 4.1857, "step": 2799 }, { "epoch": 0.028, "grad_norm": 0.8551146813140691, "learning_rate": 0.003, "loss": 4.2047, "step": 2800 }, { "epoch": 0.02801, "grad_norm": 0.8363235146801704, "learning_rate": 0.003, "loss": 4.183, "step": 2801 }, { "epoch": 0.02802, "grad_norm": 0.7702327941304277, "learning_rate": 0.003, "loss": 4.1854, "step": 2802 }, { "epoch": 0.02803, "grad_norm": 0.7820522108252399, "learning_rate": 0.003, "loss": 4.2042, "step": 2803 }, { "epoch": 0.02804, "grad_norm": 0.8315360958990581, "learning_rate": 0.003, "loss": 4.1781, "step": 2804 }, { "epoch": 0.02805, "grad_norm": 0.7877464535188091, "learning_rate": 0.003, "loss": 4.2115, "step": 2805 }, { "epoch": 0.02806, "grad_norm": 0.761746586266924, "learning_rate": 0.003, "loss": 4.1891, "step": 2806 }, { "epoch": 0.02807, "grad_norm": 0.7896019466390173, "learning_rate": 0.003, "loss": 4.1905, "step": 2807 }, { "epoch": 0.02808, "grad_norm": 0.8252866588896443, "learning_rate": 0.003, "loss": 4.1764, "step": 2808 }, { "epoch": 0.02809, "grad_norm": 0.8153674183671854, "learning_rate": 0.003, "loss": 4.2161, "step": 2809 }, { "epoch": 0.0281, "grad_norm": 0.8179736763027198, "learning_rate": 0.003, "loss": 4.1676, "step": 2810 }, { "epoch": 0.02811, "grad_norm": 0.7182358526616855, "learning_rate": 0.003, "loss": 4.1746, "step": 2811 }, { "epoch": 0.02812, "grad_norm": 0.7222785348647399, "learning_rate": 0.003, "loss": 4.1904, "step": 2812 }, { "epoch": 0.02813, "grad_norm": 0.7312392563417645, "learning_rate": 0.003, "loss": 4.182, "step": 2813 }, { "epoch": 0.02814, "grad_norm": 0.7621373960798801, "learning_rate": 0.003, "loss": 4.1842, "step": 2814 }, { "epoch": 0.02815, "grad_norm": 0.892573891744398, "learning_rate": 0.003, "loss": 4.2237, "step": 2815 }, { "epoch": 0.02816, "grad_norm": 1.0660491465880368, "learning_rate": 0.003, "loss": 4.2109, "step": 2816 }, { "epoch": 0.02817, "grad_norm": 1.1762397108284373, "learning_rate": 0.003, "loss": 4.1757, "step": 2817 }, { "epoch": 0.02818, "grad_norm": 0.8114644715240712, "learning_rate": 0.003, "loss": 4.1788, "step": 2818 }, { "epoch": 0.02819, "grad_norm": 0.7485947937835542, "learning_rate": 0.003, "loss": 4.2117, "step": 2819 }, { "epoch": 0.0282, "grad_norm": 0.7342650912183687, "learning_rate": 0.003, "loss": 4.1986, "step": 2820 }, { "epoch": 0.02821, "grad_norm": 0.908756776200653, "learning_rate": 0.003, "loss": 4.2223, "step": 2821 }, { "epoch": 0.02822, "grad_norm": 1.1149965537202893, "learning_rate": 0.003, "loss": 4.2047, "step": 2822 }, { "epoch": 0.02823, "grad_norm": 0.9035892234929067, "learning_rate": 0.003, "loss": 4.2252, "step": 2823 }, { "epoch": 0.02824, "grad_norm": 0.8715687787834273, "learning_rate": 0.003, "loss": 4.2131, "step": 2824 }, { "epoch": 0.02825, "grad_norm": 0.9190846271911702, "learning_rate": 0.003, "loss": 4.2162, "step": 2825 }, { "epoch": 0.02826, "grad_norm": 0.9656110778760747, "learning_rate": 0.003, "loss": 4.1982, "step": 2826 }, { "epoch": 0.02827, "grad_norm": 0.7985488107577273, "learning_rate": 0.003, "loss": 4.1811, "step": 2827 }, { "epoch": 0.02828, "grad_norm": 0.6338765134372335, "learning_rate": 0.003, "loss": 4.2122, "step": 2828 }, { "epoch": 0.02829, "grad_norm": 0.5839189064796922, "learning_rate": 0.003, "loss": 4.2333, "step": 2829 }, { "epoch": 0.0283, "grad_norm": 0.5171339276770696, "learning_rate": 0.003, "loss": 4.1933, "step": 2830 }, { "epoch": 0.02831, "grad_norm": 0.4981234575873454, "learning_rate": 0.003, "loss": 4.1933, "step": 2831 }, { "epoch": 0.02832, "grad_norm": 0.48374057680202603, "learning_rate": 0.003, "loss": 4.2024, "step": 2832 }, { "epoch": 0.02833, "grad_norm": 0.49701505240619126, "learning_rate": 0.003, "loss": 4.1763, "step": 2833 }, { "epoch": 0.02834, "grad_norm": 0.4375442041063234, "learning_rate": 0.003, "loss": 4.1612, "step": 2834 }, { "epoch": 0.02835, "grad_norm": 0.4492758317663205, "learning_rate": 0.003, "loss": 4.1724, "step": 2835 }, { "epoch": 0.02836, "grad_norm": 0.5303942295307315, "learning_rate": 0.003, "loss": 4.1912, "step": 2836 }, { "epoch": 0.02837, "grad_norm": 0.6807714027442211, "learning_rate": 0.003, "loss": 4.1659, "step": 2837 }, { "epoch": 0.02838, "grad_norm": 1.054384165651217, "learning_rate": 0.003, "loss": 4.1421, "step": 2838 }, { "epoch": 0.02839, "grad_norm": 1.1296655727309317, "learning_rate": 0.003, "loss": 4.1551, "step": 2839 }, { "epoch": 0.0284, "grad_norm": 0.6812539648150485, "learning_rate": 0.003, "loss": 4.1554, "step": 2840 }, { "epoch": 0.02841, "grad_norm": 0.6214085041332542, "learning_rate": 0.003, "loss": 4.1932, "step": 2841 }, { "epoch": 0.02842, "grad_norm": 0.7921932057609014, "learning_rate": 0.003, "loss": 4.1778, "step": 2842 }, { "epoch": 0.02843, "grad_norm": 0.7844292346786299, "learning_rate": 0.003, "loss": 4.2017, "step": 2843 }, { "epoch": 0.02844, "grad_norm": 0.8202507673079593, "learning_rate": 0.003, "loss": 4.1992, "step": 2844 }, { "epoch": 0.02845, "grad_norm": 0.899571446747117, "learning_rate": 0.003, "loss": 4.1842, "step": 2845 }, { "epoch": 0.02846, "grad_norm": 0.9854433598194003, "learning_rate": 0.003, "loss": 4.178, "step": 2846 }, { "epoch": 0.02847, "grad_norm": 0.9578244699339834, "learning_rate": 0.003, "loss": 4.2412, "step": 2847 }, { "epoch": 0.02848, "grad_norm": 0.8611603069570741, "learning_rate": 0.003, "loss": 4.1991, "step": 2848 }, { "epoch": 0.02849, "grad_norm": 0.8677890731777738, "learning_rate": 0.003, "loss": 4.1875, "step": 2849 }, { "epoch": 0.0285, "grad_norm": 0.9345387668269812, "learning_rate": 0.003, "loss": 4.2004, "step": 2850 }, { "epoch": 0.02851, "grad_norm": 1.022611523078909, "learning_rate": 0.003, "loss": 4.2443, "step": 2851 }, { "epoch": 0.02852, "grad_norm": 1.08179004175546, "learning_rate": 0.003, "loss": 4.2232, "step": 2852 }, { "epoch": 0.02853, "grad_norm": 0.8903248915496341, "learning_rate": 0.003, "loss": 4.1837, "step": 2853 }, { "epoch": 0.02854, "grad_norm": 0.8065798213036774, "learning_rate": 0.003, "loss": 4.222, "step": 2854 }, { "epoch": 0.02855, "grad_norm": 0.8325761138878125, "learning_rate": 0.003, "loss": 4.2217, "step": 2855 }, { "epoch": 0.02856, "grad_norm": 0.8291292758652558, "learning_rate": 0.003, "loss": 4.2197, "step": 2856 }, { "epoch": 0.02857, "grad_norm": 1.0231797870146355, "learning_rate": 0.003, "loss": 4.2081, "step": 2857 }, { "epoch": 0.02858, "grad_norm": 1.135404325827927, "learning_rate": 0.003, "loss": 4.201, "step": 2858 }, { "epoch": 0.02859, "grad_norm": 0.7053658846015558, "learning_rate": 0.003, "loss": 4.1963, "step": 2859 }, { "epoch": 0.0286, "grad_norm": 0.6736135876048694, "learning_rate": 0.003, "loss": 4.2084, "step": 2860 }, { "epoch": 0.02861, "grad_norm": 0.5759862291513759, "learning_rate": 0.003, "loss": 4.1731, "step": 2861 }, { "epoch": 0.02862, "grad_norm": 0.5616808260442542, "learning_rate": 0.003, "loss": 4.1894, "step": 2862 }, { "epoch": 0.02863, "grad_norm": 0.49788279563964805, "learning_rate": 0.003, "loss": 4.2111, "step": 2863 }, { "epoch": 0.02864, "grad_norm": 0.4647689038314713, "learning_rate": 0.003, "loss": 4.1819, "step": 2864 }, { "epoch": 0.02865, "grad_norm": 0.46260884486023074, "learning_rate": 0.003, "loss": 4.2095, "step": 2865 }, { "epoch": 0.02866, "grad_norm": 0.4834529405017652, "learning_rate": 0.003, "loss": 4.19, "step": 2866 }, { "epoch": 0.02867, "grad_norm": 0.5645551159123001, "learning_rate": 0.003, "loss": 4.1773, "step": 2867 }, { "epoch": 0.02868, "grad_norm": 0.660748140178769, "learning_rate": 0.003, "loss": 4.1865, "step": 2868 }, { "epoch": 0.02869, "grad_norm": 0.779662997391975, "learning_rate": 0.003, "loss": 4.1786, "step": 2869 }, { "epoch": 0.0287, "grad_norm": 0.8316388530903273, "learning_rate": 0.003, "loss": 4.2026, "step": 2870 }, { "epoch": 0.02871, "grad_norm": 0.7635742727135532, "learning_rate": 0.003, "loss": 4.1842, "step": 2871 }, { "epoch": 0.02872, "grad_norm": 0.7623759966436428, "learning_rate": 0.003, "loss": 4.2184, "step": 2872 }, { "epoch": 0.02873, "grad_norm": 0.7828397443503498, "learning_rate": 0.003, "loss": 4.1929, "step": 2873 }, { "epoch": 0.02874, "grad_norm": 0.7046479053883313, "learning_rate": 0.003, "loss": 4.1997, "step": 2874 }, { "epoch": 0.02875, "grad_norm": 0.6539896261531868, "learning_rate": 0.003, "loss": 4.2062, "step": 2875 }, { "epoch": 0.02876, "grad_norm": 0.5905081172799793, "learning_rate": 0.003, "loss": 4.1857, "step": 2876 }, { "epoch": 0.02877, "grad_norm": 0.5633633650090469, "learning_rate": 0.003, "loss": 4.1675, "step": 2877 }, { "epoch": 0.02878, "grad_norm": 0.6177245641938598, "learning_rate": 0.003, "loss": 4.1622, "step": 2878 }, { "epoch": 0.02879, "grad_norm": 0.6747485726027824, "learning_rate": 0.003, "loss": 4.1563, "step": 2879 }, { "epoch": 0.0288, "grad_norm": 0.9112144299725644, "learning_rate": 0.003, "loss": 4.1679, "step": 2880 }, { "epoch": 0.02881, "grad_norm": 1.1185243563448626, "learning_rate": 0.003, "loss": 4.1965, "step": 2881 }, { "epoch": 0.02882, "grad_norm": 0.7715679266943174, "learning_rate": 0.003, "loss": 4.2008, "step": 2882 }, { "epoch": 0.02883, "grad_norm": 0.8113547513229802, "learning_rate": 0.003, "loss": 4.1841, "step": 2883 }, { "epoch": 0.02884, "grad_norm": 0.9501941505139021, "learning_rate": 0.003, "loss": 4.2319, "step": 2884 }, { "epoch": 0.02885, "grad_norm": 0.8794293648016848, "learning_rate": 0.003, "loss": 4.1687, "step": 2885 }, { "epoch": 0.02886, "grad_norm": 0.6791421144716309, "learning_rate": 0.003, "loss": 4.1926, "step": 2886 }, { "epoch": 0.02887, "grad_norm": 0.7217161523796813, "learning_rate": 0.003, "loss": 4.1813, "step": 2887 }, { "epoch": 0.02888, "grad_norm": 0.6947494994368266, "learning_rate": 0.003, "loss": 4.1758, "step": 2888 }, { "epoch": 0.02889, "grad_norm": 0.736779142809904, "learning_rate": 0.003, "loss": 4.168, "step": 2889 }, { "epoch": 0.0289, "grad_norm": 0.7371012950268315, "learning_rate": 0.003, "loss": 4.1974, "step": 2890 }, { "epoch": 0.02891, "grad_norm": 0.7996974659760072, "learning_rate": 0.003, "loss": 4.217, "step": 2891 }, { "epoch": 0.02892, "grad_norm": 0.8033610158227403, "learning_rate": 0.003, "loss": 4.1701, "step": 2892 }, { "epoch": 0.02893, "grad_norm": 0.7086173042508274, "learning_rate": 0.003, "loss": 4.1738, "step": 2893 }, { "epoch": 0.02894, "grad_norm": 0.6418053792744189, "learning_rate": 0.003, "loss": 4.1778, "step": 2894 }, { "epoch": 0.02895, "grad_norm": 0.6478672327859655, "learning_rate": 0.003, "loss": 4.2197, "step": 2895 }, { "epoch": 0.02896, "grad_norm": 0.640640821327547, "learning_rate": 0.003, "loss": 4.1749, "step": 2896 }, { "epoch": 0.02897, "grad_norm": 0.5304859237589287, "learning_rate": 0.003, "loss": 4.1548, "step": 2897 }, { "epoch": 0.02898, "grad_norm": 0.5580042897470411, "learning_rate": 0.003, "loss": 4.1966, "step": 2898 }, { "epoch": 0.02899, "grad_norm": 0.7047206692629422, "learning_rate": 0.003, "loss": 4.1829, "step": 2899 }, { "epoch": 0.029, "grad_norm": 0.8635840660671906, "learning_rate": 0.003, "loss": 4.1794, "step": 2900 }, { "epoch": 0.02901, "grad_norm": 0.9189375044658029, "learning_rate": 0.003, "loss": 4.1972, "step": 2901 }, { "epoch": 0.02902, "grad_norm": 0.708585231495253, "learning_rate": 0.003, "loss": 4.2109, "step": 2902 }, { "epoch": 0.02903, "grad_norm": 0.6446440355575891, "learning_rate": 0.003, "loss": 4.1679, "step": 2903 }, { "epoch": 0.02904, "grad_norm": 0.8782661322489321, "learning_rate": 0.003, "loss": 4.1888, "step": 2904 }, { "epoch": 0.02905, "grad_norm": 0.9531560335950827, "learning_rate": 0.003, "loss": 4.2128, "step": 2905 }, { "epoch": 0.02906, "grad_norm": 0.855726406973352, "learning_rate": 0.003, "loss": 4.192, "step": 2906 }, { "epoch": 0.02907, "grad_norm": 0.8360790928903014, "learning_rate": 0.003, "loss": 4.168, "step": 2907 }, { "epoch": 0.02908, "grad_norm": 1.006275418407078, "learning_rate": 0.003, "loss": 4.1962, "step": 2908 }, { "epoch": 0.02909, "grad_norm": 1.069782048483947, "learning_rate": 0.003, "loss": 4.1953, "step": 2909 }, { "epoch": 0.0291, "grad_norm": 0.9075229413559223, "learning_rate": 0.003, "loss": 4.2047, "step": 2910 }, { "epoch": 0.02911, "grad_norm": 0.8275942965443872, "learning_rate": 0.003, "loss": 4.1994, "step": 2911 }, { "epoch": 0.02912, "grad_norm": 0.8193725896423978, "learning_rate": 0.003, "loss": 4.1688, "step": 2912 }, { "epoch": 0.02913, "grad_norm": 0.8507541445865096, "learning_rate": 0.003, "loss": 4.205, "step": 2913 }, { "epoch": 0.02914, "grad_norm": 0.8452395132518625, "learning_rate": 0.003, "loss": 4.1928, "step": 2914 }, { "epoch": 0.02915, "grad_norm": 0.9311887044384434, "learning_rate": 0.003, "loss": 4.1964, "step": 2915 }, { "epoch": 0.02916, "grad_norm": 0.8243141989685264, "learning_rate": 0.003, "loss": 4.1979, "step": 2916 }, { "epoch": 0.02917, "grad_norm": 0.8091183982304383, "learning_rate": 0.003, "loss": 4.2266, "step": 2917 }, { "epoch": 0.02918, "grad_norm": 0.7297783258634746, "learning_rate": 0.003, "loss": 4.1808, "step": 2918 }, { "epoch": 0.02919, "grad_norm": 0.7037586226632049, "learning_rate": 0.003, "loss": 4.2193, "step": 2919 }, { "epoch": 0.0292, "grad_norm": 0.7593009293515595, "learning_rate": 0.003, "loss": 4.1588, "step": 2920 }, { "epoch": 0.02921, "grad_norm": 0.7917149655134618, "learning_rate": 0.003, "loss": 4.2005, "step": 2921 }, { "epoch": 0.02922, "grad_norm": 0.9187945531692804, "learning_rate": 0.003, "loss": 4.2011, "step": 2922 }, { "epoch": 0.02923, "grad_norm": 1.0139939826189068, "learning_rate": 0.003, "loss": 4.1831, "step": 2923 }, { "epoch": 0.02924, "grad_norm": 1.2164555418920324, "learning_rate": 0.003, "loss": 4.2224, "step": 2924 }, { "epoch": 0.02925, "grad_norm": 0.7496642706044158, "learning_rate": 0.003, "loss": 4.215, "step": 2925 }, { "epoch": 0.02926, "grad_norm": 0.6478065305573352, "learning_rate": 0.003, "loss": 4.1896, "step": 2926 }, { "epoch": 0.02927, "grad_norm": 0.7254565724940762, "learning_rate": 0.003, "loss": 4.1952, "step": 2927 }, { "epoch": 0.02928, "grad_norm": 0.8685255954161141, "learning_rate": 0.003, "loss": 4.2213, "step": 2928 }, { "epoch": 0.02929, "grad_norm": 0.9237416579241535, "learning_rate": 0.003, "loss": 4.2095, "step": 2929 }, { "epoch": 0.0293, "grad_norm": 0.8372395013359059, "learning_rate": 0.003, "loss": 4.1992, "step": 2930 }, { "epoch": 0.02931, "grad_norm": 0.7932357657287773, "learning_rate": 0.003, "loss": 4.1982, "step": 2931 }, { "epoch": 0.02932, "grad_norm": 0.7073630933550545, "learning_rate": 0.003, "loss": 4.1822, "step": 2932 }, { "epoch": 0.02933, "grad_norm": 0.7308618206431018, "learning_rate": 0.003, "loss": 4.2074, "step": 2933 }, { "epoch": 0.02934, "grad_norm": 0.6003080161076961, "learning_rate": 0.003, "loss": 4.1908, "step": 2934 }, { "epoch": 0.02935, "grad_norm": 0.5616005585738562, "learning_rate": 0.003, "loss": 4.1759, "step": 2935 }, { "epoch": 0.02936, "grad_norm": 0.5762089224720499, "learning_rate": 0.003, "loss": 4.193, "step": 2936 }, { "epoch": 0.02937, "grad_norm": 0.6340624962039538, "learning_rate": 0.003, "loss": 4.1748, "step": 2937 }, { "epoch": 0.02938, "grad_norm": 0.5985017877094587, "learning_rate": 0.003, "loss": 4.1887, "step": 2938 }, { "epoch": 0.02939, "grad_norm": 0.5822648909618956, "learning_rate": 0.003, "loss": 4.1748, "step": 2939 }, { "epoch": 0.0294, "grad_norm": 0.5818506972026382, "learning_rate": 0.003, "loss": 4.1973, "step": 2940 }, { "epoch": 0.02941, "grad_norm": 0.5990628549146402, "learning_rate": 0.003, "loss": 4.1977, "step": 2941 }, { "epoch": 0.02942, "grad_norm": 0.5191939816216448, "learning_rate": 0.003, "loss": 4.1819, "step": 2942 }, { "epoch": 0.02943, "grad_norm": 0.4807951045470019, "learning_rate": 0.003, "loss": 4.1672, "step": 2943 }, { "epoch": 0.02944, "grad_norm": 0.5163327805251324, "learning_rate": 0.003, "loss": 4.1874, "step": 2944 }, { "epoch": 0.02945, "grad_norm": 0.6847252582164058, "learning_rate": 0.003, "loss": 4.1723, "step": 2945 }, { "epoch": 0.02946, "grad_norm": 1.187036076408125, "learning_rate": 0.003, "loss": 4.1786, "step": 2946 }, { "epoch": 0.02947, "grad_norm": 1.1413035518662917, "learning_rate": 0.003, "loss": 4.2213, "step": 2947 }, { "epoch": 0.02948, "grad_norm": 0.701324069469179, "learning_rate": 0.003, "loss": 4.1882, "step": 2948 }, { "epoch": 0.02949, "grad_norm": 0.8381653938972113, "learning_rate": 0.003, "loss": 4.1676, "step": 2949 }, { "epoch": 0.0295, "grad_norm": 1.0096266796093847, "learning_rate": 0.003, "loss": 4.1958, "step": 2950 }, { "epoch": 0.02951, "grad_norm": 0.8969101683640668, "learning_rate": 0.003, "loss": 4.1443, "step": 2951 }, { "epoch": 0.02952, "grad_norm": 0.7052992950548165, "learning_rate": 0.003, "loss": 4.1731, "step": 2952 }, { "epoch": 0.02953, "grad_norm": 0.7325132030616415, "learning_rate": 0.003, "loss": 4.1805, "step": 2953 }, { "epoch": 0.02954, "grad_norm": 0.750908559373497, "learning_rate": 0.003, "loss": 4.2202, "step": 2954 }, { "epoch": 0.02955, "grad_norm": 0.8349240528523505, "learning_rate": 0.003, "loss": 4.1975, "step": 2955 }, { "epoch": 0.02956, "grad_norm": 0.8858972744774044, "learning_rate": 0.003, "loss": 4.1874, "step": 2956 }, { "epoch": 0.02957, "grad_norm": 0.7827116360148241, "learning_rate": 0.003, "loss": 4.2001, "step": 2957 }, { "epoch": 0.02958, "grad_norm": 0.5580040747767873, "learning_rate": 0.003, "loss": 4.2039, "step": 2958 }, { "epoch": 0.02959, "grad_norm": 0.6281493340504504, "learning_rate": 0.003, "loss": 4.1953, "step": 2959 }, { "epoch": 0.0296, "grad_norm": 0.705986539007765, "learning_rate": 0.003, "loss": 4.2031, "step": 2960 }, { "epoch": 0.02961, "grad_norm": 0.7736093332970924, "learning_rate": 0.003, "loss": 4.176, "step": 2961 }, { "epoch": 0.02962, "grad_norm": 0.8407411528623118, "learning_rate": 0.003, "loss": 4.1877, "step": 2962 }, { "epoch": 0.02963, "grad_norm": 0.8815056405402751, "learning_rate": 0.003, "loss": 4.165, "step": 2963 }, { "epoch": 0.02964, "grad_norm": 0.8322635911509938, "learning_rate": 0.003, "loss": 4.1498, "step": 2964 }, { "epoch": 0.02965, "grad_norm": 0.7078370731458429, "learning_rate": 0.003, "loss": 4.1928, "step": 2965 }, { "epoch": 0.02966, "grad_norm": 0.8092500020138753, "learning_rate": 0.003, "loss": 4.2062, "step": 2966 }, { "epoch": 0.02967, "grad_norm": 0.8427099437715561, "learning_rate": 0.003, "loss": 4.1949, "step": 2967 }, { "epoch": 0.02968, "grad_norm": 0.8731297599816912, "learning_rate": 0.003, "loss": 4.1942, "step": 2968 }, { "epoch": 0.02969, "grad_norm": 0.9059211009671025, "learning_rate": 0.003, "loss": 4.1953, "step": 2969 }, { "epoch": 0.0297, "grad_norm": 0.9050829103933709, "learning_rate": 0.003, "loss": 4.1813, "step": 2970 }, { "epoch": 0.02971, "grad_norm": 0.8269889845972395, "learning_rate": 0.003, "loss": 4.1982, "step": 2971 }, { "epoch": 0.02972, "grad_norm": 0.9293646500807372, "learning_rate": 0.003, "loss": 4.1751, "step": 2972 }, { "epoch": 0.02973, "grad_norm": 0.8613265194586818, "learning_rate": 0.003, "loss": 4.158, "step": 2973 }, { "epoch": 0.02974, "grad_norm": 0.7340587403227173, "learning_rate": 0.003, "loss": 4.1916, "step": 2974 }, { "epoch": 0.02975, "grad_norm": 0.8605959770191105, "learning_rate": 0.003, "loss": 4.1821, "step": 2975 }, { "epoch": 0.02976, "grad_norm": 0.90470438565766, "learning_rate": 0.003, "loss": 4.1986, "step": 2976 }, { "epoch": 0.02977, "grad_norm": 0.8424713452839194, "learning_rate": 0.003, "loss": 4.1814, "step": 2977 }, { "epoch": 0.02978, "grad_norm": 0.8244805201805542, "learning_rate": 0.003, "loss": 4.2015, "step": 2978 }, { "epoch": 0.02979, "grad_norm": 0.776777955492617, "learning_rate": 0.003, "loss": 4.184, "step": 2979 }, { "epoch": 0.0298, "grad_norm": 0.8027239829996253, "learning_rate": 0.003, "loss": 4.1903, "step": 2980 }, { "epoch": 0.02981, "grad_norm": 0.8379341642233197, "learning_rate": 0.003, "loss": 4.1995, "step": 2981 }, { "epoch": 0.02982, "grad_norm": 0.8371771421098021, "learning_rate": 0.003, "loss": 4.1969, "step": 2982 }, { "epoch": 0.02983, "grad_norm": 0.963864757218071, "learning_rate": 0.003, "loss": 4.215, "step": 2983 }, { "epoch": 0.02984, "grad_norm": 0.9620939482544487, "learning_rate": 0.003, "loss": 4.1919, "step": 2984 }, { "epoch": 0.02985, "grad_norm": 0.7838854043078648, "learning_rate": 0.003, "loss": 4.2185, "step": 2985 }, { "epoch": 0.02986, "grad_norm": 0.5932011838530595, "learning_rate": 0.003, "loss": 4.1694, "step": 2986 }, { "epoch": 0.02987, "grad_norm": 0.6307482323122638, "learning_rate": 0.003, "loss": 4.1654, "step": 2987 }, { "epoch": 0.02988, "grad_norm": 0.774070001834325, "learning_rate": 0.003, "loss": 4.178, "step": 2988 }, { "epoch": 0.02989, "grad_norm": 0.9409000008465417, "learning_rate": 0.003, "loss": 4.1972, "step": 2989 }, { "epoch": 0.0299, "grad_norm": 0.8897814762817067, "learning_rate": 0.003, "loss": 4.1573, "step": 2990 }, { "epoch": 0.02991, "grad_norm": 0.7283012653732146, "learning_rate": 0.003, "loss": 4.2071, "step": 2991 }, { "epoch": 0.02992, "grad_norm": 0.8056037736979436, "learning_rate": 0.003, "loss": 4.2066, "step": 2992 }, { "epoch": 0.02993, "grad_norm": 0.8453294693115473, "learning_rate": 0.003, "loss": 4.1756, "step": 2993 }, { "epoch": 0.02994, "grad_norm": 0.7681040966163926, "learning_rate": 0.003, "loss": 4.1818, "step": 2994 }, { "epoch": 0.02995, "grad_norm": 0.737046402942997, "learning_rate": 0.003, "loss": 4.2193, "step": 2995 }, { "epoch": 0.02996, "grad_norm": 0.8016830093512028, "learning_rate": 0.003, "loss": 4.2066, "step": 2996 }, { "epoch": 0.02997, "grad_norm": 0.9076425430381243, "learning_rate": 0.003, "loss": 4.1734, "step": 2997 }, { "epoch": 0.02998, "grad_norm": 0.9119888579778009, "learning_rate": 0.003, "loss": 4.1736, "step": 2998 }, { "epoch": 0.02999, "grad_norm": 0.8734639454385718, "learning_rate": 0.003, "loss": 4.1667, "step": 2999 }, { "epoch": 0.03, "grad_norm": 0.8753881822879969, "learning_rate": 0.003, "loss": 4.2133, "step": 3000 }, { "epoch": 0.03001, "grad_norm": 0.6613925712555209, "learning_rate": 0.003, "loss": 4.1755, "step": 3001 }, { "epoch": 0.03002, "grad_norm": 0.6268534594644447, "learning_rate": 0.003, "loss": 4.1859, "step": 3002 }, { "epoch": 0.03003, "grad_norm": 0.7131894634569484, "learning_rate": 0.003, "loss": 4.1681, "step": 3003 }, { "epoch": 0.03004, "grad_norm": 0.7222298682350133, "learning_rate": 0.003, "loss": 4.1458, "step": 3004 }, { "epoch": 0.03005, "grad_norm": 0.7851659647153477, "learning_rate": 0.003, "loss": 4.1973, "step": 3005 }, { "epoch": 0.03006, "grad_norm": 0.9823948008432, "learning_rate": 0.003, "loss": 4.1985, "step": 3006 }, { "epoch": 0.03007, "grad_norm": 1.0677576033708733, "learning_rate": 0.003, "loss": 4.1791, "step": 3007 }, { "epoch": 0.03008, "grad_norm": 0.765529440508669, "learning_rate": 0.003, "loss": 4.1951, "step": 3008 }, { "epoch": 0.03009, "grad_norm": 0.7479572849164975, "learning_rate": 0.003, "loss": 4.1694, "step": 3009 }, { "epoch": 0.0301, "grad_norm": 0.7406412396521943, "learning_rate": 0.003, "loss": 4.194, "step": 3010 }, { "epoch": 0.03011, "grad_norm": 0.7262544536437181, "learning_rate": 0.003, "loss": 4.1743, "step": 3011 }, { "epoch": 0.03012, "grad_norm": 0.7544052154649398, "learning_rate": 0.003, "loss": 4.1816, "step": 3012 }, { "epoch": 0.03013, "grad_norm": 0.7162089680767648, "learning_rate": 0.003, "loss": 4.1417, "step": 3013 }, { "epoch": 0.03014, "grad_norm": 0.6940516993879415, "learning_rate": 0.003, "loss": 4.1682, "step": 3014 }, { "epoch": 0.03015, "grad_norm": 0.6844442507390428, "learning_rate": 0.003, "loss": 4.1663, "step": 3015 }, { "epoch": 0.03016, "grad_norm": 0.5615506405814443, "learning_rate": 0.003, "loss": 4.171, "step": 3016 }, { "epoch": 0.03017, "grad_norm": 0.6754454864107804, "learning_rate": 0.003, "loss": 4.1716, "step": 3017 }, { "epoch": 0.03018, "grad_norm": 0.768819694641026, "learning_rate": 0.003, "loss": 4.1698, "step": 3018 }, { "epoch": 0.03019, "grad_norm": 0.7485745058334217, "learning_rate": 0.003, "loss": 4.1872, "step": 3019 }, { "epoch": 0.0302, "grad_norm": 0.7254394591812489, "learning_rate": 0.003, "loss": 4.1737, "step": 3020 }, { "epoch": 0.03021, "grad_norm": 0.8549519124776134, "learning_rate": 0.003, "loss": 4.2019, "step": 3021 }, { "epoch": 0.03022, "grad_norm": 1.0108282452619028, "learning_rate": 0.003, "loss": 4.2049, "step": 3022 }, { "epoch": 0.03023, "grad_norm": 1.1260058425339694, "learning_rate": 0.003, "loss": 4.2268, "step": 3023 }, { "epoch": 0.03024, "grad_norm": 0.8578380008226317, "learning_rate": 0.003, "loss": 4.2105, "step": 3024 }, { "epoch": 0.03025, "grad_norm": 0.8141336248728617, "learning_rate": 0.003, "loss": 4.1943, "step": 3025 }, { "epoch": 0.03026, "grad_norm": 0.8343860321958456, "learning_rate": 0.003, "loss": 4.1674, "step": 3026 }, { "epoch": 0.03027, "grad_norm": 0.7174969120361776, "learning_rate": 0.003, "loss": 4.1858, "step": 3027 }, { "epoch": 0.03028, "grad_norm": 0.7652490077013342, "learning_rate": 0.003, "loss": 4.1482, "step": 3028 }, { "epoch": 0.03029, "grad_norm": 0.8261933696416869, "learning_rate": 0.003, "loss": 4.1934, "step": 3029 }, { "epoch": 0.0303, "grad_norm": 0.8845522708435402, "learning_rate": 0.003, "loss": 4.1774, "step": 3030 }, { "epoch": 0.03031, "grad_norm": 0.8283071846166148, "learning_rate": 0.003, "loss": 4.1606, "step": 3031 }, { "epoch": 0.03032, "grad_norm": 0.7987428540661371, "learning_rate": 0.003, "loss": 4.1851, "step": 3032 }, { "epoch": 0.03033, "grad_norm": 0.792150443356606, "learning_rate": 0.003, "loss": 4.1782, "step": 3033 }, { "epoch": 0.03034, "grad_norm": 0.7134255568100923, "learning_rate": 0.003, "loss": 4.1822, "step": 3034 }, { "epoch": 0.03035, "grad_norm": 0.5893848032969043, "learning_rate": 0.003, "loss": 4.2068, "step": 3035 }, { "epoch": 0.03036, "grad_norm": 0.5971843518344567, "learning_rate": 0.003, "loss": 4.178, "step": 3036 }, { "epoch": 0.03037, "grad_norm": 0.5650725934297753, "learning_rate": 0.003, "loss": 4.161, "step": 3037 }, { "epoch": 0.03038, "grad_norm": 0.6879606089602287, "learning_rate": 0.003, "loss": 4.1635, "step": 3038 }, { "epoch": 0.03039, "grad_norm": 0.893681525755944, "learning_rate": 0.003, "loss": 4.1741, "step": 3039 }, { "epoch": 0.0304, "grad_norm": 1.133698039918333, "learning_rate": 0.003, "loss": 4.192, "step": 3040 }, { "epoch": 0.03041, "grad_norm": 0.8486015662534218, "learning_rate": 0.003, "loss": 4.1894, "step": 3041 }, { "epoch": 0.03042, "grad_norm": 0.6236514545633368, "learning_rate": 0.003, "loss": 4.2098, "step": 3042 }, { "epoch": 0.03043, "grad_norm": 0.6770107818972176, "learning_rate": 0.003, "loss": 4.1618, "step": 3043 }, { "epoch": 0.03044, "grad_norm": 0.7747790040576381, "learning_rate": 0.003, "loss": 4.1619, "step": 3044 }, { "epoch": 0.03045, "grad_norm": 0.8128614331690203, "learning_rate": 0.003, "loss": 4.1673, "step": 3045 }, { "epoch": 0.03046, "grad_norm": 0.7607067403493086, "learning_rate": 0.003, "loss": 4.1756, "step": 3046 }, { "epoch": 0.03047, "grad_norm": 0.749167754293152, "learning_rate": 0.003, "loss": 4.1823, "step": 3047 }, { "epoch": 0.03048, "grad_norm": 0.8841149975528951, "learning_rate": 0.003, "loss": 4.1695, "step": 3048 }, { "epoch": 0.03049, "grad_norm": 0.902586510586676, "learning_rate": 0.003, "loss": 4.1608, "step": 3049 }, { "epoch": 0.0305, "grad_norm": 0.918737753442424, "learning_rate": 0.003, "loss": 4.1801, "step": 3050 }, { "epoch": 0.03051, "grad_norm": 0.8824746251699788, "learning_rate": 0.003, "loss": 4.1888, "step": 3051 }, { "epoch": 0.03052, "grad_norm": 0.8446124122566411, "learning_rate": 0.003, "loss": 4.1969, "step": 3052 }, { "epoch": 0.03053, "grad_norm": 0.7873211881465576, "learning_rate": 0.003, "loss": 4.1977, "step": 3053 }, { "epoch": 0.03054, "grad_norm": 0.8281204675386242, "learning_rate": 0.003, "loss": 4.1831, "step": 3054 }, { "epoch": 0.03055, "grad_norm": 0.8380195613474594, "learning_rate": 0.003, "loss": 4.1856, "step": 3055 }, { "epoch": 0.03056, "grad_norm": 0.8532831277535057, "learning_rate": 0.003, "loss": 4.1983, "step": 3056 }, { "epoch": 0.03057, "grad_norm": 0.8740896523349351, "learning_rate": 0.003, "loss": 4.1932, "step": 3057 }, { "epoch": 0.03058, "grad_norm": 0.9800189086277834, "learning_rate": 0.003, "loss": 4.1746, "step": 3058 }, { "epoch": 0.03059, "grad_norm": 0.9987130256226004, "learning_rate": 0.003, "loss": 4.183, "step": 3059 }, { "epoch": 0.0306, "grad_norm": 0.9698665311539535, "learning_rate": 0.003, "loss": 4.1715, "step": 3060 }, { "epoch": 0.03061, "grad_norm": 0.9452499042848614, "learning_rate": 0.003, "loss": 4.1983, "step": 3061 }, { "epoch": 0.03062, "grad_norm": 0.8545512015880272, "learning_rate": 0.003, "loss": 4.221, "step": 3062 }, { "epoch": 0.03063, "grad_norm": 0.8263018139013485, "learning_rate": 0.003, "loss": 4.1631, "step": 3063 }, { "epoch": 0.03064, "grad_norm": 0.8252780037841233, "learning_rate": 0.003, "loss": 4.196, "step": 3064 }, { "epoch": 0.03065, "grad_norm": 0.6876624447118844, "learning_rate": 0.003, "loss": 4.1593, "step": 3065 }, { "epoch": 0.03066, "grad_norm": 0.6367341430832967, "learning_rate": 0.003, "loss": 4.2021, "step": 3066 }, { "epoch": 0.03067, "grad_norm": 0.6017521480419693, "learning_rate": 0.003, "loss": 4.2046, "step": 3067 }, { "epoch": 0.03068, "grad_norm": 0.6343160042573421, "learning_rate": 0.003, "loss": 4.1735, "step": 3068 }, { "epoch": 0.03069, "grad_norm": 0.7404679311351042, "learning_rate": 0.003, "loss": 4.159, "step": 3069 }, { "epoch": 0.0307, "grad_norm": 0.9866650257634313, "learning_rate": 0.003, "loss": 4.1769, "step": 3070 }, { "epoch": 0.03071, "grad_norm": 1.1343017898084369, "learning_rate": 0.003, "loss": 4.1897, "step": 3071 }, { "epoch": 0.03072, "grad_norm": 0.8123202985082579, "learning_rate": 0.003, "loss": 4.1861, "step": 3072 }, { "epoch": 0.03073, "grad_norm": 0.7144691968854603, "learning_rate": 0.003, "loss": 4.179, "step": 3073 }, { "epoch": 0.03074, "grad_norm": 0.6377413488302888, "learning_rate": 0.003, "loss": 4.1781, "step": 3074 }, { "epoch": 0.03075, "grad_norm": 0.5884624483293499, "learning_rate": 0.003, "loss": 4.1988, "step": 3075 }, { "epoch": 0.03076, "grad_norm": 0.6087144704470088, "learning_rate": 0.003, "loss": 4.1903, "step": 3076 }, { "epoch": 0.03077, "grad_norm": 0.6783576790084909, "learning_rate": 0.003, "loss": 4.187, "step": 3077 }, { "epoch": 0.03078, "grad_norm": 0.788987640301813, "learning_rate": 0.003, "loss": 4.1748, "step": 3078 }, { "epoch": 0.03079, "grad_norm": 0.8815514064205557, "learning_rate": 0.003, "loss": 4.167, "step": 3079 }, { "epoch": 0.0308, "grad_norm": 0.9464779295877167, "learning_rate": 0.003, "loss": 4.1986, "step": 3080 }, { "epoch": 0.03081, "grad_norm": 0.8460781162865396, "learning_rate": 0.003, "loss": 4.1704, "step": 3081 }, { "epoch": 0.03082, "grad_norm": 0.661301125290537, "learning_rate": 0.003, "loss": 4.1657, "step": 3082 }, { "epoch": 0.03083, "grad_norm": 0.6453550230590483, "learning_rate": 0.003, "loss": 4.1674, "step": 3083 }, { "epoch": 0.03084, "grad_norm": 0.6680921770876664, "learning_rate": 0.003, "loss": 4.1802, "step": 3084 }, { "epoch": 0.03085, "grad_norm": 0.717538128453225, "learning_rate": 0.003, "loss": 4.1726, "step": 3085 }, { "epoch": 0.03086, "grad_norm": 0.7497027610096151, "learning_rate": 0.003, "loss": 4.1995, "step": 3086 }, { "epoch": 0.03087, "grad_norm": 0.7866492351798048, "learning_rate": 0.003, "loss": 4.1738, "step": 3087 }, { "epoch": 0.03088, "grad_norm": 0.8335106050070741, "learning_rate": 0.003, "loss": 4.1936, "step": 3088 }, { "epoch": 0.03089, "grad_norm": 0.8694998639476388, "learning_rate": 0.003, "loss": 4.1593, "step": 3089 }, { "epoch": 0.0309, "grad_norm": 0.7525867056979888, "learning_rate": 0.003, "loss": 4.1633, "step": 3090 }, { "epoch": 0.03091, "grad_norm": 0.6460307539732922, "learning_rate": 0.003, "loss": 4.1235, "step": 3091 }, { "epoch": 0.03092, "grad_norm": 0.7095786123893822, "learning_rate": 0.003, "loss": 4.1521, "step": 3092 }, { "epoch": 0.03093, "grad_norm": 0.7678464064388238, "learning_rate": 0.003, "loss": 4.1891, "step": 3093 }, { "epoch": 0.03094, "grad_norm": 0.7114573750951001, "learning_rate": 0.003, "loss": 4.1689, "step": 3094 }, { "epoch": 0.03095, "grad_norm": 0.6317515456248899, "learning_rate": 0.003, "loss": 4.1793, "step": 3095 }, { "epoch": 0.03096, "grad_norm": 0.6225934754004887, "learning_rate": 0.003, "loss": 4.1574, "step": 3096 }, { "epoch": 0.03097, "grad_norm": 0.6381632322240677, "learning_rate": 0.003, "loss": 4.1813, "step": 3097 }, { "epoch": 0.03098, "grad_norm": 0.5493766658656782, "learning_rate": 0.003, "loss": 4.1746, "step": 3098 }, { "epoch": 0.03099, "grad_norm": 0.5474679793693559, "learning_rate": 0.003, "loss": 4.1656, "step": 3099 }, { "epoch": 0.031, "grad_norm": 0.5637661199247416, "learning_rate": 0.003, "loss": 4.1627, "step": 3100 }, { "epoch": 0.03101, "grad_norm": 0.6168936290521686, "learning_rate": 0.003, "loss": 4.1562, "step": 3101 }, { "epoch": 0.03102, "grad_norm": 0.7030123347970183, "learning_rate": 0.003, "loss": 4.1584, "step": 3102 }, { "epoch": 0.03103, "grad_norm": 0.8124137266650002, "learning_rate": 0.003, "loss": 4.1488, "step": 3103 }, { "epoch": 0.03104, "grad_norm": 1.1527056277371386, "learning_rate": 0.003, "loss": 4.1653, "step": 3104 }, { "epoch": 0.03105, "grad_norm": 1.0637389066586085, "learning_rate": 0.003, "loss": 4.1487, "step": 3105 }, { "epoch": 0.03106, "grad_norm": 0.853665919221622, "learning_rate": 0.003, "loss": 4.1643, "step": 3106 }, { "epoch": 0.03107, "grad_norm": 0.6820330366060209, "learning_rate": 0.003, "loss": 4.1411, "step": 3107 }, { "epoch": 0.03108, "grad_norm": 0.7901175506821255, "learning_rate": 0.003, "loss": 4.157, "step": 3108 }, { "epoch": 0.03109, "grad_norm": 0.8999453911608919, "learning_rate": 0.003, "loss": 4.1614, "step": 3109 }, { "epoch": 0.0311, "grad_norm": 0.9752866529372375, "learning_rate": 0.003, "loss": 4.1901, "step": 3110 }, { "epoch": 0.03111, "grad_norm": 0.9099293959689805, "learning_rate": 0.003, "loss": 4.1663, "step": 3111 }, { "epoch": 0.03112, "grad_norm": 0.8946740375764577, "learning_rate": 0.003, "loss": 4.1897, "step": 3112 }, { "epoch": 0.03113, "grad_norm": 0.7357356872971288, "learning_rate": 0.003, "loss": 4.1716, "step": 3113 }, { "epoch": 0.03114, "grad_norm": 0.7508640013687926, "learning_rate": 0.003, "loss": 4.1761, "step": 3114 }, { "epoch": 0.03115, "grad_norm": 0.7602528729276722, "learning_rate": 0.003, "loss": 4.1792, "step": 3115 }, { "epoch": 0.03116, "grad_norm": 0.837948579347557, "learning_rate": 0.003, "loss": 4.1661, "step": 3116 }, { "epoch": 0.03117, "grad_norm": 0.9227910597494478, "learning_rate": 0.003, "loss": 4.1791, "step": 3117 }, { "epoch": 0.03118, "grad_norm": 0.8378702011669498, "learning_rate": 0.003, "loss": 4.1779, "step": 3118 }, { "epoch": 0.03119, "grad_norm": 0.858148891035245, "learning_rate": 0.003, "loss": 4.1947, "step": 3119 }, { "epoch": 0.0312, "grad_norm": 0.8936706763604597, "learning_rate": 0.003, "loss": 4.1985, "step": 3120 }, { "epoch": 0.03121, "grad_norm": 1.0167864501298352, "learning_rate": 0.003, "loss": 4.1846, "step": 3121 }, { "epoch": 0.03122, "grad_norm": 0.906464003938868, "learning_rate": 0.003, "loss": 4.2022, "step": 3122 }, { "epoch": 0.03123, "grad_norm": 0.8273070233817, "learning_rate": 0.003, "loss": 4.1964, "step": 3123 }, { "epoch": 0.03124, "grad_norm": 0.8276168406814042, "learning_rate": 0.003, "loss": 4.1699, "step": 3124 }, { "epoch": 0.03125, "grad_norm": 0.7854717842569616, "learning_rate": 0.003, "loss": 4.1716, "step": 3125 }, { "epoch": 0.03126, "grad_norm": 0.8477910267798683, "learning_rate": 0.003, "loss": 4.1789, "step": 3126 }, { "epoch": 0.03127, "grad_norm": 0.8759271431103222, "learning_rate": 0.003, "loss": 4.1837, "step": 3127 }, { "epoch": 0.03128, "grad_norm": 1.0163995710540448, "learning_rate": 0.003, "loss": 4.1778, "step": 3128 }, { "epoch": 0.03129, "grad_norm": 0.9119429199687523, "learning_rate": 0.003, "loss": 4.193, "step": 3129 }, { "epoch": 0.0313, "grad_norm": 0.8770385869172778, "learning_rate": 0.003, "loss": 4.1968, "step": 3130 }, { "epoch": 0.03131, "grad_norm": 0.768644055037608, "learning_rate": 0.003, "loss": 4.1747, "step": 3131 }, { "epoch": 0.03132, "grad_norm": 0.7588002202266282, "learning_rate": 0.003, "loss": 4.1972, "step": 3132 }, { "epoch": 0.03133, "grad_norm": 0.8833279390524064, "learning_rate": 0.003, "loss": 4.1902, "step": 3133 }, { "epoch": 0.03134, "grad_norm": 0.9445298443136937, "learning_rate": 0.003, "loss": 4.1915, "step": 3134 }, { "epoch": 0.03135, "grad_norm": 0.9704581871113671, "learning_rate": 0.003, "loss": 4.1649, "step": 3135 }, { "epoch": 0.03136, "grad_norm": 0.9377630840593111, "learning_rate": 0.003, "loss": 4.1782, "step": 3136 }, { "epoch": 0.03137, "grad_norm": 0.7590762953987991, "learning_rate": 0.003, "loss": 4.1745, "step": 3137 }, { "epoch": 0.03138, "grad_norm": 0.6615196510693153, "learning_rate": 0.003, "loss": 4.1722, "step": 3138 }, { "epoch": 0.03139, "grad_norm": 0.7417937957321744, "learning_rate": 0.003, "loss": 4.1818, "step": 3139 }, { "epoch": 0.0314, "grad_norm": 0.7539643313234111, "learning_rate": 0.003, "loss": 4.18, "step": 3140 }, { "epoch": 0.03141, "grad_norm": 0.6909968958896604, "learning_rate": 0.003, "loss": 4.2116, "step": 3141 }, { "epoch": 0.03142, "grad_norm": 0.7898474777523338, "learning_rate": 0.003, "loss": 4.1916, "step": 3142 }, { "epoch": 0.03143, "grad_norm": 0.7784656393609168, "learning_rate": 0.003, "loss": 4.1786, "step": 3143 }, { "epoch": 0.03144, "grad_norm": 0.9707258921774178, "learning_rate": 0.003, "loss": 4.1561, "step": 3144 }, { "epoch": 0.03145, "grad_norm": 1.0898056591342788, "learning_rate": 0.003, "loss": 4.1835, "step": 3145 }, { "epoch": 0.03146, "grad_norm": 0.8349734971045476, "learning_rate": 0.003, "loss": 4.1866, "step": 3146 }, { "epoch": 0.03147, "grad_norm": 0.7259181361450073, "learning_rate": 0.003, "loss": 4.1803, "step": 3147 }, { "epoch": 0.03148, "grad_norm": 0.687941001438256, "learning_rate": 0.003, "loss": 4.1831, "step": 3148 }, { "epoch": 0.03149, "grad_norm": 0.7440383665305609, "learning_rate": 0.003, "loss": 4.1888, "step": 3149 }, { "epoch": 0.0315, "grad_norm": 0.7343700663124108, "learning_rate": 0.003, "loss": 4.1743, "step": 3150 }, { "epoch": 0.03151, "grad_norm": 0.701384082950667, "learning_rate": 0.003, "loss": 4.1825, "step": 3151 }, { "epoch": 0.03152, "grad_norm": 0.7986992640658904, "learning_rate": 0.003, "loss": 4.1873, "step": 3152 }, { "epoch": 0.03153, "grad_norm": 0.8011793706646407, "learning_rate": 0.003, "loss": 4.1933, "step": 3153 }, { "epoch": 0.03154, "grad_norm": 0.7113372576063467, "learning_rate": 0.003, "loss": 4.1855, "step": 3154 }, { "epoch": 0.03155, "grad_norm": 0.5804198718855146, "learning_rate": 0.003, "loss": 4.1711, "step": 3155 }, { "epoch": 0.03156, "grad_norm": 0.6180565037439169, "learning_rate": 0.003, "loss": 4.1541, "step": 3156 }, { "epoch": 0.03157, "grad_norm": 0.6588053530299754, "learning_rate": 0.003, "loss": 4.1807, "step": 3157 }, { "epoch": 0.03158, "grad_norm": 0.7176039547135377, "learning_rate": 0.003, "loss": 4.1644, "step": 3158 }, { "epoch": 0.03159, "grad_norm": 0.7544412083171981, "learning_rate": 0.003, "loss": 4.167, "step": 3159 }, { "epoch": 0.0316, "grad_norm": 0.6609856171261774, "learning_rate": 0.003, "loss": 4.18, "step": 3160 }, { "epoch": 0.03161, "grad_norm": 0.601780895016729, "learning_rate": 0.003, "loss": 4.1861, "step": 3161 }, { "epoch": 0.03162, "grad_norm": 0.6040830597762422, "learning_rate": 0.003, "loss": 4.1883, "step": 3162 }, { "epoch": 0.03163, "grad_norm": 0.6320256997655519, "learning_rate": 0.003, "loss": 4.1721, "step": 3163 }, { "epoch": 0.03164, "grad_norm": 0.6389872031652813, "learning_rate": 0.003, "loss": 4.171, "step": 3164 }, { "epoch": 0.03165, "grad_norm": 0.5915242364794046, "learning_rate": 0.003, "loss": 4.1423, "step": 3165 }, { "epoch": 0.03166, "grad_norm": 0.61505912049187, "learning_rate": 0.003, "loss": 4.1544, "step": 3166 }, { "epoch": 0.03167, "grad_norm": 0.6731926870227466, "learning_rate": 0.003, "loss": 4.1466, "step": 3167 }, { "epoch": 0.03168, "grad_norm": 0.5950823041934726, "learning_rate": 0.003, "loss": 4.1714, "step": 3168 }, { "epoch": 0.03169, "grad_norm": 0.5635445406508717, "learning_rate": 0.003, "loss": 4.179, "step": 3169 }, { "epoch": 0.0317, "grad_norm": 0.6365033301443797, "learning_rate": 0.003, "loss": 4.1377, "step": 3170 }, { "epoch": 0.03171, "grad_norm": 0.7998364639202467, "learning_rate": 0.003, "loss": 4.1808, "step": 3171 }, { "epoch": 0.03172, "grad_norm": 1.0979998450099218, "learning_rate": 0.003, "loss": 4.1776, "step": 3172 }, { "epoch": 0.03173, "grad_norm": 1.1291812351216926, "learning_rate": 0.003, "loss": 4.1584, "step": 3173 }, { "epoch": 0.03174, "grad_norm": 0.8472774425289195, "learning_rate": 0.003, "loss": 4.1683, "step": 3174 }, { "epoch": 0.03175, "grad_norm": 0.6918105971567023, "learning_rate": 0.003, "loss": 4.191, "step": 3175 }, { "epoch": 0.03176, "grad_norm": 0.6745403402677648, "learning_rate": 0.003, "loss": 4.1495, "step": 3176 }, { "epoch": 0.03177, "grad_norm": 0.8711762889488271, "learning_rate": 0.003, "loss": 4.1675, "step": 3177 }, { "epoch": 0.03178, "grad_norm": 0.9618302167115096, "learning_rate": 0.003, "loss": 4.22, "step": 3178 }, { "epoch": 0.03179, "grad_norm": 0.8866746933425214, "learning_rate": 0.003, "loss": 4.1824, "step": 3179 }, { "epoch": 0.0318, "grad_norm": 0.8756075769559615, "learning_rate": 0.003, "loss": 4.1827, "step": 3180 }, { "epoch": 0.03181, "grad_norm": 0.8428496447166878, "learning_rate": 0.003, "loss": 4.1912, "step": 3181 }, { "epoch": 0.03182, "grad_norm": 0.7661700460210522, "learning_rate": 0.003, "loss": 4.1876, "step": 3182 }, { "epoch": 0.03183, "grad_norm": 0.8692473733272297, "learning_rate": 0.003, "loss": 4.1965, "step": 3183 }, { "epoch": 0.03184, "grad_norm": 0.8297755272881908, "learning_rate": 0.003, "loss": 4.1705, "step": 3184 }, { "epoch": 0.03185, "grad_norm": 0.859558340717006, "learning_rate": 0.003, "loss": 4.1915, "step": 3185 }, { "epoch": 0.03186, "grad_norm": 1.0619931666528437, "learning_rate": 0.003, "loss": 4.1964, "step": 3186 }, { "epoch": 0.03187, "grad_norm": 1.044280375901733, "learning_rate": 0.003, "loss": 4.1629, "step": 3187 }, { "epoch": 0.03188, "grad_norm": 1.076116225662036, "learning_rate": 0.003, "loss": 4.1969, "step": 3188 }, { "epoch": 0.03189, "grad_norm": 1.103493479731777, "learning_rate": 0.003, "loss": 4.1925, "step": 3189 }, { "epoch": 0.0319, "grad_norm": 1.0030010721198397, "learning_rate": 0.003, "loss": 4.1821, "step": 3190 }, { "epoch": 0.03191, "grad_norm": 0.8727680004016529, "learning_rate": 0.003, "loss": 4.2163, "step": 3191 }, { "epoch": 0.03192, "grad_norm": 0.7351212105515794, "learning_rate": 0.003, "loss": 4.1806, "step": 3192 }, { "epoch": 0.03193, "grad_norm": 0.7321735689318333, "learning_rate": 0.003, "loss": 4.182, "step": 3193 }, { "epoch": 0.03194, "grad_norm": 0.8772456813713311, "learning_rate": 0.003, "loss": 4.2075, "step": 3194 }, { "epoch": 0.03195, "grad_norm": 1.0948458483452161, "learning_rate": 0.003, "loss": 4.1816, "step": 3195 }, { "epoch": 0.03196, "grad_norm": 0.8212538098418293, "learning_rate": 0.003, "loss": 4.19, "step": 3196 }, { "epoch": 0.03197, "grad_norm": 0.6285011842982621, "learning_rate": 0.003, "loss": 4.1514, "step": 3197 }, { "epoch": 0.03198, "grad_norm": 0.6890594394182098, "learning_rate": 0.003, "loss": 4.1835, "step": 3198 }, { "epoch": 0.03199, "grad_norm": 0.7446169713513262, "learning_rate": 0.003, "loss": 4.1599, "step": 3199 }, { "epoch": 0.032, "grad_norm": 0.6287973146330949, "learning_rate": 0.003, "loss": 4.1743, "step": 3200 }, { "epoch": 0.03201, "grad_norm": 0.574028662990781, "learning_rate": 0.003, "loss": 4.1359, "step": 3201 }, { "epoch": 0.03202, "grad_norm": 0.5668166289934877, "learning_rate": 0.003, "loss": 4.1726, "step": 3202 }, { "epoch": 0.03203, "grad_norm": 0.6420200932257698, "learning_rate": 0.003, "loss": 4.158, "step": 3203 }, { "epoch": 0.03204, "grad_norm": 0.6517793013040037, "learning_rate": 0.003, "loss": 4.1736, "step": 3204 }, { "epoch": 0.03205, "grad_norm": 0.6904481493053295, "learning_rate": 0.003, "loss": 4.1691, "step": 3205 }, { "epoch": 0.03206, "grad_norm": 0.6921511390797982, "learning_rate": 0.003, "loss": 4.1638, "step": 3206 }, { "epoch": 0.03207, "grad_norm": 0.6483492465436955, "learning_rate": 0.003, "loss": 4.1414, "step": 3207 }, { "epoch": 0.03208, "grad_norm": 0.6252426422831405, "learning_rate": 0.003, "loss": 4.1649, "step": 3208 }, { "epoch": 0.03209, "grad_norm": 0.6090985425136507, "learning_rate": 0.003, "loss": 4.1993, "step": 3209 }, { "epoch": 0.0321, "grad_norm": 0.7890939094365457, "learning_rate": 0.003, "loss": 4.182, "step": 3210 }, { "epoch": 0.03211, "grad_norm": 0.9334403001508826, "learning_rate": 0.003, "loss": 4.1592, "step": 3211 }, { "epoch": 0.03212, "grad_norm": 0.9752293730760951, "learning_rate": 0.003, "loss": 4.1748, "step": 3212 }, { "epoch": 0.03213, "grad_norm": 0.9917130039556156, "learning_rate": 0.003, "loss": 4.1791, "step": 3213 }, { "epoch": 0.03214, "grad_norm": 1.1323357347525596, "learning_rate": 0.003, "loss": 4.1546, "step": 3214 }, { "epoch": 0.03215, "grad_norm": 0.9029631323088454, "learning_rate": 0.003, "loss": 4.1597, "step": 3215 }, { "epoch": 0.03216, "grad_norm": 0.8650700547902456, "learning_rate": 0.003, "loss": 4.1698, "step": 3216 }, { "epoch": 0.03217, "grad_norm": 0.7818754734092035, "learning_rate": 0.003, "loss": 4.1662, "step": 3217 }, { "epoch": 0.03218, "grad_norm": 0.7666544069164494, "learning_rate": 0.003, "loss": 4.1592, "step": 3218 }, { "epoch": 0.03219, "grad_norm": 0.6769062297915203, "learning_rate": 0.003, "loss": 4.1636, "step": 3219 }, { "epoch": 0.0322, "grad_norm": 0.8021614507600279, "learning_rate": 0.003, "loss": 4.1784, "step": 3220 }, { "epoch": 0.03221, "grad_norm": 0.8570263886415102, "learning_rate": 0.003, "loss": 4.1856, "step": 3221 }, { "epoch": 0.03222, "grad_norm": 0.8752167497732125, "learning_rate": 0.003, "loss": 4.1822, "step": 3222 }, { "epoch": 0.03223, "grad_norm": 0.7864443587317127, "learning_rate": 0.003, "loss": 4.1809, "step": 3223 }, { "epoch": 0.03224, "grad_norm": 0.7212110109835455, "learning_rate": 0.003, "loss": 4.1404, "step": 3224 }, { "epoch": 0.03225, "grad_norm": 0.654641816291879, "learning_rate": 0.003, "loss": 4.172, "step": 3225 }, { "epoch": 0.03226, "grad_norm": 0.8355512444532979, "learning_rate": 0.003, "loss": 4.1574, "step": 3226 }, { "epoch": 0.03227, "grad_norm": 0.9048367283289669, "learning_rate": 0.003, "loss": 4.182, "step": 3227 }, { "epoch": 0.03228, "grad_norm": 0.8603111512659146, "learning_rate": 0.003, "loss": 4.1544, "step": 3228 }, { "epoch": 0.03229, "grad_norm": 0.857474441474401, "learning_rate": 0.003, "loss": 4.1747, "step": 3229 }, { "epoch": 0.0323, "grad_norm": 0.7857692858326722, "learning_rate": 0.003, "loss": 4.1654, "step": 3230 }, { "epoch": 0.03231, "grad_norm": 0.8253717389435257, "learning_rate": 0.003, "loss": 4.1822, "step": 3231 }, { "epoch": 0.03232, "grad_norm": 0.8267544687317752, "learning_rate": 0.003, "loss": 4.1723, "step": 3232 }, { "epoch": 0.03233, "grad_norm": 0.9496934310450251, "learning_rate": 0.003, "loss": 4.1973, "step": 3233 }, { "epoch": 0.03234, "grad_norm": 1.0636307547806063, "learning_rate": 0.003, "loss": 4.1839, "step": 3234 }, { "epoch": 0.03235, "grad_norm": 1.1231119530164886, "learning_rate": 0.003, "loss": 4.1901, "step": 3235 }, { "epoch": 0.03236, "grad_norm": 0.8190269419203875, "learning_rate": 0.003, "loss": 4.1751, "step": 3236 }, { "epoch": 0.03237, "grad_norm": 0.8205044577561036, "learning_rate": 0.003, "loss": 4.1972, "step": 3237 }, { "epoch": 0.03238, "grad_norm": 0.78546472021138, "learning_rate": 0.003, "loss": 4.1993, "step": 3238 }, { "epoch": 0.03239, "grad_norm": 0.7702117279920518, "learning_rate": 0.003, "loss": 4.1784, "step": 3239 }, { "epoch": 0.0324, "grad_norm": 0.8238552543857671, "learning_rate": 0.003, "loss": 4.1954, "step": 3240 }, { "epoch": 0.03241, "grad_norm": 0.9026611610808934, "learning_rate": 0.003, "loss": 4.1935, "step": 3241 }, { "epoch": 0.03242, "grad_norm": 1.0678835814399164, "learning_rate": 0.003, "loss": 4.1757, "step": 3242 }, { "epoch": 0.03243, "grad_norm": 0.9396714595581975, "learning_rate": 0.003, "loss": 4.161, "step": 3243 }, { "epoch": 0.03244, "grad_norm": 0.9280118806328921, "learning_rate": 0.003, "loss": 4.1836, "step": 3244 }, { "epoch": 0.03245, "grad_norm": 0.9538979131563459, "learning_rate": 0.003, "loss": 4.1945, "step": 3245 }, { "epoch": 0.03246, "grad_norm": 0.7413512610145042, "learning_rate": 0.003, "loss": 4.1767, "step": 3246 }, { "epoch": 0.03247, "grad_norm": 0.7567991071999487, "learning_rate": 0.003, "loss": 4.1856, "step": 3247 }, { "epoch": 0.03248, "grad_norm": 0.6112013067709322, "learning_rate": 0.003, "loss": 4.1978, "step": 3248 }, { "epoch": 0.03249, "grad_norm": 0.5490973188468005, "learning_rate": 0.003, "loss": 4.16, "step": 3249 }, { "epoch": 0.0325, "grad_norm": 0.5774791971085714, "learning_rate": 0.003, "loss": 4.1924, "step": 3250 }, { "epoch": 0.03251, "grad_norm": 0.6942876437278124, "learning_rate": 0.003, "loss": 4.1879, "step": 3251 }, { "epoch": 0.03252, "grad_norm": 0.8016029308986741, "learning_rate": 0.003, "loss": 4.1599, "step": 3252 }, { "epoch": 0.03253, "grad_norm": 1.060781307197035, "learning_rate": 0.003, "loss": 4.1707, "step": 3253 }, { "epoch": 0.03254, "grad_norm": 0.9120723115867568, "learning_rate": 0.003, "loss": 4.1638, "step": 3254 }, { "epoch": 0.03255, "grad_norm": 0.656644846174074, "learning_rate": 0.003, "loss": 4.1727, "step": 3255 }, { "epoch": 0.03256, "grad_norm": 0.6737599095064468, "learning_rate": 0.003, "loss": 4.1684, "step": 3256 }, { "epoch": 0.03257, "grad_norm": 0.7252801429473028, "learning_rate": 0.003, "loss": 4.1562, "step": 3257 }, { "epoch": 0.03258, "grad_norm": 0.7056456211965811, "learning_rate": 0.003, "loss": 4.1664, "step": 3258 }, { "epoch": 0.03259, "grad_norm": 0.6702493016081907, "learning_rate": 0.003, "loss": 4.1648, "step": 3259 }, { "epoch": 0.0326, "grad_norm": 0.742639840095955, "learning_rate": 0.003, "loss": 4.1693, "step": 3260 }, { "epoch": 0.03261, "grad_norm": 0.7251275113655737, "learning_rate": 0.003, "loss": 4.1404, "step": 3261 }, { "epoch": 0.03262, "grad_norm": 0.8493290703145597, "learning_rate": 0.003, "loss": 4.1703, "step": 3262 }, { "epoch": 0.03263, "grad_norm": 0.8597025731971345, "learning_rate": 0.003, "loss": 4.1803, "step": 3263 }, { "epoch": 0.03264, "grad_norm": 0.8711807826300412, "learning_rate": 0.003, "loss": 4.1949, "step": 3264 }, { "epoch": 0.03265, "grad_norm": 0.866467942160948, "learning_rate": 0.003, "loss": 4.1874, "step": 3265 }, { "epoch": 0.03266, "grad_norm": 0.7907548571319843, "learning_rate": 0.003, "loss": 4.1421, "step": 3266 }, { "epoch": 0.03267, "grad_norm": 0.861876845682946, "learning_rate": 0.003, "loss": 4.1569, "step": 3267 }, { "epoch": 0.03268, "grad_norm": 0.8022006061060821, "learning_rate": 0.003, "loss": 4.1422, "step": 3268 }, { "epoch": 0.03269, "grad_norm": 0.7169849622143066, "learning_rate": 0.003, "loss": 4.1689, "step": 3269 }, { "epoch": 0.0327, "grad_norm": 0.7910109368111681, "learning_rate": 0.003, "loss": 4.1996, "step": 3270 }, { "epoch": 0.03271, "grad_norm": 0.8344969911176661, "learning_rate": 0.003, "loss": 4.1641, "step": 3271 }, { "epoch": 0.03272, "grad_norm": 0.872802512464233, "learning_rate": 0.003, "loss": 4.173, "step": 3272 }, { "epoch": 0.03273, "grad_norm": 1.035019938766142, "learning_rate": 0.003, "loss": 4.1763, "step": 3273 }, { "epoch": 0.03274, "grad_norm": 1.2073229183989687, "learning_rate": 0.003, "loss": 4.2034, "step": 3274 }, { "epoch": 0.03275, "grad_norm": 0.8290793472557604, "learning_rate": 0.003, "loss": 4.1932, "step": 3275 }, { "epoch": 0.03276, "grad_norm": 0.7945684592795406, "learning_rate": 0.003, "loss": 4.1727, "step": 3276 }, { "epoch": 0.03277, "grad_norm": 0.7498594076499748, "learning_rate": 0.003, "loss": 4.1678, "step": 3277 }, { "epoch": 0.03278, "grad_norm": 0.7587982978511185, "learning_rate": 0.003, "loss": 4.1765, "step": 3278 }, { "epoch": 0.03279, "grad_norm": 0.6810453324428601, "learning_rate": 0.003, "loss": 4.1616, "step": 3279 }, { "epoch": 0.0328, "grad_norm": 0.7837448527302995, "learning_rate": 0.003, "loss": 4.1508, "step": 3280 }, { "epoch": 0.03281, "grad_norm": 0.8219512369198226, "learning_rate": 0.003, "loss": 4.1987, "step": 3281 }, { "epoch": 0.03282, "grad_norm": 0.9147716603434757, "learning_rate": 0.003, "loss": 4.1562, "step": 3282 }, { "epoch": 0.03283, "grad_norm": 1.1897707873169607, "learning_rate": 0.003, "loss": 4.162, "step": 3283 }, { "epoch": 0.03284, "grad_norm": 0.9254561336830153, "learning_rate": 0.003, "loss": 4.1995, "step": 3284 }, { "epoch": 0.03285, "grad_norm": 0.7859405478984048, "learning_rate": 0.003, "loss": 4.1464, "step": 3285 }, { "epoch": 0.03286, "grad_norm": 0.8112187817191882, "learning_rate": 0.003, "loss": 4.1821, "step": 3286 }, { "epoch": 0.03287, "grad_norm": 0.8325987650512564, "learning_rate": 0.003, "loss": 4.1892, "step": 3287 }, { "epoch": 0.03288, "grad_norm": 0.7508815679592351, "learning_rate": 0.003, "loss": 4.1681, "step": 3288 }, { "epoch": 0.03289, "grad_norm": 0.7067915480483684, "learning_rate": 0.003, "loss": 4.1657, "step": 3289 }, { "epoch": 0.0329, "grad_norm": 0.6452869815142386, "learning_rate": 0.003, "loss": 4.1967, "step": 3290 }, { "epoch": 0.03291, "grad_norm": 0.7077619390726327, "learning_rate": 0.003, "loss": 4.1816, "step": 3291 }, { "epoch": 0.03292, "grad_norm": 0.764342418854837, "learning_rate": 0.003, "loss": 4.1458, "step": 3292 }, { "epoch": 0.03293, "grad_norm": 0.7980406372998824, "learning_rate": 0.003, "loss": 4.1624, "step": 3293 }, { "epoch": 0.03294, "grad_norm": 0.9021539490595357, "learning_rate": 0.003, "loss": 4.1688, "step": 3294 }, { "epoch": 0.03295, "grad_norm": 1.0144913859148905, "learning_rate": 0.003, "loss": 4.1649, "step": 3295 }, { "epoch": 0.03296, "grad_norm": 0.9261450137381763, "learning_rate": 0.003, "loss": 4.1633, "step": 3296 }, { "epoch": 0.03297, "grad_norm": 0.6064461182372102, "learning_rate": 0.003, "loss": 4.1622, "step": 3297 }, { "epoch": 0.03298, "grad_norm": 0.7850213517125518, "learning_rate": 0.003, "loss": 4.1831, "step": 3298 }, { "epoch": 0.03299, "grad_norm": 0.9549041047998519, "learning_rate": 0.003, "loss": 4.1725, "step": 3299 }, { "epoch": 0.033, "grad_norm": 1.0983996256965927, "learning_rate": 0.003, "loss": 4.1699, "step": 3300 }, { "epoch": 0.03301, "grad_norm": 0.8476481107914999, "learning_rate": 0.003, "loss": 4.2035, "step": 3301 }, { "epoch": 0.03302, "grad_norm": 0.7326575292783787, "learning_rate": 0.003, "loss": 4.1693, "step": 3302 }, { "epoch": 0.03303, "grad_norm": 0.7473518023486005, "learning_rate": 0.003, "loss": 4.1768, "step": 3303 }, { "epoch": 0.03304, "grad_norm": 0.8060023346991733, "learning_rate": 0.003, "loss": 4.1726, "step": 3304 }, { "epoch": 0.03305, "grad_norm": 0.7204535881570415, "learning_rate": 0.003, "loss": 4.1755, "step": 3305 }, { "epoch": 0.03306, "grad_norm": 0.7755543314021593, "learning_rate": 0.003, "loss": 4.1959, "step": 3306 }, { "epoch": 0.03307, "grad_norm": 0.8878094148219878, "learning_rate": 0.003, "loss": 4.1938, "step": 3307 }, { "epoch": 0.03308, "grad_norm": 0.9646699735847878, "learning_rate": 0.003, "loss": 4.1797, "step": 3308 }, { "epoch": 0.03309, "grad_norm": 1.029625480565733, "learning_rate": 0.003, "loss": 4.1708, "step": 3309 }, { "epoch": 0.0331, "grad_norm": 0.9484648928493907, "learning_rate": 0.003, "loss": 4.1735, "step": 3310 }, { "epoch": 0.03311, "grad_norm": 0.7392971191362379, "learning_rate": 0.003, "loss": 4.1858, "step": 3311 }, { "epoch": 0.03312, "grad_norm": 0.5862137687073449, "learning_rate": 0.003, "loss": 4.1502, "step": 3312 }, { "epoch": 0.03313, "grad_norm": 0.596488073707156, "learning_rate": 0.003, "loss": 4.1792, "step": 3313 }, { "epoch": 0.03314, "grad_norm": 0.627442295342148, "learning_rate": 0.003, "loss": 4.122, "step": 3314 }, { "epoch": 0.03315, "grad_norm": 0.8275855239837167, "learning_rate": 0.003, "loss": 4.1766, "step": 3315 }, { "epoch": 0.03316, "grad_norm": 0.9370582268522403, "learning_rate": 0.003, "loss": 4.1854, "step": 3316 }, { "epoch": 0.03317, "grad_norm": 0.9555220232612601, "learning_rate": 0.003, "loss": 4.176, "step": 3317 }, { "epoch": 0.03318, "grad_norm": 0.9212435845363719, "learning_rate": 0.003, "loss": 4.1593, "step": 3318 }, { "epoch": 0.03319, "grad_norm": 0.7776775672878333, "learning_rate": 0.003, "loss": 4.1535, "step": 3319 }, { "epoch": 0.0332, "grad_norm": 0.6545161157490753, "learning_rate": 0.003, "loss": 4.1717, "step": 3320 }, { "epoch": 0.03321, "grad_norm": 0.7075805828392252, "learning_rate": 0.003, "loss": 4.1886, "step": 3321 }, { "epoch": 0.03322, "grad_norm": 0.5823854673371157, "learning_rate": 0.003, "loss": 4.1369, "step": 3322 }, { "epoch": 0.03323, "grad_norm": 0.5681997988958711, "learning_rate": 0.003, "loss": 4.1781, "step": 3323 }, { "epoch": 0.03324, "grad_norm": 0.5871116115690365, "learning_rate": 0.003, "loss": 4.1444, "step": 3324 }, { "epoch": 0.03325, "grad_norm": 0.6563169766957025, "learning_rate": 0.003, "loss": 4.1787, "step": 3325 }, { "epoch": 0.03326, "grad_norm": 0.6448409982175801, "learning_rate": 0.003, "loss": 4.1374, "step": 3326 }, { "epoch": 0.03327, "grad_norm": 0.6600692635622109, "learning_rate": 0.003, "loss": 4.165, "step": 3327 }, { "epoch": 0.03328, "grad_norm": 0.5986942519205176, "learning_rate": 0.003, "loss": 4.1826, "step": 3328 }, { "epoch": 0.03329, "grad_norm": 0.5527742555822713, "learning_rate": 0.003, "loss": 4.1401, "step": 3329 }, { "epoch": 0.0333, "grad_norm": 0.5223056461707817, "learning_rate": 0.003, "loss": 4.1418, "step": 3330 }, { "epoch": 0.03331, "grad_norm": 0.7035002650156569, "learning_rate": 0.003, "loss": 4.1438, "step": 3331 }, { "epoch": 0.03332, "grad_norm": 0.9852513287674988, "learning_rate": 0.003, "loss": 4.1487, "step": 3332 }, { "epoch": 0.03333, "grad_norm": 1.2028937317307933, "learning_rate": 0.003, "loss": 4.153, "step": 3333 }, { "epoch": 0.03334, "grad_norm": 0.7217530187376259, "learning_rate": 0.003, "loss": 4.1659, "step": 3334 }, { "epoch": 0.03335, "grad_norm": 0.856060284710091, "learning_rate": 0.003, "loss": 4.1802, "step": 3335 }, { "epoch": 0.03336, "grad_norm": 0.9360301326604868, "learning_rate": 0.003, "loss": 4.1865, "step": 3336 }, { "epoch": 0.03337, "grad_norm": 0.9783080347996699, "learning_rate": 0.003, "loss": 4.1684, "step": 3337 }, { "epoch": 0.03338, "grad_norm": 0.925336540295906, "learning_rate": 0.003, "loss": 4.179, "step": 3338 }, { "epoch": 0.03339, "grad_norm": 0.8653423496229105, "learning_rate": 0.003, "loss": 4.1809, "step": 3339 }, { "epoch": 0.0334, "grad_norm": 0.8300930142413694, "learning_rate": 0.003, "loss": 4.1558, "step": 3340 }, { "epoch": 0.03341, "grad_norm": 0.7043960651078829, "learning_rate": 0.003, "loss": 4.1713, "step": 3341 }, { "epoch": 0.03342, "grad_norm": 0.7527535496540583, "learning_rate": 0.003, "loss": 4.1612, "step": 3342 }, { "epoch": 0.03343, "grad_norm": 0.7101647240608583, "learning_rate": 0.003, "loss": 4.1355, "step": 3343 }, { "epoch": 0.03344, "grad_norm": 0.6916264160905395, "learning_rate": 0.003, "loss": 4.1891, "step": 3344 }, { "epoch": 0.03345, "grad_norm": 0.6270475152140637, "learning_rate": 0.003, "loss": 4.1636, "step": 3345 }, { "epoch": 0.03346, "grad_norm": 0.6401626891249842, "learning_rate": 0.003, "loss": 4.1783, "step": 3346 }, { "epoch": 0.03347, "grad_norm": 0.7429872046530612, "learning_rate": 0.003, "loss": 4.1637, "step": 3347 }, { "epoch": 0.03348, "grad_norm": 0.9010177786070285, "learning_rate": 0.003, "loss": 4.1567, "step": 3348 }, { "epoch": 0.03349, "grad_norm": 1.2489916491427688, "learning_rate": 0.003, "loss": 4.2146, "step": 3349 }, { "epoch": 0.0335, "grad_norm": 0.7874425616197881, "learning_rate": 0.003, "loss": 4.1571, "step": 3350 }, { "epoch": 0.03351, "grad_norm": 0.6778456165956087, "learning_rate": 0.003, "loss": 4.1637, "step": 3351 }, { "epoch": 0.03352, "grad_norm": 0.6019165379094058, "learning_rate": 0.003, "loss": 4.1722, "step": 3352 }, { "epoch": 0.03353, "grad_norm": 0.5923300369029466, "learning_rate": 0.003, "loss": 4.174, "step": 3353 }, { "epoch": 0.03354, "grad_norm": 0.61539729048598, "learning_rate": 0.003, "loss": 4.1356, "step": 3354 }, { "epoch": 0.03355, "grad_norm": 0.735664235193292, "learning_rate": 0.003, "loss": 4.1652, "step": 3355 }, { "epoch": 0.03356, "grad_norm": 0.8128296289500704, "learning_rate": 0.003, "loss": 4.1888, "step": 3356 }, { "epoch": 0.03357, "grad_norm": 0.9527976996073543, "learning_rate": 0.003, "loss": 4.1679, "step": 3357 }, { "epoch": 0.03358, "grad_norm": 1.0414565142110577, "learning_rate": 0.003, "loss": 4.1895, "step": 3358 }, { "epoch": 0.03359, "grad_norm": 1.008856217533468, "learning_rate": 0.003, "loss": 4.1793, "step": 3359 }, { "epoch": 0.0336, "grad_norm": 0.8810935589863647, "learning_rate": 0.003, "loss": 4.19, "step": 3360 }, { "epoch": 0.03361, "grad_norm": 0.7183884568879403, "learning_rate": 0.003, "loss": 4.1854, "step": 3361 }, { "epoch": 0.03362, "grad_norm": 0.7267007378420227, "learning_rate": 0.003, "loss": 4.1815, "step": 3362 }, { "epoch": 0.03363, "grad_norm": 0.7801299772898054, "learning_rate": 0.003, "loss": 4.1306, "step": 3363 }, { "epoch": 0.03364, "grad_norm": 0.7966856524661053, "learning_rate": 0.003, "loss": 4.1556, "step": 3364 }, { "epoch": 0.03365, "grad_norm": 0.8847580650147542, "learning_rate": 0.003, "loss": 4.1533, "step": 3365 }, { "epoch": 0.03366, "grad_norm": 0.96095183947863, "learning_rate": 0.003, "loss": 4.1308, "step": 3366 }, { "epoch": 0.03367, "grad_norm": 1.0579464638068148, "learning_rate": 0.003, "loss": 4.1575, "step": 3367 }, { "epoch": 0.03368, "grad_norm": 0.9487908981038276, "learning_rate": 0.003, "loss": 4.2002, "step": 3368 }, { "epoch": 0.03369, "grad_norm": 0.926541854271104, "learning_rate": 0.003, "loss": 4.1569, "step": 3369 }, { "epoch": 0.0337, "grad_norm": 0.7454576740316814, "learning_rate": 0.003, "loss": 4.1891, "step": 3370 }, { "epoch": 0.03371, "grad_norm": 0.8137043749598539, "learning_rate": 0.003, "loss": 4.1272, "step": 3371 }, { "epoch": 0.03372, "grad_norm": 0.9674167173768639, "learning_rate": 0.003, "loss": 4.16, "step": 3372 }, { "epoch": 0.03373, "grad_norm": 1.2155494462899379, "learning_rate": 0.003, "loss": 4.2043, "step": 3373 }, { "epoch": 0.03374, "grad_norm": 0.990109962038484, "learning_rate": 0.003, "loss": 4.1853, "step": 3374 }, { "epoch": 0.03375, "grad_norm": 0.8775264313673968, "learning_rate": 0.003, "loss": 4.1691, "step": 3375 }, { "epoch": 0.03376, "grad_norm": 0.85672141967058, "learning_rate": 0.003, "loss": 4.1457, "step": 3376 }, { "epoch": 0.03377, "grad_norm": 0.9737179043643116, "learning_rate": 0.003, "loss": 4.1884, "step": 3377 }, { "epoch": 0.03378, "grad_norm": 1.189069284130068, "learning_rate": 0.003, "loss": 4.2031, "step": 3378 }, { "epoch": 0.03379, "grad_norm": 0.8477238820814195, "learning_rate": 0.003, "loss": 4.1659, "step": 3379 }, { "epoch": 0.0338, "grad_norm": 0.7852120011007475, "learning_rate": 0.003, "loss": 4.1908, "step": 3380 }, { "epoch": 0.03381, "grad_norm": 0.7292342587108829, "learning_rate": 0.003, "loss": 4.1928, "step": 3381 }, { "epoch": 0.03382, "grad_norm": 0.6465901203260986, "learning_rate": 0.003, "loss": 4.176, "step": 3382 }, { "epoch": 0.03383, "grad_norm": 0.6599423928010247, "learning_rate": 0.003, "loss": 4.141, "step": 3383 }, { "epoch": 0.03384, "grad_norm": 0.6191152180710551, "learning_rate": 0.003, "loss": 4.1621, "step": 3384 }, { "epoch": 0.03385, "grad_norm": 0.660294498068497, "learning_rate": 0.003, "loss": 4.1677, "step": 3385 }, { "epoch": 0.03386, "grad_norm": 0.6951272922326287, "learning_rate": 0.003, "loss": 4.1619, "step": 3386 }, { "epoch": 0.03387, "grad_norm": 0.7549085381425057, "learning_rate": 0.003, "loss": 4.1667, "step": 3387 }, { "epoch": 0.03388, "grad_norm": 0.8210352926247219, "learning_rate": 0.003, "loss": 4.1766, "step": 3388 }, { "epoch": 0.03389, "grad_norm": 0.7574190601223717, "learning_rate": 0.003, "loss": 4.192, "step": 3389 }, { "epoch": 0.0339, "grad_norm": 0.6482930381200502, "learning_rate": 0.003, "loss": 4.176, "step": 3390 }, { "epoch": 0.03391, "grad_norm": 0.6321069687041059, "learning_rate": 0.003, "loss": 4.155, "step": 3391 }, { "epoch": 0.03392, "grad_norm": 0.643651481056441, "learning_rate": 0.003, "loss": 4.1601, "step": 3392 }, { "epoch": 0.03393, "grad_norm": 0.701072754557211, "learning_rate": 0.003, "loss": 4.1704, "step": 3393 }, { "epoch": 0.03394, "grad_norm": 0.936207243661443, "learning_rate": 0.003, "loss": 4.1766, "step": 3394 }, { "epoch": 0.03395, "grad_norm": 1.148255848253888, "learning_rate": 0.003, "loss": 4.1898, "step": 3395 }, { "epoch": 0.03396, "grad_norm": 0.9888207982566978, "learning_rate": 0.003, "loss": 4.1665, "step": 3396 }, { "epoch": 0.03397, "grad_norm": 0.8996655993402819, "learning_rate": 0.003, "loss": 4.1378, "step": 3397 }, { "epoch": 0.03398, "grad_norm": 0.8176901251549881, "learning_rate": 0.003, "loss": 4.1884, "step": 3398 }, { "epoch": 0.03399, "grad_norm": 0.9597792168614336, "learning_rate": 0.003, "loss": 4.2185, "step": 3399 }, { "epoch": 0.034, "grad_norm": 0.8953838926214202, "learning_rate": 0.003, "loss": 4.1958, "step": 3400 }, { "epoch": 0.03401, "grad_norm": 0.8709033356594122, "learning_rate": 0.003, "loss": 4.1686, "step": 3401 }, { "epoch": 0.03402, "grad_norm": 0.8085455592831763, "learning_rate": 0.003, "loss": 4.1978, "step": 3402 }, { "epoch": 0.03403, "grad_norm": 0.8446940111741622, "learning_rate": 0.003, "loss": 4.157, "step": 3403 }, { "epoch": 0.03404, "grad_norm": 0.8940660013118596, "learning_rate": 0.003, "loss": 4.2009, "step": 3404 }, { "epoch": 0.03405, "grad_norm": 0.8140937963556486, "learning_rate": 0.003, "loss": 4.1779, "step": 3405 }, { "epoch": 0.03406, "grad_norm": 0.7315344108215677, "learning_rate": 0.003, "loss": 4.1602, "step": 3406 }, { "epoch": 0.03407, "grad_norm": 0.7402747619279555, "learning_rate": 0.003, "loss": 4.1747, "step": 3407 }, { "epoch": 0.03408, "grad_norm": 0.732820367759084, "learning_rate": 0.003, "loss": 4.1555, "step": 3408 }, { "epoch": 0.03409, "grad_norm": 0.7078192424381656, "learning_rate": 0.003, "loss": 4.163, "step": 3409 }, { "epoch": 0.0341, "grad_norm": 0.6568901083029279, "learning_rate": 0.003, "loss": 4.1544, "step": 3410 }, { "epoch": 0.03411, "grad_norm": 0.716531548016939, "learning_rate": 0.003, "loss": 4.1696, "step": 3411 }, { "epoch": 0.03412, "grad_norm": 0.7513105413690504, "learning_rate": 0.003, "loss": 4.1866, "step": 3412 }, { "epoch": 0.03413, "grad_norm": 0.8394550937538491, "learning_rate": 0.003, "loss": 4.1969, "step": 3413 }, { "epoch": 0.03414, "grad_norm": 0.820347608943603, "learning_rate": 0.003, "loss": 4.1853, "step": 3414 }, { "epoch": 0.03415, "grad_norm": 0.8347788267410619, "learning_rate": 0.003, "loss": 4.1684, "step": 3415 }, { "epoch": 0.03416, "grad_norm": 0.7803511833847365, "learning_rate": 0.003, "loss": 4.1695, "step": 3416 }, { "epoch": 0.03417, "grad_norm": 0.8460239573891309, "learning_rate": 0.003, "loss": 4.1658, "step": 3417 }, { "epoch": 0.03418, "grad_norm": 0.833609787064844, "learning_rate": 0.003, "loss": 4.1414, "step": 3418 }, { "epoch": 0.03419, "grad_norm": 0.8708569538552823, "learning_rate": 0.003, "loss": 4.1684, "step": 3419 }, { "epoch": 0.0342, "grad_norm": 0.9586486328814895, "learning_rate": 0.003, "loss": 4.18, "step": 3420 }, { "epoch": 0.03421, "grad_norm": 0.9216001377610666, "learning_rate": 0.003, "loss": 4.1736, "step": 3421 }, { "epoch": 0.03422, "grad_norm": 1.0431113207901683, "learning_rate": 0.003, "loss": 4.1952, "step": 3422 }, { "epoch": 0.03423, "grad_norm": 0.8615543590411862, "learning_rate": 0.003, "loss": 4.1686, "step": 3423 }, { "epoch": 0.03424, "grad_norm": 0.8731178346996326, "learning_rate": 0.003, "loss": 4.1942, "step": 3424 }, { "epoch": 0.03425, "grad_norm": 1.095469547675973, "learning_rate": 0.003, "loss": 4.1632, "step": 3425 }, { "epoch": 0.03426, "grad_norm": 1.1612745485851788, "learning_rate": 0.003, "loss": 4.1928, "step": 3426 }, { "epoch": 0.03427, "grad_norm": 0.8029034983629978, "learning_rate": 0.003, "loss": 4.1538, "step": 3427 }, { "epoch": 0.03428, "grad_norm": 0.6589054159011024, "learning_rate": 0.003, "loss": 4.1471, "step": 3428 }, { "epoch": 0.03429, "grad_norm": 0.8266299146386108, "learning_rate": 0.003, "loss": 4.1677, "step": 3429 }, { "epoch": 0.0343, "grad_norm": 0.9825330051154652, "learning_rate": 0.003, "loss": 4.1665, "step": 3430 }, { "epoch": 0.03431, "grad_norm": 0.910501700935703, "learning_rate": 0.003, "loss": 4.1613, "step": 3431 }, { "epoch": 0.03432, "grad_norm": 0.9004461828637552, "learning_rate": 0.003, "loss": 4.17, "step": 3432 }, { "epoch": 0.03433, "grad_norm": 0.8528177999939314, "learning_rate": 0.003, "loss": 4.1421, "step": 3433 }, { "epoch": 0.03434, "grad_norm": 0.89696000337961, "learning_rate": 0.003, "loss": 4.1672, "step": 3434 }, { "epoch": 0.03435, "grad_norm": 0.9861991758400855, "learning_rate": 0.003, "loss": 4.1557, "step": 3435 }, { "epoch": 0.03436, "grad_norm": 1.0131768497867766, "learning_rate": 0.003, "loss": 4.1868, "step": 3436 }, { "epoch": 0.03437, "grad_norm": 0.9466446408842145, "learning_rate": 0.003, "loss": 4.174, "step": 3437 }, { "epoch": 0.03438, "grad_norm": 0.9102848042032905, "learning_rate": 0.003, "loss": 4.1899, "step": 3438 }, { "epoch": 0.03439, "grad_norm": 0.8647708117802543, "learning_rate": 0.003, "loss": 4.1847, "step": 3439 }, { "epoch": 0.0344, "grad_norm": 0.7164211772779949, "learning_rate": 0.003, "loss": 4.1535, "step": 3440 }, { "epoch": 0.03441, "grad_norm": 0.6301651930660659, "learning_rate": 0.003, "loss": 4.1723, "step": 3441 }, { "epoch": 0.03442, "grad_norm": 0.6803560179535443, "learning_rate": 0.003, "loss": 4.1393, "step": 3442 }, { "epoch": 0.03443, "grad_norm": 0.7027557239683841, "learning_rate": 0.003, "loss": 4.1683, "step": 3443 }, { "epoch": 0.03444, "grad_norm": 0.679437494499468, "learning_rate": 0.003, "loss": 4.1792, "step": 3444 }, { "epoch": 0.03445, "grad_norm": 0.6317151895560527, "learning_rate": 0.003, "loss": 4.1393, "step": 3445 }, { "epoch": 0.03446, "grad_norm": 0.5594532664167771, "learning_rate": 0.003, "loss": 4.1262, "step": 3446 }, { "epoch": 0.03447, "grad_norm": 0.4608224201125752, "learning_rate": 0.003, "loss": 4.1725, "step": 3447 }, { "epoch": 0.03448, "grad_norm": 0.4578940242640693, "learning_rate": 0.003, "loss": 4.1424, "step": 3448 }, { "epoch": 0.03449, "grad_norm": 0.42932231306304686, "learning_rate": 0.003, "loss": 4.1274, "step": 3449 }, { "epoch": 0.0345, "grad_norm": 0.4988907778580992, "learning_rate": 0.003, "loss": 4.1635, "step": 3450 }, { "epoch": 0.03451, "grad_norm": 0.6396019203681813, "learning_rate": 0.003, "loss": 4.1364, "step": 3451 }, { "epoch": 0.03452, "grad_norm": 1.0898993664507333, "learning_rate": 0.003, "loss": 4.1757, "step": 3452 }, { "epoch": 0.03453, "grad_norm": 1.3432647343273045, "learning_rate": 0.003, "loss": 4.1673, "step": 3453 }, { "epoch": 0.03454, "grad_norm": 0.5913981853383808, "learning_rate": 0.003, "loss": 4.1344, "step": 3454 }, { "epoch": 0.03455, "grad_norm": 0.7494952955105579, "learning_rate": 0.003, "loss": 4.1362, "step": 3455 }, { "epoch": 0.03456, "grad_norm": 0.8718264272859545, "learning_rate": 0.003, "loss": 4.1597, "step": 3456 }, { "epoch": 0.03457, "grad_norm": 0.7341902897858598, "learning_rate": 0.003, "loss": 4.1168, "step": 3457 }, { "epoch": 0.03458, "grad_norm": 0.5928314274849413, "learning_rate": 0.003, "loss": 4.1526, "step": 3458 }, { "epoch": 0.03459, "grad_norm": 0.5786112106982269, "learning_rate": 0.003, "loss": 4.1792, "step": 3459 }, { "epoch": 0.0346, "grad_norm": 0.5686114896859036, "learning_rate": 0.003, "loss": 4.1768, "step": 3460 }, { "epoch": 0.03461, "grad_norm": 0.5394258978891616, "learning_rate": 0.003, "loss": 4.1554, "step": 3461 }, { "epoch": 0.03462, "grad_norm": 0.5283747456624726, "learning_rate": 0.003, "loss": 4.1468, "step": 3462 }, { "epoch": 0.03463, "grad_norm": 0.5548545176329134, "learning_rate": 0.003, "loss": 4.1315, "step": 3463 }, { "epoch": 0.03464, "grad_norm": 0.6086252098872936, "learning_rate": 0.003, "loss": 4.1482, "step": 3464 }, { "epoch": 0.03465, "grad_norm": 0.7233117130975524, "learning_rate": 0.003, "loss": 4.1559, "step": 3465 }, { "epoch": 0.03466, "grad_norm": 0.8644314342564327, "learning_rate": 0.003, "loss": 4.1605, "step": 3466 }, { "epoch": 0.03467, "grad_norm": 0.9734945351785561, "learning_rate": 0.003, "loss": 4.1479, "step": 3467 }, { "epoch": 0.03468, "grad_norm": 1.1561496198026826, "learning_rate": 0.003, "loss": 4.1675, "step": 3468 }, { "epoch": 0.03469, "grad_norm": 0.8501361165726434, "learning_rate": 0.003, "loss": 4.1585, "step": 3469 }, { "epoch": 0.0347, "grad_norm": 0.8848493282370733, "learning_rate": 0.003, "loss": 4.1734, "step": 3470 }, { "epoch": 0.03471, "grad_norm": 0.8464627960157501, "learning_rate": 0.003, "loss": 4.163, "step": 3471 }, { "epoch": 0.03472, "grad_norm": 0.8982470984794905, "learning_rate": 0.003, "loss": 4.1667, "step": 3472 }, { "epoch": 0.03473, "grad_norm": 0.9751232721396406, "learning_rate": 0.003, "loss": 4.1613, "step": 3473 }, { "epoch": 0.03474, "grad_norm": 0.8720398622820135, "learning_rate": 0.003, "loss": 4.1922, "step": 3474 }, { "epoch": 0.03475, "grad_norm": 1.0085199767036763, "learning_rate": 0.003, "loss": 4.1683, "step": 3475 }, { "epoch": 0.03476, "grad_norm": 1.087277414889208, "learning_rate": 0.003, "loss": 4.1878, "step": 3476 }, { "epoch": 0.03477, "grad_norm": 1.1234725023256988, "learning_rate": 0.003, "loss": 4.1862, "step": 3477 }, { "epoch": 0.03478, "grad_norm": 0.9746682007392135, "learning_rate": 0.003, "loss": 4.1844, "step": 3478 }, { "epoch": 0.03479, "grad_norm": 1.1292770316423675, "learning_rate": 0.003, "loss": 4.1696, "step": 3479 }, { "epoch": 0.0348, "grad_norm": 1.0966154473553906, "learning_rate": 0.003, "loss": 4.1668, "step": 3480 }, { "epoch": 0.03481, "grad_norm": 0.9630452734638144, "learning_rate": 0.003, "loss": 4.1815, "step": 3481 }, { "epoch": 0.03482, "grad_norm": 1.0282885618312065, "learning_rate": 0.003, "loss": 4.1913, "step": 3482 }, { "epoch": 0.03483, "grad_norm": 0.91830488319391, "learning_rate": 0.003, "loss": 4.1634, "step": 3483 }, { "epoch": 0.03484, "grad_norm": 0.904152019538749, "learning_rate": 0.003, "loss": 4.2088, "step": 3484 }, { "epoch": 0.03485, "grad_norm": 0.9919455357745458, "learning_rate": 0.003, "loss": 4.2056, "step": 3485 }, { "epoch": 0.03486, "grad_norm": 0.9732252412489275, "learning_rate": 0.003, "loss": 4.1777, "step": 3486 }, { "epoch": 0.03487, "grad_norm": 0.8822393408309757, "learning_rate": 0.003, "loss": 4.1698, "step": 3487 }, { "epoch": 0.03488, "grad_norm": 0.7964288583002509, "learning_rate": 0.003, "loss": 4.1617, "step": 3488 }, { "epoch": 0.03489, "grad_norm": 0.76074461149548, "learning_rate": 0.003, "loss": 4.1754, "step": 3489 }, { "epoch": 0.0349, "grad_norm": 0.6520650563876845, "learning_rate": 0.003, "loss": 4.1689, "step": 3490 }, { "epoch": 0.03491, "grad_norm": 0.6914947787275789, "learning_rate": 0.003, "loss": 4.1866, "step": 3491 }, { "epoch": 0.03492, "grad_norm": 0.5749087587448846, "learning_rate": 0.003, "loss": 4.1672, "step": 3492 }, { "epoch": 0.03493, "grad_norm": 0.542359666467121, "learning_rate": 0.003, "loss": 4.1557, "step": 3493 }, { "epoch": 0.03494, "grad_norm": 0.5345171326980744, "learning_rate": 0.003, "loss": 4.1581, "step": 3494 }, { "epoch": 0.03495, "grad_norm": 0.6065181384190205, "learning_rate": 0.003, "loss": 4.1642, "step": 3495 }, { "epoch": 0.03496, "grad_norm": 0.8772556513810549, "learning_rate": 0.003, "loss": 4.141, "step": 3496 }, { "epoch": 0.03497, "grad_norm": 1.592460211390561, "learning_rate": 0.003, "loss": 4.2001, "step": 3497 }, { "epoch": 0.03498, "grad_norm": 0.6866814285580091, "learning_rate": 0.003, "loss": 4.1572, "step": 3498 }, { "epoch": 0.03499, "grad_norm": 0.6490269076299953, "learning_rate": 0.003, "loss": 4.1576, "step": 3499 }, { "epoch": 0.035, "grad_norm": 0.7769966222047817, "learning_rate": 0.003, "loss": 4.1803, "step": 3500 }, { "epoch": 0.03501, "grad_norm": 0.8082329511105066, "learning_rate": 0.003, "loss": 4.1697, "step": 3501 }, { "epoch": 0.03502, "grad_norm": 0.8597363324605595, "learning_rate": 0.003, "loss": 4.1708, "step": 3502 }, { "epoch": 0.03503, "grad_norm": 0.7980430509608143, "learning_rate": 0.003, "loss": 4.1826, "step": 3503 }, { "epoch": 0.03504, "grad_norm": 0.8314809190142672, "learning_rate": 0.003, "loss": 4.1556, "step": 3504 }, { "epoch": 0.03505, "grad_norm": 0.7125747726013733, "learning_rate": 0.003, "loss": 4.1704, "step": 3505 }, { "epoch": 0.03506, "grad_norm": 0.6174504440202525, "learning_rate": 0.003, "loss": 4.1753, "step": 3506 }, { "epoch": 0.03507, "grad_norm": 0.7474444612701099, "learning_rate": 0.003, "loss": 4.1568, "step": 3507 }, { "epoch": 0.03508, "grad_norm": 0.744368255976828, "learning_rate": 0.003, "loss": 4.1375, "step": 3508 }, { "epoch": 0.03509, "grad_norm": 0.7883919396141444, "learning_rate": 0.003, "loss": 4.1519, "step": 3509 }, { "epoch": 0.0351, "grad_norm": 0.8999072142957483, "learning_rate": 0.003, "loss": 4.1466, "step": 3510 }, { "epoch": 0.03511, "grad_norm": 0.8704293057371069, "learning_rate": 0.003, "loss": 4.1592, "step": 3511 }, { "epoch": 0.03512, "grad_norm": 0.9815842496362013, "learning_rate": 0.003, "loss": 4.1538, "step": 3512 }, { "epoch": 0.03513, "grad_norm": 1.1066583346659893, "learning_rate": 0.003, "loss": 4.163, "step": 3513 }, { "epoch": 0.03514, "grad_norm": 0.981954170426782, "learning_rate": 0.003, "loss": 4.1757, "step": 3514 }, { "epoch": 0.03515, "grad_norm": 0.9792417977988364, "learning_rate": 0.003, "loss": 4.1869, "step": 3515 }, { "epoch": 0.03516, "grad_norm": 0.856404411633118, "learning_rate": 0.003, "loss": 4.2085, "step": 3516 }, { "epoch": 0.03517, "grad_norm": 0.7127129042609169, "learning_rate": 0.003, "loss": 4.155, "step": 3517 }, { "epoch": 0.03518, "grad_norm": 0.6898706808403468, "learning_rate": 0.003, "loss": 4.1428, "step": 3518 }, { "epoch": 0.03519, "grad_norm": 0.6402523946118677, "learning_rate": 0.003, "loss": 4.185, "step": 3519 }, { "epoch": 0.0352, "grad_norm": 0.6207863269082038, "learning_rate": 0.003, "loss": 4.1612, "step": 3520 }, { "epoch": 0.03521, "grad_norm": 0.6360401847170623, "learning_rate": 0.003, "loss": 4.1566, "step": 3521 }, { "epoch": 0.03522, "grad_norm": 0.7339165326988526, "learning_rate": 0.003, "loss": 4.159, "step": 3522 }, { "epoch": 0.03523, "grad_norm": 0.8973232567522073, "learning_rate": 0.003, "loss": 4.1303, "step": 3523 }, { "epoch": 0.03524, "grad_norm": 1.1587635718923102, "learning_rate": 0.003, "loss": 4.1833, "step": 3524 }, { "epoch": 0.03525, "grad_norm": 0.7822190217961902, "learning_rate": 0.003, "loss": 4.1417, "step": 3525 }, { "epoch": 0.03526, "grad_norm": 0.7291963363920031, "learning_rate": 0.003, "loss": 4.1485, "step": 3526 }, { "epoch": 0.03527, "grad_norm": 0.779725825468202, "learning_rate": 0.003, "loss": 4.1812, "step": 3527 }, { "epoch": 0.03528, "grad_norm": 0.8316224754491519, "learning_rate": 0.003, "loss": 4.1645, "step": 3528 }, { "epoch": 0.03529, "grad_norm": 0.8639080944248414, "learning_rate": 0.003, "loss": 4.1577, "step": 3529 }, { "epoch": 0.0353, "grad_norm": 0.8612623897066767, "learning_rate": 0.003, "loss": 4.1605, "step": 3530 }, { "epoch": 0.03531, "grad_norm": 0.8610797480294329, "learning_rate": 0.003, "loss": 4.1683, "step": 3531 }, { "epoch": 0.03532, "grad_norm": 0.9008566995316217, "learning_rate": 0.003, "loss": 4.1748, "step": 3532 }, { "epoch": 0.03533, "grad_norm": 0.7630493491147551, "learning_rate": 0.003, "loss": 4.1696, "step": 3533 }, { "epoch": 0.03534, "grad_norm": 0.7799426199854421, "learning_rate": 0.003, "loss": 4.1652, "step": 3534 }, { "epoch": 0.03535, "grad_norm": 0.7443571102643655, "learning_rate": 0.003, "loss": 4.1329, "step": 3535 }, { "epoch": 0.03536, "grad_norm": 0.9162718166038424, "learning_rate": 0.003, "loss": 4.1425, "step": 3536 }, { "epoch": 0.03537, "grad_norm": 0.9471286119751292, "learning_rate": 0.003, "loss": 4.1547, "step": 3537 }, { "epoch": 0.03538, "grad_norm": 0.949840005398208, "learning_rate": 0.003, "loss": 4.1591, "step": 3538 }, { "epoch": 0.03539, "grad_norm": 0.8941824512475612, "learning_rate": 0.003, "loss": 4.1466, "step": 3539 }, { "epoch": 0.0354, "grad_norm": 0.8416985140532036, "learning_rate": 0.003, "loss": 4.1851, "step": 3540 }, { "epoch": 0.03541, "grad_norm": 0.8643157714042061, "learning_rate": 0.003, "loss": 4.1734, "step": 3541 }, { "epoch": 0.03542, "grad_norm": 1.0638367559817616, "learning_rate": 0.003, "loss": 4.1614, "step": 3542 }, { "epoch": 0.03543, "grad_norm": 1.1652953416038865, "learning_rate": 0.003, "loss": 4.1783, "step": 3543 }, { "epoch": 0.03544, "grad_norm": 0.9946375845221913, "learning_rate": 0.003, "loss": 4.1824, "step": 3544 }, { "epoch": 0.03545, "grad_norm": 0.9626017427173171, "learning_rate": 0.003, "loss": 4.151, "step": 3545 }, { "epoch": 0.03546, "grad_norm": 0.9017009636423533, "learning_rate": 0.003, "loss": 4.1681, "step": 3546 }, { "epoch": 0.03547, "grad_norm": 0.8338308420587012, "learning_rate": 0.003, "loss": 4.1742, "step": 3547 }, { "epoch": 0.03548, "grad_norm": 0.795230968098308, "learning_rate": 0.003, "loss": 4.1761, "step": 3548 }, { "epoch": 0.03549, "grad_norm": 0.7285679046345136, "learning_rate": 0.003, "loss": 4.1663, "step": 3549 }, { "epoch": 0.0355, "grad_norm": 0.6534721210956203, "learning_rate": 0.003, "loss": 4.1208, "step": 3550 }, { "epoch": 0.03551, "grad_norm": 0.6108415927364046, "learning_rate": 0.003, "loss": 4.1778, "step": 3551 }, { "epoch": 0.03552, "grad_norm": 0.5975360738111026, "learning_rate": 0.003, "loss": 4.1357, "step": 3552 }, { "epoch": 0.03553, "grad_norm": 0.616283856055616, "learning_rate": 0.003, "loss": 4.15, "step": 3553 }, { "epoch": 0.03554, "grad_norm": 0.6492012815047297, "learning_rate": 0.003, "loss": 4.1549, "step": 3554 }, { "epoch": 0.03555, "grad_norm": 0.7788849687728241, "learning_rate": 0.003, "loss": 4.1699, "step": 3555 }, { "epoch": 0.03556, "grad_norm": 1.0575264249976415, "learning_rate": 0.003, "loss": 4.1571, "step": 3556 }, { "epoch": 0.03557, "grad_norm": 1.2637167881926659, "learning_rate": 0.003, "loss": 4.1692, "step": 3557 }, { "epoch": 0.03558, "grad_norm": 0.6413226636225998, "learning_rate": 0.003, "loss": 4.1816, "step": 3558 }, { "epoch": 0.03559, "grad_norm": 0.7092773893161065, "learning_rate": 0.003, "loss": 4.1679, "step": 3559 }, { "epoch": 0.0356, "grad_norm": 0.9184433367909671, "learning_rate": 0.003, "loss": 4.1754, "step": 3560 }, { "epoch": 0.03561, "grad_norm": 0.933975121612697, "learning_rate": 0.003, "loss": 4.2011, "step": 3561 }, { "epoch": 0.03562, "grad_norm": 0.9304359925283029, "learning_rate": 0.003, "loss": 4.164, "step": 3562 }, { "epoch": 0.03563, "grad_norm": 0.919692600735377, "learning_rate": 0.003, "loss": 4.1948, "step": 3563 }, { "epoch": 0.03564, "grad_norm": 0.7595794330578136, "learning_rate": 0.003, "loss": 4.1696, "step": 3564 }, { "epoch": 0.03565, "grad_norm": 0.6630885781605157, "learning_rate": 0.003, "loss": 4.1571, "step": 3565 }, { "epoch": 0.03566, "grad_norm": 0.6125749874747362, "learning_rate": 0.003, "loss": 4.1594, "step": 3566 }, { "epoch": 0.03567, "grad_norm": 0.7000181582049009, "learning_rate": 0.003, "loss": 4.1529, "step": 3567 }, { "epoch": 0.03568, "grad_norm": 0.668903695971405, "learning_rate": 0.003, "loss": 4.1481, "step": 3568 }, { "epoch": 0.03569, "grad_norm": 0.7198229032250428, "learning_rate": 0.003, "loss": 4.1448, "step": 3569 }, { "epoch": 0.0357, "grad_norm": 0.6904707021821862, "learning_rate": 0.003, "loss": 4.1473, "step": 3570 }, { "epoch": 0.03571, "grad_norm": 0.6259572941993182, "learning_rate": 0.003, "loss": 4.1522, "step": 3571 }, { "epoch": 0.03572, "grad_norm": 0.6307940196938667, "learning_rate": 0.003, "loss": 4.1717, "step": 3572 }, { "epoch": 0.03573, "grad_norm": 0.6869341362818634, "learning_rate": 0.003, "loss": 4.1406, "step": 3573 }, { "epoch": 0.03574, "grad_norm": 0.9390330749868537, "learning_rate": 0.003, "loss": 4.1723, "step": 3574 }, { "epoch": 0.03575, "grad_norm": 1.1544395774812188, "learning_rate": 0.003, "loss": 4.1625, "step": 3575 }, { "epoch": 0.03576, "grad_norm": 0.8068138585754847, "learning_rate": 0.003, "loss": 4.1373, "step": 3576 }, { "epoch": 0.03577, "grad_norm": 0.6631826248776513, "learning_rate": 0.003, "loss": 4.1679, "step": 3577 }, { "epoch": 0.03578, "grad_norm": 0.6113246693019829, "learning_rate": 0.003, "loss": 4.1541, "step": 3578 }, { "epoch": 0.03579, "grad_norm": 0.780311146927158, "learning_rate": 0.003, "loss": 4.1739, "step": 3579 }, { "epoch": 0.0358, "grad_norm": 0.827477686309755, "learning_rate": 0.003, "loss": 4.1335, "step": 3580 }, { "epoch": 0.03581, "grad_norm": 0.8515806061337369, "learning_rate": 0.003, "loss": 4.1442, "step": 3581 }, { "epoch": 0.03582, "grad_norm": 0.7993229318614161, "learning_rate": 0.003, "loss": 4.1436, "step": 3582 }, { "epoch": 0.03583, "grad_norm": 0.8095532689038906, "learning_rate": 0.003, "loss": 4.1456, "step": 3583 }, { "epoch": 0.03584, "grad_norm": 0.8205040467136088, "learning_rate": 0.003, "loss": 4.1372, "step": 3584 }, { "epoch": 0.03585, "grad_norm": 0.7389714832217196, "learning_rate": 0.003, "loss": 4.13, "step": 3585 }, { "epoch": 0.03586, "grad_norm": 0.6688957110779817, "learning_rate": 0.003, "loss": 4.1525, "step": 3586 }, { "epoch": 0.03587, "grad_norm": 0.7092588679384089, "learning_rate": 0.003, "loss": 4.1506, "step": 3587 }, { "epoch": 0.03588, "grad_norm": 0.7650773891669993, "learning_rate": 0.003, "loss": 4.1498, "step": 3588 }, { "epoch": 0.03589, "grad_norm": 0.6712500265923376, "learning_rate": 0.003, "loss": 4.1646, "step": 3589 }, { "epoch": 0.0359, "grad_norm": 0.7094925517899797, "learning_rate": 0.003, "loss": 4.13, "step": 3590 }, { "epoch": 0.03591, "grad_norm": 0.9772899161914086, "learning_rate": 0.003, "loss": 4.1681, "step": 3591 }, { "epoch": 0.03592, "grad_norm": 1.3584031251577058, "learning_rate": 0.003, "loss": 4.1699, "step": 3592 }, { "epoch": 0.03593, "grad_norm": 0.6739062662570949, "learning_rate": 0.003, "loss": 4.1483, "step": 3593 }, { "epoch": 0.03594, "grad_norm": 0.7388127025602351, "learning_rate": 0.003, "loss": 4.1566, "step": 3594 }, { "epoch": 0.03595, "grad_norm": 0.7746906492491002, "learning_rate": 0.003, "loss": 4.149, "step": 3595 }, { "epoch": 0.03596, "grad_norm": 0.7831810343025122, "learning_rate": 0.003, "loss": 4.1838, "step": 3596 }, { "epoch": 0.03597, "grad_norm": 0.8759604826013669, "learning_rate": 0.003, "loss": 4.1631, "step": 3597 }, { "epoch": 0.03598, "grad_norm": 0.9635123949092944, "learning_rate": 0.003, "loss": 4.1727, "step": 3598 }, { "epoch": 0.03599, "grad_norm": 0.9676585543651076, "learning_rate": 0.003, "loss": 4.1743, "step": 3599 }, { "epoch": 0.036, "grad_norm": 0.7973342092648297, "learning_rate": 0.003, "loss": 4.1881, "step": 3600 }, { "epoch": 0.03601, "grad_norm": 0.8952158475064083, "learning_rate": 0.003, "loss": 4.1641, "step": 3601 }, { "epoch": 0.03602, "grad_norm": 0.8297361733958166, "learning_rate": 0.003, "loss": 4.147, "step": 3602 }, { "epoch": 0.03603, "grad_norm": 0.8233542191650651, "learning_rate": 0.003, "loss": 4.1379, "step": 3603 }, { "epoch": 0.03604, "grad_norm": 0.8910587306080446, "learning_rate": 0.003, "loss": 4.1757, "step": 3604 }, { "epoch": 0.03605, "grad_norm": 0.8944750063674012, "learning_rate": 0.003, "loss": 4.1673, "step": 3605 }, { "epoch": 0.03606, "grad_norm": 0.8493957051258396, "learning_rate": 0.003, "loss": 4.1548, "step": 3606 }, { "epoch": 0.03607, "grad_norm": 0.8478455854512117, "learning_rate": 0.003, "loss": 4.1477, "step": 3607 }, { "epoch": 0.03608, "grad_norm": 0.8546181295422765, "learning_rate": 0.003, "loss": 4.1508, "step": 3608 }, { "epoch": 0.03609, "grad_norm": 0.8039599648016552, "learning_rate": 0.003, "loss": 4.1807, "step": 3609 }, { "epoch": 0.0361, "grad_norm": 0.7257991452948148, "learning_rate": 0.003, "loss": 4.183, "step": 3610 }, { "epoch": 0.03611, "grad_norm": 0.8324843473102026, "learning_rate": 0.003, "loss": 4.1339, "step": 3611 }, { "epoch": 0.03612, "grad_norm": 0.824602826243289, "learning_rate": 0.003, "loss": 4.1343, "step": 3612 }, { "epoch": 0.03613, "grad_norm": 1.0069180266978304, "learning_rate": 0.003, "loss": 4.182, "step": 3613 }, { "epoch": 0.03614, "grad_norm": 1.2555634931954907, "learning_rate": 0.003, "loss": 4.1726, "step": 3614 }, { "epoch": 0.03615, "grad_norm": 0.8411513212979675, "learning_rate": 0.003, "loss": 4.1693, "step": 3615 }, { "epoch": 0.03616, "grad_norm": 0.8451688294141301, "learning_rate": 0.003, "loss": 4.1775, "step": 3616 }, { "epoch": 0.03617, "grad_norm": 0.9112344841927535, "learning_rate": 0.003, "loss": 4.177, "step": 3617 }, { "epoch": 0.03618, "grad_norm": 0.9999707710039599, "learning_rate": 0.003, "loss": 4.1694, "step": 3618 }, { "epoch": 0.03619, "grad_norm": 1.0118192200700322, "learning_rate": 0.003, "loss": 4.1724, "step": 3619 }, { "epoch": 0.0362, "grad_norm": 1.0258430778394187, "learning_rate": 0.003, "loss": 4.1762, "step": 3620 }, { "epoch": 0.03621, "grad_norm": 1.0952856306278091, "learning_rate": 0.003, "loss": 4.185, "step": 3621 }, { "epoch": 0.03622, "grad_norm": 1.143882989533265, "learning_rate": 0.003, "loss": 4.1567, "step": 3622 }, { "epoch": 0.03623, "grad_norm": 0.7874622564246553, "learning_rate": 0.003, "loss": 4.1365, "step": 3623 }, { "epoch": 0.03624, "grad_norm": 0.7563110063478746, "learning_rate": 0.003, "loss": 4.1423, "step": 3624 }, { "epoch": 0.03625, "grad_norm": 0.7988179852097972, "learning_rate": 0.003, "loss": 4.178, "step": 3625 }, { "epoch": 0.03626, "grad_norm": 0.7231578892075671, "learning_rate": 0.003, "loss": 4.1549, "step": 3626 }, { "epoch": 0.03627, "grad_norm": 0.6892596641637148, "learning_rate": 0.003, "loss": 4.1529, "step": 3627 }, { "epoch": 0.03628, "grad_norm": 0.7186998166943942, "learning_rate": 0.003, "loss": 4.1606, "step": 3628 }, { "epoch": 0.03629, "grad_norm": 0.7752324983651075, "learning_rate": 0.003, "loss": 4.1787, "step": 3629 }, { "epoch": 0.0363, "grad_norm": 0.9413427411204313, "learning_rate": 0.003, "loss": 4.1685, "step": 3630 }, { "epoch": 0.03631, "grad_norm": 0.9777703909452877, "learning_rate": 0.003, "loss": 4.1642, "step": 3631 }, { "epoch": 0.03632, "grad_norm": 0.8848239522487785, "learning_rate": 0.003, "loss": 4.1579, "step": 3632 }, { "epoch": 0.03633, "grad_norm": 0.9200746928259687, "learning_rate": 0.003, "loss": 4.1929, "step": 3633 }, { "epoch": 0.03634, "grad_norm": 0.9334170004603629, "learning_rate": 0.003, "loss": 4.1669, "step": 3634 }, { "epoch": 0.03635, "grad_norm": 0.9470225102280233, "learning_rate": 0.003, "loss": 4.1558, "step": 3635 }, { "epoch": 0.03636, "grad_norm": 0.9241160510652092, "learning_rate": 0.003, "loss": 4.1754, "step": 3636 }, { "epoch": 0.03637, "grad_norm": 0.7430750130688752, "learning_rate": 0.003, "loss": 4.1589, "step": 3637 }, { "epoch": 0.03638, "grad_norm": 0.7220044156395624, "learning_rate": 0.003, "loss": 4.1645, "step": 3638 }, { "epoch": 0.03639, "grad_norm": 0.7772051390195733, "learning_rate": 0.003, "loss": 4.1882, "step": 3639 }, { "epoch": 0.0364, "grad_norm": 0.8529468509123678, "learning_rate": 0.003, "loss": 4.1847, "step": 3640 }, { "epoch": 0.03641, "grad_norm": 1.0954199441179653, "learning_rate": 0.003, "loss": 4.1735, "step": 3641 }, { "epoch": 0.03642, "grad_norm": 0.8994211892974673, "learning_rate": 0.003, "loss": 4.1879, "step": 3642 }, { "epoch": 0.03643, "grad_norm": 0.8296521561058982, "learning_rate": 0.003, "loss": 4.1481, "step": 3643 }, { "epoch": 0.03644, "grad_norm": 0.8247836328668441, "learning_rate": 0.003, "loss": 4.1762, "step": 3644 }, { "epoch": 0.03645, "grad_norm": 0.840574264399085, "learning_rate": 0.003, "loss": 4.1477, "step": 3645 }, { "epoch": 0.03646, "grad_norm": 0.8080938302750766, "learning_rate": 0.003, "loss": 4.1646, "step": 3646 }, { "epoch": 0.03647, "grad_norm": 0.9664779357124396, "learning_rate": 0.003, "loss": 4.165, "step": 3647 }, { "epoch": 0.03648, "grad_norm": 1.0928052932730679, "learning_rate": 0.003, "loss": 4.16, "step": 3648 }, { "epoch": 0.03649, "grad_norm": 0.7684467951205206, "learning_rate": 0.003, "loss": 4.1755, "step": 3649 }, { "epoch": 0.0365, "grad_norm": 0.5750234241245785, "learning_rate": 0.003, "loss": 4.148, "step": 3650 }, { "epoch": 0.03651, "grad_norm": 0.6573564310770661, "learning_rate": 0.003, "loss": 4.1708, "step": 3651 }, { "epoch": 0.03652, "grad_norm": 0.8888456033371046, "learning_rate": 0.003, "loss": 4.1579, "step": 3652 }, { "epoch": 0.03653, "grad_norm": 1.0419754908407304, "learning_rate": 0.003, "loss": 4.1477, "step": 3653 }, { "epoch": 0.03654, "grad_norm": 0.7921123605584492, "learning_rate": 0.003, "loss": 4.1421, "step": 3654 }, { "epoch": 0.03655, "grad_norm": 0.7289861914496177, "learning_rate": 0.003, "loss": 4.1668, "step": 3655 }, { "epoch": 0.03656, "grad_norm": 0.7796912479293006, "learning_rate": 0.003, "loss": 4.1547, "step": 3656 }, { "epoch": 0.03657, "grad_norm": 0.7324175109079976, "learning_rate": 0.003, "loss": 4.1261, "step": 3657 }, { "epoch": 0.03658, "grad_norm": 0.5756062562769299, "learning_rate": 0.003, "loss": 4.186, "step": 3658 }, { "epoch": 0.03659, "grad_norm": 0.6807858015149528, "learning_rate": 0.003, "loss": 4.1377, "step": 3659 }, { "epoch": 0.0366, "grad_norm": 0.7837064702433023, "learning_rate": 0.003, "loss": 4.1731, "step": 3660 }, { "epoch": 0.03661, "grad_norm": 0.7830429911526481, "learning_rate": 0.003, "loss": 4.1527, "step": 3661 }, { "epoch": 0.03662, "grad_norm": 0.7895739080121226, "learning_rate": 0.003, "loss": 4.1472, "step": 3662 }, { "epoch": 0.03663, "grad_norm": 0.893542071406176, "learning_rate": 0.003, "loss": 4.1687, "step": 3663 }, { "epoch": 0.03664, "grad_norm": 0.8754737269812044, "learning_rate": 0.003, "loss": 4.1494, "step": 3664 }, { "epoch": 0.03665, "grad_norm": 0.8736154039000673, "learning_rate": 0.003, "loss": 4.1736, "step": 3665 }, { "epoch": 0.03666, "grad_norm": 0.8365372537243944, "learning_rate": 0.003, "loss": 4.1322, "step": 3666 }, { "epoch": 0.03667, "grad_norm": 0.8675888077071373, "learning_rate": 0.003, "loss": 4.1444, "step": 3667 }, { "epoch": 0.03668, "grad_norm": 0.8544402197521644, "learning_rate": 0.003, "loss": 4.1573, "step": 3668 }, { "epoch": 0.03669, "grad_norm": 1.051770187962234, "learning_rate": 0.003, "loss": 4.1381, "step": 3669 }, { "epoch": 0.0367, "grad_norm": 1.312127243653887, "learning_rate": 0.003, "loss": 4.1894, "step": 3670 }, { "epoch": 0.03671, "grad_norm": 0.7882271613030375, "learning_rate": 0.003, "loss": 4.1628, "step": 3671 }, { "epoch": 0.03672, "grad_norm": 0.6575380088749906, "learning_rate": 0.003, "loss": 4.1614, "step": 3672 }, { "epoch": 0.03673, "grad_norm": 0.6533332497379345, "learning_rate": 0.003, "loss": 4.1705, "step": 3673 }, { "epoch": 0.03674, "grad_norm": 0.6740570121635143, "learning_rate": 0.003, "loss": 4.1281, "step": 3674 }, { "epoch": 0.03675, "grad_norm": 0.7719822820972118, "learning_rate": 0.003, "loss": 4.1729, "step": 3675 }, { "epoch": 0.03676, "grad_norm": 0.8997102742299902, "learning_rate": 0.003, "loss": 4.1318, "step": 3676 }, { "epoch": 0.03677, "grad_norm": 1.1346925943966613, "learning_rate": 0.003, "loss": 4.156, "step": 3677 }, { "epoch": 0.03678, "grad_norm": 1.3062247019240303, "learning_rate": 0.003, "loss": 4.1966, "step": 3678 }, { "epoch": 0.03679, "grad_norm": 0.7073498950931896, "learning_rate": 0.003, "loss": 4.1705, "step": 3679 }, { "epoch": 0.0368, "grad_norm": 0.7054669066644736, "learning_rate": 0.003, "loss": 4.1658, "step": 3680 }, { "epoch": 0.03681, "grad_norm": 0.6959325528820864, "learning_rate": 0.003, "loss": 4.1376, "step": 3681 }, { "epoch": 0.03682, "grad_norm": 0.6826134908981137, "learning_rate": 0.003, "loss": 4.1753, "step": 3682 }, { "epoch": 0.03683, "grad_norm": 0.9589548694394383, "learning_rate": 0.003, "loss": 4.1412, "step": 3683 }, { "epoch": 0.03684, "grad_norm": 1.3837835667463712, "learning_rate": 0.003, "loss": 4.185, "step": 3684 }, { "epoch": 0.03685, "grad_norm": 0.7029580914010543, "learning_rate": 0.003, "loss": 4.1508, "step": 3685 }, { "epoch": 0.03686, "grad_norm": 0.793184528586383, "learning_rate": 0.003, "loss": 4.173, "step": 3686 }, { "epoch": 0.03687, "grad_norm": 0.8386169263682931, "learning_rate": 0.003, "loss": 4.1944, "step": 3687 }, { "epoch": 0.03688, "grad_norm": 0.8483676049215235, "learning_rate": 0.003, "loss": 4.1841, "step": 3688 }, { "epoch": 0.03689, "grad_norm": 0.950222753372726, "learning_rate": 0.003, "loss": 4.1554, "step": 3689 }, { "epoch": 0.0369, "grad_norm": 0.9833017766191942, "learning_rate": 0.003, "loss": 4.1873, "step": 3690 }, { "epoch": 0.03691, "grad_norm": 1.0141537747097225, "learning_rate": 0.003, "loss": 4.1746, "step": 3691 }, { "epoch": 0.03692, "grad_norm": 0.8888682616492705, "learning_rate": 0.003, "loss": 4.2059, "step": 3692 }, { "epoch": 0.03693, "grad_norm": 0.9207088289233919, "learning_rate": 0.003, "loss": 4.1775, "step": 3693 }, { "epoch": 0.03694, "grad_norm": 0.8638154190597898, "learning_rate": 0.003, "loss": 4.1732, "step": 3694 }, { "epoch": 0.03695, "grad_norm": 0.7865314494927197, "learning_rate": 0.003, "loss": 4.1682, "step": 3695 }, { "epoch": 0.03696, "grad_norm": 0.811239269688311, "learning_rate": 0.003, "loss": 4.1672, "step": 3696 }, { "epoch": 0.03697, "grad_norm": 0.7527149600540498, "learning_rate": 0.003, "loss": 4.1523, "step": 3697 }, { "epoch": 0.03698, "grad_norm": 0.8165577678117001, "learning_rate": 0.003, "loss": 4.1875, "step": 3698 }, { "epoch": 0.03699, "grad_norm": 0.8971546398815292, "learning_rate": 0.003, "loss": 4.1782, "step": 3699 }, { "epoch": 0.037, "grad_norm": 0.9983619391979774, "learning_rate": 0.003, "loss": 4.1716, "step": 3700 }, { "epoch": 0.03701, "grad_norm": 0.8793417785162475, "learning_rate": 0.003, "loss": 4.1718, "step": 3701 }, { "epoch": 0.03702, "grad_norm": 0.8291238684887006, "learning_rate": 0.003, "loss": 4.1572, "step": 3702 }, { "epoch": 0.03703, "grad_norm": 0.6324908569860477, "learning_rate": 0.003, "loss": 4.1638, "step": 3703 }, { "epoch": 0.03704, "grad_norm": 0.6240031858723358, "learning_rate": 0.003, "loss": 4.1644, "step": 3704 }, { "epoch": 0.03705, "grad_norm": 0.7312113193138566, "learning_rate": 0.003, "loss": 4.1843, "step": 3705 }, { "epoch": 0.03706, "grad_norm": 0.8042606510953001, "learning_rate": 0.003, "loss": 4.1671, "step": 3706 }, { "epoch": 0.03707, "grad_norm": 0.8602980641760969, "learning_rate": 0.003, "loss": 4.1507, "step": 3707 }, { "epoch": 0.03708, "grad_norm": 0.8002371092022987, "learning_rate": 0.003, "loss": 4.1687, "step": 3708 }, { "epoch": 0.03709, "grad_norm": 0.8363140997412901, "learning_rate": 0.003, "loss": 4.1444, "step": 3709 }, { "epoch": 0.0371, "grad_norm": 0.7381408226781321, "learning_rate": 0.003, "loss": 4.1792, "step": 3710 }, { "epoch": 0.03711, "grad_norm": 0.6818914716146712, "learning_rate": 0.003, "loss": 4.1473, "step": 3711 }, { "epoch": 0.03712, "grad_norm": 0.6181242736172562, "learning_rate": 0.003, "loss": 4.1457, "step": 3712 }, { "epoch": 0.03713, "grad_norm": 0.5316945560555888, "learning_rate": 0.003, "loss": 4.1304, "step": 3713 }, { "epoch": 0.03714, "grad_norm": 0.4781488240208272, "learning_rate": 0.003, "loss": 4.1424, "step": 3714 }, { "epoch": 0.03715, "grad_norm": 0.5464841559814179, "learning_rate": 0.003, "loss": 4.1487, "step": 3715 }, { "epoch": 0.03716, "grad_norm": 0.6484458989407043, "learning_rate": 0.003, "loss": 4.1426, "step": 3716 }, { "epoch": 0.03717, "grad_norm": 0.7476388907786324, "learning_rate": 0.003, "loss": 4.1451, "step": 3717 }, { "epoch": 0.03718, "grad_norm": 1.011509883442139, "learning_rate": 0.003, "loss": 4.124, "step": 3718 }, { "epoch": 0.03719, "grad_norm": 1.2912072156164622, "learning_rate": 0.003, "loss": 4.1747, "step": 3719 }, { "epoch": 0.0372, "grad_norm": 0.5930338537992184, "learning_rate": 0.003, "loss": 4.1526, "step": 3720 }, { "epoch": 0.03721, "grad_norm": 0.830711899216099, "learning_rate": 0.003, "loss": 4.1486, "step": 3721 }, { "epoch": 0.03722, "grad_norm": 0.9431990623316837, "learning_rate": 0.003, "loss": 4.1572, "step": 3722 }, { "epoch": 0.03723, "grad_norm": 0.9432356918938554, "learning_rate": 0.003, "loss": 4.188, "step": 3723 }, { "epoch": 0.03724, "grad_norm": 0.9636776258118436, "learning_rate": 0.003, "loss": 4.1721, "step": 3724 }, { "epoch": 0.03725, "grad_norm": 0.9139015379611488, "learning_rate": 0.003, "loss": 4.1794, "step": 3725 }, { "epoch": 0.03726, "grad_norm": 0.9516147591612311, "learning_rate": 0.003, "loss": 4.1577, "step": 3726 }, { "epoch": 0.03727, "grad_norm": 0.9752852122304477, "learning_rate": 0.003, "loss": 4.1953, "step": 3727 }, { "epoch": 0.03728, "grad_norm": 0.9695953467892918, "learning_rate": 0.003, "loss": 4.1409, "step": 3728 }, { "epoch": 0.03729, "grad_norm": 0.8893393181343283, "learning_rate": 0.003, "loss": 4.1731, "step": 3729 }, { "epoch": 0.0373, "grad_norm": 0.9218487904123037, "learning_rate": 0.003, "loss": 4.1605, "step": 3730 }, { "epoch": 0.03731, "grad_norm": 0.9098551175322124, "learning_rate": 0.003, "loss": 4.1839, "step": 3731 }, { "epoch": 0.03732, "grad_norm": 0.9256072358050464, "learning_rate": 0.003, "loss": 4.1811, "step": 3732 }, { "epoch": 0.03733, "grad_norm": 0.9379985657503299, "learning_rate": 0.003, "loss": 4.1595, "step": 3733 }, { "epoch": 0.03734, "grad_norm": 0.8308298518052655, "learning_rate": 0.003, "loss": 4.1584, "step": 3734 }, { "epoch": 0.03735, "grad_norm": 0.8761219483009925, "learning_rate": 0.003, "loss": 4.1571, "step": 3735 }, { "epoch": 0.03736, "grad_norm": 1.0090875900995142, "learning_rate": 0.003, "loss": 4.1916, "step": 3736 }, { "epoch": 0.03737, "grad_norm": 0.9618846124849504, "learning_rate": 0.003, "loss": 4.1751, "step": 3737 }, { "epoch": 0.03738, "grad_norm": 0.8758346239827248, "learning_rate": 0.003, "loss": 4.1926, "step": 3738 }, { "epoch": 0.03739, "grad_norm": 0.9609527508841834, "learning_rate": 0.003, "loss": 4.1574, "step": 3739 }, { "epoch": 0.0374, "grad_norm": 1.1169937432795556, "learning_rate": 0.003, "loss": 4.1844, "step": 3740 }, { "epoch": 0.03741, "grad_norm": 0.9067264031417346, "learning_rate": 0.003, "loss": 4.1598, "step": 3741 }, { "epoch": 0.03742, "grad_norm": 0.7898423823168381, "learning_rate": 0.003, "loss": 4.151, "step": 3742 }, { "epoch": 0.03743, "grad_norm": 0.7661101135332484, "learning_rate": 0.003, "loss": 4.1462, "step": 3743 }, { "epoch": 0.03744, "grad_norm": 0.6067121401591796, "learning_rate": 0.003, "loss": 4.1712, "step": 3744 }, { "epoch": 0.03745, "grad_norm": 0.6024185554294552, "learning_rate": 0.003, "loss": 4.1531, "step": 3745 }, { "epoch": 0.03746, "grad_norm": 0.5815455713712403, "learning_rate": 0.003, "loss": 4.1601, "step": 3746 }, { "epoch": 0.03747, "grad_norm": 0.6179967126448696, "learning_rate": 0.003, "loss": 4.1659, "step": 3747 }, { "epoch": 0.03748, "grad_norm": 0.5607743590244431, "learning_rate": 0.003, "loss": 4.1341, "step": 3748 }, { "epoch": 0.03749, "grad_norm": 0.5373283353585078, "learning_rate": 0.003, "loss": 4.1453, "step": 3749 }, { "epoch": 0.0375, "grad_norm": 0.6259765189525954, "learning_rate": 0.003, "loss": 4.1675, "step": 3750 }, { "epoch": 0.03751, "grad_norm": 0.735426915588919, "learning_rate": 0.003, "loss": 4.1206, "step": 3751 }, { "epoch": 0.03752, "grad_norm": 1.1121203063280463, "learning_rate": 0.003, "loss": 4.1569, "step": 3752 }, { "epoch": 0.03753, "grad_norm": 1.1286247639907543, "learning_rate": 0.003, "loss": 4.1691, "step": 3753 }, { "epoch": 0.03754, "grad_norm": 0.654316101247794, "learning_rate": 0.003, "loss": 4.1646, "step": 3754 }, { "epoch": 0.03755, "grad_norm": 0.6144499107508359, "learning_rate": 0.003, "loss": 4.1549, "step": 3755 }, { "epoch": 0.03756, "grad_norm": 0.8558400590466524, "learning_rate": 0.003, "loss": 4.1306, "step": 3756 }, { "epoch": 0.03757, "grad_norm": 0.8999619822176426, "learning_rate": 0.003, "loss": 4.1636, "step": 3757 }, { "epoch": 0.03758, "grad_norm": 0.7120240054121859, "learning_rate": 0.003, "loss": 4.1028, "step": 3758 }, { "epoch": 0.03759, "grad_norm": 0.6839824836130421, "learning_rate": 0.003, "loss": 4.1611, "step": 3759 }, { "epoch": 0.0376, "grad_norm": 0.7584320585496591, "learning_rate": 0.003, "loss": 4.1487, "step": 3760 }, { "epoch": 0.03761, "grad_norm": 0.7956025663217148, "learning_rate": 0.003, "loss": 4.1206, "step": 3761 }, { "epoch": 0.03762, "grad_norm": 0.7804293287119324, "learning_rate": 0.003, "loss": 4.1782, "step": 3762 }, { "epoch": 0.03763, "grad_norm": 0.7176371093477065, "learning_rate": 0.003, "loss": 4.159, "step": 3763 }, { "epoch": 0.03764, "grad_norm": 0.7703219293331768, "learning_rate": 0.003, "loss": 4.1707, "step": 3764 }, { "epoch": 0.03765, "grad_norm": 0.8016568904602233, "learning_rate": 0.003, "loss": 4.191, "step": 3765 }, { "epoch": 0.03766, "grad_norm": 0.8025560943399235, "learning_rate": 0.003, "loss": 4.1326, "step": 3766 }, { "epoch": 0.03767, "grad_norm": 0.891734573749917, "learning_rate": 0.003, "loss": 4.1763, "step": 3767 }, { "epoch": 0.03768, "grad_norm": 1.043115086709847, "learning_rate": 0.003, "loss": 4.1619, "step": 3768 }, { "epoch": 0.03769, "grad_norm": 0.9976222066035275, "learning_rate": 0.003, "loss": 4.1526, "step": 3769 }, { "epoch": 0.0377, "grad_norm": 0.8470649334959093, "learning_rate": 0.003, "loss": 4.1327, "step": 3770 }, { "epoch": 0.03771, "grad_norm": 0.8274399936486776, "learning_rate": 0.003, "loss": 4.1469, "step": 3771 }, { "epoch": 0.03772, "grad_norm": 0.7871019788741513, "learning_rate": 0.003, "loss": 4.1668, "step": 3772 }, { "epoch": 0.03773, "grad_norm": 0.7390202831045702, "learning_rate": 0.003, "loss": 4.1698, "step": 3773 }, { "epoch": 0.03774, "grad_norm": 0.7113411820252128, "learning_rate": 0.003, "loss": 4.1532, "step": 3774 }, { "epoch": 0.03775, "grad_norm": 0.7435094864466253, "learning_rate": 0.003, "loss": 4.1705, "step": 3775 }, { "epoch": 0.03776, "grad_norm": 0.8919757094791062, "learning_rate": 0.003, "loss": 4.1496, "step": 3776 }, { "epoch": 0.03777, "grad_norm": 1.152057291990839, "learning_rate": 0.003, "loss": 4.1719, "step": 3777 }, { "epoch": 0.03778, "grad_norm": 1.0115617936585513, "learning_rate": 0.003, "loss": 4.1493, "step": 3778 }, { "epoch": 0.03779, "grad_norm": 0.9872755421624871, "learning_rate": 0.003, "loss": 4.1659, "step": 3779 }, { "epoch": 0.0378, "grad_norm": 0.8946795429906949, "learning_rate": 0.003, "loss": 4.164, "step": 3780 }, { "epoch": 0.03781, "grad_norm": 0.9373368557855075, "learning_rate": 0.003, "loss": 4.1956, "step": 3781 }, { "epoch": 0.03782, "grad_norm": 0.853748438513948, "learning_rate": 0.003, "loss": 4.162, "step": 3782 }, { "epoch": 0.03783, "grad_norm": 0.718548585744198, "learning_rate": 0.003, "loss": 4.164, "step": 3783 }, { "epoch": 0.03784, "grad_norm": 0.7444967401418248, "learning_rate": 0.003, "loss": 4.1631, "step": 3784 }, { "epoch": 0.03785, "grad_norm": 0.8491520079890601, "learning_rate": 0.003, "loss": 4.1505, "step": 3785 }, { "epoch": 0.03786, "grad_norm": 0.872023269439096, "learning_rate": 0.003, "loss": 4.1688, "step": 3786 }, { "epoch": 0.03787, "grad_norm": 0.8535181083648594, "learning_rate": 0.003, "loss": 4.153, "step": 3787 }, { "epoch": 0.03788, "grad_norm": 0.9108993849141537, "learning_rate": 0.003, "loss": 4.1855, "step": 3788 }, { "epoch": 0.03789, "grad_norm": 0.9794959022265245, "learning_rate": 0.003, "loss": 4.1225, "step": 3789 }, { "epoch": 0.0379, "grad_norm": 1.048224892887674, "learning_rate": 0.003, "loss": 4.1466, "step": 3790 }, { "epoch": 0.03791, "grad_norm": 0.8968035808950884, "learning_rate": 0.003, "loss": 4.1596, "step": 3791 }, { "epoch": 0.03792, "grad_norm": 0.8368499561515244, "learning_rate": 0.003, "loss": 4.1883, "step": 3792 }, { "epoch": 0.03793, "grad_norm": 0.8410884942872239, "learning_rate": 0.003, "loss": 4.1811, "step": 3793 }, { "epoch": 0.03794, "grad_norm": 0.8526793361609423, "learning_rate": 0.003, "loss": 4.1863, "step": 3794 }, { "epoch": 0.03795, "grad_norm": 0.8096867690017788, "learning_rate": 0.003, "loss": 4.1663, "step": 3795 }, { "epoch": 0.03796, "grad_norm": 0.7081089049008689, "learning_rate": 0.003, "loss": 4.145, "step": 3796 }, { "epoch": 0.03797, "grad_norm": 0.6256114559170813, "learning_rate": 0.003, "loss": 4.13, "step": 3797 }, { "epoch": 0.03798, "grad_norm": 0.5712390179880806, "learning_rate": 0.003, "loss": 4.1444, "step": 3798 }, { "epoch": 0.03799, "grad_norm": 0.6395912776533417, "learning_rate": 0.003, "loss": 4.1549, "step": 3799 }, { "epoch": 0.038, "grad_norm": 0.7194653527929541, "learning_rate": 0.003, "loss": 4.1628, "step": 3800 }, { "epoch": 0.03801, "grad_norm": 0.9532685446077319, "learning_rate": 0.003, "loss": 4.1743, "step": 3801 }, { "epoch": 0.03802, "grad_norm": 1.1563370657823024, "learning_rate": 0.003, "loss": 4.1547, "step": 3802 }, { "epoch": 0.03803, "grad_norm": 0.7722962818272365, "learning_rate": 0.003, "loss": 4.1744, "step": 3803 }, { "epoch": 0.03804, "grad_norm": 0.7050390933953039, "learning_rate": 0.003, "loss": 4.141, "step": 3804 }, { "epoch": 0.03805, "grad_norm": 0.8158853348007252, "learning_rate": 0.003, "loss": 4.158, "step": 3805 }, { "epoch": 0.03806, "grad_norm": 0.8472580689148763, "learning_rate": 0.003, "loss": 4.1592, "step": 3806 }, { "epoch": 0.03807, "grad_norm": 0.9423141657174167, "learning_rate": 0.003, "loss": 4.1545, "step": 3807 }, { "epoch": 0.03808, "grad_norm": 1.1215335628084824, "learning_rate": 0.003, "loss": 4.1793, "step": 3808 }, { "epoch": 0.03809, "grad_norm": 0.9834441734645248, "learning_rate": 0.003, "loss": 4.1571, "step": 3809 }, { "epoch": 0.0381, "grad_norm": 0.9317910497451528, "learning_rate": 0.003, "loss": 4.1413, "step": 3810 }, { "epoch": 0.03811, "grad_norm": 0.8413021683381647, "learning_rate": 0.003, "loss": 4.17, "step": 3811 }, { "epoch": 0.03812, "grad_norm": 0.772209856524563, "learning_rate": 0.003, "loss": 4.1585, "step": 3812 }, { "epoch": 0.03813, "grad_norm": 0.7904321983957339, "learning_rate": 0.003, "loss": 4.1306, "step": 3813 }, { "epoch": 0.03814, "grad_norm": 0.7457108043591989, "learning_rate": 0.003, "loss": 4.1598, "step": 3814 }, { "epoch": 0.03815, "grad_norm": 0.7103508451290527, "learning_rate": 0.003, "loss": 4.1467, "step": 3815 }, { "epoch": 0.03816, "grad_norm": 0.6613518230453459, "learning_rate": 0.003, "loss": 4.1473, "step": 3816 }, { "epoch": 0.03817, "grad_norm": 0.6678002497479578, "learning_rate": 0.003, "loss": 4.126, "step": 3817 }, { "epoch": 0.03818, "grad_norm": 0.5546400433556298, "learning_rate": 0.003, "loss": 4.1735, "step": 3818 }, { "epoch": 0.03819, "grad_norm": 0.5048588514692719, "learning_rate": 0.003, "loss": 4.1246, "step": 3819 }, { "epoch": 0.0382, "grad_norm": 0.5022259630766596, "learning_rate": 0.003, "loss": 4.1719, "step": 3820 }, { "epoch": 0.03821, "grad_norm": 0.5319841736014501, "learning_rate": 0.003, "loss": 4.1165, "step": 3821 }, { "epoch": 0.03822, "grad_norm": 0.611206545487655, "learning_rate": 0.003, "loss": 4.1392, "step": 3822 }, { "epoch": 0.03823, "grad_norm": 0.7859585505396786, "learning_rate": 0.003, "loss": 4.1387, "step": 3823 }, { "epoch": 0.03824, "grad_norm": 1.170117791068494, "learning_rate": 0.003, "loss": 4.1483, "step": 3824 }, { "epoch": 0.03825, "grad_norm": 1.1642472923233964, "learning_rate": 0.003, "loss": 4.1605, "step": 3825 }, { "epoch": 0.03826, "grad_norm": 0.8386882795746682, "learning_rate": 0.003, "loss": 4.148, "step": 3826 }, { "epoch": 0.03827, "grad_norm": 0.7280705231607668, "learning_rate": 0.003, "loss": 4.1699, "step": 3827 }, { "epoch": 0.03828, "grad_norm": 0.8654688174806783, "learning_rate": 0.003, "loss": 4.1514, "step": 3828 }, { "epoch": 0.03829, "grad_norm": 0.8475583789249039, "learning_rate": 0.003, "loss": 4.1549, "step": 3829 }, { "epoch": 0.0383, "grad_norm": 0.7947652875636247, "learning_rate": 0.003, "loss": 4.1418, "step": 3830 }, { "epoch": 0.03831, "grad_norm": 0.7168989453848753, "learning_rate": 0.003, "loss": 4.1398, "step": 3831 }, { "epoch": 0.03832, "grad_norm": 0.7200550330645938, "learning_rate": 0.003, "loss": 4.1382, "step": 3832 }, { "epoch": 0.03833, "grad_norm": 0.7587490659810894, "learning_rate": 0.003, "loss": 4.1411, "step": 3833 }, { "epoch": 0.03834, "grad_norm": 0.7136259183365191, "learning_rate": 0.003, "loss": 4.1692, "step": 3834 }, { "epoch": 0.03835, "grad_norm": 0.7659380295468856, "learning_rate": 0.003, "loss": 4.1602, "step": 3835 }, { "epoch": 0.03836, "grad_norm": 0.783318958508309, "learning_rate": 0.003, "loss": 4.1639, "step": 3836 }, { "epoch": 0.03837, "grad_norm": 0.8832315963306094, "learning_rate": 0.003, "loss": 4.1514, "step": 3837 }, { "epoch": 0.03838, "grad_norm": 0.9447883209168796, "learning_rate": 0.003, "loss": 4.1449, "step": 3838 }, { "epoch": 0.03839, "grad_norm": 1.0145507021392315, "learning_rate": 0.003, "loss": 4.1519, "step": 3839 }, { "epoch": 0.0384, "grad_norm": 1.3460191042861327, "learning_rate": 0.003, "loss": 4.1727, "step": 3840 }, { "epoch": 0.03841, "grad_norm": 0.7909608584954623, "learning_rate": 0.003, "loss": 4.1612, "step": 3841 }, { "epoch": 0.03842, "grad_norm": 0.6421152217888612, "learning_rate": 0.003, "loss": 4.1585, "step": 3842 }, { "epoch": 0.03843, "grad_norm": 0.7366594214130205, "learning_rate": 0.003, "loss": 4.1468, "step": 3843 }, { "epoch": 0.03844, "grad_norm": 0.756017458900284, "learning_rate": 0.003, "loss": 4.1537, "step": 3844 }, { "epoch": 0.03845, "grad_norm": 0.7922638056100233, "learning_rate": 0.003, "loss": 4.1552, "step": 3845 }, { "epoch": 0.03846, "grad_norm": 0.8441235594817899, "learning_rate": 0.003, "loss": 4.1721, "step": 3846 }, { "epoch": 0.03847, "grad_norm": 0.8282933496250385, "learning_rate": 0.003, "loss": 4.1618, "step": 3847 }, { "epoch": 0.03848, "grad_norm": 0.9404305490862656, "learning_rate": 0.003, "loss": 4.1493, "step": 3848 }, { "epoch": 0.03849, "grad_norm": 1.3038885588549676, "learning_rate": 0.003, "loss": 4.1779, "step": 3849 }, { "epoch": 0.0385, "grad_norm": 0.910760434256003, "learning_rate": 0.003, "loss": 4.1759, "step": 3850 }, { "epoch": 0.03851, "grad_norm": 0.9851608887041251, "learning_rate": 0.003, "loss": 4.1424, "step": 3851 }, { "epoch": 0.03852, "grad_norm": 1.115936013932957, "learning_rate": 0.003, "loss": 4.1839, "step": 3852 }, { "epoch": 0.03853, "grad_norm": 0.8625131320120981, "learning_rate": 0.003, "loss": 4.171, "step": 3853 }, { "epoch": 0.03854, "grad_norm": 0.8093220938164782, "learning_rate": 0.003, "loss": 4.1633, "step": 3854 }, { "epoch": 0.03855, "grad_norm": 0.726455705711537, "learning_rate": 0.003, "loss": 4.1793, "step": 3855 }, { "epoch": 0.03856, "grad_norm": 0.7638648169981871, "learning_rate": 0.003, "loss": 4.1493, "step": 3856 }, { "epoch": 0.03857, "grad_norm": 0.806255198654299, "learning_rate": 0.003, "loss": 4.1543, "step": 3857 }, { "epoch": 0.03858, "grad_norm": 0.831071410604779, "learning_rate": 0.003, "loss": 4.15, "step": 3858 }, { "epoch": 0.03859, "grad_norm": 0.8360163780390938, "learning_rate": 0.003, "loss": 4.1729, "step": 3859 }, { "epoch": 0.0386, "grad_norm": 0.9270300839146026, "learning_rate": 0.003, "loss": 4.1383, "step": 3860 }, { "epoch": 0.03861, "grad_norm": 1.0571665911277832, "learning_rate": 0.003, "loss": 4.1787, "step": 3861 }, { "epoch": 0.03862, "grad_norm": 0.9184603613158857, "learning_rate": 0.003, "loss": 4.1542, "step": 3862 }, { "epoch": 0.03863, "grad_norm": 0.9497546960268847, "learning_rate": 0.003, "loss": 4.1512, "step": 3863 }, { "epoch": 0.03864, "grad_norm": 1.0608016243824416, "learning_rate": 0.003, "loss": 4.1672, "step": 3864 }, { "epoch": 0.03865, "grad_norm": 0.9123627957656194, "learning_rate": 0.003, "loss": 4.173, "step": 3865 }, { "epoch": 0.03866, "grad_norm": 0.8643631111703186, "learning_rate": 0.003, "loss": 4.1491, "step": 3866 }, { "epoch": 0.03867, "grad_norm": 0.7131249254546375, "learning_rate": 0.003, "loss": 4.1275, "step": 3867 }, { "epoch": 0.03868, "grad_norm": 0.6223536314544267, "learning_rate": 0.003, "loss": 4.162, "step": 3868 }, { "epoch": 0.03869, "grad_norm": 0.6196862864928006, "learning_rate": 0.003, "loss": 4.1488, "step": 3869 }, { "epoch": 0.0387, "grad_norm": 0.6383959402879603, "learning_rate": 0.003, "loss": 4.1571, "step": 3870 }, { "epoch": 0.03871, "grad_norm": 0.7302513860623053, "learning_rate": 0.003, "loss": 4.1346, "step": 3871 }, { "epoch": 0.03872, "grad_norm": 0.8920320690566674, "learning_rate": 0.003, "loss": 4.1361, "step": 3872 }, { "epoch": 0.03873, "grad_norm": 0.9343331762751365, "learning_rate": 0.003, "loss": 4.1802, "step": 3873 }, { "epoch": 0.03874, "grad_norm": 1.0090689056461979, "learning_rate": 0.003, "loss": 4.1531, "step": 3874 }, { "epoch": 0.03875, "grad_norm": 1.0371067027925491, "learning_rate": 0.003, "loss": 4.156, "step": 3875 }, { "epoch": 0.03876, "grad_norm": 1.0801564549744878, "learning_rate": 0.003, "loss": 4.1919, "step": 3876 }, { "epoch": 0.03877, "grad_norm": 1.0484528784031752, "learning_rate": 0.003, "loss": 4.1737, "step": 3877 }, { "epoch": 0.03878, "grad_norm": 0.9887410146395552, "learning_rate": 0.003, "loss": 4.187, "step": 3878 }, { "epoch": 0.03879, "grad_norm": 0.9041370865255791, "learning_rate": 0.003, "loss": 4.1845, "step": 3879 }, { "epoch": 0.0388, "grad_norm": 0.8365742286229014, "learning_rate": 0.003, "loss": 4.144, "step": 3880 }, { "epoch": 0.03881, "grad_norm": 0.8968458423015266, "learning_rate": 0.003, "loss": 4.1712, "step": 3881 }, { "epoch": 0.03882, "grad_norm": 1.1815995202626826, "learning_rate": 0.003, "loss": 4.1894, "step": 3882 }, { "epoch": 0.03883, "grad_norm": 1.004450379233401, "learning_rate": 0.003, "loss": 4.1547, "step": 3883 }, { "epoch": 0.03884, "grad_norm": 0.8667121928698747, "learning_rate": 0.003, "loss": 4.1656, "step": 3884 }, { "epoch": 0.03885, "grad_norm": 0.6868027088778039, "learning_rate": 0.003, "loss": 4.1755, "step": 3885 }, { "epoch": 0.03886, "grad_norm": 0.7182515814021971, "learning_rate": 0.003, "loss": 4.1483, "step": 3886 }, { "epoch": 0.03887, "grad_norm": 0.737302592292138, "learning_rate": 0.003, "loss": 4.1586, "step": 3887 }, { "epoch": 0.03888, "grad_norm": 0.6947474627935248, "learning_rate": 0.003, "loss": 4.1354, "step": 3888 }, { "epoch": 0.03889, "grad_norm": 0.676048384811591, "learning_rate": 0.003, "loss": 4.1523, "step": 3889 }, { "epoch": 0.0389, "grad_norm": 0.6771755441542272, "learning_rate": 0.003, "loss": 4.1398, "step": 3890 }, { "epoch": 0.03891, "grad_norm": 0.7246456194548712, "learning_rate": 0.003, "loss": 4.149, "step": 3891 }, { "epoch": 0.03892, "grad_norm": 0.7172714775987227, "learning_rate": 0.003, "loss": 4.1606, "step": 3892 }, { "epoch": 0.03893, "grad_norm": 0.8341243893405708, "learning_rate": 0.003, "loss": 4.1388, "step": 3893 }, { "epoch": 0.03894, "grad_norm": 0.9586809206015507, "learning_rate": 0.003, "loss": 4.1519, "step": 3894 }, { "epoch": 0.03895, "grad_norm": 1.1044375554559078, "learning_rate": 0.003, "loss": 4.1352, "step": 3895 }, { "epoch": 0.03896, "grad_norm": 0.9190112275090493, "learning_rate": 0.003, "loss": 4.1578, "step": 3896 }, { "epoch": 0.03897, "grad_norm": 0.7602784227805208, "learning_rate": 0.003, "loss": 4.1256, "step": 3897 }, { "epoch": 0.03898, "grad_norm": 0.639713595234581, "learning_rate": 0.003, "loss": 4.1548, "step": 3898 }, { "epoch": 0.03899, "grad_norm": 0.7184087029382192, "learning_rate": 0.003, "loss": 4.1491, "step": 3899 }, { "epoch": 0.039, "grad_norm": 0.7814816186886983, "learning_rate": 0.003, "loss": 4.1486, "step": 3900 }, { "epoch": 0.03901, "grad_norm": 0.7162921844556793, "learning_rate": 0.003, "loss": 4.1601, "step": 3901 }, { "epoch": 0.03902, "grad_norm": 0.8155759206891086, "learning_rate": 0.003, "loss": 4.1337, "step": 3902 }, { "epoch": 0.03903, "grad_norm": 1.0636290055979989, "learning_rate": 0.003, "loss": 4.1778, "step": 3903 }, { "epoch": 0.03904, "grad_norm": 1.0177125932370377, "learning_rate": 0.003, "loss": 4.1589, "step": 3904 }, { "epoch": 0.03905, "grad_norm": 0.9442621578301694, "learning_rate": 0.003, "loss": 4.1753, "step": 3905 }, { "epoch": 0.03906, "grad_norm": 0.790601653351365, "learning_rate": 0.003, "loss": 4.1438, "step": 3906 }, { "epoch": 0.03907, "grad_norm": 0.7794484704379305, "learning_rate": 0.003, "loss": 4.156, "step": 3907 }, { "epoch": 0.03908, "grad_norm": 0.7560452632943567, "learning_rate": 0.003, "loss": 4.1382, "step": 3908 }, { "epoch": 0.03909, "grad_norm": 0.7946846197040874, "learning_rate": 0.003, "loss": 4.1459, "step": 3909 }, { "epoch": 0.0391, "grad_norm": 0.8613810775384986, "learning_rate": 0.003, "loss": 4.1768, "step": 3910 }, { "epoch": 0.03911, "grad_norm": 0.8386874382765742, "learning_rate": 0.003, "loss": 4.1609, "step": 3911 }, { "epoch": 0.03912, "grad_norm": 0.9865330307604043, "learning_rate": 0.003, "loss": 4.1673, "step": 3912 }, { "epoch": 0.03913, "grad_norm": 1.1331373891184677, "learning_rate": 0.003, "loss": 4.151, "step": 3913 }, { "epoch": 0.03914, "grad_norm": 0.7046370423377648, "learning_rate": 0.003, "loss": 4.1706, "step": 3914 }, { "epoch": 0.03915, "grad_norm": 0.7467899755928374, "learning_rate": 0.003, "loss": 4.1435, "step": 3915 }, { "epoch": 0.03916, "grad_norm": 0.892872585324721, "learning_rate": 0.003, "loss": 4.1734, "step": 3916 }, { "epoch": 0.03917, "grad_norm": 1.228738480994026, "learning_rate": 0.003, "loss": 4.1732, "step": 3917 }, { "epoch": 0.03918, "grad_norm": 0.9438157037038833, "learning_rate": 0.003, "loss": 4.1549, "step": 3918 }, { "epoch": 0.03919, "grad_norm": 0.6952503471849076, "learning_rate": 0.003, "loss": 4.1167, "step": 3919 }, { "epoch": 0.0392, "grad_norm": 0.7048182141212009, "learning_rate": 0.003, "loss": 4.1429, "step": 3920 }, { "epoch": 0.03921, "grad_norm": 0.879657398847784, "learning_rate": 0.003, "loss": 4.1435, "step": 3921 }, { "epoch": 0.03922, "grad_norm": 1.1793211640732564, "learning_rate": 0.003, "loss": 4.1943, "step": 3922 }, { "epoch": 0.03923, "grad_norm": 0.9549026205906491, "learning_rate": 0.003, "loss": 4.1448, "step": 3923 }, { "epoch": 0.03924, "grad_norm": 0.7775760685887316, "learning_rate": 0.003, "loss": 4.1477, "step": 3924 }, { "epoch": 0.03925, "grad_norm": 0.7598517272028414, "learning_rate": 0.003, "loss": 4.1522, "step": 3925 }, { "epoch": 0.03926, "grad_norm": 0.7119046881729456, "learning_rate": 0.003, "loss": 4.139, "step": 3926 }, { "epoch": 0.03927, "grad_norm": 0.8115139166245634, "learning_rate": 0.003, "loss": 4.1534, "step": 3927 }, { "epoch": 0.03928, "grad_norm": 0.8496702812010677, "learning_rate": 0.003, "loss": 4.172, "step": 3928 }, { "epoch": 0.03929, "grad_norm": 0.914389232519974, "learning_rate": 0.003, "loss": 4.1836, "step": 3929 }, { "epoch": 0.0393, "grad_norm": 0.934142877257439, "learning_rate": 0.003, "loss": 4.1699, "step": 3930 }, { "epoch": 0.03931, "grad_norm": 0.9169035963389445, "learning_rate": 0.003, "loss": 4.1746, "step": 3931 }, { "epoch": 0.03932, "grad_norm": 0.9586032495283386, "learning_rate": 0.003, "loss": 4.1767, "step": 3932 }, { "epoch": 0.03933, "grad_norm": 0.8943397447772738, "learning_rate": 0.003, "loss": 4.1604, "step": 3933 }, { "epoch": 0.03934, "grad_norm": 1.1458130542379994, "learning_rate": 0.003, "loss": 4.1591, "step": 3934 }, { "epoch": 0.03935, "grad_norm": 1.1119954153292497, "learning_rate": 0.003, "loss": 4.1564, "step": 3935 }, { "epoch": 0.03936, "grad_norm": 0.9651500208576397, "learning_rate": 0.003, "loss": 4.118, "step": 3936 }, { "epoch": 0.03937, "grad_norm": 0.9009017242821815, "learning_rate": 0.003, "loss": 4.1567, "step": 3937 }, { "epoch": 0.03938, "grad_norm": 0.682722210666222, "learning_rate": 0.003, "loss": 4.1508, "step": 3938 }, { "epoch": 0.03939, "grad_norm": 0.6257361874612599, "learning_rate": 0.003, "loss": 4.1296, "step": 3939 }, { "epoch": 0.0394, "grad_norm": 0.6910228445809036, "learning_rate": 0.003, "loss": 4.157, "step": 3940 }, { "epoch": 0.03941, "grad_norm": 0.8752420058270881, "learning_rate": 0.003, "loss": 4.1523, "step": 3941 }, { "epoch": 0.03942, "grad_norm": 1.0090937202006367, "learning_rate": 0.003, "loss": 4.1701, "step": 3942 }, { "epoch": 0.03943, "grad_norm": 0.8713788492851877, "learning_rate": 0.003, "loss": 4.1594, "step": 3943 }, { "epoch": 0.03944, "grad_norm": 0.6395719699631305, "learning_rate": 0.003, "loss": 4.1691, "step": 3944 }, { "epoch": 0.03945, "grad_norm": 0.6237828376004393, "learning_rate": 0.003, "loss": 4.1351, "step": 3945 }, { "epoch": 0.03946, "grad_norm": 0.7996865672671727, "learning_rate": 0.003, "loss": 4.1662, "step": 3946 }, { "epoch": 0.03947, "grad_norm": 0.8027734166618509, "learning_rate": 0.003, "loss": 4.1684, "step": 3947 }, { "epoch": 0.03948, "grad_norm": 0.7480672212389291, "learning_rate": 0.003, "loss": 4.185, "step": 3948 }, { "epoch": 0.03949, "grad_norm": 0.7986884544671021, "learning_rate": 0.003, "loss": 4.1278, "step": 3949 }, { "epoch": 0.0395, "grad_norm": 0.7882737042471627, "learning_rate": 0.003, "loss": 4.1581, "step": 3950 }, { "epoch": 0.03951, "grad_norm": 0.8513453245753498, "learning_rate": 0.003, "loss": 4.1908, "step": 3951 }, { "epoch": 0.03952, "grad_norm": 0.9604447646143368, "learning_rate": 0.003, "loss": 4.1733, "step": 3952 }, { "epoch": 0.03953, "grad_norm": 0.9585905884824354, "learning_rate": 0.003, "loss": 4.1514, "step": 3953 }, { "epoch": 0.03954, "grad_norm": 0.9656678821151307, "learning_rate": 0.003, "loss": 4.1554, "step": 3954 }, { "epoch": 0.03955, "grad_norm": 0.9222198486019069, "learning_rate": 0.003, "loss": 4.1565, "step": 3955 }, { "epoch": 0.03956, "grad_norm": 1.0225324631256894, "learning_rate": 0.003, "loss": 4.1615, "step": 3956 }, { "epoch": 0.03957, "grad_norm": 1.0431715292566202, "learning_rate": 0.003, "loss": 4.1665, "step": 3957 }, { "epoch": 0.03958, "grad_norm": 1.0749240135902176, "learning_rate": 0.003, "loss": 4.1477, "step": 3958 }, { "epoch": 0.03959, "grad_norm": 0.8880519487734211, "learning_rate": 0.003, "loss": 4.1433, "step": 3959 }, { "epoch": 0.0396, "grad_norm": 0.770078201609415, "learning_rate": 0.003, "loss": 4.1591, "step": 3960 }, { "epoch": 0.03961, "grad_norm": 0.7312192163132855, "learning_rate": 0.003, "loss": 4.1531, "step": 3961 }, { "epoch": 0.03962, "grad_norm": 0.8077388602218855, "learning_rate": 0.003, "loss": 4.1671, "step": 3962 }, { "epoch": 0.03963, "grad_norm": 0.9497511656553155, "learning_rate": 0.003, "loss": 4.1518, "step": 3963 }, { "epoch": 0.03964, "grad_norm": 1.09506865726512, "learning_rate": 0.003, "loss": 4.1627, "step": 3964 }, { "epoch": 0.03965, "grad_norm": 0.9655209694273282, "learning_rate": 0.003, "loss": 4.1567, "step": 3965 }, { "epoch": 0.03966, "grad_norm": 0.9060903529494769, "learning_rate": 0.003, "loss": 4.197, "step": 3966 }, { "epoch": 0.03967, "grad_norm": 0.9168805934226382, "learning_rate": 0.003, "loss": 4.1544, "step": 3967 }, { "epoch": 0.03968, "grad_norm": 0.8082674185213146, "learning_rate": 0.003, "loss": 4.1795, "step": 3968 }, { "epoch": 0.03969, "grad_norm": 0.8323391059785374, "learning_rate": 0.003, "loss": 4.1564, "step": 3969 }, { "epoch": 0.0397, "grad_norm": 0.9524651490900072, "learning_rate": 0.003, "loss": 4.1671, "step": 3970 }, { "epoch": 0.03971, "grad_norm": 0.9567901053636726, "learning_rate": 0.003, "loss": 4.1454, "step": 3971 }, { "epoch": 0.03972, "grad_norm": 0.9696093602166972, "learning_rate": 0.003, "loss": 4.1654, "step": 3972 }, { "epoch": 0.03973, "grad_norm": 0.9370133833063669, "learning_rate": 0.003, "loss": 4.1687, "step": 3973 }, { "epoch": 0.03974, "grad_norm": 0.8608341541886206, "learning_rate": 0.003, "loss": 4.1643, "step": 3974 }, { "epoch": 0.03975, "grad_norm": 0.9108884430543455, "learning_rate": 0.003, "loss": 4.1535, "step": 3975 }, { "epoch": 0.03976, "grad_norm": 0.9241700018742819, "learning_rate": 0.003, "loss": 4.1895, "step": 3976 }, { "epoch": 0.03977, "grad_norm": 0.8946194243663588, "learning_rate": 0.003, "loss": 4.1801, "step": 3977 }, { "epoch": 0.03978, "grad_norm": 0.8947344596350721, "learning_rate": 0.003, "loss": 4.1464, "step": 3978 }, { "epoch": 0.03979, "grad_norm": 0.9770682330119697, "learning_rate": 0.003, "loss": 4.166, "step": 3979 }, { "epoch": 0.0398, "grad_norm": 0.938338044149382, "learning_rate": 0.003, "loss": 4.1591, "step": 3980 }, { "epoch": 0.03981, "grad_norm": 1.0362927866956353, "learning_rate": 0.003, "loss": 4.1556, "step": 3981 }, { "epoch": 0.03982, "grad_norm": 1.1980787664669188, "learning_rate": 0.003, "loss": 4.1442, "step": 3982 }, { "epoch": 0.03983, "grad_norm": 0.8306482940322903, "learning_rate": 0.003, "loss": 4.1542, "step": 3983 }, { "epoch": 0.03984, "grad_norm": 0.7984572534714871, "learning_rate": 0.003, "loss": 4.1641, "step": 3984 }, { "epoch": 0.03985, "grad_norm": 0.7618742502189577, "learning_rate": 0.003, "loss": 4.1568, "step": 3985 }, { "epoch": 0.03986, "grad_norm": 0.7612926191359269, "learning_rate": 0.003, "loss": 4.1559, "step": 3986 }, { "epoch": 0.03987, "grad_norm": 0.8001366102763903, "learning_rate": 0.003, "loss": 4.1589, "step": 3987 }, { "epoch": 0.03988, "grad_norm": 1.0875345535497536, "learning_rate": 0.003, "loss": 4.1695, "step": 3988 }, { "epoch": 0.03989, "grad_norm": 0.9012674416468869, "learning_rate": 0.003, "loss": 4.156, "step": 3989 }, { "epoch": 0.0399, "grad_norm": 0.7809977747028035, "learning_rate": 0.003, "loss": 4.1238, "step": 3990 }, { "epoch": 0.03991, "grad_norm": 0.7695628801274876, "learning_rate": 0.003, "loss": 4.1692, "step": 3991 }, { "epoch": 0.03992, "grad_norm": 0.6898705145692119, "learning_rate": 0.003, "loss": 4.1474, "step": 3992 }, { "epoch": 0.03993, "grad_norm": 0.7439323328212757, "learning_rate": 0.003, "loss": 4.1436, "step": 3993 }, { "epoch": 0.03994, "grad_norm": 0.8108536082334781, "learning_rate": 0.003, "loss": 4.1442, "step": 3994 }, { "epoch": 0.03995, "grad_norm": 0.8023110096378178, "learning_rate": 0.003, "loss": 4.1319, "step": 3995 }, { "epoch": 0.03996, "grad_norm": 0.7568430508180715, "learning_rate": 0.003, "loss": 4.1644, "step": 3996 }, { "epoch": 0.03997, "grad_norm": 0.8365583946977457, "learning_rate": 0.003, "loss": 4.1688, "step": 3997 }, { "epoch": 0.03998, "grad_norm": 0.9142234466239328, "learning_rate": 0.003, "loss": 4.1721, "step": 3998 }, { "epoch": 0.03999, "grad_norm": 1.038283900357651, "learning_rate": 0.003, "loss": 4.1795, "step": 3999 }, { "epoch": 0.04, "grad_norm": 1.1201832946054642, "learning_rate": 0.003, "loss": 4.1848, "step": 4000 }, { "epoch": 0.04001, "grad_norm": 0.808173413358703, "learning_rate": 0.003, "loss": 4.1725, "step": 4001 }, { "epoch": 0.04002, "grad_norm": 0.6629014428039164, "learning_rate": 0.003, "loss": 4.1423, "step": 4002 }, { "epoch": 0.04003, "grad_norm": 0.7358065961733515, "learning_rate": 0.003, "loss": 4.1644, "step": 4003 }, { "epoch": 0.04004, "grad_norm": 0.8629016209004771, "learning_rate": 0.003, "loss": 4.173, "step": 4004 }, { "epoch": 0.04005, "grad_norm": 0.8785262035574583, "learning_rate": 0.003, "loss": 4.1835, "step": 4005 }, { "epoch": 0.04006, "grad_norm": 0.8675854999926088, "learning_rate": 0.003, "loss": 4.139, "step": 4006 }, { "epoch": 0.04007, "grad_norm": 0.7186371802576791, "learning_rate": 0.003, "loss": 4.1643, "step": 4007 }, { "epoch": 0.04008, "grad_norm": 0.6367987674963547, "learning_rate": 0.003, "loss": 4.1369, "step": 4008 }, { "epoch": 0.04009, "grad_norm": 0.6717991325425072, "learning_rate": 0.003, "loss": 4.1344, "step": 4009 }, { "epoch": 0.0401, "grad_norm": 1.0170810285743253, "learning_rate": 0.003, "loss": 4.1593, "step": 4010 }, { "epoch": 0.04011, "grad_norm": 1.3056483319688719, "learning_rate": 0.003, "loss": 4.1768, "step": 4011 }, { "epoch": 0.04012, "grad_norm": 0.760243524633557, "learning_rate": 0.003, "loss": 4.1595, "step": 4012 }, { "epoch": 0.04013, "grad_norm": 0.7397871657108581, "learning_rate": 0.003, "loss": 4.1563, "step": 4013 }, { "epoch": 0.04014, "grad_norm": 0.7499847916552277, "learning_rate": 0.003, "loss": 4.1409, "step": 4014 }, { "epoch": 0.04015, "grad_norm": 0.8867737502420797, "learning_rate": 0.003, "loss": 4.1225, "step": 4015 }, { "epoch": 0.04016, "grad_norm": 0.9755243040378807, "learning_rate": 0.003, "loss": 4.1449, "step": 4016 }, { "epoch": 0.04017, "grad_norm": 1.0307798897636882, "learning_rate": 0.003, "loss": 4.1884, "step": 4017 }, { "epoch": 0.04018, "grad_norm": 0.9218658209542674, "learning_rate": 0.003, "loss": 4.1539, "step": 4018 }, { "epoch": 0.04019, "grad_norm": 0.7251607327676213, "learning_rate": 0.003, "loss": 4.1289, "step": 4019 }, { "epoch": 0.0402, "grad_norm": 0.6906853299354309, "learning_rate": 0.003, "loss": 4.1606, "step": 4020 }, { "epoch": 0.04021, "grad_norm": 0.7032884976043096, "learning_rate": 0.003, "loss": 4.1343, "step": 4021 }, { "epoch": 0.04022, "grad_norm": 0.8205807169248119, "learning_rate": 0.003, "loss": 4.1352, "step": 4022 }, { "epoch": 0.04023, "grad_norm": 0.8414169267763164, "learning_rate": 0.003, "loss": 4.1364, "step": 4023 }, { "epoch": 0.04024, "grad_norm": 0.867716750706141, "learning_rate": 0.003, "loss": 4.1258, "step": 4024 }, { "epoch": 0.04025, "grad_norm": 0.8383796894954268, "learning_rate": 0.003, "loss": 4.1328, "step": 4025 }, { "epoch": 0.04026, "grad_norm": 0.8750670478361139, "learning_rate": 0.003, "loss": 4.1616, "step": 4026 }, { "epoch": 0.04027, "grad_norm": 0.8493399946851298, "learning_rate": 0.003, "loss": 4.1643, "step": 4027 }, { "epoch": 0.04028, "grad_norm": 0.8679376610707137, "learning_rate": 0.003, "loss": 4.1478, "step": 4028 }, { "epoch": 0.04029, "grad_norm": 0.7870403687200003, "learning_rate": 0.003, "loss": 4.1729, "step": 4029 }, { "epoch": 0.0403, "grad_norm": 0.7657237380129638, "learning_rate": 0.003, "loss": 4.1216, "step": 4030 }, { "epoch": 0.04031, "grad_norm": 0.8834770309040775, "learning_rate": 0.003, "loss": 4.1436, "step": 4031 }, { "epoch": 0.04032, "grad_norm": 0.971327728269083, "learning_rate": 0.003, "loss": 4.1729, "step": 4032 }, { "epoch": 0.04033, "grad_norm": 0.9938430497115239, "learning_rate": 0.003, "loss": 4.1739, "step": 4033 }, { "epoch": 0.04034, "grad_norm": 0.9507001357433588, "learning_rate": 0.003, "loss": 4.1397, "step": 4034 }, { "epoch": 0.04035, "grad_norm": 0.7647174546606386, "learning_rate": 0.003, "loss": 4.1582, "step": 4035 }, { "epoch": 0.04036, "grad_norm": 0.7526257143196984, "learning_rate": 0.003, "loss": 4.1713, "step": 4036 }, { "epoch": 0.04037, "grad_norm": 0.8211100317431537, "learning_rate": 0.003, "loss": 4.1242, "step": 4037 }, { "epoch": 0.04038, "grad_norm": 0.9809609729682468, "learning_rate": 0.003, "loss": 4.1513, "step": 4038 }, { "epoch": 0.04039, "grad_norm": 1.2927831672520989, "learning_rate": 0.003, "loss": 4.1545, "step": 4039 }, { "epoch": 0.0404, "grad_norm": 0.8695253025713267, "learning_rate": 0.003, "loss": 4.1455, "step": 4040 }, { "epoch": 0.04041, "grad_norm": 0.7950885742981303, "learning_rate": 0.003, "loss": 4.1416, "step": 4041 }, { "epoch": 0.04042, "grad_norm": 0.8571261424916939, "learning_rate": 0.003, "loss": 4.1815, "step": 4042 }, { "epoch": 0.04043, "grad_norm": 0.9275780079359417, "learning_rate": 0.003, "loss": 4.171, "step": 4043 }, { "epoch": 0.04044, "grad_norm": 0.9484978049597467, "learning_rate": 0.003, "loss": 4.1634, "step": 4044 }, { "epoch": 0.04045, "grad_norm": 0.9256569027586441, "learning_rate": 0.003, "loss": 4.1079, "step": 4045 }, { "epoch": 0.04046, "grad_norm": 0.862135206370447, "learning_rate": 0.003, "loss": 4.1554, "step": 4046 }, { "epoch": 0.04047, "grad_norm": 0.8957879714648798, "learning_rate": 0.003, "loss": 4.148, "step": 4047 }, { "epoch": 0.04048, "grad_norm": 0.915367027532899, "learning_rate": 0.003, "loss": 4.1432, "step": 4048 }, { "epoch": 0.04049, "grad_norm": 0.849479270104104, "learning_rate": 0.003, "loss": 4.1819, "step": 4049 }, { "epoch": 0.0405, "grad_norm": 0.802623437841427, "learning_rate": 0.003, "loss": 4.1744, "step": 4050 }, { "epoch": 0.04051, "grad_norm": 0.8361334119276083, "learning_rate": 0.003, "loss": 4.1544, "step": 4051 }, { "epoch": 0.04052, "grad_norm": 0.8502895680512856, "learning_rate": 0.003, "loss": 4.1628, "step": 4052 }, { "epoch": 0.04053, "grad_norm": 0.9495553848705864, "learning_rate": 0.003, "loss": 4.1153, "step": 4053 }, { "epoch": 0.04054, "grad_norm": 1.1074705319718914, "learning_rate": 0.003, "loss": 4.1421, "step": 4054 }, { "epoch": 0.04055, "grad_norm": 0.786290186382969, "learning_rate": 0.003, "loss": 4.1281, "step": 4055 }, { "epoch": 0.04056, "grad_norm": 0.6575896415910156, "learning_rate": 0.003, "loss": 4.1491, "step": 4056 }, { "epoch": 0.04057, "grad_norm": 0.7670781614470484, "learning_rate": 0.003, "loss": 4.1514, "step": 4057 }, { "epoch": 0.04058, "grad_norm": 0.833843513666775, "learning_rate": 0.003, "loss": 4.1255, "step": 4058 }, { "epoch": 0.04059, "grad_norm": 0.8114168552814227, "learning_rate": 0.003, "loss": 4.106, "step": 4059 }, { "epoch": 0.0406, "grad_norm": 0.7692714330479883, "learning_rate": 0.003, "loss": 4.1663, "step": 4060 }, { "epoch": 0.04061, "grad_norm": 0.9330465410070367, "learning_rate": 0.003, "loss": 4.1545, "step": 4061 }, { "epoch": 0.04062, "grad_norm": 0.9117817660909525, "learning_rate": 0.003, "loss": 4.1351, "step": 4062 }, { "epoch": 0.04063, "grad_norm": 0.9445008569369826, "learning_rate": 0.003, "loss": 4.1431, "step": 4063 }, { "epoch": 0.04064, "grad_norm": 1.087764691560672, "learning_rate": 0.003, "loss": 4.1619, "step": 4064 }, { "epoch": 0.04065, "grad_norm": 1.019578221772767, "learning_rate": 0.003, "loss": 4.1964, "step": 4065 }, { "epoch": 0.04066, "grad_norm": 0.8751481231375284, "learning_rate": 0.003, "loss": 4.1547, "step": 4066 }, { "epoch": 0.04067, "grad_norm": 0.8359962586744085, "learning_rate": 0.003, "loss": 4.1546, "step": 4067 }, { "epoch": 0.04068, "grad_norm": 0.8804677832654572, "learning_rate": 0.003, "loss": 4.1641, "step": 4068 }, { "epoch": 0.04069, "grad_norm": 0.8702907169533645, "learning_rate": 0.003, "loss": 4.1587, "step": 4069 }, { "epoch": 0.0407, "grad_norm": 0.8649761703309935, "learning_rate": 0.003, "loss": 4.1596, "step": 4070 }, { "epoch": 0.04071, "grad_norm": 0.8396276365063743, "learning_rate": 0.003, "loss": 4.126, "step": 4071 }, { "epoch": 0.04072, "grad_norm": 0.8818553417907129, "learning_rate": 0.003, "loss": 4.1491, "step": 4072 }, { "epoch": 0.04073, "grad_norm": 0.9077326173731153, "learning_rate": 0.003, "loss": 4.1385, "step": 4073 }, { "epoch": 0.04074, "grad_norm": 1.0214476432409059, "learning_rate": 0.003, "loss": 4.1689, "step": 4074 }, { "epoch": 0.04075, "grad_norm": 0.9920974174141465, "learning_rate": 0.003, "loss": 4.168, "step": 4075 }, { "epoch": 0.04076, "grad_norm": 0.8907761503803999, "learning_rate": 0.003, "loss": 4.1596, "step": 4076 }, { "epoch": 0.04077, "grad_norm": 0.8997616564510362, "learning_rate": 0.003, "loss": 4.1711, "step": 4077 }, { "epoch": 0.04078, "grad_norm": 0.8451943457352864, "learning_rate": 0.003, "loss": 4.1445, "step": 4078 }, { "epoch": 0.04079, "grad_norm": 0.797710300379145, "learning_rate": 0.003, "loss": 4.1525, "step": 4079 }, { "epoch": 0.0408, "grad_norm": 0.7483016121536471, "learning_rate": 0.003, "loss": 4.1443, "step": 4080 }, { "epoch": 0.04081, "grad_norm": 0.7023171834702521, "learning_rate": 0.003, "loss": 4.1223, "step": 4081 }, { "epoch": 0.04082, "grad_norm": 0.7970696730978741, "learning_rate": 0.003, "loss": 4.1597, "step": 4082 }, { "epoch": 0.04083, "grad_norm": 0.8338986570903769, "learning_rate": 0.003, "loss": 4.1382, "step": 4083 }, { "epoch": 0.04084, "grad_norm": 0.8012483337932724, "learning_rate": 0.003, "loss": 4.1743, "step": 4084 }, { "epoch": 0.04085, "grad_norm": 0.8926567018581679, "learning_rate": 0.003, "loss": 4.1539, "step": 4085 }, { "epoch": 0.04086, "grad_norm": 0.9581160933844942, "learning_rate": 0.003, "loss": 4.1789, "step": 4086 }, { "epoch": 0.04087, "grad_norm": 1.0136930052545263, "learning_rate": 0.003, "loss": 4.1619, "step": 4087 }, { "epoch": 0.04088, "grad_norm": 0.8125239420336555, "learning_rate": 0.003, "loss": 4.1368, "step": 4088 }, { "epoch": 0.04089, "grad_norm": 0.6317696576646198, "learning_rate": 0.003, "loss": 4.131, "step": 4089 }, { "epoch": 0.0409, "grad_norm": 0.7082265254467968, "learning_rate": 0.003, "loss": 4.1506, "step": 4090 }, { "epoch": 0.04091, "grad_norm": 0.7341122654124385, "learning_rate": 0.003, "loss": 4.1612, "step": 4091 }, { "epoch": 0.04092, "grad_norm": 0.8808208810851171, "learning_rate": 0.003, "loss": 4.1199, "step": 4092 }, { "epoch": 0.04093, "grad_norm": 1.0270137223618354, "learning_rate": 0.003, "loss": 4.1516, "step": 4093 }, { "epoch": 0.04094, "grad_norm": 0.9812040125714347, "learning_rate": 0.003, "loss": 4.144, "step": 4094 }, { "epoch": 0.04095, "grad_norm": 0.7496518915755831, "learning_rate": 0.003, "loss": 4.1112, "step": 4095 }, { "epoch": 0.04096, "grad_norm": 0.6920676687623359, "learning_rate": 0.003, "loss": 4.1185, "step": 4096 }, { "epoch": 0.04097, "grad_norm": 0.7659926369906308, "learning_rate": 0.003, "loss": 4.1358, "step": 4097 }, { "epoch": 0.04098, "grad_norm": 0.7973861989081666, "learning_rate": 0.003, "loss": 4.1437, "step": 4098 }, { "epoch": 0.04099, "grad_norm": 0.9266888279285345, "learning_rate": 0.003, "loss": 4.1698, "step": 4099 }, { "epoch": 0.041, "grad_norm": 0.8520041299509284, "learning_rate": 0.003, "loss": 4.1296, "step": 4100 }, { "epoch": 0.04101, "grad_norm": 0.83237018706289, "learning_rate": 0.003, "loss": 4.1245, "step": 4101 }, { "epoch": 0.04102, "grad_norm": 0.9965520477059928, "learning_rate": 0.003, "loss": 4.1505, "step": 4102 }, { "epoch": 0.04103, "grad_norm": 1.0818833033424187, "learning_rate": 0.003, "loss": 4.1836, "step": 4103 }, { "epoch": 0.04104, "grad_norm": 0.7652239209069888, "learning_rate": 0.003, "loss": 4.1545, "step": 4104 }, { "epoch": 0.04105, "grad_norm": 0.8496173505486513, "learning_rate": 0.003, "loss": 4.1264, "step": 4105 }, { "epoch": 0.04106, "grad_norm": 0.9926762870076, "learning_rate": 0.003, "loss": 4.1633, "step": 4106 }, { "epoch": 0.04107, "grad_norm": 1.1688815385892068, "learning_rate": 0.003, "loss": 4.1684, "step": 4107 }, { "epoch": 0.04108, "grad_norm": 0.9904790188927448, "learning_rate": 0.003, "loss": 4.1515, "step": 4108 }, { "epoch": 0.04109, "grad_norm": 1.0036626582575459, "learning_rate": 0.003, "loss": 4.1435, "step": 4109 }, { "epoch": 0.0411, "grad_norm": 0.9269220216164543, "learning_rate": 0.003, "loss": 4.1516, "step": 4110 }, { "epoch": 0.04111, "grad_norm": 0.9283892200368855, "learning_rate": 0.003, "loss": 4.1339, "step": 4111 }, { "epoch": 0.04112, "grad_norm": 1.1060288948934516, "learning_rate": 0.003, "loss": 4.1588, "step": 4112 }, { "epoch": 0.04113, "grad_norm": 1.0211912461653543, "learning_rate": 0.003, "loss": 4.1992, "step": 4113 }, { "epoch": 0.04114, "grad_norm": 0.917660491574899, "learning_rate": 0.003, "loss": 4.1387, "step": 4114 }, { "epoch": 0.04115, "grad_norm": 0.7897516328452319, "learning_rate": 0.003, "loss": 4.1479, "step": 4115 }, { "epoch": 0.04116, "grad_norm": 0.8456029460362186, "learning_rate": 0.003, "loss": 4.1643, "step": 4116 }, { "epoch": 0.04117, "grad_norm": 0.8944705808882064, "learning_rate": 0.003, "loss": 4.1278, "step": 4117 }, { "epoch": 0.04118, "grad_norm": 0.8832980566787416, "learning_rate": 0.003, "loss": 4.1861, "step": 4118 }, { "epoch": 0.04119, "grad_norm": 0.8067714590301097, "learning_rate": 0.003, "loss": 4.1605, "step": 4119 }, { "epoch": 0.0412, "grad_norm": 0.7737228855286454, "learning_rate": 0.003, "loss": 4.1496, "step": 4120 }, { "epoch": 0.04121, "grad_norm": 0.8389511273786252, "learning_rate": 0.003, "loss": 4.175, "step": 4121 }, { "epoch": 0.04122, "grad_norm": 0.8398694848503073, "learning_rate": 0.003, "loss": 4.1495, "step": 4122 }, { "epoch": 0.04123, "grad_norm": 0.9090933780386284, "learning_rate": 0.003, "loss": 4.1578, "step": 4123 }, { "epoch": 0.04124, "grad_norm": 0.9872711905443635, "learning_rate": 0.003, "loss": 4.1421, "step": 4124 }, { "epoch": 0.04125, "grad_norm": 1.085044262616643, "learning_rate": 0.003, "loss": 4.1685, "step": 4125 }, { "epoch": 0.04126, "grad_norm": 0.8814957273506651, "learning_rate": 0.003, "loss": 4.1556, "step": 4126 }, { "epoch": 0.04127, "grad_norm": 0.8604955271585082, "learning_rate": 0.003, "loss": 4.1452, "step": 4127 }, { "epoch": 0.04128, "grad_norm": 0.9621571347239927, "learning_rate": 0.003, "loss": 4.1288, "step": 4128 }, { "epoch": 0.04129, "grad_norm": 0.8835160766900358, "learning_rate": 0.003, "loss": 4.171, "step": 4129 }, { "epoch": 0.0413, "grad_norm": 0.8289903012402862, "learning_rate": 0.003, "loss": 4.1347, "step": 4130 }, { "epoch": 0.04131, "grad_norm": 0.8175052036927931, "learning_rate": 0.003, "loss": 4.1632, "step": 4131 }, { "epoch": 0.04132, "grad_norm": 0.8802427709327816, "learning_rate": 0.003, "loss": 4.1299, "step": 4132 }, { "epoch": 0.04133, "grad_norm": 0.9035446720390723, "learning_rate": 0.003, "loss": 4.1697, "step": 4133 }, { "epoch": 0.04134, "grad_norm": 0.7899715760468391, "learning_rate": 0.003, "loss": 4.1581, "step": 4134 }, { "epoch": 0.04135, "grad_norm": 0.7442781516869081, "learning_rate": 0.003, "loss": 4.1387, "step": 4135 }, { "epoch": 0.04136, "grad_norm": 0.7238773685007321, "learning_rate": 0.003, "loss": 4.1328, "step": 4136 }, { "epoch": 0.04137, "grad_norm": 0.7132883212877328, "learning_rate": 0.003, "loss": 4.173, "step": 4137 }, { "epoch": 0.04138, "grad_norm": 0.8062240784991308, "learning_rate": 0.003, "loss": 4.1578, "step": 4138 }, { "epoch": 0.04139, "grad_norm": 1.0359489486607962, "learning_rate": 0.003, "loss": 4.1584, "step": 4139 }, { "epoch": 0.0414, "grad_norm": 1.1920638942276114, "learning_rate": 0.003, "loss": 4.1675, "step": 4140 }, { "epoch": 0.04141, "grad_norm": 0.7605076804575376, "learning_rate": 0.003, "loss": 4.1353, "step": 4141 }, { "epoch": 0.04142, "grad_norm": 0.6390080347747942, "learning_rate": 0.003, "loss": 4.1576, "step": 4142 }, { "epoch": 0.04143, "grad_norm": 0.716027932583964, "learning_rate": 0.003, "loss": 4.145, "step": 4143 }, { "epoch": 0.04144, "grad_norm": 0.8343696675557656, "learning_rate": 0.003, "loss": 4.1459, "step": 4144 }, { "epoch": 0.04145, "grad_norm": 1.0821429598570727, "learning_rate": 0.003, "loss": 4.1347, "step": 4145 }, { "epoch": 0.04146, "grad_norm": 1.0810361946276594, "learning_rate": 0.003, "loss": 4.1595, "step": 4146 }, { "epoch": 0.04147, "grad_norm": 0.963764020070276, "learning_rate": 0.003, "loss": 4.1681, "step": 4147 }, { "epoch": 0.04148, "grad_norm": 0.8320178719420672, "learning_rate": 0.003, "loss": 4.1239, "step": 4148 }, { "epoch": 0.04149, "grad_norm": 0.7785864004070887, "learning_rate": 0.003, "loss": 4.1351, "step": 4149 }, { "epoch": 0.0415, "grad_norm": 0.8437857943811232, "learning_rate": 0.003, "loss": 4.1395, "step": 4150 }, { "epoch": 0.04151, "grad_norm": 0.9338894402887595, "learning_rate": 0.003, "loss": 4.1627, "step": 4151 }, { "epoch": 0.04152, "grad_norm": 0.9207228379208315, "learning_rate": 0.003, "loss": 4.1608, "step": 4152 }, { "epoch": 0.04153, "grad_norm": 0.8528458519081906, "learning_rate": 0.003, "loss": 4.1255, "step": 4153 }, { "epoch": 0.04154, "grad_norm": 0.7391574811644701, "learning_rate": 0.003, "loss": 4.1571, "step": 4154 }, { "epoch": 0.04155, "grad_norm": 0.75760856734852, "learning_rate": 0.003, "loss": 4.131, "step": 4155 }, { "epoch": 0.04156, "grad_norm": 0.8731952166338035, "learning_rate": 0.003, "loss": 4.1562, "step": 4156 }, { "epoch": 0.04157, "grad_norm": 0.8741618403734526, "learning_rate": 0.003, "loss": 4.1699, "step": 4157 }, { "epoch": 0.04158, "grad_norm": 0.9164866788533468, "learning_rate": 0.003, "loss": 4.1742, "step": 4158 }, { "epoch": 0.04159, "grad_norm": 0.845613595302616, "learning_rate": 0.003, "loss": 4.1426, "step": 4159 }, { "epoch": 0.0416, "grad_norm": 0.8195288287923924, "learning_rate": 0.003, "loss": 4.1518, "step": 4160 }, { "epoch": 0.04161, "grad_norm": 0.718824361447048, "learning_rate": 0.003, "loss": 4.1571, "step": 4161 }, { "epoch": 0.04162, "grad_norm": 0.7519169259607166, "learning_rate": 0.003, "loss": 4.1138, "step": 4162 }, { "epoch": 0.04163, "grad_norm": 0.7491094262311686, "learning_rate": 0.003, "loss": 4.145, "step": 4163 }, { "epoch": 0.04164, "grad_norm": 0.7993205047667109, "learning_rate": 0.003, "loss": 4.1624, "step": 4164 }, { "epoch": 0.04165, "grad_norm": 0.940572881338154, "learning_rate": 0.003, "loss": 4.1424, "step": 4165 }, { "epoch": 0.04166, "grad_norm": 1.0074718156659002, "learning_rate": 0.003, "loss": 4.1357, "step": 4166 }, { "epoch": 0.04167, "grad_norm": 0.9011941427002811, "learning_rate": 0.003, "loss": 4.1605, "step": 4167 }, { "epoch": 0.04168, "grad_norm": 0.7574197524201725, "learning_rate": 0.003, "loss": 4.1571, "step": 4168 }, { "epoch": 0.04169, "grad_norm": 0.6882002633674081, "learning_rate": 0.003, "loss": 4.1597, "step": 4169 }, { "epoch": 0.0417, "grad_norm": 0.6764254459950164, "learning_rate": 0.003, "loss": 4.1213, "step": 4170 }, { "epoch": 0.04171, "grad_norm": 0.7445116628843363, "learning_rate": 0.003, "loss": 4.1426, "step": 4171 }, { "epoch": 0.04172, "grad_norm": 0.7455792792401567, "learning_rate": 0.003, "loss": 4.1617, "step": 4172 }, { "epoch": 0.04173, "grad_norm": 0.6531022610209052, "learning_rate": 0.003, "loss": 4.1554, "step": 4173 }, { "epoch": 0.04174, "grad_norm": 0.5937692964868775, "learning_rate": 0.003, "loss": 4.1198, "step": 4174 }, { "epoch": 0.04175, "grad_norm": 0.7459694950099481, "learning_rate": 0.003, "loss": 4.172, "step": 4175 }, { "epoch": 0.04176, "grad_norm": 0.8453202156461872, "learning_rate": 0.003, "loss": 4.135, "step": 4176 }, { "epoch": 0.04177, "grad_norm": 0.9540346884999437, "learning_rate": 0.003, "loss": 4.1176, "step": 4177 }, { "epoch": 0.04178, "grad_norm": 1.04708047235552, "learning_rate": 0.003, "loss": 4.1529, "step": 4178 }, { "epoch": 0.04179, "grad_norm": 0.9852878776313095, "learning_rate": 0.003, "loss": 4.131, "step": 4179 }, { "epoch": 0.0418, "grad_norm": 0.8786478389355888, "learning_rate": 0.003, "loss": 4.1423, "step": 4180 }, { "epoch": 0.04181, "grad_norm": 0.8639731089635772, "learning_rate": 0.003, "loss": 4.1428, "step": 4181 }, { "epoch": 0.04182, "grad_norm": 0.8851770442904429, "learning_rate": 0.003, "loss": 4.1257, "step": 4182 }, { "epoch": 0.04183, "grad_norm": 0.9649954563926619, "learning_rate": 0.003, "loss": 4.137, "step": 4183 }, { "epoch": 0.04184, "grad_norm": 1.028830613529158, "learning_rate": 0.003, "loss": 4.1954, "step": 4184 }, { "epoch": 0.04185, "grad_norm": 1.1044711550852657, "learning_rate": 0.003, "loss": 4.1803, "step": 4185 }, { "epoch": 0.04186, "grad_norm": 0.7784600027105134, "learning_rate": 0.003, "loss": 4.1176, "step": 4186 }, { "epoch": 0.04187, "grad_norm": 0.9056948295325973, "learning_rate": 0.003, "loss": 4.1711, "step": 4187 }, { "epoch": 0.04188, "grad_norm": 1.1599766816413652, "learning_rate": 0.003, "loss": 4.1441, "step": 4188 }, { "epoch": 0.04189, "grad_norm": 1.1201002937284052, "learning_rate": 0.003, "loss": 4.1607, "step": 4189 }, { "epoch": 0.0419, "grad_norm": 0.7376394234035888, "learning_rate": 0.003, "loss": 4.143, "step": 4190 }, { "epoch": 0.04191, "grad_norm": 0.7634120599363466, "learning_rate": 0.003, "loss": 4.151, "step": 4191 }, { "epoch": 0.04192, "grad_norm": 0.763696759562722, "learning_rate": 0.003, "loss": 4.1308, "step": 4192 }, { "epoch": 0.04193, "grad_norm": 0.737464203974721, "learning_rate": 0.003, "loss": 4.1104, "step": 4193 }, { "epoch": 0.04194, "grad_norm": 0.8729631339185211, "learning_rate": 0.003, "loss": 4.1117, "step": 4194 }, { "epoch": 0.04195, "grad_norm": 1.0324703043706847, "learning_rate": 0.003, "loss": 4.1702, "step": 4195 }, { "epoch": 0.04196, "grad_norm": 1.1422522770241212, "learning_rate": 0.003, "loss": 4.134, "step": 4196 }, { "epoch": 0.04197, "grad_norm": 0.8189618711192417, "learning_rate": 0.003, "loss": 4.1351, "step": 4197 }, { "epoch": 0.04198, "grad_norm": 0.8230164876125401, "learning_rate": 0.003, "loss": 4.1325, "step": 4198 }, { "epoch": 0.04199, "grad_norm": 0.9071510187811559, "learning_rate": 0.003, "loss": 4.1715, "step": 4199 }, { "epoch": 0.042, "grad_norm": 0.9418882514551147, "learning_rate": 0.003, "loss": 4.1529, "step": 4200 }, { "epoch": 0.04201, "grad_norm": 1.000866294908965, "learning_rate": 0.003, "loss": 4.1309, "step": 4201 }, { "epoch": 0.04202, "grad_norm": 1.0447837092759422, "learning_rate": 0.003, "loss": 4.1533, "step": 4202 }, { "epoch": 0.04203, "grad_norm": 0.9183915096913366, "learning_rate": 0.003, "loss": 4.1691, "step": 4203 }, { "epoch": 0.04204, "grad_norm": 0.8811060274489334, "learning_rate": 0.003, "loss": 4.1356, "step": 4204 }, { "epoch": 0.04205, "grad_norm": 0.8896816083602174, "learning_rate": 0.003, "loss": 4.1641, "step": 4205 }, { "epoch": 0.04206, "grad_norm": 1.0780443908354043, "learning_rate": 0.003, "loss": 4.1613, "step": 4206 }, { "epoch": 0.04207, "grad_norm": 1.2797808490177898, "learning_rate": 0.003, "loss": 4.1573, "step": 4207 }, { "epoch": 0.04208, "grad_norm": 0.9179007982217813, "learning_rate": 0.003, "loss": 4.1587, "step": 4208 }, { "epoch": 0.04209, "grad_norm": 0.8977511024729588, "learning_rate": 0.003, "loss": 4.1592, "step": 4209 }, { "epoch": 0.0421, "grad_norm": 0.9454990443320976, "learning_rate": 0.003, "loss": 4.1243, "step": 4210 }, { "epoch": 0.04211, "grad_norm": 1.0959447073496884, "learning_rate": 0.003, "loss": 4.1634, "step": 4211 }, { "epoch": 0.04212, "grad_norm": 0.9065788803964404, "learning_rate": 0.003, "loss": 4.1506, "step": 4212 }, { "epoch": 0.04213, "grad_norm": 0.8052786745518665, "learning_rate": 0.003, "loss": 4.1284, "step": 4213 }, { "epoch": 0.04214, "grad_norm": 0.8570874487237381, "learning_rate": 0.003, "loss": 4.1372, "step": 4214 }, { "epoch": 0.04215, "grad_norm": 0.8380491615517233, "learning_rate": 0.003, "loss": 4.1566, "step": 4215 }, { "epoch": 0.04216, "grad_norm": 0.8424695888257845, "learning_rate": 0.003, "loss": 4.1206, "step": 4216 }, { "epoch": 0.04217, "grad_norm": 0.7943923888745155, "learning_rate": 0.003, "loss": 4.1403, "step": 4217 }, { "epoch": 0.04218, "grad_norm": 0.7677254830488702, "learning_rate": 0.003, "loss": 4.17, "step": 4218 }, { "epoch": 0.04219, "grad_norm": 0.6901018800578395, "learning_rate": 0.003, "loss": 4.0884, "step": 4219 }, { "epoch": 0.0422, "grad_norm": 0.6822117165527345, "learning_rate": 0.003, "loss": 4.1548, "step": 4220 }, { "epoch": 0.04221, "grad_norm": 0.6908541457405948, "learning_rate": 0.003, "loss": 4.1271, "step": 4221 }, { "epoch": 0.04222, "grad_norm": 0.6494859209716153, "learning_rate": 0.003, "loss": 4.0935, "step": 4222 }, { "epoch": 0.04223, "grad_norm": 0.6502340833989994, "learning_rate": 0.003, "loss": 4.1567, "step": 4223 }, { "epoch": 0.04224, "grad_norm": 0.6452092264025292, "learning_rate": 0.003, "loss": 4.1601, "step": 4224 }, { "epoch": 0.04225, "grad_norm": 0.6665791958458259, "learning_rate": 0.003, "loss": 4.1461, "step": 4225 }, { "epoch": 0.04226, "grad_norm": 0.6985305884300859, "learning_rate": 0.003, "loss": 4.131, "step": 4226 }, { "epoch": 0.04227, "grad_norm": 0.7856841999519473, "learning_rate": 0.003, "loss": 4.1329, "step": 4227 }, { "epoch": 0.04228, "grad_norm": 0.8773648423553363, "learning_rate": 0.003, "loss": 4.1193, "step": 4228 }, { "epoch": 0.04229, "grad_norm": 1.1631329607599121, "learning_rate": 0.003, "loss": 4.1525, "step": 4229 }, { "epoch": 0.0423, "grad_norm": 0.8856168266507708, "learning_rate": 0.003, "loss": 4.1368, "step": 4230 }, { "epoch": 0.04231, "grad_norm": 0.7750650850872894, "learning_rate": 0.003, "loss": 4.131, "step": 4231 }, { "epoch": 0.04232, "grad_norm": 0.7131605323369269, "learning_rate": 0.003, "loss": 4.1437, "step": 4232 }, { "epoch": 0.04233, "grad_norm": 0.6623267284695459, "learning_rate": 0.003, "loss": 4.0953, "step": 4233 }, { "epoch": 0.04234, "grad_norm": 0.7686135410977546, "learning_rate": 0.003, "loss": 4.1298, "step": 4234 }, { "epoch": 0.04235, "grad_norm": 0.8548823502536986, "learning_rate": 0.003, "loss": 4.1153, "step": 4235 }, { "epoch": 0.04236, "grad_norm": 0.7555623186654448, "learning_rate": 0.003, "loss": 4.1245, "step": 4236 }, { "epoch": 0.04237, "grad_norm": 0.7193078765588848, "learning_rate": 0.003, "loss": 4.0971, "step": 4237 }, { "epoch": 0.04238, "grad_norm": 0.7633904865120098, "learning_rate": 0.003, "loss": 4.1563, "step": 4238 }, { "epoch": 0.04239, "grad_norm": 0.7393743565898704, "learning_rate": 0.003, "loss": 4.1408, "step": 4239 }, { "epoch": 0.0424, "grad_norm": 0.6305401293342845, "learning_rate": 0.003, "loss": 4.1373, "step": 4240 }, { "epoch": 0.04241, "grad_norm": 0.6754271205716451, "learning_rate": 0.003, "loss": 4.1524, "step": 4241 }, { "epoch": 0.04242, "grad_norm": 0.5501650821009278, "learning_rate": 0.003, "loss": 4.1355, "step": 4242 }, { "epoch": 0.04243, "grad_norm": 0.5903485874209297, "learning_rate": 0.003, "loss": 4.1191, "step": 4243 }, { "epoch": 0.04244, "grad_norm": 0.5264249302326871, "learning_rate": 0.003, "loss": 4.1239, "step": 4244 }, { "epoch": 0.04245, "grad_norm": 0.6494172750972387, "learning_rate": 0.003, "loss": 4.1209, "step": 4245 }, { "epoch": 0.04246, "grad_norm": 1.1075046969225877, "learning_rate": 0.003, "loss": 4.1558, "step": 4246 }, { "epoch": 0.04247, "grad_norm": 1.4603989952406695, "learning_rate": 0.003, "loss": 4.1454, "step": 4247 }, { "epoch": 0.04248, "grad_norm": 0.6973373631740518, "learning_rate": 0.003, "loss": 4.1253, "step": 4248 }, { "epoch": 0.04249, "grad_norm": 0.8960151802399898, "learning_rate": 0.003, "loss": 4.1257, "step": 4249 }, { "epoch": 0.0425, "grad_norm": 1.0148476039293284, "learning_rate": 0.003, "loss": 4.1494, "step": 4250 }, { "epoch": 0.04251, "grad_norm": 1.091029088953841, "learning_rate": 0.003, "loss": 4.1103, "step": 4251 }, { "epoch": 0.04252, "grad_norm": 0.9828804779949347, "learning_rate": 0.003, "loss": 4.1604, "step": 4252 }, { "epoch": 0.04253, "grad_norm": 1.0407830888915848, "learning_rate": 0.003, "loss": 4.1223, "step": 4253 }, { "epoch": 0.04254, "grad_norm": 0.9969078243511121, "learning_rate": 0.003, "loss": 4.1498, "step": 4254 }, { "epoch": 0.04255, "grad_norm": 0.8726287911896201, "learning_rate": 0.003, "loss": 4.1246, "step": 4255 }, { "epoch": 0.04256, "grad_norm": 0.9734032369713972, "learning_rate": 0.003, "loss": 4.1419, "step": 4256 }, { "epoch": 0.04257, "grad_norm": 0.9839814555396913, "learning_rate": 0.003, "loss": 4.166, "step": 4257 }, { "epoch": 0.04258, "grad_norm": 0.9700101811814381, "learning_rate": 0.003, "loss": 4.1643, "step": 4258 }, { "epoch": 0.04259, "grad_norm": 1.1106152004837204, "learning_rate": 0.003, "loss": 4.1361, "step": 4259 }, { "epoch": 0.0426, "grad_norm": 1.0102602526109068, "learning_rate": 0.003, "loss": 4.1544, "step": 4260 }, { "epoch": 0.04261, "grad_norm": 1.0306790729362876, "learning_rate": 0.003, "loss": 4.1637, "step": 4261 }, { "epoch": 0.04262, "grad_norm": 1.2862557928903777, "learning_rate": 0.003, "loss": 4.1909, "step": 4262 }, { "epoch": 0.04263, "grad_norm": 1.0325757049146929, "learning_rate": 0.003, "loss": 4.1721, "step": 4263 }, { "epoch": 0.04264, "grad_norm": 1.0162869618759527, "learning_rate": 0.003, "loss": 4.1764, "step": 4264 }, { "epoch": 0.04265, "grad_norm": 1.0950967203268815, "learning_rate": 0.003, "loss": 4.1786, "step": 4265 }, { "epoch": 0.04266, "grad_norm": 0.948160311309965, "learning_rate": 0.003, "loss": 4.1532, "step": 4266 }, { "epoch": 0.04267, "grad_norm": 1.2348979147487773, "learning_rate": 0.003, "loss": 4.1375, "step": 4267 }, { "epoch": 0.04268, "grad_norm": 0.8778615179492638, "learning_rate": 0.003, "loss": 4.1726, "step": 4268 }, { "epoch": 0.04269, "grad_norm": 0.8286683782955315, "learning_rate": 0.003, "loss": 4.1311, "step": 4269 }, { "epoch": 0.0427, "grad_norm": 0.9176546079150403, "learning_rate": 0.003, "loss": 4.1546, "step": 4270 }, { "epoch": 0.04271, "grad_norm": 0.8811867110687683, "learning_rate": 0.003, "loss": 4.1692, "step": 4271 }, { "epoch": 0.04272, "grad_norm": 0.8863150873209689, "learning_rate": 0.003, "loss": 4.1537, "step": 4272 }, { "epoch": 0.04273, "grad_norm": 0.8585270675572245, "learning_rate": 0.003, "loss": 4.1338, "step": 4273 }, { "epoch": 0.04274, "grad_norm": 0.7731759429393139, "learning_rate": 0.003, "loss": 4.1545, "step": 4274 }, { "epoch": 0.04275, "grad_norm": 0.6354552017070271, "learning_rate": 0.003, "loss": 4.1528, "step": 4275 }, { "epoch": 0.04276, "grad_norm": 0.7646129405940127, "learning_rate": 0.003, "loss": 4.1723, "step": 4276 }, { "epoch": 0.04277, "grad_norm": 0.8927972665604179, "learning_rate": 0.003, "loss": 4.16, "step": 4277 }, { "epoch": 0.04278, "grad_norm": 1.0322138758213815, "learning_rate": 0.003, "loss": 4.1677, "step": 4278 }, { "epoch": 0.04279, "grad_norm": 0.960550547346212, "learning_rate": 0.003, "loss": 4.1397, "step": 4279 }, { "epoch": 0.0428, "grad_norm": 1.0071112981046495, "learning_rate": 0.003, "loss": 4.1241, "step": 4280 }, { "epoch": 0.04281, "grad_norm": 0.8672605322618707, "learning_rate": 0.003, "loss": 4.1751, "step": 4281 }, { "epoch": 0.04282, "grad_norm": 0.836732025387026, "learning_rate": 0.003, "loss": 4.1499, "step": 4282 }, { "epoch": 0.04283, "grad_norm": 0.8927876116515261, "learning_rate": 0.003, "loss": 4.1845, "step": 4283 }, { "epoch": 0.04284, "grad_norm": 0.8645218811194867, "learning_rate": 0.003, "loss": 4.1357, "step": 4284 }, { "epoch": 0.04285, "grad_norm": 0.842404435262628, "learning_rate": 0.003, "loss": 4.1507, "step": 4285 }, { "epoch": 0.04286, "grad_norm": 0.8812355936864655, "learning_rate": 0.003, "loss": 4.1695, "step": 4286 }, { "epoch": 0.04287, "grad_norm": 0.8580448230390728, "learning_rate": 0.003, "loss": 4.1516, "step": 4287 }, { "epoch": 0.04288, "grad_norm": 0.797143513389131, "learning_rate": 0.003, "loss": 4.1264, "step": 4288 }, { "epoch": 0.04289, "grad_norm": 0.772184705171998, "learning_rate": 0.003, "loss": 4.1343, "step": 4289 }, { "epoch": 0.0429, "grad_norm": 0.7832547421398193, "learning_rate": 0.003, "loss": 4.1196, "step": 4290 }, { "epoch": 0.04291, "grad_norm": 0.9556558247199429, "learning_rate": 0.003, "loss": 4.1288, "step": 4291 }, { "epoch": 0.04292, "grad_norm": 1.1338816182381097, "learning_rate": 0.003, "loss": 4.1412, "step": 4292 }, { "epoch": 0.04293, "grad_norm": 0.8021810217572, "learning_rate": 0.003, "loss": 4.1679, "step": 4293 }, { "epoch": 0.04294, "grad_norm": 0.6541787631431899, "learning_rate": 0.003, "loss": 4.1425, "step": 4294 }, { "epoch": 0.04295, "grad_norm": 0.7140599630716042, "learning_rate": 0.003, "loss": 4.1176, "step": 4295 }, { "epoch": 0.04296, "grad_norm": 0.7055439180689369, "learning_rate": 0.003, "loss": 4.1132, "step": 4296 }, { "epoch": 0.04297, "grad_norm": 0.7279360773174349, "learning_rate": 0.003, "loss": 4.1022, "step": 4297 }, { "epoch": 0.04298, "grad_norm": 0.8602725232980242, "learning_rate": 0.003, "loss": 4.1487, "step": 4298 }, { "epoch": 0.04299, "grad_norm": 1.1960364199120477, "learning_rate": 0.003, "loss": 4.1483, "step": 4299 }, { "epoch": 0.043, "grad_norm": 0.887735194757027, "learning_rate": 0.003, "loss": 4.1177, "step": 4300 }, { "epoch": 0.04301, "grad_norm": 0.8671235111098321, "learning_rate": 0.003, "loss": 4.1691, "step": 4301 }, { "epoch": 0.04302, "grad_norm": 0.8219777153602685, "learning_rate": 0.003, "loss": 4.1561, "step": 4302 }, { "epoch": 0.04303, "grad_norm": 0.7365454368217494, "learning_rate": 0.003, "loss": 4.1434, "step": 4303 }, { "epoch": 0.04304, "grad_norm": 0.7332880503208107, "learning_rate": 0.003, "loss": 4.1391, "step": 4304 }, { "epoch": 0.04305, "grad_norm": 0.6049518431707501, "learning_rate": 0.003, "loss": 4.1349, "step": 4305 }, { "epoch": 0.04306, "grad_norm": 0.552458178896163, "learning_rate": 0.003, "loss": 4.108, "step": 4306 }, { "epoch": 0.04307, "grad_norm": 0.6072550137534989, "learning_rate": 0.003, "loss": 4.1158, "step": 4307 }, { "epoch": 0.04308, "grad_norm": 0.8491425972676354, "learning_rate": 0.003, "loss": 4.1348, "step": 4308 }, { "epoch": 0.04309, "grad_norm": 1.1697300613438677, "learning_rate": 0.003, "loss": 4.1303, "step": 4309 }, { "epoch": 0.0431, "grad_norm": 0.9663103268632941, "learning_rate": 0.003, "loss": 4.1521, "step": 4310 }, { "epoch": 0.04311, "grad_norm": 0.8246252258530357, "learning_rate": 0.003, "loss": 4.1463, "step": 4311 }, { "epoch": 0.04312, "grad_norm": 0.8926510401981749, "learning_rate": 0.003, "loss": 4.1263, "step": 4312 }, { "epoch": 0.04313, "grad_norm": 0.9825409959593736, "learning_rate": 0.003, "loss": 4.1442, "step": 4313 }, { "epoch": 0.04314, "grad_norm": 1.135321041866459, "learning_rate": 0.003, "loss": 4.1396, "step": 4314 }, { "epoch": 0.04315, "grad_norm": 1.1351817598396858, "learning_rate": 0.003, "loss": 4.1674, "step": 4315 }, { "epoch": 0.04316, "grad_norm": 0.8853146589527483, "learning_rate": 0.003, "loss": 4.156, "step": 4316 }, { "epoch": 0.04317, "grad_norm": 0.8300765192397569, "learning_rate": 0.003, "loss": 4.1673, "step": 4317 }, { "epoch": 0.04318, "grad_norm": 0.8608939310347146, "learning_rate": 0.003, "loss": 4.1099, "step": 4318 }, { "epoch": 0.04319, "grad_norm": 0.846452814864276, "learning_rate": 0.003, "loss": 4.1072, "step": 4319 }, { "epoch": 0.0432, "grad_norm": 0.8565259497390109, "learning_rate": 0.003, "loss": 4.13, "step": 4320 }, { "epoch": 0.04321, "grad_norm": 1.003163747360344, "learning_rate": 0.003, "loss": 4.1365, "step": 4321 }, { "epoch": 0.04322, "grad_norm": 1.3687808098292402, "learning_rate": 0.003, "loss": 4.1846, "step": 4322 }, { "epoch": 0.04323, "grad_norm": 0.7765534239012071, "learning_rate": 0.003, "loss": 4.1396, "step": 4323 }, { "epoch": 0.04324, "grad_norm": 0.7409828216455313, "learning_rate": 0.003, "loss": 4.1331, "step": 4324 }, { "epoch": 0.04325, "grad_norm": 0.6593731828315239, "learning_rate": 0.003, "loss": 4.1592, "step": 4325 }, { "epoch": 0.04326, "grad_norm": 0.7626886495825427, "learning_rate": 0.003, "loss": 4.1383, "step": 4326 }, { "epoch": 0.04327, "grad_norm": 0.9753282134260721, "learning_rate": 0.003, "loss": 4.1256, "step": 4327 }, { "epoch": 0.04328, "grad_norm": 1.1122267963910306, "learning_rate": 0.003, "loss": 4.1474, "step": 4328 }, { "epoch": 0.04329, "grad_norm": 0.8812692829330543, "learning_rate": 0.003, "loss": 4.1609, "step": 4329 }, { "epoch": 0.0433, "grad_norm": 0.9117992082897464, "learning_rate": 0.003, "loss": 4.1343, "step": 4330 }, { "epoch": 0.04331, "grad_norm": 1.008387824573847, "learning_rate": 0.003, "loss": 4.132, "step": 4331 }, { "epoch": 0.04332, "grad_norm": 0.9701656948256054, "learning_rate": 0.003, "loss": 4.1817, "step": 4332 }, { "epoch": 0.04333, "grad_norm": 0.7550874154301345, "learning_rate": 0.003, "loss": 4.1464, "step": 4333 }, { "epoch": 0.04334, "grad_norm": 0.9440172266611684, "learning_rate": 0.003, "loss": 4.1374, "step": 4334 }, { "epoch": 0.04335, "grad_norm": 1.0714395710867621, "learning_rate": 0.003, "loss": 4.1307, "step": 4335 }, { "epoch": 0.04336, "grad_norm": 1.274134153372981, "learning_rate": 0.003, "loss": 4.1731, "step": 4336 }, { "epoch": 0.04337, "grad_norm": 0.919379094932585, "learning_rate": 0.003, "loss": 4.1531, "step": 4337 }, { "epoch": 0.04338, "grad_norm": 0.9113554686958292, "learning_rate": 0.003, "loss": 4.1517, "step": 4338 }, { "epoch": 0.04339, "grad_norm": 1.0175097294988102, "learning_rate": 0.003, "loss": 4.1611, "step": 4339 }, { "epoch": 0.0434, "grad_norm": 1.1823172721493573, "learning_rate": 0.003, "loss": 4.1439, "step": 4340 }, { "epoch": 0.04341, "grad_norm": 1.0738265997223806, "learning_rate": 0.003, "loss": 4.1578, "step": 4341 }, { "epoch": 0.04342, "grad_norm": 0.9859857822713426, "learning_rate": 0.003, "loss": 4.1403, "step": 4342 }, { "epoch": 0.04343, "grad_norm": 1.0178429154018418, "learning_rate": 0.003, "loss": 4.1818, "step": 4343 }, { "epoch": 0.04344, "grad_norm": 0.8898698596336355, "learning_rate": 0.003, "loss": 4.1228, "step": 4344 }, { "epoch": 0.04345, "grad_norm": 0.8224305636273126, "learning_rate": 0.003, "loss": 4.1577, "step": 4345 }, { "epoch": 0.04346, "grad_norm": 0.8036467865466507, "learning_rate": 0.003, "loss": 4.1525, "step": 4346 }, { "epoch": 0.04347, "grad_norm": 0.8647944237371374, "learning_rate": 0.003, "loss": 4.1577, "step": 4347 }, { "epoch": 0.04348, "grad_norm": 0.8408116621591446, "learning_rate": 0.003, "loss": 4.1392, "step": 4348 }, { "epoch": 0.04349, "grad_norm": 0.7541460584282086, "learning_rate": 0.003, "loss": 4.1498, "step": 4349 }, { "epoch": 0.0435, "grad_norm": 0.7511170217669931, "learning_rate": 0.003, "loss": 4.1358, "step": 4350 }, { "epoch": 0.04351, "grad_norm": 0.7203893697283558, "learning_rate": 0.003, "loss": 4.1468, "step": 4351 }, { "epoch": 0.04352, "grad_norm": 0.889935460967273, "learning_rate": 0.003, "loss": 4.1657, "step": 4352 }, { "epoch": 0.04353, "grad_norm": 1.0367752140513655, "learning_rate": 0.003, "loss": 4.1398, "step": 4353 }, { "epoch": 0.04354, "grad_norm": 1.0594985151220453, "learning_rate": 0.003, "loss": 4.1457, "step": 4354 }, { "epoch": 0.04355, "grad_norm": 1.0642370180766931, "learning_rate": 0.003, "loss": 4.157, "step": 4355 }, { "epoch": 0.04356, "grad_norm": 0.7520263014358131, "learning_rate": 0.003, "loss": 4.1511, "step": 4356 }, { "epoch": 0.04357, "grad_norm": 0.5909325929201686, "learning_rate": 0.003, "loss": 4.1345, "step": 4357 }, { "epoch": 0.04358, "grad_norm": 0.6503221006497333, "learning_rate": 0.003, "loss": 4.1232, "step": 4358 }, { "epoch": 0.04359, "grad_norm": 0.6977281953888783, "learning_rate": 0.003, "loss": 4.1442, "step": 4359 }, { "epoch": 0.0436, "grad_norm": 0.705707170893373, "learning_rate": 0.003, "loss": 4.1459, "step": 4360 }, { "epoch": 0.04361, "grad_norm": 0.6978704097912947, "learning_rate": 0.003, "loss": 4.111, "step": 4361 }, { "epoch": 0.04362, "grad_norm": 0.6552490872419106, "learning_rate": 0.003, "loss": 4.1192, "step": 4362 }, { "epoch": 0.04363, "grad_norm": 0.6258270583226841, "learning_rate": 0.003, "loss": 4.1245, "step": 4363 }, { "epoch": 0.04364, "grad_norm": 0.6273578442057955, "learning_rate": 0.003, "loss": 4.1372, "step": 4364 }, { "epoch": 0.04365, "grad_norm": 0.692759351590066, "learning_rate": 0.003, "loss": 4.1384, "step": 4365 }, { "epoch": 0.04366, "grad_norm": 0.7514191452479654, "learning_rate": 0.003, "loss": 4.0943, "step": 4366 }, { "epoch": 0.04367, "grad_norm": 0.8818255856194102, "learning_rate": 0.003, "loss": 4.1476, "step": 4367 }, { "epoch": 0.04368, "grad_norm": 1.262257310600421, "learning_rate": 0.003, "loss": 4.1443, "step": 4368 }, { "epoch": 0.04369, "grad_norm": 1.0072650700377674, "learning_rate": 0.003, "loss": 4.1343, "step": 4369 }, { "epoch": 0.0437, "grad_norm": 0.8796323488234715, "learning_rate": 0.003, "loss": 4.158, "step": 4370 }, { "epoch": 0.04371, "grad_norm": 0.8385748458074573, "learning_rate": 0.003, "loss": 4.1428, "step": 4371 }, { "epoch": 0.04372, "grad_norm": 0.9689208942990393, "learning_rate": 0.003, "loss": 4.1294, "step": 4372 }, { "epoch": 0.04373, "grad_norm": 1.0255215476071573, "learning_rate": 0.003, "loss": 4.1558, "step": 4373 }, { "epoch": 0.04374, "grad_norm": 1.0377452403485916, "learning_rate": 0.003, "loss": 4.1966, "step": 4374 }, { "epoch": 0.04375, "grad_norm": 1.225365163750092, "learning_rate": 0.003, "loss": 4.1152, "step": 4375 }, { "epoch": 0.04376, "grad_norm": 0.9746772561653527, "learning_rate": 0.003, "loss": 4.1562, "step": 4376 }, { "epoch": 0.04377, "grad_norm": 0.9884150415730085, "learning_rate": 0.003, "loss": 4.1817, "step": 4377 }, { "epoch": 0.04378, "grad_norm": 1.0487083703280902, "learning_rate": 0.003, "loss": 4.1728, "step": 4378 }, { "epoch": 0.04379, "grad_norm": 1.1241089785074756, "learning_rate": 0.003, "loss": 4.186, "step": 4379 }, { "epoch": 0.0438, "grad_norm": 1.0220079899190067, "learning_rate": 0.003, "loss": 4.1556, "step": 4380 }, { "epoch": 0.04381, "grad_norm": 0.9536314073088888, "learning_rate": 0.003, "loss": 4.143, "step": 4381 }, { "epoch": 0.04382, "grad_norm": 0.9677782727134393, "learning_rate": 0.003, "loss": 4.1552, "step": 4382 }, { "epoch": 0.04383, "grad_norm": 0.9472672654930465, "learning_rate": 0.003, "loss": 4.1335, "step": 4383 }, { "epoch": 0.04384, "grad_norm": 0.8670033507963211, "learning_rate": 0.003, "loss": 4.1382, "step": 4384 }, { "epoch": 0.04385, "grad_norm": 0.7964325834207587, "learning_rate": 0.003, "loss": 4.1743, "step": 4385 }, { "epoch": 0.04386, "grad_norm": 0.841894742549459, "learning_rate": 0.003, "loss": 4.1488, "step": 4386 }, { "epoch": 0.04387, "grad_norm": 0.7784847249978989, "learning_rate": 0.003, "loss": 4.1715, "step": 4387 }, { "epoch": 0.04388, "grad_norm": 0.8761401164059524, "learning_rate": 0.003, "loss": 4.1342, "step": 4388 }, { "epoch": 0.04389, "grad_norm": 1.0513363320918883, "learning_rate": 0.003, "loss": 4.1696, "step": 4389 }, { "epoch": 0.0439, "grad_norm": 1.018366577116525, "learning_rate": 0.003, "loss": 4.1665, "step": 4390 }, { "epoch": 0.04391, "grad_norm": 0.8408058225537212, "learning_rate": 0.003, "loss": 4.1325, "step": 4391 }, { "epoch": 0.04392, "grad_norm": 0.7944853381325269, "learning_rate": 0.003, "loss": 4.1392, "step": 4392 }, { "epoch": 0.04393, "grad_norm": 0.7864110377608735, "learning_rate": 0.003, "loss": 4.1285, "step": 4393 }, { "epoch": 0.04394, "grad_norm": 0.8003481063323544, "learning_rate": 0.003, "loss": 4.1474, "step": 4394 }, { "epoch": 0.04395, "grad_norm": 0.82434538845085, "learning_rate": 0.003, "loss": 4.1295, "step": 4395 }, { "epoch": 0.04396, "grad_norm": 0.8782690940643569, "learning_rate": 0.003, "loss": 4.1057, "step": 4396 }, { "epoch": 0.04397, "grad_norm": 0.9209528050546905, "learning_rate": 0.003, "loss": 4.1219, "step": 4397 }, { "epoch": 0.04398, "grad_norm": 0.9867795261868282, "learning_rate": 0.003, "loss": 4.1415, "step": 4398 }, { "epoch": 0.04399, "grad_norm": 1.001127729963885, "learning_rate": 0.003, "loss": 4.1499, "step": 4399 }, { "epoch": 0.044, "grad_norm": 0.9840084183939074, "learning_rate": 0.003, "loss": 4.1456, "step": 4400 }, { "epoch": 0.04401, "grad_norm": 0.9259452446977503, "learning_rate": 0.003, "loss": 4.1587, "step": 4401 }, { "epoch": 0.04402, "grad_norm": 0.941717579650296, "learning_rate": 0.003, "loss": 4.1617, "step": 4402 }, { "epoch": 0.04403, "grad_norm": 0.9973820182998915, "learning_rate": 0.003, "loss": 4.1314, "step": 4403 }, { "epoch": 0.04404, "grad_norm": 0.9391290732287068, "learning_rate": 0.003, "loss": 4.1624, "step": 4404 }, { "epoch": 0.04405, "grad_norm": 0.9771616987677118, "learning_rate": 0.003, "loss": 4.1538, "step": 4405 }, { "epoch": 0.04406, "grad_norm": 0.8645709876169484, "learning_rate": 0.003, "loss": 4.1525, "step": 4406 }, { "epoch": 0.04407, "grad_norm": 0.8057298668086074, "learning_rate": 0.003, "loss": 4.1202, "step": 4407 }, { "epoch": 0.04408, "grad_norm": 0.8826642860291924, "learning_rate": 0.003, "loss": 4.128, "step": 4408 }, { "epoch": 0.04409, "grad_norm": 0.942883777044805, "learning_rate": 0.003, "loss": 4.1621, "step": 4409 }, { "epoch": 0.0441, "grad_norm": 1.1342610134630753, "learning_rate": 0.003, "loss": 4.1509, "step": 4410 }, { "epoch": 0.04411, "grad_norm": 0.8444711866462494, "learning_rate": 0.003, "loss": 4.1275, "step": 4411 }, { "epoch": 0.04412, "grad_norm": 0.618795096815875, "learning_rate": 0.003, "loss": 4.1112, "step": 4412 }, { "epoch": 0.04413, "grad_norm": 0.604846824661422, "learning_rate": 0.003, "loss": 4.1201, "step": 4413 }, { "epoch": 0.04414, "grad_norm": 0.784437278432988, "learning_rate": 0.003, "loss": 4.1719, "step": 4414 }, { "epoch": 0.04415, "grad_norm": 1.1541784220535383, "learning_rate": 0.003, "loss": 4.1437, "step": 4415 }, { "epoch": 0.04416, "grad_norm": 0.9891248909004052, "learning_rate": 0.003, "loss": 4.1294, "step": 4416 }, { "epoch": 0.04417, "grad_norm": 0.9814607771592448, "learning_rate": 0.003, "loss": 4.1059, "step": 4417 }, { "epoch": 0.04418, "grad_norm": 1.0297483619337797, "learning_rate": 0.003, "loss": 4.1543, "step": 4418 }, { "epoch": 0.04419, "grad_norm": 0.9547057067964952, "learning_rate": 0.003, "loss": 4.1695, "step": 4419 }, { "epoch": 0.0442, "grad_norm": 0.8002338037740572, "learning_rate": 0.003, "loss": 4.1367, "step": 4420 }, { "epoch": 0.04421, "grad_norm": 0.8265892990962934, "learning_rate": 0.003, "loss": 4.1666, "step": 4421 }, { "epoch": 0.04422, "grad_norm": 1.01186758550101, "learning_rate": 0.003, "loss": 4.1412, "step": 4422 }, { "epoch": 0.04423, "grad_norm": 0.8991548755615442, "learning_rate": 0.003, "loss": 4.1558, "step": 4423 }, { "epoch": 0.04424, "grad_norm": 0.9249376765836381, "learning_rate": 0.003, "loss": 4.1084, "step": 4424 }, { "epoch": 0.04425, "grad_norm": 0.787895727766037, "learning_rate": 0.003, "loss": 4.1287, "step": 4425 }, { "epoch": 0.04426, "grad_norm": 0.8384547452759158, "learning_rate": 0.003, "loss": 4.1518, "step": 4426 }, { "epoch": 0.04427, "grad_norm": 0.8084673109174956, "learning_rate": 0.003, "loss": 4.1382, "step": 4427 }, { "epoch": 0.04428, "grad_norm": 0.7355210397111905, "learning_rate": 0.003, "loss": 4.131, "step": 4428 }, { "epoch": 0.04429, "grad_norm": 0.6972354751664229, "learning_rate": 0.003, "loss": 4.174, "step": 4429 }, { "epoch": 0.0443, "grad_norm": 0.7045383271873761, "learning_rate": 0.003, "loss": 4.1284, "step": 4430 }, { "epoch": 0.04431, "grad_norm": 0.7845800246059359, "learning_rate": 0.003, "loss": 4.1559, "step": 4431 }, { "epoch": 0.04432, "grad_norm": 0.9145736262867346, "learning_rate": 0.003, "loss": 4.1612, "step": 4432 }, { "epoch": 0.04433, "grad_norm": 0.9779689050072177, "learning_rate": 0.003, "loss": 4.1376, "step": 4433 }, { "epoch": 0.04434, "grad_norm": 1.0759117748955118, "learning_rate": 0.003, "loss": 4.1382, "step": 4434 }, { "epoch": 0.04435, "grad_norm": 1.0791198295017557, "learning_rate": 0.003, "loss": 4.1576, "step": 4435 }, { "epoch": 0.04436, "grad_norm": 0.9820518152505495, "learning_rate": 0.003, "loss": 4.1266, "step": 4436 }, { "epoch": 0.04437, "grad_norm": 0.8709315291003812, "learning_rate": 0.003, "loss": 4.1495, "step": 4437 }, { "epoch": 0.04438, "grad_norm": 0.7342179784041142, "learning_rate": 0.003, "loss": 4.1824, "step": 4438 }, { "epoch": 0.04439, "grad_norm": 0.8877734923678575, "learning_rate": 0.003, "loss": 4.155, "step": 4439 }, { "epoch": 0.0444, "grad_norm": 0.82101397614282, "learning_rate": 0.003, "loss": 4.1083, "step": 4440 }, { "epoch": 0.04441, "grad_norm": 0.8094046311622209, "learning_rate": 0.003, "loss": 4.1391, "step": 4441 }, { "epoch": 0.04442, "grad_norm": 0.8261845339551874, "learning_rate": 0.003, "loss": 4.133, "step": 4442 }, { "epoch": 0.04443, "grad_norm": 0.8838895022282738, "learning_rate": 0.003, "loss": 4.1118, "step": 4443 }, { "epoch": 0.04444, "grad_norm": 0.8720075839018505, "learning_rate": 0.003, "loss": 4.113, "step": 4444 }, { "epoch": 0.04445, "grad_norm": 0.9491882233225604, "learning_rate": 0.003, "loss": 4.1432, "step": 4445 }, { "epoch": 0.04446, "grad_norm": 0.9902050613570654, "learning_rate": 0.003, "loss": 4.1148, "step": 4446 }, { "epoch": 0.04447, "grad_norm": 0.9639238211565191, "learning_rate": 0.003, "loss": 4.1205, "step": 4447 }, { "epoch": 0.04448, "grad_norm": 1.0022912532571144, "learning_rate": 0.003, "loss": 4.1226, "step": 4448 }, { "epoch": 0.04449, "grad_norm": 1.0491136390522242, "learning_rate": 0.003, "loss": 4.1431, "step": 4449 }, { "epoch": 0.0445, "grad_norm": 1.0713493139767634, "learning_rate": 0.003, "loss": 4.1488, "step": 4450 }, { "epoch": 0.04451, "grad_norm": 0.9194026244209427, "learning_rate": 0.003, "loss": 4.1438, "step": 4451 }, { "epoch": 0.04452, "grad_norm": 0.9298997780980283, "learning_rate": 0.003, "loss": 4.1574, "step": 4452 }, { "epoch": 0.04453, "grad_norm": 0.8266906084450703, "learning_rate": 0.003, "loss": 4.167, "step": 4453 }, { "epoch": 0.04454, "grad_norm": 0.8781132654558221, "learning_rate": 0.003, "loss": 4.1456, "step": 4454 }, { "epoch": 0.04455, "grad_norm": 0.949277954775384, "learning_rate": 0.003, "loss": 4.1619, "step": 4455 }, { "epoch": 0.04456, "grad_norm": 0.9259049491169088, "learning_rate": 0.003, "loss": 4.1255, "step": 4456 }, { "epoch": 0.04457, "grad_norm": 1.1275778174317006, "learning_rate": 0.003, "loss": 4.1488, "step": 4457 }, { "epoch": 0.04458, "grad_norm": 0.9410680370701086, "learning_rate": 0.003, "loss": 4.1646, "step": 4458 }, { "epoch": 0.04459, "grad_norm": 0.9376161276388697, "learning_rate": 0.003, "loss": 4.1723, "step": 4459 }, { "epoch": 0.0446, "grad_norm": 1.04997158261314, "learning_rate": 0.003, "loss": 4.1739, "step": 4460 }, { "epoch": 0.04461, "grad_norm": 1.112982852233256, "learning_rate": 0.003, "loss": 4.174, "step": 4461 }, { "epoch": 0.04462, "grad_norm": 1.1238997045435504, "learning_rate": 0.003, "loss": 4.1339, "step": 4462 }, { "epoch": 0.04463, "grad_norm": 1.1095760340846703, "learning_rate": 0.003, "loss": 4.1465, "step": 4463 }, { "epoch": 0.04464, "grad_norm": 0.8395501354061055, "learning_rate": 0.003, "loss": 4.1571, "step": 4464 }, { "epoch": 0.04465, "grad_norm": 0.70825602546013, "learning_rate": 0.003, "loss": 4.1403, "step": 4465 }, { "epoch": 0.04466, "grad_norm": 0.6551509755636126, "learning_rate": 0.003, "loss": 4.1486, "step": 4466 }, { "epoch": 0.04467, "grad_norm": 0.6536469534418317, "learning_rate": 0.003, "loss": 4.1404, "step": 4467 }, { "epoch": 0.04468, "grad_norm": 0.7859836361638165, "learning_rate": 0.003, "loss": 4.1192, "step": 4468 }, { "epoch": 0.04469, "grad_norm": 0.9147137560720803, "learning_rate": 0.003, "loss": 4.1515, "step": 4469 }, { "epoch": 0.0447, "grad_norm": 1.0209143473160414, "learning_rate": 0.003, "loss": 4.1582, "step": 4470 }, { "epoch": 0.04471, "grad_norm": 0.8684398085030052, "learning_rate": 0.003, "loss": 4.1747, "step": 4471 }, { "epoch": 0.04472, "grad_norm": 0.7039517677091388, "learning_rate": 0.003, "loss": 4.121, "step": 4472 }, { "epoch": 0.04473, "grad_norm": 0.7016130752036785, "learning_rate": 0.003, "loss": 4.1228, "step": 4473 }, { "epoch": 0.04474, "grad_norm": 0.819372242317214, "learning_rate": 0.003, "loss": 4.1202, "step": 4474 }, { "epoch": 0.04475, "grad_norm": 0.9581916584575547, "learning_rate": 0.003, "loss": 4.1152, "step": 4475 }, { "epoch": 0.04476, "grad_norm": 1.0128184224439518, "learning_rate": 0.003, "loss": 4.1343, "step": 4476 }, { "epoch": 0.04477, "grad_norm": 0.9335743970032304, "learning_rate": 0.003, "loss": 4.1489, "step": 4477 }, { "epoch": 0.04478, "grad_norm": 0.9839256263771743, "learning_rate": 0.003, "loss": 4.1498, "step": 4478 }, { "epoch": 0.04479, "grad_norm": 1.0121152195223608, "learning_rate": 0.003, "loss": 4.1335, "step": 4479 }, { "epoch": 0.0448, "grad_norm": 1.1007103156748486, "learning_rate": 0.003, "loss": 4.1461, "step": 4480 }, { "epoch": 0.04481, "grad_norm": 0.7582518945106804, "learning_rate": 0.003, "loss": 4.1289, "step": 4481 }, { "epoch": 0.04482, "grad_norm": 0.6947887020501614, "learning_rate": 0.003, "loss": 4.15, "step": 4482 }, { "epoch": 0.04483, "grad_norm": 0.6849977721811616, "learning_rate": 0.003, "loss": 4.1374, "step": 4483 }, { "epoch": 0.04484, "grad_norm": 0.6987096200585385, "learning_rate": 0.003, "loss": 4.1514, "step": 4484 }, { "epoch": 0.04485, "grad_norm": 0.6929640245126742, "learning_rate": 0.003, "loss": 4.1581, "step": 4485 }, { "epoch": 0.04486, "grad_norm": 0.6288138860854519, "learning_rate": 0.003, "loss": 4.1182, "step": 4486 }, { "epoch": 0.04487, "grad_norm": 0.6828177453458262, "learning_rate": 0.003, "loss": 4.1276, "step": 4487 }, { "epoch": 0.04488, "grad_norm": 0.7640042454477429, "learning_rate": 0.003, "loss": 4.1472, "step": 4488 }, { "epoch": 0.04489, "grad_norm": 0.9955063503331009, "learning_rate": 0.003, "loss": 4.1542, "step": 4489 }, { "epoch": 0.0449, "grad_norm": 1.190431242593713, "learning_rate": 0.003, "loss": 4.1566, "step": 4490 }, { "epoch": 0.04491, "grad_norm": 0.729909947923579, "learning_rate": 0.003, "loss": 4.1341, "step": 4491 }, { "epoch": 0.04492, "grad_norm": 0.6337337418684384, "learning_rate": 0.003, "loss": 4.1441, "step": 4492 }, { "epoch": 0.04493, "grad_norm": 0.6901332741523903, "learning_rate": 0.003, "loss": 4.1496, "step": 4493 }, { "epoch": 0.04494, "grad_norm": 0.7753077021765635, "learning_rate": 0.003, "loss": 4.1312, "step": 4494 }, { "epoch": 0.04495, "grad_norm": 0.7970714712777819, "learning_rate": 0.003, "loss": 4.131, "step": 4495 }, { "epoch": 0.04496, "grad_norm": 0.7494590761073396, "learning_rate": 0.003, "loss": 4.1699, "step": 4496 }, { "epoch": 0.04497, "grad_norm": 0.7663642245891397, "learning_rate": 0.003, "loss": 4.148, "step": 4497 }, { "epoch": 0.04498, "grad_norm": 0.8315020221220991, "learning_rate": 0.003, "loss": 4.1238, "step": 4498 }, { "epoch": 0.04499, "grad_norm": 0.8744656956645718, "learning_rate": 0.003, "loss": 4.0979, "step": 4499 }, { "epoch": 0.045, "grad_norm": 0.8267996207966193, "learning_rate": 0.003, "loss": 4.1256, "step": 4500 }, { "epoch": 0.04501, "grad_norm": 0.8946790372935638, "learning_rate": 0.003, "loss": 4.1224, "step": 4501 }, { "epoch": 0.04502, "grad_norm": 0.9463619249258222, "learning_rate": 0.003, "loss": 4.1656, "step": 4502 }, { "epoch": 0.04503, "grad_norm": 1.0222323520895809, "learning_rate": 0.003, "loss": 4.1678, "step": 4503 }, { "epoch": 0.04504, "grad_norm": 1.0547640109337635, "learning_rate": 0.003, "loss": 4.1671, "step": 4504 }, { "epoch": 0.04505, "grad_norm": 0.9864660536132722, "learning_rate": 0.003, "loss": 4.1612, "step": 4505 }, { "epoch": 0.04506, "grad_norm": 1.0360024418113565, "learning_rate": 0.003, "loss": 4.146, "step": 4506 }, { "epoch": 0.04507, "grad_norm": 1.0278269548286743, "learning_rate": 0.003, "loss": 4.1244, "step": 4507 }, { "epoch": 0.04508, "grad_norm": 0.9639240103094384, "learning_rate": 0.003, "loss": 4.1406, "step": 4508 }, { "epoch": 0.04509, "grad_norm": 0.9853142135454404, "learning_rate": 0.003, "loss": 4.1662, "step": 4509 }, { "epoch": 0.0451, "grad_norm": 1.0935429480575658, "learning_rate": 0.003, "loss": 4.1727, "step": 4510 }, { "epoch": 0.04511, "grad_norm": 1.1102303545176306, "learning_rate": 0.003, "loss": 4.1554, "step": 4511 }, { "epoch": 0.04512, "grad_norm": 1.115670666411986, "learning_rate": 0.003, "loss": 4.1117, "step": 4512 }, { "epoch": 0.04513, "grad_norm": 0.8607532188769688, "learning_rate": 0.003, "loss": 4.135, "step": 4513 }, { "epoch": 0.04514, "grad_norm": 0.7573822293737257, "learning_rate": 0.003, "loss": 4.1412, "step": 4514 }, { "epoch": 0.04515, "grad_norm": 0.679162344485831, "learning_rate": 0.003, "loss": 4.1333, "step": 4515 }, { "epoch": 0.04516, "grad_norm": 0.7152570377504436, "learning_rate": 0.003, "loss": 4.1458, "step": 4516 }, { "epoch": 0.04517, "grad_norm": 0.7954703911013454, "learning_rate": 0.003, "loss": 4.1574, "step": 4517 }, { "epoch": 0.04518, "grad_norm": 1.0634551087943087, "learning_rate": 0.003, "loss": 4.1382, "step": 4518 }, { "epoch": 0.04519, "grad_norm": 1.2285232634943983, "learning_rate": 0.003, "loss": 4.1397, "step": 4519 }, { "epoch": 0.0452, "grad_norm": 0.8007618323918877, "learning_rate": 0.003, "loss": 4.1382, "step": 4520 }, { "epoch": 0.04521, "grad_norm": 0.6592457815722775, "learning_rate": 0.003, "loss": 4.1246, "step": 4521 }, { "epoch": 0.04522, "grad_norm": 0.8403672242278453, "learning_rate": 0.003, "loss": 4.1258, "step": 4522 }, { "epoch": 0.04523, "grad_norm": 1.1877972422707144, "learning_rate": 0.003, "loss": 4.1368, "step": 4523 }, { "epoch": 0.04524, "grad_norm": 0.8947697003509342, "learning_rate": 0.003, "loss": 4.1104, "step": 4524 }, { "epoch": 0.04525, "grad_norm": 0.7848986351194489, "learning_rate": 0.003, "loss": 4.1179, "step": 4525 }, { "epoch": 0.04526, "grad_norm": 0.7651712977009272, "learning_rate": 0.003, "loss": 4.1446, "step": 4526 }, { "epoch": 0.04527, "grad_norm": 0.6750467543871055, "learning_rate": 0.003, "loss": 4.1223, "step": 4527 }, { "epoch": 0.04528, "grad_norm": 0.7273916921816331, "learning_rate": 0.003, "loss": 4.1048, "step": 4528 }, { "epoch": 0.04529, "grad_norm": 0.8479138902943643, "learning_rate": 0.003, "loss": 4.1118, "step": 4529 }, { "epoch": 0.0453, "grad_norm": 0.9684628880640245, "learning_rate": 0.003, "loss": 4.114, "step": 4530 }, { "epoch": 0.04531, "grad_norm": 1.1134698284207358, "learning_rate": 0.003, "loss": 4.1205, "step": 4531 }, { "epoch": 0.04532, "grad_norm": 0.876191461867141, "learning_rate": 0.003, "loss": 4.1387, "step": 4532 }, { "epoch": 0.04533, "grad_norm": 0.8864267526479417, "learning_rate": 0.003, "loss": 4.1045, "step": 4533 }, { "epoch": 0.04534, "grad_norm": 0.8208227713611345, "learning_rate": 0.003, "loss": 4.1448, "step": 4534 }, { "epoch": 0.04535, "grad_norm": 0.848319739416912, "learning_rate": 0.003, "loss": 4.167, "step": 4535 }, { "epoch": 0.04536, "grad_norm": 0.6979960468142136, "learning_rate": 0.003, "loss": 4.1578, "step": 4536 }, { "epoch": 0.04537, "grad_norm": 0.7258752037017571, "learning_rate": 0.003, "loss": 4.1381, "step": 4537 }, { "epoch": 0.04538, "grad_norm": 0.7367885904517688, "learning_rate": 0.003, "loss": 4.1366, "step": 4538 }, { "epoch": 0.04539, "grad_norm": 0.7909398070219044, "learning_rate": 0.003, "loss": 4.1037, "step": 4539 }, { "epoch": 0.0454, "grad_norm": 0.9542272258252085, "learning_rate": 0.003, "loss": 4.1153, "step": 4540 }, { "epoch": 0.04541, "grad_norm": 1.342847961348652, "learning_rate": 0.003, "loss": 4.1716, "step": 4541 }, { "epoch": 0.04542, "grad_norm": 0.7341173411686917, "learning_rate": 0.003, "loss": 4.1106, "step": 4542 }, { "epoch": 0.04543, "grad_norm": 0.8199205913967764, "learning_rate": 0.003, "loss": 4.1425, "step": 4543 }, { "epoch": 0.04544, "grad_norm": 0.9991230666952743, "learning_rate": 0.003, "loss": 4.129, "step": 4544 }, { "epoch": 0.04545, "grad_norm": 1.154240542999549, "learning_rate": 0.003, "loss": 4.1457, "step": 4545 }, { "epoch": 0.04546, "grad_norm": 1.0006050135666062, "learning_rate": 0.003, "loss": 4.1274, "step": 4546 }, { "epoch": 0.04547, "grad_norm": 0.9848572236171086, "learning_rate": 0.003, "loss": 4.1334, "step": 4547 }, { "epoch": 0.04548, "grad_norm": 0.950231501902435, "learning_rate": 0.003, "loss": 4.1449, "step": 4548 }, { "epoch": 0.04549, "grad_norm": 1.0155067779223592, "learning_rate": 0.003, "loss": 4.1401, "step": 4549 }, { "epoch": 0.0455, "grad_norm": 1.0449688473715737, "learning_rate": 0.003, "loss": 4.1757, "step": 4550 }, { "epoch": 0.04551, "grad_norm": 0.901666260155581, "learning_rate": 0.003, "loss": 4.1821, "step": 4551 }, { "epoch": 0.04552, "grad_norm": 0.9049695661423774, "learning_rate": 0.003, "loss": 4.162, "step": 4552 }, { "epoch": 0.04553, "grad_norm": 0.9336311136993248, "learning_rate": 0.003, "loss": 4.1516, "step": 4553 }, { "epoch": 0.04554, "grad_norm": 0.8640303831252293, "learning_rate": 0.003, "loss": 4.1352, "step": 4554 }, { "epoch": 0.04555, "grad_norm": 0.9397425841906786, "learning_rate": 0.003, "loss": 4.1474, "step": 4555 }, { "epoch": 0.04556, "grad_norm": 0.9618493597139517, "learning_rate": 0.003, "loss": 4.152, "step": 4556 }, { "epoch": 0.04557, "grad_norm": 0.884753188680489, "learning_rate": 0.003, "loss": 4.1323, "step": 4557 }, { "epoch": 0.04558, "grad_norm": 0.7671272683143043, "learning_rate": 0.003, "loss": 4.1584, "step": 4558 }, { "epoch": 0.04559, "grad_norm": 0.6905499070998972, "learning_rate": 0.003, "loss": 4.1701, "step": 4559 }, { "epoch": 0.0456, "grad_norm": 0.6501916829076367, "learning_rate": 0.003, "loss": 4.1256, "step": 4560 }, { "epoch": 0.04561, "grad_norm": 0.7854319499541154, "learning_rate": 0.003, "loss": 4.138, "step": 4561 }, { "epoch": 0.04562, "grad_norm": 0.9267219643007959, "learning_rate": 0.003, "loss": 4.1642, "step": 4562 }, { "epoch": 0.04563, "grad_norm": 1.09883882872006, "learning_rate": 0.003, "loss": 4.1358, "step": 4563 }, { "epoch": 0.04564, "grad_norm": 0.8737747079220421, "learning_rate": 0.003, "loss": 4.1411, "step": 4564 }, { "epoch": 0.04565, "grad_norm": 0.8193816189119624, "learning_rate": 0.003, "loss": 4.1117, "step": 4565 }, { "epoch": 0.04566, "grad_norm": 0.870765536242901, "learning_rate": 0.003, "loss": 4.1465, "step": 4566 }, { "epoch": 0.04567, "grad_norm": 1.006590995989121, "learning_rate": 0.003, "loss": 4.1462, "step": 4567 }, { "epoch": 0.04568, "grad_norm": 1.042592813338758, "learning_rate": 0.003, "loss": 4.1272, "step": 4568 }, { "epoch": 0.04569, "grad_norm": 0.9179039021155998, "learning_rate": 0.003, "loss": 4.15, "step": 4569 }, { "epoch": 0.0457, "grad_norm": 0.9469764446785002, "learning_rate": 0.003, "loss": 4.1411, "step": 4570 }, { "epoch": 0.04571, "grad_norm": 0.9383380289023364, "learning_rate": 0.003, "loss": 4.1673, "step": 4571 }, { "epoch": 0.04572, "grad_norm": 1.141150830462319, "learning_rate": 0.003, "loss": 4.1469, "step": 4572 }, { "epoch": 0.04573, "grad_norm": 0.8749894272125335, "learning_rate": 0.003, "loss": 4.1135, "step": 4573 }, { "epoch": 0.04574, "grad_norm": 1.071173257933056, "learning_rate": 0.003, "loss": 4.1315, "step": 4574 }, { "epoch": 0.04575, "grad_norm": 1.1588296020864692, "learning_rate": 0.003, "loss": 4.1607, "step": 4575 }, { "epoch": 0.04576, "grad_norm": 1.0001116928921225, "learning_rate": 0.003, "loss": 4.1479, "step": 4576 }, { "epoch": 0.04577, "grad_norm": 1.1162255859546153, "learning_rate": 0.003, "loss": 4.1711, "step": 4577 }, { "epoch": 0.04578, "grad_norm": 0.8912231810669947, "learning_rate": 0.003, "loss": 4.1337, "step": 4578 }, { "epoch": 0.04579, "grad_norm": 0.7751452813751742, "learning_rate": 0.003, "loss": 4.113, "step": 4579 }, { "epoch": 0.0458, "grad_norm": 0.812467570483858, "learning_rate": 0.003, "loss": 4.1495, "step": 4580 }, { "epoch": 0.04581, "grad_norm": 0.9496499343062301, "learning_rate": 0.003, "loss": 4.1642, "step": 4581 }, { "epoch": 0.04582, "grad_norm": 1.257095124885388, "learning_rate": 0.003, "loss": 4.1743, "step": 4582 }, { "epoch": 0.04583, "grad_norm": 1.0572142245473386, "learning_rate": 0.003, "loss": 4.1413, "step": 4583 }, { "epoch": 0.04584, "grad_norm": 1.0493651940978534, "learning_rate": 0.003, "loss": 4.143, "step": 4584 }, { "epoch": 0.04585, "grad_norm": 1.0075480210769747, "learning_rate": 0.003, "loss": 4.1423, "step": 4585 }, { "epoch": 0.04586, "grad_norm": 0.9609773823228706, "learning_rate": 0.003, "loss": 4.1455, "step": 4586 }, { "epoch": 0.04587, "grad_norm": 0.8094793207666494, "learning_rate": 0.003, "loss": 4.1598, "step": 4587 }, { "epoch": 0.04588, "grad_norm": 0.6497566279929464, "learning_rate": 0.003, "loss": 4.153, "step": 4588 }, { "epoch": 0.04589, "grad_norm": 0.6395143391670874, "learning_rate": 0.003, "loss": 4.1187, "step": 4589 }, { "epoch": 0.0459, "grad_norm": 0.6017649664143976, "learning_rate": 0.003, "loss": 4.1404, "step": 4590 }, { "epoch": 0.04591, "grad_norm": 0.5348849495027335, "learning_rate": 0.003, "loss": 4.105, "step": 4591 }, { "epoch": 0.04592, "grad_norm": 0.6214267296386631, "learning_rate": 0.003, "loss": 4.1304, "step": 4592 }, { "epoch": 0.04593, "grad_norm": 0.6662326597458281, "learning_rate": 0.003, "loss": 4.1526, "step": 4593 }, { "epoch": 0.04594, "grad_norm": 0.9037274242355646, "learning_rate": 0.003, "loss": 4.1498, "step": 4594 }, { "epoch": 0.04595, "grad_norm": 1.0198140749645483, "learning_rate": 0.003, "loss": 4.1408, "step": 4595 }, { "epoch": 0.04596, "grad_norm": 0.9085054402184916, "learning_rate": 0.003, "loss": 4.1422, "step": 4596 }, { "epoch": 0.04597, "grad_norm": 0.818858115564251, "learning_rate": 0.003, "loss": 4.1439, "step": 4597 }, { "epoch": 0.04598, "grad_norm": 0.8218121161747024, "learning_rate": 0.003, "loss": 4.1774, "step": 4598 }, { "epoch": 0.04599, "grad_norm": 0.84010102233204, "learning_rate": 0.003, "loss": 4.1261, "step": 4599 }, { "epoch": 0.046, "grad_norm": 0.8552355887980282, "learning_rate": 0.003, "loss": 4.1394, "step": 4600 }, { "epoch": 0.04601, "grad_norm": 0.8950007645860975, "learning_rate": 0.003, "loss": 4.1446, "step": 4601 }, { "epoch": 0.04602, "grad_norm": 0.9057095536124716, "learning_rate": 0.003, "loss": 4.137, "step": 4602 }, { "epoch": 0.04603, "grad_norm": 0.957183477029017, "learning_rate": 0.003, "loss": 4.1534, "step": 4603 }, { "epoch": 0.04604, "grad_norm": 0.9080594849186558, "learning_rate": 0.003, "loss": 4.1496, "step": 4604 }, { "epoch": 0.04605, "grad_norm": 0.8631739974861561, "learning_rate": 0.003, "loss": 4.1569, "step": 4605 }, { "epoch": 0.04606, "grad_norm": 0.8826596745901574, "learning_rate": 0.003, "loss": 4.1212, "step": 4606 }, { "epoch": 0.04607, "grad_norm": 0.8058337880510137, "learning_rate": 0.003, "loss": 4.1298, "step": 4607 }, { "epoch": 0.04608, "grad_norm": 0.9071710574106185, "learning_rate": 0.003, "loss": 4.1128, "step": 4608 }, { "epoch": 0.04609, "grad_norm": 1.1450997362229134, "learning_rate": 0.003, "loss": 4.1442, "step": 4609 }, { "epoch": 0.0461, "grad_norm": 0.9244793191418952, "learning_rate": 0.003, "loss": 4.1194, "step": 4610 }, { "epoch": 0.04611, "grad_norm": 0.8489428980051721, "learning_rate": 0.003, "loss": 4.1642, "step": 4611 }, { "epoch": 0.04612, "grad_norm": 0.9711813069563194, "learning_rate": 0.003, "loss": 4.1554, "step": 4612 }, { "epoch": 0.04613, "grad_norm": 0.9867549615620282, "learning_rate": 0.003, "loss": 4.1465, "step": 4613 }, { "epoch": 0.04614, "grad_norm": 1.0312719455514217, "learning_rate": 0.003, "loss": 4.1103, "step": 4614 }, { "epoch": 0.04615, "grad_norm": 0.9155691103438471, "learning_rate": 0.003, "loss": 4.1472, "step": 4615 }, { "epoch": 0.04616, "grad_norm": 0.8072019861534341, "learning_rate": 0.003, "loss": 4.1246, "step": 4616 }, { "epoch": 0.04617, "grad_norm": 0.7700754980869141, "learning_rate": 0.003, "loss": 4.147, "step": 4617 }, { "epoch": 0.04618, "grad_norm": 1.0494097265203939, "learning_rate": 0.003, "loss": 4.1344, "step": 4618 }, { "epoch": 0.04619, "grad_norm": 1.2538022534808464, "learning_rate": 0.003, "loss": 4.1391, "step": 4619 }, { "epoch": 0.0462, "grad_norm": 0.8505682632402205, "learning_rate": 0.003, "loss": 4.1615, "step": 4620 }, { "epoch": 0.04621, "grad_norm": 0.8349859298521801, "learning_rate": 0.003, "loss": 4.1671, "step": 4621 }, { "epoch": 0.04622, "grad_norm": 0.8636675338766925, "learning_rate": 0.003, "loss": 4.1644, "step": 4622 }, { "epoch": 0.04623, "grad_norm": 0.9109557261630941, "learning_rate": 0.003, "loss": 4.1507, "step": 4623 }, { "epoch": 0.04624, "grad_norm": 0.8181989834734489, "learning_rate": 0.003, "loss": 4.1212, "step": 4624 }, { "epoch": 0.04625, "grad_norm": 0.836714187206783, "learning_rate": 0.003, "loss": 4.1419, "step": 4625 }, { "epoch": 0.04626, "grad_norm": 0.8974854040725699, "learning_rate": 0.003, "loss": 4.148, "step": 4626 }, { "epoch": 0.04627, "grad_norm": 0.8399612016833758, "learning_rate": 0.003, "loss": 4.1329, "step": 4627 }, { "epoch": 0.04628, "grad_norm": 0.9152472588666292, "learning_rate": 0.003, "loss": 4.1602, "step": 4628 }, { "epoch": 0.04629, "grad_norm": 0.9681854075822318, "learning_rate": 0.003, "loss": 4.1343, "step": 4629 }, { "epoch": 0.0463, "grad_norm": 1.2804113609132692, "learning_rate": 0.003, "loss": 4.1606, "step": 4630 }, { "epoch": 0.04631, "grad_norm": 0.8985924390327231, "learning_rate": 0.003, "loss": 4.1326, "step": 4631 }, { "epoch": 0.04632, "grad_norm": 0.6464228130780736, "learning_rate": 0.003, "loss": 4.1205, "step": 4632 }, { "epoch": 0.04633, "grad_norm": 0.6316825408575106, "learning_rate": 0.003, "loss": 4.1225, "step": 4633 }, { "epoch": 0.04634, "grad_norm": 0.7173406958415919, "learning_rate": 0.003, "loss": 4.1251, "step": 4634 }, { "epoch": 0.04635, "grad_norm": 0.9060915985310092, "learning_rate": 0.003, "loss": 4.131, "step": 4635 }, { "epoch": 0.04636, "grad_norm": 1.1784178209510372, "learning_rate": 0.003, "loss": 4.1746, "step": 4636 }, { "epoch": 0.04637, "grad_norm": 0.922005994009661, "learning_rate": 0.003, "loss": 4.1143, "step": 4637 }, { "epoch": 0.04638, "grad_norm": 0.7797208498501719, "learning_rate": 0.003, "loss": 4.1471, "step": 4638 }, { "epoch": 0.04639, "grad_norm": 0.9045238193488381, "learning_rate": 0.003, "loss": 4.1228, "step": 4639 }, { "epoch": 0.0464, "grad_norm": 0.8873863035754851, "learning_rate": 0.003, "loss": 4.1475, "step": 4640 }, { "epoch": 0.04641, "grad_norm": 0.9577308333714971, "learning_rate": 0.003, "loss": 4.1647, "step": 4641 }, { "epoch": 0.04642, "grad_norm": 1.053057664461161, "learning_rate": 0.003, "loss": 4.1155, "step": 4642 }, { "epoch": 0.04643, "grad_norm": 1.0102454910670058, "learning_rate": 0.003, "loss": 4.138, "step": 4643 }, { "epoch": 0.04644, "grad_norm": 1.0170448253769175, "learning_rate": 0.003, "loss": 4.1427, "step": 4644 }, { "epoch": 0.04645, "grad_norm": 1.0485498444094856, "learning_rate": 0.003, "loss": 4.1394, "step": 4645 }, { "epoch": 0.04646, "grad_norm": 1.1607087803243306, "learning_rate": 0.003, "loss": 4.1437, "step": 4646 }, { "epoch": 0.04647, "grad_norm": 0.984885992390032, "learning_rate": 0.003, "loss": 4.1376, "step": 4647 }, { "epoch": 0.04648, "grad_norm": 1.042823798476842, "learning_rate": 0.003, "loss": 4.1612, "step": 4648 }, { "epoch": 0.04649, "grad_norm": 1.1175899168625645, "learning_rate": 0.003, "loss": 4.1508, "step": 4649 }, { "epoch": 0.0465, "grad_norm": 1.1641349913018662, "learning_rate": 0.003, "loss": 4.1244, "step": 4650 }, { "epoch": 0.04651, "grad_norm": 1.1182777997481035, "learning_rate": 0.003, "loss": 4.153, "step": 4651 }, { "epoch": 0.04652, "grad_norm": 1.1639122814807727, "learning_rate": 0.003, "loss": 4.1678, "step": 4652 }, { "epoch": 0.04653, "grad_norm": 0.9052375805967465, "learning_rate": 0.003, "loss": 4.1619, "step": 4653 }, { "epoch": 0.04654, "grad_norm": 0.7923411907134336, "learning_rate": 0.003, "loss": 4.1484, "step": 4654 }, { "epoch": 0.04655, "grad_norm": 0.8924156423091137, "learning_rate": 0.003, "loss": 4.1563, "step": 4655 }, { "epoch": 0.04656, "grad_norm": 1.2505936066471501, "learning_rate": 0.003, "loss": 4.1359, "step": 4656 }, { "epoch": 0.04657, "grad_norm": 0.7705046707851887, "learning_rate": 0.003, "loss": 4.1645, "step": 4657 }, { "epoch": 0.04658, "grad_norm": 0.6721536386762502, "learning_rate": 0.003, "loss": 4.1323, "step": 4658 }, { "epoch": 0.04659, "grad_norm": 0.7136764569063401, "learning_rate": 0.003, "loss": 4.1342, "step": 4659 }, { "epoch": 0.0466, "grad_norm": 0.7116121289071756, "learning_rate": 0.003, "loss": 4.1514, "step": 4660 }, { "epoch": 0.04661, "grad_norm": 0.583275742533722, "learning_rate": 0.003, "loss": 4.1312, "step": 4661 }, { "epoch": 0.04662, "grad_norm": 0.6128088997283417, "learning_rate": 0.003, "loss": 4.1179, "step": 4662 }, { "epoch": 0.04663, "grad_norm": 0.7407134458121065, "learning_rate": 0.003, "loss": 4.1396, "step": 4663 }, { "epoch": 0.04664, "grad_norm": 0.9058911754062067, "learning_rate": 0.003, "loss": 4.1169, "step": 4664 }, { "epoch": 0.04665, "grad_norm": 1.1897493791033456, "learning_rate": 0.003, "loss": 4.1244, "step": 4665 }, { "epoch": 0.04666, "grad_norm": 0.8049445127211454, "learning_rate": 0.003, "loss": 4.1169, "step": 4666 }, { "epoch": 0.04667, "grad_norm": 0.6282022686305142, "learning_rate": 0.003, "loss": 4.1137, "step": 4667 }, { "epoch": 0.04668, "grad_norm": 0.589167104573055, "learning_rate": 0.003, "loss": 4.0868, "step": 4668 }, { "epoch": 0.04669, "grad_norm": 0.7583426981182202, "learning_rate": 0.003, "loss": 4.1473, "step": 4669 }, { "epoch": 0.0467, "grad_norm": 0.9171714392229983, "learning_rate": 0.003, "loss": 4.0986, "step": 4670 }, { "epoch": 0.04671, "grad_norm": 1.006207243577008, "learning_rate": 0.003, "loss": 4.1294, "step": 4671 }, { "epoch": 0.04672, "grad_norm": 1.063976185292903, "learning_rate": 0.003, "loss": 4.1448, "step": 4672 }, { "epoch": 0.04673, "grad_norm": 0.9424991510428172, "learning_rate": 0.003, "loss": 4.1482, "step": 4673 }, { "epoch": 0.04674, "grad_norm": 0.9346287569779962, "learning_rate": 0.003, "loss": 4.1164, "step": 4674 }, { "epoch": 0.04675, "grad_norm": 0.9394221276053155, "learning_rate": 0.003, "loss": 4.143, "step": 4675 }, { "epoch": 0.04676, "grad_norm": 1.143713314597062, "learning_rate": 0.003, "loss": 4.1526, "step": 4676 }, { "epoch": 0.04677, "grad_norm": 1.0492560762157657, "learning_rate": 0.003, "loss": 4.1572, "step": 4677 }, { "epoch": 0.04678, "grad_norm": 0.9237030884818286, "learning_rate": 0.003, "loss": 4.1102, "step": 4678 }, { "epoch": 0.04679, "grad_norm": 1.0604849270980485, "learning_rate": 0.003, "loss": 4.1502, "step": 4679 }, { "epoch": 0.0468, "grad_norm": 0.9599319014900776, "learning_rate": 0.003, "loss": 4.1206, "step": 4680 }, { "epoch": 0.04681, "grad_norm": 1.1130119095403597, "learning_rate": 0.003, "loss": 4.1766, "step": 4681 }, { "epoch": 0.04682, "grad_norm": 1.0396062685784013, "learning_rate": 0.003, "loss": 4.1404, "step": 4682 }, { "epoch": 0.04683, "grad_norm": 0.9560921327155597, "learning_rate": 0.003, "loss": 4.1339, "step": 4683 }, { "epoch": 0.04684, "grad_norm": 0.9641619437962302, "learning_rate": 0.003, "loss": 4.1543, "step": 4684 }, { "epoch": 0.04685, "grad_norm": 0.8865003876092341, "learning_rate": 0.003, "loss": 4.1364, "step": 4685 }, { "epoch": 0.04686, "grad_norm": 0.9255222896745879, "learning_rate": 0.003, "loss": 4.1467, "step": 4686 }, { "epoch": 0.04687, "grad_norm": 0.9164703207217101, "learning_rate": 0.003, "loss": 4.1737, "step": 4687 }, { "epoch": 0.04688, "grad_norm": 0.8576876310928183, "learning_rate": 0.003, "loss": 4.1534, "step": 4688 }, { "epoch": 0.04689, "grad_norm": 0.7547156448268271, "learning_rate": 0.003, "loss": 4.1655, "step": 4689 }, { "epoch": 0.0469, "grad_norm": 0.8110688622338084, "learning_rate": 0.003, "loss": 4.1169, "step": 4690 }, { "epoch": 0.04691, "grad_norm": 0.7871973858039091, "learning_rate": 0.003, "loss": 4.1284, "step": 4691 }, { "epoch": 0.04692, "grad_norm": 0.9924565689192236, "learning_rate": 0.003, "loss": 4.1389, "step": 4692 }, { "epoch": 0.04693, "grad_norm": 1.1840555444723178, "learning_rate": 0.003, "loss": 4.1699, "step": 4693 }, { "epoch": 0.04694, "grad_norm": 0.8956812289363905, "learning_rate": 0.003, "loss": 4.1282, "step": 4694 }, { "epoch": 0.04695, "grad_norm": 0.9376926331756155, "learning_rate": 0.003, "loss": 4.1166, "step": 4695 }, { "epoch": 0.04696, "grad_norm": 0.8795403664260718, "learning_rate": 0.003, "loss": 4.1578, "step": 4696 }, { "epoch": 0.04697, "grad_norm": 0.8357794249885189, "learning_rate": 0.003, "loss": 4.1396, "step": 4697 }, { "epoch": 0.04698, "grad_norm": 0.6738545963309196, "learning_rate": 0.003, "loss": 4.1382, "step": 4698 }, { "epoch": 0.04699, "grad_norm": 0.5951615468449445, "learning_rate": 0.003, "loss": 4.1238, "step": 4699 }, { "epoch": 0.047, "grad_norm": 0.6225150714509209, "learning_rate": 0.003, "loss": 4.1376, "step": 4700 }, { "epoch": 0.04701, "grad_norm": 0.6437330166954227, "learning_rate": 0.003, "loss": 4.1458, "step": 4701 }, { "epoch": 0.04702, "grad_norm": 0.7562946776902397, "learning_rate": 0.003, "loss": 4.1394, "step": 4702 }, { "epoch": 0.04703, "grad_norm": 0.9543436677847786, "learning_rate": 0.003, "loss": 4.1288, "step": 4703 }, { "epoch": 0.04704, "grad_norm": 1.4579145170633065, "learning_rate": 0.003, "loss": 4.1357, "step": 4704 }, { "epoch": 0.04705, "grad_norm": 0.6960477975136533, "learning_rate": 0.003, "loss": 4.1568, "step": 4705 }, { "epoch": 0.04706, "grad_norm": 0.5563368169699227, "learning_rate": 0.003, "loss": 4.0948, "step": 4706 }, { "epoch": 0.04707, "grad_norm": 0.7028499109080264, "learning_rate": 0.003, "loss": 4.1378, "step": 4707 }, { "epoch": 0.04708, "grad_norm": 0.868294214987963, "learning_rate": 0.003, "loss": 4.0955, "step": 4708 }, { "epoch": 0.04709, "grad_norm": 1.0806652270407238, "learning_rate": 0.003, "loss": 4.1316, "step": 4709 }, { "epoch": 0.0471, "grad_norm": 0.9045144570670249, "learning_rate": 0.003, "loss": 4.1425, "step": 4710 }, { "epoch": 0.04711, "grad_norm": 0.7929082522931609, "learning_rate": 0.003, "loss": 4.1337, "step": 4711 }, { "epoch": 0.04712, "grad_norm": 0.8998152710604039, "learning_rate": 0.003, "loss": 4.1272, "step": 4712 }, { "epoch": 0.04713, "grad_norm": 0.900819429264007, "learning_rate": 0.003, "loss": 4.1239, "step": 4713 }, { "epoch": 0.04714, "grad_norm": 0.8770728733527219, "learning_rate": 0.003, "loss": 4.1336, "step": 4714 }, { "epoch": 0.04715, "grad_norm": 0.9305052960807428, "learning_rate": 0.003, "loss": 4.1627, "step": 4715 }, { "epoch": 0.04716, "grad_norm": 1.0563148846171344, "learning_rate": 0.003, "loss": 4.1393, "step": 4716 }, { "epoch": 0.04717, "grad_norm": 1.0183123784712407, "learning_rate": 0.003, "loss": 4.158, "step": 4717 }, { "epoch": 0.04718, "grad_norm": 1.0249581976294773, "learning_rate": 0.003, "loss": 4.1235, "step": 4718 }, { "epoch": 0.04719, "grad_norm": 0.9289493979577722, "learning_rate": 0.003, "loss": 4.132, "step": 4719 }, { "epoch": 0.0472, "grad_norm": 0.8741526678227087, "learning_rate": 0.003, "loss": 4.1578, "step": 4720 }, { "epoch": 0.04721, "grad_norm": 1.2410401068900363, "learning_rate": 0.003, "loss": 4.1289, "step": 4721 }, { "epoch": 0.04722, "grad_norm": 1.1958207913096726, "learning_rate": 0.003, "loss": 4.1467, "step": 4722 }, { "epoch": 0.04723, "grad_norm": 0.983729916262391, "learning_rate": 0.003, "loss": 4.1525, "step": 4723 }, { "epoch": 0.04724, "grad_norm": 0.9846514174679839, "learning_rate": 0.003, "loss": 4.131, "step": 4724 }, { "epoch": 0.04725, "grad_norm": 1.2006797419141146, "learning_rate": 0.003, "loss": 4.1585, "step": 4725 }, { "epoch": 0.04726, "grad_norm": 0.7619404889077381, "learning_rate": 0.003, "loss": 4.138, "step": 4726 }, { "epoch": 0.04727, "grad_norm": 0.6760195770103148, "learning_rate": 0.003, "loss": 4.1183, "step": 4727 }, { "epoch": 0.04728, "grad_norm": 0.6711364467462053, "learning_rate": 0.003, "loss": 4.1259, "step": 4728 }, { "epoch": 0.04729, "grad_norm": 0.6961741239186977, "learning_rate": 0.003, "loss": 4.141, "step": 4729 }, { "epoch": 0.0473, "grad_norm": 0.7303647620564638, "learning_rate": 0.003, "loss": 4.1353, "step": 4730 }, { "epoch": 0.04731, "grad_norm": 0.712139084345733, "learning_rate": 0.003, "loss": 4.1282, "step": 4731 }, { "epoch": 0.04732, "grad_norm": 0.8077440937117389, "learning_rate": 0.003, "loss": 4.143, "step": 4732 }, { "epoch": 0.04733, "grad_norm": 1.1656166796396292, "learning_rate": 0.003, "loss": 4.1325, "step": 4733 }, { "epoch": 0.04734, "grad_norm": 1.166287688296092, "learning_rate": 0.003, "loss": 4.1433, "step": 4734 }, { "epoch": 0.04735, "grad_norm": 0.7216780900295331, "learning_rate": 0.003, "loss": 4.1516, "step": 4735 }, { "epoch": 0.04736, "grad_norm": 0.7134965352325359, "learning_rate": 0.003, "loss": 4.1348, "step": 4736 }, { "epoch": 0.04737, "grad_norm": 0.8921498388016378, "learning_rate": 0.003, "loss": 4.1039, "step": 4737 }, { "epoch": 0.04738, "grad_norm": 1.1445099509639862, "learning_rate": 0.003, "loss": 4.1524, "step": 4738 }, { "epoch": 0.04739, "grad_norm": 1.1030230453118304, "learning_rate": 0.003, "loss": 4.0873, "step": 4739 }, { "epoch": 0.0474, "grad_norm": 0.8584656214420842, "learning_rate": 0.003, "loss": 4.1535, "step": 4740 }, { "epoch": 0.04741, "grad_norm": 0.8035201218462465, "learning_rate": 0.003, "loss": 4.1414, "step": 4741 }, { "epoch": 0.04742, "grad_norm": 0.8357881665845689, "learning_rate": 0.003, "loss": 4.1292, "step": 4742 }, { "epoch": 0.04743, "grad_norm": 0.9521041649493661, "learning_rate": 0.003, "loss": 4.1522, "step": 4743 }, { "epoch": 0.04744, "grad_norm": 0.9335234952847369, "learning_rate": 0.003, "loss": 4.1154, "step": 4744 }, { "epoch": 0.04745, "grad_norm": 0.9092467912620412, "learning_rate": 0.003, "loss": 4.1178, "step": 4745 }, { "epoch": 0.04746, "grad_norm": 1.0554547866629964, "learning_rate": 0.003, "loss": 4.1384, "step": 4746 }, { "epoch": 0.04747, "grad_norm": 0.9041793629659869, "learning_rate": 0.003, "loss": 4.1109, "step": 4747 }, { "epoch": 0.04748, "grad_norm": 0.858859989561575, "learning_rate": 0.003, "loss": 4.1554, "step": 4748 }, { "epoch": 0.04749, "grad_norm": 0.9479384146537131, "learning_rate": 0.003, "loss": 4.1658, "step": 4749 }, { "epoch": 0.0475, "grad_norm": 1.0250013309916148, "learning_rate": 0.003, "loss": 4.1324, "step": 4750 }, { "epoch": 0.04751, "grad_norm": 0.9606461873821214, "learning_rate": 0.003, "loss": 4.0976, "step": 4751 }, { "epoch": 0.04752, "grad_norm": 0.9181341108641666, "learning_rate": 0.003, "loss": 4.1273, "step": 4752 }, { "epoch": 0.04753, "grad_norm": 0.8507586417545276, "learning_rate": 0.003, "loss": 4.1324, "step": 4753 }, { "epoch": 0.04754, "grad_norm": 0.8523834824717018, "learning_rate": 0.003, "loss": 4.1515, "step": 4754 }, { "epoch": 0.04755, "grad_norm": 0.789624720859558, "learning_rate": 0.003, "loss": 4.1302, "step": 4755 }, { "epoch": 0.04756, "grad_norm": 0.6798364490430269, "learning_rate": 0.003, "loss": 4.1285, "step": 4756 }, { "epoch": 0.04757, "grad_norm": 0.8431885193516194, "learning_rate": 0.003, "loss": 4.1285, "step": 4757 }, { "epoch": 0.04758, "grad_norm": 1.0193242917886531, "learning_rate": 0.003, "loss": 4.1239, "step": 4758 }, { "epoch": 0.04759, "grad_norm": 1.1675022300199527, "learning_rate": 0.003, "loss": 4.1418, "step": 4759 }, { "epoch": 0.0476, "grad_norm": 0.8374418244140791, "learning_rate": 0.003, "loss": 4.1154, "step": 4760 }, { "epoch": 0.04761, "grad_norm": 0.8004744833013939, "learning_rate": 0.003, "loss": 4.1243, "step": 4761 }, { "epoch": 0.04762, "grad_norm": 0.8950693235914801, "learning_rate": 0.003, "loss": 4.12, "step": 4762 }, { "epoch": 0.04763, "grad_norm": 0.9038705695942526, "learning_rate": 0.003, "loss": 4.1216, "step": 4763 }, { "epoch": 0.04764, "grad_norm": 1.0083140925650727, "learning_rate": 0.003, "loss": 4.115, "step": 4764 }, { "epoch": 0.04765, "grad_norm": 1.2084489092590291, "learning_rate": 0.003, "loss": 4.1734, "step": 4765 }, { "epoch": 0.04766, "grad_norm": 0.9331486718026547, "learning_rate": 0.003, "loss": 4.1369, "step": 4766 }, { "epoch": 0.04767, "grad_norm": 1.032810812597879, "learning_rate": 0.003, "loss": 4.1412, "step": 4767 }, { "epoch": 0.04768, "grad_norm": 0.8079898898652215, "learning_rate": 0.003, "loss": 4.1305, "step": 4768 }, { "epoch": 0.04769, "grad_norm": 0.8175014309895597, "learning_rate": 0.003, "loss": 4.0936, "step": 4769 }, { "epoch": 0.0477, "grad_norm": 1.040815138157364, "learning_rate": 0.003, "loss": 4.1151, "step": 4770 }, { "epoch": 0.04771, "grad_norm": 1.1500954512552088, "learning_rate": 0.003, "loss": 4.1434, "step": 4771 }, { "epoch": 0.04772, "grad_norm": 0.8897742060078679, "learning_rate": 0.003, "loss": 4.1374, "step": 4772 }, { "epoch": 0.04773, "grad_norm": 0.8351692338869225, "learning_rate": 0.003, "loss": 4.1568, "step": 4773 }, { "epoch": 0.04774, "grad_norm": 0.9151298772105244, "learning_rate": 0.003, "loss": 4.1427, "step": 4774 }, { "epoch": 0.04775, "grad_norm": 0.9751550319429745, "learning_rate": 0.003, "loss": 4.1406, "step": 4775 }, { "epoch": 0.04776, "grad_norm": 0.9638676545283447, "learning_rate": 0.003, "loss": 4.1315, "step": 4776 }, { "epoch": 0.04777, "grad_norm": 1.1646838839296059, "learning_rate": 0.003, "loss": 4.1355, "step": 4777 }, { "epoch": 0.04778, "grad_norm": 0.8824685892267762, "learning_rate": 0.003, "loss": 4.1061, "step": 4778 }, { "epoch": 0.04779, "grad_norm": 0.7623197654914352, "learning_rate": 0.003, "loss": 4.1422, "step": 4779 }, { "epoch": 0.0478, "grad_norm": 0.7803964563209768, "learning_rate": 0.003, "loss": 4.1446, "step": 4780 }, { "epoch": 0.04781, "grad_norm": 0.8677755571648247, "learning_rate": 0.003, "loss": 4.1248, "step": 4781 }, { "epoch": 0.04782, "grad_norm": 0.8977177055850454, "learning_rate": 0.003, "loss": 4.1248, "step": 4782 }, { "epoch": 0.04783, "grad_norm": 1.0119802996771465, "learning_rate": 0.003, "loss": 4.1272, "step": 4783 }, { "epoch": 0.04784, "grad_norm": 0.8774934854130646, "learning_rate": 0.003, "loss": 4.1453, "step": 4784 }, { "epoch": 0.04785, "grad_norm": 0.7857875427258052, "learning_rate": 0.003, "loss": 4.1438, "step": 4785 }, { "epoch": 0.04786, "grad_norm": 0.8820990387768732, "learning_rate": 0.003, "loss": 4.1463, "step": 4786 }, { "epoch": 0.04787, "grad_norm": 1.3348035022755884, "learning_rate": 0.003, "loss": 4.1544, "step": 4787 }, { "epoch": 0.04788, "grad_norm": 0.9722902455641531, "learning_rate": 0.003, "loss": 4.1323, "step": 4788 }, { "epoch": 0.04789, "grad_norm": 0.8249065978709161, "learning_rate": 0.003, "loss": 4.114, "step": 4789 }, { "epoch": 0.0479, "grad_norm": 0.8098922858184793, "learning_rate": 0.003, "loss": 4.1301, "step": 4790 }, { "epoch": 0.04791, "grad_norm": 0.8530020182040862, "learning_rate": 0.003, "loss": 4.1489, "step": 4791 }, { "epoch": 0.04792, "grad_norm": 0.9361191470916291, "learning_rate": 0.003, "loss": 4.1596, "step": 4792 }, { "epoch": 0.04793, "grad_norm": 0.8995706892213676, "learning_rate": 0.003, "loss": 4.1215, "step": 4793 }, { "epoch": 0.04794, "grad_norm": 0.8946924778288801, "learning_rate": 0.003, "loss": 4.1406, "step": 4794 }, { "epoch": 0.04795, "grad_norm": 0.955319338607847, "learning_rate": 0.003, "loss": 4.1433, "step": 4795 }, { "epoch": 0.04796, "grad_norm": 0.9209462528560896, "learning_rate": 0.003, "loss": 4.1112, "step": 4796 }, { "epoch": 0.04797, "grad_norm": 0.9072552403900586, "learning_rate": 0.003, "loss": 4.1462, "step": 4797 }, { "epoch": 0.04798, "grad_norm": 1.1376509845610743, "learning_rate": 0.003, "loss": 4.1115, "step": 4798 }, { "epoch": 0.04799, "grad_norm": 0.9872830450267011, "learning_rate": 0.003, "loss": 4.1212, "step": 4799 }, { "epoch": 0.048, "grad_norm": 1.0296601446082432, "learning_rate": 0.003, "loss": 4.1291, "step": 4800 }, { "epoch": 0.04801, "grad_norm": 0.9104805461757415, "learning_rate": 0.003, "loss": 4.1364, "step": 4801 }, { "epoch": 0.04802, "grad_norm": 0.9761083085564308, "learning_rate": 0.003, "loss": 4.1105, "step": 4802 }, { "epoch": 0.04803, "grad_norm": 0.9665537842868093, "learning_rate": 0.003, "loss": 4.1209, "step": 4803 }, { "epoch": 0.04804, "grad_norm": 0.9087746601238126, "learning_rate": 0.003, "loss": 4.1511, "step": 4804 }, { "epoch": 0.04805, "grad_norm": 0.8359020051915194, "learning_rate": 0.003, "loss": 4.1491, "step": 4805 }, { "epoch": 0.04806, "grad_norm": 1.0009769366999013, "learning_rate": 0.003, "loss": 4.0985, "step": 4806 }, { "epoch": 0.04807, "grad_norm": 1.3538106119491318, "learning_rate": 0.003, "loss": 4.1631, "step": 4807 }, { "epoch": 0.04808, "grad_norm": 0.7759337211545583, "learning_rate": 0.003, "loss": 4.129, "step": 4808 }, { "epoch": 0.04809, "grad_norm": 0.7867983138442217, "learning_rate": 0.003, "loss": 4.1203, "step": 4809 }, { "epoch": 0.0481, "grad_norm": 1.0460647834530599, "learning_rate": 0.003, "loss": 4.1469, "step": 4810 }, { "epoch": 0.04811, "grad_norm": 0.9588194390032568, "learning_rate": 0.003, "loss": 4.1436, "step": 4811 }, { "epoch": 0.04812, "grad_norm": 1.2072876723569714, "learning_rate": 0.003, "loss": 4.1503, "step": 4812 }, { "epoch": 0.04813, "grad_norm": 0.8442155640566389, "learning_rate": 0.003, "loss": 4.1311, "step": 4813 }, { "epoch": 0.04814, "grad_norm": 0.7342275117522243, "learning_rate": 0.003, "loss": 4.0838, "step": 4814 }, { "epoch": 0.04815, "grad_norm": 0.709374442322939, "learning_rate": 0.003, "loss": 4.1253, "step": 4815 }, { "epoch": 0.04816, "grad_norm": 0.7917399222991578, "learning_rate": 0.003, "loss": 4.135, "step": 4816 }, { "epoch": 0.04817, "grad_norm": 0.8274465563158391, "learning_rate": 0.003, "loss": 4.1101, "step": 4817 }, { "epoch": 0.04818, "grad_norm": 0.8182701860044233, "learning_rate": 0.003, "loss": 4.1239, "step": 4818 }, { "epoch": 0.04819, "grad_norm": 0.8458748903862201, "learning_rate": 0.003, "loss": 4.1283, "step": 4819 }, { "epoch": 0.0482, "grad_norm": 0.8155150451708626, "learning_rate": 0.003, "loss": 4.1167, "step": 4820 }, { "epoch": 0.04821, "grad_norm": 0.8777962638905318, "learning_rate": 0.003, "loss": 4.1466, "step": 4821 }, { "epoch": 0.04822, "grad_norm": 0.9431701355771726, "learning_rate": 0.003, "loss": 4.125, "step": 4822 }, { "epoch": 0.04823, "grad_norm": 0.9695687695883342, "learning_rate": 0.003, "loss": 4.1237, "step": 4823 }, { "epoch": 0.04824, "grad_norm": 0.991895345693445, "learning_rate": 0.003, "loss": 4.113, "step": 4824 }, { "epoch": 0.04825, "grad_norm": 1.0943797710534833, "learning_rate": 0.003, "loss": 4.1322, "step": 4825 }, { "epoch": 0.04826, "grad_norm": 0.8268466548342316, "learning_rate": 0.003, "loss": 4.1299, "step": 4826 }, { "epoch": 0.04827, "grad_norm": 0.8361639143145106, "learning_rate": 0.003, "loss": 4.1336, "step": 4827 }, { "epoch": 0.04828, "grad_norm": 1.0286267106339764, "learning_rate": 0.003, "loss": 4.1511, "step": 4828 }, { "epoch": 0.04829, "grad_norm": 1.159942400169441, "learning_rate": 0.003, "loss": 4.1364, "step": 4829 }, { "epoch": 0.0483, "grad_norm": 0.9372268357664759, "learning_rate": 0.003, "loss": 4.1427, "step": 4830 }, { "epoch": 0.04831, "grad_norm": 0.8980121041802035, "learning_rate": 0.003, "loss": 4.1082, "step": 4831 }, { "epoch": 0.04832, "grad_norm": 0.8916438529090394, "learning_rate": 0.003, "loss": 4.1312, "step": 4832 }, { "epoch": 0.04833, "grad_norm": 1.0001644468197317, "learning_rate": 0.003, "loss": 4.1421, "step": 4833 }, { "epoch": 0.04834, "grad_norm": 1.2415615713112236, "learning_rate": 0.003, "loss": 4.1197, "step": 4834 }, { "epoch": 0.04835, "grad_norm": 0.9209372163782303, "learning_rate": 0.003, "loss": 4.1495, "step": 4835 }, { "epoch": 0.04836, "grad_norm": 1.0240500494308544, "learning_rate": 0.003, "loss": 4.1217, "step": 4836 }, { "epoch": 0.04837, "grad_norm": 1.0801075970896188, "learning_rate": 0.003, "loss": 4.1502, "step": 4837 }, { "epoch": 0.04838, "grad_norm": 0.8556232959465542, "learning_rate": 0.003, "loss": 4.1533, "step": 4838 }, { "epoch": 0.04839, "grad_norm": 0.8048051834755471, "learning_rate": 0.003, "loss": 4.1369, "step": 4839 }, { "epoch": 0.0484, "grad_norm": 0.8367021532891096, "learning_rate": 0.003, "loss": 4.1415, "step": 4840 }, { "epoch": 0.04841, "grad_norm": 0.7607914132160744, "learning_rate": 0.003, "loss": 4.1369, "step": 4841 }, { "epoch": 0.04842, "grad_norm": 0.7830637400101697, "learning_rate": 0.003, "loss": 4.1503, "step": 4842 }, { "epoch": 0.04843, "grad_norm": 0.9114911514114503, "learning_rate": 0.003, "loss": 4.1092, "step": 4843 }, { "epoch": 0.04844, "grad_norm": 1.1708951262723284, "learning_rate": 0.003, "loss": 4.143, "step": 4844 }, { "epoch": 0.04845, "grad_norm": 1.0089311204905278, "learning_rate": 0.003, "loss": 4.1298, "step": 4845 }, { "epoch": 0.04846, "grad_norm": 1.0807920040383794, "learning_rate": 0.003, "loss": 4.1529, "step": 4846 }, { "epoch": 0.04847, "grad_norm": 0.8594401536252937, "learning_rate": 0.003, "loss": 4.1099, "step": 4847 }, { "epoch": 0.04848, "grad_norm": 0.7513172891372867, "learning_rate": 0.003, "loss": 4.1503, "step": 4848 }, { "epoch": 0.04849, "grad_norm": 0.751292803407457, "learning_rate": 0.003, "loss": 4.1402, "step": 4849 }, { "epoch": 0.0485, "grad_norm": 0.7959018388472837, "learning_rate": 0.003, "loss": 4.1293, "step": 4850 }, { "epoch": 0.04851, "grad_norm": 0.9962650900237886, "learning_rate": 0.003, "loss": 4.1456, "step": 4851 }, { "epoch": 0.04852, "grad_norm": 1.1834574873765151, "learning_rate": 0.003, "loss": 4.0888, "step": 4852 }, { "epoch": 0.04853, "grad_norm": 0.7889329019222011, "learning_rate": 0.003, "loss": 4.1405, "step": 4853 }, { "epoch": 0.04854, "grad_norm": 0.7364486677015084, "learning_rate": 0.003, "loss": 4.1334, "step": 4854 }, { "epoch": 0.04855, "grad_norm": 0.7363603615147029, "learning_rate": 0.003, "loss": 4.1409, "step": 4855 }, { "epoch": 0.04856, "grad_norm": 0.7141913003641922, "learning_rate": 0.003, "loss": 4.1373, "step": 4856 }, { "epoch": 0.04857, "grad_norm": 0.7697901500308619, "learning_rate": 0.003, "loss": 4.1512, "step": 4857 }, { "epoch": 0.04858, "grad_norm": 0.978947209936643, "learning_rate": 0.003, "loss": 4.1246, "step": 4858 }, { "epoch": 0.04859, "grad_norm": 1.0853261299941506, "learning_rate": 0.003, "loss": 4.164, "step": 4859 }, { "epoch": 0.0486, "grad_norm": 0.8322040486694344, "learning_rate": 0.003, "loss": 4.1406, "step": 4860 }, { "epoch": 0.04861, "grad_norm": 0.911479945440266, "learning_rate": 0.003, "loss": 4.1325, "step": 4861 }, { "epoch": 0.04862, "grad_norm": 0.9173241722504067, "learning_rate": 0.003, "loss": 4.1361, "step": 4862 }, { "epoch": 0.04863, "grad_norm": 0.8964490214389235, "learning_rate": 0.003, "loss": 4.1527, "step": 4863 }, { "epoch": 0.04864, "grad_norm": 0.972306148276167, "learning_rate": 0.003, "loss": 4.1292, "step": 4864 }, { "epoch": 0.04865, "grad_norm": 1.1291430700547054, "learning_rate": 0.003, "loss": 4.1559, "step": 4865 }, { "epoch": 0.04866, "grad_norm": 1.0797428982956703, "learning_rate": 0.003, "loss": 4.1598, "step": 4866 }, { "epoch": 0.04867, "grad_norm": 0.802691748949696, "learning_rate": 0.003, "loss": 4.1233, "step": 4867 }, { "epoch": 0.04868, "grad_norm": 0.6363552129783063, "learning_rate": 0.003, "loss": 4.1297, "step": 4868 }, { "epoch": 0.04869, "grad_norm": 0.8246675935867382, "learning_rate": 0.003, "loss": 4.1398, "step": 4869 }, { "epoch": 0.0487, "grad_norm": 0.9180468562537373, "learning_rate": 0.003, "loss": 4.1316, "step": 4870 }, { "epoch": 0.04871, "grad_norm": 1.101938319921805, "learning_rate": 0.003, "loss": 4.1561, "step": 4871 }, { "epoch": 0.04872, "grad_norm": 0.8727221393103543, "learning_rate": 0.003, "loss": 4.114, "step": 4872 }, { "epoch": 0.04873, "grad_norm": 0.9149687440430082, "learning_rate": 0.003, "loss": 4.1428, "step": 4873 }, { "epoch": 0.04874, "grad_norm": 0.9259144418955525, "learning_rate": 0.003, "loss": 4.1338, "step": 4874 }, { "epoch": 0.04875, "grad_norm": 0.8834881037341485, "learning_rate": 0.003, "loss": 4.1258, "step": 4875 }, { "epoch": 0.04876, "grad_norm": 0.933167953983561, "learning_rate": 0.003, "loss": 4.1323, "step": 4876 }, { "epoch": 0.04877, "grad_norm": 1.0482409044529428, "learning_rate": 0.003, "loss": 4.1334, "step": 4877 }, { "epoch": 0.04878, "grad_norm": 1.0364280416714626, "learning_rate": 0.003, "loss": 4.1173, "step": 4878 }, { "epoch": 0.04879, "grad_norm": 0.8888490541945427, "learning_rate": 0.003, "loss": 4.1087, "step": 4879 }, { "epoch": 0.0488, "grad_norm": 0.8177396614668453, "learning_rate": 0.003, "loss": 4.1219, "step": 4880 }, { "epoch": 0.04881, "grad_norm": 0.8123257391255282, "learning_rate": 0.003, "loss": 4.1167, "step": 4881 }, { "epoch": 0.04882, "grad_norm": 0.7975721516970518, "learning_rate": 0.003, "loss": 4.1213, "step": 4882 }, { "epoch": 0.04883, "grad_norm": 0.7955421903063317, "learning_rate": 0.003, "loss": 4.1169, "step": 4883 }, { "epoch": 0.04884, "grad_norm": 1.0189979080759815, "learning_rate": 0.003, "loss": 4.13, "step": 4884 }, { "epoch": 0.04885, "grad_norm": 1.2153408218196549, "learning_rate": 0.003, "loss": 4.1537, "step": 4885 }, { "epoch": 0.04886, "grad_norm": 0.7349864263114413, "learning_rate": 0.003, "loss": 4.1323, "step": 4886 }, { "epoch": 0.04887, "grad_norm": 0.7833103419829659, "learning_rate": 0.003, "loss": 4.1424, "step": 4887 }, { "epoch": 0.04888, "grad_norm": 0.7975450688309594, "learning_rate": 0.003, "loss": 4.1448, "step": 4888 }, { "epoch": 0.04889, "grad_norm": 1.1954327023621745, "learning_rate": 0.003, "loss": 4.1319, "step": 4889 }, { "epoch": 0.0489, "grad_norm": 1.2543113698602726, "learning_rate": 0.003, "loss": 4.1457, "step": 4890 }, { "epoch": 0.04891, "grad_norm": 0.8947016132538613, "learning_rate": 0.003, "loss": 4.1489, "step": 4891 }, { "epoch": 0.04892, "grad_norm": 0.8348670424043885, "learning_rate": 0.003, "loss": 4.1441, "step": 4892 }, { "epoch": 0.04893, "grad_norm": 0.8995677954784294, "learning_rate": 0.003, "loss": 4.1419, "step": 4893 }, { "epoch": 0.04894, "grad_norm": 1.4114151388613976, "learning_rate": 0.003, "loss": 4.1623, "step": 4894 }, { "epoch": 0.04895, "grad_norm": 1.1554803750152747, "learning_rate": 0.003, "loss": 4.134, "step": 4895 }, { "epoch": 0.04896, "grad_norm": 0.9507103715274245, "learning_rate": 0.003, "loss": 4.1441, "step": 4896 }, { "epoch": 0.04897, "grad_norm": 0.906957283921157, "learning_rate": 0.003, "loss": 4.151, "step": 4897 }, { "epoch": 0.04898, "grad_norm": 0.8944074460888909, "learning_rate": 0.003, "loss": 4.1402, "step": 4898 }, { "epoch": 0.04899, "grad_norm": 0.8638995775443565, "learning_rate": 0.003, "loss": 4.1265, "step": 4899 }, { "epoch": 0.049, "grad_norm": 0.8297845635886839, "learning_rate": 0.003, "loss": 4.1532, "step": 4900 }, { "epoch": 0.04901, "grad_norm": 0.9248298803156414, "learning_rate": 0.003, "loss": 4.1148, "step": 4901 }, { "epoch": 0.04902, "grad_norm": 1.0368926529413733, "learning_rate": 0.003, "loss": 4.1541, "step": 4902 }, { "epoch": 0.04903, "grad_norm": 0.9155640413660329, "learning_rate": 0.003, "loss": 4.1516, "step": 4903 }, { "epoch": 0.04904, "grad_norm": 0.8052932601979055, "learning_rate": 0.003, "loss": 4.1131, "step": 4904 }, { "epoch": 0.04905, "grad_norm": 0.8694803464220409, "learning_rate": 0.003, "loss": 4.1189, "step": 4905 }, { "epoch": 0.04906, "grad_norm": 1.0292003158296992, "learning_rate": 0.003, "loss": 4.1329, "step": 4906 }, { "epoch": 0.04907, "grad_norm": 1.080150716503704, "learning_rate": 0.003, "loss": 4.1175, "step": 4907 }, { "epoch": 0.04908, "grad_norm": 0.9036715675717701, "learning_rate": 0.003, "loss": 4.1285, "step": 4908 }, { "epoch": 0.04909, "grad_norm": 0.8807978117119255, "learning_rate": 0.003, "loss": 4.1342, "step": 4909 }, { "epoch": 0.0491, "grad_norm": 0.8458247064272691, "learning_rate": 0.003, "loss": 4.1265, "step": 4910 }, { "epoch": 0.04911, "grad_norm": 0.8766008605366878, "learning_rate": 0.003, "loss": 4.1228, "step": 4911 }, { "epoch": 0.04912, "grad_norm": 0.8331605702108348, "learning_rate": 0.003, "loss": 4.0972, "step": 4912 }, { "epoch": 0.04913, "grad_norm": 0.8360485345313021, "learning_rate": 0.003, "loss": 4.1235, "step": 4913 }, { "epoch": 0.04914, "grad_norm": 0.7899851119072792, "learning_rate": 0.003, "loss": 4.1276, "step": 4914 }, { "epoch": 0.04915, "grad_norm": 0.9630478684955635, "learning_rate": 0.003, "loss": 4.1381, "step": 4915 }, { "epoch": 0.04916, "grad_norm": 0.9934301086816488, "learning_rate": 0.003, "loss": 4.1375, "step": 4916 }, { "epoch": 0.04917, "grad_norm": 1.3041941957357555, "learning_rate": 0.003, "loss": 4.102, "step": 4917 }, { "epoch": 0.04918, "grad_norm": 0.7891044278279551, "learning_rate": 0.003, "loss": 4.1302, "step": 4918 }, { "epoch": 0.04919, "grad_norm": 0.866921940746236, "learning_rate": 0.003, "loss": 4.1365, "step": 4919 }, { "epoch": 0.0492, "grad_norm": 0.9289187382638477, "learning_rate": 0.003, "loss": 4.1354, "step": 4920 }, { "epoch": 0.04921, "grad_norm": 0.999976308146298, "learning_rate": 0.003, "loss": 4.1547, "step": 4921 }, { "epoch": 0.04922, "grad_norm": 1.124045561099308, "learning_rate": 0.003, "loss": 4.1365, "step": 4922 }, { "epoch": 0.04923, "grad_norm": 0.9763261912117048, "learning_rate": 0.003, "loss": 4.1117, "step": 4923 }, { "epoch": 0.04924, "grad_norm": 0.8510875610500769, "learning_rate": 0.003, "loss": 4.1611, "step": 4924 }, { "epoch": 0.04925, "grad_norm": 0.8704283410665644, "learning_rate": 0.003, "loss": 4.1127, "step": 4925 }, { "epoch": 0.04926, "grad_norm": 0.9383663644735687, "learning_rate": 0.003, "loss": 4.1169, "step": 4926 }, { "epoch": 0.04927, "grad_norm": 1.1427920258058981, "learning_rate": 0.003, "loss": 4.1548, "step": 4927 }, { "epoch": 0.04928, "grad_norm": 1.0626423088763775, "learning_rate": 0.003, "loss": 4.1151, "step": 4928 }, { "epoch": 0.04929, "grad_norm": 1.0837552807904482, "learning_rate": 0.003, "loss": 4.1353, "step": 4929 }, { "epoch": 0.0493, "grad_norm": 0.9600959111967431, "learning_rate": 0.003, "loss": 4.1253, "step": 4930 }, { "epoch": 0.04931, "grad_norm": 0.9442147578078418, "learning_rate": 0.003, "loss": 4.1055, "step": 4931 }, { "epoch": 0.04932, "grad_norm": 0.8723803650613164, "learning_rate": 0.003, "loss": 4.1546, "step": 4932 }, { "epoch": 0.04933, "grad_norm": 0.9126114021558055, "learning_rate": 0.003, "loss": 4.1313, "step": 4933 }, { "epoch": 0.04934, "grad_norm": 0.7831133724300265, "learning_rate": 0.003, "loss": 4.1594, "step": 4934 }, { "epoch": 0.04935, "grad_norm": 0.8193838082678253, "learning_rate": 0.003, "loss": 4.1556, "step": 4935 }, { "epoch": 0.04936, "grad_norm": 0.9542250458968273, "learning_rate": 0.003, "loss": 4.1312, "step": 4936 }, { "epoch": 0.04937, "grad_norm": 1.0102642975787324, "learning_rate": 0.003, "loss": 4.1293, "step": 4937 }, { "epoch": 0.04938, "grad_norm": 1.1822748470538103, "learning_rate": 0.003, "loss": 4.1433, "step": 4938 }, { "epoch": 0.04939, "grad_norm": 0.8813626480520453, "learning_rate": 0.003, "loss": 4.1465, "step": 4939 }, { "epoch": 0.0494, "grad_norm": 0.9354601920016478, "learning_rate": 0.003, "loss": 4.1534, "step": 4940 }, { "epoch": 0.04941, "grad_norm": 0.8531796298952382, "learning_rate": 0.003, "loss": 4.1179, "step": 4941 }, { "epoch": 0.04942, "grad_norm": 0.8648680617674837, "learning_rate": 0.003, "loss": 4.1322, "step": 4942 }, { "epoch": 0.04943, "grad_norm": 0.8572189947286999, "learning_rate": 0.003, "loss": 4.1537, "step": 4943 }, { "epoch": 0.04944, "grad_norm": 1.0596615074274864, "learning_rate": 0.003, "loss": 4.1644, "step": 4944 }, { "epoch": 0.04945, "grad_norm": 0.929833863296995, "learning_rate": 0.003, "loss": 4.1507, "step": 4945 }, { "epoch": 0.04946, "grad_norm": 0.9625930540154848, "learning_rate": 0.003, "loss": 4.1756, "step": 4946 }, { "epoch": 0.04947, "grad_norm": 0.9938577466365767, "learning_rate": 0.003, "loss": 4.1186, "step": 4947 }, { "epoch": 0.04948, "grad_norm": 1.0582956690263399, "learning_rate": 0.003, "loss": 4.1396, "step": 4948 }, { "epoch": 0.04949, "grad_norm": 1.0609539105003243, "learning_rate": 0.003, "loss": 4.1426, "step": 4949 }, { "epoch": 0.0495, "grad_norm": 0.8573405852337684, "learning_rate": 0.003, "loss": 4.1041, "step": 4950 }, { "epoch": 0.04951, "grad_norm": 0.9045716976759907, "learning_rate": 0.003, "loss": 4.1119, "step": 4951 }, { "epoch": 0.04952, "grad_norm": 0.9697520441480982, "learning_rate": 0.003, "loss": 4.1497, "step": 4952 }, { "epoch": 0.04953, "grad_norm": 1.0183145355055803, "learning_rate": 0.003, "loss": 4.1342, "step": 4953 }, { "epoch": 0.04954, "grad_norm": 0.9741338701993631, "learning_rate": 0.003, "loss": 4.158, "step": 4954 }, { "epoch": 0.04955, "grad_norm": 1.1025167524243962, "learning_rate": 0.003, "loss": 4.137, "step": 4955 }, { "epoch": 0.04956, "grad_norm": 0.8870510145238134, "learning_rate": 0.003, "loss": 4.1472, "step": 4956 }, { "epoch": 0.04957, "grad_norm": 0.7825537450099663, "learning_rate": 0.003, "loss": 4.1136, "step": 4957 }, { "epoch": 0.04958, "grad_norm": 0.7124964191360044, "learning_rate": 0.003, "loss": 4.1296, "step": 4958 }, { "epoch": 0.04959, "grad_norm": 0.7590903052152222, "learning_rate": 0.003, "loss": 4.1322, "step": 4959 }, { "epoch": 0.0496, "grad_norm": 0.8759268385091953, "learning_rate": 0.003, "loss": 4.1038, "step": 4960 }, { "epoch": 0.04961, "grad_norm": 1.0986531804493005, "learning_rate": 0.003, "loss": 4.1147, "step": 4961 }, { "epoch": 0.04962, "grad_norm": 0.9519760561283285, "learning_rate": 0.003, "loss": 4.1294, "step": 4962 }, { "epoch": 0.04963, "grad_norm": 0.8654145678491735, "learning_rate": 0.003, "loss": 4.1644, "step": 4963 }, { "epoch": 0.04964, "grad_norm": 0.7211388340825737, "learning_rate": 0.003, "loss": 4.1328, "step": 4964 }, { "epoch": 0.04965, "grad_norm": 0.8630218156973655, "learning_rate": 0.003, "loss": 4.1346, "step": 4965 }, { "epoch": 0.04966, "grad_norm": 1.0251817150393079, "learning_rate": 0.003, "loss": 4.1096, "step": 4966 }, { "epoch": 0.04967, "grad_norm": 1.093600352302926, "learning_rate": 0.003, "loss": 4.1381, "step": 4967 }, { "epoch": 0.04968, "grad_norm": 0.9143668717946363, "learning_rate": 0.003, "loss": 4.1578, "step": 4968 }, { "epoch": 0.04969, "grad_norm": 0.8211912863315765, "learning_rate": 0.003, "loss": 4.1303, "step": 4969 }, { "epoch": 0.0497, "grad_norm": 0.8134944463392609, "learning_rate": 0.003, "loss": 4.1253, "step": 4970 }, { "epoch": 0.04971, "grad_norm": 0.761104211586214, "learning_rate": 0.003, "loss": 4.1424, "step": 4971 }, { "epoch": 0.04972, "grad_norm": 0.8740554584097706, "learning_rate": 0.003, "loss": 4.119, "step": 4972 }, { "epoch": 0.04973, "grad_norm": 0.8620362754914105, "learning_rate": 0.003, "loss": 4.1698, "step": 4973 }, { "epoch": 0.04974, "grad_norm": 0.9442564618069658, "learning_rate": 0.003, "loss": 4.1306, "step": 4974 }, { "epoch": 0.04975, "grad_norm": 0.9778639609000918, "learning_rate": 0.003, "loss": 4.1336, "step": 4975 }, { "epoch": 0.04976, "grad_norm": 1.0420631135988534, "learning_rate": 0.003, "loss": 4.1342, "step": 4976 }, { "epoch": 0.04977, "grad_norm": 1.0602564274040003, "learning_rate": 0.003, "loss": 4.146, "step": 4977 }, { "epoch": 0.04978, "grad_norm": 0.9052204217537351, "learning_rate": 0.003, "loss": 4.139, "step": 4978 }, { "epoch": 0.04979, "grad_norm": 0.9947293611924701, "learning_rate": 0.003, "loss": 4.1496, "step": 4979 }, { "epoch": 0.0498, "grad_norm": 1.0091251108504844, "learning_rate": 0.003, "loss": 4.1407, "step": 4980 }, { "epoch": 0.04981, "grad_norm": 1.1609493803122968, "learning_rate": 0.003, "loss": 4.1227, "step": 4981 }, { "epoch": 0.04982, "grad_norm": 1.122465102066747, "learning_rate": 0.003, "loss": 4.1467, "step": 4982 }, { "epoch": 0.04983, "grad_norm": 1.3048056173322304, "learning_rate": 0.003, "loss": 4.1121, "step": 4983 }, { "epoch": 0.04984, "grad_norm": 0.9839501695087924, "learning_rate": 0.003, "loss": 4.1085, "step": 4984 }, { "epoch": 0.04985, "grad_norm": 1.0551935971857649, "learning_rate": 0.003, "loss": 4.1428, "step": 4985 }, { "epoch": 0.04986, "grad_norm": 1.0243284821155656, "learning_rate": 0.003, "loss": 4.1587, "step": 4986 }, { "epoch": 0.04987, "grad_norm": 0.7958713265484796, "learning_rate": 0.003, "loss": 4.1138, "step": 4987 }, { "epoch": 0.04988, "grad_norm": 0.7040759247040429, "learning_rate": 0.003, "loss": 4.1438, "step": 4988 }, { "epoch": 0.04989, "grad_norm": 0.8411275085142221, "learning_rate": 0.003, "loss": 4.1491, "step": 4989 }, { "epoch": 0.0499, "grad_norm": 0.8415827540961842, "learning_rate": 0.003, "loss": 4.124, "step": 4990 }, { "epoch": 0.04991, "grad_norm": 0.7469609887556586, "learning_rate": 0.003, "loss": 4.1128, "step": 4991 }, { "epoch": 0.04992, "grad_norm": 0.8249372889306039, "learning_rate": 0.003, "loss": 4.0976, "step": 4992 }, { "epoch": 0.04993, "grad_norm": 1.0566657783178806, "learning_rate": 0.003, "loss": 4.1285, "step": 4993 }, { "epoch": 0.04994, "grad_norm": 1.0996011403959378, "learning_rate": 0.003, "loss": 4.1196, "step": 4994 }, { "epoch": 0.04995, "grad_norm": 0.9432281244618136, "learning_rate": 0.003, "loss": 4.1376, "step": 4995 }, { "epoch": 0.04996, "grad_norm": 0.8902334825733927, "learning_rate": 0.003, "loss": 4.119, "step": 4996 }, { "epoch": 0.04997, "grad_norm": 0.7781105185125646, "learning_rate": 0.003, "loss": 4.1289, "step": 4997 }, { "epoch": 0.04998, "grad_norm": 0.6987900025248667, "learning_rate": 0.003, "loss": 4.1524, "step": 4998 }, { "epoch": 0.04999, "grad_norm": 0.740322084305092, "learning_rate": 0.003, "loss": 4.1116, "step": 4999 }, { "epoch": 0.05, "grad_norm": 0.8975134700731652, "learning_rate": 0.003, "loss": 4.158, "step": 5000 }, { "epoch": 0.05001, "grad_norm": 1.1125421426369673, "learning_rate": 0.003, "loss": 4.1551, "step": 5001 }, { "epoch": 0.05002, "grad_norm": 1.268946353977916, "learning_rate": 0.003, "loss": 4.1379, "step": 5002 }, { "epoch": 0.05003, "grad_norm": 0.8237186256695063, "learning_rate": 0.003, "loss": 4.1417, "step": 5003 }, { "epoch": 0.05004, "grad_norm": 0.8676428422328399, "learning_rate": 0.003, "loss": 4.1254, "step": 5004 }, { "epoch": 0.05005, "grad_norm": 0.7746806700851315, "learning_rate": 0.003, "loss": 4.1456, "step": 5005 }, { "epoch": 0.05006, "grad_norm": 0.8409396903439786, "learning_rate": 0.003, "loss": 4.1247, "step": 5006 }, { "epoch": 0.05007, "grad_norm": 0.9473102489321485, "learning_rate": 0.003, "loss": 4.1431, "step": 5007 }, { "epoch": 0.05008, "grad_norm": 1.100401498879317, "learning_rate": 0.003, "loss": 4.1132, "step": 5008 }, { "epoch": 0.05009, "grad_norm": 0.9798182802542361, "learning_rate": 0.003, "loss": 4.1205, "step": 5009 }, { "epoch": 0.0501, "grad_norm": 1.0935878325404964, "learning_rate": 0.003, "loss": 4.1384, "step": 5010 }, { "epoch": 0.05011, "grad_norm": 1.0191676731522699, "learning_rate": 0.003, "loss": 4.1388, "step": 5011 }, { "epoch": 0.05012, "grad_norm": 1.1270362384105346, "learning_rate": 0.003, "loss": 4.1333, "step": 5012 }, { "epoch": 0.05013, "grad_norm": 1.018439521580052, "learning_rate": 0.003, "loss": 4.1362, "step": 5013 }, { "epoch": 0.05014, "grad_norm": 1.0624365348167546, "learning_rate": 0.003, "loss": 4.1137, "step": 5014 }, { "epoch": 0.05015, "grad_norm": 0.9411876813426826, "learning_rate": 0.003, "loss": 4.1499, "step": 5015 }, { "epoch": 0.05016, "grad_norm": 0.9736253716196595, "learning_rate": 0.003, "loss": 4.1334, "step": 5016 }, { "epoch": 0.05017, "grad_norm": 1.1100942667063913, "learning_rate": 0.003, "loss": 4.1527, "step": 5017 }, { "epoch": 0.05018, "grad_norm": 1.0084783515734652, "learning_rate": 0.003, "loss": 4.147, "step": 5018 }, { "epoch": 0.05019, "grad_norm": 0.9759151660471137, "learning_rate": 0.003, "loss": 4.1635, "step": 5019 }, { "epoch": 0.0502, "grad_norm": 0.9723722102677356, "learning_rate": 0.003, "loss": 4.1294, "step": 5020 }, { "epoch": 0.05021, "grad_norm": 0.9049845415635082, "learning_rate": 0.003, "loss": 4.1347, "step": 5021 }, { "epoch": 0.05022, "grad_norm": 0.9107217668273613, "learning_rate": 0.003, "loss": 4.1337, "step": 5022 }, { "epoch": 0.05023, "grad_norm": 1.0887083292657602, "learning_rate": 0.003, "loss": 4.1618, "step": 5023 }, { "epoch": 0.05024, "grad_norm": 1.1150744057770852, "learning_rate": 0.003, "loss": 4.1207, "step": 5024 }, { "epoch": 0.05025, "grad_norm": 1.0031676743967801, "learning_rate": 0.003, "loss": 4.1564, "step": 5025 }, { "epoch": 0.05026, "grad_norm": 0.9402890020375552, "learning_rate": 0.003, "loss": 4.1302, "step": 5026 }, { "epoch": 0.05027, "grad_norm": 0.8731649649321198, "learning_rate": 0.003, "loss": 4.1479, "step": 5027 }, { "epoch": 0.05028, "grad_norm": 0.7951256123529497, "learning_rate": 0.003, "loss": 4.1304, "step": 5028 }, { "epoch": 0.05029, "grad_norm": 0.8539293439248877, "learning_rate": 0.003, "loss": 4.1424, "step": 5029 }, { "epoch": 0.0503, "grad_norm": 0.8281260610684518, "learning_rate": 0.003, "loss": 4.1119, "step": 5030 }, { "epoch": 0.05031, "grad_norm": 0.817963787364258, "learning_rate": 0.003, "loss": 4.1307, "step": 5031 }, { "epoch": 0.05032, "grad_norm": 0.9512198937363134, "learning_rate": 0.003, "loss": 4.1351, "step": 5032 }, { "epoch": 0.05033, "grad_norm": 0.9963022197559578, "learning_rate": 0.003, "loss": 4.1135, "step": 5033 }, { "epoch": 0.05034, "grad_norm": 1.104614966345552, "learning_rate": 0.003, "loss": 4.1216, "step": 5034 }, { "epoch": 0.05035, "grad_norm": 1.0983448584739037, "learning_rate": 0.003, "loss": 4.1412, "step": 5035 }, { "epoch": 0.05036, "grad_norm": 0.9061350341495388, "learning_rate": 0.003, "loss": 4.1213, "step": 5036 }, { "epoch": 0.05037, "grad_norm": 0.9464966069987577, "learning_rate": 0.003, "loss": 4.1164, "step": 5037 }, { "epoch": 0.05038, "grad_norm": 1.0201982381782406, "learning_rate": 0.003, "loss": 4.1289, "step": 5038 }, { "epoch": 0.05039, "grad_norm": 1.0969503993497207, "learning_rate": 0.003, "loss": 4.1453, "step": 5039 }, { "epoch": 0.0504, "grad_norm": 0.902563340773235, "learning_rate": 0.003, "loss": 4.139, "step": 5040 }, { "epoch": 0.05041, "grad_norm": 0.8273610612179048, "learning_rate": 0.003, "loss": 4.1395, "step": 5041 }, { "epoch": 0.05042, "grad_norm": 0.9503016386214178, "learning_rate": 0.003, "loss": 4.1322, "step": 5042 }, { "epoch": 0.05043, "grad_norm": 0.9388384995642503, "learning_rate": 0.003, "loss": 4.1299, "step": 5043 }, { "epoch": 0.05044, "grad_norm": 0.8905833491260514, "learning_rate": 0.003, "loss": 4.1461, "step": 5044 }, { "epoch": 0.05045, "grad_norm": 0.8430993221174582, "learning_rate": 0.003, "loss": 4.1603, "step": 5045 }, { "epoch": 0.05046, "grad_norm": 0.8169585705346127, "learning_rate": 0.003, "loss": 4.1223, "step": 5046 }, { "epoch": 0.05047, "grad_norm": 0.95893686498772, "learning_rate": 0.003, "loss": 4.134, "step": 5047 }, { "epoch": 0.05048, "grad_norm": 1.3360304246137076, "learning_rate": 0.003, "loss": 4.1378, "step": 5048 }, { "epoch": 0.05049, "grad_norm": 0.951954395506191, "learning_rate": 0.003, "loss": 4.0947, "step": 5049 }, { "epoch": 0.0505, "grad_norm": 0.8883334202277473, "learning_rate": 0.003, "loss": 4.1824, "step": 5050 }, { "epoch": 0.05051, "grad_norm": 0.849437175294796, "learning_rate": 0.003, "loss": 4.1422, "step": 5051 }, { "epoch": 0.05052, "grad_norm": 0.907407381853802, "learning_rate": 0.003, "loss": 4.1161, "step": 5052 }, { "epoch": 0.05053, "grad_norm": 0.8362515125237234, "learning_rate": 0.003, "loss": 4.0889, "step": 5053 }, { "epoch": 0.05054, "grad_norm": 0.6457242952400993, "learning_rate": 0.003, "loss": 4.1409, "step": 5054 }, { "epoch": 0.05055, "grad_norm": 0.6640040755007646, "learning_rate": 0.003, "loss": 4.1174, "step": 5055 }, { "epoch": 0.05056, "grad_norm": 0.730354009820354, "learning_rate": 0.003, "loss": 4.1334, "step": 5056 }, { "epoch": 0.05057, "grad_norm": 0.9095854829866085, "learning_rate": 0.003, "loss": 4.1397, "step": 5057 }, { "epoch": 0.05058, "grad_norm": 1.3085718232187313, "learning_rate": 0.003, "loss": 4.1505, "step": 5058 }, { "epoch": 0.05059, "grad_norm": 0.8208558372297785, "learning_rate": 0.003, "loss": 4.1067, "step": 5059 }, { "epoch": 0.0506, "grad_norm": 0.8287539156196918, "learning_rate": 0.003, "loss": 4.1408, "step": 5060 }, { "epoch": 0.05061, "grad_norm": 0.9396605720141616, "learning_rate": 0.003, "loss": 4.1213, "step": 5061 }, { "epoch": 0.05062, "grad_norm": 1.080983860701126, "learning_rate": 0.003, "loss": 4.1121, "step": 5062 }, { "epoch": 0.05063, "grad_norm": 0.9569977418540291, "learning_rate": 0.003, "loss": 4.0982, "step": 5063 }, { "epoch": 0.05064, "grad_norm": 1.075568604431901, "learning_rate": 0.003, "loss": 4.1431, "step": 5064 }, { "epoch": 0.05065, "grad_norm": 0.9243313906929234, "learning_rate": 0.003, "loss": 4.1267, "step": 5065 }, { "epoch": 0.05066, "grad_norm": 0.9374319059631955, "learning_rate": 0.003, "loss": 4.1144, "step": 5066 }, { "epoch": 0.05067, "grad_norm": 0.8481769015446154, "learning_rate": 0.003, "loss": 4.1218, "step": 5067 }, { "epoch": 0.05068, "grad_norm": 0.9303443535141828, "learning_rate": 0.003, "loss": 4.1451, "step": 5068 }, { "epoch": 0.05069, "grad_norm": 0.9770075079961184, "learning_rate": 0.003, "loss": 4.1358, "step": 5069 }, { "epoch": 0.0507, "grad_norm": 1.1996863760884935, "learning_rate": 0.003, "loss": 4.1571, "step": 5070 }, { "epoch": 0.05071, "grad_norm": 0.7185648581445719, "learning_rate": 0.003, "loss": 4.1343, "step": 5071 }, { "epoch": 0.05072, "grad_norm": 0.7935232361406919, "learning_rate": 0.003, "loss": 4.1097, "step": 5072 }, { "epoch": 0.05073, "grad_norm": 0.8603762101329337, "learning_rate": 0.003, "loss": 4.1147, "step": 5073 }, { "epoch": 0.05074, "grad_norm": 0.8776333809260097, "learning_rate": 0.003, "loss": 4.1101, "step": 5074 }, { "epoch": 0.05075, "grad_norm": 1.134189424557927, "learning_rate": 0.003, "loss": 4.1354, "step": 5075 }, { "epoch": 0.05076, "grad_norm": 1.1348672821872274, "learning_rate": 0.003, "loss": 4.1487, "step": 5076 }, { "epoch": 0.05077, "grad_norm": 0.9941313632381431, "learning_rate": 0.003, "loss": 4.1532, "step": 5077 }, { "epoch": 0.05078, "grad_norm": 0.894813842371774, "learning_rate": 0.003, "loss": 4.1412, "step": 5078 }, { "epoch": 0.05079, "grad_norm": 0.9864042658471017, "learning_rate": 0.003, "loss": 4.1151, "step": 5079 }, { "epoch": 0.0508, "grad_norm": 1.2211066026928066, "learning_rate": 0.003, "loss": 4.1176, "step": 5080 }, { "epoch": 0.05081, "grad_norm": 0.952087759769159, "learning_rate": 0.003, "loss": 4.1337, "step": 5081 }, { "epoch": 0.05082, "grad_norm": 1.0044149163525906, "learning_rate": 0.003, "loss": 4.143, "step": 5082 }, { "epoch": 0.05083, "grad_norm": 1.043147668555842, "learning_rate": 0.003, "loss": 4.1112, "step": 5083 }, { "epoch": 0.05084, "grad_norm": 0.916074260393084, "learning_rate": 0.003, "loss": 4.1253, "step": 5084 }, { "epoch": 0.05085, "grad_norm": 0.9863563460723465, "learning_rate": 0.003, "loss": 4.0964, "step": 5085 }, { "epoch": 0.05086, "grad_norm": 0.9526080378517822, "learning_rate": 0.003, "loss": 4.1059, "step": 5086 }, { "epoch": 0.05087, "grad_norm": 0.8954841316797875, "learning_rate": 0.003, "loss": 4.127, "step": 5087 }, { "epoch": 0.05088, "grad_norm": 0.6865090658283284, "learning_rate": 0.003, "loss": 4.1224, "step": 5088 }, { "epoch": 0.05089, "grad_norm": 0.7613574075539622, "learning_rate": 0.003, "loss": 4.1179, "step": 5089 }, { "epoch": 0.0509, "grad_norm": 0.8375881612111975, "learning_rate": 0.003, "loss": 4.1562, "step": 5090 }, { "epoch": 0.05091, "grad_norm": 1.088984983029717, "learning_rate": 0.003, "loss": 4.123, "step": 5091 }, { "epoch": 0.05092, "grad_norm": 0.9537871650339408, "learning_rate": 0.003, "loss": 4.1231, "step": 5092 }, { "epoch": 0.05093, "grad_norm": 0.8778865142910927, "learning_rate": 0.003, "loss": 4.1352, "step": 5093 }, { "epoch": 0.05094, "grad_norm": 0.8245219632382089, "learning_rate": 0.003, "loss": 4.0965, "step": 5094 }, { "epoch": 0.05095, "grad_norm": 0.7511141744389324, "learning_rate": 0.003, "loss": 4.1313, "step": 5095 }, { "epoch": 0.05096, "grad_norm": 0.7847143162436225, "learning_rate": 0.003, "loss": 4.1173, "step": 5096 }, { "epoch": 0.05097, "grad_norm": 0.8122517942129055, "learning_rate": 0.003, "loss": 4.1148, "step": 5097 }, { "epoch": 0.05098, "grad_norm": 0.7795564526148553, "learning_rate": 0.003, "loss": 4.1083, "step": 5098 }, { "epoch": 0.05099, "grad_norm": 0.685051675840029, "learning_rate": 0.003, "loss": 4.0823, "step": 5099 }, { "epoch": 0.051, "grad_norm": 0.7357655109108967, "learning_rate": 0.003, "loss": 4.1049, "step": 5100 }, { "epoch": 0.05101, "grad_norm": 0.7366959145534409, "learning_rate": 0.003, "loss": 4.14, "step": 5101 }, { "epoch": 0.05102, "grad_norm": 0.7876273112899445, "learning_rate": 0.003, "loss": 4.12, "step": 5102 }, { "epoch": 0.05103, "grad_norm": 0.8800742283306661, "learning_rate": 0.003, "loss": 4.134, "step": 5103 }, { "epoch": 0.05104, "grad_norm": 0.8715423090080415, "learning_rate": 0.003, "loss": 4.1156, "step": 5104 }, { "epoch": 0.05105, "grad_norm": 0.8824452296809646, "learning_rate": 0.003, "loss": 4.1041, "step": 5105 }, { "epoch": 0.05106, "grad_norm": 1.1209780196152646, "learning_rate": 0.003, "loss": 4.0869, "step": 5106 }, { "epoch": 0.05107, "grad_norm": 1.1399593567055775, "learning_rate": 0.003, "loss": 4.115, "step": 5107 }, { "epoch": 0.05108, "grad_norm": 1.4095394891532904, "learning_rate": 0.003, "loss": 4.1698, "step": 5108 }, { "epoch": 0.05109, "grad_norm": 0.7738170346317165, "learning_rate": 0.003, "loss": 4.096, "step": 5109 }, { "epoch": 0.0511, "grad_norm": 0.6999432461866981, "learning_rate": 0.003, "loss": 4.1324, "step": 5110 }, { "epoch": 0.05111, "grad_norm": 0.6582926413773714, "learning_rate": 0.003, "loss": 4.1431, "step": 5111 }, { "epoch": 0.05112, "grad_norm": 0.7551033740188957, "learning_rate": 0.003, "loss": 4.128, "step": 5112 }, { "epoch": 0.05113, "grad_norm": 1.1477812623559516, "learning_rate": 0.003, "loss": 4.1255, "step": 5113 }, { "epoch": 0.05114, "grad_norm": 1.067864380242395, "learning_rate": 0.003, "loss": 4.1017, "step": 5114 }, { "epoch": 0.05115, "grad_norm": 0.9300669029542773, "learning_rate": 0.003, "loss": 4.1297, "step": 5115 }, { "epoch": 0.05116, "grad_norm": 0.9229839891475532, "learning_rate": 0.003, "loss": 4.1356, "step": 5116 }, { "epoch": 0.05117, "grad_norm": 0.9776019657145362, "learning_rate": 0.003, "loss": 4.1199, "step": 5117 }, { "epoch": 0.05118, "grad_norm": 1.0187076693236539, "learning_rate": 0.003, "loss": 4.1208, "step": 5118 }, { "epoch": 0.05119, "grad_norm": 1.1898894115527578, "learning_rate": 0.003, "loss": 4.11, "step": 5119 }, { "epoch": 0.0512, "grad_norm": 0.7633336883976507, "learning_rate": 0.003, "loss": 4.0918, "step": 5120 }, { "epoch": 0.05121, "grad_norm": 0.8471498082951939, "learning_rate": 0.003, "loss": 4.1139, "step": 5121 }, { "epoch": 0.05122, "grad_norm": 0.9274839274688383, "learning_rate": 0.003, "loss": 4.1026, "step": 5122 }, { "epoch": 0.05123, "grad_norm": 1.1909929556885597, "learning_rate": 0.003, "loss": 4.1243, "step": 5123 }, { "epoch": 0.05124, "grad_norm": 0.815679730561404, "learning_rate": 0.003, "loss": 4.1565, "step": 5124 }, { "epoch": 0.05125, "grad_norm": 0.6789786360542248, "learning_rate": 0.003, "loss": 4.1191, "step": 5125 }, { "epoch": 0.05126, "grad_norm": 0.7440665492798663, "learning_rate": 0.003, "loss": 4.1148, "step": 5126 }, { "epoch": 0.05127, "grad_norm": 1.024011999747007, "learning_rate": 0.003, "loss": 4.1211, "step": 5127 }, { "epoch": 0.05128, "grad_norm": 1.3782586129142689, "learning_rate": 0.003, "loss": 4.1481, "step": 5128 }, { "epoch": 0.05129, "grad_norm": 0.779693198261867, "learning_rate": 0.003, "loss": 4.1056, "step": 5129 }, { "epoch": 0.0513, "grad_norm": 0.6550835083023576, "learning_rate": 0.003, "loss": 4.0883, "step": 5130 }, { "epoch": 0.05131, "grad_norm": 0.7096202778600647, "learning_rate": 0.003, "loss": 4.1263, "step": 5131 }, { "epoch": 0.05132, "grad_norm": 0.8254712872773159, "learning_rate": 0.003, "loss": 4.1147, "step": 5132 }, { "epoch": 0.05133, "grad_norm": 0.823333279464697, "learning_rate": 0.003, "loss": 4.124, "step": 5133 }, { "epoch": 0.05134, "grad_norm": 0.77855502466439, "learning_rate": 0.003, "loss": 4.1339, "step": 5134 }, { "epoch": 0.05135, "grad_norm": 0.8870106498581014, "learning_rate": 0.003, "loss": 4.1069, "step": 5135 }, { "epoch": 0.05136, "grad_norm": 1.1069868718885745, "learning_rate": 0.003, "loss": 4.098, "step": 5136 }, { "epoch": 0.05137, "grad_norm": 1.055928538799243, "learning_rate": 0.003, "loss": 4.1421, "step": 5137 }, { "epoch": 0.05138, "grad_norm": 1.2420968795396852, "learning_rate": 0.003, "loss": 4.1395, "step": 5138 }, { "epoch": 0.05139, "grad_norm": 1.1060392587125027, "learning_rate": 0.003, "loss": 4.129, "step": 5139 }, { "epoch": 0.0514, "grad_norm": 1.084456335779527, "learning_rate": 0.003, "loss": 4.1278, "step": 5140 }, { "epoch": 0.05141, "grad_norm": 0.8635486301225069, "learning_rate": 0.003, "loss": 4.1747, "step": 5141 }, { "epoch": 0.05142, "grad_norm": 0.8012211497201878, "learning_rate": 0.003, "loss": 4.1179, "step": 5142 }, { "epoch": 0.05143, "grad_norm": 0.7460263516853637, "learning_rate": 0.003, "loss": 4.12, "step": 5143 }, { "epoch": 0.05144, "grad_norm": 0.8599506765736857, "learning_rate": 0.003, "loss": 4.1086, "step": 5144 }, { "epoch": 0.05145, "grad_norm": 1.1530777313914198, "learning_rate": 0.003, "loss": 4.1152, "step": 5145 }, { "epoch": 0.05146, "grad_norm": 1.0839316629215454, "learning_rate": 0.003, "loss": 4.1076, "step": 5146 }, { "epoch": 0.05147, "grad_norm": 1.009008584231616, "learning_rate": 0.003, "loss": 4.1318, "step": 5147 }, { "epoch": 0.05148, "grad_norm": 1.0775481336458725, "learning_rate": 0.003, "loss": 4.1391, "step": 5148 }, { "epoch": 0.05149, "grad_norm": 0.8958199968824248, "learning_rate": 0.003, "loss": 4.107, "step": 5149 }, { "epoch": 0.0515, "grad_norm": 0.8559458186789074, "learning_rate": 0.003, "loss": 4.14, "step": 5150 }, { "epoch": 0.05151, "grad_norm": 0.9332584785279358, "learning_rate": 0.003, "loss": 4.1082, "step": 5151 }, { "epoch": 0.05152, "grad_norm": 0.9765825905780722, "learning_rate": 0.003, "loss": 4.1469, "step": 5152 }, { "epoch": 0.05153, "grad_norm": 0.9840577439203694, "learning_rate": 0.003, "loss": 4.1239, "step": 5153 }, { "epoch": 0.05154, "grad_norm": 1.247678420318687, "learning_rate": 0.003, "loss": 4.1479, "step": 5154 }, { "epoch": 0.05155, "grad_norm": 0.8909111698042794, "learning_rate": 0.003, "loss": 4.1354, "step": 5155 }, { "epoch": 0.05156, "grad_norm": 0.8794649041747071, "learning_rate": 0.003, "loss": 4.1605, "step": 5156 }, { "epoch": 0.05157, "grad_norm": 0.8299673058125748, "learning_rate": 0.003, "loss": 4.1296, "step": 5157 }, { "epoch": 0.05158, "grad_norm": 0.8342651114569559, "learning_rate": 0.003, "loss": 4.1251, "step": 5158 }, { "epoch": 0.05159, "grad_norm": 0.9556371282557828, "learning_rate": 0.003, "loss": 4.1096, "step": 5159 }, { "epoch": 0.0516, "grad_norm": 1.0224358360807535, "learning_rate": 0.003, "loss": 4.1183, "step": 5160 }, { "epoch": 0.05161, "grad_norm": 1.2146298773247146, "learning_rate": 0.003, "loss": 4.0964, "step": 5161 }, { "epoch": 0.05162, "grad_norm": 0.9045538434599459, "learning_rate": 0.003, "loss": 4.1206, "step": 5162 }, { "epoch": 0.05163, "grad_norm": 0.8480691004844201, "learning_rate": 0.003, "loss": 4.1363, "step": 5163 }, { "epoch": 0.05164, "grad_norm": 0.8243976283594234, "learning_rate": 0.003, "loss": 4.1136, "step": 5164 }, { "epoch": 0.05165, "grad_norm": 0.932262551471062, "learning_rate": 0.003, "loss": 4.1336, "step": 5165 }, { "epoch": 0.05166, "grad_norm": 0.8908185716264864, "learning_rate": 0.003, "loss": 4.1334, "step": 5166 }, { "epoch": 0.05167, "grad_norm": 0.8051457712052962, "learning_rate": 0.003, "loss": 4.1349, "step": 5167 }, { "epoch": 0.05168, "grad_norm": 0.9134452565395993, "learning_rate": 0.003, "loss": 4.1709, "step": 5168 }, { "epoch": 0.05169, "grad_norm": 1.132338869492368, "learning_rate": 0.003, "loss": 4.1245, "step": 5169 }, { "epoch": 0.0517, "grad_norm": 0.9742389137238414, "learning_rate": 0.003, "loss": 4.0889, "step": 5170 }, { "epoch": 0.05171, "grad_norm": 1.067704962383225, "learning_rate": 0.003, "loss": 4.1262, "step": 5171 }, { "epoch": 0.05172, "grad_norm": 0.8196743774447537, "learning_rate": 0.003, "loss": 4.1337, "step": 5172 }, { "epoch": 0.05173, "grad_norm": 0.7197901038637725, "learning_rate": 0.003, "loss": 4.1426, "step": 5173 }, { "epoch": 0.05174, "grad_norm": 0.6647802187127981, "learning_rate": 0.003, "loss": 4.1305, "step": 5174 }, { "epoch": 0.05175, "grad_norm": 0.6834092900736586, "learning_rate": 0.003, "loss": 4.1381, "step": 5175 }, { "epoch": 0.05176, "grad_norm": 0.6719272953115575, "learning_rate": 0.003, "loss": 4.102, "step": 5176 }, { "epoch": 0.05177, "grad_norm": 0.7402517697049626, "learning_rate": 0.003, "loss": 4.097, "step": 5177 }, { "epoch": 0.05178, "grad_norm": 0.9402641067102157, "learning_rate": 0.003, "loss": 4.0967, "step": 5178 }, { "epoch": 0.05179, "grad_norm": 1.1891054842856807, "learning_rate": 0.003, "loss": 4.124, "step": 5179 }, { "epoch": 0.0518, "grad_norm": 0.8356173492045658, "learning_rate": 0.003, "loss": 4.1136, "step": 5180 }, { "epoch": 0.05181, "grad_norm": 0.8309031331284943, "learning_rate": 0.003, "loss": 4.1237, "step": 5181 }, { "epoch": 0.05182, "grad_norm": 0.7669452922855337, "learning_rate": 0.003, "loss": 4.1363, "step": 5182 }, { "epoch": 0.05183, "grad_norm": 0.8821032377373486, "learning_rate": 0.003, "loss": 4.1031, "step": 5183 }, { "epoch": 0.05184, "grad_norm": 1.236399441865476, "learning_rate": 0.003, "loss": 4.1199, "step": 5184 }, { "epoch": 0.05185, "grad_norm": 1.0136590261857248, "learning_rate": 0.003, "loss": 4.1155, "step": 5185 }, { "epoch": 0.05186, "grad_norm": 0.9746961882311193, "learning_rate": 0.003, "loss": 4.1255, "step": 5186 }, { "epoch": 0.05187, "grad_norm": 0.9084848079843075, "learning_rate": 0.003, "loss": 4.1637, "step": 5187 }, { "epoch": 0.05188, "grad_norm": 1.1465706391107853, "learning_rate": 0.003, "loss": 4.1409, "step": 5188 }, { "epoch": 0.05189, "grad_norm": 1.0110195586113877, "learning_rate": 0.003, "loss": 4.1156, "step": 5189 }, { "epoch": 0.0519, "grad_norm": 0.9256433751833841, "learning_rate": 0.003, "loss": 4.1193, "step": 5190 }, { "epoch": 0.05191, "grad_norm": 0.9134067973569046, "learning_rate": 0.003, "loss": 4.118, "step": 5191 }, { "epoch": 0.05192, "grad_norm": 0.8394166491498798, "learning_rate": 0.003, "loss": 4.139, "step": 5192 }, { "epoch": 0.05193, "grad_norm": 0.8609628758113029, "learning_rate": 0.003, "loss": 4.1347, "step": 5193 }, { "epoch": 0.05194, "grad_norm": 0.9815840288645425, "learning_rate": 0.003, "loss": 4.1002, "step": 5194 }, { "epoch": 0.05195, "grad_norm": 1.0136348688047472, "learning_rate": 0.003, "loss": 4.1421, "step": 5195 }, { "epoch": 0.05196, "grad_norm": 1.1480365342383962, "learning_rate": 0.003, "loss": 4.1212, "step": 5196 }, { "epoch": 0.05197, "grad_norm": 1.012937790622735, "learning_rate": 0.003, "loss": 4.1432, "step": 5197 }, { "epoch": 0.05198, "grad_norm": 1.01624508607687, "learning_rate": 0.003, "loss": 4.1139, "step": 5198 }, { "epoch": 0.05199, "grad_norm": 1.0857847091315482, "learning_rate": 0.003, "loss": 4.1058, "step": 5199 }, { "epoch": 0.052, "grad_norm": 0.9295276501582445, "learning_rate": 0.003, "loss": 4.1351, "step": 5200 }, { "epoch": 0.05201, "grad_norm": 1.0722513948753385, "learning_rate": 0.003, "loss": 4.1361, "step": 5201 }, { "epoch": 0.05202, "grad_norm": 0.939772903818911, "learning_rate": 0.003, "loss": 4.1201, "step": 5202 }, { "epoch": 0.05203, "grad_norm": 0.9141550193477442, "learning_rate": 0.003, "loss": 4.1491, "step": 5203 }, { "epoch": 0.05204, "grad_norm": 1.1559198563879414, "learning_rate": 0.003, "loss": 4.1436, "step": 5204 }, { "epoch": 0.05205, "grad_norm": 1.0879097255219679, "learning_rate": 0.003, "loss": 4.1488, "step": 5205 }, { "epoch": 0.05206, "grad_norm": 1.116664769475625, "learning_rate": 0.003, "loss": 4.1357, "step": 5206 }, { "epoch": 0.05207, "grad_norm": 0.8807032835753495, "learning_rate": 0.003, "loss": 4.1097, "step": 5207 }, { "epoch": 0.05208, "grad_norm": 0.9569291327418711, "learning_rate": 0.003, "loss": 4.1271, "step": 5208 }, { "epoch": 0.05209, "grad_norm": 1.0490164635826829, "learning_rate": 0.003, "loss": 4.1239, "step": 5209 }, { "epoch": 0.0521, "grad_norm": 1.0917925759655678, "learning_rate": 0.003, "loss": 4.0981, "step": 5210 }, { "epoch": 0.05211, "grad_norm": 0.8008924222179035, "learning_rate": 0.003, "loss": 4.1158, "step": 5211 }, { "epoch": 0.05212, "grad_norm": 0.7902460429952818, "learning_rate": 0.003, "loss": 4.163, "step": 5212 }, { "epoch": 0.05213, "grad_norm": 0.8144312236237736, "learning_rate": 0.003, "loss": 4.1184, "step": 5213 }, { "epoch": 0.05214, "grad_norm": 0.9364089538217132, "learning_rate": 0.003, "loss": 4.1161, "step": 5214 }, { "epoch": 0.05215, "grad_norm": 1.0643005071151037, "learning_rate": 0.003, "loss": 4.1138, "step": 5215 }, { "epoch": 0.05216, "grad_norm": 0.8962708070858758, "learning_rate": 0.003, "loss": 4.1215, "step": 5216 }, { "epoch": 0.05217, "grad_norm": 0.9767411427633232, "learning_rate": 0.003, "loss": 4.1256, "step": 5217 }, { "epoch": 0.05218, "grad_norm": 1.0676104045202033, "learning_rate": 0.003, "loss": 4.1178, "step": 5218 }, { "epoch": 0.05219, "grad_norm": 1.1483320645297568, "learning_rate": 0.003, "loss": 4.1451, "step": 5219 }, { "epoch": 0.0522, "grad_norm": 0.9125640959202299, "learning_rate": 0.003, "loss": 4.11, "step": 5220 }, { "epoch": 0.05221, "grad_norm": 0.7553348099730735, "learning_rate": 0.003, "loss": 4.1156, "step": 5221 }, { "epoch": 0.05222, "grad_norm": 0.7974155600968277, "learning_rate": 0.003, "loss": 4.1265, "step": 5222 }, { "epoch": 0.05223, "grad_norm": 0.9496439860291601, "learning_rate": 0.003, "loss": 4.1454, "step": 5223 }, { "epoch": 0.05224, "grad_norm": 1.16338354437749, "learning_rate": 0.003, "loss": 4.1245, "step": 5224 }, { "epoch": 0.05225, "grad_norm": 0.7847885589485009, "learning_rate": 0.003, "loss": 4.141, "step": 5225 }, { "epoch": 0.05226, "grad_norm": 0.7629525693734982, "learning_rate": 0.003, "loss": 4.1291, "step": 5226 }, { "epoch": 0.05227, "grad_norm": 0.8426081978751812, "learning_rate": 0.003, "loss": 4.1542, "step": 5227 }, { "epoch": 0.05228, "grad_norm": 0.9894193979113189, "learning_rate": 0.003, "loss": 4.1384, "step": 5228 }, { "epoch": 0.05229, "grad_norm": 0.9793859414490865, "learning_rate": 0.003, "loss": 4.1049, "step": 5229 }, { "epoch": 0.0523, "grad_norm": 1.1720102953944842, "learning_rate": 0.003, "loss": 4.1066, "step": 5230 }, { "epoch": 0.05231, "grad_norm": 0.8997010664140895, "learning_rate": 0.003, "loss": 4.1335, "step": 5231 }, { "epoch": 0.05232, "grad_norm": 1.0693983537886569, "learning_rate": 0.003, "loss": 4.1271, "step": 5232 }, { "epoch": 0.05233, "grad_norm": 1.1956318009565163, "learning_rate": 0.003, "loss": 4.1162, "step": 5233 }, { "epoch": 0.05234, "grad_norm": 0.8983927893087599, "learning_rate": 0.003, "loss": 4.1515, "step": 5234 }, { "epoch": 0.05235, "grad_norm": 1.108065774608959, "learning_rate": 0.003, "loss": 4.1411, "step": 5235 }, { "epoch": 0.05236, "grad_norm": 0.9295908500527476, "learning_rate": 0.003, "loss": 4.116, "step": 5236 }, { "epoch": 0.05237, "grad_norm": 0.960476320153068, "learning_rate": 0.003, "loss": 4.1207, "step": 5237 }, { "epoch": 0.05238, "grad_norm": 1.057289393831159, "learning_rate": 0.003, "loss": 4.1114, "step": 5238 }, { "epoch": 0.05239, "grad_norm": 1.204033369603102, "learning_rate": 0.003, "loss": 4.1415, "step": 5239 }, { "epoch": 0.0524, "grad_norm": 0.8690312577527531, "learning_rate": 0.003, "loss": 4.1418, "step": 5240 }, { "epoch": 0.05241, "grad_norm": 0.7542309413903446, "learning_rate": 0.003, "loss": 4.1436, "step": 5241 }, { "epoch": 0.05242, "grad_norm": 0.7680576081041368, "learning_rate": 0.003, "loss": 4.1413, "step": 5242 }, { "epoch": 0.05243, "grad_norm": 0.9313741587108058, "learning_rate": 0.003, "loss": 4.1363, "step": 5243 }, { "epoch": 0.05244, "grad_norm": 1.066443957086773, "learning_rate": 0.003, "loss": 4.1163, "step": 5244 }, { "epoch": 0.05245, "grad_norm": 1.109894429188262, "learning_rate": 0.003, "loss": 4.1236, "step": 5245 }, { "epoch": 0.05246, "grad_norm": 1.2624847108853103, "learning_rate": 0.003, "loss": 4.1355, "step": 5246 }, { "epoch": 0.05247, "grad_norm": 0.7462508145722354, "learning_rate": 0.003, "loss": 4.1266, "step": 5247 }, { "epoch": 0.05248, "grad_norm": 0.8237453540746109, "learning_rate": 0.003, "loss": 4.1328, "step": 5248 }, { "epoch": 0.05249, "grad_norm": 0.9781734225083963, "learning_rate": 0.003, "loss": 4.1473, "step": 5249 }, { "epoch": 0.0525, "grad_norm": 1.095567790576504, "learning_rate": 0.003, "loss": 4.1399, "step": 5250 }, { "epoch": 0.05251, "grad_norm": 1.1083386982850694, "learning_rate": 0.003, "loss": 4.1309, "step": 5251 }, { "epoch": 0.05252, "grad_norm": 0.890687010798308, "learning_rate": 0.003, "loss": 4.1526, "step": 5252 }, { "epoch": 0.05253, "grad_norm": 0.990651302574722, "learning_rate": 0.003, "loss": 4.1211, "step": 5253 }, { "epoch": 0.05254, "grad_norm": 1.030601009707246, "learning_rate": 0.003, "loss": 4.1091, "step": 5254 }, { "epoch": 0.05255, "grad_norm": 0.9511829626380641, "learning_rate": 0.003, "loss": 4.1104, "step": 5255 }, { "epoch": 0.05256, "grad_norm": 1.0335656090038043, "learning_rate": 0.003, "loss": 4.1379, "step": 5256 }, { "epoch": 0.05257, "grad_norm": 1.1381395017816789, "learning_rate": 0.003, "loss": 4.1327, "step": 5257 }, { "epoch": 0.05258, "grad_norm": 0.8499060638291269, "learning_rate": 0.003, "loss": 4.1261, "step": 5258 }, { "epoch": 0.05259, "grad_norm": 0.7311247197745678, "learning_rate": 0.003, "loss": 4.1102, "step": 5259 }, { "epoch": 0.0526, "grad_norm": 0.6682808585819072, "learning_rate": 0.003, "loss": 4.1371, "step": 5260 }, { "epoch": 0.05261, "grad_norm": 0.7141975868734773, "learning_rate": 0.003, "loss": 4.1039, "step": 5261 }, { "epoch": 0.05262, "grad_norm": 0.9329305583466294, "learning_rate": 0.003, "loss": 4.1031, "step": 5262 }, { "epoch": 0.05263, "grad_norm": 1.0911979951064397, "learning_rate": 0.003, "loss": 4.1549, "step": 5263 }, { "epoch": 0.05264, "grad_norm": 0.9231843005490539, "learning_rate": 0.003, "loss": 4.1481, "step": 5264 }, { "epoch": 0.05265, "grad_norm": 0.8912385368222187, "learning_rate": 0.003, "loss": 4.1091, "step": 5265 }, { "epoch": 0.05266, "grad_norm": 1.0102934580335026, "learning_rate": 0.003, "loss": 4.1002, "step": 5266 }, { "epoch": 0.05267, "grad_norm": 1.0589226520761357, "learning_rate": 0.003, "loss": 4.152, "step": 5267 }, { "epoch": 0.05268, "grad_norm": 1.0573701409369491, "learning_rate": 0.003, "loss": 4.1176, "step": 5268 }, { "epoch": 0.05269, "grad_norm": 1.0527832464427491, "learning_rate": 0.003, "loss": 4.0954, "step": 5269 }, { "epoch": 0.0527, "grad_norm": 1.2210421513264824, "learning_rate": 0.003, "loss": 4.1324, "step": 5270 }, { "epoch": 0.05271, "grad_norm": 0.8659230590416374, "learning_rate": 0.003, "loss": 4.0925, "step": 5271 }, { "epoch": 0.05272, "grad_norm": 0.8213928297594013, "learning_rate": 0.003, "loss": 4.1447, "step": 5272 }, { "epoch": 0.05273, "grad_norm": 0.8661885686925771, "learning_rate": 0.003, "loss": 4.1337, "step": 5273 }, { "epoch": 0.05274, "grad_norm": 0.9513399963087238, "learning_rate": 0.003, "loss": 4.1434, "step": 5274 }, { "epoch": 0.05275, "grad_norm": 0.9767073542295025, "learning_rate": 0.003, "loss": 4.1179, "step": 5275 }, { "epoch": 0.05276, "grad_norm": 1.0855056908438585, "learning_rate": 0.003, "loss": 4.1112, "step": 5276 }, { "epoch": 0.05277, "grad_norm": 1.049255203197355, "learning_rate": 0.003, "loss": 4.1015, "step": 5277 }, { "epoch": 0.05278, "grad_norm": 1.1033706045707017, "learning_rate": 0.003, "loss": 4.1189, "step": 5278 }, { "epoch": 0.05279, "grad_norm": 0.870576303775063, "learning_rate": 0.003, "loss": 4.1148, "step": 5279 }, { "epoch": 0.0528, "grad_norm": 0.7247735805655298, "learning_rate": 0.003, "loss": 4.1066, "step": 5280 }, { "epoch": 0.05281, "grad_norm": 0.7279116242105975, "learning_rate": 0.003, "loss": 4.1199, "step": 5281 }, { "epoch": 0.05282, "grad_norm": 0.6767646560568664, "learning_rate": 0.003, "loss": 4.1115, "step": 5282 }, { "epoch": 0.05283, "grad_norm": 0.6311016160426699, "learning_rate": 0.003, "loss": 4.0959, "step": 5283 }, { "epoch": 0.05284, "grad_norm": 0.8176724475761017, "learning_rate": 0.003, "loss": 4.1268, "step": 5284 }, { "epoch": 0.05285, "grad_norm": 1.152437810958939, "learning_rate": 0.003, "loss": 4.0908, "step": 5285 }, { "epoch": 0.05286, "grad_norm": 1.0650058217265772, "learning_rate": 0.003, "loss": 4.1548, "step": 5286 }, { "epoch": 0.05287, "grad_norm": 0.7344115850568415, "learning_rate": 0.003, "loss": 4.1161, "step": 5287 }, { "epoch": 0.05288, "grad_norm": 0.7465352407381385, "learning_rate": 0.003, "loss": 4.0994, "step": 5288 }, { "epoch": 0.05289, "grad_norm": 0.9583980517974787, "learning_rate": 0.003, "loss": 4.1296, "step": 5289 }, { "epoch": 0.0529, "grad_norm": 1.1218838641820494, "learning_rate": 0.003, "loss": 4.1285, "step": 5290 }, { "epoch": 0.05291, "grad_norm": 0.9074431149136537, "learning_rate": 0.003, "loss": 4.1376, "step": 5291 }, { "epoch": 0.05292, "grad_norm": 0.8727673415619097, "learning_rate": 0.003, "loss": 4.1301, "step": 5292 }, { "epoch": 0.05293, "grad_norm": 0.8526331049950101, "learning_rate": 0.003, "loss": 4.1294, "step": 5293 }, { "epoch": 0.05294, "grad_norm": 0.9787782709064855, "learning_rate": 0.003, "loss": 4.1191, "step": 5294 }, { "epoch": 0.05295, "grad_norm": 1.0826337231740053, "learning_rate": 0.003, "loss": 4.1369, "step": 5295 }, { "epoch": 0.05296, "grad_norm": 0.8743798848979503, "learning_rate": 0.003, "loss": 4.1114, "step": 5296 }, { "epoch": 0.05297, "grad_norm": 0.8905436718147242, "learning_rate": 0.003, "loss": 4.116, "step": 5297 }, { "epoch": 0.05298, "grad_norm": 0.9159685627131691, "learning_rate": 0.003, "loss": 4.1372, "step": 5298 }, { "epoch": 0.05299, "grad_norm": 1.052082610357495, "learning_rate": 0.003, "loss": 4.116, "step": 5299 }, { "epoch": 0.053, "grad_norm": 1.254455298911267, "learning_rate": 0.003, "loss": 4.1459, "step": 5300 }, { "epoch": 0.05301, "grad_norm": 0.8905522802074054, "learning_rate": 0.003, "loss": 4.1463, "step": 5301 }, { "epoch": 0.05302, "grad_norm": 0.9386802829956024, "learning_rate": 0.003, "loss": 4.1206, "step": 5302 }, { "epoch": 0.05303, "grad_norm": 1.010270978418881, "learning_rate": 0.003, "loss": 4.1358, "step": 5303 }, { "epoch": 0.05304, "grad_norm": 1.168823963382456, "learning_rate": 0.003, "loss": 4.1259, "step": 5304 }, { "epoch": 0.05305, "grad_norm": 0.9968521036034549, "learning_rate": 0.003, "loss": 4.1311, "step": 5305 }, { "epoch": 0.05306, "grad_norm": 1.1415999424056091, "learning_rate": 0.003, "loss": 4.1436, "step": 5306 }, { "epoch": 0.05307, "grad_norm": 0.8577763401656362, "learning_rate": 0.003, "loss": 4.1333, "step": 5307 }, { "epoch": 0.05308, "grad_norm": 0.8065378663634303, "learning_rate": 0.003, "loss": 4.1436, "step": 5308 }, { "epoch": 0.05309, "grad_norm": 0.8620894674479869, "learning_rate": 0.003, "loss": 4.1092, "step": 5309 }, { "epoch": 0.0531, "grad_norm": 0.8265410133677272, "learning_rate": 0.003, "loss": 4.122, "step": 5310 }, { "epoch": 0.05311, "grad_norm": 0.840656557399505, "learning_rate": 0.003, "loss": 4.1549, "step": 5311 }, { "epoch": 0.05312, "grad_norm": 0.9477091831792939, "learning_rate": 0.003, "loss": 4.1132, "step": 5312 }, { "epoch": 0.05313, "grad_norm": 0.899466744931056, "learning_rate": 0.003, "loss": 4.0986, "step": 5313 }, { "epoch": 0.05314, "grad_norm": 0.778035937952836, "learning_rate": 0.003, "loss": 4.1225, "step": 5314 }, { "epoch": 0.05315, "grad_norm": 0.7258696044493477, "learning_rate": 0.003, "loss": 4.1265, "step": 5315 }, { "epoch": 0.05316, "grad_norm": 0.7881753340530017, "learning_rate": 0.003, "loss": 4.1286, "step": 5316 }, { "epoch": 0.05317, "grad_norm": 0.8930575208143203, "learning_rate": 0.003, "loss": 4.1315, "step": 5317 }, { "epoch": 0.05318, "grad_norm": 1.1521265057134467, "learning_rate": 0.003, "loss": 4.1055, "step": 5318 }, { "epoch": 0.05319, "grad_norm": 1.241887421255421, "learning_rate": 0.003, "loss": 4.1151, "step": 5319 }, { "epoch": 0.0532, "grad_norm": 0.8187535312680814, "learning_rate": 0.003, "loss": 4.1173, "step": 5320 }, { "epoch": 0.05321, "grad_norm": 0.8332689083765074, "learning_rate": 0.003, "loss": 4.1182, "step": 5321 }, { "epoch": 0.05322, "grad_norm": 0.9035176420820373, "learning_rate": 0.003, "loss": 4.1398, "step": 5322 }, { "epoch": 0.05323, "grad_norm": 0.8995073997128675, "learning_rate": 0.003, "loss": 4.1614, "step": 5323 }, { "epoch": 0.05324, "grad_norm": 0.8729634947781086, "learning_rate": 0.003, "loss": 4.1035, "step": 5324 }, { "epoch": 0.05325, "grad_norm": 0.8236931256668504, "learning_rate": 0.003, "loss": 4.1233, "step": 5325 }, { "epoch": 0.05326, "grad_norm": 0.9692358504853177, "learning_rate": 0.003, "loss": 4.1448, "step": 5326 }, { "epoch": 0.05327, "grad_norm": 1.176101038605182, "learning_rate": 0.003, "loss": 4.1295, "step": 5327 }, { "epoch": 0.05328, "grad_norm": 1.0752975657295243, "learning_rate": 0.003, "loss": 4.137, "step": 5328 }, { "epoch": 0.05329, "grad_norm": 0.9034182614898145, "learning_rate": 0.003, "loss": 4.1194, "step": 5329 }, { "epoch": 0.0533, "grad_norm": 0.9520954885772951, "learning_rate": 0.003, "loss": 4.1392, "step": 5330 }, { "epoch": 0.05331, "grad_norm": 1.0638411915039017, "learning_rate": 0.003, "loss": 4.1367, "step": 5331 }, { "epoch": 0.05332, "grad_norm": 1.0918447593613807, "learning_rate": 0.003, "loss": 4.153, "step": 5332 }, { "epoch": 0.05333, "grad_norm": 1.0190559714048477, "learning_rate": 0.003, "loss": 4.1214, "step": 5333 }, { "epoch": 0.05334, "grad_norm": 1.0666691935424213, "learning_rate": 0.003, "loss": 4.1119, "step": 5334 }, { "epoch": 0.05335, "grad_norm": 0.9983956702837128, "learning_rate": 0.003, "loss": 4.1417, "step": 5335 }, { "epoch": 0.05336, "grad_norm": 1.1601743720224316, "learning_rate": 0.003, "loss": 4.1541, "step": 5336 }, { "epoch": 0.05337, "grad_norm": 0.7881208672585496, "learning_rate": 0.003, "loss": 4.1362, "step": 5337 }, { "epoch": 0.05338, "grad_norm": 0.7044325113303782, "learning_rate": 0.003, "loss": 4.1153, "step": 5338 }, { "epoch": 0.05339, "grad_norm": 0.8906022550747246, "learning_rate": 0.003, "loss": 4.1105, "step": 5339 }, { "epoch": 0.0534, "grad_norm": 1.0726295657720748, "learning_rate": 0.003, "loss": 4.1323, "step": 5340 }, { "epoch": 0.05341, "grad_norm": 1.0067520134393941, "learning_rate": 0.003, "loss": 4.1366, "step": 5341 }, { "epoch": 0.05342, "grad_norm": 1.1915314362667089, "learning_rate": 0.003, "loss": 4.1291, "step": 5342 }, { "epoch": 0.05343, "grad_norm": 1.0367148889053333, "learning_rate": 0.003, "loss": 4.1394, "step": 5343 }, { "epoch": 0.05344, "grad_norm": 0.8675512667584127, "learning_rate": 0.003, "loss": 4.0968, "step": 5344 }, { "epoch": 0.05345, "grad_norm": 0.7665396986619814, "learning_rate": 0.003, "loss": 4.1076, "step": 5345 }, { "epoch": 0.05346, "grad_norm": 0.7754587564090736, "learning_rate": 0.003, "loss": 4.1263, "step": 5346 }, { "epoch": 0.05347, "grad_norm": 0.8030514376721718, "learning_rate": 0.003, "loss": 4.112, "step": 5347 }, { "epoch": 0.05348, "grad_norm": 0.8046117284167067, "learning_rate": 0.003, "loss": 4.1096, "step": 5348 }, { "epoch": 0.05349, "grad_norm": 0.849059304723128, "learning_rate": 0.003, "loss": 4.1213, "step": 5349 }, { "epoch": 0.0535, "grad_norm": 1.082067485015541, "learning_rate": 0.003, "loss": 4.1156, "step": 5350 }, { "epoch": 0.05351, "grad_norm": 1.1906159889512828, "learning_rate": 0.003, "loss": 4.1225, "step": 5351 }, { "epoch": 0.05352, "grad_norm": 1.0756028658025156, "learning_rate": 0.003, "loss": 4.1155, "step": 5352 }, { "epoch": 0.05353, "grad_norm": 1.0059552387327784, "learning_rate": 0.003, "loss": 4.1244, "step": 5353 }, { "epoch": 0.05354, "grad_norm": 1.1282632737459894, "learning_rate": 0.003, "loss": 4.114, "step": 5354 }, { "epoch": 0.05355, "grad_norm": 0.9652277661616635, "learning_rate": 0.003, "loss": 4.1049, "step": 5355 }, { "epoch": 0.05356, "grad_norm": 0.9108289260268686, "learning_rate": 0.003, "loss": 4.1412, "step": 5356 }, { "epoch": 0.05357, "grad_norm": 0.7828004246923763, "learning_rate": 0.003, "loss": 4.1259, "step": 5357 }, { "epoch": 0.05358, "grad_norm": 0.9168739207771537, "learning_rate": 0.003, "loss": 4.1163, "step": 5358 }, { "epoch": 0.05359, "grad_norm": 1.0075080046155496, "learning_rate": 0.003, "loss": 4.1299, "step": 5359 }, { "epoch": 0.0536, "grad_norm": 1.0970792845542654, "learning_rate": 0.003, "loss": 4.1282, "step": 5360 }, { "epoch": 0.05361, "grad_norm": 0.8824114030322678, "learning_rate": 0.003, "loss": 4.1317, "step": 5361 }, { "epoch": 0.05362, "grad_norm": 0.993743544871738, "learning_rate": 0.003, "loss": 4.1425, "step": 5362 }, { "epoch": 0.05363, "grad_norm": 0.9312141932011715, "learning_rate": 0.003, "loss": 4.147, "step": 5363 }, { "epoch": 0.05364, "grad_norm": 0.8022980893619517, "learning_rate": 0.003, "loss": 4.1398, "step": 5364 }, { "epoch": 0.05365, "grad_norm": 0.7351009835979198, "learning_rate": 0.003, "loss": 4.1298, "step": 5365 }, { "epoch": 0.05366, "grad_norm": 0.7464822800282958, "learning_rate": 0.003, "loss": 4.1125, "step": 5366 }, { "epoch": 0.05367, "grad_norm": 0.8388818072236447, "learning_rate": 0.003, "loss": 4.1133, "step": 5367 }, { "epoch": 0.05368, "grad_norm": 0.9344086732131297, "learning_rate": 0.003, "loss": 4.1069, "step": 5368 }, { "epoch": 0.05369, "grad_norm": 1.0848241130255325, "learning_rate": 0.003, "loss": 4.125, "step": 5369 }, { "epoch": 0.0537, "grad_norm": 1.0620467839392764, "learning_rate": 0.003, "loss": 4.1253, "step": 5370 }, { "epoch": 0.05371, "grad_norm": 1.2010940806246506, "learning_rate": 0.003, "loss": 4.1479, "step": 5371 }, { "epoch": 0.05372, "grad_norm": 0.7001840635960661, "learning_rate": 0.003, "loss": 4.1152, "step": 5372 }, { "epoch": 0.05373, "grad_norm": 0.780826293502416, "learning_rate": 0.003, "loss": 4.133, "step": 5373 }, { "epoch": 0.05374, "grad_norm": 1.0632965886449934, "learning_rate": 0.003, "loss": 4.1154, "step": 5374 }, { "epoch": 0.05375, "grad_norm": 1.1966014514160626, "learning_rate": 0.003, "loss": 4.1168, "step": 5375 }, { "epoch": 0.05376, "grad_norm": 0.8842297934560165, "learning_rate": 0.003, "loss": 4.1389, "step": 5376 }, { "epoch": 0.05377, "grad_norm": 0.778786324051496, "learning_rate": 0.003, "loss": 4.1251, "step": 5377 }, { "epoch": 0.05378, "grad_norm": 0.8596824656028225, "learning_rate": 0.003, "loss": 4.1289, "step": 5378 }, { "epoch": 0.05379, "grad_norm": 0.9934380169213879, "learning_rate": 0.003, "loss": 4.1391, "step": 5379 }, { "epoch": 0.0538, "grad_norm": 0.9638667426110514, "learning_rate": 0.003, "loss": 4.1071, "step": 5380 }, { "epoch": 0.05381, "grad_norm": 1.09393211180781, "learning_rate": 0.003, "loss": 4.1237, "step": 5381 }, { "epoch": 0.05382, "grad_norm": 0.7651739896803385, "learning_rate": 0.003, "loss": 4.135, "step": 5382 }, { "epoch": 0.05383, "grad_norm": 0.7092093674621837, "learning_rate": 0.003, "loss": 4.1301, "step": 5383 }, { "epoch": 0.05384, "grad_norm": 0.6802032392903303, "learning_rate": 0.003, "loss": 4.1096, "step": 5384 }, { "epoch": 0.05385, "grad_norm": 0.7953694996673716, "learning_rate": 0.003, "loss": 4.1378, "step": 5385 }, { "epoch": 0.05386, "grad_norm": 1.0544432753666775, "learning_rate": 0.003, "loss": 4.1246, "step": 5386 }, { "epoch": 0.05387, "grad_norm": 1.1016215087001056, "learning_rate": 0.003, "loss": 4.1131, "step": 5387 }, { "epoch": 0.05388, "grad_norm": 1.0432563739342082, "learning_rate": 0.003, "loss": 4.1506, "step": 5388 }, { "epoch": 0.05389, "grad_norm": 1.024764584061503, "learning_rate": 0.003, "loss": 4.1229, "step": 5389 }, { "epoch": 0.0539, "grad_norm": 1.1680948971945517, "learning_rate": 0.003, "loss": 4.099, "step": 5390 }, { "epoch": 0.05391, "grad_norm": 1.063797358927456, "learning_rate": 0.003, "loss": 4.1141, "step": 5391 }, { "epoch": 0.05392, "grad_norm": 1.0207694285265143, "learning_rate": 0.003, "loss": 4.1182, "step": 5392 }, { "epoch": 0.05393, "grad_norm": 1.0736035287161585, "learning_rate": 0.003, "loss": 4.1448, "step": 5393 }, { "epoch": 0.05394, "grad_norm": 1.0325659552141049, "learning_rate": 0.003, "loss": 4.1183, "step": 5394 }, { "epoch": 0.05395, "grad_norm": 0.8864130337421087, "learning_rate": 0.003, "loss": 4.1451, "step": 5395 }, { "epoch": 0.05396, "grad_norm": 0.7998033184393275, "learning_rate": 0.003, "loss": 4.1331, "step": 5396 }, { "epoch": 0.05397, "grad_norm": 0.8620800324774359, "learning_rate": 0.003, "loss": 4.1085, "step": 5397 }, { "epoch": 0.05398, "grad_norm": 0.9934924712119845, "learning_rate": 0.003, "loss": 4.1093, "step": 5398 }, { "epoch": 0.05399, "grad_norm": 1.0677747326074263, "learning_rate": 0.003, "loss": 4.1455, "step": 5399 }, { "epoch": 0.054, "grad_norm": 1.0851959853077549, "learning_rate": 0.003, "loss": 4.1543, "step": 5400 }, { "epoch": 0.05401, "grad_norm": 0.9114774952428576, "learning_rate": 0.003, "loss": 4.128, "step": 5401 }, { "epoch": 0.05402, "grad_norm": 0.7885051914531169, "learning_rate": 0.003, "loss": 4.1396, "step": 5402 }, { "epoch": 0.05403, "grad_norm": 0.8695681565380565, "learning_rate": 0.003, "loss": 4.1263, "step": 5403 }, { "epoch": 0.05404, "grad_norm": 0.8524754859405422, "learning_rate": 0.003, "loss": 4.1187, "step": 5404 }, { "epoch": 0.05405, "grad_norm": 0.925811086709952, "learning_rate": 0.003, "loss": 4.1227, "step": 5405 }, { "epoch": 0.05406, "grad_norm": 1.1266556587941388, "learning_rate": 0.003, "loss": 4.1135, "step": 5406 }, { "epoch": 0.05407, "grad_norm": 0.8810690573065147, "learning_rate": 0.003, "loss": 4.1209, "step": 5407 }, { "epoch": 0.05408, "grad_norm": 0.9725214113742383, "learning_rate": 0.003, "loss": 4.1111, "step": 5408 }, { "epoch": 0.05409, "grad_norm": 0.9978772080924893, "learning_rate": 0.003, "loss": 4.082, "step": 5409 }, { "epoch": 0.0541, "grad_norm": 1.0553800598077203, "learning_rate": 0.003, "loss": 4.1355, "step": 5410 }, { "epoch": 0.05411, "grad_norm": 0.8112327791737242, "learning_rate": 0.003, "loss": 4.1275, "step": 5411 }, { "epoch": 0.05412, "grad_norm": 0.814707239730718, "learning_rate": 0.003, "loss": 4.1219, "step": 5412 }, { "epoch": 0.05413, "grad_norm": 0.8573040534301559, "learning_rate": 0.003, "loss": 4.1167, "step": 5413 }, { "epoch": 0.05414, "grad_norm": 1.0569847685652707, "learning_rate": 0.003, "loss": 4.0999, "step": 5414 }, { "epoch": 0.05415, "grad_norm": 1.296000338820238, "learning_rate": 0.003, "loss": 4.1445, "step": 5415 }, { "epoch": 0.05416, "grad_norm": 0.7564909243699854, "learning_rate": 0.003, "loss": 4.0903, "step": 5416 }, { "epoch": 0.05417, "grad_norm": 0.6579679791300916, "learning_rate": 0.003, "loss": 4.1119, "step": 5417 }, { "epoch": 0.05418, "grad_norm": 0.7369314194114258, "learning_rate": 0.003, "loss": 4.103, "step": 5418 }, { "epoch": 0.05419, "grad_norm": 0.9712634198986607, "learning_rate": 0.003, "loss": 4.1249, "step": 5419 }, { "epoch": 0.0542, "grad_norm": 1.6010509746450199, "learning_rate": 0.003, "loss": 4.1256, "step": 5420 }, { "epoch": 0.05421, "grad_norm": 0.7403826711098679, "learning_rate": 0.003, "loss": 4.1084, "step": 5421 }, { "epoch": 0.05422, "grad_norm": 0.8348547775605684, "learning_rate": 0.003, "loss": 4.1272, "step": 5422 }, { "epoch": 0.05423, "grad_norm": 1.0466376848312227, "learning_rate": 0.003, "loss": 4.1237, "step": 5423 }, { "epoch": 0.05424, "grad_norm": 1.1320796829971231, "learning_rate": 0.003, "loss": 4.1578, "step": 5424 }, { "epoch": 0.05425, "grad_norm": 0.9903561882831523, "learning_rate": 0.003, "loss": 4.1303, "step": 5425 }, { "epoch": 0.05426, "grad_norm": 0.845398305003601, "learning_rate": 0.003, "loss": 4.11, "step": 5426 }, { "epoch": 0.05427, "grad_norm": 0.8221075578228308, "learning_rate": 0.003, "loss": 4.105, "step": 5427 }, { "epoch": 0.05428, "grad_norm": 0.8509764151091259, "learning_rate": 0.003, "loss": 4.1247, "step": 5428 }, { "epoch": 0.05429, "grad_norm": 0.9581157952224152, "learning_rate": 0.003, "loss": 4.13, "step": 5429 }, { "epoch": 0.0543, "grad_norm": 0.9436738122904814, "learning_rate": 0.003, "loss": 4.1516, "step": 5430 }, { "epoch": 0.05431, "grad_norm": 0.9842222295174526, "learning_rate": 0.003, "loss": 4.1251, "step": 5431 }, { "epoch": 0.05432, "grad_norm": 0.9492659298620857, "learning_rate": 0.003, "loss": 4.1172, "step": 5432 }, { "epoch": 0.05433, "grad_norm": 1.0409726423743026, "learning_rate": 0.003, "loss": 4.0966, "step": 5433 }, { "epoch": 0.05434, "grad_norm": 1.2078024184830263, "learning_rate": 0.003, "loss": 4.1253, "step": 5434 }, { "epoch": 0.05435, "grad_norm": 1.0505703926974785, "learning_rate": 0.003, "loss": 4.1148, "step": 5435 }, { "epoch": 0.05436, "grad_norm": 1.1873356578446181, "learning_rate": 0.003, "loss": 4.142, "step": 5436 }, { "epoch": 0.05437, "grad_norm": 0.9961938731710348, "learning_rate": 0.003, "loss": 4.1407, "step": 5437 }, { "epoch": 0.05438, "grad_norm": 1.116985738059544, "learning_rate": 0.003, "loss": 4.1315, "step": 5438 }, { "epoch": 0.05439, "grad_norm": 1.013140285858862, "learning_rate": 0.003, "loss": 4.1226, "step": 5439 }, { "epoch": 0.0544, "grad_norm": 0.97878397740082, "learning_rate": 0.003, "loss": 4.1159, "step": 5440 }, { "epoch": 0.05441, "grad_norm": 1.047952260173831, "learning_rate": 0.003, "loss": 4.1485, "step": 5441 }, { "epoch": 0.05442, "grad_norm": 1.0168380196792572, "learning_rate": 0.003, "loss": 4.11, "step": 5442 }, { "epoch": 0.05443, "grad_norm": 0.8830876213922452, "learning_rate": 0.003, "loss": 4.1231, "step": 5443 }, { "epoch": 0.05444, "grad_norm": 0.8682786365558881, "learning_rate": 0.003, "loss": 4.1114, "step": 5444 }, { "epoch": 0.05445, "grad_norm": 0.9756475817613688, "learning_rate": 0.003, "loss": 4.1128, "step": 5445 }, { "epoch": 0.05446, "grad_norm": 0.9934036218079088, "learning_rate": 0.003, "loss": 4.1167, "step": 5446 }, { "epoch": 0.05447, "grad_norm": 0.9369260577718805, "learning_rate": 0.003, "loss": 4.1074, "step": 5447 }, { "epoch": 0.05448, "grad_norm": 0.9630471244706246, "learning_rate": 0.003, "loss": 4.1466, "step": 5448 }, { "epoch": 0.05449, "grad_norm": 0.8852681416751951, "learning_rate": 0.003, "loss": 4.135, "step": 5449 }, { "epoch": 0.0545, "grad_norm": 0.8723516400396613, "learning_rate": 0.003, "loss": 4.1285, "step": 5450 }, { "epoch": 0.05451, "grad_norm": 0.7744610055329229, "learning_rate": 0.003, "loss": 4.1513, "step": 5451 }, { "epoch": 0.05452, "grad_norm": 0.7334830805607475, "learning_rate": 0.003, "loss": 4.1144, "step": 5452 }, { "epoch": 0.05453, "grad_norm": 0.6840465052326917, "learning_rate": 0.003, "loss": 4.1069, "step": 5453 }, { "epoch": 0.05454, "grad_norm": 0.7047366215451967, "learning_rate": 0.003, "loss": 4.1131, "step": 5454 }, { "epoch": 0.05455, "grad_norm": 0.7373515590693615, "learning_rate": 0.003, "loss": 4.137, "step": 5455 }, { "epoch": 0.05456, "grad_norm": 0.7466741982652629, "learning_rate": 0.003, "loss": 4.1111, "step": 5456 }, { "epoch": 0.05457, "grad_norm": 0.9180906670943273, "learning_rate": 0.003, "loss": 4.137, "step": 5457 }, { "epoch": 0.05458, "grad_norm": 1.344651162550162, "learning_rate": 0.003, "loss": 4.1195, "step": 5458 }, { "epoch": 0.05459, "grad_norm": 0.9718892443666932, "learning_rate": 0.003, "loss": 4.1279, "step": 5459 }, { "epoch": 0.0546, "grad_norm": 1.1893468224298873, "learning_rate": 0.003, "loss": 4.1244, "step": 5460 }, { "epoch": 0.05461, "grad_norm": 0.9099724886418035, "learning_rate": 0.003, "loss": 4.1198, "step": 5461 }, { "epoch": 0.05462, "grad_norm": 0.8909858664433765, "learning_rate": 0.003, "loss": 4.1452, "step": 5462 }, { "epoch": 0.05463, "grad_norm": 0.943471639674701, "learning_rate": 0.003, "loss": 4.108, "step": 5463 }, { "epoch": 0.05464, "grad_norm": 1.006359112858614, "learning_rate": 0.003, "loss": 4.1475, "step": 5464 }, { "epoch": 0.05465, "grad_norm": 0.9070328242868714, "learning_rate": 0.003, "loss": 4.1421, "step": 5465 }, { "epoch": 0.05466, "grad_norm": 0.802164321014551, "learning_rate": 0.003, "loss": 4.1469, "step": 5466 }, { "epoch": 0.05467, "grad_norm": 0.9316649207868387, "learning_rate": 0.003, "loss": 4.1292, "step": 5467 }, { "epoch": 0.05468, "grad_norm": 1.353349775155964, "learning_rate": 0.003, "loss": 4.1174, "step": 5468 }, { "epoch": 0.05469, "grad_norm": 0.9167605119347981, "learning_rate": 0.003, "loss": 4.1036, "step": 5469 }, { "epoch": 0.0547, "grad_norm": 0.978636038190918, "learning_rate": 0.003, "loss": 4.1078, "step": 5470 }, { "epoch": 0.05471, "grad_norm": 1.1336275611473108, "learning_rate": 0.003, "loss": 4.1237, "step": 5471 }, { "epoch": 0.05472, "grad_norm": 1.0636357857736143, "learning_rate": 0.003, "loss": 4.1176, "step": 5472 }, { "epoch": 0.05473, "grad_norm": 0.9598520665828051, "learning_rate": 0.003, "loss": 4.1353, "step": 5473 }, { "epoch": 0.05474, "grad_norm": 1.2065527518151113, "learning_rate": 0.003, "loss": 4.1192, "step": 5474 }, { "epoch": 0.05475, "grad_norm": 0.9321565799401134, "learning_rate": 0.003, "loss": 4.1726, "step": 5475 }, { "epoch": 0.05476, "grad_norm": 0.9847486405198328, "learning_rate": 0.003, "loss": 4.0957, "step": 5476 }, { "epoch": 0.05477, "grad_norm": 1.0242683669210109, "learning_rate": 0.003, "loss": 4.1239, "step": 5477 }, { "epoch": 0.05478, "grad_norm": 1.044507174514191, "learning_rate": 0.003, "loss": 4.1256, "step": 5478 }, { "epoch": 0.05479, "grad_norm": 1.1644979991385647, "learning_rate": 0.003, "loss": 4.1435, "step": 5479 }, { "epoch": 0.0548, "grad_norm": 1.1314239580426855, "learning_rate": 0.003, "loss": 4.1422, "step": 5480 }, { "epoch": 0.05481, "grad_norm": 1.037870781059475, "learning_rate": 0.003, "loss": 4.1251, "step": 5481 }, { "epoch": 0.05482, "grad_norm": 0.8116803859655726, "learning_rate": 0.003, "loss": 4.1141, "step": 5482 }, { "epoch": 0.05483, "grad_norm": 0.8083557186166936, "learning_rate": 0.003, "loss": 4.117, "step": 5483 }, { "epoch": 0.05484, "grad_norm": 0.9299734149282691, "learning_rate": 0.003, "loss": 4.1395, "step": 5484 }, { "epoch": 0.05485, "grad_norm": 1.4501325018415843, "learning_rate": 0.003, "loss": 4.1592, "step": 5485 }, { "epoch": 0.05486, "grad_norm": 0.6907754147935445, "learning_rate": 0.003, "loss": 4.1282, "step": 5486 }, { "epoch": 0.05487, "grad_norm": 0.771065057500987, "learning_rate": 0.003, "loss": 4.1339, "step": 5487 }, { "epoch": 0.05488, "grad_norm": 0.9398223883665248, "learning_rate": 0.003, "loss": 4.1329, "step": 5488 }, { "epoch": 0.05489, "grad_norm": 1.245734722187167, "learning_rate": 0.003, "loss": 4.1595, "step": 5489 }, { "epoch": 0.0549, "grad_norm": 1.0989670327525296, "learning_rate": 0.003, "loss": 4.1312, "step": 5490 }, { "epoch": 0.05491, "grad_norm": 0.9071445359587079, "learning_rate": 0.003, "loss": 4.0936, "step": 5491 }, { "epoch": 0.05492, "grad_norm": 0.8397438252768895, "learning_rate": 0.003, "loss": 4.1416, "step": 5492 }, { "epoch": 0.05493, "grad_norm": 0.8946981978524265, "learning_rate": 0.003, "loss": 4.1079, "step": 5493 }, { "epoch": 0.05494, "grad_norm": 0.8180794667689825, "learning_rate": 0.003, "loss": 4.1354, "step": 5494 }, { "epoch": 0.05495, "grad_norm": 0.8005989617772755, "learning_rate": 0.003, "loss": 4.1196, "step": 5495 }, { "epoch": 0.05496, "grad_norm": 0.7786588826073861, "learning_rate": 0.003, "loss": 4.1112, "step": 5496 }, { "epoch": 0.05497, "grad_norm": 0.7956028398493934, "learning_rate": 0.003, "loss": 4.1333, "step": 5497 }, { "epoch": 0.05498, "grad_norm": 0.7918941742902131, "learning_rate": 0.003, "loss": 4.1132, "step": 5498 }, { "epoch": 0.05499, "grad_norm": 0.9382930732854762, "learning_rate": 0.003, "loss": 4.1015, "step": 5499 }, { "epoch": 0.055, "grad_norm": 1.126573950621417, "learning_rate": 0.003, "loss": 4.1117, "step": 5500 }, { "epoch": 0.05501, "grad_norm": 1.003864850005161, "learning_rate": 0.003, "loss": 4.1193, "step": 5501 }, { "epoch": 0.05502, "grad_norm": 1.1422394527747446, "learning_rate": 0.003, "loss": 4.1102, "step": 5502 }, { "epoch": 0.05503, "grad_norm": 0.7665896617368361, "learning_rate": 0.003, "loss": 4.1069, "step": 5503 }, { "epoch": 0.05504, "grad_norm": 0.7411461800377359, "learning_rate": 0.003, "loss": 4.1551, "step": 5504 }, { "epoch": 0.05505, "grad_norm": 0.7963833583959748, "learning_rate": 0.003, "loss": 4.1095, "step": 5505 }, { "epoch": 0.05506, "grad_norm": 0.9257832028586459, "learning_rate": 0.003, "loss": 4.0881, "step": 5506 }, { "epoch": 0.05507, "grad_norm": 1.10578776496078, "learning_rate": 0.003, "loss": 4.1524, "step": 5507 }, { "epoch": 0.05508, "grad_norm": 1.0431087201167466, "learning_rate": 0.003, "loss": 4.1283, "step": 5508 }, { "epoch": 0.05509, "grad_norm": 1.0367422108110553, "learning_rate": 0.003, "loss": 4.1275, "step": 5509 }, { "epoch": 0.0551, "grad_norm": 1.2529842317031787, "learning_rate": 0.003, "loss": 4.1118, "step": 5510 }, { "epoch": 0.05511, "grad_norm": 0.9691541938141458, "learning_rate": 0.003, "loss": 4.1333, "step": 5511 }, { "epoch": 0.05512, "grad_norm": 0.864716463133856, "learning_rate": 0.003, "loss": 4.1352, "step": 5512 }, { "epoch": 0.05513, "grad_norm": 0.8354141565098123, "learning_rate": 0.003, "loss": 4.1195, "step": 5513 }, { "epoch": 0.05514, "grad_norm": 1.0376042919809385, "learning_rate": 0.003, "loss": 4.1334, "step": 5514 }, { "epoch": 0.05515, "grad_norm": 0.893982497355944, "learning_rate": 0.003, "loss": 4.1127, "step": 5515 }, { "epoch": 0.05516, "grad_norm": 0.8092294335938044, "learning_rate": 0.003, "loss": 4.145, "step": 5516 }, { "epoch": 0.05517, "grad_norm": 0.8335157315875446, "learning_rate": 0.003, "loss": 4.1049, "step": 5517 }, { "epoch": 0.05518, "grad_norm": 0.8631670789345905, "learning_rate": 0.003, "loss": 4.1252, "step": 5518 }, { "epoch": 0.05519, "grad_norm": 0.772502136816785, "learning_rate": 0.003, "loss": 4.15, "step": 5519 }, { "epoch": 0.0552, "grad_norm": 0.7318161157773011, "learning_rate": 0.003, "loss": 4.1198, "step": 5520 }, { "epoch": 0.05521, "grad_norm": 0.7805844973360749, "learning_rate": 0.003, "loss": 4.1341, "step": 5521 }, { "epoch": 0.05522, "grad_norm": 0.9610950814102949, "learning_rate": 0.003, "loss": 4.1492, "step": 5522 }, { "epoch": 0.05523, "grad_norm": 1.4983777342001978, "learning_rate": 0.003, "loss": 4.137, "step": 5523 }, { "epoch": 0.05524, "grad_norm": 0.8114506426427744, "learning_rate": 0.003, "loss": 4.1408, "step": 5524 }, { "epoch": 0.05525, "grad_norm": 0.8437663742868677, "learning_rate": 0.003, "loss": 4.1074, "step": 5525 }, { "epoch": 0.05526, "grad_norm": 0.966987301586662, "learning_rate": 0.003, "loss": 4.1134, "step": 5526 }, { "epoch": 0.05527, "grad_norm": 1.2725075684475797, "learning_rate": 0.003, "loss": 4.0848, "step": 5527 }, { "epoch": 0.05528, "grad_norm": 0.8486880982316782, "learning_rate": 0.003, "loss": 4.1243, "step": 5528 }, { "epoch": 0.05529, "grad_norm": 0.9285432185402387, "learning_rate": 0.003, "loss": 4.1354, "step": 5529 }, { "epoch": 0.0553, "grad_norm": 1.1784089544019016, "learning_rate": 0.003, "loss": 4.1526, "step": 5530 }, { "epoch": 0.05531, "grad_norm": 0.7869416792914739, "learning_rate": 0.003, "loss": 4.131, "step": 5531 }, { "epoch": 0.05532, "grad_norm": 0.8846989558217583, "learning_rate": 0.003, "loss": 4.1188, "step": 5532 }, { "epoch": 0.05533, "grad_norm": 0.8833239717000477, "learning_rate": 0.003, "loss": 4.1355, "step": 5533 }, { "epoch": 0.05534, "grad_norm": 0.8715591388693983, "learning_rate": 0.003, "loss": 4.1231, "step": 5534 }, { "epoch": 0.05535, "grad_norm": 0.8133286665003178, "learning_rate": 0.003, "loss": 4.134, "step": 5535 }, { "epoch": 0.05536, "grad_norm": 0.926009595893208, "learning_rate": 0.003, "loss": 4.129, "step": 5536 }, { "epoch": 0.05537, "grad_norm": 1.052877549031079, "learning_rate": 0.003, "loss": 4.1274, "step": 5537 }, { "epoch": 0.05538, "grad_norm": 0.8636003611383188, "learning_rate": 0.003, "loss": 4.1204, "step": 5538 }, { "epoch": 0.05539, "grad_norm": 0.8087447171104265, "learning_rate": 0.003, "loss": 4.1044, "step": 5539 }, { "epoch": 0.0554, "grad_norm": 0.898241169296187, "learning_rate": 0.003, "loss": 4.1631, "step": 5540 }, { "epoch": 0.05541, "grad_norm": 1.228621032338734, "learning_rate": 0.003, "loss": 4.1482, "step": 5541 }, { "epoch": 0.05542, "grad_norm": 0.908807188172974, "learning_rate": 0.003, "loss": 4.1133, "step": 5542 }, { "epoch": 0.05543, "grad_norm": 1.1675594774032338, "learning_rate": 0.003, "loss": 4.1127, "step": 5543 }, { "epoch": 0.05544, "grad_norm": 1.1933027347530332, "learning_rate": 0.003, "loss": 4.1574, "step": 5544 }, { "epoch": 0.05545, "grad_norm": 0.9571394512885866, "learning_rate": 0.003, "loss": 4.1393, "step": 5545 }, { "epoch": 0.05546, "grad_norm": 0.8590171381737576, "learning_rate": 0.003, "loss": 4.1219, "step": 5546 }, { "epoch": 0.05547, "grad_norm": 1.0666304751642304, "learning_rate": 0.003, "loss": 4.1214, "step": 5547 }, { "epoch": 0.05548, "grad_norm": 0.9704599043567833, "learning_rate": 0.003, "loss": 4.0967, "step": 5548 }, { "epoch": 0.05549, "grad_norm": 0.9865568862830414, "learning_rate": 0.003, "loss": 4.1432, "step": 5549 }, { "epoch": 0.0555, "grad_norm": 0.9885534149053431, "learning_rate": 0.003, "loss": 4.1234, "step": 5550 }, { "epoch": 0.05551, "grad_norm": 0.9749472205384399, "learning_rate": 0.003, "loss": 4.0895, "step": 5551 }, { "epoch": 0.05552, "grad_norm": 1.026944712612127, "learning_rate": 0.003, "loss": 4.1135, "step": 5552 }, { "epoch": 0.05553, "grad_norm": 1.1445015380207735, "learning_rate": 0.003, "loss": 4.1373, "step": 5553 }, { "epoch": 0.05554, "grad_norm": 1.0166458533215852, "learning_rate": 0.003, "loss": 4.1198, "step": 5554 }, { "epoch": 0.05555, "grad_norm": 1.0654603455457914, "learning_rate": 0.003, "loss": 4.1641, "step": 5555 }, { "epoch": 0.05556, "grad_norm": 1.0202184347102567, "learning_rate": 0.003, "loss": 4.1137, "step": 5556 }, { "epoch": 0.05557, "grad_norm": 1.0187438847348156, "learning_rate": 0.003, "loss": 4.1451, "step": 5557 }, { "epoch": 0.05558, "grad_norm": 1.1056836731423136, "learning_rate": 0.003, "loss": 4.1013, "step": 5558 }, { "epoch": 0.05559, "grad_norm": 1.0395503189831727, "learning_rate": 0.003, "loss": 4.1333, "step": 5559 }, { "epoch": 0.0556, "grad_norm": 1.0987107991051774, "learning_rate": 0.003, "loss": 4.1596, "step": 5560 }, { "epoch": 0.05561, "grad_norm": 0.9414302320280178, "learning_rate": 0.003, "loss": 4.1105, "step": 5561 }, { "epoch": 0.05562, "grad_norm": 1.080665983026407, "learning_rate": 0.003, "loss": 4.1262, "step": 5562 }, { "epoch": 0.05563, "grad_norm": 1.0503459917604645, "learning_rate": 0.003, "loss": 4.1356, "step": 5563 }, { "epoch": 0.05564, "grad_norm": 1.1247898961225558, "learning_rate": 0.003, "loss": 4.1203, "step": 5564 }, { "epoch": 0.05565, "grad_norm": 1.0361569490033156, "learning_rate": 0.003, "loss": 4.1594, "step": 5565 }, { "epoch": 0.05566, "grad_norm": 0.993562704860863, "learning_rate": 0.003, "loss": 4.1074, "step": 5566 }, { "epoch": 0.05567, "grad_norm": 0.9350452216561642, "learning_rate": 0.003, "loss": 4.1168, "step": 5567 }, { "epoch": 0.05568, "grad_norm": 0.9246250113104997, "learning_rate": 0.003, "loss": 4.1453, "step": 5568 }, { "epoch": 0.05569, "grad_norm": 1.1010360382158253, "learning_rate": 0.003, "loss": 4.122, "step": 5569 }, { "epoch": 0.0557, "grad_norm": 1.014573516129285, "learning_rate": 0.003, "loss": 4.1435, "step": 5570 }, { "epoch": 0.05571, "grad_norm": 1.0901500909260304, "learning_rate": 0.003, "loss": 4.1345, "step": 5571 }, { "epoch": 0.05572, "grad_norm": 1.0639295197328618, "learning_rate": 0.003, "loss": 4.1118, "step": 5572 }, { "epoch": 0.05573, "grad_norm": 1.160526318650138, "learning_rate": 0.003, "loss": 4.1559, "step": 5573 }, { "epoch": 0.05574, "grad_norm": 0.8017141002983521, "learning_rate": 0.003, "loss": 4.1171, "step": 5574 }, { "epoch": 0.05575, "grad_norm": 0.7725242603041408, "learning_rate": 0.003, "loss": 4.1173, "step": 5575 }, { "epoch": 0.05576, "grad_norm": 0.7959745021533514, "learning_rate": 0.003, "loss": 4.092, "step": 5576 }, { "epoch": 0.05577, "grad_norm": 0.8656786722945465, "learning_rate": 0.003, "loss": 4.1159, "step": 5577 }, { "epoch": 0.05578, "grad_norm": 1.129061918601625, "learning_rate": 0.003, "loss": 4.1233, "step": 5578 }, { "epoch": 0.05579, "grad_norm": 1.0600753924801507, "learning_rate": 0.003, "loss": 4.1392, "step": 5579 }, { "epoch": 0.0558, "grad_norm": 0.9542665048147827, "learning_rate": 0.003, "loss": 4.1267, "step": 5580 }, { "epoch": 0.05581, "grad_norm": 1.0248470489985253, "learning_rate": 0.003, "loss": 4.1402, "step": 5581 }, { "epoch": 0.05582, "grad_norm": 1.2221203414392114, "learning_rate": 0.003, "loss": 4.1585, "step": 5582 }, { "epoch": 0.05583, "grad_norm": 0.9829928529956432, "learning_rate": 0.003, "loss": 4.1242, "step": 5583 }, { "epoch": 0.05584, "grad_norm": 1.0416914686566905, "learning_rate": 0.003, "loss": 4.1263, "step": 5584 }, { "epoch": 0.05585, "grad_norm": 1.185036141605951, "learning_rate": 0.003, "loss": 4.1023, "step": 5585 }, { "epoch": 0.05586, "grad_norm": 0.8882890058762307, "learning_rate": 0.003, "loss": 4.1356, "step": 5586 }, { "epoch": 0.05587, "grad_norm": 0.8853504241990832, "learning_rate": 0.003, "loss": 4.1399, "step": 5587 }, { "epoch": 0.05588, "grad_norm": 0.9145372711332053, "learning_rate": 0.003, "loss": 4.1136, "step": 5588 }, { "epoch": 0.05589, "grad_norm": 1.098734879665573, "learning_rate": 0.003, "loss": 4.1284, "step": 5589 }, { "epoch": 0.0559, "grad_norm": 1.0306332486921068, "learning_rate": 0.003, "loss": 4.1255, "step": 5590 }, { "epoch": 0.05591, "grad_norm": 0.9663756117466623, "learning_rate": 0.003, "loss": 4.1338, "step": 5591 }, { "epoch": 0.05592, "grad_norm": 0.9707121490476076, "learning_rate": 0.003, "loss": 4.132, "step": 5592 }, { "epoch": 0.05593, "grad_norm": 0.9040080758689395, "learning_rate": 0.003, "loss": 4.1185, "step": 5593 }, { "epoch": 0.05594, "grad_norm": 0.8399062436913435, "learning_rate": 0.003, "loss": 4.1095, "step": 5594 }, { "epoch": 0.05595, "grad_norm": 0.7927817251689244, "learning_rate": 0.003, "loss": 4.0906, "step": 5595 }, { "epoch": 0.05596, "grad_norm": 0.7554549829981407, "learning_rate": 0.003, "loss": 4.1037, "step": 5596 }, { "epoch": 0.05597, "grad_norm": 0.8263782174776544, "learning_rate": 0.003, "loss": 4.1334, "step": 5597 }, { "epoch": 0.05598, "grad_norm": 1.0617645805745923, "learning_rate": 0.003, "loss": 4.1, "step": 5598 }, { "epoch": 0.05599, "grad_norm": 1.0105155570821656, "learning_rate": 0.003, "loss": 4.1452, "step": 5599 }, { "epoch": 0.056, "grad_norm": 1.0218581737778532, "learning_rate": 0.003, "loss": 4.0885, "step": 5600 }, { "epoch": 0.05601, "grad_norm": 1.0778280577751203, "learning_rate": 0.003, "loss": 4.1079, "step": 5601 }, { "epoch": 0.05602, "grad_norm": 0.7582012801520183, "learning_rate": 0.003, "loss": 4.0943, "step": 5602 }, { "epoch": 0.05603, "grad_norm": 0.7402260100117943, "learning_rate": 0.003, "loss": 4.1057, "step": 5603 }, { "epoch": 0.05604, "grad_norm": 0.8873650869380406, "learning_rate": 0.003, "loss": 4.1446, "step": 5604 }, { "epoch": 0.05605, "grad_norm": 1.0177321840014748, "learning_rate": 0.003, "loss": 4.0795, "step": 5605 }, { "epoch": 0.05606, "grad_norm": 1.0177155480228988, "learning_rate": 0.003, "loss": 4.1354, "step": 5606 }, { "epoch": 0.05607, "grad_norm": 1.005812598468043, "learning_rate": 0.003, "loss": 4.152, "step": 5607 }, { "epoch": 0.05608, "grad_norm": 1.345264690947568, "learning_rate": 0.003, "loss": 4.0818, "step": 5608 }, { "epoch": 0.05609, "grad_norm": 0.9420214751930821, "learning_rate": 0.003, "loss": 4.1213, "step": 5609 }, { "epoch": 0.0561, "grad_norm": 1.2061944377166507, "learning_rate": 0.003, "loss": 4.1644, "step": 5610 }, { "epoch": 0.05611, "grad_norm": 1.0861930497452696, "learning_rate": 0.003, "loss": 4.1313, "step": 5611 }, { "epoch": 0.05612, "grad_norm": 0.8693052542154518, "learning_rate": 0.003, "loss": 4.1172, "step": 5612 }, { "epoch": 0.05613, "grad_norm": 0.8252162892679517, "learning_rate": 0.003, "loss": 4.1428, "step": 5613 }, { "epoch": 0.05614, "grad_norm": 0.8162230158140894, "learning_rate": 0.003, "loss": 4.125, "step": 5614 }, { "epoch": 0.05615, "grad_norm": 0.7932415284599255, "learning_rate": 0.003, "loss": 4.09, "step": 5615 }, { "epoch": 0.05616, "grad_norm": 0.8147303370119828, "learning_rate": 0.003, "loss": 4.1067, "step": 5616 }, { "epoch": 0.05617, "grad_norm": 0.979345832369033, "learning_rate": 0.003, "loss": 4.1323, "step": 5617 }, { "epoch": 0.05618, "grad_norm": 1.1939522303555639, "learning_rate": 0.003, "loss": 4.1174, "step": 5618 }, { "epoch": 0.05619, "grad_norm": 0.9785893247462943, "learning_rate": 0.003, "loss": 4.1734, "step": 5619 }, { "epoch": 0.0562, "grad_norm": 0.8751466109554733, "learning_rate": 0.003, "loss": 4.1216, "step": 5620 }, { "epoch": 0.05621, "grad_norm": 0.9221389977798955, "learning_rate": 0.003, "loss": 4.1357, "step": 5621 }, { "epoch": 0.05622, "grad_norm": 1.0075873675251126, "learning_rate": 0.003, "loss": 4.1178, "step": 5622 }, { "epoch": 0.05623, "grad_norm": 1.1949478525553667, "learning_rate": 0.003, "loss": 4.1438, "step": 5623 }, { "epoch": 0.05624, "grad_norm": 0.8893580304093707, "learning_rate": 0.003, "loss": 4.1524, "step": 5624 }, { "epoch": 0.05625, "grad_norm": 0.9101874793046152, "learning_rate": 0.003, "loss": 4.1489, "step": 5625 }, { "epoch": 0.05626, "grad_norm": 0.8568178177963297, "learning_rate": 0.003, "loss": 4.113, "step": 5626 }, { "epoch": 0.05627, "grad_norm": 0.8445550632051778, "learning_rate": 0.003, "loss": 4.1228, "step": 5627 }, { "epoch": 0.05628, "grad_norm": 0.8395278921688747, "learning_rate": 0.003, "loss": 4.1175, "step": 5628 }, { "epoch": 0.05629, "grad_norm": 0.8238415014776045, "learning_rate": 0.003, "loss": 4.1277, "step": 5629 }, { "epoch": 0.0563, "grad_norm": 0.9266145959778516, "learning_rate": 0.003, "loss": 4.1101, "step": 5630 }, { "epoch": 0.05631, "grad_norm": 1.041049278487491, "learning_rate": 0.003, "loss": 4.0999, "step": 5631 }, { "epoch": 0.05632, "grad_norm": 1.2634529953673648, "learning_rate": 0.003, "loss": 4.1399, "step": 5632 }, { "epoch": 0.05633, "grad_norm": 0.8743820927721512, "learning_rate": 0.003, "loss": 4.1277, "step": 5633 }, { "epoch": 0.05634, "grad_norm": 0.9946748940515425, "learning_rate": 0.003, "loss": 4.1125, "step": 5634 }, { "epoch": 0.05635, "grad_norm": 1.0906818511461622, "learning_rate": 0.003, "loss": 4.1357, "step": 5635 }, { "epoch": 0.05636, "grad_norm": 0.9660961085376262, "learning_rate": 0.003, "loss": 4.1248, "step": 5636 }, { "epoch": 0.05637, "grad_norm": 0.9695551749489936, "learning_rate": 0.003, "loss": 4.1638, "step": 5637 }, { "epoch": 0.05638, "grad_norm": 0.8858313611285864, "learning_rate": 0.003, "loss": 4.1107, "step": 5638 }, { "epoch": 0.05639, "grad_norm": 0.9018585489980617, "learning_rate": 0.003, "loss": 4.1202, "step": 5639 }, { "epoch": 0.0564, "grad_norm": 0.8861660445176881, "learning_rate": 0.003, "loss": 4.1295, "step": 5640 }, { "epoch": 0.05641, "grad_norm": 0.8951943773644586, "learning_rate": 0.003, "loss": 4.0811, "step": 5641 }, { "epoch": 0.05642, "grad_norm": 0.9083819743767668, "learning_rate": 0.003, "loss": 4.1477, "step": 5642 }, { "epoch": 0.05643, "grad_norm": 0.962671004199545, "learning_rate": 0.003, "loss": 4.1087, "step": 5643 }, { "epoch": 0.05644, "grad_norm": 1.077396446555887, "learning_rate": 0.003, "loss": 4.1206, "step": 5644 }, { "epoch": 0.05645, "grad_norm": 1.40715572500545, "learning_rate": 0.003, "loss": 4.1262, "step": 5645 }, { "epoch": 0.05646, "grad_norm": 0.8175124218911509, "learning_rate": 0.003, "loss": 4.1223, "step": 5646 }, { "epoch": 0.05647, "grad_norm": 0.7832062109640497, "learning_rate": 0.003, "loss": 4.1204, "step": 5647 }, { "epoch": 0.05648, "grad_norm": 0.9630055472683711, "learning_rate": 0.003, "loss": 4.1381, "step": 5648 }, { "epoch": 0.05649, "grad_norm": 1.0631770670815335, "learning_rate": 0.003, "loss": 4.1268, "step": 5649 }, { "epoch": 0.0565, "grad_norm": 1.124204894656994, "learning_rate": 0.003, "loss": 4.1458, "step": 5650 }, { "epoch": 0.05651, "grad_norm": 1.022339961520506, "learning_rate": 0.003, "loss": 4.1098, "step": 5651 }, { "epoch": 0.05652, "grad_norm": 0.8587439092944441, "learning_rate": 0.003, "loss": 4.1043, "step": 5652 }, { "epoch": 0.05653, "grad_norm": 0.7753822534356805, "learning_rate": 0.003, "loss": 4.1302, "step": 5653 }, { "epoch": 0.05654, "grad_norm": 0.8333542595529039, "learning_rate": 0.003, "loss": 4.139, "step": 5654 }, { "epoch": 0.05655, "grad_norm": 0.9690699423615071, "learning_rate": 0.003, "loss": 4.1319, "step": 5655 }, { "epoch": 0.05656, "grad_norm": 1.1790002076186354, "learning_rate": 0.003, "loss": 4.1337, "step": 5656 }, { "epoch": 0.05657, "grad_norm": 0.8826076546461541, "learning_rate": 0.003, "loss": 4.1253, "step": 5657 }, { "epoch": 0.05658, "grad_norm": 0.8843360394533153, "learning_rate": 0.003, "loss": 4.1167, "step": 5658 }, { "epoch": 0.05659, "grad_norm": 1.0024365762558438, "learning_rate": 0.003, "loss": 4.1219, "step": 5659 }, { "epoch": 0.0566, "grad_norm": 1.0811089362032615, "learning_rate": 0.003, "loss": 4.1309, "step": 5660 }, { "epoch": 0.05661, "grad_norm": 1.0126548273143556, "learning_rate": 0.003, "loss": 4.1648, "step": 5661 }, { "epoch": 0.05662, "grad_norm": 1.1175005958432391, "learning_rate": 0.003, "loss": 4.111, "step": 5662 }, { "epoch": 0.05663, "grad_norm": 0.8123904859528547, "learning_rate": 0.003, "loss": 4.1255, "step": 5663 }, { "epoch": 0.05664, "grad_norm": 0.8486369117778391, "learning_rate": 0.003, "loss": 4.0996, "step": 5664 }, { "epoch": 0.05665, "grad_norm": 1.120613790787452, "learning_rate": 0.003, "loss": 4.1146, "step": 5665 }, { "epoch": 0.05666, "grad_norm": 1.0540993326958734, "learning_rate": 0.003, "loss": 4.1276, "step": 5666 }, { "epoch": 0.05667, "grad_norm": 0.8401662075563899, "learning_rate": 0.003, "loss": 4.0944, "step": 5667 }, { "epoch": 0.05668, "grad_norm": 0.8686439657617465, "learning_rate": 0.003, "loss": 4.1399, "step": 5668 }, { "epoch": 0.05669, "grad_norm": 0.9702787574694016, "learning_rate": 0.003, "loss": 4.0882, "step": 5669 }, { "epoch": 0.0567, "grad_norm": 1.005684782440143, "learning_rate": 0.003, "loss": 4.1283, "step": 5670 }, { "epoch": 0.05671, "grad_norm": 0.8966916733750543, "learning_rate": 0.003, "loss": 4.1121, "step": 5671 }, { "epoch": 0.05672, "grad_norm": 1.1718295219256112, "learning_rate": 0.003, "loss": 4.1199, "step": 5672 }, { "epoch": 0.05673, "grad_norm": 1.2210000503080898, "learning_rate": 0.003, "loss": 4.1025, "step": 5673 }, { "epoch": 0.05674, "grad_norm": 0.9196178304374668, "learning_rate": 0.003, "loss": 4.1317, "step": 5674 }, { "epoch": 0.05675, "grad_norm": 0.9259626109244385, "learning_rate": 0.003, "loss": 4.1204, "step": 5675 }, { "epoch": 0.05676, "grad_norm": 0.8929657175241067, "learning_rate": 0.003, "loss": 4.1289, "step": 5676 }, { "epoch": 0.05677, "grad_norm": 0.952108119774777, "learning_rate": 0.003, "loss": 4.1258, "step": 5677 }, { "epoch": 0.05678, "grad_norm": 1.1126100059487607, "learning_rate": 0.003, "loss": 4.126, "step": 5678 }, { "epoch": 0.05679, "grad_norm": 1.0364911564982762, "learning_rate": 0.003, "loss": 4.1416, "step": 5679 }, { "epoch": 0.0568, "grad_norm": 1.3191412104281883, "learning_rate": 0.003, "loss": 4.13, "step": 5680 }, { "epoch": 0.05681, "grad_norm": 0.7736101131054048, "learning_rate": 0.003, "loss": 4.1193, "step": 5681 }, { "epoch": 0.05682, "grad_norm": 0.7593487398147124, "learning_rate": 0.003, "loss": 4.1314, "step": 5682 }, { "epoch": 0.05683, "grad_norm": 0.8129218561244385, "learning_rate": 0.003, "loss": 4.0869, "step": 5683 }, { "epoch": 0.05684, "grad_norm": 1.0333497384359716, "learning_rate": 0.003, "loss": 4.1046, "step": 5684 }, { "epoch": 0.05685, "grad_norm": 1.43332981994489, "learning_rate": 0.003, "loss": 4.1523, "step": 5685 }, { "epoch": 0.05686, "grad_norm": 0.8430936458304196, "learning_rate": 0.003, "loss": 4.1449, "step": 5686 }, { "epoch": 0.05687, "grad_norm": 0.8954710874656873, "learning_rate": 0.003, "loss": 4.0975, "step": 5687 }, { "epoch": 0.05688, "grad_norm": 1.0730066166878027, "learning_rate": 0.003, "loss": 4.121, "step": 5688 }, { "epoch": 0.05689, "grad_norm": 1.08532099527152, "learning_rate": 0.003, "loss": 4.1011, "step": 5689 }, { "epoch": 0.0569, "grad_norm": 0.9067090942223068, "learning_rate": 0.003, "loss": 4.1226, "step": 5690 }, { "epoch": 0.05691, "grad_norm": 1.035726205384705, "learning_rate": 0.003, "loss": 4.1558, "step": 5691 }, { "epoch": 0.05692, "grad_norm": 0.9336873866149337, "learning_rate": 0.003, "loss": 4.1291, "step": 5692 }, { "epoch": 0.05693, "grad_norm": 0.8773378995984399, "learning_rate": 0.003, "loss": 4.1249, "step": 5693 }, { "epoch": 0.05694, "grad_norm": 0.9097271093934071, "learning_rate": 0.003, "loss": 4.1069, "step": 5694 }, { "epoch": 0.05695, "grad_norm": 1.0985181967102629, "learning_rate": 0.003, "loss": 4.11, "step": 5695 }, { "epoch": 0.05696, "grad_norm": 1.040087310747754, "learning_rate": 0.003, "loss": 4.1162, "step": 5696 }, { "epoch": 0.05697, "grad_norm": 0.9817632213668405, "learning_rate": 0.003, "loss": 4.1512, "step": 5697 }, { "epoch": 0.05698, "grad_norm": 1.0569743573990606, "learning_rate": 0.003, "loss": 4.1475, "step": 5698 }, { "epoch": 0.05699, "grad_norm": 1.0902526090916207, "learning_rate": 0.003, "loss": 4.1324, "step": 5699 }, { "epoch": 0.057, "grad_norm": 0.8928105595260728, "learning_rate": 0.003, "loss": 4.1085, "step": 5700 }, { "epoch": 0.05701, "grad_norm": 0.7369786806129934, "learning_rate": 0.003, "loss": 4.1294, "step": 5701 }, { "epoch": 0.05702, "grad_norm": 0.7065919886418875, "learning_rate": 0.003, "loss": 4.106, "step": 5702 }, { "epoch": 0.05703, "grad_norm": 0.7357229062821155, "learning_rate": 0.003, "loss": 4.1494, "step": 5703 }, { "epoch": 0.05704, "grad_norm": 0.771783060800162, "learning_rate": 0.003, "loss": 4.1087, "step": 5704 }, { "epoch": 0.05705, "grad_norm": 0.8020733794571436, "learning_rate": 0.003, "loss": 4.1077, "step": 5705 }, { "epoch": 0.05706, "grad_norm": 0.8005778172331136, "learning_rate": 0.003, "loss": 4.1043, "step": 5706 }, { "epoch": 0.05707, "grad_norm": 0.8597236364014955, "learning_rate": 0.003, "loss": 4.1054, "step": 5707 }, { "epoch": 0.05708, "grad_norm": 1.063034575015758, "learning_rate": 0.003, "loss": 4.1468, "step": 5708 }, { "epoch": 0.05709, "grad_norm": 1.0484123566982753, "learning_rate": 0.003, "loss": 4.1188, "step": 5709 }, { "epoch": 0.0571, "grad_norm": 1.1382736991524067, "learning_rate": 0.003, "loss": 4.1043, "step": 5710 }, { "epoch": 0.05711, "grad_norm": 0.9787958240785122, "learning_rate": 0.003, "loss": 4.1303, "step": 5711 }, { "epoch": 0.05712, "grad_norm": 0.9904719131026177, "learning_rate": 0.003, "loss": 4.1071, "step": 5712 }, { "epoch": 0.05713, "grad_norm": 1.0238241408520392, "learning_rate": 0.003, "loss": 4.1079, "step": 5713 }, { "epoch": 0.05714, "grad_norm": 1.0007244420082209, "learning_rate": 0.003, "loss": 4.1626, "step": 5714 }, { "epoch": 0.05715, "grad_norm": 1.1489073266379561, "learning_rate": 0.003, "loss": 4.1692, "step": 5715 }, { "epoch": 0.05716, "grad_norm": 1.1100927142481192, "learning_rate": 0.003, "loss": 4.1411, "step": 5716 }, { "epoch": 0.05717, "grad_norm": 0.8696259900880453, "learning_rate": 0.003, "loss": 4.1464, "step": 5717 }, { "epoch": 0.05718, "grad_norm": 0.9371133901319081, "learning_rate": 0.003, "loss": 4.1285, "step": 5718 }, { "epoch": 0.05719, "grad_norm": 0.931314660567155, "learning_rate": 0.003, "loss": 4.1304, "step": 5719 }, { "epoch": 0.0572, "grad_norm": 0.9782165711056069, "learning_rate": 0.003, "loss": 4.1183, "step": 5720 }, { "epoch": 0.05721, "grad_norm": 1.0886623888279425, "learning_rate": 0.003, "loss": 4.1067, "step": 5721 }, { "epoch": 0.05722, "grad_norm": 0.9997773216929959, "learning_rate": 0.003, "loss": 4.1158, "step": 5722 }, { "epoch": 0.05723, "grad_norm": 0.9514627056664299, "learning_rate": 0.003, "loss": 4.1159, "step": 5723 }, { "epoch": 0.05724, "grad_norm": 0.8230402798369292, "learning_rate": 0.003, "loss": 4.1148, "step": 5724 }, { "epoch": 0.05725, "grad_norm": 0.8987819684221542, "learning_rate": 0.003, "loss": 4.1296, "step": 5725 }, { "epoch": 0.05726, "grad_norm": 0.9892002123491518, "learning_rate": 0.003, "loss": 4.1212, "step": 5726 }, { "epoch": 0.05727, "grad_norm": 1.0469586088251535, "learning_rate": 0.003, "loss": 4.1129, "step": 5727 }, { "epoch": 0.05728, "grad_norm": 1.0973205641994594, "learning_rate": 0.003, "loss": 4.1431, "step": 5728 }, { "epoch": 0.05729, "grad_norm": 1.0064407627655658, "learning_rate": 0.003, "loss": 4.122, "step": 5729 }, { "epoch": 0.0573, "grad_norm": 1.110713311982964, "learning_rate": 0.003, "loss": 4.097, "step": 5730 }, { "epoch": 0.05731, "grad_norm": 1.0241807049928666, "learning_rate": 0.003, "loss": 4.1155, "step": 5731 }, { "epoch": 0.05732, "grad_norm": 1.0429234117415462, "learning_rate": 0.003, "loss": 4.1449, "step": 5732 }, { "epoch": 0.05733, "grad_norm": 0.9357269471087896, "learning_rate": 0.003, "loss": 4.1531, "step": 5733 }, { "epoch": 0.05734, "grad_norm": 1.1141629712170185, "learning_rate": 0.003, "loss": 4.1542, "step": 5734 }, { "epoch": 0.05735, "grad_norm": 0.8298330992050329, "learning_rate": 0.003, "loss": 4.1074, "step": 5735 }, { "epoch": 0.05736, "grad_norm": 0.8744860650234516, "learning_rate": 0.003, "loss": 4.1347, "step": 5736 }, { "epoch": 0.05737, "grad_norm": 0.8447450239136927, "learning_rate": 0.003, "loss": 4.1254, "step": 5737 }, { "epoch": 0.05738, "grad_norm": 0.9189790636785452, "learning_rate": 0.003, "loss": 4.1005, "step": 5738 }, { "epoch": 0.05739, "grad_norm": 0.9705384160287442, "learning_rate": 0.003, "loss": 4.1097, "step": 5739 }, { "epoch": 0.0574, "grad_norm": 0.9376119011389905, "learning_rate": 0.003, "loss": 4.1143, "step": 5740 }, { "epoch": 0.05741, "grad_norm": 1.1362834727779139, "learning_rate": 0.003, "loss": 4.1069, "step": 5741 }, { "epoch": 0.05742, "grad_norm": 1.2370365233043261, "learning_rate": 0.003, "loss": 4.1287, "step": 5742 }, { "epoch": 0.05743, "grad_norm": 1.3113048377687142, "learning_rate": 0.003, "loss": 4.1243, "step": 5743 }, { "epoch": 0.05744, "grad_norm": 0.8776507234514279, "learning_rate": 0.003, "loss": 4.1266, "step": 5744 }, { "epoch": 0.05745, "grad_norm": 0.9245134319819897, "learning_rate": 0.003, "loss": 4.1273, "step": 5745 }, { "epoch": 0.05746, "grad_norm": 1.0258527613597674, "learning_rate": 0.003, "loss": 4.1127, "step": 5746 }, { "epoch": 0.05747, "grad_norm": 1.0057412762907374, "learning_rate": 0.003, "loss": 4.142, "step": 5747 }, { "epoch": 0.05748, "grad_norm": 1.1704423833777504, "learning_rate": 0.003, "loss": 4.1662, "step": 5748 }, { "epoch": 0.05749, "grad_norm": 0.8033582728084808, "learning_rate": 0.003, "loss": 4.1067, "step": 5749 }, { "epoch": 0.0575, "grad_norm": 0.8202271878884321, "learning_rate": 0.003, "loss": 4.1277, "step": 5750 }, { "epoch": 0.05751, "grad_norm": 0.8323486916968836, "learning_rate": 0.003, "loss": 4.1417, "step": 5751 }, { "epoch": 0.05752, "grad_norm": 0.824247137213286, "learning_rate": 0.003, "loss": 4.1301, "step": 5752 }, { "epoch": 0.05753, "grad_norm": 0.8446556412554472, "learning_rate": 0.003, "loss": 4.1186, "step": 5753 }, { "epoch": 0.05754, "grad_norm": 1.1966126000714083, "learning_rate": 0.003, "loss": 4.1252, "step": 5754 }, { "epoch": 0.05755, "grad_norm": 1.1964226101976614, "learning_rate": 0.003, "loss": 4.109, "step": 5755 }, { "epoch": 0.05756, "grad_norm": 0.9699060929143068, "learning_rate": 0.003, "loss": 4.1407, "step": 5756 }, { "epoch": 0.05757, "grad_norm": 0.8663250789930335, "learning_rate": 0.003, "loss": 4.1228, "step": 5757 }, { "epoch": 0.05758, "grad_norm": 0.8597853728829586, "learning_rate": 0.003, "loss": 4.121, "step": 5758 }, { "epoch": 0.05759, "grad_norm": 0.8256237685530105, "learning_rate": 0.003, "loss": 4.1204, "step": 5759 }, { "epoch": 0.0576, "grad_norm": 0.7431015057566424, "learning_rate": 0.003, "loss": 4.112, "step": 5760 }, { "epoch": 0.05761, "grad_norm": 0.9128267641846647, "learning_rate": 0.003, "loss": 4.1092, "step": 5761 }, { "epoch": 0.05762, "grad_norm": 1.195756627275649, "learning_rate": 0.003, "loss": 4.0996, "step": 5762 }, { "epoch": 0.05763, "grad_norm": 1.141592654618907, "learning_rate": 0.003, "loss": 4.1152, "step": 5763 }, { "epoch": 0.05764, "grad_norm": 0.9724043681560312, "learning_rate": 0.003, "loss": 4.111, "step": 5764 }, { "epoch": 0.05765, "grad_norm": 0.9718984433800117, "learning_rate": 0.003, "loss": 4.1229, "step": 5765 }, { "epoch": 0.05766, "grad_norm": 0.8591409992621984, "learning_rate": 0.003, "loss": 4.1198, "step": 5766 }, { "epoch": 0.05767, "grad_norm": 0.9080480852024645, "learning_rate": 0.003, "loss": 4.109, "step": 5767 }, { "epoch": 0.05768, "grad_norm": 0.9292125393107181, "learning_rate": 0.003, "loss": 4.129, "step": 5768 }, { "epoch": 0.05769, "grad_norm": 1.0176905281603372, "learning_rate": 0.003, "loss": 4.1178, "step": 5769 }, { "epoch": 0.0577, "grad_norm": 1.160814957884824, "learning_rate": 0.003, "loss": 4.1257, "step": 5770 }, { "epoch": 0.05771, "grad_norm": 1.0307295215517236, "learning_rate": 0.003, "loss": 4.1307, "step": 5771 }, { "epoch": 0.05772, "grad_norm": 0.8386292091089582, "learning_rate": 0.003, "loss": 4.1343, "step": 5772 }, { "epoch": 0.05773, "grad_norm": 0.9057048000447152, "learning_rate": 0.003, "loss": 4.1294, "step": 5773 }, { "epoch": 0.05774, "grad_norm": 1.019919318588754, "learning_rate": 0.003, "loss": 4.1087, "step": 5774 }, { "epoch": 0.05775, "grad_norm": 1.0210339610557355, "learning_rate": 0.003, "loss": 4.1171, "step": 5775 }, { "epoch": 0.05776, "grad_norm": 1.1520039772321047, "learning_rate": 0.003, "loss": 4.1115, "step": 5776 }, { "epoch": 0.05777, "grad_norm": 1.010271480309236, "learning_rate": 0.003, "loss": 4.126, "step": 5777 }, { "epoch": 0.05778, "grad_norm": 1.127623586377467, "learning_rate": 0.003, "loss": 4.1136, "step": 5778 }, { "epoch": 0.05779, "grad_norm": 1.1899760224553186, "learning_rate": 0.003, "loss": 4.143, "step": 5779 }, { "epoch": 0.0578, "grad_norm": 0.9780858457407019, "learning_rate": 0.003, "loss": 4.0952, "step": 5780 }, { "epoch": 0.05781, "grad_norm": 1.0370323662272365, "learning_rate": 0.003, "loss": 4.1123, "step": 5781 }, { "epoch": 0.05782, "grad_norm": 1.113436156639303, "learning_rate": 0.003, "loss": 4.1441, "step": 5782 }, { "epoch": 0.05783, "grad_norm": 0.9966451868055693, "learning_rate": 0.003, "loss": 4.0838, "step": 5783 }, { "epoch": 0.05784, "grad_norm": 1.0113209641377017, "learning_rate": 0.003, "loss": 4.1248, "step": 5784 }, { "epoch": 0.05785, "grad_norm": 0.9893055065203489, "learning_rate": 0.003, "loss": 4.1376, "step": 5785 }, { "epoch": 0.05786, "grad_norm": 1.1620422429372013, "learning_rate": 0.003, "loss": 4.1313, "step": 5786 }, { "epoch": 0.05787, "grad_norm": 1.1194385908898823, "learning_rate": 0.003, "loss": 4.1001, "step": 5787 }, { "epoch": 0.05788, "grad_norm": 0.8497725539713678, "learning_rate": 0.003, "loss": 4.1178, "step": 5788 }, { "epoch": 0.05789, "grad_norm": 0.8685312277670877, "learning_rate": 0.003, "loss": 4.116, "step": 5789 }, { "epoch": 0.0579, "grad_norm": 1.0031081903513304, "learning_rate": 0.003, "loss": 4.1364, "step": 5790 }, { "epoch": 0.05791, "grad_norm": 0.8056020601521044, "learning_rate": 0.003, "loss": 4.1092, "step": 5791 }, { "epoch": 0.05792, "grad_norm": 0.8347661618494451, "learning_rate": 0.003, "loss": 4.112, "step": 5792 }, { "epoch": 0.05793, "grad_norm": 0.8137473201924559, "learning_rate": 0.003, "loss": 4.1025, "step": 5793 }, { "epoch": 0.05794, "grad_norm": 0.9720764458835657, "learning_rate": 0.003, "loss": 4.1177, "step": 5794 }, { "epoch": 0.05795, "grad_norm": 1.1854800461039692, "learning_rate": 0.003, "loss": 4.1547, "step": 5795 }, { "epoch": 0.05796, "grad_norm": 0.9757913448880616, "learning_rate": 0.003, "loss": 4.1208, "step": 5796 }, { "epoch": 0.05797, "grad_norm": 1.0457868316612642, "learning_rate": 0.003, "loss": 4.1212, "step": 5797 }, { "epoch": 0.05798, "grad_norm": 0.9558189857089837, "learning_rate": 0.003, "loss": 4.1013, "step": 5798 }, { "epoch": 0.05799, "grad_norm": 0.9226240851035593, "learning_rate": 0.003, "loss": 4.1332, "step": 5799 }, { "epoch": 0.058, "grad_norm": 0.9993623389955384, "learning_rate": 0.003, "loss": 4.1352, "step": 5800 }, { "epoch": 0.05801, "grad_norm": 1.0017244438128723, "learning_rate": 0.003, "loss": 4.1173, "step": 5801 }, { "epoch": 0.05802, "grad_norm": 1.0206300916624451, "learning_rate": 0.003, "loss": 4.1264, "step": 5802 }, { "epoch": 0.05803, "grad_norm": 1.2026383802301528, "learning_rate": 0.003, "loss": 4.0936, "step": 5803 }, { "epoch": 0.05804, "grad_norm": 0.8471292816498621, "learning_rate": 0.003, "loss": 4.0817, "step": 5804 }, { "epoch": 0.05805, "grad_norm": 0.9206281349289829, "learning_rate": 0.003, "loss": 4.091, "step": 5805 }, { "epoch": 0.05806, "grad_norm": 0.9920161125208922, "learning_rate": 0.003, "loss": 4.1292, "step": 5806 }, { "epoch": 0.05807, "grad_norm": 1.1340630160886371, "learning_rate": 0.003, "loss": 4.1416, "step": 5807 }, { "epoch": 0.05808, "grad_norm": 1.0539888997436435, "learning_rate": 0.003, "loss": 4.1508, "step": 5808 }, { "epoch": 0.05809, "grad_norm": 0.9733624631098186, "learning_rate": 0.003, "loss": 4.1516, "step": 5809 }, { "epoch": 0.0581, "grad_norm": 1.019182599105166, "learning_rate": 0.003, "loss": 4.0987, "step": 5810 }, { "epoch": 0.05811, "grad_norm": 0.8351011669100157, "learning_rate": 0.003, "loss": 4.1086, "step": 5811 }, { "epoch": 0.05812, "grad_norm": 0.9144093431141755, "learning_rate": 0.003, "loss": 4.1135, "step": 5812 }, { "epoch": 0.05813, "grad_norm": 0.9946389439099612, "learning_rate": 0.003, "loss": 4.1247, "step": 5813 }, { "epoch": 0.05814, "grad_norm": 1.0550398914686734, "learning_rate": 0.003, "loss": 4.1072, "step": 5814 }, { "epoch": 0.05815, "grad_norm": 1.1913066080339185, "learning_rate": 0.003, "loss": 4.0908, "step": 5815 }, { "epoch": 0.05816, "grad_norm": 1.3602272683582077, "learning_rate": 0.003, "loss": 4.1414, "step": 5816 }, { "epoch": 0.05817, "grad_norm": 1.08919306882088, "learning_rate": 0.003, "loss": 4.1512, "step": 5817 }, { "epoch": 0.05818, "grad_norm": 0.9847118464175291, "learning_rate": 0.003, "loss": 4.1302, "step": 5818 }, { "epoch": 0.05819, "grad_norm": 0.8479452899517435, "learning_rate": 0.003, "loss": 4.1178, "step": 5819 }, { "epoch": 0.0582, "grad_norm": 0.8260945784335311, "learning_rate": 0.003, "loss": 4.1211, "step": 5820 }, { "epoch": 0.05821, "grad_norm": 0.8351762734324619, "learning_rate": 0.003, "loss": 4.1317, "step": 5821 }, { "epoch": 0.05822, "grad_norm": 0.8775633899266011, "learning_rate": 0.003, "loss": 4.0966, "step": 5822 }, { "epoch": 0.05823, "grad_norm": 0.8828213129541331, "learning_rate": 0.003, "loss": 4.1249, "step": 5823 }, { "epoch": 0.05824, "grad_norm": 0.9275613652569992, "learning_rate": 0.003, "loss": 4.1288, "step": 5824 }, { "epoch": 0.05825, "grad_norm": 1.1583620136133705, "learning_rate": 0.003, "loss": 4.1575, "step": 5825 }, { "epoch": 0.05826, "grad_norm": 1.0057654996842236, "learning_rate": 0.003, "loss": 4.1039, "step": 5826 }, { "epoch": 0.05827, "grad_norm": 0.9029292549527209, "learning_rate": 0.003, "loss": 4.128, "step": 5827 }, { "epoch": 0.05828, "grad_norm": 0.84303329849258, "learning_rate": 0.003, "loss": 4.0844, "step": 5828 }, { "epoch": 0.05829, "grad_norm": 0.7396485829705943, "learning_rate": 0.003, "loss": 4.1232, "step": 5829 }, { "epoch": 0.0583, "grad_norm": 0.7632310839185854, "learning_rate": 0.003, "loss": 4.0718, "step": 5830 }, { "epoch": 0.05831, "grad_norm": 0.7643152457844856, "learning_rate": 0.003, "loss": 4.0847, "step": 5831 }, { "epoch": 0.05832, "grad_norm": 0.9136308228809447, "learning_rate": 0.003, "loss": 4.1359, "step": 5832 }, { "epoch": 0.05833, "grad_norm": 0.9801023803968252, "learning_rate": 0.003, "loss": 4.1204, "step": 5833 }, { "epoch": 0.05834, "grad_norm": 1.0650227589926176, "learning_rate": 0.003, "loss": 4.133, "step": 5834 }, { "epoch": 0.05835, "grad_norm": 1.1611792836713886, "learning_rate": 0.003, "loss": 4.1527, "step": 5835 }, { "epoch": 0.05836, "grad_norm": 1.0083690824532536, "learning_rate": 0.003, "loss": 4.1312, "step": 5836 }, { "epoch": 0.05837, "grad_norm": 1.0369684104345316, "learning_rate": 0.003, "loss": 4.1272, "step": 5837 }, { "epoch": 0.05838, "grad_norm": 1.007969306404381, "learning_rate": 0.003, "loss": 4.1275, "step": 5838 }, { "epoch": 0.05839, "grad_norm": 1.1056827923761021, "learning_rate": 0.003, "loss": 4.1307, "step": 5839 }, { "epoch": 0.0584, "grad_norm": 0.9832904688706899, "learning_rate": 0.003, "loss": 4.1472, "step": 5840 }, { "epoch": 0.05841, "grad_norm": 1.014502517643171, "learning_rate": 0.003, "loss": 4.1005, "step": 5841 }, { "epoch": 0.05842, "grad_norm": 1.1081359379902629, "learning_rate": 0.003, "loss": 4.1212, "step": 5842 }, { "epoch": 0.05843, "grad_norm": 1.0842690626582987, "learning_rate": 0.003, "loss": 4.1275, "step": 5843 }, { "epoch": 0.05844, "grad_norm": 0.8319622822533052, "learning_rate": 0.003, "loss": 4.1169, "step": 5844 }, { "epoch": 0.05845, "grad_norm": 1.0109437760206588, "learning_rate": 0.003, "loss": 4.1198, "step": 5845 }, { "epoch": 0.05846, "grad_norm": 0.9842272986484365, "learning_rate": 0.003, "loss": 4.1124, "step": 5846 }, { "epoch": 0.05847, "grad_norm": 1.039801475783821, "learning_rate": 0.003, "loss": 4.1452, "step": 5847 }, { "epoch": 0.05848, "grad_norm": 1.0562885170944185, "learning_rate": 0.003, "loss": 4.1202, "step": 5848 }, { "epoch": 0.05849, "grad_norm": 1.1734872924729944, "learning_rate": 0.003, "loss": 4.0865, "step": 5849 }, { "epoch": 0.0585, "grad_norm": 1.1879995500766252, "learning_rate": 0.003, "loss": 4.1174, "step": 5850 }, { "epoch": 0.05851, "grad_norm": 0.9294532448621421, "learning_rate": 0.003, "loss": 4.105, "step": 5851 }, { "epoch": 0.05852, "grad_norm": 0.8214805038934484, "learning_rate": 0.003, "loss": 4.1104, "step": 5852 }, { "epoch": 0.05853, "grad_norm": 1.077263514126393, "learning_rate": 0.003, "loss": 4.104, "step": 5853 }, { "epoch": 0.05854, "grad_norm": 1.006270515716583, "learning_rate": 0.003, "loss": 4.1699, "step": 5854 }, { "epoch": 0.05855, "grad_norm": 0.9464023045138196, "learning_rate": 0.003, "loss": 4.1304, "step": 5855 }, { "epoch": 0.05856, "grad_norm": 1.0642803849206124, "learning_rate": 0.003, "loss": 4.1227, "step": 5856 }, { "epoch": 0.05857, "grad_norm": 1.04585347576585, "learning_rate": 0.003, "loss": 4.1351, "step": 5857 }, { "epoch": 0.05858, "grad_norm": 1.0759346161603385, "learning_rate": 0.003, "loss": 4.1053, "step": 5858 }, { "epoch": 0.05859, "grad_norm": 0.9350487503243448, "learning_rate": 0.003, "loss": 4.1246, "step": 5859 }, { "epoch": 0.0586, "grad_norm": 0.9483831159895448, "learning_rate": 0.003, "loss": 4.1303, "step": 5860 }, { "epoch": 0.05861, "grad_norm": 0.831301332566721, "learning_rate": 0.003, "loss": 4.1424, "step": 5861 }, { "epoch": 0.05862, "grad_norm": 0.7707470466714661, "learning_rate": 0.003, "loss": 4.1299, "step": 5862 }, { "epoch": 0.05863, "grad_norm": 0.8619072544088463, "learning_rate": 0.003, "loss": 4.1329, "step": 5863 }, { "epoch": 0.05864, "grad_norm": 0.9235747550972097, "learning_rate": 0.003, "loss": 4.1105, "step": 5864 }, { "epoch": 0.05865, "grad_norm": 0.9363447226159133, "learning_rate": 0.003, "loss": 4.1093, "step": 5865 }, { "epoch": 0.05866, "grad_norm": 0.9458818904497686, "learning_rate": 0.003, "loss": 4.1298, "step": 5866 }, { "epoch": 0.05867, "grad_norm": 0.9576156779583576, "learning_rate": 0.003, "loss": 4.1125, "step": 5867 }, { "epoch": 0.05868, "grad_norm": 1.2435430439641602, "learning_rate": 0.003, "loss": 4.1065, "step": 5868 }, { "epoch": 0.05869, "grad_norm": 0.7952050411734592, "learning_rate": 0.003, "loss": 4.1097, "step": 5869 }, { "epoch": 0.0587, "grad_norm": 0.6231521938068694, "learning_rate": 0.003, "loss": 4.1031, "step": 5870 }, { "epoch": 0.05871, "grad_norm": 0.6598671272314561, "learning_rate": 0.003, "loss": 4.1065, "step": 5871 }, { "epoch": 0.05872, "grad_norm": 0.6664379166990358, "learning_rate": 0.003, "loss": 4.1015, "step": 5872 }, { "epoch": 0.05873, "grad_norm": 0.8141081232506757, "learning_rate": 0.003, "loss": 4.1177, "step": 5873 }, { "epoch": 0.05874, "grad_norm": 0.8920232475096385, "learning_rate": 0.003, "loss": 4.1095, "step": 5874 }, { "epoch": 0.05875, "grad_norm": 0.9032886130922955, "learning_rate": 0.003, "loss": 4.1236, "step": 5875 }, { "epoch": 0.05876, "grad_norm": 0.9540834088331597, "learning_rate": 0.003, "loss": 4.119, "step": 5876 }, { "epoch": 0.05877, "grad_norm": 1.2497342414913382, "learning_rate": 0.003, "loss": 4.1326, "step": 5877 }, { "epoch": 0.05878, "grad_norm": 1.157702626258542, "learning_rate": 0.003, "loss": 4.1391, "step": 5878 }, { "epoch": 0.05879, "grad_norm": 1.3519563204657061, "learning_rate": 0.003, "loss": 4.1564, "step": 5879 }, { "epoch": 0.0588, "grad_norm": 0.8959524192760782, "learning_rate": 0.003, "loss": 4.1135, "step": 5880 }, { "epoch": 0.05881, "grad_norm": 1.0117816850047339, "learning_rate": 0.003, "loss": 4.1191, "step": 5881 }, { "epoch": 0.05882, "grad_norm": 1.184013692574266, "learning_rate": 0.003, "loss": 4.1378, "step": 5882 }, { "epoch": 0.05883, "grad_norm": 0.9923400292990168, "learning_rate": 0.003, "loss": 4.1288, "step": 5883 }, { "epoch": 0.05884, "grad_norm": 1.1254965238193124, "learning_rate": 0.003, "loss": 4.1181, "step": 5884 }, { "epoch": 0.05885, "grad_norm": 1.0857352979543535, "learning_rate": 0.003, "loss": 4.1351, "step": 5885 }, { "epoch": 0.05886, "grad_norm": 1.0761862053653697, "learning_rate": 0.003, "loss": 4.1286, "step": 5886 }, { "epoch": 0.05887, "grad_norm": 0.9285727732957737, "learning_rate": 0.003, "loss": 4.1169, "step": 5887 }, { "epoch": 0.05888, "grad_norm": 0.9743669592357651, "learning_rate": 0.003, "loss": 4.0745, "step": 5888 }, { "epoch": 0.05889, "grad_norm": 1.2027520897440067, "learning_rate": 0.003, "loss": 4.1115, "step": 5889 }, { "epoch": 0.0589, "grad_norm": 1.1888490061155312, "learning_rate": 0.003, "loss": 4.1201, "step": 5890 }, { "epoch": 0.05891, "grad_norm": 1.0357674592702995, "learning_rate": 0.003, "loss": 4.1155, "step": 5891 }, { "epoch": 0.05892, "grad_norm": 0.8999706532424413, "learning_rate": 0.003, "loss": 4.1449, "step": 5892 }, { "epoch": 0.05893, "grad_norm": 0.9379818703870834, "learning_rate": 0.003, "loss": 4.1174, "step": 5893 }, { "epoch": 0.05894, "grad_norm": 0.9480316835764753, "learning_rate": 0.003, "loss": 4.0921, "step": 5894 }, { "epoch": 0.05895, "grad_norm": 1.0172489807277192, "learning_rate": 0.003, "loss": 4.1034, "step": 5895 }, { "epoch": 0.05896, "grad_norm": 1.037063209191496, "learning_rate": 0.003, "loss": 4.1312, "step": 5896 }, { "epoch": 0.05897, "grad_norm": 0.9545225500016502, "learning_rate": 0.003, "loss": 4.1348, "step": 5897 }, { "epoch": 0.05898, "grad_norm": 0.9646572312575767, "learning_rate": 0.003, "loss": 4.1296, "step": 5898 }, { "epoch": 0.05899, "grad_norm": 0.9753171874462672, "learning_rate": 0.003, "loss": 4.0932, "step": 5899 }, { "epoch": 0.059, "grad_norm": 1.0009130566634394, "learning_rate": 0.003, "loss": 4.1202, "step": 5900 }, { "epoch": 0.05901, "grad_norm": 1.080584743348167, "learning_rate": 0.003, "loss": 4.107, "step": 5901 }, { "epoch": 0.05902, "grad_norm": 1.0596981865427317, "learning_rate": 0.003, "loss": 4.1249, "step": 5902 }, { "epoch": 0.05903, "grad_norm": 0.9172095701197899, "learning_rate": 0.003, "loss": 4.1251, "step": 5903 }, { "epoch": 0.05904, "grad_norm": 0.9690723756267208, "learning_rate": 0.003, "loss": 4.1121, "step": 5904 }, { "epoch": 0.05905, "grad_norm": 1.056830689213177, "learning_rate": 0.003, "loss": 4.1153, "step": 5905 }, { "epoch": 0.05906, "grad_norm": 0.9769191100777919, "learning_rate": 0.003, "loss": 4.1138, "step": 5906 }, { "epoch": 0.05907, "grad_norm": 1.1036371067840394, "learning_rate": 0.003, "loss": 4.1083, "step": 5907 }, { "epoch": 0.05908, "grad_norm": 1.1015376780354955, "learning_rate": 0.003, "loss": 4.1307, "step": 5908 }, { "epoch": 0.05909, "grad_norm": 0.8663999061554115, "learning_rate": 0.003, "loss": 4.134, "step": 5909 }, { "epoch": 0.0591, "grad_norm": 0.7190720749570488, "learning_rate": 0.003, "loss": 4.1007, "step": 5910 }, { "epoch": 0.05911, "grad_norm": 0.766933214132005, "learning_rate": 0.003, "loss": 4.089, "step": 5911 }, { "epoch": 0.05912, "grad_norm": 0.8890912016150025, "learning_rate": 0.003, "loss": 4.1359, "step": 5912 }, { "epoch": 0.05913, "grad_norm": 1.1652210778892036, "learning_rate": 0.003, "loss": 4.1649, "step": 5913 }, { "epoch": 0.05914, "grad_norm": 0.9485707320910097, "learning_rate": 0.003, "loss": 4.1307, "step": 5914 }, { "epoch": 0.05915, "grad_norm": 0.8956370926442963, "learning_rate": 0.003, "loss": 4.1297, "step": 5915 }, { "epoch": 0.05916, "grad_norm": 0.9214734200683395, "learning_rate": 0.003, "loss": 4.1179, "step": 5916 }, { "epoch": 0.05917, "grad_norm": 1.0875072720858519, "learning_rate": 0.003, "loss": 4.102, "step": 5917 }, { "epoch": 0.05918, "grad_norm": 1.2287859167426936, "learning_rate": 0.003, "loss": 4.1201, "step": 5918 }, { "epoch": 0.05919, "grad_norm": 0.8943537238514924, "learning_rate": 0.003, "loss": 4.1079, "step": 5919 }, { "epoch": 0.0592, "grad_norm": 0.9891453949805679, "learning_rate": 0.003, "loss": 4.0924, "step": 5920 }, { "epoch": 0.05921, "grad_norm": 1.0861383341347572, "learning_rate": 0.003, "loss": 4.1406, "step": 5921 }, { "epoch": 0.05922, "grad_norm": 1.1782653575037294, "learning_rate": 0.003, "loss": 4.1343, "step": 5922 }, { "epoch": 0.05923, "grad_norm": 1.1910252107452277, "learning_rate": 0.003, "loss": 4.0986, "step": 5923 }, { "epoch": 0.05924, "grad_norm": 1.1161246036648609, "learning_rate": 0.003, "loss": 4.128, "step": 5924 }, { "epoch": 0.05925, "grad_norm": 1.1329629249127982, "learning_rate": 0.003, "loss": 4.1184, "step": 5925 }, { "epoch": 0.05926, "grad_norm": 1.1093442658198536, "learning_rate": 0.003, "loss": 4.1229, "step": 5926 }, { "epoch": 0.05927, "grad_norm": 1.1038710636777183, "learning_rate": 0.003, "loss": 4.1044, "step": 5927 }, { "epoch": 0.05928, "grad_norm": 0.9173030931279694, "learning_rate": 0.003, "loss": 4.1203, "step": 5928 }, { "epoch": 0.05929, "grad_norm": 0.9040076215100182, "learning_rate": 0.003, "loss": 4.1298, "step": 5929 }, { "epoch": 0.0593, "grad_norm": 0.9168916841143755, "learning_rate": 0.003, "loss": 4.1299, "step": 5930 }, { "epoch": 0.05931, "grad_norm": 0.9557678467541673, "learning_rate": 0.003, "loss": 4.1115, "step": 5931 }, { "epoch": 0.05932, "grad_norm": 1.0099260007035409, "learning_rate": 0.003, "loss": 4.1403, "step": 5932 }, { "epoch": 0.05933, "grad_norm": 0.9621825722693325, "learning_rate": 0.003, "loss": 4.1091, "step": 5933 }, { "epoch": 0.05934, "grad_norm": 1.0971071255328568, "learning_rate": 0.003, "loss": 4.1519, "step": 5934 }, { "epoch": 0.05935, "grad_norm": 0.9234790751127311, "learning_rate": 0.003, "loss": 4.1298, "step": 5935 }, { "epoch": 0.05936, "grad_norm": 0.9821910858313653, "learning_rate": 0.003, "loss": 4.128, "step": 5936 }, { "epoch": 0.05937, "grad_norm": 1.053295387630497, "learning_rate": 0.003, "loss": 4.1308, "step": 5937 }, { "epoch": 0.05938, "grad_norm": 0.9118058500310073, "learning_rate": 0.003, "loss": 4.1027, "step": 5938 }, { "epoch": 0.05939, "grad_norm": 0.9011024881921385, "learning_rate": 0.003, "loss": 4.1146, "step": 5939 }, { "epoch": 0.0594, "grad_norm": 0.924319370161119, "learning_rate": 0.003, "loss": 4.131, "step": 5940 }, { "epoch": 0.05941, "grad_norm": 1.0245197208399006, "learning_rate": 0.003, "loss": 4.1346, "step": 5941 }, { "epoch": 0.05942, "grad_norm": 1.0834208067154592, "learning_rate": 0.003, "loss": 4.1267, "step": 5942 }, { "epoch": 0.05943, "grad_norm": 1.1741526716357884, "learning_rate": 0.003, "loss": 4.1152, "step": 5943 }, { "epoch": 0.05944, "grad_norm": 0.9071829121363102, "learning_rate": 0.003, "loss": 4.0672, "step": 5944 }, { "epoch": 0.05945, "grad_norm": 0.7177835099971523, "learning_rate": 0.003, "loss": 4.105, "step": 5945 }, { "epoch": 0.05946, "grad_norm": 0.6696791120677105, "learning_rate": 0.003, "loss": 4.0978, "step": 5946 }, { "epoch": 0.05947, "grad_norm": 0.8016409208057959, "learning_rate": 0.003, "loss": 4.0936, "step": 5947 }, { "epoch": 0.05948, "grad_norm": 0.9133072242258121, "learning_rate": 0.003, "loss": 4.1128, "step": 5948 }, { "epoch": 0.05949, "grad_norm": 1.0761773223858178, "learning_rate": 0.003, "loss": 4.1108, "step": 5949 }, { "epoch": 0.0595, "grad_norm": 0.9836659779552777, "learning_rate": 0.003, "loss": 4.1193, "step": 5950 }, { "epoch": 0.05951, "grad_norm": 1.0199335343578013, "learning_rate": 0.003, "loss": 4.1117, "step": 5951 }, { "epoch": 0.05952, "grad_norm": 0.9743246006911659, "learning_rate": 0.003, "loss": 4.1239, "step": 5952 }, { "epoch": 0.05953, "grad_norm": 1.2016427351028913, "learning_rate": 0.003, "loss": 4.1098, "step": 5953 }, { "epoch": 0.05954, "grad_norm": 0.9106029846122331, "learning_rate": 0.003, "loss": 4.1251, "step": 5954 }, { "epoch": 0.05955, "grad_norm": 0.87535481561734, "learning_rate": 0.003, "loss": 4.1051, "step": 5955 }, { "epoch": 0.05956, "grad_norm": 0.8781809465631218, "learning_rate": 0.003, "loss": 4.0911, "step": 5956 }, { "epoch": 0.05957, "grad_norm": 0.9484771032245364, "learning_rate": 0.003, "loss": 4.0997, "step": 5957 }, { "epoch": 0.05958, "grad_norm": 1.1611462973785118, "learning_rate": 0.003, "loss": 4.1158, "step": 5958 }, { "epoch": 0.05959, "grad_norm": 1.021761889730662, "learning_rate": 0.003, "loss": 4.1123, "step": 5959 }, { "epoch": 0.0596, "grad_norm": 1.0395052086664378, "learning_rate": 0.003, "loss": 4.1291, "step": 5960 }, { "epoch": 0.05961, "grad_norm": 1.038972275353387, "learning_rate": 0.003, "loss": 4.1177, "step": 5961 }, { "epoch": 0.05962, "grad_norm": 1.1517542544342303, "learning_rate": 0.003, "loss": 4.1141, "step": 5962 }, { "epoch": 0.05963, "grad_norm": 1.0186946928220215, "learning_rate": 0.003, "loss": 4.1218, "step": 5963 }, { "epoch": 0.05964, "grad_norm": 1.0734942613507523, "learning_rate": 0.003, "loss": 4.1208, "step": 5964 }, { "epoch": 0.05965, "grad_norm": 0.99398075190589, "learning_rate": 0.003, "loss": 4.1133, "step": 5965 }, { "epoch": 0.05966, "grad_norm": 0.9660720138071446, "learning_rate": 0.003, "loss": 4.1276, "step": 5966 }, { "epoch": 0.05967, "grad_norm": 1.0547956518514383, "learning_rate": 0.003, "loss": 4.1098, "step": 5967 }, { "epoch": 0.05968, "grad_norm": 1.0422524262760788, "learning_rate": 0.003, "loss": 4.1111, "step": 5968 }, { "epoch": 0.05969, "grad_norm": 1.205455021996495, "learning_rate": 0.003, "loss": 4.1072, "step": 5969 }, { "epoch": 0.0597, "grad_norm": 0.8909956282412519, "learning_rate": 0.003, "loss": 4.106, "step": 5970 }, { "epoch": 0.05971, "grad_norm": 0.9943205977156692, "learning_rate": 0.003, "loss": 4.126, "step": 5971 }, { "epoch": 0.05972, "grad_norm": 1.1691908066767673, "learning_rate": 0.003, "loss": 4.145, "step": 5972 }, { "epoch": 0.05973, "grad_norm": 1.1397498496713665, "learning_rate": 0.003, "loss": 4.1047, "step": 5973 }, { "epoch": 0.05974, "grad_norm": 1.0955142426242983, "learning_rate": 0.003, "loss": 4.1005, "step": 5974 }, { "epoch": 0.05975, "grad_norm": 0.9528979235197279, "learning_rate": 0.003, "loss": 4.1359, "step": 5975 }, { "epoch": 0.05976, "grad_norm": 1.0192429202061466, "learning_rate": 0.003, "loss": 4.1195, "step": 5976 }, { "epoch": 0.05977, "grad_norm": 1.0732106838304054, "learning_rate": 0.003, "loss": 4.1342, "step": 5977 }, { "epoch": 0.05978, "grad_norm": 1.0142239464322467, "learning_rate": 0.003, "loss": 4.1142, "step": 5978 }, { "epoch": 0.05979, "grad_norm": 0.9137036549885006, "learning_rate": 0.003, "loss": 4.1206, "step": 5979 }, { "epoch": 0.0598, "grad_norm": 0.9017921713902286, "learning_rate": 0.003, "loss": 4.0972, "step": 5980 }, { "epoch": 0.05981, "grad_norm": 0.9418925607728872, "learning_rate": 0.003, "loss": 4.0993, "step": 5981 }, { "epoch": 0.05982, "grad_norm": 1.0614608522568088, "learning_rate": 0.003, "loss": 4.1192, "step": 5982 }, { "epoch": 0.05983, "grad_norm": 0.9276124651301009, "learning_rate": 0.003, "loss": 4.1293, "step": 5983 }, { "epoch": 0.05984, "grad_norm": 1.0099523983857785, "learning_rate": 0.003, "loss": 4.0989, "step": 5984 }, { "epoch": 0.05985, "grad_norm": 1.021527017852559, "learning_rate": 0.003, "loss": 4.1321, "step": 5985 }, { "epoch": 0.05986, "grad_norm": 1.125446892694059, "learning_rate": 0.003, "loss": 4.1342, "step": 5986 }, { "epoch": 0.05987, "grad_norm": 1.0737688373374732, "learning_rate": 0.003, "loss": 4.1047, "step": 5987 }, { "epoch": 0.05988, "grad_norm": 0.9164702997141159, "learning_rate": 0.003, "loss": 4.1099, "step": 5988 }, { "epoch": 0.05989, "grad_norm": 0.9143021900180711, "learning_rate": 0.003, "loss": 4.1077, "step": 5989 }, { "epoch": 0.0599, "grad_norm": 0.8543938538827425, "learning_rate": 0.003, "loss": 4.1136, "step": 5990 }, { "epoch": 0.05991, "grad_norm": 1.0063680270151982, "learning_rate": 0.003, "loss": 4.0852, "step": 5991 }, { "epoch": 0.05992, "grad_norm": 1.2334495218996615, "learning_rate": 0.003, "loss": 4.1186, "step": 5992 }, { "epoch": 0.05993, "grad_norm": 0.9732503663927103, "learning_rate": 0.003, "loss": 4.1229, "step": 5993 }, { "epoch": 0.05994, "grad_norm": 1.0952554321561894, "learning_rate": 0.003, "loss": 4.1077, "step": 5994 }, { "epoch": 0.05995, "grad_norm": 1.0674546890074543, "learning_rate": 0.003, "loss": 4.1131, "step": 5995 }, { "epoch": 0.05996, "grad_norm": 0.9812098630340207, "learning_rate": 0.003, "loss": 4.1448, "step": 5996 }, { "epoch": 0.05997, "grad_norm": 1.0336990678275986, "learning_rate": 0.003, "loss": 4.0907, "step": 5997 }, { "epoch": 0.05998, "grad_norm": 1.0167004385621488, "learning_rate": 0.003, "loss": 4.1414, "step": 5998 }, { "epoch": 0.05999, "grad_norm": 1.1118826651465459, "learning_rate": 0.003, "loss": 4.1104, "step": 5999 }, { "epoch": 0.06, "grad_norm": 1.0614094902208722, "learning_rate": 0.003, "loss": 4.1238, "step": 6000 }, { "epoch": 0.06001, "grad_norm": 0.9983812953779062, "learning_rate": 0.003, "loss": 4.1061, "step": 6001 }, { "epoch": 0.06002, "grad_norm": 1.0543477507195134, "learning_rate": 0.003, "loss": 4.1048, "step": 6002 }, { "epoch": 0.06003, "grad_norm": 0.7908508239243327, "learning_rate": 0.003, "loss": 4.1167, "step": 6003 }, { "epoch": 0.06004, "grad_norm": 0.7408629642754656, "learning_rate": 0.003, "loss": 4.0948, "step": 6004 }, { "epoch": 0.06005, "grad_norm": 0.7225830777674686, "learning_rate": 0.003, "loss": 4.1107, "step": 6005 }, { "epoch": 0.06006, "grad_norm": 0.8254182825933346, "learning_rate": 0.003, "loss": 4.0918, "step": 6006 }, { "epoch": 0.06007, "grad_norm": 1.0047409573685806, "learning_rate": 0.003, "loss": 4.1456, "step": 6007 }, { "epoch": 0.06008, "grad_norm": 1.3454769118513017, "learning_rate": 0.003, "loss": 4.1106, "step": 6008 }, { "epoch": 0.06009, "grad_norm": 0.8841761212056402, "learning_rate": 0.003, "loss": 4.1089, "step": 6009 }, { "epoch": 0.0601, "grad_norm": 0.8712614480464654, "learning_rate": 0.003, "loss": 4.1128, "step": 6010 }, { "epoch": 0.06011, "grad_norm": 0.9614357198294633, "learning_rate": 0.003, "loss": 4.1191, "step": 6011 }, { "epoch": 0.06012, "grad_norm": 1.0365500927855609, "learning_rate": 0.003, "loss": 4.128, "step": 6012 }, { "epoch": 0.06013, "grad_norm": 1.0079447498840501, "learning_rate": 0.003, "loss": 4.1199, "step": 6013 }, { "epoch": 0.06014, "grad_norm": 1.027171898423869, "learning_rate": 0.003, "loss": 4.1107, "step": 6014 }, { "epoch": 0.06015, "grad_norm": 0.9671591709956947, "learning_rate": 0.003, "loss": 4.1236, "step": 6015 }, { "epoch": 0.06016, "grad_norm": 0.8094084922333825, "learning_rate": 0.003, "loss": 4.1241, "step": 6016 }, { "epoch": 0.06017, "grad_norm": 0.8969442169529565, "learning_rate": 0.003, "loss": 4.1287, "step": 6017 }, { "epoch": 0.06018, "grad_norm": 0.8714590129924783, "learning_rate": 0.003, "loss": 4.1439, "step": 6018 }, { "epoch": 0.06019, "grad_norm": 0.9863848503972414, "learning_rate": 0.003, "loss": 4.1133, "step": 6019 }, { "epoch": 0.0602, "grad_norm": 0.9823451734348092, "learning_rate": 0.003, "loss": 4.1107, "step": 6020 }, { "epoch": 0.06021, "grad_norm": 1.0823387099836543, "learning_rate": 0.003, "loss": 4.1316, "step": 6021 }, { "epoch": 0.06022, "grad_norm": 1.2506242840661803, "learning_rate": 0.003, "loss": 4.1407, "step": 6022 }, { "epoch": 0.06023, "grad_norm": 1.1016631235570162, "learning_rate": 0.003, "loss": 4.1218, "step": 6023 }, { "epoch": 0.06024, "grad_norm": 1.0131110871671019, "learning_rate": 0.003, "loss": 4.1012, "step": 6024 }, { "epoch": 0.06025, "grad_norm": 0.9550711760770073, "learning_rate": 0.003, "loss": 4.1549, "step": 6025 }, { "epoch": 0.06026, "grad_norm": 1.0082742845706594, "learning_rate": 0.003, "loss": 4.1238, "step": 6026 }, { "epoch": 0.06027, "grad_norm": 1.1193983758663688, "learning_rate": 0.003, "loss": 4.1192, "step": 6027 }, { "epoch": 0.06028, "grad_norm": 1.0297321070613559, "learning_rate": 0.003, "loss": 4.1427, "step": 6028 }, { "epoch": 0.06029, "grad_norm": 1.1955788477078355, "learning_rate": 0.003, "loss": 4.1453, "step": 6029 }, { "epoch": 0.0603, "grad_norm": 0.9831040724140647, "learning_rate": 0.003, "loss": 4.1304, "step": 6030 }, { "epoch": 0.06031, "grad_norm": 1.1029620178962896, "learning_rate": 0.003, "loss": 4.1282, "step": 6031 }, { "epoch": 0.06032, "grad_norm": 1.0090020178039447, "learning_rate": 0.003, "loss": 4.1276, "step": 6032 }, { "epoch": 0.06033, "grad_norm": 1.120910029012685, "learning_rate": 0.003, "loss": 4.1408, "step": 6033 }, { "epoch": 0.06034, "grad_norm": 1.202799778814886, "learning_rate": 0.003, "loss": 4.1524, "step": 6034 }, { "epoch": 0.06035, "grad_norm": 1.2839095955018514, "learning_rate": 0.003, "loss": 4.1464, "step": 6035 }, { "epoch": 0.06036, "grad_norm": 0.8960072775662671, "learning_rate": 0.003, "loss": 4.089, "step": 6036 }, { "epoch": 0.06037, "grad_norm": 0.9251248286636486, "learning_rate": 0.003, "loss": 4.1308, "step": 6037 }, { "epoch": 0.06038, "grad_norm": 1.0231580981647335, "learning_rate": 0.003, "loss": 4.1154, "step": 6038 }, { "epoch": 0.06039, "grad_norm": 1.0065304620292819, "learning_rate": 0.003, "loss": 4.0948, "step": 6039 }, { "epoch": 0.0604, "grad_norm": 0.9234637398061142, "learning_rate": 0.003, "loss": 4.1222, "step": 6040 }, { "epoch": 0.06041, "grad_norm": 0.9399687656493915, "learning_rate": 0.003, "loss": 4.1364, "step": 6041 }, { "epoch": 0.06042, "grad_norm": 0.9871340954259973, "learning_rate": 0.003, "loss": 4.1172, "step": 6042 }, { "epoch": 0.06043, "grad_norm": 1.1211726188172257, "learning_rate": 0.003, "loss": 4.1235, "step": 6043 }, { "epoch": 0.06044, "grad_norm": 0.9672368374829318, "learning_rate": 0.003, "loss": 4.1214, "step": 6044 }, { "epoch": 0.06045, "grad_norm": 0.9720049333758409, "learning_rate": 0.003, "loss": 4.1195, "step": 6045 }, { "epoch": 0.06046, "grad_norm": 1.1307499152097553, "learning_rate": 0.003, "loss": 4.1277, "step": 6046 }, { "epoch": 0.06047, "grad_norm": 1.1577066626632935, "learning_rate": 0.003, "loss": 4.1061, "step": 6047 }, { "epoch": 0.06048, "grad_norm": 0.934752567088904, "learning_rate": 0.003, "loss": 4.1015, "step": 6048 }, { "epoch": 0.06049, "grad_norm": 0.9051391606921817, "learning_rate": 0.003, "loss": 4.0951, "step": 6049 }, { "epoch": 0.0605, "grad_norm": 1.0677162480423936, "learning_rate": 0.003, "loss": 4.1248, "step": 6050 }, { "epoch": 0.06051, "grad_norm": 0.9753098802964573, "learning_rate": 0.003, "loss": 4.1366, "step": 6051 }, { "epoch": 0.06052, "grad_norm": 0.7981061068208121, "learning_rate": 0.003, "loss": 4.129, "step": 6052 }, { "epoch": 0.06053, "grad_norm": 0.8444006516702162, "learning_rate": 0.003, "loss": 4.0885, "step": 6053 }, { "epoch": 0.06054, "grad_norm": 0.8591163485087062, "learning_rate": 0.003, "loss": 4.1239, "step": 6054 }, { "epoch": 0.06055, "grad_norm": 0.9605032090948827, "learning_rate": 0.003, "loss": 4.1182, "step": 6055 }, { "epoch": 0.06056, "grad_norm": 1.3762618972232707, "learning_rate": 0.003, "loss": 4.1179, "step": 6056 }, { "epoch": 0.06057, "grad_norm": 1.0957643635595118, "learning_rate": 0.003, "loss": 4.0995, "step": 6057 }, { "epoch": 0.06058, "grad_norm": 0.9253580114680722, "learning_rate": 0.003, "loss": 4.1114, "step": 6058 }, { "epoch": 0.06059, "grad_norm": 1.0298887868390938, "learning_rate": 0.003, "loss": 4.125, "step": 6059 }, { "epoch": 0.0606, "grad_norm": 1.1659113696201608, "learning_rate": 0.003, "loss": 4.1275, "step": 6060 }, { "epoch": 0.06061, "grad_norm": 1.003664298622335, "learning_rate": 0.003, "loss": 4.1261, "step": 6061 }, { "epoch": 0.06062, "grad_norm": 1.4291730349399376, "learning_rate": 0.003, "loss": 4.1405, "step": 6062 }, { "epoch": 0.06063, "grad_norm": 0.8331420156511541, "learning_rate": 0.003, "loss": 4.0841, "step": 6063 }, { "epoch": 0.06064, "grad_norm": 0.7542070054653189, "learning_rate": 0.003, "loss": 4.1419, "step": 6064 }, { "epoch": 0.06065, "grad_norm": 0.8207912460389019, "learning_rate": 0.003, "loss": 4.1139, "step": 6065 }, { "epoch": 0.06066, "grad_norm": 0.9450469860084448, "learning_rate": 0.003, "loss": 4.1482, "step": 6066 }, { "epoch": 0.06067, "grad_norm": 1.2813480138223348, "learning_rate": 0.003, "loss": 4.1314, "step": 6067 }, { "epoch": 0.06068, "grad_norm": 0.9754293459370826, "learning_rate": 0.003, "loss": 4.1101, "step": 6068 }, { "epoch": 0.06069, "grad_norm": 0.9381993103405896, "learning_rate": 0.003, "loss": 4.1179, "step": 6069 }, { "epoch": 0.0607, "grad_norm": 0.969119385469152, "learning_rate": 0.003, "loss": 4.1198, "step": 6070 }, { "epoch": 0.06071, "grad_norm": 1.1338080882626715, "learning_rate": 0.003, "loss": 4.0977, "step": 6071 }, { "epoch": 0.06072, "grad_norm": 0.8631213140843556, "learning_rate": 0.003, "loss": 4.099, "step": 6072 }, { "epoch": 0.06073, "grad_norm": 0.9229201198227382, "learning_rate": 0.003, "loss": 4.1299, "step": 6073 }, { "epoch": 0.06074, "grad_norm": 1.178815613946597, "learning_rate": 0.003, "loss": 4.1186, "step": 6074 }, { "epoch": 0.06075, "grad_norm": 0.9473743130203446, "learning_rate": 0.003, "loss": 4.135, "step": 6075 }, { "epoch": 0.06076, "grad_norm": 1.041670242948814, "learning_rate": 0.003, "loss": 4.1213, "step": 6076 }, { "epoch": 0.06077, "grad_norm": 1.0127128248077857, "learning_rate": 0.003, "loss": 4.1186, "step": 6077 }, { "epoch": 0.06078, "grad_norm": 0.9898812176337916, "learning_rate": 0.003, "loss": 4.1273, "step": 6078 }, { "epoch": 0.06079, "grad_norm": 0.9621264068499179, "learning_rate": 0.003, "loss": 4.1197, "step": 6079 }, { "epoch": 0.0608, "grad_norm": 0.8712893680783403, "learning_rate": 0.003, "loss": 4.1172, "step": 6080 }, { "epoch": 0.06081, "grad_norm": 0.8647897086338766, "learning_rate": 0.003, "loss": 4.099, "step": 6081 }, { "epoch": 0.06082, "grad_norm": 0.8905130772955867, "learning_rate": 0.003, "loss": 4.1442, "step": 6082 }, { "epoch": 0.06083, "grad_norm": 1.0240935462028675, "learning_rate": 0.003, "loss": 4.1324, "step": 6083 }, { "epoch": 0.06084, "grad_norm": 1.1672577186095388, "learning_rate": 0.003, "loss": 4.0984, "step": 6084 }, { "epoch": 0.06085, "grad_norm": 1.3250963261390494, "learning_rate": 0.003, "loss": 4.0884, "step": 6085 }, { "epoch": 0.06086, "grad_norm": 0.870201800346383, "learning_rate": 0.003, "loss": 4.0918, "step": 6086 }, { "epoch": 0.06087, "grad_norm": 0.8933035330011422, "learning_rate": 0.003, "loss": 4.1295, "step": 6087 }, { "epoch": 0.06088, "grad_norm": 0.9944369027728199, "learning_rate": 0.003, "loss": 4.1052, "step": 6088 }, { "epoch": 0.06089, "grad_norm": 1.188704398920773, "learning_rate": 0.003, "loss": 4.1083, "step": 6089 }, { "epoch": 0.0609, "grad_norm": 0.9777021810669053, "learning_rate": 0.003, "loss": 4.148, "step": 6090 }, { "epoch": 0.06091, "grad_norm": 1.0499229093718996, "learning_rate": 0.003, "loss": 4.1203, "step": 6091 }, { "epoch": 0.06092, "grad_norm": 0.9796736938616526, "learning_rate": 0.003, "loss": 4.102, "step": 6092 }, { "epoch": 0.06093, "grad_norm": 0.8559383646964028, "learning_rate": 0.003, "loss": 4.1123, "step": 6093 }, { "epoch": 0.06094, "grad_norm": 0.9313283566372004, "learning_rate": 0.003, "loss": 4.1156, "step": 6094 }, { "epoch": 0.06095, "grad_norm": 1.0809762060800143, "learning_rate": 0.003, "loss": 4.1212, "step": 6095 }, { "epoch": 0.06096, "grad_norm": 1.0835678303665617, "learning_rate": 0.003, "loss": 4.1125, "step": 6096 }, { "epoch": 0.06097, "grad_norm": 1.0642504020628758, "learning_rate": 0.003, "loss": 4.1284, "step": 6097 }, { "epoch": 0.06098, "grad_norm": 0.9976998221148508, "learning_rate": 0.003, "loss": 4.1002, "step": 6098 }, { "epoch": 0.06099, "grad_norm": 1.0924202946682247, "learning_rate": 0.003, "loss": 4.1258, "step": 6099 }, { "epoch": 0.061, "grad_norm": 0.8943889339335144, "learning_rate": 0.003, "loss": 4.111, "step": 6100 }, { "epoch": 0.06101, "grad_norm": 0.8414480065098904, "learning_rate": 0.003, "loss": 4.1413, "step": 6101 }, { "epoch": 0.06102, "grad_norm": 0.8157966743179256, "learning_rate": 0.003, "loss": 4.122, "step": 6102 }, { "epoch": 0.06103, "grad_norm": 0.8333573956876027, "learning_rate": 0.003, "loss": 4.1349, "step": 6103 }, { "epoch": 0.06104, "grad_norm": 0.8222742472881592, "learning_rate": 0.003, "loss": 4.1056, "step": 6104 }, { "epoch": 0.06105, "grad_norm": 0.9085226278952419, "learning_rate": 0.003, "loss": 4.1129, "step": 6105 }, { "epoch": 0.06106, "grad_norm": 1.0436819426813295, "learning_rate": 0.003, "loss": 4.1372, "step": 6106 }, { "epoch": 0.06107, "grad_norm": 1.1352646100575459, "learning_rate": 0.003, "loss": 4.1499, "step": 6107 }, { "epoch": 0.06108, "grad_norm": 1.1088556231295252, "learning_rate": 0.003, "loss": 4.1134, "step": 6108 }, { "epoch": 0.06109, "grad_norm": 1.088382498713082, "learning_rate": 0.003, "loss": 4.0941, "step": 6109 }, { "epoch": 0.0611, "grad_norm": 1.0780432441806462, "learning_rate": 0.003, "loss": 4.1136, "step": 6110 }, { "epoch": 0.06111, "grad_norm": 1.197481994266544, "learning_rate": 0.003, "loss": 4.1309, "step": 6111 }, { "epoch": 0.06112, "grad_norm": 0.9669086710118441, "learning_rate": 0.003, "loss": 4.1227, "step": 6112 }, { "epoch": 0.06113, "grad_norm": 1.129056630459804, "learning_rate": 0.003, "loss": 4.1384, "step": 6113 }, { "epoch": 0.06114, "grad_norm": 0.990545101825822, "learning_rate": 0.003, "loss": 4.0942, "step": 6114 }, { "epoch": 0.06115, "grad_norm": 0.993200814873901, "learning_rate": 0.003, "loss": 4.0954, "step": 6115 }, { "epoch": 0.06116, "grad_norm": 1.1506429666248497, "learning_rate": 0.003, "loss": 4.1159, "step": 6116 }, { "epoch": 0.06117, "grad_norm": 1.232012893903766, "learning_rate": 0.003, "loss": 4.111, "step": 6117 }, { "epoch": 0.06118, "grad_norm": 0.9333069523142945, "learning_rate": 0.003, "loss": 4.0832, "step": 6118 }, { "epoch": 0.06119, "grad_norm": 0.9309794301782843, "learning_rate": 0.003, "loss": 4.0753, "step": 6119 }, { "epoch": 0.0612, "grad_norm": 0.9889641658765227, "learning_rate": 0.003, "loss": 4.1218, "step": 6120 }, { "epoch": 0.06121, "grad_norm": 0.9429924903195458, "learning_rate": 0.003, "loss": 4.1045, "step": 6121 }, { "epoch": 0.06122, "grad_norm": 0.9263560870190729, "learning_rate": 0.003, "loss": 4.1201, "step": 6122 }, { "epoch": 0.06123, "grad_norm": 0.9349458998129097, "learning_rate": 0.003, "loss": 4.1199, "step": 6123 }, { "epoch": 0.06124, "grad_norm": 0.8727337616002643, "learning_rate": 0.003, "loss": 4.1116, "step": 6124 }, { "epoch": 0.06125, "grad_norm": 0.8116241603577743, "learning_rate": 0.003, "loss": 4.0704, "step": 6125 }, { "epoch": 0.06126, "grad_norm": 0.8298969442086128, "learning_rate": 0.003, "loss": 4.089, "step": 6126 }, { "epoch": 0.06127, "grad_norm": 1.0134811068303906, "learning_rate": 0.003, "loss": 4.0843, "step": 6127 }, { "epoch": 0.06128, "grad_norm": 1.1185496168002877, "learning_rate": 0.003, "loss": 4.1054, "step": 6128 }, { "epoch": 0.06129, "grad_norm": 0.9033340313373569, "learning_rate": 0.003, "loss": 4.1125, "step": 6129 }, { "epoch": 0.0613, "grad_norm": 0.7767340053994768, "learning_rate": 0.003, "loss": 4.1166, "step": 6130 }, { "epoch": 0.06131, "grad_norm": 0.7946814526459421, "learning_rate": 0.003, "loss": 4.1133, "step": 6131 }, { "epoch": 0.06132, "grad_norm": 0.775430711370884, "learning_rate": 0.003, "loss": 4.1112, "step": 6132 }, { "epoch": 0.06133, "grad_norm": 0.9286092538190712, "learning_rate": 0.003, "loss": 4.1063, "step": 6133 }, { "epoch": 0.06134, "grad_norm": 1.1838282950640109, "learning_rate": 0.003, "loss": 4.1024, "step": 6134 }, { "epoch": 0.06135, "grad_norm": 0.9253915308025362, "learning_rate": 0.003, "loss": 4.1239, "step": 6135 }, { "epoch": 0.06136, "grad_norm": 0.9923935179783585, "learning_rate": 0.003, "loss": 4.1066, "step": 6136 }, { "epoch": 0.06137, "grad_norm": 0.9538152751448672, "learning_rate": 0.003, "loss": 4.105, "step": 6137 }, { "epoch": 0.06138, "grad_norm": 0.9731598770998673, "learning_rate": 0.003, "loss": 4.0991, "step": 6138 }, { "epoch": 0.06139, "grad_norm": 1.1310802967564626, "learning_rate": 0.003, "loss": 4.1251, "step": 6139 }, { "epoch": 0.0614, "grad_norm": 1.0143613465951173, "learning_rate": 0.003, "loss": 4.0807, "step": 6140 }, { "epoch": 0.06141, "grad_norm": 0.9941376661663965, "learning_rate": 0.003, "loss": 4.084, "step": 6141 }, { "epoch": 0.06142, "grad_norm": 1.1125462176091532, "learning_rate": 0.003, "loss": 4.1159, "step": 6142 }, { "epoch": 0.06143, "grad_norm": 0.927598665034684, "learning_rate": 0.003, "loss": 4.1205, "step": 6143 }, { "epoch": 0.06144, "grad_norm": 1.0202798430001405, "learning_rate": 0.003, "loss": 4.1024, "step": 6144 }, { "epoch": 0.06145, "grad_norm": 1.1112123749988467, "learning_rate": 0.003, "loss": 4.1227, "step": 6145 }, { "epoch": 0.06146, "grad_norm": 0.9774352630984132, "learning_rate": 0.003, "loss": 4.0877, "step": 6146 }, { "epoch": 0.06147, "grad_norm": 1.1099041541034362, "learning_rate": 0.003, "loss": 4.1443, "step": 6147 }, { "epoch": 0.06148, "grad_norm": 1.3400741766228303, "learning_rate": 0.003, "loss": 4.1212, "step": 6148 }, { "epoch": 0.06149, "grad_norm": 0.7790383000135375, "learning_rate": 0.003, "loss": 4.1033, "step": 6149 }, { "epoch": 0.0615, "grad_norm": 0.6587385143781973, "learning_rate": 0.003, "loss": 4.1174, "step": 6150 }, { "epoch": 0.06151, "grad_norm": 0.8840013067228365, "learning_rate": 0.003, "loss": 4.1142, "step": 6151 }, { "epoch": 0.06152, "grad_norm": 1.171465834606882, "learning_rate": 0.003, "loss": 4.1077, "step": 6152 }, { "epoch": 0.06153, "grad_norm": 1.1290004133065623, "learning_rate": 0.003, "loss": 4.1331, "step": 6153 }, { "epoch": 0.06154, "grad_norm": 0.84568812643019, "learning_rate": 0.003, "loss": 4.0733, "step": 6154 }, { "epoch": 0.06155, "grad_norm": 0.7631870358160202, "learning_rate": 0.003, "loss": 4.1175, "step": 6155 }, { "epoch": 0.06156, "grad_norm": 0.8365491777650411, "learning_rate": 0.003, "loss": 4.1023, "step": 6156 }, { "epoch": 0.06157, "grad_norm": 0.962843672187252, "learning_rate": 0.003, "loss": 4.1247, "step": 6157 }, { "epoch": 0.06158, "grad_norm": 1.1397548341231054, "learning_rate": 0.003, "loss": 4.1488, "step": 6158 }, { "epoch": 0.06159, "grad_norm": 0.8950108190357373, "learning_rate": 0.003, "loss": 4.1021, "step": 6159 }, { "epoch": 0.0616, "grad_norm": 0.9839773780830863, "learning_rate": 0.003, "loss": 4.0983, "step": 6160 }, { "epoch": 0.06161, "grad_norm": 1.1856171086545502, "learning_rate": 0.003, "loss": 4.1333, "step": 6161 }, { "epoch": 0.06162, "grad_norm": 1.0668817290384003, "learning_rate": 0.003, "loss": 4.1069, "step": 6162 }, { "epoch": 0.06163, "grad_norm": 0.9679202467051432, "learning_rate": 0.003, "loss": 4.1453, "step": 6163 }, { "epoch": 0.06164, "grad_norm": 1.1789883371391081, "learning_rate": 0.003, "loss": 4.1291, "step": 6164 }, { "epoch": 0.06165, "grad_norm": 0.9378112937127264, "learning_rate": 0.003, "loss": 4.1049, "step": 6165 }, { "epoch": 0.06166, "grad_norm": 1.073717680785834, "learning_rate": 0.003, "loss": 4.1393, "step": 6166 }, { "epoch": 0.06167, "grad_norm": 1.0508780162488434, "learning_rate": 0.003, "loss": 4.1112, "step": 6167 }, { "epoch": 0.06168, "grad_norm": 1.1231275426430674, "learning_rate": 0.003, "loss": 4.1252, "step": 6168 }, { "epoch": 0.06169, "grad_norm": 0.9613639524735813, "learning_rate": 0.003, "loss": 4.1134, "step": 6169 }, { "epoch": 0.0617, "grad_norm": 0.893115062816479, "learning_rate": 0.003, "loss": 4.1167, "step": 6170 }, { "epoch": 0.06171, "grad_norm": 1.014755768462807, "learning_rate": 0.003, "loss": 4.1242, "step": 6171 }, { "epoch": 0.06172, "grad_norm": 1.1203964494434968, "learning_rate": 0.003, "loss": 4.1009, "step": 6172 }, { "epoch": 0.06173, "grad_norm": 1.1296240698689581, "learning_rate": 0.003, "loss": 4.132, "step": 6173 }, { "epoch": 0.06174, "grad_norm": 1.180308645919765, "learning_rate": 0.003, "loss": 4.1264, "step": 6174 }, { "epoch": 0.06175, "grad_norm": 0.9534486608810446, "learning_rate": 0.003, "loss": 4.1202, "step": 6175 }, { "epoch": 0.06176, "grad_norm": 1.0269321584216362, "learning_rate": 0.003, "loss": 4.119, "step": 6176 }, { "epoch": 0.06177, "grad_norm": 1.2053567559226814, "learning_rate": 0.003, "loss": 4.1177, "step": 6177 }, { "epoch": 0.06178, "grad_norm": 1.0146813059297823, "learning_rate": 0.003, "loss": 4.105, "step": 6178 }, { "epoch": 0.06179, "grad_norm": 1.1284915689903894, "learning_rate": 0.003, "loss": 4.1243, "step": 6179 }, { "epoch": 0.0618, "grad_norm": 0.9947070142693367, "learning_rate": 0.003, "loss": 4.1219, "step": 6180 }, { "epoch": 0.06181, "grad_norm": 1.0633559758739175, "learning_rate": 0.003, "loss": 4.1402, "step": 6181 }, { "epoch": 0.06182, "grad_norm": 1.1027400151633573, "learning_rate": 0.003, "loss": 4.0969, "step": 6182 }, { "epoch": 0.06183, "grad_norm": 1.0324647146802441, "learning_rate": 0.003, "loss": 4.1129, "step": 6183 }, { "epoch": 0.06184, "grad_norm": 1.1585114248380202, "learning_rate": 0.003, "loss": 4.1141, "step": 6184 }, { "epoch": 0.06185, "grad_norm": 0.9898954212712141, "learning_rate": 0.003, "loss": 4.102, "step": 6185 }, { "epoch": 0.06186, "grad_norm": 1.020362653358992, "learning_rate": 0.003, "loss": 4.1285, "step": 6186 }, { "epoch": 0.06187, "grad_norm": 1.0015357102361495, "learning_rate": 0.003, "loss": 4.1371, "step": 6187 }, { "epoch": 0.06188, "grad_norm": 1.0340445619644294, "learning_rate": 0.003, "loss": 4.1162, "step": 6188 }, { "epoch": 0.06189, "grad_norm": 1.033425949623886, "learning_rate": 0.003, "loss": 4.1031, "step": 6189 }, { "epoch": 0.0619, "grad_norm": 0.9791881079008582, "learning_rate": 0.003, "loss": 4.145, "step": 6190 }, { "epoch": 0.06191, "grad_norm": 1.0369953073953297, "learning_rate": 0.003, "loss": 4.1021, "step": 6191 }, { "epoch": 0.06192, "grad_norm": 1.0550940203599017, "learning_rate": 0.003, "loss": 4.1051, "step": 6192 }, { "epoch": 0.06193, "grad_norm": 1.004119892499567, "learning_rate": 0.003, "loss": 4.074, "step": 6193 }, { "epoch": 0.06194, "grad_norm": 1.1667997286871377, "learning_rate": 0.003, "loss": 4.1007, "step": 6194 }, { "epoch": 0.06195, "grad_norm": 0.8305937619838769, "learning_rate": 0.003, "loss": 4.1119, "step": 6195 }, { "epoch": 0.06196, "grad_norm": 0.7005743659730198, "learning_rate": 0.003, "loss": 4.1217, "step": 6196 }, { "epoch": 0.06197, "grad_norm": 0.7420600370286529, "learning_rate": 0.003, "loss": 4.106, "step": 6197 }, { "epoch": 0.06198, "grad_norm": 0.7767521916979286, "learning_rate": 0.003, "loss": 4.1326, "step": 6198 }, { "epoch": 0.06199, "grad_norm": 0.9637957082299587, "learning_rate": 0.003, "loss": 4.1252, "step": 6199 }, { "epoch": 0.062, "grad_norm": 1.436081636862012, "learning_rate": 0.003, "loss": 4.1411, "step": 6200 }, { "epoch": 0.06201, "grad_norm": 0.9348049703706455, "learning_rate": 0.003, "loss": 4.1025, "step": 6201 }, { "epoch": 0.06202, "grad_norm": 0.9670197292189014, "learning_rate": 0.003, "loss": 4.1473, "step": 6202 }, { "epoch": 0.06203, "grad_norm": 1.0733101607809545, "learning_rate": 0.003, "loss": 4.1352, "step": 6203 }, { "epoch": 0.06204, "grad_norm": 1.1583376498097138, "learning_rate": 0.003, "loss": 4.1333, "step": 6204 }, { "epoch": 0.06205, "grad_norm": 1.0094346779280738, "learning_rate": 0.003, "loss": 4.1342, "step": 6205 }, { "epoch": 0.06206, "grad_norm": 1.1703726685169265, "learning_rate": 0.003, "loss": 4.1316, "step": 6206 }, { "epoch": 0.06207, "grad_norm": 0.9450958344284114, "learning_rate": 0.003, "loss": 4.0873, "step": 6207 }, { "epoch": 0.06208, "grad_norm": 0.9939553682249204, "learning_rate": 0.003, "loss": 4.1166, "step": 6208 }, { "epoch": 0.06209, "grad_norm": 1.048829583157075, "learning_rate": 0.003, "loss": 4.1038, "step": 6209 }, { "epoch": 0.0621, "grad_norm": 1.1283576786370575, "learning_rate": 0.003, "loss": 4.1647, "step": 6210 }, { "epoch": 0.06211, "grad_norm": 1.0202411361181278, "learning_rate": 0.003, "loss": 4.1265, "step": 6211 }, { "epoch": 0.06212, "grad_norm": 0.8463803172469656, "learning_rate": 0.003, "loss": 4.127, "step": 6212 }, { "epoch": 0.06213, "grad_norm": 0.8831158988837825, "learning_rate": 0.003, "loss": 4.107, "step": 6213 }, { "epoch": 0.06214, "grad_norm": 0.9471127988954445, "learning_rate": 0.003, "loss": 4.1094, "step": 6214 }, { "epoch": 0.06215, "grad_norm": 1.1099784730474564, "learning_rate": 0.003, "loss": 4.1347, "step": 6215 }, { "epoch": 0.06216, "grad_norm": 0.8618468021683433, "learning_rate": 0.003, "loss": 4.1546, "step": 6216 }, { "epoch": 0.06217, "grad_norm": 1.0078275003589592, "learning_rate": 0.003, "loss": 4.1181, "step": 6217 }, { "epoch": 0.06218, "grad_norm": 1.2368143042206021, "learning_rate": 0.003, "loss": 4.1132, "step": 6218 }, { "epoch": 0.06219, "grad_norm": 1.054833838062907, "learning_rate": 0.003, "loss": 4.1371, "step": 6219 }, { "epoch": 0.0622, "grad_norm": 1.3062224581370208, "learning_rate": 0.003, "loss": 4.137, "step": 6220 }, { "epoch": 0.06221, "grad_norm": 0.8604885677509169, "learning_rate": 0.003, "loss": 4.1301, "step": 6221 }, { "epoch": 0.06222, "grad_norm": 0.8641494139028316, "learning_rate": 0.003, "loss": 4.1158, "step": 6222 }, { "epoch": 0.06223, "grad_norm": 0.9055413740295657, "learning_rate": 0.003, "loss": 4.1262, "step": 6223 }, { "epoch": 0.06224, "grad_norm": 0.939566222161941, "learning_rate": 0.003, "loss": 4.0953, "step": 6224 }, { "epoch": 0.06225, "grad_norm": 0.9877692010672254, "learning_rate": 0.003, "loss": 4.1228, "step": 6225 }, { "epoch": 0.06226, "grad_norm": 1.2373704185349645, "learning_rate": 0.003, "loss": 4.1047, "step": 6226 }, { "epoch": 0.06227, "grad_norm": 1.0421769966469974, "learning_rate": 0.003, "loss": 4.1156, "step": 6227 }, { "epoch": 0.06228, "grad_norm": 1.0069660017991295, "learning_rate": 0.003, "loss": 4.1204, "step": 6228 }, { "epoch": 0.06229, "grad_norm": 0.9902407890356086, "learning_rate": 0.003, "loss": 4.1025, "step": 6229 }, { "epoch": 0.0623, "grad_norm": 1.000784859461883, "learning_rate": 0.003, "loss": 4.1158, "step": 6230 }, { "epoch": 0.06231, "grad_norm": 0.9879882528823176, "learning_rate": 0.003, "loss": 4.1046, "step": 6231 }, { "epoch": 0.06232, "grad_norm": 0.9161813009855883, "learning_rate": 0.003, "loss": 4.1025, "step": 6232 }, { "epoch": 0.06233, "grad_norm": 0.9421675613078943, "learning_rate": 0.003, "loss": 4.0985, "step": 6233 }, { "epoch": 0.06234, "grad_norm": 0.9233627757140611, "learning_rate": 0.003, "loss": 4.105, "step": 6234 }, { "epoch": 0.06235, "grad_norm": 0.8773342302782378, "learning_rate": 0.003, "loss": 4.107, "step": 6235 }, { "epoch": 0.06236, "grad_norm": 0.8554767024277897, "learning_rate": 0.003, "loss": 4.1187, "step": 6236 }, { "epoch": 0.06237, "grad_norm": 0.8891401545205588, "learning_rate": 0.003, "loss": 4.1092, "step": 6237 }, { "epoch": 0.06238, "grad_norm": 0.9901683523219613, "learning_rate": 0.003, "loss": 4.1091, "step": 6238 }, { "epoch": 0.06239, "grad_norm": 1.2816193140265006, "learning_rate": 0.003, "loss": 4.099, "step": 6239 }, { "epoch": 0.0624, "grad_norm": 0.9761347780125502, "learning_rate": 0.003, "loss": 4.1073, "step": 6240 }, { "epoch": 0.06241, "grad_norm": 1.0466186560774475, "learning_rate": 0.003, "loss": 4.0848, "step": 6241 }, { "epoch": 0.06242, "grad_norm": 1.1002056251421792, "learning_rate": 0.003, "loss": 4.1194, "step": 6242 }, { "epoch": 0.06243, "grad_norm": 1.0586273465264278, "learning_rate": 0.003, "loss": 4.1044, "step": 6243 }, { "epoch": 0.06244, "grad_norm": 1.1607710935816504, "learning_rate": 0.003, "loss": 4.1426, "step": 6244 }, { "epoch": 0.06245, "grad_norm": 0.9752269488488016, "learning_rate": 0.003, "loss": 4.1148, "step": 6245 }, { "epoch": 0.06246, "grad_norm": 1.089994727247547, "learning_rate": 0.003, "loss": 4.1344, "step": 6246 }, { "epoch": 0.06247, "grad_norm": 1.0813807058658953, "learning_rate": 0.003, "loss": 4.1165, "step": 6247 }, { "epoch": 0.06248, "grad_norm": 1.0666231347706039, "learning_rate": 0.003, "loss": 4.1176, "step": 6248 }, { "epoch": 0.06249, "grad_norm": 1.0262020212481742, "learning_rate": 0.003, "loss": 4.1303, "step": 6249 }, { "epoch": 0.0625, "grad_norm": 1.2632454332795369, "learning_rate": 0.003, "loss": 4.1194, "step": 6250 }, { "epoch": 0.06251, "grad_norm": 0.9575866697235751, "learning_rate": 0.003, "loss": 4.1042, "step": 6251 }, { "epoch": 0.06252, "grad_norm": 1.0727447713803615, "learning_rate": 0.003, "loss": 4.1287, "step": 6252 }, { "epoch": 0.06253, "grad_norm": 1.275488151651881, "learning_rate": 0.003, "loss": 4.1261, "step": 6253 }, { "epoch": 0.06254, "grad_norm": 0.8302208080842566, "learning_rate": 0.003, "loss": 4.11, "step": 6254 }, { "epoch": 0.06255, "grad_norm": 0.9130697914120645, "learning_rate": 0.003, "loss": 4.1201, "step": 6255 }, { "epoch": 0.06256, "grad_norm": 0.9667951839279011, "learning_rate": 0.003, "loss": 4.0977, "step": 6256 }, { "epoch": 0.06257, "grad_norm": 0.8489314632779762, "learning_rate": 0.003, "loss": 4.0729, "step": 6257 }, { "epoch": 0.06258, "grad_norm": 0.8445325664151822, "learning_rate": 0.003, "loss": 4.126, "step": 6258 }, { "epoch": 0.06259, "grad_norm": 0.7672593438540465, "learning_rate": 0.003, "loss": 4.1391, "step": 6259 }, { "epoch": 0.0626, "grad_norm": 0.8964348171771911, "learning_rate": 0.003, "loss": 4.1107, "step": 6260 }, { "epoch": 0.06261, "grad_norm": 0.9947853097219037, "learning_rate": 0.003, "loss": 4.098, "step": 6261 }, { "epoch": 0.06262, "grad_norm": 1.2241118778779938, "learning_rate": 0.003, "loss": 4.1107, "step": 6262 }, { "epoch": 0.06263, "grad_norm": 1.1642987312142146, "learning_rate": 0.003, "loss": 4.1377, "step": 6263 }, { "epoch": 0.06264, "grad_norm": 0.8953851392238773, "learning_rate": 0.003, "loss": 4.1085, "step": 6264 }, { "epoch": 0.06265, "grad_norm": 0.8318963279886489, "learning_rate": 0.003, "loss": 4.1273, "step": 6265 }, { "epoch": 0.06266, "grad_norm": 0.8609508836654348, "learning_rate": 0.003, "loss": 4.1118, "step": 6266 }, { "epoch": 0.06267, "grad_norm": 0.8740732914198333, "learning_rate": 0.003, "loss": 4.1163, "step": 6267 }, { "epoch": 0.06268, "grad_norm": 1.05944492334423, "learning_rate": 0.003, "loss": 4.089, "step": 6268 }, { "epoch": 0.06269, "grad_norm": 1.1546078108699898, "learning_rate": 0.003, "loss": 4.0982, "step": 6269 }, { "epoch": 0.0627, "grad_norm": 1.1402430212111558, "learning_rate": 0.003, "loss": 4.1105, "step": 6270 }, { "epoch": 0.06271, "grad_norm": 0.8999820730504591, "learning_rate": 0.003, "loss": 4.1202, "step": 6271 }, { "epoch": 0.06272, "grad_norm": 0.8074571914853922, "learning_rate": 0.003, "loss": 4.1315, "step": 6272 }, { "epoch": 0.06273, "grad_norm": 0.8233642450661368, "learning_rate": 0.003, "loss": 4.127, "step": 6273 }, { "epoch": 0.06274, "grad_norm": 0.9047777840671057, "learning_rate": 0.003, "loss": 4.109, "step": 6274 }, { "epoch": 0.06275, "grad_norm": 0.9781934686902589, "learning_rate": 0.003, "loss": 4.1019, "step": 6275 }, { "epoch": 0.06276, "grad_norm": 1.04325282479346, "learning_rate": 0.003, "loss": 4.1199, "step": 6276 }, { "epoch": 0.06277, "grad_norm": 1.0621790353885807, "learning_rate": 0.003, "loss": 4.1265, "step": 6277 }, { "epoch": 0.06278, "grad_norm": 1.1452838103158116, "learning_rate": 0.003, "loss": 4.078, "step": 6278 }, { "epoch": 0.06279, "grad_norm": 0.929599885878623, "learning_rate": 0.003, "loss": 4.1327, "step": 6279 }, { "epoch": 0.0628, "grad_norm": 1.1623398174815285, "learning_rate": 0.003, "loss": 4.109, "step": 6280 }, { "epoch": 0.06281, "grad_norm": 1.151758981216677, "learning_rate": 0.003, "loss": 4.089, "step": 6281 }, { "epoch": 0.06282, "grad_norm": 1.007355594619288, "learning_rate": 0.003, "loss": 4.1092, "step": 6282 }, { "epoch": 0.06283, "grad_norm": 1.220826744113072, "learning_rate": 0.003, "loss": 4.0902, "step": 6283 }, { "epoch": 0.06284, "grad_norm": 0.8704770237141937, "learning_rate": 0.003, "loss": 4.1146, "step": 6284 }, { "epoch": 0.06285, "grad_norm": 0.8383126810956327, "learning_rate": 0.003, "loss": 4.1108, "step": 6285 }, { "epoch": 0.06286, "grad_norm": 0.8800742184339927, "learning_rate": 0.003, "loss": 4.1012, "step": 6286 }, { "epoch": 0.06287, "grad_norm": 1.0812936724403794, "learning_rate": 0.003, "loss": 4.1004, "step": 6287 }, { "epoch": 0.06288, "grad_norm": 1.2593934067897568, "learning_rate": 0.003, "loss": 4.1381, "step": 6288 }, { "epoch": 0.06289, "grad_norm": 1.1429060251374439, "learning_rate": 0.003, "loss": 4.088, "step": 6289 }, { "epoch": 0.0629, "grad_norm": 1.246481182234239, "learning_rate": 0.003, "loss": 4.1159, "step": 6290 }, { "epoch": 0.06291, "grad_norm": 0.960676109509977, "learning_rate": 0.003, "loss": 4.0841, "step": 6291 }, { "epoch": 0.06292, "grad_norm": 0.9723196671162704, "learning_rate": 0.003, "loss": 4.1103, "step": 6292 }, { "epoch": 0.06293, "grad_norm": 1.125411684140196, "learning_rate": 0.003, "loss": 4.0732, "step": 6293 }, { "epoch": 0.06294, "grad_norm": 1.0139672152671215, "learning_rate": 0.003, "loss": 4.1222, "step": 6294 }, { "epoch": 0.06295, "grad_norm": 1.2175400090011317, "learning_rate": 0.003, "loss": 4.0712, "step": 6295 }, { "epoch": 0.06296, "grad_norm": 0.8801642637892075, "learning_rate": 0.003, "loss": 4.141, "step": 6296 }, { "epoch": 0.06297, "grad_norm": 0.8995632624538455, "learning_rate": 0.003, "loss": 4.1108, "step": 6297 }, { "epoch": 0.06298, "grad_norm": 1.0073654159072412, "learning_rate": 0.003, "loss": 4.1109, "step": 6298 }, { "epoch": 0.06299, "grad_norm": 1.1330731816130224, "learning_rate": 0.003, "loss": 4.1129, "step": 6299 }, { "epoch": 0.063, "grad_norm": 1.0282659113929185, "learning_rate": 0.003, "loss": 4.1614, "step": 6300 }, { "epoch": 0.06301, "grad_norm": 1.0726720577475604, "learning_rate": 0.003, "loss": 4.1376, "step": 6301 }, { "epoch": 0.06302, "grad_norm": 1.0496260954229397, "learning_rate": 0.003, "loss": 4.1252, "step": 6302 }, { "epoch": 0.06303, "grad_norm": 1.1253852761118548, "learning_rate": 0.003, "loss": 4.1267, "step": 6303 }, { "epoch": 0.06304, "grad_norm": 0.9024308896902583, "learning_rate": 0.003, "loss": 4.1346, "step": 6304 }, { "epoch": 0.06305, "grad_norm": 0.8914762325481002, "learning_rate": 0.003, "loss": 4.1426, "step": 6305 }, { "epoch": 0.06306, "grad_norm": 1.035268496629464, "learning_rate": 0.003, "loss": 4.1064, "step": 6306 }, { "epoch": 0.06307, "grad_norm": 1.1527964693828754, "learning_rate": 0.003, "loss": 4.1053, "step": 6307 }, { "epoch": 0.06308, "grad_norm": 1.0156899019344043, "learning_rate": 0.003, "loss": 4.1221, "step": 6308 }, { "epoch": 0.06309, "grad_norm": 0.9817564469713144, "learning_rate": 0.003, "loss": 4.1091, "step": 6309 }, { "epoch": 0.0631, "grad_norm": 1.125639940555161, "learning_rate": 0.003, "loss": 4.1125, "step": 6310 }, { "epoch": 0.06311, "grad_norm": 1.146532285357419, "learning_rate": 0.003, "loss": 4.1069, "step": 6311 }, { "epoch": 0.06312, "grad_norm": 0.9764239611962195, "learning_rate": 0.003, "loss": 4.0987, "step": 6312 }, { "epoch": 0.06313, "grad_norm": 0.8815145041719636, "learning_rate": 0.003, "loss": 4.121, "step": 6313 }, { "epoch": 0.06314, "grad_norm": 0.840467663491767, "learning_rate": 0.003, "loss": 4.1249, "step": 6314 }, { "epoch": 0.06315, "grad_norm": 0.7533539654208644, "learning_rate": 0.003, "loss": 4.0989, "step": 6315 }, { "epoch": 0.06316, "grad_norm": 0.8052544193540351, "learning_rate": 0.003, "loss": 4.0696, "step": 6316 }, { "epoch": 0.06317, "grad_norm": 0.9863065368724497, "learning_rate": 0.003, "loss": 4.1032, "step": 6317 }, { "epoch": 0.06318, "grad_norm": 1.217882389975133, "learning_rate": 0.003, "loss": 4.1552, "step": 6318 }, { "epoch": 0.06319, "grad_norm": 0.70847036618039, "learning_rate": 0.003, "loss": 4.1434, "step": 6319 }, { "epoch": 0.0632, "grad_norm": 0.8377363984109976, "learning_rate": 0.003, "loss": 4.1087, "step": 6320 }, { "epoch": 0.06321, "grad_norm": 1.0324765911073792, "learning_rate": 0.003, "loss": 4.1251, "step": 6321 }, { "epoch": 0.06322, "grad_norm": 1.3181893770856248, "learning_rate": 0.003, "loss": 4.1172, "step": 6322 }, { "epoch": 0.06323, "grad_norm": 1.1690328295155383, "learning_rate": 0.003, "loss": 4.106, "step": 6323 }, { "epoch": 0.06324, "grad_norm": 1.0679893724078005, "learning_rate": 0.003, "loss": 4.1254, "step": 6324 }, { "epoch": 0.06325, "grad_norm": 0.9783736109608521, "learning_rate": 0.003, "loss": 4.1076, "step": 6325 }, { "epoch": 0.06326, "grad_norm": 1.003334131616264, "learning_rate": 0.003, "loss": 4.1216, "step": 6326 }, { "epoch": 0.06327, "grad_norm": 1.026805442964879, "learning_rate": 0.003, "loss": 4.0914, "step": 6327 }, { "epoch": 0.06328, "grad_norm": 1.076746193210205, "learning_rate": 0.003, "loss": 4.1477, "step": 6328 }, { "epoch": 0.06329, "grad_norm": 0.9881716508673234, "learning_rate": 0.003, "loss": 4.1246, "step": 6329 }, { "epoch": 0.0633, "grad_norm": 1.008130034970649, "learning_rate": 0.003, "loss": 4.1002, "step": 6330 }, { "epoch": 0.06331, "grad_norm": 1.1763834546994973, "learning_rate": 0.003, "loss": 4.1062, "step": 6331 }, { "epoch": 0.06332, "grad_norm": 1.0145017626609247, "learning_rate": 0.003, "loss": 4.1021, "step": 6332 }, { "epoch": 0.06333, "grad_norm": 1.1827671265849429, "learning_rate": 0.003, "loss": 4.1115, "step": 6333 }, { "epoch": 0.06334, "grad_norm": 1.1730909666687077, "learning_rate": 0.003, "loss": 4.1157, "step": 6334 }, { "epoch": 0.06335, "grad_norm": 1.0011266798315859, "learning_rate": 0.003, "loss": 4.0787, "step": 6335 }, { "epoch": 0.06336, "grad_norm": 1.1675286156010463, "learning_rate": 0.003, "loss": 4.146, "step": 6336 }, { "epoch": 0.06337, "grad_norm": 0.9575385222661729, "learning_rate": 0.003, "loss": 4.1428, "step": 6337 }, { "epoch": 0.06338, "grad_norm": 1.0319657712516945, "learning_rate": 0.003, "loss": 4.1292, "step": 6338 }, { "epoch": 0.06339, "grad_norm": 1.086306029136807, "learning_rate": 0.003, "loss": 4.1022, "step": 6339 }, { "epoch": 0.0634, "grad_norm": 1.1043242914825164, "learning_rate": 0.003, "loss": 4.0987, "step": 6340 }, { "epoch": 0.06341, "grad_norm": 0.8543521892094537, "learning_rate": 0.003, "loss": 4.091, "step": 6341 }, { "epoch": 0.06342, "grad_norm": 1.1724944784180746, "learning_rate": 0.003, "loss": 4.1236, "step": 6342 }, { "epoch": 0.06343, "grad_norm": 1.0104393227402537, "learning_rate": 0.003, "loss": 4.1282, "step": 6343 }, { "epoch": 0.06344, "grad_norm": 0.9854141445654092, "learning_rate": 0.003, "loss": 4.0859, "step": 6344 }, { "epoch": 0.06345, "grad_norm": 1.0901443364751084, "learning_rate": 0.003, "loss": 4.1038, "step": 6345 }, { "epoch": 0.06346, "grad_norm": 0.9719436488074422, "learning_rate": 0.003, "loss": 4.1194, "step": 6346 }, { "epoch": 0.06347, "grad_norm": 1.1087617785535473, "learning_rate": 0.003, "loss": 4.1094, "step": 6347 }, { "epoch": 0.06348, "grad_norm": 1.062332265895712, "learning_rate": 0.003, "loss": 4.1053, "step": 6348 }, { "epoch": 0.06349, "grad_norm": 1.0763754416322948, "learning_rate": 0.003, "loss": 4.1067, "step": 6349 }, { "epoch": 0.0635, "grad_norm": 0.8840755219842192, "learning_rate": 0.003, "loss": 4.0967, "step": 6350 }, { "epoch": 0.06351, "grad_norm": 0.8206117947783813, "learning_rate": 0.003, "loss": 4.119, "step": 6351 }, { "epoch": 0.06352, "grad_norm": 0.9715905967435101, "learning_rate": 0.003, "loss": 4.101, "step": 6352 }, { "epoch": 0.06353, "grad_norm": 1.3403472905771021, "learning_rate": 0.003, "loss": 4.1237, "step": 6353 }, { "epoch": 0.06354, "grad_norm": 0.7230549013634665, "learning_rate": 0.003, "loss": 4.1079, "step": 6354 }, { "epoch": 0.06355, "grad_norm": 0.6387342513840211, "learning_rate": 0.003, "loss": 4.1129, "step": 6355 }, { "epoch": 0.06356, "grad_norm": 0.8116813215118523, "learning_rate": 0.003, "loss": 4.1063, "step": 6356 }, { "epoch": 0.06357, "grad_norm": 1.0732408549550134, "learning_rate": 0.003, "loss": 4.1304, "step": 6357 }, { "epoch": 0.06358, "grad_norm": 1.4061400227332301, "learning_rate": 0.003, "loss": 4.1054, "step": 6358 }, { "epoch": 0.06359, "grad_norm": 0.816262506494798, "learning_rate": 0.003, "loss": 4.0985, "step": 6359 }, { "epoch": 0.0636, "grad_norm": 0.7856925452250502, "learning_rate": 0.003, "loss": 4.1251, "step": 6360 }, { "epoch": 0.06361, "grad_norm": 0.8172194066665021, "learning_rate": 0.003, "loss": 4.1173, "step": 6361 }, { "epoch": 0.06362, "grad_norm": 0.8937786474083863, "learning_rate": 0.003, "loss": 4.1256, "step": 6362 }, { "epoch": 0.06363, "grad_norm": 1.213011420537028, "learning_rate": 0.003, "loss": 4.1141, "step": 6363 }, { "epoch": 0.06364, "grad_norm": 0.8563828460718517, "learning_rate": 0.003, "loss": 4.106, "step": 6364 }, { "epoch": 0.06365, "grad_norm": 1.0114885163377767, "learning_rate": 0.003, "loss": 4.0969, "step": 6365 }, { "epoch": 0.06366, "grad_norm": 1.0801002256043084, "learning_rate": 0.003, "loss": 4.1261, "step": 6366 }, { "epoch": 0.06367, "grad_norm": 0.9786682306601094, "learning_rate": 0.003, "loss": 4.1056, "step": 6367 }, { "epoch": 0.06368, "grad_norm": 1.0821388307035957, "learning_rate": 0.003, "loss": 4.1443, "step": 6368 }, { "epoch": 0.06369, "grad_norm": 1.2004164285754237, "learning_rate": 0.003, "loss": 4.1524, "step": 6369 }, { "epoch": 0.0637, "grad_norm": 1.0351207503381683, "learning_rate": 0.003, "loss": 4.1074, "step": 6370 }, { "epoch": 0.06371, "grad_norm": 1.061549907692225, "learning_rate": 0.003, "loss": 4.1287, "step": 6371 }, { "epoch": 0.06372, "grad_norm": 0.9665848388085393, "learning_rate": 0.003, "loss": 4.1267, "step": 6372 }, { "epoch": 0.06373, "grad_norm": 1.0606306406187926, "learning_rate": 0.003, "loss": 4.1205, "step": 6373 }, { "epoch": 0.06374, "grad_norm": 0.9478688183673424, "learning_rate": 0.003, "loss": 4.0774, "step": 6374 }, { "epoch": 0.06375, "grad_norm": 0.9561463271072295, "learning_rate": 0.003, "loss": 4.094, "step": 6375 }, { "epoch": 0.06376, "grad_norm": 1.0954200184165113, "learning_rate": 0.003, "loss": 4.1256, "step": 6376 }, { "epoch": 0.06377, "grad_norm": 1.0149546225474946, "learning_rate": 0.003, "loss": 4.0917, "step": 6377 }, { "epoch": 0.06378, "grad_norm": 1.2530939110895587, "learning_rate": 0.003, "loss": 4.1334, "step": 6378 }, { "epoch": 0.06379, "grad_norm": 0.9536829019923387, "learning_rate": 0.003, "loss": 4.1279, "step": 6379 }, { "epoch": 0.0638, "grad_norm": 1.124365245664102, "learning_rate": 0.003, "loss": 4.1074, "step": 6380 }, { "epoch": 0.06381, "grad_norm": 1.067522258480721, "learning_rate": 0.003, "loss": 4.0911, "step": 6381 }, { "epoch": 0.06382, "grad_norm": 1.165510546851359, "learning_rate": 0.003, "loss": 4.1016, "step": 6382 }, { "epoch": 0.06383, "grad_norm": 0.9796661652974314, "learning_rate": 0.003, "loss": 4.1279, "step": 6383 }, { "epoch": 0.06384, "grad_norm": 0.9583436524182503, "learning_rate": 0.003, "loss": 4.1259, "step": 6384 }, { "epoch": 0.06385, "grad_norm": 0.9606808146113693, "learning_rate": 0.003, "loss": 4.0883, "step": 6385 }, { "epoch": 0.06386, "grad_norm": 1.0631319488352189, "learning_rate": 0.003, "loss": 4.0999, "step": 6386 }, { "epoch": 0.06387, "grad_norm": 0.9436696570931993, "learning_rate": 0.003, "loss": 4.1103, "step": 6387 }, { "epoch": 0.06388, "grad_norm": 1.0318755835894513, "learning_rate": 0.003, "loss": 4.1043, "step": 6388 }, { "epoch": 0.06389, "grad_norm": 1.0643029277751763, "learning_rate": 0.003, "loss": 4.0909, "step": 6389 }, { "epoch": 0.0639, "grad_norm": 0.9274983406171419, "learning_rate": 0.003, "loss": 4.1085, "step": 6390 }, { "epoch": 0.06391, "grad_norm": 1.1657830691809308, "learning_rate": 0.003, "loss": 4.1344, "step": 6391 }, { "epoch": 0.06392, "grad_norm": 1.0164575608007218, "learning_rate": 0.003, "loss": 4.1024, "step": 6392 }, { "epoch": 0.06393, "grad_norm": 1.0672288928145917, "learning_rate": 0.003, "loss": 4.119, "step": 6393 }, { "epoch": 0.06394, "grad_norm": 1.2905438243321443, "learning_rate": 0.003, "loss": 4.1153, "step": 6394 }, { "epoch": 0.06395, "grad_norm": 1.10123470204369, "learning_rate": 0.003, "loss": 4.1387, "step": 6395 }, { "epoch": 0.06396, "grad_norm": 1.0124081789493147, "learning_rate": 0.003, "loss": 4.1487, "step": 6396 }, { "epoch": 0.06397, "grad_norm": 0.9031540181239835, "learning_rate": 0.003, "loss": 4.1005, "step": 6397 }, { "epoch": 0.06398, "grad_norm": 0.8920146505501644, "learning_rate": 0.003, "loss": 4.092, "step": 6398 }, { "epoch": 0.06399, "grad_norm": 0.8773434678917014, "learning_rate": 0.003, "loss": 4.1232, "step": 6399 }, { "epoch": 0.064, "grad_norm": 0.9294523161088548, "learning_rate": 0.003, "loss": 4.1114, "step": 6400 }, { "epoch": 0.06401, "grad_norm": 0.8919392182837542, "learning_rate": 0.003, "loss": 4.0974, "step": 6401 }, { "epoch": 0.06402, "grad_norm": 0.8417805555156116, "learning_rate": 0.003, "loss": 4.0984, "step": 6402 }, { "epoch": 0.06403, "grad_norm": 0.9725502407864219, "learning_rate": 0.003, "loss": 4.1012, "step": 6403 }, { "epoch": 0.06404, "grad_norm": 1.0332905744103569, "learning_rate": 0.003, "loss": 4.1543, "step": 6404 }, { "epoch": 0.06405, "grad_norm": 1.2514671984682015, "learning_rate": 0.003, "loss": 4.1156, "step": 6405 }, { "epoch": 0.06406, "grad_norm": 1.1553499196304904, "learning_rate": 0.003, "loss": 4.1044, "step": 6406 }, { "epoch": 0.06407, "grad_norm": 0.9874546200949287, "learning_rate": 0.003, "loss": 4.1032, "step": 6407 }, { "epoch": 0.06408, "grad_norm": 1.0027041565820474, "learning_rate": 0.003, "loss": 4.0847, "step": 6408 }, { "epoch": 0.06409, "grad_norm": 1.0037310565561153, "learning_rate": 0.003, "loss": 4.1262, "step": 6409 }, { "epoch": 0.0641, "grad_norm": 1.0349990838623648, "learning_rate": 0.003, "loss": 4.116, "step": 6410 }, { "epoch": 0.06411, "grad_norm": 1.2505472332691576, "learning_rate": 0.003, "loss": 4.1131, "step": 6411 }, { "epoch": 0.06412, "grad_norm": 1.107423585218049, "learning_rate": 0.003, "loss": 4.1046, "step": 6412 }, { "epoch": 0.06413, "grad_norm": 1.0437351552377427, "learning_rate": 0.003, "loss": 4.1383, "step": 6413 }, { "epoch": 0.06414, "grad_norm": 1.1164278775889673, "learning_rate": 0.003, "loss": 4.0994, "step": 6414 }, { "epoch": 0.06415, "grad_norm": 0.9745118316725514, "learning_rate": 0.003, "loss": 4.0916, "step": 6415 }, { "epoch": 0.06416, "grad_norm": 0.9563092447642871, "learning_rate": 0.003, "loss": 4.1326, "step": 6416 }, { "epoch": 0.06417, "grad_norm": 0.9473328664106608, "learning_rate": 0.003, "loss": 4.1021, "step": 6417 }, { "epoch": 0.06418, "grad_norm": 1.006366955338522, "learning_rate": 0.003, "loss": 4.0935, "step": 6418 }, { "epoch": 0.06419, "grad_norm": 1.3186351537429872, "learning_rate": 0.003, "loss": 4.1086, "step": 6419 }, { "epoch": 0.0642, "grad_norm": 0.874887713674597, "learning_rate": 0.003, "loss": 4.1102, "step": 6420 }, { "epoch": 0.06421, "grad_norm": 0.793681194621428, "learning_rate": 0.003, "loss": 4.1217, "step": 6421 }, { "epoch": 0.06422, "grad_norm": 0.8959604597659665, "learning_rate": 0.003, "loss": 4.1115, "step": 6422 }, { "epoch": 0.06423, "grad_norm": 0.9509492440139267, "learning_rate": 0.003, "loss": 4.1261, "step": 6423 }, { "epoch": 0.06424, "grad_norm": 1.1536140647765203, "learning_rate": 0.003, "loss": 4.0898, "step": 6424 }, { "epoch": 0.06425, "grad_norm": 1.0770680367269025, "learning_rate": 0.003, "loss": 4.1109, "step": 6425 }, { "epoch": 0.06426, "grad_norm": 1.2598814102700444, "learning_rate": 0.003, "loss": 4.0816, "step": 6426 }, { "epoch": 0.06427, "grad_norm": 0.8655844051388307, "learning_rate": 0.003, "loss": 4.0913, "step": 6427 }, { "epoch": 0.06428, "grad_norm": 0.8184055602347383, "learning_rate": 0.003, "loss": 4.1119, "step": 6428 }, { "epoch": 0.06429, "grad_norm": 0.8554320627088341, "learning_rate": 0.003, "loss": 4.1128, "step": 6429 }, { "epoch": 0.0643, "grad_norm": 0.8949349865885768, "learning_rate": 0.003, "loss": 4.0857, "step": 6430 }, { "epoch": 0.06431, "grad_norm": 0.934335416030418, "learning_rate": 0.003, "loss": 4.1042, "step": 6431 }, { "epoch": 0.06432, "grad_norm": 0.9265117928117284, "learning_rate": 0.003, "loss": 4.0967, "step": 6432 }, { "epoch": 0.06433, "grad_norm": 1.024797684763137, "learning_rate": 0.003, "loss": 4.1274, "step": 6433 }, { "epoch": 0.06434, "grad_norm": 1.3600455167556202, "learning_rate": 0.003, "loss": 4.0844, "step": 6434 }, { "epoch": 0.06435, "grad_norm": 0.9529262641132417, "learning_rate": 0.003, "loss": 4.0954, "step": 6435 }, { "epoch": 0.06436, "grad_norm": 1.034119140498242, "learning_rate": 0.003, "loss": 4.1157, "step": 6436 }, { "epoch": 0.06437, "grad_norm": 1.0713285668150787, "learning_rate": 0.003, "loss": 4.1141, "step": 6437 }, { "epoch": 0.06438, "grad_norm": 0.9683106581176993, "learning_rate": 0.003, "loss": 4.1059, "step": 6438 }, { "epoch": 0.06439, "grad_norm": 0.9652914703033583, "learning_rate": 0.003, "loss": 4.1039, "step": 6439 }, { "epoch": 0.0644, "grad_norm": 1.0197094752336315, "learning_rate": 0.003, "loss": 4.1024, "step": 6440 }, { "epoch": 0.06441, "grad_norm": 1.0610898230757766, "learning_rate": 0.003, "loss": 4.0846, "step": 6441 }, { "epoch": 0.06442, "grad_norm": 0.9375680251241779, "learning_rate": 0.003, "loss": 4.0986, "step": 6442 }, { "epoch": 0.06443, "grad_norm": 0.882726438804907, "learning_rate": 0.003, "loss": 4.0869, "step": 6443 }, { "epoch": 0.06444, "grad_norm": 0.8666984972124473, "learning_rate": 0.003, "loss": 4.0968, "step": 6444 }, { "epoch": 0.06445, "grad_norm": 1.0588303268695138, "learning_rate": 0.003, "loss": 4.1148, "step": 6445 }, { "epoch": 0.06446, "grad_norm": 1.2128013732008698, "learning_rate": 0.003, "loss": 4.1065, "step": 6446 }, { "epoch": 0.06447, "grad_norm": 1.0289917232865164, "learning_rate": 0.003, "loss": 4.1172, "step": 6447 }, { "epoch": 0.06448, "grad_norm": 1.161447002205071, "learning_rate": 0.003, "loss": 4.1247, "step": 6448 }, { "epoch": 0.06449, "grad_norm": 1.1207683657781975, "learning_rate": 0.003, "loss": 4.15, "step": 6449 }, { "epoch": 0.0645, "grad_norm": 1.1830174596242193, "learning_rate": 0.003, "loss": 4.0992, "step": 6450 }, { "epoch": 0.06451, "grad_norm": 0.8475401638958145, "learning_rate": 0.003, "loss": 4.1394, "step": 6451 }, { "epoch": 0.06452, "grad_norm": 0.864303947498384, "learning_rate": 0.003, "loss": 4.101, "step": 6452 }, { "epoch": 0.06453, "grad_norm": 0.9576553598100936, "learning_rate": 0.003, "loss": 4.1452, "step": 6453 }, { "epoch": 0.06454, "grad_norm": 0.9814262975984199, "learning_rate": 0.003, "loss": 4.1007, "step": 6454 }, { "epoch": 0.06455, "grad_norm": 1.0787724739719247, "learning_rate": 0.003, "loss": 4.1123, "step": 6455 }, { "epoch": 0.06456, "grad_norm": 1.1228799973202512, "learning_rate": 0.003, "loss": 4.132, "step": 6456 }, { "epoch": 0.06457, "grad_norm": 1.0268253444336903, "learning_rate": 0.003, "loss": 4.1089, "step": 6457 }, { "epoch": 0.06458, "grad_norm": 1.0108645808692163, "learning_rate": 0.003, "loss": 4.1194, "step": 6458 }, { "epoch": 0.06459, "grad_norm": 1.1418149699378657, "learning_rate": 0.003, "loss": 4.1345, "step": 6459 }, { "epoch": 0.0646, "grad_norm": 0.8824017335220107, "learning_rate": 0.003, "loss": 4.1047, "step": 6460 }, { "epoch": 0.06461, "grad_norm": 1.0178895025907293, "learning_rate": 0.003, "loss": 4.149, "step": 6461 }, { "epoch": 0.06462, "grad_norm": 1.1346084629753426, "learning_rate": 0.003, "loss": 4.1018, "step": 6462 }, { "epoch": 0.06463, "grad_norm": 1.0391228751646258, "learning_rate": 0.003, "loss": 4.1021, "step": 6463 }, { "epoch": 0.06464, "grad_norm": 1.2899475407345256, "learning_rate": 0.003, "loss": 4.0989, "step": 6464 }, { "epoch": 0.06465, "grad_norm": 1.1687110082198426, "learning_rate": 0.003, "loss": 4.0972, "step": 6465 }, { "epoch": 0.06466, "grad_norm": 0.9657486632446244, "learning_rate": 0.003, "loss": 4.108, "step": 6466 }, { "epoch": 0.06467, "grad_norm": 1.0898402207089803, "learning_rate": 0.003, "loss": 4.1003, "step": 6467 }, { "epoch": 0.06468, "grad_norm": 1.0616683733064285, "learning_rate": 0.003, "loss": 4.1236, "step": 6468 }, { "epoch": 0.06469, "grad_norm": 1.0893197005186201, "learning_rate": 0.003, "loss": 4.1215, "step": 6469 }, { "epoch": 0.0647, "grad_norm": 1.0058220508191251, "learning_rate": 0.003, "loss": 4.1314, "step": 6470 }, { "epoch": 0.06471, "grad_norm": 0.956797815052642, "learning_rate": 0.003, "loss": 4.1055, "step": 6471 }, { "epoch": 0.06472, "grad_norm": 1.0128039862356037, "learning_rate": 0.003, "loss": 4.1165, "step": 6472 }, { "epoch": 0.06473, "grad_norm": 1.2167082158141116, "learning_rate": 0.003, "loss": 4.1294, "step": 6473 }, { "epoch": 0.06474, "grad_norm": 0.7988940618326229, "learning_rate": 0.003, "loss": 4.1391, "step": 6474 }, { "epoch": 0.06475, "grad_norm": 0.7962292180465622, "learning_rate": 0.003, "loss": 4.1319, "step": 6475 }, { "epoch": 0.06476, "grad_norm": 0.7713643156623257, "learning_rate": 0.003, "loss": 4.1011, "step": 6476 }, { "epoch": 0.06477, "grad_norm": 1.0364319372786666, "learning_rate": 0.003, "loss": 4.0932, "step": 6477 }, { "epoch": 0.06478, "grad_norm": 1.1397532882735162, "learning_rate": 0.003, "loss": 4.0904, "step": 6478 }, { "epoch": 0.06479, "grad_norm": 0.7537458893609261, "learning_rate": 0.003, "loss": 4.0978, "step": 6479 }, { "epoch": 0.0648, "grad_norm": 0.8777247030769344, "learning_rate": 0.003, "loss": 4.0926, "step": 6480 }, { "epoch": 0.06481, "grad_norm": 0.8860162677530811, "learning_rate": 0.003, "loss": 4.1055, "step": 6481 }, { "epoch": 0.06482, "grad_norm": 0.7666515029588695, "learning_rate": 0.003, "loss": 4.0994, "step": 6482 }, { "epoch": 0.06483, "grad_norm": 0.8203308848715672, "learning_rate": 0.003, "loss": 4.1105, "step": 6483 }, { "epoch": 0.06484, "grad_norm": 0.8079518261592182, "learning_rate": 0.003, "loss": 4.1056, "step": 6484 }, { "epoch": 0.06485, "grad_norm": 0.8931148106546561, "learning_rate": 0.003, "loss": 4.1285, "step": 6485 }, { "epoch": 0.06486, "grad_norm": 0.9951987941612116, "learning_rate": 0.003, "loss": 4.1297, "step": 6486 }, { "epoch": 0.06487, "grad_norm": 1.041100076013719, "learning_rate": 0.003, "loss": 4.1225, "step": 6487 }, { "epoch": 0.06488, "grad_norm": 1.1977548341810142, "learning_rate": 0.003, "loss": 4.1311, "step": 6488 }, { "epoch": 0.06489, "grad_norm": 1.4315443170694175, "learning_rate": 0.003, "loss": 4.1237, "step": 6489 }, { "epoch": 0.0649, "grad_norm": 0.790139556257875, "learning_rate": 0.003, "loss": 4.1012, "step": 6490 }, { "epoch": 0.06491, "grad_norm": 0.8725006774086241, "learning_rate": 0.003, "loss": 4.1091, "step": 6491 }, { "epoch": 0.06492, "grad_norm": 0.8550610838598619, "learning_rate": 0.003, "loss": 4.121, "step": 6492 }, { "epoch": 0.06493, "grad_norm": 1.0672439328854844, "learning_rate": 0.003, "loss": 4.1126, "step": 6493 }, { "epoch": 0.06494, "grad_norm": 1.1053803317543687, "learning_rate": 0.003, "loss": 4.1258, "step": 6494 }, { "epoch": 0.06495, "grad_norm": 1.2245048272685835, "learning_rate": 0.003, "loss": 4.1162, "step": 6495 }, { "epoch": 0.06496, "grad_norm": 1.1827354647076507, "learning_rate": 0.003, "loss": 4.0856, "step": 6496 }, { "epoch": 0.06497, "grad_norm": 0.8875741851301479, "learning_rate": 0.003, "loss": 4.1044, "step": 6497 }, { "epoch": 0.06498, "grad_norm": 0.9405556784618091, "learning_rate": 0.003, "loss": 4.0968, "step": 6498 }, { "epoch": 0.06499, "grad_norm": 1.1518206025009547, "learning_rate": 0.003, "loss": 4.1345, "step": 6499 }, { "epoch": 0.065, "grad_norm": 1.1135199791235904, "learning_rate": 0.003, "loss": 4.1104, "step": 6500 }, { "epoch": 0.06501, "grad_norm": 0.9426453963848523, "learning_rate": 0.003, "loss": 4.1028, "step": 6501 }, { "epoch": 0.06502, "grad_norm": 0.906946070942705, "learning_rate": 0.003, "loss": 4.1102, "step": 6502 }, { "epoch": 0.06503, "grad_norm": 0.8632347389219635, "learning_rate": 0.003, "loss": 4.1244, "step": 6503 }, { "epoch": 0.06504, "grad_norm": 0.9623491280205102, "learning_rate": 0.003, "loss": 4.1057, "step": 6504 }, { "epoch": 0.06505, "grad_norm": 1.4229366271424375, "learning_rate": 0.003, "loss": 4.1182, "step": 6505 }, { "epoch": 0.06506, "grad_norm": 1.0451736131803195, "learning_rate": 0.003, "loss": 4.0946, "step": 6506 }, { "epoch": 0.06507, "grad_norm": 1.2294334171006416, "learning_rate": 0.003, "loss": 4.0789, "step": 6507 }, { "epoch": 0.06508, "grad_norm": 0.9445531435779846, "learning_rate": 0.003, "loss": 4.1095, "step": 6508 }, { "epoch": 0.06509, "grad_norm": 0.9256926488563311, "learning_rate": 0.003, "loss": 4.0954, "step": 6509 }, { "epoch": 0.0651, "grad_norm": 1.2262032469738913, "learning_rate": 0.003, "loss": 4.1386, "step": 6510 }, { "epoch": 0.06511, "grad_norm": 1.0322387593080742, "learning_rate": 0.003, "loss": 4.0973, "step": 6511 }, { "epoch": 0.06512, "grad_norm": 1.322871770532383, "learning_rate": 0.003, "loss": 4.0921, "step": 6512 }, { "epoch": 0.06513, "grad_norm": 1.0020496514714246, "learning_rate": 0.003, "loss": 4.1176, "step": 6513 }, { "epoch": 0.06514, "grad_norm": 1.0807327376240277, "learning_rate": 0.003, "loss": 4.1059, "step": 6514 }, { "epoch": 0.06515, "grad_norm": 0.9992369872861028, "learning_rate": 0.003, "loss": 4.0905, "step": 6515 }, { "epoch": 0.06516, "grad_norm": 1.0690181996728259, "learning_rate": 0.003, "loss": 4.1105, "step": 6516 }, { "epoch": 0.06517, "grad_norm": 1.0205546370680254, "learning_rate": 0.003, "loss": 4.1091, "step": 6517 }, { "epoch": 0.06518, "grad_norm": 0.9661649072469782, "learning_rate": 0.003, "loss": 4.127, "step": 6518 }, { "epoch": 0.06519, "grad_norm": 1.1841310244092509, "learning_rate": 0.003, "loss": 4.1091, "step": 6519 }, { "epoch": 0.0652, "grad_norm": 1.0151264535711446, "learning_rate": 0.003, "loss": 4.0723, "step": 6520 }, { "epoch": 0.06521, "grad_norm": 1.1533183181347186, "learning_rate": 0.003, "loss": 4.1242, "step": 6521 }, { "epoch": 0.06522, "grad_norm": 1.0222075376967539, "learning_rate": 0.003, "loss": 4.1231, "step": 6522 }, { "epoch": 0.06523, "grad_norm": 1.1249451027703596, "learning_rate": 0.003, "loss": 4.1144, "step": 6523 }, { "epoch": 0.06524, "grad_norm": 0.9778774962433593, "learning_rate": 0.003, "loss": 4.1084, "step": 6524 }, { "epoch": 0.06525, "grad_norm": 1.038580395735807, "learning_rate": 0.003, "loss": 4.127, "step": 6525 }, { "epoch": 0.06526, "grad_norm": 1.075878133922806, "learning_rate": 0.003, "loss": 4.1126, "step": 6526 }, { "epoch": 0.06527, "grad_norm": 1.025791974704542, "learning_rate": 0.003, "loss": 4.1221, "step": 6527 }, { "epoch": 0.06528, "grad_norm": 0.9973318916151819, "learning_rate": 0.003, "loss": 4.1242, "step": 6528 }, { "epoch": 0.06529, "grad_norm": 1.2162973451905112, "learning_rate": 0.003, "loss": 4.1154, "step": 6529 }, { "epoch": 0.0653, "grad_norm": 0.9871372103084888, "learning_rate": 0.003, "loss": 4.1073, "step": 6530 }, { "epoch": 0.06531, "grad_norm": 1.1945236960287906, "learning_rate": 0.003, "loss": 4.0991, "step": 6531 }, { "epoch": 0.06532, "grad_norm": 0.9401643138302784, "learning_rate": 0.003, "loss": 4.1268, "step": 6532 }, { "epoch": 0.06533, "grad_norm": 1.1178892277715633, "learning_rate": 0.003, "loss": 4.1278, "step": 6533 }, { "epoch": 0.06534, "grad_norm": 0.9832931153486173, "learning_rate": 0.003, "loss": 4.1172, "step": 6534 }, { "epoch": 0.06535, "grad_norm": 0.8533727635187061, "learning_rate": 0.003, "loss": 4.0761, "step": 6535 }, { "epoch": 0.06536, "grad_norm": 0.7504468053256232, "learning_rate": 0.003, "loss": 4.091, "step": 6536 }, { "epoch": 0.06537, "grad_norm": 0.881013336525802, "learning_rate": 0.003, "loss": 4.1088, "step": 6537 }, { "epoch": 0.06538, "grad_norm": 1.0943393042527045, "learning_rate": 0.003, "loss": 4.1178, "step": 6538 }, { "epoch": 0.06539, "grad_norm": 1.1900608228822722, "learning_rate": 0.003, "loss": 4.0907, "step": 6539 }, { "epoch": 0.0654, "grad_norm": 0.8415174000109801, "learning_rate": 0.003, "loss": 4.1428, "step": 6540 }, { "epoch": 0.06541, "grad_norm": 0.6953210773431162, "learning_rate": 0.003, "loss": 4.1011, "step": 6541 }, { "epoch": 0.06542, "grad_norm": 0.7466595601641312, "learning_rate": 0.003, "loss": 4.0824, "step": 6542 }, { "epoch": 0.06543, "grad_norm": 0.7975588171096655, "learning_rate": 0.003, "loss": 4.1126, "step": 6543 }, { "epoch": 0.06544, "grad_norm": 0.8156245001878942, "learning_rate": 0.003, "loss": 4.1069, "step": 6544 }, { "epoch": 0.06545, "grad_norm": 0.9060264548647263, "learning_rate": 0.003, "loss": 4.114, "step": 6545 }, { "epoch": 0.06546, "grad_norm": 0.9960765646804454, "learning_rate": 0.003, "loss": 4.1009, "step": 6546 }, { "epoch": 0.06547, "grad_norm": 1.3831374261427445, "learning_rate": 0.003, "loss": 4.1087, "step": 6547 }, { "epoch": 0.06548, "grad_norm": 0.782948712181559, "learning_rate": 0.003, "loss": 4.0992, "step": 6548 }, { "epoch": 0.06549, "grad_norm": 0.8977206509308298, "learning_rate": 0.003, "loss": 4.1313, "step": 6549 }, { "epoch": 0.0655, "grad_norm": 1.120628275139245, "learning_rate": 0.003, "loss": 4.085, "step": 6550 }, { "epoch": 0.06551, "grad_norm": 1.2840718294242444, "learning_rate": 0.003, "loss": 4.0926, "step": 6551 }, { "epoch": 0.06552, "grad_norm": 1.1241386761670837, "learning_rate": 0.003, "loss": 4.1271, "step": 6552 }, { "epoch": 0.06553, "grad_norm": 0.7870751727024481, "learning_rate": 0.003, "loss": 4.0998, "step": 6553 }, { "epoch": 0.06554, "grad_norm": 0.6531268213697407, "learning_rate": 0.003, "loss": 4.1027, "step": 6554 }, { "epoch": 0.06555, "grad_norm": 0.7458288916343057, "learning_rate": 0.003, "loss": 4.0972, "step": 6555 }, { "epoch": 0.06556, "grad_norm": 0.7935619352156358, "learning_rate": 0.003, "loss": 4.1172, "step": 6556 }, { "epoch": 0.06557, "grad_norm": 0.9338290170497964, "learning_rate": 0.003, "loss": 4.0948, "step": 6557 }, { "epoch": 0.06558, "grad_norm": 1.163914741576227, "learning_rate": 0.003, "loss": 4.134, "step": 6558 }, { "epoch": 0.06559, "grad_norm": 1.030586956016889, "learning_rate": 0.003, "loss": 4.11, "step": 6559 }, { "epoch": 0.0656, "grad_norm": 1.1181941312817716, "learning_rate": 0.003, "loss": 4.1146, "step": 6560 }, { "epoch": 0.06561, "grad_norm": 1.0126984372372947, "learning_rate": 0.003, "loss": 4.1282, "step": 6561 }, { "epoch": 0.06562, "grad_norm": 1.0009991860930807, "learning_rate": 0.003, "loss": 4.0863, "step": 6562 }, { "epoch": 0.06563, "grad_norm": 1.1006051624397504, "learning_rate": 0.003, "loss": 4.1136, "step": 6563 }, { "epoch": 0.06564, "grad_norm": 1.0608902444027226, "learning_rate": 0.003, "loss": 4.1179, "step": 6564 }, { "epoch": 0.06565, "grad_norm": 1.0951281729266462, "learning_rate": 0.003, "loss": 4.0883, "step": 6565 }, { "epoch": 0.06566, "grad_norm": 1.1144896086481675, "learning_rate": 0.003, "loss": 4.1154, "step": 6566 }, { "epoch": 0.06567, "grad_norm": 1.2539223651632134, "learning_rate": 0.003, "loss": 4.1325, "step": 6567 }, { "epoch": 0.06568, "grad_norm": 1.2136692336762676, "learning_rate": 0.003, "loss": 4.1423, "step": 6568 }, { "epoch": 0.06569, "grad_norm": 1.0107611711365292, "learning_rate": 0.003, "loss": 4.0837, "step": 6569 }, { "epoch": 0.0657, "grad_norm": 1.1701756562475913, "learning_rate": 0.003, "loss": 4.1126, "step": 6570 }, { "epoch": 0.06571, "grad_norm": 1.118588414395485, "learning_rate": 0.003, "loss": 4.1175, "step": 6571 }, { "epoch": 0.06572, "grad_norm": 1.1618445605816499, "learning_rate": 0.003, "loss": 4.0856, "step": 6572 }, { "epoch": 0.06573, "grad_norm": 1.0406499653295629, "learning_rate": 0.003, "loss": 4.1103, "step": 6573 }, { "epoch": 0.06574, "grad_norm": 1.2822924507807092, "learning_rate": 0.003, "loss": 4.1044, "step": 6574 }, { "epoch": 0.06575, "grad_norm": 0.9150523557477509, "learning_rate": 0.003, "loss": 4.0927, "step": 6575 }, { "epoch": 0.06576, "grad_norm": 0.9513417232803534, "learning_rate": 0.003, "loss": 4.0791, "step": 6576 }, { "epoch": 0.06577, "grad_norm": 1.1218399136336974, "learning_rate": 0.003, "loss": 4.0929, "step": 6577 }, { "epoch": 0.06578, "grad_norm": 0.9394344957421601, "learning_rate": 0.003, "loss": 4.097, "step": 6578 }, { "epoch": 0.06579, "grad_norm": 1.0809645820357565, "learning_rate": 0.003, "loss": 4.0888, "step": 6579 }, { "epoch": 0.0658, "grad_norm": 1.214518886759514, "learning_rate": 0.003, "loss": 4.0871, "step": 6580 }, { "epoch": 0.06581, "grad_norm": 0.9638219433089225, "learning_rate": 0.003, "loss": 4.124, "step": 6581 }, { "epoch": 0.06582, "grad_norm": 0.9046848586781843, "learning_rate": 0.003, "loss": 4.1157, "step": 6582 }, { "epoch": 0.06583, "grad_norm": 0.8609622707678348, "learning_rate": 0.003, "loss": 4.0861, "step": 6583 }, { "epoch": 0.06584, "grad_norm": 0.8835791819190192, "learning_rate": 0.003, "loss": 4.0766, "step": 6584 }, { "epoch": 0.06585, "grad_norm": 0.9987318797336011, "learning_rate": 0.003, "loss": 4.1013, "step": 6585 }, { "epoch": 0.06586, "grad_norm": 1.1355073490088379, "learning_rate": 0.003, "loss": 4.1172, "step": 6586 }, { "epoch": 0.06587, "grad_norm": 1.0085175293063395, "learning_rate": 0.003, "loss": 4.1179, "step": 6587 }, { "epoch": 0.06588, "grad_norm": 0.9620876182851724, "learning_rate": 0.003, "loss": 4.1364, "step": 6588 }, { "epoch": 0.06589, "grad_norm": 1.048954389785324, "learning_rate": 0.003, "loss": 4.1253, "step": 6589 }, { "epoch": 0.0659, "grad_norm": 1.0238341074758999, "learning_rate": 0.003, "loss": 4.1349, "step": 6590 }, { "epoch": 0.06591, "grad_norm": 1.075709048948834, "learning_rate": 0.003, "loss": 4.1575, "step": 6591 }, { "epoch": 0.06592, "grad_norm": 1.1681526348561144, "learning_rate": 0.003, "loss": 4.1261, "step": 6592 }, { "epoch": 0.06593, "grad_norm": 0.9756080110323221, "learning_rate": 0.003, "loss": 4.1017, "step": 6593 }, { "epoch": 0.06594, "grad_norm": 1.0796908095925524, "learning_rate": 0.003, "loss": 4.1252, "step": 6594 }, { "epoch": 0.06595, "grad_norm": 1.0723639338576645, "learning_rate": 0.003, "loss": 4.0764, "step": 6595 }, { "epoch": 0.06596, "grad_norm": 1.2099686727476, "learning_rate": 0.003, "loss": 4.1326, "step": 6596 }, { "epoch": 0.06597, "grad_norm": 1.020495761358447, "learning_rate": 0.003, "loss": 4.1058, "step": 6597 }, { "epoch": 0.06598, "grad_norm": 1.250785346850008, "learning_rate": 0.003, "loss": 4.1453, "step": 6598 }, { "epoch": 0.06599, "grad_norm": 0.8082604467548349, "learning_rate": 0.003, "loss": 4.1178, "step": 6599 }, { "epoch": 0.066, "grad_norm": 0.7311810832534161, "learning_rate": 0.003, "loss": 4.1209, "step": 6600 }, { "epoch": 0.06601, "grad_norm": 0.8646948838436856, "learning_rate": 0.003, "loss": 4.1098, "step": 6601 }, { "epoch": 0.06602, "grad_norm": 1.2019272374710974, "learning_rate": 0.003, "loss": 4.1127, "step": 6602 }, { "epoch": 0.06603, "grad_norm": 0.99896718805981, "learning_rate": 0.003, "loss": 4.1045, "step": 6603 }, { "epoch": 0.06604, "grad_norm": 1.1797514234958721, "learning_rate": 0.003, "loss": 4.1059, "step": 6604 }, { "epoch": 0.06605, "grad_norm": 1.2720223453313813, "learning_rate": 0.003, "loss": 4.1313, "step": 6605 }, { "epoch": 0.06606, "grad_norm": 0.906064721496246, "learning_rate": 0.003, "loss": 4.1049, "step": 6606 }, { "epoch": 0.06607, "grad_norm": 0.8738244371131962, "learning_rate": 0.003, "loss": 4.0983, "step": 6607 }, { "epoch": 0.06608, "grad_norm": 0.9363228945652666, "learning_rate": 0.003, "loss": 4.0973, "step": 6608 }, { "epoch": 0.06609, "grad_norm": 1.1416719220665938, "learning_rate": 0.003, "loss": 4.1219, "step": 6609 }, { "epoch": 0.0661, "grad_norm": 1.220985108255946, "learning_rate": 0.003, "loss": 4.1245, "step": 6610 }, { "epoch": 0.06611, "grad_norm": 1.137872261131753, "learning_rate": 0.003, "loss": 4.1059, "step": 6611 }, { "epoch": 0.06612, "grad_norm": 1.1395613837870757, "learning_rate": 0.003, "loss": 4.1021, "step": 6612 }, { "epoch": 0.06613, "grad_norm": 1.191648524183824, "learning_rate": 0.003, "loss": 4.1197, "step": 6613 }, { "epoch": 0.06614, "grad_norm": 0.8858296562908814, "learning_rate": 0.003, "loss": 4.114, "step": 6614 }, { "epoch": 0.06615, "grad_norm": 0.9819337255984676, "learning_rate": 0.003, "loss": 4.0937, "step": 6615 }, { "epoch": 0.06616, "grad_norm": 1.1867192771582191, "learning_rate": 0.003, "loss": 4.1552, "step": 6616 }, { "epoch": 0.06617, "grad_norm": 0.9053344714711246, "learning_rate": 0.003, "loss": 4.1233, "step": 6617 }, { "epoch": 0.06618, "grad_norm": 0.9219240319109411, "learning_rate": 0.003, "loss": 4.1333, "step": 6618 }, { "epoch": 0.06619, "grad_norm": 0.991583160819303, "learning_rate": 0.003, "loss": 4.1026, "step": 6619 }, { "epoch": 0.0662, "grad_norm": 1.1650808913340467, "learning_rate": 0.003, "loss": 4.097, "step": 6620 }, { "epoch": 0.06621, "grad_norm": 0.9080740223120065, "learning_rate": 0.003, "loss": 4.1248, "step": 6621 }, { "epoch": 0.06622, "grad_norm": 0.8680994255398555, "learning_rate": 0.003, "loss": 4.1174, "step": 6622 }, { "epoch": 0.06623, "grad_norm": 0.8628560499230488, "learning_rate": 0.003, "loss": 4.102, "step": 6623 }, { "epoch": 0.06624, "grad_norm": 0.845810905345881, "learning_rate": 0.003, "loss": 4.1235, "step": 6624 }, { "epoch": 0.06625, "grad_norm": 0.8753283211615485, "learning_rate": 0.003, "loss": 4.1089, "step": 6625 }, { "epoch": 0.06626, "grad_norm": 0.8898864094278442, "learning_rate": 0.003, "loss": 4.1275, "step": 6626 }, { "epoch": 0.06627, "grad_norm": 0.8556887714824031, "learning_rate": 0.003, "loss": 4.115, "step": 6627 }, { "epoch": 0.06628, "grad_norm": 0.8438393299522239, "learning_rate": 0.003, "loss": 4.1156, "step": 6628 }, { "epoch": 0.06629, "grad_norm": 1.0104192170537771, "learning_rate": 0.003, "loss": 4.1141, "step": 6629 }, { "epoch": 0.0663, "grad_norm": 1.4015375334692346, "learning_rate": 0.003, "loss": 4.1415, "step": 6630 }, { "epoch": 0.06631, "grad_norm": 0.9722935737313815, "learning_rate": 0.003, "loss": 4.1089, "step": 6631 }, { "epoch": 0.06632, "grad_norm": 0.9803618166102699, "learning_rate": 0.003, "loss": 4.1123, "step": 6632 }, { "epoch": 0.06633, "grad_norm": 1.1296705060941024, "learning_rate": 0.003, "loss": 4.1345, "step": 6633 }, { "epoch": 0.06634, "grad_norm": 1.0907326109387054, "learning_rate": 0.003, "loss": 4.1204, "step": 6634 }, { "epoch": 0.06635, "grad_norm": 0.9435618555110209, "learning_rate": 0.003, "loss": 4.1341, "step": 6635 }, { "epoch": 0.06636, "grad_norm": 1.0285537900430537, "learning_rate": 0.003, "loss": 4.0925, "step": 6636 }, { "epoch": 0.06637, "grad_norm": 1.202489591650724, "learning_rate": 0.003, "loss": 4.1195, "step": 6637 }, { "epoch": 0.06638, "grad_norm": 1.017321963175585, "learning_rate": 0.003, "loss": 4.0822, "step": 6638 }, { "epoch": 0.06639, "grad_norm": 1.0524483668700617, "learning_rate": 0.003, "loss": 4.1168, "step": 6639 }, { "epoch": 0.0664, "grad_norm": 1.1482240654374392, "learning_rate": 0.003, "loss": 4.1017, "step": 6640 }, { "epoch": 0.06641, "grad_norm": 0.9809058091856554, "learning_rate": 0.003, "loss": 4.1153, "step": 6641 }, { "epoch": 0.06642, "grad_norm": 1.1082029804560825, "learning_rate": 0.003, "loss": 4.1234, "step": 6642 }, { "epoch": 0.06643, "grad_norm": 1.1934857560760654, "learning_rate": 0.003, "loss": 4.1151, "step": 6643 }, { "epoch": 0.06644, "grad_norm": 0.9229309859195602, "learning_rate": 0.003, "loss": 4.1113, "step": 6644 }, { "epoch": 0.06645, "grad_norm": 1.072671684854488, "learning_rate": 0.003, "loss": 4.098, "step": 6645 }, { "epoch": 0.06646, "grad_norm": 1.170997003327203, "learning_rate": 0.003, "loss": 4.102, "step": 6646 }, { "epoch": 0.06647, "grad_norm": 1.3361960097562062, "learning_rate": 0.003, "loss": 4.1346, "step": 6647 }, { "epoch": 0.06648, "grad_norm": 0.814630751575606, "learning_rate": 0.003, "loss": 4.0984, "step": 6648 }, { "epoch": 0.06649, "grad_norm": 0.7234763720955096, "learning_rate": 0.003, "loss": 4.1066, "step": 6649 }, { "epoch": 0.0665, "grad_norm": 0.8177448387254658, "learning_rate": 0.003, "loss": 4.1051, "step": 6650 }, { "epoch": 0.06651, "grad_norm": 0.9720558429276546, "learning_rate": 0.003, "loss": 4.0725, "step": 6651 }, { "epoch": 0.06652, "grad_norm": 1.194866309569147, "learning_rate": 0.003, "loss": 4.1084, "step": 6652 }, { "epoch": 0.06653, "grad_norm": 0.8816716043893494, "learning_rate": 0.003, "loss": 4.1276, "step": 6653 }, { "epoch": 0.06654, "grad_norm": 1.090559865889525, "learning_rate": 0.003, "loss": 4.0941, "step": 6654 }, { "epoch": 0.06655, "grad_norm": 1.0705116003746482, "learning_rate": 0.003, "loss": 4.109, "step": 6655 }, { "epoch": 0.06656, "grad_norm": 1.2331529032280468, "learning_rate": 0.003, "loss": 4.0966, "step": 6656 }, { "epoch": 0.06657, "grad_norm": 0.9218838449430262, "learning_rate": 0.003, "loss": 4.1117, "step": 6657 }, { "epoch": 0.06658, "grad_norm": 1.0193847734871568, "learning_rate": 0.003, "loss": 4.1033, "step": 6658 }, { "epoch": 0.06659, "grad_norm": 1.2471775692333547, "learning_rate": 0.003, "loss": 4.1247, "step": 6659 }, { "epoch": 0.0666, "grad_norm": 1.0479297195190378, "learning_rate": 0.003, "loss": 4.1018, "step": 6660 }, { "epoch": 0.06661, "grad_norm": 1.1007839298335338, "learning_rate": 0.003, "loss": 4.1016, "step": 6661 }, { "epoch": 0.06662, "grad_norm": 0.9250727846278579, "learning_rate": 0.003, "loss": 4.0926, "step": 6662 }, { "epoch": 0.06663, "grad_norm": 0.9825952693837798, "learning_rate": 0.003, "loss": 4.1601, "step": 6663 }, { "epoch": 0.06664, "grad_norm": 1.3207328842359807, "learning_rate": 0.003, "loss": 4.1293, "step": 6664 }, { "epoch": 0.06665, "grad_norm": 1.0063660849323326, "learning_rate": 0.003, "loss": 4.1467, "step": 6665 }, { "epoch": 0.06666, "grad_norm": 1.0152175322629062, "learning_rate": 0.003, "loss": 4.0764, "step": 6666 }, { "epoch": 0.06667, "grad_norm": 1.2629059213515375, "learning_rate": 0.003, "loss": 4.1108, "step": 6667 }, { "epoch": 0.06668, "grad_norm": 1.0230139614340956, "learning_rate": 0.003, "loss": 4.1052, "step": 6668 }, { "epoch": 0.06669, "grad_norm": 1.168998935273423, "learning_rate": 0.003, "loss": 4.1106, "step": 6669 }, { "epoch": 0.0667, "grad_norm": 0.8296397320581785, "learning_rate": 0.003, "loss": 4.1178, "step": 6670 }, { "epoch": 0.06671, "grad_norm": 1.001643971176605, "learning_rate": 0.003, "loss": 4.1143, "step": 6671 }, { "epoch": 0.06672, "grad_norm": 1.1457901096644219, "learning_rate": 0.003, "loss": 4.1024, "step": 6672 }, { "epoch": 0.06673, "grad_norm": 1.207557539094707, "learning_rate": 0.003, "loss": 4.1017, "step": 6673 }, { "epoch": 0.06674, "grad_norm": 1.182055831031003, "learning_rate": 0.003, "loss": 4.128, "step": 6674 }, { "epoch": 0.06675, "grad_norm": 1.034589515323592, "learning_rate": 0.003, "loss": 4.0683, "step": 6675 }, { "epoch": 0.06676, "grad_norm": 1.0280150455811632, "learning_rate": 0.003, "loss": 4.1206, "step": 6676 }, { "epoch": 0.06677, "grad_norm": 1.1343104818692615, "learning_rate": 0.003, "loss": 4.1346, "step": 6677 }, { "epoch": 0.06678, "grad_norm": 1.0649939504536055, "learning_rate": 0.003, "loss": 4.1415, "step": 6678 }, { "epoch": 0.06679, "grad_norm": 1.046088080295711, "learning_rate": 0.003, "loss": 4.1132, "step": 6679 }, { "epoch": 0.0668, "grad_norm": 1.1012078408727637, "learning_rate": 0.003, "loss": 4.1501, "step": 6680 }, { "epoch": 0.06681, "grad_norm": 1.0162711597155334, "learning_rate": 0.003, "loss": 4.1012, "step": 6681 }, { "epoch": 0.06682, "grad_norm": 1.2483056090722793, "learning_rate": 0.003, "loss": 4.1331, "step": 6682 }, { "epoch": 0.06683, "grad_norm": 0.9804124144487479, "learning_rate": 0.003, "loss": 4.1175, "step": 6683 }, { "epoch": 0.06684, "grad_norm": 1.0642062663332452, "learning_rate": 0.003, "loss": 4.1204, "step": 6684 }, { "epoch": 0.06685, "grad_norm": 1.2069965741700133, "learning_rate": 0.003, "loss": 4.1153, "step": 6685 }, { "epoch": 0.06686, "grad_norm": 0.8152999583568941, "learning_rate": 0.003, "loss": 4.093, "step": 6686 }, { "epoch": 0.06687, "grad_norm": 0.7938321166951481, "learning_rate": 0.003, "loss": 4.0798, "step": 6687 }, { "epoch": 0.06688, "grad_norm": 0.8347836923393193, "learning_rate": 0.003, "loss": 4.0946, "step": 6688 }, { "epoch": 0.06689, "grad_norm": 1.0457470131352429, "learning_rate": 0.003, "loss": 4.1329, "step": 6689 }, { "epoch": 0.0669, "grad_norm": 1.233670905876264, "learning_rate": 0.003, "loss": 4.1037, "step": 6690 }, { "epoch": 0.06691, "grad_norm": 0.8950769144466896, "learning_rate": 0.003, "loss": 4.1102, "step": 6691 }, { "epoch": 0.06692, "grad_norm": 0.9963325414694945, "learning_rate": 0.003, "loss": 4.0881, "step": 6692 }, { "epoch": 0.06693, "grad_norm": 1.1667705229561567, "learning_rate": 0.003, "loss": 4.1083, "step": 6693 }, { "epoch": 0.06694, "grad_norm": 1.0395357257985816, "learning_rate": 0.003, "loss": 4.1018, "step": 6694 }, { "epoch": 0.06695, "grad_norm": 1.0389020701257194, "learning_rate": 0.003, "loss": 4.1036, "step": 6695 }, { "epoch": 0.06696, "grad_norm": 0.9090318347719712, "learning_rate": 0.003, "loss": 4.1038, "step": 6696 }, { "epoch": 0.06697, "grad_norm": 1.1170620435815923, "learning_rate": 0.003, "loss": 4.1156, "step": 6697 }, { "epoch": 0.06698, "grad_norm": 1.090791170002388, "learning_rate": 0.003, "loss": 4.1051, "step": 6698 }, { "epoch": 0.06699, "grad_norm": 0.9919188684094656, "learning_rate": 0.003, "loss": 4.1013, "step": 6699 }, { "epoch": 0.067, "grad_norm": 1.049150223859985, "learning_rate": 0.003, "loss": 4.1221, "step": 6700 }, { "epoch": 0.06701, "grad_norm": 0.9745571304224953, "learning_rate": 0.003, "loss": 4.1199, "step": 6701 }, { "epoch": 0.06702, "grad_norm": 1.1042289026728087, "learning_rate": 0.003, "loss": 4.1106, "step": 6702 }, { "epoch": 0.06703, "grad_norm": 1.1602988891479034, "learning_rate": 0.003, "loss": 4.1277, "step": 6703 }, { "epoch": 0.06704, "grad_norm": 1.0541345749074789, "learning_rate": 0.003, "loss": 4.0967, "step": 6704 }, { "epoch": 0.06705, "grad_norm": 1.151339776734197, "learning_rate": 0.003, "loss": 4.1074, "step": 6705 }, { "epoch": 0.06706, "grad_norm": 0.9294993068889601, "learning_rate": 0.003, "loss": 4.1488, "step": 6706 }, { "epoch": 0.06707, "grad_norm": 1.2952523670027591, "learning_rate": 0.003, "loss": 4.0869, "step": 6707 }, { "epoch": 0.06708, "grad_norm": 1.2446326352660062, "learning_rate": 0.003, "loss": 4.1089, "step": 6708 }, { "epoch": 0.06709, "grad_norm": 1.1128123400028667, "learning_rate": 0.003, "loss": 4.1044, "step": 6709 }, { "epoch": 0.0671, "grad_norm": 1.0751514025340694, "learning_rate": 0.003, "loss": 4.1266, "step": 6710 }, { "epoch": 0.06711, "grad_norm": 0.9972877341330995, "learning_rate": 0.003, "loss": 4.1379, "step": 6711 }, { "epoch": 0.06712, "grad_norm": 1.335174960677335, "learning_rate": 0.003, "loss": 4.0973, "step": 6712 }, { "epoch": 0.06713, "grad_norm": 1.0604145566158192, "learning_rate": 0.003, "loss": 4.105, "step": 6713 }, { "epoch": 0.06714, "grad_norm": 0.9845536309503395, "learning_rate": 0.003, "loss": 4.1056, "step": 6714 }, { "epoch": 0.06715, "grad_norm": 0.98104073747809, "learning_rate": 0.003, "loss": 4.1076, "step": 6715 }, { "epoch": 0.06716, "grad_norm": 1.0253080591457324, "learning_rate": 0.003, "loss": 4.1148, "step": 6716 }, { "epoch": 0.06717, "grad_norm": 1.1656130950903412, "learning_rate": 0.003, "loss": 4.1405, "step": 6717 }, { "epoch": 0.06718, "grad_norm": 0.890282809238404, "learning_rate": 0.003, "loss": 4.0939, "step": 6718 }, { "epoch": 0.06719, "grad_norm": 1.100865501357716, "learning_rate": 0.003, "loss": 4.1, "step": 6719 }, { "epoch": 0.0672, "grad_norm": 1.2966985301388, "learning_rate": 0.003, "loss": 4.1179, "step": 6720 }, { "epoch": 0.06721, "grad_norm": 0.7681709705927325, "learning_rate": 0.003, "loss": 4.1204, "step": 6721 }, { "epoch": 0.06722, "grad_norm": 0.7579242685661051, "learning_rate": 0.003, "loss": 4.1277, "step": 6722 }, { "epoch": 0.06723, "grad_norm": 0.8463786269974299, "learning_rate": 0.003, "loss": 4.1179, "step": 6723 }, { "epoch": 0.06724, "grad_norm": 1.2355718601511936, "learning_rate": 0.003, "loss": 4.0856, "step": 6724 }, { "epoch": 0.06725, "grad_norm": 0.9973220857824812, "learning_rate": 0.003, "loss": 4.1143, "step": 6725 }, { "epoch": 0.06726, "grad_norm": 1.3121262958901847, "learning_rate": 0.003, "loss": 4.1473, "step": 6726 }, { "epoch": 0.06727, "grad_norm": 0.7506377029622459, "learning_rate": 0.003, "loss": 4.1019, "step": 6727 }, { "epoch": 0.06728, "grad_norm": 0.7375690656350768, "learning_rate": 0.003, "loss": 4.0699, "step": 6728 }, { "epoch": 0.06729, "grad_norm": 0.8237808367201397, "learning_rate": 0.003, "loss": 4.0942, "step": 6729 }, { "epoch": 0.0673, "grad_norm": 0.93797912273744, "learning_rate": 0.003, "loss": 4.1064, "step": 6730 }, { "epoch": 0.06731, "grad_norm": 0.9765511050435496, "learning_rate": 0.003, "loss": 4.0792, "step": 6731 }, { "epoch": 0.06732, "grad_norm": 1.1004645713502215, "learning_rate": 0.003, "loss": 4.1075, "step": 6732 }, { "epoch": 0.06733, "grad_norm": 1.1027248682447823, "learning_rate": 0.003, "loss": 4.1197, "step": 6733 }, { "epoch": 0.06734, "grad_norm": 1.1253735749654317, "learning_rate": 0.003, "loss": 4.1141, "step": 6734 }, { "epoch": 0.06735, "grad_norm": 1.003542734520633, "learning_rate": 0.003, "loss": 4.0897, "step": 6735 }, { "epoch": 0.06736, "grad_norm": 1.1429551938720557, "learning_rate": 0.003, "loss": 4.0999, "step": 6736 }, { "epoch": 0.06737, "grad_norm": 1.0609352949018371, "learning_rate": 0.003, "loss": 4.107, "step": 6737 }, { "epoch": 0.06738, "grad_norm": 1.0412380492368407, "learning_rate": 0.003, "loss": 4.1004, "step": 6738 }, { "epoch": 0.06739, "grad_norm": 1.225522061118244, "learning_rate": 0.003, "loss": 4.1134, "step": 6739 }, { "epoch": 0.0674, "grad_norm": 0.9920747285040594, "learning_rate": 0.003, "loss": 4.1025, "step": 6740 }, { "epoch": 0.06741, "grad_norm": 1.3728636340948883, "learning_rate": 0.003, "loss": 4.102, "step": 6741 }, { "epoch": 0.06742, "grad_norm": 1.12361874184517, "learning_rate": 0.003, "loss": 4.1229, "step": 6742 }, { "epoch": 0.06743, "grad_norm": 0.9616698612197171, "learning_rate": 0.003, "loss": 4.0871, "step": 6743 }, { "epoch": 0.06744, "grad_norm": 1.1122702070965411, "learning_rate": 0.003, "loss": 4.1266, "step": 6744 }, { "epoch": 0.06745, "grad_norm": 1.1170307488637157, "learning_rate": 0.003, "loss": 4.0993, "step": 6745 }, { "epoch": 0.06746, "grad_norm": 1.014704964150931, "learning_rate": 0.003, "loss": 4.1036, "step": 6746 }, { "epoch": 0.06747, "grad_norm": 1.0082376235883914, "learning_rate": 0.003, "loss": 4.1082, "step": 6747 }, { "epoch": 0.06748, "grad_norm": 1.0285387580619565, "learning_rate": 0.003, "loss": 4.1137, "step": 6748 }, { "epoch": 0.06749, "grad_norm": 0.9949987658288557, "learning_rate": 0.003, "loss": 4.1048, "step": 6749 }, { "epoch": 0.0675, "grad_norm": 1.012959502751053, "learning_rate": 0.003, "loss": 4.0948, "step": 6750 }, { "epoch": 0.06751, "grad_norm": 1.159863000962202, "learning_rate": 0.003, "loss": 4.1354, "step": 6751 }, { "epoch": 0.06752, "grad_norm": 1.158507578022585, "learning_rate": 0.003, "loss": 4.1294, "step": 6752 }, { "epoch": 0.06753, "grad_norm": 1.0857842005130671, "learning_rate": 0.003, "loss": 4.0887, "step": 6753 }, { "epoch": 0.06754, "grad_norm": 1.1455822073063449, "learning_rate": 0.003, "loss": 4.1133, "step": 6754 }, { "epoch": 0.06755, "grad_norm": 0.932580990979775, "learning_rate": 0.003, "loss": 4.0986, "step": 6755 }, { "epoch": 0.06756, "grad_norm": 0.945017592655439, "learning_rate": 0.003, "loss": 4.1142, "step": 6756 }, { "epoch": 0.06757, "grad_norm": 1.1087441244810934, "learning_rate": 0.003, "loss": 4.1224, "step": 6757 }, { "epoch": 0.06758, "grad_norm": 1.2032017160137123, "learning_rate": 0.003, "loss": 4.1154, "step": 6758 }, { "epoch": 0.06759, "grad_norm": 0.9725904113161775, "learning_rate": 0.003, "loss": 4.1414, "step": 6759 }, { "epoch": 0.0676, "grad_norm": 0.9994768531593359, "learning_rate": 0.003, "loss": 4.1252, "step": 6760 }, { "epoch": 0.06761, "grad_norm": 1.2158744246583477, "learning_rate": 0.003, "loss": 4.1072, "step": 6761 }, { "epoch": 0.06762, "grad_norm": 0.9948770438066381, "learning_rate": 0.003, "loss": 4.112, "step": 6762 }, { "epoch": 0.06763, "grad_norm": 1.0569961624633628, "learning_rate": 0.003, "loss": 4.1118, "step": 6763 }, { "epoch": 0.06764, "grad_norm": 1.013959466655787, "learning_rate": 0.003, "loss": 4.0962, "step": 6764 }, { "epoch": 0.06765, "grad_norm": 1.1364024067522305, "learning_rate": 0.003, "loss": 4.1118, "step": 6765 }, { "epoch": 0.06766, "grad_norm": 0.8459015884205136, "learning_rate": 0.003, "loss": 4.122, "step": 6766 }, { "epoch": 0.06767, "grad_norm": 0.880794645124822, "learning_rate": 0.003, "loss": 4.1211, "step": 6767 }, { "epoch": 0.06768, "grad_norm": 0.9020257743979438, "learning_rate": 0.003, "loss": 4.0843, "step": 6768 }, { "epoch": 0.06769, "grad_norm": 0.957067840155779, "learning_rate": 0.003, "loss": 4.1016, "step": 6769 }, { "epoch": 0.0677, "grad_norm": 0.9312965378254958, "learning_rate": 0.003, "loss": 4.1155, "step": 6770 }, { "epoch": 0.06771, "grad_norm": 0.9824294156270907, "learning_rate": 0.003, "loss": 4.1037, "step": 6771 }, { "epoch": 0.06772, "grad_norm": 1.1173456344519563, "learning_rate": 0.003, "loss": 4.1186, "step": 6772 }, { "epoch": 0.06773, "grad_norm": 0.9102193256903605, "learning_rate": 0.003, "loss": 4.1257, "step": 6773 }, { "epoch": 0.06774, "grad_norm": 0.9547527714430646, "learning_rate": 0.003, "loss": 4.0639, "step": 6774 }, { "epoch": 0.06775, "grad_norm": 1.0781058147542015, "learning_rate": 0.003, "loss": 4.1137, "step": 6775 }, { "epoch": 0.06776, "grad_norm": 1.6499486306229654, "learning_rate": 0.003, "loss": 4.153, "step": 6776 }, { "epoch": 0.06777, "grad_norm": 0.8340051804173416, "learning_rate": 0.003, "loss": 4.1254, "step": 6777 }, { "epoch": 0.06778, "grad_norm": 0.9281740251373968, "learning_rate": 0.003, "loss": 4.0716, "step": 6778 }, { "epoch": 0.06779, "grad_norm": 1.007311758088431, "learning_rate": 0.003, "loss": 4.1196, "step": 6779 }, { "epoch": 0.0678, "grad_norm": 1.0343368874975791, "learning_rate": 0.003, "loss": 4.0822, "step": 6780 }, { "epoch": 0.06781, "grad_norm": 1.0257339097012195, "learning_rate": 0.003, "loss": 4.1447, "step": 6781 }, { "epoch": 0.06782, "grad_norm": 1.208826935719483, "learning_rate": 0.003, "loss": 4.1296, "step": 6782 }, { "epoch": 0.06783, "grad_norm": 1.1183146143319012, "learning_rate": 0.003, "loss": 4.1384, "step": 6783 }, { "epoch": 0.06784, "grad_norm": 1.0684998687741116, "learning_rate": 0.003, "loss": 4.1055, "step": 6784 }, { "epoch": 0.06785, "grad_norm": 1.095620038881207, "learning_rate": 0.003, "loss": 4.1287, "step": 6785 }, { "epoch": 0.06786, "grad_norm": 1.0056898765693647, "learning_rate": 0.003, "loss": 4.0925, "step": 6786 }, { "epoch": 0.06787, "grad_norm": 1.4297140113997957, "learning_rate": 0.003, "loss": 4.0884, "step": 6787 }, { "epoch": 0.06788, "grad_norm": 0.790153682740165, "learning_rate": 0.003, "loss": 4.1265, "step": 6788 }, { "epoch": 0.06789, "grad_norm": 0.7738912403286867, "learning_rate": 0.003, "loss": 4.0877, "step": 6789 }, { "epoch": 0.0679, "grad_norm": 0.9134316450396154, "learning_rate": 0.003, "loss": 4.1181, "step": 6790 }, { "epoch": 0.06791, "grad_norm": 1.095403623939103, "learning_rate": 0.003, "loss": 4.1094, "step": 6791 }, { "epoch": 0.06792, "grad_norm": 1.0057315719319384, "learning_rate": 0.003, "loss": 4.1271, "step": 6792 }, { "epoch": 0.06793, "grad_norm": 1.0520802836117387, "learning_rate": 0.003, "loss": 4.1233, "step": 6793 }, { "epoch": 0.06794, "grad_norm": 0.930030498565471, "learning_rate": 0.003, "loss": 4.1306, "step": 6794 }, { "epoch": 0.06795, "grad_norm": 1.224689098436433, "learning_rate": 0.003, "loss": 4.0901, "step": 6795 }, { "epoch": 0.06796, "grad_norm": 1.296575828943012, "learning_rate": 0.003, "loss": 4.1244, "step": 6796 }, { "epoch": 0.06797, "grad_norm": 0.9404333531599239, "learning_rate": 0.003, "loss": 4.1372, "step": 6797 }, { "epoch": 0.06798, "grad_norm": 0.9774637017162385, "learning_rate": 0.003, "loss": 4.1046, "step": 6798 }, { "epoch": 0.06799, "grad_norm": 1.3189103861369607, "learning_rate": 0.003, "loss": 4.1445, "step": 6799 }, { "epoch": 0.068, "grad_norm": 0.8760990280464664, "learning_rate": 0.003, "loss": 4.1117, "step": 6800 }, { "epoch": 0.06801, "grad_norm": 0.7762558071690656, "learning_rate": 0.003, "loss": 4.0845, "step": 6801 }, { "epoch": 0.06802, "grad_norm": 0.8561625366506563, "learning_rate": 0.003, "loss": 4.1037, "step": 6802 }, { "epoch": 0.06803, "grad_norm": 0.9139469300515602, "learning_rate": 0.003, "loss": 4.1082, "step": 6803 }, { "epoch": 0.06804, "grad_norm": 1.1348166936740396, "learning_rate": 0.003, "loss": 4.127, "step": 6804 }, { "epoch": 0.06805, "grad_norm": 1.1939376321332151, "learning_rate": 0.003, "loss": 4.1062, "step": 6805 }, { "epoch": 0.06806, "grad_norm": 0.9515480548410903, "learning_rate": 0.003, "loss": 4.0994, "step": 6806 }, { "epoch": 0.06807, "grad_norm": 0.9471722400961542, "learning_rate": 0.003, "loss": 4.1046, "step": 6807 }, { "epoch": 0.06808, "grad_norm": 0.916585569019858, "learning_rate": 0.003, "loss": 4.1089, "step": 6808 }, { "epoch": 0.06809, "grad_norm": 1.2113066481378751, "learning_rate": 0.003, "loss": 4.0877, "step": 6809 }, { "epoch": 0.0681, "grad_norm": 1.340856802998418, "learning_rate": 0.003, "loss": 4.1112, "step": 6810 }, { "epoch": 0.06811, "grad_norm": 0.9885027786684883, "learning_rate": 0.003, "loss": 4.1374, "step": 6811 }, { "epoch": 0.06812, "grad_norm": 1.0267551949527556, "learning_rate": 0.003, "loss": 4.103, "step": 6812 }, { "epoch": 0.06813, "grad_norm": 1.040795196495314, "learning_rate": 0.003, "loss": 4.0821, "step": 6813 }, { "epoch": 0.06814, "grad_norm": 1.11293418398358, "learning_rate": 0.003, "loss": 4.1208, "step": 6814 }, { "epoch": 0.06815, "grad_norm": 1.1607149753899817, "learning_rate": 0.003, "loss": 4.1291, "step": 6815 }, { "epoch": 0.06816, "grad_norm": 0.8713926376226725, "learning_rate": 0.003, "loss": 4.1231, "step": 6816 }, { "epoch": 0.06817, "grad_norm": 0.9081597850780296, "learning_rate": 0.003, "loss": 4.0952, "step": 6817 }, { "epoch": 0.06818, "grad_norm": 0.9942673729666544, "learning_rate": 0.003, "loss": 4.1193, "step": 6818 }, { "epoch": 0.06819, "grad_norm": 1.0537961386880241, "learning_rate": 0.003, "loss": 4.1111, "step": 6819 }, { "epoch": 0.0682, "grad_norm": 1.0451146888171359, "learning_rate": 0.003, "loss": 4.1416, "step": 6820 }, { "epoch": 0.06821, "grad_norm": 1.1139335819462208, "learning_rate": 0.003, "loss": 4.1307, "step": 6821 }, { "epoch": 0.06822, "grad_norm": 1.054644343046522, "learning_rate": 0.003, "loss": 4.0913, "step": 6822 }, { "epoch": 0.06823, "grad_norm": 1.0144607819110447, "learning_rate": 0.003, "loss": 4.1084, "step": 6823 }, { "epoch": 0.06824, "grad_norm": 0.9958932547970886, "learning_rate": 0.003, "loss": 4.0837, "step": 6824 }, { "epoch": 0.06825, "grad_norm": 0.8766930589389056, "learning_rate": 0.003, "loss": 4.0977, "step": 6825 }, { "epoch": 0.06826, "grad_norm": 0.82277829993784, "learning_rate": 0.003, "loss": 4.0917, "step": 6826 }, { "epoch": 0.06827, "grad_norm": 0.9752386388524714, "learning_rate": 0.003, "loss": 4.0847, "step": 6827 }, { "epoch": 0.06828, "grad_norm": 1.205573894590776, "learning_rate": 0.003, "loss": 4.113, "step": 6828 }, { "epoch": 0.06829, "grad_norm": 1.133172664265929, "learning_rate": 0.003, "loss": 4.1181, "step": 6829 }, { "epoch": 0.0683, "grad_norm": 0.8238913017656965, "learning_rate": 0.003, "loss": 4.1025, "step": 6830 }, { "epoch": 0.06831, "grad_norm": 0.8316896827214876, "learning_rate": 0.003, "loss": 4.1253, "step": 6831 }, { "epoch": 0.06832, "grad_norm": 0.9269287661184776, "learning_rate": 0.003, "loss": 4.0941, "step": 6832 }, { "epoch": 0.06833, "grad_norm": 0.9989134257290364, "learning_rate": 0.003, "loss": 4.0809, "step": 6833 }, { "epoch": 0.06834, "grad_norm": 1.1107856107305376, "learning_rate": 0.003, "loss": 4.1194, "step": 6834 }, { "epoch": 0.06835, "grad_norm": 1.0125690092416209, "learning_rate": 0.003, "loss": 4.1158, "step": 6835 }, { "epoch": 0.06836, "grad_norm": 1.2778620275741064, "learning_rate": 0.003, "loss": 4.1469, "step": 6836 }, { "epoch": 0.06837, "grad_norm": 0.993797792799617, "learning_rate": 0.003, "loss": 4.0932, "step": 6837 }, { "epoch": 0.06838, "grad_norm": 1.385862043374062, "learning_rate": 0.003, "loss": 4.1107, "step": 6838 }, { "epoch": 0.06839, "grad_norm": 0.8693787226168019, "learning_rate": 0.003, "loss": 4.0932, "step": 6839 }, { "epoch": 0.0684, "grad_norm": 0.8191901365828969, "learning_rate": 0.003, "loss": 4.1143, "step": 6840 }, { "epoch": 0.06841, "grad_norm": 0.8891671375692035, "learning_rate": 0.003, "loss": 4.1116, "step": 6841 }, { "epoch": 0.06842, "grad_norm": 1.0307688402129975, "learning_rate": 0.003, "loss": 4.1135, "step": 6842 }, { "epoch": 0.06843, "grad_norm": 1.0868160126053539, "learning_rate": 0.003, "loss": 4.1097, "step": 6843 }, { "epoch": 0.06844, "grad_norm": 1.1395915635803298, "learning_rate": 0.003, "loss": 4.1083, "step": 6844 }, { "epoch": 0.06845, "grad_norm": 1.2053792036841005, "learning_rate": 0.003, "loss": 4.1278, "step": 6845 }, { "epoch": 0.06846, "grad_norm": 0.9394084816348595, "learning_rate": 0.003, "loss": 4.1085, "step": 6846 }, { "epoch": 0.06847, "grad_norm": 0.9659910146473426, "learning_rate": 0.003, "loss": 4.0975, "step": 6847 }, { "epoch": 0.06848, "grad_norm": 1.3652228201962051, "learning_rate": 0.003, "loss": 4.1206, "step": 6848 }, { "epoch": 0.06849, "grad_norm": 0.8944786288593367, "learning_rate": 0.003, "loss": 4.098, "step": 6849 }, { "epoch": 0.0685, "grad_norm": 0.8468146869585674, "learning_rate": 0.003, "loss": 4.0992, "step": 6850 }, { "epoch": 0.06851, "grad_norm": 0.9087271188351737, "learning_rate": 0.003, "loss": 4.0811, "step": 6851 }, { "epoch": 0.06852, "grad_norm": 1.0334027372289312, "learning_rate": 0.003, "loss": 4.0928, "step": 6852 }, { "epoch": 0.06853, "grad_norm": 0.902983888059127, "learning_rate": 0.003, "loss": 4.0857, "step": 6853 }, { "epoch": 0.06854, "grad_norm": 1.1605358694940422, "learning_rate": 0.003, "loss": 4.1266, "step": 6854 }, { "epoch": 0.06855, "grad_norm": 1.2624753429091504, "learning_rate": 0.003, "loss": 4.1411, "step": 6855 }, { "epoch": 0.06856, "grad_norm": 1.118915168359842, "learning_rate": 0.003, "loss": 4.1344, "step": 6856 }, { "epoch": 0.06857, "grad_norm": 0.967318710251076, "learning_rate": 0.003, "loss": 4.1154, "step": 6857 }, { "epoch": 0.06858, "grad_norm": 0.8993491747442931, "learning_rate": 0.003, "loss": 4.0747, "step": 6858 }, { "epoch": 0.06859, "grad_norm": 0.8640520875840221, "learning_rate": 0.003, "loss": 4.0856, "step": 6859 }, { "epoch": 0.0686, "grad_norm": 1.078288558641258, "learning_rate": 0.003, "loss": 4.0972, "step": 6860 }, { "epoch": 0.06861, "grad_norm": 1.179816478483578, "learning_rate": 0.003, "loss": 4.0935, "step": 6861 }, { "epoch": 0.06862, "grad_norm": 0.9056185959377018, "learning_rate": 0.003, "loss": 4.0868, "step": 6862 }, { "epoch": 0.06863, "grad_norm": 0.9165282656767371, "learning_rate": 0.003, "loss": 4.0952, "step": 6863 }, { "epoch": 0.06864, "grad_norm": 1.1201407060467088, "learning_rate": 0.003, "loss": 4.1103, "step": 6864 }, { "epoch": 0.06865, "grad_norm": 1.0179258305510663, "learning_rate": 0.003, "loss": 4.0916, "step": 6865 }, { "epoch": 0.06866, "grad_norm": 1.2117982840621284, "learning_rate": 0.003, "loss": 4.1195, "step": 6866 }, { "epoch": 0.06867, "grad_norm": 1.145110761860685, "learning_rate": 0.003, "loss": 4.1314, "step": 6867 }, { "epoch": 0.06868, "grad_norm": 0.8964379313752789, "learning_rate": 0.003, "loss": 4.0861, "step": 6868 }, { "epoch": 0.06869, "grad_norm": 0.8278694268577412, "learning_rate": 0.003, "loss": 4.1133, "step": 6869 }, { "epoch": 0.0687, "grad_norm": 0.9210440687808206, "learning_rate": 0.003, "loss": 4.0543, "step": 6870 }, { "epoch": 0.06871, "grad_norm": 1.2931682466657848, "learning_rate": 0.003, "loss": 4.0892, "step": 6871 }, { "epoch": 0.06872, "grad_norm": 1.158244309207232, "learning_rate": 0.003, "loss": 4.0892, "step": 6872 }, { "epoch": 0.06873, "grad_norm": 1.0509391434511997, "learning_rate": 0.003, "loss": 4.0957, "step": 6873 }, { "epoch": 0.06874, "grad_norm": 1.0160683857215573, "learning_rate": 0.003, "loss": 4.1117, "step": 6874 }, { "epoch": 0.06875, "grad_norm": 1.070894253002053, "learning_rate": 0.003, "loss": 4.1183, "step": 6875 }, { "epoch": 0.06876, "grad_norm": 1.0559885742243682, "learning_rate": 0.003, "loss": 4.1237, "step": 6876 }, { "epoch": 0.06877, "grad_norm": 0.9568759782863472, "learning_rate": 0.003, "loss": 4.0863, "step": 6877 }, { "epoch": 0.06878, "grad_norm": 1.1647575826059193, "learning_rate": 0.003, "loss": 4.0888, "step": 6878 }, { "epoch": 0.06879, "grad_norm": 1.0530843336006634, "learning_rate": 0.003, "loss": 4.0941, "step": 6879 }, { "epoch": 0.0688, "grad_norm": 1.0399591258677314, "learning_rate": 0.003, "loss": 4.1056, "step": 6880 }, { "epoch": 0.06881, "grad_norm": 0.989276355858342, "learning_rate": 0.003, "loss": 4.0919, "step": 6881 }, { "epoch": 0.06882, "grad_norm": 1.2920882835812373, "learning_rate": 0.003, "loss": 4.1033, "step": 6882 }, { "epoch": 0.06883, "grad_norm": 1.1043413882784838, "learning_rate": 0.003, "loss": 4.1245, "step": 6883 }, { "epoch": 0.06884, "grad_norm": 1.0613281788454871, "learning_rate": 0.003, "loss": 4.0873, "step": 6884 }, { "epoch": 0.06885, "grad_norm": 0.96781655650821, "learning_rate": 0.003, "loss": 4.1078, "step": 6885 }, { "epoch": 0.06886, "grad_norm": 0.9517962324408646, "learning_rate": 0.003, "loss": 4.0904, "step": 6886 }, { "epoch": 0.06887, "grad_norm": 0.9638770692400819, "learning_rate": 0.003, "loss": 4.1159, "step": 6887 }, { "epoch": 0.06888, "grad_norm": 1.0795473222518885, "learning_rate": 0.003, "loss": 4.0938, "step": 6888 }, { "epoch": 0.06889, "grad_norm": 0.959615080421356, "learning_rate": 0.003, "loss": 4.1061, "step": 6889 }, { "epoch": 0.0689, "grad_norm": 1.000200309338473, "learning_rate": 0.003, "loss": 4.0862, "step": 6890 }, { "epoch": 0.06891, "grad_norm": 1.107853464297452, "learning_rate": 0.003, "loss": 4.1259, "step": 6891 }, { "epoch": 0.06892, "grad_norm": 1.09996425163966, "learning_rate": 0.003, "loss": 4.1043, "step": 6892 }, { "epoch": 0.06893, "grad_norm": 1.1680931931384533, "learning_rate": 0.003, "loss": 4.1144, "step": 6893 }, { "epoch": 0.06894, "grad_norm": 1.105626112595867, "learning_rate": 0.003, "loss": 4.0994, "step": 6894 }, { "epoch": 0.06895, "grad_norm": 0.8725390460274325, "learning_rate": 0.003, "loss": 4.091, "step": 6895 }, { "epoch": 0.06896, "grad_norm": 1.0781792472155975, "learning_rate": 0.003, "loss": 4.0892, "step": 6896 }, { "epoch": 0.06897, "grad_norm": 1.411658300042236, "learning_rate": 0.003, "loss": 4.1405, "step": 6897 }, { "epoch": 0.06898, "grad_norm": 0.9732387067386347, "learning_rate": 0.003, "loss": 4.0577, "step": 6898 }, { "epoch": 0.06899, "grad_norm": 1.0275999231438366, "learning_rate": 0.003, "loss": 4.1369, "step": 6899 }, { "epoch": 0.069, "grad_norm": 1.0294308999954391, "learning_rate": 0.003, "loss": 4.1034, "step": 6900 }, { "epoch": 0.06901, "grad_norm": 1.1909528295828076, "learning_rate": 0.003, "loss": 4.0854, "step": 6901 }, { "epoch": 0.06902, "grad_norm": 1.014749745517845, "learning_rate": 0.003, "loss": 4.0929, "step": 6902 }, { "epoch": 0.06903, "grad_norm": 1.0651050361740424, "learning_rate": 0.003, "loss": 4.1034, "step": 6903 }, { "epoch": 0.06904, "grad_norm": 1.1039173799433109, "learning_rate": 0.003, "loss": 4.1039, "step": 6904 }, { "epoch": 0.06905, "grad_norm": 0.9239962540742471, "learning_rate": 0.003, "loss": 4.1288, "step": 6905 }, { "epoch": 0.06906, "grad_norm": 0.8654819902503579, "learning_rate": 0.003, "loss": 4.0753, "step": 6906 }, { "epoch": 0.06907, "grad_norm": 1.1296324777588087, "learning_rate": 0.003, "loss": 4.0978, "step": 6907 }, { "epoch": 0.06908, "grad_norm": 1.3859704456042667, "learning_rate": 0.003, "loss": 4.0819, "step": 6908 }, { "epoch": 0.06909, "grad_norm": 1.0521744111723947, "learning_rate": 0.003, "loss": 4.1136, "step": 6909 }, { "epoch": 0.0691, "grad_norm": 1.0884172288485674, "learning_rate": 0.003, "loss": 4.0828, "step": 6910 }, { "epoch": 0.06911, "grad_norm": 0.9789284124718082, "learning_rate": 0.003, "loss": 4.1245, "step": 6911 }, { "epoch": 0.06912, "grad_norm": 0.9437089236297371, "learning_rate": 0.003, "loss": 4.1002, "step": 6912 }, { "epoch": 0.06913, "grad_norm": 0.8815007693721976, "learning_rate": 0.003, "loss": 4.0766, "step": 6913 }, { "epoch": 0.06914, "grad_norm": 0.9983756175361831, "learning_rate": 0.003, "loss": 4.1261, "step": 6914 }, { "epoch": 0.06915, "grad_norm": 1.233420747523598, "learning_rate": 0.003, "loss": 4.1369, "step": 6915 }, { "epoch": 0.06916, "grad_norm": 1.0634074261162114, "learning_rate": 0.003, "loss": 4.103, "step": 6916 }, { "epoch": 0.06917, "grad_norm": 1.0640603780382807, "learning_rate": 0.003, "loss": 4.0882, "step": 6917 }, { "epoch": 0.06918, "grad_norm": 1.071505216595764, "learning_rate": 0.003, "loss": 4.128, "step": 6918 }, { "epoch": 0.06919, "grad_norm": 0.9525463685819003, "learning_rate": 0.003, "loss": 4.121, "step": 6919 }, { "epoch": 0.0692, "grad_norm": 1.0292478999895873, "learning_rate": 0.003, "loss": 4.1131, "step": 6920 }, { "epoch": 0.06921, "grad_norm": 1.0529101386571624, "learning_rate": 0.003, "loss": 4.1222, "step": 6921 }, { "epoch": 0.06922, "grad_norm": 1.0651675683551554, "learning_rate": 0.003, "loss": 4.116, "step": 6922 }, { "epoch": 0.06923, "grad_norm": 1.1004871894989456, "learning_rate": 0.003, "loss": 4.0992, "step": 6923 }, { "epoch": 0.06924, "grad_norm": 1.2691682406714033, "learning_rate": 0.003, "loss": 4.1405, "step": 6924 }, { "epoch": 0.06925, "grad_norm": 1.0417482481029872, "learning_rate": 0.003, "loss": 4.1309, "step": 6925 }, { "epoch": 0.06926, "grad_norm": 1.1830131179045373, "learning_rate": 0.003, "loss": 4.125, "step": 6926 }, { "epoch": 0.06927, "grad_norm": 1.0664599500199052, "learning_rate": 0.003, "loss": 4.1196, "step": 6927 }, { "epoch": 0.06928, "grad_norm": 1.017895846116543, "learning_rate": 0.003, "loss": 4.0938, "step": 6928 }, { "epoch": 0.06929, "grad_norm": 1.2322119521200885, "learning_rate": 0.003, "loss": 4.1106, "step": 6929 }, { "epoch": 0.0693, "grad_norm": 0.9444007543363695, "learning_rate": 0.003, "loss": 4.1498, "step": 6930 }, { "epoch": 0.06931, "grad_norm": 1.2022274249029867, "learning_rate": 0.003, "loss": 4.1224, "step": 6931 }, { "epoch": 0.06932, "grad_norm": 1.148339118744329, "learning_rate": 0.003, "loss": 4.0959, "step": 6932 }, { "epoch": 0.06933, "grad_norm": 1.0316199156026564, "learning_rate": 0.003, "loss": 4.132, "step": 6933 }, { "epoch": 0.06934, "grad_norm": 0.9521700971631617, "learning_rate": 0.003, "loss": 4.0953, "step": 6934 }, { "epoch": 0.06935, "grad_norm": 1.0650733326681008, "learning_rate": 0.003, "loss": 4.119, "step": 6935 }, { "epoch": 0.06936, "grad_norm": 1.0622590447464284, "learning_rate": 0.003, "loss": 4.0721, "step": 6936 }, { "epoch": 0.06937, "grad_norm": 1.0213424894082164, "learning_rate": 0.003, "loss": 4.102, "step": 6937 }, { "epoch": 0.06938, "grad_norm": 1.0634441753043125, "learning_rate": 0.003, "loss": 4.0768, "step": 6938 }, { "epoch": 0.06939, "grad_norm": 0.9122903874612669, "learning_rate": 0.003, "loss": 4.1059, "step": 6939 }, { "epoch": 0.0694, "grad_norm": 1.1312835792631537, "learning_rate": 0.003, "loss": 4.0953, "step": 6940 }, { "epoch": 0.06941, "grad_norm": 1.0379316048960034, "learning_rate": 0.003, "loss": 4.1154, "step": 6941 }, { "epoch": 0.06942, "grad_norm": 0.9970308628077427, "learning_rate": 0.003, "loss": 4.0914, "step": 6942 }, { "epoch": 0.06943, "grad_norm": 0.9814957047921263, "learning_rate": 0.003, "loss": 4.1225, "step": 6943 }, { "epoch": 0.06944, "grad_norm": 1.059683349285963, "learning_rate": 0.003, "loss": 4.1032, "step": 6944 }, { "epoch": 0.06945, "grad_norm": 1.0340425314303883, "learning_rate": 0.003, "loss": 4.1328, "step": 6945 }, { "epoch": 0.06946, "grad_norm": 1.0976200591310759, "learning_rate": 0.003, "loss": 4.1118, "step": 6946 }, { "epoch": 0.06947, "grad_norm": 1.1069722316961914, "learning_rate": 0.003, "loss": 4.1001, "step": 6947 }, { "epoch": 0.06948, "grad_norm": 0.9412592759167051, "learning_rate": 0.003, "loss": 4.1094, "step": 6948 }, { "epoch": 0.06949, "grad_norm": 0.9635427895046388, "learning_rate": 0.003, "loss": 4.1162, "step": 6949 }, { "epoch": 0.0695, "grad_norm": 1.1696562924083407, "learning_rate": 0.003, "loss": 4.0979, "step": 6950 }, { "epoch": 0.06951, "grad_norm": 0.9122948415660266, "learning_rate": 0.003, "loss": 4.1236, "step": 6951 }, { "epoch": 0.06952, "grad_norm": 1.0086842274617382, "learning_rate": 0.003, "loss": 4.1305, "step": 6952 }, { "epoch": 0.06953, "grad_norm": 0.9788527064917946, "learning_rate": 0.003, "loss": 4.1258, "step": 6953 }, { "epoch": 0.06954, "grad_norm": 1.2559346458301088, "learning_rate": 0.003, "loss": 4.0773, "step": 6954 }, { "epoch": 0.06955, "grad_norm": 0.8368392604009263, "learning_rate": 0.003, "loss": 4.1167, "step": 6955 }, { "epoch": 0.06956, "grad_norm": 0.9878198015577839, "learning_rate": 0.003, "loss": 4.1196, "step": 6956 }, { "epoch": 0.06957, "grad_norm": 1.1449000449568048, "learning_rate": 0.003, "loss": 4.0993, "step": 6957 }, { "epoch": 0.06958, "grad_norm": 1.1872333547049227, "learning_rate": 0.003, "loss": 4.1218, "step": 6958 }, { "epoch": 0.06959, "grad_norm": 1.064048588165346, "learning_rate": 0.003, "loss": 4.1114, "step": 6959 }, { "epoch": 0.0696, "grad_norm": 1.039713959773937, "learning_rate": 0.003, "loss": 4.1212, "step": 6960 }, { "epoch": 0.06961, "grad_norm": 1.0996296813655457, "learning_rate": 0.003, "loss": 4.1247, "step": 6961 }, { "epoch": 0.06962, "grad_norm": 0.9879371143151359, "learning_rate": 0.003, "loss": 4.1133, "step": 6962 }, { "epoch": 0.06963, "grad_norm": 1.0549404515531287, "learning_rate": 0.003, "loss": 4.1148, "step": 6963 }, { "epoch": 0.06964, "grad_norm": 1.1314314837164443, "learning_rate": 0.003, "loss": 4.114, "step": 6964 }, { "epoch": 0.06965, "grad_norm": 1.1674676649395428, "learning_rate": 0.003, "loss": 4.0956, "step": 6965 }, { "epoch": 0.06966, "grad_norm": 1.0504852564702463, "learning_rate": 0.003, "loss": 4.1199, "step": 6966 }, { "epoch": 0.06967, "grad_norm": 1.150124143433524, "learning_rate": 0.003, "loss": 4.1041, "step": 6967 }, { "epoch": 0.06968, "grad_norm": 1.0016009335375253, "learning_rate": 0.003, "loss": 4.1073, "step": 6968 }, { "epoch": 0.06969, "grad_norm": 1.2023901390365654, "learning_rate": 0.003, "loss": 4.1282, "step": 6969 }, { "epoch": 0.0697, "grad_norm": 0.8664049863091234, "learning_rate": 0.003, "loss": 4.089, "step": 6970 }, { "epoch": 0.06971, "grad_norm": 1.0124884250644552, "learning_rate": 0.003, "loss": 4.102, "step": 6971 }, { "epoch": 0.06972, "grad_norm": 1.2760143566845368, "learning_rate": 0.003, "loss": 4.0753, "step": 6972 }, { "epoch": 0.06973, "grad_norm": 1.0138958170391523, "learning_rate": 0.003, "loss": 4.1126, "step": 6973 }, { "epoch": 0.06974, "grad_norm": 1.1582433174589497, "learning_rate": 0.003, "loss": 4.1065, "step": 6974 }, { "epoch": 0.06975, "grad_norm": 1.1427796390190856, "learning_rate": 0.003, "loss": 4.1151, "step": 6975 }, { "epoch": 0.06976, "grad_norm": 0.9310217784301746, "learning_rate": 0.003, "loss": 4.0714, "step": 6976 }, { "epoch": 0.06977, "grad_norm": 0.9058473639895622, "learning_rate": 0.003, "loss": 4.1424, "step": 6977 }, { "epoch": 0.06978, "grad_norm": 1.0711966126178156, "learning_rate": 0.003, "loss": 4.0771, "step": 6978 }, { "epoch": 0.06979, "grad_norm": 1.0970262192518259, "learning_rate": 0.003, "loss": 4.1281, "step": 6979 }, { "epoch": 0.0698, "grad_norm": 1.0412974314007526, "learning_rate": 0.003, "loss": 4.112, "step": 6980 }, { "epoch": 0.06981, "grad_norm": 0.9795157568742333, "learning_rate": 0.003, "loss": 4.0872, "step": 6981 }, { "epoch": 0.06982, "grad_norm": 1.0944682868487863, "learning_rate": 0.003, "loss": 4.0993, "step": 6982 }, { "epoch": 0.06983, "grad_norm": 0.9694518653743476, "learning_rate": 0.003, "loss": 4.0821, "step": 6983 }, { "epoch": 0.06984, "grad_norm": 1.2844571684315234, "learning_rate": 0.003, "loss": 4.0917, "step": 6984 }, { "epoch": 0.06985, "grad_norm": 1.0332117175116347, "learning_rate": 0.003, "loss": 4.1221, "step": 6985 }, { "epoch": 0.06986, "grad_norm": 1.0819077584971564, "learning_rate": 0.003, "loss": 4.1161, "step": 6986 }, { "epoch": 0.06987, "grad_norm": 0.9731206337011348, "learning_rate": 0.003, "loss": 4.1154, "step": 6987 }, { "epoch": 0.06988, "grad_norm": 1.0718738925220694, "learning_rate": 0.003, "loss": 4.0977, "step": 6988 }, { "epoch": 0.06989, "grad_norm": 1.054505002503669, "learning_rate": 0.003, "loss": 4.0864, "step": 6989 }, { "epoch": 0.0699, "grad_norm": 1.049446778506991, "learning_rate": 0.003, "loss": 4.084, "step": 6990 }, { "epoch": 0.06991, "grad_norm": 1.0281964154708396, "learning_rate": 0.003, "loss": 4.1155, "step": 6991 }, { "epoch": 0.06992, "grad_norm": 1.1546779439763237, "learning_rate": 0.003, "loss": 4.0872, "step": 6992 }, { "epoch": 0.06993, "grad_norm": 0.9314030046479128, "learning_rate": 0.003, "loss": 4.0835, "step": 6993 }, { "epoch": 0.06994, "grad_norm": 0.9562143137910827, "learning_rate": 0.003, "loss": 4.0895, "step": 6994 }, { "epoch": 0.06995, "grad_norm": 1.1679910334460406, "learning_rate": 0.003, "loss": 4.1243, "step": 6995 }, { "epoch": 0.06996, "grad_norm": 0.9969645718957871, "learning_rate": 0.003, "loss": 4.1048, "step": 6996 }, { "epoch": 0.06997, "grad_norm": 1.1879263627500822, "learning_rate": 0.003, "loss": 4.0845, "step": 6997 }, { "epoch": 0.06998, "grad_norm": 1.0073414720644633, "learning_rate": 0.003, "loss": 4.104, "step": 6998 }, { "epoch": 0.06999, "grad_norm": 1.044274131329711, "learning_rate": 0.003, "loss": 4.0923, "step": 6999 }, { "epoch": 0.07, "grad_norm": 1.0064889519673954, "learning_rate": 0.003, "loss": 4.0958, "step": 7000 }, { "epoch": 0.07001, "grad_norm": 0.9375889716051797, "learning_rate": 0.003, "loss": 4.149, "step": 7001 }, { "epoch": 0.07002, "grad_norm": 0.9684683452814823, "learning_rate": 0.003, "loss": 4.0991, "step": 7002 }, { "epoch": 0.07003, "grad_norm": 1.1198927333018025, "learning_rate": 0.003, "loss": 4.0942, "step": 7003 }, { "epoch": 0.07004, "grad_norm": 0.8203608521817181, "learning_rate": 0.003, "loss": 4.0737, "step": 7004 }, { "epoch": 0.07005, "grad_norm": 0.9540874717299233, "learning_rate": 0.003, "loss": 4.1036, "step": 7005 }, { "epoch": 0.07006, "grad_norm": 0.9464816735202635, "learning_rate": 0.003, "loss": 4.1034, "step": 7006 }, { "epoch": 0.07007, "grad_norm": 0.941064497178461, "learning_rate": 0.003, "loss": 4.1035, "step": 7007 }, { "epoch": 0.07008, "grad_norm": 1.1484153762506384, "learning_rate": 0.003, "loss": 4.1057, "step": 7008 }, { "epoch": 0.07009, "grad_norm": 1.425852706227398, "learning_rate": 0.003, "loss": 4.1266, "step": 7009 }, { "epoch": 0.0701, "grad_norm": 1.1008237282072466, "learning_rate": 0.003, "loss": 4.0978, "step": 7010 }, { "epoch": 0.07011, "grad_norm": 1.030644753513, "learning_rate": 0.003, "loss": 4.0919, "step": 7011 }, { "epoch": 0.07012, "grad_norm": 1.0687016058276446, "learning_rate": 0.003, "loss": 4.1248, "step": 7012 }, { "epoch": 0.07013, "grad_norm": 1.1341288027770202, "learning_rate": 0.003, "loss": 4.0972, "step": 7013 }, { "epoch": 0.07014, "grad_norm": 1.0040444860045297, "learning_rate": 0.003, "loss": 4.0835, "step": 7014 }, { "epoch": 0.07015, "grad_norm": 0.9932353648755569, "learning_rate": 0.003, "loss": 4.0913, "step": 7015 }, { "epoch": 0.07016, "grad_norm": 1.2880395082608038, "learning_rate": 0.003, "loss": 4.1011, "step": 7016 }, { "epoch": 0.07017, "grad_norm": 0.7682702654491351, "learning_rate": 0.003, "loss": 4.0781, "step": 7017 }, { "epoch": 0.07018, "grad_norm": 0.810727853810661, "learning_rate": 0.003, "loss": 4.1085, "step": 7018 }, { "epoch": 0.07019, "grad_norm": 0.965846833420985, "learning_rate": 0.003, "loss": 4.0992, "step": 7019 }, { "epoch": 0.0702, "grad_norm": 1.4179828887882526, "learning_rate": 0.003, "loss": 4.0998, "step": 7020 }, { "epoch": 0.07021, "grad_norm": 1.0180276871047684, "learning_rate": 0.003, "loss": 4.1105, "step": 7021 }, { "epoch": 0.07022, "grad_norm": 1.1160919397810931, "learning_rate": 0.003, "loss": 4.1165, "step": 7022 }, { "epoch": 0.07023, "grad_norm": 0.9242970527362745, "learning_rate": 0.003, "loss": 4.1063, "step": 7023 }, { "epoch": 0.07024, "grad_norm": 0.9246329587526372, "learning_rate": 0.003, "loss": 4.0831, "step": 7024 }, { "epoch": 0.07025, "grad_norm": 1.0838078965811182, "learning_rate": 0.003, "loss": 4.1151, "step": 7025 }, { "epoch": 0.07026, "grad_norm": 1.1206485645739708, "learning_rate": 0.003, "loss": 4.0909, "step": 7026 }, { "epoch": 0.07027, "grad_norm": 1.136885864091667, "learning_rate": 0.003, "loss": 4.1011, "step": 7027 }, { "epoch": 0.07028, "grad_norm": 1.1013879817070413, "learning_rate": 0.003, "loss": 4.1, "step": 7028 }, { "epoch": 0.07029, "grad_norm": 1.0906829990632632, "learning_rate": 0.003, "loss": 4.0916, "step": 7029 }, { "epoch": 0.0703, "grad_norm": 1.180643431396029, "learning_rate": 0.003, "loss": 4.1249, "step": 7030 }, { "epoch": 0.07031, "grad_norm": 1.1456048218509032, "learning_rate": 0.003, "loss": 4.0794, "step": 7031 }, { "epoch": 0.07032, "grad_norm": 1.154341685812463, "learning_rate": 0.003, "loss": 4.1141, "step": 7032 }, { "epoch": 0.07033, "grad_norm": 0.7739985174184032, "learning_rate": 0.003, "loss": 4.1052, "step": 7033 }, { "epoch": 0.07034, "grad_norm": 0.6448433625987843, "learning_rate": 0.003, "loss": 4.0928, "step": 7034 }, { "epoch": 0.07035, "grad_norm": 0.904325986660508, "learning_rate": 0.003, "loss": 4.0981, "step": 7035 }, { "epoch": 0.07036, "grad_norm": 1.2393718269495677, "learning_rate": 0.003, "loss": 4.0935, "step": 7036 }, { "epoch": 0.07037, "grad_norm": 1.277593084466098, "learning_rate": 0.003, "loss": 4.0877, "step": 7037 }, { "epoch": 0.07038, "grad_norm": 0.8061574634404284, "learning_rate": 0.003, "loss": 4.0977, "step": 7038 }, { "epoch": 0.07039, "grad_norm": 0.8656267328817133, "learning_rate": 0.003, "loss": 4.0944, "step": 7039 }, { "epoch": 0.0704, "grad_norm": 1.0016475050937732, "learning_rate": 0.003, "loss": 4.1218, "step": 7040 }, { "epoch": 0.07041, "grad_norm": 1.1456059826411173, "learning_rate": 0.003, "loss": 4.0994, "step": 7041 }, { "epoch": 0.07042, "grad_norm": 1.0851732002617531, "learning_rate": 0.003, "loss": 4.0842, "step": 7042 }, { "epoch": 0.07043, "grad_norm": 1.0945008161058172, "learning_rate": 0.003, "loss": 4.1175, "step": 7043 }, { "epoch": 0.07044, "grad_norm": 1.0110807128605013, "learning_rate": 0.003, "loss": 4.12, "step": 7044 }, { "epoch": 0.07045, "grad_norm": 1.1320858898966155, "learning_rate": 0.003, "loss": 4.1092, "step": 7045 }, { "epoch": 0.07046, "grad_norm": 0.9871821231222185, "learning_rate": 0.003, "loss": 4.1144, "step": 7046 }, { "epoch": 0.07047, "grad_norm": 1.165418907729337, "learning_rate": 0.003, "loss": 4.1141, "step": 7047 }, { "epoch": 0.07048, "grad_norm": 1.0948822008051076, "learning_rate": 0.003, "loss": 4.1037, "step": 7048 }, { "epoch": 0.07049, "grad_norm": 1.2705996592970472, "learning_rate": 0.003, "loss": 4.1076, "step": 7049 }, { "epoch": 0.0705, "grad_norm": 0.8758701708456724, "learning_rate": 0.003, "loss": 4.1182, "step": 7050 }, { "epoch": 0.07051, "grad_norm": 0.8750281089164476, "learning_rate": 0.003, "loss": 4.1225, "step": 7051 }, { "epoch": 0.07052, "grad_norm": 1.0447997235101967, "learning_rate": 0.003, "loss": 4.0852, "step": 7052 }, { "epoch": 0.07053, "grad_norm": 1.1098900162205543, "learning_rate": 0.003, "loss": 4.0971, "step": 7053 }, { "epoch": 0.07054, "grad_norm": 0.9031483797124182, "learning_rate": 0.003, "loss": 4.0934, "step": 7054 }, { "epoch": 0.07055, "grad_norm": 0.8793147098000734, "learning_rate": 0.003, "loss": 4.0986, "step": 7055 }, { "epoch": 0.07056, "grad_norm": 1.1095062750777271, "learning_rate": 0.003, "loss": 4.0934, "step": 7056 }, { "epoch": 0.07057, "grad_norm": 1.1618977353387767, "learning_rate": 0.003, "loss": 4.1166, "step": 7057 }, { "epoch": 0.07058, "grad_norm": 1.0527234862708261, "learning_rate": 0.003, "loss": 4.0961, "step": 7058 }, { "epoch": 0.07059, "grad_norm": 1.1568117066507144, "learning_rate": 0.003, "loss": 4.1073, "step": 7059 }, { "epoch": 0.0706, "grad_norm": 1.1415615066464118, "learning_rate": 0.003, "loss": 4.1061, "step": 7060 }, { "epoch": 0.07061, "grad_norm": 1.055923444343412, "learning_rate": 0.003, "loss": 4.1005, "step": 7061 }, { "epoch": 0.07062, "grad_norm": 1.0170765646038162, "learning_rate": 0.003, "loss": 4.1346, "step": 7062 }, { "epoch": 0.07063, "grad_norm": 1.1842698666417368, "learning_rate": 0.003, "loss": 4.1148, "step": 7063 }, { "epoch": 0.07064, "grad_norm": 1.2178372221711813, "learning_rate": 0.003, "loss": 4.1252, "step": 7064 }, { "epoch": 0.07065, "grad_norm": 0.9601391763214442, "learning_rate": 0.003, "loss": 4.1117, "step": 7065 }, { "epoch": 0.07066, "grad_norm": 0.916815953425334, "learning_rate": 0.003, "loss": 4.1129, "step": 7066 }, { "epoch": 0.07067, "grad_norm": 1.186217043377779, "learning_rate": 0.003, "loss": 4.1507, "step": 7067 }, { "epoch": 0.07068, "grad_norm": 0.9796528568320318, "learning_rate": 0.003, "loss": 4.0807, "step": 7068 }, { "epoch": 0.07069, "grad_norm": 0.8983177972252203, "learning_rate": 0.003, "loss": 4.0833, "step": 7069 }, { "epoch": 0.0707, "grad_norm": 0.7769178250315103, "learning_rate": 0.003, "loss": 4.106, "step": 7070 }, { "epoch": 0.07071, "grad_norm": 0.906617130093332, "learning_rate": 0.003, "loss": 4.0857, "step": 7071 }, { "epoch": 0.07072, "grad_norm": 1.1358405076574007, "learning_rate": 0.003, "loss": 4.0978, "step": 7072 }, { "epoch": 0.07073, "grad_norm": 0.9412288770953748, "learning_rate": 0.003, "loss": 4.106, "step": 7073 }, { "epoch": 0.07074, "grad_norm": 1.0725538617168138, "learning_rate": 0.003, "loss": 4.094, "step": 7074 }, { "epoch": 0.07075, "grad_norm": 1.1672154966038255, "learning_rate": 0.003, "loss": 4.1384, "step": 7075 }, { "epoch": 0.07076, "grad_norm": 0.7062151739229583, "learning_rate": 0.003, "loss": 4.0749, "step": 7076 }, { "epoch": 0.07077, "grad_norm": 0.7739957741542802, "learning_rate": 0.003, "loss": 4.088, "step": 7077 }, { "epoch": 0.07078, "grad_norm": 1.0600410398425977, "learning_rate": 0.003, "loss": 4.1198, "step": 7078 }, { "epoch": 0.07079, "grad_norm": 1.3492326363705884, "learning_rate": 0.003, "loss": 4.1311, "step": 7079 }, { "epoch": 0.0708, "grad_norm": 0.9657503725086062, "learning_rate": 0.003, "loss": 4.0696, "step": 7080 }, { "epoch": 0.07081, "grad_norm": 1.1299019533525023, "learning_rate": 0.003, "loss": 4.1147, "step": 7081 }, { "epoch": 0.07082, "grad_norm": 0.9937833074975806, "learning_rate": 0.003, "loss": 4.0996, "step": 7082 }, { "epoch": 0.07083, "grad_norm": 1.2185933541832479, "learning_rate": 0.003, "loss": 4.0938, "step": 7083 }, { "epoch": 0.07084, "grad_norm": 0.9454456982283054, "learning_rate": 0.003, "loss": 4.1025, "step": 7084 }, { "epoch": 0.07085, "grad_norm": 0.8473361168787964, "learning_rate": 0.003, "loss": 4.1203, "step": 7085 }, { "epoch": 0.07086, "grad_norm": 0.9176320211127105, "learning_rate": 0.003, "loss": 4.0985, "step": 7086 }, { "epoch": 0.07087, "grad_norm": 1.1161177199303243, "learning_rate": 0.003, "loss": 4.0997, "step": 7087 }, { "epoch": 0.07088, "grad_norm": 1.0509489980561992, "learning_rate": 0.003, "loss": 4.0696, "step": 7088 }, { "epoch": 0.07089, "grad_norm": 0.9913619741647909, "learning_rate": 0.003, "loss": 4.0986, "step": 7089 }, { "epoch": 0.0709, "grad_norm": 1.1620589469680658, "learning_rate": 0.003, "loss": 4.105, "step": 7090 }, { "epoch": 0.07091, "grad_norm": 0.9945148846700074, "learning_rate": 0.003, "loss": 4.0926, "step": 7091 }, { "epoch": 0.07092, "grad_norm": 1.2536700163254593, "learning_rate": 0.003, "loss": 4.0984, "step": 7092 }, { "epoch": 0.07093, "grad_norm": 1.354622823433653, "learning_rate": 0.003, "loss": 4.125, "step": 7093 }, { "epoch": 0.07094, "grad_norm": 1.0630146455351173, "learning_rate": 0.003, "loss": 4.1384, "step": 7094 }, { "epoch": 0.07095, "grad_norm": 0.9454764813025995, "learning_rate": 0.003, "loss": 4.1224, "step": 7095 }, { "epoch": 0.07096, "grad_norm": 1.014738932196474, "learning_rate": 0.003, "loss": 4.0903, "step": 7096 }, { "epoch": 0.07097, "grad_norm": 1.1938770168715513, "learning_rate": 0.003, "loss": 4.1185, "step": 7097 }, { "epoch": 0.07098, "grad_norm": 0.8847286671786706, "learning_rate": 0.003, "loss": 4.1141, "step": 7098 }, { "epoch": 0.07099, "grad_norm": 1.043185377273873, "learning_rate": 0.003, "loss": 4.1178, "step": 7099 }, { "epoch": 0.071, "grad_norm": 1.1488413629431675, "learning_rate": 0.003, "loss": 4.11, "step": 7100 }, { "epoch": 0.07101, "grad_norm": 1.060125066460162, "learning_rate": 0.003, "loss": 4.0889, "step": 7101 }, { "epoch": 0.07102, "grad_norm": 1.043423213524699, "learning_rate": 0.003, "loss": 4.1043, "step": 7102 }, { "epoch": 0.07103, "grad_norm": 1.1487827868033067, "learning_rate": 0.003, "loss": 4.1185, "step": 7103 }, { "epoch": 0.07104, "grad_norm": 0.906308246298158, "learning_rate": 0.003, "loss": 4.1073, "step": 7104 }, { "epoch": 0.07105, "grad_norm": 1.0869757018932018, "learning_rate": 0.003, "loss": 4.1009, "step": 7105 }, { "epoch": 0.07106, "grad_norm": 0.994534751066206, "learning_rate": 0.003, "loss": 4.1143, "step": 7106 }, { "epoch": 0.07107, "grad_norm": 1.1286644088879794, "learning_rate": 0.003, "loss": 4.1153, "step": 7107 }, { "epoch": 0.07108, "grad_norm": 1.238356098237964, "learning_rate": 0.003, "loss": 4.126, "step": 7108 }, { "epoch": 0.07109, "grad_norm": 0.813617584956547, "learning_rate": 0.003, "loss": 4.101, "step": 7109 }, { "epoch": 0.0711, "grad_norm": 1.0042212442414669, "learning_rate": 0.003, "loss": 4.0818, "step": 7110 }, { "epoch": 0.07111, "grad_norm": 1.182160032235518, "learning_rate": 0.003, "loss": 4.0999, "step": 7111 }, { "epoch": 0.07112, "grad_norm": 1.2179586602810781, "learning_rate": 0.003, "loss": 4.1049, "step": 7112 }, { "epoch": 0.07113, "grad_norm": 1.1629649917625506, "learning_rate": 0.003, "loss": 4.1389, "step": 7113 }, { "epoch": 0.07114, "grad_norm": 1.062488521920579, "learning_rate": 0.003, "loss": 4.1132, "step": 7114 }, { "epoch": 0.07115, "grad_norm": 1.0870573312988112, "learning_rate": 0.003, "loss": 4.1009, "step": 7115 }, { "epoch": 0.07116, "grad_norm": 1.0520477518578835, "learning_rate": 0.003, "loss": 4.1036, "step": 7116 }, { "epoch": 0.07117, "grad_norm": 1.0392810415558837, "learning_rate": 0.003, "loss": 4.1196, "step": 7117 }, { "epoch": 0.07118, "grad_norm": 1.002861445960153, "learning_rate": 0.003, "loss": 4.124, "step": 7118 }, { "epoch": 0.07119, "grad_norm": 1.0589727147603938, "learning_rate": 0.003, "loss": 4.1023, "step": 7119 }, { "epoch": 0.0712, "grad_norm": 1.2060768839825937, "learning_rate": 0.003, "loss": 4.1042, "step": 7120 }, { "epoch": 0.07121, "grad_norm": 1.0484074848536764, "learning_rate": 0.003, "loss": 4.1037, "step": 7121 }, { "epoch": 0.07122, "grad_norm": 1.1680927354417303, "learning_rate": 0.003, "loss": 4.096, "step": 7122 }, { "epoch": 0.07123, "grad_norm": 1.0283240641322164, "learning_rate": 0.003, "loss": 4.087, "step": 7123 }, { "epoch": 0.07124, "grad_norm": 1.0363051726979002, "learning_rate": 0.003, "loss": 4.0842, "step": 7124 }, { "epoch": 0.07125, "grad_norm": 1.1375243622127051, "learning_rate": 0.003, "loss": 4.0934, "step": 7125 }, { "epoch": 0.07126, "grad_norm": 1.1681796706587095, "learning_rate": 0.003, "loss": 4.1313, "step": 7126 }, { "epoch": 0.07127, "grad_norm": 1.0438987028881606, "learning_rate": 0.003, "loss": 4.0967, "step": 7127 }, { "epoch": 0.07128, "grad_norm": 1.0929096590435705, "learning_rate": 0.003, "loss": 4.0825, "step": 7128 }, { "epoch": 0.07129, "grad_norm": 1.074862563144551, "learning_rate": 0.003, "loss": 4.1175, "step": 7129 }, { "epoch": 0.0713, "grad_norm": 1.0366606289855014, "learning_rate": 0.003, "loss": 4.1048, "step": 7130 }, { "epoch": 0.07131, "grad_norm": 1.0083999678728495, "learning_rate": 0.003, "loss": 4.0907, "step": 7131 }, { "epoch": 0.07132, "grad_norm": 1.1570389544595974, "learning_rate": 0.003, "loss": 4.1291, "step": 7132 }, { "epoch": 0.07133, "grad_norm": 1.0373700523560319, "learning_rate": 0.003, "loss": 4.1337, "step": 7133 }, { "epoch": 0.07134, "grad_norm": 1.1150502540574474, "learning_rate": 0.003, "loss": 4.0735, "step": 7134 }, { "epoch": 0.07135, "grad_norm": 1.0459023320389105, "learning_rate": 0.003, "loss": 4.1043, "step": 7135 }, { "epoch": 0.07136, "grad_norm": 1.0145196590306376, "learning_rate": 0.003, "loss": 4.1404, "step": 7136 }, { "epoch": 0.07137, "grad_norm": 1.25764777060964, "learning_rate": 0.003, "loss": 4.1102, "step": 7137 }, { "epoch": 0.07138, "grad_norm": 0.9231210864041297, "learning_rate": 0.003, "loss": 4.1013, "step": 7138 }, { "epoch": 0.07139, "grad_norm": 0.9792328685329048, "learning_rate": 0.003, "loss": 4.1013, "step": 7139 }, { "epoch": 0.0714, "grad_norm": 1.0188971768817803, "learning_rate": 0.003, "loss": 4.1118, "step": 7140 }, { "epoch": 0.07141, "grad_norm": 1.1228754806364862, "learning_rate": 0.003, "loss": 4.0907, "step": 7141 }, { "epoch": 0.07142, "grad_norm": 1.1921395910674482, "learning_rate": 0.003, "loss": 4.0808, "step": 7142 }, { "epoch": 0.07143, "grad_norm": 1.2166066068555654, "learning_rate": 0.003, "loss": 4.0714, "step": 7143 }, { "epoch": 0.07144, "grad_norm": 1.1851929558296612, "learning_rate": 0.003, "loss": 4.0845, "step": 7144 }, { "epoch": 0.07145, "grad_norm": 1.0160455666249097, "learning_rate": 0.003, "loss": 4.1034, "step": 7145 }, { "epoch": 0.07146, "grad_norm": 1.2365692135288116, "learning_rate": 0.003, "loss": 4.1105, "step": 7146 }, { "epoch": 0.07147, "grad_norm": 0.9464163524127707, "learning_rate": 0.003, "loss": 4.1111, "step": 7147 }, { "epoch": 0.07148, "grad_norm": 0.9187524935486385, "learning_rate": 0.003, "loss": 4.1135, "step": 7148 }, { "epoch": 0.07149, "grad_norm": 0.9915622145108923, "learning_rate": 0.003, "loss": 4.1216, "step": 7149 }, { "epoch": 0.0715, "grad_norm": 1.24972865830939, "learning_rate": 0.003, "loss": 4.1132, "step": 7150 }, { "epoch": 0.07151, "grad_norm": 1.0498265673158704, "learning_rate": 0.003, "loss": 4.1106, "step": 7151 }, { "epoch": 0.07152, "grad_norm": 1.1525404433626172, "learning_rate": 0.003, "loss": 4.129, "step": 7152 }, { "epoch": 0.07153, "grad_norm": 1.1094991540332446, "learning_rate": 0.003, "loss": 4.1127, "step": 7153 }, { "epoch": 0.07154, "grad_norm": 1.0631614580252882, "learning_rate": 0.003, "loss": 4.1438, "step": 7154 }, { "epoch": 0.07155, "grad_norm": 1.2822551903595085, "learning_rate": 0.003, "loss": 4.1114, "step": 7155 }, { "epoch": 0.07156, "grad_norm": 0.8602155152627339, "learning_rate": 0.003, "loss": 4.1119, "step": 7156 }, { "epoch": 0.07157, "grad_norm": 0.914412750635205, "learning_rate": 0.003, "loss": 4.0952, "step": 7157 }, { "epoch": 0.07158, "grad_norm": 1.3500367025031255, "learning_rate": 0.003, "loss": 4.1139, "step": 7158 }, { "epoch": 0.07159, "grad_norm": 1.0131822935746835, "learning_rate": 0.003, "loss": 4.1146, "step": 7159 }, { "epoch": 0.0716, "grad_norm": 1.107157491662557, "learning_rate": 0.003, "loss": 4.0845, "step": 7160 }, { "epoch": 0.07161, "grad_norm": 1.0609422302924651, "learning_rate": 0.003, "loss": 4.1164, "step": 7161 }, { "epoch": 0.07162, "grad_norm": 1.0617956432065383, "learning_rate": 0.003, "loss": 4.1163, "step": 7162 }, { "epoch": 0.07163, "grad_norm": 0.9961491455303284, "learning_rate": 0.003, "loss": 4.109, "step": 7163 }, { "epoch": 0.07164, "grad_norm": 1.0469051381193664, "learning_rate": 0.003, "loss": 4.0907, "step": 7164 }, { "epoch": 0.07165, "grad_norm": 1.1237775895482525, "learning_rate": 0.003, "loss": 4.0962, "step": 7165 }, { "epoch": 0.07166, "grad_norm": 1.2279690263793916, "learning_rate": 0.003, "loss": 4.1004, "step": 7166 }, { "epoch": 0.07167, "grad_norm": 1.023424584094055, "learning_rate": 0.003, "loss": 4.1139, "step": 7167 }, { "epoch": 0.07168, "grad_norm": 1.2707315933813994, "learning_rate": 0.003, "loss": 4.1359, "step": 7168 }, { "epoch": 0.07169, "grad_norm": 0.9172282319574873, "learning_rate": 0.003, "loss": 4.0963, "step": 7169 }, { "epoch": 0.0717, "grad_norm": 0.9083594703987031, "learning_rate": 0.003, "loss": 4.0871, "step": 7170 }, { "epoch": 0.07171, "grad_norm": 1.0486138594611163, "learning_rate": 0.003, "loss": 4.1162, "step": 7171 }, { "epoch": 0.07172, "grad_norm": 0.9700158502392761, "learning_rate": 0.003, "loss": 4.1153, "step": 7172 }, { "epoch": 0.07173, "grad_norm": 0.9301256536300708, "learning_rate": 0.003, "loss": 4.115, "step": 7173 }, { "epoch": 0.07174, "grad_norm": 0.8411748081354862, "learning_rate": 0.003, "loss": 4.0854, "step": 7174 }, { "epoch": 0.07175, "grad_norm": 0.978496890781848, "learning_rate": 0.003, "loss": 4.1256, "step": 7175 }, { "epoch": 0.07176, "grad_norm": 1.017883820000192, "learning_rate": 0.003, "loss": 4.1, "step": 7176 }, { "epoch": 0.07177, "grad_norm": 1.0876540425618906, "learning_rate": 0.003, "loss": 4.1099, "step": 7177 }, { "epoch": 0.07178, "grad_norm": 1.2815250814974248, "learning_rate": 0.003, "loss": 4.0738, "step": 7178 }, { "epoch": 0.07179, "grad_norm": 1.0343617750573875, "learning_rate": 0.003, "loss": 4.1043, "step": 7179 }, { "epoch": 0.0718, "grad_norm": 1.122066580167411, "learning_rate": 0.003, "loss": 4.1148, "step": 7180 }, { "epoch": 0.07181, "grad_norm": 0.9895968388675276, "learning_rate": 0.003, "loss": 4.0993, "step": 7181 }, { "epoch": 0.07182, "grad_norm": 1.140460079055109, "learning_rate": 0.003, "loss": 4.112, "step": 7182 }, { "epoch": 0.07183, "grad_norm": 1.0141873443326739, "learning_rate": 0.003, "loss": 4.1083, "step": 7183 }, { "epoch": 0.07184, "grad_norm": 1.0502527797264405, "learning_rate": 0.003, "loss": 4.0862, "step": 7184 }, { "epoch": 0.07185, "grad_norm": 1.2124587803714009, "learning_rate": 0.003, "loss": 4.0956, "step": 7185 }, { "epoch": 0.07186, "grad_norm": 0.9885002491164363, "learning_rate": 0.003, "loss": 4.1086, "step": 7186 }, { "epoch": 0.07187, "grad_norm": 1.111264868286653, "learning_rate": 0.003, "loss": 4.0986, "step": 7187 }, { "epoch": 0.07188, "grad_norm": 1.0145615669399324, "learning_rate": 0.003, "loss": 4.0655, "step": 7188 }, { "epoch": 0.07189, "grad_norm": 1.0217994358701927, "learning_rate": 0.003, "loss": 4.1135, "step": 7189 }, { "epoch": 0.0719, "grad_norm": 1.0413788946815261, "learning_rate": 0.003, "loss": 4.1107, "step": 7190 }, { "epoch": 0.07191, "grad_norm": 1.094167060143487, "learning_rate": 0.003, "loss": 4.1239, "step": 7191 }, { "epoch": 0.07192, "grad_norm": 1.0055555200473454, "learning_rate": 0.003, "loss": 4.0844, "step": 7192 }, { "epoch": 0.07193, "grad_norm": 0.9850273042601376, "learning_rate": 0.003, "loss": 4.1097, "step": 7193 }, { "epoch": 0.07194, "grad_norm": 1.0146797688856157, "learning_rate": 0.003, "loss": 4.1098, "step": 7194 }, { "epoch": 0.07195, "grad_norm": 1.1432646128077548, "learning_rate": 0.003, "loss": 4.0891, "step": 7195 }, { "epoch": 0.07196, "grad_norm": 1.0481540314236983, "learning_rate": 0.003, "loss": 4.094, "step": 7196 }, { "epoch": 0.07197, "grad_norm": 0.9542580970677246, "learning_rate": 0.003, "loss": 4.117, "step": 7197 }, { "epoch": 0.07198, "grad_norm": 1.10288485158813, "learning_rate": 0.003, "loss": 4.0802, "step": 7198 }, { "epoch": 0.07199, "grad_norm": 1.1509658058015648, "learning_rate": 0.003, "loss": 4.0943, "step": 7199 }, { "epoch": 0.072, "grad_norm": 1.0793475034183406, "learning_rate": 0.003, "loss": 4.1278, "step": 7200 }, { "epoch": 0.07201, "grad_norm": 0.9556489386143434, "learning_rate": 0.003, "loss": 4.0889, "step": 7201 }, { "epoch": 0.07202, "grad_norm": 0.9809904337905765, "learning_rate": 0.003, "loss": 4.1219, "step": 7202 }, { "epoch": 0.07203, "grad_norm": 1.1707509067401052, "learning_rate": 0.003, "loss": 4.0923, "step": 7203 }, { "epoch": 0.07204, "grad_norm": 1.3323882560669078, "learning_rate": 0.003, "loss": 4.1331, "step": 7204 }, { "epoch": 0.07205, "grad_norm": 1.1334906555538489, "learning_rate": 0.003, "loss": 4.1271, "step": 7205 }, { "epoch": 0.07206, "grad_norm": 1.1729688120105672, "learning_rate": 0.003, "loss": 4.1048, "step": 7206 }, { "epoch": 0.07207, "grad_norm": 0.8357880798460113, "learning_rate": 0.003, "loss": 4.1004, "step": 7207 }, { "epoch": 0.07208, "grad_norm": 1.0253844160329557, "learning_rate": 0.003, "loss": 4.0953, "step": 7208 }, { "epoch": 0.07209, "grad_norm": 1.2699597428691942, "learning_rate": 0.003, "loss": 4.0995, "step": 7209 }, { "epoch": 0.0721, "grad_norm": 0.844492273082808, "learning_rate": 0.003, "loss": 4.093, "step": 7210 }, { "epoch": 0.07211, "grad_norm": 0.9959930714732329, "learning_rate": 0.003, "loss": 4.1101, "step": 7211 }, { "epoch": 0.07212, "grad_norm": 1.2255608222794574, "learning_rate": 0.003, "loss": 4.1309, "step": 7212 }, { "epoch": 0.07213, "grad_norm": 1.0751280898968962, "learning_rate": 0.003, "loss": 4.1341, "step": 7213 }, { "epoch": 0.07214, "grad_norm": 1.112519303902804, "learning_rate": 0.003, "loss": 4.0849, "step": 7214 }, { "epoch": 0.07215, "grad_norm": 1.0219317668597434, "learning_rate": 0.003, "loss": 4.108, "step": 7215 }, { "epoch": 0.07216, "grad_norm": 0.9213765728970278, "learning_rate": 0.003, "loss": 4.1156, "step": 7216 }, { "epoch": 0.07217, "grad_norm": 0.8562736879877679, "learning_rate": 0.003, "loss": 4.1048, "step": 7217 }, { "epoch": 0.07218, "grad_norm": 1.0774634512005679, "learning_rate": 0.003, "loss": 4.1134, "step": 7218 }, { "epoch": 0.07219, "grad_norm": 1.1802394760260384, "learning_rate": 0.003, "loss": 4.0937, "step": 7219 }, { "epoch": 0.0722, "grad_norm": 0.919224360560586, "learning_rate": 0.003, "loss": 4.1118, "step": 7220 }, { "epoch": 0.07221, "grad_norm": 0.9309775053843943, "learning_rate": 0.003, "loss": 4.1138, "step": 7221 }, { "epoch": 0.07222, "grad_norm": 1.251585304062438, "learning_rate": 0.003, "loss": 4.1072, "step": 7222 }, { "epoch": 0.07223, "grad_norm": 1.2153975646372994, "learning_rate": 0.003, "loss": 4.1248, "step": 7223 }, { "epoch": 0.07224, "grad_norm": 0.881797645512631, "learning_rate": 0.003, "loss": 4.0794, "step": 7224 }, { "epoch": 0.07225, "grad_norm": 1.0903953232903485, "learning_rate": 0.003, "loss": 4.1121, "step": 7225 }, { "epoch": 0.07226, "grad_norm": 1.3155759554258537, "learning_rate": 0.003, "loss": 4.1166, "step": 7226 }, { "epoch": 0.07227, "grad_norm": 1.0076198414601187, "learning_rate": 0.003, "loss": 4.0908, "step": 7227 }, { "epoch": 0.07228, "grad_norm": 1.1644273856024416, "learning_rate": 0.003, "loss": 4.0879, "step": 7228 }, { "epoch": 0.07229, "grad_norm": 0.9826701191964766, "learning_rate": 0.003, "loss": 4.1052, "step": 7229 }, { "epoch": 0.0723, "grad_norm": 1.0039682774338574, "learning_rate": 0.003, "loss": 4.0768, "step": 7230 }, { "epoch": 0.07231, "grad_norm": 1.3290115071414985, "learning_rate": 0.003, "loss": 4.1198, "step": 7231 }, { "epoch": 0.07232, "grad_norm": 0.8889806292787095, "learning_rate": 0.003, "loss": 4.1044, "step": 7232 }, { "epoch": 0.07233, "grad_norm": 0.9033119332258699, "learning_rate": 0.003, "loss": 4.1063, "step": 7233 }, { "epoch": 0.07234, "grad_norm": 0.9850515096650099, "learning_rate": 0.003, "loss": 4.0861, "step": 7234 }, { "epoch": 0.07235, "grad_norm": 1.0280057066621469, "learning_rate": 0.003, "loss": 4.1324, "step": 7235 }, { "epoch": 0.07236, "grad_norm": 0.8959961876752984, "learning_rate": 0.003, "loss": 4.1287, "step": 7236 }, { "epoch": 0.07237, "grad_norm": 0.9578859446824379, "learning_rate": 0.003, "loss": 4.0914, "step": 7237 }, { "epoch": 0.07238, "grad_norm": 1.0904222667194374, "learning_rate": 0.003, "loss": 4.1223, "step": 7238 }, { "epoch": 0.07239, "grad_norm": 1.1354385440528534, "learning_rate": 0.003, "loss": 4.1233, "step": 7239 }, { "epoch": 0.0724, "grad_norm": 1.1445170478246343, "learning_rate": 0.003, "loss": 4.0808, "step": 7240 }, { "epoch": 0.07241, "grad_norm": 1.330496175036632, "learning_rate": 0.003, "loss": 4.1283, "step": 7241 }, { "epoch": 0.07242, "grad_norm": 1.0029178572683988, "learning_rate": 0.003, "loss": 4.08, "step": 7242 }, { "epoch": 0.07243, "grad_norm": 1.0419556676047186, "learning_rate": 0.003, "loss": 4.1167, "step": 7243 }, { "epoch": 0.07244, "grad_norm": 1.1405191704713464, "learning_rate": 0.003, "loss": 4.1016, "step": 7244 }, { "epoch": 0.07245, "grad_norm": 1.055957134761977, "learning_rate": 0.003, "loss": 4.1, "step": 7245 }, { "epoch": 0.07246, "grad_norm": 0.9746458798890896, "learning_rate": 0.003, "loss": 4.1078, "step": 7246 }, { "epoch": 0.07247, "grad_norm": 1.2482468143322891, "learning_rate": 0.003, "loss": 4.1325, "step": 7247 }, { "epoch": 0.07248, "grad_norm": 0.8855907919154823, "learning_rate": 0.003, "loss": 4.0924, "step": 7248 }, { "epoch": 0.07249, "grad_norm": 1.0413021832365006, "learning_rate": 0.003, "loss": 4.0745, "step": 7249 }, { "epoch": 0.0725, "grad_norm": 1.1504077090947193, "learning_rate": 0.003, "loss": 4.0915, "step": 7250 }, { "epoch": 0.07251, "grad_norm": 1.509045508344056, "learning_rate": 0.003, "loss": 4.1148, "step": 7251 }, { "epoch": 0.07252, "grad_norm": 0.7932617771234077, "learning_rate": 0.003, "loss": 4.0462, "step": 7252 }, { "epoch": 0.07253, "grad_norm": 0.8358630953519474, "learning_rate": 0.003, "loss": 4.0732, "step": 7253 }, { "epoch": 0.07254, "grad_norm": 1.1222171632357134, "learning_rate": 0.003, "loss": 4.1253, "step": 7254 }, { "epoch": 0.07255, "grad_norm": 1.0804481975183948, "learning_rate": 0.003, "loss": 4.1212, "step": 7255 }, { "epoch": 0.07256, "grad_norm": 1.2281418759865774, "learning_rate": 0.003, "loss": 4.124, "step": 7256 }, { "epoch": 0.07257, "grad_norm": 0.8609220422938224, "learning_rate": 0.003, "loss": 4.1049, "step": 7257 }, { "epoch": 0.07258, "grad_norm": 0.917114476220674, "learning_rate": 0.003, "loss": 4.0737, "step": 7258 }, { "epoch": 0.07259, "grad_norm": 1.1443689847230054, "learning_rate": 0.003, "loss": 4.106, "step": 7259 }, { "epoch": 0.0726, "grad_norm": 0.8744085214791258, "learning_rate": 0.003, "loss": 4.0793, "step": 7260 }, { "epoch": 0.07261, "grad_norm": 0.9452003703653208, "learning_rate": 0.003, "loss": 4.1068, "step": 7261 }, { "epoch": 0.07262, "grad_norm": 1.113889337528543, "learning_rate": 0.003, "loss": 4.1041, "step": 7262 }, { "epoch": 0.07263, "grad_norm": 1.1631423603603335, "learning_rate": 0.003, "loss": 4.113, "step": 7263 }, { "epoch": 0.07264, "grad_norm": 1.0786175525014503, "learning_rate": 0.003, "loss": 4.065, "step": 7264 }, { "epoch": 0.07265, "grad_norm": 1.1580180230536627, "learning_rate": 0.003, "loss": 4.1077, "step": 7265 }, { "epoch": 0.07266, "grad_norm": 0.9070949557708119, "learning_rate": 0.003, "loss": 4.1203, "step": 7266 }, { "epoch": 0.07267, "grad_norm": 1.0924109492734309, "learning_rate": 0.003, "loss": 4.1088, "step": 7267 }, { "epoch": 0.07268, "grad_norm": 1.265085711037148, "learning_rate": 0.003, "loss": 4.1246, "step": 7268 }, { "epoch": 0.07269, "grad_norm": 0.9220577386588262, "learning_rate": 0.003, "loss": 4.1021, "step": 7269 }, { "epoch": 0.0727, "grad_norm": 1.0994973872701153, "learning_rate": 0.003, "loss": 4.0919, "step": 7270 }, { "epoch": 0.07271, "grad_norm": 1.0390055174761674, "learning_rate": 0.003, "loss": 4.1003, "step": 7271 }, { "epoch": 0.07272, "grad_norm": 1.109067023286393, "learning_rate": 0.003, "loss": 4.0953, "step": 7272 }, { "epoch": 0.07273, "grad_norm": 1.190775009770159, "learning_rate": 0.003, "loss": 4.1119, "step": 7273 }, { "epoch": 0.07274, "grad_norm": 0.9643586526737157, "learning_rate": 0.003, "loss": 4.0699, "step": 7274 }, { "epoch": 0.07275, "grad_norm": 0.807588478281446, "learning_rate": 0.003, "loss": 4.0818, "step": 7275 }, { "epoch": 0.07276, "grad_norm": 0.7367459059199548, "learning_rate": 0.003, "loss": 4.0609, "step": 7276 }, { "epoch": 0.07277, "grad_norm": 0.872868798546834, "learning_rate": 0.003, "loss": 4.0958, "step": 7277 }, { "epoch": 0.07278, "grad_norm": 1.021625890463033, "learning_rate": 0.003, "loss": 4.1077, "step": 7278 }, { "epoch": 0.07279, "grad_norm": 1.1561771690179692, "learning_rate": 0.003, "loss": 4.0914, "step": 7279 }, { "epoch": 0.0728, "grad_norm": 0.995174312767239, "learning_rate": 0.003, "loss": 4.0716, "step": 7280 }, { "epoch": 0.07281, "grad_norm": 1.2231458022379895, "learning_rate": 0.003, "loss": 4.0999, "step": 7281 }, { "epoch": 0.07282, "grad_norm": 0.9752915595680061, "learning_rate": 0.003, "loss": 4.1119, "step": 7282 }, { "epoch": 0.07283, "grad_norm": 1.117893543537355, "learning_rate": 0.003, "loss": 4.1293, "step": 7283 }, { "epoch": 0.07284, "grad_norm": 1.0083611434052082, "learning_rate": 0.003, "loss": 4.0841, "step": 7284 }, { "epoch": 0.07285, "grad_norm": 0.9630552431056575, "learning_rate": 0.003, "loss": 4.1123, "step": 7285 }, { "epoch": 0.07286, "grad_norm": 0.8169119244859634, "learning_rate": 0.003, "loss": 4.0835, "step": 7286 }, { "epoch": 0.07287, "grad_norm": 0.903401863942306, "learning_rate": 0.003, "loss": 4.1082, "step": 7287 }, { "epoch": 0.07288, "grad_norm": 1.0869951380389518, "learning_rate": 0.003, "loss": 4.0852, "step": 7288 }, { "epoch": 0.07289, "grad_norm": 1.3801544823334104, "learning_rate": 0.003, "loss": 4.1174, "step": 7289 }, { "epoch": 0.0729, "grad_norm": 0.9555863894636518, "learning_rate": 0.003, "loss": 4.1294, "step": 7290 }, { "epoch": 0.07291, "grad_norm": 0.9538466741745802, "learning_rate": 0.003, "loss": 4.092, "step": 7291 }, { "epoch": 0.07292, "grad_norm": 0.9359092532053526, "learning_rate": 0.003, "loss": 4.1041, "step": 7292 }, { "epoch": 0.07293, "grad_norm": 1.095958126570705, "learning_rate": 0.003, "loss": 4.094, "step": 7293 }, { "epoch": 0.07294, "grad_norm": 1.2068433544777886, "learning_rate": 0.003, "loss": 4.0814, "step": 7294 }, { "epoch": 0.07295, "grad_norm": 0.9752948921226386, "learning_rate": 0.003, "loss": 4.0727, "step": 7295 }, { "epoch": 0.07296, "grad_norm": 1.0045088941314217, "learning_rate": 0.003, "loss": 4.0879, "step": 7296 }, { "epoch": 0.07297, "grad_norm": 1.2851924940126396, "learning_rate": 0.003, "loss": 4.1304, "step": 7297 }, { "epoch": 0.07298, "grad_norm": 0.8984769908416159, "learning_rate": 0.003, "loss": 4.0677, "step": 7298 }, { "epoch": 0.07299, "grad_norm": 1.0616115245681086, "learning_rate": 0.003, "loss": 4.1007, "step": 7299 }, { "epoch": 0.073, "grad_norm": 1.124975187025094, "learning_rate": 0.003, "loss": 4.0884, "step": 7300 }, { "epoch": 0.07301, "grad_norm": 1.0277803074493952, "learning_rate": 0.003, "loss": 4.0693, "step": 7301 }, { "epoch": 0.07302, "grad_norm": 1.2837150074018109, "learning_rate": 0.003, "loss": 4.0838, "step": 7302 }, { "epoch": 0.07303, "grad_norm": 0.964759536670288, "learning_rate": 0.003, "loss": 4.1097, "step": 7303 }, { "epoch": 0.07304, "grad_norm": 1.1222488995629787, "learning_rate": 0.003, "loss": 4.0954, "step": 7304 }, { "epoch": 0.07305, "grad_norm": 1.0053576459104663, "learning_rate": 0.003, "loss": 4.086, "step": 7305 }, { "epoch": 0.07306, "grad_norm": 0.9644307566496255, "learning_rate": 0.003, "loss": 4.1231, "step": 7306 }, { "epoch": 0.07307, "grad_norm": 1.1082915404106022, "learning_rate": 0.003, "loss": 4.1324, "step": 7307 }, { "epoch": 0.07308, "grad_norm": 0.982019032729292, "learning_rate": 0.003, "loss": 4.1019, "step": 7308 }, { "epoch": 0.07309, "grad_norm": 1.1223828148450474, "learning_rate": 0.003, "loss": 4.1101, "step": 7309 }, { "epoch": 0.0731, "grad_norm": 1.1820726188937547, "learning_rate": 0.003, "loss": 4.0677, "step": 7310 }, { "epoch": 0.07311, "grad_norm": 1.293636953613459, "learning_rate": 0.003, "loss": 4.1104, "step": 7311 }, { "epoch": 0.07312, "grad_norm": 1.1654414594535567, "learning_rate": 0.003, "loss": 4.1196, "step": 7312 }, { "epoch": 0.07313, "grad_norm": 0.955703652754267, "learning_rate": 0.003, "loss": 4.1038, "step": 7313 }, { "epoch": 0.07314, "grad_norm": 0.9882220489290244, "learning_rate": 0.003, "loss": 4.1156, "step": 7314 }, { "epoch": 0.07315, "grad_norm": 1.15014313157189, "learning_rate": 0.003, "loss": 4.1039, "step": 7315 }, { "epoch": 0.07316, "grad_norm": 1.1357410212032995, "learning_rate": 0.003, "loss": 4.1239, "step": 7316 }, { "epoch": 0.07317, "grad_norm": 1.0678436839863392, "learning_rate": 0.003, "loss": 4.1067, "step": 7317 }, { "epoch": 0.07318, "grad_norm": 1.0552939298230526, "learning_rate": 0.003, "loss": 4.0911, "step": 7318 }, { "epoch": 0.07319, "grad_norm": 1.0631930674320003, "learning_rate": 0.003, "loss": 4.1093, "step": 7319 }, { "epoch": 0.0732, "grad_norm": 1.378616480794266, "learning_rate": 0.003, "loss": 4.0949, "step": 7320 }, { "epoch": 0.07321, "grad_norm": 0.8515266821180929, "learning_rate": 0.003, "loss": 4.0929, "step": 7321 }, { "epoch": 0.07322, "grad_norm": 0.9617491536309828, "learning_rate": 0.003, "loss": 4.1151, "step": 7322 }, { "epoch": 0.07323, "grad_norm": 1.2026300700448722, "learning_rate": 0.003, "loss": 4.1065, "step": 7323 }, { "epoch": 0.07324, "grad_norm": 1.1582938023197826, "learning_rate": 0.003, "loss": 4.1018, "step": 7324 }, { "epoch": 0.07325, "grad_norm": 0.9248125901403387, "learning_rate": 0.003, "loss": 4.0758, "step": 7325 }, { "epoch": 0.07326, "grad_norm": 1.0036120714649015, "learning_rate": 0.003, "loss": 4.1154, "step": 7326 }, { "epoch": 0.07327, "grad_norm": 1.1752819874476683, "learning_rate": 0.003, "loss": 4.1201, "step": 7327 }, { "epoch": 0.07328, "grad_norm": 1.0844284580206358, "learning_rate": 0.003, "loss": 4.0763, "step": 7328 }, { "epoch": 0.07329, "grad_norm": 1.131174265281523, "learning_rate": 0.003, "loss": 4.1, "step": 7329 }, { "epoch": 0.0733, "grad_norm": 1.0902888266412114, "learning_rate": 0.003, "loss": 4.0876, "step": 7330 }, { "epoch": 0.07331, "grad_norm": 1.0697080836933746, "learning_rate": 0.003, "loss": 4.0828, "step": 7331 }, { "epoch": 0.07332, "grad_norm": 1.1425679752194804, "learning_rate": 0.003, "loss": 4.111, "step": 7332 }, { "epoch": 0.07333, "grad_norm": 1.0006191688665407, "learning_rate": 0.003, "loss": 4.1018, "step": 7333 }, { "epoch": 0.07334, "grad_norm": 0.9001558750090279, "learning_rate": 0.003, "loss": 4.0915, "step": 7334 }, { "epoch": 0.07335, "grad_norm": 0.9455956599304268, "learning_rate": 0.003, "loss": 4.0664, "step": 7335 }, { "epoch": 0.07336, "grad_norm": 1.0305099426952913, "learning_rate": 0.003, "loss": 4.116, "step": 7336 }, { "epoch": 0.07337, "grad_norm": 0.9198982480606116, "learning_rate": 0.003, "loss": 4.1026, "step": 7337 }, { "epoch": 0.07338, "grad_norm": 1.1098987695082583, "learning_rate": 0.003, "loss": 4.0931, "step": 7338 }, { "epoch": 0.07339, "grad_norm": 1.1452129437911507, "learning_rate": 0.003, "loss": 4.1276, "step": 7339 }, { "epoch": 0.0734, "grad_norm": 1.0502294889063275, "learning_rate": 0.003, "loss": 4.0832, "step": 7340 }, { "epoch": 0.07341, "grad_norm": 1.1208815982947145, "learning_rate": 0.003, "loss": 4.1039, "step": 7341 }, { "epoch": 0.07342, "grad_norm": 1.275789986817216, "learning_rate": 0.003, "loss": 4.1273, "step": 7342 }, { "epoch": 0.07343, "grad_norm": 0.9265911774890242, "learning_rate": 0.003, "loss": 4.1065, "step": 7343 }, { "epoch": 0.07344, "grad_norm": 0.9723085170850532, "learning_rate": 0.003, "loss": 4.1086, "step": 7344 }, { "epoch": 0.07345, "grad_norm": 1.310974637981528, "learning_rate": 0.003, "loss": 4.0888, "step": 7345 }, { "epoch": 0.07346, "grad_norm": 0.9859934935381308, "learning_rate": 0.003, "loss": 4.0697, "step": 7346 }, { "epoch": 0.07347, "grad_norm": 1.311642444465165, "learning_rate": 0.003, "loss": 4.103, "step": 7347 }, { "epoch": 0.07348, "grad_norm": 1.0755786167180137, "learning_rate": 0.003, "loss": 4.0998, "step": 7348 }, { "epoch": 0.07349, "grad_norm": 1.117632483088747, "learning_rate": 0.003, "loss": 4.1032, "step": 7349 }, { "epoch": 0.0735, "grad_norm": 0.9556614107870417, "learning_rate": 0.003, "loss": 4.0854, "step": 7350 }, { "epoch": 0.07351, "grad_norm": 1.0030831120053332, "learning_rate": 0.003, "loss": 4.08, "step": 7351 }, { "epoch": 0.07352, "grad_norm": 1.0839532000386127, "learning_rate": 0.003, "loss": 4.1031, "step": 7352 }, { "epoch": 0.07353, "grad_norm": 0.956388917885665, "learning_rate": 0.003, "loss": 4.1221, "step": 7353 }, { "epoch": 0.07354, "grad_norm": 1.2540325368846519, "learning_rate": 0.003, "loss": 4.0891, "step": 7354 }, { "epoch": 0.07355, "grad_norm": 0.8476486590374245, "learning_rate": 0.003, "loss": 4.105, "step": 7355 }, { "epoch": 0.07356, "grad_norm": 0.9072579050921763, "learning_rate": 0.003, "loss": 4.1082, "step": 7356 }, { "epoch": 0.07357, "grad_norm": 1.07606465118927, "learning_rate": 0.003, "loss": 4.0967, "step": 7357 }, { "epoch": 0.07358, "grad_norm": 0.9758249229656148, "learning_rate": 0.003, "loss": 4.1061, "step": 7358 }, { "epoch": 0.07359, "grad_norm": 1.2808917830241577, "learning_rate": 0.003, "loss": 4.1186, "step": 7359 }, { "epoch": 0.0736, "grad_norm": 1.0389550581293132, "learning_rate": 0.003, "loss": 4.1194, "step": 7360 }, { "epoch": 0.07361, "grad_norm": 1.4439062529022064, "learning_rate": 0.003, "loss": 4.1017, "step": 7361 }, { "epoch": 0.07362, "grad_norm": 1.0842141270597971, "learning_rate": 0.003, "loss": 4.1248, "step": 7362 }, { "epoch": 0.07363, "grad_norm": 1.0586360928887677, "learning_rate": 0.003, "loss": 4.087, "step": 7363 }, { "epoch": 0.07364, "grad_norm": 1.0790877575745619, "learning_rate": 0.003, "loss": 4.1239, "step": 7364 }, { "epoch": 0.07365, "grad_norm": 1.0487723270085267, "learning_rate": 0.003, "loss": 4.076, "step": 7365 }, { "epoch": 0.07366, "grad_norm": 0.9486821316909797, "learning_rate": 0.003, "loss": 4.1012, "step": 7366 }, { "epoch": 0.07367, "grad_norm": 1.0709893106339785, "learning_rate": 0.003, "loss": 4.1135, "step": 7367 }, { "epoch": 0.07368, "grad_norm": 1.3047096296178842, "learning_rate": 0.003, "loss": 4.1356, "step": 7368 }, { "epoch": 0.07369, "grad_norm": 0.9701406143317088, "learning_rate": 0.003, "loss": 4.0777, "step": 7369 }, { "epoch": 0.0737, "grad_norm": 0.9989447329095311, "learning_rate": 0.003, "loss": 4.1089, "step": 7370 }, { "epoch": 0.07371, "grad_norm": 0.9302946104087475, "learning_rate": 0.003, "loss": 4.1284, "step": 7371 }, { "epoch": 0.07372, "grad_norm": 1.0077914951421132, "learning_rate": 0.003, "loss": 4.1102, "step": 7372 }, { "epoch": 0.07373, "grad_norm": 1.0343422807862372, "learning_rate": 0.003, "loss": 4.1144, "step": 7373 }, { "epoch": 0.07374, "grad_norm": 1.0817869352825151, "learning_rate": 0.003, "loss": 4.1009, "step": 7374 }, { "epoch": 0.07375, "grad_norm": 0.9543927133710544, "learning_rate": 0.003, "loss": 4.1075, "step": 7375 }, { "epoch": 0.07376, "grad_norm": 1.0823157063184048, "learning_rate": 0.003, "loss": 4.1124, "step": 7376 }, { "epoch": 0.07377, "grad_norm": 1.4523320980123982, "learning_rate": 0.003, "loss": 4.1164, "step": 7377 }, { "epoch": 0.07378, "grad_norm": 0.9920344749925358, "learning_rate": 0.003, "loss": 4.1196, "step": 7378 }, { "epoch": 0.07379, "grad_norm": 1.2023630780465542, "learning_rate": 0.003, "loss": 4.094, "step": 7379 }, { "epoch": 0.0738, "grad_norm": 0.979976959424781, "learning_rate": 0.003, "loss": 4.0747, "step": 7380 }, { "epoch": 0.07381, "grad_norm": 1.2122785872569304, "learning_rate": 0.003, "loss": 4.0896, "step": 7381 }, { "epoch": 0.07382, "grad_norm": 1.0711158041090567, "learning_rate": 0.003, "loss": 4.1363, "step": 7382 }, { "epoch": 0.07383, "grad_norm": 1.0968738055753358, "learning_rate": 0.003, "loss": 4.1127, "step": 7383 }, { "epoch": 0.07384, "grad_norm": 1.0122366795759148, "learning_rate": 0.003, "loss": 4.0832, "step": 7384 }, { "epoch": 0.07385, "grad_norm": 1.1344254281864916, "learning_rate": 0.003, "loss": 4.0988, "step": 7385 }, { "epoch": 0.07386, "grad_norm": 0.9528627923235372, "learning_rate": 0.003, "loss": 4.1108, "step": 7386 }, { "epoch": 0.07387, "grad_norm": 1.0340287194364897, "learning_rate": 0.003, "loss": 4.0853, "step": 7387 }, { "epoch": 0.07388, "grad_norm": 1.0661053253287225, "learning_rate": 0.003, "loss": 4.1102, "step": 7388 }, { "epoch": 0.07389, "grad_norm": 1.0266500281332847, "learning_rate": 0.003, "loss": 4.0942, "step": 7389 }, { "epoch": 0.0739, "grad_norm": 1.129668600576453, "learning_rate": 0.003, "loss": 4.0874, "step": 7390 }, { "epoch": 0.07391, "grad_norm": 1.1431567245789562, "learning_rate": 0.003, "loss": 4.1135, "step": 7391 }, { "epoch": 0.07392, "grad_norm": 0.9263556557950041, "learning_rate": 0.003, "loss": 4.0637, "step": 7392 }, { "epoch": 0.07393, "grad_norm": 1.1810603985633787, "learning_rate": 0.003, "loss": 4.112, "step": 7393 }, { "epoch": 0.07394, "grad_norm": 1.3052057344266057, "learning_rate": 0.003, "loss": 4.1068, "step": 7394 }, { "epoch": 0.07395, "grad_norm": 0.8902451430955156, "learning_rate": 0.003, "loss": 4.0802, "step": 7395 }, { "epoch": 0.07396, "grad_norm": 0.9401136963979129, "learning_rate": 0.003, "loss": 4.0974, "step": 7396 }, { "epoch": 0.07397, "grad_norm": 1.1336701427876885, "learning_rate": 0.003, "loss": 4.1018, "step": 7397 }, { "epoch": 0.07398, "grad_norm": 1.021978406987868, "learning_rate": 0.003, "loss": 4.0967, "step": 7398 }, { "epoch": 0.07399, "grad_norm": 1.4954509140076173, "learning_rate": 0.003, "loss": 4.0844, "step": 7399 }, { "epoch": 0.074, "grad_norm": 1.0211411356410278, "learning_rate": 0.003, "loss": 4.1187, "step": 7400 }, { "epoch": 0.07401, "grad_norm": 1.0269092564499978, "learning_rate": 0.003, "loss": 4.0817, "step": 7401 }, { "epoch": 0.07402, "grad_norm": 1.4125982359822833, "learning_rate": 0.003, "loss": 4.1134, "step": 7402 }, { "epoch": 0.07403, "grad_norm": 0.9177008703221856, "learning_rate": 0.003, "loss": 4.112, "step": 7403 }, { "epoch": 0.07404, "grad_norm": 1.1319798151303864, "learning_rate": 0.003, "loss": 4.1469, "step": 7404 }, { "epoch": 0.07405, "grad_norm": 0.9497899734385127, "learning_rate": 0.003, "loss": 4.0969, "step": 7405 }, { "epoch": 0.07406, "grad_norm": 0.9312788500314818, "learning_rate": 0.003, "loss": 4.0991, "step": 7406 }, { "epoch": 0.07407, "grad_norm": 1.1720689446037251, "learning_rate": 0.003, "loss": 4.1182, "step": 7407 }, { "epoch": 0.07408, "grad_norm": 1.1799841021387516, "learning_rate": 0.003, "loss": 4.1127, "step": 7408 }, { "epoch": 0.07409, "grad_norm": 0.8985031842337684, "learning_rate": 0.003, "loss": 4.0919, "step": 7409 }, { "epoch": 0.0741, "grad_norm": 1.0335323207407554, "learning_rate": 0.003, "loss": 4.1166, "step": 7410 }, { "epoch": 0.07411, "grad_norm": 1.1752050110355394, "learning_rate": 0.003, "loss": 4.1093, "step": 7411 }, { "epoch": 0.07412, "grad_norm": 0.8792893537815811, "learning_rate": 0.003, "loss": 4.0922, "step": 7412 }, { "epoch": 0.07413, "grad_norm": 0.8217721513844498, "learning_rate": 0.003, "loss": 4.0923, "step": 7413 }, { "epoch": 0.07414, "grad_norm": 1.0066515265982865, "learning_rate": 0.003, "loss": 4.095, "step": 7414 }, { "epoch": 0.07415, "grad_norm": 1.1016182405167252, "learning_rate": 0.003, "loss": 4.0997, "step": 7415 }, { "epoch": 0.07416, "grad_norm": 1.1590514698613215, "learning_rate": 0.003, "loss": 4.0827, "step": 7416 }, { "epoch": 0.07417, "grad_norm": 1.2339180292510161, "learning_rate": 0.003, "loss": 4.106, "step": 7417 }, { "epoch": 0.07418, "grad_norm": 0.9874762325328805, "learning_rate": 0.003, "loss": 4.0841, "step": 7418 }, { "epoch": 0.07419, "grad_norm": 0.9246629621622819, "learning_rate": 0.003, "loss": 4.0627, "step": 7419 }, { "epoch": 0.0742, "grad_norm": 1.1184657030620395, "learning_rate": 0.003, "loss": 4.1114, "step": 7420 }, { "epoch": 0.07421, "grad_norm": 1.0677774940176092, "learning_rate": 0.003, "loss": 4.0825, "step": 7421 }, { "epoch": 0.07422, "grad_norm": 1.1836820756336819, "learning_rate": 0.003, "loss": 4.1064, "step": 7422 }, { "epoch": 0.07423, "grad_norm": 0.8743651545923469, "learning_rate": 0.003, "loss": 4.0784, "step": 7423 }, { "epoch": 0.07424, "grad_norm": 0.7825771879450308, "learning_rate": 0.003, "loss": 4.1019, "step": 7424 }, { "epoch": 0.07425, "grad_norm": 0.870565586334107, "learning_rate": 0.003, "loss": 4.0667, "step": 7425 }, { "epoch": 0.07426, "grad_norm": 0.8918705011152487, "learning_rate": 0.003, "loss": 4.066, "step": 7426 }, { "epoch": 0.07427, "grad_norm": 0.8411163701679026, "learning_rate": 0.003, "loss": 4.0701, "step": 7427 }, { "epoch": 0.07428, "grad_norm": 0.93864995125254, "learning_rate": 0.003, "loss": 4.1256, "step": 7428 }, { "epoch": 0.07429, "grad_norm": 1.375676042770668, "learning_rate": 0.003, "loss": 4.1109, "step": 7429 }, { "epoch": 0.0743, "grad_norm": 1.2901201769037816, "learning_rate": 0.003, "loss": 4.0949, "step": 7430 }, { "epoch": 0.07431, "grad_norm": 1.0568232104600719, "learning_rate": 0.003, "loss": 4.0844, "step": 7431 }, { "epoch": 0.07432, "grad_norm": 1.1699681424250212, "learning_rate": 0.003, "loss": 4.1097, "step": 7432 }, { "epoch": 0.07433, "grad_norm": 0.8912336353896766, "learning_rate": 0.003, "loss": 4.0873, "step": 7433 }, { "epoch": 0.07434, "grad_norm": 0.9043967976875844, "learning_rate": 0.003, "loss": 4.0608, "step": 7434 }, { "epoch": 0.07435, "grad_norm": 0.9722703409572532, "learning_rate": 0.003, "loss": 4.0989, "step": 7435 }, { "epoch": 0.07436, "grad_norm": 1.2699630591839492, "learning_rate": 0.003, "loss": 4.1265, "step": 7436 }, { "epoch": 0.07437, "grad_norm": 1.2023886828716306, "learning_rate": 0.003, "loss": 4.0582, "step": 7437 }, { "epoch": 0.07438, "grad_norm": 1.1478584610411458, "learning_rate": 0.003, "loss": 4.1073, "step": 7438 }, { "epoch": 0.07439, "grad_norm": 0.9559788657368175, "learning_rate": 0.003, "loss": 4.0922, "step": 7439 }, { "epoch": 0.0744, "grad_norm": 0.9360959694218371, "learning_rate": 0.003, "loss": 4.0963, "step": 7440 }, { "epoch": 0.07441, "grad_norm": 1.1994542701856334, "learning_rate": 0.003, "loss": 4.1363, "step": 7441 }, { "epoch": 0.07442, "grad_norm": 1.148317105389101, "learning_rate": 0.003, "loss": 4.1122, "step": 7442 }, { "epoch": 0.07443, "grad_norm": 1.1828609361416955, "learning_rate": 0.003, "loss": 4.0958, "step": 7443 }, { "epoch": 0.07444, "grad_norm": 0.9984270524869665, "learning_rate": 0.003, "loss": 4.1146, "step": 7444 }, { "epoch": 0.07445, "grad_norm": 1.1046109735024745, "learning_rate": 0.003, "loss": 4.0991, "step": 7445 }, { "epoch": 0.07446, "grad_norm": 1.077502936625141, "learning_rate": 0.003, "loss": 4.0806, "step": 7446 }, { "epoch": 0.07447, "grad_norm": 1.3996128849618712, "learning_rate": 0.003, "loss": 4.0951, "step": 7447 }, { "epoch": 0.07448, "grad_norm": 1.0287142278411547, "learning_rate": 0.003, "loss": 4.0829, "step": 7448 }, { "epoch": 0.07449, "grad_norm": 1.2427549674708467, "learning_rate": 0.003, "loss": 4.1051, "step": 7449 }, { "epoch": 0.0745, "grad_norm": 0.9870702348058904, "learning_rate": 0.003, "loss": 4.1089, "step": 7450 }, { "epoch": 0.07451, "grad_norm": 1.0952820057543522, "learning_rate": 0.003, "loss": 4.1052, "step": 7451 }, { "epoch": 0.07452, "grad_norm": 1.0656349431590586, "learning_rate": 0.003, "loss": 4.0756, "step": 7452 }, { "epoch": 0.07453, "grad_norm": 1.0968374507716108, "learning_rate": 0.003, "loss": 4.1052, "step": 7453 }, { "epoch": 0.07454, "grad_norm": 1.1404103114047044, "learning_rate": 0.003, "loss": 4.0871, "step": 7454 }, { "epoch": 0.07455, "grad_norm": 1.0359953314455825, "learning_rate": 0.003, "loss": 4.0953, "step": 7455 }, { "epoch": 0.07456, "grad_norm": 1.1415929983014337, "learning_rate": 0.003, "loss": 4.1019, "step": 7456 }, { "epoch": 0.07457, "grad_norm": 1.1474751867174888, "learning_rate": 0.003, "loss": 4.094, "step": 7457 }, { "epoch": 0.07458, "grad_norm": 0.9827108066913763, "learning_rate": 0.003, "loss": 4.1046, "step": 7458 }, { "epoch": 0.07459, "grad_norm": 1.0934454393079214, "learning_rate": 0.003, "loss": 4.0911, "step": 7459 }, { "epoch": 0.0746, "grad_norm": 0.98268777973652, "learning_rate": 0.003, "loss": 4.103, "step": 7460 }, { "epoch": 0.07461, "grad_norm": 0.9816039441879064, "learning_rate": 0.003, "loss": 4.1116, "step": 7461 }, { "epoch": 0.07462, "grad_norm": 1.3051378208274698, "learning_rate": 0.003, "loss": 4.0809, "step": 7462 }, { "epoch": 0.07463, "grad_norm": 1.02310463494236, "learning_rate": 0.003, "loss": 4.1332, "step": 7463 }, { "epoch": 0.07464, "grad_norm": 1.2909266493327525, "learning_rate": 0.003, "loss": 4.0857, "step": 7464 }, { "epoch": 0.07465, "grad_norm": 0.9501490683976144, "learning_rate": 0.003, "loss": 4.104, "step": 7465 }, { "epoch": 0.07466, "grad_norm": 1.088750961064017, "learning_rate": 0.003, "loss": 4.1219, "step": 7466 }, { "epoch": 0.07467, "grad_norm": 1.1372556677278904, "learning_rate": 0.003, "loss": 4.092, "step": 7467 }, { "epoch": 0.07468, "grad_norm": 1.3878690319885936, "learning_rate": 0.003, "loss": 4.1147, "step": 7468 }, { "epoch": 0.07469, "grad_norm": 0.9635630973090923, "learning_rate": 0.003, "loss": 4.0744, "step": 7469 }, { "epoch": 0.0747, "grad_norm": 1.0927607741202252, "learning_rate": 0.003, "loss": 4.1264, "step": 7470 }, { "epoch": 0.07471, "grad_norm": 0.9589717096614562, "learning_rate": 0.003, "loss": 4.1062, "step": 7471 }, { "epoch": 0.07472, "grad_norm": 1.08022929264285, "learning_rate": 0.003, "loss": 4.1101, "step": 7472 }, { "epoch": 0.07473, "grad_norm": 1.1109856773117461, "learning_rate": 0.003, "loss": 4.1156, "step": 7473 }, { "epoch": 0.07474, "grad_norm": 1.1365304830035643, "learning_rate": 0.003, "loss": 4.0991, "step": 7474 }, { "epoch": 0.07475, "grad_norm": 1.2667706944106274, "learning_rate": 0.003, "loss": 4.1157, "step": 7475 }, { "epoch": 0.07476, "grad_norm": 0.9098797976710105, "learning_rate": 0.003, "loss": 4.0769, "step": 7476 }, { "epoch": 0.07477, "grad_norm": 1.0418500545925362, "learning_rate": 0.003, "loss": 4.1134, "step": 7477 }, { "epoch": 0.07478, "grad_norm": 1.086194005911626, "learning_rate": 0.003, "loss": 4.0591, "step": 7478 }, { "epoch": 0.07479, "grad_norm": 1.152475534521128, "learning_rate": 0.003, "loss": 4.0787, "step": 7479 }, { "epoch": 0.0748, "grad_norm": 1.2078785501955323, "learning_rate": 0.003, "loss": 4.0995, "step": 7480 }, { "epoch": 0.07481, "grad_norm": 0.9625364564038152, "learning_rate": 0.003, "loss": 4.1003, "step": 7481 }, { "epoch": 0.07482, "grad_norm": 1.1403385669598296, "learning_rate": 0.003, "loss": 4.1074, "step": 7482 }, { "epoch": 0.07483, "grad_norm": 1.182656884639201, "learning_rate": 0.003, "loss": 4.1327, "step": 7483 }, { "epoch": 0.07484, "grad_norm": 1.0500570511572587, "learning_rate": 0.003, "loss": 4.11, "step": 7484 }, { "epoch": 0.07485, "grad_norm": 1.1939538114667945, "learning_rate": 0.003, "loss": 4.1097, "step": 7485 }, { "epoch": 0.07486, "grad_norm": 0.8599612813420194, "learning_rate": 0.003, "loss": 4.0851, "step": 7486 }, { "epoch": 0.07487, "grad_norm": 0.906249363877033, "learning_rate": 0.003, "loss": 4.1022, "step": 7487 }, { "epoch": 0.07488, "grad_norm": 0.9120721853548994, "learning_rate": 0.003, "loss": 4.1112, "step": 7488 }, { "epoch": 0.07489, "grad_norm": 1.1756362495916781, "learning_rate": 0.003, "loss": 4.1267, "step": 7489 }, { "epoch": 0.0749, "grad_norm": 1.178043554191812, "learning_rate": 0.003, "loss": 4.0967, "step": 7490 }, { "epoch": 0.07491, "grad_norm": 1.177445877528429, "learning_rate": 0.003, "loss": 4.103, "step": 7491 }, { "epoch": 0.07492, "grad_norm": 1.2883563550681214, "learning_rate": 0.003, "loss": 4.1274, "step": 7492 }, { "epoch": 0.07493, "grad_norm": 0.9792550641752528, "learning_rate": 0.003, "loss": 4.1104, "step": 7493 }, { "epoch": 0.07494, "grad_norm": 1.192575626109686, "learning_rate": 0.003, "loss": 4.1253, "step": 7494 }, { "epoch": 0.07495, "grad_norm": 1.109696103067543, "learning_rate": 0.003, "loss": 4.1022, "step": 7495 }, { "epoch": 0.07496, "grad_norm": 1.0076672522580505, "learning_rate": 0.003, "loss": 4.0912, "step": 7496 }, { "epoch": 0.07497, "grad_norm": 1.2073380480227682, "learning_rate": 0.003, "loss": 4.1061, "step": 7497 }, { "epoch": 0.07498, "grad_norm": 1.0463371240066046, "learning_rate": 0.003, "loss": 4.0957, "step": 7498 }, { "epoch": 0.07499, "grad_norm": 1.1632287228456402, "learning_rate": 0.003, "loss": 4.0983, "step": 7499 }, { "epoch": 0.075, "grad_norm": 0.9397310085287326, "learning_rate": 0.003, "loss": 4.1133, "step": 7500 }, { "epoch": 0.07501, "grad_norm": 1.020093818883204, "learning_rate": 0.003, "loss": 4.09, "step": 7501 }, { "epoch": 0.07502, "grad_norm": 1.22722234628339, "learning_rate": 0.003, "loss": 4.0896, "step": 7502 }, { "epoch": 0.07503, "grad_norm": 1.033227705567588, "learning_rate": 0.003, "loss": 4.0876, "step": 7503 }, { "epoch": 0.07504, "grad_norm": 1.171817783999695, "learning_rate": 0.003, "loss": 4.0929, "step": 7504 }, { "epoch": 0.07505, "grad_norm": 1.0270833409022972, "learning_rate": 0.003, "loss": 4.077, "step": 7505 }, { "epoch": 0.07506, "grad_norm": 1.2198086432393653, "learning_rate": 0.003, "loss": 4.0938, "step": 7506 }, { "epoch": 0.07507, "grad_norm": 0.8648598507253091, "learning_rate": 0.003, "loss": 4.0742, "step": 7507 }, { "epoch": 0.07508, "grad_norm": 0.9930152349407657, "learning_rate": 0.003, "loss": 4.1154, "step": 7508 }, { "epoch": 0.07509, "grad_norm": 1.1790407090479906, "learning_rate": 0.003, "loss": 4.1022, "step": 7509 }, { "epoch": 0.0751, "grad_norm": 0.9191788328705034, "learning_rate": 0.003, "loss": 4.0925, "step": 7510 }, { "epoch": 0.07511, "grad_norm": 1.0740498003601395, "learning_rate": 0.003, "loss": 4.0888, "step": 7511 }, { "epoch": 0.07512, "grad_norm": 0.8772452583013857, "learning_rate": 0.003, "loss": 4.1212, "step": 7512 }, { "epoch": 0.07513, "grad_norm": 0.9911439111577278, "learning_rate": 0.003, "loss": 4.1064, "step": 7513 }, { "epoch": 0.07514, "grad_norm": 1.5156680730273393, "learning_rate": 0.003, "loss": 4.0788, "step": 7514 }, { "epoch": 0.07515, "grad_norm": 0.9253063304788648, "learning_rate": 0.003, "loss": 4.1296, "step": 7515 }, { "epoch": 0.07516, "grad_norm": 1.113149970267038, "learning_rate": 0.003, "loss": 4.0841, "step": 7516 }, { "epoch": 0.07517, "grad_norm": 1.243025113042767, "learning_rate": 0.003, "loss": 4.1124, "step": 7517 }, { "epoch": 0.07518, "grad_norm": 0.8855817369758254, "learning_rate": 0.003, "loss": 4.0857, "step": 7518 }, { "epoch": 0.07519, "grad_norm": 0.9112412797077198, "learning_rate": 0.003, "loss": 4.0962, "step": 7519 }, { "epoch": 0.0752, "grad_norm": 1.116396451974829, "learning_rate": 0.003, "loss": 4.1036, "step": 7520 }, { "epoch": 0.07521, "grad_norm": 1.0324392071556105, "learning_rate": 0.003, "loss": 4.1034, "step": 7521 }, { "epoch": 0.07522, "grad_norm": 1.302244987987702, "learning_rate": 0.003, "loss": 4.0955, "step": 7522 }, { "epoch": 0.07523, "grad_norm": 0.9120906340155054, "learning_rate": 0.003, "loss": 4.0841, "step": 7523 }, { "epoch": 0.07524, "grad_norm": 1.1575722806063664, "learning_rate": 0.003, "loss": 4.1188, "step": 7524 }, { "epoch": 0.07525, "grad_norm": 1.273634189592798, "learning_rate": 0.003, "loss": 4.1278, "step": 7525 }, { "epoch": 0.07526, "grad_norm": 0.9217593936350734, "learning_rate": 0.003, "loss": 4.119, "step": 7526 }, { "epoch": 0.07527, "grad_norm": 1.063249795293575, "learning_rate": 0.003, "loss": 4.1213, "step": 7527 }, { "epoch": 0.07528, "grad_norm": 1.0381592400998108, "learning_rate": 0.003, "loss": 4.1257, "step": 7528 }, { "epoch": 0.07529, "grad_norm": 1.0503623433090992, "learning_rate": 0.003, "loss": 4.0929, "step": 7529 }, { "epoch": 0.0753, "grad_norm": 1.0575571009050915, "learning_rate": 0.003, "loss": 4.0926, "step": 7530 }, { "epoch": 0.07531, "grad_norm": 1.0284966628634444, "learning_rate": 0.003, "loss": 4.1221, "step": 7531 }, { "epoch": 0.07532, "grad_norm": 1.017348507432173, "learning_rate": 0.003, "loss": 4.1089, "step": 7532 }, { "epoch": 0.07533, "grad_norm": 1.228626600553027, "learning_rate": 0.003, "loss": 4.1019, "step": 7533 }, { "epoch": 0.07534, "grad_norm": 1.1401992033872428, "learning_rate": 0.003, "loss": 4.1035, "step": 7534 }, { "epoch": 0.07535, "grad_norm": 1.1630915222125624, "learning_rate": 0.003, "loss": 4.1133, "step": 7535 }, { "epoch": 0.07536, "grad_norm": 1.012891605331372, "learning_rate": 0.003, "loss": 4.0905, "step": 7536 }, { "epoch": 0.07537, "grad_norm": 1.2029828166327519, "learning_rate": 0.003, "loss": 4.0847, "step": 7537 }, { "epoch": 0.07538, "grad_norm": 1.0177281645657348, "learning_rate": 0.003, "loss": 4.1002, "step": 7538 }, { "epoch": 0.07539, "grad_norm": 1.2772627744588205, "learning_rate": 0.003, "loss": 4.0799, "step": 7539 }, { "epoch": 0.0754, "grad_norm": 0.9294525785897821, "learning_rate": 0.003, "loss": 4.0941, "step": 7540 }, { "epoch": 0.07541, "grad_norm": 1.0828336834647083, "learning_rate": 0.003, "loss": 4.0761, "step": 7541 }, { "epoch": 0.07542, "grad_norm": 1.1780063549119208, "learning_rate": 0.003, "loss": 4.0997, "step": 7542 }, { "epoch": 0.07543, "grad_norm": 1.0749429588406836, "learning_rate": 0.003, "loss": 4.1152, "step": 7543 }, { "epoch": 0.07544, "grad_norm": 1.0841966667755707, "learning_rate": 0.003, "loss": 4.1097, "step": 7544 }, { "epoch": 0.07545, "grad_norm": 0.9980448156387751, "learning_rate": 0.003, "loss": 4.1343, "step": 7545 }, { "epoch": 0.07546, "grad_norm": 1.2664949065709301, "learning_rate": 0.003, "loss": 4.098, "step": 7546 }, { "epoch": 0.07547, "grad_norm": 1.062678674841793, "learning_rate": 0.003, "loss": 4.112, "step": 7547 }, { "epoch": 0.07548, "grad_norm": 1.5159617863041532, "learning_rate": 0.003, "loss": 4.1355, "step": 7548 }, { "epoch": 0.07549, "grad_norm": 0.9217954430172313, "learning_rate": 0.003, "loss": 4.1264, "step": 7549 }, { "epoch": 0.0755, "grad_norm": 0.9469672042958677, "learning_rate": 0.003, "loss": 4.1245, "step": 7550 }, { "epoch": 0.07551, "grad_norm": 1.1010649743264453, "learning_rate": 0.003, "loss": 4.1158, "step": 7551 }, { "epoch": 0.07552, "grad_norm": 0.979006740783685, "learning_rate": 0.003, "loss": 4.1151, "step": 7552 }, { "epoch": 0.07553, "grad_norm": 1.2202632658315713, "learning_rate": 0.003, "loss": 4.1234, "step": 7553 }, { "epoch": 0.07554, "grad_norm": 1.0141175000304337, "learning_rate": 0.003, "loss": 4.1183, "step": 7554 }, { "epoch": 0.07555, "grad_norm": 1.2381367341566416, "learning_rate": 0.003, "loss": 4.1133, "step": 7555 }, { "epoch": 0.07556, "grad_norm": 1.1130694211087335, "learning_rate": 0.003, "loss": 4.1263, "step": 7556 }, { "epoch": 0.07557, "grad_norm": 1.220415386983482, "learning_rate": 0.003, "loss": 4.1006, "step": 7557 }, { "epoch": 0.07558, "grad_norm": 0.9103727107756383, "learning_rate": 0.003, "loss": 4.0969, "step": 7558 }, { "epoch": 0.07559, "grad_norm": 1.161068539868024, "learning_rate": 0.003, "loss": 4.0679, "step": 7559 }, { "epoch": 0.0756, "grad_norm": 1.2157119525881248, "learning_rate": 0.003, "loss": 4.1117, "step": 7560 }, { "epoch": 0.07561, "grad_norm": 0.8483939083465544, "learning_rate": 0.003, "loss": 4.0913, "step": 7561 }, { "epoch": 0.07562, "grad_norm": 0.8203601465351547, "learning_rate": 0.003, "loss": 4.0687, "step": 7562 }, { "epoch": 0.07563, "grad_norm": 1.0211935844350564, "learning_rate": 0.003, "loss": 4.1023, "step": 7563 }, { "epoch": 0.07564, "grad_norm": 1.243412295176371, "learning_rate": 0.003, "loss": 4.1278, "step": 7564 }, { "epoch": 0.07565, "grad_norm": 0.8572336614367662, "learning_rate": 0.003, "loss": 4.1014, "step": 7565 }, { "epoch": 0.07566, "grad_norm": 0.9904342760557475, "learning_rate": 0.003, "loss": 4.0994, "step": 7566 }, { "epoch": 0.07567, "grad_norm": 1.3580247317393612, "learning_rate": 0.003, "loss": 4.1183, "step": 7567 }, { "epoch": 0.07568, "grad_norm": 0.9605002083972642, "learning_rate": 0.003, "loss": 4.1087, "step": 7568 }, { "epoch": 0.07569, "grad_norm": 1.1090261834271542, "learning_rate": 0.003, "loss": 4.0884, "step": 7569 }, { "epoch": 0.0757, "grad_norm": 1.1287563780825858, "learning_rate": 0.003, "loss": 4.0891, "step": 7570 }, { "epoch": 0.07571, "grad_norm": 1.1003145238464291, "learning_rate": 0.003, "loss": 4.0818, "step": 7571 }, { "epoch": 0.07572, "grad_norm": 1.1635794583045487, "learning_rate": 0.003, "loss": 4.0822, "step": 7572 }, { "epoch": 0.07573, "grad_norm": 1.1124514629320816, "learning_rate": 0.003, "loss": 4.0987, "step": 7573 }, { "epoch": 0.07574, "grad_norm": 1.1040878656253004, "learning_rate": 0.003, "loss": 4.1126, "step": 7574 }, { "epoch": 0.07575, "grad_norm": 1.1860641536671122, "learning_rate": 0.003, "loss": 4.0933, "step": 7575 }, { "epoch": 0.07576, "grad_norm": 0.9866929732016607, "learning_rate": 0.003, "loss": 4.0861, "step": 7576 }, { "epoch": 0.07577, "grad_norm": 1.176366123038857, "learning_rate": 0.003, "loss": 4.1224, "step": 7577 }, { "epoch": 0.07578, "grad_norm": 0.9002821233286987, "learning_rate": 0.003, "loss": 4.1016, "step": 7578 }, { "epoch": 0.07579, "grad_norm": 1.0018491936918001, "learning_rate": 0.003, "loss": 4.0885, "step": 7579 }, { "epoch": 0.0758, "grad_norm": 1.1996191726980248, "learning_rate": 0.003, "loss": 4.1172, "step": 7580 }, { "epoch": 0.07581, "grad_norm": 1.0390536601921203, "learning_rate": 0.003, "loss": 4.0985, "step": 7581 }, { "epoch": 0.07582, "grad_norm": 1.3008748412907003, "learning_rate": 0.003, "loss": 4.1021, "step": 7582 }, { "epoch": 0.07583, "grad_norm": 0.9378232162415077, "learning_rate": 0.003, "loss": 4.0941, "step": 7583 }, { "epoch": 0.07584, "grad_norm": 0.9365440192341151, "learning_rate": 0.003, "loss": 4.138, "step": 7584 }, { "epoch": 0.07585, "grad_norm": 1.164036895229275, "learning_rate": 0.003, "loss": 4.1082, "step": 7585 }, { "epoch": 0.07586, "grad_norm": 1.0329782543130193, "learning_rate": 0.003, "loss": 4.0862, "step": 7586 }, { "epoch": 0.07587, "grad_norm": 1.377018066910138, "learning_rate": 0.003, "loss": 4.1337, "step": 7587 }, { "epoch": 0.07588, "grad_norm": 1.051480991233041, "learning_rate": 0.003, "loss": 4.1009, "step": 7588 }, { "epoch": 0.07589, "grad_norm": 1.0376720795682441, "learning_rate": 0.003, "loss": 4.1068, "step": 7589 }, { "epoch": 0.0759, "grad_norm": 1.220277986042294, "learning_rate": 0.003, "loss": 4.1093, "step": 7590 }, { "epoch": 0.07591, "grad_norm": 1.0320383144822298, "learning_rate": 0.003, "loss": 4.111, "step": 7591 }, { "epoch": 0.07592, "grad_norm": 1.113224947516195, "learning_rate": 0.003, "loss": 4.1142, "step": 7592 }, { "epoch": 0.07593, "grad_norm": 1.0193178300169947, "learning_rate": 0.003, "loss": 4.0811, "step": 7593 }, { "epoch": 0.07594, "grad_norm": 1.0127757135657491, "learning_rate": 0.003, "loss": 4.0829, "step": 7594 }, { "epoch": 0.07595, "grad_norm": 1.2522485470517928, "learning_rate": 0.003, "loss": 4.0884, "step": 7595 }, { "epoch": 0.07596, "grad_norm": 0.8461175020076812, "learning_rate": 0.003, "loss": 4.0717, "step": 7596 }, { "epoch": 0.07597, "grad_norm": 1.0614723159182662, "learning_rate": 0.003, "loss": 4.116, "step": 7597 }, { "epoch": 0.07598, "grad_norm": 1.2432661985382218, "learning_rate": 0.003, "loss": 4.1142, "step": 7598 }, { "epoch": 0.07599, "grad_norm": 0.9724406876800284, "learning_rate": 0.003, "loss": 4.1121, "step": 7599 }, { "epoch": 0.076, "grad_norm": 1.0294698539571896, "learning_rate": 0.003, "loss": 4.1005, "step": 7600 }, { "epoch": 0.07601, "grad_norm": 1.2075561785390323, "learning_rate": 0.003, "loss": 4.1236, "step": 7601 }, { "epoch": 0.07602, "grad_norm": 0.908367171679946, "learning_rate": 0.003, "loss": 4.0928, "step": 7602 }, { "epoch": 0.07603, "grad_norm": 0.819105434643421, "learning_rate": 0.003, "loss": 4.0926, "step": 7603 }, { "epoch": 0.07604, "grad_norm": 0.940096955111327, "learning_rate": 0.003, "loss": 4.1097, "step": 7604 }, { "epoch": 0.07605, "grad_norm": 1.1869813844923396, "learning_rate": 0.003, "loss": 4.1054, "step": 7605 }, { "epoch": 0.07606, "grad_norm": 1.319999351024039, "learning_rate": 0.003, "loss": 4.1612, "step": 7606 }, { "epoch": 0.07607, "grad_norm": 0.9928524946267742, "learning_rate": 0.003, "loss": 4.1131, "step": 7607 }, { "epoch": 0.07608, "grad_norm": 1.2163879425361146, "learning_rate": 0.003, "loss": 4.1334, "step": 7608 }, { "epoch": 0.07609, "grad_norm": 1.0931474918077035, "learning_rate": 0.003, "loss": 4.1031, "step": 7609 }, { "epoch": 0.0761, "grad_norm": 1.2939128848551429, "learning_rate": 0.003, "loss": 4.102, "step": 7610 }, { "epoch": 0.07611, "grad_norm": 0.9268786250276437, "learning_rate": 0.003, "loss": 4.1245, "step": 7611 }, { "epoch": 0.07612, "grad_norm": 1.09828222716354, "learning_rate": 0.003, "loss": 4.0928, "step": 7612 }, { "epoch": 0.07613, "grad_norm": 1.1664511437430545, "learning_rate": 0.003, "loss": 4.0886, "step": 7613 }, { "epoch": 0.07614, "grad_norm": 1.0431360064756068, "learning_rate": 0.003, "loss": 4.0734, "step": 7614 }, { "epoch": 0.07615, "grad_norm": 1.4227723083965413, "learning_rate": 0.003, "loss": 4.1075, "step": 7615 }, { "epoch": 0.07616, "grad_norm": 0.8909517959568098, "learning_rate": 0.003, "loss": 4.1002, "step": 7616 }, { "epoch": 0.07617, "grad_norm": 0.8509788287430593, "learning_rate": 0.003, "loss": 4.0625, "step": 7617 }, { "epoch": 0.07618, "grad_norm": 1.1167760380486436, "learning_rate": 0.003, "loss": 4.1378, "step": 7618 }, { "epoch": 0.07619, "grad_norm": 1.11193937966075, "learning_rate": 0.003, "loss": 4.1128, "step": 7619 }, { "epoch": 0.0762, "grad_norm": 1.2309465997063567, "learning_rate": 0.003, "loss": 4.0888, "step": 7620 }, { "epoch": 0.07621, "grad_norm": 0.8811305994875692, "learning_rate": 0.003, "loss": 4.1161, "step": 7621 }, { "epoch": 0.07622, "grad_norm": 0.9686305823838591, "learning_rate": 0.003, "loss": 4.1151, "step": 7622 }, { "epoch": 0.07623, "grad_norm": 0.9243277071622491, "learning_rate": 0.003, "loss": 4.0935, "step": 7623 }, { "epoch": 0.07624, "grad_norm": 1.191977900320886, "learning_rate": 0.003, "loss": 4.1062, "step": 7624 }, { "epoch": 0.07625, "grad_norm": 1.026341609349241, "learning_rate": 0.003, "loss": 4.0857, "step": 7625 }, { "epoch": 0.07626, "grad_norm": 1.711585880657544, "learning_rate": 0.003, "loss": 4.114, "step": 7626 }, { "epoch": 0.07627, "grad_norm": 1.1367491841820259, "learning_rate": 0.003, "loss": 4.117, "step": 7627 }, { "epoch": 0.07628, "grad_norm": 0.988369876898564, "learning_rate": 0.003, "loss": 4.0776, "step": 7628 }, { "epoch": 0.07629, "grad_norm": 1.121398232821057, "learning_rate": 0.003, "loss": 4.1029, "step": 7629 }, { "epoch": 0.0763, "grad_norm": 1.0136207435839946, "learning_rate": 0.003, "loss": 4.0856, "step": 7630 }, { "epoch": 0.07631, "grad_norm": 1.4267741493920876, "learning_rate": 0.003, "loss": 4.0994, "step": 7631 }, { "epoch": 0.07632, "grad_norm": 1.1159763620561827, "learning_rate": 0.003, "loss": 4.0859, "step": 7632 }, { "epoch": 0.07633, "grad_norm": 1.3484061208170044, "learning_rate": 0.003, "loss": 4.0975, "step": 7633 }, { "epoch": 0.07634, "grad_norm": 0.9621844147865416, "learning_rate": 0.003, "loss": 4.0814, "step": 7634 }, { "epoch": 0.07635, "grad_norm": 0.9183557192199527, "learning_rate": 0.003, "loss": 4.1015, "step": 7635 }, { "epoch": 0.07636, "grad_norm": 0.9807300765346603, "learning_rate": 0.003, "loss": 4.081, "step": 7636 }, { "epoch": 0.07637, "grad_norm": 0.987741492699573, "learning_rate": 0.003, "loss": 4.1195, "step": 7637 }, { "epoch": 0.07638, "grad_norm": 1.150598587756748, "learning_rate": 0.003, "loss": 4.1018, "step": 7638 }, { "epoch": 0.07639, "grad_norm": 1.1008913618099967, "learning_rate": 0.003, "loss": 4.1065, "step": 7639 }, { "epoch": 0.0764, "grad_norm": 1.069231691804542, "learning_rate": 0.003, "loss": 4.1348, "step": 7640 }, { "epoch": 0.07641, "grad_norm": 1.0412024223182812, "learning_rate": 0.003, "loss": 4.1288, "step": 7641 }, { "epoch": 0.07642, "grad_norm": 1.042062368982713, "learning_rate": 0.003, "loss": 4.1208, "step": 7642 }, { "epoch": 0.07643, "grad_norm": 1.2951167861590105, "learning_rate": 0.003, "loss": 4.1075, "step": 7643 }, { "epoch": 0.07644, "grad_norm": 1.0584656072218666, "learning_rate": 0.003, "loss": 4.0868, "step": 7644 }, { "epoch": 0.07645, "grad_norm": 1.072877867189409, "learning_rate": 0.003, "loss": 4.1025, "step": 7645 }, { "epoch": 0.07646, "grad_norm": 1.3048350731050382, "learning_rate": 0.003, "loss": 4.1072, "step": 7646 }, { "epoch": 0.07647, "grad_norm": 1.1612332984131792, "learning_rate": 0.003, "loss": 4.0847, "step": 7647 }, { "epoch": 0.07648, "grad_norm": 0.8560991825716354, "learning_rate": 0.003, "loss": 4.1004, "step": 7648 }, { "epoch": 0.07649, "grad_norm": 0.9881761378783187, "learning_rate": 0.003, "loss": 4.0996, "step": 7649 }, { "epoch": 0.0765, "grad_norm": 1.363753027922075, "learning_rate": 0.003, "loss": 4.1001, "step": 7650 }, { "epoch": 0.07651, "grad_norm": 0.8946406952460046, "learning_rate": 0.003, "loss": 4.0954, "step": 7651 }, { "epoch": 0.07652, "grad_norm": 0.9508520934841937, "learning_rate": 0.003, "loss": 4.0921, "step": 7652 }, { "epoch": 0.07653, "grad_norm": 1.1285258062605226, "learning_rate": 0.003, "loss": 4.0788, "step": 7653 }, { "epoch": 0.07654, "grad_norm": 1.066906075262894, "learning_rate": 0.003, "loss": 4.1045, "step": 7654 }, { "epoch": 0.07655, "grad_norm": 1.1447845805737322, "learning_rate": 0.003, "loss": 4.1133, "step": 7655 }, { "epoch": 0.07656, "grad_norm": 1.0026793162461074, "learning_rate": 0.003, "loss": 4.1133, "step": 7656 }, { "epoch": 0.07657, "grad_norm": 1.253277049833007, "learning_rate": 0.003, "loss": 4.1232, "step": 7657 }, { "epoch": 0.07658, "grad_norm": 0.9868834909744868, "learning_rate": 0.003, "loss": 4.1017, "step": 7658 }, { "epoch": 0.07659, "grad_norm": 1.3930524753298466, "learning_rate": 0.003, "loss": 4.1023, "step": 7659 }, { "epoch": 0.0766, "grad_norm": 1.147552881069107, "learning_rate": 0.003, "loss": 4.1419, "step": 7660 }, { "epoch": 0.07661, "grad_norm": 1.097273792838392, "learning_rate": 0.003, "loss": 4.0989, "step": 7661 }, { "epoch": 0.07662, "grad_norm": 0.9919267467777996, "learning_rate": 0.003, "loss": 4.1246, "step": 7662 }, { "epoch": 0.07663, "grad_norm": 0.9604902804967356, "learning_rate": 0.003, "loss": 4.117, "step": 7663 }, { "epoch": 0.07664, "grad_norm": 0.9241313914036474, "learning_rate": 0.003, "loss": 4.1069, "step": 7664 }, { "epoch": 0.07665, "grad_norm": 1.104796598248223, "learning_rate": 0.003, "loss": 4.0834, "step": 7665 }, { "epoch": 0.07666, "grad_norm": 1.3127029491687616, "learning_rate": 0.003, "loss": 4.1062, "step": 7666 }, { "epoch": 0.07667, "grad_norm": 1.2964036322532557, "learning_rate": 0.003, "loss": 4.1132, "step": 7667 }, { "epoch": 0.07668, "grad_norm": 0.9580286300224677, "learning_rate": 0.003, "loss": 4.1004, "step": 7668 }, { "epoch": 0.07669, "grad_norm": 0.9745998727959247, "learning_rate": 0.003, "loss": 4.1129, "step": 7669 }, { "epoch": 0.0767, "grad_norm": 1.0011917702341462, "learning_rate": 0.003, "loss": 4.0783, "step": 7670 }, { "epoch": 0.07671, "grad_norm": 1.0199987802111787, "learning_rate": 0.003, "loss": 4.1162, "step": 7671 }, { "epoch": 0.07672, "grad_norm": 0.9129909978014048, "learning_rate": 0.003, "loss": 4.0975, "step": 7672 }, { "epoch": 0.07673, "grad_norm": 1.0607294077470495, "learning_rate": 0.003, "loss": 4.1345, "step": 7673 }, { "epoch": 0.07674, "grad_norm": 1.0920171975344224, "learning_rate": 0.003, "loss": 4.0999, "step": 7674 }, { "epoch": 0.07675, "grad_norm": 1.0565727558971971, "learning_rate": 0.003, "loss": 4.0919, "step": 7675 }, { "epoch": 0.07676, "grad_norm": 1.2725834956197855, "learning_rate": 0.003, "loss": 4.096, "step": 7676 }, { "epoch": 0.07677, "grad_norm": 1.2598694460071018, "learning_rate": 0.003, "loss": 4.0986, "step": 7677 }, { "epoch": 0.07678, "grad_norm": 1.325358583295913, "learning_rate": 0.003, "loss": 4.1318, "step": 7678 }, { "epoch": 0.07679, "grad_norm": 0.9349674003951666, "learning_rate": 0.003, "loss": 4.1194, "step": 7679 }, { "epoch": 0.0768, "grad_norm": 1.0596741567867978, "learning_rate": 0.003, "loss": 4.0915, "step": 7680 }, { "epoch": 0.07681, "grad_norm": 1.1277581537166605, "learning_rate": 0.003, "loss": 4.105, "step": 7681 }, { "epoch": 0.07682, "grad_norm": 1.083698074015161, "learning_rate": 0.003, "loss": 4.0908, "step": 7682 }, { "epoch": 0.07683, "grad_norm": 0.9949340311875353, "learning_rate": 0.003, "loss": 4.104, "step": 7683 }, { "epoch": 0.07684, "grad_norm": 1.053359494756575, "learning_rate": 0.003, "loss": 4.1183, "step": 7684 }, { "epoch": 0.07685, "grad_norm": 1.0752188724225478, "learning_rate": 0.003, "loss": 4.0955, "step": 7685 }, { "epoch": 0.07686, "grad_norm": 0.9963305559685445, "learning_rate": 0.003, "loss": 4.0966, "step": 7686 }, { "epoch": 0.07687, "grad_norm": 1.1200067453730849, "learning_rate": 0.003, "loss": 4.1103, "step": 7687 }, { "epoch": 0.07688, "grad_norm": 1.4005284389753019, "learning_rate": 0.003, "loss": 4.1095, "step": 7688 }, { "epoch": 0.07689, "grad_norm": 0.9511902260435283, "learning_rate": 0.003, "loss": 4.0848, "step": 7689 }, { "epoch": 0.0769, "grad_norm": 1.1187995671857418, "learning_rate": 0.003, "loss": 4.1174, "step": 7690 }, { "epoch": 0.07691, "grad_norm": 1.2231215859628386, "learning_rate": 0.003, "loss": 4.1125, "step": 7691 }, { "epoch": 0.07692, "grad_norm": 0.8299137991876565, "learning_rate": 0.003, "loss": 4.1122, "step": 7692 }, { "epoch": 0.07693, "grad_norm": 0.9082081822910427, "learning_rate": 0.003, "loss": 4.0875, "step": 7693 }, { "epoch": 0.07694, "grad_norm": 1.0916728801306739, "learning_rate": 0.003, "loss": 4.0936, "step": 7694 }, { "epoch": 0.07695, "grad_norm": 0.9838815880287163, "learning_rate": 0.003, "loss": 4.1138, "step": 7695 }, { "epoch": 0.07696, "grad_norm": 1.155813106601287, "learning_rate": 0.003, "loss": 4.052, "step": 7696 }, { "epoch": 0.07697, "grad_norm": 1.0839130165247974, "learning_rate": 0.003, "loss": 4.131, "step": 7697 }, { "epoch": 0.07698, "grad_norm": 1.1680104322789127, "learning_rate": 0.003, "loss": 4.1076, "step": 7698 }, { "epoch": 0.07699, "grad_norm": 1.0447890613446116, "learning_rate": 0.003, "loss": 4.1322, "step": 7699 }, { "epoch": 0.077, "grad_norm": 1.0704008306301402, "learning_rate": 0.003, "loss": 4.1059, "step": 7700 }, { "epoch": 0.07701, "grad_norm": 1.202553730312753, "learning_rate": 0.003, "loss": 4.0922, "step": 7701 }, { "epoch": 0.07702, "grad_norm": 1.0984179830351732, "learning_rate": 0.003, "loss": 4.092, "step": 7702 }, { "epoch": 0.07703, "grad_norm": 1.0470287328541912, "learning_rate": 0.003, "loss": 4.1061, "step": 7703 }, { "epoch": 0.07704, "grad_norm": 1.1097420437759358, "learning_rate": 0.003, "loss": 4.1083, "step": 7704 }, { "epoch": 0.07705, "grad_norm": 1.1144201249659018, "learning_rate": 0.003, "loss": 4.0905, "step": 7705 }, { "epoch": 0.07706, "grad_norm": 1.0217698112069022, "learning_rate": 0.003, "loss": 4.1088, "step": 7706 }, { "epoch": 0.07707, "grad_norm": 1.2109348836998135, "learning_rate": 0.003, "loss": 4.1099, "step": 7707 }, { "epoch": 0.07708, "grad_norm": 1.0553701436633687, "learning_rate": 0.003, "loss": 4.1202, "step": 7708 }, { "epoch": 0.07709, "grad_norm": 1.0487666289699271, "learning_rate": 0.003, "loss": 4.1068, "step": 7709 }, { "epoch": 0.0771, "grad_norm": 1.2192442347910983, "learning_rate": 0.003, "loss": 4.0903, "step": 7710 }, { "epoch": 0.07711, "grad_norm": 0.8886343142082175, "learning_rate": 0.003, "loss": 4.0881, "step": 7711 }, { "epoch": 0.07712, "grad_norm": 1.097685818072634, "learning_rate": 0.003, "loss": 4.1002, "step": 7712 }, { "epoch": 0.07713, "grad_norm": 1.0647709451424068, "learning_rate": 0.003, "loss": 4.0759, "step": 7713 }, { "epoch": 0.07714, "grad_norm": 1.1091271113459429, "learning_rate": 0.003, "loss": 4.0771, "step": 7714 }, { "epoch": 0.07715, "grad_norm": 1.1833117992139233, "learning_rate": 0.003, "loss": 4.1048, "step": 7715 }, { "epoch": 0.07716, "grad_norm": 1.3072350917328062, "learning_rate": 0.003, "loss": 4.089, "step": 7716 }, { "epoch": 0.07717, "grad_norm": 1.1929125586163642, "learning_rate": 0.003, "loss": 4.0875, "step": 7717 }, { "epoch": 0.07718, "grad_norm": 0.9794043684657727, "learning_rate": 0.003, "loss": 4.1233, "step": 7718 }, { "epoch": 0.07719, "grad_norm": 0.975112457349595, "learning_rate": 0.003, "loss": 4.0922, "step": 7719 }, { "epoch": 0.0772, "grad_norm": 1.1897527208548928, "learning_rate": 0.003, "loss": 4.1289, "step": 7720 }, { "epoch": 0.07721, "grad_norm": 0.9330747049002993, "learning_rate": 0.003, "loss": 4.083, "step": 7721 }, { "epoch": 0.07722, "grad_norm": 1.1116995201555617, "learning_rate": 0.003, "loss": 4.1187, "step": 7722 }, { "epoch": 0.07723, "grad_norm": 0.9908749389404667, "learning_rate": 0.003, "loss": 4.1192, "step": 7723 }, { "epoch": 0.07724, "grad_norm": 1.3797674628535543, "learning_rate": 0.003, "loss": 4.1102, "step": 7724 }, { "epoch": 0.07725, "grad_norm": 0.8365626531454406, "learning_rate": 0.003, "loss": 4.0891, "step": 7725 }, { "epoch": 0.07726, "grad_norm": 0.8890967555331794, "learning_rate": 0.003, "loss": 4.116, "step": 7726 }, { "epoch": 0.07727, "grad_norm": 1.0153125391727966, "learning_rate": 0.003, "loss": 4.0797, "step": 7727 }, { "epoch": 0.07728, "grad_norm": 1.2781185982208176, "learning_rate": 0.003, "loss": 4.0914, "step": 7728 }, { "epoch": 0.07729, "grad_norm": 1.3348745122018035, "learning_rate": 0.003, "loss": 4.1172, "step": 7729 }, { "epoch": 0.0773, "grad_norm": 1.1346003292608733, "learning_rate": 0.003, "loss": 4.1084, "step": 7730 }, { "epoch": 0.07731, "grad_norm": 1.1382685641315156, "learning_rate": 0.003, "loss": 4.0985, "step": 7731 }, { "epoch": 0.07732, "grad_norm": 0.909477352989243, "learning_rate": 0.003, "loss": 4.1025, "step": 7732 }, { "epoch": 0.07733, "grad_norm": 0.9380438060395494, "learning_rate": 0.003, "loss": 4.1062, "step": 7733 }, { "epoch": 0.07734, "grad_norm": 1.182723768892894, "learning_rate": 0.003, "loss": 4.1143, "step": 7734 }, { "epoch": 0.07735, "grad_norm": 0.956856860551835, "learning_rate": 0.003, "loss": 4.0829, "step": 7735 }, { "epoch": 0.07736, "grad_norm": 1.2503508123110225, "learning_rate": 0.003, "loss": 4.1132, "step": 7736 }, { "epoch": 0.07737, "grad_norm": 0.9753617896147677, "learning_rate": 0.003, "loss": 4.072, "step": 7737 }, { "epoch": 0.07738, "grad_norm": 1.165817507529701, "learning_rate": 0.003, "loss": 4.0965, "step": 7738 }, { "epoch": 0.07739, "grad_norm": 1.0658659573641052, "learning_rate": 0.003, "loss": 4.0986, "step": 7739 }, { "epoch": 0.0774, "grad_norm": 1.3038985858407492, "learning_rate": 0.003, "loss": 4.0859, "step": 7740 }, { "epoch": 0.07741, "grad_norm": 0.8212293829001464, "learning_rate": 0.003, "loss": 4.0823, "step": 7741 }, { "epoch": 0.07742, "grad_norm": 0.8961041116710575, "learning_rate": 0.003, "loss": 4.0738, "step": 7742 }, { "epoch": 0.07743, "grad_norm": 1.078993168136095, "learning_rate": 0.003, "loss": 4.0952, "step": 7743 }, { "epoch": 0.07744, "grad_norm": 1.2392253006965184, "learning_rate": 0.003, "loss": 4.0969, "step": 7744 }, { "epoch": 0.07745, "grad_norm": 1.092921186916631, "learning_rate": 0.003, "loss": 4.0961, "step": 7745 }, { "epoch": 0.07746, "grad_norm": 0.9498762317732079, "learning_rate": 0.003, "loss": 4.0989, "step": 7746 }, { "epoch": 0.07747, "grad_norm": 1.0777857631437682, "learning_rate": 0.003, "loss": 4.1033, "step": 7747 }, { "epoch": 0.07748, "grad_norm": 1.2501420325170711, "learning_rate": 0.003, "loss": 4.131, "step": 7748 }, { "epoch": 0.07749, "grad_norm": 1.1659362799933979, "learning_rate": 0.003, "loss": 4.1228, "step": 7749 }, { "epoch": 0.0775, "grad_norm": 1.0057776955954294, "learning_rate": 0.003, "loss": 4.0802, "step": 7750 }, { "epoch": 0.07751, "grad_norm": 1.0908977077309605, "learning_rate": 0.003, "loss": 4.0991, "step": 7751 }, { "epoch": 0.07752, "grad_norm": 1.0146076688175378, "learning_rate": 0.003, "loss": 4.1126, "step": 7752 }, { "epoch": 0.07753, "grad_norm": 1.363739323470198, "learning_rate": 0.003, "loss": 4.0897, "step": 7753 }, { "epoch": 0.07754, "grad_norm": 0.9532027349687967, "learning_rate": 0.003, "loss": 4.0945, "step": 7754 }, { "epoch": 0.07755, "grad_norm": 1.2117236044034614, "learning_rate": 0.003, "loss": 4.1049, "step": 7755 }, { "epoch": 0.07756, "grad_norm": 0.9695361568369883, "learning_rate": 0.003, "loss": 4.115, "step": 7756 }, { "epoch": 0.07757, "grad_norm": 0.9161470471426952, "learning_rate": 0.003, "loss": 4.0803, "step": 7757 }, { "epoch": 0.07758, "grad_norm": 1.048139217552709, "learning_rate": 0.003, "loss": 4.0824, "step": 7758 }, { "epoch": 0.07759, "grad_norm": 1.0718445202257145, "learning_rate": 0.003, "loss": 4.092, "step": 7759 }, { "epoch": 0.0776, "grad_norm": 1.0941281483649858, "learning_rate": 0.003, "loss": 4.075, "step": 7760 }, { "epoch": 0.07761, "grad_norm": 1.16599825792614, "learning_rate": 0.003, "loss": 4.1191, "step": 7761 }, { "epoch": 0.07762, "grad_norm": 1.1135150780281224, "learning_rate": 0.003, "loss": 4.0976, "step": 7762 }, { "epoch": 0.07763, "grad_norm": 1.2834428160255196, "learning_rate": 0.003, "loss": 4.1149, "step": 7763 }, { "epoch": 0.07764, "grad_norm": 1.0594672168271608, "learning_rate": 0.003, "loss": 4.1218, "step": 7764 }, { "epoch": 0.07765, "grad_norm": 1.1099903530222162, "learning_rate": 0.003, "loss": 4.1017, "step": 7765 }, { "epoch": 0.07766, "grad_norm": 1.1139847548103408, "learning_rate": 0.003, "loss": 4.0855, "step": 7766 }, { "epoch": 0.07767, "grad_norm": 1.0721664614212691, "learning_rate": 0.003, "loss": 4.0832, "step": 7767 }, { "epoch": 0.07768, "grad_norm": 1.1111257160722119, "learning_rate": 0.003, "loss": 4.0945, "step": 7768 }, { "epoch": 0.07769, "grad_norm": 1.3832464631797665, "learning_rate": 0.003, "loss": 4.1056, "step": 7769 }, { "epoch": 0.0777, "grad_norm": 0.843477392576227, "learning_rate": 0.003, "loss": 4.1105, "step": 7770 }, { "epoch": 0.07771, "grad_norm": 0.8510362431061711, "learning_rate": 0.003, "loss": 4.1101, "step": 7771 }, { "epoch": 0.07772, "grad_norm": 0.8897772192499654, "learning_rate": 0.003, "loss": 4.1006, "step": 7772 }, { "epoch": 0.07773, "grad_norm": 1.013378086803394, "learning_rate": 0.003, "loss": 4.0704, "step": 7773 }, { "epoch": 0.07774, "grad_norm": 1.0168273891280863, "learning_rate": 0.003, "loss": 4.1001, "step": 7774 }, { "epoch": 0.07775, "grad_norm": 1.072622212301239, "learning_rate": 0.003, "loss": 4.1108, "step": 7775 }, { "epoch": 0.07776, "grad_norm": 1.0581485073003125, "learning_rate": 0.003, "loss": 4.0966, "step": 7776 }, { "epoch": 0.07777, "grad_norm": 1.1302210145024325, "learning_rate": 0.003, "loss": 4.1173, "step": 7777 }, { "epoch": 0.07778, "grad_norm": 1.024976398839125, "learning_rate": 0.003, "loss": 4.1171, "step": 7778 }, { "epoch": 0.07779, "grad_norm": 1.1717175996834288, "learning_rate": 0.003, "loss": 4.0828, "step": 7779 }, { "epoch": 0.0778, "grad_norm": 1.427357253672805, "learning_rate": 0.003, "loss": 4.0809, "step": 7780 }, { "epoch": 0.07781, "grad_norm": 1.0969858418068819, "learning_rate": 0.003, "loss": 4.0667, "step": 7781 }, { "epoch": 0.07782, "grad_norm": 1.2968713137185408, "learning_rate": 0.003, "loss": 4.0909, "step": 7782 }, { "epoch": 0.07783, "grad_norm": 0.8773486594119719, "learning_rate": 0.003, "loss": 4.0899, "step": 7783 }, { "epoch": 0.07784, "grad_norm": 1.0342548469691255, "learning_rate": 0.003, "loss": 4.0903, "step": 7784 }, { "epoch": 0.07785, "grad_norm": 1.10039126806273, "learning_rate": 0.003, "loss": 4.1206, "step": 7785 }, { "epoch": 0.07786, "grad_norm": 1.0452193011897093, "learning_rate": 0.003, "loss": 4.0851, "step": 7786 }, { "epoch": 0.07787, "grad_norm": 1.1642092124226817, "learning_rate": 0.003, "loss": 4.074, "step": 7787 }, { "epoch": 0.07788, "grad_norm": 1.1434848747699677, "learning_rate": 0.003, "loss": 4.0907, "step": 7788 }, { "epoch": 0.07789, "grad_norm": 1.257771451993656, "learning_rate": 0.003, "loss": 4.0768, "step": 7789 }, { "epoch": 0.0779, "grad_norm": 1.1643784681384861, "learning_rate": 0.003, "loss": 4.1285, "step": 7790 }, { "epoch": 0.07791, "grad_norm": 0.8236374738012967, "learning_rate": 0.003, "loss": 4.0816, "step": 7791 }, { "epoch": 0.07792, "grad_norm": 1.0332387020792073, "learning_rate": 0.003, "loss": 4.0716, "step": 7792 }, { "epoch": 0.07793, "grad_norm": 1.1512556053819554, "learning_rate": 0.003, "loss": 4.0767, "step": 7793 }, { "epoch": 0.07794, "grad_norm": 1.03917363342737, "learning_rate": 0.003, "loss": 4.1239, "step": 7794 }, { "epoch": 0.07795, "grad_norm": 1.6696662165864793, "learning_rate": 0.003, "loss": 4.0798, "step": 7795 }, { "epoch": 0.07796, "grad_norm": 0.787952991787084, "learning_rate": 0.003, "loss": 4.0891, "step": 7796 }, { "epoch": 0.07797, "grad_norm": 0.8948697180629456, "learning_rate": 0.003, "loss": 4.084, "step": 7797 }, { "epoch": 0.07798, "grad_norm": 0.9493585597786888, "learning_rate": 0.003, "loss": 4.1203, "step": 7798 }, { "epoch": 0.07799, "grad_norm": 1.2066941416147947, "learning_rate": 0.003, "loss": 4.1283, "step": 7799 }, { "epoch": 0.078, "grad_norm": 1.0241472782325425, "learning_rate": 0.003, "loss": 4.0887, "step": 7800 }, { "epoch": 0.07801, "grad_norm": 1.0033620445871885, "learning_rate": 0.003, "loss": 4.0972, "step": 7801 }, { "epoch": 0.07802, "grad_norm": 1.0483115871730873, "learning_rate": 0.003, "loss": 4.0817, "step": 7802 }, { "epoch": 0.07803, "grad_norm": 1.1661836582075822, "learning_rate": 0.003, "loss": 4.0931, "step": 7803 }, { "epoch": 0.07804, "grad_norm": 1.0045845234300916, "learning_rate": 0.003, "loss": 4.115, "step": 7804 }, { "epoch": 0.07805, "grad_norm": 1.3719059597176528, "learning_rate": 0.003, "loss": 4.1117, "step": 7805 }, { "epoch": 0.07806, "grad_norm": 1.1373415665521494, "learning_rate": 0.003, "loss": 4.1191, "step": 7806 }, { "epoch": 0.07807, "grad_norm": 1.1540676505275376, "learning_rate": 0.003, "loss": 4.1231, "step": 7807 }, { "epoch": 0.07808, "grad_norm": 1.304033145674968, "learning_rate": 0.003, "loss": 4.0918, "step": 7808 }, { "epoch": 0.07809, "grad_norm": 1.1020823842953529, "learning_rate": 0.003, "loss": 4.1039, "step": 7809 }, { "epoch": 0.0781, "grad_norm": 1.081555316618965, "learning_rate": 0.003, "loss": 4.0832, "step": 7810 }, { "epoch": 0.07811, "grad_norm": 0.9962888054522434, "learning_rate": 0.003, "loss": 4.0946, "step": 7811 }, { "epoch": 0.07812, "grad_norm": 1.1186570479651625, "learning_rate": 0.003, "loss": 4.0993, "step": 7812 }, { "epoch": 0.07813, "grad_norm": 1.1090198553450583, "learning_rate": 0.003, "loss": 4.1117, "step": 7813 }, { "epoch": 0.07814, "grad_norm": 1.2848083778524626, "learning_rate": 0.003, "loss": 4.1436, "step": 7814 }, { "epoch": 0.07815, "grad_norm": 0.9594867164410406, "learning_rate": 0.003, "loss": 4.1034, "step": 7815 }, { "epoch": 0.07816, "grad_norm": 1.0723647489392034, "learning_rate": 0.003, "loss": 4.0933, "step": 7816 }, { "epoch": 0.07817, "grad_norm": 1.2086534813788286, "learning_rate": 0.003, "loss": 4.0958, "step": 7817 }, { "epoch": 0.07818, "grad_norm": 1.2308305090547291, "learning_rate": 0.003, "loss": 4.088, "step": 7818 }, { "epoch": 0.07819, "grad_norm": 1.1260665579293765, "learning_rate": 0.003, "loss": 4.0767, "step": 7819 }, { "epoch": 0.0782, "grad_norm": 1.037032588684538, "learning_rate": 0.003, "loss": 4.1191, "step": 7820 }, { "epoch": 0.07821, "grad_norm": 1.1246879048340699, "learning_rate": 0.003, "loss": 4.0885, "step": 7821 }, { "epoch": 0.07822, "grad_norm": 1.0979965409828816, "learning_rate": 0.003, "loss": 4.0781, "step": 7822 }, { "epoch": 0.07823, "grad_norm": 1.1545268347310245, "learning_rate": 0.003, "loss": 4.0959, "step": 7823 }, { "epoch": 0.07824, "grad_norm": 0.98432541262975, "learning_rate": 0.003, "loss": 4.1239, "step": 7824 }, { "epoch": 0.07825, "grad_norm": 1.256001106118303, "learning_rate": 0.003, "loss": 4.0913, "step": 7825 }, { "epoch": 0.07826, "grad_norm": 1.0371372534263357, "learning_rate": 0.003, "loss": 4.0815, "step": 7826 }, { "epoch": 0.07827, "grad_norm": 1.3685741094922728, "learning_rate": 0.003, "loss": 4.1127, "step": 7827 }, { "epoch": 0.07828, "grad_norm": 1.0775990135733946, "learning_rate": 0.003, "loss": 4.1138, "step": 7828 }, { "epoch": 0.07829, "grad_norm": 1.3362851668698614, "learning_rate": 0.003, "loss": 4.1213, "step": 7829 }, { "epoch": 0.0783, "grad_norm": 0.9305364252486059, "learning_rate": 0.003, "loss": 4.0819, "step": 7830 }, { "epoch": 0.07831, "grad_norm": 0.8515151977423888, "learning_rate": 0.003, "loss": 4.0864, "step": 7831 }, { "epoch": 0.07832, "grad_norm": 0.8728982928691557, "learning_rate": 0.003, "loss": 4.0968, "step": 7832 }, { "epoch": 0.07833, "grad_norm": 0.9928756485225986, "learning_rate": 0.003, "loss": 4.0943, "step": 7833 }, { "epoch": 0.07834, "grad_norm": 1.4351901450584381, "learning_rate": 0.003, "loss": 4.1542, "step": 7834 }, { "epoch": 0.07835, "grad_norm": 0.9793084700860754, "learning_rate": 0.003, "loss": 4.1065, "step": 7835 }, { "epoch": 0.07836, "grad_norm": 1.2828041473172216, "learning_rate": 0.003, "loss": 4.0977, "step": 7836 }, { "epoch": 0.07837, "grad_norm": 1.0306628659997845, "learning_rate": 0.003, "loss": 4.1037, "step": 7837 }, { "epoch": 0.07838, "grad_norm": 1.2201424308363111, "learning_rate": 0.003, "loss": 4.065, "step": 7838 }, { "epoch": 0.07839, "grad_norm": 0.8727982004597001, "learning_rate": 0.003, "loss": 4.0846, "step": 7839 }, { "epoch": 0.0784, "grad_norm": 0.9518942667143618, "learning_rate": 0.003, "loss": 4.1125, "step": 7840 }, { "epoch": 0.07841, "grad_norm": 1.1002413813695149, "learning_rate": 0.003, "loss": 4.094, "step": 7841 }, { "epoch": 0.07842, "grad_norm": 1.3289013602682165, "learning_rate": 0.003, "loss": 4.0907, "step": 7842 }, { "epoch": 0.07843, "grad_norm": 1.1995549008047959, "learning_rate": 0.003, "loss": 4.1039, "step": 7843 }, { "epoch": 0.07844, "grad_norm": 1.302842725312235, "learning_rate": 0.003, "loss": 4.1077, "step": 7844 }, { "epoch": 0.07845, "grad_norm": 0.8615424064159155, "learning_rate": 0.003, "loss": 4.1102, "step": 7845 }, { "epoch": 0.07846, "grad_norm": 1.1032730283055294, "learning_rate": 0.003, "loss": 4.1301, "step": 7846 }, { "epoch": 0.07847, "grad_norm": 1.1716909284747095, "learning_rate": 0.003, "loss": 4.0811, "step": 7847 }, { "epoch": 0.07848, "grad_norm": 1.0298840210031728, "learning_rate": 0.003, "loss": 4.119, "step": 7848 }, { "epoch": 0.07849, "grad_norm": 0.9361652769304023, "learning_rate": 0.003, "loss": 4.126, "step": 7849 }, { "epoch": 0.0785, "grad_norm": 0.9654921254423353, "learning_rate": 0.003, "loss": 4.0953, "step": 7850 }, { "epoch": 0.07851, "grad_norm": 1.3042027618759802, "learning_rate": 0.003, "loss": 4.0905, "step": 7851 }, { "epoch": 0.07852, "grad_norm": 1.2431447724018592, "learning_rate": 0.003, "loss": 4.1292, "step": 7852 }, { "epoch": 0.07853, "grad_norm": 1.083275442045458, "learning_rate": 0.003, "loss": 4.093, "step": 7853 }, { "epoch": 0.07854, "grad_norm": 0.9580563046958107, "learning_rate": 0.003, "loss": 4.1075, "step": 7854 }, { "epoch": 0.07855, "grad_norm": 0.9661427902084164, "learning_rate": 0.003, "loss": 4.0952, "step": 7855 }, { "epoch": 0.07856, "grad_norm": 1.1404243379470476, "learning_rate": 0.003, "loss": 4.1035, "step": 7856 }, { "epoch": 0.07857, "grad_norm": 1.1278970478970323, "learning_rate": 0.003, "loss": 4.1202, "step": 7857 }, { "epoch": 0.07858, "grad_norm": 1.2400683738024636, "learning_rate": 0.003, "loss": 4.113, "step": 7858 }, { "epoch": 0.07859, "grad_norm": 1.1367394355297402, "learning_rate": 0.003, "loss": 4.0826, "step": 7859 }, { "epoch": 0.0786, "grad_norm": 1.1458758894227754, "learning_rate": 0.003, "loss": 4.096, "step": 7860 }, { "epoch": 0.07861, "grad_norm": 1.020642921999076, "learning_rate": 0.003, "loss": 4.1145, "step": 7861 }, { "epoch": 0.07862, "grad_norm": 0.9153329211108876, "learning_rate": 0.003, "loss": 4.1321, "step": 7862 }, { "epoch": 0.07863, "grad_norm": 1.0169957598242134, "learning_rate": 0.003, "loss": 4.067, "step": 7863 }, { "epoch": 0.07864, "grad_norm": 1.259541799929286, "learning_rate": 0.003, "loss": 4.1166, "step": 7864 }, { "epoch": 0.07865, "grad_norm": 1.053664052392776, "learning_rate": 0.003, "loss": 4.0865, "step": 7865 }, { "epoch": 0.07866, "grad_norm": 1.1541810165983357, "learning_rate": 0.003, "loss": 4.0947, "step": 7866 }, { "epoch": 0.07867, "grad_norm": 1.0834227683128044, "learning_rate": 0.003, "loss": 4.0832, "step": 7867 }, { "epoch": 0.07868, "grad_norm": 1.1142727145292364, "learning_rate": 0.003, "loss": 4.1176, "step": 7868 }, { "epoch": 0.07869, "grad_norm": 1.3213620312341967, "learning_rate": 0.003, "loss": 4.1037, "step": 7869 }, { "epoch": 0.0787, "grad_norm": 0.9543811915443509, "learning_rate": 0.003, "loss": 4.078, "step": 7870 }, { "epoch": 0.07871, "grad_norm": 1.0556891019646535, "learning_rate": 0.003, "loss": 4.1321, "step": 7871 }, { "epoch": 0.07872, "grad_norm": 1.0423505032259703, "learning_rate": 0.003, "loss": 4.1261, "step": 7872 }, { "epoch": 0.07873, "grad_norm": 1.2630689394264598, "learning_rate": 0.003, "loss": 4.0867, "step": 7873 }, { "epoch": 0.07874, "grad_norm": 1.0744222348469838, "learning_rate": 0.003, "loss": 4.0978, "step": 7874 }, { "epoch": 0.07875, "grad_norm": 1.1590033383841085, "learning_rate": 0.003, "loss": 4.1099, "step": 7875 }, { "epoch": 0.07876, "grad_norm": 1.0641603834704565, "learning_rate": 0.003, "loss": 4.1085, "step": 7876 }, { "epoch": 0.07877, "grad_norm": 1.07771416779964, "learning_rate": 0.003, "loss": 4.0941, "step": 7877 }, { "epoch": 0.07878, "grad_norm": 1.0105372101646315, "learning_rate": 0.003, "loss": 4.1039, "step": 7878 }, { "epoch": 0.07879, "grad_norm": 0.964112703343037, "learning_rate": 0.003, "loss": 4.1048, "step": 7879 }, { "epoch": 0.0788, "grad_norm": 1.0161410845205, "learning_rate": 0.003, "loss": 4.0879, "step": 7880 }, { "epoch": 0.07881, "grad_norm": 1.1773807921941188, "learning_rate": 0.003, "loss": 4.1029, "step": 7881 }, { "epoch": 0.07882, "grad_norm": 1.10321775987289, "learning_rate": 0.003, "loss": 4.0936, "step": 7882 }, { "epoch": 0.07883, "grad_norm": 1.369091366939899, "learning_rate": 0.003, "loss": 4.0823, "step": 7883 }, { "epoch": 0.07884, "grad_norm": 0.9760109355931481, "learning_rate": 0.003, "loss": 4.0989, "step": 7884 }, { "epoch": 0.07885, "grad_norm": 0.9854215860262345, "learning_rate": 0.003, "loss": 4.0876, "step": 7885 }, { "epoch": 0.07886, "grad_norm": 1.0764050852490537, "learning_rate": 0.003, "loss": 4.1153, "step": 7886 }, { "epoch": 0.07887, "grad_norm": 1.1764071385853805, "learning_rate": 0.003, "loss": 4.0926, "step": 7887 }, { "epoch": 0.07888, "grad_norm": 1.209620298516619, "learning_rate": 0.003, "loss": 4.106, "step": 7888 }, { "epoch": 0.07889, "grad_norm": 1.0297274496770714, "learning_rate": 0.003, "loss": 4.0716, "step": 7889 }, { "epoch": 0.0789, "grad_norm": 1.1786330089012873, "learning_rate": 0.003, "loss": 4.0908, "step": 7890 }, { "epoch": 0.07891, "grad_norm": 1.0533061689526215, "learning_rate": 0.003, "loss": 4.0864, "step": 7891 }, { "epoch": 0.07892, "grad_norm": 1.1205645980083818, "learning_rate": 0.003, "loss": 4.0856, "step": 7892 }, { "epoch": 0.07893, "grad_norm": 1.0523794156863215, "learning_rate": 0.003, "loss": 4.1082, "step": 7893 }, { "epoch": 0.07894, "grad_norm": 0.9699886855362052, "learning_rate": 0.003, "loss": 4.0785, "step": 7894 }, { "epoch": 0.07895, "grad_norm": 0.9833094035853261, "learning_rate": 0.003, "loss": 4.077, "step": 7895 }, { "epoch": 0.07896, "grad_norm": 0.9776394904354138, "learning_rate": 0.003, "loss": 4.0887, "step": 7896 }, { "epoch": 0.07897, "grad_norm": 1.1062378440010827, "learning_rate": 0.003, "loss": 4.1185, "step": 7897 }, { "epoch": 0.07898, "grad_norm": 1.0700004254123947, "learning_rate": 0.003, "loss": 4.0999, "step": 7898 }, { "epoch": 0.07899, "grad_norm": 1.293605829425846, "learning_rate": 0.003, "loss": 4.0858, "step": 7899 }, { "epoch": 0.079, "grad_norm": 1.0712959158479753, "learning_rate": 0.003, "loss": 4.0863, "step": 7900 }, { "epoch": 0.07901, "grad_norm": 1.3719650813885331, "learning_rate": 0.003, "loss": 4.1227, "step": 7901 }, { "epoch": 0.07902, "grad_norm": 1.0271054136067208, "learning_rate": 0.003, "loss": 4.0707, "step": 7902 }, { "epoch": 0.07903, "grad_norm": 1.4611371181707402, "learning_rate": 0.003, "loss": 4.0956, "step": 7903 }, { "epoch": 0.07904, "grad_norm": 0.8252296257800624, "learning_rate": 0.003, "loss": 4.071, "step": 7904 }, { "epoch": 0.07905, "grad_norm": 0.7895305759219005, "learning_rate": 0.003, "loss": 4.0584, "step": 7905 }, { "epoch": 0.07906, "grad_norm": 0.9276855253246521, "learning_rate": 0.003, "loss": 4.0761, "step": 7906 }, { "epoch": 0.07907, "grad_norm": 1.0888746411346901, "learning_rate": 0.003, "loss": 4.1025, "step": 7907 }, { "epoch": 0.07908, "grad_norm": 1.1846497029397596, "learning_rate": 0.003, "loss": 4.1073, "step": 7908 }, { "epoch": 0.07909, "grad_norm": 1.3242319782732728, "learning_rate": 0.003, "loss": 4.1117, "step": 7909 }, { "epoch": 0.0791, "grad_norm": 0.9082616315727167, "learning_rate": 0.003, "loss": 4.0879, "step": 7910 }, { "epoch": 0.07911, "grad_norm": 0.977177222280417, "learning_rate": 0.003, "loss": 4.1098, "step": 7911 }, { "epoch": 0.07912, "grad_norm": 1.134278943251006, "learning_rate": 0.003, "loss": 4.087, "step": 7912 }, { "epoch": 0.07913, "grad_norm": 0.9726471608410041, "learning_rate": 0.003, "loss": 4.0732, "step": 7913 }, { "epoch": 0.07914, "grad_norm": 1.0400459476467385, "learning_rate": 0.003, "loss": 4.0889, "step": 7914 }, { "epoch": 0.07915, "grad_norm": 1.1254579113656433, "learning_rate": 0.003, "loss": 4.082, "step": 7915 }, { "epoch": 0.07916, "grad_norm": 1.170036571371052, "learning_rate": 0.003, "loss": 4.1113, "step": 7916 }, { "epoch": 0.07917, "grad_norm": 0.9982422946503648, "learning_rate": 0.003, "loss": 4.1077, "step": 7917 }, { "epoch": 0.07918, "grad_norm": 1.334554846263995, "learning_rate": 0.003, "loss": 4.1118, "step": 7918 }, { "epoch": 0.07919, "grad_norm": 0.9593512516129795, "learning_rate": 0.003, "loss": 4.0627, "step": 7919 }, { "epoch": 0.0792, "grad_norm": 1.2945155738424934, "learning_rate": 0.003, "loss": 4.1006, "step": 7920 }, { "epoch": 0.07921, "grad_norm": 0.998743031279866, "learning_rate": 0.003, "loss": 4.1017, "step": 7921 }, { "epoch": 0.07922, "grad_norm": 1.2856678381292854, "learning_rate": 0.003, "loss": 4.0856, "step": 7922 }, { "epoch": 0.07923, "grad_norm": 1.3839771793815328, "learning_rate": 0.003, "loss": 4.089, "step": 7923 }, { "epoch": 0.07924, "grad_norm": 0.843757956272009, "learning_rate": 0.003, "loss": 4.0794, "step": 7924 }, { "epoch": 0.07925, "grad_norm": 0.8712759778946663, "learning_rate": 0.003, "loss": 4.1073, "step": 7925 }, { "epoch": 0.07926, "grad_norm": 0.8792339530446677, "learning_rate": 0.003, "loss": 4.1161, "step": 7926 }, { "epoch": 0.07927, "grad_norm": 0.9425668093513904, "learning_rate": 0.003, "loss": 4.0557, "step": 7927 }, { "epoch": 0.07928, "grad_norm": 1.181163559854013, "learning_rate": 0.003, "loss": 4.1098, "step": 7928 }, { "epoch": 0.07929, "grad_norm": 1.0055246739263888, "learning_rate": 0.003, "loss": 4.0849, "step": 7929 }, { "epoch": 0.0793, "grad_norm": 1.1521126305735454, "learning_rate": 0.003, "loss": 4.0738, "step": 7930 }, { "epoch": 0.07931, "grad_norm": 1.1425888894120744, "learning_rate": 0.003, "loss": 4.1103, "step": 7931 }, { "epoch": 0.07932, "grad_norm": 1.2461666977358548, "learning_rate": 0.003, "loss": 4.1154, "step": 7932 }, { "epoch": 0.07933, "grad_norm": 1.0256002678148357, "learning_rate": 0.003, "loss": 4.0987, "step": 7933 }, { "epoch": 0.07934, "grad_norm": 1.3007738245290619, "learning_rate": 0.003, "loss": 4.1253, "step": 7934 }, { "epoch": 0.07935, "grad_norm": 0.9296174844539437, "learning_rate": 0.003, "loss": 4.0996, "step": 7935 }, { "epoch": 0.07936, "grad_norm": 0.8725870895380266, "learning_rate": 0.003, "loss": 4.0912, "step": 7936 }, { "epoch": 0.07937, "grad_norm": 0.932587169003262, "learning_rate": 0.003, "loss": 4.0595, "step": 7937 }, { "epoch": 0.07938, "grad_norm": 1.0963665524323598, "learning_rate": 0.003, "loss": 4.0757, "step": 7938 }, { "epoch": 0.07939, "grad_norm": 1.1192645927998164, "learning_rate": 0.003, "loss": 4.0828, "step": 7939 }, { "epoch": 0.0794, "grad_norm": 1.0887510653304764, "learning_rate": 0.003, "loss": 4.0882, "step": 7940 }, { "epoch": 0.07941, "grad_norm": 1.568700579038047, "learning_rate": 0.003, "loss": 4.0887, "step": 7941 }, { "epoch": 0.07942, "grad_norm": 0.9367815838429774, "learning_rate": 0.003, "loss": 4.075, "step": 7942 }, { "epoch": 0.07943, "grad_norm": 0.989902176784745, "learning_rate": 0.003, "loss": 4.082, "step": 7943 }, { "epoch": 0.07944, "grad_norm": 1.2806804391718558, "learning_rate": 0.003, "loss": 4.1317, "step": 7944 }, { "epoch": 0.07945, "grad_norm": 1.0465168074755917, "learning_rate": 0.003, "loss": 4.1036, "step": 7945 }, { "epoch": 0.07946, "grad_norm": 1.1723891827338213, "learning_rate": 0.003, "loss": 4.1063, "step": 7946 }, { "epoch": 0.07947, "grad_norm": 1.1681881065505169, "learning_rate": 0.003, "loss": 4.1247, "step": 7947 }, { "epoch": 0.07948, "grad_norm": 1.0403361408707281, "learning_rate": 0.003, "loss": 4.1027, "step": 7948 }, { "epoch": 0.07949, "grad_norm": 1.089737847328847, "learning_rate": 0.003, "loss": 4.1179, "step": 7949 }, { "epoch": 0.0795, "grad_norm": 1.0006486459411923, "learning_rate": 0.003, "loss": 4.0745, "step": 7950 }, { "epoch": 0.07951, "grad_norm": 1.0643791943109515, "learning_rate": 0.003, "loss": 4.0779, "step": 7951 }, { "epoch": 0.07952, "grad_norm": 1.1537519015439461, "learning_rate": 0.003, "loss": 4.1085, "step": 7952 }, { "epoch": 0.07953, "grad_norm": 1.183449549789871, "learning_rate": 0.003, "loss": 4.0774, "step": 7953 }, { "epoch": 0.07954, "grad_norm": 0.9709964660164878, "learning_rate": 0.003, "loss": 4.1149, "step": 7954 }, { "epoch": 0.07955, "grad_norm": 1.2278848117081071, "learning_rate": 0.003, "loss": 4.116, "step": 7955 }, { "epoch": 0.07956, "grad_norm": 1.1583662422335557, "learning_rate": 0.003, "loss": 4.0997, "step": 7956 }, { "epoch": 0.07957, "grad_norm": 1.0170490448671368, "learning_rate": 0.003, "loss": 4.0855, "step": 7957 }, { "epoch": 0.07958, "grad_norm": 1.084771863517993, "learning_rate": 0.003, "loss": 4.0962, "step": 7958 }, { "epoch": 0.07959, "grad_norm": 1.1488835770046757, "learning_rate": 0.003, "loss": 4.1015, "step": 7959 }, { "epoch": 0.0796, "grad_norm": 1.2272254053247194, "learning_rate": 0.003, "loss": 4.0996, "step": 7960 }, { "epoch": 0.07961, "grad_norm": 0.7550769269887291, "learning_rate": 0.003, "loss": 4.0766, "step": 7961 }, { "epoch": 0.07962, "grad_norm": 0.7402534276864938, "learning_rate": 0.003, "loss": 4.0986, "step": 7962 }, { "epoch": 0.07963, "grad_norm": 0.8779302221658243, "learning_rate": 0.003, "loss": 4.0657, "step": 7963 }, { "epoch": 0.07964, "grad_norm": 1.0462430118892019, "learning_rate": 0.003, "loss": 4.0903, "step": 7964 }, { "epoch": 0.07965, "grad_norm": 1.1593091162794427, "learning_rate": 0.003, "loss": 4.1039, "step": 7965 }, { "epoch": 0.07966, "grad_norm": 1.0541722292533413, "learning_rate": 0.003, "loss": 4.0796, "step": 7966 }, { "epoch": 0.07967, "grad_norm": 1.1496986651653818, "learning_rate": 0.003, "loss": 4.1054, "step": 7967 }, { "epoch": 0.07968, "grad_norm": 1.2051439905100614, "learning_rate": 0.003, "loss": 4.0782, "step": 7968 }, { "epoch": 0.07969, "grad_norm": 1.166558887648512, "learning_rate": 0.003, "loss": 4.1204, "step": 7969 }, { "epoch": 0.0797, "grad_norm": 0.9177683584105356, "learning_rate": 0.003, "loss": 4.0929, "step": 7970 }, { "epoch": 0.07971, "grad_norm": 1.1341545449465498, "learning_rate": 0.003, "loss": 4.0842, "step": 7971 }, { "epoch": 0.07972, "grad_norm": 1.5081463937449056, "learning_rate": 0.003, "loss": 4.0989, "step": 7972 }, { "epoch": 0.07973, "grad_norm": 0.8474584989165531, "learning_rate": 0.003, "loss": 4.1, "step": 7973 }, { "epoch": 0.07974, "grad_norm": 1.1289338429631244, "learning_rate": 0.003, "loss": 4.0904, "step": 7974 }, { "epoch": 0.07975, "grad_norm": 1.0951365579102945, "learning_rate": 0.003, "loss": 4.1214, "step": 7975 }, { "epoch": 0.07976, "grad_norm": 1.202331783937366, "learning_rate": 0.003, "loss": 4.1087, "step": 7976 }, { "epoch": 0.07977, "grad_norm": 1.129368184698674, "learning_rate": 0.003, "loss": 4.0986, "step": 7977 }, { "epoch": 0.07978, "grad_norm": 1.1036298314531698, "learning_rate": 0.003, "loss": 4.1033, "step": 7978 }, { "epoch": 0.07979, "grad_norm": 1.0740326562297844, "learning_rate": 0.003, "loss": 4.0822, "step": 7979 }, { "epoch": 0.0798, "grad_norm": 1.2013980109141247, "learning_rate": 0.003, "loss": 4.115, "step": 7980 }, { "epoch": 0.07981, "grad_norm": 1.2324688029721227, "learning_rate": 0.003, "loss": 4.0655, "step": 7981 }, { "epoch": 0.07982, "grad_norm": 1.102179948446623, "learning_rate": 0.003, "loss": 4.1, "step": 7982 }, { "epoch": 0.07983, "grad_norm": 1.0997229370918842, "learning_rate": 0.003, "loss": 4.126, "step": 7983 }, { "epoch": 0.07984, "grad_norm": 1.1774300218492646, "learning_rate": 0.003, "loss": 4.1067, "step": 7984 }, { "epoch": 0.07985, "grad_norm": 1.0396930643543236, "learning_rate": 0.003, "loss": 4.0662, "step": 7985 }, { "epoch": 0.07986, "grad_norm": 0.9072544573320482, "learning_rate": 0.003, "loss": 4.1073, "step": 7986 }, { "epoch": 0.07987, "grad_norm": 1.014408265411552, "learning_rate": 0.003, "loss": 4.0812, "step": 7987 }, { "epoch": 0.07988, "grad_norm": 1.0940592348211349, "learning_rate": 0.003, "loss": 4.1003, "step": 7988 }, { "epoch": 0.07989, "grad_norm": 1.5166715738459051, "learning_rate": 0.003, "loss": 4.0776, "step": 7989 }, { "epoch": 0.0799, "grad_norm": 1.1113923958563618, "learning_rate": 0.003, "loss": 4.0935, "step": 7990 }, { "epoch": 0.07991, "grad_norm": 1.1217722300895914, "learning_rate": 0.003, "loss": 4.093, "step": 7991 }, { "epoch": 0.07992, "grad_norm": 0.9491402602543485, "learning_rate": 0.003, "loss": 4.0887, "step": 7992 }, { "epoch": 0.07993, "grad_norm": 0.955914940013595, "learning_rate": 0.003, "loss": 4.0813, "step": 7993 }, { "epoch": 0.07994, "grad_norm": 0.9095882982710126, "learning_rate": 0.003, "loss": 4.0995, "step": 7994 }, { "epoch": 0.07995, "grad_norm": 1.1501864531666672, "learning_rate": 0.003, "loss": 4.1057, "step": 7995 }, { "epoch": 0.07996, "grad_norm": 1.2288505894745418, "learning_rate": 0.003, "loss": 4.0989, "step": 7996 }, { "epoch": 0.07997, "grad_norm": 1.0079549128442122, "learning_rate": 0.003, "loss": 4.0908, "step": 7997 }, { "epoch": 0.07998, "grad_norm": 1.2919012167956998, "learning_rate": 0.003, "loss": 4.1264, "step": 7998 }, { "epoch": 0.07999, "grad_norm": 0.7692306732693436, "learning_rate": 0.003, "loss": 4.0886, "step": 7999 }, { "epoch": 0.08, "grad_norm": 0.7658749344062428, "learning_rate": 0.003, "loss": 4.094, "step": 8000 }, { "epoch": 0.08001, "grad_norm": 1.1172290952445811, "learning_rate": 0.003, "loss": 4.0872, "step": 8001 }, { "epoch": 0.08002, "grad_norm": 1.2957889022004523, "learning_rate": 0.003, "loss": 4.0802, "step": 8002 }, { "epoch": 0.08003, "grad_norm": 1.0457934117693226, "learning_rate": 0.003, "loss": 4.1042, "step": 8003 }, { "epoch": 0.08004, "grad_norm": 1.0305139871514264, "learning_rate": 0.003, "loss": 4.0836, "step": 8004 }, { "epoch": 0.08005, "grad_norm": 1.1419018400508258, "learning_rate": 0.003, "loss": 4.1135, "step": 8005 }, { "epoch": 0.08006, "grad_norm": 1.0896797731228673, "learning_rate": 0.003, "loss": 4.0901, "step": 8006 }, { "epoch": 0.08007, "grad_norm": 1.228265310250077, "learning_rate": 0.003, "loss": 4.0925, "step": 8007 }, { "epoch": 0.08008, "grad_norm": 1.1176908197364457, "learning_rate": 0.003, "loss": 4.074, "step": 8008 }, { "epoch": 0.08009, "grad_norm": 1.2643872060654957, "learning_rate": 0.003, "loss": 4.0987, "step": 8009 }, { "epoch": 0.0801, "grad_norm": 1.0093075739010895, "learning_rate": 0.003, "loss": 4.0856, "step": 8010 }, { "epoch": 0.08011, "grad_norm": 1.2086973478631764, "learning_rate": 0.003, "loss": 4.1013, "step": 8011 }, { "epoch": 0.08012, "grad_norm": 1.1284095226139361, "learning_rate": 0.003, "loss": 4.1026, "step": 8012 }, { "epoch": 0.08013, "grad_norm": 1.0806491574385775, "learning_rate": 0.003, "loss": 4.1017, "step": 8013 }, { "epoch": 0.08014, "grad_norm": 1.0499299018409107, "learning_rate": 0.003, "loss": 4.0853, "step": 8014 }, { "epoch": 0.08015, "grad_norm": 1.1832836384348369, "learning_rate": 0.003, "loss": 4.1183, "step": 8015 }, { "epoch": 0.08016, "grad_norm": 1.2639744517182545, "learning_rate": 0.003, "loss": 4.0714, "step": 8016 }, { "epoch": 0.08017, "grad_norm": 1.1523916089493353, "learning_rate": 0.003, "loss": 4.0788, "step": 8017 }, { "epoch": 0.08018, "grad_norm": 1.184155355677051, "learning_rate": 0.003, "loss": 4.0916, "step": 8018 }, { "epoch": 0.08019, "grad_norm": 0.8903633933008729, "learning_rate": 0.003, "loss": 4.1168, "step": 8019 }, { "epoch": 0.0802, "grad_norm": 0.9520940918042209, "learning_rate": 0.003, "loss": 4.0809, "step": 8020 }, { "epoch": 0.08021, "grad_norm": 1.1295461972729675, "learning_rate": 0.003, "loss": 4.0634, "step": 8021 }, { "epoch": 0.08022, "grad_norm": 1.002556264564961, "learning_rate": 0.003, "loss": 4.1026, "step": 8022 }, { "epoch": 0.08023, "grad_norm": 1.1786108591329556, "learning_rate": 0.003, "loss": 4.0941, "step": 8023 }, { "epoch": 0.08024, "grad_norm": 0.9555442073542494, "learning_rate": 0.003, "loss": 4.1078, "step": 8024 }, { "epoch": 0.08025, "grad_norm": 1.2962171433165834, "learning_rate": 0.003, "loss": 4.0838, "step": 8025 }, { "epoch": 0.08026, "grad_norm": 0.9010195783376608, "learning_rate": 0.003, "loss": 4.0875, "step": 8026 }, { "epoch": 0.08027, "grad_norm": 1.0238470434941584, "learning_rate": 0.003, "loss": 4.0737, "step": 8027 }, { "epoch": 0.08028, "grad_norm": 0.9548697731707966, "learning_rate": 0.003, "loss": 4.0859, "step": 8028 }, { "epoch": 0.08029, "grad_norm": 1.0023059334224902, "learning_rate": 0.003, "loss": 4.1106, "step": 8029 }, { "epoch": 0.0803, "grad_norm": 1.252440818965578, "learning_rate": 0.003, "loss": 4.0842, "step": 8030 }, { "epoch": 0.08031, "grad_norm": 1.068095590321888, "learning_rate": 0.003, "loss": 4.0989, "step": 8031 }, { "epoch": 0.08032, "grad_norm": 1.3578842748881137, "learning_rate": 0.003, "loss": 4.0912, "step": 8032 }, { "epoch": 0.08033, "grad_norm": 0.9060347362052149, "learning_rate": 0.003, "loss": 4.112, "step": 8033 }, { "epoch": 0.08034, "grad_norm": 1.0537972743762867, "learning_rate": 0.003, "loss": 4.1106, "step": 8034 }, { "epoch": 0.08035, "grad_norm": 1.319790888337074, "learning_rate": 0.003, "loss": 4.1239, "step": 8035 }, { "epoch": 0.08036, "grad_norm": 0.9259950251223968, "learning_rate": 0.003, "loss": 4.1316, "step": 8036 }, { "epoch": 0.08037, "grad_norm": 0.9375805778403902, "learning_rate": 0.003, "loss": 4.0925, "step": 8037 }, { "epoch": 0.08038, "grad_norm": 1.0239195144037003, "learning_rate": 0.003, "loss": 4.0916, "step": 8038 }, { "epoch": 0.08039, "grad_norm": 1.3195500348248446, "learning_rate": 0.003, "loss": 4.1079, "step": 8039 }, { "epoch": 0.0804, "grad_norm": 1.1082428212704758, "learning_rate": 0.003, "loss": 4.1099, "step": 8040 }, { "epoch": 0.08041, "grad_norm": 1.1596745215156437, "learning_rate": 0.003, "loss": 4.1201, "step": 8041 }, { "epoch": 0.08042, "grad_norm": 1.064287573359169, "learning_rate": 0.003, "loss": 4.1021, "step": 8042 }, { "epoch": 0.08043, "grad_norm": 1.1696931645456978, "learning_rate": 0.003, "loss": 4.099, "step": 8043 }, { "epoch": 0.08044, "grad_norm": 1.0352811531836574, "learning_rate": 0.003, "loss": 4.0786, "step": 8044 }, { "epoch": 0.08045, "grad_norm": 1.2260629551905426, "learning_rate": 0.003, "loss": 4.0875, "step": 8045 }, { "epoch": 0.08046, "grad_norm": 1.2395334777899678, "learning_rate": 0.003, "loss": 4.1011, "step": 8046 }, { "epoch": 0.08047, "grad_norm": 1.1355116433579298, "learning_rate": 0.003, "loss": 4.0952, "step": 8047 }, { "epoch": 0.08048, "grad_norm": 1.0459530983468805, "learning_rate": 0.003, "loss": 4.104, "step": 8048 }, { "epoch": 0.08049, "grad_norm": 0.9552089247433029, "learning_rate": 0.003, "loss": 4.0834, "step": 8049 }, { "epoch": 0.0805, "grad_norm": 1.076195243985042, "learning_rate": 0.003, "loss": 4.1121, "step": 8050 }, { "epoch": 0.08051, "grad_norm": 1.1851350313945763, "learning_rate": 0.003, "loss": 4.0869, "step": 8051 }, { "epoch": 0.08052, "grad_norm": 0.8808885981989337, "learning_rate": 0.003, "loss": 4.0651, "step": 8052 }, { "epoch": 0.08053, "grad_norm": 0.8774625163986162, "learning_rate": 0.003, "loss": 4.1056, "step": 8053 }, { "epoch": 0.08054, "grad_norm": 1.0416380425987897, "learning_rate": 0.003, "loss": 4.0796, "step": 8054 }, { "epoch": 0.08055, "grad_norm": 1.171907766302584, "learning_rate": 0.003, "loss": 4.0635, "step": 8055 }, { "epoch": 0.08056, "grad_norm": 0.9849653204085805, "learning_rate": 0.003, "loss": 4.0933, "step": 8056 }, { "epoch": 0.08057, "grad_norm": 0.9694296024333688, "learning_rate": 0.003, "loss": 4.0956, "step": 8057 }, { "epoch": 0.08058, "grad_norm": 1.2661816817540663, "learning_rate": 0.003, "loss": 4.1007, "step": 8058 }, { "epoch": 0.08059, "grad_norm": 1.091215180973451, "learning_rate": 0.003, "loss": 4.0782, "step": 8059 }, { "epoch": 0.0806, "grad_norm": 1.0117982176263614, "learning_rate": 0.003, "loss": 4.0856, "step": 8060 }, { "epoch": 0.08061, "grad_norm": 0.9611026577563779, "learning_rate": 0.003, "loss": 4.0705, "step": 8061 }, { "epoch": 0.08062, "grad_norm": 1.1817220754515882, "learning_rate": 0.003, "loss": 4.1051, "step": 8062 }, { "epoch": 0.08063, "grad_norm": 1.2318418776684246, "learning_rate": 0.003, "loss": 4.1387, "step": 8063 }, { "epoch": 0.08064, "grad_norm": 1.1485414527276185, "learning_rate": 0.003, "loss": 4.0918, "step": 8064 }, { "epoch": 0.08065, "grad_norm": 1.1378965393033746, "learning_rate": 0.003, "loss": 4.1036, "step": 8065 }, { "epoch": 0.08066, "grad_norm": 1.1922787895771274, "learning_rate": 0.003, "loss": 4.1211, "step": 8066 }, { "epoch": 0.08067, "grad_norm": 1.153807083794397, "learning_rate": 0.003, "loss": 4.1082, "step": 8067 }, { "epoch": 0.08068, "grad_norm": 1.027030031849749, "learning_rate": 0.003, "loss": 4.1029, "step": 8068 }, { "epoch": 0.08069, "grad_norm": 1.2677666258950757, "learning_rate": 0.003, "loss": 4.1076, "step": 8069 }, { "epoch": 0.0807, "grad_norm": 1.1788831903370685, "learning_rate": 0.003, "loss": 4.0939, "step": 8070 }, { "epoch": 0.08071, "grad_norm": 1.1349772426555031, "learning_rate": 0.003, "loss": 4.0915, "step": 8071 }, { "epoch": 0.08072, "grad_norm": 1.2643432586832963, "learning_rate": 0.003, "loss": 4.0853, "step": 8072 }, { "epoch": 0.08073, "grad_norm": 1.004717571451655, "learning_rate": 0.003, "loss": 4.0841, "step": 8073 }, { "epoch": 0.08074, "grad_norm": 1.0281952059306105, "learning_rate": 0.003, "loss": 4.0679, "step": 8074 }, { "epoch": 0.08075, "grad_norm": 1.1302334862263013, "learning_rate": 0.003, "loss": 4.1182, "step": 8075 }, { "epoch": 0.08076, "grad_norm": 0.8849239138901812, "learning_rate": 0.003, "loss": 4.0661, "step": 8076 }, { "epoch": 0.08077, "grad_norm": 1.0904929838277595, "learning_rate": 0.003, "loss": 4.121, "step": 8077 }, { "epoch": 0.08078, "grad_norm": 1.2580142950738245, "learning_rate": 0.003, "loss": 4.1121, "step": 8078 }, { "epoch": 0.08079, "grad_norm": 1.0647573839248337, "learning_rate": 0.003, "loss": 4.0971, "step": 8079 }, { "epoch": 0.0808, "grad_norm": 1.009377860450751, "learning_rate": 0.003, "loss": 4.0873, "step": 8080 }, { "epoch": 0.08081, "grad_norm": 1.1462226057928848, "learning_rate": 0.003, "loss": 4.1018, "step": 8081 }, { "epoch": 0.08082, "grad_norm": 0.8533433204287125, "learning_rate": 0.003, "loss": 4.0604, "step": 8082 }, { "epoch": 0.08083, "grad_norm": 0.9405341530283867, "learning_rate": 0.003, "loss": 4.0836, "step": 8083 }, { "epoch": 0.08084, "grad_norm": 1.1885955017782868, "learning_rate": 0.003, "loss": 4.0953, "step": 8084 }, { "epoch": 0.08085, "grad_norm": 1.059349902006872, "learning_rate": 0.003, "loss": 4.1229, "step": 8085 }, { "epoch": 0.08086, "grad_norm": 1.1246370076606642, "learning_rate": 0.003, "loss": 4.0749, "step": 8086 }, { "epoch": 0.08087, "grad_norm": 0.8911927530693307, "learning_rate": 0.003, "loss": 4.0817, "step": 8087 }, { "epoch": 0.08088, "grad_norm": 1.1142808988853259, "learning_rate": 0.003, "loss": 4.1006, "step": 8088 }, { "epoch": 0.08089, "grad_norm": 1.211243629124428, "learning_rate": 0.003, "loss": 4.0659, "step": 8089 }, { "epoch": 0.0809, "grad_norm": 1.1971598603451767, "learning_rate": 0.003, "loss": 4.0952, "step": 8090 }, { "epoch": 0.08091, "grad_norm": 1.2219998111646206, "learning_rate": 0.003, "loss": 4.1176, "step": 8091 }, { "epoch": 0.08092, "grad_norm": 1.0863330722445514, "learning_rate": 0.003, "loss": 4.1132, "step": 8092 }, { "epoch": 0.08093, "grad_norm": 1.1734069502989566, "learning_rate": 0.003, "loss": 4.0799, "step": 8093 }, { "epoch": 0.08094, "grad_norm": 0.9707437526356893, "learning_rate": 0.003, "loss": 4.0723, "step": 8094 }, { "epoch": 0.08095, "grad_norm": 1.2867032457790912, "learning_rate": 0.003, "loss": 4.0732, "step": 8095 }, { "epoch": 0.08096, "grad_norm": 1.0927893033822331, "learning_rate": 0.003, "loss": 4.0989, "step": 8096 }, { "epoch": 0.08097, "grad_norm": 1.1439771673315244, "learning_rate": 0.003, "loss": 4.0906, "step": 8097 }, { "epoch": 0.08098, "grad_norm": 1.196138267953642, "learning_rate": 0.003, "loss": 4.1225, "step": 8098 }, { "epoch": 0.08099, "grad_norm": 1.0160152850154716, "learning_rate": 0.003, "loss": 4.092, "step": 8099 }, { "epoch": 0.081, "grad_norm": 1.0887542330321427, "learning_rate": 0.003, "loss": 4.0982, "step": 8100 }, { "epoch": 0.08101, "grad_norm": 1.1829287676768256, "learning_rate": 0.003, "loss": 4.1222, "step": 8101 }, { "epoch": 0.08102, "grad_norm": 1.140589086395366, "learning_rate": 0.003, "loss": 4.097, "step": 8102 }, { "epoch": 0.08103, "grad_norm": 1.1358979019750919, "learning_rate": 0.003, "loss": 4.0985, "step": 8103 }, { "epoch": 0.08104, "grad_norm": 1.1288134341200402, "learning_rate": 0.003, "loss": 4.099, "step": 8104 }, { "epoch": 0.08105, "grad_norm": 1.0135805089098007, "learning_rate": 0.003, "loss": 4.1144, "step": 8105 }, { "epoch": 0.08106, "grad_norm": 1.2455233908534715, "learning_rate": 0.003, "loss": 4.1118, "step": 8106 }, { "epoch": 0.08107, "grad_norm": 1.0325069605488597, "learning_rate": 0.003, "loss": 4.1042, "step": 8107 }, { "epoch": 0.08108, "grad_norm": 1.0751154807719285, "learning_rate": 0.003, "loss": 4.0501, "step": 8108 }, { "epoch": 0.08109, "grad_norm": 1.025582036833368, "learning_rate": 0.003, "loss": 4.0949, "step": 8109 }, { "epoch": 0.0811, "grad_norm": 1.1308151638204902, "learning_rate": 0.003, "loss": 4.1134, "step": 8110 }, { "epoch": 0.08111, "grad_norm": 1.1259700518707663, "learning_rate": 0.003, "loss": 4.1048, "step": 8111 }, { "epoch": 0.08112, "grad_norm": 1.0161036704370565, "learning_rate": 0.003, "loss": 4.0731, "step": 8112 }, { "epoch": 0.08113, "grad_norm": 1.0188779126166405, "learning_rate": 0.003, "loss": 4.0888, "step": 8113 }, { "epoch": 0.08114, "grad_norm": 0.9793445412752302, "learning_rate": 0.003, "loss": 4.1156, "step": 8114 }, { "epoch": 0.08115, "grad_norm": 1.0652029124997249, "learning_rate": 0.003, "loss": 4.1094, "step": 8115 }, { "epoch": 0.08116, "grad_norm": 1.5790703567249575, "learning_rate": 0.003, "loss": 4.1277, "step": 8116 }, { "epoch": 0.08117, "grad_norm": 1.3300782853173372, "learning_rate": 0.003, "loss": 4.1266, "step": 8117 }, { "epoch": 0.08118, "grad_norm": 1.0830907844791484, "learning_rate": 0.003, "loss": 4.1059, "step": 8118 }, { "epoch": 0.08119, "grad_norm": 1.0533574564377777, "learning_rate": 0.003, "loss": 4.0962, "step": 8119 }, { "epoch": 0.0812, "grad_norm": 1.0429356977745372, "learning_rate": 0.003, "loss": 4.0771, "step": 8120 }, { "epoch": 0.08121, "grad_norm": 1.0610900493162627, "learning_rate": 0.003, "loss": 4.1047, "step": 8121 }, { "epoch": 0.08122, "grad_norm": 1.0524119493247375, "learning_rate": 0.003, "loss": 4.1004, "step": 8122 }, { "epoch": 0.08123, "grad_norm": 1.054031758211702, "learning_rate": 0.003, "loss": 4.0951, "step": 8123 }, { "epoch": 0.08124, "grad_norm": 1.2911908500296434, "learning_rate": 0.003, "loss": 4.1113, "step": 8124 }, { "epoch": 0.08125, "grad_norm": 1.0027166818569144, "learning_rate": 0.003, "loss": 4.1255, "step": 8125 }, { "epoch": 0.08126, "grad_norm": 1.2287258658702658, "learning_rate": 0.003, "loss": 4.0875, "step": 8126 }, { "epoch": 0.08127, "grad_norm": 1.2494299585369655, "learning_rate": 0.003, "loss": 4.088, "step": 8127 }, { "epoch": 0.08128, "grad_norm": 1.0401015237932247, "learning_rate": 0.003, "loss": 4.1078, "step": 8128 }, { "epoch": 0.08129, "grad_norm": 1.1620438566222164, "learning_rate": 0.003, "loss": 4.0819, "step": 8129 }, { "epoch": 0.0813, "grad_norm": 1.1249593770127122, "learning_rate": 0.003, "loss": 4.0799, "step": 8130 }, { "epoch": 0.08131, "grad_norm": 1.1903713747016034, "learning_rate": 0.003, "loss": 4.1024, "step": 8131 }, { "epoch": 0.08132, "grad_norm": 1.127606083040239, "learning_rate": 0.003, "loss": 4.0916, "step": 8132 }, { "epoch": 0.08133, "grad_norm": 1.0591042809058853, "learning_rate": 0.003, "loss": 4.0833, "step": 8133 }, { "epoch": 0.08134, "grad_norm": 1.0352971364527501, "learning_rate": 0.003, "loss": 4.0817, "step": 8134 }, { "epoch": 0.08135, "grad_norm": 1.0953299118626243, "learning_rate": 0.003, "loss": 4.083, "step": 8135 }, { "epoch": 0.08136, "grad_norm": 1.082062792719415, "learning_rate": 0.003, "loss": 4.1123, "step": 8136 }, { "epoch": 0.08137, "grad_norm": 1.024569523425406, "learning_rate": 0.003, "loss": 4.0526, "step": 8137 }, { "epoch": 0.08138, "grad_norm": 1.0882241989313444, "learning_rate": 0.003, "loss": 4.0832, "step": 8138 }, { "epoch": 0.08139, "grad_norm": 1.0250913668500887, "learning_rate": 0.003, "loss": 4.0933, "step": 8139 }, { "epoch": 0.0814, "grad_norm": 1.1805058404248698, "learning_rate": 0.003, "loss": 4.1085, "step": 8140 }, { "epoch": 0.08141, "grad_norm": 1.0554436593079368, "learning_rate": 0.003, "loss": 4.1024, "step": 8141 }, { "epoch": 0.08142, "grad_norm": 1.1633473045475347, "learning_rate": 0.003, "loss": 4.1162, "step": 8142 }, { "epoch": 0.08143, "grad_norm": 1.0891327651830105, "learning_rate": 0.003, "loss": 4.1083, "step": 8143 }, { "epoch": 0.08144, "grad_norm": 1.2506038115523703, "learning_rate": 0.003, "loss": 4.0813, "step": 8144 }, { "epoch": 0.08145, "grad_norm": 1.0856039619112743, "learning_rate": 0.003, "loss": 4.0849, "step": 8145 }, { "epoch": 0.08146, "grad_norm": 1.1661061378191588, "learning_rate": 0.003, "loss": 4.1186, "step": 8146 }, { "epoch": 0.08147, "grad_norm": 1.1101233976675375, "learning_rate": 0.003, "loss": 4.1183, "step": 8147 }, { "epoch": 0.08148, "grad_norm": 0.8889453883522072, "learning_rate": 0.003, "loss": 4.07, "step": 8148 }, { "epoch": 0.08149, "grad_norm": 0.9830514476631846, "learning_rate": 0.003, "loss": 4.0821, "step": 8149 }, { "epoch": 0.0815, "grad_norm": 1.1043176784528583, "learning_rate": 0.003, "loss": 4.0774, "step": 8150 }, { "epoch": 0.08151, "grad_norm": 1.1479592567463672, "learning_rate": 0.003, "loss": 4.106, "step": 8151 }, { "epoch": 0.08152, "grad_norm": 1.4269496832332111, "learning_rate": 0.003, "loss": 4.1118, "step": 8152 }, { "epoch": 0.08153, "grad_norm": 0.9417973264848024, "learning_rate": 0.003, "loss": 4.0812, "step": 8153 }, { "epoch": 0.08154, "grad_norm": 1.1174301212473075, "learning_rate": 0.003, "loss": 4.099, "step": 8154 }, { "epoch": 0.08155, "grad_norm": 1.1514382782073405, "learning_rate": 0.003, "loss": 4.1144, "step": 8155 }, { "epoch": 0.08156, "grad_norm": 1.2437840352931884, "learning_rate": 0.003, "loss": 4.0977, "step": 8156 }, { "epoch": 0.08157, "grad_norm": 1.24374265378239, "learning_rate": 0.003, "loss": 4.047, "step": 8157 }, { "epoch": 0.08158, "grad_norm": 1.2391145299742439, "learning_rate": 0.003, "loss": 4.1289, "step": 8158 }, { "epoch": 0.08159, "grad_norm": 0.9209362202403073, "learning_rate": 0.003, "loss": 4.1224, "step": 8159 }, { "epoch": 0.0816, "grad_norm": 0.9496193056630835, "learning_rate": 0.003, "loss": 4.0926, "step": 8160 }, { "epoch": 0.08161, "grad_norm": 1.1810669050285874, "learning_rate": 0.003, "loss": 4.1044, "step": 8161 }, { "epoch": 0.08162, "grad_norm": 0.8943628515512069, "learning_rate": 0.003, "loss": 4.0983, "step": 8162 }, { "epoch": 0.08163, "grad_norm": 1.2794592979642219, "learning_rate": 0.003, "loss": 4.0622, "step": 8163 }, { "epoch": 0.08164, "grad_norm": 0.9636518073823616, "learning_rate": 0.003, "loss": 4.0862, "step": 8164 }, { "epoch": 0.08165, "grad_norm": 1.08459249431884, "learning_rate": 0.003, "loss": 4.0933, "step": 8165 }, { "epoch": 0.08166, "grad_norm": 1.0603325319395356, "learning_rate": 0.003, "loss": 4.0951, "step": 8166 }, { "epoch": 0.08167, "grad_norm": 1.1850345601678152, "learning_rate": 0.003, "loss": 4.0985, "step": 8167 }, { "epoch": 0.08168, "grad_norm": 0.95521911177411, "learning_rate": 0.003, "loss": 4.0661, "step": 8168 }, { "epoch": 0.08169, "grad_norm": 1.22286947393468, "learning_rate": 0.003, "loss": 4.106, "step": 8169 }, { "epoch": 0.0817, "grad_norm": 1.0781374703616826, "learning_rate": 0.003, "loss": 4.0951, "step": 8170 }, { "epoch": 0.08171, "grad_norm": 1.1894843647102566, "learning_rate": 0.003, "loss": 4.0843, "step": 8171 }, { "epoch": 0.08172, "grad_norm": 1.0536990256821024, "learning_rate": 0.003, "loss": 4.1077, "step": 8172 }, { "epoch": 0.08173, "grad_norm": 1.2178651572012762, "learning_rate": 0.003, "loss": 4.1314, "step": 8173 }, { "epoch": 0.08174, "grad_norm": 1.0060322968259616, "learning_rate": 0.003, "loss": 4.1235, "step": 8174 }, { "epoch": 0.08175, "grad_norm": 1.1986954728048305, "learning_rate": 0.003, "loss": 4.0941, "step": 8175 }, { "epoch": 0.08176, "grad_norm": 0.9545643256470665, "learning_rate": 0.003, "loss": 4.0664, "step": 8176 }, { "epoch": 0.08177, "grad_norm": 1.0761729611403241, "learning_rate": 0.003, "loss": 4.0799, "step": 8177 }, { "epoch": 0.08178, "grad_norm": 1.0563026676258491, "learning_rate": 0.003, "loss": 4.0921, "step": 8178 }, { "epoch": 0.08179, "grad_norm": 1.2148124269546543, "learning_rate": 0.003, "loss": 4.1227, "step": 8179 }, { "epoch": 0.0818, "grad_norm": 1.169362110071581, "learning_rate": 0.003, "loss": 4.0961, "step": 8180 }, { "epoch": 0.08181, "grad_norm": 1.0955307239949632, "learning_rate": 0.003, "loss": 4.0847, "step": 8181 }, { "epoch": 0.08182, "grad_norm": 1.1144317578395646, "learning_rate": 0.003, "loss": 4.1287, "step": 8182 }, { "epoch": 0.08183, "grad_norm": 1.0295955999206605, "learning_rate": 0.003, "loss": 4.0854, "step": 8183 }, { "epoch": 0.08184, "grad_norm": 1.0947095795077642, "learning_rate": 0.003, "loss": 4.0808, "step": 8184 }, { "epoch": 0.08185, "grad_norm": 1.1371091387650618, "learning_rate": 0.003, "loss": 4.102, "step": 8185 }, { "epoch": 0.08186, "grad_norm": 1.2471904492463743, "learning_rate": 0.003, "loss": 4.1, "step": 8186 }, { "epoch": 0.08187, "grad_norm": 1.1930155717879962, "learning_rate": 0.003, "loss": 4.0693, "step": 8187 }, { "epoch": 0.08188, "grad_norm": 1.0342424236497922, "learning_rate": 0.003, "loss": 4.1164, "step": 8188 }, { "epoch": 0.08189, "grad_norm": 1.0352329127755682, "learning_rate": 0.003, "loss": 4.1117, "step": 8189 }, { "epoch": 0.0819, "grad_norm": 1.1527070200939569, "learning_rate": 0.003, "loss": 4.0883, "step": 8190 }, { "epoch": 0.08191, "grad_norm": 0.8792668377328183, "learning_rate": 0.003, "loss": 4.0855, "step": 8191 }, { "epoch": 0.08192, "grad_norm": 1.0467393413650283, "learning_rate": 0.003, "loss": 4.0502, "step": 8192 }, { "epoch": 0.08193, "grad_norm": 1.2051924602645199, "learning_rate": 0.003, "loss": 4.0952, "step": 8193 }, { "epoch": 0.08194, "grad_norm": 0.9287872876699927, "learning_rate": 0.003, "loss": 4.099, "step": 8194 }, { "epoch": 0.08195, "grad_norm": 1.1076662233195484, "learning_rate": 0.003, "loss": 4.086, "step": 8195 }, { "epoch": 0.08196, "grad_norm": 0.9526697950640942, "learning_rate": 0.003, "loss": 4.1135, "step": 8196 }, { "epoch": 0.08197, "grad_norm": 1.0534752143661108, "learning_rate": 0.003, "loss": 4.0974, "step": 8197 }, { "epoch": 0.08198, "grad_norm": 1.385183829879785, "learning_rate": 0.003, "loss": 4.0958, "step": 8198 }, { "epoch": 0.08199, "grad_norm": 1.0493902630605674, "learning_rate": 0.003, "loss": 4.1001, "step": 8199 }, { "epoch": 0.082, "grad_norm": 1.1705668502686706, "learning_rate": 0.003, "loss": 4.0757, "step": 8200 }, { "epoch": 0.08201, "grad_norm": 1.0066907003222516, "learning_rate": 0.003, "loss": 4.0824, "step": 8201 }, { "epoch": 0.08202, "grad_norm": 1.2820389328790676, "learning_rate": 0.003, "loss": 4.079, "step": 8202 }, { "epoch": 0.08203, "grad_norm": 0.8350201264359747, "learning_rate": 0.003, "loss": 4.1019, "step": 8203 }, { "epoch": 0.08204, "grad_norm": 0.9463964782125038, "learning_rate": 0.003, "loss": 4.0881, "step": 8204 }, { "epoch": 0.08205, "grad_norm": 1.2945714728843314, "learning_rate": 0.003, "loss": 4.0948, "step": 8205 }, { "epoch": 0.08206, "grad_norm": 0.9038682308269772, "learning_rate": 0.003, "loss": 4.1187, "step": 8206 }, { "epoch": 0.08207, "grad_norm": 1.1933488063660513, "learning_rate": 0.003, "loss": 4.0946, "step": 8207 }, { "epoch": 0.08208, "grad_norm": 1.0853521570058549, "learning_rate": 0.003, "loss": 4.0961, "step": 8208 }, { "epoch": 0.08209, "grad_norm": 1.1921605079877917, "learning_rate": 0.003, "loss": 4.1142, "step": 8209 }, { "epoch": 0.0821, "grad_norm": 1.0851985553856465, "learning_rate": 0.003, "loss": 4.0945, "step": 8210 }, { "epoch": 0.08211, "grad_norm": 1.181977847238575, "learning_rate": 0.003, "loss": 4.0972, "step": 8211 }, { "epoch": 0.08212, "grad_norm": 1.1493226942987702, "learning_rate": 0.003, "loss": 4.0946, "step": 8212 }, { "epoch": 0.08213, "grad_norm": 1.046684484682388, "learning_rate": 0.003, "loss": 4.0768, "step": 8213 }, { "epoch": 0.08214, "grad_norm": 1.250975258065935, "learning_rate": 0.003, "loss": 4.114, "step": 8214 }, { "epoch": 0.08215, "grad_norm": 1.154855618057501, "learning_rate": 0.003, "loss": 4.1181, "step": 8215 }, { "epoch": 0.08216, "grad_norm": 1.0257914000202106, "learning_rate": 0.003, "loss": 4.0927, "step": 8216 }, { "epoch": 0.08217, "grad_norm": 1.0800945344509303, "learning_rate": 0.003, "loss": 4.1242, "step": 8217 }, { "epoch": 0.08218, "grad_norm": 1.1371989960222333, "learning_rate": 0.003, "loss": 4.0886, "step": 8218 }, { "epoch": 0.08219, "grad_norm": 1.1092426389119456, "learning_rate": 0.003, "loss": 4.083, "step": 8219 }, { "epoch": 0.0822, "grad_norm": 1.2715561919775418, "learning_rate": 0.003, "loss": 4.097, "step": 8220 }, { "epoch": 0.08221, "grad_norm": 1.0407722220048563, "learning_rate": 0.003, "loss": 4.1117, "step": 8221 }, { "epoch": 0.08222, "grad_norm": 1.3022172548610993, "learning_rate": 0.003, "loss": 4.1023, "step": 8222 }, { "epoch": 0.08223, "grad_norm": 0.8922838680442248, "learning_rate": 0.003, "loss": 4.1074, "step": 8223 }, { "epoch": 0.08224, "grad_norm": 0.8581824078692487, "learning_rate": 0.003, "loss": 4.0988, "step": 8224 }, { "epoch": 0.08225, "grad_norm": 0.8472746859295258, "learning_rate": 0.003, "loss": 4.1115, "step": 8225 }, { "epoch": 0.08226, "grad_norm": 0.9428073992848998, "learning_rate": 0.003, "loss": 4.0752, "step": 8226 }, { "epoch": 0.08227, "grad_norm": 1.2120835674691295, "learning_rate": 0.003, "loss": 4.1225, "step": 8227 }, { "epoch": 0.08228, "grad_norm": 0.9976142438793715, "learning_rate": 0.003, "loss": 4.1058, "step": 8228 }, { "epoch": 0.08229, "grad_norm": 1.449903058079096, "learning_rate": 0.003, "loss": 4.0809, "step": 8229 }, { "epoch": 0.0823, "grad_norm": 0.9384378582006289, "learning_rate": 0.003, "loss": 4.1159, "step": 8230 }, { "epoch": 0.08231, "grad_norm": 1.0676173907140094, "learning_rate": 0.003, "loss": 4.0697, "step": 8231 }, { "epoch": 0.08232, "grad_norm": 1.3410926543201007, "learning_rate": 0.003, "loss": 4.088, "step": 8232 }, { "epoch": 0.08233, "grad_norm": 1.0127914202688533, "learning_rate": 0.003, "loss": 4.0923, "step": 8233 }, { "epoch": 0.08234, "grad_norm": 1.160477985307097, "learning_rate": 0.003, "loss": 4.1098, "step": 8234 }, { "epoch": 0.08235, "grad_norm": 1.0668413961583754, "learning_rate": 0.003, "loss": 4.1278, "step": 8235 }, { "epoch": 0.08236, "grad_norm": 0.9901916985746556, "learning_rate": 0.003, "loss": 4.0785, "step": 8236 }, { "epoch": 0.08237, "grad_norm": 1.0644853370451144, "learning_rate": 0.003, "loss": 4.0944, "step": 8237 }, { "epoch": 0.08238, "grad_norm": 1.004834581027316, "learning_rate": 0.003, "loss": 4.1006, "step": 8238 }, { "epoch": 0.08239, "grad_norm": 1.0868937448101426, "learning_rate": 0.003, "loss": 4.0759, "step": 8239 }, { "epoch": 0.0824, "grad_norm": 1.2795995013293346, "learning_rate": 0.003, "loss": 4.1171, "step": 8240 }, { "epoch": 0.08241, "grad_norm": 1.0885372674354117, "learning_rate": 0.003, "loss": 4.097, "step": 8241 }, { "epoch": 0.08242, "grad_norm": 1.3352614702888232, "learning_rate": 0.003, "loss": 4.0693, "step": 8242 }, { "epoch": 0.08243, "grad_norm": 0.9699411728441731, "learning_rate": 0.003, "loss": 4.116, "step": 8243 }, { "epoch": 0.08244, "grad_norm": 1.0933561697473708, "learning_rate": 0.003, "loss": 4.0892, "step": 8244 }, { "epoch": 0.08245, "grad_norm": 1.338593294596529, "learning_rate": 0.003, "loss": 4.0945, "step": 8245 }, { "epoch": 0.08246, "grad_norm": 0.9397562612350752, "learning_rate": 0.003, "loss": 4.0895, "step": 8246 }, { "epoch": 0.08247, "grad_norm": 1.0486818288187878, "learning_rate": 0.003, "loss": 4.0962, "step": 8247 }, { "epoch": 0.08248, "grad_norm": 1.1040734775180387, "learning_rate": 0.003, "loss": 4.0898, "step": 8248 }, { "epoch": 0.08249, "grad_norm": 1.0083132980867908, "learning_rate": 0.003, "loss": 4.1072, "step": 8249 }, { "epoch": 0.0825, "grad_norm": 1.1974518119321447, "learning_rate": 0.003, "loss": 4.085, "step": 8250 }, { "epoch": 0.08251, "grad_norm": 0.9667133442070936, "learning_rate": 0.003, "loss": 4.1011, "step": 8251 }, { "epoch": 0.08252, "grad_norm": 1.1201280872751502, "learning_rate": 0.003, "loss": 4.1077, "step": 8252 }, { "epoch": 0.08253, "grad_norm": 0.998409555504509, "learning_rate": 0.003, "loss": 4.0994, "step": 8253 }, { "epoch": 0.08254, "grad_norm": 1.3018028543957068, "learning_rate": 0.003, "loss": 4.0891, "step": 8254 }, { "epoch": 0.08255, "grad_norm": 0.9111483127940284, "learning_rate": 0.003, "loss": 4.0764, "step": 8255 }, { "epoch": 0.08256, "grad_norm": 1.0988385606276057, "learning_rate": 0.003, "loss": 4.1132, "step": 8256 }, { "epoch": 0.08257, "grad_norm": 1.3331206787781442, "learning_rate": 0.003, "loss": 4.0929, "step": 8257 }, { "epoch": 0.08258, "grad_norm": 1.118558216630687, "learning_rate": 0.003, "loss": 4.09, "step": 8258 }, { "epoch": 0.08259, "grad_norm": 1.2391360739497914, "learning_rate": 0.003, "loss": 4.1284, "step": 8259 }, { "epoch": 0.0826, "grad_norm": 1.2260663334599986, "learning_rate": 0.003, "loss": 4.0753, "step": 8260 }, { "epoch": 0.08261, "grad_norm": 1.0507114341081063, "learning_rate": 0.003, "loss": 4.0869, "step": 8261 }, { "epoch": 0.08262, "grad_norm": 1.2994656575599821, "learning_rate": 0.003, "loss": 4.1309, "step": 8262 }, { "epoch": 0.08263, "grad_norm": 0.9294316792288047, "learning_rate": 0.003, "loss": 4.0914, "step": 8263 }, { "epoch": 0.08264, "grad_norm": 1.0817062665366421, "learning_rate": 0.003, "loss": 4.1046, "step": 8264 }, { "epoch": 0.08265, "grad_norm": 1.3031114166024542, "learning_rate": 0.003, "loss": 4.0978, "step": 8265 }, { "epoch": 0.08266, "grad_norm": 0.9187895794684968, "learning_rate": 0.003, "loss": 4.0932, "step": 8266 }, { "epoch": 0.08267, "grad_norm": 1.084320300187528, "learning_rate": 0.003, "loss": 4.0977, "step": 8267 }, { "epoch": 0.08268, "grad_norm": 1.3152711121489948, "learning_rate": 0.003, "loss": 4.0858, "step": 8268 }, { "epoch": 0.08269, "grad_norm": 1.0774842752138507, "learning_rate": 0.003, "loss": 4.1153, "step": 8269 }, { "epoch": 0.0827, "grad_norm": 1.1132630336583187, "learning_rate": 0.003, "loss": 4.0905, "step": 8270 }, { "epoch": 0.08271, "grad_norm": 0.9988446582678208, "learning_rate": 0.003, "loss": 4.1084, "step": 8271 }, { "epoch": 0.08272, "grad_norm": 1.1318532368830336, "learning_rate": 0.003, "loss": 4.1008, "step": 8272 }, { "epoch": 0.08273, "grad_norm": 1.2166172102778954, "learning_rate": 0.003, "loss": 4.0936, "step": 8273 }, { "epoch": 0.08274, "grad_norm": 0.9006690473275145, "learning_rate": 0.003, "loss": 4.1048, "step": 8274 }, { "epoch": 0.08275, "grad_norm": 1.2927983262436373, "learning_rate": 0.003, "loss": 4.0868, "step": 8275 }, { "epoch": 0.08276, "grad_norm": 1.2815263368790923, "learning_rate": 0.003, "loss": 4.0764, "step": 8276 }, { "epoch": 0.08277, "grad_norm": 0.8505292867566394, "learning_rate": 0.003, "loss": 4.073, "step": 8277 }, { "epoch": 0.08278, "grad_norm": 1.047826620944776, "learning_rate": 0.003, "loss": 4.0875, "step": 8278 }, { "epoch": 0.08279, "grad_norm": 1.2347360947843489, "learning_rate": 0.003, "loss": 4.0935, "step": 8279 }, { "epoch": 0.0828, "grad_norm": 0.9255694974902187, "learning_rate": 0.003, "loss": 4.1137, "step": 8280 }, { "epoch": 0.08281, "grad_norm": 0.8671297701457938, "learning_rate": 0.003, "loss": 4.0724, "step": 8281 }, { "epoch": 0.08282, "grad_norm": 1.005419393978364, "learning_rate": 0.003, "loss": 4.087, "step": 8282 }, { "epoch": 0.08283, "grad_norm": 1.1740422382225844, "learning_rate": 0.003, "loss": 4.1084, "step": 8283 }, { "epoch": 0.08284, "grad_norm": 1.0602566479523916, "learning_rate": 0.003, "loss": 4.084, "step": 8284 }, { "epoch": 0.08285, "grad_norm": 1.281256167393328, "learning_rate": 0.003, "loss": 4.1069, "step": 8285 }, { "epoch": 0.08286, "grad_norm": 1.2698218556683902, "learning_rate": 0.003, "loss": 4.119, "step": 8286 }, { "epoch": 0.08287, "grad_norm": 1.0905009067792837, "learning_rate": 0.003, "loss": 4.0938, "step": 8287 }, { "epoch": 0.08288, "grad_norm": 1.2351695328697756, "learning_rate": 0.003, "loss": 4.0792, "step": 8288 }, { "epoch": 0.08289, "grad_norm": 0.9062126664137129, "learning_rate": 0.003, "loss": 4.0949, "step": 8289 }, { "epoch": 0.0829, "grad_norm": 0.8229690552504854, "learning_rate": 0.003, "loss": 4.0681, "step": 8290 }, { "epoch": 0.08291, "grad_norm": 1.035719318529572, "learning_rate": 0.003, "loss": 4.0855, "step": 8291 }, { "epoch": 0.08292, "grad_norm": 1.321975709341898, "learning_rate": 0.003, "loss": 4.0988, "step": 8292 }, { "epoch": 0.08293, "grad_norm": 0.956397612925301, "learning_rate": 0.003, "loss": 4.1013, "step": 8293 }, { "epoch": 0.08294, "grad_norm": 1.1005920349573157, "learning_rate": 0.003, "loss": 4.089, "step": 8294 }, { "epoch": 0.08295, "grad_norm": 1.1827646432424894, "learning_rate": 0.003, "loss": 4.0916, "step": 8295 }, { "epoch": 0.08296, "grad_norm": 0.9893580937327027, "learning_rate": 0.003, "loss": 4.0893, "step": 8296 }, { "epoch": 0.08297, "grad_norm": 1.0584769255248867, "learning_rate": 0.003, "loss": 4.1027, "step": 8297 }, { "epoch": 0.08298, "grad_norm": 1.1520347568565248, "learning_rate": 0.003, "loss": 4.0716, "step": 8298 }, { "epoch": 0.08299, "grad_norm": 1.5127165191333787, "learning_rate": 0.003, "loss": 4.0974, "step": 8299 }, { "epoch": 0.083, "grad_norm": 0.988716971696361, "learning_rate": 0.003, "loss": 4.1257, "step": 8300 }, { "epoch": 0.08301, "grad_norm": 1.2189370746107486, "learning_rate": 0.003, "loss": 4.1084, "step": 8301 }, { "epoch": 0.08302, "grad_norm": 1.1627940955744562, "learning_rate": 0.003, "loss": 4.1195, "step": 8302 }, { "epoch": 0.08303, "grad_norm": 1.0548746777232683, "learning_rate": 0.003, "loss": 4.0855, "step": 8303 }, { "epoch": 0.08304, "grad_norm": 1.1160211910286122, "learning_rate": 0.003, "loss": 4.1176, "step": 8304 }, { "epoch": 0.08305, "grad_norm": 1.2052489612764943, "learning_rate": 0.003, "loss": 4.0904, "step": 8305 }, { "epoch": 0.08306, "grad_norm": 1.0824749055776, "learning_rate": 0.003, "loss": 4.0999, "step": 8306 }, { "epoch": 0.08307, "grad_norm": 1.3467892345516064, "learning_rate": 0.003, "loss": 4.0896, "step": 8307 }, { "epoch": 0.08308, "grad_norm": 1.1553201736475212, "learning_rate": 0.003, "loss": 4.0719, "step": 8308 }, { "epoch": 0.08309, "grad_norm": 1.1190267883548892, "learning_rate": 0.003, "loss": 4.0686, "step": 8309 }, { "epoch": 0.0831, "grad_norm": 1.1176202436275529, "learning_rate": 0.003, "loss": 4.0959, "step": 8310 }, { "epoch": 0.08311, "grad_norm": 0.8361916910739546, "learning_rate": 0.003, "loss": 4.0849, "step": 8311 }, { "epoch": 0.08312, "grad_norm": 1.0224578245449776, "learning_rate": 0.003, "loss": 4.1156, "step": 8312 }, { "epoch": 0.08313, "grad_norm": 1.2877054818229416, "learning_rate": 0.003, "loss": 4.0826, "step": 8313 }, { "epoch": 0.08314, "grad_norm": 1.2226376630948521, "learning_rate": 0.003, "loss": 4.11, "step": 8314 }, { "epoch": 0.08315, "grad_norm": 1.1227844150001849, "learning_rate": 0.003, "loss": 4.0629, "step": 8315 }, { "epoch": 0.08316, "grad_norm": 1.0714770724372067, "learning_rate": 0.003, "loss": 4.0843, "step": 8316 }, { "epoch": 0.08317, "grad_norm": 1.156610721764529, "learning_rate": 0.003, "loss": 4.0822, "step": 8317 }, { "epoch": 0.08318, "grad_norm": 1.0610246046060168, "learning_rate": 0.003, "loss": 4.0754, "step": 8318 }, { "epoch": 0.08319, "grad_norm": 1.0429084261396977, "learning_rate": 0.003, "loss": 4.0754, "step": 8319 }, { "epoch": 0.0832, "grad_norm": 1.0243220134183482, "learning_rate": 0.003, "loss": 4.0719, "step": 8320 }, { "epoch": 0.08321, "grad_norm": 1.1448330627177232, "learning_rate": 0.003, "loss": 4.1063, "step": 8321 }, { "epoch": 0.08322, "grad_norm": 1.1972256232351257, "learning_rate": 0.003, "loss": 4.0877, "step": 8322 }, { "epoch": 0.08323, "grad_norm": 1.156455898352581, "learning_rate": 0.003, "loss": 4.0703, "step": 8323 }, { "epoch": 0.08324, "grad_norm": 1.1787235098823745, "learning_rate": 0.003, "loss": 4.0985, "step": 8324 }, { "epoch": 0.08325, "grad_norm": 1.1828573586679785, "learning_rate": 0.003, "loss": 4.1022, "step": 8325 }, { "epoch": 0.08326, "grad_norm": 1.0054340916036382, "learning_rate": 0.003, "loss": 4.0851, "step": 8326 }, { "epoch": 0.08327, "grad_norm": 1.2407155491184743, "learning_rate": 0.003, "loss": 4.0843, "step": 8327 }, { "epoch": 0.08328, "grad_norm": 1.0406068984959647, "learning_rate": 0.003, "loss": 4.0863, "step": 8328 }, { "epoch": 0.08329, "grad_norm": 1.1046752234336579, "learning_rate": 0.003, "loss": 4.0833, "step": 8329 }, { "epoch": 0.0833, "grad_norm": 1.074379949234231, "learning_rate": 0.003, "loss": 4.0985, "step": 8330 }, { "epoch": 0.08331, "grad_norm": 1.3380178938643743, "learning_rate": 0.003, "loss": 4.0894, "step": 8331 }, { "epoch": 0.08332, "grad_norm": 1.0510812462170238, "learning_rate": 0.003, "loss": 4.1039, "step": 8332 }, { "epoch": 0.08333, "grad_norm": 1.1840932096423546, "learning_rate": 0.003, "loss": 4.1081, "step": 8333 }, { "epoch": 0.08334, "grad_norm": 1.0785750284205469, "learning_rate": 0.003, "loss": 4.1039, "step": 8334 }, { "epoch": 0.08335, "grad_norm": 1.1947059923861136, "learning_rate": 0.003, "loss": 4.0992, "step": 8335 }, { "epoch": 0.08336, "grad_norm": 1.0556584760576144, "learning_rate": 0.003, "loss": 4.1212, "step": 8336 }, { "epoch": 0.08337, "grad_norm": 1.072592749257867, "learning_rate": 0.003, "loss": 4.1092, "step": 8337 }, { "epoch": 0.08338, "grad_norm": 0.972461982559215, "learning_rate": 0.003, "loss": 4.0968, "step": 8338 }, { "epoch": 0.08339, "grad_norm": 1.0573142171642655, "learning_rate": 0.003, "loss": 4.0951, "step": 8339 }, { "epoch": 0.0834, "grad_norm": 1.0328161551983648, "learning_rate": 0.003, "loss": 4.0938, "step": 8340 }, { "epoch": 0.08341, "grad_norm": 1.0749745185516024, "learning_rate": 0.003, "loss": 4.1246, "step": 8341 }, { "epoch": 0.08342, "grad_norm": 1.1598258763657079, "learning_rate": 0.003, "loss": 4.0685, "step": 8342 }, { "epoch": 0.08343, "grad_norm": 1.200654151812054, "learning_rate": 0.003, "loss": 4.0766, "step": 8343 }, { "epoch": 0.08344, "grad_norm": 1.074601761075456, "learning_rate": 0.003, "loss": 4.0906, "step": 8344 }, { "epoch": 0.08345, "grad_norm": 1.2407518752084736, "learning_rate": 0.003, "loss": 4.0846, "step": 8345 }, { "epoch": 0.08346, "grad_norm": 0.8878117997949128, "learning_rate": 0.003, "loss": 4.0968, "step": 8346 }, { "epoch": 0.08347, "grad_norm": 1.0334765028748847, "learning_rate": 0.003, "loss": 4.0949, "step": 8347 }, { "epoch": 0.08348, "grad_norm": 0.9992128280318741, "learning_rate": 0.003, "loss": 4.0579, "step": 8348 }, { "epoch": 0.08349, "grad_norm": 1.063209503148335, "learning_rate": 0.003, "loss": 4.1032, "step": 8349 }, { "epoch": 0.0835, "grad_norm": 1.1905818892426057, "learning_rate": 0.003, "loss": 4.1318, "step": 8350 }, { "epoch": 0.08351, "grad_norm": 1.1772176254153641, "learning_rate": 0.003, "loss": 4.1144, "step": 8351 }, { "epoch": 0.08352, "grad_norm": 1.2115245247856177, "learning_rate": 0.003, "loss": 4.1105, "step": 8352 }, { "epoch": 0.08353, "grad_norm": 1.0366342198498524, "learning_rate": 0.003, "loss": 4.0932, "step": 8353 }, { "epoch": 0.08354, "grad_norm": 1.3335689200164156, "learning_rate": 0.003, "loss": 4.104, "step": 8354 }, { "epoch": 0.08355, "grad_norm": 0.9642656094519331, "learning_rate": 0.003, "loss": 4.073, "step": 8355 }, { "epoch": 0.08356, "grad_norm": 1.1411059061577467, "learning_rate": 0.003, "loss": 4.0996, "step": 8356 }, { "epoch": 0.08357, "grad_norm": 0.9934662625503926, "learning_rate": 0.003, "loss": 4.1047, "step": 8357 }, { "epoch": 0.08358, "grad_norm": 1.1371302192567387, "learning_rate": 0.003, "loss": 4.097, "step": 8358 }, { "epoch": 0.08359, "grad_norm": 0.9740680031348249, "learning_rate": 0.003, "loss": 4.0687, "step": 8359 }, { "epoch": 0.0836, "grad_norm": 1.1419609591681779, "learning_rate": 0.003, "loss": 4.0775, "step": 8360 }, { "epoch": 0.08361, "grad_norm": 1.0646469303885506, "learning_rate": 0.003, "loss": 4.0966, "step": 8361 }, { "epoch": 0.08362, "grad_norm": 1.1094407683581557, "learning_rate": 0.003, "loss": 4.1126, "step": 8362 }, { "epoch": 0.08363, "grad_norm": 1.1295628536998819, "learning_rate": 0.003, "loss": 4.1266, "step": 8363 }, { "epoch": 0.08364, "grad_norm": 1.2583129130125206, "learning_rate": 0.003, "loss": 4.0913, "step": 8364 }, { "epoch": 0.08365, "grad_norm": 1.1878894378097133, "learning_rate": 0.003, "loss": 4.1062, "step": 8365 }, { "epoch": 0.08366, "grad_norm": 1.3686033802340978, "learning_rate": 0.003, "loss": 4.1216, "step": 8366 }, { "epoch": 0.08367, "grad_norm": 0.9011404948936148, "learning_rate": 0.003, "loss": 4.092, "step": 8367 }, { "epoch": 0.08368, "grad_norm": 1.0882608213924376, "learning_rate": 0.003, "loss": 4.1092, "step": 8368 }, { "epoch": 0.08369, "grad_norm": 1.3599006033461152, "learning_rate": 0.003, "loss": 4.1054, "step": 8369 }, { "epoch": 0.0837, "grad_norm": 0.7503193168469512, "learning_rate": 0.003, "loss": 4.09, "step": 8370 }, { "epoch": 0.08371, "grad_norm": 0.9831335397603639, "learning_rate": 0.003, "loss": 4.0988, "step": 8371 }, { "epoch": 0.08372, "grad_norm": 1.4830324351803355, "learning_rate": 0.003, "loss": 4.1279, "step": 8372 }, { "epoch": 0.08373, "grad_norm": 0.7578377445625349, "learning_rate": 0.003, "loss": 4.0957, "step": 8373 }, { "epoch": 0.08374, "grad_norm": 1.0871032375475826, "learning_rate": 0.003, "loss": 4.0788, "step": 8374 }, { "epoch": 0.08375, "grad_norm": 1.02193912735729, "learning_rate": 0.003, "loss": 4.0835, "step": 8375 }, { "epoch": 0.08376, "grad_norm": 1.1806307187497338, "learning_rate": 0.003, "loss": 4.0913, "step": 8376 }, { "epoch": 0.08377, "grad_norm": 1.0233421107165972, "learning_rate": 0.003, "loss": 4.0682, "step": 8377 }, { "epoch": 0.08378, "grad_norm": 1.0515534170348302, "learning_rate": 0.003, "loss": 4.0949, "step": 8378 }, { "epoch": 0.08379, "grad_norm": 1.3247691739476297, "learning_rate": 0.003, "loss": 4.095, "step": 8379 }, { "epoch": 0.0838, "grad_norm": 0.9259173435236405, "learning_rate": 0.003, "loss": 4.0864, "step": 8380 }, { "epoch": 0.08381, "grad_norm": 0.9616554946157291, "learning_rate": 0.003, "loss": 4.0938, "step": 8381 }, { "epoch": 0.08382, "grad_norm": 1.038118730756911, "learning_rate": 0.003, "loss": 4.1101, "step": 8382 }, { "epoch": 0.08383, "grad_norm": 1.2117364000112238, "learning_rate": 0.003, "loss": 4.0765, "step": 8383 }, { "epoch": 0.08384, "grad_norm": 0.9896634834173473, "learning_rate": 0.003, "loss": 4.0808, "step": 8384 }, { "epoch": 0.08385, "grad_norm": 1.2713083862055092, "learning_rate": 0.003, "loss": 4.0825, "step": 8385 }, { "epoch": 0.08386, "grad_norm": 1.040476983684724, "learning_rate": 0.003, "loss": 4.0902, "step": 8386 }, { "epoch": 0.08387, "grad_norm": 1.2968269677913182, "learning_rate": 0.003, "loss": 4.0677, "step": 8387 }, { "epoch": 0.08388, "grad_norm": 0.9493766094384419, "learning_rate": 0.003, "loss": 4.063, "step": 8388 }, { "epoch": 0.08389, "grad_norm": 0.9193393415229424, "learning_rate": 0.003, "loss": 4.1042, "step": 8389 }, { "epoch": 0.0839, "grad_norm": 1.2288668412945116, "learning_rate": 0.003, "loss": 4.0747, "step": 8390 }, { "epoch": 0.08391, "grad_norm": 1.3874236422255661, "learning_rate": 0.003, "loss": 4.1135, "step": 8391 }, { "epoch": 0.08392, "grad_norm": 1.107145105182999, "learning_rate": 0.003, "loss": 4.0801, "step": 8392 }, { "epoch": 0.08393, "grad_norm": 1.0476788926336602, "learning_rate": 0.003, "loss": 4.1097, "step": 8393 }, { "epoch": 0.08394, "grad_norm": 1.0534947679685096, "learning_rate": 0.003, "loss": 4.0678, "step": 8394 }, { "epoch": 0.08395, "grad_norm": 1.1208698213755766, "learning_rate": 0.003, "loss": 4.0791, "step": 8395 }, { "epoch": 0.08396, "grad_norm": 0.928086307132362, "learning_rate": 0.003, "loss": 4.1067, "step": 8396 }, { "epoch": 0.08397, "grad_norm": 1.1521765849653989, "learning_rate": 0.003, "loss": 4.0836, "step": 8397 }, { "epoch": 0.08398, "grad_norm": 1.3972790523786438, "learning_rate": 0.003, "loss": 4.1061, "step": 8398 }, { "epoch": 0.08399, "grad_norm": 1.086059690207596, "learning_rate": 0.003, "loss": 4.0891, "step": 8399 }, { "epoch": 0.084, "grad_norm": 1.0596488413781482, "learning_rate": 0.003, "loss": 4.0656, "step": 8400 }, { "epoch": 0.08401, "grad_norm": 1.0555266948649673, "learning_rate": 0.003, "loss": 4.0781, "step": 8401 }, { "epoch": 0.08402, "grad_norm": 1.1373781516408072, "learning_rate": 0.003, "loss": 4.0776, "step": 8402 }, { "epoch": 0.08403, "grad_norm": 0.966784512965234, "learning_rate": 0.003, "loss": 4.0862, "step": 8403 }, { "epoch": 0.08404, "grad_norm": 1.3105083482065643, "learning_rate": 0.003, "loss": 4.089, "step": 8404 }, { "epoch": 0.08405, "grad_norm": 1.1525915739023551, "learning_rate": 0.003, "loss": 4.0798, "step": 8405 }, { "epoch": 0.08406, "grad_norm": 1.0591140746101446, "learning_rate": 0.003, "loss": 4.0661, "step": 8406 }, { "epoch": 0.08407, "grad_norm": 1.2144407868343654, "learning_rate": 0.003, "loss": 4.1449, "step": 8407 }, { "epoch": 0.08408, "grad_norm": 1.1819589739488847, "learning_rate": 0.003, "loss": 4.0694, "step": 8408 }, { "epoch": 0.08409, "grad_norm": 1.0441350913888352, "learning_rate": 0.003, "loss": 4.1072, "step": 8409 }, { "epoch": 0.0841, "grad_norm": 1.1383642745480276, "learning_rate": 0.003, "loss": 4.106, "step": 8410 }, { "epoch": 0.08411, "grad_norm": 1.1073161454597245, "learning_rate": 0.003, "loss": 4.1294, "step": 8411 }, { "epoch": 0.08412, "grad_norm": 1.0480383171994219, "learning_rate": 0.003, "loss": 4.1063, "step": 8412 }, { "epoch": 0.08413, "grad_norm": 1.1492294487682813, "learning_rate": 0.003, "loss": 4.0687, "step": 8413 }, { "epoch": 0.08414, "grad_norm": 1.0210002105972802, "learning_rate": 0.003, "loss": 4.1081, "step": 8414 }, { "epoch": 0.08415, "grad_norm": 1.0465311769849186, "learning_rate": 0.003, "loss": 4.0667, "step": 8415 }, { "epoch": 0.08416, "grad_norm": 1.0796543627431838, "learning_rate": 0.003, "loss": 4.1017, "step": 8416 }, { "epoch": 0.08417, "grad_norm": 1.079430987633751, "learning_rate": 0.003, "loss": 4.1011, "step": 8417 }, { "epoch": 0.08418, "grad_norm": 1.0105681425463648, "learning_rate": 0.003, "loss": 4.0848, "step": 8418 }, { "epoch": 0.08419, "grad_norm": 1.247355704527725, "learning_rate": 0.003, "loss": 4.0965, "step": 8419 }, { "epoch": 0.0842, "grad_norm": 1.2105784595860498, "learning_rate": 0.003, "loss": 4.0776, "step": 8420 }, { "epoch": 0.08421, "grad_norm": 1.1492243371213764, "learning_rate": 0.003, "loss": 4.067, "step": 8421 }, { "epoch": 0.08422, "grad_norm": 0.9577188639693539, "learning_rate": 0.003, "loss": 4.1034, "step": 8422 }, { "epoch": 0.08423, "grad_norm": 1.034036484869677, "learning_rate": 0.003, "loss": 4.0883, "step": 8423 }, { "epoch": 0.08424, "grad_norm": 1.1242868596920017, "learning_rate": 0.003, "loss": 4.0851, "step": 8424 }, { "epoch": 0.08425, "grad_norm": 0.8441325876866451, "learning_rate": 0.003, "loss": 4.0969, "step": 8425 }, { "epoch": 0.08426, "grad_norm": 0.9813337516253374, "learning_rate": 0.003, "loss": 4.0831, "step": 8426 }, { "epoch": 0.08427, "grad_norm": 1.2605532328539706, "learning_rate": 0.003, "loss": 4.1191, "step": 8427 }, { "epoch": 0.08428, "grad_norm": 0.9704205073413669, "learning_rate": 0.003, "loss": 4.0906, "step": 8428 }, { "epoch": 0.08429, "grad_norm": 1.2140012352375567, "learning_rate": 0.003, "loss": 4.0899, "step": 8429 }, { "epoch": 0.0843, "grad_norm": 1.3884303580055675, "learning_rate": 0.003, "loss": 4.0661, "step": 8430 }, { "epoch": 0.08431, "grad_norm": 1.142875897516256, "learning_rate": 0.003, "loss": 4.0802, "step": 8431 }, { "epoch": 0.08432, "grad_norm": 1.1142933348631512, "learning_rate": 0.003, "loss": 4.0667, "step": 8432 }, { "epoch": 0.08433, "grad_norm": 1.1373903002999843, "learning_rate": 0.003, "loss": 4.1065, "step": 8433 }, { "epoch": 0.08434, "grad_norm": 1.1107740861581872, "learning_rate": 0.003, "loss": 4.0555, "step": 8434 }, { "epoch": 0.08435, "grad_norm": 1.2124601254077134, "learning_rate": 0.003, "loss": 4.0916, "step": 8435 }, { "epoch": 0.08436, "grad_norm": 1.0006866274530002, "learning_rate": 0.003, "loss": 4.088, "step": 8436 }, { "epoch": 0.08437, "grad_norm": 1.1576379043773006, "learning_rate": 0.003, "loss": 4.1034, "step": 8437 }, { "epoch": 0.08438, "grad_norm": 1.0862541258017986, "learning_rate": 0.003, "loss": 4.0856, "step": 8438 }, { "epoch": 0.08439, "grad_norm": 0.904884805175504, "learning_rate": 0.003, "loss": 4.0991, "step": 8439 }, { "epoch": 0.0844, "grad_norm": 1.1078688851927403, "learning_rate": 0.003, "loss": 4.1122, "step": 8440 }, { "epoch": 0.08441, "grad_norm": 1.1844909779663761, "learning_rate": 0.003, "loss": 4.0869, "step": 8441 }, { "epoch": 0.08442, "grad_norm": 1.2262528613979544, "learning_rate": 0.003, "loss": 4.0747, "step": 8442 }, { "epoch": 0.08443, "grad_norm": 1.0879224552984719, "learning_rate": 0.003, "loss": 4.0928, "step": 8443 }, { "epoch": 0.08444, "grad_norm": 1.0908579778214895, "learning_rate": 0.003, "loss": 4.1235, "step": 8444 }, { "epoch": 0.08445, "grad_norm": 1.134907437950465, "learning_rate": 0.003, "loss": 4.0726, "step": 8445 }, { "epoch": 0.08446, "grad_norm": 1.0581120143027056, "learning_rate": 0.003, "loss": 4.0914, "step": 8446 }, { "epoch": 0.08447, "grad_norm": 1.1471818659392075, "learning_rate": 0.003, "loss": 4.1032, "step": 8447 }, { "epoch": 0.08448, "grad_norm": 1.0030963036118246, "learning_rate": 0.003, "loss": 4.0991, "step": 8448 }, { "epoch": 0.08449, "grad_norm": 1.0449812545173407, "learning_rate": 0.003, "loss": 4.1094, "step": 8449 }, { "epoch": 0.0845, "grad_norm": 0.9500089816000273, "learning_rate": 0.003, "loss": 4.0781, "step": 8450 }, { "epoch": 0.08451, "grad_norm": 1.0300366626148232, "learning_rate": 0.003, "loss": 4.1075, "step": 8451 }, { "epoch": 0.08452, "grad_norm": 1.0714746920985359, "learning_rate": 0.003, "loss": 4.0871, "step": 8452 }, { "epoch": 0.08453, "grad_norm": 1.094370289054861, "learning_rate": 0.003, "loss": 4.1092, "step": 8453 }, { "epoch": 0.08454, "grad_norm": 1.1120630928897233, "learning_rate": 0.003, "loss": 4.1034, "step": 8454 }, { "epoch": 0.08455, "grad_norm": 1.2560544276208887, "learning_rate": 0.003, "loss": 4.0804, "step": 8455 }, { "epoch": 0.08456, "grad_norm": 0.9752461801922623, "learning_rate": 0.003, "loss": 4.0851, "step": 8456 }, { "epoch": 0.08457, "grad_norm": 1.27526851227219, "learning_rate": 0.003, "loss": 4.1261, "step": 8457 }, { "epoch": 0.08458, "grad_norm": 1.3055821184674892, "learning_rate": 0.003, "loss": 4.0972, "step": 8458 }, { "epoch": 0.08459, "grad_norm": 1.0932295290544094, "learning_rate": 0.003, "loss": 4.0762, "step": 8459 }, { "epoch": 0.0846, "grad_norm": 1.1813495224526782, "learning_rate": 0.003, "loss": 4.1042, "step": 8460 }, { "epoch": 0.08461, "grad_norm": 1.0721351403498962, "learning_rate": 0.003, "loss": 4.1073, "step": 8461 }, { "epoch": 0.08462, "grad_norm": 1.3346700222485708, "learning_rate": 0.003, "loss": 4.1122, "step": 8462 }, { "epoch": 0.08463, "grad_norm": 0.9392142093785416, "learning_rate": 0.003, "loss": 4.0742, "step": 8463 }, { "epoch": 0.08464, "grad_norm": 1.102351898512901, "learning_rate": 0.003, "loss": 4.1139, "step": 8464 }, { "epoch": 0.08465, "grad_norm": 1.0020672839019136, "learning_rate": 0.003, "loss": 4.0931, "step": 8465 }, { "epoch": 0.08466, "grad_norm": 1.2416686816572848, "learning_rate": 0.003, "loss": 4.1161, "step": 8466 }, { "epoch": 0.08467, "grad_norm": 1.011177944354335, "learning_rate": 0.003, "loss": 4.1118, "step": 8467 }, { "epoch": 0.08468, "grad_norm": 0.9939470606527164, "learning_rate": 0.003, "loss": 4.1035, "step": 8468 }, { "epoch": 0.08469, "grad_norm": 1.1517379990943246, "learning_rate": 0.003, "loss": 4.1095, "step": 8469 }, { "epoch": 0.0847, "grad_norm": 1.2134216312298631, "learning_rate": 0.003, "loss": 4.0959, "step": 8470 }, { "epoch": 0.08471, "grad_norm": 1.0402951838186536, "learning_rate": 0.003, "loss": 4.0741, "step": 8471 }, { "epoch": 0.08472, "grad_norm": 1.1880832032771211, "learning_rate": 0.003, "loss": 4.0852, "step": 8472 }, { "epoch": 0.08473, "grad_norm": 1.1386393088356936, "learning_rate": 0.003, "loss": 4.0811, "step": 8473 }, { "epoch": 0.08474, "grad_norm": 1.0623433425885904, "learning_rate": 0.003, "loss": 4.0637, "step": 8474 }, { "epoch": 0.08475, "grad_norm": 1.4615195992095813, "learning_rate": 0.003, "loss": 4.0994, "step": 8475 }, { "epoch": 0.08476, "grad_norm": 1.400034147068075, "learning_rate": 0.003, "loss": 4.1193, "step": 8476 }, { "epoch": 0.08477, "grad_norm": 0.9222788274944164, "learning_rate": 0.003, "loss": 4.0931, "step": 8477 }, { "epoch": 0.08478, "grad_norm": 0.9721124921620972, "learning_rate": 0.003, "loss": 4.0871, "step": 8478 }, { "epoch": 0.08479, "grad_norm": 1.12880457289452, "learning_rate": 0.003, "loss": 4.1093, "step": 8479 }, { "epoch": 0.0848, "grad_norm": 1.0391225480274222, "learning_rate": 0.003, "loss": 4.1078, "step": 8480 }, { "epoch": 0.08481, "grad_norm": 1.188111048658957, "learning_rate": 0.003, "loss": 4.0752, "step": 8481 }, { "epoch": 0.08482, "grad_norm": 1.2272677578307725, "learning_rate": 0.003, "loss": 4.1035, "step": 8482 }, { "epoch": 0.08483, "grad_norm": 0.9551332686819325, "learning_rate": 0.003, "loss": 4.0919, "step": 8483 }, { "epoch": 0.08484, "grad_norm": 1.5024989588505933, "learning_rate": 0.003, "loss": 4.0755, "step": 8484 }, { "epoch": 0.08485, "grad_norm": 1.1577932362053887, "learning_rate": 0.003, "loss": 4.1015, "step": 8485 }, { "epoch": 0.08486, "grad_norm": 1.0658532503557656, "learning_rate": 0.003, "loss": 4.1217, "step": 8486 }, { "epoch": 0.08487, "grad_norm": 1.2131550444413173, "learning_rate": 0.003, "loss": 4.1192, "step": 8487 }, { "epoch": 0.08488, "grad_norm": 1.1266967628313658, "learning_rate": 0.003, "loss": 4.063, "step": 8488 }, { "epoch": 0.08489, "grad_norm": 1.1425330146082786, "learning_rate": 0.003, "loss": 4.1035, "step": 8489 }, { "epoch": 0.0849, "grad_norm": 1.0411187983617134, "learning_rate": 0.003, "loss": 4.1083, "step": 8490 }, { "epoch": 0.08491, "grad_norm": 1.2636596972864644, "learning_rate": 0.003, "loss": 4.1108, "step": 8491 }, { "epoch": 0.08492, "grad_norm": 0.9418480617171554, "learning_rate": 0.003, "loss": 4.0881, "step": 8492 }, { "epoch": 0.08493, "grad_norm": 1.0704335834033358, "learning_rate": 0.003, "loss": 4.0984, "step": 8493 }, { "epoch": 0.08494, "grad_norm": 0.9203629097780305, "learning_rate": 0.003, "loss": 4.0816, "step": 8494 }, { "epoch": 0.08495, "grad_norm": 1.1526449050049299, "learning_rate": 0.003, "loss": 4.0979, "step": 8495 }, { "epoch": 0.08496, "grad_norm": 1.263784431170235, "learning_rate": 0.003, "loss": 4.0708, "step": 8496 }, { "epoch": 0.08497, "grad_norm": 1.0711188981690378, "learning_rate": 0.003, "loss": 4.0863, "step": 8497 }, { "epoch": 0.08498, "grad_norm": 1.247652954241833, "learning_rate": 0.003, "loss": 4.1013, "step": 8498 }, { "epoch": 0.08499, "grad_norm": 1.0480915872768224, "learning_rate": 0.003, "loss": 4.1188, "step": 8499 }, { "epoch": 0.085, "grad_norm": 0.9860067174717102, "learning_rate": 0.003, "loss": 4.0555, "step": 8500 }, { "epoch": 0.08501, "grad_norm": 1.2116292642399038, "learning_rate": 0.003, "loss": 4.0884, "step": 8501 }, { "epoch": 0.08502, "grad_norm": 0.9240148489827791, "learning_rate": 0.003, "loss": 4.0921, "step": 8502 }, { "epoch": 0.08503, "grad_norm": 1.075740001423965, "learning_rate": 0.003, "loss": 4.0939, "step": 8503 }, { "epoch": 0.08504, "grad_norm": 1.2241979651093111, "learning_rate": 0.003, "loss": 4.1259, "step": 8504 }, { "epoch": 0.08505, "grad_norm": 1.1189328001416305, "learning_rate": 0.003, "loss": 4.0857, "step": 8505 }, { "epoch": 0.08506, "grad_norm": 1.2782740508359858, "learning_rate": 0.003, "loss": 4.1063, "step": 8506 }, { "epoch": 0.08507, "grad_norm": 0.9869939621726643, "learning_rate": 0.003, "loss": 4.0713, "step": 8507 }, { "epoch": 0.08508, "grad_norm": 1.3233689266261979, "learning_rate": 0.003, "loss": 4.0876, "step": 8508 }, { "epoch": 0.08509, "grad_norm": 1.1313632101488496, "learning_rate": 0.003, "loss": 4.124, "step": 8509 }, { "epoch": 0.0851, "grad_norm": 1.1719684297456017, "learning_rate": 0.003, "loss": 4.0941, "step": 8510 }, { "epoch": 0.08511, "grad_norm": 1.029569375700711, "learning_rate": 0.003, "loss": 4.0987, "step": 8511 }, { "epoch": 0.08512, "grad_norm": 1.0768752140455045, "learning_rate": 0.003, "loss": 4.1264, "step": 8512 }, { "epoch": 0.08513, "grad_norm": 1.0949952948920811, "learning_rate": 0.003, "loss": 4.0892, "step": 8513 }, { "epoch": 0.08514, "grad_norm": 1.0899619602404949, "learning_rate": 0.003, "loss": 4.0938, "step": 8514 }, { "epoch": 0.08515, "grad_norm": 0.9411338192057704, "learning_rate": 0.003, "loss": 4.0813, "step": 8515 }, { "epoch": 0.08516, "grad_norm": 1.0725560537444447, "learning_rate": 0.003, "loss": 4.1183, "step": 8516 }, { "epoch": 0.08517, "grad_norm": 1.1508250226748429, "learning_rate": 0.003, "loss": 4.0894, "step": 8517 }, { "epoch": 0.08518, "grad_norm": 1.2110527808200953, "learning_rate": 0.003, "loss": 4.0837, "step": 8518 }, { "epoch": 0.08519, "grad_norm": 1.1151939229482213, "learning_rate": 0.003, "loss": 4.0848, "step": 8519 }, { "epoch": 0.0852, "grad_norm": 1.058765428917515, "learning_rate": 0.003, "loss": 4.0616, "step": 8520 }, { "epoch": 0.08521, "grad_norm": 1.562898533414411, "learning_rate": 0.003, "loss": 4.1139, "step": 8521 }, { "epoch": 0.08522, "grad_norm": 0.9343614192489725, "learning_rate": 0.003, "loss": 4.059, "step": 8522 }, { "epoch": 0.08523, "grad_norm": 0.9687273425374774, "learning_rate": 0.003, "loss": 4.0788, "step": 8523 }, { "epoch": 0.08524, "grad_norm": 1.1935185487859767, "learning_rate": 0.003, "loss": 4.0905, "step": 8524 }, { "epoch": 0.08525, "grad_norm": 1.1707008624166617, "learning_rate": 0.003, "loss": 4.0852, "step": 8525 }, { "epoch": 0.08526, "grad_norm": 1.140262331690332, "learning_rate": 0.003, "loss": 4.0881, "step": 8526 }, { "epoch": 0.08527, "grad_norm": 1.1910316786163704, "learning_rate": 0.003, "loss": 4.0907, "step": 8527 }, { "epoch": 0.08528, "grad_norm": 1.25069179782291, "learning_rate": 0.003, "loss": 4.077, "step": 8528 }, { "epoch": 0.08529, "grad_norm": 0.8931136112866865, "learning_rate": 0.003, "loss": 4.092, "step": 8529 }, { "epoch": 0.0853, "grad_norm": 1.098918408213405, "learning_rate": 0.003, "loss": 4.0935, "step": 8530 }, { "epoch": 0.08531, "grad_norm": 1.2764274322869384, "learning_rate": 0.003, "loss": 4.0718, "step": 8531 }, { "epoch": 0.08532, "grad_norm": 1.085018277020142, "learning_rate": 0.003, "loss": 4.072, "step": 8532 }, { "epoch": 0.08533, "grad_norm": 1.4334361641261024, "learning_rate": 0.003, "loss": 4.1118, "step": 8533 }, { "epoch": 0.08534, "grad_norm": 1.004055489226347, "learning_rate": 0.003, "loss": 4.099, "step": 8534 }, { "epoch": 0.08535, "grad_norm": 1.1653627072213197, "learning_rate": 0.003, "loss": 4.0871, "step": 8535 }, { "epoch": 0.08536, "grad_norm": 1.0516909042494484, "learning_rate": 0.003, "loss": 4.1031, "step": 8536 }, { "epoch": 0.08537, "grad_norm": 1.1601019959999184, "learning_rate": 0.003, "loss": 4.0706, "step": 8537 }, { "epoch": 0.08538, "grad_norm": 1.2230647833271613, "learning_rate": 0.003, "loss": 4.077, "step": 8538 }, { "epoch": 0.08539, "grad_norm": 1.3530142917773134, "learning_rate": 0.003, "loss": 4.1189, "step": 8539 }, { "epoch": 0.0854, "grad_norm": 1.1128122417965394, "learning_rate": 0.003, "loss": 4.0636, "step": 8540 }, { "epoch": 0.08541, "grad_norm": 1.0675207047672388, "learning_rate": 0.003, "loss": 4.0729, "step": 8541 }, { "epoch": 0.08542, "grad_norm": 1.0336315662780386, "learning_rate": 0.003, "loss": 4.1081, "step": 8542 }, { "epoch": 0.08543, "grad_norm": 1.2081270118051715, "learning_rate": 0.003, "loss": 4.0998, "step": 8543 }, { "epoch": 0.08544, "grad_norm": 0.9074808211335781, "learning_rate": 0.003, "loss": 4.0733, "step": 8544 }, { "epoch": 0.08545, "grad_norm": 1.0203074076002299, "learning_rate": 0.003, "loss": 4.0844, "step": 8545 }, { "epoch": 0.08546, "grad_norm": 1.3480134319252597, "learning_rate": 0.003, "loss": 4.0996, "step": 8546 }, { "epoch": 0.08547, "grad_norm": 0.8532216900844057, "learning_rate": 0.003, "loss": 4.1037, "step": 8547 }, { "epoch": 0.08548, "grad_norm": 1.0598267597981388, "learning_rate": 0.003, "loss": 4.1104, "step": 8548 }, { "epoch": 0.08549, "grad_norm": 1.1890143144588734, "learning_rate": 0.003, "loss": 4.1074, "step": 8549 }, { "epoch": 0.0855, "grad_norm": 1.0305051389885018, "learning_rate": 0.003, "loss": 4.1136, "step": 8550 }, { "epoch": 0.08551, "grad_norm": 1.2021160867834288, "learning_rate": 0.003, "loss": 4.1151, "step": 8551 }, { "epoch": 0.08552, "grad_norm": 1.229391199817145, "learning_rate": 0.003, "loss": 4.0972, "step": 8552 }, { "epoch": 0.08553, "grad_norm": 1.2277388086687184, "learning_rate": 0.003, "loss": 4.1049, "step": 8553 }, { "epoch": 0.08554, "grad_norm": 1.0009999269129783, "learning_rate": 0.003, "loss": 4.0625, "step": 8554 }, { "epoch": 0.08555, "grad_norm": 1.3281536314404834, "learning_rate": 0.003, "loss": 4.1213, "step": 8555 }, { "epoch": 0.08556, "grad_norm": 1.0941894161427708, "learning_rate": 0.003, "loss": 4.1047, "step": 8556 }, { "epoch": 0.08557, "grad_norm": 1.0074987192557203, "learning_rate": 0.003, "loss": 4.0702, "step": 8557 }, { "epoch": 0.08558, "grad_norm": 1.3120755419752181, "learning_rate": 0.003, "loss": 4.103, "step": 8558 }, { "epoch": 0.08559, "grad_norm": 1.112855109141421, "learning_rate": 0.003, "loss": 4.1002, "step": 8559 }, { "epoch": 0.0856, "grad_norm": 1.1802008883688693, "learning_rate": 0.003, "loss": 4.0972, "step": 8560 }, { "epoch": 0.08561, "grad_norm": 1.1435400866015262, "learning_rate": 0.003, "loss": 4.0884, "step": 8561 }, { "epoch": 0.08562, "grad_norm": 1.2157748046066845, "learning_rate": 0.003, "loss": 4.0658, "step": 8562 }, { "epoch": 0.08563, "grad_norm": 0.9184521377121538, "learning_rate": 0.003, "loss": 4.0872, "step": 8563 }, { "epoch": 0.08564, "grad_norm": 1.3069415671600118, "learning_rate": 0.003, "loss": 4.0896, "step": 8564 }, { "epoch": 0.08565, "grad_norm": 1.2906376644935875, "learning_rate": 0.003, "loss": 4.0804, "step": 8565 }, { "epoch": 0.08566, "grad_norm": 0.8897753639609908, "learning_rate": 0.003, "loss": 4.0906, "step": 8566 }, { "epoch": 0.08567, "grad_norm": 0.9167516320833566, "learning_rate": 0.003, "loss": 4.0966, "step": 8567 }, { "epoch": 0.08568, "grad_norm": 0.9734453972672918, "learning_rate": 0.003, "loss": 4.1135, "step": 8568 }, { "epoch": 0.08569, "grad_norm": 1.1471493076406067, "learning_rate": 0.003, "loss": 4.124, "step": 8569 }, { "epoch": 0.0857, "grad_norm": 1.0305087020424901, "learning_rate": 0.003, "loss": 4.1174, "step": 8570 }, { "epoch": 0.08571, "grad_norm": 1.1530264666685826, "learning_rate": 0.003, "loss": 4.1088, "step": 8571 }, { "epoch": 0.08572, "grad_norm": 1.072811478724214, "learning_rate": 0.003, "loss": 4.0829, "step": 8572 }, { "epoch": 0.08573, "grad_norm": 1.0010697460369764, "learning_rate": 0.003, "loss": 4.0889, "step": 8573 }, { "epoch": 0.08574, "grad_norm": 1.3554961872841174, "learning_rate": 0.003, "loss": 4.0903, "step": 8574 }, { "epoch": 0.08575, "grad_norm": 1.0594679145189, "learning_rate": 0.003, "loss": 4.1074, "step": 8575 }, { "epoch": 0.08576, "grad_norm": 1.531733599884536, "learning_rate": 0.003, "loss": 4.1063, "step": 8576 }, { "epoch": 0.08577, "grad_norm": 0.946013162845909, "learning_rate": 0.003, "loss": 4.0982, "step": 8577 }, { "epoch": 0.08578, "grad_norm": 0.9822723663412125, "learning_rate": 0.003, "loss": 4.0642, "step": 8578 }, { "epoch": 0.08579, "grad_norm": 1.264222600454884, "learning_rate": 0.003, "loss": 4.1001, "step": 8579 }, { "epoch": 0.0858, "grad_norm": 1.2028419299274247, "learning_rate": 0.003, "loss": 4.0905, "step": 8580 }, { "epoch": 0.08581, "grad_norm": 1.166999098358251, "learning_rate": 0.003, "loss": 4.0809, "step": 8581 }, { "epoch": 0.08582, "grad_norm": 1.1946130776246302, "learning_rate": 0.003, "loss": 4.1072, "step": 8582 }, { "epoch": 0.08583, "grad_norm": 0.8275926145019445, "learning_rate": 0.003, "loss": 4.112, "step": 8583 }, { "epoch": 0.08584, "grad_norm": 0.9601828393882058, "learning_rate": 0.003, "loss": 4.1016, "step": 8584 }, { "epoch": 0.08585, "grad_norm": 1.1418282893849723, "learning_rate": 0.003, "loss": 4.1102, "step": 8585 }, { "epoch": 0.08586, "grad_norm": 0.9692499003978388, "learning_rate": 0.003, "loss": 4.1235, "step": 8586 }, { "epoch": 0.08587, "grad_norm": 1.3248373211325242, "learning_rate": 0.003, "loss": 4.1038, "step": 8587 }, { "epoch": 0.08588, "grad_norm": 0.9942213910299912, "learning_rate": 0.003, "loss": 4.0806, "step": 8588 }, { "epoch": 0.08589, "grad_norm": 1.280584293268826, "learning_rate": 0.003, "loss": 4.0997, "step": 8589 }, { "epoch": 0.0859, "grad_norm": 0.8301437868115007, "learning_rate": 0.003, "loss": 4.083, "step": 8590 }, { "epoch": 0.08591, "grad_norm": 0.9369694398863401, "learning_rate": 0.003, "loss": 4.0886, "step": 8591 }, { "epoch": 0.08592, "grad_norm": 1.275721161123058, "learning_rate": 0.003, "loss": 4.1196, "step": 8592 }, { "epoch": 0.08593, "grad_norm": 0.8955252636026287, "learning_rate": 0.003, "loss": 4.1308, "step": 8593 }, { "epoch": 0.08594, "grad_norm": 1.4011732140313142, "learning_rate": 0.003, "loss": 4.0911, "step": 8594 }, { "epoch": 0.08595, "grad_norm": 1.0600791952772353, "learning_rate": 0.003, "loss": 4.0964, "step": 8595 }, { "epoch": 0.08596, "grad_norm": 1.267568542822607, "learning_rate": 0.003, "loss": 4.115, "step": 8596 }, { "epoch": 0.08597, "grad_norm": 1.0586434977398884, "learning_rate": 0.003, "loss": 4.1114, "step": 8597 }, { "epoch": 0.08598, "grad_norm": 1.27561385748398, "learning_rate": 0.003, "loss": 4.0879, "step": 8598 }, { "epoch": 0.08599, "grad_norm": 1.1065342806433314, "learning_rate": 0.003, "loss": 4.1054, "step": 8599 }, { "epoch": 0.086, "grad_norm": 1.1348977213487659, "learning_rate": 0.003, "loss": 4.1282, "step": 8600 }, { "epoch": 0.08601, "grad_norm": 1.176279738052218, "learning_rate": 0.003, "loss": 4.1122, "step": 8601 }, { "epoch": 0.08602, "grad_norm": 1.1064233817022224, "learning_rate": 0.003, "loss": 4.121, "step": 8602 }, { "epoch": 0.08603, "grad_norm": 1.0688250082333606, "learning_rate": 0.003, "loss": 4.1114, "step": 8603 }, { "epoch": 0.08604, "grad_norm": 0.9619884458380277, "learning_rate": 0.003, "loss": 4.0901, "step": 8604 }, { "epoch": 0.08605, "grad_norm": 1.1539583918871326, "learning_rate": 0.003, "loss": 4.1099, "step": 8605 }, { "epoch": 0.08606, "grad_norm": 1.3009621210548399, "learning_rate": 0.003, "loss": 4.0835, "step": 8606 }, { "epoch": 0.08607, "grad_norm": 0.9519847108591609, "learning_rate": 0.003, "loss": 4.1196, "step": 8607 }, { "epoch": 0.08608, "grad_norm": 1.1165576004738105, "learning_rate": 0.003, "loss": 4.095, "step": 8608 }, { "epoch": 0.08609, "grad_norm": 1.2359873790117464, "learning_rate": 0.003, "loss": 4.1239, "step": 8609 }, { "epoch": 0.0861, "grad_norm": 0.8527902351660679, "learning_rate": 0.003, "loss": 4.1134, "step": 8610 }, { "epoch": 0.08611, "grad_norm": 1.0660073775504049, "learning_rate": 0.003, "loss": 4.0948, "step": 8611 }, { "epoch": 0.08612, "grad_norm": 1.1340056451512006, "learning_rate": 0.003, "loss": 4.1349, "step": 8612 }, { "epoch": 0.08613, "grad_norm": 1.3268150264154202, "learning_rate": 0.003, "loss": 4.1125, "step": 8613 }, { "epoch": 0.08614, "grad_norm": 1.1093722096483616, "learning_rate": 0.003, "loss": 4.0914, "step": 8614 }, { "epoch": 0.08615, "grad_norm": 1.3664781719705619, "learning_rate": 0.003, "loss": 4.0977, "step": 8615 }, { "epoch": 0.08616, "grad_norm": 0.9384106102804197, "learning_rate": 0.003, "loss": 4.0728, "step": 8616 }, { "epoch": 0.08617, "grad_norm": 1.0492336110550373, "learning_rate": 0.003, "loss": 4.0834, "step": 8617 }, { "epoch": 0.08618, "grad_norm": 1.1146964560183108, "learning_rate": 0.003, "loss": 4.0983, "step": 8618 }, { "epoch": 0.08619, "grad_norm": 1.194030607125967, "learning_rate": 0.003, "loss": 4.1152, "step": 8619 }, { "epoch": 0.0862, "grad_norm": 1.2682058300146288, "learning_rate": 0.003, "loss": 4.1037, "step": 8620 }, { "epoch": 0.08621, "grad_norm": 1.0546915391496692, "learning_rate": 0.003, "loss": 4.0863, "step": 8621 }, { "epoch": 0.08622, "grad_norm": 1.201336209536038, "learning_rate": 0.003, "loss": 4.1162, "step": 8622 }, { "epoch": 0.08623, "grad_norm": 0.8286267707907774, "learning_rate": 0.003, "loss": 4.0998, "step": 8623 }, { "epoch": 0.08624, "grad_norm": 0.8050569535552256, "learning_rate": 0.003, "loss": 4.0808, "step": 8624 }, { "epoch": 0.08625, "grad_norm": 1.0090223121995023, "learning_rate": 0.003, "loss": 4.0765, "step": 8625 }, { "epoch": 0.08626, "grad_norm": 1.3836049551194456, "learning_rate": 0.003, "loss": 4.1241, "step": 8626 }, { "epoch": 0.08627, "grad_norm": 0.9396004260224532, "learning_rate": 0.003, "loss": 4.0926, "step": 8627 }, { "epoch": 0.08628, "grad_norm": 1.0940158992898368, "learning_rate": 0.003, "loss": 4.1015, "step": 8628 }, { "epoch": 0.08629, "grad_norm": 1.3217515207758788, "learning_rate": 0.003, "loss": 4.0992, "step": 8629 }, { "epoch": 0.0863, "grad_norm": 1.0401740449406907, "learning_rate": 0.003, "loss": 4.0836, "step": 8630 }, { "epoch": 0.08631, "grad_norm": 1.1695727742458422, "learning_rate": 0.003, "loss": 4.1062, "step": 8631 }, { "epoch": 0.08632, "grad_norm": 0.9840368958621316, "learning_rate": 0.003, "loss": 4.0817, "step": 8632 }, { "epoch": 0.08633, "grad_norm": 1.2699264390946416, "learning_rate": 0.003, "loss": 4.1261, "step": 8633 }, { "epoch": 0.08634, "grad_norm": 0.954835252306559, "learning_rate": 0.003, "loss": 4.1046, "step": 8634 }, { "epoch": 0.08635, "grad_norm": 1.0949764020553094, "learning_rate": 0.003, "loss": 4.0972, "step": 8635 }, { "epoch": 0.08636, "grad_norm": 1.2803777327613919, "learning_rate": 0.003, "loss": 4.1016, "step": 8636 }, { "epoch": 0.08637, "grad_norm": 1.0805026136910605, "learning_rate": 0.003, "loss": 4.1243, "step": 8637 }, { "epoch": 0.08638, "grad_norm": 1.1432278312866255, "learning_rate": 0.003, "loss": 4.1139, "step": 8638 }, { "epoch": 0.08639, "grad_norm": 1.1025532522142478, "learning_rate": 0.003, "loss": 4.0922, "step": 8639 }, { "epoch": 0.0864, "grad_norm": 0.9490378890886939, "learning_rate": 0.003, "loss": 4.094, "step": 8640 }, { "epoch": 0.08641, "grad_norm": 1.285895447896166, "learning_rate": 0.003, "loss": 4.1029, "step": 8641 }, { "epoch": 0.08642, "grad_norm": 1.3217026018289355, "learning_rate": 0.003, "loss": 4.1073, "step": 8642 }, { "epoch": 0.08643, "grad_norm": 1.1095193491690498, "learning_rate": 0.003, "loss": 4.0907, "step": 8643 }, { "epoch": 0.08644, "grad_norm": 1.0954213548252971, "learning_rate": 0.003, "loss": 4.0726, "step": 8644 }, { "epoch": 0.08645, "grad_norm": 1.0836314486145702, "learning_rate": 0.003, "loss": 4.0844, "step": 8645 }, { "epoch": 0.08646, "grad_norm": 1.2812668159767462, "learning_rate": 0.003, "loss": 4.0746, "step": 8646 }, { "epoch": 0.08647, "grad_norm": 1.1030847393878342, "learning_rate": 0.003, "loss": 4.0577, "step": 8647 }, { "epoch": 0.08648, "grad_norm": 1.111034140124417, "learning_rate": 0.003, "loss": 4.0948, "step": 8648 }, { "epoch": 0.08649, "grad_norm": 1.1538481841911161, "learning_rate": 0.003, "loss": 4.121, "step": 8649 }, { "epoch": 0.0865, "grad_norm": 1.0152874441487634, "learning_rate": 0.003, "loss": 4.0723, "step": 8650 }, { "epoch": 0.08651, "grad_norm": 1.0994427850562234, "learning_rate": 0.003, "loss": 4.0992, "step": 8651 }, { "epoch": 0.08652, "grad_norm": 1.127861382204831, "learning_rate": 0.003, "loss": 4.0723, "step": 8652 }, { "epoch": 0.08653, "grad_norm": 1.07505411159091, "learning_rate": 0.003, "loss": 4.0779, "step": 8653 }, { "epoch": 0.08654, "grad_norm": 1.3496208467560102, "learning_rate": 0.003, "loss": 4.1008, "step": 8654 }, { "epoch": 0.08655, "grad_norm": 1.01647617649097, "learning_rate": 0.003, "loss": 4.0758, "step": 8655 }, { "epoch": 0.08656, "grad_norm": 1.4209597404124044, "learning_rate": 0.003, "loss": 4.1165, "step": 8656 }, { "epoch": 0.08657, "grad_norm": 0.9690796077947841, "learning_rate": 0.003, "loss": 4.1186, "step": 8657 }, { "epoch": 0.08658, "grad_norm": 1.217443551789007, "learning_rate": 0.003, "loss": 4.0823, "step": 8658 }, { "epoch": 0.08659, "grad_norm": 0.9222153095355206, "learning_rate": 0.003, "loss": 4.0889, "step": 8659 }, { "epoch": 0.0866, "grad_norm": 1.042457381694646, "learning_rate": 0.003, "loss": 4.0858, "step": 8660 }, { "epoch": 0.08661, "grad_norm": 1.28321325520703, "learning_rate": 0.003, "loss": 4.0944, "step": 8661 }, { "epoch": 0.08662, "grad_norm": 0.9921356540740303, "learning_rate": 0.003, "loss": 4.0691, "step": 8662 }, { "epoch": 0.08663, "grad_norm": 1.1816308894921619, "learning_rate": 0.003, "loss": 4.1103, "step": 8663 }, { "epoch": 0.08664, "grad_norm": 0.9604118665551328, "learning_rate": 0.003, "loss": 4.0925, "step": 8664 }, { "epoch": 0.08665, "grad_norm": 1.103577915598327, "learning_rate": 0.003, "loss": 4.0988, "step": 8665 }, { "epoch": 0.08666, "grad_norm": 1.0912951429700728, "learning_rate": 0.003, "loss": 4.0934, "step": 8666 }, { "epoch": 0.08667, "grad_norm": 1.089798083822531, "learning_rate": 0.003, "loss": 4.0828, "step": 8667 }, { "epoch": 0.08668, "grad_norm": 1.0154192284190815, "learning_rate": 0.003, "loss": 4.0826, "step": 8668 }, { "epoch": 0.08669, "grad_norm": 1.1674462530445848, "learning_rate": 0.003, "loss": 4.0868, "step": 8669 }, { "epoch": 0.0867, "grad_norm": 1.2856109028709737, "learning_rate": 0.003, "loss": 4.0897, "step": 8670 }, { "epoch": 0.08671, "grad_norm": 1.0904458700486965, "learning_rate": 0.003, "loss": 4.1003, "step": 8671 }, { "epoch": 0.08672, "grad_norm": 1.4024483060796498, "learning_rate": 0.003, "loss": 4.0912, "step": 8672 }, { "epoch": 0.08673, "grad_norm": 0.8669257920104226, "learning_rate": 0.003, "loss": 4.0916, "step": 8673 }, { "epoch": 0.08674, "grad_norm": 0.8388966263742093, "learning_rate": 0.003, "loss": 4.0967, "step": 8674 }, { "epoch": 0.08675, "grad_norm": 0.9635039333587543, "learning_rate": 0.003, "loss": 4.1026, "step": 8675 }, { "epoch": 0.08676, "grad_norm": 1.2344963320644802, "learning_rate": 0.003, "loss": 4.0805, "step": 8676 }, { "epoch": 0.08677, "grad_norm": 1.0387887353279868, "learning_rate": 0.003, "loss": 4.0786, "step": 8677 }, { "epoch": 0.08678, "grad_norm": 1.1527070962116546, "learning_rate": 0.003, "loss": 4.0902, "step": 8678 }, { "epoch": 0.08679, "grad_norm": 1.1029487438754473, "learning_rate": 0.003, "loss": 4.0995, "step": 8679 }, { "epoch": 0.0868, "grad_norm": 1.3124639540897858, "learning_rate": 0.003, "loss": 4.0647, "step": 8680 }, { "epoch": 0.08681, "grad_norm": 1.2437793839980147, "learning_rate": 0.003, "loss": 4.0833, "step": 8681 }, { "epoch": 0.08682, "grad_norm": 1.0540929823400549, "learning_rate": 0.003, "loss": 4.0873, "step": 8682 }, { "epoch": 0.08683, "grad_norm": 1.2790371263369233, "learning_rate": 0.003, "loss": 4.1144, "step": 8683 }, { "epoch": 0.08684, "grad_norm": 0.8657522896104256, "learning_rate": 0.003, "loss": 4.096, "step": 8684 }, { "epoch": 0.08685, "grad_norm": 1.0517928037944657, "learning_rate": 0.003, "loss": 4.096, "step": 8685 }, { "epoch": 0.08686, "grad_norm": 1.1161496633955428, "learning_rate": 0.003, "loss": 4.1017, "step": 8686 }, { "epoch": 0.08687, "grad_norm": 1.0797615023578369, "learning_rate": 0.003, "loss": 4.1039, "step": 8687 }, { "epoch": 0.08688, "grad_norm": 1.2204131791670518, "learning_rate": 0.003, "loss": 4.094, "step": 8688 }, { "epoch": 0.08689, "grad_norm": 1.1023472021408198, "learning_rate": 0.003, "loss": 4.0548, "step": 8689 }, { "epoch": 0.0869, "grad_norm": 1.2096005457249266, "learning_rate": 0.003, "loss": 4.0995, "step": 8690 }, { "epoch": 0.08691, "grad_norm": 1.0159491010398511, "learning_rate": 0.003, "loss": 4.0716, "step": 8691 }, { "epoch": 0.08692, "grad_norm": 0.9501982626279768, "learning_rate": 0.003, "loss": 4.093, "step": 8692 }, { "epoch": 0.08693, "grad_norm": 1.1527647539457258, "learning_rate": 0.003, "loss": 4.0735, "step": 8693 }, { "epoch": 0.08694, "grad_norm": 1.070928960423378, "learning_rate": 0.003, "loss": 4.0832, "step": 8694 }, { "epoch": 0.08695, "grad_norm": 1.207001157436488, "learning_rate": 0.003, "loss": 4.0868, "step": 8695 }, { "epoch": 0.08696, "grad_norm": 1.053492800660972, "learning_rate": 0.003, "loss": 4.1123, "step": 8696 }, { "epoch": 0.08697, "grad_norm": 1.2985291128404224, "learning_rate": 0.003, "loss": 4.1112, "step": 8697 }, { "epoch": 0.08698, "grad_norm": 1.0631378422403852, "learning_rate": 0.003, "loss": 4.0807, "step": 8698 }, { "epoch": 0.08699, "grad_norm": 1.1823377411491516, "learning_rate": 0.003, "loss": 4.0978, "step": 8699 }, { "epoch": 0.087, "grad_norm": 1.0378537893217477, "learning_rate": 0.003, "loss": 4.1183, "step": 8700 }, { "epoch": 0.08701, "grad_norm": 1.2051775798271545, "learning_rate": 0.003, "loss": 4.1029, "step": 8701 }, { "epoch": 0.08702, "grad_norm": 1.034446479405826, "learning_rate": 0.003, "loss": 4.0901, "step": 8702 }, { "epoch": 0.08703, "grad_norm": 0.9988920281954514, "learning_rate": 0.003, "loss": 4.0746, "step": 8703 }, { "epoch": 0.08704, "grad_norm": 1.1646013905661423, "learning_rate": 0.003, "loss": 4.0572, "step": 8704 }, { "epoch": 0.08705, "grad_norm": 1.0961253024762871, "learning_rate": 0.003, "loss": 4.1131, "step": 8705 }, { "epoch": 0.08706, "grad_norm": 1.1967169806769928, "learning_rate": 0.003, "loss": 4.0852, "step": 8706 }, { "epoch": 0.08707, "grad_norm": 0.9474678629718546, "learning_rate": 0.003, "loss": 4.0836, "step": 8707 }, { "epoch": 0.08708, "grad_norm": 1.1516705610260332, "learning_rate": 0.003, "loss": 4.0959, "step": 8708 }, { "epoch": 0.08709, "grad_norm": 1.2887917026425864, "learning_rate": 0.003, "loss": 4.109, "step": 8709 }, { "epoch": 0.0871, "grad_norm": 1.049402400833251, "learning_rate": 0.003, "loss": 4.083, "step": 8710 }, { "epoch": 0.08711, "grad_norm": 1.267428077933279, "learning_rate": 0.003, "loss": 4.1003, "step": 8711 }, { "epoch": 0.08712, "grad_norm": 0.9108996994268164, "learning_rate": 0.003, "loss": 4.073, "step": 8712 }, { "epoch": 0.08713, "grad_norm": 1.129357093503024, "learning_rate": 0.003, "loss": 4.1034, "step": 8713 }, { "epoch": 0.08714, "grad_norm": 1.1837985448712545, "learning_rate": 0.003, "loss": 4.0932, "step": 8714 }, { "epoch": 0.08715, "grad_norm": 1.0341252713206417, "learning_rate": 0.003, "loss": 4.1002, "step": 8715 }, { "epoch": 0.08716, "grad_norm": 1.195648131905206, "learning_rate": 0.003, "loss": 4.1037, "step": 8716 }, { "epoch": 0.08717, "grad_norm": 1.100628894474599, "learning_rate": 0.003, "loss": 4.1088, "step": 8717 }, { "epoch": 0.08718, "grad_norm": 1.2704156956713528, "learning_rate": 0.003, "loss": 4.1044, "step": 8718 }, { "epoch": 0.08719, "grad_norm": 1.0463156811456091, "learning_rate": 0.003, "loss": 4.0878, "step": 8719 }, { "epoch": 0.0872, "grad_norm": 1.1258377527829428, "learning_rate": 0.003, "loss": 4.0779, "step": 8720 }, { "epoch": 0.08721, "grad_norm": 0.9919059576280482, "learning_rate": 0.003, "loss": 4.097, "step": 8721 }, { "epoch": 0.08722, "grad_norm": 1.1228247316933804, "learning_rate": 0.003, "loss": 4.0886, "step": 8722 }, { "epoch": 0.08723, "grad_norm": 1.2656029984955952, "learning_rate": 0.003, "loss": 4.0874, "step": 8723 }, { "epoch": 0.08724, "grad_norm": 1.2555608900266475, "learning_rate": 0.003, "loss": 4.0932, "step": 8724 }, { "epoch": 0.08725, "grad_norm": 1.0587890320142601, "learning_rate": 0.003, "loss": 4.1012, "step": 8725 }, { "epoch": 0.08726, "grad_norm": 1.1684635337571947, "learning_rate": 0.003, "loss": 4.0849, "step": 8726 }, { "epoch": 0.08727, "grad_norm": 1.0747000877168544, "learning_rate": 0.003, "loss": 4.0767, "step": 8727 }, { "epoch": 0.08728, "grad_norm": 1.0900120073540935, "learning_rate": 0.003, "loss": 4.0612, "step": 8728 }, { "epoch": 0.08729, "grad_norm": 1.045934652730821, "learning_rate": 0.003, "loss": 4.0577, "step": 8729 }, { "epoch": 0.0873, "grad_norm": 1.2476511295842083, "learning_rate": 0.003, "loss": 4.1019, "step": 8730 }, { "epoch": 0.08731, "grad_norm": 1.1697153837448115, "learning_rate": 0.003, "loss": 4.0984, "step": 8731 }, { "epoch": 0.08732, "grad_norm": 1.1538724837794856, "learning_rate": 0.003, "loss": 4.0816, "step": 8732 }, { "epoch": 0.08733, "grad_norm": 1.027194286522456, "learning_rate": 0.003, "loss": 4.0775, "step": 8733 }, { "epoch": 0.08734, "grad_norm": 1.22602238055874, "learning_rate": 0.003, "loss": 4.1164, "step": 8734 }, { "epoch": 0.08735, "grad_norm": 1.1763552197930867, "learning_rate": 0.003, "loss": 4.0976, "step": 8735 }, { "epoch": 0.08736, "grad_norm": 1.0792760623837168, "learning_rate": 0.003, "loss": 4.0989, "step": 8736 }, { "epoch": 0.08737, "grad_norm": 1.1921416599279708, "learning_rate": 0.003, "loss": 4.1186, "step": 8737 }, { "epoch": 0.08738, "grad_norm": 1.1452855639758666, "learning_rate": 0.003, "loss": 4.0733, "step": 8738 }, { "epoch": 0.08739, "grad_norm": 1.0439716447507283, "learning_rate": 0.003, "loss": 4.0912, "step": 8739 }, { "epoch": 0.0874, "grad_norm": 1.0420639689753401, "learning_rate": 0.003, "loss": 4.0739, "step": 8740 }, { "epoch": 0.08741, "grad_norm": 1.0712505938261823, "learning_rate": 0.003, "loss": 4.0769, "step": 8741 }, { "epoch": 0.08742, "grad_norm": 1.0422480953499131, "learning_rate": 0.003, "loss": 4.0771, "step": 8742 }, { "epoch": 0.08743, "grad_norm": 1.1752533479503575, "learning_rate": 0.003, "loss": 4.0746, "step": 8743 }, { "epoch": 0.08744, "grad_norm": 1.0901393452167087, "learning_rate": 0.003, "loss": 4.1135, "step": 8744 }, { "epoch": 0.08745, "grad_norm": 1.2208848202535578, "learning_rate": 0.003, "loss": 4.0933, "step": 8745 }, { "epoch": 0.08746, "grad_norm": 1.0770248033885392, "learning_rate": 0.003, "loss": 4.0957, "step": 8746 }, { "epoch": 0.08747, "grad_norm": 1.354397110036037, "learning_rate": 0.003, "loss": 4.1018, "step": 8747 }, { "epoch": 0.08748, "grad_norm": 1.0312216209666822, "learning_rate": 0.003, "loss": 4.0871, "step": 8748 }, { "epoch": 0.08749, "grad_norm": 1.1323882557666598, "learning_rate": 0.003, "loss": 4.0877, "step": 8749 }, { "epoch": 0.0875, "grad_norm": 1.0381850401908939, "learning_rate": 0.003, "loss": 4.0887, "step": 8750 }, { "epoch": 0.08751, "grad_norm": 1.2178940429665637, "learning_rate": 0.003, "loss": 4.0971, "step": 8751 }, { "epoch": 0.08752, "grad_norm": 0.8993679957101482, "learning_rate": 0.003, "loss": 4.1113, "step": 8752 }, { "epoch": 0.08753, "grad_norm": 1.0370682749447258, "learning_rate": 0.003, "loss": 4.1158, "step": 8753 }, { "epoch": 0.08754, "grad_norm": 1.2042388022732353, "learning_rate": 0.003, "loss": 4.1003, "step": 8754 }, { "epoch": 0.08755, "grad_norm": 1.1184706165321316, "learning_rate": 0.003, "loss": 4.0832, "step": 8755 }, { "epoch": 0.08756, "grad_norm": 1.1561679243035003, "learning_rate": 0.003, "loss": 4.086, "step": 8756 }, { "epoch": 0.08757, "grad_norm": 1.024513208226419, "learning_rate": 0.003, "loss": 4.114, "step": 8757 }, { "epoch": 0.08758, "grad_norm": 1.1184437711252289, "learning_rate": 0.003, "loss": 4.0741, "step": 8758 }, { "epoch": 0.08759, "grad_norm": 1.177181255250541, "learning_rate": 0.003, "loss": 4.0921, "step": 8759 }, { "epoch": 0.0876, "grad_norm": 1.0936241217545724, "learning_rate": 0.003, "loss": 4.0895, "step": 8760 }, { "epoch": 0.08761, "grad_norm": 1.252279806451399, "learning_rate": 0.003, "loss": 4.1255, "step": 8761 }, { "epoch": 0.08762, "grad_norm": 1.2123177273975052, "learning_rate": 0.003, "loss": 4.0732, "step": 8762 }, { "epoch": 0.08763, "grad_norm": 1.0369156484787974, "learning_rate": 0.003, "loss": 4.0699, "step": 8763 }, { "epoch": 0.08764, "grad_norm": 1.0335684893203534, "learning_rate": 0.003, "loss": 4.0691, "step": 8764 }, { "epoch": 0.08765, "grad_norm": 1.1945644551620531, "learning_rate": 0.003, "loss": 4.0542, "step": 8765 }, { "epoch": 0.08766, "grad_norm": 0.9950694287958622, "learning_rate": 0.003, "loss": 4.1037, "step": 8766 }, { "epoch": 0.08767, "grad_norm": 1.371292755326346, "learning_rate": 0.003, "loss": 4.0964, "step": 8767 }, { "epoch": 0.08768, "grad_norm": 0.7677810209753791, "learning_rate": 0.003, "loss": 4.0798, "step": 8768 }, { "epoch": 0.08769, "grad_norm": 0.7426354441341513, "learning_rate": 0.003, "loss": 4.0703, "step": 8769 }, { "epoch": 0.0877, "grad_norm": 0.8480980442986015, "learning_rate": 0.003, "loss": 4.0903, "step": 8770 }, { "epoch": 0.08771, "grad_norm": 1.2161480975231598, "learning_rate": 0.003, "loss": 4.0679, "step": 8771 }, { "epoch": 0.08772, "grad_norm": 1.1051348226912412, "learning_rate": 0.003, "loss": 4.0764, "step": 8772 }, { "epoch": 0.08773, "grad_norm": 1.2177584535671466, "learning_rate": 0.003, "loss": 4.109, "step": 8773 }, { "epoch": 0.08774, "grad_norm": 0.8702308282274693, "learning_rate": 0.003, "loss": 4.0839, "step": 8774 }, { "epoch": 0.08775, "grad_norm": 0.9929217162163861, "learning_rate": 0.003, "loss": 4.0689, "step": 8775 }, { "epoch": 0.08776, "grad_norm": 1.263113346655494, "learning_rate": 0.003, "loss": 4.0913, "step": 8776 }, { "epoch": 0.08777, "grad_norm": 1.3257815908902322, "learning_rate": 0.003, "loss": 4.106, "step": 8777 }, { "epoch": 0.08778, "grad_norm": 1.269385296284594, "learning_rate": 0.003, "loss": 4.0914, "step": 8778 }, { "epoch": 0.08779, "grad_norm": 0.9282639434675246, "learning_rate": 0.003, "loss": 4.1027, "step": 8779 }, { "epoch": 0.0878, "grad_norm": 1.0191602256022942, "learning_rate": 0.003, "loss": 4.0678, "step": 8780 }, { "epoch": 0.08781, "grad_norm": 1.1997723203364192, "learning_rate": 0.003, "loss": 4.0824, "step": 8781 }, { "epoch": 0.08782, "grad_norm": 1.2176855894509315, "learning_rate": 0.003, "loss": 4.083, "step": 8782 }, { "epoch": 0.08783, "grad_norm": 1.1652524255615615, "learning_rate": 0.003, "loss": 4.0994, "step": 8783 }, { "epoch": 0.08784, "grad_norm": 1.2085793286606386, "learning_rate": 0.003, "loss": 4.0881, "step": 8784 }, { "epoch": 0.08785, "grad_norm": 0.9571651778611358, "learning_rate": 0.003, "loss": 4.0552, "step": 8785 }, { "epoch": 0.08786, "grad_norm": 1.1582560830850317, "learning_rate": 0.003, "loss": 4.1107, "step": 8786 }, { "epoch": 0.08787, "grad_norm": 0.9758954458203004, "learning_rate": 0.003, "loss": 4.0706, "step": 8787 }, { "epoch": 0.08788, "grad_norm": 1.135243821696432, "learning_rate": 0.003, "loss": 4.1103, "step": 8788 }, { "epoch": 0.08789, "grad_norm": 1.2150079947158996, "learning_rate": 0.003, "loss": 4.0804, "step": 8789 }, { "epoch": 0.0879, "grad_norm": 1.1623522733964076, "learning_rate": 0.003, "loss": 4.1026, "step": 8790 }, { "epoch": 0.08791, "grad_norm": 1.2156744840046996, "learning_rate": 0.003, "loss": 4.0897, "step": 8791 }, { "epoch": 0.08792, "grad_norm": 0.9980442564190244, "learning_rate": 0.003, "loss": 4.0956, "step": 8792 }, { "epoch": 0.08793, "grad_norm": 1.0903970342772615, "learning_rate": 0.003, "loss": 4.0862, "step": 8793 }, { "epoch": 0.08794, "grad_norm": 1.3109945906614189, "learning_rate": 0.003, "loss": 4.0862, "step": 8794 }, { "epoch": 0.08795, "grad_norm": 1.0586317024452636, "learning_rate": 0.003, "loss": 4.0824, "step": 8795 }, { "epoch": 0.08796, "grad_norm": 1.2308071734445476, "learning_rate": 0.003, "loss": 4.0931, "step": 8796 }, { "epoch": 0.08797, "grad_norm": 1.093667193233928, "learning_rate": 0.003, "loss": 4.0665, "step": 8797 }, { "epoch": 0.08798, "grad_norm": 1.239645975324972, "learning_rate": 0.003, "loss": 4.1101, "step": 8798 }, { "epoch": 0.08799, "grad_norm": 1.1829608196970796, "learning_rate": 0.003, "loss": 4.0648, "step": 8799 }, { "epoch": 0.088, "grad_norm": 1.179310389918988, "learning_rate": 0.003, "loss": 4.111, "step": 8800 }, { "epoch": 0.08801, "grad_norm": 1.0740673093501871, "learning_rate": 0.003, "loss": 4.0513, "step": 8801 }, { "epoch": 0.08802, "grad_norm": 1.0524632137944494, "learning_rate": 0.003, "loss": 4.0994, "step": 8802 }, { "epoch": 0.08803, "grad_norm": 1.2242810222141456, "learning_rate": 0.003, "loss": 4.0675, "step": 8803 }, { "epoch": 0.08804, "grad_norm": 1.0060241245553985, "learning_rate": 0.003, "loss": 4.07, "step": 8804 }, { "epoch": 0.08805, "grad_norm": 1.343503230878226, "learning_rate": 0.003, "loss": 4.0673, "step": 8805 }, { "epoch": 0.08806, "grad_norm": 0.920693396297294, "learning_rate": 0.003, "loss": 4.0738, "step": 8806 }, { "epoch": 0.08807, "grad_norm": 0.9856376499764958, "learning_rate": 0.003, "loss": 4.0929, "step": 8807 }, { "epoch": 0.08808, "grad_norm": 1.3858125490769972, "learning_rate": 0.003, "loss": 4.0886, "step": 8808 }, { "epoch": 0.08809, "grad_norm": 1.0746767723125061, "learning_rate": 0.003, "loss": 4.0844, "step": 8809 }, { "epoch": 0.0881, "grad_norm": 1.232148978357337, "learning_rate": 0.003, "loss": 4.0944, "step": 8810 }, { "epoch": 0.08811, "grad_norm": 0.9811405734907467, "learning_rate": 0.003, "loss": 4.0874, "step": 8811 }, { "epoch": 0.08812, "grad_norm": 1.3329184858707495, "learning_rate": 0.003, "loss": 4.1024, "step": 8812 }, { "epoch": 0.08813, "grad_norm": 0.9265847297855418, "learning_rate": 0.003, "loss": 4.0596, "step": 8813 }, { "epoch": 0.08814, "grad_norm": 0.9570432521573592, "learning_rate": 0.003, "loss": 4.0991, "step": 8814 }, { "epoch": 0.08815, "grad_norm": 1.1665709914992801, "learning_rate": 0.003, "loss": 4.1029, "step": 8815 }, { "epoch": 0.08816, "grad_norm": 0.9874536553232731, "learning_rate": 0.003, "loss": 4.0849, "step": 8816 }, { "epoch": 0.08817, "grad_norm": 1.4382139291553788, "learning_rate": 0.003, "loss": 4.0812, "step": 8817 }, { "epoch": 0.08818, "grad_norm": 1.1605884282639565, "learning_rate": 0.003, "loss": 4.0853, "step": 8818 }, { "epoch": 0.08819, "grad_norm": 1.134115351493515, "learning_rate": 0.003, "loss": 4.1194, "step": 8819 }, { "epoch": 0.0882, "grad_norm": 0.9996745730470702, "learning_rate": 0.003, "loss": 4.0898, "step": 8820 }, { "epoch": 0.08821, "grad_norm": 1.2619153343104192, "learning_rate": 0.003, "loss": 4.1261, "step": 8821 }, { "epoch": 0.08822, "grad_norm": 1.016060665012236, "learning_rate": 0.003, "loss": 4.0566, "step": 8822 }, { "epoch": 0.08823, "grad_norm": 1.2486817945224273, "learning_rate": 0.003, "loss": 4.0931, "step": 8823 }, { "epoch": 0.08824, "grad_norm": 1.0324860890887555, "learning_rate": 0.003, "loss": 4.0859, "step": 8824 }, { "epoch": 0.08825, "grad_norm": 1.154701898108942, "learning_rate": 0.003, "loss": 4.0977, "step": 8825 }, { "epoch": 0.08826, "grad_norm": 1.0402884792853095, "learning_rate": 0.003, "loss": 4.0734, "step": 8826 }, { "epoch": 0.08827, "grad_norm": 1.1370723262382887, "learning_rate": 0.003, "loss": 4.1006, "step": 8827 }, { "epoch": 0.08828, "grad_norm": 1.195942427398043, "learning_rate": 0.003, "loss": 4.1032, "step": 8828 }, { "epoch": 0.08829, "grad_norm": 1.1015741941643187, "learning_rate": 0.003, "loss": 4.0989, "step": 8829 }, { "epoch": 0.0883, "grad_norm": 1.24287314028591, "learning_rate": 0.003, "loss": 4.0909, "step": 8830 }, { "epoch": 0.08831, "grad_norm": 1.1540903706193593, "learning_rate": 0.003, "loss": 4.0897, "step": 8831 }, { "epoch": 0.08832, "grad_norm": 1.0841781176133298, "learning_rate": 0.003, "loss": 4.0644, "step": 8832 }, { "epoch": 0.08833, "grad_norm": 1.1620178341397036, "learning_rate": 0.003, "loss": 4.0847, "step": 8833 }, { "epoch": 0.08834, "grad_norm": 1.285096206379007, "learning_rate": 0.003, "loss": 4.0804, "step": 8834 }, { "epoch": 0.08835, "grad_norm": 1.2220110993481752, "learning_rate": 0.003, "loss": 4.1159, "step": 8835 }, { "epoch": 0.08836, "grad_norm": 1.2824438182664897, "learning_rate": 0.003, "loss": 4.1028, "step": 8836 }, { "epoch": 0.08837, "grad_norm": 1.030557537546806, "learning_rate": 0.003, "loss": 4.0751, "step": 8837 }, { "epoch": 0.08838, "grad_norm": 1.2592905047467666, "learning_rate": 0.003, "loss": 4.1013, "step": 8838 }, { "epoch": 0.08839, "grad_norm": 0.9948192453122057, "learning_rate": 0.003, "loss": 4.0826, "step": 8839 }, { "epoch": 0.0884, "grad_norm": 0.9604246294508401, "learning_rate": 0.003, "loss": 4.0766, "step": 8840 }, { "epoch": 0.08841, "grad_norm": 1.0821685274255017, "learning_rate": 0.003, "loss": 4.104, "step": 8841 }, { "epoch": 0.08842, "grad_norm": 1.2149930539564309, "learning_rate": 0.003, "loss": 4.0846, "step": 8842 }, { "epoch": 0.08843, "grad_norm": 1.1535664957187073, "learning_rate": 0.003, "loss": 4.0827, "step": 8843 }, { "epoch": 0.08844, "grad_norm": 1.0643701786404371, "learning_rate": 0.003, "loss": 4.0685, "step": 8844 }, { "epoch": 0.08845, "grad_norm": 1.1519022410200033, "learning_rate": 0.003, "loss": 4.054, "step": 8845 }, { "epoch": 0.08846, "grad_norm": 1.3347722844417247, "learning_rate": 0.003, "loss": 4.128, "step": 8846 }, { "epoch": 0.08847, "grad_norm": 1.045478931502332, "learning_rate": 0.003, "loss": 4.1145, "step": 8847 }, { "epoch": 0.08848, "grad_norm": 1.0931523912517, "learning_rate": 0.003, "loss": 4.113, "step": 8848 }, { "epoch": 0.08849, "grad_norm": 1.085871448202411, "learning_rate": 0.003, "loss": 4.0871, "step": 8849 }, { "epoch": 0.0885, "grad_norm": 0.9450454887357006, "learning_rate": 0.003, "loss": 4.0968, "step": 8850 }, { "epoch": 0.08851, "grad_norm": 1.0428404856331992, "learning_rate": 0.003, "loss": 4.1004, "step": 8851 }, { "epoch": 0.08852, "grad_norm": 1.1178846171093337, "learning_rate": 0.003, "loss": 4.0962, "step": 8852 }, { "epoch": 0.08853, "grad_norm": 1.0673247680057139, "learning_rate": 0.003, "loss": 4.1065, "step": 8853 }, { "epoch": 0.08854, "grad_norm": 1.4370218662311567, "learning_rate": 0.003, "loss": 4.1298, "step": 8854 }, { "epoch": 0.08855, "grad_norm": 1.2424246600081954, "learning_rate": 0.003, "loss": 4.0846, "step": 8855 }, { "epoch": 0.08856, "grad_norm": 1.1952803481942424, "learning_rate": 0.003, "loss": 4.1005, "step": 8856 }, { "epoch": 0.08857, "grad_norm": 1.0610730671993456, "learning_rate": 0.003, "loss": 4.0995, "step": 8857 }, { "epoch": 0.08858, "grad_norm": 1.204237016674479, "learning_rate": 0.003, "loss": 4.0886, "step": 8858 }, { "epoch": 0.08859, "grad_norm": 0.985430125390967, "learning_rate": 0.003, "loss": 4.0963, "step": 8859 }, { "epoch": 0.0886, "grad_norm": 1.1113027683439096, "learning_rate": 0.003, "loss": 4.1115, "step": 8860 }, { "epoch": 0.08861, "grad_norm": 1.0509820877432436, "learning_rate": 0.003, "loss": 4.0826, "step": 8861 }, { "epoch": 0.08862, "grad_norm": 1.0753189064520456, "learning_rate": 0.003, "loss": 4.0787, "step": 8862 }, { "epoch": 0.08863, "grad_norm": 1.146796776128645, "learning_rate": 0.003, "loss": 4.0968, "step": 8863 }, { "epoch": 0.08864, "grad_norm": 1.2419423716308708, "learning_rate": 0.003, "loss": 4.1052, "step": 8864 }, { "epoch": 0.08865, "grad_norm": 0.99866038915533, "learning_rate": 0.003, "loss": 4.0975, "step": 8865 }, { "epoch": 0.08866, "grad_norm": 1.223417079345496, "learning_rate": 0.003, "loss": 4.0742, "step": 8866 }, { "epoch": 0.08867, "grad_norm": 0.9179947374618549, "learning_rate": 0.003, "loss": 4.0992, "step": 8867 }, { "epoch": 0.08868, "grad_norm": 1.0120882698093776, "learning_rate": 0.003, "loss": 4.076, "step": 8868 }, { "epoch": 0.08869, "grad_norm": 1.4761625946066304, "learning_rate": 0.003, "loss": 4.1472, "step": 8869 }, { "epoch": 0.0887, "grad_norm": 1.0013311950887647, "learning_rate": 0.003, "loss": 4.0725, "step": 8870 }, { "epoch": 0.08871, "grad_norm": 1.1905133400701076, "learning_rate": 0.003, "loss": 4.0677, "step": 8871 }, { "epoch": 0.08872, "grad_norm": 1.225762665576794, "learning_rate": 0.003, "loss": 4.1298, "step": 8872 }, { "epoch": 0.08873, "grad_norm": 1.1051400960493805, "learning_rate": 0.003, "loss": 4.0781, "step": 8873 }, { "epoch": 0.08874, "grad_norm": 1.1725074717579642, "learning_rate": 0.003, "loss": 4.0912, "step": 8874 }, { "epoch": 0.08875, "grad_norm": 1.1425706121868302, "learning_rate": 0.003, "loss": 4.1095, "step": 8875 }, { "epoch": 0.08876, "grad_norm": 1.1324360400730162, "learning_rate": 0.003, "loss": 4.0652, "step": 8876 }, { "epoch": 0.08877, "grad_norm": 1.1193149417853414, "learning_rate": 0.003, "loss": 4.0748, "step": 8877 }, { "epoch": 0.08878, "grad_norm": 1.1813382330480562, "learning_rate": 0.003, "loss": 4.0889, "step": 8878 }, { "epoch": 0.08879, "grad_norm": 1.0591929235147335, "learning_rate": 0.003, "loss": 4.078, "step": 8879 }, { "epoch": 0.0888, "grad_norm": 1.1938270432922649, "learning_rate": 0.003, "loss": 4.0771, "step": 8880 }, { "epoch": 0.08881, "grad_norm": 1.15951980851196, "learning_rate": 0.003, "loss": 4.0637, "step": 8881 }, { "epoch": 0.08882, "grad_norm": 1.1269173518743505, "learning_rate": 0.003, "loss": 4.0972, "step": 8882 }, { "epoch": 0.08883, "grad_norm": 0.9256998003722761, "learning_rate": 0.003, "loss": 4.1316, "step": 8883 }, { "epoch": 0.08884, "grad_norm": 0.9928338510945068, "learning_rate": 0.003, "loss": 4.0851, "step": 8884 }, { "epoch": 0.08885, "grad_norm": 1.235862690491941, "learning_rate": 0.003, "loss": 4.0845, "step": 8885 }, { "epoch": 0.08886, "grad_norm": 1.025957142318475, "learning_rate": 0.003, "loss": 4.0883, "step": 8886 }, { "epoch": 0.08887, "grad_norm": 1.1645758366066534, "learning_rate": 0.003, "loss": 4.0838, "step": 8887 }, { "epoch": 0.08888, "grad_norm": 1.0979405833970721, "learning_rate": 0.003, "loss": 4.0793, "step": 8888 }, { "epoch": 0.08889, "grad_norm": 1.1926742138622266, "learning_rate": 0.003, "loss": 4.0805, "step": 8889 }, { "epoch": 0.0889, "grad_norm": 0.9283655625939753, "learning_rate": 0.003, "loss": 4.078, "step": 8890 }, { "epoch": 0.08891, "grad_norm": 0.9922525991886726, "learning_rate": 0.003, "loss": 4.0583, "step": 8891 }, { "epoch": 0.08892, "grad_norm": 1.2511385500851198, "learning_rate": 0.003, "loss": 4.1119, "step": 8892 }, { "epoch": 0.08893, "grad_norm": 1.0877694715722532, "learning_rate": 0.003, "loss": 4.0821, "step": 8893 }, { "epoch": 0.08894, "grad_norm": 1.2425342003563593, "learning_rate": 0.003, "loss": 4.0741, "step": 8894 }, { "epoch": 0.08895, "grad_norm": 1.1708965119096941, "learning_rate": 0.003, "loss": 4.086, "step": 8895 }, { "epoch": 0.08896, "grad_norm": 1.148852163720437, "learning_rate": 0.003, "loss": 4.0681, "step": 8896 }, { "epoch": 0.08897, "grad_norm": 0.9829003899946172, "learning_rate": 0.003, "loss": 4.0716, "step": 8897 }, { "epoch": 0.08898, "grad_norm": 1.1577433657757135, "learning_rate": 0.003, "loss": 4.0883, "step": 8898 }, { "epoch": 0.08899, "grad_norm": 1.0212815115160019, "learning_rate": 0.003, "loss": 4.086, "step": 8899 }, { "epoch": 0.089, "grad_norm": 1.304229793351174, "learning_rate": 0.003, "loss": 4.0806, "step": 8900 }, { "epoch": 0.08901, "grad_norm": 0.9490335114750387, "learning_rate": 0.003, "loss": 4.0876, "step": 8901 }, { "epoch": 0.08902, "grad_norm": 1.0053164397125514, "learning_rate": 0.003, "loss": 4.09, "step": 8902 }, { "epoch": 0.08903, "grad_norm": 1.2031637408189797, "learning_rate": 0.003, "loss": 4.0866, "step": 8903 }, { "epoch": 0.08904, "grad_norm": 1.0348101241532264, "learning_rate": 0.003, "loss": 4.0945, "step": 8904 }, { "epoch": 0.08905, "grad_norm": 1.2411377788704367, "learning_rate": 0.003, "loss": 4.1106, "step": 8905 }, { "epoch": 0.08906, "grad_norm": 1.2368523131379077, "learning_rate": 0.003, "loss": 4.0982, "step": 8906 }, { "epoch": 0.08907, "grad_norm": 1.196052564590542, "learning_rate": 0.003, "loss": 4.0904, "step": 8907 }, { "epoch": 0.08908, "grad_norm": 1.149069619968232, "learning_rate": 0.003, "loss": 4.1105, "step": 8908 }, { "epoch": 0.08909, "grad_norm": 1.2196952014342493, "learning_rate": 0.003, "loss": 4.106, "step": 8909 }, { "epoch": 0.0891, "grad_norm": 0.9657789763518829, "learning_rate": 0.003, "loss": 4.0823, "step": 8910 }, { "epoch": 0.08911, "grad_norm": 1.226501539468004, "learning_rate": 0.003, "loss": 4.0923, "step": 8911 }, { "epoch": 0.08912, "grad_norm": 0.9831269607454067, "learning_rate": 0.003, "loss": 4.0866, "step": 8912 }, { "epoch": 0.08913, "grad_norm": 1.0432328581446457, "learning_rate": 0.003, "loss": 4.1081, "step": 8913 }, { "epoch": 0.08914, "grad_norm": 1.2445813702184239, "learning_rate": 0.003, "loss": 4.1051, "step": 8914 }, { "epoch": 0.08915, "grad_norm": 0.8348608288238978, "learning_rate": 0.003, "loss": 4.1046, "step": 8915 }, { "epoch": 0.08916, "grad_norm": 1.0277019185383869, "learning_rate": 0.003, "loss": 4.0971, "step": 8916 }, { "epoch": 0.08917, "grad_norm": 1.29836858008687, "learning_rate": 0.003, "loss": 4.0874, "step": 8917 }, { "epoch": 0.08918, "grad_norm": 1.1457493663465181, "learning_rate": 0.003, "loss": 4.0933, "step": 8918 }, { "epoch": 0.08919, "grad_norm": 1.3902905052344592, "learning_rate": 0.003, "loss": 4.1088, "step": 8919 }, { "epoch": 0.0892, "grad_norm": 1.1302254715099984, "learning_rate": 0.003, "loss": 4.0758, "step": 8920 }, { "epoch": 0.08921, "grad_norm": 1.144164260144895, "learning_rate": 0.003, "loss": 4.0787, "step": 8921 }, { "epoch": 0.08922, "grad_norm": 1.2102096555366344, "learning_rate": 0.003, "loss": 4.107, "step": 8922 }, { "epoch": 0.08923, "grad_norm": 1.2889411146695433, "learning_rate": 0.003, "loss": 4.1137, "step": 8923 }, { "epoch": 0.08924, "grad_norm": 1.1914185822394034, "learning_rate": 0.003, "loss": 4.0848, "step": 8924 }, { "epoch": 0.08925, "grad_norm": 1.1563115618267714, "learning_rate": 0.003, "loss": 4.0924, "step": 8925 }, { "epoch": 0.08926, "grad_norm": 1.0599030969488006, "learning_rate": 0.003, "loss": 4.0757, "step": 8926 }, { "epoch": 0.08927, "grad_norm": 1.2872549938271667, "learning_rate": 0.003, "loss": 4.0944, "step": 8927 }, { "epoch": 0.08928, "grad_norm": 1.0434873158845894, "learning_rate": 0.003, "loss": 4.0811, "step": 8928 }, { "epoch": 0.08929, "grad_norm": 1.0870121366596754, "learning_rate": 0.003, "loss": 4.0899, "step": 8929 }, { "epoch": 0.0893, "grad_norm": 1.1211050004018694, "learning_rate": 0.003, "loss": 4.0994, "step": 8930 }, { "epoch": 0.08931, "grad_norm": 1.065437355820969, "learning_rate": 0.003, "loss": 4.0812, "step": 8931 }, { "epoch": 0.08932, "grad_norm": 1.4361757687132564, "learning_rate": 0.003, "loss": 4.1038, "step": 8932 }, { "epoch": 0.08933, "grad_norm": 1.2264163195022284, "learning_rate": 0.003, "loss": 4.0944, "step": 8933 }, { "epoch": 0.08934, "grad_norm": 0.9643110026877505, "learning_rate": 0.003, "loss": 4.0858, "step": 8934 }, { "epoch": 0.08935, "grad_norm": 1.0313524255500566, "learning_rate": 0.003, "loss": 4.1024, "step": 8935 }, { "epoch": 0.08936, "grad_norm": 1.100550149766114, "learning_rate": 0.003, "loss": 4.0665, "step": 8936 }, { "epoch": 0.08937, "grad_norm": 1.074917096686148, "learning_rate": 0.003, "loss": 4.0701, "step": 8937 }, { "epoch": 0.08938, "grad_norm": 1.3433517938631507, "learning_rate": 0.003, "loss": 4.1159, "step": 8938 }, { "epoch": 0.08939, "grad_norm": 0.8971118463380504, "learning_rate": 0.003, "loss": 4.1223, "step": 8939 }, { "epoch": 0.0894, "grad_norm": 1.0390958399705494, "learning_rate": 0.003, "loss": 4.0878, "step": 8940 }, { "epoch": 0.08941, "grad_norm": 1.2711873162688037, "learning_rate": 0.003, "loss": 4.1062, "step": 8941 }, { "epoch": 0.08942, "grad_norm": 1.1680587830752844, "learning_rate": 0.003, "loss": 4.0806, "step": 8942 }, { "epoch": 0.08943, "grad_norm": 1.077559364926768, "learning_rate": 0.003, "loss": 4.0947, "step": 8943 }, { "epoch": 0.08944, "grad_norm": 1.1778138324586276, "learning_rate": 0.003, "loss": 4.0692, "step": 8944 }, { "epoch": 0.08945, "grad_norm": 1.0950154110091395, "learning_rate": 0.003, "loss": 4.0637, "step": 8945 }, { "epoch": 0.08946, "grad_norm": 0.9413771676176436, "learning_rate": 0.003, "loss": 4.1107, "step": 8946 }, { "epoch": 0.08947, "grad_norm": 1.116304589986322, "learning_rate": 0.003, "loss": 4.0922, "step": 8947 }, { "epoch": 0.08948, "grad_norm": 1.1348251492332606, "learning_rate": 0.003, "loss": 4.0613, "step": 8948 }, { "epoch": 0.08949, "grad_norm": 1.0167111128911273, "learning_rate": 0.003, "loss": 4.0826, "step": 8949 }, { "epoch": 0.0895, "grad_norm": 1.1114205516835516, "learning_rate": 0.003, "loss": 4.0808, "step": 8950 }, { "epoch": 0.08951, "grad_norm": 1.0311274627489682, "learning_rate": 0.003, "loss": 4.0813, "step": 8951 }, { "epoch": 0.08952, "grad_norm": 1.2420772425442637, "learning_rate": 0.003, "loss": 4.1023, "step": 8952 }, { "epoch": 0.08953, "grad_norm": 0.9702294606818496, "learning_rate": 0.003, "loss": 4.0894, "step": 8953 }, { "epoch": 0.08954, "grad_norm": 1.0514539971922081, "learning_rate": 0.003, "loss": 4.0544, "step": 8954 }, { "epoch": 0.08955, "grad_norm": 1.1175561786688144, "learning_rate": 0.003, "loss": 4.089, "step": 8955 }, { "epoch": 0.08956, "grad_norm": 1.3697725831400098, "learning_rate": 0.003, "loss": 4.0985, "step": 8956 }, { "epoch": 0.08957, "grad_norm": 0.8881912667968426, "learning_rate": 0.003, "loss": 4.0987, "step": 8957 }, { "epoch": 0.08958, "grad_norm": 1.2990213460731657, "learning_rate": 0.003, "loss": 4.0627, "step": 8958 }, { "epoch": 0.08959, "grad_norm": 1.1275379507438985, "learning_rate": 0.003, "loss": 4.0828, "step": 8959 }, { "epoch": 0.0896, "grad_norm": 1.1686764581323634, "learning_rate": 0.003, "loss": 4.0569, "step": 8960 }, { "epoch": 0.08961, "grad_norm": 1.2008879374871733, "learning_rate": 0.003, "loss": 4.0803, "step": 8961 }, { "epoch": 0.08962, "grad_norm": 0.9166635174392294, "learning_rate": 0.003, "loss": 4.0787, "step": 8962 }, { "epoch": 0.08963, "grad_norm": 0.9529723483512812, "learning_rate": 0.003, "loss": 4.1113, "step": 8963 }, { "epoch": 0.08964, "grad_norm": 1.1181882536744028, "learning_rate": 0.003, "loss": 4.0946, "step": 8964 }, { "epoch": 0.08965, "grad_norm": 1.1324629678961362, "learning_rate": 0.003, "loss": 4.1065, "step": 8965 }, { "epoch": 0.08966, "grad_norm": 1.04799498443996, "learning_rate": 0.003, "loss": 4.0741, "step": 8966 }, { "epoch": 0.08967, "grad_norm": 1.193797147891738, "learning_rate": 0.003, "loss": 4.1212, "step": 8967 }, { "epoch": 0.08968, "grad_norm": 1.0573943300927882, "learning_rate": 0.003, "loss": 4.0876, "step": 8968 }, { "epoch": 0.08969, "grad_norm": 1.386096847311672, "learning_rate": 0.003, "loss": 4.0925, "step": 8969 }, { "epoch": 0.0897, "grad_norm": 1.2251660107604352, "learning_rate": 0.003, "loss": 4.1001, "step": 8970 }, { "epoch": 0.08971, "grad_norm": 1.170614763215056, "learning_rate": 0.003, "loss": 4.0892, "step": 8971 }, { "epoch": 0.08972, "grad_norm": 1.1175989967353661, "learning_rate": 0.003, "loss": 4.077, "step": 8972 }, { "epoch": 0.08973, "grad_norm": 1.233829185362789, "learning_rate": 0.003, "loss": 4.1143, "step": 8973 }, { "epoch": 0.08974, "grad_norm": 1.109937499784889, "learning_rate": 0.003, "loss": 4.1054, "step": 8974 }, { "epoch": 0.08975, "grad_norm": 1.210997922226824, "learning_rate": 0.003, "loss": 4.0895, "step": 8975 }, { "epoch": 0.08976, "grad_norm": 1.00399822862685, "learning_rate": 0.003, "loss": 4.0985, "step": 8976 }, { "epoch": 0.08977, "grad_norm": 1.183420824192127, "learning_rate": 0.003, "loss": 4.0892, "step": 8977 }, { "epoch": 0.08978, "grad_norm": 1.2611176663658683, "learning_rate": 0.003, "loss": 4.0958, "step": 8978 }, { "epoch": 0.08979, "grad_norm": 1.0685727933905735, "learning_rate": 0.003, "loss": 4.0714, "step": 8979 }, { "epoch": 0.0898, "grad_norm": 1.1240486549449662, "learning_rate": 0.003, "loss": 4.0956, "step": 8980 }, { "epoch": 0.08981, "grad_norm": 1.306291543933293, "learning_rate": 0.003, "loss": 4.074, "step": 8981 }, { "epoch": 0.08982, "grad_norm": 0.8432622834426845, "learning_rate": 0.003, "loss": 4.0747, "step": 8982 }, { "epoch": 0.08983, "grad_norm": 0.9177109499908207, "learning_rate": 0.003, "loss": 4.0595, "step": 8983 }, { "epoch": 0.08984, "grad_norm": 1.2277287537315837, "learning_rate": 0.003, "loss": 4.1126, "step": 8984 }, { "epoch": 0.08985, "grad_norm": 1.075207515725915, "learning_rate": 0.003, "loss": 4.0936, "step": 8985 }, { "epoch": 0.08986, "grad_norm": 1.2740566021576438, "learning_rate": 0.003, "loss": 4.1002, "step": 8986 }, { "epoch": 0.08987, "grad_norm": 1.1459099503667163, "learning_rate": 0.003, "loss": 4.0814, "step": 8987 }, { "epoch": 0.08988, "grad_norm": 1.064348413271581, "learning_rate": 0.003, "loss": 4.0791, "step": 8988 }, { "epoch": 0.08989, "grad_norm": 1.1793693743335125, "learning_rate": 0.003, "loss": 4.1127, "step": 8989 }, { "epoch": 0.0899, "grad_norm": 1.0425071968697808, "learning_rate": 0.003, "loss": 4.0775, "step": 8990 }, { "epoch": 0.08991, "grad_norm": 1.468796617381801, "learning_rate": 0.003, "loss": 4.0926, "step": 8991 }, { "epoch": 0.08992, "grad_norm": 0.9327563758012993, "learning_rate": 0.003, "loss": 4.0714, "step": 8992 }, { "epoch": 0.08993, "grad_norm": 1.0525922979652393, "learning_rate": 0.003, "loss": 4.0831, "step": 8993 }, { "epoch": 0.08994, "grad_norm": 1.267662146582978, "learning_rate": 0.003, "loss": 4.0891, "step": 8994 }, { "epoch": 0.08995, "grad_norm": 1.0986748559783313, "learning_rate": 0.003, "loss": 4.0919, "step": 8995 }, { "epoch": 0.08996, "grad_norm": 1.3115338127669567, "learning_rate": 0.003, "loss": 4.1178, "step": 8996 }, { "epoch": 0.08997, "grad_norm": 0.9557599873271017, "learning_rate": 0.003, "loss": 4.0854, "step": 8997 }, { "epoch": 0.08998, "grad_norm": 1.139031347368501, "learning_rate": 0.003, "loss": 4.0735, "step": 8998 }, { "epoch": 0.08999, "grad_norm": 1.076392466563895, "learning_rate": 0.003, "loss": 4.1009, "step": 8999 }, { "epoch": 0.09, "grad_norm": 1.218252301925029, "learning_rate": 0.003, "loss": 4.0965, "step": 9000 }, { "epoch": 0.09001, "grad_norm": 1.1108783016947812, "learning_rate": 0.003, "loss": 4.0887, "step": 9001 }, { "epoch": 0.09002, "grad_norm": 1.1195942660042726, "learning_rate": 0.003, "loss": 4.0776, "step": 9002 }, { "epoch": 0.09003, "grad_norm": 1.203270184084982, "learning_rate": 0.003, "loss": 4.0667, "step": 9003 }, { "epoch": 0.09004, "grad_norm": 1.2021233361621204, "learning_rate": 0.003, "loss": 4.0865, "step": 9004 }, { "epoch": 0.09005, "grad_norm": 0.9743559860743248, "learning_rate": 0.003, "loss": 4.1056, "step": 9005 }, { "epoch": 0.09006, "grad_norm": 1.1823704601559941, "learning_rate": 0.003, "loss": 4.0876, "step": 9006 }, { "epoch": 0.09007, "grad_norm": 1.123233185721264, "learning_rate": 0.003, "loss": 4.1116, "step": 9007 }, { "epoch": 0.09008, "grad_norm": 1.0757862834677538, "learning_rate": 0.003, "loss": 4.0957, "step": 9008 }, { "epoch": 0.09009, "grad_norm": 1.266049145604807, "learning_rate": 0.003, "loss": 4.1, "step": 9009 }, { "epoch": 0.0901, "grad_norm": 1.1564480352920141, "learning_rate": 0.003, "loss": 4.0886, "step": 9010 }, { "epoch": 0.09011, "grad_norm": 1.2954757514768573, "learning_rate": 0.003, "loss": 4.0886, "step": 9011 }, { "epoch": 0.09012, "grad_norm": 0.9656390078636606, "learning_rate": 0.003, "loss": 4.0843, "step": 9012 }, { "epoch": 0.09013, "grad_norm": 1.153233909037288, "learning_rate": 0.003, "loss": 4.0778, "step": 9013 }, { "epoch": 0.09014, "grad_norm": 1.052108855153509, "learning_rate": 0.003, "loss": 4.1142, "step": 9014 }, { "epoch": 0.09015, "grad_norm": 1.2385359460251653, "learning_rate": 0.003, "loss": 4.1273, "step": 9015 }, { "epoch": 0.09016, "grad_norm": 1.0770867919896612, "learning_rate": 0.003, "loss": 4.1066, "step": 9016 }, { "epoch": 0.09017, "grad_norm": 1.0339851137855502, "learning_rate": 0.003, "loss": 4.1071, "step": 9017 }, { "epoch": 0.09018, "grad_norm": 1.3115774956660293, "learning_rate": 0.003, "loss": 4.106, "step": 9018 }, { "epoch": 0.09019, "grad_norm": 1.063765856905231, "learning_rate": 0.003, "loss": 4.074, "step": 9019 }, { "epoch": 0.0902, "grad_norm": 1.171303481692876, "learning_rate": 0.003, "loss": 4.0915, "step": 9020 }, { "epoch": 0.09021, "grad_norm": 1.0064142048132063, "learning_rate": 0.003, "loss": 4.1031, "step": 9021 }, { "epoch": 0.09022, "grad_norm": 1.0413311438329733, "learning_rate": 0.003, "loss": 4.093, "step": 9022 }, { "epoch": 0.09023, "grad_norm": 1.1452677126031137, "learning_rate": 0.003, "loss": 4.0877, "step": 9023 }, { "epoch": 0.09024, "grad_norm": 1.1850056876367443, "learning_rate": 0.003, "loss": 4.0738, "step": 9024 }, { "epoch": 0.09025, "grad_norm": 1.0640895906055132, "learning_rate": 0.003, "loss": 4.0931, "step": 9025 }, { "epoch": 0.09026, "grad_norm": 1.0784240586445417, "learning_rate": 0.003, "loss": 4.0855, "step": 9026 }, { "epoch": 0.09027, "grad_norm": 1.0642912702715661, "learning_rate": 0.003, "loss": 4.0801, "step": 9027 }, { "epoch": 0.09028, "grad_norm": 1.2655742298981867, "learning_rate": 0.003, "loss": 4.0993, "step": 9028 }, { "epoch": 0.09029, "grad_norm": 1.1293787235945678, "learning_rate": 0.003, "loss": 4.0887, "step": 9029 }, { "epoch": 0.0903, "grad_norm": 1.2521967234284181, "learning_rate": 0.003, "loss": 4.1175, "step": 9030 }, { "epoch": 0.09031, "grad_norm": 1.2715061899665723, "learning_rate": 0.003, "loss": 4.0684, "step": 9031 }, { "epoch": 0.09032, "grad_norm": 1.1227033191663636, "learning_rate": 0.003, "loss": 4.1186, "step": 9032 }, { "epoch": 0.09033, "grad_norm": 1.2698924690086315, "learning_rate": 0.003, "loss": 4.062, "step": 9033 }, { "epoch": 0.09034, "grad_norm": 1.0279738017452675, "learning_rate": 0.003, "loss": 4.0832, "step": 9034 }, { "epoch": 0.09035, "grad_norm": 0.9188646060088369, "learning_rate": 0.003, "loss": 4.0819, "step": 9035 }, { "epoch": 0.09036, "grad_norm": 1.0006330136135604, "learning_rate": 0.003, "loss": 4.0768, "step": 9036 }, { "epoch": 0.09037, "grad_norm": 1.229414309206932, "learning_rate": 0.003, "loss": 4.0872, "step": 9037 }, { "epoch": 0.09038, "grad_norm": 1.056028513175944, "learning_rate": 0.003, "loss": 4.083, "step": 9038 }, { "epoch": 0.09039, "grad_norm": 1.1567627754431467, "learning_rate": 0.003, "loss": 4.0976, "step": 9039 }, { "epoch": 0.0904, "grad_norm": 1.2620704881278146, "learning_rate": 0.003, "loss": 4.0711, "step": 9040 }, { "epoch": 0.09041, "grad_norm": 0.9332791175909212, "learning_rate": 0.003, "loss": 4.0623, "step": 9041 }, { "epoch": 0.09042, "grad_norm": 1.0031397413450402, "learning_rate": 0.003, "loss": 4.0671, "step": 9042 }, { "epoch": 0.09043, "grad_norm": 1.0842259914712198, "learning_rate": 0.003, "loss": 4.073, "step": 9043 }, { "epoch": 0.09044, "grad_norm": 1.1766907490463179, "learning_rate": 0.003, "loss": 4.0725, "step": 9044 }, { "epoch": 0.09045, "grad_norm": 1.1641848477623784, "learning_rate": 0.003, "loss": 4.0838, "step": 9045 }, { "epoch": 0.09046, "grad_norm": 0.9648345631484123, "learning_rate": 0.003, "loss": 4.0543, "step": 9046 }, { "epoch": 0.09047, "grad_norm": 1.0156565200283614, "learning_rate": 0.003, "loss": 4.0775, "step": 9047 }, { "epoch": 0.09048, "grad_norm": 1.2711228411111342, "learning_rate": 0.003, "loss": 4.0851, "step": 9048 }, { "epoch": 0.09049, "grad_norm": 0.9295851773809277, "learning_rate": 0.003, "loss": 4.1037, "step": 9049 }, { "epoch": 0.0905, "grad_norm": 1.1530883099484381, "learning_rate": 0.003, "loss": 4.0869, "step": 9050 }, { "epoch": 0.09051, "grad_norm": 1.058484140380266, "learning_rate": 0.003, "loss": 4.0795, "step": 9051 }, { "epoch": 0.09052, "grad_norm": 1.3221517826262894, "learning_rate": 0.003, "loss": 4.0979, "step": 9052 }, { "epoch": 0.09053, "grad_norm": 0.9783869911947403, "learning_rate": 0.003, "loss": 4.0727, "step": 9053 }, { "epoch": 0.09054, "grad_norm": 1.31217086863143, "learning_rate": 0.003, "loss": 4.0788, "step": 9054 }, { "epoch": 0.09055, "grad_norm": 1.15712884508595, "learning_rate": 0.003, "loss": 4.0672, "step": 9055 }, { "epoch": 0.09056, "grad_norm": 1.3601399184462666, "learning_rate": 0.003, "loss": 4.0759, "step": 9056 }, { "epoch": 0.09057, "grad_norm": 0.9702143476288941, "learning_rate": 0.003, "loss": 4.0979, "step": 9057 }, { "epoch": 0.09058, "grad_norm": 1.005445525878324, "learning_rate": 0.003, "loss": 4.1045, "step": 9058 }, { "epoch": 0.09059, "grad_norm": 1.343638060042953, "learning_rate": 0.003, "loss": 4.0989, "step": 9059 }, { "epoch": 0.0906, "grad_norm": 1.061498240726219, "learning_rate": 0.003, "loss": 4.107, "step": 9060 }, { "epoch": 0.09061, "grad_norm": 1.2119494291256259, "learning_rate": 0.003, "loss": 4.0983, "step": 9061 }, { "epoch": 0.09062, "grad_norm": 1.1153933410174173, "learning_rate": 0.003, "loss": 4.0659, "step": 9062 }, { "epoch": 0.09063, "grad_norm": 1.0509431689336337, "learning_rate": 0.003, "loss": 4.0794, "step": 9063 }, { "epoch": 0.09064, "grad_norm": 1.0456015290402727, "learning_rate": 0.003, "loss": 4.1104, "step": 9064 }, { "epoch": 0.09065, "grad_norm": 1.2005228086487294, "learning_rate": 0.003, "loss": 4.1067, "step": 9065 }, { "epoch": 0.09066, "grad_norm": 1.251274599678117, "learning_rate": 0.003, "loss": 4.097, "step": 9066 }, { "epoch": 0.09067, "grad_norm": 1.1743203475476542, "learning_rate": 0.003, "loss": 4.0891, "step": 9067 }, { "epoch": 0.09068, "grad_norm": 0.9880369735330882, "learning_rate": 0.003, "loss": 4.0988, "step": 9068 }, { "epoch": 0.09069, "grad_norm": 1.1738728040556163, "learning_rate": 0.003, "loss": 4.1397, "step": 9069 }, { "epoch": 0.0907, "grad_norm": 1.0377157330168174, "learning_rate": 0.003, "loss": 4.1178, "step": 9070 }, { "epoch": 0.09071, "grad_norm": 1.1057934780286178, "learning_rate": 0.003, "loss": 4.0976, "step": 9071 }, { "epoch": 0.09072, "grad_norm": 1.0188511735448431, "learning_rate": 0.003, "loss": 4.0708, "step": 9072 }, { "epoch": 0.09073, "grad_norm": 1.315640376041937, "learning_rate": 0.003, "loss": 4.1049, "step": 9073 }, { "epoch": 0.09074, "grad_norm": 1.101736279121598, "learning_rate": 0.003, "loss": 4.0708, "step": 9074 }, { "epoch": 0.09075, "grad_norm": 1.3946631521180286, "learning_rate": 0.003, "loss": 4.1137, "step": 9075 }, { "epoch": 0.09076, "grad_norm": 1.1148038465546715, "learning_rate": 0.003, "loss": 4.1152, "step": 9076 }, { "epoch": 0.09077, "grad_norm": 1.3284754703148802, "learning_rate": 0.003, "loss": 4.1146, "step": 9077 }, { "epoch": 0.09078, "grad_norm": 0.9783046559145829, "learning_rate": 0.003, "loss": 4.086, "step": 9078 }, { "epoch": 0.09079, "grad_norm": 0.913093721909462, "learning_rate": 0.003, "loss": 4.0941, "step": 9079 }, { "epoch": 0.0908, "grad_norm": 1.095737687813746, "learning_rate": 0.003, "loss": 4.0776, "step": 9080 }, { "epoch": 0.09081, "grad_norm": 1.2614609976958837, "learning_rate": 0.003, "loss": 4.1057, "step": 9081 }, { "epoch": 0.09082, "grad_norm": 1.2075594634986682, "learning_rate": 0.003, "loss": 4.079, "step": 9082 }, { "epoch": 0.09083, "grad_norm": 1.2122949862145338, "learning_rate": 0.003, "loss": 4.0975, "step": 9083 }, { "epoch": 0.09084, "grad_norm": 0.9500851087827136, "learning_rate": 0.003, "loss": 4.0662, "step": 9084 }, { "epoch": 0.09085, "grad_norm": 1.0476982169455582, "learning_rate": 0.003, "loss": 4.1066, "step": 9085 }, { "epoch": 0.09086, "grad_norm": 1.2554444552905066, "learning_rate": 0.003, "loss": 4.0835, "step": 9086 }, { "epoch": 0.09087, "grad_norm": 0.9421781553632245, "learning_rate": 0.003, "loss": 4.0754, "step": 9087 }, { "epoch": 0.09088, "grad_norm": 1.4131215511951543, "learning_rate": 0.003, "loss": 4.0851, "step": 9088 }, { "epoch": 0.09089, "grad_norm": 1.1108096893759776, "learning_rate": 0.003, "loss": 4.0912, "step": 9089 }, { "epoch": 0.0909, "grad_norm": 1.1952675721480868, "learning_rate": 0.003, "loss": 4.0818, "step": 9090 }, { "epoch": 0.09091, "grad_norm": 1.0724547606838648, "learning_rate": 0.003, "loss": 4.0636, "step": 9091 }, { "epoch": 0.09092, "grad_norm": 1.2051575454817625, "learning_rate": 0.003, "loss": 4.1245, "step": 9092 }, { "epoch": 0.09093, "grad_norm": 1.3267657738954277, "learning_rate": 0.003, "loss": 4.0824, "step": 9093 }, { "epoch": 0.09094, "grad_norm": 1.1254991850990028, "learning_rate": 0.003, "loss": 4.0873, "step": 9094 }, { "epoch": 0.09095, "grad_norm": 1.076280006184989, "learning_rate": 0.003, "loss": 4.0966, "step": 9095 }, { "epoch": 0.09096, "grad_norm": 1.2975257873577617, "learning_rate": 0.003, "loss": 4.0775, "step": 9096 }, { "epoch": 0.09097, "grad_norm": 0.8764623283131802, "learning_rate": 0.003, "loss": 4.084, "step": 9097 }, { "epoch": 0.09098, "grad_norm": 0.9942352270948855, "learning_rate": 0.003, "loss": 4.0592, "step": 9098 }, { "epoch": 0.09099, "grad_norm": 1.2726874710768534, "learning_rate": 0.003, "loss": 4.075, "step": 9099 }, { "epoch": 0.091, "grad_norm": 0.9642055558382768, "learning_rate": 0.003, "loss": 4.0608, "step": 9100 }, { "epoch": 0.09101, "grad_norm": 1.1013949140781738, "learning_rate": 0.003, "loss": 4.0831, "step": 9101 }, { "epoch": 0.09102, "grad_norm": 1.0913604869756948, "learning_rate": 0.003, "loss": 4.0934, "step": 9102 }, { "epoch": 0.09103, "grad_norm": 1.2031231056541927, "learning_rate": 0.003, "loss": 4.0604, "step": 9103 }, { "epoch": 0.09104, "grad_norm": 1.0988423095239437, "learning_rate": 0.003, "loss": 4.0863, "step": 9104 }, { "epoch": 0.09105, "grad_norm": 1.0423092013378354, "learning_rate": 0.003, "loss": 4.066, "step": 9105 }, { "epoch": 0.09106, "grad_norm": 1.2438248747188299, "learning_rate": 0.003, "loss": 4.0685, "step": 9106 }, { "epoch": 0.09107, "grad_norm": 1.1483179803808778, "learning_rate": 0.003, "loss": 4.0794, "step": 9107 }, { "epoch": 0.09108, "grad_norm": 1.326132597957969, "learning_rate": 0.003, "loss": 4.0737, "step": 9108 }, { "epoch": 0.09109, "grad_norm": 1.0113040032192664, "learning_rate": 0.003, "loss": 4.0687, "step": 9109 }, { "epoch": 0.0911, "grad_norm": 1.030048752615744, "learning_rate": 0.003, "loss": 4.0798, "step": 9110 }, { "epoch": 0.09111, "grad_norm": 1.2571878851477611, "learning_rate": 0.003, "loss": 4.1057, "step": 9111 }, { "epoch": 0.09112, "grad_norm": 0.9793382607265296, "learning_rate": 0.003, "loss": 4.0668, "step": 9112 }, { "epoch": 0.09113, "grad_norm": 1.1758336523804527, "learning_rate": 0.003, "loss": 4.0566, "step": 9113 }, { "epoch": 0.09114, "grad_norm": 1.0634406149265316, "learning_rate": 0.003, "loss": 4.0764, "step": 9114 }, { "epoch": 0.09115, "grad_norm": 1.2160595915766383, "learning_rate": 0.003, "loss": 4.0834, "step": 9115 }, { "epoch": 0.09116, "grad_norm": 1.0742814994403547, "learning_rate": 0.003, "loss": 4.1286, "step": 9116 }, { "epoch": 0.09117, "grad_norm": 1.168852556344174, "learning_rate": 0.003, "loss": 4.1029, "step": 9117 }, { "epoch": 0.09118, "grad_norm": 1.2618508344525157, "learning_rate": 0.003, "loss": 4.092, "step": 9118 }, { "epoch": 0.09119, "grad_norm": 0.977470788596692, "learning_rate": 0.003, "loss": 4.0807, "step": 9119 }, { "epoch": 0.0912, "grad_norm": 1.2131827050884871, "learning_rate": 0.003, "loss": 4.0997, "step": 9120 }, { "epoch": 0.09121, "grad_norm": 0.9797942266581282, "learning_rate": 0.003, "loss": 4.1182, "step": 9121 }, { "epoch": 0.09122, "grad_norm": 1.3195467632222808, "learning_rate": 0.003, "loss": 4.1106, "step": 9122 }, { "epoch": 0.09123, "grad_norm": 1.2794360152215094, "learning_rate": 0.003, "loss": 4.0647, "step": 9123 }, { "epoch": 0.09124, "grad_norm": 1.0383850216227262, "learning_rate": 0.003, "loss": 4.0801, "step": 9124 }, { "epoch": 0.09125, "grad_norm": 1.1895964256162657, "learning_rate": 0.003, "loss": 4.1042, "step": 9125 }, { "epoch": 0.09126, "grad_norm": 1.1565205073495568, "learning_rate": 0.003, "loss": 4.0578, "step": 9126 }, { "epoch": 0.09127, "grad_norm": 0.9870065472498529, "learning_rate": 0.003, "loss": 4.0923, "step": 9127 }, { "epoch": 0.09128, "grad_norm": 1.3924191556687682, "learning_rate": 0.003, "loss": 4.0966, "step": 9128 }, { "epoch": 0.09129, "grad_norm": 1.0415308726364658, "learning_rate": 0.003, "loss": 4.0804, "step": 9129 }, { "epoch": 0.0913, "grad_norm": 1.2239110824416417, "learning_rate": 0.003, "loss": 4.0831, "step": 9130 }, { "epoch": 0.09131, "grad_norm": 1.1333067848208729, "learning_rate": 0.003, "loss": 4.0786, "step": 9131 }, { "epoch": 0.09132, "grad_norm": 1.0124991406398278, "learning_rate": 0.003, "loss": 4.106, "step": 9132 }, { "epoch": 0.09133, "grad_norm": 1.1465094851053432, "learning_rate": 0.003, "loss": 4.0874, "step": 9133 }, { "epoch": 0.09134, "grad_norm": 1.2509363219325378, "learning_rate": 0.003, "loss": 4.0924, "step": 9134 }, { "epoch": 0.09135, "grad_norm": 1.1392240077626206, "learning_rate": 0.003, "loss": 4.1003, "step": 9135 }, { "epoch": 0.09136, "grad_norm": 1.083710930327429, "learning_rate": 0.003, "loss": 4.0802, "step": 9136 }, { "epoch": 0.09137, "grad_norm": 1.2335339938885863, "learning_rate": 0.003, "loss": 4.1181, "step": 9137 }, { "epoch": 0.09138, "grad_norm": 1.101274143648628, "learning_rate": 0.003, "loss": 4.0844, "step": 9138 }, { "epoch": 0.09139, "grad_norm": 1.2239809585615162, "learning_rate": 0.003, "loss": 4.0879, "step": 9139 }, { "epoch": 0.0914, "grad_norm": 0.9619789123487413, "learning_rate": 0.003, "loss": 4.0865, "step": 9140 }, { "epoch": 0.09141, "grad_norm": 1.1383798391014412, "learning_rate": 0.003, "loss": 4.1214, "step": 9141 }, { "epoch": 0.09142, "grad_norm": 1.0505265979852703, "learning_rate": 0.003, "loss": 4.1072, "step": 9142 }, { "epoch": 0.09143, "grad_norm": 1.0997295836100855, "learning_rate": 0.003, "loss": 4.0602, "step": 9143 }, { "epoch": 0.09144, "grad_norm": 1.0581118806506116, "learning_rate": 0.003, "loss": 4.0849, "step": 9144 }, { "epoch": 0.09145, "grad_norm": 1.234848911993916, "learning_rate": 0.003, "loss": 4.1144, "step": 9145 }, { "epoch": 0.09146, "grad_norm": 1.08478997063052, "learning_rate": 0.003, "loss": 4.1048, "step": 9146 }, { "epoch": 0.09147, "grad_norm": 1.2618896424614257, "learning_rate": 0.003, "loss": 4.0916, "step": 9147 }, { "epoch": 0.09148, "grad_norm": 1.123169224398365, "learning_rate": 0.003, "loss": 4.1023, "step": 9148 }, { "epoch": 0.09149, "grad_norm": 1.0847323522173526, "learning_rate": 0.003, "loss": 4.0463, "step": 9149 }, { "epoch": 0.0915, "grad_norm": 1.1680298895237302, "learning_rate": 0.003, "loss": 4.0558, "step": 9150 }, { "epoch": 0.09151, "grad_norm": 1.3844852453386007, "learning_rate": 0.003, "loss": 4.0928, "step": 9151 }, { "epoch": 0.09152, "grad_norm": 1.0218893177216748, "learning_rate": 0.003, "loss": 4.0966, "step": 9152 }, { "epoch": 0.09153, "grad_norm": 0.9965527672502461, "learning_rate": 0.003, "loss": 4.0923, "step": 9153 }, { "epoch": 0.09154, "grad_norm": 1.0995384198383622, "learning_rate": 0.003, "loss": 4.0986, "step": 9154 }, { "epoch": 0.09155, "grad_norm": 1.0410969716281233, "learning_rate": 0.003, "loss": 4.1026, "step": 9155 }, { "epoch": 0.09156, "grad_norm": 1.2600779934860684, "learning_rate": 0.003, "loss": 4.1036, "step": 9156 }, { "epoch": 0.09157, "grad_norm": 0.9324524194147646, "learning_rate": 0.003, "loss": 4.0498, "step": 9157 }, { "epoch": 0.09158, "grad_norm": 1.1514601056640672, "learning_rate": 0.003, "loss": 4.0831, "step": 9158 }, { "epoch": 0.09159, "grad_norm": 1.2870193873680913, "learning_rate": 0.003, "loss": 4.0959, "step": 9159 }, { "epoch": 0.0916, "grad_norm": 1.0560405419821877, "learning_rate": 0.003, "loss": 4.0711, "step": 9160 }, { "epoch": 0.09161, "grad_norm": 1.0299875085861723, "learning_rate": 0.003, "loss": 4.0749, "step": 9161 }, { "epoch": 0.09162, "grad_norm": 1.116367131505824, "learning_rate": 0.003, "loss": 4.1133, "step": 9162 }, { "epoch": 0.09163, "grad_norm": 0.9983394688004362, "learning_rate": 0.003, "loss": 4.0626, "step": 9163 }, { "epoch": 0.09164, "grad_norm": 1.1592721035304163, "learning_rate": 0.003, "loss": 4.0857, "step": 9164 }, { "epoch": 0.09165, "grad_norm": 1.0344539200329619, "learning_rate": 0.003, "loss": 4.1122, "step": 9165 }, { "epoch": 0.09166, "grad_norm": 1.0474394105993658, "learning_rate": 0.003, "loss": 4.0863, "step": 9166 }, { "epoch": 0.09167, "grad_norm": 1.1658008703847873, "learning_rate": 0.003, "loss": 4.0779, "step": 9167 }, { "epoch": 0.09168, "grad_norm": 1.0606777041009616, "learning_rate": 0.003, "loss": 4.0687, "step": 9168 }, { "epoch": 0.09169, "grad_norm": 1.200603861024009, "learning_rate": 0.003, "loss": 4.0997, "step": 9169 }, { "epoch": 0.0917, "grad_norm": 1.0944668053816515, "learning_rate": 0.003, "loss": 4.0772, "step": 9170 }, { "epoch": 0.09171, "grad_norm": 1.0821405986570527, "learning_rate": 0.003, "loss": 4.1139, "step": 9171 }, { "epoch": 0.09172, "grad_norm": 1.145946204001904, "learning_rate": 0.003, "loss": 4.1157, "step": 9172 }, { "epoch": 0.09173, "grad_norm": 1.3491933499222881, "learning_rate": 0.003, "loss": 4.0875, "step": 9173 }, { "epoch": 0.09174, "grad_norm": 1.1851092298888857, "learning_rate": 0.003, "loss": 4.1014, "step": 9174 }, { "epoch": 0.09175, "grad_norm": 1.307879610936908, "learning_rate": 0.003, "loss": 4.097, "step": 9175 }, { "epoch": 0.09176, "grad_norm": 1.0397697110301758, "learning_rate": 0.003, "loss": 4.0974, "step": 9176 }, { "epoch": 0.09177, "grad_norm": 1.322756400789418, "learning_rate": 0.003, "loss": 4.0557, "step": 9177 }, { "epoch": 0.09178, "grad_norm": 1.0359489238564688, "learning_rate": 0.003, "loss": 4.0988, "step": 9178 }, { "epoch": 0.09179, "grad_norm": 1.387284506907797, "learning_rate": 0.003, "loss": 4.0963, "step": 9179 }, { "epoch": 0.0918, "grad_norm": 1.0856625764441654, "learning_rate": 0.003, "loss": 4.0878, "step": 9180 }, { "epoch": 0.09181, "grad_norm": 1.3055834964911002, "learning_rate": 0.003, "loss": 4.0839, "step": 9181 }, { "epoch": 0.09182, "grad_norm": 1.1092289866442044, "learning_rate": 0.003, "loss": 4.0852, "step": 9182 }, { "epoch": 0.09183, "grad_norm": 1.1885236545545457, "learning_rate": 0.003, "loss": 4.0786, "step": 9183 }, { "epoch": 0.09184, "grad_norm": 0.9487275740138595, "learning_rate": 0.003, "loss": 4.074, "step": 9184 }, { "epoch": 0.09185, "grad_norm": 1.2244446488284848, "learning_rate": 0.003, "loss": 4.1114, "step": 9185 }, { "epoch": 0.09186, "grad_norm": 0.9495737584675268, "learning_rate": 0.003, "loss": 4.0496, "step": 9186 }, { "epoch": 0.09187, "grad_norm": 1.1296622323739962, "learning_rate": 0.003, "loss": 4.1225, "step": 9187 }, { "epoch": 0.09188, "grad_norm": 1.201509024595302, "learning_rate": 0.003, "loss": 4.0993, "step": 9188 }, { "epoch": 0.09189, "grad_norm": 1.2702535000447714, "learning_rate": 0.003, "loss": 4.105, "step": 9189 }, { "epoch": 0.0919, "grad_norm": 1.1985062138264047, "learning_rate": 0.003, "loss": 4.1166, "step": 9190 }, { "epoch": 0.09191, "grad_norm": 1.0017418634767983, "learning_rate": 0.003, "loss": 4.0884, "step": 9191 }, { "epoch": 0.09192, "grad_norm": 1.0025901701320967, "learning_rate": 0.003, "loss": 4.0734, "step": 9192 }, { "epoch": 0.09193, "grad_norm": 1.3829381034381623, "learning_rate": 0.003, "loss": 4.0746, "step": 9193 }, { "epoch": 0.09194, "grad_norm": 0.9677385774606069, "learning_rate": 0.003, "loss": 4.0878, "step": 9194 }, { "epoch": 0.09195, "grad_norm": 1.6506761595471497, "learning_rate": 0.003, "loss": 4.1039, "step": 9195 }, { "epoch": 0.09196, "grad_norm": 1.001256536135946, "learning_rate": 0.003, "loss": 4.1103, "step": 9196 }, { "epoch": 0.09197, "grad_norm": 1.160229001768998, "learning_rate": 0.003, "loss": 4.0947, "step": 9197 }, { "epoch": 0.09198, "grad_norm": 1.2451954616247143, "learning_rate": 0.003, "loss": 4.0977, "step": 9198 }, { "epoch": 0.09199, "grad_norm": 1.234886981466303, "learning_rate": 0.003, "loss": 4.1039, "step": 9199 }, { "epoch": 0.092, "grad_norm": 1.4720829204629136, "learning_rate": 0.003, "loss": 4.1055, "step": 9200 }, { "epoch": 0.09201, "grad_norm": 0.9054048830224165, "learning_rate": 0.003, "loss": 4.0875, "step": 9201 }, { "epoch": 0.09202, "grad_norm": 0.916669785043941, "learning_rate": 0.003, "loss": 4.0998, "step": 9202 }, { "epoch": 0.09203, "grad_norm": 1.0582633355745965, "learning_rate": 0.003, "loss": 4.1007, "step": 9203 }, { "epoch": 0.09204, "grad_norm": 1.435908065217812, "learning_rate": 0.003, "loss": 4.114, "step": 9204 }, { "epoch": 0.09205, "grad_norm": 0.9513665704460509, "learning_rate": 0.003, "loss": 4.0891, "step": 9205 }, { "epoch": 0.09206, "grad_norm": 1.1687418270227496, "learning_rate": 0.003, "loss": 4.1024, "step": 9206 }, { "epoch": 0.09207, "grad_norm": 1.092936507823944, "learning_rate": 0.003, "loss": 4.1192, "step": 9207 }, { "epoch": 0.09208, "grad_norm": 1.0812988391688032, "learning_rate": 0.003, "loss": 4.0742, "step": 9208 }, { "epoch": 0.09209, "grad_norm": 1.092551499098981, "learning_rate": 0.003, "loss": 4.0769, "step": 9209 }, { "epoch": 0.0921, "grad_norm": 0.963997770849915, "learning_rate": 0.003, "loss": 4.0985, "step": 9210 }, { "epoch": 0.09211, "grad_norm": 1.22240394776083, "learning_rate": 0.003, "loss": 4.0843, "step": 9211 }, { "epoch": 0.09212, "grad_norm": 0.9848334695913099, "learning_rate": 0.003, "loss": 4.103, "step": 9212 }, { "epoch": 0.09213, "grad_norm": 1.1751785205728014, "learning_rate": 0.003, "loss": 4.092, "step": 9213 }, { "epoch": 0.09214, "grad_norm": 0.9488549068764679, "learning_rate": 0.003, "loss": 4.0681, "step": 9214 }, { "epoch": 0.09215, "grad_norm": 1.5103788355550805, "learning_rate": 0.003, "loss": 4.0925, "step": 9215 }, { "epoch": 0.09216, "grad_norm": 1.0379900464298588, "learning_rate": 0.003, "loss": 4.0942, "step": 9216 }, { "epoch": 0.09217, "grad_norm": 1.1644131128976074, "learning_rate": 0.003, "loss": 4.0566, "step": 9217 }, { "epoch": 0.09218, "grad_norm": 1.209017912778444, "learning_rate": 0.003, "loss": 4.1124, "step": 9218 }, { "epoch": 0.09219, "grad_norm": 1.1356937772580649, "learning_rate": 0.003, "loss": 4.0936, "step": 9219 }, { "epoch": 0.0922, "grad_norm": 1.0245612101785528, "learning_rate": 0.003, "loss": 4.089, "step": 9220 }, { "epoch": 0.09221, "grad_norm": 1.1979930204364009, "learning_rate": 0.003, "loss": 4.108, "step": 9221 }, { "epoch": 0.09222, "grad_norm": 1.0364515076478473, "learning_rate": 0.003, "loss": 4.0881, "step": 9222 }, { "epoch": 0.09223, "grad_norm": 1.098278277703503, "learning_rate": 0.003, "loss": 4.0851, "step": 9223 }, { "epoch": 0.09224, "grad_norm": 1.0253467498355422, "learning_rate": 0.003, "loss": 4.1004, "step": 9224 }, { "epoch": 0.09225, "grad_norm": 1.2208294239001167, "learning_rate": 0.003, "loss": 4.0846, "step": 9225 }, { "epoch": 0.09226, "grad_norm": 0.982628865592051, "learning_rate": 0.003, "loss": 4.0681, "step": 9226 }, { "epoch": 0.09227, "grad_norm": 1.2469564366340415, "learning_rate": 0.003, "loss": 4.0929, "step": 9227 }, { "epoch": 0.09228, "grad_norm": 0.9686273648297514, "learning_rate": 0.003, "loss": 4.0882, "step": 9228 }, { "epoch": 0.09229, "grad_norm": 1.6332491084756777, "learning_rate": 0.003, "loss": 4.1054, "step": 9229 }, { "epoch": 0.0923, "grad_norm": 0.9688872939255809, "learning_rate": 0.003, "loss": 4.0947, "step": 9230 }, { "epoch": 0.09231, "grad_norm": 1.071956008784058, "learning_rate": 0.003, "loss": 4.0914, "step": 9231 }, { "epoch": 0.09232, "grad_norm": 1.0691740349202605, "learning_rate": 0.003, "loss": 4.1016, "step": 9232 }, { "epoch": 0.09233, "grad_norm": 1.1757930687992064, "learning_rate": 0.003, "loss": 4.114, "step": 9233 }, { "epoch": 0.09234, "grad_norm": 1.164102159428301, "learning_rate": 0.003, "loss": 4.1207, "step": 9234 }, { "epoch": 0.09235, "grad_norm": 1.0114788239049175, "learning_rate": 0.003, "loss": 4.0808, "step": 9235 }, { "epoch": 0.09236, "grad_norm": 1.404798045906275, "learning_rate": 0.003, "loss": 4.1296, "step": 9236 }, { "epoch": 0.09237, "grad_norm": 1.136045166122303, "learning_rate": 0.003, "loss": 4.0882, "step": 9237 }, { "epoch": 0.09238, "grad_norm": 1.318478567865631, "learning_rate": 0.003, "loss": 4.0724, "step": 9238 }, { "epoch": 0.09239, "grad_norm": 1.2045711416898381, "learning_rate": 0.003, "loss": 4.0774, "step": 9239 }, { "epoch": 0.0924, "grad_norm": 1.098429972645525, "learning_rate": 0.003, "loss": 4.0825, "step": 9240 }, { "epoch": 0.09241, "grad_norm": 1.0429115598947243, "learning_rate": 0.003, "loss": 4.0881, "step": 9241 }, { "epoch": 0.09242, "grad_norm": 1.0965896753183049, "learning_rate": 0.003, "loss": 4.1027, "step": 9242 }, { "epoch": 0.09243, "grad_norm": 1.150697191723102, "learning_rate": 0.003, "loss": 4.1055, "step": 9243 }, { "epoch": 0.09244, "grad_norm": 0.9965888915579897, "learning_rate": 0.003, "loss": 4.0886, "step": 9244 }, { "epoch": 0.09245, "grad_norm": 1.1167730900283515, "learning_rate": 0.003, "loss": 4.0732, "step": 9245 }, { "epoch": 0.09246, "grad_norm": 1.0193728752692073, "learning_rate": 0.003, "loss": 4.0858, "step": 9246 }, { "epoch": 0.09247, "grad_norm": 1.0219816554123302, "learning_rate": 0.003, "loss": 4.0756, "step": 9247 }, { "epoch": 0.09248, "grad_norm": 1.1080989732024906, "learning_rate": 0.003, "loss": 4.089, "step": 9248 }, { "epoch": 0.09249, "grad_norm": 1.103023479040818, "learning_rate": 0.003, "loss": 4.1377, "step": 9249 }, { "epoch": 0.0925, "grad_norm": 0.9904590912992994, "learning_rate": 0.003, "loss": 4.1052, "step": 9250 }, { "epoch": 0.09251, "grad_norm": 1.08798994175361, "learning_rate": 0.003, "loss": 4.1066, "step": 9251 }, { "epoch": 0.09252, "grad_norm": 1.2064555294662573, "learning_rate": 0.003, "loss": 4.1189, "step": 9252 }, { "epoch": 0.09253, "grad_norm": 1.0462445137910321, "learning_rate": 0.003, "loss": 4.0878, "step": 9253 }, { "epoch": 0.09254, "grad_norm": 1.2715061499946645, "learning_rate": 0.003, "loss": 4.1091, "step": 9254 }, { "epoch": 0.09255, "grad_norm": 1.2438088592426244, "learning_rate": 0.003, "loss": 4.0698, "step": 9255 }, { "epoch": 0.09256, "grad_norm": 1.2544838328606913, "learning_rate": 0.003, "loss": 4.1045, "step": 9256 }, { "epoch": 0.09257, "grad_norm": 1.0171563098654022, "learning_rate": 0.003, "loss": 4.089, "step": 9257 }, { "epoch": 0.09258, "grad_norm": 1.2557192439580471, "learning_rate": 0.003, "loss": 4.0833, "step": 9258 }, { "epoch": 0.09259, "grad_norm": 1.0493174212094476, "learning_rate": 0.003, "loss": 4.1034, "step": 9259 }, { "epoch": 0.0926, "grad_norm": 1.419614696139921, "learning_rate": 0.003, "loss": 4.0968, "step": 9260 }, { "epoch": 0.09261, "grad_norm": 1.3455975545498775, "learning_rate": 0.003, "loss": 4.0746, "step": 9261 }, { "epoch": 0.09262, "grad_norm": 0.924216487979214, "learning_rate": 0.003, "loss": 4.0723, "step": 9262 }, { "epoch": 0.09263, "grad_norm": 1.1329672668552027, "learning_rate": 0.003, "loss": 4.0974, "step": 9263 }, { "epoch": 0.09264, "grad_norm": 1.1778087554299006, "learning_rate": 0.003, "loss": 4.1327, "step": 9264 }, { "epoch": 0.09265, "grad_norm": 0.9448279086828678, "learning_rate": 0.003, "loss": 4.0922, "step": 9265 }, { "epoch": 0.09266, "grad_norm": 1.056098935824549, "learning_rate": 0.003, "loss": 4.0952, "step": 9266 }, { "epoch": 0.09267, "grad_norm": 1.0254847874505146, "learning_rate": 0.003, "loss": 4.0716, "step": 9267 }, { "epoch": 0.09268, "grad_norm": 1.4009869680406688, "learning_rate": 0.003, "loss": 4.0865, "step": 9268 }, { "epoch": 0.09269, "grad_norm": 0.7636038650685317, "learning_rate": 0.003, "loss": 4.0849, "step": 9269 }, { "epoch": 0.0927, "grad_norm": 0.9369345979518385, "learning_rate": 0.003, "loss": 4.1127, "step": 9270 }, { "epoch": 0.09271, "grad_norm": 1.380947841962622, "learning_rate": 0.003, "loss": 4.1061, "step": 9271 }, { "epoch": 0.09272, "grad_norm": 1.0988768526691668, "learning_rate": 0.003, "loss": 4.0845, "step": 9272 }, { "epoch": 0.09273, "grad_norm": 1.201945787507501, "learning_rate": 0.003, "loss": 4.104, "step": 9273 }, { "epoch": 0.09274, "grad_norm": 1.110016280769841, "learning_rate": 0.003, "loss": 4.0915, "step": 9274 }, { "epoch": 0.09275, "grad_norm": 1.2820959206746452, "learning_rate": 0.003, "loss": 4.1131, "step": 9275 }, { "epoch": 0.09276, "grad_norm": 0.9452322962716858, "learning_rate": 0.003, "loss": 4.0953, "step": 9276 }, { "epoch": 0.09277, "grad_norm": 0.9631660827302326, "learning_rate": 0.003, "loss": 4.0883, "step": 9277 }, { "epoch": 0.09278, "grad_norm": 1.14895196868688, "learning_rate": 0.003, "loss": 4.0687, "step": 9278 }, { "epoch": 0.09279, "grad_norm": 1.0107540563165534, "learning_rate": 0.003, "loss": 4.092, "step": 9279 }, { "epoch": 0.0928, "grad_norm": 1.2063554014208668, "learning_rate": 0.003, "loss": 4.0896, "step": 9280 }, { "epoch": 0.09281, "grad_norm": 1.2176906534765644, "learning_rate": 0.003, "loss": 4.0976, "step": 9281 }, { "epoch": 0.09282, "grad_norm": 1.1177750119704892, "learning_rate": 0.003, "loss": 4.1199, "step": 9282 }, { "epoch": 0.09283, "grad_norm": 1.163517359272145, "learning_rate": 0.003, "loss": 4.1117, "step": 9283 }, { "epoch": 0.09284, "grad_norm": 1.0946434369290665, "learning_rate": 0.003, "loss": 4.0887, "step": 9284 }, { "epoch": 0.09285, "grad_norm": 1.0823393825924086, "learning_rate": 0.003, "loss": 4.0787, "step": 9285 }, { "epoch": 0.09286, "grad_norm": 1.1806916666698575, "learning_rate": 0.003, "loss": 4.0982, "step": 9286 }, { "epoch": 0.09287, "grad_norm": 1.1880978680931515, "learning_rate": 0.003, "loss": 4.0954, "step": 9287 }, { "epoch": 0.09288, "grad_norm": 1.0280458966744943, "learning_rate": 0.003, "loss": 4.0943, "step": 9288 }, { "epoch": 0.09289, "grad_norm": 1.165817168542551, "learning_rate": 0.003, "loss": 4.0661, "step": 9289 }, { "epoch": 0.0929, "grad_norm": 1.0529747069060469, "learning_rate": 0.003, "loss": 4.0926, "step": 9290 }, { "epoch": 0.09291, "grad_norm": 1.1911530919173343, "learning_rate": 0.003, "loss": 4.092, "step": 9291 }, { "epoch": 0.09292, "grad_norm": 1.0341998160006975, "learning_rate": 0.003, "loss": 4.107, "step": 9292 }, { "epoch": 0.09293, "grad_norm": 1.2373473437286293, "learning_rate": 0.003, "loss": 4.0734, "step": 9293 }, { "epoch": 0.09294, "grad_norm": 1.0331578332846882, "learning_rate": 0.003, "loss": 4.0881, "step": 9294 }, { "epoch": 0.09295, "grad_norm": 1.4023676676067414, "learning_rate": 0.003, "loss": 4.0971, "step": 9295 }, { "epoch": 0.09296, "grad_norm": 0.9248252353770195, "learning_rate": 0.003, "loss": 4.0754, "step": 9296 }, { "epoch": 0.09297, "grad_norm": 1.1620446186083104, "learning_rate": 0.003, "loss": 4.0917, "step": 9297 }, { "epoch": 0.09298, "grad_norm": 1.4922806015618209, "learning_rate": 0.003, "loss": 4.1003, "step": 9298 }, { "epoch": 0.09299, "grad_norm": 1.0988995125095076, "learning_rate": 0.003, "loss": 4.0917, "step": 9299 }, { "epoch": 0.093, "grad_norm": 0.9445100928298968, "learning_rate": 0.003, "loss": 4.0881, "step": 9300 }, { "epoch": 0.09301, "grad_norm": 0.9841352473435087, "learning_rate": 0.003, "loss": 4.0833, "step": 9301 }, { "epoch": 0.09302, "grad_norm": 1.2917413918777942, "learning_rate": 0.003, "loss": 4.1175, "step": 9302 }, { "epoch": 0.09303, "grad_norm": 1.1883645310803719, "learning_rate": 0.003, "loss": 4.0872, "step": 9303 }, { "epoch": 0.09304, "grad_norm": 1.3732633784531907, "learning_rate": 0.003, "loss": 4.0843, "step": 9304 }, { "epoch": 0.09305, "grad_norm": 0.9439791924939734, "learning_rate": 0.003, "loss": 4.0816, "step": 9305 }, { "epoch": 0.09306, "grad_norm": 1.0780768668152263, "learning_rate": 0.003, "loss": 4.0866, "step": 9306 }, { "epoch": 0.09307, "grad_norm": 1.1260710089048631, "learning_rate": 0.003, "loss": 4.0908, "step": 9307 }, { "epoch": 0.09308, "grad_norm": 1.1930964774850155, "learning_rate": 0.003, "loss": 4.0747, "step": 9308 }, { "epoch": 0.09309, "grad_norm": 1.1034586208598085, "learning_rate": 0.003, "loss": 4.0967, "step": 9309 }, { "epoch": 0.0931, "grad_norm": 1.066001517551609, "learning_rate": 0.003, "loss": 4.102, "step": 9310 }, { "epoch": 0.09311, "grad_norm": 1.2800453382483366, "learning_rate": 0.003, "loss": 4.0907, "step": 9311 }, { "epoch": 0.09312, "grad_norm": 0.9839849279310616, "learning_rate": 0.003, "loss": 4.0877, "step": 9312 }, { "epoch": 0.09313, "grad_norm": 1.1207976506853647, "learning_rate": 0.003, "loss": 4.1194, "step": 9313 }, { "epoch": 0.09314, "grad_norm": 1.0422474086769304, "learning_rate": 0.003, "loss": 4.0908, "step": 9314 }, { "epoch": 0.09315, "grad_norm": 1.3153342391219134, "learning_rate": 0.003, "loss": 4.0945, "step": 9315 }, { "epoch": 0.09316, "grad_norm": 0.9724487500802593, "learning_rate": 0.003, "loss": 4.0872, "step": 9316 }, { "epoch": 0.09317, "grad_norm": 1.2842443211720584, "learning_rate": 0.003, "loss": 4.1096, "step": 9317 }, { "epoch": 0.09318, "grad_norm": 1.2767684341502084, "learning_rate": 0.003, "loss": 4.0803, "step": 9318 }, { "epoch": 0.09319, "grad_norm": 0.8177780821558324, "learning_rate": 0.003, "loss": 4.1119, "step": 9319 }, { "epoch": 0.0932, "grad_norm": 0.8736538476508046, "learning_rate": 0.003, "loss": 4.0886, "step": 9320 }, { "epoch": 0.09321, "grad_norm": 1.0764894227017647, "learning_rate": 0.003, "loss": 4.0837, "step": 9321 }, { "epoch": 0.09322, "grad_norm": 1.106669548495908, "learning_rate": 0.003, "loss": 4.06, "step": 9322 }, { "epoch": 0.09323, "grad_norm": 0.9109911357340257, "learning_rate": 0.003, "loss": 4.0853, "step": 9323 }, { "epoch": 0.09324, "grad_norm": 1.0121524528258052, "learning_rate": 0.003, "loss": 4.1085, "step": 9324 }, { "epoch": 0.09325, "grad_norm": 1.1953363852776204, "learning_rate": 0.003, "loss": 4.0571, "step": 9325 }, { "epoch": 0.09326, "grad_norm": 1.2214118230644457, "learning_rate": 0.003, "loss": 4.0744, "step": 9326 }, { "epoch": 0.09327, "grad_norm": 1.3516713372464368, "learning_rate": 0.003, "loss": 4.089, "step": 9327 }, { "epoch": 0.09328, "grad_norm": 1.358690648608492, "learning_rate": 0.003, "loss": 4.0948, "step": 9328 }, { "epoch": 0.09329, "grad_norm": 0.8240895443762108, "learning_rate": 0.003, "loss": 4.0845, "step": 9329 }, { "epoch": 0.0933, "grad_norm": 0.9127779555741105, "learning_rate": 0.003, "loss": 4.1008, "step": 9330 }, { "epoch": 0.09331, "grad_norm": 1.2795917276154456, "learning_rate": 0.003, "loss": 4.078, "step": 9331 }, { "epoch": 0.09332, "grad_norm": 1.1571382152960765, "learning_rate": 0.003, "loss": 4.0997, "step": 9332 }, { "epoch": 0.09333, "grad_norm": 1.534670784931041, "learning_rate": 0.003, "loss": 4.0833, "step": 9333 }, { "epoch": 0.09334, "grad_norm": 0.8244035475472739, "learning_rate": 0.003, "loss": 4.0986, "step": 9334 }, { "epoch": 0.09335, "grad_norm": 0.9041976617795457, "learning_rate": 0.003, "loss": 4.0805, "step": 9335 }, { "epoch": 0.09336, "grad_norm": 1.2406100259880315, "learning_rate": 0.003, "loss": 4.0769, "step": 9336 }, { "epoch": 0.09337, "grad_norm": 1.0485091876703885, "learning_rate": 0.003, "loss": 4.0894, "step": 9337 }, { "epoch": 0.09338, "grad_norm": 1.489821206610966, "learning_rate": 0.003, "loss": 4.0996, "step": 9338 }, { "epoch": 0.09339, "grad_norm": 0.964634627652276, "learning_rate": 0.003, "loss": 4.0735, "step": 9339 }, { "epoch": 0.0934, "grad_norm": 1.0533122624515205, "learning_rate": 0.003, "loss": 4.0671, "step": 9340 }, { "epoch": 0.09341, "grad_norm": 1.1590715738463193, "learning_rate": 0.003, "loss": 4.0964, "step": 9341 }, { "epoch": 0.09342, "grad_norm": 1.2626146558793703, "learning_rate": 0.003, "loss": 4.0854, "step": 9342 }, { "epoch": 0.09343, "grad_norm": 1.0502100202087143, "learning_rate": 0.003, "loss": 4.0875, "step": 9343 }, { "epoch": 0.09344, "grad_norm": 1.283106408396065, "learning_rate": 0.003, "loss": 4.0945, "step": 9344 }, { "epoch": 0.09345, "grad_norm": 0.9876362836581257, "learning_rate": 0.003, "loss": 4.1088, "step": 9345 }, { "epoch": 0.09346, "grad_norm": 1.217302791244378, "learning_rate": 0.003, "loss": 4.0997, "step": 9346 }, { "epoch": 0.09347, "grad_norm": 0.922861237259809, "learning_rate": 0.003, "loss": 4.0815, "step": 9347 }, { "epoch": 0.09348, "grad_norm": 1.0359926662539227, "learning_rate": 0.003, "loss": 4.1005, "step": 9348 }, { "epoch": 0.09349, "grad_norm": 1.2124928441737841, "learning_rate": 0.003, "loss": 4.1043, "step": 9349 }, { "epoch": 0.0935, "grad_norm": 1.2233172467818763, "learning_rate": 0.003, "loss": 4.0895, "step": 9350 }, { "epoch": 0.09351, "grad_norm": 1.1007285345731772, "learning_rate": 0.003, "loss": 4.0591, "step": 9351 }, { "epoch": 0.09352, "grad_norm": 1.1629721595908953, "learning_rate": 0.003, "loss": 4.0906, "step": 9352 }, { "epoch": 0.09353, "grad_norm": 1.1644970866221465, "learning_rate": 0.003, "loss": 4.068, "step": 9353 }, { "epoch": 0.09354, "grad_norm": 1.2882587498408007, "learning_rate": 0.003, "loss": 4.0837, "step": 9354 }, { "epoch": 0.09355, "grad_norm": 1.206734517803683, "learning_rate": 0.003, "loss": 4.1063, "step": 9355 }, { "epoch": 0.09356, "grad_norm": 1.1528835236818287, "learning_rate": 0.003, "loss": 4.0919, "step": 9356 }, { "epoch": 0.09357, "grad_norm": 1.0992532104364745, "learning_rate": 0.003, "loss": 4.0891, "step": 9357 }, { "epoch": 0.09358, "grad_norm": 1.0561375952353425, "learning_rate": 0.003, "loss": 4.0891, "step": 9358 }, { "epoch": 0.09359, "grad_norm": 1.1750434512246604, "learning_rate": 0.003, "loss": 4.1049, "step": 9359 }, { "epoch": 0.0936, "grad_norm": 1.0984683955298284, "learning_rate": 0.003, "loss": 4.0951, "step": 9360 }, { "epoch": 0.09361, "grad_norm": 1.3400256540724003, "learning_rate": 0.003, "loss": 4.1013, "step": 9361 }, { "epoch": 0.09362, "grad_norm": 1.2780797275614484, "learning_rate": 0.003, "loss": 4.131, "step": 9362 }, { "epoch": 0.09363, "grad_norm": 1.0599674321980093, "learning_rate": 0.003, "loss": 4.1124, "step": 9363 }, { "epoch": 0.09364, "grad_norm": 1.1102228277609347, "learning_rate": 0.003, "loss": 4.0819, "step": 9364 }, { "epoch": 0.09365, "grad_norm": 1.0688813344095076, "learning_rate": 0.003, "loss": 4.1051, "step": 9365 }, { "epoch": 0.09366, "grad_norm": 1.3133131086901089, "learning_rate": 0.003, "loss": 4.0581, "step": 9366 }, { "epoch": 0.09367, "grad_norm": 1.0992701868502373, "learning_rate": 0.003, "loss": 4.0877, "step": 9367 }, { "epoch": 0.09368, "grad_norm": 1.3194143703533834, "learning_rate": 0.003, "loss": 4.1021, "step": 9368 }, { "epoch": 0.09369, "grad_norm": 1.0629018593754467, "learning_rate": 0.003, "loss": 4.0658, "step": 9369 }, { "epoch": 0.0937, "grad_norm": 1.0256226402002402, "learning_rate": 0.003, "loss": 4.1078, "step": 9370 }, { "epoch": 0.09371, "grad_norm": 1.088497881262939, "learning_rate": 0.003, "loss": 4.097, "step": 9371 }, { "epoch": 0.09372, "grad_norm": 1.1654530492594952, "learning_rate": 0.003, "loss": 4.077, "step": 9372 }, { "epoch": 0.09373, "grad_norm": 1.1828421853475708, "learning_rate": 0.003, "loss": 4.1286, "step": 9373 }, { "epoch": 0.09374, "grad_norm": 0.9578727686445765, "learning_rate": 0.003, "loss": 4.1225, "step": 9374 }, { "epoch": 0.09375, "grad_norm": 1.1087522061787056, "learning_rate": 0.003, "loss": 4.0655, "step": 9375 }, { "epoch": 0.09376, "grad_norm": 1.075961470799943, "learning_rate": 0.003, "loss": 4.1104, "step": 9376 }, { "epoch": 0.09377, "grad_norm": 1.037345966623672, "learning_rate": 0.003, "loss": 4.0899, "step": 9377 }, { "epoch": 0.09378, "grad_norm": 1.1148154963944317, "learning_rate": 0.003, "loss": 4.0999, "step": 9378 }, { "epoch": 0.09379, "grad_norm": 1.0376386791840873, "learning_rate": 0.003, "loss": 4.0635, "step": 9379 }, { "epoch": 0.0938, "grad_norm": 1.4017661260371703, "learning_rate": 0.003, "loss": 4.0735, "step": 9380 }, { "epoch": 0.09381, "grad_norm": 1.0029510870258844, "learning_rate": 0.003, "loss": 4.0733, "step": 9381 }, { "epoch": 0.09382, "grad_norm": 1.2388574249097994, "learning_rate": 0.003, "loss": 4.0993, "step": 9382 }, { "epoch": 0.09383, "grad_norm": 1.1132940713186557, "learning_rate": 0.003, "loss": 4.0829, "step": 9383 }, { "epoch": 0.09384, "grad_norm": 1.1756128136659143, "learning_rate": 0.003, "loss": 4.1107, "step": 9384 }, { "epoch": 0.09385, "grad_norm": 1.332429527105117, "learning_rate": 0.003, "loss": 4.0968, "step": 9385 }, { "epoch": 0.09386, "grad_norm": 0.8812918899253777, "learning_rate": 0.003, "loss": 4.0575, "step": 9386 }, { "epoch": 0.09387, "grad_norm": 1.0670364087317739, "learning_rate": 0.003, "loss": 4.0959, "step": 9387 }, { "epoch": 0.09388, "grad_norm": 1.235244434189283, "learning_rate": 0.003, "loss": 4.0958, "step": 9388 }, { "epoch": 0.09389, "grad_norm": 1.1706315387987256, "learning_rate": 0.003, "loss": 4.071, "step": 9389 }, { "epoch": 0.0939, "grad_norm": 1.1499760335019689, "learning_rate": 0.003, "loss": 4.0993, "step": 9390 }, { "epoch": 0.09391, "grad_norm": 1.2531320429573474, "learning_rate": 0.003, "loss": 4.0797, "step": 9391 }, { "epoch": 0.09392, "grad_norm": 1.0209475725286998, "learning_rate": 0.003, "loss": 4.1117, "step": 9392 }, { "epoch": 0.09393, "grad_norm": 1.359927438767316, "learning_rate": 0.003, "loss": 4.1062, "step": 9393 }, { "epoch": 0.09394, "grad_norm": 0.9466831704667089, "learning_rate": 0.003, "loss": 4.0872, "step": 9394 }, { "epoch": 0.09395, "grad_norm": 1.335400407810145, "learning_rate": 0.003, "loss": 4.0983, "step": 9395 }, { "epoch": 0.09396, "grad_norm": 1.0971675919494583, "learning_rate": 0.003, "loss": 4.1229, "step": 9396 }, { "epoch": 0.09397, "grad_norm": 1.074597575443764, "learning_rate": 0.003, "loss": 4.0738, "step": 9397 }, { "epoch": 0.09398, "grad_norm": 1.0573510523767233, "learning_rate": 0.003, "loss": 4.1074, "step": 9398 }, { "epoch": 0.09399, "grad_norm": 1.1461006871137773, "learning_rate": 0.003, "loss": 4.0977, "step": 9399 }, { "epoch": 0.094, "grad_norm": 1.2871999841337791, "learning_rate": 0.003, "loss": 4.0916, "step": 9400 }, { "epoch": 0.09401, "grad_norm": 1.0662993758000978, "learning_rate": 0.003, "loss": 4.1041, "step": 9401 }, { "epoch": 0.09402, "grad_norm": 0.985967078282072, "learning_rate": 0.003, "loss": 4.0884, "step": 9402 }, { "epoch": 0.09403, "grad_norm": 1.1318935874061076, "learning_rate": 0.003, "loss": 4.1139, "step": 9403 }, { "epoch": 0.09404, "grad_norm": 1.1693036846085634, "learning_rate": 0.003, "loss": 4.072, "step": 9404 }, { "epoch": 0.09405, "grad_norm": 1.080452148293103, "learning_rate": 0.003, "loss": 4.1086, "step": 9405 }, { "epoch": 0.09406, "grad_norm": 1.1205331802372656, "learning_rate": 0.003, "loss": 4.1069, "step": 9406 }, { "epoch": 0.09407, "grad_norm": 1.0444311736562197, "learning_rate": 0.003, "loss": 4.0903, "step": 9407 }, { "epoch": 0.09408, "grad_norm": 1.176470642135001, "learning_rate": 0.003, "loss": 4.0755, "step": 9408 }, { "epoch": 0.09409, "grad_norm": 1.155930558160252, "learning_rate": 0.003, "loss": 4.1015, "step": 9409 }, { "epoch": 0.0941, "grad_norm": 0.9223475406864473, "learning_rate": 0.003, "loss": 4.1037, "step": 9410 }, { "epoch": 0.09411, "grad_norm": 1.2829733075715253, "learning_rate": 0.003, "loss": 4.118, "step": 9411 }, { "epoch": 0.09412, "grad_norm": 1.078581058592096, "learning_rate": 0.003, "loss": 4.1064, "step": 9412 }, { "epoch": 0.09413, "grad_norm": 1.0432780765545573, "learning_rate": 0.003, "loss": 4.078, "step": 9413 }, { "epoch": 0.09414, "grad_norm": 1.0829176810329322, "learning_rate": 0.003, "loss": 4.1035, "step": 9414 }, { "epoch": 0.09415, "grad_norm": 1.3236052288392064, "learning_rate": 0.003, "loss": 4.0716, "step": 9415 }, { "epoch": 0.09416, "grad_norm": 1.2301785581024418, "learning_rate": 0.003, "loss": 4.045, "step": 9416 }, { "epoch": 0.09417, "grad_norm": 1.1007603067639056, "learning_rate": 0.003, "loss": 4.0779, "step": 9417 }, { "epoch": 0.09418, "grad_norm": 1.2070889380081715, "learning_rate": 0.003, "loss": 4.08, "step": 9418 }, { "epoch": 0.09419, "grad_norm": 1.0931219718566882, "learning_rate": 0.003, "loss": 4.1126, "step": 9419 }, { "epoch": 0.0942, "grad_norm": 0.9277318132635548, "learning_rate": 0.003, "loss": 4.0827, "step": 9420 }, { "epoch": 0.09421, "grad_norm": 0.9913968322046562, "learning_rate": 0.003, "loss": 4.082, "step": 9421 }, { "epoch": 0.09422, "grad_norm": 1.265447990074078, "learning_rate": 0.003, "loss": 4.0912, "step": 9422 }, { "epoch": 0.09423, "grad_norm": 0.9313054820006388, "learning_rate": 0.003, "loss": 4.0334, "step": 9423 }, { "epoch": 0.09424, "grad_norm": 1.2579896425286146, "learning_rate": 0.003, "loss": 4.0925, "step": 9424 }, { "epoch": 0.09425, "grad_norm": 1.105309071721738, "learning_rate": 0.003, "loss": 4.1101, "step": 9425 }, { "epoch": 0.09426, "grad_norm": 1.3115119791195045, "learning_rate": 0.003, "loss": 4.1195, "step": 9426 }, { "epoch": 0.09427, "grad_norm": 1.184811038469732, "learning_rate": 0.003, "loss": 4.1001, "step": 9427 }, { "epoch": 0.09428, "grad_norm": 1.019043659850143, "learning_rate": 0.003, "loss": 4.1075, "step": 9428 }, { "epoch": 0.09429, "grad_norm": 1.1829465834546578, "learning_rate": 0.003, "loss": 4.0812, "step": 9429 }, { "epoch": 0.0943, "grad_norm": 1.1269673551587713, "learning_rate": 0.003, "loss": 4.0929, "step": 9430 }, { "epoch": 0.09431, "grad_norm": 1.1911164137240267, "learning_rate": 0.003, "loss": 4.1017, "step": 9431 }, { "epoch": 0.09432, "grad_norm": 1.0130186081326145, "learning_rate": 0.003, "loss": 4.084, "step": 9432 }, { "epoch": 0.09433, "grad_norm": 1.0546005003528371, "learning_rate": 0.003, "loss": 4.0874, "step": 9433 }, { "epoch": 0.09434, "grad_norm": 0.9341349184124893, "learning_rate": 0.003, "loss": 4.0621, "step": 9434 }, { "epoch": 0.09435, "grad_norm": 1.1083137995222276, "learning_rate": 0.003, "loss": 4.0852, "step": 9435 }, { "epoch": 0.09436, "grad_norm": 1.0654190303786433, "learning_rate": 0.003, "loss": 4.0642, "step": 9436 }, { "epoch": 0.09437, "grad_norm": 1.584058741422896, "learning_rate": 0.003, "loss": 4.0971, "step": 9437 }, { "epoch": 0.09438, "grad_norm": 0.9750544774777391, "learning_rate": 0.003, "loss": 4.1082, "step": 9438 }, { "epoch": 0.09439, "grad_norm": 1.339680164780197, "learning_rate": 0.003, "loss": 4.1071, "step": 9439 }, { "epoch": 0.0944, "grad_norm": 1.1756294393140456, "learning_rate": 0.003, "loss": 4.087, "step": 9440 }, { "epoch": 0.09441, "grad_norm": 1.014182140362658, "learning_rate": 0.003, "loss": 4.0858, "step": 9441 }, { "epoch": 0.09442, "grad_norm": 1.1662162468777555, "learning_rate": 0.003, "loss": 4.089, "step": 9442 }, { "epoch": 0.09443, "grad_norm": 0.9553948736102283, "learning_rate": 0.003, "loss": 4.1113, "step": 9443 }, { "epoch": 0.09444, "grad_norm": 1.0709848010805125, "learning_rate": 0.003, "loss": 4.1004, "step": 9444 }, { "epoch": 0.09445, "grad_norm": 1.134319736024161, "learning_rate": 0.003, "loss": 4.0929, "step": 9445 }, { "epoch": 0.09446, "grad_norm": 1.0318135027042934, "learning_rate": 0.003, "loss": 4.0787, "step": 9446 }, { "epoch": 0.09447, "grad_norm": 1.3736137282415484, "learning_rate": 0.003, "loss": 4.1081, "step": 9447 }, { "epoch": 0.09448, "grad_norm": 0.9759871083573741, "learning_rate": 0.003, "loss": 4.0958, "step": 9448 }, { "epoch": 0.09449, "grad_norm": 1.3188281568720621, "learning_rate": 0.003, "loss": 4.0916, "step": 9449 }, { "epoch": 0.0945, "grad_norm": 1.009402232799297, "learning_rate": 0.003, "loss": 4.0858, "step": 9450 }, { "epoch": 0.09451, "grad_norm": 1.2792820747685412, "learning_rate": 0.003, "loss": 4.1016, "step": 9451 }, { "epoch": 0.09452, "grad_norm": 1.0990815298004843, "learning_rate": 0.003, "loss": 4.1064, "step": 9452 }, { "epoch": 0.09453, "grad_norm": 0.8910360106491243, "learning_rate": 0.003, "loss": 4.0497, "step": 9453 }, { "epoch": 0.09454, "grad_norm": 1.1110202820179365, "learning_rate": 0.003, "loss": 4.0766, "step": 9454 }, { "epoch": 0.09455, "grad_norm": 1.034632480340015, "learning_rate": 0.003, "loss": 4.0864, "step": 9455 }, { "epoch": 0.09456, "grad_norm": 1.282774041514997, "learning_rate": 0.003, "loss": 4.0883, "step": 9456 }, { "epoch": 0.09457, "grad_norm": 1.1710239575027193, "learning_rate": 0.003, "loss": 4.1031, "step": 9457 }, { "epoch": 0.09458, "grad_norm": 1.1948449206009228, "learning_rate": 0.003, "loss": 4.0775, "step": 9458 }, { "epoch": 0.09459, "grad_norm": 1.4329498523084134, "learning_rate": 0.003, "loss": 4.1023, "step": 9459 }, { "epoch": 0.0946, "grad_norm": 1.0480088763983315, "learning_rate": 0.003, "loss": 4.1111, "step": 9460 }, { "epoch": 0.09461, "grad_norm": 1.1347178145478178, "learning_rate": 0.003, "loss": 4.1213, "step": 9461 }, { "epoch": 0.09462, "grad_norm": 1.2915927303703032, "learning_rate": 0.003, "loss": 4.073, "step": 9462 }, { "epoch": 0.09463, "grad_norm": 0.8991215034828245, "learning_rate": 0.003, "loss": 4.0735, "step": 9463 }, { "epoch": 0.09464, "grad_norm": 1.217189952244973, "learning_rate": 0.003, "loss": 4.0798, "step": 9464 }, { "epoch": 0.09465, "grad_norm": 1.108024828467612, "learning_rate": 0.003, "loss": 4.0829, "step": 9465 }, { "epoch": 0.09466, "grad_norm": 1.254719512836333, "learning_rate": 0.003, "loss": 4.0967, "step": 9466 }, { "epoch": 0.09467, "grad_norm": 1.0657274107021417, "learning_rate": 0.003, "loss": 4.0852, "step": 9467 }, { "epoch": 0.09468, "grad_norm": 1.307853078083217, "learning_rate": 0.003, "loss": 4.0951, "step": 9468 }, { "epoch": 0.09469, "grad_norm": 1.0966231654158667, "learning_rate": 0.003, "loss": 4.0633, "step": 9469 }, { "epoch": 0.0947, "grad_norm": 0.9873023581515457, "learning_rate": 0.003, "loss": 4.072, "step": 9470 }, { "epoch": 0.09471, "grad_norm": 1.0072444332654882, "learning_rate": 0.003, "loss": 4.1054, "step": 9471 }, { "epoch": 0.09472, "grad_norm": 1.217023364844429, "learning_rate": 0.003, "loss": 4.0895, "step": 9472 }, { "epoch": 0.09473, "grad_norm": 1.2084006922919652, "learning_rate": 0.003, "loss": 4.082, "step": 9473 }, { "epoch": 0.09474, "grad_norm": 1.1341037440802257, "learning_rate": 0.003, "loss": 4.0759, "step": 9474 }, { "epoch": 0.09475, "grad_norm": 1.3957096693032005, "learning_rate": 0.003, "loss": 4.0857, "step": 9475 }, { "epoch": 0.09476, "grad_norm": 1.0893090004792627, "learning_rate": 0.003, "loss": 4.0815, "step": 9476 }, { "epoch": 0.09477, "grad_norm": 1.0407040979188704, "learning_rate": 0.003, "loss": 4.0774, "step": 9477 }, { "epoch": 0.09478, "grad_norm": 1.0265803874051365, "learning_rate": 0.003, "loss": 4.063, "step": 9478 }, { "epoch": 0.09479, "grad_norm": 1.3147642641286537, "learning_rate": 0.003, "loss": 4.0866, "step": 9479 }, { "epoch": 0.0948, "grad_norm": 1.0903590993744465, "learning_rate": 0.003, "loss": 4.068, "step": 9480 }, { "epoch": 0.09481, "grad_norm": 1.1338195989175865, "learning_rate": 0.003, "loss": 4.1109, "step": 9481 }, { "epoch": 0.09482, "grad_norm": 1.1488847919171767, "learning_rate": 0.003, "loss": 4.074, "step": 9482 }, { "epoch": 0.09483, "grad_norm": 1.0895462508145521, "learning_rate": 0.003, "loss": 4.049, "step": 9483 }, { "epoch": 0.09484, "grad_norm": 1.0370843870115312, "learning_rate": 0.003, "loss": 4.0999, "step": 9484 }, { "epoch": 0.09485, "grad_norm": 1.2723261565988722, "learning_rate": 0.003, "loss": 4.09, "step": 9485 }, { "epoch": 0.09486, "grad_norm": 0.965502657440432, "learning_rate": 0.003, "loss": 4.0968, "step": 9486 }, { "epoch": 0.09487, "grad_norm": 1.342344919200925, "learning_rate": 0.003, "loss": 4.0923, "step": 9487 }, { "epoch": 0.09488, "grad_norm": 0.9393715311377178, "learning_rate": 0.003, "loss": 4.1302, "step": 9488 }, { "epoch": 0.09489, "grad_norm": 1.0372158171386994, "learning_rate": 0.003, "loss": 4.0961, "step": 9489 }, { "epoch": 0.0949, "grad_norm": 1.2723730449344455, "learning_rate": 0.003, "loss": 4.0979, "step": 9490 }, { "epoch": 0.09491, "grad_norm": 1.1531858839360378, "learning_rate": 0.003, "loss": 4.0744, "step": 9491 }, { "epoch": 0.09492, "grad_norm": 1.0917875041437035, "learning_rate": 0.003, "loss": 4.0869, "step": 9492 }, { "epoch": 0.09493, "grad_norm": 1.0917450412795457, "learning_rate": 0.003, "loss": 4.0621, "step": 9493 }, { "epoch": 0.09494, "grad_norm": 1.300674879254043, "learning_rate": 0.003, "loss": 4.1029, "step": 9494 }, { "epoch": 0.09495, "grad_norm": 1.0477357995905079, "learning_rate": 0.003, "loss": 4.0878, "step": 9495 }, { "epoch": 0.09496, "grad_norm": 1.308017981995178, "learning_rate": 0.003, "loss": 4.1079, "step": 9496 }, { "epoch": 0.09497, "grad_norm": 1.01346574500382, "learning_rate": 0.003, "loss": 4.0832, "step": 9497 }, { "epoch": 0.09498, "grad_norm": 1.2616680695782678, "learning_rate": 0.003, "loss": 4.0649, "step": 9498 }, { "epoch": 0.09499, "grad_norm": 1.0581943895228205, "learning_rate": 0.003, "loss": 4.0709, "step": 9499 }, { "epoch": 0.095, "grad_norm": 1.2120883263505524, "learning_rate": 0.003, "loss": 4.1228, "step": 9500 }, { "epoch": 0.09501, "grad_norm": 1.1721476598291545, "learning_rate": 0.003, "loss": 4.0838, "step": 9501 }, { "epoch": 0.09502, "grad_norm": 1.1119140712765754, "learning_rate": 0.003, "loss": 4.0794, "step": 9502 }, { "epoch": 0.09503, "grad_norm": 1.0214668770994324, "learning_rate": 0.003, "loss": 4.0394, "step": 9503 }, { "epoch": 0.09504, "grad_norm": 1.2759200888998818, "learning_rate": 0.003, "loss": 4.0994, "step": 9504 }, { "epoch": 0.09505, "grad_norm": 1.043511269779943, "learning_rate": 0.003, "loss": 4.0808, "step": 9505 }, { "epoch": 0.09506, "grad_norm": 1.3204985861928111, "learning_rate": 0.003, "loss": 4.1019, "step": 9506 }, { "epoch": 0.09507, "grad_norm": 0.9924814468245581, "learning_rate": 0.003, "loss": 4.1082, "step": 9507 }, { "epoch": 0.09508, "grad_norm": 1.0657775357354207, "learning_rate": 0.003, "loss": 4.0819, "step": 9508 }, { "epoch": 0.09509, "grad_norm": 1.0941505499696234, "learning_rate": 0.003, "loss": 4.081, "step": 9509 }, { "epoch": 0.0951, "grad_norm": 1.1405032961817592, "learning_rate": 0.003, "loss": 4.0906, "step": 9510 }, { "epoch": 0.09511, "grad_norm": 1.1584629996450322, "learning_rate": 0.003, "loss": 4.0799, "step": 9511 }, { "epoch": 0.09512, "grad_norm": 1.1526523340688448, "learning_rate": 0.003, "loss": 4.0851, "step": 9512 }, { "epoch": 0.09513, "grad_norm": 1.012931520636649, "learning_rate": 0.003, "loss": 4.0725, "step": 9513 }, { "epoch": 0.09514, "grad_norm": 1.3001297494253556, "learning_rate": 0.003, "loss": 4.0886, "step": 9514 }, { "epoch": 0.09515, "grad_norm": 1.0909618097726363, "learning_rate": 0.003, "loss": 4.0956, "step": 9515 }, { "epoch": 0.09516, "grad_norm": 1.1840771159783676, "learning_rate": 0.003, "loss": 4.0854, "step": 9516 }, { "epoch": 0.09517, "grad_norm": 1.2901035070823237, "learning_rate": 0.003, "loss": 4.0897, "step": 9517 }, { "epoch": 0.09518, "grad_norm": 1.071713800777275, "learning_rate": 0.003, "loss": 4.0985, "step": 9518 }, { "epoch": 0.09519, "grad_norm": 1.2588935345722223, "learning_rate": 0.003, "loss": 4.1155, "step": 9519 }, { "epoch": 0.0952, "grad_norm": 1.150982327799799, "learning_rate": 0.003, "loss": 4.0567, "step": 9520 }, { "epoch": 0.09521, "grad_norm": 1.0941667959158383, "learning_rate": 0.003, "loss": 4.0662, "step": 9521 }, { "epoch": 0.09522, "grad_norm": 1.2457329144942104, "learning_rate": 0.003, "loss": 4.0756, "step": 9522 }, { "epoch": 0.09523, "grad_norm": 1.0685171393986306, "learning_rate": 0.003, "loss": 4.0762, "step": 9523 }, { "epoch": 0.09524, "grad_norm": 1.059073090332289, "learning_rate": 0.003, "loss": 4.0702, "step": 9524 }, { "epoch": 0.09525, "grad_norm": 1.1317859296460144, "learning_rate": 0.003, "loss": 4.0908, "step": 9525 }, { "epoch": 0.09526, "grad_norm": 1.0358007539350973, "learning_rate": 0.003, "loss": 4.08, "step": 9526 }, { "epoch": 0.09527, "grad_norm": 1.1025319126502247, "learning_rate": 0.003, "loss": 4.0828, "step": 9527 }, { "epoch": 0.09528, "grad_norm": 0.9906183944830572, "learning_rate": 0.003, "loss": 4.1013, "step": 9528 }, { "epoch": 0.09529, "grad_norm": 1.2857359082767073, "learning_rate": 0.003, "loss": 4.0639, "step": 9529 }, { "epoch": 0.0953, "grad_norm": 1.096634688209568, "learning_rate": 0.003, "loss": 4.0862, "step": 9530 }, { "epoch": 0.09531, "grad_norm": 1.1987032031232345, "learning_rate": 0.003, "loss": 4.0762, "step": 9531 }, { "epoch": 0.09532, "grad_norm": 1.0394624430870323, "learning_rate": 0.003, "loss": 4.0653, "step": 9532 }, { "epoch": 0.09533, "grad_norm": 1.4862811291217963, "learning_rate": 0.003, "loss": 4.0712, "step": 9533 }, { "epoch": 0.09534, "grad_norm": 0.9465852817142347, "learning_rate": 0.003, "loss": 4.0682, "step": 9534 }, { "epoch": 0.09535, "grad_norm": 1.1139397441123668, "learning_rate": 0.003, "loss": 4.082, "step": 9535 }, { "epoch": 0.09536, "grad_norm": 1.0512790302527368, "learning_rate": 0.003, "loss": 4.0978, "step": 9536 }, { "epoch": 0.09537, "grad_norm": 1.1891251252289299, "learning_rate": 0.003, "loss": 4.1086, "step": 9537 }, { "epoch": 0.09538, "grad_norm": 1.0883015405092293, "learning_rate": 0.003, "loss": 4.0879, "step": 9538 }, { "epoch": 0.09539, "grad_norm": 1.175924001059813, "learning_rate": 0.003, "loss": 4.1051, "step": 9539 }, { "epoch": 0.0954, "grad_norm": 1.134543356215364, "learning_rate": 0.003, "loss": 4.0674, "step": 9540 }, { "epoch": 0.09541, "grad_norm": 0.9943534853511077, "learning_rate": 0.003, "loss": 4.0876, "step": 9541 }, { "epoch": 0.09542, "grad_norm": 1.2269677886331591, "learning_rate": 0.003, "loss": 4.0651, "step": 9542 }, { "epoch": 0.09543, "grad_norm": 1.1058198763957707, "learning_rate": 0.003, "loss": 4.079, "step": 9543 }, { "epoch": 0.09544, "grad_norm": 1.3515105723732794, "learning_rate": 0.003, "loss": 4.0817, "step": 9544 }, { "epoch": 0.09545, "grad_norm": 0.9298293002067652, "learning_rate": 0.003, "loss": 4.086, "step": 9545 }, { "epoch": 0.09546, "grad_norm": 1.1998465227374622, "learning_rate": 0.003, "loss": 4.1061, "step": 9546 }, { "epoch": 0.09547, "grad_norm": 1.2539130424447145, "learning_rate": 0.003, "loss": 4.098, "step": 9547 }, { "epoch": 0.09548, "grad_norm": 1.0141650543050034, "learning_rate": 0.003, "loss": 4.0834, "step": 9548 }, { "epoch": 0.09549, "grad_norm": 1.3549810054680689, "learning_rate": 0.003, "loss": 4.1249, "step": 9549 }, { "epoch": 0.0955, "grad_norm": 1.0786660493312055, "learning_rate": 0.003, "loss": 4.0629, "step": 9550 }, { "epoch": 0.09551, "grad_norm": 1.572522511143651, "learning_rate": 0.003, "loss": 4.0802, "step": 9551 }, { "epoch": 0.09552, "grad_norm": 1.0306417789967723, "learning_rate": 0.003, "loss": 4.1405, "step": 9552 }, { "epoch": 0.09553, "grad_norm": 1.2709722537106287, "learning_rate": 0.003, "loss": 4.0997, "step": 9553 }, { "epoch": 0.09554, "grad_norm": 0.9366968321825859, "learning_rate": 0.003, "loss": 4.0664, "step": 9554 }, { "epoch": 0.09555, "grad_norm": 1.0415773988291444, "learning_rate": 0.003, "loss": 4.0723, "step": 9555 }, { "epoch": 0.09556, "grad_norm": 1.1872790399809248, "learning_rate": 0.003, "loss": 4.0871, "step": 9556 }, { "epoch": 0.09557, "grad_norm": 1.0918558120913107, "learning_rate": 0.003, "loss": 4.1047, "step": 9557 }, { "epoch": 0.09558, "grad_norm": 1.227846572726307, "learning_rate": 0.003, "loss": 4.0932, "step": 9558 }, { "epoch": 0.09559, "grad_norm": 1.2386135780006378, "learning_rate": 0.003, "loss": 4.1189, "step": 9559 }, { "epoch": 0.0956, "grad_norm": 1.0618652016313734, "learning_rate": 0.003, "loss": 4.0758, "step": 9560 }, { "epoch": 0.09561, "grad_norm": 1.119890477915479, "learning_rate": 0.003, "loss": 4.1061, "step": 9561 }, { "epoch": 0.09562, "grad_norm": 1.016523442218028, "learning_rate": 0.003, "loss": 4.0526, "step": 9562 }, { "epoch": 0.09563, "grad_norm": 1.1785565929403774, "learning_rate": 0.003, "loss": 4.1249, "step": 9563 }, { "epoch": 0.09564, "grad_norm": 0.9337669267700384, "learning_rate": 0.003, "loss": 4.1066, "step": 9564 }, { "epoch": 0.09565, "grad_norm": 1.0780917400747017, "learning_rate": 0.003, "loss": 4.0767, "step": 9565 }, { "epoch": 0.09566, "grad_norm": 1.195493305871725, "learning_rate": 0.003, "loss": 4.092, "step": 9566 }, { "epoch": 0.09567, "grad_norm": 1.156333194607271, "learning_rate": 0.003, "loss": 4.0867, "step": 9567 }, { "epoch": 0.09568, "grad_norm": 1.1350215045863368, "learning_rate": 0.003, "loss": 4.0907, "step": 9568 }, { "epoch": 0.09569, "grad_norm": 1.2348564410977743, "learning_rate": 0.003, "loss": 4.0984, "step": 9569 }, { "epoch": 0.0957, "grad_norm": 1.0560118100188587, "learning_rate": 0.003, "loss": 4.0933, "step": 9570 }, { "epoch": 0.09571, "grad_norm": 1.1086095630369532, "learning_rate": 0.003, "loss": 4.0927, "step": 9571 }, { "epoch": 0.09572, "grad_norm": 1.0193857583153085, "learning_rate": 0.003, "loss": 4.0849, "step": 9572 }, { "epoch": 0.09573, "grad_norm": 1.2887839228489493, "learning_rate": 0.003, "loss": 4.0915, "step": 9573 }, { "epoch": 0.09574, "grad_norm": 1.0614328188485582, "learning_rate": 0.003, "loss": 4.0885, "step": 9574 }, { "epoch": 0.09575, "grad_norm": 1.4319941739250786, "learning_rate": 0.003, "loss": 4.0765, "step": 9575 }, { "epoch": 0.09576, "grad_norm": 0.8356787684459427, "learning_rate": 0.003, "loss": 4.0687, "step": 9576 }, { "epoch": 0.09577, "grad_norm": 0.9948210584003812, "learning_rate": 0.003, "loss": 4.0572, "step": 9577 }, { "epoch": 0.09578, "grad_norm": 1.3921854113957979, "learning_rate": 0.003, "loss": 4.0629, "step": 9578 }, { "epoch": 0.09579, "grad_norm": 1.0864312621607588, "learning_rate": 0.003, "loss": 4.0663, "step": 9579 }, { "epoch": 0.0958, "grad_norm": 1.1156666180695558, "learning_rate": 0.003, "loss": 4.0996, "step": 9580 }, { "epoch": 0.09581, "grad_norm": 1.0322074629753206, "learning_rate": 0.003, "loss": 4.1318, "step": 9581 }, { "epoch": 0.09582, "grad_norm": 1.1315310163551284, "learning_rate": 0.003, "loss": 4.0751, "step": 9582 }, { "epoch": 0.09583, "grad_norm": 1.115818784532022, "learning_rate": 0.003, "loss": 4.0628, "step": 9583 }, { "epoch": 0.09584, "grad_norm": 1.0450540579600909, "learning_rate": 0.003, "loss": 4.0839, "step": 9584 }, { "epoch": 0.09585, "grad_norm": 1.036975398253159, "learning_rate": 0.003, "loss": 4.0667, "step": 9585 }, { "epoch": 0.09586, "grad_norm": 1.1315376448061392, "learning_rate": 0.003, "loss": 4.1021, "step": 9586 }, { "epoch": 0.09587, "grad_norm": 0.8801716853178653, "learning_rate": 0.003, "loss": 4.0446, "step": 9587 }, { "epoch": 0.09588, "grad_norm": 1.1979096575235726, "learning_rate": 0.003, "loss": 4.0878, "step": 9588 }, { "epoch": 0.09589, "grad_norm": 1.238757685421794, "learning_rate": 0.003, "loss": 4.09, "step": 9589 }, { "epoch": 0.0959, "grad_norm": 1.1267556925308015, "learning_rate": 0.003, "loss": 4.1013, "step": 9590 }, { "epoch": 0.09591, "grad_norm": 1.130584073648821, "learning_rate": 0.003, "loss": 4.1296, "step": 9591 }, { "epoch": 0.09592, "grad_norm": 1.0738133917355688, "learning_rate": 0.003, "loss": 4.0941, "step": 9592 }, { "epoch": 0.09593, "grad_norm": 1.1776732863268307, "learning_rate": 0.003, "loss": 4.1073, "step": 9593 }, { "epoch": 0.09594, "grad_norm": 1.0443759893082762, "learning_rate": 0.003, "loss": 4.0721, "step": 9594 }, { "epoch": 0.09595, "grad_norm": 1.3547206567603112, "learning_rate": 0.003, "loss": 4.085, "step": 9595 }, { "epoch": 0.09596, "grad_norm": 0.9752910801651283, "learning_rate": 0.003, "loss": 4.0786, "step": 9596 }, { "epoch": 0.09597, "grad_norm": 1.4882322897015248, "learning_rate": 0.003, "loss": 4.0933, "step": 9597 }, { "epoch": 0.09598, "grad_norm": 1.1972088601549762, "learning_rate": 0.003, "loss": 4.1001, "step": 9598 }, { "epoch": 0.09599, "grad_norm": 1.2882807539335117, "learning_rate": 0.003, "loss": 4.0782, "step": 9599 }, { "epoch": 0.096, "grad_norm": 1.2197457372448262, "learning_rate": 0.003, "loss": 4.0821, "step": 9600 }, { "epoch": 0.09601, "grad_norm": 1.0026983299265844, "learning_rate": 0.003, "loss": 4.083, "step": 9601 }, { "epoch": 0.09602, "grad_norm": 1.0359963184067358, "learning_rate": 0.003, "loss": 4.1124, "step": 9602 }, { "epoch": 0.09603, "grad_norm": 1.0929372806073887, "learning_rate": 0.003, "loss": 4.0917, "step": 9603 }, { "epoch": 0.09604, "grad_norm": 1.048976558304099, "learning_rate": 0.003, "loss": 4.0546, "step": 9604 }, { "epoch": 0.09605, "grad_norm": 1.133185053036612, "learning_rate": 0.003, "loss": 4.1014, "step": 9605 }, { "epoch": 0.09606, "grad_norm": 1.3381830624612678, "learning_rate": 0.003, "loss": 4.102, "step": 9606 }, { "epoch": 0.09607, "grad_norm": 1.0571967293380204, "learning_rate": 0.003, "loss": 4.1066, "step": 9607 }, { "epoch": 0.09608, "grad_norm": 1.2880171482068876, "learning_rate": 0.003, "loss": 4.1229, "step": 9608 }, { "epoch": 0.09609, "grad_norm": 1.1090349823779482, "learning_rate": 0.003, "loss": 4.0937, "step": 9609 }, { "epoch": 0.0961, "grad_norm": 1.0287658024244641, "learning_rate": 0.003, "loss": 4.0899, "step": 9610 }, { "epoch": 0.09611, "grad_norm": 1.0738364101326499, "learning_rate": 0.003, "loss": 4.0882, "step": 9611 }, { "epoch": 0.09612, "grad_norm": 1.282326636020608, "learning_rate": 0.003, "loss": 4.0947, "step": 9612 }, { "epoch": 0.09613, "grad_norm": 1.3701329621538962, "learning_rate": 0.003, "loss": 4.1095, "step": 9613 }, { "epoch": 0.09614, "grad_norm": 1.1634185313174656, "learning_rate": 0.003, "loss": 4.0968, "step": 9614 }, { "epoch": 0.09615, "grad_norm": 1.1017589773949448, "learning_rate": 0.003, "loss": 4.0689, "step": 9615 }, { "epoch": 0.09616, "grad_norm": 1.1956757667716347, "learning_rate": 0.003, "loss": 4.0975, "step": 9616 }, { "epoch": 0.09617, "grad_norm": 1.0483299135427413, "learning_rate": 0.003, "loss": 4.0681, "step": 9617 }, { "epoch": 0.09618, "grad_norm": 1.0207002766117625, "learning_rate": 0.003, "loss": 4.0881, "step": 9618 }, { "epoch": 0.09619, "grad_norm": 1.1687767625123595, "learning_rate": 0.003, "loss": 4.0847, "step": 9619 }, { "epoch": 0.0962, "grad_norm": 1.2124142030797413, "learning_rate": 0.003, "loss": 4.0747, "step": 9620 }, { "epoch": 0.09621, "grad_norm": 1.0427796712758857, "learning_rate": 0.003, "loss": 4.0907, "step": 9621 }, { "epoch": 0.09622, "grad_norm": 1.1321557809894192, "learning_rate": 0.003, "loss": 4.0779, "step": 9622 }, { "epoch": 0.09623, "grad_norm": 1.080551373298356, "learning_rate": 0.003, "loss": 4.0761, "step": 9623 }, { "epoch": 0.09624, "grad_norm": 1.2231836247934575, "learning_rate": 0.003, "loss": 4.093, "step": 9624 }, { "epoch": 0.09625, "grad_norm": 0.9743054244901846, "learning_rate": 0.003, "loss": 4.0653, "step": 9625 }, { "epoch": 0.09626, "grad_norm": 1.1144708979343558, "learning_rate": 0.003, "loss": 4.078, "step": 9626 }, { "epoch": 0.09627, "grad_norm": 1.0555475417358153, "learning_rate": 0.003, "loss": 4.0746, "step": 9627 }, { "epoch": 0.09628, "grad_norm": 1.1998838196216295, "learning_rate": 0.003, "loss": 4.0843, "step": 9628 }, { "epoch": 0.09629, "grad_norm": 1.29917367128637, "learning_rate": 0.003, "loss": 4.1017, "step": 9629 }, { "epoch": 0.0963, "grad_norm": 1.3338654943176473, "learning_rate": 0.003, "loss": 4.0889, "step": 9630 }, { "epoch": 0.09631, "grad_norm": 1.133478955698145, "learning_rate": 0.003, "loss": 4.1218, "step": 9631 }, { "epoch": 0.09632, "grad_norm": 1.2159206106180984, "learning_rate": 0.003, "loss": 4.0567, "step": 9632 }, { "epoch": 0.09633, "grad_norm": 1.3893586132647013, "learning_rate": 0.003, "loss": 4.0795, "step": 9633 }, { "epoch": 0.09634, "grad_norm": 1.0477658736391087, "learning_rate": 0.003, "loss": 4.0608, "step": 9634 }, { "epoch": 0.09635, "grad_norm": 1.0397822704281512, "learning_rate": 0.003, "loss": 4.0977, "step": 9635 }, { "epoch": 0.09636, "grad_norm": 1.2637642360376349, "learning_rate": 0.003, "loss": 4.1085, "step": 9636 }, { "epoch": 0.09637, "grad_norm": 0.9333047922469344, "learning_rate": 0.003, "loss": 4.0699, "step": 9637 }, { "epoch": 0.09638, "grad_norm": 1.1332885470676644, "learning_rate": 0.003, "loss": 4.0648, "step": 9638 }, { "epoch": 0.09639, "grad_norm": 1.160189932847423, "learning_rate": 0.003, "loss": 4.0741, "step": 9639 }, { "epoch": 0.0964, "grad_norm": 1.175168237055286, "learning_rate": 0.003, "loss": 4.1189, "step": 9640 }, { "epoch": 0.09641, "grad_norm": 1.130167062520008, "learning_rate": 0.003, "loss": 4.0572, "step": 9641 }, { "epoch": 0.09642, "grad_norm": 1.0245340911895193, "learning_rate": 0.003, "loss": 4.109, "step": 9642 }, { "epoch": 0.09643, "grad_norm": 1.2253638168085181, "learning_rate": 0.003, "loss": 4.088, "step": 9643 }, { "epoch": 0.09644, "grad_norm": 1.0444677924566768, "learning_rate": 0.003, "loss": 4.1031, "step": 9644 }, { "epoch": 0.09645, "grad_norm": 1.1654182478122446, "learning_rate": 0.003, "loss": 4.0956, "step": 9645 }, { "epoch": 0.09646, "grad_norm": 1.1443250863557537, "learning_rate": 0.003, "loss": 4.0902, "step": 9646 }, { "epoch": 0.09647, "grad_norm": 1.167791156046563, "learning_rate": 0.003, "loss": 4.0504, "step": 9647 }, { "epoch": 0.09648, "grad_norm": 1.1947555409014852, "learning_rate": 0.003, "loss": 4.1242, "step": 9648 }, { "epoch": 0.09649, "grad_norm": 1.2907277869077691, "learning_rate": 0.003, "loss": 4.0641, "step": 9649 }, { "epoch": 0.0965, "grad_norm": 1.0649464399284445, "learning_rate": 0.003, "loss": 4.0707, "step": 9650 }, { "epoch": 0.09651, "grad_norm": 1.167094081630458, "learning_rate": 0.003, "loss": 4.0925, "step": 9651 }, { "epoch": 0.09652, "grad_norm": 1.0329533755389784, "learning_rate": 0.003, "loss": 4.08, "step": 9652 }, { "epoch": 0.09653, "grad_norm": 1.2268437129588268, "learning_rate": 0.003, "loss": 4.1075, "step": 9653 }, { "epoch": 0.09654, "grad_norm": 1.1469799145575592, "learning_rate": 0.003, "loss": 4.059, "step": 9654 }, { "epoch": 0.09655, "grad_norm": 1.2112360987709447, "learning_rate": 0.003, "loss": 4.0942, "step": 9655 }, { "epoch": 0.09656, "grad_norm": 1.0727040957696266, "learning_rate": 0.003, "loss": 4.102, "step": 9656 }, { "epoch": 0.09657, "grad_norm": 1.1776494445532388, "learning_rate": 0.003, "loss": 4.1, "step": 9657 }, { "epoch": 0.09658, "grad_norm": 1.1303799155061847, "learning_rate": 0.003, "loss": 4.0778, "step": 9658 }, { "epoch": 0.09659, "grad_norm": 1.2571601860110444, "learning_rate": 0.003, "loss": 4.1112, "step": 9659 }, { "epoch": 0.0966, "grad_norm": 1.0693884401717517, "learning_rate": 0.003, "loss": 4.1007, "step": 9660 }, { "epoch": 0.09661, "grad_norm": 1.004175075932187, "learning_rate": 0.003, "loss": 4.0682, "step": 9661 }, { "epoch": 0.09662, "grad_norm": 1.2266057039354872, "learning_rate": 0.003, "loss": 4.0789, "step": 9662 }, { "epoch": 0.09663, "grad_norm": 0.9543532741319573, "learning_rate": 0.003, "loss": 4.1053, "step": 9663 }, { "epoch": 0.09664, "grad_norm": 1.3902940709187628, "learning_rate": 0.003, "loss": 4.1064, "step": 9664 }, { "epoch": 0.09665, "grad_norm": 0.9943600842844004, "learning_rate": 0.003, "loss": 4.0818, "step": 9665 }, { "epoch": 0.09666, "grad_norm": 1.1758873634953837, "learning_rate": 0.003, "loss": 4.101, "step": 9666 }, { "epoch": 0.09667, "grad_norm": 0.9845801901236461, "learning_rate": 0.003, "loss": 4.0969, "step": 9667 }, { "epoch": 0.09668, "grad_norm": 1.218866323888133, "learning_rate": 0.003, "loss": 4.0788, "step": 9668 }, { "epoch": 0.09669, "grad_norm": 1.153506103770982, "learning_rate": 0.003, "loss": 4.0836, "step": 9669 }, { "epoch": 0.0967, "grad_norm": 1.1368038601399784, "learning_rate": 0.003, "loss": 4.0988, "step": 9670 }, { "epoch": 0.09671, "grad_norm": 1.4006249826823316, "learning_rate": 0.003, "loss": 4.0845, "step": 9671 }, { "epoch": 0.09672, "grad_norm": 0.9161827917716768, "learning_rate": 0.003, "loss": 4.1058, "step": 9672 }, { "epoch": 0.09673, "grad_norm": 1.0835717715634787, "learning_rate": 0.003, "loss": 4.106, "step": 9673 }, { "epoch": 0.09674, "grad_norm": 1.0399068019858488, "learning_rate": 0.003, "loss": 4.1367, "step": 9674 }, { "epoch": 0.09675, "grad_norm": 1.0469774866570443, "learning_rate": 0.003, "loss": 4.0726, "step": 9675 }, { "epoch": 0.09676, "grad_norm": 1.2806357843149219, "learning_rate": 0.003, "loss": 4.0955, "step": 9676 }, { "epoch": 0.09677, "grad_norm": 1.009853061676496, "learning_rate": 0.003, "loss": 4.0509, "step": 9677 }, { "epoch": 0.09678, "grad_norm": 1.1540167046393022, "learning_rate": 0.003, "loss": 4.1157, "step": 9678 }, { "epoch": 0.09679, "grad_norm": 1.0771299268306656, "learning_rate": 0.003, "loss": 4.103, "step": 9679 }, { "epoch": 0.0968, "grad_norm": 1.2351879811153652, "learning_rate": 0.003, "loss": 4.0941, "step": 9680 }, { "epoch": 0.09681, "grad_norm": 1.0473582360721734, "learning_rate": 0.003, "loss": 4.0985, "step": 9681 }, { "epoch": 0.09682, "grad_norm": 1.1020137212667507, "learning_rate": 0.003, "loss": 4.0839, "step": 9682 }, { "epoch": 0.09683, "grad_norm": 1.1966346222569244, "learning_rate": 0.003, "loss": 4.0792, "step": 9683 }, { "epoch": 0.09684, "grad_norm": 1.2461509613901234, "learning_rate": 0.003, "loss": 4.1094, "step": 9684 }, { "epoch": 0.09685, "grad_norm": 1.1552960648766668, "learning_rate": 0.003, "loss": 4.0892, "step": 9685 }, { "epoch": 0.09686, "grad_norm": 1.0019986633187414, "learning_rate": 0.003, "loss": 4.0918, "step": 9686 }, { "epoch": 0.09687, "grad_norm": 1.3405619727570803, "learning_rate": 0.003, "loss": 4.0997, "step": 9687 }, { "epoch": 0.09688, "grad_norm": 0.9274861169023544, "learning_rate": 0.003, "loss": 4.1183, "step": 9688 }, { "epoch": 0.09689, "grad_norm": 1.0221235959428148, "learning_rate": 0.003, "loss": 4.0765, "step": 9689 }, { "epoch": 0.0969, "grad_norm": 1.24021442454874, "learning_rate": 0.003, "loss": 4.0961, "step": 9690 }, { "epoch": 0.09691, "grad_norm": 1.0627861292781198, "learning_rate": 0.003, "loss": 4.1072, "step": 9691 }, { "epoch": 0.09692, "grad_norm": 1.2282733803147996, "learning_rate": 0.003, "loss": 4.1056, "step": 9692 }, { "epoch": 0.09693, "grad_norm": 1.1357687713326148, "learning_rate": 0.003, "loss": 4.068, "step": 9693 }, { "epoch": 0.09694, "grad_norm": 1.5228079240920682, "learning_rate": 0.003, "loss": 4.0911, "step": 9694 }, { "epoch": 0.09695, "grad_norm": 1.0140611378984512, "learning_rate": 0.003, "loss": 4.0975, "step": 9695 }, { "epoch": 0.09696, "grad_norm": 1.3988967695816317, "learning_rate": 0.003, "loss": 4.0892, "step": 9696 }, { "epoch": 0.09697, "grad_norm": 0.9961426808869265, "learning_rate": 0.003, "loss": 4.0978, "step": 9697 }, { "epoch": 0.09698, "grad_norm": 1.1658655050212485, "learning_rate": 0.003, "loss": 4.1085, "step": 9698 }, { "epoch": 0.09699, "grad_norm": 1.0747064995848443, "learning_rate": 0.003, "loss": 4.0795, "step": 9699 }, { "epoch": 0.097, "grad_norm": 1.3446822786713644, "learning_rate": 0.003, "loss": 4.1236, "step": 9700 }, { "epoch": 0.09701, "grad_norm": 0.8646827591218209, "learning_rate": 0.003, "loss": 4.0916, "step": 9701 }, { "epoch": 0.09702, "grad_norm": 1.068401196167419, "learning_rate": 0.003, "loss": 4.0742, "step": 9702 }, { "epoch": 0.09703, "grad_norm": 1.3695048401575418, "learning_rate": 0.003, "loss": 4.091, "step": 9703 }, { "epoch": 0.09704, "grad_norm": 1.108979607117454, "learning_rate": 0.003, "loss": 4.091, "step": 9704 }, { "epoch": 0.09705, "grad_norm": 1.225092292145554, "learning_rate": 0.003, "loss": 4.0817, "step": 9705 }, { "epoch": 0.09706, "grad_norm": 0.8926937513877116, "learning_rate": 0.003, "loss": 4.0931, "step": 9706 }, { "epoch": 0.09707, "grad_norm": 1.0510954237421852, "learning_rate": 0.003, "loss": 4.0582, "step": 9707 }, { "epoch": 0.09708, "grad_norm": 1.2904843277354863, "learning_rate": 0.003, "loss": 4.0628, "step": 9708 }, { "epoch": 0.09709, "grad_norm": 0.9595584155271577, "learning_rate": 0.003, "loss": 4.0916, "step": 9709 }, { "epoch": 0.0971, "grad_norm": 1.4667900800195917, "learning_rate": 0.003, "loss": 4.0898, "step": 9710 }, { "epoch": 0.09711, "grad_norm": 0.8316555422617203, "learning_rate": 0.003, "loss": 4.0885, "step": 9711 }, { "epoch": 0.09712, "grad_norm": 0.9443133434214869, "learning_rate": 0.003, "loss": 4.1253, "step": 9712 }, { "epoch": 0.09713, "grad_norm": 1.3054408816764862, "learning_rate": 0.003, "loss": 4.0576, "step": 9713 }, { "epoch": 0.09714, "grad_norm": 0.9463113118510074, "learning_rate": 0.003, "loss": 4.0704, "step": 9714 }, { "epoch": 0.09715, "grad_norm": 0.9711389233571197, "learning_rate": 0.003, "loss": 4.0949, "step": 9715 }, { "epoch": 0.09716, "grad_norm": 1.2214846668210972, "learning_rate": 0.003, "loss": 4.0651, "step": 9716 }, { "epoch": 0.09717, "grad_norm": 1.0151000634517393, "learning_rate": 0.003, "loss": 4.1156, "step": 9717 }, { "epoch": 0.09718, "grad_norm": 1.4019736336963942, "learning_rate": 0.003, "loss": 4.094, "step": 9718 }, { "epoch": 0.09719, "grad_norm": 1.3303176085468154, "learning_rate": 0.003, "loss": 4.0801, "step": 9719 }, { "epoch": 0.0972, "grad_norm": 1.1313634808344284, "learning_rate": 0.003, "loss": 4.0739, "step": 9720 }, { "epoch": 0.09721, "grad_norm": 1.2746221665923179, "learning_rate": 0.003, "loss": 4.1114, "step": 9721 }, { "epoch": 0.09722, "grad_norm": 0.892644565526494, "learning_rate": 0.003, "loss": 4.0834, "step": 9722 }, { "epoch": 0.09723, "grad_norm": 0.9551155117082032, "learning_rate": 0.003, "loss": 4.0507, "step": 9723 }, { "epoch": 0.09724, "grad_norm": 1.19901007204341, "learning_rate": 0.003, "loss": 4.0756, "step": 9724 }, { "epoch": 0.09725, "grad_norm": 1.1178173811513445, "learning_rate": 0.003, "loss": 4.0321, "step": 9725 }, { "epoch": 0.09726, "grad_norm": 1.3145957793294916, "learning_rate": 0.003, "loss": 4.0963, "step": 9726 }, { "epoch": 0.09727, "grad_norm": 0.9933567371016185, "learning_rate": 0.003, "loss": 4.0809, "step": 9727 }, { "epoch": 0.09728, "grad_norm": 1.1580376397503649, "learning_rate": 0.003, "loss": 4.0782, "step": 9728 }, { "epoch": 0.09729, "grad_norm": 0.9180435605034515, "learning_rate": 0.003, "loss": 4.0776, "step": 9729 }, { "epoch": 0.0973, "grad_norm": 1.1007443018077367, "learning_rate": 0.003, "loss": 4.0905, "step": 9730 }, { "epoch": 0.09731, "grad_norm": 1.3412829735980611, "learning_rate": 0.003, "loss": 4.0467, "step": 9731 }, { "epoch": 0.09732, "grad_norm": 1.270840362188865, "learning_rate": 0.003, "loss": 4.0905, "step": 9732 }, { "epoch": 0.09733, "grad_norm": 1.1412059933750884, "learning_rate": 0.003, "loss": 4.0751, "step": 9733 }, { "epoch": 0.09734, "grad_norm": 1.02676220561073, "learning_rate": 0.003, "loss": 4.1044, "step": 9734 }, { "epoch": 0.09735, "grad_norm": 1.131065828153314, "learning_rate": 0.003, "loss": 4.1027, "step": 9735 }, { "epoch": 0.09736, "grad_norm": 1.047750185255067, "learning_rate": 0.003, "loss": 4.0776, "step": 9736 }, { "epoch": 0.09737, "grad_norm": 1.0697443599529262, "learning_rate": 0.003, "loss": 4.0334, "step": 9737 }, { "epoch": 0.09738, "grad_norm": 1.085106830258747, "learning_rate": 0.003, "loss": 4.0998, "step": 9738 }, { "epoch": 0.09739, "grad_norm": 1.0775630395031826, "learning_rate": 0.003, "loss": 4.0953, "step": 9739 }, { "epoch": 0.0974, "grad_norm": 1.1335140983423713, "learning_rate": 0.003, "loss": 4.0858, "step": 9740 }, { "epoch": 0.09741, "grad_norm": 1.1750684356779817, "learning_rate": 0.003, "loss": 4.131, "step": 9741 }, { "epoch": 0.09742, "grad_norm": 1.3393502821906345, "learning_rate": 0.003, "loss": 4.0742, "step": 9742 }, { "epoch": 0.09743, "grad_norm": 0.8102089351238771, "learning_rate": 0.003, "loss": 4.0934, "step": 9743 }, { "epoch": 0.09744, "grad_norm": 0.9575256130991701, "learning_rate": 0.003, "loss": 4.0557, "step": 9744 }, { "epoch": 0.09745, "grad_norm": 1.3204084853812648, "learning_rate": 0.003, "loss": 4.118, "step": 9745 }, { "epoch": 0.09746, "grad_norm": 1.1649043514340434, "learning_rate": 0.003, "loss": 4.0699, "step": 9746 }, { "epoch": 0.09747, "grad_norm": 1.2014671262072751, "learning_rate": 0.003, "loss": 4.0883, "step": 9747 }, { "epoch": 0.09748, "grad_norm": 1.1721400848514163, "learning_rate": 0.003, "loss": 4.1202, "step": 9748 }, { "epoch": 0.09749, "grad_norm": 0.9427633072020334, "learning_rate": 0.003, "loss": 4.0578, "step": 9749 }, { "epoch": 0.0975, "grad_norm": 1.2436393502737977, "learning_rate": 0.003, "loss": 4.0753, "step": 9750 }, { "epoch": 0.09751, "grad_norm": 0.9618810694053324, "learning_rate": 0.003, "loss": 4.0433, "step": 9751 }, { "epoch": 0.09752, "grad_norm": 1.113371972687393, "learning_rate": 0.003, "loss": 4.0928, "step": 9752 }, { "epoch": 0.09753, "grad_norm": 1.1960548546905854, "learning_rate": 0.003, "loss": 4.0477, "step": 9753 }, { "epoch": 0.09754, "grad_norm": 1.1085263084596224, "learning_rate": 0.003, "loss": 4.0951, "step": 9754 }, { "epoch": 0.09755, "grad_norm": 1.2291799079319456, "learning_rate": 0.003, "loss": 4.0693, "step": 9755 }, { "epoch": 0.09756, "grad_norm": 1.1359054436090914, "learning_rate": 0.003, "loss": 4.0709, "step": 9756 }, { "epoch": 0.09757, "grad_norm": 1.315833425223894, "learning_rate": 0.003, "loss": 4.0847, "step": 9757 }, { "epoch": 0.09758, "grad_norm": 0.9137304753604574, "learning_rate": 0.003, "loss": 4.0768, "step": 9758 }, { "epoch": 0.09759, "grad_norm": 1.1037158944833518, "learning_rate": 0.003, "loss": 4.0814, "step": 9759 }, { "epoch": 0.0976, "grad_norm": 1.27332720944019, "learning_rate": 0.003, "loss": 4.0897, "step": 9760 }, { "epoch": 0.09761, "grad_norm": 1.0871017800919058, "learning_rate": 0.003, "loss": 4.0772, "step": 9761 }, { "epoch": 0.09762, "grad_norm": 1.4097920604765457, "learning_rate": 0.003, "loss": 4.1072, "step": 9762 }, { "epoch": 0.09763, "grad_norm": 1.0543709201081772, "learning_rate": 0.003, "loss": 4.055, "step": 9763 }, { "epoch": 0.09764, "grad_norm": 1.3174829695931187, "learning_rate": 0.003, "loss": 4.061, "step": 9764 }, { "epoch": 0.09765, "grad_norm": 1.0572347247872953, "learning_rate": 0.003, "loss": 4.1136, "step": 9765 }, { "epoch": 0.09766, "grad_norm": 1.2330553046659682, "learning_rate": 0.003, "loss": 4.0896, "step": 9766 }, { "epoch": 0.09767, "grad_norm": 1.1251012136521312, "learning_rate": 0.003, "loss": 4.1021, "step": 9767 }, { "epoch": 0.09768, "grad_norm": 1.207157802526716, "learning_rate": 0.003, "loss": 4.0736, "step": 9768 }, { "epoch": 0.09769, "grad_norm": 1.1489618911216242, "learning_rate": 0.003, "loss": 4.1081, "step": 9769 }, { "epoch": 0.0977, "grad_norm": 1.0543591815664535, "learning_rate": 0.003, "loss": 4.1022, "step": 9770 }, { "epoch": 0.09771, "grad_norm": 1.0711936690272732, "learning_rate": 0.003, "loss": 4.0735, "step": 9771 }, { "epoch": 0.09772, "grad_norm": 1.0665693361733348, "learning_rate": 0.003, "loss": 4.0698, "step": 9772 }, { "epoch": 0.09773, "grad_norm": 1.2726662562614708, "learning_rate": 0.003, "loss": 4.0918, "step": 9773 }, { "epoch": 0.09774, "grad_norm": 1.056717609884744, "learning_rate": 0.003, "loss": 4.1061, "step": 9774 }, { "epoch": 0.09775, "grad_norm": 1.300907210522145, "learning_rate": 0.003, "loss": 4.1071, "step": 9775 }, { "epoch": 0.09776, "grad_norm": 1.1616423803262845, "learning_rate": 0.003, "loss": 4.0755, "step": 9776 }, { "epoch": 0.09777, "grad_norm": 1.1101219869879306, "learning_rate": 0.003, "loss": 4.1008, "step": 9777 }, { "epoch": 0.09778, "grad_norm": 1.1601171778844346, "learning_rate": 0.003, "loss": 4.0622, "step": 9778 }, { "epoch": 0.09779, "grad_norm": 1.0863832514735121, "learning_rate": 0.003, "loss": 4.0745, "step": 9779 }, { "epoch": 0.0978, "grad_norm": 1.0996842022521216, "learning_rate": 0.003, "loss": 4.0723, "step": 9780 }, { "epoch": 0.09781, "grad_norm": 1.2011284768026025, "learning_rate": 0.003, "loss": 4.0912, "step": 9781 }, { "epoch": 0.09782, "grad_norm": 1.1251539953516647, "learning_rate": 0.003, "loss": 4.0882, "step": 9782 }, { "epoch": 0.09783, "grad_norm": 1.0937091196835855, "learning_rate": 0.003, "loss": 4.0628, "step": 9783 }, { "epoch": 0.09784, "grad_norm": 1.401979999650129, "learning_rate": 0.003, "loss": 4.0476, "step": 9784 }, { "epoch": 0.09785, "grad_norm": 1.2484479648172329, "learning_rate": 0.003, "loss": 4.0859, "step": 9785 }, { "epoch": 0.09786, "grad_norm": 0.9741916665639245, "learning_rate": 0.003, "loss": 4.0687, "step": 9786 }, { "epoch": 0.09787, "grad_norm": 0.9768176115546604, "learning_rate": 0.003, "loss": 4.0668, "step": 9787 }, { "epoch": 0.09788, "grad_norm": 1.2418178056224656, "learning_rate": 0.003, "loss": 4.0934, "step": 9788 }, { "epoch": 0.09789, "grad_norm": 1.0229569058564036, "learning_rate": 0.003, "loss": 4.0601, "step": 9789 }, { "epoch": 0.0979, "grad_norm": 1.3990216170496972, "learning_rate": 0.003, "loss": 4.0994, "step": 9790 }, { "epoch": 0.09791, "grad_norm": 1.054322946861501, "learning_rate": 0.003, "loss": 4.0963, "step": 9791 }, { "epoch": 0.09792, "grad_norm": 1.0079326867694463, "learning_rate": 0.003, "loss": 4.0643, "step": 9792 }, { "epoch": 0.09793, "grad_norm": 1.1905527226997223, "learning_rate": 0.003, "loss": 4.0811, "step": 9793 }, { "epoch": 0.09794, "grad_norm": 1.2912865761568724, "learning_rate": 0.003, "loss": 4.094, "step": 9794 }, { "epoch": 0.09795, "grad_norm": 1.0946945932266972, "learning_rate": 0.003, "loss": 4.1032, "step": 9795 }, { "epoch": 0.09796, "grad_norm": 1.2185676619450705, "learning_rate": 0.003, "loss": 4.0804, "step": 9796 }, { "epoch": 0.09797, "grad_norm": 1.283197500753962, "learning_rate": 0.003, "loss": 4.084, "step": 9797 }, { "epoch": 0.09798, "grad_norm": 0.9618699050164604, "learning_rate": 0.003, "loss": 4.0745, "step": 9798 }, { "epoch": 0.09799, "grad_norm": 1.0714888118856691, "learning_rate": 0.003, "loss": 4.1065, "step": 9799 }, { "epoch": 0.098, "grad_norm": 1.1761297292426933, "learning_rate": 0.003, "loss": 4.0841, "step": 9800 }, { "epoch": 0.09801, "grad_norm": 1.2281890490435963, "learning_rate": 0.003, "loss": 4.0895, "step": 9801 }, { "epoch": 0.09802, "grad_norm": 1.0664551725656253, "learning_rate": 0.003, "loss": 4.0746, "step": 9802 }, { "epoch": 0.09803, "grad_norm": 1.1046441878202422, "learning_rate": 0.003, "loss": 4.0675, "step": 9803 }, { "epoch": 0.09804, "grad_norm": 0.9496888889795306, "learning_rate": 0.003, "loss": 4.0695, "step": 9804 }, { "epoch": 0.09805, "grad_norm": 1.0844565308342573, "learning_rate": 0.003, "loss": 4.064, "step": 9805 }, { "epoch": 0.09806, "grad_norm": 1.0849479136330835, "learning_rate": 0.003, "loss": 4.0602, "step": 9806 }, { "epoch": 0.09807, "grad_norm": 1.1895584455614043, "learning_rate": 0.003, "loss": 4.0806, "step": 9807 }, { "epoch": 0.09808, "grad_norm": 1.0288089992322405, "learning_rate": 0.003, "loss": 4.1084, "step": 9808 }, { "epoch": 0.09809, "grad_norm": 1.2608877275032968, "learning_rate": 0.003, "loss": 4.0851, "step": 9809 }, { "epoch": 0.0981, "grad_norm": 1.1758889828454, "learning_rate": 0.003, "loss": 4.1096, "step": 9810 }, { "epoch": 0.09811, "grad_norm": 1.0784427152170986, "learning_rate": 0.003, "loss": 4.1081, "step": 9811 }, { "epoch": 0.09812, "grad_norm": 1.2805804980790638, "learning_rate": 0.003, "loss": 4.0839, "step": 9812 }, { "epoch": 0.09813, "grad_norm": 1.224801700658708, "learning_rate": 0.003, "loss": 4.0924, "step": 9813 }, { "epoch": 0.09814, "grad_norm": 1.341835488155258, "learning_rate": 0.003, "loss": 4.0766, "step": 9814 }, { "epoch": 0.09815, "grad_norm": 1.2555265063780896, "learning_rate": 0.003, "loss": 4.045, "step": 9815 }, { "epoch": 0.09816, "grad_norm": 1.1577054447465116, "learning_rate": 0.003, "loss": 4.0504, "step": 9816 }, { "epoch": 0.09817, "grad_norm": 1.2693818580526928, "learning_rate": 0.003, "loss": 4.073, "step": 9817 }, { "epoch": 0.09818, "grad_norm": 1.2054892593089115, "learning_rate": 0.003, "loss": 4.0971, "step": 9818 }, { "epoch": 0.09819, "grad_norm": 1.0065370155399669, "learning_rate": 0.003, "loss": 4.0945, "step": 9819 }, { "epoch": 0.0982, "grad_norm": 1.05278766664134, "learning_rate": 0.003, "loss": 4.0773, "step": 9820 }, { "epoch": 0.09821, "grad_norm": 1.107489019798911, "learning_rate": 0.003, "loss": 4.0709, "step": 9821 }, { "epoch": 0.09822, "grad_norm": 1.3877140403407857, "learning_rate": 0.003, "loss": 4.0957, "step": 9822 }, { "epoch": 0.09823, "grad_norm": 1.0026511211775835, "learning_rate": 0.003, "loss": 4.0996, "step": 9823 }, { "epoch": 0.09824, "grad_norm": 1.033939959190294, "learning_rate": 0.003, "loss": 4.0894, "step": 9824 }, { "epoch": 0.09825, "grad_norm": 1.2849763015452975, "learning_rate": 0.003, "loss": 4.0841, "step": 9825 }, { "epoch": 0.09826, "grad_norm": 1.0313620019572372, "learning_rate": 0.003, "loss": 4.0606, "step": 9826 }, { "epoch": 0.09827, "grad_norm": 1.1878648695251066, "learning_rate": 0.003, "loss": 4.0713, "step": 9827 }, { "epoch": 0.09828, "grad_norm": 1.2119833843882561, "learning_rate": 0.003, "loss": 4.0888, "step": 9828 }, { "epoch": 0.09829, "grad_norm": 1.1477303105835792, "learning_rate": 0.003, "loss": 4.0802, "step": 9829 }, { "epoch": 0.0983, "grad_norm": 1.0635731826395092, "learning_rate": 0.003, "loss": 4.0813, "step": 9830 }, { "epoch": 0.09831, "grad_norm": 1.293170125069096, "learning_rate": 0.003, "loss": 4.0963, "step": 9831 }, { "epoch": 0.09832, "grad_norm": 0.9769068590933736, "learning_rate": 0.003, "loss": 4.0516, "step": 9832 }, { "epoch": 0.09833, "grad_norm": 1.1407951573643962, "learning_rate": 0.003, "loss": 4.1015, "step": 9833 }, { "epoch": 0.09834, "grad_norm": 1.1059375659080322, "learning_rate": 0.003, "loss": 4.089, "step": 9834 }, { "epoch": 0.09835, "grad_norm": 1.168854618660162, "learning_rate": 0.003, "loss": 4.0853, "step": 9835 }, { "epoch": 0.09836, "grad_norm": 0.9730803504426415, "learning_rate": 0.003, "loss": 4.0486, "step": 9836 }, { "epoch": 0.09837, "grad_norm": 1.2347465653452903, "learning_rate": 0.003, "loss": 4.0444, "step": 9837 }, { "epoch": 0.09838, "grad_norm": 1.1476121202950844, "learning_rate": 0.003, "loss": 4.0752, "step": 9838 }, { "epoch": 0.09839, "grad_norm": 1.2696542282974184, "learning_rate": 0.003, "loss": 4.1093, "step": 9839 }, { "epoch": 0.0984, "grad_norm": 1.0724548362610222, "learning_rate": 0.003, "loss": 4.0806, "step": 9840 }, { "epoch": 0.09841, "grad_norm": 1.0152332779536721, "learning_rate": 0.003, "loss": 4.0921, "step": 9841 }, { "epoch": 0.09842, "grad_norm": 1.1771386010629616, "learning_rate": 0.003, "loss": 4.072, "step": 9842 }, { "epoch": 0.09843, "grad_norm": 0.9479731922186927, "learning_rate": 0.003, "loss": 4.0735, "step": 9843 }, { "epoch": 0.09844, "grad_norm": 1.17928222050337, "learning_rate": 0.003, "loss": 4.0588, "step": 9844 }, { "epoch": 0.09845, "grad_norm": 1.0128480063087133, "learning_rate": 0.003, "loss": 4.0728, "step": 9845 }, { "epoch": 0.09846, "grad_norm": 1.332545163467496, "learning_rate": 0.003, "loss": 4.0631, "step": 9846 }, { "epoch": 0.09847, "grad_norm": 1.1738880704546049, "learning_rate": 0.003, "loss": 4.0861, "step": 9847 }, { "epoch": 0.09848, "grad_norm": 1.1376872338122561, "learning_rate": 0.003, "loss": 4.0631, "step": 9848 }, { "epoch": 0.09849, "grad_norm": 1.1281981446522997, "learning_rate": 0.003, "loss": 4.0749, "step": 9849 }, { "epoch": 0.0985, "grad_norm": 1.3198147193649996, "learning_rate": 0.003, "loss": 4.1125, "step": 9850 }, { "epoch": 0.09851, "grad_norm": 0.9562453134354812, "learning_rate": 0.003, "loss": 4.1088, "step": 9851 }, { "epoch": 0.09852, "grad_norm": 1.3740365526158849, "learning_rate": 0.003, "loss": 4.0915, "step": 9852 }, { "epoch": 0.09853, "grad_norm": 1.0095709459654698, "learning_rate": 0.003, "loss": 4.0777, "step": 9853 }, { "epoch": 0.09854, "grad_norm": 1.2468967661465074, "learning_rate": 0.003, "loss": 4.062, "step": 9854 }, { "epoch": 0.09855, "grad_norm": 1.0743143852792683, "learning_rate": 0.003, "loss": 4.0514, "step": 9855 }, { "epoch": 0.09856, "grad_norm": 1.3474930801551417, "learning_rate": 0.003, "loss": 4.1161, "step": 9856 }, { "epoch": 0.09857, "grad_norm": 1.12785581796951, "learning_rate": 0.003, "loss": 4.0822, "step": 9857 }, { "epoch": 0.09858, "grad_norm": 1.0233629542205291, "learning_rate": 0.003, "loss": 4.0749, "step": 9858 }, { "epoch": 0.09859, "grad_norm": 1.3543380171211834, "learning_rate": 0.003, "loss": 4.1244, "step": 9859 }, { "epoch": 0.0986, "grad_norm": 0.9674230733367689, "learning_rate": 0.003, "loss": 4.076, "step": 9860 }, { "epoch": 0.09861, "grad_norm": 1.2444572560734313, "learning_rate": 0.003, "loss": 4.0912, "step": 9861 }, { "epoch": 0.09862, "grad_norm": 1.0081407117477676, "learning_rate": 0.003, "loss": 4.1014, "step": 9862 }, { "epoch": 0.09863, "grad_norm": 1.3463742496274695, "learning_rate": 0.003, "loss": 4.1201, "step": 9863 }, { "epoch": 0.09864, "grad_norm": 1.1319604981703015, "learning_rate": 0.003, "loss": 4.1138, "step": 9864 }, { "epoch": 0.09865, "grad_norm": 1.174713167251643, "learning_rate": 0.003, "loss": 4.0932, "step": 9865 }, { "epoch": 0.09866, "grad_norm": 1.1897800123686184, "learning_rate": 0.003, "loss": 4.0822, "step": 9866 }, { "epoch": 0.09867, "grad_norm": 1.0641636117425668, "learning_rate": 0.003, "loss": 4.0628, "step": 9867 }, { "epoch": 0.09868, "grad_norm": 1.0050044319605966, "learning_rate": 0.003, "loss": 4.07, "step": 9868 }, { "epoch": 0.09869, "grad_norm": 1.2371922187774345, "learning_rate": 0.003, "loss": 4.0755, "step": 9869 }, { "epoch": 0.0987, "grad_norm": 1.0979751060301617, "learning_rate": 0.003, "loss": 4.067, "step": 9870 }, { "epoch": 0.09871, "grad_norm": 1.075197663211, "learning_rate": 0.003, "loss": 4.1416, "step": 9871 }, { "epoch": 0.09872, "grad_norm": 1.3162216075817015, "learning_rate": 0.003, "loss": 4.1034, "step": 9872 }, { "epoch": 0.09873, "grad_norm": 0.9695957927551048, "learning_rate": 0.003, "loss": 4.0835, "step": 9873 }, { "epoch": 0.09874, "grad_norm": 1.141162495266731, "learning_rate": 0.003, "loss": 4.1051, "step": 9874 }, { "epoch": 0.09875, "grad_norm": 1.3599541714097816, "learning_rate": 0.003, "loss": 4.0807, "step": 9875 }, { "epoch": 0.09876, "grad_norm": 1.2469087193309278, "learning_rate": 0.003, "loss": 4.086, "step": 9876 }, { "epoch": 0.09877, "grad_norm": 1.1735176016241307, "learning_rate": 0.003, "loss": 4.0785, "step": 9877 }, { "epoch": 0.09878, "grad_norm": 1.281465889737412, "learning_rate": 0.003, "loss": 4.0673, "step": 9878 }, { "epoch": 0.09879, "grad_norm": 1.041824837320346, "learning_rate": 0.003, "loss": 4.0702, "step": 9879 }, { "epoch": 0.0988, "grad_norm": 1.0776427620189757, "learning_rate": 0.003, "loss": 4.0621, "step": 9880 }, { "epoch": 0.09881, "grad_norm": 0.9551291296156181, "learning_rate": 0.003, "loss": 4.0707, "step": 9881 }, { "epoch": 0.09882, "grad_norm": 1.2407348799554545, "learning_rate": 0.003, "loss": 4.1003, "step": 9882 }, { "epoch": 0.09883, "grad_norm": 1.0724628106467027, "learning_rate": 0.003, "loss": 4.078, "step": 9883 }, { "epoch": 0.09884, "grad_norm": 1.2312626740236658, "learning_rate": 0.003, "loss": 4.0678, "step": 9884 }, { "epoch": 0.09885, "grad_norm": 1.1214546593036556, "learning_rate": 0.003, "loss": 4.0755, "step": 9885 }, { "epoch": 0.09886, "grad_norm": 1.2006637515281076, "learning_rate": 0.003, "loss": 4.065, "step": 9886 }, { "epoch": 0.09887, "grad_norm": 1.0990165349102805, "learning_rate": 0.003, "loss": 4.0992, "step": 9887 }, { "epoch": 0.09888, "grad_norm": 1.2350170155045428, "learning_rate": 0.003, "loss": 4.1174, "step": 9888 }, { "epoch": 0.09889, "grad_norm": 1.362476733084362, "learning_rate": 0.003, "loss": 4.0683, "step": 9889 }, { "epoch": 0.0989, "grad_norm": 0.7819269212449568, "learning_rate": 0.003, "loss": 4.0794, "step": 9890 }, { "epoch": 0.09891, "grad_norm": 1.048931288056113, "learning_rate": 0.003, "loss": 4.0912, "step": 9891 }, { "epoch": 0.09892, "grad_norm": 1.107313711237237, "learning_rate": 0.003, "loss": 4.0538, "step": 9892 }, { "epoch": 0.09893, "grad_norm": 0.966456965078536, "learning_rate": 0.003, "loss": 4.0819, "step": 9893 }, { "epoch": 0.09894, "grad_norm": 1.2879625194596298, "learning_rate": 0.003, "loss": 4.1019, "step": 9894 }, { "epoch": 0.09895, "grad_norm": 1.010388942214523, "learning_rate": 0.003, "loss": 4.0967, "step": 9895 }, { "epoch": 0.09896, "grad_norm": 1.469756639451844, "learning_rate": 0.003, "loss": 4.0737, "step": 9896 }, { "epoch": 0.09897, "grad_norm": 0.9724964169587819, "learning_rate": 0.003, "loss": 4.0806, "step": 9897 }, { "epoch": 0.09898, "grad_norm": 1.110062957321316, "learning_rate": 0.003, "loss": 4.0781, "step": 9898 }, { "epoch": 0.09899, "grad_norm": 1.1012435868630626, "learning_rate": 0.003, "loss": 4.0759, "step": 9899 }, { "epoch": 0.099, "grad_norm": 1.1331555108422586, "learning_rate": 0.003, "loss": 4.1127, "step": 9900 }, { "epoch": 0.09901, "grad_norm": 1.240416809700766, "learning_rate": 0.003, "loss": 4.0988, "step": 9901 }, { "epoch": 0.09902, "grad_norm": 1.2370388742960592, "learning_rate": 0.003, "loss": 4.0961, "step": 9902 }, { "epoch": 0.09903, "grad_norm": 1.1397273835793245, "learning_rate": 0.003, "loss": 4.0812, "step": 9903 }, { "epoch": 0.09904, "grad_norm": 1.0060094292673782, "learning_rate": 0.003, "loss": 4.0648, "step": 9904 }, { "epoch": 0.09905, "grad_norm": 1.3423527405582918, "learning_rate": 0.003, "loss": 4.0915, "step": 9905 }, { "epoch": 0.09906, "grad_norm": 1.1284117962253848, "learning_rate": 0.003, "loss": 4.0993, "step": 9906 }, { "epoch": 0.09907, "grad_norm": 1.3696783895343145, "learning_rate": 0.003, "loss": 4.0552, "step": 9907 }, { "epoch": 0.09908, "grad_norm": 1.2581458797401222, "learning_rate": 0.003, "loss": 4.1027, "step": 9908 }, { "epoch": 0.09909, "grad_norm": 1.1631948089672632, "learning_rate": 0.003, "loss": 4.0968, "step": 9909 }, { "epoch": 0.0991, "grad_norm": 1.044196951382137, "learning_rate": 0.003, "loss": 4.0894, "step": 9910 }, { "epoch": 0.09911, "grad_norm": 1.1236345691959386, "learning_rate": 0.003, "loss": 4.0942, "step": 9911 }, { "epoch": 0.09912, "grad_norm": 1.1346415688903262, "learning_rate": 0.003, "loss": 4.0803, "step": 9912 }, { "epoch": 0.09913, "grad_norm": 1.135605780851906, "learning_rate": 0.003, "loss": 4.1068, "step": 9913 }, { "epoch": 0.09914, "grad_norm": 0.9738857919843333, "learning_rate": 0.003, "loss": 4.0597, "step": 9914 }, { "epoch": 0.09915, "grad_norm": 1.1826516493578803, "learning_rate": 0.003, "loss": 4.076, "step": 9915 }, { "epoch": 0.09916, "grad_norm": 1.0458826743541352, "learning_rate": 0.003, "loss": 4.0967, "step": 9916 }, { "epoch": 0.09917, "grad_norm": 1.2069309328906987, "learning_rate": 0.003, "loss": 4.1002, "step": 9917 }, { "epoch": 0.09918, "grad_norm": 1.093532181480916, "learning_rate": 0.003, "loss": 4.0601, "step": 9918 }, { "epoch": 0.09919, "grad_norm": 1.0556920632213929, "learning_rate": 0.003, "loss": 4.109, "step": 9919 }, { "epoch": 0.0992, "grad_norm": 1.2088706818022557, "learning_rate": 0.003, "loss": 4.0955, "step": 9920 }, { "epoch": 0.09921, "grad_norm": 1.16100001226493, "learning_rate": 0.003, "loss": 4.0421, "step": 9921 }, { "epoch": 0.09922, "grad_norm": 1.3171688806014334, "learning_rate": 0.003, "loss": 4.0866, "step": 9922 }, { "epoch": 0.09923, "grad_norm": 1.1526997426645762, "learning_rate": 0.003, "loss": 4.0918, "step": 9923 }, { "epoch": 0.09924, "grad_norm": 0.9461549921276146, "learning_rate": 0.003, "loss": 4.0469, "step": 9924 }, { "epoch": 0.09925, "grad_norm": 0.9614350868486845, "learning_rate": 0.003, "loss": 4.0523, "step": 9925 }, { "epoch": 0.09926, "grad_norm": 1.2208880766976027, "learning_rate": 0.003, "loss": 4.0837, "step": 9926 }, { "epoch": 0.09927, "grad_norm": 1.0350078279501103, "learning_rate": 0.003, "loss": 4.1119, "step": 9927 }, { "epoch": 0.09928, "grad_norm": 1.290699463729032, "learning_rate": 0.003, "loss": 4.075, "step": 9928 }, { "epoch": 0.09929, "grad_norm": 0.9234887678298359, "learning_rate": 0.003, "loss": 4.0796, "step": 9929 }, { "epoch": 0.0993, "grad_norm": 1.0780902297276813, "learning_rate": 0.003, "loss": 4.0635, "step": 9930 }, { "epoch": 0.09931, "grad_norm": 1.1709209020702502, "learning_rate": 0.003, "loss": 4.0792, "step": 9931 }, { "epoch": 0.09932, "grad_norm": 1.0557018974394423, "learning_rate": 0.003, "loss": 4.0858, "step": 9932 }, { "epoch": 0.09933, "grad_norm": 1.277625174006262, "learning_rate": 0.003, "loss": 4.0885, "step": 9933 }, { "epoch": 0.09934, "grad_norm": 1.1428722092685848, "learning_rate": 0.003, "loss": 4.0907, "step": 9934 }, { "epoch": 0.09935, "grad_norm": 1.3283362969267656, "learning_rate": 0.003, "loss": 4.0688, "step": 9935 }, { "epoch": 0.09936, "grad_norm": 1.2116884651385684, "learning_rate": 0.003, "loss": 4.0848, "step": 9936 }, { "epoch": 0.09937, "grad_norm": 1.1129728817050744, "learning_rate": 0.003, "loss": 4.0509, "step": 9937 }, { "epoch": 0.09938, "grad_norm": 1.076087084617254, "learning_rate": 0.003, "loss": 4.0905, "step": 9938 }, { "epoch": 0.09939, "grad_norm": 1.1984980026743222, "learning_rate": 0.003, "loss": 4.0896, "step": 9939 }, { "epoch": 0.0994, "grad_norm": 1.0631816350352683, "learning_rate": 0.003, "loss": 4.1072, "step": 9940 }, { "epoch": 0.09941, "grad_norm": 1.0349558778639734, "learning_rate": 0.003, "loss": 4.0741, "step": 9941 }, { "epoch": 0.09942, "grad_norm": 1.3832430713974921, "learning_rate": 0.003, "loss": 4.0873, "step": 9942 }, { "epoch": 0.09943, "grad_norm": 1.2127412232640076, "learning_rate": 0.003, "loss": 4.0303, "step": 9943 }, { "epoch": 0.09944, "grad_norm": 1.1014547273341169, "learning_rate": 0.003, "loss": 4.1101, "step": 9944 }, { "epoch": 0.09945, "grad_norm": 1.1126941512319144, "learning_rate": 0.003, "loss": 4.0467, "step": 9945 }, { "epoch": 0.09946, "grad_norm": 1.143588221605322, "learning_rate": 0.003, "loss": 4.1016, "step": 9946 }, { "epoch": 0.09947, "grad_norm": 1.1319703116283337, "learning_rate": 0.003, "loss": 4.0881, "step": 9947 }, { "epoch": 0.09948, "grad_norm": 1.184782613822203, "learning_rate": 0.003, "loss": 4.0922, "step": 9948 }, { "epoch": 0.09949, "grad_norm": 1.2660838975989452, "learning_rate": 0.003, "loss": 4.0939, "step": 9949 }, { "epoch": 0.0995, "grad_norm": 1.0530905713274332, "learning_rate": 0.003, "loss": 4.0811, "step": 9950 }, { "epoch": 0.09951, "grad_norm": 1.2458577796529136, "learning_rate": 0.003, "loss": 4.0604, "step": 9951 }, { "epoch": 0.09952, "grad_norm": 1.086892599123822, "learning_rate": 0.003, "loss": 4.0495, "step": 9952 }, { "epoch": 0.09953, "grad_norm": 1.2738241159815291, "learning_rate": 0.003, "loss": 4.0987, "step": 9953 }, { "epoch": 0.09954, "grad_norm": 1.1805167886930188, "learning_rate": 0.003, "loss": 4.0563, "step": 9954 }, { "epoch": 0.09955, "grad_norm": 1.159573083100745, "learning_rate": 0.003, "loss": 4.0609, "step": 9955 }, { "epoch": 0.09956, "grad_norm": 1.03695422736725, "learning_rate": 0.003, "loss": 4.0666, "step": 9956 }, { "epoch": 0.09957, "grad_norm": 1.2902593078263713, "learning_rate": 0.003, "loss": 4.0835, "step": 9957 }, { "epoch": 0.09958, "grad_norm": 1.0541302884915535, "learning_rate": 0.003, "loss": 4.0938, "step": 9958 }, { "epoch": 0.09959, "grad_norm": 1.2123162248067152, "learning_rate": 0.003, "loss": 4.1, "step": 9959 }, { "epoch": 0.0996, "grad_norm": 0.8893684589311062, "learning_rate": 0.003, "loss": 4.0788, "step": 9960 }, { "epoch": 0.09961, "grad_norm": 1.0680186135141085, "learning_rate": 0.003, "loss": 4.0863, "step": 9961 }, { "epoch": 0.09962, "grad_norm": 1.0769231089037596, "learning_rate": 0.003, "loss": 4.095, "step": 9962 }, { "epoch": 0.09963, "grad_norm": 1.157455437174743, "learning_rate": 0.003, "loss": 4.0944, "step": 9963 }, { "epoch": 0.09964, "grad_norm": 1.167599421268358, "learning_rate": 0.003, "loss": 4.0638, "step": 9964 }, { "epoch": 0.09965, "grad_norm": 1.0009527785807748, "learning_rate": 0.003, "loss": 4.0699, "step": 9965 }, { "epoch": 0.09966, "grad_norm": 1.1837576069622928, "learning_rate": 0.003, "loss": 4.0956, "step": 9966 }, { "epoch": 0.09967, "grad_norm": 1.3602843120067374, "learning_rate": 0.003, "loss": 4.0914, "step": 9967 }, { "epoch": 0.09968, "grad_norm": 1.2627589377158193, "learning_rate": 0.003, "loss": 4.0806, "step": 9968 }, { "epoch": 0.09969, "grad_norm": 1.0719830176194494, "learning_rate": 0.003, "loss": 4.0798, "step": 9969 }, { "epoch": 0.0997, "grad_norm": 1.0574384689662466, "learning_rate": 0.003, "loss": 4.0712, "step": 9970 }, { "epoch": 0.09971, "grad_norm": 1.229985803627891, "learning_rate": 0.003, "loss": 4.0826, "step": 9971 }, { "epoch": 0.09972, "grad_norm": 1.022626878926486, "learning_rate": 0.003, "loss": 4.0669, "step": 9972 }, { "epoch": 0.09973, "grad_norm": 1.3166393251005053, "learning_rate": 0.003, "loss": 4.084, "step": 9973 }, { "epoch": 0.09974, "grad_norm": 0.9811358575164996, "learning_rate": 0.003, "loss": 4.0759, "step": 9974 }, { "epoch": 0.09975, "grad_norm": 1.3113017482993325, "learning_rate": 0.003, "loss": 4.0941, "step": 9975 }, { "epoch": 0.09976, "grad_norm": 0.9954690018023183, "learning_rate": 0.003, "loss": 4.0621, "step": 9976 }, { "epoch": 0.09977, "grad_norm": 1.3210366296189624, "learning_rate": 0.003, "loss": 4.0623, "step": 9977 }, { "epoch": 0.09978, "grad_norm": 1.006727568887328, "learning_rate": 0.003, "loss": 4.0676, "step": 9978 }, { "epoch": 0.09979, "grad_norm": 1.4046031646009436, "learning_rate": 0.003, "loss": 4.0908, "step": 9979 }, { "epoch": 0.0998, "grad_norm": 1.1228742990032854, "learning_rate": 0.003, "loss": 4.0825, "step": 9980 }, { "epoch": 0.09981, "grad_norm": 1.0350610981605155, "learning_rate": 0.003, "loss": 4.0836, "step": 9981 }, { "epoch": 0.09982, "grad_norm": 1.1337520050085472, "learning_rate": 0.003, "loss": 4.092, "step": 9982 }, { "epoch": 0.09983, "grad_norm": 1.21368108908721, "learning_rate": 0.003, "loss": 4.0803, "step": 9983 }, { "epoch": 0.09984, "grad_norm": 1.0936050467548077, "learning_rate": 0.003, "loss": 4.1007, "step": 9984 }, { "epoch": 0.09985, "grad_norm": 1.3692702486619674, "learning_rate": 0.003, "loss": 4.0981, "step": 9985 }, { "epoch": 0.09986, "grad_norm": 1.0255191220694087, "learning_rate": 0.003, "loss": 4.0985, "step": 9986 }, { "epoch": 0.09987, "grad_norm": 1.283385904792894, "learning_rate": 0.003, "loss": 4.0716, "step": 9987 }, { "epoch": 0.09988, "grad_norm": 0.9973600391953245, "learning_rate": 0.003, "loss": 4.1066, "step": 9988 }, { "epoch": 0.09989, "grad_norm": 1.3224856674170413, "learning_rate": 0.003, "loss": 4.0804, "step": 9989 }, { "epoch": 0.0999, "grad_norm": 1.0342040264635208, "learning_rate": 0.003, "loss": 4.0835, "step": 9990 }, { "epoch": 0.09991, "grad_norm": 1.2010872038379294, "learning_rate": 0.003, "loss": 4.1087, "step": 9991 }, { "epoch": 0.09992, "grad_norm": 1.1028513651017675, "learning_rate": 0.003, "loss": 4.0631, "step": 9992 }, { "epoch": 0.09993, "grad_norm": 1.2704048114109967, "learning_rate": 0.003, "loss": 4.0837, "step": 9993 }, { "epoch": 0.09994, "grad_norm": 1.1812143976118226, "learning_rate": 0.003, "loss": 4.1011, "step": 9994 }, { "epoch": 0.09995, "grad_norm": 1.412624941712437, "learning_rate": 0.003, "loss": 4.1039, "step": 9995 }, { "epoch": 0.09996, "grad_norm": 1.1199866494266424, "learning_rate": 0.003, "loss": 4.0784, "step": 9996 }, { "epoch": 0.09997, "grad_norm": 1.108151659835287, "learning_rate": 0.003, "loss": 4.0842, "step": 9997 }, { "epoch": 0.09998, "grad_norm": 1.0462352704362803, "learning_rate": 0.003, "loss": 4.0903, "step": 9998 }, { "epoch": 0.09999, "grad_norm": 1.5166747268978646, "learning_rate": 0.003, "loss": 4.0941, "step": 9999 }, { "epoch": 0.1, "grad_norm": 1.0728207634540272, "learning_rate": 0.003, "loss": 4.0922, "step": 10000 }, { "epoch": 0.10001, "grad_norm": 1.1674018884204582, "learning_rate": 0.003, "loss": 4.0764, "step": 10001 }, { "epoch": 0.10002, "grad_norm": 1.0527973953758787, "learning_rate": 0.003, "loss": 4.1026, "step": 10002 }, { "epoch": 0.10003, "grad_norm": 1.2928164974147542, "learning_rate": 0.003, "loss": 4.0816, "step": 10003 }, { "epoch": 0.10004, "grad_norm": 0.8969958806342154, "learning_rate": 0.003, "loss": 4.0667, "step": 10004 }, { "epoch": 0.10005, "grad_norm": 1.1574074627438358, "learning_rate": 0.003, "loss": 4.0824, "step": 10005 }, { "epoch": 0.10006, "grad_norm": 1.3771495047557027, "learning_rate": 0.003, "loss": 4.0909, "step": 10006 }, { "epoch": 0.10007, "grad_norm": 1.167333352905923, "learning_rate": 0.003, "loss": 4.0635, "step": 10007 }, { "epoch": 0.10008, "grad_norm": 1.2023273467074431, "learning_rate": 0.003, "loss": 4.0975, "step": 10008 }, { "epoch": 0.10009, "grad_norm": 1.074482670712341, "learning_rate": 0.003, "loss": 4.0776, "step": 10009 }, { "epoch": 0.1001, "grad_norm": 1.211996301608577, "learning_rate": 0.003, "loss": 4.1003, "step": 10010 }, { "epoch": 0.10011, "grad_norm": 0.9009540566883109, "learning_rate": 0.003, "loss": 4.0761, "step": 10011 }, { "epoch": 0.10012, "grad_norm": 1.0564974927278068, "learning_rate": 0.003, "loss": 4.0795, "step": 10012 }, { "epoch": 0.10013, "grad_norm": 1.3350438104824471, "learning_rate": 0.003, "loss": 4.1063, "step": 10013 }, { "epoch": 0.10014, "grad_norm": 1.0822425068020562, "learning_rate": 0.003, "loss": 4.0919, "step": 10014 }, { "epoch": 0.10015, "grad_norm": 1.019372729465579, "learning_rate": 0.003, "loss": 4.1089, "step": 10015 }, { "epoch": 0.10016, "grad_norm": 1.1682122132964539, "learning_rate": 0.003, "loss": 4.0754, "step": 10016 }, { "epoch": 0.10017, "grad_norm": 1.0636692414598121, "learning_rate": 0.003, "loss": 4.0933, "step": 10017 }, { "epoch": 0.10018, "grad_norm": 1.153668222529384, "learning_rate": 0.003, "loss": 4.0963, "step": 10018 }, { "epoch": 0.10019, "grad_norm": 1.1863489301505494, "learning_rate": 0.003, "loss": 4.0929, "step": 10019 }, { "epoch": 0.1002, "grad_norm": 1.1426419458719907, "learning_rate": 0.003, "loss": 4.0589, "step": 10020 }, { "epoch": 0.10021, "grad_norm": 1.2306087761188012, "learning_rate": 0.003, "loss": 4.0989, "step": 10021 }, { "epoch": 0.10022, "grad_norm": 1.0297341287287447, "learning_rate": 0.003, "loss": 4.0791, "step": 10022 }, { "epoch": 0.10023, "grad_norm": 1.2320108667621523, "learning_rate": 0.003, "loss": 4.0933, "step": 10023 }, { "epoch": 0.10024, "grad_norm": 1.0916466193494474, "learning_rate": 0.003, "loss": 4.1001, "step": 10024 }, { "epoch": 0.10025, "grad_norm": 1.1464372389238344, "learning_rate": 0.003, "loss": 4.0897, "step": 10025 }, { "epoch": 0.10026, "grad_norm": 1.2105092192886502, "learning_rate": 0.003, "loss": 4.0704, "step": 10026 }, { "epoch": 0.10027, "grad_norm": 1.0593198000420767, "learning_rate": 0.003, "loss": 4.0616, "step": 10027 }, { "epoch": 0.10028, "grad_norm": 1.2124348310175164, "learning_rate": 0.003, "loss": 4.1178, "step": 10028 }, { "epoch": 0.10029, "grad_norm": 1.3449665326048608, "learning_rate": 0.003, "loss": 4.0836, "step": 10029 }, { "epoch": 0.1003, "grad_norm": 1.13252213762222, "learning_rate": 0.003, "loss": 4.089, "step": 10030 }, { "epoch": 0.10031, "grad_norm": 1.208957749908409, "learning_rate": 0.003, "loss": 4.1191, "step": 10031 }, { "epoch": 0.10032, "grad_norm": 1.0843050238714709, "learning_rate": 0.003, "loss": 4.072, "step": 10032 }, { "epoch": 0.10033, "grad_norm": 1.1118594428514184, "learning_rate": 0.003, "loss": 4.0767, "step": 10033 }, { "epoch": 0.10034, "grad_norm": 1.1454301997431835, "learning_rate": 0.003, "loss": 4.0687, "step": 10034 }, { "epoch": 0.10035, "grad_norm": 1.2373694067703571, "learning_rate": 0.003, "loss": 4.0819, "step": 10035 }, { "epoch": 0.10036, "grad_norm": 1.234927886756976, "learning_rate": 0.003, "loss": 4.053, "step": 10036 }, { "epoch": 0.10037, "grad_norm": 1.183130686021368, "learning_rate": 0.003, "loss": 4.0713, "step": 10037 }, { "epoch": 0.10038, "grad_norm": 1.1581871376157453, "learning_rate": 0.003, "loss": 4.1012, "step": 10038 }, { "epoch": 0.10039, "grad_norm": 1.0805442201078972, "learning_rate": 0.003, "loss": 4.0766, "step": 10039 }, { "epoch": 0.1004, "grad_norm": 1.0644531472003973, "learning_rate": 0.003, "loss": 4.0795, "step": 10040 }, { "epoch": 0.10041, "grad_norm": 1.1689125902658166, "learning_rate": 0.003, "loss": 4.0923, "step": 10041 }, { "epoch": 0.10042, "grad_norm": 0.9131143869964985, "learning_rate": 0.003, "loss": 4.1087, "step": 10042 }, { "epoch": 0.10043, "grad_norm": 1.1125978766470848, "learning_rate": 0.003, "loss": 4.109, "step": 10043 }, { "epoch": 0.10044, "grad_norm": 1.1554321741109186, "learning_rate": 0.003, "loss": 4.0974, "step": 10044 }, { "epoch": 0.10045, "grad_norm": 0.9476401166924924, "learning_rate": 0.003, "loss": 4.0651, "step": 10045 }, { "epoch": 0.10046, "grad_norm": 1.472333260084941, "learning_rate": 0.003, "loss": 4.1057, "step": 10046 }, { "epoch": 0.10047, "grad_norm": 0.9478781520731763, "learning_rate": 0.003, "loss": 4.0867, "step": 10047 }, { "epoch": 0.10048, "grad_norm": 1.2498078049575723, "learning_rate": 0.003, "loss": 4.0941, "step": 10048 }, { "epoch": 0.10049, "grad_norm": 1.1847876378870514, "learning_rate": 0.003, "loss": 4.0628, "step": 10049 }, { "epoch": 0.1005, "grad_norm": 1.4042126593877933, "learning_rate": 0.003, "loss": 4.0964, "step": 10050 }, { "epoch": 0.10051, "grad_norm": 1.1738786717009158, "learning_rate": 0.003, "loss": 4.0854, "step": 10051 }, { "epoch": 0.10052, "grad_norm": 1.0152151745073639, "learning_rate": 0.003, "loss": 4.0726, "step": 10052 }, { "epoch": 0.10053, "grad_norm": 1.158818668265983, "learning_rate": 0.003, "loss": 4.0677, "step": 10053 }, { "epoch": 0.10054, "grad_norm": 1.3629148640865112, "learning_rate": 0.003, "loss": 4.102, "step": 10054 }, { "epoch": 0.10055, "grad_norm": 0.9977236962370847, "learning_rate": 0.003, "loss": 4.099, "step": 10055 }, { "epoch": 0.10056, "grad_norm": 1.3482816658394208, "learning_rate": 0.003, "loss": 4.116, "step": 10056 }, { "epoch": 0.10057, "grad_norm": 0.9539265856157081, "learning_rate": 0.003, "loss": 4.095, "step": 10057 }, { "epoch": 0.10058, "grad_norm": 1.1076892954413688, "learning_rate": 0.003, "loss": 4.0933, "step": 10058 }, { "epoch": 0.10059, "grad_norm": 1.0197333845789176, "learning_rate": 0.003, "loss": 4.0727, "step": 10059 }, { "epoch": 0.1006, "grad_norm": 1.2717378385736515, "learning_rate": 0.003, "loss": 4.0957, "step": 10060 }, { "epoch": 0.10061, "grad_norm": 1.1525924759121828, "learning_rate": 0.003, "loss": 4.0704, "step": 10061 }, { "epoch": 0.10062, "grad_norm": 1.1978969573423361, "learning_rate": 0.003, "loss": 4.066, "step": 10062 }, { "epoch": 0.10063, "grad_norm": 1.1840545782953635, "learning_rate": 0.003, "loss": 4.0706, "step": 10063 }, { "epoch": 0.10064, "grad_norm": 1.0449591042434434, "learning_rate": 0.003, "loss": 4.0798, "step": 10064 }, { "epoch": 0.10065, "grad_norm": 1.212776165261585, "learning_rate": 0.003, "loss": 4.1055, "step": 10065 }, { "epoch": 0.10066, "grad_norm": 1.2286144576629734, "learning_rate": 0.003, "loss": 4.0812, "step": 10066 }, { "epoch": 0.10067, "grad_norm": 1.1815206209397082, "learning_rate": 0.003, "loss": 4.0932, "step": 10067 }, { "epoch": 0.10068, "grad_norm": 1.089605118069462, "learning_rate": 0.003, "loss": 4.0829, "step": 10068 }, { "epoch": 0.10069, "grad_norm": 1.2751774093739108, "learning_rate": 0.003, "loss": 4.0759, "step": 10069 }, { "epoch": 0.1007, "grad_norm": 1.0518315664497935, "learning_rate": 0.003, "loss": 4.0634, "step": 10070 }, { "epoch": 0.10071, "grad_norm": 1.0075982956011604, "learning_rate": 0.003, "loss": 4.0758, "step": 10071 }, { "epoch": 0.10072, "grad_norm": 0.9328071534298303, "learning_rate": 0.003, "loss": 4.075, "step": 10072 }, { "epoch": 0.10073, "grad_norm": 1.0491017217453746, "learning_rate": 0.003, "loss": 4.0908, "step": 10073 }, { "epoch": 0.10074, "grad_norm": 1.1418471177068579, "learning_rate": 0.003, "loss": 4.063, "step": 10074 }, { "epoch": 0.10075, "grad_norm": 0.9615639895687752, "learning_rate": 0.003, "loss": 4.1053, "step": 10075 }, { "epoch": 0.10076, "grad_norm": 1.186944099321991, "learning_rate": 0.003, "loss": 4.0925, "step": 10076 }, { "epoch": 0.10077, "grad_norm": 1.1045264647874289, "learning_rate": 0.003, "loss": 4.1057, "step": 10077 }, { "epoch": 0.10078, "grad_norm": 1.3580204083886844, "learning_rate": 0.003, "loss": 4.0634, "step": 10078 }, { "epoch": 0.10079, "grad_norm": 1.119001442915295, "learning_rate": 0.003, "loss": 4.0821, "step": 10079 }, { "epoch": 0.1008, "grad_norm": 1.6069412879920508, "learning_rate": 0.003, "loss": 4.0855, "step": 10080 }, { "epoch": 0.10081, "grad_norm": 1.0604760790650596, "learning_rate": 0.003, "loss": 4.1064, "step": 10081 }, { "epoch": 0.10082, "grad_norm": 1.1115804845269284, "learning_rate": 0.003, "loss": 4.0816, "step": 10082 }, { "epoch": 0.10083, "grad_norm": 1.238741868903038, "learning_rate": 0.003, "loss": 4.0986, "step": 10083 }, { "epoch": 0.10084, "grad_norm": 1.2314664198702576, "learning_rate": 0.003, "loss": 4.0573, "step": 10084 }, { "epoch": 0.10085, "grad_norm": 1.1833832812757277, "learning_rate": 0.003, "loss": 4.0728, "step": 10085 }, { "epoch": 0.10086, "grad_norm": 1.0137645005741707, "learning_rate": 0.003, "loss": 4.0861, "step": 10086 }, { "epoch": 0.10087, "grad_norm": 1.231342789321368, "learning_rate": 0.003, "loss": 4.1073, "step": 10087 }, { "epoch": 0.10088, "grad_norm": 0.9848663896765558, "learning_rate": 0.003, "loss": 4.077, "step": 10088 }, { "epoch": 0.10089, "grad_norm": 1.2766688039526612, "learning_rate": 0.003, "loss": 4.0919, "step": 10089 }, { "epoch": 0.1009, "grad_norm": 1.0665591732242785, "learning_rate": 0.003, "loss": 4.1106, "step": 10090 }, { "epoch": 0.10091, "grad_norm": 1.3844890839324862, "learning_rate": 0.003, "loss": 4.0832, "step": 10091 }, { "epoch": 0.10092, "grad_norm": 0.9703410910939055, "learning_rate": 0.003, "loss": 4.0942, "step": 10092 }, { "epoch": 0.10093, "grad_norm": 1.1081454322650999, "learning_rate": 0.003, "loss": 4.0839, "step": 10093 }, { "epoch": 0.10094, "grad_norm": 1.2702596843147151, "learning_rate": 0.003, "loss": 4.0746, "step": 10094 }, { "epoch": 0.10095, "grad_norm": 1.3241180876871614, "learning_rate": 0.003, "loss": 4.0663, "step": 10095 }, { "epoch": 0.10096, "grad_norm": 1.063782838125781, "learning_rate": 0.003, "loss": 4.0818, "step": 10096 }, { "epoch": 0.10097, "grad_norm": 1.173590221046823, "learning_rate": 0.003, "loss": 4.0776, "step": 10097 }, { "epoch": 0.10098, "grad_norm": 1.1388690036673108, "learning_rate": 0.003, "loss": 4.0751, "step": 10098 }, { "epoch": 0.10099, "grad_norm": 1.016521589387514, "learning_rate": 0.003, "loss": 4.0718, "step": 10099 }, { "epoch": 0.101, "grad_norm": 1.124890576751929, "learning_rate": 0.003, "loss": 4.0902, "step": 10100 }, { "epoch": 0.10101, "grad_norm": 1.0702676449322102, "learning_rate": 0.003, "loss": 4.0714, "step": 10101 }, { "epoch": 0.10102, "grad_norm": 1.0594937403569906, "learning_rate": 0.003, "loss": 4.0952, "step": 10102 }, { "epoch": 0.10103, "grad_norm": 1.1189786533695918, "learning_rate": 0.003, "loss": 4.0557, "step": 10103 }, { "epoch": 0.10104, "grad_norm": 1.1621062267666284, "learning_rate": 0.003, "loss": 4.0872, "step": 10104 }, { "epoch": 0.10105, "grad_norm": 1.1586349551608832, "learning_rate": 0.003, "loss": 4.0791, "step": 10105 }, { "epoch": 0.10106, "grad_norm": 1.2449125737307989, "learning_rate": 0.003, "loss": 4.0879, "step": 10106 }, { "epoch": 0.10107, "grad_norm": 1.0103660592944312, "learning_rate": 0.003, "loss": 4.0816, "step": 10107 }, { "epoch": 0.10108, "grad_norm": 1.2430486420035323, "learning_rate": 0.003, "loss": 4.0993, "step": 10108 }, { "epoch": 0.10109, "grad_norm": 1.060115161096371, "learning_rate": 0.003, "loss": 4.0623, "step": 10109 }, { "epoch": 0.1011, "grad_norm": 1.0405717245223134, "learning_rate": 0.003, "loss": 4.0979, "step": 10110 }, { "epoch": 0.10111, "grad_norm": 1.166298936720804, "learning_rate": 0.003, "loss": 4.0966, "step": 10111 }, { "epoch": 0.10112, "grad_norm": 1.2285570938753498, "learning_rate": 0.003, "loss": 4.0735, "step": 10112 }, { "epoch": 0.10113, "grad_norm": 1.014117471053155, "learning_rate": 0.003, "loss": 4.0489, "step": 10113 }, { "epoch": 0.10114, "grad_norm": 1.1714584848865766, "learning_rate": 0.003, "loss": 4.0539, "step": 10114 }, { "epoch": 0.10115, "grad_norm": 1.171191771839053, "learning_rate": 0.003, "loss": 4.0816, "step": 10115 }, { "epoch": 0.10116, "grad_norm": 1.1268191412711248, "learning_rate": 0.003, "loss": 4.0743, "step": 10116 }, { "epoch": 0.10117, "grad_norm": 1.056060456963882, "learning_rate": 0.003, "loss": 4.0649, "step": 10117 }, { "epoch": 0.10118, "grad_norm": 1.14423470231977, "learning_rate": 0.003, "loss": 4.1048, "step": 10118 }, { "epoch": 0.10119, "grad_norm": 1.2978002036951515, "learning_rate": 0.003, "loss": 4.0729, "step": 10119 }, { "epoch": 0.1012, "grad_norm": 0.9773861726106657, "learning_rate": 0.003, "loss": 4.0673, "step": 10120 }, { "epoch": 0.10121, "grad_norm": 1.2298033697165904, "learning_rate": 0.003, "loss": 4.0932, "step": 10121 }, { "epoch": 0.10122, "grad_norm": 1.2027309543987579, "learning_rate": 0.003, "loss": 4.0581, "step": 10122 }, { "epoch": 0.10123, "grad_norm": 1.0867591534868284, "learning_rate": 0.003, "loss": 4.0719, "step": 10123 }, { "epoch": 0.10124, "grad_norm": 1.2447200356116115, "learning_rate": 0.003, "loss": 4.1042, "step": 10124 }, { "epoch": 0.10125, "grad_norm": 1.0661379327703835, "learning_rate": 0.003, "loss": 4.09, "step": 10125 }, { "epoch": 0.10126, "grad_norm": 1.1293357584285106, "learning_rate": 0.003, "loss": 4.0702, "step": 10126 }, { "epoch": 0.10127, "grad_norm": 1.0276292424876197, "learning_rate": 0.003, "loss": 4.0954, "step": 10127 }, { "epoch": 0.10128, "grad_norm": 1.3109229855803073, "learning_rate": 0.003, "loss": 4.0977, "step": 10128 }, { "epoch": 0.10129, "grad_norm": 1.175979220904674, "learning_rate": 0.003, "loss": 4.0988, "step": 10129 }, { "epoch": 0.1013, "grad_norm": 1.046008366778478, "learning_rate": 0.003, "loss": 4.0508, "step": 10130 }, { "epoch": 0.10131, "grad_norm": 1.2889435751549339, "learning_rate": 0.003, "loss": 4.0768, "step": 10131 }, { "epoch": 0.10132, "grad_norm": 1.3251036219526253, "learning_rate": 0.003, "loss": 4.0927, "step": 10132 }, { "epoch": 0.10133, "grad_norm": 1.1995187882520537, "learning_rate": 0.003, "loss": 4.0709, "step": 10133 }, { "epoch": 0.10134, "grad_norm": 0.9472900131078198, "learning_rate": 0.003, "loss": 4.0639, "step": 10134 }, { "epoch": 0.10135, "grad_norm": 1.1108816955880372, "learning_rate": 0.003, "loss": 4.082, "step": 10135 }, { "epoch": 0.10136, "grad_norm": 1.078636334857174, "learning_rate": 0.003, "loss": 4.0608, "step": 10136 }, { "epoch": 0.10137, "grad_norm": 1.3105555106578144, "learning_rate": 0.003, "loss": 4.0722, "step": 10137 }, { "epoch": 0.10138, "grad_norm": 1.1199739555288444, "learning_rate": 0.003, "loss": 4.0844, "step": 10138 }, { "epoch": 0.10139, "grad_norm": 1.3912021727977202, "learning_rate": 0.003, "loss": 4.0504, "step": 10139 }, { "epoch": 0.1014, "grad_norm": 1.0491432390341304, "learning_rate": 0.003, "loss": 4.1085, "step": 10140 }, { "epoch": 0.10141, "grad_norm": 1.1476014585804746, "learning_rate": 0.003, "loss": 4.0712, "step": 10141 }, { "epoch": 0.10142, "grad_norm": 1.059954567327284, "learning_rate": 0.003, "loss": 4.0661, "step": 10142 }, { "epoch": 0.10143, "grad_norm": 1.418874522825989, "learning_rate": 0.003, "loss": 4.0612, "step": 10143 }, { "epoch": 0.10144, "grad_norm": 1.0628985240296076, "learning_rate": 0.003, "loss": 4.0899, "step": 10144 }, { "epoch": 0.10145, "grad_norm": 1.2768614034132377, "learning_rate": 0.003, "loss": 4.0801, "step": 10145 }, { "epoch": 0.10146, "grad_norm": 0.8767978842753446, "learning_rate": 0.003, "loss": 4.0537, "step": 10146 }, { "epoch": 0.10147, "grad_norm": 0.9766014836284787, "learning_rate": 0.003, "loss": 4.0804, "step": 10147 }, { "epoch": 0.10148, "grad_norm": 1.2106871464655664, "learning_rate": 0.003, "loss": 4.0881, "step": 10148 }, { "epoch": 0.10149, "grad_norm": 0.888207092902911, "learning_rate": 0.003, "loss": 4.0424, "step": 10149 }, { "epoch": 0.1015, "grad_norm": 1.2922590715806803, "learning_rate": 0.003, "loss": 4.074, "step": 10150 }, { "epoch": 0.10151, "grad_norm": 1.237040116341225, "learning_rate": 0.003, "loss": 4.0867, "step": 10151 }, { "epoch": 0.10152, "grad_norm": 1.0625251992129978, "learning_rate": 0.003, "loss": 4.08, "step": 10152 }, { "epoch": 0.10153, "grad_norm": 1.075075130630272, "learning_rate": 0.003, "loss": 4.0866, "step": 10153 }, { "epoch": 0.10154, "grad_norm": 1.1080444014221436, "learning_rate": 0.003, "loss": 4.0915, "step": 10154 }, { "epoch": 0.10155, "grad_norm": 1.2551412407034876, "learning_rate": 0.003, "loss": 4.0911, "step": 10155 }, { "epoch": 0.10156, "grad_norm": 1.0021536921323417, "learning_rate": 0.003, "loss": 4.0615, "step": 10156 }, { "epoch": 0.10157, "grad_norm": 1.175049036818617, "learning_rate": 0.003, "loss": 4.0673, "step": 10157 }, { "epoch": 0.10158, "grad_norm": 1.0293363891023952, "learning_rate": 0.003, "loss": 4.0835, "step": 10158 }, { "epoch": 0.10159, "grad_norm": 1.0548857368096765, "learning_rate": 0.003, "loss": 4.0607, "step": 10159 }, { "epoch": 0.1016, "grad_norm": 1.329175614790365, "learning_rate": 0.003, "loss": 4.0978, "step": 10160 }, { "epoch": 0.10161, "grad_norm": 0.9488327406094123, "learning_rate": 0.003, "loss": 4.0672, "step": 10161 }, { "epoch": 0.10162, "grad_norm": 1.4323787828675487, "learning_rate": 0.003, "loss": 4.0946, "step": 10162 }, { "epoch": 0.10163, "grad_norm": 1.1936836804121034, "learning_rate": 0.003, "loss": 4.0872, "step": 10163 }, { "epoch": 0.10164, "grad_norm": 1.406321899502577, "learning_rate": 0.003, "loss": 4.1087, "step": 10164 }, { "epoch": 0.10165, "grad_norm": 1.2261032646604577, "learning_rate": 0.003, "loss": 4.0776, "step": 10165 }, { "epoch": 0.10166, "grad_norm": 0.9461556092810096, "learning_rate": 0.003, "loss": 4.092, "step": 10166 }, { "epoch": 0.10167, "grad_norm": 1.1845316724357464, "learning_rate": 0.003, "loss": 4.0881, "step": 10167 }, { "epoch": 0.10168, "grad_norm": 0.975484573387496, "learning_rate": 0.003, "loss": 4.0531, "step": 10168 }, { "epoch": 0.10169, "grad_norm": 1.283116632412761, "learning_rate": 0.003, "loss": 4.0872, "step": 10169 }, { "epoch": 0.1017, "grad_norm": 0.95631152337951, "learning_rate": 0.003, "loss": 4.0847, "step": 10170 }, { "epoch": 0.10171, "grad_norm": 1.1139812130735123, "learning_rate": 0.003, "loss": 4.0648, "step": 10171 }, { "epoch": 0.10172, "grad_norm": 1.2961802093469639, "learning_rate": 0.003, "loss": 4.0965, "step": 10172 }, { "epoch": 0.10173, "grad_norm": 1.1726631483761973, "learning_rate": 0.003, "loss": 4.0907, "step": 10173 }, { "epoch": 0.10174, "grad_norm": 1.2122860250041871, "learning_rate": 0.003, "loss": 4.0715, "step": 10174 }, { "epoch": 0.10175, "grad_norm": 1.050830198920568, "learning_rate": 0.003, "loss": 4.0694, "step": 10175 }, { "epoch": 0.10176, "grad_norm": 1.3590915054220583, "learning_rate": 0.003, "loss": 4.0854, "step": 10176 }, { "epoch": 0.10177, "grad_norm": 1.0726629179146294, "learning_rate": 0.003, "loss": 4.0707, "step": 10177 }, { "epoch": 0.10178, "grad_norm": 1.2545949381444286, "learning_rate": 0.003, "loss": 4.0909, "step": 10178 }, { "epoch": 0.10179, "grad_norm": 1.0047716022040467, "learning_rate": 0.003, "loss": 4.0853, "step": 10179 }, { "epoch": 0.1018, "grad_norm": 1.3165911693506096, "learning_rate": 0.003, "loss": 4.0827, "step": 10180 }, { "epoch": 0.10181, "grad_norm": 1.1402632334004243, "learning_rate": 0.003, "loss": 4.0818, "step": 10181 }, { "epoch": 0.10182, "grad_norm": 1.1578192434246495, "learning_rate": 0.003, "loss": 4.0923, "step": 10182 }, { "epoch": 0.10183, "grad_norm": 1.141684268742947, "learning_rate": 0.003, "loss": 4.0982, "step": 10183 }, { "epoch": 0.10184, "grad_norm": 1.1923829372124475, "learning_rate": 0.003, "loss": 4.1041, "step": 10184 }, { "epoch": 0.10185, "grad_norm": 1.0715432709277495, "learning_rate": 0.003, "loss": 4.0625, "step": 10185 }, { "epoch": 0.10186, "grad_norm": 1.2311029741984303, "learning_rate": 0.003, "loss": 4.0969, "step": 10186 }, { "epoch": 0.10187, "grad_norm": 1.0152401242124782, "learning_rate": 0.003, "loss": 4.0766, "step": 10187 }, { "epoch": 0.10188, "grad_norm": 1.148194763035427, "learning_rate": 0.003, "loss": 4.0752, "step": 10188 }, { "epoch": 0.10189, "grad_norm": 1.2204798728816166, "learning_rate": 0.003, "loss": 4.0498, "step": 10189 }, { "epoch": 0.1019, "grad_norm": 0.9145267434089139, "learning_rate": 0.003, "loss": 4.0794, "step": 10190 }, { "epoch": 0.10191, "grad_norm": 1.136204850799004, "learning_rate": 0.003, "loss": 4.0925, "step": 10191 }, { "epoch": 0.10192, "grad_norm": 1.373348499431395, "learning_rate": 0.003, "loss": 4.075, "step": 10192 }, { "epoch": 0.10193, "grad_norm": 1.2291098930583644, "learning_rate": 0.003, "loss": 4.0847, "step": 10193 }, { "epoch": 0.10194, "grad_norm": 1.077193674838292, "learning_rate": 0.003, "loss": 4.097, "step": 10194 }, { "epoch": 0.10195, "grad_norm": 1.0523312544463759, "learning_rate": 0.003, "loss": 4.0979, "step": 10195 }, { "epoch": 0.10196, "grad_norm": 1.1638451574486604, "learning_rate": 0.003, "loss": 4.0923, "step": 10196 }, { "epoch": 0.10197, "grad_norm": 1.1208148962054274, "learning_rate": 0.003, "loss": 4.0916, "step": 10197 }, { "epoch": 0.10198, "grad_norm": 1.2235618242687356, "learning_rate": 0.003, "loss": 4.1126, "step": 10198 }, { "epoch": 0.10199, "grad_norm": 1.1266562751363347, "learning_rate": 0.003, "loss": 4.0859, "step": 10199 }, { "epoch": 0.102, "grad_norm": 1.3174022901674332, "learning_rate": 0.003, "loss": 4.0806, "step": 10200 }, { "epoch": 0.10201, "grad_norm": 1.1387589562083924, "learning_rate": 0.003, "loss": 4.0849, "step": 10201 }, { "epoch": 0.10202, "grad_norm": 1.1561654783381967, "learning_rate": 0.003, "loss": 4.0525, "step": 10202 }, { "epoch": 0.10203, "grad_norm": 1.239835426005678, "learning_rate": 0.003, "loss": 4.0753, "step": 10203 }, { "epoch": 0.10204, "grad_norm": 0.9884086947734628, "learning_rate": 0.003, "loss": 4.0981, "step": 10204 }, { "epoch": 0.10205, "grad_norm": 1.2302368861346946, "learning_rate": 0.003, "loss": 4.082, "step": 10205 }, { "epoch": 0.10206, "grad_norm": 1.1093982540522616, "learning_rate": 0.003, "loss": 4.0915, "step": 10206 }, { "epoch": 0.10207, "grad_norm": 1.339176634282011, "learning_rate": 0.003, "loss": 4.0859, "step": 10207 }, { "epoch": 0.10208, "grad_norm": 1.0896168191058506, "learning_rate": 0.003, "loss": 4.0482, "step": 10208 }, { "epoch": 0.10209, "grad_norm": 1.3023601380111545, "learning_rate": 0.003, "loss": 4.0951, "step": 10209 }, { "epoch": 0.1021, "grad_norm": 1.1319714724397607, "learning_rate": 0.003, "loss": 4.0657, "step": 10210 }, { "epoch": 0.10211, "grad_norm": 1.0388223208698757, "learning_rate": 0.003, "loss": 4.0984, "step": 10211 }, { "epoch": 0.10212, "grad_norm": 1.1553139116759326, "learning_rate": 0.003, "loss": 4.0826, "step": 10212 }, { "epoch": 0.10213, "grad_norm": 0.9724491568977133, "learning_rate": 0.003, "loss": 4.0544, "step": 10213 }, { "epoch": 0.10214, "grad_norm": 1.1061949747140027, "learning_rate": 0.003, "loss": 4.0891, "step": 10214 }, { "epoch": 0.10215, "grad_norm": 1.2796200746117097, "learning_rate": 0.003, "loss": 4.1081, "step": 10215 }, { "epoch": 0.10216, "grad_norm": 0.931370707864801, "learning_rate": 0.003, "loss": 4.089, "step": 10216 }, { "epoch": 0.10217, "grad_norm": 1.3218802452586387, "learning_rate": 0.003, "loss": 4.0807, "step": 10217 }, { "epoch": 0.10218, "grad_norm": 1.433367299708013, "learning_rate": 0.003, "loss": 4.1155, "step": 10218 }, { "epoch": 0.10219, "grad_norm": 1.0503834994854093, "learning_rate": 0.003, "loss": 4.0516, "step": 10219 }, { "epoch": 0.1022, "grad_norm": 1.0554447867190975, "learning_rate": 0.003, "loss": 4.0884, "step": 10220 }, { "epoch": 0.10221, "grad_norm": 1.2932737417269615, "learning_rate": 0.003, "loss": 4.1073, "step": 10221 }, { "epoch": 0.10222, "grad_norm": 1.1891985207325497, "learning_rate": 0.003, "loss": 4.0893, "step": 10222 }, { "epoch": 0.10223, "grad_norm": 1.2076064076458244, "learning_rate": 0.003, "loss": 4.0972, "step": 10223 }, { "epoch": 0.10224, "grad_norm": 0.9557299648740953, "learning_rate": 0.003, "loss": 4.0872, "step": 10224 }, { "epoch": 0.10225, "grad_norm": 1.204468346015853, "learning_rate": 0.003, "loss": 4.088, "step": 10225 }, { "epoch": 0.10226, "grad_norm": 1.2441047386472726, "learning_rate": 0.003, "loss": 4.1353, "step": 10226 }, { "epoch": 0.10227, "grad_norm": 0.9380172757036935, "learning_rate": 0.003, "loss": 4.0892, "step": 10227 }, { "epoch": 0.10228, "grad_norm": 1.2223945943385395, "learning_rate": 0.003, "loss": 4.0987, "step": 10228 }, { "epoch": 0.10229, "grad_norm": 1.316102481302476, "learning_rate": 0.003, "loss": 4.0666, "step": 10229 }, { "epoch": 0.1023, "grad_norm": 0.9082697141003253, "learning_rate": 0.003, "loss": 4.0843, "step": 10230 }, { "epoch": 0.10231, "grad_norm": 0.96075390670666, "learning_rate": 0.003, "loss": 4.0864, "step": 10231 }, { "epoch": 0.10232, "grad_norm": 1.15482596015767, "learning_rate": 0.003, "loss": 4.0924, "step": 10232 }, { "epoch": 0.10233, "grad_norm": 1.0928828537089608, "learning_rate": 0.003, "loss": 4.0639, "step": 10233 }, { "epoch": 0.10234, "grad_norm": 0.944162993108171, "learning_rate": 0.003, "loss": 4.0933, "step": 10234 }, { "epoch": 0.10235, "grad_norm": 1.1235742192351625, "learning_rate": 0.003, "loss": 4.0781, "step": 10235 }, { "epoch": 0.10236, "grad_norm": 1.4527585424607339, "learning_rate": 0.003, "loss": 4.1198, "step": 10236 }, { "epoch": 0.10237, "grad_norm": 0.7729456938299287, "learning_rate": 0.003, "loss": 4.0728, "step": 10237 }, { "epoch": 0.10238, "grad_norm": 0.826808362714121, "learning_rate": 0.003, "loss": 4.0675, "step": 10238 }, { "epoch": 0.10239, "grad_norm": 1.0529521678348204, "learning_rate": 0.003, "loss": 4.1141, "step": 10239 }, { "epoch": 0.1024, "grad_norm": 1.2682105340898882, "learning_rate": 0.003, "loss": 4.0408, "step": 10240 }, { "epoch": 0.10241, "grad_norm": 1.0205246994429473, "learning_rate": 0.003, "loss": 4.0695, "step": 10241 }, { "epoch": 0.10242, "grad_norm": 1.2302603927232567, "learning_rate": 0.003, "loss": 4.0764, "step": 10242 }, { "epoch": 0.10243, "grad_norm": 1.403810456792893, "learning_rate": 0.003, "loss": 4.0946, "step": 10243 }, { "epoch": 0.10244, "grad_norm": 1.0296307975878158, "learning_rate": 0.003, "loss": 4.0757, "step": 10244 }, { "epoch": 0.10245, "grad_norm": 1.1353767920298885, "learning_rate": 0.003, "loss": 4.0633, "step": 10245 }, { "epoch": 0.10246, "grad_norm": 1.2374393275711797, "learning_rate": 0.003, "loss": 4.1108, "step": 10246 }, { "epoch": 0.10247, "grad_norm": 1.1135987774709135, "learning_rate": 0.003, "loss": 4.1113, "step": 10247 }, { "epoch": 0.10248, "grad_norm": 1.202901096533362, "learning_rate": 0.003, "loss": 4.1027, "step": 10248 }, { "epoch": 0.10249, "grad_norm": 1.1617662718599118, "learning_rate": 0.003, "loss": 4.0896, "step": 10249 }, { "epoch": 0.1025, "grad_norm": 1.0837696177852594, "learning_rate": 0.003, "loss": 4.0945, "step": 10250 }, { "epoch": 0.10251, "grad_norm": 0.9925390043486788, "learning_rate": 0.003, "loss": 4.0806, "step": 10251 }, { "epoch": 0.10252, "grad_norm": 1.3752058241211946, "learning_rate": 0.003, "loss": 4.0657, "step": 10252 }, { "epoch": 0.10253, "grad_norm": 1.030876728203176, "learning_rate": 0.003, "loss": 4.0836, "step": 10253 }, { "epoch": 0.10254, "grad_norm": 1.2700450563127943, "learning_rate": 0.003, "loss": 4.0726, "step": 10254 }, { "epoch": 0.10255, "grad_norm": 1.2562782348470567, "learning_rate": 0.003, "loss": 4.0846, "step": 10255 }, { "epoch": 0.10256, "grad_norm": 1.2603460522592087, "learning_rate": 0.003, "loss": 4.0794, "step": 10256 }, { "epoch": 0.10257, "grad_norm": 1.080959098447801, "learning_rate": 0.003, "loss": 4.0691, "step": 10257 }, { "epoch": 0.10258, "grad_norm": 1.034112972768806, "learning_rate": 0.003, "loss": 4.0813, "step": 10258 }, { "epoch": 0.10259, "grad_norm": 1.2521568837250108, "learning_rate": 0.003, "loss": 4.0802, "step": 10259 }, { "epoch": 0.1026, "grad_norm": 1.265948982738962, "learning_rate": 0.003, "loss": 4.0908, "step": 10260 }, { "epoch": 0.10261, "grad_norm": 1.382410009056688, "learning_rate": 0.003, "loss": 4.0875, "step": 10261 }, { "epoch": 0.10262, "grad_norm": 1.0064611117395248, "learning_rate": 0.003, "loss": 4.0855, "step": 10262 }, { "epoch": 0.10263, "grad_norm": 1.1945625123239736, "learning_rate": 0.003, "loss": 4.0678, "step": 10263 }, { "epoch": 0.10264, "grad_norm": 1.14540664203943, "learning_rate": 0.003, "loss": 4.0715, "step": 10264 }, { "epoch": 0.10265, "grad_norm": 1.2367135364640374, "learning_rate": 0.003, "loss": 4.0795, "step": 10265 }, { "epoch": 0.10266, "grad_norm": 1.1122316758156898, "learning_rate": 0.003, "loss": 4.0774, "step": 10266 }, { "epoch": 0.10267, "grad_norm": 1.0701772786087305, "learning_rate": 0.003, "loss": 4.1025, "step": 10267 }, { "epoch": 0.10268, "grad_norm": 1.0731388472404364, "learning_rate": 0.003, "loss": 4.0662, "step": 10268 }, { "epoch": 0.10269, "grad_norm": 1.3059731702245454, "learning_rate": 0.003, "loss": 4.0804, "step": 10269 }, { "epoch": 0.1027, "grad_norm": 0.9631098566908819, "learning_rate": 0.003, "loss": 4.0787, "step": 10270 }, { "epoch": 0.10271, "grad_norm": 1.152291992671532, "learning_rate": 0.003, "loss": 4.095, "step": 10271 }, { "epoch": 0.10272, "grad_norm": 1.0656130885686521, "learning_rate": 0.003, "loss": 4.0575, "step": 10272 }, { "epoch": 0.10273, "grad_norm": 1.1249057999871102, "learning_rate": 0.003, "loss": 4.0961, "step": 10273 }, { "epoch": 0.10274, "grad_norm": 1.0611993649516465, "learning_rate": 0.003, "loss": 4.0696, "step": 10274 }, { "epoch": 0.10275, "grad_norm": 1.2992200726524803, "learning_rate": 0.003, "loss": 4.1023, "step": 10275 }, { "epoch": 0.10276, "grad_norm": 1.054504040345836, "learning_rate": 0.003, "loss": 4.0875, "step": 10276 }, { "epoch": 0.10277, "grad_norm": 1.4833995346186497, "learning_rate": 0.003, "loss": 4.0861, "step": 10277 }, { "epoch": 0.10278, "grad_norm": 0.8234878421729, "learning_rate": 0.003, "loss": 4.077, "step": 10278 }, { "epoch": 0.10279, "grad_norm": 0.9428712206050871, "learning_rate": 0.003, "loss": 4.0833, "step": 10279 }, { "epoch": 0.1028, "grad_norm": 1.4592341165022211, "learning_rate": 0.003, "loss": 4.1023, "step": 10280 }, { "epoch": 0.10281, "grad_norm": 1.156632111531649, "learning_rate": 0.003, "loss": 4.075, "step": 10281 }, { "epoch": 0.10282, "grad_norm": 1.2933653483347347, "learning_rate": 0.003, "loss": 4.072, "step": 10282 }, { "epoch": 0.10283, "grad_norm": 1.1659712151559902, "learning_rate": 0.003, "loss": 4.0897, "step": 10283 }, { "epoch": 0.10284, "grad_norm": 1.2280066889594574, "learning_rate": 0.003, "loss": 4.078, "step": 10284 }, { "epoch": 0.10285, "grad_norm": 1.1488050547183473, "learning_rate": 0.003, "loss": 4.0709, "step": 10285 }, { "epoch": 0.10286, "grad_norm": 1.323738392473033, "learning_rate": 0.003, "loss": 4.0889, "step": 10286 }, { "epoch": 0.10287, "grad_norm": 1.1218747518818815, "learning_rate": 0.003, "loss": 4.0722, "step": 10287 }, { "epoch": 0.10288, "grad_norm": 1.1164468731289583, "learning_rate": 0.003, "loss": 4.1158, "step": 10288 }, { "epoch": 0.10289, "grad_norm": 1.1198194850749519, "learning_rate": 0.003, "loss": 4.1056, "step": 10289 }, { "epoch": 0.1029, "grad_norm": 1.1190422077436522, "learning_rate": 0.003, "loss": 4.1027, "step": 10290 }, { "epoch": 0.10291, "grad_norm": 1.1083018333936319, "learning_rate": 0.003, "loss": 4.0694, "step": 10291 }, { "epoch": 0.10292, "grad_norm": 1.1728920878257667, "learning_rate": 0.003, "loss": 4.1085, "step": 10292 }, { "epoch": 0.10293, "grad_norm": 1.0234079345708287, "learning_rate": 0.003, "loss": 4.085, "step": 10293 }, { "epoch": 0.10294, "grad_norm": 1.3026166075083452, "learning_rate": 0.003, "loss": 4.1116, "step": 10294 }, { "epoch": 0.10295, "grad_norm": 1.09022240976423, "learning_rate": 0.003, "loss": 4.0953, "step": 10295 }, { "epoch": 0.10296, "grad_norm": 1.1713940517600216, "learning_rate": 0.003, "loss": 4.0588, "step": 10296 }, { "epoch": 0.10297, "grad_norm": 1.3662518963462131, "learning_rate": 0.003, "loss": 4.1019, "step": 10297 }, { "epoch": 0.10298, "grad_norm": 0.8570513244454326, "learning_rate": 0.003, "loss": 4.113, "step": 10298 }, { "epoch": 0.10299, "grad_norm": 0.9887649979400258, "learning_rate": 0.003, "loss": 4.0806, "step": 10299 }, { "epoch": 0.103, "grad_norm": 1.3373972917994166, "learning_rate": 0.003, "loss": 4.0973, "step": 10300 }, { "epoch": 0.10301, "grad_norm": 1.1238719201347804, "learning_rate": 0.003, "loss": 4.1049, "step": 10301 }, { "epoch": 0.10302, "grad_norm": 1.5150408743774066, "learning_rate": 0.003, "loss": 4.1212, "step": 10302 }, { "epoch": 0.10303, "grad_norm": 1.05105694956177, "learning_rate": 0.003, "loss": 4.0778, "step": 10303 }, { "epoch": 0.10304, "grad_norm": 1.259009056860587, "learning_rate": 0.003, "loss": 4.1011, "step": 10304 }, { "epoch": 0.10305, "grad_norm": 1.021533029187029, "learning_rate": 0.003, "loss": 4.0766, "step": 10305 }, { "epoch": 0.10306, "grad_norm": 1.2456339894412451, "learning_rate": 0.003, "loss": 4.108, "step": 10306 }, { "epoch": 0.10307, "grad_norm": 0.9300215044673347, "learning_rate": 0.003, "loss": 4.0627, "step": 10307 }, { "epoch": 0.10308, "grad_norm": 1.134643171364985, "learning_rate": 0.003, "loss": 4.0776, "step": 10308 }, { "epoch": 0.10309, "grad_norm": 1.178158076352255, "learning_rate": 0.003, "loss": 4.0647, "step": 10309 }, { "epoch": 0.1031, "grad_norm": 1.3390469838383294, "learning_rate": 0.003, "loss": 4.0684, "step": 10310 }, { "epoch": 0.10311, "grad_norm": 1.1157468121360072, "learning_rate": 0.003, "loss": 4.0785, "step": 10311 }, { "epoch": 0.10312, "grad_norm": 1.2246339507992163, "learning_rate": 0.003, "loss": 4.0829, "step": 10312 }, { "epoch": 0.10313, "grad_norm": 1.0604563014042496, "learning_rate": 0.003, "loss": 4.0555, "step": 10313 }, { "epoch": 0.10314, "grad_norm": 1.1819775023994816, "learning_rate": 0.003, "loss": 4.0911, "step": 10314 }, { "epoch": 0.10315, "grad_norm": 1.0248694025643161, "learning_rate": 0.003, "loss": 4.0774, "step": 10315 }, { "epoch": 0.10316, "grad_norm": 1.2689573088204367, "learning_rate": 0.003, "loss": 4.1202, "step": 10316 }, { "epoch": 0.10317, "grad_norm": 1.0715328778410222, "learning_rate": 0.003, "loss": 4.0714, "step": 10317 }, { "epoch": 0.10318, "grad_norm": 1.242756989409096, "learning_rate": 0.003, "loss": 4.065, "step": 10318 }, { "epoch": 0.10319, "grad_norm": 1.083638891028703, "learning_rate": 0.003, "loss": 4.0811, "step": 10319 }, { "epoch": 0.1032, "grad_norm": 1.1672554434212687, "learning_rate": 0.003, "loss": 4.0895, "step": 10320 }, { "epoch": 0.10321, "grad_norm": 1.1707775081338792, "learning_rate": 0.003, "loss": 4.09, "step": 10321 }, { "epoch": 0.10322, "grad_norm": 0.986097296114779, "learning_rate": 0.003, "loss": 4.1064, "step": 10322 }, { "epoch": 0.10323, "grad_norm": 1.2429808744997115, "learning_rate": 0.003, "loss": 4.0953, "step": 10323 }, { "epoch": 0.10324, "grad_norm": 1.2546021069410018, "learning_rate": 0.003, "loss": 4.0923, "step": 10324 }, { "epoch": 0.10325, "grad_norm": 1.5501024626716682, "learning_rate": 0.003, "loss": 4.0794, "step": 10325 }, { "epoch": 0.10326, "grad_norm": 1.0117683155283734, "learning_rate": 0.003, "loss": 4.0881, "step": 10326 }, { "epoch": 0.10327, "grad_norm": 1.3447969269996676, "learning_rate": 0.003, "loss": 4.0863, "step": 10327 }, { "epoch": 0.10328, "grad_norm": 0.9361989040980547, "learning_rate": 0.003, "loss": 4.0644, "step": 10328 }, { "epoch": 0.10329, "grad_norm": 1.1057788815729743, "learning_rate": 0.003, "loss": 4.0956, "step": 10329 }, { "epoch": 0.1033, "grad_norm": 1.2098144965183406, "learning_rate": 0.003, "loss": 4.0926, "step": 10330 }, { "epoch": 0.10331, "grad_norm": 1.1005575275556037, "learning_rate": 0.003, "loss": 4.0652, "step": 10331 }, { "epoch": 0.10332, "grad_norm": 1.2796524139321448, "learning_rate": 0.003, "loss": 4.0739, "step": 10332 }, { "epoch": 0.10333, "grad_norm": 1.0127273578550784, "learning_rate": 0.003, "loss": 4.0759, "step": 10333 }, { "epoch": 0.10334, "grad_norm": 1.1144472965857977, "learning_rate": 0.003, "loss": 4.1079, "step": 10334 }, { "epoch": 0.10335, "grad_norm": 1.0218677548334034, "learning_rate": 0.003, "loss": 4.0744, "step": 10335 }, { "epoch": 0.10336, "grad_norm": 1.2463894208631197, "learning_rate": 0.003, "loss": 4.0992, "step": 10336 }, { "epoch": 0.10337, "grad_norm": 1.1402067778691138, "learning_rate": 0.003, "loss": 4.084, "step": 10337 }, { "epoch": 0.10338, "grad_norm": 1.076909964879423, "learning_rate": 0.003, "loss": 4.0711, "step": 10338 }, { "epoch": 0.10339, "grad_norm": 1.07400837774445, "learning_rate": 0.003, "loss": 4.0522, "step": 10339 }, { "epoch": 0.1034, "grad_norm": 1.3597158531075113, "learning_rate": 0.003, "loss": 4.0636, "step": 10340 }, { "epoch": 0.10341, "grad_norm": 1.1489541697757224, "learning_rate": 0.003, "loss": 4.072, "step": 10341 }, { "epoch": 0.10342, "grad_norm": 1.1362409836381153, "learning_rate": 0.003, "loss": 4.0569, "step": 10342 }, { "epoch": 0.10343, "grad_norm": 1.2193846234659038, "learning_rate": 0.003, "loss": 4.0844, "step": 10343 }, { "epoch": 0.10344, "grad_norm": 1.1013460412198315, "learning_rate": 0.003, "loss": 4.0872, "step": 10344 }, { "epoch": 0.10345, "grad_norm": 1.1986398777019778, "learning_rate": 0.003, "loss": 4.0735, "step": 10345 }, { "epoch": 0.10346, "grad_norm": 1.2259535634191354, "learning_rate": 0.003, "loss": 4.0583, "step": 10346 }, { "epoch": 0.10347, "grad_norm": 1.3180066632365137, "learning_rate": 0.003, "loss": 4.0795, "step": 10347 }, { "epoch": 0.10348, "grad_norm": 1.0507937399513352, "learning_rate": 0.003, "loss": 4.0974, "step": 10348 }, { "epoch": 0.10349, "grad_norm": 1.3233595709960158, "learning_rate": 0.003, "loss": 4.0845, "step": 10349 }, { "epoch": 0.1035, "grad_norm": 1.1369428252105478, "learning_rate": 0.003, "loss": 4.0818, "step": 10350 }, { "epoch": 0.10351, "grad_norm": 0.9967600558942973, "learning_rate": 0.003, "loss": 4.0854, "step": 10351 }, { "epoch": 0.10352, "grad_norm": 1.1650604603706625, "learning_rate": 0.003, "loss": 4.083, "step": 10352 }, { "epoch": 0.10353, "grad_norm": 0.8988347203171979, "learning_rate": 0.003, "loss": 4.1102, "step": 10353 }, { "epoch": 0.10354, "grad_norm": 1.0825657617706126, "learning_rate": 0.003, "loss": 4.0762, "step": 10354 }, { "epoch": 0.10355, "grad_norm": 1.1940812907579037, "learning_rate": 0.003, "loss": 4.0743, "step": 10355 }, { "epoch": 0.10356, "grad_norm": 1.2644949563081056, "learning_rate": 0.003, "loss": 4.0916, "step": 10356 }, { "epoch": 0.10357, "grad_norm": 1.2076295759178497, "learning_rate": 0.003, "loss": 4.0593, "step": 10357 }, { "epoch": 0.10358, "grad_norm": 1.0173897880605756, "learning_rate": 0.003, "loss": 4.0616, "step": 10358 }, { "epoch": 0.10359, "grad_norm": 1.472853270790185, "learning_rate": 0.003, "loss": 4.0862, "step": 10359 }, { "epoch": 0.1036, "grad_norm": 0.9144735283899198, "learning_rate": 0.003, "loss": 4.0843, "step": 10360 }, { "epoch": 0.10361, "grad_norm": 1.2350910698680162, "learning_rate": 0.003, "loss": 4.0498, "step": 10361 }, { "epoch": 0.10362, "grad_norm": 1.144218155547618, "learning_rate": 0.003, "loss": 4.0653, "step": 10362 }, { "epoch": 0.10363, "grad_norm": 1.1359921765261243, "learning_rate": 0.003, "loss": 4.0713, "step": 10363 }, { "epoch": 0.10364, "grad_norm": 1.1753318212686652, "learning_rate": 0.003, "loss": 4.085, "step": 10364 }, { "epoch": 0.10365, "grad_norm": 1.0614783369827396, "learning_rate": 0.003, "loss": 4.1062, "step": 10365 }, { "epoch": 0.10366, "grad_norm": 1.3653564221120145, "learning_rate": 0.003, "loss": 4.0952, "step": 10366 }, { "epoch": 0.10367, "grad_norm": 1.0235396618427333, "learning_rate": 0.003, "loss": 4.1025, "step": 10367 }, { "epoch": 0.10368, "grad_norm": 1.3294725069518951, "learning_rate": 0.003, "loss": 4.0633, "step": 10368 }, { "epoch": 0.10369, "grad_norm": 1.0195789150289067, "learning_rate": 0.003, "loss": 4.0774, "step": 10369 }, { "epoch": 0.1037, "grad_norm": 1.599898624102901, "learning_rate": 0.003, "loss": 4.0882, "step": 10370 }, { "epoch": 0.10371, "grad_norm": 0.9491921215914594, "learning_rate": 0.003, "loss": 4.0585, "step": 10371 }, { "epoch": 0.10372, "grad_norm": 0.9802561557279339, "learning_rate": 0.003, "loss": 4.0689, "step": 10372 }, { "epoch": 0.10373, "grad_norm": 1.3466915502832943, "learning_rate": 0.003, "loss": 4.0977, "step": 10373 }, { "epoch": 0.10374, "grad_norm": 1.2192167320083085, "learning_rate": 0.003, "loss": 4.0836, "step": 10374 }, { "epoch": 0.10375, "grad_norm": 1.1799499752387714, "learning_rate": 0.003, "loss": 4.0726, "step": 10375 }, { "epoch": 0.10376, "grad_norm": 1.0836478765285256, "learning_rate": 0.003, "loss": 4.0721, "step": 10376 }, { "epoch": 0.10377, "grad_norm": 1.4820696523099477, "learning_rate": 0.003, "loss": 4.096, "step": 10377 }, { "epoch": 0.10378, "grad_norm": 0.9821149971943706, "learning_rate": 0.003, "loss": 4.0544, "step": 10378 }, { "epoch": 0.10379, "grad_norm": 1.0077055514699162, "learning_rate": 0.003, "loss": 4.0615, "step": 10379 }, { "epoch": 0.1038, "grad_norm": 1.3503476406368762, "learning_rate": 0.003, "loss": 4.1151, "step": 10380 }, { "epoch": 0.10381, "grad_norm": 1.050288186407046, "learning_rate": 0.003, "loss": 4.0871, "step": 10381 }, { "epoch": 0.10382, "grad_norm": 1.2049600489251269, "learning_rate": 0.003, "loss": 4.0662, "step": 10382 }, { "epoch": 0.10383, "grad_norm": 0.99486063450449, "learning_rate": 0.003, "loss": 4.0911, "step": 10383 }, { "epoch": 0.10384, "grad_norm": 1.2964579440731385, "learning_rate": 0.003, "loss": 4.1003, "step": 10384 }, { "epoch": 0.10385, "grad_norm": 1.271716496730901, "learning_rate": 0.003, "loss": 4.0896, "step": 10385 }, { "epoch": 0.10386, "grad_norm": 1.1667002374884987, "learning_rate": 0.003, "loss": 4.0936, "step": 10386 }, { "epoch": 0.10387, "grad_norm": 1.1834225987437577, "learning_rate": 0.003, "loss": 4.0914, "step": 10387 }, { "epoch": 0.10388, "grad_norm": 1.1304175541301462, "learning_rate": 0.003, "loss": 4.0797, "step": 10388 }, { "epoch": 0.10389, "grad_norm": 1.256719322090126, "learning_rate": 0.003, "loss": 4.0987, "step": 10389 }, { "epoch": 0.1039, "grad_norm": 1.192656971682307, "learning_rate": 0.003, "loss": 4.0568, "step": 10390 }, { "epoch": 0.10391, "grad_norm": 1.1851085258903133, "learning_rate": 0.003, "loss": 4.0813, "step": 10391 }, { "epoch": 0.10392, "grad_norm": 0.926270846236646, "learning_rate": 0.003, "loss": 4.0794, "step": 10392 }, { "epoch": 0.10393, "grad_norm": 1.0638718913865592, "learning_rate": 0.003, "loss": 4.0721, "step": 10393 }, { "epoch": 0.10394, "grad_norm": 1.2688290938922746, "learning_rate": 0.003, "loss": 4.0843, "step": 10394 }, { "epoch": 0.10395, "grad_norm": 0.9888514003130423, "learning_rate": 0.003, "loss": 4.1049, "step": 10395 }, { "epoch": 0.10396, "grad_norm": 1.387707708861817, "learning_rate": 0.003, "loss": 4.093, "step": 10396 }, { "epoch": 0.10397, "grad_norm": 1.004707297031302, "learning_rate": 0.003, "loss": 4.095, "step": 10397 }, { "epoch": 0.10398, "grad_norm": 1.4190078354572566, "learning_rate": 0.003, "loss": 4.098, "step": 10398 }, { "epoch": 0.10399, "grad_norm": 0.9423223908023143, "learning_rate": 0.003, "loss": 4.0932, "step": 10399 }, { "epoch": 0.104, "grad_norm": 1.1895178529282744, "learning_rate": 0.003, "loss": 4.0879, "step": 10400 }, { "epoch": 0.10401, "grad_norm": 1.1718657484527286, "learning_rate": 0.003, "loss": 4.0808, "step": 10401 }, { "epoch": 0.10402, "grad_norm": 1.211910599497182, "learning_rate": 0.003, "loss": 4.075, "step": 10402 }, { "epoch": 0.10403, "grad_norm": 1.2045298508618671, "learning_rate": 0.003, "loss": 4.0801, "step": 10403 }, { "epoch": 0.10404, "grad_norm": 1.060678391068161, "learning_rate": 0.003, "loss": 4.077, "step": 10404 }, { "epoch": 0.10405, "grad_norm": 0.9070948966110515, "learning_rate": 0.003, "loss": 4.0825, "step": 10405 }, { "epoch": 0.10406, "grad_norm": 1.1432558088594225, "learning_rate": 0.003, "loss": 4.0742, "step": 10406 }, { "epoch": 0.10407, "grad_norm": 1.0400913783180197, "learning_rate": 0.003, "loss": 4.0965, "step": 10407 }, { "epoch": 0.10408, "grad_norm": 1.0531303855940235, "learning_rate": 0.003, "loss": 4.065, "step": 10408 }, { "epoch": 0.10409, "grad_norm": 1.1393712037271202, "learning_rate": 0.003, "loss": 4.0894, "step": 10409 }, { "epoch": 0.1041, "grad_norm": 1.0860395180685651, "learning_rate": 0.003, "loss": 4.115, "step": 10410 }, { "epoch": 0.10411, "grad_norm": 1.1306337552458763, "learning_rate": 0.003, "loss": 4.0758, "step": 10411 }, { "epoch": 0.10412, "grad_norm": 1.300910075580706, "learning_rate": 0.003, "loss": 4.0454, "step": 10412 }, { "epoch": 0.10413, "grad_norm": 0.9914524798821533, "learning_rate": 0.003, "loss": 4.0942, "step": 10413 }, { "epoch": 0.10414, "grad_norm": 1.5336050089105886, "learning_rate": 0.003, "loss": 4.0943, "step": 10414 }, { "epoch": 0.10415, "grad_norm": 1.213025943980225, "learning_rate": 0.003, "loss": 4.0947, "step": 10415 }, { "epoch": 0.10416, "grad_norm": 1.1306085072323393, "learning_rate": 0.003, "loss": 4.1037, "step": 10416 }, { "epoch": 0.10417, "grad_norm": 1.3363346046785622, "learning_rate": 0.003, "loss": 4.1157, "step": 10417 }, { "epoch": 0.10418, "grad_norm": 1.239558459468119, "learning_rate": 0.003, "loss": 4.0661, "step": 10418 }, { "epoch": 0.10419, "grad_norm": 1.0659408116986282, "learning_rate": 0.003, "loss": 4.1264, "step": 10419 }, { "epoch": 0.1042, "grad_norm": 1.2990563819605447, "learning_rate": 0.003, "loss": 4.1077, "step": 10420 }, { "epoch": 0.10421, "grad_norm": 1.045572541998183, "learning_rate": 0.003, "loss": 4.0736, "step": 10421 }, { "epoch": 0.10422, "grad_norm": 1.2573704949094036, "learning_rate": 0.003, "loss": 4.0875, "step": 10422 }, { "epoch": 0.10423, "grad_norm": 1.0570604162814328, "learning_rate": 0.003, "loss": 4.0976, "step": 10423 }, { "epoch": 0.10424, "grad_norm": 1.2160452336095209, "learning_rate": 0.003, "loss": 4.0793, "step": 10424 }, { "epoch": 0.10425, "grad_norm": 0.9409949308423432, "learning_rate": 0.003, "loss": 4.0773, "step": 10425 }, { "epoch": 0.10426, "grad_norm": 1.3636682193227165, "learning_rate": 0.003, "loss": 4.081, "step": 10426 }, { "epoch": 0.10427, "grad_norm": 1.0254345901674822, "learning_rate": 0.003, "loss": 4.0929, "step": 10427 }, { "epoch": 0.10428, "grad_norm": 1.061685565843801, "learning_rate": 0.003, "loss": 4.0654, "step": 10428 }, { "epoch": 0.10429, "grad_norm": 1.1842401380873595, "learning_rate": 0.003, "loss": 4.0665, "step": 10429 }, { "epoch": 0.1043, "grad_norm": 1.0907336276654527, "learning_rate": 0.003, "loss": 4.065, "step": 10430 }, { "epoch": 0.10431, "grad_norm": 1.2487690617445066, "learning_rate": 0.003, "loss": 4.0827, "step": 10431 }, { "epoch": 0.10432, "grad_norm": 1.1487619453121192, "learning_rate": 0.003, "loss": 4.0854, "step": 10432 }, { "epoch": 0.10433, "grad_norm": 1.1353525358762642, "learning_rate": 0.003, "loss": 4.0801, "step": 10433 }, { "epoch": 0.10434, "grad_norm": 1.182062089492732, "learning_rate": 0.003, "loss": 4.0746, "step": 10434 }, { "epoch": 0.10435, "grad_norm": 1.2919162960064097, "learning_rate": 0.003, "loss": 4.0666, "step": 10435 }, { "epoch": 0.10436, "grad_norm": 0.8784182498056757, "learning_rate": 0.003, "loss": 4.0796, "step": 10436 }, { "epoch": 0.10437, "grad_norm": 1.0964932058185943, "learning_rate": 0.003, "loss": 4.0904, "step": 10437 }, { "epoch": 0.10438, "grad_norm": 1.2113358892839388, "learning_rate": 0.003, "loss": 4.0816, "step": 10438 }, { "epoch": 0.10439, "grad_norm": 0.9411748843655284, "learning_rate": 0.003, "loss": 4.0899, "step": 10439 }, { "epoch": 0.1044, "grad_norm": 1.3960224048377663, "learning_rate": 0.003, "loss": 4.0987, "step": 10440 }, { "epoch": 0.10441, "grad_norm": 1.115230953127737, "learning_rate": 0.003, "loss": 4.0699, "step": 10441 }, { "epoch": 0.10442, "grad_norm": 1.3839057582946162, "learning_rate": 0.003, "loss": 4.0718, "step": 10442 }, { "epoch": 0.10443, "grad_norm": 0.9918161777706388, "learning_rate": 0.003, "loss": 4.1031, "step": 10443 }, { "epoch": 0.10444, "grad_norm": 1.275224814509774, "learning_rate": 0.003, "loss": 4.0925, "step": 10444 }, { "epoch": 0.10445, "grad_norm": 1.1568911084251285, "learning_rate": 0.003, "loss": 4.1043, "step": 10445 }, { "epoch": 0.10446, "grad_norm": 1.1098848204188037, "learning_rate": 0.003, "loss": 4.0622, "step": 10446 }, { "epoch": 0.10447, "grad_norm": 1.3475023444604641, "learning_rate": 0.003, "loss": 4.0774, "step": 10447 }, { "epoch": 0.10448, "grad_norm": 1.0874203874677135, "learning_rate": 0.003, "loss": 4.0823, "step": 10448 }, { "epoch": 0.10449, "grad_norm": 1.0170769666280837, "learning_rate": 0.003, "loss": 4.0736, "step": 10449 }, { "epoch": 0.1045, "grad_norm": 1.1462867838652293, "learning_rate": 0.003, "loss": 4.1031, "step": 10450 }, { "epoch": 0.10451, "grad_norm": 1.1467278614404408, "learning_rate": 0.003, "loss": 4.0583, "step": 10451 }, { "epoch": 0.10452, "grad_norm": 1.0082005144502915, "learning_rate": 0.003, "loss": 4.0795, "step": 10452 }, { "epoch": 0.10453, "grad_norm": 1.3099977100561864, "learning_rate": 0.003, "loss": 4.069, "step": 10453 }, { "epoch": 0.10454, "grad_norm": 0.9399797422935121, "learning_rate": 0.003, "loss": 4.066, "step": 10454 }, { "epoch": 0.10455, "grad_norm": 1.1633812167389073, "learning_rate": 0.003, "loss": 4.0691, "step": 10455 }, { "epoch": 0.10456, "grad_norm": 0.8441790544250756, "learning_rate": 0.003, "loss": 4.0913, "step": 10456 }, { "epoch": 0.10457, "grad_norm": 1.2142260614952807, "learning_rate": 0.003, "loss": 4.0917, "step": 10457 }, { "epoch": 0.10458, "grad_norm": 1.7525087370523817, "learning_rate": 0.003, "loss": 4.1103, "step": 10458 }, { "epoch": 0.10459, "grad_norm": 1.0260708858250815, "learning_rate": 0.003, "loss": 4.0739, "step": 10459 }, { "epoch": 0.1046, "grad_norm": 1.1569113558478397, "learning_rate": 0.003, "loss": 4.1043, "step": 10460 }, { "epoch": 0.10461, "grad_norm": 1.1702933967791436, "learning_rate": 0.003, "loss": 4.0687, "step": 10461 }, { "epoch": 0.10462, "grad_norm": 1.13316121198954, "learning_rate": 0.003, "loss": 4.0942, "step": 10462 }, { "epoch": 0.10463, "grad_norm": 1.2833369061026105, "learning_rate": 0.003, "loss": 4.0686, "step": 10463 }, { "epoch": 0.10464, "grad_norm": 0.9714472006978334, "learning_rate": 0.003, "loss": 4.0962, "step": 10464 }, { "epoch": 0.10465, "grad_norm": 1.4026424196175369, "learning_rate": 0.003, "loss": 4.0934, "step": 10465 }, { "epoch": 0.10466, "grad_norm": 1.0745827757635804, "learning_rate": 0.003, "loss": 4.0825, "step": 10466 }, { "epoch": 0.10467, "grad_norm": 1.144442451198987, "learning_rate": 0.003, "loss": 4.0805, "step": 10467 }, { "epoch": 0.10468, "grad_norm": 1.084471302601593, "learning_rate": 0.003, "loss": 4.0631, "step": 10468 }, { "epoch": 0.10469, "grad_norm": 1.2833962066273863, "learning_rate": 0.003, "loss": 4.0936, "step": 10469 }, { "epoch": 0.1047, "grad_norm": 1.193232193819357, "learning_rate": 0.003, "loss": 4.0762, "step": 10470 }, { "epoch": 0.10471, "grad_norm": 0.9239517653004048, "learning_rate": 0.003, "loss": 4.0864, "step": 10471 }, { "epoch": 0.10472, "grad_norm": 1.075646980282147, "learning_rate": 0.003, "loss": 4.0649, "step": 10472 }, { "epoch": 0.10473, "grad_norm": 1.4461175206428305, "learning_rate": 0.003, "loss": 4.0746, "step": 10473 }, { "epoch": 0.10474, "grad_norm": 0.9482031218732518, "learning_rate": 0.003, "loss": 4.07, "step": 10474 }, { "epoch": 0.10475, "grad_norm": 1.0455087032787922, "learning_rate": 0.003, "loss": 4.069, "step": 10475 }, { "epoch": 0.10476, "grad_norm": 1.3144286848750573, "learning_rate": 0.003, "loss": 4.0898, "step": 10476 }, { "epoch": 0.10477, "grad_norm": 1.0497100127868861, "learning_rate": 0.003, "loss": 4.061, "step": 10477 }, { "epoch": 0.10478, "grad_norm": 1.3195516134146819, "learning_rate": 0.003, "loss": 4.0722, "step": 10478 }, { "epoch": 0.10479, "grad_norm": 1.2063061015916368, "learning_rate": 0.003, "loss": 4.0851, "step": 10479 }, { "epoch": 0.1048, "grad_norm": 1.098171732041185, "learning_rate": 0.003, "loss": 4.0638, "step": 10480 }, { "epoch": 0.10481, "grad_norm": 1.085191930367436, "learning_rate": 0.003, "loss": 4.0817, "step": 10481 }, { "epoch": 0.10482, "grad_norm": 1.1856723926494253, "learning_rate": 0.003, "loss": 4.0736, "step": 10482 }, { "epoch": 0.10483, "grad_norm": 1.0735567994984079, "learning_rate": 0.003, "loss": 4.0595, "step": 10483 }, { "epoch": 0.10484, "grad_norm": 1.1245294485238906, "learning_rate": 0.003, "loss": 4.0721, "step": 10484 }, { "epoch": 0.10485, "grad_norm": 1.290202581949648, "learning_rate": 0.003, "loss": 4.0769, "step": 10485 }, { "epoch": 0.10486, "grad_norm": 1.355625022483555, "learning_rate": 0.003, "loss": 4.0532, "step": 10486 }, { "epoch": 0.10487, "grad_norm": 1.2683269217963558, "learning_rate": 0.003, "loss": 4.0926, "step": 10487 }, { "epoch": 0.10488, "grad_norm": 1.0944164469031583, "learning_rate": 0.003, "loss": 4.0744, "step": 10488 }, { "epoch": 0.10489, "grad_norm": 1.2964886991431226, "learning_rate": 0.003, "loss": 4.0848, "step": 10489 }, { "epoch": 0.1049, "grad_norm": 0.8463459006306381, "learning_rate": 0.003, "loss": 4.0878, "step": 10490 }, { "epoch": 0.10491, "grad_norm": 0.8321150941800822, "learning_rate": 0.003, "loss": 4.0917, "step": 10491 }, { "epoch": 0.10492, "grad_norm": 1.0614150475628232, "learning_rate": 0.003, "loss": 4.0776, "step": 10492 }, { "epoch": 0.10493, "grad_norm": 1.3288957408778774, "learning_rate": 0.003, "loss": 4.085, "step": 10493 }, { "epoch": 0.10494, "grad_norm": 1.0381729235961423, "learning_rate": 0.003, "loss": 4.0447, "step": 10494 }, { "epoch": 0.10495, "grad_norm": 1.1504603418116, "learning_rate": 0.003, "loss": 4.0649, "step": 10495 }, { "epoch": 0.10496, "grad_norm": 1.0621219335371417, "learning_rate": 0.003, "loss": 4.0825, "step": 10496 }, { "epoch": 0.10497, "grad_norm": 1.1857329505281322, "learning_rate": 0.003, "loss": 4.0653, "step": 10497 }, { "epoch": 0.10498, "grad_norm": 1.2097394740072545, "learning_rate": 0.003, "loss": 4.0932, "step": 10498 }, { "epoch": 0.10499, "grad_norm": 1.2327086515615175, "learning_rate": 0.003, "loss": 4.0604, "step": 10499 }, { "epoch": 0.105, "grad_norm": 1.0287208162269712, "learning_rate": 0.003, "loss": 4.1016, "step": 10500 }, { "epoch": 0.10501, "grad_norm": 1.3067433688127361, "learning_rate": 0.003, "loss": 4.0671, "step": 10501 }, { "epoch": 0.10502, "grad_norm": 0.8389192771053627, "learning_rate": 0.003, "loss": 4.0595, "step": 10502 }, { "epoch": 0.10503, "grad_norm": 0.8885610051367872, "learning_rate": 0.003, "loss": 4.0992, "step": 10503 }, { "epoch": 0.10504, "grad_norm": 1.129677206792387, "learning_rate": 0.003, "loss": 4.0624, "step": 10504 }, { "epoch": 0.10505, "grad_norm": 1.4582269862849158, "learning_rate": 0.003, "loss": 4.0805, "step": 10505 }, { "epoch": 0.10506, "grad_norm": 0.9728461765979113, "learning_rate": 0.003, "loss": 4.0761, "step": 10506 }, { "epoch": 0.10507, "grad_norm": 1.0183753353440348, "learning_rate": 0.003, "loss": 4.0816, "step": 10507 }, { "epoch": 0.10508, "grad_norm": 1.0906651316555487, "learning_rate": 0.003, "loss": 4.0575, "step": 10508 }, { "epoch": 0.10509, "grad_norm": 1.1755984747198869, "learning_rate": 0.003, "loss": 4.071, "step": 10509 }, { "epoch": 0.1051, "grad_norm": 1.2954430703614785, "learning_rate": 0.003, "loss": 4.0844, "step": 10510 }, { "epoch": 0.10511, "grad_norm": 1.1692617069825422, "learning_rate": 0.003, "loss": 4.0545, "step": 10511 }, { "epoch": 0.10512, "grad_norm": 1.1742433903751759, "learning_rate": 0.003, "loss": 4.1121, "step": 10512 }, { "epoch": 0.10513, "grad_norm": 1.0256581785137762, "learning_rate": 0.003, "loss": 4.0466, "step": 10513 }, { "epoch": 0.10514, "grad_norm": 1.3408802641679947, "learning_rate": 0.003, "loss": 4.0426, "step": 10514 }, { "epoch": 0.10515, "grad_norm": 1.0567380514922469, "learning_rate": 0.003, "loss": 4.087, "step": 10515 }, { "epoch": 0.10516, "grad_norm": 1.396812036919821, "learning_rate": 0.003, "loss": 4.0526, "step": 10516 }, { "epoch": 0.10517, "grad_norm": 1.2013341558124457, "learning_rate": 0.003, "loss": 4.1073, "step": 10517 }, { "epoch": 0.10518, "grad_norm": 1.1212373729312872, "learning_rate": 0.003, "loss": 4.1014, "step": 10518 }, { "epoch": 0.10519, "grad_norm": 1.0783056687463477, "learning_rate": 0.003, "loss": 4.057, "step": 10519 }, { "epoch": 0.1052, "grad_norm": 1.3837530732267258, "learning_rate": 0.003, "loss": 4.1108, "step": 10520 }, { "epoch": 0.10521, "grad_norm": 1.201989834218592, "learning_rate": 0.003, "loss": 4.0795, "step": 10521 }, { "epoch": 0.10522, "grad_norm": 1.184508196310951, "learning_rate": 0.003, "loss": 4.0901, "step": 10522 }, { "epoch": 0.10523, "grad_norm": 1.1274254742305352, "learning_rate": 0.003, "loss": 4.1045, "step": 10523 }, { "epoch": 0.10524, "grad_norm": 1.0024063475584597, "learning_rate": 0.003, "loss": 4.0735, "step": 10524 }, { "epoch": 0.10525, "grad_norm": 1.2163573465379631, "learning_rate": 0.003, "loss": 4.0794, "step": 10525 }, { "epoch": 0.10526, "grad_norm": 1.0786932134805922, "learning_rate": 0.003, "loss": 4.0742, "step": 10526 }, { "epoch": 0.10527, "grad_norm": 1.0145686562796887, "learning_rate": 0.003, "loss": 4.1032, "step": 10527 }, { "epoch": 0.10528, "grad_norm": 1.1055592765968865, "learning_rate": 0.003, "loss": 4.0544, "step": 10528 }, { "epoch": 0.10529, "grad_norm": 1.1306994783428688, "learning_rate": 0.003, "loss": 4.0667, "step": 10529 }, { "epoch": 0.1053, "grad_norm": 1.2836574349271654, "learning_rate": 0.003, "loss": 4.0964, "step": 10530 }, { "epoch": 0.10531, "grad_norm": 1.1432115850931945, "learning_rate": 0.003, "loss": 4.0738, "step": 10531 }, { "epoch": 0.10532, "grad_norm": 1.2618053347146072, "learning_rate": 0.003, "loss": 4.0886, "step": 10532 }, { "epoch": 0.10533, "grad_norm": 0.981706408047873, "learning_rate": 0.003, "loss": 4.0806, "step": 10533 }, { "epoch": 0.10534, "grad_norm": 1.234979426310262, "learning_rate": 0.003, "loss": 4.085, "step": 10534 }, { "epoch": 0.10535, "grad_norm": 0.8845202900772148, "learning_rate": 0.003, "loss": 4.0548, "step": 10535 }, { "epoch": 0.10536, "grad_norm": 0.9068303046180969, "learning_rate": 0.003, "loss": 4.0471, "step": 10536 }, { "epoch": 0.10537, "grad_norm": 1.1135901044513736, "learning_rate": 0.003, "loss": 4.0756, "step": 10537 }, { "epoch": 0.10538, "grad_norm": 1.4681092701245535, "learning_rate": 0.003, "loss": 4.0803, "step": 10538 }, { "epoch": 0.10539, "grad_norm": 1.023925708133604, "learning_rate": 0.003, "loss": 4.0661, "step": 10539 }, { "epoch": 0.1054, "grad_norm": 1.3058416784344284, "learning_rate": 0.003, "loss": 4.0504, "step": 10540 }, { "epoch": 0.10541, "grad_norm": 0.8823316840514877, "learning_rate": 0.003, "loss": 4.086, "step": 10541 }, { "epoch": 0.10542, "grad_norm": 1.0792554022628849, "learning_rate": 0.003, "loss": 4.0755, "step": 10542 }, { "epoch": 0.10543, "grad_norm": 1.1580977814345135, "learning_rate": 0.003, "loss": 4.0781, "step": 10543 }, { "epoch": 0.10544, "grad_norm": 0.9819845632535975, "learning_rate": 0.003, "loss": 4.0589, "step": 10544 }, { "epoch": 0.10545, "grad_norm": 1.2479955950739714, "learning_rate": 0.003, "loss": 4.094, "step": 10545 }, { "epoch": 0.10546, "grad_norm": 1.1609585761944425, "learning_rate": 0.003, "loss": 4.1128, "step": 10546 }, { "epoch": 0.10547, "grad_norm": 1.3531425668991288, "learning_rate": 0.003, "loss": 4.0675, "step": 10547 }, { "epoch": 0.10548, "grad_norm": 1.0063446389136106, "learning_rate": 0.003, "loss": 4.0688, "step": 10548 }, { "epoch": 0.10549, "grad_norm": 1.0325407798781339, "learning_rate": 0.003, "loss": 4.0666, "step": 10549 }, { "epoch": 0.1055, "grad_norm": 1.1423351291584458, "learning_rate": 0.003, "loss": 4.0753, "step": 10550 }, { "epoch": 0.10551, "grad_norm": 1.3503659758788984, "learning_rate": 0.003, "loss": 4.1028, "step": 10551 }, { "epoch": 0.10552, "grad_norm": 1.032045363172022, "learning_rate": 0.003, "loss": 4.0998, "step": 10552 }, { "epoch": 0.10553, "grad_norm": 1.2490414067587199, "learning_rate": 0.003, "loss": 4.0568, "step": 10553 }, { "epoch": 0.10554, "grad_norm": 1.1538988693639156, "learning_rate": 0.003, "loss": 4.0782, "step": 10554 }, { "epoch": 0.10555, "grad_norm": 1.0159744608616759, "learning_rate": 0.003, "loss": 4.0929, "step": 10555 }, { "epoch": 0.10556, "grad_norm": 1.1185713174374186, "learning_rate": 0.003, "loss": 4.0805, "step": 10556 }, { "epoch": 0.10557, "grad_norm": 1.4202293030936732, "learning_rate": 0.003, "loss": 4.0807, "step": 10557 }, { "epoch": 0.10558, "grad_norm": 1.0266908330606117, "learning_rate": 0.003, "loss": 4.0563, "step": 10558 }, { "epoch": 0.10559, "grad_norm": 1.3782557360428902, "learning_rate": 0.003, "loss": 4.0684, "step": 10559 }, { "epoch": 0.1056, "grad_norm": 0.9892321589682566, "learning_rate": 0.003, "loss": 4.0626, "step": 10560 }, { "epoch": 0.10561, "grad_norm": 1.4029111054013261, "learning_rate": 0.003, "loss": 4.08, "step": 10561 }, { "epoch": 0.10562, "grad_norm": 1.0593993799476236, "learning_rate": 0.003, "loss": 4.0565, "step": 10562 }, { "epoch": 0.10563, "grad_norm": 1.351876246310265, "learning_rate": 0.003, "loss": 4.1037, "step": 10563 }, { "epoch": 0.10564, "grad_norm": 1.164144715391898, "learning_rate": 0.003, "loss": 4.0997, "step": 10564 }, { "epoch": 0.10565, "grad_norm": 1.0477122601932365, "learning_rate": 0.003, "loss": 4.0849, "step": 10565 }, { "epoch": 0.10566, "grad_norm": 1.0978684188474286, "learning_rate": 0.003, "loss": 4.0926, "step": 10566 }, { "epoch": 0.10567, "grad_norm": 1.1015952703676826, "learning_rate": 0.003, "loss": 4.1121, "step": 10567 }, { "epoch": 0.10568, "grad_norm": 1.1622739145924461, "learning_rate": 0.003, "loss": 4.108, "step": 10568 }, { "epoch": 0.10569, "grad_norm": 1.300806287594485, "learning_rate": 0.003, "loss": 4.1107, "step": 10569 }, { "epoch": 0.1057, "grad_norm": 1.1367400633037008, "learning_rate": 0.003, "loss": 4.0747, "step": 10570 }, { "epoch": 0.10571, "grad_norm": 1.0314579163465747, "learning_rate": 0.003, "loss": 4.0836, "step": 10571 }, { "epoch": 0.10572, "grad_norm": 1.2863399997244334, "learning_rate": 0.003, "loss": 4.092, "step": 10572 }, { "epoch": 0.10573, "grad_norm": 1.2736178186569311, "learning_rate": 0.003, "loss": 4.0751, "step": 10573 }, { "epoch": 0.10574, "grad_norm": 1.1340756812991262, "learning_rate": 0.003, "loss": 4.0875, "step": 10574 }, { "epoch": 0.10575, "grad_norm": 1.1035950954361586, "learning_rate": 0.003, "loss": 4.0609, "step": 10575 }, { "epoch": 0.10576, "grad_norm": 1.1926319049621437, "learning_rate": 0.003, "loss": 4.1028, "step": 10576 }, { "epoch": 0.10577, "grad_norm": 1.2414399446888533, "learning_rate": 0.003, "loss": 4.074, "step": 10577 }, { "epoch": 0.10578, "grad_norm": 1.1865528097901785, "learning_rate": 0.003, "loss": 4.0854, "step": 10578 }, { "epoch": 0.10579, "grad_norm": 1.03281303097576, "learning_rate": 0.003, "loss": 4.04, "step": 10579 }, { "epoch": 0.1058, "grad_norm": 1.355791716984305, "learning_rate": 0.003, "loss": 4.1008, "step": 10580 }, { "epoch": 0.10581, "grad_norm": 0.9732650815642186, "learning_rate": 0.003, "loss": 4.0959, "step": 10581 }, { "epoch": 0.10582, "grad_norm": 1.3168947077970985, "learning_rate": 0.003, "loss": 4.0803, "step": 10582 }, { "epoch": 0.10583, "grad_norm": 1.126918732108372, "learning_rate": 0.003, "loss": 4.0769, "step": 10583 }, { "epoch": 0.10584, "grad_norm": 1.236038574060123, "learning_rate": 0.003, "loss": 4.0648, "step": 10584 }, { "epoch": 0.10585, "grad_norm": 0.9764226588991661, "learning_rate": 0.003, "loss": 4.1164, "step": 10585 }, { "epoch": 0.10586, "grad_norm": 1.4613436318029804, "learning_rate": 0.003, "loss": 4.1013, "step": 10586 }, { "epoch": 0.10587, "grad_norm": 0.8912353673808929, "learning_rate": 0.003, "loss": 4.1042, "step": 10587 }, { "epoch": 0.10588, "grad_norm": 1.1660830397020308, "learning_rate": 0.003, "loss": 4.0632, "step": 10588 }, { "epoch": 0.10589, "grad_norm": 1.2040748292723822, "learning_rate": 0.003, "loss": 4.086, "step": 10589 }, { "epoch": 0.1059, "grad_norm": 1.0023369501968133, "learning_rate": 0.003, "loss": 4.07, "step": 10590 }, { "epoch": 0.10591, "grad_norm": 1.3963382291928614, "learning_rate": 0.003, "loss": 4.069, "step": 10591 }, { "epoch": 0.10592, "grad_norm": 1.1181609850538723, "learning_rate": 0.003, "loss": 4.1205, "step": 10592 }, { "epoch": 0.10593, "grad_norm": 1.0780726062707058, "learning_rate": 0.003, "loss": 4.0533, "step": 10593 }, { "epoch": 0.10594, "grad_norm": 1.0189897758129178, "learning_rate": 0.003, "loss": 4.0836, "step": 10594 }, { "epoch": 0.10595, "grad_norm": 1.1796416670163832, "learning_rate": 0.003, "loss": 4.0791, "step": 10595 }, { "epoch": 0.10596, "grad_norm": 1.0541631031801153, "learning_rate": 0.003, "loss": 4.088, "step": 10596 }, { "epoch": 0.10597, "grad_norm": 1.2244230665198172, "learning_rate": 0.003, "loss": 4.084, "step": 10597 }, { "epoch": 0.10598, "grad_norm": 1.1989598614086323, "learning_rate": 0.003, "loss": 4.0959, "step": 10598 }, { "epoch": 0.10599, "grad_norm": 1.3731573352295356, "learning_rate": 0.003, "loss": 4.0779, "step": 10599 }, { "epoch": 0.106, "grad_norm": 1.1170959555314157, "learning_rate": 0.003, "loss": 4.0835, "step": 10600 }, { "epoch": 0.10601, "grad_norm": 1.0774495893470961, "learning_rate": 0.003, "loss": 4.0823, "step": 10601 }, { "epoch": 0.10602, "grad_norm": 1.2011499694569694, "learning_rate": 0.003, "loss": 4.1017, "step": 10602 }, { "epoch": 0.10603, "grad_norm": 1.0407850956499687, "learning_rate": 0.003, "loss": 4.0846, "step": 10603 }, { "epoch": 0.10604, "grad_norm": 1.1901650467323437, "learning_rate": 0.003, "loss": 4.1149, "step": 10604 }, { "epoch": 0.10605, "grad_norm": 1.3123752042362158, "learning_rate": 0.003, "loss": 4.071, "step": 10605 }, { "epoch": 0.10606, "grad_norm": 1.1092558216886073, "learning_rate": 0.003, "loss": 4.0891, "step": 10606 }, { "epoch": 0.10607, "grad_norm": 1.4070031238725738, "learning_rate": 0.003, "loss": 4.117, "step": 10607 }, { "epoch": 0.10608, "grad_norm": 0.9636786414342008, "learning_rate": 0.003, "loss": 4.0791, "step": 10608 }, { "epoch": 0.10609, "grad_norm": 1.069934629136634, "learning_rate": 0.003, "loss": 4.0937, "step": 10609 }, { "epoch": 0.1061, "grad_norm": 1.3188057349981706, "learning_rate": 0.003, "loss": 4.0741, "step": 10610 }, { "epoch": 0.10611, "grad_norm": 1.0898838550748384, "learning_rate": 0.003, "loss": 4.0412, "step": 10611 }, { "epoch": 0.10612, "grad_norm": 1.0798870153432583, "learning_rate": 0.003, "loss": 4.0661, "step": 10612 }, { "epoch": 0.10613, "grad_norm": 1.2461876109787824, "learning_rate": 0.003, "loss": 4.07, "step": 10613 }, { "epoch": 0.10614, "grad_norm": 1.160551479100354, "learning_rate": 0.003, "loss": 4.0903, "step": 10614 }, { "epoch": 0.10615, "grad_norm": 1.0023266443176357, "learning_rate": 0.003, "loss": 4.0882, "step": 10615 }, { "epoch": 0.10616, "grad_norm": 1.454836687976706, "learning_rate": 0.003, "loss": 4.1119, "step": 10616 }, { "epoch": 0.10617, "grad_norm": 1.1059202732456763, "learning_rate": 0.003, "loss": 4.0996, "step": 10617 }, { "epoch": 0.10618, "grad_norm": 1.1665251025931167, "learning_rate": 0.003, "loss": 4.0853, "step": 10618 }, { "epoch": 0.10619, "grad_norm": 1.1209740969721296, "learning_rate": 0.003, "loss": 4.0868, "step": 10619 }, { "epoch": 0.1062, "grad_norm": 1.066063305689871, "learning_rate": 0.003, "loss": 4.0683, "step": 10620 }, { "epoch": 0.10621, "grad_norm": 1.2008845335202172, "learning_rate": 0.003, "loss": 4.0672, "step": 10621 }, { "epoch": 0.10622, "grad_norm": 1.161660223257581, "learning_rate": 0.003, "loss": 4.1015, "step": 10622 }, { "epoch": 0.10623, "grad_norm": 1.6711171820413637, "learning_rate": 0.003, "loss": 4.103, "step": 10623 }, { "epoch": 0.10624, "grad_norm": 0.8884554897648648, "learning_rate": 0.003, "loss": 4.0719, "step": 10624 }, { "epoch": 0.10625, "grad_norm": 0.8383413684344952, "learning_rate": 0.003, "loss": 4.066, "step": 10625 }, { "epoch": 0.10626, "grad_norm": 1.2129767496942065, "learning_rate": 0.003, "loss": 4.0578, "step": 10626 }, { "epoch": 0.10627, "grad_norm": 1.186003854674095, "learning_rate": 0.003, "loss": 4.0929, "step": 10627 }, { "epoch": 0.10628, "grad_norm": 1.487670135436451, "learning_rate": 0.003, "loss": 4.0762, "step": 10628 }, { "epoch": 0.10629, "grad_norm": 1.1376454723598795, "learning_rate": 0.003, "loss": 4.0782, "step": 10629 }, { "epoch": 0.1063, "grad_norm": 0.9890545429620673, "learning_rate": 0.003, "loss": 4.0632, "step": 10630 }, { "epoch": 0.10631, "grad_norm": 1.134570590643702, "learning_rate": 0.003, "loss": 4.0963, "step": 10631 }, { "epoch": 0.10632, "grad_norm": 1.049562191577295, "learning_rate": 0.003, "loss": 4.0479, "step": 10632 }, { "epoch": 0.10633, "grad_norm": 1.3046806308376384, "learning_rate": 0.003, "loss": 4.0668, "step": 10633 }, { "epoch": 0.10634, "grad_norm": 1.2510913324277757, "learning_rate": 0.003, "loss": 4.0808, "step": 10634 }, { "epoch": 0.10635, "grad_norm": 1.0580335535557992, "learning_rate": 0.003, "loss": 4.0714, "step": 10635 }, { "epoch": 0.10636, "grad_norm": 1.139536280056406, "learning_rate": 0.003, "loss": 4.0652, "step": 10636 }, { "epoch": 0.10637, "grad_norm": 1.219965449308225, "learning_rate": 0.003, "loss": 4.0758, "step": 10637 }, { "epoch": 0.10638, "grad_norm": 1.0682385047561698, "learning_rate": 0.003, "loss": 4.0785, "step": 10638 }, { "epoch": 0.10639, "grad_norm": 1.014126460631885, "learning_rate": 0.003, "loss": 4.0755, "step": 10639 }, { "epoch": 0.1064, "grad_norm": 1.3296713286352824, "learning_rate": 0.003, "loss": 4.0723, "step": 10640 }, { "epoch": 0.10641, "grad_norm": 0.850836262660311, "learning_rate": 0.003, "loss": 4.0751, "step": 10641 }, { "epoch": 0.10642, "grad_norm": 0.9780187533229948, "learning_rate": 0.003, "loss": 4.0674, "step": 10642 }, { "epoch": 0.10643, "grad_norm": 1.2420781102234009, "learning_rate": 0.003, "loss": 4.0611, "step": 10643 }, { "epoch": 0.10644, "grad_norm": 1.295759972755444, "learning_rate": 0.003, "loss": 4.0993, "step": 10644 }, { "epoch": 0.10645, "grad_norm": 1.4513465623405362, "learning_rate": 0.003, "loss": 4.0727, "step": 10645 }, { "epoch": 0.10646, "grad_norm": 1.20752416924251, "learning_rate": 0.003, "loss": 4.0755, "step": 10646 }, { "epoch": 0.10647, "grad_norm": 1.0423488355215869, "learning_rate": 0.003, "loss": 4.0781, "step": 10647 }, { "epoch": 0.10648, "grad_norm": 1.1509516111491, "learning_rate": 0.003, "loss": 4.0805, "step": 10648 }, { "epoch": 0.10649, "grad_norm": 1.1565147136273588, "learning_rate": 0.003, "loss": 4.0736, "step": 10649 }, { "epoch": 0.1065, "grad_norm": 1.0850380169542058, "learning_rate": 0.003, "loss": 4.1076, "step": 10650 }, { "epoch": 0.10651, "grad_norm": 1.2937739072167689, "learning_rate": 0.003, "loss": 4.0455, "step": 10651 }, { "epoch": 0.10652, "grad_norm": 1.1400309243264664, "learning_rate": 0.003, "loss": 4.0809, "step": 10652 }, { "epoch": 0.10653, "grad_norm": 1.3398794194480168, "learning_rate": 0.003, "loss": 4.0869, "step": 10653 }, { "epoch": 0.10654, "grad_norm": 0.900721665595705, "learning_rate": 0.003, "loss": 4.0732, "step": 10654 }, { "epoch": 0.10655, "grad_norm": 0.9738388665304021, "learning_rate": 0.003, "loss": 4.08, "step": 10655 }, { "epoch": 0.10656, "grad_norm": 0.9972631904494703, "learning_rate": 0.003, "loss": 4.0641, "step": 10656 }, { "epoch": 0.10657, "grad_norm": 1.3116692550014106, "learning_rate": 0.003, "loss": 4.0885, "step": 10657 }, { "epoch": 0.10658, "grad_norm": 1.1119940904641739, "learning_rate": 0.003, "loss": 4.0651, "step": 10658 }, { "epoch": 0.10659, "grad_norm": 1.2235179754264862, "learning_rate": 0.003, "loss": 4.0922, "step": 10659 }, { "epoch": 0.1066, "grad_norm": 1.0639452001222656, "learning_rate": 0.003, "loss": 4.0655, "step": 10660 }, { "epoch": 0.10661, "grad_norm": 1.2330959592550323, "learning_rate": 0.003, "loss": 4.0924, "step": 10661 }, { "epoch": 0.10662, "grad_norm": 1.286976857529244, "learning_rate": 0.003, "loss": 4.0861, "step": 10662 }, { "epoch": 0.10663, "grad_norm": 1.2171565929941812, "learning_rate": 0.003, "loss": 4.0724, "step": 10663 }, { "epoch": 0.10664, "grad_norm": 1.0935451659013982, "learning_rate": 0.003, "loss": 4.0725, "step": 10664 }, { "epoch": 0.10665, "grad_norm": 1.2188284949481023, "learning_rate": 0.003, "loss": 4.0788, "step": 10665 }, { "epoch": 0.10666, "grad_norm": 1.397262797478184, "learning_rate": 0.003, "loss": 4.0664, "step": 10666 }, { "epoch": 0.10667, "grad_norm": 1.1459926815948558, "learning_rate": 0.003, "loss": 4.0462, "step": 10667 }, { "epoch": 0.10668, "grad_norm": 1.3576298480008717, "learning_rate": 0.003, "loss": 4.0738, "step": 10668 }, { "epoch": 0.10669, "grad_norm": 0.9923081391132149, "learning_rate": 0.003, "loss": 4.0429, "step": 10669 }, { "epoch": 0.1067, "grad_norm": 1.0964860970428374, "learning_rate": 0.003, "loss": 4.0616, "step": 10670 }, { "epoch": 0.10671, "grad_norm": 1.1424381515252007, "learning_rate": 0.003, "loss": 4.073, "step": 10671 }, { "epoch": 0.10672, "grad_norm": 1.1371330916319933, "learning_rate": 0.003, "loss": 4.0821, "step": 10672 }, { "epoch": 0.10673, "grad_norm": 1.1158052623087227, "learning_rate": 0.003, "loss": 4.0718, "step": 10673 }, { "epoch": 0.10674, "grad_norm": 1.0989876469829485, "learning_rate": 0.003, "loss": 4.0652, "step": 10674 }, { "epoch": 0.10675, "grad_norm": 1.2350446853004926, "learning_rate": 0.003, "loss": 4.106, "step": 10675 }, { "epoch": 0.10676, "grad_norm": 1.091518153578478, "learning_rate": 0.003, "loss": 4.0937, "step": 10676 }, { "epoch": 0.10677, "grad_norm": 1.0887157550467645, "learning_rate": 0.003, "loss": 4.0413, "step": 10677 }, { "epoch": 0.10678, "grad_norm": 1.4079144898312181, "learning_rate": 0.003, "loss": 4.0776, "step": 10678 }, { "epoch": 0.10679, "grad_norm": 1.4385828914523937, "learning_rate": 0.003, "loss": 4.0802, "step": 10679 }, { "epoch": 0.1068, "grad_norm": 1.2632143752877412, "learning_rate": 0.003, "loss": 4.1004, "step": 10680 }, { "epoch": 0.10681, "grad_norm": 1.2711449191855915, "learning_rate": 0.003, "loss": 4.0681, "step": 10681 }, { "epoch": 0.10682, "grad_norm": 1.1502078722655877, "learning_rate": 0.003, "loss": 4.1003, "step": 10682 }, { "epoch": 0.10683, "grad_norm": 1.1689157301365982, "learning_rate": 0.003, "loss": 4.0854, "step": 10683 }, { "epoch": 0.10684, "grad_norm": 1.089658606167191, "learning_rate": 0.003, "loss": 4.0872, "step": 10684 }, { "epoch": 0.10685, "grad_norm": 1.2028062154824137, "learning_rate": 0.003, "loss": 4.0997, "step": 10685 }, { "epoch": 0.10686, "grad_norm": 1.1532956950445268, "learning_rate": 0.003, "loss": 4.0879, "step": 10686 }, { "epoch": 0.10687, "grad_norm": 1.1798981113852622, "learning_rate": 0.003, "loss": 4.0828, "step": 10687 }, { "epoch": 0.10688, "grad_norm": 1.2487741120499263, "learning_rate": 0.003, "loss": 4.0881, "step": 10688 }, { "epoch": 0.10689, "grad_norm": 1.1511081846376128, "learning_rate": 0.003, "loss": 4.0795, "step": 10689 }, { "epoch": 0.1069, "grad_norm": 1.0698421224764414, "learning_rate": 0.003, "loss": 4.0804, "step": 10690 }, { "epoch": 0.10691, "grad_norm": 1.0392026107147678, "learning_rate": 0.003, "loss": 4.1098, "step": 10691 }, { "epoch": 0.10692, "grad_norm": 1.166887026494509, "learning_rate": 0.003, "loss": 4.0811, "step": 10692 }, { "epoch": 0.10693, "grad_norm": 1.2943326893489389, "learning_rate": 0.003, "loss": 4.0563, "step": 10693 }, { "epoch": 0.10694, "grad_norm": 1.188867274604835, "learning_rate": 0.003, "loss": 4.0721, "step": 10694 }, { "epoch": 0.10695, "grad_norm": 1.2966549339357967, "learning_rate": 0.003, "loss": 4.0642, "step": 10695 }, { "epoch": 0.10696, "grad_norm": 1.1063571292520855, "learning_rate": 0.003, "loss": 4.0879, "step": 10696 }, { "epoch": 0.10697, "grad_norm": 1.2644135704444233, "learning_rate": 0.003, "loss": 4.0908, "step": 10697 }, { "epoch": 0.10698, "grad_norm": 0.8829702212715093, "learning_rate": 0.003, "loss": 4.0905, "step": 10698 }, { "epoch": 0.10699, "grad_norm": 1.1161125424172864, "learning_rate": 0.003, "loss": 4.1081, "step": 10699 }, { "epoch": 0.107, "grad_norm": 1.1207907296680133, "learning_rate": 0.003, "loss": 4.1017, "step": 10700 }, { "epoch": 0.10701, "grad_norm": 1.3700136520118256, "learning_rate": 0.003, "loss": 4.0688, "step": 10701 }, { "epoch": 0.10702, "grad_norm": 0.9950362140263944, "learning_rate": 0.003, "loss": 4.0614, "step": 10702 }, { "epoch": 0.10703, "grad_norm": 1.510859550825327, "learning_rate": 0.003, "loss": 4.0984, "step": 10703 }, { "epoch": 0.10704, "grad_norm": 0.7815725976781704, "learning_rate": 0.003, "loss": 4.0721, "step": 10704 }, { "epoch": 0.10705, "grad_norm": 0.9710303251899548, "learning_rate": 0.003, "loss": 4.0778, "step": 10705 }, { "epoch": 0.10706, "grad_norm": 1.3959458546094592, "learning_rate": 0.003, "loss": 4.0966, "step": 10706 }, { "epoch": 0.10707, "grad_norm": 1.5291106936962824, "learning_rate": 0.003, "loss": 4.0671, "step": 10707 }, { "epoch": 0.10708, "grad_norm": 0.8437013934373869, "learning_rate": 0.003, "loss": 4.0877, "step": 10708 }, { "epoch": 0.10709, "grad_norm": 1.1558904845934623, "learning_rate": 0.003, "loss": 4.0964, "step": 10709 }, { "epoch": 0.1071, "grad_norm": 1.29722970020292, "learning_rate": 0.003, "loss": 4.0825, "step": 10710 }, { "epoch": 0.10711, "grad_norm": 0.859426982103532, "learning_rate": 0.003, "loss": 4.0818, "step": 10711 }, { "epoch": 0.10712, "grad_norm": 1.0814119668627848, "learning_rate": 0.003, "loss": 4.1004, "step": 10712 }, { "epoch": 0.10713, "grad_norm": 1.0836003319414476, "learning_rate": 0.003, "loss": 4.0766, "step": 10713 }, { "epoch": 0.10714, "grad_norm": 1.2002969744808742, "learning_rate": 0.003, "loss": 4.1, "step": 10714 }, { "epoch": 0.10715, "grad_norm": 1.235398778887394, "learning_rate": 0.003, "loss": 4.0726, "step": 10715 }, { "epoch": 0.10716, "grad_norm": 1.2848874979455636, "learning_rate": 0.003, "loss": 4.0928, "step": 10716 }, { "epoch": 0.10717, "grad_norm": 1.3187911005800033, "learning_rate": 0.003, "loss": 4.0869, "step": 10717 }, { "epoch": 0.10718, "grad_norm": 1.0754491902530563, "learning_rate": 0.003, "loss": 4.046, "step": 10718 }, { "epoch": 0.10719, "grad_norm": 1.1780564966095928, "learning_rate": 0.003, "loss": 4.099, "step": 10719 }, { "epoch": 0.1072, "grad_norm": 0.9569717886019223, "learning_rate": 0.003, "loss": 4.1111, "step": 10720 }, { "epoch": 0.10721, "grad_norm": 1.2052252644508454, "learning_rate": 0.003, "loss": 4.0882, "step": 10721 }, { "epoch": 0.10722, "grad_norm": 1.1518598964675328, "learning_rate": 0.003, "loss": 4.0664, "step": 10722 }, { "epoch": 0.10723, "grad_norm": 1.2581208059738584, "learning_rate": 0.003, "loss": 4.0857, "step": 10723 }, { "epoch": 0.10724, "grad_norm": 0.9900557775494677, "learning_rate": 0.003, "loss": 4.0677, "step": 10724 }, { "epoch": 0.10725, "grad_norm": 1.2056121145888294, "learning_rate": 0.003, "loss": 4.0535, "step": 10725 }, { "epoch": 0.10726, "grad_norm": 1.2315129488700127, "learning_rate": 0.003, "loss": 4.1, "step": 10726 }, { "epoch": 0.10727, "grad_norm": 1.2188154408180971, "learning_rate": 0.003, "loss": 4.0905, "step": 10727 }, { "epoch": 0.10728, "grad_norm": 1.2995915472073434, "learning_rate": 0.003, "loss": 4.1036, "step": 10728 }, { "epoch": 0.10729, "grad_norm": 1.170488160136393, "learning_rate": 0.003, "loss": 4.0728, "step": 10729 }, { "epoch": 0.1073, "grad_norm": 1.23873916978082, "learning_rate": 0.003, "loss": 4.071, "step": 10730 }, { "epoch": 0.10731, "grad_norm": 1.0873846640281402, "learning_rate": 0.003, "loss": 4.0733, "step": 10731 }, { "epoch": 0.10732, "grad_norm": 1.2585084327632297, "learning_rate": 0.003, "loss": 4.0938, "step": 10732 }, { "epoch": 0.10733, "grad_norm": 1.097532457246418, "learning_rate": 0.003, "loss": 4.0944, "step": 10733 }, { "epoch": 0.10734, "grad_norm": 1.426673691228244, "learning_rate": 0.003, "loss": 4.0832, "step": 10734 }, { "epoch": 0.10735, "grad_norm": 1.0017834686251834, "learning_rate": 0.003, "loss": 4.0834, "step": 10735 }, { "epoch": 0.10736, "grad_norm": 1.3745884537168405, "learning_rate": 0.003, "loss": 4.0495, "step": 10736 }, { "epoch": 0.10737, "grad_norm": 1.0946666388549016, "learning_rate": 0.003, "loss": 4.076, "step": 10737 }, { "epoch": 0.10738, "grad_norm": 1.2950358443757186, "learning_rate": 0.003, "loss": 4.0635, "step": 10738 }, { "epoch": 0.10739, "grad_norm": 1.0666381621479504, "learning_rate": 0.003, "loss": 4.0807, "step": 10739 }, { "epoch": 0.1074, "grad_norm": 0.9465427423268675, "learning_rate": 0.003, "loss": 4.0605, "step": 10740 }, { "epoch": 0.10741, "grad_norm": 1.0827188551900107, "learning_rate": 0.003, "loss": 4.091, "step": 10741 }, { "epoch": 0.10742, "grad_norm": 1.16078214440933, "learning_rate": 0.003, "loss": 4.0759, "step": 10742 }, { "epoch": 0.10743, "grad_norm": 1.241621485105003, "learning_rate": 0.003, "loss": 4.0998, "step": 10743 }, { "epoch": 0.10744, "grad_norm": 1.1416330699562267, "learning_rate": 0.003, "loss": 4.0764, "step": 10744 }, { "epoch": 0.10745, "grad_norm": 1.1898306550711573, "learning_rate": 0.003, "loss": 4.0734, "step": 10745 }, { "epoch": 0.10746, "grad_norm": 1.1218560056235294, "learning_rate": 0.003, "loss": 4.0765, "step": 10746 }, { "epoch": 0.10747, "grad_norm": 1.3355326489416064, "learning_rate": 0.003, "loss": 4.0863, "step": 10747 }, { "epoch": 0.10748, "grad_norm": 1.2377403051630251, "learning_rate": 0.003, "loss": 4.1057, "step": 10748 }, { "epoch": 0.10749, "grad_norm": 1.1643784959264663, "learning_rate": 0.003, "loss": 4.0857, "step": 10749 }, { "epoch": 0.1075, "grad_norm": 1.0817261939571676, "learning_rate": 0.003, "loss": 4.0924, "step": 10750 }, { "epoch": 0.10751, "grad_norm": 1.071412951472623, "learning_rate": 0.003, "loss": 4.1033, "step": 10751 }, { "epoch": 0.10752, "grad_norm": 1.129314768018961, "learning_rate": 0.003, "loss": 4.0686, "step": 10752 }, { "epoch": 0.10753, "grad_norm": 1.3150908604706937, "learning_rate": 0.003, "loss": 4.0692, "step": 10753 }, { "epoch": 0.10754, "grad_norm": 1.125804848010354, "learning_rate": 0.003, "loss": 4.0739, "step": 10754 }, { "epoch": 0.10755, "grad_norm": 0.9977931603665849, "learning_rate": 0.003, "loss": 4.0895, "step": 10755 }, { "epoch": 0.10756, "grad_norm": 1.2436945170808427, "learning_rate": 0.003, "loss": 4.0756, "step": 10756 }, { "epoch": 0.10757, "grad_norm": 0.8883670690464539, "learning_rate": 0.003, "loss": 4.0805, "step": 10757 }, { "epoch": 0.10758, "grad_norm": 1.0612366629583811, "learning_rate": 0.003, "loss": 4.0835, "step": 10758 }, { "epoch": 0.10759, "grad_norm": 1.2769764962748154, "learning_rate": 0.003, "loss": 4.0531, "step": 10759 }, { "epoch": 0.1076, "grad_norm": 1.060806806921856, "learning_rate": 0.003, "loss": 4.0828, "step": 10760 }, { "epoch": 0.10761, "grad_norm": 1.4172996097599664, "learning_rate": 0.003, "loss": 4.0764, "step": 10761 }, { "epoch": 0.10762, "grad_norm": 0.964200936262678, "learning_rate": 0.003, "loss": 4.0898, "step": 10762 }, { "epoch": 0.10763, "grad_norm": 1.1448760828760276, "learning_rate": 0.003, "loss": 4.0748, "step": 10763 }, { "epoch": 0.10764, "grad_norm": 1.177398561595663, "learning_rate": 0.003, "loss": 4.0868, "step": 10764 }, { "epoch": 0.10765, "grad_norm": 1.3016816642607325, "learning_rate": 0.003, "loss": 4.082, "step": 10765 }, { "epoch": 0.10766, "grad_norm": 1.2312984478898017, "learning_rate": 0.003, "loss": 4.0706, "step": 10766 }, { "epoch": 0.10767, "grad_norm": 1.0270267866285048, "learning_rate": 0.003, "loss": 4.0725, "step": 10767 }, { "epoch": 0.10768, "grad_norm": 1.337822246381765, "learning_rate": 0.003, "loss": 4.0727, "step": 10768 }, { "epoch": 0.10769, "grad_norm": 1.1389782040182617, "learning_rate": 0.003, "loss": 4.0813, "step": 10769 }, { "epoch": 0.1077, "grad_norm": 1.0710292813632354, "learning_rate": 0.003, "loss": 4.0843, "step": 10770 }, { "epoch": 0.10771, "grad_norm": 1.2348971326728477, "learning_rate": 0.003, "loss": 4.0919, "step": 10771 }, { "epoch": 0.10772, "grad_norm": 1.161967747578216, "learning_rate": 0.003, "loss": 4.0677, "step": 10772 }, { "epoch": 0.10773, "grad_norm": 1.3634709444585764, "learning_rate": 0.003, "loss": 4.09, "step": 10773 }, { "epoch": 0.10774, "grad_norm": 1.0806304958698216, "learning_rate": 0.003, "loss": 4.0863, "step": 10774 }, { "epoch": 0.10775, "grad_norm": 1.2438172089730135, "learning_rate": 0.003, "loss": 4.0953, "step": 10775 }, { "epoch": 0.10776, "grad_norm": 0.9949587102464901, "learning_rate": 0.003, "loss": 4.0652, "step": 10776 }, { "epoch": 0.10777, "grad_norm": 1.3006582082099583, "learning_rate": 0.003, "loss": 4.0852, "step": 10777 }, { "epoch": 0.10778, "grad_norm": 1.0692179806323354, "learning_rate": 0.003, "loss": 4.0667, "step": 10778 }, { "epoch": 0.10779, "grad_norm": 1.2464760909389696, "learning_rate": 0.003, "loss": 4.0574, "step": 10779 }, { "epoch": 0.1078, "grad_norm": 1.1691551207648234, "learning_rate": 0.003, "loss": 4.1025, "step": 10780 }, { "epoch": 0.10781, "grad_norm": 1.196080726419258, "learning_rate": 0.003, "loss": 4.0596, "step": 10781 }, { "epoch": 0.10782, "grad_norm": 1.0266639938567623, "learning_rate": 0.003, "loss": 4.0771, "step": 10782 }, { "epoch": 0.10783, "grad_norm": 1.1797973687393533, "learning_rate": 0.003, "loss": 4.0854, "step": 10783 }, { "epoch": 0.10784, "grad_norm": 1.2600032626739337, "learning_rate": 0.003, "loss": 4.1055, "step": 10784 }, { "epoch": 0.10785, "grad_norm": 1.3016460761536592, "learning_rate": 0.003, "loss": 4.0843, "step": 10785 }, { "epoch": 0.10786, "grad_norm": 1.1906263662241257, "learning_rate": 0.003, "loss": 4.0992, "step": 10786 }, { "epoch": 0.10787, "grad_norm": 1.0763915235301704, "learning_rate": 0.003, "loss": 4.0601, "step": 10787 }, { "epoch": 0.10788, "grad_norm": 1.4651066745900185, "learning_rate": 0.003, "loss": 4.0781, "step": 10788 }, { "epoch": 0.10789, "grad_norm": 0.9201304719705409, "learning_rate": 0.003, "loss": 4.0302, "step": 10789 }, { "epoch": 0.1079, "grad_norm": 1.232011515029564, "learning_rate": 0.003, "loss": 4.0983, "step": 10790 }, { "epoch": 0.10791, "grad_norm": 1.1591988872666497, "learning_rate": 0.003, "loss": 4.1031, "step": 10791 }, { "epoch": 0.10792, "grad_norm": 0.9966425554361823, "learning_rate": 0.003, "loss": 4.0616, "step": 10792 }, { "epoch": 0.10793, "grad_norm": 1.1991131105714399, "learning_rate": 0.003, "loss": 4.0757, "step": 10793 }, { "epoch": 0.10794, "grad_norm": 1.1337112341368891, "learning_rate": 0.003, "loss": 4.0965, "step": 10794 }, { "epoch": 0.10795, "grad_norm": 1.5909269743356649, "learning_rate": 0.003, "loss": 4.0547, "step": 10795 }, { "epoch": 0.10796, "grad_norm": 0.8416057431798984, "learning_rate": 0.003, "loss": 4.0739, "step": 10796 }, { "epoch": 0.10797, "grad_norm": 1.045224639779817, "learning_rate": 0.003, "loss": 4.0573, "step": 10797 }, { "epoch": 0.10798, "grad_norm": 1.2293165519213025, "learning_rate": 0.003, "loss": 4.08, "step": 10798 }, { "epoch": 0.10799, "grad_norm": 1.3020130984066611, "learning_rate": 0.003, "loss": 4.0542, "step": 10799 }, { "epoch": 0.108, "grad_norm": 1.186604828872371, "learning_rate": 0.003, "loss": 4.0936, "step": 10800 }, { "epoch": 0.10801, "grad_norm": 1.097067294795651, "learning_rate": 0.003, "loss": 4.08, "step": 10801 }, { "epoch": 0.10802, "grad_norm": 1.2450394305274315, "learning_rate": 0.003, "loss": 4.0821, "step": 10802 }, { "epoch": 0.10803, "grad_norm": 0.9949904324744665, "learning_rate": 0.003, "loss": 4.0906, "step": 10803 }, { "epoch": 0.10804, "grad_norm": 1.2360068450090314, "learning_rate": 0.003, "loss": 4.0857, "step": 10804 }, { "epoch": 0.10805, "grad_norm": 1.1744858298406828, "learning_rate": 0.003, "loss": 4.0787, "step": 10805 }, { "epoch": 0.10806, "grad_norm": 1.2654260319802026, "learning_rate": 0.003, "loss": 4.061, "step": 10806 }, { "epoch": 0.10807, "grad_norm": 1.3100632848185885, "learning_rate": 0.003, "loss": 4.063, "step": 10807 }, { "epoch": 0.10808, "grad_norm": 1.1062041850452753, "learning_rate": 0.003, "loss": 4.1571, "step": 10808 }, { "epoch": 0.10809, "grad_norm": 0.9827220295477982, "learning_rate": 0.003, "loss": 4.0808, "step": 10809 }, { "epoch": 0.1081, "grad_norm": 1.0524862530261532, "learning_rate": 0.003, "loss": 4.0765, "step": 10810 }, { "epoch": 0.10811, "grad_norm": 1.249344963668008, "learning_rate": 0.003, "loss": 4.107, "step": 10811 }, { "epoch": 0.10812, "grad_norm": 0.9730113013840976, "learning_rate": 0.003, "loss": 4.0492, "step": 10812 }, { "epoch": 0.10813, "grad_norm": 1.2459709042802147, "learning_rate": 0.003, "loss": 4.0714, "step": 10813 }, { "epoch": 0.10814, "grad_norm": 1.1460293966725768, "learning_rate": 0.003, "loss": 4.0806, "step": 10814 }, { "epoch": 0.10815, "grad_norm": 1.019469842228145, "learning_rate": 0.003, "loss": 4.0799, "step": 10815 }, { "epoch": 0.10816, "grad_norm": 1.2233373876851823, "learning_rate": 0.003, "loss": 4.0872, "step": 10816 }, { "epoch": 0.10817, "grad_norm": 1.2413354353726054, "learning_rate": 0.003, "loss": 4.0822, "step": 10817 }, { "epoch": 0.10818, "grad_norm": 1.1987889670937286, "learning_rate": 0.003, "loss": 4.0787, "step": 10818 }, { "epoch": 0.10819, "grad_norm": 1.1933924411078074, "learning_rate": 0.003, "loss": 4.0814, "step": 10819 }, { "epoch": 0.1082, "grad_norm": 1.2359549573144932, "learning_rate": 0.003, "loss": 4.079, "step": 10820 }, { "epoch": 0.10821, "grad_norm": 1.1880375232350786, "learning_rate": 0.003, "loss": 4.0375, "step": 10821 }, { "epoch": 0.10822, "grad_norm": 1.0189809993081607, "learning_rate": 0.003, "loss": 4.0585, "step": 10822 }, { "epoch": 0.10823, "grad_norm": 1.2875865120552497, "learning_rate": 0.003, "loss": 4.054, "step": 10823 }, { "epoch": 0.10824, "grad_norm": 1.101554063341351, "learning_rate": 0.003, "loss": 4.0826, "step": 10824 }, { "epoch": 0.10825, "grad_norm": 1.0262557252237896, "learning_rate": 0.003, "loss": 4.0551, "step": 10825 }, { "epoch": 0.10826, "grad_norm": 1.46748660981515, "learning_rate": 0.003, "loss": 4.0463, "step": 10826 }, { "epoch": 0.10827, "grad_norm": 0.9468098890621249, "learning_rate": 0.003, "loss": 4.103, "step": 10827 }, { "epoch": 0.10828, "grad_norm": 1.1989102371714668, "learning_rate": 0.003, "loss": 4.1142, "step": 10828 }, { "epoch": 0.10829, "grad_norm": 0.9733274516871876, "learning_rate": 0.003, "loss": 4.0595, "step": 10829 }, { "epoch": 0.1083, "grad_norm": 1.521393355716284, "learning_rate": 0.003, "loss": 4.1052, "step": 10830 }, { "epoch": 0.10831, "grad_norm": 1.0648565577154092, "learning_rate": 0.003, "loss": 4.0715, "step": 10831 }, { "epoch": 0.10832, "grad_norm": 1.670795345847375, "learning_rate": 0.003, "loss": 4.1274, "step": 10832 }, { "epoch": 0.10833, "grad_norm": 1.1214446923451595, "learning_rate": 0.003, "loss": 4.0866, "step": 10833 }, { "epoch": 0.10834, "grad_norm": 1.3418726310171458, "learning_rate": 0.003, "loss": 4.0548, "step": 10834 }, { "epoch": 0.10835, "grad_norm": 1.1958076729736424, "learning_rate": 0.003, "loss": 4.0869, "step": 10835 }, { "epoch": 0.10836, "grad_norm": 1.0348374383064476, "learning_rate": 0.003, "loss": 4.0908, "step": 10836 }, { "epoch": 0.10837, "grad_norm": 1.263763456536624, "learning_rate": 0.003, "loss": 4.093, "step": 10837 }, { "epoch": 0.10838, "grad_norm": 1.079535399309003, "learning_rate": 0.003, "loss": 4.1104, "step": 10838 }, { "epoch": 0.10839, "grad_norm": 1.0280742647698429, "learning_rate": 0.003, "loss": 4.0848, "step": 10839 }, { "epoch": 0.1084, "grad_norm": 1.3474646277249114, "learning_rate": 0.003, "loss": 4.1151, "step": 10840 }, { "epoch": 0.10841, "grad_norm": 1.4499211742939313, "learning_rate": 0.003, "loss": 4.0982, "step": 10841 }, { "epoch": 0.10842, "grad_norm": 1.3290309322316827, "learning_rate": 0.003, "loss": 4.1063, "step": 10842 }, { "epoch": 0.10843, "grad_norm": 1.0528497774700494, "learning_rate": 0.003, "loss": 4.087, "step": 10843 }, { "epoch": 0.10844, "grad_norm": 1.260209051962453, "learning_rate": 0.003, "loss": 4.1176, "step": 10844 }, { "epoch": 0.10845, "grad_norm": 1.2502627934282764, "learning_rate": 0.003, "loss": 4.1099, "step": 10845 }, { "epoch": 0.10846, "grad_norm": 0.9786619469735314, "learning_rate": 0.003, "loss": 4.0778, "step": 10846 }, { "epoch": 0.10847, "grad_norm": 1.1772139936380883, "learning_rate": 0.003, "loss": 4.1044, "step": 10847 }, { "epoch": 0.10848, "grad_norm": 1.1154749608556416, "learning_rate": 0.003, "loss": 4.0539, "step": 10848 }, { "epoch": 0.10849, "grad_norm": 1.4255927279548497, "learning_rate": 0.003, "loss": 4.0677, "step": 10849 }, { "epoch": 0.1085, "grad_norm": 0.9591807418215578, "learning_rate": 0.003, "loss": 4.0736, "step": 10850 }, { "epoch": 0.10851, "grad_norm": 1.1500794089470685, "learning_rate": 0.003, "loss": 4.0871, "step": 10851 }, { "epoch": 0.10852, "grad_norm": 1.1733854286996765, "learning_rate": 0.003, "loss": 4.0925, "step": 10852 }, { "epoch": 0.10853, "grad_norm": 1.3167899561992558, "learning_rate": 0.003, "loss": 4.0979, "step": 10853 }, { "epoch": 0.10854, "grad_norm": 1.0983208572180652, "learning_rate": 0.003, "loss": 4.0597, "step": 10854 }, { "epoch": 0.10855, "grad_norm": 1.1066364502124872, "learning_rate": 0.003, "loss": 4.0589, "step": 10855 }, { "epoch": 0.10856, "grad_norm": 1.211760377665726, "learning_rate": 0.003, "loss": 4.0843, "step": 10856 }, { "epoch": 0.10857, "grad_norm": 0.9064617457621427, "learning_rate": 0.003, "loss": 4.0674, "step": 10857 }, { "epoch": 0.10858, "grad_norm": 1.3610637638848726, "learning_rate": 0.003, "loss": 4.0901, "step": 10858 }, { "epoch": 0.10859, "grad_norm": 0.9870487625663688, "learning_rate": 0.003, "loss": 4.0563, "step": 10859 }, { "epoch": 0.1086, "grad_norm": 1.1299893307649607, "learning_rate": 0.003, "loss": 4.0705, "step": 10860 }, { "epoch": 0.10861, "grad_norm": 0.8905455480301787, "learning_rate": 0.003, "loss": 4.0914, "step": 10861 }, { "epoch": 0.10862, "grad_norm": 1.1931426253312616, "learning_rate": 0.003, "loss": 4.124, "step": 10862 }, { "epoch": 0.10863, "grad_norm": 1.3545454024620263, "learning_rate": 0.003, "loss": 4.0993, "step": 10863 }, { "epoch": 0.10864, "grad_norm": 1.0496058644337216, "learning_rate": 0.003, "loss": 4.0733, "step": 10864 }, { "epoch": 0.10865, "grad_norm": 1.3595045084756816, "learning_rate": 0.003, "loss": 4.0903, "step": 10865 }, { "epoch": 0.10866, "grad_norm": 1.0384779967197104, "learning_rate": 0.003, "loss": 4.0873, "step": 10866 }, { "epoch": 0.10867, "grad_norm": 1.3330568756729129, "learning_rate": 0.003, "loss": 4.0588, "step": 10867 }, { "epoch": 0.10868, "grad_norm": 0.9664414853324744, "learning_rate": 0.003, "loss": 4.055, "step": 10868 }, { "epoch": 0.10869, "grad_norm": 1.2749960456735239, "learning_rate": 0.003, "loss": 4.0854, "step": 10869 }, { "epoch": 0.1087, "grad_norm": 1.109639104543151, "learning_rate": 0.003, "loss": 4.0883, "step": 10870 }, { "epoch": 0.10871, "grad_norm": 1.2792810165388266, "learning_rate": 0.003, "loss": 4.071, "step": 10871 }, { "epoch": 0.10872, "grad_norm": 1.1238748789860165, "learning_rate": 0.003, "loss": 4.0708, "step": 10872 }, { "epoch": 0.10873, "grad_norm": 1.0909837652576535, "learning_rate": 0.003, "loss": 4.0854, "step": 10873 }, { "epoch": 0.10874, "grad_norm": 1.3464645036836225, "learning_rate": 0.003, "loss": 4.078, "step": 10874 }, { "epoch": 0.10875, "grad_norm": 1.2481986896710018, "learning_rate": 0.003, "loss": 4.1016, "step": 10875 }, { "epoch": 0.10876, "grad_norm": 1.283555944231391, "learning_rate": 0.003, "loss": 4.0403, "step": 10876 }, { "epoch": 0.10877, "grad_norm": 1.141382207544397, "learning_rate": 0.003, "loss": 4.0679, "step": 10877 }, { "epoch": 0.10878, "grad_norm": 1.0151640131393809, "learning_rate": 0.003, "loss": 4.0676, "step": 10878 }, { "epoch": 0.10879, "grad_norm": 1.3476780151695977, "learning_rate": 0.003, "loss": 4.0867, "step": 10879 }, { "epoch": 0.1088, "grad_norm": 1.1041206730298196, "learning_rate": 0.003, "loss": 4.0685, "step": 10880 }, { "epoch": 0.10881, "grad_norm": 1.0387519214779812, "learning_rate": 0.003, "loss": 4.0495, "step": 10881 }, { "epoch": 0.10882, "grad_norm": 1.4934301337368312, "learning_rate": 0.003, "loss": 4.0793, "step": 10882 }, { "epoch": 0.10883, "grad_norm": 1.1675196985897143, "learning_rate": 0.003, "loss": 4.0978, "step": 10883 }, { "epoch": 0.10884, "grad_norm": 1.0064153324913705, "learning_rate": 0.003, "loss": 4.0786, "step": 10884 }, { "epoch": 0.10885, "grad_norm": 1.2330812158547564, "learning_rate": 0.003, "loss": 4.0736, "step": 10885 }, { "epoch": 0.10886, "grad_norm": 1.116359574003855, "learning_rate": 0.003, "loss": 4.0603, "step": 10886 }, { "epoch": 0.10887, "grad_norm": 1.2550204984733917, "learning_rate": 0.003, "loss": 4.1043, "step": 10887 }, { "epoch": 0.10888, "grad_norm": 0.9007148413092196, "learning_rate": 0.003, "loss": 4.0433, "step": 10888 }, { "epoch": 0.10889, "grad_norm": 1.1800673903239454, "learning_rate": 0.003, "loss": 4.0614, "step": 10889 }, { "epoch": 0.1089, "grad_norm": 1.2503116665259741, "learning_rate": 0.003, "loss": 4.0608, "step": 10890 }, { "epoch": 0.10891, "grad_norm": 0.9694142487528438, "learning_rate": 0.003, "loss": 4.0827, "step": 10891 }, { "epoch": 0.10892, "grad_norm": 1.2388945396491844, "learning_rate": 0.003, "loss": 4.0832, "step": 10892 }, { "epoch": 0.10893, "grad_norm": 1.0902349584357809, "learning_rate": 0.003, "loss": 4.0639, "step": 10893 }, { "epoch": 0.10894, "grad_norm": 1.1154048190089856, "learning_rate": 0.003, "loss": 4.0639, "step": 10894 }, { "epoch": 0.10895, "grad_norm": 1.0622891299100115, "learning_rate": 0.003, "loss": 4.0996, "step": 10895 }, { "epoch": 0.10896, "grad_norm": 0.9948833067306971, "learning_rate": 0.003, "loss": 4.0759, "step": 10896 }, { "epoch": 0.10897, "grad_norm": 1.2959590809617756, "learning_rate": 0.003, "loss": 4.0652, "step": 10897 }, { "epoch": 0.10898, "grad_norm": 0.9015807760549023, "learning_rate": 0.003, "loss": 4.0775, "step": 10898 }, { "epoch": 0.10899, "grad_norm": 1.1824188153007684, "learning_rate": 0.003, "loss": 4.071, "step": 10899 }, { "epoch": 0.109, "grad_norm": 1.1692001862445531, "learning_rate": 0.003, "loss": 4.0973, "step": 10900 }, { "epoch": 0.10901, "grad_norm": 1.0149117448150018, "learning_rate": 0.003, "loss": 4.0679, "step": 10901 }, { "epoch": 0.10902, "grad_norm": 1.1839491756763185, "learning_rate": 0.003, "loss": 4.0692, "step": 10902 }, { "epoch": 0.10903, "grad_norm": 1.0179984934097102, "learning_rate": 0.003, "loss": 4.084, "step": 10903 }, { "epoch": 0.10904, "grad_norm": 1.4614628598796946, "learning_rate": 0.003, "loss": 4.0688, "step": 10904 }, { "epoch": 0.10905, "grad_norm": 1.1953547562934568, "learning_rate": 0.003, "loss": 4.0575, "step": 10905 }, { "epoch": 0.10906, "grad_norm": 1.636601542722464, "learning_rate": 0.003, "loss": 4.1064, "step": 10906 }, { "epoch": 0.10907, "grad_norm": 0.7528713036455322, "learning_rate": 0.003, "loss": 4.075, "step": 10907 }, { "epoch": 0.10908, "grad_norm": 1.0389917267494386, "learning_rate": 0.003, "loss": 4.08, "step": 10908 }, { "epoch": 0.10909, "grad_norm": 1.7925209163614575, "learning_rate": 0.003, "loss": 4.081, "step": 10909 }, { "epoch": 0.1091, "grad_norm": 0.970094220814019, "learning_rate": 0.003, "loss": 4.0795, "step": 10910 }, { "epoch": 0.10911, "grad_norm": 1.179875808883414, "learning_rate": 0.003, "loss": 4.0889, "step": 10911 }, { "epoch": 0.10912, "grad_norm": 1.2181742635726007, "learning_rate": 0.003, "loss": 4.0871, "step": 10912 }, { "epoch": 0.10913, "grad_norm": 1.1496508513916157, "learning_rate": 0.003, "loss": 4.101, "step": 10913 }, { "epoch": 0.10914, "grad_norm": 1.1005537168343975, "learning_rate": 0.003, "loss": 4.0762, "step": 10914 }, { "epoch": 0.10915, "grad_norm": 1.206635927488834, "learning_rate": 0.003, "loss": 4.0829, "step": 10915 }, { "epoch": 0.10916, "grad_norm": 0.956242769077545, "learning_rate": 0.003, "loss": 4.0722, "step": 10916 }, { "epoch": 0.10917, "grad_norm": 1.041315048532933, "learning_rate": 0.003, "loss": 4.0913, "step": 10917 }, { "epoch": 0.10918, "grad_norm": 1.122023578338097, "learning_rate": 0.003, "loss": 4.0914, "step": 10918 }, { "epoch": 0.10919, "grad_norm": 1.0972674985128863, "learning_rate": 0.003, "loss": 4.0703, "step": 10919 }, { "epoch": 0.1092, "grad_norm": 1.2457136032987424, "learning_rate": 0.003, "loss": 4.0762, "step": 10920 }, { "epoch": 0.10921, "grad_norm": 1.3141612985198625, "learning_rate": 0.003, "loss": 4.0956, "step": 10921 }, { "epoch": 0.10922, "grad_norm": 1.132872008720541, "learning_rate": 0.003, "loss": 4.081, "step": 10922 }, { "epoch": 0.10923, "grad_norm": 1.1124959252954467, "learning_rate": 0.003, "loss": 4.0428, "step": 10923 }, { "epoch": 0.10924, "grad_norm": 0.9957527052591222, "learning_rate": 0.003, "loss": 4.0509, "step": 10924 }, { "epoch": 0.10925, "grad_norm": 1.2351433504353926, "learning_rate": 0.003, "loss": 4.0849, "step": 10925 }, { "epoch": 0.10926, "grad_norm": 1.1421529222430997, "learning_rate": 0.003, "loss": 4.0424, "step": 10926 }, { "epoch": 0.10927, "grad_norm": 1.100095566570253, "learning_rate": 0.003, "loss": 4.0856, "step": 10927 }, { "epoch": 0.10928, "grad_norm": 1.1198941608177244, "learning_rate": 0.003, "loss": 4.0692, "step": 10928 }, { "epoch": 0.10929, "grad_norm": 1.1434762329489463, "learning_rate": 0.003, "loss": 4.0568, "step": 10929 }, { "epoch": 0.1093, "grad_norm": 1.3132536106629562, "learning_rate": 0.003, "loss": 4.0925, "step": 10930 }, { "epoch": 0.10931, "grad_norm": 1.1417515606575013, "learning_rate": 0.003, "loss": 4.0907, "step": 10931 }, { "epoch": 0.10932, "grad_norm": 1.248357377123286, "learning_rate": 0.003, "loss": 4.0841, "step": 10932 }, { "epoch": 0.10933, "grad_norm": 1.0008989985258727, "learning_rate": 0.003, "loss": 4.0903, "step": 10933 }, { "epoch": 0.10934, "grad_norm": 1.1789044443234111, "learning_rate": 0.003, "loss": 4.1205, "step": 10934 }, { "epoch": 0.10935, "grad_norm": 1.0501842927343423, "learning_rate": 0.003, "loss": 4.0907, "step": 10935 }, { "epoch": 0.10936, "grad_norm": 1.1345208975658883, "learning_rate": 0.003, "loss": 4.05, "step": 10936 }, { "epoch": 0.10937, "grad_norm": 0.953442335452271, "learning_rate": 0.003, "loss": 4.0985, "step": 10937 }, { "epoch": 0.10938, "grad_norm": 1.1377302461450274, "learning_rate": 0.003, "loss": 4.0948, "step": 10938 }, { "epoch": 0.10939, "grad_norm": 1.443996061455607, "learning_rate": 0.003, "loss": 4.0819, "step": 10939 }, { "epoch": 0.1094, "grad_norm": 1.3938331183018506, "learning_rate": 0.003, "loss": 4.1125, "step": 10940 }, { "epoch": 0.10941, "grad_norm": 0.9701809916187377, "learning_rate": 0.003, "loss": 4.0985, "step": 10941 }, { "epoch": 0.10942, "grad_norm": 1.2668069397873494, "learning_rate": 0.003, "loss": 4.096, "step": 10942 }, { "epoch": 0.10943, "grad_norm": 1.2083114473027807, "learning_rate": 0.003, "loss": 4.0713, "step": 10943 }, { "epoch": 0.10944, "grad_norm": 0.915154558843135, "learning_rate": 0.003, "loss": 4.0786, "step": 10944 }, { "epoch": 0.10945, "grad_norm": 1.1018952265095763, "learning_rate": 0.003, "loss": 4.0818, "step": 10945 }, { "epoch": 0.10946, "grad_norm": 1.2555065686402815, "learning_rate": 0.003, "loss": 4.0929, "step": 10946 }, { "epoch": 0.10947, "grad_norm": 0.9217730848513727, "learning_rate": 0.003, "loss": 4.0605, "step": 10947 }, { "epoch": 0.10948, "grad_norm": 1.2211889008294665, "learning_rate": 0.003, "loss": 4.0677, "step": 10948 }, { "epoch": 0.10949, "grad_norm": 1.2459604011150405, "learning_rate": 0.003, "loss": 4.0648, "step": 10949 }, { "epoch": 0.1095, "grad_norm": 1.1768814916877195, "learning_rate": 0.003, "loss": 4.061, "step": 10950 }, { "epoch": 0.10951, "grad_norm": 1.018547814321062, "learning_rate": 0.003, "loss": 4.0581, "step": 10951 }, { "epoch": 0.10952, "grad_norm": 1.3164194264217637, "learning_rate": 0.003, "loss": 4.0904, "step": 10952 }, { "epoch": 0.10953, "grad_norm": 1.1706035941961062, "learning_rate": 0.003, "loss": 4.0817, "step": 10953 }, { "epoch": 0.10954, "grad_norm": 1.337196572028656, "learning_rate": 0.003, "loss": 4.1074, "step": 10954 }, { "epoch": 0.10955, "grad_norm": 1.0738506043022735, "learning_rate": 0.003, "loss": 4.0713, "step": 10955 }, { "epoch": 0.10956, "grad_norm": 1.2189719121937406, "learning_rate": 0.003, "loss": 4.0952, "step": 10956 }, { "epoch": 0.10957, "grad_norm": 0.9514863447924589, "learning_rate": 0.003, "loss": 4.0546, "step": 10957 }, { "epoch": 0.10958, "grad_norm": 1.2891491203795653, "learning_rate": 0.003, "loss": 4.0783, "step": 10958 }, { "epoch": 0.10959, "grad_norm": 1.0703009561907915, "learning_rate": 0.003, "loss": 4.0875, "step": 10959 }, { "epoch": 0.1096, "grad_norm": 1.3554792243974731, "learning_rate": 0.003, "loss": 4.0569, "step": 10960 }, { "epoch": 0.10961, "grad_norm": 1.0350064870118225, "learning_rate": 0.003, "loss": 4.0641, "step": 10961 }, { "epoch": 0.10962, "grad_norm": 1.2722625320372223, "learning_rate": 0.003, "loss": 4.1015, "step": 10962 }, { "epoch": 0.10963, "grad_norm": 1.0687986810017533, "learning_rate": 0.003, "loss": 4.09, "step": 10963 }, { "epoch": 0.10964, "grad_norm": 1.146695596346475, "learning_rate": 0.003, "loss": 4.0919, "step": 10964 }, { "epoch": 0.10965, "grad_norm": 1.115830007762537, "learning_rate": 0.003, "loss": 4.0448, "step": 10965 }, { "epoch": 0.10966, "grad_norm": 1.089456831868218, "learning_rate": 0.003, "loss": 4.0741, "step": 10966 }, { "epoch": 0.10967, "grad_norm": 1.2396652030097792, "learning_rate": 0.003, "loss": 4.0974, "step": 10967 }, { "epoch": 0.10968, "grad_norm": 1.1821538893519532, "learning_rate": 0.003, "loss": 4.103, "step": 10968 }, { "epoch": 0.10969, "grad_norm": 1.2414507644573376, "learning_rate": 0.003, "loss": 4.1028, "step": 10969 }, { "epoch": 0.1097, "grad_norm": 1.1096500583314424, "learning_rate": 0.003, "loss": 4.0799, "step": 10970 }, { "epoch": 0.10971, "grad_norm": 1.3266075353839237, "learning_rate": 0.003, "loss": 4.0993, "step": 10971 }, { "epoch": 0.10972, "grad_norm": 1.1288752735161007, "learning_rate": 0.003, "loss": 4.123, "step": 10972 }, { "epoch": 0.10973, "grad_norm": 1.1893059979730722, "learning_rate": 0.003, "loss": 4.0778, "step": 10973 }, { "epoch": 0.10974, "grad_norm": 1.1016118927584362, "learning_rate": 0.003, "loss": 4.0679, "step": 10974 }, { "epoch": 0.10975, "grad_norm": 1.1496691228515794, "learning_rate": 0.003, "loss": 4.088, "step": 10975 }, { "epoch": 0.10976, "grad_norm": 1.1396501642870858, "learning_rate": 0.003, "loss": 4.0665, "step": 10976 }, { "epoch": 0.10977, "grad_norm": 1.154739766385983, "learning_rate": 0.003, "loss": 4.0837, "step": 10977 }, { "epoch": 0.10978, "grad_norm": 1.2202424954301483, "learning_rate": 0.003, "loss": 4.0935, "step": 10978 }, { "epoch": 0.10979, "grad_norm": 1.024190700243081, "learning_rate": 0.003, "loss": 4.0748, "step": 10979 }, { "epoch": 0.1098, "grad_norm": 1.2310691922569879, "learning_rate": 0.003, "loss": 4.073, "step": 10980 }, { "epoch": 0.10981, "grad_norm": 1.1718885237175614, "learning_rate": 0.003, "loss": 4.0889, "step": 10981 }, { "epoch": 0.10982, "grad_norm": 1.1343921348884538, "learning_rate": 0.003, "loss": 4.0518, "step": 10982 }, { "epoch": 0.10983, "grad_norm": 1.220869472852959, "learning_rate": 0.003, "loss": 4.0559, "step": 10983 }, { "epoch": 0.10984, "grad_norm": 1.2531776119674387, "learning_rate": 0.003, "loss": 4.1018, "step": 10984 }, { "epoch": 0.10985, "grad_norm": 1.0520148216339635, "learning_rate": 0.003, "loss": 4.0798, "step": 10985 }, { "epoch": 0.10986, "grad_norm": 1.2151773663575407, "learning_rate": 0.003, "loss": 4.0814, "step": 10986 }, { "epoch": 0.10987, "grad_norm": 1.1474346188515292, "learning_rate": 0.003, "loss": 4.0951, "step": 10987 }, { "epoch": 0.10988, "grad_norm": 1.2191252270848074, "learning_rate": 0.003, "loss": 4.0894, "step": 10988 }, { "epoch": 0.10989, "grad_norm": 1.0468863444800465, "learning_rate": 0.003, "loss": 4.0756, "step": 10989 }, { "epoch": 0.1099, "grad_norm": 1.102740386736203, "learning_rate": 0.003, "loss": 4.0746, "step": 10990 }, { "epoch": 0.10991, "grad_norm": 1.0709787382103662, "learning_rate": 0.003, "loss": 4.08, "step": 10991 }, { "epoch": 0.10992, "grad_norm": 1.3471933663495328, "learning_rate": 0.003, "loss": 4.0767, "step": 10992 }, { "epoch": 0.10993, "grad_norm": 0.8777327978040282, "learning_rate": 0.003, "loss": 4.0607, "step": 10993 }, { "epoch": 0.10994, "grad_norm": 1.0607706676678479, "learning_rate": 0.003, "loss": 4.1107, "step": 10994 }, { "epoch": 0.10995, "grad_norm": 1.2261575756169905, "learning_rate": 0.003, "loss": 4.0882, "step": 10995 }, { "epoch": 0.10996, "grad_norm": 1.0839466931853257, "learning_rate": 0.003, "loss": 4.0786, "step": 10996 }, { "epoch": 0.10997, "grad_norm": 1.200556428480252, "learning_rate": 0.003, "loss": 4.1083, "step": 10997 }, { "epoch": 0.10998, "grad_norm": 1.1891628270375623, "learning_rate": 0.003, "loss": 4.111, "step": 10998 }, { "epoch": 0.10999, "grad_norm": 1.1803068575793494, "learning_rate": 0.003, "loss": 4.0747, "step": 10999 }, { "epoch": 0.11, "grad_norm": 1.216378777563462, "learning_rate": 0.003, "loss": 4.1113, "step": 11000 }, { "epoch": 0.11001, "grad_norm": 1.289656496874306, "learning_rate": 0.003, "loss": 4.1076, "step": 11001 }, { "epoch": 0.11002, "grad_norm": 1.1492884131117849, "learning_rate": 0.003, "loss": 4.0889, "step": 11002 }, { "epoch": 0.11003, "grad_norm": 1.1083998300787332, "learning_rate": 0.003, "loss": 4.0807, "step": 11003 }, { "epoch": 0.11004, "grad_norm": 1.1328457610360267, "learning_rate": 0.003, "loss": 4.0806, "step": 11004 }, { "epoch": 0.11005, "grad_norm": 1.184919916402421, "learning_rate": 0.003, "loss": 4.0576, "step": 11005 }, { "epoch": 0.11006, "grad_norm": 1.2225122116520737, "learning_rate": 0.003, "loss": 4.0879, "step": 11006 }, { "epoch": 0.11007, "grad_norm": 1.2692158192366592, "learning_rate": 0.003, "loss": 4.077, "step": 11007 }, { "epoch": 0.11008, "grad_norm": 1.2185410602014857, "learning_rate": 0.003, "loss": 4.0907, "step": 11008 }, { "epoch": 0.11009, "grad_norm": 1.1291190266911624, "learning_rate": 0.003, "loss": 4.0611, "step": 11009 }, { "epoch": 0.1101, "grad_norm": 1.033167010262578, "learning_rate": 0.003, "loss": 4.037, "step": 11010 }, { "epoch": 0.11011, "grad_norm": 1.2126335789070828, "learning_rate": 0.003, "loss": 4.0647, "step": 11011 }, { "epoch": 0.11012, "grad_norm": 1.1834850607714638, "learning_rate": 0.003, "loss": 4.0671, "step": 11012 }, { "epoch": 0.11013, "grad_norm": 1.1749484121679226, "learning_rate": 0.003, "loss": 4.0831, "step": 11013 }, { "epoch": 0.11014, "grad_norm": 1.0223261088307614, "learning_rate": 0.003, "loss": 4.0797, "step": 11014 }, { "epoch": 0.11015, "grad_norm": 1.2933058857253439, "learning_rate": 0.003, "loss": 4.062, "step": 11015 }, { "epoch": 0.11016, "grad_norm": 1.06013065227872, "learning_rate": 0.003, "loss": 4.0762, "step": 11016 }, { "epoch": 0.11017, "grad_norm": 1.3486859594905953, "learning_rate": 0.003, "loss": 4.061, "step": 11017 }, { "epoch": 0.11018, "grad_norm": 0.986916211339272, "learning_rate": 0.003, "loss": 4.091, "step": 11018 }, { "epoch": 0.11019, "grad_norm": 1.167384153564792, "learning_rate": 0.003, "loss": 4.0754, "step": 11019 }, { "epoch": 0.1102, "grad_norm": 1.1052461201632215, "learning_rate": 0.003, "loss": 4.0497, "step": 11020 }, { "epoch": 0.11021, "grad_norm": 1.2800251869251464, "learning_rate": 0.003, "loss": 4.0598, "step": 11021 }, { "epoch": 0.11022, "grad_norm": 1.3272037366109815, "learning_rate": 0.003, "loss": 4.0921, "step": 11022 }, { "epoch": 0.11023, "grad_norm": 1.1668592711484271, "learning_rate": 0.003, "loss": 4.0947, "step": 11023 }, { "epoch": 0.11024, "grad_norm": 1.1585458235363237, "learning_rate": 0.003, "loss": 4.0958, "step": 11024 }, { "epoch": 0.11025, "grad_norm": 1.0635883194519844, "learning_rate": 0.003, "loss": 4.0742, "step": 11025 }, { "epoch": 0.11026, "grad_norm": 1.2126176125490782, "learning_rate": 0.003, "loss": 4.0738, "step": 11026 }, { "epoch": 0.11027, "grad_norm": 1.0696229205857228, "learning_rate": 0.003, "loss": 4.0592, "step": 11027 }, { "epoch": 0.11028, "grad_norm": 1.1450257642123574, "learning_rate": 0.003, "loss": 4.066, "step": 11028 }, { "epoch": 0.11029, "grad_norm": 1.126193649245872, "learning_rate": 0.003, "loss": 4.0571, "step": 11029 }, { "epoch": 0.1103, "grad_norm": 1.3229077788056023, "learning_rate": 0.003, "loss": 4.0811, "step": 11030 }, { "epoch": 0.11031, "grad_norm": 0.9020140652726835, "learning_rate": 0.003, "loss": 4.0703, "step": 11031 }, { "epoch": 0.11032, "grad_norm": 1.1991614113373614, "learning_rate": 0.003, "loss": 4.0904, "step": 11032 }, { "epoch": 0.11033, "grad_norm": 1.293323738809132, "learning_rate": 0.003, "loss": 4.0796, "step": 11033 }, { "epoch": 0.11034, "grad_norm": 0.9940537745625097, "learning_rate": 0.003, "loss": 4.0442, "step": 11034 }, { "epoch": 0.11035, "grad_norm": 1.1648199986703267, "learning_rate": 0.003, "loss": 4.0884, "step": 11035 }, { "epoch": 0.11036, "grad_norm": 1.2311792624685047, "learning_rate": 0.003, "loss": 4.1033, "step": 11036 }, { "epoch": 0.11037, "grad_norm": 1.031689829336888, "learning_rate": 0.003, "loss": 4.0454, "step": 11037 }, { "epoch": 0.11038, "grad_norm": 1.4327397680815341, "learning_rate": 0.003, "loss": 4.0793, "step": 11038 }, { "epoch": 0.11039, "grad_norm": 1.1444715457989634, "learning_rate": 0.003, "loss": 4.0791, "step": 11039 }, { "epoch": 0.1104, "grad_norm": 1.414163564427113, "learning_rate": 0.003, "loss": 4.1148, "step": 11040 }, { "epoch": 0.11041, "grad_norm": 1.2181604281798424, "learning_rate": 0.003, "loss": 4.0962, "step": 11041 }, { "epoch": 0.11042, "grad_norm": 1.049232482218127, "learning_rate": 0.003, "loss": 4.0986, "step": 11042 }, { "epoch": 0.11043, "grad_norm": 1.3071397803017475, "learning_rate": 0.003, "loss": 4.0884, "step": 11043 }, { "epoch": 0.11044, "grad_norm": 0.9255042719347787, "learning_rate": 0.003, "loss": 4.0743, "step": 11044 }, { "epoch": 0.11045, "grad_norm": 1.155005012874666, "learning_rate": 0.003, "loss": 4.066, "step": 11045 }, { "epoch": 0.11046, "grad_norm": 0.9592257000206171, "learning_rate": 0.003, "loss": 4.0602, "step": 11046 }, { "epoch": 0.11047, "grad_norm": 1.1407107616989076, "learning_rate": 0.003, "loss": 4.0582, "step": 11047 }, { "epoch": 0.11048, "grad_norm": 1.2702583077104692, "learning_rate": 0.003, "loss": 4.0796, "step": 11048 }, { "epoch": 0.11049, "grad_norm": 1.1664236578426006, "learning_rate": 0.003, "loss": 4.0682, "step": 11049 }, { "epoch": 0.1105, "grad_norm": 1.1022552350618229, "learning_rate": 0.003, "loss": 4.0723, "step": 11050 }, { "epoch": 0.11051, "grad_norm": 1.1980343626935672, "learning_rate": 0.003, "loss": 4.0895, "step": 11051 }, { "epoch": 0.11052, "grad_norm": 1.1563108049431163, "learning_rate": 0.003, "loss": 4.0866, "step": 11052 }, { "epoch": 0.11053, "grad_norm": 1.363084009888808, "learning_rate": 0.003, "loss": 4.081, "step": 11053 }, { "epoch": 0.11054, "grad_norm": 1.0707123910939973, "learning_rate": 0.003, "loss": 4.0745, "step": 11054 }, { "epoch": 0.11055, "grad_norm": 1.1497782594709365, "learning_rate": 0.003, "loss": 4.0773, "step": 11055 }, { "epoch": 0.11056, "grad_norm": 1.1540871060383255, "learning_rate": 0.003, "loss": 4.08, "step": 11056 }, { "epoch": 0.11057, "grad_norm": 1.0737054231011558, "learning_rate": 0.003, "loss": 4.058, "step": 11057 }, { "epoch": 0.11058, "grad_norm": 1.2385487272101865, "learning_rate": 0.003, "loss": 4.0967, "step": 11058 }, { "epoch": 0.11059, "grad_norm": 1.1050454322125165, "learning_rate": 0.003, "loss": 4.081, "step": 11059 }, { "epoch": 0.1106, "grad_norm": 1.3081187989296468, "learning_rate": 0.003, "loss": 4.0875, "step": 11060 }, { "epoch": 0.11061, "grad_norm": 0.97080304591906, "learning_rate": 0.003, "loss": 4.0918, "step": 11061 }, { "epoch": 0.11062, "grad_norm": 1.399622532788519, "learning_rate": 0.003, "loss": 4.0919, "step": 11062 }, { "epoch": 0.11063, "grad_norm": 1.0870843815603293, "learning_rate": 0.003, "loss": 4.0586, "step": 11063 }, { "epoch": 0.11064, "grad_norm": 1.1939902003866765, "learning_rate": 0.003, "loss": 4.0767, "step": 11064 }, { "epoch": 0.11065, "grad_norm": 1.1477894467567924, "learning_rate": 0.003, "loss": 4.062, "step": 11065 }, { "epoch": 0.11066, "grad_norm": 1.0221153810183439, "learning_rate": 0.003, "loss": 4.0902, "step": 11066 }, { "epoch": 0.11067, "grad_norm": 1.1756356134617003, "learning_rate": 0.003, "loss": 4.0968, "step": 11067 }, { "epoch": 0.11068, "grad_norm": 1.0162716761143804, "learning_rate": 0.003, "loss": 4.0682, "step": 11068 }, { "epoch": 0.11069, "grad_norm": 1.0810707049598247, "learning_rate": 0.003, "loss": 4.0581, "step": 11069 }, { "epoch": 0.1107, "grad_norm": 1.1967629587539852, "learning_rate": 0.003, "loss": 4.0877, "step": 11070 }, { "epoch": 0.11071, "grad_norm": 1.200254444855109, "learning_rate": 0.003, "loss": 4.0867, "step": 11071 }, { "epoch": 0.11072, "grad_norm": 1.132231184817727, "learning_rate": 0.003, "loss": 4.0656, "step": 11072 }, { "epoch": 0.11073, "grad_norm": 1.165720812170398, "learning_rate": 0.003, "loss": 4.1108, "step": 11073 }, { "epoch": 0.11074, "grad_norm": 1.4490821475131048, "learning_rate": 0.003, "loss": 4.0752, "step": 11074 }, { "epoch": 0.11075, "grad_norm": 1.2757756702169034, "learning_rate": 0.003, "loss": 4.1033, "step": 11075 }, { "epoch": 0.11076, "grad_norm": 0.9824431493913524, "learning_rate": 0.003, "loss": 4.0719, "step": 11076 }, { "epoch": 0.11077, "grad_norm": 1.4698769662624136, "learning_rate": 0.003, "loss": 4.1345, "step": 11077 }, { "epoch": 0.11078, "grad_norm": 1.0045970042994936, "learning_rate": 0.003, "loss": 4.0924, "step": 11078 }, { "epoch": 0.11079, "grad_norm": 1.3847834009536861, "learning_rate": 0.003, "loss": 4.081, "step": 11079 }, { "epoch": 0.1108, "grad_norm": 0.8797002169260058, "learning_rate": 0.003, "loss": 4.0688, "step": 11080 }, { "epoch": 0.11081, "grad_norm": 1.0901215929148171, "learning_rate": 0.003, "loss": 4.0867, "step": 11081 }, { "epoch": 0.11082, "grad_norm": 1.279624718766799, "learning_rate": 0.003, "loss": 4.0901, "step": 11082 }, { "epoch": 0.11083, "grad_norm": 1.167559863023754, "learning_rate": 0.003, "loss": 4.0746, "step": 11083 }, { "epoch": 0.11084, "grad_norm": 1.2821184130733287, "learning_rate": 0.003, "loss": 4.0731, "step": 11084 }, { "epoch": 0.11085, "grad_norm": 0.897330302931402, "learning_rate": 0.003, "loss": 4.0766, "step": 11085 }, { "epoch": 0.11086, "grad_norm": 1.0432451750288771, "learning_rate": 0.003, "loss": 4.1024, "step": 11086 }, { "epoch": 0.11087, "grad_norm": 1.022519793193149, "learning_rate": 0.003, "loss": 4.0816, "step": 11087 }, { "epoch": 0.11088, "grad_norm": 1.0535630318913338, "learning_rate": 0.003, "loss": 4.0781, "step": 11088 }, { "epoch": 0.11089, "grad_norm": 1.3968923566997862, "learning_rate": 0.003, "loss": 4.0798, "step": 11089 }, { "epoch": 0.1109, "grad_norm": 1.3674588309755793, "learning_rate": 0.003, "loss": 4.0624, "step": 11090 }, { "epoch": 0.11091, "grad_norm": 1.1591441198161583, "learning_rate": 0.003, "loss": 4.0687, "step": 11091 }, { "epoch": 0.11092, "grad_norm": 1.1480313147385102, "learning_rate": 0.003, "loss": 4.0739, "step": 11092 }, { "epoch": 0.11093, "grad_norm": 1.1701761236562274, "learning_rate": 0.003, "loss": 4.0848, "step": 11093 }, { "epoch": 0.11094, "grad_norm": 1.189369778108736, "learning_rate": 0.003, "loss": 4.0784, "step": 11094 }, { "epoch": 0.11095, "grad_norm": 1.295774630396128, "learning_rate": 0.003, "loss": 4.0974, "step": 11095 }, { "epoch": 0.11096, "grad_norm": 1.2094272273358757, "learning_rate": 0.003, "loss": 4.0913, "step": 11096 }, { "epoch": 0.11097, "grad_norm": 1.1580677336324876, "learning_rate": 0.003, "loss": 4.0937, "step": 11097 }, { "epoch": 0.11098, "grad_norm": 1.2363250718260648, "learning_rate": 0.003, "loss": 4.0745, "step": 11098 }, { "epoch": 0.11099, "grad_norm": 0.833144240622923, "learning_rate": 0.003, "loss": 4.0978, "step": 11099 }, { "epoch": 0.111, "grad_norm": 1.053843619719335, "learning_rate": 0.003, "loss": 4.062, "step": 11100 }, { "epoch": 0.11101, "grad_norm": 1.146692066640518, "learning_rate": 0.003, "loss": 4.0856, "step": 11101 }, { "epoch": 0.11102, "grad_norm": 1.1457348265065703, "learning_rate": 0.003, "loss": 4.0717, "step": 11102 }, { "epoch": 0.11103, "grad_norm": 1.3834421526344465, "learning_rate": 0.003, "loss": 4.078, "step": 11103 }, { "epoch": 0.11104, "grad_norm": 0.8438504572465533, "learning_rate": 0.003, "loss": 4.0665, "step": 11104 }, { "epoch": 0.11105, "grad_norm": 0.922149256979776, "learning_rate": 0.003, "loss": 4.0687, "step": 11105 }, { "epoch": 0.11106, "grad_norm": 1.259771858112156, "learning_rate": 0.003, "loss": 4.0681, "step": 11106 }, { "epoch": 0.11107, "grad_norm": 1.0197544280538555, "learning_rate": 0.003, "loss": 4.0656, "step": 11107 }, { "epoch": 0.11108, "grad_norm": 1.274947463280985, "learning_rate": 0.003, "loss": 4.0957, "step": 11108 }, { "epoch": 0.11109, "grad_norm": 1.3399035789882239, "learning_rate": 0.003, "loss": 4.0955, "step": 11109 }, { "epoch": 0.1111, "grad_norm": 1.09604148282697, "learning_rate": 0.003, "loss": 4.0875, "step": 11110 }, { "epoch": 0.11111, "grad_norm": 1.3240783623002959, "learning_rate": 0.003, "loss": 4.0885, "step": 11111 }, { "epoch": 0.11112, "grad_norm": 0.9601801261401435, "learning_rate": 0.003, "loss": 4.0999, "step": 11112 }, { "epoch": 0.11113, "grad_norm": 1.3202629173276619, "learning_rate": 0.003, "loss": 4.1021, "step": 11113 }, { "epoch": 0.11114, "grad_norm": 1.027638567450629, "learning_rate": 0.003, "loss": 4.1166, "step": 11114 }, { "epoch": 0.11115, "grad_norm": 1.2143442400495823, "learning_rate": 0.003, "loss": 4.0906, "step": 11115 }, { "epoch": 0.11116, "grad_norm": 1.104943840701868, "learning_rate": 0.003, "loss": 4.0705, "step": 11116 }, { "epoch": 0.11117, "grad_norm": 1.1174412623468781, "learning_rate": 0.003, "loss": 4.0986, "step": 11117 }, { "epoch": 0.11118, "grad_norm": 1.336239897944748, "learning_rate": 0.003, "loss": 4.0718, "step": 11118 }, { "epoch": 0.11119, "grad_norm": 1.2056116868391393, "learning_rate": 0.003, "loss": 4.0982, "step": 11119 }, { "epoch": 0.1112, "grad_norm": 1.2520239209685728, "learning_rate": 0.003, "loss": 4.0735, "step": 11120 }, { "epoch": 0.11121, "grad_norm": 1.256131905227129, "learning_rate": 0.003, "loss": 4.0585, "step": 11121 }, { "epoch": 0.11122, "grad_norm": 1.0098916379815241, "learning_rate": 0.003, "loss": 4.0936, "step": 11122 }, { "epoch": 0.11123, "grad_norm": 1.3292004460141258, "learning_rate": 0.003, "loss": 4.101, "step": 11123 }, { "epoch": 0.11124, "grad_norm": 0.9452229243493425, "learning_rate": 0.003, "loss": 4.063, "step": 11124 }, { "epoch": 0.11125, "grad_norm": 1.6146840509394205, "learning_rate": 0.003, "loss": 4.101, "step": 11125 }, { "epoch": 0.11126, "grad_norm": 0.9071079822837661, "learning_rate": 0.003, "loss": 4.0726, "step": 11126 }, { "epoch": 0.11127, "grad_norm": 1.1078521847315095, "learning_rate": 0.003, "loss": 4.0842, "step": 11127 }, { "epoch": 0.11128, "grad_norm": 1.159721875586419, "learning_rate": 0.003, "loss": 4.0592, "step": 11128 }, { "epoch": 0.11129, "grad_norm": 1.0396207654106058, "learning_rate": 0.003, "loss": 4.0677, "step": 11129 }, { "epoch": 0.1113, "grad_norm": 1.1290974026996876, "learning_rate": 0.003, "loss": 4.0493, "step": 11130 }, { "epoch": 0.11131, "grad_norm": 1.07403026241833, "learning_rate": 0.003, "loss": 4.0774, "step": 11131 }, { "epoch": 0.11132, "grad_norm": 1.3933698121617073, "learning_rate": 0.003, "loss": 4.0795, "step": 11132 }, { "epoch": 0.11133, "grad_norm": 1.1725955117249347, "learning_rate": 0.003, "loss": 4.0636, "step": 11133 }, { "epoch": 0.11134, "grad_norm": 1.3174881140272372, "learning_rate": 0.003, "loss": 4.0575, "step": 11134 }, { "epoch": 0.11135, "grad_norm": 1.1736131979152669, "learning_rate": 0.003, "loss": 4.0801, "step": 11135 }, { "epoch": 0.11136, "grad_norm": 1.048053300051631, "learning_rate": 0.003, "loss": 4.0481, "step": 11136 }, { "epoch": 0.11137, "grad_norm": 1.102288655476633, "learning_rate": 0.003, "loss": 4.0735, "step": 11137 }, { "epoch": 0.11138, "grad_norm": 1.108742784150663, "learning_rate": 0.003, "loss": 4.058, "step": 11138 }, { "epoch": 0.11139, "grad_norm": 1.173686326578977, "learning_rate": 0.003, "loss": 4.0936, "step": 11139 }, { "epoch": 0.1114, "grad_norm": 1.2774571058823576, "learning_rate": 0.003, "loss": 4.0606, "step": 11140 }, { "epoch": 0.11141, "grad_norm": 1.0850920923430565, "learning_rate": 0.003, "loss": 4.0827, "step": 11141 }, { "epoch": 0.11142, "grad_norm": 1.2722820226326228, "learning_rate": 0.003, "loss": 4.0839, "step": 11142 }, { "epoch": 0.11143, "grad_norm": 0.9960056462520913, "learning_rate": 0.003, "loss": 4.0717, "step": 11143 }, { "epoch": 0.11144, "grad_norm": 1.1203106074026263, "learning_rate": 0.003, "loss": 4.0495, "step": 11144 }, { "epoch": 0.11145, "grad_norm": 1.1966501459479226, "learning_rate": 0.003, "loss": 4.0874, "step": 11145 }, { "epoch": 0.11146, "grad_norm": 1.1984315813802482, "learning_rate": 0.003, "loss": 4.0818, "step": 11146 }, { "epoch": 0.11147, "grad_norm": 1.0764740147161755, "learning_rate": 0.003, "loss": 4.0983, "step": 11147 }, { "epoch": 0.11148, "grad_norm": 1.3121591191772344, "learning_rate": 0.003, "loss": 4.0837, "step": 11148 }, { "epoch": 0.11149, "grad_norm": 1.1646342972823798, "learning_rate": 0.003, "loss": 4.0608, "step": 11149 }, { "epoch": 0.1115, "grad_norm": 1.331730803049615, "learning_rate": 0.003, "loss": 4.0742, "step": 11150 }, { "epoch": 0.11151, "grad_norm": 0.9860918080871273, "learning_rate": 0.003, "loss": 4.0952, "step": 11151 }, { "epoch": 0.11152, "grad_norm": 1.2815747677037665, "learning_rate": 0.003, "loss": 4.066, "step": 11152 }, { "epoch": 0.11153, "grad_norm": 1.0277548994206205, "learning_rate": 0.003, "loss": 4.0894, "step": 11153 }, { "epoch": 0.11154, "grad_norm": 1.2541170294769153, "learning_rate": 0.003, "loss": 4.0879, "step": 11154 }, { "epoch": 0.11155, "grad_norm": 0.9600444972450275, "learning_rate": 0.003, "loss": 4.0922, "step": 11155 }, { "epoch": 0.11156, "grad_norm": 1.2549006486500096, "learning_rate": 0.003, "loss": 4.0901, "step": 11156 }, { "epoch": 0.11157, "grad_norm": 1.045895721445441, "learning_rate": 0.003, "loss": 4.0933, "step": 11157 }, { "epoch": 0.11158, "grad_norm": 1.281534872473627, "learning_rate": 0.003, "loss": 4.0982, "step": 11158 }, { "epoch": 0.11159, "grad_norm": 0.9835671110843429, "learning_rate": 0.003, "loss": 4.0838, "step": 11159 }, { "epoch": 0.1116, "grad_norm": 1.3993735657738384, "learning_rate": 0.003, "loss": 4.0979, "step": 11160 }, { "epoch": 0.11161, "grad_norm": 1.1542062042669807, "learning_rate": 0.003, "loss": 4.0658, "step": 11161 }, { "epoch": 0.11162, "grad_norm": 1.2233265143236163, "learning_rate": 0.003, "loss": 4.0802, "step": 11162 }, { "epoch": 0.11163, "grad_norm": 1.1421093108971783, "learning_rate": 0.003, "loss": 4.0788, "step": 11163 }, { "epoch": 0.11164, "grad_norm": 1.0930501156203938, "learning_rate": 0.003, "loss": 4.0836, "step": 11164 }, { "epoch": 0.11165, "grad_norm": 1.1391165974539779, "learning_rate": 0.003, "loss": 4.0825, "step": 11165 }, { "epoch": 0.11166, "grad_norm": 1.1192418056700695, "learning_rate": 0.003, "loss": 4.0851, "step": 11166 }, { "epoch": 0.11167, "grad_norm": 1.3010174658966451, "learning_rate": 0.003, "loss": 4.0883, "step": 11167 }, { "epoch": 0.11168, "grad_norm": 1.3028159385520257, "learning_rate": 0.003, "loss": 4.0756, "step": 11168 }, { "epoch": 0.11169, "grad_norm": 0.9401279630622993, "learning_rate": 0.003, "loss": 4.0713, "step": 11169 }, { "epoch": 0.1117, "grad_norm": 0.9373861405274122, "learning_rate": 0.003, "loss": 4.0796, "step": 11170 }, { "epoch": 0.11171, "grad_norm": 1.0889118677948568, "learning_rate": 0.003, "loss": 4.0638, "step": 11171 }, { "epoch": 0.11172, "grad_norm": 1.1942612460598918, "learning_rate": 0.003, "loss": 4.065, "step": 11172 }, { "epoch": 0.11173, "grad_norm": 1.098306651564804, "learning_rate": 0.003, "loss": 4.0864, "step": 11173 }, { "epoch": 0.11174, "grad_norm": 1.3399889218813388, "learning_rate": 0.003, "loss": 4.0959, "step": 11174 }, { "epoch": 0.11175, "grad_norm": 1.1531186410717533, "learning_rate": 0.003, "loss": 4.0916, "step": 11175 }, { "epoch": 0.11176, "grad_norm": 1.209712286758474, "learning_rate": 0.003, "loss": 4.0589, "step": 11176 }, { "epoch": 0.11177, "grad_norm": 0.9782456245193254, "learning_rate": 0.003, "loss": 4.0998, "step": 11177 }, { "epoch": 0.11178, "grad_norm": 1.1558283929711195, "learning_rate": 0.003, "loss": 4.0934, "step": 11178 }, { "epoch": 0.11179, "grad_norm": 1.2772922260185524, "learning_rate": 0.003, "loss": 4.0755, "step": 11179 }, { "epoch": 0.1118, "grad_norm": 0.9989572762152379, "learning_rate": 0.003, "loss": 4.078, "step": 11180 }, { "epoch": 0.11181, "grad_norm": 1.2637439132331842, "learning_rate": 0.003, "loss": 4.0925, "step": 11181 }, { "epoch": 0.11182, "grad_norm": 1.2297280930942402, "learning_rate": 0.003, "loss": 4.079, "step": 11182 }, { "epoch": 0.11183, "grad_norm": 1.03577064834795, "learning_rate": 0.003, "loss": 4.0831, "step": 11183 }, { "epoch": 0.11184, "grad_norm": 1.2063110896089673, "learning_rate": 0.003, "loss": 4.0822, "step": 11184 }, { "epoch": 0.11185, "grad_norm": 0.9773849831210153, "learning_rate": 0.003, "loss": 4.0779, "step": 11185 }, { "epoch": 0.11186, "grad_norm": 1.18402790441611, "learning_rate": 0.003, "loss": 4.0626, "step": 11186 }, { "epoch": 0.11187, "grad_norm": 1.4741121362262628, "learning_rate": 0.003, "loss": 4.0535, "step": 11187 }, { "epoch": 0.11188, "grad_norm": 1.2388308800713173, "learning_rate": 0.003, "loss": 4.0712, "step": 11188 }, { "epoch": 0.11189, "grad_norm": 1.2911692078858614, "learning_rate": 0.003, "loss": 4.1121, "step": 11189 }, { "epoch": 0.1119, "grad_norm": 1.0218562441800818, "learning_rate": 0.003, "loss": 4.1037, "step": 11190 }, { "epoch": 0.11191, "grad_norm": 1.2931489036686594, "learning_rate": 0.003, "loss": 4.0704, "step": 11191 }, { "epoch": 0.11192, "grad_norm": 1.0861872408771538, "learning_rate": 0.003, "loss": 4.1046, "step": 11192 }, { "epoch": 0.11193, "grad_norm": 1.114209549974762, "learning_rate": 0.003, "loss": 4.0775, "step": 11193 }, { "epoch": 0.11194, "grad_norm": 1.209019727027277, "learning_rate": 0.003, "loss": 4.1123, "step": 11194 }, { "epoch": 0.11195, "grad_norm": 1.0108126210804838, "learning_rate": 0.003, "loss": 4.0638, "step": 11195 }, { "epoch": 0.11196, "grad_norm": 1.20589123068435, "learning_rate": 0.003, "loss": 4.0785, "step": 11196 }, { "epoch": 0.11197, "grad_norm": 1.0243731523960853, "learning_rate": 0.003, "loss": 4.0716, "step": 11197 }, { "epoch": 0.11198, "grad_norm": 1.405136226654536, "learning_rate": 0.003, "loss": 4.0772, "step": 11198 }, { "epoch": 0.11199, "grad_norm": 0.9997432541113855, "learning_rate": 0.003, "loss": 4.077, "step": 11199 }, { "epoch": 0.112, "grad_norm": 1.2367199253922476, "learning_rate": 0.003, "loss": 4.0506, "step": 11200 }, { "epoch": 0.11201, "grad_norm": 1.0631224885717738, "learning_rate": 0.003, "loss": 4.0753, "step": 11201 }, { "epoch": 0.11202, "grad_norm": 1.1185478468019074, "learning_rate": 0.003, "loss": 4.0811, "step": 11202 }, { "epoch": 0.11203, "grad_norm": 1.1490403919062497, "learning_rate": 0.003, "loss": 4.072, "step": 11203 }, { "epoch": 0.11204, "grad_norm": 0.9717178345845396, "learning_rate": 0.003, "loss": 4.0555, "step": 11204 }, { "epoch": 0.11205, "grad_norm": 1.1687038083031553, "learning_rate": 0.003, "loss": 4.0787, "step": 11205 }, { "epoch": 0.11206, "grad_norm": 1.073448923677053, "learning_rate": 0.003, "loss": 4.0934, "step": 11206 }, { "epoch": 0.11207, "grad_norm": 1.2772128792027855, "learning_rate": 0.003, "loss": 4.0722, "step": 11207 }, { "epoch": 0.11208, "grad_norm": 1.004257416373169, "learning_rate": 0.003, "loss": 4.084, "step": 11208 }, { "epoch": 0.11209, "grad_norm": 1.414543278622611, "learning_rate": 0.003, "loss": 4.0769, "step": 11209 }, { "epoch": 0.1121, "grad_norm": 1.1533718420414074, "learning_rate": 0.003, "loss": 4.0634, "step": 11210 }, { "epoch": 0.11211, "grad_norm": 1.232881271126248, "learning_rate": 0.003, "loss": 4.0986, "step": 11211 }, { "epoch": 0.11212, "grad_norm": 1.3454643836906734, "learning_rate": 0.003, "loss": 4.0862, "step": 11212 }, { "epoch": 0.11213, "grad_norm": 1.1964442622903273, "learning_rate": 0.003, "loss": 4.0841, "step": 11213 }, { "epoch": 0.11214, "grad_norm": 1.016582264231967, "learning_rate": 0.003, "loss": 4.1037, "step": 11214 }, { "epoch": 0.11215, "grad_norm": 1.2465471638780516, "learning_rate": 0.003, "loss": 4.113, "step": 11215 }, { "epoch": 0.11216, "grad_norm": 0.9731189335286383, "learning_rate": 0.003, "loss": 4.0645, "step": 11216 }, { "epoch": 0.11217, "grad_norm": 1.2585237890766032, "learning_rate": 0.003, "loss": 4.0821, "step": 11217 }, { "epoch": 0.11218, "grad_norm": 1.375187414405389, "learning_rate": 0.003, "loss": 4.0847, "step": 11218 }, { "epoch": 0.11219, "grad_norm": 1.21500013057869, "learning_rate": 0.003, "loss": 4.0823, "step": 11219 }, { "epoch": 0.1122, "grad_norm": 1.0237848543455739, "learning_rate": 0.003, "loss": 4.0413, "step": 11220 }, { "epoch": 0.11221, "grad_norm": 1.0314896825666124, "learning_rate": 0.003, "loss": 4.0768, "step": 11221 }, { "epoch": 0.11222, "grad_norm": 1.3233867169080658, "learning_rate": 0.003, "loss": 4.1044, "step": 11222 }, { "epoch": 0.11223, "grad_norm": 1.0975586830375574, "learning_rate": 0.003, "loss": 4.0697, "step": 11223 }, { "epoch": 0.11224, "grad_norm": 1.1930854617321562, "learning_rate": 0.003, "loss": 4.0875, "step": 11224 }, { "epoch": 0.11225, "grad_norm": 1.0911558326089497, "learning_rate": 0.003, "loss": 4.102, "step": 11225 }, { "epoch": 0.11226, "grad_norm": 1.2351625865634561, "learning_rate": 0.003, "loss": 4.0921, "step": 11226 }, { "epoch": 0.11227, "grad_norm": 1.1749371871274827, "learning_rate": 0.003, "loss": 4.0674, "step": 11227 }, { "epoch": 0.11228, "grad_norm": 1.198635523522791, "learning_rate": 0.003, "loss": 4.0796, "step": 11228 }, { "epoch": 0.11229, "grad_norm": 1.1240002402811142, "learning_rate": 0.003, "loss": 4.0788, "step": 11229 }, { "epoch": 0.1123, "grad_norm": 1.2241592922697206, "learning_rate": 0.003, "loss": 4.0851, "step": 11230 }, { "epoch": 0.11231, "grad_norm": 1.1970424174019643, "learning_rate": 0.003, "loss": 4.0741, "step": 11231 }, { "epoch": 0.11232, "grad_norm": 1.1421340748635413, "learning_rate": 0.003, "loss": 4.0665, "step": 11232 }, { "epoch": 0.11233, "grad_norm": 1.1759250032252293, "learning_rate": 0.003, "loss": 4.0861, "step": 11233 }, { "epoch": 0.11234, "grad_norm": 1.1609578732577739, "learning_rate": 0.003, "loss": 4.0714, "step": 11234 }, { "epoch": 0.11235, "grad_norm": 1.1821129395196315, "learning_rate": 0.003, "loss": 4.0815, "step": 11235 }, { "epoch": 0.11236, "grad_norm": 1.1070416228627675, "learning_rate": 0.003, "loss": 4.0923, "step": 11236 }, { "epoch": 0.11237, "grad_norm": 1.3664160052213254, "learning_rate": 0.003, "loss": 4.0847, "step": 11237 }, { "epoch": 0.11238, "grad_norm": 0.9119492467214586, "learning_rate": 0.003, "loss": 4.0562, "step": 11238 }, { "epoch": 0.11239, "grad_norm": 1.0622389881528407, "learning_rate": 0.003, "loss": 4.0815, "step": 11239 }, { "epoch": 0.1124, "grad_norm": 1.1894727674355798, "learning_rate": 0.003, "loss": 4.0818, "step": 11240 }, { "epoch": 0.11241, "grad_norm": 0.9537664095570878, "learning_rate": 0.003, "loss": 4.0795, "step": 11241 }, { "epoch": 0.11242, "grad_norm": 1.2108180500354753, "learning_rate": 0.003, "loss": 4.0831, "step": 11242 }, { "epoch": 0.11243, "grad_norm": 1.0377598646551163, "learning_rate": 0.003, "loss": 4.0785, "step": 11243 }, { "epoch": 0.11244, "grad_norm": 1.2667536423936097, "learning_rate": 0.003, "loss": 4.0763, "step": 11244 }, { "epoch": 0.11245, "grad_norm": 1.0199261735712357, "learning_rate": 0.003, "loss": 4.0922, "step": 11245 }, { "epoch": 0.11246, "grad_norm": 1.380042950316513, "learning_rate": 0.003, "loss": 4.0769, "step": 11246 }, { "epoch": 0.11247, "grad_norm": 1.2233265055955735, "learning_rate": 0.003, "loss": 4.0586, "step": 11247 }, { "epoch": 0.11248, "grad_norm": 1.177520883310014, "learning_rate": 0.003, "loss": 4.1064, "step": 11248 }, { "epoch": 0.11249, "grad_norm": 1.0352426374271515, "learning_rate": 0.003, "loss": 4.0869, "step": 11249 }, { "epoch": 0.1125, "grad_norm": 1.2681183552927422, "learning_rate": 0.003, "loss": 4.0671, "step": 11250 }, { "epoch": 0.11251, "grad_norm": 1.1477755872427362, "learning_rate": 0.003, "loss": 4.0668, "step": 11251 }, { "epoch": 0.11252, "grad_norm": 1.517706889182652, "learning_rate": 0.003, "loss": 4.1032, "step": 11252 }, { "epoch": 0.11253, "grad_norm": 1.1177213037118356, "learning_rate": 0.003, "loss": 4.0839, "step": 11253 }, { "epoch": 0.11254, "grad_norm": 1.3343373703779187, "learning_rate": 0.003, "loss": 4.0811, "step": 11254 }, { "epoch": 0.11255, "grad_norm": 1.1567401753985436, "learning_rate": 0.003, "loss": 4.0898, "step": 11255 }, { "epoch": 0.11256, "grad_norm": 1.0806058389994764, "learning_rate": 0.003, "loss": 4.0834, "step": 11256 }, { "epoch": 0.11257, "grad_norm": 1.133158277986818, "learning_rate": 0.003, "loss": 4.0735, "step": 11257 }, { "epoch": 0.11258, "grad_norm": 1.1883695225812174, "learning_rate": 0.003, "loss": 4.0758, "step": 11258 }, { "epoch": 0.11259, "grad_norm": 0.9681430811253013, "learning_rate": 0.003, "loss": 4.1043, "step": 11259 }, { "epoch": 0.1126, "grad_norm": 1.1934967669263257, "learning_rate": 0.003, "loss": 4.0616, "step": 11260 }, { "epoch": 0.11261, "grad_norm": 1.085936306228125, "learning_rate": 0.003, "loss": 4.07, "step": 11261 }, { "epoch": 0.11262, "grad_norm": 1.28556430915771, "learning_rate": 0.003, "loss": 4.0867, "step": 11262 }, { "epoch": 0.11263, "grad_norm": 1.0671406124670375, "learning_rate": 0.003, "loss": 4.0622, "step": 11263 }, { "epoch": 0.11264, "grad_norm": 1.261192487487167, "learning_rate": 0.003, "loss": 4.0753, "step": 11264 }, { "epoch": 0.11265, "grad_norm": 1.0739516983789419, "learning_rate": 0.003, "loss": 4.0516, "step": 11265 }, { "epoch": 0.11266, "grad_norm": 1.2903709025877286, "learning_rate": 0.003, "loss": 4.0925, "step": 11266 }, { "epoch": 0.11267, "grad_norm": 1.1865966885308028, "learning_rate": 0.003, "loss": 4.1016, "step": 11267 }, { "epoch": 0.11268, "grad_norm": 1.2393487002892085, "learning_rate": 0.003, "loss": 4.0713, "step": 11268 }, { "epoch": 0.11269, "grad_norm": 1.131721357284378, "learning_rate": 0.003, "loss": 4.0507, "step": 11269 }, { "epoch": 0.1127, "grad_norm": 1.0623045335870454, "learning_rate": 0.003, "loss": 4.075, "step": 11270 }, { "epoch": 0.11271, "grad_norm": 1.298629433282519, "learning_rate": 0.003, "loss": 4.0864, "step": 11271 }, { "epoch": 0.11272, "grad_norm": 1.1024801052028652, "learning_rate": 0.003, "loss": 4.0664, "step": 11272 }, { "epoch": 0.11273, "grad_norm": 1.1799581337189502, "learning_rate": 0.003, "loss": 4.0908, "step": 11273 }, { "epoch": 0.11274, "grad_norm": 1.0744504496225524, "learning_rate": 0.003, "loss": 4.0707, "step": 11274 }, { "epoch": 0.11275, "grad_norm": 1.2446994109409262, "learning_rate": 0.003, "loss": 4.0458, "step": 11275 }, { "epoch": 0.11276, "grad_norm": 1.0883069074518403, "learning_rate": 0.003, "loss": 4.0826, "step": 11276 }, { "epoch": 0.11277, "grad_norm": 1.1088581538356497, "learning_rate": 0.003, "loss": 4.0693, "step": 11277 }, { "epoch": 0.11278, "grad_norm": 1.1320400571531242, "learning_rate": 0.003, "loss": 4.0972, "step": 11278 }, { "epoch": 0.11279, "grad_norm": 1.1882710772080816, "learning_rate": 0.003, "loss": 4.0673, "step": 11279 }, { "epoch": 0.1128, "grad_norm": 1.1828816974011485, "learning_rate": 0.003, "loss": 4.0841, "step": 11280 }, { "epoch": 0.11281, "grad_norm": 1.2189576526947539, "learning_rate": 0.003, "loss": 4.0847, "step": 11281 }, { "epoch": 0.11282, "grad_norm": 1.0520825905755233, "learning_rate": 0.003, "loss": 4.0723, "step": 11282 }, { "epoch": 0.11283, "grad_norm": 1.2397599842533094, "learning_rate": 0.003, "loss": 4.0856, "step": 11283 }, { "epoch": 0.11284, "grad_norm": 1.025617880523038, "learning_rate": 0.003, "loss": 4.061, "step": 11284 }, { "epoch": 0.11285, "grad_norm": 1.1821805898447082, "learning_rate": 0.003, "loss": 4.0776, "step": 11285 }, { "epoch": 0.11286, "grad_norm": 1.1165620824376317, "learning_rate": 0.003, "loss": 4.0869, "step": 11286 }, { "epoch": 0.11287, "grad_norm": 1.1072689494019334, "learning_rate": 0.003, "loss": 4.0615, "step": 11287 }, { "epoch": 0.11288, "grad_norm": 1.2825093591266885, "learning_rate": 0.003, "loss": 4.0533, "step": 11288 }, { "epoch": 0.11289, "grad_norm": 1.3094637479991706, "learning_rate": 0.003, "loss": 4.1014, "step": 11289 }, { "epoch": 0.1129, "grad_norm": 1.2293468957577327, "learning_rate": 0.003, "loss": 4.0647, "step": 11290 }, { "epoch": 0.11291, "grad_norm": 1.3881096727893352, "learning_rate": 0.003, "loss": 4.1009, "step": 11291 }, { "epoch": 0.11292, "grad_norm": 1.1828219431674265, "learning_rate": 0.003, "loss": 4.0792, "step": 11292 }, { "epoch": 0.11293, "grad_norm": 1.0432111324738476, "learning_rate": 0.003, "loss": 4.0967, "step": 11293 }, { "epoch": 0.11294, "grad_norm": 1.2028521119807711, "learning_rate": 0.003, "loss": 4.0503, "step": 11294 }, { "epoch": 0.11295, "grad_norm": 1.1560410520532847, "learning_rate": 0.003, "loss": 4.082, "step": 11295 }, { "epoch": 0.11296, "grad_norm": 1.241080361905138, "learning_rate": 0.003, "loss": 4.0624, "step": 11296 }, { "epoch": 0.11297, "grad_norm": 1.102149499553362, "learning_rate": 0.003, "loss": 4.0999, "step": 11297 }, { "epoch": 0.11298, "grad_norm": 1.4981514137553442, "learning_rate": 0.003, "loss": 4.0897, "step": 11298 }, { "epoch": 0.11299, "grad_norm": 1.2732016237261758, "learning_rate": 0.003, "loss": 4.0779, "step": 11299 }, { "epoch": 0.113, "grad_norm": 1.0855217853652195, "learning_rate": 0.003, "loss": 4.0966, "step": 11300 }, { "epoch": 0.11301, "grad_norm": 1.144721962347034, "learning_rate": 0.003, "loss": 4.0628, "step": 11301 }, { "epoch": 0.11302, "grad_norm": 1.163283114829239, "learning_rate": 0.003, "loss": 4.0703, "step": 11302 }, { "epoch": 0.11303, "grad_norm": 1.5095251442036801, "learning_rate": 0.003, "loss": 4.0825, "step": 11303 }, { "epoch": 0.11304, "grad_norm": 0.8655844076550182, "learning_rate": 0.003, "loss": 4.0793, "step": 11304 }, { "epoch": 0.11305, "grad_norm": 0.9768503913687301, "learning_rate": 0.003, "loss": 4.0559, "step": 11305 }, { "epoch": 0.11306, "grad_norm": 1.1118481212396893, "learning_rate": 0.003, "loss": 4.0568, "step": 11306 }, { "epoch": 0.11307, "grad_norm": 1.2557851707694494, "learning_rate": 0.003, "loss": 4.0814, "step": 11307 }, { "epoch": 0.11308, "grad_norm": 1.0876765492352787, "learning_rate": 0.003, "loss": 4.072, "step": 11308 }, { "epoch": 0.11309, "grad_norm": 1.3197206224683027, "learning_rate": 0.003, "loss": 4.0844, "step": 11309 }, { "epoch": 0.1131, "grad_norm": 1.1527574893555428, "learning_rate": 0.003, "loss": 4.0566, "step": 11310 }, { "epoch": 0.11311, "grad_norm": 1.0055020429352322, "learning_rate": 0.003, "loss": 4.0786, "step": 11311 }, { "epoch": 0.11312, "grad_norm": 1.2278396206335767, "learning_rate": 0.003, "loss": 4.0831, "step": 11312 }, { "epoch": 0.11313, "grad_norm": 1.0076037063843581, "learning_rate": 0.003, "loss": 4.0595, "step": 11313 }, { "epoch": 0.11314, "grad_norm": 1.4336445287934063, "learning_rate": 0.003, "loss": 4.0987, "step": 11314 }, { "epoch": 0.11315, "grad_norm": 0.8419355428055376, "learning_rate": 0.003, "loss": 4.0798, "step": 11315 }, { "epoch": 0.11316, "grad_norm": 0.9841916247201852, "learning_rate": 0.003, "loss": 4.0586, "step": 11316 }, { "epoch": 0.11317, "grad_norm": 1.3680597035869113, "learning_rate": 0.003, "loss": 4.0894, "step": 11317 }, { "epoch": 0.11318, "grad_norm": 1.393252247457501, "learning_rate": 0.003, "loss": 4.0954, "step": 11318 }, { "epoch": 0.11319, "grad_norm": 1.2698890335191317, "learning_rate": 0.003, "loss": 4.0904, "step": 11319 }, { "epoch": 0.1132, "grad_norm": 1.054983155096937, "learning_rate": 0.003, "loss": 4.0814, "step": 11320 }, { "epoch": 0.11321, "grad_norm": 1.1642598738916654, "learning_rate": 0.003, "loss": 4.0479, "step": 11321 }, { "epoch": 0.11322, "grad_norm": 1.2880074580030998, "learning_rate": 0.003, "loss": 4.0877, "step": 11322 }, { "epoch": 0.11323, "grad_norm": 0.9696344981062905, "learning_rate": 0.003, "loss": 4.0347, "step": 11323 }, { "epoch": 0.11324, "grad_norm": 1.2317118080416707, "learning_rate": 0.003, "loss": 4.0759, "step": 11324 }, { "epoch": 0.11325, "grad_norm": 1.0240155450785555, "learning_rate": 0.003, "loss": 4.0923, "step": 11325 }, { "epoch": 0.11326, "grad_norm": 1.1909002991650128, "learning_rate": 0.003, "loss": 4.0881, "step": 11326 }, { "epoch": 0.11327, "grad_norm": 1.1363333235538478, "learning_rate": 0.003, "loss": 4.0877, "step": 11327 }, { "epoch": 0.11328, "grad_norm": 1.153386328873901, "learning_rate": 0.003, "loss": 4.0791, "step": 11328 }, { "epoch": 0.11329, "grad_norm": 1.1843625290435011, "learning_rate": 0.003, "loss": 4.0812, "step": 11329 }, { "epoch": 0.1133, "grad_norm": 1.1713278236704114, "learning_rate": 0.003, "loss": 4.0811, "step": 11330 }, { "epoch": 0.11331, "grad_norm": 1.4738186710191181, "learning_rate": 0.003, "loss": 4.0711, "step": 11331 }, { "epoch": 0.11332, "grad_norm": 1.1168929903796028, "learning_rate": 0.003, "loss": 4.0986, "step": 11332 }, { "epoch": 0.11333, "grad_norm": 1.4447556787111548, "learning_rate": 0.003, "loss": 4.0417, "step": 11333 }, { "epoch": 0.11334, "grad_norm": 1.0926804158099357, "learning_rate": 0.003, "loss": 4.0671, "step": 11334 }, { "epoch": 0.11335, "grad_norm": 1.2927439581888547, "learning_rate": 0.003, "loss": 4.0629, "step": 11335 }, { "epoch": 0.11336, "grad_norm": 1.077028010839177, "learning_rate": 0.003, "loss": 4.0589, "step": 11336 }, { "epoch": 0.11337, "grad_norm": 1.2918345218299065, "learning_rate": 0.003, "loss": 4.0723, "step": 11337 }, { "epoch": 0.11338, "grad_norm": 1.0598579343301027, "learning_rate": 0.003, "loss": 4.0843, "step": 11338 }, { "epoch": 0.11339, "grad_norm": 1.1257298398963314, "learning_rate": 0.003, "loss": 4.1108, "step": 11339 }, { "epoch": 0.1134, "grad_norm": 1.3162052695908748, "learning_rate": 0.003, "loss": 4.083, "step": 11340 }, { "epoch": 0.11341, "grad_norm": 1.131065545796142, "learning_rate": 0.003, "loss": 4.0642, "step": 11341 }, { "epoch": 0.11342, "grad_norm": 1.1871289356923436, "learning_rate": 0.003, "loss": 4.0633, "step": 11342 }, { "epoch": 0.11343, "grad_norm": 1.2208326094667281, "learning_rate": 0.003, "loss": 4.0834, "step": 11343 }, { "epoch": 0.11344, "grad_norm": 1.147121675994465, "learning_rate": 0.003, "loss": 4.0918, "step": 11344 }, { "epoch": 0.11345, "grad_norm": 1.110093784860901, "learning_rate": 0.003, "loss": 4.0954, "step": 11345 }, { "epoch": 0.11346, "grad_norm": 1.2018669323129128, "learning_rate": 0.003, "loss": 4.0826, "step": 11346 }, { "epoch": 0.11347, "grad_norm": 0.9854247275018928, "learning_rate": 0.003, "loss": 4.0615, "step": 11347 }, { "epoch": 0.11348, "grad_norm": 1.3540848648186976, "learning_rate": 0.003, "loss": 4.0869, "step": 11348 }, { "epoch": 0.11349, "grad_norm": 0.9209528701178629, "learning_rate": 0.003, "loss": 4.0791, "step": 11349 }, { "epoch": 0.1135, "grad_norm": 1.183193945927848, "learning_rate": 0.003, "loss": 4.1036, "step": 11350 }, { "epoch": 0.11351, "grad_norm": 1.2241071788224156, "learning_rate": 0.003, "loss": 4.0798, "step": 11351 }, { "epoch": 0.11352, "grad_norm": 1.264128877176907, "learning_rate": 0.003, "loss": 4.0472, "step": 11352 }, { "epoch": 0.11353, "grad_norm": 1.3375943105466253, "learning_rate": 0.003, "loss": 4.063, "step": 11353 }, { "epoch": 0.11354, "grad_norm": 0.971453513895793, "learning_rate": 0.003, "loss": 4.0674, "step": 11354 }, { "epoch": 0.11355, "grad_norm": 1.1992359612359424, "learning_rate": 0.003, "loss": 4.1127, "step": 11355 }, { "epoch": 0.11356, "grad_norm": 1.12742305868137, "learning_rate": 0.003, "loss": 4.074, "step": 11356 }, { "epoch": 0.11357, "grad_norm": 1.057282156982112, "learning_rate": 0.003, "loss": 4.0477, "step": 11357 }, { "epoch": 0.11358, "grad_norm": 1.3540534987747048, "learning_rate": 0.003, "loss": 4.0825, "step": 11358 }, { "epoch": 0.11359, "grad_norm": 1.054180251810891, "learning_rate": 0.003, "loss": 4.1003, "step": 11359 }, { "epoch": 0.1136, "grad_norm": 1.177256362048272, "learning_rate": 0.003, "loss": 4.0863, "step": 11360 }, { "epoch": 0.11361, "grad_norm": 1.1430591013777645, "learning_rate": 0.003, "loss": 4.0487, "step": 11361 }, { "epoch": 0.11362, "grad_norm": 1.2545721554637832, "learning_rate": 0.003, "loss": 4.0783, "step": 11362 }, { "epoch": 0.11363, "grad_norm": 1.1487676491455483, "learning_rate": 0.003, "loss": 4.0674, "step": 11363 }, { "epoch": 0.11364, "grad_norm": 1.174028562853811, "learning_rate": 0.003, "loss": 4.0935, "step": 11364 }, { "epoch": 0.11365, "grad_norm": 1.0524531488709266, "learning_rate": 0.003, "loss": 4.0758, "step": 11365 }, { "epoch": 0.11366, "grad_norm": 1.1127606638729275, "learning_rate": 0.003, "loss": 4.0752, "step": 11366 }, { "epoch": 0.11367, "grad_norm": 1.2889056949638662, "learning_rate": 0.003, "loss": 4.051, "step": 11367 }, { "epoch": 0.11368, "grad_norm": 1.1945595400318936, "learning_rate": 0.003, "loss": 4.099, "step": 11368 }, { "epoch": 0.11369, "grad_norm": 1.121435622675818, "learning_rate": 0.003, "loss": 4.0909, "step": 11369 }, { "epoch": 0.1137, "grad_norm": 1.2517077973859847, "learning_rate": 0.003, "loss": 4.1092, "step": 11370 }, { "epoch": 0.11371, "grad_norm": 1.0949921154536846, "learning_rate": 0.003, "loss": 4.0784, "step": 11371 }, { "epoch": 0.11372, "grad_norm": 1.247312554142341, "learning_rate": 0.003, "loss": 4.0705, "step": 11372 }, { "epoch": 0.11373, "grad_norm": 1.0455564948736882, "learning_rate": 0.003, "loss": 4.0839, "step": 11373 }, { "epoch": 0.11374, "grad_norm": 1.2309575034906135, "learning_rate": 0.003, "loss": 4.085, "step": 11374 }, { "epoch": 0.11375, "grad_norm": 1.1997157049677671, "learning_rate": 0.003, "loss": 4.0905, "step": 11375 }, { "epoch": 0.11376, "grad_norm": 1.2922369438044266, "learning_rate": 0.003, "loss": 4.1163, "step": 11376 }, { "epoch": 0.11377, "grad_norm": 1.227221408176285, "learning_rate": 0.003, "loss": 4.0789, "step": 11377 }, { "epoch": 0.11378, "grad_norm": 1.5411142967060618, "learning_rate": 0.003, "loss": 4.0738, "step": 11378 }, { "epoch": 0.11379, "grad_norm": 0.9410473447301275, "learning_rate": 0.003, "loss": 4.0581, "step": 11379 }, { "epoch": 0.1138, "grad_norm": 1.1332538562404029, "learning_rate": 0.003, "loss": 4.054, "step": 11380 }, { "epoch": 0.11381, "grad_norm": 1.4056385188743654, "learning_rate": 0.003, "loss": 4.0427, "step": 11381 }, { "epoch": 0.11382, "grad_norm": 1.1076341193368915, "learning_rate": 0.003, "loss": 4.0871, "step": 11382 }, { "epoch": 0.11383, "grad_norm": 1.3425460788316952, "learning_rate": 0.003, "loss": 4.0628, "step": 11383 }, { "epoch": 0.11384, "grad_norm": 0.9299747587551522, "learning_rate": 0.003, "loss": 4.0718, "step": 11384 }, { "epoch": 0.11385, "grad_norm": 1.1616465110678822, "learning_rate": 0.003, "loss": 4.0626, "step": 11385 }, { "epoch": 0.11386, "grad_norm": 1.228364230665424, "learning_rate": 0.003, "loss": 4.0931, "step": 11386 }, { "epoch": 0.11387, "grad_norm": 1.026775499618035, "learning_rate": 0.003, "loss": 4.066, "step": 11387 }, { "epoch": 0.11388, "grad_norm": 1.186841490398931, "learning_rate": 0.003, "loss": 4.0679, "step": 11388 }, { "epoch": 0.11389, "grad_norm": 1.113607604079879, "learning_rate": 0.003, "loss": 4.0784, "step": 11389 }, { "epoch": 0.1139, "grad_norm": 1.214021323948272, "learning_rate": 0.003, "loss": 4.0646, "step": 11390 }, { "epoch": 0.11391, "grad_norm": 1.0617336666676447, "learning_rate": 0.003, "loss": 4.0577, "step": 11391 }, { "epoch": 0.11392, "grad_norm": 1.4256327524001133, "learning_rate": 0.003, "loss": 4.0938, "step": 11392 }, { "epoch": 0.11393, "grad_norm": 0.9429026567842537, "learning_rate": 0.003, "loss": 4.0462, "step": 11393 }, { "epoch": 0.11394, "grad_norm": 1.0149734142468037, "learning_rate": 0.003, "loss": 4.0439, "step": 11394 }, { "epoch": 0.11395, "grad_norm": 0.9422668319518931, "learning_rate": 0.003, "loss": 4.08, "step": 11395 }, { "epoch": 0.11396, "grad_norm": 1.1401996080256407, "learning_rate": 0.003, "loss": 4.0844, "step": 11396 }, { "epoch": 0.11397, "grad_norm": 1.177214766567936, "learning_rate": 0.003, "loss": 4.0674, "step": 11397 }, { "epoch": 0.11398, "grad_norm": 1.0110021900524433, "learning_rate": 0.003, "loss": 4.0508, "step": 11398 }, { "epoch": 0.11399, "grad_norm": 1.1768839261262272, "learning_rate": 0.003, "loss": 4.0915, "step": 11399 }, { "epoch": 0.114, "grad_norm": 1.1414020916346157, "learning_rate": 0.003, "loss": 4.107, "step": 11400 }, { "epoch": 0.11401, "grad_norm": 1.150374703617575, "learning_rate": 0.003, "loss": 4.0662, "step": 11401 }, { "epoch": 0.11402, "grad_norm": 1.215556312236231, "learning_rate": 0.003, "loss": 4.064, "step": 11402 }, { "epoch": 0.11403, "grad_norm": 1.0440764034627348, "learning_rate": 0.003, "loss": 4.042, "step": 11403 }, { "epoch": 0.11404, "grad_norm": 1.6023256769426333, "learning_rate": 0.003, "loss": 4.0918, "step": 11404 }, { "epoch": 0.11405, "grad_norm": 1.294742133665827, "learning_rate": 0.003, "loss": 4.0874, "step": 11405 }, { "epoch": 0.11406, "grad_norm": 0.9897834597352938, "learning_rate": 0.003, "loss": 4.0985, "step": 11406 }, { "epoch": 0.11407, "grad_norm": 1.3648306468527558, "learning_rate": 0.003, "loss": 4.0754, "step": 11407 }, { "epoch": 0.11408, "grad_norm": 1.017340568621194, "learning_rate": 0.003, "loss": 4.0885, "step": 11408 }, { "epoch": 0.11409, "grad_norm": 1.3555212966368582, "learning_rate": 0.003, "loss": 4.0948, "step": 11409 }, { "epoch": 0.1141, "grad_norm": 1.0123596831049593, "learning_rate": 0.003, "loss": 4.0828, "step": 11410 }, { "epoch": 0.11411, "grad_norm": 1.4084356670391436, "learning_rate": 0.003, "loss": 4.1008, "step": 11411 }, { "epoch": 0.11412, "grad_norm": 1.0910871007698164, "learning_rate": 0.003, "loss": 4.0923, "step": 11412 }, { "epoch": 0.11413, "grad_norm": 0.9821281307103067, "learning_rate": 0.003, "loss": 4.068, "step": 11413 }, { "epoch": 0.11414, "grad_norm": 1.0281020978417275, "learning_rate": 0.003, "loss": 4.0828, "step": 11414 }, { "epoch": 0.11415, "grad_norm": 1.27061682180603, "learning_rate": 0.003, "loss": 4.0867, "step": 11415 }, { "epoch": 0.11416, "grad_norm": 1.3825399870725403, "learning_rate": 0.003, "loss": 4.0819, "step": 11416 }, { "epoch": 0.11417, "grad_norm": 1.0349060713735456, "learning_rate": 0.003, "loss": 4.0623, "step": 11417 }, { "epoch": 0.11418, "grad_norm": 1.247625981100148, "learning_rate": 0.003, "loss": 4.0775, "step": 11418 }, { "epoch": 0.11419, "grad_norm": 1.076240310475421, "learning_rate": 0.003, "loss": 4.0561, "step": 11419 }, { "epoch": 0.1142, "grad_norm": 1.126178785427172, "learning_rate": 0.003, "loss": 4.056, "step": 11420 }, { "epoch": 0.11421, "grad_norm": 1.1656854594192045, "learning_rate": 0.003, "loss": 4.0472, "step": 11421 }, { "epoch": 0.11422, "grad_norm": 1.2802536744397197, "learning_rate": 0.003, "loss": 4.0858, "step": 11422 }, { "epoch": 0.11423, "grad_norm": 1.058189762030975, "learning_rate": 0.003, "loss": 4.1078, "step": 11423 }, { "epoch": 0.11424, "grad_norm": 1.254947080462852, "learning_rate": 0.003, "loss": 4.104, "step": 11424 }, { "epoch": 0.11425, "grad_norm": 1.0493001159378452, "learning_rate": 0.003, "loss": 4.0816, "step": 11425 }, { "epoch": 0.11426, "grad_norm": 1.4885665242657737, "learning_rate": 0.003, "loss": 4.1081, "step": 11426 }, { "epoch": 0.11427, "grad_norm": 0.9720233775254423, "learning_rate": 0.003, "loss": 4.0741, "step": 11427 }, { "epoch": 0.11428, "grad_norm": 1.2866999811480075, "learning_rate": 0.003, "loss": 4.089, "step": 11428 }, { "epoch": 0.11429, "grad_norm": 1.2926212573442324, "learning_rate": 0.003, "loss": 4.1181, "step": 11429 }, { "epoch": 0.1143, "grad_norm": 1.0602107668095546, "learning_rate": 0.003, "loss": 4.097, "step": 11430 }, { "epoch": 0.11431, "grad_norm": 1.3490347702536993, "learning_rate": 0.003, "loss": 4.1032, "step": 11431 }, { "epoch": 0.11432, "grad_norm": 1.3200116165463798, "learning_rate": 0.003, "loss": 4.0622, "step": 11432 }, { "epoch": 0.11433, "grad_norm": 1.0646935424901862, "learning_rate": 0.003, "loss": 4.1142, "step": 11433 }, { "epoch": 0.11434, "grad_norm": 1.053308075790006, "learning_rate": 0.003, "loss": 4.1086, "step": 11434 }, { "epoch": 0.11435, "grad_norm": 1.0987766943842554, "learning_rate": 0.003, "loss": 4.0699, "step": 11435 }, { "epoch": 0.11436, "grad_norm": 1.078887782161941, "learning_rate": 0.003, "loss": 4.0861, "step": 11436 }, { "epoch": 0.11437, "grad_norm": 1.2236619685979702, "learning_rate": 0.003, "loss": 4.0412, "step": 11437 }, { "epoch": 0.11438, "grad_norm": 0.9965897884378967, "learning_rate": 0.003, "loss": 4.0765, "step": 11438 }, { "epoch": 0.11439, "grad_norm": 1.5171415237819303, "learning_rate": 0.003, "loss": 4.0645, "step": 11439 }, { "epoch": 0.1144, "grad_norm": 0.9254183689955512, "learning_rate": 0.003, "loss": 4.0867, "step": 11440 }, { "epoch": 0.11441, "grad_norm": 1.1802455391859032, "learning_rate": 0.003, "loss": 4.07, "step": 11441 }, { "epoch": 0.11442, "grad_norm": 1.1304408432442188, "learning_rate": 0.003, "loss": 4.0772, "step": 11442 }, { "epoch": 0.11443, "grad_norm": 1.2616223153469455, "learning_rate": 0.003, "loss": 4.0852, "step": 11443 }, { "epoch": 0.11444, "grad_norm": 0.9898632535928127, "learning_rate": 0.003, "loss": 4.0611, "step": 11444 }, { "epoch": 0.11445, "grad_norm": 1.1106024369621383, "learning_rate": 0.003, "loss": 4.0893, "step": 11445 }, { "epoch": 0.11446, "grad_norm": 1.1313665518951734, "learning_rate": 0.003, "loss": 4.0782, "step": 11446 }, { "epoch": 0.11447, "grad_norm": 1.1055046914749922, "learning_rate": 0.003, "loss": 4.1165, "step": 11447 }, { "epoch": 0.11448, "grad_norm": 1.2700027524427357, "learning_rate": 0.003, "loss": 4.1123, "step": 11448 }, { "epoch": 0.11449, "grad_norm": 1.1747973362905308, "learning_rate": 0.003, "loss": 4.0784, "step": 11449 }, { "epoch": 0.1145, "grad_norm": 1.2448867650140631, "learning_rate": 0.003, "loss": 4.0946, "step": 11450 }, { "epoch": 0.11451, "grad_norm": 1.4021973572768907, "learning_rate": 0.003, "loss": 4.0935, "step": 11451 }, { "epoch": 0.11452, "grad_norm": 1.2978912346016391, "learning_rate": 0.003, "loss": 4.0652, "step": 11452 }, { "epoch": 0.11453, "grad_norm": 1.2161204587703507, "learning_rate": 0.003, "loss": 4.0732, "step": 11453 }, { "epoch": 0.11454, "grad_norm": 1.347884771284405, "learning_rate": 0.003, "loss": 4.0771, "step": 11454 }, { "epoch": 0.11455, "grad_norm": 1.0701199335942853, "learning_rate": 0.003, "loss": 4.0477, "step": 11455 }, { "epoch": 0.11456, "grad_norm": 1.087637663877047, "learning_rate": 0.003, "loss": 4.0713, "step": 11456 }, { "epoch": 0.11457, "grad_norm": 1.1001486271515806, "learning_rate": 0.003, "loss": 4.0627, "step": 11457 }, { "epoch": 0.11458, "grad_norm": 1.0288611708922306, "learning_rate": 0.003, "loss": 4.0878, "step": 11458 }, { "epoch": 0.11459, "grad_norm": 1.2930415256059284, "learning_rate": 0.003, "loss": 4.0761, "step": 11459 }, { "epoch": 0.1146, "grad_norm": 0.9396094840689607, "learning_rate": 0.003, "loss": 4.0599, "step": 11460 }, { "epoch": 0.11461, "grad_norm": 1.3243731266257484, "learning_rate": 0.003, "loss": 4.0956, "step": 11461 }, { "epoch": 0.11462, "grad_norm": 1.0499960484767916, "learning_rate": 0.003, "loss": 4.0807, "step": 11462 }, { "epoch": 0.11463, "grad_norm": 1.366084137306469, "learning_rate": 0.003, "loss": 4.0784, "step": 11463 }, { "epoch": 0.11464, "grad_norm": 0.9389564810897638, "learning_rate": 0.003, "loss": 4.0785, "step": 11464 }, { "epoch": 0.11465, "grad_norm": 1.20440081026079, "learning_rate": 0.003, "loss": 4.0918, "step": 11465 }, { "epoch": 0.11466, "grad_norm": 1.149724890423667, "learning_rate": 0.003, "loss": 4.0751, "step": 11466 }, { "epoch": 0.11467, "grad_norm": 1.2853468225896403, "learning_rate": 0.003, "loss": 4.0938, "step": 11467 }, { "epoch": 0.11468, "grad_norm": 1.1978301632485409, "learning_rate": 0.003, "loss": 4.0777, "step": 11468 }, { "epoch": 0.11469, "grad_norm": 1.0202858294727375, "learning_rate": 0.003, "loss": 4.0786, "step": 11469 }, { "epoch": 0.1147, "grad_norm": 1.409755040044635, "learning_rate": 0.003, "loss": 4.0698, "step": 11470 }, { "epoch": 0.11471, "grad_norm": 1.1159553996794154, "learning_rate": 0.003, "loss": 4.0752, "step": 11471 }, { "epoch": 0.11472, "grad_norm": 1.114063611856724, "learning_rate": 0.003, "loss": 4.0666, "step": 11472 }, { "epoch": 0.11473, "grad_norm": 1.2182970670688382, "learning_rate": 0.003, "loss": 4.0795, "step": 11473 }, { "epoch": 0.11474, "grad_norm": 1.0696376544415567, "learning_rate": 0.003, "loss": 4.0791, "step": 11474 }, { "epoch": 0.11475, "grad_norm": 1.332890292488662, "learning_rate": 0.003, "loss": 4.0593, "step": 11475 }, { "epoch": 0.11476, "grad_norm": 0.9756801900159172, "learning_rate": 0.003, "loss": 4.0689, "step": 11476 }, { "epoch": 0.11477, "grad_norm": 1.372607682848459, "learning_rate": 0.003, "loss": 4.1171, "step": 11477 }, { "epoch": 0.11478, "grad_norm": 0.8255639146406556, "learning_rate": 0.003, "loss": 4.0494, "step": 11478 }, { "epoch": 0.11479, "grad_norm": 1.1664926036649057, "learning_rate": 0.003, "loss": 4.0608, "step": 11479 }, { "epoch": 0.1148, "grad_norm": 1.294357451851071, "learning_rate": 0.003, "loss": 4.1077, "step": 11480 }, { "epoch": 0.11481, "grad_norm": 1.3833008263992455, "learning_rate": 0.003, "loss": 4.0641, "step": 11481 }, { "epoch": 0.11482, "grad_norm": 1.1468410323877665, "learning_rate": 0.003, "loss": 4.1107, "step": 11482 }, { "epoch": 0.11483, "grad_norm": 0.9887223520013214, "learning_rate": 0.003, "loss": 4.0526, "step": 11483 }, { "epoch": 0.11484, "grad_norm": 1.1103043081093302, "learning_rate": 0.003, "loss": 4.0747, "step": 11484 }, { "epoch": 0.11485, "grad_norm": 1.2539182845087775, "learning_rate": 0.003, "loss": 4.0725, "step": 11485 }, { "epoch": 0.11486, "grad_norm": 1.229758733901623, "learning_rate": 0.003, "loss": 4.0849, "step": 11486 }, { "epoch": 0.11487, "grad_norm": 1.1836079272602194, "learning_rate": 0.003, "loss": 4.0797, "step": 11487 }, { "epoch": 0.11488, "grad_norm": 0.9349775387277866, "learning_rate": 0.003, "loss": 4.0822, "step": 11488 }, { "epoch": 0.11489, "grad_norm": 1.1135837477366786, "learning_rate": 0.003, "loss": 4.0779, "step": 11489 }, { "epoch": 0.1149, "grad_norm": 1.0788610305126276, "learning_rate": 0.003, "loss": 4.0778, "step": 11490 }, { "epoch": 0.11491, "grad_norm": 1.0996128966845133, "learning_rate": 0.003, "loss": 4.0445, "step": 11491 }, { "epoch": 0.11492, "grad_norm": 1.1962298260701187, "learning_rate": 0.003, "loss": 4.0814, "step": 11492 }, { "epoch": 0.11493, "grad_norm": 1.0941724572100726, "learning_rate": 0.003, "loss": 4.077, "step": 11493 }, { "epoch": 0.11494, "grad_norm": 1.3475798987524423, "learning_rate": 0.003, "loss": 4.0638, "step": 11494 }, { "epoch": 0.11495, "grad_norm": 1.1329316783150218, "learning_rate": 0.003, "loss": 4.0795, "step": 11495 }, { "epoch": 0.11496, "grad_norm": 1.1896099564690321, "learning_rate": 0.003, "loss": 4.0621, "step": 11496 }, { "epoch": 0.11497, "grad_norm": 1.0928050136095804, "learning_rate": 0.003, "loss": 4.061, "step": 11497 }, { "epoch": 0.11498, "grad_norm": 1.0963983050897443, "learning_rate": 0.003, "loss": 4.0725, "step": 11498 }, { "epoch": 0.11499, "grad_norm": 1.2008513248037702, "learning_rate": 0.003, "loss": 4.0784, "step": 11499 }, { "epoch": 0.115, "grad_norm": 1.308613469975689, "learning_rate": 0.003, "loss": 4.1015, "step": 11500 }, { "epoch": 0.11501, "grad_norm": 1.3175611324257215, "learning_rate": 0.003, "loss": 4.101, "step": 11501 }, { "epoch": 0.11502, "grad_norm": 1.026242356134211, "learning_rate": 0.003, "loss": 4.087, "step": 11502 }, { "epoch": 0.11503, "grad_norm": 1.1805651466017921, "learning_rate": 0.003, "loss": 4.0862, "step": 11503 }, { "epoch": 0.11504, "grad_norm": 0.9738010615945621, "learning_rate": 0.003, "loss": 4.0701, "step": 11504 }, { "epoch": 0.11505, "grad_norm": 1.3092225762310343, "learning_rate": 0.003, "loss": 4.0773, "step": 11505 }, { "epoch": 0.11506, "grad_norm": 0.97474226050319, "learning_rate": 0.003, "loss": 4.0981, "step": 11506 }, { "epoch": 0.11507, "grad_norm": 1.1334909402353455, "learning_rate": 0.003, "loss": 4.0939, "step": 11507 }, { "epoch": 0.11508, "grad_norm": 1.2437644494475641, "learning_rate": 0.003, "loss": 4.0686, "step": 11508 }, { "epoch": 0.11509, "grad_norm": 1.127927044502533, "learning_rate": 0.003, "loss": 4.0654, "step": 11509 }, { "epoch": 0.1151, "grad_norm": 1.3443112351268103, "learning_rate": 0.003, "loss": 4.066, "step": 11510 }, { "epoch": 0.11511, "grad_norm": 1.1679143581179199, "learning_rate": 0.003, "loss": 4.0566, "step": 11511 }, { "epoch": 0.11512, "grad_norm": 1.1817297226139298, "learning_rate": 0.003, "loss": 4.0649, "step": 11512 }, { "epoch": 0.11513, "grad_norm": 1.3774151922566873, "learning_rate": 0.003, "loss": 4.0942, "step": 11513 }, { "epoch": 0.11514, "grad_norm": 1.064266101329151, "learning_rate": 0.003, "loss": 4.0563, "step": 11514 }, { "epoch": 0.11515, "grad_norm": 1.3813630987722414, "learning_rate": 0.003, "loss": 4.1194, "step": 11515 }, { "epoch": 0.11516, "grad_norm": 1.0477884800845256, "learning_rate": 0.003, "loss": 4.0954, "step": 11516 }, { "epoch": 0.11517, "grad_norm": 1.321817198455129, "learning_rate": 0.003, "loss": 4.0725, "step": 11517 }, { "epoch": 0.11518, "grad_norm": 0.9354471219736039, "learning_rate": 0.003, "loss": 4.1007, "step": 11518 }, { "epoch": 0.11519, "grad_norm": 1.2091105494550873, "learning_rate": 0.003, "loss": 4.0709, "step": 11519 }, { "epoch": 0.1152, "grad_norm": 1.0806128156529098, "learning_rate": 0.003, "loss": 4.0435, "step": 11520 }, { "epoch": 0.11521, "grad_norm": 1.2408933034442884, "learning_rate": 0.003, "loss": 4.0829, "step": 11521 }, { "epoch": 0.11522, "grad_norm": 1.2467169335127846, "learning_rate": 0.003, "loss": 4.0837, "step": 11522 }, { "epoch": 0.11523, "grad_norm": 1.2856824199643657, "learning_rate": 0.003, "loss": 4.0864, "step": 11523 }, { "epoch": 0.11524, "grad_norm": 1.264360674227239, "learning_rate": 0.003, "loss": 4.0863, "step": 11524 }, { "epoch": 0.11525, "grad_norm": 1.0508319704976452, "learning_rate": 0.003, "loss": 4.0924, "step": 11525 }, { "epoch": 0.11526, "grad_norm": 1.3112540782269941, "learning_rate": 0.003, "loss": 4.0534, "step": 11526 }, { "epoch": 0.11527, "grad_norm": 1.0643098598317824, "learning_rate": 0.003, "loss": 4.0862, "step": 11527 }, { "epoch": 0.11528, "grad_norm": 1.1906495679674107, "learning_rate": 0.003, "loss": 4.061, "step": 11528 }, { "epoch": 0.11529, "grad_norm": 1.2980704217448864, "learning_rate": 0.003, "loss": 4.0923, "step": 11529 }, { "epoch": 0.1153, "grad_norm": 0.8867430435194795, "learning_rate": 0.003, "loss": 4.0362, "step": 11530 }, { "epoch": 0.11531, "grad_norm": 1.1040483708431459, "learning_rate": 0.003, "loss": 4.0847, "step": 11531 }, { "epoch": 0.11532, "grad_norm": 1.1805903226827017, "learning_rate": 0.003, "loss": 4.0722, "step": 11532 }, { "epoch": 0.11533, "grad_norm": 1.1523593683566506, "learning_rate": 0.003, "loss": 4.0603, "step": 11533 }, { "epoch": 0.11534, "grad_norm": 1.2466735154295447, "learning_rate": 0.003, "loss": 4.0611, "step": 11534 }, { "epoch": 0.11535, "grad_norm": 1.1515859901279828, "learning_rate": 0.003, "loss": 4.044, "step": 11535 }, { "epoch": 0.11536, "grad_norm": 0.9773817166626956, "learning_rate": 0.003, "loss": 4.0615, "step": 11536 }, { "epoch": 0.11537, "grad_norm": 1.353165757233726, "learning_rate": 0.003, "loss": 4.0813, "step": 11537 }, { "epoch": 0.11538, "grad_norm": 1.0286698146467392, "learning_rate": 0.003, "loss": 4.0938, "step": 11538 }, { "epoch": 0.11539, "grad_norm": 1.3615233983899735, "learning_rate": 0.003, "loss": 4.0627, "step": 11539 }, { "epoch": 0.1154, "grad_norm": 0.923404459886176, "learning_rate": 0.003, "loss": 4.0958, "step": 11540 }, { "epoch": 0.11541, "grad_norm": 1.179077597795612, "learning_rate": 0.003, "loss": 4.0987, "step": 11541 }, { "epoch": 0.11542, "grad_norm": 1.0504999776520778, "learning_rate": 0.003, "loss": 4.0985, "step": 11542 }, { "epoch": 0.11543, "grad_norm": 1.266392988607479, "learning_rate": 0.003, "loss": 4.0758, "step": 11543 }, { "epoch": 0.11544, "grad_norm": 1.0690194520990461, "learning_rate": 0.003, "loss": 4.0643, "step": 11544 }, { "epoch": 0.11545, "grad_norm": 1.0187394776585554, "learning_rate": 0.003, "loss": 4.0518, "step": 11545 }, { "epoch": 0.11546, "grad_norm": 1.1881403653169518, "learning_rate": 0.003, "loss": 4.0922, "step": 11546 }, { "epoch": 0.11547, "grad_norm": 0.9061876375465832, "learning_rate": 0.003, "loss": 4.0588, "step": 11547 }, { "epoch": 0.11548, "grad_norm": 1.280760949288439, "learning_rate": 0.003, "loss": 4.0638, "step": 11548 }, { "epoch": 0.11549, "grad_norm": 1.6926012022473007, "learning_rate": 0.003, "loss": 4.0971, "step": 11549 }, { "epoch": 0.1155, "grad_norm": 1.3743997496349738, "learning_rate": 0.003, "loss": 4.0708, "step": 11550 }, { "epoch": 0.11551, "grad_norm": 0.9575390298144556, "learning_rate": 0.003, "loss": 4.0808, "step": 11551 }, { "epoch": 0.11552, "grad_norm": 1.3025097000865644, "learning_rate": 0.003, "loss": 4.086, "step": 11552 }, { "epoch": 0.11553, "grad_norm": 1.174925247280143, "learning_rate": 0.003, "loss": 4.0607, "step": 11553 }, { "epoch": 0.11554, "grad_norm": 1.1507902802997487, "learning_rate": 0.003, "loss": 4.0714, "step": 11554 }, { "epoch": 0.11555, "grad_norm": 1.3723048067418877, "learning_rate": 0.003, "loss": 4.0772, "step": 11555 }, { "epoch": 0.11556, "grad_norm": 1.1434625203555215, "learning_rate": 0.003, "loss": 4.0525, "step": 11556 }, { "epoch": 0.11557, "grad_norm": 1.274192421285598, "learning_rate": 0.003, "loss": 4.0741, "step": 11557 }, { "epoch": 0.11558, "grad_norm": 1.21004073015434, "learning_rate": 0.003, "loss": 4.08, "step": 11558 }, { "epoch": 0.11559, "grad_norm": 1.0819390577627044, "learning_rate": 0.003, "loss": 4.0694, "step": 11559 }, { "epoch": 0.1156, "grad_norm": 1.2733871597157078, "learning_rate": 0.003, "loss": 4.1022, "step": 11560 }, { "epoch": 0.11561, "grad_norm": 1.1210160300210403, "learning_rate": 0.003, "loss": 4.0739, "step": 11561 }, { "epoch": 0.11562, "grad_norm": 1.341917662056347, "learning_rate": 0.003, "loss": 4.0516, "step": 11562 }, { "epoch": 0.11563, "grad_norm": 0.9899531411900627, "learning_rate": 0.003, "loss": 4.0678, "step": 11563 }, { "epoch": 0.11564, "grad_norm": 1.3331094512680763, "learning_rate": 0.003, "loss": 4.0757, "step": 11564 }, { "epoch": 0.11565, "grad_norm": 1.074378038012777, "learning_rate": 0.003, "loss": 4.0881, "step": 11565 }, { "epoch": 0.11566, "grad_norm": 1.0921225706498394, "learning_rate": 0.003, "loss": 4.0457, "step": 11566 }, { "epoch": 0.11567, "grad_norm": 1.1686265160176652, "learning_rate": 0.003, "loss": 4.0674, "step": 11567 }, { "epoch": 0.11568, "grad_norm": 1.1345061017130718, "learning_rate": 0.003, "loss": 4.0782, "step": 11568 }, { "epoch": 0.11569, "grad_norm": 1.2407044804857985, "learning_rate": 0.003, "loss": 4.083, "step": 11569 }, { "epoch": 0.1157, "grad_norm": 1.2933894276546998, "learning_rate": 0.003, "loss": 4.0436, "step": 11570 }, { "epoch": 0.11571, "grad_norm": 1.0144524727039599, "learning_rate": 0.003, "loss": 4.0767, "step": 11571 }, { "epoch": 0.11572, "grad_norm": 1.154824048951517, "learning_rate": 0.003, "loss": 4.0313, "step": 11572 }, { "epoch": 0.11573, "grad_norm": 1.3344606188631556, "learning_rate": 0.003, "loss": 4.0876, "step": 11573 }, { "epoch": 0.11574, "grad_norm": 0.9324521525913374, "learning_rate": 0.003, "loss": 4.0392, "step": 11574 }, { "epoch": 0.11575, "grad_norm": 0.9666961252283515, "learning_rate": 0.003, "loss": 4.0744, "step": 11575 }, { "epoch": 0.11576, "grad_norm": 1.0929762780109733, "learning_rate": 0.003, "loss": 4.041, "step": 11576 }, { "epoch": 0.11577, "grad_norm": 1.359434801271052, "learning_rate": 0.003, "loss": 4.06, "step": 11577 }, { "epoch": 0.11578, "grad_norm": 1.0266321837498404, "learning_rate": 0.003, "loss": 4.0924, "step": 11578 }, { "epoch": 0.11579, "grad_norm": 1.1097851636303517, "learning_rate": 0.003, "loss": 4.0621, "step": 11579 }, { "epoch": 0.1158, "grad_norm": 1.2531227471152486, "learning_rate": 0.003, "loss": 4.0701, "step": 11580 }, { "epoch": 0.11581, "grad_norm": 1.3107090415682447, "learning_rate": 0.003, "loss": 4.0406, "step": 11581 }, { "epoch": 0.11582, "grad_norm": 1.132608200106952, "learning_rate": 0.003, "loss": 4.065, "step": 11582 }, { "epoch": 0.11583, "grad_norm": 1.1646643953883302, "learning_rate": 0.003, "loss": 4.0527, "step": 11583 }, { "epoch": 0.11584, "grad_norm": 0.966463751002327, "learning_rate": 0.003, "loss": 4.0836, "step": 11584 }, { "epoch": 0.11585, "grad_norm": 1.1856859379676656, "learning_rate": 0.003, "loss": 4.1024, "step": 11585 }, { "epoch": 0.11586, "grad_norm": 1.086649497913007, "learning_rate": 0.003, "loss": 4.0602, "step": 11586 }, { "epoch": 0.11587, "grad_norm": 1.2690379379929062, "learning_rate": 0.003, "loss": 4.032, "step": 11587 }, { "epoch": 0.11588, "grad_norm": 0.9935718042905032, "learning_rate": 0.003, "loss": 4.076, "step": 11588 }, { "epoch": 0.11589, "grad_norm": 1.3313725583096991, "learning_rate": 0.003, "loss": 4.0624, "step": 11589 }, { "epoch": 0.1159, "grad_norm": 1.0259547188310283, "learning_rate": 0.003, "loss": 4.0522, "step": 11590 }, { "epoch": 0.11591, "grad_norm": 1.3133936881217214, "learning_rate": 0.003, "loss": 4.0767, "step": 11591 }, { "epoch": 0.11592, "grad_norm": 1.0213838493879515, "learning_rate": 0.003, "loss": 4.0796, "step": 11592 }, { "epoch": 0.11593, "grad_norm": 1.347456693219393, "learning_rate": 0.003, "loss": 4.08, "step": 11593 }, { "epoch": 0.11594, "grad_norm": 1.1679330089514939, "learning_rate": 0.003, "loss": 4.0546, "step": 11594 }, { "epoch": 0.11595, "grad_norm": 1.104230185366043, "learning_rate": 0.003, "loss": 4.092, "step": 11595 }, { "epoch": 0.11596, "grad_norm": 1.3096538728464635, "learning_rate": 0.003, "loss": 4.0663, "step": 11596 }, { "epoch": 0.11597, "grad_norm": 1.0708176834350167, "learning_rate": 0.003, "loss": 4.0955, "step": 11597 }, { "epoch": 0.11598, "grad_norm": 1.419218488468854, "learning_rate": 0.003, "loss": 4.0774, "step": 11598 }, { "epoch": 0.11599, "grad_norm": 1.059367189303041, "learning_rate": 0.003, "loss": 4.0647, "step": 11599 }, { "epoch": 0.116, "grad_norm": 1.4108671475415633, "learning_rate": 0.003, "loss": 4.0705, "step": 11600 }, { "epoch": 0.11601, "grad_norm": 0.9733638184815594, "learning_rate": 0.003, "loss": 4.0959, "step": 11601 }, { "epoch": 0.11602, "grad_norm": 1.0820043730117697, "learning_rate": 0.003, "loss": 4.043, "step": 11602 }, { "epoch": 0.11603, "grad_norm": 1.2147531231150952, "learning_rate": 0.003, "loss": 4.0606, "step": 11603 }, { "epoch": 0.11604, "grad_norm": 1.123040949818344, "learning_rate": 0.003, "loss": 4.0628, "step": 11604 }, { "epoch": 0.11605, "grad_norm": 1.3485877186705435, "learning_rate": 0.003, "loss": 4.0759, "step": 11605 }, { "epoch": 0.11606, "grad_norm": 1.0159588779281732, "learning_rate": 0.003, "loss": 4.061, "step": 11606 }, { "epoch": 0.11607, "grad_norm": 1.31908257646201, "learning_rate": 0.003, "loss": 4.0752, "step": 11607 }, { "epoch": 0.11608, "grad_norm": 0.9972867655077241, "learning_rate": 0.003, "loss": 4.0891, "step": 11608 }, { "epoch": 0.11609, "grad_norm": 1.3061558323354592, "learning_rate": 0.003, "loss": 4.1141, "step": 11609 }, { "epoch": 0.1161, "grad_norm": 0.971407556763974, "learning_rate": 0.003, "loss": 4.0925, "step": 11610 }, { "epoch": 0.11611, "grad_norm": 1.2215246724221014, "learning_rate": 0.003, "loss": 4.0938, "step": 11611 }, { "epoch": 0.11612, "grad_norm": 1.1214563125973465, "learning_rate": 0.003, "loss": 4.1022, "step": 11612 }, { "epoch": 0.11613, "grad_norm": 1.3463323837268832, "learning_rate": 0.003, "loss": 4.067, "step": 11613 }, { "epoch": 0.11614, "grad_norm": 1.005861871950001, "learning_rate": 0.003, "loss": 4.0784, "step": 11614 }, { "epoch": 0.11615, "grad_norm": 1.1173298064399497, "learning_rate": 0.003, "loss": 4.0724, "step": 11615 }, { "epoch": 0.11616, "grad_norm": 1.0865141677282626, "learning_rate": 0.003, "loss": 4.0508, "step": 11616 }, { "epoch": 0.11617, "grad_norm": 1.2021344904783986, "learning_rate": 0.003, "loss": 4.0916, "step": 11617 }, { "epoch": 0.11618, "grad_norm": 1.2654706652266698, "learning_rate": 0.003, "loss": 4.0715, "step": 11618 }, { "epoch": 0.11619, "grad_norm": 1.1995899472436178, "learning_rate": 0.003, "loss": 4.0687, "step": 11619 }, { "epoch": 0.1162, "grad_norm": 1.0558948472377412, "learning_rate": 0.003, "loss": 4.066, "step": 11620 }, { "epoch": 0.11621, "grad_norm": 1.2607446375842715, "learning_rate": 0.003, "loss": 4.0935, "step": 11621 }, { "epoch": 0.11622, "grad_norm": 1.115655721531063, "learning_rate": 0.003, "loss": 4.0834, "step": 11622 }, { "epoch": 0.11623, "grad_norm": 1.2795840516663242, "learning_rate": 0.003, "loss": 4.0802, "step": 11623 }, { "epoch": 0.11624, "grad_norm": 1.1604711054182177, "learning_rate": 0.003, "loss": 4.068, "step": 11624 }, { "epoch": 0.11625, "grad_norm": 1.1954532583033348, "learning_rate": 0.003, "loss": 4.0588, "step": 11625 }, { "epoch": 0.11626, "grad_norm": 1.020769225323057, "learning_rate": 0.003, "loss": 4.0626, "step": 11626 }, { "epoch": 0.11627, "grad_norm": 1.3014178927477036, "learning_rate": 0.003, "loss": 4.0657, "step": 11627 }, { "epoch": 0.11628, "grad_norm": 1.1723273023585614, "learning_rate": 0.003, "loss": 4.1149, "step": 11628 }, { "epoch": 0.11629, "grad_norm": 1.3072394900620368, "learning_rate": 0.003, "loss": 4.0738, "step": 11629 }, { "epoch": 0.1163, "grad_norm": 1.2846287717965974, "learning_rate": 0.003, "loss": 4.06, "step": 11630 }, { "epoch": 0.11631, "grad_norm": 0.9780232880461375, "learning_rate": 0.003, "loss": 4.0582, "step": 11631 }, { "epoch": 0.11632, "grad_norm": 1.1660762073465643, "learning_rate": 0.003, "loss": 4.069, "step": 11632 }, { "epoch": 0.11633, "grad_norm": 1.110005484921728, "learning_rate": 0.003, "loss": 4.0552, "step": 11633 }, { "epoch": 0.11634, "grad_norm": 1.1452153297503211, "learning_rate": 0.003, "loss": 4.0863, "step": 11634 }, { "epoch": 0.11635, "grad_norm": 0.9829449454658096, "learning_rate": 0.003, "loss": 4.0682, "step": 11635 }, { "epoch": 0.11636, "grad_norm": 1.3147897318417492, "learning_rate": 0.003, "loss": 4.0714, "step": 11636 }, { "epoch": 0.11637, "grad_norm": 1.0265219018313658, "learning_rate": 0.003, "loss": 4.0739, "step": 11637 }, { "epoch": 0.11638, "grad_norm": 1.21791936895603, "learning_rate": 0.003, "loss": 4.0638, "step": 11638 }, { "epoch": 0.11639, "grad_norm": 1.1939433767765795, "learning_rate": 0.003, "loss": 4.0542, "step": 11639 }, { "epoch": 0.1164, "grad_norm": 1.3414467709428268, "learning_rate": 0.003, "loss": 4.1041, "step": 11640 }, { "epoch": 0.11641, "grad_norm": 1.0941144662922078, "learning_rate": 0.003, "loss": 4.0813, "step": 11641 }, { "epoch": 0.11642, "grad_norm": 1.2308907705165233, "learning_rate": 0.003, "loss": 4.071, "step": 11642 }, { "epoch": 0.11643, "grad_norm": 0.9691303413285445, "learning_rate": 0.003, "loss": 4.0874, "step": 11643 }, { "epoch": 0.11644, "grad_norm": 1.08208704933791, "learning_rate": 0.003, "loss": 4.0779, "step": 11644 }, { "epoch": 0.11645, "grad_norm": 1.1657566901506018, "learning_rate": 0.003, "loss": 4.0903, "step": 11645 }, { "epoch": 0.11646, "grad_norm": 1.0026554853780012, "learning_rate": 0.003, "loss": 4.0918, "step": 11646 }, { "epoch": 0.11647, "grad_norm": 1.3278869384524574, "learning_rate": 0.003, "loss": 4.0633, "step": 11647 }, { "epoch": 0.11648, "grad_norm": 1.1526164101217382, "learning_rate": 0.003, "loss": 4.0866, "step": 11648 }, { "epoch": 0.11649, "grad_norm": 1.1702419673180477, "learning_rate": 0.003, "loss": 4.069, "step": 11649 }, { "epoch": 0.1165, "grad_norm": 1.2199436298530248, "learning_rate": 0.003, "loss": 4.0633, "step": 11650 }, { "epoch": 0.11651, "grad_norm": 1.0170923163006491, "learning_rate": 0.003, "loss": 4.0807, "step": 11651 }, { "epoch": 0.11652, "grad_norm": 1.1947281587929344, "learning_rate": 0.003, "loss": 4.0441, "step": 11652 }, { "epoch": 0.11653, "grad_norm": 1.4536369469192147, "learning_rate": 0.003, "loss": 4.0886, "step": 11653 }, { "epoch": 0.11654, "grad_norm": 0.986442918965962, "learning_rate": 0.003, "loss": 4.0505, "step": 11654 }, { "epoch": 0.11655, "grad_norm": 1.3730559055015554, "learning_rate": 0.003, "loss": 4.079, "step": 11655 }, { "epoch": 0.11656, "grad_norm": 0.987156630639679, "learning_rate": 0.003, "loss": 4.0682, "step": 11656 }, { "epoch": 0.11657, "grad_norm": 1.3704453079105388, "learning_rate": 0.003, "loss": 4.081, "step": 11657 }, { "epoch": 0.11658, "grad_norm": 1.1986780205189669, "learning_rate": 0.003, "loss": 4.1009, "step": 11658 }, { "epoch": 0.11659, "grad_norm": 1.2081230426683085, "learning_rate": 0.003, "loss": 4.0811, "step": 11659 }, { "epoch": 0.1166, "grad_norm": 1.2151928435575177, "learning_rate": 0.003, "loss": 4.0679, "step": 11660 }, { "epoch": 0.11661, "grad_norm": 1.2077636774696505, "learning_rate": 0.003, "loss": 4.0882, "step": 11661 }, { "epoch": 0.11662, "grad_norm": 1.1200976843627546, "learning_rate": 0.003, "loss": 4.0707, "step": 11662 }, { "epoch": 0.11663, "grad_norm": 1.3007439222307542, "learning_rate": 0.003, "loss": 4.0558, "step": 11663 }, { "epoch": 0.11664, "grad_norm": 0.9318794503160776, "learning_rate": 0.003, "loss": 4.0926, "step": 11664 }, { "epoch": 0.11665, "grad_norm": 1.098989623725052, "learning_rate": 0.003, "loss": 4.0427, "step": 11665 }, { "epoch": 0.11666, "grad_norm": 1.1501714554377738, "learning_rate": 0.003, "loss": 4.0704, "step": 11666 }, { "epoch": 0.11667, "grad_norm": 1.2824033405427517, "learning_rate": 0.003, "loss": 4.1042, "step": 11667 }, { "epoch": 0.11668, "grad_norm": 0.9966630289505742, "learning_rate": 0.003, "loss": 4.0643, "step": 11668 }, { "epoch": 0.11669, "grad_norm": 1.3169369638514503, "learning_rate": 0.003, "loss": 4.0805, "step": 11669 }, { "epoch": 0.1167, "grad_norm": 1.0024602739049056, "learning_rate": 0.003, "loss": 4.0778, "step": 11670 }, { "epoch": 0.11671, "grad_norm": 1.4408581457072513, "learning_rate": 0.003, "loss": 4.0642, "step": 11671 }, { "epoch": 0.11672, "grad_norm": 0.8851874480879731, "learning_rate": 0.003, "loss": 4.0991, "step": 11672 }, { "epoch": 0.11673, "grad_norm": 1.1906471160370373, "learning_rate": 0.003, "loss": 4.0696, "step": 11673 }, { "epoch": 0.11674, "grad_norm": 1.1719886567944, "learning_rate": 0.003, "loss": 4.0719, "step": 11674 }, { "epoch": 0.11675, "grad_norm": 1.3878506215850128, "learning_rate": 0.003, "loss": 4.0784, "step": 11675 }, { "epoch": 0.11676, "grad_norm": 1.0904085137277049, "learning_rate": 0.003, "loss": 4.0816, "step": 11676 }, { "epoch": 0.11677, "grad_norm": 1.2116152858637508, "learning_rate": 0.003, "loss": 4.0933, "step": 11677 }, { "epoch": 0.11678, "grad_norm": 1.343503928364903, "learning_rate": 0.003, "loss": 4.0893, "step": 11678 }, { "epoch": 0.11679, "grad_norm": 0.937076872190132, "learning_rate": 0.003, "loss": 4.0679, "step": 11679 }, { "epoch": 0.1168, "grad_norm": 1.0519616690909357, "learning_rate": 0.003, "loss": 4.0947, "step": 11680 }, { "epoch": 0.11681, "grad_norm": 1.2549289807301722, "learning_rate": 0.003, "loss": 4.0676, "step": 11681 }, { "epoch": 0.11682, "grad_norm": 0.9579540582874558, "learning_rate": 0.003, "loss": 4.0706, "step": 11682 }, { "epoch": 0.11683, "grad_norm": 1.2352716839598907, "learning_rate": 0.003, "loss": 4.1017, "step": 11683 }, { "epoch": 0.11684, "grad_norm": 1.2156498815663923, "learning_rate": 0.003, "loss": 4.0798, "step": 11684 }, { "epoch": 0.11685, "grad_norm": 1.0363718832082467, "learning_rate": 0.003, "loss": 4.0696, "step": 11685 }, { "epoch": 0.11686, "grad_norm": 1.3470193086825817, "learning_rate": 0.003, "loss": 4.0853, "step": 11686 }, { "epoch": 0.11687, "grad_norm": 1.336107332557728, "learning_rate": 0.003, "loss": 4.0934, "step": 11687 }, { "epoch": 0.11688, "grad_norm": 1.148676523864226, "learning_rate": 0.003, "loss": 4.0892, "step": 11688 }, { "epoch": 0.11689, "grad_norm": 1.2732728738773667, "learning_rate": 0.003, "loss": 4.0899, "step": 11689 }, { "epoch": 0.1169, "grad_norm": 0.9761027416337912, "learning_rate": 0.003, "loss": 4.0766, "step": 11690 }, { "epoch": 0.11691, "grad_norm": 1.194763358774187, "learning_rate": 0.003, "loss": 4.0605, "step": 11691 }, { "epoch": 0.11692, "grad_norm": 1.1155053422367112, "learning_rate": 0.003, "loss": 4.0657, "step": 11692 }, { "epoch": 0.11693, "grad_norm": 1.0844912377130445, "learning_rate": 0.003, "loss": 4.0781, "step": 11693 }, { "epoch": 0.11694, "grad_norm": 1.360219760433983, "learning_rate": 0.003, "loss": 4.0723, "step": 11694 }, { "epoch": 0.11695, "grad_norm": 1.083480128114835, "learning_rate": 0.003, "loss": 4.0762, "step": 11695 }, { "epoch": 0.11696, "grad_norm": 1.209946409459336, "learning_rate": 0.003, "loss": 4.0818, "step": 11696 }, { "epoch": 0.11697, "grad_norm": 1.3042665518425522, "learning_rate": 0.003, "loss": 4.0503, "step": 11697 }, { "epoch": 0.11698, "grad_norm": 1.054254079024093, "learning_rate": 0.003, "loss": 4.0645, "step": 11698 }, { "epoch": 0.11699, "grad_norm": 1.4255708234494604, "learning_rate": 0.003, "loss": 4.0758, "step": 11699 }, { "epoch": 0.117, "grad_norm": 0.9222276852629732, "learning_rate": 0.003, "loss": 4.0743, "step": 11700 }, { "epoch": 0.11701, "grad_norm": 1.0913345591584294, "learning_rate": 0.003, "loss": 4.0757, "step": 11701 }, { "epoch": 0.11702, "grad_norm": 1.2159577552464076, "learning_rate": 0.003, "loss": 4.0767, "step": 11702 }, { "epoch": 0.11703, "grad_norm": 1.2755220220242691, "learning_rate": 0.003, "loss": 4.0589, "step": 11703 }, { "epoch": 0.11704, "grad_norm": 1.1010394356103956, "learning_rate": 0.003, "loss": 4.0956, "step": 11704 }, { "epoch": 0.11705, "grad_norm": 1.181555190974642, "learning_rate": 0.003, "loss": 4.0624, "step": 11705 }, { "epoch": 0.11706, "grad_norm": 1.355731758915925, "learning_rate": 0.003, "loss": 4.0945, "step": 11706 }, { "epoch": 0.11707, "grad_norm": 1.0451505789149136, "learning_rate": 0.003, "loss": 4.0738, "step": 11707 }, { "epoch": 0.11708, "grad_norm": 1.1674014223340736, "learning_rate": 0.003, "loss": 4.0601, "step": 11708 }, { "epoch": 0.11709, "grad_norm": 1.2271039669137753, "learning_rate": 0.003, "loss": 4.07, "step": 11709 }, { "epoch": 0.1171, "grad_norm": 1.290536541283088, "learning_rate": 0.003, "loss": 4.0778, "step": 11710 }, { "epoch": 0.11711, "grad_norm": 1.192079023146588, "learning_rate": 0.003, "loss": 4.0473, "step": 11711 }, { "epoch": 0.11712, "grad_norm": 0.9538550952961976, "learning_rate": 0.003, "loss": 4.1077, "step": 11712 }, { "epoch": 0.11713, "grad_norm": 1.1694628178896578, "learning_rate": 0.003, "loss": 4.0775, "step": 11713 }, { "epoch": 0.11714, "grad_norm": 1.1266996139387908, "learning_rate": 0.003, "loss": 4.0643, "step": 11714 }, { "epoch": 0.11715, "grad_norm": 1.101302435927173, "learning_rate": 0.003, "loss": 4.0786, "step": 11715 }, { "epoch": 0.11716, "grad_norm": 1.416114823275292, "learning_rate": 0.003, "loss": 4.0454, "step": 11716 }, { "epoch": 0.11717, "grad_norm": 0.8770688127299036, "learning_rate": 0.003, "loss": 4.0533, "step": 11717 }, { "epoch": 0.11718, "grad_norm": 0.9247194966585638, "learning_rate": 0.003, "loss": 4.0706, "step": 11718 }, { "epoch": 0.11719, "grad_norm": 1.1420126400773882, "learning_rate": 0.003, "loss": 4.0549, "step": 11719 }, { "epoch": 0.1172, "grad_norm": 1.2430497400012208, "learning_rate": 0.003, "loss": 4.0598, "step": 11720 }, { "epoch": 0.11721, "grad_norm": 1.3437462196938905, "learning_rate": 0.003, "loss": 4.0746, "step": 11721 }, { "epoch": 0.11722, "grad_norm": 1.1611268169534532, "learning_rate": 0.003, "loss": 4.0923, "step": 11722 }, { "epoch": 0.11723, "grad_norm": 1.1589328188926054, "learning_rate": 0.003, "loss": 4.0677, "step": 11723 }, { "epoch": 0.11724, "grad_norm": 1.1234276251597817, "learning_rate": 0.003, "loss": 4.0585, "step": 11724 }, { "epoch": 0.11725, "grad_norm": 1.392314288157773, "learning_rate": 0.003, "loss": 4.0721, "step": 11725 }, { "epoch": 0.11726, "grad_norm": 1.2016337841307816, "learning_rate": 0.003, "loss": 4.0793, "step": 11726 }, { "epoch": 0.11727, "grad_norm": 1.1672506920181773, "learning_rate": 0.003, "loss": 4.0982, "step": 11727 }, { "epoch": 0.11728, "grad_norm": 1.177675554505713, "learning_rate": 0.003, "loss": 4.0573, "step": 11728 }, { "epoch": 0.11729, "grad_norm": 1.1596721790007607, "learning_rate": 0.003, "loss": 4.0867, "step": 11729 }, { "epoch": 0.1173, "grad_norm": 1.2114792058895776, "learning_rate": 0.003, "loss": 4.0868, "step": 11730 }, { "epoch": 0.11731, "grad_norm": 1.2697478329728649, "learning_rate": 0.003, "loss": 4.0816, "step": 11731 }, { "epoch": 0.11732, "grad_norm": 1.2046703193379353, "learning_rate": 0.003, "loss": 4.0698, "step": 11732 }, { "epoch": 0.11733, "grad_norm": 1.1587625630956098, "learning_rate": 0.003, "loss": 4.0868, "step": 11733 }, { "epoch": 0.11734, "grad_norm": 0.9835383110437321, "learning_rate": 0.003, "loss": 4.0791, "step": 11734 }, { "epoch": 0.11735, "grad_norm": 1.1354652923916457, "learning_rate": 0.003, "loss": 4.0618, "step": 11735 }, { "epoch": 0.11736, "grad_norm": 1.1795068819900612, "learning_rate": 0.003, "loss": 4.0831, "step": 11736 }, { "epoch": 0.11737, "grad_norm": 0.9590724542826325, "learning_rate": 0.003, "loss": 4.0562, "step": 11737 }, { "epoch": 0.11738, "grad_norm": 1.2529064551602063, "learning_rate": 0.003, "loss": 4.0905, "step": 11738 }, { "epoch": 0.11739, "grad_norm": 0.9692882883964946, "learning_rate": 0.003, "loss": 4.0795, "step": 11739 }, { "epoch": 0.1174, "grad_norm": 1.3363662322865228, "learning_rate": 0.003, "loss": 4.0613, "step": 11740 }, { "epoch": 0.11741, "grad_norm": 1.1388596723511775, "learning_rate": 0.003, "loss": 4.073, "step": 11741 }, { "epoch": 0.11742, "grad_norm": 1.2061449922738792, "learning_rate": 0.003, "loss": 4.094, "step": 11742 }, { "epoch": 0.11743, "grad_norm": 1.2673513330553907, "learning_rate": 0.003, "loss": 4.0782, "step": 11743 }, { "epoch": 0.11744, "grad_norm": 0.9992619777221167, "learning_rate": 0.003, "loss": 4.0722, "step": 11744 }, { "epoch": 0.11745, "grad_norm": 1.4369939742498439, "learning_rate": 0.003, "loss": 4.0791, "step": 11745 }, { "epoch": 0.11746, "grad_norm": 1.0352605145126343, "learning_rate": 0.003, "loss": 4.052, "step": 11746 }, { "epoch": 0.11747, "grad_norm": 1.372321493331476, "learning_rate": 0.003, "loss": 4.0734, "step": 11747 }, { "epoch": 0.11748, "grad_norm": 1.391876422983213, "learning_rate": 0.003, "loss": 4.0711, "step": 11748 }, { "epoch": 0.11749, "grad_norm": 1.135363815029667, "learning_rate": 0.003, "loss": 4.0706, "step": 11749 }, { "epoch": 0.1175, "grad_norm": 1.242728276892211, "learning_rate": 0.003, "loss": 4.0961, "step": 11750 }, { "epoch": 0.11751, "grad_norm": 1.1148012860516774, "learning_rate": 0.003, "loss": 4.0588, "step": 11751 }, { "epoch": 0.11752, "grad_norm": 1.3070790394082443, "learning_rate": 0.003, "loss": 4.0751, "step": 11752 }, { "epoch": 0.11753, "grad_norm": 0.9992661935620242, "learning_rate": 0.003, "loss": 4.1025, "step": 11753 }, { "epoch": 0.11754, "grad_norm": 1.134067039387444, "learning_rate": 0.003, "loss": 4.0807, "step": 11754 }, { "epoch": 0.11755, "grad_norm": 0.9589451796726519, "learning_rate": 0.003, "loss": 4.0947, "step": 11755 }, { "epoch": 0.11756, "grad_norm": 1.3568027476997095, "learning_rate": 0.003, "loss": 4.0968, "step": 11756 }, { "epoch": 0.11757, "grad_norm": 1.1644018380445698, "learning_rate": 0.003, "loss": 4.0766, "step": 11757 }, { "epoch": 0.11758, "grad_norm": 1.2216645892129108, "learning_rate": 0.003, "loss": 4.1069, "step": 11758 }, { "epoch": 0.11759, "grad_norm": 1.1649690820165117, "learning_rate": 0.003, "loss": 4.0959, "step": 11759 }, { "epoch": 0.1176, "grad_norm": 1.4111637333354248, "learning_rate": 0.003, "loss": 4.0811, "step": 11760 }, { "epoch": 0.11761, "grad_norm": 1.1386523484876632, "learning_rate": 0.003, "loss": 4.0652, "step": 11761 }, { "epoch": 0.11762, "grad_norm": 1.309097793898003, "learning_rate": 0.003, "loss": 4.0716, "step": 11762 }, { "epoch": 0.11763, "grad_norm": 1.0977987428654228, "learning_rate": 0.003, "loss": 4.0721, "step": 11763 }, { "epoch": 0.11764, "grad_norm": 1.3897224575122786, "learning_rate": 0.003, "loss": 4.0834, "step": 11764 }, { "epoch": 0.11765, "grad_norm": 1.1526270512131422, "learning_rate": 0.003, "loss": 4.09, "step": 11765 }, { "epoch": 0.11766, "grad_norm": 1.0482334575456738, "learning_rate": 0.003, "loss": 4.0799, "step": 11766 }, { "epoch": 0.11767, "grad_norm": 1.1680362709673526, "learning_rate": 0.003, "loss": 4.0868, "step": 11767 }, { "epoch": 0.11768, "grad_norm": 0.8998671971903655, "learning_rate": 0.003, "loss": 4.0681, "step": 11768 }, { "epoch": 0.11769, "grad_norm": 1.2109642683247706, "learning_rate": 0.003, "loss": 4.0749, "step": 11769 }, { "epoch": 0.1177, "grad_norm": 1.1568339451894927, "learning_rate": 0.003, "loss": 4.0763, "step": 11770 }, { "epoch": 0.11771, "grad_norm": 1.2552008359510307, "learning_rate": 0.003, "loss": 4.0518, "step": 11771 }, { "epoch": 0.11772, "grad_norm": 1.140799491460487, "learning_rate": 0.003, "loss": 4.0834, "step": 11772 }, { "epoch": 0.11773, "grad_norm": 1.3359287045550823, "learning_rate": 0.003, "loss": 4.0861, "step": 11773 }, { "epoch": 0.11774, "grad_norm": 0.9020151031780287, "learning_rate": 0.003, "loss": 4.0681, "step": 11774 }, { "epoch": 0.11775, "grad_norm": 1.039904881152444, "learning_rate": 0.003, "loss": 4.0668, "step": 11775 }, { "epoch": 0.11776, "grad_norm": 1.1772503869017166, "learning_rate": 0.003, "loss": 4.0696, "step": 11776 }, { "epoch": 0.11777, "grad_norm": 1.0978295001721685, "learning_rate": 0.003, "loss": 4.0742, "step": 11777 }, { "epoch": 0.11778, "grad_norm": 1.2932395259715683, "learning_rate": 0.003, "loss": 4.0927, "step": 11778 }, { "epoch": 0.11779, "grad_norm": 1.062639311748465, "learning_rate": 0.003, "loss": 4.0718, "step": 11779 }, { "epoch": 0.1178, "grad_norm": 1.5522142112093957, "learning_rate": 0.003, "loss": 4.067, "step": 11780 }, { "epoch": 0.11781, "grad_norm": 1.1901849556182456, "learning_rate": 0.003, "loss": 4.0638, "step": 11781 }, { "epoch": 0.11782, "grad_norm": 1.1360741091738087, "learning_rate": 0.003, "loss": 4.0897, "step": 11782 }, { "epoch": 0.11783, "grad_norm": 1.2956834409184492, "learning_rate": 0.003, "loss": 4.1031, "step": 11783 }, { "epoch": 0.11784, "grad_norm": 0.9080434495652585, "learning_rate": 0.003, "loss": 4.0585, "step": 11784 }, { "epoch": 0.11785, "grad_norm": 1.3873971037919273, "learning_rate": 0.003, "loss": 4.0741, "step": 11785 }, { "epoch": 0.11786, "grad_norm": 0.9221289765526292, "learning_rate": 0.003, "loss": 4.0677, "step": 11786 }, { "epoch": 0.11787, "grad_norm": 1.0909130831395086, "learning_rate": 0.003, "loss": 4.0954, "step": 11787 }, { "epoch": 0.11788, "grad_norm": 1.1162084751400934, "learning_rate": 0.003, "loss": 4.0614, "step": 11788 }, { "epoch": 0.11789, "grad_norm": 1.0164636722393448, "learning_rate": 0.003, "loss": 4.056, "step": 11789 }, { "epoch": 0.1179, "grad_norm": 1.289296214700985, "learning_rate": 0.003, "loss": 4.0934, "step": 11790 }, { "epoch": 0.11791, "grad_norm": 1.197526365357239, "learning_rate": 0.003, "loss": 4.0662, "step": 11791 }, { "epoch": 0.11792, "grad_norm": 1.3794954189831017, "learning_rate": 0.003, "loss": 4.0881, "step": 11792 }, { "epoch": 0.11793, "grad_norm": 1.3379948580447394, "learning_rate": 0.003, "loss": 4.0715, "step": 11793 }, { "epoch": 0.11794, "grad_norm": 0.9973265523988945, "learning_rate": 0.003, "loss": 4.0727, "step": 11794 }, { "epoch": 0.11795, "grad_norm": 1.0920690685918633, "learning_rate": 0.003, "loss": 4.0836, "step": 11795 }, { "epoch": 0.11796, "grad_norm": 1.175601375502935, "learning_rate": 0.003, "loss": 4.0974, "step": 11796 }, { "epoch": 0.11797, "grad_norm": 1.1703415868934268, "learning_rate": 0.003, "loss": 4.0573, "step": 11797 }, { "epoch": 0.11798, "grad_norm": 1.3008008929070882, "learning_rate": 0.003, "loss": 4.0728, "step": 11798 }, { "epoch": 0.11799, "grad_norm": 1.0314930400630682, "learning_rate": 0.003, "loss": 4.0998, "step": 11799 }, { "epoch": 0.118, "grad_norm": 1.1375274780427134, "learning_rate": 0.003, "loss": 4.0841, "step": 11800 }, { "epoch": 0.11801, "grad_norm": 1.0380180520004971, "learning_rate": 0.003, "loss": 4.0417, "step": 11801 }, { "epoch": 0.11802, "grad_norm": 1.2499064995107485, "learning_rate": 0.003, "loss": 4.0743, "step": 11802 }, { "epoch": 0.11803, "grad_norm": 0.9206344380911405, "learning_rate": 0.003, "loss": 4.0896, "step": 11803 }, { "epoch": 0.11804, "grad_norm": 1.0563731974738848, "learning_rate": 0.003, "loss": 4.0825, "step": 11804 }, { "epoch": 0.11805, "grad_norm": 1.2138022475057453, "learning_rate": 0.003, "loss": 4.0602, "step": 11805 }, { "epoch": 0.11806, "grad_norm": 1.2367306427185214, "learning_rate": 0.003, "loss": 4.0817, "step": 11806 }, { "epoch": 0.11807, "grad_norm": 1.2203777945761236, "learning_rate": 0.003, "loss": 4.0808, "step": 11807 }, { "epoch": 0.11808, "grad_norm": 1.0586284436372069, "learning_rate": 0.003, "loss": 4.0595, "step": 11808 }, { "epoch": 0.11809, "grad_norm": 1.1556376689324874, "learning_rate": 0.003, "loss": 4.0927, "step": 11809 }, { "epoch": 0.1181, "grad_norm": 1.0478388340716422, "learning_rate": 0.003, "loss": 4.0688, "step": 11810 }, { "epoch": 0.11811, "grad_norm": 0.9796072523473666, "learning_rate": 0.003, "loss": 4.087, "step": 11811 }, { "epoch": 0.11812, "grad_norm": 1.3267520540589823, "learning_rate": 0.003, "loss": 4.0853, "step": 11812 }, { "epoch": 0.11813, "grad_norm": 1.0624381801166245, "learning_rate": 0.003, "loss": 4.0783, "step": 11813 }, { "epoch": 0.11814, "grad_norm": 1.4611375132742075, "learning_rate": 0.003, "loss": 4.0768, "step": 11814 }, { "epoch": 0.11815, "grad_norm": 1.5778823206110453, "learning_rate": 0.003, "loss": 4.0983, "step": 11815 }, { "epoch": 0.11816, "grad_norm": 1.2354906198845372, "learning_rate": 0.003, "loss": 4.0616, "step": 11816 }, { "epoch": 0.11817, "grad_norm": 1.1693476173681472, "learning_rate": 0.003, "loss": 4.0605, "step": 11817 }, { "epoch": 0.11818, "grad_norm": 1.1420531165461065, "learning_rate": 0.003, "loss": 4.0791, "step": 11818 }, { "epoch": 0.11819, "grad_norm": 1.2909973798684644, "learning_rate": 0.003, "loss": 4.0462, "step": 11819 }, { "epoch": 0.1182, "grad_norm": 1.2598501868393714, "learning_rate": 0.003, "loss": 4.0861, "step": 11820 }, { "epoch": 0.11821, "grad_norm": 1.0835462615430216, "learning_rate": 0.003, "loss": 4.0395, "step": 11821 }, { "epoch": 0.11822, "grad_norm": 1.2044057463788023, "learning_rate": 0.003, "loss": 4.0827, "step": 11822 }, { "epoch": 0.11823, "grad_norm": 1.0910825742504715, "learning_rate": 0.003, "loss": 4.0664, "step": 11823 }, { "epoch": 0.11824, "grad_norm": 1.2386646366156195, "learning_rate": 0.003, "loss": 4.065, "step": 11824 }, { "epoch": 0.11825, "grad_norm": 1.0416502202053222, "learning_rate": 0.003, "loss": 4.0552, "step": 11825 }, { "epoch": 0.11826, "grad_norm": 1.2848901046610308, "learning_rate": 0.003, "loss": 4.0805, "step": 11826 }, { "epoch": 0.11827, "grad_norm": 1.2023286965580762, "learning_rate": 0.003, "loss": 4.0886, "step": 11827 }, { "epoch": 0.11828, "grad_norm": 1.2356902108526424, "learning_rate": 0.003, "loss": 4.091, "step": 11828 }, { "epoch": 0.11829, "grad_norm": 1.088816813835146, "learning_rate": 0.003, "loss": 4.0925, "step": 11829 }, { "epoch": 0.1183, "grad_norm": 1.3892098766726853, "learning_rate": 0.003, "loss": 4.086, "step": 11830 }, { "epoch": 0.11831, "grad_norm": 1.1409473182564889, "learning_rate": 0.003, "loss": 4.079, "step": 11831 }, { "epoch": 0.11832, "grad_norm": 0.9492869584044819, "learning_rate": 0.003, "loss": 4.0537, "step": 11832 }, { "epoch": 0.11833, "grad_norm": 1.1940304123684042, "learning_rate": 0.003, "loss": 4.1049, "step": 11833 }, { "epoch": 0.11834, "grad_norm": 1.3422321290193753, "learning_rate": 0.003, "loss": 4.1046, "step": 11834 }, { "epoch": 0.11835, "grad_norm": 0.9584281942908637, "learning_rate": 0.003, "loss": 4.0575, "step": 11835 }, { "epoch": 0.11836, "grad_norm": 1.2123845815097705, "learning_rate": 0.003, "loss": 4.0384, "step": 11836 }, { "epoch": 0.11837, "grad_norm": 1.2698491999443162, "learning_rate": 0.003, "loss": 4.0784, "step": 11837 }, { "epoch": 0.11838, "grad_norm": 1.3122355602083897, "learning_rate": 0.003, "loss": 4.0589, "step": 11838 }, { "epoch": 0.11839, "grad_norm": 1.105366259392173, "learning_rate": 0.003, "loss": 4.0719, "step": 11839 }, { "epoch": 0.1184, "grad_norm": 1.200875820871659, "learning_rate": 0.003, "loss": 4.0663, "step": 11840 }, { "epoch": 0.11841, "grad_norm": 1.1742575594715312, "learning_rate": 0.003, "loss": 4.0767, "step": 11841 }, { "epoch": 0.11842, "grad_norm": 1.0467822742428252, "learning_rate": 0.003, "loss": 4.0594, "step": 11842 }, { "epoch": 0.11843, "grad_norm": 1.2674758216305462, "learning_rate": 0.003, "loss": 4.0731, "step": 11843 }, { "epoch": 0.11844, "grad_norm": 0.8946428012259164, "learning_rate": 0.003, "loss": 4.053, "step": 11844 }, { "epoch": 0.11845, "grad_norm": 1.5005655593944138, "learning_rate": 0.003, "loss": 4.0934, "step": 11845 }, { "epoch": 0.11846, "grad_norm": 0.985039493861634, "learning_rate": 0.003, "loss": 4.0765, "step": 11846 }, { "epoch": 0.11847, "grad_norm": 1.420682429044482, "learning_rate": 0.003, "loss": 4.0926, "step": 11847 }, { "epoch": 0.11848, "grad_norm": 1.1624676672252878, "learning_rate": 0.003, "loss": 4.1019, "step": 11848 }, { "epoch": 0.11849, "grad_norm": 1.3452780755205915, "learning_rate": 0.003, "loss": 4.0988, "step": 11849 }, { "epoch": 0.1185, "grad_norm": 1.0704095122382435, "learning_rate": 0.003, "loss": 4.0786, "step": 11850 }, { "epoch": 0.11851, "grad_norm": 1.0919642793213178, "learning_rate": 0.003, "loss": 4.0607, "step": 11851 }, { "epoch": 0.11852, "grad_norm": 1.4400788622045355, "learning_rate": 0.003, "loss": 4.0694, "step": 11852 }, { "epoch": 0.11853, "grad_norm": 1.1511166394484542, "learning_rate": 0.003, "loss": 4.08, "step": 11853 }, { "epoch": 0.11854, "grad_norm": 1.2657967835115171, "learning_rate": 0.003, "loss": 4.0636, "step": 11854 }, { "epoch": 0.11855, "grad_norm": 1.061771798870032, "learning_rate": 0.003, "loss": 4.0916, "step": 11855 }, { "epoch": 0.11856, "grad_norm": 1.1199224960482557, "learning_rate": 0.003, "loss": 4.0587, "step": 11856 }, { "epoch": 0.11857, "grad_norm": 1.22166661194193, "learning_rate": 0.003, "loss": 4.0731, "step": 11857 }, { "epoch": 0.11858, "grad_norm": 1.0497049888388341, "learning_rate": 0.003, "loss": 4.1052, "step": 11858 }, { "epoch": 0.11859, "grad_norm": 1.3246403958345156, "learning_rate": 0.003, "loss": 4.0906, "step": 11859 }, { "epoch": 0.1186, "grad_norm": 1.0904610992730863, "learning_rate": 0.003, "loss": 4.0732, "step": 11860 }, { "epoch": 0.11861, "grad_norm": 1.1682035816854246, "learning_rate": 0.003, "loss": 4.0666, "step": 11861 }, { "epoch": 0.11862, "grad_norm": 1.129212585599079, "learning_rate": 0.003, "loss": 4.0915, "step": 11862 }, { "epoch": 0.11863, "grad_norm": 1.0838812912698164, "learning_rate": 0.003, "loss": 4.0951, "step": 11863 }, { "epoch": 0.11864, "grad_norm": 1.0755256728268683, "learning_rate": 0.003, "loss": 4.0728, "step": 11864 }, { "epoch": 0.11865, "grad_norm": 1.2686653387821047, "learning_rate": 0.003, "loss": 4.0764, "step": 11865 }, { "epoch": 0.11866, "grad_norm": 1.082041014866532, "learning_rate": 0.003, "loss": 4.0765, "step": 11866 }, { "epoch": 0.11867, "grad_norm": 1.224353513755153, "learning_rate": 0.003, "loss": 4.0786, "step": 11867 }, { "epoch": 0.11868, "grad_norm": 1.3885837099920573, "learning_rate": 0.003, "loss": 4.0756, "step": 11868 }, { "epoch": 0.11869, "grad_norm": 1.1683078148841186, "learning_rate": 0.003, "loss": 4.0837, "step": 11869 }, { "epoch": 0.1187, "grad_norm": 1.464726853426062, "learning_rate": 0.003, "loss": 4.0903, "step": 11870 }, { "epoch": 0.11871, "grad_norm": 0.9331683604512, "learning_rate": 0.003, "loss": 4.0602, "step": 11871 }, { "epoch": 0.11872, "grad_norm": 0.9556768456530206, "learning_rate": 0.003, "loss": 4.0883, "step": 11872 }, { "epoch": 0.11873, "grad_norm": 1.1912510423155762, "learning_rate": 0.003, "loss": 4.0592, "step": 11873 }, { "epoch": 0.11874, "grad_norm": 1.1284462075181023, "learning_rate": 0.003, "loss": 4.0398, "step": 11874 }, { "epoch": 0.11875, "grad_norm": 1.3263536085256413, "learning_rate": 0.003, "loss": 4.0862, "step": 11875 }, { "epoch": 0.11876, "grad_norm": 1.338889444852054, "learning_rate": 0.003, "loss": 4.0554, "step": 11876 }, { "epoch": 0.11877, "grad_norm": 1.2191734588981227, "learning_rate": 0.003, "loss": 4.102, "step": 11877 }, { "epoch": 0.11878, "grad_norm": 1.1458563136253266, "learning_rate": 0.003, "loss": 4.0745, "step": 11878 }, { "epoch": 0.11879, "grad_norm": 1.0174231879677096, "learning_rate": 0.003, "loss": 4.0889, "step": 11879 }, { "epoch": 0.1188, "grad_norm": 1.2214455142653529, "learning_rate": 0.003, "loss": 4.0772, "step": 11880 }, { "epoch": 0.11881, "grad_norm": 1.006528896792693, "learning_rate": 0.003, "loss": 4.0677, "step": 11881 }, { "epoch": 0.11882, "grad_norm": 1.4040376973932613, "learning_rate": 0.003, "loss": 4.0663, "step": 11882 }, { "epoch": 0.11883, "grad_norm": 1.028507125117693, "learning_rate": 0.003, "loss": 4.0936, "step": 11883 }, { "epoch": 0.11884, "grad_norm": 1.215953435707132, "learning_rate": 0.003, "loss": 4.067, "step": 11884 }, { "epoch": 0.11885, "grad_norm": 1.1170954606641776, "learning_rate": 0.003, "loss": 4.1036, "step": 11885 }, { "epoch": 0.11886, "grad_norm": 1.0222720334577216, "learning_rate": 0.003, "loss": 4.0722, "step": 11886 }, { "epoch": 0.11887, "grad_norm": 1.1071360827090961, "learning_rate": 0.003, "loss": 4.0538, "step": 11887 }, { "epoch": 0.11888, "grad_norm": 1.3166253130255614, "learning_rate": 0.003, "loss": 4.1053, "step": 11888 }, { "epoch": 0.11889, "grad_norm": 1.2640837305337336, "learning_rate": 0.003, "loss": 4.0921, "step": 11889 }, { "epoch": 0.1189, "grad_norm": 1.0385008572054317, "learning_rate": 0.003, "loss": 4.0944, "step": 11890 }, { "epoch": 0.11891, "grad_norm": 1.2492202823351668, "learning_rate": 0.003, "loss": 4.0733, "step": 11891 }, { "epoch": 0.11892, "grad_norm": 1.2917337053073015, "learning_rate": 0.003, "loss": 4.0801, "step": 11892 }, { "epoch": 0.11893, "grad_norm": 1.133444321773112, "learning_rate": 0.003, "loss": 4.0771, "step": 11893 }, { "epoch": 0.11894, "grad_norm": 1.2816714042210788, "learning_rate": 0.003, "loss": 4.0551, "step": 11894 }, { "epoch": 0.11895, "grad_norm": 1.1511928847349535, "learning_rate": 0.003, "loss": 4.0653, "step": 11895 }, { "epoch": 0.11896, "grad_norm": 1.1723599609058912, "learning_rate": 0.003, "loss": 4.0925, "step": 11896 }, { "epoch": 0.11897, "grad_norm": 1.0912912177129093, "learning_rate": 0.003, "loss": 4.0799, "step": 11897 }, { "epoch": 0.11898, "grad_norm": 1.2583528065403673, "learning_rate": 0.003, "loss": 4.0926, "step": 11898 }, { "epoch": 0.11899, "grad_norm": 1.2696257492886351, "learning_rate": 0.003, "loss": 4.0465, "step": 11899 }, { "epoch": 0.119, "grad_norm": 1.1657896014255376, "learning_rate": 0.003, "loss": 4.0744, "step": 11900 }, { "epoch": 0.11901, "grad_norm": 1.2337083368973147, "learning_rate": 0.003, "loss": 4.0939, "step": 11901 }, { "epoch": 0.11902, "grad_norm": 1.0673704845935785, "learning_rate": 0.003, "loss": 4.0556, "step": 11902 }, { "epoch": 0.11903, "grad_norm": 1.2072025133556539, "learning_rate": 0.003, "loss": 4.0766, "step": 11903 }, { "epoch": 0.11904, "grad_norm": 0.9622886881189869, "learning_rate": 0.003, "loss": 4.0722, "step": 11904 }, { "epoch": 0.11905, "grad_norm": 1.5013454950863534, "learning_rate": 0.003, "loss": 4.0742, "step": 11905 }, { "epoch": 0.11906, "grad_norm": 0.9027233765240715, "learning_rate": 0.003, "loss": 4.0713, "step": 11906 }, { "epoch": 0.11907, "grad_norm": 1.2764946041770708, "learning_rate": 0.003, "loss": 4.0916, "step": 11907 }, { "epoch": 0.11908, "grad_norm": 1.0055649877091324, "learning_rate": 0.003, "loss": 4.0913, "step": 11908 }, { "epoch": 0.11909, "grad_norm": 1.5417677310497, "learning_rate": 0.003, "loss": 4.0488, "step": 11909 }, { "epoch": 0.1191, "grad_norm": 1.1311757539323997, "learning_rate": 0.003, "loss": 4.0938, "step": 11910 }, { "epoch": 0.11911, "grad_norm": 1.1679816226515805, "learning_rate": 0.003, "loss": 4.0775, "step": 11911 }, { "epoch": 0.11912, "grad_norm": 1.0541786047876225, "learning_rate": 0.003, "loss": 4.1158, "step": 11912 }, { "epoch": 0.11913, "grad_norm": 1.2322905506078932, "learning_rate": 0.003, "loss": 4.0987, "step": 11913 }, { "epoch": 0.11914, "grad_norm": 1.3491037394150307, "learning_rate": 0.003, "loss": 4.1039, "step": 11914 }, { "epoch": 0.11915, "grad_norm": 0.8938748705932589, "learning_rate": 0.003, "loss": 4.0829, "step": 11915 }, { "epoch": 0.11916, "grad_norm": 1.0615958706145021, "learning_rate": 0.003, "loss": 4.0945, "step": 11916 }, { "epoch": 0.11917, "grad_norm": 1.3243478475913795, "learning_rate": 0.003, "loss": 4.0976, "step": 11917 }, { "epoch": 0.11918, "grad_norm": 0.9752861321174783, "learning_rate": 0.003, "loss": 4.0771, "step": 11918 }, { "epoch": 0.11919, "grad_norm": 1.1617839452425116, "learning_rate": 0.003, "loss": 4.0851, "step": 11919 }, { "epoch": 0.1192, "grad_norm": 1.2323714936097194, "learning_rate": 0.003, "loss": 4.0875, "step": 11920 }, { "epoch": 0.11921, "grad_norm": 1.2292482817665966, "learning_rate": 0.003, "loss": 4.0466, "step": 11921 }, { "epoch": 0.11922, "grad_norm": 1.2663608044619241, "learning_rate": 0.003, "loss": 4.0962, "step": 11922 }, { "epoch": 0.11923, "grad_norm": 1.0587798818379033, "learning_rate": 0.003, "loss": 4.0334, "step": 11923 }, { "epoch": 0.11924, "grad_norm": 1.0895636196749694, "learning_rate": 0.003, "loss": 4.0648, "step": 11924 }, { "epoch": 0.11925, "grad_norm": 1.1325718267024283, "learning_rate": 0.003, "loss": 4.0844, "step": 11925 }, { "epoch": 0.11926, "grad_norm": 1.3151618929161546, "learning_rate": 0.003, "loss": 4.0766, "step": 11926 }, { "epoch": 0.11927, "grad_norm": 1.001849668904841, "learning_rate": 0.003, "loss": 4.0832, "step": 11927 }, { "epoch": 0.11928, "grad_norm": 1.4619761719316582, "learning_rate": 0.003, "loss": 4.0738, "step": 11928 }, { "epoch": 0.11929, "grad_norm": 0.930796532554024, "learning_rate": 0.003, "loss": 4.0735, "step": 11929 }, { "epoch": 0.1193, "grad_norm": 1.2378138922895818, "learning_rate": 0.003, "loss": 4.0789, "step": 11930 }, { "epoch": 0.11931, "grad_norm": 1.146907374465275, "learning_rate": 0.003, "loss": 4.0696, "step": 11931 }, { "epoch": 0.11932, "grad_norm": 0.928290416437596, "learning_rate": 0.003, "loss": 4.0804, "step": 11932 }, { "epoch": 0.11933, "grad_norm": 1.0338660433415998, "learning_rate": 0.003, "loss": 4.0525, "step": 11933 }, { "epoch": 0.11934, "grad_norm": 1.2049870696278178, "learning_rate": 0.003, "loss": 4.0953, "step": 11934 }, { "epoch": 0.11935, "grad_norm": 1.3410424655125115, "learning_rate": 0.003, "loss": 4.0659, "step": 11935 }, { "epoch": 0.11936, "grad_norm": 1.0913007430410453, "learning_rate": 0.003, "loss": 4.0713, "step": 11936 }, { "epoch": 0.11937, "grad_norm": 1.3956547335981675, "learning_rate": 0.003, "loss": 4.092, "step": 11937 }, { "epoch": 0.11938, "grad_norm": 1.1526516474361628, "learning_rate": 0.003, "loss": 4.0486, "step": 11938 }, { "epoch": 0.11939, "grad_norm": 1.0661020181841054, "learning_rate": 0.003, "loss": 4.0618, "step": 11939 }, { "epoch": 0.1194, "grad_norm": 1.215621630171941, "learning_rate": 0.003, "loss": 4.0877, "step": 11940 }, { "epoch": 0.11941, "grad_norm": 1.3620176372301322, "learning_rate": 0.003, "loss": 4.0848, "step": 11941 }, { "epoch": 0.11942, "grad_norm": 1.076803965161367, "learning_rate": 0.003, "loss": 4.0911, "step": 11942 }, { "epoch": 0.11943, "grad_norm": 1.3137769906580026, "learning_rate": 0.003, "loss": 4.0695, "step": 11943 }, { "epoch": 0.11944, "grad_norm": 1.2734546183235818, "learning_rate": 0.003, "loss": 4.0745, "step": 11944 }, { "epoch": 0.11945, "grad_norm": 1.240548250296292, "learning_rate": 0.003, "loss": 4.0453, "step": 11945 }, { "epoch": 0.11946, "grad_norm": 1.2086492743738333, "learning_rate": 0.003, "loss": 4.0596, "step": 11946 }, { "epoch": 0.11947, "grad_norm": 1.0632751865289534, "learning_rate": 0.003, "loss": 4.087, "step": 11947 }, { "epoch": 0.11948, "grad_norm": 1.2100696179246582, "learning_rate": 0.003, "loss": 4.0786, "step": 11948 }, { "epoch": 0.11949, "grad_norm": 1.2304485972274908, "learning_rate": 0.003, "loss": 4.092, "step": 11949 }, { "epoch": 0.1195, "grad_norm": 1.1614419796519644, "learning_rate": 0.003, "loss": 4.0691, "step": 11950 }, { "epoch": 0.11951, "grad_norm": 0.9685120514135239, "learning_rate": 0.003, "loss": 4.0685, "step": 11951 }, { "epoch": 0.11952, "grad_norm": 1.1708872350226684, "learning_rate": 0.003, "loss": 4.0828, "step": 11952 }, { "epoch": 0.11953, "grad_norm": 1.1972019050800324, "learning_rate": 0.003, "loss": 4.0841, "step": 11953 }, { "epoch": 0.11954, "grad_norm": 1.1990332756494675, "learning_rate": 0.003, "loss": 4.0495, "step": 11954 }, { "epoch": 0.11955, "grad_norm": 1.1167038436736807, "learning_rate": 0.003, "loss": 4.0676, "step": 11955 }, { "epoch": 0.11956, "grad_norm": 1.138929420282013, "learning_rate": 0.003, "loss": 4.0871, "step": 11956 }, { "epoch": 0.11957, "grad_norm": 1.2896652846169314, "learning_rate": 0.003, "loss": 4.0584, "step": 11957 }, { "epoch": 0.11958, "grad_norm": 1.1849380305493635, "learning_rate": 0.003, "loss": 4.0687, "step": 11958 }, { "epoch": 0.11959, "grad_norm": 0.9725598923562588, "learning_rate": 0.003, "loss": 4.0397, "step": 11959 }, { "epoch": 0.1196, "grad_norm": 1.2310019243117136, "learning_rate": 0.003, "loss": 4.0575, "step": 11960 }, { "epoch": 0.11961, "grad_norm": 1.1828789776611688, "learning_rate": 0.003, "loss": 4.049, "step": 11961 }, { "epoch": 0.11962, "grad_norm": 1.207568363453171, "learning_rate": 0.003, "loss": 4.0907, "step": 11962 }, { "epoch": 0.11963, "grad_norm": 1.23441993888683, "learning_rate": 0.003, "loss": 4.0707, "step": 11963 }, { "epoch": 0.11964, "grad_norm": 1.081282831387257, "learning_rate": 0.003, "loss": 4.0611, "step": 11964 }, { "epoch": 0.11965, "grad_norm": 1.0644562331563587, "learning_rate": 0.003, "loss": 4.076, "step": 11965 }, { "epoch": 0.11966, "grad_norm": 1.2165695726517882, "learning_rate": 0.003, "loss": 4.0672, "step": 11966 }, { "epoch": 0.11967, "grad_norm": 1.1643580839905228, "learning_rate": 0.003, "loss": 4.0444, "step": 11967 }, { "epoch": 0.11968, "grad_norm": 1.0694960681873658, "learning_rate": 0.003, "loss": 4.0459, "step": 11968 }, { "epoch": 0.11969, "grad_norm": 1.2798763528913761, "learning_rate": 0.003, "loss": 4.0898, "step": 11969 }, { "epoch": 0.1197, "grad_norm": 1.0104359886256886, "learning_rate": 0.003, "loss": 4.0764, "step": 11970 }, { "epoch": 0.11971, "grad_norm": 1.4908390782821748, "learning_rate": 0.003, "loss": 4.0774, "step": 11971 }, { "epoch": 0.11972, "grad_norm": 1.1262404046440087, "learning_rate": 0.003, "loss": 4.0514, "step": 11972 }, { "epoch": 0.11973, "grad_norm": 1.3700582843926272, "learning_rate": 0.003, "loss": 4.0729, "step": 11973 }, { "epoch": 0.11974, "grad_norm": 1.042113991640309, "learning_rate": 0.003, "loss": 4.0594, "step": 11974 }, { "epoch": 0.11975, "grad_norm": 1.3250275421778095, "learning_rate": 0.003, "loss": 4.0661, "step": 11975 }, { "epoch": 0.11976, "grad_norm": 0.8820121746589544, "learning_rate": 0.003, "loss": 4.0642, "step": 11976 }, { "epoch": 0.11977, "grad_norm": 0.9874024178086407, "learning_rate": 0.003, "loss": 4.0627, "step": 11977 }, { "epoch": 0.11978, "grad_norm": 1.2509197830941192, "learning_rate": 0.003, "loss": 4.0732, "step": 11978 }, { "epoch": 0.11979, "grad_norm": 1.4301678995271991, "learning_rate": 0.003, "loss": 4.0633, "step": 11979 }, { "epoch": 0.1198, "grad_norm": 1.2225856918692968, "learning_rate": 0.003, "loss": 4.109, "step": 11980 }, { "epoch": 0.11981, "grad_norm": 0.9750623949395286, "learning_rate": 0.003, "loss": 4.0909, "step": 11981 }, { "epoch": 0.11982, "grad_norm": 1.1936375123088359, "learning_rate": 0.003, "loss": 4.078, "step": 11982 }, { "epoch": 0.11983, "grad_norm": 1.1280323217958304, "learning_rate": 0.003, "loss": 4.0919, "step": 11983 }, { "epoch": 0.11984, "grad_norm": 1.427743469061658, "learning_rate": 0.003, "loss": 4.0612, "step": 11984 }, { "epoch": 0.11985, "grad_norm": 1.003738628950787, "learning_rate": 0.003, "loss": 4.0707, "step": 11985 }, { "epoch": 0.11986, "grad_norm": 1.4284484153405892, "learning_rate": 0.003, "loss": 4.0864, "step": 11986 }, { "epoch": 0.11987, "grad_norm": 1.0369046324770255, "learning_rate": 0.003, "loss": 4.0744, "step": 11987 }, { "epoch": 0.11988, "grad_norm": 1.0507093817761055, "learning_rate": 0.003, "loss": 4.0548, "step": 11988 }, { "epoch": 0.11989, "grad_norm": 1.2157923701396773, "learning_rate": 0.003, "loss": 4.0886, "step": 11989 }, { "epoch": 0.1199, "grad_norm": 1.163075834187405, "learning_rate": 0.003, "loss": 4.0702, "step": 11990 }, { "epoch": 0.11991, "grad_norm": 1.1397923959795142, "learning_rate": 0.003, "loss": 4.0787, "step": 11991 }, { "epoch": 0.11992, "grad_norm": 1.1696029570490936, "learning_rate": 0.003, "loss": 4.1058, "step": 11992 }, { "epoch": 0.11993, "grad_norm": 1.0055737816247683, "learning_rate": 0.003, "loss": 4.0575, "step": 11993 }, { "epoch": 0.11994, "grad_norm": 1.4353057542413166, "learning_rate": 0.003, "loss": 4.0937, "step": 11994 }, { "epoch": 0.11995, "grad_norm": 1.0107043389029158, "learning_rate": 0.003, "loss": 4.0751, "step": 11995 }, { "epoch": 0.11996, "grad_norm": 1.3901873327466496, "learning_rate": 0.003, "loss": 4.0616, "step": 11996 }, { "epoch": 0.11997, "grad_norm": 1.0343379072321632, "learning_rate": 0.003, "loss": 4.0656, "step": 11997 }, { "epoch": 0.11998, "grad_norm": 1.1937023604692176, "learning_rate": 0.003, "loss": 4.0805, "step": 11998 }, { "epoch": 0.11999, "grad_norm": 1.40236106905162, "learning_rate": 0.003, "loss": 4.0982, "step": 11999 }, { "epoch": 0.12, "grad_norm": 1.026542270252335, "learning_rate": 0.003, "loss": 4.0638, "step": 12000 }, { "epoch": 0.12001, "grad_norm": 1.1423095300856636, "learning_rate": 0.003, "loss": 4.0678, "step": 12001 }, { "epoch": 0.12002, "grad_norm": 1.3492444697177328, "learning_rate": 0.003, "loss": 4.0659, "step": 12002 }, { "epoch": 0.12003, "grad_norm": 1.03747627559624, "learning_rate": 0.003, "loss": 4.0603, "step": 12003 }, { "epoch": 0.12004, "grad_norm": 1.098077915902504, "learning_rate": 0.003, "loss": 4.0648, "step": 12004 }, { "epoch": 0.12005, "grad_norm": 1.1083046035044317, "learning_rate": 0.003, "loss": 4.061, "step": 12005 }, { "epoch": 0.12006, "grad_norm": 1.2956516804135265, "learning_rate": 0.003, "loss": 4.0632, "step": 12006 }, { "epoch": 0.12007, "grad_norm": 1.218187648921116, "learning_rate": 0.003, "loss": 4.0725, "step": 12007 }, { "epoch": 0.12008, "grad_norm": 1.2671842173164862, "learning_rate": 0.003, "loss": 4.1, "step": 12008 }, { "epoch": 0.12009, "grad_norm": 1.0001998120171887, "learning_rate": 0.003, "loss": 4.0715, "step": 12009 }, { "epoch": 0.1201, "grad_norm": 1.2404288265445556, "learning_rate": 0.003, "loss": 4.0618, "step": 12010 }, { "epoch": 0.12011, "grad_norm": 1.186709338504087, "learning_rate": 0.003, "loss": 4.0593, "step": 12011 }, { "epoch": 0.12012, "grad_norm": 0.946874831717622, "learning_rate": 0.003, "loss": 4.0491, "step": 12012 }, { "epoch": 0.12013, "grad_norm": 1.2537579467676088, "learning_rate": 0.003, "loss": 4.061, "step": 12013 }, { "epoch": 0.12014, "grad_norm": 0.9993314650918684, "learning_rate": 0.003, "loss": 4.0556, "step": 12014 }, { "epoch": 0.12015, "grad_norm": 1.4004563110169088, "learning_rate": 0.003, "loss": 4.0764, "step": 12015 }, { "epoch": 0.12016, "grad_norm": 1.156141295287908, "learning_rate": 0.003, "loss": 4.0804, "step": 12016 }, { "epoch": 0.12017, "grad_norm": 1.2964729871447291, "learning_rate": 0.003, "loss": 4.0852, "step": 12017 }, { "epoch": 0.12018, "grad_norm": 1.0999613690530028, "learning_rate": 0.003, "loss": 4.0186, "step": 12018 }, { "epoch": 0.12019, "grad_norm": 1.1599848241675983, "learning_rate": 0.003, "loss": 4.0699, "step": 12019 }, { "epoch": 0.1202, "grad_norm": 1.1798981477348156, "learning_rate": 0.003, "loss": 4.0853, "step": 12020 }, { "epoch": 0.12021, "grad_norm": 1.0775703260675573, "learning_rate": 0.003, "loss": 4.0819, "step": 12021 }, { "epoch": 0.12022, "grad_norm": 1.180706310665044, "learning_rate": 0.003, "loss": 4.0392, "step": 12022 }, { "epoch": 0.12023, "grad_norm": 1.008178597261202, "learning_rate": 0.003, "loss": 4.0491, "step": 12023 }, { "epoch": 0.12024, "grad_norm": 1.22542908123363, "learning_rate": 0.003, "loss": 4.0578, "step": 12024 }, { "epoch": 0.12025, "grad_norm": 0.9947915674056316, "learning_rate": 0.003, "loss": 4.0597, "step": 12025 }, { "epoch": 0.12026, "grad_norm": 1.5178238222188867, "learning_rate": 0.003, "loss": 4.0878, "step": 12026 }, { "epoch": 0.12027, "grad_norm": 1.1898559508214746, "learning_rate": 0.003, "loss": 4.0672, "step": 12027 }, { "epoch": 0.12028, "grad_norm": 1.423286550054181, "learning_rate": 0.003, "loss": 4.0739, "step": 12028 }, { "epoch": 0.12029, "grad_norm": 1.1405204709773993, "learning_rate": 0.003, "loss": 4.0649, "step": 12029 }, { "epoch": 0.1203, "grad_norm": 1.200577829785475, "learning_rate": 0.003, "loss": 4.0683, "step": 12030 }, { "epoch": 0.12031, "grad_norm": 0.9370161339686052, "learning_rate": 0.003, "loss": 4.0752, "step": 12031 }, { "epoch": 0.12032, "grad_norm": 1.1990570439275474, "learning_rate": 0.003, "loss": 4.069, "step": 12032 }, { "epoch": 0.12033, "grad_norm": 1.0987107400462872, "learning_rate": 0.003, "loss": 4.074, "step": 12033 }, { "epoch": 0.12034, "grad_norm": 1.1543926616087123, "learning_rate": 0.003, "loss": 4.07, "step": 12034 }, { "epoch": 0.12035, "grad_norm": 1.0314508365784196, "learning_rate": 0.003, "loss": 4.0437, "step": 12035 }, { "epoch": 0.12036, "grad_norm": 1.1627404611666088, "learning_rate": 0.003, "loss": 4.0844, "step": 12036 }, { "epoch": 0.12037, "grad_norm": 1.3173377590411293, "learning_rate": 0.003, "loss": 4.0854, "step": 12037 }, { "epoch": 0.12038, "grad_norm": 1.513419415873118, "learning_rate": 0.003, "loss": 4.088, "step": 12038 }, { "epoch": 0.12039, "grad_norm": 0.8914448292836505, "learning_rate": 0.003, "loss": 4.0487, "step": 12039 }, { "epoch": 0.1204, "grad_norm": 1.1146273437131011, "learning_rate": 0.003, "loss": 4.0821, "step": 12040 }, { "epoch": 0.12041, "grad_norm": 1.3983349421031432, "learning_rate": 0.003, "loss": 4.0677, "step": 12041 }, { "epoch": 0.12042, "grad_norm": 1.0342958303049343, "learning_rate": 0.003, "loss": 4.0495, "step": 12042 }, { "epoch": 0.12043, "grad_norm": 1.3699563847933571, "learning_rate": 0.003, "loss": 4.101, "step": 12043 }, { "epoch": 0.12044, "grad_norm": 1.1499691529560814, "learning_rate": 0.003, "loss": 4.0707, "step": 12044 }, { "epoch": 0.12045, "grad_norm": 1.1554235526088477, "learning_rate": 0.003, "loss": 4.0805, "step": 12045 }, { "epoch": 0.12046, "grad_norm": 1.1070632794794475, "learning_rate": 0.003, "loss": 4.0705, "step": 12046 }, { "epoch": 0.12047, "grad_norm": 1.1989324046303953, "learning_rate": 0.003, "loss": 4.0899, "step": 12047 }, { "epoch": 0.12048, "grad_norm": 1.3734754037694892, "learning_rate": 0.003, "loss": 4.077, "step": 12048 }, { "epoch": 0.12049, "grad_norm": 0.9346259551884931, "learning_rate": 0.003, "loss": 4.0644, "step": 12049 }, { "epoch": 0.1205, "grad_norm": 1.2286880245496783, "learning_rate": 0.003, "loss": 4.1016, "step": 12050 }, { "epoch": 0.12051, "grad_norm": 1.2931944632957222, "learning_rate": 0.003, "loss": 4.0886, "step": 12051 }, { "epoch": 0.12052, "grad_norm": 1.186952328950266, "learning_rate": 0.003, "loss": 4.0869, "step": 12052 }, { "epoch": 0.12053, "grad_norm": 1.0423335340296913, "learning_rate": 0.003, "loss": 4.078, "step": 12053 }, { "epoch": 0.12054, "grad_norm": 1.291648787743366, "learning_rate": 0.003, "loss": 4.099, "step": 12054 }, { "epoch": 0.12055, "grad_norm": 1.0061073650939143, "learning_rate": 0.003, "loss": 4.0923, "step": 12055 }, { "epoch": 0.12056, "grad_norm": 1.3839942243577954, "learning_rate": 0.003, "loss": 4.0665, "step": 12056 }, { "epoch": 0.12057, "grad_norm": 0.9752307709453086, "learning_rate": 0.003, "loss": 4.0763, "step": 12057 }, { "epoch": 0.12058, "grad_norm": 1.2270914306104124, "learning_rate": 0.003, "loss": 4.1001, "step": 12058 }, { "epoch": 0.12059, "grad_norm": 1.0029707921281812, "learning_rate": 0.003, "loss": 4.0933, "step": 12059 }, { "epoch": 0.1206, "grad_norm": 1.1581116882653377, "learning_rate": 0.003, "loss": 4.0743, "step": 12060 }, { "epoch": 0.12061, "grad_norm": 1.16058489656729, "learning_rate": 0.003, "loss": 4.0998, "step": 12061 }, { "epoch": 0.12062, "grad_norm": 1.0890114409959768, "learning_rate": 0.003, "loss": 4.0554, "step": 12062 }, { "epoch": 0.12063, "grad_norm": 1.4169170730642937, "learning_rate": 0.003, "loss": 4.0887, "step": 12063 }, { "epoch": 0.12064, "grad_norm": 1.6320055790805699, "learning_rate": 0.003, "loss": 4.0554, "step": 12064 }, { "epoch": 0.12065, "grad_norm": 1.0678708229046874, "learning_rate": 0.003, "loss": 4.069, "step": 12065 }, { "epoch": 0.12066, "grad_norm": 1.4154088742505124, "learning_rate": 0.003, "loss": 4.0735, "step": 12066 }, { "epoch": 0.12067, "grad_norm": 1.0580579174780345, "learning_rate": 0.003, "loss": 4.1071, "step": 12067 }, { "epoch": 0.12068, "grad_norm": 1.1573381957667936, "learning_rate": 0.003, "loss": 4.067, "step": 12068 }, { "epoch": 0.12069, "grad_norm": 1.2075686702373492, "learning_rate": 0.003, "loss": 4.0678, "step": 12069 }, { "epoch": 0.1207, "grad_norm": 1.180040557288595, "learning_rate": 0.003, "loss": 4.0559, "step": 12070 }, { "epoch": 0.12071, "grad_norm": 1.3809455658461673, "learning_rate": 0.003, "loss": 4.0747, "step": 12071 }, { "epoch": 0.12072, "grad_norm": 0.959954898153743, "learning_rate": 0.003, "loss": 4.1134, "step": 12072 }, { "epoch": 0.12073, "grad_norm": 1.2776334910268239, "learning_rate": 0.003, "loss": 4.0798, "step": 12073 }, { "epoch": 0.12074, "grad_norm": 1.3060968377539421, "learning_rate": 0.003, "loss": 4.0833, "step": 12074 }, { "epoch": 0.12075, "grad_norm": 0.9904143169299933, "learning_rate": 0.003, "loss": 4.024, "step": 12075 }, { "epoch": 0.12076, "grad_norm": 1.3380997950361384, "learning_rate": 0.003, "loss": 4.0743, "step": 12076 }, { "epoch": 0.12077, "grad_norm": 1.162997245731524, "learning_rate": 0.003, "loss": 4.0292, "step": 12077 }, { "epoch": 0.12078, "grad_norm": 1.2680231106066544, "learning_rate": 0.003, "loss": 4.0563, "step": 12078 }, { "epoch": 0.12079, "grad_norm": 1.106780471919048, "learning_rate": 0.003, "loss": 4.0663, "step": 12079 }, { "epoch": 0.1208, "grad_norm": 1.3749209213072149, "learning_rate": 0.003, "loss": 4.0879, "step": 12080 }, { "epoch": 0.12081, "grad_norm": 1.0047272466191906, "learning_rate": 0.003, "loss": 4.0746, "step": 12081 }, { "epoch": 0.12082, "grad_norm": 1.2394249155802766, "learning_rate": 0.003, "loss": 4.0705, "step": 12082 }, { "epoch": 0.12083, "grad_norm": 1.1454349115566007, "learning_rate": 0.003, "loss": 4.0606, "step": 12083 }, { "epoch": 0.12084, "grad_norm": 1.2837907309804073, "learning_rate": 0.003, "loss": 4.0851, "step": 12084 }, { "epoch": 0.12085, "grad_norm": 1.1247969860112579, "learning_rate": 0.003, "loss": 4.0814, "step": 12085 }, { "epoch": 0.12086, "grad_norm": 1.281210661018199, "learning_rate": 0.003, "loss": 4.0893, "step": 12086 }, { "epoch": 0.12087, "grad_norm": 1.2889817385160782, "learning_rate": 0.003, "loss": 4.1129, "step": 12087 }, { "epoch": 0.12088, "grad_norm": 0.9718926809926418, "learning_rate": 0.003, "loss": 4.062, "step": 12088 }, { "epoch": 0.12089, "grad_norm": 1.1986561179093216, "learning_rate": 0.003, "loss": 4.0875, "step": 12089 }, { "epoch": 0.1209, "grad_norm": 1.0917914177599881, "learning_rate": 0.003, "loss": 4.0513, "step": 12090 }, { "epoch": 0.12091, "grad_norm": 1.3732130609897086, "learning_rate": 0.003, "loss": 4.0767, "step": 12091 }, { "epoch": 0.12092, "grad_norm": 0.9824315718137555, "learning_rate": 0.003, "loss": 4.0616, "step": 12092 }, { "epoch": 0.12093, "grad_norm": 1.4181887491487821, "learning_rate": 0.003, "loss": 4.0996, "step": 12093 }, { "epoch": 0.12094, "grad_norm": 1.0585609680051242, "learning_rate": 0.003, "loss": 4.0905, "step": 12094 }, { "epoch": 0.12095, "grad_norm": 1.3200300643697576, "learning_rate": 0.003, "loss": 4.0915, "step": 12095 }, { "epoch": 0.12096, "grad_norm": 1.1183009094442204, "learning_rate": 0.003, "loss": 4.0706, "step": 12096 }, { "epoch": 0.12097, "grad_norm": 1.0695065139721762, "learning_rate": 0.003, "loss": 4.0987, "step": 12097 }, { "epoch": 0.12098, "grad_norm": 1.301116622282068, "learning_rate": 0.003, "loss": 4.0606, "step": 12098 }, { "epoch": 0.12099, "grad_norm": 1.0517241296656699, "learning_rate": 0.003, "loss": 4.0522, "step": 12099 }, { "epoch": 0.121, "grad_norm": 1.124424750696396, "learning_rate": 0.003, "loss": 4.0567, "step": 12100 }, { "epoch": 0.12101, "grad_norm": 1.057831288939637, "learning_rate": 0.003, "loss": 4.0774, "step": 12101 }, { "epoch": 0.12102, "grad_norm": 1.1101237754091218, "learning_rate": 0.003, "loss": 4.0704, "step": 12102 }, { "epoch": 0.12103, "grad_norm": 0.9612959979260564, "learning_rate": 0.003, "loss": 4.069, "step": 12103 }, { "epoch": 0.12104, "grad_norm": 1.2061186388314846, "learning_rate": 0.003, "loss": 4.0495, "step": 12104 }, { "epoch": 0.12105, "grad_norm": 0.9678635059546692, "learning_rate": 0.003, "loss": 4.0639, "step": 12105 }, { "epoch": 0.12106, "grad_norm": 1.3505427330387154, "learning_rate": 0.003, "loss": 4.0612, "step": 12106 }, { "epoch": 0.12107, "grad_norm": 1.1774391224451677, "learning_rate": 0.003, "loss": 4.102, "step": 12107 }, { "epoch": 0.12108, "grad_norm": 1.2054664808823852, "learning_rate": 0.003, "loss": 4.0719, "step": 12108 }, { "epoch": 0.12109, "grad_norm": 1.3781623075709073, "learning_rate": 0.003, "loss": 4.0581, "step": 12109 }, { "epoch": 0.1211, "grad_norm": 1.2233097675635973, "learning_rate": 0.003, "loss": 4.0862, "step": 12110 }, { "epoch": 0.12111, "grad_norm": 1.2557489541050741, "learning_rate": 0.003, "loss": 4.0706, "step": 12111 }, { "epoch": 0.12112, "grad_norm": 1.110510268205828, "learning_rate": 0.003, "loss": 4.0907, "step": 12112 }, { "epoch": 0.12113, "grad_norm": 1.1942343402586795, "learning_rate": 0.003, "loss": 4.0861, "step": 12113 }, { "epoch": 0.12114, "grad_norm": 1.1603250584499352, "learning_rate": 0.003, "loss": 4.0691, "step": 12114 }, { "epoch": 0.12115, "grad_norm": 1.3816403360973941, "learning_rate": 0.003, "loss": 4.0376, "step": 12115 }, { "epoch": 0.12116, "grad_norm": 0.9444863512485694, "learning_rate": 0.003, "loss": 4.1, "step": 12116 }, { "epoch": 0.12117, "grad_norm": 1.080848550044189, "learning_rate": 0.003, "loss": 4.0934, "step": 12117 }, { "epoch": 0.12118, "grad_norm": 1.1995298311432099, "learning_rate": 0.003, "loss": 4.0435, "step": 12118 }, { "epoch": 0.12119, "grad_norm": 1.3314668602134827, "learning_rate": 0.003, "loss": 4.0941, "step": 12119 }, { "epoch": 0.1212, "grad_norm": 1.1715889058212041, "learning_rate": 0.003, "loss": 4.0486, "step": 12120 }, { "epoch": 0.12121, "grad_norm": 1.3881009117703809, "learning_rate": 0.003, "loss": 4.0852, "step": 12121 }, { "epoch": 0.12122, "grad_norm": 1.1412877222655404, "learning_rate": 0.003, "loss": 4.0937, "step": 12122 }, { "epoch": 0.12123, "grad_norm": 1.2581104621689234, "learning_rate": 0.003, "loss": 4.0653, "step": 12123 }, { "epoch": 0.12124, "grad_norm": 1.0606037278927645, "learning_rate": 0.003, "loss": 4.0676, "step": 12124 }, { "epoch": 0.12125, "grad_norm": 1.1190145081517746, "learning_rate": 0.003, "loss": 4.085, "step": 12125 }, { "epoch": 0.12126, "grad_norm": 1.1601406262785494, "learning_rate": 0.003, "loss": 4.0742, "step": 12126 }, { "epoch": 0.12127, "grad_norm": 1.2080466202977465, "learning_rate": 0.003, "loss": 4.0648, "step": 12127 }, { "epoch": 0.12128, "grad_norm": 1.1227257472864887, "learning_rate": 0.003, "loss": 4.0898, "step": 12128 }, { "epoch": 0.12129, "grad_norm": 1.1102649147398123, "learning_rate": 0.003, "loss": 4.0589, "step": 12129 }, { "epoch": 0.1213, "grad_norm": 1.1537725039029854, "learning_rate": 0.003, "loss": 4.087, "step": 12130 }, { "epoch": 0.12131, "grad_norm": 1.3114867170222135, "learning_rate": 0.003, "loss": 4.094, "step": 12131 }, { "epoch": 0.12132, "grad_norm": 0.9234569238063531, "learning_rate": 0.003, "loss": 4.0783, "step": 12132 }, { "epoch": 0.12133, "grad_norm": 1.1279748123445532, "learning_rate": 0.003, "loss": 4.071, "step": 12133 }, { "epoch": 0.12134, "grad_norm": 1.2600065947038894, "learning_rate": 0.003, "loss": 4.0945, "step": 12134 }, { "epoch": 0.12135, "grad_norm": 1.007780976926589, "learning_rate": 0.003, "loss": 4.0598, "step": 12135 }, { "epoch": 0.12136, "grad_norm": 1.1984153877860733, "learning_rate": 0.003, "loss": 4.0736, "step": 12136 }, { "epoch": 0.12137, "grad_norm": 1.423530162485099, "learning_rate": 0.003, "loss": 4.1151, "step": 12137 }, { "epoch": 0.12138, "grad_norm": 1.145701795096461, "learning_rate": 0.003, "loss": 4.0949, "step": 12138 }, { "epoch": 0.12139, "grad_norm": 1.189173877147463, "learning_rate": 0.003, "loss": 4.0942, "step": 12139 }, { "epoch": 0.1214, "grad_norm": 1.1973702326318612, "learning_rate": 0.003, "loss": 4.0958, "step": 12140 }, { "epoch": 0.12141, "grad_norm": 1.167840594810345, "learning_rate": 0.003, "loss": 4.068, "step": 12141 }, { "epoch": 0.12142, "grad_norm": 1.220199217906067, "learning_rate": 0.003, "loss": 4.083, "step": 12142 }, { "epoch": 0.12143, "grad_norm": 1.1368354349474534, "learning_rate": 0.003, "loss": 4.0834, "step": 12143 }, { "epoch": 0.12144, "grad_norm": 1.1779110345400967, "learning_rate": 0.003, "loss": 4.0984, "step": 12144 }, { "epoch": 0.12145, "grad_norm": 1.2251180573005898, "learning_rate": 0.003, "loss": 4.0946, "step": 12145 }, { "epoch": 0.12146, "grad_norm": 0.9548607030454273, "learning_rate": 0.003, "loss": 4.0879, "step": 12146 }, { "epoch": 0.12147, "grad_norm": 1.1049973878812982, "learning_rate": 0.003, "loss": 4.0783, "step": 12147 }, { "epoch": 0.12148, "grad_norm": 1.0970442611358633, "learning_rate": 0.003, "loss": 4.0883, "step": 12148 }, { "epoch": 0.12149, "grad_norm": 1.2126640085318225, "learning_rate": 0.003, "loss": 4.0721, "step": 12149 }, { "epoch": 0.1215, "grad_norm": 1.0151653599442538, "learning_rate": 0.003, "loss": 4.0637, "step": 12150 }, { "epoch": 0.12151, "grad_norm": 1.2551528836697066, "learning_rate": 0.003, "loss": 4.0462, "step": 12151 }, { "epoch": 0.12152, "grad_norm": 1.1346936623381856, "learning_rate": 0.003, "loss": 4.0518, "step": 12152 }, { "epoch": 0.12153, "grad_norm": 1.2703669585817352, "learning_rate": 0.003, "loss": 4.1049, "step": 12153 }, { "epoch": 0.12154, "grad_norm": 1.349583252239081, "learning_rate": 0.003, "loss": 4.0397, "step": 12154 }, { "epoch": 0.12155, "grad_norm": 1.304156342689166, "learning_rate": 0.003, "loss": 4.0529, "step": 12155 }, { "epoch": 0.12156, "grad_norm": 1.171188520618665, "learning_rate": 0.003, "loss": 4.0699, "step": 12156 }, { "epoch": 0.12157, "grad_norm": 1.0722828495955472, "learning_rate": 0.003, "loss": 4.0251, "step": 12157 }, { "epoch": 0.12158, "grad_norm": 1.1483576341362236, "learning_rate": 0.003, "loss": 4.0711, "step": 12158 }, { "epoch": 0.12159, "grad_norm": 1.1198251651835405, "learning_rate": 0.003, "loss": 4.0716, "step": 12159 }, { "epoch": 0.1216, "grad_norm": 1.2441043778471383, "learning_rate": 0.003, "loss": 4.0632, "step": 12160 }, { "epoch": 0.12161, "grad_norm": 1.0390582605247063, "learning_rate": 0.003, "loss": 4.076, "step": 12161 }, { "epoch": 0.12162, "grad_norm": 1.0943988455876759, "learning_rate": 0.003, "loss": 4.0648, "step": 12162 }, { "epoch": 0.12163, "grad_norm": 1.1896450292303948, "learning_rate": 0.003, "loss": 4.0688, "step": 12163 }, { "epoch": 0.12164, "grad_norm": 1.1110737554572037, "learning_rate": 0.003, "loss": 4.067, "step": 12164 }, { "epoch": 0.12165, "grad_norm": 1.2722228329526097, "learning_rate": 0.003, "loss": 4.0518, "step": 12165 }, { "epoch": 0.12166, "grad_norm": 0.9537638120975451, "learning_rate": 0.003, "loss": 4.0762, "step": 12166 }, { "epoch": 0.12167, "grad_norm": 1.168497152953338, "learning_rate": 0.003, "loss": 4.0624, "step": 12167 }, { "epoch": 0.12168, "grad_norm": 1.1096490497883948, "learning_rate": 0.003, "loss": 4.0601, "step": 12168 }, { "epoch": 0.12169, "grad_norm": 1.3891119687227524, "learning_rate": 0.003, "loss": 4.0902, "step": 12169 }, { "epoch": 0.1217, "grad_norm": 1.2381209928430519, "learning_rate": 0.003, "loss": 4.0577, "step": 12170 }, { "epoch": 0.12171, "grad_norm": 1.075557009585934, "learning_rate": 0.003, "loss": 4.0543, "step": 12171 }, { "epoch": 0.12172, "grad_norm": 1.252001700722133, "learning_rate": 0.003, "loss": 4.0901, "step": 12172 }, { "epoch": 0.12173, "grad_norm": 1.1991927553926924, "learning_rate": 0.003, "loss": 4.0729, "step": 12173 }, { "epoch": 0.12174, "grad_norm": 1.0706837329851342, "learning_rate": 0.003, "loss": 4.0903, "step": 12174 }, { "epoch": 0.12175, "grad_norm": 1.198039780467412, "learning_rate": 0.003, "loss": 4.0824, "step": 12175 }, { "epoch": 0.12176, "grad_norm": 1.3827857084183623, "learning_rate": 0.003, "loss": 4.0623, "step": 12176 }, { "epoch": 0.12177, "grad_norm": 1.0018996448559927, "learning_rate": 0.003, "loss": 4.0693, "step": 12177 }, { "epoch": 0.12178, "grad_norm": 1.2754499630795086, "learning_rate": 0.003, "loss": 4.0853, "step": 12178 }, { "epoch": 0.12179, "grad_norm": 1.141006379500696, "learning_rate": 0.003, "loss": 4.0748, "step": 12179 }, { "epoch": 0.1218, "grad_norm": 1.1554481432840713, "learning_rate": 0.003, "loss": 4.0548, "step": 12180 }, { "epoch": 0.12181, "grad_norm": 1.1287874762838688, "learning_rate": 0.003, "loss": 4.0526, "step": 12181 }, { "epoch": 0.12182, "grad_norm": 1.1406212931173205, "learning_rate": 0.003, "loss": 4.0761, "step": 12182 }, { "epoch": 0.12183, "grad_norm": 1.044135481875383, "learning_rate": 0.003, "loss": 4.0411, "step": 12183 }, { "epoch": 0.12184, "grad_norm": 1.2984165649244284, "learning_rate": 0.003, "loss": 4.0812, "step": 12184 }, { "epoch": 0.12185, "grad_norm": 1.0484508917758921, "learning_rate": 0.003, "loss": 4.0569, "step": 12185 }, { "epoch": 0.12186, "grad_norm": 1.3348507005714336, "learning_rate": 0.003, "loss": 4.0662, "step": 12186 }, { "epoch": 0.12187, "grad_norm": 1.3208776230573516, "learning_rate": 0.003, "loss": 4.0859, "step": 12187 }, { "epoch": 0.12188, "grad_norm": 1.2186571875290615, "learning_rate": 0.003, "loss": 4.0788, "step": 12188 }, { "epoch": 0.12189, "grad_norm": 1.2492837808487232, "learning_rate": 0.003, "loss": 4.0658, "step": 12189 }, { "epoch": 0.1219, "grad_norm": 1.1334787027314297, "learning_rate": 0.003, "loss": 4.0783, "step": 12190 }, { "epoch": 0.12191, "grad_norm": 1.2246108948049745, "learning_rate": 0.003, "loss": 4.0938, "step": 12191 }, { "epoch": 0.12192, "grad_norm": 1.2110322419568589, "learning_rate": 0.003, "loss": 4.0884, "step": 12192 }, { "epoch": 0.12193, "grad_norm": 1.0261731131892984, "learning_rate": 0.003, "loss": 4.0936, "step": 12193 }, { "epoch": 0.12194, "grad_norm": 1.3792295929539276, "learning_rate": 0.003, "loss": 4.0711, "step": 12194 }, { "epoch": 0.12195, "grad_norm": 0.9497590834552457, "learning_rate": 0.003, "loss": 4.0597, "step": 12195 }, { "epoch": 0.12196, "grad_norm": 1.1971816454138282, "learning_rate": 0.003, "loss": 4.0567, "step": 12196 }, { "epoch": 0.12197, "grad_norm": 1.346176011354049, "learning_rate": 0.003, "loss": 4.06, "step": 12197 }, { "epoch": 0.12198, "grad_norm": 0.9845918447685015, "learning_rate": 0.003, "loss": 4.0717, "step": 12198 }, { "epoch": 0.12199, "grad_norm": 1.4013619145319482, "learning_rate": 0.003, "loss": 4.0919, "step": 12199 }, { "epoch": 0.122, "grad_norm": 1.3402558991670583, "learning_rate": 0.003, "loss": 4.08, "step": 12200 }, { "epoch": 0.12201, "grad_norm": 1.0917890609387033, "learning_rate": 0.003, "loss": 4.0668, "step": 12201 }, { "epoch": 0.12202, "grad_norm": 1.3977198473853139, "learning_rate": 0.003, "loss": 4.0952, "step": 12202 }, { "epoch": 0.12203, "grad_norm": 1.1663142701608584, "learning_rate": 0.003, "loss": 4.0943, "step": 12203 }, { "epoch": 0.12204, "grad_norm": 1.3067073939428604, "learning_rate": 0.003, "loss": 4.1008, "step": 12204 }, { "epoch": 0.12205, "grad_norm": 0.9953106606493801, "learning_rate": 0.003, "loss": 4.0638, "step": 12205 }, { "epoch": 0.12206, "grad_norm": 1.1961413384082726, "learning_rate": 0.003, "loss": 4.0514, "step": 12206 }, { "epoch": 0.12207, "grad_norm": 1.1660261913008847, "learning_rate": 0.003, "loss": 4.1179, "step": 12207 }, { "epoch": 0.12208, "grad_norm": 1.3567657303292768, "learning_rate": 0.003, "loss": 4.0946, "step": 12208 }, { "epoch": 0.12209, "grad_norm": 1.1940065501612556, "learning_rate": 0.003, "loss": 4.0762, "step": 12209 }, { "epoch": 0.1221, "grad_norm": 1.3451533385151508, "learning_rate": 0.003, "loss": 4.0754, "step": 12210 }, { "epoch": 0.12211, "grad_norm": 1.0861834534287251, "learning_rate": 0.003, "loss": 4.0604, "step": 12211 }, { "epoch": 0.12212, "grad_norm": 1.1170231539539128, "learning_rate": 0.003, "loss": 4.0683, "step": 12212 }, { "epoch": 0.12213, "grad_norm": 1.1842563818995908, "learning_rate": 0.003, "loss": 4.0848, "step": 12213 }, { "epoch": 0.12214, "grad_norm": 1.2412122957607226, "learning_rate": 0.003, "loss": 4.0801, "step": 12214 }, { "epoch": 0.12215, "grad_norm": 1.281614029600869, "learning_rate": 0.003, "loss": 4.0978, "step": 12215 }, { "epoch": 0.12216, "grad_norm": 1.1905902787785605, "learning_rate": 0.003, "loss": 4.0713, "step": 12216 }, { "epoch": 0.12217, "grad_norm": 1.15855892402375, "learning_rate": 0.003, "loss": 4.0578, "step": 12217 }, { "epoch": 0.12218, "grad_norm": 1.1656183934254463, "learning_rate": 0.003, "loss": 4.0516, "step": 12218 }, { "epoch": 0.12219, "grad_norm": 1.205973966373365, "learning_rate": 0.003, "loss": 4.0951, "step": 12219 }, { "epoch": 0.1222, "grad_norm": 0.9515308914661088, "learning_rate": 0.003, "loss": 4.0708, "step": 12220 }, { "epoch": 0.12221, "grad_norm": 1.2099897259709462, "learning_rate": 0.003, "loss": 4.0751, "step": 12221 }, { "epoch": 0.12222, "grad_norm": 1.3622068216886205, "learning_rate": 0.003, "loss": 4.0594, "step": 12222 }, { "epoch": 0.12223, "grad_norm": 1.1289481339729313, "learning_rate": 0.003, "loss": 4.0798, "step": 12223 }, { "epoch": 0.12224, "grad_norm": 1.1788306971210525, "learning_rate": 0.003, "loss": 4.0524, "step": 12224 }, { "epoch": 0.12225, "grad_norm": 1.2714621284495276, "learning_rate": 0.003, "loss": 4.0738, "step": 12225 }, { "epoch": 0.12226, "grad_norm": 1.172042908502254, "learning_rate": 0.003, "loss": 4.0746, "step": 12226 }, { "epoch": 0.12227, "grad_norm": 1.0272185949615782, "learning_rate": 0.003, "loss": 4.047, "step": 12227 }, { "epoch": 0.12228, "grad_norm": 1.2050509278203998, "learning_rate": 0.003, "loss": 4.0369, "step": 12228 }, { "epoch": 0.12229, "grad_norm": 1.133457829662722, "learning_rate": 0.003, "loss": 4.0714, "step": 12229 }, { "epoch": 0.1223, "grad_norm": 1.1844731245673616, "learning_rate": 0.003, "loss": 4.0966, "step": 12230 }, { "epoch": 0.12231, "grad_norm": 1.1279404374429298, "learning_rate": 0.003, "loss": 4.0694, "step": 12231 }, { "epoch": 0.12232, "grad_norm": 1.2244882868041893, "learning_rate": 0.003, "loss": 4.0688, "step": 12232 }, { "epoch": 0.12233, "grad_norm": 1.0501181368810093, "learning_rate": 0.003, "loss": 4.0592, "step": 12233 }, { "epoch": 0.12234, "grad_norm": 1.3746819695663268, "learning_rate": 0.003, "loss": 4.1022, "step": 12234 }, { "epoch": 0.12235, "grad_norm": 1.1124087609485749, "learning_rate": 0.003, "loss": 4.0605, "step": 12235 }, { "epoch": 0.12236, "grad_norm": 1.1498676818007616, "learning_rate": 0.003, "loss": 4.0959, "step": 12236 }, { "epoch": 0.12237, "grad_norm": 1.2311318127440682, "learning_rate": 0.003, "loss": 4.0682, "step": 12237 }, { "epoch": 0.12238, "grad_norm": 1.6033985780973559, "learning_rate": 0.003, "loss": 4.0562, "step": 12238 }, { "epoch": 0.12239, "grad_norm": 1.072128730565727, "learning_rate": 0.003, "loss": 4.0852, "step": 12239 }, { "epoch": 0.1224, "grad_norm": 1.3035753237150192, "learning_rate": 0.003, "loss": 4.0968, "step": 12240 }, { "epoch": 0.12241, "grad_norm": 1.1989449538493253, "learning_rate": 0.003, "loss": 4.0496, "step": 12241 }, { "epoch": 0.12242, "grad_norm": 1.114035699582397, "learning_rate": 0.003, "loss": 4.0751, "step": 12242 }, { "epoch": 0.12243, "grad_norm": 1.1325432549923347, "learning_rate": 0.003, "loss": 4.0746, "step": 12243 }, { "epoch": 0.12244, "grad_norm": 1.1685657840734007, "learning_rate": 0.003, "loss": 4.0917, "step": 12244 }, { "epoch": 0.12245, "grad_norm": 1.2109037845732278, "learning_rate": 0.003, "loss": 4.0975, "step": 12245 }, { "epoch": 0.12246, "grad_norm": 1.2960492703053292, "learning_rate": 0.003, "loss": 4.0731, "step": 12246 }, { "epoch": 0.12247, "grad_norm": 1.0630001470423178, "learning_rate": 0.003, "loss": 4.0544, "step": 12247 }, { "epoch": 0.12248, "grad_norm": 1.439686551242132, "learning_rate": 0.003, "loss": 4.0717, "step": 12248 }, { "epoch": 0.12249, "grad_norm": 0.9360021317225822, "learning_rate": 0.003, "loss": 4.0333, "step": 12249 }, { "epoch": 0.1225, "grad_norm": 1.1608835012568275, "learning_rate": 0.003, "loss": 4.089, "step": 12250 }, { "epoch": 0.12251, "grad_norm": 1.152778853567974, "learning_rate": 0.003, "loss": 4.0936, "step": 12251 }, { "epoch": 0.12252, "grad_norm": 1.3070190135020505, "learning_rate": 0.003, "loss": 4.08, "step": 12252 }, { "epoch": 0.12253, "grad_norm": 1.1703555806685912, "learning_rate": 0.003, "loss": 4.0741, "step": 12253 }, { "epoch": 0.12254, "grad_norm": 1.2753604282061932, "learning_rate": 0.003, "loss": 4.0749, "step": 12254 }, { "epoch": 0.12255, "grad_norm": 1.2012730702627195, "learning_rate": 0.003, "loss": 4.0711, "step": 12255 }, { "epoch": 0.12256, "grad_norm": 1.2685624175175607, "learning_rate": 0.003, "loss": 4.0738, "step": 12256 }, { "epoch": 0.12257, "grad_norm": 1.1982293227603562, "learning_rate": 0.003, "loss": 4.0961, "step": 12257 }, { "epoch": 0.12258, "grad_norm": 1.3454858034418558, "learning_rate": 0.003, "loss": 4.0622, "step": 12258 }, { "epoch": 0.12259, "grad_norm": 1.2663687933849272, "learning_rate": 0.003, "loss": 4.0707, "step": 12259 }, { "epoch": 0.1226, "grad_norm": 1.2182908134730372, "learning_rate": 0.003, "loss": 4.0558, "step": 12260 }, { "epoch": 0.12261, "grad_norm": 1.2478259449383944, "learning_rate": 0.003, "loss": 4.0604, "step": 12261 }, { "epoch": 0.12262, "grad_norm": 1.0688438965810219, "learning_rate": 0.003, "loss": 4.0504, "step": 12262 }, { "epoch": 0.12263, "grad_norm": 1.2976972823949195, "learning_rate": 0.003, "loss": 4.0451, "step": 12263 }, { "epoch": 0.12264, "grad_norm": 1.148927289043632, "learning_rate": 0.003, "loss": 4.0818, "step": 12264 }, { "epoch": 0.12265, "grad_norm": 0.9320813625265878, "learning_rate": 0.003, "loss": 4.0542, "step": 12265 }, { "epoch": 0.12266, "grad_norm": 1.1132041468172307, "learning_rate": 0.003, "loss": 4.04, "step": 12266 }, { "epoch": 0.12267, "grad_norm": 1.099340406144865, "learning_rate": 0.003, "loss": 4.0826, "step": 12267 }, { "epoch": 0.12268, "grad_norm": 1.065205149602413, "learning_rate": 0.003, "loss": 4.0648, "step": 12268 }, { "epoch": 0.12269, "grad_norm": 1.1588607244578912, "learning_rate": 0.003, "loss": 4.0597, "step": 12269 }, { "epoch": 0.1227, "grad_norm": 1.1704601580263845, "learning_rate": 0.003, "loss": 4.0779, "step": 12270 }, { "epoch": 0.12271, "grad_norm": 1.3898945239488298, "learning_rate": 0.003, "loss": 4.0594, "step": 12271 }, { "epoch": 0.12272, "grad_norm": 0.95609587772018, "learning_rate": 0.003, "loss": 4.0351, "step": 12272 }, { "epoch": 0.12273, "grad_norm": 1.0888462421073244, "learning_rate": 0.003, "loss": 4.0736, "step": 12273 }, { "epoch": 0.12274, "grad_norm": 1.2938651078465015, "learning_rate": 0.003, "loss": 4.0532, "step": 12274 }, { "epoch": 0.12275, "grad_norm": 1.0515210845199563, "learning_rate": 0.003, "loss": 4.0865, "step": 12275 }, { "epoch": 0.12276, "grad_norm": 1.1681798074447074, "learning_rate": 0.003, "loss": 4.068, "step": 12276 }, { "epoch": 0.12277, "grad_norm": 1.0957574398621501, "learning_rate": 0.003, "loss": 4.0761, "step": 12277 }, { "epoch": 0.12278, "grad_norm": 1.2402753483112188, "learning_rate": 0.003, "loss": 4.0813, "step": 12278 }, { "epoch": 0.12279, "grad_norm": 1.3253921086663587, "learning_rate": 0.003, "loss": 4.0364, "step": 12279 }, { "epoch": 0.1228, "grad_norm": 0.9446132722770976, "learning_rate": 0.003, "loss": 4.0736, "step": 12280 }, { "epoch": 0.12281, "grad_norm": 1.2543549659079503, "learning_rate": 0.003, "loss": 4.0673, "step": 12281 }, { "epoch": 0.12282, "grad_norm": 1.044464140315054, "learning_rate": 0.003, "loss": 4.0968, "step": 12282 }, { "epoch": 0.12283, "grad_norm": 1.4448428457498539, "learning_rate": 0.003, "loss": 4.0638, "step": 12283 }, { "epoch": 0.12284, "grad_norm": 0.9323247854530715, "learning_rate": 0.003, "loss": 4.0487, "step": 12284 }, { "epoch": 0.12285, "grad_norm": 1.2183938038468276, "learning_rate": 0.003, "loss": 4.0838, "step": 12285 }, { "epoch": 0.12286, "grad_norm": 1.1168696478309292, "learning_rate": 0.003, "loss": 4.0806, "step": 12286 }, { "epoch": 0.12287, "grad_norm": 1.4118380090895937, "learning_rate": 0.003, "loss": 4.0872, "step": 12287 }, { "epoch": 0.12288, "grad_norm": 1.034995448495016, "learning_rate": 0.003, "loss": 4.0924, "step": 12288 }, { "epoch": 0.12289, "grad_norm": 1.2572160704779352, "learning_rate": 0.003, "loss": 4.0467, "step": 12289 }, { "epoch": 0.1229, "grad_norm": 1.0887530584287088, "learning_rate": 0.003, "loss": 4.0652, "step": 12290 }, { "epoch": 0.12291, "grad_norm": 1.2194821967348717, "learning_rate": 0.003, "loss": 4.0753, "step": 12291 }, { "epoch": 0.12292, "grad_norm": 1.1774785860755754, "learning_rate": 0.003, "loss": 4.0637, "step": 12292 }, { "epoch": 0.12293, "grad_norm": 1.290622218740054, "learning_rate": 0.003, "loss": 4.0879, "step": 12293 }, { "epoch": 0.12294, "grad_norm": 1.2162922983304212, "learning_rate": 0.003, "loss": 4.0872, "step": 12294 }, { "epoch": 0.12295, "grad_norm": 1.1223430685307745, "learning_rate": 0.003, "loss": 4.0827, "step": 12295 }, { "epoch": 0.12296, "grad_norm": 1.3312238598478945, "learning_rate": 0.003, "loss": 4.0755, "step": 12296 }, { "epoch": 0.12297, "grad_norm": 1.2024661509907468, "learning_rate": 0.003, "loss": 4.0667, "step": 12297 }, { "epoch": 0.12298, "grad_norm": 0.9552507213085488, "learning_rate": 0.003, "loss": 4.0791, "step": 12298 }, { "epoch": 0.12299, "grad_norm": 1.107910154506854, "learning_rate": 0.003, "loss": 4.0621, "step": 12299 }, { "epoch": 0.123, "grad_norm": 0.9669575672708685, "learning_rate": 0.003, "loss": 4.079, "step": 12300 }, { "epoch": 0.12301, "grad_norm": 1.2230981789950965, "learning_rate": 0.003, "loss": 4.098, "step": 12301 }, { "epoch": 0.12302, "grad_norm": 0.9706414365878976, "learning_rate": 0.003, "loss": 4.0855, "step": 12302 }, { "epoch": 0.12303, "grad_norm": 1.3413519996827952, "learning_rate": 0.003, "loss": 4.1424, "step": 12303 }, { "epoch": 0.12304, "grad_norm": 1.0182058543155155, "learning_rate": 0.003, "loss": 4.0534, "step": 12304 }, { "epoch": 0.12305, "grad_norm": 1.2254298651040276, "learning_rate": 0.003, "loss": 4.0765, "step": 12305 }, { "epoch": 0.12306, "grad_norm": 1.1502601972603381, "learning_rate": 0.003, "loss": 4.0757, "step": 12306 }, { "epoch": 0.12307, "grad_norm": 1.1914233088333748, "learning_rate": 0.003, "loss": 4.0703, "step": 12307 }, { "epoch": 0.12308, "grad_norm": 1.5601911292837407, "learning_rate": 0.003, "loss": 4.0975, "step": 12308 }, { "epoch": 0.12309, "grad_norm": 1.1028187394945874, "learning_rate": 0.003, "loss": 4.0734, "step": 12309 }, { "epoch": 0.1231, "grad_norm": 1.540945453323211, "learning_rate": 0.003, "loss": 4.0846, "step": 12310 }, { "epoch": 0.12311, "grad_norm": 1.060524542842975, "learning_rate": 0.003, "loss": 4.0615, "step": 12311 }, { "epoch": 0.12312, "grad_norm": 1.1688223005429994, "learning_rate": 0.003, "loss": 4.092, "step": 12312 }, { "epoch": 0.12313, "grad_norm": 0.9494509325263587, "learning_rate": 0.003, "loss": 4.0978, "step": 12313 }, { "epoch": 0.12314, "grad_norm": 1.231785054785133, "learning_rate": 0.003, "loss": 4.0717, "step": 12314 }, { "epoch": 0.12315, "grad_norm": 1.2738306714406045, "learning_rate": 0.003, "loss": 4.1012, "step": 12315 }, { "epoch": 0.12316, "grad_norm": 1.1829541714808394, "learning_rate": 0.003, "loss": 4.0873, "step": 12316 }, { "epoch": 0.12317, "grad_norm": 1.4545722081228218, "learning_rate": 0.003, "loss": 4.0476, "step": 12317 }, { "epoch": 0.12318, "grad_norm": 1.1860778929682056, "learning_rate": 0.003, "loss": 4.098, "step": 12318 }, { "epoch": 0.12319, "grad_norm": 1.105919643488645, "learning_rate": 0.003, "loss": 4.0777, "step": 12319 }, { "epoch": 0.1232, "grad_norm": 1.2407488668861022, "learning_rate": 0.003, "loss": 4.0999, "step": 12320 }, { "epoch": 0.12321, "grad_norm": 1.1156137272992521, "learning_rate": 0.003, "loss": 4.0673, "step": 12321 }, { "epoch": 0.12322, "grad_norm": 1.45072845598213, "learning_rate": 0.003, "loss": 4.0794, "step": 12322 }, { "epoch": 0.12323, "grad_norm": 1.194969851622858, "learning_rate": 0.003, "loss": 4.0708, "step": 12323 }, { "epoch": 0.12324, "grad_norm": 1.2640609944605872, "learning_rate": 0.003, "loss": 4.0645, "step": 12324 }, { "epoch": 0.12325, "grad_norm": 0.9388456031153591, "learning_rate": 0.003, "loss": 4.0362, "step": 12325 }, { "epoch": 0.12326, "grad_norm": 1.282589163964992, "learning_rate": 0.003, "loss": 4.0733, "step": 12326 }, { "epoch": 0.12327, "grad_norm": 1.3107507611608642, "learning_rate": 0.003, "loss": 4.0463, "step": 12327 }, { "epoch": 0.12328, "grad_norm": 1.1448524193305218, "learning_rate": 0.003, "loss": 4.0418, "step": 12328 }, { "epoch": 0.12329, "grad_norm": 1.2593319977525492, "learning_rate": 0.003, "loss": 4.089, "step": 12329 }, { "epoch": 0.1233, "grad_norm": 1.0922788769721852, "learning_rate": 0.003, "loss": 4.055, "step": 12330 }, { "epoch": 0.12331, "grad_norm": 1.326047175565969, "learning_rate": 0.003, "loss": 4.0662, "step": 12331 }, { "epoch": 0.12332, "grad_norm": 1.0559325620610494, "learning_rate": 0.003, "loss": 4.0545, "step": 12332 }, { "epoch": 0.12333, "grad_norm": 1.333598551232034, "learning_rate": 0.003, "loss": 4.07, "step": 12333 }, { "epoch": 0.12334, "grad_norm": 1.070779532164327, "learning_rate": 0.003, "loss": 4.0828, "step": 12334 }, { "epoch": 0.12335, "grad_norm": 1.337160726790681, "learning_rate": 0.003, "loss": 4.0793, "step": 12335 }, { "epoch": 0.12336, "grad_norm": 1.045959131867625, "learning_rate": 0.003, "loss": 4.0785, "step": 12336 }, { "epoch": 0.12337, "grad_norm": 1.1652231452234632, "learning_rate": 0.003, "loss": 4.0775, "step": 12337 }, { "epoch": 0.12338, "grad_norm": 1.1153823730527561, "learning_rate": 0.003, "loss": 4.0657, "step": 12338 }, { "epoch": 0.12339, "grad_norm": 1.1801428814861659, "learning_rate": 0.003, "loss": 4.0809, "step": 12339 }, { "epoch": 0.1234, "grad_norm": 1.1411467633416132, "learning_rate": 0.003, "loss": 4.1012, "step": 12340 }, { "epoch": 0.12341, "grad_norm": 1.183971895065385, "learning_rate": 0.003, "loss": 4.0878, "step": 12341 }, { "epoch": 0.12342, "grad_norm": 1.2148932364453693, "learning_rate": 0.003, "loss": 4.0635, "step": 12342 }, { "epoch": 0.12343, "grad_norm": 1.3997236811583327, "learning_rate": 0.003, "loss": 4.0651, "step": 12343 }, { "epoch": 0.12344, "grad_norm": 1.3906783136925027, "learning_rate": 0.003, "loss": 4.0811, "step": 12344 }, { "epoch": 0.12345, "grad_norm": 1.1929763591115197, "learning_rate": 0.003, "loss": 4.0405, "step": 12345 }, { "epoch": 0.12346, "grad_norm": 1.0421233278605289, "learning_rate": 0.003, "loss": 4.0532, "step": 12346 }, { "epoch": 0.12347, "grad_norm": 1.2864293081553977, "learning_rate": 0.003, "loss": 4.0748, "step": 12347 }, { "epoch": 0.12348, "grad_norm": 1.0784610258983045, "learning_rate": 0.003, "loss": 4.0775, "step": 12348 }, { "epoch": 0.12349, "grad_norm": 1.1614271899541546, "learning_rate": 0.003, "loss": 4.0597, "step": 12349 }, { "epoch": 0.1235, "grad_norm": 0.9254677323202702, "learning_rate": 0.003, "loss": 4.0563, "step": 12350 }, { "epoch": 0.12351, "grad_norm": 1.2856580170551943, "learning_rate": 0.003, "loss": 4.0854, "step": 12351 }, { "epoch": 0.12352, "grad_norm": 0.9532348430578328, "learning_rate": 0.003, "loss": 4.0587, "step": 12352 }, { "epoch": 0.12353, "grad_norm": 1.3028059426238214, "learning_rate": 0.003, "loss": 4.0531, "step": 12353 }, { "epoch": 0.12354, "grad_norm": 1.1371787413729926, "learning_rate": 0.003, "loss": 4.0657, "step": 12354 }, { "epoch": 0.12355, "grad_norm": 1.4618199369215297, "learning_rate": 0.003, "loss": 4.0687, "step": 12355 }, { "epoch": 0.12356, "grad_norm": 1.066228952067004, "learning_rate": 0.003, "loss": 4.0583, "step": 12356 }, { "epoch": 0.12357, "grad_norm": 1.1062125587234808, "learning_rate": 0.003, "loss": 4.0535, "step": 12357 }, { "epoch": 0.12358, "grad_norm": 1.1954685144933828, "learning_rate": 0.003, "loss": 4.0642, "step": 12358 }, { "epoch": 0.12359, "grad_norm": 1.0600695155564492, "learning_rate": 0.003, "loss": 4.0454, "step": 12359 }, { "epoch": 0.1236, "grad_norm": 1.2525463893335433, "learning_rate": 0.003, "loss": 4.0565, "step": 12360 }, { "epoch": 0.12361, "grad_norm": 1.1782542376101623, "learning_rate": 0.003, "loss": 4.0608, "step": 12361 }, { "epoch": 0.12362, "grad_norm": 1.1600706645438057, "learning_rate": 0.003, "loss": 4.0821, "step": 12362 }, { "epoch": 0.12363, "grad_norm": 1.137380421954321, "learning_rate": 0.003, "loss": 4.0613, "step": 12363 }, { "epoch": 0.12364, "grad_norm": 1.1280376424988865, "learning_rate": 0.003, "loss": 4.0596, "step": 12364 }, { "epoch": 0.12365, "grad_norm": 1.1698311964249937, "learning_rate": 0.003, "loss": 4.0561, "step": 12365 }, { "epoch": 0.12366, "grad_norm": 1.3353646434819473, "learning_rate": 0.003, "loss": 4.1267, "step": 12366 }, { "epoch": 0.12367, "grad_norm": 1.2863268829142191, "learning_rate": 0.003, "loss": 4.0942, "step": 12367 }, { "epoch": 0.12368, "grad_norm": 1.0029004777191795, "learning_rate": 0.003, "loss": 4.0728, "step": 12368 }, { "epoch": 0.12369, "grad_norm": 1.4233370290194403, "learning_rate": 0.003, "loss": 4.0943, "step": 12369 }, { "epoch": 0.1237, "grad_norm": 1.1184490331663481, "learning_rate": 0.003, "loss": 4.0797, "step": 12370 }, { "epoch": 0.12371, "grad_norm": 1.2770423154469277, "learning_rate": 0.003, "loss": 4.0644, "step": 12371 }, { "epoch": 0.12372, "grad_norm": 0.9439049554373123, "learning_rate": 0.003, "loss": 4.0644, "step": 12372 }, { "epoch": 0.12373, "grad_norm": 1.2224924524375946, "learning_rate": 0.003, "loss": 4.0878, "step": 12373 }, { "epoch": 0.12374, "grad_norm": 1.0573524541014572, "learning_rate": 0.003, "loss": 4.0718, "step": 12374 }, { "epoch": 0.12375, "grad_norm": 1.138607119920257, "learning_rate": 0.003, "loss": 4.0591, "step": 12375 }, { "epoch": 0.12376, "grad_norm": 1.0254491056459425, "learning_rate": 0.003, "loss": 4.0892, "step": 12376 }, { "epoch": 0.12377, "grad_norm": 1.1953961864241487, "learning_rate": 0.003, "loss": 4.0401, "step": 12377 }, { "epoch": 0.12378, "grad_norm": 1.239262872294417, "learning_rate": 0.003, "loss": 4.1009, "step": 12378 }, { "epoch": 0.12379, "grad_norm": 1.3245694630712719, "learning_rate": 0.003, "loss": 4.0637, "step": 12379 }, { "epoch": 0.1238, "grad_norm": 1.0423837674916894, "learning_rate": 0.003, "loss": 4.0808, "step": 12380 }, { "epoch": 0.12381, "grad_norm": 1.1667607294537654, "learning_rate": 0.003, "loss": 4.0783, "step": 12381 }, { "epoch": 0.12382, "grad_norm": 1.1767961078546874, "learning_rate": 0.003, "loss": 4.0634, "step": 12382 }, { "epoch": 0.12383, "grad_norm": 1.187882008900835, "learning_rate": 0.003, "loss": 4.0525, "step": 12383 }, { "epoch": 0.12384, "grad_norm": 1.0051234424869664, "learning_rate": 0.003, "loss": 4.0728, "step": 12384 }, { "epoch": 0.12385, "grad_norm": 1.2969760483111112, "learning_rate": 0.003, "loss": 4.0598, "step": 12385 }, { "epoch": 0.12386, "grad_norm": 1.0697051420470138, "learning_rate": 0.003, "loss": 4.0747, "step": 12386 }, { "epoch": 0.12387, "grad_norm": 1.3510261709568954, "learning_rate": 0.003, "loss": 4.0727, "step": 12387 }, { "epoch": 0.12388, "grad_norm": 1.0318070562588793, "learning_rate": 0.003, "loss": 4.0572, "step": 12388 }, { "epoch": 0.12389, "grad_norm": 1.6188747792730196, "learning_rate": 0.003, "loss": 4.1061, "step": 12389 }, { "epoch": 0.1239, "grad_norm": 0.946262468912621, "learning_rate": 0.003, "loss": 4.1156, "step": 12390 }, { "epoch": 0.12391, "grad_norm": 1.3309245924925017, "learning_rate": 0.003, "loss": 4.044, "step": 12391 }, { "epoch": 0.12392, "grad_norm": 1.2873771647331873, "learning_rate": 0.003, "loss": 4.0733, "step": 12392 }, { "epoch": 0.12393, "grad_norm": 1.157658538183153, "learning_rate": 0.003, "loss": 4.086, "step": 12393 }, { "epoch": 0.12394, "grad_norm": 1.2264491677633973, "learning_rate": 0.003, "loss": 4.0734, "step": 12394 }, { "epoch": 0.12395, "grad_norm": 0.9912588373523377, "learning_rate": 0.003, "loss": 4.0405, "step": 12395 }, { "epoch": 0.12396, "grad_norm": 1.3829782770017245, "learning_rate": 0.003, "loss": 4.1015, "step": 12396 }, { "epoch": 0.12397, "grad_norm": 0.9969910644951696, "learning_rate": 0.003, "loss": 4.0863, "step": 12397 }, { "epoch": 0.12398, "grad_norm": 1.3826345300514964, "learning_rate": 0.003, "loss": 4.108, "step": 12398 }, { "epoch": 0.12399, "grad_norm": 1.0440178845749568, "learning_rate": 0.003, "loss": 4.1056, "step": 12399 }, { "epoch": 0.124, "grad_norm": 1.349613962975738, "learning_rate": 0.003, "loss": 4.1233, "step": 12400 }, { "epoch": 0.12401, "grad_norm": 1.0142913897995074, "learning_rate": 0.003, "loss": 4.0556, "step": 12401 }, { "epoch": 0.12402, "grad_norm": 1.3609411930461677, "learning_rate": 0.003, "loss": 4.0932, "step": 12402 }, { "epoch": 0.12403, "grad_norm": 1.1827122840577082, "learning_rate": 0.003, "loss": 4.0731, "step": 12403 }, { "epoch": 0.12404, "grad_norm": 1.2215790114579566, "learning_rate": 0.003, "loss": 4.0758, "step": 12404 }, { "epoch": 0.12405, "grad_norm": 1.243523400765216, "learning_rate": 0.003, "loss": 4.0836, "step": 12405 }, { "epoch": 0.12406, "grad_norm": 1.1446590993375974, "learning_rate": 0.003, "loss": 4.0662, "step": 12406 }, { "epoch": 0.12407, "grad_norm": 1.2028933130830788, "learning_rate": 0.003, "loss": 4.0632, "step": 12407 }, { "epoch": 0.12408, "grad_norm": 1.0805977575687995, "learning_rate": 0.003, "loss": 4.0823, "step": 12408 }, { "epoch": 0.12409, "grad_norm": 1.2076277518877603, "learning_rate": 0.003, "loss": 4.0493, "step": 12409 }, { "epoch": 0.1241, "grad_norm": 1.0360928232504336, "learning_rate": 0.003, "loss": 4.0839, "step": 12410 }, { "epoch": 0.12411, "grad_norm": 1.2538038581764603, "learning_rate": 0.003, "loss": 4.0569, "step": 12411 }, { "epoch": 0.12412, "grad_norm": 1.117318743300164, "learning_rate": 0.003, "loss": 4.0449, "step": 12412 }, { "epoch": 0.12413, "grad_norm": 1.2765684340945003, "learning_rate": 0.003, "loss": 4.0974, "step": 12413 }, { "epoch": 0.12414, "grad_norm": 1.1275644577185353, "learning_rate": 0.003, "loss": 4.0557, "step": 12414 }, { "epoch": 0.12415, "grad_norm": 1.1486775459125977, "learning_rate": 0.003, "loss": 4.071, "step": 12415 }, { "epoch": 0.12416, "grad_norm": 1.116843978464965, "learning_rate": 0.003, "loss": 4.0913, "step": 12416 }, { "epoch": 0.12417, "grad_norm": 1.0780602688508731, "learning_rate": 0.003, "loss": 4.0864, "step": 12417 }, { "epoch": 0.12418, "grad_norm": 1.2625972608543656, "learning_rate": 0.003, "loss": 4.0809, "step": 12418 }, { "epoch": 0.12419, "grad_norm": 1.136511891701055, "learning_rate": 0.003, "loss": 4.0504, "step": 12419 }, { "epoch": 0.1242, "grad_norm": 1.335064760846712, "learning_rate": 0.003, "loss": 4.0426, "step": 12420 }, { "epoch": 0.12421, "grad_norm": 1.1493168734621346, "learning_rate": 0.003, "loss": 4.095, "step": 12421 }, { "epoch": 0.12422, "grad_norm": 1.2595484150746281, "learning_rate": 0.003, "loss": 4.0704, "step": 12422 }, { "epoch": 0.12423, "grad_norm": 1.1535911104465582, "learning_rate": 0.003, "loss": 4.0799, "step": 12423 }, { "epoch": 0.12424, "grad_norm": 1.1185599234146864, "learning_rate": 0.003, "loss": 4.0307, "step": 12424 }, { "epoch": 0.12425, "grad_norm": 0.9901883288896391, "learning_rate": 0.003, "loss": 4.0564, "step": 12425 }, { "epoch": 0.12426, "grad_norm": 1.2859790206056052, "learning_rate": 0.003, "loss": 4.0586, "step": 12426 }, { "epoch": 0.12427, "grad_norm": 1.2239228314237462, "learning_rate": 0.003, "loss": 4.0714, "step": 12427 }, { "epoch": 0.12428, "grad_norm": 1.223034801205773, "learning_rate": 0.003, "loss": 4.0887, "step": 12428 }, { "epoch": 0.12429, "grad_norm": 1.1267875607961046, "learning_rate": 0.003, "loss": 4.0481, "step": 12429 }, { "epoch": 0.1243, "grad_norm": 1.2112429229194317, "learning_rate": 0.003, "loss": 4.0818, "step": 12430 }, { "epoch": 0.12431, "grad_norm": 1.2009002242700133, "learning_rate": 0.003, "loss": 4.0617, "step": 12431 }, { "epoch": 0.12432, "grad_norm": 1.1621112911264861, "learning_rate": 0.003, "loss": 4.0778, "step": 12432 }, { "epoch": 0.12433, "grad_norm": 1.4528639075118381, "learning_rate": 0.003, "loss": 4.0875, "step": 12433 }, { "epoch": 0.12434, "grad_norm": 0.9828266663691128, "learning_rate": 0.003, "loss": 4.0999, "step": 12434 }, { "epoch": 0.12435, "grad_norm": 1.208413930102256, "learning_rate": 0.003, "loss": 4.0728, "step": 12435 }, { "epoch": 0.12436, "grad_norm": 1.2434917891523287, "learning_rate": 0.003, "loss": 4.05, "step": 12436 }, { "epoch": 0.12437, "grad_norm": 1.240598712732967, "learning_rate": 0.003, "loss": 4.05, "step": 12437 }, { "epoch": 0.12438, "grad_norm": 1.029094900595442, "learning_rate": 0.003, "loss": 4.0447, "step": 12438 }, { "epoch": 0.12439, "grad_norm": 1.349558368050492, "learning_rate": 0.003, "loss": 4.0875, "step": 12439 }, { "epoch": 0.1244, "grad_norm": 0.9220520856370474, "learning_rate": 0.003, "loss": 4.0566, "step": 12440 }, { "epoch": 0.12441, "grad_norm": 1.191546944432022, "learning_rate": 0.003, "loss": 4.0502, "step": 12441 }, { "epoch": 0.12442, "grad_norm": 1.0966955815185557, "learning_rate": 0.003, "loss": 4.075, "step": 12442 }, { "epoch": 0.12443, "grad_norm": 1.181104240860222, "learning_rate": 0.003, "loss": 4.0801, "step": 12443 }, { "epoch": 0.12444, "grad_norm": 1.106511036608127, "learning_rate": 0.003, "loss": 4.0657, "step": 12444 }, { "epoch": 0.12445, "grad_norm": 1.2160044974004254, "learning_rate": 0.003, "loss": 4.0847, "step": 12445 }, { "epoch": 0.12446, "grad_norm": 1.2540860300867798, "learning_rate": 0.003, "loss": 4.0837, "step": 12446 }, { "epoch": 0.12447, "grad_norm": 1.5920465287533634, "learning_rate": 0.003, "loss": 4.1002, "step": 12447 }, { "epoch": 0.12448, "grad_norm": 1.3714492953122197, "learning_rate": 0.003, "loss": 4.0904, "step": 12448 }, { "epoch": 0.12449, "grad_norm": 1.1988615033271683, "learning_rate": 0.003, "loss": 4.107, "step": 12449 }, { "epoch": 0.1245, "grad_norm": 1.2820507313484732, "learning_rate": 0.003, "loss": 4.063, "step": 12450 }, { "epoch": 0.12451, "grad_norm": 1.0123391719593466, "learning_rate": 0.003, "loss": 4.0342, "step": 12451 }, { "epoch": 0.12452, "grad_norm": 1.1708447718933537, "learning_rate": 0.003, "loss": 4.0846, "step": 12452 }, { "epoch": 0.12453, "grad_norm": 1.0224703740085388, "learning_rate": 0.003, "loss": 4.0903, "step": 12453 }, { "epoch": 0.12454, "grad_norm": 1.096286692309235, "learning_rate": 0.003, "loss": 4.0397, "step": 12454 }, { "epoch": 0.12455, "grad_norm": 1.3979018490361652, "learning_rate": 0.003, "loss": 4.0823, "step": 12455 }, { "epoch": 0.12456, "grad_norm": 1.146467679758208, "learning_rate": 0.003, "loss": 4.0493, "step": 12456 }, { "epoch": 0.12457, "grad_norm": 1.5371802869904305, "learning_rate": 0.003, "loss": 4.0867, "step": 12457 }, { "epoch": 0.12458, "grad_norm": 0.9429196785460733, "learning_rate": 0.003, "loss": 4.053, "step": 12458 }, { "epoch": 0.12459, "grad_norm": 1.126333732451378, "learning_rate": 0.003, "loss": 4.0872, "step": 12459 }, { "epoch": 0.1246, "grad_norm": 1.3427249346932093, "learning_rate": 0.003, "loss": 4.101, "step": 12460 }, { "epoch": 0.12461, "grad_norm": 1.2335821482057865, "learning_rate": 0.003, "loss": 4.0707, "step": 12461 }, { "epoch": 0.12462, "grad_norm": 1.2589350122049023, "learning_rate": 0.003, "loss": 4.0685, "step": 12462 }, { "epoch": 0.12463, "grad_norm": 1.1078088631731713, "learning_rate": 0.003, "loss": 4.0695, "step": 12463 }, { "epoch": 0.12464, "grad_norm": 1.156194221905579, "learning_rate": 0.003, "loss": 4.0835, "step": 12464 }, { "epoch": 0.12465, "grad_norm": 1.1287746508449288, "learning_rate": 0.003, "loss": 4.0512, "step": 12465 }, { "epoch": 0.12466, "grad_norm": 1.1136569055957326, "learning_rate": 0.003, "loss": 4.0931, "step": 12466 }, { "epoch": 0.12467, "grad_norm": 1.0799338044455242, "learning_rate": 0.003, "loss": 4.0802, "step": 12467 }, { "epoch": 0.12468, "grad_norm": 1.3289748461710855, "learning_rate": 0.003, "loss": 4.0967, "step": 12468 }, { "epoch": 0.12469, "grad_norm": 1.1151078645112802, "learning_rate": 0.003, "loss": 4.0531, "step": 12469 }, { "epoch": 0.1247, "grad_norm": 1.1463534088936385, "learning_rate": 0.003, "loss": 4.0414, "step": 12470 }, { "epoch": 0.12471, "grad_norm": 1.1519553387047017, "learning_rate": 0.003, "loss": 4.0921, "step": 12471 }, { "epoch": 0.12472, "grad_norm": 1.2285550661794005, "learning_rate": 0.003, "loss": 4.0891, "step": 12472 }, { "epoch": 0.12473, "grad_norm": 1.1215923088146995, "learning_rate": 0.003, "loss": 4.0577, "step": 12473 }, { "epoch": 0.12474, "grad_norm": 1.231438287323106, "learning_rate": 0.003, "loss": 4.083, "step": 12474 }, { "epoch": 0.12475, "grad_norm": 1.0679941372411055, "learning_rate": 0.003, "loss": 4.063, "step": 12475 }, { "epoch": 0.12476, "grad_norm": 1.0239810114957488, "learning_rate": 0.003, "loss": 4.0943, "step": 12476 }, { "epoch": 0.12477, "grad_norm": 1.3657673364526786, "learning_rate": 0.003, "loss": 4.0519, "step": 12477 }, { "epoch": 0.12478, "grad_norm": 1.066101071076436, "learning_rate": 0.003, "loss": 4.0469, "step": 12478 }, { "epoch": 0.12479, "grad_norm": 1.7640966435197194, "learning_rate": 0.003, "loss": 4.0863, "step": 12479 }, { "epoch": 0.1248, "grad_norm": 1.102522439741497, "learning_rate": 0.003, "loss": 4.075, "step": 12480 }, { "epoch": 0.12481, "grad_norm": 1.241907969502746, "learning_rate": 0.003, "loss": 4.0576, "step": 12481 }, { "epoch": 0.12482, "grad_norm": 1.1094473374941114, "learning_rate": 0.003, "loss": 4.0878, "step": 12482 }, { "epoch": 0.12483, "grad_norm": 1.3305339634481412, "learning_rate": 0.003, "loss": 4.0907, "step": 12483 }, { "epoch": 0.12484, "grad_norm": 1.04860359051801, "learning_rate": 0.003, "loss": 4.0304, "step": 12484 }, { "epoch": 0.12485, "grad_norm": 1.3466472306173978, "learning_rate": 0.003, "loss": 4.0898, "step": 12485 }, { "epoch": 0.12486, "grad_norm": 1.091955860046152, "learning_rate": 0.003, "loss": 4.0518, "step": 12486 }, { "epoch": 0.12487, "grad_norm": 1.1542843548616597, "learning_rate": 0.003, "loss": 4.0371, "step": 12487 }, { "epoch": 0.12488, "grad_norm": 1.2977298649449636, "learning_rate": 0.003, "loss": 4.0798, "step": 12488 }, { "epoch": 0.12489, "grad_norm": 1.1502135615273064, "learning_rate": 0.003, "loss": 4.0585, "step": 12489 }, { "epoch": 0.1249, "grad_norm": 1.1954733894997076, "learning_rate": 0.003, "loss": 4.0756, "step": 12490 }, { "epoch": 0.12491, "grad_norm": 1.1477409651605421, "learning_rate": 0.003, "loss": 4.0758, "step": 12491 }, { "epoch": 0.12492, "grad_norm": 1.0229908159016856, "learning_rate": 0.003, "loss": 4.0881, "step": 12492 }, { "epoch": 0.12493, "grad_norm": 1.397010869196908, "learning_rate": 0.003, "loss": 4.0885, "step": 12493 }, { "epoch": 0.12494, "grad_norm": 1.0552851041740259, "learning_rate": 0.003, "loss": 4.057, "step": 12494 }, { "epoch": 0.12495, "grad_norm": 1.2568137758291984, "learning_rate": 0.003, "loss": 4.0789, "step": 12495 }, { "epoch": 0.12496, "grad_norm": 1.3314768441604752, "learning_rate": 0.003, "loss": 4.0586, "step": 12496 }, { "epoch": 0.12497, "grad_norm": 1.4693373016661582, "learning_rate": 0.003, "loss": 4.0761, "step": 12497 }, { "epoch": 0.12498, "grad_norm": 0.940180612411933, "learning_rate": 0.003, "loss": 4.0701, "step": 12498 }, { "epoch": 0.12499, "grad_norm": 1.0852382733465447, "learning_rate": 0.003, "loss": 4.0766, "step": 12499 }, { "epoch": 0.125, "grad_norm": 1.5622120700812359, "learning_rate": 0.003, "loss": 4.124, "step": 12500 }, { "epoch": 0.12501, "grad_norm": 0.8739362980405638, "learning_rate": 0.003, "loss": 4.0846, "step": 12501 }, { "epoch": 0.12502, "grad_norm": 1.0636747915337694, "learning_rate": 0.003, "loss": 4.0717, "step": 12502 }, { "epoch": 0.12503, "grad_norm": 1.2278657194272873, "learning_rate": 0.003, "loss": 4.0839, "step": 12503 }, { "epoch": 0.12504, "grad_norm": 1.0974608415259097, "learning_rate": 0.003, "loss": 4.0822, "step": 12504 }, { "epoch": 0.12505, "grad_norm": 1.1248766463340456, "learning_rate": 0.003, "loss": 4.0924, "step": 12505 }, { "epoch": 0.12506, "grad_norm": 1.2399103972271988, "learning_rate": 0.003, "loss": 4.0829, "step": 12506 }, { "epoch": 0.12507, "grad_norm": 1.1756145223100918, "learning_rate": 0.003, "loss": 4.0536, "step": 12507 }, { "epoch": 0.12508, "grad_norm": 1.1015940953540795, "learning_rate": 0.003, "loss": 4.0744, "step": 12508 }, { "epoch": 0.12509, "grad_norm": 1.1879648693043818, "learning_rate": 0.003, "loss": 4.0779, "step": 12509 }, { "epoch": 0.1251, "grad_norm": 1.1637957279316695, "learning_rate": 0.003, "loss": 4.0433, "step": 12510 }, { "epoch": 0.12511, "grad_norm": 1.2494606772760055, "learning_rate": 0.003, "loss": 4.0628, "step": 12511 }, { "epoch": 0.12512, "grad_norm": 1.2181112953250357, "learning_rate": 0.003, "loss": 4.0736, "step": 12512 }, { "epoch": 0.12513, "grad_norm": 0.9729638548913785, "learning_rate": 0.003, "loss": 4.0762, "step": 12513 }, { "epoch": 0.12514, "grad_norm": 1.2901127381666835, "learning_rate": 0.003, "loss": 4.0731, "step": 12514 }, { "epoch": 0.12515, "grad_norm": 1.0976102482057357, "learning_rate": 0.003, "loss": 4.069, "step": 12515 }, { "epoch": 0.12516, "grad_norm": 1.336702885012363, "learning_rate": 0.003, "loss": 4.0849, "step": 12516 }, { "epoch": 0.12517, "grad_norm": 1.0272505207107212, "learning_rate": 0.003, "loss": 4.0381, "step": 12517 }, { "epoch": 0.12518, "grad_norm": 1.3057449911660892, "learning_rate": 0.003, "loss": 4.0941, "step": 12518 }, { "epoch": 0.12519, "grad_norm": 1.218532581003206, "learning_rate": 0.003, "loss": 4.0779, "step": 12519 }, { "epoch": 0.1252, "grad_norm": 1.0277171173468325, "learning_rate": 0.003, "loss": 4.0312, "step": 12520 }, { "epoch": 0.12521, "grad_norm": 1.48322118713419, "learning_rate": 0.003, "loss": 4.0858, "step": 12521 }, { "epoch": 0.12522, "grad_norm": 1.028286503421473, "learning_rate": 0.003, "loss": 4.1146, "step": 12522 }, { "epoch": 0.12523, "grad_norm": 1.2785485432284658, "learning_rate": 0.003, "loss": 4.0649, "step": 12523 }, { "epoch": 0.12524, "grad_norm": 1.133692151244233, "learning_rate": 0.003, "loss": 4.0533, "step": 12524 }, { "epoch": 0.12525, "grad_norm": 1.3082450711743057, "learning_rate": 0.003, "loss": 4.0517, "step": 12525 }, { "epoch": 0.12526, "grad_norm": 1.2367359920672203, "learning_rate": 0.003, "loss": 4.1063, "step": 12526 }, { "epoch": 0.12527, "grad_norm": 1.1937292021887054, "learning_rate": 0.003, "loss": 4.1004, "step": 12527 }, { "epoch": 0.12528, "grad_norm": 1.2382152188842006, "learning_rate": 0.003, "loss": 4.0599, "step": 12528 }, { "epoch": 0.12529, "grad_norm": 1.3361864008371442, "learning_rate": 0.003, "loss": 4.0611, "step": 12529 }, { "epoch": 0.1253, "grad_norm": 0.9484016556199034, "learning_rate": 0.003, "loss": 4.0482, "step": 12530 }, { "epoch": 0.12531, "grad_norm": 1.1821540840825915, "learning_rate": 0.003, "loss": 4.0708, "step": 12531 }, { "epoch": 0.12532, "grad_norm": 1.270341277351583, "learning_rate": 0.003, "loss": 4.0605, "step": 12532 }, { "epoch": 0.12533, "grad_norm": 1.0751666876115358, "learning_rate": 0.003, "loss": 4.0692, "step": 12533 }, { "epoch": 0.12534, "grad_norm": 1.1701969786843704, "learning_rate": 0.003, "loss": 4.0638, "step": 12534 }, { "epoch": 0.12535, "grad_norm": 1.065142223277472, "learning_rate": 0.003, "loss": 4.0767, "step": 12535 }, { "epoch": 0.12536, "grad_norm": 1.269846923699688, "learning_rate": 0.003, "loss": 4.0139, "step": 12536 }, { "epoch": 0.12537, "grad_norm": 1.1512438095355702, "learning_rate": 0.003, "loss": 4.0725, "step": 12537 }, { "epoch": 0.12538, "grad_norm": 1.0280753584333013, "learning_rate": 0.003, "loss": 4.0688, "step": 12538 }, { "epoch": 0.12539, "grad_norm": 1.1432091793641852, "learning_rate": 0.003, "loss": 4.0979, "step": 12539 }, { "epoch": 0.1254, "grad_norm": 1.0965031416949658, "learning_rate": 0.003, "loss": 4.0683, "step": 12540 }, { "epoch": 0.12541, "grad_norm": 1.3324295530793964, "learning_rate": 0.003, "loss": 4.0516, "step": 12541 }, { "epoch": 0.12542, "grad_norm": 1.082712427398565, "learning_rate": 0.003, "loss": 4.0744, "step": 12542 }, { "epoch": 0.12543, "grad_norm": 1.4584132011678437, "learning_rate": 0.003, "loss": 4.0391, "step": 12543 }, { "epoch": 0.12544, "grad_norm": 1.0347081253877564, "learning_rate": 0.003, "loss": 4.1032, "step": 12544 }, { "epoch": 0.12545, "grad_norm": 1.4080593441623668, "learning_rate": 0.003, "loss": 4.0741, "step": 12545 }, { "epoch": 0.12546, "grad_norm": 0.94172210940559, "learning_rate": 0.003, "loss": 4.0597, "step": 12546 }, { "epoch": 0.12547, "grad_norm": 1.204417126633597, "learning_rate": 0.003, "loss": 4.0996, "step": 12547 }, { "epoch": 0.12548, "grad_norm": 1.153669862341393, "learning_rate": 0.003, "loss": 4.0533, "step": 12548 }, { "epoch": 0.12549, "grad_norm": 1.1623807046879422, "learning_rate": 0.003, "loss": 4.0824, "step": 12549 }, { "epoch": 0.1255, "grad_norm": 1.2223824905007339, "learning_rate": 0.003, "loss": 4.0646, "step": 12550 }, { "epoch": 0.12551, "grad_norm": 1.401806814284153, "learning_rate": 0.003, "loss": 4.091, "step": 12551 }, { "epoch": 0.12552, "grad_norm": 1.0804692663089763, "learning_rate": 0.003, "loss": 4.043, "step": 12552 }, { "epoch": 0.12553, "grad_norm": 1.3223499261165146, "learning_rate": 0.003, "loss": 4.0898, "step": 12553 }, { "epoch": 0.12554, "grad_norm": 1.2048586744932688, "learning_rate": 0.003, "loss": 4.0672, "step": 12554 }, { "epoch": 0.12555, "grad_norm": 1.220463489847063, "learning_rate": 0.003, "loss": 4.0851, "step": 12555 }, { "epoch": 0.12556, "grad_norm": 1.2298530213918342, "learning_rate": 0.003, "loss": 4.103, "step": 12556 }, { "epoch": 0.12557, "grad_norm": 1.107477371434991, "learning_rate": 0.003, "loss": 4.0901, "step": 12557 }, { "epoch": 0.12558, "grad_norm": 1.4953971908411758, "learning_rate": 0.003, "loss": 4.0848, "step": 12558 }, { "epoch": 0.12559, "grad_norm": 1.0919959852714403, "learning_rate": 0.003, "loss": 4.0501, "step": 12559 }, { "epoch": 0.1256, "grad_norm": 1.4031108439000246, "learning_rate": 0.003, "loss": 4.0749, "step": 12560 }, { "epoch": 0.12561, "grad_norm": 0.9274982786809287, "learning_rate": 0.003, "loss": 4.0433, "step": 12561 }, { "epoch": 0.12562, "grad_norm": 1.148283669382533, "learning_rate": 0.003, "loss": 4.0665, "step": 12562 }, { "epoch": 0.12563, "grad_norm": 1.252732398453347, "learning_rate": 0.003, "loss": 4.0714, "step": 12563 }, { "epoch": 0.12564, "grad_norm": 1.179501544509447, "learning_rate": 0.003, "loss": 4.0771, "step": 12564 }, { "epoch": 0.12565, "grad_norm": 1.2630777174253471, "learning_rate": 0.003, "loss": 4.0837, "step": 12565 }, { "epoch": 0.12566, "grad_norm": 1.1104232624786845, "learning_rate": 0.003, "loss": 4.0819, "step": 12566 }, { "epoch": 0.12567, "grad_norm": 1.0172810898817368, "learning_rate": 0.003, "loss": 4.1082, "step": 12567 }, { "epoch": 0.12568, "grad_norm": 1.1101095264680614, "learning_rate": 0.003, "loss": 4.0639, "step": 12568 }, { "epoch": 0.12569, "grad_norm": 1.0660138431796577, "learning_rate": 0.003, "loss": 4.0645, "step": 12569 }, { "epoch": 0.1257, "grad_norm": 1.1675985526701762, "learning_rate": 0.003, "loss": 4.1186, "step": 12570 }, { "epoch": 0.12571, "grad_norm": 1.1076673801819517, "learning_rate": 0.003, "loss": 4.0837, "step": 12571 }, { "epoch": 0.12572, "grad_norm": 1.2810009421141986, "learning_rate": 0.003, "loss": 4.0865, "step": 12572 }, { "epoch": 0.12573, "grad_norm": 1.3484034745623017, "learning_rate": 0.003, "loss": 4.0561, "step": 12573 }, { "epoch": 0.12574, "grad_norm": 1.3426946654575465, "learning_rate": 0.003, "loss": 4.0869, "step": 12574 }, { "epoch": 0.12575, "grad_norm": 0.9506284253662771, "learning_rate": 0.003, "loss": 4.1076, "step": 12575 }, { "epoch": 0.12576, "grad_norm": 1.3356304158256906, "learning_rate": 0.003, "loss": 4.0808, "step": 12576 }, { "epoch": 0.12577, "grad_norm": 1.3499833136159347, "learning_rate": 0.003, "loss": 4.0622, "step": 12577 }, { "epoch": 0.12578, "grad_norm": 0.9853562356103978, "learning_rate": 0.003, "loss": 4.0601, "step": 12578 }, { "epoch": 0.12579, "grad_norm": 1.1326976276614282, "learning_rate": 0.003, "loss": 4.081, "step": 12579 }, { "epoch": 0.1258, "grad_norm": 1.0176354143472115, "learning_rate": 0.003, "loss": 4.0771, "step": 12580 }, { "epoch": 0.12581, "grad_norm": 1.4072549274790551, "learning_rate": 0.003, "loss": 4.0469, "step": 12581 }, { "epoch": 0.12582, "grad_norm": 0.7865868508456874, "learning_rate": 0.003, "loss": 4.0364, "step": 12582 }, { "epoch": 0.12583, "grad_norm": 0.9827649168754795, "learning_rate": 0.003, "loss": 4.0865, "step": 12583 }, { "epoch": 0.12584, "grad_norm": 1.3261940918286663, "learning_rate": 0.003, "loss": 4.059, "step": 12584 }, { "epoch": 0.12585, "grad_norm": 1.0096186221785435, "learning_rate": 0.003, "loss": 4.1009, "step": 12585 }, { "epoch": 0.12586, "grad_norm": 1.550758321674492, "learning_rate": 0.003, "loss": 4.0832, "step": 12586 }, { "epoch": 0.12587, "grad_norm": 1.28775237574974, "learning_rate": 0.003, "loss": 4.1042, "step": 12587 }, { "epoch": 0.12588, "grad_norm": 1.212541757461834, "learning_rate": 0.003, "loss": 4.0605, "step": 12588 }, { "epoch": 0.12589, "grad_norm": 1.0041117070411152, "learning_rate": 0.003, "loss": 4.0749, "step": 12589 }, { "epoch": 0.1259, "grad_norm": 1.3697511255613661, "learning_rate": 0.003, "loss": 4.0687, "step": 12590 }, { "epoch": 0.12591, "grad_norm": 1.0498516362975538, "learning_rate": 0.003, "loss": 4.0617, "step": 12591 }, { "epoch": 0.12592, "grad_norm": 1.4301823880591849, "learning_rate": 0.003, "loss": 4.0733, "step": 12592 }, { "epoch": 0.12593, "grad_norm": 0.9182078521998195, "learning_rate": 0.003, "loss": 4.0768, "step": 12593 }, { "epoch": 0.12594, "grad_norm": 1.1837839051014718, "learning_rate": 0.003, "loss": 4.0844, "step": 12594 }, { "epoch": 0.12595, "grad_norm": 1.298135818800885, "learning_rate": 0.003, "loss": 4.1093, "step": 12595 }, { "epoch": 0.12596, "grad_norm": 1.041094948367079, "learning_rate": 0.003, "loss": 4.042, "step": 12596 }, { "epoch": 0.12597, "grad_norm": 1.490598566038195, "learning_rate": 0.003, "loss": 4.0902, "step": 12597 }, { "epoch": 0.12598, "grad_norm": 1.0931752922510034, "learning_rate": 0.003, "loss": 4.0962, "step": 12598 }, { "epoch": 0.12599, "grad_norm": 1.2504967445814124, "learning_rate": 0.003, "loss": 4.1034, "step": 12599 }, { "epoch": 0.126, "grad_norm": 1.1377759522818098, "learning_rate": 0.003, "loss": 4.0966, "step": 12600 }, { "epoch": 0.12601, "grad_norm": 1.15656521401536, "learning_rate": 0.003, "loss": 4.1128, "step": 12601 }, { "epoch": 0.12602, "grad_norm": 0.9867574592186018, "learning_rate": 0.003, "loss": 4.0863, "step": 12602 }, { "epoch": 0.12603, "grad_norm": 1.26686466484779, "learning_rate": 0.003, "loss": 4.0304, "step": 12603 }, { "epoch": 0.12604, "grad_norm": 1.091406672024988, "learning_rate": 0.003, "loss": 4.0985, "step": 12604 }, { "epoch": 0.12605, "grad_norm": 1.1478402161573151, "learning_rate": 0.003, "loss": 4.0698, "step": 12605 }, { "epoch": 0.12606, "grad_norm": 1.2518292789824488, "learning_rate": 0.003, "loss": 4.0889, "step": 12606 }, { "epoch": 0.12607, "grad_norm": 1.2143300345493522, "learning_rate": 0.003, "loss": 4.0658, "step": 12607 }, { "epoch": 0.12608, "grad_norm": 1.1601393056615503, "learning_rate": 0.003, "loss": 4.0628, "step": 12608 }, { "epoch": 0.12609, "grad_norm": 1.2908871565774243, "learning_rate": 0.003, "loss": 4.073, "step": 12609 }, { "epoch": 0.1261, "grad_norm": 1.0781328575412508, "learning_rate": 0.003, "loss": 4.0723, "step": 12610 }, { "epoch": 0.12611, "grad_norm": 1.3641588513571903, "learning_rate": 0.003, "loss": 4.0911, "step": 12611 }, { "epoch": 0.12612, "grad_norm": 0.9713706238293194, "learning_rate": 0.003, "loss": 4.0616, "step": 12612 }, { "epoch": 0.12613, "grad_norm": 1.2393486901278956, "learning_rate": 0.003, "loss": 4.0399, "step": 12613 }, { "epoch": 0.12614, "grad_norm": 1.1093799942750657, "learning_rate": 0.003, "loss": 4.0846, "step": 12614 }, { "epoch": 0.12615, "grad_norm": 1.270999707985446, "learning_rate": 0.003, "loss": 4.0987, "step": 12615 }, { "epoch": 0.12616, "grad_norm": 1.1626868005815518, "learning_rate": 0.003, "loss": 4.0491, "step": 12616 }, { "epoch": 0.12617, "grad_norm": 1.1501465323341438, "learning_rate": 0.003, "loss": 4.0577, "step": 12617 }, { "epoch": 0.12618, "grad_norm": 1.0542480027114456, "learning_rate": 0.003, "loss": 4.0547, "step": 12618 }, { "epoch": 0.12619, "grad_norm": 1.210139597853682, "learning_rate": 0.003, "loss": 4.0629, "step": 12619 }, { "epoch": 0.1262, "grad_norm": 1.2551499777634003, "learning_rate": 0.003, "loss": 4.0865, "step": 12620 }, { "epoch": 0.12621, "grad_norm": 1.1108012833404652, "learning_rate": 0.003, "loss": 4.0669, "step": 12621 }, { "epoch": 0.12622, "grad_norm": 1.0156488990422134, "learning_rate": 0.003, "loss": 4.071, "step": 12622 }, { "epoch": 0.12623, "grad_norm": 1.1962566895669906, "learning_rate": 0.003, "loss": 4.0767, "step": 12623 }, { "epoch": 0.12624, "grad_norm": 1.0158639537186367, "learning_rate": 0.003, "loss": 4.0604, "step": 12624 }, { "epoch": 0.12625, "grad_norm": 1.3317713116392733, "learning_rate": 0.003, "loss": 4.0761, "step": 12625 }, { "epoch": 0.12626, "grad_norm": 0.9373068860721429, "learning_rate": 0.003, "loss": 4.0537, "step": 12626 }, { "epoch": 0.12627, "grad_norm": 1.2071017440177256, "learning_rate": 0.003, "loss": 4.0686, "step": 12627 }, { "epoch": 0.12628, "grad_norm": 1.2255144596796357, "learning_rate": 0.003, "loss": 4.0495, "step": 12628 }, { "epoch": 0.12629, "grad_norm": 1.199131934289616, "learning_rate": 0.003, "loss": 4.0963, "step": 12629 }, { "epoch": 0.1263, "grad_norm": 1.262215865484067, "learning_rate": 0.003, "loss": 4.057, "step": 12630 }, { "epoch": 0.12631, "grad_norm": 1.0301353256706713, "learning_rate": 0.003, "loss": 4.0625, "step": 12631 }, { "epoch": 0.12632, "grad_norm": 1.1811291528410854, "learning_rate": 0.003, "loss": 4.0714, "step": 12632 }, { "epoch": 0.12633, "grad_norm": 1.2521403597119023, "learning_rate": 0.003, "loss": 4.0608, "step": 12633 }, { "epoch": 0.12634, "grad_norm": 1.104551521815721, "learning_rate": 0.003, "loss": 4.0771, "step": 12634 }, { "epoch": 0.12635, "grad_norm": 1.1228874241048405, "learning_rate": 0.003, "loss": 4.1076, "step": 12635 }, { "epoch": 0.12636, "grad_norm": 1.217972941351292, "learning_rate": 0.003, "loss": 4.0743, "step": 12636 }, { "epoch": 0.12637, "grad_norm": 1.5946769636593079, "learning_rate": 0.003, "loss": 4.0865, "step": 12637 }, { "epoch": 0.12638, "grad_norm": 0.9998925575565182, "learning_rate": 0.003, "loss": 4.0727, "step": 12638 }, { "epoch": 0.12639, "grad_norm": 1.4039675600512114, "learning_rate": 0.003, "loss": 4.0875, "step": 12639 }, { "epoch": 0.1264, "grad_norm": 1.003036680605145, "learning_rate": 0.003, "loss": 4.0708, "step": 12640 }, { "epoch": 0.12641, "grad_norm": 1.4049383754523674, "learning_rate": 0.003, "loss": 4.0787, "step": 12641 }, { "epoch": 0.12642, "grad_norm": 1.305155487945457, "learning_rate": 0.003, "loss": 4.0728, "step": 12642 }, { "epoch": 0.12643, "grad_norm": 0.8031734594976118, "learning_rate": 0.003, "loss": 4.0844, "step": 12643 }, { "epoch": 0.12644, "grad_norm": 0.8521750456478882, "learning_rate": 0.003, "loss": 4.0585, "step": 12644 }, { "epoch": 0.12645, "grad_norm": 1.2685751637446443, "learning_rate": 0.003, "loss": 4.0798, "step": 12645 }, { "epoch": 0.12646, "grad_norm": 1.3968989411592685, "learning_rate": 0.003, "loss": 4.0606, "step": 12646 }, { "epoch": 0.12647, "grad_norm": 1.073636686301602, "learning_rate": 0.003, "loss": 4.0499, "step": 12647 }, { "epoch": 0.12648, "grad_norm": 1.1945443103769646, "learning_rate": 0.003, "loss": 4.1042, "step": 12648 }, { "epoch": 0.12649, "grad_norm": 1.190570374890015, "learning_rate": 0.003, "loss": 4.0451, "step": 12649 }, { "epoch": 0.1265, "grad_norm": 1.166870543567209, "learning_rate": 0.003, "loss": 4.0995, "step": 12650 }, { "epoch": 0.12651, "grad_norm": 1.0606374742427602, "learning_rate": 0.003, "loss": 4.076, "step": 12651 }, { "epoch": 0.12652, "grad_norm": 1.0566125104734068, "learning_rate": 0.003, "loss": 4.0559, "step": 12652 }, { "epoch": 0.12653, "grad_norm": 1.2680473842459052, "learning_rate": 0.003, "loss": 4.0697, "step": 12653 }, { "epoch": 0.12654, "grad_norm": 1.466838845584005, "learning_rate": 0.003, "loss": 4.0615, "step": 12654 }, { "epoch": 0.12655, "grad_norm": 1.1214819051890923, "learning_rate": 0.003, "loss": 4.0778, "step": 12655 }, { "epoch": 0.12656, "grad_norm": 1.0688303461370197, "learning_rate": 0.003, "loss": 4.0965, "step": 12656 }, { "epoch": 0.12657, "grad_norm": 1.259173862912746, "learning_rate": 0.003, "loss": 4.0685, "step": 12657 }, { "epoch": 0.12658, "grad_norm": 1.0776915341427002, "learning_rate": 0.003, "loss": 4.0963, "step": 12658 }, { "epoch": 0.12659, "grad_norm": 1.2629615894034179, "learning_rate": 0.003, "loss": 4.0478, "step": 12659 }, { "epoch": 0.1266, "grad_norm": 1.2892363491649281, "learning_rate": 0.003, "loss": 4.061, "step": 12660 }, { "epoch": 0.12661, "grad_norm": 1.3285413004792743, "learning_rate": 0.003, "loss": 4.0737, "step": 12661 }, { "epoch": 0.12662, "grad_norm": 1.180371463792909, "learning_rate": 0.003, "loss": 4.0666, "step": 12662 }, { "epoch": 0.12663, "grad_norm": 1.208193650593098, "learning_rate": 0.003, "loss": 4.0555, "step": 12663 }, { "epoch": 0.12664, "grad_norm": 1.0500808440531122, "learning_rate": 0.003, "loss": 4.0619, "step": 12664 }, { "epoch": 0.12665, "grad_norm": 1.1607253819424002, "learning_rate": 0.003, "loss": 4.0805, "step": 12665 }, { "epoch": 0.12666, "grad_norm": 1.022073776534333, "learning_rate": 0.003, "loss": 4.0621, "step": 12666 }, { "epoch": 0.12667, "grad_norm": 1.2749582096083807, "learning_rate": 0.003, "loss": 4.0529, "step": 12667 }, { "epoch": 0.12668, "grad_norm": 1.1541879354072204, "learning_rate": 0.003, "loss": 4.0618, "step": 12668 }, { "epoch": 0.12669, "grad_norm": 1.0729069740078794, "learning_rate": 0.003, "loss": 4.0935, "step": 12669 }, { "epoch": 0.1267, "grad_norm": 1.0752540096809793, "learning_rate": 0.003, "loss": 4.039, "step": 12670 }, { "epoch": 0.12671, "grad_norm": 1.3556142957785262, "learning_rate": 0.003, "loss": 4.0844, "step": 12671 }, { "epoch": 0.12672, "grad_norm": 1.0754056558863596, "learning_rate": 0.003, "loss": 4.0566, "step": 12672 }, { "epoch": 0.12673, "grad_norm": 1.4216700371204218, "learning_rate": 0.003, "loss": 4.0865, "step": 12673 }, { "epoch": 0.12674, "grad_norm": 1.130987561730525, "learning_rate": 0.003, "loss": 4.069, "step": 12674 }, { "epoch": 0.12675, "grad_norm": 1.3325284780537245, "learning_rate": 0.003, "loss": 4.0965, "step": 12675 }, { "epoch": 0.12676, "grad_norm": 1.1970970779957901, "learning_rate": 0.003, "loss": 4.0586, "step": 12676 }, { "epoch": 0.12677, "grad_norm": 1.0600549073306158, "learning_rate": 0.003, "loss": 4.0447, "step": 12677 }, { "epoch": 0.12678, "grad_norm": 1.1510265499571368, "learning_rate": 0.003, "loss": 4.0718, "step": 12678 }, { "epoch": 0.12679, "grad_norm": 1.1025950728150287, "learning_rate": 0.003, "loss": 4.0806, "step": 12679 }, { "epoch": 0.1268, "grad_norm": 1.5977246430944323, "learning_rate": 0.003, "loss": 4.0841, "step": 12680 }, { "epoch": 0.12681, "grad_norm": 1.2572186245403059, "learning_rate": 0.003, "loss": 4.0899, "step": 12681 }, { "epoch": 0.12682, "grad_norm": 1.0544016504397105, "learning_rate": 0.003, "loss": 4.0747, "step": 12682 }, { "epoch": 0.12683, "grad_norm": 1.277683584940182, "learning_rate": 0.003, "loss": 4.0946, "step": 12683 }, { "epoch": 0.12684, "grad_norm": 1.1984352962734288, "learning_rate": 0.003, "loss": 4.093, "step": 12684 }, { "epoch": 0.12685, "grad_norm": 0.974734679415916, "learning_rate": 0.003, "loss": 4.0769, "step": 12685 }, { "epoch": 0.12686, "grad_norm": 1.1643936922516476, "learning_rate": 0.003, "loss": 4.088, "step": 12686 }, { "epoch": 0.12687, "grad_norm": 1.2262435368031364, "learning_rate": 0.003, "loss": 4.0502, "step": 12687 }, { "epoch": 0.12688, "grad_norm": 1.2296156118738668, "learning_rate": 0.003, "loss": 4.0695, "step": 12688 }, { "epoch": 0.12689, "grad_norm": 1.0892106921141844, "learning_rate": 0.003, "loss": 4.0589, "step": 12689 }, { "epoch": 0.1269, "grad_norm": 1.0366641496791478, "learning_rate": 0.003, "loss": 4.0626, "step": 12690 }, { "epoch": 0.12691, "grad_norm": 1.3342956070153213, "learning_rate": 0.003, "loss": 4.0569, "step": 12691 }, { "epoch": 0.12692, "grad_norm": 0.9619591761019171, "learning_rate": 0.003, "loss": 4.087, "step": 12692 }, { "epoch": 0.12693, "grad_norm": 1.4034682137840007, "learning_rate": 0.003, "loss": 4.0672, "step": 12693 }, { "epoch": 0.12694, "grad_norm": 1.0726913636880717, "learning_rate": 0.003, "loss": 4.0444, "step": 12694 }, { "epoch": 0.12695, "grad_norm": 1.6632780527157962, "learning_rate": 0.003, "loss": 4.0937, "step": 12695 }, { "epoch": 0.12696, "grad_norm": 0.8681943038818645, "learning_rate": 0.003, "loss": 4.0598, "step": 12696 }, { "epoch": 0.12697, "grad_norm": 1.0063370084866963, "learning_rate": 0.003, "loss": 4.0381, "step": 12697 }, { "epoch": 0.12698, "grad_norm": 1.3100329034788503, "learning_rate": 0.003, "loss": 4.0585, "step": 12698 }, { "epoch": 0.12699, "grad_norm": 1.133944589846256, "learning_rate": 0.003, "loss": 4.0458, "step": 12699 }, { "epoch": 0.127, "grad_norm": 1.1630734414695099, "learning_rate": 0.003, "loss": 4.0597, "step": 12700 }, { "epoch": 0.12701, "grad_norm": 1.09819934090099, "learning_rate": 0.003, "loss": 4.0552, "step": 12701 }, { "epoch": 0.12702, "grad_norm": 1.2941794036977707, "learning_rate": 0.003, "loss": 4.0774, "step": 12702 }, { "epoch": 0.12703, "grad_norm": 1.2328994449759347, "learning_rate": 0.003, "loss": 4.0775, "step": 12703 }, { "epoch": 0.12704, "grad_norm": 1.1596742990374647, "learning_rate": 0.003, "loss": 4.0835, "step": 12704 }, { "epoch": 0.12705, "grad_norm": 1.249626557969983, "learning_rate": 0.003, "loss": 4.0757, "step": 12705 }, { "epoch": 0.12706, "grad_norm": 1.0664166312741485, "learning_rate": 0.003, "loss": 4.0588, "step": 12706 }, { "epoch": 0.12707, "grad_norm": 1.213254277457309, "learning_rate": 0.003, "loss": 4.0792, "step": 12707 }, { "epoch": 0.12708, "grad_norm": 0.9958840449855603, "learning_rate": 0.003, "loss": 4.0867, "step": 12708 }, { "epoch": 0.12709, "grad_norm": 1.3146805427924255, "learning_rate": 0.003, "loss": 4.079, "step": 12709 }, { "epoch": 0.1271, "grad_norm": 1.0398187077801997, "learning_rate": 0.003, "loss": 4.0741, "step": 12710 }, { "epoch": 0.12711, "grad_norm": 1.1626402744939957, "learning_rate": 0.003, "loss": 4.0607, "step": 12711 }, { "epoch": 0.12712, "grad_norm": 1.1473284230800858, "learning_rate": 0.003, "loss": 4.0589, "step": 12712 }, { "epoch": 0.12713, "grad_norm": 1.0848536100335324, "learning_rate": 0.003, "loss": 4.0701, "step": 12713 }, { "epoch": 0.12714, "grad_norm": 1.3831331715427961, "learning_rate": 0.003, "loss": 4.0968, "step": 12714 }, { "epoch": 0.12715, "grad_norm": 1.1338759456068388, "learning_rate": 0.003, "loss": 4.0434, "step": 12715 }, { "epoch": 0.12716, "grad_norm": 1.1864130921092912, "learning_rate": 0.003, "loss": 4.0726, "step": 12716 }, { "epoch": 0.12717, "grad_norm": 1.201188020850399, "learning_rate": 0.003, "loss": 4.0853, "step": 12717 }, { "epoch": 0.12718, "grad_norm": 1.1889820047256625, "learning_rate": 0.003, "loss": 4.0656, "step": 12718 }, { "epoch": 0.12719, "grad_norm": 1.3267095900035595, "learning_rate": 0.003, "loss": 4.089, "step": 12719 }, { "epoch": 0.1272, "grad_norm": 1.105599328700241, "learning_rate": 0.003, "loss": 4.036, "step": 12720 }, { "epoch": 0.12721, "grad_norm": 1.340115460536244, "learning_rate": 0.003, "loss": 4.09, "step": 12721 }, { "epoch": 0.12722, "grad_norm": 1.206281956639625, "learning_rate": 0.003, "loss": 4.0829, "step": 12722 }, { "epoch": 0.12723, "grad_norm": 1.3126374431751684, "learning_rate": 0.003, "loss": 4.0845, "step": 12723 }, { "epoch": 0.12724, "grad_norm": 1.3455412140019556, "learning_rate": 0.003, "loss": 4.0741, "step": 12724 }, { "epoch": 0.12725, "grad_norm": 1.2989919649637907, "learning_rate": 0.003, "loss": 4.0803, "step": 12725 }, { "epoch": 0.12726, "grad_norm": 1.138944369119727, "learning_rate": 0.003, "loss": 4.0602, "step": 12726 }, { "epoch": 0.12727, "grad_norm": 1.1531095634219064, "learning_rate": 0.003, "loss": 4.0662, "step": 12727 }, { "epoch": 0.12728, "grad_norm": 1.2911306924099242, "learning_rate": 0.003, "loss": 4.0845, "step": 12728 }, { "epoch": 0.12729, "grad_norm": 1.0833222993254583, "learning_rate": 0.003, "loss": 4.0805, "step": 12729 }, { "epoch": 0.1273, "grad_norm": 1.4432297163898637, "learning_rate": 0.003, "loss": 4.0772, "step": 12730 }, { "epoch": 0.12731, "grad_norm": 0.9822744898425566, "learning_rate": 0.003, "loss": 4.0452, "step": 12731 }, { "epoch": 0.12732, "grad_norm": 1.1457318578638307, "learning_rate": 0.003, "loss": 4.05, "step": 12732 }, { "epoch": 0.12733, "grad_norm": 1.3808953292491515, "learning_rate": 0.003, "loss": 4.0888, "step": 12733 }, { "epoch": 0.12734, "grad_norm": 1.2028234266526825, "learning_rate": 0.003, "loss": 4.0667, "step": 12734 }, { "epoch": 0.12735, "grad_norm": 0.9884754749991274, "learning_rate": 0.003, "loss": 4.0664, "step": 12735 }, { "epoch": 0.12736, "grad_norm": 1.2084403474756913, "learning_rate": 0.003, "loss": 4.0863, "step": 12736 }, { "epoch": 0.12737, "grad_norm": 1.0040944184692593, "learning_rate": 0.003, "loss": 4.0794, "step": 12737 }, { "epoch": 0.12738, "grad_norm": 1.2811809584095792, "learning_rate": 0.003, "loss": 4.0858, "step": 12738 }, { "epoch": 0.12739, "grad_norm": 1.0230493648988197, "learning_rate": 0.003, "loss": 4.0761, "step": 12739 }, { "epoch": 0.1274, "grad_norm": 1.4005792410516809, "learning_rate": 0.003, "loss": 4.065, "step": 12740 }, { "epoch": 0.12741, "grad_norm": 1.1006908025651114, "learning_rate": 0.003, "loss": 4.0782, "step": 12741 }, { "epoch": 0.12742, "grad_norm": 1.2419269613043473, "learning_rate": 0.003, "loss": 4.0551, "step": 12742 }, { "epoch": 0.12743, "grad_norm": 1.056815602258476, "learning_rate": 0.003, "loss": 4.0577, "step": 12743 }, { "epoch": 0.12744, "grad_norm": 1.0891562763539673, "learning_rate": 0.003, "loss": 4.0581, "step": 12744 }, { "epoch": 0.12745, "grad_norm": 1.2863861344562593, "learning_rate": 0.003, "loss": 4.0864, "step": 12745 }, { "epoch": 0.12746, "grad_norm": 1.1750782483639854, "learning_rate": 0.003, "loss": 4.0931, "step": 12746 }, { "epoch": 0.12747, "grad_norm": 1.0367537228548345, "learning_rate": 0.003, "loss": 4.0826, "step": 12747 }, { "epoch": 0.12748, "grad_norm": 1.0320366689138416, "learning_rate": 0.003, "loss": 4.0836, "step": 12748 }, { "epoch": 0.12749, "grad_norm": 1.3090270380465214, "learning_rate": 0.003, "loss": 4.0775, "step": 12749 }, { "epoch": 0.1275, "grad_norm": 0.9623742918617262, "learning_rate": 0.003, "loss": 4.079, "step": 12750 }, { "epoch": 0.12751, "grad_norm": 1.4293745653200822, "learning_rate": 0.003, "loss": 4.1016, "step": 12751 }, { "epoch": 0.12752, "grad_norm": 0.9835449704041017, "learning_rate": 0.003, "loss": 4.078, "step": 12752 }, { "epoch": 0.12753, "grad_norm": 1.1922279569087715, "learning_rate": 0.003, "loss": 4.0718, "step": 12753 }, { "epoch": 0.12754, "grad_norm": 1.2462405781423709, "learning_rate": 0.003, "loss": 4.105, "step": 12754 }, { "epoch": 0.12755, "grad_norm": 1.2474327637473313, "learning_rate": 0.003, "loss": 4.0711, "step": 12755 }, { "epoch": 0.12756, "grad_norm": 1.2417319598905454, "learning_rate": 0.003, "loss": 4.0882, "step": 12756 }, { "epoch": 0.12757, "grad_norm": 1.0348760164518949, "learning_rate": 0.003, "loss": 4.0736, "step": 12757 }, { "epoch": 0.12758, "grad_norm": 1.2812274353318962, "learning_rate": 0.003, "loss": 4.0697, "step": 12758 }, { "epoch": 0.12759, "grad_norm": 1.259984768621498, "learning_rate": 0.003, "loss": 4.069, "step": 12759 }, { "epoch": 0.1276, "grad_norm": 1.4602602011892685, "learning_rate": 0.003, "loss": 4.0886, "step": 12760 }, { "epoch": 0.12761, "grad_norm": 0.9397846125753508, "learning_rate": 0.003, "loss": 4.0675, "step": 12761 }, { "epoch": 0.12762, "grad_norm": 1.186600256474887, "learning_rate": 0.003, "loss": 4.0453, "step": 12762 }, { "epoch": 0.12763, "grad_norm": 1.073769792838965, "learning_rate": 0.003, "loss": 4.0367, "step": 12763 }, { "epoch": 0.12764, "grad_norm": 1.275205021998926, "learning_rate": 0.003, "loss": 4.1016, "step": 12764 }, { "epoch": 0.12765, "grad_norm": 1.1742251437935147, "learning_rate": 0.003, "loss": 4.0786, "step": 12765 }, { "epoch": 0.12766, "grad_norm": 1.3329399824900785, "learning_rate": 0.003, "loss": 4.0862, "step": 12766 }, { "epoch": 0.12767, "grad_norm": 1.0791021912483265, "learning_rate": 0.003, "loss": 4.0705, "step": 12767 }, { "epoch": 0.12768, "grad_norm": 1.4595623801311226, "learning_rate": 0.003, "loss": 4.0975, "step": 12768 }, { "epoch": 0.12769, "grad_norm": 1.1736033398782468, "learning_rate": 0.003, "loss": 4.0722, "step": 12769 }, { "epoch": 0.1277, "grad_norm": 1.1062460564248988, "learning_rate": 0.003, "loss": 4.0335, "step": 12770 }, { "epoch": 0.12771, "grad_norm": 1.041613507025483, "learning_rate": 0.003, "loss": 4.0673, "step": 12771 }, { "epoch": 0.12772, "grad_norm": 1.3741621938740265, "learning_rate": 0.003, "loss": 4.0717, "step": 12772 }, { "epoch": 0.12773, "grad_norm": 0.9883536324275963, "learning_rate": 0.003, "loss": 4.0459, "step": 12773 }, { "epoch": 0.12774, "grad_norm": 1.409100278621397, "learning_rate": 0.003, "loss": 4.0309, "step": 12774 }, { "epoch": 0.12775, "grad_norm": 0.9143556891914499, "learning_rate": 0.003, "loss": 4.0458, "step": 12775 }, { "epoch": 0.12776, "grad_norm": 1.1985009871078336, "learning_rate": 0.003, "loss": 4.067, "step": 12776 }, { "epoch": 0.12777, "grad_norm": 1.196130551271495, "learning_rate": 0.003, "loss": 4.0676, "step": 12777 }, { "epoch": 0.12778, "grad_norm": 1.0073557140282612, "learning_rate": 0.003, "loss": 4.0688, "step": 12778 }, { "epoch": 0.12779, "grad_norm": 1.405770620626207, "learning_rate": 0.003, "loss": 4.0973, "step": 12779 }, { "epoch": 0.1278, "grad_norm": 1.0804071655753458, "learning_rate": 0.003, "loss": 4.0723, "step": 12780 }, { "epoch": 0.12781, "grad_norm": 1.0997196151059192, "learning_rate": 0.003, "loss": 4.0516, "step": 12781 }, { "epoch": 0.12782, "grad_norm": 1.1494608347279651, "learning_rate": 0.003, "loss": 4.086, "step": 12782 }, { "epoch": 0.12783, "grad_norm": 1.0321321835252084, "learning_rate": 0.003, "loss": 4.0752, "step": 12783 }, { "epoch": 0.12784, "grad_norm": 1.1384291428999396, "learning_rate": 0.003, "loss": 4.0705, "step": 12784 }, { "epoch": 0.12785, "grad_norm": 1.288495970579294, "learning_rate": 0.003, "loss": 4.049, "step": 12785 }, { "epoch": 0.12786, "grad_norm": 1.363284159380646, "learning_rate": 0.003, "loss": 4.0772, "step": 12786 }, { "epoch": 0.12787, "grad_norm": 1.3524790224997603, "learning_rate": 0.003, "loss": 4.0798, "step": 12787 }, { "epoch": 0.12788, "grad_norm": 1.0276998746728125, "learning_rate": 0.003, "loss": 4.0683, "step": 12788 }, { "epoch": 0.12789, "grad_norm": 1.318856149931721, "learning_rate": 0.003, "loss": 4.0651, "step": 12789 }, { "epoch": 0.1279, "grad_norm": 1.2069682807094462, "learning_rate": 0.003, "loss": 4.0927, "step": 12790 }, { "epoch": 0.12791, "grad_norm": 1.1184456166514853, "learning_rate": 0.003, "loss": 4.0733, "step": 12791 }, { "epoch": 0.12792, "grad_norm": 1.1067357652558798, "learning_rate": 0.003, "loss": 4.0726, "step": 12792 }, { "epoch": 0.12793, "grad_norm": 1.4670707203875561, "learning_rate": 0.003, "loss": 4.0772, "step": 12793 }, { "epoch": 0.12794, "grad_norm": 0.8727727545953691, "learning_rate": 0.003, "loss": 4.0651, "step": 12794 }, { "epoch": 0.12795, "grad_norm": 0.9961457976446118, "learning_rate": 0.003, "loss": 4.0666, "step": 12795 }, { "epoch": 0.12796, "grad_norm": 1.254421658016817, "learning_rate": 0.003, "loss": 4.0513, "step": 12796 }, { "epoch": 0.12797, "grad_norm": 1.24442927219961, "learning_rate": 0.003, "loss": 4.0524, "step": 12797 }, { "epoch": 0.12798, "grad_norm": 1.3697473984511896, "learning_rate": 0.003, "loss": 4.0495, "step": 12798 }, { "epoch": 0.12799, "grad_norm": 0.9884266532498901, "learning_rate": 0.003, "loss": 4.081, "step": 12799 }, { "epoch": 0.128, "grad_norm": 1.1109350516665366, "learning_rate": 0.003, "loss": 4.0511, "step": 12800 }, { "epoch": 0.12801, "grad_norm": 1.3380461520808835, "learning_rate": 0.003, "loss": 4.0741, "step": 12801 }, { "epoch": 0.12802, "grad_norm": 0.9970239959885036, "learning_rate": 0.003, "loss": 4.1064, "step": 12802 }, { "epoch": 0.12803, "grad_norm": 1.3235631353916777, "learning_rate": 0.003, "loss": 4.0571, "step": 12803 }, { "epoch": 0.12804, "grad_norm": 0.8483000630193144, "learning_rate": 0.003, "loss": 4.0668, "step": 12804 }, { "epoch": 0.12805, "grad_norm": 1.1064973927009718, "learning_rate": 0.003, "loss": 4.0781, "step": 12805 }, { "epoch": 0.12806, "grad_norm": 1.2732880584031754, "learning_rate": 0.003, "loss": 4.0759, "step": 12806 }, { "epoch": 0.12807, "grad_norm": 1.2812021949362313, "learning_rate": 0.003, "loss": 4.0896, "step": 12807 }, { "epoch": 0.12808, "grad_norm": 1.1966594964800683, "learning_rate": 0.003, "loss": 4.0728, "step": 12808 }, { "epoch": 0.12809, "grad_norm": 1.16669330019199, "learning_rate": 0.003, "loss": 4.0614, "step": 12809 }, { "epoch": 0.1281, "grad_norm": 1.2485541233691577, "learning_rate": 0.003, "loss": 4.068, "step": 12810 }, { "epoch": 0.12811, "grad_norm": 1.3432781017291722, "learning_rate": 0.003, "loss": 4.0765, "step": 12811 }, { "epoch": 0.12812, "grad_norm": 1.1129839438989915, "learning_rate": 0.003, "loss": 4.0756, "step": 12812 }, { "epoch": 0.12813, "grad_norm": 1.0945004724599967, "learning_rate": 0.003, "loss": 4.0774, "step": 12813 }, { "epoch": 0.12814, "grad_norm": 1.3305159561639528, "learning_rate": 0.003, "loss": 4.0886, "step": 12814 }, { "epoch": 0.12815, "grad_norm": 0.9993067828031007, "learning_rate": 0.003, "loss": 4.0386, "step": 12815 }, { "epoch": 0.12816, "grad_norm": 1.1632692943418896, "learning_rate": 0.003, "loss": 4.0614, "step": 12816 }, { "epoch": 0.12817, "grad_norm": 1.0089713823630224, "learning_rate": 0.003, "loss": 4.064, "step": 12817 }, { "epoch": 0.12818, "grad_norm": 1.467243264855757, "learning_rate": 0.003, "loss": 4.0603, "step": 12818 }, { "epoch": 0.12819, "grad_norm": 1.1672674505745604, "learning_rate": 0.003, "loss": 4.08, "step": 12819 }, { "epoch": 0.1282, "grad_norm": 1.0441202538190926, "learning_rate": 0.003, "loss": 4.0376, "step": 12820 }, { "epoch": 0.12821, "grad_norm": 1.3899115952334344, "learning_rate": 0.003, "loss": 4.0751, "step": 12821 }, { "epoch": 0.12822, "grad_norm": 1.1284507672330641, "learning_rate": 0.003, "loss": 4.1015, "step": 12822 }, { "epoch": 0.12823, "grad_norm": 1.5657702324506915, "learning_rate": 0.003, "loss": 4.0989, "step": 12823 }, { "epoch": 0.12824, "grad_norm": 0.8546350417815805, "learning_rate": 0.003, "loss": 4.0748, "step": 12824 }, { "epoch": 0.12825, "grad_norm": 1.0705539272876288, "learning_rate": 0.003, "loss": 4.0788, "step": 12825 }, { "epoch": 0.12826, "grad_norm": 1.4048058479185337, "learning_rate": 0.003, "loss": 4.0863, "step": 12826 }, { "epoch": 0.12827, "grad_norm": 1.2121907285812725, "learning_rate": 0.003, "loss": 4.076, "step": 12827 }, { "epoch": 0.12828, "grad_norm": 1.2009641061982406, "learning_rate": 0.003, "loss": 4.1033, "step": 12828 }, { "epoch": 0.12829, "grad_norm": 1.1627578672696073, "learning_rate": 0.003, "loss": 4.0586, "step": 12829 }, { "epoch": 0.1283, "grad_norm": 1.0991143035535262, "learning_rate": 0.003, "loss": 4.0778, "step": 12830 }, { "epoch": 0.12831, "grad_norm": 1.115238829257928, "learning_rate": 0.003, "loss": 4.0561, "step": 12831 }, { "epoch": 0.12832, "grad_norm": 1.1877303958826508, "learning_rate": 0.003, "loss": 4.0684, "step": 12832 }, { "epoch": 0.12833, "grad_norm": 1.2737008439070627, "learning_rate": 0.003, "loss": 4.1135, "step": 12833 }, { "epoch": 0.12834, "grad_norm": 0.9863971781458649, "learning_rate": 0.003, "loss": 4.0604, "step": 12834 }, { "epoch": 0.12835, "grad_norm": 1.217349751553351, "learning_rate": 0.003, "loss": 4.0619, "step": 12835 }, { "epoch": 0.12836, "grad_norm": 0.948173851737352, "learning_rate": 0.003, "loss": 4.0627, "step": 12836 }, { "epoch": 0.12837, "grad_norm": 1.3598810604323968, "learning_rate": 0.003, "loss": 4.1047, "step": 12837 }, { "epoch": 0.12838, "grad_norm": 1.138930648132535, "learning_rate": 0.003, "loss": 4.0734, "step": 12838 }, { "epoch": 0.12839, "grad_norm": 1.0006638436626087, "learning_rate": 0.003, "loss": 4.0581, "step": 12839 }, { "epoch": 0.1284, "grad_norm": 1.3073913436699431, "learning_rate": 0.003, "loss": 4.0647, "step": 12840 }, { "epoch": 0.12841, "grad_norm": 1.3275864218910867, "learning_rate": 0.003, "loss": 4.041, "step": 12841 }, { "epoch": 0.12842, "grad_norm": 1.1493120075388126, "learning_rate": 0.003, "loss": 4.0711, "step": 12842 }, { "epoch": 0.12843, "grad_norm": 1.2017852678334084, "learning_rate": 0.003, "loss": 4.0749, "step": 12843 }, { "epoch": 0.12844, "grad_norm": 1.057771641595308, "learning_rate": 0.003, "loss": 4.0671, "step": 12844 }, { "epoch": 0.12845, "grad_norm": 1.3808357093166381, "learning_rate": 0.003, "loss": 4.0609, "step": 12845 }, { "epoch": 0.12846, "grad_norm": 1.1008203357281807, "learning_rate": 0.003, "loss": 4.068, "step": 12846 }, { "epoch": 0.12847, "grad_norm": 1.3152387191500043, "learning_rate": 0.003, "loss": 4.0868, "step": 12847 }, { "epoch": 0.12848, "grad_norm": 1.3343889831602418, "learning_rate": 0.003, "loss": 4.0884, "step": 12848 }, { "epoch": 0.12849, "grad_norm": 1.1294171865863156, "learning_rate": 0.003, "loss": 4.0413, "step": 12849 }, { "epoch": 0.1285, "grad_norm": 1.1586576580299826, "learning_rate": 0.003, "loss": 4.0539, "step": 12850 }, { "epoch": 0.12851, "grad_norm": 1.1504795266991583, "learning_rate": 0.003, "loss": 4.0635, "step": 12851 }, { "epoch": 0.12852, "grad_norm": 1.1865678989137314, "learning_rate": 0.003, "loss": 4.0688, "step": 12852 }, { "epoch": 0.12853, "grad_norm": 1.2176250634540564, "learning_rate": 0.003, "loss": 4.0735, "step": 12853 }, { "epoch": 0.12854, "grad_norm": 1.0883767807314606, "learning_rate": 0.003, "loss": 4.0893, "step": 12854 }, { "epoch": 0.12855, "grad_norm": 1.0889257086818527, "learning_rate": 0.003, "loss": 4.0377, "step": 12855 }, { "epoch": 0.12856, "grad_norm": 1.2076962938762292, "learning_rate": 0.003, "loss": 4.0589, "step": 12856 }, { "epoch": 0.12857, "grad_norm": 1.0516746139614412, "learning_rate": 0.003, "loss": 4.0475, "step": 12857 }, { "epoch": 0.12858, "grad_norm": 1.4350109548800867, "learning_rate": 0.003, "loss": 4.093, "step": 12858 }, { "epoch": 0.12859, "grad_norm": 0.9591249658291462, "learning_rate": 0.003, "loss": 4.0495, "step": 12859 }, { "epoch": 0.1286, "grad_norm": 1.5555432130245146, "learning_rate": 0.003, "loss": 4.0811, "step": 12860 }, { "epoch": 0.12861, "grad_norm": 1.1690526647953225, "learning_rate": 0.003, "loss": 4.0837, "step": 12861 }, { "epoch": 0.12862, "grad_norm": 1.2929818536041606, "learning_rate": 0.003, "loss": 4.0701, "step": 12862 }, { "epoch": 0.12863, "grad_norm": 1.0584395412571626, "learning_rate": 0.003, "loss": 4.09, "step": 12863 }, { "epoch": 0.12864, "grad_norm": 1.1070018585167076, "learning_rate": 0.003, "loss": 4.0805, "step": 12864 }, { "epoch": 0.12865, "grad_norm": 1.1516901687191021, "learning_rate": 0.003, "loss": 4.0644, "step": 12865 }, { "epoch": 0.12866, "grad_norm": 1.171564827066514, "learning_rate": 0.003, "loss": 4.0842, "step": 12866 }, { "epoch": 0.12867, "grad_norm": 1.0159836410101402, "learning_rate": 0.003, "loss": 4.0525, "step": 12867 }, { "epoch": 0.12868, "grad_norm": 1.2280536990300008, "learning_rate": 0.003, "loss": 4.0714, "step": 12868 }, { "epoch": 0.12869, "grad_norm": 1.4569571047621208, "learning_rate": 0.003, "loss": 4.0493, "step": 12869 }, { "epoch": 0.1287, "grad_norm": 1.2578352058524276, "learning_rate": 0.003, "loss": 4.0801, "step": 12870 }, { "epoch": 0.12871, "grad_norm": 1.159664751084178, "learning_rate": 0.003, "loss": 4.1094, "step": 12871 }, { "epoch": 0.12872, "grad_norm": 1.04319532085096, "learning_rate": 0.003, "loss": 4.0594, "step": 12872 }, { "epoch": 0.12873, "grad_norm": 1.0788402168566222, "learning_rate": 0.003, "loss": 4.0615, "step": 12873 }, { "epoch": 0.12874, "grad_norm": 1.068336320457033, "learning_rate": 0.003, "loss": 4.0792, "step": 12874 }, { "epoch": 0.12875, "grad_norm": 1.3543454544421982, "learning_rate": 0.003, "loss": 4.0906, "step": 12875 }, { "epoch": 0.12876, "grad_norm": 1.3794616530476864, "learning_rate": 0.003, "loss": 4.0712, "step": 12876 }, { "epoch": 0.12877, "grad_norm": 1.044127299764906, "learning_rate": 0.003, "loss": 4.0593, "step": 12877 }, { "epoch": 0.12878, "grad_norm": 1.0555971510389346, "learning_rate": 0.003, "loss": 4.0784, "step": 12878 }, { "epoch": 0.12879, "grad_norm": 1.049314355759846, "learning_rate": 0.003, "loss": 4.0412, "step": 12879 }, { "epoch": 0.1288, "grad_norm": 1.3798266253275924, "learning_rate": 0.003, "loss": 4.0553, "step": 12880 }, { "epoch": 0.12881, "grad_norm": 0.9047032299045279, "learning_rate": 0.003, "loss": 4.0743, "step": 12881 }, { "epoch": 0.12882, "grad_norm": 1.237667655709752, "learning_rate": 0.003, "loss": 4.0672, "step": 12882 }, { "epoch": 0.12883, "grad_norm": 1.0513962810178894, "learning_rate": 0.003, "loss": 4.0754, "step": 12883 }, { "epoch": 0.12884, "grad_norm": 1.1654304128154607, "learning_rate": 0.003, "loss": 4.0782, "step": 12884 }, { "epoch": 0.12885, "grad_norm": 1.223231574979754, "learning_rate": 0.003, "loss": 4.0544, "step": 12885 }, { "epoch": 0.12886, "grad_norm": 1.3370106029329338, "learning_rate": 0.003, "loss": 4.0735, "step": 12886 }, { "epoch": 0.12887, "grad_norm": 1.2360740240021264, "learning_rate": 0.003, "loss": 4.0929, "step": 12887 }, { "epoch": 0.12888, "grad_norm": 1.2319839349362194, "learning_rate": 0.003, "loss": 4.0652, "step": 12888 }, { "epoch": 0.12889, "grad_norm": 1.3882447262004995, "learning_rate": 0.003, "loss": 4.0962, "step": 12889 }, { "epoch": 0.1289, "grad_norm": 0.9886714804166923, "learning_rate": 0.003, "loss": 4.0708, "step": 12890 }, { "epoch": 0.12891, "grad_norm": 1.277581788852025, "learning_rate": 0.003, "loss": 4.0576, "step": 12891 }, { "epoch": 0.12892, "grad_norm": 0.9658180522667765, "learning_rate": 0.003, "loss": 4.0622, "step": 12892 }, { "epoch": 0.12893, "grad_norm": 1.071268157458069, "learning_rate": 0.003, "loss": 4.1138, "step": 12893 }, { "epoch": 0.12894, "grad_norm": 1.352899796848344, "learning_rate": 0.003, "loss": 4.0859, "step": 12894 }, { "epoch": 0.12895, "grad_norm": 1.0521509767290662, "learning_rate": 0.003, "loss": 4.0837, "step": 12895 }, { "epoch": 0.12896, "grad_norm": 1.5389168201105246, "learning_rate": 0.003, "loss": 4.0671, "step": 12896 }, { "epoch": 0.12897, "grad_norm": 1.0686414383777203, "learning_rate": 0.003, "loss": 4.0271, "step": 12897 }, { "epoch": 0.12898, "grad_norm": 1.2169757972025348, "learning_rate": 0.003, "loss": 4.0679, "step": 12898 }, { "epoch": 0.12899, "grad_norm": 1.308036689238183, "learning_rate": 0.003, "loss": 4.1004, "step": 12899 }, { "epoch": 0.129, "grad_norm": 1.2307611483435221, "learning_rate": 0.003, "loss": 4.0652, "step": 12900 }, { "epoch": 0.12901, "grad_norm": 1.0337449423782072, "learning_rate": 0.003, "loss": 4.0744, "step": 12901 }, { "epoch": 0.12902, "grad_norm": 1.2459216760216396, "learning_rate": 0.003, "loss": 4.0752, "step": 12902 }, { "epoch": 0.12903, "grad_norm": 1.3137652546230592, "learning_rate": 0.003, "loss": 4.0968, "step": 12903 }, { "epoch": 0.12904, "grad_norm": 1.2224418199363158, "learning_rate": 0.003, "loss": 4.0402, "step": 12904 }, { "epoch": 0.12905, "grad_norm": 1.0773334695030383, "learning_rate": 0.003, "loss": 4.0581, "step": 12905 }, { "epoch": 0.12906, "grad_norm": 1.1445121047428575, "learning_rate": 0.003, "loss": 4.0627, "step": 12906 }, { "epoch": 0.12907, "grad_norm": 1.1651006592214215, "learning_rate": 0.003, "loss": 4.0802, "step": 12907 }, { "epoch": 0.12908, "grad_norm": 1.235356870985851, "learning_rate": 0.003, "loss": 4.038, "step": 12908 }, { "epoch": 0.12909, "grad_norm": 1.0979832086059393, "learning_rate": 0.003, "loss": 4.0762, "step": 12909 }, { "epoch": 0.1291, "grad_norm": 1.099055647655412, "learning_rate": 0.003, "loss": 4.0648, "step": 12910 }, { "epoch": 0.12911, "grad_norm": 1.3345734515203431, "learning_rate": 0.003, "loss": 4.0749, "step": 12911 }, { "epoch": 0.12912, "grad_norm": 1.2055307153529464, "learning_rate": 0.003, "loss": 4.0451, "step": 12912 }, { "epoch": 0.12913, "grad_norm": 0.9378615333877254, "learning_rate": 0.003, "loss": 4.0718, "step": 12913 }, { "epoch": 0.12914, "grad_norm": 1.2759949524150114, "learning_rate": 0.003, "loss": 4.0829, "step": 12914 }, { "epoch": 0.12915, "grad_norm": 1.4032020521873743, "learning_rate": 0.003, "loss": 4.0505, "step": 12915 }, { "epoch": 0.12916, "grad_norm": 1.4120384567652084, "learning_rate": 0.003, "loss": 4.0931, "step": 12916 }, { "epoch": 0.12917, "grad_norm": 1.1960338866006393, "learning_rate": 0.003, "loss": 4.0817, "step": 12917 }, { "epoch": 0.12918, "grad_norm": 1.0388195462652365, "learning_rate": 0.003, "loss": 4.0941, "step": 12918 }, { "epoch": 0.12919, "grad_norm": 1.303206159930564, "learning_rate": 0.003, "loss": 4.0773, "step": 12919 }, { "epoch": 0.1292, "grad_norm": 1.2077059437923325, "learning_rate": 0.003, "loss": 4.0696, "step": 12920 }, { "epoch": 0.12921, "grad_norm": 1.0952104661367275, "learning_rate": 0.003, "loss": 4.0732, "step": 12921 }, { "epoch": 0.12922, "grad_norm": 1.344907820003931, "learning_rate": 0.003, "loss": 4.0762, "step": 12922 }, { "epoch": 0.12923, "grad_norm": 1.155637762860284, "learning_rate": 0.003, "loss": 4.0264, "step": 12923 }, { "epoch": 0.12924, "grad_norm": 1.3436189844739663, "learning_rate": 0.003, "loss": 4.0892, "step": 12924 }, { "epoch": 0.12925, "grad_norm": 1.0353235582247116, "learning_rate": 0.003, "loss": 4.066, "step": 12925 }, { "epoch": 0.12926, "grad_norm": 1.2229537570230653, "learning_rate": 0.003, "loss": 4.0733, "step": 12926 }, { "epoch": 0.12927, "grad_norm": 1.0059810201492685, "learning_rate": 0.003, "loss": 4.0625, "step": 12927 }, { "epoch": 0.12928, "grad_norm": 1.3482657563536515, "learning_rate": 0.003, "loss": 4.0779, "step": 12928 }, { "epoch": 0.12929, "grad_norm": 1.0034702187552473, "learning_rate": 0.003, "loss": 4.0504, "step": 12929 }, { "epoch": 0.1293, "grad_norm": 1.237325267234494, "learning_rate": 0.003, "loss": 4.0768, "step": 12930 }, { "epoch": 0.12931, "grad_norm": 1.217620116486049, "learning_rate": 0.003, "loss": 4.0708, "step": 12931 }, { "epoch": 0.12932, "grad_norm": 1.0866115666571554, "learning_rate": 0.003, "loss": 4.0624, "step": 12932 }, { "epoch": 0.12933, "grad_norm": 1.196918851275972, "learning_rate": 0.003, "loss": 4.0801, "step": 12933 }, { "epoch": 0.12934, "grad_norm": 1.118890903625114, "learning_rate": 0.003, "loss": 4.0788, "step": 12934 }, { "epoch": 0.12935, "grad_norm": 1.2573756682627455, "learning_rate": 0.003, "loss": 4.0709, "step": 12935 }, { "epoch": 0.12936, "grad_norm": 0.9737075399054111, "learning_rate": 0.003, "loss": 4.0786, "step": 12936 }, { "epoch": 0.12937, "grad_norm": 1.213197985342146, "learning_rate": 0.003, "loss": 4.0507, "step": 12937 }, { "epoch": 0.12938, "grad_norm": 1.2727782592637276, "learning_rate": 0.003, "loss": 4.0875, "step": 12938 }, { "epoch": 0.12939, "grad_norm": 1.0975732780763439, "learning_rate": 0.003, "loss": 4.0512, "step": 12939 }, { "epoch": 0.1294, "grad_norm": 1.2000156851268204, "learning_rate": 0.003, "loss": 4.069, "step": 12940 }, { "epoch": 0.12941, "grad_norm": 1.1312255444762633, "learning_rate": 0.003, "loss": 4.0673, "step": 12941 }, { "epoch": 0.12942, "grad_norm": 1.2019313243678091, "learning_rate": 0.003, "loss": 4.0597, "step": 12942 }, { "epoch": 0.12943, "grad_norm": 1.1117822536192914, "learning_rate": 0.003, "loss": 4.0647, "step": 12943 }, { "epoch": 0.12944, "grad_norm": 1.1989374864817226, "learning_rate": 0.003, "loss": 4.0768, "step": 12944 }, { "epoch": 0.12945, "grad_norm": 1.15357346799443, "learning_rate": 0.003, "loss": 4.0826, "step": 12945 }, { "epoch": 0.12946, "grad_norm": 1.3358287172664458, "learning_rate": 0.003, "loss": 4.0962, "step": 12946 }, { "epoch": 0.12947, "grad_norm": 1.2171654215607066, "learning_rate": 0.003, "loss": 4.0644, "step": 12947 }, { "epoch": 0.12948, "grad_norm": 1.1992249878175494, "learning_rate": 0.003, "loss": 4.0516, "step": 12948 }, { "epoch": 0.12949, "grad_norm": 1.2569764949972773, "learning_rate": 0.003, "loss": 4.0664, "step": 12949 }, { "epoch": 0.1295, "grad_norm": 1.2619205520569914, "learning_rate": 0.003, "loss": 4.0663, "step": 12950 }, { "epoch": 0.12951, "grad_norm": 1.1394939514320945, "learning_rate": 0.003, "loss": 4.0626, "step": 12951 }, { "epoch": 0.12952, "grad_norm": 1.155842315189784, "learning_rate": 0.003, "loss": 4.0574, "step": 12952 }, { "epoch": 0.12953, "grad_norm": 1.1707677310635063, "learning_rate": 0.003, "loss": 4.0793, "step": 12953 }, { "epoch": 0.12954, "grad_norm": 1.0933718841461246, "learning_rate": 0.003, "loss": 4.0426, "step": 12954 }, { "epoch": 0.12955, "grad_norm": 1.1535013083902386, "learning_rate": 0.003, "loss": 4.0822, "step": 12955 }, { "epoch": 0.12956, "grad_norm": 1.2597637233126602, "learning_rate": 0.003, "loss": 4.0587, "step": 12956 }, { "epoch": 0.12957, "grad_norm": 1.2129477062167615, "learning_rate": 0.003, "loss": 4.0584, "step": 12957 }, { "epoch": 0.12958, "grad_norm": 1.0259513574852872, "learning_rate": 0.003, "loss": 4.0631, "step": 12958 }, { "epoch": 0.12959, "grad_norm": 1.3098981298516374, "learning_rate": 0.003, "loss": 4.0916, "step": 12959 }, { "epoch": 0.1296, "grad_norm": 1.0705475071474728, "learning_rate": 0.003, "loss": 4.0613, "step": 12960 }, { "epoch": 0.12961, "grad_norm": 1.2082787123272298, "learning_rate": 0.003, "loss": 4.0954, "step": 12961 }, { "epoch": 0.12962, "grad_norm": 0.9752946105506849, "learning_rate": 0.003, "loss": 4.0749, "step": 12962 }, { "epoch": 0.12963, "grad_norm": 1.427170678310923, "learning_rate": 0.003, "loss": 4.0813, "step": 12963 }, { "epoch": 0.12964, "grad_norm": 0.9801046548980611, "learning_rate": 0.003, "loss": 4.064, "step": 12964 }, { "epoch": 0.12965, "grad_norm": 1.4189584638318544, "learning_rate": 0.003, "loss": 4.0544, "step": 12965 }, { "epoch": 0.12966, "grad_norm": 0.9692416621823341, "learning_rate": 0.003, "loss": 4.0637, "step": 12966 }, { "epoch": 0.12967, "grad_norm": 1.2840081091300894, "learning_rate": 0.003, "loss": 4.0245, "step": 12967 }, { "epoch": 0.12968, "grad_norm": 1.0219463558374213, "learning_rate": 0.003, "loss": 4.0822, "step": 12968 }, { "epoch": 0.12969, "grad_norm": 1.3432349293723256, "learning_rate": 0.003, "loss": 4.0656, "step": 12969 }, { "epoch": 0.1297, "grad_norm": 0.9788760612331872, "learning_rate": 0.003, "loss": 4.0462, "step": 12970 }, { "epoch": 0.12971, "grad_norm": 1.1150769051237421, "learning_rate": 0.003, "loss": 4.0672, "step": 12971 }, { "epoch": 0.12972, "grad_norm": 1.5177506434977075, "learning_rate": 0.003, "loss": 4.083, "step": 12972 }, { "epoch": 0.12973, "grad_norm": 1.0099978634477083, "learning_rate": 0.003, "loss": 4.1072, "step": 12973 }, { "epoch": 0.12974, "grad_norm": 1.1922998133660232, "learning_rate": 0.003, "loss": 4.0727, "step": 12974 }, { "epoch": 0.12975, "grad_norm": 1.2232258790717931, "learning_rate": 0.003, "loss": 4.0739, "step": 12975 }, { "epoch": 0.12976, "grad_norm": 1.101427794606135, "learning_rate": 0.003, "loss": 4.0913, "step": 12976 }, { "epoch": 0.12977, "grad_norm": 1.304895856234629, "learning_rate": 0.003, "loss": 4.0977, "step": 12977 }, { "epoch": 0.12978, "grad_norm": 1.2628731587815114, "learning_rate": 0.003, "loss": 4.063, "step": 12978 }, { "epoch": 0.12979, "grad_norm": 1.6235453994304816, "learning_rate": 0.003, "loss": 4.0619, "step": 12979 }, { "epoch": 0.1298, "grad_norm": 1.2048853555128978, "learning_rate": 0.003, "loss": 4.0563, "step": 12980 }, { "epoch": 0.12981, "grad_norm": 1.2375469153589247, "learning_rate": 0.003, "loss": 4.0807, "step": 12981 }, { "epoch": 0.12982, "grad_norm": 1.0542275359671747, "learning_rate": 0.003, "loss": 4.1048, "step": 12982 }, { "epoch": 0.12983, "grad_norm": 1.2239658195260248, "learning_rate": 0.003, "loss": 4.0507, "step": 12983 }, { "epoch": 0.12984, "grad_norm": 1.4114467551974932, "learning_rate": 0.003, "loss": 4.0839, "step": 12984 }, { "epoch": 0.12985, "grad_norm": 1.0277566905987274, "learning_rate": 0.003, "loss": 4.0748, "step": 12985 }, { "epoch": 0.12986, "grad_norm": 1.2165699490367765, "learning_rate": 0.003, "loss": 4.0572, "step": 12986 }, { "epoch": 0.12987, "grad_norm": 1.0726865407475268, "learning_rate": 0.003, "loss": 4.0757, "step": 12987 }, { "epoch": 0.12988, "grad_norm": 1.3509187805816179, "learning_rate": 0.003, "loss": 4.0782, "step": 12988 }, { "epoch": 0.12989, "grad_norm": 0.9369681453881235, "learning_rate": 0.003, "loss": 4.0672, "step": 12989 }, { "epoch": 0.1299, "grad_norm": 1.0227497888593235, "learning_rate": 0.003, "loss": 4.0767, "step": 12990 }, { "epoch": 0.12991, "grad_norm": 1.2447174630530253, "learning_rate": 0.003, "loss": 4.0575, "step": 12991 }, { "epoch": 0.12992, "grad_norm": 1.2955141038836735, "learning_rate": 0.003, "loss": 4.0837, "step": 12992 }, { "epoch": 0.12993, "grad_norm": 1.0410231425868315, "learning_rate": 0.003, "loss": 4.051, "step": 12993 }, { "epoch": 0.12994, "grad_norm": 1.2099567963574025, "learning_rate": 0.003, "loss": 4.0551, "step": 12994 }, { "epoch": 0.12995, "grad_norm": 1.154147808313667, "learning_rate": 0.003, "loss": 4.0472, "step": 12995 }, { "epoch": 0.12996, "grad_norm": 1.2765151728742004, "learning_rate": 0.003, "loss": 4.0414, "step": 12996 }, { "epoch": 0.12997, "grad_norm": 0.9682271673833246, "learning_rate": 0.003, "loss": 4.0744, "step": 12997 }, { "epoch": 0.12998, "grad_norm": 1.1146858882428259, "learning_rate": 0.003, "loss": 4.0618, "step": 12998 }, { "epoch": 0.12999, "grad_norm": 1.1852155319301692, "learning_rate": 0.003, "loss": 4.0449, "step": 12999 }, { "epoch": 0.13, "grad_norm": 1.0976339654389153, "learning_rate": 0.003, "loss": 4.0612, "step": 13000 }, { "epoch": 0.13001, "grad_norm": 1.1719840711361706, "learning_rate": 0.003, "loss": 4.0685, "step": 13001 }, { "epoch": 0.13002, "grad_norm": 1.2395325556126913, "learning_rate": 0.003, "loss": 4.051, "step": 13002 }, { "epoch": 0.13003, "grad_norm": 1.2171854677821783, "learning_rate": 0.003, "loss": 4.0721, "step": 13003 }, { "epoch": 0.13004, "grad_norm": 1.2637731269041053, "learning_rate": 0.003, "loss": 4.0604, "step": 13004 }, { "epoch": 0.13005, "grad_norm": 1.1017358129235937, "learning_rate": 0.003, "loss": 4.059, "step": 13005 }, { "epoch": 0.13006, "grad_norm": 1.2406184371706632, "learning_rate": 0.003, "loss": 4.0401, "step": 13006 }, { "epoch": 0.13007, "grad_norm": 1.1825684407984873, "learning_rate": 0.003, "loss": 4.085, "step": 13007 }, { "epoch": 0.13008, "grad_norm": 1.3488824485332012, "learning_rate": 0.003, "loss": 4.0901, "step": 13008 }, { "epoch": 0.13009, "grad_norm": 1.2234427376099881, "learning_rate": 0.003, "loss": 4.0727, "step": 13009 }, { "epoch": 0.1301, "grad_norm": 1.08591708301902, "learning_rate": 0.003, "loss": 4.0689, "step": 13010 }, { "epoch": 0.13011, "grad_norm": 1.3407717742473235, "learning_rate": 0.003, "loss": 4.0769, "step": 13011 }, { "epoch": 0.13012, "grad_norm": 1.2155971810878954, "learning_rate": 0.003, "loss": 4.0907, "step": 13012 }, { "epoch": 0.13013, "grad_norm": 1.0879271257155427, "learning_rate": 0.003, "loss": 4.0906, "step": 13013 }, { "epoch": 0.13014, "grad_norm": 1.1851693890243746, "learning_rate": 0.003, "loss": 4.0696, "step": 13014 }, { "epoch": 0.13015, "grad_norm": 1.199017353126804, "learning_rate": 0.003, "loss": 4.0783, "step": 13015 }, { "epoch": 0.13016, "grad_norm": 1.1776438621405436, "learning_rate": 0.003, "loss": 4.0518, "step": 13016 }, { "epoch": 0.13017, "grad_norm": 0.9992607017416343, "learning_rate": 0.003, "loss": 4.0794, "step": 13017 }, { "epoch": 0.13018, "grad_norm": 1.3721850549640682, "learning_rate": 0.003, "loss": 4.0609, "step": 13018 }, { "epoch": 0.13019, "grad_norm": 1.1592138262428602, "learning_rate": 0.003, "loss": 4.0866, "step": 13019 }, { "epoch": 0.1302, "grad_norm": 1.3905239030194738, "learning_rate": 0.003, "loss": 4.0704, "step": 13020 }, { "epoch": 0.13021, "grad_norm": 1.0426943295167048, "learning_rate": 0.003, "loss": 4.067, "step": 13021 }, { "epoch": 0.13022, "grad_norm": 1.337761810483605, "learning_rate": 0.003, "loss": 4.0905, "step": 13022 }, { "epoch": 0.13023, "grad_norm": 1.3428463476352528, "learning_rate": 0.003, "loss": 4.0881, "step": 13023 }, { "epoch": 0.13024, "grad_norm": 1.0184689002960445, "learning_rate": 0.003, "loss": 4.0238, "step": 13024 }, { "epoch": 0.13025, "grad_norm": 1.567374367389213, "learning_rate": 0.003, "loss": 4.081, "step": 13025 }, { "epoch": 0.13026, "grad_norm": 0.9682298144928962, "learning_rate": 0.003, "loss": 4.0855, "step": 13026 }, { "epoch": 0.13027, "grad_norm": 1.2481611156368244, "learning_rate": 0.003, "loss": 4.0596, "step": 13027 }, { "epoch": 0.13028, "grad_norm": 1.1118337049905826, "learning_rate": 0.003, "loss": 4.1026, "step": 13028 }, { "epoch": 0.13029, "grad_norm": 1.0678002232745039, "learning_rate": 0.003, "loss": 4.0541, "step": 13029 }, { "epoch": 0.1303, "grad_norm": 1.3028948228940662, "learning_rate": 0.003, "loss": 4.1037, "step": 13030 }, { "epoch": 0.13031, "grad_norm": 1.1345815629618317, "learning_rate": 0.003, "loss": 4.0614, "step": 13031 }, { "epoch": 0.13032, "grad_norm": 1.2764060035685951, "learning_rate": 0.003, "loss": 4.0695, "step": 13032 }, { "epoch": 0.13033, "grad_norm": 1.0880223457154283, "learning_rate": 0.003, "loss": 4.0626, "step": 13033 }, { "epoch": 0.13034, "grad_norm": 1.2493974225524753, "learning_rate": 0.003, "loss": 4.0989, "step": 13034 }, { "epoch": 0.13035, "grad_norm": 1.192649822548278, "learning_rate": 0.003, "loss": 4.0901, "step": 13035 }, { "epoch": 0.13036, "grad_norm": 1.1621841236091006, "learning_rate": 0.003, "loss": 4.0674, "step": 13036 }, { "epoch": 0.13037, "grad_norm": 1.1555538619464063, "learning_rate": 0.003, "loss": 4.0657, "step": 13037 }, { "epoch": 0.13038, "grad_norm": 1.1040581259118034, "learning_rate": 0.003, "loss": 4.0903, "step": 13038 }, { "epoch": 0.13039, "grad_norm": 1.2449707400894137, "learning_rate": 0.003, "loss": 4.0942, "step": 13039 }, { "epoch": 0.1304, "grad_norm": 1.1186950779568379, "learning_rate": 0.003, "loss": 4.0578, "step": 13040 }, { "epoch": 0.13041, "grad_norm": 1.2095185684633825, "learning_rate": 0.003, "loss": 4.0755, "step": 13041 }, { "epoch": 0.13042, "grad_norm": 1.0596708796928151, "learning_rate": 0.003, "loss": 4.0775, "step": 13042 }, { "epoch": 0.13043, "grad_norm": 1.327168194888388, "learning_rate": 0.003, "loss": 4.0774, "step": 13043 }, { "epoch": 0.13044, "grad_norm": 1.0088888183450886, "learning_rate": 0.003, "loss": 4.0431, "step": 13044 }, { "epoch": 0.13045, "grad_norm": 1.373055801348955, "learning_rate": 0.003, "loss": 4.0648, "step": 13045 }, { "epoch": 0.13046, "grad_norm": 1.039158109370196, "learning_rate": 0.003, "loss": 4.0742, "step": 13046 }, { "epoch": 0.13047, "grad_norm": 1.254604689950872, "learning_rate": 0.003, "loss": 4.0723, "step": 13047 }, { "epoch": 0.13048, "grad_norm": 1.303497464207566, "learning_rate": 0.003, "loss": 4.036, "step": 13048 }, { "epoch": 0.13049, "grad_norm": 1.1142815098414798, "learning_rate": 0.003, "loss": 4.0352, "step": 13049 }, { "epoch": 0.1305, "grad_norm": 1.0922039155513472, "learning_rate": 0.003, "loss": 4.0336, "step": 13050 }, { "epoch": 0.13051, "grad_norm": 1.2383441895824563, "learning_rate": 0.003, "loss": 4.0932, "step": 13051 }, { "epoch": 0.13052, "grad_norm": 0.871361235889519, "learning_rate": 0.003, "loss": 4.0487, "step": 13052 }, { "epoch": 0.13053, "grad_norm": 1.20706424684798, "learning_rate": 0.003, "loss": 4.0441, "step": 13053 }, { "epoch": 0.13054, "grad_norm": 1.1275445937801618, "learning_rate": 0.003, "loss": 4.0722, "step": 13054 }, { "epoch": 0.13055, "grad_norm": 1.3058414663434768, "learning_rate": 0.003, "loss": 4.0512, "step": 13055 }, { "epoch": 0.13056, "grad_norm": 1.1606410345404876, "learning_rate": 0.003, "loss": 4.0473, "step": 13056 }, { "epoch": 0.13057, "grad_norm": 1.1159909777718091, "learning_rate": 0.003, "loss": 4.0697, "step": 13057 }, { "epoch": 0.13058, "grad_norm": 1.3532554025376324, "learning_rate": 0.003, "loss": 4.0794, "step": 13058 }, { "epoch": 0.13059, "grad_norm": 0.9671037931541167, "learning_rate": 0.003, "loss": 4.0935, "step": 13059 }, { "epoch": 0.1306, "grad_norm": 1.3555157417557786, "learning_rate": 0.003, "loss": 4.077, "step": 13060 }, { "epoch": 0.13061, "grad_norm": 1.1339550840622015, "learning_rate": 0.003, "loss": 4.0459, "step": 13061 }, { "epoch": 0.13062, "grad_norm": 1.3155454591800106, "learning_rate": 0.003, "loss": 4.0671, "step": 13062 }, { "epoch": 0.13063, "grad_norm": 0.9497245187691132, "learning_rate": 0.003, "loss": 4.0321, "step": 13063 }, { "epoch": 0.13064, "grad_norm": 1.1112703813027285, "learning_rate": 0.003, "loss": 4.1008, "step": 13064 }, { "epoch": 0.13065, "grad_norm": 1.3094402100725666, "learning_rate": 0.003, "loss": 4.0661, "step": 13065 }, { "epoch": 0.13066, "grad_norm": 1.207723488602815, "learning_rate": 0.003, "loss": 4.0694, "step": 13066 }, { "epoch": 0.13067, "grad_norm": 1.138535241928582, "learning_rate": 0.003, "loss": 4.1175, "step": 13067 }, { "epoch": 0.13068, "grad_norm": 1.125570761311579, "learning_rate": 0.003, "loss": 4.0508, "step": 13068 }, { "epoch": 0.13069, "grad_norm": 1.1380573095984057, "learning_rate": 0.003, "loss": 4.0716, "step": 13069 }, { "epoch": 0.1307, "grad_norm": 1.3993482428247788, "learning_rate": 0.003, "loss": 4.0829, "step": 13070 }, { "epoch": 0.13071, "grad_norm": 1.0460800989467518, "learning_rate": 0.003, "loss": 4.0575, "step": 13071 }, { "epoch": 0.13072, "grad_norm": 1.3210821096496814, "learning_rate": 0.003, "loss": 4.0624, "step": 13072 }, { "epoch": 0.13073, "grad_norm": 1.2782443653088875, "learning_rate": 0.003, "loss": 4.0628, "step": 13073 }, { "epoch": 0.13074, "grad_norm": 1.0264864205435482, "learning_rate": 0.003, "loss": 4.0515, "step": 13074 }, { "epoch": 0.13075, "grad_norm": 1.4539548194948315, "learning_rate": 0.003, "loss": 4.0893, "step": 13075 }, { "epoch": 0.13076, "grad_norm": 1.0571972127694433, "learning_rate": 0.003, "loss": 4.0821, "step": 13076 }, { "epoch": 0.13077, "grad_norm": 1.4052143426558386, "learning_rate": 0.003, "loss": 4.0755, "step": 13077 }, { "epoch": 0.13078, "grad_norm": 1.0168093488205356, "learning_rate": 0.003, "loss": 4.0602, "step": 13078 }, { "epoch": 0.13079, "grad_norm": 1.0372755501789055, "learning_rate": 0.003, "loss": 4.038, "step": 13079 }, { "epoch": 0.1308, "grad_norm": 1.1618249789461472, "learning_rate": 0.003, "loss": 4.0676, "step": 13080 }, { "epoch": 0.13081, "grad_norm": 1.0624452352346405, "learning_rate": 0.003, "loss": 4.0754, "step": 13081 }, { "epoch": 0.13082, "grad_norm": 1.531261452855752, "learning_rate": 0.003, "loss": 4.0583, "step": 13082 }, { "epoch": 0.13083, "grad_norm": 0.9416551787171488, "learning_rate": 0.003, "loss": 4.0634, "step": 13083 }, { "epoch": 0.13084, "grad_norm": 1.0731902948525347, "learning_rate": 0.003, "loss": 4.0762, "step": 13084 }, { "epoch": 0.13085, "grad_norm": 1.3053143377615222, "learning_rate": 0.003, "loss": 4.0854, "step": 13085 }, { "epoch": 0.13086, "grad_norm": 1.268815155355026, "learning_rate": 0.003, "loss": 4.0551, "step": 13086 }, { "epoch": 0.13087, "grad_norm": 1.0761772837233665, "learning_rate": 0.003, "loss": 4.0561, "step": 13087 }, { "epoch": 0.13088, "grad_norm": 1.274969036571922, "learning_rate": 0.003, "loss": 4.0794, "step": 13088 }, { "epoch": 0.13089, "grad_norm": 1.324164809138917, "learning_rate": 0.003, "loss": 4.0745, "step": 13089 }, { "epoch": 0.1309, "grad_norm": 1.016357532614868, "learning_rate": 0.003, "loss": 4.0467, "step": 13090 }, { "epoch": 0.13091, "grad_norm": 1.2212894941990593, "learning_rate": 0.003, "loss": 4.0732, "step": 13091 }, { "epoch": 0.13092, "grad_norm": 1.1327866585335682, "learning_rate": 0.003, "loss": 4.0754, "step": 13092 }, { "epoch": 0.13093, "grad_norm": 1.2464559565085225, "learning_rate": 0.003, "loss": 4.0819, "step": 13093 }, { "epoch": 0.13094, "grad_norm": 1.1703567866448106, "learning_rate": 0.003, "loss": 4.0832, "step": 13094 }, { "epoch": 0.13095, "grad_norm": 1.1254181586581946, "learning_rate": 0.003, "loss": 4.0702, "step": 13095 }, { "epoch": 0.13096, "grad_norm": 1.4898858598531313, "learning_rate": 0.003, "loss": 4.0801, "step": 13096 }, { "epoch": 0.13097, "grad_norm": 1.0993970060752314, "learning_rate": 0.003, "loss": 4.097, "step": 13097 }, { "epoch": 0.13098, "grad_norm": 1.2468958183914973, "learning_rate": 0.003, "loss": 4.08, "step": 13098 }, { "epoch": 0.13099, "grad_norm": 1.1356271282152646, "learning_rate": 0.003, "loss": 4.08, "step": 13099 }, { "epoch": 0.131, "grad_norm": 1.164359375650946, "learning_rate": 0.003, "loss": 4.0742, "step": 13100 }, { "epoch": 0.13101, "grad_norm": 1.2026038979244935, "learning_rate": 0.003, "loss": 4.0711, "step": 13101 }, { "epoch": 0.13102, "grad_norm": 1.239000477800128, "learning_rate": 0.003, "loss": 4.0614, "step": 13102 }, { "epoch": 0.13103, "grad_norm": 1.0921048185721713, "learning_rate": 0.003, "loss": 4.0855, "step": 13103 }, { "epoch": 0.13104, "grad_norm": 1.1912772826367217, "learning_rate": 0.003, "loss": 4.0724, "step": 13104 }, { "epoch": 0.13105, "grad_norm": 1.2496469503092782, "learning_rate": 0.003, "loss": 4.0614, "step": 13105 }, { "epoch": 0.13106, "grad_norm": 1.1050279696637815, "learning_rate": 0.003, "loss": 4.0486, "step": 13106 }, { "epoch": 0.13107, "grad_norm": 1.0460931373647986, "learning_rate": 0.003, "loss": 4.0561, "step": 13107 }, { "epoch": 0.13108, "grad_norm": 1.15041158164609, "learning_rate": 0.003, "loss": 4.05, "step": 13108 }, { "epoch": 0.13109, "grad_norm": 1.202937178450559, "learning_rate": 0.003, "loss": 4.0647, "step": 13109 }, { "epoch": 0.1311, "grad_norm": 1.3292885960129865, "learning_rate": 0.003, "loss": 4.0741, "step": 13110 }, { "epoch": 0.13111, "grad_norm": 1.037369122357519, "learning_rate": 0.003, "loss": 4.0494, "step": 13111 }, { "epoch": 0.13112, "grad_norm": 1.2547884682511412, "learning_rate": 0.003, "loss": 4.0667, "step": 13112 }, { "epoch": 0.13113, "grad_norm": 1.1019106389585718, "learning_rate": 0.003, "loss": 4.1, "step": 13113 }, { "epoch": 0.13114, "grad_norm": 1.203936727523638, "learning_rate": 0.003, "loss": 4.038, "step": 13114 }, { "epoch": 0.13115, "grad_norm": 1.162395306383368, "learning_rate": 0.003, "loss": 4.0794, "step": 13115 }, { "epoch": 0.13116, "grad_norm": 1.2454704246661696, "learning_rate": 0.003, "loss": 4.0742, "step": 13116 }, { "epoch": 0.13117, "grad_norm": 1.1536892102972363, "learning_rate": 0.003, "loss": 4.0614, "step": 13117 }, { "epoch": 0.13118, "grad_norm": 1.236290395505284, "learning_rate": 0.003, "loss": 4.0912, "step": 13118 }, { "epoch": 0.13119, "grad_norm": 1.1220643138722115, "learning_rate": 0.003, "loss": 4.0502, "step": 13119 }, { "epoch": 0.1312, "grad_norm": 1.2630931565665404, "learning_rate": 0.003, "loss": 4.0848, "step": 13120 }, { "epoch": 0.13121, "grad_norm": 1.0232700484349586, "learning_rate": 0.003, "loss": 4.0592, "step": 13121 }, { "epoch": 0.13122, "grad_norm": 1.3329053565683004, "learning_rate": 0.003, "loss": 4.0661, "step": 13122 }, { "epoch": 0.13123, "grad_norm": 1.1981636851001758, "learning_rate": 0.003, "loss": 4.0617, "step": 13123 }, { "epoch": 0.13124, "grad_norm": 1.122876647482152, "learning_rate": 0.003, "loss": 4.0571, "step": 13124 }, { "epoch": 0.13125, "grad_norm": 1.0983788389474274, "learning_rate": 0.003, "loss": 4.0637, "step": 13125 }, { "epoch": 0.13126, "grad_norm": 1.189172427531487, "learning_rate": 0.003, "loss": 4.0729, "step": 13126 }, { "epoch": 0.13127, "grad_norm": 1.2610958209663654, "learning_rate": 0.003, "loss": 4.0517, "step": 13127 }, { "epoch": 0.13128, "grad_norm": 1.177518134913952, "learning_rate": 0.003, "loss": 4.0359, "step": 13128 }, { "epoch": 0.13129, "grad_norm": 1.2070916925420854, "learning_rate": 0.003, "loss": 4.0642, "step": 13129 }, { "epoch": 0.1313, "grad_norm": 1.021589725438157, "learning_rate": 0.003, "loss": 4.0751, "step": 13130 }, { "epoch": 0.13131, "grad_norm": 1.416014144166172, "learning_rate": 0.003, "loss": 4.0736, "step": 13131 }, { "epoch": 0.13132, "grad_norm": 1.1578426046991976, "learning_rate": 0.003, "loss": 4.079, "step": 13132 }, { "epoch": 0.13133, "grad_norm": 1.1432614142886794, "learning_rate": 0.003, "loss": 4.0955, "step": 13133 }, { "epoch": 0.13134, "grad_norm": 1.1401084280990508, "learning_rate": 0.003, "loss": 4.0508, "step": 13134 }, { "epoch": 0.13135, "grad_norm": 1.2440658002569782, "learning_rate": 0.003, "loss": 4.06, "step": 13135 }, { "epoch": 0.13136, "grad_norm": 1.052418668704034, "learning_rate": 0.003, "loss": 4.0685, "step": 13136 }, { "epoch": 0.13137, "grad_norm": 1.303485320067905, "learning_rate": 0.003, "loss": 4.0518, "step": 13137 }, { "epoch": 0.13138, "grad_norm": 1.2187999745892937, "learning_rate": 0.003, "loss": 4.049, "step": 13138 }, { "epoch": 0.13139, "grad_norm": 1.2058533722431775, "learning_rate": 0.003, "loss": 4.0745, "step": 13139 }, { "epoch": 0.1314, "grad_norm": 1.0408781113472696, "learning_rate": 0.003, "loss": 4.0758, "step": 13140 }, { "epoch": 0.13141, "grad_norm": 1.1411681778399358, "learning_rate": 0.003, "loss": 4.0721, "step": 13141 }, { "epoch": 0.13142, "grad_norm": 1.2319320845876192, "learning_rate": 0.003, "loss": 4.0962, "step": 13142 }, { "epoch": 0.13143, "grad_norm": 1.085798298116153, "learning_rate": 0.003, "loss": 4.0541, "step": 13143 }, { "epoch": 0.13144, "grad_norm": 1.3545254009295953, "learning_rate": 0.003, "loss": 4.0741, "step": 13144 }, { "epoch": 0.13145, "grad_norm": 0.9143757032556452, "learning_rate": 0.003, "loss": 4.0387, "step": 13145 }, { "epoch": 0.13146, "grad_norm": 1.4720284030388406, "learning_rate": 0.003, "loss": 4.0587, "step": 13146 }, { "epoch": 0.13147, "grad_norm": 1.1731835884741624, "learning_rate": 0.003, "loss": 4.0824, "step": 13147 }, { "epoch": 0.13148, "grad_norm": 1.6066667376482742, "learning_rate": 0.003, "loss": 4.099, "step": 13148 }, { "epoch": 0.13149, "grad_norm": 0.9410727696498126, "learning_rate": 0.003, "loss": 4.0642, "step": 13149 }, { "epoch": 0.1315, "grad_norm": 1.1983778287162503, "learning_rate": 0.003, "loss": 4.0742, "step": 13150 }, { "epoch": 0.13151, "grad_norm": 1.0307962254410346, "learning_rate": 0.003, "loss": 4.0688, "step": 13151 }, { "epoch": 0.13152, "grad_norm": 1.43694201892184, "learning_rate": 0.003, "loss": 4.0785, "step": 13152 }, { "epoch": 0.13153, "grad_norm": 1.0677993579040252, "learning_rate": 0.003, "loss": 4.0591, "step": 13153 }, { "epoch": 0.13154, "grad_norm": 1.268461574067591, "learning_rate": 0.003, "loss": 4.0415, "step": 13154 }, { "epoch": 0.13155, "grad_norm": 1.3489756824284524, "learning_rate": 0.003, "loss": 4.0677, "step": 13155 }, { "epoch": 0.13156, "grad_norm": 1.2563872234790574, "learning_rate": 0.003, "loss": 4.0664, "step": 13156 }, { "epoch": 0.13157, "grad_norm": 1.1144314498494128, "learning_rate": 0.003, "loss": 4.1076, "step": 13157 }, { "epoch": 0.13158, "grad_norm": 1.1621371252015023, "learning_rate": 0.003, "loss": 4.0641, "step": 13158 }, { "epoch": 0.13159, "grad_norm": 1.1953764462629082, "learning_rate": 0.003, "loss": 4.059, "step": 13159 }, { "epoch": 0.1316, "grad_norm": 1.3534961536701053, "learning_rate": 0.003, "loss": 4.0342, "step": 13160 }, { "epoch": 0.13161, "grad_norm": 1.242098722764693, "learning_rate": 0.003, "loss": 4.0622, "step": 13161 }, { "epoch": 0.13162, "grad_norm": 1.1870910181990675, "learning_rate": 0.003, "loss": 4.0763, "step": 13162 }, { "epoch": 0.13163, "grad_norm": 1.2544164663661443, "learning_rate": 0.003, "loss": 4.1104, "step": 13163 }, { "epoch": 0.13164, "grad_norm": 0.9468907101214062, "learning_rate": 0.003, "loss": 4.0477, "step": 13164 }, { "epoch": 0.13165, "grad_norm": 1.147766455124515, "learning_rate": 0.003, "loss": 4.0818, "step": 13165 }, { "epoch": 0.13166, "grad_norm": 1.173114027596755, "learning_rate": 0.003, "loss": 4.0701, "step": 13166 }, { "epoch": 0.13167, "grad_norm": 1.3007314649436021, "learning_rate": 0.003, "loss": 4.0882, "step": 13167 }, { "epoch": 0.13168, "grad_norm": 1.2323954828024348, "learning_rate": 0.003, "loss": 4.0908, "step": 13168 }, { "epoch": 0.13169, "grad_norm": 1.1484314440408978, "learning_rate": 0.003, "loss": 4.0435, "step": 13169 }, { "epoch": 0.1317, "grad_norm": 1.2450412020280535, "learning_rate": 0.003, "loss": 4.0761, "step": 13170 }, { "epoch": 0.13171, "grad_norm": 1.0654001535175357, "learning_rate": 0.003, "loss": 4.0971, "step": 13171 }, { "epoch": 0.13172, "grad_norm": 1.2522723227148644, "learning_rate": 0.003, "loss": 4.0608, "step": 13172 }, { "epoch": 0.13173, "grad_norm": 1.2519635803570286, "learning_rate": 0.003, "loss": 4.0423, "step": 13173 }, { "epoch": 0.13174, "grad_norm": 1.3000796495416944, "learning_rate": 0.003, "loss": 4.0962, "step": 13174 }, { "epoch": 0.13175, "grad_norm": 1.1218033849114384, "learning_rate": 0.003, "loss": 4.0293, "step": 13175 }, { "epoch": 0.13176, "grad_norm": 1.40399595321406, "learning_rate": 0.003, "loss": 4.0663, "step": 13176 }, { "epoch": 0.13177, "grad_norm": 1.034644584531117, "learning_rate": 0.003, "loss": 4.0735, "step": 13177 }, { "epoch": 0.13178, "grad_norm": 1.2692271154725556, "learning_rate": 0.003, "loss": 4.0647, "step": 13178 }, { "epoch": 0.13179, "grad_norm": 1.0060631692783792, "learning_rate": 0.003, "loss": 4.0662, "step": 13179 }, { "epoch": 0.1318, "grad_norm": 1.2995122597023647, "learning_rate": 0.003, "loss": 4.0595, "step": 13180 }, { "epoch": 0.13181, "grad_norm": 1.0182110416638797, "learning_rate": 0.003, "loss": 4.0423, "step": 13181 }, { "epoch": 0.13182, "grad_norm": 1.3172261033849268, "learning_rate": 0.003, "loss": 4.0516, "step": 13182 }, { "epoch": 0.13183, "grad_norm": 1.1460498770288394, "learning_rate": 0.003, "loss": 4.0708, "step": 13183 }, { "epoch": 0.13184, "grad_norm": 1.4216084949568217, "learning_rate": 0.003, "loss": 4.0775, "step": 13184 }, { "epoch": 0.13185, "grad_norm": 1.147631551120462, "learning_rate": 0.003, "loss": 4.0887, "step": 13185 }, { "epoch": 0.13186, "grad_norm": 1.1662040796241357, "learning_rate": 0.003, "loss": 4.0846, "step": 13186 }, { "epoch": 0.13187, "grad_norm": 1.0624324039471014, "learning_rate": 0.003, "loss": 4.0438, "step": 13187 }, { "epoch": 0.13188, "grad_norm": 1.4689423539802247, "learning_rate": 0.003, "loss": 4.0757, "step": 13188 }, { "epoch": 0.13189, "grad_norm": 1.2125059062651504, "learning_rate": 0.003, "loss": 4.0416, "step": 13189 }, { "epoch": 0.1319, "grad_norm": 1.294366777943818, "learning_rate": 0.003, "loss": 4.1072, "step": 13190 }, { "epoch": 0.13191, "grad_norm": 1.2504632282545622, "learning_rate": 0.003, "loss": 4.1012, "step": 13191 }, { "epoch": 0.13192, "grad_norm": 1.2500300419282089, "learning_rate": 0.003, "loss": 4.0431, "step": 13192 }, { "epoch": 0.13193, "grad_norm": 1.188141854015912, "learning_rate": 0.003, "loss": 4.0252, "step": 13193 }, { "epoch": 0.13194, "grad_norm": 1.1643243566980073, "learning_rate": 0.003, "loss": 4.0702, "step": 13194 }, { "epoch": 0.13195, "grad_norm": 1.2181816479686556, "learning_rate": 0.003, "loss": 4.0592, "step": 13195 }, { "epoch": 0.13196, "grad_norm": 1.3827877725312507, "learning_rate": 0.003, "loss": 4.03, "step": 13196 }, { "epoch": 0.13197, "grad_norm": 1.1016038971829898, "learning_rate": 0.003, "loss": 4.0792, "step": 13197 }, { "epoch": 0.13198, "grad_norm": 1.1855740411996358, "learning_rate": 0.003, "loss": 4.0627, "step": 13198 }, { "epoch": 0.13199, "grad_norm": 1.2171685480656886, "learning_rate": 0.003, "loss": 4.0716, "step": 13199 }, { "epoch": 0.132, "grad_norm": 1.1741359782355432, "learning_rate": 0.003, "loss": 4.0779, "step": 13200 }, { "epoch": 0.13201, "grad_norm": 1.0440591109644022, "learning_rate": 0.003, "loss": 4.0392, "step": 13201 }, { "epoch": 0.13202, "grad_norm": 1.274471138754622, "learning_rate": 0.003, "loss": 4.0807, "step": 13202 }, { "epoch": 0.13203, "grad_norm": 1.0243281102287765, "learning_rate": 0.003, "loss": 4.073, "step": 13203 }, { "epoch": 0.13204, "grad_norm": 1.2860425256286774, "learning_rate": 0.003, "loss": 4.0652, "step": 13204 }, { "epoch": 0.13205, "grad_norm": 1.1348776907669282, "learning_rate": 0.003, "loss": 4.0789, "step": 13205 }, { "epoch": 0.13206, "grad_norm": 1.1730696996692072, "learning_rate": 0.003, "loss": 4.058, "step": 13206 }, { "epoch": 0.13207, "grad_norm": 1.1865654537115866, "learning_rate": 0.003, "loss": 4.0703, "step": 13207 }, { "epoch": 0.13208, "grad_norm": 1.171465135259456, "learning_rate": 0.003, "loss": 4.0409, "step": 13208 }, { "epoch": 0.13209, "grad_norm": 1.167747396466667, "learning_rate": 0.003, "loss": 4.072, "step": 13209 }, { "epoch": 0.1321, "grad_norm": 1.3028876053170206, "learning_rate": 0.003, "loss": 4.1076, "step": 13210 }, { "epoch": 0.13211, "grad_norm": 1.0152159430030967, "learning_rate": 0.003, "loss": 4.0847, "step": 13211 }, { "epoch": 0.13212, "grad_norm": 1.3183786319105706, "learning_rate": 0.003, "loss": 4.0714, "step": 13212 }, { "epoch": 0.13213, "grad_norm": 1.0998785599181296, "learning_rate": 0.003, "loss": 4.083, "step": 13213 }, { "epoch": 0.13214, "grad_norm": 1.3318000426642578, "learning_rate": 0.003, "loss": 4.0598, "step": 13214 }, { "epoch": 0.13215, "grad_norm": 1.0581639289197025, "learning_rate": 0.003, "loss": 4.0707, "step": 13215 }, { "epoch": 0.13216, "grad_norm": 1.2798620619411303, "learning_rate": 0.003, "loss": 4.0707, "step": 13216 }, { "epoch": 0.13217, "grad_norm": 1.1679712848756147, "learning_rate": 0.003, "loss": 4.0596, "step": 13217 }, { "epoch": 0.13218, "grad_norm": 1.0639680271963092, "learning_rate": 0.003, "loss": 4.07, "step": 13218 }, { "epoch": 0.13219, "grad_norm": 1.080567226953388, "learning_rate": 0.003, "loss": 4.116, "step": 13219 }, { "epoch": 0.1322, "grad_norm": 1.126118010598661, "learning_rate": 0.003, "loss": 4.1015, "step": 13220 }, { "epoch": 0.13221, "grad_norm": 1.3291393029914746, "learning_rate": 0.003, "loss": 4.0787, "step": 13221 }, { "epoch": 0.13222, "grad_norm": 1.1108425950802159, "learning_rate": 0.003, "loss": 4.0651, "step": 13222 }, { "epoch": 0.13223, "grad_norm": 1.2382546339313611, "learning_rate": 0.003, "loss": 4.0692, "step": 13223 }, { "epoch": 0.13224, "grad_norm": 1.1371082689527137, "learning_rate": 0.003, "loss": 4.0489, "step": 13224 }, { "epoch": 0.13225, "grad_norm": 1.3903131092309453, "learning_rate": 0.003, "loss": 4.0457, "step": 13225 }, { "epoch": 0.13226, "grad_norm": 1.1597561076975533, "learning_rate": 0.003, "loss": 4.0704, "step": 13226 }, { "epoch": 0.13227, "grad_norm": 1.042148751678186, "learning_rate": 0.003, "loss": 4.0832, "step": 13227 }, { "epoch": 0.13228, "grad_norm": 1.2992142654925891, "learning_rate": 0.003, "loss": 4.0695, "step": 13228 }, { "epoch": 0.13229, "grad_norm": 0.8828692668277073, "learning_rate": 0.003, "loss": 4.0724, "step": 13229 }, { "epoch": 0.1323, "grad_norm": 1.3203625380909305, "learning_rate": 0.003, "loss": 4.0577, "step": 13230 }, { "epoch": 0.13231, "grad_norm": 1.0816150876660933, "learning_rate": 0.003, "loss": 4.0467, "step": 13231 }, { "epoch": 0.13232, "grad_norm": 1.0725194297395733, "learning_rate": 0.003, "loss": 4.067, "step": 13232 }, { "epoch": 0.13233, "grad_norm": 1.1413962335441905, "learning_rate": 0.003, "loss": 4.0401, "step": 13233 }, { "epoch": 0.13234, "grad_norm": 1.2134453387384976, "learning_rate": 0.003, "loss": 4.0536, "step": 13234 }, { "epoch": 0.13235, "grad_norm": 1.247755762674242, "learning_rate": 0.003, "loss": 4.0647, "step": 13235 }, { "epoch": 0.13236, "grad_norm": 1.0189650430636072, "learning_rate": 0.003, "loss": 4.0338, "step": 13236 }, { "epoch": 0.13237, "grad_norm": 1.4715555159366667, "learning_rate": 0.003, "loss": 4.0798, "step": 13237 }, { "epoch": 0.13238, "grad_norm": 1.0375021199817416, "learning_rate": 0.003, "loss": 4.0656, "step": 13238 }, { "epoch": 0.13239, "grad_norm": 1.4209193503839939, "learning_rate": 0.003, "loss": 4.0776, "step": 13239 }, { "epoch": 0.1324, "grad_norm": 1.1020368731645678, "learning_rate": 0.003, "loss": 4.0745, "step": 13240 }, { "epoch": 0.13241, "grad_norm": 1.324787650406456, "learning_rate": 0.003, "loss": 4.0651, "step": 13241 }, { "epoch": 0.13242, "grad_norm": 1.2556118004452719, "learning_rate": 0.003, "loss": 4.0961, "step": 13242 }, { "epoch": 0.13243, "grad_norm": 1.0985048995218638, "learning_rate": 0.003, "loss": 4.0601, "step": 13243 }, { "epoch": 0.13244, "grad_norm": 1.2957851512955199, "learning_rate": 0.003, "loss": 4.0428, "step": 13244 }, { "epoch": 0.13245, "grad_norm": 1.0214523845328929, "learning_rate": 0.003, "loss": 4.0237, "step": 13245 }, { "epoch": 0.13246, "grad_norm": 1.1986691549585764, "learning_rate": 0.003, "loss": 4.0454, "step": 13246 }, { "epoch": 0.13247, "grad_norm": 1.4288685128409502, "learning_rate": 0.003, "loss": 4.0869, "step": 13247 }, { "epoch": 0.13248, "grad_norm": 1.0236478696036977, "learning_rate": 0.003, "loss": 4.0848, "step": 13248 }, { "epoch": 0.13249, "grad_norm": 1.292811480699596, "learning_rate": 0.003, "loss": 4.0502, "step": 13249 }, { "epoch": 0.1325, "grad_norm": 0.9843660778914611, "learning_rate": 0.003, "loss": 4.0629, "step": 13250 }, { "epoch": 0.13251, "grad_norm": 1.2233385186478687, "learning_rate": 0.003, "loss": 4.0591, "step": 13251 }, { "epoch": 0.13252, "grad_norm": 1.1244979257071788, "learning_rate": 0.003, "loss": 4.0774, "step": 13252 }, { "epoch": 0.13253, "grad_norm": 1.3177861997563658, "learning_rate": 0.003, "loss": 4.034, "step": 13253 }, { "epoch": 0.13254, "grad_norm": 1.0244642998401052, "learning_rate": 0.003, "loss": 4.0904, "step": 13254 }, { "epoch": 0.13255, "grad_norm": 1.2270378253803171, "learning_rate": 0.003, "loss": 4.0839, "step": 13255 }, { "epoch": 0.13256, "grad_norm": 1.2548833887046018, "learning_rate": 0.003, "loss": 4.0833, "step": 13256 }, { "epoch": 0.13257, "grad_norm": 1.2879983769150376, "learning_rate": 0.003, "loss": 4.0745, "step": 13257 }, { "epoch": 0.13258, "grad_norm": 1.0310468271487272, "learning_rate": 0.003, "loss": 4.0623, "step": 13258 }, { "epoch": 0.13259, "grad_norm": 1.2970101383442554, "learning_rate": 0.003, "loss": 4.092, "step": 13259 }, { "epoch": 0.1326, "grad_norm": 1.2461330317886523, "learning_rate": 0.003, "loss": 4.0769, "step": 13260 }, { "epoch": 0.13261, "grad_norm": 1.1113100025589566, "learning_rate": 0.003, "loss": 4.0697, "step": 13261 }, { "epoch": 0.13262, "grad_norm": 1.3786132650631386, "learning_rate": 0.003, "loss": 4.0607, "step": 13262 }, { "epoch": 0.13263, "grad_norm": 1.0032645606950847, "learning_rate": 0.003, "loss": 4.0536, "step": 13263 }, { "epoch": 0.13264, "grad_norm": 1.3399680984267557, "learning_rate": 0.003, "loss": 4.0746, "step": 13264 }, { "epoch": 0.13265, "grad_norm": 1.017522080857777, "learning_rate": 0.003, "loss": 4.062, "step": 13265 }, { "epoch": 0.13266, "grad_norm": 1.64686624277984, "learning_rate": 0.003, "loss": 4.0834, "step": 13266 }, { "epoch": 0.13267, "grad_norm": 0.8593432409687214, "learning_rate": 0.003, "loss": 4.0728, "step": 13267 }, { "epoch": 0.13268, "grad_norm": 1.1885456279714617, "learning_rate": 0.003, "loss": 4.078, "step": 13268 }, { "epoch": 0.13269, "grad_norm": 1.1230075687986711, "learning_rate": 0.003, "loss": 4.0499, "step": 13269 }, { "epoch": 0.1327, "grad_norm": 1.3107914102384386, "learning_rate": 0.003, "loss": 4.0651, "step": 13270 }, { "epoch": 0.13271, "grad_norm": 1.1820667544165129, "learning_rate": 0.003, "loss": 4.0816, "step": 13271 }, { "epoch": 0.13272, "grad_norm": 1.3241595677319673, "learning_rate": 0.003, "loss": 4.0503, "step": 13272 }, { "epoch": 0.13273, "grad_norm": 1.115483229779608, "learning_rate": 0.003, "loss": 4.0615, "step": 13273 }, { "epoch": 0.13274, "grad_norm": 1.0905548862218508, "learning_rate": 0.003, "loss": 4.0932, "step": 13274 }, { "epoch": 0.13275, "grad_norm": 1.1888232473964362, "learning_rate": 0.003, "loss": 4.0844, "step": 13275 }, { "epoch": 0.13276, "grad_norm": 1.176879818014123, "learning_rate": 0.003, "loss": 4.0547, "step": 13276 }, { "epoch": 0.13277, "grad_norm": 1.1947106153364264, "learning_rate": 0.003, "loss": 4.0753, "step": 13277 }, { "epoch": 0.13278, "grad_norm": 1.341444256794462, "learning_rate": 0.003, "loss": 4.0639, "step": 13278 }, { "epoch": 0.13279, "grad_norm": 0.945105729131143, "learning_rate": 0.003, "loss": 4.0813, "step": 13279 }, { "epoch": 0.1328, "grad_norm": 1.0371380072191345, "learning_rate": 0.003, "loss": 4.0432, "step": 13280 }, { "epoch": 0.13281, "grad_norm": 1.3667997773870617, "learning_rate": 0.003, "loss": 4.0795, "step": 13281 }, { "epoch": 0.13282, "grad_norm": 1.1626626485048392, "learning_rate": 0.003, "loss": 4.0941, "step": 13282 }, { "epoch": 0.13283, "grad_norm": 1.1427129167806978, "learning_rate": 0.003, "loss": 4.0775, "step": 13283 }, { "epoch": 0.13284, "grad_norm": 1.101227744769405, "learning_rate": 0.003, "loss": 4.0683, "step": 13284 }, { "epoch": 0.13285, "grad_norm": 1.149143830642704, "learning_rate": 0.003, "loss": 4.051, "step": 13285 }, { "epoch": 0.13286, "grad_norm": 1.1520063370772313, "learning_rate": 0.003, "loss": 4.0989, "step": 13286 }, { "epoch": 0.13287, "grad_norm": 1.2596745982290913, "learning_rate": 0.003, "loss": 4.0774, "step": 13287 }, { "epoch": 0.13288, "grad_norm": 1.3171987184882712, "learning_rate": 0.003, "loss": 4.0747, "step": 13288 }, { "epoch": 0.13289, "grad_norm": 1.0863020590124632, "learning_rate": 0.003, "loss": 4.0902, "step": 13289 }, { "epoch": 0.1329, "grad_norm": 1.2508202256063077, "learning_rate": 0.003, "loss": 4.0925, "step": 13290 }, { "epoch": 0.13291, "grad_norm": 1.1495342256166174, "learning_rate": 0.003, "loss": 4.0711, "step": 13291 }, { "epoch": 0.13292, "grad_norm": 1.0969413588048633, "learning_rate": 0.003, "loss": 4.0748, "step": 13292 }, { "epoch": 0.13293, "grad_norm": 1.2066481663915818, "learning_rate": 0.003, "loss": 4.0616, "step": 13293 }, { "epoch": 0.13294, "grad_norm": 1.310751051697473, "learning_rate": 0.003, "loss": 4.0755, "step": 13294 }, { "epoch": 0.13295, "grad_norm": 1.1537848448961812, "learning_rate": 0.003, "loss": 4.0765, "step": 13295 }, { "epoch": 0.13296, "grad_norm": 1.2863255538835128, "learning_rate": 0.003, "loss": 4.1097, "step": 13296 }, { "epoch": 0.13297, "grad_norm": 1.1418414868931317, "learning_rate": 0.003, "loss": 4.0676, "step": 13297 }, { "epoch": 0.13298, "grad_norm": 1.4867602662556318, "learning_rate": 0.003, "loss": 4.0639, "step": 13298 }, { "epoch": 0.13299, "grad_norm": 0.9831475534156756, "learning_rate": 0.003, "loss": 4.1155, "step": 13299 }, { "epoch": 0.133, "grad_norm": 1.2855457081631143, "learning_rate": 0.003, "loss": 4.08, "step": 13300 }, { "epoch": 0.13301, "grad_norm": 0.9964339126410291, "learning_rate": 0.003, "loss": 4.0607, "step": 13301 }, { "epoch": 0.13302, "grad_norm": 1.4105926735675485, "learning_rate": 0.003, "loss": 4.0326, "step": 13302 }, { "epoch": 0.13303, "grad_norm": 1.124060560334596, "learning_rate": 0.003, "loss": 4.0824, "step": 13303 }, { "epoch": 0.13304, "grad_norm": 1.1347947355935488, "learning_rate": 0.003, "loss": 4.0724, "step": 13304 }, { "epoch": 0.13305, "grad_norm": 1.1243413098280783, "learning_rate": 0.003, "loss": 4.0972, "step": 13305 }, { "epoch": 0.13306, "grad_norm": 1.2185377831088966, "learning_rate": 0.003, "loss": 4.0735, "step": 13306 }, { "epoch": 0.13307, "grad_norm": 1.1696030730137574, "learning_rate": 0.003, "loss": 4.0532, "step": 13307 }, { "epoch": 0.13308, "grad_norm": 1.2161013629889335, "learning_rate": 0.003, "loss": 4.0641, "step": 13308 }, { "epoch": 0.13309, "grad_norm": 1.4482005394663107, "learning_rate": 0.003, "loss": 4.0508, "step": 13309 }, { "epoch": 0.1331, "grad_norm": 1.2558887517933217, "learning_rate": 0.003, "loss": 4.0632, "step": 13310 }, { "epoch": 0.13311, "grad_norm": 0.9864618715968552, "learning_rate": 0.003, "loss": 4.047, "step": 13311 }, { "epoch": 0.13312, "grad_norm": 1.2173793332268772, "learning_rate": 0.003, "loss": 4.0815, "step": 13312 }, { "epoch": 0.13313, "grad_norm": 1.2436277242839906, "learning_rate": 0.003, "loss": 4.0551, "step": 13313 }, { "epoch": 0.13314, "grad_norm": 1.3875133554853671, "learning_rate": 0.003, "loss": 4.0694, "step": 13314 }, { "epoch": 0.13315, "grad_norm": 0.9751721952947197, "learning_rate": 0.003, "loss": 4.0804, "step": 13315 }, { "epoch": 0.13316, "grad_norm": 1.278534885777062, "learning_rate": 0.003, "loss": 4.0582, "step": 13316 }, { "epoch": 0.13317, "grad_norm": 1.01636143477108, "learning_rate": 0.003, "loss": 4.0769, "step": 13317 }, { "epoch": 0.13318, "grad_norm": 1.3649983840486495, "learning_rate": 0.003, "loss": 4.077, "step": 13318 }, { "epoch": 0.13319, "grad_norm": 0.9565038889965672, "learning_rate": 0.003, "loss": 4.0895, "step": 13319 }, { "epoch": 0.1332, "grad_norm": 1.262266957645931, "learning_rate": 0.003, "loss": 4.067, "step": 13320 }, { "epoch": 0.13321, "grad_norm": 1.1835388605510213, "learning_rate": 0.003, "loss": 4.0817, "step": 13321 }, { "epoch": 0.13322, "grad_norm": 1.2606599407501045, "learning_rate": 0.003, "loss": 4.0679, "step": 13322 }, { "epoch": 0.13323, "grad_norm": 1.552563990129336, "learning_rate": 0.003, "loss": 4.0853, "step": 13323 }, { "epoch": 0.13324, "grad_norm": 1.046510957547619, "learning_rate": 0.003, "loss": 4.0847, "step": 13324 }, { "epoch": 0.13325, "grad_norm": 1.110530797228649, "learning_rate": 0.003, "loss": 4.0661, "step": 13325 }, { "epoch": 0.13326, "grad_norm": 1.0935804074363407, "learning_rate": 0.003, "loss": 4.1024, "step": 13326 }, { "epoch": 0.13327, "grad_norm": 1.2812629789819532, "learning_rate": 0.003, "loss": 4.0702, "step": 13327 }, { "epoch": 0.13328, "grad_norm": 0.937706571734598, "learning_rate": 0.003, "loss": 4.0779, "step": 13328 }, { "epoch": 0.13329, "grad_norm": 1.504345922785764, "learning_rate": 0.003, "loss": 4.1024, "step": 13329 }, { "epoch": 0.1333, "grad_norm": 0.942919545368339, "learning_rate": 0.003, "loss": 4.0938, "step": 13330 }, { "epoch": 0.13331, "grad_norm": 1.3860701408176805, "learning_rate": 0.003, "loss": 4.0887, "step": 13331 }, { "epoch": 0.13332, "grad_norm": 1.1649499190977812, "learning_rate": 0.003, "loss": 4.071, "step": 13332 }, { "epoch": 0.13333, "grad_norm": 1.150537119298711, "learning_rate": 0.003, "loss": 4.076, "step": 13333 }, { "epoch": 0.13334, "grad_norm": 1.1921957935384304, "learning_rate": 0.003, "loss": 4.0834, "step": 13334 }, { "epoch": 0.13335, "grad_norm": 1.2058355277788635, "learning_rate": 0.003, "loss": 4.0467, "step": 13335 }, { "epoch": 0.13336, "grad_norm": 1.14521866323671, "learning_rate": 0.003, "loss": 4.0668, "step": 13336 }, { "epoch": 0.13337, "grad_norm": 1.1264333972428362, "learning_rate": 0.003, "loss": 4.0629, "step": 13337 }, { "epoch": 0.13338, "grad_norm": 1.4305065758976705, "learning_rate": 0.003, "loss": 4.1101, "step": 13338 }, { "epoch": 0.13339, "grad_norm": 1.3457442251358365, "learning_rate": 0.003, "loss": 4.0811, "step": 13339 }, { "epoch": 0.1334, "grad_norm": 1.0819564548054041, "learning_rate": 0.003, "loss": 4.0543, "step": 13340 }, { "epoch": 0.13341, "grad_norm": 1.1564015759857842, "learning_rate": 0.003, "loss": 4.1099, "step": 13341 }, { "epoch": 0.13342, "grad_norm": 1.2673644877503936, "learning_rate": 0.003, "loss": 4.0904, "step": 13342 }, { "epoch": 0.13343, "grad_norm": 1.2180577385368219, "learning_rate": 0.003, "loss": 4.066, "step": 13343 }, { "epoch": 0.13344, "grad_norm": 1.0635239571968256, "learning_rate": 0.003, "loss": 4.1039, "step": 13344 }, { "epoch": 0.13345, "grad_norm": 1.259260532413804, "learning_rate": 0.003, "loss": 4.1126, "step": 13345 }, { "epoch": 0.13346, "grad_norm": 1.0411929273126528, "learning_rate": 0.003, "loss": 4.07, "step": 13346 }, { "epoch": 0.13347, "grad_norm": 1.1397615490181243, "learning_rate": 0.003, "loss": 4.0542, "step": 13347 }, { "epoch": 0.13348, "grad_norm": 1.259020935793382, "learning_rate": 0.003, "loss": 4.0892, "step": 13348 }, { "epoch": 0.13349, "grad_norm": 1.1319109432970802, "learning_rate": 0.003, "loss": 4.0743, "step": 13349 }, { "epoch": 0.1335, "grad_norm": 1.2067967507361204, "learning_rate": 0.003, "loss": 4.0622, "step": 13350 }, { "epoch": 0.13351, "grad_norm": 1.1658964781212067, "learning_rate": 0.003, "loss": 4.0802, "step": 13351 }, { "epoch": 0.13352, "grad_norm": 1.3254913743020704, "learning_rate": 0.003, "loss": 4.0557, "step": 13352 }, { "epoch": 0.13353, "grad_norm": 1.1830811251199571, "learning_rate": 0.003, "loss": 4.0832, "step": 13353 }, { "epoch": 0.13354, "grad_norm": 1.3095382839092642, "learning_rate": 0.003, "loss": 4.0788, "step": 13354 }, { "epoch": 0.13355, "grad_norm": 1.4355120103751222, "learning_rate": 0.003, "loss": 4.0692, "step": 13355 }, { "epoch": 0.13356, "grad_norm": 0.882950527780688, "learning_rate": 0.003, "loss": 4.0743, "step": 13356 }, { "epoch": 0.13357, "grad_norm": 0.992528936217066, "learning_rate": 0.003, "loss": 4.0645, "step": 13357 }, { "epoch": 0.13358, "grad_norm": 1.3482841530983989, "learning_rate": 0.003, "loss": 4.0615, "step": 13358 }, { "epoch": 0.13359, "grad_norm": 1.0200084021979765, "learning_rate": 0.003, "loss": 4.0844, "step": 13359 }, { "epoch": 0.1336, "grad_norm": 1.3198266502439855, "learning_rate": 0.003, "loss": 4.0809, "step": 13360 }, { "epoch": 0.13361, "grad_norm": 0.9318948648450732, "learning_rate": 0.003, "loss": 4.0564, "step": 13361 }, { "epoch": 0.13362, "grad_norm": 1.156104443980015, "learning_rate": 0.003, "loss": 4.0575, "step": 13362 }, { "epoch": 0.13363, "grad_norm": 1.1348916514353162, "learning_rate": 0.003, "loss": 4.0682, "step": 13363 }, { "epoch": 0.13364, "grad_norm": 0.9545398557086426, "learning_rate": 0.003, "loss": 4.0661, "step": 13364 }, { "epoch": 0.13365, "grad_norm": 1.0430415277592333, "learning_rate": 0.003, "loss": 4.0621, "step": 13365 }, { "epoch": 0.13366, "grad_norm": 1.2973396135907518, "learning_rate": 0.003, "loss": 4.0797, "step": 13366 }, { "epoch": 0.13367, "grad_norm": 1.377584331147787, "learning_rate": 0.003, "loss": 4.0541, "step": 13367 }, { "epoch": 0.13368, "grad_norm": 1.087111394889965, "learning_rate": 0.003, "loss": 4.0876, "step": 13368 }, { "epoch": 0.13369, "grad_norm": 1.1399775172501416, "learning_rate": 0.003, "loss": 4.0739, "step": 13369 }, { "epoch": 0.1337, "grad_norm": 1.2782889127256056, "learning_rate": 0.003, "loss": 4.0739, "step": 13370 }, { "epoch": 0.13371, "grad_norm": 1.2107464720288288, "learning_rate": 0.003, "loss": 4.0734, "step": 13371 }, { "epoch": 0.13372, "grad_norm": 1.1235269728784218, "learning_rate": 0.003, "loss": 4.0712, "step": 13372 }, { "epoch": 0.13373, "grad_norm": 1.171530183194923, "learning_rate": 0.003, "loss": 4.062, "step": 13373 }, { "epoch": 0.13374, "grad_norm": 1.0460985774606695, "learning_rate": 0.003, "loss": 4.0976, "step": 13374 }, { "epoch": 0.13375, "grad_norm": 1.2202546470687783, "learning_rate": 0.003, "loss": 4.0499, "step": 13375 }, { "epoch": 0.13376, "grad_norm": 1.2928630380745547, "learning_rate": 0.003, "loss": 4.0884, "step": 13376 }, { "epoch": 0.13377, "grad_norm": 1.1104800307893876, "learning_rate": 0.003, "loss": 4.0521, "step": 13377 }, { "epoch": 0.13378, "grad_norm": 1.3133149278619125, "learning_rate": 0.003, "loss": 4.0636, "step": 13378 }, { "epoch": 0.13379, "grad_norm": 1.192621777611117, "learning_rate": 0.003, "loss": 4.0603, "step": 13379 }, { "epoch": 0.1338, "grad_norm": 1.6324962573798982, "learning_rate": 0.003, "loss": 4.0654, "step": 13380 }, { "epoch": 0.13381, "grad_norm": 0.9608650701463821, "learning_rate": 0.003, "loss": 4.0426, "step": 13381 }, { "epoch": 0.13382, "grad_norm": 1.309755125207003, "learning_rate": 0.003, "loss": 4.0783, "step": 13382 }, { "epoch": 0.13383, "grad_norm": 1.2184602263012025, "learning_rate": 0.003, "loss": 4.0623, "step": 13383 }, { "epoch": 0.13384, "grad_norm": 0.9707313098371623, "learning_rate": 0.003, "loss": 4.0283, "step": 13384 }, { "epoch": 0.13385, "grad_norm": 1.3512516690476681, "learning_rate": 0.003, "loss": 4.0792, "step": 13385 }, { "epoch": 0.13386, "grad_norm": 1.0809748766020124, "learning_rate": 0.003, "loss": 4.0771, "step": 13386 }, { "epoch": 0.13387, "grad_norm": 1.2485092151247863, "learning_rate": 0.003, "loss": 4.0615, "step": 13387 }, { "epoch": 0.13388, "grad_norm": 1.3101065144013673, "learning_rate": 0.003, "loss": 4.0862, "step": 13388 }, { "epoch": 0.13389, "grad_norm": 1.084908616260319, "learning_rate": 0.003, "loss": 4.0744, "step": 13389 }, { "epoch": 0.1339, "grad_norm": 1.22079628713085, "learning_rate": 0.003, "loss": 4.0668, "step": 13390 }, { "epoch": 0.13391, "grad_norm": 1.0202426905455475, "learning_rate": 0.003, "loss": 4.0495, "step": 13391 }, { "epoch": 0.13392, "grad_norm": 1.2992335676237536, "learning_rate": 0.003, "loss": 4.0698, "step": 13392 }, { "epoch": 0.13393, "grad_norm": 1.092089891274764, "learning_rate": 0.003, "loss": 4.0837, "step": 13393 }, { "epoch": 0.13394, "grad_norm": 1.3867392115270731, "learning_rate": 0.003, "loss": 4.0817, "step": 13394 }, { "epoch": 0.13395, "grad_norm": 0.9820590058309596, "learning_rate": 0.003, "loss": 4.0581, "step": 13395 }, { "epoch": 0.13396, "grad_norm": 1.3005697214378344, "learning_rate": 0.003, "loss": 4.0608, "step": 13396 }, { "epoch": 0.13397, "grad_norm": 1.279191656902853, "learning_rate": 0.003, "loss": 4.0745, "step": 13397 }, { "epoch": 0.13398, "grad_norm": 1.299460212780781, "learning_rate": 0.003, "loss": 4.062, "step": 13398 }, { "epoch": 0.13399, "grad_norm": 1.123847793349164, "learning_rate": 0.003, "loss": 4.0369, "step": 13399 }, { "epoch": 0.134, "grad_norm": 1.254970037907744, "learning_rate": 0.003, "loss": 4.0977, "step": 13400 }, { "epoch": 0.13401, "grad_norm": 1.2751182808912296, "learning_rate": 0.003, "loss": 4.0643, "step": 13401 }, { "epoch": 0.13402, "grad_norm": 1.2877165752219555, "learning_rate": 0.003, "loss": 4.0247, "step": 13402 }, { "epoch": 0.13403, "grad_norm": 1.0003900002349755, "learning_rate": 0.003, "loss": 4.105, "step": 13403 }, { "epoch": 0.13404, "grad_norm": 1.0931429176530132, "learning_rate": 0.003, "loss": 4.0973, "step": 13404 }, { "epoch": 0.13405, "grad_norm": 1.2238882247414367, "learning_rate": 0.003, "loss": 4.0611, "step": 13405 }, { "epoch": 0.13406, "grad_norm": 1.226888309982652, "learning_rate": 0.003, "loss": 4.0813, "step": 13406 }, { "epoch": 0.13407, "grad_norm": 1.5015375172841177, "learning_rate": 0.003, "loss": 4.088, "step": 13407 }, { "epoch": 0.13408, "grad_norm": 1.0494091926556364, "learning_rate": 0.003, "loss": 4.0699, "step": 13408 }, { "epoch": 0.13409, "grad_norm": 1.2981650249555947, "learning_rate": 0.003, "loss": 4.0755, "step": 13409 }, { "epoch": 0.1341, "grad_norm": 1.139531561438358, "learning_rate": 0.003, "loss": 4.0511, "step": 13410 }, { "epoch": 0.13411, "grad_norm": 1.107127473358016, "learning_rate": 0.003, "loss": 4.0829, "step": 13411 }, { "epoch": 0.13412, "grad_norm": 1.1067004651124173, "learning_rate": 0.003, "loss": 4.0731, "step": 13412 }, { "epoch": 0.13413, "grad_norm": 1.2818189360675667, "learning_rate": 0.003, "loss": 4.0836, "step": 13413 }, { "epoch": 0.13414, "grad_norm": 1.0920613489076882, "learning_rate": 0.003, "loss": 4.0941, "step": 13414 }, { "epoch": 0.13415, "grad_norm": 1.323974517486657, "learning_rate": 0.003, "loss": 4.0713, "step": 13415 }, { "epoch": 0.13416, "grad_norm": 1.1510418258436874, "learning_rate": 0.003, "loss": 4.0751, "step": 13416 }, { "epoch": 0.13417, "grad_norm": 1.3101953345153434, "learning_rate": 0.003, "loss": 4.0531, "step": 13417 }, { "epoch": 0.13418, "grad_norm": 1.2484858107402523, "learning_rate": 0.003, "loss": 4.069, "step": 13418 }, { "epoch": 0.13419, "grad_norm": 1.2008316192186337, "learning_rate": 0.003, "loss": 4.087, "step": 13419 }, { "epoch": 0.1342, "grad_norm": 1.1571115511829444, "learning_rate": 0.003, "loss": 4.0583, "step": 13420 }, { "epoch": 0.13421, "grad_norm": 1.3922202445034606, "learning_rate": 0.003, "loss": 4.0889, "step": 13421 }, { "epoch": 0.13422, "grad_norm": 1.0939348589277045, "learning_rate": 0.003, "loss": 4.0923, "step": 13422 }, { "epoch": 0.13423, "grad_norm": 1.3721463765694832, "learning_rate": 0.003, "loss": 4.0792, "step": 13423 }, { "epoch": 0.13424, "grad_norm": 0.9950380943555217, "learning_rate": 0.003, "loss": 4.0645, "step": 13424 }, { "epoch": 0.13425, "grad_norm": 1.4568519919068281, "learning_rate": 0.003, "loss": 4.0896, "step": 13425 }, { "epoch": 0.13426, "grad_norm": 0.8845714068799299, "learning_rate": 0.003, "loss": 4.0402, "step": 13426 }, { "epoch": 0.13427, "grad_norm": 0.961609139691443, "learning_rate": 0.003, "loss": 4.0781, "step": 13427 }, { "epoch": 0.13428, "grad_norm": 1.3462502809623975, "learning_rate": 0.003, "loss": 4.0954, "step": 13428 }, { "epoch": 0.13429, "grad_norm": 1.1194356350584982, "learning_rate": 0.003, "loss": 4.0857, "step": 13429 }, { "epoch": 0.1343, "grad_norm": 1.4920878363470662, "learning_rate": 0.003, "loss": 4.0562, "step": 13430 }, { "epoch": 0.13431, "grad_norm": 0.8483834854771911, "learning_rate": 0.003, "loss": 4.0742, "step": 13431 }, { "epoch": 0.13432, "grad_norm": 0.9112242246450506, "learning_rate": 0.003, "loss": 4.0637, "step": 13432 }, { "epoch": 0.13433, "grad_norm": 1.2665685261460575, "learning_rate": 0.003, "loss": 4.0735, "step": 13433 }, { "epoch": 0.13434, "grad_norm": 1.0610030221051627, "learning_rate": 0.003, "loss": 4.075, "step": 13434 }, { "epoch": 0.13435, "grad_norm": 1.321751025696581, "learning_rate": 0.003, "loss": 4.105, "step": 13435 }, { "epoch": 0.13436, "grad_norm": 0.9682699838920557, "learning_rate": 0.003, "loss": 4.0891, "step": 13436 }, { "epoch": 0.13437, "grad_norm": 1.3736444982158735, "learning_rate": 0.003, "loss": 4.0725, "step": 13437 }, { "epoch": 0.13438, "grad_norm": 1.2448295072637303, "learning_rate": 0.003, "loss": 4.0895, "step": 13438 }, { "epoch": 0.13439, "grad_norm": 1.2473219844564283, "learning_rate": 0.003, "loss": 4.0876, "step": 13439 }, { "epoch": 0.1344, "grad_norm": 1.3130340499660091, "learning_rate": 0.003, "loss": 4.0969, "step": 13440 }, { "epoch": 0.13441, "grad_norm": 1.240881935789378, "learning_rate": 0.003, "loss": 4.0569, "step": 13441 }, { "epoch": 0.13442, "grad_norm": 1.269013493595987, "learning_rate": 0.003, "loss": 4.0715, "step": 13442 }, { "epoch": 0.13443, "grad_norm": 1.2069173559517443, "learning_rate": 0.003, "loss": 4.0829, "step": 13443 }, { "epoch": 0.13444, "grad_norm": 1.2916260889649558, "learning_rate": 0.003, "loss": 4.0606, "step": 13444 }, { "epoch": 0.13445, "grad_norm": 1.021728453694052, "learning_rate": 0.003, "loss": 4.0605, "step": 13445 }, { "epoch": 0.13446, "grad_norm": 1.1582896508308027, "learning_rate": 0.003, "loss": 4.0818, "step": 13446 }, { "epoch": 0.13447, "grad_norm": 1.2018218800263953, "learning_rate": 0.003, "loss": 4.1036, "step": 13447 }, { "epoch": 0.13448, "grad_norm": 1.1793134666029423, "learning_rate": 0.003, "loss": 4.0781, "step": 13448 }, { "epoch": 0.13449, "grad_norm": 1.113403756290677, "learning_rate": 0.003, "loss": 4.0775, "step": 13449 }, { "epoch": 0.1345, "grad_norm": 1.2679419571020423, "learning_rate": 0.003, "loss": 4.0183, "step": 13450 }, { "epoch": 0.13451, "grad_norm": 1.0504262222413014, "learning_rate": 0.003, "loss": 4.0624, "step": 13451 }, { "epoch": 0.13452, "grad_norm": 1.6461457947162657, "learning_rate": 0.003, "loss": 4.0691, "step": 13452 }, { "epoch": 0.13453, "grad_norm": 0.8415228653785007, "learning_rate": 0.003, "loss": 4.0664, "step": 13453 }, { "epoch": 0.13454, "grad_norm": 0.9853824307930039, "learning_rate": 0.003, "loss": 4.0695, "step": 13454 }, { "epoch": 0.13455, "grad_norm": 1.3711398037232354, "learning_rate": 0.003, "loss": 4.0963, "step": 13455 }, { "epoch": 0.13456, "grad_norm": 0.9299209880239284, "learning_rate": 0.003, "loss": 4.0874, "step": 13456 }, { "epoch": 0.13457, "grad_norm": 1.2201305236616733, "learning_rate": 0.003, "loss": 4.0521, "step": 13457 }, { "epoch": 0.13458, "grad_norm": 1.042117831521888, "learning_rate": 0.003, "loss": 4.0965, "step": 13458 }, { "epoch": 0.13459, "grad_norm": 1.2885917369252293, "learning_rate": 0.003, "loss": 4.0975, "step": 13459 }, { "epoch": 0.1346, "grad_norm": 1.1186516456657598, "learning_rate": 0.003, "loss": 4.0818, "step": 13460 }, { "epoch": 0.13461, "grad_norm": 1.1909986315329697, "learning_rate": 0.003, "loss": 4.0769, "step": 13461 }, { "epoch": 0.13462, "grad_norm": 1.2576997493640354, "learning_rate": 0.003, "loss": 4.0537, "step": 13462 }, { "epoch": 0.13463, "grad_norm": 1.155181532510466, "learning_rate": 0.003, "loss": 4.0841, "step": 13463 }, { "epoch": 0.13464, "grad_norm": 1.5427338937478448, "learning_rate": 0.003, "loss": 4.0943, "step": 13464 }, { "epoch": 0.13465, "grad_norm": 1.1015368366752862, "learning_rate": 0.003, "loss": 4.0531, "step": 13465 }, { "epoch": 0.13466, "grad_norm": 1.328116987003986, "learning_rate": 0.003, "loss": 4.0653, "step": 13466 }, { "epoch": 0.13467, "grad_norm": 1.4210595457619721, "learning_rate": 0.003, "loss": 4.1008, "step": 13467 }, { "epoch": 0.13468, "grad_norm": 1.0154008116906554, "learning_rate": 0.003, "loss": 4.0724, "step": 13468 }, { "epoch": 0.13469, "grad_norm": 1.3831223411143163, "learning_rate": 0.003, "loss": 4.0814, "step": 13469 }, { "epoch": 0.1347, "grad_norm": 1.1187264304497462, "learning_rate": 0.003, "loss": 4.0617, "step": 13470 }, { "epoch": 0.13471, "grad_norm": 1.3256152725200183, "learning_rate": 0.003, "loss": 4.0717, "step": 13471 }, { "epoch": 0.13472, "grad_norm": 1.0436438983185985, "learning_rate": 0.003, "loss": 4.0657, "step": 13472 }, { "epoch": 0.13473, "grad_norm": 1.3890226571349744, "learning_rate": 0.003, "loss": 4.0404, "step": 13473 }, { "epoch": 0.13474, "grad_norm": 0.9523790124680513, "learning_rate": 0.003, "loss": 4.0759, "step": 13474 }, { "epoch": 0.13475, "grad_norm": 1.2705342694225594, "learning_rate": 0.003, "loss": 4.0776, "step": 13475 }, { "epoch": 0.13476, "grad_norm": 1.0416515369439097, "learning_rate": 0.003, "loss": 4.0647, "step": 13476 }, { "epoch": 0.13477, "grad_norm": 1.213808961134788, "learning_rate": 0.003, "loss": 4.068, "step": 13477 }, { "epoch": 0.13478, "grad_norm": 1.2567203891908283, "learning_rate": 0.003, "loss": 4.084, "step": 13478 }, { "epoch": 0.13479, "grad_norm": 1.0627266915077016, "learning_rate": 0.003, "loss": 4.0577, "step": 13479 }, { "epoch": 0.1348, "grad_norm": 1.1942181777864018, "learning_rate": 0.003, "loss": 4.0835, "step": 13480 }, { "epoch": 0.13481, "grad_norm": 0.9086917010183528, "learning_rate": 0.003, "loss": 4.0802, "step": 13481 }, { "epoch": 0.13482, "grad_norm": 1.0518581902163413, "learning_rate": 0.003, "loss": 4.0506, "step": 13482 }, { "epoch": 0.13483, "grad_norm": 1.1908018688437683, "learning_rate": 0.003, "loss": 4.0328, "step": 13483 }, { "epoch": 0.13484, "grad_norm": 1.1141590588201575, "learning_rate": 0.003, "loss": 4.0731, "step": 13484 }, { "epoch": 0.13485, "grad_norm": 1.254355944042286, "learning_rate": 0.003, "loss": 4.0946, "step": 13485 }, { "epoch": 0.13486, "grad_norm": 1.3334461860370763, "learning_rate": 0.003, "loss": 4.0805, "step": 13486 }, { "epoch": 0.13487, "grad_norm": 1.2665983361640198, "learning_rate": 0.003, "loss": 4.0623, "step": 13487 }, { "epoch": 0.13488, "grad_norm": 1.0654884519609067, "learning_rate": 0.003, "loss": 4.0719, "step": 13488 }, { "epoch": 0.13489, "grad_norm": 1.7471630662376938, "learning_rate": 0.003, "loss": 4.064, "step": 13489 }, { "epoch": 0.1349, "grad_norm": 1.08610677249902, "learning_rate": 0.003, "loss": 4.0733, "step": 13490 }, { "epoch": 0.13491, "grad_norm": 1.6699300578111145, "learning_rate": 0.003, "loss": 4.0928, "step": 13491 }, { "epoch": 0.13492, "grad_norm": 1.0051323094271105, "learning_rate": 0.003, "loss": 4.0836, "step": 13492 }, { "epoch": 0.13493, "grad_norm": 1.1366130593463755, "learning_rate": 0.003, "loss": 4.0535, "step": 13493 }, { "epoch": 0.13494, "grad_norm": 1.0985346160202802, "learning_rate": 0.003, "loss": 4.0663, "step": 13494 }, { "epoch": 0.13495, "grad_norm": 1.3406522988660494, "learning_rate": 0.003, "loss": 4.1061, "step": 13495 }, { "epoch": 0.13496, "grad_norm": 1.28654771486778, "learning_rate": 0.003, "loss": 4.0986, "step": 13496 }, { "epoch": 0.13497, "grad_norm": 1.0071406929819673, "learning_rate": 0.003, "loss": 4.0599, "step": 13497 }, { "epoch": 0.13498, "grad_norm": 1.4246452225409638, "learning_rate": 0.003, "loss": 4.0838, "step": 13498 }, { "epoch": 0.13499, "grad_norm": 0.9972509121748382, "learning_rate": 0.003, "loss": 4.0369, "step": 13499 }, { "epoch": 0.135, "grad_norm": 1.189967531780217, "learning_rate": 0.003, "loss": 4.0664, "step": 13500 }, { "epoch": 0.13501, "grad_norm": 1.1493270847412234, "learning_rate": 0.003, "loss": 4.0958, "step": 13501 }, { "epoch": 0.13502, "grad_norm": 1.189309815156667, "learning_rate": 0.003, "loss": 4.0857, "step": 13502 }, { "epoch": 0.13503, "grad_norm": 1.2288407869403217, "learning_rate": 0.003, "loss": 4.0671, "step": 13503 }, { "epoch": 0.13504, "grad_norm": 1.1103535353418132, "learning_rate": 0.003, "loss": 4.0657, "step": 13504 }, { "epoch": 0.13505, "grad_norm": 1.4194568999011024, "learning_rate": 0.003, "loss": 4.094, "step": 13505 }, { "epoch": 0.13506, "grad_norm": 1.220666174798732, "learning_rate": 0.003, "loss": 4.0817, "step": 13506 }, { "epoch": 0.13507, "grad_norm": 1.1415944662630442, "learning_rate": 0.003, "loss": 4.0895, "step": 13507 }, { "epoch": 0.13508, "grad_norm": 1.1661369922608253, "learning_rate": 0.003, "loss": 4.0583, "step": 13508 }, { "epoch": 0.13509, "grad_norm": 1.3463481120750507, "learning_rate": 0.003, "loss": 4.0625, "step": 13509 }, { "epoch": 0.1351, "grad_norm": 1.0168206869032022, "learning_rate": 0.003, "loss": 4.0737, "step": 13510 }, { "epoch": 0.13511, "grad_norm": 1.1550104408646904, "learning_rate": 0.003, "loss": 4.0669, "step": 13511 }, { "epoch": 0.13512, "grad_norm": 1.4461693850672317, "learning_rate": 0.003, "loss": 4.0698, "step": 13512 }, { "epoch": 0.13513, "grad_norm": 1.0399061773826157, "learning_rate": 0.003, "loss": 4.0613, "step": 13513 }, { "epoch": 0.13514, "grad_norm": 1.342808826553646, "learning_rate": 0.003, "loss": 4.0884, "step": 13514 }, { "epoch": 0.13515, "grad_norm": 1.032416494086337, "learning_rate": 0.003, "loss": 4.0761, "step": 13515 }, { "epoch": 0.13516, "grad_norm": 1.3650543147519953, "learning_rate": 0.003, "loss": 4.0307, "step": 13516 }, { "epoch": 0.13517, "grad_norm": 1.0965216566801397, "learning_rate": 0.003, "loss": 4.069, "step": 13517 }, { "epoch": 0.13518, "grad_norm": 1.099857419661087, "learning_rate": 0.003, "loss": 4.0714, "step": 13518 }, { "epoch": 0.13519, "grad_norm": 1.1411897913570885, "learning_rate": 0.003, "loss": 4.0787, "step": 13519 }, { "epoch": 0.1352, "grad_norm": 1.2794171237422962, "learning_rate": 0.003, "loss": 4.0437, "step": 13520 }, { "epoch": 0.13521, "grad_norm": 0.9811240347362448, "learning_rate": 0.003, "loss": 4.0719, "step": 13521 }, { "epoch": 0.13522, "grad_norm": 1.268850909392575, "learning_rate": 0.003, "loss": 4.082, "step": 13522 }, { "epoch": 0.13523, "grad_norm": 1.1586115250772768, "learning_rate": 0.003, "loss": 4.0639, "step": 13523 }, { "epoch": 0.13524, "grad_norm": 0.9997886279609471, "learning_rate": 0.003, "loss": 4.0734, "step": 13524 }, { "epoch": 0.13525, "grad_norm": 1.2578207296594535, "learning_rate": 0.003, "loss": 4.0767, "step": 13525 }, { "epoch": 0.13526, "grad_norm": 1.1211549330059851, "learning_rate": 0.003, "loss": 4.0621, "step": 13526 }, { "epoch": 0.13527, "grad_norm": 1.3489510791240398, "learning_rate": 0.003, "loss": 4.0712, "step": 13527 }, { "epoch": 0.13528, "grad_norm": 0.9549824449680652, "learning_rate": 0.003, "loss": 4.0617, "step": 13528 }, { "epoch": 0.13529, "grad_norm": 1.402123824992118, "learning_rate": 0.003, "loss": 4.1072, "step": 13529 }, { "epoch": 0.1353, "grad_norm": 0.9658857549632947, "learning_rate": 0.003, "loss": 4.0863, "step": 13530 }, { "epoch": 0.13531, "grad_norm": 1.2385811105356495, "learning_rate": 0.003, "loss": 4.0839, "step": 13531 }, { "epoch": 0.13532, "grad_norm": 1.3384536701437566, "learning_rate": 0.003, "loss": 4.1002, "step": 13532 }, { "epoch": 0.13533, "grad_norm": 1.1271439594468355, "learning_rate": 0.003, "loss": 4.0601, "step": 13533 }, { "epoch": 0.13534, "grad_norm": 1.4053236824732496, "learning_rate": 0.003, "loss": 4.1023, "step": 13534 }, { "epoch": 0.13535, "grad_norm": 1.4210254217327944, "learning_rate": 0.003, "loss": 4.0799, "step": 13535 }, { "epoch": 0.13536, "grad_norm": 1.1757317921279626, "learning_rate": 0.003, "loss": 4.063, "step": 13536 }, { "epoch": 0.13537, "grad_norm": 1.1826886943855999, "learning_rate": 0.003, "loss": 4.066, "step": 13537 }, { "epoch": 0.13538, "grad_norm": 1.3867148362658737, "learning_rate": 0.003, "loss": 4.0922, "step": 13538 }, { "epoch": 0.13539, "grad_norm": 1.339948436143428, "learning_rate": 0.003, "loss": 4.0774, "step": 13539 }, { "epoch": 0.1354, "grad_norm": 1.0840175756391561, "learning_rate": 0.003, "loss": 4.0587, "step": 13540 }, { "epoch": 0.13541, "grad_norm": 1.1034967651255367, "learning_rate": 0.003, "loss": 4.0965, "step": 13541 }, { "epoch": 0.13542, "grad_norm": 1.495915605936551, "learning_rate": 0.003, "loss": 4.0917, "step": 13542 }, { "epoch": 0.13543, "grad_norm": 1.0025020971735863, "learning_rate": 0.003, "loss": 4.0556, "step": 13543 }, { "epoch": 0.13544, "grad_norm": 1.4015261150506153, "learning_rate": 0.003, "loss": 4.0791, "step": 13544 }, { "epoch": 0.13545, "grad_norm": 1.0777302094105228, "learning_rate": 0.003, "loss": 4.0684, "step": 13545 }, { "epoch": 0.13546, "grad_norm": 1.0075832217606906, "learning_rate": 0.003, "loss": 4.0641, "step": 13546 }, { "epoch": 0.13547, "grad_norm": 1.4568098120896713, "learning_rate": 0.003, "loss": 4.0418, "step": 13547 }, { "epoch": 0.13548, "grad_norm": 1.0073230103121065, "learning_rate": 0.003, "loss": 4.0747, "step": 13548 }, { "epoch": 0.13549, "grad_norm": 1.1906217853371375, "learning_rate": 0.003, "loss": 4.0562, "step": 13549 }, { "epoch": 0.1355, "grad_norm": 0.9433059001387746, "learning_rate": 0.003, "loss": 4.0477, "step": 13550 }, { "epoch": 0.13551, "grad_norm": 1.1362882888979027, "learning_rate": 0.003, "loss": 4.068, "step": 13551 }, { "epoch": 0.13552, "grad_norm": 1.1704663789178789, "learning_rate": 0.003, "loss": 4.0749, "step": 13552 }, { "epoch": 0.13553, "grad_norm": 1.2410924551269613, "learning_rate": 0.003, "loss": 4.1047, "step": 13553 }, { "epoch": 0.13554, "grad_norm": 1.1613234355481612, "learning_rate": 0.003, "loss": 4.067, "step": 13554 }, { "epoch": 0.13555, "grad_norm": 1.0467286749826417, "learning_rate": 0.003, "loss": 4.0365, "step": 13555 }, { "epoch": 0.13556, "grad_norm": 1.3129682484329963, "learning_rate": 0.003, "loss": 4.0604, "step": 13556 }, { "epoch": 0.13557, "grad_norm": 1.308745439163254, "learning_rate": 0.003, "loss": 4.0661, "step": 13557 }, { "epoch": 0.13558, "grad_norm": 1.2234312370273055, "learning_rate": 0.003, "loss": 4.0491, "step": 13558 }, { "epoch": 0.13559, "grad_norm": 0.9438356557960328, "learning_rate": 0.003, "loss": 4.0797, "step": 13559 }, { "epoch": 0.1356, "grad_norm": 1.1359668216145018, "learning_rate": 0.003, "loss": 4.047, "step": 13560 }, { "epoch": 0.13561, "grad_norm": 1.2025332339088028, "learning_rate": 0.003, "loss": 4.0729, "step": 13561 }, { "epoch": 0.13562, "grad_norm": 1.0278362689972806, "learning_rate": 0.003, "loss": 4.0442, "step": 13562 }, { "epoch": 0.13563, "grad_norm": 1.2927399972064535, "learning_rate": 0.003, "loss": 4.0202, "step": 13563 }, { "epoch": 0.13564, "grad_norm": 1.1454504278844562, "learning_rate": 0.003, "loss": 4.0857, "step": 13564 }, { "epoch": 0.13565, "grad_norm": 1.2713561682590202, "learning_rate": 0.003, "loss": 4.0622, "step": 13565 }, { "epoch": 0.13566, "grad_norm": 1.0101189277867662, "learning_rate": 0.003, "loss": 4.0611, "step": 13566 }, { "epoch": 0.13567, "grad_norm": 1.3452338901105434, "learning_rate": 0.003, "loss": 4.0695, "step": 13567 }, { "epoch": 0.13568, "grad_norm": 0.9651810314708721, "learning_rate": 0.003, "loss": 4.0609, "step": 13568 }, { "epoch": 0.13569, "grad_norm": 1.200150574860146, "learning_rate": 0.003, "loss": 4.0705, "step": 13569 }, { "epoch": 0.1357, "grad_norm": 1.3013228175830704, "learning_rate": 0.003, "loss": 4.0741, "step": 13570 }, { "epoch": 0.13571, "grad_norm": 1.1413524786109808, "learning_rate": 0.003, "loss": 4.0728, "step": 13571 }, { "epoch": 0.13572, "grad_norm": 1.3872233678122348, "learning_rate": 0.003, "loss": 4.0799, "step": 13572 }, { "epoch": 0.13573, "grad_norm": 1.19536948436539, "learning_rate": 0.003, "loss": 4.0534, "step": 13573 }, { "epoch": 0.13574, "grad_norm": 1.2507510937540127, "learning_rate": 0.003, "loss": 4.0772, "step": 13574 }, { "epoch": 0.13575, "grad_norm": 1.0919209378269294, "learning_rate": 0.003, "loss": 4.0419, "step": 13575 }, { "epoch": 0.13576, "grad_norm": 1.4568843241344767, "learning_rate": 0.003, "loss": 4.0795, "step": 13576 }, { "epoch": 0.13577, "grad_norm": 1.0313914316793116, "learning_rate": 0.003, "loss": 4.0796, "step": 13577 }, { "epoch": 0.13578, "grad_norm": 1.2021084920918845, "learning_rate": 0.003, "loss": 4.0807, "step": 13578 }, { "epoch": 0.13579, "grad_norm": 1.180819236115947, "learning_rate": 0.003, "loss": 4.0687, "step": 13579 }, { "epoch": 0.1358, "grad_norm": 1.112970308942345, "learning_rate": 0.003, "loss": 4.0676, "step": 13580 }, { "epoch": 0.13581, "grad_norm": 1.6581105119328479, "learning_rate": 0.003, "loss": 4.0569, "step": 13581 }, { "epoch": 0.13582, "grad_norm": 1.2212266178712439, "learning_rate": 0.003, "loss": 4.0886, "step": 13582 }, { "epoch": 0.13583, "grad_norm": 1.2713278799534757, "learning_rate": 0.003, "loss": 4.0725, "step": 13583 }, { "epoch": 0.13584, "grad_norm": 0.9492981552071117, "learning_rate": 0.003, "loss": 4.0688, "step": 13584 }, { "epoch": 0.13585, "grad_norm": 1.376007204854274, "learning_rate": 0.003, "loss": 4.0578, "step": 13585 }, { "epoch": 0.13586, "grad_norm": 1.0569112183452425, "learning_rate": 0.003, "loss": 4.1088, "step": 13586 }, { "epoch": 0.13587, "grad_norm": 1.3311495423571125, "learning_rate": 0.003, "loss": 4.0563, "step": 13587 }, { "epoch": 0.13588, "grad_norm": 1.0424856372298525, "learning_rate": 0.003, "loss": 4.0657, "step": 13588 }, { "epoch": 0.13589, "grad_norm": 1.3682385130528956, "learning_rate": 0.003, "loss": 4.0788, "step": 13589 }, { "epoch": 0.1359, "grad_norm": 1.201664787748843, "learning_rate": 0.003, "loss": 4.043, "step": 13590 }, { "epoch": 0.13591, "grad_norm": 1.3490298062373562, "learning_rate": 0.003, "loss": 4.0589, "step": 13591 }, { "epoch": 0.13592, "grad_norm": 0.9562478284952731, "learning_rate": 0.003, "loss": 4.0571, "step": 13592 }, { "epoch": 0.13593, "grad_norm": 1.2375419040612659, "learning_rate": 0.003, "loss": 4.0715, "step": 13593 }, { "epoch": 0.13594, "grad_norm": 1.0476579348639876, "learning_rate": 0.003, "loss": 4.0828, "step": 13594 }, { "epoch": 0.13595, "grad_norm": 1.0717762364233978, "learning_rate": 0.003, "loss": 4.0677, "step": 13595 }, { "epoch": 0.13596, "grad_norm": 1.0671422537293738, "learning_rate": 0.003, "loss": 4.0785, "step": 13596 }, { "epoch": 0.13597, "grad_norm": 1.3256644481635962, "learning_rate": 0.003, "loss": 4.0743, "step": 13597 }, { "epoch": 0.13598, "grad_norm": 1.144294080622358, "learning_rate": 0.003, "loss": 4.1036, "step": 13598 }, { "epoch": 0.13599, "grad_norm": 1.0859542569821918, "learning_rate": 0.003, "loss": 4.0842, "step": 13599 }, { "epoch": 0.136, "grad_norm": 1.1406897699892184, "learning_rate": 0.003, "loss": 4.0234, "step": 13600 }, { "epoch": 0.13601, "grad_norm": 1.247529096109531, "learning_rate": 0.003, "loss": 4.0854, "step": 13601 }, { "epoch": 0.13602, "grad_norm": 1.1840828561107422, "learning_rate": 0.003, "loss": 4.0757, "step": 13602 }, { "epoch": 0.13603, "grad_norm": 1.2623904567135522, "learning_rate": 0.003, "loss": 4.0836, "step": 13603 }, { "epoch": 0.13604, "grad_norm": 0.9505781321776695, "learning_rate": 0.003, "loss": 4.0598, "step": 13604 }, { "epoch": 0.13605, "grad_norm": 1.2276234061362843, "learning_rate": 0.003, "loss": 4.024, "step": 13605 }, { "epoch": 0.13606, "grad_norm": 0.997265695680624, "learning_rate": 0.003, "loss": 4.0601, "step": 13606 }, { "epoch": 0.13607, "grad_norm": 1.2553216295764094, "learning_rate": 0.003, "loss": 4.102, "step": 13607 }, { "epoch": 0.13608, "grad_norm": 1.200664004006863, "learning_rate": 0.003, "loss": 4.0851, "step": 13608 }, { "epoch": 0.13609, "grad_norm": 1.0895193259639602, "learning_rate": 0.003, "loss": 4.0493, "step": 13609 }, { "epoch": 0.1361, "grad_norm": 1.4530026766767474, "learning_rate": 0.003, "loss": 4.0744, "step": 13610 }, { "epoch": 0.13611, "grad_norm": 1.2262235822245915, "learning_rate": 0.003, "loss": 4.0643, "step": 13611 }, { "epoch": 0.13612, "grad_norm": 1.36036589820018, "learning_rate": 0.003, "loss": 4.0841, "step": 13612 }, { "epoch": 0.13613, "grad_norm": 1.1091710118023062, "learning_rate": 0.003, "loss": 4.0669, "step": 13613 }, { "epoch": 0.13614, "grad_norm": 1.1614141717257438, "learning_rate": 0.003, "loss": 4.0587, "step": 13614 }, { "epoch": 0.13615, "grad_norm": 1.3336021860149063, "learning_rate": 0.003, "loss": 4.0801, "step": 13615 }, { "epoch": 0.13616, "grad_norm": 1.153599402627099, "learning_rate": 0.003, "loss": 4.0539, "step": 13616 }, { "epoch": 0.13617, "grad_norm": 1.1603709609738238, "learning_rate": 0.003, "loss": 4.0625, "step": 13617 }, { "epoch": 0.13618, "grad_norm": 1.1346068059085979, "learning_rate": 0.003, "loss": 4.0879, "step": 13618 }, { "epoch": 0.13619, "grad_norm": 1.0892165185334561, "learning_rate": 0.003, "loss": 4.0813, "step": 13619 }, { "epoch": 0.1362, "grad_norm": 1.1292986742158317, "learning_rate": 0.003, "loss": 4.0841, "step": 13620 }, { "epoch": 0.13621, "grad_norm": 1.560107881154133, "learning_rate": 0.003, "loss": 4.0691, "step": 13621 }, { "epoch": 0.13622, "grad_norm": 1.1694445330416534, "learning_rate": 0.003, "loss": 4.066, "step": 13622 }, { "epoch": 0.13623, "grad_norm": 1.614640768254796, "learning_rate": 0.003, "loss": 4.0837, "step": 13623 }, { "epoch": 0.13624, "grad_norm": 0.9008587605126983, "learning_rate": 0.003, "loss": 4.0686, "step": 13624 }, { "epoch": 0.13625, "grad_norm": 1.0479880204115588, "learning_rate": 0.003, "loss": 4.0697, "step": 13625 }, { "epoch": 0.13626, "grad_norm": 1.2980084288540925, "learning_rate": 0.003, "loss": 4.078, "step": 13626 }, { "epoch": 0.13627, "grad_norm": 1.2328245026092544, "learning_rate": 0.003, "loss": 4.051, "step": 13627 }, { "epoch": 0.13628, "grad_norm": 1.2485744616304153, "learning_rate": 0.003, "loss": 4.0765, "step": 13628 }, { "epoch": 0.13629, "grad_norm": 1.1805256275888583, "learning_rate": 0.003, "loss": 4.0548, "step": 13629 }, { "epoch": 0.1363, "grad_norm": 1.1925719533196926, "learning_rate": 0.003, "loss": 4.0382, "step": 13630 }, { "epoch": 0.13631, "grad_norm": 1.0630848397550998, "learning_rate": 0.003, "loss": 4.0557, "step": 13631 }, { "epoch": 0.13632, "grad_norm": 1.350964628485614, "learning_rate": 0.003, "loss": 4.1045, "step": 13632 }, { "epoch": 0.13633, "grad_norm": 0.9968793398396996, "learning_rate": 0.003, "loss": 4.0642, "step": 13633 }, { "epoch": 0.13634, "grad_norm": 1.412486851774138, "learning_rate": 0.003, "loss": 4.0668, "step": 13634 }, { "epoch": 0.13635, "grad_norm": 1.03583980831029, "learning_rate": 0.003, "loss": 4.0908, "step": 13635 }, { "epoch": 0.13636, "grad_norm": 1.3373321063845935, "learning_rate": 0.003, "loss": 4.0715, "step": 13636 }, { "epoch": 0.13637, "grad_norm": 1.241854730615486, "learning_rate": 0.003, "loss": 4.059, "step": 13637 }, { "epoch": 0.13638, "grad_norm": 1.255189863858571, "learning_rate": 0.003, "loss": 4.0965, "step": 13638 }, { "epoch": 0.13639, "grad_norm": 1.294169323504986, "learning_rate": 0.003, "loss": 4.077, "step": 13639 }, { "epoch": 0.1364, "grad_norm": 1.3626068385772372, "learning_rate": 0.003, "loss": 4.0622, "step": 13640 }, { "epoch": 0.13641, "grad_norm": 1.1096206915740712, "learning_rate": 0.003, "loss": 4.097, "step": 13641 }, { "epoch": 0.13642, "grad_norm": 1.1413504000739507, "learning_rate": 0.003, "loss": 4.0742, "step": 13642 }, { "epoch": 0.13643, "grad_norm": 1.356345520757034, "learning_rate": 0.003, "loss": 4.0869, "step": 13643 }, { "epoch": 0.13644, "grad_norm": 1.0209199059256513, "learning_rate": 0.003, "loss": 4.0433, "step": 13644 }, { "epoch": 0.13645, "grad_norm": 1.173159769080088, "learning_rate": 0.003, "loss": 4.048, "step": 13645 }, { "epoch": 0.13646, "grad_norm": 0.9623922125269074, "learning_rate": 0.003, "loss": 4.0333, "step": 13646 }, { "epoch": 0.13647, "grad_norm": 1.3193897614870387, "learning_rate": 0.003, "loss": 4.0773, "step": 13647 }, { "epoch": 0.13648, "grad_norm": 1.0635369222409417, "learning_rate": 0.003, "loss": 4.0338, "step": 13648 }, { "epoch": 0.13649, "grad_norm": 1.1228393873951896, "learning_rate": 0.003, "loss": 4.0718, "step": 13649 }, { "epoch": 0.1365, "grad_norm": 1.3158969083041163, "learning_rate": 0.003, "loss": 4.0724, "step": 13650 }, { "epoch": 0.13651, "grad_norm": 1.0029865082671081, "learning_rate": 0.003, "loss": 4.0724, "step": 13651 }, { "epoch": 0.13652, "grad_norm": 1.3868079902769934, "learning_rate": 0.003, "loss": 4.0669, "step": 13652 }, { "epoch": 0.13653, "grad_norm": 1.1745509739886861, "learning_rate": 0.003, "loss": 4.0689, "step": 13653 }, { "epoch": 0.13654, "grad_norm": 1.4310037147962442, "learning_rate": 0.003, "loss": 4.0696, "step": 13654 }, { "epoch": 0.13655, "grad_norm": 1.315973174976202, "learning_rate": 0.003, "loss": 4.0496, "step": 13655 }, { "epoch": 0.13656, "grad_norm": 1.1774246017219323, "learning_rate": 0.003, "loss": 4.0699, "step": 13656 }, { "epoch": 0.13657, "grad_norm": 1.2043319701834125, "learning_rate": 0.003, "loss": 4.0602, "step": 13657 }, { "epoch": 0.13658, "grad_norm": 1.0467647848338972, "learning_rate": 0.003, "loss": 4.0945, "step": 13658 }, { "epoch": 0.13659, "grad_norm": 1.1918431884790026, "learning_rate": 0.003, "loss": 4.0696, "step": 13659 }, { "epoch": 0.1366, "grad_norm": 1.0430180175805952, "learning_rate": 0.003, "loss": 4.1008, "step": 13660 }, { "epoch": 0.13661, "grad_norm": 1.1816583087991992, "learning_rate": 0.003, "loss": 4.0645, "step": 13661 }, { "epoch": 0.13662, "grad_norm": 1.1048636429833953, "learning_rate": 0.003, "loss": 4.0796, "step": 13662 }, { "epoch": 0.13663, "grad_norm": 1.4735721937903374, "learning_rate": 0.003, "loss": 4.0846, "step": 13663 }, { "epoch": 0.13664, "grad_norm": 1.2188453846802905, "learning_rate": 0.003, "loss": 4.0362, "step": 13664 }, { "epoch": 0.13665, "grad_norm": 1.3376154423362254, "learning_rate": 0.003, "loss": 4.0889, "step": 13665 }, { "epoch": 0.13666, "grad_norm": 1.1007384584021513, "learning_rate": 0.003, "loss": 4.0806, "step": 13666 }, { "epoch": 0.13667, "grad_norm": 1.3815169014254765, "learning_rate": 0.003, "loss": 4.0836, "step": 13667 }, { "epoch": 0.13668, "grad_norm": 1.064453029465472, "learning_rate": 0.003, "loss": 4.0832, "step": 13668 }, { "epoch": 0.13669, "grad_norm": 1.2765847403316573, "learning_rate": 0.003, "loss": 4.062, "step": 13669 }, { "epoch": 0.1367, "grad_norm": 0.9455494847940942, "learning_rate": 0.003, "loss": 4.0736, "step": 13670 }, { "epoch": 0.13671, "grad_norm": 1.2949353596820141, "learning_rate": 0.003, "loss": 4.0802, "step": 13671 }, { "epoch": 0.13672, "grad_norm": 1.1328451712184167, "learning_rate": 0.003, "loss": 4.0435, "step": 13672 }, { "epoch": 0.13673, "grad_norm": 1.141295914638846, "learning_rate": 0.003, "loss": 4.0658, "step": 13673 }, { "epoch": 0.13674, "grad_norm": 1.2513018708628711, "learning_rate": 0.003, "loss": 4.0795, "step": 13674 }, { "epoch": 0.13675, "grad_norm": 1.0576007704115173, "learning_rate": 0.003, "loss": 4.0453, "step": 13675 }, { "epoch": 0.13676, "grad_norm": 1.536526245377778, "learning_rate": 0.003, "loss": 4.0347, "step": 13676 }, { "epoch": 0.13677, "grad_norm": 1.0623188298061705, "learning_rate": 0.003, "loss": 4.0723, "step": 13677 }, { "epoch": 0.13678, "grad_norm": 1.0695343095334628, "learning_rate": 0.003, "loss": 4.0597, "step": 13678 }, { "epoch": 0.13679, "grad_norm": 1.255050117367722, "learning_rate": 0.003, "loss": 4.0738, "step": 13679 }, { "epoch": 0.1368, "grad_norm": 1.143266529110637, "learning_rate": 0.003, "loss": 4.0478, "step": 13680 }, { "epoch": 0.13681, "grad_norm": 1.1790345274322582, "learning_rate": 0.003, "loss": 4.0511, "step": 13681 }, { "epoch": 0.13682, "grad_norm": 0.8769317544083803, "learning_rate": 0.003, "loss": 4.0863, "step": 13682 }, { "epoch": 0.13683, "grad_norm": 0.9956358013379742, "learning_rate": 0.003, "loss": 4.0603, "step": 13683 }, { "epoch": 0.13684, "grad_norm": 1.2610983337776362, "learning_rate": 0.003, "loss": 4.071, "step": 13684 }, { "epoch": 0.13685, "grad_norm": 1.224229969634879, "learning_rate": 0.003, "loss": 4.0484, "step": 13685 }, { "epoch": 0.13686, "grad_norm": 1.2852057042781868, "learning_rate": 0.003, "loss": 4.0715, "step": 13686 }, { "epoch": 0.13687, "grad_norm": 1.1287402541109783, "learning_rate": 0.003, "loss": 4.0591, "step": 13687 }, { "epoch": 0.13688, "grad_norm": 1.3066891799267129, "learning_rate": 0.003, "loss": 4.0595, "step": 13688 }, { "epoch": 0.13689, "grad_norm": 1.0248433613412835, "learning_rate": 0.003, "loss": 4.0843, "step": 13689 }, { "epoch": 0.1369, "grad_norm": 1.367176090369949, "learning_rate": 0.003, "loss": 4.0923, "step": 13690 }, { "epoch": 0.13691, "grad_norm": 0.9683325719123493, "learning_rate": 0.003, "loss": 4.0833, "step": 13691 }, { "epoch": 0.13692, "grad_norm": 1.441647625810317, "learning_rate": 0.003, "loss": 4.0605, "step": 13692 }, { "epoch": 0.13693, "grad_norm": 1.065749153683736, "learning_rate": 0.003, "loss": 4.0664, "step": 13693 }, { "epoch": 0.13694, "grad_norm": 1.3819720275263285, "learning_rate": 0.003, "loss": 4.0541, "step": 13694 }, { "epoch": 0.13695, "grad_norm": 1.0748305770889302, "learning_rate": 0.003, "loss": 4.0733, "step": 13695 }, { "epoch": 0.13696, "grad_norm": 1.3677968293134606, "learning_rate": 0.003, "loss": 4.0771, "step": 13696 }, { "epoch": 0.13697, "grad_norm": 1.147089897423671, "learning_rate": 0.003, "loss": 4.0507, "step": 13697 }, { "epoch": 0.13698, "grad_norm": 1.235718274909368, "learning_rate": 0.003, "loss": 4.0563, "step": 13698 }, { "epoch": 0.13699, "grad_norm": 1.0225627832242097, "learning_rate": 0.003, "loss": 4.0749, "step": 13699 }, { "epoch": 0.137, "grad_norm": 1.1623363764699286, "learning_rate": 0.003, "loss": 4.0561, "step": 13700 }, { "epoch": 0.13701, "grad_norm": 1.0890333717262264, "learning_rate": 0.003, "loss": 4.0963, "step": 13701 }, { "epoch": 0.13702, "grad_norm": 1.1302786432260996, "learning_rate": 0.003, "loss": 4.051, "step": 13702 }, { "epoch": 0.13703, "grad_norm": 1.0776517926981157, "learning_rate": 0.003, "loss": 4.0438, "step": 13703 }, { "epoch": 0.13704, "grad_norm": 1.3785458069212284, "learning_rate": 0.003, "loss": 4.0742, "step": 13704 }, { "epoch": 0.13705, "grad_norm": 1.2512617040619354, "learning_rate": 0.003, "loss": 4.0758, "step": 13705 }, { "epoch": 0.13706, "grad_norm": 1.3928377499783793, "learning_rate": 0.003, "loss": 4.0683, "step": 13706 }, { "epoch": 0.13707, "grad_norm": 1.118559237908763, "learning_rate": 0.003, "loss": 4.0636, "step": 13707 }, { "epoch": 0.13708, "grad_norm": 1.2029639746066683, "learning_rate": 0.003, "loss": 4.0609, "step": 13708 }, { "epoch": 0.13709, "grad_norm": 1.1127653339477992, "learning_rate": 0.003, "loss": 4.0536, "step": 13709 }, { "epoch": 0.1371, "grad_norm": 1.2691378804015563, "learning_rate": 0.003, "loss": 4.0328, "step": 13710 }, { "epoch": 0.13711, "grad_norm": 1.0512236630876264, "learning_rate": 0.003, "loss": 4.0568, "step": 13711 }, { "epoch": 0.13712, "grad_norm": 1.4292441372063016, "learning_rate": 0.003, "loss": 4.0477, "step": 13712 }, { "epoch": 0.13713, "grad_norm": 0.9978745751173643, "learning_rate": 0.003, "loss": 4.0701, "step": 13713 }, { "epoch": 0.13714, "grad_norm": 1.2432364841852932, "learning_rate": 0.003, "loss": 4.059, "step": 13714 }, { "epoch": 0.13715, "grad_norm": 1.297845034586291, "learning_rate": 0.003, "loss": 4.0863, "step": 13715 }, { "epoch": 0.13716, "grad_norm": 1.2900468090400596, "learning_rate": 0.003, "loss": 4.0408, "step": 13716 }, { "epoch": 0.13717, "grad_norm": 1.181269532330925, "learning_rate": 0.003, "loss": 4.0661, "step": 13717 }, { "epoch": 0.13718, "grad_norm": 1.329669457108706, "learning_rate": 0.003, "loss": 4.0851, "step": 13718 }, { "epoch": 0.13719, "grad_norm": 1.3658069391182663, "learning_rate": 0.003, "loss": 4.0657, "step": 13719 }, { "epoch": 0.1372, "grad_norm": 1.1011609524331176, "learning_rate": 0.003, "loss": 4.0636, "step": 13720 }, { "epoch": 0.13721, "grad_norm": 1.069565039739558, "learning_rate": 0.003, "loss": 4.0526, "step": 13721 }, { "epoch": 0.13722, "grad_norm": 1.0644531780194006, "learning_rate": 0.003, "loss": 4.0474, "step": 13722 }, { "epoch": 0.13723, "grad_norm": 1.4266250579306097, "learning_rate": 0.003, "loss": 4.0718, "step": 13723 }, { "epoch": 0.13724, "grad_norm": 1.3837474539387862, "learning_rate": 0.003, "loss": 4.0812, "step": 13724 }, { "epoch": 0.13725, "grad_norm": 1.2266512686805247, "learning_rate": 0.003, "loss": 4.0909, "step": 13725 }, { "epoch": 0.13726, "grad_norm": 1.0802335853435896, "learning_rate": 0.003, "loss": 4.0712, "step": 13726 }, { "epoch": 0.13727, "grad_norm": 1.2520063776990553, "learning_rate": 0.003, "loss": 4.0836, "step": 13727 }, { "epoch": 0.13728, "grad_norm": 1.1506232520668933, "learning_rate": 0.003, "loss": 4.0888, "step": 13728 }, { "epoch": 0.13729, "grad_norm": 1.191036533468555, "learning_rate": 0.003, "loss": 4.055, "step": 13729 }, { "epoch": 0.1373, "grad_norm": 1.0785639678596188, "learning_rate": 0.003, "loss": 4.068, "step": 13730 }, { "epoch": 0.13731, "grad_norm": 1.3274351259060906, "learning_rate": 0.003, "loss": 4.0675, "step": 13731 }, { "epoch": 0.13732, "grad_norm": 1.1462342271727504, "learning_rate": 0.003, "loss": 4.0276, "step": 13732 }, { "epoch": 0.13733, "grad_norm": 1.1965092703547526, "learning_rate": 0.003, "loss": 4.0802, "step": 13733 }, { "epoch": 0.13734, "grad_norm": 1.0113223792794093, "learning_rate": 0.003, "loss": 4.0575, "step": 13734 }, { "epoch": 0.13735, "grad_norm": 1.4309755113711486, "learning_rate": 0.003, "loss": 4.078, "step": 13735 }, { "epoch": 0.13736, "grad_norm": 1.2237061269968448, "learning_rate": 0.003, "loss": 4.0708, "step": 13736 }, { "epoch": 0.13737, "grad_norm": 1.2061844744891848, "learning_rate": 0.003, "loss": 4.0575, "step": 13737 }, { "epoch": 0.13738, "grad_norm": 1.1160780969915798, "learning_rate": 0.003, "loss": 4.062, "step": 13738 }, { "epoch": 0.13739, "grad_norm": 1.1317129219921174, "learning_rate": 0.003, "loss": 4.0579, "step": 13739 }, { "epoch": 0.1374, "grad_norm": 1.2730856773538404, "learning_rate": 0.003, "loss": 4.0541, "step": 13740 }, { "epoch": 0.13741, "grad_norm": 1.2716542917352498, "learning_rate": 0.003, "loss": 4.0537, "step": 13741 }, { "epoch": 0.13742, "grad_norm": 1.1083900184558646, "learning_rate": 0.003, "loss": 4.0508, "step": 13742 }, { "epoch": 0.13743, "grad_norm": 1.3454270303632545, "learning_rate": 0.003, "loss": 4.0366, "step": 13743 }, { "epoch": 0.13744, "grad_norm": 1.034111525761471, "learning_rate": 0.003, "loss": 4.0366, "step": 13744 }, { "epoch": 0.13745, "grad_norm": 1.2830839448043023, "learning_rate": 0.003, "loss": 4.0525, "step": 13745 }, { "epoch": 0.13746, "grad_norm": 1.0135350503384342, "learning_rate": 0.003, "loss": 4.061, "step": 13746 }, { "epoch": 0.13747, "grad_norm": 1.2599057190334695, "learning_rate": 0.003, "loss": 4.0505, "step": 13747 }, { "epoch": 0.13748, "grad_norm": 1.140178222134933, "learning_rate": 0.003, "loss": 4.1005, "step": 13748 }, { "epoch": 0.13749, "grad_norm": 1.2357363339772103, "learning_rate": 0.003, "loss": 4.0582, "step": 13749 }, { "epoch": 0.1375, "grad_norm": 1.072744201221382, "learning_rate": 0.003, "loss": 4.0795, "step": 13750 }, { "epoch": 0.13751, "grad_norm": 1.1219993471230616, "learning_rate": 0.003, "loss": 4.0924, "step": 13751 }, { "epoch": 0.13752, "grad_norm": 1.0795350179180665, "learning_rate": 0.003, "loss": 4.0352, "step": 13752 }, { "epoch": 0.13753, "grad_norm": 1.2898162353050031, "learning_rate": 0.003, "loss": 4.0568, "step": 13753 }, { "epoch": 0.13754, "grad_norm": 1.2443028895509332, "learning_rate": 0.003, "loss": 4.0783, "step": 13754 }, { "epoch": 0.13755, "grad_norm": 1.196620422488927, "learning_rate": 0.003, "loss": 4.0713, "step": 13755 }, { "epoch": 0.13756, "grad_norm": 1.0545696215897438, "learning_rate": 0.003, "loss": 4.054, "step": 13756 }, { "epoch": 0.13757, "grad_norm": 1.5430740459539571, "learning_rate": 0.003, "loss": 4.0629, "step": 13757 }, { "epoch": 0.13758, "grad_norm": 1.055078720882959, "learning_rate": 0.003, "loss": 4.0714, "step": 13758 }, { "epoch": 0.13759, "grad_norm": 1.4273948140349615, "learning_rate": 0.003, "loss": 4.0511, "step": 13759 }, { "epoch": 0.1376, "grad_norm": 1.1561993353954598, "learning_rate": 0.003, "loss": 4.0887, "step": 13760 }, { "epoch": 0.13761, "grad_norm": 1.0385151262168988, "learning_rate": 0.003, "loss": 4.0286, "step": 13761 }, { "epoch": 0.13762, "grad_norm": 1.0751004141967082, "learning_rate": 0.003, "loss": 4.0411, "step": 13762 }, { "epoch": 0.13763, "grad_norm": 1.4191592640008177, "learning_rate": 0.003, "loss": 4.1066, "step": 13763 }, { "epoch": 0.13764, "grad_norm": 1.284386984507721, "learning_rate": 0.003, "loss": 4.094, "step": 13764 }, { "epoch": 0.13765, "grad_norm": 1.092072644588252, "learning_rate": 0.003, "loss": 4.067, "step": 13765 }, { "epoch": 0.13766, "grad_norm": 1.3050467595815576, "learning_rate": 0.003, "loss": 4.0439, "step": 13766 }, { "epoch": 0.13767, "grad_norm": 1.0362345927564356, "learning_rate": 0.003, "loss": 4.0455, "step": 13767 }, { "epoch": 0.13768, "grad_norm": 1.1589269088035632, "learning_rate": 0.003, "loss": 4.0692, "step": 13768 }, { "epoch": 0.13769, "grad_norm": 0.940109702074667, "learning_rate": 0.003, "loss": 4.0491, "step": 13769 }, { "epoch": 0.1377, "grad_norm": 1.2589329742785462, "learning_rate": 0.003, "loss": 4.0679, "step": 13770 }, { "epoch": 0.13771, "grad_norm": 1.1183296378587286, "learning_rate": 0.003, "loss": 4.076, "step": 13771 }, { "epoch": 0.13772, "grad_norm": 1.2373485396105985, "learning_rate": 0.003, "loss": 4.0813, "step": 13772 }, { "epoch": 0.13773, "grad_norm": 1.1013092832854794, "learning_rate": 0.003, "loss": 4.0381, "step": 13773 }, { "epoch": 0.13774, "grad_norm": 1.1323220409061479, "learning_rate": 0.003, "loss": 4.0664, "step": 13774 }, { "epoch": 0.13775, "grad_norm": 1.2697176050393486, "learning_rate": 0.003, "loss": 4.0911, "step": 13775 }, { "epoch": 0.13776, "grad_norm": 1.2289730726276897, "learning_rate": 0.003, "loss": 4.0139, "step": 13776 }, { "epoch": 0.13777, "grad_norm": 1.0859061023105696, "learning_rate": 0.003, "loss": 4.0421, "step": 13777 }, { "epoch": 0.13778, "grad_norm": 1.431775187944301, "learning_rate": 0.003, "loss": 4.042, "step": 13778 }, { "epoch": 0.13779, "grad_norm": 1.023077064814121, "learning_rate": 0.003, "loss": 4.0386, "step": 13779 }, { "epoch": 0.1378, "grad_norm": 1.4857316253141342, "learning_rate": 0.003, "loss": 4.0735, "step": 13780 }, { "epoch": 0.13781, "grad_norm": 0.9571007767783862, "learning_rate": 0.003, "loss": 4.0723, "step": 13781 }, { "epoch": 0.13782, "grad_norm": 1.242294225250045, "learning_rate": 0.003, "loss": 4.0583, "step": 13782 }, { "epoch": 0.13783, "grad_norm": 1.2180590672415628, "learning_rate": 0.003, "loss": 4.1137, "step": 13783 }, { "epoch": 0.13784, "grad_norm": 1.321609816960609, "learning_rate": 0.003, "loss": 4.0815, "step": 13784 }, { "epoch": 0.13785, "grad_norm": 1.0398449333073807, "learning_rate": 0.003, "loss": 4.0678, "step": 13785 }, { "epoch": 0.13786, "grad_norm": 1.2166767032863475, "learning_rate": 0.003, "loss": 4.0523, "step": 13786 }, { "epoch": 0.13787, "grad_norm": 1.4130608082527525, "learning_rate": 0.003, "loss": 4.0756, "step": 13787 }, { "epoch": 0.13788, "grad_norm": 1.102039258721679, "learning_rate": 0.003, "loss": 4.0378, "step": 13788 }, { "epoch": 0.13789, "grad_norm": 1.2791281383107909, "learning_rate": 0.003, "loss": 4.0949, "step": 13789 }, { "epoch": 0.1379, "grad_norm": 1.0757574152524256, "learning_rate": 0.003, "loss": 4.0558, "step": 13790 }, { "epoch": 0.13791, "grad_norm": 1.2327339175446506, "learning_rate": 0.003, "loss": 4.0844, "step": 13791 }, { "epoch": 0.13792, "grad_norm": 1.4027215777211544, "learning_rate": 0.003, "loss": 4.0729, "step": 13792 }, { "epoch": 0.13793, "grad_norm": 1.3955735712252815, "learning_rate": 0.003, "loss": 4.0767, "step": 13793 }, { "epoch": 0.13794, "grad_norm": 1.279540475431449, "learning_rate": 0.003, "loss": 4.0975, "step": 13794 }, { "epoch": 0.13795, "grad_norm": 1.2178098654806002, "learning_rate": 0.003, "loss": 4.0792, "step": 13795 }, { "epoch": 0.13796, "grad_norm": 1.0539717585069517, "learning_rate": 0.003, "loss": 4.0515, "step": 13796 }, { "epoch": 0.13797, "grad_norm": 1.1084300591943956, "learning_rate": 0.003, "loss": 4.0785, "step": 13797 }, { "epoch": 0.13798, "grad_norm": 1.1013695496932099, "learning_rate": 0.003, "loss": 4.0597, "step": 13798 }, { "epoch": 0.13799, "grad_norm": 1.2816063137585847, "learning_rate": 0.003, "loss": 4.0574, "step": 13799 }, { "epoch": 0.138, "grad_norm": 1.2913325263741893, "learning_rate": 0.003, "loss": 4.0748, "step": 13800 }, { "epoch": 0.13801, "grad_norm": 1.0391772873458964, "learning_rate": 0.003, "loss": 4.0503, "step": 13801 }, { "epoch": 0.13802, "grad_norm": 0.976617087548208, "learning_rate": 0.003, "loss": 4.0533, "step": 13802 }, { "epoch": 0.13803, "grad_norm": 1.1665271871136662, "learning_rate": 0.003, "loss": 4.0714, "step": 13803 }, { "epoch": 0.13804, "grad_norm": 1.0006403922231308, "learning_rate": 0.003, "loss": 4.0892, "step": 13804 }, { "epoch": 0.13805, "grad_norm": 1.3645898219329173, "learning_rate": 0.003, "loss": 4.0547, "step": 13805 }, { "epoch": 0.13806, "grad_norm": 1.3277857270183633, "learning_rate": 0.003, "loss": 4.0595, "step": 13806 }, { "epoch": 0.13807, "grad_norm": 1.0644790355228766, "learning_rate": 0.003, "loss": 4.0755, "step": 13807 }, { "epoch": 0.13808, "grad_norm": 1.6318741175584281, "learning_rate": 0.003, "loss": 4.0865, "step": 13808 }, { "epoch": 0.13809, "grad_norm": 0.7879688329732388, "learning_rate": 0.003, "loss": 4.0608, "step": 13809 }, { "epoch": 0.1381, "grad_norm": 1.0661121011213612, "learning_rate": 0.003, "loss": 4.0457, "step": 13810 }, { "epoch": 0.13811, "grad_norm": 1.4052756870562733, "learning_rate": 0.003, "loss": 4.0975, "step": 13811 }, { "epoch": 0.13812, "grad_norm": 1.0490862665519338, "learning_rate": 0.003, "loss": 4.0807, "step": 13812 }, { "epoch": 0.13813, "grad_norm": 1.3622389610202286, "learning_rate": 0.003, "loss": 4.0659, "step": 13813 }, { "epoch": 0.13814, "grad_norm": 1.0425545423063145, "learning_rate": 0.003, "loss": 4.0853, "step": 13814 }, { "epoch": 0.13815, "grad_norm": 1.2575941945973617, "learning_rate": 0.003, "loss": 4.0451, "step": 13815 }, { "epoch": 0.13816, "grad_norm": 1.2265374473742845, "learning_rate": 0.003, "loss": 4.0653, "step": 13816 }, { "epoch": 0.13817, "grad_norm": 1.038745981724648, "learning_rate": 0.003, "loss": 4.0733, "step": 13817 }, { "epoch": 0.13818, "grad_norm": 1.4731932890584123, "learning_rate": 0.003, "loss": 4.0619, "step": 13818 }, { "epoch": 0.13819, "grad_norm": 0.870903805105368, "learning_rate": 0.003, "loss": 4.0549, "step": 13819 }, { "epoch": 0.1382, "grad_norm": 1.2175904467110854, "learning_rate": 0.003, "loss": 4.0685, "step": 13820 }, { "epoch": 0.13821, "grad_norm": 1.1400424745695283, "learning_rate": 0.003, "loss": 4.0775, "step": 13821 }, { "epoch": 0.13822, "grad_norm": 1.1995296242121614, "learning_rate": 0.003, "loss": 4.0914, "step": 13822 }, { "epoch": 0.13823, "grad_norm": 1.5178825762562025, "learning_rate": 0.003, "loss": 4.0731, "step": 13823 }, { "epoch": 0.13824, "grad_norm": 1.2214546895126566, "learning_rate": 0.003, "loss": 4.0925, "step": 13824 }, { "epoch": 0.13825, "grad_norm": 1.2664357163506683, "learning_rate": 0.003, "loss": 4.0623, "step": 13825 }, { "epoch": 0.13826, "grad_norm": 1.0589329402026078, "learning_rate": 0.003, "loss": 4.0746, "step": 13826 }, { "epoch": 0.13827, "grad_norm": 1.1789523161858053, "learning_rate": 0.003, "loss": 4.0809, "step": 13827 }, { "epoch": 0.13828, "grad_norm": 1.5475286557737047, "learning_rate": 0.003, "loss": 4.0511, "step": 13828 }, { "epoch": 0.13829, "grad_norm": 0.8351369152104988, "learning_rate": 0.003, "loss": 4.0317, "step": 13829 }, { "epoch": 0.1383, "grad_norm": 0.9956043388064251, "learning_rate": 0.003, "loss": 4.0567, "step": 13830 }, { "epoch": 0.13831, "grad_norm": 1.3769018795018921, "learning_rate": 0.003, "loss": 4.0451, "step": 13831 }, { "epoch": 0.13832, "grad_norm": 1.1845831230408836, "learning_rate": 0.003, "loss": 4.0589, "step": 13832 }, { "epoch": 0.13833, "grad_norm": 1.2332693899003906, "learning_rate": 0.003, "loss": 4.0568, "step": 13833 }, { "epoch": 0.13834, "grad_norm": 1.1534843571421134, "learning_rate": 0.003, "loss": 4.0695, "step": 13834 }, { "epoch": 0.13835, "grad_norm": 1.14869899572321, "learning_rate": 0.003, "loss": 4.0571, "step": 13835 }, { "epoch": 0.13836, "grad_norm": 1.0403880611532428, "learning_rate": 0.003, "loss": 4.0679, "step": 13836 }, { "epoch": 0.13837, "grad_norm": 1.130107909605646, "learning_rate": 0.003, "loss": 4.0737, "step": 13837 }, { "epoch": 0.13838, "grad_norm": 1.1789301956378286, "learning_rate": 0.003, "loss": 4.0275, "step": 13838 }, { "epoch": 0.13839, "grad_norm": 1.0178110819703394, "learning_rate": 0.003, "loss": 4.0761, "step": 13839 }, { "epoch": 0.1384, "grad_norm": 1.3602920796271976, "learning_rate": 0.003, "loss": 4.0789, "step": 13840 }, { "epoch": 0.13841, "grad_norm": 1.086448902654459, "learning_rate": 0.003, "loss": 4.0538, "step": 13841 }, { "epoch": 0.13842, "grad_norm": 1.1826934581231108, "learning_rate": 0.003, "loss": 4.0657, "step": 13842 }, { "epoch": 0.13843, "grad_norm": 1.2588360759412796, "learning_rate": 0.003, "loss": 4.0711, "step": 13843 }, { "epoch": 0.13844, "grad_norm": 1.0960611948512418, "learning_rate": 0.003, "loss": 4.0557, "step": 13844 }, { "epoch": 0.13845, "grad_norm": 1.066118581321973, "learning_rate": 0.003, "loss": 4.0722, "step": 13845 }, { "epoch": 0.13846, "grad_norm": 1.238514866519139, "learning_rate": 0.003, "loss": 4.087, "step": 13846 }, { "epoch": 0.13847, "grad_norm": 1.3378615113114039, "learning_rate": 0.003, "loss": 4.044, "step": 13847 }, { "epoch": 0.13848, "grad_norm": 1.107126510372777, "learning_rate": 0.003, "loss": 4.0656, "step": 13848 }, { "epoch": 0.13849, "grad_norm": 1.1923907811098338, "learning_rate": 0.003, "loss": 4.0674, "step": 13849 }, { "epoch": 0.1385, "grad_norm": 1.2036075929545467, "learning_rate": 0.003, "loss": 4.0643, "step": 13850 }, { "epoch": 0.13851, "grad_norm": 1.3265920242843754, "learning_rate": 0.003, "loss": 4.0469, "step": 13851 }, { "epoch": 0.13852, "grad_norm": 1.0783781712805554, "learning_rate": 0.003, "loss": 4.0664, "step": 13852 }, { "epoch": 0.13853, "grad_norm": 1.300679877981243, "learning_rate": 0.003, "loss": 4.0657, "step": 13853 }, { "epoch": 0.13854, "grad_norm": 0.979441386732072, "learning_rate": 0.003, "loss": 4.0753, "step": 13854 }, { "epoch": 0.13855, "grad_norm": 1.2626027757975171, "learning_rate": 0.003, "loss": 4.0585, "step": 13855 }, { "epoch": 0.13856, "grad_norm": 1.0324943518145326, "learning_rate": 0.003, "loss": 4.0855, "step": 13856 }, { "epoch": 0.13857, "grad_norm": 1.2208611478898446, "learning_rate": 0.003, "loss": 4.0598, "step": 13857 }, { "epoch": 0.13858, "grad_norm": 1.4150408537958241, "learning_rate": 0.003, "loss": 4.0655, "step": 13858 }, { "epoch": 0.13859, "grad_norm": 1.1934719812277916, "learning_rate": 0.003, "loss": 4.073, "step": 13859 }, { "epoch": 0.1386, "grad_norm": 1.1097521653095541, "learning_rate": 0.003, "loss": 4.0565, "step": 13860 }, { "epoch": 0.13861, "grad_norm": 1.0591341409277695, "learning_rate": 0.003, "loss": 4.0789, "step": 13861 }, { "epoch": 0.13862, "grad_norm": 1.311078165099448, "learning_rate": 0.003, "loss": 4.0905, "step": 13862 }, { "epoch": 0.13863, "grad_norm": 1.075891909671555, "learning_rate": 0.003, "loss": 4.1015, "step": 13863 }, { "epoch": 0.13864, "grad_norm": 1.266066950103621, "learning_rate": 0.003, "loss": 4.0906, "step": 13864 }, { "epoch": 0.13865, "grad_norm": 1.3438837530062315, "learning_rate": 0.003, "loss": 4.0843, "step": 13865 }, { "epoch": 0.13866, "grad_norm": 0.934191573347225, "learning_rate": 0.003, "loss": 4.0415, "step": 13866 }, { "epoch": 0.13867, "grad_norm": 1.2095735572592086, "learning_rate": 0.003, "loss": 4.0553, "step": 13867 }, { "epoch": 0.13868, "grad_norm": 1.3806980271358968, "learning_rate": 0.003, "loss": 4.078, "step": 13868 }, { "epoch": 0.13869, "grad_norm": 0.8708374532186117, "learning_rate": 0.003, "loss": 4.043, "step": 13869 }, { "epoch": 0.1387, "grad_norm": 1.1395576474767661, "learning_rate": 0.003, "loss": 4.0346, "step": 13870 }, { "epoch": 0.13871, "grad_norm": 1.1544323853403469, "learning_rate": 0.003, "loss": 4.0395, "step": 13871 }, { "epoch": 0.13872, "grad_norm": 1.1905739230752463, "learning_rate": 0.003, "loss": 4.0535, "step": 13872 }, { "epoch": 0.13873, "grad_norm": 1.23088045055871, "learning_rate": 0.003, "loss": 4.0555, "step": 13873 }, { "epoch": 0.13874, "grad_norm": 1.0749892857645922, "learning_rate": 0.003, "loss": 4.0606, "step": 13874 }, { "epoch": 0.13875, "grad_norm": 1.6296366511374096, "learning_rate": 0.003, "loss": 4.0736, "step": 13875 }, { "epoch": 0.13876, "grad_norm": 1.0406320129171167, "learning_rate": 0.003, "loss": 4.0795, "step": 13876 }, { "epoch": 0.13877, "grad_norm": 1.5276874377748664, "learning_rate": 0.003, "loss": 4.063, "step": 13877 }, { "epoch": 0.13878, "grad_norm": 0.9085018293664214, "learning_rate": 0.003, "loss": 4.0367, "step": 13878 }, { "epoch": 0.13879, "grad_norm": 1.198722324986746, "learning_rate": 0.003, "loss": 4.0896, "step": 13879 }, { "epoch": 0.1388, "grad_norm": 1.4900679753925414, "learning_rate": 0.003, "loss": 4.085, "step": 13880 }, { "epoch": 0.13881, "grad_norm": 0.9588686404421287, "learning_rate": 0.003, "loss": 4.0613, "step": 13881 }, { "epoch": 0.13882, "grad_norm": 1.016556436833725, "learning_rate": 0.003, "loss": 4.0595, "step": 13882 }, { "epoch": 0.13883, "grad_norm": 1.3591825267573006, "learning_rate": 0.003, "loss": 4.0577, "step": 13883 }, { "epoch": 0.13884, "grad_norm": 0.9969580757172046, "learning_rate": 0.003, "loss": 4.0851, "step": 13884 }, { "epoch": 0.13885, "grad_norm": 1.376113519006804, "learning_rate": 0.003, "loss": 4.113, "step": 13885 }, { "epoch": 0.13886, "grad_norm": 1.130956073216537, "learning_rate": 0.003, "loss": 4.0358, "step": 13886 }, { "epoch": 0.13887, "grad_norm": 1.3813816134763195, "learning_rate": 0.003, "loss": 4.0609, "step": 13887 }, { "epoch": 0.13888, "grad_norm": 1.0133877966264333, "learning_rate": 0.003, "loss": 4.0731, "step": 13888 }, { "epoch": 0.13889, "grad_norm": 1.3471834838681127, "learning_rate": 0.003, "loss": 4.0841, "step": 13889 }, { "epoch": 0.1389, "grad_norm": 1.0201858599279923, "learning_rate": 0.003, "loss": 4.077, "step": 13890 }, { "epoch": 0.13891, "grad_norm": 1.2828078609119384, "learning_rate": 0.003, "loss": 4.0808, "step": 13891 }, { "epoch": 0.13892, "grad_norm": 1.11863619216386, "learning_rate": 0.003, "loss": 4.0849, "step": 13892 }, { "epoch": 0.13893, "grad_norm": 1.603158276512601, "learning_rate": 0.003, "loss": 4.0621, "step": 13893 }, { "epoch": 0.13894, "grad_norm": 1.1969476992562909, "learning_rate": 0.003, "loss": 4.0628, "step": 13894 }, { "epoch": 0.13895, "grad_norm": 1.5423365160858793, "learning_rate": 0.003, "loss": 4.07, "step": 13895 }, { "epoch": 0.13896, "grad_norm": 1.077928852196919, "learning_rate": 0.003, "loss": 4.0909, "step": 13896 }, { "epoch": 0.13897, "grad_norm": 1.026728940213821, "learning_rate": 0.003, "loss": 4.0662, "step": 13897 }, { "epoch": 0.13898, "grad_norm": 1.2677260856492953, "learning_rate": 0.003, "loss": 4.0494, "step": 13898 }, { "epoch": 0.13899, "grad_norm": 1.2272126840482172, "learning_rate": 0.003, "loss": 4.0626, "step": 13899 }, { "epoch": 0.139, "grad_norm": 1.287321008248166, "learning_rate": 0.003, "loss": 4.097, "step": 13900 }, { "epoch": 0.13901, "grad_norm": 1.0518284235161448, "learning_rate": 0.003, "loss": 4.0864, "step": 13901 }, { "epoch": 0.13902, "grad_norm": 1.1496082951367534, "learning_rate": 0.003, "loss": 4.0683, "step": 13902 }, { "epoch": 0.13903, "grad_norm": 1.0881877055239924, "learning_rate": 0.003, "loss": 4.0402, "step": 13903 }, { "epoch": 0.13904, "grad_norm": 1.0367206105761448, "learning_rate": 0.003, "loss": 4.0563, "step": 13904 }, { "epoch": 0.13905, "grad_norm": 1.3221464375084346, "learning_rate": 0.003, "loss": 4.0858, "step": 13905 }, { "epoch": 0.13906, "grad_norm": 1.0677619426077882, "learning_rate": 0.003, "loss": 4.025, "step": 13906 }, { "epoch": 0.13907, "grad_norm": 1.4128252294951043, "learning_rate": 0.003, "loss": 4.0623, "step": 13907 }, { "epoch": 0.13908, "grad_norm": 0.880496247210434, "learning_rate": 0.003, "loss": 4.0491, "step": 13908 }, { "epoch": 0.13909, "grad_norm": 1.0017699963385105, "learning_rate": 0.003, "loss": 4.0531, "step": 13909 }, { "epoch": 0.1391, "grad_norm": 1.2636365868887092, "learning_rate": 0.003, "loss": 4.0761, "step": 13910 }, { "epoch": 0.13911, "grad_norm": 1.225275355478596, "learning_rate": 0.003, "loss": 4.0618, "step": 13911 }, { "epoch": 0.13912, "grad_norm": 1.3060925427906085, "learning_rate": 0.003, "loss": 4.0839, "step": 13912 }, { "epoch": 0.13913, "grad_norm": 1.2074339416830029, "learning_rate": 0.003, "loss": 4.0592, "step": 13913 }, { "epoch": 0.13914, "grad_norm": 1.2732718468049447, "learning_rate": 0.003, "loss": 4.0726, "step": 13914 }, { "epoch": 0.13915, "grad_norm": 1.059581908738672, "learning_rate": 0.003, "loss": 4.0706, "step": 13915 }, { "epoch": 0.13916, "grad_norm": 1.3835418931483388, "learning_rate": 0.003, "loss": 4.0777, "step": 13916 }, { "epoch": 0.13917, "grad_norm": 1.286371224707917, "learning_rate": 0.003, "loss": 4.0873, "step": 13917 }, { "epoch": 0.13918, "grad_norm": 1.1491760972411582, "learning_rate": 0.003, "loss": 4.0892, "step": 13918 }, { "epoch": 0.13919, "grad_norm": 1.0381404485028898, "learning_rate": 0.003, "loss": 4.0493, "step": 13919 }, { "epoch": 0.1392, "grad_norm": 1.2936934515418033, "learning_rate": 0.003, "loss": 4.0722, "step": 13920 }, { "epoch": 0.13921, "grad_norm": 1.1336491850307007, "learning_rate": 0.003, "loss": 4.0683, "step": 13921 }, { "epoch": 0.13922, "grad_norm": 1.032283544406887, "learning_rate": 0.003, "loss": 4.0528, "step": 13922 }, { "epoch": 0.13923, "grad_norm": 1.1350421647630262, "learning_rate": 0.003, "loss": 4.0769, "step": 13923 }, { "epoch": 0.13924, "grad_norm": 1.2181477345896563, "learning_rate": 0.003, "loss": 4.0665, "step": 13924 }, { "epoch": 0.13925, "grad_norm": 1.2935392461959319, "learning_rate": 0.003, "loss": 4.0662, "step": 13925 }, { "epoch": 0.13926, "grad_norm": 1.580693431878095, "learning_rate": 0.003, "loss": 4.0792, "step": 13926 }, { "epoch": 0.13927, "grad_norm": 1.0568452383884184, "learning_rate": 0.003, "loss": 4.0577, "step": 13927 }, { "epoch": 0.13928, "grad_norm": 1.1826999735195407, "learning_rate": 0.003, "loss": 4.0787, "step": 13928 }, { "epoch": 0.13929, "grad_norm": 1.4389017526386207, "learning_rate": 0.003, "loss": 4.0497, "step": 13929 }, { "epoch": 0.1393, "grad_norm": 1.0264331694045916, "learning_rate": 0.003, "loss": 4.0642, "step": 13930 }, { "epoch": 0.13931, "grad_norm": 1.3517532712896185, "learning_rate": 0.003, "loss": 4.0048, "step": 13931 }, { "epoch": 0.13932, "grad_norm": 1.06929122881059, "learning_rate": 0.003, "loss": 4.0245, "step": 13932 }, { "epoch": 0.13933, "grad_norm": 1.362157964664218, "learning_rate": 0.003, "loss": 4.0407, "step": 13933 }, { "epoch": 0.13934, "grad_norm": 1.0026056408211013, "learning_rate": 0.003, "loss": 4.0745, "step": 13934 }, { "epoch": 0.13935, "grad_norm": 1.4217178830899353, "learning_rate": 0.003, "loss": 4.075, "step": 13935 }, { "epoch": 0.13936, "grad_norm": 1.038671947066456, "learning_rate": 0.003, "loss": 4.065, "step": 13936 }, { "epoch": 0.13937, "grad_norm": 1.2887008768297814, "learning_rate": 0.003, "loss": 4.048, "step": 13937 }, { "epoch": 0.13938, "grad_norm": 1.2099968579918394, "learning_rate": 0.003, "loss": 4.057, "step": 13938 }, { "epoch": 0.13939, "grad_norm": 1.3522984630469852, "learning_rate": 0.003, "loss": 4.0784, "step": 13939 }, { "epoch": 0.1394, "grad_norm": 1.1961466683874793, "learning_rate": 0.003, "loss": 4.044, "step": 13940 }, { "epoch": 0.13941, "grad_norm": 1.306101262984324, "learning_rate": 0.003, "loss": 4.0685, "step": 13941 }, { "epoch": 0.13942, "grad_norm": 0.9711576861323818, "learning_rate": 0.003, "loss": 4.0676, "step": 13942 }, { "epoch": 0.13943, "grad_norm": 1.1791976699448328, "learning_rate": 0.003, "loss": 4.1138, "step": 13943 }, { "epoch": 0.13944, "grad_norm": 1.0763256910193317, "learning_rate": 0.003, "loss": 4.0836, "step": 13944 }, { "epoch": 0.13945, "grad_norm": 1.190284236924628, "learning_rate": 0.003, "loss": 4.0724, "step": 13945 }, { "epoch": 0.13946, "grad_norm": 1.1505593702223464, "learning_rate": 0.003, "loss": 4.0813, "step": 13946 }, { "epoch": 0.13947, "grad_norm": 1.3241486709444608, "learning_rate": 0.003, "loss": 4.0825, "step": 13947 }, { "epoch": 0.13948, "grad_norm": 1.2436621418152587, "learning_rate": 0.003, "loss": 4.0813, "step": 13948 }, { "epoch": 0.13949, "grad_norm": 1.0489075729453334, "learning_rate": 0.003, "loss": 4.0797, "step": 13949 }, { "epoch": 0.1395, "grad_norm": 1.1671792480190273, "learning_rate": 0.003, "loss": 4.0858, "step": 13950 }, { "epoch": 0.13951, "grad_norm": 1.2678027178461682, "learning_rate": 0.003, "loss": 4.0546, "step": 13951 }, { "epoch": 0.13952, "grad_norm": 1.1699449496959624, "learning_rate": 0.003, "loss": 4.0659, "step": 13952 }, { "epoch": 0.13953, "grad_norm": 1.3098685402383983, "learning_rate": 0.003, "loss": 4.0658, "step": 13953 }, { "epoch": 0.13954, "grad_norm": 1.224454648402645, "learning_rate": 0.003, "loss": 4.0392, "step": 13954 }, { "epoch": 0.13955, "grad_norm": 1.3048386632435085, "learning_rate": 0.003, "loss": 4.0834, "step": 13955 }, { "epoch": 0.13956, "grad_norm": 1.0227877046412943, "learning_rate": 0.003, "loss": 4.0764, "step": 13956 }, { "epoch": 0.13957, "grad_norm": 1.1985956342806694, "learning_rate": 0.003, "loss": 4.089, "step": 13957 }, { "epoch": 0.13958, "grad_norm": 1.183718979418515, "learning_rate": 0.003, "loss": 4.079, "step": 13958 }, { "epoch": 0.13959, "grad_norm": 1.2330220548490956, "learning_rate": 0.003, "loss": 4.0734, "step": 13959 }, { "epoch": 0.1396, "grad_norm": 1.125662126079691, "learning_rate": 0.003, "loss": 4.0536, "step": 13960 }, { "epoch": 0.13961, "grad_norm": 1.285516911003546, "learning_rate": 0.003, "loss": 4.0684, "step": 13961 }, { "epoch": 0.13962, "grad_norm": 1.1264208265619262, "learning_rate": 0.003, "loss": 4.0796, "step": 13962 }, { "epoch": 0.13963, "grad_norm": 1.031640488901549, "learning_rate": 0.003, "loss": 4.07, "step": 13963 }, { "epoch": 0.13964, "grad_norm": 1.2802365534499318, "learning_rate": 0.003, "loss": 4.0532, "step": 13964 }, { "epoch": 0.13965, "grad_norm": 1.1884413630126052, "learning_rate": 0.003, "loss": 4.0412, "step": 13965 }, { "epoch": 0.13966, "grad_norm": 1.1690090473186747, "learning_rate": 0.003, "loss": 4.0655, "step": 13966 }, { "epoch": 0.13967, "grad_norm": 1.1366104707513918, "learning_rate": 0.003, "loss": 4.051, "step": 13967 }, { "epoch": 0.13968, "grad_norm": 1.2198371288381218, "learning_rate": 0.003, "loss": 4.039, "step": 13968 }, { "epoch": 0.13969, "grad_norm": 1.120964014292074, "learning_rate": 0.003, "loss": 4.0402, "step": 13969 }, { "epoch": 0.1397, "grad_norm": 1.6853975112623714, "learning_rate": 0.003, "loss": 4.0553, "step": 13970 }, { "epoch": 0.13971, "grad_norm": 0.8516396043457239, "learning_rate": 0.003, "loss": 4.0816, "step": 13971 }, { "epoch": 0.13972, "grad_norm": 1.3989485640785961, "learning_rate": 0.003, "loss": 4.0833, "step": 13972 }, { "epoch": 0.13973, "grad_norm": 1.2911245355553416, "learning_rate": 0.003, "loss": 4.0943, "step": 13973 }, { "epoch": 0.13974, "grad_norm": 1.3562710136515597, "learning_rate": 0.003, "loss": 4.0785, "step": 13974 }, { "epoch": 0.13975, "grad_norm": 1.0626370665776999, "learning_rate": 0.003, "loss": 4.0542, "step": 13975 }, { "epoch": 0.13976, "grad_norm": 1.3882714398556264, "learning_rate": 0.003, "loss": 4.0477, "step": 13976 }, { "epoch": 0.13977, "grad_norm": 1.07602319752501, "learning_rate": 0.003, "loss": 4.0533, "step": 13977 }, { "epoch": 0.13978, "grad_norm": 1.204983796527482, "learning_rate": 0.003, "loss": 4.1228, "step": 13978 }, { "epoch": 0.13979, "grad_norm": 1.0239793207772239, "learning_rate": 0.003, "loss": 4.068, "step": 13979 }, { "epoch": 0.1398, "grad_norm": 1.4068167574457042, "learning_rate": 0.003, "loss": 4.0774, "step": 13980 }, { "epoch": 0.13981, "grad_norm": 0.9272392246853138, "learning_rate": 0.003, "loss": 4.0661, "step": 13981 }, { "epoch": 0.13982, "grad_norm": 1.2235259694039406, "learning_rate": 0.003, "loss": 4.0743, "step": 13982 }, { "epoch": 0.13983, "grad_norm": 1.0285946068778307, "learning_rate": 0.003, "loss": 4.0826, "step": 13983 }, { "epoch": 0.13984, "grad_norm": 1.3088445090796035, "learning_rate": 0.003, "loss": 4.0652, "step": 13984 }, { "epoch": 0.13985, "grad_norm": 1.0166766015149529, "learning_rate": 0.003, "loss": 4.0696, "step": 13985 }, { "epoch": 0.13986, "grad_norm": 1.2031489036038912, "learning_rate": 0.003, "loss": 4.0903, "step": 13986 }, { "epoch": 0.13987, "grad_norm": 1.0570294895755987, "learning_rate": 0.003, "loss": 4.0418, "step": 13987 }, { "epoch": 0.13988, "grad_norm": 1.4843615987129113, "learning_rate": 0.003, "loss": 4.0735, "step": 13988 }, { "epoch": 0.13989, "grad_norm": 1.172269746382644, "learning_rate": 0.003, "loss": 4.0608, "step": 13989 }, { "epoch": 0.1399, "grad_norm": 1.1866403118247013, "learning_rate": 0.003, "loss": 4.0653, "step": 13990 }, { "epoch": 0.13991, "grad_norm": 1.1538755961087634, "learning_rate": 0.003, "loss": 4.0759, "step": 13991 }, { "epoch": 0.13992, "grad_norm": 1.3588394984928112, "learning_rate": 0.003, "loss": 4.0744, "step": 13992 }, { "epoch": 0.13993, "grad_norm": 1.1448255332881148, "learning_rate": 0.003, "loss": 4.0411, "step": 13993 }, { "epoch": 0.13994, "grad_norm": 1.1296028881145412, "learning_rate": 0.003, "loss": 4.0626, "step": 13994 }, { "epoch": 0.13995, "grad_norm": 1.3123610288025107, "learning_rate": 0.003, "loss": 4.0619, "step": 13995 }, { "epoch": 0.13996, "grad_norm": 1.025069752380935, "learning_rate": 0.003, "loss": 4.0642, "step": 13996 }, { "epoch": 0.13997, "grad_norm": 1.5053800598899543, "learning_rate": 0.003, "loss": 4.0743, "step": 13997 }, { "epoch": 0.13998, "grad_norm": 0.953506370355428, "learning_rate": 0.003, "loss": 4.0718, "step": 13998 }, { "epoch": 0.13999, "grad_norm": 1.0708954418212988, "learning_rate": 0.003, "loss": 4.077, "step": 13999 }, { "epoch": 0.14, "grad_norm": 1.0853424429098648, "learning_rate": 0.003, "loss": 4.0443, "step": 14000 }, { "epoch": 0.14001, "grad_norm": 1.1751322615896895, "learning_rate": 0.003, "loss": 4.0639, "step": 14001 }, { "epoch": 0.14002, "grad_norm": 1.2627706928641533, "learning_rate": 0.003, "loss": 4.0801, "step": 14002 }, { "epoch": 0.14003, "grad_norm": 1.256909647986896, "learning_rate": 0.003, "loss": 4.0822, "step": 14003 }, { "epoch": 0.14004, "grad_norm": 1.2356056386980325, "learning_rate": 0.003, "loss": 4.0686, "step": 14004 }, { "epoch": 0.14005, "grad_norm": 1.1453273029563364, "learning_rate": 0.003, "loss": 4.0628, "step": 14005 }, { "epoch": 0.14006, "grad_norm": 1.2198936454607405, "learning_rate": 0.003, "loss": 4.0699, "step": 14006 }, { "epoch": 0.14007, "grad_norm": 1.1391939945267104, "learning_rate": 0.003, "loss": 4.0789, "step": 14007 }, { "epoch": 0.14008, "grad_norm": 1.3187676811830558, "learning_rate": 0.003, "loss": 4.0845, "step": 14008 }, { "epoch": 0.14009, "grad_norm": 0.9599819243120673, "learning_rate": 0.003, "loss": 4.0648, "step": 14009 }, { "epoch": 0.1401, "grad_norm": 1.2606158811736639, "learning_rate": 0.003, "loss": 4.0797, "step": 14010 }, { "epoch": 0.14011, "grad_norm": 1.3055698944912726, "learning_rate": 0.003, "loss": 4.0494, "step": 14011 }, { "epoch": 0.14012, "grad_norm": 1.2797345567603078, "learning_rate": 0.003, "loss": 4.0699, "step": 14012 }, { "epoch": 0.14013, "grad_norm": 1.4950526637194654, "learning_rate": 0.003, "loss": 4.0645, "step": 14013 }, { "epoch": 0.14014, "grad_norm": 0.9905350914594043, "learning_rate": 0.003, "loss": 4.0585, "step": 14014 }, { "epoch": 0.14015, "grad_norm": 1.2965093091361792, "learning_rate": 0.003, "loss": 4.0533, "step": 14015 }, { "epoch": 0.14016, "grad_norm": 1.2418385758571846, "learning_rate": 0.003, "loss": 4.083, "step": 14016 }, { "epoch": 0.14017, "grad_norm": 1.2348271237915762, "learning_rate": 0.003, "loss": 4.0712, "step": 14017 }, { "epoch": 0.14018, "grad_norm": 1.2745667416649695, "learning_rate": 0.003, "loss": 4.0651, "step": 14018 }, { "epoch": 0.14019, "grad_norm": 1.082411174356095, "learning_rate": 0.003, "loss": 4.075, "step": 14019 }, { "epoch": 0.1402, "grad_norm": 1.2153197263791202, "learning_rate": 0.003, "loss": 4.0306, "step": 14020 }, { "epoch": 0.14021, "grad_norm": 1.0910587248310606, "learning_rate": 0.003, "loss": 4.0784, "step": 14021 }, { "epoch": 0.14022, "grad_norm": 1.2500389515470083, "learning_rate": 0.003, "loss": 4.0431, "step": 14022 }, { "epoch": 0.14023, "grad_norm": 1.0380589880341928, "learning_rate": 0.003, "loss": 4.0553, "step": 14023 }, { "epoch": 0.14024, "grad_norm": 1.4434775858039697, "learning_rate": 0.003, "loss": 4.0846, "step": 14024 }, { "epoch": 0.14025, "grad_norm": 0.9907731511394811, "learning_rate": 0.003, "loss": 4.0404, "step": 14025 }, { "epoch": 0.14026, "grad_norm": 1.4208867005861814, "learning_rate": 0.003, "loss": 4.081, "step": 14026 }, { "epoch": 0.14027, "grad_norm": 1.115862229281504, "learning_rate": 0.003, "loss": 4.0409, "step": 14027 }, { "epoch": 0.14028, "grad_norm": 1.1944116583170195, "learning_rate": 0.003, "loss": 4.0459, "step": 14028 }, { "epoch": 0.14029, "grad_norm": 1.1740283527298874, "learning_rate": 0.003, "loss": 4.0405, "step": 14029 }, { "epoch": 0.1403, "grad_norm": 1.2636790423414466, "learning_rate": 0.003, "loss": 4.0653, "step": 14030 }, { "epoch": 0.14031, "grad_norm": 1.0205070534553105, "learning_rate": 0.003, "loss": 4.0513, "step": 14031 }, { "epoch": 0.14032, "grad_norm": 1.148178017331799, "learning_rate": 0.003, "loss": 4.0588, "step": 14032 }, { "epoch": 0.14033, "grad_norm": 1.101090766483742, "learning_rate": 0.003, "loss": 4.0908, "step": 14033 }, { "epoch": 0.14034, "grad_norm": 1.180918011469458, "learning_rate": 0.003, "loss": 4.0563, "step": 14034 }, { "epoch": 0.14035, "grad_norm": 1.3016727533830463, "learning_rate": 0.003, "loss": 4.0998, "step": 14035 }, { "epoch": 0.14036, "grad_norm": 1.1174539382869852, "learning_rate": 0.003, "loss": 4.0597, "step": 14036 }, { "epoch": 0.14037, "grad_norm": 1.1302127058999685, "learning_rate": 0.003, "loss": 4.0977, "step": 14037 }, { "epoch": 0.14038, "grad_norm": 1.1432418603164374, "learning_rate": 0.003, "loss": 4.1043, "step": 14038 }, { "epoch": 0.14039, "grad_norm": 1.0127791234279393, "learning_rate": 0.003, "loss": 4.0874, "step": 14039 }, { "epoch": 0.1404, "grad_norm": 1.2078468546327734, "learning_rate": 0.003, "loss": 4.0732, "step": 14040 }, { "epoch": 0.14041, "grad_norm": 1.0044245340634046, "learning_rate": 0.003, "loss": 4.0631, "step": 14041 }, { "epoch": 0.14042, "grad_norm": 1.3387862921051845, "learning_rate": 0.003, "loss": 4.0851, "step": 14042 }, { "epoch": 0.14043, "grad_norm": 1.1765964167320597, "learning_rate": 0.003, "loss": 4.0498, "step": 14043 }, { "epoch": 0.14044, "grad_norm": 1.1365245865954814, "learning_rate": 0.003, "loss": 4.0336, "step": 14044 }, { "epoch": 0.14045, "grad_norm": 1.2670225146315648, "learning_rate": 0.003, "loss": 4.0711, "step": 14045 }, { "epoch": 0.14046, "grad_norm": 1.1551713629150766, "learning_rate": 0.003, "loss": 4.0685, "step": 14046 }, { "epoch": 0.14047, "grad_norm": 1.2263739906716984, "learning_rate": 0.003, "loss": 4.0594, "step": 14047 }, { "epoch": 0.14048, "grad_norm": 1.3292121038850169, "learning_rate": 0.003, "loss": 4.0624, "step": 14048 }, { "epoch": 0.14049, "grad_norm": 1.1707901228667954, "learning_rate": 0.003, "loss": 4.0615, "step": 14049 }, { "epoch": 0.1405, "grad_norm": 1.1820577056492885, "learning_rate": 0.003, "loss": 4.0709, "step": 14050 }, { "epoch": 0.14051, "grad_norm": 1.4970989890341049, "learning_rate": 0.003, "loss": 4.0416, "step": 14051 }, { "epoch": 0.14052, "grad_norm": 1.1880382310046214, "learning_rate": 0.003, "loss": 4.0782, "step": 14052 }, { "epoch": 0.14053, "grad_norm": 1.0418475121645348, "learning_rate": 0.003, "loss": 4.0833, "step": 14053 }, { "epoch": 0.14054, "grad_norm": 1.3826547250959087, "learning_rate": 0.003, "loss": 4.0737, "step": 14054 }, { "epoch": 0.14055, "grad_norm": 1.1989247615972918, "learning_rate": 0.003, "loss": 4.0431, "step": 14055 }, { "epoch": 0.14056, "grad_norm": 1.4922771838130722, "learning_rate": 0.003, "loss": 4.0882, "step": 14056 }, { "epoch": 0.14057, "grad_norm": 0.9828593072226139, "learning_rate": 0.003, "loss": 4.0499, "step": 14057 }, { "epoch": 0.14058, "grad_norm": 1.242153543410688, "learning_rate": 0.003, "loss": 4.0895, "step": 14058 }, { "epoch": 0.14059, "grad_norm": 1.2644191548856771, "learning_rate": 0.003, "loss": 4.0897, "step": 14059 }, { "epoch": 0.1406, "grad_norm": 1.0402160183989233, "learning_rate": 0.003, "loss": 4.0511, "step": 14060 }, { "epoch": 0.14061, "grad_norm": 1.4643926891587589, "learning_rate": 0.003, "loss": 4.0781, "step": 14061 }, { "epoch": 0.14062, "grad_norm": 1.2854550032648755, "learning_rate": 0.003, "loss": 4.0697, "step": 14062 }, { "epoch": 0.14063, "grad_norm": 1.1399715066086162, "learning_rate": 0.003, "loss": 4.0664, "step": 14063 }, { "epoch": 0.14064, "grad_norm": 1.2613905152553255, "learning_rate": 0.003, "loss": 4.0868, "step": 14064 }, { "epoch": 0.14065, "grad_norm": 1.109738252666644, "learning_rate": 0.003, "loss": 4.0747, "step": 14065 }, { "epoch": 0.14066, "grad_norm": 1.1536706142441462, "learning_rate": 0.003, "loss": 4.0689, "step": 14066 }, { "epoch": 0.14067, "grad_norm": 1.1686760221894372, "learning_rate": 0.003, "loss": 4.0598, "step": 14067 }, { "epoch": 0.14068, "grad_norm": 1.4431129370117735, "learning_rate": 0.003, "loss": 4.0478, "step": 14068 }, { "epoch": 0.14069, "grad_norm": 1.1529287834339563, "learning_rate": 0.003, "loss": 4.0729, "step": 14069 }, { "epoch": 0.1407, "grad_norm": 1.2272293396189673, "learning_rate": 0.003, "loss": 4.0803, "step": 14070 }, { "epoch": 0.14071, "grad_norm": 1.2113547139493819, "learning_rate": 0.003, "loss": 4.0717, "step": 14071 }, { "epoch": 0.14072, "grad_norm": 1.0976937568041476, "learning_rate": 0.003, "loss": 4.0788, "step": 14072 }, { "epoch": 0.14073, "grad_norm": 1.247524526764744, "learning_rate": 0.003, "loss": 4.0776, "step": 14073 }, { "epoch": 0.14074, "grad_norm": 1.089618174756747, "learning_rate": 0.003, "loss": 4.0771, "step": 14074 }, { "epoch": 0.14075, "grad_norm": 1.700931146159072, "learning_rate": 0.003, "loss": 4.0869, "step": 14075 }, { "epoch": 0.14076, "grad_norm": 1.1481484438903082, "learning_rate": 0.003, "loss": 4.0601, "step": 14076 }, { "epoch": 0.14077, "grad_norm": 1.7389127274356944, "learning_rate": 0.003, "loss": 4.0751, "step": 14077 }, { "epoch": 0.14078, "grad_norm": 0.91208467488463, "learning_rate": 0.003, "loss": 4.0744, "step": 14078 }, { "epoch": 0.14079, "grad_norm": 1.058450877881687, "learning_rate": 0.003, "loss": 4.0394, "step": 14079 }, { "epoch": 0.1408, "grad_norm": 1.5232097577975936, "learning_rate": 0.003, "loss": 4.0731, "step": 14080 }, { "epoch": 0.14081, "grad_norm": 1.1325448640909364, "learning_rate": 0.003, "loss": 4.0586, "step": 14081 }, { "epoch": 0.14082, "grad_norm": 1.3692201824713788, "learning_rate": 0.003, "loss": 4.0833, "step": 14082 }, { "epoch": 0.14083, "grad_norm": 1.2405415707110088, "learning_rate": 0.003, "loss": 4.0832, "step": 14083 }, { "epoch": 0.14084, "grad_norm": 1.2612958054984984, "learning_rate": 0.003, "loss": 4.0311, "step": 14084 }, { "epoch": 0.14085, "grad_norm": 1.0115092048586234, "learning_rate": 0.003, "loss": 4.0526, "step": 14085 }, { "epoch": 0.14086, "grad_norm": 1.1961231758905728, "learning_rate": 0.003, "loss": 4.0606, "step": 14086 }, { "epoch": 0.14087, "grad_norm": 1.1376853253553598, "learning_rate": 0.003, "loss": 4.0751, "step": 14087 }, { "epoch": 0.14088, "grad_norm": 1.2719190721417335, "learning_rate": 0.003, "loss": 4.0771, "step": 14088 }, { "epoch": 0.14089, "grad_norm": 0.9257838863610739, "learning_rate": 0.003, "loss": 4.0628, "step": 14089 }, { "epoch": 0.1409, "grad_norm": 1.0136199776793982, "learning_rate": 0.003, "loss": 4.0809, "step": 14090 }, { "epoch": 0.14091, "grad_norm": 1.1143236991560046, "learning_rate": 0.003, "loss": 4.0539, "step": 14091 }, { "epoch": 0.14092, "grad_norm": 1.4618438256914872, "learning_rate": 0.003, "loss": 4.0905, "step": 14092 }, { "epoch": 0.14093, "grad_norm": 1.2310701034540763, "learning_rate": 0.003, "loss": 4.0658, "step": 14093 }, { "epoch": 0.14094, "grad_norm": 1.3875878617357045, "learning_rate": 0.003, "loss": 4.0401, "step": 14094 }, { "epoch": 0.14095, "grad_norm": 1.2401622519093114, "learning_rate": 0.003, "loss": 4.084, "step": 14095 }, { "epoch": 0.14096, "grad_norm": 1.100162709342287, "learning_rate": 0.003, "loss": 4.0701, "step": 14096 }, { "epoch": 0.14097, "grad_norm": 1.3648806080597096, "learning_rate": 0.003, "loss": 4.0636, "step": 14097 }, { "epoch": 0.14098, "grad_norm": 0.9119748179992018, "learning_rate": 0.003, "loss": 4.0723, "step": 14098 }, { "epoch": 0.14099, "grad_norm": 1.101659224646554, "learning_rate": 0.003, "loss": 4.0764, "step": 14099 }, { "epoch": 0.141, "grad_norm": 1.2489497461395942, "learning_rate": 0.003, "loss": 4.0979, "step": 14100 }, { "epoch": 0.14101, "grad_norm": 1.087271878838567, "learning_rate": 0.003, "loss": 4.0635, "step": 14101 }, { "epoch": 0.14102, "grad_norm": 1.538799319935321, "learning_rate": 0.003, "loss": 4.0701, "step": 14102 }, { "epoch": 0.14103, "grad_norm": 1.0619706577492025, "learning_rate": 0.003, "loss": 4.073, "step": 14103 }, { "epoch": 0.14104, "grad_norm": 1.3883605928555496, "learning_rate": 0.003, "loss": 4.0533, "step": 14104 }, { "epoch": 0.14105, "grad_norm": 1.091052895958119, "learning_rate": 0.003, "loss": 4.0417, "step": 14105 }, { "epoch": 0.14106, "grad_norm": 1.26460143028129, "learning_rate": 0.003, "loss": 4.0652, "step": 14106 }, { "epoch": 0.14107, "grad_norm": 1.0241565414578302, "learning_rate": 0.003, "loss": 4.0498, "step": 14107 }, { "epoch": 0.14108, "grad_norm": 1.4581418517518614, "learning_rate": 0.003, "loss": 4.0762, "step": 14108 }, { "epoch": 0.14109, "grad_norm": 1.020482183692946, "learning_rate": 0.003, "loss": 4.073, "step": 14109 }, { "epoch": 0.1411, "grad_norm": 1.2695426847696474, "learning_rate": 0.003, "loss": 4.0535, "step": 14110 }, { "epoch": 0.14111, "grad_norm": 1.042396582666057, "learning_rate": 0.003, "loss": 4.0753, "step": 14111 }, { "epoch": 0.14112, "grad_norm": 1.3471325666922835, "learning_rate": 0.003, "loss": 4.0791, "step": 14112 }, { "epoch": 0.14113, "grad_norm": 1.1013474928848515, "learning_rate": 0.003, "loss": 4.0456, "step": 14113 }, { "epoch": 0.14114, "grad_norm": 1.3049329414297008, "learning_rate": 0.003, "loss": 4.0709, "step": 14114 }, { "epoch": 0.14115, "grad_norm": 1.165601398901598, "learning_rate": 0.003, "loss": 4.0695, "step": 14115 }, { "epoch": 0.14116, "grad_norm": 1.2845637290855145, "learning_rate": 0.003, "loss": 4.0724, "step": 14116 }, { "epoch": 0.14117, "grad_norm": 1.215084813674362, "learning_rate": 0.003, "loss": 4.0568, "step": 14117 }, { "epoch": 0.14118, "grad_norm": 1.267146370467284, "learning_rate": 0.003, "loss": 4.0661, "step": 14118 }, { "epoch": 0.14119, "grad_norm": 1.2088876244998399, "learning_rate": 0.003, "loss": 4.0667, "step": 14119 }, { "epoch": 0.1412, "grad_norm": 1.0370689493716259, "learning_rate": 0.003, "loss": 4.0489, "step": 14120 }, { "epoch": 0.14121, "grad_norm": 1.3567784912838432, "learning_rate": 0.003, "loss": 4.0664, "step": 14121 }, { "epoch": 0.14122, "grad_norm": 1.1358911454763512, "learning_rate": 0.003, "loss": 4.0864, "step": 14122 }, { "epoch": 0.14123, "grad_norm": 1.444138133727849, "learning_rate": 0.003, "loss": 4.0633, "step": 14123 }, { "epoch": 0.14124, "grad_norm": 1.019323927670304, "learning_rate": 0.003, "loss": 4.0654, "step": 14124 }, { "epoch": 0.14125, "grad_norm": 1.0620644030242081, "learning_rate": 0.003, "loss": 4.0425, "step": 14125 }, { "epoch": 0.14126, "grad_norm": 1.3640724702723905, "learning_rate": 0.003, "loss": 4.0178, "step": 14126 }, { "epoch": 0.14127, "grad_norm": 1.062696357746494, "learning_rate": 0.003, "loss": 4.0646, "step": 14127 }, { "epoch": 0.14128, "grad_norm": 1.2543281286826446, "learning_rate": 0.003, "loss": 4.0689, "step": 14128 }, { "epoch": 0.14129, "grad_norm": 1.2307060795050326, "learning_rate": 0.003, "loss": 4.0549, "step": 14129 }, { "epoch": 0.1413, "grad_norm": 1.0676327695640522, "learning_rate": 0.003, "loss": 4.0831, "step": 14130 }, { "epoch": 0.14131, "grad_norm": 1.1381901901492002, "learning_rate": 0.003, "loss": 4.0597, "step": 14131 }, { "epoch": 0.14132, "grad_norm": 1.1621490624517714, "learning_rate": 0.003, "loss": 4.094, "step": 14132 }, { "epoch": 0.14133, "grad_norm": 0.9949905353372355, "learning_rate": 0.003, "loss": 4.0321, "step": 14133 }, { "epoch": 0.14134, "grad_norm": 1.3635458799768592, "learning_rate": 0.003, "loss": 4.0968, "step": 14134 }, { "epoch": 0.14135, "grad_norm": 0.9973050565209519, "learning_rate": 0.003, "loss": 4.0836, "step": 14135 }, { "epoch": 0.14136, "grad_norm": 1.5237844040165016, "learning_rate": 0.003, "loss": 4.0687, "step": 14136 }, { "epoch": 0.14137, "grad_norm": 1.2408921889432498, "learning_rate": 0.003, "loss": 4.0813, "step": 14137 }, { "epoch": 0.14138, "grad_norm": 1.1249081956948144, "learning_rate": 0.003, "loss": 4.0704, "step": 14138 }, { "epoch": 0.14139, "grad_norm": 1.089470920215599, "learning_rate": 0.003, "loss": 4.0622, "step": 14139 }, { "epoch": 0.1414, "grad_norm": 1.766406695149215, "learning_rate": 0.003, "loss": 4.0496, "step": 14140 }, { "epoch": 0.14141, "grad_norm": 0.9474434010605183, "learning_rate": 0.003, "loss": 4.0726, "step": 14141 }, { "epoch": 0.14142, "grad_norm": 1.1543755248262677, "learning_rate": 0.003, "loss": 4.0592, "step": 14142 }, { "epoch": 0.14143, "grad_norm": 1.15551118814518, "learning_rate": 0.003, "loss": 4.0521, "step": 14143 }, { "epoch": 0.14144, "grad_norm": 1.0949168441885477, "learning_rate": 0.003, "loss": 4.0512, "step": 14144 }, { "epoch": 0.14145, "grad_norm": 1.2158154688963252, "learning_rate": 0.003, "loss": 4.0817, "step": 14145 }, { "epoch": 0.14146, "grad_norm": 1.1930896585816888, "learning_rate": 0.003, "loss": 4.0826, "step": 14146 }, { "epoch": 0.14147, "grad_norm": 1.3306246715158585, "learning_rate": 0.003, "loss": 4.0807, "step": 14147 }, { "epoch": 0.14148, "grad_norm": 0.9676846760335293, "learning_rate": 0.003, "loss": 4.0707, "step": 14148 }, { "epoch": 0.14149, "grad_norm": 1.212415931754474, "learning_rate": 0.003, "loss": 4.0791, "step": 14149 }, { "epoch": 0.1415, "grad_norm": 1.452342777360195, "learning_rate": 0.003, "loss": 4.1047, "step": 14150 }, { "epoch": 0.14151, "grad_norm": 1.2710544608686976, "learning_rate": 0.003, "loss": 4.0847, "step": 14151 }, { "epoch": 0.14152, "grad_norm": 1.0482634458726003, "learning_rate": 0.003, "loss": 4.0473, "step": 14152 }, { "epoch": 0.14153, "grad_norm": 1.4379195293470144, "learning_rate": 0.003, "loss": 4.0468, "step": 14153 }, { "epoch": 0.14154, "grad_norm": 1.191609048243655, "learning_rate": 0.003, "loss": 4.0428, "step": 14154 }, { "epoch": 0.14155, "grad_norm": 1.3787592896305192, "learning_rate": 0.003, "loss": 4.0796, "step": 14155 }, { "epoch": 0.14156, "grad_norm": 0.9893806391740821, "learning_rate": 0.003, "loss": 4.0469, "step": 14156 }, { "epoch": 0.14157, "grad_norm": 1.272420136765873, "learning_rate": 0.003, "loss": 4.0882, "step": 14157 }, { "epoch": 0.14158, "grad_norm": 1.2617920027622527, "learning_rate": 0.003, "loss": 4.0835, "step": 14158 }, { "epoch": 0.14159, "grad_norm": 1.1700761289369743, "learning_rate": 0.003, "loss": 4.073, "step": 14159 }, { "epoch": 0.1416, "grad_norm": 1.122186994806511, "learning_rate": 0.003, "loss": 4.0472, "step": 14160 }, { "epoch": 0.14161, "grad_norm": 1.1404942329267413, "learning_rate": 0.003, "loss": 4.1088, "step": 14161 }, { "epoch": 0.14162, "grad_norm": 1.085778004031023, "learning_rate": 0.003, "loss": 4.0478, "step": 14162 }, { "epoch": 0.14163, "grad_norm": 1.0694076701689075, "learning_rate": 0.003, "loss": 4.0609, "step": 14163 }, { "epoch": 0.14164, "grad_norm": 1.1191905529863895, "learning_rate": 0.003, "loss": 4.0545, "step": 14164 }, { "epoch": 0.14165, "grad_norm": 1.5149121636829188, "learning_rate": 0.003, "loss": 4.0962, "step": 14165 }, { "epoch": 0.14166, "grad_norm": 1.1636360103645447, "learning_rate": 0.003, "loss": 4.085, "step": 14166 }, { "epoch": 0.14167, "grad_norm": 1.2826324722743676, "learning_rate": 0.003, "loss": 4.049, "step": 14167 }, { "epoch": 0.14168, "grad_norm": 1.1435295912881631, "learning_rate": 0.003, "loss": 4.0708, "step": 14168 }, { "epoch": 0.14169, "grad_norm": 1.0783640644185406, "learning_rate": 0.003, "loss": 4.0736, "step": 14169 }, { "epoch": 0.1417, "grad_norm": 1.2931780213121669, "learning_rate": 0.003, "loss": 4.0731, "step": 14170 }, { "epoch": 0.14171, "grad_norm": 1.1902891825456416, "learning_rate": 0.003, "loss": 4.0754, "step": 14171 }, { "epoch": 0.14172, "grad_norm": 1.284651646093003, "learning_rate": 0.003, "loss": 4.0519, "step": 14172 }, { "epoch": 0.14173, "grad_norm": 1.1197725075637848, "learning_rate": 0.003, "loss": 4.0792, "step": 14173 }, { "epoch": 0.14174, "grad_norm": 1.32959635212142, "learning_rate": 0.003, "loss": 4.0783, "step": 14174 }, { "epoch": 0.14175, "grad_norm": 1.0929271176927622, "learning_rate": 0.003, "loss": 4.087, "step": 14175 }, { "epoch": 0.14176, "grad_norm": 1.107336305247538, "learning_rate": 0.003, "loss": 4.0567, "step": 14176 }, { "epoch": 0.14177, "grad_norm": 1.0345033569018824, "learning_rate": 0.003, "loss": 4.1002, "step": 14177 }, { "epoch": 0.14178, "grad_norm": 1.119196119261943, "learning_rate": 0.003, "loss": 4.0315, "step": 14178 }, { "epoch": 0.14179, "grad_norm": 1.4634545139397872, "learning_rate": 0.003, "loss": 4.0827, "step": 14179 }, { "epoch": 0.1418, "grad_norm": 1.027755596902636, "learning_rate": 0.003, "loss": 4.0533, "step": 14180 }, { "epoch": 0.14181, "grad_norm": 1.6493733952615102, "learning_rate": 0.003, "loss": 4.0825, "step": 14181 }, { "epoch": 0.14182, "grad_norm": 1.199597131516992, "learning_rate": 0.003, "loss": 4.0703, "step": 14182 }, { "epoch": 0.14183, "grad_norm": 1.2057489298199282, "learning_rate": 0.003, "loss": 4.0657, "step": 14183 }, { "epoch": 0.14184, "grad_norm": 1.2224922049388296, "learning_rate": 0.003, "loss": 4.0888, "step": 14184 }, { "epoch": 0.14185, "grad_norm": 1.1297684255979763, "learning_rate": 0.003, "loss": 4.084, "step": 14185 }, { "epoch": 0.14186, "grad_norm": 1.2586919988023766, "learning_rate": 0.003, "loss": 4.0723, "step": 14186 }, { "epoch": 0.14187, "grad_norm": 1.0496742596269901, "learning_rate": 0.003, "loss": 4.0779, "step": 14187 }, { "epoch": 0.14188, "grad_norm": 1.2778464566434349, "learning_rate": 0.003, "loss": 4.0516, "step": 14188 }, { "epoch": 0.14189, "grad_norm": 1.091293365195764, "learning_rate": 0.003, "loss": 4.0606, "step": 14189 }, { "epoch": 0.1419, "grad_norm": 1.181619627239842, "learning_rate": 0.003, "loss": 4.0723, "step": 14190 }, { "epoch": 0.14191, "grad_norm": 1.1808431279254916, "learning_rate": 0.003, "loss": 4.0687, "step": 14191 }, { "epoch": 0.14192, "grad_norm": 1.1471303523369158, "learning_rate": 0.003, "loss": 4.1122, "step": 14192 }, { "epoch": 0.14193, "grad_norm": 1.1733350139049805, "learning_rate": 0.003, "loss": 4.0586, "step": 14193 }, { "epoch": 0.14194, "grad_norm": 1.2063705463703653, "learning_rate": 0.003, "loss": 4.1014, "step": 14194 }, { "epoch": 0.14195, "grad_norm": 1.4150917924345123, "learning_rate": 0.003, "loss": 4.0916, "step": 14195 }, { "epoch": 0.14196, "grad_norm": 1.1644947084397448, "learning_rate": 0.003, "loss": 4.0704, "step": 14196 }, { "epoch": 0.14197, "grad_norm": 1.3360296380209244, "learning_rate": 0.003, "loss": 4.0759, "step": 14197 }, { "epoch": 0.14198, "grad_norm": 1.3234276563046434, "learning_rate": 0.003, "loss": 4.0708, "step": 14198 }, { "epoch": 0.14199, "grad_norm": 1.655669932559523, "learning_rate": 0.003, "loss": 4.0765, "step": 14199 }, { "epoch": 0.142, "grad_norm": 0.8645260924147421, "learning_rate": 0.003, "loss": 4.0692, "step": 14200 }, { "epoch": 0.14201, "grad_norm": 1.231467452337542, "learning_rate": 0.003, "loss": 4.0409, "step": 14201 }, { "epoch": 0.14202, "grad_norm": 1.3296047105728226, "learning_rate": 0.003, "loss": 4.0794, "step": 14202 }, { "epoch": 0.14203, "grad_norm": 1.0361973528335804, "learning_rate": 0.003, "loss": 4.0618, "step": 14203 }, { "epoch": 0.14204, "grad_norm": 1.2130661454494625, "learning_rate": 0.003, "loss": 4.063, "step": 14204 }, { "epoch": 0.14205, "grad_norm": 1.0396218101989512, "learning_rate": 0.003, "loss": 4.0606, "step": 14205 }, { "epoch": 0.14206, "grad_norm": 1.1370721026997253, "learning_rate": 0.003, "loss": 4.0375, "step": 14206 }, { "epoch": 0.14207, "grad_norm": 1.0778254858539933, "learning_rate": 0.003, "loss": 4.0855, "step": 14207 }, { "epoch": 0.14208, "grad_norm": 1.2817697240467347, "learning_rate": 0.003, "loss": 4.0484, "step": 14208 }, { "epoch": 0.14209, "grad_norm": 1.571986516035277, "learning_rate": 0.003, "loss": 4.0648, "step": 14209 }, { "epoch": 0.1421, "grad_norm": 1.2369047244446083, "learning_rate": 0.003, "loss": 4.0603, "step": 14210 }, { "epoch": 0.14211, "grad_norm": 1.122695466629707, "learning_rate": 0.003, "loss": 4.0386, "step": 14211 }, { "epoch": 0.14212, "grad_norm": 1.2753495336204912, "learning_rate": 0.003, "loss": 4.0592, "step": 14212 }, { "epoch": 0.14213, "grad_norm": 1.2800777890439599, "learning_rate": 0.003, "loss": 4.0713, "step": 14213 }, { "epoch": 0.14214, "grad_norm": 1.1483107106387056, "learning_rate": 0.003, "loss": 4.0587, "step": 14214 }, { "epoch": 0.14215, "grad_norm": 1.1398680420847587, "learning_rate": 0.003, "loss": 4.0454, "step": 14215 }, { "epoch": 0.14216, "grad_norm": 1.1207686970460085, "learning_rate": 0.003, "loss": 4.0653, "step": 14216 }, { "epoch": 0.14217, "grad_norm": 1.0481411536484901, "learning_rate": 0.003, "loss": 4.0574, "step": 14217 }, { "epoch": 0.14218, "grad_norm": 1.2157894765444972, "learning_rate": 0.003, "loss": 4.058, "step": 14218 }, { "epoch": 0.14219, "grad_norm": 1.175092816881649, "learning_rate": 0.003, "loss": 4.0494, "step": 14219 }, { "epoch": 0.1422, "grad_norm": 1.0675976028832213, "learning_rate": 0.003, "loss": 4.0388, "step": 14220 }, { "epoch": 0.14221, "grad_norm": 1.4064689804811845, "learning_rate": 0.003, "loss": 4.053, "step": 14221 }, { "epoch": 0.14222, "grad_norm": 1.31466024641552, "learning_rate": 0.003, "loss": 4.0925, "step": 14222 }, { "epoch": 0.14223, "grad_norm": 1.071967340068421, "learning_rate": 0.003, "loss": 4.064, "step": 14223 }, { "epoch": 0.14224, "grad_norm": 1.3461964597737213, "learning_rate": 0.003, "loss": 4.025, "step": 14224 }, { "epoch": 0.14225, "grad_norm": 0.8939320576619639, "learning_rate": 0.003, "loss": 4.0486, "step": 14225 }, { "epoch": 0.14226, "grad_norm": 1.1882540605637555, "learning_rate": 0.003, "loss": 4.0601, "step": 14226 }, { "epoch": 0.14227, "grad_norm": 1.1193579100454676, "learning_rate": 0.003, "loss": 4.0581, "step": 14227 }, { "epoch": 0.14228, "grad_norm": 1.5711398723592591, "learning_rate": 0.003, "loss": 4.0556, "step": 14228 }, { "epoch": 0.14229, "grad_norm": 1.071306204940862, "learning_rate": 0.003, "loss": 4.0572, "step": 14229 }, { "epoch": 0.1423, "grad_norm": 1.199926718935403, "learning_rate": 0.003, "loss": 4.0405, "step": 14230 }, { "epoch": 0.14231, "grad_norm": 1.2968772399813384, "learning_rate": 0.003, "loss": 4.0833, "step": 14231 }, { "epoch": 0.14232, "grad_norm": 1.2110709858588478, "learning_rate": 0.003, "loss": 4.059, "step": 14232 }, { "epoch": 0.14233, "grad_norm": 1.076850702138636, "learning_rate": 0.003, "loss": 4.0924, "step": 14233 }, { "epoch": 0.14234, "grad_norm": 1.2137894987587472, "learning_rate": 0.003, "loss": 4.0489, "step": 14234 }, { "epoch": 0.14235, "grad_norm": 0.9700130811788708, "learning_rate": 0.003, "loss": 4.0555, "step": 14235 }, { "epoch": 0.14236, "grad_norm": 1.2284258014361296, "learning_rate": 0.003, "loss": 4.0493, "step": 14236 }, { "epoch": 0.14237, "grad_norm": 1.0625057397406377, "learning_rate": 0.003, "loss": 4.0547, "step": 14237 }, { "epoch": 0.14238, "grad_norm": 1.3454226153455804, "learning_rate": 0.003, "loss": 4.0839, "step": 14238 }, { "epoch": 0.14239, "grad_norm": 1.0430985593988913, "learning_rate": 0.003, "loss": 4.0956, "step": 14239 }, { "epoch": 0.1424, "grad_norm": 1.2549151547175872, "learning_rate": 0.003, "loss": 4.0699, "step": 14240 }, { "epoch": 0.14241, "grad_norm": 1.105889056985785, "learning_rate": 0.003, "loss": 4.0932, "step": 14241 }, { "epoch": 0.14242, "grad_norm": 1.2534217071630724, "learning_rate": 0.003, "loss": 4.0594, "step": 14242 }, { "epoch": 0.14243, "grad_norm": 1.490414740246, "learning_rate": 0.003, "loss": 4.0697, "step": 14243 }, { "epoch": 0.14244, "grad_norm": 1.1990047833883828, "learning_rate": 0.003, "loss": 4.0601, "step": 14244 }, { "epoch": 0.14245, "grad_norm": 1.2362137951101504, "learning_rate": 0.003, "loss": 4.0667, "step": 14245 }, { "epoch": 0.14246, "grad_norm": 1.2196722993987894, "learning_rate": 0.003, "loss": 4.077, "step": 14246 }, { "epoch": 0.14247, "grad_norm": 1.2809217326424946, "learning_rate": 0.003, "loss": 4.0559, "step": 14247 }, { "epoch": 0.14248, "grad_norm": 1.0180462757614577, "learning_rate": 0.003, "loss": 4.0628, "step": 14248 }, { "epoch": 0.14249, "grad_norm": 1.293475424515683, "learning_rate": 0.003, "loss": 4.0219, "step": 14249 }, { "epoch": 0.1425, "grad_norm": 1.0327402727410269, "learning_rate": 0.003, "loss": 4.0333, "step": 14250 }, { "epoch": 0.14251, "grad_norm": 1.464622467671724, "learning_rate": 0.003, "loss": 4.0731, "step": 14251 }, { "epoch": 0.14252, "grad_norm": 1.1079023727495274, "learning_rate": 0.003, "loss": 4.0754, "step": 14252 }, { "epoch": 0.14253, "grad_norm": 1.1738022294070674, "learning_rate": 0.003, "loss": 4.0969, "step": 14253 }, { "epoch": 0.14254, "grad_norm": 1.1749821721531346, "learning_rate": 0.003, "loss": 4.0572, "step": 14254 }, { "epoch": 0.14255, "grad_norm": 1.155641367291191, "learning_rate": 0.003, "loss": 4.0554, "step": 14255 }, { "epoch": 0.14256, "grad_norm": 1.4718813972164393, "learning_rate": 0.003, "loss": 4.058, "step": 14256 }, { "epoch": 0.14257, "grad_norm": 1.1861704431299545, "learning_rate": 0.003, "loss": 4.0713, "step": 14257 }, { "epoch": 0.14258, "grad_norm": 1.3177585554808537, "learning_rate": 0.003, "loss": 4.0544, "step": 14258 }, { "epoch": 0.14259, "grad_norm": 1.2620995819080834, "learning_rate": 0.003, "loss": 4.0724, "step": 14259 }, { "epoch": 0.1426, "grad_norm": 1.0326300047426755, "learning_rate": 0.003, "loss": 4.0562, "step": 14260 }, { "epoch": 0.14261, "grad_norm": 1.445018691737256, "learning_rate": 0.003, "loss": 4.0655, "step": 14261 }, { "epoch": 0.14262, "grad_norm": 0.9199967785969957, "learning_rate": 0.003, "loss": 4.0415, "step": 14262 }, { "epoch": 0.14263, "grad_norm": 1.2215931395439386, "learning_rate": 0.003, "loss": 4.0745, "step": 14263 }, { "epoch": 0.14264, "grad_norm": 1.1709372196894714, "learning_rate": 0.003, "loss": 4.0599, "step": 14264 }, { "epoch": 0.14265, "grad_norm": 1.232346357358036, "learning_rate": 0.003, "loss": 4.0866, "step": 14265 }, { "epoch": 0.14266, "grad_norm": 1.3335474977640058, "learning_rate": 0.003, "loss": 4.0842, "step": 14266 }, { "epoch": 0.14267, "grad_norm": 1.0131622771709141, "learning_rate": 0.003, "loss": 4.0812, "step": 14267 }, { "epoch": 0.14268, "grad_norm": 1.5364409505944565, "learning_rate": 0.003, "loss": 4.0498, "step": 14268 }, { "epoch": 0.14269, "grad_norm": 0.8789420939411684, "learning_rate": 0.003, "loss": 4.0899, "step": 14269 }, { "epoch": 0.1427, "grad_norm": 1.2583510459232634, "learning_rate": 0.003, "loss": 4.0838, "step": 14270 }, { "epoch": 0.14271, "grad_norm": 1.0326989710149879, "learning_rate": 0.003, "loss": 4.0617, "step": 14271 }, { "epoch": 0.14272, "grad_norm": 1.2486351214299949, "learning_rate": 0.003, "loss": 4.0903, "step": 14272 }, { "epoch": 0.14273, "grad_norm": 1.0690904813796103, "learning_rate": 0.003, "loss": 4.0642, "step": 14273 }, { "epoch": 0.14274, "grad_norm": 1.2556787600710217, "learning_rate": 0.003, "loss": 4.0773, "step": 14274 }, { "epoch": 0.14275, "grad_norm": 1.3684099959816238, "learning_rate": 0.003, "loss": 4.0737, "step": 14275 }, { "epoch": 0.14276, "grad_norm": 1.062957034604356, "learning_rate": 0.003, "loss": 4.0724, "step": 14276 }, { "epoch": 0.14277, "grad_norm": 1.258125035002094, "learning_rate": 0.003, "loss": 4.0841, "step": 14277 }, { "epoch": 0.14278, "grad_norm": 1.07996135461999, "learning_rate": 0.003, "loss": 4.0445, "step": 14278 }, { "epoch": 0.14279, "grad_norm": 1.3070816281444275, "learning_rate": 0.003, "loss": 4.0948, "step": 14279 }, { "epoch": 0.1428, "grad_norm": 1.0147384420988028, "learning_rate": 0.003, "loss": 4.0847, "step": 14280 }, { "epoch": 0.14281, "grad_norm": 1.233363304571996, "learning_rate": 0.003, "loss": 4.0714, "step": 14281 }, { "epoch": 0.14282, "grad_norm": 1.1563333766935289, "learning_rate": 0.003, "loss": 4.0742, "step": 14282 }, { "epoch": 0.14283, "grad_norm": 1.1194835920626935, "learning_rate": 0.003, "loss": 4.0413, "step": 14283 }, { "epoch": 0.14284, "grad_norm": 1.2053877604243153, "learning_rate": 0.003, "loss": 4.0409, "step": 14284 }, { "epoch": 0.14285, "grad_norm": 1.1034257992664742, "learning_rate": 0.003, "loss": 4.048, "step": 14285 }, { "epoch": 0.14286, "grad_norm": 1.3143939008549936, "learning_rate": 0.003, "loss": 4.0561, "step": 14286 }, { "epoch": 0.14287, "grad_norm": 1.3366976752252706, "learning_rate": 0.003, "loss": 4.0505, "step": 14287 }, { "epoch": 0.14288, "grad_norm": 1.1310654369469988, "learning_rate": 0.003, "loss": 4.0524, "step": 14288 }, { "epoch": 0.14289, "grad_norm": 1.2265929427028606, "learning_rate": 0.003, "loss": 4.0574, "step": 14289 }, { "epoch": 0.1429, "grad_norm": 1.2046181480804943, "learning_rate": 0.003, "loss": 4.0319, "step": 14290 }, { "epoch": 0.14291, "grad_norm": 1.1558269131326242, "learning_rate": 0.003, "loss": 4.066, "step": 14291 }, { "epoch": 0.14292, "grad_norm": 1.0649877913921748, "learning_rate": 0.003, "loss": 4.0598, "step": 14292 }, { "epoch": 0.14293, "grad_norm": 1.4234779215332185, "learning_rate": 0.003, "loss": 4.0426, "step": 14293 }, { "epoch": 0.14294, "grad_norm": 1.082763152735558, "learning_rate": 0.003, "loss": 4.0882, "step": 14294 }, { "epoch": 0.14295, "grad_norm": 1.173731979995803, "learning_rate": 0.003, "loss": 4.0659, "step": 14295 }, { "epoch": 0.14296, "grad_norm": 1.1618342398990031, "learning_rate": 0.003, "loss": 4.0642, "step": 14296 }, { "epoch": 0.14297, "grad_norm": 1.398643937597955, "learning_rate": 0.003, "loss": 4.0813, "step": 14297 }, { "epoch": 0.14298, "grad_norm": 1.1297895969250424, "learning_rate": 0.003, "loss": 4.0454, "step": 14298 }, { "epoch": 0.14299, "grad_norm": 1.0644719244957406, "learning_rate": 0.003, "loss": 4.07, "step": 14299 }, { "epoch": 0.143, "grad_norm": 1.1861565471135649, "learning_rate": 0.003, "loss": 4.0871, "step": 14300 }, { "epoch": 0.14301, "grad_norm": 1.0575205910749435, "learning_rate": 0.003, "loss": 4.0767, "step": 14301 }, { "epoch": 0.14302, "grad_norm": 1.3843489024489457, "learning_rate": 0.003, "loss": 4.0346, "step": 14302 }, { "epoch": 0.14303, "grad_norm": 0.9955134745867775, "learning_rate": 0.003, "loss": 4.0688, "step": 14303 }, { "epoch": 0.14304, "grad_norm": 1.363422538376115, "learning_rate": 0.003, "loss": 4.0663, "step": 14304 }, { "epoch": 0.14305, "grad_norm": 1.1986656569895822, "learning_rate": 0.003, "loss": 4.0607, "step": 14305 }, { "epoch": 0.14306, "grad_norm": 1.2880639273858205, "learning_rate": 0.003, "loss": 4.0611, "step": 14306 }, { "epoch": 0.14307, "grad_norm": 1.1620878358597082, "learning_rate": 0.003, "loss": 4.0804, "step": 14307 }, { "epoch": 0.14308, "grad_norm": 1.1767622378675848, "learning_rate": 0.003, "loss": 4.0845, "step": 14308 }, { "epoch": 0.14309, "grad_norm": 1.19414029814871, "learning_rate": 0.003, "loss": 4.0905, "step": 14309 }, { "epoch": 0.1431, "grad_norm": 1.1167610645284485, "learning_rate": 0.003, "loss": 4.0509, "step": 14310 }, { "epoch": 0.14311, "grad_norm": 1.1773561919335593, "learning_rate": 0.003, "loss": 4.0562, "step": 14311 }, { "epoch": 0.14312, "grad_norm": 1.1455104222251367, "learning_rate": 0.003, "loss": 4.0524, "step": 14312 }, { "epoch": 0.14313, "grad_norm": 1.1359866568220862, "learning_rate": 0.003, "loss": 4.0697, "step": 14313 }, { "epoch": 0.14314, "grad_norm": 1.420754054680384, "learning_rate": 0.003, "loss": 4.0574, "step": 14314 }, { "epoch": 0.14315, "grad_norm": 1.0646963383903258, "learning_rate": 0.003, "loss": 4.0649, "step": 14315 }, { "epoch": 0.14316, "grad_norm": 1.2763403852729251, "learning_rate": 0.003, "loss": 4.0594, "step": 14316 }, { "epoch": 0.14317, "grad_norm": 1.094937836624008, "learning_rate": 0.003, "loss": 4.0604, "step": 14317 }, { "epoch": 0.14318, "grad_norm": 1.2742582445618376, "learning_rate": 0.003, "loss": 4.0697, "step": 14318 }, { "epoch": 0.14319, "grad_norm": 1.2092886240243337, "learning_rate": 0.003, "loss": 4.0746, "step": 14319 }, { "epoch": 0.1432, "grad_norm": 1.1382478374397993, "learning_rate": 0.003, "loss": 4.0725, "step": 14320 }, { "epoch": 0.14321, "grad_norm": 1.2418323837035394, "learning_rate": 0.003, "loss": 4.04, "step": 14321 }, { "epoch": 0.14322, "grad_norm": 1.1518922494326984, "learning_rate": 0.003, "loss": 4.0657, "step": 14322 }, { "epoch": 0.14323, "grad_norm": 1.1400534779865883, "learning_rate": 0.003, "loss": 4.036, "step": 14323 }, { "epoch": 0.14324, "grad_norm": 1.0692597026270678, "learning_rate": 0.003, "loss": 4.0593, "step": 14324 }, { "epoch": 0.14325, "grad_norm": 1.2840580875617695, "learning_rate": 0.003, "loss": 4.0367, "step": 14325 }, { "epoch": 0.14326, "grad_norm": 1.2816960237064363, "learning_rate": 0.003, "loss": 4.0836, "step": 14326 }, { "epoch": 0.14327, "grad_norm": 1.1003045057564285, "learning_rate": 0.003, "loss": 4.0586, "step": 14327 }, { "epoch": 0.14328, "grad_norm": 1.3623564374963049, "learning_rate": 0.003, "loss": 4.1072, "step": 14328 }, { "epoch": 0.14329, "grad_norm": 1.1472976404966218, "learning_rate": 0.003, "loss": 4.0512, "step": 14329 }, { "epoch": 0.1433, "grad_norm": 1.3164838448329237, "learning_rate": 0.003, "loss": 4.0772, "step": 14330 }, { "epoch": 0.14331, "grad_norm": 1.1756095255143983, "learning_rate": 0.003, "loss": 4.0495, "step": 14331 }, { "epoch": 0.14332, "grad_norm": 1.3444697447616594, "learning_rate": 0.003, "loss": 4.0644, "step": 14332 }, { "epoch": 0.14333, "grad_norm": 1.0284512757583038, "learning_rate": 0.003, "loss": 4.0525, "step": 14333 }, { "epoch": 0.14334, "grad_norm": 1.1859319506845603, "learning_rate": 0.003, "loss": 4.0669, "step": 14334 }, { "epoch": 0.14335, "grad_norm": 1.3257963016987404, "learning_rate": 0.003, "loss": 4.0518, "step": 14335 }, { "epoch": 0.14336, "grad_norm": 1.1837739932179037, "learning_rate": 0.003, "loss": 4.0656, "step": 14336 }, { "epoch": 0.14337, "grad_norm": 1.5042978304051147, "learning_rate": 0.003, "loss": 4.0832, "step": 14337 }, { "epoch": 0.14338, "grad_norm": 1.0186872922432264, "learning_rate": 0.003, "loss": 4.0515, "step": 14338 }, { "epoch": 0.14339, "grad_norm": 1.1228133682719914, "learning_rate": 0.003, "loss": 4.0735, "step": 14339 }, { "epoch": 0.1434, "grad_norm": 1.2933482445652345, "learning_rate": 0.003, "loss": 4.0521, "step": 14340 }, { "epoch": 0.14341, "grad_norm": 1.2985215677679838, "learning_rate": 0.003, "loss": 4.0884, "step": 14341 }, { "epoch": 0.14342, "grad_norm": 1.2023861963450682, "learning_rate": 0.003, "loss": 4.0606, "step": 14342 }, { "epoch": 0.14343, "grad_norm": 1.1193205301312157, "learning_rate": 0.003, "loss": 4.1037, "step": 14343 }, { "epoch": 0.14344, "grad_norm": 1.2517525510158198, "learning_rate": 0.003, "loss": 4.1045, "step": 14344 }, { "epoch": 0.14345, "grad_norm": 0.987524666521823, "learning_rate": 0.003, "loss": 4.0608, "step": 14345 }, { "epoch": 0.14346, "grad_norm": 1.4945311055814365, "learning_rate": 0.003, "loss": 4.0749, "step": 14346 }, { "epoch": 0.14347, "grad_norm": 0.925123815103442, "learning_rate": 0.003, "loss": 4.08, "step": 14347 }, { "epoch": 0.14348, "grad_norm": 1.2561977804757412, "learning_rate": 0.003, "loss": 4.087, "step": 14348 }, { "epoch": 0.14349, "grad_norm": 1.3819098921980475, "learning_rate": 0.003, "loss": 4.0607, "step": 14349 }, { "epoch": 0.1435, "grad_norm": 1.2217818750913538, "learning_rate": 0.003, "loss": 4.062, "step": 14350 }, { "epoch": 0.14351, "grad_norm": 1.4013634989561357, "learning_rate": 0.003, "loss": 4.0761, "step": 14351 }, { "epoch": 0.14352, "grad_norm": 1.0295581584090119, "learning_rate": 0.003, "loss": 4.0643, "step": 14352 }, { "epoch": 0.14353, "grad_norm": 1.4056870647209194, "learning_rate": 0.003, "loss": 4.067, "step": 14353 }, { "epoch": 0.14354, "grad_norm": 1.0034190206026332, "learning_rate": 0.003, "loss": 4.0345, "step": 14354 }, { "epoch": 0.14355, "grad_norm": 1.316326546089397, "learning_rate": 0.003, "loss": 4.0507, "step": 14355 }, { "epoch": 0.14356, "grad_norm": 0.9754900843938116, "learning_rate": 0.003, "loss": 4.061, "step": 14356 }, { "epoch": 0.14357, "grad_norm": 1.297848544306027, "learning_rate": 0.003, "loss": 4.0539, "step": 14357 }, { "epoch": 0.14358, "grad_norm": 1.279855276088557, "learning_rate": 0.003, "loss": 4.05, "step": 14358 }, { "epoch": 0.14359, "grad_norm": 1.2715881490998309, "learning_rate": 0.003, "loss": 4.0745, "step": 14359 }, { "epoch": 0.1436, "grad_norm": 1.0772515093805437, "learning_rate": 0.003, "loss": 4.0538, "step": 14360 }, { "epoch": 0.14361, "grad_norm": 1.326509901035539, "learning_rate": 0.003, "loss": 4.0339, "step": 14361 }, { "epoch": 0.14362, "grad_norm": 1.0928487982487218, "learning_rate": 0.003, "loss": 4.1031, "step": 14362 }, { "epoch": 0.14363, "grad_norm": 1.1140734060002717, "learning_rate": 0.003, "loss": 4.0505, "step": 14363 }, { "epoch": 0.14364, "grad_norm": 1.2634878968911296, "learning_rate": 0.003, "loss": 4.0789, "step": 14364 }, { "epoch": 0.14365, "grad_norm": 0.9166447285257685, "learning_rate": 0.003, "loss": 4.0561, "step": 14365 }, { "epoch": 0.14366, "grad_norm": 1.1893186520089365, "learning_rate": 0.003, "loss": 4.0667, "step": 14366 }, { "epoch": 0.14367, "grad_norm": 1.4496297340179196, "learning_rate": 0.003, "loss": 4.0476, "step": 14367 }, { "epoch": 0.14368, "grad_norm": 0.9584473542956278, "learning_rate": 0.003, "loss": 4.0548, "step": 14368 }, { "epoch": 0.14369, "grad_norm": 1.1633714002749225, "learning_rate": 0.003, "loss": 4.0492, "step": 14369 }, { "epoch": 0.1437, "grad_norm": 1.1387893863445588, "learning_rate": 0.003, "loss": 4.0431, "step": 14370 }, { "epoch": 0.14371, "grad_norm": 1.1966518735212899, "learning_rate": 0.003, "loss": 4.0596, "step": 14371 }, { "epoch": 0.14372, "grad_norm": 1.1451166210976724, "learning_rate": 0.003, "loss": 4.06, "step": 14372 }, { "epoch": 0.14373, "grad_norm": 1.2668958911872625, "learning_rate": 0.003, "loss": 4.0821, "step": 14373 }, { "epoch": 0.14374, "grad_norm": 1.1588675101541697, "learning_rate": 0.003, "loss": 4.07, "step": 14374 }, { "epoch": 0.14375, "grad_norm": 1.006739453380164, "learning_rate": 0.003, "loss": 4.0798, "step": 14375 }, { "epoch": 0.14376, "grad_norm": 1.3155615412750452, "learning_rate": 0.003, "loss": 4.0714, "step": 14376 }, { "epoch": 0.14377, "grad_norm": 1.3150776956736385, "learning_rate": 0.003, "loss": 4.0642, "step": 14377 }, { "epoch": 0.14378, "grad_norm": 1.3536875500683214, "learning_rate": 0.003, "loss": 4.057, "step": 14378 }, { "epoch": 0.14379, "grad_norm": 1.1112812989222873, "learning_rate": 0.003, "loss": 4.0595, "step": 14379 }, { "epoch": 0.1438, "grad_norm": 1.1198013178022161, "learning_rate": 0.003, "loss": 4.055, "step": 14380 }, { "epoch": 0.14381, "grad_norm": 1.270568937258633, "learning_rate": 0.003, "loss": 4.047, "step": 14381 }, { "epoch": 0.14382, "grad_norm": 0.9293268891593124, "learning_rate": 0.003, "loss": 4.0471, "step": 14382 }, { "epoch": 0.14383, "grad_norm": 1.3597104760585619, "learning_rate": 0.003, "loss": 4.0674, "step": 14383 }, { "epoch": 0.14384, "grad_norm": 1.0496305274028601, "learning_rate": 0.003, "loss": 4.0582, "step": 14384 }, { "epoch": 0.14385, "grad_norm": 1.2067881598096581, "learning_rate": 0.003, "loss": 4.072, "step": 14385 }, { "epoch": 0.14386, "grad_norm": 1.1605057054446346, "learning_rate": 0.003, "loss": 4.0666, "step": 14386 }, { "epoch": 0.14387, "grad_norm": 1.4127073565580401, "learning_rate": 0.003, "loss": 4.048, "step": 14387 }, { "epoch": 0.14388, "grad_norm": 1.0999021637295774, "learning_rate": 0.003, "loss": 4.0644, "step": 14388 }, { "epoch": 0.14389, "grad_norm": 1.235426272669938, "learning_rate": 0.003, "loss": 4.0909, "step": 14389 }, { "epoch": 0.1439, "grad_norm": 1.2021659561157405, "learning_rate": 0.003, "loss": 4.0661, "step": 14390 }, { "epoch": 0.14391, "grad_norm": 1.1205366066590694, "learning_rate": 0.003, "loss": 4.0531, "step": 14391 }, { "epoch": 0.14392, "grad_norm": 1.2062595819672428, "learning_rate": 0.003, "loss": 4.0597, "step": 14392 }, { "epoch": 0.14393, "grad_norm": 1.2771123386515066, "learning_rate": 0.003, "loss": 4.0427, "step": 14393 }, { "epoch": 0.14394, "grad_norm": 1.1521172255235568, "learning_rate": 0.003, "loss": 4.064, "step": 14394 }, { "epoch": 0.14395, "grad_norm": 1.171395943337496, "learning_rate": 0.003, "loss": 4.0584, "step": 14395 }, { "epoch": 0.14396, "grad_norm": 1.2084041986998417, "learning_rate": 0.003, "loss": 4.0707, "step": 14396 }, { "epoch": 0.14397, "grad_norm": 0.9788561303985829, "learning_rate": 0.003, "loss": 4.0355, "step": 14397 }, { "epoch": 0.14398, "grad_norm": 1.0686881244883093, "learning_rate": 0.003, "loss": 4.048, "step": 14398 }, { "epoch": 0.14399, "grad_norm": 1.3496435545184504, "learning_rate": 0.003, "loss": 4.0427, "step": 14399 }, { "epoch": 0.144, "grad_norm": 0.9615424273870988, "learning_rate": 0.003, "loss": 4.0605, "step": 14400 }, { "epoch": 0.14401, "grad_norm": 1.3318751453700992, "learning_rate": 0.003, "loss": 4.0531, "step": 14401 }, { "epoch": 0.14402, "grad_norm": 1.096003181810339, "learning_rate": 0.003, "loss": 4.0796, "step": 14402 }, { "epoch": 0.14403, "grad_norm": 1.11545228527377, "learning_rate": 0.003, "loss": 4.0777, "step": 14403 }, { "epoch": 0.14404, "grad_norm": 1.2041662877713406, "learning_rate": 0.003, "loss": 4.0796, "step": 14404 }, { "epoch": 0.14405, "grad_norm": 1.0513210905801005, "learning_rate": 0.003, "loss": 4.0713, "step": 14405 }, { "epoch": 0.14406, "grad_norm": 1.1606525548776172, "learning_rate": 0.003, "loss": 4.0847, "step": 14406 }, { "epoch": 0.14407, "grad_norm": 1.2641343564802707, "learning_rate": 0.003, "loss": 4.0729, "step": 14407 }, { "epoch": 0.14408, "grad_norm": 1.247940316562981, "learning_rate": 0.003, "loss": 4.0884, "step": 14408 }, { "epoch": 0.14409, "grad_norm": 1.2041270097613936, "learning_rate": 0.003, "loss": 4.0387, "step": 14409 }, { "epoch": 0.1441, "grad_norm": 1.0415938224350967, "learning_rate": 0.003, "loss": 4.0322, "step": 14410 }, { "epoch": 0.14411, "grad_norm": 1.2666962054012227, "learning_rate": 0.003, "loss": 4.0492, "step": 14411 }, { "epoch": 0.14412, "grad_norm": 1.1439247392330354, "learning_rate": 0.003, "loss": 4.0431, "step": 14412 }, { "epoch": 0.14413, "grad_norm": 1.2602037342993857, "learning_rate": 0.003, "loss": 4.0809, "step": 14413 }, { "epoch": 0.14414, "grad_norm": 1.1622242094178536, "learning_rate": 0.003, "loss": 4.058, "step": 14414 }, { "epoch": 0.14415, "grad_norm": 1.2105148648661788, "learning_rate": 0.003, "loss": 4.0247, "step": 14415 }, { "epoch": 0.14416, "grad_norm": 1.1207045739791284, "learning_rate": 0.003, "loss": 4.038, "step": 14416 }, { "epoch": 0.14417, "grad_norm": 1.4836229313973075, "learning_rate": 0.003, "loss": 4.0519, "step": 14417 }, { "epoch": 0.14418, "grad_norm": 0.9855947054257272, "learning_rate": 0.003, "loss": 4.0711, "step": 14418 }, { "epoch": 0.14419, "grad_norm": 1.3469614476056524, "learning_rate": 0.003, "loss": 4.0704, "step": 14419 }, { "epoch": 0.1442, "grad_norm": 1.1425115038010174, "learning_rate": 0.003, "loss": 4.0829, "step": 14420 }, { "epoch": 0.14421, "grad_norm": 1.2457255262964664, "learning_rate": 0.003, "loss": 4.0802, "step": 14421 }, { "epoch": 0.14422, "grad_norm": 1.1336401634870403, "learning_rate": 0.003, "loss": 4.0688, "step": 14422 }, { "epoch": 0.14423, "grad_norm": 1.0490914614301334, "learning_rate": 0.003, "loss": 4.0692, "step": 14423 }, { "epoch": 0.14424, "grad_norm": 1.2515768162712582, "learning_rate": 0.003, "loss": 4.0341, "step": 14424 }, { "epoch": 0.14425, "grad_norm": 1.177888633839058, "learning_rate": 0.003, "loss": 4.0403, "step": 14425 }, { "epoch": 0.14426, "grad_norm": 1.4308659781289155, "learning_rate": 0.003, "loss": 4.0734, "step": 14426 }, { "epoch": 0.14427, "grad_norm": 1.2623096956474995, "learning_rate": 0.003, "loss": 4.0721, "step": 14427 }, { "epoch": 0.14428, "grad_norm": 1.2325810704553675, "learning_rate": 0.003, "loss": 4.0531, "step": 14428 }, { "epoch": 0.14429, "grad_norm": 1.2871405107187142, "learning_rate": 0.003, "loss": 4.0935, "step": 14429 }, { "epoch": 0.1443, "grad_norm": 1.2934040487110332, "learning_rate": 0.003, "loss": 4.0815, "step": 14430 }, { "epoch": 0.14431, "grad_norm": 1.2189751427410236, "learning_rate": 0.003, "loss": 4.0869, "step": 14431 }, { "epoch": 0.14432, "grad_norm": 1.412281996903719, "learning_rate": 0.003, "loss": 4.0772, "step": 14432 }, { "epoch": 0.14433, "grad_norm": 1.0195358371839056, "learning_rate": 0.003, "loss": 4.0611, "step": 14433 }, { "epoch": 0.14434, "grad_norm": 1.2086298050739679, "learning_rate": 0.003, "loss": 4.0557, "step": 14434 }, { "epoch": 0.14435, "grad_norm": 1.0830766577799509, "learning_rate": 0.003, "loss": 4.075, "step": 14435 }, { "epoch": 0.14436, "grad_norm": 1.2975847416280557, "learning_rate": 0.003, "loss": 4.0887, "step": 14436 }, { "epoch": 0.14437, "grad_norm": 0.9731705441371323, "learning_rate": 0.003, "loss": 4.0609, "step": 14437 }, { "epoch": 0.14438, "grad_norm": 1.2630870673775951, "learning_rate": 0.003, "loss": 4.0798, "step": 14438 }, { "epoch": 0.14439, "grad_norm": 1.1358145259025993, "learning_rate": 0.003, "loss": 4.0976, "step": 14439 }, { "epoch": 0.1444, "grad_norm": 1.3018920164838934, "learning_rate": 0.003, "loss": 4.0755, "step": 14440 }, { "epoch": 0.14441, "grad_norm": 0.9520262031343586, "learning_rate": 0.003, "loss": 4.085, "step": 14441 }, { "epoch": 0.14442, "grad_norm": 1.37672864168369, "learning_rate": 0.003, "loss": 4.069, "step": 14442 }, { "epoch": 0.14443, "grad_norm": 1.0880208080473694, "learning_rate": 0.003, "loss": 4.0386, "step": 14443 }, { "epoch": 0.14444, "grad_norm": 1.3299459072739503, "learning_rate": 0.003, "loss": 4.0562, "step": 14444 }, { "epoch": 0.14445, "grad_norm": 1.0251162319901956, "learning_rate": 0.003, "loss": 4.0684, "step": 14445 }, { "epoch": 0.14446, "grad_norm": 1.4596973738649757, "learning_rate": 0.003, "loss": 4.0496, "step": 14446 }, { "epoch": 0.14447, "grad_norm": 1.0297129556679012, "learning_rate": 0.003, "loss": 4.0881, "step": 14447 }, { "epoch": 0.14448, "grad_norm": 1.4479733966210386, "learning_rate": 0.003, "loss": 4.0765, "step": 14448 }, { "epoch": 0.14449, "grad_norm": 1.2439657330802247, "learning_rate": 0.003, "loss": 4.0377, "step": 14449 }, { "epoch": 0.1445, "grad_norm": 1.0180874319876076, "learning_rate": 0.003, "loss": 4.0751, "step": 14450 }, { "epoch": 0.14451, "grad_norm": 1.328472310811886, "learning_rate": 0.003, "loss": 4.0548, "step": 14451 }, { "epoch": 0.14452, "grad_norm": 1.1103269484626483, "learning_rate": 0.003, "loss": 4.0551, "step": 14452 }, { "epoch": 0.14453, "grad_norm": 1.2902268585395242, "learning_rate": 0.003, "loss": 4.0612, "step": 14453 }, { "epoch": 0.14454, "grad_norm": 1.2104492374632532, "learning_rate": 0.003, "loss": 4.0551, "step": 14454 }, { "epoch": 0.14455, "grad_norm": 1.2072891677667257, "learning_rate": 0.003, "loss": 4.039, "step": 14455 }, { "epoch": 0.14456, "grad_norm": 1.196612239151037, "learning_rate": 0.003, "loss": 4.0285, "step": 14456 }, { "epoch": 0.14457, "grad_norm": 1.2599557462766016, "learning_rate": 0.003, "loss": 4.0693, "step": 14457 }, { "epoch": 0.14458, "grad_norm": 1.3945832178739188, "learning_rate": 0.003, "loss": 4.0532, "step": 14458 }, { "epoch": 0.14459, "grad_norm": 1.1456726401275719, "learning_rate": 0.003, "loss": 4.067, "step": 14459 }, { "epoch": 0.1446, "grad_norm": 1.3704195688415575, "learning_rate": 0.003, "loss": 4.0696, "step": 14460 }, { "epoch": 0.14461, "grad_norm": 0.883410802755949, "learning_rate": 0.003, "loss": 4.0556, "step": 14461 }, { "epoch": 0.14462, "grad_norm": 0.9383054724629, "learning_rate": 0.003, "loss": 4.0501, "step": 14462 }, { "epoch": 0.14463, "grad_norm": 1.18692137658522, "learning_rate": 0.003, "loss": 4.075, "step": 14463 }, { "epoch": 0.14464, "grad_norm": 1.13057186010462, "learning_rate": 0.003, "loss": 4.0813, "step": 14464 }, { "epoch": 0.14465, "grad_norm": 1.4486273262774236, "learning_rate": 0.003, "loss": 4.0692, "step": 14465 }, { "epoch": 0.14466, "grad_norm": 1.0363963440455068, "learning_rate": 0.003, "loss": 4.0618, "step": 14466 }, { "epoch": 0.14467, "grad_norm": 1.2446597742072119, "learning_rate": 0.003, "loss": 4.1064, "step": 14467 }, { "epoch": 0.14468, "grad_norm": 1.2922347688614726, "learning_rate": 0.003, "loss": 4.0563, "step": 14468 }, { "epoch": 0.14469, "grad_norm": 1.2097053980473647, "learning_rate": 0.003, "loss": 4.0654, "step": 14469 }, { "epoch": 0.1447, "grad_norm": 1.351741293527175, "learning_rate": 0.003, "loss": 4.0887, "step": 14470 }, { "epoch": 0.14471, "grad_norm": 1.0702467467520114, "learning_rate": 0.003, "loss": 4.0579, "step": 14471 }, { "epoch": 0.14472, "grad_norm": 1.28495992989329, "learning_rate": 0.003, "loss": 4.0596, "step": 14472 }, { "epoch": 0.14473, "grad_norm": 1.0662656580764747, "learning_rate": 0.003, "loss": 4.0396, "step": 14473 }, { "epoch": 0.14474, "grad_norm": 1.197696239350392, "learning_rate": 0.003, "loss": 4.072, "step": 14474 }, { "epoch": 0.14475, "grad_norm": 1.129624119151968, "learning_rate": 0.003, "loss": 4.0931, "step": 14475 }, { "epoch": 0.14476, "grad_norm": 1.2455647657848186, "learning_rate": 0.003, "loss": 4.0889, "step": 14476 }, { "epoch": 0.14477, "grad_norm": 1.1819455647373525, "learning_rate": 0.003, "loss": 4.0753, "step": 14477 }, { "epoch": 0.14478, "grad_norm": 1.3905545875085565, "learning_rate": 0.003, "loss": 4.0554, "step": 14478 }, { "epoch": 0.14479, "grad_norm": 1.170762766452118, "learning_rate": 0.003, "loss": 4.0514, "step": 14479 }, { "epoch": 0.1448, "grad_norm": 1.2513212509048317, "learning_rate": 0.003, "loss": 4.0534, "step": 14480 }, { "epoch": 0.14481, "grad_norm": 0.9761595945600573, "learning_rate": 0.003, "loss": 4.0749, "step": 14481 }, { "epoch": 0.14482, "grad_norm": 1.0174914668918622, "learning_rate": 0.003, "loss": 4.0516, "step": 14482 }, { "epoch": 0.14483, "grad_norm": 1.359980196411554, "learning_rate": 0.003, "loss": 4.0538, "step": 14483 }, { "epoch": 0.14484, "grad_norm": 1.0902831742596368, "learning_rate": 0.003, "loss": 4.0465, "step": 14484 }, { "epoch": 0.14485, "grad_norm": 1.2593625056395699, "learning_rate": 0.003, "loss": 4.0562, "step": 14485 }, { "epoch": 0.14486, "grad_norm": 0.9821002792210466, "learning_rate": 0.003, "loss": 4.0601, "step": 14486 }, { "epoch": 0.14487, "grad_norm": 1.2255247495177521, "learning_rate": 0.003, "loss": 4.0588, "step": 14487 }, { "epoch": 0.14488, "grad_norm": 1.2069811086964375, "learning_rate": 0.003, "loss": 4.0874, "step": 14488 }, { "epoch": 0.14489, "grad_norm": 1.0643181195045752, "learning_rate": 0.003, "loss": 4.0566, "step": 14489 }, { "epoch": 0.1449, "grad_norm": 1.326577652449159, "learning_rate": 0.003, "loss": 4.0599, "step": 14490 }, { "epoch": 0.14491, "grad_norm": 1.0539774466861362, "learning_rate": 0.003, "loss": 4.0591, "step": 14491 }, { "epoch": 0.14492, "grad_norm": 1.2060034543920808, "learning_rate": 0.003, "loss": 4.0694, "step": 14492 }, { "epoch": 0.14493, "grad_norm": 1.255149818281733, "learning_rate": 0.003, "loss": 4.072, "step": 14493 }, { "epoch": 0.14494, "grad_norm": 1.5354233835193087, "learning_rate": 0.003, "loss": 4.091, "step": 14494 }, { "epoch": 0.14495, "grad_norm": 0.9562276465593462, "learning_rate": 0.003, "loss": 4.0704, "step": 14495 }, { "epoch": 0.14496, "grad_norm": 1.2394086556118171, "learning_rate": 0.003, "loss": 4.0542, "step": 14496 }, { "epoch": 0.14497, "grad_norm": 1.3351260622140366, "learning_rate": 0.003, "loss": 4.0803, "step": 14497 }, { "epoch": 0.14498, "grad_norm": 1.0365048636710072, "learning_rate": 0.003, "loss": 4.0574, "step": 14498 }, { "epoch": 0.14499, "grad_norm": 1.3895783940738689, "learning_rate": 0.003, "loss": 4.0878, "step": 14499 }, { "epoch": 0.145, "grad_norm": 1.092441014869526, "learning_rate": 0.003, "loss": 4.1018, "step": 14500 }, { "epoch": 0.14501, "grad_norm": 1.278832843735235, "learning_rate": 0.003, "loss": 4.0795, "step": 14501 }, { "epoch": 0.14502, "grad_norm": 1.2593853894413225, "learning_rate": 0.003, "loss": 4.0684, "step": 14502 }, { "epoch": 0.14503, "grad_norm": 1.0732786142231292, "learning_rate": 0.003, "loss": 4.0893, "step": 14503 }, { "epoch": 0.14504, "grad_norm": 1.3895213148451642, "learning_rate": 0.003, "loss": 4.0754, "step": 14504 }, { "epoch": 0.14505, "grad_norm": 1.0759802486567076, "learning_rate": 0.003, "loss": 4.0447, "step": 14505 }, { "epoch": 0.14506, "grad_norm": 1.5076770877355903, "learning_rate": 0.003, "loss": 4.0531, "step": 14506 }, { "epoch": 0.14507, "grad_norm": 1.100514514015529, "learning_rate": 0.003, "loss": 4.0365, "step": 14507 }, { "epoch": 0.14508, "grad_norm": 1.3372686597544416, "learning_rate": 0.003, "loss": 4.0499, "step": 14508 }, { "epoch": 0.14509, "grad_norm": 1.0311444282019597, "learning_rate": 0.003, "loss": 4.0785, "step": 14509 }, { "epoch": 0.1451, "grad_norm": 1.1733465513027923, "learning_rate": 0.003, "loss": 4.0806, "step": 14510 }, { "epoch": 0.14511, "grad_norm": 1.174627063648199, "learning_rate": 0.003, "loss": 4.0557, "step": 14511 }, { "epoch": 0.14512, "grad_norm": 1.0887227832658273, "learning_rate": 0.003, "loss": 4.1037, "step": 14512 }, { "epoch": 0.14513, "grad_norm": 1.3383210590853012, "learning_rate": 0.003, "loss": 4.0545, "step": 14513 }, { "epoch": 0.14514, "grad_norm": 1.199637874835041, "learning_rate": 0.003, "loss": 4.0871, "step": 14514 }, { "epoch": 0.14515, "grad_norm": 1.2820711854534685, "learning_rate": 0.003, "loss": 4.0759, "step": 14515 }, { "epoch": 0.14516, "grad_norm": 1.2169255591432746, "learning_rate": 0.003, "loss": 4.0853, "step": 14516 }, { "epoch": 0.14517, "grad_norm": 1.4259356736559685, "learning_rate": 0.003, "loss": 4.0649, "step": 14517 }, { "epoch": 0.14518, "grad_norm": 1.085508315724117, "learning_rate": 0.003, "loss": 4.0748, "step": 14518 }, { "epoch": 0.14519, "grad_norm": 1.2337796428518382, "learning_rate": 0.003, "loss": 4.0776, "step": 14519 }, { "epoch": 0.1452, "grad_norm": 1.2381036650493278, "learning_rate": 0.003, "loss": 4.052, "step": 14520 }, { "epoch": 0.14521, "grad_norm": 1.222946112707692, "learning_rate": 0.003, "loss": 4.0657, "step": 14521 }, { "epoch": 0.14522, "grad_norm": 0.9838979388637205, "learning_rate": 0.003, "loss": 4.0262, "step": 14522 }, { "epoch": 0.14523, "grad_norm": 1.2459678211650227, "learning_rate": 0.003, "loss": 4.0688, "step": 14523 }, { "epoch": 0.14524, "grad_norm": 0.9726485250787739, "learning_rate": 0.003, "loss": 4.0853, "step": 14524 }, { "epoch": 0.14525, "grad_norm": 1.30750556386179, "learning_rate": 0.003, "loss": 4.043, "step": 14525 }, { "epoch": 0.14526, "grad_norm": 1.3415239021375474, "learning_rate": 0.003, "loss": 4.0593, "step": 14526 }, { "epoch": 0.14527, "grad_norm": 1.2152644096308234, "learning_rate": 0.003, "loss": 4.068, "step": 14527 }, { "epoch": 0.14528, "grad_norm": 1.0561859514540088, "learning_rate": 0.003, "loss": 4.0595, "step": 14528 }, { "epoch": 0.14529, "grad_norm": 1.1541866884114087, "learning_rate": 0.003, "loss": 4.0679, "step": 14529 }, { "epoch": 0.1453, "grad_norm": 1.2411854612012745, "learning_rate": 0.003, "loss": 4.0609, "step": 14530 }, { "epoch": 0.14531, "grad_norm": 1.2655536383647337, "learning_rate": 0.003, "loss": 4.0597, "step": 14531 }, { "epoch": 0.14532, "grad_norm": 1.6007632252405208, "learning_rate": 0.003, "loss": 4.0769, "step": 14532 }, { "epoch": 0.14533, "grad_norm": 1.013734516566645, "learning_rate": 0.003, "loss": 4.0881, "step": 14533 }, { "epoch": 0.14534, "grad_norm": 1.2815320279732259, "learning_rate": 0.003, "loss": 4.0793, "step": 14534 }, { "epoch": 0.14535, "grad_norm": 1.1098958620616504, "learning_rate": 0.003, "loss": 4.0829, "step": 14535 }, { "epoch": 0.14536, "grad_norm": 1.262519752412867, "learning_rate": 0.003, "loss": 4.0807, "step": 14536 }, { "epoch": 0.14537, "grad_norm": 1.229410593450138, "learning_rate": 0.003, "loss": 4.0499, "step": 14537 }, { "epoch": 0.14538, "grad_norm": 1.0249854737352893, "learning_rate": 0.003, "loss": 4.0706, "step": 14538 }, { "epoch": 0.14539, "grad_norm": 1.2106568712291543, "learning_rate": 0.003, "loss": 4.0585, "step": 14539 }, { "epoch": 0.1454, "grad_norm": 1.3375194318072543, "learning_rate": 0.003, "loss": 4.0716, "step": 14540 }, { "epoch": 0.14541, "grad_norm": 1.2808277806276076, "learning_rate": 0.003, "loss": 4.0669, "step": 14541 }, { "epoch": 0.14542, "grad_norm": 1.3645526732280286, "learning_rate": 0.003, "loss": 4.0685, "step": 14542 }, { "epoch": 0.14543, "grad_norm": 0.927561164935603, "learning_rate": 0.003, "loss": 4.0597, "step": 14543 }, { "epoch": 0.14544, "grad_norm": 1.2374565780761364, "learning_rate": 0.003, "loss": 4.0864, "step": 14544 }, { "epoch": 0.14545, "grad_norm": 1.172172317630963, "learning_rate": 0.003, "loss": 4.0472, "step": 14545 }, { "epoch": 0.14546, "grad_norm": 1.0587197270489002, "learning_rate": 0.003, "loss": 4.0694, "step": 14546 }, { "epoch": 0.14547, "grad_norm": 1.0765614881728571, "learning_rate": 0.003, "loss": 4.0809, "step": 14547 }, { "epoch": 0.14548, "grad_norm": 1.301588512798978, "learning_rate": 0.003, "loss": 4.0502, "step": 14548 }, { "epoch": 0.14549, "grad_norm": 1.2205407961821984, "learning_rate": 0.003, "loss": 4.0787, "step": 14549 }, { "epoch": 0.1455, "grad_norm": 1.1145718192779168, "learning_rate": 0.003, "loss": 4.0713, "step": 14550 }, { "epoch": 0.14551, "grad_norm": 1.179621630415264, "learning_rate": 0.003, "loss": 4.0672, "step": 14551 }, { "epoch": 0.14552, "grad_norm": 1.1880988740837897, "learning_rate": 0.003, "loss": 4.0856, "step": 14552 }, { "epoch": 0.14553, "grad_norm": 1.1808497016223631, "learning_rate": 0.003, "loss": 4.0821, "step": 14553 }, { "epoch": 0.14554, "grad_norm": 1.2605311608691536, "learning_rate": 0.003, "loss": 4.0564, "step": 14554 }, { "epoch": 0.14555, "grad_norm": 1.0056741277557708, "learning_rate": 0.003, "loss": 4.047, "step": 14555 }, { "epoch": 0.14556, "grad_norm": 1.22080608763233, "learning_rate": 0.003, "loss": 4.0583, "step": 14556 }, { "epoch": 0.14557, "grad_norm": 1.2786913420968125, "learning_rate": 0.003, "loss": 4.0639, "step": 14557 }, { "epoch": 0.14558, "grad_norm": 1.0155114758361838, "learning_rate": 0.003, "loss": 4.0638, "step": 14558 }, { "epoch": 0.14559, "grad_norm": 1.346244687983044, "learning_rate": 0.003, "loss": 4.0569, "step": 14559 }, { "epoch": 0.1456, "grad_norm": 0.9875529737539662, "learning_rate": 0.003, "loss": 4.0459, "step": 14560 }, { "epoch": 0.14561, "grad_norm": 1.3121646791886872, "learning_rate": 0.003, "loss": 4.0657, "step": 14561 }, { "epoch": 0.14562, "grad_norm": 1.2255825121213335, "learning_rate": 0.003, "loss": 4.0604, "step": 14562 }, { "epoch": 0.14563, "grad_norm": 1.267275515399711, "learning_rate": 0.003, "loss": 4.0781, "step": 14563 }, { "epoch": 0.14564, "grad_norm": 1.4494582533880669, "learning_rate": 0.003, "loss": 4.0311, "step": 14564 }, { "epoch": 0.14565, "grad_norm": 0.9840388509427782, "learning_rate": 0.003, "loss": 4.0832, "step": 14565 }, { "epoch": 0.14566, "grad_norm": 1.1418983249704502, "learning_rate": 0.003, "loss": 4.087, "step": 14566 }, { "epoch": 0.14567, "grad_norm": 1.1638000329298885, "learning_rate": 0.003, "loss": 4.0694, "step": 14567 }, { "epoch": 0.14568, "grad_norm": 1.2215265284809167, "learning_rate": 0.003, "loss": 4.0744, "step": 14568 }, { "epoch": 0.14569, "grad_norm": 1.278431476453285, "learning_rate": 0.003, "loss": 4.0982, "step": 14569 }, { "epoch": 0.1457, "grad_norm": 1.0550827420241753, "learning_rate": 0.003, "loss": 4.0656, "step": 14570 }, { "epoch": 0.14571, "grad_norm": 1.3879840743112057, "learning_rate": 0.003, "loss": 4.0655, "step": 14571 }, { "epoch": 0.14572, "grad_norm": 1.1270340714390352, "learning_rate": 0.003, "loss": 4.0578, "step": 14572 }, { "epoch": 0.14573, "grad_norm": 1.0760229161533796, "learning_rate": 0.003, "loss": 4.0582, "step": 14573 }, { "epoch": 0.14574, "grad_norm": 1.3566984962366937, "learning_rate": 0.003, "loss": 4.0615, "step": 14574 }, { "epoch": 0.14575, "grad_norm": 1.0191836638824994, "learning_rate": 0.003, "loss": 4.0629, "step": 14575 }, { "epoch": 0.14576, "grad_norm": 1.2673730075312062, "learning_rate": 0.003, "loss": 4.0647, "step": 14576 }, { "epoch": 0.14577, "grad_norm": 1.176399466905642, "learning_rate": 0.003, "loss": 4.055, "step": 14577 }, { "epoch": 0.14578, "grad_norm": 1.170054090010369, "learning_rate": 0.003, "loss": 4.0591, "step": 14578 }, { "epoch": 0.14579, "grad_norm": 1.060788554025084, "learning_rate": 0.003, "loss": 4.0386, "step": 14579 }, { "epoch": 0.1458, "grad_norm": 1.2276134150389053, "learning_rate": 0.003, "loss": 4.0757, "step": 14580 }, { "epoch": 0.14581, "grad_norm": 1.3744823581511467, "learning_rate": 0.003, "loss": 4.033, "step": 14581 }, { "epoch": 0.14582, "grad_norm": 1.2962302966317039, "learning_rate": 0.003, "loss": 4.0606, "step": 14582 }, { "epoch": 0.14583, "grad_norm": 1.1529484222967947, "learning_rate": 0.003, "loss": 4.0494, "step": 14583 }, { "epoch": 0.14584, "grad_norm": 1.2619267189116472, "learning_rate": 0.003, "loss": 4.078, "step": 14584 }, { "epoch": 0.14585, "grad_norm": 1.299212021768132, "learning_rate": 0.003, "loss": 4.079, "step": 14585 }, { "epoch": 0.14586, "grad_norm": 1.2045962939061319, "learning_rate": 0.003, "loss": 4.0405, "step": 14586 }, { "epoch": 0.14587, "grad_norm": 1.082534399396641, "learning_rate": 0.003, "loss": 4.0669, "step": 14587 }, { "epoch": 0.14588, "grad_norm": 1.3499879459859705, "learning_rate": 0.003, "loss": 4.066, "step": 14588 }, { "epoch": 0.14589, "grad_norm": 1.0895206117845913, "learning_rate": 0.003, "loss": 4.0253, "step": 14589 }, { "epoch": 0.1459, "grad_norm": 1.3209310273651989, "learning_rate": 0.003, "loss": 4.0772, "step": 14590 }, { "epoch": 0.14591, "grad_norm": 1.0109199459082603, "learning_rate": 0.003, "loss": 4.0386, "step": 14591 }, { "epoch": 0.14592, "grad_norm": 1.3237592900981143, "learning_rate": 0.003, "loss": 4.0711, "step": 14592 }, { "epoch": 0.14593, "grad_norm": 1.2350511124004422, "learning_rate": 0.003, "loss": 4.0878, "step": 14593 }, { "epoch": 0.14594, "grad_norm": 1.3199162857821065, "learning_rate": 0.003, "loss": 4.0666, "step": 14594 }, { "epoch": 0.14595, "grad_norm": 1.1754752002000035, "learning_rate": 0.003, "loss": 4.066, "step": 14595 }, { "epoch": 0.14596, "grad_norm": 1.1615829473896362, "learning_rate": 0.003, "loss": 4.0694, "step": 14596 }, { "epoch": 0.14597, "grad_norm": 1.2390998641870503, "learning_rate": 0.003, "loss": 4.0989, "step": 14597 }, { "epoch": 0.14598, "grad_norm": 1.1328430900161557, "learning_rate": 0.003, "loss": 4.1069, "step": 14598 }, { "epoch": 0.14599, "grad_norm": 1.2876515222337628, "learning_rate": 0.003, "loss": 4.0368, "step": 14599 }, { "epoch": 0.146, "grad_norm": 1.0675039173430005, "learning_rate": 0.003, "loss": 4.0776, "step": 14600 }, { "epoch": 0.14601, "grad_norm": 1.1882138300330307, "learning_rate": 0.003, "loss": 4.078, "step": 14601 }, { "epoch": 0.14602, "grad_norm": 0.9090556479930595, "learning_rate": 0.003, "loss": 4.0464, "step": 14602 }, { "epoch": 0.14603, "grad_norm": 1.244940507479042, "learning_rate": 0.003, "loss": 4.0879, "step": 14603 }, { "epoch": 0.14604, "grad_norm": 1.382787613320107, "learning_rate": 0.003, "loss": 4.0662, "step": 14604 }, { "epoch": 0.14605, "grad_norm": 1.2648621268954807, "learning_rate": 0.003, "loss": 4.0475, "step": 14605 }, { "epoch": 0.14606, "grad_norm": 1.189385499215917, "learning_rate": 0.003, "loss": 4.0549, "step": 14606 }, { "epoch": 0.14607, "grad_norm": 1.2313921134389831, "learning_rate": 0.003, "loss": 4.0691, "step": 14607 }, { "epoch": 0.14608, "grad_norm": 1.2198898892078907, "learning_rate": 0.003, "loss": 4.0481, "step": 14608 }, { "epoch": 0.14609, "grad_norm": 1.2774933099669845, "learning_rate": 0.003, "loss": 4.0261, "step": 14609 }, { "epoch": 0.1461, "grad_norm": 1.0161959784494827, "learning_rate": 0.003, "loss": 4.0538, "step": 14610 }, { "epoch": 0.14611, "grad_norm": 1.1696317224606492, "learning_rate": 0.003, "loss": 4.0804, "step": 14611 }, { "epoch": 0.14612, "grad_norm": 1.2497934127953523, "learning_rate": 0.003, "loss": 4.0766, "step": 14612 }, { "epoch": 0.14613, "grad_norm": 1.1170236643383034, "learning_rate": 0.003, "loss": 4.0504, "step": 14613 }, { "epoch": 0.14614, "grad_norm": 1.2429169519224235, "learning_rate": 0.003, "loss": 4.0673, "step": 14614 }, { "epoch": 0.14615, "grad_norm": 0.9616778368294037, "learning_rate": 0.003, "loss": 4.0552, "step": 14615 }, { "epoch": 0.14616, "grad_norm": 1.372503444062351, "learning_rate": 0.003, "loss": 4.0658, "step": 14616 }, { "epoch": 0.14617, "grad_norm": 1.2010856800097343, "learning_rate": 0.003, "loss": 4.0735, "step": 14617 }, { "epoch": 0.14618, "grad_norm": 1.4562051310403952, "learning_rate": 0.003, "loss": 4.0679, "step": 14618 }, { "epoch": 0.14619, "grad_norm": 1.0531531805489553, "learning_rate": 0.003, "loss": 4.0676, "step": 14619 }, { "epoch": 0.1462, "grad_norm": 1.3799675814826888, "learning_rate": 0.003, "loss": 4.0846, "step": 14620 }, { "epoch": 0.14621, "grad_norm": 0.9903401853565199, "learning_rate": 0.003, "loss": 4.0885, "step": 14621 }, { "epoch": 0.14622, "grad_norm": 1.4470191974780509, "learning_rate": 0.003, "loss": 4.0745, "step": 14622 }, { "epoch": 0.14623, "grad_norm": 1.086507214320285, "learning_rate": 0.003, "loss": 4.0688, "step": 14623 }, { "epoch": 0.14624, "grad_norm": 1.3369608097583916, "learning_rate": 0.003, "loss": 4.083, "step": 14624 }, { "epoch": 0.14625, "grad_norm": 1.22527580058732, "learning_rate": 0.003, "loss": 4.0862, "step": 14625 }, { "epoch": 0.14626, "grad_norm": 1.1552527438449494, "learning_rate": 0.003, "loss": 4.086, "step": 14626 }, { "epoch": 0.14627, "grad_norm": 1.1882929095233759, "learning_rate": 0.003, "loss": 4.0751, "step": 14627 }, { "epoch": 0.14628, "grad_norm": 1.0708626022018357, "learning_rate": 0.003, "loss": 4.0789, "step": 14628 }, { "epoch": 0.14629, "grad_norm": 1.183950295621809, "learning_rate": 0.003, "loss": 4.0438, "step": 14629 }, { "epoch": 0.1463, "grad_norm": 1.3449094679059561, "learning_rate": 0.003, "loss": 4.1035, "step": 14630 }, { "epoch": 0.14631, "grad_norm": 1.0995889188378865, "learning_rate": 0.003, "loss": 4.0577, "step": 14631 }, { "epoch": 0.14632, "grad_norm": 1.3625103167580896, "learning_rate": 0.003, "loss": 4.0417, "step": 14632 }, { "epoch": 0.14633, "grad_norm": 1.187469929229442, "learning_rate": 0.003, "loss": 4.062, "step": 14633 }, { "epoch": 0.14634, "grad_norm": 1.3842712959457022, "learning_rate": 0.003, "loss": 4.0881, "step": 14634 }, { "epoch": 0.14635, "grad_norm": 1.1657763904446767, "learning_rate": 0.003, "loss": 4.0772, "step": 14635 }, { "epoch": 0.14636, "grad_norm": 1.0820491943073718, "learning_rate": 0.003, "loss": 4.0573, "step": 14636 }, { "epoch": 0.14637, "grad_norm": 1.2288146011263648, "learning_rate": 0.003, "loss": 4.0931, "step": 14637 }, { "epoch": 0.14638, "grad_norm": 1.0388897379918565, "learning_rate": 0.003, "loss": 4.0424, "step": 14638 }, { "epoch": 0.14639, "grad_norm": 1.4093240022149505, "learning_rate": 0.003, "loss": 4.0445, "step": 14639 }, { "epoch": 0.1464, "grad_norm": 1.0009021916306446, "learning_rate": 0.003, "loss": 4.0492, "step": 14640 }, { "epoch": 0.14641, "grad_norm": 1.6059976781374665, "learning_rate": 0.003, "loss": 4.0855, "step": 14641 }, { "epoch": 0.14642, "grad_norm": 1.005407837664379, "learning_rate": 0.003, "loss": 4.0572, "step": 14642 }, { "epoch": 0.14643, "grad_norm": 1.359081219147045, "learning_rate": 0.003, "loss": 4.0848, "step": 14643 }, { "epoch": 0.14644, "grad_norm": 1.225943668673731, "learning_rate": 0.003, "loss": 4.0351, "step": 14644 }, { "epoch": 0.14645, "grad_norm": 1.1407681364901596, "learning_rate": 0.003, "loss": 4.0812, "step": 14645 }, { "epoch": 0.14646, "grad_norm": 1.3033431686765575, "learning_rate": 0.003, "loss": 4.0803, "step": 14646 }, { "epoch": 0.14647, "grad_norm": 1.140009102359671, "learning_rate": 0.003, "loss": 4.0436, "step": 14647 }, { "epoch": 0.14648, "grad_norm": 1.1058876269366287, "learning_rate": 0.003, "loss": 4.0777, "step": 14648 }, { "epoch": 0.14649, "grad_norm": 1.5635184765600387, "learning_rate": 0.003, "loss": 4.0822, "step": 14649 }, { "epoch": 0.1465, "grad_norm": 1.1411867898542467, "learning_rate": 0.003, "loss": 4.073, "step": 14650 }, { "epoch": 0.14651, "grad_norm": 1.3139899588159343, "learning_rate": 0.003, "loss": 4.0953, "step": 14651 }, { "epoch": 0.14652, "grad_norm": 1.0367864731998273, "learning_rate": 0.003, "loss": 4.0536, "step": 14652 }, { "epoch": 0.14653, "grad_norm": 1.3422765458253958, "learning_rate": 0.003, "loss": 4.0903, "step": 14653 }, { "epoch": 0.14654, "grad_norm": 1.126354692445179, "learning_rate": 0.003, "loss": 4.0591, "step": 14654 }, { "epoch": 0.14655, "grad_norm": 1.2419398434083484, "learning_rate": 0.003, "loss": 4.0483, "step": 14655 }, { "epoch": 0.14656, "grad_norm": 1.0078000026030147, "learning_rate": 0.003, "loss": 4.0592, "step": 14656 }, { "epoch": 0.14657, "grad_norm": 1.2022185194869899, "learning_rate": 0.003, "loss": 4.0429, "step": 14657 }, { "epoch": 0.14658, "grad_norm": 1.1633967049791696, "learning_rate": 0.003, "loss": 4.0339, "step": 14658 }, { "epoch": 0.14659, "grad_norm": 1.1010487324351983, "learning_rate": 0.003, "loss": 4.0589, "step": 14659 }, { "epoch": 0.1466, "grad_norm": 1.0010589971993746, "learning_rate": 0.003, "loss": 4.0612, "step": 14660 }, { "epoch": 0.14661, "grad_norm": 1.3786396459595662, "learning_rate": 0.003, "loss": 4.074, "step": 14661 }, { "epoch": 0.14662, "grad_norm": 1.2255796984428124, "learning_rate": 0.003, "loss": 4.0484, "step": 14662 }, { "epoch": 0.14663, "grad_norm": 1.1980058196702907, "learning_rate": 0.003, "loss": 4.0884, "step": 14663 }, { "epoch": 0.14664, "grad_norm": 1.186504821107635, "learning_rate": 0.003, "loss": 4.0384, "step": 14664 }, { "epoch": 0.14665, "grad_norm": 1.0413325935959892, "learning_rate": 0.003, "loss": 4.0539, "step": 14665 }, { "epoch": 0.14666, "grad_norm": 1.2490642421528309, "learning_rate": 0.003, "loss": 4.059, "step": 14666 }, { "epoch": 0.14667, "grad_norm": 0.965613004754449, "learning_rate": 0.003, "loss": 4.0568, "step": 14667 }, { "epoch": 0.14668, "grad_norm": 1.1489845934784366, "learning_rate": 0.003, "loss": 4.0917, "step": 14668 }, { "epoch": 0.14669, "grad_norm": 1.4148464166474155, "learning_rate": 0.003, "loss": 4.0897, "step": 14669 }, { "epoch": 0.1467, "grad_norm": 1.1174324667613753, "learning_rate": 0.003, "loss": 4.0663, "step": 14670 }, { "epoch": 0.14671, "grad_norm": 1.4705757103421846, "learning_rate": 0.003, "loss": 4.0873, "step": 14671 }, { "epoch": 0.14672, "grad_norm": 1.1503001407497926, "learning_rate": 0.003, "loss": 4.0696, "step": 14672 }, { "epoch": 0.14673, "grad_norm": 1.2246606338903052, "learning_rate": 0.003, "loss": 4.0714, "step": 14673 }, { "epoch": 0.14674, "grad_norm": 1.2790334951540812, "learning_rate": 0.003, "loss": 4.0351, "step": 14674 }, { "epoch": 0.14675, "grad_norm": 1.1307066190918795, "learning_rate": 0.003, "loss": 4.0677, "step": 14675 }, { "epoch": 0.14676, "grad_norm": 1.2569590416028746, "learning_rate": 0.003, "loss": 4.0634, "step": 14676 }, { "epoch": 0.14677, "grad_norm": 1.0829300408271783, "learning_rate": 0.003, "loss": 4.0478, "step": 14677 }, { "epoch": 0.14678, "grad_norm": 1.269313016423024, "learning_rate": 0.003, "loss": 4.0768, "step": 14678 }, { "epoch": 0.14679, "grad_norm": 0.9549088118669827, "learning_rate": 0.003, "loss": 4.0671, "step": 14679 }, { "epoch": 0.1468, "grad_norm": 1.4753734474615985, "learning_rate": 0.003, "loss": 4.0945, "step": 14680 }, { "epoch": 0.14681, "grad_norm": 0.9408217562756533, "learning_rate": 0.003, "loss": 4.0535, "step": 14681 }, { "epoch": 0.14682, "grad_norm": 1.4576604514636082, "learning_rate": 0.003, "loss": 4.0644, "step": 14682 }, { "epoch": 0.14683, "grad_norm": 1.1046458235008672, "learning_rate": 0.003, "loss": 4.058, "step": 14683 }, { "epoch": 0.14684, "grad_norm": 1.2793133353188606, "learning_rate": 0.003, "loss": 4.0621, "step": 14684 }, { "epoch": 0.14685, "grad_norm": 1.1469602156718897, "learning_rate": 0.003, "loss": 4.0809, "step": 14685 }, { "epoch": 0.14686, "grad_norm": 1.346386790762256, "learning_rate": 0.003, "loss": 4.0456, "step": 14686 }, { "epoch": 0.14687, "grad_norm": 1.5623657287016393, "learning_rate": 0.003, "loss": 4.0726, "step": 14687 }, { "epoch": 0.14688, "grad_norm": 1.2249617734399507, "learning_rate": 0.003, "loss": 4.0647, "step": 14688 }, { "epoch": 0.14689, "grad_norm": 1.1500344130353606, "learning_rate": 0.003, "loss": 4.0761, "step": 14689 }, { "epoch": 0.1469, "grad_norm": 1.105124624695385, "learning_rate": 0.003, "loss": 4.0532, "step": 14690 }, { "epoch": 0.14691, "grad_norm": 1.2645902920592447, "learning_rate": 0.003, "loss": 4.0481, "step": 14691 }, { "epoch": 0.14692, "grad_norm": 1.0712841892770806, "learning_rate": 0.003, "loss": 4.061, "step": 14692 }, { "epoch": 0.14693, "grad_norm": 1.134902703042126, "learning_rate": 0.003, "loss": 4.0629, "step": 14693 }, { "epoch": 0.14694, "grad_norm": 1.1108776270674083, "learning_rate": 0.003, "loss": 4.0554, "step": 14694 }, { "epoch": 0.14695, "grad_norm": 1.059649463246095, "learning_rate": 0.003, "loss": 4.0885, "step": 14695 }, { "epoch": 0.14696, "grad_norm": 1.3014136466686084, "learning_rate": 0.003, "loss": 4.0519, "step": 14696 }, { "epoch": 0.14697, "grad_norm": 1.0449429689531058, "learning_rate": 0.003, "loss": 4.0575, "step": 14697 }, { "epoch": 0.14698, "grad_norm": 1.4253373081448815, "learning_rate": 0.003, "loss": 4.052, "step": 14698 }, { "epoch": 0.14699, "grad_norm": 1.297590735587724, "learning_rate": 0.003, "loss": 4.029, "step": 14699 }, { "epoch": 0.147, "grad_norm": 1.539234247196825, "learning_rate": 0.003, "loss": 4.0814, "step": 14700 }, { "epoch": 0.14701, "grad_norm": 1.0035249133499857, "learning_rate": 0.003, "loss": 4.0557, "step": 14701 }, { "epoch": 0.14702, "grad_norm": 1.0865586183742486, "learning_rate": 0.003, "loss": 4.0577, "step": 14702 }, { "epoch": 0.14703, "grad_norm": 1.1942781752177534, "learning_rate": 0.003, "loss": 4.064, "step": 14703 }, { "epoch": 0.14704, "grad_norm": 1.3027030455420896, "learning_rate": 0.003, "loss": 4.0601, "step": 14704 }, { "epoch": 0.14705, "grad_norm": 1.0922265779538505, "learning_rate": 0.003, "loss": 4.0564, "step": 14705 }, { "epoch": 0.14706, "grad_norm": 1.2461119997772603, "learning_rate": 0.003, "loss": 4.0679, "step": 14706 }, { "epoch": 0.14707, "grad_norm": 1.0979541624663782, "learning_rate": 0.003, "loss": 4.0723, "step": 14707 }, { "epoch": 0.14708, "grad_norm": 1.0504304539572533, "learning_rate": 0.003, "loss": 4.0536, "step": 14708 }, { "epoch": 0.14709, "grad_norm": 1.2043107179364494, "learning_rate": 0.003, "loss": 4.0192, "step": 14709 }, { "epoch": 0.1471, "grad_norm": 1.1676241791623339, "learning_rate": 0.003, "loss": 4.0498, "step": 14710 }, { "epoch": 0.14711, "grad_norm": 1.210562928850982, "learning_rate": 0.003, "loss": 4.0713, "step": 14711 }, { "epoch": 0.14712, "grad_norm": 1.2477750222221868, "learning_rate": 0.003, "loss": 4.0338, "step": 14712 }, { "epoch": 0.14713, "grad_norm": 1.0360888983840815, "learning_rate": 0.003, "loss": 4.0293, "step": 14713 }, { "epoch": 0.14714, "grad_norm": 1.2496428384055074, "learning_rate": 0.003, "loss": 4.0484, "step": 14714 }, { "epoch": 0.14715, "grad_norm": 1.1964986772086843, "learning_rate": 0.003, "loss": 4.0869, "step": 14715 }, { "epoch": 0.14716, "grad_norm": 1.3580341432365273, "learning_rate": 0.003, "loss": 4.0946, "step": 14716 }, { "epoch": 0.14717, "grad_norm": 1.0931902152639112, "learning_rate": 0.003, "loss": 4.0785, "step": 14717 }, { "epoch": 0.14718, "grad_norm": 1.3500717654172187, "learning_rate": 0.003, "loss": 4.0346, "step": 14718 }, { "epoch": 0.14719, "grad_norm": 1.1276161718533264, "learning_rate": 0.003, "loss": 4.0332, "step": 14719 }, { "epoch": 0.1472, "grad_norm": 1.132809481240904, "learning_rate": 0.003, "loss": 4.059, "step": 14720 }, { "epoch": 0.14721, "grad_norm": 1.3768027297001184, "learning_rate": 0.003, "loss": 4.062, "step": 14721 }, { "epoch": 0.14722, "grad_norm": 0.9530890757395711, "learning_rate": 0.003, "loss": 4.0737, "step": 14722 }, { "epoch": 0.14723, "grad_norm": 1.0827926656367546, "learning_rate": 0.003, "loss": 4.0409, "step": 14723 }, { "epoch": 0.14724, "grad_norm": 1.1187348846776448, "learning_rate": 0.003, "loss": 4.0648, "step": 14724 }, { "epoch": 0.14725, "grad_norm": 1.5315708362193274, "learning_rate": 0.003, "loss": 4.0716, "step": 14725 }, { "epoch": 0.14726, "grad_norm": 0.9385900526540597, "learning_rate": 0.003, "loss": 4.067, "step": 14726 }, { "epoch": 0.14727, "grad_norm": 1.2573944213072323, "learning_rate": 0.003, "loss": 4.0621, "step": 14727 }, { "epoch": 0.14728, "grad_norm": 1.2554160816324138, "learning_rate": 0.003, "loss": 4.0524, "step": 14728 }, { "epoch": 0.14729, "grad_norm": 1.1407009187347024, "learning_rate": 0.003, "loss": 4.0628, "step": 14729 }, { "epoch": 0.1473, "grad_norm": 1.1233097547412818, "learning_rate": 0.003, "loss": 4.0522, "step": 14730 }, { "epoch": 0.14731, "grad_norm": 1.1880767072159575, "learning_rate": 0.003, "loss": 4.0826, "step": 14731 }, { "epoch": 0.14732, "grad_norm": 1.1293569513781274, "learning_rate": 0.003, "loss": 4.0848, "step": 14732 }, { "epoch": 0.14733, "grad_norm": 1.0806254931971826, "learning_rate": 0.003, "loss": 4.0608, "step": 14733 }, { "epoch": 0.14734, "grad_norm": 1.2303839421783829, "learning_rate": 0.003, "loss": 4.0752, "step": 14734 }, { "epoch": 0.14735, "grad_norm": 1.1038210616468882, "learning_rate": 0.003, "loss": 4.0798, "step": 14735 }, { "epoch": 0.14736, "grad_norm": 1.4684827863386878, "learning_rate": 0.003, "loss": 4.0751, "step": 14736 }, { "epoch": 0.14737, "grad_norm": 0.9586319555713341, "learning_rate": 0.003, "loss": 4.055, "step": 14737 }, { "epoch": 0.14738, "grad_norm": 1.3252229291648583, "learning_rate": 0.003, "loss": 4.0782, "step": 14738 }, { "epoch": 0.14739, "grad_norm": 1.0321800907435257, "learning_rate": 0.003, "loss": 4.0515, "step": 14739 }, { "epoch": 0.1474, "grad_norm": 1.2514911163218652, "learning_rate": 0.003, "loss": 4.0636, "step": 14740 }, { "epoch": 0.14741, "grad_norm": 1.098563671851662, "learning_rate": 0.003, "loss": 4.0562, "step": 14741 }, { "epoch": 0.14742, "grad_norm": 1.2164774786994783, "learning_rate": 0.003, "loss": 4.0604, "step": 14742 }, { "epoch": 0.14743, "grad_norm": 1.3127094532072934, "learning_rate": 0.003, "loss": 4.0861, "step": 14743 }, { "epoch": 0.14744, "grad_norm": 1.6661720086832064, "learning_rate": 0.003, "loss": 4.0625, "step": 14744 }, { "epoch": 0.14745, "grad_norm": 1.1512949942038893, "learning_rate": 0.003, "loss": 4.0939, "step": 14745 }, { "epoch": 0.14746, "grad_norm": 1.1407807110626125, "learning_rate": 0.003, "loss": 4.0348, "step": 14746 }, { "epoch": 0.14747, "grad_norm": 1.252229562812047, "learning_rate": 0.003, "loss": 4.0757, "step": 14747 }, { "epoch": 0.14748, "grad_norm": 1.2099183759820158, "learning_rate": 0.003, "loss": 4.0714, "step": 14748 }, { "epoch": 0.14749, "grad_norm": 1.1472340119243152, "learning_rate": 0.003, "loss": 4.0679, "step": 14749 }, { "epoch": 0.1475, "grad_norm": 1.2165033715530738, "learning_rate": 0.003, "loss": 4.0772, "step": 14750 }, { "epoch": 0.14751, "grad_norm": 1.2680735466183857, "learning_rate": 0.003, "loss": 4.0457, "step": 14751 }, { "epoch": 0.14752, "grad_norm": 1.1487999400150248, "learning_rate": 0.003, "loss": 4.0912, "step": 14752 }, { "epoch": 0.14753, "grad_norm": 1.0689480660851851, "learning_rate": 0.003, "loss": 4.0411, "step": 14753 }, { "epoch": 0.14754, "grad_norm": 1.1452910366618543, "learning_rate": 0.003, "loss": 4.049, "step": 14754 }, { "epoch": 0.14755, "grad_norm": 1.07772546469753, "learning_rate": 0.003, "loss": 4.0905, "step": 14755 }, { "epoch": 0.14756, "grad_norm": 1.665425389885917, "learning_rate": 0.003, "loss": 4.0649, "step": 14756 }, { "epoch": 0.14757, "grad_norm": 0.9772616298363169, "learning_rate": 0.003, "loss": 4.0733, "step": 14757 }, { "epoch": 0.14758, "grad_norm": 1.2775909224056938, "learning_rate": 0.003, "loss": 4.0777, "step": 14758 }, { "epoch": 0.14759, "grad_norm": 1.0999525314929848, "learning_rate": 0.003, "loss": 4.0801, "step": 14759 }, { "epoch": 0.1476, "grad_norm": 1.1310319747116835, "learning_rate": 0.003, "loss": 4.067, "step": 14760 }, { "epoch": 0.14761, "grad_norm": 1.4201830819922163, "learning_rate": 0.003, "loss": 4.0741, "step": 14761 }, { "epoch": 0.14762, "grad_norm": 1.0700611618861156, "learning_rate": 0.003, "loss": 4.0627, "step": 14762 }, { "epoch": 0.14763, "grad_norm": 1.2204969586231962, "learning_rate": 0.003, "loss": 4.0573, "step": 14763 }, { "epoch": 0.14764, "grad_norm": 1.1176178847569394, "learning_rate": 0.003, "loss": 4.0527, "step": 14764 }, { "epoch": 0.14765, "grad_norm": 1.3089771832935406, "learning_rate": 0.003, "loss": 4.0671, "step": 14765 }, { "epoch": 0.14766, "grad_norm": 1.1554055840105277, "learning_rate": 0.003, "loss": 4.0714, "step": 14766 }, { "epoch": 0.14767, "grad_norm": 1.6609072176542494, "learning_rate": 0.003, "loss": 4.1078, "step": 14767 }, { "epoch": 0.14768, "grad_norm": 1.246840048758139, "learning_rate": 0.003, "loss": 4.0698, "step": 14768 }, { "epoch": 0.14769, "grad_norm": 1.0241407891675958, "learning_rate": 0.003, "loss": 4.044, "step": 14769 }, { "epoch": 0.1477, "grad_norm": 1.4294137772300697, "learning_rate": 0.003, "loss": 4.1007, "step": 14770 }, { "epoch": 0.14771, "grad_norm": 1.1823718803966925, "learning_rate": 0.003, "loss": 4.066, "step": 14771 }, { "epoch": 0.14772, "grad_norm": 1.0938050717152028, "learning_rate": 0.003, "loss": 4.0514, "step": 14772 }, { "epoch": 0.14773, "grad_norm": 1.2027402156460119, "learning_rate": 0.003, "loss": 4.0833, "step": 14773 }, { "epoch": 0.14774, "grad_norm": 1.0051254526180466, "learning_rate": 0.003, "loss": 4.0719, "step": 14774 }, { "epoch": 0.14775, "grad_norm": 1.3381363065056453, "learning_rate": 0.003, "loss": 4.0673, "step": 14775 }, { "epoch": 0.14776, "grad_norm": 0.984399063906672, "learning_rate": 0.003, "loss": 4.0479, "step": 14776 }, { "epoch": 0.14777, "grad_norm": 1.4129631638050066, "learning_rate": 0.003, "loss": 4.07, "step": 14777 }, { "epoch": 0.14778, "grad_norm": 1.4308281610427214, "learning_rate": 0.003, "loss": 4.0854, "step": 14778 }, { "epoch": 0.14779, "grad_norm": 0.955116698206692, "learning_rate": 0.003, "loss": 4.0583, "step": 14779 }, { "epoch": 0.1478, "grad_norm": 1.0683568116905549, "learning_rate": 0.003, "loss": 4.0619, "step": 14780 }, { "epoch": 0.14781, "grad_norm": 1.5234885998554772, "learning_rate": 0.003, "loss": 4.0405, "step": 14781 }, { "epoch": 0.14782, "grad_norm": 1.4803704733366099, "learning_rate": 0.003, "loss": 4.0896, "step": 14782 }, { "epoch": 0.14783, "grad_norm": 1.1172227373523123, "learning_rate": 0.003, "loss": 4.0758, "step": 14783 }, { "epoch": 0.14784, "grad_norm": 1.1037040152748328, "learning_rate": 0.003, "loss": 4.0497, "step": 14784 }, { "epoch": 0.14785, "grad_norm": 1.2164798999169917, "learning_rate": 0.003, "loss": 4.0479, "step": 14785 }, { "epoch": 0.14786, "grad_norm": 1.1894087594233522, "learning_rate": 0.003, "loss": 4.0647, "step": 14786 }, { "epoch": 0.14787, "grad_norm": 1.1490904386061846, "learning_rate": 0.003, "loss": 4.056, "step": 14787 }, { "epoch": 0.14788, "grad_norm": 1.3607172116554545, "learning_rate": 0.003, "loss": 4.03, "step": 14788 }, { "epoch": 0.14789, "grad_norm": 1.1301412998785627, "learning_rate": 0.003, "loss": 4.0747, "step": 14789 }, { "epoch": 0.1479, "grad_norm": 1.3003399748050446, "learning_rate": 0.003, "loss": 4.0617, "step": 14790 }, { "epoch": 0.14791, "grad_norm": 1.1210026984885646, "learning_rate": 0.003, "loss": 4.0704, "step": 14791 }, { "epoch": 0.14792, "grad_norm": 1.0584248662119486, "learning_rate": 0.003, "loss": 4.0565, "step": 14792 }, { "epoch": 0.14793, "grad_norm": 0.9511820155749877, "learning_rate": 0.003, "loss": 4.0656, "step": 14793 }, { "epoch": 0.14794, "grad_norm": 1.3841734053580776, "learning_rate": 0.003, "loss": 4.0743, "step": 14794 }, { "epoch": 0.14795, "grad_norm": 1.2774076165864907, "learning_rate": 0.003, "loss": 4.112, "step": 14795 }, { "epoch": 0.14796, "grad_norm": 1.170578313981786, "learning_rate": 0.003, "loss": 4.0311, "step": 14796 }, { "epoch": 0.14797, "grad_norm": 1.0440748467826015, "learning_rate": 0.003, "loss": 4.0691, "step": 14797 }, { "epoch": 0.14798, "grad_norm": 1.4750922628950431, "learning_rate": 0.003, "loss": 4.0846, "step": 14798 }, { "epoch": 0.14799, "grad_norm": 1.1759901251394247, "learning_rate": 0.003, "loss": 4.0699, "step": 14799 }, { "epoch": 0.148, "grad_norm": 1.2119667526250828, "learning_rate": 0.003, "loss": 4.0505, "step": 14800 }, { "epoch": 0.14801, "grad_norm": 1.2058549245284778, "learning_rate": 0.003, "loss": 4.0803, "step": 14801 }, { "epoch": 0.14802, "grad_norm": 1.2094443172169136, "learning_rate": 0.003, "loss": 4.0658, "step": 14802 }, { "epoch": 0.14803, "grad_norm": 1.4077305093241759, "learning_rate": 0.003, "loss": 4.0818, "step": 14803 }, { "epoch": 0.14804, "grad_norm": 0.8884195461234354, "learning_rate": 0.003, "loss": 4.0473, "step": 14804 }, { "epoch": 0.14805, "grad_norm": 1.2496416639608543, "learning_rate": 0.003, "loss": 4.0612, "step": 14805 }, { "epoch": 0.14806, "grad_norm": 1.1645065589649986, "learning_rate": 0.003, "loss": 4.0517, "step": 14806 }, { "epoch": 0.14807, "grad_norm": 1.069211692346441, "learning_rate": 0.003, "loss": 4.0294, "step": 14807 }, { "epoch": 0.14808, "grad_norm": 1.3547410373763051, "learning_rate": 0.003, "loss": 4.0899, "step": 14808 }, { "epoch": 0.14809, "grad_norm": 0.9696174075795378, "learning_rate": 0.003, "loss": 4.0272, "step": 14809 }, { "epoch": 0.1481, "grad_norm": 1.0054300191239651, "learning_rate": 0.003, "loss": 4.0656, "step": 14810 }, { "epoch": 0.14811, "grad_norm": 1.3234824372039642, "learning_rate": 0.003, "loss": 4.0706, "step": 14811 }, { "epoch": 0.14812, "grad_norm": 1.1056347507868667, "learning_rate": 0.003, "loss": 4.0123, "step": 14812 }, { "epoch": 0.14813, "grad_norm": 1.2361695219915105, "learning_rate": 0.003, "loss": 4.0737, "step": 14813 }, { "epoch": 0.14814, "grad_norm": 1.2422214003764849, "learning_rate": 0.003, "loss": 4.0554, "step": 14814 }, { "epoch": 0.14815, "grad_norm": 1.4860662128172797, "learning_rate": 0.003, "loss": 4.097, "step": 14815 }, { "epoch": 0.14816, "grad_norm": 1.095204934951978, "learning_rate": 0.003, "loss": 4.0706, "step": 14816 }, { "epoch": 0.14817, "grad_norm": 1.0549027373607387, "learning_rate": 0.003, "loss": 4.0667, "step": 14817 }, { "epoch": 0.14818, "grad_norm": 1.4257982540616287, "learning_rate": 0.003, "loss": 4.0335, "step": 14818 }, { "epoch": 0.14819, "grad_norm": 1.0022689449589357, "learning_rate": 0.003, "loss": 4.0591, "step": 14819 }, { "epoch": 0.1482, "grad_norm": 1.430846028214134, "learning_rate": 0.003, "loss": 4.0557, "step": 14820 }, { "epoch": 0.14821, "grad_norm": 1.222262674104428, "learning_rate": 0.003, "loss": 4.0449, "step": 14821 }, { "epoch": 0.14822, "grad_norm": 1.0746599463987394, "learning_rate": 0.003, "loss": 4.0287, "step": 14822 }, { "epoch": 0.14823, "grad_norm": 1.2059917161328615, "learning_rate": 0.003, "loss": 4.0456, "step": 14823 }, { "epoch": 0.14824, "grad_norm": 1.258274546832546, "learning_rate": 0.003, "loss": 4.0943, "step": 14824 }, { "epoch": 0.14825, "grad_norm": 1.1351255207782565, "learning_rate": 0.003, "loss": 4.0586, "step": 14825 }, { "epoch": 0.14826, "grad_norm": 1.1827242960166198, "learning_rate": 0.003, "loss": 4.062, "step": 14826 }, { "epoch": 0.14827, "grad_norm": 1.075962939782082, "learning_rate": 0.003, "loss": 4.077, "step": 14827 }, { "epoch": 0.14828, "grad_norm": 1.2771551815105726, "learning_rate": 0.003, "loss": 4.0779, "step": 14828 }, { "epoch": 0.14829, "grad_norm": 1.1121069099361611, "learning_rate": 0.003, "loss": 4.0662, "step": 14829 }, { "epoch": 0.1483, "grad_norm": 1.0363683408331907, "learning_rate": 0.003, "loss": 4.0178, "step": 14830 }, { "epoch": 0.14831, "grad_norm": 1.3818657844749311, "learning_rate": 0.003, "loss": 4.0833, "step": 14831 }, { "epoch": 0.14832, "grad_norm": 1.0801904208892987, "learning_rate": 0.003, "loss": 4.0839, "step": 14832 }, { "epoch": 0.14833, "grad_norm": 1.4360579789876684, "learning_rate": 0.003, "loss": 4.0783, "step": 14833 }, { "epoch": 0.14834, "grad_norm": 1.1531827140412088, "learning_rate": 0.003, "loss": 4.0695, "step": 14834 }, { "epoch": 0.14835, "grad_norm": 1.201525696367292, "learning_rate": 0.003, "loss": 4.0671, "step": 14835 }, { "epoch": 0.14836, "grad_norm": 1.0297719350270018, "learning_rate": 0.003, "loss": 4.0648, "step": 14836 }, { "epoch": 0.14837, "grad_norm": 1.3917502124564454, "learning_rate": 0.003, "loss": 4.074, "step": 14837 }, { "epoch": 0.14838, "grad_norm": 0.9675947716587852, "learning_rate": 0.003, "loss": 4.0574, "step": 14838 }, { "epoch": 0.14839, "grad_norm": 1.0119005973054083, "learning_rate": 0.003, "loss": 4.0732, "step": 14839 }, { "epoch": 0.1484, "grad_norm": 1.325765639484017, "learning_rate": 0.003, "loss": 4.0507, "step": 14840 }, { "epoch": 0.14841, "grad_norm": 1.1782336088419427, "learning_rate": 0.003, "loss": 4.0323, "step": 14841 }, { "epoch": 0.14842, "grad_norm": 1.183012556136071, "learning_rate": 0.003, "loss": 4.0543, "step": 14842 }, { "epoch": 0.14843, "grad_norm": 1.3779721557010383, "learning_rate": 0.003, "loss": 4.0627, "step": 14843 }, { "epoch": 0.14844, "grad_norm": 1.1266000946049708, "learning_rate": 0.003, "loss": 4.0533, "step": 14844 }, { "epoch": 0.14845, "grad_norm": 1.1059549469170604, "learning_rate": 0.003, "loss": 4.068, "step": 14845 }, { "epoch": 0.14846, "grad_norm": 1.1584890768547578, "learning_rate": 0.003, "loss": 4.0431, "step": 14846 }, { "epoch": 0.14847, "grad_norm": 1.3109204457867119, "learning_rate": 0.003, "loss": 4.0504, "step": 14847 }, { "epoch": 0.14848, "grad_norm": 1.033589018387824, "learning_rate": 0.003, "loss": 4.0781, "step": 14848 }, { "epoch": 0.14849, "grad_norm": 1.4595391695249693, "learning_rate": 0.003, "loss": 4.0663, "step": 14849 }, { "epoch": 0.1485, "grad_norm": 1.102563723807411, "learning_rate": 0.003, "loss": 4.047, "step": 14850 }, { "epoch": 0.14851, "grad_norm": 1.2581620355315035, "learning_rate": 0.003, "loss": 4.0545, "step": 14851 }, { "epoch": 0.14852, "grad_norm": 1.1974611892483713, "learning_rate": 0.003, "loss": 4.0511, "step": 14852 }, { "epoch": 0.14853, "grad_norm": 1.1440570289725562, "learning_rate": 0.003, "loss": 4.07, "step": 14853 }, { "epoch": 0.14854, "grad_norm": 1.4276809502727477, "learning_rate": 0.003, "loss": 4.0754, "step": 14854 }, { "epoch": 0.14855, "grad_norm": 1.0365841652167092, "learning_rate": 0.003, "loss": 4.0804, "step": 14855 }, { "epoch": 0.14856, "grad_norm": 1.3196888860347857, "learning_rate": 0.003, "loss": 4.0558, "step": 14856 }, { "epoch": 0.14857, "grad_norm": 1.2096838857761567, "learning_rate": 0.003, "loss": 4.0825, "step": 14857 }, { "epoch": 0.14858, "grad_norm": 1.1539413366906002, "learning_rate": 0.003, "loss": 4.0674, "step": 14858 }, { "epoch": 0.14859, "grad_norm": 1.1643571950626377, "learning_rate": 0.003, "loss": 4.0754, "step": 14859 }, { "epoch": 0.1486, "grad_norm": 1.1594988548082985, "learning_rate": 0.003, "loss": 4.0572, "step": 14860 }, { "epoch": 0.14861, "grad_norm": 1.3592161138599952, "learning_rate": 0.003, "loss": 4.0705, "step": 14861 }, { "epoch": 0.14862, "grad_norm": 1.0451287814201298, "learning_rate": 0.003, "loss": 4.0385, "step": 14862 }, { "epoch": 0.14863, "grad_norm": 1.5454792569609612, "learning_rate": 0.003, "loss": 4.0763, "step": 14863 }, { "epoch": 0.14864, "grad_norm": 0.8998383802889542, "learning_rate": 0.003, "loss": 4.0613, "step": 14864 }, { "epoch": 0.14865, "grad_norm": 1.1008145659534658, "learning_rate": 0.003, "loss": 4.0928, "step": 14865 }, { "epoch": 0.14866, "grad_norm": 1.4084066436583653, "learning_rate": 0.003, "loss": 4.0534, "step": 14866 }, { "epoch": 0.14867, "grad_norm": 1.1237411421162848, "learning_rate": 0.003, "loss": 4.0341, "step": 14867 }, { "epoch": 0.14868, "grad_norm": 1.2005867055587227, "learning_rate": 0.003, "loss": 4.0581, "step": 14868 }, { "epoch": 0.14869, "grad_norm": 1.1066652314635532, "learning_rate": 0.003, "loss": 4.0675, "step": 14869 }, { "epoch": 0.1487, "grad_norm": 0.9622231423033722, "learning_rate": 0.003, "loss": 4.0512, "step": 14870 }, { "epoch": 0.14871, "grad_norm": 1.1404170600749484, "learning_rate": 0.003, "loss": 4.076, "step": 14871 }, { "epoch": 0.14872, "grad_norm": 1.3609976338398784, "learning_rate": 0.003, "loss": 4.0798, "step": 14872 }, { "epoch": 0.14873, "grad_norm": 1.2893910141485294, "learning_rate": 0.003, "loss": 4.0856, "step": 14873 }, { "epoch": 0.14874, "grad_norm": 1.2931408987911757, "learning_rate": 0.003, "loss": 4.037, "step": 14874 }, { "epoch": 0.14875, "grad_norm": 1.471807512574448, "learning_rate": 0.003, "loss": 4.0713, "step": 14875 }, { "epoch": 0.14876, "grad_norm": 1.2090411680675062, "learning_rate": 0.003, "loss": 4.0719, "step": 14876 }, { "epoch": 0.14877, "grad_norm": 1.0829059112698427, "learning_rate": 0.003, "loss": 4.0798, "step": 14877 }, { "epoch": 0.14878, "grad_norm": 1.4129900132682789, "learning_rate": 0.003, "loss": 4.0703, "step": 14878 }, { "epoch": 0.14879, "grad_norm": 0.8776243325674575, "learning_rate": 0.003, "loss": 4.0782, "step": 14879 }, { "epoch": 0.1488, "grad_norm": 1.0485285841302954, "learning_rate": 0.003, "loss": 4.0486, "step": 14880 }, { "epoch": 0.14881, "grad_norm": 1.35981606046848, "learning_rate": 0.003, "loss": 4.058, "step": 14881 }, { "epoch": 0.14882, "grad_norm": 1.0436195240498212, "learning_rate": 0.003, "loss": 4.0763, "step": 14882 }, { "epoch": 0.14883, "grad_norm": 1.3226492069477507, "learning_rate": 0.003, "loss": 4.0722, "step": 14883 }, { "epoch": 0.14884, "grad_norm": 0.9986253500027842, "learning_rate": 0.003, "loss": 4.0609, "step": 14884 }, { "epoch": 0.14885, "grad_norm": 1.5201063299213164, "learning_rate": 0.003, "loss": 4.0725, "step": 14885 }, { "epoch": 0.14886, "grad_norm": 0.9757732233985466, "learning_rate": 0.003, "loss": 4.0461, "step": 14886 }, { "epoch": 0.14887, "grad_norm": 1.4476204357522973, "learning_rate": 0.003, "loss": 4.0675, "step": 14887 }, { "epoch": 0.14888, "grad_norm": 1.27074467688112, "learning_rate": 0.003, "loss": 4.1077, "step": 14888 }, { "epoch": 0.14889, "grad_norm": 1.1092463388578182, "learning_rate": 0.003, "loss": 4.0748, "step": 14889 }, { "epoch": 0.1489, "grad_norm": 1.214671239122057, "learning_rate": 0.003, "loss": 4.061, "step": 14890 }, { "epoch": 0.14891, "grad_norm": 1.0775666477090526, "learning_rate": 0.003, "loss": 4.0454, "step": 14891 }, { "epoch": 0.14892, "grad_norm": 1.1798038251945897, "learning_rate": 0.003, "loss": 4.0697, "step": 14892 }, { "epoch": 0.14893, "grad_norm": 1.046311938985408, "learning_rate": 0.003, "loss": 4.0528, "step": 14893 }, { "epoch": 0.14894, "grad_norm": 1.2238737392714125, "learning_rate": 0.003, "loss": 4.0803, "step": 14894 }, { "epoch": 0.14895, "grad_norm": 1.1095924412059364, "learning_rate": 0.003, "loss": 4.0901, "step": 14895 }, { "epoch": 0.14896, "grad_norm": 1.2527084956854004, "learning_rate": 0.003, "loss": 4.0722, "step": 14896 }, { "epoch": 0.14897, "grad_norm": 1.1403618171116803, "learning_rate": 0.003, "loss": 4.0694, "step": 14897 }, { "epoch": 0.14898, "grad_norm": 1.12014813866488, "learning_rate": 0.003, "loss": 4.0666, "step": 14898 }, { "epoch": 0.14899, "grad_norm": 1.1011878648338038, "learning_rate": 0.003, "loss": 4.0587, "step": 14899 }, { "epoch": 0.149, "grad_norm": 1.322568954041676, "learning_rate": 0.003, "loss": 4.0689, "step": 14900 }, { "epoch": 0.14901, "grad_norm": 1.1483799144543763, "learning_rate": 0.003, "loss": 4.0737, "step": 14901 }, { "epoch": 0.14902, "grad_norm": 1.710842623398834, "learning_rate": 0.003, "loss": 4.0877, "step": 14902 }, { "epoch": 0.14903, "grad_norm": 1.244171088897236, "learning_rate": 0.003, "loss": 4.0792, "step": 14903 }, { "epoch": 0.14904, "grad_norm": 1.2131362467533762, "learning_rate": 0.003, "loss": 4.0926, "step": 14904 }, { "epoch": 0.14905, "grad_norm": 1.2296471336790669, "learning_rate": 0.003, "loss": 4.0688, "step": 14905 }, { "epoch": 0.14906, "grad_norm": 1.2984408456967405, "learning_rate": 0.003, "loss": 4.0691, "step": 14906 }, { "epoch": 0.14907, "grad_norm": 1.0883845872709204, "learning_rate": 0.003, "loss": 4.0693, "step": 14907 }, { "epoch": 0.14908, "grad_norm": 1.0840404616703376, "learning_rate": 0.003, "loss": 4.0842, "step": 14908 }, { "epoch": 0.14909, "grad_norm": 1.181748263320111, "learning_rate": 0.003, "loss": 4.0634, "step": 14909 }, { "epoch": 0.1491, "grad_norm": 1.2663945605998859, "learning_rate": 0.003, "loss": 4.071, "step": 14910 }, { "epoch": 0.14911, "grad_norm": 1.187648049989691, "learning_rate": 0.003, "loss": 4.0847, "step": 14911 }, { "epoch": 0.14912, "grad_norm": 1.3811751228991052, "learning_rate": 0.003, "loss": 4.0826, "step": 14912 }, { "epoch": 0.14913, "grad_norm": 1.0219001247634045, "learning_rate": 0.003, "loss": 4.0427, "step": 14913 }, { "epoch": 0.14914, "grad_norm": 1.1593021387672653, "learning_rate": 0.003, "loss": 4.0636, "step": 14914 }, { "epoch": 0.14915, "grad_norm": 1.1568549264162649, "learning_rate": 0.003, "loss": 4.0584, "step": 14915 }, { "epoch": 0.14916, "grad_norm": 1.3202195197549011, "learning_rate": 0.003, "loss": 4.0752, "step": 14916 }, { "epoch": 0.14917, "grad_norm": 0.9020011592700239, "learning_rate": 0.003, "loss": 4.0844, "step": 14917 }, { "epoch": 0.14918, "grad_norm": 1.215424432279608, "learning_rate": 0.003, "loss": 4.0634, "step": 14918 }, { "epoch": 0.14919, "grad_norm": 1.1570968488310853, "learning_rate": 0.003, "loss": 4.0849, "step": 14919 }, { "epoch": 0.1492, "grad_norm": 1.1472721476899068, "learning_rate": 0.003, "loss": 4.0522, "step": 14920 }, { "epoch": 0.14921, "grad_norm": 1.0858959326819342, "learning_rate": 0.003, "loss": 4.0702, "step": 14921 }, { "epoch": 0.14922, "grad_norm": 1.2509352187724492, "learning_rate": 0.003, "loss": 4.0624, "step": 14922 }, { "epoch": 0.14923, "grad_norm": 1.2945174545370837, "learning_rate": 0.003, "loss": 4.0331, "step": 14923 }, { "epoch": 0.14924, "grad_norm": 1.1346005544615452, "learning_rate": 0.003, "loss": 4.041, "step": 14924 }, { "epoch": 0.14925, "grad_norm": 1.1845853865243938, "learning_rate": 0.003, "loss": 4.0548, "step": 14925 }, { "epoch": 0.14926, "grad_norm": 1.1090740148524028, "learning_rate": 0.003, "loss": 4.0494, "step": 14926 }, { "epoch": 0.14927, "grad_norm": 1.358250117436443, "learning_rate": 0.003, "loss": 4.0449, "step": 14927 }, { "epoch": 0.14928, "grad_norm": 1.3453422985843648, "learning_rate": 0.003, "loss": 4.0869, "step": 14928 }, { "epoch": 0.14929, "grad_norm": 1.1702428717059403, "learning_rate": 0.003, "loss": 4.0633, "step": 14929 }, { "epoch": 0.1493, "grad_norm": 1.1656251202564534, "learning_rate": 0.003, "loss": 4.0532, "step": 14930 }, { "epoch": 0.14931, "grad_norm": 1.098970939588012, "learning_rate": 0.003, "loss": 4.0537, "step": 14931 }, { "epoch": 0.14932, "grad_norm": 1.279436473773213, "learning_rate": 0.003, "loss": 4.0679, "step": 14932 }, { "epoch": 0.14933, "grad_norm": 1.0961977273192847, "learning_rate": 0.003, "loss": 4.0505, "step": 14933 }, { "epoch": 0.14934, "grad_norm": 1.5309509003949007, "learning_rate": 0.003, "loss": 4.0811, "step": 14934 }, { "epoch": 0.14935, "grad_norm": 1.1528254041188803, "learning_rate": 0.003, "loss": 4.0651, "step": 14935 }, { "epoch": 0.14936, "grad_norm": 1.2167835400582947, "learning_rate": 0.003, "loss": 4.0539, "step": 14936 }, { "epoch": 0.14937, "grad_norm": 1.3413070347149811, "learning_rate": 0.003, "loss": 4.0595, "step": 14937 }, { "epoch": 0.14938, "grad_norm": 1.0683864300767574, "learning_rate": 0.003, "loss": 4.0452, "step": 14938 }, { "epoch": 0.14939, "grad_norm": 1.2664966641085762, "learning_rate": 0.003, "loss": 4.0441, "step": 14939 }, { "epoch": 0.1494, "grad_norm": 1.0407237225742514, "learning_rate": 0.003, "loss": 4.059, "step": 14940 }, { "epoch": 0.14941, "grad_norm": 1.3931658434384966, "learning_rate": 0.003, "loss": 4.1025, "step": 14941 }, { "epoch": 0.14942, "grad_norm": 1.1109856813737475, "learning_rate": 0.003, "loss": 4.0411, "step": 14942 }, { "epoch": 0.14943, "grad_norm": 1.2908550200559337, "learning_rate": 0.003, "loss": 4.0644, "step": 14943 }, { "epoch": 0.14944, "grad_norm": 1.186703869958113, "learning_rate": 0.003, "loss": 4.0537, "step": 14944 }, { "epoch": 0.14945, "grad_norm": 1.1522749739993157, "learning_rate": 0.003, "loss": 4.0383, "step": 14945 }, { "epoch": 0.14946, "grad_norm": 1.0896949660175999, "learning_rate": 0.003, "loss": 4.0683, "step": 14946 }, { "epoch": 0.14947, "grad_norm": 1.3521263557717234, "learning_rate": 0.003, "loss": 4.0616, "step": 14947 }, { "epoch": 0.14948, "grad_norm": 1.1608496283133627, "learning_rate": 0.003, "loss": 4.0648, "step": 14948 }, { "epoch": 0.14949, "grad_norm": 1.2930816709791524, "learning_rate": 0.003, "loss": 4.0608, "step": 14949 }, { "epoch": 0.1495, "grad_norm": 0.985064802221724, "learning_rate": 0.003, "loss": 4.0761, "step": 14950 }, { "epoch": 0.14951, "grad_norm": 1.3801184624443306, "learning_rate": 0.003, "loss": 4.0304, "step": 14951 }, { "epoch": 0.14952, "grad_norm": 1.2195738258753523, "learning_rate": 0.003, "loss": 4.0654, "step": 14952 }, { "epoch": 0.14953, "grad_norm": 1.1345932271080388, "learning_rate": 0.003, "loss": 4.0592, "step": 14953 }, { "epoch": 0.14954, "grad_norm": 1.5557880953706804, "learning_rate": 0.003, "loss": 4.1055, "step": 14954 }, { "epoch": 0.14955, "grad_norm": 0.9071133427738178, "learning_rate": 0.003, "loss": 4.03, "step": 14955 }, { "epoch": 0.14956, "grad_norm": 1.468982543843281, "learning_rate": 0.003, "loss": 4.0552, "step": 14956 }, { "epoch": 0.14957, "grad_norm": 1.0423328924259037, "learning_rate": 0.003, "loss": 4.0455, "step": 14957 }, { "epoch": 0.14958, "grad_norm": 1.300583567325534, "learning_rate": 0.003, "loss": 4.0953, "step": 14958 }, { "epoch": 0.14959, "grad_norm": 1.0981910385097837, "learning_rate": 0.003, "loss": 4.0334, "step": 14959 }, { "epoch": 0.1496, "grad_norm": 1.0490751947643875, "learning_rate": 0.003, "loss": 4.0639, "step": 14960 }, { "epoch": 0.14961, "grad_norm": 1.2816163773619358, "learning_rate": 0.003, "loss": 4.0662, "step": 14961 }, { "epoch": 0.14962, "grad_norm": 1.1466675209377373, "learning_rate": 0.003, "loss": 4.0546, "step": 14962 }, { "epoch": 0.14963, "grad_norm": 1.290152566123964, "learning_rate": 0.003, "loss": 4.0684, "step": 14963 }, { "epoch": 0.14964, "grad_norm": 1.261020616938976, "learning_rate": 0.003, "loss": 4.0592, "step": 14964 }, { "epoch": 0.14965, "grad_norm": 1.4649546679785381, "learning_rate": 0.003, "loss": 4.0628, "step": 14965 }, { "epoch": 0.14966, "grad_norm": 1.1441136056324674, "learning_rate": 0.003, "loss": 4.0756, "step": 14966 }, { "epoch": 0.14967, "grad_norm": 1.2019750797486521, "learning_rate": 0.003, "loss": 4.0921, "step": 14967 }, { "epoch": 0.14968, "grad_norm": 1.2306693484236046, "learning_rate": 0.003, "loss": 4.0716, "step": 14968 }, { "epoch": 0.14969, "grad_norm": 1.2777053540849892, "learning_rate": 0.003, "loss": 4.0777, "step": 14969 }, { "epoch": 0.1497, "grad_norm": 1.1084346925337791, "learning_rate": 0.003, "loss": 4.0769, "step": 14970 }, { "epoch": 0.14971, "grad_norm": 1.3229335658679369, "learning_rate": 0.003, "loss": 4.0587, "step": 14971 }, { "epoch": 0.14972, "grad_norm": 1.095538056703894, "learning_rate": 0.003, "loss": 4.0708, "step": 14972 }, { "epoch": 0.14973, "grad_norm": 1.204405878347447, "learning_rate": 0.003, "loss": 4.0189, "step": 14973 }, { "epoch": 0.14974, "grad_norm": 1.2368040421511077, "learning_rate": 0.003, "loss": 4.0475, "step": 14974 }, { "epoch": 0.14975, "grad_norm": 1.4194541955071356, "learning_rate": 0.003, "loss": 4.0739, "step": 14975 }, { "epoch": 0.14976, "grad_norm": 1.0096839483393467, "learning_rate": 0.003, "loss": 4.0559, "step": 14976 }, { "epoch": 0.14977, "grad_norm": 1.3009380635532586, "learning_rate": 0.003, "loss": 4.0875, "step": 14977 }, { "epoch": 0.14978, "grad_norm": 1.0562909428594593, "learning_rate": 0.003, "loss": 4.0689, "step": 14978 }, { "epoch": 0.14979, "grad_norm": 1.2309819538095335, "learning_rate": 0.003, "loss": 4.0623, "step": 14979 }, { "epoch": 0.1498, "grad_norm": 1.1181451782884846, "learning_rate": 0.003, "loss": 4.0582, "step": 14980 }, { "epoch": 0.14981, "grad_norm": 1.1433863177365757, "learning_rate": 0.003, "loss": 4.0768, "step": 14981 }, { "epoch": 0.14982, "grad_norm": 1.2583042755829557, "learning_rate": 0.003, "loss": 4.0631, "step": 14982 }, { "epoch": 0.14983, "grad_norm": 1.1190272068614602, "learning_rate": 0.003, "loss": 4.0509, "step": 14983 }, { "epoch": 0.14984, "grad_norm": 1.4018613975193144, "learning_rate": 0.003, "loss": 4.0501, "step": 14984 }, { "epoch": 0.14985, "grad_norm": 1.0417216427534308, "learning_rate": 0.003, "loss": 4.0686, "step": 14985 }, { "epoch": 0.14986, "grad_norm": 1.134960157527509, "learning_rate": 0.003, "loss": 4.0569, "step": 14986 }, { "epoch": 0.14987, "grad_norm": 1.1174795370938495, "learning_rate": 0.003, "loss": 4.0519, "step": 14987 }, { "epoch": 0.14988, "grad_norm": 1.296639112026413, "learning_rate": 0.003, "loss": 4.0458, "step": 14988 }, { "epoch": 0.14989, "grad_norm": 1.3261095373498002, "learning_rate": 0.003, "loss": 4.0674, "step": 14989 }, { "epoch": 0.1499, "grad_norm": 1.2075226522964326, "learning_rate": 0.003, "loss": 4.0554, "step": 14990 }, { "epoch": 0.14991, "grad_norm": 1.008897068961723, "learning_rate": 0.003, "loss": 4.0481, "step": 14991 }, { "epoch": 0.14992, "grad_norm": 1.1942021518709378, "learning_rate": 0.003, "loss": 4.0453, "step": 14992 }, { "epoch": 0.14993, "grad_norm": 1.2211219368278374, "learning_rate": 0.003, "loss": 4.0516, "step": 14993 }, { "epoch": 0.14994, "grad_norm": 1.0870179584685973, "learning_rate": 0.003, "loss": 4.0681, "step": 14994 }, { "epoch": 0.14995, "grad_norm": 1.399608332516027, "learning_rate": 0.003, "loss": 4.0772, "step": 14995 }, { "epoch": 0.14996, "grad_norm": 1.0137612423797608, "learning_rate": 0.003, "loss": 4.0552, "step": 14996 }, { "epoch": 0.14997, "grad_norm": 1.3744351672232484, "learning_rate": 0.003, "loss": 4.0935, "step": 14997 }, { "epoch": 0.14998, "grad_norm": 0.9940359625848266, "learning_rate": 0.003, "loss": 4.0618, "step": 14998 }, { "epoch": 0.14999, "grad_norm": 1.4399413719162288, "learning_rate": 0.003, "loss": 4.0897, "step": 14999 }, { "epoch": 0.15, "grad_norm": 1.4974736632895418, "learning_rate": 0.003, "loss": 4.0882, "step": 15000 }, { "epoch": 0.15001, "grad_norm": 1.0668423995174225, "learning_rate": 0.003, "loss": 4.0622, "step": 15001 }, { "epoch": 0.15002, "grad_norm": 1.2110486697491223, "learning_rate": 0.003, "loss": 4.0425, "step": 15002 }, { "epoch": 0.15003, "grad_norm": 1.1795308324685607, "learning_rate": 0.003, "loss": 4.0913, "step": 15003 }, { "epoch": 0.15004, "grad_norm": 1.182856925873376, "learning_rate": 0.003, "loss": 4.06, "step": 15004 }, { "epoch": 0.15005, "grad_norm": 1.2054796107168688, "learning_rate": 0.003, "loss": 4.0925, "step": 15005 }, { "epoch": 0.15006, "grad_norm": 1.1233127843365838, "learning_rate": 0.003, "loss": 4.0952, "step": 15006 }, { "epoch": 0.15007, "grad_norm": 1.190740139784129, "learning_rate": 0.003, "loss": 4.0938, "step": 15007 }, { "epoch": 0.15008, "grad_norm": 1.1628208665821074, "learning_rate": 0.003, "loss": 4.0501, "step": 15008 }, { "epoch": 0.15009, "grad_norm": 1.184497287054005, "learning_rate": 0.003, "loss": 4.0401, "step": 15009 }, { "epoch": 0.1501, "grad_norm": 1.2170802962909146, "learning_rate": 0.003, "loss": 4.0509, "step": 15010 }, { "epoch": 0.15011, "grad_norm": 1.2448713816234198, "learning_rate": 0.003, "loss": 4.0682, "step": 15011 }, { "epoch": 0.15012, "grad_norm": 1.2795966193025246, "learning_rate": 0.003, "loss": 4.0464, "step": 15012 }, { "epoch": 0.15013, "grad_norm": 1.1253323596450409, "learning_rate": 0.003, "loss": 4.0733, "step": 15013 }, { "epoch": 0.15014, "grad_norm": 1.3692104101867797, "learning_rate": 0.003, "loss": 4.045, "step": 15014 }, { "epoch": 0.15015, "grad_norm": 1.2394377326991453, "learning_rate": 0.003, "loss": 4.0565, "step": 15015 }, { "epoch": 0.15016, "grad_norm": 1.1450312041451869, "learning_rate": 0.003, "loss": 4.0385, "step": 15016 }, { "epoch": 0.15017, "grad_norm": 1.1937362084722123, "learning_rate": 0.003, "loss": 4.0668, "step": 15017 }, { "epoch": 0.15018, "grad_norm": 1.288688457822522, "learning_rate": 0.003, "loss": 4.0548, "step": 15018 }, { "epoch": 0.15019, "grad_norm": 1.1409664304451768, "learning_rate": 0.003, "loss": 4.0584, "step": 15019 }, { "epoch": 0.1502, "grad_norm": 1.0873027688479973, "learning_rate": 0.003, "loss": 4.0507, "step": 15020 }, { "epoch": 0.15021, "grad_norm": 1.2665845047581592, "learning_rate": 0.003, "loss": 4.0702, "step": 15021 }, { "epoch": 0.15022, "grad_norm": 0.9745123669411565, "learning_rate": 0.003, "loss": 4.0796, "step": 15022 }, { "epoch": 0.15023, "grad_norm": 1.297378059616036, "learning_rate": 0.003, "loss": 4.0486, "step": 15023 }, { "epoch": 0.15024, "grad_norm": 1.0679672509082843, "learning_rate": 0.003, "loss": 4.0384, "step": 15024 }, { "epoch": 0.15025, "grad_norm": 1.3637170088458421, "learning_rate": 0.003, "loss": 4.0447, "step": 15025 }, { "epoch": 0.15026, "grad_norm": 1.204927160864958, "learning_rate": 0.003, "loss": 4.0405, "step": 15026 }, { "epoch": 0.15027, "grad_norm": 1.2156307138708313, "learning_rate": 0.003, "loss": 4.0808, "step": 15027 }, { "epoch": 0.15028, "grad_norm": 1.1531164890056576, "learning_rate": 0.003, "loss": 4.0719, "step": 15028 }, { "epoch": 0.15029, "grad_norm": 1.2553312157873282, "learning_rate": 0.003, "loss": 4.046, "step": 15029 }, { "epoch": 0.1503, "grad_norm": 0.9522168739397687, "learning_rate": 0.003, "loss": 4.0205, "step": 15030 }, { "epoch": 0.15031, "grad_norm": 1.226122357521874, "learning_rate": 0.003, "loss": 4.0527, "step": 15031 }, { "epoch": 0.15032, "grad_norm": 1.5474125500292253, "learning_rate": 0.003, "loss": 4.0409, "step": 15032 }, { "epoch": 0.15033, "grad_norm": 1.2647460146585578, "learning_rate": 0.003, "loss": 4.0536, "step": 15033 }, { "epoch": 0.15034, "grad_norm": 1.3907590251108537, "learning_rate": 0.003, "loss": 4.0769, "step": 15034 }, { "epoch": 0.15035, "grad_norm": 1.0234787803166052, "learning_rate": 0.003, "loss": 4.0554, "step": 15035 }, { "epoch": 0.15036, "grad_norm": 1.337378293856488, "learning_rate": 0.003, "loss": 4.064, "step": 15036 }, { "epoch": 0.15037, "grad_norm": 1.1095014826228464, "learning_rate": 0.003, "loss": 4.0917, "step": 15037 }, { "epoch": 0.15038, "grad_norm": 1.2913904076297182, "learning_rate": 0.003, "loss": 4.0785, "step": 15038 }, { "epoch": 0.15039, "grad_norm": 1.0887255711713681, "learning_rate": 0.003, "loss": 4.0669, "step": 15039 }, { "epoch": 0.1504, "grad_norm": 1.2547438431290638, "learning_rate": 0.003, "loss": 4.0658, "step": 15040 }, { "epoch": 0.15041, "grad_norm": 1.4240124528037976, "learning_rate": 0.003, "loss": 4.1082, "step": 15041 }, { "epoch": 0.15042, "grad_norm": 1.1673943418670996, "learning_rate": 0.003, "loss": 4.0355, "step": 15042 }, { "epoch": 0.15043, "grad_norm": 1.0867675070481164, "learning_rate": 0.003, "loss": 4.0709, "step": 15043 }, { "epoch": 0.15044, "grad_norm": 1.1782378462470475, "learning_rate": 0.003, "loss": 4.058, "step": 15044 }, { "epoch": 0.15045, "grad_norm": 1.0872039407960756, "learning_rate": 0.003, "loss": 4.0742, "step": 15045 }, { "epoch": 0.15046, "grad_norm": 1.3736289236684958, "learning_rate": 0.003, "loss": 4.0841, "step": 15046 }, { "epoch": 0.15047, "grad_norm": 1.1308513861955674, "learning_rate": 0.003, "loss": 4.051, "step": 15047 }, { "epoch": 0.15048, "grad_norm": 1.2090821969797114, "learning_rate": 0.003, "loss": 4.0551, "step": 15048 }, { "epoch": 0.15049, "grad_norm": 1.1373820398442709, "learning_rate": 0.003, "loss": 4.0853, "step": 15049 }, { "epoch": 0.1505, "grad_norm": 1.0984446995600496, "learning_rate": 0.003, "loss": 4.0715, "step": 15050 }, { "epoch": 0.15051, "grad_norm": 1.155054800503449, "learning_rate": 0.003, "loss": 4.0302, "step": 15051 }, { "epoch": 0.15052, "grad_norm": 1.3827231447355273, "learning_rate": 0.003, "loss": 4.0801, "step": 15052 }, { "epoch": 0.15053, "grad_norm": 1.1833574261390736, "learning_rate": 0.003, "loss": 4.0814, "step": 15053 }, { "epoch": 0.15054, "grad_norm": 1.195554739703261, "learning_rate": 0.003, "loss": 4.0761, "step": 15054 }, { "epoch": 0.15055, "grad_norm": 1.2453938734519396, "learning_rate": 0.003, "loss": 4.0631, "step": 15055 }, { "epoch": 0.15056, "grad_norm": 1.2110376922895725, "learning_rate": 0.003, "loss": 4.0174, "step": 15056 }, { "epoch": 0.15057, "grad_norm": 1.2151309358317508, "learning_rate": 0.003, "loss": 4.0691, "step": 15057 }, { "epoch": 0.15058, "grad_norm": 1.259729714323035, "learning_rate": 0.003, "loss": 4.0914, "step": 15058 }, { "epoch": 0.15059, "grad_norm": 1.1680098473605123, "learning_rate": 0.003, "loss": 4.0569, "step": 15059 }, { "epoch": 0.1506, "grad_norm": 1.0365628276921912, "learning_rate": 0.003, "loss": 4.0441, "step": 15060 }, { "epoch": 0.15061, "grad_norm": 1.1860587795685422, "learning_rate": 0.003, "loss": 4.0818, "step": 15061 }, { "epoch": 0.15062, "grad_norm": 1.1652134132806091, "learning_rate": 0.003, "loss": 4.0722, "step": 15062 }, { "epoch": 0.15063, "grad_norm": 1.1893670897157664, "learning_rate": 0.003, "loss": 4.078, "step": 15063 }, { "epoch": 0.15064, "grad_norm": 1.2641455031540747, "learning_rate": 0.003, "loss": 4.0586, "step": 15064 }, { "epoch": 0.15065, "grad_norm": 1.179907937763059, "learning_rate": 0.003, "loss": 4.0539, "step": 15065 }, { "epoch": 0.15066, "grad_norm": 1.256981536680123, "learning_rate": 0.003, "loss": 4.083, "step": 15066 }, { "epoch": 0.15067, "grad_norm": 1.0820909858204335, "learning_rate": 0.003, "loss": 4.0907, "step": 15067 }, { "epoch": 0.15068, "grad_norm": 1.184433524088628, "learning_rate": 0.003, "loss": 4.0592, "step": 15068 }, { "epoch": 0.15069, "grad_norm": 1.2716471033861365, "learning_rate": 0.003, "loss": 4.0648, "step": 15069 }, { "epoch": 0.1507, "grad_norm": 1.1683331028574926, "learning_rate": 0.003, "loss": 4.0421, "step": 15070 }, { "epoch": 0.15071, "grad_norm": 1.1797972267504941, "learning_rate": 0.003, "loss": 4.0637, "step": 15071 }, { "epoch": 0.15072, "grad_norm": 1.1302381466298992, "learning_rate": 0.003, "loss": 4.0681, "step": 15072 }, { "epoch": 0.15073, "grad_norm": 1.355350169867995, "learning_rate": 0.003, "loss": 4.0521, "step": 15073 }, { "epoch": 0.15074, "grad_norm": 1.044651911324449, "learning_rate": 0.003, "loss": 4.0666, "step": 15074 }, { "epoch": 0.15075, "grad_norm": 1.3046654742749575, "learning_rate": 0.003, "loss": 4.0743, "step": 15075 }, { "epoch": 0.15076, "grad_norm": 1.1648922397030492, "learning_rate": 0.003, "loss": 4.0471, "step": 15076 }, { "epoch": 0.15077, "grad_norm": 1.2806126751874845, "learning_rate": 0.003, "loss": 4.0343, "step": 15077 }, { "epoch": 0.15078, "grad_norm": 1.3528938353220739, "learning_rate": 0.003, "loss": 4.044, "step": 15078 }, { "epoch": 0.15079, "grad_norm": 1.1861142100163455, "learning_rate": 0.003, "loss": 4.0745, "step": 15079 }, { "epoch": 0.1508, "grad_norm": 1.0589951871882235, "learning_rate": 0.003, "loss": 4.0736, "step": 15080 }, { "epoch": 0.15081, "grad_norm": 1.2318643421068618, "learning_rate": 0.003, "loss": 4.0912, "step": 15081 }, { "epoch": 0.15082, "grad_norm": 1.123727186298089, "learning_rate": 0.003, "loss": 4.0651, "step": 15082 }, { "epoch": 0.15083, "grad_norm": 1.3178690838827816, "learning_rate": 0.003, "loss": 4.0491, "step": 15083 }, { "epoch": 0.15084, "grad_norm": 1.0237784168087798, "learning_rate": 0.003, "loss": 4.0533, "step": 15084 }, { "epoch": 0.15085, "grad_norm": 1.492155585068999, "learning_rate": 0.003, "loss": 4.0813, "step": 15085 }, { "epoch": 0.15086, "grad_norm": 1.1925963482272997, "learning_rate": 0.003, "loss": 4.0311, "step": 15086 }, { "epoch": 0.15087, "grad_norm": 1.4319500990788507, "learning_rate": 0.003, "loss": 4.0722, "step": 15087 }, { "epoch": 0.15088, "grad_norm": 1.060355543294847, "learning_rate": 0.003, "loss": 4.0555, "step": 15088 }, { "epoch": 0.15089, "grad_norm": 1.3104618148468035, "learning_rate": 0.003, "loss": 4.064, "step": 15089 }, { "epoch": 0.1509, "grad_norm": 1.2251162778987241, "learning_rate": 0.003, "loss": 4.0442, "step": 15090 }, { "epoch": 0.15091, "grad_norm": 1.306080204600944, "learning_rate": 0.003, "loss": 4.0861, "step": 15091 }, { "epoch": 0.15092, "grad_norm": 1.0915870298734536, "learning_rate": 0.003, "loss": 4.0849, "step": 15092 }, { "epoch": 0.15093, "grad_norm": 1.2119632574467158, "learning_rate": 0.003, "loss": 4.0406, "step": 15093 }, { "epoch": 0.15094, "grad_norm": 1.1595876781351768, "learning_rate": 0.003, "loss": 4.0603, "step": 15094 }, { "epoch": 0.15095, "grad_norm": 1.119325709139666, "learning_rate": 0.003, "loss": 4.0683, "step": 15095 }, { "epoch": 0.15096, "grad_norm": 1.2044471540428243, "learning_rate": 0.003, "loss": 4.0444, "step": 15096 }, { "epoch": 0.15097, "grad_norm": 1.4245175575214675, "learning_rate": 0.003, "loss": 4.0592, "step": 15097 }, { "epoch": 0.15098, "grad_norm": 1.5517681537973362, "learning_rate": 0.003, "loss": 4.0549, "step": 15098 }, { "epoch": 0.15099, "grad_norm": 1.0067593744034253, "learning_rate": 0.003, "loss": 4.0505, "step": 15099 }, { "epoch": 0.151, "grad_norm": 1.2089884789065866, "learning_rate": 0.003, "loss": 4.0358, "step": 15100 }, { "epoch": 0.15101, "grad_norm": 1.105057199052122, "learning_rate": 0.003, "loss": 4.0462, "step": 15101 }, { "epoch": 0.15102, "grad_norm": 1.3101763159722803, "learning_rate": 0.003, "loss": 4.0791, "step": 15102 }, { "epoch": 0.15103, "grad_norm": 1.0291772237772652, "learning_rate": 0.003, "loss": 4.0653, "step": 15103 }, { "epoch": 0.15104, "grad_norm": 1.3908177507943922, "learning_rate": 0.003, "loss": 4.0439, "step": 15104 }, { "epoch": 0.15105, "grad_norm": 0.993120529030314, "learning_rate": 0.003, "loss": 4.0501, "step": 15105 }, { "epoch": 0.15106, "grad_norm": 1.2450894355942614, "learning_rate": 0.003, "loss": 4.0625, "step": 15106 }, { "epoch": 0.15107, "grad_norm": 1.056113877015572, "learning_rate": 0.003, "loss": 4.0553, "step": 15107 }, { "epoch": 0.15108, "grad_norm": 1.1662206734963574, "learning_rate": 0.003, "loss": 4.0692, "step": 15108 }, { "epoch": 0.15109, "grad_norm": 1.2621098326636795, "learning_rate": 0.003, "loss": 4.0625, "step": 15109 }, { "epoch": 0.1511, "grad_norm": 1.3570631825072197, "learning_rate": 0.003, "loss": 4.0365, "step": 15110 }, { "epoch": 0.15111, "grad_norm": 1.3311454757868875, "learning_rate": 0.003, "loss": 4.0691, "step": 15111 }, { "epoch": 0.15112, "grad_norm": 1.2989363196563721, "learning_rate": 0.003, "loss": 4.0529, "step": 15112 }, { "epoch": 0.15113, "grad_norm": 0.9189229359882644, "learning_rate": 0.003, "loss": 4.0608, "step": 15113 }, { "epoch": 0.15114, "grad_norm": 1.240809467869954, "learning_rate": 0.003, "loss": 4.0465, "step": 15114 }, { "epoch": 0.15115, "grad_norm": 1.1928550783856977, "learning_rate": 0.003, "loss": 4.0548, "step": 15115 }, { "epoch": 0.15116, "grad_norm": 1.2624829089617984, "learning_rate": 0.003, "loss": 4.07, "step": 15116 }, { "epoch": 0.15117, "grad_norm": 1.0273366211589472, "learning_rate": 0.003, "loss": 4.0511, "step": 15117 }, { "epoch": 0.15118, "grad_norm": 1.3046658005508183, "learning_rate": 0.003, "loss": 4.06, "step": 15118 }, { "epoch": 0.15119, "grad_norm": 1.159606861497468, "learning_rate": 0.003, "loss": 4.0753, "step": 15119 }, { "epoch": 0.1512, "grad_norm": 1.3059467524421076, "learning_rate": 0.003, "loss": 4.0899, "step": 15120 }, { "epoch": 0.15121, "grad_norm": 1.029847433937725, "learning_rate": 0.003, "loss": 4.0373, "step": 15121 }, { "epoch": 0.15122, "grad_norm": 1.4482892460541101, "learning_rate": 0.003, "loss": 4.0701, "step": 15122 }, { "epoch": 0.15123, "grad_norm": 1.0429287950728148, "learning_rate": 0.003, "loss": 4.043, "step": 15123 }, { "epoch": 0.15124, "grad_norm": 1.2259003329089468, "learning_rate": 0.003, "loss": 4.0376, "step": 15124 }, { "epoch": 0.15125, "grad_norm": 1.278091232104381, "learning_rate": 0.003, "loss": 4.0792, "step": 15125 }, { "epoch": 0.15126, "grad_norm": 1.1938494318875001, "learning_rate": 0.003, "loss": 4.0691, "step": 15126 }, { "epoch": 0.15127, "grad_norm": 1.2589849654916394, "learning_rate": 0.003, "loss": 4.0812, "step": 15127 }, { "epoch": 0.15128, "grad_norm": 1.1330547145294245, "learning_rate": 0.003, "loss": 4.0769, "step": 15128 }, { "epoch": 0.15129, "grad_norm": 1.097005960935907, "learning_rate": 0.003, "loss": 4.0864, "step": 15129 }, { "epoch": 0.1513, "grad_norm": 1.1729004214319463, "learning_rate": 0.003, "loss": 4.0413, "step": 15130 }, { "epoch": 0.15131, "grad_norm": 1.1726540565553825, "learning_rate": 0.003, "loss": 4.0599, "step": 15131 }, { "epoch": 0.15132, "grad_norm": 1.2424736420971347, "learning_rate": 0.003, "loss": 4.0691, "step": 15132 }, { "epoch": 0.15133, "grad_norm": 1.1448268613315058, "learning_rate": 0.003, "loss": 4.0571, "step": 15133 }, { "epoch": 0.15134, "grad_norm": 1.5708460693967223, "learning_rate": 0.003, "loss": 4.0586, "step": 15134 }, { "epoch": 0.15135, "grad_norm": 1.040378867053243, "learning_rate": 0.003, "loss": 4.0527, "step": 15135 }, { "epoch": 0.15136, "grad_norm": 1.4040593854991856, "learning_rate": 0.003, "loss": 4.0318, "step": 15136 }, { "epoch": 0.15137, "grad_norm": 1.2011938592676896, "learning_rate": 0.003, "loss": 4.0475, "step": 15137 }, { "epoch": 0.15138, "grad_norm": 1.2367047396246862, "learning_rate": 0.003, "loss": 4.1022, "step": 15138 }, { "epoch": 0.15139, "grad_norm": 1.1293395957439654, "learning_rate": 0.003, "loss": 4.0533, "step": 15139 }, { "epoch": 0.1514, "grad_norm": 1.3090085398590288, "learning_rate": 0.003, "loss": 4.0463, "step": 15140 }, { "epoch": 0.15141, "grad_norm": 1.0075436032183622, "learning_rate": 0.003, "loss": 4.0486, "step": 15141 }, { "epoch": 0.15142, "grad_norm": 1.33714212691153, "learning_rate": 0.003, "loss": 4.0713, "step": 15142 }, { "epoch": 0.15143, "grad_norm": 1.1663913366871692, "learning_rate": 0.003, "loss": 4.0531, "step": 15143 }, { "epoch": 0.15144, "grad_norm": 1.2132377498419114, "learning_rate": 0.003, "loss": 4.0737, "step": 15144 }, { "epoch": 0.15145, "grad_norm": 1.2259208725380872, "learning_rate": 0.003, "loss": 4.0459, "step": 15145 }, { "epoch": 0.15146, "grad_norm": 1.297997547043007, "learning_rate": 0.003, "loss": 4.0732, "step": 15146 }, { "epoch": 0.15147, "grad_norm": 1.1330532631224162, "learning_rate": 0.003, "loss": 4.0406, "step": 15147 }, { "epoch": 0.15148, "grad_norm": 1.2123133697932726, "learning_rate": 0.003, "loss": 4.0649, "step": 15148 }, { "epoch": 0.15149, "grad_norm": 1.2721907366832415, "learning_rate": 0.003, "loss": 4.0867, "step": 15149 }, { "epoch": 0.1515, "grad_norm": 1.2258759438061497, "learning_rate": 0.003, "loss": 4.0685, "step": 15150 }, { "epoch": 0.15151, "grad_norm": 1.2837753617747198, "learning_rate": 0.003, "loss": 4.0539, "step": 15151 }, { "epoch": 0.15152, "grad_norm": 1.0306162316863614, "learning_rate": 0.003, "loss": 4.0983, "step": 15152 }, { "epoch": 0.15153, "grad_norm": 1.5644759764698157, "learning_rate": 0.003, "loss": 4.0774, "step": 15153 }, { "epoch": 0.15154, "grad_norm": 0.9980410349143073, "learning_rate": 0.003, "loss": 4.0452, "step": 15154 }, { "epoch": 0.15155, "grad_norm": 1.3128438968987235, "learning_rate": 0.003, "loss": 4.0752, "step": 15155 }, { "epoch": 0.15156, "grad_norm": 1.044644352242352, "learning_rate": 0.003, "loss": 4.0659, "step": 15156 }, { "epoch": 0.15157, "grad_norm": 1.190633698957856, "learning_rate": 0.003, "loss": 4.0593, "step": 15157 }, { "epoch": 0.15158, "grad_norm": 1.1669918387521137, "learning_rate": 0.003, "loss": 4.0663, "step": 15158 }, { "epoch": 0.15159, "grad_norm": 1.222507008110017, "learning_rate": 0.003, "loss": 4.0782, "step": 15159 }, { "epoch": 0.1516, "grad_norm": 1.1100243033486945, "learning_rate": 0.003, "loss": 4.0489, "step": 15160 }, { "epoch": 0.15161, "grad_norm": 1.2258618521306344, "learning_rate": 0.003, "loss": 4.0615, "step": 15161 }, { "epoch": 0.15162, "grad_norm": 1.4331452700729912, "learning_rate": 0.003, "loss": 4.0986, "step": 15162 }, { "epoch": 0.15163, "grad_norm": 1.4096854957474376, "learning_rate": 0.003, "loss": 4.0633, "step": 15163 }, { "epoch": 0.15164, "grad_norm": 1.0366396321569484, "learning_rate": 0.003, "loss": 4.0528, "step": 15164 }, { "epoch": 0.15165, "grad_norm": 1.192406791153559, "learning_rate": 0.003, "loss": 4.035, "step": 15165 }, { "epoch": 0.15166, "grad_norm": 1.2189832259548985, "learning_rate": 0.003, "loss": 4.0368, "step": 15166 }, { "epoch": 0.15167, "grad_norm": 0.9551976170429402, "learning_rate": 0.003, "loss": 4.0482, "step": 15167 }, { "epoch": 0.15168, "grad_norm": 1.119167860282972, "learning_rate": 0.003, "loss": 4.0546, "step": 15168 }, { "epoch": 0.15169, "grad_norm": 1.1447472033605182, "learning_rate": 0.003, "loss": 4.0472, "step": 15169 }, { "epoch": 0.1517, "grad_norm": 1.2429156332586646, "learning_rate": 0.003, "loss": 4.0424, "step": 15170 }, { "epoch": 0.15171, "grad_norm": 1.298998789476462, "learning_rate": 0.003, "loss": 4.0626, "step": 15171 }, { "epoch": 0.15172, "grad_norm": 1.3241459773471154, "learning_rate": 0.003, "loss": 4.0575, "step": 15172 }, { "epoch": 0.15173, "grad_norm": 1.464002389595656, "learning_rate": 0.003, "loss": 4.0526, "step": 15173 }, { "epoch": 0.15174, "grad_norm": 1.1619733096151428, "learning_rate": 0.003, "loss": 4.0485, "step": 15174 }, { "epoch": 0.15175, "grad_norm": 1.220311346333655, "learning_rate": 0.003, "loss": 4.0681, "step": 15175 }, { "epoch": 0.15176, "grad_norm": 1.1774755736623603, "learning_rate": 0.003, "loss": 4.0496, "step": 15176 }, { "epoch": 0.15177, "grad_norm": 1.270705325277135, "learning_rate": 0.003, "loss": 4.0519, "step": 15177 }, { "epoch": 0.15178, "grad_norm": 1.0688786362101994, "learning_rate": 0.003, "loss": 4.0282, "step": 15178 }, { "epoch": 0.15179, "grad_norm": 1.433993614691496, "learning_rate": 0.003, "loss": 4.058, "step": 15179 }, { "epoch": 0.1518, "grad_norm": 1.045346819367408, "learning_rate": 0.003, "loss": 4.0369, "step": 15180 }, { "epoch": 0.15181, "grad_norm": 1.370623206374627, "learning_rate": 0.003, "loss": 4.049, "step": 15181 }, { "epoch": 0.15182, "grad_norm": 1.1695371902586094, "learning_rate": 0.003, "loss": 4.0726, "step": 15182 }, { "epoch": 0.15183, "grad_norm": 1.3886462368807206, "learning_rate": 0.003, "loss": 4.0423, "step": 15183 }, { "epoch": 0.15184, "grad_norm": 0.9645218148634334, "learning_rate": 0.003, "loss": 4.0631, "step": 15184 }, { "epoch": 0.15185, "grad_norm": 1.2966774954952045, "learning_rate": 0.003, "loss": 4.0709, "step": 15185 }, { "epoch": 0.15186, "grad_norm": 1.2335208988336277, "learning_rate": 0.003, "loss": 4.0566, "step": 15186 }, { "epoch": 0.15187, "grad_norm": 1.0050835608908404, "learning_rate": 0.003, "loss": 4.0719, "step": 15187 }, { "epoch": 0.15188, "grad_norm": 1.3701899653186935, "learning_rate": 0.003, "loss": 4.0754, "step": 15188 }, { "epoch": 0.15189, "grad_norm": 1.135170710607172, "learning_rate": 0.003, "loss": 4.0625, "step": 15189 }, { "epoch": 0.1519, "grad_norm": 1.2432212879200064, "learning_rate": 0.003, "loss": 4.0821, "step": 15190 }, { "epoch": 0.15191, "grad_norm": 1.1103180631951515, "learning_rate": 0.003, "loss": 4.0948, "step": 15191 }, { "epoch": 0.15192, "grad_norm": 1.2425724660435107, "learning_rate": 0.003, "loss": 4.063, "step": 15192 }, { "epoch": 0.15193, "grad_norm": 1.3533195363117116, "learning_rate": 0.003, "loss": 4.0645, "step": 15193 }, { "epoch": 0.15194, "grad_norm": 1.3515973455544246, "learning_rate": 0.003, "loss": 4.0708, "step": 15194 }, { "epoch": 0.15195, "grad_norm": 1.0261684737287875, "learning_rate": 0.003, "loss": 4.0426, "step": 15195 }, { "epoch": 0.15196, "grad_norm": 1.5674528909788603, "learning_rate": 0.003, "loss": 4.0932, "step": 15196 }, { "epoch": 0.15197, "grad_norm": 1.3803955355175979, "learning_rate": 0.003, "loss": 4.0611, "step": 15197 }, { "epoch": 0.15198, "grad_norm": 1.5062575871981334, "learning_rate": 0.003, "loss": 4.081, "step": 15198 }, { "epoch": 0.15199, "grad_norm": 1.024630024329231, "learning_rate": 0.003, "loss": 4.0703, "step": 15199 }, { "epoch": 0.152, "grad_norm": 1.213495646655414, "learning_rate": 0.003, "loss": 4.0705, "step": 15200 }, { "epoch": 0.15201, "grad_norm": 1.4510169131842288, "learning_rate": 0.003, "loss": 4.07, "step": 15201 }, { "epoch": 0.15202, "grad_norm": 1.1302118547381454, "learning_rate": 0.003, "loss": 4.0668, "step": 15202 }, { "epoch": 0.15203, "grad_norm": 1.2725092157742763, "learning_rate": 0.003, "loss": 4.0547, "step": 15203 }, { "epoch": 0.15204, "grad_norm": 1.137982520437394, "learning_rate": 0.003, "loss": 4.0905, "step": 15204 }, { "epoch": 0.15205, "grad_norm": 1.0223284267568893, "learning_rate": 0.003, "loss": 4.0616, "step": 15205 }, { "epoch": 0.15206, "grad_norm": 1.0565281836081886, "learning_rate": 0.003, "loss": 4.0456, "step": 15206 }, { "epoch": 0.15207, "grad_norm": 1.3043821397289166, "learning_rate": 0.003, "loss": 4.0422, "step": 15207 }, { "epoch": 0.15208, "grad_norm": 1.105080361488173, "learning_rate": 0.003, "loss": 4.0264, "step": 15208 }, { "epoch": 0.15209, "grad_norm": 1.1399072306971487, "learning_rate": 0.003, "loss": 4.0829, "step": 15209 }, { "epoch": 0.1521, "grad_norm": 1.1386602307645344, "learning_rate": 0.003, "loss": 4.0666, "step": 15210 }, { "epoch": 0.15211, "grad_norm": 1.0134889061744425, "learning_rate": 0.003, "loss": 4.0729, "step": 15211 }, { "epoch": 0.15212, "grad_norm": 1.2881827316598067, "learning_rate": 0.003, "loss": 4.0699, "step": 15212 }, { "epoch": 0.15213, "grad_norm": 0.9346737937438413, "learning_rate": 0.003, "loss": 4.0364, "step": 15213 }, { "epoch": 0.15214, "grad_norm": 1.2436543960136706, "learning_rate": 0.003, "loss": 4.0616, "step": 15214 }, { "epoch": 0.15215, "grad_norm": 1.1459849786740162, "learning_rate": 0.003, "loss": 4.051, "step": 15215 }, { "epoch": 0.15216, "grad_norm": 1.2205207180453597, "learning_rate": 0.003, "loss": 4.0372, "step": 15216 }, { "epoch": 0.15217, "grad_norm": 1.5972490115073228, "learning_rate": 0.003, "loss": 4.067, "step": 15217 }, { "epoch": 0.15218, "grad_norm": 1.0397158701269222, "learning_rate": 0.003, "loss": 4.0372, "step": 15218 }, { "epoch": 0.15219, "grad_norm": 1.454284438026809, "learning_rate": 0.003, "loss": 4.0747, "step": 15219 }, { "epoch": 0.1522, "grad_norm": 1.0013808946685574, "learning_rate": 0.003, "loss": 4.066, "step": 15220 }, { "epoch": 0.15221, "grad_norm": 1.3739928964687649, "learning_rate": 0.003, "loss": 4.0869, "step": 15221 }, { "epoch": 0.15222, "grad_norm": 1.0891144380179014, "learning_rate": 0.003, "loss": 4.0686, "step": 15222 }, { "epoch": 0.15223, "grad_norm": 1.526998946193764, "learning_rate": 0.003, "loss": 4.0429, "step": 15223 }, { "epoch": 0.15224, "grad_norm": 1.330605515094238, "learning_rate": 0.003, "loss": 4.0849, "step": 15224 }, { "epoch": 0.15225, "grad_norm": 0.9326977690364996, "learning_rate": 0.003, "loss": 4.0571, "step": 15225 }, { "epoch": 0.15226, "grad_norm": 1.299351030717249, "learning_rate": 0.003, "loss": 4.0675, "step": 15226 }, { "epoch": 0.15227, "grad_norm": 1.205549490078649, "learning_rate": 0.003, "loss": 4.0685, "step": 15227 }, { "epoch": 0.15228, "grad_norm": 1.3237257889809773, "learning_rate": 0.003, "loss": 4.0474, "step": 15228 }, { "epoch": 0.15229, "grad_norm": 1.1757579002320533, "learning_rate": 0.003, "loss": 4.0268, "step": 15229 }, { "epoch": 0.1523, "grad_norm": 1.0704092999538808, "learning_rate": 0.003, "loss": 4.0826, "step": 15230 }, { "epoch": 0.15231, "grad_norm": 1.3457605300447015, "learning_rate": 0.003, "loss": 4.0764, "step": 15231 }, { "epoch": 0.15232, "grad_norm": 1.0255019166629449, "learning_rate": 0.003, "loss": 4.0472, "step": 15232 }, { "epoch": 0.15233, "grad_norm": 1.4300673251118192, "learning_rate": 0.003, "loss": 4.0566, "step": 15233 }, { "epoch": 0.15234, "grad_norm": 1.0734449075243604, "learning_rate": 0.003, "loss": 4.0568, "step": 15234 }, { "epoch": 0.15235, "grad_norm": 1.473003403443299, "learning_rate": 0.003, "loss": 4.0534, "step": 15235 }, { "epoch": 0.15236, "grad_norm": 1.0636588121743216, "learning_rate": 0.003, "loss": 4.0866, "step": 15236 }, { "epoch": 0.15237, "grad_norm": 1.216439505946412, "learning_rate": 0.003, "loss": 4.0584, "step": 15237 }, { "epoch": 0.15238, "grad_norm": 1.1762090328789125, "learning_rate": 0.003, "loss": 4.043, "step": 15238 }, { "epoch": 0.15239, "grad_norm": 1.2784513886299516, "learning_rate": 0.003, "loss": 4.0493, "step": 15239 }, { "epoch": 0.1524, "grad_norm": 1.0374994043846635, "learning_rate": 0.003, "loss": 4.0575, "step": 15240 }, { "epoch": 0.15241, "grad_norm": 1.221527477801365, "learning_rate": 0.003, "loss": 4.0893, "step": 15241 }, { "epoch": 0.15242, "grad_norm": 1.4062968215005007, "learning_rate": 0.003, "loss": 4.0924, "step": 15242 }, { "epoch": 0.15243, "grad_norm": 1.0214884999824123, "learning_rate": 0.003, "loss": 4.0641, "step": 15243 }, { "epoch": 0.15244, "grad_norm": 1.263794425337306, "learning_rate": 0.003, "loss": 4.0825, "step": 15244 }, { "epoch": 0.15245, "grad_norm": 1.0302338675328382, "learning_rate": 0.003, "loss": 4.0635, "step": 15245 }, { "epoch": 0.15246, "grad_norm": 1.136932342643618, "learning_rate": 0.003, "loss": 4.0514, "step": 15246 }, { "epoch": 0.15247, "grad_norm": 1.3867514442765265, "learning_rate": 0.003, "loss": 4.059, "step": 15247 }, { "epoch": 0.15248, "grad_norm": 1.343751673856623, "learning_rate": 0.003, "loss": 4.0845, "step": 15248 }, { "epoch": 0.15249, "grad_norm": 1.2947833796560833, "learning_rate": 0.003, "loss": 4.0679, "step": 15249 }, { "epoch": 0.1525, "grad_norm": 0.8717763519936194, "learning_rate": 0.003, "loss": 4.0687, "step": 15250 }, { "epoch": 0.15251, "grad_norm": 1.2661112616588408, "learning_rate": 0.003, "loss": 4.1, "step": 15251 }, { "epoch": 0.15252, "grad_norm": 1.4149494227736545, "learning_rate": 0.003, "loss": 4.0928, "step": 15252 }, { "epoch": 0.15253, "grad_norm": 1.1173745271801705, "learning_rate": 0.003, "loss": 4.0361, "step": 15253 }, { "epoch": 0.15254, "grad_norm": 1.0403524152240142, "learning_rate": 0.003, "loss": 4.0572, "step": 15254 }, { "epoch": 0.15255, "grad_norm": 1.3111776704502551, "learning_rate": 0.003, "loss": 4.0602, "step": 15255 }, { "epoch": 0.15256, "grad_norm": 1.0953509594301363, "learning_rate": 0.003, "loss": 4.0602, "step": 15256 }, { "epoch": 0.15257, "grad_norm": 1.1169720379175168, "learning_rate": 0.003, "loss": 4.0524, "step": 15257 }, { "epoch": 0.15258, "grad_norm": 1.301900403801355, "learning_rate": 0.003, "loss": 4.0709, "step": 15258 }, { "epoch": 0.15259, "grad_norm": 1.106392313755815, "learning_rate": 0.003, "loss": 4.0596, "step": 15259 }, { "epoch": 0.1526, "grad_norm": 1.7125824444678244, "learning_rate": 0.003, "loss": 4.0417, "step": 15260 }, { "epoch": 0.15261, "grad_norm": 1.249930993267981, "learning_rate": 0.003, "loss": 4.0717, "step": 15261 }, { "epoch": 0.15262, "grad_norm": 1.149457244300075, "learning_rate": 0.003, "loss": 4.0672, "step": 15262 }, { "epoch": 0.15263, "grad_norm": 1.3977655537172164, "learning_rate": 0.003, "loss": 4.0593, "step": 15263 }, { "epoch": 0.15264, "grad_norm": 1.007233880569925, "learning_rate": 0.003, "loss": 4.0463, "step": 15264 }, { "epoch": 0.15265, "grad_norm": 1.223511465635572, "learning_rate": 0.003, "loss": 4.0705, "step": 15265 }, { "epoch": 0.15266, "grad_norm": 1.1709854606116625, "learning_rate": 0.003, "loss": 4.0401, "step": 15266 }, { "epoch": 0.15267, "grad_norm": 1.060570100692955, "learning_rate": 0.003, "loss": 4.0555, "step": 15267 }, { "epoch": 0.15268, "grad_norm": 1.170916883693262, "learning_rate": 0.003, "loss": 4.0638, "step": 15268 }, { "epoch": 0.15269, "grad_norm": 1.13507419180462, "learning_rate": 0.003, "loss": 4.0519, "step": 15269 }, { "epoch": 0.1527, "grad_norm": 1.3497710321548095, "learning_rate": 0.003, "loss": 4.0577, "step": 15270 }, { "epoch": 0.15271, "grad_norm": 1.0625889576185044, "learning_rate": 0.003, "loss": 4.0742, "step": 15271 }, { "epoch": 0.15272, "grad_norm": 1.2984489575320157, "learning_rate": 0.003, "loss": 4.0456, "step": 15272 }, { "epoch": 0.15273, "grad_norm": 0.9872476347269957, "learning_rate": 0.003, "loss": 4.029, "step": 15273 }, { "epoch": 0.15274, "grad_norm": 1.2339173710754143, "learning_rate": 0.003, "loss": 4.0462, "step": 15274 }, { "epoch": 0.15275, "grad_norm": 1.4088951536106782, "learning_rate": 0.003, "loss": 4.0644, "step": 15275 }, { "epoch": 0.15276, "grad_norm": 1.0339165586886583, "learning_rate": 0.003, "loss": 4.0772, "step": 15276 }, { "epoch": 0.15277, "grad_norm": 1.276415098913713, "learning_rate": 0.003, "loss": 4.0318, "step": 15277 }, { "epoch": 0.15278, "grad_norm": 1.2089381455177826, "learning_rate": 0.003, "loss": 4.0581, "step": 15278 }, { "epoch": 0.15279, "grad_norm": 1.2323476700147227, "learning_rate": 0.003, "loss": 4.0637, "step": 15279 }, { "epoch": 0.1528, "grad_norm": 1.0261540015381687, "learning_rate": 0.003, "loss": 4.0626, "step": 15280 }, { "epoch": 0.15281, "grad_norm": 1.4081233348548858, "learning_rate": 0.003, "loss": 4.0443, "step": 15281 }, { "epoch": 0.15282, "grad_norm": 0.9943608644134634, "learning_rate": 0.003, "loss": 4.0685, "step": 15282 }, { "epoch": 0.15283, "grad_norm": 1.4353065923668198, "learning_rate": 0.003, "loss": 4.0788, "step": 15283 }, { "epoch": 0.15284, "grad_norm": 1.1158782446879179, "learning_rate": 0.003, "loss": 4.0723, "step": 15284 }, { "epoch": 0.15285, "grad_norm": 1.378278155658196, "learning_rate": 0.003, "loss": 4.0798, "step": 15285 }, { "epoch": 0.15286, "grad_norm": 1.0485484157388185, "learning_rate": 0.003, "loss": 4.0638, "step": 15286 }, { "epoch": 0.15287, "grad_norm": 1.237331451668235, "learning_rate": 0.003, "loss": 4.0476, "step": 15287 }, { "epoch": 0.15288, "grad_norm": 1.1032372363197314, "learning_rate": 0.003, "loss": 4.0599, "step": 15288 }, { "epoch": 0.15289, "grad_norm": 1.2870811875647183, "learning_rate": 0.003, "loss": 4.0457, "step": 15289 }, { "epoch": 0.1529, "grad_norm": 1.232656198062413, "learning_rate": 0.003, "loss": 4.0675, "step": 15290 }, { "epoch": 0.15291, "grad_norm": 1.0414566423936231, "learning_rate": 0.003, "loss": 4.0816, "step": 15291 }, { "epoch": 0.15292, "grad_norm": 1.6205413571895269, "learning_rate": 0.003, "loss": 4.0964, "step": 15292 }, { "epoch": 0.15293, "grad_norm": 0.971170882527214, "learning_rate": 0.003, "loss": 4.0795, "step": 15293 }, { "epoch": 0.15294, "grad_norm": 1.348736972371927, "learning_rate": 0.003, "loss": 4.0869, "step": 15294 }, { "epoch": 0.15295, "grad_norm": 1.0345246527365257, "learning_rate": 0.003, "loss": 4.0756, "step": 15295 }, { "epoch": 0.15296, "grad_norm": 1.3846095991384118, "learning_rate": 0.003, "loss": 4.0965, "step": 15296 }, { "epoch": 0.15297, "grad_norm": 1.1386240471237765, "learning_rate": 0.003, "loss": 4.0637, "step": 15297 }, { "epoch": 0.15298, "grad_norm": 1.1205088253245086, "learning_rate": 0.003, "loss": 4.0869, "step": 15298 }, { "epoch": 0.15299, "grad_norm": 1.1018736843383212, "learning_rate": 0.003, "loss": 4.0836, "step": 15299 }, { "epoch": 0.153, "grad_norm": 1.2649416067278314, "learning_rate": 0.003, "loss": 4.0944, "step": 15300 }, { "epoch": 0.15301, "grad_norm": 1.3461673681342328, "learning_rate": 0.003, "loss": 4.0509, "step": 15301 }, { "epoch": 0.15302, "grad_norm": 1.5236898373506294, "learning_rate": 0.003, "loss": 4.0411, "step": 15302 }, { "epoch": 0.15303, "grad_norm": 1.421307585707187, "learning_rate": 0.003, "loss": 4.096, "step": 15303 }, { "epoch": 0.15304, "grad_norm": 1.0770487276285168, "learning_rate": 0.003, "loss": 4.0931, "step": 15304 }, { "epoch": 0.15305, "grad_norm": 1.2803489328118753, "learning_rate": 0.003, "loss": 4.0539, "step": 15305 }, { "epoch": 0.15306, "grad_norm": 1.173623701273869, "learning_rate": 0.003, "loss": 4.0465, "step": 15306 }, { "epoch": 0.15307, "grad_norm": 1.3290158642222427, "learning_rate": 0.003, "loss": 4.0352, "step": 15307 }, { "epoch": 0.15308, "grad_norm": 1.2386634096017313, "learning_rate": 0.003, "loss": 4.0693, "step": 15308 }, { "epoch": 0.15309, "grad_norm": 0.9787141666673741, "learning_rate": 0.003, "loss": 4.0645, "step": 15309 }, { "epoch": 0.1531, "grad_norm": 1.274687433764411, "learning_rate": 0.003, "loss": 4.0802, "step": 15310 }, { "epoch": 0.15311, "grad_norm": 1.2427929343705486, "learning_rate": 0.003, "loss": 4.0662, "step": 15311 }, { "epoch": 0.15312, "grad_norm": 0.9266597260781467, "learning_rate": 0.003, "loss": 4.0546, "step": 15312 }, { "epoch": 0.15313, "grad_norm": 1.2328317591820357, "learning_rate": 0.003, "loss": 4.0809, "step": 15313 }, { "epoch": 0.15314, "grad_norm": 1.0652956670195421, "learning_rate": 0.003, "loss": 4.0676, "step": 15314 }, { "epoch": 0.15315, "grad_norm": 1.2815764068774655, "learning_rate": 0.003, "loss": 4.0816, "step": 15315 }, { "epoch": 0.15316, "grad_norm": 1.1158678154899941, "learning_rate": 0.003, "loss": 4.0849, "step": 15316 }, { "epoch": 0.15317, "grad_norm": 1.0487746743010973, "learning_rate": 0.003, "loss": 4.0857, "step": 15317 }, { "epoch": 0.15318, "grad_norm": 1.2063070441600514, "learning_rate": 0.003, "loss": 4.0618, "step": 15318 }, { "epoch": 0.15319, "grad_norm": 1.1941164133060218, "learning_rate": 0.003, "loss": 4.0402, "step": 15319 }, { "epoch": 0.1532, "grad_norm": 1.362717851724568, "learning_rate": 0.003, "loss": 4.0682, "step": 15320 }, { "epoch": 0.15321, "grad_norm": 1.1536782993394188, "learning_rate": 0.003, "loss": 4.0683, "step": 15321 }, { "epoch": 0.15322, "grad_norm": 1.3971050672436391, "learning_rate": 0.003, "loss": 4.0851, "step": 15322 }, { "epoch": 0.15323, "grad_norm": 1.0713944798806787, "learning_rate": 0.003, "loss": 4.0749, "step": 15323 }, { "epoch": 0.15324, "grad_norm": 1.4856335726244028, "learning_rate": 0.003, "loss": 4.0638, "step": 15324 }, { "epoch": 0.15325, "grad_norm": 0.9987382371883158, "learning_rate": 0.003, "loss": 4.0839, "step": 15325 }, { "epoch": 0.15326, "grad_norm": 1.1336635102779324, "learning_rate": 0.003, "loss": 4.0645, "step": 15326 }, { "epoch": 0.15327, "grad_norm": 1.1534136647775999, "learning_rate": 0.003, "loss": 4.0845, "step": 15327 }, { "epoch": 0.15328, "grad_norm": 1.122548485263682, "learning_rate": 0.003, "loss": 4.0316, "step": 15328 }, { "epoch": 0.15329, "grad_norm": 1.1152038630462762, "learning_rate": 0.003, "loss": 4.0549, "step": 15329 }, { "epoch": 0.1533, "grad_norm": 1.411285240833272, "learning_rate": 0.003, "loss": 4.0645, "step": 15330 }, { "epoch": 0.15331, "grad_norm": 1.1149263633444346, "learning_rate": 0.003, "loss": 4.0515, "step": 15331 }, { "epoch": 0.15332, "grad_norm": 1.1647721531569564, "learning_rate": 0.003, "loss": 4.0486, "step": 15332 }, { "epoch": 0.15333, "grad_norm": 1.1676832997788698, "learning_rate": 0.003, "loss": 4.0753, "step": 15333 }, { "epoch": 0.15334, "grad_norm": 1.2533323067325388, "learning_rate": 0.003, "loss": 4.0439, "step": 15334 }, { "epoch": 0.15335, "grad_norm": 1.225618474950692, "learning_rate": 0.003, "loss": 4.0901, "step": 15335 }, { "epoch": 0.15336, "grad_norm": 1.126694619597195, "learning_rate": 0.003, "loss": 4.0635, "step": 15336 }, { "epoch": 0.15337, "grad_norm": 1.3866292038106043, "learning_rate": 0.003, "loss": 4.0682, "step": 15337 }, { "epoch": 0.15338, "grad_norm": 1.0129310147081365, "learning_rate": 0.003, "loss": 4.0803, "step": 15338 }, { "epoch": 0.15339, "grad_norm": 1.3182981482150227, "learning_rate": 0.003, "loss": 4.0666, "step": 15339 }, { "epoch": 0.1534, "grad_norm": 1.0536027013917006, "learning_rate": 0.003, "loss": 4.0617, "step": 15340 }, { "epoch": 0.15341, "grad_norm": 1.3112499607823593, "learning_rate": 0.003, "loss": 4.0823, "step": 15341 }, { "epoch": 0.15342, "grad_norm": 1.1314261343534997, "learning_rate": 0.003, "loss": 4.0703, "step": 15342 }, { "epoch": 0.15343, "grad_norm": 1.4555014481375317, "learning_rate": 0.003, "loss": 4.0557, "step": 15343 }, { "epoch": 0.15344, "grad_norm": 1.0460042053898686, "learning_rate": 0.003, "loss": 4.0729, "step": 15344 }, { "epoch": 0.15345, "grad_norm": 1.418197133113981, "learning_rate": 0.003, "loss": 4.087, "step": 15345 }, { "epoch": 0.15346, "grad_norm": 0.8737250485311147, "learning_rate": 0.003, "loss": 4.0589, "step": 15346 }, { "epoch": 0.15347, "grad_norm": 1.1180711173208815, "learning_rate": 0.003, "loss": 4.0408, "step": 15347 }, { "epoch": 0.15348, "grad_norm": 1.1934974290749425, "learning_rate": 0.003, "loss": 4.0738, "step": 15348 }, { "epoch": 0.15349, "grad_norm": 1.2919597559085614, "learning_rate": 0.003, "loss": 4.0653, "step": 15349 }, { "epoch": 0.1535, "grad_norm": 1.2062159759058706, "learning_rate": 0.003, "loss": 4.0712, "step": 15350 }, { "epoch": 0.15351, "grad_norm": 1.1297498995697322, "learning_rate": 0.003, "loss": 4.022, "step": 15351 }, { "epoch": 0.15352, "grad_norm": 1.4763402501322123, "learning_rate": 0.003, "loss": 4.0589, "step": 15352 }, { "epoch": 0.15353, "grad_norm": 1.1012937369280016, "learning_rate": 0.003, "loss": 4.0512, "step": 15353 }, { "epoch": 0.15354, "grad_norm": 1.1566135011916077, "learning_rate": 0.003, "loss": 4.0638, "step": 15354 }, { "epoch": 0.15355, "grad_norm": 1.4332475868537249, "learning_rate": 0.003, "loss": 4.076, "step": 15355 }, { "epoch": 0.15356, "grad_norm": 1.120567900992729, "learning_rate": 0.003, "loss": 4.0474, "step": 15356 }, { "epoch": 0.15357, "grad_norm": 1.2898623769989723, "learning_rate": 0.003, "loss": 4.077, "step": 15357 }, { "epoch": 0.15358, "grad_norm": 1.1751466448499435, "learning_rate": 0.003, "loss": 4.0489, "step": 15358 }, { "epoch": 0.15359, "grad_norm": 1.1583971676769786, "learning_rate": 0.003, "loss": 4.0373, "step": 15359 }, { "epoch": 0.1536, "grad_norm": 1.0556389547544935, "learning_rate": 0.003, "loss": 4.0514, "step": 15360 }, { "epoch": 0.15361, "grad_norm": 1.5931953804066483, "learning_rate": 0.003, "loss": 4.0709, "step": 15361 }, { "epoch": 0.15362, "grad_norm": 1.0044840368714696, "learning_rate": 0.003, "loss": 4.0618, "step": 15362 }, { "epoch": 0.15363, "grad_norm": 1.4726867281373388, "learning_rate": 0.003, "loss": 4.1001, "step": 15363 }, { "epoch": 0.15364, "grad_norm": 1.0650785921619903, "learning_rate": 0.003, "loss": 4.0519, "step": 15364 }, { "epoch": 0.15365, "grad_norm": 1.0864983987638424, "learning_rate": 0.003, "loss": 4.0779, "step": 15365 }, { "epoch": 0.15366, "grad_norm": 1.299374370037176, "learning_rate": 0.003, "loss": 4.0934, "step": 15366 }, { "epoch": 0.15367, "grad_norm": 1.0634044848199793, "learning_rate": 0.003, "loss": 4.0377, "step": 15367 }, { "epoch": 0.15368, "grad_norm": 1.1976379564094535, "learning_rate": 0.003, "loss": 4.0557, "step": 15368 }, { "epoch": 0.15369, "grad_norm": 1.2144313440821537, "learning_rate": 0.003, "loss": 4.045, "step": 15369 }, { "epoch": 0.1537, "grad_norm": 1.4634358111483572, "learning_rate": 0.003, "loss": 4.0294, "step": 15370 }, { "epoch": 0.15371, "grad_norm": 0.9345829397981521, "learning_rate": 0.003, "loss": 4.0612, "step": 15371 }, { "epoch": 0.15372, "grad_norm": 1.2087216895515303, "learning_rate": 0.003, "loss": 4.0591, "step": 15372 }, { "epoch": 0.15373, "grad_norm": 1.2333298012088245, "learning_rate": 0.003, "loss": 4.0683, "step": 15373 }, { "epoch": 0.15374, "grad_norm": 1.3674821509295068, "learning_rate": 0.003, "loss": 4.0711, "step": 15374 }, { "epoch": 0.15375, "grad_norm": 1.099436386068766, "learning_rate": 0.003, "loss": 4.055, "step": 15375 }, { "epoch": 0.15376, "grad_norm": 1.3148981477185882, "learning_rate": 0.003, "loss": 4.0772, "step": 15376 }, { "epoch": 0.15377, "grad_norm": 1.0775682871515682, "learning_rate": 0.003, "loss": 4.0528, "step": 15377 }, { "epoch": 0.15378, "grad_norm": 1.2954200985903883, "learning_rate": 0.003, "loss": 4.06, "step": 15378 }, { "epoch": 0.15379, "grad_norm": 1.304413999793567, "learning_rate": 0.003, "loss": 4.0908, "step": 15379 }, { "epoch": 0.1538, "grad_norm": 1.0702911542230267, "learning_rate": 0.003, "loss": 4.0544, "step": 15380 }, { "epoch": 0.15381, "grad_norm": 1.2258801818381888, "learning_rate": 0.003, "loss": 4.0536, "step": 15381 }, { "epoch": 0.15382, "grad_norm": 1.4176031465193795, "learning_rate": 0.003, "loss": 4.0761, "step": 15382 }, { "epoch": 0.15383, "grad_norm": 1.098414082125524, "learning_rate": 0.003, "loss": 4.0836, "step": 15383 }, { "epoch": 0.15384, "grad_norm": 1.3295703898676385, "learning_rate": 0.003, "loss": 4.0617, "step": 15384 }, { "epoch": 0.15385, "grad_norm": 0.981606428503668, "learning_rate": 0.003, "loss": 4.0749, "step": 15385 }, { "epoch": 0.15386, "grad_norm": 1.4435323663276558, "learning_rate": 0.003, "loss": 4.0392, "step": 15386 }, { "epoch": 0.15387, "grad_norm": 0.9267463710947388, "learning_rate": 0.003, "loss": 4.036, "step": 15387 }, { "epoch": 0.15388, "grad_norm": 1.3629792496605608, "learning_rate": 0.003, "loss": 4.0816, "step": 15388 }, { "epoch": 0.15389, "grad_norm": 1.1319156951659917, "learning_rate": 0.003, "loss": 4.0447, "step": 15389 }, { "epoch": 0.1539, "grad_norm": 1.234161943185369, "learning_rate": 0.003, "loss": 4.0916, "step": 15390 }, { "epoch": 0.15391, "grad_norm": 1.3987267358306366, "learning_rate": 0.003, "loss": 4.0686, "step": 15391 }, { "epoch": 0.15392, "grad_norm": 1.2372901652941222, "learning_rate": 0.003, "loss": 4.0613, "step": 15392 }, { "epoch": 0.15393, "grad_norm": 1.1868185575964618, "learning_rate": 0.003, "loss": 4.0731, "step": 15393 }, { "epoch": 0.15394, "grad_norm": 1.2551610808976825, "learning_rate": 0.003, "loss": 4.0401, "step": 15394 }, { "epoch": 0.15395, "grad_norm": 1.4110003770316828, "learning_rate": 0.003, "loss": 4.0801, "step": 15395 }, { "epoch": 0.15396, "grad_norm": 1.0311286977956948, "learning_rate": 0.003, "loss": 4.0312, "step": 15396 }, { "epoch": 0.15397, "grad_norm": 1.3176922148726504, "learning_rate": 0.003, "loss": 4.0954, "step": 15397 }, { "epoch": 0.15398, "grad_norm": 1.119210963152209, "learning_rate": 0.003, "loss": 4.0365, "step": 15398 }, { "epoch": 0.15399, "grad_norm": 1.1585313167937918, "learning_rate": 0.003, "loss": 4.0272, "step": 15399 }, { "epoch": 0.154, "grad_norm": 1.1718901018265493, "learning_rate": 0.003, "loss": 4.0427, "step": 15400 }, { "epoch": 0.15401, "grad_norm": 1.2295322444343344, "learning_rate": 0.003, "loss": 4.058, "step": 15401 }, { "epoch": 0.15402, "grad_norm": 1.2411543091235737, "learning_rate": 0.003, "loss": 4.0537, "step": 15402 }, { "epoch": 0.15403, "grad_norm": 1.0399534030778381, "learning_rate": 0.003, "loss": 4.0361, "step": 15403 }, { "epoch": 0.15404, "grad_norm": 1.345481710843217, "learning_rate": 0.003, "loss": 4.0433, "step": 15404 }, { "epoch": 0.15405, "grad_norm": 1.0782567322337016, "learning_rate": 0.003, "loss": 4.0969, "step": 15405 }, { "epoch": 0.15406, "grad_norm": 1.3770651287882087, "learning_rate": 0.003, "loss": 4.0548, "step": 15406 }, { "epoch": 0.15407, "grad_norm": 1.1791156159933935, "learning_rate": 0.003, "loss": 4.0311, "step": 15407 }, { "epoch": 0.15408, "grad_norm": 1.2591927464401316, "learning_rate": 0.003, "loss": 4.0782, "step": 15408 }, { "epoch": 0.15409, "grad_norm": 1.1787754971139592, "learning_rate": 0.003, "loss": 4.088, "step": 15409 }, { "epoch": 0.1541, "grad_norm": 1.2428022611947518, "learning_rate": 0.003, "loss": 4.0771, "step": 15410 }, { "epoch": 0.15411, "grad_norm": 1.1173057583099402, "learning_rate": 0.003, "loss": 4.063, "step": 15411 }, { "epoch": 0.15412, "grad_norm": 1.252529165084718, "learning_rate": 0.003, "loss": 4.058, "step": 15412 }, { "epoch": 0.15413, "grad_norm": 1.250156560920163, "learning_rate": 0.003, "loss": 4.0971, "step": 15413 }, { "epoch": 0.15414, "grad_norm": 1.2241752494741582, "learning_rate": 0.003, "loss": 4.0612, "step": 15414 }, { "epoch": 0.15415, "grad_norm": 1.2707570259367225, "learning_rate": 0.003, "loss": 4.0752, "step": 15415 }, { "epoch": 0.15416, "grad_norm": 1.426633007592224, "learning_rate": 0.003, "loss": 4.0819, "step": 15416 }, { "epoch": 0.15417, "grad_norm": 1.5907670541264263, "learning_rate": 0.003, "loss": 4.0614, "step": 15417 }, { "epoch": 0.15418, "grad_norm": 0.9091544803403122, "learning_rate": 0.003, "loss": 4.0284, "step": 15418 }, { "epoch": 0.15419, "grad_norm": 1.047751881825804, "learning_rate": 0.003, "loss": 4.0489, "step": 15419 }, { "epoch": 0.1542, "grad_norm": 1.4170082204482124, "learning_rate": 0.003, "loss": 4.0591, "step": 15420 }, { "epoch": 0.15421, "grad_norm": 1.5733130067127954, "learning_rate": 0.003, "loss": 4.072, "step": 15421 }, { "epoch": 0.15422, "grad_norm": 1.1820234369725136, "learning_rate": 0.003, "loss": 4.049, "step": 15422 }, { "epoch": 0.15423, "grad_norm": 1.1659647288488164, "learning_rate": 0.003, "loss": 4.0216, "step": 15423 }, { "epoch": 0.15424, "grad_norm": 1.0479943433146457, "learning_rate": 0.003, "loss": 4.06, "step": 15424 }, { "epoch": 0.15425, "grad_norm": 1.4604447382383878, "learning_rate": 0.003, "loss": 4.0538, "step": 15425 }, { "epoch": 0.15426, "grad_norm": 1.0268914205354007, "learning_rate": 0.003, "loss": 4.0367, "step": 15426 }, { "epoch": 0.15427, "grad_norm": 1.1655023774090285, "learning_rate": 0.003, "loss": 4.0595, "step": 15427 }, { "epoch": 0.15428, "grad_norm": 1.2744321487748451, "learning_rate": 0.003, "loss": 4.0853, "step": 15428 }, { "epoch": 0.15429, "grad_norm": 1.044821370728945, "learning_rate": 0.003, "loss": 4.0468, "step": 15429 }, { "epoch": 0.1543, "grad_norm": 1.336660814677801, "learning_rate": 0.003, "loss": 4.0668, "step": 15430 }, { "epoch": 0.15431, "grad_norm": 1.0115445320314047, "learning_rate": 0.003, "loss": 4.0866, "step": 15431 }, { "epoch": 0.15432, "grad_norm": 1.3471221521419243, "learning_rate": 0.003, "loss": 4.0745, "step": 15432 }, { "epoch": 0.15433, "grad_norm": 1.218306878259319, "learning_rate": 0.003, "loss": 4.0691, "step": 15433 }, { "epoch": 0.15434, "grad_norm": 1.2667733712071507, "learning_rate": 0.003, "loss": 4.0456, "step": 15434 }, { "epoch": 0.15435, "grad_norm": 1.515485139828299, "learning_rate": 0.003, "loss": 4.0583, "step": 15435 }, { "epoch": 0.15436, "grad_norm": 1.0256710056302927, "learning_rate": 0.003, "loss": 4.0709, "step": 15436 }, { "epoch": 0.15437, "grad_norm": 1.3457950188332866, "learning_rate": 0.003, "loss": 4.0597, "step": 15437 }, { "epoch": 0.15438, "grad_norm": 1.0676201322248198, "learning_rate": 0.003, "loss": 4.0916, "step": 15438 }, { "epoch": 0.15439, "grad_norm": 1.2795541407972861, "learning_rate": 0.003, "loss": 4.0541, "step": 15439 }, { "epoch": 0.1544, "grad_norm": 1.0362570095334005, "learning_rate": 0.003, "loss": 4.0371, "step": 15440 }, { "epoch": 0.15441, "grad_norm": 1.3352973418819123, "learning_rate": 0.003, "loss": 4.0654, "step": 15441 }, { "epoch": 0.15442, "grad_norm": 1.5627360313033476, "learning_rate": 0.003, "loss": 4.0525, "step": 15442 }, { "epoch": 0.15443, "grad_norm": 0.8954312056496571, "learning_rate": 0.003, "loss": 4.0334, "step": 15443 }, { "epoch": 0.15444, "grad_norm": 0.9003385139611089, "learning_rate": 0.003, "loss": 4.0501, "step": 15444 }, { "epoch": 0.15445, "grad_norm": 1.1641098471539841, "learning_rate": 0.003, "loss": 4.0229, "step": 15445 }, { "epoch": 0.15446, "grad_norm": 1.038599211201421, "learning_rate": 0.003, "loss": 4.0464, "step": 15446 }, { "epoch": 0.15447, "grad_norm": 1.3357295990512643, "learning_rate": 0.003, "loss": 4.0731, "step": 15447 }, { "epoch": 0.15448, "grad_norm": 1.1546627490334793, "learning_rate": 0.003, "loss": 4.0845, "step": 15448 }, { "epoch": 0.15449, "grad_norm": 1.1299525347651804, "learning_rate": 0.003, "loss": 4.0475, "step": 15449 }, { "epoch": 0.1545, "grad_norm": 1.407049291550648, "learning_rate": 0.003, "loss": 4.0781, "step": 15450 }, { "epoch": 0.15451, "grad_norm": 1.0122402823659429, "learning_rate": 0.003, "loss": 4.0621, "step": 15451 }, { "epoch": 0.15452, "grad_norm": 1.3964304058717305, "learning_rate": 0.003, "loss": 4.0408, "step": 15452 }, { "epoch": 0.15453, "grad_norm": 1.1379627627656579, "learning_rate": 0.003, "loss": 4.0531, "step": 15453 }, { "epoch": 0.15454, "grad_norm": 1.1716005696161556, "learning_rate": 0.003, "loss": 4.0461, "step": 15454 }, { "epoch": 0.15455, "grad_norm": 1.0512955853342485, "learning_rate": 0.003, "loss": 4.087, "step": 15455 }, { "epoch": 0.15456, "grad_norm": 1.477149573778637, "learning_rate": 0.003, "loss": 4.0422, "step": 15456 }, { "epoch": 0.15457, "grad_norm": 1.0542623589033444, "learning_rate": 0.003, "loss": 4.0701, "step": 15457 }, { "epoch": 0.15458, "grad_norm": 1.4777046926994861, "learning_rate": 0.003, "loss": 4.0724, "step": 15458 }, { "epoch": 0.15459, "grad_norm": 1.0398416915025779, "learning_rate": 0.003, "loss": 4.0606, "step": 15459 }, { "epoch": 0.1546, "grad_norm": 1.3202079609830617, "learning_rate": 0.003, "loss": 4.0485, "step": 15460 }, { "epoch": 0.15461, "grad_norm": 1.172625188072539, "learning_rate": 0.003, "loss": 4.0398, "step": 15461 }, { "epoch": 0.15462, "grad_norm": 1.3455025181894154, "learning_rate": 0.003, "loss": 4.0489, "step": 15462 }, { "epoch": 0.15463, "grad_norm": 1.1862746747288408, "learning_rate": 0.003, "loss": 4.0332, "step": 15463 }, { "epoch": 0.15464, "grad_norm": 1.3410554902084484, "learning_rate": 0.003, "loss": 4.0578, "step": 15464 }, { "epoch": 0.15465, "grad_norm": 0.7670962679225621, "learning_rate": 0.003, "loss": 4.0409, "step": 15465 }, { "epoch": 0.15466, "grad_norm": 1.2231600802959381, "learning_rate": 0.003, "loss": 4.0674, "step": 15466 }, { "epoch": 0.15467, "grad_norm": 1.3683096303657931, "learning_rate": 0.003, "loss": 4.0678, "step": 15467 }, { "epoch": 0.15468, "grad_norm": 1.0633665339305103, "learning_rate": 0.003, "loss": 4.0408, "step": 15468 }, { "epoch": 0.15469, "grad_norm": 1.1569441747859937, "learning_rate": 0.003, "loss": 4.0788, "step": 15469 }, { "epoch": 0.1547, "grad_norm": 1.2900173704913636, "learning_rate": 0.003, "loss": 4.0713, "step": 15470 }, { "epoch": 0.15471, "grad_norm": 0.8564636912847731, "learning_rate": 0.003, "loss": 4.0578, "step": 15471 }, { "epoch": 0.15472, "grad_norm": 0.958179425574608, "learning_rate": 0.003, "loss": 4.0449, "step": 15472 }, { "epoch": 0.15473, "grad_norm": 1.2239514432578584, "learning_rate": 0.003, "loss": 4.086, "step": 15473 }, { "epoch": 0.15474, "grad_norm": 1.0718699054270044, "learning_rate": 0.003, "loss": 4.0506, "step": 15474 }, { "epoch": 0.15475, "grad_norm": 1.1614611181808692, "learning_rate": 0.003, "loss": 4.0691, "step": 15475 }, { "epoch": 0.15476, "grad_norm": 1.3226572290841543, "learning_rate": 0.003, "loss": 4.0887, "step": 15476 }, { "epoch": 0.15477, "grad_norm": 1.2292997948946798, "learning_rate": 0.003, "loss": 4.0695, "step": 15477 }, { "epoch": 0.15478, "grad_norm": 1.1967741002548855, "learning_rate": 0.003, "loss": 4.0587, "step": 15478 }, { "epoch": 0.15479, "grad_norm": 1.3506286726490333, "learning_rate": 0.003, "loss": 4.0533, "step": 15479 }, { "epoch": 0.1548, "grad_norm": 1.2115615904440533, "learning_rate": 0.003, "loss": 4.0693, "step": 15480 }, { "epoch": 0.15481, "grad_norm": 1.2256804181231584, "learning_rate": 0.003, "loss": 4.0676, "step": 15481 }, { "epoch": 0.15482, "grad_norm": 1.19838963798865, "learning_rate": 0.003, "loss": 4.0507, "step": 15482 }, { "epoch": 0.15483, "grad_norm": 1.140780741362575, "learning_rate": 0.003, "loss": 4.0732, "step": 15483 }, { "epoch": 0.15484, "grad_norm": 1.3496605791591862, "learning_rate": 0.003, "loss": 4.0984, "step": 15484 }, { "epoch": 0.15485, "grad_norm": 1.4588788825008616, "learning_rate": 0.003, "loss": 4.0903, "step": 15485 }, { "epoch": 0.15486, "grad_norm": 0.9756831933110768, "learning_rate": 0.003, "loss": 4.0747, "step": 15486 }, { "epoch": 0.15487, "grad_norm": 1.1689926631074683, "learning_rate": 0.003, "loss": 4.0579, "step": 15487 }, { "epoch": 0.15488, "grad_norm": 1.2231147673104008, "learning_rate": 0.003, "loss": 4.0661, "step": 15488 }, { "epoch": 0.15489, "grad_norm": 1.2436681537852814, "learning_rate": 0.003, "loss": 4.0543, "step": 15489 }, { "epoch": 0.1549, "grad_norm": 1.1856690267859664, "learning_rate": 0.003, "loss": 4.0658, "step": 15490 }, { "epoch": 0.15491, "grad_norm": 1.1528720868109508, "learning_rate": 0.003, "loss": 4.0446, "step": 15491 }, { "epoch": 0.15492, "grad_norm": 1.2449752035398804, "learning_rate": 0.003, "loss": 4.0612, "step": 15492 }, { "epoch": 0.15493, "grad_norm": 1.2889529574754994, "learning_rate": 0.003, "loss": 4.058, "step": 15493 }, { "epoch": 0.15494, "grad_norm": 1.1847213418178684, "learning_rate": 0.003, "loss": 4.0508, "step": 15494 }, { "epoch": 0.15495, "grad_norm": 1.1044995912636943, "learning_rate": 0.003, "loss": 4.0821, "step": 15495 }, { "epoch": 0.15496, "grad_norm": 1.385225884920843, "learning_rate": 0.003, "loss": 4.0646, "step": 15496 }, { "epoch": 0.15497, "grad_norm": 1.1573457439443438, "learning_rate": 0.003, "loss": 4.0557, "step": 15497 }, { "epoch": 0.15498, "grad_norm": 1.0805457747390674, "learning_rate": 0.003, "loss": 4.0092, "step": 15498 }, { "epoch": 0.15499, "grad_norm": 1.245268042241472, "learning_rate": 0.003, "loss": 4.0486, "step": 15499 }, { "epoch": 0.155, "grad_norm": 1.0798871257694764, "learning_rate": 0.003, "loss": 4.0542, "step": 15500 }, { "epoch": 0.15501, "grad_norm": 1.3062139606925551, "learning_rate": 0.003, "loss": 4.0485, "step": 15501 }, { "epoch": 0.15502, "grad_norm": 1.0361323349085323, "learning_rate": 0.003, "loss": 4.0584, "step": 15502 }, { "epoch": 0.15503, "grad_norm": 1.3788101018795207, "learning_rate": 0.003, "loss": 4.0541, "step": 15503 }, { "epoch": 0.15504, "grad_norm": 1.1210327336799735, "learning_rate": 0.003, "loss": 4.077, "step": 15504 }, { "epoch": 0.15505, "grad_norm": 1.1745939313235108, "learning_rate": 0.003, "loss": 4.0512, "step": 15505 }, { "epoch": 0.15506, "grad_norm": 1.1844245158495095, "learning_rate": 0.003, "loss": 4.0482, "step": 15506 }, { "epoch": 0.15507, "grad_norm": 1.4079253310507698, "learning_rate": 0.003, "loss": 4.0446, "step": 15507 }, { "epoch": 0.15508, "grad_norm": 1.0722617625019373, "learning_rate": 0.003, "loss": 4.0539, "step": 15508 }, { "epoch": 0.15509, "grad_norm": 1.4814012345526648, "learning_rate": 0.003, "loss": 4.0423, "step": 15509 }, { "epoch": 0.1551, "grad_norm": 0.9902479457488543, "learning_rate": 0.003, "loss": 4.0794, "step": 15510 }, { "epoch": 0.15511, "grad_norm": 1.524427659657888, "learning_rate": 0.003, "loss": 4.0998, "step": 15511 }, { "epoch": 0.15512, "grad_norm": 1.2749984682610411, "learning_rate": 0.003, "loss": 4.0967, "step": 15512 }, { "epoch": 0.15513, "grad_norm": 1.1272472479632105, "learning_rate": 0.003, "loss": 4.0514, "step": 15513 }, { "epoch": 0.15514, "grad_norm": 1.241946440642086, "learning_rate": 0.003, "loss": 4.0944, "step": 15514 }, { "epoch": 0.15515, "grad_norm": 1.1902532115752005, "learning_rate": 0.003, "loss": 4.0585, "step": 15515 }, { "epoch": 0.15516, "grad_norm": 1.0176104417246836, "learning_rate": 0.003, "loss": 4.0712, "step": 15516 }, { "epoch": 0.15517, "grad_norm": 1.1905163593066492, "learning_rate": 0.003, "loss": 4.0626, "step": 15517 }, { "epoch": 0.15518, "grad_norm": 1.2953977991788461, "learning_rate": 0.003, "loss": 4.066, "step": 15518 }, { "epoch": 0.15519, "grad_norm": 1.3096714320460605, "learning_rate": 0.003, "loss": 4.0478, "step": 15519 }, { "epoch": 0.1552, "grad_norm": 0.9938131193534318, "learning_rate": 0.003, "loss": 4.0448, "step": 15520 }, { "epoch": 0.15521, "grad_norm": 1.2698929503832623, "learning_rate": 0.003, "loss": 4.0903, "step": 15521 }, { "epoch": 0.15522, "grad_norm": 1.2406608158792833, "learning_rate": 0.003, "loss": 4.087, "step": 15522 }, { "epoch": 0.15523, "grad_norm": 1.3692338528521906, "learning_rate": 0.003, "loss": 4.0816, "step": 15523 }, { "epoch": 0.15524, "grad_norm": 0.982338615316267, "learning_rate": 0.003, "loss": 4.0723, "step": 15524 }, { "epoch": 0.15525, "grad_norm": 1.1851100242763046, "learning_rate": 0.003, "loss": 4.0628, "step": 15525 }, { "epoch": 0.15526, "grad_norm": 1.5136744226877126, "learning_rate": 0.003, "loss": 4.0677, "step": 15526 }, { "epoch": 0.15527, "grad_norm": 1.128916057984434, "learning_rate": 0.003, "loss": 4.0519, "step": 15527 }, { "epoch": 0.15528, "grad_norm": 1.2309658190479194, "learning_rate": 0.003, "loss": 4.04, "step": 15528 }, { "epoch": 0.15529, "grad_norm": 1.1849305960657193, "learning_rate": 0.003, "loss": 4.0551, "step": 15529 }, { "epoch": 0.1553, "grad_norm": 1.2412067053524356, "learning_rate": 0.003, "loss": 4.0912, "step": 15530 }, { "epoch": 0.15531, "grad_norm": 0.9597583454661515, "learning_rate": 0.003, "loss": 4.0503, "step": 15531 }, { "epoch": 0.15532, "grad_norm": 1.3845715607138387, "learning_rate": 0.003, "loss": 4.0972, "step": 15532 }, { "epoch": 0.15533, "grad_norm": 1.189346116306467, "learning_rate": 0.003, "loss": 4.079, "step": 15533 }, { "epoch": 0.15534, "grad_norm": 1.2230592177407105, "learning_rate": 0.003, "loss": 4.0861, "step": 15534 }, { "epoch": 0.15535, "grad_norm": 1.1659354130678887, "learning_rate": 0.003, "loss": 4.0364, "step": 15535 }, { "epoch": 0.15536, "grad_norm": 1.1925263639024195, "learning_rate": 0.003, "loss": 4.0543, "step": 15536 }, { "epoch": 0.15537, "grad_norm": 1.0822641134617847, "learning_rate": 0.003, "loss": 4.0801, "step": 15537 }, { "epoch": 0.15538, "grad_norm": 1.161595204822747, "learning_rate": 0.003, "loss": 4.058, "step": 15538 }, { "epoch": 0.15539, "grad_norm": 1.3862142005882838, "learning_rate": 0.003, "loss": 4.0986, "step": 15539 }, { "epoch": 0.1554, "grad_norm": 1.229526343493269, "learning_rate": 0.003, "loss": 4.0985, "step": 15540 }, { "epoch": 0.15541, "grad_norm": 1.3488023480037135, "learning_rate": 0.003, "loss": 4.0898, "step": 15541 }, { "epoch": 0.15542, "grad_norm": 1.1274545055092502, "learning_rate": 0.003, "loss": 4.0396, "step": 15542 }, { "epoch": 0.15543, "grad_norm": 1.175224591693728, "learning_rate": 0.003, "loss": 4.0361, "step": 15543 }, { "epoch": 0.15544, "grad_norm": 1.0955511795444062, "learning_rate": 0.003, "loss": 4.1061, "step": 15544 }, { "epoch": 0.15545, "grad_norm": 1.2112066874099896, "learning_rate": 0.003, "loss": 4.0363, "step": 15545 }, { "epoch": 0.15546, "grad_norm": 1.1832663010198956, "learning_rate": 0.003, "loss": 4.0813, "step": 15546 }, { "epoch": 0.15547, "grad_norm": 0.9592680884857658, "learning_rate": 0.003, "loss": 4.0574, "step": 15547 }, { "epoch": 0.15548, "grad_norm": 1.2826548106039155, "learning_rate": 0.003, "loss": 4.0721, "step": 15548 }, { "epoch": 0.15549, "grad_norm": 1.180632470945826, "learning_rate": 0.003, "loss": 4.077, "step": 15549 }, { "epoch": 0.1555, "grad_norm": 1.1456310983281182, "learning_rate": 0.003, "loss": 4.0659, "step": 15550 }, { "epoch": 0.15551, "grad_norm": 1.2055256823594842, "learning_rate": 0.003, "loss": 4.0763, "step": 15551 }, { "epoch": 0.15552, "grad_norm": 1.199809677412245, "learning_rate": 0.003, "loss": 4.0452, "step": 15552 }, { "epoch": 0.15553, "grad_norm": 1.0730626887780514, "learning_rate": 0.003, "loss": 4.021, "step": 15553 }, { "epoch": 0.15554, "grad_norm": 1.1391559272958152, "learning_rate": 0.003, "loss": 4.0473, "step": 15554 }, { "epoch": 0.15555, "grad_norm": 1.1054273997346828, "learning_rate": 0.003, "loss": 4.0232, "step": 15555 }, { "epoch": 0.15556, "grad_norm": 1.3381741488249064, "learning_rate": 0.003, "loss": 4.0419, "step": 15556 }, { "epoch": 0.15557, "grad_norm": 1.0129453484587203, "learning_rate": 0.003, "loss": 4.0623, "step": 15557 }, { "epoch": 0.15558, "grad_norm": 1.3550247336343197, "learning_rate": 0.003, "loss": 4.062, "step": 15558 }, { "epoch": 0.15559, "grad_norm": 1.0800661526811102, "learning_rate": 0.003, "loss": 4.0495, "step": 15559 }, { "epoch": 0.1556, "grad_norm": 1.2794205298372396, "learning_rate": 0.003, "loss": 4.0763, "step": 15560 }, { "epoch": 0.15561, "grad_norm": 1.2484059967266228, "learning_rate": 0.003, "loss": 4.0488, "step": 15561 }, { "epoch": 0.15562, "grad_norm": 1.2579795851592477, "learning_rate": 0.003, "loss": 4.0599, "step": 15562 }, { "epoch": 0.15563, "grad_norm": 1.3985523212529833, "learning_rate": 0.003, "loss": 4.0848, "step": 15563 }, { "epoch": 0.15564, "grad_norm": 1.299900491125251, "learning_rate": 0.003, "loss": 4.0657, "step": 15564 }, { "epoch": 0.15565, "grad_norm": 1.22289263232539, "learning_rate": 0.003, "loss": 4.0504, "step": 15565 }, { "epoch": 0.15566, "grad_norm": 1.1682975186023075, "learning_rate": 0.003, "loss": 4.0505, "step": 15566 }, { "epoch": 0.15567, "grad_norm": 1.2231282085928956, "learning_rate": 0.003, "loss": 4.055, "step": 15567 }, { "epoch": 0.15568, "grad_norm": 1.2840712432383061, "learning_rate": 0.003, "loss": 4.0577, "step": 15568 }, { "epoch": 0.15569, "grad_norm": 1.2048914943418152, "learning_rate": 0.003, "loss": 4.0386, "step": 15569 }, { "epoch": 0.1557, "grad_norm": 1.1334564677078711, "learning_rate": 0.003, "loss": 4.0429, "step": 15570 }, { "epoch": 0.15571, "grad_norm": 1.2492044466984566, "learning_rate": 0.003, "loss": 4.0468, "step": 15571 }, { "epoch": 0.15572, "grad_norm": 1.2787685966793079, "learning_rate": 0.003, "loss": 4.0638, "step": 15572 }, { "epoch": 0.15573, "grad_norm": 1.2109642972554564, "learning_rate": 0.003, "loss": 4.0731, "step": 15573 }, { "epoch": 0.15574, "grad_norm": 1.2469054441137288, "learning_rate": 0.003, "loss": 4.0772, "step": 15574 }, { "epoch": 0.15575, "grad_norm": 1.3721895705961396, "learning_rate": 0.003, "loss": 4.0509, "step": 15575 }, { "epoch": 0.15576, "grad_norm": 1.183178923658135, "learning_rate": 0.003, "loss": 4.0654, "step": 15576 }, { "epoch": 0.15577, "grad_norm": 1.2709919241182668, "learning_rate": 0.003, "loss": 4.0688, "step": 15577 }, { "epoch": 0.15578, "grad_norm": 1.1787930793479175, "learning_rate": 0.003, "loss": 4.0646, "step": 15578 }, { "epoch": 0.15579, "grad_norm": 1.3582338253437232, "learning_rate": 0.003, "loss": 4.0514, "step": 15579 }, { "epoch": 0.1558, "grad_norm": 1.319750457522074, "learning_rate": 0.003, "loss": 4.0863, "step": 15580 }, { "epoch": 0.15581, "grad_norm": 1.1439198612979784, "learning_rate": 0.003, "loss": 4.0476, "step": 15581 }, { "epoch": 0.15582, "grad_norm": 1.207775361596036, "learning_rate": 0.003, "loss": 4.0688, "step": 15582 }, { "epoch": 0.15583, "grad_norm": 1.2628986971845215, "learning_rate": 0.003, "loss": 4.05, "step": 15583 }, { "epoch": 0.15584, "grad_norm": 1.208959099329531, "learning_rate": 0.003, "loss": 4.0341, "step": 15584 }, { "epoch": 0.15585, "grad_norm": 1.217579157488508, "learning_rate": 0.003, "loss": 4.0871, "step": 15585 }, { "epoch": 0.15586, "grad_norm": 1.1710200109179474, "learning_rate": 0.003, "loss": 4.0176, "step": 15586 }, { "epoch": 0.15587, "grad_norm": 1.1298927902719316, "learning_rate": 0.003, "loss": 4.0337, "step": 15587 }, { "epoch": 0.15588, "grad_norm": 1.351308668801087, "learning_rate": 0.003, "loss": 4.0492, "step": 15588 }, { "epoch": 0.15589, "grad_norm": 1.1533687643932535, "learning_rate": 0.003, "loss": 4.0253, "step": 15589 }, { "epoch": 0.1559, "grad_norm": 1.3577624193107098, "learning_rate": 0.003, "loss": 4.0484, "step": 15590 }, { "epoch": 0.15591, "grad_norm": 1.0378217988790521, "learning_rate": 0.003, "loss": 4.0603, "step": 15591 }, { "epoch": 0.15592, "grad_norm": 1.3241158196415868, "learning_rate": 0.003, "loss": 4.0819, "step": 15592 }, { "epoch": 0.15593, "grad_norm": 0.9357014159701508, "learning_rate": 0.003, "loss": 4.019, "step": 15593 }, { "epoch": 0.15594, "grad_norm": 1.3290171088512335, "learning_rate": 0.003, "loss": 4.046, "step": 15594 }, { "epoch": 0.15595, "grad_norm": 1.1905298663725166, "learning_rate": 0.003, "loss": 4.034, "step": 15595 }, { "epoch": 0.15596, "grad_norm": 1.1842922519903687, "learning_rate": 0.003, "loss": 4.0342, "step": 15596 }, { "epoch": 0.15597, "grad_norm": 1.393171668065257, "learning_rate": 0.003, "loss": 4.0751, "step": 15597 }, { "epoch": 0.15598, "grad_norm": 1.2093019393117734, "learning_rate": 0.003, "loss": 4.0526, "step": 15598 }, { "epoch": 0.15599, "grad_norm": 1.3478666873443483, "learning_rate": 0.003, "loss": 4.0391, "step": 15599 }, { "epoch": 0.156, "grad_norm": 1.1657516417400606, "learning_rate": 0.003, "loss": 4.0357, "step": 15600 }, { "epoch": 0.15601, "grad_norm": 1.358518532556128, "learning_rate": 0.003, "loss": 4.0593, "step": 15601 }, { "epoch": 0.15602, "grad_norm": 1.1803239294661774, "learning_rate": 0.003, "loss": 4.0739, "step": 15602 }, { "epoch": 0.15603, "grad_norm": 1.2955861708868923, "learning_rate": 0.003, "loss": 4.0378, "step": 15603 }, { "epoch": 0.15604, "grad_norm": 1.0399032409655542, "learning_rate": 0.003, "loss": 4.0575, "step": 15604 }, { "epoch": 0.15605, "grad_norm": 1.2609466127421523, "learning_rate": 0.003, "loss": 4.0664, "step": 15605 }, { "epoch": 0.15606, "grad_norm": 0.9914988599779099, "learning_rate": 0.003, "loss": 4.0626, "step": 15606 }, { "epoch": 0.15607, "grad_norm": 1.4370190397876632, "learning_rate": 0.003, "loss": 4.0524, "step": 15607 }, { "epoch": 0.15608, "grad_norm": 1.1799076597681437, "learning_rate": 0.003, "loss": 4.0653, "step": 15608 }, { "epoch": 0.15609, "grad_norm": 1.4752831292983322, "learning_rate": 0.003, "loss": 4.0791, "step": 15609 }, { "epoch": 0.1561, "grad_norm": 1.039199974330079, "learning_rate": 0.003, "loss": 4.0439, "step": 15610 }, { "epoch": 0.15611, "grad_norm": 1.2253487449069982, "learning_rate": 0.003, "loss": 4.0744, "step": 15611 }, { "epoch": 0.15612, "grad_norm": 1.093840574448093, "learning_rate": 0.003, "loss": 4.0423, "step": 15612 }, { "epoch": 0.15613, "grad_norm": 1.3571446766506403, "learning_rate": 0.003, "loss": 4.0411, "step": 15613 }, { "epoch": 0.15614, "grad_norm": 1.0553760636201928, "learning_rate": 0.003, "loss": 4.0448, "step": 15614 }, { "epoch": 0.15615, "grad_norm": 1.3235541151541643, "learning_rate": 0.003, "loss": 4.0459, "step": 15615 }, { "epoch": 0.15616, "grad_norm": 1.2661053326679264, "learning_rate": 0.003, "loss": 4.0633, "step": 15616 }, { "epoch": 0.15617, "grad_norm": 0.9442433454080968, "learning_rate": 0.003, "loss": 4.0798, "step": 15617 }, { "epoch": 0.15618, "grad_norm": 1.5305838600867183, "learning_rate": 0.003, "loss": 4.0463, "step": 15618 }, { "epoch": 0.15619, "grad_norm": 1.0971756564789878, "learning_rate": 0.003, "loss": 4.0719, "step": 15619 }, { "epoch": 0.1562, "grad_norm": 1.534084990212608, "learning_rate": 0.003, "loss": 4.0625, "step": 15620 }, { "epoch": 0.15621, "grad_norm": 1.0651697006964178, "learning_rate": 0.003, "loss": 4.0468, "step": 15621 }, { "epoch": 0.15622, "grad_norm": 1.2181410597137075, "learning_rate": 0.003, "loss": 4.0864, "step": 15622 }, { "epoch": 0.15623, "grad_norm": 1.3182947944513597, "learning_rate": 0.003, "loss": 4.0515, "step": 15623 }, { "epoch": 0.15624, "grad_norm": 1.0913383790969486, "learning_rate": 0.003, "loss": 4.0881, "step": 15624 }, { "epoch": 0.15625, "grad_norm": 1.3833835876314324, "learning_rate": 0.003, "loss": 4.0609, "step": 15625 }, { "epoch": 0.15626, "grad_norm": 0.9309552488047658, "learning_rate": 0.003, "loss": 4.059, "step": 15626 }, { "epoch": 0.15627, "grad_norm": 1.1570848674176404, "learning_rate": 0.003, "loss": 4.0692, "step": 15627 }, { "epoch": 0.15628, "grad_norm": 1.176935804098361, "learning_rate": 0.003, "loss": 4.042, "step": 15628 }, { "epoch": 0.15629, "grad_norm": 1.0719613956642227, "learning_rate": 0.003, "loss": 4.0539, "step": 15629 }, { "epoch": 0.1563, "grad_norm": 1.212478073422822, "learning_rate": 0.003, "loss": 4.0822, "step": 15630 }, { "epoch": 0.15631, "grad_norm": 1.2547829857423518, "learning_rate": 0.003, "loss": 4.0643, "step": 15631 }, { "epoch": 0.15632, "grad_norm": 1.435656586125539, "learning_rate": 0.003, "loss": 4.0833, "step": 15632 }, { "epoch": 0.15633, "grad_norm": 1.2244113407724508, "learning_rate": 0.003, "loss": 4.0844, "step": 15633 }, { "epoch": 0.15634, "grad_norm": 1.3451385172699777, "learning_rate": 0.003, "loss": 4.0863, "step": 15634 }, { "epoch": 0.15635, "grad_norm": 0.9828019537442368, "learning_rate": 0.003, "loss": 4.061, "step": 15635 }, { "epoch": 0.15636, "grad_norm": 1.309784793149415, "learning_rate": 0.003, "loss": 4.0644, "step": 15636 }, { "epoch": 0.15637, "grad_norm": 1.3062190678521401, "learning_rate": 0.003, "loss": 4.091, "step": 15637 }, { "epoch": 0.15638, "grad_norm": 1.165473176393098, "learning_rate": 0.003, "loss": 4.0379, "step": 15638 }, { "epoch": 0.15639, "grad_norm": 1.1670306993408306, "learning_rate": 0.003, "loss": 4.0432, "step": 15639 }, { "epoch": 0.1564, "grad_norm": 1.1486624058525117, "learning_rate": 0.003, "loss": 4.0672, "step": 15640 }, { "epoch": 0.15641, "grad_norm": 1.139740792872974, "learning_rate": 0.003, "loss": 4.0434, "step": 15641 }, { "epoch": 0.15642, "grad_norm": 1.1265148410314216, "learning_rate": 0.003, "loss": 4.0695, "step": 15642 }, { "epoch": 0.15643, "grad_norm": 1.275213876165096, "learning_rate": 0.003, "loss": 4.0812, "step": 15643 }, { "epoch": 0.15644, "grad_norm": 1.0717084212859374, "learning_rate": 0.003, "loss": 4.0487, "step": 15644 }, { "epoch": 0.15645, "grad_norm": 1.583705740950635, "learning_rate": 0.003, "loss": 4.0808, "step": 15645 }, { "epoch": 0.15646, "grad_norm": 1.102649284741737, "learning_rate": 0.003, "loss": 4.028, "step": 15646 }, { "epoch": 0.15647, "grad_norm": 1.2455052029010671, "learning_rate": 0.003, "loss": 4.0473, "step": 15647 }, { "epoch": 0.15648, "grad_norm": 1.1972247406763779, "learning_rate": 0.003, "loss": 4.0777, "step": 15648 }, { "epoch": 0.15649, "grad_norm": 1.226549854995466, "learning_rate": 0.003, "loss": 4.0665, "step": 15649 }, { "epoch": 0.1565, "grad_norm": 1.1847153962415675, "learning_rate": 0.003, "loss": 4.0168, "step": 15650 }, { "epoch": 0.15651, "grad_norm": 1.2756284529825144, "learning_rate": 0.003, "loss": 4.0709, "step": 15651 }, { "epoch": 0.15652, "grad_norm": 1.041978770124662, "learning_rate": 0.003, "loss": 4.08, "step": 15652 }, { "epoch": 0.15653, "grad_norm": 1.4074424570284112, "learning_rate": 0.003, "loss": 4.057, "step": 15653 }, { "epoch": 0.15654, "grad_norm": 1.0815867887693429, "learning_rate": 0.003, "loss": 4.0547, "step": 15654 }, { "epoch": 0.15655, "grad_norm": 1.1911344281142204, "learning_rate": 0.003, "loss": 4.1043, "step": 15655 }, { "epoch": 0.15656, "grad_norm": 1.3363273582630495, "learning_rate": 0.003, "loss": 4.0555, "step": 15656 }, { "epoch": 0.15657, "grad_norm": 1.4034885185783113, "learning_rate": 0.003, "loss": 4.0671, "step": 15657 }, { "epoch": 0.15658, "grad_norm": 1.3287131511830883, "learning_rate": 0.003, "loss": 4.072, "step": 15658 }, { "epoch": 0.15659, "grad_norm": 1.272868305526661, "learning_rate": 0.003, "loss": 4.0798, "step": 15659 }, { "epoch": 0.1566, "grad_norm": 1.3610441579515373, "learning_rate": 0.003, "loss": 4.0771, "step": 15660 }, { "epoch": 0.15661, "grad_norm": 1.1070335416027541, "learning_rate": 0.003, "loss": 4.0524, "step": 15661 }, { "epoch": 0.15662, "grad_norm": 1.137121323673713, "learning_rate": 0.003, "loss": 4.0481, "step": 15662 }, { "epoch": 0.15663, "grad_norm": 1.1688201307165267, "learning_rate": 0.003, "loss": 4.0581, "step": 15663 }, { "epoch": 0.15664, "grad_norm": 1.1884300419398568, "learning_rate": 0.003, "loss": 4.0724, "step": 15664 }, { "epoch": 0.15665, "grad_norm": 1.1153395770391112, "learning_rate": 0.003, "loss": 4.0646, "step": 15665 }, { "epoch": 0.15666, "grad_norm": 1.3654060929395346, "learning_rate": 0.003, "loss": 4.0534, "step": 15666 }, { "epoch": 0.15667, "grad_norm": 1.056118733006509, "learning_rate": 0.003, "loss": 4.0609, "step": 15667 }, { "epoch": 0.15668, "grad_norm": 1.3586731584276457, "learning_rate": 0.003, "loss": 4.0459, "step": 15668 }, { "epoch": 0.15669, "grad_norm": 1.1796131792325408, "learning_rate": 0.003, "loss": 4.056, "step": 15669 }, { "epoch": 0.1567, "grad_norm": 1.5477999205246045, "learning_rate": 0.003, "loss": 4.0454, "step": 15670 }, { "epoch": 0.15671, "grad_norm": 1.0676276941822342, "learning_rate": 0.003, "loss": 4.0577, "step": 15671 }, { "epoch": 0.15672, "grad_norm": 1.2235310848942378, "learning_rate": 0.003, "loss": 4.0548, "step": 15672 }, { "epoch": 0.15673, "grad_norm": 1.1832277832368536, "learning_rate": 0.003, "loss": 4.0827, "step": 15673 }, { "epoch": 0.15674, "grad_norm": 1.1765804606495207, "learning_rate": 0.003, "loss": 4.0611, "step": 15674 }, { "epoch": 0.15675, "grad_norm": 1.2041687186960341, "learning_rate": 0.003, "loss": 4.0643, "step": 15675 }, { "epoch": 0.15676, "grad_norm": 1.0160506887649807, "learning_rate": 0.003, "loss": 4.0743, "step": 15676 }, { "epoch": 0.15677, "grad_norm": 1.2126187401023711, "learning_rate": 0.003, "loss": 4.0366, "step": 15677 }, { "epoch": 0.15678, "grad_norm": 0.9716988481086387, "learning_rate": 0.003, "loss": 4.0621, "step": 15678 }, { "epoch": 0.15679, "grad_norm": 1.2613322000198528, "learning_rate": 0.003, "loss": 4.0742, "step": 15679 }, { "epoch": 0.1568, "grad_norm": 1.0382859046183213, "learning_rate": 0.003, "loss": 4.0506, "step": 15680 }, { "epoch": 0.15681, "grad_norm": 1.3604619130798723, "learning_rate": 0.003, "loss": 4.0981, "step": 15681 }, { "epoch": 0.15682, "grad_norm": 1.1817349333719727, "learning_rate": 0.003, "loss": 4.0576, "step": 15682 }, { "epoch": 0.15683, "grad_norm": 1.1425818004623922, "learning_rate": 0.003, "loss": 4.053, "step": 15683 }, { "epoch": 0.15684, "grad_norm": 1.4118023735842253, "learning_rate": 0.003, "loss": 4.0776, "step": 15684 }, { "epoch": 0.15685, "grad_norm": 1.0797767177312902, "learning_rate": 0.003, "loss": 4.0614, "step": 15685 }, { "epoch": 0.15686, "grad_norm": 1.3559641912318932, "learning_rate": 0.003, "loss": 4.0785, "step": 15686 }, { "epoch": 0.15687, "grad_norm": 1.180834137733882, "learning_rate": 0.003, "loss": 4.0791, "step": 15687 }, { "epoch": 0.15688, "grad_norm": 1.2901718630974206, "learning_rate": 0.003, "loss": 4.0509, "step": 15688 }, { "epoch": 0.15689, "grad_norm": 1.057343694698553, "learning_rate": 0.003, "loss": 4.0771, "step": 15689 }, { "epoch": 0.1569, "grad_norm": 1.4531352327451286, "learning_rate": 0.003, "loss": 4.0512, "step": 15690 }, { "epoch": 0.15691, "grad_norm": 1.059809813801114, "learning_rate": 0.003, "loss": 4.0547, "step": 15691 }, { "epoch": 0.15692, "grad_norm": 1.326585675340708, "learning_rate": 0.003, "loss": 4.0429, "step": 15692 }, { "epoch": 0.15693, "grad_norm": 1.0546296617877726, "learning_rate": 0.003, "loss": 4.0473, "step": 15693 }, { "epoch": 0.15694, "grad_norm": 1.1803547660970806, "learning_rate": 0.003, "loss": 4.0742, "step": 15694 }, { "epoch": 0.15695, "grad_norm": 1.1205652925533187, "learning_rate": 0.003, "loss": 4.0582, "step": 15695 }, { "epoch": 0.15696, "grad_norm": 1.0476974385425437, "learning_rate": 0.003, "loss": 4.0542, "step": 15696 }, { "epoch": 0.15697, "grad_norm": 1.284802117087022, "learning_rate": 0.003, "loss": 4.0571, "step": 15697 }, { "epoch": 0.15698, "grad_norm": 1.0799119338890766, "learning_rate": 0.003, "loss": 4.0686, "step": 15698 }, { "epoch": 0.15699, "grad_norm": 1.3706470506218387, "learning_rate": 0.003, "loss": 4.1048, "step": 15699 }, { "epoch": 0.157, "grad_norm": 1.0082630839398643, "learning_rate": 0.003, "loss": 4.0363, "step": 15700 }, { "epoch": 0.15701, "grad_norm": 1.445676153125903, "learning_rate": 0.003, "loss": 4.0517, "step": 15701 }, { "epoch": 0.15702, "grad_norm": 1.064353835615799, "learning_rate": 0.003, "loss": 4.0576, "step": 15702 }, { "epoch": 0.15703, "grad_norm": 1.2484061812785103, "learning_rate": 0.003, "loss": 4.0698, "step": 15703 }, { "epoch": 0.15704, "grad_norm": 1.4801219200898403, "learning_rate": 0.003, "loss": 4.0727, "step": 15704 }, { "epoch": 0.15705, "grad_norm": 1.2584117781793354, "learning_rate": 0.003, "loss": 4.0512, "step": 15705 }, { "epoch": 0.15706, "grad_norm": 1.3361614955363434, "learning_rate": 0.003, "loss": 4.0484, "step": 15706 }, { "epoch": 0.15707, "grad_norm": 1.2896463924521695, "learning_rate": 0.003, "loss": 4.0653, "step": 15707 }, { "epoch": 0.15708, "grad_norm": 1.1868240849069513, "learning_rate": 0.003, "loss": 4.0656, "step": 15708 }, { "epoch": 0.15709, "grad_norm": 1.2353964823213757, "learning_rate": 0.003, "loss": 4.0651, "step": 15709 }, { "epoch": 0.1571, "grad_norm": 1.3438687182860078, "learning_rate": 0.003, "loss": 4.0865, "step": 15710 }, { "epoch": 0.15711, "grad_norm": 1.3063674960988663, "learning_rate": 0.003, "loss": 4.0521, "step": 15711 }, { "epoch": 0.15712, "grad_norm": 1.1174297258576622, "learning_rate": 0.003, "loss": 4.069, "step": 15712 }, { "epoch": 0.15713, "grad_norm": 1.2206113182977072, "learning_rate": 0.003, "loss": 4.0448, "step": 15713 }, { "epoch": 0.15714, "grad_norm": 0.9891530851223427, "learning_rate": 0.003, "loss": 4.0498, "step": 15714 }, { "epoch": 0.15715, "grad_norm": 1.5402493043887262, "learning_rate": 0.003, "loss": 4.0515, "step": 15715 }, { "epoch": 0.15716, "grad_norm": 1.0024584399723404, "learning_rate": 0.003, "loss": 4.0812, "step": 15716 }, { "epoch": 0.15717, "grad_norm": 1.5360847032225067, "learning_rate": 0.003, "loss": 4.0727, "step": 15717 }, { "epoch": 0.15718, "grad_norm": 1.063704056965126, "learning_rate": 0.003, "loss": 4.0642, "step": 15718 }, { "epoch": 0.15719, "grad_norm": 1.342678153024432, "learning_rate": 0.003, "loss": 4.0706, "step": 15719 }, { "epoch": 0.1572, "grad_norm": 1.1265734694726142, "learning_rate": 0.003, "loss": 4.068, "step": 15720 }, { "epoch": 0.15721, "grad_norm": 1.2662125532950392, "learning_rate": 0.003, "loss": 4.0563, "step": 15721 }, { "epoch": 0.15722, "grad_norm": 1.1696121939389466, "learning_rate": 0.003, "loss": 4.0527, "step": 15722 }, { "epoch": 0.15723, "grad_norm": 1.1858634514564075, "learning_rate": 0.003, "loss": 4.0847, "step": 15723 }, { "epoch": 0.15724, "grad_norm": 1.2401541222621788, "learning_rate": 0.003, "loss": 4.0582, "step": 15724 }, { "epoch": 0.15725, "grad_norm": 1.1757637701745292, "learning_rate": 0.003, "loss": 4.0341, "step": 15725 }, { "epoch": 0.15726, "grad_norm": 1.1936334840612361, "learning_rate": 0.003, "loss": 4.0344, "step": 15726 }, { "epoch": 0.15727, "grad_norm": 0.9496985596585019, "learning_rate": 0.003, "loss": 4.0597, "step": 15727 }, { "epoch": 0.15728, "grad_norm": 1.3178393298136768, "learning_rate": 0.003, "loss": 4.0681, "step": 15728 }, { "epoch": 0.15729, "grad_norm": 1.375278498748954, "learning_rate": 0.003, "loss": 4.0717, "step": 15729 }, { "epoch": 0.1573, "grad_norm": 1.224750356040554, "learning_rate": 0.003, "loss": 4.0684, "step": 15730 }, { "epoch": 0.15731, "grad_norm": 1.1842857042512631, "learning_rate": 0.003, "loss": 4.0422, "step": 15731 }, { "epoch": 0.15732, "grad_norm": 1.276922944243024, "learning_rate": 0.003, "loss": 4.0383, "step": 15732 }, { "epoch": 0.15733, "grad_norm": 1.1616866651523663, "learning_rate": 0.003, "loss": 4.0429, "step": 15733 }, { "epoch": 0.15734, "grad_norm": 1.271758163927367, "learning_rate": 0.003, "loss": 4.0713, "step": 15734 }, { "epoch": 0.15735, "grad_norm": 1.0451445922159956, "learning_rate": 0.003, "loss": 4.0666, "step": 15735 }, { "epoch": 0.15736, "grad_norm": 1.12669786390475, "learning_rate": 0.003, "loss": 4.0646, "step": 15736 }, { "epoch": 0.15737, "grad_norm": 1.081969957241331, "learning_rate": 0.003, "loss": 4.0727, "step": 15737 }, { "epoch": 0.15738, "grad_norm": 1.3554078152104214, "learning_rate": 0.003, "loss": 4.0646, "step": 15738 }, { "epoch": 0.15739, "grad_norm": 1.1487176707527527, "learning_rate": 0.003, "loss": 4.0575, "step": 15739 }, { "epoch": 0.1574, "grad_norm": 1.3685503250200426, "learning_rate": 0.003, "loss": 4.0622, "step": 15740 }, { "epoch": 0.15741, "grad_norm": 1.1669559736978832, "learning_rate": 0.003, "loss": 4.0655, "step": 15741 }, { "epoch": 0.15742, "grad_norm": 1.2886763102225725, "learning_rate": 0.003, "loss": 4.0557, "step": 15742 }, { "epoch": 0.15743, "grad_norm": 1.3166304422510648, "learning_rate": 0.003, "loss": 4.0508, "step": 15743 }, { "epoch": 0.15744, "grad_norm": 1.4377739852078564, "learning_rate": 0.003, "loss": 4.0396, "step": 15744 }, { "epoch": 0.15745, "grad_norm": 1.0733206043203236, "learning_rate": 0.003, "loss": 4.0633, "step": 15745 }, { "epoch": 0.15746, "grad_norm": 1.302632631536053, "learning_rate": 0.003, "loss": 4.0457, "step": 15746 }, { "epoch": 0.15747, "grad_norm": 1.1271031367568127, "learning_rate": 0.003, "loss": 4.0662, "step": 15747 }, { "epoch": 0.15748, "grad_norm": 1.249087683227782, "learning_rate": 0.003, "loss": 4.0545, "step": 15748 }, { "epoch": 0.15749, "grad_norm": 0.9972463264110328, "learning_rate": 0.003, "loss": 4.0517, "step": 15749 }, { "epoch": 0.1575, "grad_norm": 1.29105919726399, "learning_rate": 0.003, "loss": 4.0516, "step": 15750 }, { "epoch": 0.15751, "grad_norm": 0.9443062712650601, "learning_rate": 0.003, "loss": 4.0435, "step": 15751 }, { "epoch": 0.15752, "grad_norm": 1.230321224780944, "learning_rate": 0.003, "loss": 4.0758, "step": 15752 }, { "epoch": 0.15753, "grad_norm": 1.1393553130448237, "learning_rate": 0.003, "loss": 4.0783, "step": 15753 }, { "epoch": 0.15754, "grad_norm": 1.1904480545785132, "learning_rate": 0.003, "loss": 4.0696, "step": 15754 }, { "epoch": 0.15755, "grad_norm": 1.068113412072056, "learning_rate": 0.003, "loss": 4.0497, "step": 15755 }, { "epoch": 0.15756, "grad_norm": 1.5412000458981607, "learning_rate": 0.003, "loss": 4.094, "step": 15756 }, { "epoch": 0.15757, "grad_norm": 1.1302032111036926, "learning_rate": 0.003, "loss": 4.0313, "step": 15757 }, { "epoch": 0.15758, "grad_norm": 1.4361441260651815, "learning_rate": 0.003, "loss": 4.0772, "step": 15758 }, { "epoch": 0.15759, "grad_norm": 1.185263160492388, "learning_rate": 0.003, "loss": 4.0663, "step": 15759 }, { "epoch": 0.1576, "grad_norm": 1.1936371622024107, "learning_rate": 0.003, "loss": 4.0323, "step": 15760 }, { "epoch": 0.15761, "grad_norm": 1.2967703711658523, "learning_rate": 0.003, "loss": 4.0212, "step": 15761 }, { "epoch": 0.15762, "grad_norm": 1.1455304588777764, "learning_rate": 0.003, "loss": 4.0477, "step": 15762 }, { "epoch": 0.15763, "grad_norm": 1.4246356741152995, "learning_rate": 0.003, "loss": 4.0611, "step": 15763 }, { "epoch": 0.15764, "grad_norm": 0.9812512488800511, "learning_rate": 0.003, "loss": 4.0193, "step": 15764 }, { "epoch": 0.15765, "grad_norm": 1.0685826322848317, "learning_rate": 0.003, "loss": 4.0648, "step": 15765 }, { "epoch": 0.15766, "grad_norm": 1.442939948848548, "learning_rate": 0.003, "loss": 4.0613, "step": 15766 }, { "epoch": 0.15767, "grad_norm": 1.0450538289646265, "learning_rate": 0.003, "loss": 4.0771, "step": 15767 }, { "epoch": 0.15768, "grad_norm": 1.355954874873054, "learning_rate": 0.003, "loss": 4.0571, "step": 15768 }, { "epoch": 0.15769, "grad_norm": 0.9825217012228542, "learning_rate": 0.003, "loss": 4.0774, "step": 15769 }, { "epoch": 0.1577, "grad_norm": 1.2113682167958801, "learning_rate": 0.003, "loss": 4.071, "step": 15770 }, { "epoch": 0.15771, "grad_norm": 1.1721148970936335, "learning_rate": 0.003, "loss": 4.0638, "step": 15771 }, { "epoch": 0.15772, "grad_norm": 1.170800408276965, "learning_rate": 0.003, "loss": 4.0939, "step": 15772 }, { "epoch": 0.15773, "grad_norm": 1.4323770606252804, "learning_rate": 0.003, "loss": 4.0763, "step": 15773 }, { "epoch": 0.15774, "grad_norm": 1.1113737649682414, "learning_rate": 0.003, "loss": 4.0588, "step": 15774 }, { "epoch": 0.15775, "grad_norm": 1.2427021366448472, "learning_rate": 0.003, "loss": 4.0387, "step": 15775 }, { "epoch": 0.15776, "grad_norm": 1.2605703342585746, "learning_rate": 0.003, "loss": 4.0542, "step": 15776 }, { "epoch": 0.15777, "grad_norm": 1.1862357476961214, "learning_rate": 0.003, "loss": 4.0455, "step": 15777 }, { "epoch": 0.15778, "grad_norm": 1.1628968647887679, "learning_rate": 0.003, "loss": 4.0515, "step": 15778 }, { "epoch": 0.15779, "grad_norm": 1.135395616114251, "learning_rate": 0.003, "loss": 4.047, "step": 15779 }, { "epoch": 0.1578, "grad_norm": 1.2833211835181242, "learning_rate": 0.003, "loss": 4.0734, "step": 15780 }, { "epoch": 0.15781, "grad_norm": 1.219054473115084, "learning_rate": 0.003, "loss": 4.0039, "step": 15781 }, { "epoch": 0.15782, "grad_norm": 1.3614090599157573, "learning_rate": 0.003, "loss": 4.08, "step": 15782 }, { "epoch": 0.15783, "grad_norm": 1.1023328220848818, "learning_rate": 0.003, "loss": 4.0518, "step": 15783 }, { "epoch": 0.15784, "grad_norm": 1.3507662672711478, "learning_rate": 0.003, "loss": 4.0531, "step": 15784 }, { "epoch": 0.15785, "grad_norm": 1.277368068591102, "learning_rate": 0.003, "loss": 4.0704, "step": 15785 }, { "epoch": 0.15786, "grad_norm": 1.2831362884284865, "learning_rate": 0.003, "loss": 4.0451, "step": 15786 }, { "epoch": 0.15787, "grad_norm": 1.2327201269925094, "learning_rate": 0.003, "loss": 4.0706, "step": 15787 }, { "epoch": 0.15788, "grad_norm": 1.3013672361448259, "learning_rate": 0.003, "loss": 4.0834, "step": 15788 }, { "epoch": 0.15789, "grad_norm": 1.1977439741562261, "learning_rate": 0.003, "loss": 4.0521, "step": 15789 }, { "epoch": 0.1579, "grad_norm": 1.3018864470607883, "learning_rate": 0.003, "loss": 4.065, "step": 15790 }, { "epoch": 0.15791, "grad_norm": 1.028971036425535, "learning_rate": 0.003, "loss": 4.0737, "step": 15791 }, { "epoch": 0.15792, "grad_norm": 1.316986973616366, "learning_rate": 0.003, "loss": 4.0706, "step": 15792 }, { "epoch": 0.15793, "grad_norm": 1.0587295674584893, "learning_rate": 0.003, "loss": 4.0916, "step": 15793 }, { "epoch": 0.15794, "grad_norm": 1.4539141011250214, "learning_rate": 0.003, "loss": 4.0844, "step": 15794 }, { "epoch": 0.15795, "grad_norm": 1.301643316539598, "learning_rate": 0.003, "loss": 4.0815, "step": 15795 }, { "epoch": 0.15796, "grad_norm": 1.240351388241122, "learning_rate": 0.003, "loss": 4.0717, "step": 15796 }, { "epoch": 0.15797, "grad_norm": 1.2002682131440259, "learning_rate": 0.003, "loss": 4.0283, "step": 15797 }, { "epoch": 0.15798, "grad_norm": 1.056315126378722, "learning_rate": 0.003, "loss": 4.0788, "step": 15798 }, { "epoch": 0.15799, "grad_norm": 1.4008096337287652, "learning_rate": 0.003, "loss": 4.0461, "step": 15799 }, { "epoch": 0.158, "grad_norm": 1.1423018732564754, "learning_rate": 0.003, "loss": 4.0438, "step": 15800 }, { "epoch": 0.15801, "grad_norm": 1.171708557390231, "learning_rate": 0.003, "loss": 4.054, "step": 15801 }, { "epoch": 0.15802, "grad_norm": 0.9453108222646237, "learning_rate": 0.003, "loss": 4.0536, "step": 15802 }, { "epoch": 0.15803, "grad_norm": 1.1751428344710695, "learning_rate": 0.003, "loss": 4.0329, "step": 15803 }, { "epoch": 0.15804, "grad_norm": 1.306193556816233, "learning_rate": 0.003, "loss": 4.0572, "step": 15804 }, { "epoch": 0.15805, "grad_norm": 1.030461592193479, "learning_rate": 0.003, "loss": 4.0763, "step": 15805 }, { "epoch": 0.15806, "grad_norm": 1.245678365494413, "learning_rate": 0.003, "loss": 4.0766, "step": 15806 }, { "epoch": 0.15807, "grad_norm": 1.1662603675502075, "learning_rate": 0.003, "loss": 4.0745, "step": 15807 }, { "epoch": 0.15808, "grad_norm": 1.3387633832600623, "learning_rate": 0.003, "loss": 4.0441, "step": 15808 }, { "epoch": 0.15809, "grad_norm": 1.240465003088661, "learning_rate": 0.003, "loss": 4.0582, "step": 15809 }, { "epoch": 0.1581, "grad_norm": 1.209119805612708, "learning_rate": 0.003, "loss": 4.0605, "step": 15810 }, { "epoch": 0.15811, "grad_norm": 1.3162005653315245, "learning_rate": 0.003, "loss": 4.0769, "step": 15811 }, { "epoch": 0.15812, "grad_norm": 1.0192654482072048, "learning_rate": 0.003, "loss": 4.0466, "step": 15812 }, { "epoch": 0.15813, "grad_norm": 1.1285453507352443, "learning_rate": 0.003, "loss": 4.0595, "step": 15813 }, { "epoch": 0.15814, "grad_norm": 1.0936354239745236, "learning_rate": 0.003, "loss": 4.0671, "step": 15814 }, { "epoch": 0.15815, "grad_norm": 1.1330306276771145, "learning_rate": 0.003, "loss": 4.0508, "step": 15815 }, { "epoch": 0.15816, "grad_norm": 1.3275264189382847, "learning_rate": 0.003, "loss": 4.0735, "step": 15816 }, { "epoch": 0.15817, "grad_norm": 1.2532559093461328, "learning_rate": 0.003, "loss": 4.0462, "step": 15817 }, { "epoch": 0.15818, "grad_norm": 1.312860969230454, "learning_rate": 0.003, "loss": 4.0502, "step": 15818 }, { "epoch": 0.15819, "grad_norm": 1.2683436307454612, "learning_rate": 0.003, "loss": 4.0194, "step": 15819 }, { "epoch": 0.1582, "grad_norm": 1.0569106961087773, "learning_rate": 0.003, "loss": 4.0935, "step": 15820 }, { "epoch": 0.15821, "grad_norm": 1.3513616887372883, "learning_rate": 0.003, "loss": 4.0635, "step": 15821 }, { "epoch": 0.15822, "grad_norm": 1.1322845703855622, "learning_rate": 0.003, "loss": 4.0644, "step": 15822 }, { "epoch": 0.15823, "grad_norm": 1.3709174657164709, "learning_rate": 0.003, "loss": 4.0518, "step": 15823 }, { "epoch": 0.15824, "grad_norm": 1.0535463272210899, "learning_rate": 0.003, "loss": 4.0693, "step": 15824 }, { "epoch": 0.15825, "grad_norm": 1.2813176007090163, "learning_rate": 0.003, "loss": 4.0729, "step": 15825 }, { "epoch": 0.15826, "grad_norm": 1.16816717861855, "learning_rate": 0.003, "loss": 4.0573, "step": 15826 }, { "epoch": 0.15827, "grad_norm": 1.5224553137595678, "learning_rate": 0.003, "loss": 4.0408, "step": 15827 }, { "epoch": 0.15828, "grad_norm": 1.0193731373062342, "learning_rate": 0.003, "loss": 4.0376, "step": 15828 }, { "epoch": 0.15829, "grad_norm": 1.3539267964637687, "learning_rate": 0.003, "loss": 4.0691, "step": 15829 }, { "epoch": 0.1583, "grad_norm": 0.8968372894061043, "learning_rate": 0.003, "loss": 4.07, "step": 15830 }, { "epoch": 0.15831, "grad_norm": 1.393815292827624, "learning_rate": 0.003, "loss": 4.0618, "step": 15831 }, { "epoch": 0.15832, "grad_norm": 1.435291386789168, "learning_rate": 0.003, "loss": 4.0749, "step": 15832 }, { "epoch": 0.15833, "grad_norm": 1.089936174306673, "learning_rate": 0.003, "loss": 4.0392, "step": 15833 }, { "epoch": 0.15834, "grad_norm": 1.3391268534335066, "learning_rate": 0.003, "loss": 4.0757, "step": 15834 }, { "epoch": 0.15835, "grad_norm": 1.1837628962250406, "learning_rate": 0.003, "loss": 4.0599, "step": 15835 }, { "epoch": 0.15836, "grad_norm": 1.2177305620915122, "learning_rate": 0.003, "loss": 4.0534, "step": 15836 }, { "epoch": 0.15837, "grad_norm": 1.2467793854853642, "learning_rate": 0.003, "loss": 4.0591, "step": 15837 }, { "epoch": 0.15838, "grad_norm": 1.054378133249128, "learning_rate": 0.003, "loss": 4.0614, "step": 15838 }, { "epoch": 0.15839, "grad_norm": 1.2407604744450886, "learning_rate": 0.003, "loss": 4.0442, "step": 15839 }, { "epoch": 0.1584, "grad_norm": 1.0413195885226798, "learning_rate": 0.003, "loss": 4.0484, "step": 15840 }, { "epoch": 0.15841, "grad_norm": 1.5920472604532057, "learning_rate": 0.003, "loss": 4.0826, "step": 15841 }, { "epoch": 0.15842, "grad_norm": 1.1342416919557914, "learning_rate": 0.003, "loss": 4.0629, "step": 15842 }, { "epoch": 0.15843, "grad_norm": 1.393455769747034, "learning_rate": 0.003, "loss": 4.0755, "step": 15843 }, { "epoch": 0.15844, "grad_norm": 1.2981772033429155, "learning_rate": 0.003, "loss": 4.072, "step": 15844 }, { "epoch": 0.15845, "grad_norm": 1.0921851196401828, "learning_rate": 0.003, "loss": 4.0844, "step": 15845 }, { "epoch": 0.15846, "grad_norm": 1.290066187713939, "learning_rate": 0.003, "loss": 4.0673, "step": 15846 }, { "epoch": 0.15847, "grad_norm": 1.2975045304029453, "learning_rate": 0.003, "loss": 4.0569, "step": 15847 }, { "epoch": 0.15848, "grad_norm": 0.9903688537248176, "learning_rate": 0.003, "loss": 4.0412, "step": 15848 }, { "epoch": 0.15849, "grad_norm": 1.6142623172682926, "learning_rate": 0.003, "loss": 4.0795, "step": 15849 }, { "epoch": 0.1585, "grad_norm": 0.985378213981797, "learning_rate": 0.003, "loss": 4.0882, "step": 15850 }, { "epoch": 0.15851, "grad_norm": 1.3318945081444562, "learning_rate": 0.003, "loss": 4.0621, "step": 15851 }, { "epoch": 0.15852, "grad_norm": 1.1256886562107775, "learning_rate": 0.003, "loss": 4.0389, "step": 15852 }, { "epoch": 0.15853, "grad_norm": 1.296262931962511, "learning_rate": 0.003, "loss": 4.0529, "step": 15853 }, { "epoch": 0.15854, "grad_norm": 1.2919035541279742, "learning_rate": 0.003, "loss": 4.0441, "step": 15854 }, { "epoch": 0.15855, "grad_norm": 1.3608356785547568, "learning_rate": 0.003, "loss": 4.0657, "step": 15855 }, { "epoch": 0.15856, "grad_norm": 1.2760762183953474, "learning_rate": 0.003, "loss": 4.0894, "step": 15856 }, { "epoch": 0.15857, "grad_norm": 1.1121899650478617, "learning_rate": 0.003, "loss": 4.0586, "step": 15857 }, { "epoch": 0.15858, "grad_norm": 1.363693216241739, "learning_rate": 0.003, "loss": 4.0826, "step": 15858 }, { "epoch": 0.15859, "grad_norm": 1.0983535650642993, "learning_rate": 0.003, "loss": 4.0852, "step": 15859 }, { "epoch": 0.1586, "grad_norm": 1.4565610602528292, "learning_rate": 0.003, "loss": 4.0607, "step": 15860 }, { "epoch": 0.15861, "grad_norm": 1.1219134267588653, "learning_rate": 0.003, "loss": 4.0551, "step": 15861 }, { "epoch": 0.15862, "grad_norm": 0.9972620894806278, "learning_rate": 0.003, "loss": 4.0826, "step": 15862 }, { "epoch": 0.15863, "grad_norm": 1.3566778235730177, "learning_rate": 0.003, "loss": 4.0418, "step": 15863 }, { "epoch": 0.15864, "grad_norm": 1.112461176260205, "learning_rate": 0.003, "loss": 4.0697, "step": 15864 }, { "epoch": 0.15865, "grad_norm": 1.1576143537930752, "learning_rate": 0.003, "loss": 4.0509, "step": 15865 }, { "epoch": 0.15866, "grad_norm": 1.0190099145159142, "learning_rate": 0.003, "loss": 4.0546, "step": 15866 }, { "epoch": 0.15867, "grad_norm": 1.402365337674979, "learning_rate": 0.003, "loss": 4.0675, "step": 15867 }, { "epoch": 0.15868, "grad_norm": 1.053246113011527, "learning_rate": 0.003, "loss": 4.0337, "step": 15868 }, { "epoch": 0.15869, "grad_norm": 1.2852503700895828, "learning_rate": 0.003, "loss": 4.0358, "step": 15869 }, { "epoch": 0.1587, "grad_norm": 1.132776331866809, "learning_rate": 0.003, "loss": 4.0422, "step": 15870 }, { "epoch": 0.15871, "grad_norm": 1.3188795308390149, "learning_rate": 0.003, "loss": 4.0683, "step": 15871 }, { "epoch": 0.15872, "grad_norm": 1.212507733266001, "learning_rate": 0.003, "loss": 4.0788, "step": 15872 }, { "epoch": 0.15873, "grad_norm": 1.308345556908113, "learning_rate": 0.003, "loss": 4.0702, "step": 15873 }, { "epoch": 0.15874, "grad_norm": 1.2011406331704382, "learning_rate": 0.003, "loss": 4.0388, "step": 15874 }, { "epoch": 0.15875, "grad_norm": 1.4341293399013142, "learning_rate": 0.003, "loss": 4.0394, "step": 15875 }, { "epoch": 0.15876, "grad_norm": 1.546959866294749, "learning_rate": 0.003, "loss": 4.0752, "step": 15876 }, { "epoch": 0.15877, "grad_norm": 1.0426492002607934, "learning_rate": 0.003, "loss": 4.079, "step": 15877 }, { "epoch": 0.15878, "grad_norm": 1.3141174054097085, "learning_rate": 0.003, "loss": 4.0492, "step": 15878 }, { "epoch": 0.15879, "grad_norm": 1.079309116038525, "learning_rate": 0.003, "loss": 4.0705, "step": 15879 }, { "epoch": 0.1588, "grad_norm": 1.1522466875849964, "learning_rate": 0.003, "loss": 4.0828, "step": 15880 }, { "epoch": 0.15881, "grad_norm": 1.3014708659787106, "learning_rate": 0.003, "loss": 4.0521, "step": 15881 }, { "epoch": 0.15882, "grad_norm": 1.0808883773056064, "learning_rate": 0.003, "loss": 4.0315, "step": 15882 }, { "epoch": 0.15883, "grad_norm": 1.2566124828480216, "learning_rate": 0.003, "loss": 4.0458, "step": 15883 }, { "epoch": 0.15884, "grad_norm": 1.2658895711654297, "learning_rate": 0.003, "loss": 4.0743, "step": 15884 }, { "epoch": 0.15885, "grad_norm": 0.9571184866919882, "learning_rate": 0.003, "loss": 4.0445, "step": 15885 }, { "epoch": 0.15886, "grad_norm": 1.283364138119938, "learning_rate": 0.003, "loss": 4.0786, "step": 15886 }, { "epoch": 0.15887, "grad_norm": 1.0923717676531308, "learning_rate": 0.003, "loss": 4.0811, "step": 15887 }, { "epoch": 0.15888, "grad_norm": 1.0991231881499555, "learning_rate": 0.003, "loss": 4.0532, "step": 15888 }, { "epoch": 0.15889, "grad_norm": 1.3272415803409467, "learning_rate": 0.003, "loss": 4.0888, "step": 15889 }, { "epoch": 0.1589, "grad_norm": 1.1785840209706901, "learning_rate": 0.003, "loss": 4.0887, "step": 15890 }, { "epoch": 0.15891, "grad_norm": 1.107506409072357, "learning_rate": 0.003, "loss": 4.0892, "step": 15891 }, { "epoch": 0.15892, "grad_norm": 1.2298490379923164, "learning_rate": 0.003, "loss": 4.0538, "step": 15892 }, { "epoch": 0.15893, "grad_norm": 1.1870014015172905, "learning_rate": 0.003, "loss": 4.0487, "step": 15893 }, { "epoch": 0.15894, "grad_norm": 1.7188385883097401, "learning_rate": 0.003, "loss": 4.0561, "step": 15894 }, { "epoch": 0.15895, "grad_norm": 1.2362021387221938, "learning_rate": 0.003, "loss": 4.071, "step": 15895 }, { "epoch": 0.15896, "grad_norm": 1.34789117487494, "learning_rate": 0.003, "loss": 4.0405, "step": 15896 }, { "epoch": 0.15897, "grad_norm": 1.0695391433208734, "learning_rate": 0.003, "loss": 4.0268, "step": 15897 }, { "epoch": 0.15898, "grad_norm": 1.1777839125106804, "learning_rate": 0.003, "loss": 4.0596, "step": 15898 }, { "epoch": 0.15899, "grad_norm": 1.0976417218398744, "learning_rate": 0.003, "loss": 4.0244, "step": 15899 }, { "epoch": 0.159, "grad_norm": 1.1801152956563512, "learning_rate": 0.003, "loss": 4.0287, "step": 15900 }, { "epoch": 0.15901, "grad_norm": 1.0112695513469703, "learning_rate": 0.003, "loss": 4.068, "step": 15901 }, { "epoch": 0.15902, "grad_norm": 1.4063076193955033, "learning_rate": 0.003, "loss": 4.1067, "step": 15902 }, { "epoch": 0.15903, "grad_norm": 1.0667191964445872, "learning_rate": 0.003, "loss": 4.0611, "step": 15903 }, { "epoch": 0.15904, "grad_norm": 1.3380817870927166, "learning_rate": 0.003, "loss": 4.0719, "step": 15904 }, { "epoch": 0.15905, "grad_norm": 1.0974570739841378, "learning_rate": 0.003, "loss": 4.0532, "step": 15905 }, { "epoch": 0.15906, "grad_norm": 1.426638793800246, "learning_rate": 0.003, "loss": 4.0863, "step": 15906 }, { "epoch": 0.15907, "grad_norm": 1.2647011527595533, "learning_rate": 0.003, "loss": 4.0682, "step": 15907 }, { "epoch": 0.15908, "grad_norm": 1.2032650178773125, "learning_rate": 0.003, "loss": 4.0557, "step": 15908 }, { "epoch": 0.15909, "grad_norm": 1.1950608703430268, "learning_rate": 0.003, "loss": 4.0738, "step": 15909 }, { "epoch": 0.1591, "grad_norm": 1.4601945009357429, "learning_rate": 0.003, "loss": 4.0723, "step": 15910 }, { "epoch": 0.15911, "grad_norm": 1.1067183552727093, "learning_rate": 0.003, "loss": 4.0699, "step": 15911 }, { "epoch": 0.15912, "grad_norm": 1.417817886596112, "learning_rate": 0.003, "loss": 4.0685, "step": 15912 }, { "epoch": 0.15913, "grad_norm": 1.2132424767181238, "learning_rate": 0.003, "loss": 4.0436, "step": 15913 }, { "epoch": 0.15914, "grad_norm": 1.1538329650841557, "learning_rate": 0.003, "loss": 4.09, "step": 15914 }, { "epoch": 0.15915, "grad_norm": 1.274924356409109, "learning_rate": 0.003, "loss": 4.0397, "step": 15915 }, { "epoch": 0.15916, "grad_norm": 1.2825839286168426, "learning_rate": 0.003, "loss": 4.0722, "step": 15916 }, { "epoch": 0.15917, "grad_norm": 1.1133627843857545, "learning_rate": 0.003, "loss": 4.0642, "step": 15917 }, { "epoch": 0.15918, "grad_norm": 1.185840545901858, "learning_rate": 0.003, "loss": 4.0684, "step": 15918 }, { "epoch": 0.15919, "grad_norm": 1.0864911009183758, "learning_rate": 0.003, "loss": 4.077, "step": 15919 }, { "epoch": 0.1592, "grad_norm": 1.3851073990719955, "learning_rate": 0.003, "loss": 4.064, "step": 15920 }, { "epoch": 0.15921, "grad_norm": 0.9656982199781505, "learning_rate": 0.003, "loss": 4.0903, "step": 15921 }, { "epoch": 0.15922, "grad_norm": 1.1446516677778402, "learning_rate": 0.003, "loss": 4.0579, "step": 15922 }, { "epoch": 0.15923, "grad_norm": 1.0606920603331644, "learning_rate": 0.003, "loss": 4.061, "step": 15923 }, { "epoch": 0.15924, "grad_norm": 1.3304356753224753, "learning_rate": 0.003, "loss": 4.0734, "step": 15924 }, { "epoch": 0.15925, "grad_norm": 1.2508962554946679, "learning_rate": 0.003, "loss": 4.0388, "step": 15925 }, { "epoch": 0.15926, "grad_norm": 1.4449873882960205, "learning_rate": 0.003, "loss": 4.0543, "step": 15926 }, { "epoch": 0.15927, "grad_norm": 1.0599813172496817, "learning_rate": 0.003, "loss": 4.0243, "step": 15927 }, { "epoch": 0.15928, "grad_norm": 1.2402991954860334, "learning_rate": 0.003, "loss": 4.0578, "step": 15928 }, { "epoch": 0.15929, "grad_norm": 1.4530991867031946, "learning_rate": 0.003, "loss": 4.0742, "step": 15929 }, { "epoch": 0.1593, "grad_norm": 1.069378091186554, "learning_rate": 0.003, "loss": 4.0623, "step": 15930 }, { "epoch": 0.15931, "grad_norm": 1.2095128452010742, "learning_rate": 0.003, "loss": 4.0589, "step": 15931 }, { "epoch": 0.15932, "grad_norm": 1.2083494374648298, "learning_rate": 0.003, "loss": 4.0547, "step": 15932 }, { "epoch": 0.15933, "grad_norm": 1.3423400065295152, "learning_rate": 0.003, "loss": 4.0424, "step": 15933 }, { "epoch": 0.15934, "grad_norm": 1.064303273571097, "learning_rate": 0.003, "loss": 4.0489, "step": 15934 }, { "epoch": 0.15935, "grad_norm": 1.4546372575740287, "learning_rate": 0.003, "loss": 4.0635, "step": 15935 }, { "epoch": 0.15936, "grad_norm": 1.0870248420346131, "learning_rate": 0.003, "loss": 4.0491, "step": 15936 }, { "epoch": 0.15937, "grad_norm": 1.2222674384931338, "learning_rate": 0.003, "loss": 4.0637, "step": 15937 }, { "epoch": 0.15938, "grad_norm": 1.2353233739383782, "learning_rate": 0.003, "loss": 4.0528, "step": 15938 }, { "epoch": 0.15939, "grad_norm": 1.2656560872610918, "learning_rate": 0.003, "loss": 4.0623, "step": 15939 }, { "epoch": 0.1594, "grad_norm": 1.3038681911989853, "learning_rate": 0.003, "loss": 4.0482, "step": 15940 }, { "epoch": 0.15941, "grad_norm": 0.9973340686690166, "learning_rate": 0.003, "loss": 4.0408, "step": 15941 }, { "epoch": 0.15942, "grad_norm": 1.422979066690338, "learning_rate": 0.003, "loss": 4.0279, "step": 15942 }, { "epoch": 0.15943, "grad_norm": 1.198234746778187, "learning_rate": 0.003, "loss": 4.0555, "step": 15943 }, { "epoch": 0.15944, "grad_norm": 1.261941686794253, "learning_rate": 0.003, "loss": 4.0715, "step": 15944 }, { "epoch": 0.15945, "grad_norm": 1.185765211650111, "learning_rate": 0.003, "loss": 4.066, "step": 15945 }, { "epoch": 0.15946, "grad_norm": 1.2466689219040417, "learning_rate": 0.003, "loss": 4.0513, "step": 15946 }, { "epoch": 0.15947, "grad_norm": 0.9899503882038371, "learning_rate": 0.003, "loss": 4.0446, "step": 15947 }, { "epoch": 0.15948, "grad_norm": 1.3385743779929415, "learning_rate": 0.003, "loss": 4.0697, "step": 15948 }, { "epoch": 0.15949, "grad_norm": 1.0195047180234955, "learning_rate": 0.003, "loss": 4.0581, "step": 15949 }, { "epoch": 0.1595, "grad_norm": 1.4255360401605368, "learning_rate": 0.003, "loss": 4.0707, "step": 15950 }, { "epoch": 0.15951, "grad_norm": 1.38313794297874, "learning_rate": 0.003, "loss": 4.0628, "step": 15951 }, { "epoch": 0.15952, "grad_norm": 1.176495580840481, "learning_rate": 0.003, "loss": 4.0572, "step": 15952 }, { "epoch": 0.15953, "grad_norm": 1.4944983800535494, "learning_rate": 0.003, "loss": 4.0842, "step": 15953 }, { "epoch": 0.15954, "grad_norm": 1.263338265873482, "learning_rate": 0.003, "loss": 4.0962, "step": 15954 }, { "epoch": 0.15955, "grad_norm": 1.0441497380391978, "learning_rate": 0.003, "loss": 4.0466, "step": 15955 }, { "epoch": 0.15956, "grad_norm": 1.3221945378646158, "learning_rate": 0.003, "loss": 4.0686, "step": 15956 }, { "epoch": 0.15957, "grad_norm": 0.9623396164481901, "learning_rate": 0.003, "loss": 4.0564, "step": 15957 }, { "epoch": 0.15958, "grad_norm": 1.1978286873707253, "learning_rate": 0.003, "loss": 4.0503, "step": 15958 }, { "epoch": 0.15959, "grad_norm": 1.2541150992104266, "learning_rate": 0.003, "loss": 4.0714, "step": 15959 }, { "epoch": 0.1596, "grad_norm": 1.4053468007862848, "learning_rate": 0.003, "loss": 4.064, "step": 15960 }, { "epoch": 0.15961, "grad_norm": 1.2022954055802446, "learning_rate": 0.003, "loss": 4.0532, "step": 15961 }, { "epoch": 0.15962, "grad_norm": 1.2899479012614354, "learning_rate": 0.003, "loss": 4.0668, "step": 15962 }, { "epoch": 0.15963, "grad_norm": 1.0104604306834757, "learning_rate": 0.003, "loss": 4.0541, "step": 15963 }, { "epoch": 0.15964, "grad_norm": 1.3701365986520497, "learning_rate": 0.003, "loss": 4.0379, "step": 15964 }, { "epoch": 0.15965, "grad_norm": 0.915145979140205, "learning_rate": 0.003, "loss": 4.0681, "step": 15965 }, { "epoch": 0.15966, "grad_norm": 1.3395650250329996, "learning_rate": 0.003, "loss": 4.0424, "step": 15966 }, { "epoch": 0.15967, "grad_norm": 1.2875086574703656, "learning_rate": 0.003, "loss": 4.0617, "step": 15967 }, { "epoch": 0.15968, "grad_norm": 1.2504696248888783, "learning_rate": 0.003, "loss": 4.068, "step": 15968 }, { "epoch": 0.15969, "grad_norm": 1.0215547112285952, "learning_rate": 0.003, "loss": 4.0508, "step": 15969 }, { "epoch": 0.1597, "grad_norm": 1.2703756402320694, "learning_rate": 0.003, "loss": 4.0824, "step": 15970 }, { "epoch": 0.15971, "grad_norm": 1.1870985294748964, "learning_rate": 0.003, "loss": 4.0645, "step": 15971 }, { "epoch": 0.15972, "grad_norm": 1.577281483064767, "learning_rate": 0.003, "loss": 4.0911, "step": 15972 }, { "epoch": 0.15973, "grad_norm": 1.1997563643841984, "learning_rate": 0.003, "loss": 4.059, "step": 15973 }, { "epoch": 0.15974, "grad_norm": 1.1111073471566701, "learning_rate": 0.003, "loss": 4.0757, "step": 15974 }, { "epoch": 0.15975, "grad_norm": 1.1634488039787199, "learning_rate": 0.003, "loss": 4.062, "step": 15975 }, { "epoch": 0.15976, "grad_norm": 1.2411269453094047, "learning_rate": 0.003, "loss": 4.0625, "step": 15976 }, { "epoch": 0.15977, "grad_norm": 1.461638922771171, "learning_rate": 0.003, "loss": 4.0736, "step": 15977 }, { "epoch": 0.15978, "grad_norm": 0.86336412630456, "learning_rate": 0.003, "loss": 4.0588, "step": 15978 }, { "epoch": 0.15979, "grad_norm": 1.1476716496753052, "learning_rate": 0.003, "loss": 4.0294, "step": 15979 }, { "epoch": 0.1598, "grad_norm": 1.189440127812191, "learning_rate": 0.003, "loss": 4.0485, "step": 15980 }, { "epoch": 0.15981, "grad_norm": 1.2509862236836928, "learning_rate": 0.003, "loss": 4.0513, "step": 15981 }, { "epoch": 0.15982, "grad_norm": 1.237805705201756, "learning_rate": 0.003, "loss": 4.0808, "step": 15982 }, { "epoch": 0.15983, "grad_norm": 1.3810342421390938, "learning_rate": 0.003, "loss": 4.0387, "step": 15983 }, { "epoch": 0.15984, "grad_norm": 1.1936090245544986, "learning_rate": 0.003, "loss": 4.0412, "step": 15984 }, { "epoch": 0.15985, "grad_norm": 1.5557605543282316, "learning_rate": 0.003, "loss": 4.0757, "step": 15985 }, { "epoch": 0.15986, "grad_norm": 1.290156158048075, "learning_rate": 0.003, "loss": 4.0439, "step": 15986 }, { "epoch": 0.15987, "grad_norm": 1.0745807071056974, "learning_rate": 0.003, "loss": 4.0519, "step": 15987 }, { "epoch": 0.15988, "grad_norm": 1.506036125111331, "learning_rate": 0.003, "loss": 4.0777, "step": 15988 }, { "epoch": 0.15989, "grad_norm": 0.9712888397562033, "learning_rate": 0.003, "loss": 4.0465, "step": 15989 }, { "epoch": 0.1599, "grad_norm": 1.4077697682259613, "learning_rate": 0.003, "loss": 4.0442, "step": 15990 }, { "epoch": 0.15991, "grad_norm": 1.156952447010478, "learning_rate": 0.003, "loss": 4.0803, "step": 15991 }, { "epoch": 0.15992, "grad_norm": 1.234014646961057, "learning_rate": 0.003, "loss": 4.0573, "step": 15992 }, { "epoch": 0.15993, "grad_norm": 0.9377390955202223, "learning_rate": 0.003, "loss": 4.0624, "step": 15993 }, { "epoch": 0.15994, "grad_norm": 1.5044058374437788, "learning_rate": 0.003, "loss": 4.0642, "step": 15994 }, { "epoch": 0.15995, "grad_norm": 1.047958226650085, "learning_rate": 0.003, "loss": 4.0459, "step": 15995 }, { "epoch": 0.15996, "grad_norm": 1.352017504920686, "learning_rate": 0.003, "loss": 4.0632, "step": 15996 }, { "epoch": 0.15997, "grad_norm": 1.02018421874671, "learning_rate": 0.003, "loss": 4.047, "step": 15997 }, { "epoch": 0.15998, "grad_norm": 1.3361840507251346, "learning_rate": 0.003, "loss": 4.0639, "step": 15998 }, { "epoch": 0.15999, "grad_norm": 1.123832468238917, "learning_rate": 0.003, "loss": 4.0609, "step": 15999 }, { "epoch": 0.16, "grad_norm": 1.2282973173546792, "learning_rate": 0.003, "loss": 4.0783, "step": 16000 }, { "epoch": 0.16001, "grad_norm": 1.185876019053862, "learning_rate": 0.003, "loss": 4.0579, "step": 16001 }, { "epoch": 0.16002, "grad_norm": 1.0161405965512897, "learning_rate": 0.003, "loss": 4.0487, "step": 16002 }, { "epoch": 0.16003, "grad_norm": 1.3867974768957554, "learning_rate": 0.003, "loss": 4.0947, "step": 16003 }, { "epoch": 0.16004, "grad_norm": 1.2610512502089228, "learning_rate": 0.003, "loss": 4.0548, "step": 16004 }, { "epoch": 0.16005, "grad_norm": 1.290496838186172, "learning_rate": 0.003, "loss": 4.0496, "step": 16005 }, { "epoch": 0.16006, "grad_norm": 1.1766003813701296, "learning_rate": 0.003, "loss": 4.0403, "step": 16006 }, { "epoch": 0.16007, "grad_norm": 1.3807222232910399, "learning_rate": 0.003, "loss": 4.0503, "step": 16007 }, { "epoch": 0.16008, "grad_norm": 1.0151941528455057, "learning_rate": 0.003, "loss": 4.0766, "step": 16008 }, { "epoch": 0.16009, "grad_norm": 1.3315177267931306, "learning_rate": 0.003, "loss": 4.0525, "step": 16009 }, { "epoch": 0.1601, "grad_norm": 0.9928579574156897, "learning_rate": 0.003, "loss": 4.0416, "step": 16010 }, { "epoch": 0.16011, "grad_norm": 1.3602863221481476, "learning_rate": 0.003, "loss": 4.0599, "step": 16011 }, { "epoch": 0.16012, "grad_norm": 1.0132912108523175, "learning_rate": 0.003, "loss": 4.0799, "step": 16012 }, { "epoch": 0.16013, "grad_norm": 1.5088117293451413, "learning_rate": 0.003, "loss": 4.0996, "step": 16013 }, { "epoch": 0.16014, "grad_norm": 1.1583704828044479, "learning_rate": 0.003, "loss": 4.0569, "step": 16014 }, { "epoch": 0.16015, "grad_norm": 1.2705045977034461, "learning_rate": 0.003, "loss": 4.0504, "step": 16015 }, { "epoch": 0.16016, "grad_norm": 1.317851900152992, "learning_rate": 0.003, "loss": 4.0448, "step": 16016 }, { "epoch": 0.16017, "grad_norm": 1.1853045034271175, "learning_rate": 0.003, "loss": 4.0729, "step": 16017 }, { "epoch": 0.16018, "grad_norm": 1.2364527640293232, "learning_rate": 0.003, "loss": 4.0461, "step": 16018 }, { "epoch": 0.16019, "grad_norm": 1.2285386520990271, "learning_rate": 0.003, "loss": 4.026, "step": 16019 }, { "epoch": 0.1602, "grad_norm": 0.8858904555018807, "learning_rate": 0.003, "loss": 4.056, "step": 16020 }, { "epoch": 0.16021, "grad_norm": 1.350921027596688, "learning_rate": 0.003, "loss": 4.0554, "step": 16021 }, { "epoch": 0.16022, "grad_norm": 1.4351101543673916, "learning_rate": 0.003, "loss": 4.0558, "step": 16022 }, { "epoch": 0.16023, "grad_norm": 1.255697509081209, "learning_rate": 0.003, "loss": 4.0553, "step": 16023 }, { "epoch": 0.16024, "grad_norm": 1.1824335549248934, "learning_rate": 0.003, "loss": 4.0704, "step": 16024 }, { "epoch": 0.16025, "grad_norm": 1.2662263515950418, "learning_rate": 0.003, "loss": 4.0665, "step": 16025 }, { "epoch": 0.16026, "grad_norm": 1.1399840534697918, "learning_rate": 0.003, "loss": 4.0299, "step": 16026 }, { "epoch": 0.16027, "grad_norm": 1.295445347906111, "learning_rate": 0.003, "loss": 4.0334, "step": 16027 }, { "epoch": 0.16028, "grad_norm": 1.0959290848383678, "learning_rate": 0.003, "loss": 4.0722, "step": 16028 }, { "epoch": 0.16029, "grad_norm": 1.2500054271602945, "learning_rate": 0.003, "loss": 4.0717, "step": 16029 }, { "epoch": 0.1603, "grad_norm": 0.7828254220254027, "learning_rate": 0.003, "loss": 4.0441, "step": 16030 }, { "epoch": 0.16031, "grad_norm": 0.9367431986280713, "learning_rate": 0.003, "loss": 4.0519, "step": 16031 }, { "epoch": 0.16032, "grad_norm": 1.1556516536007895, "learning_rate": 0.003, "loss": 4.0674, "step": 16032 }, { "epoch": 0.16033, "grad_norm": 1.208116076995649, "learning_rate": 0.003, "loss": 4.0543, "step": 16033 }, { "epoch": 0.16034, "grad_norm": 1.2607954671386226, "learning_rate": 0.003, "loss": 4.0566, "step": 16034 }, { "epoch": 0.16035, "grad_norm": 1.4102971953230792, "learning_rate": 0.003, "loss": 4.0707, "step": 16035 }, { "epoch": 0.16036, "grad_norm": 1.3184640225874744, "learning_rate": 0.003, "loss": 4.0465, "step": 16036 }, { "epoch": 0.16037, "grad_norm": 1.0168064820061287, "learning_rate": 0.003, "loss": 4.056, "step": 16037 }, { "epoch": 0.16038, "grad_norm": 1.3409307266038692, "learning_rate": 0.003, "loss": 4.0575, "step": 16038 }, { "epoch": 0.16039, "grad_norm": 1.2873512931259226, "learning_rate": 0.003, "loss": 4.043, "step": 16039 }, { "epoch": 0.1604, "grad_norm": 1.0602613953704065, "learning_rate": 0.003, "loss": 4.0598, "step": 16040 }, { "epoch": 0.16041, "grad_norm": 1.1722729803123337, "learning_rate": 0.003, "loss": 4.0485, "step": 16041 }, { "epoch": 0.16042, "grad_norm": 1.2144026204334712, "learning_rate": 0.003, "loss": 4.0764, "step": 16042 }, { "epoch": 0.16043, "grad_norm": 1.2670937938441484, "learning_rate": 0.003, "loss": 4.0552, "step": 16043 }, { "epoch": 0.16044, "grad_norm": 1.36102212031014, "learning_rate": 0.003, "loss": 4.0626, "step": 16044 }, { "epoch": 0.16045, "grad_norm": 1.313454575506855, "learning_rate": 0.003, "loss": 4.0624, "step": 16045 }, { "epoch": 0.16046, "grad_norm": 1.077842899825297, "learning_rate": 0.003, "loss": 4.054, "step": 16046 }, { "epoch": 0.16047, "grad_norm": 1.2856899362195513, "learning_rate": 0.003, "loss": 4.0662, "step": 16047 }, { "epoch": 0.16048, "grad_norm": 1.1620704440883474, "learning_rate": 0.003, "loss": 4.0553, "step": 16048 }, { "epoch": 0.16049, "grad_norm": 1.1791352016767631, "learning_rate": 0.003, "loss": 4.0563, "step": 16049 }, { "epoch": 0.1605, "grad_norm": 1.2276731565409649, "learning_rate": 0.003, "loss": 4.0455, "step": 16050 }, { "epoch": 0.16051, "grad_norm": 1.215460082915433, "learning_rate": 0.003, "loss": 4.0611, "step": 16051 }, { "epoch": 0.16052, "grad_norm": 1.3024479744349649, "learning_rate": 0.003, "loss": 4.0426, "step": 16052 }, { "epoch": 0.16053, "grad_norm": 1.0434738105117647, "learning_rate": 0.003, "loss": 4.0716, "step": 16053 }, { "epoch": 0.16054, "grad_norm": 1.4922296754524016, "learning_rate": 0.003, "loss": 4.0371, "step": 16054 }, { "epoch": 0.16055, "grad_norm": 1.1139652143755538, "learning_rate": 0.003, "loss": 4.0712, "step": 16055 }, { "epoch": 0.16056, "grad_norm": 1.0482729569435472, "learning_rate": 0.003, "loss": 4.0767, "step": 16056 }, { "epoch": 0.16057, "grad_norm": 1.6201160330504003, "learning_rate": 0.003, "loss": 4.0779, "step": 16057 }, { "epoch": 0.16058, "grad_norm": 1.1310803320642941, "learning_rate": 0.003, "loss": 4.0523, "step": 16058 }, { "epoch": 0.16059, "grad_norm": 1.6347875117596131, "learning_rate": 0.003, "loss": 4.0629, "step": 16059 }, { "epoch": 0.1606, "grad_norm": 1.2376701938294354, "learning_rate": 0.003, "loss": 4.0786, "step": 16060 }, { "epoch": 0.16061, "grad_norm": 0.9828200049395512, "learning_rate": 0.003, "loss": 4.0833, "step": 16061 }, { "epoch": 0.16062, "grad_norm": 1.2588410567839852, "learning_rate": 0.003, "loss": 4.0632, "step": 16062 }, { "epoch": 0.16063, "grad_norm": 1.3762150425011925, "learning_rate": 0.003, "loss": 4.0721, "step": 16063 }, { "epoch": 0.16064, "grad_norm": 1.246496347742565, "learning_rate": 0.003, "loss": 4.0854, "step": 16064 }, { "epoch": 0.16065, "grad_norm": 1.196117012172844, "learning_rate": 0.003, "loss": 4.0561, "step": 16065 }, { "epoch": 0.16066, "grad_norm": 1.4471735569176785, "learning_rate": 0.003, "loss": 4.0496, "step": 16066 }, { "epoch": 0.16067, "grad_norm": 0.8566399612016478, "learning_rate": 0.003, "loss": 4.0753, "step": 16067 }, { "epoch": 0.16068, "grad_norm": 1.072391558803118, "learning_rate": 0.003, "loss": 4.0406, "step": 16068 }, { "epoch": 0.16069, "grad_norm": 1.2635911866254046, "learning_rate": 0.003, "loss": 4.0717, "step": 16069 }, { "epoch": 0.1607, "grad_norm": 1.1597614951049424, "learning_rate": 0.003, "loss": 4.0644, "step": 16070 }, { "epoch": 0.16071, "grad_norm": 1.2302599861777737, "learning_rate": 0.003, "loss": 4.0321, "step": 16071 }, { "epoch": 0.16072, "grad_norm": 1.3428540526302917, "learning_rate": 0.003, "loss": 4.0825, "step": 16072 }, { "epoch": 0.16073, "grad_norm": 1.085160823232457, "learning_rate": 0.003, "loss": 4.0618, "step": 16073 }, { "epoch": 0.16074, "grad_norm": 1.351673584398568, "learning_rate": 0.003, "loss": 4.0593, "step": 16074 }, { "epoch": 0.16075, "grad_norm": 1.1102246000748848, "learning_rate": 0.003, "loss": 4.0776, "step": 16075 }, { "epoch": 0.16076, "grad_norm": 1.4375465430859315, "learning_rate": 0.003, "loss": 4.0877, "step": 16076 }, { "epoch": 0.16077, "grad_norm": 1.216029659979363, "learning_rate": 0.003, "loss": 4.0459, "step": 16077 }, { "epoch": 0.16078, "grad_norm": 1.170066835033616, "learning_rate": 0.003, "loss": 4.026, "step": 16078 }, { "epoch": 0.16079, "grad_norm": 1.1498334678374798, "learning_rate": 0.003, "loss": 4.0802, "step": 16079 }, { "epoch": 0.1608, "grad_norm": 1.0705153383453425, "learning_rate": 0.003, "loss": 4.0365, "step": 16080 }, { "epoch": 0.16081, "grad_norm": 1.4351203776646833, "learning_rate": 0.003, "loss": 4.0821, "step": 16081 }, { "epoch": 0.16082, "grad_norm": 1.1773355247461257, "learning_rate": 0.003, "loss": 4.0318, "step": 16082 }, { "epoch": 0.16083, "grad_norm": 1.3970482590289004, "learning_rate": 0.003, "loss": 4.0378, "step": 16083 }, { "epoch": 0.16084, "grad_norm": 1.0310870012110842, "learning_rate": 0.003, "loss": 4.0734, "step": 16084 }, { "epoch": 0.16085, "grad_norm": 1.4005190205008389, "learning_rate": 0.003, "loss": 4.0362, "step": 16085 }, { "epoch": 0.16086, "grad_norm": 1.297700056125098, "learning_rate": 0.003, "loss": 4.0324, "step": 16086 }, { "epoch": 0.16087, "grad_norm": 1.0731629299873047, "learning_rate": 0.003, "loss": 4.0454, "step": 16087 }, { "epoch": 0.16088, "grad_norm": 1.2658779579928179, "learning_rate": 0.003, "loss": 4.0528, "step": 16088 }, { "epoch": 0.16089, "grad_norm": 1.1329288764364365, "learning_rate": 0.003, "loss": 4.0446, "step": 16089 }, { "epoch": 0.1609, "grad_norm": 1.2426179867228542, "learning_rate": 0.003, "loss": 4.06, "step": 16090 }, { "epoch": 0.16091, "grad_norm": 1.0263552094840165, "learning_rate": 0.003, "loss": 4.0333, "step": 16091 }, { "epoch": 0.16092, "grad_norm": 1.2416680279212071, "learning_rate": 0.003, "loss": 4.0241, "step": 16092 }, { "epoch": 0.16093, "grad_norm": 1.260700386795223, "learning_rate": 0.003, "loss": 4.0237, "step": 16093 }, { "epoch": 0.16094, "grad_norm": 1.2875913173260094, "learning_rate": 0.003, "loss": 4.0586, "step": 16094 }, { "epoch": 0.16095, "grad_norm": 1.3656342469786085, "learning_rate": 0.003, "loss": 4.0387, "step": 16095 }, { "epoch": 0.16096, "grad_norm": 1.0897669652984088, "learning_rate": 0.003, "loss": 4.0588, "step": 16096 }, { "epoch": 0.16097, "grad_norm": 1.483569404996074, "learning_rate": 0.003, "loss": 4.0408, "step": 16097 }, { "epoch": 0.16098, "grad_norm": 0.9930638791381724, "learning_rate": 0.003, "loss": 4.0617, "step": 16098 }, { "epoch": 0.16099, "grad_norm": 1.332688523816265, "learning_rate": 0.003, "loss": 4.0659, "step": 16099 }, { "epoch": 0.161, "grad_norm": 1.1332123227431758, "learning_rate": 0.003, "loss": 4.0533, "step": 16100 }, { "epoch": 0.16101, "grad_norm": 1.4142004288182568, "learning_rate": 0.003, "loss": 4.065, "step": 16101 }, { "epoch": 0.16102, "grad_norm": 1.0554095342428538, "learning_rate": 0.003, "loss": 4.0177, "step": 16102 }, { "epoch": 0.16103, "grad_norm": 1.5078649726370579, "learning_rate": 0.003, "loss": 4.0484, "step": 16103 }, { "epoch": 0.16104, "grad_norm": 1.1804539195656147, "learning_rate": 0.003, "loss": 4.0584, "step": 16104 }, { "epoch": 0.16105, "grad_norm": 1.3659350489796371, "learning_rate": 0.003, "loss": 4.0542, "step": 16105 }, { "epoch": 0.16106, "grad_norm": 1.2005682397916537, "learning_rate": 0.003, "loss": 4.0581, "step": 16106 }, { "epoch": 0.16107, "grad_norm": 1.3341577103762357, "learning_rate": 0.003, "loss": 4.0634, "step": 16107 }, { "epoch": 0.16108, "grad_norm": 1.3639069812836295, "learning_rate": 0.003, "loss": 4.0819, "step": 16108 }, { "epoch": 0.16109, "grad_norm": 1.2025429222345108, "learning_rate": 0.003, "loss": 4.0682, "step": 16109 }, { "epoch": 0.1611, "grad_norm": 1.2097383388696323, "learning_rate": 0.003, "loss": 4.0591, "step": 16110 }, { "epoch": 0.16111, "grad_norm": 1.2458176475290417, "learning_rate": 0.003, "loss": 4.0615, "step": 16111 }, { "epoch": 0.16112, "grad_norm": 1.2368323118535012, "learning_rate": 0.003, "loss": 4.0666, "step": 16112 }, { "epoch": 0.16113, "grad_norm": 1.142566868935622, "learning_rate": 0.003, "loss": 4.0705, "step": 16113 }, { "epoch": 0.16114, "grad_norm": 1.1802710966943095, "learning_rate": 0.003, "loss": 4.052, "step": 16114 }, { "epoch": 0.16115, "grad_norm": 1.1055928491050242, "learning_rate": 0.003, "loss": 4.0617, "step": 16115 }, { "epoch": 0.16116, "grad_norm": 1.187695141379374, "learning_rate": 0.003, "loss": 4.0297, "step": 16116 }, { "epoch": 0.16117, "grad_norm": 1.031688892996054, "learning_rate": 0.003, "loss": 4.0688, "step": 16117 }, { "epoch": 0.16118, "grad_norm": 1.2575076087246992, "learning_rate": 0.003, "loss": 4.0479, "step": 16118 }, { "epoch": 0.16119, "grad_norm": 0.9999568372935328, "learning_rate": 0.003, "loss": 4.0574, "step": 16119 }, { "epoch": 0.1612, "grad_norm": 1.4193557198363458, "learning_rate": 0.003, "loss": 4.0599, "step": 16120 }, { "epoch": 0.16121, "grad_norm": 0.9899605942335881, "learning_rate": 0.003, "loss": 4.0673, "step": 16121 }, { "epoch": 0.16122, "grad_norm": 1.447924369250344, "learning_rate": 0.003, "loss": 4.0878, "step": 16122 }, { "epoch": 0.16123, "grad_norm": 1.0232117979767148, "learning_rate": 0.003, "loss": 4.0918, "step": 16123 }, { "epoch": 0.16124, "grad_norm": 1.4723857913910692, "learning_rate": 0.003, "loss": 4.0629, "step": 16124 }, { "epoch": 0.16125, "grad_norm": 1.0783329673441044, "learning_rate": 0.003, "loss": 4.0533, "step": 16125 }, { "epoch": 0.16126, "grad_norm": 1.2898787396854268, "learning_rate": 0.003, "loss": 4.06, "step": 16126 }, { "epoch": 0.16127, "grad_norm": 1.1780275482968812, "learning_rate": 0.003, "loss": 4.0722, "step": 16127 }, { "epoch": 0.16128, "grad_norm": 1.3930139902311514, "learning_rate": 0.003, "loss": 4.0766, "step": 16128 }, { "epoch": 0.16129, "grad_norm": 1.2315390005845666, "learning_rate": 0.003, "loss": 4.0782, "step": 16129 }, { "epoch": 0.1613, "grad_norm": 1.1571113291929978, "learning_rate": 0.003, "loss": 4.0807, "step": 16130 }, { "epoch": 0.16131, "grad_norm": 1.3384100128422542, "learning_rate": 0.003, "loss": 4.0604, "step": 16131 }, { "epoch": 0.16132, "grad_norm": 1.072585086800576, "learning_rate": 0.003, "loss": 4.0671, "step": 16132 }, { "epoch": 0.16133, "grad_norm": 1.2729336759623386, "learning_rate": 0.003, "loss": 4.0633, "step": 16133 }, { "epoch": 0.16134, "grad_norm": 1.114749540234282, "learning_rate": 0.003, "loss": 4.0647, "step": 16134 }, { "epoch": 0.16135, "grad_norm": 1.1175124107985823, "learning_rate": 0.003, "loss": 4.0696, "step": 16135 }, { "epoch": 0.16136, "grad_norm": 1.2246547602788969, "learning_rate": 0.003, "loss": 4.0483, "step": 16136 }, { "epoch": 0.16137, "grad_norm": 1.2857717063706393, "learning_rate": 0.003, "loss": 4.0587, "step": 16137 }, { "epoch": 0.16138, "grad_norm": 1.4942308360429648, "learning_rate": 0.003, "loss": 4.0886, "step": 16138 }, { "epoch": 0.16139, "grad_norm": 1.1375748258649176, "learning_rate": 0.003, "loss": 4.0769, "step": 16139 }, { "epoch": 0.1614, "grad_norm": 1.1923552179035142, "learning_rate": 0.003, "loss": 4.0624, "step": 16140 }, { "epoch": 0.16141, "grad_norm": 1.5189022469135416, "learning_rate": 0.003, "loss": 4.0532, "step": 16141 }, { "epoch": 0.16142, "grad_norm": 1.2905301683105843, "learning_rate": 0.003, "loss": 4.0482, "step": 16142 }, { "epoch": 0.16143, "grad_norm": 1.0586844543041596, "learning_rate": 0.003, "loss": 4.0504, "step": 16143 }, { "epoch": 0.16144, "grad_norm": 1.078529221858877, "learning_rate": 0.003, "loss": 4.0886, "step": 16144 }, { "epoch": 0.16145, "grad_norm": 1.2734398060290861, "learning_rate": 0.003, "loss": 4.0538, "step": 16145 }, { "epoch": 0.16146, "grad_norm": 1.3953782131238706, "learning_rate": 0.003, "loss": 4.0659, "step": 16146 }, { "epoch": 0.16147, "grad_norm": 1.1646308253134419, "learning_rate": 0.003, "loss": 4.076, "step": 16147 }, { "epoch": 0.16148, "grad_norm": 1.368083019000008, "learning_rate": 0.003, "loss": 4.0578, "step": 16148 }, { "epoch": 0.16149, "grad_norm": 1.240713218394061, "learning_rate": 0.003, "loss": 4.0478, "step": 16149 }, { "epoch": 0.1615, "grad_norm": 1.2105345191615164, "learning_rate": 0.003, "loss": 4.0792, "step": 16150 }, { "epoch": 0.16151, "grad_norm": 1.117951094917481, "learning_rate": 0.003, "loss": 4.064, "step": 16151 }, { "epoch": 0.16152, "grad_norm": 1.3586250303202676, "learning_rate": 0.003, "loss": 4.0613, "step": 16152 }, { "epoch": 0.16153, "grad_norm": 1.1797380254267351, "learning_rate": 0.003, "loss": 4.0577, "step": 16153 }, { "epoch": 0.16154, "grad_norm": 1.2663320152817172, "learning_rate": 0.003, "loss": 4.0951, "step": 16154 }, { "epoch": 0.16155, "grad_norm": 1.1101839680137688, "learning_rate": 0.003, "loss": 4.0535, "step": 16155 }, { "epoch": 0.16156, "grad_norm": 1.4659568094993767, "learning_rate": 0.003, "loss": 4.066, "step": 16156 }, { "epoch": 0.16157, "grad_norm": 1.1840417445148899, "learning_rate": 0.003, "loss": 4.0575, "step": 16157 }, { "epoch": 0.16158, "grad_norm": 1.2275331549066915, "learning_rate": 0.003, "loss": 4.0437, "step": 16158 }, { "epoch": 0.16159, "grad_norm": 1.2497562642871187, "learning_rate": 0.003, "loss": 4.0855, "step": 16159 }, { "epoch": 0.1616, "grad_norm": 1.3215853560109796, "learning_rate": 0.003, "loss": 4.0563, "step": 16160 }, { "epoch": 0.16161, "grad_norm": 1.1276735764177956, "learning_rate": 0.003, "loss": 4.062, "step": 16161 }, { "epoch": 0.16162, "grad_norm": 1.2316294802425705, "learning_rate": 0.003, "loss": 4.03, "step": 16162 }, { "epoch": 0.16163, "grad_norm": 0.9928474126809026, "learning_rate": 0.003, "loss": 4.0471, "step": 16163 }, { "epoch": 0.16164, "grad_norm": 1.2823445831755897, "learning_rate": 0.003, "loss": 4.0482, "step": 16164 }, { "epoch": 0.16165, "grad_norm": 1.079049904434384, "learning_rate": 0.003, "loss": 4.0512, "step": 16165 }, { "epoch": 0.16166, "grad_norm": 1.3013455832012046, "learning_rate": 0.003, "loss": 4.0765, "step": 16166 }, { "epoch": 0.16167, "grad_norm": 1.1097297541578093, "learning_rate": 0.003, "loss": 4.0646, "step": 16167 }, { "epoch": 0.16168, "grad_norm": 1.3489914890034207, "learning_rate": 0.003, "loss": 4.0636, "step": 16168 }, { "epoch": 0.16169, "grad_norm": 1.2096691901994352, "learning_rate": 0.003, "loss": 4.0374, "step": 16169 }, { "epoch": 0.1617, "grad_norm": 1.3065672763313205, "learning_rate": 0.003, "loss": 4.0271, "step": 16170 }, { "epoch": 0.16171, "grad_norm": 1.341880768268709, "learning_rate": 0.003, "loss": 4.0721, "step": 16171 }, { "epoch": 0.16172, "grad_norm": 1.2454783442636452, "learning_rate": 0.003, "loss": 4.0598, "step": 16172 }, { "epoch": 0.16173, "grad_norm": 1.2473409038756866, "learning_rate": 0.003, "loss": 4.0907, "step": 16173 }, { "epoch": 0.16174, "grad_norm": 1.1579721074325866, "learning_rate": 0.003, "loss": 4.0719, "step": 16174 }, { "epoch": 0.16175, "grad_norm": 1.0720426369036318, "learning_rate": 0.003, "loss": 4.0333, "step": 16175 }, { "epoch": 0.16176, "grad_norm": 1.2194859985899926, "learning_rate": 0.003, "loss": 4.0634, "step": 16176 }, { "epoch": 0.16177, "grad_norm": 1.1420179520039173, "learning_rate": 0.003, "loss": 4.0695, "step": 16177 }, { "epoch": 0.16178, "grad_norm": 1.142372388101929, "learning_rate": 0.003, "loss": 4.063, "step": 16178 }, { "epoch": 0.16179, "grad_norm": 1.305081858798519, "learning_rate": 0.003, "loss": 4.0504, "step": 16179 }, { "epoch": 0.1618, "grad_norm": 1.270635324379897, "learning_rate": 0.003, "loss": 4.0572, "step": 16180 }, { "epoch": 0.16181, "grad_norm": 1.2908644938680864, "learning_rate": 0.003, "loss": 4.0423, "step": 16181 }, { "epoch": 0.16182, "grad_norm": 1.0091203657975452, "learning_rate": 0.003, "loss": 4.0338, "step": 16182 }, { "epoch": 0.16183, "grad_norm": 1.5364701904545728, "learning_rate": 0.003, "loss": 4.0716, "step": 16183 }, { "epoch": 0.16184, "grad_norm": 1.0304836363577676, "learning_rate": 0.003, "loss": 4.0436, "step": 16184 }, { "epoch": 0.16185, "grad_norm": 1.4540775782228723, "learning_rate": 0.003, "loss": 4.0958, "step": 16185 }, { "epoch": 0.16186, "grad_norm": 0.9617748173512339, "learning_rate": 0.003, "loss": 4.045, "step": 16186 }, { "epoch": 0.16187, "grad_norm": 1.217977899288468, "learning_rate": 0.003, "loss": 4.0814, "step": 16187 }, { "epoch": 0.16188, "grad_norm": 1.487051711688237, "learning_rate": 0.003, "loss": 4.0876, "step": 16188 }, { "epoch": 0.16189, "grad_norm": 1.17910209302436, "learning_rate": 0.003, "loss": 4.0778, "step": 16189 }, { "epoch": 0.1619, "grad_norm": 1.1446373184625196, "learning_rate": 0.003, "loss": 4.025, "step": 16190 }, { "epoch": 0.16191, "grad_norm": 1.284875588091632, "learning_rate": 0.003, "loss": 4.0675, "step": 16191 }, { "epoch": 0.16192, "grad_norm": 1.313551319883505, "learning_rate": 0.003, "loss": 4.0582, "step": 16192 }, { "epoch": 0.16193, "grad_norm": 1.2333812069105783, "learning_rate": 0.003, "loss": 4.0685, "step": 16193 }, { "epoch": 0.16194, "grad_norm": 1.2048765218001227, "learning_rate": 0.003, "loss": 4.0476, "step": 16194 }, { "epoch": 0.16195, "grad_norm": 1.1685474837899368, "learning_rate": 0.003, "loss": 4.053, "step": 16195 }, { "epoch": 0.16196, "grad_norm": 1.234379565341423, "learning_rate": 0.003, "loss": 4.0577, "step": 16196 }, { "epoch": 0.16197, "grad_norm": 1.2416619629065064, "learning_rate": 0.003, "loss": 4.0775, "step": 16197 }, { "epoch": 0.16198, "grad_norm": 1.1506895479509611, "learning_rate": 0.003, "loss": 4.0538, "step": 16198 }, { "epoch": 0.16199, "grad_norm": 1.3069220528527965, "learning_rate": 0.003, "loss": 4.0601, "step": 16199 }, { "epoch": 0.162, "grad_norm": 1.2141660935638423, "learning_rate": 0.003, "loss": 4.0837, "step": 16200 }, { "epoch": 0.16201, "grad_norm": 1.277566042717582, "learning_rate": 0.003, "loss": 4.0727, "step": 16201 }, { "epoch": 0.16202, "grad_norm": 1.1887268494189691, "learning_rate": 0.003, "loss": 4.0599, "step": 16202 }, { "epoch": 0.16203, "grad_norm": 1.2395698830375579, "learning_rate": 0.003, "loss": 4.0455, "step": 16203 }, { "epoch": 0.16204, "grad_norm": 1.2176320361343518, "learning_rate": 0.003, "loss": 4.0867, "step": 16204 }, { "epoch": 0.16205, "grad_norm": 1.4108530624303173, "learning_rate": 0.003, "loss": 4.0633, "step": 16205 }, { "epoch": 0.16206, "grad_norm": 1.134903228670893, "learning_rate": 0.003, "loss": 4.0526, "step": 16206 }, { "epoch": 0.16207, "grad_norm": 1.2099066685612638, "learning_rate": 0.003, "loss": 4.0685, "step": 16207 }, { "epoch": 0.16208, "grad_norm": 1.2046684455575636, "learning_rate": 0.003, "loss": 4.0378, "step": 16208 }, { "epoch": 0.16209, "grad_norm": 1.20432078452343, "learning_rate": 0.003, "loss": 4.0675, "step": 16209 }, { "epoch": 0.1621, "grad_norm": 1.1205535134930835, "learning_rate": 0.003, "loss": 4.0567, "step": 16210 }, { "epoch": 0.16211, "grad_norm": 1.3164407472701114, "learning_rate": 0.003, "loss": 4.055, "step": 16211 }, { "epoch": 0.16212, "grad_norm": 0.9504364598306422, "learning_rate": 0.003, "loss": 4.0558, "step": 16212 }, { "epoch": 0.16213, "grad_norm": 1.4074031635387623, "learning_rate": 0.003, "loss": 4.0548, "step": 16213 }, { "epoch": 0.16214, "grad_norm": 1.00809562486534, "learning_rate": 0.003, "loss": 4.0799, "step": 16214 }, { "epoch": 0.16215, "grad_norm": 1.334635201219315, "learning_rate": 0.003, "loss": 4.0911, "step": 16215 }, { "epoch": 0.16216, "grad_norm": 1.1974054616067982, "learning_rate": 0.003, "loss": 4.0438, "step": 16216 }, { "epoch": 0.16217, "grad_norm": 1.1746897452771488, "learning_rate": 0.003, "loss": 4.0782, "step": 16217 }, { "epoch": 0.16218, "grad_norm": 1.2765648358837007, "learning_rate": 0.003, "loss": 4.0417, "step": 16218 }, { "epoch": 0.16219, "grad_norm": 1.004880048598137, "learning_rate": 0.003, "loss": 4.0523, "step": 16219 }, { "epoch": 0.1622, "grad_norm": 1.4440937736237733, "learning_rate": 0.003, "loss": 4.0487, "step": 16220 }, { "epoch": 0.16221, "grad_norm": 1.2105566172203854, "learning_rate": 0.003, "loss": 4.0646, "step": 16221 }, { "epoch": 0.16222, "grad_norm": 1.3202740298730096, "learning_rate": 0.003, "loss": 4.0564, "step": 16222 }, { "epoch": 0.16223, "grad_norm": 1.280287335011057, "learning_rate": 0.003, "loss": 4.0818, "step": 16223 }, { "epoch": 0.16224, "grad_norm": 1.2458770506199293, "learning_rate": 0.003, "loss": 4.0509, "step": 16224 }, { "epoch": 0.16225, "grad_norm": 1.362888596559668, "learning_rate": 0.003, "loss": 4.0428, "step": 16225 }, { "epoch": 0.16226, "grad_norm": 1.1366893412204746, "learning_rate": 0.003, "loss": 4.0507, "step": 16226 }, { "epoch": 0.16227, "grad_norm": 1.4216958867902325, "learning_rate": 0.003, "loss": 4.0565, "step": 16227 }, { "epoch": 0.16228, "grad_norm": 1.381805161860225, "learning_rate": 0.003, "loss": 4.0621, "step": 16228 }, { "epoch": 0.16229, "grad_norm": 1.1295034159836423, "learning_rate": 0.003, "loss": 4.0753, "step": 16229 }, { "epoch": 0.1623, "grad_norm": 1.0784970122984692, "learning_rate": 0.003, "loss": 4.0294, "step": 16230 }, { "epoch": 0.16231, "grad_norm": 1.1336871653186584, "learning_rate": 0.003, "loss": 4.0571, "step": 16231 }, { "epoch": 0.16232, "grad_norm": 1.2501064912218984, "learning_rate": 0.003, "loss": 4.0582, "step": 16232 }, { "epoch": 0.16233, "grad_norm": 1.1649618958076693, "learning_rate": 0.003, "loss": 4.0375, "step": 16233 }, { "epoch": 0.16234, "grad_norm": 1.0368502437332756, "learning_rate": 0.003, "loss": 4.0653, "step": 16234 }, { "epoch": 0.16235, "grad_norm": 1.3480478678997219, "learning_rate": 0.003, "loss": 4.0767, "step": 16235 }, { "epoch": 0.16236, "grad_norm": 1.1367294472111962, "learning_rate": 0.003, "loss": 4.0523, "step": 16236 }, { "epoch": 0.16237, "grad_norm": 1.0659105492586416, "learning_rate": 0.003, "loss": 4.0804, "step": 16237 }, { "epoch": 0.16238, "grad_norm": 1.3748269521555407, "learning_rate": 0.003, "loss": 4.061, "step": 16238 }, { "epoch": 0.16239, "grad_norm": 1.3121789858444732, "learning_rate": 0.003, "loss": 4.0825, "step": 16239 }, { "epoch": 0.1624, "grad_norm": 1.4056845341506228, "learning_rate": 0.003, "loss": 4.0548, "step": 16240 }, { "epoch": 0.16241, "grad_norm": 1.1679735202194832, "learning_rate": 0.003, "loss": 4.0905, "step": 16241 }, { "epoch": 0.16242, "grad_norm": 1.319184163343352, "learning_rate": 0.003, "loss": 4.0622, "step": 16242 }, { "epoch": 0.16243, "grad_norm": 1.2125087661596101, "learning_rate": 0.003, "loss": 4.0727, "step": 16243 }, { "epoch": 0.16244, "grad_norm": 1.2719103951861044, "learning_rate": 0.003, "loss": 4.0267, "step": 16244 }, { "epoch": 0.16245, "grad_norm": 1.1984346997620228, "learning_rate": 0.003, "loss": 4.0429, "step": 16245 }, { "epoch": 0.16246, "grad_norm": 1.1340434776667687, "learning_rate": 0.003, "loss": 4.0548, "step": 16246 }, { "epoch": 0.16247, "grad_norm": 1.1140869732254772, "learning_rate": 0.003, "loss": 4.0548, "step": 16247 }, { "epoch": 0.16248, "grad_norm": 1.227453917661384, "learning_rate": 0.003, "loss": 4.0463, "step": 16248 }, { "epoch": 0.16249, "grad_norm": 1.1925920328393311, "learning_rate": 0.003, "loss": 4.0629, "step": 16249 }, { "epoch": 0.1625, "grad_norm": 1.1399032518800811, "learning_rate": 0.003, "loss": 4.035, "step": 16250 }, { "epoch": 0.16251, "grad_norm": 1.2797498532161469, "learning_rate": 0.003, "loss": 4.0546, "step": 16251 }, { "epoch": 0.16252, "grad_norm": 1.019199769972669, "learning_rate": 0.003, "loss": 4.0491, "step": 16252 }, { "epoch": 0.16253, "grad_norm": 1.452625321554583, "learning_rate": 0.003, "loss": 4.0747, "step": 16253 }, { "epoch": 0.16254, "grad_norm": 1.144178826389817, "learning_rate": 0.003, "loss": 4.0296, "step": 16254 }, { "epoch": 0.16255, "grad_norm": 1.4030724132011527, "learning_rate": 0.003, "loss": 4.0498, "step": 16255 }, { "epoch": 0.16256, "grad_norm": 1.3017431523623153, "learning_rate": 0.003, "loss": 4.0634, "step": 16256 }, { "epoch": 0.16257, "grad_norm": 1.208476282338132, "learning_rate": 0.003, "loss": 4.0513, "step": 16257 }, { "epoch": 0.16258, "grad_norm": 1.3275210983008279, "learning_rate": 0.003, "loss": 4.0655, "step": 16258 }, { "epoch": 0.16259, "grad_norm": 1.1927129114231794, "learning_rate": 0.003, "loss": 4.0576, "step": 16259 }, { "epoch": 0.1626, "grad_norm": 1.184898032853809, "learning_rate": 0.003, "loss": 4.0543, "step": 16260 }, { "epoch": 0.16261, "grad_norm": 1.3221817149951827, "learning_rate": 0.003, "loss": 4.0316, "step": 16261 }, { "epoch": 0.16262, "grad_norm": 1.0597447397778326, "learning_rate": 0.003, "loss": 4.0675, "step": 16262 }, { "epoch": 0.16263, "grad_norm": 1.5015000416488335, "learning_rate": 0.003, "loss": 4.0627, "step": 16263 }, { "epoch": 0.16264, "grad_norm": 0.9484849035292638, "learning_rate": 0.003, "loss": 4.0626, "step": 16264 }, { "epoch": 0.16265, "grad_norm": 1.2497923166040579, "learning_rate": 0.003, "loss": 4.0529, "step": 16265 }, { "epoch": 0.16266, "grad_norm": 1.052695750134391, "learning_rate": 0.003, "loss": 4.0619, "step": 16266 }, { "epoch": 0.16267, "grad_norm": 1.323958844792137, "learning_rate": 0.003, "loss": 4.0524, "step": 16267 }, { "epoch": 0.16268, "grad_norm": 1.067533482038374, "learning_rate": 0.003, "loss": 4.0728, "step": 16268 }, { "epoch": 0.16269, "grad_norm": 1.3651789367884446, "learning_rate": 0.003, "loss": 4.0722, "step": 16269 }, { "epoch": 0.1627, "grad_norm": 1.2776129239950722, "learning_rate": 0.003, "loss": 4.078, "step": 16270 }, { "epoch": 0.16271, "grad_norm": 1.6778241915864978, "learning_rate": 0.003, "loss": 4.0758, "step": 16271 }, { "epoch": 0.16272, "grad_norm": 0.9700957980425218, "learning_rate": 0.003, "loss": 4.0433, "step": 16272 }, { "epoch": 0.16273, "grad_norm": 1.2646988723773898, "learning_rate": 0.003, "loss": 4.0587, "step": 16273 }, { "epoch": 0.16274, "grad_norm": 1.4089292011783563, "learning_rate": 0.003, "loss": 4.0586, "step": 16274 }, { "epoch": 0.16275, "grad_norm": 1.166694981145315, "learning_rate": 0.003, "loss": 4.0963, "step": 16275 }, { "epoch": 0.16276, "grad_norm": 1.108640007705579, "learning_rate": 0.003, "loss": 4.0407, "step": 16276 }, { "epoch": 0.16277, "grad_norm": 1.202577317559944, "learning_rate": 0.003, "loss": 4.0637, "step": 16277 }, { "epoch": 0.16278, "grad_norm": 1.0624388101346522, "learning_rate": 0.003, "loss": 4.0527, "step": 16278 }, { "epoch": 0.16279, "grad_norm": 1.5170265623031196, "learning_rate": 0.003, "loss": 4.0641, "step": 16279 }, { "epoch": 0.1628, "grad_norm": 1.0856494741830944, "learning_rate": 0.003, "loss": 4.073, "step": 16280 }, { "epoch": 0.16281, "grad_norm": 1.3976086613417265, "learning_rate": 0.003, "loss": 4.0554, "step": 16281 }, { "epoch": 0.16282, "grad_norm": 1.0250708449718593, "learning_rate": 0.003, "loss": 4.0403, "step": 16282 }, { "epoch": 0.16283, "grad_norm": 1.4883504341428777, "learning_rate": 0.003, "loss": 4.0291, "step": 16283 }, { "epoch": 0.16284, "grad_norm": 1.1823939648448438, "learning_rate": 0.003, "loss": 4.0472, "step": 16284 }, { "epoch": 0.16285, "grad_norm": 1.1090593980612389, "learning_rate": 0.003, "loss": 4.0689, "step": 16285 }, { "epoch": 0.16286, "grad_norm": 1.255754121545388, "learning_rate": 0.003, "loss": 4.0804, "step": 16286 }, { "epoch": 0.16287, "grad_norm": 1.132016724713699, "learning_rate": 0.003, "loss": 4.0649, "step": 16287 }, { "epoch": 0.16288, "grad_norm": 1.272722794652534, "learning_rate": 0.003, "loss": 4.0669, "step": 16288 }, { "epoch": 0.16289, "grad_norm": 1.3094349363350137, "learning_rate": 0.003, "loss": 4.0695, "step": 16289 }, { "epoch": 0.1629, "grad_norm": 1.2304301370076913, "learning_rate": 0.003, "loss": 4.0538, "step": 16290 }, { "epoch": 0.16291, "grad_norm": 1.3643880545394098, "learning_rate": 0.003, "loss": 4.0624, "step": 16291 }, { "epoch": 0.16292, "grad_norm": 1.0740680933845081, "learning_rate": 0.003, "loss": 4.0634, "step": 16292 }, { "epoch": 0.16293, "grad_norm": 1.3050220676600632, "learning_rate": 0.003, "loss": 4.0589, "step": 16293 }, { "epoch": 0.16294, "grad_norm": 1.0184695674557405, "learning_rate": 0.003, "loss": 4.039, "step": 16294 }, { "epoch": 0.16295, "grad_norm": 1.6244685049589114, "learning_rate": 0.003, "loss": 4.0616, "step": 16295 }, { "epoch": 0.16296, "grad_norm": 0.9898495972924228, "learning_rate": 0.003, "loss": 4.0762, "step": 16296 }, { "epoch": 0.16297, "grad_norm": 1.373468676826956, "learning_rate": 0.003, "loss": 4.0682, "step": 16297 }, { "epoch": 0.16298, "grad_norm": 1.0752429483583343, "learning_rate": 0.003, "loss": 4.0485, "step": 16298 }, { "epoch": 0.16299, "grad_norm": 1.1489424900245766, "learning_rate": 0.003, "loss": 4.0774, "step": 16299 }, { "epoch": 0.163, "grad_norm": 1.3025797423455028, "learning_rate": 0.003, "loss": 4.0763, "step": 16300 }, { "epoch": 0.16301, "grad_norm": 1.0566881385453588, "learning_rate": 0.003, "loss": 4.0286, "step": 16301 }, { "epoch": 0.16302, "grad_norm": 1.2898357916465581, "learning_rate": 0.003, "loss": 4.0739, "step": 16302 }, { "epoch": 0.16303, "grad_norm": 1.1588859621825356, "learning_rate": 0.003, "loss": 4.0715, "step": 16303 }, { "epoch": 0.16304, "grad_norm": 1.1608499531827632, "learning_rate": 0.003, "loss": 4.0635, "step": 16304 }, { "epoch": 0.16305, "grad_norm": 1.123357233759964, "learning_rate": 0.003, "loss": 4.0693, "step": 16305 }, { "epoch": 0.16306, "grad_norm": 1.4211787825524893, "learning_rate": 0.003, "loss": 4.0621, "step": 16306 }, { "epoch": 0.16307, "grad_norm": 1.4222413627859347, "learning_rate": 0.003, "loss": 4.0521, "step": 16307 }, { "epoch": 0.16308, "grad_norm": 1.4241146551019983, "learning_rate": 0.003, "loss": 4.047, "step": 16308 }, { "epoch": 0.16309, "grad_norm": 1.2873547417279407, "learning_rate": 0.003, "loss": 4.0611, "step": 16309 }, { "epoch": 0.1631, "grad_norm": 1.0979961313056252, "learning_rate": 0.003, "loss": 4.0674, "step": 16310 }, { "epoch": 0.16311, "grad_norm": 1.5158301454154433, "learning_rate": 0.003, "loss": 4.0677, "step": 16311 }, { "epoch": 0.16312, "grad_norm": 0.9623413202787349, "learning_rate": 0.003, "loss": 4.1004, "step": 16312 }, { "epoch": 0.16313, "grad_norm": 1.1579853235171516, "learning_rate": 0.003, "loss": 4.0517, "step": 16313 }, { "epoch": 0.16314, "grad_norm": 1.2370693069383596, "learning_rate": 0.003, "loss": 4.0601, "step": 16314 }, { "epoch": 0.16315, "grad_norm": 1.223963284605398, "learning_rate": 0.003, "loss": 4.0331, "step": 16315 }, { "epoch": 0.16316, "grad_norm": 1.2131062801443062, "learning_rate": 0.003, "loss": 4.0616, "step": 16316 }, { "epoch": 0.16317, "grad_norm": 0.9333364883918731, "learning_rate": 0.003, "loss": 4.0602, "step": 16317 }, { "epoch": 0.16318, "grad_norm": 1.2545637003766508, "learning_rate": 0.003, "loss": 4.0665, "step": 16318 }, { "epoch": 0.16319, "grad_norm": 1.1637564320008495, "learning_rate": 0.003, "loss": 4.0638, "step": 16319 }, { "epoch": 0.1632, "grad_norm": 1.3930824654232326, "learning_rate": 0.003, "loss": 4.0668, "step": 16320 }, { "epoch": 0.16321, "grad_norm": 1.2701909486548193, "learning_rate": 0.003, "loss": 4.0727, "step": 16321 }, { "epoch": 0.16322, "grad_norm": 1.1504704692579468, "learning_rate": 0.003, "loss": 4.0655, "step": 16322 }, { "epoch": 0.16323, "grad_norm": 1.2619124017098002, "learning_rate": 0.003, "loss": 4.0475, "step": 16323 }, { "epoch": 0.16324, "grad_norm": 1.6371242880230603, "learning_rate": 0.003, "loss": 4.0808, "step": 16324 }, { "epoch": 0.16325, "grad_norm": 1.2092492938774504, "learning_rate": 0.003, "loss": 4.0434, "step": 16325 }, { "epoch": 0.16326, "grad_norm": 1.4449491267608436, "learning_rate": 0.003, "loss": 4.0787, "step": 16326 }, { "epoch": 0.16327, "grad_norm": 1.2168703017267488, "learning_rate": 0.003, "loss": 4.0312, "step": 16327 }, { "epoch": 0.16328, "grad_norm": 1.353767460624721, "learning_rate": 0.003, "loss": 4.0823, "step": 16328 }, { "epoch": 0.16329, "grad_norm": 1.0766855812631326, "learning_rate": 0.003, "loss": 4.0766, "step": 16329 }, { "epoch": 0.1633, "grad_norm": 1.2225130893227987, "learning_rate": 0.003, "loss": 4.0683, "step": 16330 }, { "epoch": 0.16331, "grad_norm": 1.2415697408617543, "learning_rate": 0.003, "loss": 4.0643, "step": 16331 }, { "epoch": 0.16332, "grad_norm": 1.14481606629091, "learning_rate": 0.003, "loss": 4.062, "step": 16332 }, { "epoch": 0.16333, "grad_norm": 1.1365437547684951, "learning_rate": 0.003, "loss": 4.0535, "step": 16333 }, { "epoch": 0.16334, "grad_norm": 1.2608703109317938, "learning_rate": 0.003, "loss": 4.0775, "step": 16334 }, { "epoch": 0.16335, "grad_norm": 1.1967000746733578, "learning_rate": 0.003, "loss": 4.0903, "step": 16335 }, { "epoch": 0.16336, "grad_norm": 1.1930161262170862, "learning_rate": 0.003, "loss": 4.0802, "step": 16336 }, { "epoch": 0.16337, "grad_norm": 1.2430520582370888, "learning_rate": 0.003, "loss": 4.0662, "step": 16337 }, { "epoch": 0.16338, "grad_norm": 1.3995785930958728, "learning_rate": 0.003, "loss": 4.0761, "step": 16338 }, { "epoch": 0.16339, "grad_norm": 1.2550047638086808, "learning_rate": 0.003, "loss": 4.0671, "step": 16339 }, { "epoch": 0.1634, "grad_norm": 1.0181579197625517, "learning_rate": 0.003, "loss": 4.0485, "step": 16340 }, { "epoch": 0.16341, "grad_norm": 1.1941714205548857, "learning_rate": 0.003, "loss": 4.0348, "step": 16341 }, { "epoch": 0.16342, "grad_norm": 1.1161593175141173, "learning_rate": 0.003, "loss": 4.0456, "step": 16342 }, { "epoch": 0.16343, "grad_norm": 1.3615843906357468, "learning_rate": 0.003, "loss": 4.047, "step": 16343 }, { "epoch": 0.16344, "grad_norm": 1.017842175059377, "learning_rate": 0.003, "loss": 4.0365, "step": 16344 }, { "epoch": 0.16345, "grad_norm": 1.1738558883336643, "learning_rate": 0.003, "loss": 4.0681, "step": 16345 }, { "epoch": 0.16346, "grad_norm": 1.2112578199939257, "learning_rate": 0.003, "loss": 4.0628, "step": 16346 }, { "epoch": 0.16347, "grad_norm": 1.2394809936950661, "learning_rate": 0.003, "loss": 4.082, "step": 16347 }, { "epoch": 0.16348, "grad_norm": 1.475630322606032, "learning_rate": 0.003, "loss": 4.08, "step": 16348 }, { "epoch": 0.16349, "grad_norm": 1.0248623737151932, "learning_rate": 0.003, "loss": 4.0734, "step": 16349 }, { "epoch": 0.1635, "grad_norm": 1.4091242271820457, "learning_rate": 0.003, "loss": 4.05, "step": 16350 }, { "epoch": 0.16351, "grad_norm": 1.1368378857737884, "learning_rate": 0.003, "loss": 4.0572, "step": 16351 }, { "epoch": 0.16352, "grad_norm": 1.1872056869253942, "learning_rate": 0.003, "loss": 4.0289, "step": 16352 }, { "epoch": 0.16353, "grad_norm": 1.1185431794190572, "learning_rate": 0.003, "loss": 4.0302, "step": 16353 }, { "epoch": 0.16354, "grad_norm": 1.137493645130181, "learning_rate": 0.003, "loss": 4.0903, "step": 16354 }, { "epoch": 0.16355, "grad_norm": 1.2358146820289326, "learning_rate": 0.003, "loss": 4.0859, "step": 16355 }, { "epoch": 0.16356, "grad_norm": 1.2315889636488526, "learning_rate": 0.003, "loss": 4.0336, "step": 16356 }, { "epoch": 0.16357, "grad_norm": 1.2802968208976975, "learning_rate": 0.003, "loss": 4.0648, "step": 16357 }, { "epoch": 0.16358, "grad_norm": 1.260811899864562, "learning_rate": 0.003, "loss": 4.0805, "step": 16358 }, { "epoch": 0.16359, "grad_norm": 1.2121622704122177, "learning_rate": 0.003, "loss": 4.0683, "step": 16359 }, { "epoch": 0.1636, "grad_norm": 1.2357583365276035, "learning_rate": 0.003, "loss": 4.0721, "step": 16360 }, { "epoch": 0.16361, "grad_norm": 1.377904643878386, "learning_rate": 0.003, "loss": 4.0417, "step": 16361 }, { "epoch": 0.16362, "grad_norm": 0.9684549664648887, "learning_rate": 0.003, "loss": 4.0797, "step": 16362 }, { "epoch": 0.16363, "grad_norm": 1.4179259020766957, "learning_rate": 0.003, "loss": 4.0611, "step": 16363 }, { "epoch": 0.16364, "grad_norm": 1.1206211379617204, "learning_rate": 0.003, "loss": 4.0765, "step": 16364 }, { "epoch": 0.16365, "grad_norm": 1.279809554655433, "learning_rate": 0.003, "loss": 4.0695, "step": 16365 }, { "epoch": 0.16366, "grad_norm": 1.2594313513466797, "learning_rate": 0.003, "loss": 4.0682, "step": 16366 }, { "epoch": 0.16367, "grad_norm": 1.225775077542689, "learning_rate": 0.003, "loss": 4.0408, "step": 16367 }, { "epoch": 0.16368, "grad_norm": 1.3272212082797992, "learning_rate": 0.003, "loss": 4.0655, "step": 16368 }, { "epoch": 0.16369, "grad_norm": 1.2976282058453512, "learning_rate": 0.003, "loss": 4.0693, "step": 16369 }, { "epoch": 0.1637, "grad_norm": 1.234289289259054, "learning_rate": 0.003, "loss": 4.0467, "step": 16370 }, { "epoch": 0.16371, "grad_norm": 1.3518875427870496, "learning_rate": 0.003, "loss": 4.0852, "step": 16371 }, { "epoch": 0.16372, "grad_norm": 1.1901678123588724, "learning_rate": 0.003, "loss": 4.0587, "step": 16372 }, { "epoch": 0.16373, "grad_norm": 1.1922646263251482, "learning_rate": 0.003, "loss": 4.0403, "step": 16373 }, { "epoch": 0.16374, "grad_norm": 1.118598784078167, "learning_rate": 0.003, "loss": 4.0481, "step": 16374 }, { "epoch": 0.16375, "grad_norm": 1.3563234908303308, "learning_rate": 0.003, "loss": 4.0717, "step": 16375 }, { "epoch": 0.16376, "grad_norm": 0.9629005182541244, "learning_rate": 0.003, "loss": 4.0474, "step": 16376 }, { "epoch": 0.16377, "grad_norm": 1.3759361678038555, "learning_rate": 0.003, "loss": 4.0686, "step": 16377 }, { "epoch": 0.16378, "grad_norm": 1.1391749543532, "learning_rate": 0.003, "loss": 4.0378, "step": 16378 }, { "epoch": 0.16379, "grad_norm": 1.3822263931996204, "learning_rate": 0.003, "loss": 4.0562, "step": 16379 }, { "epoch": 0.1638, "grad_norm": 1.1339954320059182, "learning_rate": 0.003, "loss": 4.0584, "step": 16380 }, { "epoch": 0.16381, "grad_norm": 1.3466622107327346, "learning_rate": 0.003, "loss": 4.0578, "step": 16381 }, { "epoch": 0.16382, "grad_norm": 1.1597951508934399, "learning_rate": 0.003, "loss": 4.0681, "step": 16382 }, { "epoch": 0.16383, "grad_norm": 1.2302866378170052, "learning_rate": 0.003, "loss": 4.0626, "step": 16383 }, { "epoch": 0.16384, "grad_norm": 1.2515905665860958, "learning_rate": 0.003, "loss": 4.0603, "step": 16384 }, { "epoch": 0.16385, "grad_norm": 1.0422015271915603, "learning_rate": 0.003, "loss": 4.0663, "step": 16385 }, { "epoch": 0.16386, "grad_norm": 1.29635466700889, "learning_rate": 0.003, "loss": 4.0594, "step": 16386 }, { "epoch": 0.16387, "grad_norm": 1.0875135908370575, "learning_rate": 0.003, "loss": 4.055, "step": 16387 }, { "epoch": 0.16388, "grad_norm": 1.2718309619067318, "learning_rate": 0.003, "loss": 4.0579, "step": 16388 }, { "epoch": 0.16389, "grad_norm": 1.0022806628307235, "learning_rate": 0.003, "loss": 4.0589, "step": 16389 }, { "epoch": 0.1639, "grad_norm": 1.4790777363163723, "learning_rate": 0.003, "loss": 4.0892, "step": 16390 }, { "epoch": 0.16391, "grad_norm": 1.14263807272635, "learning_rate": 0.003, "loss": 4.071, "step": 16391 }, { "epoch": 0.16392, "grad_norm": 1.1865540155604621, "learning_rate": 0.003, "loss": 4.0369, "step": 16392 }, { "epoch": 0.16393, "grad_norm": 1.415860201754588, "learning_rate": 0.003, "loss": 4.0498, "step": 16393 }, { "epoch": 0.16394, "grad_norm": 1.3899506802540258, "learning_rate": 0.003, "loss": 4.0968, "step": 16394 }, { "epoch": 0.16395, "grad_norm": 1.5882227448719515, "learning_rate": 0.003, "loss": 4.0644, "step": 16395 }, { "epoch": 0.16396, "grad_norm": 1.08999481689989, "learning_rate": 0.003, "loss": 4.0421, "step": 16396 }, { "epoch": 0.16397, "grad_norm": 1.2994181147414126, "learning_rate": 0.003, "loss": 4.069, "step": 16397 }, { "epoch": 0.16398, "grad_norm": 1.1958925573152632, "learning_rate": 0.003, "loss": 4.0667, "step": 16398 }, { "epoch": 0.16399, "grad_norm": 1.086394744003385, "learning_rate": 0.003, "loss": 4.0558, "step": 16399 }, { "epoch": 0.164, "grad_norm": 1.2930250221392108, "learning_rate": 0.003, "loss": 4.0777, "step": 16400 }, { "epoch": 0.16401, "grad_norm": 0.8959460745072638, "learning_rate": 0.003, "loss": 4.0431, "step": 16401 }, { "epoch": 0.16402, "grad_norm": 1.227152163682272, "learning_rate": 0.003, "loss": 4.0991, "step": 16402 }, { "epoch": 0.16403, "grad_norm": 1.2293512458095264, "learning_rate": 0.003, "loss": 4.0537, "step": 16403 }, { "epoch": 0.16404, "grad_norm": 1.0173681014886449, "learning_rate": 0.003, "loss": 4.0513, "step": 16404 }, { "epoch": 0.16405, "grad_norm": 1.3262938317259634, "learning_rate": 0.003, "loss": 4.0663, "step": 16405 }, { "epoch": 0.16406, "grad_norm": 1.054899121048987, "learning_rate": 0.003, "loss": 4.0601, "step": 16406 }, { "epoch": 0.16407, "grad_norm": 1.4160598366644352, "learning_rate": 0.003, "loss": 4.0951, "step": 16407 }, { "epoch": 0.16408, "grad_norm": 0.9675264790932663, "learning_rate": 0.003, "loss": 4.0564, "step": 16408 }, { "epoch": 0.16409, "grad_norm": 1.1201656449148025, "learning_rate": 0.003, "loss": 4.0357, "step": 16409 }, { "epoch": 0.1641, "grad_norm": 1.41925141094191, "learning_rate": 0.003, "loss": 4.0655, "step": 16410 }, { "epoch": 0.16411, "grad_norm": 1.466931189921125, "learning_rate": 0.003, "loss": 4.0574, "step": 16411 }, { "epoch": 0.16412, "grad_norm": 1.1504773378613948, "learning_rate": 0.003, "loss": 4.0583, "step": 16412 }, { "epoch": 0.16413, "grad_norm": 1.362644924490457, "learning_rate": 0.003, "loss": 4.068, "step": 16413 }, { "epoch": 0.16414, "grad_norm": 1.0361407075459195, "learning_rate": 0.003, "loss": 4.0338, "step": 16414 }, { "epoch": 0.16415, "grad_norm": 1.3555931613478198, "learning_rate": 0.003, "loss": 4.055, "step": 16415 }, { "epoch": 0.16416, "grad_norm": 1.231546061828751, "learning_rate": 0.003, "loss": 4.0679, "step": 16416 }, { "epoch": 0.16417, "grad_norm": 1.1195223025704868, "learning_rate": 0.003, "loss": 4.0521, "step": 16417 }, { "epoch": 0.16418, "grad_norm": 1.3278966187876984, "learning_rate": 0.003, "loss": 4.0657, "step": 16418 }, { "epoch": 0.16419, "grad_norm": 1.1440588893808925, "learning_rate": 0.003, "loss": 4.0725, "step": 16419 }, { "epoch": 0.1642, "grad_norm": 1.4434601081654999, "learning_rate": 0.003, "loss": 4.1053, "step": 16420 }, { "epoch": 0.16421, "grad_norm": 1.0226496351857584, "learning_rate": 0.003, "loss": 4.0535, "step": 16421 }, { "epoch": 0.16422, "grad_norm": 1.5841365599632116, "learning_rate": 0.003, "loss": 4.0383, "step": 16422 }, { "epoch": 0.16423, "grad_norm": 1.1434420576338018, "learning_rate": 0.003, "loss": 4.0378, "step": 16423 }, { "epoch": 0.16424, "grad_norm": 1.2985223382883153, "learning_rate": 0.003, "loss": 4.0647, "step": 16424 }, { "epoch": 0.16425, "grad_norm": 1.132505631613156, "learning_rate": 0.003, "loss": 4.0535, "step": 16425 }, { "epoch": 0.16426, "grad_norm": 1.3534589432652555, "learning_rate": 0.003, "loss": 4.0887, "step": 16426 }, { "epoch": 0.16427, "grad_norm": 1.4277535109802608, "learning_rate": 0.003, "loss": 4.0503, "step": 16427 }, { "epoch": 0.16428, "grad_norm": 1.196476803073257, "learning_rate": 0.003, "loss": 4.0662, "step": 16428 }, { "epoch": 0.16429, "grad_norm": 1.1345684075559352, "learning_rate": 0.003, "loss": 4.0577, "step": 16429 }, { "epoch": 0.1643, "grad_norm": 1.4574862318885673, "learning_rate": 0.003, "loss": 4.0594, "step": 16430 }, { "epoch": 0.16431, "grad_norm": 1.0817848202585982, "learning_rate": 0.003, "loss": 4.0623, "step": 16431 }, { "epoch": 0.16432, "grad_norm": 1.3027331901469046, "learning_rate": 0.003, "loss": 4.0777, "step": 16432 }, { "epoch": 0.16433, "grad_norm": 1.179901742427836, "learning_rate": 0.003, "loss": 4.0757, "step": 16433 }, { "epoch": 0.16434, "grad_norm": 1.1242233806228457, "learning_rate": 0.003, "loss": 4.0577, "step": 16434 }, { "epoch": 0.16435, "grad_norm": 1.1178056065486728, "learning_rate": 0.003, "loss": 4.0348, "step": 16435 }, { "epoch": 0.16436, "grad_norm": 1.198810463998239, "learning_rate": 0.003, "loss": 4.0606, "step": 16436 }, { "epoch": 0.16437, "grad_norm": 1.0265981605669419, "learning_rate": 0.003, "loss": 4.0776, "step": 16437 }, { "epoch": 0.16438, "grad_norm": 1.3442598281984819, "learning_rate": 0.003, "loss": 4.0588, "step": 16438 }, { "epoch": 0.16439, "grad_norm": 1.1147868203594866, "learning_rate": 0.003, "loss": 4.0607, "step": 16439 }, { "epoch": 0.1644, "grad_norm": 1.2091652208444614, "learning_rate": 0.003, "loss": 4.0493, "step": 16440 }, { "epoch": 0.16441, "grad_norm": 1.2902464501166047, "learning_rate": 0.003, "loss": 4.0635, "step": 16441 }, { "epoch": 0.16442, "grad_norm": 1.1058223615753715, "learning_rate": 0.003, "loss": 4.0303, "step": 16442 }, { "epoch": 0.16443, "grad_norm": 1.4162825495350395, "learning_rate": 0.003, "loss": 4.0909, "step": 16443 }, { "epoch": 0.16444, "grad_norm": 0.9704567342020411, "learning_rate": 0.003, "loss": 4.0692, "step": 16444 }, { "epoch": 0.16445, "grad_norm": 1.2929890493890461, "learning_rate": 0.003, "loss": 4.0626, "step": 16445 }, { "epoch": 0.16446, "grad_norm": 1.312806546807894, "learning_rate": 0.003, "loss": 4.0497, "step": 16446 }, { "epoch": 0.16447, "grad_norm": 1.3934869343729224, "learning_rate": 0.003, "loss": 4.0786, "step": 16447 }, { "epoch": 0.16448, "grad_norm": 1.238306320379035, "learning_rate": 0.003, "loss": 4.077, "step": 16448 }, { "epoch": 0.16449, "grad_norm": 1.046299047424111, "learning_rate": 0.003, "loss": 4.0411, "step": 16449 }, { "epoch": 0.1645, "grad_norm": 1.37940166459787, "learning_rate": 0.003, "loss": 4.0622, "step": 16450 }, { "epoch": 0.16451, "grad_norm": 0.9769709791788973, "learning_rate": 0.003, "loss": 4.0616, "step": 16451 }, { "epoch": 0.16452, "grad_norm": 1.4713445311775446, "learning_rate": 0.003, "loss": 4.0471, "step": 16452 }, { "epoch": 0.16453, "grad_norm": 1.0333409989710807, "learning_rate": 0.003, "loss": 4.0771, "step": 16453 }, { "epoch": 0.16454, "grad_norm": 1.275234893906959, "learning_rate": 0.003, "loss": 4.0857, "step": 16454 }, { "epoch": 0.16455, "grad_norm": 1.2147269810274521, "learning_rate": 0.003, "loss": 4.0874, "step": 16455 }, { "epoch": 0.16456, "grad_norm": 1.3637218724665405, "learning_rate": 0.003, "loss": 4.0759, "step": 16456 }, { "epoch": 0.16457, "grad_norm": 1.104648426622214, "learning_rate": 0.003, "loss": 4.0846, "step": 16457 }, { "epoch": 0.16458, "grad_norm": 1.2502549309947466, "learning_rate": 0.003, "loss": 4.0521, "step": 16458 }, { "epoch": 0.16459, "grad_norm": 1.1989193222086119, "learning_rate": 0.003, "loss": 4.0596, "step": 16459 }, { "epoch": 0.1646, "grad_norm": 1.1567167697347025, "learning_rate": 0.003, "loss": 4.0695, "step": 16460 }, { "epoch": 0.16461, "grad_norm": 1.2334558387358596, "learning_rate": 0.003, "loss": 4.0403, "step": 16461 }, { "epoch": 0.16462, "grad_norm": 1.2588965894970137, "learning_rate": 0.003, "loss": 4.0466, "step": 16462 }, { "epoch": 0.16463, "grad_norm": 1.0711040285980924, "learning_rate": 0.003, "loss": 4.0567, "step": 16463 }, { "epoch": 0.16464, "grad_norm": 1.3324142770374714, "learning_rate": 0.003, "loss": 4.044, "step": 16464 }, { "epoch": 0.16465, "grad_norm": 1.2924020103371354, "learning_rate": 0.003, "loss": 4.0727, "step": 16465 }, { "epoch": 0.16466, "grad_norm": 1.2271686092882501, "learning_rate": 0.003, "loss": 4.0465, "step": 16466 }, { "epoch": 0.16467, "grad_norm": 1.1391827547004902, "learning_rate": 0.003, "loss": 4.0635, "step": 16467 }, { "epoch": 0.16468, "grad_norm": 1.2727746840360479, "learning_rate": 0.003, "loss": 4.0519, "step": 16468 }, { "epoch": 0.16469, "grad_norm": 1.1221544359042348, "learning_rate": 0.003, "loss": 4.0739, "step": 16469 }, { "epoch": 0.1647, "grad_norm": 1.308102129076781, "learning_rate": 0.003, "loss": 4.0598, "step": 16470 }, { "epoch": 0.16471, "grad_norm": 1.3310089253793593, "learning_rate": 0.003, "loss": 4.0603, "step": 16471 }, { "epoch": 0.16472, "grad_norm": 1.099131548971726, "learning_rate": 0.003, "loss": 4.0782, "step": 16472 }, { "epoch": 0.16473, "grad_norm": 1.1507369134549297, "learning_rate": 0.003, "loss": 4.074, "step": 16473 }, { "epoch": 0.16474, "grad_norm": 1.1696433055407094, "learning_rate": 0.003, "loss": 4.0647, "step": 16474 }, { "epoch": 0.16475, "grad_norm": 1.185818918775687, "learning_rate": 0.003, "loss": 4.08, "step": 16475 }, { "epoch": 0.16476, "grad_norm": 1.0511754226018224, "learning_rate": 0.003, "loss": 4.0628, "step": 16476 }, { "epoch": 0.16477, "grad_norm": 1.342190845266813, "learning_rate": 0.003, "loss": 4.0679, "step": 16477 }, { "epoch": 0.16478, "grad_norm": 1.396898197039214, "learning_rate": 0.003, "loss": 4.0907, "step": 16478 }, { "epoch": 0.16479, "grad_norm": 1.0029197019363287, "learning_rate": 0.003, "loss": 4.054, "step": 16479 }, { "epoch": 0.1648, "grad_norm": 1.3474363421958357, "learning_rate": 0.003, "loss": 4.0399, "step": 16480 }, { "epoch": 0.16481, "grad_norm": 1.203370400537817, "learning_rate": 0.003, "loss": 4.0621, "step": 16481 }, { "epoch": 0.16482, "grad_norm": 1.3496925563552822, "learning_rate": 0.003, "loss": 4.0487, "step": 16482 }, { "epoch": 0.16483, "grad_norm": 1.140456352470189, "learning_rate": 0.003, "loss": 4.0373, "step": 16483 }, { "epoch": 0.16484, "grad_norm": 1.161661118843948, "learning_rate": 0.003, "loss": 4.0459, "step": 16484 }, { "epoch": 0.16485, "grad_norm": 1.2702972031072666, "learning_rate": 0.003, "loss": 4.0704, "step": 16485 }, { "epoch": 0.16486, "grad_norm": 1.214327740218, "learning_rate": 0.003, "loss": 4.0372, "step": 16486 }, { "epoch": 0.16487, "grad_norm": 1.4686085233106276, "learning_rate": 0.003, "loss": 4.0688, "step": 16487 }, { "epoch": 0.16488, "grad_norm": 1.1129341580421621, "learning_rate": 0.003, "loss": 4.086, "step": 16488 }, { "epoch": 0.16489, "grad_norm": 1.4217588281003763, "learning_rate": 0.003, "loss": 4.0696, "step": 16489 }, { "epoch": 0.1649, "grad_norm": 1.0703469821845155, "learning_rate": 0.003, "loss": 4.0639, "step": 16490 }, { "epoch": 0.16491, "grad_norm": 1.265477567719717, "learning_rate": 0.003, "loss": 4.0388, "step": 16491 }, { "epoch": 0.16492, "grad_norm": 1.0621408450657313, "learning_rate": 0.003, "loss": 4.0446, "step": 16492 }, { "epoch": 0.16493, "grad_norm": 1.3972589457929456, "learning_rate": 0.003, "loss": 4.0778, "step": 16493 }, { "epoch": 0.16494, "grad_norm": 1.2650926087062098, "learning_rate": 0.003, "loss": 4.0487, "step": 16494 }, { "epoch": 0.16495, "grad_norm": 1.2978586043422418, "learning_rate": 0.003, "loss": 4.0912, "step": 16495 }, { "epoch": 0.16496, "grad_norm": 1.261472305227728, "learning_rate": 0.003, "loss": 4.0493, "step": 16496 }, { "epoch": 0.16497, "grad_norm": 1.242683064248222, "learning_rate": 0.003, "loss": 4.0211, "step": 16497 }, { "epoch": 0.16498, "grad_norm": 1.2036963193234782, "learning_rate": 0.003, "loss": 4.0108, "step": 16498 }, { "epoch": 0.16499, "grad_norm": 1.3948241551460334, "learning_rate": 0.003, "loss": 4.0717, "step": 16499 }, { "epoch": 0.165, "grad_norm": 1.07581359154424, "learning_rate": 0.003, "loss": 4.0389, "step": 16500 }, { "epoch": 0.16501, "grad_norm": 1.5118180487772566, "learning_rate": 0.003, "loss": 4.074, "step": 16501 }, { "epoch": 0.16502, "grad_norm": 1.261475590974746, "learning_rate": 0.003, "loss": 4.0632, "step": 16502 }, { "epoch": 0.16503, "grad_norm": 1.0255140844117148, "learning_rate": 0.003, "loss": 4.0435, "step": 16503 }, { "epoch": 0.16504, "grad_norm": 1.3639751753138154, "learning_rate": 0.003, "loss": 4.0758, "step": 16504 }, { "epoch": 0.16505, "grad_norm": 1.202677925288568, "learning_rate": 0.003, "loss": 4.0685, "step": 16505 }, { "epoch": 0.16506, "grad_norm": 1.2378749997616612, "learning_rate": 0.003, "loss": 4.0796, "step": 16506 }, { "epoch": 0.16507, "grad_norm": 1.145146155706816, "learning_rate": 0.003, "loss": 4.0624, "step": 16507 }, { "epoch": 0.16508, "grad_norm": 1.2535730286186197, "learning_rate": 0.003, "loss": 4.0454, "step": 16508 }, { "epoch": 0.16509, "grad_norm": 1.1298757008410567, "learning_rate": 0.003, "loss": 4.0376, "step": 16509 }, { "epoch": 0.1651, "grad_norm": 1.038127001875258, "learning_rate": 0.003, "loss": 4.0334, "step": 16510 }, { "epoch": 0.16511, "grad_norm": 1.2620275203759714, "learning_rate": 0.003, "loss": 4.0588, "step": 16511 }, { "epoch": 0.16512, "grad_norm": 1.1271879091413812, "learning_rate": 0.003, "loss": 4.0708, "step": 16512 }, { "epoch": 0.16513, "grad_norm": 1.3324389570308686, "learning_rate": 0.003, "loss": 4.0606, "step": 16513 }, { "epoch": 0.16514, "grad_norm": 1.0241500602776423, "learning_rate": 0.003, "loss": 4.0358, "step": 16514 }, { "epoch": 0.16515, "grad_norm": 1.2954261214680667, "learning_rate": 0.003, "loss": 4.0599, "step": 16515 }, { "epoch": 0.16516, "grad_norm": 1.224561218138641, "learning_rate": 0.003, "loss": 4.0444, "step": 16516 }, { "epoch": 0.16517, "grad_norm": 1.4856014637916959, "learning_rate": 0.003, "loss": 4.0526, "step": 16517 }, { "epoch": 0.16518, "grad_norm": 1.3083785465492175, "learning_rate": 0.003, "loss": 4.0518, "step": 16518 }, { "epoch": 0.16519, "grad_norm": 1.1754221139693608, "learning_rate": 0.003, "loss": 4.0685, "step": 16519 }, { "epoch": 0.1652, "grad_norm": 1.383733782221718, "learning_rate": 0.003, "loss": 4.0388, "step": 16520 }, { "epoch": 0.16521, "grad_norm": 0.9224264945042131, "learning_rate": 0.003, "loss": 4.0463, "step": 16521 }, { "epoch": 0.16522, "grad_norm": 1.271845436140457, "learning_rate": 0.003, "loss": 4.0428, "step": 16522 }, { "epoch": 0.16523, "grad_norm": 1.2684184328179449, "learning_rate": 0.003, "loss": 4.0758, "step": 16523 }, { "epoch": 0.16524, "grad_norm": 1.2957251132249745, "learning_rate": 0.003, "loss": 4.04, "step": 16524 }, { "epoch": 0.16525, "grad_norm": 1.0759420818376795, "learning_rate": 0.003, "loss": 4.0428, "step": 16525 }, { "epoch": 0.16526, "grad_norm": 1.3680200743833646, "learning_rate": 0.003, "loss": 4.0598, "step": 16526 }, { "epoch": 0.16527, "grad_norm": 1.3403723906917697, "learning_rate": 0.003, "loss": 4.081, "step": 16527 }, { "epoch": 0.16528, "grad_norm": 1.08893530152959, "learning_rate": 0.003, "loss": 4.0295, "step": 16528 }, { "epoch": 0.16529, "grad_norm": 1.3497970090602143, "learning_rate": 0.003, "loss": 4.0663, "step": 16529 }, { "epoch": 0.1653, "grad_norm": 1.0818049101829756, "learning_rate": 0.003, "loss": 4.0511, "step": 16530 }, { "epoch": 0.16531, "grad_norm": 1.4560504817993078, "learning_rate": 0.003, "loss": 4.0615, "step": 16531 }, { "epoch": 0.16532, "grad_norm": 1.2019258840381728, "learning_rate": 0.003, "loss": 4.0616, "step": 16532 }, { "epoch": 0.16533, "grad_norm": 1.2671708842962617, "learning_rate": 0.003, "loss": 4.0544, "step": 16533 }, { "epoch": 0.16534, "grad_norm": 1.1933625613664025, "learning_rate": 0.003, "loss": 4.0729, "step": 16534 }, { "epoch": 0.16535, "grad_norm": 1.0915886188920134, "learning_rate": 0.003, "loss": 4.0525, "step": 16535 }, { "epoch": 0.16536, "grad_norm": 1.2468246178670828, "learning_rate": 0.003, "loss": 4.054, "step": 16536 }, { "epoch": 0.16537, "grad_norm": 1.0420479871195687, "learning_rate": 0.003, "loss": 4.0704, "step": 16537 }, { "epoch": 0.16538, "grad_norm": 1.4846408040185886, "learning_rate": 0.003, "loss": 4.0685, "step": 16538 }, { "epoch": 0.16539, "grad_norm": 1.0010507915854987, "learning_rate": 0.003, "loss": 4.0335, "step": 16539 }, { "epoch": 0.1654, "grad_norm": 1.3841859663118798, "learning_rate": 0.003, "loss": 4.0565, "step": 16540 }, { "epoch": 0.16541, "grad_norm": 0.9995675238909607, "learning_rate": 0.003, "loss": 4.0558, "step": 16541 }, { "epoch": 0.16542, "grad_norm": 1.4135590670613585, "learning_rate": 0.003, "loss": 4.0639, "step": 16542 }, { "epoch": 0.16543, "grad_norm": 1.271032768090048, "learning_rate": 0.003, "loss": 4.0582, "step": 16543 }, { "epoch": 0.16544, "grad_norm": 1.0003598220576706, "learning_rate": 0.003, "loss": 4.0323, "step": 16544 }, { "epoch": 0.16545, "grad_norm": 1.4438228510842843, "learning_rate": 0.003, "loss": 4.0769, "step": 16545 }, { "epoch": 0.16546, "grad_norm": 1.079194189276615, "learning_rate": 0.003, "loss": 4.0505, "step": 16546 }, { "epoch": 0.16547, "grad_norm": 1.3462259069028204, "learning_rate": 0.003, "loss": 4.0683, "step": 16547 }, { "epoch": 0.16548, "grad_norm": 1.0604888414253373, "learning_rate": 0.003, "loss": 4.0441, "step": 16548 }, { "epoch": 0.16549, "grad_norm": 1.4315815511781533, "learning_rate": 0.003, "loss": 4.0534, "step": 16549 }, { "epoch": 0.1655, "grad_norm": 1.279012800053747, "learning_rate": 0.003, "loss": 4.0535, "step": 16550 }, { "epoch": 0.16551, "grad_norm": 1.3147197009267735, "learning_rate": 0.003, "loss": 4.0598, "step": 16551 }, { "epoch": 0.16552, "grad_norm": 1.3008364923026423, "learning_rate": 0.003, "loss": 4.0814, "step": 16552 }, { "epoch": 0.16553, "grad_norm": 1.0376654244687642, "learning_rate": 0.003, "loss": 4.068, "step": 16553 }, { "epoch": 0.16554, "grad_norm": 1.3682159327472294, "learning_rate": 0.003, "loss": 4.0531, "step": 16554 }, { "epoch": 0.16555, "grad_norm": 1.287686464491986, "learning_rate": 0.003, "loss": 4.0645, "step": 16555 }, { "epoch": 0.16556, "grad_norm": 1.0984911002136095, "learning_rate": 0.003, "loss": 4.0515, "step": 16556 }, { "epoch": 0.16557, "grad_norm": 1.4013179191300427, "learning_rate": 0.003, "loss": 4.0641, "step": 16557 }, { "epoch": 0.16558, "grad_norm": 1.2721433708285537, "learning_rate": 0.003, "loss": 4.064, "step": 16558 }, { "epoch": 0.16559, "grad_norm": 1.2072626658190684, "learning_rate": 0.003, "loss": 4.0741, "step": 16559 }, { "epoch": 0.1656, "grad_norm": 1.2627153176527717, "learning_rate": 0.003, "loss": 4.0708, "step": 16560 }, { "epoch": 0.16561, "grad_norm": 1.3613505028934798, "learning_rate": 0.003, "loss": 4.0852, "step": 16561 }, { "epoch": 0.16562, "grad_norm": 1.3669212213985202, "learning_rate": 0.003, "loss": 4.076, "step": 16562 }, { "epoch": 0.16563, "grad_norm": 1.1351984881474968, "learning_rate": 0.003, "loss": 4.0563, "step": 16563 }, { "epoch": 0.16564, "grad_norm": 1.245779541760801, "learning_rate": 0.003, "loss": 4.0239, "step": 16564 }, { "epoch": 0.16565, "grad_norm": 1.2270767489407162, "learning_rate": 0.003, "loss": 4.0637, "step": 16565 }, { "epoch": 0.16566, "grad_norm": 1.1773840818283083, "learning_rate": 0.003, "loss": 4.0847, "step": 16566 }, { "epoch": 0.16567, "grad_norm": 1.1912047515356559, "learning_rate": 0.003, "loss": 4.0389, "step": 16567 }, { "epoch": 0.16568, "grad_norm": 1.123840748354702, "learning_rate": 0.003, "loss": 4.028, "step": 16568 }, { "epoch": 0.16569, "grad_norm": 1.1655040178671072, "learning_rate": 0.003, "loss": 4.0565, "step": 16569 }, { "epoch": 0.1657, "grad_norm": 1.3526719963038978, "learning_rate": 0.003, "loss": 4.0566, "step": 16570 }, { "epoch": 0.16571, "grad_norm": 1.1875691743114833, "learning_rate": 0.003, "loss": 4.0409, "step": 16571 }, { "epoch": 0.16572, "grad_norm": 1.2235636981375877, "learning_rate": 0.003, "loss": 4.0679, "step": 16572 }, { "epoch": 0.16573, "grad_norm": 1.2977232871300532, "learning_rate": 0.003, "loss": 4.0669, "step": 16573 }, { "epoch": 0.16574, "grad_norm": 1.2268435006440799, "learning_rate": 0.003, "loss": 4.0312, "step": 16574 }, { "epoch": 0.16575, "grad_norm": 1.158616634691927, "learning_rate": 0.003, "loss": 4.0841, "step": 16575 }, { "epoch": 0.16576, "grad_norm": 1.2259858847843446, "learning_rate": 0.003, "loss": 4.0591, "step": 16576 }, { "epoch": 0.16577, "grad_norm": 1.109482144570921, "learning_rate": 0.003, "loss": 4.0589, "step": 16577 }, { "epoch": 0.16578, "grad_norm": 1.2766512889347732, "learning_rate": 0.003, "loss": 4.0603, "step": 16578 }, { "epoch": 0.16579, "grad_norm": 1.0916250750966277, "learning_rate": 0.003, "loss": 4.0396, "step": 16579 }, { "epoch": 0.1658, "grad_norm": 1.3784453360171725, "learning_rate": 0.003, "loss": 4.0587, "step": 16580 }, { "epoch": 0.16581, "grad_norm": 1.3106198650486158, "learning_rate": 0.003, "loss": 4.0484, "step": 16581 }, { "epoch": 0.16582, "grad_norm": 1.1538738897341496, "learning_rate": 0.003, "loss": 4.0532, "step": 16582 }, { "epoch": 0.16583, "grad_norm": 1.289280959030002, "learning_rate": 0.003, "loss": 4.0699, "step": 16583 }, { "epoch": 0.16584, "grad_norm": 1.1419639947253375, "learning_rate": 0.003, "loss": 4.0588, "step": 16584 }, { "epoch": 0.16585, "grad_norm": 1.1909428219339122, "learning_rate": 0.003, "loss": 4.0316, "step": 16585 }, { "epoch": 0.16586, "grad_norm": 1.233121475694777, "learning_rate": 0.003, "loss": 4.0689, "step": 16586 }, { "epoch": 0.16587, "grad_norm": 1.281209517667142, "learning_rate": 0.003, "loss": 4.0688, "step": 16587 }, { "epoch": 0.16588, "grad_norm": 1.307368701030308, "learning_rate": 0.003, "loss": 4.0378, "step": 16588 }, { "epoch": 0.16589, "grad_norm": 1.0210678628509378, "learning_rate": 0.003, "loss": 4.0566, "step": 16589 }, { "epoch": 0.1659, "grad_norm": 1.50612650511528, "learning_rate": 0.003, "loss": 4.0459, "step": 16590 }, { "epoch": 0.16591, "grad_norm": 0.9982600987829718, "learning_rate": 0.003, "loss": 4.0905, "step": 16591 }, { "epoch": 0.16592, "grad_norm": 1.4759984528929282, "learning_rate": 0.003, "loss": 4.0701, "step": 16592 }, { "epoch": 0.16593, "grad_norm": 1.2897250493468178, "learning_rate": 0.003, "loss": 4.0672, "step": 16593 }, { "epoch": 0.16594, "grad_norm": 1.5208296147653992, "learning_rate": 0.003, "loss": 4.0504, "step": 16594 }, { "epoch": 0.16595, "grad_norm": 1.0583249028775887, "learning_rate": 0.003, "loss": 4.0812, "step": 16595 }, { "epoch": 0.16596, "grad_norm": 1.2899447233941335, "learning_rate": 0.003, "loss": 4.0579, "step": 16596 }, { "epoch": 0.16597, "grad_norm": 1.0402907849884682, "learning_rate": 0.003, "loss": 4.055, "step": 16597 }, { "epoch": 0.16598, "grad_norm": 1.2329279897475582, "learning_rate": 0.003, "loss": 4.0226, "step": 16598 }, { "epoch": 0.16599, "grad_norm": 1.2710444937587715, "learning_rate": 0.003, "loss": 4.0873, "step": 16599 }, { "epoch": 0.166, "grad_norm": 1.0563254861048152, "learning_rate": 0.003, "loss": 4.0198, "step": 16600 }, { "epoch": 0.16601, "grad_norm": 1.2066109622890218, "learning_rate": 0.003, "loss": 4.0933, "step": 16601 }, { "epoch": 0.16602, "grad_norm": 1.2329930980591854, "learning_rate": 0.003, "loss": 4.0486, "step": 16602 }, { "epoch": 0.16603, "grad_norm": 1.2790475248579785, "learning_rate": 0.003, "loss": 4.0957, "step": 16603 }, { "epoch": 0.16604, "grad_norm": 1.225644885990106, "learning_rate": 0.003, "loss": 4.0537, "step": 16604 }, { "epoch": 0.16605, "grad_norm": 1.244570324311943, "learning_rate": 0.003, "loss": 4.072, "step": 16605 }, { "epoch": 0.16606, "grad_norm": 1.524337993014525, "learning_rate": 0.003, "loss": 4.0686, "step": 16606 }, { "epoch": 0.16607, "grad_norm": 1.0802970479558118, "learning_rate": 0.003, "loss": 4.0508, "step": 16607 }, { "epoch": 0.16608, "grad_norm": 1.479143935511754, "learning_rate": 0.003, "loss": 4.0613, "step": 16608 }, { "epoch": 0.16609, "grad_norm": 1.0502722454074231, "learning_rate": 0.003, "loss": 4.0562, "step": 16609 }, { "epoch": 0.1661, "grad_norm": 1.09474150083467, "learning_rate": 0.003, "loss": 4.0287, "step": 16610 }, { "epoch": 0.16611, "grad_norm": 1.2960369195096926, "learning_rate": 0.003, "loss": 4.073, "step": 16611 }, { "epoch": 0.16612, "grad_norm": 1.1297372361576241, "learning_rate": 0.003, "loss": 4.0517, "step": 16612 }, { "epoch": 0.16613, "grad_norm": 1.4419374093664417, "learning_rate": 0.003, "loss": 4.0497, "step": 16613 }, { "epoch": 0.16614, "grad_norm": 0.9300048151347583, "learning_rate": 0.003, "loss": 4.0709, "step": 16614 }, { "epoch": 0.16615, "grad_norm": 1.2192941217633135, "learning_rate": 0.003, "loss": 4.0257, "step": 16615 }, { "epoch": 0.16616, "grad_norm": 1.3430237358940136, "learning_rate": 0.003, "loss": 4.1127, "step": 16616 }, { "epoch": 0.16617, "grad_norm": 1.1587947661119633, "learning_rate": 0.003, "loss": 4.0372, "step": 16617 }, { "epoch": 0.16618, "grad_norm": 1.0769665339153838, "learning_rate": 0.003, "loss": 4.06, "step": 16618 }, { "epoch": 0.16619, "grad_norm": 1.1729264327104374, "learning_rate": 0.003, "loss": 4.0471, "step": 16619 }, { "epoch": 0.1662, "grad_norm": 1.3988852778387986, "learning_rate": 0.003, "loss": 4.0671, "step": 16620 }, { "epoch": 0.16621, "grad_norm": 0.998962651394616, "learning_rate": 0.003, "loss": 4.0597, "step": 16621 }, { "epoch": 0.16622, "grad_norm": 1.2830511904376256, "learning_rate": 0.003, "loss": 4.0244, "step": 16622 }, { "epoch": 0.16623, "grad_norm": 1.1378218315230693, "learning_rate": 0.003, "loss": 4.0893, "step": 16623 }, { "epoch": 0.16624, "grad_norm": 1.3807881831736453, "learning_rate": 0.003, "loss": 4.0772, "step": 16624 }, { "epoch": 0.16625, "grad_norm": 0.9858965160582824, "learning_rate": 0.003, "loss": 4.0608, "step": 16625 }, { "epoch": 0.16626, "grad_norm": 1.7187760223123387, "learning_rate": 0.003, "loss": 4.0891, "step": 16626 }, { "epoch": 0.16627, "grad_norm": 1.2227095831403285, "learning_rate": 0.003, "loss": 4.0758, "step": 16627 }, { "epoch": 0.16628, "grad_norm": 1.2230579568409723, "learning_rate": 0.003, "loss": 4.0734, "step": 16628 }, { "epoch": 0.16629, "grad_norm": 1.156960828242787, "learning_rate": 0.003, "loss": 4.1007, "step": 16629 }, { "epoch": 0.1663, "grad_norm": 1.1059706078495835, "learning_rate": 0.003, "loss": 4.0431, "step": 16630 }, { "epoch": 0.16631, "grad_norm": 1.2862955782814594, "learning_rate": 0.003, "loss": 4.0775, "step": 16631 }, { "epoch": 0.16632, "grad_norm": 1.2112016157602854, "learning_rate": 0.003, "loss": 4.0578, "step": 16632 }, { "epoch": 0.16633, "grad_norm": 1.3286289272363898, "learning_rate": 0.003, "loss": 4.0752, "step": 16633 }, { "epoch": 0.16634, "grad_norm": 1.1538531993132737, "learning_rate": 0.003, "loss": 4.0705, "step": 16634 }, { "epoch": 0.16635, "grad_norm": 1.2576693976330668, "learning_rate": 0.003, "loss": 4.0472, "step": 16635 }, { "epoch": 0.16636, "grad_norm": 1.1026365269531988, "learning_rate": 0.003, "loss": 4.0053, "step": 16636 }, { "epoch": 0.16637, "grad_norm": 1.548931553914938, "learning_rate": 0.003, "loss": 4.0265, "step": 16637 }, { "epoch": 0.16638, "grad_norm": 1.117314280310054, "learning_rate": 0.003, "loss": 4.0578, "step": 16638 }, { "epoch": 0.16639, "grad_norm": 1.5756770892701766, "learning_rate": 0.003, "loss": 4.0799, "step": 16639 }, { "epoch": 0.1664, "grad_norm": 0.9983836493429521, "learning_rate": 0.003, "loss": 4.0815, "step": 16640 }, { "epoch": 0.16641, "grad_norm": 1.0402184994585415, "learning_rate": 0.003, "loss": 4.0831, "step": 16641 }, { "epoch": 0.16642, "grad_norm": 1.251995281802957, "learning_rate": 0.003, "loss": 4.0689, "step": 16642 }, { "epoch": 0.16643, "grad_norm": 1.1515925373876876, "learning_rate": 0.003, "loss": 4.0407, "step": 16643 }, { "epoch": 0.16644, "grad_norm": 1.309248830234756, "learning_rate": 0.003, "loss": 4.0839, "step": 16644 }, { "epoch": 0.16645, "grad_norm": 0.970468496424474, "learning_rate": 0.003, "loss": 4.0426, "step": 16645 }, { "epoch": 0.16646, "grad_norm": 1.3792261702226625, "learning_rate": 0.003, "loss": 4.0578, "step": 16646 }, { "epoch": 0.16647, "grad_norm": 1.4702319843822849, "learning_rate": 0.003, "loss": 4.057, "step": 16647 }, { "epoch": 0.16648, "grad_norm": 1.0627611009219051, "learning_rate": 0.003, "loss": 4.0271, "step": 16648 }, { "epoch": 0.16649, "grad_norm": 1.4639207361370845, "learning_rate": 0.003, "loss": 4.0948, "step": 16649 }, { "epoch": 0.1665, "grad_norm": 0.9345316126168038, "learning_rate": 0.003, "loss": 4.0675, "step": 16650 }, { "epoch": 0.16651, "grad_norm": 1.1455019229999, "learning_rate": 0.003, "loss": 4.0321, "step": 16651 }, { "epoch": 0.16652, "grad_norm": 1.2811163596097883, "learning_rate": 0.003, "loss": 4.0736, "step": 16652 }, { "epoch": 0.16653, "grad_norm": 1.3224004194363164, "learning_rate": 0.003, "loss": 4.0589, "step": 16653 }, { "epoch": 0.16654, "grad_norm": 1.3176969368553564, "learning_rate": 0.003, "loss": 4.0601, "step": 16654 }, { "epoch": 0.16655, "grad_norm": 1.0965596801476316, "learning_rate": 0.003, "loss": 4.0405, "step": 16655 }, { "epoch": 0.16656, "grad_norm": 1.4494251414404067, "learning_rate": 0.003, "loss": 4.0478, "step": 16656 }, { "epoch": 0.16657, "grad_norm": 1.0490595911504679, "learning_rate": 0.003, "loss": 4.0582, "step": 16657 }, { "epoch": 0.16658, "grad_norm": 1.5623421520822351, "learning_rate": 0.003, "loss": 4.0602, "step": 16658 }, { "epoch": 0.16659, "grad_norm": 1.0680816050788435, "learning_rate": 0.003, "loss": 4.0816, "step": 16659 }, { "epoch": 0.1666, "grad_norm": 1.2744220397032522, "learning_rate": 0.003, "loss": 4.0553, "step": 16660 }, { "epoch": 0.16661, "grad_norm": 1.2102224257041356, "learning_rate": 0.003, "loss": 4.043, "step": 16661 }, { "epoch": 0.16662, "grad_norm": 1.2072423576390283, "learning_rate": 0.003, "loss": 4.054, "step": 16662 }, { "epoch": 0.16663, "grad_norm": 1.142951208437605, "learning_rate": 0.003, "loss": 4.0509, "step": 16663 }, { "epoch": 0.16664, "grad_norm": 1.2096617764630802, "learning_rate": 0.003, "loss": 4.0687, "step": 16664 }, { "epoch": 0.16665, "grad_norm": 1.3496536003210005, "learning_rate": 0.003, "loss": 4.0871, "step": 16665 }, { "epoch": 0.16666, "grad_norm": 1.1119443012894936, "learning_rate": 0.003, "loss": 4.0433, "step": 16666 }, { "epoch": 0.16667, "grad_norm": 1.3637097231304605, "learning_rate": 0.003, "loss": 4.0567, "step": 16667 }, { "epoch": 0.16668, "grad_norm": 1.136525798824719, "learning_rate": 0.003, "loss": 4.0576, "step": 16668 }, { "epoch": 0.16669, "grad_norm": 1.2871812450875617, "learning_rate": 0.003, "loss": 4.0774, "step": 16669 }, { "epoch": 0.1667, "grad_norm": 1.0484800005031734, "learning_rate": 0.003, "loss": 4.0424, "step": 16670 }, { "epoch": 0.16671, "grad_norm": 1.359449088041725, "learning_rate": 0.003, "loss": 4.0517, "step": 16671 }, { "epoch": 0.16672, "grad_norm": 0.9942863342115491, "learning_rate": 0.003, "loss": 4.0499, "step": 16672 }, { "epoch": 0.16673, "grad_norm": 1.3423120246085614, "learning_rate": 0.003, "loss": 4.0259, "step": 16673 }, { "epoch": 0.16674, "grad_norm": 1.2140808730736126, "learning_rate": 0.003, "loss": 4.0524, "step": 16674 }, { "epoch": 0.16675, "grad_norm": 1.4548008551067455, "learning_rate": 0.003, "loss": 4.069, "step": 16675 }, { "epoch": 0.16676, "grad_norm": 1.1394767621560034, "learning_rate": 0.003, "loss": 4.0392, "step": 16676 }, { "epoch": 0.16677, "grad_norm": 1.109650645696266, "learning_rate": 0.003, "loss": 4.0471, "step": 16677 }, { "epoch": 0.16678, "grad_norm": 1.3453213444110206, "learning_rate": 0.003, "loss": 4.0697, "step": 16678 }, { "epoch": 0.16679, "grad_norm": 0.9973657059977181, "learning_rate": 0.003, "loss": 4.0632, "step": 16679 }, { "epoch": 0.1668, "grad_norm": 1.405969876693654, "learning_rate": 0.003, "loss": 4.0774, "step": 16680 }, { "epoch": 0.16681, "grad_norm": 1.4372242054083253, "learning_rate": 0.003, "loss": 4.0509, "step": 16681 }, { "epoch": 0.16682, "grad_norm": 1.2049277306896602, "learning_rate": 0.003, "loss": 4.0316, "step": 16682 }, { "epoch": 0.16683, "grad_norm": 1.1270606338060425, "learning_rate": 0.003, "loss": 4.0529, "step": 16683 }, { "epoch": 0.16684, "grad_norm": 1.3861618443258414, "learning_rate": 0.003, "loss": 4.0532, "step": 16684 }, { "epoch": 0.16685, "grad_norm": 1.300637850400735, "learning_rate": 0.003, "loss": 4.0699, "step": 16685 }, { "epoch": 0.16686, "grad_norm": 1.4305195354924658, "learning_rate": 0.003, "loss": 4.0492, "step": 16686 }, { "epoch": 0.16687, "grad_norm": 0.9834443407385918, "learning_rate": 0.003, "loss": 4.0723, "step": 16687 }, { "epoch": 0.16688, "grad_norm": 1.1516380475585228, "learning_rate": 0.003, "loss": 4.0623, "step": 16688 }, { "epoch": 0.16689, "grad_norm": 1.3003089848062008, "learning_rate": 0.003, "loss": 4.0419, "step": 16689 }, { "epoch": 0.1669, "grad_norm": 1.1630158367131944, "learning_rate": 0.003, "loss": 4.0414, "step": 16690 }, { "epoch": 0.16691, "grad_norm": 1.3116129953554654, "learning_rate": 0.003, "loss": 4.07, "step": 16691 }, { "epoch": 0.16692, "grad_norm": 1.2074662314667017, "learning_rate": 0.003, "loss": 4.0476, "step": 16692 }, { "epoch": 0.16693, "grad_norm": 1.117522657261956, "learning_rate": 0.003, "loss": 4.0306, "step": 16693 }, { "epoch": 0.16694, "grad_norm": 1.2954370481690003, "learning_rate": 0.003, "loss": 4.0611, "step": 16694 }, { "epoch": 0.16695, "grad_norm": 1.1841582254798, "learning_rate": 0.003, "loss": 4.0725, "step": 16695 }, { "epoch": 0.16696, "grad_norm": 1.1602177828544977, "learning_rate": 0.003, "loss": 4.0374, "step": 16696 }, { "epoch": 0.16697, "grad_norm": 1.1429155185627244, "learning_rate": 0.003, "loss": 4.0474, "step": 16697 }, { "epoch": 0.16698, "grad_norm": 1.345459979414429, "learning_rate": 0.003, "loss": 4.0614, "step": 16698 }, { "epoch": 0.16699, "grad_norm": 1.0878738142344293, "learning_rate": 0.003, "loss": 4.0363, "step": 16699 }, { "epoch": 0.167, "grad_norm": 1.4486378160079911, "learning_rate": 0.003, "loss": 4.0613, "step": 16700 }, { "epoch": 0.16701, "grad_norm": 1.1832505013898313, "learning_rate": 0.003, "loss": 4.0513, "step": 16701 }, { "epoch": 0.16702, "grad_norm": 1.2511837552947567, "learning_rate": 0.003, "loss": 4.0277, "step": 16702 }, { "epoch": 0.16703, "grad_norm": 1.3032500450570588, "learning_rate": 0.003, "loss": 4.0228, "step": 16703 }, { "epoch": 0.16704, "grad_norm": 1.2286025139769188, "learning_rate": 0.003, "loss": 4.0892, "step": 16704 }, { "epoch": 0.16705, "grad_norm": 1.2664712081187757, "learning_rate": 0.003, "loss": 4.0528, "step": 16705 }, { "epoch": 0.16706, "grad_norm": 1.145412191225435, "learning_rate": 0.003, "loss": 4.0471, "step": 16706 }, { "epoch": 0.16707, "grad_norm": 1.0940826598922664, "learning_rate": 0.003, "loss": 4.0477, "step": 16707 }, { "epoch": 0.16708, "grad_norm": 1.4093829939888982, "learning_rate": 0.003, "loss": 4.0468, "step": 16708 }, { "epoch": 0.16709, "grad_norm": 1.1847099195012318, "learning_rate": 0.003, "loss": 4.0743, "step": 16709 }, { "epoch": 0.1671, "grad_norm": 1.2357493273984, "learning_rate": 0.003, "loss": 4.0403, "step": 16710 }, { "epoch": 0.16711, "grad_norm": 1.1098081847606114, "learning_rate": 0.003, "loss": 4.0293, "step": 16711 }, { "epoch": 0.16712, "grad_norm": 1.5584462871650122, "learning_rate": 0.003, "loss": 4.0195, "step": 16712 }, { "epoch": 0.16713, "grad_norm": 1.0848551739989956, "learning_rate": 0.003, "loss": 4.0481, "step": 16713 }, { "epoch": 0.16714, "grad_norm": 1.4922464999775857, "learning_rate": 0.003, "loss": 4.0792, "step": 16714 }, { "epoch": 0.16715, "grad_norm": 1.0532986551389765, "learning_rate": 0.003, "loss": 4.0658, "step": 16715 }, { "epoch": 0.16716, "grad_norm": 1.2566759003383463, "learning_rate": 0.003, "loss": 4.0636, "step": 16716 }, { "epoch": 0.16717, "grad_norm": 1.3231547729401212, "learning_rate": 0.003, "loss": 4.0556, "step": 16717 }, { "epoch": 0.16718, "grad_norm": 1.0533646976294528, "learning_rate": 0.003, "loss": 4.0466, "step": 16718 }, { "epoch": 0.16719, "grad_norm": 1.2786192435431603, "learning_rate": 0.003, "loss": 4.0602, "step": 16719 }, { "epoch": 0.1672, "grad_norm": 1.2718527016177648, "learning_rate": 0.003, "loss": 4.055, "step": 16720 }, { "epoch": 0.16721, "grad_norm": 1.3344911184399857, "learning_rate": 0.003, "loss": 4.0592, "step": 16721 }, { "epoch": 0.16722, "grad_norm": 1.1602486472769074, "learning_rate": 0.003, "loss": 4.04, "step": 16722 }, { "epoch": 0.16723, "grad_norm": 1.251733459671244, "learning_rate": 0.003, "loss": 4.045, "step": 16723 }, { "epoch": 0.16724, "grad_norm": 1.2158861810514705, "learning_rate": 0.003, "loss": 4.0934, "step": 16724 }, { "epoch": 0.16725, "grad_norm": 1.471518052860995, "learning_rate": 0.003, "loss": 4.0927, "step": 16725 }, { "epoch": 0.16726, "grad_norm": 1.0095992941540062, "learning_rate": 0.003, "loss": 4.0304, "step": 16726 }, { "epoch": 0.16727, "grad_norm": 1.324918748254166, "learning_rate": 0.003, "loss": 4.0469, "step": 16727 }, { "epoch": 0.16728, "grad_norm": 1.0844912101007658, "learning_rate": 0.003, "loss": 4.0651, "step": 16728 }, { "epoch": 0.16729, "grad_norm": 1.3535669261603624, "learning_rate": 0.003, "loss": 4.059, "step": 16729 }, { "epoch": 0.1673, "grad_norm": 1.2555576937702524, "learning_rate": 0.003, "loss": 4.1071, "step": 16730 }, { "epoch": 0.16731, "grad_norm": 1.3869000673568552, "learning_rate": 0.003, "loss": 4.0557, "step": 16731 }, { "epoch": 0.16732, "grad_norm": 1.217822597439683, "learning_rate": 0.003, "loss": 4.0635, "step": 16732 }, { "epoch": 0.16733, "grad_norm": 1.1401702983763804, "learning_rate": 0.003, "loss": 4.0494, "step": 16733 }, { "epoch": 0.16734, "grad_norm": 1.5260237630917093, "learning_rate": 0.003, "loss": 4.0654, "step": 16734 }, { "epoch": 0.16735, "grad_norm": 1.0878394273854046, "learning_rate": 0.003, "loss": 4.0931, "step": 16735 }, { "epoch": 0.16736, "grad_norm": 1.4571879519148716, "learning_rate": 0.003, "loss": 4.0758, "step": 16736 }, { "epoch": 0.16737, "grad_norm": 1.1484215831393614, "learning_rate": 0.003, "loss": 4.057, "step": 16737 }, { "epoch": 0.16738, "grad_norm": 1.239249105824225, "learning_rate": 0.003, "loss": 4.066, "step": 16738 }, { "epoch": 0.16739, "grad_norm": 1.0846360676996116, "learning_rate": 0.003, "loss": 4.0636, "step": 16739 }, { "epoch": 0.1674, "grad_norm": 1.2457041173856414, "learning_rate": 0.003, "loss": 4.0437, "step": 16740 }, { "epoch": 0.16741, "grad_norm": 1.278139066819567, "learning_rate": 0.003, "loss": 4.0866, "step": 16741 }, { "epoch": 0.16742, "grad_norm": 1.288984975340617, "learning_rate": 0.003, "loss": 4.0428, "step": 16742 }, { "epoch": 0.16743, "grad_norm": 1.2454372914172789, "learning_rate": 0.003, "loss": 4.066, "step": 16743 }, { "epoch": 0.16744, "grad_norm": 1.1742155484056467, "learning_rate": 0.003, "loss": 4.0873, "step": 16744 }, { "epoch": 0.16745, "grad_norm": 1.085445817969519, "learning_rate": 0.003, "loss": 4.0492, "step": 16745 }, { "epoch": 0.16746, "grad_norm": 1.2708236741095844, "learning_rate": 0.003, "loss": 4.0905, "step": 16746 }, { "epoch": 0.16747, "grad_norm": 1.2218806583737991, "learning_rate": 0.003, "loss": 4.0559, "step": 16747 }, { "epoch": 0.16748, "grad_norm": 1.2064425335775943, "learning_rate": 0.003, "loss": 4.0785, "step": 16748 }, { "epoch": 0.16749, "grad_norm": 1.1619865034622123, "learning_rate": 0.003, "loss": 4.062, "step": 16749 }, { "epoch": 0.1675, "grad_norm": 1.1704000966985963, "learning_rate": 0.003, "loss": 4.0511, "step": 16750 }, { "epoch": 0.16751, "grad_norm": 1.1183024761571165, "learning_rate": 0.003, "loss": 4.0537, "step": 16751 }, { "epoch": 0.16752, "grad_norm": 1.0937304186087908, "learning_rate": 0.003, "loss": 4.0512, "step": 16752 }, { "epoch": 0.16753, "grad_norm": 1.373737523421354, "learning_rate": 0.003, "loss": 4.0602, "step": 16753 }, { "epoch": 0.16754, "grad_norm": 1.3196540953296025, "learning_rate": 0.003, "loss": 4.0714, "step": 16754 }, { "epoch": 0.16755, "grad_norm": 1.1139155806499885, "learning_rate": 0.003, "loss": 4.0779, "step": 16755 }, { "epoch": 0.16756, "grad_norm": 1.4359115891977452, "learning_rate": 0.003, "loss": 4.0291, "step": 16756 }, { "epoch": 0.16757, "grad_norm": 1.1165242308607433, "learning_rate": 0.003, "loss": 4.0601, "step": 16757 }, { "epoch": 0.16758, "grad_norm": 1.182216776928444, "learning_rate": 0.003, "loss": 4.0521, "step": 16758 }, { "epoch": 0.16759, "grad_norm": 1.4008740145971497, "learning_rate": 0.003, "loss": 4.0403, "step": 16759 }, { "epoch": 0.1676, "grad_norm": 1.1061524789097879, "learning_rate": 0.003, "loss": 4.0643, "step": 16760 }, { "epoch": 0.16761, "grad_norm": 1.2007991816682586, "learning_rate": 0.003, "loss": 4.0761, "step": 16761 }, { "epoch": 0.16762, "grad_norm": 0.9883652542426329, "learning_rate": 0.003, "loss": 4.0579, "step": 16762 }, { "epoch": 0.16763, "grad_norm": 1.2734810036538666, "learning_rate": 0.003, "loss": 4.0707, "step": 16763 }, { "epoch": 0.16764, "grad_norm": 0.999922945265591, "learning_rate": 0.003, "loss": 4.0297, "step": 16764 }, { "epoch": 0.16765, "grad_norm": 1.4319142076681646, "learning_rate": 0.003, "loss": 4.0726, "step": 16765 }, { "epoch": 0.16766, "grad_norm": 1.1343143488760452, "learning_rate": 0.003, "loss": 4.1002, "step": 16766 }, { "epoch": 0.16767, "grad_norm": 1.2168925527908758, "learning_rate": 0.003, "loss": 4.05, "step": 16767 }, { "epoch": 0.16768, "grad_norm": 1.1988282700392212, "learning_rate": 0.003, "loss": 4.0832, "step": 16768 }, { "epoch": 0.16769, "grad_norm": 1.1145099318152145, "learning_rate": 0.003, "loss": 4.0566, "step": 16769 }, { "epoch": 0.1677, "grad_norm": 1.5275396374681, "learning_rate": 0.003, "loss": 4.0625, "step": 16770 }, { "epoch": 0.16771, "grad_norm": 1.192165268592216, "learning_rate": 0.003, "loss": 4.0829, "step": 16771 }, { "epoch": 0.16772, "grad_norm": 1.2292219207859179, "learning_rate": 0.003, "loss": 4.0513, "step": 16772 }, { "epoch": 0.16773, "grad_norm": 1.3214290398435105, "learning_rate": 0.003, "loss": 4.0248, "step": 16773 }, { "epoch": 0.16774, "grad_norm": 1.1882177696146676, "learning_rate": 0.003, "loss": 4.0676, "step": 16774 }, { "epoch": 0.16775, "grad_norm": 1.548376013128292, "learning_rate": 0.003, "loss": 4.0407, "step": 16775 }, { "epoch": 0.16776, "grad_norm": 1.182191136580024, "learning_rate": 0.003, "loss": 4.0589, "step": 16776 }, { "epoch": 0.16777, "grad_norm": 1.4569800869382106, "learning_rate": 0.003, "loss": 4.0691, "step": 16777 }, { "epoch": 0.16778, "grad_norm": 1.0865748237703814, "learning_rate": 0.003, "loss": 4.089, "step": 16778 }, { "epoch": 0.16779, "grad_norm": 1.277386244617626, "learning_rate": 0.003, "loss": 4.0788, "step": 16779 }, { "epoch": 0.1678, "grad_norm": 1.0391810445636955, "learning_rate": 0.003, "loss": 4.0644, "step": 16780 }, { "epoch": 0.16781, "grad_norm": 1.1643208648906385, "learning_rate": 0.003, "loss": 4.1017, "step": 16781 }, { "epoch": 0.16782, "grad_norm": 1.1873531329716722, "learning_rate": 0.003, "loss": 4.0641, "step": 16782 }, { "epoch": 0.16783, "grad_norm": 1.1934136985580575, "learning_rate": 0.003, "loss": 4.0802, "step": 16783 }, { "epoch": 0.16784, "grad_norm": 1.2671358079736963, "learning_rate": 0.003, "loss": 4.0553, "step": 16784 }, { "epoch": 0.16785, "grad_norm": 1.3854175483820694, "learning_rate": 0.003, "loss": 4.0776, "step": 16785 }, { "epoch": 0.16786, "grad_norm": 1.2081305889906775, "learning_rate": 0.003, "loss": 4.0887, "step": 16786 }, { "epoch": 0.16787, "grad_norm": 1.214557089350369, "learning_rate": 0.003, "loss": 4.053, "step": 16787 }, { "epoch": 0.16788, "grad_norm": 1.2221538985193494, "learning_rate": 0.003, "loss": 4.0892, "step": 16788 }, { "epoch": 0.16789, "grad_norm": 1.2767535267757835, "learning_rate": 0.003, "loss": 4.0736, "step": 16789 }, { "epoch": 0.1679, "grad_norm": 1.1098423119957812, "learning_rate": 0.003, "loss": 4.0333, "step": 16790 }, { "epoch": 0.16791, "grad_norm": 1.4444174542229276, "learning_rate": 0.003, "loss": 4.0842, "step": 16791 }, { "epoch": 0.16792, "grad_norm": 1.151166125718121, "learning_rate": 0.003, "loss": 4.0245, "step": 16792 }, { "epoch": 0.16793, "grad_norm": 1.456350573735467, "learning_rate": 0.003, "loss": 4.0531, "step": 16793 }, { "epoch": 0.16794, "grad_norm": 1.313794112815368, "learning_rate": 0.003, "loss": 4.0562, "step": 16794 }, { "epoch": 0.16795, "grad_norm": 1.0236949668333608, "learning_rate": 0.003, "loss": 4.0486, "step": 16795 }, { "epoch": 0.16796, "grad_norm": 1.4150976958572334, "learning_rate": 0.003, "loss": 4.0292, "step": 16796 }, { "epoch": 0.16797, "grad_norm": 1.137530296809528, "learning_rate": 0.003, "loss": 4.0685, "step": 16797 }, { "epoch": 0.16798, "grad_norm": 1.4597846986507814, "learning_rate": 0.003, "loss": 4.0697, "step": 16798 }, { "epoch": 0.16799, "grad_norm": 1.0377322869212624, "learning_rate": 0.003, "loss": 4.0699, "step": 16799 }, { "epoch": 0.168, "grad_norm": 1.2591973812317476, "learning_rate": 0.003, "loss": 4.0512, "step": 16800 }, { "epoch": 0.16801, "grad_norm": 1.1764894782139463, "learning_rate": 0.003, "loss": 4.0621, "step": 16801 }, { "epoch": 0.16802, "grad_norm": 1.2632774612926907, "learning_rate": 0.003, "loss": 4.067, "step": 16802 }, { "epoch": 0.16803, "grad_norm": 1.29296779956215, "learning_rate": 0.003, "loss": 4.0954, "step": 16803 }, { "epoch": 0.16804, "grad_norm": 1.0144603330901674, "learning_rate": 0.003, "loss": 4.0659, "step": 16804 }, { "epoch": 0.16805, "grad_norm": 1.2248370621605944, "learning_rate": 0.003, "loss": 4.0644, "step": 16805 }, { "epoch": 0.16806, "grad_norm": 1.606685061239406, "learning_rate": 0.003, "loss": 4.0888, "step": 16806 }, { "epoch": 0.16807, "grad_norm": 1.1500315710151148, "learning_rate": 0.003, "loss": 4.0385, "step": 16807 }, { "epoch": 0.16808, "grad_norm": 1.3947512644262978, "learning_rate": 0.003, "loss": 4.0668, "step": 16808 }, { "epoch": 0.16809, "grad_norm": 1.1025214300653976, "learning_rate": 0.003, "loss": 4.0651, "step": 16809 }, { "epoch": 0.1681, "grad_norm": 1.3507202635811593, "learning_rate": 0.003, "loss": 4.0794, "step": 16810 }, { "epoch": 0.16811, "grad_norm": 1.1370812976046, "learning_rate": 0.003, "loss": 4.0873, "step": 16811 }, { "epoch": 0.16812, "grad_norm": 1.3641818804777335, "learning_rate": 0.003, "loss": 4.0429, "step": 16812 }, { "epoch": 0.16813, "grad_norm": 1.2657995069383365, "learning_rate": 0.003, "loss": 4.0552, "step": 16813 }, { "epoch": 0.16814, "grad_norm": 1.0623824621060314, "learning_rate": 0.003, "loss": 4.0342, "step": 16814 }, { "epoch": 0.16815, "grad_norm": 1.2610460250197402, "learning_rate": 0.003, "loss": 4.1043, "step": 16815 }, { "epoch": 0.16816, "grad_norm": 1.2207105633397042, "learning_rate": 0.003, "loss": 4.0539, "step": 16816 }, { "epoch": 0.16817, "grad_norm": 1.2889344728156151, "learning_rate": 0.003, "loss": 4.0584, "step": 16817 }, { "epoch": 0.16818, "grad_norm": 1.405483760348606, "learning_rate": 0.003, "loss": 4.0683, "step": 16818 }, { "epoch": 0.16819, "grad_norm": 1.0386857054925762, "learning_rate": 0.003, "loss": 4.0321, "step": 16819 }, { "epoch": 0.1682, "grad_norm": 1.2465438352281049, "learning_rate": 0.003, "loss": 4.0529, "step": 16820 }, { "epoch": 0.16821, "grad_norm": 1.196637130653281, "learning_rate": 0.003, "loss": 4.0278, "step": 16821 }, { "epoch": 0.16822, "grad_norm": 1.457820682101557, "learning_rate": 0.003, "loss": 4.0573, "step": 16822 }, { "epoch": 0.16823, "grad_norm": 1.1613232889985434, "learning_rate": 0.003, "loss": 4.047, "step": 16823 }, { "epoch": 0.16824, "grad_norm": 1.2706295316842788, "learning_rate": 0.003, "loss": 4.0764, "step": 16824 }, { "epoch": 0.16825, "grad_norm": 1.0911487428394981, "learning_rate": 0.003, "loss": 4.0601, "step": 16825 }, { "epoch": 0.16826, "grad_norm": 1.3595954191392943, "learning_rate": 0.003, "loss": 4.0621, "step": 16826 }, { "epoch": 0.16827, "grad_norm": 1.2658568820915035, "learning_rate": 0.003, "loss": 4.0556, "step": 16827 }, { "epoch": 0.16828, "grad_norm": 1.397019207943614, "learning_rate": 0.003, "loss": 4.0696, "step": 16828 }, { "epoch": 0.16829, "grad_norm": 1.1829917724750292, "learning_rate": 0.003, "loss": 4.0665, "step": 16829 }, { "epoch": 0.1683, "grad_norm": 1.1737431792927684, "learning_rate": 0.003, "loss": 4.0596, "step": 16830 }, { "epoch": 0.16831, "grad_norm": 1.2377992906490987, "learning_rate": 0.003, "loss": 4.0373, "step": 16831 }, { "epoch": 0.16832, "grad_norm": 1.2105804851112891, "learning_rate": 0.003, "loss": 4.0572, "step": 16832 }, { "epoch": 0.16833, "grad_norm": 1.1670172496235744, "learning_rate": 0.003, "loss": 4.03, "step": 16833 }, { "epoch": 0.16834, "grad_norm": 1.111923517385695, "learning_rate": 0.003, "loss": 4.0425, "step": 16834 }, { "epoch": 0.16835, "grad_norm": 1.2673999166716663, "learning_rate": 0.003, "loss": 4.0109, "step": 16835 }, { "epoch": 0.16836, "grad_norm": 1.2584695185196386, "learning_rate": 0.003, "loss": 4.069, "step": 16836 }, { "epoch": 0.16837, "grad_norm": 1.0391253683909778, "learning_rate": 0.003, "loss": 4.0487, "step": 16837 }, { "epoch": 0.16838, "grad_norm": 1.2686470581121168, "learning_rate": 0.003, "loss": 4.0849, "step": 16838 }, { "epoch": 0.16839, "grad_norm": 0.939093786741332, "learning_rate": 0.003, "loss": 4.0441, "step": 16839 }, { "epoch": 0.1684, "grad_norm": 1.2721900411497336, "learning_rate": 0.003, "loss": 4.0494, "step": 16840 }, { "epoch": 0.16841, "grad_norm": 1.2357322793990546, "learning_rate": 0.003, "loss": 4.0609, "step": 16841 }, { "epoch": 0.16842, "grad_norm": 1.1881231361127453, "learning_rate": 0.003, "loss": 4.0564, "step": 16842 }, { "epoch": 0.16843, "grad_norm": 1.1508538543573568, "learning_rate": 0.003, "loss": 4.0689, "step": 16843 }, { "epoch": 0.16844, "grad_norm": 1.2773939686381024, "learning_rate": 0.003, "loss": 4.0846, "step": 16844 }, { "epoch": 0.16845, "grad_norm": 1.2152273111502037, "learning_rate": 0.003, "loss": 4.0724, "step": 16845 }, { "epoch": 0.16846, "grad_norm": 1.3766362139974697, "learning_rate": 0.003, "loss": 4.0799, "step": 16846 }, { "epoch": 0.16847, "grad_norm": 1.378993780999002, "learning_rate": 0.003, "loss": 4.0426, "step": 16847 }, { "epoch": 0.16848, "grad_norm": 1.5109246400409688, "learning_rate": 0.003, "loss": 4.0475, "step": 16848 }, { "epoch": 0.16849, "grad_norm": 0.953824324235403, "learning_rate": 0.003, "loss": 4.0383, "step": 16849 }, { "epoch": 0.1685, "grad_norm": 1.339281249581703, "learning_rate": 0.003, "loss": 4.076, "step": 16850 }, { "epoch": 0.16851, "grad_norm": 1.2998593399676501, "learning_rate": 0.003, "loss": 4.0413, "step": 16851 }, { "epoch": 0.16852, "grad_norm": 1.0164201352560711, "learning_rate": 0.003, "loss": 4.0581, "step": 16852 }, { "epoch": 0.16853, "grad_norm": 1.4406245656231773, "learning_rate": 0.003, "loss": 4.0843, "step": 16853 }, { "epoch": 0.16854, "grad_norm": 0.9835109188131523, "learning_rate": 0.003, "loss": 4.0575, "step": 16854 }, { "epoch": 0.16855, "grad_norm": 1.3767727285210112, "learning_rate": 0.003, "loss": 4.0575, "step": 16855 }, { "epoch": 0.16856, "grad_norm": 1.1044606829707675, "learning_rate": 0.003, "loss": 4.0656, "step": 16856 }, { "epoch": 0.16857, "grad_norm": 1.299670854958639, "learning_rate": 0.003, "loss": 4.0215, "step": 16857 }, { "epoch": 0.16858, "grad_norm": 1.2416317130185637, "learning_rate": 0.003, "loss": 4.0325, "step": 16858 }, { "epoch": 0.16859, "grad_norm": 1.1737138450195503, "learning_rate": 0.003, "loss": 4.0561, "step": 16859 }, { "epoch": 0.1686, "grad_norm": 1.3630157587564866, "learning_rate": 0.003, "loss": 4.0779, "step": 16860 }, { "epoch": 0.16861, "grad_norm": 1.1138327516009972, "learning_rate": 0.003, "loss": 4.0653, "step": 16861 }, { "epoch": 0.16862, "grad_norm": 1.4377750117381896, "learning_rate": 0.003, "loss": 4.0633, "step": 16862 }, { "epoch": 0.16863, "grad_norm": 1.087949320127546, "learning_rate": 0.003, "loss": 4.05, "step": 16863 }, { "epoch": 0.16864, "grad_norm": 1.4837405757909432, "learning_rate": 0.003, "loss": 4.0631, "step": 16864 }, { "epoch": 0.16865, "grad_norm": 1.533739270840856, "learning_rate": 0.003, "loss": 4.0394, "step": 16865 }, { "epoch": 0.16866, "grad_norm": 0.9784322019819925, "learning_rate": 0.003, "loss": 4.0608, "step": 16866 }, { "epoch": 0.16867, "grad_norm": 1.299062648690208, "learning_rate": 0.003, "loss": 4.0535, "step": 16867 }, { "epoch": 0.16868, "grad_norm": 1.2283298033113765, "learning_rate": 0.003, "loss": 4.0531, "step": 16868 }, { "epoch": 0.16869, "grad_norm": 1.1134405692621459, "learning_rate": 0.003, "loss": 4.0486, "step": 16869 }, { "epoch": 0.1687, "grad_norm": 1.1747016739767506, "learning_rate": 0.003, "loss": 4.0768, "step": 16870 }, { "epoch": 0.16871, "grad_norm": 1.2950017560021867, "learning_rate": 0.003, "loss": 4.0803, "step": 16871 }, { "epoch": 0.16872, "grad_norm": 1.0149044248103394, "learning_rate": 0.003, "loss": 4.0446, "step": 16872 }, { "epoch": 0.16873, "grad_norm": 1.391205949700118, "learning_rate": 0.003, "loss": 4.0559, "step": 16873 }, { "epoch": 0.16874, "grad_norm": 0.9455363272123867, "learning_rate": 0.003, "loss": 4.0376, "step": 16874 }, { "epoch": 0.16875, "grad_norm": 1.129067712212612, "learning_rate": 0.003, "loss": 4.0725, "step": 16875 }, { "epoch": 0.16876, "grad_norm": 1.3755510568160323, "learning_rate": 0.003, "loss": 4.0706, "step": 16876 }, { "epoch": 0.16877, "grad_norm": 1.1459983143566221, "learning_rate": 0.003, "loss": 4.0425, "step": 16877 }, { "epoch": 0.16878, "grad_norm": 1.1875243488588099, "learning_rate": 0.003, "loss": 4.038, "step": 16878 }, { "epoch": 0.16879, "grad_norm": 1.5235497085078586, "learning_rate": 0.003, "loss": 4.0353, "step": 16879 }, { "epoch": 0.1688, "grad_norm": 1.2689704443388405, "learning_rate": 0.003, "loss": 4.086, "step": 16880 }, { "epoch": 0.16881, "grad_norm": 1.3657585585182919, "learning_rate": 0.003, "loss": 4.0483, "step": 16881 }, { "epoch": 0.16882, "grad_norm": 1.2518826080151741, "learning_rate": 0.003, "loss": 4.0624, "step": 16882 }, { "epoch": 0.16883, "grad_norm": 1.2398375070189747, "learning_rate": 0.003, "loss": 4.0732, "step": 16883 }, { "epoch": 0.16884, "grad_norm": 1.2043682351070955, "learning_rate": 0.003, "loss": 4.0598, "step": 16884 }, { "epoch": 0.16885, "grad_norm": 1.1290648098157186, "learning_rate": 0.003, "loss": 4.0293, "step": 16885 }, { "epoch": 0.16886, "grad_norm": 1.1214426109760152, "learning_rate": 0.003, "loss": 4.0545, "step": 16886 }, { "epoch": 0.16887, "grad_norm": 1.1907149719752812, "learning_rate": 0.003, "loss": 4.0794, "step": 16887 }, { "epoch": 0.16888, "grad_norm": 1.171586890870058, "learning_rate": 0.003, "loss": 4.0837, "step": 16888 }, { "epoch": 0.16889, "grad_norm": 1.2615857750256425, "learning_rate": 0.003, "loss": 4.0569, "step": 16889 }, { "epoch": 0.1689, "grad_norm": 1.1590161604632432, "learning_rate": 0.003, "loss": 4.0654, "step": 16890 }, { "epoch": 0.16891, "grad_norm": 1.201379688982921, "learning_rate": 0.003, "loss": 4.0438, "step": 16891 }, { "epoch": 0.16892, "grad_norm": 1.2085583376593114, "learning_rate": 0.003, "loss": 4.0327, "step": 16892 }, { "epoch": 0.16893, "grad_norm": 1.2833285113963626, "learning_rate": 0.003, "loss": 4.0734, "step": 16893 }, { "epoch": 0.16894, "grad_norm": 1.2010659385222722, "learning_rate": 0.003, "loss": 4.0543, "step": 16894 }, { "epoch": 0.16895, "grad_norm": 1.268911100327642, "learning_rate": 0.003, "loss": 4.08, "step": 16895 }, { "epoch": 0.16896, "grad_norm": 1.4269464293669927, "learning_rate": 0.003, "loss": 4.057, "step": 16896 }, { "epoch": 0.16897, "grad_norm": 1.0134031667736194, "learning_rate": 0.003, "loss": 4.0387, "step": 16897 }, { "epoch": 0.16898, "grad_norm": 1.3552971992354175, "learning_rate": 0.003, "loss": 4.054, "step": 16898 }, { "epoch": 0.16899, "grad_norm": 1.1969400108620583, "learning_rate": 0.003, "loss": 4.0659, "step": 16899 }, { "epoch": 0.169, "grad_norm": 1.2440903886771624, "learning_rate": 0.003, "loss": 4.0312, "step": 16900 }, { "epoch": 0.16901, "grad_norm": 1.2832883433021833, "learning_rate": 0.003, "loss": 4.053, "step": 16901 }, { "epoch": 0.16902, "grad_norm": 1.27406726673779, "learning_rate": 0.003, "loss": 4.0539, "step": 16902 }, { "epoch": 0.16903, "grad_norm": 1.2260976174013318, "learning_rate": 0.003, "loss": 4.047, "step": 16903 }, { "epoch": 0.16904, "grad_norm": 1.1966634019522435, "learning_rate": 0.003, "loss": 4.0957, "step": 16904 }, { "epoch": 0.16905, "grad_norm": 1.1490515794269593, "learning_rate": 0.003, "loss": 4.0603, "step": 16905 }, { "epoch": 0.16906, "grad_norm": 1.198462401149591, "learning_rate": 0.003, "loss": 4.056, "step": 16906 }, { "epoch": 0.16907, "grad_norm": 1.3693505230975247, "learning_rate": 0.003, "loss": 4.0478, "step": 16907 }, { "epoch": 0.16908, "grad_norm": 1.2447177579825452, "learning_rate": 0.003, "loss": 4.0587, "step": 16908 }, { "epoch": 0.16909, "grad_norm": 1.2208258128228069, "learning_rate": 0.003, "loss": 4.0507, "step": 16909 }, { "epoch": 0.1691, "grad_norm": 1.1390878433907936, "learning_rate": 0.003, "loss": 4.0097, "step": 16910 }, { "epoch": 0.16911, "grad_norm": 1.2772889738612396, "learning_rate": 0.003, "loss": 4.0484, "step": 16911 }, { "epoch": 0.16912, "grad_norm": 1.0495387417713018, "learning_rate": 0.003, "loss": 4.0685, "step": 16912 }, { "epoch": 0.16913, "grad_norm": 1.236094976534584, "learning_rate": 0.003, "loss": 4.052, "step": 16913 }, { "epoch": 0.16914, "grad_norm": 1.3062207475022636, "learning_rate": 0.003, "loss": 4.0652, "step": 16914 }, { "epoch": 0.16915, "grad_norm": 1.064390364463006, "learning_rate": 0.003, "loss": 4.0557, "step": 16915 }, { "epoch": 0.16916, "grad_norm": 1.5191869092136066, "learning_rate": 0.003, "loss": 4.0783, "step": 16916 }, { "epoch": 0.16917, "grad_norm": 1.3670090944649158, "learning_rate": 0.003, "loss": 4.0437, "step": 16917 }, { "epoch": 0.16918, "grad_norm": 1.2314801927745396, "learning_rate": 0.003, "loss": 4.0734, "step": 16918 }, { "epoch": 0.16919, "grad_norm": 1.2714919494982853, "learning_rate": 0.003, "loss": 4.0573, "step": 16919 }, { "epoch": 0.1692, "grad_norm": 1.0245371421311447, "learning_rate": 0.003, "loss": 4.0299, "step": 16920 }, { "epoch": 0.16921, "grad_norm": 1.399187969255701, "learning_rate": 0.003, "loss": 4.0641, "step": 16921 }, { "epoch": 0.16922, "grad_norm": 1.1394392772369148, "learning_rate": 0.003, "loss": 4.0612, "step": 16922 }, { "epoch": 0.16923, "grad_norm": 1.2358651424794553, "learning_rate": 0.003, "loss": 4.0575, "step": 16923 }, { "epoch": 0.16924, "grad_norm": 1.274086921485851, "learning_rate": 0.003, "loss": 4.0645, "step": 16924 }, { "epoch": 0.16925, "grad_norm": 1.2177939284397616, "learning_rate": 0.003, "loss": 4.0728, "step": 16925 }, { "epoch": 0.16926, "grad_norm": 1.2178018449888186, "learning_rate": 0.003, "loss": 4.0519, "step": 16926 }, { "epoch": 0.16927, "grad_norm": 1.2757120385924865, "learning_rate": 0.003, "loss": 4.0355, "step": 16927 }, { "epoch": 0.16928, "grad_norm": 1.2282608548823535, "learning_rate": 0.003, "loss": 4.0597, "step": 16928 }, { "epoch": 0.16929, "grad_norm": 1.3189168954328372, "learning_rate": 0.003, "loss": 4.0438, "step": 16929 }, { "epoch": 0.1693, "grad_norm": 1.092074440286679, "learning_rate": 0.003, "loss": 4.0593, "step": 16930 }, { "epoch": 0.16931, "grad_norm": 1.196966518540893, "learning_rate": 0.003, "loss": 4.0754, "step": 16931 }, { "epoch": 0.16932, "grad_norm": 1.320981547763483, "learning_rate": 0.003, "loss": 4.0616, "step": 16932 }, { "epoch": 0.16933, "grad_norm": 1.1825425271113728, "learning_rate": 0.003, "loss": 4.0667, "step": 16933 }, { "epoch": 0.16934, "grad_norm": 1.3499495322767026, "learning_rate": 0.003, "loss": 4.075, "step": 16934 }, { "epoch": 0.16935, "grad_norm": 1.0805460557226116, "learning_rate": 0.003, "loss": 4.0688, "step": 16935 }, { "epoch": 0.16936, "grad_norm": 1.2386571350265883, "learning_rate": 0.003, "loss": 4.0317, "step": 16936 }, { "epoch": 0.16937, "grad_norm": 1.0856980807139698, "learning_rate": 0.003, "loss": 4.0276, "step": 16937 }, { "epoch": 0.16938, "grad_norm": 1.2197054766341284, "learning_rate": 0.003, "loss": 4.0444, "step": 16938 }, { "epoch": 0.16939, "grad_norm": 1.2668817756255695, "learning_rate": 0.003, "loss": 4.0551, "step": 16939 }, { "epoch": 0.1694, "grad_norm": 1.1224968209854305, "learning_rate": 0.003, "loss": 4.0635, "step": 16940 }, { "epoch": 0.16941, "grad_norm": 1.2091332805281065, "learning_rate": 0.003, "loss": 4.0717, "step": 16941 }, { "epoch": 0.16942, "grad_norm": 1.184514981822149, "learning_rate": 0.003, "loss": 4.0544, "step": 16942 }, { "epoch": 0.16943, "grad_norm": 1.1619732892487817, "learning_rate": 0.003, "loss": 4.0142, "step": 16943 }, { "epoch": 0.16944, "grad_norm": 1.2575293378460344, "learning_rate": 0.003, "loss": 4.0557, "step": 16944 }, { "epoch": 0.16945, "grad_norm": 1.2800738467900814, "learning_rate": 0.003, "loss": 4.0449, "step": 16945 }, { "epoch": 0.16946, "grad_norm": 0.9543077777657314, "learning_rate": 0.003, "loss": 4.0608, "step": 16946 }, { "epoch": 0.16947, "grad_norm": 1.2539968033580715, "learning_rate": 0.003, "loss": 4.0644, "step": 16947 }, { "epoch": 0.16948, "grad_norm": 1.134897054004596, "learning_rate": 0.003, "loss": 4.0661, "step": 16948 }, { "epoch": 0.16949, "grad_norm": 1.1783067609268147, "learning_rate": 0.003, "loss": 4.0173, "step": 16949 }, { "epoch": 0.1695, "grad_norm": 1.3061849316390377, "learning_rate": 0.003, "loss": 4.0696, "step": 16950 }, { "epoch": 0.16951, "grad_norm": 1.3587387222632443, "learning_rate": 0.003, "loss": 4.0601, "step": 16951 }, { "epoch": 0.16952, "grad_norm": 1.3807827782986366, "learning_rate": 0.003, "loss": 4.0496, "step": 16952 }, { "epoch": 0.16953, "grad_norm": 1.443310911870979, "learning_rate": 0.003, "loss": 4.0651, "step": 16953 }, { "epoch": 0.16954, "grad_norm": 1.1588162955104133, "learning_rate": 0.003, "loss": 4.095, "step": 16954 }, { "epoch": 0.16955, "grad_norm": 1.366344017441944, "learning_rate": 0.003, "loss": 4.0774, "step": 16955 }, { "epoch": 0.16956, "grad_norm": 1.0687092197667276, "learning_rate": 0.003, "loss": 4.0573, "step": 16956 }, { "epoch": 0.16957, "grad_norm": 1.5070998108760714, "learning_rate": 0.003, "loss": 4.074, "step": 16957 }, { "epoch": 0.16958, "grad_norm": 1.0440279203002143, "learning_rate": 0.003, "loss": 4.0426, "step": 16958 }, { "epoch": 0.16959, "grad_norm": 1.2029541519854532, "learning_rate": 0.003, "loss": 4.0366, "step": 16959 }, { "epoch": 0.1696, "grad_norm": 1.1639026967557624, "learning_rate": 0.003, "loss": 4.0488, "step": 16960 }, { "epoch": 0.16961, "grad_norm": 1.2879390201346683, "learning_rate": 0.003, "loss": 4.0413, "step": 16961 }, { "epoch": 0.16962, "grad_norm": 1.2338983857667059, "learning_rate": 0.003, "loss": 4.0867, "step": 16962 }, { "epoch": 0.16963, "grad_norm": 1.334995760107833, "learning_rate": 0.003, "loss": 4.0772, "step": 16963 }, { "epoch": 0.16964, "grad_norm": 1.2032484577130451, "learning_rate": 0.003, "loss": 4.0752, "step": 16964 }, { "epoch": 0.16965, "grad_norm": 1.3254940084493645, "learning_rate": 0.003, "loss": 4.0313, "step": 16965 }, { "epoch": 0.16966, "grad_norm": 1.1808584653625602, "learning_rate": 0.003, "loss": 4.0744, "step": 16966 }, { "epoch": 0.16967, "grad_norm": 1.0956024495103562, "learning_rate": 0.003, "loss": 4.0527, "step": 16967 }, { "epoch": 0.16968, "grad_norm": 1.3582823693965818, "learning_rate": 0.003, "loss": 4.0764, "step": 16968 }, { "epoch": 0.16969, "grad_norm": 1.3127841660646782, "learning_rate": 0.003, "loss": 4.0745, "step": 16969 }, { "epoch": 0.1697, "grad_norm": 1.0590613504253792, "learning_rate": 0.003, "loss": 4.0575, "step": 16970 }, { "epoch": 0.16971, "grad_norm": 1.3370003005267381, "learning_rate": 0.003, "loss": 4.0721, "step": 16971 }, { "epoch": 0.16972, "grad_norm": 1.0825998424017913, "learning_rate": 0.003, "loss": 4.066, "step": 16972 }, { "epoch": 0.16973, "grad_norm": 1.5805086190217519, "learning_rate": 0.003, "loss": 4.0826, "step": 16973 }, { "epoch": 0.16974, "grad_norm": 1.21296940946939, "learning_rate": 0.003, "loss": 4.0443, "step": 16974 }, { "epoch": 0.16975, "grad_norm": 1.2413744805373763, "learning_rate": 0.003, "loss": 4.0657, "step": 16975 }, { "epoch": 0.16976, "grad_norm": 1.1799095411296254, "learning_rate": 0.003, "loss": 4.0156, "step": 16976 }, { "epoch": 0.16977, "grad_norm": 1.3159578433009453, "learning_rate": 0.003, "loss": 4.0554, "step": 16977 }, { "epoch": 0.16978, "grad_norm": 1.5239983621956583, "learning_rate": 0.003, "loss": 4.0474, "step": 16978 }, { "epoch": 0.16979, "grad_norm": 1.2117505051118325, "learning_rate": 0.003, "loss": 4.0447, "step": 16979 }, { "epoch": 0.1698, "grad_norm": 1.1247154415502227, "learning_rate": 0.003, "loss": 4.0474, "step": 16980 }, { "epoch": 0.16981, "grad_norm": 1.3699757772248973, "learning_rate": 0.003, "loss": 4.0353, "step": 16981 }, { "epoch": 0.16982, "grad_norm": 1.017653817874775, "learning_rate": 0.003, "loss": 4.0768, "step": 16982 }, { "epoch": 0.16983, "grad_norm": 1.4629027360363662, "learning_rate": 0.003, "loss": 4.0683, "step": 16983 }, { "epoch": 0.16984, "grad_norm": 0.9247191316273929, "learning_rate": 0.003, "loss": 4.0802, "step": 16984 }, { "epoch": 0.16985, "grad_norm": 1.4410991040670977, "learning_rate": 0.003, "loss": 4.0592, "step": 16985 }, { "epoch": 0.16986, "grad_norm": 1.3009971277751673, "learning_rate": 0.003, "loss": 4.0664, "step": 16986 }, { "epoch": 0.16987, "grad_norm": 1.1143033414580263, "learning_rate": 0.003, "loss": 4.0919, "step": 16987 }, { "epoch": 0.16988, "grad_norm": 1.5558709985350159, "learning_rate": 0.003, "loss": 4.065, "step": 16988 }, { "epoch": 0.16989, "grad_norm": 1.1190764199677528, "learning_rate": 0.003, "loss": 4.0489, "step": 16989 }, { "epoch": 0.1699, "grad_norm": 1.271684890657497, "learning_rate": 0.003, "loss": 4.0578, "step": 16990 }, { "epoch": 0.16991, "grad_norm": 1.1418945148210984, "learning_rate": 0.003, "loss": 4.0576, "step": 16991 }, { "epoch": 0.16992, "grad_norm": 1.3841242182659466, "learning_rate": 0.003, "loss": 4.0731, "step": 16992 }, { "epoch": 0.16993, "grad_norm": 0.9437527262546115, "learning_rate": 0.003, "loss": 4.0703, "step": 16993 }, { "epoch": 0.16994, "grad_norm": 1.2806800712476167, "learning_rate": 0.003, "loss": 4.0788, "step": 16994 }, { "epoch": 0.16995, "grad_norm": 1.0295146916247497, "learning_rate": 0.003, "loss": 4.0275, "step": 16995 }, { "epoch": 0.16996, "grad_norm": 1.367398593535846, "learning_rate": 0.003, "loss": 4.0516, "step": 16996 }, { "epoch": 0.16997, "grad_norm": 1.114513362033281, "learning_rate": 0.003, "loss": 4.0752, "step": 16997 }, { "epoch": 0.16998, "grad_norm": 1.6710717846457315, "learning_rate": 0.003, "loss": 4.0872, "step": 16998 }, { "epoch": 0.16999, "grad_norm": 1.08533786983493, "learning_rate": 0.003, "loss": 4.0794, "step": 16999 }, { "epoch": 0.17, "grad_norm": 1.1300516478506852, "learning_rate": 0.003, "loss": 4.0245, "step": 17000 }, { "epoch": 0.17001, "grad_norm": 1.4854644865056545, "learning_rate": 0.003, "loss": 4.0576, "step": 17001 }, { "epoch": 0.17002, "grad_norm": 1.0647905698786158, "learning_rate": 0.003, "loss": 4.074, "step": 17002 }, { "epoch": 0.17003, "grad_norm": 1.4684027361951262, "learning_rate": 0.003, "loss": 4.0343, "step": 17003 }, { "epoch": 0.17004, "grad_norm": 1.0702179810626333, "learning_rate": 0.003, "loss": 4.0615, "step": 17004 }, { "epoch": 0.17005, "grad_norm": 1.109196939589509, "learning_rate": 0.003, "loss": 4.0529, "step": 17005 }, { "epoch": 0.17006, "grad_norm": 1.1788556480619246, "learning_rate": 0.003, "loss": 4.0832, "step": 17006 }, { "epoch": 0.17007, "grad_norm": 1.1618337354719386, "learning_rate": 0.003, "loss": 4.0401, "step": 17007 }, { "epoch": 0.17008, "grad_norm": 1.4112447351543733, "learning_rate": 0.003, "loss": 4.09, "step": 17008 }, { "epoch": 0.17009, "grad_norm": 1.0210019746504395, "learning_rate": 0.003, "loss": 4.044, "step": 17009 }, { "epoch": 0.1701, "grad_norm": 1.5186241163420218, "learning_rate": 0.003, "loss": 4.0623, "step": 17010 }, { "epoch": 0.17011, "grad_norm": 0.9681540699367649, "learning_rate": 0.003, "loss": 4.0541, "step": 17011 }, { "epoch": 0.17012, "grad_norm": 1.347101933964949, "learning_rate": 0.003, "loss": 4.0778, "step": 17012 }, { "epoch": 0.17013, "grad_norm": 1.320421310437298, "learning_rate": 0.003, "loss": 4.0607, "step": 17013 }, { "epoch": 0.17014, "grad_norm": 1.2416515545235678, "learning_rate": 0.003, "loss": 4.0499, "step": 17014 }, { "epoch": 0.17015, "grad_norm": 1.2594764298389944, "learning_rate": 0.003, "loss": 4.0574, "step": 17015 }, { "epoch": 0.17016, "grad_norm": 1.0665417499892844, "learning_rate": 0.003, "loss": 4.0502, "step": 17016 }, { "epoch": 0.17017, "grad_norm": 1.2560262315788144, "learning_rate": 0.003, "loss": 4.0654, "step": 17017 }, { "epoch": 0.17018, "grad_norm": 1.1509264809125181, "learning_rate": 0.003, "loss": 4.0807, "step": 17018 }, { "epoch": 0.17019, "grad_norm": 1.2088345627948451, "learning_rate": 0.003, "loss": 4.0435, "step": 17019 }, { "epoch": 0.1702, "grad_norm": 1.411378690404628, "learning_rate": 0.003, "loss": 4.0684, "step": 17020 }, { "epoch": 0.17021, "grad_norm": 1.1801370251695185, "learning_rate": 0.003, "loss": 4.0514, "step": 17021 }, { "epoch": 0.17022, "grad_norm": 1.4776693771393323, "learning_rate": 0.003, "loss": 4.0554, "step": 17022 }, { "epoch": 0.17023, "grad_norm": 1.0551018726002896, "learning_rate": 0.003, "loss": 4.0647, "step": 17023 }, { "epoch": 0.17024, "grad_norm": 1.2469741530408414, "learning_rate": 0.003, "loss": 4.051, "step": 17024 }, { "epoch": 0.17025, "grad_norm": 1.1500908920402089, "learning_rate": 0.003, "loss": 4.0273, "step": 17025 }, { "epoch": 0.17026, "grad_norm": 1.201045732731781, "learning_rate": 0.003, "loss": 4.0322, "step": 17026 }, { "epoch": 0.17027, "grad_norm": 1.1096566430181591, "learning_rate": 0.003, "loss": 4.0417, "step": 17027 }, { "epoch": 0.17028, "grad_norm": 1.4834179558886718, "learning_rate": 0.003, "loss": 4.0588, "step": 17028 }, { "epoch": 0.17029, "grad_norm": 1.1165586509695928, "learning_rate": 0.003, "loss": 4.0819, "step": 17029 }, { "epoch": 0.1703, "grad_norm": 1.274416067501693, "learning_rate": 0.003, "loss": 4.0498, "step": 17030 }, { "epoch": 0.17031, "grad_norm": 1.056483448533938, "learning_rate": 0.003, "loss": 4.0719, "step": 17031 }, { "epoch": 0.17032, "grad_norm": 1.6005804495978606, "learning_rate": 0.003, "loss": 4.0595, "step": 17032 }, { "epoch": 0.17033, "grad_norm": 0.860041113410521, "learning_rate": 0.003, "loss": 4.0497, "step": 17033 }, { "epoch": 0.17034, "grad_norm": 1.0279815590719537, "learning_rate": 0.003, "loss": 4.0476, "step": 17034 }, { "epoch": 0.17035, "grad_norm": 1.666759738599301, "learning_rate": 0.003, "loss": 4.0434, "step": 17035 }, { "epoch": 0.17036, "grad_norm": 1.0416404861330888, "learning_rate": 0.003, "loss": 4.0403, "step": 17036 }, { "epoch": 0.17037, "grad_norm": 1.336235499498572, "learning_rate": 0.003, "loss": 4.0558, "step": 17037 }, { "epoch": 0.17038, "grad_norm": 1.1070780681384462, "learning_rate": 0.003, "loss": 4.056, "step": 17038 }, { "epoch": 0.17039, "grad_norm": 1.1474080014229298, "learning_rate": 0.003, "loss": 4.0461, "step": 17039 }, { "epoch": 0.1704, "grad_norm": 1.2356096392683003, "learning_rate": 0.003, "loss": 4.0844, "step": 17040 }, { "epoch": 0.17041, "grad_norm": 1.2116338646318794, "learning_rate": 0.003, "loss": 4.0648, "step": 17041 }, { "epoch": 0.17042, "grad_norm": 1.3185251194346252, "learning_rate": 0.003, "loss": 4.0481, "step": 17042 }, { "epoch": 0.17043, "grad_norm": 1.1305643242555485, "learning_rate": 0.003, "loss": 4.0451, "step": 17043 }, { "epoch": 0.17044, "grad_norm": 1.4747500406152476, "learning_rate": 0.003, "loss": 4.0796, "step": 17044 }, { "epoch": 0.17045, "grad_norm": 0.9367047255629887, "learning_rate": 0.003, "loss": 4.0662, "step": 17045 }, { "epoch": 0.17046, "grad_norm": 1.4376927861567141, "learning_rate": 0.003, "loss": 4.0811, "step": 17046 }, { "epoch": 0.17047, "grad_norm": 1.1354376573084526, "learning_rate": 0.003, "loss": 4.0674, "step": 17047 }, { "epoch": 0.17048, "grad_norm": 1.2809581279291067, "learning_rate": 0.003, "loss": 4.0885, "step": 17048 }, { "epoch": 0.17049, "grad_norm": 1.1060134587814072, "learning_rate": 0.003, "loss": 4.0321, "step": 17049 }, { "epoch": 0.1705, "grad_norm": 1.3903949900172548, "learning_rate": 0.003, "loss": 4.0569, "step": 17050 }, { "epoch": 0.17051, "grad_norm": 1.2102423319200646, "learning_rate": 0.003, "loss": 4.0502, "step": 17051 }, { "epoch": 0.17052, "grad_norm": 1.392974983130736, "learning_rate": 0.003, "loss": 4.0844, "step": 17052 }, { "epoch": 0.17053, "grad_norm": 1.2761802143619123, "learning_rate": 0.003, "loss": 4.0734, "step": 17053 }, { "epoch": 0.17054, "grad_norm": 1.340852438654309, "learning_rate": 0.003, "loss": 4.098, "step": 17054 }, { "epoch": 0.17055, "grad_norm": 1.1047860580331121, "learning_rate": 0.003, "loss": 4.0766, "step": 17055 }, { "epoch": 0.17056, "grad_norm": 1.2925607206559326, "learning_rate": 0.003, "loss": 4.06, "step": 17056 }, { "epoch": 0.17057, "grad_norm": 1.2118379864560753, "learning_rate": 0.003, "loss": 4.0404, "step": 17057 }, { "epoch": 0.17058, "grad_norm": 1.2102751990134006, "learning_rate": 0.003, "loss": 4.0554, "step": 17058 }, { "epoch": 0.17059, "grad_norm": 1.2238964106513677, "learning_rate": 0.003, "loss": 4.0314, "step": 17059 }, { "epoch": 0.1706, "grad_norm": 1.2902083032121994, "learning_rate": 0.003, "loss": 4.0603, "step": 17060 }, { "epoch": 0.17061, "grad_norm": 0.9884657706767966, "learning_rate": 0.003, "loss": 4.0693, "step": 17061 }, { "epoch": 0.17062, "grad_norm": 1.362392521026528, "learning_rate": 0.003, "loss": 4.0236, "step": 17062 }, { "epoch": 0.17063, "grad_norm": 1.033157284902731, "learning_rate": 0.003, "loss": 4.0455, "step": 17063 }, { "epoch": 0.17064, "grad_norm": 1.3522412439083653, "learning_rate": 0.003, "loss": 4.0645, "step": 17064 }, { "epoch": 0.17065, "grad_norm": 1.1503250545725914, "learning_rate": 0.003, "loss": 4.0629, "step": 17065 }, { "epoch": 0.17066, "grad_norm": 1.2561473842039022, "learning_rate": 0.003, "loss": 4.0525, "step": 17066 }, { "epoch": 0.17067, "grad_norm": 1.4375575271359797, "learning_rate": 0.003, "loss": 4.078, "step": 17067 }, { "epoch": 0.17068, "grad_norm": 1.024327865973433, "learning_rate": 0.003, "loss": 4.0736, "step": 17068 }, { "epoch": 0.17069, "grad_norm": 1.4739457588751657, "learning_rate": 0.003, "loss": 4.0837, "step": 17069 }, { "epoch": 0.1707, "grad_norm": 1.042496759524706, "learning_rate": 0.003, "loss": 4.066, "step": 17070 }, { "epoch": 0.17071, "grad_norm": 1.2787623283171865, "learning_rate": 0.003, "loss": 4.047, "step": 17071 }, { "epoch": 0.17072, "grad_norm": 1.2739738294358849, "learning_rate": 0.003, "loss": 4.0521, "step": 17072 }, { "epoch": 0.17073, "grad_norm": 1.3069607361903082, "learning_rate": 0.003, "loss": 4.0689, "step": 17073 }, { "epoch": 0.17074, "grad_norm": 1.1109209840193457, "learning_rate": 0.003, "loss": 4.0651, "step": 17074 }, { "epoch": 0.17075, "grad_norm": 1.3578483278539972, "learning_rate": 0.003, "loss": 4.0601, "step": 17075 }, { "epoch": 0.17076, "grad_norm": 1.2057365219041163, "learning_rate": 0.003, "loss": 4.0374, "step": 17076 }, { "epoch": 0.17077, "grad_norm": 1.1902615298987598, "learning_rate": 0.003, "loss": 4.0726, "step": 17077 }, { "epoch": 0.17078, "grad_norm": 1.0977206185792603, "learning_rate": 0.003, "loss": 4.0517, "step": 17078 }, { "epoch": 0.17079, "grad_norm": 1.2100246532932626, "learning_rate": 0.003, "loss": 4.0556, "step": 17079 }, { "epoch": 0.1708, "grad_norm": 1.1747207972823042, "learning_rate": 0.003, "loss": 4.0434, "step": 17080 }, { "epoch": 0.17081, "grad_norm": 1.4256099582840815, "learning_rate": 0.003, "loss": 4.0503, "step": 17081 }, { "epoch": 0.17082, "grad_norm": 1.0428864359588268, "learning_rate": 0.003, "loss": 4.0914, "step": 17082 }, { "epoch": 0.17083, "grad_norm": 1.4943451862253978, "learning_rate": 0.003, "loss": 4.0661, "step": 17083 }, { "epoch": 0.17084, "grad_norm": 1.119894011493226, "learning_rate": 0.003, "loss": 4.0661, "step": 17084 }, { "epoch": 0.17085, "grad_norm": 1.4031043390497535, "learning_rate": 0.003, "loss": 4.0719, "step": 17085 }, { "epoch": 0.17086, "grad_norm": 1.1318173853206333, "learning_rate": 0.003, "loss": 4.0495, "step": 17086 }, { "epoch": 0.17087, "grad_norm": 1.627373600531823, "learning_rate": 0.003, "loss": 4.06, "step": 17087 }, { "epoch": 0.17088, "grad_norm": 1.2141086452195955, "learning_rate": 0.003, "loss": 4.0674, "step": 17088 }, { "epoch": 0.17089, "grad_norm": 1.2285668262442404, "learning_rate": 0.003, "loss": 4.0463, "step": 17089 }, { "epoch": 0.1709, "grad_norm": 1.074949899276511, "learning_rate": 0.003, "loss": 4.0494, "step": 17090 }, { "epoch": 0.17091, "grad_norm": 1.4761983475605394, "learning_rate": 0.003, "loss": 4.0892, "step": 17091 }, { "epoch": 0.17092, "grad_norm": 1.0780631791435658, "learning_rate": 0.003, "loss": 4.0309, "step": 17092 }, { "epoch": 0.17093, "grad_norm": 1.6138046600098406, "learning_rate": 0.003, "loss": 4.0581, "step": 17093 }, { "epoch": 0.17094, "grad_norm": 0.9592767113879691, "learning_rate": 0.003, "loss": 4.0686, "step": 17094 }, { "epoch": 0.17095, "grad_norm": 1.267181289397601, "learning_rate": 0.003, "loss": 4.0388, "step": 17095 }, { "epoch": 0.17096, "grad_norm": 1.1983977944981508, "learning_rate": 0.003, "loss": 4.0537, "step": 17096 }, { "epoch": 0.17097, "grad_norm": 1.7081677681732466, "learning_rate": 0.003, "loss": 4.0483, "step": 17097 }, { "epoch": 0.17098, "grad_norm": 1.0094432444386825, "learning_rate": 0.003, "loss": 4.0818, "step": 17098 }, { "epoch": 0.17099, "grad_norm": 1.2879564300679882, "learning_rate": 0.003, "loss": 4.0321, "step": 17099 }, { "epoch": 0.171, "grad_norm": 1.2241451007568611, "learning_rate": 0.003, "loss": 4.0562, "step": 17100 }, { "epoch": 0.17101, "grad_norm": 1.2005359846033563, "learning_rate": 0.003, "loss": 4.0465, "step": 17101 }, { "epoch": 0.17102, "grad_norm": 1.2775872159975104, "learning_rate": 0.003, "loss": 4.0517, "step": 17102 }, { "epoch": 0.17103, "grad_norm": 1.1794188282881768, "learning_rate": 0.003, "loss": 4.0619, "step": 17103 }, { "epoch": 0.17104, "grad_norm": 1.1650774374183848, "learning_rate": 0.003, "loss": 4.0543, "step": 17104 }, { "epoch": 0.17105, "grad_norm": 1.2195995474646042, "learning_rate": 0.003, "loss": 4.0392, "step": 17105 }, { "epoch": 0.17106, "grad_norm": 1.2893412901033197, "learning_rate": 0.003, "loss": 4.0629, "step": 17106 }, { "epoch": 0.17107, "grad_norm": 1.319909128826875, "learning_rate": 0.003, "loss": 4.0836, "step": 17107 }, { "epoch": 0.17108, "grad_norm": 1.1690540591132834, "learning_rate": 0.003, "loss": 4.06, "step": 17108 }, { "epoch": 0.17109, "grad_norm": 1.2115142944406738, "learning_rate": 0.003, "loss": 4.0528, "step": 17109 }, { "epoch": 0.1711, "grad_norm": 1.2627038138659699, "learning_rate": 0.003, "loss": 4.0623, "step": 17110 }, { "epoch": 0.17111, "grad_norm": 1.6278553032524519, "learning_rate": 0.003, "loss": 4.0526, "step": 17111 }, { "epoch": 0.17112, "grad_norm": 1.0070548525930745, "learning_rate": 0.003, "loss": 4.0403, "step": 17112 }, { "epoch": 0.17113, "grad_norm": 1.1407931098727693, "learning_rate": 0.003, "loss": 4.0393, "step": 17113 }, { "epoch": 0.17114, "grad_norm": 1.5466415497004846, "learning_rate": 0.003, "loss": 4.0589, "step": 17114 }, { "epoch": 0.17115, "grad_norm": 1.3913316031898841, "learning_rate": 0.003, "loss": 4.0611, "step": 17115 }, { "epoch": 0.17116, "grad_norm": 1.1203436523118306, "learning_rate": 0.003, "loss": 4.0522, "step": 17116 }, { "epoch": 0.17117, "grad_norm": 1.137359047504489, "learning_rate": 0.003, "loss": 4.0437, "step": 17117 }, { "epoch": 0.17118, "grad_norm": 1.2331556996086932, "learning_rate": 0.003, "loss": 4.0552, "step": 17118 }, { "epoch": 0.17119, "grad_norm": 1.269085644507956, "learning_rate": 0.003, "loss": 4.0664, "step": 17119 }, { "epoch": 0.1712, "grad_norm": 1.1478520730090644, "learning_rate": 0.003, "loss": 4.0557, "step": 17120 }, { "epoch": 0.17121, "grad_norm": 1.096765511893826, "learning_rate": 0.003, "loss": 4.0548, "step": 17121 }, { "epoch": 0.17122, "grad_norm": 1.394911999429465, "learning_rate": 0.003, "loss": 4.0497, "step": 17122 }, { "epoch": 0.17123, "grad_norm": 1.1459402376211092, "learning_rate": 0.003, "loss": 4.0451, "step": 17123 }, { "epoch": 0.17124, "grad_norm": 1.3012976253681308, "learning_rate": 0.003, "loss": 4.0177, "step": 17124 }, { "epoch": 0.17125, "grad_norm": 1.1823411170357148, "learning_rate": 0.003, "loss": 4.088, "step": 17125 }, { "epoch": 0.17126, "grad_norm": 1.3339255381643982, "learning_rate": 0.003, "loss": 4.0471, "step": 17126 }, { "epoch": 0.17127, "grad_norm": 1.114045506324468, "learning_rate": 0.003, "loss": 4.0594, "step": 17127 }, { "epoch": 0.17128, "grad_norm": 1.2408804559454076, "learning_rate": 0.003, "loss": 4.0595, "step": 17128 }, { "epoch": 0.17129, "grad_norm": 1.1429967648203054, "learning_rate": 0.003, "loss": 4.0353, "step": 17129 }, { "epoch": 0.1713, "grad_norm": 1.1435463857854034, "learning_rate": 0.003, "loss": 4.0475, "step": 17130 }, { "epoch": 0.17131, "grad_norm": 1.1192740481677632, "learning_rate": 0.003, "loss": 4.0391, "step": 17131 }, { "epoch": 0.17132, "grad_norm": 1.1649496145325733, "learning_rate": 0.003, "loss": 4.0459, "step": 17132 }, { "epoch": 0.17133, "grad_norm": 1.1653298575255848, "learning_rate": 0.003, "loss": 4.0863, "step": 17133 }, { "epoch": 0.17134, "grad_norm": 1.2927260128753215, "learning_rate": 0.003, "loss": 4.0693, "step": 17134 }, { "epoch": 0.17135, "grad_norm": 1.1052873686354054, "learning_rate": 0.003, "loss": 4.0593, "step": 17135 }, { "epoch": 0.17136, "grad_norm": 1.4240461447369932, "learning_rate": 0.003, "loss": 4.0476, "step": 17136 }, { "epoch": 0.17137, "grad_norm": 1.0025371795503932, "learning_rate": 0.003, "loss": 4.0566, "step": 17137 }, { "epoch": 0.17138, "grad_norm": 1.2783951799923268, "learning_rate": 0.003, "loss": 4.0525, "step": 17138 }, { "epoch": 0.17139, "grad_norm": 1.2915244743297567, "learning_rate": 0.003, "loss": 4.0645, "step": 17139 }, { "epoch": 0.1714, "grad_norm": 1.2239294795654907, "learning_rate": 0.003, "loss": 4.0838, "step": 17140 }, { "epoch": 0.17141, "grad_norm": 1.287674278636702, "learning_rate": 0.003, "loss": 4.0494, "step": 17141 }, { "epoch": 0.17142, "grad_norm": 1.2329873107473168, "learning_rate": 0.003, "loss": 4.0258, "step": 17142 }, { "epoch": 0.17143, "grad_norm": 1.1853654198506531, "learning_rate": 0.003, "loss": 4.0486, "step": 17143 }, { "epoch": 0.17144, "grad_norm": 1.6144709209863437, "learning_rate": 0.003, "loss": 4.0553, "step": 17144 }, { "epoch": 0.17145, "grad_norm": 1.0842925800239687, "learning_rate": 0.003, "loss": 4.0353, "step": 17145 }, { "epoch": 0.17146, "grad_norm": 1.4143467272752115, "learning_rate": 0.003, "loss": 4.0598, "step": 17146 }, { "epoch": 0.17147, "grad_norm": 0.836879174133386, "learning_rate": 0.003, "loss": 4.0323, "step": 17147 }, { "epoch": 0.17148, "grad_norm": 1.1927653211616587, "learning_rate": 0.003, "loss": 4.0621, "step": 17148 }, { "epoch": 0.17149, "grad_norm": 1.1736106709041612, "learning_rate": 0.003, "loss": 4.0572, "step": 17149 }, { "epoch": 0.1715, "grad_norm": 1.4105477692900008, "learning_rate": 0.003, "loss": 4.0526, "step": 17150 }, { "epoch": 0.17151, "grad_norm": 1.2359755066435034, "learning_rate": 0.003, "loss": 4.0632, "step": 17151 }, { "epoch": 0.17152, "grad_norm": 1.4072251037006216, "learning_rate": 0.003, "loss": 4.0459, "step": 17152 }, { "epoch": 0.17153, "grad_norm": 1.2121102719247743, "learning_rate": 0.003, "loss": 4.0509, "step": 17153 }, { "epoch": 0.17154, "grad_norm": 1.0708423088931183, "learning_rate": 0.003, "loss": 4.0473, "step": 17154 }, { "epoch": 0.17155, "grad_norm": 1.4164021755321157, "learning_rate": 0.003, "loss": 4.0552, "step": 17155 }, { "epoch": 0.17156, "grad_norm": 1.0757005546234806, "learning_rate": 0.003, "loss": 4.0799, "step": 17156 }, { "epoch": 0.17157, "grad_norm": 1.4312245312473366, "learning_rate": 0.003, "loss": 4.0652, "step": 17157 }, { "epoch": 0.17158, "grad_norm": 1.0794673781354265, "learning_rate": 0.003, "loss": 4.0329, "step": 17158 }, { "epoch": 0.17159, "grad_norm": 1.415711859144589, "learning_rate": 0.003, "loss": 4.078, "step": 17159 }, { "epoch": 0.1716, "grad_norm": 1.062912828210717, "learning_rate": 0.003, "loss": 4.0497, "step": 17160 }, { "epoch": 0.17161, "grad_norm": 1.4631637785064096, "learning_rate": 0.003, "loss": 4.0943, "step": 17161 }, { "epoch": 0.17162, "grad_norm": 1.0663778939242714, "learning_rate": 0.003, "loss": 4.0881, "step": 17162 }, { "epoch": 0.17163, "grad_norm": 1.6987545202361207, "learning_rate": 0.003, "loss": 4.0428, "step": 17163 }, { "epoch": 0.17164, "grad_norm": 1.0323738877625745, "learning_rate": 0.003, "loss": 4.0598, "step": 17164 }, { "epoch": 0.17165, "grad_norm": 1.3288184898116433, "learning_rate": 0.003, "loss": 4.0681, "step": 17165 }, { "epoch": 0.17166, "grad_norm": 1.1158937376781226, "learning_rate": 0.003, "loss": 4.0699, "step": 17166 }, { "epoch": 0.17167, "grad_norm": 1.26856294914553, "learning_rate": 0.003, "loss": 4.0688, "step": 17167 }, { "epoch": 0.17168, "grad_norm": 1.2245825834066786, "learning_rate": 0.003, "loss": 4.0448, "step": 17168 }, { "epoch": 0.17169, "grad_norm": 1.2851830135787832, "learning_rate": 0.003, "loss": 4.0855, "step": 17169 }, { "epoch": 0.1717, "grad_norm": 0.9162180990378845, "learning_rate": 0.003, "loss": 4.0624, "step": 17170 }, { "epoch": 0.17171, "grad_norm": 1.2643111081821041, "learning_rate": 0.003, "loss": 4.0668, "step": 17171 }, { "epoch": 0.17172, "grad_norm": 1.3660061580806508, "learning_rate": 0.003, "loss": 4.0659, "step": 17172 }, { "epoch": 0.17173, "grad_norm": 1.4266468625355122, "learning_rate": 0.003, "loss": 4.0316, "step": 17173 }, { "epoch": 0.17174, "grad_norm": 1.3602724508990756, "learning_rate": 0.003, "loss": 4.0824, "step": 17174 }, { "epoch": 0.17175, "grad_norm": 0.9471462822131239, "learning_rate": 0.003, "loss": 4.0598, "step": 17175 }, { "epoch": 0.17176, "grad_norm": 1.2282409267406897, "learning_rate": 0.003, "loss": 4.0566, "step": 17176 }, { "epoch": 0.17177, "grad_norm": 1.3409203348563423, "learning_rate": 0.003, "loss": 4.0374, "step": 17177 }, { "epoch": 0.17178, "grad_norm": 1.2178308315203064, "learning_rate": 0.003, "loss": 4.0596, "step": 17178 }, { "epoch": 0.17179, "grad_norm": 1.0951387805996513, "learning_rate": 0.003, "loss": 4.0462, "step": 17179 }, { "epoch": 0.1718, "grad_norm": 1.2908997990033155, "learning_rate": 0.003, "loss": 4.0795, "step": 17180 }, { "epoch": 0.17181, "grad_norm": 1.1445767219936012, "learning_rate": 0.003, "loss": 4.0467, "step": 17181 }, { "epoch": 0.17182, "grad_norm": 1.1380557271009788, "learning_rate": 0.003, "loss": 4.0312, "step": 17182 }, { "epoch": 0.17183, "grad_norm": 1.275095676428426, "learning_rate": 0.003, "loss": 4.0614, "step": 17183 }, { "epoch": 0.17184, "grad_norm": 1.2225802772918613, "learning_rate": 0.003, "loss": 4.0681, "step": 17184 }, { "epoch": 0.17185, "grad_norm": 1.6827976521471713, "learning_rate": 0.003, "loss": 4.0409, "step": 17185 }, { "epoch": 0.17186, "grad_norm": 1.120126198618937, "learning_rate": 0.003, "loss": 4.054, "step": 17186 }, { "epoch": 0.17187, "grad_norm": 1.2685479238932016, "learning_rate": 0.003, "loss": 4.0751, "step": 17187 }, { "epoch": 0.17188, "grad_norm": 1.2273429494321608, "learning_rate": 0.003, "loss": 4.0837, "step": 17188 }, { "epoch": 0.17189, "grad_norm": 1.175547919825017, "learning_rate": 0.003, "loss": 4.0359, "step": 17189 }, { "epoch": 0.1719, "grad_norm": 1.4270859354766496, "learning_rate": 0.003, "loss": 4.0517, "step": 17190 }, { "epoch": 0.17191, "grad_norm": 1.21603731282858, "learning_rate": 0.003, "loss": 4.0873, "step": 17191 }, { "epoch": 0.17192, "grad_norm": 1.1269431238317735, "learning_rate": 0.003, "loss": 4.0564, "step": 17192 }, { "epoch": 0.17193, "grad_norm": 1.0547513507423332, "learning_rate": 0.003, "loss": 4.0418, "step": 17193 }, { "epoch": 0.17194, "grad_norm": 1.3598685088552334, "learning_rate": 0.003, "loss": 4.0454, "step": 17194 }, { "epoch": 0.17195, "grad_norm": 1.1824352169266816, "learning_rate": 0.003, "loss": 4.0465, "step": 17195 }, { "epoch": 0.17196, "grad_norm": 1.293936300977291, "learning_rate": 0.003, "loss": 4.038, "step": 17196 }, { "epoch": 0.17197, "grad_norm": 1.3017028808489668, "learning_rate": 0.003, "loss": 4.0493, "step": 17197 }, { "epoch": 0.17198, "grad_norm": 1.2113699499764334, "learning_rate": 0.003, "loss": 4.1044, "step": 17198 }, { "epoch": 0.17199, "grad_norm": 1.3628410320018296, "learning_rate": 0.003, "loss": 4.0664, "step": 17199 }, { "epoch": 0.172, "grad_norm": 1.1963881630514184, "learning_rate": 0.003, "loss": 4.0691, "step": 17200 }, { "epoch": 0.17201, "grad_norm": 1.2352001983736032, "learning_rate": 0.003, "loss": 4.0488, "step": 17201 }, { "epoch": 0.17202, "grad_norm": 1.2540179225481027, "learning_rate": 0.003, "loss": 4.0383, "step": 17202 }, { "epoch": 0.17203, "grad_norm": 1.1253322468332425, "learning_rate": 0.003, "loss": 4.0612, "step": 17203 }, { "epoch": 0.17204, "grad_norm": 1.2995373145495206, "learning_rate": 0.003, "loss": 4.0457, "step": 17204 }, { "epoch": 0.17205, "grad_norm": 1.1866529753856296, "learning_rate": 0.003, "loss": 4.0487, "step": 17205 }, { "epoch": 0.17206, "grad_norm": 1.268599132845124, "learning_rate": 0.003, "loss": 4.0692, "step": 17206 }, { "epoch": 0.17207, "grad_norm": 1.2904055430536574, "learning_rate": 0.003, "loss": 4.0683, "step": 17207 }, { "epoch": 0.17208, "grad_norm": 1.3318610535485012, "learning_rate": 0.003, "loss": 4.0648, "step": 17208 }, { "epoch": 0.17209, "grad_norm": 1.2517193849636388, "learning_rate": 0.003, "loss": 4.0513, "step": 17209 }, { "epoch": 0.1721, "grad_norm": 1.359154555885252, "learning_rate": 0.003, "loss": 4.0295, "step": 17210 }, { "epoch": 0.17211, "grad_norm": 1.153342960723365, "learning_rate": 0.003, "loss": 4.0408, "step": 17211 }, { "epoch": 0.17212, "grad_norm": 1.2571055996638105, "learning_rate": 0.003, "loss": 4.0522, "step": 17212 }, { "epoch": 0.17213, "grad_norm": 0.9998084123979417, "learning_rate": 0.003, "loss": 4.0693, "step": 17213 }, { "epoch": 0.17214, "grad_norm": 1.2799901188175968, "learning_rate": 0.003, "loss": 4.0537, "step": 17214 }, { "epoch": 0.17215, "grad_norm": 0.9469252917559144, "learning_rate": 0.003, "loss": 4.0416, "step": 17215 }, { "epoch": 0.17216, "grad_norm": 1.4177754942789411, "learning_rate": 0.003, "loss": 4.0391, "step": 17216 }, { "epoch": 0.17217, "grad_norm": 1.280720785201341, "learning_rate": 0.003, "loss": 4.018, "step": 17217 }, { "epoch": 0.17218, "grad_norm": 1.1338877661442035, "learning_rate": 0.003, "loss": 4.0236, "step": 17218 }, { "epoch": 0.17219, "grad_norm": 1.4341819613445428, "learning_rate": 0.003, "loss": 4.0346, "step": 17219 }, { "epoch": 0.1722, "grad_norm": 0.8765015139996959, "learning_rate": 0.003, "loss": 4.0178, "step": 17220 }, { "epoch": 0.17221, "grad_norm": 1.1053439625899335, "learning_rate": 0.003, "loss": 4.0544, "step": 17221 }, { "epoch": 0.17222, "grad_norm": 1.4071027750109097, "learning_rate": 0.003, "loss": 4.0655, "step": 17222 }, { "epoch": 0.17223, "grad_norm": 1.0695360583694458, "learning_rate": 0.003, "loss": 4.0162, "step": 17223 }, { "epoch": 0.17224, "grad_norm": 1.4011965510898687, "learning_rate": 0.003, "loss": 4.0849, "step": 17224 }, { "epoch": 0.17225, "grad_norm": 1.179663336846994, "learning_rate": 0.003, "loss": 4.055, "step": 17225 }, { "epoch": 0.17226, "grad_norm": 1.2039026930037622, "learning_rate": 0.003, "loss": 4.0801, "step": 17226 }, { "epoch": 0.17227, "grad_norm": 1.2446959723832283, "learning_rate": 0.003, "loss": 4.0545, "step": 17227 }, { "epoch": 0.17228, "grad_norm": 1.2130664411452436, "learning_rate": 0.003, "loss": 4.0462, "step": 17228 }, { "epoch": 0.17229, "grad_norm": 1.279387723205202, "learning_rate": 0.003, "loss": 4.0699, "step": 17229 }, { "epoch": 0.1723, "grad_norm": 1.2646549830766647, "learning_rate": 0.003, "loss": 4.0649, "step": 17230 }, { "epoch": 0.17231, "grad_norm": 1.1476763293666046, "learning_rate": 0.003, "loss": 4.0437, "step": 17231 }, { "epoch": 0.17232, "grad_norm": 1.2667751820120776, "learning_rate": 0.003, "loss": 4.0537, "step": 17232 }, { "epoch": 0.17233, "grad_norm": 1.33390482237294, "learning_rate": 0.003, "loss": 4.0485, "step": 17233 }, { "epoch": 0.17234, "grad_norm": 1.2556664716742125, "learning_rate": 0.003, "loss": 4.039, "step": 17234 }, { "epoch": 0.17235, "grad_norm": 1.1846048324078389, "learning_rate": 0.003, "loss": 4.0248, "step": 17235 }, { "epoch": 0.17236, "grad_norm": 1.0903822402335819, "learning_rate": 0.003, "loss": 4.0834, "step": 17236 }, { "epoch": 0.17237, "grad_norm": 1.331981127769312, "learning_rate": 0.003, "loss": 4.0502, "step": 17237 }, { "epoch": 0.17238, "grad_norm": 1.1383088320999761, "learning_rate": 0.003, "loss": 4.0862, "step": 17238 }, { "epoch": 0.17239, "grad_norm": 1.3565717249246343, "learning_rate": 0.003, "loss": 4.0652, "step": 17239 }, { "epoch": 0.1724, "grad_norm": 1.0118820735186458, "learning_rate": 0.003, "loss": 4.0016, "step": 17240 }, { "epoch": 0.17241, "grad_norm": 1.3922275482107882, "learning_rate": 0.003, "loss": 4.0819, "step": 17241 }, { "epoch": 0.17242, "grad_norm": 1.1687677777125651, "learning_rate": 0.003, "loss": 4.0546, "step": 17242 }, { "epoch": 0.17243, "grad_norm": 1.2327638917695471, "learning_rate": 0.003, "loss": 4.0519, "step": 17243 }, { "epoch": 0.17244, "grad_norm": 1.339453077024193, "learning_rate": 0.003, "loss": 4.0994, "step": 17244 }, { "epoch": 0.17245, "grad_norm": 1.5186126884766142, "learning_rate": 0.003, "loss": 4.0788, "step": 17245 }, { "epoch": 0.17246, "grad_norm": 1.05824978685617, "learning_rate": 0.003, "loss": 4.0642, "step": 17246 }, { "epoch": 0.17247, "grad_norm": 1.255995773051049, "learning_rate": 0.003, "loss": 4.0605, "step": 17247 }, { "epoch": 0.17248, "grad_norm": 1.2315503113323412, "learning_rate": 0.003, "loss": 4.0823, "step": 17248 }, { "epoch": 0.17249, "grad_norm": 0.9318070395997096, "learning_rate": 0.003, "loss": 4.0366, "step": 17249 }, { "epoch": 0.1725, "grad_norm": 1.1908491186156405, "learning_rate": 0.003, "loss": 4.0814, "step": 17250 }, { "epoch": 0.17251, "grad_norm": 1.1915153514630226, "learning_rate": 0.003, "loss": 4.0527, "step": 17251 }, { "epoch": 0.17252, "grad_norm": 1.1563311259365638, "learning_rate": 0.003, "loss": 4.0777, "step": 17252 }, { "epoch": 0.17253, "grad_norm": 1.270705816063968, "learning_rate": 0.003, "loss": 4.0477, "step": 17253 }, { "epoch": 0.17254, "grad_norm": 1.0384070607139333, "learning_rate": 0.003, "loss": 4.0258, "step": 17254 }, { "epoch": 0.17255, "grad_norm": 1.4061055380075593, "learning_rate": 0.003, "loss": 4.0578, "step": 17255 }, { "epoch": 0.17256, "grad_norm": 1.3552131673979657, "learning_rate": 0.003, "loss": 4.059, "step": 17256 }, { "epoch": 0.17257, "grad_norm": 1.3011398340871603, "learning_rate": 0.003, "loss": 4.0482, "step": 17257 }, { "epoch": 0.17258, "grad_norm": 1.1959542406063346, "learning_rate": 0.003, "loss": 4.0573, "step": 17258 }, { "epoch": 0.17259, "grad_norm": 1.2030988997548404, "learning_rate": 0.003, "loss": 4.0646, "step": 17259 }, { "epoch": 0.1726, "grad_norm": 1.3521643865300224, "learning_rate": 0.003, "loss": 4.0471, "step": 17260 }, { "epoch": 0.17261, "grad_norm": 1.1479353012867946, "learning_rate": 0.003, "loss": 4.0531, "step": 17261 }, { "epoch": 0.17262, "grad_norm": 1.4021285788014592, "learning_rate": 0.003, "loss": 4.0523, "step": 17262 }, { "epoch": 0.17263, "grad_norm": 1.097132875295684, "learning_rate": 0.003, "loss": 4.0649, "step": 17263 }, { "epoch": 0.17264, "grad_norm": 1.468951819612964, "learning_rate": 0.003, "loss": 4.0446, "step": 17264 }, { "epoch": 0.17265, "grad_norm": 1.173350174634137, "learning_rate": 0.003, "loss": 4.0323, "step": 17265 }, { "epoch": 0.17266, "grad_norm": 1.3197111849349104, "learning_rate": 0.003, "loss": 4.0409, "step": 17266 }, { "epoch": 0.17267, "grad_norm": 1.1182822292349939, "learning_rate": 0.003, "loss": 4.0556, "step": 17267 }, { "epoch": 0.17268, "grad_norm": 1.408288619100258, "learning_rate": 0.003, "loss": 4.0743, "step": 17268 }, { "epoch": 0.17269, "grad_norm": 1.0378872949043103, "learning_rate": 0.003, "loss": 4.0497, "step": 17269 }, { "epoch": 0.1727, "grad_norm": 1.323592295870236, "learning_rate": 0.003, "loss": 4.0487, "step": 17270 }, { "epoch": 0.17271, "grad_norm": 1.1529028314777992, "learning_rate": 0.003, "loss": 4.0782, "step": 17271 }, { "epoch": 0.17272, "grad_norm": 1.3556256576605632, "learning_rate": 0.003, "loss": 4.0503, "step": 17272 }, { "epoch": 0.17273, "grad_norm": 1.3919913794708458, "learning_rate": 0.003, "loss": 4.0479, "step": 17273 }, { "epoch": 0.17274, "grad_norm": 1.0873879746321833, "learning_rate": 0.003, "loss": 4.065, "step": 17274 }, { "epoch": 0.17275, "grad_norm": 1.216810707377631, "learning_rate": 0.003, "loss": 4.0286, "step": 17275 }, { "epoch": 0.17276, "grad_norm": 1.0698529071889722, "learning_rate": 0.003, "loss": 4.0533, "step": 17276 }, { "epoch": 0.17277, "grad_norm": 1.350392714923593, "learning_rate": 0.003, "loss": 4.0646, "step": 17277 }, { "epoch": 0.17278, "grad_norm": 1.104244064909029, "learning_rate": 0.003, "loss": 4.0446, "step": 17278 }, { "epoch": 0.17279, "grad_norm": 1.2353810154030969, "learning_rate": 0.003, "loss": 4.0526, "step": 17279 }, { "epoch": 0.1728, "grad_norm": 1.4745630962674738, "learning_rate": 0.003, "loss": 4.0392, "step": 17280 }, { "epoch": 0.17281, "grad_norm": 1.1006183337244388, "learning_rate": 0.003, "loss": 4.0105, "step": 17281 }, { "epoch": 0.17282, "grad_norm": 1.1945203534698534, "learning_rate": 0.003, "loss": 4.0395, "step": 17282 }, { "epoch": 0.17283, "grad_norm": 1.1795349161011404, "learning_rate": 0.003, "loss": 4.0619, "step": 17283 }, { "epoch": 0.17284, "grad_norm": 1.1510241962711578, "learning_rate": 0.003, "loss": 4.05, "step": 17284 }, { "epoch": 0.17285, "grad_norm": 1.2127775054776724, "learning_rate": 0.003, "loss": 4.0714, "step": 17285 }, { "epoch": 0.17286, "grad_norm": 1.1555907178306937, "learning_rate": 0.003, "loss": 4.0584, "step": 17286 }, { "epoch": 0.17287, "grad_norm": 1.2965122071414634, "learning_rate": 0.003, "loss": 4.052, "step": 17287 }, { "epoch": 0.17288, "grad_norm": 0.99620776378164, "learning_rate": 0.003, "loss": 4.0412, "step": 17288 }, { "epoch": 0.17289, "grad_norm": 1.4769229074340224, "learning_rate": 0.003, "loss": 4.0666, "step": 17289 }, { "epoch": 0.1729, "grad_norm": 1.0784906715158313, "learning_rate": 0.003, "loss": 4.0583, "step": 17290 }, { "epoch": 0.17291, "grad_norm": 1.3670448482541506, "learning_rate": 0.003, "loss": 4.0568, "step": 17291 }, { "epoch": 0.17292, "grad_norm": 0.8972122750790381, "learning_rate": 0.003, "loss": 4.0339, "step": 17292 }, { "epoch": 0.17293, "grad_norm": 1.1533510954730688, "learning_rate": 0.003, "loss": 4.0591, "step": 17293 }, { "epoch": 0.17294, "grad_norm": 1.5653563273040265, "learning_rate": 0.003, "loss": 4.0726, "step": 17294 }, { "epoch": 0.17295, "grad_norm": 1.2090100400889283, "learning_rate": 0.003, "loss": 4.0362, "step": 17295 }, { "epoch": 0.17296, "grad_norm": 1.671789202098, "learning_rate": 0.003, "loss": 4.0488, "step": 17296 }, { "epoch": 0.17297, "grad_norm": 0.9866922678480271, "learning_rate": 0.003, "loss": 4.0707, "step": 17297 }, { "epoch": 0.17298, "grad_norm": 1.1652443394502294, "learning_rate": 0.003, "loss": 4.0371, "step": 17298 }, { "epoch": 0.17299, "grad_norm": 1.2859266559722047, "learning_rate": 0.003, "loss": 4.0625, "step": 17299 }, { "epoch": 0.173, "grad_norm": 1.0738415213616017, "learning_rate": 0.003, "loss": 4.0793, "step": 17300 }, { "epoch": 0.17301, "grad_norm": 1.4778060060409284, "learning_rate": 0.003, "loss": 4.0673, "step": 17301 }, { "epoch": 0.17302, "grad_norm": 0.9797779022071426, "learning_rate": 0.003, "loss": 4.0581, "step": 17302 }, { "epoch": 0.17303, "grad_norm": 1.3111110817384728, "learning_rate": 0.003, "loss": 4.0637, "step": 17303 }, { "epoch": 0.17304, "grad_norm": 1.2661216619762456, "learning_rate": 0.003, "loss": 4.0613, "step": 17304 }, { "epoch": 0.17305, "grad_norm": 1.4112246801763713, "learning_rate": 0.003, "loss": 4.0443, "step": 17305 }, { "epoch": 0.17306, "grad_norm": 1.1174961321385242, "learning_rate": 0.003, "loss": 4.0337, "step": 17306 }, { "epoch": 0.17307, "grad_norm": 1.485030346166996, "learning_rate": 0.003, "loss": 4.0656, "step": 17307 }, { "epoch": 0.17308, "grad_norm": 0.9852911769361422, "learning_rate": 0.003, "loss": 4.073, "step": 17308 }, { "epoch": 0.17309, "grad_norm": 1.3304128966550497, "learning_rate": 0.003, "loss": 4.0919, "step": 17309 }, { "epoch": 0.1731, "grad_norm": 1.0230794649394752, "learning_rate": 0.003, "loss": 4.0861, "step": 17310 }, { "epoch": 0.17311, "grad_norm": 1.3701800499998642, "learning_rate": 0.003, "loss": 4.0933, "step": 17311 }, { "epoch": 0.17312, "grad_norm": 1.3044692407123164, "learning_rate": 0.003, "loss": 4.0661, "step": 17312 }, { "epoch": 0.17313, "grad_norm": 1.0733473023342064, "learning_rate": 0.003, "loss": 4.0698, "step": 17313 }, { "epoch": 0.17314, "grad_norm": 1.6499598634474362, "learning_rate": 0.003, "loss": 4.0871, "step": 17314 }, { "epoch": 0.17315, "grad_norm": 1.1770519491082734, "learning_rate": 0.003, "loss": 4.0589, "step": 17315 }, { "epoch": 0.17316, "grad_norm": 1.311011617413508, "learning_rate": 0.003, "loss": 4.0494, "step": 17316 }, { "epoch": 0.17317, "grad_norm": 1.1164103479989218, "learning_rate": 0.003, "loss": 4.0651, "step": 17317 }, { "epoch": 0.17318, "grad_norm": 1.5049104682902432, "learning_rate": 0.003, "loss": 4.0963, "step": 17318 }, { "epoch": 0.17319, "grad_norm": 1.1184468457167045, "learning_rate": 0.003, "loss": 4.0409, "step": 17319 }, { "epoch": 0.1732, "grad_norm": 1.2817762642047776, "learning_rate": 0.003, "loss": 4.0474, "step": 17320 }, { "epoch": 0.17321, "grad_norm": 1.2009349227032486, "learning_rate": 0.003, "loss": 4.074, "step": 17321 }, { "epoch": 0.17322, "grad_norm": 1.046563884718884, "learning_rate": 0.003, "loss": 4.0615, "step": 17322 }, { "epoch": 0.17323, "grad_norm": 1.3383076588281584, "learning_rate": 0.003, "loss": 4.0595, "step": 17323 }, { "epoch": 0.17324, "grad_norm": 1.016047648867495, "learning_rate": 0.003, "loss": 4.0762, "step": 17324 }, { "epoch": 0.17325, "grad_norm": 1.5114589017916467, "learning_rate": 0.003, "loss": 4.0424, "step": 17325 }, { "epoch": 0.17326, "grad_norm": 0.9455930518856717, "learning_rate": 0.003, "loss": 4.0429, "step": 17326 }, { "epoch": 0.17327, "grad_norm": 1.234539836178666, "learning_rate": 0.003, "loss": 4.0959, "step": 17327 }, { "epoch": 0.17328, "grad_norm": 1.1030151397789953, "learning_rate": 0.003, "loss": 4.0437, "step": 17328 }, { "epoch": 0.17329, "grad_norm": 1.193455723878663, "learning_rate": 0.003, "loss": 4.0803, "step": 17329 }, { "epoch": 0.1733, "grad_norm": 1.1870621228560552, "learning_rate": 0.003, "loss": 4.0663, "step": 17330 }, { "epoch": 0.17331, "grad_norm": 1.181074107227128, "learning_rate": 0.003, "loss": 4.0631, "step": 17331 }, { "epoch": 0.17332, "grad_norm": 1.2535305004121582, "learning_rate": 0.003, "loss": 4.0863, "step": 17332 }, { "epoch": 0.17333, "grad_norm": 1.1677861497915796, "learning_rate": 0.003, "loss": 4.0686, "step": 17333 }, { "epoch": 0.17334, "grad_norm": 1.3835848238520283, "learning_rate": 0.003, "loss": 4.042, "step": 17334 }, { "epoch": 0.17335, "grad_norm": 1.001867916778088, "learning_rate": 0.003, "loss": 4.0497, "step": 17335 }, { "epoch": 0.17336, "grad_norm": 1.1941878147378127, "learning_rate": 0.003, "loss": 4.0771, "step": 17336 }, { "epoch": 0.17337, "grad_norm": 1.4436640146540043, "learning_rate": 0.003, "loss": 4.0812, "step": 17337 }, { "epoch": 0.17338, "grad_norm": 1.2000636357731604, "learning_rate": 0.003, "loss": 4.0274, "step": 17338 }, { "epoch": 0.17339, "grad_norm": 1.425669125341233, "learning_rate": 0.003, "loss": 4.088, "step": 17339 }, { "epoch": 0.1734, "grad_norm": 1.1067087319978788, "learning_rate": 0.003, "loss": 4.0856, "step": 17340 }, { "epoch": 0.17341, "grad_norm": 1.5781509645823653, "learning_rate": 0.003, "loss": 4.0471, "step": 17341 }, { "epoch": 0.17342, "grad_norm": 1.3895482867320283, "learning_rate": 0.003, "loss": 4.0527, "step": 17342 }, { "epoch": 0.17343, "grad_norm": 1.394289370505111, "learning_rate": 0.003, "loss": 4.0822, "step": 17343 }, { "epoch": 0.17344, "grad_norm": 1.1254299309636389, "learning_rate": 0.003, "loss": 4.0608, "step": 17344 }, { "epoch": 0.17345, "grad_norm": 1.3039505706052283, "learning_rate": 0.003, "loss": 4.0538, "step": 17345 }, { "epoch": 0.17346, "grad_norm": 1.0314295078687075, "learning_rate": 0.003, "loss": 4.0676, "step": 17346 }, { "epoch": 0.17347, "grad_norm": 1.428709188191048, "learning_rate": 0.003, "loss": 4.0414, "step": 17347 }, { "epoch": 0.17348, "grad_norm": 1.0377680148142387, "learning_rate": 0.003, "loss": 4.0425, "step": 17348 }, { "epoch": 0.17349, "grad_norm": 1.2438023988122742, "learning_rate": 0.003, "loss": 4.0569, "step": 17349 }, { "epoch": 0.1735, "grad_norm": 1.0543355503086393, "learning_rate": 0.003, "loss": 4.0453, "step": 17350 }, { "epoch": 0.17351, "grad_norm": 1.3034446575461487, "learning_rate": 0.003, "loss": 4.0569, "step": 17351 }, { "epoch": 0.17352, "grad_norm": 1.0396478433823113, "learning_rate": 0.003, "loss": 4.051, "step": 17352 }, { "epoch": 0.17353, "grad_norm": 1.3464171912001612, "learning_rate": 0.003, "loss": 4.0182, "step": 17353 }, { "epoch": 0.17354, "grad_norm": 1.2469167332453905, "learning_rate": 0.003, "loss": 4.0763, "step": 17354 }, { "epoch": 0.17355, "grad_norm": 1.3267021457687498, "learning_rate": 0.003, "loss": 4.0758, "step": 17355 }, { "epoch": 0.17356, "grad_norm": 1.0710192732798718, "learning_rate": 0.003, "loss": 4.0681, "step": 17356 }, { "epoch": 0.17357, "grad_norm": 1.0544001713144933, "learning_rate": 0.003, "loss": 4.0832, "step": 17357 }, { "epoch": 0.17358, "grad_norm": 1.0685268509528416, "learning_rate": 0.003, "loss": 4.0307, "step": 17358 }, { "epoch": 0.17359, "grad_norm": 1.1413721600516764, "learning_rate": 0.003, "loss": 4.0369, "step": 17359 }, { "epoch": 0.1736, "grad_norm": 1.0506366536428504, "learning_rate": 0.003, "loss": 4.0475, "step": 17360 }, { "epoch": 0.17361, "grad_norm": 1.3447342054920186, "learning_rate": 0.003, "loss": 4.0531, "step": 17361 }, { "epoch": 0.17362, "grad_norm": 1.0637095744551777, "learning_rate": 0.003, "loss": 4.0521, "step": 17362 }, { "epoch": 0.17363, "grad_norm": 1.7388362693003458, "learning_rate": 0.003, "loss": 4.0936, "step": 17363 }, { "epoch": 0.17364, "grad_norm": 0.9831432521140705, "learning_rate": 0.003, "loss": 4.0216, "step": 17364 }, { "epoch": 0.17365, "grad_norm": 1.4705586825909038, "learning_rate": 0.003, "loss": 4.0464, "step": 17365 }, { "epoch": 0.17366, "grad_norm": 1.2134443365457936, "learning_rate": 0.003, "loss": 4.0508, "step": 17366 }, { "epoch": 0.17367, "grad_norm": 1.0760335322232322, "learning_rate": 0.003, "loss": 4.0356, "step": 17367 }, { "epoch": 0.17368, "grad_norm": 1.4050896742620886, "learning_rate": 0.003, "loss": 4.0362, "step": 17368 }, { "epoch": 0.17369, "grad_norm": 1.1519306351139893, "learning_rate": 0.003, "loss": 4.0583, "step": 17369 }, { "epoch": 0.1737, "grad_norm": 1.2715432556807118, "learning_rate": 0.003, "loss": 4.0767, "step": 17370 }, { "epoch": 0.17371, "grad_norm": 1.0798038752130645, "learning_rate": 0.003, "loss": 4.0658, "step": 17371 }, { "epoch": 0.17372, "grad_norm": 1.1916009628730764, "learning_rate": 0.003, "loss": 4.0488, "step": 17372 }, { "epoch": 0.17373, "grad_norm": 1.34387916507962, "learning_rate": 0.003, "loss": 4.0851, "step": 17373 }, { "epoch": 0.17374, "grad_norm": 1.2660161673516706, "learning_rate": 0.003, "loss": 4.094, "step": 17374 }, { "epoch": 0.17375, "grad_norm": 1.2639639529750466, "learning_rate": 0.003, "loss": 4.0656, "step": 17375 }, { "epoch": 0.17376, "grad_norm": 0.9883293538639081, "learning_rate": 0.003, "loss": 4.0435, "step": 17376 }, { "epoch": 0.17377, "grad_norm": 1.349228294150077, "learning_rate": 0.003, "loss": 4.0608, "step": 17377 }, { "epoch": 0.17378, "grad_norm": 1.1976298692014888, "learning_rate": 0.003, "loss": 4.0595, "step": 17378 }, { "epoch": 0.17379, "grad_norm": 1.2593558266892997, "learning_rate": 0.003, "loss": 4.0293, "step": 17379 }, { "epoch": 0.1738, "grad_norm": 1.125174432735208, "learning_rate": 0.003, "loss": 4.0166, "step": 17380 }, { "epoch": 0.17381, "grad_norm": 1.134657811220434, "learning_rate": 0.003, "loss": 4.048, "step": 17381 }, { "epoch": 0.17382, "grad_norm": 1.3860411347672301, "learning_rate": 0.003, "loss": 4.0925, "step": 17382 }, { "epoch": 0.17383, "grad_norm": 1.1571068925438313, "learning_rate": 0.003, "loss": 4.0373, "step": 17383 }, { "epoch": 0.17384, "grad_norm": 1.3121596939110471, "learning_rate": 0.003, "loss": 4.0605, "step": 17384 }, { "epoch": 0.17385, "grad_norm": 1.1660777370818731, "learning_rate": 0.003, "loss": 4.0429, "step": 17385 }, { "epoch": 0.17386, "grad_norm": 1.359590834657511, "learning_rate": 0.003, "loss": 4.0524, "step": 17386 }, { "epoch": 0.17387, "grad_norm": 1.2396381385843909, "learning_rate": 0.003, "loss": 4.0601, "step": 17387 }, { "epoch": 0.17388, "grad_norm": 1.3750258841566776, "learning_rate": 0.003, "loss": 4.0488, "step": 17388 }, { "epoch": 0.17389, "grad_norm": 1.0431582342894719, "learning_rate": 0.003, "loss": 4.0477, "step": 17389 }, { "epoch": 0.1739, "grad_norm": 1.092776645824461, "learning_rate": 0.003, "loss": 4.0459, "step": 17390 }, { "epoch": 0.17391, "grad_norm": 1.2212321930368153, "learning_rate": 0.003, "loss": 4.0579, "step": 17391 }, { "epoch": 0.17392, "grad_norm": 1.1880519929514128, "learning_rate": 0.003, "loss": 4.0381, "step": 17392 }, { "epoch": 0.17393, "grad_norm": 1.2987233694070297, "learning_rate": 0.003, "loss": 4.0574, "step": 17393 }, { "epoch": 0.17394, "grad_norm": 1.1547617681859796, "learning_rate": 0.003, "loss": 4.0603, "step": 17394 }, { "epoch": 0.17395, "grad_norm": 1.18312823952414, "learning_rate": 0.003, "loss": 4.0265, "step": 17395 }, { "epoch": 0.17396, "grad_norm": 1.2214029007933558, "learning_rate": 0.003, "loss": 4.0285, "step": 17396 }, { "epoch": 0.17397, "grad_norm": 1.1923965263096545, "learning_rate": 0.003, "loss": 4.052, "step": 17397 }, { "epoch": 0.17398, "grad_norm": 1.1939984511215258, "learning_rate": 0.003, "loss": 4.0862, "step": 17398 }, { "epoch": 0.17399, "grad_norm": 1.3508573792006229, "learning_rate": 0.003, "loss": 4.0569, "step": 17399 }, { "epoch": 0.174, "grad_norm": 0.9756792054531213, "learning_rate": 0.003, "loss": 4.0533, "step": 17400 }, { "epoch": 0.17401, "grad_norm": 1.540516399421629, "learning_rate": 0.003, "loss": 4.0369, "step": 17401 }, { "epoch": 0.17402, "grad_norm": 1.1929414416125332, "learning_rate": 0.003, "loss": 4.0459, "step": 17402 }, { "epoch": 0.17403, "grad_norm": 1.456731591899645, "learning_rate": 0.003, "loss": 4.0738, "step": 17403 }, { "epoch": 0.17404, "grad_norm": 1.3061742791024789, "learning_rate": 0.003, "loss": 4.0538, "step": 17404 }, { "epoch": 0.17405, "grad_norm": 1.0258277823038597, "learning_rate": 0.003, "loss": 4.0377, "step": 17405 }, { "epoch": 0.17406, "grad_norm": 1.3363912977895924, "learning_rate": 0.003, "loss": 4.0608, "step": 17406 }, { "epoch": 0.17407, "grad_norm": 1.0277188435368505, "learning_rate": 0.003, "loss": 4.0221, "step": 17407 }, { "epoch": 0.17408, "grad_norm": 1.4616916007944838, "learning_rate": 0.003, "loss": 4.079, "step": 17408 }, { "epoch": 0.17409, "grad_norm": 1.0754104873918373, "learning_rate": 0.003, "loss": 4.0679, "step": 17409 }, { "epoch": 0.1741, "grad_norm": 1.3433569144712922, "learning_rate": 0.003, "loss": 4.065, "step": 17410 }, { "epoch": 0.17411, "grad_norm": 1.2290776012565996, "learning_rate": 0.003, "loss": 4.0444, "step": 17411 }, { "epoch": 0.17412, "grad_norm": 1.4395562337305694, "learning_rate": 0.003, "loss": 4.0554, "step": 17412 }, { "epoch": 0.17413, "grad_norm": 1.059776161876866, "learning_rate": 0.003, "loss": 4.0264, "step": 17413 }, { "epoch": 0.17414, "grad_norm": 1.1961370968530112, "learning_rate": 0.003, "loss": 4.0528, "step": 17414 }, { "epoch": 0.17415, "grad_norm": 1.1513711456485125, "learning_rate": 0.003, "loss": 4.0709, "step": 17415 }, { "epoch": 0.17416, "grad_norm": 1.1713676939812365, "learning_rate": 0.003, "loss": 4.0574, "step": 17416 }, { "epoch": 0.17417, "grad_norm": 1.21215240137682, "learning_rate": 0.003, "loss": 4.0551, "step": 17417 }, { "epoch": 0.17418, "grad_norm": 1.5039122488959036, "learning_rate": 0.003, "loss": 4.0494, "step": 17418 }, { "epoch": 0.17419, "grad_norm": 1.0521888970913764, "learning_rate": 0.003, "loss": 4.0528, "step": 17419 }, { "epoch": 0.1742, "grad_norm": 1.7427240299517976, "learning_rate": 0.003, "loss": 4.0419, "step": 17420 }, { "epoch": 0.17421, "grad_norm": 1.2071011119302868, "learning_rate": 0.003, "loss": 4.061, "step": 17421 }, { "epoch": 0.17422, "grad_norm": 1.1554356510982162, "learning_rate": 0.003, "loss": 4.0552, "step": 17422 }, { "epoch": 0.17423, "grad_norm": 1.0665614824926966, "learning_rate": 0.003, "loss": 4.0564, "step": 17423 }, { "epoch": 0.17424, "grad_norm": 1.0319775786698153, "learning_rate": 0.003, "loss": 4.0589, "step": 17424 }, { "epoch": 0.17425, "grad_norm": 1.3222593694531697, "learning_rate": 0.003, "loss": 4.0762, "step": 17425 }, { "epoch": 0.17426, "grad_norm": 1.0752339619562283, "learning_rate": 0.003, "loss": 4.0454, "step": 17426 }, { "epoch": 0.17427, "grad_norm": 1.1739220640278247, "learning_rate": 0.003, "loss": 4.0535, "step": 17427 }, { "epoch": 0.17428, "grad_norm": 1.2056576134654182, "learning_rate": 0.003, "loss": 4.0439, "step": 17428 }, { "epoch": 0.17429, "grad_norm": 1.2450995026317917, "learning_rate": 0.003, "loss": 4.0965, "step": 17429 }, { "epoch": 0.1743, "grad_norm": 1.31570224111731, "learning_rate": 0.003, "loss": 4.0377, "step": 17430 }, { "epoch": 0.17431, "grad_norm": 1.0332357869599802, "learning_rate": 0.003, "loss": 4.0238, "step": 17431 }, { "epoch": 0.17432, "grad_norm": 1.5605484066922581, "learning_rate": 0.003, "loss": 4.0613, "step": 17432 }, { "epoch": 0.17433, "grad_norm": 1.3264933956197713, "learning_rate": 0.003, "loss": 4.071, "step": 17433 }, { "epoch": 0.17434, "grad_norm": 1.2936850640111643, "learning_rate": 0.003, "loss": 4.0648, "step": 17434 }, { "epoch": 0.17435, "grad_norm": 1.2244355500203998, "learning_rate": 0.003, "loss": 4.0782, "step": 17435 }, { "epoch": 0.17436, "grad_norm": 1.2315533707703679, "learning_rate": 0.003, "loss": 4.0217, "step": 17436 }, { "epoch": 0.17437, "grad_norm": 1.2586522237325786, "learning_rate": 0.003, "loss": 4.0444, "step": 17437 }, { "epoch": 0.17438, "grad_norm": 1.1841362837337046, "learning_rate": 0.003, "loss": 4.066, "step": 17438 }, { "epoch": 0.17439, "grad_norm": 1.2176337631673573, "learning_rate": 0.003, "loss": 4.0627, "step": 17439 }, { "epoch": 0.1744, "grad_norm": 0.8833942887639723, "learning_rate": 0.003, "loss": 4.0462, "step": 17440 }, { "epoch": 0.17441, "grad_norm": 1.0862041245266658, "learning_rate": 0.003, "loss": 4.0428, "step": 17441 }, { "epoch": 0.17442, "grad_norm": 1.5753903842429755, "learning_rate": 0.003, "loss": 4.0479, "step": 17442 }, { "epoch": 0.17443, "grad_norm": 1.24787047520156, "learning_rate": 0.003, "loss": 4.0607, "step": 17443 }, { "epoch": 0.17444, "grad_norm": 1.5739716818429952, "learning_rate": 0.003, "loss": 4.0243, "step": 17444 }, { "epoch": 0.17445, "grad_norm": 0.9822706554680806, "learning_rate": 0.003, "loss": 4.0425, "step": 17445 }, { "epoch": 0.17446, "grad_norm": 1.1543423233155667, "learning_rate": 0.003, "loss": 4.052, "step": 17446 }, { "epoch": 0.17447, "grad_norm": 1.3171341261232, "learning_rate": 0.003, "loss": 4.0484, "step": 17447 }, { "epoch": 0.17448, "grad_norm": 1.0885413195518518, "learning_rate": 0.003, "loss": 4.0757, "step": 17448 }, { "epoch": 0.17449, "grad_norm": 1.5900315349528589, "learning_rate": 0.003, "loss": 4.0658, "step": 17449 }, { "epoch": 0.1745, "grad_norm": 1.0495111283484926, "learning_rate": 0.003, "loss": 4.0523, "step": 17450 }, { "epoch": 0.17451, "grad_norm": 1.2986543620469437, "learning_rate": 0.003, "loss": 4.0301, "step": 17451 }, { "epoch": 0.17452, "grad_norm": 1.1124923261037898, "learning_rate": 0.003, "loss": 4.063, "step": 17452 }, { "epoch": 0.17453, "grad_norm": 1.1499695987840877, "learning_rate": 0.003, "loss": 4.0717, "step": 17453 }, { "epoch": 0.17454, "grad_norm": 1.3980024711271888, "learning_rate": 0.003, "loss": 4.0641, "step": 17454 }, { "epoch": 0.17455, "grad_norm": 1.1463215124019255, "learning_rate": 0.003, "loss": 4.0484, "step": 17455 }, { "epoch": 0.17456, "grad_norm": 1.416338854831411, "learning_rate": 0.003, "loss": 4.0456, "step": 17456 }, { "epoch": 0.17457, "grad_norm": 1.0268140875842684, "learning_rate": 0.003, "loss": 4.0628, "step": 17457 }, { "epoch": 0.17458, "grad_norm": 1.2748176967428673, "learning_rate": 0.003, "loss": 4.0668, "step": 17458 }, { "epoch": 0.17459, "grad_norm": 1.1358222810883827, "learning_rate": 0.003, "loss": 4.059, "step": 17459 }, { "epoch": 0.1746, "grad_norm": 1.1405717283917192, "learning_rate": 0.003, "loss": 4.0524, "step": 17460 }, { "epoch": 0.17461, "grad_norm": 1.1963364224680375, "learning_rate": 0.003, "loss": 4.0552, "step": 17461 }, { "epoch": 0.17462, "grad_norm": 1.2574941412065117, "learning_rate": 0.003, "loss": 4.0721, "step": 17462 }, { "epoch": 0.17463, "grad_norm": 1.1582105387470332, "learning_rate": 0.003, "loss": 4.0303, "step": 17463 }, { "epoch": 0.17464, "grad_norm": 1.1285020721400938, "learning_rate": 0.003, "loss": 4.0733, "step": 17464 }, { "epoch": 0.17465, "grad_norm": 1.1918468012067522, "learning_rate": 0.003, "loss": 4.054, "step": 17465 }, { "epoch": 0.17466, "grad_norm": 1.2707539191826716, "learning_rate": 0.003, "loss": 4.0471, "step": 17466 }, { "epoch": 0.17467, "grad_norm": 1.1256103256155763, "learning_rate": 0.003, "loss": 4.0551, "step": 17467 }, { "epoch": 0.17468, "grad_norm": 1.2050832262061841, "learning_rate": 0.003, "loss": 4.0516, "step": 17468 }, { "epoch": 0.17469, "grad_norm": 1.1627586590453378, "learning_rate": 0.003, "loss": 4.0596, "step": 17469 }, { "epoch": 0.1747, "grad_norm": 1.4769393152301924, "learning_rate": 0.003, "loss": 4.0803, "step": 17470 }, { "epoch": 0.17471, "grad_norm": 1.4019072501158703, "learning_rate": 0.003, "loss": 4.055, "step": 17471 }, { "epoch": 0.17472, "grad_norm": 1.274208524354242, "learning_rate": 0.003, "loss": 4.0706, "step": 17472 }, { "epoch": 0.17473, "grad_norm": 1.0939524218488357, "learning_rate": 0.003, "loss": 4.055, "step": 17473 }, { "epoch": 0.17474, "grad_norm": 1.034345914998027, "learning_rate": 0.003, "loss": 4.049, "step": 17474 }, { "epoch": 0.17475, "grad_norm": 1.6101262917447974, "learning_rate": 0.003, "loss": 4.0475, "step": 17475 }, { "epoch": 0.17476, "grad_norm": 0.9626673674877094, "learning_rate": 0.003, "loss": 4.0625, "step": 17476 }, { "epoch": 0.17477, "grad_norm": 1.480219461509833, "learning_rate": 0.003, "loss": 4.054, "step": 17477 }, { "epoch": 0.17478, "grad_norm": 1.016698696410458, "learning_rate": 0.003, "loss": 4.0387, "step": 17478 }, { "epoch": 0.17479, "grad_norm": 1.3362403645797698, "learning_rate": 0.003, "loss": 4.0449, "step": 17479 }, { "epoch": 0.1748, "grad_norm": 1.1132221595496428, "learning_rate": 0.003, "loss": 4.0692, "step": 17480 }, { "epoch": 0.17481, "grad_norm": 1.3258542160701117, "learning_rate": 0.003, "loss": 4.0447, "step": 17481 }, { "epoch": 0.17482, "grad_norm": 1.2929608645634088, "learning_rate": 0.003, "loss": 4.034, "step": 17482 }, { "epoch": 0.17483, "grad_norm": 1.434721460999265, "learning_rate": 0.003, "loss": 4.074, "step": 17483 }, { "epoch": 0.17484, "grad_norm": 1.0592250933041438, "learning_rate": 0.003, "loss": 4.061, "step": 17484 }, { "epoch": 0.17485, "grad_norm": 1.5433120517065146, "learning_rate": 0.003, "loss": 4.0634, "step": 17485 }, { "epoch": 0.17486, "grad_norm": 1.0705572036394353, "learning_rate": 0.003, "loss": 4.0918, "step": 17486 }, { "epoch": 0.17487, "grad_norm": 1.3827707405959053, "learning_rate": 0.003, "loss": 4.0369, "step": 17487 }, { "epoch": 0.17488, "grad_norm": 1.142033885926698, "learning_rate": 0.003, "loss": 4.0496, "step": 17488 }, { "epoch": 0.17489, "grad_norm": 1.2753747853269362, "learning_rate": 0.003, "loss": 4.0423, "step": 17489 }, { "epoch": 0.1749, "grad_norm": 1.2093357794938853, "learning_rate": 0.003, "loss": 4.0366, "step": 17490 }, { "epoch": 0.17491, "grad_norm": 1.1658421812276745, "learning_rate": 0.003, "loss": 4.0287, "step": 17491 }, { "epoch": 0.17492, "grad_norm": 1.2155218871293862, "learning_rate": 0.003, "loss": 4.0292, "step": 17492 }, { "epoch": 0.17493, "grad_norm": 1.1373968714050275, "learning_rate": 0.003, "loss": 4.0579, "step": 17493 }, { "epoch": 0.17494, "grad_norm": 1.448814901417094, "learning_rate": 0.003, "loss": 4.0488, "step": 17494 }, { "epoch": 0.17495, "grad_norm": 1.0393948111752698, "learning_rate": 0.003, "loss": 4.0327, "step": 17495 }, { "epoch": 0.17496, "grad_norm": 1.2805671539513424, "learning_rate": 0.003, "loss": 4.0426, "step": 17496 }, { "epoch": 0.17497, "grad_norm": 1.0441830037144317, "learning_rate": 0.003, "loss": 4.059, "step": 17497 }, { "epoch": 0.17498, "grad_norm": 1.2992945648252001, "learning_rate": 0.003, "loss": 4.0841, "step": 17498 }, { "epoch": 0.17499, "grad_norm": 1.1967694643430729, "learning_rate": 0.003, "loss": 4.0445, "step": 17499 }, { "epoch": 0.175, "grad_norm": 1.4493269513423794, "learning_rate": 0.003, "loss": 4.0667, "step": 17500 }, { "epoch": 0.17501, "grad_norm": 1.1608281281318578, "learning_rate": 0.003, "loss": 4.0651, "step": 17501 }, { "epoch": 0.17502, "grad_norm": 1.3398676043732376, "learning_rate": 0.003, "loss": 4.0603, "step": 17502 }, { "epoch": 0.17503, "grad_norm": 1.256590649603026, "learning_rate": 0.003, "loss": 4.0891, "step": 17503 }, { "epoch": 0.17504, "grad_norm": 1.2671038513783908, "learning_rate": 0.003, "loss": 4.0561, "step": 17504 }, { "epoch": 0.17505, "grad_norm": 1.1787987263178186, "learning_rate": 0.003, "loss": 4.0489, "step": 17505 }, { "epoch": 0.17506, "grad_norm": 1.1479484073222257, "learning_rate": 0.003, "loss": 4.0373, "step": 17506 }, { "epoch": 0.17507, "grad_norm": 1.3052434632621392, "learning_rate": 0.003, "loss": 4.0326, "step": 17507 }, { "epoch": 0.17508, "grad_norm": 1.1784053215658497, "learning_rate": 0.003, "loss": 4.049, "step": 17508 }, { "epoch": 0.17509, "grad_norm": 1.3909370878692535, "learning_rate": 0.003, "loss": 4.051, "step": 17509 }, { "epoch": 0.1751, "grad_norm": 1.0533720768723058, "learning_rate": 0.003, "loss": 4.0363, "step": 17510 }, { "epoch": 0.17511, "grad_norm": 1.4538669744073178, "learning_rate": 0.003, "loss": 4.0521, "step": 17511 }, { "epoch": 0.17512, "grad_norm": 0.9633325199231935, "learning_rate": 0.003, "loss": 4.0587, "step": 17512 }, { "epoch": 0.17513, "grad_norm": 1.1919300616810415, "learning_rate": 0.003, "loss": 4.0774, "step": 17513 }, { "epoch": 0.17514, "grad_norm": 1.146252261093832, "learning_rate": 0.003, "loss": 4.0669, "step": 17514 }, { "epoch": 0.17515, "grad_norm": 1.368024356258058, "learning_rate": 0.003, "loss": 4.0536, "step": 17515 }, { "epoch": 0.17516, "grad_norm": 1.2166841015921095, "learning_rate": 0.003, "loss": 4.0547, "step": 17516 }, { "epoch": 0.17517, "grad_norm": 1.3091680163057102, "learning_rate": 0.003, "loss": 4.0727, "step": 17517 }, { "epoch": 0.17518, "grad_norm": 1.2396443494024652, "learning_rate": 0.003, "loss": 4.0513, "step": 17518 }, { "epoch": 0.17519, "grad_norm": 1.1627180611853813, "learning_rate": 0.003, "loss": 4.0218, "step": 17519 }, { "epoch": 0.1752, "grad_norm": 1.6198315221427717, "learning_rate": 0.003, "loss": 4.0591, "step": 17520 }, { "epoch": 0.17521, "grad_norm": 0.840168992554525, "learning_rate": 0.003, "loss": 4.0287, "step": 17521 }, { "epoch": 0.17522, "grad_norm": 0.9590914720353984, "learning_rate": 0.003, "loss": 4.0376, "step": 17522 }, { "epoch": 0.17523, "grad_norm": 1.3983920320796508, "learning_rate": 0.003, "loss": 4.0655, "step": 17523 }, { "epoch": 0.17524, "grad_norm": 1.0315801148551125, "learning_rate": 0.003, "loss": 4.0523, "step": 17524 }, { "epoch": 0.17525, "grad_norm": 1.3434704317693753, "learning_rate": 0.003, "loss": 4.056, "step": 17525 }, { "epoch": 0.17526, "grad_norm": 1.2293442992262185, "learning_rate": 0.003, "loss": 4.052, "step": 17526 }, { "epoch": 0.17527, "grad_norm": 1.2841900805269686, "learning_rate": 0.003, "loss": 4.0304, "step": 17527 }, { "epoch": 0.17528, "grad_norm": 1.0836105577783346, "learning_rate": 0.003, "loss": 4.0362, "step": 17528 }, { "epoch": 0.17529, "grad_norm": 1.3544433389629666, "learning_rate": 0.003, "loss": 4.0571, "step": 17529 }, { "epoch": 0.1753, "grad_norm": 1.2382887236243227, "learning_rate": 0.003, "loss": 4.0354, "step": 17530 }, { "epoch": 0.17531, "grad_norm": 1.5347719322095512, "learning_rate": 0.003, "loss": 4.0887, "step": 17531 }, { "epoch": 0.17532, "grad_norm": 1.330589852802039, "learning_rate": 0.003, "loss": 4.0086, "step": 17532 }, { "epoch": 0.17533, "grad_norm": 1.1626480484934163, "learning_rate": 0.003, "loss": 4.0598, "step": 17533 }, { "epoch": 0.17534, "grad_norm": 1.335722699882295, "learning_rate": 0.003, "loss": 4.0371, "step": 17534 }, { "epoch": 0.17535, "grad_norm": 1.2927036804776442, "learning_rate": 0.003, "loss": 4.0187, "step": 17535 }, { "epoch": 0.17536, "grad_norm": 1.1470715866277477, "learning_rate": 0.003, "loss": 4.0324, "step": 17536 }, { "epoch": 0.17537, "grad_norm": 1.4091680550006398, "learning_rate": 0.003, "loss": 4.0537, "step": 17537 }, { "epoch": 0.17538, "grad_norm": 1.0395124989106723, "learning_rate": 0.003, "loss": 4.0813, "step": 17538 }, { "epoch": 0.17539, "grad_norm": 1.1193328696694718, "learning_rate": 0.003, "loss": 4.0338, "step": 17539 }, { "epoch": 0.1754, "grad_norm": 1.0572607792899362, "learning_rate": 0.003, "loss": 4.084, "step": 17540 }, { "epoch": 0.17541, "grad_norm": 1.127607101416577, "learning_rate": 0.003, "loss": 4.0726, "step": 17541 }, { "epoch": 0.17542, "grad_norm": 1.1000821175888242, "learning_rate": 0.003, "loss": 4.053, "step": 17542 }, { "epoch": 0.17543, "grad_norm": 1.1518160078872373, "learning_rate": 0.003, "loss": 4.0703, "step": 17543 }, { "epoch": 0.17544, "grad_norm": 1.0542688960623041, "learning_rate": 0.003, "loss": 4.0606, "step": 17544 }, { "epoch": 0.17545, "grad_norm": 1.328051050266375, "learning_rate": 0.003, "loss": 4.0498, "step": 17545 }, { "epoch": 0.17546, "grad_norm": 1.2527923949541515, "learning_rate": 0.003, "loss": 4.0628, "step": 17546 }, { "epoch": 0.17547, "grad_norm": 1.4186973412639106, "learning_rate": 0.003, "loss": 4.0374, "step": 17547 }, { "epoch": 0.17548, "grad_norm": 1.1633900552347904, "learning_rate": 0.003, "loss": 4.0335, "step": 17548 }, { "epoch": 0.17549, "grad_norm": 1.4780459216676087, "learning_rate": 0.003, "loss": 4.0424, "step": 17549 }, { "epoch": 0.1755, "grad_norm": 1.0172920362123463, "learning_rate": 0.003, "loss": 4.0334, "step": 17550 }, { "epoch": 0.17551, "grad_norm": 1.3685240676184935, "learning_rate": 0.003, "loss": 4.0474, "step": 17551 }, { "epoch": 0.17552, "grad_norm": 1.0368441281610936, "learning_rate": 0.003, "loss": 4.0824, "step": 17552 }, { "epoch": 0.17553, "grad_norm": 1.4058036979030497, "learning_rate": 0.003, "loss": 4.0595, "step": 17553 }, { "epoch": 0.17554, "grad_norm": 1.1194347132437152, "learning_rate": 0.003, "loss": 4.0543, "step": 17554 }, { "epoch": 0.17555, "grad_norm": 1.5655268291642233, "learning_rate": 0.003, "loss": 4.0301, "step": 17555 }, { "epoch": 0.17556, "grad_norm": 1.1736440422000152, "learning_rate": 0.003, "loss": 4.0649, "step": 17556 }, { "epoch": 0.17557, "grad_norm": 1.170099664234999, "learning_rate": 0.003, "loss": 4.0427, "step": 17557 }, { "epoch": 0.17558, "grad_norm": 1.1925769624962441, "learning_rate": 0.003, "loss": 4.0741, "step": 17558 }, { "epoch": 0.17559, "grad_norm": 1.4664331252751481, "learning_rate": 0.003, "loss": 4.074, "step": 17559 }, { "epoch": 0.1756, "grad_norm": 1.165334631183967, "learning_rate": 0.003, "loss": 4.0236, "step": 17560 }, { "epoch": 0.17561, "grad_norm": 1.2493506552134022, "learning_rate": 0.003, "loss": 4.0349, "step": 17561 }, { "epoch": 0.17562, "grad_norm": 1.1387507616222472, "learning_rate": 0.003, "loss": 4.0616, "step": 17562 }, { "epoch": 0.17563, "grad_norm": 1.230383826478518, "learning_rate": 0.003, "loss": 4.0439, "step": 17563 }, { "epoch": 0.17564, "grad_norm": 1.3822001105484043, "learning_rate": 0.003, "loss": 4.0583, "step": 17564 }, { "epoch": 0.17565, "grad_norm": 1.1913997662718807, "learning_rate": 0.003, "loss": 4.0494, "step": 17565 }, { "epoch": 0.17566, "grad_norm": 1.174785593420601, "learning_rate": 0.003, "loss": 4.028, "step": 17566 }, { "epoch": 0.17567, "grad_norm": 1.2781363720580885, "learning_rate": 0.003, "loss": 4.0651, "step": 17567 }, { "epoch": 0.17568, "grad_norm": 1.430901217235836, "learning_rate": 0.003, "loss": 4.0428, "step": 17568 }, { "epoch": 0.17569, "grad_norm": 1.4839028594353691, "learning_rate": 0.003, "loss": 4.0693, "step": 17569 }, { "epoch": 0.1757, "grad_norm": 1.0595950525634856, "learning_rate": 0.003, "loss": 4.0437, "step": 17570 }, { "epoch": 0.17571, "grad_norm": 1.397745750793812, "learning_rate": 0.003, "loss": 4.0552, "step": 17571 }, { "epoch": 0.17572, "grad_norm": 0.8888418747520094, "learning_rate": 0.003, "loss": 4.051, "step": 17572 }, { "epoch": 0.17573, "grad_norm": 1.1291369602671173, "learning_rate": 0.003, "loss": 4.0458, "step": 17573 }, { "epoch": 0.17574, "grad_norm": 1.345867183091558, "learning_rate": 0.003, "loss": 4.0543, "step": 17574 }, { "epoch": 0.17575, "grad_norm": 1.2251985628025621, "learning_rate": 0.003, "loss": 4.0385, "step": 17575 }, { "epoch": 0.17576, "grad_norm": 1.1472025451571806, "learning_rate": 0.003, "loss": 4.0647, "step": 17576 }, { "epoch": 0.17577, "grad_norm": 1.310219249105373, "learning_rate": 0.003, "loss": 4.0349, "step": 17577 }, { "epoch": 0.17578, "grad_norm": 1.115732655745892, "learning_rate": 0.003, "loss": 4.067, "step": 17578 }, { "epoch": 0.17579, "grad_norm": 1.4050434236000149, "learning_rate": 0.003, "loss": 4.0587, "step": 17579 }, { "epoch": 0.1758, "grad_norm": 1.1561917260482364, "learning_rate": 0.003, "loss": 4.0592, "step": 17580 }, { "epoch": 0.17581, "grad_norm": 1.1093752538129786, "learning_rate": 0.003, "loss": 4.0806, "step": 17581 }, { "epoch": 0.17582, "grad_norm": 1.1966739403321605, "learning_rate": 0.003, "loss": 4.0825, "step": 17582 }, { "epoch": 0.17583, "grad_norm": 1.1597078275261965, "learning_rate": 0.003, "loss": 4.0614, "step": 17583 }, { "epoch": 0.17584, "grad_norm": 1.245543602017537, "learning_rate": 0.003, "loss": 4.0638, "step": 17584 }, { "epoch": 0.17585, "grad_norm": 1.2971524283976426, "learning_rate": 0.003, "loss": 4.0677, "step": 17585 }, { "epoch": 0.17586, "grad_norm": 1.279460954671343, "learning_rate": 0.003, "loss": 4.0492, "step": 17586 }, { "epoch": 0.17587, "grad_norm": 0.9850573623267598, "learning_rate": 0.003, "loss": 4.0576, "step": 17587 }, { "epoch": 0.17588, "grad_norm": 1.5538238797994108, "learning_rate": 0.003, "loss": 4.1036, "step": 17588 }, { "epoch": 0.17589, "grad_norm": 1.4729239462062387, "learning_rate": 0.003, "loss": 4.0448, "step": 17589 }, { "epoch": 0.1759, "grad_norm": 1.3982049000883698, "learning_rate": 0.003, "loss": 4.0687, "step": 17590 }, { "epoch": 0.17591, "grad_norm": 1.1514404548821, "learning_rate": 0.003, "loss": 4.0724, "step": 17591 }, { "epoch": 0.17592, "grad_norm": 1.3239635036333988, "learning_rate": 0.003, "loss": 4.0475, "step": 17592 }, { "epoch": 0.17593, "grad_norm": 1.1577004099456845, "learning_rate": 0.003, "loss": 4.0674, "step": 17593 }, { "epoch": 0.17594, "grad_norm": 1.48381322508598, "learning_rate": 0.003, "loss": 4.0595, "step": 17594 }, { "epoch": 0.17595, "grad_norm": 1.2818816514500897, "learning_rate": 0.003, "loss": 4.0472, "step": 17595 }, { "epoch": 0.17596, "grad_norm": 1.1230113393426457, "learning_rate": 0.003, "loss": 4.0592, "step": 17596 }, { "epoch": 0.17597, "grad_norm": 1.3729943591307712, "learning_rate": 0.003, "loss": 4.0567, "step": 17597 }, { "epoch": 0.17598, "grad_norm": 0.9914762352146483, "learning_rate": 0.003, "loss": 4.0829, "step": 17598 }, { "epoch": 0.17599, "grad_norm": 1.3774066199231678, "learning_rate": 0.003, "loss": 4.0945, "step": 17599 }, { "epoch": 0.176, "grad_norm": 1.0965748648443667, "learning_rate": 0.003, "loss": 4.0462, "step": 17600 }, { "epoch": 0.17601, "grad_norm": 1.3602734721447638, "learning_rate": 0.003, "loss": 4.0666, "step": 17601 }, { "epoch": 0.17602, "grad_norm": 1.1284290528234537, "learning_rate": 0.003, "loss": 4.0464, "step": 17602 }, { "epoch": 0.17603, "grad_norm": 1.4470060608861322, "learning_rate": 0.003, "loss": 4.0588, "step": 17603 }, { "epoch": 0.17604, "grad_norm": 1.3326749037952799, "learning_rate": 0.003, "loss": 4.0532, "step": 17604 }, { "epoch": 0.17605, "grad_norm": 1.1992346349580116, "learning_rate": 0.003, "loss": 4.0771, "step": 17605 }, { "epoch": 0.17606, "grad_norm": 1.204252474404044, "learning_rate": 0.003, "loss": 4.0592, "step": 17606 }, { "epoch": 0.17607, "grad_norm": 1.3357520365039544, "learning_rate": 0.003, "loss": 4.0545, "step": 17607 }, { "epoch": 0.17608, "grad_norm": 1.0331392547237794, "learning_rate": 0.003, "loss": 4.0313, "step": 17608 }, { "epoch": 0.17609, "grad_norm": 1.3546960054869321, "learning_rate": 0.003, "loss": 4.0474, "step": 17609 }, { "epoch": 0.1761, "grad_norm": 1.121500244461343, "learning_rate": 0.003, "loss": 4.0274, "step": 17610 }, { "epoch": 0.17611, "grad_norm": 1.6978483805985116, "learning_rate": 0.003, "loss": 4.0619, "step": 17611 }, { "epoch": 0.17612, "grad_norm": 1.2286940762229328, "learning_rate": 0.003, "loss": 3.9987, "step": 17612 }, { "epoch": 0.17613, "grad_norm": 1.1652848259594104, "learning_rate": 0.003, "loss": 4.0484, "step": 17613 }, { "epoch": 0.17614, "grad_norm": 1.273266706629538, "learning_rate": 0.003, "loss": 4.0284, "step": 17614 }, { "epoch": 0.17615, "grad_norm": 1.3824239141339554, "learning_rate": 0.003, "loss": 4.0295, "step": 17615 }, { "epoch": 0.17616, "grad_norm": 0.9745262884860337, "learning_rate": 0.003, "loss": 4.0505, "step": 17616 }, { "epoch": 0.17617, "grad_norm": 1.4242361731724766, "learning_rate": 0.003, "loss": 4.0374, "step": 17617 }, { "epoch": 0.17618, "grad_norm": 0.8912212954298186, "learning_rate": 0.003, "loss": 4.0755, "step": 17618 }, { "epoch": 0.17619, "grad_norm": 1.1166246583787072, "learning_rate": 0.003, "loss": 4.0798, "step": 17619 }, { "epoch": 0.1762, "grad_norm": 1.4365710019264393, "learning_rate": 0.003, "loss": 4.0419, "step": 17620 }, { "epoch": 0.17621, "grad_norm": 0.9786269404450421, "learning_rate": 0.003, "loss": 4.0379, "step": 17621 }, { "epoch": 0.17622, "grad_norm": 1.2822849544308548, "learning_rate": 0.003, "loss": 4.0643, "step": 17622 }, { "epoch": 0.17623, "grad_norm": 1.2558430140495684, "learning_rate": 0.003, "loss": 4.0561, "step": 17623 }, { "epoch": 0.17624, "grad_norm": 1.0919536116804685, "learning_rate": 0.003, "loss": 4.0637, "step": 17624 }, { "epoch": 0.17625, "grad_norm": 1.234739985489756, "learning_rate": 0.003, "loss": 4.0439, "step": 17625 }, { "epoch": 0.17626, "grad_norm": 1.0727220983044485, "learning_rate": 0.003, "loss": 4.0372, "step": 17626 }, { "epoch": 0.17627, "grad_norm": 1.3477234653911756, "learning_rate": 0.003, "loss": 4.0816, "step": 17627 }, { "epoch": 0.17628, "grad_norm": 1.0251531491230008, "learning_rate": 0.003, "loss": 4.0501, "step": 17628 }, { "epoch": 0.17629, "grad_norm": 1.5822116001438553, "learning_rate": 0.003, "loss": 4.0808, "step": 17629 }, { "epoch": 0.1763, "grad_norm": 1.1129092524525874, "learning_rate": 0.003, "loss": 4.0541, "step": 17630 }, { "epoch": 0.17631, "grad_norm": 1.3659215961959672, "learning_rate": 0.003, "loss": 4.0758, "step": 17631 }, { "epoch": 0.17632, "grad_norm": 1.0435482840136914, "learning_rate": 0.003, "loss": 4.1079, "step": 17632 }, { "epoch": 0.17633, "grad_norm": 1.3355283738661576, "learning_rate": 0.003, "loss": 4.0517, "step": 17633 }, { "epoch": 0.17634, "grad_norm": 1.274431936531252, "learning_rate": 0.003, "loss": 4.0453, "step": 17634 }, { "epoch": 0.17635, "grad_norm": 1.2090461791871303, "learning_rate": 0.003, "loss": 4.038, "step": 17635 }, { "epoch": 0.17636, "grad_norm": 1.3685920892283052, "learning_rate": 0.003, "loss": 4.0608, "step": 17636 }, { "epoch": 0.17637, "grad_norm": 1.2009704327649677, "learning_rate": 0.003, "loss": 4.0441, "step": 17637 }, { "epoch": 0.17638, "grad_norm": 1.2632615719779379, "learning_rate": 0.003, "loss": 4.0781, "step": 17638 }, { "epoch": 0.17639, "grad_norm": 1.1397733894136788, "learning_rate": 0.003, "loss": 4.0534, "step": 17639 }, { "epoch": 0.1764, "grad_norm": 1.4079776352686797, "learning_rate": 0.003, "loss": 4.0709, "step": 17640 }, { "epoch": 0.17641, "grad_norm": 1.039036046539604, "learning_rate": 0.003, "loss": 4.0492, "step": 17641 }, { "epoch": 0.17642, "grad_norm": 1.2930599131435265, "learning_rate": 0.003, "loss": 4.0172, "step": 17642 }, { "epoch": 0.17643, "grad_norm": 1.168619643131443, "learning_rate": 0.003, "loss": 4.0585, "step": 17643 }, { "epoch": 0.17644, "grad_norm": 1.66000628596868, "learning_rate": 0.003, "loss": 4.0581, "step": 17644 }, { "epoch": 0.17645, "grad_norm": 1.0035728656425535, "learning_rate": 0.003, "loss": 4.0767, "step": 17645 }, { "epoch": 0.17646, "grad_norm": 1.3657648068501418, "learning_rate": 0.003, "loss": 4.0516, "step": 17646 }, { "epoch": 0.17647, "grad_norm": 1.2531256308702934, "learning_rate": 0.003, "loss": 4.0505, "step": 17647 }, { "epoch": 0.17648, "grad_norm": 1.1741662655175689, "learning_rate": 0.003, "loss": 4.0671, "step": 17648 }, { "epoch": 0.17649, "grad_norm": 1.3786238821849517, "learning_rate": 0.003, "loss": 4.0447, "step": 17649 }, { "epoch": 0.1765, "grad_norm": 1.2318838699530237, "learning_rate": 0.003, "loss": 4.075, "step": 17650 }, { "epoch": 0.17651, "grad_norm": 1.3729766156990553, "learning_rate": 0.003, "loss": 4.0571, "step": 17651 }, { "epoch": 0.17652, "grad_norm": 1.271866798739319, "learning_rate": 0.003, "loss": 4.0735, "step": 17652 }, { "epoch": 0.17653, "grad_norm": 1.0108262950648896, "learning_rate": 0.003, "loss": 4.0641, "step": 17653 }, { "epoch": 0.17654, "grad_norm": 1.4361150452615088, "learning_rate": 0.003, "loss": 4.0685, "step": 17654 }, { "epoch": 0.17655, "grad_norm": 1.3401112621851474, "learning_rate": 0.003, "loss": 4.0749, "step": 17655 }, { "epoch": 0.17656, "grad_norm": 1.0152148181993021, "learning_rate": 0.003, "loss": 4.0498, "step": 17656 }, { "epoch": 0.17657, "grad_norm": 1.5114430360780986, "learning_rate": 0.003, "loss": 4.0246, "step": 17657 }, { "epoch": 0.17658, "grad_norm": 1.111390450685094, "learning_rate": 0.003, "loss": 4.0454, "step": 17658 }, { "epoch": 0.17659, "grad_norm": 1.338600690986281, "learning_rate": 0.003, "loss": 4.0536, "step": 17659 }, { "epoch": 0.1766, "grad_norm": 1.0702854368560606, "learning_rate": 0.003, "loss": 4.0869, "step": 17660 }, { "epoch": 0.17661, "grad_norm": 1.2839299019526675, "learning_rate": 0.003, "loss": 4.0585, "step": 17661 }, { "epoch": 0.17662, "grad_norm": 1.3771072187069797, "learning_rate": 0.003, "loss": 4.0317, "step": 17662 }, { "epoch": 0.17663, "grad_norm": 1.1019750514542517, "learning_rate": 0.003, "loss": 4.0535, "step": 17663 }, { "epoch": 0.17664, "grad_norm": 1.249229399378219, "learning_rate": 0.003, "loss": 4.0482, "step": 17664 }, { "epoch": 0.17665, "grad_norm": 1.0222927806103053, "learning_rate": 0.003, "loss": 4.0647, "step": 17665 }, { "epoch": 0.17666, "grad_norm": 1.1079364755619592, "learning_rate": 0.003, "loss": 4.0582, "step": 17666 }, { "epoch": 0.17667, "grad_norm": 1.4148312123252131, "learning_rate": 0.003, "loss": 4.0441, "step": 17667 }, { "epoch": 0.17668, "grad_norm": 1.3043529386896926, "learning_rate": 0.003, "loss": 4.0861, "step": 17668 }, { "epoch": 0.17669, "grad_norm": 1.3097333085299816, "learning_rate": 0.003, "loss": 4.0577, "step": 17669 }, { "epoch": 0.1767, "grad_norm": 1.3170883756693887, "learning_rate": 0.003, "loss": 4.0444, "step": 17670 }, { "epoch": 0.17671, "grad_norm": 1.3114065044602923, "learning_rate": 0.003, "loss": 4.0747, "step": 17671 }, { "epoch": 0.17672, "grad_norm": 1.2785908563031507, "learning_rate": 0.003, "loss": 4.0592, "step": 17672 }, { "epoch": 0.17673, "grad_norm": 1.4310817681056016, "learning_rate": 0.003, "loss": 4.0369, "step": 17673 }, { "epoch": 0.17674, "grad_norm": 1.0281852852331894, "learning_rate": 0.003, "loss": 4.0634, "step": 17674 }, { "epoch": 0.17675, "grad_norm": 1.2569150977830006, "learning_rate": 0.003, "loss": 4.0698, "step": 17675 }, { "epoch": 0.17676, "grad_norm": 1.1352305456951268, "learning_rate": 0.003, "loss": 4.0598, "step": 17676 }, { "epoch": 0.17677, "grad_norm": 1.3484709075956498, "learning_rate": 0.003, "loss": 4.0609, "step": 17677 }, { "epoch": 0.17678, "grad_norm": 1.1328023324347336, "learning_rate": 0.003, "loss": 4.0853, "step": 17678 }, { "epoch": 0.17679, "grad_norm": 1.2653804097778356, "learning_rate": 0.003, "loss": 4.0376, "step": 17679 }, { "epoch": 0.1768, "grad_norm": 1.0475305282319034, "learning_rate": 0.003, "loss": 4.0274, "step": 17680 }, { "epoch": 0.17681, "grad_norm": 1.3737179652172784, "learning_rate": 0.003, "loss": 4.0925, "step": 17681 }, { "epoch": 0.17682, "grad_norm": 1.2535426664356284, "learning_rate": 0.003, "loss": 4.0301, "step": 17682 }, { "epoch": 0.17683, "grad_norm": 1.2279889096925745, "learning_rate": 0.003, "loss": 4.0768, "step": 17683 }, { "epoch": 0.17684, "grad_norm": 1.1955588371680803, "learning_rate": 0.003, "loss": 4.0733, "step": 17684 }, { "epoch": 0.17685, "grad_norm": 1.4865881201944695, "learning_rate": 0.003, "loss": 4.0816, "step": 17685 }, { "epoch": 0.17686, "grad_norm": 1.0444206015567041, "learning_rate": 0.003, "loss": 4.0631, "step": 17686 }, { "epoch": 0.17687, "grad_norm": 1.4115457653886876, "learning_rate": 0.003, "loss": 4.0429, "step": 17687 }, { "epoch": 0.17688, "grad_norm": 1.0142565536116115, "learning_rate": 0.003, "loss": 4.0412, "step": 17688 }, { "epoch": 0.17689, "grad_norm": 1.2686762194693473, "learning_rate": 0.003, "loss": 4.0481, "step": 17689 }, { "epoch": 0.1769, "grad_norm": 1.1771112520826768, "learning_rate": 0.003, "loss": 4.0771, "step": 17690 }, { "epoch": 0.17691, "grad_norm": 1.1278410458658201, "learning_rate": 0.003, "loss": 4.0634, "step": 17691 }, { "epoch": 0.17692, "grad_norm": 1.6059003840813524, "learning_rate": 0.003, "loss": 4.0526, "step": 17692 }, { "epoch": 0.17693, "grad_norm": 0.9920783057554169, "learning_rate": 0.003, "loss": 4.0758, "step": 17693 }, { "epoch": 0.17694, "grad_norm": 1.4282914897729126, "learning_rate": 0.003, "loss": 4.0483, "step": 17694 }, { "epoch": 0.17695, "grad_norm": 1.1648853447259635, "learning_rate": 0.003, "loss": 4.0618, "step": 17695 }, { "epoch": 0.17696, "grad_norm": 1.580463939936142, "learning_rate": 0.003, "loss": 4.0582, "step": 17696 }, { "epoch": 0.17697, "grad_norm": 1.117583895703195, "learning_rate": 0.003, "loss": 4.0604, "step": 17697 }, { "epoch": 0.17698, "grad_norm": 1.2340128342748578, "learning_rate": 0.003, "loss": 4.0621, "step": 17698 }, { "epoch": 0.17699, "grad_norm": 1.1396362622344216, "learning_rate": 0.003, "loss": 4.0498, "step": 17699 }, { "epoch": 0.177, "grad_norm": 1.214027466886459, "learning_rate": 0.003, "loss": 4.0587, "step": 17700 }, { "epoch": 0.17701, "grad_norm": 1.1328025500727548, "learning_rate": 0.003, "loss": 4.0586, "step": 17701 }, { "epoch": 0.17702, "grad_norm": 1.408872578584067, "learning_rate": 0.003, "loss": 4.0626, "step": 17702 }, { "epoch": 0.17703, "grad_norm": 1.1604123137397944, "learning_rate": 0.003, "loss": 4.0178, "step": 17703 }, { "epoch": 0.17704, "grad_norm": 1.5697108988592918, "learning_rate": 0.003, "loss": 4.059, "step": 17704 }, { "epoch": 0.17705, "grad_norm": 1.2145111871919732, "learning_rate": 0.003, "loss": 4.0427, "step": 17705 }, { "epoch": 0.17706, "grad_norm": 1.1636825149371066, "learning_rate": 0.003, "loss": 4.0609, "step": 17706 }, { "epoch": 0.17707, "grad_norm": 1.144255277297568, "learning_rate": 0.003, "loss": 4.0517, "step": 17707 }, { "epoch": 0.17708, "grad_norm": 1.1035256098491526, "learning_rate": 0.003, "loss": 4.045, "step": 17708 }, { "epoch": 0.17709, "grad_norm": 1.321224817587734, "learning_rate": 0.003, "loss": 4.0901, "step": 17709 }, { "epoch": 0.1771, "grad_norm": 1.144107581755462, "learning_rate": 0.003, "loss": 4.0527, "step": 17710 }, { "epoch": 0.17711, "grad_norm": 1.182481438002469, "learning_rate": 0.003, "loss": 4.0558, "step": 17711 }, { "epoch": 0.17712, "grad_norm": 1.3135315910739842, "learning_rate": 0.003, "loss": 4.0215, "step": 17712 }, { "epoch": 0.17713, "grad_norm": 1.3106809535007589, "learning_rate": 0.003, "loss": 4.0386, "step": 17713 }, { "epoch": 0.17714, "grad_norm": 1.370874815697478, "learning_rate": 0.003, "loss": 4.0647, "step": 17714 }, { "epoch": 0.17715, "grad_norm": 1.1333885622735582, "learning_rate": 0.003, "loss": 4.0402, "step": 17715 }, { "epoch": 0.17716, "grad_norm": 1.2091495521461906, "learning_rate": 0.003, "loss": 4.0732, "step": 17716 }, { "epoch": 0.17717, "grad_norm": 1.330653285021552, "learning_rate": 0.003, "loss": 4.0818, "step": 17717 }, { "epoch": 0.17718, "grad_norm": 1.3424982564609935, "learning_rate": 0.003, "loss": 4.0826, "step": 17718 }, { "epoch": 0.17719, "grad_norm": 1.0814534319686981, "learning_rate": 0.003, "loss": 4.0257, "step": 17719 }, { "epoch": 0.1772, "grad_norm": 1.2538717772839065, "learning_rate": 0.003, "loss": 4.0411, "step": 17720 }, { "epoch": 0.17721, "grad_norm": 1.109582526488981, "learning_rate": 0.003, "loss": 4.0592, "step": 17721 }, { "epoch": 0.17722, "grad_norm": 1.519905307423152, "learning_rate": 0.003, "loss": 4.0326, "step": 17722 }, { "epoch": 0.17723, "grad_norm": 1.0296511825449017, "learning_rate": 0.003, "loss": 4.0575, "step": 17723 }, { "epoch": 0.17724, "grad_norm": 1.3566963687284888, "learning_rate": 0.003, "loss": 4.0564, "step": 17724 }, { "epoch": 0.17725, "grad_norm": 1.1788101028553641, "learning_rate": 0.003, "loss": 4.0605, "step": 17725 }, { "epoch": 0.17726, "grad_norm": 1.421231601442113, "learning_rate": 0.003, "loss": 4.047, "step": 17726 }, { "epoch": 0.17727, "grad_norm": 1.2076056839101652, "learning_rate": 0.003, "loss": 4.0467, "step": 17727 }, { "epoch": 0.17728, "grad_norm": 1.520651972884749, "learning_rate": 0.003, "loss": 4.0752, "step": 17728 }, { "epoch": 0.17729, "grad_norm": 1.1488636254034916, "learning_rate": 0.003, "loss": 4.0578, "step": 17729 }, { "epoch": 0.1773, "grad_norm": 1.1346597045234943, "learning_rate": 0.003, "loss": 4.0354, "step": 17730 }, { "epoch": 0.17731, "grad_norm": 1.306299544302792, "learning_rate": 0.003, "loss": 4.0549, "step": 17731 }, { "epoch": 0.17732, "grad_norm": 1.2966838820232638, "learning_rate": 0.003, "loss": 4.0781, "step": 17732 }, { "epoch": 0.17733, "grad_norm": 1.2096264718522176, "learning_rate": 0.003, "loss": 4.0429, "step": 17733 }, { "epoch": 0.17734, "grad_norm": 1.4028764472622368, "learning_rate": 0.003, "loss": 4.0889, "step": 17734 }, { "epoch": 0.17735, "grad_norm": 0.9606968967699793, "learning_rate": 0.003, "loss": 4.0486, "step": 17735 }, { "epoch": 0.17736, "grad_norm": 1.2020753944731186, "learning_rate": 0.003, "loss": 4.0621, "step": 17736 }, { "epoch": 0.17737, "grad_norm": 1.092798913708364, "learning_rate": 0.003, "loss": 4.0413, "step": 17737 }, { "epoch": 0.17738, "grad_norm": 1.1665994160643567, "learning_rate": 0.003, "loss": 4.0616, "step": 17738 }, { "epoch": 0.17739, "grad_norm": 1.1739179621104348, "learning_rate": 0.003, "loss": 4.0888, "step": 17739 }, { "epoch": 0.1774, "grad_norm": 1.171815396210287, "learning_rate": 0.003, "loss": 4.0626, "step": 17740 }, { "epoch": 0.17741, "grad_norm": 1.3307553846157683, "learning_rate": 0.003, "loss": 4.0692, "step": 17741 }, { "epoch": 0.17742, "grad_norm": 1.1873860826085796, "learning_rate": 0.003, "loss": 4.0547, "step": 17742 }, { "epoch": 0.17743, "grad_norm": 1.7621063120413198, "learning_rate": 0.003, "loss": 4.1058, "step": 17743 }, { "epoch": 0.17744, "grad_norm": 1.0028118946892863, "learning_rate": 0.003, "loss": 4.0428, "step": 17744 }, { "epoch": 0.17745, "grad_norm": 1.3005509065026561, "learning_rate": 0.003, "loss": 4.0668, "step": 17745 }, { "epoch": 0.17746, "grad_norm": 1.223731103567378, "learning_rate": 0.003, "loss": 4.0746, "step": 17746 }, { "epoch": 0.17747, "grad_norm": 1.1780862268274155, "learning_rate": 0.003, "loss": 4.0642, "step": 17747 }, { "epoch": 0.17748, "grad_norm": 1.3037820890798655, "learning_rate": 0.003, "loss": 4.0682, "step": 17748 }, { "epoch": 0.17749, "grad_norm": 1.2787151216940233, "learning_rate": 0.003, "loss": 4.0371, "step": 17749 }, { "epoch": 0.1775, "grad_norm": 1.258189733789853, "learning_rate": 0.003, "loss": 4.0617, "step": 17750 }, { "epoch": 0.17751, "grad_norm": 1.2173859470415813, "learning_rate": 0.003, "loss": 4.0814, "step": 17751 }, { "epoch": 0.17752, "grad_norm": 1.0831928331721983, "learning_rate": 0.003, "loss": 4.0467, "step": 17752 }, { "epoch": 0.17753, "grad_norm": 1.1724284204882491, "learning_rate": 0.003, "loss": 4.0488, "step": 17753 }, { "epoch": 0.17754, "grad_norm": 1.1634608471119323, "learning_rate": 0.003, "loss": 4.0812, "step": 17754 }, { "epoch": 0.17755, "grad_norm": 1.312550749239694, "learning_rate": 0.003, "loss": 4.0718, "step": 17755 }, { "epoch": 0.17756, "grad_norm": 1.2343635267456727, "learning_rate": 0.003, "loss": 4.0295, "step": 17756 }, { "epoch": 0.17757, "grad_norm": 1.6113516556311576, "learning_rate": 0.003, "loss": 4.0745, "step": 17757 }, { "epoch": 0.17758, "grad_norm": 1.3210645551520415, "learning_rate": 0.003, "loss": 4.0647, "step": 17758 }, { "epoch": 0.17759, "grad_norm": 1.059033798774361, "learning_rate": 0.003, "loss": 4.0558, "step": 17759 }, { "epoch": 0.1776, "grad_norm": 1.5521546207512071, "learning_rate": 0.003, "loss": 4.0422, "step": 17760 }, { "epoch": 0.17761, "grad_norm": 1.1829313243367725, "learning_rate": 0.003, "loss": 4.0564, "step": 17761 }, { "epoch": 0.17762, "grad_norm": 1.680215690648254, "learning_rate": 0.003, "loss": 4.0733, "step": 17762 }, { "epoch": 0.17763, "grad_norm": 0.9782282263638936, "learning_rate": 0.003, "loss": 4.0531, "step": 17763 }, { "epoch": 0.17764, "grad_norm": 1.3291797239860548, "learning_rate": 0.003, "loss": 4.071, "step": 17764 }, { "epoch": 0.17765, "grad_norm": 1.5131838203870505, "learning_rate": 0.003, "loss": 4.068, "step": 17765 }, { "epoch": 0.17766, "grad_norm": 1.1618004425482447, "learning_rate": 0.003, "loss": 4.0593, "step": 17766 }, { "epoch": 0.17767, "grad_norm": 1.314309478723575, "learning_rate": 0.003, "loss": 4.072, "step": 17767 }, { "epoch": 0.17768, "grad_norm": 1.1771406581385115, "learning_rate": 0.003, "loss": 4.0429, "step": 17768 }, { "epoch": 0.17769, "grad_norm": 1.081005441102454, "learning_rate": 0.003, "loss": 4.0718, "step": 17769 }, { "epoch": 0.1777, "grad_norm": 1.3551935041036287, "learning_rate": 0.003, "loss": 4.0381, "step": 17770 }, { "epoch": 0.17771, "grad_norm": 1.2143773829771678, "learning_rate": 0.003, "loss": 4.0453, "step": 17771 }, { "epoch": 0.17772, "grad_norm": 1.141660551470356, "learning_rate": 0.003, "loss": 4.0542, "step": 17772 }, { "epoch": 0.17773, "grad_norm": 1.332139537228624, "learning_rate": 0.003, "loss": 4.05, "step": 17773 }, { "epoch": 0.17774, "grad_norm": 1.157054587182458, "learning_rate": 0.003, "loss": 4.0615, "step": 17774 }, { "epoch": 0.17775, "grad_norm": 1.2408157631694359, "learning_rate": 0.003, "loss": 4.0194, "step": 17775 }, { "epoch": 0.17776, "grad_norm": 1.1591098606009527, "learning_rate": 0.003, "loss": 4.0543, "step": 17776 }, { "epoch": 0.17777, "grad_norm": 1.2447238780446748, "learning_rate": 0.003, "loss": 4.0299, "step": 17777 }, { "epoch": 0.17778, "grad_norm": 1.3893890679657246, "learning_rate": 0.003, "loss": 4.052, "step": 17778 }, { "epoch": 0.17779, "grad_norm": 1.258303519382804, "learning_rate": 0.003, "loss": 4.0311, "step": 17779 }, { "epoch": 0.1778, "grad_norm": 1.159148796376476, "learning_rate": 0.003, "loss": 4.063, "step": 17780 }, { "epoch": 0.17781, "grad_norm": 1.233847865044257, "learning_rate": 0.003, "loss": 4.0385, "step": 17781 }, { "epoch": 0.17782, "grad_norm": 1.163521453872752, "learning_rate": 0.003, "loss": 4.0583, "step": 17782 }, { "epoch": 0.17783, "grad_norm": 1.2348060635009395, "learning_rate": 0.003, "loss": 4.0336, "step": 17783 }, { "epoch": 0.17784, "grad_norm": 1.1449346808237173, "learning_rate": 0.003, "loss": 4.0611, "step": 17784 }, { "epoch": 0.17785, "grad_norm": 1.2755708596167314, "learning_rate": 0.003, "loss": 4.0381, "step": 17785 }, { "epoch": 0.17786, "grad_norm": 1.0950855712195076, "learning_rate": 0.003, "loss": 4.0702, "step": 17786 }, { "epoch": 0.17787, "grad_norm": 1.449139170957916, "learning_rate": 0.003, "loss": 4.0662, "step": 17787 }, { "epoch": 0.17788, "grad_norm": 1.1825995204290765, "learning_rate": 0.003, "loss": 4.0607, "step": 17788 }, { "epoch": 0.17789, "grad_norm": 1.4782552630292467, "learning_rate": 0.003, "loss": 4.0493, "step": 17789 }, { "epoch": 0.1779, "grad_norm": 1.0884511898359692, "learning_rate": 0.003, "loss": 4.0281, "step": 17790 }, { "epoch": 0.17791, "grad_norm": 1.303993288342492, "learning_rate": 0.003, "loss": 4.0529, "step": 17791 }, { "epoch": 0.17792, "grad_norm": 1.4776915153971397, "learning_rate": 0.003, "loss": 4.0276, "step": 17792 }, { "epoch": 0.17793, "grad_norm": 1.143160794688882, "learning_rate": 0.003, "loss": 4.0817, "step": 17793 }, { "epoch": 0.17794, "grad_norm": 1.337585706731177, "learning_rate": 0.003, "loss": 4.0624, "step": 17794 }, { "epoch": 0.17795, "grad_norm": 1.2882867499124373, "learning_rate": 0.003, "loss": 4.0497, "step": 17795 }, { "epoch": 0.17796, "grad_norm": 1.3099175579503028, "learning_rate": 0.003, "loss": 4.0376, "step": 17796 }, { "epoch": 0.17797, "grad_norm": 1.3772907450722711, "learning_rate": 0.003, "loss": 4.0728, "step": 17797 }, { "epoch": 0.17798, "grad_norm": 1.1439080417379976, "learning_rate": 0.003, "loss": 4.0548, "step": 17798 }, { "epoch": 0.17799, "grad_norm": 1.4427349945651953, "learning_rate": 0.003, "loss": 4.0793, "step": 17799 }, { "epoch": 0.178, "grad_norm": 0.8359998493602386, "learning_rate": 0.003, "loss": 4.0489, "step": 17800 }, { "epoch": 0.17801, "grad_norm": 1.3170836659544407, "learning_rate": 0.003, "loss": 4.0373, "step": 17801 }, { "epoch": 0.17802, "grad_norm": 1.3913854503988206, "learning_rate": 0.003, "loss": 4.0801, "step": 17802 }, { "epoch": 0.17803, "grad_norm": 1.117097811135875, "learning_rate": 0.003, "loss": 4.0631, "step": 17803 }, { "epoch": 0.17804, "grad_norm": 1.44672804441292, "learning_rate": 0.003, "loss": 4.0747, "step": 17804 }, { "epoch": 0.17805, "grad_norm": 1.081482875995203, "learning_rate": 0.003, "loss": 4.0628, "step": 17805 }, { "epoch": 0.17806, "grad_norm": 1.2817128313436066, "learning_rate": 0.003, "loss": 4.0546, "step": 17806 }, { "epoch": 0.17807, "grad_norm": 1.1442125206487275, "learning_rate": 0.003, "loss": 4.0461, "step": 17807 }, { "epoch": 0.17808, "grad_norm": 1.2389433793066236, "learning_rate": 0.003, "loss": 4.027, "step": 17808 }, { "epoch": 0.17809, "grad_norm": 1.1412907392971439, "learning_rate": 0.003, "loss": 4.0748, "step": 17809 }, { "epoch": 0.1781, "grad_norm": 1.4502744583052565, "learning_rate": 0.003, "loss": 4.0543, "step": 17810 }, { "epoch": 0.17811, "grad_norm": 0.8816630115732121, "learning_rate": 0.003, "loss": 4.0608, "step": 17811 }, { "epoch": 0.17812, "grad_norm": 1.243229745885222, "learning_rate": 0.003, "loss": 4.0613, "step": 17812 }, { "epoch": 0.17813, "grad_norm": 1.2535565267594253, "learning_rate": 0.003, "loss": 4.0964, "step": 17813 }, { "epoch": 0.17814, "grad_norm": 1.2126124047186526, "learning_rate": 0.003, "loss": 4.0741, "step": 17814 }, { "epoch": 0.17815, "grad_norm": 1.298381161029849, "learning_rate": 0.003, "loss": 4.0505, "step": 17815 }, { "epoch": 0.17816, "grad_norm": 1.200970343801902, "learning_rate": 0.003, "loss": 4.0373, "step": 17816 }, { "epoch": 0.17817, "grad_norm": 1.4408631650324562, "learning_rate": 0.003, "loss": 4.0673, "step": 17817 }, { "epoch": 0.17818, "grad_norm": 1.3058479181801004, "learning_rate": 0.003, "loss": 4.0507, "step": 17818 }, { "epoch": 0.17819, "grad_norm": 1.2483198465754957, "learning_rate": 0.003, "loss": 4.0722, "step": 17819 }, { "epoch": 0.1782, "grad_norm": 1.2421489369861138, "learning_rate": 0.003, "loss": 4.0768, "step": 17820 }, { "epoch": 0.17821, "grad_norm": 1.2208967380802178, "learning_rate": 0.003, "loss": 4.0694, "step": 17821 }, { "epoch": 0.17822, "grad_norm": 1.2559317071733944, "learning_rate": 0.003, "loss": 4.0414, "step": 17822 }, { "epoch": 0.17823, "grad_norm": 1.2096541669674137, "learning_rate": 0.003, "loss": 4.0326, "step": 17823 }, { "epoch": 0.17824, "grad_norm": 1.2152591273481994, "learning_rate": 0.003, "loss": 4.0322, "step": 17824 }, { "epoch": 0.17825, "grad_norm": 1.2091785587737627, "learning_rate": 0.003, "loss": 4.052, "step": 17825 }, { "epoch": 0.17826, "grad_norm": 1.1591984087625793, "learning_rate": 0.003, "loss": 4.0569, "step": 17826 }, { "epoch": 0.17827, "grad_norm": 1.2241813736032778, "learning_rate": 0.003, "loss": 4.0629, "step": 17827 }, { "epoch": 0.17828, "grad_norm": 1.3272339228045538, "learning_rate": 0.003, "loss": 4.052, "step": 17828 }, { "epoch": 0.17829, "grad_norm": 1.139992611986561, "learning_rate": 0.003, "loss": 4.0599, "step": 17829 }, { "epoch": 0.1783, "grad_norm": 1.5554212661070375, "learning_rate": 0.003, "loss": 4.0194, "step": 17830 }, { "epoch": 0.17831, "grad_norm": 0.9181997975303219, "learning_rate": 0.003, "loss": 4.0683, "step": 17831 }, { "epoch": 0.17832, "grad_norm": 1.1076323918128805, "learning_rate": 0.003, "loss": 4.0607, "step": 17832 }, { "epoch": 0.17833, "grad_norm": 1.3858914559464381, "learning_rate": 0.003, "loss": 4.0524, "step": 17833 }, { "epoch": 0.17834, "grad_norm": 1.1699376628371638, "learning_rate": 0.003, "loss": 4.0468, "step": 17834 }, { "epoch": 0.17835, "grad_norm": 1.3609949015398057, "learning_rate": 0.003, "loss": 4.0722, "step": 17835 }, { "epoch": 0.17836, "grad_norm": 1.0474430728969828, "learning_rate": 0.003, "loss": 4.055, "step": 17836 }, { "epoch": 0.17837, "grad_norm": 1.517426880051347, "learning_rate": 0.003, "loss": 4.0532, "step": 17837 }, { "epoch": 0.17838, "grad_norm": 1.1199970353263173, "learning_rate": 0.003, "loss": 4.0751, "step": 17838 }, { "epoch": 0.17839, "grad_norm": 1.5003042927264383, "learning_rate": 0.003, "loss": 4.0825, "step": 17839 }, { "epoch": 0.1784, "grad_norm": 1.1665549956682206, "learning_rate": 0.003, "loss": 4.0715, "step": 17840 }, { "epoch": 0.17841, "grad_norm": 1.3668209475230961, "learning_rate": 0.003, "loss": 4.0338, "step": 17841 }, { "epoch": 0.17842, "grad_norm": 1.3351548331053524, "learning_rate": 0.003, "loss": 4.0328, "step": 17842 }, { "epoch": 0.17843, "grad_norm": 1.2465284262700684, "learning_rate": 0.003, "loss": 4.0178, "step": 17843 }, { "epoch": 0.17844, "grad_norm": 1.4059048434887123, "learning_rate": 0.003, "loss": 4.0942, "step": 17844 }, { "epoch": 0.17845, "grad_norm": 1.1641418528709417, "learning_rate": 0.003, "loss": 4.0648, "step": 17845 }, { "epoch": 0.17846, "grad_norm": 1.2066772712782268, "learning_rate": 0.003, "loss": 4.063, "step": 17846 }, { "epoch": 0.17847, "grad_norm": 1.1941036074655749, "learning_rate": 0.003, "loss": 4.055, "step": 17847 }, { "epoch": 0.17848, "grad_norm": 1.2433204115339367, "learning_rate": 0.003, "loss": 4.0565, "step": 17848 }, { "epoch": 0.17849, "grad_norm": 1.3093484254756549, "learning_rate": 0.003, "loss": 4.0849, "step": 17849 }, { "epoch": 0.1785, "grad_norm": 1.1826914800534813, "learning_rate": 0.003, "loss": 4.0769, "step": 17850 }, { "epoch": 0.17851, "grad_norm": 1.2097111418164714, "learning_rate": 0.003, "loss": 4.0762, "step": 17851 }, { "epoch": 0.17852, "grad_norm": 1.2434591953396055, "learning_rate": 0.003, "loss": 4.0632, "step": 17852 }, { "epoch": 0.17853, "grad_norm": 1.2430392272630955, "learning_rate": 0.003, "loss": 4.0628, "step": 17853 }, { "epoch": 0.17854, "grad_norm": 1.1597079985384222, "learning_rate": 0.003, "loss": 4.0381, "step": 17854 }, { "epoch": 0.17855, "grad_norm": 1.1952883665850529, "learning_rate": 0.003, "loss": 4.0392, "step": 17855 }, { "epoch": 0.17856, "grad_norm": 1.2904642448505192, "learning_rate": 0.003, "loss": 4.0756, "step": 17856 }, { "epoch": 0.17857, "grad_norm": 1.2101126895353091, "learning_rate": 0.003, "loss": 4.0516, "step": 17857 }, { "epoch": 0.17858, "grad_norm": 1.2557265330953655, "learning_rate": 0.003, "loss": 4.0474, "step": 17858 }, { "epoch": 0.17859, "grad_norm": 1.3445512913386632, "learning_rate": 0.003, "loss": 4.0663, "step": 17859 }, { "epoch": 0.1786, "grad_norm": 1.3347337979593992, "learning_rate": 0.003, "loss": 4.037, "step": 17860 }, { "epoch": 0.17861, "grad_norm": 1.1674711177676054, "learning_rate": 0.003, "loss": 4.084, "step": 17861 }, { "epoch": 0.17862, "grad_norm": 1.179269192911582, "learning_rate": 0.003, "loss": 4.0468, "step": 17862 }, { "epoch": 0.17863, "grad_norm": 1.1267366647848631, "learning_rate": 0.003, "loss": 4.0559, "step": 17863 }, { "epoch": 0.17864, "grad_norm": 1.3658451603592885, "learning_rate": 0.003, "loss": 4.0816, "step": 17864 }, { "epoch": 0.17865, "grad_norm": 1.183338712902706, "learning_rate": 0.003, "loss": 4.0283, "step": 17865 }, { "epoch": 0.17866, "grad_norm": 1.426545758295184, "learning_rate": 0.003, "loss": 4.0545, "step": 17866 }, { "epoch": 0.17867, "grad_norm": 1.0203541867654127, "learning_rate": 0.003, "loss": 4.0593, "step": 17867 }, { "epoch": 0.17868, "grad_norm": 1.349649210302117, "learning_rate": 0.003, "loss": 4.0573, "step": 17868 }, { "epoch": 0.17869, "grad_norm": 1.0430308736743832, "learning_rate": 0.003, "loss": 4.0588, "step": 17869 }, { "epoch": 0.1787, "grad_norm": 1.4004712405709534, "learning_rate": 0.003, "loss": 4.0463, "step": 17870 }, { "epoch": 0.17871, "grad_norm": 1.1322203274855553, "learning_rate": 0.003, "loss": 4.0499, "step": 17871 }, { "epoch": 0.17872, "grad_norm": 1.3211443371500873, "learning_rate": 0.003, "loss": 4.0436, "step": 17872 }, { "epoch": 0.17873, "grad_norm": 1.458967552716695, "learning_rate": 0.003, "loss": 4.0473, "step": 17873 }, { "epoch": 0.17874, "grad_norm": 1.3076101482343323, "learning_rate": 0.003, "loss": 4.0513, "step": 17874 }, { "epoch": 0.17875, "grad_norm": 1.18593223186616, "learning_rate": 0.003, "loss": 4.086, "step": 17875 }, { "epoch": 0.17876, "grad_norm": 1.1109432963991712, "learning_rate": 0.003, "loss": 4.0693, "step": 17876 }, { "epoch": 0.17877, "grad_norm": 1.4205139155536228, "learning_rate": 0.003, "loss": 4.0502, "step": 17877 }, { "epoch": 0.17878, "grad_norm": 1.0616259733822684, "learning_rate": 0.003, "loss": 4.0197, "step": 17878 }, { "epoch": 0.17879, "grad_norm": 1.33721088716122, "learning_rate": 0.003, "loss": 4.0522, "step": 17879 }, { "epoch": 0.1788, "grad_norm": 1.045484556026667, "learning_rate": 0.003, "loss": 4.0333, "step": 17880 }, { "epoch": 0.17881, "grad_norm": 1.423347840598154, "learning_rate": 0.003, "loss": 4.0666, "step": 17881 }, { "epoch": 0.17882, "grad_norm": 1.0699509572458517, "learning_rate": 0.003, "loss": 4.0301, "step": 17882 }, { "epoch": 0.17883, "grad_norm": 1.4800768903141235, "learning_rate": 0.003, "loss": 4.0565, "step": 17883 }, { "epoch": 0.17884, "grad_norm": 1.2282277128798569, "learning_rate": 0.003, "loss": 4.0294, "step": 17884 }, { "epoch": 0.17885, "grad_norm": 1.458533419609548, "learning_rate": 0.003, "loss": 4.0531, "step": 17885 }, { "epoch": 0.17886, "grad_norm": 1.291266728601586, "learning_rate": 0.003, "loss": 4.0771, "step": 17886 }, { "epoch": 0.17887, "grad_norm": 1.4931675431109201, "learning_rate": 0.003, "loss": 4.0486, "step": 17887 }, { "epoch": 0.17888, "grad_norm": 1.3043677984540665, "learning_rate": 0.003, "loss": 4.0778, "step": 17888 }, { "epoch": 0.17889, "grad_norm": 1.0833629842955117, "learning_rate": 0.003, "loss": 4.0717, "step": 17889 }, { "epoch": 0.1789, "grad_norm": 1.3650054650821548, "learning_rate": 0.003, "loss": 4.0897, "step": 17890 }, { "epoch": 0.17891, "grad_norm": 1.1478210263972732, "learning_rate": 0.003, "loss": 4.0585, "step": 17891 }, { "epoch": 0.17892, "grad_norm": 1.2251283044225163, "learning_rate": 0.003, "loss": 4.0991, "step": 17892 }, { "epoch": 0.17893, "grad_norm": 1.3066400717250508, "learning_rate": 0.003, "loss": 4.0517, "step": 17893 }, { "epoch": 0.17894, "grad_norm": 1.1998657253539133, "learning_rate": 0.003, "loss": 4.0575, "step": 17894 }, { "epoch": 0.17895, "grad_norm": 1.2697499227104956, "learning_rate": 0.003, "loss": 4.0918, "step": 17895 }, { "epoch": 0.17896, "grad_norm": 1.2260165025111005, "learning_rate": 0.003, "loss": 4.0755, "step": 17896 }, { "epoch": 0.17897, "grad_norm": 1.2328994647871134, "learning_rate": 0.003, "loss": 4.0743, "step": 17897 }, { "epoch": 0.17898, "grad_norm": 1.1355480189018692, "learning_rate": 0.003, "loss": 4.0282, "step": 17898 }, { "epoch": 0.17899, "grad_norm": 1.4180974474725112, "learning_rate": 0.003, "loss": 4.0624, "step": 17899 }, { "epoch": 0.179, "grad_norm": 1.299799815272747, "learning_rate": 0.003, "loss": 4.0765, "step": 17900 }, { "epoch": 0.17901, "grad_norm": 1.1427514002836991, "learning_rate": 0.003, "loss": 4.0613, "step": 17901 }, { "epoch": 0.17902, "grad_norm": 1.4421237669924571, "learning_rate": 0.003, "loss": 4.0572, "step": 17902 }, { "epoch": 0.17903, "grad_norm": 1.143470683758552, "learning_rate": 0.003, "loss": 4.0558, "step": 17903 }, { "epoch": 0.17904, "grad_norm": 1.2890418395082373, "learning_rate": 0.003, "loss": 4.0579, "step": 17904 }, { "epoch": 0.17905, "grad_norm": 1.231504588880156, "learning_rate": 0.003, "loss": 4.0503, "step": 17905 }, { "epoch": 0.17906, "grad_norm": 1.1615472894176582, "learning_rate": 0.003, "loss": 4.0337, "step": 17906 }, { "epoch": 0.17907, "grad_norm": 1.1718428948631765, "learning_rate": 0.003, "loss": 4.0161, "step": 17907 }, { "epoch": 0.17908, "grad_norm": 1.2373777391072323, "learning_rate": 0.003, "loss": 4.0661, "step": 17908 }, { "epoch": 0.17909, "grad_norm": 1.4080376200108604, "learning_rate": 0.003, "loss": 4.0791, "step": 17909 }, { "epoch": 0.1791, "grad_norm": 1.0077488587335164, "learning_rate": 0.003, "loss": 4.0539, "step": 17910 }, { "epoch": 0.17911, "grad_norm": 1.5016356252161176, "learning_rate": 0.003, "loss": 4.048, "step": 17911 }, { "epoch": 0.17912, "grad_norm": 1.1234412426190965, "learning_rate": 0.003, "loss": 4.0381, "step": 17912 }, { "epoch": 0.17913, "grad_norm": 1.3905931376690837, "learning_rate": 0.003, "loss": 4.0768, "step": 17913 }, { "epoch": 0.17914, "grad_norm": 0.9897384707940726, "learning_rate": 0.003, "loss": 4.0432, "step": 17914 }, { "epoch": 0.17915, "grad_norm": 1.6719850108086427, "learning_rate": 0.003, "loss": 4.0648, "step": 17915 }, { "epoch": 0.17916, "grad_norm": 1.1694364300723652, "learning_rate": 0.003, "loss": 4.0765, "step": 17916 }, { "epoch": 0.17917, "grad_norm": 1.3665114206250535, "learning_rate": 0.003, "loss": 4.0847, "step": 17917 }, { "epoch": 0.17918, "grad_norm": 1.4539978671274996, "learning_rate": 0.003, "loss": 4.0532, "step": 17918 }, { "epoch": 0.17919, "grad_norm": 1.0078089353484048, "learning_rate": 0.003, "loss": 4.0364, "step": 17919 }, { "epoch": 0.1792, "grad_norm": 1.2324802885155937, "learning_rate": 0.003, "loss": 4.0717, "step": 17920 }, { "epoch": 0.17921, "grad_norm": 1.2576899133452208, "learning_rate": 0.003, "loss": 4.0525, "step": 17921 }, { "epoch": 0.17922, "grad_norm": 1.2547391534459178, "learning_rate": 0.003, "loss": 4.061, "step": 17922 }, { "epoch": 0.17923, "grad_norm": 1.1905937404588032, "learning_rate": 0.003, "loss": 4.0595, "step": 17923 }, { "epoch": 0.17924, "grad_norm": 1.3320548580499494, "learning_rate": 0.003, "loss": 4.0675, "step": 17924 }, { "epoch": 0.17925, "grad_norm": 1.1673613294904652, "learning_rate": 0.003, "loss": 4.0429, "step": 17925 }, { "epoch": 0.17926, "grad_norm": 1.0821393933109273, "learning_rate": 0.003, "loss": 4.0334, "step": 17926 }, { "epoch": 0.17927, "grad_norm": 1.3049139324124037, "learning_rate": 0.003, "loss": 4.0715, "step": 17927 }, { "epoch": 0.17928, "grad_norm": 1.279645265800928, "learning_rate": 0.003, "loss": 4.0603, "step": 17928 }, { "epoch": 0.17929, "grad_norm": 1.2267947142349527, "learning_rate": 0.003, "loss": 4.0437, "step": 17929 }, { "epoch": 0.1793, "grad_norm": 1.2725727547444488, "learning_rate": 0.003, "loss": 4.0616, "step": 17930 }, { "epoch": 0.17931, "grad_norm": 1.2168153164979243, "learning_rate": 0.003, "loss": 4.0474, "step": 17931 }, { "epoch": 0.17932, "grad_norm": 1.1107128817347829, "learning_rate": 0.003, "loss": 4.0595, "step": 17932 }, { "epoch": 0.17933, "grad_norm": 1.1207404804594867, "learning_rate": 0.003, "loss": 4.0354, "step": 17933 }, { "epoch": 0.17934, "grad_norm": 1.2833669444402074, "learning_rate": 0.003, "loss": 4.0838, "step": 17934 }, { "epoch": 0.17935, "grad_norm": 1.0390032883015055, "learning_rate": 0.003, "loss": 4.0672, "step": 17935 }, { "epoch": 0.17936, "grad_norm": 1.1763755061412124, "learning_rate": 0.003, "loss": 4.0591, "step": 17936 }, { "epoch": 0.17937, "grad_norm": 1.4458590632263786, "learning_rate": 0.003, "loss": 4.0729, "step": 17937 }, { "epoch": 0.17938, "grad_norm": 1.0989058007592827, "learning_rate": 0.003, "loss": 4.0672, "step": 17938 }, { "epoch": 0.17939, "grad_norm": 1.326712895987331, "learning_rate": 0.003, "loss": 4.0816, "step": 17939 }, { "epoch": 0.1794, "grad_norm": 1.0218133507438032, "learning_rate": 0.003, "loss": 4.0375, "step": 17940 }, { "epoch": 0.17941, "grad_norm": 1.3050696388879868, "learning_rate": 0.003, "loss": 4.0475, "step": 17941 }, { "epoch": 0.17942, "grad_norm": 1.0433657660442555, "learning_rate": 0.003, "loss": 4.0299, "step": 17942 }, { "epoch": 0.17943, "grad_norm": 1.3186778687120766, "learning_rate": 0.003, "loss": 4.0392, "step": 17943 }, { "epoch": 0.17944, "grad_norm": 1.155891861610521, "learning_rate": 0.003, "loss": 4.0692, "step": 17944 }, { "epoch": 0.17945, "grad_norm": 1.6330229213431369, "learning_rate": 0.003, "loss": 4.0377, "step": 17945 }, { "epoch": 0.17946, "grad_norm": 0.962515382284713, "learning_rate": 0.003, "loss": 4.0558, "step": 17946 }, { "epoch": 0.17947, "grad_norm": 1.2324918576067574, "learning_rate": 0.003, "loss": 4.0493, "step": 17947 }, { "epoch": 0.17948, "grad_norm": 1.3618482939584695, "learning_rate": 0.003, "loss": 4.0533, "step": 17948 }, { "epoch": 0.17949, "grad_norm": 1.1950894831066303, "learning_rate": 0.003, "loss": 4.0659, "step": 17949 }, { "epoch": 0.1795, "grad_norm": 1.2517382845383047, "learning_rate": 0.003, "loss": 4.0797, "step": 17950 }, { "epoch": 0.17951, "grad_norm": 1.277703868914862, "learning_rate": 0.003, "loss": 4.0532, "step": 17951 }, { "epoch": 0.17952, "grad_norm": 1.0964228888869958, "learning_rate": 0.003, "loss": 4.0262, "step": 17952 }, { "epoch": 0.17953, "grad_norm": 1.1939341821407763, "learning_rate": 0.003, "loss": 4.0723, "step": 17953 }, { "epoch": 0.17954, "grad_norm": 1.1651267305984425, "learning_rate": 0.003, "loss": 4.0611, "step": 17954 }, { "epoch": 0.17955, "grad_norm": 1.3850039040149604, "learning_rate": 0.003, "loss": 4.0455, "step": 17955 }, { "epoch": 0.17956, "grad_norm": 1.1177304808547166, "learning_rate": 0.003, "loss": 4.0326, "step": 17956 }, { "epoch": 0.17957, "grad_norm": 1.426185111467826, "learning_rate": 0.003, "loss": 4.0354, "step": 17957 }, { "epoch": 0.17958, "grad_norm": 1.0222384263820816, "learning_rate": 0.003, "loss": 4.0496, "step": 17958 }, { "epoch": 0.17959, "grad_norm": 1.4837637148519538, "learning_rate": 0.003, "loss": 4.0641, "step": 17959 }, { "epoch": 0.1796, "grad_norm": 1.031401407932794, "learning_rate": 0.003, "loss": 4.0715, "step": 17960 }, { "epoch": 0.17961, "grad_norm": 1.4946769120816015, "learning_rate": 0.003, "loss": 4.0721, "step": 17961 }, { "epoch": 0.17962, "grad_norm": 1.1092444101815782, "learning_rate": 0.003, "loss": 4.0471, "step": 17962 }, { "epoch": 0.17963, "grad_norm": 1.364482573566523, "learning_rate": 0.003, "loss": 4.0475, "step": 17963 }, { "epoch": 0.17964, "grad_norm": 1.0107775822137524, "learning_rate": 0.003, "loss": 4.0632, "step": 17964 }, { "epoch": 0.17965, "grad_norm": 1.321693710368358, "learning_rate": 0.003, "loss": 4.0449, "step": 17965 }, { "epoch": 0.17966, "grad_norm": 1.3516587372727524, "learning_rate": 0.003, "loss": 4.029, "step": 17966 }, { "epoch": 0.17967, "grad_norm": 1.4901704315784046, "learning_rate": 0.003, "loss": 4.0665, "step": 17967 }, { "epoch": 0.17968, "grad_norm": 1.1203814778283905, "learning_rate": 0.003, "loss": 4.0545, "step": 17968 }, { "epoch": 0.17969, "grad_norm": 1.411285830228734, "learning_rate": 0.003, "loss": 4.0604, "step": 17969 }, { "epoch": 0.1797, "grad_norm": 1.26186049354943, "learning_rate": 0.003, "loss": 4.0591, "step": 17970 }, { "epoch": 0.17971, "grad_norm": 1.2830855891996265, "learning_rate": 0.003, "loss": 4.0613, "step": 17971 }, { "epoch": 0.17972, "grad_norm": 1.1050419981201274, "learning_rate": 0.003, "loss": 4.0851, "step": 17972 }, { "epoch": 0.17973, "grad_norm": 1.3510607889599966, "learning_rate": 0.003, "loss": 4.0673, "step": 17973 }, { "epoch": 0.17974, "grad_norm": 1.3193656073693985, "learning_rate": 0.003, "loss": 4.0227, "step": 17974 }, { "epoch": 0.17975, "grad_norm": 1.475721440137748, "learning_rate": 0.003, "loss": 4.0593, "step": 17975 }, { "epoch": 0.17976, "grad_norm": 1.0470951869725496, "learning_rate": 0.003, "loss": 4.0363, "step": 17976 }, { "epoch": 0.17977, "grad_norm": 1.381573519833511, "learning_rate": 0.003, "loss": 4.0476, "step": 17977 }, { "epoch": 0.17978, "grad_norm": 1.2287920507101606, "learning_rate": 0.003, "loss": 4.0614, "step": 17978 }, { "epoch": 0.17979, "grad_norm": 1.049816581057559, "learning_rate": 0.003, "loss": 4.0388, "step": 17979 }, { "epoch": 0.1798, "grad_norm": 1.3932567849099906, "learning_rate": 0.003, "loss": 4.0424, "step": 17980 }, { "epoch": 0.17981, "grad_norm": 0.9571817753127221, "learning_rate": 0.003, "loss": 3.9816, "step": 17981 }, { "epoch": 0.17982, "grad_norm": 1.33453258245156, "learning_rate": 0.003, "loss": 4.0641, "step": 17982 }, { "epoch": 0.17983, "grad_norm": 1.0808435904112632, "learning_rate": 0.003, "loss": 4.0611, "step": 17983 }, { "epoch": 0.17984, "grad_norm": 1.3283266299336365, "learning_rate": 0.003, "loss": 4.0661, "step": 17984 }, { "epoch": 0.17985, "grad_norm": 1.1088742473039481, "learning_rate": 0.003, "loss": 4.047, "step": 17985 }, { "epoch": 0.17986, "grad_norm": 1.0956769108826203, "learning_rate": 0.003, "loss": 4.0552, "step": 17986 }, { "epoch": 0.17987, "grad_norm": 1.1649275307222553, "learning_rate": 0.003, "loss": 4.0608, "step": 17987 }, { "epoch": 0.17988, "grad_norm": 1.303071401589229, "learning_rate": 0.003, "loss": 4.0764, "step": 17988 }, { "epoch": 0.17989, "grad_norm": 1.240823923601934, "learning_rate": 0.003, "loss": 4.0696, "step": 17989 }, { "epoch": 0.1799, "grad_norm": 1.6323677339061988, "learning_rate": 0.003, "loss": 4.093, "step": 17990 }, { "epoch": 0.17991, "grad_norm": 1.2061400821525412, "learning_rate": 0.003, "loss": 4.0713, "step": 17991 }, { "epoch": 0.17992, "grad_norm": 1.184887958492287, "learning_rate": 0.003, "loss": 4.0563, "step": 17992 }, { "epoch": 0.17993, "grad_norm": 1.5001225441924497, "learning_rate": 0.003, "loss": 4.0618, "step": 17993 }, { "epoch": 0.17994, "grad_norm": 1.1464491904627079, "learning_rate": 0.003, "loss": 4.0435, "step": 17994 }, { "epoch": 0.17995, "grad_norm": 1.3643988758344954, "learning_rate": 0.003, "loss": 4.0922, "step": 17995 }, { "epoch": 0.17996, "grad_norm": 1.2115026512711184, "learning_rate": 0.003, "loss": 4.0386, "step": 17996 }, { "epoch": 0.17997, "grad_norm": 1.3257856655430436, "learning_rate": 0.003, "loss": 4.0705, "step": 17997 }, { "epoch": 0.17998, "grad_norm": 1.2256908288383672, "learning_rate": 0.003, "loss": 4.0439, "step": 17998 }, { "epoch": 0.17999, "grad_norm": 1.363555714171468, "learning_rate": 0.003, "loss": 4.0683, "step": 17999 }, { "epoch": 0.18, "grad_norm": 1.164308022587496, "learning_rate": 0.003, "loss": 4.0558, "step": 18000 }, { "epoch": 0.18001, "grad_norm": 1.3412044136812997, "learning_rate": 0.003, "loss": 4.0367, "step": 18001 }, { "epoch": 0.18002, "grad_norm": 1.0335057488033608, "learning_rate": 0.003, "loss": 4.0011, "step": 18002 }, { "epoch": 0.18003, "grad_norm": 1.4237231980653955, "learning_rate": 0.003, "loss": 4.0622, "step": 18003 }, { "epoch": 0.18004, "grad_norm": 1.2158323562143312, "learning_rate": 0.003, "loss": 4.065, "step": 18004 }, { "epoch": 0.18005, "grad_norm": 1.1791576182188304, "learning_rate": 0.003, "loss": 4.0585, "step": 18005 }, { "epoch": 0.18006, "grad_norm": 1.1951902681655582, "learning_rate": 0.003, "loss": 4.0648, "step": 18006 }, { "epoch": 0.18007, "grad_norm": 1.172380680181894, "learning_rate": 0.003, "loss": 4.0635, "step": 18007 }, { "epoch": 0.18008, "grad_norm": 1.2765435380928838, "learning_rate": 0.003, "loss": 4.0621, "step": 18008 }, { "epoch": 0.18009, "grad_norm": 1.2416008822680729, "learning_rate": 0.003, "loss": 4.024, "step": 18009 }, { "epoch": 0.1801, "grad_norm": 1.525421875150215, "learning_rate": 0.003, "loss": 4.0515, "step": 18010 }, { "epoch": 0.18011, "grad_norm": 1.021853060087861, "learning_rate": 0.003, "loss": 4.0503, "step": 18011 }, { "epoch": 0.18012, "grad_norm": 1.3502946125471011, "learning_rate": 0.003, "loss": 4.0611, "step": 18012 }, { "epoch": 0.18013, "grad_norm": 1.239070355051635, "learning_rate": 0.003, "loss": 4.0711, "step": 18013 }, { "epoch": 0.18014, "grad_norm": 1.3423907417513339, "learning_rate": 0.003, "loss": 4.0454, "step": 18014 }, { "epoch": 0.18015, "grad_norm": 1.2316428670209958, "learning_rate": 0.003, "loss": 4.1037, "step": 18015 }, { "epoch": 0.18016, "grad_norm": 1.2381433964966237, "learning_rate": 0.003, "loss": 4.065, "step": 18016 }, { "epoch": 0.18017, "grad_norm": 1.1709753510909788, "learning_rate": 0.003, "loss": 4.0552, "step": 18017 }, { "epoch": 0.18018, "grad_norm": 1.2412957293555624, "learning_rate": 0.003, "loss": 4.0493, "step": 18018 }, { "epoch": 0.18019, "grad_norm": 1.1756494709968706, "learning_rate": 0.003, "loss": 4.0361, "step": 18019 }, { "epoch": 0.1802, "grad_norm": 1.199461034209744, "learning_rate": 0.003, "loss": 4.056, "step": 18020 }, { "epoch": 0.18021, "grad_norm": 1.159139826162093, "learning_rate": 0.003, "loss": 4.0803, "step": 18021 }, { "epoch": 0.18022, "grad_norm": 1.2105068534819363, "learning_rate": 0.003, "loss": 4.0419, "step": 18022 }, { "epoch": 0.18023, "grad_norm": 1.2302619521320286, "learning_rate": 0.003, "loss": 4.0356, "step": 18023 }, { "epoch": 0.18024, "grad_norm": 1.1376319466362441, "learning_rate": 0.003, "loss": 4.0421, "step": 18024 }, { "epoch": 0.18025, "grad_norm": 1.3512522055142933, "learning_rate": 0.003, "loss": 4.0285, "step": 18025 }, { "epoch": 0.18026, "grad_norm": 1.3164740883781272, "learning_rate": 0.003, "loss": 4.0777, "step": 18026 }, { "epoch": 0.18027, "grad_norm": 1.2879961756010196, "learning_rate": 0.003, "loss": 4.0499, "step": 18027 }, { "epoch": 0.18028, "grad_norm": 1.1804963011805325, "learning_rate": 0.003, "loss": 4.0261, "step": 18028 }, { "epoch": 0.18029, "grad_norm": 1.4838431177520457, "learning_rate": 0.003, "loss": 4.0684, "step": 18029 }, { "epoch": 0.1803, "grad_norm": 1.0351592429171272, "learning_rate": 0.003, "loss": 4.0776, "step": 18030 }, { "epoch": 0.18031, "grad_norm": 1.2113149016403093, "learning_rate": 0.003, "loss": 4.0427, "step": 18031 }, { "epoch": 0.18032, "grad_norm": 1.148497153568014, "learning_rate": 0.003, "loss": 4.1188, "step": 18032 }, { "epoch": 0.18033, "grad_norm": 1.2622573843454679, "learning_rate": 0.003, "loss": 4.0508, "step": 18033 }, { "epoch": 0.18034, "grad_norm": 1.1921723760443994, "learning_rate": 0.003, "loss": 4.0582, "step": 18034 }, { "epoch": 0.18035, "grad_norm": 1.6950407168602217, "learning_rate": 0.003, "loss": 4.0634, "step": 18035 }, { "epoch": 0.18036, "grad_norm": 1.0196026972145558, "learning_rate": 0.003, "loss": 4.0674, "step": 18036 }, { "epoch": 0.18037, "grad_norm": 1.5396249695351576, "learning_rate": 0.003, "loss": 4.0689, "step": 18037 }, { "epoch": 0.18038, "grad_norm": 0.9897671671610535, "learning_rate": 0.003, "loss": 4.0612, "step": 18038 }, { "epoch": 0.18039, "grad_norm": 1.3512283651462276, "learning_rate": 0.003, "loss": 4.0611, "step": 18039 }, { "epoch": 0.1804, "grad_norm": 1.1484591871116598, "learning_rate": 0.003, "loss": 4.0567, "step": 18040 }, { "epoch": 0.18041, "grad_norm": 1.2147665348576069, "learning_rate": 0.003, "loss": 4.0536, "step": 18041 }, { "epoch": 0.18042, "grad_norm": 1.2380319471707066, "learning_rate": 0.003, "loss": 4.0725, "step": 18042 }, { "epoch": 0.18043, "grad_norm": 1.2160999869057305, "learning_rate": 0.003, "loss": 4.0669, "step": 18043 }, { "epoch": 0.18044, "grad_norm": 1.349050585184057, "learning_rate": 0.003, "loss": 4.1029, "step": 18044 }, { "epoch": 0.18045, "grad_norm": 1.3352668030417683, "learning_rate": 0.003, "loss": 4.0603, "step": 18045 }, { "epoch": 0.18046, "grad_norm": 1.1045455583294432, "learning_rate": 0.003, "loss": 4.0375, "step": 18046 }, { "epoch": 0.18047, "grad_norm": 1.338558036270574, "learning_rate": 0.003, "loss": 4.0611, "step": 18047 }, { "epoch": 0.18048, "grad_norm": 1.2568708707224872, "learning_rate": 0.003, "loss": 4.0534, "step": 18048 }, { "epoch": 0.18049, "grad_norm": 1.2333058752609791, "learning_rate": 0.003, "loss": 4.0805, "step": 18049 }, { "epoch": 0.1805, "grad_norm": 1.2138238824506928, "learning_rate": 0.003, "loss": 4.0616, "step": 18050 }, { "epoch": 0.18051, "grad_norm": 1.203369092557718, "learning_rate": 0.003, "loss": 4.0592, "step": 18051 }, { "epoch": 0.18052, "grad_norm": 1.1492783619007025, "learning_rate": 0.003, "loss": 4.0617, "step": 18052 }, { "epoch": 0.18053, "grad_norm": 1.294912972265915, "learning_rate": 0.003, "loss": 4.0489, "step": 18053 }, { "epoch": 0.18054, "grad_norm": 1.1068522222740473, "learning_rate": 0.003, "loss": 4.0322, "step": 18054 }, { "epoch": 0.18055, "grad_norm": 1.2042374149239887, "learning_rate": 0.003, "loss": 4.068, "step": 18055 }, { "epoch": 0.18056, "grad_norm": 1.1564275150288272, "learning_rate": 0.003, "loss": 4.0418, "step": 18056 }, { "epoch": 0.18057, "grad_norm": 1.0611837217930469, "learning_rate": 0.003, "loss": 4.0424, "step": 18057 }, { "epoch": 0.18058, "grad_norm": 1.4900755342779948, "learning_rate": 0.003, "loss": 4.0515, "step": 18058 }, { "epoch": 0.18059, "grad_norm": 1.2733962853773153, "learning_rate": 0.003, "loss": 4.0417, "step": 18059 }, { "epoch": 0.1806, "grad_norm": 1.2119750026885194, "learning_rate": 0.003, "loss": 4.0223, "step": 18060 }, { "epoch": 0.18061, "grad_norm": 1.2263481938044916, "learning_rate": 0.003, "loss": 4.0644, "step": 18061 }, { "epoch": 0.18062, "grad_norm": 1.2387285682421108, "learning_rate": 0.003, "loss": 4.0492, "step": 18062 }, { "epoch": 0.18063, "grad_norm": 1.4175924684006194, "learning_rate": 0.003, "loss": 4.0468, "step": 18063 }, { "epoch": 0.18064, "grad_norm": 1.063975173407913, "learning_rate": 0.003, "loss": 4.0221, "step": 18064 }, { "epoch": 0.18065, "grad_norm": 1.5299678144357638, "learning_rate": 0.003, "loss": 4.0643, "step": 18065 }, { "epoch": 0.18066, "grad_norm": 1.1752826025686958, "learning_rate": 0.003, "loss": 4.0387, "step": 18066 }, { "epoch": 0.18067, "grad_norm": 1.3438190004736612, "learning_rate": 0.003, "loss": 4.0641, "step": 18067 }, { "epoch": 0.18068, "grad_norm": 1.2332253208533215, "learning_rate": 0.003, "loss": 4.028, "step": 18068 }, { "epoch": 0.18069, "grad_norm": 1.107062614861532, "learning_rate": 0.003, "loss": 4.0469, "step": 18069 }, { "epoch": 0.1807, "grad_norm": 1.26988589284218, "learning_rate": 0.003, "loss": 4.0301, "step": 18070 }, { "epoch": 0.18071, "grad_norm": 1.2597509582459114, "learning_rate": 0.003, "loss": 4.0456, "step": 18071 }, { "epoch": 0.18072, "grad_norm": 1.2984968024076675, "learning_rate": 0.003, "loss": 4.0646, "step": 18072 }, { "epoch": 0.18073, "grad_norm": 1.1404329439653857, "learning_rate": 0.003, "loss": 4.0869, "step": 18073 }, { "epoch": 0.18074, "grad_norm": 1.320504001181121, "learning_rate": 0.003, "loss": 4.0599, "step": 18074 }, { "epoch": 0.18075, "grad_norm": 1.1885836109920593, "learning_rate": 0.003, "loss": 4.0499, "step": 18075 }, { "epoch": 0.18076, "grad_norm": 1.2229886387747457, "learning_rate": 0.003, "loss": 4.0517, "step": 18076 }, { "epoch": 0.18077, "grad_norm": 1.0857055300924845, "learning_rate": 0.003, "loss": 4.0693, "step": 18077 }, { "epoch": 0.18078, "grad_norm": 1.3553751900839532, "learning_rate": 0.003, "loss": 4.0355, "step": 18078 }, { "epoch": 0.18079, "grad_norm": 1.2476777391744713, "learning_rate": 0.003, "loss": 4.0499, "step": 18079 }, { "epoch": 0.1808, "grad_norm": 1.381771594339528, "learning_rate": 0.003, "loss": 4.0579, "step": 18080 }, { "epoch": 0.18081, "grad_norm": 1.0374523519668841, "learning_rate": 0.003, "loss": 4.0849, "step": 18081 }, { "epoch": 0.18082, "grad_norm": 1.3655064736167601, "learning_rate": 0.003, "loss": 4.0538, "step": 18082 }, { "epoch": 0.18083, "grad_norm": 1.3394540211727788, "learning_rate": 0.003, "loss": 4.0482, "step": 18083 }, { "epoch": 0.18084, "grad_norm": 1.2450092834886002, "learning_rate": 0.003, "loss": 4.0586, "step": 18084 }, { "epoch": 0.18085, "grad_norm": 1.1342345617675242, "learning_rate": 0.003, "loss": 4.0263, "step": 18085 }, { "epoch": 0.18086, "grad_norm": 1.117787528449565, "learning_rate": 0.003, "loss": 4.0603, "step": 18086 }, { "epoch": 0.18087, "grad_norm": 1.4351247873622142, "learning_rate": 0.003, "loss": 4.0644, "step": 18087 }, { "epoch": 0.18088, "grad_norm": 1.1237382278508348, "learning_rate": 0.003, "loss": 4.0595, "step": 18088 }, { "epoch": 0.18089, "grad_norm": 1.6307934694970347, "learning_rate": 0.003, "loss": 4.037, "step": 18089 }, { "epoch": 0.1809, "grad_norm": 0.9957714737265753, "learning_rate": 0.003, "loss": 4.0723, "step": 18090 }, { "epoch": 0.18091, "grad_norm": 1.4826744600099264, "learning_rate": 0.003, "loss": 4.0638, "step": 18091 }, { "epoch": 0.18092, "grad_norm": 1.0616821669304313, "learning_rate": 0.003, "loss": 4.0613, "step": 18092 }, { "epoch": 0.18093, "grad_norm": 1.2013092222755266, "learning_rate": 0.003, "loss": 4.0778, "step": 18093 }, { "epoch": 0.18094, "grad_norm": 1.270532891813035, "learning_rate": 0.003, "loss": 4.0381, "step": 18094 }, { "epoch": 0.18095, "grad_norm": 1.146642489000094, "learning_rate": 0.003, "loss": 4.0251, "step": 18095 }, { "epoch": 0.18096, "grad_norm": 1.3894318324246953, "learning_rate": 0.003, "loss": 4.051, "step": 18096 }, { "epoch": 0.18097, "grad_norm": 1.3435900704186123, "learning_rate": 0.003, "loss": 4.0554, "step": 18097 }, { "epoch": 0.18098, "grad_norm": 1.3240453622481962, "learning_rate": 0.003, "loss": 4.0552, "step": 18098 }, { "epoch": 0.18099, "grad_norm": 1.2954307780137428, "learning_rate": 0.003, "loss": 4.0504, "step": 18099 }, { "epoch": 0.181, "grad_norm": 1.4478304170711598, "learning_rate": 0.003, "loss": 4.0785, "step": 18100 }, { "epoch": 0.18101, "grad_norm": 1.1157493662902294, "learning_rate": 0.003, "loss": 4.0457, "step": 18101 }, { "epoch": 0.18102, "grad_norm": 1.4174307426902089, "learning_rate": 0.003, "loss": 4.0585, "step": 18102 }, { "epoch": 0.18103, "grad_norm": 1.0507780911728946, "learning_rate": 0.003, "loss": 4.0638, "step": 18103 }, { "epoch": 0.18104, "grad_norm": 1.4682717700412773, "learning_rate": 0.003, "loss": 4.0613, "step": 18104 }, { "epoch": 0.18105, "grad_norm": 1.0560006411757137, "learning_rate": 0.003, "loss": 4.0749, "step": 18105 }, { "epoch": 0.18106, "grad_norm": 1.2451914317863495, "learning_rate": 0.003, "loss": 4.0521, "step": 18106 }, { "epoch": 0.18107, "grad_norm": 1.1641326286115536, "learning_rate": 0.003, "loss": 4.034, "step": 18107 }, { "epoch": 0.18108, "grad_norm": 1.214895668988106, "learning_rate": 0.003, "loss": 4.0672, "step": 18108 }, { "epoch": 0.18109, "grad_norm": 0.9797829546107352, "learning_rate": 0.003, "loss": 4.0418, "step": 18109 }, { "epoch": 0.1811, "grad_norm": 1.3979027989683943, "learning_rate": 0.003, "loss": 4.0929, "step": 18110 }, { "epoch": 0.18111, "grad_norm": 1.1773445699754366, "learning_rate": 0.003, "loss": 4.0387, "step": 18111 }, { "epoch": 0.18112, "grad_norm": 1.491013110380474, "learning_rate": 0.003, "loss": 4.063, "step": 18112 }, { "epoch": 0.18113, "grad_norm": 1.4326619946319281, "learning_rate": 0.003, "loss": 4.0954, "step": 18113 }, { "epoch": 0.18114, "grad_norm": 1.1473072257356771, "learning_rate": 0.003, "loss": 4.0623, "step": 18114 }, { "epoch": 0.18115, "grad_norm": 1.3187217651431622, "learning_rate": 0.003, "loss": 4.0435, "step": 18115 }, { "epoch": 0.18116, "grad_norm": 1.1026689359745885, "learning_rate": 0.003, "loss": 4.0929, "step": 18116 }, { "epoch": 0.18117, "grad_norm": 1.407834913836384, "learning_rate": 0.003, "loss": 4.0875, "step": 18117 }, { "epoch": 0.18118, "grad_norm": 0.9974611519559162, "learning_rate": 0.003, "loss": 4.0503, "step": 18118 }, { "epoch": 0.18119, "grad_norm": 1.3584033870556507, "learning_rate": 0.003, "loss": 4.0606, "step": 18119 }, { "epoch": 0.1812, "grad_norm": 1.1575947584923405, "learning_rate": 0.003, "loss": 4.0486, "step": 18120 }, { "epoch": 0.18121, "grad_norm": 1.3993242415779288, "learning_rate": 0.003, "loss": 4.0343, "step": 18121 }, { "epoch": 0.18122, "grad_norm": 1.2113842574798337, "learning_rate": 0.003, "loss": 4.0482, "step": 18122 }, { "epoch": 0.18123, "grad_norm": 1.110930997452562, "learning_rate": 0.003, "loss": 4.0172, "step": 18123 }, { "epoch": 0.18124, "grad_norm": 1.267077278650209, "learning_rate": 0.003, "loss": 4.0349, "step": 18124 }, { "epoch": 0.18125, "grad_norm": 1.248509228043099, "learning_rate": 0.003, "loss": 4.0622, "step": 18125 }, { "epoch": 0.18126, "grad_norm": 1.23188913831077, "learning_rate": 0.003, "loss": 4.0682, "step": 18126 }, { "epoch": 0.18127, "grad_norm": 1.259012015805301, "learning_rate": 0.003, "loss": 4.0835, "step": 18127 }, { "epoch": 0.18128, "grad_norm": 1.0563186995646476, "learning_rate": 0.003, "loss": 4.0333, "step": 18128 }, { "epoch": 0.18129, "grad_norm": 1.329551586318536, "learning_rate": 0.003, "loss": 4.0379, "step": 18129 }, { "epoch": 0.1813, "grad_norm": 1.2869493911917629, "learning_rate": 0.003, "loss": 4.0512, "step": 18130 }, { "epoch": 0.18131, "grad_norm": 1.2765876629558837, "learning_rate": 0.003, "loss": 4.0566, "step": 18131 }, { "epoch": 0.18132, "grad_norm": 1.1801299514195327, "learning_rate": 0.003, "loss": 4.0662, "step": 18132 }, { "epoch": 0.18133, "grad_norm": 1.424653965297774, "learning_rate": 0.003, "loss": 4.0537, "step": 18133 }, { "epoch": 0.18134, "grad_norm": 1.0888569962987793, "learning_rate": 0.003, "loss": 4.0733, "step": 18134 }, { "epoch": 0.18135, "grad_norm": 1.5304092847839716, "learning_rate": 0.003, "loss": 4.061, "step": 18135 }, { "epoch": 0.18136, "grad_norm": 1.1698168815889853, "learning_rate": 0.003, "loss": 4.047, "step": 18136 }, { "epoch": 0.18137, "grad_norm": 1.279134688994487, "learning_rate": 0.003, "loss": 4.0815, "step": 18137 }, { "epoch": 0.18138, "grad_norm": 1.1713600574739995, "learning_rate": 0.003, "loss": 4.0503, "step": 18138 }, { "epoch": 0.18139, "grad_norm": 1.348354173926797, "learning_rate": 0.003, "loss": 4.0549, "step": 18139 }, { "epoch": 0.1814, "grad_norm": 1.1903681611134498, "learning_rate": 0.003, "loss": 4.0464, "step": 18140 }, { "epoch": 0.18141, "grad_norm": 1.396601393949861, "learning_rate": 0.003, "loss": 4.0517, "step": 18141 }, { "epoch": 0.18142, "grad_norm": 1.1114557185986427, "learning_rate": 0.003, "loss": 4.0339, "step": 18142 }, { "epoch": 0.18143, "grad_norm": 1.363436301360319, "learning_rate": 0.003, "loss": 4.0389, "step": 18143 }, { "epoch": 0.18144, "grad_norm": 1.1996189226751284, "learning_rate": 0.003, "loss": 4.0688, "step": 18144 }, { "epoch": 0.18145, "grad_norm": 1.5186331509259658, "learning_rate": 0.003, "loss": 4.0538, "step": 18145 }, { "epoch": 0.18146, "grad_norm": 0.9823695488471641, "learning_rate": 0.003, "loss": 4.0386, "step": 18146 }, { "epoch": 0.18147, "grad_norm": 1.3796416762623962, "learning_rate": 0.003, "loss": 4.0545, "step": 18147 }, { "epoch": 0.18148, "grad_norm": 1.2882762858680572, "learning_rate": 0.003, "loss": 4.0402, "step": 18148 }, { "epoch": 0.18149, "grad_norm": 1.1096876382926448, "learning_rate": 0.003, "loss": 4.0556, "step": 18149 }, { "epoch": 0.1815, "grad_norm": 1.4933113427119287, "learning_rate": 0.003, "loss": 4.0525, "step": 18150 }, { "epoch": 0.18151, "grad_norm": 1.053805286179601, "learning_rate": 0.003, "loss": 4.0702, "step": 18151 }, { "epoch": 0.18152, "grad_norm": 1.3146079074140908, "learning_rate": 0.003, "loss": 4.0535, "step": 18152 }, { "epoch": 0.18153, "grad_norm": 1.0868174410400009, "learning_rate": 0.003, "loss": 4.0743, "step": 18153 }, { "epoch": 0.18154, "grad_norm": 1.264663576873165, "learning_rate": 0.003, "loss": 4.034, "step": 18154 }, { "epoch": 0.18155, "grad_norm": 1.2457024950840911, "learning_rate": 0.003, "loss": 4.0754, "step": 18155 }, { "epoch": 0.18156, "grad_norm": 1.2908701271650775, "learning_rate": 0.003, "loss": 4.0659, "step": 18156 }, { "epoch": 0.18157, "grad_norm": 1.5574366991134188, "learning_rate": 0.003, "loss": 4.0749, "step": 18157 }, { "epoch": 0.18158, "grad_norm": 1.1082657444457111, "learning_rate": 0.003, "loss": 4.0341, "step": 18158 }, { "epoch": 0.18159, "grad_norm": 1.6012985022268318, "learning_rate": 0.003, "loss": 4.0569, "step": 18159 }, { "epoch": 0.1816, "grad_norm": 1.0897029935236637, "learning_rate": 0.003, "loss": 4.0656, "step": 18160 }, { "epoch": 0.18161, "grad_norm": 1.3739888336510175, "learning_rate": 0.003, "loss": 4.0595, "step": 18161 }, { "epoch": 0.18162, "grad_norm": 1.075768081948606, "learning_rate": 0.003, "loss": 4.0724, "step": 18162 }, { "epoch": 0.18163, "grad_norm": 1.0848080036670058, "learning_rate": 0.003, "loss": 4.0493, "step": 18163 }, { "epoch": 0.18164, "grad_norm": 1.251003885452719, "learning_rate": 0.003, "loss": 4.0603, "step": 18164 }, { "epoch": 0.18165, "grad_norm": 1.2802533359731652, "learning_rate": 0.003, "loss": 4.0567, "step": 18165 }, { "epoch": 0.18166, "grad_norm": 1.1654615542109659, "learning_rate": 0.003, "loss": 4.0384, "step": 18166 }, { "epoch": 0.18167, "grad_norm": 1.3373001401801745, "learning_rate": 0.003, "loss": 4.0648, "step": 18167 }, { "epoch": 0.18168, "grad_norm": 1.1409164703407304, "learning_rate": 0.003, "loss": 4.0528, "step": 18168 }, { "epoch": 0.18169, "grad_norm": 1.468935684660133, "learning_rate": 0.003, "loss": 4.0666, "step": 18169 }, { "epoch": 0.1817, "grad_norm": 1.121908993117358, "learning_rate": 0.003, "loss": 4.0423, "step": 18170 }, { "epoch": 0.18171, "grad_norm": 1.3702058817884184, "learning_rate": 0.003, "loss": 4.0714, "step": 18171 }, { "epoch": 0.18172, "grad_norm": 1.1760654252535787, "learning_rate": 0.003, "loss": 4.051, "step": 18172 }, { "epoch": 0.18173, "grad_norm": 1.1683759422998519, "learning_rate": 0.003, "loss": 4.0387, "step": 18173 }, { "epoch": 0.18174, "grad_norm": 1.2596310269817508, "learning_rate": 0.003, "loss": 4.0745, "step": 18174 }, { "epoch": 0.18175, "grad_norm": 1.0022911390677356, "learning_rate": 0.003, "loss": 4.072, "step": 18175 }, { "epoch": 0.18176, "grad_norm": 1.5245780473996084, "learning_rate": 0.003, "loss": 4.076, "step": 18176 }, { "epoch": 0.18177, "grad_norm": 1.0375269397207594, "learning_rate": 0.003, "loss": 4.0712, "step": 18177 }, { "epoch": 0.18178, "grad_norm": 1.7900866802540607, "learning_rate": 0.003, "loss": 4.0922, "step": 18178 }, { "epoch": 0.18179, "grad_norm": 0.8776712208774177, "learning_rate": 0.003, "loss": 4.0293, "step": 18179 }, { "epoch": 0.1818, "grad_norm": 1.2002975729982084, "learning_rate": 0.003, "loss": 4.0666, "step": 18180 }, { "epoch": 0.18181, "grad_norm": 1.3797133184577732, "learning_rate": 0.003, "loss": 4.0683, "step": 18181 }, { "epoch": 0.18182, "grad_norm": 1.0868714237281438, "learning_rate": 0.003, "loss": 4.036, "step": 18182 }, { "epoch": 0.18183, "grad_norm": 1.3460658814721587, "learning_rate": 0.003, "loss": 4.0578, "step": 18183 }, { "epoch": 0.18184, "grad_norm": 1.366988467703209, "learning_rate": 0.003, "loss": 4.0807, "step": 18184 }, { "epoch": 0.18185, "grad_norm": 1.4201669622193525, "learning_rate": 0.003, "loss": 4.086, "step": 18185 }, { "epoch": 0.18186, "grad_norm": 1.205017084823888, "learning_rate": 0.003, "loss": 4.0593, "step": 18186 }, { "epoch": 0.18187, "grad_norm": 1.3172298536356926, "learning_rate": 0.003, "loss": 4.0675, "step": 18187 }, { "epoch": 0.18188, "grad_norm": 1.0620127686630847, "learning_rate": 0.003, "loss": 4.0674, "step": 18188 }, { "epoch": 0.18189, "grad_norm": 1.4387558417071202, "learning_rate": 0.003, "loss": 4.0521, "step": 18189 }, { "epoch": 0.1819, "grad_norm": 1.2840994198713898, "learning_rate": 0.003, "loss": 4.0566, "step": 18190 }, { "epoch": 0.18191, "grad_norm": 1.2090175105525351, "learning_rate": 0.003, "loss": 4.0565, "step": 18191 }, { "epoch": 0.18192, "grad_norm": 1.1974107121217237, "learning_rate": 0.003, "loss": 4.0685, "step": 18192 }, { "epoch": 0.18193, "grad_norm": 1.308380356147569, "learning_rate": 0.003, "loss": 4.0428, "step": 18193 }, { "epoch": 0.18194, "grad_norm": 1.0468549651928476, "learning_rate": 0.003, "loss": 4.0449, "step": 18194 }, { "epoch": 0.18195, "grad_norm": 1.2222161372847502, "learning_rate": 0.003, "loss": 4.0609, "step": 18195 }, { "epoch": 0.18196, "grad_norm": 1.203429438902244, "learning_rate": 0.003, "loss": 4.0408, "step": 18196 }, { "epoch": 0.18197, "grad_norm": 1.1828547791402275, "learning_rate": 0.003, "loss": 4.0486, "step": 18197 }, { "epoch": 0.18198, "grad_norm": 1.254164207569211, "learning_rate": 0.003, "loss": 4.0443, "step": 18198 }, { "epoch": 0.18199, "grad_norm": 1.2540510591540648, "learning_rate": 0.003, "loss": 4.06, "step": 18199 }, { "epoch": 0.182, "grad_norm": 1.1693680555263508, "learning_rate": 0.003, "loss": 4.0402, "step": 18200 }, { "epoch": 0.18201, "grad_norm": 1.5424127788744206, "learning_rate": 0.003, "loss": 4.0755, "step": 18201 }, { "epoch": 0.18202, "grad_norm": 1.0666229449001072, "learning_rate": 0.003, "loss": 4.041, "step": 18202 }, { "epoch": 0.18203, "grad_norm": 1.483172598404617, "learning_rate": 0.003, "loss": 4.0525, "step": 18203 }, { "epoch": 0.18204, "grad_norm": 1.1751492384475146, "learning_rate": 0.003, "loss": 4.0328, "step": 18204 }, { "epoch": 0.18205, "grad_norm": 1.2448157635226145, "learning_rate": 0.003, "loss": 4.0267, "step": 18205 }, { "epoch": 0.18206, "grad_norm": 1.7781994750984322, "learning_rate": 0.003, "loss": 4.0457, "step": 18206 }, { "epoch": 0.18207, "grad_norm": 1.3792367114824855, "learning_rate": 0.003, "loss": 4.0627, "step": 18207 }, { "epoch": 0.18208, "grad_norm": 1.0562640052095869, "learning_rate": 0.003, "loss": 4.0716, "step": 18208 }, { "epoch": 0.18209, "grad_norm": 1.3334679284779756, "learning_rate": 0.003, "loss": 4.0471, "step": 18209 }, { "epoch": 0.1821, "grad_norm": 1.1901289354630578, "learning_rate": 0.003, "loss": 4.0606, "step": 18210 }, { "epoch": 0.18211, "grad_norm": 1.4574478094780343, "learning_rate": 0.003, "loss": 4.0498, "step": 18211 }, { "epoch": 0.18212, "grad_norm": 1.2568982149335541, "learning_rate": 0.003, "loss": 4.0404, "step": 18212 }, { "epoch": 0.18213, "grad_norm": 1.0991050921729633, "learning_rate": 0.003, "loss": 4.0632, "step": 18213 }, { "epoch": 0.18214, "grad_norm": 1.2358437166942335, "learning_rate": 0.003, "loss": 4.0422, "step": 18214 }, { "epoch": 0.18215, "grad_norm": 1.1466128425869118, "learning_rate": 0.003, "loss": 4.0592, "step": 18215 }, { "epoch": 0.18216, "grad_norm": 1.1505190831323802, "learning_rate": 0.003, "loss": 4.0246, "step": 18216 }, { "epoch": 0.18217, "grad_norm": 1.3262682912182426, "learning_rate": 0.003, "loss": 4.0675, "step": 18217 }, { "epoch": 0.18218, "grad_norm": 1.0954661064915585, "learning_rate": 0.003, "loss": 4.0207, "step": 18218 }, { "epoch": 0.18219, "grad_norm": 1.3173900128684979, "learning_rate": 0.003, "loss": 4.067, "step": 18219 }, { "epoch": 0.1822, "grad_norm": 1.0356806303502464, "learning_rate": 0.003, "loss": 4.0619, "step": 18220 }, { "epoch": 0.18221, "grad_norm": 1.3347460450586923, "learning_rate": 0.003, "loss": 4.0814, "step": 18221 }, { "epoch": 0.18222, "grad_norm": 1.04383992299412, "learning_rate": 0.003, "loss": 4.0686, "step": 18222 }, { "epoch": 0.18223, "grad_norm": 1.269495196604096, "learning_rate": 0.003, "loss": 4.0749, "step": 18223 }, { "epoch": 0.18224, "grad_norm": 1.2044940236297053, "learning_rate": 0.003, "loss": 4.036, "step": 18224 }, { "epoch": 0.18225, "grad_norm": 1.380718818266551, "learning_rate": 0.003, "loss": 4.0542, "step": 18225 }, { "epoch": 0.18226, "grad_norm": 1.3422067657627699, "learning_rate": 0.003, "loss": 4.0755, "step": 18226 }, { "epoch": 0.18227, "grad_norm": 1.183494487628006, "learning_rate": 0.003, "loss": 4.0819, "step": 18227 }, { "epoch": 0.18228, "grad_norm": 1.1868594272546387, "learning_rate": 0.003, "loss": 4.064, "step": 18228 }, { "epoch": 0.18229, "grad_norm": 1.0844426746752334, "learning_rate": 0.003, "loss": 4.0371, "step": 18229 }, { "epoch": 0.1823, "grad_norm": 1.3500625981204193, "learning_rate": 0.003, "loss": 4.0733, "step": 18230 }, { "epoch": 0.18231, "grad_norm": 1.1111751784718151, "learning_rate": 0.003, "loss": 4.0409, "step": 18231 }, { "epoch": 0.18232, "grad_norm": 1.2567781682210177, "learning_rate": 0.003, "loss": 4.0614, "step": 18232 }, { "epoch": 0.18233, "grad_norm": 1.068747246417859, "learning_rate": 0.003, "loss": 4.0559, "step": 18233 }, { "epoch": 0.18234, "grad_norm": 1.280503411083223, "learning_rate": 0.003, "loss": 4.0416, "step": 18234 }, { "epoch": 0.18235, "grad_norm": 1.121330812361253, "learning_rate": 0.003, "loss": 4.076, "step": 18235 }, { "epoch": 0.18236, "grad_norm": 1.1211308075295416, "learning_rate": 0.003, "loss": 4.0614, "step": 18236 }, { "epoch": 0.18237, "grad_norm": 1.4690225127213803, "learning_rate": 0.003, "loss": 4.065, "step": 18237 }, { "epoch": 0.18238, "grad_norm": 1.1570290902385694, "learning_rate": 0.003, "loss": 4.0545, "step": 18238 }, { "epoch": 0.18239, "grad_norm": 1.4223512705051753, "learning_rate": 0.003, "loss": 4.0433, "step": 18239 }, { "epoch": 0.1824, "grad_norm": 1.1734552617111975, "learning_rate": 0.003, "loss": 4.0309, "step": 18240 }, { "epoch": 0.18241, "grad_norm": 1.723514897872368, "learning_rate": 0.003, "loss": 4.0771, "step": 18241 }, { "epoch": 0.18242, "grad_norm": 1.160824652772994, "learning_rate": 0.003, "loss": 4.0233, "step": 18242 }, { "epoch": 0.18243, "grad_norm": 1.7053584970983728, "learning_rate": 0.003, "loss": 4.0388, "step": 18243 }, { "epoch": 0.18244, "grad_norm": 1.2443513513362243, "learning_rate": 0.003, "loss": 4.0691, "step": 18244 }, { "epoch": 0.18245, "grad_norm": 1.1734999692543786, "learning_rate": 0.003, "loss": 4.0692, "step": 18245 }, { "epoch": 0.18246, "grad_norm": 1.3092224123559955, "learning_rate": 0.003, "loss": 4.0536, "step": 18246 }, { "epoch": 0.18247, "grad_norm": 1.4217802514625206, "learning_rate": 0.003, "loss": 4.0663, "step": 18247 }, { "epoch": 0.18248, "grad_norm": 1.059889898722704, "learning_rate": 0.003, "loss": 4.0773, "step": 18248 }, { "epoch": 0.18249, "grad_norm": 1.3825667571008604, "learning_rate": 0.003, "loss": 4.0558, "step": 18249 }, { "epoch": 0.1825, "grad_norm": 1.2257225759372845, "learning_rate": 0.003, "loss": 4.0778, "step": 18250 }, { "epoch": 0.18251, "grad_norm": 1.5244673134581954, "learning_rate": 0.003, "loss": 4.0754, "step": 18251 }, { "epoch": 0.18252, "grad_norm": 1.0797095673183785, "learning_rate": 0.003, "loss": 4.0923, "step": 18252 }, { "epoch": 0.18253, "grad_norm": 1.327777557684387, "learning_rate": 0.003, "loss": 4.0464, "step": 18253 }, { "epoch": 0.18254, "grad_norm": 1.0489593877567802, "learning_rate": 0.003, "loss": 4.0667, "step": 18254 }, { "epoch": 0.18255, "grad_norm": 1.5240857198603717, "learning_rate": 0.003, "loss": 4.0585, "step": 18255 }, { "epoch": 0.18256, "grad_norm": 1.136629056612958, "learning_rate": 0.003, "loss": 4.0556, "step": 18256 }, { "epoch": 0.18257, "grad_norm": 0.9934615078496511, "learning_rate": 0.003, "loss": 4.0544, "step": 18257 }, { "epoch": 0.18258, "grad_norm": 1.3380337871466155, "learning_rate": 0.003, "loss": 4.0801, "step": 18258 }, { "epoch": 0.18259, "grad_norm": 1.0183344292377945, "learning_rate": 0.003, "loss": 4.0543, "step": 18259 }, { "epoch": 0.1826, "grad_norm": 1.337678866827291, "learning_rate": 0.003, "loss": 4.0253, "step": 18260 }, { "epoch": 0.18261, "grad_norm": 1.4272146872387483, "learning_rate": 0.003, "loss": 4.039, "step": 18261 }, { "epoch": 0.18262, "grad_norm": 1.1892082207166188, "learning_rate": 0.003, "loss": 4.0515, "step": 18262 }, { "epoch": 0.18263, "grad_norm": 1.130988447114421, "learning_rate": 0.003, "loss": 4.0573, "step": 18263 }, { "epoch": 0.18264, "grad_norm": 1.3096296248501902, "learning_rate": 0.003, "loss": 4.0713, "step": 18264 }, { "epoch": 0.18265, "grad_norm": 1.2888662142148892, "learning_rate": 0.003, "loss": 4.0617, "step": 18265 }, { "epoch": 0.18266, "grad_norm": 1.267280829020684, "learning_rate": 0.003, "loss": 4.0469, "step": 18266 }, { "epoch": 0.18267, "grad_norm": 1.0603504202519463, "learning_rate": 0.003, "loss": 4.0575, "step": 18267 }, { "epoch": 0.18268, "grad_norm": 1.2371568894298932, "learning_rate": 0.003, "loss": 4.0565, "step": 18268 }, { "epoch": 0.18269, "grad_norm": 1.1722973922098998, "learning_rate": 0.003, "loss": 4.0635, "step": 18269 }, { "epoch": 0.1827, "grad_norm": 1.3027346280041303, "learning_rate": 0.003, "loss": 4.0506, "step": 18270 }, { "epoch": 0.18271, "grad_norm": 1.2309646097171965, "learning_rate": 0.003, "loss": 4.0678, "step": 18271 }, { "epoch": 0.18272, "grad_norm": 1.3663051952039242, "learning_rate": 0.003, "loss": 4.0549, "step": 18272 }, { "epoch": 0.18273, "grad_norm": 1.2285679725594594, "learning_rate": 0.003, "loss": 4.039, "step": 18273 }, { "epoch": 0.18274, "grad_norm": 1.097664724710302, "learning_rate": 0.003, "loss": 4.0741, "step": 18274 }, { "epoch": 0.18275, "grad_norm": 1.3283120714766745, "learning_rate": 0.003, "loss": 4.0615, "step": 18275 }, { "epoch": 0.18276, "grad_norm": 0.9972392283599518, "learning_rate": 0.003, "loss": 4.0522, "step": 18276 }, { "epoch": 0.18277, "grad_norm": 1.243113304546045, "learning_rate": 0.003, "loss": 4.0642, "step": 18277 }, { "epoch": 0.18278, "grad_norm": 1.2489426903513285, "learning_rate": 0.003, "loss": 4.0535, "step": 18278 }, { "epoch": 0.18279, "grad_norm": 1.2452532494113835, "learning_rate": 0.003, "loss": 4.0612, "step": 18279 }, { "epoch": 0.1828, "grad_norm": 1.2500726378055314, "learning_rate": 0.003, "loss": 4.0601, "step": 18280 }, { "epoch": 0.18281, "grad_norm": 1.2552143875100266, "learning_rate": 0.003, "loss": 4.0769, "step": 18281 }, { "epoch": 0.18282, "grad_norm": 1.2464722209299044, "learning_rate": 0.003, "loss": 4.0619, "step": 18282 }, { "epoch": 0.18283, "grad_norm": 1.1204390881075939, "learning_rate": 0.003, "loss": 4.065, "step": 18283 }, { "epoch": 0.18284, "grad_norm": 1.4257660571352764, "learning_rate": 0.003, "loss": 4.0803, "step": 18284 }, { "epoch": 0.18285, "grad_norm": 1.1537045570323736, "learning_rate": 0.003, "loss": 4.0322, "step": 18285 }, { "epoch": 0.18286, "grad_norm": 1.3376390225902903, "learning_rate": 0.003, "loss": 4.0516, "step": 18286 }, { "epoch": 0.18287, "grad_norm": 1.1905243620450867, "learning_rate": 0.003, "loss": 4.0661, "step": 18287 }, { "epoch": 0.18288, "grad_norm": 1.4219983758559431, "learning_rate": 0.003, "loss": 4.0653, "step": 18288 }, { "epoch": 0.18289, "grad_norm": 1.491007229393689, "learning_rate": 0.003, "loss": 4.0459, "step": 18289 }, { "epoch": 0.1829, "grad_norm": 0.9372579397340259, "learning_rate": 0.003, "loss": 4.0195, "step": 18290 }, { "epoch": 0.18291, "grad_norm": 1.2249712767153973, "learning_rate": 0.003, "loss": 4.0291, "step": 18291 }, { "epoch": 0.18292, "grad_norm": 1.1668551361500483, "learning_rate": 0.003, "loss": 4.0755, "step": 18292 }, { "epoch": 0.18293, "grad_norm": 1.2483664929708789, "learning_rate": 0.003, "loss": 4.0584, "step": 18293 }, { "epoch": 0.18294, "grad_norm": 1.3767312298327286, "learning_rate": 0.003, "loss": 4.0542, "step": 18294 }, { "epoch": 0.18295, "grad_norm": 1.1326350231472428, "learning_rate": 0.003, "loss": 4.0451, "step": 18295 }, { "epoch": 0.18296, "grad_norm": 1.10728669872332, "learning_rate": 0.003, "loss": 4.0676, "step": 18296 }, { "epoch": 0.18297, "grad_norm": 1.5379390762954344, "learning_rate": 0.003, "loss": 4.0743, "step": 18297 }, { "epoch": 0.18298, "grad_norm": 1.0414777100691703, "learning_rate": 0.003, "loss": 4.0864, "step": 18298 }, { "epoch": 0.18299, "grad_norm": 1.5429002917083006, "learning_rate": 0.003, "loss": 4.0477, "step": 18299 }, { "epoch": 0.183, "grad_norm": 1.1100658287599119, "learning_rate": 0.003, "loss": 4.0611, "step": 18300 }, { "epoch": 0.18301, "grad_norm": 1.231714394432137, "learning_rate": 0.003, "loss": 4.0523, "step": 18301 }, { "epoch": 0.18302, "grad_norm": 1.2259113285668939, "learning_rate": 0.003, "loss": 4.0768, "step": 18302 }, { "epoch": 0.18303, "grad_norm": 1.1364295934492379, "learning_rate": 0.003, "loss": 4.0566, "step": 18303 }, { "epoch": 0.18304, "grad_norm": 1.489554152808706, "learning_rate": 0.003, "loss": 4.0838, "step": 18304 }, { "epoch": 0.18305, "grad_norm": 0.9800170074568549, "learning_rate": 0.003, "loss": 4.0365, "step": 18305 }, { "epoch": 0.18306, "grad_norm": 1.5025478067001217, "learning_rate": 0.003, "loss": 4.0575, "step": 18306 }, { "epoch": 0.18307, "grad_norm": 1.152243239338405, "learning_rate": 0.003, "loss": 4.0276, "step": 18307 }, { "epoch": 0.18308, "grad_norm": 1.3602754061144529, "learning_rate": 0.003, "loss": 4.0398, "step": 18308 }, { "epoch": 0.18309, "grad_norm": 1.2800181471469514, "learning_rate": 0.003, "loss": 4.0595, "step": 18309 }, { "epoch": 0.1831, "grad_norm": 1.2850650195814035, "learning_rate": 0.003, "loss": 4.0641, "step": 18310 }, { "epoch": 0.18311, "grad_norm": 1.226801174414026, "learning_rate": 0.003, "loss": 4.0454, "step": 18311 }, { "epoch": 0.18312, "grad_norm": 1.0931000755517517, "learning_rate": 0.003, "loss": 4.0673, "step": 18312 }, { "epoch": 0.18313, "grad_norm": 1.2551201885435146, "learning_rate": 0.003, "loss": 3.9859, "step": 18313 }, { "epoch": 0.18314, "grad_norm": 0.9381602413085871, "learning_rate": 0.003, "loss": 4.0414, "step": 18314 }, { "epoch": 0.18315, "grad_norm": 1.291920088716176, "learning_rate": 0.003, "loss": 4.0567, "step": 18315 }, { "epoch": 0.18316, "grad_norm": 1.2293352344284154, "learning_rate": 0.003, "loss": 4.0662, "step": 18316 }, { "epoch": 0.18317, "grad_norm": 1.2691610893685448, "learning_rate": 0.003, "loss": 4.0721, "step": 18317 }, { "epoch": 0.18318, "grad_norm": 1.2031632877889293, "learning_rate": 0.003, "loss": 4.0532, "step": 18318 }, { "epoch": 0.18319, "grad_norm": 1.4372889545623073, "learning_rate": 0.003, "loss": 4.0174, "step": 18319 }, { "epoch": 0.1832, "grad_norm": 1.1938822809284222, "learning_rate": 0.003, "loss": 4.0453, "step": 18320 }, { "epoch": 0.18321, "grad_norm": 1.7396964838428048, "learning_rate": 0.003, "loss": 4.1234, "step": 18321 }, { "epoch": 0.18322, "grad_norm": 1.0167582841136749, "learning_rate": 0.003, "loss": 4.0626, "step": 18322 }, { "epoch": 0.18323, "grad_norm": 1.2967181482217018, "learning_rate": 0.003, "loss": 4.0512, "step": 18323 }, { "epoch": 0.18324, "grad_norm": 1.40156900342504, "learning_rate": 0.003, "loss": 4.0693, "step": 18324 }, { "epoch": 0.18325, "grad_norm": 1.4238864556716415, "learning_rate": 0.003, "loss": 4.0566, "step": 18325 }, { "epoch": 0.18326, "grad_norm": 1.3421617265103192, "learning_rate": 0.003, "loss": 4.0508, "step": 18326 }, { "epoch": 0.18327, "grad_norm": 1.117911370027948, "learning_rate": 0.003, "loss": 4.0525, "step": 18327 }, { "epoch": 0.18328, "grad_norm": 1.3451995827335028, "learning_rate": 0.003, "loss": 4.0694, "step": 18328 }, { "epoch": 0.18329, "grad_norm": 1.1784696999814355, "learning_rate": 0.003, "loss": 4.033, "step": 18329 }, { "epoch": 0.1833, "grad_norm": 1.390971450708566, "learning_rate": 0.003, "loss": 4.0941, "step": 18330 }, { "epoch": 0.18331, "grad_norm": 1.3913771267533268, "learning_rate": 0.003, "loss": 4.0351, "step": 18331 }, { "epoch": 0.18332, "grad_norm": 1.0795999422508706, "learning_rate": 0.003, "loss": 4.0412, "step": 18332 }, { "epoch": 0.18333, "grad_norm": 1.3322619358634882, "learning_rate": 0.003, "loss": 4.0638, "step": 18333 }, { "epoch": 0.18334, "grad_norm": 1.1542061585922139, "learning_rate": 0.003, "loss": 4.0244, "step": 18334 }, { "epoch": 0.18335, "grad_norm": 1.4284341864109296, "learning_rate": 0.003, "loss": 4.0629, "step": 18335 }, { "epoch": 0.18336, "grad_norm": 1.0439042550047468, "learning_rate": 0.003, "loss": 4.0259, "step": 18336 }, { "epoch": 0.18337, "grad_norm": 1.3615724668800908, "learning_rate": 0.003, "loss": 4.055, "step": 18337 }, { "epoch": 0.18338, "grad_norm": 1.1382971968373383, "learning_rate": 0.003, "loss": 4.0319, "step": 18338 }, { "epoch": 0.18339, "grad_norm": 1.4439725470651312, "learning_rate": 0.003, "loss": 4.0643, "step": 18339 }, { "epoch": 0.1834, "grad_norm": 1.3157856314429814, "learning_rate": 0.003, "loss": 4.0533, "step": 18340 }, { "epoch": 0.18341, "grad_norm": 0.9721477717900843, "learning_rate": 0.003, "loss": 4.0287, "step": 18341 }, { "epoch": 0.18342, "grad_norm": 1.3463242621740237, "learning_rate": 0.003, "loss": 4.0587, "step": 18342 }, { "epoch": 0.18343, "grad_norm": 1.0596979320175746, "learning_rate": 0.003, "loss": 4.0335, "step": 18343 }, { "epoch": 0.18344, "grad_norm": 1.4004903286620822, "learning_rate": 0.003, "loss": 4.0722, "step": 18344 }, { "epoch": 0.18345, "grad_norm": 1.1448159920144843, "learning_rate": 0.003, "loss": 4.0517, "step": 18345 }, { "epoch": 0.18346, "grad_norm": 1.3386523052482129, "learning_rate": 0.003, "loss": 4.054, "step": 18346 }, { "epoch": 0.18347, "grad_norm": 1.063078283040304, "learning_rate": 0.003, "loss": 4.0502, "step": 18347 }, { "epoch": 0.18348, "grad_norm": 1.459775990951888, "learning_rate": 0.003, "loss": 4.0706, "step": 18348 }, { "epoch": 0.18349, "grad_norm": 1.2590137029632196, "learning_rate": 0.003, "loss": 4.0547, "step": 18349 }, { "epoch": 0.1835, "grad_norm": 1.394849360214779, "learning_rate": 0.003, "loss": 4.0569, "step": 18350 }, { "epoch": 0.18351, "grad_norm": 1.2674196745764508, "learning_rate": 0.003, "loss": 4.0541, "step": 18351 }, { "epoch": 0.18352, "grad_norm": 1.1597828357200477, "learning_rate": 0.003, "loss": 4.042, "step": 18352 }, { "epoch": 0.18353, "grad_norm": 1.3406387081704234, "learning_rate": 0.003, "loss": 4.0559, "step": 18353 }, { "epoch": 0.18354, "grad_norm": 1.1076191932627248, "learning_rate": 0.003, "loss": 4.0536, "step": 18354 }, { "epoch": 0.18355, "grad_norm": 1.4097792549029675, "learning_rate": 0.003, "loss": 4.0609, "step": 18355 }, { "epoch": 0.18356, "grad_norm": 1.1904630552071125, "learning_rate": 0.003, "loss": 4.0282, "step": 18356 }, { "epoch": 0.18357, "grad_norm": 1.337396066701302, "learning_rate": 0.003, "loss": 4.0395, "step": 18357 }, { "epoch": 0.18358, "grad_norm": 1.2589228951948699, "learning_rate": 0.003, "loss": 4.0491, "step": 18358 }, { "epoch": 0.18359, "grad_norm": 1.108926826809863, "learning_rate": 0.003, "loss": 4.0808, "step": 18359 }, { "epoch": 0.1836, "grad_norm": 1.3603760767508253, "learning_rate": 0.003, "loss": 4.0516, "step": 18360 }, { "epoch": 0.18361, "grad_norm": 1.1226725968804887, "learning_rate": 0.003, "loss": 4.0227, "step": 18361 }, { "epoch": 0.18362, "grad_norm": 1.2720267602985238, "learning_rate": 0.003, "loss": 4.0435, "step": 18362 }, { "epoch": 0.18363, "grad_norm": 1.1342031278953895, "learning_rate": 0.003, "loss": 4.0365, "step": 18363 }, { "epoch": 0.18364, "grad_norm": 1.295159021261614, "learning_rate": 0.003, "loss": 4.088, "step": 18364 }, { "epoch": 0.18365, "grad_norm": 1.1735434226929458, "learning_rate": 0.003, "loss": 4.047, "step": 18365 }, { "epoch": 0.18366, "grad_norm": 1.2954442953501812, "learning_rate": 0.003, "loss": 4.0476, "step": 18366 }, { "epoch": 0.18367, "grad_norm": 1.2740847313393036, "learning_rate": 0.003, "loss": 4.0549, "step": 18367 }, { "epoch": 0.18368, "grad_norm": 1.2140895280074016, "learning_rate": 0.003, "loss": 4.0298, "step": 18368 }, { "epoch": 0.18369, "grad_norm": 1.3933726149998125, "learning_rate": 0.003, "loss": 4.0557, "step": 18369 }, { "epoch": 0.1837, "grad_norm": 1.1258452789286704, "learning_rate": 0.003, "loss": 4.0296, "step": 18370 }, { "epoch": 0.18371, "grad_norm": 1.3659543472847637, "learning_rate": 0.003, "loss": 4.03, "step": 18371 }, { "epoch": 0.18372, "grad_norm": 1.222245331243453, "learning_rate": 0.003, "loss": 4.061, "step": 18372 }, { "epoch": 0.18373, "grad_norm": 1.7141417251122142, "learning_rate": 0.003, "loss": 4.0829, "step": 18373 }, { "epoch": 0.18374, "grad_norm": 0.9671025556018095, "learning_rate": 0.003, "loss": 4.047, "step": 18374 }, { "epoch": 0.18375, "grad_norm": 1.1872740727230664, "learning_rate": 0.003, "loss": 4.0503, "step": 18375 }, { "epoch": 0.18376, "grad_norm": 1.3334669979762708, "learning_rate": 0.003, "loss": 4.0815, "step": 18376 }, { "epoch": 0.18377, "grad_norm": 1.1194892071827662, "learning_rate": 0.003, "loss": 4.03, "step": 18377 }, { "epoch": 0.18378, "grad_norm": 1.5590549469648822, "learning_rate": 0.003, "loss": 4.0531, "step": 18378 }, { "epoch": 0.18379, "grad_norm": 1.1596940620535392, "learning_rate": 0.003, "loss": 4.069, "step": 18379 }, { "epoch": 0.1838, "grad_norm": 1.29515950308488, "learning_rate": 0.003, "loss": 4.0742, "step": 18380 }, { "epoch": 0.18381, "grad_norm": 0.9829611319668723, "learning_rate": 0.003, "loss": 4.0795, "step": 18381 }, { "epoch": 0.18382, "grad_norm": 1.5561889709198071, "learning_rate": 0.003, "loss": 4.0717, "step": 18382 }, { "epoch": 0.18383, "grad_norm": 1.1900066307598778, "learning_rate": 0.003, "loss": 4.0582, "step": 18383 }, { "epoch": 0.18384, "grad_norm": 1.3426263061815311, "learning_rate": 0.003, "loss": 4.0844, "step": 18384 }, { "epoch": 0.18385, "grad_norm": 1.3912776433001495, "learning_rate": 0.003, "loss": 4.0544, "step": 18385 }, { "epoch": 0.18386, "grad_norm": 1.2472173440990098, "learning_rate": 0.003, "loss": 4.0323, "step": 18386 }, { "epoch": 0.18387, "grad_norm": 1.0997827846903279, "learning_rate": 0.003, "loss": 4.0528, "step": 18387 }, { "epoch": 0.18388, "grad_norm": 1.5915345944293096, "learning_rate": 0.003, "loss": 4.0711, "step": 18388 }, { "epoch": 0.18389, "grad_norm": 1.0540192254937788, "learning_rate": 0.003, "loss": 4.0689, "step": 18389 }, { "epoch": 0.1839, "grad_norm": 1.349088059231571, "learning_rate": 0.003, "loss": 4.061, "step": 18390 }, { "epoch": 0.18391, "grad_norm": 1.0832034858737989, "learning_rate": 0.003, "loss": 4.0418, "step": 18391 }, { "epoch": 0.18392, "grad_norm": 1.3619004635892382, "learning_rate": 0.003, "loss": 4.0338, "step": 18392 }, { "epoch": 0.18393, "grad_norm": 1.0458881019397792, "learning_rate": 0.003, "loss": 4.0641, "step": 18393 }, { "epoch": 0.18394, "grad_norm": 1.3173913973229336, "learning_rate": 0.003, "loss": 4.042, "step": 18394 }, { "epoch": 0.18395, "grad_norm": 1.2713040555271644, "learning_rate": 0.003, "loss": 4.042, "step": 18395 }, { "epoch": 0.18396, "grad_norm": 1.4819357793803813, "learning_rate": 0.003, "loss": 4.0882, "step": 18396 }, { "epoch": 0.18397, "grad_norm": 1.091929480347712, "learning_rate": 0.003, "loss": 4.0757, "step": 18397 }, { "epoch": 0.18398, "grad_norm": 1.5549150779413397, "learning_rate": 0.003, "loss": 4.0523, "step": 18398 }, { "epoch": 0.18399, "grad_norm": 1.2504656996458987, "learning_rate": 0.003, "loss": 4.0475, "step": 18399 }, { "epoch": 0.184, "grad_norm": 1.2328847983295108, "learning_rate": 0.003, "loss": 4.0511, "step": 18400 }, { "epoch": 0.18401, "grad_norm": 1.1143207423651258, "learning_rate": 0.003, "loss": 4.0586, "step": 18401 }, { "epoch": 0.18402, "grad_norm": 1.1704892853025841, "learning_rate": 0.003, "loss": 4.0494, "step": 18402 }, { "epoch": 0.18403, "grad_norm": 1.2814790421906408, "learning_rate": 0.003, "loss": 4.0183, "step": 18403 }, { "epoch": 0.18404, "grad_norm": 1.0653977131709254, "learning_rate": 0.003, "loss": 4.0456, "step": 18404 }, { "epoch": 0.18405, "grad_norm": 1.4286666051912853, "learning_rate": 0.003, "loss": 4.0506, "step": 18405 }, { "epoch": 0.18406, "grad_norm": 1.2242947705176372, "learning_rate": 0.003, "loss": 4.0635, "step": 18406 }, { "epoch": 0.18407, "grad_norm": 1.2965782130321206, "learning_rate": 0.003, "loss": 4.0414, "step": 18407 }, { "epoch": 0.18408, "grad_norm": 1.364762697056736, "learning_rate": 0.003, "loss": 4.0649, "step": 18408 }, { "epoch": 0.18409, "grad_norm": 1.307486344537389, "learning_rate": 0.003, "loss": 4.0689, "step": 18409 }, { "epoch": 0.1841, "grad_norm": 1.1484488387476597, "learning_rate": 0.003, "loss": 4.0479, "step": 18410 }, { "epoch": 0.18411, "grad_norm": 1.4953546714049815, "learning_rate": 0.003, "loss": 4.077, "step": 18411 }, { "epoch": 0.18412, "grad_norm": 1.2321071738116949, "learning_rate": 0.003, "loss": 4.0826, "step": 18412 }, { "epoch": 0.18413, "grad_norm": 1.145249847322974, "learning_rate": 0.003, "loss": 4.0371, "step": 18413 }, { "epoch": 0.18414, "grad_norm": 1.2487494626464846, "learning_rate": 0.003, "loss": 4.0502, "step": 18414 }, { "epoch": 0.18415, "grad_norm": 1.0883548355445094, "learning_rate": 0.003, "loss": 4.0414, "step": 18415 }, { "epoch": 0.18416, "grad_norm": 1.2299959920294103, "learning_rate": 0.003, "loss": 4.0397, "step": 18416 }, { "epoch": 0.18417, "grad_norm": 1.2072193836859442, "learning_rate": 0.003, "loss": 4.0807, "step": 18417 }, { "epoch": 0.18418, "grad_norm": 1.1419089881924183, "learning_rate": 0.003, "loss": 4.0255, "step": 18418 }, { "epoch": 0.18419, "grad_norm": 1.291328812747233, "learning_rate": 0.003, "loss": 4.0352, "step": 18419 }, { "epoch": 0.1842, "grad_norm": 1.0516099710007751, "learning_rate": 0.003, "loss": 4.0684, "step": 18420 }, { "epoch": 0.18421, "grad_norm": 1.6372782377502058, "learning_rate": 0.003, "loss": 4.066, "step": 18421 }, { "epoch": 0.18422, "grad_norm": 1.1970379761625947, "learning_rate": 0.003, "loss": 4.0536, "step": 18422 }, { "epoch": 0.18423, "grad_norm": 1.3665717406752702, "learning_rate": 0.003, "loss": 4.0362, "step": 18423 }, { "epoch": 0.18424, "grad_norm": 1.050578301481084, "learning_rate": 0.003, "loss": 4.067, "step": 18424 }, { "epoch": 0.18425, "grad_norm": 1.3996093984009, "learning_rate": 0.003, "loss": 4.0168, "step": 18425 }, { "epoch": 0.18426, "grad_norm": 1.1612361973455871, "learning_rate": 0.003, "loss": 4.0494, "step": 18426 }, { "epoch": 0.18427, "grad_norm": 1.088657031452232, "learning_rate": 0.003, "loss": 4.0629, "step": 18427 }, { "epoch": 0.18428, "grad_norm": 1.4412719606226663, "learning_rate": 0.003, "loss": 4.0597, "step": 18428 }, { "epoch": 0.18429, "grad_norm": 1.1368657759095244, "learning_rate": 0.003, "loss": 4.0614, "step": 18429 }, { "epoch": 0.1843, "grad_norm": 1.1075714101497987, "learning_rate": 0.003, "loss": 4.0275, "step": 18430 }, { "epoch": 0.18431, "grad_norm": 1.2733237994192719, "learning_rate": 0.003, "loss": 4.0425, "step": 18431 }, { "epoch": 0.18432, "grad_norm": 1.1232848243686768, "learning_rate": 0.003, "loss": 4.0425, "step": 18432 }, { "epoch": 0.18433, "grad_norm": 1.4071867153137911, "learning_rate": 0.003, "loss": 4.0391, "step": 18433 }, { "epoch": 0.18434, "grad_norm": 1.2290071096289996, "learning_rate": 0.003, "loss": 4.056, "step": 18434 }, { "epoch": 0.18435, "grad_norm": 1.032723486247934, "learning_rate": 0.003, "loss": 4.0554, "step": 18435 }, { "epoch": 0.18436, "grad_norm": 1.2846682751341674, "learning_rate": 0.003, "loss": 4.0691, "step": 18436 }, { "epoch": 0.18437, "grad_norm": 1.0787708489962597, "learning_rate": 0.003, "loss": 4.0799, "step": 18437 }, { "epoch": 0.18438, "grad_norm": 1.464669613675582, "learning_rate": 0.003, "loss": 4.0523, "step": 18438 }, { "epoch": 0.18439, "grad_norm": 1.2595339936840588, "learning_rate": 0.003, "loss": 4.0359, "step": 18439 }, { "epoch": 0.1844, "grad_norm": 1.1910280353388745, "learning_rate": 0.003, "loss": 4.044, "step": 18440 }, { "epoch": 0.18441, "grad_norm": 1.401596779857551, "learning_rate": 0.003, "loss": 4.0522, "step": 18441 }, { "epoch": 0.18442, "grad_norm": 1.1663080858959465, "learning_rate": 0.003, "loss": 4.0167, "step": 18442 }, { "epoch": 0.18443, "grad_norm": 1.4213127571437272, "learning_rate": 0.003, "loss": 4.0708, "step": 18443 }, { "epoch": 0.18444, "grad_norm": 1.2784713167394093, "learning_rate": 0.003, "loss": 4.0669, "step": 18444 }, { "epoch": 0.18445, "grad_norm": 1.2699967101061032, "learning_rate": 0.003, "loss": 4.0579, "step": 18445 }, { "epoch": 0.18446, "grad_norm": 1.4811331385648756, "learning_rate": 0.003, "loss": 4.0613, "step": 18446 }, { "epoch": 0.18447, "grad_norm": 1.2417564796737708, "learning_rate": 0.003, "loss": 4.0657, "step": 18447 }, { "epoch": 0.18448, "grad_norm": 1.43635534847114, "learning_rate": 0.003, "loss": 4.0588, "step": 18448 }, { "epoch": 0.18449, "grad_norm": 1.215379106673042, "learning_rate": 0.003, "loss": 4.0475, "step": 18449 }, { "epoch": 0.1845, "grad_norm": 1.2157904836130864, "learning_rate": 0.003, "loss": 4.0369, "step": 18450 }, { "epoch": 0.18451, "grad_norm": 1.329311485228502, "learning_rate": 0.003, "loss": 4.0592, "step": 18451 }, { "epoch": 0.18452, "grad_norm": 1.0561661848277206, "learning_rate": 0.003, "loss": 4.0376, "step": 18452 }, { "epoch": 0.18453, "grad_norm": 1.2937290983083072, "learning_rate": 0.003, "loss": 4.095, "step": 18453 }, { "epoch": 0.18454, "grad_norm": 1.0068479492925069, "learning_rate": 0.003, "loss": 4.0526, "step": 18454 }, { "epoch": 0.18455, "grad_norm": 1.2853261379702334, "learning_rate": 0.003, "loss": 4.0276, "step": 18455 }, { "epoch": 0.18456, "grad_norm": 1.1472678057552477, "learning_rate": 0.003, "loss": 4.0605, "step": 18456 }, { "epoch": 0.18457, "grad_norm": 1.518079009396172, "learning_rate": 0.003, "loss": 4.0856, "step": 18457 }, { "epoch": 0.18458, "grad_norm": 1.0893014638648486, "learning_rate": 0.003, "loss": 4.0816, "step": 18458 }, { "epoch": 0.18459, "grad_norm": 1.395160488488644, "learning_rate": 0.003, "loss": 4.0403, "step": 18459 }, { "epoch": 0.1846, "grad_norm": 1.279374257931241, "learning_rate": 0.003, "loss": 4.0274, "step": 18460 }, { "epoch": 0.18461, "grad_norm": 1.268182481391018, "learning_rate": 0.003, "loss": 4.0531, "step": 18461 }, { "epoch": 0.18462, "grad_norm": 1.340110662770171, "learning_rate": 0.003, "loss": 4.061, "step": 18462 }, { "epoch": 0.18463, "grad_norm": 1.0870891437013017, "learning_rate": 0.003, "loss": 4.072, "step": 18463 }, { "epoch": 0.18464, "grad_norm": 1.3560092280503788, "learning_rate": 0.003, "loss": 4.0721, "step": 18464 }, { "epoch": 0.18465, "grad_norm": 1.146359035106978, "learning_rate": 0.003, "loss": 4.0757, "step": 18465 }, { "epoch": 0.18466, "grad_norm": 1.3008419389754802, "learning_rate": 0.003, "loss": 4.0464, "step": 18466 }, { "epoch": 0.18467, "grad_norm": 1.2050250922697034, "learning_rate": 0.003, "loss": 4.0698, "step": 18467 }, { "epoch": 0.18468, "grad_norm": 1.3916879294092328, "learning_rate": 0.003, "loss": 4.0726, "step": 18468 }, { "epoch": 0.18469, "grad_norm": 1.6267142323330905, "learning_rate": 0.003, "loss": 4.0887, "step": 18469 }, { "epoch": 0.1847, "grad_norm": 1.1109695202765093, "learning_rate": 0.003, "loss": 4.0271, "step": 18470 }, { "epoch": 0.18471, "grad_norm": 1.2428610086781493, "learning_rate": 0.003, "loss": 4.0652, "step": 18471 }, { "epoch": 0.18472, "grad_norm": 1.4164641965220681, "learning_rate": 0.003, "loss": 4.0431, "step": 18472 }, { "epoch": 0.18473, "grad_norm": 1.2862536050892301, "learning_rate": 0.003, "loss": 4.0522, "step": 18473 }, { "epoch": 0.18474, "grad_norm": 1.372252339990451, "learning_rate": 0.003, "loss": 4.0468, "step": 18474 }, { "epoch": 0.18475, "grad_norm": 1.2771637418850568, "learning_rate": 0.003, "loss": 4.0547, "step": 18475 }, { "epoch": 0.18476, "grad_norm": 0.9798307090773521, "learning_rate": 0.003, "loss": 4.0561, "step": 18476 }, { "epoch": 0.18477, "grad_norm": 1.370160757472685, "learning_rate": 0.003, "loss": 4.0408, "step": 18477 }, { "epoch": 0.18478, "grad_norm": 1.0822830726748474, "learning_rate": 0.003, "loss": 4.0604, "step": 18478 }, { "epoch": 0.18479, "grad_norm": 1.3446721670686375, "learning_rate": 0.003, "loss": 4.0447, "step": 18479 }, { "epoch": 0.1848, "grad_norm": 1.2880062992123016, "learning_rate": 0.003, "loss": 4.0199, "step": 18480 }, { "epoch": 0.18481, "grad_norm": 1.1764977833596155, "learning_rate": 0.003, "loss": 4.0529, "step": 18481 }, { "epoch": 0.18482, "grad_norm": 1.2670672336263902, "learning_rate": 0.003, "loss": 4.0202, "step": 18482 }, { "epoch": 0.18483, "grad_norm": 1.3699901726556798, "learning_rate": 0.003, "loss": 4.0751, "step": 18483 }, { "epoch": 0.18484, "grad_norm": 1.1748915505926445, "learning_rate": 0.003, "loss": 4.0331, "step": 18484 }, { "epoch": 0.18485, "grad_norm": 1.162132788634817, "learning_rate": 0.003, "loss": 4.0494, "step": 18485 }, { "epoch": 0.18486, "grad_norm": 1.1761824497637563, "learning_rate": 0.003, "loss": 4.0121, "step": 18486 }, { "epoch": 0.18487, "grad_norm": 1.1980158306891817, "learning_rate": 0.003, "loss": 4.0391, "step": 18487 }, { "epoch": 0.18488, "grad_norm": 1.2258039379616448, "learning_rate": 0.003, "loss": 4.0494, "step": 18488 }, { "epoch": 0.18489, "grad_norm": 1.4208962807298569, "learning_rate": 0.003, "loss": 4.0641, "step": 18489 }, { "epoch": 0.1849, "grad_norm": 1.0200113458498734, "learning_rate": 0.003, "loss": 4.0641, "step": 18490 }, { "epoch": 0.18491, "grad_norm": 1.4595091737327908, "learning_rate": 0.003, "loss": 4.0537, "step": 18491 }, { "epoch": 0.18492, "grad_norm": 1.3168285517146034, "learning_rate": 0.003, "loss": 4.0407, "step": 18492 }, { "epoch": 0.18493, "grad_norm": 1.4532712875747906, "learning_rate": 0.003, "loss": 4.0608, "step": 18493 }, { "epoch": 0.18494, "grad_norm": 1.0459214766605567, "learning_rate": 0.003, "loss": 4.063, "step": 18494 }, { "epoch": 0.18495, "grad_norm": 1.1381218722401705, "learning_rate": 0.003, "loss": 4.0448, "step": 18495 }, { "epoch": 0.18496, "grad_norm": 1.5386479772919055, "learning_rate": 0.003, "loss": 4.0165, "step": 18496 }, { "epoch": 0.18497, "grad_norm": 1.2299892544229238, "learning_rate": 0.003, "loss": 4.0332, "step": 18497 }, { "epoch": 0.18498, "grad_norm": 1.2335831638757717, "learning_rate": 0.003, "loss": 4.0474, "step": 18498 }, { "epoch": 0.18499, "grad_norm": 1.2669312066929301, "learning_rate": 0.003, "loss": 4.0605, "step": 18499 }, { "epoch": 0.185, "grad_norm": 1.087188823010408, "learning_rate": 0.003, "loss": 4.069, "step": 18500 }, { "epoch": 0.18501, "grad_norm": 1.1783337651243124, "learning_rate": 0.003, "loss": 4.0173, "step": 18501 }, { "epoch": 0.18502, "grad_norm": 1.0736642518059751, "learning_rate": 0.003, "loss": 4.038, "step": 18502 }, { "epoch": 0.18503, "grad_norm": 1.4393107138897194, "learning_rate": 0.003, "loss": 4.0566, "step": 18503 }, { "epoch": 0.18504, "grad_norm": 1.1090883746507503, "learning_rate": 0.003, "loss": 4.0567, "step": 18504 }, { "epoch": 0.18505, "grad_norm": 1.4161606762569607, "learning_rate": 0.003, "loss": 4.0343, "step": 18505 }, { "epoch": 0.18506, "grad_norm": 1.2049783538307672, "learning_rate": 0.003, "loss": 4.0813, "step": 18506 }, { "epoch": 0.18507, "grad_norm": 1.2081699483481163, "learning_rate": 0.003, "loss": 4.0377, "step": 18507 }, { "epoch": 0.18508, "grad_norm": 1.2776863645276184, "learning_rate": 0.003, "loss": 4.0615, "step": 18508 }, { "epoch": 0.18509, "grad_norm": 1.421387032589169, "learning_rate": 0.003, "loss": 4.0356, "step": 18509 }, { "epoch": 0.1851, "grad_norm": 1.1131341934615544, "learning_rate": 0.003, "loss": 4.0808, "step": 18510 }, { "epoch": 0.18511, "grad_norm": 1.38719870177101, "learning_rate": 0.003, "loss": 4.0964, "step": 18511 }, { "epoch": 0.18512, "grad_norm": 1.3032146929287052, "learning_rate": 0.003, "loss": 4.0682, "step": 18512 }, { "epoch": 0.18513, "grad_norm": 1.2852777986544128, "learning_rate": 0.003, "loss": 4.0673, "step": 18513 }, { "epoch": 0.18514, "grad_norm": 1.0331253160795906, "learning_rate": 0.003, "loss": 4.078, "step": 18514 }, { "epoch": 0.18515, "grad_norm": 1.5607563072743642, "learning_rate": 0.003, "loss": 4.0252, "step": 18515 }, { "epoch": 0.18516, "grad_norm": 1.311212303194748, "learning_rate": 0.003, "loss": 4.0625, "step": 18516 }, { "epoch": 0.18517, "grad_norm": 1.1417369515028657, "learning_rate": 0.003, "loss": 4.0521, "step": 18517 }, { "epoch": 0.18518, "grad_norm": 1.4972229558162757, "learning_rate": 0.003, "loss": 4.034, "step": 18518 }, { "epoch": 0.18519, "grad_norm": 1.1670225381994737, "learning_rate": 0.003, "loss": 4.0402, "step": 18519 }, { "epoch": 0.1852, "grad_norm": 1.6450574769424848, "learning_rate": 0.003, "loss": 4.0519, "step": 18520 }, { "epoch": 0.18521, "grad_norm": 1.047966343768503, "learning_rate": 0.003, "loss": 4.0548, "step": 18521 }, { "epoch": 0.18522, "grad_norm": 1.26719944704278, "learning_rate": 0.003, "loss": 4.054, "step": 18522 }, { "epoch": 0.18523, "grad_norm": 1.1959792480142655, "learning_rate": 0.003, "loss": 4.0443, "step": 18523 }, { "epoch": 0.18524, "grad_norm": 1.3138523037586216, "learning_rate": 0.003, "loss": 4.0739, "step": 18524 }, { "epoch": 0.18525, "grad_norm": 1.0237531776331572, "learning_rate": 0.003, "loss": 4.0431, "step": 18525 }, { "epoch": 0.18526, "grad_norm": 1.286349474483322, "learning_rate": 0.003, "loss": 4.0486, "step": 18526 }, { "epoch": 0.18527, "grad_norm": 1.2670213943868798, "learning_rate": 0.003, "loss": 4.0671, "step": 18527 }, { "epoch": 0.18528, "grad_norm": 1.1188335528948643, "learning_rate": 0.003, "loss": 4.075, "step": 18528 }, { "epoch": 0.18529, "grad_norm": 1.4593728460814455, "learning_rate": 0.003, "loss": 4.0379, "step": 18529 }, { "epoch": 0.1853, "grad_norm": 1.3874596643152133, "learning_rate": 0.003, "loss": 4.0528, "step": 18530 }, { "epoch": 0.18531, "grad_norm": 1.5584581857765323, "learning_rate": 0.003, "loss": 4.0444, "step": 18531 }, { "epoch": 0.18532, "grad_norm": 1.0277005199484917, "learning_rate": 0.003, "loss": 4.0717, "step": 18532 }, { "epoch": 0.18533, "grad_norm": 1.381896949372854, "learning_rate": 0.003, "loss": 4.0842, "step": 18533 }, { "epoch": 0.18534, "grad_norm": 1.2713297855170362, "learning_rate": 0.003, "loss": 4.0626, "step": 18534 }, { "epoch": 0.18535, "grad_norm": 1.1657663540513583, "learning_rate": 0.003, "loss": 4.074, "step": 18535 }, { "epoch": 0.18536, "grad_norm": 1.3492976594869548, "learning_rate": 0.003, "loss": 4.0452, "step": 18536 }, { "epoch": 0.18537, "grad_norm": 1.0301590695489486, "learning_rate": 0.003, "loss": 4.0604, "step": 18537 }, { "epoch": 0.18538, "grad_norm": 1.2991531867485961, "learning_rate": 0.003, "loss": 4.0597, "step": 18538 }, { "epoch": 0.18539, "grad_norm": 1.2309409745880733, "learning_rate": 0.003, "loss": 4.0689, "step": 18539 }, { "epoch": 0.1854, "grad_norm": 1.6172072791048684, "learning_rate": 0.003, "loss": 4.0699, "step": 18540 }, { "epoch": 0.18541, "grad_norm": 1.038151461419373, "learning_rate": 0.003, "loss": 4.0387, "step": 18541 }, { "epoch": 0.18542, "grad_norm": 1.6681079800201724, "learning_rate": 0.003, "loss": 4.0593, "step": 18542 }, { "epoch": 0.18543, "grad_norm": 0.9142406340517287, "learning_rate": 0.003, "loss": 4.0533, "step": 18543 }, { "epoch": 0.18544, "grad_norm": 1.1290954549194858, "learning_rate": 0.003, "loss": 4.0641, "step": 18544 }, { "epoch": 0.18545, "grad_norm": 1.2390811169384381, "learning_rate": 0.003, "loss": 4.0248, "step": 18545 }, { "epoch": 0.18546, "grad_norm": 1.3684541946848185, "learning_rate": 0.003, "loss": 4.0418, "step": 18546 }, { "epoch": 0.18547, "grad_norm": 0.9376309640085142, "learning_rate": 0.003, "loss": 4.0693, "step": 18547 }, { "epoch": 0.18548, "grad_norm": 1.169731857420868, "learning_rate": 0.003, "loss": 4.0693, "step": 18548 }, { "epoch": 0.18549, "grad_norm": 1.2633582357108424, "learning_rate": 0.003, "loss": 4.0721, "step": 18549 }, { "epoch": 0.1855, "grad_norm": 1.2898919233025385, "learning_rate": 0.003, "loss": 4.0591, "step": 18550 }, { "epoch": 0.18551, "grad_norm": 1.2996495701078847, "learning_rate": 0.003, "loss": 4.0489, "step": 18551 }, { "epoch": 0.18552, "grad_norm": 1.645188039124316, "learning_rate": 0.003, "loss": 4.0945, "step": 18552 }, { "epoch": 0.18553, "grad_norm": 0.9700590043202127, "learning_rate": 0.003, "loss": 4.0469, "step": 18553 }, { "epoch": 0.18554, "grad_norm": 1.3383196501748178, "learning_rate": 0.003, "loss": 4.052, "step": 18554 }, { "epoch": 0.18555, "grad_norm": 1.4447324218137332, "learning_rate": 0.003, "loss": 4.0572, "step": 18555 }, { "epoch": 0.18556, "grad_norm": 1.2238794234024961, "learning_rate": 0.003, "loss": 4.0674, "step": 18556 }, { "epoch": 0.18557, "grad_norm": 1.228237793895757, "learning_rate": 0.003, "loss": 4.0263, "step": 18557 }, { "epoch": 0.18558, "grad_norm": 1.2377248740570805, "learning_rate": 0.003, "loss": 4.0409, "step": 18558 }, { "epoch": 0.18559, "grad_norm": 1.1775863527255352, "learning_rate": 0.003, "loss": 4.0388, "step": 18559 }, { "epoch": 0.1856, "grad_norm": 1.2862630217843478, "learning_rate": 0.003, "loss": 4.0472, "step": 18560 }, { "epoch": 0.18561, "grad_norm": 1.2725546242503096, "learning_rate": 0.003, "loss": 4.0379, "step": 18561 }, { "epoch": 0.18562, "grad_norm": 1.0084923210327843, "learning_rate": 0.003, "loss": 4.0406, "step": 18562 }, { "epoch": 0.18563, "grad_norm": 1.4549473612904706, "learning_rate": 0.003, "loss": 4.023, "step": 18563 }, { "epoch": 0.18564, "grad_norm": 1.0898082300447982, "learning_rate": 0.003, "loss": 4.0291, "step": 18564 }, { "epoch": 0.18565, "grad_norm": 1.4204817236107417, "learning_rate": 0.003, "loss": 4.0704, "step": 18565 }, { "epoch": 0.18566, "grad_norm": 1.1306017397286994, "learning_rate": 0.003, "loss": 4.0326, "step": 18566 }, { "epoch": 0.18567, "grad_norm": 1.3545883527371785, "learning_rate": 0.003, "loss": 4.05, "step": 18567 }, { "epoch": 0.18568, "grad_norm": 1.0700768308961939, "learning_rate": 0.003, "loss": 4.0637, "step": 18568 }, { "epoch": 0.18569, "grad_norm": 1.253490566049985, "learning_rate": 0.003, "loss": 4.0632, "step": 18569 }, { "epoch": 0.1857, "grad_norm": 1.1080382610625297, "learning_rate": 0.003, "loss": 4.0698, "step": 18570 }, { "epoch": 0.18571, "grad_norm": 1.4826227076272758, "learning_rate": 0.003, "loss": 4.0794, "step": 18571 }, { "epoch": 0.18572, "grad_norm": 1.0275343283728213, "learning_rate": 0.003, "loss": 4.0492, "step": 18572 }, { "epoch": 0.18573, "grad_norm": 1.5618169718041988, "learning_rate": 0.003, "loss": 4.0668, "step": 18573 }, { "epoch": 0.18574, "grad_norm": 1.2097067630440304, "learning_rate": 0.003, "loss": 4.0648, "step": 18574 }, { "epoch": 0.18575, "grad_norm": 1.2138181501319916, "learning_rate": 0.003, "loss": 4.0855, "step": 18575 }, { "epoch": 0.18576, "grad_norm": 1.2300603103536538, "learning_rate": 0.003, "loss": 4.0652, "step": 18576 }, { "epoch": 0.18577, "grad_norm": 1.1092966404884428, "learning_rate": 0.003, "loss": 4.0314, "step": 18577 }, { "epoch": 0.18578, "grad_norm": 1.4220184020364897, "learning_rate": 0.003, "loss": 4.0344, "step": 18578 }, { "epoch": 0.18579, "grad_norm": 1.508446898391007, "learning_rate": 0.003, "loss": 4.0866, "step": 18579 }, { "epoch": 0.1858, "grad_norm": 1.4423422577068636, "learning_rate": 0.003, "loss": 4.0288, "step": 18580 }, { "epoch": 0.18581, "grad_norm": 1.0395036848085926, "learning_rate": 0.003, "loss": 4.0252, "step": 18581 }, { "epoch": 0.18582, "grad_norm": 1.4666658372450543, "learning_rate": 0.003, "loss": 4.0619, "step": 18582 }, { "epoch": 0.18583, "grad_norm": 1.2325909275221758, "learning_rate": 0.003, "loss": 4.0496, "step": 18583 }, { "epoch": 0.18584, "grad_norm": 1.3082951147599138, "learning_rate": 0.003, "loss": 4.0454, "step": 18584 }, { "epoch": 0.18585, "grad_norm": 1.1768231850320106, "learning_rate": 0.003, "loss": 4.0395, "step": 18585 }, { "epoch": 0.18586, "grad_norm": 1.3390239461857918, "learning_rate": 0.003, "loss": 4.0385, "step": 18586 }, { "epoch": 0.18587, "grad_norm": 1.299623772556699, "learning_rate": 0.003, "loss": 4.0322, "step": 18587 }, { "epoch": 0.18588, "grad_norm": 1.255595666077245, "learning_rate": 0.003, "loss": 4.0437, "step": 18588 }, { "epoch": 0.18589, "grad_norm": 1.4111729802737794, "learning_rate": 0.003, "loss": 4.0593, "step": 18589 }, { "epoch": 0.1859, "grad_norm": 1.0180266304943313, "learning_rate": 0.003, "loss": 4.0529, "step": 18590 }, { "epoch": 0.18591, "grad_norm": 1.3972753800116477, "learning_rate": 0.003, "loss": 4.0291, "step": 18591 }, { "epoch": 0.18592, "grad_norm": 1.1621225220737084, "learning_rate": 0.003, "loss": 4.0592, "step": 18592 }, { "epoch": 0.18593, "grad_norm": 1.3732411657292347, "learning_rate": 0.003, "loss": 4.0797, "step": 18593 }, { "epoch": 0.18594, "grad_norm": 1.209977099499059, "learning_rate": 0.003, "loss": 4.0723, "step": 18594 }, { "epoch": 0.18595, "grad_norm": 1.0367443408157266, "learning_rate": 0.003, "loss": 4.0569, "step": 18595 }, { "epoch": 0.18596, "grad_norm": 1.180761823307239, "learning_rate": 0.003, "loss": 4.0892, "step": 18596 }, { "epoch": 0.18597, "grad_norm": 1.2569061231225465, "learning_rate": 0.003, "loss": 4.083, "step": 18597 }, { "epoch": 0.18598, "grad_norm": 1.2470900105281038, "learning_rate": 0.003, "loss": 4.0857, "step": 18598 }, { "epoch": 0.18599, "grad_norm": 1.1336668930254277, "learning_rate": 0.003, "loss": 4.0351, "step": 18599 }, { "epoch": 0.186, "grad_norm": 1.409078569317884, "learning_rate": 0.003, "loss": 4.0508, "step": 18600 }, { "epoch": 0.18601, "grad_norm": 0.997962942227669, "learning_rate": 0.003, "loss": 4.0265, "step": 18601 }, { "epoch": 0.18602, "grad_norm": 1.460422818822215, "learning_rate": 0.003, "loss": 4.0311, "step": 18602 }, { "epoch": 0.18603, "grad_norm": 1.0546011794234764, "learning_rate": 0.003, "loss": 4.021, "step": 18603 }, { "epoch": 0.18604, "grad_norm": 1.5160687672117177, "learning_rate": 0.003, "loss": 4.0889, "step": 18604 }, { "epoch": 0.18605, "grad_norm": 1.2771163674586319, "learning_rate": 0.003, "loss": 4.0693, "step": 18605 }, { "epoch": 0.18606, "grad_norm": 1.3208345170662767, "learning_rate": 0.003, "loss": 4.0774, "step": 18606 }, { "epoch": 0.18607, "grad_norm": 1.1516522908857025, "learning_rate": 0.003, "loss": 4.0784, "step": 18607 }, { "epoch": 0.18608, "grad_norm": 1.2046464737978206, "learning_rate": 0.003, "loss": 4.0505, "step": 18608 }, { "epoch": 0.18609, "grad_norm": 1.5262495264400762, "learning_rate": 0.003, "loss": 4.0718, "step": 18609 }, { "epoch": 0.1861, "grad_norm": 1.0641298642658337, "learning_rate": 0.003, "loss": 4.0796, "step": 18610 }, { "epoch": 0.18611, "grad_norm": 1.5932848573565546, "learning_rate": 0.003, "loss": 4.061, "step": 18611 }, { "epoch": 0.18612, "grad_norm": 1.002904295640288, "learning_rate": 0.003, "loss": 4.0638, "step": 18612 }, { "epoch": 0.18613, "grad_norm": 1.4969747288804227, "learning_rate": 0.003, "loss": 4.0971, "step": 18613 }, { "epoch": 0.18614, "grad_norm": 1.1470765262778144, "learning_rate": 0.003, "loss": 4.0178, "step": 18614 }, { "epoch": 0.18615, "grad_norm": 1.1680529552859376, "learning_rate": 0.003, "loss": 4.0795, "step": 18615 }, { "epoch": 0.18616, "grad_norm": 1.1606542101751762, "learning_rate": 0.003, "loss": 4.0451, "step": 18616 }, { "epoch": 0.18617, "grad_norm": 1.1986020675670659, "learning_rate": 0.003, "loss": 4.0604, "step": 18617 }, { "epoch": 0.18618, "grad_norm": 1.4659025348896637, "learning_rate": 0.003, "loss": 4.0785, "step": 18618 }, { "epoch": 0.18619, "grad_norm": 1.1797749440013026, "learning_rate": 0.003, "loss": 4.0553, "step": 18619 }, { "epoch": 0.1862, "grad_norm": 1.3628153636650342, "learning_rate": 0.003, "loss": 4.0783, "step": 18620 }, { "epoch": 0.18621, "grad_norm": 1.2827901915534006, "learning_rate": 0.003, "loss": 4.0537, "step": 18621 }, { "epoch": 0.18622, "grad_norm": 1.388778940140245, "learning_rate": 0.003, "loss": 4.0558, "step": 18622 }, { "epoch": 0.18623, "grad_norm": 1.1919220173396643, "learning_rate": 0.003, "loss": 4.0463, "step": 18623 }, { "epoch": 0.18624, "grad_norm": 1.336295009143112, "learning_rate": 0.003, "loss": 4.0546, "step": 18624 }, { "epoch": 0.18625, "grad_norm": 1.2187632702612323, "learning_rate": 0.003, "loss": 4.053, "step": 18625 }, { "epoch": 0.18626, "grad_norm": 1.3712027530620414, "learning_rate": 0.003, "loss": 4.0601, "step": 18626 }, { "epoch": 0.18627, "grad_norm": 1.2382458749170933, "learning_rate": 0.003, "loss": 4.0743, "step": 18627 }, { "epoch": 0.18628, "grad_norm": 1.1962023404453752, "learning_rate": 0.003, "loss": 4.0394, "step": 18628 }, { "epoch": 0.18629, "grad_norm": 1.283440987751412, "learning_rate": 0.003, "loss": 4.0462, "step": 18629 }, { "epoch": 0.1863, "grad_norm": 1.0614466943285759, "learning_rate": 0.003, "loss": 4.0445, "step": 18630 }, { "epoch": 0.18631, "grad_norm": 1.3246924380965295, "learning_rate": 0.003, "loss": 4.0879, "step": 18631 }, { "epoch": 0.18632, "grad_norm": 1.295372257801994, "learning_rate": 0.003, "loss": 4.0391, "step": 18632 }, { "epoch": 0.18633, "grad_norm": 1.2585393271056065, "learning_rate": 0.003, "loss": 4.0473, "step": 18633 }, { "epoch": 0.18634, "grad_norm": 1.4536916864523137, "learning_rate": 0.003, "loss": 4.0631, "step": 18634 }, { "epoch": 0.18635, "grad_norm": 1.322218953388777, "learning_rate": 0.003, "loss": 4.0834, "step": 18635 }, { "epoch": 0.18636, "grad_norm": 1.3423310818076708, "learning_rate": 0.003, "loss": 4.0405, "step": 18636 }, { "epoch": 0.18637, "grad_norm": 1.1432741798710784, "learning_rate": 0.003, "loss": 4.0688, "step": 18637 }, { "epoch": 0.18638, "grad_norm": 1.4349547761611217, "learning_rate": 0.003, "loss": 4.065, "step": 18638 }, { "epoch": 0.18639, "grad_norm": 1.0468928856209745, "learning_rate": 0.003, "loss": 4.0758, "step": 18639 }, { "epoch": 0.1864, "grad_norm": 1.2412971534074615, "learning_rate": 0.003, "loss": 4.0566, "step": 18640 }, { "epoch": 0.18641, "grad_norm": 1.2683840160903244, "learning_rate": 0.003, "loss": 4.0335, "step": 18641 }, { "epoch": 0.18642, "grad_norm": 1.2784065112434306, "learning_rate": 0.003, "loss": 4.0321, "step": 18642 }, { "epoch": 0.18643, "grad_norm": 1.0127344608529372, "learning_rate": 0.003, "loss": 4.0646, "step": 18643 }, { "epoch": 0.18644, "grad_norm": 1.3747573847249897, "learning_rate": 0.003, "loss": 4.0402, "step": 18644 }, { "epoch": 0.18645, "grad_norm": 1.0808831372998855, "learning_rate": 0.003, "loss": 4.041, "step": 18645 }, { "epoch": 0.18646, "grad_norm": 1.420689805250137, "learning_rate": 0.003, "loss": 4.0455, "step": 18646 }, { "epoch": 0.18647, "grad_norm": 1.345166995863017, "learning_rate": 0.003, "loss": 4.0616, "step": 18647 }, { "epoch": 0.18648, "grad_norm": 1.2971319501526732, "learning_rate": 0.003, "loss": 4.0404, "step": 18648 }, { "epoch": 0.18649, "grad_norm": 1.2443932292658246, "learning_rate": 0.003, "loss": 4.0762, "step": 18649 }, { "epoch": 0.1865, "grad_norm": 1.2414565982922923, "learning_rate": 0.003, "loss": 4.0727, "step": 18650 }, { "epoch": 0.18651, "grad_norm": 1.34014236981556, "learning_rate": 0.003, "loss": 4.0474, "step": 18651 }, { "epoch": 0.18652, "grad_norm": 1.0791936562175262, "learning_rate": 0.003, "loss": 4.0719, "step": 18652 }, { "epoch": 0.18653, "grad_norm": 1.2294366527608758, "learning_rate": 0.003, "loss": 4.0324, "step": 18653 }, { "epoch": 0.18654, "grad_norm": 1.2654637882095023, "learning_rate": 0.003, "loss": 4.0393, "step": 18654 }, { "epoch": 0.18655, "grad_norm": 1.3101259069674958, "learning_rate": 0.003, "loss": 4.0515, "step": 18655 }, { "epoch": 0.18656, "grad_norm": 1.2869356795084852, "learning_rate": 0.003, "loss": 4.0594, "step": 18656 }, { "epoch": 0.18657, "grad_norm": 1.2162196991906062, "learning_rate": 0.003, "loss": 4.0614, "step": 18657 }, { "epoch": 0.18658, "grad_norm": 1.2460610463275514, "learning_rate": 0.003, "loss": 4.0515, "step": 18658 }, { "epoch": 0.18659, "grad_norm": 1.1774219485727546, "learning_rate": 0.003, "loss": 4.0505, "step": 18659 }, { "epoch": 0.1866, "grad_norm": 1.4596058612447504, "learning_rate": 0.003, "loss": 4.0378, "step": 18660 }, { "epoch": 0.18661, "grad_norm": 0.9977181994708096, "learning_rate": 0.003, "loss": 4.0196, "step": 18661 }, { "epoch": 0.18662, "grad_norm": 1.5588672641369112, "learning_rate": 0.003, "loss": 4.0575, "step": 18662 }, { "epoch": 0.18663, "grad_norm": 1.0160923890565297, "learning_rate": 0.003, "loss": 4.0609, "step": 18663 }, { "epoch": 0.18664, "grad_norm": 1.3804578276728894, "learning_rate": 0.003, "loss": 4.0563, "step": 18664 }, { "epoch": 0.18665, "grad_norm": 1.5290662221388045, "learning_rate": 0.003, "loss": 4.0536, "step": 18665 }, { "epoch": 0.18666, "grad_norm": 1.1849130713284057, "learning_rate": 0.003, "loss": 4.0653, "step": 18666 }, { "epoch": 0.18667, "grad_norm": 1.32853994281615, "learning_rate": 0.003, "loss": 4.0657, "step": 18667 }, { "epoch": 0.18668, "grad_norm": 1.1182458060812699, "learning_rate": 0.003, "loss": 4.0628, "step": 18668 }, { "epoch": 0.18669, "grad_norm": 1.5520326454039035, "learning_rate": 0.003, "loss": 4.0416, "step": 18669 }, { "epoch": 0.1867, "grad_norm": 1.019058986085553, "learning_rate": 0.003, "loss": 4.047, "step": 18670 }, { "epoch": 0.18671, "grad_norm": 1.4160518584743231, "learning_rate": 0.003, "loss": 4.0708, "step": 18671 }, { "epoch": 0.18672, "grad_norm": 1.0368478405689707, "learning_rate": 0.003, "loss": 4.0558, "step": 18672 }, { "epoch": 0.18673, "grad_norm": 1.3670502550654924, "learning_rate": 0.003, "loss": 4.0644, "step": 18673 }, { "epoch": 0.18674, "grad_norm": 1.126542538632425, "learning_rate": 0.003, "loss": 4.0353, "step": 18674 }, { "epoch": 0.18675, "grad_norm": 1.2764599594968868, "learning_rate": 0.003, "loss": 4.0516, "step": 18675 }, { "epoch": 0.18676, "grad_norm": 1.3132500344481821, "learning_rate": 0.003, "loss": 4.0522, "step": 18676 }, { "epoch": 0.18677, "grad_norm": 1.201393565565788, "learning_rate": 0.003, "loss": 4.0642, "step": 18677 }, { "epoch": 0.18678, "grad_norm": 1.1121205736381374, "learning_rate": 0.003, "loss": 4.0534, "step": 18678 }, { "epoch": 0.18679, "grad_norm": 1.292409613879842, "learning_rate": 0.003, "loss": 4.0398, "step": 18679 }, { "epoch": 0.1868, "grad_norm": 1.1330555155779942, "learning_rate": 0.003, "loss": 4.0507, "step": 18680 }, { "epoch": 0.18681, "grad_norm": 1.5906447922596547, "learning_rate": 0.003, "loss": 4.0628, "step": 18681 }, { "epoch": 0.18682, "grad_norm": 1.1764539128829046, "learning_rate": 0.003, "loss": 4.056, "step": 18682 }, { "epoch": 0.18683, "grad_norm": 1.474986996228695, "learning_rate": 0.003, "loss": 4.0511, "step": 18683 }, { "epoch": 0.18684, "grad_norm": 1.2686172753821547, "learning_rate": 0.003, "loss": 4.076, "step": 18684 }, { "epoch": 0.18685, "grad_norm": 1.0932632235147117, "learning_rate": 0.003, "loss": 4.0444, "step": 18685 }, { "epoch": 0.18686, "grad_norm": 1.3377980575479242, "learning_rate": 0.003, "loss": 4.0669, "step": 18686 }, { "epoch": 0.18687, "grad_norm": 1.1281891101466395, "learning_rate": 0.003, "loss": 4.0819, "step": 18687 }, { "epoch": 0.18688, "grad_norm": 1.2612420416188836, "learning_rate": 0.003, "loss": 4.0531, "step": 18688 }, { "epoch": 0.18689, "grad_norm": 1.3658504975467765, "learning_rate": 0.003, "loss": 4.0335, "step": 18689 }, { "epoch": 0.1869, "grad_norm": 1.0824198490941663, "learning_rate": 0.003, "loss": 4.0523, "step": 18690 }, { "epoch": 0.18691, "grad_norm": 1.5137347285081706, "learning_rate": 0.003, "loss": 4.0726, "step": 18691 }, { "epoch": 0.18692, "grad_norm": 1.199086493296236, "learning_rate": 0.003, "loss": 4.0419, "step": 18692 }, { "epoch": 0.18693, "grad_norm": 1.221970076674539, "learning_rate": 0.003, "loss": 4.051, "step": 18693 }, { "epoch": 0.18694, "grad_norm": 1.4117607798223235, "learning_rate": 0.003, "loss": 4.0841, "step": 18694 }, { "epoch": 0.18695, "grad_norm": 1.3890068014115582, "learning_rate": 0.003, "loss": 4.0411, "step": 18695 }, { "epoch": 0.18696, "grad_norm": 1.3429995283722804, "learning_rate": 0.003, "loss": 4.0734, "step": 18696 }, { "epoch": 0.18697, "grad_norm": 1.1531051577775693, "learning_rate": 0.003, "loss": 4.0332, "step": 18697 }, { "epoch": 0.18698, "grad_norm": 1.3306981034164977, "learning_rate": 0.003, "loss": 4.0587, "step": 18698 }, { "epoch": 0.18699, "grad_norm": 1.343078886524723, "learning_rate": 0.003, "loss": 4.0532, "step": 18699 }, { "epoch": 0.187, "grad_norm": 1.0108194495872993, "learning_rate": 0.003, "loss": 4.0373, "step": 18700 }, { "epoch": 0.18701, "grad_norm": 1.3489124319503016, "learning_rate": 0.003, "loss": 4.0603, "step": 18701 }, { "epoch": 0.18702, "grad_norm": 1.0487537741166186, "learning_rate": 0.003, "loss": 4.0531, "step": 18702 }, { "epoch": 0.18703, "grad_norm": 1.1623226772074409, "learning_rate": 0.003, "loss": 4.0537, "step": 18703 }, { "epoch": 0.18704, "grad_norm": 1.3563915104551696, "learning_rate": 0.003, "loss": 4.042, "step": 18704 }, { "epoch": 0.18705, "grad_norm": 1.4090253126103949, "learning_rate": 0.003, "loss": 4.0704, "step": 18705 }, { "epoch": 0.18706, "grad_norm": 1.0781544362157394, "learning_rate": 0.003, "loss": 4.0305, "step": 18706 }, { "epoch": 0.18707, "grad_norm": 1.4090107424738676, "learning_rate": 0.003, "loss": 4.0452, "step": 18707 }, { "epoch": 0.18708, "grad_norm": 1.1798980777848695, "learning_rate": 0.003, "loss": 4.0192, "step": 18708 }, { "epoch": 0.18709, "grad_norm": 1.2834313985151309, "learning_rate": 0.003, "loss": 4.0559, "step": 18709 }, { "epoch": 0.1871, "grad_norm": 1.2516894700577164, "learning_rate": 0.003, "loss": 4.0498, "step": 18710 }, { "epoch": 0.18711, "grad_norm": 1.265294176703849, "learning_rate": 0.003, "loss": 4.0532, "step": 18711 }, { "epoch": 0.18712, "grad_norm": 1.183201878916828, "learning_rate": 0.003, "loss": 4.0652, "step": 18712 }, { "epoch": 0.18713, "grad_norm": 1.4076155148906417, "learning_rate": 0.003, "loss": 4.0699, "step": 18713 }, { "epoch": 0.18714, "grad_norm": 1.1285627861130147, "learning_rate": 0.003, "loss": 4.0349, "step": 18714 }, { "epoch": 0.18715, "grad_norm": 1.1434289243119355, "learning_rate": 0.003, "loss": 4.0618, "step": 18715 }, { "epoch": 0.18716, "grad_norm": 1.1301370706376264, "learning_rate": 0.003, "loss": 4.0428, "step": 18716 }, { "epoch": 0.18717, "grad_norm": 1.379301811151608, "learning_rate": 0.003, "loss": 4.0215, "step": 18717 }, { "epoch": 0.18718, "grad_norm": 1.111389298349418, "learning_rate": 0.003, "loss": 4.0303, "step": 18718 }, { "epoch": 0.18719, "grad_norm": 1.4741722058225644, "learning_rate": 0.003, "loss": 4.072, "step": 18719 }, { "epoch": 0.1872, "grad_norm": 1.1450588816578233, "learning_rate": 0.003, "loss": 4.0645, "step": 18720 }, { "epoch": 0.18721, "grad_norm": 1.418618013788135, "learning_rate": 0.003, "loss": 4.0515, "step": 18721 }, { "epoch": 0.18722, "grad_norm": 1.097306672519934, "learning_rate": 0.003, "loss": 4.0454, "step": 18722 }, { "epoch": 0.18723, "grad_norm": 1.1003636866584108, "learning_rate": 0.003, "loss": 4.0752, "step": 18723 }, { "epoch": 0.18724, "grad_norm": 1.328837159162091, "learning_rate": 0.003, "loss": 4.0509, "step": 18724 }, { "epoch": 0.18725, "grad_norm": 1.0331274012525158, "learning_rate": 0.003, "loss": 4.0411, "step": 18725 }, { "epoch": 0.18726, "grad_norm": 1.2773483347630517, "learning_rate": 0.003, "loss": 4.0592, "step": 18726 }, { "epoch": 0.18727, "grad_norm": 1.14599339799992, "learning_rate": 0.003, "loss": 4.0303, "step": 18727 }, { "epoch": 0.18728, "grad_norm": 1.4346416890229334, "learning_rate": 0.003, "loss": 4.0651, "step": 18728 }, { "epoch": 0.18729, "grad_norm": 1.407287027352288, "learning_rate": 0.003, "loss": 4.0964, "step": 18729 }, { "epoch": 0.1873, "grad_norm": 1.2160554604571705, "learning_rate": 0.003, "loss": 4.041, "step": 18730 }, { "epoch": 0.18731, "grad_norm": 1.7322312946905274, "learning_rate": 0.003, "loss": 4.0572, "step": 18731 }, { "epoch": 0.18732, "grad_norm": 0.956403957669077, "learning_rate": 0.003, "loss": 4.0593, "step": 18732 }, { "epoch": 0.18733, "grad_norm": 1.341215213111897, "learning_rate": 0.003, "loss": 4.0494, "step": 18733 }, { "epoch": 0.18734, "grad_norm": 1.3155922164273968, "learning_rate": 0.003, "loss": 4.0588, "step": 18734 }, { "epoch": 0.18735, "grad_norm": 1.0627323976815257, "learning_rate": 0.003, "loss": 4.0608, "step": 18735 }, { "epoch": 0.18736, "grad_norm": 1.5730089758668246, "learning_rate": 0.003, "loss": 4.0717, "step": 18736 }, { "epoch": 0.18737, "grad_norm": 0.9824040045509904, "learning_rate": 0.003, "loss": 4.0764, "step": 18737 }, { "epoch": 0.18738, "grad_norm": 1.6083545327331183, "learning_rate": 0.003, "loss": 4.0438, "step": 18738 }, { "epoch": 0.18739, "grad_norm": 1.1029472411719976, "learning_rate": 0.003, "loss": 4.0784, "step": 18739 }, { "epoch": 0.1874, "grad_norm": 1.3746299168459029, "learning_rate": 0.003, "loss": 4.0243, "step": 18740 }, { "epoch": 0.18741, "grad_norm": 1.195778597348714, "learning_rate": 0.003, "loss": 4.056, "step": 18741 }, { "epoch": 0.18742, "grad_norm": 1.2358515578569471, "learning_rate": 0.003, "loss": 4.0466, "step": 18742 }, { "epoch": 0.18743, "grad_norm": 1.3240559875389892, "learning_rate": 0.003, "loss": 4.0629, "step": 18743 }, { "epoch": 0.18744, "grad_norm": 1.1268808956342007, "learning_rate": 0.003, "loss": 4.0395, "step": 18744 }, { "epoch": 0.18745, "grad_norm": 1.4062233643366688, "learning_rate": 0.003, "loss": 4.0442, "step": 18745 }, { "epoch": 0.18746, "grad_norm": 1.021994536965376, "learning_rate": 0.003, "loss": 4.0306, "step": 18746 }, { "epoch": 0.18747, "grad_norm": 1.281103835925785, "learning_rate": 0.003, "loss": 4.0811, "step": 18747 }, { "epoch": 0.18748, "grad_norm": 1.1180020105028619, "learning_rate": 0.003, "loss": 4.0767, "step": 18748 }, { "epoch": 0.18749, "grad_norm": 1.4942830191803198, "learning_rate": 0.003, "loss": 4.0601, "step": 18749 }, { "epoch": 0.1875, "grad_norm": 1.1061109217475096, "learning_rate": 0.003, "loss": 4.0795, "step": 18750 }, { "epoch": 0.18751, "grad_norm": 1.641043630101888, "learning_rate": 0.003, "loss": 4.0624, "step": 18751 }, { "epoch": 0.18752, "grad_norm": 1.1827677693278353, "learning_rate": 0.003, "loss": 4.0583, "step": 18752 }, { "epoch": 0.18753, "grad_norm": 1.6576891155457214, "learning_rate": 0.003, "loss": 4.0882, "step": 18753 }, { "epoch": 0.18754, "grad_norm": 0.9166261845829198, "learning_rate": 0.003, "loss": 4.0777, "step": 18754 }, { "epoch": 0.18755, "grad_norm": 1.1448762486717943, "learning_rate": 0.003, "loss": 4.0359, "step": 18755 }, { "epoch": 0.18756, "grad_norm": 1.4223527798625488, "learning_rate": 0.003, "loss": 4.073, "step": 18756 }, { "epoch": 0.18757, "grad_norm": 1.392805424459787, "learning_rate": 0.003, "loss": 4.0619, "step": 18757 }, { "epoch": 0.18758, "grad_norm": 1.0679035726418673, "learning_rate": 0.003, "loss": 4.0502, "step": 18758 }, { "epoch": 0.18759, "grad_norm": 1.3142145374211716, "learning_rate": 0.003, "loss": 4.0905, "step": 18759 }, { "epoch": 0.1876, "grad_norm": 1.1928279399139405, "learning_rate": 0.003, "loss": 4.0387, "step": 18760 }, { "epoch": 0.18761, "grad_norm": 1.1299659658986525, "learning_rate": 0.003, "loss": 4.0542, "step": 18761 }, { "epoch": 0.18762, "grad_norm": 1.1132795872327033, "learning_rate": 0.003, "loss": 4.0749, "step": 18762 }, { "epoch": 0.18763, "grad_norm": 1.5519728527597751, "learning_rate": 0.003, "loss": 4.0622, "step": 18763 }, { "epoch": 0.18764, "grad_norm": 1.068029752545123, "learning_rate": 0.003, "loss": 4.0786, "step": 18764 }, { "epoch": 0.18765, "grad_norm": 1.4216790590547768, "learning_rate": 0.003, "loss": 4.04, "step": 18765 }, { "epoch": 0.18766, "grad_norm": 1.2092731561974441, "learning_rate": 0.003, "loss": 4.0625, "step": 18766 }, { "epoch": 0.18767, "grad_norm": 1.1278818613591106, "learning_rate": 0.003, "loss": 4.0679, "step": 18767 }, { "epoch": 0.18768, "grad_norm": 1.2642280565529451, "learning_rate": 0.003, "loss": 4.0514, "step": 18768 }, { "epoch": 0.18769, "grad_norm": 1.0210999890823318, "learning_rate": 0.003, "loss": 4.0607, "step": 18769 }, { "epoch": 0.1877, "grad_norm": 1.559365397727483, "learning_rate": 0.003, "loss": 4.0893, "step": 18770 }, { "epoch": 0.18771, "grad_norm": 0.9757599576723669, "learning_rate": 0.003, "loss": 4.0747, "step": 18771 }, { "epoch": 0.18772, "grad_norm": 1.57422857102648, "learning_rate": 0.003, "loss": 4.0877, "step": 18772 }, { "epoch": 0.18773, "grad_norm": 1.004663288361435, "learning_rate": 0.003, "loss": 4.0639, "step": 18773 }, { "epoch": 0.18774, "grad_norm": 1.6048384508059539, "learning_rate": 0.003, "loss": 4.0512, "step": 18774 }, { "epoch": 0.18775, "grad_norm": 1.4036700003248082, "learning_rate": 0.003, "loss": 4.0744, "step": 18775 }, { "epoch": 0.18776, "grad_norm": 1.1564717371974478, "learning_rate": 0.003, "loss": 4.0743, "step": 18776 }, { "epoch": 0.18777, "grad_norm": 1.423201262806476, "learning_rate": 0.003, "loss": 4.0606, "step": 18777 }, { "epoch": 0.18778, "grad_norm": 1.2151965986827458, "learning_rate": 0.003, "loss": 4.0831, "step": 18778 }, { "epoch": 0.18779, "grad_norm": 1.2812432269648473, "learning_rate": 0.003, "loss": 4.0362, "step": 18779 }, { "epoch": 0.1878, "grad_norm": 1.247415236644388, "learning_rate": 0.003, "loss": 4.0425, "step": 18780 }, { "epoch": 0.18781, "grad_norm": 1.3298421943655812, "learning_rate": 0.003, "loss": 4.0509, "step": 18781 }, { "epoch": 0.18782, "grad_norm": 1.291192897252953, "learning_rate": 0.003, "loss": 4.018, "step": 18782 }, { "epoch": 0.18783, "grad_norm": 1.1717124515165933, "learning_rate": 0.003, "loss": 4.0514, "step": 18783 }, { "epoch": 0.18784, "grad_norm": 1.169639893608424, "learning_rate": 0.003, "loss": 4.0552, "step": 18784 }, { "epoch": 0.18785, "grad_norm": 1.0843001053025956, "learning_rate": 0.003, "loss": 4.0725, "step": 18785 }, { "epoch": 0.18786, "grad_norm": 1.2616116562972313, "learning_rate": 0.003, "loss": 4.0542, "step": 18786 }, { "epoch": 0.18787, "grad_norm": 1.1914570782081966, "learning_rate": 0.003, "loss": 4.0493, "step": 18787 }, { "epoch": 0.18788, "grad_norm": 1.457602004564518, "learning_rate": 0.003, "loss": 4.0591, "step": 18788 }, { "epoch": 0.18789, "grad_norm": 1.0264531082510175, "learning_rate": 0.003, "loss": 4.0586, "step": 18789 }, { "epoch": 0.1879, "grad_norm": 1.3495833279426392, "learning_rate": 0.003, "loss": 4.0573, "step": 18790 }, { "epoch": 0.18791, "grad_norm": 1.22976012385877, "learning_rate": 0.003, "loss": 4.0308, "step": 18791 }, { "epoch": 0.18792, "grad_norm": 1.1837343742085296, "learning_rate": 0.003, "loss": 4.0506, "step": 18792 }, { "epoch": 0.18793, "grad_norm": 1.2220676815952656, "learning_rate": 0.003, "loss": 4.0768, "step": 18793 }, { "epoch": 0.18794, "grad_norm": 1.3643574760850015, "learning_rate": 0.003, "loss": 4.0522, "step": 18794 }, { "epoch": 0.18795, "grad_norm": 1.1508421010536982, "learning_rate": 0.003, "loss": 4.0645, "step": 18795 }, { "epoch": 0.18796, "grad_norm": 1.297945274980459, "learning_rate": 0.003, "loss": 4.0717, "step": 18796 }, { "epoch": 0.18797, "grad_norm": 1.134148435951273, "learning_rate": 0.003, "loss": 4.0529, "step": 18797 }, { "epoch": 0.18798, "grad_norm": 1.2610109456665957, "learning_rate": 0.003, "loss": 4.0564, "step": 18798 }, { "epoch": 0.18799, "grad_norm": 1.2548997534372939, "learning_rate": 0.003, "loss": 4.051, "step": 18799 }, { "epoch": 0.188, "grad_norm": 1.384054339922946, "learning_rate": 0.003, "loss": 4.0536, "step": 18800 }, { "epoch": 0.18801, "grad_norm": 1.0618489518064131, "learning_rate": 0.003, "loss": 4.0358, "step": 18801 }, { "epoch": 0.18802, "grad_norm": 1.2730200819602449, "learning_rate": 0.003, "loss": 4.0413, "step": 18802 }, { "epoch": 0.18803, "grad_norm": 1.2770460140006974, "learning_rate": 0.003, "loss": 4.05, "step": 18803 }, { "epoch": 0.18804, "grad_norm": 0.9867148284171002, "learning_rate": 0.003, "loss": 4.0409, "step": 18804 }, { "epoch": 0.18805, "grad_norm": 1.4204970430346897, "learning_rate": 0.003, "loss": 4.0865, "step": 18805 }, { "epoch": 0.18806, "grad_norm": 1.0189924063256883, "learning_rate": 0.003, "loss": 4.0603, "step": 18806 }, { "epoch": 0.18807, "grad_norm": 1.2809889790216966, "learning_rate": 0.003, "loss": 4.032, "step": 18807 }, { "epoch": 0.18808, "grad_norm": 1.0096811490728688, "learning_rate": 0.003, "loss": 4.039, "step": 18808 }, { "epoch": 0.18809, "grad_norm": 1.39817487077181, "learning_rate": 0.003, "loss": 4.0529, "step": 18809 }, { "epoch": 0.1881, "grad_norm": 1.414326729617747, "learning_rate": 0.003, "loss": 4.0403, "step": 18810 }, { "epoch": 0.18811, "grad_norm": 1.5941626255541983, "learning_rate": 0.003, "loss": 4.0714, "step": 18811 }, { "epoch": 0.18812, "grad_norm": 1.0536188164043787, "learning_rate": 0.003, "loss": 4.0265, "step": 18812 }, { "epoch": 0.18813, "grad_norm": 1.3324731310272937, "learning_rate": 0.003, "loss": 4.041, "step": 18813 }, { "epoch": 0.18814, "grad_norm": 1.1925920227492695, "learning_rate": 0.003, "loss": 4.0476, "step": 18814 }, { "epoch": 0.18815, "grad_norm": 1.3120883229036413, "learning_rate": 0.003, "loss": 4.0652, "step": 18815 }, { "epoch": 0.18816, "grad_norm": 1.0930340123629445, "learning_rate": 0.003, "loss": 4.0281, "step": 18816 }, { "epoch": 0.18817, "grad_norm": 1.219252902262019, "learning_rate": 0.003, "loss": 4.0955, "step": 18817 }, { "epoch": 0.18818, "grad_norm": 1.5524946483590683, "learning_rate": 0.003, "loss": 4.0723, "step": 18818 }, { "epoch": 0.18819, "grad_norm": 0.9897498018958503, "learning_rate": 0.003, "loss": 4.0641, "step": 18819 }, { "epoch": 0.1882, "grad_norm": 1.5411202902175514, "learning_rate": 0.003, "loss": 4.0604, "step": 18820 }, { "epoch": 0.18821, "grad_norm": 0.9875784929877088, "learning_rate": 0.003, "loss": 4.0782, "step": 18821 }, { "epoch": 0.18822, "grad_norm": 1.201853807644429, "learning_rate": 0.003, "loss": 4.0696, "step": 18822 }, { "epoch": 0.18823, "grad_norm": 1.419840414175821, "learning_rate": 0.003, "loss": 4.0626, "step": 18823 }, { "epoch": 0.18824, "grad_norm": 1.4980577234133672, "learning_rate": 0.003, "loss": 4.0391, "step": 18824 }, { "epoch": 0.18825, "grad_norm": 1.3236118047628036, "learning_rate": 0.003, "loss": 4.0524, "step": 18825 }, { "epoch": 0.18826, "grad_norm": 1.2847401073788443, "learning_rate": 0.003, "loss": 4.0591, "step": 18826 }, { "epoch": 0.18827, "grad_norm": 1.3538219367720525, "learning_rate": 0.003, "loss": 4.071, "step": 18827 }, { "epoch": 0.18828, "grad_norm": 1.1769730216139096, "learning_rate": 0.003, "loss": 4.0493, "step": 18828 }, { "epoch": 0.18829, "grad_norm": 0.9967557172414647, "learning_rate": 0.003, "loss": 4.0588, "step": 18829 }, { "epoch": 0.1883, "grad_norm": 1.5240229295468803, "learning_rate": 0.003, "loss": 4.0625, "step": 18830 }, { "epoch": 0.18831, "grad_norm": 0.9906777777489462, "learning_rate": 0.003, "loss": 4.0318, "step": 18831 }, { "epoch": 0.18832, "grad_norm": 1.5082207878120748, "learning_rate": 0.003, "loss": 4.0823, "step": 18832 }, { "epoch": 0.18833, "grad_norm": 0.9477446461010499, "learning_rate": 0.003, "loss": 4.0329, "step": 18833 }, { "epoch": 0.18834, "grad_norm": 1.3717536216498554, "learning_rate": 0.003, "loss": 4.0827, "step": 18834 }, { "epoch": 0.18835, "grad_norm": 1.0891391557208812, "learning_rate": 0.003, "loss": 4.0478, "step": 18835 }, { "epoch": 0.18836, "grad_norm": 1.5045117096900225, "learning_rate": 0.003, "loss": 4.0629, "step": 18836 }, { "epoch": 0.18837, "grad_norm": 1.4637726562877373, "learning_rate": 0.003, "loss": 4.0714, "step": 18837 }, { "epoch": 0.18838, "grad_norm": 1.2250820711558688, "learning_rate": 0.003, "loss": 4.0668, "step": 18838 }, { "epoch": 0.18839, "grad_norm": 1.1610335799901077, "learning_rate": 0.003, "loss": 4.0513, "step": 18839 }, { "epoch": 0.1884, "grad_norm": 1.3300297533103165, "learning_rate": 0.003, "loss": 4.053, "step": 18840 }, { "epoch": 0.18841, "grad_norm": 1.3617342778490928, "learning_rate": 0.003, "loss": 4.0571, "step": 18841 }, { "epoch": 0.18842, "grad_norm": 1.3820649819828323, "learning_rate": 0.003, "loss": 4.0642, "step": 18842 }, { "epoch": 0.18843, "grad_norm": 1.16220397310607, "learning_rate": 0.003, "loss": 4.0635, "step": 18843 }, { "epoch": 0.18844, "grad_norm": 1.1410995537572468, "learning_rate": 0.003, "loss": 4.079, "step": 18844 }, { "epoch": 0.18845, "grad_norm": 1.363632859651913, "learning_rate": 0.003, "loss": 4.0517, "step": 18845 }, { "epoch": 0.18846, "grad_norm": 1.4013719562590952, "learning_rate": 0.003, "loss": 4.0469, "step": 18846 }, { "epoch": 0.18847, "grad_norm": 1.0040823809056576, "learning_rate": 0.003, "loss": 4.0489, "step": 18847 }, { "epoch": 0.18848, "grad_norm": 1.7138847323654889, "learning_rate": 0.003, "loss": 4.0438, "step": 18848 }, { "epoch": 0.18849, "grad_norm": 1.1974881816633904, "learning_rate": 0.003, "loss": 4.0662, "step": 18849 }, { "epoch": 0.1885, "grad_norm": 1.4071990004696164, "learning_rate": 0.003, "loss": 4.0567, "step": 18850 }, { "epoch": 0.18851, "grad_norm": 1.2945320517992853, "learning_rate": 0.003, "loss": 4.0583, "step": 18851 }, { "epoch": 0.18852, "grad_norm": 1.0493911494113357, "learning_rate": 0.003, "loss": 4.082, "step": 18852 }, { "epoch": 0.18853, "grad_norm": 1.3522301267229937, "learning_rate": 0.003, "loss": 4.0653, "step": 18853 }, { "epoch": 0.18854, "grad_norm": 0.9479585527222019, "learning_rate": 0.003, "loss": 4.0673, "step": 18854 }, { "epoch": 0.18855, "grad_norm": 1.3052992350117785, "learning_rate": 0.003, "loss": 4.068, "step": 18855 }, { "epoch": 0.18856, "grad_norm": 1.1088881925585756, "learning_rate": 0.003, "loss": 4.0267, "step": 18856 }, { "epoch": 0.18857, "grad_norm": 1.1951391458949168, "learning_rate": 0.003, "loss": 4.0544, "step": 18857 }, { "epoch": 0.18858, "grad_norm": 1.2872035984060555, "learning_rate": 0.003, "loss": 4.0509, "step": 18858 }, { "epoch": 0.18859, "grad_norm": 1.0738444028740626, "learning_rate": 0.003, "loss": 4.0479, "step": 18859 }, { "epoch": 0.1886, "grad_norm": 1.423952097069054, "learning_rate": 0.003, "loss": 4.082, "step": 18860 }, { "epoch": 0.18861, "grad_norm": 1.1861884938319123, "learning_rate": 0.003, "loss": 4.074, "step": 18861 }, { "epoch": 0.18862, "grad_norm": 1.268656998919176, "learning_rate": 0.003, "loss": 4.0707, "step": 18862 }, { "epoch": 0.18863, "grad_norm": 1.5208838043266313, "learning_rate": 0.003, "loss": 4.0356, "step": 18863 }, { "epoch": 0.18864, "grad_norm": 1.0769050208064912, "learning_rate": 0.003, "loss": 4.0674, "step": 18864 }, { "epoch": 0.18865, "grad_norm": 1.4268680552190511, "learning_rate": 0.003, "loss": 4.0612, "step": 18865 }, { "epoch": 0.18866, "grad_norm": 1.0360871095610265, "learning_rate": 0.003, "loss": 4.0554, "step": 18866 }, { "epoch": 0.18867, "grad_norm": 1.4852075792511028, "learning_rate": 0.003, "loss": 4.0568, "step": 18867 }, { "epoch": 0.18868, "grad_norm": 1.0618590015090037, "learning_rate": 0.003, "loss": 4.0797, "step": 18868 }, { "epoch": 0.18869, "grad_norm": 1.5082378931522273, "learning_rate": 0.003, "loss": 4.0657, "step": 18869 }, { "epoch": 0.1887, "grad_norm": 1.1699590064692302, "learning_rate": 0.003, "loss": 4.0491, "step": 18870 }, { "epoch": 0.18871, "grad_norm": 1.5945378827543812, "learning_rate": 0.003, "loss": 4.0827, "step": 18871 }, { "epoch": 0.18872, "grad_norm": 1.1131528037116456, "learning_rate": 0.003, "loss": 4.0566, "step": 18872 }, { "epoch": 0.18873, "grad_norm": 1.1786422363921647, "learning_rate": 0.003, "loss": 4.0424, "step": 18873 }, { "epoch": 0.18874, "grad_norm": 1.2345214530184712, "learning_rate": 0.003, "loss": 4.0664, "step": 18874 }, { "epoch": 0.18875, "grad_norm": 1.290648570863387, "learning_rate": 0.003, "loss": 4.0934, "step": 18875 }, { "epoch": 0.18876, "grad_norm": 1.2236955589023566, "learning_rate": 0.003, "loss": 4.05, "step": 18876 }, { "epoch": 0.18877, "grad_norm": 1.1482019620159905, "learning_rate": 0.003, "loss": 4.0588, "step": 18877 }, { "epoch": 0.18878, "grad_norm": 1.312168673740298, "learning_rate": 0.003, "loss": 4.0333, "step": 18878 }, { "epoch": 0.18879, "grad_norm": 1.3563338453714813, "learning_rate": 0.003, "loss": 4.0582, "step": 18879 }, { "epoch": 0.1888, "grad_norm": 1.0704749748380464, "learning_rate": 0.003, "loss": 4.0553, "step": 18880 }, { "epoch": 0.18881, "grad_norm": 1.2655347387144644, "learning_rate": 0.003, "loss": 4.0368, "step": 18881 }, { "epoch": 0.18882, "grad_norm": 1.2141701531674876, "learning_rate": 0.003, "loss": 4.081, "step": 18882 }, { "epoch": 0.18883, "grad_norm": 1.2624455567851907, "learning_rate": 0.003, "loss": 4.0398, "step": 18883 }, { "epoch": 0.18884, "grad_norm": 1.406371652440986, "learning_rate": 0.003, "loss": 4.064, "step": 18884 }, { "epoch": 0.18885, "grad_norm": 0.9983916971461474, "learning_rate": 0.003, "loss": 4.0247, "step": 18885 }, { "epoch": 0.18886, "grad_norm": 1.4333871584870326, "learning_rate": 0.003, "loss": 4.0581, "step": 18886 }, { "epoch": 0.18887, "grad_norm": 1.1255837722806947, "learning_rate": 0.003, "loss": 4.0617, "step": 18887 }, { "epoch": 0.18888, "grad_norm": 1.409572666244752, "learning_rate": 0.003, "loss": 4.0752, "step": 18888 }, { "epoch": 0.18889, "grad_norm": 1.1808092271215411, "learning_rate": 0.003, "loss": 4.0343, "step": 18889 }, { "epoch": 0.1889, "grad_norm": 1.294565705597222, "learning_rate": 0.003, "loss": 4.0665, "step": 18890 }, { "epoch": 0.18891, "grad_norm": 1.2245215397964557, "learning_rate": 0.003, "loss": 4.0586, "step": 18891 }, { "epoch": 0.18892, "grad_norm": 1.057239401631099, "learning_rate": 0.003, "loss": 4.0533, "step": 18892 }, { "epoch": 0.18893, "grad_norm": 1.479448868735589, "learning_rate": 0.003, "loss": 4.0358, "step": 18893 }, { "epoch": 0.18894, "grad_norm": 1.0000969539239213, "learning_rate": 0.003, "loss": 4.0448, "step": 18894 }, { "epoch": 0.18895, "grad_norm": 1.5510575764076522, "learning_rate": 0.003, "loss": 4.0415, "step": 18895 }, { "epoch": 0.18896, "grad_norm": 1.013820609190286, "learning_rate": 0.003, "loss": 4.0372, "step": 18896 }, { "epoch": 0.18897, "grad_norm": 1.339914850079941, "learning_rate": 0.003, "loss": 4.0368, "step": 18897 }, { "epoch": 0.18898, "grad_norm": 1.0905513146887738, "learning_rate": 0.003, "loss": 4.0738, "step": 18898 }, { "epoch": 0.18899, "grad_norm": 1.4980920304789715, "learning_rate": 0.003, "loss": 4.0519, "step": 18899 }, { "epoch": 0.189, "grad_norm": 1.220822162538095, "learning_rate": 0.003, "loss": 4.0329, "step": 18900 }, { "epoch": 0.18901, "grad_norm": 1.2984675363326095, "learning_rate": 0.003, "loss": 4.0519, "step": 18901 }, { "epoch": 0.18902, "grad_norm": 1.1393971083261454, "learning_rate": 0.003, "loss": 4.052, "step": 18902 }, { "epoch": 0.18903, "grad_norm": 1.2798411582421874, "learning_rate": 0.003, "loss": 4.0653, "step": 18903 }, { "epoch": 0.18904, "grad_norm": 1.29715988715658, "learning_rate": 0.003, "loss": 4.0791, "step": 18904 }, { "epoch": 0.18905, "grad_norm": 1.1510916069001649, "learning_rate": 0.003, "loss": 4.0582, "step": 18905 }, { "epoch": 0.18906, "grad_norm": 1.2876896440316996, "learning_rate": 0.003, "loss": 4.0599, "step": 18906 }, { "epoch": 0.18907, "grad_norm": 1.0620968999573859, "learning_rate": 0.003, "loss": 4.0386, "step": 18907 }, { "epoch": 0.18908, "grad_norm": 1.2586156432936058, "learning_rate": 0.003, "loss": 4.0386, "step": 18908 }, { "epoch": 0.18909, "grad_norm": 1.1054595703171823, "learning_rate": 0.003, "loss": 4.0619, "step": 18909 }, { "epoch": 0.1891, "grad_norm": 1.6090156306881913, "learning_rate": 0.003, "loss": 4.0797, "step": 18910 }, { "epoch": 0.18911, "grad_norm": 1.0370237279217571, "learning_rate": 0.003, "loss": 4.0824, "step": 18911 }, { "epoch": 0.18912, "grad_norm": 1.3108642891143027, "learning_rate": 0.003, "loss": 4.0681, "step": 18912 }, { "epoch": 0.18913, "grad_norm": 1.166700841813181, "learning_rate": 0.003, "loss": 4.0521, "step": 18913 }, { "epoch": 0.18914, "grad_norm": 1.3118894510467247, "learning_rate": 0.003, "loss": 4.071, "step": 18914 }, { "epoch": 0.18915, "grad_norm": 1.3672575167176313, "learning_rate": 0.003, "loss": 4.0418, "step": 18915 }, { "epoch": 0.18916, "grad_norm": 1.1851340868517821, "learning_rate": 0.003, "loss": 4.0624, "step": 18916 }, { "epoch": 0.18917, "grad_norm": 1.394840501538389, "learning_rate": 0.003, "loss": 4.065, "step": 18917 }, { "epoch": 0.18918, "grad_norm": 1.172796389585038, "learning_rate": 0.003, "loss": 4.038, "step": 18918 }, { "epoch": 0.18919, "grad_norm": 1.4820948479111926, "learning_rate": 0.003, "loss": 4.0859, "step": 18919 }, { "epoch": 0.1892, "grad_norm": 1.1087980018754855, "learning_rate": 0.003, "loss": 4.0549, "step": 18920 }, { "epoch": 0.18921, "grad_norm": 1.6150820672105983, "learning_rate": 0.003, "loss": 4.0468, "step": 18921 }, { "epoch": 0.18922, "grad_norm": 1.1334047373684482, "learning_rate": 0.003, "loss": 4.0511, "step": 18922 }, { "epoch": 0.18923, "grad_norm": 1.2049292682676265, "learning_rate": 0.003, "loss": 4.0417, "step": 18923 }, { "epoch": 0.18924, "grad_norm": 1.3865621325880826, "learning_rate": 0.003, "loss": 4.0585, "step": 18924 }, { "epoch": 0.18925, "grad_norm": 1.1170851788113854, "learning_rate": 0.003, "loss": 4.0369, "step": 18925 }, { "epoch": 0.18926, "grad_norm": 1.6191176988281037, "learning_rate": 0.003, "loss": 4.0943, "step": 18926 }, { "epoch": 0.18927, "grad_norm": 0.9493332963228506, "learning_rate": 0.003, "loss": 4.0594, "step": 18927 }, { "epoch": 0.18928, "grad_norm": 1.144685556081795, "learning_rate": 0.003, "loss": 4.0228, "step": 18928 }, { "epoch": 0.18929, "grad_norm": 1.3412926968213719, "learning_rate": 0.003, "loss": 4.0768, "step": 18929 }, { "epoch": 0.1893, "grad_norm": 1.2014837923534043, "learning_rate": 0.003, "loss": 4.0353, "step": 18930 }, { "epoch": 0.18931, "grad_norm": 1.247210746382838, "learning_rate": 0.003, "loss": 4.0569, "step": 18931 }, { "epoch": 0.18932, "grad_norm": 1.0874898122123102, "learning_rate": 0.003, "loss": 4.0923, "step": 18932 }, { "epoch": 0.18933, "grad_norm": 1.1680747946964996, "learning_rate": 0.003, "loss": 4.0326, "step": 18933 }, { "epoch": 0.18934, "grad_norm": 1.3061225339713496, "learning_rate": 0.003, "loss": 4.0602, "step": 18934 }, { "epoch": 0.18935, "grad_norm": 1.0774394467048682, "learning_rate": 0.003, "loss": 4.0323, "step": 18935 }, { "epoch": 0.18936, "grad_norm": 1.2662585394487187, "learning_rate": 0.003, "loss": 4.0745, "step": 18936 }, { "epoch": 0.18937, "grad_norm": 1.1485257564065396, "learning_rate": 0.003, "loss": 4.0287, "step": 18937 }, { "epoch": 0.18938, "grad_norm": 1.3073638993766212, "learning_rate": 0.003, "loss": 4.0627, "step": 18938 }, { "epoch": 0.18939, "grad_norm": 1.1820949992693355, "learning_rate": 0.003, "loss": 4.0535, "step": 18939 }, { "epoch": 0.1894, "grad_norm": 1.2770017063627936, "learning_rate": 0.003, "loss": 4.0798, "step": 18940 }, { "epoch": 0.18941, "grad_norm": 1.3485118903633273, "learning_rate": 0.003, "loss": 4.0456, "step": 18941 }, { "epoch": 0.18942, "grad_norm": 1.299939845997475, "learning_rate": 0.003, "loss": 4.0667, "step": 18942 }, { "epoch": 0.18943, "grad_norm": 1.101719247255625, "learning_rate": 0.003, "loss": 4.0412, "step": 18943 }, { "epoch": 0.18944, "grad_norm": 1.4394683795789305, "learning_rate": 0.003, "loss": 4.0513, "step": 18944 }, { "epoch": 0.18945, "grad_norm": 0.9588952070702902, "learning_rate": 0.003, "loss": 4.0188, "step": 18945 }, { "epoch": 0.18946, "grad_norm": 1.2430544770767809, "learning_rate": 0.003, "loss": 4.0341, "step": 18946 }, { "epoch": 0.18947, "grad_norm": 1.3361267616006969, "learning_rate": 0.003, "loss": 4.0395, "step": 18947 }, { "epoch": 0.18948, "grad_norm": 1.3336918407281249, "learning_rate": 0.003, "loss": 4.0654, "step": 18948 }, { "epoch": 0.18949, "grad_norm": 1.1394460864552343, "learning_rate": 0.003, "loss": 4.0553, "step": 18949 }, { "epoch": 0.1895, "grad_norm": 1.324063928905688, "learning_rate": 0.003, "loss": 4.0794, "step": 18950 }, { "epoch": 0.18951, "grad_norm": 1.3409462464835011, "learning_rate": 0.003, "loss": 4.0506, "step": 18951 }, { "epoch": 0.18952, "grad_norm": 1.2524298601876103, "learning_rate": 0.003, "loss": 4.0302, "step": 18952 }, { "epoch": 0.18953, "grad_norm": 1.162892549071531, "learning_rate": 0.003, "loss": 4.0486, "step": 18953 }, { "epoch": 0.18954, "grad_norm": 1.4673924290785485, "learning_rate": 0.003, "loss": 4.0561, "step": 18954 }, { "epoch": 0.18955, "grad_norm": 1.1855174825877477, "learning_rate": 0.003, "loss": 4.0264, "step": 18955 }, { "epoch": 0.18956, "grad_norm": 1.3151638032539035, "learning_rate": 0.003, "loss": 4.0582, "step": 18956 }, { "epoch": 0.18957, "grad_norm": 1.0229964035825971, "learning_rate": 0.003, "loss": 4.024, "step": 18957 }, { "epoch": 0.18958, "grad_norm": 1.305533065097813, "learning_rate": 0.003, "loss": 4.0325, "step": 18958 }, { "epoch": 0.18959, "grad_norm": 1.203761119232079, "learning_rate": 0.003, "loss": 4.0202, "step": 18959 }, { "epoch": 0.1896, "grad_norm": 1.2164254639553058, "learning_rate": 0.003, "loss": 4.0482, "step": 18960 }, { "epoch": 0.18961, "grad_norm": 1.2777264410334181, "learning_rate": 0.003, "loss": 4.0587, "step": 18961 }, { "epoch": 0.18962, "grad_norm": 1.399564003481491, "learning_rate": 0.003, "loss": 4.091, "step": 18962 }, { "epoch": 0.18963, "grad_norm": 1.2219866179380663, "learning_rate": 0.003, "loss": 4.0666, "step": 18963 }, { "epoch": 0.18964, "grad_norm": 1.2931070300778063, "learning_rate": 0.003, "loss": 4.0741, "step": 18964 }, { "epoch": 0.18965, "grad_norm": 1.2408788270459041, "learning_rate": 0.003, "loss": 4.0656, "step": 18965 }, { "epoch": 0.18966, "grad_norm": 1.3852451172811868, "learning_rate": 0.003, "loss": 4.0405, "step": 18966 }, { "epoch": 0.18967, "grad_norm": 1.181955340350945, "learning_rate": 0.003, "loss": 4.0467, "step": 18967 }, { "epoch": 0.18968, "grad_norm": 1.388009976154795, "learning_rate": 0.003, "loss": 4.0766, "step": 18968 }, { "epoch": 0.18969, "grad_norm": 1.452508863615844, "learning_rate": 0.003, "loss": 4.068, "step": 18969 }, { "epoch": 0.1897, "grad_norm": 1.0539613115032418, "learning_rate": 0.003, "loss": 4.0257, "step": 18970 }, { "epoch": 0.18971, "grad_norm": 1.5283629289875311, "learning_rate": 0.003, "loss": 4.0357, "step": 18971 }, { "epoch": 0.18972, "grad_norm": 1.0020586289626954, "learning_rate": 0.003, "loss": 4.0563, "step": 18972 }, { "epoch": 0.18973, "grad_norm": 1.4643755552488762, "learning_rate": 0.003, "loss": 4.0283, "step": 18973 }, { "epoch": 0.18974, "grad_norm": 1.082547674257327, "learning_rate": 0.003, "loss": 4.0714, "step": 18974 }, { "epoch": 0.18975, "grad_norm": 1.4776300371215036, "learning_rate": 0.003, "loss": 4.048, "step": 18975 }, { "epoch": 0.18976, "grad_norm": 1.3135592405403211, "learning_rate": 0.003, "loss": 4.0856, "step": 18976 }, { "epoch": 0.18977, "grad_norm": 1.0745658921355028, "learning_rate": 0.003, "loss": 4.0774, "step": 18977 }, { "epoch": 0.18978, "grad_norm": 1.2457878005667065, "learning_rate": 0.003, "loss": 4.0531, "step": 18978 }, { "epoch": 0.18979, "grad_norm": 1.0044668622723179, "learning_rate": 0.003, "loss": 4.0455, "step": 18979 }, { "epoch": 0.1898, "grad_norm": 1.4943251717141453, "learning_rate": 0.003, "loss": 4.0406, "step": 18980 }, { "epoch": 0.18981, "grad_norm": 1.1274084795099037, "learning_rate": 0.003, "loss": 4.0579, "step": 18981 }, { "epoch": 0.18982, "grad_norm": 1.306218374717761, "learning_rate": 0.003, "loss": 4.0471, "step": 18982 }, { "epoch": 0.18983, "grad_norm": 1.195507438217964, "learning_rate": 0.003, "loss": 4.0663, "step": 18983 }, { "epoch": 0.18984, "grad_norm": 1.3489812646578905, "learning_rate": 0.003, "loss": 4.0591, "step": 18984 }, { "epoch": 0.18985, "grad_norm": 1.079978511793761, "learning_rate": 0.003, "loss": 4.048, "step": 18985 }, { "epoch": 0.18986, "grad_norm": 1.8198031846834766, "learning_rate": 0.003, "loss": 4.0243, "step": 18986 }, { "epoch": 0.18987, "grad_norm": 0.9543303051945693, "learning_rate": 0.003, "loss": 4.0256, "step": 18987 }, { "epoch": 0.18988, "grad_norm": 1.2918817749052591, "learning_rate": 0.003, "loss": 4.0604, "step": 18988 }, { "epoch": 0.18989, "grad_norm": 1.2298887897140032, "learning_rate": 0.003, "loss": 4.0561, "step": 18989 }, { "epoch": 0.1899, "grad_norm": 1.4214273857280832, "learning_rate": 0.003, "loss": 4.0384, "step": 18990 }, { "epoch": 0.18991, "grad_norm": 0.9867915594580235, "learning_rate": 0.003, "loss": 4.0525, "step": 18991 }, { "epoch": 0.18992, "grad_norm": 1.1982962929557852, "learning_rate": 0.003, "loss": 4.0412, "step": 18992 }, { "epoch": 0.18993, "grad_norm": 1.2484423183571205, "learning_rate": 0.003, "loss": 4.0752, "step": 18993 }, { "epoch": 0.18994, "grad_norm": 1.1323142694208788, "learning_rate": 0.003, "loss": 4.0554, "step": 18994 }, { "epoch": 0.18995, "grad_norm": 1.1463045485943708, "learning_rate": 0.003, "loss": 4.0, "step": 18995 }, { "epoch": 0.18996, "grad_norm": 1.0377533028547177, "learning_rate": 0.003, "loss": 4.0518, "step": 18996 }, { "epoch": 0.18997, "grad_norm": 1.3611288030082336, "learning_rate": 0.003, "loss": 4.0679, "step": 18997 }, { "epoch": 0.18998, "grad_norm": 1.1713704658513688, "learning_rate": 0.003, "loss": 4.0632, "step": 18998 }, { "epoch": 0.18999, "grad_norm": 1.4184447797254487, "learning_rate": 0.003, "loss": 4.0385, "step": 18999 }, { "epoch": 0.19, "grad_norm": 0.9938167299526427, "learning_rate": 0.003, "loss": 4.0717, "step": 19000 }, { "epoch": 0.19001, "grad_norm": 1.3314032409593017, "learning_rate": 0.003, "loss": 4.0656, "step": 19001 }, { "epoch": 0.19002, "grad_norm": 1.1479969333151756, "learning_rate": 0.003, "loss": 4.0366, "step": 19002 }, { "epoch": 0.19003, "grad_norm": 1.2375409505772659, "learning_rate": 0.003, "loss": 4.0555, "step": 19003 }, { "epoch": 0.19004, "grad_norm": 1.3176797601185788, "learning_rate": 0.003, "loss": 4.0585, "step": 19004 }, { "epoch": 0.19005, "grad_norm": 1.6342456579206885, "learning_rate": 0.003, "loss": 4.0513, "step": 19005 }, { "epoch": 0.19006, "grad_norm": 1.1753775466652139, "learning_rate": 0.003, "loss": 4.0629, "step": 19006 }, { "epoch": 0.19007, "grad_norm": 1.2277475711315688, "learning_rate": 0.003, "loss": 4.0636, "step": 19007 }, { "epoch": 0.19008, "grad_norm": 1.4782021717418683, "learning_rate": 0.003, "loss": 4.0513, "step": 19008 }, { "epoch": 0.19009, "grad_norm": 1.0861563056624512, "learning_rate": 0.003, "loss": 4.0706, "step": 19009 }, { "epoch": 0.1901, "grad_norm": 1.2797915735224965, "learning_rate": 0.003, "loss": 4.0477, "step": 19010 }, { "epoch": 0.19011, "grad_norm": 1.3737834290748083, "learning_rate": 0.003, "loss": 4.0484, "step": 19011 }, { "epoch": 0.19012, "grad_norm": 1.3596128313334717, "learning_rate": 0.003, "loss": 4.0429, "step": 19012 }, { "epoch": 0.19013, "grad_norm": 1.1798694422318736, "learning_rate": 0.003, "loss": 4.0338, "step": 19013 }, { "epoch": 0.19014, "grad_norm": 1.3600351848591157, "learning_rate": 0.003, "loss": 4.0532, "step": 19014 }, { "epoch": 0.19015, "grad_norm": 1.141414457718081, "learning_rate": 0.003, "loss": 4.0159, "step": 19015 }, { "epoch": 0.19016, "grad_norm": 1.262402004064144, "learning_rate": 0.003, "loss": 4.0434, "step": 19016 }, { "epoch": 0.19017, "grad_norm": 1.2130498633124398, "learning_rate": 0.003, "loss": 4.0765, "step": 19017 }, { "epoch": 0.19018, "grad_norm": 1.271567940825599, "learning_rate": 0.003, "loss": 4.0605, "step": 19018 }, { "epoch": 0.19019, "grad_norm": 1.1325178362829071, "learning_rate": 0.003, "loss": 4.0451, "step": 19019 }, { "epoch": 0.1902, "grad_norm": 1.2074662414846844, "learning_rate": 0.003, "loss": 4.0593, "step": 19020 }, { "epoch": 0.19021, "grad_norm": 1.1450482817426657, "learning_rate": 0.003, "loss": 4.0537, "step": 19021 }, { "epoch": 0.19022, "grad_norm": 1.312247265532254, "learning_rate": 0.003, "loss": 4.0832, "step": 19022 }, { "epoch": 0.19023, "grad_norm": 1.3091949037421626, "learning_rate": 0.003, "loss": 4.1027, "step": 19023 }, { "epoch": 0.19024, "grad_norm": 1.1872720134607715, "learning_rate": 0.003, "loss": 4.0415, "step": 19024 }, { "epoch": 0.19025, "grad_norm": 1.2670890663034708, "learning_rate": 0.003, "loss": 4.0544, "step": 19025 }, { "epoch": 0.19026, "grad_norm": 1.3355338650971906, "learning_rate": 0.003, "loss": 4.062, "step": 19026 }, { "epoch": 0.19027, "grad_norm": 1.1281748348642044, "learning_rate": 0.003, "loss": 4.073, "step": 19027 }, { "epoch": 0.19028, "grad_norm": 1.5137878227413488, "learning_rate": 0.003, "loss": 4.0512, "step": 19028 }, { "epoch": 0.19029, "grad_norm": 1.0493984454266818, "learning_rate": 0.003, "loss": 4.0245, "step": 19029 }, { "epoch": 0.1903, "grad_norm": 1.5058467634901882, "learning_rate": 0.003, "loss": 4.065, "step": 19030 }, { "epoch": 0.19031, "grad_norm": 1.45001970840602, "learning_rate": 0.003, "loss": 4.0531, "step": 19031 }, { "epoch": 0.19032, "grad_norm": 1.4059804607395356, "learning_rate": 0.003, "loss": 4.0558, "step": 19032 }, { "epoch": 0.19033, "grad_norm": 1.4382315050248882, "learning_rate": 0.003, "loss": 4.0341, "step": 19033 }, { "epoch": 0.19034, "grad_norm": 1.2144508213612755, "learning_rate": 0.003, "loss": 4.056, "step": 19034 }, { "epoch": 0.19035, "grad_norm": 1.2834551998532726, "learning_rate": 0.003, "loss": 4.0316, "step": 19035 }, { "epoch": 0.19036, "grad_norm": 1.1145058006863604, "learning_rate": 0.003, "loss": 4.0477, "step": 19036 }, { "epoch": 0.19037, "grad_norm": 1.2924159032683007, "learning_rate": 0.003, "loss": 4.0319, "step": 19037 }, { "epoch": 0.19038, "grad_norm": 1.1242842475684267, "learning_rate": 0.003, "loss": 4.0626, "step": 19038 }, { "epoch": 0.19039, "grad_norm": 1.320438816371859, "learning_rate": 0.003, "loss": 4.0575, "step": 19039 }, { "epoch": 0.1904, "grad_norm": 1.0389084205361234, "learning_rate": 0.003, "loss": 4.0229, "step": 19040 }, { "epoch": 0.19041, "grad_norm": 1.4772991214516211, "learning_rate": 0.003, "loss": 4.0546, "step": 19041 }, { "epoch": 0.19042, "grad_norm": 1.0942789445482135, "learning_rate": 0.003, "loss": 4.0636, "step": 19042 }, { "epoch": 0.19043, "grad_norm": 1.4041406453379557, "learning_rate": 0.003, "loss": 4.0692, "step": 19043 }, { "epoch": 0.19044, "grad_norm": 1.3233649763436737, "learning_rate": 0.003, "loss": 4.0554, "step": 19044 }, { "epoch": 0.19045, "grad_norm": 1.3186020535932237, "learning_rate": 0.003, "loss": 4.0668, "step": 19045 }, { "epoch": 0.19046, "grad_norm": 1.139676960059712, "learning_rate": 0.003, "loss": 4.0364, "step": 19046 }, { "epoch": 0.19047, "grad_norm": 1.30211529961854, "learning_rate": 0.003, "loss": 4.0489, "step": 19047 }, { "epoch": 0.19048, "grad_norm": 1.2421629614746557, "learning_rate": 0.003, "loss": 4.0696, "step": 19048 }, { "epoch": 0.19049, "grad_norm": 1.453451977114375, "learning_rate": 0.003, "loss": 4.0778, "step": 19049 }, { "epoch": 0.1905, "grad_norm": 0.9663541331808644, "learning_rate": 0.003, "loss": 4.0741, "step": 19050 }, { "epoch": 0.19051, "grad_norm": 1.2498652674040842, "learning_rate": 0.003, "loss": 4.0606, "step": 19051 }, { "epoch": 0.19052, "grad_norm": 1.3002205859726348, "learning_rate": 0.003, "loss": 4.0683, "step": 19052 }, { "epoch": 0.19053, "grad_norm": 1.4834338460002425, "learning_rate": 0.003, "loss": 4.0339, "step": 19053 }, { "epoch": 0.19054, "grad_norm": 1.2333198115803878, "learning_rate": 0.003, "loss": 4.049, "step": 19054 }, { "epoch": 0.19055, "grad_norm": 1.271025096267169, "learning_rate": 0.003, "loss": 4.0491, "step": 19055 }, { "epoch": 0.19056, "grad_norm": 1.3713791298987748, "learning_rate": 0.003, "loss": 4.0442, "step": 19056 }, { "epoch": 0.19057, "grad_norm": 1.2974018074502591, "learning_rate": 0.003, "loss": 4.053, "step": 19057 }, { "epoch": 0.19058, "grad_norm": 1.3168209422171067, "learning_rate": 0.003, "loss": 4.0702, "step": 19058 }, { "epoch": 0.19059, "grad_norm": 1.1841553199839385, "learning_rate": 0.003, "loss": 4.0592, "step": 19059 }, { "epoch": 0.1906, "grad_norm": 1.2722570368049224, "learning_rate": 0.003, "loss": 4.0631, "step": 19060 }, { "epoch": 0.19061, "grad_norm": 1.3041753642371723, "learning_rate": 0.003, "loss": 4.0564, "step": 19061 }, { "epoch": 0.19062, "grad_norm": 1.1612477238430865, "learning_rate": 0.003, "loss": 4.0447, "step": 19062 }, { "epoch": 0.19063, "grad_norm": 1.2736861822164462, "learning_rate": 0.003, "loss": 4.0154, "step": 19063 }, { "epoch": 0.19064, "grad_norm": 1.0833764979127176, "learning_rate": 0.003, "loss": 4.0528, "step": 19064 }, { "epoch": 0.19065, "grad_norm": 1.3819328838021085, "learning_rate": 0.003, "loss": 4.0693, "step": 19065 }, { "epoch": 0.19066, "grad_norm": 1.2713583019321415, "learning_rate": 0.003, "loss": 4.0573, "step": 19066 }, { "epoch": 0.19067, "grad_norm": 1.4667943448149092, "learning_rate": 0.003, "loss": 4.0523, "step": 19067 }, { "epoch": 0.19068, "grad_norm": 1.1869598216617847, "learning_rate": 0.003, "loss": 4.0513, "step": 19068 }, { "epoch": 0.19069, "grad_norm": 1.1738634779373909, "learning_rate": 0.003, "loss": 4.0647, "step": 19069 }, { "epoch": 0.1907, "grad_norm": 1.2267084541346251, "learning_rate": 0.003, "loss": 4.0381, "step": 19070 }, { "epoch": 0.19071, "grad_norm": 1.4894837196590038, "learning_rate": 0.003, "loss": 4.064, "step": 19071 }, { "epoch": 0.19072, "grad_norm": 1.452084826976423, "learning_rate": 0.003, "loss": 4.0727, "step": 19072 }, { "epoch": 0.19073, "grad_norm": 1.1464432083818423, "learning_rate": 0.003, "loss": 4.0535, "step": 19073 }, { "epoch": 0.19074, "grad_norm": 1.3021489620983207, "learning_rate": 0.003, "loss": 4.075, "step": 19074 }, { "epoch": 0.19075, "grad_norm": 1.0786772835814227, "learning_rate": 0.003, "loss": 4.0517, "step": 19075 }, { "epoch": 0.19076, "grad_norm": 1.5499784953997167, "learning_rate": 0.003, "loss": 4.0496, "step": 19076 }, { "epoch": 0.19077, "grad_norm": 1.2342223626061197, "learning_rate": 0.003, "loss": 4.0761, "step": 19077 }, { "epoch": 0.19078, "grad_norm": 1.1209715371426334, "learning_rate": 0.003, "loss": 4.0325, "step": 19078 }, { "epoch": 0.19079, "grad_norm": 1.2676347243302521, "learning_rate": 0.003, "loss": 4.0675, "step": 19079 }, { "epoch": 0.1908, "grad_norm": 1.1221281508566292, "learning_rate": 0.003, "loss": 4.0224, "step": 19080 }, { "epoch": 0.19081, "grad_norm": 1.260891655079296, "learning_rate": 0.003, "loss": 4.0316, "step": 19081 }, { "epoch": 0.19082, "grad_norm": 1.1017830975718048, "learning_rate": 0.003, "loss": 4.0481, "step": 19082 }, { "epoch": 0.19083, "grad_norm": 1.388700561330615, "learning_rate": 0.003, "loss": 4.0397, "step": 19083 }, { "epoch": 0.19084, "grad_norm": 1.34690955993557, "learning_rate": 0.003, "loss": 4.0558, "step": 19084 }, { "epoch": 0.19085, "grad_norm": 1.145241746931564, "learning_rate": 0.003, "loss": 4.0573, "step": 19085 }, { "epoch": 0.19086, "grad_norm": 1.270284682902497, "learning_rate": 0.003, "loss": 4.0445, "step": 19086 }, { "epoch": 0.19087, "grad_norm": 1.1500679421006432, "learning_rate": 0.003, "loss": 4.06, "step": 19087 }, { "epoch": 0.19088, "grad_norm": 1.5543662258719404, "learning_rate": 0.003, "loss": 4.0869, "step": 19088 }, { "epoch": 0.19089, "grad_norm": 0.8755120821039072, "learning_rate": 0.003, "loss": 4.0351, "step": 19089 }, { "epoch": 0.1909, "grad_norm": 1.4335401457780075, "learning_rate": 0.003, "loss": 4.0552, "step": 19090 }, { "epoch": 0.19091, "grad_norm": 1.385490472183994, "learning_rate": 0.003, "loss": 4.0706, "step": 19091 }, { "epoch": 0.19092, "grad_norm": 1.6226202331853985, "learning_rate": 0.003, "loss": 4.0325, "step": 19092 }, { "epoch": 0.19093, "grad_norm": 1.1890689330629287, "learning_rate": 0.003, "loss": 4.0508, "step": 19093 }, { "epoch": 0.19094, "grad_norm": 1.1981371155081517, "learning_rate": 0.003, "loss": 4.0697, "step": 19094 }, { "epoch": 0.19095, "grad_norm": 1.2893668593081675, "learning_rate": 0.003, "loss": 4.0634, "step": 19095 }, { "epoch": 0.19096, "grad_norm": 1.4960662294561393, "learning_rate": 0.003, "loss": 4.0701, "step": 19096 }, { "epoch": 0.19097, "grad_norm": 0.9725147576688425, "learning_rate": 0.003, "loss": 4.0615, "step": 19097 }, { "epoch": 0.19098, "grad_norm": 1.3685030885529894, "learning_rate": 0.003, "loss": 4.0449, "step": 19098 }, { "epoch": 0.19099, "grad_norm": 1.0622915877891217, "learning_rate": 0.003, "loss": 4.05, "step": 19099 }, { "epoch": 0.191, "grad_norm": 1.464669870198639, "learning_rate": 0.003, "loss": 4.087, "step": 19100 }, { "epoch": 0.19101, "grad_norm": 1.118779923692166, "learning_rate": 0.003, "loss": 4.0721, "step": 19101 }, { "epoch": 0.19102, "grad_norm": 1.2976349635461937, "learning_rate": 0.003, "loss": 4.0428, "step": 19102 }, { "epoch": 0.19103, "grad_norm": 1.1300313503024064, "learning_rate": 0.003, "loss": 4.0477, "step": 19103 }, { "epoch": 0.19104, "grad_norm": 1.2789988461407795, "learning_rate": 0.003, "loss": 4.0735, "step": 19104 }, { "epoch": 0.19105, "grad_norm": 1.2132765671340873, "learning_rate": 0.003, "loss": 4.0696, "step": 19105 }, { "epoch": 0.19106, "grad_norm": 1.2121676434130617, "learning_rate": 0.003, "loss": 4.0461, "step": 19106 }, { "epoch": 0.19107, "grad_norm": 1.265660415292461, "learning_rate": 0.003, "loss": 4.0596, "step": 19107 }, { "epoch": 0.19108, "grad_norm": 1.4014520774130477, "learning_rate": 0.003, "loss": 4.0593, "step": 19108 }, { "epoch": 0.19109, "grad_norm": 1.1756389944148185, "learning_rate": 0.003, "loss": 4.0825, "step": 19109 }, { "epoch": 0.1911, "grad_norm": 1.5469899981517996, "learning_rate": 0.003, "loss": 4.0707, "step": 19110 }, { "epoch": 0.19111, "grad_norm": 1.0752369359541158, "learning_rate": 0.003, "loss": 4.0458, "step": 19111 }, { "epoch": 0.19112, "grad_norm": 1.2913248218854734, "learning_rate": 0.003, "loss": 4.0711, "step": 19112 }, { "epoch": 0.19113, "grad_norm": 1.3731946415703815, "learning_rate": 0.003, "loss": 4.0685, "step": 19113 }, { "epoch": 0.19114, "grad_norm": 1.1777399714521335, "learning_rate": 0.003, "loss": 4.0302, "step": 19114 }, { "epoch": 0.19115, "grad_norm": 1.3621232493478581, "learning_rate": 0.003, "loss": 4.0552, "step": 19115 }, { "epoch": 0.19116, "grad_norm": 1.2428027654546105, "learning_rate": 0.003, "loss": 4.0718, "step": 19116 }, { "epoch": 0.19117, "grad_norm": 1.3920508306613928, "learning_rate": 0.003, "loss": 4.044, "step": 19117 }, { "epoch": 0.19118, "grad_norm": 1.1046257840388285, "learning_rate": 0.003, "loss": 4.0543, "step": 19118 }, { "epoch": 0.19119, "grad_norm": 1.3008624308439076, "learning_rate": 0.003, "loss": 4.0986, "step": 19119 }, { "epoch": 0.1912, "grad_norm": 1.0358300100952855, "learning_rate": 0.003, "loss": 4.0707, "step": 19120 }, { "epoch": 0.19121, "grad_norm": 1.2014291619295017, "learning_rate": 0.003, "loss": 4.0639, "step": 19121 }, { "epoch": 0.19122, "grad_norm": 1.3298764689448466, "learning_rate": 0.003, "loss": 4.0636, "step": 19122 }, { "epoch": 0.19123, "grad_norm": 1.2413092514637005, "learning_rate": 0.003, "loss": 4.0445, "step": 19123 }, { "epoch": 0.19124, "grad_norm": 1.4638470284624077, "learning_rate": 0.003, "loss": 4.0876, "step": 19124 }, { "epoch": 0.19125, "grad_norm": 1.2100119342079405, "learning_rate": 0.003, "loss": 4.0442, "step": 19125 }, { "epoch": 0.19126, "grad_norm": 1.327011420791088, "learning_rate": 0.003, "loss": 4.0296, "step": 19126 }, { "epoch": 0.19127, "grad_norm": 1.18672195015303, "learning_rate": 0.003, "loss": 4.0567, "step": 19127 }, { "epoch": 0.19128, "grad_norm": 1.2471773665078512, "learning_rate": 0.003, "loss": 4.0492, "step": 19128 }, { "epoch": 0.19129, "grad_norm": 1.1345648663615966, "learning_rate": 0.003, "loss": 4.0575, "step": 19129 }, { "epoch": 0.1913, "grad_norm": 1.249516390824633, "learning_rate": 0.003, "loss": 4.0715, "step": 19130 }, { "epoch": 0.19131, "grad_norm": 1.5043471304993827, "learning_rate": 0.003, "loss": 4.0531, "step": 19131 }, { "epoch": 0.19132, "grad_norm": 1.0754592345618441, "learning_rate": 0.003, "loss": 4.0469, "step": 19132 }, { "epoch": 0.19133, "grad_norm": 1.4711542619175533, "learning_rate": 0.003, "loss": 4.0325, "step": 19133 }, { "epoch": 0.19134, "grad_norm": 1.2202325514640542, "learning_rate": 0.003, "loss": 4.0856, "step": 19134 }, { "epoch": 0.19135, "grad_norm": 1.3349597063538905, "learning_rate": 0.003, "loss": 4.0716, "step": 19135 }, { "epoch": 0.19136, "grad_norm": 1.4219213256390757, "learning_rate": 0.003, "loss": 4.0534, "step": 19136 }, { "epoch": 0.19137, "grad_norm": 1.244969908564544, "learning_rate": 0.003, "loss": 4.026, "step": 19137 }, { "epoch": 0.19138, "grad_norm": 1.5286432652037785, "learning_rate": 0.003, "loss": 4.0645, "step": 19138 }, { "epoch": 0.19139, "grad_norm": 0.9895873720911743, "learning_rate": 0.003, "loss": 4.0803, "step": 19139 }, { "epoch": 0.1914, "grad_norm": 1.3173501475465692, "learning_rate": 0.003, "loss": 4.059, "step": 19140 }, { "epoch": 0.19141, "grad_norm": 1.2158916545602216, "learning_rate": 0.003, "loss": 4.0142, "step": 19141 }, { "epoch": 0.19142, "grad_norm": 1.1792868374867513, "learning_rate": 0.003, "loss": 4.0483, "step": 19142 }, { "epoch": 0.19143, "grad_norm": 1.2818615849270238, "learning_rate": 0.003, "loss": 4.0436, "step": 19143 }, { "epoch": 0.19144, "grad_norm": 1.5747934446702487, "learning_rate": 0.003, "loss": 4.0651, "step": 19144 }, { "epoch": 0.19145, "grad_norm": 1.1546309654274602, "learning_rate": 0.003, "loss": 4.0601, "step": 19145 }, { "epoch": 0.19146, "grad_norm": 1.2410780386836373, "learning_rate": 0.003, "loss": 4.0444, "step": 19146 }, { "epoch": 0.19147, "grad_norm": 1.1409885878401673, "learning_rate": 0.003, "loss": 4.055, "step": 19147 }, { "epoch": 0.19148, "grad_norm": 1.3436084461851314, "learning_rate": 0.003, "loss": 4.0642, "step": 19148 }, { "epoch": 0.19149, "grad_norm": 1.1699891038377133, "learning_rate": 0.003, "loss": 4.0417, "step": 19149 }, { "epoch": 0.1915, "grad_norm": 1.4725718423048586, "learning_rate": 0.003, "loss": 4.0533, "step": 19150 }, { "epoch": 0.19151, "grad_norm": 1.1693307627587524, "learning_rate": 0.003, "loss": 4.0643, "step": 19151 }, { "epoch": 0.19152, "grad_norm": 1.3657279001419607, "learning_rate": 0.003, "loss": 4.0305, "step": 19152 }, { "epoch": 0.19153, "grad_norm": 0.905534039598116, "learning_rate": 0.003, "loss": 4.0563, "step": 19153 }, { "epoch": 0.19154, "grad_norm": 1.2419837627588104, "learning_rate": 0.003, "loss": 4.0709, "step": 19154 }, { "epoch": 0.19155, "grad_norm": 1.4400848465098008, "learning_rate": 0.003, "loss": 4.0283, "step": 19155 }, { "epoch": 0.19156, "grad_norm": 1.2188726153413414, "learning_rate": 0.003, "loss": 4.0654, "step": 19156 }, { "epoch": 0.19157, "grad_norm": 1.3653131530262828, "learning_rate": 0.003, "loss": 4.0501, "step": 19157 }, { "epoch": 0.19158, "grad_norm": 1.3064082256285416, "learning_rate": 0.003, "loss": 4.0606, "step": 19158 }, { "epoch": 0.19159, "grad_norm": 1.090331273149747, "learning_rate": 0.003, "loss": 4.0769, "step": 19159 }, { "epoch": 0.1916, "grad_norm": 1.6293185159130206, "learning_rate": 0.003, "loss": 4.0438, "step": 19160 }, { "epoch": 0.19161, "grad_norm": 0.8287655198162486, "learning_rate": 0.003, "loss": 4.0454, "step": 19161 }, { "epoch": 0.19162, "grad_norm": 1.158652812687935, "learning_rate": 0.003, "loss": 4.0408, "step": 19162 }, { "epoch": 0.19163, "grad_norm": 1.248177992876627, "learning_rate": 0.003, "loss": 4.0484, "step": 19163 }, { "epoch": 0.19164, "grad_norm": 1.5455163093491473, "learning_rate": 0.003, "loss": 4.1117, "step": 19164 }, { "epoch": 0.19165, "grad_norm": 1.1528992311916821, "learning_rate": 0.003, "loss": 4.0418, "step": 19165 }, { "epoch": 0.19166, "grad_norm": 1.3662058618697008, "learning_rate": 0.003, "loss": 4.0524, "step": 19166 }, { "epoch": 0.19167, "grad_norm": 1.3966410134540599, "learning_rate": 0.003, "loss": 4.0777, "step": 19167 }, { "epoch": 0.19168, "grad_norm": 1.0744464750352165, "learning_rate": 0.003, "loss": 4.0235, "step": 19168 }, { "epoch": 0.19169, "grad_norm": 1.3750293897026336, "learning_rate": 0.003, "loss": 4.02, "step": 19169 }, { "epoch": 0.1917, "grad_norm": 1.1321040873817443, "learning_rate": 0.003, "loss": 4.0771, "step": 19170 }, { "epoch": 0.19171, "grad_norm": 1.4050662826257483, "learning_rate": 0.003, "loss": 4.0313, "step": 19171 }, { "epoch": 0.19172, "grad_norm": 1.2842379826432408, "learning_rate": 0.003, "loss": 4.0505, "step": 19172 }, { "epoch": 0.19173, "grad_norm": 1.4611207157151929, "learning_rate": 0.003, "loss": 4.0589, "step": 19173 }, { "epoch": 0.19174, "grad_norm": 0.9763677160665539, "learning_rate": 0.003, "loss": 4.0557, "step": 19174 }, { "epoch": 0.19175, "grad_norm": 1.190942187768536, "learning_rate": 0.003, "loss": 4.0933, "step": 19175 }, { "epoch": 0.19176, "grad_norm": 1.2381022913461883, "learning_rate": 0.003, "loss": 4.0469, "step": 19176 }, { "epoch": 0.19177, "grad_norm": 1.22330454111681, "learning_rate": 0.003, "loss": 4.0542, "step": 19177 }, { "epoch": 0.19178, "grad_norm": 1.2691358808693634, "learning_rate": 0.003, "loss": 4.0375, "step": 19178 }, { "epoch": 0.19179, "grad_norm": 1.2382625954537447, "learning_rate": 0.003, "loss": 4.0773, "step": 19179 }, { "epoch": 0.1918, "grad_norm": 1.4497885036032783, "learning_rate": 0.003, "loss": 4.0397, "step": 19180 }, { "epoch": 0.19181, "grad_norm": 1.134016258479399, "learning_rate": 0.003, "loss": 4.057, "step": 19181 }, { "epoch": 0.19182, "grad_norm": 1.35760379413228, "learning_rate": 0.003, "loss": 4.0583, "step": 19182 }, { "epoch": 0.19183, "grad_norm": 1.2957311215263954, "learning_rate": 0.003, "loss": 4.0649, "step": 19183 }, { "epoch": 0.19184, "grad_norm": 1.310887624893315, "learning_rate": 0.003, "loss": 4.0619, "step": 19184 }, { "epoch": 0.19185, "grad_norm": 1.4510683143720915, "learning_rate": 0.003, "loss": 4.0388, "step": 19185 }, { "epoch": 0.19186, "grad_norm": 1.2682977145649312, "learning_rate": 0.003, "loss": 4.0232, "step": 19186 }, { "epoch": 0.19187, "grad_norm": 1.575695837497368, "learning_rate": 0.003, "loss": 4.0739, "step": 19187 }, { "epoch": 0.19188, "grad_norm": 1.1916183188246712, "learning_rate": 0.003, "loss": 4.0475, "step": 19188 }, { "epoch": 0.19189, "grad_norm": 1.2328848455590875, "learning_rate": 0.003, "loss": 4.0266, "step": 19189 }, { "epoch": 0.1919, "grad_norm": 1.2209844780962866, "learning_rate": 0.003, "loss": 4.0506, "step": 19190 }, { "epoch": 0.19191, "grad_norm": 1.2732974615205064, "learning_rate": 0.003, "loss": 4.0515, "step": 19191 }, { "epoch": 0.19192, "grad_norm": 1.1609500022343568, "learning_rate": 0.003, "loss": 4.0132, "step": 19192 }, { "epoch": 0.19193, "grad_norm": 1.2852569005801213, "learning_rate": 0.003, "loss": 4.0539, "step": 19193 }, { "epoch": 0.19194, "grad_norm": 1.1691452818388632, "learning_rate": 0.003, "loss": 4.0372, "step": 19194 }, { "epoch": 0.19195, "grad_norm": 1.3702680762038288, "learning_rate": 0.003, "loss": 4.0521, "step": 19195 }, { "epoch": 0.19196, "grad_norm": 1.1694488089282924, "learning_rate": 0.003, "loss": 4.0504, "step": 19196 }, { "epoch": 0.19197, "grad_norm": 1.2473752069715442, "learning_rate": 0.003, "loss": 4.0348, "step": 19197 }, { "epoch": 0.19198, "grad_norm": 1.3688613100206077, "learning_rate": 0.003, "loss": 4.0388, "step": 19198 }, { "epoch": 0.19199, "grad_norm": 1.0233257482532478, "learning_rate": 0.003, "loss": 4.0692, "step": 19199 }, { "epoch": 0.192, "grad_norm": 1.561881594874595, "learning_rate": 0.003, "loss": 4.097, "step": 19200 }, { "epoch": 0.19201, "grad_norm": 1.0375767253882549, "learning_rate": 0.003, "loss": 4.0369, "step": 19201 }, { "epoch": 0.19202, "grad_norm": 1.4407943662007237, "learning_rate": 0.003, "loss": 4.0686, "step": 19202 }, { "epoch": 0.19203, "grad_norm": 1.2342685010454892, "learning_rate": 0.003, "loss": 4.0342, "step": 19203 }, { "epoch": 0.19204, "grad_norm": 1.4208038113256212, "learning_rate": 0.003, "loss": 4.0455, "step": 19204 }, { "epoch": 0.19205, "grad_norm": 1.3487435507325976, "learning_rate": 0.003, "loss": 4.0875, "step": 19205 }, { "epoch": 0.19206, "grad_norm": 1.1336355167725256, "learning_rate": 0.003, "loss": 4.0312, "step": 19206 }, { "epoch": 0.19207, "grad_norm": 1.3828811875751037, "learning_rate": 0.003, "loss": 4.0731, "step": 19207 }, { "epoch": 0.19208, "grad_norm": 0.9345781028127893, "learning_rate": 0.003, "loss": 4.0633, "step": 19208 }, { "epoch": 0.19209, "grad_norm": 1.333644111030958, "learning_rate": 0.003, "loss": 4.0778, "step": 19209 }, { "epoch": 0.1921, "grad_norm": 1.0937521969016353, "learning_rate": 0.003, "loss": 4.0624, "step": 19210 }, { "epoch": 0.19211, "grad_norm": 1.3620631814238218, "learning_rate": 0.003, "loss": 4.0386, "step": 19211 }, { "epoch": 0.19212, "grad_norm": 1.0229306203607218, "learning_rate": 0.003, "loss": 4.0562, "step": 19212 }, { "epoch": 0.19213, "grad_norm": 1.606807081278299, "learning_rate": 0.003, "loss": 4.0662, "step": 19213 }, { "epoch": 0.19214, "grad_norm": 1.0327706718080056, "learning_rate": 0.003, "loss": 4.0419, "step": 19214 }, { "epoch": 0.19215, "grad_norm": 1.631649256616336, "learning_rate": 0.003, "loss": 4.0863, "step": 19215 }, { "epoch": 0.19216, "grad_norm": 1.207976476510446, "learning_rate": 0.003, "loss": 4.0553, "step": 19216 }, { "epoch": 0.19217, "grad_norm": 1.3048150254088753, "learning_rate": 0.003, "loss": 4.0375, "step": 19217 }, { "epoch": 0.19218, "grad_norm": 1.3380053116505832, "learning_rate": 0.003, "loss": 4.1072, "step": 19218 }, { "epoch": 0.19219, "grad_norm": 1.0679714142709387, "learning_rate": 0.003, "loss": 4.0368, "step": 19219 }, { "epoch": 0.1922, "grad_norm": 1.4958575564073844, "learning_rate": 0.003, "loss": 4.0336, "step": 19220 }, { "epoch": 0.19221, "grad_norm": 1.117812451045177, "learning_rate": 0.003, "loss": 4.0906, "step": 19221 }, { "epoch": 0.19222, "grad_norm": 1.2419866483185766, "learning_rate": 0.003, "loss": 4.0491, "step": 19222 }, { "epoch": 0.19223, "grad_norm": 1.3768988883459534, "learning_rate": 0.003, "loss": 4.0627, "step": 19223 }, { "epoch": 0.19224, "grad_norm": 1.069206529611852, "learning_rate": 0.003, "loss": 4.0312, "step": 19224 }, { "epoch": 0.19225, "grad_norm": 1.3544075558994118, "learning_rate": 0.003, "loss": 4.0714, "step": 19225 }, { "epoch": 0.19226, "grad_norm": 1.130254621717398, "learning_rate": 0.003, "loss": 4.0455, "step": 19226 }, { "epoch": 0.19227, "grad_norm": 1.2701516631007297, "learning_rate": 0.003, "loss": 4.0651, "step": 19227 }, { "epoch": 0.19228, "grad_norm": 1.141542494705014, "learning_rate": 0.003, "loss": 4.0475, "step": 19228 }, { "epoch": 0.19229, "grad_norm": 1.7278661090592242, "learning_rate": 0.003, "loss": 4.045, "step": 19229 }, { "epoch": 0.1923, "grad_norm": 0.8602523351380327, "learning_rate": 0.003, "loss": 4.0621, "step": 19230 }, { "epoch": 0.19231, "grad_norm": 1.0008049284928373, "learning_rate": 0.003, "loss": 4.0621, "step": 19231 }, { "epoch": 0.19232, "grad_norm": 1.6441796376548221, "learning_rate": 0.003, "loss": 4.0548, "step": 19232 }, { "epoch": 0.19233, "grad_norm": 1.06348186022257, "learning_rate": 0.003, "loss": 4.05, "step": 19233 }, { "epoch": 0.19234, "grad_norm": 1.5056261816852763, "learning_rate": 0.003, "loss": 4.0502, "step": 19234 }, { "epoch": 0.19235, "grad_norm": 1.059974211670759, "learning_rate": 0.003, "loss": 4.071, "step": 19235 }, { "epoch": 0.19236, "grad_norm": 1.4293178717403514, "learning_rate": 0.003, "loss": 4.0796, "step": 19236 }, { "epoch": 0.19237, "grad_norm": 1.3128062170109287, "learning_rate": 0.003, "loss": 4.0568, "step": 19237 }, { "epoch": 0.19238, "grad_norm": 1.2427623419926968, "learning_rate": 0.003, "loss": 4.0476, "step": 19238 }, { "epoch": 0.19239, "grad_norm": 1.3914305120241952, "learning_rate": 0.003, "loss": 4.0573, "step": 19239 }, { "epoch": 0.1924, "grad_norm": 1.1844701559061308, "learning_rate": 0.003, "loss": 4.0299, "step": 19240 }, { "epoch": 0.19241, "grad_norm": 1.3358176920204796, "learning_rate": 0.003, "loss": 4.0544, "step": 19241 }, { "epoch": 0.19242, "grad_norm": 1.2900014037234846, "learning_rate": 0.003, "loss": 4.064, "step": 19242 }, { "epoch": 0.19243, "grad_norm": 1.303591170485071, "learning_rate": 0.003, "loss": 4.0772, "step": 19243 }, { "epoch": 0.19244, "grad_norm": 1.259426315541422, "learning_rate": 0.003, "loss": 4.0512, "step": 19244 }, { "epoch": 0.19245, "grad_norm": 1.09817722925374, "learning_rate": 0.003, "loss": 4.0498, "step": 19245 }, { "epoch": 0.19246, "grad_norm": 1.4539360635886593, "learning_rate": 0.003, "loss": 4.0633, "step": 19246 }, { "epoch": 0.19247, "grad_norm": 1.3018116569648512, "learning_rate": 0.003, "loss": 4.045, "step": 19247 }, { "epoch": 0.19248, "grad_norm": 1.1432748974280003, "learning_rate": 0.003, "loss": 4.0314, "step": 19248 }, { "epoch": 0.19249, "grad_norm": 1.4700394550063518, "learning_rate": 0.003, "loss": 4.0315, "step": 19249 }, { "epoch": 0.1925, "grad_norm": 0.9896072843511874, "learning_rate": 0.003, "loss": 4.0641, "step": 19250 }, { "epoch": 0.19251, "grad_norm": 1.3138089628103964, "learning_rate": 0.003, "loss": 4.0707, "step": 19251 }, { "epoch": 0.19252, "grad_norm": 1.0817784890206963, "learning_rate": 0.003, "loss": 4.0379, "step": 19252 }, { "epoch": 0.19253, "grad_norm": 1.3153319585561318, "learning_rate": 0.003, "loss": 4.0671, "step": 19253 }, { "epoch": 0.19254, "grad_norm": 1.2170125022548548, "learning_rate": 0.003, "loss": 4.1, "step": 19254 }, { "epoch": 0.19255, "grad_norm": 1.2694224062537878, "learning_rate": 0.003, "loss": 4.0782, "step": 19255 }, { "epoch": 0.19256, "grad_norm": 1.2453691515563632, "learning_rate": 0.003, "loss": 4.0509, "step": 19256 }, { "epoch": 0.19257, "grad_norm": 1.2349272540389504, "learning_rate": 0.003, "loss": 4.0269, "step": 19257 }, { "epoch": 0.19258, "grad_norm": 1.2688291821025708, "learning_rate": 0.003, "loss": 4.0458, "step": 19258 }, { "epoch": 0.19259, "grad_norm": 1.2700821480023927, "learning_rate": 0.003, "loss": 4.0617, "step": 19259 }, { "epoch": 0.1926, "grad_norm": 1.3579286957148133, "learning_rate": 0.003, "loss": 4.0601, "step": 19260 }, { "epoch": 0.19261, "grad_norm": 1.4739489420098513, "learning_rate": 0.003, "loss": 4.0475, "step": 19261 }, { "epoch": 0.19262, "grad_norm": 0.8635668416528791, "learning_rate": 0.003, "loss": 4.0326, "step": 19262 }, { "epoch": 0.19263, "grad_norm": 1.327948158133326, "learning_rate": 0.003, "loss": 4.0582, "step": 19263 }, { "epoch": 0.19264, "grad_norm": 1.3390826730360608, "learning_rate": 0.003, "loss": 4.0626, "step": 19264 }, { "epoch": 0.19265, "grad_norm": 1.217363022053805, "learning_rate": 0.003, "loss": 4.05, "step": 19265 }, { "epoch": 0.19266, "grad_norm": 1.182243398179855, "learning_rate": 0.003, "loss": 4.0576, "step": 19266 }, { "epoch": 0.19267, "grad_norm": 1.4727125262172842, "learning_rate": 0.003, "loss": 4.0459, "step": 19267 }, { "epoch": 0.19268, "grad_norm": 1.0565346665110982, "learning_rate": 0.003, "loss": 4.0333, "step": 19268 }, { "epoch": 0.19269, "grad_norm": 1.318652043931812, "learning_rate": 0.003, "loss": 4.0649, "step": 19269 }, { "epoch": 0.1927, "grad_norm": 1.1072118070146826, "learning_rate": 0.003, "loss": 4.0468, "step": 19270 }, { "epoch": 0.19271, "grad_norm": 1.362157172652143, "learning_rate": 0.003, "loss": 4.0605, "step": 19271 }, { "epoch": 0.19272, "grad_norm": 1.1281091123143556, "learning_rate": 0.003, "loss": 4.0565, "step": 19272 }, { "epoch": 0.19273, "grad_norm": 1.292803290024137, "learning_rate": 0.003, "loss": 4.0589, "step": 19273 }, { "epoch": 0.19274, "grad_norm": 1.2353272280198817, "learning_rate": 0.003, "loss": 4.0815, "step": 19274 }, { "epoch": 0.19275, "grad_norm": 1.1438153015330619, "learning_rate": 0.003, "loss": 4.0395, "step": 19275 }, { "epoch": 0.19276, "grad_norm": 1.4514607843346614, "learning_rate": 0.003, "loss": 4.0564, "step": 19276 }, { "epoch": 0.19277, "grad_norm": 1.2813060281552016, "learning_rate": 0.003, "loss": 4.0928, "step": 19277 }, { "epoch": 0.19278, "grad_norm": 1.6930936661015001, "learning_rate": 0.003, "loss": 4.0746, "step": 19278 }, { "epoch": 0.19279, "grad_norm": 1.1007912171244085, "learning_rate": 0.003, "loss": 4.0315, "step": 19279 }, { "epoch": 0.1928, "grad_norm": 1.1826340704770322, "learning_rate": 0.003, "loss": 4.0679, "step": 19280 }, { "epoch": 0.19281, "grad_norm": 1.2769700503974448, "learning_rate": 0.003, "loss": 4.0529, "step": 19281 }, { "epoch": 0.19282, "grad_norm": 1.2924742929851223, "learning_rate": 0.003, "loss": 4.0442, "step": 19282 }, { "epoch": 0.19283, "grad_norm": 1.234831979304908, "learning_rate": 0.003, "loss": 4.0401, "step": 19283 }, { "epoch": 0.19284, "grad_norm": 1.2207811629021978, "learning_rate": 0.003, "loss": 4.0793, "step": 19284 }, { "epoch": 0.19285, "grad_norm": 1.2816252739865883, "learning_rate": 0.003, "loss": 4.0366, "step": 19285 }, { "epoch": 0.19286, "grad_norm": 1.209800978176332, "learning_rate": 0.003, "loss": 4.0208, "step": 19286 }, { "epoch": 0.19287, "grad_norm": 1.2372222252209188, "learning_rate": 0.003, "loss": 4.0448, "step": 19287 }, { "epoch": 0.19288, "grad_norm": 1.158777792355288, "learning_rate": 0.003, "loss": 4.0587, "step": 19288 }, { "epoch": 0.19289, "grad_norm": 1.359679154466316, "learning_rate": 0.003, "loss": 4.0396, "step": 19289 }, { "epoch": 0.1929, "grad_norm": 0.960495143521634, "learning_rate": 0.003, "loss": 4.0591, "step": 19290 }, { "epoch": 0.19291, "grad_norm": 1.4301214559126807, "learning_rate": 0.003, "loss": 4.0759, "step": 19291 }, { "epoch": 0.19292, "grad_norm": 1.217408751778196, "learning_rate": 0.003, "loss": 4.0546, "step": 19292 }, { "epoch": 0.19293, "grad_norm": 1.3874768024193218, "learning_rate": 0.003, "loss": 4.0509, "step": 19293 }, { "epoch": 0.19294, "grad_norm": 1.332721404459955, "learning_rate": 0.003, "loss": 4.0803, "step": 19294 }, { "epoch": 0.19295, "grad_norm": 1.2173838719965513, "learning_rate": 0.003, "loss": 4.0387, "step": 19295 }, { "epoch": 0.19296, "grad_norm": 1.2916412791849425, "learning_rate": 0.003, "loss": 4.0377, "step": 19296 }, { "epoch": 0.19297, "grad_norm": 1.296134930093045, "learning_rate": 0.003, "loss": 4.0507, "step": 19297 }, { "epoch": 0.19298, "grad_norm": 1.1697870555772907, "learning_rate": 0.003, "loss": 4.0325, "step": 19298 }, { "epoch": 0.19299, "grad_norm": 1.1415884021127054, "learning_rate": 0.003, "loss": 4.0348, "step": 19299 }, { "epoch": 0.193, "grad_norm": 1.3121021912821669, "learning_rate": 0.003, "loss": 4.0565, "step": 19300 }, { "epoch": 0.19301, "grad_norm": 1.2277104069030647, "learning_rate": 0.003, "loss": 4.0807, "step": 19301 }, { "epoch": 0.19302, "grad_norm": 1.168389647950443, "learning_rate": 0.003, "loss": 4.0498, "step": 19302 }, { "epoch": 0.19303, "grad_norm": 1.199071860372418, "learning_rate": 0.003, "loss": 4.0695, "step": 19303 }, { "epoch": 0.19304, "grad_norm": 1.6197362151787464, "learning_rate": 0.003, "loss": 4.0618, "step": 19304 }, { "epoch": 0.19305, "grad_norm": 1.3872492287681017, "learning_rate": 0.003, "loss": 4.0578, "step": 19305 }, { "epoch": 0.19306, "grad_norm": 1.1968252615562611, "learning_rate": 0.003, "loss": 4.0489, "step": 19306 }, { "epoch": 0.19307, "grad_norm": 1.4302807502281258, "learning_rate": 0.003, "loss": 4.0683, "step": 19307 }, { "epoch": 0.19308, "grad_norm": 1.130676365717826, "learning_rate": 0.003, "loss": 4.0451, "step": 19308 }, { "epoch": 0.19309, "grad_norm": 1.4403273412737394, "learning_rate": 0.003, "loss": 4.064, "step": 19309 }, { "epoch": 0.1931, "grad_norm": 1.0710180781601069, "learning_rate": 0.003, "loss": 4.06, "step": 19310 }, { "epoch": 0.19311, "grad_norm": 1.3604383674442513, "learning_rate": 0.003, "loss": 4.0497, "step": 19311 }, { "epoch": 0.19312, "grad_norm": 1.093479747707285, "learning_rate": 0.003, "loss": 4.0497, "step": 19312 }, { "epoch": 0.19313, "grad_norm": 1.3405809261925952, "learning_rate": 0.003, "loss": 4.0875, "step": 19313 }, { "epoch": 0.19314, "grad_norm": 1.0784131713588263, "learning_rate": 0.003, "loss": 4.0575, "step": 19314 }, { "epoch": 0.19315, "grad_norm": 1.1059936570643827, "learning_rate": 0.003, "loss": 4.0217, "step": 19315 }, { "epoch": 0.19316, "grad_norm": 1.2920533396885687, "learning_rate": 0.003, "loss": 4.0375, "step": 19316 }, { "epoch": 0.19317, "grad_norm": 1.119322072880016, "learning_rate": 0.003, "loss": 4.0375, "step": 19317 }, { "epoch": 0.19318, "grad_norm": 1.2110968387282244, "learning_rate": 0.003, "loss": 4.0407, "step": 19318 }, { "epoch": 0.19319, "grad_norm": 1.2244841200983902, "learning_rate": 0.003, "loss": 4.0495, "step": 19319 }, { "epoch": 0.1932, "grad_norm": 0.9959893152714806, "learning_rate": 0.003, "loss": 4.049, "step": 19320 }, { "epoch": 0.19321, "grad_norm": 1.7272148138632006, "learning_rate": 0.003, "loss": 4.056, "step": 19321 }, { "epoch": 0.19322, "grad_norm": 1.266154287794963, "learning_rate": 0.003, "loss": 4.0745, "step": 19322 }, { "epoch": 0.19323, "grad_norm": 1.5853357914133883, "learning_rate": 0.003, "loss": 4.0523, "step": 19323 }, { "epoch": 0.19324, "grad_norm": 1.1858059760863204, "learning_rate": 0.003, "loss": 4.0717, "step": 19324 }, { "epoch": 0.19325, "grad_norm": 1.3031647803358464, "learning_rate": 0.003, "loss": 4.056, "step": 19325 }, { "epoch": 0.19326, "grad_norm": 1.4328634660837625, "learning_rate": 0.003, "loss": 4.0709, "step": 19326 }, { "epoch": 0.19327, "grad_norm": 1.2369860960951975, "learning_rate": 0.003, "loss": 4.0741, "step": 19327 }, { "epoch": 0.19328, "grad_norm": 1.2614163252370505, "learning_rate": 0.003, "loss": 4.0737, "step": 19328 }, { "epoch": 0.19329, "grad_norm": 1.162931386262037, "learning_rate": 0.003, "loss": 4.0702, "step": 19329 }, { "epoch": 0.1933, "grad_norm": 1.330709905124812, "learning_rate": 0.003, "loss": 4.026, "step": 19330 }, { "epoch": 0.19331, "grad_norm": 1.158193294800109, "learning_rate": 0.003, "loss": 4.0446, "step": 19331 }, { "epoch": 0.19332, "grad_norm": 1.3349312226282135, "learning_rate": 0.003, "loss": 4.0495, "step": 19332 }, { "epoch": 0.19333, "grad_norm": 1.3713505684636866, "learning_rate": 0.003, "loss": 4.0598, "step": 19333 }, { "epoch": 0.19334, "grad_norm": 1.4973548136509773, "learning_rate": 0.003, "loss": 4.0166, "step": 19334 }, { "epoch": 0.19335, "grad_norm": 1.2475000046473035, "learning_rate": 0.003, "loss": 4.0283, "step": 19335 }, { "epoch": 0.19336, "grad_norm": 1.4186242337546051, "learning_rate": 0.003, "loss": 4.041, "step": 19336 }, { "epoch": 0.19337, "grad_norm": 1.0438410940110374, "learning_rate": 0.003, "loss": 4.0449, "step": 19337 }, { "epoch": 0.19338, "grad_norm": 1.281900229011639, "learning_rate": 0.003, "loss": 4.0833, "step": 19338 }, { "epoch": 0.19339, "grad_norm": 1.179938674265604, "learning_rate": 0.003, "loss": 4.0507, "step": 19339 }, { "epoch": 0.1934, "grad_norm": 1.3683485649923222, "learning_rate": 0.003, "loss": 4.0563, "step": 19340 }, { "epoch": 0.19341, "grad_norm": 1.093382737839294, "learning_rate": 0.003, "loss": 4.0267, "step": 19341 }, { "epoch": 0.19342, "grad_norm": 1.5809442527601474, "learning_rate": 0.003, "loss": 4.0723, "step": 19342 }, { "epoch": 0.19343, "grad_norm": 0.9839961061261212, "learning_rate": 0.003, "loss": 4.0574, "step": 19343 }, { "epoch": 0.19344, "grad_norm": 1.513270827948988, "learning_rate": 0.003, "loss": 4.0626, "step": 19344 }, { "epoch": 0.19345, "grad_norm": 1.1007128440719929, "learning_rate": 0.003, "loss": 4.0689, "step": 19345 }, { "epoch": 0.19346, "grad_norm": 1.2374639283484485, "learning_rate": 0.003, "loss": 4.0263, "step": 19346 }, { "epoch": 0.19347, "grad_norm": 1.3191176372983453, "learning_rate": 0.003, "loss": 4.0727, "step": 19347 }, { "epoch": 0.19348, "grad_norm": 1.430232717602349, "learning_rate": 0.003, "loss": 4.1079, "step": 19348 }, { "epoch": 0.19349, "grad_norm": 1.3784803258619929, "learning_rate": 0.003, "loss": 4.0445, "step": 19349 }, { "epoch": 0.1935, "grad_norm": 1.1225933161395825, "learning_rate": 0.003, "loss": 4.0592, "step": 19350 }, { "epoch": 0.19351, "grad_norm": 1.340969744963439, "learning_rate": 0.003, "loss": 4.0657, "step": 19351 }, { "epoch": 0.19352, "grad_norm": 1.4495783809268945, "learning_rate": 0.003, "loss": 4.0593, "step": 19352 }, { "epoch": 0.19353, "grad_norm": 1.7421593307521097, "learning_rate": 0.003, "loss": 4.0624, "step": 19353 }, { "epoch": 0.19354, "grad_norm": 0.9658945600154332, "learning_rate": 0.003, "loss": 4.0646, "step": 19354 }, { "epoch": 0.19355, "grad_norm": 1.205921178736843, "learning_rate": 0.003, "loss": 4.0609, "step": 19355 }, { "epoch": 0.19356, "grad_norm": 1.2757237919629527, "learning_rate": 0.003, "loss": 4.0639, "step": 19356 }, { "epoch": 0.19357, "grad_norm": 1.3411673563909947, "learning_rate": 0.003, "loss": 4.0346, "step": 19357 }, { "epoch": 0.19358, "grad_norm": 1.2848950228013367, "learning_rate": 0.003, "loss": 4.0675, "step": 19358 }, { "epoch": 0.19359, "grad_norm": 1.3013320004570212, "learning_rate": 0.003, "loss": 4.0414, "step": 19359 }, { "epoch": 0.1936, "grad_norm": 1.0435278072734142, "learning_rate": 0.003, "loss": 4.0729, "step": 19360 }, { "epoch": 0.19361, "grad_norm": 1.287288538323467, "learning_rate": 0.003, "loss": 4.054, "step": 19361 }, { "epoch": 0.19362, "grad_norm": 1.110274706607819, "learning_rate": 0.003, "loss": 4.0378, "step": 19362 }, { "epoch": 0.19363, "grad_norm": 1.3060179286263724, "learning_rate": 0.003, "loss": 4.0668, "step": 19363 }, { "epoch": 0.19364, "grad_norm": 1.0599462956406744, "learning_rate": 0.003, "loss": 4.0493, "step": 19364 }, { "epoch": 0.19365, "grad_norm": 1.4188548021615779, "learning_rate": 0.003, "loss": 4.0658, "step": 19365 }, { "epoch": 0.19366, "grad_norm": 0.9977169189433731, "learning_rate": 0.003, "loss": 4.0575, "step": 19366 }, { "epoch": 0.19367, "grad_norm": 1.6418750673442695, "learning_rate": 0.003, "loss": 4.0658, "step": 19367 }, { "epoch": 0.19368, "grad_norm": 0.915761451677153, "learning_rate": 0.003, "loss": 4.0358, "step": 19368 }, { "epoch": 0.19369, "grad_norm": 1.36982814408899, "learning_rate": 0.003, "loss": 4.0555, "step": 19369 }, { "epoch": 0.1937, "grad_norm": 1.4218831422973166, "learning_rate": 0.003, "loss": 4.0589, "step": 19370 }, { "epoch": 0.19371, "grad_norm": 1.1392214472527857, "learning_rate": 0.003, "loss": 4.0538, "step": 19371 }, { "epoch": 0.19372, "grad_norm": 1.2130464242599384, "learning_rate": 0.003, "loss": 4.0542, "step": 19372 }, { "epoch": 0.19373, "grad_norm": 1.4550828458250995, "learning_rate": 0.003, "loss": 4.0727, "step": 19373 }, { "epoch": 0.19374, "grad_norm": 1.21073894464927, "learning_rate": 0.003, "loss": 4.0395, "step": 19374 }, { "epoch": 0.19375, "grad_norm": 1.4476681629819754, "learning_rate": 0.003, "loss": 4.0651, "step": 19375 }, { "epoch": 0.19376, "grad_norm": 1.4949961065263637, "learning_rate": 0.003, "loss": 4.0625, "step": 19376 }, { "epoch": 0.19377, "grad_norm": 1.1271841630809518, "learning_rate": 0.003, "loss": 4.0674, "step": 19377 }, { "epoch": 0.19378, "grad_norm": 1.3807963456598449, "learning_rate": 0.003, "loss": 4.0844, "step": 19378 }, { "epoch": 0.19379, "grad_norm": 1.0596898741955287, "learning_rate": 0.003, "loss": 4.0199, "step": 19379 }, { "epoch": 0.1938, "grad_norm": 1.4072597453787985, "learning_rate": 0.003, "loss": 4.0664, "step": 19380 }, { "epoch": 0.19381, "grad_norm": 0.9550860981623658, "learning_rate": 0.003, "loss": 4.0446, "step": 19381 }, { "epoch": 0.19382, "grad_norm": 1.4785497463244524, "learning_rate": 0.003, "loss": 4.0765, "step": 19382 }, { "epoch": 0.19383, "grad_norm": 1.0397212461472258, "learning_rate": 0.003, "loss": 4.0315, "step": 19383 }, { "epoch": 0.19384, "grad_norm": 1.412362570967319, "learning_rate": 0.003, "loss": 4.0977, "step": 19384 }, { "epoch": 0.19385, "grad_norm": 1.2552527247210523, "learning_rate": 0.003, "loss": 4.0633, "step": 19385 }, { "epoch": 0.19386, "grad_norm": 0.9757782727735469, "learning_rate": 0.003, "loss": 4.018, "step": 19386 }, { "epoch": 0.19387, "grad_norm": 1.5585007953127197, "learning_rate": 0.003, "loss": 4.0737, "step": 19387 }, { "epoch": 0.19388, "grad_norm": 1.2592213733025597, "learning_rate": 0.003, "loss": 4.0603, "step": 19388 }, { "epoch": 0.19389, "grad_norm": 1.0797725144848942, "learning_rate": 0.003, "loss": 4.0629, "step": 19389 }, { "epoch": 0.1939, "grad_norm": 1.1894014821171297, "learning_rate": 0.003, "loss": 4.0673, "step": 19390 }, { "epoch": 0.19391, "grad_norm": 1.3322313361724551, "learning_rate": 0.003, "loss": 4.0662, "step": 19391 }, { "epoch": 0.19392, "grad_norm": 1.3131930257013624, "learning_rate": 0.003, "loss": 4.0532, "step": 19392 }, { "epoch": 0.19393, "grad_norm": 1.2090617157853354, "learning_rate": 0.003, "loss": 4.0363, "step": 19393 }, { "epoch": 0.19394, "grad_norm": 1.2370155160264173, "learning_rate": 0.003, "loss": 4.0412, "step": 19394 }, { "epoch": 0.19395, "grad_norm": 1.1644921328866482, "learning_rate": 0.003, "loss": 4.0483, "step": 19395 }, { "epoch": 0.19396, "grad_norm": 1.2434079928932653, "learning_rate": 0.003, "loss": 4.057, "step": 19396 }, { "epoch": 0.19397, "grad_norm": 1.0364099563185734, "learning_rate": 0.003, "loss": 4.0566, "step": 19397 }, { "epoch": 0.19398, "grad_norm": 1.836051257592041, "learning_rate": 0.003, "loss": 4.0558, "step": 19398 }, { "epoch": 0.19399, "grad_norm": 1.005551153961616, "learning_rate": 0.003, "loss": 4.068, "step": 19399 }, { "epoch": 0.194, "grad_norm": 1.2868116846193804, "learning_rate": 0.003, "loss": 4.0598, "step": 19400 }, { "epoch": 0.19401, "grad_norm": 1.1521465841753658, "learning_rate": 0.003, "loss": 4.0484, "step": 19401 }, { "epoch": 0.19402, "grad_norm": 1.300040789113978, "learning_rate": 0.003, "loss": 4.0497, "step": 19402 }, { "epoch": 0.19403, "grad_norm": 1.1206071310857861, "learning_rate": 0.003, "loss": 4.0483, "step": 19403 }, { "epoch": 0.19404, "grad_norm": 1.2770638746309435, "learning_rate": 0.003, "loss": 4.0612, "step": 19404 }, { "epoch": 0.19405, "grad_norm": 1.1000698654299443, "learning_rate": 0.003, "loss": 4.0538, "step": 19405 }, { "epoch": 0.19406, "grad_norm": 1.3706451663448482, "learning_rate": 0.003, "loss": 4.0273, "step": 19406 }, { "epoch": 0.19407, "grad_norm": 1.322881375041662, "learning_rate": 0.003, "loss": 4.057, "step": 19407 }, { "epoch": 0.19408, "grad_norm": 1.256616975675363, "learning_rate": 0.003, "loss": 4.0545, "step": 19408 }, { "epoch": 0.19409, "grad_norm": 1.157319297162313, "learning_rate": 0.003, "loss": 4.0642, "step": 19409 }, { "epoch": 0.1941, "grad_norm": 1.2604883125122113, "learning_rate": 0.003, "loss": 4.0708, "step": 19410 }, { "epoch": 0.19411, "grad_norm": 1.2465910754730276, "learning_rate": 0.003, "loss": 4.0759, "step": 19411 }, { "epoch": 0.19412, "grad_norm": 1.2902604302149527, "learning_rate": 0.003, "loss": 4.0432, "step": 19412 }, { "epoch": 0.19413, "grad_norm": 1.2428339542153704, "learning_rate": 0.003, "loss": 4.0566, "step": 19413 }, { "epoch": 0.19414, "grad_norm": 1.641619198006355, "learning_rate": 0.003, "loss": 4.0448, "step": 19414 }, { "epoch": 0.19415, "grad_norm": 1.1884015575175786, "learning_rate": 0.003, "loss": 4.0172, "step": 19415 }, { "epoch": 0.19416, "grad_norm": 1.4006931979815456, "learning_rate": 0.003, "loss": 4.0495, "step": 19416 }, { "epoch": 0.19417, "grad_norm": 1.1684221143708646, "learning_rate": 0.003, "loss": 4.069, "step": 19417 }, { "epoch": 0.19418, "grad_norm": 1.4080870540690456, "learning_rate": 0.003, "loss": 4.0387, "step": 19418 }, { "epoch": 0.19419, "grad_norm": 1.2924752860876128, "learning_rate": 0.003, "loss": 4.0481, "step": 19419 }, { "epoch": 0.1942, "grad_norm": 1.1857590064382597, "learning_rate": 0.003, "loss": 4.0336, "step": 19420 }, { "epoch": 0.19421, "grad_norm": 1.2224389378484004, "learning_rate": 0.003, "loss": 4.0398, "step": 19421 }, { "epoch": 0.19422, "grad_norm": 1.5321972622089721, "learning_rate": 0.003, "loss": 4.0594, "step": 19422 }, { "epoch": 0.19423, "grad_norm": 0.9439131538154166, "learning_rate": 0.003, "loss": 4.057, "step": 19423 }, { "epoch": 0.19424, "grad_norm": 1.0779829074698497, "learning_rate": 0.003, "loss": 4.0568, "step": 19424 }, { "epoch": 0.19425, "grad_norm": 1.4187367087645348, "learning_rate": 0.003, "loss": 4.048, "step": 19425 }, { "epoch": 0.19426, "grad_norm": 1.2518657015206214, "learning_rate": 0.003, "loss": 4.0653, "step": 19426 }, { "epoch": 0.19427, "grad_norm": 1.2975650890704618, "learning_rate": 0.003, "loss": 4.0445, "step": 19427 }, { "epoch": 0.19428, "grad_norm": 1.2504288611968093, "learning_rate": 0.003, "loss": 4.0377, "step": 19428 }, { "epoch": 0.19429, "grad_norm": 1.0626854032608262, "learning_rate": 0.003, "loss": 4.0599, "step": 19429 }, { "epoch": 0.1943, "grad_norm": 1.453944232695203, "learning_rate": 0.003, "loss": 4.0804, "step": 19430 }, { "epoch": 0.19431, "grad_norm": 1.1132415563210114, "learning_rate": 0.003, "loss": 4.0178, "step": 19431 }, { "epoch": 0.19432, "grad_norm": 1.330780831477712, "learning_rate": 0.003, "loss": 4.0605, "step": 19432 }, { "epoch": 0.19433, "grad_norm": 1.2264764129711287, "learning_rate": 0.003, "loss": 4.05, "step": 19433 }, { "epoch": 0.19434, "grad_norm": 1.119671719753815, "learning_rate": 0.003, "loss": 4.0635, "step": 19434 }, { "epoch": 0.19435, "grad_norm": 1.4390097295078403, "learning_rate": 0.003, "loss": 4.0686, "step": 19435 }, { "epoch": 0.19436, "grad_norm": 1.0497880754472968, "learning_rate": 0.003, "loss": 4.067, "step": 19436 }, { "epoch": 0.19437, "grad_norm": 1.3619036851797004, "learning_rate": 0.003, "loss": 4.0351, "step": 19437 }, { "epoch": 0.19438, "grad_norm": 1.1321332926155272, "learning_rate": 0.003, "loss": 4.0695, "step": 19438 }, { "epoch": 0.19439, "grad_norm": 1.3594115203490629, "learning_rate": 0.003, "loss": 4.0702, "step": 19439 }, { "epoch": 0.1944, "grad_norm": 1.1652985939647977, "learning_rate": 0.003, "loss": 4.0811, "step": 19440 }, { "epoch": 0.19441, "grad_norm": 1.3598903677041336, "learning_rate": 0.003, "loss": 4.0853, "step": 19441 }, { "epoch": 0.19442, "grad_norm": 1.2438339406836934, "learning_rate": 0.003, "loss": 4.0677, "step": 19442 }, { "epoch": 0.19443, "grad_norm": 1.2762886643194773, "learning_rate": 0.003, "loss": 4.0802, "step": 19443 }, { "epoch": 0.19444, "grad_norm": 1.1623881767163036, "learning_rate": 0.003, "loss": 4.0397, "step": 19444 }, { "epoch": 0.19445, "grad_norm": 1.5096350045671685, "learning_rate": 0.003, "loss": 4.0717, "step": 19445 }, { "epoch": 0.19446, "grad_norm": 1.1648388825625222, "learning_rate": 0.003, "loss": 4.0517, "step": 19446 }, { "epoch": 0.19447, "grad_norm": 1.4692385629412785, "learning_rate": 0.003, "loss": 4.0419, "step": 19447 }, { "epoch": 0.19448, "grad_norm": 1.3199067754255398, "learning_rate": 0.003, "loss": 4.0454, "step": 19448 }, { "epoch": 0.19449, "grad_norm": 1.1597482093135931, "learning_rate": 0.003, "loss": 4.0622, "step": 19449 }, { "epoch": 0.1945, "grad_norm": 1.2452753354825359, "learning_rate": 0.003, "loss": 4.0419, "step": 19450 }, { "epoch": 0.19451, "grad_norm": 1.1868649662414918, "learning_rate": 0.003, "loss": 4.0555, "step": 19451 }, { "epoch": 0.19452, "grad_norm": 1.245608401244629, "learning_rate": 0.003, "loss": 4.031, "step": 19452 }, { "epoch": 0.19453, "grad_norm": 1.2706894870815892, "learning_rate": 0.003, "loss": 4.0692, "step": 19453 }, { "epoch": 0.19454, "grad_norm": 1.1640633667913651, "learning_rate": 0.003, "loss": 4.0738, "step": 19454 }, { "epoch": 0.19455, "grad_norm": 1.3139941101204728, "learning_rate": 0.003, "loss": 4.0673, "step": 19455 }, { "epoch": 0.19456, "grad_norm": 1.2765744305834983, "learning_rate": 0.003, "loss": 4.0825, "step": 19456 }, { "epoch": 0.19457, "grad_norm": 1.0679899936936712, "learning_rate": 0.003, "loss": 4.0566, "step": 19457 }, { "epoch": 0.19458, "grad_norm": 1.156260712455033, "learning_rate": 0.003, "loss": 4.0554, "step": 19458 }, { "epoch": 0.19459, "grad_norm": 1.043446962809172, "learning_rate": 0.003, "loss": 4.0481, "step": 19459 }, { "epoch": 0.1946, "grad_norm": 1.7189150230765415, "learning_rate": 0.003, "loss": 4.0482, "step": 19460 }, { "epoch": 0.19461, "grad_norm": 1.0548421420952567, "learning_rate": 0.003, "loss": 4.1019, "step": 19461 }, { "epoch": 0.19462, "grad_norm": 1.5763941439210714, "learning_rate": 0.003, "loss": 4.0773, "step": 19462 }, { "epoch": 0.19463, "grad_norm": 0.9633473320637942, "learning_rate": 0.003, "loss": 4.0681, "step": 19463 }, { "epoch": 0.19464, "grad_norm": 1.3369137592285703, "learning_rate": 0.003, "loss": 4.0333, "step": 19464 }, { "epoch": 0.19465, "grad_norm": 1.5895673104733836, "learning_rate": 0.003, "loss": 4.0641, "step": 19465 }, { "epoch": 0.19466, "grad_norm": 1.3282123036176765, "learning_rate": 0.003, "loss": 4.0743, "step": 19466 }, { "epoch": 0.19467, "grad_norm": 1.3285557790501594, "learning_rate": 0.003, "loss": 4.0734, "step": 19467 }, { "epoch": 0.19468, "grad_norm": 1.6404474223744476, "learning_rate": 0.003, "loss": 4.0486, "step": 19468 }, { "epoch": 0.19469, "grad_norm": 1.0272368445693343, "learning_rate": 0.003, "loss": 4.0512, "step": 19469 }, { "epoch": 0.1947, "grad_norm": 1.3470462955974183, "learning_rate": 0.003, "loss": 4.0658, "step": 19470 }, { "epoch": 0.19471, "grad_norm": 1.1352798820585221, "learning_rate": 0.003, "loss": 4.0659, "step": 19471 }, { "epoch": 0.19472, "grad_norm": 1.2741577834484807, "learning_rate": 0.003, "loss": 4.0665, "step": 19472 }, { "epoch": 0.19473, "grad_norm": 1.0338226235999233, "learning_rate": 0.003, "loss": 4.0535, "step": 19473 }, { "epoch": 0.19474, "grad_norm": 1.4778316675976706, "learning_rate": 0.003, "loss": 4.0733, "step": 19474 }, { "epoch": 0.19475, "grad_norm": 1.001802548083004, "learning_rate": 0.003, "loss": 4.0563, "step": 19475 }, { "epoch": 0.19476, "grad_norm": 1.2776564898157308, "learning_rate": 0.003, "loss": 4.0486, "step": 19476 }, { "epoch": 0.19477, "grad_norm": 0.9566690015761019, "learning_rate": 0.003, "loss": 4.0543, "step": 19477 }, { "epoch": 0.19478, "grad_norm": 1.7124043767568116, "learning_rate": 0.003, "loss": 4.0433, "step": 19478 }, { "epoch": 0.19479, "grad_norm": 1.3346500199645615, "learning_rate": 0.003, "loss": 4.0558, "step": 19479 }, { "epoch": 0.1948, "grad_norm": 1.2822789974464504, "learning_rate": 0.003, "loss": 4.064, "step": 19480 }, { "epoch": 0.19481, "grad_norm": 1.2003608141339481, "learning_rate": 0.003, "loss": 4.0748, "step": 19481 }, { "epoch": 0.19482, "grad_norm": 1.275137873917073, "learning_rate": 0.003, "loss": 4.0709, "step": 19482 }, { "epoch": 0.19483, "grad_norm": 1.2061150927595439, "learning_rate": 0.003, "loss": 4.0429, "step": 19483 }, { "epoch": 0.19484, "grad_norm": 1.3083050736246036, "learning_rate": 0.003, "loss": 4.0676, "step": 19484 }, { "epoch": 0.19485, "grad_norm": 1.4213799670596485, "learning_rate": 0.003, "loss": 4.0733, "step": 19485 }, { "epoch": 0.19486, "grad_norm": 1.1501040588090954, "learning_rate": 0.003, "loss": 4.065, "step": 19486 }, { "epoch": 0.19487, "grad_norm": 1.1549563528980238, "learning_rate": 0.003, "loss": 4.0708, "step": 19487 }, { "epoch": 0.19488, "grad_norm": 1.4431058402236616, "learning_rate": 0.003, "loss": 4.0422, "step": 19488 }, { "epoch": 0.19489, "grad_norm": 1.2175355473059875, "learning_rate": 0.003, "loss": 4.0562, "step": 19489 }, { "epoch": 0.1949, "grad_norm": 1.3138477593838491, "learning_rate": 0.003, "loss": 4.0684, "step": 19490 }, { "epoch": 0.19491, "grad_norm": 1.026785003255267, "learning_rate": 0.003, "loss": 4.0569, "step": 19491 }, { "epoch": 0.19492, "grad_norm": 1.443968589105436, "learning_rate": 0.003, "loss": 4.091, "step": 19492 }, { "epoch": 0.19493, "grad_norm": 1.2559761385162542, "learning_rate": 0.003, "loss": 4.058, "step": 19493 }, { "epoch": 0.19494, "grad_norm": 1.2258978094864679, "learning_rate": 0.003, "loss": 4.0681, "step": 19494 }, { "epoch": 0.19495, "grad_norm": 1.3268032918441472, "learning_rate": 0.003, "loss": 4.0366, "step": 19495 }, { "epoch": 0.19496, "grad_norm": 1.172929669839517, "learning_rate": 0.003, "loss": 4.0407, "step": 19496 }, { "epoch": 0.19497, "grad_norm": 1.8611157687150652, "learning_rate": 0.003, "loss": 4.0703, "step": 19497 }, { "epoch": 0.19498, "grad_norm": 0.8175248819424565, "learning_rate": 0.003, "loss": 4.0326, "step": 19498 }, { "epoch": 0.19499, "grad_norm": 1.148929410552156, "learning_rate": 0.003, "loss": 4.072, "step": 19499 }, { "epoch": 0.195, "grad_norm": 1.4298450383026386, "learning_rate": 0.003, "loss": 4.0664, "step": 19500 }, { "epoch": 0.19501, "grad_norm": 1.352448920737841, "learning_rate": 0.003, "loss": 4.0845, "step": 19501 }, { "epoch": 0.19502, "grad_norm": 1.0639703947120964, "learning_rate": 0.003, "loss": 4.0698, "step": 19502 }, { "epoch": 0.19503, "grad_norm": 1.3500421848525783, "learning_rate": 0.003, "loss": 4.0368, "step": 19503 }, { "epoch": 0.19504, "grad_norm": 1.2014125271111171, "learning_rate": 0.003, "loss": 4.0535, "step": 19504 }, { "epoch": 0.19505, "grad_norm": 1.116248970796125, "learning_rate": 0.003, "loss": 4.0606, "step": 19505 }, { "epoch": 0.19506, "grad_norm": 1.3352551390859568, "learning_rate": 0.003, "loss": 4.0582, "step": 19506 }, { "epoch": 0.19507, "grad_norm": 1.2268556128174655, "learning_rate": 0.003, "loss": 4.0592, "step": 19507 }, { "epoch": 0.19508, "grad_norm": 1.52270928368884, "learning_rate": 0.003, "loss": 4.0465, "step": 19508 }, { "epoch": 0.19509, "grad_norm": 0.9903829036789055, "learning_rate": 0.003, "loss": 4.0675, "step": 19509 }, { "epoch": 0.1951, "grad_norm": 1.6486221557732899, "learning_rate": 0.003, "loss": 4.0618, "step": 19510 }, { "epoch": 0.19511, "grad_norm": 1.1343000920670525, "learning_rate": 0.003, "loss": 4.0387, "step": 19511 }, { "epoch": 0.19512, "grad_norm": 1.3747103241346732, "learning_rate": 0.003, "loss": 4.0728, "step": 19512 }, { "epoch": 0.19513, "grad_norm": 1.3077240373564807, "learning_rate": 0.003, "loss": 4.0773, "step": 19513 }, { "epoch": 0.19514, "grad_norm": 1.3304880069667193, "learning_rate": 0.003, "loss": 4.0899, "step": 19514 }, { "epoch": 0.19515, "grad_norm": 1.3365639153795414, "learning_rate": 0.003, "loss": 4.0534, "step": 19515 }, { "epoch": 0.19516, "grad_norm": 1.2805735532922602, "learning_rate": 0.003, "loss": 4.0476, "step": 19516 }, { "epoch": 0.19517, "grad_norm": 1.176542850338708, "learning_rate": 0.003, "loss": 4.0407, "step": 19517 }, { "epoch": 0.19518, "grad_norm": 1.373636340662171, "learning_rate": 0.003, "loss": 4.0602, "step": 19518 }, { "epoch": 0.19519, "grad_norm": 1.2370671038084415, "learning_rate": 0.003, "loss": 4.0552, "step": 19519 }, { "epoch": 0.1952, "grad_norm": 1.1049544219671612, "learning_rate": 0.003, "loss": 4.0697, "step": 19520 }, { "epoch": 0.19521, "grad_norm": 1.3532990901663196, "learning_rate": 0.003, "loss": 4.0369, "step": 19521 }, { "epoch": 0.19522, "grad_norm": 1.3889945459144393, "learning_rate": 0.003, "loss": 4.0809, "step": 19522 }, { "epoch": 0.19523, "grad_norm": 1.1596723900487986, "learning_rate": 0.003, "loss": 4.0574, "step": 19523 }, { "epoch": 0.19524, "grad_norm": 1.1650541156874012, "learning_rate": 0.003, "loss": 4.0672, "step": 19524 }, { "epoch": 0.19525, "grad_norm": 1.257471344778045, "learning_rate": 0.003, "loss": 4.0301, "step": 19525 }, { "epoch": 0.19526, "grad_norm": 1.2079351500457816, "learning_rate": 0.003, "loss": 4.0395, "step": 19526 }, { "epoch": 0.19527, "grad_norm": 1.2256036351388304, "learning_rate": 0.003, "loss": 4.0549, "step": 19527 }, { "epoch": 0.19528, "grad_norm": 1.246772224839127, "learning_rate": 0.003, "loss": 4.0516, "step": 19528 }, { "epoch": 0.19529, "grad_norm": 1.2693597734212752, "learning_rate": 0.003, "loss": 4.0386, "step": 19529 }, { "epoch": 0.1953, "grad_norm": 0.9784867447087106, "learning_rate": 0.003, "loss": 4.0426, "step": 19530 }, { "epoch": 0.19531, "grad_norm": 1.1659601986877428, "learning_rate": 0.003, "loss": 4.0532, "step": 19531 }, { "epoch": 0.19532, "grad_norm": 1.3708558472221481, "learning_rate": 0.003, "loss": 4.0873, "step": 19532 }, { "epoch": 0.19533, "grad_norm": 1.2520196238389958, "learning_rate": 0.003, "loss": 4.0654, "step": 19533 }, { "epoch": 0.19534, "grad_norm": 1.1855313119014532, "learning_rate": 0.003, "loss": 4.0798, "step": 19534 }, { "epoch": 0.19535, "grad_norm": 1.3884757533251968, "learning_rate": 0.003, "loss": 4.0849, "step": 19535 }, { "epoch": 0.19536, "grad_norm": 1.1952833344636515, "learning_rate": 0.003, "loss": 4.0207, "step": 19536 }, { "epoch": 0.19537, "grad_norm": 1.2574406153793336, "learning_rate": 0.003, "loss": 4.0347, "step": 19537 }, { "epoch": 0.19538, "grad_norm": 1.1060869917154688, "learning_rate": 0.003, "loss": 4.0643, "step": 19538 }, { "epoch": 0.19539, "grad_norm": 1.4285291372266062, "learning_rate": 0.003, "loss": 4.0623, "step": 19539 }, { "epoch": 0.1954, "grad_norm": 1.4360456654895173, "learning_rate": 0.003, "loss": 4.0496, "step": 19540 }, { "epoch": 0.19541, "grad_norm": 1.518265568289758, "learning_rate": 0.003, "loss": 4.0625, "step": 19541 }, { "epoch": 0.19542, "grad_norm": 1.2775972527094612, "learning_rate": 0.003, "loss": 4.0761, "step": 19542 }, { "epoch": 0.19543, "grad_norm": 1.218283698591187, "learning_rate": 0.003, "loss": 4.0778, "step": 19543 }, { "epoch": 0.19544, "grad_norm": 1.3374266934193224, "learning_rate": 0.003, "loss": 4.0549, "step": 19544 }, { "epoch": 0.19545, "grad_norm": 1.2582913334864856, "learning_rate": 0.003, "loss": 4.0567, "step": 19545 }, { "epoch": 0.19546, "grad_norm": 1.4337878676516518, "learning_rate": 0.003, "loss": 4.0556, "step": 19546 }, { "epoch": 0.19547, "grad_norm": 1.1073651769316895, "learning_rate": 0.003, "loss": 4.0529, "step": 19547 }, { "epoch": 0.19548, "grad_norm": 1.410688848816578, "learning_rate": 0.003, "loss": 4.0619, "step": 19548 }, { "epoch": 0.19549, "grad_norm": 1.1377670523166983, "learning_rate": 0.003, "loss": 4.0441, "step": 19549 }, { "epoch": 0.1955, "grad_norm": 1.2137808592369639, "learning_rate": 0.003, "loss": 4.0236, "step": 19550 }, { "epoch": 0.19551, "grad_norm": 1.3996923080260375, "learning_rate": 0.003, "loss": 4.0815, "step": 19551 }, { "epoch": 0.19552, "grad_norm": 1.4013579938780023, "learning_rate": 0.003, "loss": 4.0371, "step": 19552 }, { "epoch": 0.19553, "grad_norm": 1.1686342591779189, "learning_rate": 0.003, "loss": 4.0609, "step": 19553 }, { "epoch": 0.19554, "grad_norm": 1.3472445469317165, "learning_rate": 0.003, "loss": 4.0173, "step": 19554 }, { "epoch": 0.19555, "grad_norm": 1.2043089863323246, "learning_rate": 0.003, "loss": 4.0484, "step": 19555 }, { "epoch": 0.19556, "grad_norm": 1.0969614136516466, "learning_rate": 0.003, "loss": 4.0239, "step": 19556 }, { "epoch": 0.19557, "grad_norm": 1.5492311727960493, "learning_rate": 0.003, "loss": 4.0645, "step": 19557 }, { "epoch": 0.19558, "grad_norm": 0.9987137999123095, "learning_rate": 0.003, "loss": 4.0346, "step": 19558 }, { "epoch": 0.19559, "grad_norm": 1.4186798103122438, "learning_rate": 0.003, "loss": 4.0498, "step": 19559 }, { "epoch": 0.1956, "grad_norm": 1.2227299263360691, "learning_rate": 0.003, "loss": 4.0814, "step": 19560 }, { "epoch": 0.19561, "grad_norm": 1.2878151342007549, "learning_rate": 0.003, "loss": 4.0577, "step": 19561 }, { "epoch": 0.19562, "grad_norm": 1.2513526923807567, "learning_rate": 0.003, "loss": 4.0276, "step": 19562 }, { "epoch": 0.19563, "grad_norm": 1.1845288870786257, "learning_rate": 0.003, "loss": 4.0367, "step": 19563 }, { "epoch": 0.19564, "grad_norm": 1.2968329398378675, "learning_rate": 0.003, "loss": 4.0619, "step": 19564 }, { "epoch": 0.19565, "grad_norm": 1.1556297886996172, "learning_rate": 0.003, "loss": 4.0227, "step": 19565 }, { "epoch": 0.19566, "grad_norm": 1.5926717140053277, "learning_rate": 0.003, "loss": 4.0634, "step": 19566 }, { "epoch": 0.19567, "grad_norm": 1.0154673277724815, "learning_rate": 0.003, "loss": 4.0415, "step": 19567 }, { "epoch": 0.19568, "grad_norm": 1.5961582019473584, "learning_rate": 0.003, "loss": 4.0574, "step": 19568 }, { "epoch": 0.19569, "grad_norm": 1.0970845859088805, "learning_rate": 0.003, "loss": 4.0496, "step": 19569 }, { "epoch": 0.1957, "grad_norm": 1.382216240127931, "learning_rate": 0.003, "loss": 4.063, "step": 19570 }, { "epoch": 0.19571, "grad_norm": 1.2870224499187535, "learning_rate": 0.003, "loss": 4.063, "step": 19571 }, { "epoch": 0.19572, "grad_norm": 1.2231550655720813, "learning_rate": 0.003, "loss": 4.0493, "step": 19572 }, { "epoch": 0.19573, "grad_norm": 1.0646521546788703, "learning_rate": 0.003, "loss": 4.0382, "step": 19573 }, { "epoch": 0.19574, "grad_norm": 1.3969937544755526, "learning_rate": 0.003, "loss": 4.0543, "step": 19574 }, { "epoch": 0.19575, "grad_norm": 1.379948980968305, "learning_rate": 0.003, "loss": 4.0836, "step": 19575 }, { "epoch": 0.19576, "grad_norm": 1.0433605833792832, "learning_rate": 0.003, "loss": 4.0341, "step": 19576 }, { "epoch": 0.19577, "grad_norm": 1.5545742346773992, "learning_rate": 0.003, "loss": 4.074, "step": 19577 }, { "epoch": 0.19578, "grad_norm": 1.1495336764895239, "learning_rate": 0.003, "loss": 4.0473, "step": 19578 }, { "epoch": 0.19579, "grad_norm": 1.2713073092805014, "learning_rate": 0.003, "loss": 4.0529, "step": 19579 }, { "epoch": 0.1958, "grad_norm": 1.2475666394907055, "learning_rate": 0.003, "loss": 4.0672, "step": 19580 }, { "epoch": 0.19581, "grad_norm": 1.081462776113956, "learning_rate": 0.003, "loss": 4.0055, "step": 19581 }, { "epoch": 0.19582, "grad_norm": 1.642268700127265, "learning_rate": 0.003, "loss": 4.0785, "step": 19582 }, { "epoch": 0.19583, "grad_norm": 1.03323585563909, "learning_rate": 0.003, "loss": 4.0601, "step": 19583 }, { "epoch": 0.19584, "grad_norm": 1.55716996864311, "learning_rate": 0.003, "loss": 4.0613, "step": 19584 }, { "epoch": 0.19585, "grad_norm": 0.993583704461042, "learning_rate": 0.003, "loss": 4.0518, "step": 19585 }, { "epoch": 0.19586, "grad_norm": 1.2703165301059727, "learning_rate": 0.003, "loss": 4.0499, "step": 19586 }, { "epoch": 0.19587, "grad_norm": 1.3644708922683342, "learning_rate": 0.003, "loss": 4.0837, "step": 19587 }, { "epoch": 0.19588, "grad_norm": 1.3114723319710333, "learning_rate": 0.003, "loss": 4.0431, "step": 19588 }, { "epoch": 0.19589, "grad_norm": 1.2118999445324037, "learning_rate": 0.003, "loss": 4.0677, "step": 19589 }, { "epoch": 0.1959, "grad_norm": 1.2095389613176255, "learning_rate": 0.003, "loss": 4.0805, "step": 19590 }, { "epoch": 0.19591, "grad_norm": 1.286135984669697, "learning_rate": 0.003, "loss": 4.0783, "step": 19591 }, { "epoch": 0.19592, "grad_norm": 1.2367742585855162, "learning_rate": 0.003, "loss": 4.0722, "step": 19592 }, { "epoch": 0.19593, "grad_norm": 1.1296819706661674, "learning_rate": 0.003, "loss": 4.0596, "step": 19593 }, { "epoch": 0.19594, "grad_norm": 1.328571573474519, "learning_rate": 0.003, "loss": 4.076, "step": 19594 }, { "epoch": 0.19595, "grad_norm": 1.1078568494276055, "learning_rate": 0.003, "loss": 4.0436, "step": 19595 }, { "epoch": 0.19596, "grad_norm": 1.35862584683534, "learning_rate": 0.003, "loss": 4.0485, "step": 19596 }, { "epoch": 0.19597, "grad_norm": 1.2569322781269223, "learning_rate": 0.003, "loss": 4.0791, "step": 19597 }, { "epoch": 0.19598, "grad_norm": 1.468685773176229, "learning_rate": 0.003, "loss": 4.0399, "step": 19598 }, { "epoch": 0.19599, "grad_norm": 1.2630812981750894, "learning_rate": 0.003, "loss": 4.0309, "step": 19599 }, { "epoch": 0.196, "grad_norm": 1.2238107822625526, "learning_rate": 0.003, "loss": 4.0259, "step": 19600 }, { "epoch": 0.19601, "grad_norm": 1.2472312210806207, "learning_rate": 0.003, "loss": 4.0341, "step": 19601 }, { "epoch": 0.19602, "grad_norm": 1.3741733077287006, "learning_rate": 0.003, "loss": 4.0276, "step": 19602 }, { "epoch": 0.19603, "grad_norm": 1.1794829060341756, "learning_rate": 0.003, "loss": 4.0579, "step": 19603 }, { "epoch": 0.19604, "grad_norm": 1.3311347113042369, "learning_rate": 0.003, "loss": 4.049, "step": 19604 }, { "epoch": 0.19605, "grad_norm": 1.0357739621590778, "learning_rate": 0.003, "loss": 4.0126, "step": 19605 }, { "epoch": 0.19606, "grad_norm": 1.5275152034988269, "learning_rate": 0.003, "loss": 4.0445, "step": 19606 }, { "epoch": 0.19607, "grad_norm": 1.076429551433992, "learning_rate": 0.003, "loss": 4.0231, "step": 19607 }, { "epoch": 0.19608, "grad_norm": 1.3647193929111205, "learning_rate": 0.003, "loss": 4.0438, "step": 19608 }, { "epoch": 0.19609, "grad_norm": 1.4867220985587168, "learning_rate": 0.003, "loss": 4.0662, "step": 19609 }, { "epoch": 0.1961, "grad_norm": 1.125587708642376, "learning_rate": 0.003, "loss": 4.0562, "step": 19610 }, { "epoch": 0.19611, "grad_norm": 1.3631568088824906, "learning_rate": 0.003, "loss": 4.0598, "step": 19611 }, { "epoch": 0.19612, "grad_norm": 1.2624153115040977, "learning_rate": 0.003, "loss": 4.0698, "step": 19612 }, { "epoch": 0.19613, "grad_norm": 1.2075066353474955, "learning_rate": 0.003, "loss": 4.0678, "step": 19613 }, { "epoch": 0.19614, "grad_norm": 1.3088675832743637, "learning_rate": 0.003, "loss": 4.0234, "step": 19614 }, { "epoch": 0.19615, "grad_norm": 1.121251813389829, "learning_rate": 0.003, "loss": 4.0445, "step": 19615 }, { "epoch": 0.19616, "grad_norm": 1.1065842533281147, "learning_rate": 0.003, "loss": 4.0448, "step": 19616 }, { "epoch": 0.19617, "grad_norm": 1.1859974265552933, "learning_rate": 0.003, "loss": 4.0852, "step": 19617 }, { "epoch": 0.19618, "grad_norm": 1.4820292716416597, "learning_rate": 0.003, "loss": 4.0468, "step": 19618 }, { "epoch": 0.19619, "grad_norm": 1.0404900909759272, "learning_rate": 0.003, "loss": 4.0582, "step": 19619 }, { "epoch": 0.1962, "grad_norm": 1.5799678557313894, "learning_rate": 0.003, "loss": 4.0597, "step": 19620 }, { "epoch": 0.19621, "grad_norm": 1.626377600840289, "learning_rate": 0.003, "loss": 4.0463, "step": 19621 }, { "epoch": 0.19622, "grad_norm": 1.2952931979556077, "learning_rate": 0.003, "loss": 4.0589, "step": 19622 }, { "epoch": 0.19623, "grad_norm": 1.2740424205319891, "learning_rate": 0.003, "loss": 4.0455, "step": 19623 }, { "epoch": 0.19624, "grad_norm": 1.4104946822740858, "learning_rate": 0.003, "loss": 4.061, "step": 19624 }, { "epoch": 0.19625, "grad_norm": 1.1955443930403675, "learning_rate": 0.003, "loss": 4.0526, "step": 19625 }, { "epoch": 0.19626, "grad_norm": 1.3927373880231353, "learning_rate": 0.003, "loss": 4.069, "step": 19626 }, { "epoch": 0.19627, "grad_norm": 1.0953666384205296, "learning_rate": 0.003, "loss": 4.0411, "step": 19627 }, { "epoch": 0.19628, "grad_norm": 1.5311802781921875, "learning_rate": 0.003, "loss": 4.0413, "step": 19628 }, { "epoch": 0.19629, "grad_norm": 1.1746433054095549, "learning_rate": 0.003, "loss": 4.0436, "step": 19629 }, { "epoch": 0.1963, "grad_norm": 1.2335569066702672, "learning_rate": 0.003, "loss": 4.0208, "step": 19630 }, { "epoch": 0.19631, "grad_norm": 1.269593398527897, "learning_rate": 0.003, "loss": 4.0604, "step": 19631 }, { "epoch": 0.19632, "grad_norm": 1.514708223805417, "learning_rate": 0.003, "loss": 4.0759, "step": 19632 }, { "epoch": 0.19633, "grad_norm": 1.252573199409988, "learning_rate": 0.003, "loss": 4.0354, "step": 19633 }, { "epoch": 0.19634, "grad_norm": 1.1973954065559616, "learning_rate": 0.003, "loss": 4.075, "step": 19634 }, { "epoch": 0.19635, "grad_norm": 1.493911201291848, "learning_rate": 0.003, "loss": 4.0603, "step": 19635 }, { "epoch": 0.19636, "grad_norm": 1.088139048031911, "learning_rate": 0.003, "loss": 4.0699, "step": 19636 }, { "epoch": 0.19637, "grad_norm": 1.2257101321716657, "learning_rate": 0.003, "loss": 4.0299, "step": 19637 }, { "epoch": 0.19638, "grad_norm": 1.246686123057356, "learning_rate": 0.003, "loss": 4.0014, "step": 19638 }, { "epoch": 0.19639, "grad_norm": 1.4004374784771214, "learning_rate": 0.003, "loss": 4.0926, "step": 19639 }, { "epoch": 0.1964, "grad_norm": 1.2275817958209077, "learning_rate": 0.003, "loss": 4.0336, "step": 19640 }, { "epoch": 0.19641, "grad_norm": 1.2209983282623578, "learning_rate": 0.003, "loss": 4.0707, "step": 19641 }, { "epoch": 0.19642, "grad_norm": 1.4039034259534702, "learning_rate": 0.003, "loss": 4.0431, "step": 19642 }, { "epoch": 0.19643, "grad_norm": 1.1166387643455105, "learning_rate": 0.003, "loss": 4.0653, "step": 19643 }, { "epoch": 0.19644, "grad_norm": 1.1532866416283507, "learning_rate": 0.003, "loss": 4.0693, "step": 19644 }, { "epoch": 0.19645, "grad_norm": 1.1511586224034764, "learning_rate": 0.003, "loss": 4.0608, "step": 19645 }, { "epoch": 0.19646, "grad_norm": 1.2214661318176832, "learning_rate": 0.003, "loss": 4.0726, "step": 19646 }, { "epoch": 0.19647, "grad_norm": 1.1854282536885827, "learning_rate": 0.003, "loss": 4.0467, "step": 19647 }, { "epoch": 0.19648, "grad_norm": 1.5656721018163773, "learning_rate": 0.003, "loss": 4.0693, "step": 19648 }, { "epoch": 0.19649, "grad_norm": 0.9549202035089439, "learning_rate": 0.003, "loss": 4.0292, "step": 19649 }, { "epoch": 0.1965, "grad_norm": 1.3043409972453064, "learning_rate": 0.003, "loss": 4.0585, "step": 19650 }, { "epoch": 0.19651, "grad_norm": 1.338903895964893, "learning_rate": 0.003, "loss": 4.0512, "step": 19651 }, { "epoch": 0.19652, "grad_norm": 1.1193027749921456, "learning_rate": 0.003, "loss": 4.0641, "step": 19652 }, { "epoch": 0.19653, "grad_norm": 1.294277668693427, "learning_rate": 0.003, "loss": 4.0311, "step": 19653 }, { "epoch": 0.19654, "grad_norm": 1.3469191435912236, "learning_rate": 0.003, "loss": 4.0532, "step": 19654 }, { "epoch": 0.19655, "grad_norm": 1.2635214490978228, "learning_rate": 0.003, "loss": 4.0466, "step": 19655 }, { "epoch": 0.19656, "grad_norm": 1.039754857917406, "learning_rate": 0.003, "loss": 4.0344, "step": 19656 }, { "epoch": 0.19657, "grad_norm": 1.425960245599324, "learning_rate": 0.003, "loss": 4.0449, "step": 19657 }, { "epoch": 0.19658, "grad_norm": 1.1131270129876956, "learning_rate": 0.003, "loss": 4.0821, "step": 19658 }, { "epoch": 0.19659, "grad_norm": 1.4233979198300244, "learning_rate": 0.003, "loss": 4.0731, "step": 19659 }, { "epoch": 0.1966, "grad_norm": 1.1268682959377379, "learning_rate": 0.003, "loss": 4.0373, "step": 19660 }, { "epoch": 0.19661, "grad_norm": 1.2983658300625234, "learning_rate": 0.003, "loss": 4.0458, "step": 19661 }, { "epoch": 0.19662, "grad_norm": 1.3793633866642898, "learning_rate": 0.003, "loss": 4.0781, "step": 19662 }, { "epoch": 0.19663, "grad_norm": 1.1681532953099747, "learning_rate": 0.003, "loss": 4.06, "step": 19663 }, { "epoch": 0.19664, "grad_norm": 1.449327499358008, "learning_rate": 0.003, "loss": 4.0463, "step": 19664 }, { "epoch": 0.19665, "grad_norm": 1.1594026577520937, "learning_rate": 0.003, "loss": 4.0379, "step": 19665 }, { "epoch": 0.19666, "grad_norm": 1.304588843645449, "learning_rate": 0.003, "loss": 4.0531, "step": 19666 }, { "epoch": 0.19667, "grad_norm": 1.2855175972244166, "learning_rate": 0.003, "loss": 4.0686, "step": 19667 }, { "epoch": 0.19668, "grad_norm": 1.2108562775295844, "learning_rate": 0.003, "loss": 4.0341, "step": 19668 }, { "epoch": 0.19669, "grad_norm": 1.4881363232865699, "learning_rate": 0.003, "loss": 4.0317, "step": 19669 }, { "epoch": 0.1967, "grad_norm": 1.1050320650830274, "learning_rate": 0.003, "loss": 4.0521, "step": 19670 }, { "epoch": 0.19671, "grad_norm": 1.4028067934145172, "learning_rate": 0.003, "loss": 4.0436, "step": 19671 }, { "epoch": 0.19672, "grad_norm": 1.212968583047561, "learning_rate": 0.003, "loss": 4.0253, "step": 19672 }, { "epoch": 0.19673, "grad_norm": 1.4005758245894164, "learning_rate": 0.003, "loss": 4.03, "step": 19673 }, { "epoch": 0.19674, "grad_norm": 1.1259823388693067, "learning_rate": 0.003, "loss": 4.0255, "step": 19674 }, { "epoch": 0.19675, "grad_norm": 1.4615329880964987, "learning_rate": 0.003, "loss": 4.0552, "step": 19675 }, { "epoch": 0.19676, "grad_norm": 1.0969160536054376, "learning_rate": 0.003, "loss": 4.0546, "step": 19676 }, { "epoch": 0.19677, "grad_norm": 1.4081030226763, "learning_rate": 0.003, "loss": 4.0484, "step": 19677 }, { "epoch": 0.19678, "grad_norm": 1.2211659347447854, "learning_rate": 0.003, "loss": 4.0614, "step": 19678 }, { "epoch": 0.19679, "grad_norm": 1.2150982107622408, "learning_rate": 0.003, "loss": 4.0335, "step": 19679 }, { "epoch": 0.1968, "grad_norm": 1.3535734950333507, "learning_rate": 0.003, "loss": 4.0393, "step": 19680 }, { "epoch": 0.19681, "grad_norm": 1.0663404840851425, "learning_rate": 0.003, "loss": 4.0805, "step": 19681 }, { "epoch": 0.19682, "grad_norm": 1.6754522163616776, "learning_rate": 0.003, "loss": 4.0539, "step": 19682 }, { "epoch": 0.19683, "grad_norm": 1.0209211426190017, "learning_rate": 0.003, "loss": 4.0498, "step": 19683 }, { "epoch": 0.19684, "grad_norm": 1.4384155112003565, "learning_rate": 0.003, "loss": 4.0518, "step": 19684 }, { "epoch": 0.19685, "grad_norm": 1.183663031201126, "learning_rate": 0.003, "loss": 4.0124, "step": 19685 }, { "epoch": 0.19686, "grad_norm": 1.204767719156397, "learning_rate": 0.003, "loss": 4.054, "step": 19686 }, { "epoch": 0.19687, "grad_norm": 1.43264678315374, "learning_rate": 0.003, "loss": 4.051, "step": 19687 }, { "epoch": 0.19688, "grad_norm": 1.1444222737791538, "learning_rate": 0.003, "loss": 4.0738, "step": 19688 }, { "epoch": 0.19689, "grad_norm": 1.4164282948887503, "learning_rate": 0.003, "loss": 4.0608, "step": 19689 }, { "epoch": 0.1969, "grad_norm": 1.519323637865488, "learning_rate": 0.003, "loss": 4.0838, "step": 19690 }, { "epoch": 0.19691, "grad_norm": 0.9758922904826194, "learning_rate": 0.003, "loss": 4.0314, "step": 19691 }, { "epoch": 0.19692, "grad_norm": 1.2979414816054384, "learning_rate": 0.003, "loss": 4.0585, "step": 19692 }, { "epoch": 0.19693, "grad_norm": 1.2027431932617603, "learning_rate": 0.003, "loss": 4.0468, "step": 19693 }, { "epoch": 0.19694, "grad_norm": 1.483541597195135, "learning_rate": 0.003, "loss": 4.044, "step": 19694 }, { "epoch": 0.19695, "grad_norm": 1.2092502940393162, "learning_rate": 0.003, "loss": 4.0697, "step": 19695 }, { "epoch": 0.19696, "grad_norm": 1.4774850821717136, "learning_rate": 0.003, "loss": 4.0723, "step": 19696 }, { "epoch": 0.19697, "grad_norm": 1.326064854971596, "learning_rate": 0.003, "loss": 4.0486, "step": 19697 }, { "epoch": 0.19698, "grad_norm": 1.0462582412112698, "learning_rate": 0.003, "loss": 4.08, "step": 19698 }, { "epoch": 0.19699, "grad_norm": 1.2129076652215423, "learning_rate": 0.003, "loss": 4.0266, "step": 19699 }, { "epoch": 0.197, "grad_norm": 0.9404415450425605, "learning_rate": 0.003, "loss": 4.053, "step": 19700 }, { "epoch": 0.19701, "grad_norm": 1.408164938206551, "learning_rate": 0.003, "loss": 4.0832, "step": 19701 }, { "epoch": 0.19702, "grad_norm": 1.140221276354727, "learning_rate": 0.003, "loss": 4.0668, "step": 19702 }, { "epoch": 0.19703, "grad_norm": 1.632658533157582, "learning_rate": 0.003, "loss": 4.0815, "step": 19703 }, { "epoch": 0.19704, "grad_norm": 1.2428620826128585, "learning_rate": 0.003, "loss": 4.032, "step": 19704 }, { "epoch": 0.19705, "grad_norm": 1.2176041406759814, "learning_rate": 0.003, "loss": 4.0333, "step": 19705 }, { "epoch": 0.19706, "grad_norm": 1.1040196987269462, "learning_rate": 0.003, "loss": 4.0153, "step": 19706 }, { "epoch": 0.19707, "grad_norm": 1.3294950566418005, "learning_rate": 0.003, "loss": 4.061, "step": 19707 }, { "epoch": 0.19708, "grad_norm": 1.2000338311222096, "learning_rate": 0.003, "loss": 4.0641, "step": 19708 }, { "epoch": 0.19709, "grad_norm": 1.3011272832919174, "learning_rate": 0.003, "loss": 4.0854, "step": 19709 }, { "epoch": 0.1971, "grad_norm": 1.0625031410202341, "learning_rate": 0.003, "loss": 4.0624, "step": 19710 }, { "epoch": 0.19711, "grad_norm": 1.3300242910674192, "learning_rate": 0.003, "loss": 4.0704, "step": 19711 }, { "epoch": 0.19712, "grad_norm": 1.0724203569381963, "learning_rate": 0.003, "loss": 4.0599, "step": 19712 }, { "epoch": 0.19713, "grad_norm": 1.4760362987564652, "learning_rate": 0.003, "loss": 4.0582, "step": 19713 }, { "epoch": 0.19714, "grad_norm": 1.2615842808043052, "learning_rate": 0.003, "loss": 4.0203, "step": 19714 }, { "epoch": 0.19715, "grad_norm": 1.4767863603989584, "learning_rate": 0.003, "loss": 4.0569, "step": 19715 }, { "epoch": 0.19716, "grad_norm": 1.3187426984567565, "learning_rate": 0.003, "loss": 4.0358, "step": 19716 }, { "epoch": 0.19717, "grad_norm": 1.1333722085269455, "learning_rate": 0.003, "loss": 4.0774, "step": 19717 }, { "epoch": 0.19718, "grad_norm": 1.2921747578348115, "learning_rate": 0.003, "loss": 4.0508, "step": 19718 }, { "epoch": 0.19719, "grad_norm": 1.1755894496904107, "learning_rate": 0.003, "loss": 4.0051, "step": 19719 }, { "epoch": 0.1972, "grad_norm": 1.342887844017686, "learning_rate": 0.003, "loss": 4.0177, "step": 19720 }, { "epoch": 0.19721, "grad_norm": 1.1299536859301806, "learning_rate": 0.003, "loss": 4.0375, "step": 19721 }, { "epoch": 0.19722, "grad_norm": 1.2841992171190308, "learning_rate": 0.003, "loss": 4.0613, "step": 19722 }, { "epoch": 0.19723, "grad_norm": 1.2866555374877549, "learning_rate": 0.003, "loss": 4.0636, "step": 19723 }, { "epoch": 0.19724, "grad_norm": 1.3985165773499388, "learning_rate": 0.003, "loss": 4.0455, "step": 19724 }, { "epoch": 0.19725, "grad_norm": 1.0846348006817275, "learning_rate": 0.003, "loss": 4.0415, "step": 19725 }, { "epoch": 0.19726, "grad_norm": 1.3040928333214943, "learning_rate": 0.003, "loss": 4.0573, "step": 19726 }, { "epoch": 0.19727, "grad_norm": 1.2191004795301124, "learning_rate": 0.003, "loss": 4.0452, "step": 19727 }, { "epoch": 0.19728, "grad_norm": 1.240838042793232, "learning_rate": 0.003, "loss": 4.0626, "step": 19728 }, { "epoch": 0.19729, "grad_norm": 1.7369481182987327, "learning_rate": 0.003, "loss": 4.0568, "step": 19729 }, { "epoch": 0.1973, "grad_norm": 1.012198468014683, "learning_rate": 0.003, "loss": 4.0556, "step": 19730 }, { "epoch": 0.19731, "grad_norm": 1.4397242497122429, "learning_rate": 0.003, "loss": 4.0801, "step": 19731 }, { "epoch": 0.19732, "grad_norm": 1.4325529884833712, "learning_rate": 0.003, "loss": 4.0622, "step": 19732 }, { "epoch": 0.19733, "grad_norm": 1.072002107789731, "learning_rate": 0.003, "loss": 4.0483, "step": 19733 }, { "epoch": 0.19734, "grad_norm": 1.4621090283171523, "learning_rate": 0.003, "loss": 4.0487, "step": 19734 }, { "epoch": 0.19735, "grad_norm": 1.216349933331565, "learning_rate": 0.003, "loss": 4.0676, "step": 19735 }, { "epoch": 0.19736, "grad_norm": 1.4335498129154791, "learning_rate": 0.003, "loss": 4.0523, "step": 19736 }, { "epoch": 0.19737, "grad_norm": 1.162866042862082, "learning_rate": 0.003, "loss": 4.0322, "step": 19737 }, { "epoch": 0.19738, "grad_norm": 1.1751061853343154, "learning_rate": 0.003, "loss": 4.0431, "step": 19738 }, { "epoch": 0.19739, "grad_norm": 1.2320589920594738, "learning_rate": 0.003, "loss": 4.0619, "step": 19739 }, { "epoch": 0.1974, "grad_norm": 1.2162642898797495, "learning_rate": 0.003, "loss": 4.0341, "step": 19740 }, { "epoch": 0.19741, "grad_norm": 1.5012653945264107, "learning_rate": 0.003, "loss": 4.0591, "step": 19741 }, { "epoch": 0.19742, "grad_norm": 1.2301414165363693, "learning_rate": 0.003, "loss": 4.0487, "step": 19742 }, { "epoch": 0.19743, "grad_norm": 1.520156501009919, "learning_rate": 0.003, "loss": 4.0548, "step": 19743 }, { "epoch": 0.19744, "grad_norm": 1.0940999798891307, "learning_rate": 0.003, "loss": 4.0498, "step": 19744 }, { "epoch": 0.19745, "grad_norm": 1.213718260293644, "learning_rate": 0.003, "loss": 4.0382, "step": 19745 }, { "epoch": 0.19746, "grad_norm": 1.335698961189427, "learning_rate": 0.003, "loss": 4.0629, "step": 19746 }, { "epoch": 0.19747, "grad_norm": 1.3992921968338952, "learning_rate": 0.003, "loss": 4.0664, "step": 19747 }, { "epoch": 0.19748, "grad_norm": 1.3038221562173062, "learning_rate": 0.003, "loss": 4.0479, "step": 19748 }, { "epoch": 0.19749, "grad_norm": 1.131098670339918, "learning_rate": 0.003, "loss": 4.0536, "step": 19749 }, { "epoch": 0.1975, "grad_norm": 1.332030843151734, "learning_rate": 0.003, "loss": 4.0608, "step": 19750 }, { "epoch": 0.19751, "grad_norm": 1.092283739981877, "learning_rate": 0.003, "loss": 4.0651, "step": 19751 }, { "epoch": 0.19752, "grad_norm": 1.302600775651643, "learning_rate": 0.003, "loss": 4.0642, "step": 19752 }, { "epoch": 0.19753, "grad_norm": 1.4128837585792693, "learning_rate": 0.003, "loss": 4.0505, "step": 19753 }, { "epoch": 0.19754, "grad_norm": 1.3196083484063799, "learning_rate": 0.003, "loss": 4.0519, "step": 19754 }, { "epoch": 0.19755, "grad_norm": 1.2206351818074985, "learning_rate": 0.003, "loss": 4.077, "step": 19755 }, { "epoch": 0.19756, "grad_norm": 1.1734556955156739, "learning_rate": 0.003, "loss": 4.0166, "step": 19756 }, { "epoch": 0.19757, "grad_norm": 1.2553909696677223, "learning_rate": 0.003, "loss": 4.0386, "step": 19757 }, { "epoch": 0.19758, "grad_norm": 1.3092843491839787, "learning_rate": 0.003, "loss": 4.0928, "step": 19758 }, { "epoch": 0.19759, "grad_norm": 1.1460844748190464, "learning_rate": 0.003, "loss": 4.0444, "step": 19759 }, { "epoch": 0.1976, "grad_norm": 1.6616584576681912, "learning_rate": 0.003, "loss": 4.1023, "step": 19760 }, { "epoch": 0.19761, "grad_norm": 1.0073287619168259, "learning_rate": 0.003, "loss": 4.0514, "step": 19761 }, { "epoch": 0.19762, "grad_norm": 1.6368543289617503, "learning_rate": 0.003, "loss": 4.0694, "step": 19762 }, { "epoch": 0.19763, "grad_norm": 1.0704247305937595, "learning_rate": 0.003, "loss": 4.0435, "step": 19763 }, { "epoch": 0.19764, "grad_norm": 1.5161658380137082, "learning_rate": 0.003, "loss": 4.0629, "step": 19764 }, { "epoch": 0.19765, "grad_norm": 1.1206543468688372, "learning_rate": 0.003, "loss": 4.0421, "step": 19765 }, { "epoch": 0.19766, "grad_norm": 1.2946853858835212, "learning_rate": 0.003, "loss": 4.0888, "step": 19766 }, { "epoch": 0.19767, "grad_norm": 1.3963559045296543, "learning_rate": 0.003, "loss": 4.0522, "step": 19767 }, { "epoch": 0.19768, "grad_norm": 1.1659693910680011, "learning_rate": 0.003, "loss": 4.0685, "step": 19768 }, { "epoch": 0.19769, "grad_norm": 1.263219690264999, "learning_rate": 0.003, "loss": 4.0657, "step": 19769 }, { "epoch": 0.1977, "grad_norm": 1.1237394968167047, "learning_rate": 0.003, "loss": 4.0876, "step": 19770 }, { "epoch": 0.19771, "grad_norm": 1.4099145685647496, "learning_rate": 0.003, "loss": 4.0733, "step": 19771 }, { "epoch": 0.19772, "grad_norm": 1.1953522455301766, "learning_rate": 0.003, "loss": 4.058, "step": 19772 }, { "epoch": 0.19773, "grad_norm": 1.295100610042665, "learning_rate": 0.003, "loss": 4.0265, "step": 19773 }, { "epoch": 0.19774, "grad_norm": 1.137876500914938, "learning_rate": 0.003, "loss": 4.0613, "step": 19774 }, { "epoch": 0.19775, "grad_norm": 1.2001488009196706, "learning_rate": 0.003, "loss": 4.0381, "step": 19775 }, { "epoch": 0.19776, "grad_norm": 1.2092359549959708, "learning_rate": 0.003, "loss": 4.0359, "step": 19776 }, { "epoch": 0.19777, "grad_norm": 1.34434557789657, "learning_rate": 0.003, "loss": 4.0272, "step": 19777 }, { "epoch": 0.19778, "grad_norm": 1.2289769972887201, "learning_rate": 0.003, "loss": 4.0497, "step": 19778 }, { "epoch": 0.19779, "grad_norm": 1.3665721151529493, "learning_rate": 0.003, "loss": 4.0732, "step": 19779 }, { "epoch": 0.1978, "grad_norm": 1.200114494842136, "learning_rate": 0.003, "loss": 4.054, "step": 19780 }, { "epoch": 0.19781, "grad_norm": 1.3027910997422532, "learning_rate": 0.003, "loss": 4.0485, "step": 19781 }, { "epoch": 0.19782, "grad_norm": 1.482047104503544, "learning_rate": 0.003, "loss": 4.0432, "step": 19782 }, { "epoch": 0.19783, "grad_norm": 1.2591000303376083, "learning_rate": 0.003, "loss": 4.0574, "step": 19783 }, { "epoch": 0.19784, "grad_norm": 1.3366242503251673, "learning_rate": 0.003, "loss": 4.0641, "step": 19784 }, { "epoch": 0.19785, "grad_norm": 1.0752888957210784, "learning_rate": 0.003, "loss": 4.0582, "step": 19785 }, { "epoch": 0.19786, "grad_norm": 1.443814361354114, "learning_rate": 0.003, "loss": 4.0527, "step": 19786 }, { "epoch": 0.19787, "grad_norm": 1.1848301940578947, "learning_rate": 0.003, "loss": 4.0455, "step": 19787 }, { "epoch": 0.19788, "grad_norm": 1.2137564648369716, "learning_rate": 0.003, "loss": 4.0595, "step": 19788 }, { "epoch": 0.19789, "grad_norm": 1.434319996226596, "learning_rate": 0.003, "loss": 4.0679, "step": 19789 }, { "epoch": 0.1979, "grad_norm": 1.2083466466277641, "learning_rate": 0.003, "loss": 4.0183, "step": 19790 }, { "epoch": 0.19791, "grad_norm": 1.1864184491836716, "learning_rate": 0.003, "loss": 4.0633, "step": 19791 }, { "epoch": 0.19792, "grad_norm": 1.1504282266698773, "learning_rate": 0.003, "loss": 4.0761, "step": 19792 }, { "epoch": 0.19793, "grad_norm": 1.359723089458863, "learning_rate": 0.003, "loss": 4.0524, "step": 19793 }, { "epoch": 0.19794, "grad_norm": 1.1286923137424962, "learning_rate": 0.003, "loss": 4.0523, "step": 19794 }, { "epoch": 0.19795, "grad_norm": 1.3713746016133905, "learning_rate": 0.003, "loss": 4.0577, "step": 19795 }, { "epoch": 0.19796, "grad_norm": 1.274236183603281, "learning_rate": 0.003, "loss": 4.046, "step": 19796 }, { "epoch": 0.19797, "grad_norm": 1.4731164051604642, "learning_rate": 0.003, "loss": 4.0406, "step": 19797 }, { "epoch": 0.19798, "grad_norm": 1.1204714469424668, "learning_rate": 0.003, "loss": 4.053, "step": 19798 }, { "epoch": 0.19799, "grad_norm": 1.4618269616945994, "learning_rate": 0.003, "loss": 4.0103, "step": 19799 }, { "epoch": 0.198, "grad_norm": 1.2709446431944755, "learning_rate": 0.003, "loss": 4.0834, "step": 19800 }, { "epoch": 0.19801, "grad_norm": 1.4320673965655029, "learning_rate": 0.003, "loss": 4.0516, "step": 19801 }, { "epoch": 0.19802, "grad_norm": 1.1044847641006195, "learning_rate": 0.003, "loss": 4.075, "step": 19802 }, { "epoch": 0.19803, "grad_norm": 1.34151957210351, "learning_rate": 0.003, "loss": 4.0521, "step": 19803 }, { "epoch": 0.19804, "grad_norm": 1.3141402337987576, "learning_rate": 0.003, "loss": 4.0539, "step": 19804 }, { "epoch": 0.19805, "grad_norm": 1.2197903583977774, "learning_rate": 0.003, "loss": 4.0261, "step": 19805 }, { "epoch": 0.19806, "grad_norm": 1.3001406850014425, "learning_rate": 0.003, "loss": 4.0664, "step": 19806 }, { "epoch": 0.19807, "grad_norm": 1.276210948948388, "learning_rate": 0.003, "loss": 4.0472, "step": 19807 }, { "epoch": 0.19808, "grad_norm": 1.3200298625015325, "learning_rate": 0.003, "loss": 4.1025, "step": 19808 }, { "epoch": 0.19809, "grad_norm": 1.1835970967927212, "learning_rate": 0.003, "loss": 4.0589, "step": 19809 }, { "epoch": 0.1981, "grad_norm": 1.5677071896388945, "learning_rate": 0.003, "loss": 4.0606, "step": 19810 }, { "epoch": 0.19811, "grad_norm": 0.9995657286317865, "learning_rate": 0.003, "loss": 4.0236, "step": 19811 }, { "epoch": 0.19812, "grad_norm": 1.6820081353806127, "learning_rate": 0.003, "loss": 4.0385, "step": 19812 }, { "epoch": 0.19813, "grad_norm": 1.2837414160665404, "learning_rate": 0.003, "loss": 4.0499, "step": 19813 }, { "epoch": 0.19814, "grad_norm": 1.2874766422848056, "learning_rate": 0.003, "loss": 4.0608, "step": 19814 }, { "epoch": 0.19815, "grad_norm": 1.227080582477793, "learning_rate": 0.003, "loss": 4.0473, "step": 19815 }, { "epoch": 0.19816, "grad_norm": 1.273447687133168, "learning_rate": 0.003, "loss": 4.0469, "step": 19816 }, { "epoch": 0.19817, "grad_norm": 1.0630905492942202, "learning_rate": 0.003, "loss": 4.0453, "step": 19817 }, { "epoch": 0.19818, "grad_norm": 1.3489130559244074, "learning_rate": 0.003, "loss": 4.0385, "step": 19818 }, { "epoch": 0.19819, "grad_norm": 1.051338096722153, "learning_rate": 0.003, "loss": 4.0368, "step": 19819 }, { "epoch": 0.1982, "grad_norm": 1.490436613752286, "learning_rate": 0.003, "loss": 4.0593, "step": 19820 }, { "epoch": 0.19821, "grad_norm": 1.162533122975096, "learning_rate": 0.003, "loss": 4.0625, "step": 19821 }, { "epoch": 0.19822, "grad_norm": 1.2323568924771044, "learning_rate": 0.003, "loss": 4.0325, "step": 19822 }, { "epoch": 0.19823, "grad_norm": 1.4005699745657514, "learning_rate": 0.003, "loss": 4.0236, "step": 19823 }, { "epoch": 0.19824, "grad_norm": 1.2512895475599117, "learning_rate": 0.003, "loss": 4.0678, "step": 19824 }, { "epoch": 0.19825, "grad_norm": 1.4769621946202904, "learning_rate": 0.003, "loss": 4.0441, "step": 19825 }, { "epoch": 0.19826, "grad_norm": 0.9742481471592784, "learning_rate": 0.003, "loss": 4.047, "step": 19826 }, { "epoch": 0.19827, "grad_norm": 1.441936380925012, "learning_rate": 0.003, "loss": 4.0323, "step": 19827 }, { "epoch": 0.19828, "grad_norm": 1.3288185839554574, "learning_rate": 0.003, "loss": 4.0485, "step": 19828 }, { "epoch": 0.19829, "grad_norm": 1.230331722424284, "learning_rate": 0.003, "loss": 4.0263, "step": 19829 }, { "epoch": 0.1983, "grad_norm": 1.2536637743366936, "learning_rate": 0.003, "loss": 4.0357, "step": 19830 }, { "epoch": 0.19831, "grad_norm": 1.0284825575681031, "learning_rate": 0.003, "loss": 4.0556, "step": 19831 }, { "epoch": 0.19832, "grad_norm": 1.5022375842577895, "learning_rate": 0.003, "loss": 4.0526, "step": 19832 }, { "epoch": 0.19833, "grad_norm": 1.1806101547354726, "learning_rate": 0.003, "loss": 4.0401, "step": 19833 }, { "epoch": 0.19834, "grad_norm": 1.5733560786177732, "learning_rate": 0.003, "loss": 4.0285, "step": 19834 }, { "epoch": 0.19835, "grad_norm": 1.0162032290423761, "learning_rate": 0.003, "loss": 4.0465, "step": 19835 }, { "epoch": 0.19836, "grad_norm": 1.3385074775758588, "learning_rate": 0.003, "loss": 4.0375, "step": 19836 }, { "epoch": 0.19837, "grad_norm": 1.2916607796056685, "learning_rate": 0.003, "loss": 4.0504, "step": 19837 }, { "epoch": 0.19838, "grad_norm": 1.1405657623166203, "learning_rate": 0.003, "loss": 4.0408, "step": 19838 }, { "epoch": 0.19839, "grad_norm": 1.3169298747833968, "learning_rate": 0.003, "loss": 4.0303, "step": 19839 }, { "epoch": 0.1984, "grad_norm": 1.3005917176876103, "learning_rate": 0.003, "loss": 4.0645, "step": 19840 }, { "epoch": 0.19841, "grad_norm": 1.1950476180369751, "learning_rate": 0.003, "loss": 4.0504, "step": 19841 }, { "epoch": 0.19842, "grad_norm": 1.5084819204641784, "learning_rate": 0.003, "loss": 4.0595, "step": 19842 }, { "epoch": 0.19843, "grad_norm": 1.3434051931634823, "learning_rate": 0.003, "loss": 4.0606, "step": 19843 }, { "epoch": 0.19844, "grad_norm": 1.1731326757972111, "learning_rate": 0.003, "loss": 4.044, "step": 19844 }, { "epoch": 0.19845, "grad_norm": 1.340524498590452, "learning_rate": 0.003, "loss": 4.0474, "step": 19845 }, { "epoch": 0.19846, "grad_norm": 1.1333914521541875, "learning_rate": 0.003, "loss": 4.0467, "step": 19846 }, { "epoch": 0.19847, "grad_norm": 1.3156608871762097, "learning_rate": 0.003, "loss": 4.041, "step": 19847 }, { "epoch": 0.19848, "grad_norm": 1.1448220768181383, "learning_rate": 0.003, "loss": 4.024, "step": 19848 }, { "epoch": 0.19849, "grad_norm": 1.2130472682156814, "learning_rate": 0.003, "loss": 4.0447, "step": 19849 }, { "epoch": 0.1985, "grad_norm": 1.6346768634941562, "learning_rate": 0.003, "loss": 4.044, "step": 19850 }, { "epoch": 0.19851, "grad_norm": 1.254898576966238, "learning_rate": 0.003, "loss": 4.0474, "step": 19851 }, { "epoch": 0.19852, "grad_norm": 1.1760033171758264, "learning_rate": 0.003, "loss": 4.0424, "step": 19852 }, { "epoch": 0.19853, "grad_norm": 1.3457811542948783, "learning_rate": 0.003, "loss": 4.0323, "step": 19853 }, { "epoch": 0.19854, "grad_norm": 1.058916187847548, "learning_rate": 0.003, "loss": 4.0418, "step": 19854 }, { "epoch": 0.19855, "grad_norm": 1.7990623310783709, "learning_rate": 0.003, "loss": 4.0428, "step": 19855 }, { "epoch": 0.19856, "grad_norm": 0.8837071904663756, "learning_rate": 0.003, "loss": 4.0611, "step": 19856 }, { "epoch": 0.19857, "grad_norm": 1.142372906591354, "learning_rate": 0.003, "loss": 4.042, "step": 19857 }, { "epoch": 0.19858, "grad_norm": 1.2797239297455683, "learning_rate": 0.003, "loss": 4.0588, "step": 19858 }, { "epoch": 0.19859, "grad_norm": 1.3144649280053993, "learning_rate": 0.003, "loss": 4.0556, "step": 19859 }, { "epoch": 0.1986, "grad_norm": 1.0841021383129212, "learning_rate": 0.003, "loss": 4.055, "step": 19860 }, { "epoch": 0.19861, "grad_norm": 1.2014366394398417, "learning_rate": 0.003, "loss": 4.0598, "step": 19861 }, { "epoch": 0.19862, "grad_norm": 1.0775585816687974, "learning_rate": 0.003, "loss": 4.0482, "step": 19862 }, { "epoch": 0.19863, "grad_norm": 1.1743469652799556, "learning_rate": 0.003, "loss": 4.0622, "step": 19863 }, { "epoch": 0.19864, "grad_norm": 1.410076067702952, "learning_rate": 0.003, "loss": 4.0697, "step": 19864 }, { "epoch": 0.19865, "grad_norm": 1.1785401252329866, "learning_rate": 0.003, "loss": 4.0594, "step": 19865 }, { "epoch": 0.19866, "grad_norm": 1.495714534192095, "learning_rate": 0.003, "loss": 4.0212, "step": 19866 }, { "epoch": 0.19867, "grad_norm": 1.5079690053059533, "learning_rate": 0.003, "loss": 4.0602, "step": 19867 }, { "epoch": 0.19868, "grad_norm": 1.3028722347832058, "learning_rate": 0.003, "loss": 4.0672, "step": 19868 }, { "epoch": 0.19869, "grad_norm": 1.2229180419660335, "learning_rate": 0.003, "loss": 4.0634, "step": 19869 }, { "epoch": 0.1987, "grad_norm": 1.3000083287845383, "learning_rate": 0.003, "loss": 4.0565, "step": 19870 }, { "epoch": 0.19871, "grad_norm": 1.2806149133519935, "learning_rate": 0.003, "loss": 4.0546, "step": 19871 }, { "epoch": 0.19872, "grad_norm": 1.4813382010985194, "learning_rate": 0.003, "loss": 4.0394, "step": 19872 }, { "epoch": 0.19873, "grad_norm": 1.107017891996426, "learning_rate": 0.003, "loss": 4.0715, "step": 19873 }, { "epoch": 0.19874, "grad_norm": 1.6276836704174702, "learning_rate": 0.003, "loss": 4.0644, "step": 19874 }, { "epoch": 0.19875, "grad_norm": 0.9401675544626849, "learning_rate": 0.003, "loss": 4.0349, "step": 19875 }, { "epoch": 0.19876, "grad_norm": 1.1573380857665263, "learning_rate": 0.003, "loss": 4.0559, "step": 19876 }, { "epoch": 0.19877, "grad_norm": 1.5833019989287398, "learning_rate": 0.003, "loss": 4.0631, "step": 19877 }, { "epoch": 0.19878, "grad_norm": 1.00550242586545, "learning_rate": 0.003, "loss": 4.0557, "step": 19878 }, { "epoch": 0.19879, "grad_norm": 1.7589818815100986, "learning_rate": 0.003, "loss": 4.0312, "step": 19879 }, { "epoch": 0.1988, "grad_norm": 1.2298928930003272, "learning_rate": 0.003, "loss": 4.0632, "step": 19880 }, { "epoch": 0.19881, "grad_norm": 1.255587711606598, "learning_rate": 0.003, "loss": 4.0573, "step": 19881 }, { "epoch": 0.19882, "grad_norm": 1.221522282609823, "learning_rate": 0.003, "loss": 4.0864, "step": 19882 }, { "epoch": 0.19883, "grad_norm": 1.2605578122182348, "learning_rate": 0.003, "loss": 4.0623, "step": 19883 }, { "epoch": 0.19884, "grad_norm": 1.3609947205276998, "learning_rate": 0.003, "loss": 4.0711, "step": 19884 }, { "epoch": 0.19885, "grad_norm": 1.1422373658539642, "learning_rate": 0.003, "loss": 4.0692, "step": 19885 }, { "epoch": 0.19886, "grad_norm": 1.4430124926734533, "learning_rate": 0.003, "loss": 4.0402, "step": 19886 }, { "epoch": 0.19887, "grad_norm": 1.0289263684445262, "learning_rate": 0.003, "loss": 4.0797, "step": 19887 }, { "epoch": 0.19888, "grad_norm": 1.2489489645216596, "learning_rate": 0.003, "loss": 4.0517, "step": 19888 }, { "epoch": 0.19889, "grad_norm": 1.1521252967917146, "learning_rate": 0.003, "loss": 4.0442, "step": 19889 }, { "epoch": 0.1989, "grad_norm": 1.342265374894575, "learning_rate": 0.003, "loss": 4.0697, "step": 19890 }, { "epoch": 0.19891, "grad_norm": 1.1780633659571689, "learning_rate": 0.003, "loss": 4.0549, "step": 19891 }, { "epoch": 0.19892, "grad_norm": 1.260831484289609, "learning_rate": 0.003, "loss": 4.0337, "step": 19892 }, { "epoch": 0.19893, "grad_norm": 1.218498024755539, "learning_rate": 0.003, "loss": 4.0536, "step": 19893 }, { "epoch": 0.19894, "grad_norm": 1.4019787340199494, "learning_rate": 0.003, "loss": 4.0527, "step": 19894 }, { "epoch": 0.19895, "grad_norm": 1.023096803623969, "learning_rate": 0.003, "loss": 4.0723, "step": 19895 }, { "epoch": 0.19896, "grad_norm": 1.5085660726002734, "learning_rate": 0.003, "loss": 4.0491, "step": 19896 }, { "epoch": 0.19897, "grad_norm": 1.2362003169101425, "learning_rate": 0.003, "loss": 4.0294, "step": 19897 }, { "epoch": 0.19898, "grad_norm": 1.2063931465920943, "learning_rate": 0.003, "loss": 4.0604, "step": 19898 }, { "epoch": 0.19899, "grad_norm": 1.255079633207274, "learning_rate": 0.003, "loss": 4.0254, "step": 19899 }, { "epoch": 0.199, "grad_norm": 1.0795335302369486, "learning_rate": 0.003, "loss": 4.0405, "step": 19900 }, { "epoch": 0.19901, "grad_norm": 1.517738283239204, "learning_rate": 0.003, "loss": 4.0436, "step": 19901 }, { "epoch": 0.19902, "grad_norm": 1.275472442563502, "learning_rate": 0.003, "loss": 4.0535, "step": 19902 }, { "epoch": 0.19903, "grad_norm": 1.598886659373521, "learning_rate": 0.003, "loss": 4.0774, "step": 19903 }, { "epoch": 0.19904, "grad_norm": 1.174823330451563, "learning_rate": 0.003, "loss": 4.0609, "step": 19904 }, { "epoch": 0.19905, "grad_norm": 1.3753719989298585, "learning_rate": 0.003, "loss": 4.063, "step": 19905 }, { "epoch": 0.19906, "grad_norm": 1.2271065069522846, "learning_rate": 0.003, "loss": 4.0392, "step": 19906 }, { "epoch": 0.19907, "grad_norm": 1.311042604189716, "learning_rate": 0.003, "loss": 4.0611, "step": 19907 }, { "epoch": 0.19908, "grad_norm": 1.215407641539844, "learning_rate": 0.003, "loss": 4.011, "step": 19908 }, { "epoch": 0.19909, "grad_norm": 1.4285042785820834, "learning_rate": 0.003, "loss": 4.0403, "step": 19909 }, { "epoch": 0.1991, "grad_norm": 1.032240696399498, "learning_rate": 0.003, "loss": 4.0413, "step": 19910 }, { "epoch": 0.19911, "grad_norm": 1.5493249948776597, "learning_rate": 0.003, "loss": 4.0672, "step": 19911 }, { "epoch": 0.19912, "grad_norm": 1.2445896337671107, "learning_rate": 0.003, "loss": 4.0607, "step": 19912 }, { "epoch": 0.19913, "grad_norm": 1.2623958234009334, "learning_rate": 0.003, "loss": 4.0639, "step": 19913 }, { "epoch": 0.19914, "grad_norm": 1.175614375207124, "learning_rate": 0.003, "loss": 4.0419, "step": 19914 }, { "epoch": 0.19915, "grad_norm": 1.4829885795294195, "learning_rate": 0.003, "loss": 4.0564, "step": 19915 }, { "epoch": 0.19916, "grad_norm": 1.0508722651509181, "learning_rate": 0.003, "loss": 4.0517, "step": 19916 }, { "epoch": 0.19917, "grad_norm": 1.4794229065211975, "learning_rate": 0.003, "loss": 4.0712, "step": 19917 }, { "epoch": 0.19918, "grad_norm": 0.9297647168578973, "learning_rate": 0.003, "loss": 4.0587, "step": 19918 }, { "epoch": 0.19919, "grad_norm": 1.4133945135714843, "learning_rate": 0.003, "loss": 4.0594, "step": 19919 }, { "epoch": 0.1992, "grad_norm": 1.2317095796741588, "learning_rate": 0.003, "loss": 4.0703, "step": 19920 }, { "epoch": 0.19921, "grad_norm": 1.3173487800851562, "learning_rate": 0.003, "loss": 4.0693, "step": 19921 }, { "epoch": 0.19922, "grad_norm": 1.123948441025127, "learning_rate": 0.003, "loss": 4.0474, "step": 19922 }, { "epoch": 0.19923, "grad_norm": 1.378439216093473, "learning_rate": 0.003, "loss": 4.0758, "step": 19923 }, { "epoch": 0.19924, "grad_norm": 1.3997852719612403, "learning_rate": 0.003, "loss": 4.0622, "step": 19924 }, { "epoch": 0.19925, "grad_norm": 1.159786235179167, "learning_rate": 0.003, "loss": 4.0725, "step": 19925 }, { "epoch": 0.19926, "grad_norm": 1.3538870408494252, "learning_rate": 0.003, "loss": 4.06, "step": 19926 }, { "epoch": 0.19927, "grad_norm": 1.392540317577808, "learning_rate": 0.003, "loss": 4.0557, "step": 19927 }, { "epoch": 0.19928, "grad_norm": 1.5589978943880642, "learning_rate": 0.003, "loss": 4.0645, "step": 19928 }, { "epoch": 0.19929, "grad_norm": 1.335774839412378, "learning_rate": 0.003, "loss": 4.0552, "step": 19929 }, { "epoch": 0.1993, "grad_norm": 1.1407088420493976, "learning_rate": 0.003, "loss": 4.0448, "step": 19930 }, { "epoch": 0.19931, "grad_norm": 1.4056955897389483, "learning_rate": 0.003, "loss": 4.0884, "step": 19931 }, { "epoch": 0.19932, "grad_norm": 1.240044423909628, "learning_rate": 0.003, "loss": 4.0589, "step": 19932 }, { "epoch": 0.19933, "grad_norm": 1.3847437309510247, "learning_rate": 0.003, "loss": 4.0527, "step": 19933 }, { "epoch": 0.19934, "grad_norm": 1.1917714786704507, "learning_rate": 0.003, "loss": 4.0651, "step": 19934 }, { "epoch": 0.19935, "grad_norm": 1.1149699870223053, "learning_rate": 0.003, "loss": 4.0339, "step": 19935 }, { "epoch": 0.19936, "grad_norm": 1.1342568304392888, "learning_rate": 0.003, "loss": 4.0652, "step": 19936 }, { "epoch": 0.19937, "grad_norm": 1.2700039155178806, "learning_rate": 0.003, "loss": 4.0525, "step": 19937 }, { "epoch": 0.19938, "grad_norm": 1.2080604366635694, "learning_rate": 0.003, "loss": 4.0439, "step": 19938 }, { "epoch": 0.19939, "grad_norm": 1.244308802459882, "learning_rate": 0.003, "loss": 4.0479, "step": 19939 }, { "epoch": 0.1994, "grad_norm": 1.1106149313546723, "learning_rate": 0.003, "loss": 4.0254, "step": 19940 }, { "epoch": 0.19941, "grad_norm": 1.3238101030344678, "learning_rate": 0.003, "loss": 4.068, "step": 19941 }, { "epoch": 0.19942, "grad_norm": 1.2034382199853526, "learning_rate": 0.003, "loss": 4.0341, "step": 19942 }, { "epoch": 0.19943, "grad_norm": 1.4013367377250077, "learning_rate": 0.003, "loss": 4.0758, "step": 19943 }, { "epoch": 0.19944, "grad_norm": 1.2972300230868017, "learning_rate": 0.003, "loss": 4.0762, "step": 19944 }, { "epoch": 0.19945, "grad_norm": 1.2951811204708457, "learning_rate": 0.003, "loss": 4.0525, "step": 19945 }, { "epoch": 0.19946, "grad_norm": 1.1610586091426425, "learning_rate": 0.003, "loss": 4.0599, "step": 19946 }, { "epoch": 0.19947, "grad_norm": 1.4178084078057516, "learning_rate": 0.003, "loss": 4.0397, "step": 19947 }, { "epoch": 0.19948, "grad_norm": 1.158927526770478, "learning_rate": 0.003, "loss": 4.0471, "step": 19948 }, { "epoch": 0.19949, "grad_norm": 1.2809085892791512, "learning_rate": 0.003, "loss": 4.0391, "step": 19949 }, { "epoch": 0.1995, "grad_norm": 1.3862229637166266, "learning_rate": 0.003, "loss": 4.0384, "step": 19950 }, { "epoch": 0.19951, "grad_norm": 1.1954504386101603, "learning_rate": 0.003, "loss": 4.0482, "step": 19951 }, { "epoch": 0.19952, "grad_norm": 1.3199456984155447, "learning_rate": 0.003, "loss": 4.0746, "step": 19952 }, { "epoch": 0.19953, "grad_norm": 1.2481296971378508, "learning_rate": 0.003, "loss": 4.0321, "step": 19953 }, { "epoch": 0.19954, "grad_norm": 1.2697107338936542, "learning_rate": 0.003, "loss": 4.0288, "step": 19954 }, { "epoch": 0.19955, "grad_norm": 1.119725423462046, "learning_rate": 0.003, "loss": 4.0379, "step": 19955 }, { "epoch": 0.19956, "grad_norm": 1.2760293725356533, "learning_rate": 0.003, "loss": 4.0425, "step": 19956 }, { "epoch": 0.19957, "grad_norm": 1.297012595989181, "learning_rate": 0.003, "loss": 4.0384, "step": 19957 }, { "epoch": 0.19958, "grad_norm": 1.419427840824775, "learning_rate": 0.003, "loss": 4.0725, "step": 19958 }, { "epoch": 0.19959, "grad_norm": 0.9602902187756038, "learning_rate": 0.003, "loss": 4.0536, "step": 19959 }, { "epoch": 0.1996, "grad_norm": 1.1223572375743875, "learning_rate": 0.003, "loss": 4.0357, "step": 19960 }, { "epoch": 0.19961, "grad_norm": 1.5950858182495724, "learning_rate": 0.003, "loss": 4.0647, "step": 19961 }, { "epoch": 0.19962, "grad_norm": 1.1441204681554322, "learning_rate": 0.003, "loss": 4.0634, "step": 19962 }, { "epoch": 0.19963, "grad_norm": 1.3126206493847443, "learning_rate": 0.003, "loss": 4.062, "step": 19963 }, { "epoch": 0.19964, "grad_norm": 1.2433338163350853, "learning_rate": 0.003, "loss": 4.0471, "step": 19964 }, { "epoch": 0.19965, "grad_norm": 1.291102879209652, "learning_rate": 0.003, "loss": 4.0627, "step": 19965 }, { "epoch": 0.19966, "grad_norm": 1.0747583379896155, "learning_rate": 0.003, "loss": 4.048, "step": 19966 }, { "epoch": 0.19967, "grad_norm": 1.519028210916497, "learning_rate": 0.003, "loss": 4.06, "step": 19967 }, { "epoch": 0.19968, "grad_norm": 1.0716160858687132, "learning_rate": 0.003, "loss": 4.0378, "step": 19968 }, { "epoch": 0.19969, "grad_norm": 1.5164868648659249, "learning_rate": 0.003, "loss": 4.0391, "step": 19969 }, { "epoch": 0.1997, "grad_norm": 0.9582219395470151, "learning_rate": 0.003, "loss": 4.0372, "step": 19970 }, { "epoch": 0.19971, "grad_norm": 1.3474662875829415, "learning_rate": 0.003, "loss": 4.0168, "step": 19971 }, { "epoch": 0.19972, "grad_norm": 1.4013781080276175, "learning_rate": 0.003, "loss": 4.0571, "step": 19972 }, { "epoch": 0.19973, "grad_norm": 1.1454916139054667, "learning_rate": 0.003, "loss": 4.0627, "step": 19973 }, { "epoch": 0.19974, "grad_norm": 1.3984598096784422, "learning_rate": 0.003, "loss": 4.035, "step": 19974 }, { "epoch": 0.19975, "grad_norm": 1.3501707571712709, "learning_rate": 0.003, "loss": 4.0344, "step": 19975 }, { "epoch": 0.19976, "grad_norm": 1.3484662570777235, "learning_rate": 0.003, "loss": 4.0403, "step": 19976 }, { "epoch": 0.19977, "grad_norm": 1.216492280996437, "learning_rate": 0.003, "loss": 4.0135, "step": 19977 }, { "epoch": 0.19978, "grad_norm": 1.142709040714013, "learning_rate": 0.003, "loss": 4.0769, "step": 19978 }, { "epoch": 0.19979, "grad_norm": 1.1135564681824013, "learning_rate": 0.003, "loss": 4.0536, "step": 19979 }, { "epoch": 0.1998, "grad_norm": 1.2621270377951224, "learning_rate": 0.003, "loss": 4.0416, "step": 19980 }, { "epoch": 0.19981, "grad_norm": 1.403693140830989, "learning_rate": 0.003, "loss": 4.0406, "step": 19981 }, { "epoch": 0.19982, "grad_norm": 0.9981059833386207, "learning_rate": 0.003, "loss": 4.0567, "step": 19982 }, { "epoch": 0.19983, "grad_norm": 1.3993533271308505, "learning_rate": 0.003, "loss": 4.0969, "step": 19983 }, { "epoch": 0.19984, "grad_norm": 1.3028828668644656, "learning_rate": 0.003, "loss": 4.0654, "step": 19984 }, { "epoch": 0.19985, "grad_norm": 1.4637659181918143, "learning_rate": 0.003, "loss": 4.0419, "step": 19985 }, { "epoch": 0.19986, "grad_norm": 1.1807978110160766, "learning_rate": 0.003, "loss": 4.0468, "step": 19986 }, { "epoch": 0.19987, "grad_norm": 1.4742560009047105, "learning_rate": 0.003, "loss": 4.028, "step": 19987 }, { "epoch": 0.19988, "grad_norm": 1.173404595311201, "learning_rate": 0.003, "loss": 4.003, "step": 19988 }, { "epoch": 0.19989, "grad_norm": 1.25242865522054, "learning_rate": 0.003, "loss": 4.0523, "step": 19989 }, { "epoch": 0.1999, "grad_norm": 1.387684480624249, "learning_rate": 0.003, "loss": 4.0562, "step": 19990 }, { "epoch": 0.19991, "grad_norm": 1.3890228225877637, "learning_rate": 0.003, "loss": 4.0446, "step": 19991 }, { "epoch": 0.19992, "grad_norm": 1.0908316718165803, "learning_rate": 0.003, "loss": 4.0253, "step": 19992 }, { "epoch": 0.19993, "grad_norm": 1.3039394189034819, "learning_rate": 0.003, "loss": 4.0708, "step": 19993 }, { "epoch": 0.19994, "grad_norm": 1.3524185493509755, "learning_rate": 0.003, "loss": 4.072, "step": 19994 }, { "epoch": 0.19995, "grad_norm": 1.1431826798538707, "learning_rate": 0.003, "loss": 4.0431, "step": 19995 }, { "epoch": 0.19996, "grad_norm": 1.0803834019026506, "learning_rate": 0.003, "loss": 4.0191, "step": 19996 }, { "epoch": 0.19997, "grad_norm": 1.285890308629624, "learning_rate": 0.003, "loss": 4.0309, "step": 19997 }, { "epoch": 0.19998, "grad_norm": 1.3028078375512837, "learning_rate": 0.003, "loss": 4.0623, "step": 19998 }, { "epoch": 0.19999, "grad_norm": 1.6457298761704569, "learning_rate": 0.003, "loss": 4.0306, "step": 19999 }, { "epoch": 0.2, "grad_norm": 1.115562086778755, "learning_rate": 0.003, "loss": 4.0246, "step": 20000 }, { "epoch": 0.20001, "grad_norm": 1.622288790814596, "learning_rate": 0.003, "loss": 4.0279, "step": 20001 }, { "epoch": 0.20002, "grad_norm": 1.165374122105647, "learning_rate": 0.003, "loss": 4.0426, "step": 20002 }, { "epoch": 0.20003, "grad_norm": 1.180965129177058, "learning_rate": 0.003, "loss": 4.0643, "step": 20003 }, { "epoch": 0.20004, "grad_norm": 1.3909555533551616, "learning_rate": 0.003, "loss": 4.0349, "step": 20004 }, { "epoch": 0.20005, "grad_norm": 1.0214383584852353, "learning_rate": 0.003, "loss": 4.06, "step": 20005 }, { "epoch": 0.20006, "grad_norm": 1.4877283163255348, "learning_rate": 0.003, "loss": 4.0222, "step": 20006 }, { "epoch": 0.20007, "grad_norm": 0.9872070090617107, "learning_rate": 0.003, "loss": 4.0234, "step": 20007 }, { "epoch": 0.20008, "grad_norm": 1.4735250417481307, "learning_rate": 0.003, "loss": 4.0563, "step": 20008 }, { "epoch": 0.20009, "grad_norm": 1.1796954083047972, "learning_rate": 0.003, "loss": 4.0724, "step": 20009 }, { "epoch": 0.2001, "grad_norm": 1.3488507948704809, "learning_rate": 0.003, "loss": 4.047, "step": 20010 }, { "epoch": 0.20011, "grad_norm": 1.3140881598709688, "learning_rate": 0.003, "loss": 4.077, "step": 20011 }, { "epoch": 0.20012, "grad_norm": 1.289546318132687, "learning_rate": 0.003, "loss": 4.0712, "step": 20012 }, { "epoch": 0.20013, "grad_norm": 1.500309834972784, "learning_rate": 0.003, "loss": 4.0842, "step": 20013 }, { "epoch": 0.20014, "grad_norm": 1.1858787717683152, "learning_rate": 0.003, "loss": 4.0579, "step": 20014 }, { "epoch": 0.20015, "grad_norm": 1.0725535067917935, "learning_rate": 0.003, "loss": 4.0429, "step": 20015 }, { "epoch": 0.20016, "grad_norm": 1.597372267607848, "learning_rate": 0.003, "loss": 4.0686, "step": 20016 }, { "epoch": 0.20017, "grad_norm": 1.0563564460372217, "learning_rate": 0.003, "loss": 4.0201, "step": 20017 }, { "epoch": 0.20018, "grad_norm": 1.2592839008600343, "learning_rate": 0.003, "loss": 4.0431, "step": 20018 }, { "epoch": 0.20019, "grad_norm": 1.100640324338181, "learning_rate": 0.003, "loss": 4.0548, "step": 20019 }, { "epoch": 0.2002, "grad_norm": 1.4228491575051412, "learning_rate": 0.003, "loss": 4.053, "step": 20020 }, { "epoch": 0.20021, "grad_norm": 1.179568307485374, "learning_rate": 0.003, "loss": 4.0822, "step": 20021 }, { "epoch": 0.20022, "grad_norm": 1.3440274696691585, "learning_rate": 0.003, "loss": 4.0534, "step": 20022 }, { "epoch": 0.20023, "grad_norm": 1.054115868379596, "learning_rate": 0.003, "loss": 4.0493, "step": 20023 }, { "epoch": 0.20024, "grad_norm": 1.2388225483904531, "learning_rate": 0.003, "loss": 4.0431, "step": 20024 }, { "epoch": 0.20025, "grad_norm": 1.2029896621757317, "learning_rate": 0.003, "loss": 4.0299, "step": 20025 }, { "epoch": 0.20026, "grad_norm": 1.5468913707142669, "learning_rate": 0.003, "loss": 4.0491, "step": 20026 }, { "epoch": 0.20027, "grad_norm": 1.2897726030241574, "learning_rate": 0.003, "loss": 4.052, "step": 20027 }, { "epoch": 0.20028, "grad_norm": 1.3801929628271197, "learning_rate": 0.003, "loss": 4.0314, "step": 20028 }, { "epoch": 0.20029, "grad_norm": 1.111478986976431, "learning_rate": 0.003, "loss": 4.0232, "step": 20029 }, { "epoch": 0.2003, "grad_norm": 1.2733116930929462, "learning_rate": 0.003, "loss": 4.0516, "step": 20030 }, { "epoch": 0.20031, "grad_norm": 1.1679315621127557, "learning_rate": 0.003, "loss": 4.0529, "step": 20031 }, { "epoch": 0.20032, "grad_norm": 1.2552177994447329, "learning_rate": 0.003, "loss": 4.0489, "step": 20032 }, { "epoch": 0.20033, "grad_norm": 1.013463527645874, "learning_rate": 0.003, "loss": 4.0448, "step": 20033 }, { "epoch": 0.20034, "grad_norm": 1.4306374729371047, "learning_rate": 0.003, "loss": 4.0431, "step": 20034 }, { "epoch": 0.20035, "grad_norm": 1.316492598084679, "learning_rate": 0.003, "loss": 4.0521, "step": 20035 }, { "epoch": 0.20036, "grad_norm": 1.5313705166049483, "learning_rate": 0.003, "loss": 4.0904, "step": 20036 }, { "epoch": 0.20037, "grad_norm": 1.1156570219061208, "learning_rate": 0.003, "loss": 4.0404, "step": 20037 }, { "epoch": 0.20038, "grad_norm": 1.2566360302027615, "learning_rate": 0.003, "loss": 4.051, "step": 20038 }, { "epoch": 0.20039, "grad_norm": 1.2317306888095088, "learning_rate": 0.003, "loss": 4.0285, "step": 20039 }, { "epoch": 0.2004, "grad_norm": 1.3288950327435645, "learning_rate": 0.003, "loss": 4.0726, "step": 20040 }, { "epoch": 0.20041, "grad_norm": 1.3179608954967692, "learning_rate": 0.003, "loss": 4.0585, "step": 20041 }, { "epoch": 0.20042, "grad_norm": 1.3907809117743692, "learning_rate": 0.003, "loss": 4.0969, "step": 20042 }, { "epoch": 0.20043, "grad_norm": 1.2781584573939455, "learning_rate": 0.003, "loss": 4.0538, "step": 20043 }, { "epoch": 0.20044, "grad_norm": 1.3675301151990953, "learning_rate": 0.003, "loss": 4.0458, "step": 20044 }, { "epoch": 0.20045, "grad_norm": 1.270993402537075, "learning_rate": 0.003, "loss": 4.0351, "step": 20045 }, { "epoch": 0.20046, "grad_norm": 1.2574780035973352, "learning_rate": 0.003, "loss": 4.0264, "step": 20046 }, { "epoch": 0.20047, "grad_norm": 1.3772785672275094, "learning_rate": 0.003, "loss": 4.0331, "step": 20047 }, { "epoch": 0.20048, "grad_norm": 1.0130786444460946, "learning_rate": 0.003, "loss": 4.0547, "step": 20048 }, { "epoch": 0.20049, "grad_norm": 1.4723445410063, "learning_rate": 0.003, "loss": 4.0419, "step": 20049 }, { "epoch": 0.2005, "grad_norm": 0.9607882441038933, "learning_rate": 0.003, "loss": 4.037, "step": 20050 }, { "epoch": 0.20051, "grad_norm": 1.3958089165986418, "learning_rate": 0.003, "loss": 4.0526, "step": 20051 }, { "epoch": 0.20052, "grad_norm": 1.1082284128077051, "learning_rate": 0.003, "loss": 4.0403, "step": 20052 }, { "epoch": 0.20053, "grad_norm": 1.2324738316505863, "learning_rate": 0.003, "loss": 4.0398, "step": 20053 }, { "epoch": 0.20054, "grad_norm": 1.1701801286715698, "learning_rate": 0.003, "loss": 4.0593, "step": 20054 }, { "epoch": 0.20055, "grad_norm": 1.5767109989106038, "learning_rate": 0.003, "loss": 4.0433, "step": 20055 }, { "epoch": 0.20056, "grad_norm": 1.1822714727528925, "learning_rate": 0.003, "loss": 4.0678, "step": 20056 }, { "epoch": 0.20057, "grad_norm": 1.512624540405217, "learning_rate": 0.003, "loss": 4.0458, "step": 20057 }, { "epoch": 0.20058, "grad_norm": 0.9959903059781854, "learning_rate": 0.003, "loss": 4.0575, "step": 20058 }, { "epoch": 0.20059, "grad_norm": 1.3754117091411615, "learning_rate": 0.003, "loss": 4.0719, "step": 20059 }, { "epoch": 0.2006, "grad_norm": 1.0748141614161064, "learning_rate": 0.003, "loss": 4.0262, "step": 20060 }, { "epoch": 0.20061, "grad_norm": 1.531186154248781, "learning_rate": 0.003, "loss": 4.0742, "step": 20061 }, { "epoch": 0.20062, "grad_norm": 1.1058611864267804, "learning_rate": 0.003, "loss": 4.0717, "step": 20062 }, { "epoch": 0.20063, "grad_norm": 1.524239022010734, "learning_rate": 0.003, "loss": 4.0937, "step": 20063 }, { "epoch": 0.20064, "grad_norm": 1.1334731641616702, "learning_rate": 0.003, "loss": 4.0879, "step": 20064 }, { "epoch": 0.20065, "grad_norm": 1.4810691498472939, "learning_rate": 0.003, "loss": 4.0684, "step": 20065 }, { "epoch": 0.20066, "grad_norm": 1.0674910488827514, "learning_rate": 0.003, "loss": 4.0225, "step": 20066 }, { "epoch": 0.20067, "grad_norm": 1.6207322617765014, "learning_rate": 0.003, "loss": 4.0654, "step": 20067 }, { "epoch": 0.20068, "grad_norm": 1.2086115569280247, "learning_rate": 0.003, "loss": 4.041, "step": 20068 }, { "epoch": 0.20069, "grad_norm": 1.2904239609284227, "learning_rate": 0.003, "loss": 4.0686, "step": 20069 }, { "epoch": 0.2007, "grad_norm": 1.0621367796085464, "learning_rate": 0.003, "loss": 4.0806, "step": 20070 }, { "epoch": 0.20071, "grad_norm": 1.4303644222363705, "learning_rate": 0.003, "loss": 4.0567, "step": 20071 }, { "epoch": 0.20072, "grad_norm": 1.1062087157842548, "learning_rate": 0.003, "loss": 4.0131, "step": 20072 }, { "epoch": 0.20073, "grad_norm": 1.3451566024188915, "learning_rate": 0.003, "loss": 4.0205, "step": 20073 }, { "epoch": 0.20074, "grad_norm": 1.1083985823857898, "learning_rate": 0.003, "loss": 4.0465, "step": 20074 }, { "epoch": 0.20075, "grad_norm": 1.6352535934136272, "learning_rate": 0.003, "loss": 4.0581, "step": 20075 }, { "epoch": 0.20076, "grad_norm": 1.002940017778515, "learning_rate": 0.003, "loss": 4.0621, "step": 20076 }, { "epoch": 0.20077, "grad_norm": 1.439017091942511, "learning_rate": 0.003, "loss": 4.0308, "step": 20077 }, { "epoch": 0.20078, "grad_norm": 1.3782679495157797, "learning_rate": 0.003, "loss": 4.0732, "step": 20078 }, { "epoch": 0.20079, "grad_norm": 1.3827209292167564, "learning_rate": 0.003, "loss": 4.0422, "step": 20079 }, { "epoch": 0.2008, "grad_norm": 1.4801421508733992, "learning_rate": 0.003, "loss": 4.0682, "step": 20080 }, { "epoch": 0.20081, "grad_norm": 1.2299642601354424, "learning_rate": 0.003, "loss": 4.045, "step": 20081 }, { "epoch": 0.20082, "grad_norm": 1.0748645186589552, "learning_rate": 0.003, "loss": 4.0624, "step": 20082 }, { "epoch": 0.20083, "grad_norm": 1.6764368194458739, "learning_rate": 0.003, "loss": 4.0624, "step": 20083 }, { "epoch": 0.20084, "grad_norm": 1.1209269924804597, "learning_rate": 0.003, "loss": 4.0385, "step": 20084 }, { "epoch": 0.20085, "grad_norm": 1.4128168768585003, "learning_rate": 0.003, "loss": 4.0738, "step": 20085 }, { "epoch": 0.20086, "grad_norm": 0.8986786012657335, "learning_rate": 0.003, "loss": 4.0394, "step": 20086 }, { "epoch": 0.20087, "grad_norm": 1.3521795476722067, "learning_rate": 0.003, "loss": 4.062, "step": 20087 }, { "epoch": 0.20088, "grad_norm": 1.235014884231785, "learning_rate": 0.003, "loss": 4.0323, "step": 20088 }, { "epoch": 0.20089, "grad_norm": 1.2032672957274169, "learning_rate": 0.003, "loss": 4.0527, "step": 20089 }, { "epoch": 0.2009, "grad_norm": 1.4142481020944606, "learning_rate": 0.003, "loss": 4.0515, "step": 20090 }, { "epoch": 0.20091, "grad_norm": 1.3443612249758137, "learning_rate": 0.003, "loss": 4.06, "step": 20091 }, { "epoch": 0.20092, "grad_norm": 1.2835404914601647, "learning_rate": 0.003, "loss": 4.0668, "step": 20092 }, { "epoch": 0.20093, "grad_norm": 1.1366687506192215, "learning_rate": 0.003, "loss": 4.0509, "step": 20093 }, { "epoch": 0.20094, "grad_norm": 1.38504527416011, "learning_rate": 0.003, "loss": 4.0331, "step": 20094 }, { "epoch": 0.20095, "grad_norm": 1.28205422920678, "learning_rate": 0.003, "loss": 4.0771, "step": 20095 }, { "epoch": 0.20096, "grad_norm": 1.1859031552274366, "learning_rate": 0.003, "loss": 4.0303, "step": 20096 }, { "epoch": 0.20097, "grad_norm": 1.0980781630590808, "learning_rate": 0.003, "loss": 4.0508, "step": 20097 }, { "epoch": 0.20098, "grad_norm": 1.3229800200284534, "learning_rate": 0.003, "loss": 4.0378, "step": 20098 }, { "epoch": 0.20099, "grad_norm": 1.1651813466999181, "learning_rate": 0.003, "loss": 4.0594, "step": 20099 }, { "epoch": 0.201, "grad_norm": 1.4175521333513585, "learning_rate": 0.003, "loss": 4.0423, "step": 20100 }, { "epoch": 0.20101, "grad_norm": 1.2325766987635356, "learning_rate": 0.003, "loss": 4.0312, "step": 20101 }, { "epoch": 0.20102, "grad_norm": 1.3466611075080315, "learning_rate": 0.003, "loss": 4.078, "step": 20102 }, { "epoch": 0.20103, "grad_norm": 1.034269239083285, "learning_rate": 0.003, "loss": 4.051, "step": 20103 }, { "epoch": 0.20104, "grad_norm": 1.376615743184664, "learning_rate": 0.003, "loss": 4.0685, "step": 20104 }, { "epoch": 0.20105, "grad_norm": 1.0123650904183055, "learning_rate": 0.003, "loss": 4.0559, "step": 20105 }, { "epoch": 0.20106, "grad_norm": 1.486929100142425, "learning_rate": 0.003, "loss": 4.0629, "step": 20106 }, { "epoch": 0.20107, "grad_norm": 1.1278985311114458, "learning_rate": 0.003, "loss": 4.058, "step": 20107 }, { "epoch": 0.20108, "grad_norm": 1.3273253735382757, "learning_rate": 0.003, "loss": 4.0786, "step": 20108 }, { "epoch": 0.20109, "grad_norm": 1.4065619334016792, "learning_rate": 0.003, "loss": 4.0735, "step": 20109 }, { "epoch": 0.2011, "grad_norm": 1.1120393553676875, "learning_rate": 0.003, "loss": 4.0526, "step": 20110 }, { "epoch": 0.20111, "grad_norm": 1.4423889109397692, "learning_rate": 0.003, "loss": 4.0572, "step": 20111 }, { "epoch": 0.20112, "grad_norm": 1.1855066575374555, "learning_rate": 0.003, "loss": 4.0384, "step": 20112 }, { "epoch": 0.20113, "grad_norm": 1.5549470082093861, "learning_rate": 0.003, "loss": 4.0669, "step": 20113 }, { "epoch": 0.20114, "grad_norm": 1.137511816050523, "learning_rate": 0.003, "loss": 4.0395, "step": 20114 }, { "epoch": 0.20115, "grad_norm": 1.3462244667642502, "learning_rate": 0.003, "loss": 4.0855, "step": 20115 }, { "epoch": 0.20116, "grad_norm": 1.3206899218544779, "learning_rate": 0.003, "loss": 4.0375, "step": 20116 }, { "epoch": 0.20117, "grad_norm": 1.1252241220562909, "learning_rate": 0.003, "loss": 4.0379, "step": 20117 }, { "epoch": 0.20118, "grad_norm": 1.3005460086834528, "learning_rate": 0.003, "loss": 4.0507, "step": 20118 }, { "epoch": 0.20119, "grad_norm": 1.10493921156302, "learning_rate": 0.003, "loss": 4.0556, "step": 20119 }, { "epoch": 0.2012, "grad_norm": 1.3581398182718027, "learning_rate": 0.003, "loss": 4.0505, "step": 20120 }, { "epoch": 0.20121, "grad_norm": 1.2797980078882876, "learning_rate": 0.003, "loss": 4.0647, "step": 20121 }, { "epoch": 0.20122, "grad_norm": 1.4728301874267207, "learning_rate": 0.003, "loss": 4.0882, "step": 20122 }, { "epoch": 0.20123, "grad_norm": 0.9950607551670232, "learning_rate": 0.003, "loss": 4.0686, "step": 20123 }, { "epoch": 0.20124, "grad_norm": 1.5816024665862407, "learning_rate": 0.003, "loss": 4.0512, "step": 20124 }, { "epoch": 0.20125, "grad_norm": 1.1421784593217872, "learning_rate": 0.003, "loss": 4.0427, "step": 20125 }, { "epoch": 0.20126, "grad_norm": 1.5559806391426019, "learning_rate": 0.003, "loss": 4.0403, "step": 20126 }, { "epoch": 0.20127, "grad_norm": 1.1535837996632359, "learning_rate": 0.003, "loss": 4.047, "step": 20127 }, { "epoch": 0.20128, "grad_norm": 1.3551460341143398, "learning_rate": 0.003, "loss": 4.0785, "step": 20128 }, { "epoch": 0.20129, "grad_norm": 0.9418879294863098, "learning_rate": 0.003, "loss": 4.058, "step": 20129 }, { "epoch": 0.2013, "grad_norm": 1.3655003356694182, "learning_rate": 0.003, "loss": 4.0798, "step": 20130 }, { "epoch": 0.20131, "grad_norm": 1.2759535807730578, "learning_rate": 0.003, "loss": 4.0649, "step": 20131 }, { "epoch": 0.20132, "grad_norm": 1.198035312885684, "learning_rate": 0.003, "loss": 4.0735, "step": 20132 }, { "epoch": 0.20133, "grad_norm": 1.2929011908222066, "learning_rate": 0.003, "loss": 4.0565, "step": 20133 }, { "epoch": 0.20134, "grad_norm": 1.318354331958291, "learning_rate": 0.003, "loss": 4.0428, "step": 20134 }, { "epoch": 0.20135, "grad_norm": 1.8884102519126955, "learning_rate": 0.003, "loss": 4.07, "step": 20135 }, { "epoch": 0.20136, "grad_norm": 1.061299313732376, "learning_rate": 0.003, "loss": 4.0742, "step": 20136 }, { "epoch": 0.20137, "grad_norm": 1.267913944663337, "learning_rate": 0.003, "loss": 4.0286, "step": 20137 }, { "epoch": 0.20138, "grad_norm": 1.1492742229813098, "learning_rate": 0.003, "loss": 4.0655, "step": 20138 }, { "epoch": 0.20139, "grad_norm": 1.3433772181787036, "learning_rate": 0.003, "loss": 4.0547, "step": 20139 }, { "epoch": 0.2014, "grad_norm": 1.1855937629839706, "learning_rate": 0.003, "loss": 4.032, "step": 20140 }, { "epoch": 0.20141, "grad_norm": 1.5788519463679762, "learning_rate": 0.003, "loss": 4.0536, "step": 20141 }, { "epoch": 0.20142, "grad_norm": 1.1655157739928947, "learning_rate": 0.003, "loss": 4.0376, "step": 20142 }, { "epoch": 0.20143, "grad_norm": 1.3720381633727898, "learning_rate": 0.003, "loss": 4.0427, "step": 20143 }, { "epoch": 0.20144, "grad_norm": 1.0623338765075157, "learning_rate": 0.003, "loss": 4.0325, "step": 20144 }, { "epoch": 0.20145, "grad_norm": 1.3952375725860666, "learning_rate": 0.003, "loss": 4.0613, "step": 20145 }, { "epoch": 0.20146, "grad_norm": 1.148310575339641, "learning_rate": 0.003, "loss": 4.0755, "step": 20146 }, { "epoch": 0.20147, "grad_norm": 1.430096768068745, "learning_rate": 0.003, "loss": 4.0601, "step": 20147 }, { "epoch": 0.20148, "grad_norm": 1.1339120576674702, "learning_rate": 0.003, "loss": 4.0577, "step": 20148 }, { "epoch": 0.20149, "grad_norm": 1.264807104959973, "learning_rate": 0.003, "loss": 4.0438, "step": 20149 }, { "epoch": 0.2015, "grad_norm": 1.179430688158245, "learning_rate": 0.003, "loss": 4.0571, "step": 20150 }, { "epoch": 0.20151, "grad_norm": 1.2961004328981822, "learning_rate": 0.003, "loss": 4.0488, "step": 20151 }, { "epoch": 0.20152, "grad_norm": 1.0940237875476264, "learning_rate": 0.003, "loss": 4.0606, "step": 20152 }, { "epoch": 0.20153, "grad_norm": 1.286709407766725, "learning_rate": 0.003, "loss": 4.0353, "step": 20153 }, { "epoch": 0.20154, "grad_norm": 1.3227627119690266, "learning_rate": 0.003, "loss": 4.075, "step": 20154 }, { "epoch": 0.20155, "grad_norm": 1.6105251847809123, "learning_rate": 0.003, "loss": 4.0481, "step": 20155 }, { "epoch": 0.20156, "grad_norm": 1.0747232683539776, "learning_rate": 0.003, "loss": 4.0528, "step": 20156 }, { "epoch": 0.20157, "grad_norm": 1.3600242060341243, "learning_rate": 0.003, "loss": 4.0281, "step": 20157 }, { "epoch": 0.20158, "grad_norm": 1.2727788318316686, "learning_rate": 0.003, "loss": 4.0274, "step": 20158 }, { "epoch": 0.20159, "grad_norm": 1.449762729363481, "learning_rate": 0.003, "loss": 4.0196, "step": 20159 }, { "epoch": 0.2016, "grad_norm": 1.2522029223183724, "learning_rate": 0.003, "loss": 4.0287, "step": 20160 }, { "epoch": 0.20161, "grad_norm": 1.470939394443757, "learning_rate": 0.003, "loss": 4.0559, "step": 20161 }, { "epoch": 0.20162, "grad_norm": 0.9780713632241691, "learning_rate": 0.003, "loss": 4.0616, "step": 20162 }, { "epoch": 0.20163, "grad_norm": 1.4069970555682791, "learning_rate": 0.003, "loss": 4.0673, "step": 20163 }, { "epoch": 0.20164, "grad_norm": 1.1120523232113189, "learning_rate": 0.003, "loss": 4.0868, "step": 20164 }, { "epoch": 0.20165, "grad_norm": 1.3154390323043668, "learning_rate": 0.003, "loss": 4.0273, "step": 20165 }, { "epoch": 0.20166, "grad_norm": 1.2081791962327835, "learning_rate": 0.003, "loss": 4.0668, "step": 20166 }, { "epoch": 0.20167, "grad_norm": 1.382220430187685, "learning_rate": 0.003, "loss": 4.029, "step": 20167 }, { "epoch": 0.20168, "grad_norm": 1.2383371057025319, "learning_rate": 0.003, "loss": 4.0473, "step": 20168 }, { "epoch": 0.20169, "grad_norm": 1.2679904971753184, "learning_rate": 0.003, "loss": 4.0405, "step": 20169 }, { "epoch": 0.2017, "grad_norm": 1.3384110300248249, "learning_rate": 0.003, "loss": 4.051, "step": 20170 }, { "epoch": 0.20171, "grad_norm": 0.9326477901198288, "learning_rate": 0.003, "loss": 4.0673, "step": 20171 }, { "epoch": 0.20172, "grad_norm": 1.350582451313348, "learning_rate": 0.003, "loss": 4.0657, "step": 20172 }, { "epoch": 0.20173, "grad_norm": 1.340611180510728, "learning_rate": 0.003, "loss": 4.0054, "step": 20173 }, { "epoch": 0.20174, "grad_norm": 1.2952026073736063, "learning_rate": 0.003, "loss": 4.0256, "step": 20174 }, { "epoch": 0.20175, "grad_norm": 1.2413191354769146, "learning_rate": 0.003, "loss": 4.0526, "step": 20175 }, { "epoch": 0.20176, "grad_norm": 1.2481837574056514, "learning_rate": 0.003, "loss": 4.0514, "step": 20176 }, { "epoch": 0.20177, "grad_norm": 1.2435819576658815, "learning_rate": 0.003, "loss": 4.0588, "step": 20177 }, { "epoch": 0.20178, "grad_norm": 1.0742089009050002, "learning_rate": 0.003, "loss": 4.0388, "step": 20178 }, { "epoch": 0.20179, "grad_norm": 1.4036356251900837, "learning_rate": 0.003, "loss": 4.0587, "step": 20179 }, { "epoch": 0.2018, "grad_norm": 1.1926696153443954, "learning_rate": 0.003, "loss": 4.0571, "step": 20180 }, { "epoch": 0.20181, "grad_norm": 1.491775506134129, "learning_rate": 0.003, "loss": 4.0387, "step": 20181 }, { "epoch": 0.20182, "grad_norm": 1.1372863797918262, "learning_rate": 0.003, "loss": 4.0522, "step": 20182 }, { "epoch": 0.20183, "grad_norm": 1.3711382987272935, "learning_rate": 0.003, "loss": 4.0401, "step": 20183 }, { "epoch": 0.20184, "grad_norm": 1.0709976530119787, "learning_rate": 0.003, "loss": 4.0858, "step": 20184 }, { "epoch": 0.20185, "grad_norm": 1.3723070346701574, "learning_rate": 0.003, "loss": 4.0324, "step": 20185 }, { "epoch": 0.20186, "grad_norm": 1.528028266743247, "learning_rate": 0.003, "loss": 4.0602, "step": 20186 }, { "epoch": 0.20187, "grad_norm": 1.4037706732194764, "learning_rate": 0.003, "loss": 4.0522, "step": 20187 }, { "epoch": 0.20188, "grad_norm": 1.149409673743734, "learning_rate": 0.003, "loss": 4.0708, "step": 20188 }, { "epoch": 0.20189, "grad_norm": 1.2286043874357866, "learning_rate": 0.003, "loss": 4.0303, "step": 20189 }, { "epoch": 0.2019, "grad_norm": 1.1478341184973142, "learning_rate": 0.003, "loss": 4.0563, "step": 20190 }, { "epoch": 0.20191, "grad_norm": 1.2120327248254865, "learning_rate": 0.003, "loss": 4.0777, "step": 20191 }, { "epoch": 0.20192, "grad_norm": 1.1481050371361636, "learning_rate": 0.003, "loss": 4.0553, "step": 20192 }, { "epoch": 0.20193, "grad_norm": 1.2131783050485487, "learning_rate": 0.003, "loss": 4.0306, "step": 20193 }, { "epoch": 0.20194, "grad_norm": 1.1697391913516844, "learning_rate": 0.003, "loss": 4.0584, "step": 20194 }, { "epoch": 0.20195, "grad_norm": 1.4817698697937207, "learning_rate": 0.003, "loss": 4.1085, "step": 20195 }, { "epoch": 0.20196, "grad_norm": 1.0721068277211925, "learning_rate": 0.003, "loss": 4.0366, "step": 20196 }, { "epoch": 0.20197, "grad_norm": 1.5394627898345763, "learning_rate": 0.003, "loss": 4.0531, "step": 20197 }, { "epoch": 0.20198, "grad_norm": 1.1113258936770396, "learning_rate": 0.003, "loss": 4.026, "step": 20198 }, { "epoch": 0.20199, "grad_norm": 1.5233099851925598, "learning_rate": 0.003, "loss": 4.048, "step": 20199 }, { "epoch": 0.202, "grad_norm": 1.1817220138550921, "learning_rate": 0.003, "loss": 4.0769, "step": 20200 }, { "epoch": 0.20201, "grad_norm": 1.2445238084487324, "learning_rate": 0.003, "loss": 4.0457, "step": 20201 }, { "epoch": 0.20202, "grad_norm": 1.2647594216143536, "learning_rate": 0.003, "loss": 4.0947, "step": 20202 }, { "epoch": 0.20203, "grad_norm": 1.2384518932185162, "learning_rate": 0.003, "loss": 4.0493, "step": 20203 }, { "epoch": 0.20204, "grad_norm": 1.0317194075839051, "learning_rate": 0.003, "loss": 4.072, "step": 20204 }, { "epoch": 0.20205, "grad_norm": 1.323480590891624, "learning_rate": 0.003, "loss": 4.0664, "step": 20205 }, { "epoch": 0.20206, "grad_norm": 1.0463189815345117, "learning_rate": 0.003, "loss": 4.0054, "step": 20206 }, { "epoch": 0.20207, "grad_norm": 1.495579252481595, "learning_rate": 0.003, "loss": 4.045, "step": 20207 }, { "epoch": 0.20208, "grad_norm": 1.079861742292042, "learning_rate": 0.003, "loss": 4.0322, "step": 20208 }, { "epoch": 0.20209, "grad_norm": 1.2910326664187293, "learning_rate": 0.003, "loss": 4.0552, "step": 20209 }, { "epoch": 0.2021, "grad_norm": 1.4701760393593748, "learning_rate": 0.003, "loss": 4.0588, "step": 20210 }, { "epoch": 0.20211, "grad_norm": 1.3310843617338952, "learning_rate": 0.003, "loss": 4.0833, "step": 20211 }, { "epoch": 0.20212, "grad_norm": 1.5347552587814175, "learning_rate": 0.003, "loss": 4.0569, "step": 20212 }, { "epoch": 0.20213, "grad_norm": 1.1585200462266632, "learning_rate": 0.003, "loss": 4.0452, "step": 20213 }, { "epoch": 0.20214, "grad_norm": 1.3185943826230864, "learning_rate": 0.003, "loss": 4.0148, "step": 20214 }, { "epoch": 0.20215, "grad_norm": 1.3251462472306097, "learning_rate": 0.003, "loss": 4.0517, "step": 20215 }, { "epoch": 0.20216, "grad_norm": 1.1553789059246724, "learning_rate": 0.003, "loss": 4.0818, "step": 20216 }, { "epoch": 0.20217, "grad_norm": 1.0744917015208617, "learning_rate": 0.003, "loss": 4.0313, "step": 20217 }, { "epoch": 0.20218, "grad_norm": 1.5741468499145435, "learning_rate": 0.003, "loss": 4.0603, "step": 20218 }, { "epoch": 0.20219, "grad_norm": 0.991116419691916, "learning_rate": 0.003, "loss": 4.0322, "step": 20219 }, { "epoch": 0.2022, "grad_norm": 1.467538659804866, "learning_rate": 0.003, "loss": 4.0249, "step": 20220 }, { "epoch": 0.20221, "grad_norm": 1.3377664250726413, "learning_rate": 0.003, "loss": 4.0622, "step": 20221 }, { "epoch": 0.20222, "grad_norm": 1.3989307178459185, "learning_rate": 0.003, "loss": 4.0539, "step": 20222 }, { "epoch": 0.20223, "grad_norm": 1.1634816025495518, "learning_rate": 0.003, "loss": 4.0442, "step": 20223 }, { "epoch": 0.20224, "grad_norm": 1.4188224020522449, "learning_rate": 0.003, "loss": 4.0883, "step": 20224 }, { "epoch": 0.20225, "grad_norm": 1.0811208864402644, "learning_rate": 0.003, "loss": 4.0305, "step": 20225 }, { "epoch": 0.20226, "grad_norm": 1.4280355181481164, "learning_rate": 0.003, "loss": 4.0615, "step": 20226 }, { "epoch": 0.20227, "grad_norm": 1.067544445702645, "learning_rate": 0.003, "loss": 4.0328, "step": 20227 }, { "epoch": 0.20228, "grad_norm": 1.6110678887820848, "learning_rate": 0.003, "loss": 4.0356, "step": 20228 }, { "epoch": 0.20229, "grad_norm": 0.9769871544490712, "learning_rate": 0.003, "loss": 4.0413, "step": 20229 }, { "epoch": 0.2023, "grad_norm": 1.3872537306793618, "learning_rate": 0.003, "loss": 4.0436, "step": 20230 }, { "epoch": 0.20231, "grad_norm": 1.0661111881765468, "learning_rate": 0.003, "loss": 4.0754, "step": 20231 }, { "epoch": 0.20232, "grad_norm": 1.2983706866034253, "learning_rate": 0.003, "loss": 4.0199, "step": 20232 }, { "epoch": 0.20233, "grad_norm": 1.3006655842830865, "learning_rate": 0.003, "loss": 4.0691, "step": 20233 }, { "epoch": 0.20234, "grad_norm": 1.6780930215987617, "learning_rate": 0.003, "loss": 4.0716, "step": 20234 }, { "epoch": 0.20235, "grad_norm": 1.143017211264876, "learning_rate": 0.003, "loss": 4.0443, "step": 20235 }, { "epoch": 0.20236, "grad_norm": 1.369127090739487, "learning_rate": 0.003, "loss": 4.0585, "step": 20236 }, { "epoch": 0.20237, "grad_norm": 1.0208193547565565, "learning_rate": 0.003, "loss": 4.0492, "step": 20237 }, { "epoch": 0.20238, "grad_norm": 1.2853649168833825, "learning_rate": 0.003, "loss": 4.0466, "step": 20238 }, { "epoch": 0.20239, "grad_norm": 1.2039176355369394, "learning_rate": 0.003, "loss": 4.0435, "step": 20239 }, { "epoch": 0.2024, "grad_norm": 1.2742021486780266, "learning_rate": 0.003, "loss": 4.0442, "step": 20240 }, { "epoch": 0.20241, "grad_norm": 1.0981347079706447, "learning_rate": 0.003, "loss": 4.0622, "step": 20241 }, { "epoch": 0.20242, "grad_norm": 1.4868618409719494, "learning_rate": 0.003, "loss": 4.0734, "step": 20242 }, { "epoch": 0.20243, "grad_norm": 1.1534983644467036, "learning_rate": 0.003, "loss": 4.0155, "step": 20243 }, { "epoch": 0.20244, "grad_norm": 1.3505490920042413, "learning_rate": 0.003, "loss": 4.0775, "step": 20244 }, { "epoch": 0.20245, "grad_norm": 1.2754207396280954, "learning_rate": 0.003, "loss": 4.039, "step": 20245 }, { "epoch": 0.20246, "grad_norm": 1.2481516555763843, "learning_rate": 0.003, "loss": 4.0401, "step": 20246 }, { "epoch": 0.20247, "grad_norm": 1.3683820050551518, "learning_rate": 0.003, "loss": 4.0786, "step": 20247 }, { "epoch": 0.20248, "grad_norm": 1.3825827423130315, "learning_rate": 0.003, "loss": 4.0383, "step": 20248 }, { "epoch": 0.20249, "grad_norm": 1.188296243989267, "learning_rate": 0.003, "loss": 4.0604, "step": 20249 }, { "epoch": 0.2025, "grad_norm": 1.2789938257566154, "learning_rate": 0.003, "loss": 4.0452, "step": 20250 }, { "epoch": 0.20251, "grad_norm": 1.2262139492662278, "learning_rate": 0.003, "loss": 4.0338, "step": 20251 }, { "epoch": 0.20252, "grad_norm": 1.1547178266304627, "learning_rate": 0.003, "loss": 4.0594, "step": 20252 }, { "epoch": 0.20253, "grad_norm": 1.7121811575875046, "learning_rate": 0.003, "loss": 4.0509, "step": 20253 }, { "epoch": 0.20254, "grad_norm": 1.1515771332390023, "learning_rate": 0.003, "loss": 4.0956, "step": 20254 }, { "epoch": 0.20255, "grad_norm": 1.5168764501142065, "learning_rate": 0.003, "loss": 4.0466, "step": 20255 }, { "epoch": 0.20256, "grad_norm": 1.0447316920803935, "learning_rate": 0.003, "loss": 4.0581, "step": 20256 }, { "epoch": 0.20257, "grad_norm": 1.5409148718058376, "learning_rate": 0.003, "loss": 4.0407, "step": 20257 }, { "epoch": 0.20258, "grad_norm": 1.254626363776836, "learning_rate": 0.003, "loss": 4.0209, "step": 20258 }, { "epoch": 0.20259, "grad_norm": 1.3004260472547127, "learning_rate": 0.003, "loss": 4.0616, "step": 20259 }, { "epoch": 0.2026, "grad_norm": 1.425302205282692, "learning_rate": 0.003, "loss": 4.0616, "step": 20260 }, { "epoch": 0.20261, "grad_norm": 1.1133046134525415, "learning_rate": 0.003, "loss": 4.08, "step": 20261 }, { "epoch": 0.20262, "grad_norm": 1.2249565414123091, "learning_rate": 0.003, "loss": 4.0514, "step": 20262 }, { "epoch": 0.20263, "grad_norm": 1.2147479847044318, "learning_rate": 0.003, "loss": 4.0509, "step": 20263 }, { "epoch": 0.20264, "grad_norm": 1.2785029752457941, "learning_rate": 0.003, "loss": 4.0654, "step": 20264 }, { "epoch": 0.20265, "grad_norm": 1.356830248487893, "learning_rate": 0.003, "loss": 4.0233, "step": 20265 }, { "epoch": 0.20266, "grad_norm": 1.0981050254445741, "learning_rate": 0.003, "loss": 4.0339, "step": 20266 }, { "epoch": 0.20267, "grad_norm": 1.1983540241053046, "learning_rate": 0.003, "loss": 4.0528, "step": 20267 }, { "epoch": 0.20268, "grad_norm": 1.2823777515572727, "learning_rate": 0.003, "loss": 4.0644, "step": 20268 }, { "epoch": 0.20269, "grad_norm": 1.1623789358938565, "learning_rate": 0.003, "loss": 4.0445, "step": 20269 }, { "epoch": 0.2027, "grad_norm": 1.3864113772264623, "learning_rate": 0.003, "loss": 4.0547, "step": 20270 }, { "epoch": 0.20271, "grad_norm": 1.0095425180500388, "learning_rate": 0.003, "loss": 4.0686, "step": 20271 }, { "epoch": 0.20272, "grad_norm": 1.4567747087753842, "learning_rate": 0.003, "loss": 4.0591, "step": 20272 }, { "epoch": 0.20273, "grad_norm": 1.2986456339209054, "learning_rate": 0.003, "loss": 4.0377, "step": 20273 }, { "epoch": 0.20274, "grad_norm": 1.5445124404317825, "learning_rate": 0.003, "loss": 4.0749, "step": 20274 }, { "epoch": 0.20275, "grad_norm": 1.1873407876828512, "learning_rate": 0.003, "loss": 4.049, "step": 20275 }, { "epoch": 0.20276, "grad_norm": 1.2862597617477896, "learning_rate": 0.003, "loss": 4.0616, "step": 20276 }, { "epoch": 0.20277, "grad_norm": 1.0957498635637768, "learning_rate": 0.003, "loss": 4.0744, "step": 20277 }, { "epoch": 0.20278, "grad_norm": 1.4062364206465463, "learning_rate": 0.003, "loss": 4.0252, "step": 20278 }, { "epoch": 0.20279, "grad_norm": 1.2413485576763343, "learning_rate": 0.003, "loss": 4.0521, "step": 20279 }, { "epoch": 0.2028, "grad_norm": 1.384671038415917, "learning_rate": 0.003, "loss": 4.013, "step": 20280 }, { "epoch": 0.20281, "grad_norm": 1.0885068965446765, "learning_rate": 0.003, "loss": 4.0302, "step": 20281 }, { "epoch": 0.20282, "grad_norm": 1.3977988221248578, "learning_rate": 0.003, "loss": 4.0562, "step": 20282 }, { "epoch": 0.20283, "grad_norm": 1.1067110693550775, "learning_rate": 0.003, "loss": 4.0697, "step": 20283 }, { "epoch": 0.20284, "grad_norm": 1.3549528720449708, "learning_rate": 0.003, "loss": 4.0519, "step": 20284 }, { "epoch": 0.20285, "grad_norm": 1.1860419143984546, "learning_rate": 0.003, "loss": 4.0455, "step": 20285 }, { "epoch": 0.20286, "grad_norm": 1.2059777955000675, "learning_rate": 0.003, "loss": 4.0414, "step": 20286 }, { "epoch": 0.20287, "grad_norm": 1.2452324495460783, "learning_rate": 0.003, "loss": 4.0225, "step": 20287 }, { "epoch": 0.20288, "grad_norm": 1.2008982965029176, "learning_rate": 0.003, "loss": 4.0492, "step": 20288 }, { "epoch": 0.20289, "grad_norm": 1.2053420925394356, "learning_rate": 0.003, "loss": 4.0635, "step": 20289 }, { "epoch": 0.2029, "grad_norm": 1.1930940105972596, "learning_rate": 0.003, "loss": 4.0625, "step": 20290 }, { "epoch": 0.20291, "grad_norm": 1.1511373172127057, "learning_rate": 0.003, "loss": 4.0294, "step": 20291 }, { "epoch": 0.20292, "grad_norm": 1.5235344977033483, "learning_rate": 0.003, "loss": 4.0542, "step": 20292 }, { "epoch": 0.20293, "grad_norm": 1.0133916595069772, "learning_rate": 0.003, "loss": 4.0574, "step": 20293 }, { "epoch": 0.20294, "grad_norm": 1.449860175797969, "learning_rate": 0.003, "loss": 4.0323, "step": 20294 }, { "epoch": 0.20295, "grad_norm": 1.166260469765599, "learning_rate": 0.003, "loss": 4.0577, "step": 20295 }, { "epoch": 0.20296, "grad_norm": 1.7094333232663563, "learning_rate": 0.003, "loss": 4.0392, "step": 20296 }, { "epoch": 0.20297, "grad_norm": 1.0802573259244068, "learning_rate": 0.003, "loss": 4.0609, "step": 20297 }, { "epoch": 0.20298, "grad_norm": 1.3950396969201893, "learning_rate": 0.003, "loss": 4.0573, "step": 20298 }, { "epoch": 0.20299, "grad_norm": 1.2776938243492255, "learning_rate": 0.003, "loss": 4.0544, "step": 20299 }, { "epoch": 0.203, "grad_norm": 1.1235479306361127, "learning_rate": 0.003, "loss": 4.0677, "step": 20300 }, { "epoch": 0.20301, "grad_norm": 1.3193150170315688, "learning_rate": 0.003, "loss": 4.0501, "step": 20301 }, { "epoch": 0.20302, "grad_norm": 1.3359995387119015, "learning_rate": 0.003, "loss": 4.0655, "step": 20302 }, { "epoch": 0.20303, "grad_norm": 1.3579836294915681, "learning_rate": 0.003, "loss": 4.0744, "step": 20303 }, { "epoch": 0.20304, "grad_norm": 1.3162768399519815, "learning_rate": 0.003, "loss": 4.0432, "step": 20304 }, { "epoch": 0.20305, "grad_norm": 1.3250263807513758, "learning_rate": 0.003, "loss": 4.0713, "step": 20305 }, { "epoch": 0.20306, "grad_norm": 1.2816788958797483, "learning_rate": 0.003, "loss": 4.0295, "step": 20306 }, { "epoch": 0.20307, "grad_norm": 1.326718838954274, "learning_rate": 0.003, "loss": 4.0595, "step": 20307 }, { "epoch": 0.20308, "grad_norm": 1.592020573675577, "learning_rate": 0.003, "loss": 4.0523, "step": 20308 }, { "epoch": 0.20309, "grad_norm": 1.1498016645153961, "learning_rate": 0.003, "loss": 4.0414, "step": 20309 }, { "epoch": 0.2031, "grad_norm": 1.3609823524921811, "learning_rate": 0.003, "loss": 4.0396, "step": 20310 }, { "epoch": 0.20311, "grad_norm": 1.0477184836568727, "learning_rate": 0.003, "loss": 4.0405, "step": 20311 }, { "epoch": 0.20312, "grad_norm": 1.696289530479139, "learning_rate": 0.003, "loss": 4.036, "step": 20312 }, { "epoch": 0.20313, "grad_norm": 1.3710431911272913, "learning_rate": 0.003, "loss": 4.0579, "step": 20313 }, { "epoch": 0.20314, "grad_norm": 1.089592730722583, "learning_rate": 0.003, "loss": 4.0584, "step": 20314 }, { "epoch": 0.20315, "grad_norm": 1.2866317286088467, "learning_rate": 0.003, "loss": 4.061, "step": 20315 }, { "epoch": 0.20316, "grad_norm": 1.1346409696551878, "learning_rate": 0.003, "loss": 4.035, "step": 20316 }, { "epoch": 0.20317, "grad_norm": 1.3248740676276995, "learning_rate": 0.003, "loss": 4.0451, "step": 20317 }, { "epoch": 0.20318, "grad_norm": 1.1648010683216785, "learning_rate": 0.003, "loss": 4.0474, "step": 20318 }, { "epoch": 0.20319, "grad_norm": 1.2382848380382208, "learning_rate": 0.003, "loss": 4.0571, "step": 20319 }, { "epoch": 0.2032, "grad_norm": 1.1936347470703148, "learning_rate": 0.003, "loss": 4.0452, "step": 20320 }, { "epoch": 0.20321, "grad_norm": 1.5168939617426216, "learning_rate": 0.003, "loss": 4.0565, "step": 20321 }, { "epoch": 0.20322, "grad_norm": 1.243897914432, "learning_rate": 0.003, "loss": 4.0457, "step": 20322 }, { "epoch": 0.20323, "grad_norm": 1.335384991277468, "learning_rate": 0.003, "loss": 4.0691, "step": 20323 }, { "epoch": 0.20324, "grad_norm": 1.2671372121627433, "learning_rate": 0.003, "loss": 4.0597, "step": 20324 }, { "epoch": 0.20325, "grad_norm": 1.3591566627374516, "learning_rate": 0.003, "loss": 4.0415, "step": 20325 }, { "epoch": 0.20326, "grad_norm": 1.1685880487870943, "learning_rate": 0.003, "loss": 4.0059, "step": 20326 }, { "epoch": 0.20327, "grad_norm": 1.2474832230538342, "learning_rate": 0.003, "loss": 4.058, "step": 20327 }, { "epoch": 0.20328, "grad_norm": 1.218543277114442, "learning_rate": 0.003, "loss": 4.0687, "step": 20328 }, { "epoch": 0.20329, "grad_norm": 1.3341045186550506, "learning_rate": 0.003, "loss": 4.0252, "step": 20329 }, { "epoch": 0.2033, "grad_norm": 1.267574594995304, "learning_rate": 0.003, "loss": 4.0524, "step": 20330 }, { "epoch": 0.20331, "grad_norm": 1.2795384215934296, "learning_rate": 0.003, "loss": 4.0411, "step": 20331 }, { "epoch": 0.20332, "grad_norm": 1.3637715732263345, "learning_rate": 0.003, "loss": 4.0384, "step": 20332 }, { "epoch": 0.20333, "grad_norm": 1.358986591353964, "learning_rate": 0.003, "loss": 4.0663, "step": 20333 }, { "epoch": 0.20334, "grad_norm": 1.1231116626020192, "learning_rate": 0.003, "loss": 4.0348, "step": 20334 }, { "epoch": 0.20335, "grad_norm": 1.13442333370607, "learning_rate": 0.003, "loss": 4.0136, "step": 20335 }, { "epoch": 0.20336, "grad_norm": 1.3593483790810181, "learning_rate": 0.003, "loss": 4.0478, "step": 20336 }, { "epoch": 0.20337, "grad_norm": 1.2023125147943492, "learning_rate": 0.003, "loss": 4.0394, "step": 20337 }, { "epoch": 0.20338, "grad_norm": 1.5018889719909159, "learning_rate": 0.003, "loss": 4.0518, "step": 20338 }, { "epoch": 0.20339, "grad_norm": 1.1479734371080672, "learning_rate": 0.003, "loss": 4.0555, "step": 20339 }, { "epoch": 0.2034, "grad_norm": 1.3136111992465354, "learning_rate": 0.003, "loss": 4.0775, "step": 20340 }, { "epoch": 0.20341, "grad_norm": 1.3910643534364964, "learning_rate": 0.003, "loss": 4.0315, "step": 20341 }, { "epoch": 0.20342, "grad_norm": 1.1428898175511215, "learning_rate": 0.003, "loss": 4.0197, "step": 20342 }, { "epoch": 0.20343, "grad_norm": 1.4647864618752329, "learning_rate": 0.003, "loss": 4.0659, "step": 20343 }, { "epoch": 0.20344, "grad_norm": 1.1926491616247035, "learning_rate": 0.003, "loss": 4.0417, "step": 20344 }, { "epoch": 0.20345, "grad_norm": 1.246382195862455, "learning_rate": 0.003, "loss": 4.0421, "step": 20345 }, { "epoch": 0.20346, "grad_norm": 1.2274302854273504, "learning_rate": 0.003, "loss": 4.0124, "step": 20346 }, { "epoch": 0.20347, "grad_norm": 1.2504913828412314, "learning_rate": 0.003, "loss": 4.0643, "step": 20347 }, { "epoch": 0.20348, "grad_norm": 1.372360129836187, "learning_rate": 0.003, "loss": 4.0628, "step": 20348 }, { "epoch": 0.20349, "grad_norm": 1.289719034252838, "learning_rate": 0.003, "loss": 4.0589, "step": 20349 }, { "epoch": 0.2035, "grad_norm": 1.6733341302170084, "learning_rate": 0.003, "loss": 4.0631, "step": 20350 }, { "epoch": 0.20351, "grad_norm": 1.2187008495803562, "learning_rate": 0.003, "loss": 4.0328, "step": 20351 }, { "epoch": 0.20352, "grad_norm": 1.3169621407501653, "learning_rate": 0.003, "loss": 4.0513, "step": 20352 }, { "epoch": 0.20353, "grad_norm": 1.4691001197294091, "learning_rate": 0.003, "loss": 4.0525, "step": 20353 }, { "epoch": 0.20354, "grad_norm": 1.0503063097982503, "learning_rate": 0.003, "loss": 4.0267, "step": 20354 }, { "epoch": 0.20355, "grad_norm": 1.3117587120224454, "learning_rate": 0.003, "loss": 4.0526, "step": 20355 }, { "epoch": 0.20356, "grad_norm": 1.1385371164477698, "learning_rate": 0.003, "loss": 4.0301, "step": 20356 }, { "epoch": 0.20357, "grad_norm": 1.3306978257381137, "learning_rate": 0.003, "loss": 4.0518, "step": 20357 }, { "epoch": 0.20358, "grad_norm": 1.435511112935323, "learning_rate": 0.003, "loss": 4.0666, "step": 20358 }, { "epoch": 0.20359, "grad_norm": 1.2726098866315974, "learning_rate": 0.003, "loss": 4.0365, "step": 20359 }, { "epoch": 0.2036, "grad_norm": 1.4639642814763292, "learning_rate": 0.003, "loss": 4.0508, "step": 20360 }, { "epoch": 0.20361, "grad_norm": 1.2133872583870144, "learning_rate": 0.003, "loss": 4.0398, "step": 20361 }, { "epoch": 0.20362, "grad_norm": 1.3707965450986, "learning_rate": 0.003, "loss": 4.0472, "step": 20362 }, { "epoch": 0.20363, "grad_norm": 1.137050348760085, "learning_rate": 0.003, "loss": 4.0073, "step": 20363 }, { "epoch": 0.20364, "grad_norm": 1.2245817580588365, "learning_rate": 0.003, "loss": 4.0455, "step": 20364 }, { "epoch": 0.20365, "grad_norm": 1.2015734355515697, "learning_rate": 0.003, "loss": 4.0455, "step": 20365 }, { "epoch": 0.20366, "grad_norm": 1.2583926218200907, "learning_rate": 0.003, "loss": 4.051, "step": 20366 }, { "epoch": 0.20367, "grad_norm": 1.137470719888171, "learning_rate": 0.003, "loss": 4.0509, "step": 20367 }, { "epoch": 0.20368, "grad_norm": 1.412201035519502, "learning_rate": 0.003, "loss": 4.0365, "step": 20368 }, { "epoch": 0.20369, "grad_norm": 1.3518580392474349, "learning_rate": 0.003, "loss": 4.0726, "step": 20369 }, { "epoch": 0.2037, "grad_norm": 1.5867531436060096, "learning_rate": 0.003, "loss": 4.0786, "step": 20370 }, { "epoch": 0.20371, "grad_norm": 1.0850388843692231, "learning_rate": 0.003, "loss": 4.0454, "step": 20371 }, { "epoch": 0.20372, "grad_norm": 1.2594058474456522, "learning_rate": 0.003, "loss": 4.0262, "step": 20372 }, { "epoch": 0.20373, "grad_norm": 1.132405443425768, "learning_rate": 0.003, "loss": 4.0645, "step": 20373 }, { "epoch": 0.20374, "grad_norm": 1.2085162478440552, "learning_rate": 0.003, "loss": 4.0558, "step": 20374 }, { "epoch": 0.20375, "grad_norm": 1.3309587794895381, "learning_rate": 0.003, "loss": 4.0478, "step": 20375 }, { "epoch": 0.20376, "grad_norm": 1.518814487811154, "learning_rate": 0.003, "loss": 4.0486, "step": 20376 }, { "epoch": 0.20377, "grad_norm": 1.1174678164788465, "learning_rate": 0.003, "loss": 4.0674, "step": 20377 }, { "epoch": 0.20378, "grad_norm": 1.5506355617390695, "learning_rate": 0.003, "loss": 4.0485, "step": 20378 }, { "epoch": 0.20379, "grad_norm": 1.066648758046894, "learning_rate": 0.003, "loss": 4.0729, "step": 20379 }, { "epoch": 0.2038, "grad_norm": 1.2907516251263023, "learning_rate": 0.003, "loss": 4.0552, "step": 20380 }, { "epoch": 0.20381, "grad_norm": 1.2177344809594945, "learning_rate": 0.003, "loss": 4.0157, "step": 20381 }, { "epoch": 0.20382, "grad_norm": 1.4399129113062175, "learning_rate": 0.003, "loss": 4.0396, "step": 20382 }, { "epoch": 0.20383, "grad_norm": 1.139629489947086, "learning_rate": 0.003, "loss": 4.0643, "step": 20383 }, { "epoch": 0.20384, "grad_norm": 1.3217997643163781, "learning_rate": 0.003, "loss": 4.0797, "step": 20384 }, { "epoch": 0.20385, "grad_norm": 1.2047773926287186, "learning_rate": 0.003, "loss": 4.0565, "step": 20385 }, { "epoch": 0.20386, "grad_norm": 1.1547469008316145, "learning_rate": 0.003, "loss": 4.0463, "step": 20386 }, { "epoch": 0.20387, "grad_norm": 1.3116613988192434, "learning_rate": 0.003, "loss": 4.0603, "step": 20387 }, { "epoch": 0.20388, "grad_norm": 1.1018342578418183, "learning_rate": 0.003, "loss": 4.0313, "step": 20388 }, { "epoch": 0.20389, "grad_norm": 1.2453373265547718, "learning_rate": 0.003, "loss": 4.0277, "step": 20389 }, { "epoch": 0.2039, "grad_norm": 1.2475157243348953, "learning_rate": 0.003, "loss": 4.0719, "step": 20390 }, { "epoch": 0.20391, "grad_norm": 1.3824501706722891, "learning_rate": 0.003, "loss": 4.0619, "step": 20391 }, { "epoch": 0.20392, "grad_norm": 1.36766174684596, "learning_rate": 0.003, "loss": 4.0813, "step": 20392 }, { "epoch": 0.20393, "grad_norm": 1.234481097722599, "learning_rate": 0.003, "loss": 4.0648, "step": 20393 }, { "epoch": 0.20394, "grad_norm": 1.150615707215384, "learning_rate": 0.003, "loss": 4.0475, "step": 20394 }, { "epoch": 0.20395, "grad_norm": 1.4406174400976866, "learning_rate": 0.003, "loss": 4.0773, "step": 20395 }, { "epoch": 0.20396, "grad_norm": 1.045138828697066, "learning_rate": 0.003, "loss": 4.0148, "step": 20396 }, { "epoch": 0.20397, "grad_norm": 1.643939395888882, "learning_rate": 0.003, "loss": 4.0887, "step": 20397 }, { "epoch": 0.20398, "grad_norm": 0.9437710813220958, "learning_rate": 0.003, "loss": 4.0633, "step": 20398 }, { "epoch": 0.20399, "grad_norm": 1.5709103193037224, "learning_rate": 0.003, "loss": 4.0706, "step": 20399 }, { "epoch": 0.204, "grad_norm": 1.3603139040282624, "learning_rate": 0.003, "loss": 4.0938, "step": 20400 }, { "epoch": 0.20401, "grad_norm": 1.258317441464475, "learning_rate": 0.003, "loss": 4.0513, "step": 20401 }, { "epoch": 0.20402, "grad_norm": 1.2323831932165392, "learning_rate": 0.003, "loss": 4.0513, "step": 20402 }, { "epoch": 0.20403, "grad_norm": 1.1613228064934027, "learning_rate": 0.003, "loss": 4.0366, "step": 20403 }, { "epoch": 0.20404, "grad_norm": 1.1157951639671577, "learning_rate": 0.003, "loss": 4.0533, "step": 20404 }, { "epoch": 0.20405, "grad_norm": 1.117386643961847, "learning_rate": 0.003, "loss": 4.0342, "step": 20405 }, { "epoch": 0.20406, "grad_norm": 1.142677385878031, "learning_rate": 0.003, "loss": 4.0501, "step": 20406 }, { "epoch": 0.20407, "grad_norm": 1.4502256122374817, "learning_rate": 0.003, "loss": 4.0743, "step": 20407 }, { "epoch": 0.20408, "grad_norm": 1.2028422026661283, "learning_rate": 0.003, "loss": 4.0562, "step": 20408 }, { "epoch": 0.20409, "grad_norm": 1.5946736655945457, "learning_rate": 0.003, "loss": 4.0563, "step": 20409 }, { "epoch": 0.2041, "grad_norm": 1.1774295096789895, "learning_rate": 0.003, "loss": 4.0455, "step": 20410 }, { "epoch": 0.20411, "grad_norm": 1.3394453182476518, "learning_rate": 0.003, "loss": 4.0331, "step": 20411 }, { "epoch": 0.20412, "grad_norm": 1.2014049871896877, "learning_rate": 0.003, "loss": 4.0205, "step": 20412 }, { "epoch": 0.20413, "grad_norm": 1.1046436883020136, "learning_rate": 0.003, "loss": 4.0474, "step": 20413 }, { "epoch": 0.20414, "grad_norm": 1.5143551635735701, "learning_rate": 0.003, "loss": 4.0395, "step": 20414 }, { "epoch": 0.20415, "grad_norm": 1.2081076575150453, "learning_rate": 0.003, "loss": 4.0281, "step": 20415 }, { "epoch": 0.20416, "grad_norm": 1.3398110451492105, "learning_rate": 0.003, "loss": 4.0748, "step": 20416 }, { "epoch": 0.20417, "grad_norm": 1.0986812247155366, "learning_rate": 0.003, "loss": 4.0391, "step": 20417 }, { "epoch": 0.20418, "grad_norm": 1.5545693709449835, "learning_rate": 0.003, "loss": 4.046, "step": 20418 }, { "epoch": 0.20419, "grad_norm": 1.141388079534894, "learning_rate": 0.003, "loss": 4.0524, "step": 20419 }, { "epoch": 0.2042, "grad_norm": 1.3293185978350845, "learning_rate": 0.003, "loss": 4.038, "step": 20420 }, { "epoch": 0.20421, "grad_norm": 1.1445835926864036, "learning_rate": 0.003, "loss": 4.0502, "step": 20421 }, { "epoch": 0.20422, "grad_norm": 1.7085839702569687, "learning_rate": 0.003, "loss": 4.0662, "step": 20422 }, { "epoch": 0.20423, "grad_norm": 1.129517781431287, "learning_rate": 0.003, "loss": 4.0392, "step": 20423 }, { "epoch": 0.20424, "grad_norm": 1.4261103639533557, "learning_rate": 0.003, "loss": 4.0451, "step": 20424 }, { "epoch": 0.20425, "grad_norm": 1.1996507960900666, "learning_rate": 0.003, "loss": 4.053, "step": 20425 }, { "epoch": 0.20426, "grad_norm": 1.2758568684985025, "learning_rate": 0.003, "loss": 4.0518, "step": 20426 }, { "epoch": 0.20427, "grad_norm": 1.1207260094818572, "learning_rate": 0.003, "loss": 4.0567, "step": 20427 }, { "epoch": 0.20428, "grad_norm": 1.3128081076573406, "learning_rate": 0.003, "loss": 4.0604, "step": 20428 }, { "epoch": 0.20429, "grad_norm": 1.3320582459592007, "learning_rate": 0.003, "loss": 4.0207, "step": 20429 }, { "epoch": 0.2043, "grad_norm": 1.2680088635640516, "learning_rate": 0.003, "loss": 4.079, "step": 20430 }, { "epoch": 0.20431, "grad_norm": 1.2788587267600544, "learning_rate": 0.003, "loss": 4.0453, "step": 20431 }, { "epoch": 0.20432, "grad_norm": 1.2588374516855552, "learning_rate": 0.003, "loss": 4.0446, "step": 20432 }, { "epoch": 0.20433, "grad_norm": 1.2353153784507898, "learning_rate": 0.003, "loss": 4.0454, "step": 20433 }, { "epoch": 0.20434, "grad_norm": 1.229581316148024, "learning_rate": 0.003, "loss": 4.0602, "step": 20434 }, { "epoch": 0.20435, "grad_norm": 1.4257389393320417, "learning_rate": 0.003, "loss": 4.0603, "step": 20435 }, { "epoch": 0.20436, "grad_norm": 1.2496937224156124, "learning_rate": 0.003, "loss": 4.033, "step": 20436 }, { "epoch": 0.20437, "grad_norm": 1.1178812642826619, "learning_rate": 0.003, "loss": 4.0411, "step": 20437 }, { "epoch": 0.20438, "grad_norm": 1.241905961365824, "learning_rate": 0.003, "loss": 4.0304, "step": 20438 }, { "epoch": 0.20439, "grad_norm": 1.2702669223841556, "learning_rate": 0.003, "loss": 4.0556, "step": 20439 }, { "epoch": 0.2044, "grad_norm": 1.3862685914354573, "learning_rate": 0.003, "loss": 4.0434, "step": 20440 }, { "epoch": 0.20441, "grad_norm": 1.2178800769182598, "learning_rate": 0.003, "loss": 4.0479, "step": 20441 }, { "epoch": 0.20442, "grad_norm": 1.2758757529344358, "learning_rate": 0.003, "loss": 4.0493, "step": 20442 }, { "epoch": 0.20443, "grad_norm": 1.2717016156474636, "learning_rate": 0.003, "loss": 4.0561, "step": 20443 }, { "epoch": 0.20444, "grad_norm": 1.1874295826494963, "learning_rate": 0.003, "loss": 4.047, "step": 20444 }, { "epoch": 0.20445, "grad_norm": 1.273127108312738, "learning_rate": 0.003, "loss": 4.024, "step": 20445 }, { "epoch": 0.20446, "grad_norm": 1.3033196593525216, "learning_rate": 0.003, "loss": 4.0466, "step": 20446 }, { "epoch": 0.20447, "grad_norm": 1.3573380259627446, "learning_rate": 0.003, "loss": 4.0733, "step": 20447 }, { "epoch": 0.20448, "grad_norm": 1.3280882364089757, "learning_rate": 0.003, "loss": 4.0502, "step": 20448 }, { "epoch": 0.20449, "grad_norm": 1.1733062924191375, "learning_rate": 0.003, "loss": 4.0329, "step": 20449 }, { "epoch": 0.2045, "grad_norm": 1.2328609540930164, "learning_rate": 0.003, "loss": 4.0498, "step": 20450 }, { "epoch": 0.20451, "grad_norm": 1.3820397298453064, "learning_rate": 0.003, "loss": 4.0233, "step": 20451 }, { "epoch": 0.20452, "grad_norm": 1.1611758200547153, "learning_rate": 0.003, "loss": 4.0218, "step": 20452 }, { "epoch": 0.20453, "grad_norm": 1.1965128389229713, "learning_rate": 0.003, "loss": 4.0328, "step": 20453 }, { "epoch": 0.20454, "grad_norm": 1.1817630791136045, "learning_rate": 0.003, "loss": 4.0622, "step": 20454 }, { "epoch": 0.20455, "grad_norm": 1.702877011445011, "learning_rate": 0.003, "loss": 4.0541, "step": 20455 }, { "epoch": 0.20456, "grad_norm": 1.194996344264795, "learning_rate": 0.003, "loss": 4.0152, "step": 20456 }, { "epoch": 0.20457, "grad_norm": 1.5014411928263622, "learning_rate": 0.003, "loss": 4.0335, "step": 20457 }, { "epoch": 0.20458, "grad_norm": 1.351084444433871, "learning_rate": 0.003, "loss": 4.0362, "step": 20458 }, { "epoch": 0.20459, "grad_norm": 1.3053422526345337, "learning_rate": 0.003, "loss": 4.0268, "step": 20459 }, { "epoch": 0.2046, "grad_norm": 1.1632725678267068, "learning_rate": 0.003, "loss": 4.0381, "step": 20460 }, { "epoch": 0.20461, "grad_norm": 1.4148776664413345, "learning_rate": 0.003, "loss": 4.0525, "step": 20461 }, { "epoch": 0.20462, "grad_norm": 1.0753270302692137, "learning_rate": 0.003, "loss": 4.0431, "step": 20462 }, { "epoch": 0.20463, "grad_norm": 1.244881027428857, "learning_rate": 0.003, "loss": 4.0336, "step": 20463 }, { "epoch": 0.20464, "grad_norm": 1.605785444033807, "learning_rate": 0.003, "loss": 4.0828, "step": 20464 }, { "epoch": 0.20465, "grad_norm": 1.0265654918416915, "learning_rate": 0.003, "loss": 4.0507, "step": 20465 }, { "epoch": 0.20466, "grad_norm": 1.4597953303483058, "learning_rate": 0.003, "loss": 4.0803, "step": 20466 }, { "epoch": 0.20467, "grad_norm": 1.4019596632247724, "learning_rate": 0.003, "loss": 4.0466, "step": 20467 }, { "epoch": 0.20468, "grad_norm": 1.1972777119646112, "learning_rate": 0.003, "loss": 4.061, "step": 20468 }, { "epoch": 0.20469, "grad_norm": 1.1827608046664349, "learning_rate": 0.003, "loss": 4.0582, "step": 20469 }, { "epoch": 0.2047, "grad_norm": 1.2217603760295772, "learning_rate": 0.003, "loss": 4.0303, "step": 20470 }, { "epoch": 0.20471, "grad_norm": 1.1165912816299286, "learning_rate": 0.003, "loss": 4.0312, "step": 20471 }, { "epoch": 0.20472, "grad_norm": 1.3192129060455589, "learning_rate": 0.003, "loss": 4.0578, "step": 20472 }, { "epoch": 0.20473, "grad_norm": 1.220922299063871, "learning_rate": 0.003, "loss": 4.0536, "step": 20473 }, { "epoch": 0.20474, "grad_norm": 1.3695141554657906, "learning_rate": 0.003, "loss": 4.0541, "step": 20474 }, { "epoch": 0.20475, "grad_norm": 1.2980958215607803, "learning_rate": 0.003, "loss": 4.0693, "step": 20475 }, { "epoch": 0.20476, "grad_norm": 1.1630131801927541, "learning_rate": 0.003, "loss": 4.058, "step": 20476 }, { "epoch": 0.20477, "grad_norm": 1.3306418530414286, "learning_rate": 0.003, "loss": 4.0258, "step": 20477 }, { "epoch": 0.20478, "grad_norm": 1.292662469015786, "learning_rate": 0.003, "loss": 4.035, "step": 20478 }, { "epoch": 0.20479, "grad_norm": 1.429433414427323, "learning_rate": 0.003, "loss": 4.0836, "step": 20479 }, { "epoch": 0.2048, "grad_norm": 1.259234210940857, "learning_rate": 0.003, "loss": 4.0579, "step": 20480 }, { "epoch": 0.20481, "grad_norm": 1.6757163183846884, "learning_rate": 0.003, "loss": 4.0061, "step": 20481 }, { "epoch": 0.20482, "grad_norm": 1.0300496487687896, "learning_rate": 0.003, "loss": 4.0388, "step": 20482 }, { "epoch": 0.20483, "grad_norm": 1.4611030214175802, "learning_rate": 0.003, "loss": 4.0628, "step": 20483 }, { "epoch": 0.20484, "grad_norm": 1.1961501734660809, "learning_rate": 0.003, "loss": 4.0707, "step": 20484 }, { "epoch": 0.20485, "grad_norm": 1.6365170923707129, "learning_rate": 0.003, "loss": 4.0866, "step": 20485 }, { "epoch": 0.20486, "grad_norm": 0.9255837236365103, "learning_rate": 0.003, "loss": 4.0444, "step": 20486 }, { "epoch": 0.20487, "grad_norm": 1.2907347708489552, "learning_rate": 0.003, "loss": 4.0653, "step": 20487 }, { "epoch": 0.20488, "grad_norm": 1.211527144515664, "learning_rate": 0.003, "loss": 4.0621, "step": 20488 }, { "epoch": 0.20489, "grad_norm": 1.2567952590179416, "learning_rate": 0.003, "loss": 4.0563, "step": 20489 }, { "epoch": 0.2049, "grad_norm": 1.2447501540709187, "learning_rate": 0.003, "loss": 4.0585, "step": 20490 }, { "epoch": 0.20491, "grad_norm": 1.5771043171932784, "learning_rate": 0.003, "loss": 4.0662, "step": 20491 }, { "epoch": 0.20492, "grad_norm": 1.2944710811943263, "learning_rate": 0.003, "loss": 4.0495, "step": 20492 }, { "epoch": 0.20493, "grad_norm": 1.1536166957557195, "learning_rate": 0.003, "loss": 4.0609, "step": 20493 }, { "epoch": 0.20494, "grad_norm": 1.230412031789269, "learning_rate": 0.003, "loss": 4.0517, "step": 20494 }, { "epoch": 0.20495, "grad_norm": 1.4309706890344465, "learning_rate": 0.003, "loss": 4.0274, "step": 20495 }, { "epoch": 0.20496, "grad_norm": 1.239470233245192, "learning_rate": 0.003, "loss": 4.0464, "step": 20496 }, { "epoch": 0.20497, "grad_norm": 1.3717068737602895, "learning_rate": 0.003, "loss": 4.055, "step": 20497 }, { "epoch": 0.20498, "grad_norm": 1.0206873885980308, "learning_rate": 0.003, "loss": 4.0533, "step": 20498 }, { "epoch": 0.20499, "grad_norm": 1.4824021249275585, "learning_rate": 0.003, "loss": 4.084, "step": 20499 }, { "epoch": 0.205, "grad_norm": 1.0350841616123798, "learning_rate": 0.003, "loss": 4.0615, "step": 20500 }, { "epoch": 0.20501, "grad_norm": 1.3121771019769726, "learning_rate": 0.003, "loss": 4.0616, "step": 20501 }, { "epoch": 0.20502, "grad_norm": 1.1708296290415547, "learning_rate": 0.003, "loss": 4.0434, "step": 20502 }, { "epoch": 0.20503, "grad_norm": 1.3824308379400518, "learning_rate": 0.003, "loss": 4.0512, "step": 20503 }, { "epoch": 0.20504, "grad_norm": 1.2070001077828993, "learning_rate": 0.003, "loss": 4.0713, "step": 20504 }, { "epoch": 0.20505, "grad_norm": 1.4755787704240584, "learning_rate": 0.003, "loss": 4.0385, "step": 20505 }, { "epoch": 0.20506, "grad_norm": 1.2193032792463148, "learning_rate": 0.003, "loss": 4.1092, "step": 20506 }, { "epoch": 0.20507, "grad_norm": 1.1078265884640162, "learning_rate": 0.003, "loss": 4.0679, "step": 20507 }, { "epoch": 0.20508, "grad_norm": 1.5052162221790877, "learning_rate": 0.003, "loss": 4.0609, "step": 20508 }, { "epoch": 0.20509, "grad_norm": 1.1666948310567067, "learning_rate": 0.003, "loss": 4.0752, "step": 20509 }, { "epoch": 0.2051, "grad_norm": 1.4079526417493138, "learning_rate": 0.003, "loss": 4.0536, "step": 20510 }, { "epoch": 0.20511, "grad_norm": 1.1709527819746652, "learning_rate": 0.003, "loss": 4.0528, "step": 20511 }, { "epoch": 0.20512, "grad_norm": 1.3605758979989855, "learning_rate": 0.003, "loss": 4.0296, "step": 20512 }, { "epoch": 0.20513, "grad_norm": 1.2534739820163656, "learning_rate": 0.003, "loss": 4.0761, "step": 20513 }, { "epoch": 0.20514, "grad_norm": 1.1809491957128397, "learning_rate": 0.003, "loss": 4.0542, "step": 20514 }, { "epoch": 0.20515, "grad_norm": 1.3620554584835562, "learning_rate": 0.003, "loss": 4.056, "step": 20515 }, { "epoch": 0.20516, "grad_norm": 1.296887173066892, "learning_rate": 0.003, "loss": 4.0306, "step": 20516 }, { "epoch": 0.20517, "grad_norm": 1.318459153043808, "learning_rate": 0.003, "loss": 4.064, "step": 20517 }, { "epoch": 0.20518, "grad_norm": 1.4895743766572598, "learning_rate": 0.003, "loss": 4.0394, "step": 20518 }, { "epoch": 0.20519, "grad_norm": 1.332194058789682, "learning_rate": 0.003, "loss": 4.0625, "step": 20519 }, { "epoch": 0.2052, "grad_norm": 1.3867458319313875, "learning_rate": 0.003, "loss": 4.0463, "step": 20520 }, { "epoch": 0.20521, "grad_norm": 1.1098151444373003, "learning_rate": 0.003, "loss": 4.0213, "step": 20521 }, { "epoch": 0.20522, "grad_norm": 1.330463477670165, "learning_rate": 0.003, "loss": 4.06, "step": 20522 }, { "epoch": 0.20523, "grad_norm": 1.1888056787678798, "learning_rate": 0.003, "loss": 4.062, "step": 20523 }, { "epoch": 0.20524, "grad_norm": 1.282382010427326, "learning_rate": 0.003, "loss": 4.0483, "step": 20524 }, { "epoch": 0.20525, "grad_norm": 1.3060109899235448, "learning_rate": 0.003, "loss": 4.0497, "step": 20525 }, { "epoch": 0.20526, "grad_norm": 1.2621685969087504, "learning_rate": 0.003, "loss": 4.0393, "step": 20526 }, { "epoch": 0.20527, "grad_norm": 1.180382961412499, "learning_rate": 0.003, "loss": 4.0487, "step": 20527 }, { "epoch": 0.20528, "grad_norm": 1.2882221410943004, "learning_rate": 0.003, "loss": 4.0381, "step": 20528 }, { "epoch": 0.20529, "grad_norm": 1.1572585682565248, "learning_rate": 0.003, "loss": 4.064, "step": 20529 }, { "epoch": 0.2053, "grad_norm": 1.3687852153418119, "learning_rate": 0.003, "loss": 4.052, "step": 20530 }, { "epoch": 0.20531, "grad_norm": 1.1473823648403227, "learning_rate": 0.003, "loss": 4.0628, "step": 20531 }, { "epoch": 0.20532, "grad_norm": 1.2134733635286072, "learning_rate": 0.003, "loss": 4.0184, "step": 20532 }, { "epoch": 0.20533, "grad_norm": 1.1899592582746903, "learning_rate": 0.003, "loss": 4.0498, "step": 20533 }, { "epoch": 0.20534, "grad_norm": 1.388609446741199, "learning_rate": 0.003, "loss": 4.0595, "step": 20534 }, { "epoch": 0.20535, "grad_norm": 1.4700216421888237, "learning_rate": 0.003, "loss": 4.0473, "step": 20535 }, { "epoch": 0.20536, "grad_norm": 1.1139304987306389, "learning_rate": 0.003, "loss": 4.0371, "step": 20536 }, { "epoch": 0.20537, "grad_norm": 1.666631669804934, "learning_rate": 0.003, "loss": 4.0436, "step": 20537 }, { "epoch": 0.20538, "grad_norm": 1.0872075776099392, "learning_rate": 0.003, "loss": 4.0285, "step": 20538 }, { "epoch": 0.20539, "grad_norm": 1.387006541861756, "learning_rate": 0.003, "loss": 4.0542, "step": 20539 }, { "epoch": 0.2054, "grad_norm": 1.2152337637991244, "learning_rate": 0.003, "loss": 4.0596, "step": 20540 }, { "epoch": 0.20541, "grad_norm": 1.464417464579699, "learning_rate": 0.003, "loss": 4.035, "step": 20541 }, { "epoch": 0.20542, "grad_norm": 1.2962152186801776, "learning_rate": 0.003, "loss": 4.1038, "step": 20542 }, { "epoch": 0.20543, "grad_norm": 1.2587650889566364, "learning_rate": 0.003, "loss": 4.0371, "step": 20543 }, { "epoch": 0.20544, "grad_norm": 1.2278624251115824, "learning_rate": 0.003, "loss": 4.037, "step": 20544 }, { "epoch": 0.20545, "grad_norm": 1.3033201276713364, "learning_rate": 0.003, "loss": 4.0376, "step": 20545 }, { "epoch": 0.20546, "grad_norm": 1.3703718316768527, "learning_rate": 0.003, "loss": 4.011, "step": 20546 }, { "epoch": 0.20547, "grad_norm": 1.2482927032005222, "learning_rate": 0.003, "loss": 4.0279, "step": 20547 }, { "epoch": 0.20548, "grad_norm": 1.2524111624915995, "learning_rate": 0.003, "loss": 4.0723, "step": 20548 }, { "epoch": 0.20549, "grad_norm": 1.3874528947117495, "learning_rate": 0.003, "loss": 4.0558, "step": 20549 }, { "epoch": 0.2055, "grad_norm": 1.2424791935123651, "learning_rate": 0.003, "loss": 4.0657, "step": 20550 }, { "epoch": 0.20551, "grad_norm": 1.2245096080080125, "learning_rate": 0.003, "loss": 4.0463, "step": 20551 }, { "epoch": 0.20552, "grad_norm": 1.565644611290424, "learning_rate": 0.003, "loss": 4.0502, "step": 20552 }, { "epoch": 0.20553, "grad_norm": 0.8983082530583695, "learning_rate": 0.003, "loss": 4.0206, "step": 20553 }, { "epoch": 0.20554, "grad_norm": 1.4622981748993382, "learning_rate": 0.003, "loss": 4.0604, "step": 20554 }, { "epoch": 0.20555, "grad_norm": 1.3713023786716179, "learning_rate": 0.003, "loss": 4.0722, "step": 20555 }, { "epoch": 0.20556, "grad_norm": 1.2223116811646306, "learning_rate": 0.003, "loss": 4.0311, "step": 20556 }, { "epoch": 0.20557, "grad_norm": 1.2230211390947725, "learning_rate": 0.003, "loss": 4.0506, "step": 20557 }, { "epoch": 0.20558, "grad_norm": 1.2727476956328838, "learning_rate": 0.003, "loss": 4.0749, "step": 20558 }, { "epoch": 0.20559, "grad_norm": 1.2630263689417707, "learning_rate": 0.003, "loss": 4.0394, "step": 20559 }, { "epoch": 0.2056, "grad_norm": 1.380809886600488, "learning_rate": 0.003, "loss": 4.0612, "step": 20560 }, { "epoch": 0.20561, "grad_norm": 1.2041721567299215, "learning_rate": 0.003, "loss": 4.0504, "step": 20561 }, { "epoch": 0.20562, "grad_norm": 1.3675561477488738, "learning_rate": 0.003, "loss": 4.0539, "step": 20562 }, { "epoch": 0.20563, "grad_norm": 0.9813567807250267, "learning_rate": 0.003, "loss": 4.0653, "step": 20563 }, { "epoch": 0.20564, "grad_norm": 1.2156187944006267, "learning_rate": 0.003, "loss": 4.0377, "step": 20564 }, { "epoch": 0.20565, "grad_norm": 1.17726874163147, "learning_rate": 0.003, "loss": 4.0353, "step": 20565 }, { "epoch": 0.20566, "grad_norm": 1.252404813673878, "learning_rate": 0.003, "loss": 4.0474, "step": 20566 }, { "epoch": 0.20567, "grad_norm": 1.3644825816059427, "learning_rate": 0.003, "loss": 4.0271, "step": 20567 }, { "epoch": 0.20568, "grad_norm": 1.2680271459853363, "learning_rate": 0.003, "loss": 4.023, "step": 20568 }, { "epoch": 0.20569, "grad_norm": 1.251376240785194, "learning_rate": 0.003, "loss": 4.0279, "step": 20569 }, { "epoch": 0.2057, "grad_norm": 1.3928814421434723, "learning_rate": 0.003, "loss": 4.0471, "step": 20570 }, { "epoch": 0.20571, "grad_norm": 1.149725386259138, "learning_rate": 0.003, "loss": 4.0483, "step": 20571 }, { "epoch": 0.20572, "grad_norm": 1.406423417673629, "learning_rate": 0.003, "loss": 4.0554, "step": 20572 }, { "epoch": 0.20573, "grad_norm": 1.2594626237260136, "learning_rate": 0.003, "loss": 4.0576, "step": 20573 }, { "epoch": 0.20574, "grad_norm": 1.214761971646534, "learning_rate": 0.003, "loss": 4.0474, "step": 20574 }, { "epoch": 0.20575, "grad_norm": 1.3867559215345646, "learning_rate": 0.003, "loss": 4.0717, "step": 20575 }, { "epoch": 0.20576, "grad_norm": 1.0350722602622613, "learning_rate": 0.003, "loss": 4.0274, "step": 20576 }, { "epoch": 0.20577, "grad_norm": 1.4415700347605132, "learning_rate": 0.003, "loss": 4.035, "step": 20577 }, { "epoch": 0.20578, "grad_norm": 1.2257186853441844, "learning_rate": 0.003, "loss": 4.0588, "step": 20578 }, { "epoch": 0.20579, "grad_norm": 1.4617773787267274, "learning_rate": 0.003, "loss": 4.0381, "step": 20579 }, { "epoch": 0.2058, "grad_norm": 1.0446844849935397, "learning_rate": 0.003, "loss": 4.0637, "step": 20580 }, { "epoch": 0.20581, "grad_norm": 1.2309633880667266, "learning_rate": 0.003, "loss": 4.0412, "step": 20581 }, { "epoch": 0.20582, "grad_norm": 1.2501727853012718, "learning_rate": 0.003, "loss": 4.0647, "step": 20582 }, { "epoch": 0.20583, "grad_norm": 1.554489605582137, "learning_rate": 0.003, "loss": 4.0281, "step": 20583 }, { "epoch": 0.20584, "grad_norm": 1.1688257115667773, "learning_rate": 0.003, "loss": 4.0538, "step": 20584 }, { "epoch": 0.20585, "grad_norm": 1.657144677567979, "learning_rate": 0.003, "loss": 4.0531, "step": 20585 }, { "epoch": 0.20586, "grad_norm": 0.9750985458908387, "learning_rate": 0.003, "loss": 4.0367, "step": 20586 }, { "epoch": 0.20587, "grad_norm": 1.3610198926495998, "learning_rate": 0.003, "loss": 4.0652, "step": 20587 }, { "epoch": 0.20588, "grad_norm": 1.194014890403564, "learning_rate": 0.003, "loss": 4.0388, "step": 20588 }, { "epoch": 0.20589, "grad_norm": 1.2440849695940555, "learning_rate": 0.003, "loss": 4.0429, "step": 20589 }, { "epoch": 0.2059, "grad_norm": 1.182737702079615, "learning_rate": 0.003, "loss": 4.0657, "step": 20590 }, { "epoch": 0.20591, "grad_norm": 1.1991048748933788, "learning_rate": 0.003, "loss": 4.041, "step": 20591 }, { "epoch": 0.20592, "grad_norm": 1.369921693717321, "learning_rate": 0.003, "loss": 4.0513, "step": 20592 }, { "epoch": 0.20593, "grad_norm": 1.4253420981695712, "learning_rate": 0.003, "loss": 4.0505, "step": 20593 }, { "epoch": 0.20594, "grad_norm": 1.239517134925124, "learning_rate": 0.003, "loss": 4.0461, "step": 20594 }, { "epoch": 0.20595, "grad_norm": 1.2045848075684387, "learning_rate": 0.003, "loss": 4.0473, "step": 20595 }, { "epoch": 0.20596, "grad_norm": 1.3277529608869314, "learning_rate": 0.003, "loss": 4.0712, "step": 20596 }, { "epoch": 0.20597, "grad_norm": 1.092490748652337, "learning_rate": 0.003, "loss": 4.0205, "step": 20597 }, { "epoch": 0.20598, "grad_norm": 1.2148708086439042, "learning_rate": 0.003, "loss": 4.0565, "step": 20598 }, { "epoch": 0.20599, "grad_norm": 1.456604647874599, "learning_rate": 0.003, "loss": 4.1006, "step": 20599 }, { "epoch": 0.206, "grad_norm": 1.185871450619726, "learning_rate": 0.003, "loss": 4.0351, "step": 20600 }, { "epoch": 0.20601, "grad_norm": 1.3666582857210787, "learning_rate": 0.003, "loss": 4.061, "step": 20601 }, { "epoch": 0.20602, "grad_norm": 1.2210227052199842, "learning_rate": 0.003, "loss": 4.0481, "step": 20602 }, { "epoch": 0.20603, "grad_norm": 1.4549005179978116, "learning_rate": 0.003, "loss": 4.0318, "step": 20603 }, { "epoch": 0.20604, "grad_norm": 1.150989627895327, "learning_rate": 0.003, "loss": 4.0492, "step": 20604 }, { "epoch": 0.20605, "grad_norm": 1.5770164514930232, "learning_rate": 0.003, "loss": 4.0716, "step": 20605 }, { "epoch": 0.20606, "grad_norm": 1.0607505870665006, "learning_rate": 0.003, "loss": 4.0191, "step": 20606 }, { "epoch": 0.20607, "grad_norm": 1.70352693192207, "learning_rate": 0.003, "loss": 4.0891, "step": 20607 }, { "epoch": 0.20608, "grad_norm": 1.1033050008611927, "learning_rate": 0.003, "loss": 4.0365, "step": 20608 }, { "epoch": 0.20609, "grad_norm": 1.235798120813274, "learning_rate": 0.003, "loss": 4.0627, "step": 20609 }, { "epoch": 0.2061, "grad_norm": 1.6060560230983314, "learning_rate": 0.003, "loss": 4.052, "step": 20610 }, { "epoch": 0.20611, "grad_norm": 0.8996221193574458, "learning_rate": 0.003, "loss": 4.0691, "step": 20611 }, { "epoch": 0.20612, "grad_norm": 1.4191226435691027, "learning_rate": 0.003, "loss": 4.0757, "step": 20612 }, { "epoch": 0.20613, "grad_norm": 1.2155417941742577, "learning_rate": 0.003, "loss": 4.032, "step": 20613 }, { "epoch": 0.20614, "grad_norm": 1.3933365393758117, "learning_rate": 0.003, "loss": 4.0805, "step": 20614 }, { "epoch": 0.20615, "grad_norm": 1.259737007470902, "learning_rate": 0.003, "loss": 4.0225, "step": 20615 }, { "epoch": 0.20616, "grad_norm": 1.4224039150471577, "learning_rate": 0.003, "loss": 4.0698, "step": 20616 }, { "epoch": 0.20617, "grad_norm": 1.1925649227479738, "learning_rate": 0.003, "loss": 4.0405, "step": 20617 }, { "epoch": 0.20618, "grad_norm": 1.2215486795495927, "learning_rate": 0.003, "loss": 4.0552, "step": 20618 }, { "epoch": 0.20619, "grad_norm": 1.3938969321035501, "learning_rate": 0.003, "loss": 4.0939, "step": 20619 }, { "epoch": 0.2062, "grad_norm": 1.627778881202133, "learning_rate": 0.003, "loss": 4.057, "step": 20620 }, { "epoch": 0.20621, "grad_norm": 1.2438999679741989, "learning_rate": 0.003, "loss": 4.0654, "step": 20621 }, { "epoch": 0.20622, "grad_norm": 1.0670015597144602, "learning_rate": 0.003, "loss": 4.0344, "step": 20622 }, { "epoch": 0.20623, "grad_norm": 1.4980924500585362, "learning_rate": 0.003, "loss": 4.0479, "step": 20623 }, { "epoch": 0.20624, "grad_norm": 1.158117494940876, "learning_rate": 0.003, "loss": 4.0384, "step": 20624 }, { "epoch": 0.20625, "grad_norm": 1.4354739798432996, "learning_rate": 0.003, "loss": 4.036, "step": 20625 }, { "epoch": 0.20626, "grad_norm": 1.0875577049487977, "learning_rate": 0.003, "loss": 4.0792, "step": 20626 }, { "epoch": 0.20627, "grad_norm": 1.2212342891564674, "learning_rate": 0.003, "loss": 4.0596, "step": 20627 }, { "epoch": 0.20628, "grad_norm": 1.1224429716310118, "learning_rate": 0.003, "loss": 4.0776, "step": 20628 }, { "epoch": 0.20629, "grad_norm": 1.1185573262472182, "learning_rate": 0.003, "loss": 4.0508, "step": 20629 }, { "epoch": 0.2063, "grad_norm": 1.3897139436188553, "learning_rate": 0.003, "loss": 4.0309, "step": 20630 }, { "epoch": 0.20631, "grad_norm": 1.317437915728774, "learning_rate": 0.003, "loss": 4.048, "step": 20631 }, { "epoch": 0.20632, "grad_norm": 1.2560902403608283, "learning_rate": 0.003, "loss": 4.0529, "step": 20632 }, { "epoch": 0.20633, "grad_norm": 1.2556066526959209, "learning_rate": 0.003, "loss": 4.0421, "step": 20633 }, { "epoch": 0.20634, "grad_norm": 1.29749439461352, "learning_rate": 0.003, "loss": 4.0501, "step": 20634 }, { "epoch": 0.20635, "grad_norm": 1.2046627949955206, "learning_rate": 0.003, "loss": 4.0314, "step": 20635 }, { "epoch": 0.20636, "grad_norm": 1.4682069421201127, "learning_rate": 0.003, "loss": 4.0776, "step": 20636 }, { "epoch": 0.20637, "grad_norm": 1.181759682172407, "learning_rate": 0.003, "loss": 4.0274, "step": 20637 }, { "epoch": 0.20638, "grad_norm": 1.420208549746017, "learning_rate": 0.003, "loss": 4.0356, "step": 20638 }, { "epoch": 0.20639, "grad_norm": 1.0997561521912844, "learning_rate": 0.003, "loss": 4.0239, "step": 20639 }, { "epoch": 0.2064, "grad_norm": 1.3971745611642852, "learning_rate": 0.003, "loss": 4.061, "step": 20640 }, { "epoch": 0.20641, "grad_norm": 1.1517191748984548, "learning_rate": 0.003, "loss": 4.0455, "step": 20641 }, { "epoch": 0.20642, "grad_norm": 1.3854882769852916, "learning_rate": 0.003, "loss": 4.0556, "step": 20642 }, { "epoch": 0.20643, "grad_norm": 1.2170386478062296, "learning_rate": 0.003, "loss": 4.0512, "step": 20643 }, { "epoch": 0.20644, "grad_norm": 1.281003095728175, "learning_rate": 0.003, "loss": 4.068, "step": 20644 }, { "epoch": 0.20645, "grad_norm": 1.2251108945855154, "learning_rate": 0.003, "loss": 4.0652, "step": 20645 }, { "epoch": 0.20646, "grad_norm": 1.38774337747246, "learning_rate": 0.003, "loss": 4.0581, "step": 20646 }, { "epoch": 0.20647, "grad_norm": 1.2719846999883633, "learning_rate": 0.003, "loss": 4.0562, "step": 20647 }, { "epoch": 0.20648, "grad_norm": 1.3384524298496236, "learning_rate": 0.003, "loss": 4.0242, "step": 20648 }, { "epoch": 0.20649, "grad_norm": 1.5083236211350421, "learning_rate": 0.003, "loss": 4.054, "step": 20649 }, { "epoch": 0.2065, "grad_norm": 1.1610057430015674, "learning_rate": 0.003, "loss": 4.0141, "step": 20650 }, { "epoch": 0.20651, "grad_norm": 1.2686265477241983, "learning_rate": 0.003, "loss": 4.0521, "step": 20651 }, { "epoch": 0.20652, "grad_norm": 1.2170028195733724, "learning_rate": 0.003, "loss": 3.9959, "step": 20652 }, { "epoch": 0.20653, "grad_norm": 1.1984071586143492, "learning_rate": 0.003, "loss": 4.0455, "step": 20653 }, { "epoch": 0.20654, "grad_norm": 1.2865578494221055, "learning_rate": 0.003, "loss": 4.0634, "step": 20654 }, { "epoch": 0.20655, "grad_norm": 1.0735489017586273, "learning_rate": 0.003, "loss": 4.042, "step": 20655 }, { "epoch": 0.20656, "grad_norm": 1.6011180870459096, "learning_rate": 0.003, "loss": 4.0523, "step": 20656 }, { "epoch": 0.20657, "grad_norm": 1.0945711017845394, "learning_rate": 0.003, "loss": 4.0823, "step": 20657 }, { "epoch": 0.20658, "grad_norm": 1.632189199418318, "learning_rate": 0.003, "loss": 4.0586, "step": 20658 }, { "epoch": 0.20659, "grad_norm": 1.248305388631274, "learning_rate": 0.003, "loss": 4.0599, "step": 20659 }, { "epoch": 0.2066, "grad_norm": 1.4236851837706253, "learning_rate": 0.003, "loss": 4.0323, "step": 20660 }, { "epoch": 0.20661, "grad_norm": 1.177042760140521, "learning_rate": 0.003, "loss": 4.0776, "step": 20661 }, { "epoch": 0.20662, "grad_norm": 1.2280027663932278, "learning_rate": 0.003, "loss": 4.0439, "step": 20662 }, { "epoch": 0.20663, "grad_norm": 1.4467956014829126, "learning_rate": 0.003, "loss": 4.0418, "step": 20663 }, { "epoch": 0.20664, "grad_norm": 1.231375671997567, "learning_rate": 0.003, "loss": 4.019, "step": 20664 }, { "epoch": 0.20665, "grad_norm": 1.1612978286709583, "learning_rate": 0.003, "loss": 4.0434, "step": 20665 }, { "epoch": 0.20666, "grad_norm": 1.1931450829525216, "learning_rate": 0.003, "loss": 4.0527, "step": 20666 }, { "epoch": 0.20667, "grad_norm": 1.3875364993532027, "learning_rate": 0.003, "loss": 4.0785, "step": 20667 }, { "epoch": 0.20668, "grad_norm": 1.2775509160710945, "learning_rate": 0.003, "loss": 4.0601, "step": 20668 }, { "epoch": 0.20669, "grad_norm": 1.2290026315694929, "learning_rate": 0.003, "loss": 4.0124, "step": 20669 }, { "epoch": 0.2067, "grad_norm": 1.3993429867831688, "learning_rate": 0.003, "loss": 4.0725, "step": 20670 }, { "epoch": 0.20671, "grad_norm": 1.2247149046568575, "learning_rate": 0.003, "loss": 4.0426, "step": 20671 }, { "epoch": 0.20672, "grad_norm": 1.5992138460318421, "learning_rate": 0.003, "loss": 4.0737, "step": 20672 }, { "epoch": 0.20673, "grad_norm": 1.217819248244983, "learning_rate": 0.003, "loss": 4.0529, "step": 20673 }, { "epoch": 0.20674, "grad_norm": 1.174585184075093, "learning_rate": 0.003, "loss": 4.0234, "step": 20674 }, { "epoch": 0.20675, "grad_norm": 1.5199680898632342, "learning_rate": 0.003, "loss": 4.0726, "step": 20675 }, { "epoch": 0.20676, "grad_norm": 1.2770945799134772, "learning_rate": 0.003, "loss": 4.0354, "step": 20676 }, { "epoch": 0.20677, "grad_norm": 1.230372537574802, "learning_rate": 0.003, "loss": 4.014, "step": 20677 }, { "epoch": 0.20678, "grad_norm": 1.2001912213070292, "learning_rate": 0.003, "loss": 4.0492, "step": 20678 }, { "epoch": 0.20679, "grad_norm": 1.2545792444304698, "learning_rate": 0.003, "loss": 4.08, "step": 20679 }, { "epoch": 0.2068, "grad_norm": 1.2036745332235377, "learning_rate": 0.003, "loss": 4.048, "step": 20680 }, { "epoch": 0.20681, "grad_norm": 1.156025537281944, "learning_rate": 0.003, "loss": 4.0458, "step": 20681 }, { "epoch": 0.20682, "grad_norm": 1.2147853047790425, "learning_rate": 0.003, "loss": 4.009, "step": 20682 }, { "epoch": 0.20683, "grad_norm": 1.2951899681457213, "learning_rate": 0.003, "loss": 4.0365, "step": 20683 }, { "epoch": 0.20684, "grad_norm": 1.2828579337821484, "learning_rate": 0.003, "loss": 4.0578, "step": 20684 }, { "epoch": 0.20685, "grad_norm": 1.2365241593515177, "learning_rate": 0.003, "loss": 4.0261, "step": 20685 }, { "epoch": 0.20686, "grad_norm": 1.415664458003797, "learning_rate": 0.003, "loss": 4.0604, "step": 20686 }, { "epoch": 0.20687, "grad_norm": 1.3797239423240015, "learning_rate": 0.003, "loss": 4.0219, "step": 20687 }, { "epoch": 0.20688, "grad_norm": 1.1438572845436745, "learning_rate": 0.003, "loss": 4.0432, "step": 20688 }, { "epoch": 0.20689, "grad_norm": 1.3461755101404702, "learning_rate": 0.003, "loss": 4.0385, "step": 20689 }, { "epoch": 0.2069, "grad_norm": 1.312201354100928, "learning_rate": 0.003, "loss": 4.0524, "step": 20690 }, { "epoch": 0.20691, "grad_norm": 1.3174299393061977, "learning_rate": 0.003, "loss": 4.0697, "step": 20691 }, { "epoch": 0.20692, "grad_norm": 1.1573339127994, "learning_rate": 0.003, "loss": 4.0585, "step": 20692 }, { "epoch": 0.20693, "grad_norm": 1.5660673222263295, "learning_rate": 0.003, "loss": 4.0314, "step": 20693 }, { "epoch": 0.20694, "grad_norm": 1.0905513582817086, "learning_rate": 0.003, "loss": 4.0352, "step": 20694 }, { "epoch": 0.20695, "grad_norm": 1.3212477241955471, "learning_rate": 0.003, "loss": 4.0397, "step": 20695 }, { "epoch": 0.20696, "grad_norm": 1.073198136565413, "learning_rate": 0.003, "loss": 4.0468, "step": 20696 }, { "epoch": 0.20697, "grad_norm": 1.300374772924821, "learning_rate": 0.003, "loss": 4.0413, "step": 20697 }, { "epoch": 0.20698, "grad_norm": 1.0767247923857175, "learning_rate": 0.003, "loss": 4.0452, "step": 20698 }, { "epoch": 0.20699, "grad_norm": 1.5065065597585965, "learning_rate": 0.003, "loss": 4.073, "step": 20699 }, { "epoch": 0.207, "grad_norm": 1.2454389004086728, "learning_rate": 0.003, "loss": 4.0367, "step": 20700 }, { "epoch": 0.20701, "grad_norm": 1.3314144646308688, "learning_rate": 0.003, "loss": 4.0614, "step": 20701 }, { "epoch": 0.20702, "grad_norm": 1.3694677377990165, "learning_rate": 0.003, "loss": 4.0688, "step": 20702 }, { "epoch": 0.20703, "grad_norm": 1.4647675342775082, "learning_rate": 0.003, "loss": 4.0384, "step": 20703 }, { "epoch": 0.20704, "grad_norm": 1.2431562429480945, "learning_rate": 0.003, "loss": 4.0505, "step": 20704 }, { "epoch": 0.20705, "grad_norm": 1.3133051510352354, "learning_rate": 0.003, "loss": 4.0414, "step": 20705 }, { "epoch": 0.20706, "grad_norm": 1.4612118003039232, "learning_rate": 0.003, "loss": 4.0362, "step": 20706 }, { "epoch": 0.20707, "grad_norm": 1.118376984859685, "learning_rate": 0.003, "loss": 4.0392, "step": 20707 }, { "epoch": 0.20708, "grad_norm": 1.5201178905586379, "learning_rate": 0.003, "loss": 4.0679, "step": 20708 }, { "epoch": 0.20709, "grad_norm": 0.9611235534335962, "learning_rate": 0.003, "loss": 4.0565, "step": 20709 }, { "epoch": 0.2071, "grad_norm": 1.3960161951593277, "learning_rate": 0.003, "loss": 4.0317, "step": 20710 }, { "epoch": 0.20711, "grad_norm": 1.1151622481979715, "learning_rate": 0.003, "loss": 4.0669, "step": 20711 }, { "epoch": 0.20712, "grad_norm": 1.5674117359355935, "learning_rate": 0.003, "loss": 4.077, "step": 20712 }, { "epoch": 0.20713, "grad_norm": 1.0274111294218384, "learning_rate": 0.003, "loss": 4.0377, "step": 20713 }, { "epoch": 0.20714, "grad_norm": 1.581769560964899, "learning_rate": 0.003, "loss": 4.0592, "step": 20714 }, { "epoch": 0.20715, "grad_norm": 1.084185138019809, "learning_rate": 0.003, "loss": 4.0501, "step": 20715 }, { "epoch": 0.20716, "grad_norm": 1.4325211186606996, "learning_rate": 0.003, "loss": 4.0402, "step": 20716 }, { "epoch": 0.20717, "grad_norm": 1.3918793460894499, "learning_rate": 0.003, "loss": 4.0484, "step": 20717 }, { "epoch": 0.20718, "grad_norm": 1.6126234769973353, "learning_rate": 0.003, "loss": 4.0739, "step": 20718 }, { "epoch": 0.20719, "grad_norm": 1.2739890863916845, "learning_rate": 0.003, "loss": 4.0688, "step": 20719 }, { "epoch": 0.2072, "grad_norm": 1.3784070284601277, "learning_rate": 0.003, "loss": 4.0528, "step": 20720 }, { "epoch": 0.20721, "grad_norm": 1.327251701653354, "learning_rate": 0.003, "loss": 4.0905, "step": 20721 }, { "epoch": 0.20722, "grad_norm": 1.1155978897483336, "learning_rate": 0.003, "loss": 4.0594, "step": 20722 }, { "epoch": 0.20723, "grad_norm": 1.315794850057153, "learning_rate": 0.003, "loss": 4.069, "step": 20723 }, { "epoch": 0.20724, "grad_norm": 1.3772555675504765, "learning_rate": 0.003, "loss": 4.0488, "step": 20724 }, { "epoch": 0.20725, "grad_norm": 1.1192654698576237, "learning_rate": 0.003, "loss": 4.0805, "step": 20725 }, { "epoch": 0.20726, "grad_norm": 1.5072056858156686, "learning_rate": 0.003, "loss": 4.0473, "step": 20726 }, { "epoch": 0.20727, "grad_norm": 1.2860033041347505, "learning_rate": 0.003, "loss": 4.041, "step": 20727 }, { "epoch": 0.20728, "grad_norm": 1.1638881920695072, "learning_rate": 0.003, "loss": 4.047, "step": 20728 }, { "epoch": 0.20729, "grad_norm": 1.4448337864291165, "learning_rate": 0.003, "loss": 4.0337, "step": 20729 }, { "epoch": 0.2073, "grad_norm": 1.1178446923668317, "learning_rate": 0.003, "loss": 4.0486, "step": 20730 }, { "epoch": 0.20731, "grad_norm": 1.3855596548490468, "learning_rate": 0.003, "loss": 4.0308, "step": 20731 }, { "epoch": 0.20732, "grad_norm": 1.0902998889131228, "learning_rate": 0.003, "loss": 4.0032, "step": 20732 }, { "epoch": 0.20733, "grad_norm": 1.3623174030487706, "learning_rate": 0.003, "loss": 4.0306, "step": 20733 }, { "epoch": 0.20734, "grad_norm": 1.2207749874969758, "learning_rate": 0.003, "loss": 4.0516, "step": 20734 }, { "epoch": 0.20735, "grad_norm": 1.316201299828994, "learning_rate": 0.003, "loss": 4.0763, "step": 20735 }, { "epoch": 0.20736, "grad_norm": 1.1638952152907889, "learning_rate": 0.003, "loss": 4.0498, "step": 20736 }, { "epoch": 0.20737, "grad_norm": 1.2916292397240166, "learning_rate": 0.003, "loss": 4.0491, "step": 20737 }, { "epoch": 0.20738, "grad_norm": 1.1862988132039978, "learning_rate": 0.003, "loss": 4.0593, "step": 20738 }, { "epoch": 0.20739, "grad_norm": 1.3710172962165703, "learning_rate": 0.003, "loss": 4.0496, "step": 20739 }, { "epoch": 0.2074, "grad_norm": 1.1044965581222408, "learning_rate": 0.003, "loss": 4.0508, "step": 20740 }, { "epoch": 0.20741, "grad_norm": 1.270604851444255, "learning_rate": 0.003, "loss": 4.0533, "step": 20741 }, { "epoch": 0.20742, "grad_norm": 1.1771694767018996, "learning_rate": 0.003, "loss": 4.0362, "step": 20742 }, { "epoch": 0.20743, "grad_norm": 1.4379249444804114, "learning_rate": 0.003, "loss": 4.0637, "step": 20743 }, { "epoch": 0.20744, "grad_norm": 1.0065490802911512, "learning_rate": 0.003, "loss": 4.0801, "step": 20744 }, { "epoch": 0.20745, "grad_norm": 1.606752109344879, "learning_rate": 0.003, "loss": 4.0455, "step": 20745 }, { "epoch": 0.20746, "grad_norm": 1.6169098553926398, "learning_rate": 0.003, "loss": 4.0514, "step": 20746 }, { "epoch": 0.20747, "grad_norm": 1.0463943810491074, "learning_rate": 0.003, "loss": 4.0611, "step": 20747 }, { "epoch": 0.20748, "grad_norm": 1.3687174936811755, "learning_rate": 0.003, "loss": 4.0475, "step": 20748 }, { "epoch": 0.20749, "grad_norm": 1.1107697487186206, "learning_rate": 0.003, "loss": 4.0513, "step": 20749 }, { "epoch": 0.2075, "grad_norm": 1.1774792649263777, "learning_rate": 0.003, "loss": 4.0782, "step": 20750 }, { "epoch": 0.20751, "grad_norm": 1.2807190015587913, "learning_rate": 0.003, "loss": 4.0505, "step": 20751 }, { "epoch": 0.20752, "grad_norm": 1.3254048160794856, "learning_rate": 0.003, "loss": 4.0865, "step": 20752 }, { "epoch": 0.20753, "grad_norm": 1.3495750153155786, "learning_rate": 0.003, "loss": 4.0634, "step": 20753 }, { "epoch": 0.20754, "grad_norm": 1.357838787640671, "learning_rate": 0.003, "loss": 4.0574, "step": 20754 }, { "epoch": 0.20755, "grad_norm": 1.1509300649152578, "learning_rate": 0.003, "loss": 4.0449, "step": 20755 }, { "epoch": 0.20756, "grad_norm": 1.4739648216529362, "learning_rate": 0.003, "loss": 4.0446, "step": 20756 }, { "epoch": 0.20757, "grad_norm": 1.0317305907954093, "learning_rate": 0.003, "loss": 4.0492, "step": 20757 }, { "epoch": 0.20758, "grad_norm": 1.5550783556663623, "learning_rate": 0.003, "loss": 4.0544, "step": 20758 }, { "epoch": 0.20759, "grad_norm": 1.049816633874253, "learning_rate": 0.003, "loss": 4.0329, "step": 20759 }, { "epoch": 0.2076, "grad_norm": 1.3850869771822945, "learning_rate": 0.003, "loss": 4.0214, "step": 20760 }, { "epoch": 0.20761, "grad_norm": 1.18946485064815, "learning_rate": 0.003, "loss": 4.0451, "step": 20761 }, { "epoch": 0.20762, "grad_norm": 1.4499318233215346, "learning_rate": 0.003, "loss": 4.0621, "step": 20762 }, { "epoch": 0.20763, "grad_norm": 0.9098224927031964, "learning_rate": 0.003, "loss": 4.046, "step": 20763 }, { "epoch": 0.20764, "grad_norm": 1.336208800715881, "learning_rate": 0.003, "loss": 4.0697, "step": 20764 }, { "epoch": 0.20765, "grad_norm": 1.3240349176179802, "learning_rate": 0.003, "loss": 4.0443, "step": 20765 }, { "epoch": 0.20766, "grad_norm": 1.4657133881894573, "learning_rate": 0.003, "loss": 4.0615, "step": 20766 }, { "epoch": 0.20767, "grad_norm": 1.463064796644083, "learning_rate": 0.003, "loss": 4.0769, "step": 20767 }, { "epoch": 0.20768, "grad_norm": 1.0376490916670396, "learning_rate": 0.003, "loss": 4.0698, "step": 20768 }, { "epoch": 0.20769, "grad_norm": 1.5355933849330268, "learning_rate": 0.003, "loss": 4.051, "step": 20769 }, { "epoch": 0.2077, "grad_norm": 1.1550285482192808, "learning_rate": 0.003, "loss": 4.0553, "step": 20770 }, { "epoch": 0.20771, "grad_norm": 1.6016164048712531, "learning_rate": 0.003, "loss": 4.0126, "step": 20771 }, { "epoch": 0.20772, "grad_norm": 1.0556020067156495, "learning_rate": 0.003, "loss": 4.0287, "step": 20772 }, { "epoch": 0.20773, "grad_norm": 1.1085310083632471, "learning_rate": 0.003, "loss": 4.0528, "step": 20773 }, { "epoch": 0.20774, "grad_norm": 1.3511133330470035, "learning_rate": 0.003, "loss": 4.0483, "step": 20774 }, { "epoch": 0.20775, "grad_norm": 1.0777696954278422, "learning_rate": 0.003, "loss": 4.049, "step": 20775 }, { "epoch": 0.20776, "grad_norm": 1.4478514990309528, "learning_rate": 0.003, "loss": 4.0478, "step": 20776 }, { "epoch": 0.20777, "grad_norm": 1.2627413233954994, "learning_rate": 0.003, "loss": 4.0841, "step": 20777 }, { "epoch": 0.20778, "grad_norm": 1.1945623095279247, "learning_rate": 0.003, "loss": 4.0668, "step": 20778 }, { "epoch": 0.20779, "grad_norm": 1.350820223734434, "learning_rate": 0.003, "loss": 4.0512, "step": 20779 }, { "epoch": 0.2078, "grad_norm": 1.233820204919974, "learning_rate": 0.003, "loss": 4.0333, "step": 20780 }, { "epoch": 0.20781, "grad_norm": 1.424115334207775, "learning_rate": 0.003, "loss": 4.0664, "step": 20781 }, { "epoch": 0.20782, "grad_norm": 1.1560346264308283, "learning_rate": 0.003, "loss": 4.0493, "step": 20782 }, { "epoch": 0.20783, "grad_norm": 1.9797191617411831, "learning_rate": 0.003, "loss": 4.0571, "step": 20783 }, { "epoch": 0.20784, "grad_norm": 1.1826811649741138, "learning_rate": 0.003, "loss": 4.0365, "step": 20784 }, { "epoch": 0.20785, "grad_norm": 1.3513571715731507, "learning_rate": 0.003, "loss": 4.0405, "step": 20785 }, { "epoch": 0.20786, "grad_norm": 1.2581827014596552, "learning_rate": 0.003, "loss": 4.0732, "step": 20786 }, { "epoch": 0.20787, "grad_norm": 1.199670279474661, "learning_rate": 0.003, "loss": 4.0405, "step": 20787 }, { "epoch": 0.20788, "grad_norm": 1.2612259754453077, "learning_rate": 0.003, "loss": 4.0253, "step": 20788 }, { "epoch": 0.20789, "grad_norm": 1.4795022118185028, "learning_rate": 0.003, "loss": 4.0724, "step": 20789 }, { "epoch": 0.2079, "grad_norm": 1.1137152810385258, "learning_rate": 0.003, "loss": 4.0576, "step": 20790 }, { "epoch": 0.20791, "grad_norm": 1.116026403579865, "learning_rate": 0.003, "loss": 4.0509, "step": 20791 }, { "epoch": 0.20792, "grad_norm": 1.4050517941276137, "learning_rate": 0.003, "loss": 4.06, "step": 20792 }, { "epoch": 0.20793, "grad_norm": 1.3092047542552736, "learning_rate": 0.003, "loss": 4.0792, "step": 20793 }, { "epoch": 0.20794, "grad_norm": 1.2555634846244876, "learning_rate": 0.003, "loss": 4.0603, "step": 20794 }, { "epoch": 0.20795, "grad_norm": 1.1152539238852723, "learning_rate": 0.003, "loss": 4.0419, "step": 20795 }, { "epoch": 0.20796, "grad_norm": 1.3704306356685125, "learning_rate": 0.003, "loss": 4.0542, "step": 20796 }, { "epoch": 0.20797, "grad_norm": 1.3005254051176414, "learning_rate": 0.003, "loss": 4.0049, "step": 20797 }, { "epoch": 0.20798, "grad_norm": 1.1293085128932312, "learning_rate": 0.003, "loss": 4.0462, "step": 20798 }, { "epoch": 0.20799, "grad_norm": 1.2954795612006205, "learning_rate": 0.003, "loss": 4.0492, "step": 20799 }, { "epoch": 0.208, "grad_norm": 1.287234029685038, "learning_rate": 0.003, "loss": 4.0376, "step": 20800 }, { "epoch": 0.20801, "grad_norm": 1.1639186714194185, "learning_rate": 0.003, "loss": 4.0503, "step": 20801 }, { "epoch": 0.20802, "grad_norm": 1.3642931017731221, "learning_rate": 0.003, "loss": 4.0659, "step": 20802 }, { "epoch": 0.20803, "grad_norm": 1.300376417837977, "learning_rate": 0.003, "loss": 4.0351, "step": 20803 }, { "epoch": 0.20804, "grad_norm": 1.19418609442342, "learning_rate": 0.003, "loss": 4.0784, "step": 20804 }, { "epoch": 0.20805, "grad_norm": 1.474440167259748, "learning_rate": 0.003, "loss": 4.0451, "step": 20805 }, { "epoch": 0.20806, "grad_norm": 1.1675361434908935, "learning_rate": 0.003, "loss": 4.0731, "step": 20806 }, { "epoch": 0.20807, "grad_norm": 1.4549767555633737, "learning_rate": 0.003, "loss": 4.0433, "step": 20807 }, { "epoch": 0.20808, "grad_norm": 1.2582960110328578, "learning_rate": 0.003, "loss": 4.0263, "step": 20808 }, { "epoch": 0.20809, "grad_norm": 1.293121104939769, "learning_rate": 0.003, "loss": 4.0535, "step": 20809 }, { "epoch": 0.2081, "grad_norm": 1.3426327364168016, "learning_rate": 0.003, "loss": 4.0562, "step": 20810 }, { "epoch": 0.20811, "grad_norm": 1.0024443980199094, "learning_rate": 0.003, "loss": 4.045, "step": 20811 }, { "epoch": 0.20812, "grad_norm": 1.335153296270808, "learning_rate": 0.003, "loss": 4.0464, "step": 20812 }, { "epoch": 0.20813, "grad_norm": 1.1369543975249898, "learning_rate": 0.003, "loss": 4.0667, "step": 20813 }, { "epoch": 0.20814, "grad_norm": 1.2746262554435819, "learning_rate": 0.003, "loss": 4.0709, "step": 20814 }, { "epoch": 0.20815, "grad_norm": 1.3587407132341922, "learning_rate": 0.003, "loss": 4.0544, "step": 20815 }, { "epoch": 0.20816, "grad_norm": 1.1291126681104129, "learning_rate": 0.003, "loss": 4.0523, "step": 20816 }, { "epoch": 0.20817, "grad_norm": 1.4650140314166336, "learning_rate": 0.003, "loss": 4.0607, "step": 20817 }, { "epoch": 0.20818, "grad_norm": 1.113597228538697, "learning_rate": 0.003, "loss": 4.0873, "step": 20818 }, { "epoch": 0.20819, "grad_norm": 1.6122646829652605, "learning_rate": 0.003, "loss": 4.047, "step": 20819 }, { "epoch": 0.2082, "grad_norm": 1.3561395147569244, "learning_rate": 0.003, "loss": 4.0606, "step": 20820 }, { "epoch": 0.20821, "grad_norm": 1.6761463283533349, "learning_rate": 0.003, "loss": 4.0404, "step": 20821 }, { "epoch": 0.20822, "grad_norm": 1.0185067312374685, "learning_rate": 0.003, "loss": 4.0818, "step": 20822 }, { "epoch": 0.20823, "grad_norm": 1.293982685093887, "learning_rate": 0.003, "loss": 4.0475, "step": 20823 }, { "epoch": 0.20824, "grad_norm": 1.1625700519350355, "learning_rate": 0.003, "loss": 4.085, "step": 20824 }, { "epoch": 0.20825, "grad_norm": 1.2888389139611907, "learning_rate": 0.003, "loss": 4.0766, "step": 20825 }, { "epoch": 0.20826, "grad_norm": 1.116572675993816, "learning_rate": 0.003, "loss": 4.0764, "step": 20826 }, { "epoch": 0.20827, "grad_norm": 1.3177369257552722, "learning_rate": 0.003, "loss": 4.0444, "step": 20827 }, { "epoch": 0.20828, "grad_norm": 1.1371442329404948, "learning_rate": 0.003, "loss": 4.0525, "step": 20828 }, { "epoch": 0.20829, "grad_norm": 1.2947468624123517, "learning_rate": 0.003, "loss": 4.0542, "step": 20829 }, { "epoch": 0.2083, "grad_norm": 1.2602130502568567, "learning_rate": 0.003, "loss": 4.0326, "step": 20830 }, { "epoch": 0.20831, "grad_norm": 1.4506110797716467, "learning_rate": 0.003, "loss": 4.0643, "step": 20831 }, { "epoch": 0.20832, "grad_norm": 1.0928195888686325, "learning_rate": 0.003, "loss": 4.0327, "step": 20832 }, { "epoch": 0.20833, "grad_norm": 1.4604003874374663, "learning_rate": 0.003, "loss": 4.0849, "step": 20833 }, { "epoch": 0.20834, "grad_norm": 1.2252945089312588, "learning_rate": 0.003, "loss": 4.0341, "step": 20834 }, { "epoch": 0.20835, "grad_norm": 1.3495567279436074, "learning_rate": 0.003, "loss": 4.0676, "step": 20835 }, { "epoch": 0.20836, "grad_norm": 1.1663520726944048, "learning_rate": 0.003, "loss": 4.0577, "step": 20836 }, { "epoch": 0.20837, "grad_norm": 1.3042798883210238, "learning_rate": 0.003, "loss": 4.0819, "step": 20837 }, { "epoch": 0.20838, "grad_norm": 1.234572062163786, "learning_rate": 0.003, "loss": 4.0198, "step": 20838 }, { "epoch": 0.20839, "grad_norm": 1.3203252337818405, "learning_rate": 0.003, "loss": 4.0557, "step": 20839 }, { "epoch": 0.2084, "grad_norm": 1.0404060047258217, "learning_rate": 0.003, "loss": 4.0266, "step": 20840 }, { "epoch": 0.20841, "grad_norm": 1.5256692288077685, "learning_rate": 0.003, "loss": 4.0397, "step": 20841 }, { "epoch": 0.20842, "grad_norm": 1.6727543163276237, "learning_rate": 0.003, "loss": 4.0773, "step": 20842 }, { "epoch": 0.20843, "grad_norm": 1.2483890401340039, "learning_rate": 0.003, "loss": 4.0317, "step": 20843 }, { "epoch": 0.20844, "grad_norm": 1.2585950608498393, "learning_rate": 0.003, "loss": 4.0402, "step": 20844 }, { "epoch": 0.20845, "grad_norm": 1.3047930143249864, "learning_rate": 0.003, "loss": 4.0524, "step": 20845 }, { "epoch": 0.20846, "grad_norm": 1.3345669483117277, "learning_rate": 0.003, "loss": 4.0421, "step": 20846 }, { "epoch": 0.20847, "grad_norm": 1.3647467882722444, "learning_rate": 0.003, "loss": 4.0773, "step": 20847 }, { "epoch": 0.20848, "grad_norm": 0.9861317635079706, "learning_rate": 0.003, "loss": 4.0307, "step": 20848 }, { "epoch": 0.20849, "grad_norm": 1.3763661514454046, "learning_rate": 0.003, "loss": 4.0525, "step": 20849 }, { "epoch": 0.2085, "grad_norm": 1.3161831768920973, "learning_rate": 0.003, "loss": 4.053, "step": 20850 }, { "epoch": 0.20851, "grad_norm": 1.2378064888547666, "learning_rate": 0.003, "loss": 4.0368, "step": 20851 }, { "epoch": 0.20852, "grad_norm": 1.216971027708619, "learning_rate": 0.003, "loss": 4.0594, "step": 20852 }, { "epoch": 0.20853, "grad_norm": 1.2657044247692195, "learning_rate": 0.003, "loss": 4.0431, "step": 20853 }, { "epoch": 0.20854, "grad_norm": 1.3156640153098798, "learning_rate": 0.003, "loss": 4.0704, "step": 20854 }, { "epoch": 0.20855, "grad_norm": 1.3470086450334378, "learning_rate": 0.003, "loss": 4.0465, "step": 20855 }, { "epoch": 0.20856, "grad_norm": 1.1077005630643735, "learning_rate": 0.003, "loss": 4.037, "step": 20856 }, { "epoch": 0.20857, "grad_norm": 2.0150889721151577, "learning_rate": 0.003, "loss": 4.0754, "step": 20857 }, { "epoch": 0.20858, "grad_norm": 1.0494119576201082, "learning_rate": 0.003, "loss": 4.0747, "step": 20858 }, { "epoch": 0.20859, "grad_norm": 1.4376968690517848, "learning_rate": 0.003, "loss": 4.0963, "step": 20859 }, { "epoch": 0.2086, "grad_norm": 1.2228033500769202, "learning_rate": 0.003, "loss": 4.0609, "step": 20860 }, { "epoch": 0.20861, "grad_norm": 1.3024082179552416, "learning_rate": 0.003, "loss": 4.0512, "step": 20861 }, { "epoch": 0.20862, "grad_norm": 1.3337153907875594, "learning_rate": 0.003, "loss": 4.0505, "step": 20862 }, { "epoch": 0.20863, "grad_norm": 1.3523027874716267, "learning_rate": 0.003, "loss": 4.068, "step": 20863 }, { "epoch": 0.20864, "grad_norm": 1.1000205902889388, "learning_rate": 0.003, "loss": 4.0281, "step": 20864 }, { "epoch": 0.20865, "grad_norm": 1.3449720085528347, "learning_rate": 0.003, "loss": 4.0586, "step": 20865 }, { "epoch": 0.20866, "grad_norm": 1.2382383546368607, "learning_rate": 0.003, "loss": 4.0554, "step": 20866 }, { "epoch": 0.20867, "grad_norm": 1.1269017647330786, "learning_rate": 0.003, "loss": 4.0426, "step": 20867 }, { "epoch": 0.20868, "grad_norm": 1.2986669015988064, "learning_rate": 0.003, "loss": 4.0463, "step": 20868 }, { "epoch": 0.20869, "grad_norm": 1.1455552607424562, "learning_rate": 0.003, "loss": 4.0671, "step": 20869 }, { "epoch": 0.2087, "grad_norm": 1.3253962829092347, "learning_rate": 0.003, "loss": 4.0566, "step": 20870 }, { "epoch": 0.20871, "grad_norm": 1.0568251837872027, "learning_rate": 0.003, "loss": 4.0736, "step": 20871 }, { "epoch": 0.20872, "grad_norm": 1.2689662179831813, "learning_rate": 0.003, "loss": 4.0877, "step": 20872 }, { "epoch": 0.20873, "grad_norm": 1.0137833497000508, "learning_rate": 0.003, "loss": 4.0482, "step": 20873 }, { "epoch": 0.20874, "grad_norm": 1.3925150194736946, "learning_rate": 0.003, "loss": 4.0704, "step": 20874 }, { "epoch": 0.20875, "grad_norm": 1.147944154950515, "learning_rate": 0.003, "loss": 4.0429, "step": 20875 }, { "epoch": 0.20876, "grad_norm": 1.4880598077123546, "learning_rate": 0.003, "loss": 4.0706, "step": 20876 }, { "epoch": 0.20877, "grad_norm": 1.4617326494717378, "learning_rate": 0.003, "loss": 4.0463, "step": 20877 }, { "epoch": 0.20878, "grad_norm": 1.3219327544624884, "learning_rate": 0.003, "loss": 4.0647, "step": 20878 }, { "epoch": 0.20879, "grad_norm": 1.420988445786985, "learning_rate": 0.003, "loss": 4.0517, "step": 20879 }, { "epoch": 0.2088, "grad_norm": 1.1368159597668355, "learning_rate": 0.003, "loss": 4.0567, "step": 20880 }, { "epoch": 0.20881, "grad_norm": 1.1650311538443099, "learning_rate": 0.003, "loss": 4.0639, "step": 20881 }, { "epoch": 0.20882, "grad_norm": 1.4596155026540338, "learning_rate": 0.003, "loss": 4.0713, "step": 20882 }, { "epoch": 0.20883, "grad_norm": 1.2613695008600023, "learning_rate": 0.003, "loss": 4.0541, "step": 20883 }, { "epoch": 0.20884, "grad_norm": 1.2692564484497293, "learning_rate": 0.003, "loss": 4.081, "step": 20884 }, { "epoch": 0.20885, "grad_norm": 1.292229202469805, "learning_rate": 0.003, "loss": 4.0507, "step": 20885 }, { "epoch": 0.20886, "grad_norm": 1.1647777743341188, "learning_rate": 0.003, "loss": 4.0384, "step": 20886 }, { "epoch": 0.20887, "grad_norm": 1.3424953059606388, "learning_rate": 0.003, "loss": 4.0939, "step": 20887 }, { "epoch": 0.20888, "grad_norm": 1.4193618803799186, "learning_rate": 0.003, "loss": 4.0762, "step": 20888 }, { "epoch": 0.20889, "grad_norm": 1.1278547256220763, "learning_rate": 0.003, "loss": 4.0557, "step": 20889 }, { "epoch": 0.2089, "grad_norm": 1.3975877767352827, "learning_rate": 0.003, "loss": 4.027, "step": 20890 }, { "epoch": 0.20891, "grad_norm": 1.162881771907803, "learning_rate": 0.003, "loss": 4.0361, "step": 20891 }, { "epoch": 0.20892, "grad_norm": 1.5056194371792304, "learning_rate": 0.003, "loss": 4.0626, "step": 20892 }, { "epoch": 0.20893, "grad_norm": 1.2736401249513964, "learning_rate": 0.003, "loss": 4.0366, "step": 20893 }, { "epoch": 0.20894, "grad_norm": 1.4500261051793482, "learning_rate": 0.003, "loss": 4.0103, "step": 20894 }, { "epoch": 0.20895, "grad_norm": 1.179569673900195, "learning_rate": 0.003, "loss": 4.0729, "step": 20895 }, { "epoch": 0.20896, "grad_norm": 1.229599686937581, "learning_rate": 0.003, "loss": 4.0618, "step": 20896 }, { "epoch": 0.20897, "grad_norm": 1.2413269333270043, "learning_rate": 0.003, "loss": 4.037, "step": 20897 }, { "epoch": 0.20898, "grad_norm": 1.1355251062525624, "learning_rate": 0.003, "loss": 4.0436, "step": 20898 }, { "epoch": 0.20899, "grad_norm": 1.326473613801506, "learning_rate": 0.003, "loss": 4.0603, "step": 20899 }, { "epoch": 0.209, "grad_norm": 1.2938071798412598, "learning_rate": 0.003, "loss": 4.0367, "step": 20900 }, { "epoch": 0.20901, "grad_norm": 1.5539919449076114, "learning_rate": 0.003, "loss": 4.0826, "step": 20901 }, { "epoch": 0.20902, "grad_norm": 1.0175792191757047, "learning_rate": 0.003, "loss": 4.0379, "step": 20902 }, { "epoch": 0.20903, "grad_norm": 1.4185608575024329, "learning_rate": 0.003, "loss": 4.0302, "step": 20903 }, { "epoch": 0.20904, "grad_norm": 1.489692037723099, "learning_rate": 0.003, "loss": 4.049, "step": 20904 }, { "epoch": 0.20905, "grad_norm": 1.1518373995991453, "learning_rate": 0.003, "loss": 4.0878, "step": 20905 }, { "epoch": 0.20906, "grad_norm": 1.2745302902631523, "learning_rate": 0.003, "loss": 4.0646, "step": 20906 }, { "epoch": 0.20907, "grad_norm": 1.1981105025956502, "learning_rate": 0.003, "loss": 4.06, "step": 20907 }, { "epoch": 0.20908, "grad_norm": 1.1071198808766434, "learning_rate": 0.003, "loss": 4.0233, "step": 20908 }, { "epoch": 0.20909, "grad_norm": 1.2480676753967612, "learning_rate": 0.003, "loss": 4.0271, "step": 20909 }, { "epoch": 0.2091, "grad_norm": 1.2125172864613187, "learning_rate": 0.003, "loss": 4.0607, "step": 20910 }, { "epoch": 0.20911, "grad_norm": 1.328490722810735, "learning_rate": 0.003, "loss": 4.0596, "step": 20911 }, { "epoch": 0.20912, "grad_norm": 1.1249015135299267, "learning_rate": 0.003, "loss": 4.041, "step": 20912 }, { "epoch": 0.20913, "grad_norm": 1.2203451005770125, "learning_rate": 0.003, "loss": 4.0585, "step": 20913 }, { "epoch": 0.20914, "grad_norm": 1.4007068660764677, "learning_rate": 0.003, "loss": 4.0503, "step": 20914 }, { "epoch": 0.20915, "grad_norm": 1.096838057456493, "learning_rate": 0.003, "loss": 4.0572, "step": 20915 }, { "epoch": 0.20916, "grad_norm": 1.462002736596207, "learning_rate": 0.003, "loss": 4.0856, "step": 20916 }, { "epoch": 0.20917, "grad_norm": 1.2023236355995262, "learning_rate": 0.003, "loss": 4.0239, "step": 20917 }, { "epoch": 0.20918, "grad_norm": 1.2963698128726884, "learning_rate": 0.003, "loss": 4.0485, "step": 20918 }, { "epoch": 0.20919, "grad_norm": 1.228596871930989, "learning_rate": 0.003, "loss": 4.0635, "step": 20919 }, { "epoch": 0.2092, "grad_norm": 1.3763175882958565, "learning_rate": 0.003, "loss": 4.0655, "step": 20920 }, { "epoch": 0.20921, "grad_norm": 1.2404096739675687, "learning_rate": 0.003, "loss": 4.0534, "step": 20921 }, { "epoch": 0.20922, "grad_norm": 1.1403168687067564, "learning_rate": 0.003, "loss": 4.0377, "step": 20922 }, { "epoch": 0.20923, "grad_norm": 1.3997924724596653, "learning_rate": 0.003, "loss": 4.0853, "step": 20923 }, { "epoch": 0.20924, "grad_norm": 1.1041797216252074, "learning_rate": 0.003, "loss": 4.0319, "step": 20924 }, { "epoch": 0.20925, "grad_norm": 1.3711841401346831, "learning_rate": 0.003, "loss": 4.0737, "step": 20925 }, { "epoch": 0.20926, "grad_norm": 1.0450269013241913, "learning_rate": 0.003, "loss": 4.0522, "step": 20926 }, { "epoch": 0.20927, "grad_norm": 1.2769383700182384, "learning_rate": 0.003, "loss": 4.0522, "step": 20927 }, { "epoch": 0.20928, "grad_norm": 1.0877225315431993, "learning_rate": 0.003, "loss": 4.0282, "step": 20928 }, { "epoch": 0.20929, "grad_norm": 1.6300139792461399, "learning_rate": 0.003, "loss": 4.0692, "step": 20929 }, { "epoch": 0.2093, "grad_norm": 1.3242783820254749, "learning_rate": 0.003, "loss": 4.0412, "step": 20930 }, { "epoch": 0.20931, "grad_norm": 1.257103843444017, "learning_rate": 0.003, "loss": 4.0662, "step": 20931 }, { "epoch": 0.20932, "grad_norm": 1.521557788811263, "learning_rate": 0.003, "loss": 4.0482, "step": 20932 }, { "epoch": 0.20933, "grad_norm": 1.318340273370988, "learning_rate": 0.003, "loss": 4.0292, "step": 20933 }, { "epoch": 0.20934, "grad_norm": 1.3417539283120765, "learning_rate": 0.003, "loss": 4.071, "step": 20934 }, { "epoch": 0.20935, "grad_norm": 1.3259221373793442, "learning_rate": 0.003, "loss": 4.0682, "step": 20935 }, { "epoch": 0.20936, "grad_norm": 1.2800407480604998, "learning_rate": 0.003, "loss": 4.0132, "step": 20936 }, { "epoch": 0.20937, "grad_norm": 0.9876674260510029, "learning_rate": 0.003, "loss": 4.0398, "step": 20937 }, { "epoch": 0.20938, "grad_norm": 1.4116121790748142, "learning_rate": 0.003, "loss": 4.0489, "step": 20938 }, { "epoch": 0.20939, "grad_norm": 1.2277894657211132, "learning_rate": 0.003, "loss": 4.0556, "step": 20939 }, { "epoch": 0.2094, "grad_norm": 1.3662094509846838, "learning_rate": 0.003, "loss": 4.0575, "step": 20940 }, { "epoch": 0.20941, "grad_norm": 1.115121969618257, "learning_rate": 0.003, "loss": 4.057, "step": 20941 }, { "epoch": 0.20942, "grad_norm": 1.5119266528992794, "learning_rate": 0.003, "loss": 4.0521, "step": 20942 }, { "epoch": 0.20943, "grad_norm": 1.3523607537848863, "learning_rate": 0.003, "loss": 4.0565, "step": 20943 }, { "epoch": 0.20944, "grad_norm": 1.3515926466848913, "learning_rate": 0.003, "loss": 4.0446, "step": 20944 }, { "epoch": 0.20945, "grad_norm": 1.1832739989508942, "learning_rate": 0.003, "loss": 4.0689, "step": 20945 }, { "epoch": 0.20946, "grad_norm": 1.2097681134785438, "learning_rate": 0.003, "loss": 4.0579, "step": 20946 }, { "epoch": 0.20947, "grad_norm": 1.1714307763907412, "learning_rate": 0.003, "loss": 4.0351, "step": 20947 }, { "epoch": 0.20948, "grad_norm": 1.352952022914992, "learning_rate": 0.003, "loss": 4.0323, "step": 20948 }, { "epoch": 0.20949, "grad_norm": 1.415737718595553, "learning_rate": 0.003, "loss": 4.0455, "step": 20949 }, { "epoch": 0.2095, "grad_norm": 1.058417143122475, "learning_rate": 0.003, "loss": 4.0492, "step": 20950 }, { "epoch": 0.20951, "grad_norm": 1.2579147483083288, "learning_rate": 0.003, "loss": 4.0578, "step": 20951 }, { "epoch": 0.20952, "grad_norm": 1.474085819945999, "learning_rate": 0.003, "loss": 4.0858, "step": 20952 }, { "epoch": 0.20953, "grad_norm": 0.9452978892338686, "learning_rate": 0.003, "loss": 4.0153, "step": 20953 }, { "epoch": 0.20954, "grad_norm": 1.2645678482354852, "learning_rate": 0.003, "loss": 4.0127, "step": 20954 }, { "epoch": 0.20955, "grad_norm": 1.3400150004615905, "learning_rate": 0.003, "loss": 4.0368, "step": 20955 }, { "epoch": 0.20956, "grad_norm": 1.133802275603527, "learning_rate": 0.003, "loss": 4.0408, "step": 20956 }, { "epoch": 0.20957, "grad_norm": 1.2247995439196313, "learning_rate": 0.003, "loss": 4.0319, "step": 20957 }, { "epoch": 0.20958, "grad_norm": 1.3174201355530692, "learning_rate": 0.003, "loss": 4.0714, "step": 20958 }, { "epoch": 0.20959, "grad_norm": 1.2321339496179788, "learning_rate": 0.003, "loss": 4.0426, "step": 20959 }, { "epoch": 0.2096, "grad_norm": 1.345487502302491, "learning_rate": 0.003, "loss": 4.0396, "step": 20960 }, { "epoch": 0.20961, "grad_norm": 1.232371302399167, "learning_rate": 0.003, "loss": 4.0372, "step": 20961 }, { "epoch": 0.20962, "grad_norm": 1.4100997616180153, "learning_rate": 0.003, "loss": 4.056, "step": 20962 }, { "epoch": 0.20963, "grad_norm": 1.214560339721232, "learning_rate": 0.003, "loss": 4.0325, "step": 20963 }, { "epoch": 0.20964, "grad_norm": 1.3867937404423791, "learning_rate": 0.003, "loss": 4.0476, "step": 20964 }, { "epoch": 0.20965, "grad_norm": 1.269943982429771, "learning_rate": 0.003, "loss": 4.0479, "step": 20965 }, { "epoch": 0.20966, "grad_norm": 1.3386216228181835, "learning_rate": 0.003, "loss": 4.0226, "step": 20966 }, { "epoch": 0.20967, "grad_norm": 1.227105428563871, "learning_rate": 0.003, "loss": 4.0318, "step": 20967 }, { "epoch": 0.20968, "grad_norm": 1.4264916694640326, "learning_rate": 0.003, "loss": 4.0565, "step": 20968 }, { "epoch": 0.20969, "grad_norm": 0.9743109180159732, "learning_rate": 0.003, "loss": 4.0415, "step": 20969 }, { "epoch": 0.2097, "grad_norm": 1.5037940025653431, "learning_rate": 0.003, "loss": 4.0476, "step": 20970 }, { "epoch": 0.20971, "grad_norm": 1.1352251405098386, "learning_rate": 0.003, "loss": 4.0423, "step": 20971 }, { "epoch": 0.20972, "grad_norm": 1.3717446928455301, "learning_rate": 0.003, "loss": 4.0668, "step": 20972 }, { "epoch": 0.20973, "grad_norm": 1.183300212395144, "learning_rate": 0.003, "loss": 4.0298, "step": 20973 }, { "epoch": 0.20974, "grad_norm": 1.627985998429152, "learning_rate": 0.003, "loss": 4.0727, "step": 20974 }, { "epoch": 0.20975, "grad_norm": 1.189942525999578, "learning_rate": 0.003, "loss": 4.0789, "step": 20975 }, { "epoch": 0.20976, "grad_norm": 1.3191092046966775, "learning_rate": 0.003, "loss": 4.0344, "step": 20976 }, { "epoch": 0.20977, "grad_norm": 1.2108052685829256, "learning_rate": 0.003, "loss": 4.0798, "step": 20977 }, { "epoch": 0.20978, "grad_norm": 1.1163403429301721, "learning_rate": 0.003, "loss": 4.0269, "step": 20978 }, { "epoch": 0.20979, "grad_norm": 1.2791139974316421, "learning_rate": 0.003, "loss": 4.0905, "step": 20979 }, { "epoch": 0.2098, "grad_norm": 1.3739885718908496, "learning_rate": 0.003, "loss": 4.0525, "step": 20980 }, { "epoch": 0.20981, "grad_norm": 1.1176967990345994, "learning_rate": 0.003, "loss": 4.0205, "step": 20981 }, { "epoch": 0.20982, "grad_norm": 1.7555812723347777, "learning_rate": 0.003, "loss": 4.0709, "step": 20982 }, { "epoch": 0.20983, "grad_norm": 1.0190490633305063, "learning_rate": 0.003, "loss": 4.0452, "step": 20983 }, { "epoch": 0.20984, "grad_norm": 1.647968629574487, "learning_rate": 0.003, "loss": 4.0473, "step": 20984 }, { "epoch": 0.20985, "grad_norm": 1.228601296940512, "learning_rate": 0.003, "loss": 4.088, "step": 20985 }, { "epoch": 0.20986, "grad_norm": 1.1780593895154643, "learning_rate": 0.003, "loss": 4.058, "step": 20986 }, { "epoch": 0.20987, "grad_norm": 1.5167477665623446, "learning_rate": 0.003, "loss": 4.0723, "step": 20987 }, { "epoch": 0.20988, "grad_norm": 1.2027453692295034, "learning_rate": 0.003, "loss": 4.0472, "step": 20988 }, { "epoch": 0.20989, "grad_norm": 1.3819784422414434, "learning_rate": 0.003, "loss": 4.0427, "step": 20989 }, { "epoch": 0.2099, "grad_norm": 1.431069095891436, "learning_rate": 0.003, "loss": 4.0631, "step": 20990 }, { "epoch": 0.20991, "grad_norm": 1.1856362937113927, "learning_rate": 0.003, "loss": 4.0684, "step": 20991 }, { "epoch": 0.20992, "grad_norm": 1.4579232113790843, "learning_rate": 0.003, "loss": 4.0715, "step": 20992 }, { "epoch": 0.20993, "grad_norm": 1.0877371552863573, "learning_rate": 0.003, "loss": 4.1089, "step": 20993 }, { "epoch": 0.20994, "grad_norm": 1.0799867989318765, "learning_rate": 0.003, "loss": 4.0648, "step": 20994 }, { "epoch": 0.20995, "grad_norm": 1.2679371046599244, "learning_rate": 0.003, "loss": 4.0314, "step": 20995 }, { "epoch": 0.20996, "grad_norm": 1.1247101100163748, "learning_rate": 0.003, "loss": 4.0538, "step": 20996 }, { "epoch": 0.20997, "grad_norm": 1.2723423812425403, "learning_rate": 0.003, "loss": 4.0357, "step": 20997 }, { "epoch": 0.20998, "grad_norm": 1.1523814759851905, "learning_rate": 0.003, "loss": 4.0432, "step": 20998 }, { "epoch": 0.20999, "grad_norm": 1.4168831860950526, "learning_rate": 0.003, "loss": 4.0615, "step": 20999 }, { "epoch": 0.21, "grad_norm": 0.9648126597864406, "learning_rate": 0.003, "loss": 4.08, "step": 21000 }, { "epoch": 0.21001, "grad_norm": 1.54406208440241, "learning_rate": 0.003, "loss": 4.0527, "step": 21001 }, { "epoch": 0.21002, "grad_norm": 1.128820519320507, "learning_rate": 0.003, "loss": 4.059, "step": 21002 }, { "epoch": 0.21003, "grad_norm": 1.2746340201716286, "learning_rate": 0.003, "loss": 4.0461, "step": 21003 }, { "epoch": 0.21004, "grad_norm": 1.2857510270154682, "learning_rate": 0.003, "loss": 4.0358, "step": 21004 }, { "epoch": 0.21005, "grad_norm": 1.1850808119042948, "learning_rate": 0.003, "loss": 4.048, "step": 21005 }, { "epoch": 0.21006, "grad_norm": 1.5522083439081036, "learning_rate": 0.003, "loss": 4.0516, "step": 21006 }, { "epoch": 0.21007, "grad_norm": 1.260774149470961, "learning_rate": 0.003, "loss": 4.0491, "step": 21007 }, { "epoch": 0.21008, "grad_norm": 1.5370298278427954, "learning_rate": 0.003, "loss": 4.0455, "step": 21008 }, { "epoch": 0.21009, "grad_norm": 1.2243091526973395, "learning_rate": 0.003, "loss": 4.0496, "step": 21009 }, { "epoch": 0.2101, "grad_norm": 1.3225787370878426, "learning_rate": 0.003, "loss": 4.0338, "step": 21010 }, { "epoch": 0.21011, "grad_norm": 1.18069554223696, "learning_rate": 0.003, "loss": 4.0417, "step": 21011 }, { "epoch": 0.21012, "grad_norm": 1.5722260010050817, "learning_rate": 0.003, "loss": 4.0733, "step": 21012 }, { "epoch": 0.21013, "grad_norm": 1.2113411973015873, "learning_rate": 0.003, "loss": 4.049, "step": 21013 }, { "epoch": 0.21014, "grad_norm": 1.3628036017370306, "learning_rate": 0.003, "loss": 4.0646, "step": 21014 }, { "epoch": 0.21015, "grad_norm": 1.1750802878262518, "learning_rate": 0.003, "loss": 4.0398, "step": 21015 }, { "epoch": 0.21016, "grad_norm": 1.4427593106096692, "learning_rate": 0.003, "loss": 4.0707, "step": 21016 }, { "epoch": 0.21017, "grad_norm": 1.0377068324397796, "learning_rate": 0.003, "loss": 4.033, "step": 21017 }, { "epoch": 0.21018, "grad_norm": 1.6716123251557335, "learning_rate": 0.003, "loss": 4.0493, "step": 21018 }, { "epoch": 0.21019, "grad_norm": 1.0599154480065698, "learning_rate": 0.003, "loss": 4.043, "step": 21019 }, { "epoch": 0.2102, "grad_norm": 1.5586356522325036, "learning_rate": 0.003, "loss": 4.0552, "step": 21020 }, { "epoch": 0.21021, "grad_norm": 1.2503098443526723, "learning_rate": 0.003, "loss": 4.0745, "step": 21021 }, { "epoch": 0.21022, "grad_norm": 1.2825989183117132, "learning_rate": 0.003, "loss": 4.051, "step": 21022 }, { "epoch": 0.21023, "grad_norm": 1.23175707597897, "learning_rate": 0.003, "loss": 4.0285, "step": 21023 }, { "epoch": 0.21024, "grad_norm": 1.3429031287970576, "learning_rate": 0.003, "loss": 4.059, "step": 21024 }, { "epoch": 0.21025, "grad_norm": 1.2119921684369184, "learning_rate": 0.003, "loss": 4.0507, "step": 21025 }, { "epoch": 0.21026, "grad_norm": 1.4827436505288472, "learning_rate": 0.003, "loss": 4.0667, "step": 21026 }, { "epoch": 0.21027, "grad_norm": 1.3281124155898993, "learning_rate": 0.003, "loss": 4.0848, "step": 21027 }, { "epoch": 0.21028, "grad_norm": 1.3266289315697783, "learning_rate": 0.003, "loss": 4.0524, "step": 21028 }, { "epoch": 0.21029, "grad_norm": 1.172763625947714, "learning_rate": 0.003, "loss": 4.0716, "step": 21029 }, { "epoch": 0.2103, "grad_norm": 1.2856114185127379, "learning_rate": 0.003, "loss": 4.0649, "step": 21030 }, { "epoch": 0.21031, "grad_norm": 1.1196273421901068, "learning_rate": 0.003, "loss": 4.0454, "step": 21031 }, { "epoch": 0.21032, "grad_norm": 1.4086759644049058, "learning_rate": 0.003, "loss": 4.0512, "step": 21032 }, { "epoch": 0.21033, "grad_norm": 1.2229488440703944, "learning_rate": 0.003, "loss": 4.0491, "step": 21033 }, { "epoch": 0.21034, "grad_norm": 1.3050149701049576, "learning_rate": 0.003, "loss": 4.0383, "step": 21034 }, { "epoch": 0.21035, "grad_norm": 1.561400770706906, "learning_rate": 0.003, "loss": 4.0447, "step": 21035 }, { "epoch": 0.21036, "grad_norm": 1.0559325603740646, "learning_rate": 0.003, "loss": 4.034, "step": 21036 }, { "epoch": 0.21037, "grad_norm": 1.3934052758308075, "learning_rate": 0.003, "loss": 4.0541, "step": 21037 }, { "epoch": 0.21038, "grad_norm": 1.1612544978492942, "learning_rate": 0.003, "loss": 4.0672, "step": 21038 }, { "epoch": 0.21039, "grad_norm": 1.4088570787549488, "learning_rate": 0.003, "loss": 4.0411, "step": 21039 }, { "epoch": 0.2104, "grad_norm": 1.2569459965148915, "learning_rate": 0.003, "loss": 4.0628, "step": 21040 }, { "epoch": 0.21041, "grad_norm": 1.3039754397941128, "learning_rate": 0.003, "loss": 4.0493, "step": 21041 }, { "epoch": 0.21042, "grad_norm": 1.1099797720966442, "learning_rate": 0.003, "loss": 4.0507, "step": 21042 }, { "epoch": 0.21043, "grad_norm": 1.2845449228632493, "learning_rate": 0.003, "loss": 4.0507, "step": 21043 }, { "epoch": 0.21044, "grad_norm": 1.3554885841652318, "learning_rate": 0.003, "loss": 4.0458, "step": 21044 }, { "epoch": 0.21045, "grad_norm": 1.230013105795136, "learning_rate": 0.003, "loss": 4.0284, "step": 21045 }, { "epoch": 0.21046, "grad_norm": 1.5314007698323049, "learning_rate": 0.003, "loss": 4.0512, "step": 21046 }, { "epoch": 0.21047, "grad_norm": 1.0067461744861579, "learning_rate": 0.003, "loss": 4.0246, "step": 21047 }, { "epoch": 0.21048, "grad_norm": 1.584802777434922, "learning_rate": 0.003, "loss": 4.0666, "step": 21048 }, { "epoch": 0.21049, "grad_norm": 1.1263449954189204, "learning_rate": 0.003, "loss": 4.0401, "step": 21049 }, { "epoch": 0.2105, "grad_norm": 1.6591382458406492, "learning_rate": 0.003, "loss": 4.0559, "step": 21050 }, { "epoch": 0.21051, "grad_norm": 1.05067562949757, "learning_rate": 0.003, "loss": 4.0899, "step": 21051 }, { "epoch": 0.21052, "grad_norm": 1.4121258978825222, "learning_rate": 0.003, "loss": 4.0455, "step": 21052 }, { "epoch": 0.21053, "grad_norm": 1.1885772521432727, "learning_rate": 0.003, "loss": 4.0381, "step": 21053 }, { "epoch": 0.21054, "grad_norm": 1.2910866675596309, "learning_rate": 0.003, "loss": 4.0761, "step": 21054 }, { "epoch": 0.21055, "grad_norm": 1.303537308251438, "learning_rate": 0.003, "loss": 4.069, "step": 21055 }, { "epoch": 0.21056, "grad_norm": 1.3372199163127803, "learning_rate": 0.003, "loss": 4.0614, "step": 21056 }, { "epoch": 0.21057, "grad_norm": 1.1842089486023717, "learning_rate": 0.003, "loss": 4.0403, "step": 21057 }, { "epoch": 0.21058, "grad_norm": 1.1150311159984807, "learning_rate": 0.003, "loss": 4.0501, "step": 21058 }, { "epoch": 0.21059, "grad_norm": 1.616958463659875, "learning_rate": 0.003, "loss": 4.0466, "step": 21059 }, { "epoch": 0.2106, "grad_norm": 1.00862729220954, "learning_rate": 0.003, "loss": 4.0593, "step": 21060 }, { "epoch": 0.21061, "grad_norm": 1.502498230598199, "learning_rate": 0.003, "loss": 4.0652, "step": 21061 }, { "epoch": 0.21062, "grad_norm": 1.0766915542174063, "learning_rate": 0.003, "loss": 4.0399, "step": 21062 }, { "epoch": 0.21063, "grad_norm": 1.5675211012398373, "learning_rate": 0.003, "loss": 4.0846, "step": 21063 }, { "epoch": 0.21064, "grad_norm": 1.0380204363407695, "learning_rate": 0.003, "loss": 4.0699, "step": 21064 }, { "epoch": 0.21065, "grad_norm": 1.469412240069509, "learning_rate": 0.003, "loss": 4.0664, "step": 21065 }, { "epoch": 0.21066, "grad_norm": 1.1353878096784409, "learning_rate": 0.003, "loss": 4.0536, "step": 21066 }, { "epoch": 0.21067, "grad_norm": 1.262367418845785, "learning_rate": 0.003, "loss": 4.0483, "step": 21067 }, { "epoch": 0.21068, "grad_norm": 1.161593396240763, "learning_rate": 0.003, "loss": 4.0496, "step": 21068 }, { "epoch": 0.21069, "grad_norm": 1.2757299030587597, "learning_rate": 0.003, "loss": 4.0688, "step": 21069 }, { "epoch": 0.2107, "grad_norm": 1.2448132233635982, "learning_rate": 0.003, "loss": 4.0626, "step": 21070 }, { "epoch": 0.21071, "grad_norm": 1.442376030384632, "learning_rate": 0.003, "loss": 4.0296, "step": 21071 }, { "epoch": 0.21072, "grad_norm": 1.4587049850944842, "learning_rate": 0.003, "loss": 4.0563, "step": 21072 }, { "epoch": 0.21073, "grad_norm": 1.192299761850415, "learning_rate": 0.003, "loss": 4.0697, "step": 21073 }, { "epoch": 0.21074, "grad_norm": 1.26884910425769, "learning_rate": 0.003, "loss": 4.058, "step": 21074 }, { "epoch": 0.21075, "grad_norm": 1.2893964103039317, "learning_rate": 0.003, "loss": 4.0689, "step": 21075 }, { "epoch": 0.21076, "grad_norm": 1.1787456309208952, "learning_rate": 0.003, "loss": 4.0516, "step": 21076 }, { "epoch": 0.21077, "grad_norm": 1.138603335750006, "learning_rate": 0.003, "loss": 4.0419, "step": 21077 }, { "epoch": 0.21078, "grad_norm": 1.1600173037126666, "learning_rate": 0.003, "loss": 4.0411, "step": 21078 }, { "epoch": 0.21079, "grad_norm": 1.3422085299958455, "learning_rate": 0.003, "loss": 4.0445, "step": 21079 }, { "epoch": 0.2108, "grad_norm": 1.106362513409859, "learning_rate": 0.003, "loss": 4.0093, "step": 21080 }, { "epoch": 0.21081, "grad_norm": 1.4805500589637395, "learning_rate": 0.003, "loss": 4.0686, "step": 21081 }, { "epoch": 0.21082, "grad_norm": 0.934243209161164, "learning_rate": 0.003, "loss": 4.0442, "step": 21082 }, { "epoch": 0.21083, "grad_norm": 1.3886084879612093, "learning_rate": 0.003, "loss": 4.065, "step": 21083 }, { "epoch": 0.21084, "grad_norm": 1.1952447634633065, "learning_rate": 0.003, "loss": 4.0161, "step": 21084 }, { "epoch": 0.21085, "grad_norm": 1.271509636174596, "learning_rate": 0.003, "loss": 4.0468, "step": 21085 }, { "epoch": 0.21086, "grad_norm": 1.1432380501669652, "learning_rate": 0.003, "loss": 4.0292, "step": 21086 }, { "epoch": 0.21087, "grad_norm": 1.5734936022836352, "learning_rate": 0.003, "loss": 4.072, "step": 21087 }, { "epoch": 0.21088, "grad_norm": 1.062576420079353, "learning_rate": 0.003, "loss": 4.0298, "step": 21088 }, { "epoch": 0.21089, "grad_norm": 1.577088852900612, "learning_rate": 0.003, "loss": 4.0774, "step": 21089 }, { "epoch": 0.2109, "grad_norm": 1.3262250573900358, "learning_rate": 0.003, "loss": 4.0601, "step": 21090 }, { "epoch": 0.21091, "grad_norm": 1.4790908407427124, "learning_rate": 0.003, "loss": 4.0384, "step": 21091 }, { "epoch": 0.21092, "grad_norm": 1.2811464635095229, "learning_rate": 0.003, "loss": 4.0708, "step": 21092 }, { "epoch": 0.21093, "grad_norm": 1.0974674810300278, "learning_rate": 0.003, "loss": 4.0686, "step": 21093 }, { "epoch": 0.21094, "grad_norm": 1.3751477595513788, "learning_rate": 0.003, "loss": 4.0507, "step": 21094 }, { "epoch": 0.21095, "grad_norm": 1.2412138108559143, "learning_rate": 0.003, "loss": 4.0653, "step": 21095 }, { "epoch": 0.21096, "grad_norm": 1.1591859582479014, "learning_rate": 0.003, "loss": 4.0342, "step": 21096 }, { "epoch": 0.21097, "grad_norm": 1.3530582912094091, "learning_rate": 0.003, "loss": 4.0473, "step": 21097 }, { "epoch": 0.21098, "grad_norm": 1.1617085781215324, "learning_rate": 0.003, "loss": 4.0489, "step": 21098 }, { "epoch": 0.21099, "grad_norm": 1.4129967228300409, "learning_rate": 0.003, "loss": 4.0357, "step": 21099 }, { "epoch": 0.211, "grad_norm": 1.0766297210176785, "learning_rate": 0.003, "loss": 4.045, "step": 21100 }, { "epoch": 0.21101, "grad_norm": 1.5789553526879387, "learning_rate": 0.003, "loss": 4.0163, "step": 21101 }, { "epoch": 0.21102, "grad_norm": 1.110150691244987, "learning_rate": 0.003, "loss": 4.0288, "step": 21102 }, { "epoch": 0.21103, "grad_norm": 1.3936425498083649, "learning_rate": 0.003, "loss": 4.0589, "step": 21103 }, { "epoch": 0.21104, "grad_norm": 1.2694800492278608, "learning_rate": 0.003, "loss": 4.0471, "step": 21104 }, { "epoch": 0.21105, "grad_norm": 1.4180659771764696, "learning_rate": 0.003, "loss": 4.0709, "step": 21105 }, { "epoch": 0.21106, "grad_norm": 1.1710647247911583, "learning_rate": 0.003, "loss": 4.0413, "step": 21106 }, { "epoch": 0.21107, "grad_norm": 1.3933621823096176, "learning_rate": 0.003, "loss": 4.0628, "step": 21107 }, { "epoch": 0.21108, "grad_norm": 1.201541250227036, "learning_rate": 0.003, "loss": 4.061, "step": 21108 }, { "epoch": 0.21109, "grad_norm": 1.2044639273867823, "learning_rate": 0.003, "loss": 4.0634, "step": 21109 }, { "epoch": 0.2111, "grad_norm": 1.3735940076719475, "learning_rate": 0.003, "loss": 4.0552, "step": 21110 }, { "epoch": 0.21111, "grad_norm": 1.2542327173484145, "learning_rate": 0.003, "loss": 4.0314, "step": 21111 }, { "epoch": 0.21112, "grad_norm": 1.6773848266159919, "learning_rate": 0.003, "loss": 4.0704, "step": 21112 }, { "epoch": 0.21113, "grad_norm": 1.2495999851941781, "learning_rate": 0.003, "loss": 4.0867, "step": 21113 }, { "epoch": 0.21114, "grad_norm": 1.283840621076478, "learning_rate": 0.003, "loss": 4.0762, "step": 21114 }, { "epoch": 0.21115, "grad_norm": 1.2666174821831944, "learning_rate": 0.003, "loss": 4.0661, "step": 21115 }, { "epoch": 0.21116, "grad_norm": 1.252920124476882, "learning_rate": 0.003, "loss": 4.0712, "step": 21116 }, { "epoch": 0.21117, "grad_norm": 1.1195946630367066, "learning_rate": 0.003, "loss": 4.0509, "step": 21117 }, { "epoch": 0.21118, "grad_norm": 1.2807588933818483, "learning_rate": 0.003, "loss": 4.013, "step": 21118 }, { "epoch": 0.21119, "grad_norm": 1.2286394305042434, "learning_rate": 0.003, "loss": 4.0487, "step": 21119 }, { "epoch": 0.2112, "grad_norm": 1.1650873552914147, "learning_rate": 0.003, "loss": 4.0723, "step": 21120 }, { "epoch": 0.21121, "grad_norm": 1.2531458893285583, "learning_rate": 0.003, "loss": 4.0402, "step": 21121 }, { "epoch": 0.21122, "grad_norm": 1.3200496972039166, "learning_rate": 0.003, "loss": 4.0306, "step": 21122 }, { "epoch": 0.21123, "grad_norm": 1.2464324409865812, "learning_rate": 0.003, "loss": 4.0482, "step": 21123 }, { "epoch": 0.21124, "grad_norm": 1.4351654375211487, "learning_rate": 0.003, "loss": 4.0413, "step": 21124 }, { "epoch": 0.21125, "grad_norm": 1.139776590390933, "learning_rate": 0.003, "loss": 4.0651, "step": 21125 }, { "epoch": 0.21126, "grad_norm": 1.3978253357428427, "learning_rate": 0.003, "loss": 4.0616, "step": 21126 }, { "epoch": 0.21127, "grad_norm": 1.2922171671346667, "learning_rate": 0.003, "loss": 4.0633, "step": 21127 }, { "epoch": 0.21128, "grad_norm": 1.091379730592485, "learning_rate": 0.003, "loss": 4.0464, "step": 21128 }, { "epoch": 0.21129, "grad_norm": 1.4712654455929264, "learning_rate": 0.003, "loss": 4.0683, "step": 21129 }, { "epoch": 0.2113, "grad_norm": 0.9216631097573316, "learning_rate": 0.003, "loss": 4.0423, "step": 21130 }, { "epoch": 0.21131, "grad_norm": 1.407904789855863, "learning_rate": 0.003, "loss": 4.0569, "step": 21131 }, { "epoch": 0.21132, "grad_norm": 1.115428868593496, "learning_rate": 0.003, "loss": 4.0575, "step": 21132 }, { "epoch": 0.21133, "grad_norm": 1.3006217391820132, "learning_rate": 0.003, "loss": 4.018, "step": 21133 }, { "epoch": 0.21134, "grad_norm": 1.0998440559746168, "learning_rate": 0.003, "loss": 4.0504, "step": 21134 }, { "epoch": 0.21135, "grad_norm": 1.2710129216531116, "learning_rate": 0.003, "loss": 4.0534, "step": 21135 }, { "epoch": 0.21136, "grad_norm": 1.3880099841956677, "learning_rate": 0.003, "loss": 4.0554, "step": 21136 }, { "epoch": 0.21137, "grad_norm": 1.274201487226877, "learning_rate": 0.003, "loss": 4.0431, "step": 21137 }, { "epoch": 0.21138, "grad_norm": 1.4527859691316656, "learning_rate": 0.003, "loss": 4.0224, "step": 21138 }, { "epoch": 0.21139, "grad_norm": 1.3381869632235348, "learning_rate": 0.003, "loss": 4.0482, "step": 21139 }, { "epoch": 0.2114, "grad_norm": 1.2203439761517438, "learning_rate": 0.003, "loss": 4.0621, "step": 21140 }, { "epoch": 0.21141, "grad_norm": 1.2587623540614699, "learning_rate": 0.003, "loss": 4.0399, "step": 21141 }, { "epoch": 0.21142, "grad_norm": 1.3981945231759798, "learning_rate": 0.003, "loss": 4.0524, "step": 21142 }, { "epoch": 0.21143, "grad_norm": 1.271008692230558, "learning_rate": 0.003, "loss": 4.0305, "step": 21143 }, { "epoch": 0.21144, "grad_norm": 1.1259055061671566, "learning_rate": 0.003, "loss": 4.0647, "step": 21144 }, { "epoch": 0.21145, "grad_norm": 1.2996841724908386, "learning_rate": 0.003, "loss": 4.0298, "step": 21145 }, { "epoch": 0.21146, "grad_norm": 1.1888755607807513, "learning_rate": 0.003, "loss": 4.0456, "step": 21146 }, { "epoch": 0.21147, "grad_norm": 1.2862247813508612, "learning_rate": 0.003, "loss": 4.0776, "step": 21147 }, { "epoch": 0.21148, "grad_norm": 1.1688646974365517, "learning_rate": 0.003, "loss": 4.0393, "step": 21148 }, { "epoch": 0.21149, "grad_norm": 1.363403802643418, "learning_rate": 0.003, "loss": 4.0885, "step": 21149 }, { "epoch": 0.2115, "grad_norm": 1.2336213830020937, "learning_rate": 0.003, "loss": 4.0545, "step": 21150 }, { "epoch": 0.21151, "grad_norm": 1.3083225835331127, "learning_rate": 0.003, "loss": 4.0494, "step": 21151 }, { "epoch": 0.21152, "grad_norm": 1.2041426169518765, "learning_rate": 0.003, "loss": 4.0324, "step": 21152 }, { "epoch": 0.21153, "grad_norm": 1.2308593899786529, "learning_rate": 0.003, "loss": 4.0388, "step": 21153 }, { "epoch": 0.21154, "grad_norm": 1.5824931816806196, "learning_rate": 0.003, "loss": 4.0365, "step": 21154 }, { "epoch": 0.21155, "grad_norm": 1.332366931832797, "learning_rate": 0.003, "loss": 4.0542, "step": 21155 }, { "epoch": 0.21156, "grad_norm": 1.4667661411092565, "learning_rate": 0.003, "loss": 4.0468, "step": 21156 }, { "epoch": 0.21157, "grad_norm": 1.1830738359144912, "learning_rate": 0.003, "loss": 4.0292, "step": 21157 }, { "epoch": 0.21158, "grad_norm": 1.4195587534035023, "learning_rate": 0.003, "loss": 4.0481, "step": 21158 }, { "epoch": 0.21159, "grad_norm": 1.3111519755366765, "learning_rate": 0.003, "loss": 4.0246, "step": 21159 }, { "epoch": 0.2116, "grad_norm": 1.300192919042313, "learning_rate": 0.003, "loss": 4.0551, "step": 21160 }, { "epoch": 0.21161, "grad_norm": 1.2209088854363133, "learning_rate": 0.003, "loss": 4.0364, "step": 21161 }, { "epoch": 0.21162, "grad_norm": 1.2583148642570114, "learning_rate": 0.003, "loss": 4.0452, "step": 21162 }, { "epoch": 0.21163, "grad_norm": 1.4025946969762597, "learning_rate": 0.003, "loss": 4.039, "step": 21163 }, { "epoch": 0.21164, "grad_norm": 1.130491503867519, "learning_rate": 0.003, "loss": 4.0665, "step": 21164 }, { "epoch": 0.21165, "grad_norm": 1.3669270369568793, "learning_rate": 0.003, "loss": 4.0473, "step": 21165 }, { "epoch": 0.21166, "grad_norm": 1.1992647571186519, "learning_rate": 0.003, "loss": 4.0298, "step": 21166 }, { "epoch": 0.21167, "grad_norm": 1.4464082027049883, "learning_rate": 0.003, "loss": 4.0496, "step": 21167 }, { "epoch": 0.21168, "grad_norm": 1.3387509078840587, "learning_rate": 0.003, "loss": 4.0479, "step": 21168 }, { "epoch": 0.21169, "grad_norm": 1.1012420365333515, "learning_rate": 0.003, "loss": 4.0465, "step": 21169 }, { "epoch": 0.2117, "grad_norm": 1.2878223957665704, "learning_rate": 0.003, "loss": 4.0519, "step": 21170 }, { "epoch": 0.21171, "grad_norm": 1.1488393297719888, "learning_rate": 0.003, "loss": 4.068, "step": 21171 }, { "epoch": 0.21172, "grad_norm": 1.3890064273623446, "learning_rate": 0.003, "loss": 4.0731, "step": 21172 }, { "epoch": 0.21173, "grad_norm": 1.153878422157314, "learning_rate": 0.003, "loss": 4.0512, "step": 21173 }, { "epoch": 0.21174, "grad_norm": 1.4317029178580625, "learning_rate": 0.003, "loss": 4.0569, "step": 21174 }, { "epoch": 0.21175, "grad_norm": 1.217479259645463, "learning_rate": 0.003, "loss": 4.0606, "step": 21175 }, { "epoch": 0.21176, "grad_norm": 1.1898684451219492, "learning_rate": 0.003, "loss": 4.0615, "step": 21176 }, { "epoch": 0.21177, "grad_norm": 1.2770393642886273, "learning_rate": 0.003, "loss": 4.0438, "step": 21177 }, { "epoch": 0.21178, "grad_norm": 1.233163842181116, "learning_rate": 0.003, "loss": 4.0128, "step": 21178 }, { "epoch": 0.21179, "grad_norm": 1.1305383322900322, "learning_rate": 0.003, "loss": 4.0638, "step": 21179 }, { "epoch": 0.2118, "grad_norm": 1.2239972747122971, "learning_rate": 0.003, "loss": 4.0333, "step": 21180 }, { "epoch": 0.21181, "grad_norm": 1.3121720290800374, "learning_rate": 0.003, "loss": 4.023, "step": 21181 }, { "epoch": 0.21182, "grad_norm": 1.5015691005417624, "learning_rate": 0.003, "loss": 4.0509, "step": 21182 }, { "epoch": 0.21183, "grad_norm": 1.4855617135506352, "learning_rate": 0.003, "loss": 4.0468, "step": 21183 }, { "epoch": 0.21184, "grad_norm": 1.363701504514968, "learning_rate": 0.003, "loss": 4.0662, "step": 21184 }, { "epoch": 0.21185, "grad_norm": 1.159770894014433, "learning_rate": 0.003, "loss": 4.0478, "step": 21185 }, { "epoch": 0.21186, "grad_norm": 1.4774361893297834, "learning_rate": 0.003, "loss": 4.0741, "step": 21186 }, { "epoch": 0.21187, "grad_norm": 1.1007616891834122, "learning_rate": 0.003, "loss": 4.0325, "step": 21187 }, { "epoch": 0.21188, "grad_norm": 1.2487380713948097, "learning_rate": 0.003, "loss": 4.0321, "step": 21188 }, { "epoch": 0.21189, "grad_norm": 1.3618169855089544, "learning_rate": 0.003, "loss": 4.0237, "step": 21189 }, { "epoch": 0.2119, "grad_norm": 1.201925785707879, "learning_rate": 0.003, "loss": 4.0556, "step": 21190 }, { "epoch": 0.21191, "grad_norm": 1.3024386752170385, "learning_rate": 0.003, "loss": 4.0517, "step": 21191 }, { "epoch": 0.21192, "grad_norm": 1.1672279223852937, "learning_rate": 0.003, "loss": 4.0396, "step": 21192 }, { "epoch": 0.21193, "grad_norm": 1.3705558796138086, "learning_rate": 0.003, "loss": 4.0233, "step": 21193 }, { "epoch": 0.21194, "grad_norm": 1.4762161050778222, "learning_rate": 0.003, "loss": 4.0834, "step": 21194 }, { "epoch": 0.21195, "grad_norm": 1.2062317526880664, "learning_rate": 0.003, "loss": 4.0495, "step": 21195 }, { "epoch": 0.21196, "grad_norm": 1.340626791633665, "learning_rate": 0.003, "loss": 4.0691, "step": 21196 }, { "epoch": 0.21197, "grad_norm": 1.2354014350222875, "learning_rate": 0.003, "loss": 4.0553, "step": 21197 }, { "epoch": 0.21198, "grad_norm": 1.5150865936875821, "learning_rate": 0.003, "loss": 4.0503, "step": 21198 }, { "epoch": 0.21199, "grad_norm": 1.1791380457988314, "learning_rate": 0.003, "loss": 4.0362, "step": 21199 }, { "epoch": 0.212, "grad_norm": 1.4960503083917507, "learning_rate": 0.003, "loss": 4.0397, "step": 21200 }, { "epoch": 0.21201, "grad_norm": 1.0226116936688727, "learning_rate": 0.003, "loss": 4.0373, "step": 21201 }, { "epoch": 0.21202, "grad_norm": 1.4619565029751385, "learning_rate": 0.003, "loss": 4.0402, "step": 21202 }, { "epoch": 0.21203, "grad_norm": 1.0822958599251975, "learning_rate": 0.003, "loss": 4.052, "step": 21203 }, { "epoch": 0.21204, "grad_norm": 1.44743948852276, "learning_rate": 0.003, "loss": 4.0555, "step": 21204 }, { "epoch": 0.21205, "grad_norm": 1.0735995023147238, "learning_rate": 0.003, "loss": 4.0474, "step": 21205 }, { "epoch": 0.21206, "grad_norm": 1.4273899604773974, "learning_rate": 0.003, "loss": 4.0517, "step": 21206 }, { "epoch": 0.21207, "grad_norm": 1.0711027752697948, "learning_rate": 0.003, "loss": 4.0857, "step": 21207 }, { "epoch": 0.21208, "grad_norm": 1.5263229391706197, "learning_rate": 0.003, "loss": 4.087, "step": 21208 }, { "epoch": 0.21209, "grad_norm": 0.9509188211685536, "learning_rate": 0.003, "loss": 4.0381, "step": 21209 }, { "epoch": 0.2121, "grad_norm": 1.1490255252840045, "learning_rate": 0.003, "loss": 4.0523, "step": 21210 }, { "epoch": 0.21211, "grad_norm": 1.3422143862668483, "learning_rate": 0.003, "loss": 4.0606, "step": 21211 }, { "epoch": 0.21212, "grad_norm": 1.2539022969418465, "learning_rate": 0.003, "loss": 4.0488, "step": 21212 }, { "epoch": 0.21213, "grad_norm": 1.8629431414796072, "learning_rate": 0.003, "loss": 4.0647, "step": 21213 }, { "epoch": 0.21214, "grad_norm": 1.0436144942927696, "learning_rate": 0.003, "loss": 4.0385, "step": 21214 }, { "epoch": 0.21215, "grad_norm": 1.2701470718243433, "learning_rate": 0.003, "loss": 4.0279, "step": 21215 }, { "epoch": 0.21216, "grad_norm": 1.2025971490312357, "learning_rate": 0.003, "loss": 4.0443, "step": 21216 }, { "epoch": 0.21217, "grad_norm": 1.1863459257692672, "learning_rate": 0.003, "loss": 4.03, "step": 21217 }, { "epoch": 0.21218, "grad_norm": 1.2931599020212343, "learning_rate": 0.003, "loss": 4.0463, "step": 21218 }, { "epoch": 0.21219, "grad_norm": 1.166605540713005, "learning_rate": 0.003, "loss": 4.06, "step": 21219 }, { "epoch": 0.2122, "grad_norm": 1.3880922740795296, "learning_rate": 0.003, "loss": 4.0776, "step": 21220 }, { "epoch": 0.21221, "grad_norm": 1.2206694900056876, "learning_rate": 0.003, "loss": 4.0562, "step": 21221 }, { "epoch": 0.21222, "grad_norm": 1.3585826498780238, "learning_rate": 0.003, "loss": 4.0464, "step": 21222 }, { "epoch": 0.21223, "grad_norm": 1.317578632988024, "learning_rate": 0.003, "loss": 4.0526, "step": 21223 }, { "epoch": 0.21224, "grad_norm": 1.2874456251618371, "learning_rate": 0.003, "loss": 4.0376, "step": 21224 }, { "epoch": 0.21225, "grad_norm": 1.241267859205183, "learning_rate": 0.003, "loss": 4.019, "step": 21225 }, { "epoch": 0.21226, "grad_norm": 1.3982627897056497, "learning_rate": 0.003, "loss": 4.0607, "step": 21226 }, { "epoch": 0.21227, "grad_norm": 1.395875156197999, "learning_rate": 0.003, "loss": 4.0656, "step": 21227 }, { "epoch": 0.21228, "grad_norm": 1.355288621035149, "learning_rate": 0.003, "loss": 4.0514, "step": 21228 }, { "epoch": 0.21229, "grad_norm": 1.3051451099291878, "learning_rate": 0.003, "loss": 4.0552, "step": 21229 }, { "epoch": 0.2123, "grad_norm": 1.4068723537818764, "learning_rate": 0.003, "loss": 4.0455, "step": 21230 }, { "epoch": 0.21231, "grad_norm": 1.1396637306136057, "learning_rate": 0.003, "loss": 4.0667, "step": 21231 }, { "epoch": 0.21232, "grad_norm": 1.3511158184996983, "learning_rate": 0.003, "loss": 4.0229, "step": 21232 }, { "epoch": 0.21233, "grad_norm": 1.2450881997501164, "learning_rate": 0.003, "loss": 4.0472, "step": 21233 }, { "epoch": 0.21234, "grad_norm": 1.2643725323698822, "learning_rate": 0.003, "loss": 4.0487, "step": 21234 }, { "epoch": 0.21235, "grad_norm": 1.2409392883847383, "learning_rate": 0.003, "loss": 4.0535, "step": 21235 }, { "epoch": 0.21236, "grad_norm": 1.6763642327401815, "learning_rate": 0.003, "loss": 4.0351, "step": 21236 }, { "epoch": 0.21237, "grad_norm": 1.2025489336682045, "learning_rate": 0.003, "loss": 4.0297, "step": 21237 }, { "epoch": 0.21238, "grad_norm": 1.3302806801590354, "learning_rate": 0.003, "loss": 4.0559, "step": 21238 }, { "epoch": 0.21239, "grad_norm": 1.1525403685678053, "learning_rate": 0.003, "loss": 4.0422, "step": 21239 }, { "epoch": 0.2124, "grad_norm": 1.3638264634999357, "learning_rate": 0.003, "loss": 4.0286, "step": 21240 }, { "epoch": 0.21241, "grad_norm": 1.0902909893581767, "learning_rate": 0.003, "loss": 4.0621, "step": 21241 }, { "epoch": 0.21242, "grad_norm": 1.3797642168018844, "learning_rate": 0.003, "loss": 4.0382, "step": 21242 }, { "epoch": 0.21243, "grad_norm": 1.2206955407043554, "learning_rate": 0.003, "loss": 4.0385, "step": 21243 }, { "epoch": 0.21244, "grad_norm": 1.5156553552816339, "learning_rate": 0.003, "loss": 4.045, "step": 21244 }, { "epoch": 0.21245, "grad_norm": 0.9515384315876183, "learning_rate": 0.003, "loss": 4.0472, "step": 21245 }, { "epoch": 0.21246, "grad_norm": 1.361146046697627, "learning_rate": 0.003, "loss": 4.056, "step": 21246 }, { "epoch": 0.21247, "grad_norm": 1.5913645215564736, "learning_rate": 0.003, "loss": 4.055, "step": 21247 }, { "epoch": 0.21248, "grad_norm": 1.4370052161228561, "learning_rate": 0.003, "loss": 4.0438, "step": 21248 }, { "epoch": 0.21249, "grad_norm": 1.2309898212193815, "learning_rate": 0.003, "loss": 4.0648, "step": 21249 }, { "epoch": 0.2125, "grad_norm": 1.4114840886409648, "learning_rate": 0.003, "loss": 4.0779, "step": 21250 }, { "epoch": 0.21251, "grad_norm": 1.1537287913840821, "learning_rate": 0.003, "loss": 4.0663, "step": 21251 }, { "epoch": 0.21252, "grad_norm": 1.2553623291546478, "learning_rate": 0.003, "loss": 4.0661, "step": 21252 }, { "epoch": 0.21253, "grad_norm": 1.3921081641908704, "learning_rate": 0.003, "loss": 4.0516, "step": 21253 }, { "epoch": 0.21254, "grad_norm": 1.1053487916938944, "learning_rate": 0.003, "loss": 4.0331, "step": 21254 }, { "epoch": 0.21255, "grad_norm": 1.8061320320678969, "learning_rate": 0.003, "loss": 4.0314, "step": 21255 }, { "epoch": 0.21256, "grad_norm": 0.8969966868615841, "learning_rate": 0.003, "loss": 4.0331, "step": 21256 }, { "epoch": 0.21257, "grad_norm": 1.205780940445784, "learning_rate": 0.003, "loss": 4.0591, "step": 21257 }, { "epoch": 0.21258, "grad_norm": 1.3600689626589804, "learning_rate": 0.003, "loss": 4.0598, "step": 21258 }, { "epoch": 0.21259, "grad_norm": 1.3242629843533986, "learning_rate": 0.003, "loss": 4.0486, "step": 21259 }, { "epoch": 0.2126, "grad_norm": 1.3834817194882478, "learning_rate": 0.003, "loss": 4.0618, "step": 21260 }, { "epoch": 0.21261, "grad_norm": 1.0838249822303874, "learning_rate": 0.003, "loss": 4.0579, "step": 21261 }, { "epoch": 0.21262, "grad_norm": 1.5128461510514026, "learning_rate": 0.003, "loss": 4.0372, "step": 21262 }, { "epoch": 0.21263, "grad_norm": 1.1827106561363243, "learning_rate": 0.003, "loss": 4.0093, "step": 21263 }, { "epoch": 0.21264, "grad_norm": 1.3315292703566595, "learning_rate": 0.003, "loss": 4.0186, "step": 21264 }, { "epoch": 0.21265, "grad_norm": 1.1555146028756413, "learning_rate": 0.003, "loss": 4.0446, "step": 21265 }, { "epoch": 0.21266, "grad_norm": 1.4866501663257243, "learning_rate": 0.003, "loss": 4.0508, "step": 21266 }, { "epoch": 0.21267, "grad_norm": 1.4051336614763095, "learning_rate": 0.003, "loss": 4.0508, "step": 21267 }, { "epoch": 0.21268, "grad_norm": 1.1328052502105757, "learning_rate": 0.003, "loss": 4.0485, "step": 21268 }, { "epoch": 0.21269, "grad_norm": 1.2898245831393798, "learning_rate": 0.003, "loss": 4.066, "step": 21269 }, { "epoch": 0.2127, "grad_norm": 1.3953673594523057, "learning_rate": 0.003, "loss": 4.0234, "step": 21270 }, { "epoch": 0.21271, "grad_norm": 1.131002090499698, "learning_rate": 0.003, "loss": 4.07, "step": 21271 }, { "epoch": 0.21272, "grad_norm": 1.4609953205099613, "learning_rate": 0.003, "loss": 4.0289, "step": 21272 }, { "epoch": 0.21273, "grad_norm": 1.1210795678224974, "learning_rate": 0.003, "loss": 4.0474, "step": 21273 }, { "epoch": 0.21274, "grad_norm": 1.410929257076832, "learning_rate": 0.003, "loss": 4.0838, "step": 21274 }, { "epoch": 0.21275, "grad_norm": 1.2311509749398506, "learning_rate": 0.003, "loss": 4.0094, "step": 21275 }, { "epoch": 0.21276, "grad_norm": 1.4049179781701777, "learning_rate": 0.003, "loss": 4.048, "step": 21276 }, { "epoch": 0.21277, "grad_norm": 1.099506667330958, "learning_rate": 0.003, "loss": 4.0195, "step": 21277 }, { "epoch": 0.21278, "grad_norm": 1.3875113272449338, "learning_rate": 0.003, "loss": 4.0499, "step": 21278 }, { "epoch": 0.21279, "grad_norm": 1.2558224446512296, "learning_rate": 0.003, "loss": 4.0463, "step": 21279 }, { "epoch": 0.2128, "grad_norm": 1.3544694177374053, "learning_rate": 0.003, "loss": 4.0308, "step": 21280 }, { "epoch": 0.21281, "grad_norm": 1.2628164865838394, "learning_rate": 0.003, "loss": 4.0435, "step": 21281 }, { "epoch": 0.21282, "grad_norm": 1.094287994161794, "learning_rate": 0.003, "loss": 4.0584, "step": 21282 }, { "epoch": 0.21283, "grad_norm": 1.2896996630861157, "learning_rate": 0.003, "loss": 4.0226, "step": 21283 }, { "epoch": 0.21284, "grad_norm": 1.1074798369391314, "learning_rate": 0.003, "loss": 4.0412, "step": 21284 }, { "epoch": 0.21285, "grad_norm": 1.3538423443692602, "learning_rate": 0.003, "loss": 4.0776, "step": 21285 }, { "epoch": 0.21286, "grad_norm": 1.1110113965840402, "learning_rate": 0.003, "loss": 4.0302, "step": 21286 }, { "epoch": 0.21287, "grad_norm": 1.2221320233309492, "learning_rate": 0.003, "loss": 4.0336, "step": 21287 }, { "epoch": 0.21288, "grad_norm": 1.2271627263198994, "learning_rate": 0.003, "loss": 4.0462, "step": 21288 }, { "epoch": 0.21289, "grad_norm": 1.2211200333212375, "learning_rate": 0.003, "loss": 4.0393, "step": 21289 }, { "epoch": 0.2129, "grad_norm": 1.1597470812427897, "learning_rate": 0.003, "loss": 4.0159, "step": 21290 }, { "epoch": 0.21291, "grad_norm": 1.1865759125638362, "learning_rate": 0.003, "loss": 4.0524, "step": 21291 }, { "epoch": 0.21292, "grad_norm": 1.2334931450457307, "learning_rate": 0.003, "loss": 4.0363, "step": 21292 }, { "epoch": 0.21293, "grad_norm": 1.3137351543194178, "learning_rate": 0.003, "loss": 4.0433, "step": 21293 }, { "epoch": 0.21294, "grad_norm": 1.2310978816330074, "learning_rate": 0.003, "loss": 4.0683, "step": 21294 }, { "epoch": 0.21295, "grad_norm": 1.4940754079648613, "learning_rate": 0.003, "loss": 4.0604, "step": 21295 }, { "epoch": 0.21296, "grad_norm": 1.3409897686951235, "learning_rate": 0.003, "loss": 4.0546, "step": 21296 }, { "epoch": 0.21297, "grad_norm": 1.2704738218133316, "learning_rate": 0.003, "loss": 4.0453, "step": 21297 }, { "epoch": 0.21298, "grad_norm": 1.2609362505399053, "learning_rate": 0.003, "loss": 4.0851, "step": 21298 }, { "epoch": 0.21299, "grad_norm": 1.225084252513662, "learning_rate": 0.003, "loss": 4.0574, "step": 21299 }, { "epoch": 0.213, "grad_norm": 1.20988265938044, "learning_rate": 0.003, "loss": 4.0502, "step": 21300 }, { "epoch": 0.21301, "grad_norm": 1.4497084415618406, "learning_rate": 0.003, "loss": 4.0429, "step": 21301 }, { "epoch": 0.21302, "grad_norm": 1.193484205546864, "learning_rate": 0.003, "loss": 4.0359, "step": 21302 }, { "epoch": 0.21303, "grad_norm": 1.2345370413341212, "learning_rate": 0.003, "loss": 4.0439, "step": 21303 }, { "epoch": 0.21304, "grad_norm": 1.318776212154148, "learning_rate": 0.003, "loss": 4.0264, "step": 21304 }, { "epoch": 0.21305, "grad_norm": 1.6794035477830318, "learning_rate": 0.003, "loss": 4.038, "step": 21305 }, { "epoch": 0.21306, "grad_norm": 1.298167289102819, "learning_rate": 0.003, "loss": 4.0437, "step": 21306 }, { "epoch": 0.21307, "grad_norm": 1.1408218010914621, "learning_rate": 0.003, "loss": 4.0261, "step": 21307 }, { "epoch": 0.21308, "grad_norm": 1.4345539439571031, "learning_rate": 0.003, "loss": 4.0565, "step": 21308 }, { "epoch": 0.21309, "grad_norm": 1.08308841421971, "learning_rate": 0.003, "loss": 4.0041, "step": 21309 }, { "epoch": 0.2131, "grad_norm": 1.323940917710682, "learning_rate": 0.003, "loss": 4.0237, "step": 21310 }, { "epoch": 0.21311, "grad_norm": 1.2696573808223026, "learning_rate": 0.003, "loss": 4.0444, "step": 21311 }, { "epoch": 0.21312, "grad_norm": 1.1996852266168259, "learning_rate": 0.003, "loss": 4.038, "step": 21312 }, { "epoch": 0.21313, "grad_norm": 1.369723073919903, "learning_rate": 0.003, "loss": 4.0505, "step": 21313 }, { "epoch": 0.21314, "grad_norm": 1.28562190993862, "learning_rate": 0.003, "loss": 4.0414, "step": 21314 }, { "epoch": 0.21315, "grad_norm": 1.446931202538823, "learning_rate": 0.003, "loss": 4.0631, "step": 21315 }, { "epoch": 0.21316, "grad_norm": 1.2156284322875552, "learning_rate": 0.003, "loss": 4.0643, "step": 21316 }, { "epoch": 0.21317, "grad_norm": 1.3079028561141957, "learning_rate": 0.003, "loss": 4.0721, "step": 21317 }, { "epoch": 0.21318, "grad_norm": 1.3649474090891291, "learning_rate": 0.003, "loss": 4.0605, "step": 21318 }, { "epoch": 0.21319, "grad_norm": 1.2606745758198419, "learning_rate": 0.003, "loss": 4.021, "step": 21319 }, { "epoch": 0.2132, "grad_norm": 1.4084878363128626, "learning_rate": 0.003, "loss": 4.0629, "step": 21320 }, { "epoch": 0.21321, "grad_norm": 1.2936532541666024, "learning_rate": 0.003, "loss": 4.0383, "step": 21321 }, { "epoch": 0.21322, "grad_norm": 1.38838107459198, "learning_rate": 0.003, "loss": 4.0499, "step": 21322 }, { "epoch": 0.21323, "grad_norm": 1.1583770792135066, "learning_rate": 0.003, "loss": 4.0341, "step": 21323 }, { "epoch": 0.21324, "grad_norm": 1.2580253757229194, "learning_rate": 0.003, "loss": 4.0434, "step": 21324 }, { "epoch": 0.21325, "grad_norm": 1.2897936310118077, "learning_rate": 0.003, "loss": 4.0317, "step": 21325 }, { "epoch": 0.21326, "grad_norm": 1.3398874636543947, "learning_rate": 0.003, "loss": 4.0639, "step": 21326 }, { "epoch": 0.21327, "grad_norm": 1.2076014768358647, "learning_rate": 0.003, "loss": 4.0271, "step": 21327 }, { "epoch": 0.21328, "grad_norm": 1.2883759373541221, "learning_rate": 0.003, "loss": 4.0207, "step": 21328 }, { "epoch": 0.21329, "grad_norm": 1.3162283800326189, "learning_rate": 0.003, "loss": 4.0395, "step": 21329 }, { "epoch": 0.2133, "grad_norm": 1.250449034147522, "learning_rate": 0.003, "loss": 4.054, "step": 21330 }, { "epoch": 0.21331, "grad_norm": 1.4278049635796872, "learning_rate": 0.003, "loss": 4.0556, "step": 21331 }, { "epoch": 0.21332, "grad_norm": 1.14298792973109, "learning_rate": 0.003, "loss": 4.0525, "step": 21332 }, { "epoch": 0.21333, "grad_norm": 1.6626489515816607, "learning_rate": 0.003, "loss": 4.0439, "step": 21333 }, { "epoch": 0.21334, "grad_norm": 1.2030828894352494, "learning_rate": 0.003, "loss": 4.0331, "step": 21334 }, { "epoch": 0.21335, "grad_norm": 1.349062438237261, "learning_rate": 0.003, "loss": 4.052, "step": 21335 }, { "epoch": 0.21336, "grad_norm": 1.3795838496859136, "learning_rate": 0.003, "loss": 4.0429, "step": 21336 }, { "epoch": 0.21337, "grad_norm": 1.0649483702192992, "learning_rate": 0.003, "loss": 4.052, "step": 21337 }, { "epoch": 0.21338, "grad_norm": 1.508413228100052, "learning_rate": 0.003, "loss": 4.0672, "step": 21338 }, { "epoch": 0.21339, "grad_norm": 1.0765460212141766, "learning_rate": 0.003, "loss": 4.0518, "step": 21339 }, { "epoch": 0.2134, "grad_norm": 1.4888043428112916, "learning_rate": 0.003, "loss": 4.0641, "step": 21340 }, { "epoch": 0.21341, "grad_norm": 0.9995688586010715, "learning_rate": 0.003, "loss": 4.0629, "step": 21341 }, { "epoch": 0.21342, "grad_norm": 1.3972463254192167, "learning_rate": 0.003, "loss": 4.0629, "step": 21342 }, { "epoch": 0.21343, "grad_norm": 1.107999971944543, "learning_rate": 0.003, "loss": 4.0598, "step": 21343 }, { "epoch": 0.21344, "grad_norm": 1.3448387423992543, "learning_rate": 0.003, "loss": 4.0564, "step": 21344 }, { "epoch": 0.21345, "grad_norm": 1.2637795248816244, "learning_rate": 0.003, "loss": 4.062, "step": 21345 }, { "epoch": 0.21346, "grad_norm": 1.0118101759825742, "learning_rate": 0.003, "loss": 4.0435, "step": 21346 }, { "epoch": 0.21347, "grad_norm": 1.4459521110402487, "learning_rate": 0.003, "loss": 4.0716, "step": 21347 }, { "epoch": 0.21348, "grad_norm": 1.0311863301319009, "learning_rate": 0.003, "loss": 4.0596, "step": 21348 }, { "epoch": 0.21349, "grad_norm": 1.8475944530552786, "learning_rate": 0.003, "loss": 4.0571, "step": 21349 }, { "epoch": 0.2135, "grad_norm": 1.138540427507579, "learning_rate": 0.003, "loss": 4.0524, "step": 21350 }, { "epoch": 0.21351, "grad_norm": 1.3721821472192584, "learning_rate": 0.003, "loss": 4.0634, "step": 21351 }, { "epoch": 0.21352, "grad_norm": 1.4142680422434566, "learning_rate": 0.003, "loss": 4.0631, "step": 21352 }, { "epoch": 0.21353, "grad_norm": 1.3751571157419424, "learning_rate": 0.003, "loss": 4.0546, "step": 21353 }, { "epoch": 0.21354, "grad_norm": 1.1920477240206362, "learning_rate": 0.003, "loss": 4.0607, "step": 21354 }, { "epoch": 0.21355, "grad_norm": 1.3629304224733283, "learning_rate": 0.003, "loss": 4.0389, "step": 21355 }, { "epoch": 0.21356, "grad_norm": 1.2024918478466937, "learning_rate": 0.003, "loss": 4.0601, "step": 21356 }, { "epoch": 0.21357, "grad_norm": 1.1978202577753532, "learning_rate": 0.003, "loss": 4.0667, "step": 21357 }, { "epoch": 0.21358, "grad_norm": 1.1474077755882799, "learning_rate": 0.003, "loss": 4.0288, "step": 21358 }, { "epoch": 0.21359, "grad_norm": 1.3273962868251463, "learning_rate": 0.003, "loss": 4.0482, "step": 21359 }, { "epoch": 0.2136, "grad_norm": 1.2976154120167893, "learning_rate": 0.003, "loss": 4.0498, "step": 21360 }, { "epoch": 0.21361, "grad_norm": 1.8210705771178908, "learning_rate": 0.003, "loss": 4.0555, "step": 21361 }, { "epoch": 0.21362, "grad_norm": 1.0567342453250417, "learning_rate": 0.003, "loss": 4.0631, "step": 21362 }, { "epoch": 0.21363, "grad_norm": 1.5856935209757512, "learning_rate": 0.003, "loss": 4.0652, "step": 21363 }, { "epoch": 0.21364, "grad_norm": 1.2636879371466914, "learning_rate": 0.003, "loss": 4.0581, "step": 21364 }, { "epoch": 0.21365, "grad_norm": 1.2705957404981245, "learning_rate": 0.003, "loss": 4.0465, "step": 21365 }, { "epoch": 0.21366, "grad_norm": 1.2541463528391474, "learning_rate": 0.003, "loss": 4.062, "step": 21366 }, { "epoch": 0.21367, "grad_norm": 1.3865654261817257, "learning_rate": 0.003, "loss": 4.0481, "step": 21367 }, { "epoch": 0.21368, "grad_norm": 1.4439535448527894, "learning_rate": 0.003, "loss": 4.064, "step": 21368 }, { "epoch": 0.21369, "grad_norm": 1.627448150006948, "learning_rate": 0.003, "loss": 4.0319, "step": 21369 }, { "epoch": 0.2137, "grad_norm": 1.0886093563909551, "learning_rate": 0.003, "loss": 4.0878, "step": 21370 }, { "epoch": 0.21371, "grad_norm": 1.282950003680138, "learning_rate": 0.003, "loss": 4.0731, "step": 21371 }, { "epoch": 0.21372, "grad_norm": 1.312882928122349, "learning_rate": 0.003, "loss": 4.0861, "step": 21372 }, { "epoch": 0.21373, "grad_norm": 1.4587166915301903, "learning_rate": 0.003, "loss": 4.0599, "step": 21373 }, { "epoch": 0.21374, "grad_norm": 1.2093497165791725, "learning_rate": 0.003, "loss": 4.0426, "step": 21374 }, { "epoch": 0.21375, "grad_norm": 1.2743170494067226, "learning_rate": 0.003, "loss": 4.0349, "step": 21375 }, { "epoch": 0.21376, "grad_norm": 1.3201371885098845, "learning_rate": 0.003, "loss": 4.0371, "step": 21376 }, { "epoch": 0.21377, "grad_norm": 1.260209169861331, "learning_rate": 0.003, "loss": 4.0674, "step": 21377 }, { "epoch": 0.21378, "grad_norm": 1.4904707031555913, "learning_rate": 0.003, "loss": 4.0448, "step": 21378 }, { "epoch": 0.21379, "grad_norm": 1.0340463707054346, "learning_rate": 0.003, "loss": 4.0501, "step": 21379 }, { "epoch": 0.2138, "grad_norm": 1.5150284932481612, "learning_rate": 0.003, "loss": 4.0559, "step": 21380 }, { "epoch": 0.21381, "grad_norm": 1.1273990458283891, "learning_rate": 0.003, "loss": 4.0418, "step": 21381 }, { "epoch": 0.21382, "grad_norm": 1.3210726311942573, "learning_rate": 0.003, "loss": 4.0168, "step": 21382 }, { "epoch": 0.21383, "grad_norm": 1.1069027042636328, "learning_rate": 0.003, "loss": 4.0447, "step": 21383 }, { "epoch": 0.21384, "grad_norm": 1.3092370982697676, "learning_rate": 0.003, "loss": 4.0295, "step": 21384 }, { "epoch": 0.21385, "grad_norm": 1.6472511093567705, "learning_rate": 0.003, "loss": 4.0421, "step": 21385 }, { "epoch": 0.21386, "grad_norm": 1.1252433906386323, "learning_rate": 0.003, "loss": 4.0384, "step": 21386 }, { "epoch": 0.21387, "grad_norm": 1.624161635575557, "learning_rate": 0.003, "loss": 4.0357, "step": 21387 }, { "epoch": 0.21388, "grad_norm": 0.9656518628738969, "learning_rate": 0.003, "loss": 4.0387, "step": 21388 }, { "epoch": 0.21389, "grad_norm": 1.51171735229888, "learning_rate": 0.003, "loss": 4.0693, "step": 21389 }, { "epoch": 0.2139, "grad_norm": 1.2277851492776417, "learning_rate": 0.003, "loss": 4.032, "step": 21390 }, { "epoch": 0.21391, "grad_norm": 1.584747677568358, "learning_rate": 0.003, "loss": 4.0439, "step": 21391 }, { "epoch": 0.21392, "grad_norm": 1.202180154252793, "learning_rate": 0.003, "loss": 4.0565, "step": 21392 }, { "epoch": 0.21393, "grad_norm": 1.1697463543298947, "learning_rate": 0.003, "loss": 4.0462, "step": 21393 }, { "epoch": 0.21394, "grad_norm": 1.4453891958788705, "learning_rate": 0.003, "loss": 4.0725, "step": 21394 }, { "epoch": 0.21395, "grad_norm": 1.0284231831226134, "learning_rate": 0.003, "loss": 4.0509, "step": 21395 }, { "epoch": 0.21396, "grad_norm": 1.3694082143995245, "learning_rate": 0.003, "loss": 4.0403, "step": 21396 }, { "epoch": 0.21397, "grad_norm": 1.3419886484533383, "learning_rate": 0.003, "loss": 4.0638, "step": 21397 }, { "epoch": 0.21398, "grad_norm": 1.1957459993332311, "learning_rate": 0.003, "loss": 4.0498, "step": 21398 }, { "epoch": 0.21399, "grad_norm": 1.2523630190898598, "learning_rate": 0.003, "loss": 4.0497, "step": 21399 }, { "epoch": 0.214, "grad_norm": 1.2590213585149514, "learning_rate": 0.003, "loss": 4.0787, "step": 21400 }, { "epoch": 0.21401, "grad_norm": 1.2158586422954252, "learning_rate": 0.003, "loss": 4.0556, "step": 21401 }, { "epoch": 0.21402, "grad_norm": 1.1927925220634177, "learning_rate": 0.003, "loss": 4.051, "step": 21402 }, { "epoch": 0.21403, "grad_norm": 1.3230329876727829, "learning_rate": 0.003, "loss": 4.0667, "step": 21403 }, { "epoch": 0.21404, "grad_norm": 1.3948266556870443, "learning_rate": 0.003, "loss": 4.0561, "step": 21404 }, { "epoch": 0.21405, "grad_norm": 1.4600556127552742, "learning_rate": 0.003, "loss": 4.0442, "step": 21405 }, { "epoch": 0.21406, "grad_norm": 1.023712788436025, "learning_rate": 0.003, "loss": 4.0267, "step": 21406 }, { "epoch": 0.21407, "grad_norm": 1.59571712199933, "learning_rate": 0.003, "loss": 4.0723, "step": 21407 }, { "epoch": 0.21408, "grad_norm": 0.9265689920715735, "learning_rate": 0.003, "loss": 4.0148, "step": 21408 }, { "epoch": 0.21409, "grad_norm": 1.2559292429582913, "learning_rate": 0.003, "loss": 4.0646, "step": 21409 }, { "epoch": 0.2141, "grad_norm": 1.356898234285401, "learning_rate": 0.003, "loss": 4.0376, "step": 21410 }, { "epoch": 0.21411, "grad_norm": 1.3822189229917954, "learning_rate": 0.003, "loss": 4.0532, "step": 21411 }, { "epoch": 0.21412, "grad_norm": 1.1456331216536773, "learning_rate": 0.003, "loss": 4.0592, "step": 21412 }, { "epoch": 0.21413, "grad_norm": 1.3884648494073832, "learning_rate": 0.003, "loss": 4.0613, "step": 21413 }, { "epoch": 0.21414, "grad_norm": 1.3388728832041317, "learning_rate": 0.003, "loss": 4.0125, "step": 21414 }, { "epoch": 0.21415, "grad_norm": 1.083806039595083, "learning_rate": 0.003, "loss": 4.0279, "step": 21415 }, { "epoch": 0.21416, "grad_norm": 1.3370142619299101, "learning_rate": 0.003, "loss": 4.0353, "step": 21416 }, { "epoch": 0.21417, "grad_norm": 1.2235260078732255, "learning_rate": 0.003, "loss": 4.042, "step": 21417 }, { "epoch": 0.21418, "grad_norm": 1.2795875579430278, "learning_rate": 0.003, "loss": 4.0608, "step": 21418 }, { "epoch": 0.21419, "grad_norm": 1.4309625949091171, "learning_rate": 0.003, "loss": 4.0526, "step": 21419 }, { "epoch": 0.2142, "grad_norm": 1.585382629070091, "learning_rate": 0.003, "loss": 4.0517, "step": 21420 }, { "epoch": 0.21421, "grad_norm": 1.006334571201957, "learning_rate": 0.003, "loss": 4.0539, "step": 21421 }, { "epoch": 0.21422, "grad_norm": 1.4230217197820794, "learning_rate": 0.003, "loss": 4.0578, "step": 21422 }, { "epoch": 0.21423, "grad_norm": 1.1473431232363076, "learning_rate": 0.003, "loss": 4.0616, "step": 21423 }, { "epoch": 0.21424, "grad_norm": 1.406963978924778, "learning_rate": 0.003, "loss": 4.0486, "step": 21424 }, { "epoch": 0.21425, "grad_norm": 1.2102770648595627, "learning_rate": 0.003, "loss": 4.0751, "step": 21425 }, { "epoch": 0.21426, "grad_norm": 1.362648793885832, "learning_rate": 0.003, "loss": 4.0723, "step": 21426 }, { "epoch": 0.21427, "grad_norm": 1.2221966631120762, "learning_rate": 0.003, "loss": 4.0644, "step": 21427 }, { "epoch": 0.21428, "grad_norm": 1.2148550249578498, "learning_rate": 0.003, "loss": 4.0427, "step": 21428 }, { "epoch": 0.21429, "grad_norm": 1.3770428805474857, "learning_rate": 0.003, "loss": 4.0426, "step": 21429 }, { "epoch": 0.2143, "grad_norm": 1.3735600093893532, "learning_rate": 0.003, "loss": 4.0449, "step": 21430 }, { "epoch": 0.21431, "grad_norm": 1.247480959233962, "learning_rate": 0.003, "loss": 4.0681, "step": 21431 }, { "epoch": 0.21432, "grad_norm": 1.0875437244340893, "learning_rate": 0.003, "loss": 4.0626, "step": 21432 }, { "epoch": 0.21433, "grad_norm": 1.5089941568181857, "learning_rate": 0.003, "loss": 4.0485, "step": 21433 }, { "epoch": 0.21434, "grad_norm": 1.0593319177878011, "learning_rate": 0.003, "loss": 4.0233, "step": 21434 }, { "epoch": 0.21435, "grad_norm": 1.4208541942764379, "learning_rate": 0.003, "loss": 4.0473, "step": 21435 }, { "epoch": 0.21436, "grad_norm": 0.9393903080155143, "learning_rate": 0.003, "loss": 4.0392, "step": 21436 }, { "epoch": 0.21437, "grad_norm": 1.591779273895513, "learning_rate": 0.003, "loss": 4.0403, "step": 21437 }, { "epoch": 0.21438, "grad_norm": 1.1622175569073974, "learning_rate": 0.003, "loss": 4.0329, "step": 21438 }, { "epoch": 0.21439, "grad_norm": 1.6971122172312312, "learning_rate": 0.003, "loss": 4.0755, "step": 21439 }, { "epoch": 0.2144, "grad_norm": 1.1780489518286923, "learning_rate": 0.003, "loss": 4.0278, "step": 21440 }, { "epoch": 0.21441, "grad_norm": 1.0365230744754172, "learning_rate": 0.003, "loss": 4.0302, "step": 21441 }, { "epoch": 0.21442, "grad_norm": 1.4104533794855865, "learning_rate": 0.003, "loss": 4.0788, "step": 21442 }, { "epoch": 0.21443, "grad_norm": 1.2555781924185174, "learning_rate": 0.003, "loss": 4.0756, "step": 21443 }, { "epoch": 0.21444, "grad_norm": 1.4190495250147561, "learning_rate": 0.003, "loss": 4.0903, "step": 21444 }, { "epoch": 0.21445, "grad_norm": 1.2594779709007058, "learning_rate": 0.003, "loss": 4.0771, "step": 21445 }, { "epoch": 0.21446, "grad_norm": 1.437264589849899, "learning_rate": 0.003, "loss": 4.0555, "step": 21446 }, { "epoch": 0.21447, "grad_norm": 1.1470295091155245, "learning_rate": 0.003, "loss": 4.0248, "step": 21447 }, { "epoch": 0.21448, "grad_norm": 1.3882002714959472, "learning_rate": 0.003, "loss": 4.062, "step": 21448 }, { "epoch": 0.21449, "grad_norm": 1.1807616504826879, "learning_rate": 0.003, "loss": 4.0557, "step": 21449 }, { "epoch": 0.2145, "grad_norm": 1.4204086149297122, "learning_rate": 0.003, "loss": 4.0422, "step": 21450 }, { "epoch": 0.21451, "grad_norm": 1.4210985255809614, "learning_rate": 0.003, "loss": 4.0586, "step": 21451 }, { "epoch": 0.21452, "grad_norm": 1.2066524038320814, "learning_rate": 0.003, "loss": 4.0425, "step": 21452 }, { "epoch": 0.21453, "grad_norm": 1.299796867171121, "learning_rate": 0.003, "loss": 4.0382, "step": 21453 }, { "epoch": 0.21454, "grad_norm": 1.2914211157329647, "learning_rate": 0.003, "loss": 4.0267, "step": 21454 }, { "epoch": 0.21455, "grad_norm": 1.1828566623496566, "learning_rate": 0.003, "loss": 4.0239, "step": 21455 }, { "epoch": 0.21456, "grad_norm": 1.485362907689217, "learning_rate": 0.003, "loss": 4.0698, "step": 21456 }, { "epoch": 0.21457, "grad_norm": 1.1259998176997275, "learning_rate": 0.003, "loss": 4.0423, "step": 21457 }, { "epoch": 0.21458, "grad_norm": 1.4734118462883794, "learning_rate": 0.003, "loss": 4.0364, "step": 21458 }, { "epoch": 0.21459, "grad_norm": 0.9283127651333246, "learning_rate": 0.003, "loss": 4.0578, "step": 21459 }, { "epoch": 0.2146, "grad_norm": 1.514546398235819, "learning_rate": 0.003, "loss": 4.0665, "step": 21460 }, { "epoch": 0.21461, "grad_norm": 1.2310494697427181, "learning_rate": 0.003, "loss": 4.0533, "step": 21461 }, { "epoch": 0.21462, "grad_norm": 1.520764891934006, "learning_rate": 0.003, "loss": 4.078, "step": 21462 }, { "epoch": 0.21463, "grad_norm": 1.1134509205383405, "learning_rate": 0.003, "loss": 4.0749, "step": 21463 }, { "epoch": 0.21464, "grad_norm": 1.2882346731976746, "learning_rate": 0.003, "loss": 4.0658, "step": 21464 }, { "epoch": 0.21465, "grad_norm": 1.244103986944709, "learning_rate": 0.003, "loss": 4.053, "step": 21465 }, { "epoch": 0.21466, "grad_norm": 1.1535959575311907, "learning_rate": 0.003, "loss": 4.063, "step": 21466 }, { "epoch": 0.21467, "grad_norm": 1.300012609803945, "learning_rate": 0.003, "loss": 4.0718, "step": 21467 }, { "epoch": 0.21468, "grad_norm": 1.2087191925302314, "learning_rate": 0.003, "loss": 4.0508, "step": 21468 }, { "epoch": 0.21469, "grad_norm": 1.2723417012323934, "learning_rate": 0.003, "loss": 4.0846, "step": 21469 }, { "epoch": 0.2147, "grad_norm": 1.1233641225791804, "learning_rate": 0.003, "loss": 4.0756, "step": 21470 }, { "epoch": 0.21471, "grad_norm": 1.569980926663672, "learning_rate": 0.003, "loss": 4.0476, "step": 21471 }, { "epoch": 0.21472, "grad_norm": 1.1506424120774437, "learning_rate": 0.003, "loss": 4.0684, "step": 21472 }, { "epoch": 0.21473, "grad_norm": 1.484760860744661, "learning_rate": 0.003, "loss": 4.0489, "step": 21473 }, { "epoch": 0.21474, "grad_norm": 1.4899737721885717, "learning_rate": 0.003, "loss": 4.0553, "step": 21474 }, { "epoch": 0.21475, "grad_norm": 1.489509118493849, "learning_rate": 0.003, "loss": 4.0508, "step": 21475 }, { "epoch": 0.21476, "grad_norm": 1.31366767297891, "learning_rate": 0.003, "loss": 4.0777, "step": 21476 }, { "epoch": 0.21477, "grad_norm": 1.2960671083487345, "learning_rate": 0.003, "loss": 4.0755, "step": 21477 }, { "epoch": 0.21478, "grad_norm": 1.2062205712519865, "learning_rate": 0.003, "loss": 4.0205, "step": 21478 }, { "epoch": 0.21479, "grad_norm": 1.370323358808596, "learning_rate": 0.003, "loss": 4.0323, "step": 21479 }, { "epoch": 0.2148, "grad_norm": 1.1825382882281281, "learning_rate": 0.003, "loss": 4.0567, "step": 21480 }, { "epoch": 0.21481, "grad_norm": 1.459331329537895, "learning_rate": 0.003, "loss": 4.0297, "step": 21481 }, { "epoch": 0.21482, "grad_norm": 1.2949174695844587, "learning_rate": 0.003, "loss": 4.0313, "step": 21482 }, { "epoch": 0.21483, "grad_norm": 1.375259940854289, "learning_rate": 0.003, "loss": 4.0538, "step": 21483 }, { "epoch": 0.21484, "grad_norm": 1.2303477301289663, "learning_rate": 0.003, "loss": 4.0605, "step": 21484 }, { "epoch": 0.21485, "grad_norm": 1.4353712203613151, "learning_rate": 0.003, "loss": 4.0521, "step": 21485 }, { "epoch": 0.21486, "grad_norm": 1.0818882035804576, "learning_rate": 0.003, "loss": 4.0745, "step": 21486 }, { "epoch": 0.21487, "grad_norm": 1.400946805712582, "learning_rate": 0.003, "loss": 4.0519, "step": 21487 }, { "epoch": 0.21488, "grad_norm": 1.2237243568913674, "learning_rate": 0.003, "loss": 4.0856, "step": 21488 }, { "epoch": 0.21489, "grad_norm": 1.5907722556953077, "learning_rate": 0.003, "loss": 4.0584, "step": 21489 }, { "epoch": 0.2149, "grad_norm": 1.1326397112931466, "learning_rate": 0.003, "loss": 4.0501, "step": 21490 }, { "epoch": 0.21491, "grad_norm": 1.5141931094356682, "learning_rate": 0.003, "loss": 4.0953, "step": 21491 }, { "epoch": 0.21492, "grad_norm": 1.2468582875027279, "learning_rate": 0.003, "loss": 4.05, "step": 21492 }, { "epoch": 0.21493, "grad_norm": 1.58904364529496, "learning_rate": 0.003, "loss": 4.0625, "step": 21493 }, { "epoch": 0.21494, "grad_norm": 0.9441584505932659, "learning_rate": 0.003, "loss": 4.0889, "step": 21494 }, { "epoch": 0.21495, "grad_norm": 1.2017863682929744, "learning_rate": 0.003, "loss": 4.0523, "step": 21495 }, { "epoch": 0.21496, "grad_norm": 1.4082893274140686, "learning_rate": 0.003, "loss": 4.0404, "step": 21496 }, { "epoch": 0.21497, "grad_norm": 1.1384979025041544, "learning_rate": 0.003, "loss": 4.0495, "step": 21497 }, { "epoch": 0.21498, "grad_norm": 1.4893106830259322, "learning_rate": 0.003, "loss": 4.0762, "step": 21498 }, { "epoch": 0.21499, "grad_norm": 1.3061178650236966, "learning_rate": 0.003, "loss": 4.0366, "step": 21499 }, { "epoch": 0.215, "grad_norm": 1.1975960147897236, "learning_rate": 0.003, "loss": 4.0323, "step": 21500 }, { "epoch": 0.21501, "grad_norm": 1.378108295364783, "learning_rate": 0.003, "loss": 4.0765, "step": 21501 }, { "epoch": 0.21502, "grad_norm": 1.2509148359913587, "learning_rate": 0.003, "loss": 4.0377, "step": 21502 }, { "epoch": 0.21503, "grad_norm": 1.1785855504452174, "learning_rate": 0.003, "loss": 4.0418, "step": 21503 }, { "epoch": 0.21504, "grad_norm": 1.2981264827377557, "learning_rate": 0.003, "loss": 4.0389, "step": 21504 }, { "epoch": 0.21505, "grad_norm": 1.1593302304472886, "learning_rate": 0.003, "loss": 4.0339, "step": 21505 }, { "epoch": 0.21506, "grad_norm": 1.4530196824357662, "learning_rate": 0.003, "loss": 4.0198, "step": 21506 }, { "epoch": 0.21507, "grad_norm": 1.131644225116241, "learning_rate": 0.003, "loss": 4.0447, "step": 21507 }, { "epoch": 0.21508, "grad_norm": 1.2982821610148942, "learning_rate": 0.003, "loss": 4.0492, "step": 21508 }, { "epoch": 0.21509, "grad_norm": 1.1850714169582515, "learning_rate": 0.003, "loss": 4.0609, "step": 21509 }, { "epoch": 0.2151, "grad_norm": 1.3086252014539759, "learning_rate": 0.003, "loss": 4.0395, "step": 21510 }, { "epoch": 0.21511, "grad_norm": 1.3489796934325005, "learning_rate": 0.003, "loss": 4.0684, "step": 21511 }, { "epoch": 0.21512, "grad_norm": 1.4437118489819842, "learning_rate": 0.003, "loss": 4.0082, "step": 21512 }, { "epoch": 0.21513, "grad_norm": 1.4002119949049092, "learning_rate": 0.003, "loss": 4.0697, "step": 21513 }, { "epoch": 0.21514, "grad_norm": 1.1188591397030128, "learning_rate": 0.003, "loss": 4.0218, "step": 21514 }, { "epoch": 0.21515, "grad_norm": 1.4327792047917798, "learning_rate": 0.003, "loss": 4.0701, "step": 21515 }, { "epoch": 0.21516, "grad_norm": 1.341617127907377, "learning_rate": 0.003, "loss": 4.0553, "step": 21516 }, { "epoch": 0.21517, "grad_norm": 1.164987996259884, "learning_rate": 0.003, "loss": 4.0442, "step": 21517 }, { "epoch": 0.21518, "grad_norm": 1.2281640200108996, "learning_rate": 0.003, "loss": 4.0354, "step": 21518 }, { "epoch": 0.21519, "grad_norm": 1.312899207326584, "learning_rate": 0.003, "loss": 4.0504, "step": 21519 }, { "epoch": 0.2152, "grad_norm": 1.206296180283065, "learning_rate": 0.003, "loss": 4.0349, "step": 21520 }, { "epoch": 0.21521, "grad_norm": 1.459840488704888, "learning_rate": 0.003, "loss": 4.0478, "step": 21521 }, { "epoch": 0.21522, "grad_norm": 0.9606316711795931, "learning_rate": 0.003, "loss": 4.0352, "step": 21522 }, { "epoch": 0.21523, "grad_norm": 1.547110083453148, "learning_rate": 0.003, "loss": 4.0431, "step": 21523 }, { "epoch": 0.21524, "grad_norm": 1.0169027031505786, "learning_rate": 0.003, "loss": 4.0643, "step": 21524 }, { "epoch": 0.21525, "grad_norm": 1.4369074392414622, "learning_rate": 0.003, "loss": 4.0606, "step": 21525 }, { "epoch": 0.21526, "grad_norm": 1.1627215327156861, "learning_rate": 0.003, "loss": 4.0458, "step": 21526 }, { "epoch": 0.21527, "grad_norm": 1.4412975392393965, "learning_rate": 0.003, "loss": 4.0651, "step": 21527 }, { "epoch": 0.21528, "grad_norm": 1.091798184460252, "learning_rate": 0.003, "loss": 4.0291, "step": 21528 }, { "epoch": 0.21529, "grad_norm": 1.4442076048064423, "learning_rate": 0.003, "loss": 4.0552, "step": 21529 }, { "epoch": 0.2153, "grad_norm": 1.326773243090699, "learning_rate": 0.003, "loss": 4.0635, "step": 21530 }, { "epoch": 0.21531, "grad_norm": 1.4728032312949157, "learning_rate": 0.003, "loss": 4.0246, "step": 21531 }, { "epoch": 0.21532, "grad_norm": 1.3554900969417256, "learning_rate": 0.003, "loss": 4.0612, "step": 21532 }, { "epoch": 0.21533, "grad_norm": 1.3109923420859935, "learning_rate": 0.003, "loss": 4.025, "step": 21533 }, { "epoch": 0.21534, "grad_norm": 1.3344710645017832, "learning_rate": 0.003, "loss": 4.0312, "step": 21534 }, { "epoch": 0.21535, "grad_norm": 1.4446752836141827, "learning_rate": 0.003, "loss": 4.0786, "step": 21535 }, { "epoch": 0.21536, "grad_norm": 1.2802692177491153, "learning_rate": 0.003, "loss": 4.0718, "step": 21536 }, { "epoch": 0.21537, "grad_norm": 1.240002120002768, "learning_rate": 0.003, "loss": 4.0363, "step": 21537 }, { "epoch": 0.21538, "grad_norm": 1.2479876865674657, "learning_rate": 0.003, "loss": 4.0536, "step": 21538 }, { "epoch": 0.21539, "grad_norm": 1.2777882246417078, "learning_rate": 0.003, "loss": 4.0577, "step": 21539 }, { "epoch": 0.2154, "grad_norm": 1.216095670864655, "learning_rate": 0.003, "loss": 4.0461, "step": 21540 }, { "epoch": 0.21541, "grad_norm": 1.2426701966452454, "learning_rate": 0.003, "loss": 4.0492, "step": 21541 }, { "epoch": 0.21542, "grad_norm": 1.6144032344125891, "learning_rate": 0.003, "loss": 4.0347, "step": 21542 }, { "epoch": 0.21543, "grad_norm": 1.2786230481168372, "learning_rate": 0.003, "loss": 4.0399, "step": 21543 }, { "epoch": 0.21544, "grad_norm": 1.3659857976232632, "learning_rate": 0.003, "loss": 4.0414, "step": 21544 }, { "epoch": 0.21545, "grad_norm": 1.3709623322979243, "learning_rate": 0.003, "loss": 4.0524, "step": 21545 }, { "epoch": 0.21546, "grad_norm": 1.1588598521743951, "learning_rate": 0.003, "loss": 4.035, "step": 21546 }, { "epoch": 0.21547, "grad_norm": 1.3220473388613287, "learning_rate": 0.003, "loss": 4.0487, "step": 21547 }, { "epoch": 0.21548, "grad_norm": 1.2082416254846837, "learning_rate": 0.003, "loss": 4.0541, "step": 21548 }, { "epoch": 0.21549, "grad_norm": 1.2889883211256423, "learning_rate": 0.003, "loss": 4.0506, "step": 21549 }, { "epoch": 0.2155, "grad_norm": 1.1773015288618838, "learning_rate": 0.003, "loss": 4.0558, "step": 21550 }, { "epoch": 0.21551, "grad_norm": 1.3413792697844495, "learning_rate": 0.003, "loss": 4.0374, "step": 21551 }, { "epoch": 0.21552, "grad_norm": 1.2138099463872778, "learning_rate": 0.003, "loss": 4.0308, "step": 21552 }, { "epoch": 0.21553, "grad_norm": 1.322098344705854, "learning_rate": 0.003, "loss": 4.0335, "step": 21553 }, { "epoch": 0.21554, "grad_norm": 1.1972945432618907, "learning_rate": 0.003, "loss": 4.0462, "step": 21554 }, { "epoch": 0.21555, "grad_norm": 1.3546799217413956, "learning_rate": 0.003, "loss": 4.0494, "step": 21555 }, { "epoch": 0.21556, "grad_norm": 1.4130727000380368, "learning_rate": 0.003, "loss": 4.0542, "step": 21556 }, { "epoch": 0.21557, "grad_norm": 1.1751027556247067, "learning_rate": 0.003, "loss": 4.0384, "step": 21557 }, { "epoch": 0.21558, "grad_norm": 1.4329701254594025, "learning_rate": 0.003, "loss": 4.0411, "step": 21558 }, { "epoch": 0.21559, "grad_norm": 1.1449424741576872, "learning_rate": 0.003, "loss": 4.0729, "step": 21559 }, { "epoch": 0.2156, "grad_norm": 1.4630223754035958, "learning_rate": 0.003, "loss": 4.0478, "step": 21560 }, { "epoch": 0.21561, "grad_norm": 1.2564142417719932, "learning_rate": 0.003, "loss": 4.0793, "step": 21561 }, { "epoch": 0.21562, "grad_norm": 1.2367252119445342, "learning_rate": 0.003, "loss": 4.0692, "step": 21562 }, { "epoch": 0.21563, "grad_norm": 1.2917252238179042, "learning_rate": 0.003, "loss": 4.0587, "step": 21563 }, { "epoch": 0.21564, "grad_norm": 1.3584215993866915, "learning_rate": 0.003, "loss": 4.0209, "step": 21564 }, { "epoch": 0.21565, "grad_norm": 1.1646880532013897, "learning_rate": 0.003, "loss": 4.0788, "step": 21565 }, { "epoch": 0.21566, "grad_norm": 1.4133766722188805, "learning_rate": 0.003, "loss": 4.0624, "step": 21566 }, { "epoch": 0.21567, "grad_norm": 1.382450781353812, "learning_rate": 0.003, "loss": 4.0467, "step": 21567 }, { "epoch": 0.21568, "grad_norm": 1.0461181073194628, "learning_rate": 0.003, "loss": 4.0588, "step": 21568 }, { "epoch": 0.21569, "grad_norm": 1.4468496782759792, "learning_rate": 0.003, "loss": 4.0455, "step": 21569 }, { "epoch": 0.2157, "grad_norm": 1.0693770762476509, "learning_rate": 0.003, "loss": 4.041, "step": 21570 }, { "epoch": 0.21571, "grad_norm": 1.3436574081692902, "learning_rate": 0.003, "loss": 4.0371, "step": 21571 }, { "epoch": 0.21572, "grad_norm": 1.1077115140885394, "learning_rate": 0.003, "loss": 4.0652, "step": 21572 }, { "epoch": 0.21573, "grad_norm": 1.474968349036726, "learning_rate": 0.003, "loss": 4.0428, "step": 21573 }, { "epoch": 0.21574, "grad_norm": 1.2243913229395103, "learning_rate": 0.003, "loss": 4.0448, "step": 21574 }, { "epoch": 0.21575, "grad_norm": 1.4357631117853142, "learning_rate": 0.003, "loss": 4.0616, "step": 21575 }, { "epoch": 0.21576, "grad_norm": 1.143636442416954, "learning_rate": 0.003, "loss": 4.0429, "step": 21576 }, { "epoch": 0.21577, "grad_norm": 1.3307684153775725, "learning_rate": 0.003, "loss": 4.0644, "step": 21577 }, { "epoch": 0.21578, "grad_norm": 1.4348354005676, "learning_rate": 0.003, "loss": 4.052, "step": 21578 }, { "epoch": 0.21579, "grad_norm": 1.1869246906858275, "learning_rate": 0.003, "loss": 4.0597, "step": 21579 }, { "epoch": 0.2158, "grad_norm": 1.3228739877660194, "learning_rate": 0.003, "loss": 4.0202, "step": 21580 }, { "epoch": 0.21581, "grad_norm": 1.259811713759345, "learning_rate": 0.003, "loss": 4.0468, "step": 21581 }, { "epoch": 0.21582, "grad_norm": 1.2068431794784156, "learning_rate": 0.003, "loss": 4.0425, "step": 21582 }, { "epoch": 0.21583, "grad_norm": 1.1254360834494639, "learning_rate": 0.003, "loss": 4.0438, "step": 21583 }, { "epoch": 0.21584, "grad_norm": 1.4300677552421914, "learning_rate": 0.003, "loss": 4.0566, "step": 21584 }, { "epoch": 0.21585, "grad_norm": 1.2716059702652318, "learning_rate": 0.003, "loss": 4.0668, "step": 21585 }, { "epoch": 0.21586, "grad_norm": 1.3081695594695006, "learning_rate": 0.003, "loss": 4.0286, "step": 21586 }, { "epoch": 0.21587, "grad_norm": 1.2917306653430283, "learning_rate": 0.003, "loss": 4.0881, "step": 21587 }, { "epoch": 0.21588, "grad_norm": 1.2968082982162343, "learning_rate": 0.003, "loss": 4.0437, "step": 21588 }, { "epoch": 0.21589, "grad_norm": 1.2790088314939978, "learning_rate": 0.003, "loss": 4.0522, "step": 21589 }, { "epoch": 0.2159, "grad_norm": 1.2423031177996033, "learning_rate": 0.003, "loss": 4.0393, "step": 21590 }, { "epoch": 0.21591, "grad_norm": 1.3732270439759997, "learning_rate": 0.003, "loss": 4.0399, "step": 21591 }, { "epoch": 0.21592, "grad_norm": 1.3319770501913595, "learning_rate": 0.003, "loss": 4.0492, "step": 21592 }, { "epoch": 0.21593, "grad_norm": 1.3052330638964627, "learning_rate": 0.003, "loss": 4.0714, "step": 21593 }, { "epoch": 0.21594, "grad_norm": 1.3956728189002865, "learning_rate": 0.003, "loss": 4.0232, "step": 21594 }, { "epoch": 0.21595, "grad_norm": 1.4167772078378211, "learning_rate": 0.003, "loss": 4.0518, "step": 21595 }, { "epoch": 0.21596, "grad_norm": 1.307222877586128, "learning_rate": 0.003, "loss": 4.0357, "step": 21596 }, { "epoch": 0.21597, "grad_norm": 1.2620104264140628, "learning_rate": 0.003, "loss": 4.0252, "step": 21597 }, { "epoch": 0.21598, "grad_norm": 1.4415953331243176, "learning_rate": 0.003, "loss": 4.0721, "step": 21598 }, { "epoch": 0.21599, "grad_norm": 1.1689909013279838, "learning_rate": 0.003, "loss": 4.0416, "step": 21599 }, { "epoch": 0.216, "grad_norm": 1.1752426271453542, "learning_rate": 0.003, "loss": 4.0653, "step": 21600 }, { "epoch": 0.21601, "grad_norm": 1.4373277102072644, "learning_rate": 0.003, "loss": 4.0409, "step": 21601 }, { "epoch": 0.21602, "grad_norm": 1.139326460570499, "learning_rate": 0.003, "loss": 3.9979, "step": 21602 }, { "epoch": 0.21603, "grad_norm": 1.702342005023596, "learning_rate": 0.003, "loss": 4.0546, "step": 21603 }, { "epoch": 0.21604, "grad_norm": 0.9181373709243951, "learning_rate": 0.003, "loss": 4.0293, "step": 21604 }, { "epoch": 0.21605, "grad_norm": 1.4156053304286162, "learning_rate": 0.003, "loss": 4.0379, "step": 21605 }, { "epoch": 0.21606, "grad_norm": 1.299322167838569, "learning_rate": 0.003, "loss": 4.0579, "step": 21606 }, { "epoch": 0.21607, "grad_norm": 1.4494695676496232, "learning_rate": 0.003, "loss": 4.0589, "step": 21607 }, { "epoch": 0.21608, "grad_norm": 1.187093459511788, "learning_rate": 0.003, "loss": 4.0523, "step": 21608 }, { "epoch": 0.21609, "grad_norm": 1.307214047059684, "learning_rate": 0.003, "loss": 4.0451, "step": 21609 }, { "epoch": 0.2161, "grad_norm": 1.305218544830906, "learning_rate": 0.003, "loss": 4.0377, "step": 21610 }, { "epoch": 0.21611, "grad_norm": 1.1245976597135463, "learning_rate": 0.003, "loss": 4.0512, "step": 21611 }, { "epoch": 0.21612, "grad_norm": 1.4216159000344961, "learning_rate": 0.003, "loss": 4.0405, "step": 21612 }, { "epoch": 0.21613, "grad_norm": 1.1299949348152787, "learning_rate": 0.003, "loss": 4.0291, "step": 21613 }, { "epoch": 0.21614, "grad_norm": 1.4791347438098315, "learning_rate": 0.003, "loss": 4.0673, "step": 21614 }, { "epoch": 0.21615, "grad_norm": 1.0638591599537033, "learning_rate": 0.003, "loss": 4.052, "step": 21615 }, { "epoch": 0.21616, "grad_norm": 1.3475699073183858, "learning_rate": 0.003, "loss": 4.0485, "step": 21616 }, { "epoch": 0.21617, "grad_norm": 1.2316611239869664, "learning_rate": 0.003, "loss": 4.0507, "step": 21617 }, { "epoch": 0.21618, "grad_norm": 1.5065327402318887, "learning_rate": 0.003, "loss": 4.0776, "step": 21618 }, { "epoch": 0.21619, "grad_norm": 1.4096142512965315, "learning_rate": 0.003, "loss": 4.0668, "step": 21619 }, { "epoch": 0.2162, "grad_norm": 1.159454546769651, "learning_rate": 0.003, "loss": 4.062, "step": 21620 }, { "epoch": 0.21621, "grad_norm": 1.317540642950509, "learning_rate": 0.003, "loss": 4.0495, "step": 21621 }, { "epoch": 0.21622, "grad_norm": 1.1865350298335586, "learning_rate": 0.003, "loss": 4.0811, "step": 21622 }, { "epoch": 0.21623, "grad_norm": 1.3959728152423938, "learning_rate": 0.003, "loss": 4.0728, "step": 21623 }, { "epoch": 0.21624, "grad_norm": 1.0967286789828294, "learning_rate": 0.003, "loss": 4.0557, "step": 21624 }, { "epoch": 0.21625, "grad_norm": 1.4107174860021268, "learning_rate": 0.003, "loss": 4.0436, "step": 21625 }, { "epoch": 0.21626, "grad_norm": 1.43536578196684, "learning_rate": 0.003, "loss": 4.0393, "step": 21626 }, { "epoch": 0.21627, "grad_norm": 1.3778884389546275, "learning_rate": 0.003, "loss": 4.0614, "step": 21627 }, { "epoch": 0.21628, "grad_norm": 1.145262126204291, "learning_rate": 0.003, "loss": 3.9985, "step": 21628 }, { "epoch": 0.21629, "grad_norm": 1.329957496149722, "learning_rate": 0.003, "loss": 4.0484, "step": 21629 }, { "epoch": 0.2163, "grad_norm": 1.2073280019111943, "learning_rate": 0.003, "loss": 4.0363, "step": 21630 }, { "epoch": 0.21631, "grad_norm": 1.4143764551609217, "learning_rate": 0.003, "loss": 4.051, "step": 21631 }, { "epoch": 0.21632, "grad_norm": 1.2345148189041288, "learning_rate": 0.003, "loss": 4.0453, "step": 21632 }, { "epoch": 0.21633, "grad_norm": 1.2452968459023985, "learning_rate": 0.003, "loss": 4.0674, "step": 21633 }, { "epoch": 0.21634, "grad_norm": 1.4200986048996227, "learning_rate": 0.003, "loss": 4.0648, "step": 21634 }, { "epoch": 0.21635, "grad_norm": 1.3058008842472895, "learning_rate": 0.003, "loss": 4.0717, "step": 21635 }, { "epoch": 0.21636, "grad_norm": 1.1026251764083932, "learning_rate": 0.003, "loss": 4.0557, "step": 21636 }, { "epoch": 0.21637, "grad_norm": 1.3835744114032524, "learning_rate": 0.003, "loss": 4.058, "step": 21637 }, { "epoch": 0.21638, "grad_norm": 1.1163499998284347, "learning_rate": 0.003, "loss": 4.0416, "step": 21638 }, { "epoch": 0.21639, "grad_norm": 1.7718147706196286, "learning_rate": 0.003, "loss": 4.0494, "step": 21639 }, { "epoch": 0.2164, "grad_norm": 0.9960949690844515, "learning_rate": 0.003, "loss": 4.0197, "step": 21640 }, { "epoch": 0.21641, "grad_norm": 1.598915141672048, "learning_rate": 0.003, "loss": 4.0763, "step": 21641 }, { "epoch": 0.21642, "grad_norm": 1.3501190162195924, "learning_rate": 0.003, "loss": 4.0493, "step": 21642 }, { "epoch": 0.21643, "grad_norm": 1.11708654672925, "learning_rate": 0.003, "loss": 4.0542, "step": 21643 }, { "epoch": 0.21644, "grad_norm": 1.1837617744724265, "learning_rate": 0.003, "loss": 4.0464, "step": 21644 }, { "epoch": 0.21645, "grad_norm": 1.3930522195397887, "learning_rate": 0.003, "loss": 4.0786, "step": 21645 }, { "epoch": 0.21646, "grad_norm": 1.1971351458603388, "learning_rate": 0.003, "loss": 4.0533, "step": 21646 }, { "epoch": 0.21647, "grad_norm": 1.195709723980143, "learning_rate": 0.003, "loss": 4.037, "step": 21647 }, { "epoch": 0.21648, "grad_norm": 1.2392818652695932, "learning_rate": 0.003, "loss": 4.0487, "step": 21648 }, { "epoch": 0.21649, "grad_norm": 1.2836118144686024, "learning_rate": 0.003, "loss": 4.0477, "step": 21649 }, { "epoch": 0.2165, "grad_norm": 1.3915139268510568, "learning_rate": 0.003, "loss": 4.0657, "step": 21650 }, { "epoch": 0.21651, "grad_norm": 1.1292513375476736, "learning_rate": 0.003, "loss": 4.0538, "step": 21651 }, { "epoch": 0.21652, "grad_norm": 1.6038647277746723, "learning_rate": 0.003, "loss": 4.0157, "step": 21652 }, { "epoch": 0.21653, "grad_norm": 0.979494980453447, "learning_rate": 0.003, "loss": 4.06, "step": 21653 }, { "epoch": 0.21654, "grad_norm": 1.5388938433260417, "learning_rate": 0.003, "loss": 4.0275, "step": 21654 }, { "epoch": 0.21655, "grad_norm": 1.3595512608491056, "learning_rate": 0.003, "loss": 4.0683, "step": 21655 }, { "epoch": 0.21656, "grad_norm": 1.526590089855836, "learning_rate": 0.003, "loss": 4.0544, "step": 21656 }, { "epoch": 0.21657, "grad_norm": 1.0464180350081598, "learning_rate": 0.003, "loss": 4.0303, "step": 21657 }, { "epoch": 0.21658, "grad_norm": 1.4781300459318107, "learning_rate": 0.003, "loss": 4.0674, "step": 21658 }, { "epoch": 0.21659, "grad_norm": 1.0687999139567819, "learning_rate": 0.003, "loss": 4.0393, "step": 21659 }, { "epoch": 0.2166, "grad_norm": 1.5910343276132453, "learning_rate": 0.003, "loss": 4.0542, "step": 21660 }, { "epoch": 0.21661, "grad_norm": 1.054474179530512, "learning_rate": 0.003, "loss": 4.0716, "step": 21661 }, { "epoch": 0.21662, "grad_norm": 1.389793991058751, "learning_rate": 0.003, "loss": 4.0579, "step": 21662 }, { "epoch": 0.21663, "grad_norm": 1.098142249375085, "learning_rate": 0.003, "loss": 4.0613, "step": 21663 }, { "epoch": 0.21664, "grad_norm": 1.4766492013725177, "learning_rate": 0.003, "loss": 4.0609, "step": 21664 }, { "epoch": 0.21665, "grad_norm": 1.1644620344863594, "learning_rate": 0.003, "loss": 4.0732, "step": 21665 }, { "epoch": 0.21666, "grad_norm": 1.3802003071818196, "learning_rate": 0.003, "loss": 4.037, "step": 21666 }, { "epoch": 0.21667, "grad_norm": 1.3349039152220499, "learning_rate": 0.003, "loss": 4.052, "step": 21667 }, { "epoch": 0.21668, "grad_norm": 1.2044820082090815, "learning_rate": 0.003, "loss": 4.0358, "step": 21668 }, { "epoch": 0.21669, "grad_norm": 1.3231136323465995, "learning_rate": 0.003, "loss": 4.0254, "step": 21669 }, { "epoch": 0.2167, "grad_norm": 1.2623644530106173, "learning_rate": 0.003, "loss": 4.0559, "step": 21670 }, { "epoch": 0.21671, "grad_norm": 1.4624627172783418, "learning_rate": 0.003, "loss": 4.0726, "step": 21671 }, { "epoch": 0.21672, "grad_norm": 1.149299095223078, "learning_rate": 0.003, "loss": 4.0105, "step": 21672 }, { "epoch": 0.21673, "grad_norm": 1.2030043550332916, "learning_rate": 0.003, "loss": 4.0352, "step": 21673 }, { "epoch": 0.21674, "grad_norm": 1.2909773238033704, "learning_rate": 0.003, "loss": 4.0099, "step": 21674 }, { "epoch": 0.21675, "grad_norm": 1.2679498641166909, "learning_rate": 0.003, "loss": 4.0202, "step": 21675 }, { "epoch": 0.21676, "grad_norm": 1.2613544205889584, "learning_rate": 0.003, "loss": 4.0548, "step": 21676 }, { "epoch": 0.21677, "grad_norm": 1.0937616178822396, "learning_rate": 0.003, "loss": 4.0535, "step": 21677 }, { "epoch": 0.21678, "grad_norm": 1.4788878273843822, "learning_rate": 0.003, "loss": 4.0434, "step": 21678 }, { "epoch": 0.21679, "grad_norm": 1.048627147398995, "learning_rate": 0.003, "loss": 4.0639, "step": 21679 }, { "epoch": 0.2168, "grad_norm": 1.437251322030383, "learning_rate": 0.003, "loss": 4.0429, "step": 21680 }, { "epoch": 0.21681, "grad_norm": 1.1656455097893734, "learning_rate": 0.003, "loss": 4.0403, "step": 21681 }, { "epoch": 0.21682, "grad_norm": 1.2233444905006519, "learning_rate": 0.003, "loss": 4.0429, "step": 21682 }, { "epoch": 0.21683, "grad_norm": 1.1120374871767775, "learning_rate": 0.003, "loss": 4.0726, "step": 21683 }, { "epoch": 0.21684, "grad_norm": 1.5661729770110275, "learning_rate": 0.003, "loss": 4.0517, "step": 21684 }, { "epoch": 0.21685, "grad_norm": 1.4661846422040847, "learning_rate": 0.003, "loss": 4.0541, "step": 21685 }, { "epoch": 0.21686, "grad_norm": 1.4717736535512747, "learning_rate": 0.003, "loss": 4.0412, "step": 21686 }, { "epoch": 0.21687, "grad_norm": 1.1563436294708824, "learning_rate": 0.003, "loss": 4.0194, "step": 21687 }, { "epoch": 0.21688, "grad_norm": 1.118294700225654, "learning_rate": 0.003, "loss": 4.0591, "step": 21688 }, { "epoch": 0.21689, "grad_norm": 1.5012816566404177, "learning_rate": 0.003, "loss": 4.0378, "step": 21689 }, { "epoch": 0.2169, "grad_norm": 1.3687644289169574, "learning_rate": 0.003, "loss": 4.0471, "step": 21690 }, { "epoch": 0.21691, "grad_norm": 1.1850811915893817, "learning_rate": 0.003, "loss": 4.019, "step": 21691 }, { "epoch": 0.21692, "grad_norm": 1.2438480460026475, "learning_rate": 0.003, "loss": 4.0539, "step": 21692 }, { "epoch": 0.21693, "grad_norm": 1.2771512200175685, "learning_rate": 0.003, "loss": 4.0377, "step": 21693 }, { "epoch": 0.21694, "grad_norm": 1.2909528453072165, "learning_rate": 0.003, "loss": 4.028, "step": 21694 }, { "epoch": 0.21695, "grad_norm": 1.1863761593004463, "learning_rate": 0.003, "loss": 4.0556, "step": 21695 }, { "epoch": 0.21696, "grad_norm": 1.1697625793188724, "learning_rate": 0.003, "loss": 4.0484, "step": 21696 }, { "epoch": 0.21697, "grad_norm": 1.3451951255757777, "learning_rate": 0.003, "loss": 4.0493, "step": 21697 }, { "epoch": 0.21698, "grad_norm": 1.2960474918538587, "learning_rate": 0.003, "loss": 3.9967, "step": 21698 }, { "epoch": 0.21699, "grad_norm": 1.2345397804210092, "learning_rate": 0.003, "loss": 4.0504, "step": 21699 }, { "epoch": 0.217, "grad_norm": 1.3326197741245571, "learning_rate": 0.003, "loss": 4.0437, "step": 21700 }, { "epoch": 0.21701, "grad_norm": 1.3130926308827144, "learning_rate": 0.003, "loss": 4.064, "step": 21701 }, { "epoch": 0.21702, "grad_norm": 1.2045128299819818, "learning_rate": 0.003, "loss": 4.0512, "step": 21702 }, { "epoch": 0.21703, "grad_norm": 1.2567238761158683, "learning_rate": 0.003, "loss": 4.0322, "step": 21703 }, { "epoch": 0.21704, "grad_norm": 1.1901544425293324, "learning_rate": 0.003, "loss": 4.0442, "step": 21704 }, { "epoch": 0.21705, "grad_norm": 1.2804168037424124, "learning_rate": 0.003, "loss": 4.027, "step": 21705 }, { "epoch": 0.21706, "grad_norm": 1.3042417045037906, "learning_rate": 0.003, "loss": 4.0333, "step": 21706 }, { "epoch": 0.21707, "grad_norm": 1.462729231419984, "learning_rate": 0.003, "loss": 4.0709, "step": 21707 }, { "epoch": 0.21708, "grad_norm": 1.0520057436004748, "learning_rate": 0.003, "loss": 4.0358, "step": 21708 }, { "epoch": 0.21709, "grad_norm": 1.465433317820797, "learning_rate": 0.003, "loss": 4.035, "step": 21709 }, { "epoch": 0.2171, "grad_norm": 1.142849578505775, "learning_rate": 0.003, "loss": 4.0512, "step": 21710 }, { "epoch": 0.21711, "grad_norm": 1.872884346632127, "learning_rate": 0.003, "loss": 4.0533, "step": 21711 }, { "epoch": 0.21712, "grad_norm": 1.260572130773345, "learning_rate": 0.003, "loss": 4.0421, "step": 21712 }, { "epoch": 0.21713, "grad_norm": 1.1845132205454723, "learning_rate": 0.003, "loss": 4.0708, "step": 21713 }, { "epoch": 0.21714, "grad_norm": 1.5152055655833143, "learning_rate": 0.003, "loss": 4.0295, "step": 21714 }, { "epoch": 0.21715, "grad_norm": 1.1536665825124806, "learning_rate": 0.003, "loss": 4.0461, "step": 21715 }, { "epoch": 0.21716, "grad_norm": 1.2851235738906504, "learning_rate": 0.003, "loss": 4.0451, "step": 21716 }, { "epoch": 0.21717, "grad_norm": 1.6653586909130602, "learning_rate": 0.003, "loss": 4.0485, "step": 21717 }, { "epoch": 0.21718, "grad_norm": 1.1457946051654446, "learning_rate": 0.003, "loss": 4.0359, "step": 21718 }, { "epoch": 0.21719, "grad_norm": 1.4192257902255103, "learning_rate": 0.003, "loss": 4.0428, "step": 21719 }, { "epoch": 0.2172, "grad_norm": 1.2438092755574954, "learning_rate": 0.003, "loss": 4.0499, "step": 21720 }, { "epoch": 0.21721, "grad_norm": 1.32219183146982, "learning_rate": 0.003, "loss": 4.0672, "step": 21721 }, { "epoch": 0.21722, "grad_norm": 1.284226692704832, "learning_rate": 0.003, "loss": 4.0202, "step": 21722 }, { "epoch": 0.21723, "grad_norm": 1.4497404618246834, "learning_rate": 0.003, "loss": 4.0476, "step": 21723 }, { "epoch": 0.21724, "grad_norm": 1.2108837005354196, "learning_rate": 0.003, "loss": 4.0595, "step": 21724 }, { "epoch": 0.21725, "grad_norm": 1.2882882117901246, "learning_rate": 0.003, "loss": 4.0426, "step": 21725 }, { "epoch": 0.21726, "grad_norm": 1.1053404936096287, "learning_rate": 0.003, "loss": 4.0311, "step": 21726 }, { "epoch": 0.21727, "grad_norm": 1.5178973976789325, "learning_rate": 0.003, "loss": 4.0614, "step": 21727 }, { "epoch": 0.21728, "grad_norm": 1.1428639071682594, "learning_rate": 0.003, "loss": 4.0576, "step": 21728 }, { "epoch": 0.21729, "grad_norm": 1.3564810985494395, "learning_rate": 0.003, "loss": 4.0453, "step": 21729 }, { "epoch": 0.2173, "grad_norm": 1.1904550337860347, "learning_rate": 0.003, "loss": 4.0296, "step": 21730 }, { "epoch": 0.21731, "grad_norm": 1.4848382747182223, "learning_rate": 0.003, "loss": 4.0931, "step": 21731 }, { "epoch": 0.21732, "grad_norm": 1.1988120980535053, "learning_rate": 0.003, "loss": 4.0593, "step": 21732 }, { "epoch": 0.21733, "grad_norm": 1.4498829318594544, "learning_rate": 0.003, "loss": 4.0738, "step": 21733 }, { "epoch": 0.21734, "grad_norm": 1.1800732265472247, "learning_rate": 0.003, "loss": 4.0581, "step": 21734 }, { "epoch": 0.21735, "grad_norm": 1.2274318114915934, "learning_rate": 0.003, "loss": 4.0417, "step": 21735 }, { "epoch": 0.21736, "grad_norm": 1.3328993248200016, "learning_rate": 0.003, "loss": 4.0315, "step": 21736 }, { "epoch": 0.21737, "grad_norm": 1.1868578841688844, "learning_rate": 0.003, "loss": 4.0249, "step": 21737 }, { "epoch": 0.21738, "grad_norm": 1.1156789035775732, "learning_rate": 0.003, "loss": 4.0616, "step": 21738 }, { "epoch": 0.21739, "grad_norm": 1.5106478469827882, "learning_rate": 0.003, "loss": 4.0532, "step": 21739 }, { "epoch": 0.2174, "grad_norm": 1.1993640107254842, "learning_rate": 0.003, "loss": 4.0117, "step": 21740 }, { "epoch": 0.21741, "grad_norm": 1.6821834633986894, "learning_rate": 0.003, "loss": 4.0636, "step": 21741 }, { "epoch": 0.21742, "grad_norm": 1.1557534760532018, "learning_rate": 0.003, "loss": 4.067, "step": 21742 }, { "epoch": 0.21743, "grad_norm": 1.3357601424742491, "learning_rate": 0.003, "loss": 4.0805, "step": 21743 }, { "epoch": 0.21744, "grad_norm": 1.263381980222935, "learning_rate": 0.003, "loss": 4.0617, "step": 21744 }, { "epoch": 0.21745, "grad_norm": 1.491219457287799, "learning_rate": 0.003, "loss": 4.0536, "step": 21745 }, { "epoch": 0.21746, "grad_norm": 0.9872060673395434, "learning_rate": 0.003, "loss": 4.0349, "step": 21746 }, { "epoch": 0.21747, "grad_norm": 1.1932658773275, "learning_rate": 0.003, "loss": 4.0683, "step": 21747 }, { "epoch": 0.21748, "grad_norm": 1.333675384596158, "learning_rate": 0.003, "loss": 4.0432, "step": 21748 }, { "epoch": 0.21749, "grad_norm": 1.470193364690084, "learning_rate": 0.003, "loss": 4.0262, "step": 21749 }, { "epoch": 0.2175, "grad_norm": 1.0614461446475574, "learning_rate": 0.003, "loss": 4.0405, "step": 21750 }, { "epoch": 0.21751, "grad_norm": 1.4165992659360809, "learning_rate": 0.003, "loss": 4.0189, "step": 21751 }, { "epoch": 0.21752, "grad_norm": 1.1924084246654956, "learning_rate": 0.003, "loss": 4.0623, "step": 21752 }, { "epoch": 0.21753, "grad_norm": 1.313097715920257, "learning_rate": 0.003, "loss": 4.0554, "step": 21753 }, { "epoch": 0.21754, "grad_norm": 1.270571883977382, "learning_rate": 0.003, "loss": 4.0334, "step": 21754 }, { "epoch": 0.21755, "grad_norm": 1.272599828364782, "learning_rate": 0.003, "loss": 4.0673, "step": 21755 }, { "epoch": 0.21756, "grad_norm": 1.4556803151817337, "learning_rate": 0.003, "loss": 4.0167, "step": 21756 }, { "epoch": 0.21757, "grad_norm": 1.4361716690834565, "learning_rate": 0.003, "loss": 4.0549, "step": 21757 }, { "epoch": 0.21758, "grad_norm": 1.3398933909134998, "learning_rate": 0.003, "loss": 4.0602, "step": 21758 }, { "epoch": 0.21759, "grad_norm": 1.4170590315386171, "learning_rate": 0.003, "loss": 4.0605, "step": 21759 }, { "epoch": 0.2176, "grad_norm": 1.2193683819262564, "learning_rate": 0.003, "loss": 4.0243, "step": 21760 }, { "epoch": 0.21761, "grad_norm": 1.28055552804489, "learning_rate": 0.003, "loss": 4.0166, "step": 21761 }, { "epoch": 0.21762, "grad_norm": 1.2245662664859995, "learning_rate": 0.003, "loss": 4.0336, "step": 21762 }, { "epoch": 0.21763, "grad_norm": 1.4084482449929674, "learning_rate": 0.003, "loss": 4.059, "step": 21763 }, { "epoch": 0.21764, "grad_norm": 1.1585647833835193, "learning_rate": 0.003, "loss": 4.0532, "step": 21764 }, { "epoch": 0.21765, "grad_norm": 1.2115678362913604, "learning_rate": 0.003, "loss": 4.0603, "step": 21765 }, { "epoch": 0.21766, "grad_norm": 1.157946350652913, "learning_rate": 0.003, "loss": 4.0401, "step": 21766 }, { "epoch": 0.21767, "grad_norm": 1.499516708908824, "learning_rate": 0.003, "loss": 4.0606, "step": 21767 }, { "epoch": 0.21768, "grad_norm": 1.452932108659569, "learning_rate": 0.003, "loss": 4.0669, "step": 21768 }, { "epoch": 0.21769, "grad_norm": 1.1759988580555416, "learning_rate": 0.003, "loss": 4.0528, "step": 21769 }, { "epoch": 0.2177, "grad_norm": 1.4267740862588505, "learning_rate": 0.003, "loss": 4.0558, "step": 21770 }, { "epoch": 0.21771, "grad_norm": 1.1640836432646118, "learning_rate": 0.003, "loss": 4.0444, "step": 21771 }, { "epoch": 0.21772, "grad_norm": 1.3022378991827772, "learning_rate": 0.003, "loss": 4.0367, "step": 21772 }, { "epoch": 0.21773, "grad_norm": 1.3072925323817297, "learning_rate": 0.003, "loss": 4.0698, "step": 21773 }, { "epoch": 0.21774, "grad_norm": 1.2105045238472119, "learning_rate": 0.003, "loss": 4.038, "step": 21774 }, { "epoch": 0.21775, "grad_norm": 1.5017155364835821, "learning_rate": 0.003, "loss": 4.0616, "step": 21775 }, { "epoch": 0.21776, "grad_norm": 1.1197846090932555, "learning_rate": 0.003, "loss": 4.0238, "step": 21776 }, { "epoch": 0.21777, "grad_norm": 1.9095875506733158, "learning_rate": 0.003, "loss": 4.0625, "step": 21777 }, { "epoch": 0.21778, "grad_norm": 0.9170804396826207, "learning_rate": 0.003, "loss": 4.016, "step": 21778 }, { "epoch": 0.21779, "grad_norm": 1.3377267722755934, "learning_rate": 0.003, "loss": 4.0373, "step": 21779 }, { "epoch": 0.2178, "grad_norm": 1.243052792917148, "learning_rate": 0.003, "loss": 4.0655, "step": 21780 }, { "epoch": 0.21781, "grad_norm": 1.382652263301088, "learning_rate": 0.003, "loss": 4.0686, "step": 21781 }, { "epoch": 0.21782, "grad_norm": 1.1051630228725697, "learning_rate": 0.003, "loss": 4.0243, "step": 21782 }, { "epoch": 0.21783, "grad_norm": 1.1404321590529605, "learning_rate": 0.003, "loss": 4.022, "step": 21783 }, { "epoch": 0.21784, "grad_norm": 1.301785248799035, "learning_rate": 0.003, "loss": 4.0475, "step": 21784 }, { "epoch": 0.21785, "grad_norm": 1.0912646971413142, "learning_rate": 0.003, "loss": 4.0485, "step": 21785 }, { "epoch": 0.21786, "grad_norm": 1.4845960158302227, "learning_rate": 0.003, "loss": 4.0551, "step": 21786 }, { "epoch": 0.21787, "grad_norm": 1.4692435735597091, "learning_rate": 0.003, "loss": 4.0536, "step": 21787 }, { "epoch": 0.21788, "grad_norm": 1.371294947139874, "learning_rate": 0.003, "loss": 4.0709, "step": 21788 }, { "epoch": 0.21789, "grad_norm": 1.340268924973976, "learning_rate": 0.003, "loss": 4.0546, "step": 21789 }, { "epoch": 0.2179, "grad_norm": 1.3030649934143335, "learning_rate": 0.003, "loss": 4.0334, "step": 21790 }, { "epoch": 0.21791, "grad_norm": 1.1655853202551025, "learning_rate": 0.003, "loss": 4.0499, "step": 21791 }, { "epoch": 0.21792, "grad_norm": 1.417186448806512, "learning_rate": 0.003, "loss": 4.0294, "step": 21792 }, { "epoch": 0.21793, "grad_norm": 1.200047988513548, "learning_rate": 0.003, "loss": 4.0629, "step": 21793 }, { "epoch": 0.21794, "grad_norm": 1.3164823798491363, "learning_rate": 0.003, "loss": 4.0645, "step": 21794 }, { "epoch": 0.21795, "grad_norm": 1.3345678999834834, "learning_rate": 0.003, "loss": 4.0549, "step": 21795 }, { "epoch": 0.21796, "grad_norm": 1.2137007865353142, "learning_rate": 0.003, "loss": 3.9997, "step": 21796 }, { "epoch": 0.21797, "grad_norm": 1.0818115182812484, "learning_rate": 0.003, "loss": 4.0461, "step": 21797 }, { "epoch": 0.21798, "grad_norm": 1.3995812927115454, "learning_rate": 0.003, "loss": 4.0481, "step": 21798 }, { "epoch": 0.21799, "grad_norm": 1.0887305321876026, "learning_rate": 0.003, "loss": 4.0328, "step": 21799 }, { "epoch": 0.218, "grad_norm": 1.696440647290099, "learning_rate": 0.003, "loss": 4.0456, "step": 21800 }, { "epoch": 0.21801, "grad_norm": 1.2730282446637662, "learning_rate": 0.003, "loss": 4.0498, "step": 21801 }, { "epoch": 0.21802, "grad_norm": 1.45578713110437, "learning_rate": 0.003, "loss": 4.0545, "step": 21802 }, { "epoch": 0.21803, "grad_norm": 1.446971481281067, "learning_rate": 0.003, "loss": 4.0748, "step": 21803 }, { "epoch": 0.21804, "grad_norm": 0.965512232012288, "learning_rate": 0.003, "loss": 4.0584, "step": 21804 }, { "epoch": 0.21805, "grad_norm": 1.628209411214888, "learning_rate": 0.003, "loss": 4.0627, "step": 21805 }, { "epoch": 0.21806, "grad_norm": 1.1474627984919128, "learning_rate": 0.003, "loss": 4.0652, "step": 21806 }, { "epoch": 0.21807, "grad_norm": 1.196026867362031, "learning_rate": 0.003, "loss": 4.0485, "step": 21807 }, { "epoch": 0.21808, "grad_norm": 1.2443965979525626, "learning_rate": 0.003, "loss": 4.0419, "step": 21808 }, { "epoch": 0.21809, "grad_norm": 1.5809297041238688, "learning_rate": 0.003, "loss": 4.0835, "step": 21809 }, { "epoch": 0.2181, "grad_norm": 1.0872077699541336, "learning_rate": 0.003, "loss": 4.0345, "step": 21810 }, { "epoch": 0.21811, "grad_norm": 1.6236580541857166, "learning_rate": 0.003, "loss": 4.0638, "step": 21811 }, { "epoch": 0.21812, "grad_norm": 1.4540955870785828, "learning_rate": 0.003, "loss": 4.0332, "step": 21812 }, { "epoch": 0.21813, "grad_norm": 1.1880065957311807, "learning_rate": 0.003, "loss": 4.0545, "step": 21813 }, { "epoch": 0.21814, "grad_norm": 1.2187188051396427, "learning_rate": 0.003, "loss": 4.0351, "step": 21814 }, { "epoch": 0.21815, "grad_norm": 1.399988740367689, "learning_rate": 0.003, "loss": 4.0259, "step": 21815 }, { "epoch": 0.21816, "grad_norm": 1.3127841473145994, "learning_rate": 0.003, "loss": 4.0587, "step": 21816 }, { "epoch": 0.21817, "grad_norm": 1.2047107452225148, "learning_rate": 0.003, "loss": 4.0282, "step": 21817 }, { "epoch": 0.21818, "grad_norm": 1.3808457862570687, "learning_rate": 0.003, "loss": 4.0429, "step": 21818 }, { "epoch": 0.21819, "grad_norm": 1.1588561724439312, "learning_rate": 0.003, "loss": 4.0515, "step": 21819 }, { "epoch": 0.2182, "grad_norm": 1.4001257307576287, "learning_rate": 0.003, "loss": 4.0536, "step": 21820 }, { "epoch": 0.21821, "grad_norm": 1.02993503751475, "learning_rate": 0.003, "loss": 4.058, "step": 21821 }, { "epoch": 0.21822, "grad_norm": 1.4803856847048495, "learning_rate": 0.003, "loss": 4.0219, "step": 21822 }, { "epoch": 0.21823, "grad_norm": 1.1216293175586112, "learning_rate": 0.003, "loss": 4.0768, "step": 21823 }, { "epoch": 0.21824, "grad_norm": 1.5049522871633374, "learning_rate": 0.003, "loss": 4.0432, "step": 21824 }, { "epoch": 0.21825, "grad_norm": 1.0412979291171327, "learning_rate": 0.003, "loss": 4.0596, "step": 21825 }, { "epoch": 0.21826, "grad_norm": 1.4578299600424, "learning_rate": 0.003, "loss": 4.0542, "step": 21826 }, { "epoch": 0.21827, "grad_norm": 1.1201321192220774, "learning_rate": 0.003, "loss": 4.0592, "step": 21827 }, { "epoch": 0.21828, "grad_norm": 1.4657247929765522, "learning_rate": 0.003, "loss": 4.068, "step": 21828 }, { "epoch": 0.21829, "grad_norm": 1.3090297312739307, "learning_rate": 0.003, "loss": 4.0652, "step": 21829 }, { "epoch": 0.2183, "grad_norm": 1.3781285015558673, "learning_rate": 0.003, "loss": 4.054, "step": 21830 }, { "epoch": 0.21831, "grad_norm": 1.21686698334116, "learning_rate": 0.003, "loss": 4.0616, "step": 21831 }, { "epoch": 0.21832, "grad_norm": 1.2566016580753985, "learning_rate": 0.003, "loss": 4.0148, "step": 21832 }, { "epoch": 0.21833, "grad_norm": 1.4487463646569787, "learning_rate": 0.003, "loss": 4.0337, "step": 21833 }, { "epoch": 0.21834, "grad_norm": 1.0677199207991201, "learning_rate": 0.003, "loss": 4.0588, "step": 21834 }, { "epoch": 0.21835, "grad_norm": 1.6613710385839113, "learning_rate": 0.003, "loss": 4.0542, "step": 21835 }, { "epoch": 0.21836, "grad_norm": 1.097299801532908, "learning_rate": 0.003, "loss": 4.0491, "step": 21836 }, { "epoch": 0.21837, "grad_norm": 1.495664002050768, "learning_rate": 0.003, "loss": 4.0354, "step": 21837 }, { "epoch": 0.21838, "grad_norm": 1.1129084471167503, "learning_rate": 0.003, "loss": 4.0666, "step": 21838 }, { "epoch": 0.21839, "grad_norm": 1.5530672756532087, "learning_rate": 0.003, "loss": 4.0497, "step": 21839 }, { "epoch": 0.2184, "grad_norm": 1.1848727302751756, "learning_rate": 0.003, "loss": 4.0225, "step": 21840 }, { "epoch": 0.21841, "grad_norm": 1.340491877693265, "learning_rate": 0.003, "loss": 4.0868, "step": 21841 }, { "epoch": 0.21842, "grad_norm": 1.3901836643028558, "learning_rate": 0.003, "loss": 4.0596, "step": 21842 }, { "epoch": 0.21843, "grad_norm": 1.0581420731535256, "learning_rate": 0.003, "loss": 4.0358, "step": 21843 }, { "epoch": 0.21844, "grad_norm": 1.4694580602654814, "learning_rate": 0.003, "loss": 4.0495, "step": 21844 }, { "epoch": 0.21845, "grad_norm": 1.2246254850083362, "learning_rate": 0.003, "loss": 4.0791, "step": 21845 }, { "epoch": 0.21846, "grad_norm": 1.1792676147967918, "learning_rate": 0.003, "loss": 4.0282, "step": 21846 }, { "epoch": 0.21847, "grad_norm": 1.2479741589582791, "learning_rate": 0.003, "loss": 4.0227, "step": 21847 }, { "epoch": 0.21848, "grad_norm": 1.2744055254108633, "learning_rate": 0.003, "loss": 4.0445, "step": 21848 }, { "epoch": 0.21849, "grad_norm": 1.6129612207504957, "learning_rate": 0.003, "loss": 4.0304, "step": 21849 }, { "epoch": 0.2185, "grad_norm": 1.3219532561820633, "learning_rate": 0.003, "loss": 4.0381, "step": 21850 }, { "epoch": 0.21851, "grad_norm": 1.3018424802157766, "learning_rate": 0.003, "loss": 4.0541, "step": 21851 }, { "epoch": 0.21852, "grad_norm": 1.3628580496989582, "learning_rate": 0.003, "loss": 4.0271, "step": 21852 }, { "epoch": 0.21853, "grad_norm": 1.1376207302177839, "learning_rate": 0.003, "loss": 4.0502, "step": 21853 }, { "epoch": 0.21854, "grad_norm": 1.2694024402813655, "learning_rate": 0.003, "loss": 4.0629, "step": 21854 }, { "epoch": 0.21855, "grad_norm": 1.3893402012442544, "learning_rate": 0.003, "loss": 4.0586, "step": 21855 }, { "epoch": 0.21856, "grad_norm": 1.197721354828964, "learning_rate": 0.003, "loss": 4.0631, "step": 21856 }, { "epoch": 0.21857, "grad_norm": 1.3381212107701923, "learning_rate": 0.003, "loss": 4.0303, "step": 21857 }, { "epoch": 0.21858, "grad_norm": 0.9806807309955483, "learning_rate": 0.003, "loss": 4.0094, "step": 21858 }, { "epoch": 0.21859, "grad_norm": 1.56375667778978, "learning_rate": 0.003, "loss": 4.0543, "step": 21859 }, { "epoch": 0.2186, "grad_norm": 1.3261575869782218, "learning_rate": 0.003, "loss": 4.0395, "step": 21860 }, { "epoch": 0.21861, "grad_norm": 1.261235160635816, "learning_rate": 0.003, "loss": 4.0383, "step": 21861 }, { "epoch": 0.21862, "grad_norm": 1.1608136640008393, "learning_rate": 0.003, "loss": 4.0341, "step": 21862 }, { "epoch": 0.21863, "grad_norm": 1.4585535655625999, "learning_rate": 0.003, "loss": 4.0356, "step": 21863 }, { "epoch": 0.21864, "grad_norm": 1.2657939021710913, "learning_rate": 0.003, "loss": 4.0041, "step": 21864 }, { "epoch": 0.21865, "grad_norm": 1.1623417582456166, "learning_rate": 0.003, "loss": 4.0682, "step": 21865 }, { "epoch": 0.21866, "grad_norm": 1.3911128896185476, "learning_rate": 0.003, "loss": 4.0354, "step": 21866 }, { "epoch": 0.21867, "grad_norm": 1.1625706876939281, "learning_rate": 0.003, "loss": 4.0611, "step": 21867 }, { "epoch": 0.21868, "grad_norm": 1.3387012015290805, "learning_rate": 0.003, "loss": 4.039, "step": 21868 }, { "epoch": 0.21869, "grad_norm": 1.314110378401501, "learning_rate": 0.003, "loss": 4.0464, "step": 21869 }, { "epoch": 0.2187, "grad_norm": 0.9999991750378531, "learning_rate": 0.003, "loss": 4.0158, "step": 21870 }, { "epoch": 0.21871, "grad_norm": 1.5012812845497798, "learning_rate": 0.003, "loss": 4.0572, "step": 21871 }, { "epoch": 0.21872, "grad_norm": 1.2033947428833383, "learning_rate": 0.003, "loss": 4.0446, "step": 21872 }, { "epoch": 0.21873, "grad_norm": 1.1355665356272322, "learning_rate": 0.003, "loss": 4.0293, "step": 21873 }, { "epoch": 0.21874, "grad_norm": 1.3781499258525007, "learning_rate": 0.003, "loss": 4.0604, "step": 21874 }, { "epoch": 0.21875, "grad_norm": 1.1076317553859993, "learning_rate": 0.003, "loss": 4.0453, "step": 21875 }, { "epoch": 0.21876, "grad_norm": 1.312420119942699, "learning_rate": 0.003, "loss": 4.0332, "step": 21876 }, { "epoch": 0.21877, "grad_norm": 1.260117273022279, "learning_rate": 0.003, "loss": 4.0763, "step": 21877 }, { "epoch": 0.21878, "grad_norm": 1.2736011384829615, "learning_rate": 0.003, "loss": 4.0211, "step": 21878 }, { "epoch": 0.21879, "grad_norm": 1.6563759600413785, "learning_rate": 0.003, "loss": 4.0573, "step": 21879 }, { "epoch": 0.2188, "grad_norm": 1.214488166570031, "learning_rate": 0.003, "loss": 4.0343, "step": 21880 }, { "epoch": 0.21881, "grad_norm": 1.7308811759401588, "learning_rate": 0.003, "loss": 4.0716, "step": 21881 }, { "epoch": 0.21882, "grad_norm": 1.2657519703370974, "learning_rate": 0.003, "loss": 4.0296, "step": 21882 }, { "epoch": 0.21883, "grad_norm": 1.2021609269771085, "learning_rate": 0.003, "loss": 4.0209, "step": 21883 }, { "epoch": 0.21884, "grad_norm": 1.5947524084221563, "learning_rate": 0.003, "loss": 4.069, "step": 21884 }, { "epoch": 0.21885, "grad_norm": 1.1360520615617342, "learning_rate": 0.003, "loss": 4.0551, "step": 21885 }, { "epoch": 0.21886, "grad_norm": 1.7329244709863258, "learning_rate": 0.003, "loss": 4.1144, "step": 21886 }, { "epoch": 0.21887, "grad_norm": 0.8703951983084132, "learning_rate": 0.003, "loss": 4.0266, "step": 21887 }, { "epoch": 0.21888, "grad_norm": 1.243980851234002, "learning_rate": 0.003, "loss": 4.0337, "step": 21888 }, { "epoch": 0.21889, "grad_norm": 1.278766372745449, "learning_rate": 0.003, "loss": 4.0421, "step": 21889 }, { "epoch": 0.2189, "grad_norm": 1.6718724133847218, "learning_rate": 0.003, "loss": 4.0451, "step": 21890 }, { "epoch": 0.21891, "grad_norm": 1.145674094050216, "learning_rate": 0.003, "loss": 4.0499, "step": 21891 }, { "epoch": 0.21892, "grad_norm": 1.1523355795045476, "learning_rate": 0.003, "loss": 4.0772, "step": 21892 }, { "epoch": 0.21893, "grad_norm": 1.2014596209351205, "learning_rate": 0.003, "loss": 4.0242, "step": 21893 }, { "epoch": 0.21894, "grad_norm": 1.3958011540508617, "learning_rate": 0.003, "loss": 4.02, "step": 21894 }, { "epoch": 0.21895, "grad_norm": 1.2393398588122093, "learning_rate": 0.003, "loss": 4.0329, "step": 21895 }, { "epoch": 0.21896, "grad_norm": 1.3347735006817876, "learning_rate": 0.003, "loss": 4.0563, "step": 21896 }, { "epoch": 0.21897, "grad_norm": 1.1324565067433832, "learning_rate": 0.003, "loss": 4.0465, "step": 21897 }, { "epoch": 0.21898, "grad_norm": 1.2039209931932169, "learning_rate": 0.003, "loss": 4.0598, "step": 21898 }, { "epoch": 0.21899, "grad_norm": 1.221239275920937, "learning_rate": 0.003, "loss": 4.0439, "step": 21899 }, { "epoch": 0.219, "grad_norm": 1.3846023565342906, "learning_rate": 0.003, "loss": 4.0642, "step": 21900 }, { "epoch": 0.21901, "grad_norm": 1.2097233432992271, "learning_rate": 0.003, "loss": 4.0429, "step": 21901 }, { "epoch": 0.21902, "grad_norm": 1.5307952520411225, "learning_rate": 0.003, "loss": 4.0454, "step": 21902 }, { "epoch": 0.21903, "grad_norm": 1.1426380684974333, "learning_rate": 0.003, "loss": 4.0447, "step": 21903 }, { "epoch": 0.21904, "grad_norm": 1.4425273570195094, "learning_rate": 0.003, "loss": 4.0347, "step": 21904 }, { "epoch": 0.21905, "grad_norm": 1.2135500938696013, "learning_rate": 0.003, "loss": 4.036, "step": 21905 }, { "epoch": 0.21906, "grad_norm": 1.4791938122224157, "learning_rate": 0.003, "loss": 4.0411, "step": 21906 }, { "epoch": 0.21907, "grad_norm": 1.2385415384800407, "learning_rate": 0.003, "loss": 4.0302, "step": 21907 }, { "epoch": 0.21908, "grad_norm": 1.2204377059545255, "learning_rate": 0.003, "loss": 4.046, "step": 21908 }, { "epoch": 0.21909, "grad_norm": 1.4729856305254043, "learning_rate": 0.003, "loss": 4.0325, "step": 21909 }, { "epoch": 0.2191, "grad_norm": 1.1580053596916617, "learning_rate": 0.003, "loss": 4.0747, "step": 21910 }, { "epoch": 0.21911, "grad_norm": 1.4685645178871252, "learning_rate": 0.003, "loss": 4.0562, "step": 21911 }, { "epoch": 0.21912, "grad_norm": 1.1270720819698852, "learning_rate": 0.003, "loss": 4.0425, "step": 21912 }, { "epoch": 0.21913, "grad_norm": 1.4187154875858043, "learning_rate": 0.003, "loss": 4.0518, "step": 21913 }, { "epoch": 0.21914, "grad_norm": 1.110683647538591, "learning_rate": 0.003, "loss": 4.0643, "step": 21914 }, { "epoch": 0.21915, "grad_norm": 1.547151058553336, "learning_rate": 0.003, "loss": 4.081, "step": 21915 }, { "epoch": 0.21916, "grad_norm": 1.1328535661260921, "learning_rate": 0.003, "loss": 4.0692, "step": 21916 }, { "epoch": 0.21917, "grad_norm": 1.372936148828246, "learning_rate": 0.003, "loss": 4.0621, "step": 21917 }, { "epoch": 0.21918, "grad_norm": 1.4214768619666374, "learning_rate": 0.003, "loss": 4.0404, "step": 21918 }, { "epoch": 0.21919, "grad_norm": 1.3312621969706127, "learning_rate": 0.003, "loss": 4.0775, "step": 21919 }, { "epoch": 0.2192, "grad_norm": 1.245775162625903, "learning_rate": 0.003, "loss": 4.0308, "step": 21920 }, { "epoch": 0.21921, "grad_norm": 1.3279234469523127, "learning_rate": 0.003, "loss": 4.0221, "step": 21921 }, { "epoch": 0.21922, "grad_norm": 1.4554362869485136, "learning_rate": 0.003, "loss": 4.0324, "step": 21922 }, { "epoch": 0.21923, "grad_norm": 1.2809881640912781, "learning_rate": 0.003, "loss": 4.0334, "step": 21923 }, { "epoch": 0.21924, "grad_norm": 1.354197059303451, "learning_rate": 0.003, "loss": 4.0812, "step": 21924 }, { "epoch": 0.21925, "grad_norm": 1.2019574280100456, "learning_rate": 0.003, "loss": 4.0646, "step": 21925 }, { "epoch": 0.21926, "grad_norm": 1.1380669860661248, "learning_rate": 0.003, "loss": 4.0319, "step": 21926 }, { "epoch": 0.21927, "grad_norm": 1.4336369214880056, "learning_rate": 0.003, "loss": 4.0559, "step": 21927 }, { "epoch": 0.21928, "grad_norm": 1.1654398339180754, "learning_rate": 0.003, "loss": 4.0742, "step": 21928 }, { "epoch": 0.21929, "grad_norm": 1.4196392347750213, "learning_rate": 0.003, "loss": 4.0845, "step": 21929 }, { "epoch": 0.2193, "grad_norm": 1.0921325701490219, "learning_rate": 0.003, "loss": 4.0373, "step": 21930 }, { "epoch": 0.21931, "grad_norm": 1.4207894374679884, "learning_rate": 0.003, "loss": 4.0584, "step": 21931 }, { "epoch": 0.21932, "grad_norm": 1.2890895537088862, "learning_rate": 0.003, "loss": 4.0366, "step": 21932 }, { "epoch": 0.21933, "grad_norm": 1.7586636158569737, "learning_rate": 0.003, "loss": 4.059, "step": 21933 }, { "epoch": 0.21934, "grad_norm": 1.1272703711037795, "learning_rate": 0.003, "loss": 4.0568, "step": 21934 }, { "epoch": 0.21935, "grad_norm": 1.2564020111487622, "learning_rate": 0.003, "loss": 4.0428, "step": 21935 }, { "epoch": 0.21936, "grad_norm": 1.1981853982583015, "learning_rate": 0.003, "loss": 4.065, "step": 21936 }, { "epoch": 0.21937, "grad_norm": 1.2212148482772398, "learning_rate": 0.003, "loss": 4.0292, "step": 21937 }, { "epoch": 0.21938, "grad_norm": 1.190855324309062, "learning_rate": 0.003, "loss": 4.0844, "step": 21938 }, { "epoch": 0.21939, "grad_norm": 1.4013183516912304, "learning_rate": 0.003, "loss": 4.0444, "step": 21939 }, { "epoch": 0.2194, "grad_norm": 1.134892224515106, "learning_rate": 0.003, "loss": 4.0532, "step": 21940 }, { "epoch": 0.21941, "grad_norm": 1.301908871170717, "learning_rate": 0.003, "loss": 4.0859, "step": 21941 }, { "epoch": 0.21942, "grad_norm": 1.1932679515891222, "learning_rate": 0.003, "loss": 4.0398, "step": 21942 }, { "epoch": 0.21943, "grad_norm": 1.3558064809702928, "learning_rate": 0.003, "loss": 4.0262, "step": 21943 }, { "epoch": 0.21944, "grad_norm": 1.21968860972223, "learning_rate": 0.003, "loss": 4.0693, "step": 21944 }, { "epoch": 0.21945, "grad_norm": 1.3174656195034329, "learning_rate": 0.003, "loss": 4.0385, "step": 21945 }, { "epoch": 0.21946, "grad_norm": 1.3446320833830796, "learning_rate": 0.003, "loss": 4.0429, "step": 21946 }, { "epoch": 0.21947, "grad_norm": 1.6524268096393475, "learning_rate": 0.003, "loss": 4.0752, "step": 21947 }, { "epoch": 0.21948, "grad_norm": 1.2762365175000143, "learning_rate": 0.003, "loss": 4.0521, "step": 21948 }, { "epoch": 0.21949, "grad_norm": 1.4106233865194429, "learning_rate": 0.003, "loss": 4.0558, "step": 21949 }, { "epoch": 0.2195, "grad_norm": 1.2292659867037987, "learning_rate": 0.003, "loss": 4.0175, "step": 21950 }, { "epoch": 0.21951, "grad_norm": 1.5629695436942541, "learning_rate": 0.003, "loss": 4.0256, "step": 21951 }, { "epoch": 0.21952, "grad_norm": 1.0618016722660668, "learning_rate": 0.003, "loss": 4.0351, "step": 21952 }, { "epoch": 0.21953, "grad_norm": 1.426553492728306, "learning_rate": 0.003, "loss": 4.045, "step": 21953 }, { "epoch": 0.21954, "grad_norm": 1.2761199006119757, "learning_rate": 0.003, "loss": 4.0538, "step": 21954 }, { "epoch": 0.21955, "grad_norm": 1.2847758400514797, "learning_rate": 0.003, "loss": 4.0435, "step": 21955 }, { "epoch": 0.21956, "grad_norm": 1.2110449349287116, "learning_rate": 0.003, "loss": 4.0426, "step": 21956 }, { "epoch": 0.21957, "grad_norm": 1.30524349065345, "learning_rate": 0.003, "loss": 4.026, "step": 21957 }, { "epoch": 0.21958, "grad_norm": 1.0488535033113748, "learning_rate": 0.003, "loss": 4.0318, "step": 21958 }, { "epoch": 0.21959, "grad_norm": 1.3720199191394824, "learning_rate": 0.003, "loss": 4.0398, "step": 21959 }, { "epoch": 0.2196, "grad_norm": 1.1758659575453692, "learning_rate": 0.003, "loss": 4.054, "step": 21960 }, { "epoch": 0.21961, "grad_norm": 1.3792286043631028, "learning_rate": 0.003, "loss": 4.0732, "step": 21961 }, { "epoch": 0.21962, "grad_norm": 1.2129583578478145, "learning_rate": 0.003, "loss": 4.0442, "step": 21962 }, { "epoch": 0.21963, "grad_norm": 1.4881039884514828, "learning_rate": 0.003, "loss": 4.0691, "step": 21963 }, { "epoch": 0.21964, "grad_norm": 1.2358647782456937, "learning_rate": 0.003, "loss": 4.0195, "step": 21964 }, { "epoch": 0.21965, "grad_norm": 1.884750601200753, "learning_rate": 0.003, "loss": 4.0324, "step": 21965 }, { "epoch": 0.21966, "grad_norm": 1.115206786447901, "learning_rate": 0.003, "loss": 4.0652, "step": 21966 }, { "epoch": 0.21967, "grad_norm": 1.1835173732623296, "learning_rate": 0.003, "loss": 4.0517, "step": 21967 }, { "epoch": 0.21968, "grad_norm": 1.2952447568803054, "learning_rate": 0.003, "loss": 4.0379, "step": 21968 }, { "epoch": 0.21969, "grad_norm": 1.1173047032920922, "learning_rate": 0.003, "loss": 4.063, "step": 21969 }, { "epoch": 0.2197, "grad_norm": 1.376174741074243, "learning_rate": 0.003, "loss": 4.0569, "step": 21970 }, { "epoch": 0.21971, "grad_norm": 1.2395593887738778, "learning_rate": 0.003, "loss": 4.0533, "step": 21971 }, { "epoch": 0.21972, "grad_norm": 1.2924060549475613, "learning_rate": 0.003, "loss": 4.056, "step": 21972 }, { "epoch": 0.21973, "grad_norm": 1.3933521899632437, "learning_rate": 0.003, "loss": 4.0655, "step": 21973 }, { "epoch": 0.21974, "grad_norm": 1.4489330442733404, "learning_rate": 0.003, "loss": 4.0874, "step": 21974 }, { "epoch": 0.21975, "grad_norm": 1.3402591915023505, "learning_rate": 0.003, "loss": 4.022, "step": 21975 }, { "epoch": 0.21976, "grad_norm": 1.1361412490524445, "learning_rate": 0.003, "loss": 4.0417, "step": 21976 }, { "epoch": 0.21977, "grad_norm": 1.3929749248907122, "learning_rate": 0.003, "loss": 4.0782, "step": 21977 }, { "epoch": 0.21978, "grad_norm": 1.3121856380564516, "learning_rate": 0.003, "loss": 4.0825, "step": 21978 }, { "epoch": 0.21979, "grad_norm": 1.2598021636776788, "learning_rate": 0.003, "loss": 4.0592, "step": 21979 }, { "epoch": 0.2198, "grad_norm": 1.3883881259052413, "learning_rate": 0.003, "loss": 4.0286, "step": 21980 }, { "epoch": 0.21981, "grad_norm": 1.3650733820778755, "learning_rate": 0.003, "loss": 4.0547, "step": 21981 }, { "epoch": 0.21982, "grad_norm": 1.2185180588385176, "learning_rate": 0.003, "loss": 4.0511, "step": 21982 }, { "epoch": 0.21983, "grad_norm": 1.405136322415539, "learning_rate": 0.003, "loss": 4.0634, "step": 21983 }, { "epoch": 0.21984, "grad_norm": 1.1392482188167596, "learning_rate": 0.003, "loss": 4.0464, "step": 21984 }, { "epoch": 0.21985, "grad_norm": 1.515030162640994, "learning_rate": 0.003, "loss": 4.0607, "step": 21985 }, { "epoch": 0.21986, "grad_norm": 1.1278694570469598, "learning_rate": 0.003, "loss": 4.0613, "step": 21986 }, { "epoch": 0.21987, "grad_norm": 1.5710847577934575, "learning_rate": 0.003, "loss": 4.0603, "step": 21987 }, { "epoch": 0.21988, "grad_norm": 1.7399635521600438, "learning_rate": 0.003, "loss": 4.0501, "step": 21988 }, { "epoch": 0.21989, "grad_norm": 1.4419622166426336, "learning_rate": 0.003, "loss": 4.0655, "step": 21989 }, { "epoch": 0.2199, "grad_norm": 1.1323052650451653, "learning_rate": 0.003, "loss": 4.0554, "step": 21990 }, { "epoch": 0.21991, "grad_norm": 1.3395754195874743, "learning_rate": 0.003, "loss": 4.0555, "step": 21991 }, { "epoch": 0.21992, "grad_norm": 1.105614324419437, "learning_rate": 0.003, "loss": 4.0365, "step": 21992 }, { "epoch": 0.21993, "grad_norm": 1.3279786482364344, "learning_rate": 0.003, "loss": 4.0305, "step": 21993 }, { "epoch": 0.21994, "grad_norm": 1.1997867243597922, "learning_rate": 0.003, "loss": 4.0364, "step": 21994 }, { "epoch": 0.21995, "grad_norm": 1.2714493569781429, "learning_rate": 0.003, "loss": 4.0343, "step": 21995 }, { "epoch": 0.21996, "grad_norm": 1.159293005396773, "learning_rate": 0.003, "loss": 4.0691, "step": 21996 }, { "epoch": 0.21997, "grad_norm": 1.282598504075124, "learning_rate": 0.003, "loss": 4.0765, "step": 21997 }, { "epoch": 0.21998, "grad_norm": 1.1958611546236209, "learning_rate": 0.003, "loss": 4.0443, "step": 21998 }, { "epoch": 0.21999, "grad_norm": 1.3805033783662493, "learning_rate": 0.003, "loss": 4.0489, "step": 21999 }, { "epoch": 0.22, "grad_norm": 1.2006099392204475, "learning_rate": 0.003, "loss": 4.0534, "step": 22000 }, { "epoch": 0.22001, "grad_norm": 1.6004264636382097, "learning_rate": 0.003, "loss": 4.0633, "step": 22001 }, { "epoch": 0.22002, "grad_norm": 1.1289168964241154, "learning_rate": 0.003, "loss": 4.0629, "step": 22002 }, { "epoch": 0.22003, "grad_norm": 1.310963994707133, "learning_rate": 0.003, "loss": 4.0752, "step": 22003 }, { "epoch": 0.22004, "grad_norm": 1.1871691746716844, "learning_rate": 0.003, "loss": 4.0506, "step": 22004 }, { "epoch": 0.22005, "grad_norm": 1.3996480373055058, "learning_rate": 0.003, "loss": 4.0671, "step": 22005 }, { "epoch": 0.22006, "grad_norm": 1.4427349936785299, "learning_rate": 0.003, "loss": 4.0564, "step": 22006 }, { "epoch": 0.22007, "grad_norm": 1.03229091907169, "learning_rate": 0.003, "loss": 4.0256, "step": 22007 }, { "epoch": 0.22008, "grad_norm": 1.979030657273191, "learning_rate": 0.003, "loss": 4.0471, "step": 22008 }, { "epoch": 0.22009, "grad_norm": 1.0712747011990373, "learning_rate": 0.003, "loss": 4.0341, "step": 22009 }, { "epoch": 0.2201, "grad_norm": 1.220404774144469, "learning_rate": 0.003, "loss": 4.0456, "step": 22010 }, { "epoch": 0.22011, "grad_norm": 1.3574732714764237, "learning_rate": 0.003, "loss": 4.0572, "step": 22011 }, { "epoch": 0.22012, "grad_norm": 1.2587755650236263, "learning_rate": 0.003, "loss": 4.0516, "step": 22012 }, { "epoch": 0.22013, "grad_norm": 1.3788069975716672, "learning_rate": 0.003, "loss": 4.0319, "step": 22013 }, { "epoch": 0.22014, "grad_norm": 1.3649281869718972, "learning_rate": 0.003, "loss": 4.0335, "step": 22014 }, { "epoch": 0.22015, "grad_norm": 1.3188502903880825, "learning_rate": 0.003, "loss": 4.0534, "step": 22015 }, { "epoch": 0.22016, "grad_norm": 1.2532977210053056, "learning_rate": 0.003, "loss": 4.0664, "step": 22016 }, { "epoch": 0.22017, "grad_norm": 1.2454509901370068, "learning_rate": 0.003, "loss": 4.0604, "step": 22017 }, { "epoch": 0.22018, "grad_norm": 1.2219080505179116, "learning_rate": 0.003, "loss": 4.0596, "step": 22018 }, { "epoch": 0.22019, "grad_norm": 1.227557250799576, "learning_rate": 0.003, "loss": 4.0392, "step": 22019 }, { "epoch": 0.2202, "grad_norm": 1.4217461690897921, "learning_rate": 0.003, "loss": 4.0552, "step": 22020 }, { "epoch": 0.22021, "grad_norm": 1.1745684753740826, "learning_rate": 0.003, "loss": 4.0824, "step": 22021 }, { "epoch": 0.22022, "grad_norm": 1.3316959051174804, "learning_rate": 0.003, "loss": 4.0358, "step": 22022 }, { "epoch": 0.22023, "grad_norm": 1.132020313014203, "learning_rate": 0.003, "loss": 4.049, "step": 22023 }, { "epoch": 0.22024, "grad_norm": 1.391354676802887, "learning_rate": 0.003, "loss": 4.062, "step": 22024 }, { "epoch": 0.22025, "grad_norm": 1.2009920679822612, "learning_rate": 0.003, "loss": 4.0328, "step": 22025 }, { "epoch": 0.22026, "grad_norm": 1.3050770476118223, "learning_rate": 0.003, "loss": 4.0464, "step": 22026 }, { "epoch": 0.22027, "grad_norm": 1.2710427521381245, "learning_rate": 0.003, "loss": 4.0723, "step": 22027 }, { "epoch": 0.22028, "grad_norm": 1.512835456341759, "learning_rate": 0.003, "loss": 4.0689, "step": 22028 }, { "epoch": 0.22029, "grad_norm": 1.0541873228922884, "learning_rate": 0.003, "loss": 4.0453, "step": 22029 }, { "epoch": 0.2203, "grad_norm": 1.5742314943388278, "learning_rate": 0.003, "loss": 4.0667, "step": 22030 }, { "epoch": 0.22031, "grad_norm": 1.28883727437431, "learning_rate": 0.003, "loss": 4.0419, "step": 22031 }, { "epoch": 0.22032, "grad_norm": 1.3461851868328158, "learning_rate": 0.003, "loss": 4.0695, "step": 22032 }, { "epoch": 0.22033, "grad_norm": 1.334434768183276, "learning_rate": 0.003, "loss": 4.0657, "step": 22033 }, { "epoch": 0.22034, "grad_norm": 1.1809934324254372, "learning_rate": 0.003, "loss": 4.0476, "step": 22034 }, { "epoch": 0.22035, "grad_norm": 1.3568723623825492, "learning_rate": 0.003, "loss": 4.0598, "step": 22035 }, { "epoch": 0.22036, "grad_norm": 1.1667312162928247, "learning_rate": 0.003, "loss": 4.067, "step": 22036 }, { "epoch": 0.22037, "grad_norm": 1.3838957074239497, "learning_rate": 0.003, "loss": 4.071, "step": 22037 }, { "epoch": 0.22038, "grad_norm": 1.064283880214727, "learning_rate": 0.003, "loss": 4.0492, "step": 22038 }, { "epoch": 0.22039, "grad_norm": 1.5433932879794727, "learning_rate": 0.003, "loss": 4.088, "step": 22039 }, { "epoch": 0.2204, "grad_norm": 1.0299755896576255, "learning_rate": 0.003, "loss": 4.056, "step": 22040 }, { "epoch": 0.22041, "grad_norm": 1.3830542724373722, "learning_rate": 0.003, "loss": 4.0495, "step": 22041 }, { "epoch": 0.22042, "grad_norm": 1.3030790728249382, "learning_rate": 0.003, "loss": 4.0662, "step": 22042 }, { "epoch": 0.22043, "grad_norm": 1.1550599226862925, "learning_rate": 0.003, "loss": 4.0473, "step": 22043 }, { "epoch": 0.22044, "grad_norm": 1.3085847492254559, "learning_rate": 0.003, "loss": 4.0531, "step": 22044 }, { "epoch": 0.22045, "grad_norm": 1.1182409119330847, "learning_rate": 0.003, "loss": 4.0636, "step": 22045 }, { "epoch": 0.22046, "grad_norm": 1.5194677321955776, "learning_rate": 0.003, "loss": 4.0307, "step": 22046 }, { "epoch": 0.22047, "grad_norm": 1.2218773568374732, "learning_rate": 0.003, "loss": 4.0585, "step": 22047 }, { "epoch": 0.22048, "grad_norm": 1.576138031865719, "learning_rate": 0.003, "loss": 4.0606, "step": 22048 }, { "epoch": 0.22049, "grad_norm": 1.239135079172457, "learning_rate": 0.003, "loss": 4.0322, "step": 22049 }, { "epoch": 0.2205, "grad_norm": 1.1697429723915767, "learning_rate": 0.003, "loss": 4.0196, "step": 22050 }, { "epoch": 0.22051, "grad_norm": 1.2253034134469498, "learning_rate": 0.003, "loss": 4.0265, "step": 22051 }, { "epoch": 0.22052, "grad_norm": 1.249720705142228, "learning_rate": 0.003, "loss": 4.0365, "step": 22052 }, { "epoch": 0.22053, "grad_norm": 1.3785068123116304, "learning_rate": 0.003, "loss": 4.0317, "step": 22053 }, { "epoch": 0.22054, "grad_norm": 0.9985964205797072, "learning_rate": 0.003, "loss": 4.0282, "step": 22054 }, { "epoch": 0.22055, "grad_norm": 1.289746726837481, "learning_rate": 0.003, "loss": 4.01, "step": 22055 }, { "epoch": 0.22056, "grad_norm": 1.2888530434697942, "learning_rate": 0.003, "loss": 4.0608, "step": 22056 }, { "epoch": 0.22057, "grad_norm": 1.4008643616024095, "learning_rate": 0.003, "loss": 4.0664, "step": 22057 }, { "epoch": 0.22058, "grad_norm": 1.3198347378582163, "learning_rate": 0.003, "loss": 4.0459, "step": 22058 }, { "epoch": 0.22059, "grad_norm": 1.184716181331145, "learning_rate": 0.003, "loss": 4.0198, "step": 22059 }, { "epoch": 0.2206, "grad_norm": 1.2195267100300138, "learning_rate": 0.003, "loss": 4.0593, "step": 22060 }, { "epoch": 0.22061, "grad_norm": 1.3366811902149836, "learning_rate": 0.003, "loss": 4.0273, "step": 22061 }, { "epoch": 0.22062, "grad_norm": 1.3207975326265124, "learning_rate": 0.003, "loss": 4.0191, "step": 22062 }, { "epoch": 0.22063, "grad_norm": 1.542595951567761, "learning_rate": 0.003, "loss": 4.0377, "step": 22063 }, { "epoch": 0.22064, "grad_norm": 1.059672367589958, "learning_rate": 0.003, "loss": 4.0559, "step": 22064 }, { "epoch": 0.22065, "grad_norm": 1.3609537408360166, "learning_rate": 0.003, "loss": 4.0058, "step": 22065 }, { "epoch": 0.22066, "grad_norm": 1.2922115423020513, "learning_rate": 0.003, "loss": 4.0149, "step": 22066 }, { "epoch": 0.22067, "grad_norm": 1.3724490104352296, "learning_rate": 0.003, "loss": 4.0505, "step": 22067 }, { "epoch": 0.22068, "grad_norm": 1.2304597950909055, "learning_rate": 0.003, "loss": 4.0186, "step": 22068 }, { "epoch": 0.22069, "grad_norm": 1.4060815997573863, "learning_rate": 0.003, "loss": 4.05, "step": 22069 }, { "epoch": 0.2207, "grad_norm": 1.3168133631060361, "learning_rate": 0.003, "loss": 4.0443, "step": 22070 }, { "epoch": 0.22071, "grad_norm": 1.1000232768231184, "learning_rate": 0.003, "loss": 4.0565, "step": 22071 }, { "epoch": 0.22072, "grad_norm": 1.2985061789221102, "learning_rate": 0.003, "loss": 4.0617, "step": 22072 }, { "epoch": 0.22073, "grad_norm": 1.3343167037257264, "learning_rate": 0.003, "loss": 4.0653, "step": 22073 }, { "epoch": 0.22074, "grad_norm": 1.5818526276304599, "learning_rate": 0.003, "loss": 4.05, "step": 22074 }, { "epoch": 0.22075, "grad_norm": 1.2370323629411177, "learning_rate": 0.003, "loss": 4.0338, "step": 22075 }, { "epoch": 0.22076, "grad_norm": 1.2236372247133998, "learning_rate": 0.003, "loss": 4.0795, "step": 22076 }, { "epoch": 0.22077, "grad_norm": 1.4676775252690224, "learning_rate": 0.003, "loss": 4.0483, "step": 22077 }, { "epoch": 0.22078, "grad_norm": 1.2644907515701838, "learning_rate": 0.003, "loss": 4.0414, "step": 22078 }, { "epoch": 0.22079, "grad_norm": 1.2902745386762855, "learning_rate": 0.003, "loss": 4.0549, "step": 22079 }, { "epoch": 0.2208, "grad_norm": 1.3074531077802405, "learning_rate": 0.003, "loss": 4.0496, "step": 22080 }, { "epoch": 0.22081, "grad_norm": 1.324892085960681, "learning_rate": 0.003, "loss": 4.0516, "step": 22081 }, { "epoch": 0.22082, "grad_norm": 1.227464840408412, "learning_rate": 0.003, "loss": 4.05, "step": 22082 }, { "epoch": 0.22083, "grad_norm": 1.4387552546135856, "learning_rate": 0.003, "loss": 4.0439, "step": 22083 }, { "epoch": 0.22084, "grad_norm": 1.0736599163266853, "learning_rate": 0.003, "loss": 4.0143, "step": 22084 }, { "epoch": 0.22085, "grad_norm": 1.4733321761073723, "learning_rate": 0.003, "loss": 4.0505, "step": 22085 }, { "epoch": 0.22086, "grad_norm": 0.9524379310571377, "learning_rate": 0.003, "loss": 4.0619, "step": 22086 }, { "epoch": 0.22087, "grad_norm": 1.3604123053335961, "learning_rate": 0.003, "loss": 4.0533, "step": 22087 }, { "epoch": 0.22088, "grad_norm": 1.229120424668029, "learning_rate": 0.003, "loss": 4.05, "step": 22088 }, { "epoch": 0.22089, "grad_norm": 1.0602883349987202, "learning_rate": 0.003, "loss": 4.0594, "step": 22089 }, { "epoch": 0.2209, "grad_norm": 1.3339922136285414, "learning_rate": 0.003, "loss": 4.0605, "step": 22090 }, { "epoch": 0.22091, "grad_norm": 1.1983700160262636, "learning_rate": 0.003, "loss": 4.0453, "step": 22091 }, { "epoch": 0.22092, "grad_norm": 1.4439405413795647, "learning_rate": 0.003, "loss": 4.0612, "step": 22092 }, { "epoch": 0.22093, "grad_norm": 1.2562344082226966, "learning_rate": 0.003, "loss": 4.0406, "step": 22093 }, { "epoch": 0.22094, "grad_norm": 1.3410934863378028, "learning_rate": 0.003, "loss": 4.0426, "step": 22094 }, { "epoch": 0.22095, "grad_norm": 1.5791392824791675, "learning_rate": 0.003, "loss": 4.0243, "step": 22095 }, { "epoch": 0.22096, "grad_norm": 1.2348082012269177, "learning_rate": 0.003, "loss": 4.0819, "step": 22096 }, { "epoch": 0.22097, "grad_norm": 1.2668458739879667, "learning_rate": 0.003, "loss": 4.0338, "step": 22097 }, { "epoch": 0.22098, "grad_norm": 1.33665490855949, "learning_rate": 0.003, "loss": 4.0221, "step": 22098 }, { "epoch": 0.22099, "grad_norm": 1.3129218902648512, "learning_rate": 0.003, "loss": 4.0261, "step": 22099 }, { "epoch": 0.221, "grad_norm": 1.1421527114272347, "learning_rate": 0.003, "loss": 4.0514, "step": 22100 }, { "epoch": 0.22101, "grad_norm": 1.293725981566844, "learning_rate": 0.003, "loss": 4.0632, "step": 22101 }, { "epoch": 0.22102, "grad_norm": 1.1726341143874381, "learning_rate": 0.003, "loss": 4.0384, "step": 22102 }, { "epoch": 0.22103, "grad_norm": 1.362723647667269, "learning_rate": 0.003, "loss": 4.0695, "step": 22103 }, { "epoch": 0.22104, "grad_norm": 1.2387148849942935, "learning_rate": 0.003, "loss": 4.0699, "step": 22104 }, { "epoch": 0.22105, "grad_norm": 1.4540185131777992, "learning_rate": 0.003, "loss": 4.0441, "step": 22105 }, { "epoch": 0.22106, "grad_norm": 1.271059598551409, "learning_rate": 0.003, "loss": 4.0417, "step": 22106 }, { "epoch": 0.22107, "grad_norm": 1.2978366936481063, "learning_rate": 0.003, "loss": 4.0572, "step": 22107 }, { "epoch": 0.22108, "grad_norm": 1.178265137297338, "learning_rate": 0.003, "loss": 4.0272, "step": 22108 }, { "epoch": 0.22109, "grad_norm": 1.3938851889024702, "learning_rate": 0.003, "loss": 4.0534, "step": 22109 }, { "epoch": 0.2211, "grad_norm": 1.2962696073975482, "learning_rate": 0.003, "loss": 4.0559, "step": 22110 }, { "epoch": 0.22111, "grad_norm": 1.1512034284164956, "learning_rate": 0.003, "loss": 4.009, "step": 22111 }, { "epoch": 0.22112, "grad_norm": 1.3157831848230688, "learning_rate": 0.003, "loss": 4.0292, "step": 22112 }, { "epoch": 0.22113, "grad_norm": 1.327099346972767, "learning_rate": 0.003, "loss": 4.0565, "step": 22113 }, { "epoch": 0.22114, "grad_norm": 1.418072769481291, "learning_rate": 0.003, "loss": 4.0417, "step": 22114 }, { "epoch": 0.22115, "grad_norm": 1.120536080965957, "learning_rate": 0.003, "loss": 4.0425, "step": 22115 }, { "epoch": 0.22116, "grad_norm": 1.401330009769329, "learning_rate": 0.003, "loss": 4.0329, "step": 22116 }, { "epoch": 0.22117, "grad_norm": 1.1719044404065446, "learning_rate": 0.003, "loss": 4.0335, "step": 22117 }, { "epoch": 0.22118, "grad_norm": 1.3289814894533063, "learning_rate": 0.003, "loss": 4.0785, "step": 22118 }, { "epoch": 0.22119, "grad_norm": 1.2830149264163095, "learning_rate": 0.003, "loss": 4.056, "step": 22119 }, { "epoch": 0.2212, "grad_norm": 1.2561297299636018, "learning_rate": 0.003, "loss": 4.0663, "step": 22120 }, { "epoch": 0.22121, "grad_norm": 1.26843521352646, "learning_rate": 0.003, "loss": 4.0491, "step": 22121 }, { "epoch": 0.22122, "grad_norm": 1.4501851741246976, "learning_rate": 0.003, "loss": 4.0312, "step": 22122 }, { "epoch": 0.22123, "grad_norm": 1.067518429602658, "learning_rate": 0.003, "loss": 4.0184, "step": 22123 }, { "epoch": 0.22124, "grad_norm": 1.2479888401514971, "learning_rate": 0.003, "loss": 4.0411, "step": 22124 }, { "epoch": 0.22125, "grad_norm": 1.3969267245385515, "learning_rate": 0.003, "loss": 4.0542, "step": 22125 }, { "epoch": 0.22126, "grad_norm": 1.2327623436297128, "learning_rate": 0.003, "loss": 4.0654, "step": 22126 }, { "epoch": 0.22127, "grad_norm": 1.5941374199152674, "learning_rate": 0.003, "loss": 4.072, "step": 22127 }, { "epoch": 0.22128, "grad_norm": 1.2497619135028593, "learning_rate": 0.003, "loss": 4.0364, "step": 22128 }, { "epoch": 0.22129, "grad_norm": 1.3817912621007702, "learning_rate": 0.003, "loss": 4.0532, "step": 22129 }, { "epoch": 0.2213, "grad_norm": 1.228494123558605, "learning_rate": 0.003, "loss": 4.0367, "step": 22130 }, { "epoch": 0.22131, "grad_norm": 1.3183425194046234, "learning_rate": 0.003, "loss": 4.0209, "step": 22131 }, { "epoch": 0.22132, "grad_norm": 1.4042877045462423, "learning_rate": 0.003, "loss": 4.045, "step": 22132 }, { "epoch": 0.22133, "grad_norm": 1.2713861340476686, "learning_rate": 0.003, "loss": 4.0344, "step": 22133 }, { "epoch": 0.22134, "grad_norm": 1.264998720335586, "learning_rate": 0.003, "loss": 4.0379, "step": 22134 }, { "epoch": 0.22135, "grad_norm": 1.0702543077013797, "learning_rate": 0.003, "loss": 4.0225, "step": 22135 }, { "epoch": 0.22136, "grad_norm": 1.2781884544972073, "learning_rate": 0.003, "loss": 4.0544, "step": 22136 }, { "epoch": 0.22137, "grad_norm": 1.2852709215884592, "learning_rate": 0.003, "loss": 4.0473, "step": 22137 }, { "epoch": 0.22138, "grad_norm": 1.1005312711969961, "learning_rate": 0.003, "loss": 4.0323, "step": 22138 }, { "epoch": 0.22139, "grad_norm": 1.4759152117594718, "learning_rate": 0.003, "loss": 4.0423, "step": 22139 }, { "epoch": 0.2214, "grad_norm": 1.1324323721310299, "learning_rate": 0.003, "loss": 4.05, "step": 22140 }, { "epoch": 0.22141, "grad_norm": 1.6236663750577598, "learning_rate": 0.003, "loss": 4.0288, "step": 22141 }, { "epoch": 0.22142, "grad_norm": 1.1052368982921688, "learning_rate": 0.003, "loss": 4.0703, "step": 22142 }, { "epoch": 0.22143, "grad_norm": 1.5905769657683608, "learning_rate": 0.003, "loss": 4.069, "step": 22143 }, { "epoch": 0.22144, "grad_norm": 1.290551305451846, "learning_rate": 0.003, "loss": 4.0556, "step": 22144 }, { "epoch": 0.22145, "grad_norm": 1.6117881526943743, "learning_rate": 0.003, "loss": 4.0624, "step": 22145 }, { "epoch": 0.22146, "grad_norm": 1.1898382046669767, "learning_rate": 0.003, "loss": 4.0539, "step": 22146 }, { "epoch": 0.22147, "grad_norm": 1.2688902828105881, "learning_rate": 0.003, "loss": 4.0404, "step": 22147 }, { "epoch": 0.22148, "grad_norm": 1.4496171386489354, "learning_rate": 0.003, "loss": 4.05, "step": 22148 }, { "epoch": 0.22149, "grad_norm": 1.2586881822922122, "learning_rate": 0.003, "loss": 4.0535, "step": 22149 }, { "epoch": 0.2215, "grad_norm": 1.2137604380996958, "learning_rate": 0.003, "loss": 4.0141, "step": 22150 }, { "epoch": 0.22151, "grad_norm": 1.3586147786780414, "learning_rate": 0.003, "loss": 4.0434, "step": 22151 }, { "epoch": 0.22152, "grad_norm": 1.3470812827794485, "learning_rate": 0.003, "loss": 4.0859, "step": 22152 }, { "epoch": 0.22153, "grad_norm": 1.111077917048213, "learning_rate": 0.003, "loss": 4.0345, "step": 22153 }, { "epoch": 0.22154, "grad_norm": 1.2526395333727327, "learning_rate": 0.003, "loss": 4.0442, "step": 22154 }, { "epoch": 0.22155, "grad_norm": 1.2096794324801892, "learning_rate": 0.003, "loss": 4.0432, "step": 22155 }, { "epoch": 0.22156, "grad_norm": 1.4582326550309401, "learning_rate": 0.003, "loss": 4.0627, "step": 22156 }, { "epoch": 0.22157, "grad_norm": 1.409000980007499, "learning_rate": 0.003, "loss": 4.066, "step": 22157 }, { "epoch": 0.22158, "grad_norm": 1.3329337922666435, "learning_rate": 0.003, "loss": 4.0804, "step": 22158 }, { "epoch": 0.22159, "grad_norm": 1.365377698047997, "learning_rate": 0.003, "loss": 4.0642, "step": 22159 }, { "epoch": 0.2216, "grad_norm": 1.1750227017032777, "learning_rate": 0.003, "loss": 4.0506, "step": 22160 }, { "epoch": 0.22161, "grad_norm": 1.5052042818096398, "learning_rate": 0.003, "loss": 4.0495, "step": 22161 }, { "epoch": 0.22162, "grad_norm": 1.2857515138723257, "learning_rate": 0.003, "loss": 4.0539, "step": 22162 }, { "epoch": 0.22163, "grad_norm": 1.317674462309269, "learning_rate": 0.003, "loss": 4.0723, "step": 22163 }, { "epoch": 0.22164, "grad_norm": 1.4494465627418465, "learning_rate": 0.003, "loss": 4.0554, "step": 22164 }, { "epoch": 0.22165, "grad_norm": 1.1890524232637567, "learning_rate": 0.003, "loss": 4.0281, "step": 22165 }, { "epoch": 0.22166, "grad_norm": 1.245053588698817, "learning_rate": 0.003, "loss": 4.0307, "step": 22166 }, { "epoch": 0.22167, "grad_norm": 1.2127224222811872, "learning_rate": 0.003, "loss": 4.0358, "step": 22167 }, { "epoch": 0.22168, "grad_norm": 1.3495825122136302, "learning_rate": 0.003, "loss": 4.0279, "step": 22168 }, { "epoch": 0.22169, "grad_norm": 1.3198287182724973, "learning_rate": 0.003, "loss": 4.0527, "step": 22169 }, { "epoch": 0.2217, "grad_norm": 1.1676461875714192, "learning_rate": 0.003, "loss": 4.0586, "step": 22170 }, { "epoch": 0.22171, "grad_norm": 1.3226166710355396, "learning_rate": 0.003, "loss": 4.0666, "step": 22171 }, { "epoch": 0.22172, "grad_norm": 1.2403691143544944, "learning_rate": 0.003, "loss": 4.0492, "step": 22172 }, { "epoch": 0.22173, "grad_norm": 1.2596956337110792, "learning_rate": 0.003, "loss": 4.0366, "step": 22173 }, { "epoch": 0.22174, "grad_norm": 1.2764309524916242, "learning_rate": 0.003, "loss": 4.0567, "step": 22174 }, { "epoch": 0.22175, "grad_norm": 1.48595336118084, "learning_rate": 0.003, "loss": 4.0339, "step": 22175 }, { "epoch": 0.22176, "grad_norm": 1.2256372106366105, "learning_rate": 0.003, "loss": 4.0806, "step": 22176 }, { "epoch": 0.22177, "grad_norm": 1.3040329225333143, "learning_rate": 0.003, "loss": 4.078, "step": 22177 }, { "epoch": 0.22178, "grad_norm": 1.3292123033342786, "learning_rate": 0.003, "loss": 4.0719, "step": 22178 }, { "epoch": 0.22179, "grad_norm": 1.2179206276720262, "learning_rate": 0.003, "loss": 4.0493, "step": 22179 }, { "epoch": 0.2218, "grad_norm": 1.1196439625291523, "learning_rate": 0.003, "loss": 4.0473, "step": 22180 }, { "epoch": 0.22181, "grad_norm": 1.4174064312799082, "learning_rate": 0.003, "loss": 4.0507, "step": 22181 }, { "epoch": 0.22182, "grad_norm": 1.4953810749362755, "learning_rate": 0.003, "loss": 4.0476, "step": 22182 }, { "epoch": 0.22183, "grad_norm": 1.2424138147622161, "learning_rate": 0.003, "loss": 4.0377, "step": 22183 }, { "epoch": 0.22184, "grad_norm": 1.203758898472911, "learning_rate": 0.003, "loss": 4.0303, "step": 22184 }, { "epoch": 0.22185, "grad_norm": 1.2753021228370445, "learning_rate": 0.003, "loss": 4.0329, "step": 22185 }, { "epoch": 0.22186, "grad_norm": 1.5446446901416815, "learning_rate": 0.003, "loss": 4.054, "step": 22186 }, { "epoch": 0.22187, "grad_norm": 1.286948814251965, "learning_rate": 0.003, "loss": 4.0177, "step": 22187 }, { "epoch": 0.22188, "grad_norm": 1.3190321978728676, "learning_rate": 0.003, "loss": 4.0765, "step": 22188 }, { "epoch": 0.22189, "grad_norm": 1.382468196040536, "learning_rate": 0.003, "loss": 4.0297, "step": 22189 }, { "epoch": 0.2219, "grad_norm": 1.1252654706006864, "learning_rate": 0.003, "loss": 4.0446, "step": 22190 }, { "epoch": 0.22191, "grad_norm": 1.3471660412782678, "learning_rate": 0.003, "loss": 4.0252, "step": 22191 }, { "epoch": 0.22192, "grad_norm": 1.0705751076007068, "learning_rate": 0.003, "loss": 4.0314, "step": 22192 }, { "epoch": 0.22193, "grad_norm": 1.6493932185604472, "learning_rate": 0.003, "loss": 4.0582, "step": 22193 }, { "epoch": 0.22194, "grad_norm": 1.1736238809221848, "learning_rate": 0.003, "loss": 4.0625, "step": 22194 }, { "epoch": 0.22195, "grad_norm": 1.2157897684409515, "learning_rate": 0.003, "loss": 4.0212, "step": 22195 }, { "epoch": 0.22196, "grad_norm": 1.2794623068668234, "learning_rate": 0.003, "loss": 4.0728, "step": 22196 }, { "epoch": 0.22197, "grad_norm": 1.3300454281223202, "learning_rate": 0.003, "loss": 4.0446, "step": 22197 }, { "epoch": 0.22198, "grad_norm": 1.3339497746025875, "learning_rate": 0.003, "loss": 4.0436, "step": 22198 }, { "epoch": 0.22199, "grad_norm": 1.0741702760335805, "learning_rate": 0.003, "loss": 4.0332, "step": 22199 }, { "epoch": 0.222, "grad_norm": 1.3166946481347264, "learning_rate": 0.003, "loss": 4.0408, "step": 22200 }, { "epoch": 0.22201, "grad_norm": 1.3751429508002253, "learning_rate": 0.003, "loss": 4.076, "step": 22201 }, { "epoch": 0.22202, "grad_norm": 1.3557418090919628, "learning_rate": 0.003, "loss": 4.0432, "step": 22202 }, { "epoch": 0.22203, "grad_norm": 1.2601051449173593, "learning_rate": 0.003, "loss": 4.0515, "step": 22203 }, { "epoch": 0.22204, "grad_norm": 1.3010076869110623, "learning_rate": 0.003, "loss": 4.0304, "step": 22204 }, { "epoch": 0.22205, "grad_norm": 1.2402110537802256, "learning_rate": 0.003, "loss": 4.0623, "step": 22205 }, { "epoch": 0.22206, "grad_norm": 1.2812950776919276, "learning_rate": 0.003, "loss": 4.038, "step": 22206 }, { "epoch": 0.22207, "grad_norm": 1.387955277094656, "learning_rate": 0.003, "loss": 4.0654, "step": 22207 }, { "epoch": 0.22208, "grad_norm": 1.267722103524495, "learning_rate": 0.003, "loss": 4.0595, "step": 22208 }, { "epoch": 0.22209, "grad_norm": 1.3558599220332246, "learning_rate": 0.003, "loss": 4.0404, "step": 22209 }, { "epoch": 0.2221, "grad_norm": 1.4674635859547762, "learning_rate": 0.003, "loss": 4.05, "step": 22210 }, { "epoch": 0.22211, "grad_norm": 1.2803201729098241, "learning_rate": 0.003, "loss": 4.0558, "step": 22211 }, { "epoch": 0.22212, "grad_norm": 1.2872499174108878, "learning_rate": 0.003, "loss": 4.0454, "step": 22212 }, { "epoch": 0.22213, "grad_norm": 1.534661195185381, "learning_rate": 0.003, "loss": 4.0449, "step": 22213 }, { "epoch": 0.22214, "grad_norm": 1.206148335199384, "learning_rate": 0.003, "loss": 4.0503, "step": 22214 }, { "epoch": 0.22215, "grad_norm": 1.4653518529019343, "learning_rate": 0.003, "loss": 4.0478, "step": 22215 }, { "epoch": 0.22216, "grad_norm": 1.3758519702839136, "learning_rate": 0.003, "loss": 4.0562, "step": 22216 }, { "epoch": 0.22217, "grad_norm": 0.9965818671762745, "learning_rate": 0.003, "loss": 4.0553, "step": 22217 }, { "epoch": 0.22218, "grad_norm": 1.4390991131991475, "learning_rate": 0.003, "loss": 4.0628, "step": 22218 }, { "epoch": 0.22219, "grad_norm": 1.0430885363107065, "learning_rate": 0.003, "loss": 4.0175, "step": 22219 }, { "epoch": 0.2222, "grad_norm": 1.6980517978677148, "learning_rate": 0.003, "loss": 4.0576, "step": 22220 }, { "epoch": 0.22221, "grad_norm": 1.016124124849197, "learning_rate": 0.003, "loss": 4.0583, "step": 22221 }, { "epoch": 0.22222, "grad_norm": 1.4373105670424555, "learning_rate": 0.003, "loss": 4.0748, "step": 22222 }, { "epoch": 0.22223, "grad_norm": 1.2350703992552048, "learning_rate": 0.003, "loss": 4.0586, "step": 22223 }, { "epoch": 0.22224, "grad_norm": 1.1116514156313353, "learning_rate": 0.003, "loss": 4.0458, "step": 22224 }, { "epoch": 0.22225, "grad_norm": 1.1848877751606608, "learning_rate": 0.003, "loss": 4.0313, "step": 22225 }, { "epoch": 0.22226, "grad_norm": 1.2034576921297468, "learning_rate": 0.003, "loss": 4.0371, "step": 22226 }, { "epoch": 0.22227, "grad_norm": 1.4162655732776661, "learning_rate": 0.003, "loss": 4.0355, "step": 22227 }, { "epoch": 0.22228, "grad_norm": 1.133867009897121, "learning_rate": 0.003, "loss": 4.0566, "step": 22228 }, { "epoch": 0.22229, "grad_norm": 1.584053996594265, "learning_rate": 0.003, "loss": 4.0582, "step": 22229 }, { "epoch": 0.2223, "grad_norm": 1.293766270217981, "learning_rate": 0.003, "loss": 4.0359, "step": 22230 }, { "epoch": 0.22231, "grad_norm": 1.325285564953847, "learning_rate": 0.003, "loss": 4.0509, "step": 22231 }, { "epoch": 0.22232, "grad_norm": 1.2219685925488346, "learning_rate": 0.003, "loss": 4.0652, "step": 22232 }, { "epoch": 0.22233, "grad_norm": 1.2736953396915642, "learning_rate": 0.003, "loss": 4.0409, "step": 22233 }, { "epoch": 0.22234, "grad_norm": 1.4948282289064123, "learning_rate": 0.003, "loss": 4.0619, "step": 22234 }, { "epoch": 0.22235, "grad_norm": 1.4841260398697191, "learning_rate": 0.003, "loss": 4.0376, "step": 22235 }, { "epoch": 0.22236, "grad_norm": 1.3580554324399594, "learning_rate": 0.003, "loss": 4.0417, "step": 22236 }, { "epoch": 0.22237, "grad_norm": 1.2282339725710454, "learning_rate": 0.003, "loss": 4.0566, "step": 22237 }, { "epoch": 0.22238, "grad_norm": 1.3470326964917405, "learning_rate": 0.003, "loss": 4.0675, "step": 22238 }, { "epoch": 0.22239, "grad_norm": 1.4522168393080932, "learning_rate": 0.003, "loss": 4.0566, "step": 22239 }, { "epoch": 0.2224, "grad_norm": 1.1399714328058317, "learning_rate": 0.003, "loss": 4.0388, "step": 22240 }, { "epoch": 0.22241, "grad_norm": 1.4846788031412537, "learning_rate": 0.003, "loss": 4.0294, "step": 22241 }, { "epoch": 0.22242, "grad_norm": 1.050301085182913, "learning_rate": 0.003, "loss": 4.0218, "step": 22242 }, { "epoch": 0.22243, "grad_norm": 1.5609731667227993, "learning_rate": 0.003, "loss": 4.079, "step": 22243 }, { "epoch": 0.22244, "grad_norm": 1.1309386771706527, "learning_rate": 0.003, "loss": 4.035, "step": 22244 }, { "epoch": 0.22245, "grad_norm": 1.6179429538297576, "learning_rate": 0.003, "loss": 4.0441, "step": 22245 }, { "epoch": 0.22246, "grad_norm": 1.2934334521502595, "learning_rate": 0.003, "loss": 4.051, "step": 22246 }, { "epoch": 0.22247, "grad_norm": 1.4519038796624146, "learning_rate": 0.003, "loss": 4.067, "step": 22247 }, { "epoch": 0.22248, "grad_norm": 1.1833006425804269, "learning_rate": 0.003, "loss": 4.041, "step": 22248 }, { "epoch": 0.22249, "grad_norm": 1.3246829441844563, "learning_rate": 0.003, "loss": 4.0355, "step": 22249 }, { "epoch": 0.2225, "grad_norm": 1.0861351098106586, "learning_rate": 0.003, "loss": 4.0323, "step": 22250 }, { "epoch": 0.22251, "grad_norm": 1.877304766376981, "learning_rate": 0.003, "loss": 4.0432, "step": 22251 }, { "epoch": 0.22252, "grad_norm": 0.9286239406522893, "learning_rate": 0.003, "loss": 4.0427, "step": 22252 }, { "epoch": 0.22253, "grad_norm": 1.4307041713117088, "learning_rate": 0.003, "loss": 4.0269, "step": 22253 }, { "epoch": 0.22254, "grad_norm": 1.435639142730211, "learning_rate": 0.003, "loss": 4.0642, "step": 22254 }, { "epoch": 0.22255, "grad_norm": 1.5054286514642126, "learning_rate": 0.003, "loss": 4.082, "step": 22255 }, { "epoch": 0.22256, "grad_norm": 1.0504286982391589, "learning_rate": 0.003, "loss": 4.0541, "step": 22256 }, { "epoch": 0.22257, "grad_norm": 1.2221700419293187, "learning_rate": 0.003, "loss": 4.0667, "step": 22257 }, { "epoch": 0.22258, "grad_norm": 1.5719172508623824, "learning_rate": 0.003, "loss": 4.0448, "step": 22258 }, { "epoch": 0.22259, "grad_norm": 1.0656455147623263, "learning_rate": 0.003, "loss": 4.0496, "step": 22259 }, { "epoch": 0.2226, "grad_norm": 1.6173941270830197, "learning_rate": 0.003, "loss": 4.0622, "step": 22260 }, { "epoch": 0.22261, "grad_norm": 1.0519273454191997, "learning_rate": 0.003, "loss": 4.0688, "step": 22261 }, { "epoch": 0.22262, "grad_norm": 1.5023187322203926, "learning_rate": 0.003, "loss": 4.0634, "step": 22262 }, { "epoch": 0.22263, "grad_norm": 0.9452941415252466, "learning_rate": 0.003, "loss": 4.0682, "step": 22263 }, { "epoch": 0.22264, "grad_norm": 1.2330828125061868, "learning_rate": 0.003, "loss": 4.0679, "step": 22264 }, { "epoch": 0.22265, "grad_norm": 1.2745966699823645, "learning_rate": 0.003, "loss": 4.0859, "step": 22265 }, { "epoch": 0.22266, "grad_norm": 1.2830923093107793, "learning_rate": 0.003, "loss": 4.0557, "step": 22266 }, { "epoch": 0.22267, "grad_norm": 1.1763273594841448, "learning_rate": 0.003, "loss": 4.0234, "step": 22267 }, { "epoch": 0.22268, "grad_norm": 1.2877466306358871, "learning_rate": 0.003, "loss": 4.0357, "step": 22268 }, { "epoch": 0.22269, "grad_norm": 1.464868480327683, "learning_rate": 0.003, "loss": 4.0511, "step": 22269 }, { "epoch": 0.2227, "grad_norm": 1.0312615050265248, "learning_rate": 0.003, "loss": 4.0603, "step": 22270 }, { "epoch": 0.22271, "grad_norm": 1.4502149804268774, "learning_rate": 0.003, "loss": 4.0686, "step": 22271 }, { "epoch": 0.22272, "grad_norm": 1.155633312870541, "learning_rate": 0.003, "loss": 4.0424, "step": 22272 }, { "epoch": 0.22273, "grad_norm": 1.4255298658899878, "learning_rate": 0.003, "loss": 4.072, "step": 22273 }, { "epoch": 0.22274, "grad_norm": 1.4504068707265383, "learning_rate": 0.003, "loss": 4.0818, "step": 22274 }, { "epoch": 0.22275, "grad_norm": 1.3851491233174398, "learning_rate": 0.003, "loss": 4.0617, "step": 22275 }, { "epoch": 0.22276, "grad_norm": 1.2922839019261136, "learning_rate": 0.003, "loss": 4.0533, "step": 22276 }, { "epoch": 0.22277, "grad_norm": 1.4410317816557514, "learning_rate": 0.003, "loss": 4.0608, "step": 22277 }, { "epoch": 0.22278, "grad_norm": 1.336970797140703, "learning_rate": 0.003, "loss": 4.0407, "step": 22278 }, { "epoch": 0.22279, "grad_norm": 1.2350932710515, "learning_rate": 0.003, "loss": 4.0299, "step": 22279 }, { "epoch": 0.2228, "grad_norm": 1.0855331358867362, "learning_rate": 0.003, "loss": 4.0294, "step": 22280 }, { "epoch": 0.22281, "grad_norm": 1.452992882821028, "learning_rate": 0.003, "loss": 4.0682, "step": 22281 }, { "epoch": 0.22282, "grad_norm": 1.2090584854632997, "learning_rate": 0.003, "loss": 4.0568, "step": 22282 }, { "epoch": 0.22283, "grad_norm": 1.3079075769214261, "learning_rate": 0.003, "loss": 4.0397, "step": 22283 }, { "epoch": 0.22284, "grad_norm": 1.2317660787996143, "learning_rate": 0.003, "loss": 4.076, "step": 22284 }, { "epoch": 0.22285, "grad_norm": 1.2386508563305105, "learning_rate": 0.003, "loss": 4.0689, "step": 22285 }, { "epoch": 0.22286, "grad_norm": 1.2384648024035614, "learning_rate": 0.003, "loss": 4.0397, "step": 22286 }, { "epoch": 0.22287, "grad_norm": 1.7077043579283722, "learning_rate": 0.003, "loss": 4.0548, "step": 22287 }, { "epoch": 0.22288, "grad_norm": 1.1340765158806927, "learning_rate": 0.003, "loss": 4.0443, "step": 22288 }, { "epoch": 0.22289, "grad_norm": 1.2469906059671239, "learning_rate": 0.003, "loss": 4.0487, "step": 22289 }, { "epoch": 0.2229, "grad_norm": 1.2846156754206628, "learning_rate": 0.003, "loss": 4.0256, "step": 22290 }, { "epoch": 0.22291, "grad_norm": 1.0916856529044876, "learning_rate": 0.003, "loss": 4.0373, "step": 22291 }, { "epoch": 0.22292, "grad_norm": 1.6297288285047618, "learning_rate": 0.003, "loss": 4.035, "step": 22292 }, { "epoch": 0.22293, "grad_norm": 1.0844638564461677, "learning_rate": 0.003, "loss": 4.0465, "step": 22293 }, { "epoch": 0.22294, "grad_norm": 1.359361726134664, "learning_rate": 0.003, "loss": 4.0262, "step": 22294 }, { "epoch": 0.22295, "grad_norm": 1.249457508093861, "learning_rate": 0.003, "loss": 4.0635, "step": 22295 }, { "epoch": 0.22296, "grad_norm": 1.4601642774573411, "learning_rate": 0.003, "loss": 4.0487, "step": 22296 }, { "epoch": 0.22297, "grad_norm": 1.2309937735568728, "learning_rate": 0.003, "loss": 4.0339, "step": 22297 }, { "epoch": 0.22298, "grad_norm": 1.1975980985484633, "learning_rate": 0.003, "loss": 4.0706, "step": 22298 }, { "epoch": 0.22299, "grad_norm": 1.2023968725377303, "learning_rate": 0.003, "loss": 4.0559, "step": 22299 }, { "epoch": 0.223, "grad_norm": 1.276569396931763, "learning_rate": 0.003, "loss": 4.0628, "step": 22300 }, { "epoch": 0.22301, "grad_norm": 1.5279585850236914, "learning_rate": 0.003, "loss": 4.0215, "step": 22301 }, { "epoch": 0.22302, "grad_norm": 1.2117977118371657, "learning_rate": 0.003, "loss": 4.051, "step": 22302 }, { "epoch": 0.22303, "grad_norm": 1.2715957643290374, "learning_rate": 0.003, "loss": 3.9883, "step": 22303 }, { "epoch": 0.22304, "grad_norm": 1.1552053751556521, "learning_rate": 0.003, "loss": 4.038, "step": 22304 }, { "epoch": 0.22305, "grad_norm": 1.578301985762087, "learning_rate": 0.003, "loss": 4.0388, "step": 22305 }, { "epoch": 0.22306, "grad_norm": 1.1582304903209346, "learning_rate": 0.003, "loss": 4.0229, "step": 22306 }, { "epoch": 0.22307, "grad_norm": 1.2956039149251017, "learning_rate": 0.003, "loss": 4.0597, "step": 22307 }, { "epoch": 0.22308, "grad_norm": 1.2236909258652096, "learning_rate": 0.003, "loss": 4.0507, "step": 22308 }, { "epoch": 0.22309, "grad_norm": 1.2852634680259556, "learning_rate": 0.003, "loss": 4.025, "step": 22309 }, { "epoch": 0.2231, "grad_norm": 1.0868540714500894, "learning_rate": 0.003, "loss": 4.0443, "step": 22310 }, { "epoch": 0.22311, "grad_norm": 1.182212080244588, "learning_rate": 0.003, "loss": 4.0649, "step": 22311 }, { "epoch": 0.22312, "grad_norm": 1.3121510630486868, "learning_rate": 0.003, "loss": 4.0338, "step": 22312 }, { "epoch": 0.22313, "grad_norm": 1.0167860947785579, "learning_rate": 0.003, "loss": 4.0363, "step": 22313 }, { "epoch": 0.22314, "grad_norm": 1.4314189082804962, "learning_rate": 0.003, "loss": 4.0416, "step": 22314 }, { "epoch": 0.22315, "grad_norm": 1.0938782114272059, "learning_rate": 0.003, "loss": 4.0594, "step": 22315 }, { "epoch": 0.22316, "grad_norm": 1.4573906816415934, "learning_rate": 0.003, "loss": 4.067, "step": 22316 }, { "epoch": 0.22317, "grad_norm": 1.3304554290767985, "learning_rate": 0.003, "loss": 4.0608, "step": 22317 }, { "epoch": 0.22318, "grad_norm": 1.4129225666875442, "learning_rate": 0.003, "loss": 4.055, "step": 22318 }, { "epoch": 0.22319, "grad_norm": 1.6290619628830438, "learning_rate": 0.003, "loss": 4.0715, "step": 22319 }, { "epoch": 0.2232, "grad_norm": 1.2383722469772172, "learning_rate": 0.003, "loss": 4.063, "step": 22320 }, { "epoch": 0.22321, "grad_norm": 1.3661791730130857, "learning_rate": 0.003, "loss": 4.0697, "step": 22321 }, { "epoch": 0.22322, "grad_norm": 1.168435559990707, "learning_rate": 0.003, "loss": 4.0557, "step": 22322 }, { "epoch": 0.22323, "grad_norm": 1.412346641501458, "learning_rate": 0.003, "loss": 4.0702, "step": 22323 }, { "epoch": 0.22324, "grad_norm": 1.23847662842489, "learning_rate": 0.003, "loss": 4.0451, "step": 22324 }, { "epoch": 0.22325, "grad_norm": 1.13232533705196, "learning_rate": 0.003, "loss": 4.0393, "step": 22325 }, { "epoch": 0.22326, "grad_norm": 1.0809228457746634, "learning_rate": 0.003, "loss": 4.0473, "step": 22326 }, { "epoch": 0.22327, "grad_norm": 1.2593200103397935, "learning_rate": 0.003, "loss": 4.0318, "step": 22327 }, { "epoch": 0.22328, "grad_norm": 1.1832379244327795, "learning_rate": 0.003, "loss": 4.0662, "step": 22328 }, { "epoch": 0.22329, "grad_norm": 1.3634986343403839, "learning_rate": 0.003, "loss": 4.0514, "step": 22329 }, { "epoch": 0.2233, "grad_norm": 1.3001802022990911, "learning_rate": 0.003, "loss": 4.0662, "step": 22330 }, { "epoch": 0.22331, "grad_norm": 1.4709011222083932, "learning_rate": 0.003, "loss": 4.0516, "step": 22331 }, { "epoch": 0.22332, "grad_norm": 1.4709580366309234, "learning_rate": 0.003, "loss": 4.0406, "step": 22332 }, { "epoch": 0.22333, "grad_norm": 1.1163882905309481, "learning_rate": 0.003, "loss": 4.0392, "step": 22333 }, { "epoch": 0.22334, "grad_norm": 1.4893043270331274, "learning_rate": 0.003, "loss": 4.087, "step": 22334 }, { "epoch": 0.22335, "grad_norm": 0.9839768616454792, "learning_rate": 0.003, "loss": 4.0401, "step": 22335 }, { "epoch": 0.22336, "grad_norm": 1.5790218564132446, "learning_rate": 0.003, "loss": 4.0757, "step": 22336 }, { "epoch": 0.22337, "grad_norm": 1.1386007221622676, "learning_rate": 0.003, "loss": 4.0603, "step": 22337 }, { "epoch": 0.22338, "grad_norm": 1.4932045660837323, "learning_rate": 0.003, "loss": 4.058, "step": 22338 }, { "epoch": 0.22339, "grad_norm": 1.2818113626498306, "learning_rate": 0.003, "loss": 4.0479, "step": 22339 }, { "epoch": 0.2234, "grad_norm": 1.3015829509065822, "learning_rate": 0.003, "loss": 4.027, "step": 22340 }, { "epoch": 0.22341, "grad_norm": 1.0940565336446755, "learning_rate": 0.003, "loss": 4.018, "step": 22341 }, { "epoch": 0.22342, "grad_norm": 1.4679803696931557, "learning_rate": 0.003, "loss": 4.0256, "step": 22342 }, { "epoch": 0.22343, "grad_norm": 1.3606075206541697, "learning_rate": 0.003, "loss": 4.0594, "step": 22343 }, { "epoch": 0.22344, "grad_norm": 1.2753880982610875, "learning_rate": 0.003, "loss": 4.0487, "step": 22344 }, { "epoch": 0.22345, "grad_norm": 1.1863030220797137, "learning_rate": 0.003, "loss": 4.0409, "step": 22345 }, { "epoch": 0.22346, "grad_norm": 1.3796537921896148, "learning_rate": 0.003, "loss": 4.0737, "step": 22346 }, { "epoch": 0.22347, "grad_norm": 1.3850505212625757, "learning_rate": 0.003, "loss": 4.0325, "step": 22347 }, { "epoch": 0.22348, "grad_norm": 1.1212723584802724, "learning_rate": 0.003, "loss": 4.0477, "step": 22348 }, { "epoch": 0.22349, "grad_norm": 1.4953190800799308, "learning_rate": 0.003, "loss": 4.0434, "step": 22349 }, { "epoch": 0.2235, "grad_norm": 1.2159937041121087, "learning_rate": 0.003, "loss": 4.0523, "step": 22350 }, { "epoch": 0.22351, "grad_norm": 1.3013062519303582, "learning_rate": 0.003, "loss": 4.0723, "step": 22351 }, { "epoch": 0.22352, "grad_norm": 1.0749860878580382, "learning_rate": 0.003, "loss": 4.0611, "step": 22352 }, { "epoch": 0.22353, "grad_norm": 1.4021783584701677, "learning_rate": 0.003, "loss": 4.04, "step": 22353 }, { "epoch": 0.22354, "grad_norm": 1.3528154375174712, "learning_rate": 0.003, "loss": 4.0627, "step": 22354 }, { "epoch": 0.22355, "grad_norm": 1.3334532021261907, "learning_rate": 0.003, "loss": 4.0452, "step": 22355 }, { "epoch": 0.22356, "grad_norm": 1.4016203227488941, "learning_rate": 0.003, "loss": 4.0625, "step": 22356 }, { "epoch": 0.22357, "grad_norm": 1.1047671489127466, "learning_rate": 0.003, "loss": 4.0462, "step": 22357 }, { "epoch": 0.22358, "grad_norm": 1.2573296160077105, "learning_rate": 0.003, "loss": 4.0844, "step": 22358 }, { "epoch": 0.22359, "grad_norm": 1.4888231359476305, "learning_rate": 0.003, "loss": 4.0641, "step": 22359 }, { "epoch": 0.2236, "grad_norm": 1.2463718734796545, "learning_rate": 0.003, "loss": 4.0439, "step": 22360 }, { "epoch": 0.22361, "grad_norm": 1.404590659680424, "learning_rate": 0.003, "loss": 4.0506, "step": 22361 }, { "epoch": 0.22362, "grad_norm": 1.2813413033547938, "learning_rate": 0.003, "loss": 4.0417, "step": 22362 }, { "epoch": 0.22363, "grad_norm": 1.4883101044999885, "learning_rate": 0.003, "loss": 4.0247, "step": 22363 }, { "epoch": 0.22364, "grad_norm": 1.1796491598106018, "learning_rate": 0.003, "loss": 4.0322, "step": 22364 }, { "epoch": 0.22365, "grad_norm": 1.478516703458727, "learning_rate": 0.003, "loss": 4.0415, "step": 22365 }, { "epoch": 0.22366, "grad_norm": 1.0224712330143668, "learning_rate": 0.003, "loss": 4.0106, "step": 22366 }, { "epoch": 0.22367, "grad_norm": 1.3751666660674458, "learning_rate": 0.003, "loss": 4.0422, "step": 22367 }, { "epoch": 0.22368, "grad_norm": 1.135934042305452, "learning_rate": 0.003, "loss": 4.0638, "step": 22368 }, { "epoch": 0.22369, "grad_norm": 1.522880405573611, "learning_rate": 0.003, "loss": 4.0523, "step": 22369 }, { "epoch": 0.2237, "grad_norm": 1.3445328112617103, "learning_rate": 0.003, "loss": 4.0531, "step": 22370 }, { "epoch": 0.22371, "grad_norm": 1.418514430795147, "learning_rate": 0.003, "loss": 4.0606, "step": 22371 }, { "epoch": 0.22372, "grad_norm": 1.2207692837978814, "learning_rate": 0.003, "loss": 4.0657, "step": 22372 }, { "epoch": 0.22373, "grad_norm": 1.2769624870617733, "learning_rate": 0.003, "loss": 4.0671, "step": 22373 }, { "epoch": 0.22374, "grad_norm": 1.0708031813127818, "learning_rate": 0.003, "loss": 4.06, "step": 22374 }, { "epoch": 0.22375, "grad_norm": 1.3730281568778038, "learning_rate": 0.003, "loss": 4.0311, "step": 22375 }, { "epoch": 0.22376, "grad_norm": 1.0477212952218362, "learning_rate": 0.003, "loss": 4.0276, "step": 22376 }, { "epoch": 0.22377, "grad_norm": 1.7332241105349706, "learning_rate": 0.003, "loss": 4.011, "step": 22377 }, { "epoch": 0.22378, "grad_norm": 1.0521585596074527, "learning_rate": 0.003, "loss": 4.0417, "step": 22378 }, { "epoch": 0.22379, "grad_norm": 1.490163928963922, "learning_rate": 0.003, "loss": 4.0385, "step": 22379 }, { "epoch": 0.2238, "grad_norm": 1.1353874945847433, "learning_rate": 0.003, "loss": 4.055, "step": 22380 }, { "epoch": 0.22381, "grad_norm": 1.3622848418232558, "learning_rate": 0.003, "loss": 4.016, "step": 22381 }, { "epoch": 0.22382, "grad_norm": 1.4320302775215485, "learning_rate": 0.003, "loss": 4.0648, "step": 22382 }, { "epoch": 0.22383, "grad_norm": 1.3346976943076352, "learning_rate": 0.003, "loss": 4.0744, "step": 22383 }, { "epoch": 0.22384, "grad_norm": 1.3432305941947116, "learning_rate": 0.003, "loss": 4.0336, "step": 22384 }, { "epoch": 0.22385, "grad_norm": 1.116094994647951, "learning_rate": 0.003, "loss": 4.0538, "step": 22385 }, { "epoch": 0.22386, "grad_norm": 1.369139885417438, "learning_rate": 0.003, "loss": 4.0784, "step": 22386 }, { "epoch": 0.22387, "grad_norm": 1.2171968643952806, "learning_rate": 0.003, "loss": 4.0296, "step": 22387 }, { "epoch": 0.22388, "grad_norm": 1.5021210991763598, "learning_rate": 0.003, "loss": 4.081, "step": 22388 }, { "epoch": 0.22389, "grad_norm": 0.9165710159106715, "learning_rate": 0.003, "loss": 4.0397, "step": 22389 }, { "epoch": 0.2239, "grad_norm": 1.3538788722876058, "learning_rate": 0.003, "loss": 4.0628, "step": 22390 }, { "epoch": 0.22391, "grad_norm": 1.2920509847169448, "learning_rate": 0.003, "loss": 4.0416, "step": 22391 }, { "epoch": 0.22392, "grad_norm": 1.2823088339124191, "learning_rate": 0.003, "loss": 4.0365, "step": 22392 }, { "epoch": 0.22393, "grad_norm": 1.1256445035900986, "learning_rate": 0.003, "loss": 4.0376, "step": 22393 }, { "epoch": 0.22394, "grad_norm": 1.491436558545818, "learning_rate": 0.003, "loss": 4.0541, "step": 22394 }, { "epoch": 0.22395, "grad_norm": 1.3985742371270176, "learning_rate": 0.003, "loss": 4.075, "step": 22395 }, { "epoch": 0.22396, "grad_norm": 1.146971564066309, "learning_rate": 0.003, "loss": 4.0628, "step": 22396 }, { "epoch": 0.22397, "grad_norm": 1.2971131257252893, "learning_rate": 0.003, "loss": 4.0252, "step": 22397 }, { "epoch": 0.22398, "grad_norm": 1.3626163416376342, "learning_rate": 0.003, "loss": 4.0265, "step": 22398 }, { "epoch": 0.22399, "grad_norm": 1.1601142495852304, "learning_rate": 0.003, "loss": 4.0678, "step": 22399 }, { "epoch": 0.224, "grad_norm": 1.4788745770494174, "learning_rate": 0.003, "loss": 4.0498, "step": 22400 }, { "epoch": 0.22401, "grad_norm": 1.115497940758771, "learning_rate": 0.003, "loss": 4.0857, "step": 22401 }, { "epoch": 0.22402, "grad_norm": 1.3175398356922252, "learning_rate": 0.003, "loss": 4.0411, "step": 22402 }, { "epoch": 0.22403, "grad_norm": 1.3867457091875925, "learning_rate": 0.003, "loss": 4.0703, "step": 22403 }, { "epoch": 0.22404, "grad_norm": 1.2188372464295825, "learning_rate": 0.003, "loss": 4.0556, "step": 22404 }, { "epoch": 0.22405, "grad_norm": 1.6136079424523424, "learning_rate": 0.003, "loss": 4.0278, "step": 22405 }, { "epoch": 0.22406, "grad_norm": 1.2840510951647703, "learning_rate": 0.003, "loss": 4.0472, "step": 22406 }, { "epoch": 0.22407, "grad_norm": 1.1922017910090996, "learning_rate": 0.003, "loss": 4.0233, "step": 22407 }, { "epoch": 0.22408, "grad_norm": 1.2513253776576456, "learning_rate": 0.003, "loss": 4.0693, "step": 22408 }, { "epoch": 0.22409, "grad_norm": 0.9935214065796066, "learning_rate": 0.003, "loss": 4.0567, "step": 22409 }, { "epoch": 0.2241, "grad_norm": 1.311400695830719, "learning_rate": 0.003, "loss": 4.053, "step": 22410 }, { "epoch": 0.22411, "grad_norm": 1.176252463303334, "learning_rate": 0.003, "loss": 4.0435, "step": 22411 }, { "epoch": 0.22412, "grad_norm": 1.2844675406192732, "learning_rate": 0.003, "loss": 4.0287, "step": 22412 }, { "epoch": 0.22413, "grad_norm": 1.226472730943966, "learning_rate": 0.003, "loss": 4.0322, "step": 22413 }, { "epoch": 0.22414, "grad_norm": 1.4417331936987128, "learning_rate": 0.003, "loss": 4.0636, "step": 22414 }, { "epoch": 0.22415, "grad_norm": 1.4405129959311636, "learning_rate": 0.003, "loss": 4.0206, "step": 22415 }, { "epoch": 0.22416, "grad_norm": 1.2116370472850526, "learning_rate": 0.003, "loss": 4.0077, "step": 22416 }, { "epoch": 0.22417, "grad_norm": 1.467541063276362, "learning_rate": 0.003, "loss": 4.0588, "step": 22417 }, { "epoch": 0.22418, "grad_norm": 1.4105104108935589, "learning_rate": 0.003, "loss": 4.0506, "step": 22418 }, { "epoch": 0.22419, "grad_norm": 1.1739765243612241, "learning_rate": 0.003, "loss": 4.045, "step": 22419 }, { "epoch": 0.2242, "grad_norm": 1.4885789058986547, "learning_rate": 0.003, "loss": 4.068, "step": 22420 }, { "epoch": 0.22421, "grad_norm": 1.037911003556224, "learning_rate": 0.003, "loss": 4.0262, "step": 22421 }, { "epoch": 0.22422, "grad_norm": 1.3592019525985986, "learning_rate": 0.003, "loss": 4.0704, "step": 22422 }, { "epoch": 0.22423, "grad_norm": 1.3884009919816935, "learning_rate": 0.003, "loss": 4.0588, "step": 22423 }, { "epoch": 0.22424, "grad_norm": 1.518517321208513, "learning_rate": 0.003, "loss": 4.0189, "step": 22424 }, { "epoch": 0.22425, "grad_norm": 1.1687833287461964, "learning_rate": 0.003, "loss": 4.0333, "step": 22425 }, { "epoch": 0.22426, "grad_norm": 1.365890176260934, "learning_rate": 0.003, "loss": 4.0327, "step": 22426 }, { "epoch": 0.22427, "grad_norm": 1.133197490558877, "learning_rate": 0.003, "loss": 4.064, "step": 22427 }, { "epoch": 0.22428, "grad_norm": 1.5122539036707174, "learning_rate": 0.003, "loss": 4.0486, "step": 22428 }, { "epoch": 0.22429, "grad_norm": 1.0362710842338592, "learning_rate": 0.003, "loss": 4.0476, "step": 22429 }, { "epoch": 0.2243, "grad_norm": 1.584004024651577, "learning_rate": 0.003, "loss": 4.0383, "step": 22430 }, { "epoch": 0.22431, "grad_norm": 1.1423398695142792, "learning_rate": 0.003, "loss": 4.0402, "step": 22431 }, { "epoch": 0.22432, "grad_norm": 1.3993501986507162, "learning_rate": 0.003, "loss": 4.0426, "step": 22432 }, { "epoch": 0.22433, "grad_norm": 1.114000224834582, "learning_rate": 0.003, "loss": 4.0337, "step": 22433 }, { "epoch": 0.22434, "grad_norm": 1.318832448588106, "learning_rate": 0.003, "loss": 4.0465, "step": 22434 }, { "epoch": 0.22435, "grad_norm": 1.1834170443200742, "learning_rate": 0.003, "loss": 4.0772, "step": 22435 }, { "epoch": 0.22436, "grad_norm": 1.2640764447417074, "learning_rate": 0.003, "loss": 4.0446, "step": 22436 }, { "epoch": 0.22437, "grad_norm": 1.377324457859247, "learning_rate": 0.003, "loss": 4.0463, "step": 22437 }, { "epoch": 0.22438, "grad_norm": 1.158898418135193, "learning_rate": 0.003, "loss": 4.0267, "step": 22438 }, { "epoch": 0.22439, "grad_norm": 1.3749488508273955, "learning_rate": 0.003, "loss": 4.0771, "step": 22439 }, { "epoch": 0.2244, "grad_norm": 1.322779772421797, "learning_rate": 0.003, "loss": 4.071, "step": 22440 }, { "epoch": 0.22441, "grad_norm": 1.2278097001260544, "learning_rate": 0.003, "loss": 4.0562, "step": 22441 }, { "epoch": 0.22442, "grad_norm": 1.2261115415486579, "learning_rate": 0.003, "loss": 4.0654, "step": 22442 }, { "epoch": 0.22443, "grad_norm": 1.4746898038594523, "learning_rate": 0.003, "loss": 4.0322, "step": 22443 }, { "epoch": 0.22444, "grad_norm": 1.3500624604968665, "learning_rate": 0.003, "loss": 4.0152, "step": 22444 }, { "epoch": 0.22445, "grad_norm": 1.3502637523670458, "learning_rate": 0.003, "loss": 4.039, "step": 22445 }, { "epoch": 0.22446, "grad_norm": 1.2719783765555996, "learning_rate": 0.003, "loss": 4.0673, "step": 22446 }, { "epoch": 0.22447, "grad_norm": 1.3869575617555128, "learning_rate": 0.003, "loss": 4.0472, "step": 22447 }, { "epoch": 0.22448, "grad_norm": 1.225287527657583, "learning_rate": 0.003, "loss": 4.0499, "step": 22448 }, { "epoch": 0.22449, "grad_norm": 1.3169393455999998, "learning_rate": 0.003, "loss": 4.0334, "step": 22449 }, { "epoch": 0.2245, "grad_norm": 1.4657992689735533, "learning_rate": 0.003, "loss": 4.0651, "step": 22450 }, { "epoch": 0.22451, "grad_norm": 1.4944982135719134, "learning_rate": 0.003, "loss": 4.0406, "step": 22451 }, { "epoch": 0.22452, "grad_norm": 1.1613291719364822, "learning_rate": 0.003, "loss": 4.0561, "step": 22452 }, { "epoch": 0.22453, "grad_norm": 1.5470304026709796, "learning_rate": 0.003, "loss": 4.0646, "step": 22453 }, { "epoch": 0.22454, "grad_norm": 1.0105540330389218, "learning_rate": 0.003, "loss": 4.0479, "step": 22454 }, { "epoch": 0.22455, "grad_norm": 1.4572158881289734, "learning_rate": 0.003, "loss": 4.0178, "step": 22455 }, { "epoch": 0.22456, "grad_norm": 1.0906508336760523, "learning_rate": 0.003, "loss": 4.0441, "step": 22456 }, { "epoch": 0.22457, "grad_norm": 1.7741978830452934, "learning_rate": 0.003, "loss": 4.0735, "step": 22457 }, { "epoch": 0.22458, "grad_norm": 1.2198289454157725, "learning_rate": 0.003, "loss": 4.0611, "step": 22458 }, { "epoch": 0.22459, "grad_norm": 1.4733642872878125, "learning_rate": 0.003, "loss": 4.0367, "step": 22459 }, { "epoch": 0.2246, "grad_norm": 1.32696670690676, "learning_rate": 0.003, "loss": 4.0679, "step": 22460 }, { "epoch": 0.22461, "grad_norm": 1.1886104756898297, "learning_rate": 0.003, "loss": 4.0551, "step": 22461 }, { "epoch": 0.22462, "grad_norm": 1.2191357719141387, "learning_rate": 0.003, "loss": 4.051, "step": 22462 }, { "epoch": 0.22463, "grad_norm": 1.4188923527117434, "learning_rate": 0.003, "loss": 4.0668, "step": 22463 }, { "epoch": 0.22464, "grad_norm": 1.2805984183571502, "learning_rate": 0.003, "loss": 4.0794, "step": 22464 }, { "epoch": 0.22465, "grad_norm": 1.5753496965318274, "learning_rate": 0.003, "loss": 4.0784, "step": 22465 }, { "epoch": 0.22466, "grad_norm": 1.179880806487591, "learning_rate": 0.003, "loss": 4.0592, "step": 22466 }, { "epoch": 0.22467, "grad_norm": 1.3485754220347155, "learning_rate": 0.003, "loss": 4.0444, "step": 22467 }, { "epoch": 0.22468, "grad_norm": 1.2708371386466257, "learning_rate": 0.003, "loss": 4.0554, "step": 22468 }, { "epoch": 0.22469, "grad_norm": 1.1944117439714106, "learning_rate": 0.003, "loss": 4.0824, "step": 22469 }, { "epoch": 0.2247, "grad_norm": 1.2766852560716921, "learning_rate": 0.003, "loss": 4.0522, "step": 22470 }, { "epoch": 0.22471, "grad_norm": 1.7035429052101791, "learning_rate": 0.003, "loss": 4.0708, "step": 22471 }, { "epoch": 0.22472, "grad_norm": 1.0915261183000333, "learning_rate": 0.003, "loss": 4.0835, "step": 22472 }, { "epoch": 0.22473, "grad_norm": 1.4883615868165714, "learning_rate": 0.003, "loss": 4.047, "step": 22473 }, { "epoch": 0.22474, "grad_norm": 1.0578550955517174, "learning_rate": 0.003, "loss": 4.0484, "step": 22474 }, { "epoch": 0.22475, "grad_norm": 1.2875892117294654, "learning_rate": 0.003, "loss": 4.0388, "step": 22475 }, { "epoch": 0.22476, "grad_norm": 1.2169411525097316, "learning_rate": 0.003, "loss": 4.0676, "step": 22476 }, { "epoch": 0.22477, "grad_norm": 1.4483348325305532, "learning_rate": 0.003, "loss": 4.0295, "step": 22477 }, { "epoch": 0.22478, "grad_norm": 1.2229456460115564, "learning_rate": 0.003, "loss": 4.0604, "step": 22478 }, { "epoch": 0.22479, "grad_norm": 1.1979696729402194, "learning_rate": 0.003, "loss": 4.054, "step": 22479 }, { "epoch": 0.2248, "grad_norm": 1.3704086152133736, "learning_rate": 0.003, "loss": 4.0621, "step": 22480 }, { "epoch": 0.22481, "grad_norm": 1.2069377418342675, "learning_rate": 0.003, "loss": 4.0591, "step": 22481 }, { "epoch": 0.22482, "grad_norm": 1.3379856851864262, "learning_rate": 0.003, "loss": 4.0391, "step": 22482 }, { "epoch": 0.22483, "grad_norm": 1.1664769506298684, "learning_rate": 0.003, "loss": 4.0552, "step": 22483 }, { "epoch": 0.22484, "grad_norm": 1.5041251557210558, "learning_rate": 0.003, "loss": 4.0617, "step": 22484 }, { "epoch": 0.22485, "grad_norm": 1.0180817941224296, "learning_rate": 0.003, "loss": 4.0263, "step": 22485 }, { "epoch": 0.22486, "grad_norm": 1.6643621587920203, "learning_rate": 0.003, "loss": 4.0869, "step": 22486 }, { "epoch": 0.22487, "grad_norm": 1.1126847126724073, "learning_rate": 0.003, "loss": 4.0662, "step": 22487 }, { "epoch": 0.22488, "grad_norm": 1.3776327714744627, "learning_rate": 0.003, "loss": 4.0771, "step": 22488 }, { "epoch": 0.22489, "grad_norm": 1.434849507538738, "learning_rate": 0.003, "loss": 4.0605, "step": 22489 }, { "epoch": 0.2249, "grad_norm": 1.3887483141725754, "learning_rate": 0.003, "loss": 4.0576, "step": 22490 }, { "epoch": 0.22491, "grad_norm": 1.1684750215251973, "learning_rate": 0.003, "loss": 4.0459, "step": 22491 }, { "epoch": 0.22492, "grad_norm": 1.1722986701929685, "learning_rate": 0.003, "loss": 4.0574, "step": 22492 }, { "epoch": 0.22493, "grad_norm": 1.4917275414576998, "learning_rate": 0.003, "loss": 4.0167, "step": 22493 }, { "epoch": 0.22494, "grad_norm": 1.3392680617555675, "learning_rate": 0.003, "loss": 4.0487, "step": 22494 }, { "epoch": 0.22495, "grad_norm": 1.3363684423707247, "learning_rate": 0.003, "loss": 4.0732, "step": 22495 }, { "epoch": 0.22496, "grad_norm": 1.1204296554549116, "learning_rate": 0.003, "loss": 4.0806, "step": 22496 }, { "epoch": 0.22497, "grad_norm": 1.4715952734942133, "learning_rate": 0.003, "loss": 4.0594, "step": 22497 }, { "epoch": 0.22498, "grad_norm": 1.1286796848027008, "learning_rate": 0.003, "loss": 4.0376, "step": 22498 }, { "epoch": 0.22499, "grad_norm": 1.372096531276422, "learning_rate": 0.003, "loss": 4.0347, "step": 22499 }, { "epoch": 0.225, "grad_norm": 1.1486641500096484, "learning_rate": 0.003, "loss": 4.0413, "step": 22500 }, { "epoch": 0.22501, "grad_norm": 1.3723721393832513, "learning_rate": 0.003, "loss": 4.0398, "step": 22501 }, { "epoch": 0.22502, "grad_norm": 1.3046545766839284, "learning_rate": 0.003, "loss": 4.06, "step": 22502 }, { "epoch": 0.22503, "grad_norm": 1.3284179722379172, "learning_rate": 0.003, "loss": 4.0674, "step": 22503 }, { "epoch": 0.22504, "grad_norm": 1.1119344957399873, "learning_rate": 0.003, "loss": 4.0533, "step": 22504 }, { "epoch": 0.22505, "grad_norm": 1.578456863799798, "learning_rate": 0.003, "loss": 4.0636, "step": 22505 }, { "epoch": 0.22506, "grad_norm": 1.2077848169170577, "learning_rate": 0.003, "loss": 4.0668, "step": 22506 }, { "epoch": 0.22507, "grad_norm": 1.2051758062685873, "learning_rate": 0.003, "loss": 4.0543, "step": 22507 }, { "epoch": 0.22508, "grad_norm": 1.280602849013952, "learning_rate": 0.003, "loss": 4.0395, "step": 22508 }, { "epoch": 0.22509, "grad_norm": 1.352851133581328, "learning_rate": 0.003, "loss": 4.0613, "step": 22509 }, { "epoch": 0.2251, "grad_norm": 1.0192495616969712, "learning_rate": 0.003, "loss": 4.0502, "step": 22510 }, { "epoch": 0.22511, "grad_norm": 1.5177906883465204, "learning_rate": 0.003, "loss": 4.0276, "step": 22511 }, { "epoch": 0.22512, "grad_norm": 1.1457818258301886, "learning_rate": 0.003, "loss": 4.0578, "step": 22512 }, { "epoch": 0.22513, "grad_norm": 1.3491140413787326, "learning_rate": 0.003, "loss": 4.042, "step": 22513 }, { "epoch": 0.22514, "grad_norm": 0.99994817126937, "learning_rate": 0.003, "loss": 4.0088, "step": 22514 }, { "epoch": 0.22515, "grad_norm": 1.414576842473423, "learning_rate": 0.003, "loss": 4.0553, "step": 22515 }, { "epoch": 0.22516, "grad_norm": 1.0583718157620126, "learning_rate": 0.003, "loss": 4.071, "step": 22516 }, { "epoch": 0.22517, "grad_norm": 1.51180811624445, "learning_rate": 0.003, "loss": 4.064, "step": 22517 }, { "epoch": 0.22518, "grad_norm": 1.0395624435698443, "learning_rate": 0.003, "loss": 4.0449, "step": 22518 }, { "epoch": 0.22519, "grad_norm": 1.4144988736342254, "learning_rate": 0.003, "loss": 4.0538, "step": 22519 }, { "epoch": 0.2252, "grad_norm": 1.4898113967313333, "learning_rate": 0.003, "loss": 4.0279, "step": 22520 }, { "epoch": 0.22521, "grad_norm": 1.2624273689852192, "learning_rate": 0.003, "loss": 4.0451, "step": 22521 }, { "epoch": 0.22522, "grad_norm": 1.4067702398160626, "learning_rate": 0.003, "loss": 4.0496, "step": 22522 }, { "epoch": 0.22523, "grad_norm": 1.1773782057120372, "learning_rate": 0.003, "loss": 4.0537, "step": 22523 }, { "epoch": 0.22524, "grad_norm": 1.3744929248938713, "learning_rate": 0.003, "loss": 4.0477, "step": 22524 }, { "epoch": 0.22525, "grad_norm": 1.036773671963765, "learning_rate": 0.003, "loss": 4.0382, "step": 22525 }, { "epoch": 0.22526, "grad_norm": 1.3639080767800937, "learning_rate": 0.003, "loss": 4.0717, "step": 22526 }, { "epoch": 0.22527, "grad_norm": 1.252612508165742, "learning_rate": 0.003, "loss": 4.0676, "step": 22527 }, { "epoch": 0.22528, "grad_norm": 1.4278657114127173, "learning_rate": 0.003, "loss": 4.0643, "step": 22528 }, { "epoch": 0.22529, "grad_norm": 0.9999350331633271, "learning_rate": 0.003, "loss": 4.0404, "step": 22529 }, { "epoch": 0.2253, "grad_norm": 1.2817891196627766, "learning_rate": 0.003, "loss": 4.0399, "step": 22530 }, { "epoch": 0.22531, "grad_norm": 1.4696249469244116, "learning_rate": 0.003, "loss": 4.0751, "step": 22531 }, { "epoch": 0.22532, "grad_norm": 1.302138195690749, "learning_rate": 0.003, "loss": 4.0477, "step": 22532 }, { "epoch": 0.22533, "grad_norm": 1.3981459768298665, "learning_rate": 0.003, "loss": 4.0202, "step": 22533 }, { "epoch": 0.22534, "grad_norm": 1.3194697626184033, "learning_rate": 0.003, "loss": 4.0267, "step": 22534 }, { "epoch": 0.22535, "grad_norm": 1.2430679840444745, "learning_rate": 0.003, "loss": 4.0601, "step": 22535 }, { "epoch": 0.22536, "grad_norm": 1.2846080870555665, "learning_rate": 0.003, "loss": 4.0569, "step": 22536 }, { "epoch": 0.22537, "grad_norm": 1.1769136551752915, "learning_rate": 0.003, "loss": 4.0522, "step": 22537 }, { "epoch": 0.22538, "grad_norm": 1.2321372670348028, "learning_rate": 0.003, "loss": 4.0691, "step": 22538 }, { "epoch": 0.22539, "grad_norm": 1.5042893783921505, "learning_rate": 0.003, "loss": 4.0299, "step": 22539 }, { "epoch": 0.2254, "grad_norm": 1.124989172031265, "learning_rate": 0.003, "loss": 4.032, "step": 22540 }, { "epoch": 0.22541, "grad_norm": 1.5375434287648648, "learning_rate": 0.003, "loss": 4.0644, "step": 22541 }, { "epoch": 0.22542, "grad_norm": 1.1460432617510374, "learning_rate": 0.003, "loss": 4.0494, "step": 22542 }, { "epoch": 0.22543, "grad_norm": 1.229328340830702, "learning_rate": 0.003, "loss": 4.0568, "step": 22543 }, { "epoch": 0.22544, "grad_norm": 1.2005061825366072, "learning_rate": 0.003, "loss": 4.0269, "step": 22544 }, { "epoch": 0.22545, "grad_norm": 1.2614278890985622, "learning_rate": 0.003, "loss": 4.0581, "step": 22545 }, { "epoch": 0.22546, "grad_norm": 1.3068986510878424, "learning_rate": 0.003, "loss": 4.0182, "step": 22546 }, { "epoch": 0.22547, "grad_norm": 1.3270799932059623, "learning_rate": 0.003, "loss": 4.0426, "step": 22547 }, { "epoch": 0.22548, "grad_norm": 1.6015301694475044, "learning_rate": 0.003, "loss": 4.0539, "step": 22548 }, { "epoch": 0.22549, "grad_norm": 0.9898988981525878, "learning_rate": 0.003, "loss": 4.0251, "step": 22549 }, { "epoch": 0.2255, "grad_norm": 1.5398743726609867, "learning_rate": 0.003, "loss": 4.058, "step": 22550 }, { "epoch": 0.22551, "grad_norm": 1.0979781519615899, "learning_rate": 0.003, "loss": 4.0529, "step": 22551 }, { "epoch": 0.22552, "grad_norm": 1.2099441771834618, "learning_rate": 0.003, "loss": 4.0493, "step": 22552 }, { "epoch": 0.22553, "grad_norm": 1.3489680531585568, "learning_rate": 0.003, "loss": 4.0313, "step": 22553 }, { "epoch": 0.22554, "grad_norm": 1.324302858070193, "learning_rate": 0.003, "loss": 4.072, "step": 22554 }, { "epoch": 0.22555, "grad_norm": 1.2575044614820337, "learning_rate": 0.003, "loss": 4.0604, "step": 22555 }, { "epoch": 0.22556, "grad_norm": 1.2626124900806674, "learning_rate": 0.003, "loss": 4.0519, "step": 22556 }, { "epoch": 0.22557, "grad_norm": 1.2154527676652886, "learning_rate": 0.003, "loss": 4.0428, "step": 22557 }, { "epoch": 0.22558, "grad_norm": 1.5566588781201214, "learning_rate": 0.003, "loss": 4.0644, "step": 22558 }, { "epoch": 0.22559, "grad_norm": 1.2210561493024892, "learning_rate": 0.003, "loss": 4.0376, "step": 22559 }, { "epoch": 0.2256, "grad_norm": 1.4062844461091686, "learning_rate": 0.003, "loss": 4.0764, "step": 22560 }, { "epoch": 0.22561, "grad_norm": 1.1072881788800877, "learning_rate": 0.003, "loss": 4.0717, "step": 22561 }, { "epoch": 0.22562, "grad_norm": 1.697164349991573, "learning_rate": 0.003, "loss": 4.0784, "step": 22562 }, { "epoch": 0.22563, "grad_norm": 1.0823401616660748, "learning_rate": 0.003, "loss": 4.0371, "step": 22563 }, { "epoch": 0.22564, "grad_norm": 1.4578122990061382, "learning_rate": 0.003, "loss": 4.0372, "step": 22564 }, { "epoch": 0.22565, "grad_norm": 1.0274068168740182, "learning_rate": 0.003, "loss": 4.0745, "step": 22565 }, { "epoch": 0.22566, "grad_norm": 1.3415441629728502, "learning_rate": 0.003, "loss": 4.0461, "step": 22566 }, { "epoch": 0.22567, "grad_norm": 1.1806346198089603, "learning_rate": 0.003, "loss": 4.0626, "step": 22567 }, { "epoch": 0.22568, "grad_norm": 1.4665297214099582, "learning_rate": 0.003, "loss": 4.0342, "step": 22568 }, { "epoch": 0.22569, "grad_norm": 1.0734620731968774, "learning_rate": 0.003, "loss": 4.0358, "step": 22569 }, { "epoch": 0.2257, "grad_norm": 1.5077966401944736, "learning_rate": 0.003, "loss": 4.0215, "step": 22570 }, { "epoch": 0.22571, "grad_norm": 1.2047552478719217, "learning_rate": 0.003, "loss": 4.0522, "step": 22571 }, { "epoch": 0.22572, "grad_norm": 1.525018125012932, "learning_rate": 0.003, "loss": 4.0734, "step": 22572 }, { "epoch": 0.22573, "grad_norm": 1.0652457912081579, "learning_rate": 0.003, "loss": 4.0397, "step": 22573 }, { "epoch": 0.22574, "grad_norm": 1.2114860422416958, "learning_rate": 0.003, "loss": 4.0568, "step": 22574 }, { "epoch": 0.22575, "grad_norm": 1.3818053420657728, "learning_rate": 0.003, "loss": 4.0688, "step": 22575 }, { "epoch": 0.22576, "grad_norm": 1.0639394818485324, "learning_rate": 0.003, "loss": 4.0268, "step": 22576 }, { "epoch": 0.22577, "grad_norm": 1.3868765056606982, "learning_rate": 0.003, "loss": 4.0533, "step": 22577 }, { "epoch": 0.22578, "grad_norm": 1.520924933073174, "learning_rate": 0.003, "loss": 4.0615, "step": 22578 }, { "epoch": 0.22579, "grad_norm": 1.26000523465386, "learning_rate": 0.003, "loss": 4.0539, "step": 22579 }, { "epoch": 0.2258, "grad_norm": 1.3881665919508095, "learning_rate": 0.003, "loss": 4.032, "step": 22580 }, { "epoch": 0.22581, "grad_norm": 1.2315326602596626, "learning_rate": 0.003, "loss": 4.0377, "step": 22581 }, { "epoch": 0.22582, "grad_norm": 1.1901480425516753, "learning_rate": 0.003, "loss": 4.0527, "step": 22582 }, { "epoch": 0.22583, "grad_norm": 1.130467926058565, "learning_rate": 0.003, "loss": 4.0615, "step": 22583 }, { "epoch": 0.22584, "grad_norm": 1.406528403899872, "learning_rate": 0.003, "loss": 4.0736, "step": 22584 }, { "epoch": 0.22585, "grad_norm": 1.542422584892502, "learning_rate": 0.003, "loss": 4.082, "step": 22585 }, { "epoch": 0.22586, "grad_norm": 1.593864830777442, "learning_rate": 0.003, "loss": 4.0698, "step": 22586 }, { "epoch": 0.22587, "grad_norm": 1.3854399345300927, "learning_rate": 0.003, "loss": 4.0644, "step": 22587 }, { "epoch": 0.22588, "grad_norm": 1.2856110491107158, "learning_rate": 0.003, "loss": 4.0609, "step": 22588 }, { "epoch": 0.22589, "grad_norm": 1.2276983551534677, "learning_rate": 0.003, "loss": 4.0348, "step": 22589 }, { "epoch": 0.2259, "grad_norm": 1.4277051343967955, "learning_rate": 0.003, "loss": 4.0369, "step": 22590 }, { "epoch": 0.22591, "grad_norm": 1.1407240152746587, "learning_rate": 0.003, "loss": 4.0539, "step": 22591 }, { "epoch": 0.22592, "grad_norm": 1.2619386720652488, "learning_rate": 0.003, "loss": 4.0722, "step": 22592 }, { "epoch": 0.22593, "grad_norm": 1.313989982492832, "learning_rate": 0.003, "loss": 4.0277, "step": 22593 }, { "epoch": 0.22594, "grad_norm": 1.1822401340355067, "learning_rate": 0.003, "loss": 4.0385, "step": 22594 }, { "epoch": 0.22595, "grad_norm": 1.278760915465666, "learning_rate": 0.003, "loss": 4.0351, "step": 22595 }, { "epoch": 0.22596, "grad_norm": 1.2367738374208836, "learning_rate": 0.003, "loss": 4.0433, "step": 22596 }, { "epoch": 0.22597, "grad_norm": 1.300515051860066, "learning_rate": 0.003, "loss": 4.0484, "step": 22597 }, { "epoch": 0.22598, "grad_norm": 1.2797126858848549, "learning_rate": 0.003, "loss": 4.0571, "step": 22598 }, { "epoch": 0.22599, "grad_norm": 1.308836644674331, "learning_rate": 0.003, "loss": 4.034, "step": 22599 }, { "epoch": 0.226, "grad_norm": 1.1671181400459543, "learning_rate": 0.003, "loss": 4.0517, "step": 22600 }, { "epoch": 0.22601, "grad_norm": 1.3042428869441687, "learning_rate": 0.003, "loss": 4.0464, "step": 22601 }, { "epoch": 0.22602, "grad_norm": 1.2135366414762496, "learning_rate": 0.003, "loss": 4.0439, "step": 22602 }, { "epoch": 0.22603, "grad_norm": 1.26134826915533, "learning_rate": 0.003, "loss": 4.0288, "step": 22603 }, { "epoch": 0.22604, "grad_norm": 1.2354811227773834, "learning_rate": 0.003, "loss": 4.0539, "step": 22604 }, { "epoch": 0.22605, "grad_norm": 1.2536625176725607, "learning_rate": 0.003, "loss": 4.0463, "step": 22605 }, { "epoch": 0.22606, "grad_norm": 1.313000819827062, "learning_rate": 0.003, "loss": 4.0727, "step": 22606 }, { "epoch": 0.22607, "grad_norm": 1.4713753017199727, "learning_rate": 0.003, "loss": 4.0574, "step": 22607 }, { "epoch": 0.22608, "grad_norm": 1.3520838679471667, "learning_rate": 0.003, "loss": 4.0571, "step": 22608 }, { "epoch": 0.22609, "grad_norm": 1.5455726355290877, "learning_rate": 0.003, "loss": 4.0822, "step": 22609 }, { "epoch": 0.2261, "grad_norm": 1.2182019153494124, "learning_rate": 0.003, "loss": 4.0362, "step": 22610 }, { "epoch": 0.22611, "grad_norm": 1.3561505169364962, "learning_rate": 0.003, "loss": 4.0114, "step": 22611 }, { "epoch": 0.22612, "grad_norm": 1.3133920050258785, "learning_rate": 0.003, "loss": 4.0464, "step": 22612 }, { "epoch": 0.22613, "grad_norm": 1.312449719643275, "learning_rate": 0.003, "loss": 4.0797, "step": 22613 }, { "epoch": 0.22614, "grad_norm": 1.074657696380918, "learning_rate": 0.003, "loss": 4.0479, "step": 22614 }, { "epoch": 0.22615, "grad_norm": 1.442241840495215, "learning_rate": 0.003, "loss": 4.0437, "step": 22615 }, { "epoch": 0.22616, "grad_norm": 1.0601450853962522, "learning_rate": 0.003, "loss": 4.0257, "step": 22616 }, { "epoch": 0.22617, "grad_norm": 1.5620336776709003, "learning_rate": 0.003, "loss": 4.0346, "step": 22617 }, { "epoch": 0.22618, "grad_norm": 0.9689079372536944, "learning_rate": 0.003, "loss": 4.0728, "step": 22618 }, { "epoch": 0.22619, "grad_norm": 1.6808687314107054, "learning_rate": 0.003, "loss": 4.0621, "step": 22619 }, { "epoch": 0.2262, "grad_norm": 1.1640894040723138, "learning_rate": 0.003, "loss": 4.0483, "step": 22620 }, { "epoch": 0.22621, "grad_norm": 1.344668278352046, "learning_rate": 0.003, "loss": 4.0433, "step": 22621 }, { "epoch": 0.22622, "grad_norm": 1.44299896409508, "learning_rate": 0.003, "loss": 4.0668, "step": 22622 }, { "epoch": 0.22623, "grad_norm": 1.3721375096943011, "learning_rate": 0.003, "loss": 4.0368, "step": 22623 }, { "epoch": 0.22624, "grad_norm": 1.1522684824387952, "learning_rate": 0.003, "loss": 4.043, "step": 22624 }, { "epoch": 0.22625, "grad_norm": 1.2260741731087532, "learning_rate": 0.003, "loss": 4.0515, "step": 22625 }, { "epoch": 0.22626, "grad_norm": 1.3070977190046338, "learning_rate": 0.003, "loss": 4.0419, "step": 22626 }, { "epoch": 0.22627, "grad_norm": 1.2589188365447337, "learning_rate": 0.003, "loss": 4.046, "step": 22627 }, { "epoch": 0.22628, "grad_norm": 1.3944764788183477, "learning_rate": 0.003, "loss": 4.0769, "step": 22628 }, { "epoch": 0.22629, "grad_norm": 1.1662320467262188, "learning_rate": 0.003, "loss": 4.0586, "step": 22629 }, { "epoch": 0.2263, "grad_norm": 1.3279237232083825, "learning_rate": 0.003, "loss": 4.035, "step": 22630 }, { "epoch": 0.22631, "grad_norm": 1.4282357337112004, "learning_rate": 0.003, "loss": 4.0678, "step": 22631 }, { "epoch": 0.22632, "grad_norm": 1.2661735843623207, "learning_rate": 0.003, "loss": 4.0377, "step": 22632 }, { "epoch": 0.22633, "grad_norm": 1.6257038340349381, "learning_rate": 0.003, "loss": 4.0783, "step": 22633 }, { "epoch": 0.22634, "grad_norm": 0.9648899165313575, "learning_rate": 0.003, "loss": 4.0344, "step": 22634 }, { "epoch": 0.22635, "grad_norm": 1.371901664366491, "learning_rate": 0.003, "loss": 4.0348, "step": 22635 }, { "epoch": 0.22636, "grad_norm": 1.0438516546415813, "learning_rate": 0.003, "loss": 4.0445, "step": 22636 }, { "epoch": 0.22637, "grad_norm": 1.4310376589392284, "learning_rate": 0.003, "loss": 4.0615, "step": 22637 }, { "epoch": 0.22638, "grad_norm": 1.1567361263821145, "learning_rate": 0.003, "loss": 4.0666, "step": 22638 }, { "epoch": 0.22639, "grad_norm": 1.2405414390662812, "learning_rate": 0.003, "loss": 4.0498, "step": 22639 }, { "epoch": 0.2264, "grad_norm": 1.3410653625276123, "learning_rate": 0.003, "loss": 4.0581, "step": 22640 }, { "epoch": 0.22641, "grad_norm": 1.4188255889704002, "learning_rate": 0.003, "loss": 4.0608, "step": 22641 }, { "epoch": 0.22642, "grad_norm": 1.3474829338391447, "learning_rate": 0.003, "loss": 4.0445, "step": 22642 }, { "epoch": 0.22643, "grad_norm": 1.2841709957180494, "learning_rate": 0.003, "loss": 4.0681, "step": 22643 }, { "epoch": 0.22644, "grad_norm": 1.2520534260697076, "learning_rate": 0.003, "loss": 4.0353, "step": 22644 }, { "epoch": 0.22645, "grad_norm": 1.5517767716720081, "learning_rate": 0.003, "loss": 4.0418, "step": 22645 }, { "epoch": 0.22646, "grad_norm": 1.0330104694980737, "learning_rate": 0.003, "loss": 4.076, "step": 22646 }, { "epoch": 0.22647, "grad_norm": 1.4845952140019196, "learning_rate": 0.003, "loss": 4.0139, "step": 22647 }, { "epoch": 0.22648, "grad_norm": 1.1610074054596695, "learning_rate": 0.003, "loss": 4.084, "step": 22648 }, { "epoch": 0.22649, "grad_norm": 1.3770728628981728, "learning_rate": 0.003, "loss": 4.0315, "step": 22649 }, { "epoch": 0.2265, "grad_norm": 1.1936327852921174, "learning_rate": 0.003, "loss": 4.017, "step": 22650 }, { "epoch": 0.22651, "grad_norm": 1.331939898988897, "learning_rate": 0.003, "loss": 4.0425, "step": 22651 }, { "epoch": 0.22652, "grad_norm": 1.1178861395283695, "learning_rate": 0.003, "loss": 4.0571, "step": 22652 }, { "epoch": 0.22653, "grad_norm": 1.354185330857788, "learning_rate": 0.003, "loss": 4.0546, "step": 22653 }, { "epoch": 0.22654, "grad_norm": 1.4125164056890323, "learning_rate": 0.003, "loss": 4.0347, "step": 22654 }, { "epoch": 0.22655, "grad_norm": 1.232991593856581, "learning_rate": 0.003, "loss": 4.0422, "step": 22655 }, { "epoch": 0.22656, "grad_norm": 1.5464579689092532, "learning_rate": 0.003, "loss": 4.0376, "step": 22656 }, { "epoch": 0.22657, "grad_norm": 1.0228004325978513, "learning_rate": 0.003, "loss": 4.064, "step": 22657 }, { "epoch": 0.22658, "grad_norm": 1.216542862857828, "learning_rate": 0.003, "loss": 4.057, "step": 22658 }, { "epoch": 0.22659, "grad_norm": 1.3763995609191408, "learning_rate": 0.003, "loss": 4.0248, "step": 22659 }, { "epoch": 0.2266, "grad_norm": 1.6924043104114337, "learning_rate": 0.003, "loss": 4.051, "step": 22660 }, { "epoch": 0.22661, "grad_norm": 1.11683907313906, "learning_rate": 0.003, "loss": 4.0696, "step": 22661 }, { "epoch": 0.22662, "grad_norm": 1.6378062181657183, "learning_rate": 0.003, "loss": 4.0388, "step": 22662 }, { "epoch": 0.22663, "grad_norm": 1.1164816037937753, "learning_rate": 0.003, "loss": 4.0326, "step": 22663 }, { "epoch": 0.22664, "grad_norm": 1.5921206462792123, "learning_rate": 0.003, "loss": 4.0591, "step": 22664 }, { "epoch": 0.22665, "grad_norm": 0.9708473646983161, "learning_rate": 0.003, "loss": 4.0733, "step": 22665 }, { "epoch": 0.22666, "grad_norm": 1.3820591197284526, "learning_rate": 0.003, "loss": 4.072, "step": 22666 }, { "epoch": 0.22667, "grad_norm": 1.109630020127074, "learning_rate": 0.003, "loss": 4.0402, "step": 22667 }, { "epoch": 0.22668, "grad_norm": 1.3925172522796756, "learning_rate": 0.003, "loss": 4.0245, "step": 22668 }, { "epoch": 0.22669, "grad_norm": 1.3126031415495094, "learning_rate": 0.003, "loss": 4.0327, "step": 22669 }, { "epoch": 0.2267, "grad_norm": 1.148292833387036, "learning_rate": 0.003, "loss": 4.0414, "step": 22670 }, { "epoch": 0.22671, "grad_norm": 1.370358115085091, "learning_rate": 0.003, "loss": 4.0554, "step": 22671 }, { "epoch": 0.22672, "grad_norm": 1.3196396087006252, "learning_rate": 0.003, "loss": 4.0515, "step": 22672 }, { "epoch": 0.22673, "grad_norm": 1.518693881349011, "learning_rate": 0.003, "loss": 4.0119, "step": 22673 }, { "epoch": 0.22674, "grad_norm": 1.2770842397458222, "learning_rate": 0.003, "loss": 4.0319, "step": 22674 }, { "epoch": 0.22675, "grad_norm": 1.4607556447767738, "learning_rate": 0.003, "loss": 4.0778, "step": 22675 }, { "epoch": 0.22676, "grad_norm": 1.1421132126999296, "learning_rate": 0.003, "loss": 4.0358, "step": 22676 }, { "epoch": 0.22677, "grad_norm": 1.324177575118353, "learning_rate": 0.003, "loss": 4.0492, "step": 22677 }, { "epoch": 0.22678, "grad_norm": 1.0253559359484354, "learning_rate": 0.003, "loss": 4.0481, "step": 22678 }, { "epoch": 0.22679, "grad_norm": 1.5325664911066361, "learning_rate": 0.003, "loss": 4.0417, "step": 22679 }, { "epoch": 0.2268, "grad_norm": 1.0671308318897104, "learning_rate": 0.003, "loss": 4.04, "step": 22680 }, { "epoch": 0.22681, "grad_norm": 1.6768360576583528, "learning_rate": 0.003, "loss": 4.0576, "step": 22681 }, { "epoch": 0.22682, "grad_norm": 1.125262484743771, "learning_rate": 0.003, "loss": 4.0554, "step": 22682 }, { "epoch": 0.22683, "grad_norm": 1.3855513862543525, "learning_rate": 0.003, "loss": 4.0482, "step": 22683 }, { "epoch": 0.22684, "grad_norm": 1.235809625591013, "learning_rate": 0.003, "loss": 4.0624, "step": 22684 }, { "epoch": 0.22685, "grad_norm": 1.2529885237322402, "learning_rate": 0.003, "loss": 4.048, "step": 22685 }, { "epoch": 0.22686, "grad_norm": 1.3013013941449454, "learning_rate": 0.003, "loss": 4.0359, "step": 22686 }, { "epoch": 0.22687, "grad_norm": 1.2866448254940237, "learning_rate": 0.003, "loss": 4.0631, "step": 22687 }, { "epoch": 0.22688, "grad_norm": 1.3333273010739528, "learning_rate": 0.003, "loss": 4.0546, "step": 22688 }, { "epoch": 0.22689, "grad_norm": 1.3697817826252374, "learning_rate": 0.003, "loss": 4.0605, "step": 22689 }, { "epoch": 0.2269, "grad_norm": 1.1749598944282096, "learning_rate": 0.003, "loss": 4.0531, "step": 22690 }, { "epoch": 0.22691, "grad_norm": 1.3354900475986349, "learning_rate": 0.003, "loss": 4.0778, "step": 22691 }, { "epoch": 0.22692, "grad_norm": 1.2378429742426724, "learning_rate": 0.003, "loss": 4.04, "step": 22692 }, { "epoch": 0.22693, "grad_norm": 1.4046528514027574, "learning_rate": 0.003, "loss": 4.0377, "step": 22693 }, { "epoch": 0.22694, "grad_norm": 1.104016045879578, "learning_rate": 0.003, "loss": 4.0483, "step": 22694 }, { "epoch": 0.22695, "grad_norm": 1.5853447545804227, "learning_rate": 0.003, "loss": 4.0444, "step": 22695 }, { "epoch": 0.22696, "grad_norm": 1.2017974352948775, "learning_rate": 0.003, "loss": 4.0466, "step": 22696 }, { "epoch": 0.22697, "grad_norm": 1.2639960734509235, "learning_rate": 0.003, "loss": 4.0706, "step": 22697 }, { "epoch": 0.22698, "grad_norm": 1.3598055873871733, "learning_rate": 0.003, "loss": 4.0219, "step": 22698 }, { "epoch": 0.22699, "grad_norm": 1.2155778228686982, "learning_rate": 0.003, "loss": 4.0334, "step": 22699 }, { "epoch": 0.227, "grad_norm": 1.5096492497099996, "learning_rate": 0.003, "loss": 4.0444, "step": 22700 }, { "epoch": 0.22701, "grad_norm": 1.1438661949627844, "learning_rate": 0.003, "loss": 4.0717, "step": 22701 }, { "epoch": 0.22702, "grad_norm": 1.4522348604048556, "learning_rate": 0.003, "loss": 4.0592, "step": 22702 }, { "epoch": 0.22703, "grad_norm": 1.0230340496577157, "learning_rate": 0.003, "loss": 4.0368, "step": 22703 }, { "epoch": 0.22704, "grad_norm": 1.4613033441247205, "learning_rate": 0.003, "loss": 4.0569, "step": 22704 }, { "epoch": 0.22705, "grad_norm": 1.063308297536425, "learning_rate": 0.003, "loss": 4.0593, "step": 22705 }, { "epoch": 0.22706, "grad_norm": 1.5973493003176626, "learning_rate": 0.003, "loss": 4.045, "step": 22706 }, { "epoch": 0.22707, "grad_norm": 1.1719072994723225, "learning_rate": 0.003, "loss": 4.0594, "step": 22707 }, { "epoch": 0.22708, "grad_norm": 1.3647035739638114, "learning_rate": 0.003, "loss": 4.0114, "step": 22708 }, { "epoch": 0.22709, "grad_norm": 1.21264256468456, "learning_rate": 0.003, "loss": 4.0599, "step": 22709 }, { "epoch": 0.2271, "grad_norm": 1.4957409118180294, "learning_rate": 0.003, "loss": 4.0438, "step": 22710 }, { "epoch": 0.22711, "grad_norm": 1.3074504903035473, "learning_rate": 0.003, "loss": 4.0543, "step": 22711 }, { "epoch": 0.22712, "grad_norm": 1.2299409912407897, "learning_rate": 0.003, "loss": 4.0316, "step": 22712 }, { "epoch": 0.22713, "grad_norm": 1.4756516454197908, "learning_rate": 0.003, "loss": 4.0386, "step": 22713 }, { "epoch": 0.22714, "grad_norm": 1.2180637235702896, "learning_rate": 0.003, "loss": 4.0664, "step": 22714 }, { "epoch": 0.22715, "grad_norm": 1.3062867497873833, "learning_rate": 0.003, "loss": 4.0291, "step": 22715 }, { "epoch": 0.22716, "grad_norm": 1.1861936183326716, "learning_rate": 0.003, "loss": 4.0415, "step": 22716 }, { "epoch": 0.22717, "grad_norm": 1.2698126370628935, "learning_rate": 0.003, "loss": 4.0729, "step": 22717 }, { "epoch": 0.22718, "grad_norm": 1.142857025678499, "learning_rate": 0.003, "loss": 4.0606, "step": 22718 }, { "epoch": 0.22719, "grad_norm": 1.2838286118946398, "learning_rate": 0.003, "loss": 4.058, "step": 22719 }, { "epoch": 0.2272, "grad_norm": 1.3379258913071415, "learning_rate": 0.003, "loss": 4.0572, "step": 22720 }, { "epoch": 0.22721, "grad_norm": 1.1583924755668984, "learning_rate": 0.003, "loss": 4.0388, "step": 22721 }, { "epoch": 0.22722, "grad_norm": 1.338776068657845, "learning_rate": 0.003, "loss": 4.0484, "step": 22722 }, { "epoch": 0.22723, "grad_norm": 1.193967684373097, "learning_rate": 0.003, "loss": 4.0465, "step": 22723 }, { "epoch": 0.22724, "grad_norm": 1.2453421739638189, "learning_rate": 0.003, "loss": 4.0444, "step": 22724 }, { "epoch": 0.22725, "grad_norm": 1.3277986491013223, "learning_rate": 0.003, "loss": 4.0678, "step": 22725 }, { "epoch": 0.22726, "grad_norm": 1.1606456992588032, "learning_rate": 0.003, "loss": 4.046, "step": 22726 }, { "epoch": 0.22727, "grad_norm": 1.6014790251768964, "learning_rate": 0.003, "loss": 4.0679, "step": 22727 }, { "epoch": 0.22728, "grad_norm": 1.1772438646256707, "learning_rate": 0.003, "loss": 4.0562, "step": 22728 }, { "epoch": 0.22729, "grad_norm": 1.5753193741401947, "learning_rate": 0.003, "loss": 4.0566, "step": 22729 }, { "epoch": 0.2273, "grad_norm": 1.2244001581798778, "learning_rate": 0.003, "loss": 4.06, "step": 22730 }, { "epoch": 0.22731, "grad_norm": 1.3049762139181278, "learning_rate": 0.003, "loss": 4.0419, "step": 22731 }, { "epoch": 0.22732, "grad_norm": 1.4257523619725556, "learning_rate": 0.003, "loss": 4.0748, "step": 22732 }, { "epoch": 0.22733, "grad_norm": 1.1300033013606043, "learning_rate": 0.003, "loss": 4.041, "step": 22733 }, { "epoch": 0.22734, "grad_norm": 1.2789000183009511, "learning_rate": 0.003, "loss": 4.06, "step": 22734 }, { "epoch": 0.22735, "grad_norm": 1.3787011030907839, "learning_rate": 0.003, "loss": 4.0455, "step": 22735 }, { "epoch": 0.22736, "grad_norm": 1.1534785050906549, "learning_rate": 0.003, "loss": 4.0364, "step": 22736 }, { "epoch": 0.22737, "grad_norm": 1.2556409246801243, "learning_rate": 0.003, "loss": 4.0441, "step": 22737 }, { "epoch": 0.22738, "grad_norm": 1.5023911495167528, "learning_rate": 0.003, "loss": 4.0797, "step": 22738 }, { "epoch": 0.22739, "grad_norm": 1.089950690583449, "learning_rate": 0.003, "loss": 4.0776, "step": 22739 }, { "epoch": 0.2274, "grad_norm": 1.388094751486672, "learning_rate": 0.003, "loss": 4.0518, "step": 22740 }, { "epoch": 0.22741, "grad_norm": 1.1041767662188406, "learning_rate": 0.003, "loss": 4.0505, "step": 22741 }, { "epoch": 0.22742, "grad_norm": 1.577536409684799, "learning_rate": 0.003, "loss": 4.0503, "step": 22742 }, { "epoch": 0.22743, "grad_norm": 1.5105192793850055, "learning_rate": 0.003, "loss": 4.0575, "step": 22743 }, { "epoch": 0.22744, "grad_norm": 1.3115780040631255, "learning_rate": 0.003, "loss": 4.0661, "step": 22744 }, { "epoch": 0.22745, "grad_norm": 1.4678226325592918, "learning_rate": 0.003, "loss": 4.049, "step": 22745 }, { "epoch": 0.22746, "grad_norm": 1.2946841486547571, "learning_rate": 0.003, "loss": 4.0592, "step": 22746 }, { "epoch": 0.22747, "grad_norm": 1.0972080533745272, "learning_rate": 0.003, "loss": 4.0395, "step": 22747 }, { "epoch": 0.22748, "grad_norm": 1.4098976880591747, "learning_rate": 0.003, "loss": 4.0254, "step": 22748 }, { "epoch": 0.22749, "grad_norm": 0.9706114850723467, "learning_rate": 0.003, "loss": 4.0212, "step": 22749 }, { "epoch": 0.2275, "grad_norm": 1.7768992997687312, "learning_rate": 0.003, "loss": 4.0501, "step": 22750 }, { "epoch": 0.22751, "grad_norm": 1.5976243640323782, "learning_rate": 0.003, "loss": 4.0584, "step": 22751 }, { "epoch": 0.22752, "grad_norm": 1.208176486669619, "learning_rate": 0.003, "loss": 4.0394, "step": 22752 }, { "epoch": 0.22753, "grad_norm": 1.3220546996996818, "learning_rate": 0.003, "loss": 4.0203, "step": 22753 }, { "epoch": 0.22754, "grad_norm": 1.2117167761290428, "learning_rate": 0.003, "loss": 4.0527, "step": 22754 }, { "epoch": 0.22755, "grad_norm": 1.5439747829457355, "learning_rate": 0.003, "loss": 4.0581, "step": 22755 }, { "epoch": 0.22756, "grad_norm": 1.2509292818980267, "learning_rate": 0.003, "loss": 4.0559, "step": 22756 }, { "epoch": 0.22757, "grad_norm": 1.353529833415662, "learning_rate": 0.003, "loss": 4.0537, "step": 22757 }, { "epoch": 0.22758, "grad_norm": 1.22119780866865, "learning_rate": 0.003, "loss": 4.0675, "step": 22758 }, { "epoch": 0.22759, "grad_norm": 1.2846534764837474, "learning_rate": 0.003, "loss": 4.0476, "step": 22759 }, { "epoch": 0.2276, "grad_norm": 1.2337042940346146, "learning_rate": 0.003, "loss": 4.0626, "step": 22760 }, { "epoch": 0.22761, "grad_norm": 1.334364310105172, "learning_rate": 0.003, "loss": 4.0644, "step": 22761 }, { "epoch": 0.22762, "grad_norm": 1.0545672102456172, "learning_rate": 0.003, "loss": 4.0788, "step": 22762 }, { "epoch": 0.22763, "grad_norm": 1.4889896620876106, "learning_rate": 0.003, "loss": 4.0722, "step": 22763 }, { "epoch": 0.22764, "grad_norm": 1.1729072021046347, "learning_rate": 0.003, "loss": 4.0661, "step": 22764 }, { "epoch": 0.22765, "grad_norm": 1.205257120595343, "learning_rate": 0.003, "loss": 4.0372, "step": 22765 }, { "epoch": 0.22766, "grad_norm": 1.1970006365925647, "learning_rate": 0.003, "loss": 4.0603, "step": 22766 }, { "epoch": 0.22767, "grad_norm": 1.3730117095039738, "learning_rate": 0.003, "loss": 4.0569, "step": 22767 }, { "epoch": 0.22768, "grad_norm": 1.0673400445379233, "learning_rate": 0.003, "loss": 4.0805, "step": 22768 }, { "epoch": 0.22769, "grad_norm": 1.389501589959973, "learning_rate": 0.003, "loss": 4.0394, "step": 22769 }, { "epoch": 0.2277, "grad_norm": 1.1542861521096173, "learning_rate": 0.003, "loss": 4.0447, "step": 22770 }, { "epoch": 0.22771, "grad_norm": 1.189489063108176, "learning_rate": 0.003, "loss": 4.0575, "step": 22771 }, { "epoch": 0.22772, "grad_norm": 1.2319560086110348, "learning_rate": 0.003, "loss": 4.0533, "step": 22772 }, { "epoch": 0.22773, "grad_norm": 1.3255860520775213, "learning_rate": 0.003, "loss": 4.0396, "step": 22773 }, { "epoch": 0.22774, "grad_norm": 1.6902334263287309, "learning_rate": 0.003, "loss": 4.0353, "step": 22774 }, { "epoch": 0.22775, "grad_norm": 1.3989969471055916, "learning_rate": 0.003, "loss": 4.0466, "step": 22775 }, { "epoch": 0.22776, "grad_norm": 1.2443381445096287, "learning_rate": 0.003, "loss": 4.0562, "step": 22776 }, { "epoch": 0.22777, "grad_norm": 1.4293361992662825, "learning_rate": 0.003, "loss": 4.0531, "step": 22777 }, { "epoch": 0.22778, "grad_norm": 1.3550414229713295, "learning_rate": 0.003, "loss": 4.0312, "step": 22778 }, { "epoch": 0.22779, "grad_norm": 1.3182152091949482, "learning_rate": 0.003, "loss": 4.0517, "step": 22779 }, { "epoch": 0.2278, "grad_norm": 1.2091175791261757, "learning_rate": 0.003, "loss": 4.0633, "step": 22780 }, { "epoch": 0.22781, "grad_norm": 1.115419144054797, "learning_rate": 0.003, "loss": 4.0461, "step": 22781 }, { "epoch": 0.22782, "grad_norm": 1.3232712279784384, "learning_rate": 0.003, "loss": 4.0585, "step": 22782 }, { "epoch": 0.22783, "grad_norm": 1.2005645416277266, "learning_rate": 0.003, "loss": 4.0319, "step": 22783 }, { "epoch": 0.22784, "grad_norm": 1.5097870070482657, "learning_rate": 0.003, "loss": 4.0665, "step": 22784 }, { "epoch": 0.22785, "grad_norm": 1.0688476025622304, "learning_rate": 0.003, "loss": 4.0379, "step": 22785 }, { "epoch": 0.22786, "grad_norm": 1.3204882262420305, "learning_rate": 0.003, "loss": 4.0056, "step": 22786 }, { "epoch": 0.22787, "grad_norm": 1.013031325356019, "learning_rate": 0.003, "loss": 4.0631, "step": 22787 }, { "epoch": 0.22788, "grad_norm": 1.3314224675818025, "learning_rate": 0.003, "loss": 4.0683, "step": 22788 }, { "epoch": 0.22789, "grad_norm": 1.0705568142322244, "learning_rate": 0.003, "loss": 4.062, "step": 22789 }, { "epoch": 0.2279, "grad_norm": 1.2883963183099956, "learning_rate": 0.003, "loss": 4.071, "step": 22790 }, { "epoch": 0.22791, "grad_norm": 1.3430730702231144, "learning_rate": 0.003, "loss": 4.0635, "step": 22791 }, { "epoch": 0.22792, "grad_norm": 1.2845662137609606, "learning_rate": 0.003, "loss": 4.0348, "step": 22792 }, { "epoch": 0.22793, "grad_norm": 1.121581724185981, "learning_rate": 0.003, "loss": 4.0178, "step": 22793 }, { "epoch": 0.22794, "grad_norm": 1.4068842948849558, "learning_rate": 0.003, "loss": 4.047, "step": 22794 }, { "epoch": 0.22795, "grad_norm": 1.3629735629978736, "learning_rate": 0.003, "loss": 4.0993, "step": 22795 }, { "epoch": 0.22796, "grad_norm": 1.4727742275118698, "learning_rate": 0.003, "loss": 4.0528, "step": 22796 }, { "epoch": 0.22797, "grad_norm": 1.5091489239273752, "learning_rate": 0.003, "loss": 4.051, "step": 22797 }, { "epoch": 0.22798, "grad_norm": 1.257883528943881, "learning_rate": 0.003, "loss": 4.0633, "step": 22798 }, { "epoch": 0.22799, "grad_norm": 1.3645306525309961, "learning_rate": 0.003, "loss": 4.0552, "step": 22799 }, { "epoch": 0.228, "grad_norm": 1.2680722205796182, "learning_rate": 0.003, "loss": 4.0355, "step": 22800 }, { "epoch": 0.22801, "grad_norm": 1.3520412484618418, "learning_rate": 0.003, "loss": 4.0548, "step": 22801 }, { "epoch": 0.22802, "grad_norm": 1.1886074017749466, "learning_rate": 0.003, "loss": 4.0433, "step": 22802 }, { "epoch": 0.22803, "grad_norm": 1.1935355561367111, "learning_rate": 0.003, "loss": 4.058, "step": 22803 }, { "epoch": 0.22804, "grad_norm": 1.3201940689028522, "learning_rate": 0.003, "loss": 4.0481, "step": 22804 }, { "epoch": 0.22805, "grad_norm": 1.0934454414120258, "learning_rate": 0.003, "loss": 4.0293, "step": 22805 }, { "epoch": 0.22806, "grad_norm": 1.3482588962313506, "learning_rate": 0.003, "loss": 4.0423, "step": 22806 }, { "epoch": 0.22807, "grad_norm": 1.1237577331274458, "learning_rate": 0.003, "loss": 4.0359, "step": 22807 }, { "epoch": 0.22808, "grad_norm": 1.50413009597263, "learning_rate": 0.003, "loss": 4.0682, "step": 22808 }, { "epoch": 0.22809, "grad_norm": 1.0720239819102442, "learning_rate": 0.003, "loss": 4.0649, "step": 22809 }, { "epoch": 0.2281, "grad_norm": 1.34377944310639, "learning_rate": 0.003, "loss": 4.0539, "step": 22810 }, { "epoch": 0.22811, "grad_norm": 1.2586971636561433, "learning_rate": 0.003, "loss": 4.0493, "step": 22811 }, { "epoch": 0.22812, "grad_norm": 1.326930680078535, "learning_rate": 0.003, "loss": 4.049, "step": 22812 }, { "epoch": 0.22813, "grad_norm": 1.2083646576642904, "learning_rate": 0.003, "loss": 4.0209, "step": 22813 }, { "epoch": 0.22814, "grad_norm": 1.5251985459951989, "learning_rate": 0.003, "loss": 4.0328, "step": 22814 }, { "epoch": 0.22815, "grad_norm": 1.0479209292349305, "learning_rate": 0.003, "loss": 4.0499, "step": 22815 }, { "epoch": 0.22816, "grad_norm": 1.5954414927716987, "learning_rate": 0.003, "loss": 4.0631, "step": 22816 }, { "epoch": 0.22817, "grad_norm": 1.0961782039927528, "learning_rate": 0.003, "loss": 4.0165, "step": 22817 }, { "epoch": 0.22818, "grad_norm": 1.401675335635331, "learning_rate": 0.003, "loss": 4.0342, "step": 22818 }, { "epoch": 0.22819, "grad_norm": 1.2230643459470691, "learning_rate": 0.003, "loss": 4.0358, "step": 22819 }, { "epoch": 0.2282, "grad_norm": 1.3305995274495828, "learning_rate": 0.003, "loss": 4.0379, "step": 22820 }, { "epoch": 0.22821, "grad_norm": 1.2484291763090232, "learning_rate": 0.003, "loss": 4.0265, "step": 22821 }, { "epoch": 0.22822, "grad_norm": 1.3922284995542147, "learning_rate": 0.003, "loss": 4.045, "step": 22822 }, { "epoch": 0.22823, "grad_norm": 1.331782803765551, "learning_rate": 0.003, "loss": 4.0404, "step": 22823 }, { "epoch": 0.22824, "grad_norm": 1.2936263790312528, "learning_rate": 0.003, "loss": 4.0437, "step": 22824 }, { "epoch": 0.22825, "grad_norm": 1.2238490062297172, "learning_rate": 0.003, "loss": 4.0206, "step": 22825 }, { "epoch": 0.22826, "grad_norm": 1.3873058960375435, "learning_rate": 0.003, "loss": 4.0415, "step": 22826 }, { "epoch": 0.22827, "grad_norm": 1.2621561196069349, "learning_rate": 0.003, "loss": 4.0607, "step": 22827 }, { "epoch": 0.22828, "grad_norm": 1.3191413240495282, "learning_rate": 0.003, "loss": 4.0488, "step": 22828 }, { "epoch": 0.22829, "grad_norm": 1.1449900765963297, "learning_rate": 0.003, "loss": 4.0181, "step": 22829 }, { "epoch": 0.2283, "grad_norm": 1.5723864078810423, "learning_rate": 0.003, "loss": 4.0532, "step": 22830 }, { "epoch": 0.22831, "grad_norm": 0.9270222188318371, "learning_rate": 0.003, "loss": 4.0267, "step": 22831 }, { "epoch": 0.22832, "grad_norm": 1.59614219616322, "learning_rate": 0.003, "loss": 4.0606, "step": 22832 }, { "epoch": 0.22833, "grad_norm": 1.0525211230538725, "learning_rate": 0.003, "loss": 4.045, "step": 22833 }, { "epoch": 0.22834, "grad_norm": 1.3192374056002631, "learning_rate": 0.003, "loss": 4.0704, "step": 22834 }, { "epoch": 0.22835, "grad_norm": 1.194476809455396, "learning_rate": 0.003, "loss": 4.011, "step": 22835 }, { "epoch": 0.22836, "grad_norm": 1.3874118940182827, "learning_rate": 0.003, "loss": 4.0836, "step": 22836 }, { "epoch": 0.22837, "grad_norm": 1.3744848109449395, "learning_rate": 0.003, "loss": 4.0655, "step": 22837 }, { "epoch": 0.22838, "grad_norm": 1.3423859284378152, "learning_rate": 0.003, "loss": 4.096, "step": 22838 }, { "epoch": 0.22839, "grad_norm": 1.198426965913589, "learning_rate": 0.003, "loss": 4.0436, "step": 22839 }, { "epoch": 0.2284, "grad_norm": 1.493836846331223, "learning_rate": 0.003, "loss": 4.0515, "step": 22840 }, { "epoch": 0.22841, "grad_norm": 1.3312138491836896, "learning_rate": 0.003, "loss": 4.0431, "step": 22841 }, { "epoch": 0.22842, "grad_norm": 1.5003509037800031, "learning_rate": 0.003, "loss": 4.0212, "step": 22842 }, { "epoch": 0.22843, "grad_norm": 1.205974901084229, "learning_rate": 0.003, "loss": 4.0537, "step": 22843 }, { "epoch": 0.22844, "grad_norm": 1.3298920680682018, "learning_rate": 0.003, "loss": 4.0266, "step": 22844 }, { "epoch": 0.22845, "grad_norm": 1.3421983269805815, "learning_rate": 0.003, "loss": 4.0178, "step": 22845 }, { "epoch": 0.22846, "grad_norm": 1.166847014841375, "learning_rate": 0.003, "loss": 4.0304, "step": 22846 }, { "epoch": 0.22847, "grad_norm": 1.4362377752411226, "learning_rate": 0.003, "loss": 4.063, "step": 22847 }, { "epoch": 0.22848, "grad_norm": 1.299435089518182, "learning_rate": 0.003, "loss": 4.0793, "step": 22848 }, { "epoch": 0.22849, "grad_norm": 1.3610712132367684, "learning_rate": 0.003, "loss": 4.0541, "step": 22849 }, { "epoch": 0.2285, "grad_norm": 1.2032297939998098, "learning_rate": 0.003, "loss": 4.0424, "step": 22850 }, { "epoch": 0.22851, "grad_norm": 1.3579332227391925, "learning_rate": 0.003, "loss": 4.0346, "step": 22851 }, { "epoch": 0.22852, "grad_norm": 1.0683078712159948, "learning_rate": 0.003, "loss": 4.0108, "step": 22852 }, { "epoch": 0.22853, "grad_norm": 1.6043031294033705, "learning_rate": 0.003, "loss": 4.0491, "step": 22853 }, { "epoch": 0.22854, "grad_norm": 1.1035714288783613, "learning_rate": 0.003, "loss": 4.0514, "step": 22854 }, { "epoch": 0.22855, "grad_norm": 1.46243142753709, "learning_rate": 0.003, "loss": 4.0799, "step": 22855 }, { "epoch": 0.22856, "grad_norm": 1.1288347218629557, "learning_rate": 0.003, "loss": 4.0804, "step": 22856 }, { "epoch": 0.22857, "grad_norm": 1.4360025524383433, "learning_rate": 0.003, "loss": 4.0295, "step": 22857 }, { "epoch": 0.22858, "grad_norm": 1.177066293382833, "learning_rate": 0.003, "loss": 4.0482, "step": 22858 }, { "epoch": 0.22859, "grad_norm": 1.6349447769258982, "learning_rate": 0.003, "loss": 4.046, "step": 22859 }, { "epoch": 0.2286, "grad_norm": 1.0161831975431206, "learning_rate": 0.003, "loss": 4.0761, "step": 22860 }, { "epoch": 0.22861, "grad_norm": 1.4289237473634253, "learning_rate": 0.003, "loss": 4.0343, "step": 22861 }, { "epoch": 0.22862, "grad_norm": 1.1343028599970997, "learning_rate": 0.003, "loss": 4.0752, "step": 22862 }, { "epoch": 0.22863, "grad_norm": 1.3729139119942362, "learning_rate": 0.003, "loss": 4.0756, "step": 22863 }, { "epoch": 0.22864, "grad_norm": 1.0333333687581276, "learning_rate": 0.003, "loss": 4.0579, "step": 22864 }, { "epoch": 0.22865, "grad_norm": 1.3602636566405157, "learning_rate": 0.003, "loss": 4.0357, "step": 22865 }, { "epoch": 0.22866, "grad_norm": 1.2408380139459476, "learning_rate": 0.003, "loss": 4.0495, "step": 22866 }, { "epoch": 0.22867, "grad_norm": 1.432421938149444, "learning_rate": 0.003, "loss": 4.0379, "step": 22867 }, { "epoch": 0.22868, "grad_norm": 1.2288204950857995, "learning_rate": 0.003, "loss": 4.0912, "step": 22868 }, { "epoch": 0.22869, "grad_norm": 1.2689135070146418, "learning_rate": 0.003, "loss": 4.0645, "step": 22869 }, { "epoch": 0.2287, "grad_norm": 1.3511059785670072, "learning_rate": 0.003, "loss": 4.0542, "step": 22870 }, { "epoch": 0.22871, "grad_norm": 1.1712551823346788, "learning_rate": 0.003, "loss": 4.026, "step": 22871 }, { "epoch": 0.22872, "grad_norm": 1.29458357449702, "learning_rate": 0.003, "loss": 4.0225, "step": 22872 }, { "epoch": 0.22873, "grad_norm": 1.4486140513751453, "learning_rate": 0.003, "loss": 4.0641, "step": 22873 }, { "epoch": 0.22874, "grad_norm": 1.3324991638906907, "learning_rate": 0.003, "loss": 4.0486, "step": 22874 }, { "epoch": 0.22875, "grad_norm": 1.4379664962729155, "learning_rate": 0.003, "loss": 4.0147, "step": 22875 }, { "epoch": 0.22876, "grad_norm": 1.4621484333769805, "learning_rate": 0.003, "loss": 4.017, "step": 22876 }, { "epoch": 0.22877, "grad_norm": 1.3807068471846857, "learning_rate": 0.003, "loss": 4.0552, "step": 22877 }, { "epoch": 0.22878, "grad_norm": 1.1550550968192956, "learning_rate": 0.003, "loss": 4.0552, "step": 22878 }, { "epoch": 0.22879, "grad_norm": 1.2105770412043309, "learning_rate": 0.003, "loss": 4.0237, "step": 22879 }, { "epoch": 0.2288, "grad_norm": 1.25293275091284, "learning_rate": 0.003, "loss": 4.0414, "step": 22880 }, { "epoch": 0.22881, "grad_norm": 1.1599415415362224, "learning_rate": 0.003, "loss": 4.0552, "step": 22881 }, { "epoch": 0.22882, "grad_norm": 1.356880953365146, "learning_rate": 0.003, "loss": 4.0483, "step": 22882 }, { "epoch": 0.22883, "grad_norm": 1.3682136629109296, "learning_rate": 0.003, "loss": 4.0221, "step": 22883 }, { "epoch": 0.22884, "grad_norm": 1.5085045762272722, "learning_rate": 0.003, "loss": 4.0352, "step": 22884 }, { "epoch": 0.22885, "grad_norm": 1.1662355438875212, "learning_rate": 0.003, "loss": 4.0352, "step": 22885 }, { "epoch": 0.22886, "grad_norm": 1.3460202999737598, "learning_rate": 0.003, "loss": 4.0645, "step": 22886 }, { "epoch": 0.22887, "grad_norm": 1.2062139329278307, "learning_rate": 0.003, "loss": 4.0364, "step": 22887 }, { "epoch": 0.22888, "grad_norm": 1.3084987947299225, "learning_rate": 0.003, "loss": 4.0099, "step": 22888 }, { "epoch": 0.22889, "grad_norm": 1.2595171984947928, "learning_rate": 0.003, "loss": 4.0519, "step": 22889 }, { "epoch": 0.2289, "grad_norm": 1.15468884380117, "learning_rate": 0.003, "loss": 4.0466, "step": 22890 }, { "epoch": 0.22891, "grad_norm": 1.2945936313729483, "learning_rate": 0.003, "loss": 4.0466, "step": 22891 }, { "epoch": 0.22892, "grad_norm": 1.4032639666308364, "learning_rate": 0.003, "loss": 4.0156, "step": 22892 }, { "epoch": 0.22893, "grad_norm": 1.1528761000140313, "learning_rate": 0.003, "loss": 4.0319, "step": 22893 }, { "epoch": 0.22894, "grad_norm": 1.2795750717064307, "learning_rate": 0.003, "loss": 4.0739, "step": 22894 }, { "epoch": 0.22895, "grad_norm": 1.1631033517975322, "learning_rate": 0.003, "loss": 4.0401, "step": 22895 }, { "epoch": 0.22896, "grad_norm": 1.347506369499531, "learning_rate": 0.003, "loss": 4.0537, "step": 22896 }, { "epoch": 0.22897, "grad_norm": 1.1553451292295511, "learning_rate": 0.003, "loss": 4.0043, "step": 22897 }, { "epoch": 0.22898, "grad_norm": 1.5172033899346415, "learning_rate": 0.003, "loss": 4.0303, "step": 22898 }, { "epoch": 0.22899, "grad_norm": 1.4964559666410877, "learning_rate": 0.003, "loss": 4.0481, "step": 22899 }, { "epoch": 0.229, "grad_norm": 1.0558874002499534, "learning_rate": 0.003, "loss": 4.0232, "step": 22900 }, { "epoch": 0.22901, "grad_norm": 1.5504407578681763, "learning_rate": 0.003, "loss": 4.032, "step": 22901 }, { "epoch": 0.22902, "grad_norm": 1.323277562639125, "learning_rate": 0.003, "loss": 4.0473, "step": 22902 }, { "epoch": 0.22903, "grad_norm": 1.3913718127575616, "learning_rate": 0.003, "loss": 4.0712, "step": 22903 }, { "epoch": 0.22904, "grad_norm": 1.285094747518903, "learning_rate": 0.003, "loss": 4.0087, "step": 22904 }, { "epoch": 0.22905, "grad_norm": 1.2167559944301476, "learning_rate": 0.003, "loss": 4.0482, "step": 22905 }, { "epoch": 0.22906, "grad_norm": 1.234721589474369, "learning_rate": 0.003, "loss": 4.0566, "step": 22906 }, { "epoch": 0.22907, "grad_norm": 1.177360279709588, "learning_rate": 0.003, "loss": 4.0525, "step": 22907 }, { "epoch": 0.22908, "grad_norm": 1.3467052007901217, "learning_rate": 0.003, "loss": 4.0492, "step": 22908 }, { "epoch": 0.22909, "grad_norm": 1.2610553027562679, "learning_rate": 0.003, "loss": 4.0521, "step": 22909 }, { "epoch": 0.2291, "grad_norm": 1.2236670577041506, "learning_rate": 0.003, "loss": 4.0397, "step": 22910 }, { "epoch": 0.22911, "grad_norm": 1.455089190712894, "learning_rate": 0.003, "loss": 4.0352, "step": 22911 }, { "epoch": 0.22912, "grad_norm": 1.2685255647901654, "learning_rate": 0.003, "loss": 4.0706, "step": 22912 }, { "epoch": 0.22913, "grad_norm": 1.3934264149350137, "learning_rate": 0.003, "loss": 4.0487, "step": 22913 }, { "epoch": 0.22914, "grad_norm": 1.21597673338576, "learning_rate": 0.003, "loss": 4.0642, "step": 22914 }, { "epoch": 0.22915, "grad_norm": 1.3134165822709327, "learning_rate": 0.003, "loss": 4.0519, "step": 22915 }, { "epoch": 0.22916, "grad_norm": 1.2909098050528915, "learning_rate": 0.003, "loss": 4.0529, "step": 22916 }, { "epoch": 0.22917, "grad_norm": 1.4851169396437347, "learning_rate": 0.003, "loss": 4.0308, "step": 22917 }, { "epoch": 0.22918, "grad_norm": 1.2788011360923555, "learning_rate": 0.003, "loss": 4.0642, "step": 22918 }, { "epoch": 0.22919, "grad_norm": 1.3221037668694506, "learning_rate": 0.003, "loss": 4.0443, "step": 22919 }, { "epoch": 0.2292, "grad_norm": 1.5433161913020008, "learning_rate": 0.003, "loss": 4.075, "step": 22920 }, { "epoch": 0.22921, "grad_norm": 1.2422243655526495, "learning_rate": 0.003, "loss": 4.0513, "step": 22921 }, { "epoch": 0.22922, "grad_norm": 1.2065210526072263, "learning_rate": 0.003, "loss": 4.045, "step": 22922 }, { "epoch": 0.22923, "grad_norm": 1.1431078244331845, "learning_rate": 0.003, "loss": 4.0406, "step": 22923 }, { "epoch": 0.22924, "grad_norm": 1.397505579264044, "learning_rate": 0.003, "loss": 4.0348, "step": 22924 }, { "epoch": 0.22925, "grad_norm": 1.1318338746097767, "learning_rate": 0.003, "loss": 4.0388, "step": 22925 }, { "epoch": 0.22926, "grad_norm": 1.5262692206953534, "learning_rate": 0.003, "loss": 4.0773, "step": 22926 }, { "epoch": 0.22927, "grad_norm": 1.2262615429393733, "learning_rate": 0.003, "loss": 4.0528, "step": 22927 }, { "epoch": 0.22928, "grad_norm": 1.389586082414331, "learning_rate": 0.003, "loss": 4.0372, "step": 22928 }, { "epoch": 0.22929, "grad_norm": 1.2247846332805663, "learning_rate": 0.003, "loss": 4.0591, "step": 22929 }, { "epoch": 0.2293, "grad_norm": 1.3678563895752596, "learning_rate": 0.003, "loss": 4.0152, "step": 22930 }, { "epoch": 0.22931, "grad_norm": 1.3002273140429303, "learning_rate": 0.003, "loss": 4.0503, "step": 22931 }, { "epoch": 0.22932, "grad_norm": 1.1659909609621704, "learning_rate": 0.003, "loss": 4.0052, "step": 22932 }, { "epoch": 0.22933, "grad_norm": 1.3198104999793905, "learning_rate": 0.003, "loss": 4.0585, "step": 22933 }, { "epoch": 0.22934, "grad_norm": 1.2667220713990213, "learning_rate": 0.003, "loss": 4.0337, "step": 22934 }, { "epoch": 0.22935, "grad_norm": 1.3144717008615145, "learning_rate": 0.003, "loss": 4.0102, "step": 22935 }, { "epoch": 0.22936, "grad_norm": 1.2291184178785146, "learning_rate": 0.003, "loss": 4.0393, "step": 22936 }, { "epoch": 0.22937, "grad_norm": 1.64357979360127, "learning_rate": 0.003, "loss": 4.0437, "step": 22937 }, { "epoch": 0.22938, "grad_norm": 0.9299019781222478, "learning_rate": 0.003, "loss": 4.0353, "step": 22938 }, { "epoch": 0.22939, "grad_norm": 1.4757190447923763, "learning_rate": 0.003, "loss": 4.0451, "step": 22939 }, { "epoch": 0.2294, "grad_norm": 1.3135447530156399, "learning_rate": 0.003, "loss": 4.0317, "step": 22940 }, { "epoch": 0.22941, "grad_norm": 1.4439066531604614, "learning_rate": 0.003, "loss": 4.0655, "step": 22941 }, { "epoch": 0.22942, "grad_norm": 1.261143192288765, "learning_rate": 0.003, "loss": 4.0036, "step": 22942 }, { "epoch": 0.22943, "grad_norm": 1.295165090004534, "learning_rate": 0.003, "loss": 4.0242, "step": 22943 }, { "epoch": 0.22944, "grad_norm": 1.1359597889951656, "learning_rate": 0.003, "loss": 4.0488, "step": 22944 }, { "epoch": 0.22945, "grad_norm": 1.3768127618117638, "learning_rate": 0.003, "loss": 4.0375, "step": 22945 }, { "epoch": 0.22946, "grad_norm": 1.3472858451524485, "learning_rate": 0.003, "loss": 4.0318, "step": 22946 }, { "epoch": 0.22947, "grad_norm": 1.0278995294613298, "learning_rate": 0.003, "loss": 4.0414, "step": 22947 }, { "epoch": 0.22948, "grad_norm": 1.5433280328381243, "learning_rate": 0.003, "loss": 4.079, "step": 22948 }, { "epoch": 0.22949, "grad_norm": 1.1117553909390965, "learning_rate": 0.003, "loss": 4.0523, "step": 22949 }, { "epoch": 0.2295, "grad_norm": 1.4920059526550298, "learning_rate": 0.003, "loss": 4.0527, "step": 22950 }, { "epoch": 0.22951, "grad_norm": 0.993928301649757, "learning_rate": 0.003, "loss": 4.0563, "step": 22951 }, { "epoch": 0.22952, "grad_norm": 1.4335446024072818, "learning_rate": 0.003, "loss": 4.0792, "step": 22952 }, { "epoch": 0.22953, "grad_norm": 1.2702029848287875, "learning_rate": 0.003, "loss": 4.0319, "step": 22953 }, { "epoch": 0.22954, "grad_norm": 1.422156469107183, "learning_rate": 0.003, "loss": 4.0566, "step": 22954 }, { "epoch": 0.22955, "grad_norm": 1.5017688053877793, "learning_rate": 0.003, "loss": 4.0547, "step": 22955 }, { "epoch": 0.22956, "grad_norm": 1.1832689528429465, "learning_rate": 0.003, "loss": 4.0526, "step": 22956 }, { "epoch": 0.22957, "grad_norm": 1.3202163592733576, "learning_rate": 0.003, "loss": 4.0512, "step": 22957 }, { "epoch": 0.22958, "grad_norm": 1.2773921373579953, "learning_rate": 0.003, "loss": 4.0571, "step": 22958 }, { "epoch": 0.22959, "grad_norm": 1.375776974065839, "learning_rate": 0.003, "loss": 4.0426, "step": 22959 }, { "epoch": 0.2296, "grad_norm": 1.173674651647164, "learning_rate": 0.003, "loss": 3.9901, "step": 22960 }, { "epoch": 0.22961, "grad_norm": 1.137614899206587, "learning_rate": 0.003, "loss": 4.0428, "step": 22961 }, { "epoch": 0.22962, "grad_norm": 1.2713892856399858, "learning_rate": 0.003, "loss": 4.0633, "step": 22962 }, { "epoch": 0.22963, "grad_norm": 1.1695322673610669, "learning_rate": 0.003, "loss": 4.0621, "step": 22963 }, { "epoch": 0.22964, "grad_norm": 1.4291921316022251, "learning_rate": 0.003, "loss": 4.0359, "step": 22964 }, { "epoch": 0.22965, "grad_norm": 1.2511351489355886, "learning_rate": 0.003, "loss": 4.05, "step": 22965 }, { "epoch": 0.22966, "grad_norm": 1.577778526342389, "learning_rate": 0.003, "loss": 4.0326, "step": 22966 }, { "epoch": 0.22967, "grad_norm": 1.3302290332583357, "learning_rate": 0.003, "loss": 4.0767, "step": 22967 }, { "epoch": 0.22968, "grad_norm": 1.0985676142568965, "learning_rate": 0.003, "loss": 4.0458, "step": 22968 }, { "epoch": 0.22969, "grad_norm": 1.256983007767673, "learning_rate": 0.003, "loss": 4.0664, "step": 22969 }, { "epoch": 0.2297, "grad_norm": 1.4345333632950201, "learning_rate": 0.003, "loss": 4.0315, "step": 22970 }, { "epoch": 0.22971, "grad_norm": 1.400421469249285, "learning_rate": 0.003, "loss": 4.0715, "step": 22971 }, { "epoch": 0.22972, "grad_norm": 1.1891576143755764, "learning_rate": 0.003, "loss": 4.0517, "step": 22972 }, { "epoch": 0.22973, "grad_norm": 1.1912768620700784, "learning_rate": 0.003, "loss": 4.0299, "step": 22973 }, { "epoch": 0.22974, "grad_norm": 1.3218740906006485, "learning_rate": 0.003, "loss": 4.0648, "step": 22974 }, { "epoch": 0.22975, "grad_norm": 1.4549893013901607, "learning_rate": 0.003, "loss": 4.0572, "step": 22975 }, { "epoch": 0.22976, "grad_norm": 1.2714160560273011, "learning_rate": 0.003, "loss": 4.0464, "step": 22976 }, { "epoch": 0.22977, "grad_norm": 1.4377271393819862, "learning_rate": 0.003, "loss": 4.0505, "step": 22977 }, { "epoch": 0.22978, "grad_norm": 1.2509815525636199, "learning_rate": 0.003, "loss": 4.0434, "step": 22978 }, { "epoch": 0.22979, "grad_norm": 1.1503119946235718, "learning_rate": 0.003, "loss": 4.0447, "step": 22979 }, { "epoch": 0.2298, "grad_norm": 1.4611575540202049, "learning_rate": 0.003, "loss": 4.0639, "step": 22980 }, { "epoch": 0.22981, "grad_norm": 0.9394057189840224, "learning_rate": 0.003, "loss": 4.0667, "step": 22981 }, { "epoch": 0.22982, "grad_norm": 1.3346211156308416, "learning_rate": 0.003, "loss": 4.0442, "step": 22982 }, { "epoch": 0.22983, "grad_norm": 1.2155765157826248, "learning_rate": 0.003, "loss": 4.0631, "step": 22983 }, { "epoch": 0.22984, "grad_norm": 1.5584576073336336, "learning_rate": 0.003, "loss": 4.0528, "step": 22984 }, { "epoch": 0.22985, "grad_norm": 1.0796233892676972, "learning_rate": 0.003, "loss": 4.0468, "step": 22985 }, { "epoch": 0.22986, "grad_norm": 1.4850812765520647, "learning_rate": 0.003, "loss": 4.0336, "step": 22986 }, { "epoch": 0.22987, "grad_norm": 1.444019742132719, "learning_rate": 0.003, "loss": 4.0602, "step": 22987 }, { "epoch": 0.22988, "grad_norm": 1.2808928261785606, "learning_rate": 0.003, "loss": 4.0468, "step": 22988 }, { "epoch": 0.22989, "grad_norm": 1.2188613157029755, "learning_rate": 0.003, "loss": 4.0311, "step": 22989 }, { "epoch": 0.2299, "grad_norm": 1.3536175566759292, "learning_rate": 0.003, "loss": 4.0527, "step": 22990 }, { "epoch": 0.22991, "grad_norm": 1.4217767897760307, "learning_rate": 0.003, "loss": 4.0721, "step": 22991 }, { "epoch": 0.22992, "grad_norm": 1.589064135663114, "learning_rate": 0.003, "loss": 4.0356, "step": 22992 }, { "epoch": 0.22993, "grad_norm": 1.205479423377943, "learning_rate": 0.003, "loss": 4.0872, "step": 22993 }, { "epoch": 0.22994, "grad_norm": 1.3154851765166542, "learning_rate": 0.003, "loss": 4.0766, "step": 22994 }, { "epoch": 0.22995, "grad_norm": 1.3085659744248501, "learning_rate": 0.003, "loss": 4.0571, "step": 22995 }, { "epoch": 0.22996, "grad_norm": 1.0965907825938541, "learning_rate": 0.003, "loss": 4.0423, "step": 22996 }, { "epoch": 0.22997, "grad_norm": 1.4478601107402107, "learning_rate": 0.003, "loss": 4.0167, "step": 22997 }, { "epoch": 0.22998, "grad_norm": 1.116756667172281, "learning_rate": 0.003, "loss": 4.0618, "step": 22998 }, { "epoch": 0.22999, "grad_norm": 1.4968404198900556, "learning_rate": 0.003, "loss": 4.027, "step": 22999 }, { "epoch": 0.23, "grad_norm": 1.2554377085216648, "learning_rate": 0.003, "loss": 4.0546, "step": 23000 }, { "epoch": 0.23001, "grad_norm": 1.3627960256708616, "learning_rate": 0.003, "loss": 4.0671, "step": 23001 }, { "epoch": 0.23002, "grad_norm": 1.3359068277671018, "learning_rate": 0.003, "loss": 4.0466, "step": 23002 }, { "epoch": 0.23003, "grad_norm": 1.2518805679187248, "learning_rate": 0.003, "loss": 4.0317, "step": 23003 }, { "epoch": 0.23004, "grad_norm": 1.4332295048459291, "learning_rate": 0.003, "loss": 4.0618, "step": 23004 }, { "epoch": 0.23005, "grad_norm": 1.0469876339429016, "learning_rate": 0.003, "loss": 4.0591, "step": 23005 }, { "epoch": 0.23006, "grad_norm": 1.3750879167506647, "learning_rate": 0.003, "loss": 4.0548, "step": 23006 }, { "epoch": 0.23007, "grad_norm": 1.102596208160135, "learning_rate": 0.003, "loss": 4.0412, "step": 23007 }, { "epoch": 0.23008, "grad_norm": 1.4818339892917813, "learning_rate": 0.003, "loss": 4.0594, "step": 23008 }, { "epoch": 0.23009, "grad_norm": 1.2458263803707696, "learning_rate": 0.003, "loss": 4.0521, "step": 23009 }, { "epoch": 0.2301, "grad_norm": 1.2125471959171752, "learning_rate": 0.003, "loss": 4.0379, "step": 23010 }, { "epoch": 0.23011, "grad_norm": 1.3658388882001644, "learning_rate": 0.003, "loss": 4.0543, "step": 23011 }, { "epoch": 0.23012, "grad_norm": 1.353890222092544, "learning_rate": 0.003, "loss": 4.0326, "step": 23012 }, { "epoch": 0.23013, "grad_norm": 1.6489701885843449, "learning_rate": 0.003, "loss": 4.0529, "step": 23013 }, { "epoch": 0.23014, "grad_norm": 1.088813020669272, "learning_rate": 0.003, "loss": 4.027, "step": 23014 }, { "epoch": 0.23015, "grad_norm": 1.4024324704957338, "learning_rate": 0.003, "loss": 4.0427, "step": 23015 }, { "epoch": 0.23016, "grad_norm": 1.2345054187348519, "learning_rate": 0.003, "loss": 4.0337, "step": 23016 }, { "epoch": 0.23017, "grad_norm": 1.5140625233767058, "learning_rate": 0.003, "loss": 4.0614, "step": 23017 }, { "epoch": 0.23018, "grad_norm": 1.0472359944970655, "learning_rate": 0.003, "loss": 4.0197, "step": 23018 }, { "epoch": 0.23019, "grad_norm": 1.5599569042159376, "learning_rate": 0.003, "loss": 4.0252, "step": 23019 }, { "epoch": 0.2302, "grad_norm": 1.1192254538437625, "learning_rate": 0.003, "loss": 4.055, "step": 23020 }, { "epoch": 0.23021, "grad_norm": 1.672551497850161, "learning_rate": 0.003, "loss": 4.0354, "step": 23021 }, { "epoch": 0.23022, "grad_norm": 1.3578940341893924, "learning_rate": 0.003, "loss": 4.0418, "step": 23022 }, { "epoch": 0.23023, "grad_norm": 1.0490870663434972, "learning_rate": 0.003, "loss": 4.0514, "step": 23023 }, { "epoch": 0.23024, "grad_norm": 1.4833999301909206, "learning_rate": 0.003, "loss": 4.0201, "step": 23024 }, { "epoch": 0.23025, "grad_norm": 1.1904474999043988, "learning_rate": 0.003, "loss": 4.0363, "step": 23025 }, { "epoch": 0.23026, "grad_norm": 1.22093948956686, "learning_rate": 0.003, "loss": 4.05, "step": 23026 }, { "epoch": 0.23027, "grad_norm": 1.283267201391806, "learning_rate": 0.003, "loss": 4.0499, "step": 23027 }, { "epoch": 0.23028, "grad_norm": 1.3487053628796137, "learning_rate": 0.003, "loss": 4.0537, "step": 23028 }, { "epoch": 0.23029, "grad_norm": 1.3663018286440518, "learning_rate": 0.003, "loss": 4.0211, "step": 23029 }, { "epoch": 0.2303, "grad_norm": 1.3141607999694707, "learning_rate": 0.003, "loss": 4.0344, "step": 23030 }, { "epoch": 0.23031, "grad_norm": 1.3381526802135097, "learning_rate": 0.003, "loss": 4.0391, "step": 23031 }, { "epoch": 0.23032, "grad_norm": 1.147102598297343, "learning_rate": 0.003, "loss": 4.0444, "step": 23032 }, { "epoch": 0.23033, "grad_norm": 1.1537468439259586, "learning_rate": 0.003, "loss": 4.0374, "step": 23033 }, { "epoch": 0.23034, "grad_norm": 1.2070762700690738, "learning_rate": 0.003, "loss": 4.0385, "step": 23034 }, { "epoch": 0.23035, "grad_norm": 1.3833601060265663, "learning_rate": 0.003, "loss": 4.0558, "step": 23035 }, { "epoch": 0.23036, "grad_norm": 1.4867932786299043, "learning_rate": 0.003, "loss": 4.0192, "step": 23036 }, { "epoch": 0.23037, "grad_norm": 1.0908046132988751, "learning_rate": 0.003, "loss": 4.0684, "step": 23037 }, { "epoch": 0.23038, "grad_norm": 1.2936273598672328, "learning_rate": 0.003, "loss": 4.0451, "step": 23038 }, { "epoch": 0.23039, "grad_norm": 1.3801343216314799, "learning_rate": 0.003, "loss": 4.0261, "step": 23039 }, { "epoch": 0.2304, "grad_norm": 1.1626614428690953, "learning_rate": 0.003, "loss": 4.0649, "step": 23040 }, { "epoch": 0.23041, "grad_norm": 1.3698695749482124, "learning_rate": 0.003, "loss": 4.0688, "step": 23041 }, { "epoch": 0.23042, "grad_norm": 1.2378153196003894, "learning_rate": 0.003, "loss": 3.9983, "step": 23042 }, { "epoch": 0.23043, "grad_norm": 1.577881438326901, "learning_rate": 0.003, "loss": 4.0247, "step": 23043 }, { "epoch": 0.23044, "grad_norm": 1.2460208256319625, "learning_rate": 0.003, "loss": 4.023, "step": 23044 }, { "epoch": 0.23045, "grad_norm": 1.4455431353142316, "learning_rate": 0.003, "loss": 4.0407, "step": 23045 }, { "epoch": 0.23046, "grad_norm": 1.2795392871288738, "learning_rate": 0.003, "loss": 4.0449, "step": 23046 }, { "epoch": 0.23047, "grad_norm": 1.3033283084011695, "learning_rate": 0.003, "loss": 4.0459, "step": 23047 }, { "epoch": 0.23048, "grad_norm": 1.3811992126023218, "learning_rate": 0.003, "loss": 4.0278, "step": 23048 }, { "epoch": 0.23049, "grad_norm": 1.1930207159835433, "learning_rate": 0.003, "loss": 4.0339, "step": 23049 }, { "epoch": 0.2305, "grad_norm": 1.4585760788983548, "learning_rate": 0.003, "loss": 4.0313, "step": 23050 }, { "epoch": 0.23051, "grad_norm": 1.358262142646777, "learning_rate": 0.003, "loss": 4.0375, "step": 23051 }, { "epoch": 0.23052, "grad_norm": 1.3419042260630287, "learning_rate": 0.003, "loss": 4.0558, "step": 23052 }, { "epoch": 0.23053, "grad_norm": 1.4121503378881262, "learning_rate": 0.003, "loss": 4.0629, "step": 23053 }, { "epoch": 0.23054, "grad_norm": 1.1599489122641409, "learning_rate": 0.003, "loss": 4.052, "step": 23054 }, { "epoch": 0.23055, "grad_norm": 1.2760740138601372, "learning_rate": 0.003, "loss": 4.0523, "step": 23055 }, { "epoch": 0.23056, "grad_norm": 1.096737656844803, "learning_rate": 0.003, "loss": 4.0239, "step": 23056 }, { "epoch": 0.23057, "grad_norm": 1.583486108065741, "learning_rate": 0.003, "loss": 4.0814, "step": 23057 }, { "epoch": 0.23058, "grad_norm": 1.1276380647128694, "learning_rate": 0.003, "loss": 4.0114, "step": 23058 }, { "epoch": 0.23059, "grad_norm": 1.5101198500495259, "learning_rate": 0.003, "loss": 4.0255, "step": 23059 }, { "epoch": 0.2306, "grad_norm": 1.0406442245597265, "learning_rate": 0.003, "loss": 4.0442, "step": 23060 }, { "epoch": 0.23061, "grad_norm": 1.6043601362782616, "learning_rate": 0.003, "loss": 4.0873, "step": 23061 }, { "epoch": 0.23062, "grad_norm": 1.1131308412584864, "learning_rate": 0.003, "loss": 4.0207, "step": 23062 }, { "epoch": 0.23063, "grad_norm": 1.3691056020948003, "learning_rate": 0.003, "loss": 4.0648, "step": 23063 }, { "epoch": 0.23064, "grad_norm": 1.1932208340525927, "learning_rate": 0.003, "loss": 4.0497, "step": 23064 }, { "epoch": 0.23065, "grad_norm": 1.4098137332654574, "learning_rate": 0.003, "loss": 4.044, "step": 23065 }, { "epoch": 0.23066, "grad_norm": 1.3104140321873772, "learning_rate": 0.003, "loss": 4.0543, "step": 23066 }, { "epoch": 0.23067, "grad_norm": 1.3614850118300021, "learning_rate": 0.003, "loss": 4.0375, "step": 23067 }, { "epoch": 0.23068, "grad_norm": 1.2275383517188325, "learning_rate": 0.003, "loss": 4.0249, "step": 23068 }, { "epoch": 0.23069, "grad_norm": 1.5518594230598315, "learning_rate": 0.003, "loss": 4.0561, "step": 23069 }, { "epoch": 0.2307, "grad_norm": 1.1123928582963447, "learning_rate": 0.003, "loss": 4.05, "step": 23070 }, { "epoch": 0.23071, "grad_norm": 1.372531640792637, "learning_rate": 0.003, "loss": 4.0656, "step": 23071 }, { "epoch": 0.23072, "grad_norm": 1.0690552375552709, "learning_rate": 0.003, "loss": 4.032, "step": 23072 }, { "epoch": 0.23073, "grad_norm": 1.5236579195250115, "learning_rate": 0.003, "loss": 4.0544, "step": 23073 }, { "epoch": 0.23074, "grad_norm": 1.221992758364506, "learning_rate": 0.003, "loss": 4.0254, "step": 23074 }, { "epoch": 0.23075, "grad_norm": 1.483652994064702, "learning_rate": 0.003, "loss": 4.0731, "step": 23075 }, { "epoch": 0.23076, "grad_norm": 1.2949621355925267, "learning_rate": 0.003, "loss": 4.0634, "step": 23076 }, { "epoch": 0.23077, "grad_norm": 1.220737057512629, "learning_rate": 0.003, "loss": 4.0392, "step": 23077 }, { "epoch": 0.23078, "grad_norm": 1.283862303259275, "learning_rate": 0.003, "loss": 4.0462, "step": 23078 }, { "epoch": 0.23079, "grad_norm": 1.2600780269031073, "learning_rate": 0.003, "loss": 4.0124, "step": 23079 }, { "epoch": 0.2308, "grad_norm": 1.3530737646299207, "learning_rate": 0.003, "loss": 4.0659, "step": 23080 }, { "epoch": 0.23081, "grad_norm": 1.253509029242236, "learning_rate": 0.003, "loss": 4.0612, "step": 23081 }, { "epoch": 0.23082, "grad_norm": 1.1352375655895686, "learning_rate": 0.003, "loss": 4.0291, "step": 23082 }, { "epoch": 0.23083, "grad_norm": 1.4336850022395355, "learning_rate": 0.003, "loss": 4.0444, "step": 23083 }, { "epoch": 0.23084, "grad_norm": 1.3638809663352245, "learning_rate": 0.003, "loss": 4.0512, "step": 23084 }, { "epoch": 0.23085, "grad_norm": 1.207880423206292, "learning_rate": 0.003, "loss": 4.0187, "step": 23085 }, { "epoch": 0.23086, "grad_norm": 1.1490890793397543, "learning_rate": 0.003, "loss": 4.0317, "step": 23086 }, { "epoch": 0.23087, "grad_norm": 1.5840436288048507, "learning_rate": 0.003, "loss": 4.0422, "step": 23087 }, { "epoch": 0.23088, "grad_norm": 1.0133310027332907, "learning_rate": 0.003, "loss": 4.0412, "step": 23088 }, { "epoch": 0.23089, "grad_norm": 1.4733120744342365, "learning_rate": 0.003, "loss": 4.0222, "step": 23089 }, { "epoch": 0.2309, "grad_norm": 1.0767299521882396, "learning_rate": 0.003, "loss": 4.0265, "step": 23090 }, { "epoch": 0.23091, "grad_norm": 1.4570100154875747, "learning_rate": 0.003, "loss": 4.072, "step": 23091 }, { "epoch": 0.23092, "grad_norm": 1.0391219016099587, "learning_rate": 0.003, "loss": 4.0535, "step": 23092 }, { "epoch": 0.23093, "grad_norm": 1.4364810847065508, "learning_rate": 0.003, "loss": 4.0629, "step": 23093 }, { "epoch": 0.23094, "grad_norm": 1.3610556377897853, "learning_rate": 0.003, "loss": 4.0551, "step": 23094 }, { "epoch": 0.23095, "grad_norm": 1.3988893696193503, "learning_rate": 0.003, "loss": 4.0453, "step": 23095 }, { "epoch": 0.23096, "grad_norm": 1.1029080686008839, "learning_rate": 0.003, "loss": 4.0401, "step": 23096 }, { "epoch": 0.23097, "grad_norm": 1.4107123479155623, "learning_rate": 0.003, "loss": 4.0392, "step": 23097 }, { "epoch": 0.23098, "grad_norm": 1.343328213577928, "learning_rate": 0.003, "loss": 4.037, "step": 23098 }, { "epoch": 0.23099, "grad_norm": 1.3654904195614725, "learning_rate": 0.003, "loss": 4.0569, "step": 23099 }, { "epoch": 0.231, "grad_norm": 1.3704007991239546, "learning_rate": 0.003, "loss": 4.0628, "step": 23100 }, { "epoch": 0.23101, "grad_norm": 1.1214932744145265, "learning_rate": 0.003, "loss": 4.0605, "step": 23101 }, { "epoch": 0.23102, "grad_norm": 1.1804347778162774, "learning_rate": 0.003, "loss": 4.0439, "step": 23102 }, { "epoch": 0.23103, "grad_norm": 1.4653494874052122, "learning_rate": 0.003, "loss": 4.06, "step": 23103 }, { "epoch": 0.23104, "grad_norm": 1.202233869245019, "learning_rate": 0.003, "loss": 4.0475, "step": 23104 }, { "epoch": 0.23105, "grad_norm": 1.2266739737655035, "learning_rate": 0.003, "loss": 4.0631, "step": 23105 }, { "epoch": 0.23106, "grad_norm": 1.0978365678608377, "learning_rate": 0.003, "loss": 4.0456, "step": 23106 }, { "epoch": 0.23107, "grad_norm": 1.5587540284358783, "learning_rate": 0.003, "loss": 4.0755, "step": 23107 }, { "epoch": 0.23108, "grad_norm": 1.0250454122787522, "learning_rate": 0.003, "loss": 4.0, "step": 23108 }, { "epoch": 0.23109, "grad_norm": 1.400043931350464, "learning_rate": 0.003, "loss": 4.0461, "step": 23109 }, { "epoch": 0.2311, "grad_norm": 1.1424968938310303, "learning_rate": 0.003, "loss": 4.0495, "step": 23110 }, { "epoch": 0.23111, "grad_norm": 1.3105726010209262, "learning_rate": 0.003, "loss": 4.037, "step": 23111 }, { "epoch": 0.23112, "grad_norm": 1.2783236071241297, "learning_rate": 0.003, "loss": 4.0924, "step": 23112 }, { "epoch": 0.23113, "grad_norm": 1.098052421859732, "learning_rate": 0.003, "loss": 4.0383, "step": 23113 }, { "epoch": 0.23114, "grad_norm": 1.4886710941589372, "learning_rate": 0.003, "loss": 4.032, "step": 23114 }, { "epoch": 0.23115, "grad_norm": 1.2584381049217355, "learning_rate": 0.003, "loss": 4.072, "step": 23115 }, { "epoch": 0.23116, "grad_norm": 1.5741984533301971, "learning_rate": 0.003, "loss": 4.0326, "step": 23116 }, { "epoch": 0.23117, "grad_norm": 1.0653124310499962, "learning_rate": 0.003, "loss": 4.0609, "step": 23117 }, { "epoch": 0.23118, "grad_norm": 1.4895368143505685, "learning_rate": 0.003, "loss": 4.0433, "step": 23118 }, { "epoch": 0.23119, "grad_norm": 1.184337772614455, "learning_rate": 0.003, "loss": 4.0668, "step": 23119 }, { "epoch": 0.2312, "grad_norm": 1.3718758827598803, "learning_rate": 0.003, "loss": 4.0611, "step": 23120 }, { "epoch": 0.23121, "grad_norm": 1.3362888794689922, "learning_rate": 0.003, "loss": 4.0406, "step": 23121 }, { "epoch": 0.23122, "grad_norm": 1.323590297729312, "learning_rate": 0.003, "loss": 4.0476, "step": 23122 }, { "epoch": 0.23123, "grad_norm": 1.3800072954163674, "learning_rate": 0.003, "loss": 4.0539, "step": 23123 }, { "epoch": 0.23124, "grad_norm": 1.32135895563633, "learning_rate": 0.003, "loss": 4.0517, "step": 23124 }, { "epoch": 0.23125, "grad_norm": 1.3187542779296537, "learning_rate": 0.003, "loss": 4.0661, "step": 23125 }, { "epoch": 0.23126, "grad_norm": 1.155357502688108, "learning_rate": 0.003, "loss": 4.0646, "step": 23126 }, { "epoch": 0.23127, "grad_norm": 1.4408205504184937, "learning_rate": 0.003, "loss": 4.0427, "step": 23127 }, { "epoch": 0.23128, "grad_norm": 1.1680508849965598, "learning_rate": 0.003, "loss": 4.0207, "step": 23128 }, { "epoch": 0.23129, "grad_norm": 1.3068099641234334, "learning_rate": 0.003, "loss": 4.0382, "step": 23129 }, { "epoch": 0.2313, "grad_norm": 1.255595033145102, "learning_rate": 0.003, "loss": 4.0345, "step": 23130 }, { "epoch": 0.23131, "grad_norm": 1.256673276997919, "learning_rate": 0.003, "loss": 4.0517, "step": 23131 }, { "epoch": 0.23132, "grad_norm": 1.1999917681709251, "learning_rate": 0.003, "loss": 4.0296, "step": 23132 }, { "epoch": 0.23133, "grad_norm": 1.2180017020237284, "learning_rate": 0.003, "loss": 4.0397, "step": 23133 }, { "epoch": 0.23134, "grad_norm": 1.4270176461206947, "learning_rate": 0.003, "loss": 4.0391, "step": 23134 }, { "epoch": 0.23135, "grad_norm": 1.1126326837084093, "learning_rate": 0.003, "loss": 4.0553, "step": 23135 }, { "epoch": 0.23136, "grad_norm": 1.1601847464346535, "learning_rate": 0.003, "loss": 4.0085, "step": 23136 }, { "epoch": 0.23137, "grad_norm": 1.1876828983408563, "learning_rate": 0.003, "loss": 4.0265, "step": 23137 }, { "epoch": 0.23138, "grad_norm": 1.2480541814908157, "learning_rate": 0.003, "loss": 4.0153, "step": 23138 }, { "epoch": 0.23139, "grad_norm": 1.2603734939723252, "learning_rate": 0.003, "loss": 4.0784, "step": 23139 }, { "epoch": 0.2314, "grad_norm": 1.2501705941310617, "learning_rate": 0.003, "loss": 4.0227, "step": 23140 }, { "epoch": 0.23141, "grad_norm": 1.3280082885540658, "learning_rate": 0.003, "loss": 4.0413, "step": 23141 }, { "epoch": 0.23142, "grad_norm": 1.2586145728752787, "learning_rate": 0.003, "loss": 4.0331, "step": 23142 }, { "epoch": 0.23143, "grad_norm": 1.3034597497152454, "learning_rate": 0.003, "loss": 4.01, "step": 23143 }, { "epoch": 0.23144, "grad_norm": 1.4402599066569184, "learning_rate": 0.003, "loss": 4.0302, "step": 23144 }, { "epoch": 0.23145, "grad_norm": 1.69681052767605, "learning_rate": 0.003, "loss": 4.0293, "step": 23145 }, { "epoch": 0.23146, "grad_norm": 1.4345291431594693, "learning_rate": 0.003, "loss": 4.085, "step": 23146 }, { "epoch": 0.23147, "grad_norm": 1.1958868206230282, "learning_rate": 0.003, "loss": 4.0594, "step": 23147 }, { "epoch": 0.23148, "grad_norm": 1.2495945557166919, "learning_rate": 0.003, "loss": 4.0902, "step": 23148 }, { "epoch": 0.23149, "grad_norm": 1.440018602006004, "learning_rate": 0.003, "loss": 4.0529, "step": 23149 }, { "epoch": 0.2315, "grad_norm": 1.1795806910438877, "learning_rate": 0.003, "loss": 4.0511, "step": 23150 }, { "epoch": 0.23151, "grad_norm": 1.496192189070443, "learning_rate": 0.003, "loss": 4.0489, "step": 23151 }, { "epoch": 0.23152, "grad_norm": 1.1909053961441836, "learning_rate": 0.003, "loss": 4.0285, "step": 23152 }, { "epoch": 0.23153, "grad_norm": 1.3962953025431974, "learning_rate": 0.003, "loss": 4.0824, "step": 23153 }, { "epoch": 0.23154, "grad_norm": 1.1675557846803757, "learning_rate": 0.003, "loss": 4.0464, "step": 23154 }, { "epoch": 0.23155, "grad_norm": 1.3015424301448135, "learning_rate": 0.003, "loss": 4.0238, "step": 23155 }, { "epoch": 0.23156, "grad_norm": 1.5827241926282818, "learning_rate": 0.003, "loss": 4.0325, "step": 23156 }, { "epoch": 0.23157, "grad_norm": 1.3530596449290646, "learning_rate": 0.003, "loss": 4.058, "step": 23157 }, { "epoch": 0.23158, "grad_norm": 1.2588614609207367, "learning_rate": 0.003, "loss": 4.0465, "step": 23158 }, { "epoch": 0.23159, "grad_norm": 1.1243603405515727, "learning_rate": 0.003, "loss": 4.0381, "step": 23159 }, { "epoch": 0.2316, "grad_norm": 1.5716867859951131, "learning_rate": 0.003, "loss": 4.045, "step": 23160 }, { "epoch": 0.23161, "grad_norm": 1.2191364918003271, "learning_rate": 0.003, "loss": 4.0254, "step": 23161 }, { "epoch": 0.23162, "grad_norm": 1.3556083560268222, "learning_rate": 0.003, "loss": 4.0408, "step": 23162 }, { "epoch": 0.23163, "grad_norm": 1.428243446122848, "learning_rate": 0.003, "loss": 4.0785, "step": 23163 }, { "epoch": 0.23164, "grad_norm": 1.1248025309685776, "learning_rate": 0.003, "loss": 4.0684, "step": 23164 }, { "epoch": 0.23165, "grad_norm": 1.4494731533647893, "learning_rate": 0.003, "loss": 4.055, "step": 23165 }, { "epoch": 0.23166, "grad_norm": 1.1067007282731376, "learning_rate": 0.003, "loss": 4.0408, "step": 23166 }, { "epoch": 0.23167, "grad_norm": 1.634272919421246, "learning_rate": 0.003, "loss": 4.0491, "step": 23167 }, { "epoch": 0.23168, "grad_norm": 1.1041896195392928, "learning_rate": 0.003, "loss": 4.0137, "step": 23168 }, { "epoch": 0.23169, "grad_norm": 1.3448074533014545, "learning_rate": 0.003, "loss": 4.0043, "step": 23169 }, { "epoch": 0.2317, "grad_norm": 1.3057330465981378, "learning_rate": 0.003, "loss": 4.0525, "step": 23170 }, { "epoch": 0.23171, "grad_norm": 1.2471084396318384, "learning_rate": 0.003, "loss": 4.0538, "step": 23171 }, { "epoch": 0.23172, "grad_norm": 1.334405673948641, "learning_rate": 0.003, "loss": 4.045, "step": 23172 }, { "epoch": 0.23173, "grad_norm": 1.1157727143956997, "learning_rate": 0.003, "loss": 4.0335, "step": 23173 }, { "epoch": 0.23174, "grad_norm": 1.3155148133688945, "learning_rate": 0.003, "loss": 4.0357, "step": 23174 }, { "epoch": 0.23175, "grad_norm": 1.176538074429829, "learning_rate": 0.003, "loss": 4.0501, "step": 23175 }, { "epoch": 0.23176, "grad_norm": 1.2847015156755377, "learning_rate": 0.003, "loss": 4.0303, "step": 23176 }, { "epoch": 0.23177, "grad_norm": 1.4650567470933238, "learning_rate": 0.003, "loss": 4.0471, "step": 23177 }, { "epoch": 0.23178, "grad_norm": 1.4779059182089649, "learning_rate": 0.003, "loss": 4.0365, "step": 23178 }, { "epoch": 0.23179, "grad_norm": 1.1127730668606501, "learning_rate": 0.003, "loss": 4.0487, "step": 23179 }, { "epoch": 0.2318, "grad_norm": 1.6486347967805015, "learning_rate": 0.003, "loss": 4.0556, "step": 23180 }, { "epoch": 0.23181, "grad_norm": 1.0985864177337061, "learning_rate": 0.003, "loss": 4.0662, "step": 23181 }, { "epoch": 0.23182, "grad_norm": 1.2383694147162692, "learning_rate": 0.003, "loss": 4.0367, "step": 23182 }, { "epoch": 0.23183, "grad_norm": 1.3524554549919405, "learning_rate": 0.003, "loss": 4.0587, "step": 23183 }, { "epoch": 0.23184, "grad_norm": 1.205740605623693, "learning_rate": 0.003, "loss": 4.0483, "step": 23184 }, { "epoch": 0.23185, "grad_norm": 1.3614646044473162, "learning_rate": 0.003, "loss": 4.0636, "step": 23185 }, { "epoch": 0.23186, "grad_norm": 1.0473391191618198, "learning_rate": 0.003, "loss": 4.0186, "step": 23186 }, { "epoch": 0.23187, "grad_norm": 1.4147875306952638, "learning_rate": 0.003, "loss": 4.0518, "step": 23187 }, { "epoch": 0.23188, "grad_norm": 1.171074279477168, "learning_rate": 0.003, "loss": 4.0876, "step": 23188 }, { "epoch": 0.23189, "grad_norm": 1.2003817706389435, "learning_rate": 0.003, "loss": 4.0204, "step": 23189 }, { "epoch": 0.2319, "grad_norm": 1.2329943281245004, "learning_rate": 0.003, "loss": 4.0237, "step": 23190 }, { "epoch": 0.23191, "grad_norm": 1.282464087951763, "learning_rate": 0.003, "loss": 4.0509, "step": 23191 }, { "epoch": 0.23192, "grad_norm": 1.4175510640335267, "learning_rate": 0.003, "loss": 4.0682, "step": 23192 }, { "epoch": 0.23193, "grad_norm": 1.2231555605608266, "learning_rate": 0.003, "loss": 4.0478, "step": 23193 }, { "epoch": 0.23194, "grad_norm": 1.464929973426186, "learning_rate": 0.003, "loss": 4.0106, "step": 23194 }, { "epoch": 0.23195, "grad_norm": 1.2891837757244058, "learning_rate": 0.003, "loss": 4.0651, "step": 23195 }, { "epoch": 0.23196, "grad_norm": 1.284554852519397, "learning_rate": 0.003, "loss": 4.0682, "step": 23196 }, { "epoch": 0.23197, "grad_norm": 1.1835775203695522, "learning_rate": 0.003, "loss": 4.0475, "step": 23197 }, { "epoch": 0.23198, "grad_norm": 1.3405997143577533, "learning_rate": 0.003, "loss": 4.0266, "step": 23198 }, { "epoch": 0.23199, "grad_norm": 1.3221228242110958, "learning_rate": 0.003, "loss": 4.0535, "step": 23199 }, { "epoch": 0.232, "grad_norm": 1.2650841488173297, "learning_rate": 0.003, "loss": 4.0458, "step": 23200 }, { "epoch": 0.23201, "grad_norm": 1.2434903559984762, "learning_rate": 0.003, "loss": 4.0387, "step": 23201 }, { "epoch": 0.23202, "grad_norm": 1.2103808769700988, "learning_rate": 0.003, "loss": 4.0465, "step": 23202 }, { "epoch": 0.23203, "grad_norm": 1.2757788894125792, "learning_rate": 0.003, "loss": 4.0575, "step": 23203 }, { "epoch": 0.23204, "grad_norm": 1.2843571037396353, "learning_rate": 0.003, "loss": 4.0321, "step": 23204 }, { "epoch": 0.23205, "grad_norm": 1.3123870931340436, "learning_rate": 0.003, "loss": 4.0531, "step": 23205 }, { "epoch": 0.23206, "grad_norm": 1.1504312830448848, "learning_rate": 0.003, "loss": 4.0433, "step": 23206 }, { "epoch": 0.23207, "grad_norm": 1.4494120792756744, "learning_rate": 0.003, "loss": 4.0694, "step": 23207 }, { "epoch": 0.23208, "grad_norm": 1.2321102679461702, "learning_rate": 0.003, "loss": 4.057, "step": 23208 }, { "epoch": 0.23209, "grad_norm": 1.4467505845454234, "learning_rate": 0.003, "loss": 4.0292, "step": 23209 }, { "epoch": 0.2321, "grad_norm": 1.5921456379945131, "learning_rate": 0.003, "loss": 4.0546, "step": 23210 }, { "epoch": 0.23211, "grad_norm": 1.25486819305933, "learning_rate": 0.003, "loss": 4.0633, "step": 23211 }, { "epoch": 0.23212, "grad_norm": 1.318159553687812, "learning_rate": 0.003, "loss": 4.041, "step": 23212 }, { "epoch": 0.23213, "grad_norm": 1.3522028549171818, "learning_rate": 0.003, "loss": 4.0752, "step": 23213 }, { "epoch": 0.23214, "grad_norm": 1.2425751876346849, "learning_rate": 0.003, "loss": 4.0206, "step": 23214 }, { "epoch": 0.23215, "grad_norm": 1.0371338136696169, "learning_rate": 0.003, "loss": 4.0349, "step": 23215 }, { "epoch": 0.23216, "grad_norm": 1.1216127249129784, "learning_rate": 0.003, "loss": 4.0173, "step": 23216 }, { "epoch": 0.23217, "grad_norm": 1.3631832004698994, "learning_rate": 0.003, "loss": 4.0474, "step": 23217 }, { "epoch": 0.23218, "grad_norm": 1.4806873039900939, "learning_rate": 0.003, "loss": 4.0321, "step": 23218 }, { "epoch": 0.23219, "grad_norm": 1.3258900547927888, "learning_rate": 0.003, "loss": 4.0485, "step": 23219 }, { "epoch": 0.2322, "grad_norm": 1.325757282319854, "learning_rate": 0.003, "loss": 4.056, "step": 23220 }, { "epoch": 0.23221, "grad_norm": 1.3677893861856718, "learning_rate": 0.003, "loss": 4.0341, "step": 23221 }, { "epoch": 0.23222, "grad_norm": 1.2091666637750367, "learning_rate": 0.003, "loss": 4.0179, "step": 23222 }, { "epoch": 0.23223, "grad_norm": 1.407492764753761, "learning_rate": 0.003, "loss": 4.0425, "step": 23223 }, { "epoch": 0.23224, "grad_norm": 0.9543968454000411, "learning_rate": 0.003, "loss": 4.0344, "step": 23224 }, { "epoch": 0.23225, "grad_norm": 1.424205440931275, "learning_rate": 0.003, "loss": 4.0625, "step": 23225 }, { "epoch": 0.23226, "grad_norm": 1.1789994954770877, "learning_rate": 0.003, "loss": 4.0222, "step": 23226 }, { "epoch": 0.23227, "grad_norm": 1.500989460201003, "learning_rate": 0.003, "loss": 4.0525, "step": 23227 }, { "epoch": 0.23228, "grad_norm": 1.0752262785965898, "learning_rate": 0.003, "loss": 4.0583, "step": 23228 }, { "epoch": 0.23229, "grad_norm": 1.2616444613266276, "learning_rate": 0.003, "loss": 4.0356, "step": 23229 }, { "epoch": 0.2323, "grad_norm": 1.371139188479844, "learning_rate": 0.003, "loss": 4.0549, "step": 23230 }, { "epoch": 0.23231, "grad_norm": 1.2715010105964422, "learning_rate": 0.003, "loss": 4.0142, "step": 23231 }, { "epoch": 0.23232, "grad_norm": 1.128084975286615, "learning_rate": 0.003, "loss": 4.0525, "step": 23232 }, { "epoch": 0.23233, "grad_norm": 1.3631250374049788, "learning_rate": 0.003, "loss": 4.0606, "step": 23233 }, { "epoch": 0.23234, "grad_norm": 1.2555405702095657, "learning_rate": 0.003, "loss": 4.0481, "step": 23234 }, { "epoch": 0.23235, "grad_norm": 1.5995130541276383, "learning_rate": 0.003, "loss": 4.0323, "step": 23235 }, { "epoch": 0.23236, "grad_norm": 1.262016998384145, "learning_rate": 0.003, "loss": 4.0139, "step": 23236 }, { "epoch": 0.23237, "grad_norm": 1.366410789894868, "learning_rate": 0.003, "loss": 4.0421, "step": 23237 }, { "epoch": 0.23238, "grad_norm": 1.234722136798343, "learning_rate": 0.003, "loss": 4.063, "step": 23238 }, { "epoch": 0.23239, "grad_norm": 1.4539394754451245, "learning_rate": 0.003, "loss": 4.0817, "step": 23239 }, { "epoch": 0.2324, "grad_norm": 1.1737704757671368, "learning_rate": 0.003, "loss": 4.0455, "step": 23240 }, { "epoch": 0.23241, "grad_norm": 1.6214539718520233, "learning_rate": 0.003, "loss": 4.0536, "step": 23241 }, { "epoch": 0.23242, "grad_norm": 1.088284562816423, "learning_rate": 0.003, "loss": 4.0507, "step": 23242 }, { "epoch": 0.23243, "grad_norm": 1.5156893674451846, "learning_rate": 0.003, "loss": 4.0652, "step": 23243 }, { "epoch": 0.23244, "grad_norm": 1.1443822468829565, "learning_rate": 0.003, "loss": 4.0443, "step": 23244 }, { "epoch": 0.23245, "grad_norm": 1.3462592350267242, "learning_rate": 0.003, "loss": 4.0468, "step": 23245 }, { "epoch": 0.23246, "grad_norm": 1.2110783221956347, "learning_rate": 0.003, "loss": 4.0693, "step": 23246 }, { "epoch": 0.23247, "grad_norm": 1.3039759113548803, "learning_rate": 0.003, "loss": 4.0484, "step": 23247 }, { "epoch": 0.23248, "grad_norm": 1.07813170533897, "learning_rate": 0.003, "loss": 4.0556, "step": 23248 }, { "epoch": 0.23249, "grad_norm": 1.848094974981773, "learning_rate": 0.003, "loss": 4.0718, "step": 23249 }, { "epoch": 0.2325, "grad_norm": 1.071732617437444, "learning_rate": 0.003, "loss": 4.0923, "step": 23250 }, { "epoch": 0.23251, "grad_norm": 1.6644558775451108, "learning_rate": 0.003, "loss": 4.072, "step": 23251 }, { "epoch": 0.23252, "grad_norm": 1.186053105959287, "learning_rate": 0.003, "loss": 4.0558, "step": 23252 }, { "epoch": 0.23253, "grad_norm": 1.2267337511924554, "learning_rate": 0.003, "loss": 4.0431, "step": 23253 }, { "epoch": 0.23254, "grad_norm": 1.2007454933825648, "learning_rate": 0.003, "loss": 4.0604, "step": 23254 }, { "epoch": 0.23255, "grad_norm": 1.3137764399401184, "learning_rate": 0.003, "loss": 4.0543, "step": 23255 }, { "epoch": 0.23256, "grad_norm": 1.1356408678775007, "learning_rate": 0.003, "loss": 4.0385, "step": 23256 }, { "epoch": 0.23257, "grad_norm": 1.5515651222082167, "learning_rate": 0.003, "loss": 4.0441, "step": 23257 }, { "epoch": 0.23258, "grad_norm": 1.3651398703632192, "learning_rate": 0.003, "loss": 4.0117, "step": 23258 }, { "epoch": 0.23259, "grad_norm": 1.4503685628261243, "learning_rate": 0.003, "loss": 4.0392, "step": 23259 }, { "epoch": 0.2326, "grad_norm": 1.1816761800865787, "learning_rate": 0.003, "loss": 4.0702, "step": 23260 }, { "epoch": 0.23261, "grad_norm": 1.371236268002453, "learning_rate": 0.003, "loss": 4.0437, "step": 23261 }, { "epoch": 0.23262, "grad_norm": 1.2417484239071042, "learning_rate": 0.003, "loss": 4.022, "step": 23262 }, { "epoch": 0.23263, "grad_norm": 1.3351348679475814, "learning_rate": 0.003, "loss": 4.0493, "step": 23263 }, { "epoch": 0.23264, "grad_norm": 1.1375114553965955, "learning_rate": 0.003, "loss": 4.0403, "step": 23264 }, { "epoch": 0.23265, "grad_norm": 1.152906481026978, "learning_rate": 0.003, "loss": 4.0556, "step": 23265 }, { "epoch": 0.23266, "grad_norm": 1.392860597586105, "learning_rate": 0.003, "loss": 4.0406, "step": 23266 }, { "epoch": 0.23267, "grad_norm": 1.1876590651180388, "learning_rate": 0.003, "loss": 4.0446, "step": 23267 }, { "epoch": 0.23268, "grad_norm": 1.2643494790083054, "learning_rate": 0.003, "loss": 4.0383, "step": 23268 }, { "epoch": 0.23269, "grad_norm": 1.2981487311165842, "learning_rate": 0.003, "loss": 4.0554, "step": 23269 }, { "epoch": 0.2327, "grad_norm": 1.37569174521526, "learning_rate": 0.003, "loss": 4.0572, "step": 23270 }, { "epoch": 0.23271, "grad_norm": 1.3189397526707127, "learning_rate": 0.003, "loss": 4.0331, "step": 23271 }, { "epoch": 0.23272, "grad_norm": 1.3527989469943307, "learning_rate": 0.003, "loss": 4.0641, "step": 23272 }, { "epoch": 0.23273, "grad_norm": 1.2582985951640966, "learning_rate": 0.003, "loss": 4.0498, "step": 23273 }, { "epoch": 0.23274, "grad_norm": 1.2558499925607352, "learning_rate": 0.003, "loss": 4.0256, "step": 23274 }, { "epoch": 0.23275, "grad_norm": 1.4347250406858223, "learning_rate": 0.003, "loss": 4.0475, "step": 23275 }, { "epoch": 0.23276, "grad_norm": 1.2249077678685385, "learning_rate": 0.003, "loss": 4.0722, "step": 23276 }, { "epoch": 0.23277, "grad_norm": 1.6413836240818787, "learning_rate": 0.003, "loss": 4.0542, "step": 23277 }, { "epoch": 0.23278, "grad_norm": 1.2936745852776543, "learning_rate": 0.003, "loss": 4.0448, "step": 23278 }, { "epoch": 0.23279, "grad_norm": 1.2424349656014362, "learning_rate": 0.003, "loss": 4.0403, "step": 23279 }, { "epoch": 0.2328, "grad_norm": 1.355113918644518, "learning_rate": 0.003, "loss": 4.0747, "step": 23280 }, { "epoch": 0.23281, "grad_norm": 1.2320748259439178, "learning_rate": 0.003, "loss": 4.0463, "step": 23281 }, { "epoch": 0.23282, "grad_norm": 1.5111517506175998, "learning_rate": 0.003, "loss": 4.0392, "step": 23282 }, { "epoch": 0.23283, "grad_norm": 1.276988340170519, "learning_rate": 0.003, "loss": 4.0366, "step": 23283 }, { "epoch": 0.23284, "grad_norm": 1.1501325406200669, "learning_rate": 0.003, "loss": 4.0369, "step": 23284 }, { "epoch": 0.23285, "grad_norm": 1.5871858862380583, "learning_rate": 0.003, "loss": 4.0566, "step": 23285 }, { "epoch": 0.23286, "grad_norm": 0.9997980241356003, "learning_rate": 0.003, "loss": 4.0136, "step": 23286 }, { "epoch": 0.23287, "grad_norm": 1.5871920968229956, "learning_rate": 0.003, "loss": 4.0454, "step": 23287 }, { "epoch": 0.23288, "grad_norm": 1.0883272856605712, "learning_rate": 0.003, "loss": 4.0577, "step": 23288 }, { "epoch": 0.23289, "grad_norm": 1.3025211560375816, "learning_rate": 0.003, "loss": 4.0391, "step": 23289 }, { "epoch": 0.2329, "grad_norm": 1.2242693005816543, "learning_rate": 0.003, "loss": 4.077, "step": 23290 }, { "epoch": 0.23291, "grad_norm": 1.3148277382310716, "learning_rate": 0.003, "loss": 4.0587, "step": 23291 }, { "epoch": 0.23292, "grad_norm": 1.1256728129228435, "learning_rate": 0.003, "loss": 4.0484, "step": 23292 }, { "epoch": 0.23293, "grad_norm": 1.524326752503377, "learning_rate": 0.003, "loss": 4.0491, "step": 23293 }, { "epoch": 0.23294, "grad_norm": 1.3728702520680502, "learning_rate": 0.003, "loss": 4.0533, "step": 23294 }, { "epoch": 0.23295, "grad_norm": 1.3817397381305565, "learning_rate": 0.003, "loss": 4.0629, "step": 23295 }, { "epoch": 0.23296, "grad_norm": 1.0052534286348047, "learning_rate": 0.003, "loss": 4.0172, "step": 23296 }, { "epoch": 0.23297, "grad_norm": 1.587897522327039, "learning_rate": 0.003, "loss": 4.0505, "step": 23297 }, { "epoch": 0.23298, "grad_norm": 1.0623389759525195, "learning_rate": 0.003, "loss": 4.046, "step": 23298 }, { "epoch": 0.23299, "grad_norm": 1.3452812993470136, "learning_rate": 0.003, "loss": 4.0555, "step": 23299 }, { "epoch": 0.233, "grad_norm": 1.112843801250024, "learning_rate": 0.003, "loss": 4.0821, "step": 23300 }, { "epoch": 0.23301, "grad_norm": 1.2807395161155217, "learning_rate": 0.003, "loss": 4.0454, "step": 23301 }, { "epoch": 0.23302, "grad_norm": 1.2393823980255212, "learning_rate": 0.003, "loss": 4.0577, "step": 23302 }, { "epoch": 0.23303, "grad_norm": 1.177388458780208, "learning_rate": 0.003, "loss": 3.9993, "step": 23303 }, { "epoch": 0.23304, "grad_norm": 1.4993408883728558, "learning_rate": 0.003, "loss": 4.0595, "step": 23304 }, { "epoch": 0.23305, "grad_norm": 1.3478072590870311, "learning_rate": 0.003, "loss": 4.0537, "step": 23305 }, { "epoch": 0.23306, "grad_norm": 1.2621195994423227, "learning_rate": 0.003, "loss": 4.0419, "step": 23306 }, { "epoch": 0.23307, "grad_norm": 1.2399002505059544, "learning_rate": 0.003, "loss": 4.0503, "step": 23307 }, { "epoch": 0.23308, "grad_norm": 1.1400015294578412, "learning_rate": 0.003, "loss": 4.0206, "step": 23308 }, { "epoch": 0.23309, "grad_norm": 1.4750528418226974, "learning_rate": 0.003, "loss": 4.0603, "step": 23309 }, { "epoch": 0.2331, "grad_norm": 1.1650374850103553, "learning_rate": 0.003, "loss": 4.0497, "step": 23310 }, { "epoch": 0.23311, "grad_norm": 1.426177695369418, "learning_rate": 0.003, "loss": 4.038, "step": 23311 }, { "epoch": 0.23312, "grad_norm": 1.3577724159397382, "learning_rate": 0.003, "loss": 4.0456, "step": 23312 }, { "epoch": 0.23313, "grad_norm": 1.4658900785144333, "learning_rate": 0.003, "loss": 4.0408, "step": 23313 }, { "epoch": 0.23314, "grad_norm": 1.2162271156431093, "learning_rate": 0.003, "loss": 4.067, "step": 23314 }, { "epoch": 0.23315, "grad_norm": 1.5732571181404769, "learning_rate": 0.003, "loss": 4.0336, "step": 23315 }, { "epoch": 0.23316, "grad_norm": 1.2566505485428612, "learning_rate": 0.003, "loss": 4.0664, "step": 23316 }, { "epoch": 0.23317, "grad_norm": 1.4479241168886905, "learning_rate": 0.003, "loss": 4.083, "step": 23317 }, { "epoch": 0.23318, "grad_norm": 1.204350903825152, "learning_rate": 0.003, "loss": 4.0486, "step": 23318 }, { "epoch": 0.23319, "grad_norm": 1.2414904171838381, "learning_rate": 0.003, "loss": 4.0272, "step": 23319 }, { "epoch": 0.2332, "grad_norm": 1.2229429594840069, "learning_rate": 0.003, "loss": 4.0518, "step": 23320 }, { "epoch": 0.23321, "grad_norm": 1.3985446855655157, "learning_rate": 0.003, "loss": 4.0535, "step": 23321 }, { "epoch": 0.23322, "grad_norm": 1.2873223423469282, "learning_rate": 0.003, "loss": 4.0352, "step": 23322 }, { "epoch": 0.23323, "grad_norm": 1.2732889145538597, "learning_rate": 0.003, "loss": 4.0484, "step": 23323 }, { "epoch": 0.23324, "grad_norm": 1.3982124959311553, "learning_rate": 0.003, "loss": 4.0663, "step": 23324 }, { "epoch": 0.23325, "grad_norm": 1.2716832460961498, "learning_rate": 0.003, "loss": 4.0514, "step": 23325 }, { "epoch": 0.23326, "grad_norm": 1.3899609151312846, "learning_rate": 0.003, "loss": 4.0503, "step": 23326 }, { "epoch": 0.23327, "grad_norm": 1.271183161385945, "learning_rate": 0.003, "loss": 4.0061, "step": 23327 }, { "epoch": 0.23328, "grad_norm": 1.2059489982588343, "learning_rate": 0.003, "loss": 4.0374, "step": 23328 }, { "epoch": 0.23329, "grad_norm": 1.4973451702797238, "learning_rate": 0.003, "loss": 4.038, "step": 23329 }, { "epoch": 0.2333, "grad_norm": 1.336286258154857, "learning_rate": 0.003, "loss": 4.0271, "step": 23330 }, { "epoch": 0.23331, "grad_norm": 1.3218764464893902, "learning_rate": 0.003, "loss": 4.0486, "step": 23331 }, { "epoch": 0.23332, "grad_norm": 1.189667149685002, "learning_rate": 0.003, "loss": 4.0161, "step": 23332 }, { "epoch": 0.23333, "grad_norm": 1.2322615568078854, "learning_rate": 0.003, "loss": 4.0213, "step": 23333 }, { "epoch": 0.23334, "grad_norm": 1.3163948969762989, "learning_rate": 0.003, "loss": 4.0235, "step": 23334 }, { "epoch": 0.23335, "grad_norm": 1.186609084019781, "learning_rate": 0.003, "loss": 4.0437, "step": 23335 }, { "epoch": 0.23336, "grad_norm": 1.5092553303414036, "learning_rate": 0.003, "loss": 4.036, "step": 23336 }, { "epoch": 0.23337, "grad_norm": 0.9346544544768131, "learning_rate": 0.003, "loss": 4.0352, "step": 23337 }, { "epoch": 0.23338, "grad_norm": 1.4567871344287167, "learning_rate": 0.003, "loss": 4.0489, "step": 23338 }, { "epoch": 0.23339, "grad_norm": 1.2192066288744325, "learning_rate": 0.003, "loss": 4.0464, "step": 23339 }, { "epoch": 0.2334, "grad_norm": 1.4134711947502165, "learning_rate": 0.003, "loss": 4.0603, "step": 23340 }, { "epoch": 0.23341, "grad_norm": 1.0416974593078356, "learning_rate": 0.003, "loss": 4.0483, "step": 23341 }, { "epoch": 0.23342, "grad_norm": 1.4811986235247379, "learning_rate": 0.003, "loss": 4.0312, "step": 23342 }, { "epoch": 0.23343, "grad_norm": 1.2937531192280796, "learning_rate": 0.003, "loss": 4.0579, "step": 23343 }, { "epoch": 0.23344, "grad_norm": 1.2491946624829435, "learning_rate": 0.003, "loss": 4.0398, "step": 23344 }, { "epoch": 0.23345, "grad_norm": 1.4541947258751813, "learning_rate": 0.003, "loss": 4.0417, "step": 23345 }, { "epoch": 0.23346, "grad_norm": 1.3063125324576894, "learning_rate": 0.003, "loss": 4.0663, "step": 23346 }, { "epoch": 0.23347, "grad_norm": 1.2654606405347482, "learning_rate": 0.003, "loss": 3.9985, "step": 23347 }, { "epoch": 0.23348, "grad_norm": 1.352898458264151, "learning_rate": 0.003, "loss": 4.0649, "step": 23348 }, { "epoch": 0.23349, "grad_norm": 1.419736958452378, "learning_rate": 0.003, "loss": 4.0085, "step": 23349 }, { "epoch": 0.2335, "grad_norm": 1.2239250662377954, "learning_rate": 0.003, "loss": 4.0184, "step": 23350 }, { "epoch": 0.23351, "grad_norm": 1.4556471497734236, "learning_rate": 0.003, "loss": 4.0199, "step": 23351 }, { "epoch": 0.23352, "grad_norm": 1.2775522244513904, "learning_rate": 0.003, "loss": 4.0743, "step": 23352 }, { "epoch": 0.23353, "grad_norm": 1.4390638945032457, "learning_rate": 0.003, "loss": 4.052, "step": 23353 }, { "epoch": 0.23354, "grad_norm": 1.2013876104286578, "learning_rate": 0.003, "loss": 4.066, "step": 23354 }, { "epoch": 0.23355, "grad_norm": 1.460624431573545, "learning_rate": 0.003, "loss": 4.0231, "step": 23355 }, { "epoch": 0.23356, "grad_norm": 1.1786276934316429, "learning_rate": 0.003, "loss": 4.0549, "step": 23356 }, { "epoch": 0.23357, "grad_norm": 1.446583890113212, "learning_rate": 0.003, "loss": 4.0657, "step": 23357 }, { "epoch": 0.23358, "grad_norm": 1.3394048256342341, "learning_rate": 0.003, "loss": 4.0401, "step": 23358 }, { "epoch": 0.23359, "grad_norm": 1.4911245231435557, "learning_rate": 0.003, "loss": 4.045, "step": 23359 }, { "epoch": 0.2336, "grad_norm": 1.3385068391904777, "learning_rate": 0.003, "loss": 4.0612, "step": 23360 }, { "epoch": 0.23361, "grad_norm": 1.245599564200335, "learning_rate": 0.003, "loss": 4.0698, "step": 23361 }, { "epoch": 0.23362, "grad_norm": 1.0990189591725963, "learning_rate": 0.003, "loss": 4.0613, "step": 23362 }, { "epoch": 0.23363, "grad_norm": 1.26945861378687, "learning_rate": 0.003, "loss": 4.0331, "step": 23363 }, { "epoch": 0.23364, "grad_norm": 1.438677382730055, "learning_rate": 0.003, "loss": 4.0668, "step": 23364 }, { "epoch": 0.23365, "grad_norm": 1.1688671630588279, "learning_rate": 0.003, "loss": 4.039, "step": 23365 }, { "epoch": 0.23366, "grad_norm": 1.2190584746599422, "learning_rate": 0.003, "loss": 4.0162, "step": 23366 }, { "epoch": 0.23367, "grad_norm": 1.3516268608402984, "learning_rate": 0.003, "loss": 4.0717, "step": 23367 }, { "epoch": 0.23368, "grad_norm": 1.0832683915887389, "learning_rate": 0.003, "loss": 4.0301, "step": 23368 }, { "epoch": 0.23369, "grad_norm": 1.388129636200871, "learning_rate": 0.003, "loss": 4.0192, "step": 23369 }, { "epoch": 0.2337, "grad_norm": 1.249431964971318, "learning_rate": 0.003, "loss": 4.0366, "step": 23370 }, { "epoch": 0.23371, "grad_norm": 1.4278674904087199, "learning_rate": 0.003, "loss": 4.0545, "step": 23371 }, { "epoch": 0.23372, "grad_norm": 1.2706777979832753, "learning_rate": 0.003, "loss": 4.0425, "step": 23372 }, { "epoch": 0.23373, "grad_norm": 1.4755016394086917, "learning_rate": 0.003, "loss": 4.0449, "step": 23373 }, { "epoch": 0.23374, "grad_norm": 1.4776964671313162, "learning_rate": 0.003, "loss": 4.0301, "step": 23374 }, { "epoch": 0.23375, "grad_norm": 1.1164524099187323, "learning_rate": 0.003, "loss": 4.0258, "step": 23375 }, { "epoch": 0.23376, "grad_norm": 1.4127839087747904, "learning_rate": 0.003, "loss": 4.0285, "step": 23376 }, { "epoch": 0.23377, "grad_norm": 1.6199554185054557, "learning_rate": 0.003, "loss": 4.0583, "step": 23377 }, { "epoch": 0.23378, "grad_norm": 1.0902052277350867, "learning_rate": 0.003, "loss": 4.0639, "step": 23378 }, { "epoch": 0.23379, "grad_norm": 1.3211830886924998, "learning_rate": 0.003, "loss": 4.0361, "step": 23379 }, { "epoch": 0.2338, "grad_norm": 1.2891991985832214, "learning_rate": 0.003, "loss": 4.0667, "step": 23380 }, { "epoch": 0.23381, "grad_norm": 1.1498749882727495, "learning_rate": 0.003, "loss": 4.0517, "step": 23381 }, { "epoch": 0.23382, "grad_norm": 1.2456215243805242, "learning_rate": 0.003, "loss": 4.0541, "step": 23382 }, { "epoch": 0.23383, "grad_norm": 1.281259435831229, "learning_rate": 0.003, "loss": 4.0325, "step": 23383 }, { "epoch": 0.23384, "grad_norm": 1.4055223207724385, "learning_rate": 0.003, "loss": 4.025, "step": 23384 }, { "epoch": 0.23385, "grad_norm": 1.4293121693887338, "learning_rate": 0.003, "loss": 4.0407, "step": 23385 }, { "epoch": 0.23386, "grad_norm": 1.5747444673804685, "learning_rate": 0.003, "loss": 4.0378, "step": 23386 }, { "epoch": 0.23387, "grad_norm": 1.1126669344311293, "learning_rate": 0.003, "loss": 4.0379, "step": 23387 }, { "epoch": 0.23388, "grad_norm": 1.3099554552952772, "learning_rate": 0.003, "loss": 4.0389, "step": 23388 }, { "epoch": 0.23389, "grad_norm": 1.1230297737483306, "learning_rate": 0.003, "loss": 4.0328, "step": 23389 }, { "epoch": 0.2339, "grad_norm": 1.6483998893832372, "learning_rate": 0.003, "loss": 4.0511, "step": 23390 }, { "epoch": 0.23391, "grad_norm": 0.9630058891032882, "learning_rate": 0.003, "loss": 4.0542, "step": 23391 }, { "epoch": 0.23392, "grad_norm": 1.4841588187258903, "learning_rate": 0.003, "loss": 4.0846, "step": 23392 }, { "epoch": 0.23393, "grad_norm": 1.4368091500267326, "learning_rate": 0.003, "loss": 4.0932, "step": 23393 }, { "epoch": 0.23394, "grad_norm": 1.224397933931255, "learning_rate": 0.003, "loss": 4.0506, "step": 23394 }, { "epoch": 0.23395, "grad_norm": 1.4468463742130788, "learning_rate": 0.003, "loss": 4.0694, "step": 23395 }, { "epoch": 0.23396, "grad_norm": 1.0249706697937349, "learning_rate": 0.003, "loss": 4.0441, "step": 23396 }, { "epoch": 0.23397, "grad_norm": 1.3713755836938135, "learning_rate": 0.003, "loss": 4.0516, "step": 23397 }, { "epoch": 0.23398, "grad_norm": 1.1611916618142029, "learning_rate": 0.003, "loss": 4.0528, "step": 23398 }, { "epoch": 0.23399, "grad_norm": 1.4914468952398496, "learning_rate": 0.003, "loss": 4.0823, "step": 23399 }, { "epoch": 0.234, "grad_norm": 1.264474854899204, "learning_rate": 0.003, "loss": 4.0601, "step": 23400 }, { "epoch": 0.23401, "grad_norm": 1.1446779052357057, "learning_rate": 0.003, "loss": 4.0401, "step": 23401 }, { "epoch": 0.23402, "grad_norm": 1.4639520243379562, "learning_rate": 0.003, "loss": 4.0755, "step": 23402 }, { "epoch": 0.23403, "grad_norm": 1.3668044246529565, "learning_rate": 0.003, "loss": 4.0516, "step": 23403 }, { "epoch": 0.23404, "grad_norm": 1.5335095337809348, "learning_rate": 0.003, "loss": 4.0241, "step": 23404 }, { "epoch": 0.23405, "grad_norm": 1.4608038413972224, "learning_rate": 0.003, "loss": 4.0408, "step": 23405 }, { "epoch": 0.23406, "grad_norm": 1.1831709899512708, "learning_rate": 0.003, "loss": 4.0501, "step": 23406 }, { "epoch": 0.23407, "grad_norm": 1.5170908258570104, "learning_rate": 0.003, "loss": 4.0566, "step": 23407 }, { "epoch": 0.23408, "grad_norm": 1.2951979560311104, "learning_rate": 0.003, "loss": 4.0337, "step": 23408 }, { "epoch": 0.23409, "grad_norm": 1.4367463373648688, "learning_rate": 0.003, "loss": 4.0295, "step": 23409 }, { "epoch": 0.2341, "grad_norm": 1.123484772307956, "learning_rate": 0.003, "loss": 4.0505, "step": 23410 }, { "epoch": 0.23411, "grad_norm": 1.4166871511080863, "learning_rate": 0.003, "loss": 4.042, "step": 23411 }, { "epoch": 0.23412, "grad_norm": 1.270901708599592, "learning_rate": 0.003, "loss": 4.037, "step": 23412 }, { "epoch": 0.23413, "grad_norm": 1.2685136248395315, "learning_rate": 0.003, "loss": 4.0457, "step": 23413 }, { "epoch": 0.23414, "grad_norm": 1.44777247024794, "learning_rate": 0.003, "loss": 4.0678, "step": 23414 }, { "epoch": 0.23415, "grad_norm": 1.1173661883047519, "learning_rate": 0.003, "loss": 4.049, "step": 23415 }, { "epoch": 0.23416, "grad_norm": 1.3467413258436467, "learning_rate": 0.003, "loss": 4.069, "step": 23416 }, { "epoch": 0.23417, "grad_norm": 1.1930537002459254, "learning_rate": 0.003, "loss": 4.0317, "step": 23417 }, { "epoch": 0.23418, "grad_norm": 1.5650903952247919, "learning_rate": 0.003, "loss": 4.0611, "step": 23418 }, { "epoch": 0.23419, "grad_norm": 1.266998349523785, "learning_rate": 0.003, "loss": 4.0442, "step": 23419 }, { "epoch": 0.2342, "grad_norm": 1.2893704775377108, "learning_rate": 0.003, "loss": 4.0706, "step": 23420 }, { "epoch": 0.23421, "grad_norm": 1.1791439191452007, "learning_rate": 0.003, "loss": 4.041, "step": 23421 }, { "epoch": 0.23422, "grad_norm": 1.4245509050405656, "learning_rate": 0.003, "loss": 4.0355, "step": 23422 }, { "epoch": 0.23423, "grad_norm": 1.307165228171369, "learning_rate": 0.003, "loss": 4.0435, "step": 23423 }, { "epoch": 0.23424, "grad_norm": 1.2025462135710496, "learning_rate": 0.003, "loss": 4.0669, "step": 23424 }, { "epoch": 0.23425, "grad_norm": 1.1276283362529052, "learning_rate": 0.003, "loss": 4.0599, "step": 23425 }, { "epoch": 0.23426, "grad_norm": 1.6340636956456778, "learning_rate": 0.003, "loss": 4.0519, "step": 23426 }, { "epoch": 0.23427, "grad_norm": 1.0473584773050157, "learning_rate": 0.003, "loss": 4.0403, "step": 23427 }, { "epoch": 0.23428, "grad_norm": 1.41795237037583, "learning_rate": 0.003, "loss": 4.0191, "step": 23428 }, { "epoch": 0.23429, "grad_norm": 1.251689438580586, "learning_rate": 0.003, "loss": 4.0611, "step": 23429 }, { "epoch": 0.2343, "grad_norm": 1.24312453461535, "learning_rate": 0.003, "loss": 4.0204, "step": 23430 }, { "epoch": 0.23431, "grad_norm": 1.5292958131803067, "learning_rate": 0.003, "loss": 4.0577, "step": 23431 }, { "epoch": 0.23432, "grad_norm": 1.0026100040387071, "learning_rate": 0.003, "loss": 4.0308, "step": 23432 }, { "epoch": 0.23433, "grad_norm": 1.4850961014902797, "learning_rate": 0.003, "loss": 4.0595, "step": 23433 }, { "epoch": 0.23434, "grad_norm": 1.312183879029588, "learning_rate": 0.003, "loss": 4.0441, "step": 23434 }, { "epoch": 0.23435, "grad_norm": 1.3949258439329408, "learning_rate": 0.003, "loss": 4.0514, "step": 23435 }, { "epoch": 0.23436, "grad_norm": 1.0606214454185012, "learning_rate": 0.003, "loss": 4.0219, "step": 23436 }, { "epoch": 0.23437, "grad_norm": 1.309998350683547, "learning_rate": 0.003, "loss": 4.0397, "step": 23437 }, { "epoch": 0.23438, "grad_norm": 1.6471810202782498, "learning_rate": 0.003, "loss": 4.1063, "step": 23438 }, { "epoch": 0.23439, "grad_norm": 0.8944518120383201, "learning_rate": 0.003, "loss": 4.0424, "step": 23439 }, { "epoch": 0.2344, "grad_norm": 1.4210544849423057, "learning_rate": 0.003, "loss": 4.0664, "step": 23440 }, { "epoch": 0.23441, "grad_norm": 1.2186446707547596, "learning_rate": 0.003, "loss": 4.0586, "step": 23441 }, { "epoch": 0.23442, "grad_norm": 1.494003855319048, "learning_rate": 0.003, "loss": 4.0813, "step": 23442 }, { "epoch": 0.23443, "grad_norm": 1.054176325159872, "learning_rate": 0.003, "loss": 4.0459, "step": 23443 }, { "epoch": 0.23444, "grad_norm": 1.2423595007967576, "learning_rate": 0.003, "loss": 4.0255, "step": 23444 }, { "epoch": 0.23445, "grad_norm": 1.2816517990823693, "learning_rate": 0.003, "loss": 4.0639, "step": 23445 }, { "epoch": 0.23446, "grad_norm": 1.1626938469678996, "learning_rate": 0.003, "loss": 4.0734, "step": 23446 }, { "epoch": 0.23447, "grad_norm": 1.4096546631106326, "learning_rate": 0.003, "loss": 4.0598, "step": 23447 }, { "epoch": 0.23448, "grad_norm": 1.3479844981950733, "learning_rate": 0.003, "loss": 4.0236, "step": 23448 }, { "epoch": 0.23449, "grad_norm": 1.2963676549501473, "learning_rate": 0.003, "loss": 4.0753, "step": 23449 }, { "epoch": 0.2345, "grad_norm": 1.4253073749545635, "learning_rate": 0.003, "loss": 4.0573, "step": 23450 }, { "epoch": 0.23451, "grad_norm": 1.4171133991274112, "learning_rate": 0.003, "loss": 4.0572, "step": 23451 }, { "epoch": 0.23452, "grad_norm": 1.1785833076166021, "learning_rate": 0.003, "loss": 4.0537, "step": 23452 }, { "epoch": 0.23453, "grad_norm": 1.2528821116649267, "learning_rate": 0.003, "loss": 4.0412, "step": 23453 }, { "epoch": 0.23454, "grad_norm": 1.2101387604618083, "learning_rate": 0.003, "loss": 4.0473, "step": 23454 }, { "epoch": 0.23455, "grad_norm": 1.195179141503579, "learning_rate": 0.003, "loss": 4.0329, "step": 23455 }, { "epoch": 0.23456, "grad_norm": 1.4079311615674788, "learning_rate": 0.003, "loss": 4.039, "step": 23456 }, { "epoch": 0.23457, "grad_norm": 1.2056265379259656, "learning_rate": 0.003, "loss": 4.0406, "step": 23457 }, { "epoch": 0.23458, "grad_norm": 1.3973622134010926, "learning_rate": 0.003, "loss": 4.0095, "step": 23458 }, { "epoch": 0.23459, "grad_norm": 1.3555146070260597, "learning_rate": 0.003, "loss": 4.029, "step": 23459 }, { "epoch": 0.2346, "grad_norm": 1.0957213427482388, "learning_rate": 0.003, "loss": 4.0467, "step": 23460 }, { "epoch": 0.23461, "grad_norm": 1.3887908367780266, "learning_rate": 0.003, "loss": 4.0474, "step": 23461 }, { "epoch": 0.23462, "grad_norm": 1.2282466648526902, "learning_rate": 0.003, "loss": 4.0648, "step": 23462 }, { "epoch": 0.23463, "grad_norm": 1.3751474104027561, "learning_rate": 0.003, "loss": 4.0306, "step": 23463 }, { "epoch": 0.23464, "grad_norm": 1.1243949525543444, "learning_rate": 0.003, "loss": 4.0483, "step": 23464 }, { "epoch": 0.23465, "grad_norm": 1.3575816508528418, "learning_rate": 0.003, "loss": 4.0566, "step": 23465 }, { "epoch": 0.23466, "grad_norm": 1.2282764477527965, "learning_rate": 0.003, "loss": 4.0454, "step": 23466 }, { "epoch": 0.23467, "grad_norm": 1.4233168149193143, "learning_rate": 0.003, "loss": 4.0537, "step": 23467 }, { "epoch": 0.23468, "grad_norm": 1.4474934405616533, "learning_rate": 0.003, "loss": 4.0375, "step": 23468 }, { "epoch": 0.23469, "grad_norm": 1.4218676326800723, "learning_rate": 0.003, "loss": 4.0189, "step": 23469 }, { "epoch": 0.2347, "grad_norm": 1.664032858766595, "learning_rate": 0.003, "loss": 4.0417, "step": 23470 }, { "epoch": 0.23471, "grad_norm": 1.102384835105614, "learning_rate": 0.003, "loss": 4.0614, "step": 23471 }, { "epoch": 0.23472, "grad_norm": 1.558404434106586, "learning_rate": 0.003, "loss": 4.0413, "step": 23472 }, { "epoch": 0.23473, "grad_norm": 1.4654681533016647, "learning_rate": 0.003, "loss": 4.0497, "step": 23473 }, { "epoch": 0.23474, "grad_norm": 1.0643767449235741, "learning_rate": 0.003, "loss": 4.0526, "step": 23474 }, { "epoch": 0.23475, "grad_norm": 1.3229509118922687, "learning_rate": 0.003, "loss": 4.0289, "step": 23475 }, { "epoch": 0.23476, "grad_norm": 1.1436970071364307, "learning_rate": 0.003, "loss": 4.0507, "step": 23476 }, { "epoch": 0.23477, "grad_norm": 1.299810402344442, "learning_rate": 0.003, "loss": 4.0501, "step": 23477 }, { "epoch": 0.23478, "grad_norm": 1.410001754430526, "learning_rate": 0.003, "loss": 4.0731, "step": 23478 }, { "epoch": 0.23479, "grad_norm": 1.556235327561, "learning_rate": 0.003, "loss": 4.0561, "step": 23479 }, { "epoch": 0.2348, "grad_norm": 1.2825652799276612, "learning_rate": 0.003, "loss": 4.0998, "step": 23480 }, { "epoch": 0.23481, "grad_norm": 1.2644227757042912, "learning_rate": 0.003, "loss": 4.0377, "step": 23481 }, { "epoch": 0.23482, "grad_norm": 1.3954059311583669, "learning_rate": 0.003, "loss": 4.0496, "step": 23482 }, { "epoch": 0.23483, "grad_norm": 1.2727811322791007, "learning_rate": 0.003, "loss": 4.0253, "step": 23483 }, { "epoch": 0.23484, "grad_norm": 1.229768603415025, "learning_rate": 0.003, "loss": 4.0189, "step": 23484 }, { "epoch": 0.23485, "grad_norm": 1.3334129651079165, "learning_rate": 0.003, "loss": 4.0127, "step": 23485 }, { "epoch": 0.23486, "grad_norm": 1.13762202483748, "learning_rate": 0.003, "loss": 4.035, "step": 23486 }, { "epoch": 0.23487, "grad_norm": 1.5323904826671573, "learning_rate": 0.003, "loss": 4.0583, "step": 23487 }, { "epoch": 0.23488, "grad_norm": 1.054834634325706, "learning_rate": 0.003, "loss": 4.0382, "step": 23488 }, { "epoch": 0.23489, "grad_norm": 1.5246077700703264, "learning_rate": 0.003, "loss": 4.0636, "step": 23489 }, { "epoch": 0.2349, "grad_norm": 1.7121003945182949, "learning_rate": 0.003, "loss": 4.0756, "step": 23490 }, { "epoch": 0.23491, "grad_norm": 1.0920165177493513, "learning_rate": 0.003, "loss": 4.0436, "step": 23491 }, { "epoch": 0.23492, "grad_norm": 1.3298167289889824, "learning_rate": 0.003, "loss": 4.0078, "step": 23492 }, { "epoch": 0.23493, "grad_norm": 1.3828459675807627, "learning_rate": 0.003, "loss": 4.0269, "step": 23493 }, { "epoch": 0.23494, "grad_norm": 1.221615658964387, "learning_rate": 0.003, "loss": 4.0524, "step": 23494 }, { "epoch": 0.23495, "grad_norm": 1.4375394642189678, "learning_rate": 0.003, "loss": 4.0726, "step": 23495 }, { "epoch": 0.23496, "grad_norm": 1.1857413863217876, "learning_rate": 0.003, "loss": 4.0435, "step": 23496 }, { "epoch": 0.23497, "grad_norm": 1.4778501948927982, "learning_rate": 0.003, "loss": 4.0227, "step": 23497 }, { "epoch": 0.23498, "grad_norm": 1.0787725661340666, "learning_rate": 0.003, "loss": 4.0232, "step": 23498 }, { "epoch": 0.23499, "grad_norm": 1.3473963181966018, "learning_rate": 0.003, "loss": 4.0443, "step": 23499 }, { "epoch": 0.235, "grad_norm": 1.2036853420996367, "learning_rate": 0.003, "loss": 4.065, "step": 23500 }, { "epoch": 0.23501, "grad_norm": 1.223826477588823, "learning_rate": 0.003, "loss": 4.0266, "step": 23501 }, { "epoch": 0.23502, "grad_norm": 1.364407731652484, "learning_rate": 0.003, "loss": 4.062, "step": 23502 }, { "epoch": 0.23503, "grad_norm": 1.1295936347117093, "learning_rate": 0.003, "loss": 4.0255, "step": 23503 }, { "epoch": 0.23504, "grad_norm": 1.6254081267240414, "learning_rate": 0.003, "loss": 4.047, "step": 23504 }, { "epoch": 0.23505, "grad_norm": 1.1570251855775018, "learning_rate": 0.003, "loss": 4.0868, "step": 23505 }, { "epoch": 0.23506, "grad_norm": 1.48082942571748, "learning_rate": 0.003, "loss": 4.0359, "step": 23506 }, { "epoch": 0.23507, "grad_norm": 1.1792640412381283, "learning_rate": 0.003, "loss": 4.0378, "step": 23507 }, { "epoch": 0.23508, "grad_norm": 1.2190174513725403, "learning_rate": 0.003, "loss": 4.0754, "step": 23508 }, { "epoch": 0.23509, "grad_norm": 1.2423780823318002, "learning_rate": 0.003, "loss": 4.047, "step": 23509 }, { "epoch": 0.2351, "grad_norm": 1.1311859862325964, "learning_rate": 0.003, "loss": 4.0419, "step": 23510 }, { "epoch": 0.23511, "grad_norm": 1.3564786712449024, "learning_rate": 0.003, "loss": 4.05, "step": 23511 }, { "epoch": 0.23512, "grad_norm": 1.2417781889293547, "learning_rate": 0.003, "loss": 4.0658, "step": 23512 }, { "epoch": 0.23513, "grad_norm": 1.344399604598758, "learning_rate": 0.003, "loss": 4.0507, "step": 23513 }, { "epoch": 0.23514, "grad_norm": 1.7041998176781976, "learning_rate": 0.003, "loss": 4.0375, "step": 23514 }, { "epoch": 0.23515, "grad_norm": 1.6275217000215694, "learning_rate": 0.003, "loss": 4.0552, "step": 23515 }, { "epoch": 0.23516, "grad_norm": 1.1442946514548158, "learning_rate": 0.003, "loss": 4.0307, "step": 23516 }, { "epoch": 0.23517, "grad_norm": 1.5762637464493032, "learning_rate": 0.003, "loss": 4.0489, "step": 23517 }, { "epoch": 0.23518, "grad_norm": 1.2188318647418979, "learning_rate": 0.003, "loss": 4.0512, "step": 23518 }, { "epoch": 0.23519, "grad_norm": 1.3055805380645995, "learning_rate": 0.003, "loss": 4.0479, "step": 23519 }, { "epoch": 0.2352, "grad_norm": 1.2737345581728645, "learning_rate": 0.003, "loss": 4.0853, "step": 23520 }, { "epoch": 0.23521, "grad_norm": 1.2852203267288158, "learning_rate": 0.003, "loss": 4.0245, "step": 23521 }, { "epoch": 0.23522, "grad_norm": 1.1579369875805805, "learning_rate": 0.003, "loss": 4.0229, "step": 23522 }, { "epoch": 0.23523, "grad_norm": 1.4677316102183895, "learning_rate": 0.003, "loss": 4.0474, "step": 23523 }, { "epoch": 0.23524, "grad_norm": 1.1430443031419644, "learning_rate": 0.003, "loss": 4.098, "step": 23524 }, { "epoch": 0.23525, "grad_norm": 1.375666600673755, "learning_rate": 0.003, "loss": 4.0557, "step": 23525 }, { "epoch": 0.23526, "grad_norm": 1.272444260566715, "learning_rate": 0.003, "loss": 4.0528, "step": 23526 }, { "epoch": 0.23527, "grad_norm": 1.115615574247192, "learning_rate": 0.003, "loss": 4.0445, "step": 23527 }, { "epoch": 0.23528, "grad_norm": 1.3639004778067785, "learning_rate": 0.003, "loss": 4.0199, "step": 23528 }, { "epoch": 0.23529, "grad_norm": 1.4220766235449773, "learning_rate": 0.003, "loss": 4.0681, "step": 23529 }, { "epoch": 0.2353, "grad_norm": 1.4496060048333888, "learning_rate": 0.003, "loss": 4.0458, "step": 23530 }, { "epoch": 0.23531, "grad_norm": 1.1550650266203935, "learning_rate": 0.003, "loss": 3.9924, "step": 23531 }, { "epoch": 0.23532, "grad_norm": 1.4439004272198512, "learning_rate": 0.003, "loss": 4.0827, "step": 23532 }, { "epoch": 0.23533, "grad_norm": 1.0282806112006082, "learning_rate": 0.003, "loss": 4.0587, "step": 23533 }, { "epoch": 0.23534, "grad_norm": 1.5221722552914796, "learning_rate": 0.003, "loss": 4.0699, "step": 23534 }, { "epoch": 0.23535, "grad_norm": 1.142393286351315, "learning_rate": 0.003, "loss": 4.0574, "step": 23535 }, { "epoch": 0.23536, "grad_norm": 1.3858806885932367, "learning_rate": 0.003, "loss": 4.0232, "step": 23536 }, { "epoch": 0.23537, "grad_norm": 1.2887760091250255, "learning_rate": 0.003, "loss": 4.0397, "step": 23537 }, { "epoch": 0.23538, "grad_norm": 1.268491137227561, "learning_rate": 0.003, "loss": 4.0167, "step": 23538 }, { "epoch": 0.23539, "grad_norm": 1.3745065016018787, "learning_rate": 0.003, "loss": 4.047, "step": 23539 }, { "epoch": 0.2354, "grad_norm": 1.2379029621010118, "learning_rate": 0.003, "loss": 4.0669, "step": 23540 }, { "epoch": 0.23541, "grad_norm": 1.4016614074941376, "learning_rate": 0.003, "loss": 4.0511, "step": 23541 }, { "epoch": 0.23542, "grad_norm": 1.1048640443595201, "learning_rate": 0.003, "loss": 4.0411, "step": 23542 }, { "epoch": 0.23543, "grad_norm": 1.3942164398860328, "learning_rate": 0.003, "loss": 4.0474, "step": 23543 }, { "epoch": 0.23544, "grad_norm": 1.2790291542295937, "learning_rate": 0.003, "loss": 4.0332, "step": 23544 }, { "epoch": 0.23545, "grad_norm": 1.3862190297543544, "learning_rate": 0.003, "loss": 4.0332, "step": 23545 }, { "epoch": 0.23546, "grad_norm": 1.2095973979703265, "learning_rate": 0.003, "loss": 4.026, "step": 23546 }, { "epoch": 0.23547, "grad_norm": 1.5758682439787788, "learning_rate": 0.003, "loss": 4.0567, "step": 23547 }, { "epoch": 0.23548, "grad_norm": 1.2828088939122402, "learning_rate": 0.003, "loss": 4.0602, "step": 23548 }, { "epoch": 0.23549, "grad_norm": 1.4679174046132317, "learning_rate": 0.003, "loss": 4.0605, "step": 23549 }, { "epoch": 0.2355, "grad_norm": 1.0639583078929091, "learning_rate": 0.003, "loss": 4.0316, "step": 23550 }, { "epoch": 0.23551, "grad_norm": 1.556049948036192, "learning_rate": 0.003, "loss": 4.0347, "step": 23551 }, { "epoch": 0.23552, "grad_norm": 1.248536199129532, "learning_rate": 0.003, "loss": 4.0465, "step": 23552 }, { "epoch": 0.23553, "grad_norm": 1.5064653223459554, "learning_rate": 0.003, "loss": 4.0493, "step": 23553 }, { "epoch": 0.23554, "grad_norm": 1.1439828566478294, "learning_rate": 0.003, "loss": 4.0645, "step": 23554 }, { "epoch": 0.23555, "grad_norm": 1.713054405907999, "learning_rate": 0.003, "loss": 4.0548, "step": 23555 }, { "epoch": 0.23556, "grad_norm": 1.0470829391495002, "learning_rate": 0.003, "loss": 4.0186, "step": 23556 }, { "epoch": 0.23557, "grad_norm": 1.366403003368298, "learning_rate": 0.003, "loss": 4.0201, "step": 23557 }, { "epoch": 0.23558, "grad_norm": 1.2510403515167263, "learning_rate": 0.003, "loss": 4.0512, "step": 23558 }, { "epoch": 0.23559, "grad_norm": 1.5170501567192332, "learning_rate": 0.003, "loss": 4.0228, "step": 23559 }, { "epoch": 0.2356, "grad_norm": 1.5571243941431558, "learning_rate": 0.003, "loss": 4.07, "step": 23560 }, { "epoch": 0.23561, "grad_norm": 1.2034867092975143, "learning_rate": 0.003, "loss": 4.0515, "step": 23561 }, { "epoch": 0.23562, "grad_norm": 1.2143662031534421, "learning_rate": 0.003, "loss": 4.0401, "step": 23562 }, { "epoch": 0.23563, "grad_norm": 1.188212442106662, "learning_rate": 0.003, "loss": 4.0502, "step": 23563 }, { "epoch": 0.23564, "grad_norm": 1.3677431216054141, "learning_rate": 0.003, "loss": 4.0569, "step": 23564 }, { "epoch": 0.23565, "grad_norm": 1.101889247978777, "learning_rate": 0.003, "loss": 4.0273, "step": 23565 }, { "epoch": 0.23566, "grad_norm": 1.2717039396998226, "learning_rate": 0.003, "loss": 4.0338, "step": 23566 }, { "epoch": 0.23567, "grad_norm": 1.1856496646020174, "learning_rate": 0.003, "loss": 4.0457, "step": 23567 }, { "epoch": 0.23568, "grad_norm": 1.2855123764395402, "learning_rate": 0.003, "loss": 4.0515, "step": 23568 }, { "epoch": 0.23569, "grad_norm": 1.3967086315002388, "learning_rate": 0.003, "loss": 4.0712, "step": 23569 }, { "epoch": 0.2357, "grad_norm": 1.3082550029704532, "learning_rate": 0.003, "loss": 4.0423, "step": 23570 }, { "epoch": 0.23571, "grad_norm": 1.7442156804502371, "learning_rate": 0.003, "loss": 4.0541, "step": 23571 }, { "epoch": 0.23572, "grad_norm": 1.1207225006713477, "learning_rate": 0.003, "loss": 4.0321, "step": 23572 }, { "epoch": 0.23573, "grad_norm": 1.2888850203454607, "learning_rate": 0.003, "loss": 4.0475, "step": 23573 }, { "epoch": 0.23574, "grad_norm": 1.1377771881493937, "learning_rate": 0.003, "loss": 4.0213, "step": 23574 }, { "epoch": 0.23575, "grad_norm": 1.4572297206713176, "learning_rate": 0.003, "loss": 4.0719, "step": 23575 }, { "epoch": 0.23576, "grad_norm": 1.0875871086980016, "learning_rate": 0.003, "loss": 4.0541, "step": 23576 }, { "epoch": 0.23577, "grad_norm": 1.5637828316273845, "learning_rate": 0.003, "loss": 4.048, "step": 23577 }, { "epoch": 0.23578, "grad_norm": 1.1770711611707216, "learning_rate": 0.003, "loss": 4.0522, "step": 23578 }, { "epoch": 0.23579, "grad_norm": 1.820602698168912, "learning_rate": 0.003, "loss": 4.0304, "step": 23579 }, { "epoch": 0.2358, "grad_norm": 1.171823469053568, "learning_rate": 0.003, "loss": 4.0389, "step": 23580 }, { "epoch": 0.23581, "grad_norm": 1.2935324234332046, "learning_rate": 0.003, "loss": 4.0434, "step": 23581 }, { "epoch": 0.23582, "grad_norm": 1.3044972822919219, "learning_rate": 0.003, "loss": 4.072, "step": 23582 }, { "epoch": 0.23583, "grad_norm": 1.238231255195047, "learning_rate": 0.003, "loss": 4.0179, "step": 23583 }, { "epoch": 0.23584, "grad_norm": 1.4004548670675205, "learning_rate": 0.003, "loss": 4.0445, "step": 23584 }, { "epoch": 0.23585, "grad_norm": 1.0231016296401765, "learning_rate": 0.003, "loss": 4.0318, "step": 23585 }, { "epoch": 0.23586, "grad_norm": 1.4767443977757324, "learning_rate": 0.003, "loss": 4.0365, "step": 23586 }, { "epoch": 0.23587, "grad_norm": 1.5119451251525389, "learning_rate": 0.003, "loss": 4.0581, "step": 23587 }, { "epoch": 0.23588, "grad_norm": 1.1989637011059378, "learning_rate": 0.003, "loss": 4.0356, "step": 23588 }, { "epoch": 0.23589, "grad_norm": 1.5270406155035605, "learning_rate": 0.003, "loss": 4.0588, "step": 23589 }, { "epoch": 0.2359, "grad_norm": 1.269032908594783, "learning_rate": 0.003, "loss": 4.051, "step": 23590 }, { "epoch": 0.23591, "grad_norm": 1.6225865933820671, "learning_rate": 0.003, "loss": 4.0582, "step": 23591 }, { "epoch": 0.23592, "grad_norm": 1.08473728649285, "learning_rate": 0.003, "loss": 4.0794, "step": 23592 }, { "epoch": 0.23593, "grad_norm": 1.3425190691770323, "learning_rate": 0.003, "loss": 4.0494, "step": 23593 }, { "epoch": 0.23594, "grad_norm": 1.2466637972544092, "learning_rate": 0.003, "loss": 4.0673, "step": 23594 }, { "epoch": 0.23595, "grad_norm": 1.2785123651003583, "learning_rate": 0.003, "loss": 4.0631, "step": 23595 }, { "epoch": 0.23596, "grad_norm": 1.4788456954119416, "learning_rate": 0.003, "loss": 4.068, "step": 23596 }, { "epoch": 0.23597, "grad_norm": 1.0538422980668722, "learning_rate": 0.003, "loss": 4.0607, "step": 23597 }, { "epoch": 0.23598, "grad_norm": 1.7099588121690856, "learning_rate": 0.003, "loss": 4.0714, "step": 23598 }, { "epoch": 0.23599, "grad_norm": 1.003066627245704, "learning_rate": 0.003, "loss": 4.0412, "step": 23599 }, { "epoch": 0.236, "grad_norm": 1.9791699549245332, "learning_rate": 0.003, "loss": 4.0916, "step": 23600 }, { "epoch": 0.23601, "grad_norm": 1.2652429275722812, "learning_rate": 0.003, "loss": 4.0479, "step": 23601 }, { "epoch": 0.23602, "grad_norm": 1.1216889993479584, "learning_rate": 0.003, "loss": 4.052, "step": 23602 }, { "epoch": 0.23603, "grad_norm": 1.394702868129797, "learning_rate": 0.003, "loss": 4.0917, "step": 23603 }, { "epoch": 0.23604, "grad_norm": 1.2191916467988568, "learning_rate": 0.003, "loss": 4.0613, "step": 23604 }, { "epoch": 0.23605, "grad_norm": 1.548537069103473, "learning_rate": 0.003, "loss": 4.0799, "step": 23605 }, { "epoch": 0.23606, "grad_norm": 1.1493443935645882, "learning_rate": 0.003, "loss": 4.0514, "step": 23606 }, { "epoch": 0.23607, "grad_norm": 1.3073424183189124, "learning_rate": 0.003, "loss": 4.0715, "step": 23607 }, { "epoch": 0.23608, "grad_norm": 1.4108859714119362, "learning_rate": 0.003, "loss": 4.0337, "step": 23608 }, { "epoch": 0.23609, "grad_norm": 1.1970254370362037, "learning_rate": 0.003, "loss": 4.0521, "step": 23609 }, { "epoch": 0.2361, "grad_norm": 1.0952019716437682, "learning_rate": 0.003, "loss": 4.0328, "step": 23610 }, { "epoch": 0.23611, "grad_norm": 1.3565307731217888, "learning_rate": 0.003, "loss": 4.0356, "step": 23611 }, { "epoch": 0.23612, "grad_norm": 1.3095323280163718, "learning_rate": 0.003, "loss": 4.0544, "step": 23612 }, { "epoch": 0.23613, "grad_norm": 1.2567105525421158, "learning_rate": 0.003, "loss": 4.0617, "step": 23613 }, { "epoch": 0.23614, "grad_norm": 1.3368337684650289, "learning_rate": 0.003, "loss": 4.0269, "step": 23614 }, { "epoch": 0.23615, "grad_norm": 1.34774045634497, "learning_rate": 0.003, "loss": 4.0494, "step": 23615 }, { "epoch": 0.23616, "grad_norm": 1.3199422151102764, "learning_rate": 0.003, "loss": 4.0288, "step": 23616 }, { "epoch": 0.23617, "grad_norm": 1.2142054499970667, "learning_rate": 0.003, "loss": 4.0567, "step": 23617 }, { "epoch": 0.23618, "grad_norm": 1.4447440314536921, "learning_rate": 0.003, "loss": 4.019, "step": 23618 }, { "epoch": 0.23619, "grad_norm": 1.1610116807628668, "learning_rate": 0.003, "loss": 4.0226, "step": 23619 }, { "epoch": 0.2362, "grad_norm": 1.2968802701414694, "learning_rate": 0.003, "loss": 4.0398, "step": 23620 }, { "epoch": 0.23621, "grad_norm": 1.411809158090715, "learning_rate": 0.003, "loss": 4.0422, "step": 23621 }, { "epoch": 0.23622, "grad_norm": 1.0018815118641649, "learning_rate": 0.003, "loss": 4.0397, "step": 23622 }, { "epoch": 0.23623, "grad_norm": 1.3635308876029366, "learning_rate": 0.003, "loss": 4.0112, "step": 23623 }, { "epoch": 0.23624, "grad_norm": 1.10965217854949, "learning_rate": 0.003, "loss": 4.0505, "step": 23624 }, { "epoch": 0.23625, "grad_norm": 1.7488481805435205, "learning_rate": 0.003, "loss": 4.062, "step": 23625 }, { "epoch": 0.23626, "grad_norm": 1.0859923544661083, "learning_rate": 0.003, "loss": 4.07, "step": 23626 }, { "epoch": 0.23627, "grad_norm": 1.2384710397490812, "learning_rate": 0.003, "loss": 4.0334, "step": 23627 }, { "epoch": 0.23628, "grad_norm": 1.2155553676919961, "learning_rate": 0.003, "loss": 4.0261, "step": 23628 }, { "epoch": 0.23629, "grad_norm": 1.0133477747370572, "learning_rate": 0.003, "loss": 4.0556, "step": 23629 }, { "epoch": 0.2363, "grad_norm": 1.4125211879000776, "learning_rate": 0.003, "loss": 4.0668, "step": 23630 }, { "epoch": 0.23631, "grad_norm": 1.3163483627954522, "learning_rate": 0.003, "loss": 4.0223, "step": 23631 }, { "epoch": 0.23632, "grad_norm": 1.5009282652314748, "learning_rate": 0.003, "loss": 4.0422, "step": 23632 }, { "epoch": 0.23633, "grad_norm": 1.461536087930038, "learning_rate": 0.003, "loss": 4.0466, "step": 23633 }, { "epoch": 0.23634, "grad_norm": 1.3360946359702124, "learning_rate": 0.003, "loss": 4.0828, "step": 23634 }, { "epoch": 0.23635, "grad_norm": 1.6226273615839206, "learning_rate": 0.003, "loss": 4.0524, "step": 23635 }, { "epoch": 0.23636, "grad_norm": 0.946379008272811, "learning_rate": 0.003, "loss": 4.0436, "step": 23636 }, { "epoch": 0.23637, "grad_norm": 1.2687992609766823, "learning_rate": 0.003, "loss": 4.0483, "step": 23637 }, { "epoch": 0.23638, "grad_norm": 1.3144689070142017, "learning_rate": 0.003, "loss": 4.0534, "step": 23638 }, { "epoch": 0.23639, "grad_norm": 1.1833706412679799, "learning_rate": 0.003, "loss": 4.0346, "step": 23639 }, { "epoch": 0.2364, "grad_norm": 1.2643846450362188, "learning_rate": 0.003, "loss": 4.0167, "step": 23640 }, { "epoch": 0.23641, "grad_norm": 1.2836297052401486, "learning_rate": 0.003, "loss": 4.0231, "step": 23641 }, { "epoch": 0.23642, "grad_norm": 1.350728821115279, "learning_rate": 0.003, "loss": 4.0601, "step": 23642 }, { "epoch": 0.23643, "grad_norm": 1.195000845841808, "learning_rate": 0.003, "loss": 4.08, "step": 23643 }, { "epoch": 0.23644, "grad_norm": 1.4629558253860953, "learning_rate": 0.003, "loss": 4.0758, "step": 23644 }, { "epoch": 0.23645, "grad_norm": 1.093914745066469, "learning_rate": 0.003, "loss": 4.0464, "step": 23645 }, { "epoch": 0.23646, "grad_norm": 1.3877748865140367, "learning_rate": 0.003, "loss": 4.0483, "step": 23646 }, { "epoch": 0.23647, "grad_norm": 1.264758900032, "learning_rate": 0.003, "loss": 4.0363, "step": 23647 }, { "epoch": 0.23648, "grad_norm": 1.5739334685255446, "learning_rate": 0.003, "loss": 4.0477, "step": 23648 }, { "epoch": 0.23649, "grad_norm": 1.0311298235821733, "learning_rate": 0.003, "loss": 4.0512, "step": 23649 }, { "epoch": 0.2365, "grad_norm": 1.6397741977463254, "learning_rate": 0.003, "loss": 4.047, "step": 23650 }, { "epoch": 0.23651, "grad_norm": 1.0232840837952728, "learning_rate": 0.003, "loss": 4.0534, "step": 23651 }, { "epoch": 0.23652, "grad_norm": 1.629541944846788, "learning_rate": 0.003, "loss": 4.0328, "step": 23652 }, { "epoch": 0.23653, "grad_norm": 1.177581598969811, "learning_rate": 0.003, "loss": 4.067, "step": 23653 }, { "epoch": 0.23654, "grad_norm": 1.1235775126833063, "learning_rate": 0.003, "loss": 4.016, "step": 23654 }, { "epoch": 0.23655, "grad_norm": 1.2878011900426842, "learning_rate": 0.003, "loss": 4.0505, "step": 23655 }, { "epoch": 0.23656, "grad_norm": 1.1940669610180175, "learning_rate": 0.003, "loss": 4.0303, "step": 23656 }, { "epoch": 0.23657, "grad_norm": 1.206844629046312, "learning_rate": 0.003, "loss": 4.0449, "step": 23657 }, { "epoch": 0.23658, "grad_norm": 1.4371758952293205, "learning_rate": 0.003, "loss": 4.0699, "step": 23658 }, { "epoch": 0.23659, "grad_norm": 1.2523693023114035, "learning_rate": 0.003, "loss": 4.0439, "step": 23659 }, { "epoch": 0.2366, "grad_norm": 1.462068175556941, "learning_rate": 0.003, "loss": 4.0713, "step": 23660 }, { "epoch": 0.23661, "grad_norm": 1.481961618084773, "learning_rate": 0.003, "loss": 4.0491, "step": 23661 }, { "epoch": 0.23662, "grad_norm": 1.269959000324739, "learning_rate": 0.003, "loss": 4.0589, "step": 23662 }, { "epoch": 0.23663, "grad_norm": 1.2376874660119221, "learning_rate": 0.003, "loss": 4.0465, "step": 23663 }, { "epoch": 0.23664, "grad_norm": 1.465345469478452, "learning_rate": 0.003, "loss": 4.0585, "step": 23664 }, { "epoch": 0.23665, "grad_norm": 1.2198269537172273, "learning_rate": 0.003, "loss": 4.0494, "step": 23665 }, { "epoch": 0.23666, "grad_norm": 1.2361049303348048, "learning_rate": 0.003, "loss": 4.1018, "step": 23666 }, { "epoch": 0.23667, "grad_norm": 1.2426580260886706, "learning_rate": 0.003, "loss": 4.0458, "step": 23667 }, { "epoch": 0.23668, "grad_norm": 1.0462775177853703, "learning_rate": 0.003, "loss": 4.015, "step": 23668 }, { "epoch": 0.23669, "grad_norm": 1.4892376243472623, "learning_rate": 0.003, "loss": 4.0573, "step": 23669 }, { "epoch": 0.2367, "grad_norm": 1.227079176290411, "learning_rate": 0.003, "loss": 4.0141, "step": 23670 }, { "epoch": 0.23671, "grad_norm": 1.6896739591178331, "learning_rate": 0.003, "loss": 4.0535, "step": 23671 }, { "epoch": 0.23672, "grad_norm": 1.166689132632801, "learning_rate": 0.003, "loss": 4.0547, "step": 23672 }, { "epoch": 0.23673, "grad_norm": 1.4623926653176087, "learning_rate": 0.003, "loss": 4.018, "step": 23673 }, { "epoch": 0.23674, "grad_norm": 1.1253555650291334, "learning_rate": 0.003, "loss": 4.0426, "step": 23674 }, { "epoch": 0.23675, "grad_norm": 1.4424946620000527, "learning_rate": 0.003, "loss": 4.035, "step": 23675 }, { "epoch": 0.23676, "grad_norm": 1.1775474693912467, "learning_rate": 0.003, "loss": 4.08, "step": 23676 }, { "epoch": 0.23677, "grad_norm": 1.2019398953946223, "learning_rate": 0.003, "loss": 4.0426, "step": 23677 }, { "epoch": 0.23678, "grad_norm": 1.2624286059830971, "learning_rate": 0.003, "loss": 4.0555, "step": 23678 }, { "epoch": 0.23679, "grad_norm": 1.353696407430561, "learning_rate": 0.003, "loss": 4.04, "step": 23679 }, { "epoch": 0.2368, "grad_norm": 1.4063596347236127, "learning_rate": 0.003, "loss": 4.0037, "step": 23680 }, { "epoch": 0.23681, "grad_norm": 1.0400918466665534, "learning_rate": 0.003, "loss": 4.0195, "step": 23681 }, { "epoch": 0.23682, "grad_norm": 1.7751778071967164, "learning_rate": 0.003, "loss": 4.025, "step": 23682 }, { "epoch": 0.23683, "grad_norm": 1.1440786910415348, "learning_rate": 0.003, "loss": 4.0663, "step": 23683 }, { "epoch": 0.23684, "grad_norm": 1.374883240005277, "learning_rate": 0.003, "loss": 4.048, "step": 23684 }, { "epoch": 0.23685, "grad_norm": 1.0745801359741007, "learning_rate": 0.003, "loss": 4.0316, "step": 23685 }, { "epoch": 0.23686, "grad_norm": 1.5386474927017875, "learning_rate": 0.003, "loss": 4.0721, "step": 23686 }, { "epoch": 0.23687, "grad_norm": 1.1690145054557997, "learning_rate": 0.003, "loss": 4.0443, "step": 23687 }, { "epoch": 0.23688, "grad_norm": 1.356249541312422, "learning_rate": 0.003, "loss": 4.0589, "step": 23688 }, { "epoch": 0.23689, "grad_norm": 1.0793811140424336, "learning_rate": 0.003, "loss": 4.0481, "step": 23689 }, { "epoch": 0.2369, "grad_norm": 1.3056075597987458, "learning_rate": 0.003, "loss": 4.0686, "step": 23690 }, { "epoch": 0.23691, "grad_norm": 1.1297859569042559, "learning_rate": 0.003, "loss": 4.0248, "step": 23691 }, { "epoch": 0.23692, "grad_norm": 1.220226725732296, "learning_rate": 0.003, "loss": 4.0389, "step": 23692 }, { "epoch": 0.23693, "grad_norm": 1.4298512360059767, "learning_rate": 0.003, "loss": 4.0459, "step": 23693 }, { "epoch": 0.23694, "grad_norm": 1.2220316175695687, "learning_rate": 0.003, "loss": 4.0097, "step": 23694 }, { "epoch": 0.23695, "grad_norm": 1.2735929220920998, "learning_rate": 0.003, "loss": 4.0583, "step": 23695 }, { "epoch": 0.23696, "grad_norm": 1.3782266394623592, "learning_rate": 0.003, "loss": 4.0479, "step": 23696 }, { "epoch": 0.23697, "grad_norm": 1.2610129440776563, "learning_rate": 0.003, "loss": 4.0307, "step": 23697 }, { "epoch": 0.23698, "grad_norm": 1.3325089768412006, "learning_rate": 0.003, "loss": 4.0217, "step": 23698 }, { "epoch": 0.23699, "grad_norm": 1.2869215805307337, "learning_rate": 0.003, "loss": 4.0143, "step": 23699 }, { "epoch": 0.237, "grad_norm": 1.284724772908701, "learning_rate": 0.003, "loss": 4.0229, "step": 23700 }, { "epoch": 0.23701, "grad_norm": 1.5090489407528482, "learning_rate": 0.003, "loss": 4.0174, "step": 23701 }, { "epoch": 0.23702, "grad_norm": 1.3073404647654603, "learning_rate": 0.003, "loss": 4.0392, "step": 23702 }, { "epoch": 0.23703, "grad_norm": 1.1558030374791959, "learning_rate": 0.003, "loss": 4.0341, "step": 23703 }, { "epoch": 0.23704, "grad_norm": 1.4692820358607068, "learning_rate": 0.003, "loss": 4.0517, "step": 23704 }, { "epoch": 0.23705, "grad_norm": 0.9417283807886037, "learning_rate": 0.003, "loss": 4.0127, "step": 23705 }, { "epoch": 0.23706, "grad_norm": 1.5206367146642286, "learning_rate": 0.003, "loss": 4.0191, "step": 23706 }, { "epoch": 0.23707, "grad_norm": 1.2779932353414432, "learning_rate": 0.003, "loss": 4.0445, "step": 23707 }, { "epoch": 0.23708, "grad_norm": 1.545714886319892, "learning_rate": 0.003, "loss": 4.0544, "step": 23708 }, { "epoch": 0.23709, "grad_norm": 1.3814072514689995, "learning_rate": 0.003, "loss": 4.059, "step": 23709 }, { "epoch": 0.2371, "grad_norm": 1.2825906797456401, "learning_rate": 0.003, "loss": 4.0331, "step": 23710 }, { "epoch": 0.23711, "grad_norm": 1.2494070433367912, "learning_rate": 0.003, "loss": 4.0539, "step": 23711 }, { "epoch": 0.23712, "grad_norm": 1.30400311295083, "learning_rate": 0.003, "loss": 4.0516, "step": 23712 }, { "epoch": 0.23713, "grad_norm": 1.2562514373354503, "learning_rate": 0.003, "loss": 4.0407, "step": 23713 }, { "epoch": 0.23714, "grad_norm": 1.2832290456993536, "learning_rate": 0.003, "loss": 4.0621, "step": 23714 }, { "epoch": 0.23715, "grad_norm": 1.3039640135418236, "learning_rate": 0.003, "loss": 4.0352, "step": 23715 }, { "epoch": 0.23716, "grad_norm": 1.222289107950515, "learning_rate": 0.003, "loss": 4.0472, "step": 23716 }, { "epoch": 0.23717, "grad_norm": 1.5755722655773658, "learning_rate": 0.003, "loss": 4.0565, "step": 23717 }, { "epoch": 0.23718, "grad_norm": 1.1780195442081456, "learning_rate": 0.003, "loss": 4.0633, "step": 23718 }, { "epoch": 0.23719, "grad_norm": 1.4001038713573974, "learning_rate": 0.003, "loss": 4.062, "step": 23719 }, { "epoch": 0.2372, "grad_norm": 1.120587085206351, "learning_rate": 0.003, "loss": 4.0452, "step": 23720 }, { "epoch": 0.23721, "grad_norm": 1.113506003852723, "learning_rate": 0.003, "loss": 4.0089, "step": 23721 }, { "epoch": 0.23722, "grad_norm": 1.1902530140764338, "learning_rate": 0.003, "loss": 4.0426, "step": 23722 }, { "epoch": 0.23723, "grad_norm": 1.5474127615484905, "learning_rate": 0.003, "loss": 4.0665, "step": 23723 }, { "epoch": 0.23724, "grad_norm": 1.2084857165780836, "learning_rate": 0.003, "loss": 4.0609, "step": 23724 }, { "epoch": 0.23725, "grad_norm": 1.3756020473215986, "learning_rate": 0.003, "loss": 4.0499, "step": 23725 }, { "epoch": 0.23726, "grad_norm": 1.2561203221931616, "learning_rate": 0.003, "loss": 4.0495, "step": 23726 }, { "epoch": 0.23727, "grad_norm": 1.5585393178990727, "learning_rate": 0.003, "loss": 4.0484, "step": 23727 }, { "epoch": 0.23728, "grad_norm": 1.1033787870196319, "learning_rate": 0.003, "loss": 3.9958, "step": 23728 }, { "epoch": 0.23729, "grad_norm": 1.5542847328246168, "learning_rate": 0.003, "loss": 4.078, "step": 23729 }, { "epoch": 0.2373, "grad_norm": 1.0816735241659021, "learning_rate": 0.003, "loss": 4.0375, "step": 23730 }, { "epoch": 0.23731, "grad_norm": 1.3889145716303541, "learning_rate": 0.003, "loss": 4.0446, "step": 23731 }, { "epoch": 0.23732, "grad_norm": 1.2593014438570076, "learning_rate": 0.003, "loss": 4.0517, "step": 23732 }, { "epoch": 0.23733, "grad_norm": 1.3321997521846503, "learning_rate": 0.003, "loss": 4.0627, "step": 23733 }, { "epoch": 0.23734, "grad_norm": 1.1744229189687176, "learning_rate": 0.003, "loss": 4.0383, "step": 23734 }, { "epoch": 0.23735, "grad_norm": 1.457656813230531, "learning_rate": 0.003, "loss": 4.0477, "step": 23735 }, { "epoch": 0.23736, "grad_norm": 1.1030522299872654, "learning_rate": 0.003, "loss": 4.0134, "step": 23736 }, { "epoch": 0.23737, "grad_norm": 1.424446338633868, "learning_rate": 0.003, "loss": 4.0394, "step": 23737 }, { "epoch": 0.23738, "grad_norm": 1.2894176123642855, "learning_rate": 0.003, "loss": 4.0553, "step": 23738 }, { "epoch": 0.23739, "grad_norm": 1.3409103580448187, "learning_rate": 0.003, "loss": 4.0565, "step": 23739 }, { "epoch": 0.2374, "grad_norm": 1.4057281996507969, "learning_rate": 0.003, "loss": 4.0589, "step": 23740 }, { "epoch": 0.23741, "grad_norm": 1.115664020440304, "learning_rate": 0.003, "loss": 4.0302, "step": 23741 }, { "epoch": 0.23742, "grad_norm": 1.3408818434111343, "learning_rate": 0.003, "loss": 4.0451, "step": 23742 }, { "epoch": 0.23743, "grad_norm": 1.3386866261604355, "learning_rate": 0.003, "loss": 4.0501, "step": 23743 }, { "epoch": 0.23744, "grad_norm": 1.8501301660584664, "learning_rate": 0.003, "loss": 4.052, "step": 23744 }, { "epoch": 0.23745, "grad_norm": 1.3553097999844423, "learning_rate": 0.003, "loss": 4.0539, "step": 23745 }, { "epoch": 0.23746, "grad_norm": 1.146161360704984, "learning_rate": 0.003, "loss": 4.0572, "step": 23746 }, { "epoch": 0.23747, "grad_norm": 1.332728259047778, "learning_rate": 0.003, "loss": 4.0472, "step": 23747 }, { "epoch": 0.23748, "grad_norm": 1.4008393918644393, "learning_rate": 0.003, "loss": 4.0247, "step": 23748 }, { "epoch": 0.23749, "grad_norm": 1.480555260688311, "learning_rate": 0.003, "loss": 4.0355, "step": 23749 }, { "epoch": 0.2375, "grad_norm": 1.208726374303708, "learning_rate": 0.003, "loss": 4.0459, "step": 23750 }, { "epoch": 0.23751, "grad_norm": 1.3185136266661017, "learning_rate": 0.003, "loss": 4.0594, "step": 23751 }, { "epoch": 0.23752, "grad_norm": 1.198548566656275, "learning_rate": 0.003, "loss": 4.0358, "step": 23752 }, { "epoch": 0.23753, "grad_norm": 1.3611165764660003, "learning_rate": 0.003, "loss": 4.0491, "step": 23753 }, { "epoch": 0.23754, "grad_norm": 1.1900264302317343, "learning_rate": 0.003, "loss": 4.0455, "step": 23754 }, { "epoch": 0.23755, "grad_norm": 1.3856426778562365, "learning_rate": 0.003, "loss": 4.0367, "step": 23755 }, { "epoch": 0.23756, "grad_norm": 1.1452716099963869, "learning_rate": 0.003, "loss": 4.036, "step": 23756 }, { "epoch": 0.23757, "grad_norm": 1.2830380403883017, "learning_rate": 0.003, "loss": 4.0448, "step": 23757 }, { "epoch": 0.23758, "grad_norm": 1.263668970515008, "learning_rate": 0.003, "loss": 4.0508, "step": 23758 }, { "epoch": 0.23759, "grad_norm": 1.345236670290279, "learning_rate": 0.003, "loss": 4.0375, "step": 23759 }, { "epoch": 0.2376, "grad_norm": 1.292150063376207, "learning_rate": 0.003, "loss": 4.0333, "step": 23760 }, { "epoch": 0.23761, "grad_norm": 1.1808226954554617, "learning_rate": 0.003, "loss": 4.0571, "step": 23761 }, { "epoch": 0.23762, "grad_norm": 1.2465742158976219, "learning_rate": 0.003, "loss": 4.0322, "step": 23762 }, { "epoch": 0.23763, "grad_norm": 1.261865763600032, "learning_rate": 0.003, "loss": 4.041, "step": 23763 }, { "epoch": 0.23764, "grad_norm": 1.1471541620118675, "learning_rate": 0.003, "loss": 4.0027, "step": 23764 }, { "epoch": 0.23765, "grad_norm": 1.5774345085773744, "learning_rate": 0.003, "loss": 4.0503, "step": 23765 }, { "epoch": 0.23766, "grad_norm": 1.1769892792339534, "learning_rate": 0.003, "loss": 4.031, "step": 23766 }, { "epoch": 0.23767, "grad_norm": 1.5802083274816956, "learning_rate": 0.003, "loss": 4.0208, "step": 23767 }, { "epoch": 0.23768, "grad_norm": 1.0660115952853646, "learning_rate": 0.003, "loss": 4.0329, "step": 23768 }, { "epoch": 0.23769, "grad_norm": 1.096845953913119, "learning_rate": 0.003, "loss": 4.0399, "step": 23769 }, { "epoch": 0.2377, "grad_norm": 1.6428235416320027, "learning_rate": 0.003, "loss": 4.044, "step": 23770 }, { "epoch": 0.23771, "grad_norm": 1.2268917946899078, "learning_rate": 0.003, "loss": 4.026, "step": 23771 }, { "epoch": 0.23772, "grad_norm": 1.481420221928791, "learning_rate": 0.003, "loss": 4.0248, "step": 23772 }, { "epoch": 0.23773, "grad_norm": 1.2066449681381066, "learning_rate": 0.003, "loss": 4.0299, "step": 23773 }, { "epoch": 0.23774, "grad_norm": 1.3852901385359326, "learning_rate": 0.003, "loss": 4.0452, "step": 23774 }, { "epoch": 0.23775, "grad_norm": 1.2098759784121849, "learning_rate": 0.003, "loss": 4.0478, "step": 23775 }, { "epoch": 0.23776, "grad_norm": 1.35072879631268, "learning_rate": 0.003, "loss": 4.0433, "step": 23776 }, { "epoch": 0.23777, "grad_norm": 1.0212375161493594, "learning_rate": 0.003, "loss": 4.0439, "step": 23777 }, { "epoch": 0.23778, "grad_norm": 1.5916799860075879, "learning_rate": 0.003, "loss": 4.0334, "step": 23778 }, { "epoch": 0.23779, "grad_norm": 1.2941785183522605, "learning_rate": 0.003, "loss": 4.0332, "step": 23779 }, { "epoch": 0.2378, "grad_norm": 1.389404951829121, "learning_rate": 0.003, "loss": 3.9991, "step": 23780 }, { "epoch": 0.23781, "grad_norm": 1.23583468677105, "learning_rate": 0.003, "loss": 4.0668, "step": 23781 }, { "epoch": 0.23782, "grad_norm": 1.1860316882252406, "learning_rate": 0.003, "loss": 4.0583, "step": 23782 }, { "epoch": 0.23783, "grad_norm": 1.3522011828676728, "learning_rate": 0.003, "loss": 4.0491, "step": 23783 }, { "epoch": 0.23784, "grad_norm": 1.4237710626629256, "learning_rate": 0.003, "loss": 4.0483, "step": 23784 }, { "epoch": 0.23785, "grad_norm": 1.4850213377224577, "learning_rate": 0.003, "loss": 4.0241, "step": 23785 }, { "epoch": 0.23786, "grad_norm": 1.367666695197778, "learning_rate": 0.003, "loss": 4.0566, "step": 23786 }, { "epoch": 0.23787, "grad_norm": 1.5293138274104534, "learning_rate": 0.003, "loss": 4.0494, "step": 23787 }, { "epoch": 0.23788, "grad_norm": 1.0747528320121753, "learning_rate": 0.003, "loss": 4.0471, "step": 23788 }, { "epoch": 0.23789, "grad_norm": 1.661082382912658, "learning_rate": 0.003, "loss": 4.0358, "step": 23789 }, { "epoch": 0.2379, "grad_norm": 1.0991220074339236, "learning_rate": 0.003, "loss": 4.0606, "step": 23790 }, { "epoch": 0.23791, "grad_norm": 1.6177981022933468, "learning_rate": 0.003, "loss": 4.0305, "step": 23791 }, { "epoch": 0.23792, "grad_norm": 1.2496492678688773, "learning_rate": 0.003, "loss": 4.0384, "step": 23792 }, { "epoch": 0.23793, "grad_norm": 1.270639004872499, "learning_rate": 0.003, "loss": 4.0321, "step": 23793 }, { "epoch": 0.23794, "grad_norm": 1.396290220103336, "learning_rate": 0.003, "loss": 4.0737, "step": 23794 }, { "epoch": 0.23795, "grad_norm": 1.2041502942283608, "learning_rate": 0.003, "loss": 4.0596, "step": 23795 }, { "epoch": 0.23796, "grad_norm": 1.4160771650035684, "learning_rate": 0.003, "loss": 4.0778, "step": 23796 }, { "epoch": 0.23797, "grad_norm": 1.2315464099960576, "learning_rate": 0.003, "loss": 4.0265, "step": 23797 }, { "epoch": 0.23798, "grad_norm": 1.4166731645042991, "learning_rate": 0.003, "loss": 4.025, "step": 23798 }, { "epoch": 0.23799, "grad_norm": 1.1756818998517438, "learning_rate": 0.003, "loss": 4.0629, "step": 23799 }, { "epoch": 0.238, "grad_norm": 1.3068080497004517, "learning_rate": 0.003, "loss": 4.0356, "step": 23800 }, { "epoch": 0.23801, "grad_norm": 1.0583396063313557, "learning_rate": 0.003, "loss": 4.0854, "step": 23801 }, { "epoch": 0.23802, "grad_norm": 1.4194066619768246, "learning_rate": 0.003, "loss": 4.0424, "step": 23802 }, { "epoch": 0.23803, "grad_norm": 0.9248461116780902, "learning_rate": 0.003, "loss": 4.0538, "step": 23803 }, { "epoch": 0.23804, "grad_norm": 1.5982794283903097, "learning_rate": 0.003, "loss": 4.0426, "step": 23804 }, { "epoch": 0.23805, "grad_norm": 1.5213386475082742, "learning_rate": 0.003, "loss": 4.0333, "step": 23805 }, { "epoch": 0.23806, "grad_norm": 1.296312226802364, "learning_rate": 0.003, "loss": 4.0446, "step": 23806 }, { "epoch": 0.23807, "grad_norm": 1.2186993021713506, "learning_rate": 0.003, "loss": 4.0283, "step": 23807 }, { "epoch": 0.23808, "grad_norm": 1.5581545633728306, "learning_rate": 0.003, "loss": 4.0311, "step": 23808 }, { "epoch": 0.23809, "grad_norm": 1.0839293973349182, "learning_rate": 0.003, "loss": 4.0404, "step": 23809 }, { "epoch": 0.2381, "grad_norm": 1.4127437288819267, "learning_rate": 0.003, "loss": 4.073, "step": 23810 }, { "epoch": 0.23811, "grad_norm": 0.9772413731718163, "learning_rate": 0.003, "loss": 4.0565, "step": 23811 }, { "epoch": 0.23812, "grad_norm": 1.2074401285427336, "learning_rate": 0.003, "loss": 4.0543, "step": 23812 }, { "epoch": 0.23813, "grad_norm": 1.2826858301353392, "learning_rate": 0.003, "loss": 4.0576, "step": 23813 }, { "epoch": 0.23814, "grad_norm": 1.3832318258593754, "learning_rate": 0.003, "loss": 4.0816, "step": 23814 }, { "epoch": 0.23815, "grad_norm": 1.4700508680675612, "learning_rate": 0.003, "loss": 4.0741, "step": 23815 }, { "epoch": 0.23816, "grad_norm": 1.0910979547311042, "learning_rate": 0.003, "loss": 4.0359, "step": 23816 }, { "epoch": 0.23817, "grad_norm": 1.5588218804038259, "learning_rate": 0.003, "loss": 4.0989, "step": 23817 }, { "epoch": 0.23818, "grad_norm": 1.3386480825097924, "learning_rate": 0.003, "loss": 4.0585, "step": 23818 }, { "epoch": 0.23819, "grad_norm": 1.5134689319234114, "learning_rate": 0.003, "loss": 4.0638, "step": 23819 }, { "epoch": 0.2382, "grad_norm": 1.241625547934892, "learning_rate": 0.003, "loss": 4.0525, "step": 23820 }, { "epoch": 0.23821, "grad_norm": 1.3534401482195397, "learning_rate": 0.003, "loss": 4.0577, "step": 23821 }, { "epoch": 0.23822, "grad_norm": 1.1628791461240653, "learning_rate": 0.003, "loss": 4.0644, "step": 23822 }, { "epoch": 0.23823, "grad_norm": 1.5041129084766678, "learning_rate": 0.003, "loss": 4.0567, "step": 23823 }, { "epoch": 0.23824, "grad_norm": 1.1665011979447957, "learning_rate": 0.003, "loss": 4.0524, "step": 23824 }, { "epoch": 0.23825, "grad_norm": 1.2121294063615315, "learning_rate": 0.003, "loss": 4.0375, "step": 23825 }, { "epoch": 0.23826, "grad_norm": 1.2558788993358732, "learning_rate": 0.003, "loss": 4.0168, "step": 23826 }, { "epoch": 0.23827, "grad_norm": 1.2735924026925578, "learning_rate": 0.003, "loss": 4.0227, "step": 23827 }, { "epoch": 0.23828, "grad_norm": 1.3808488526296263, "learning_rate": 0.003, "loss": 4.0326, "step": 23828 }, { "epoch": 0.23829, "grad_norm": 1.2224072738269918, "learning_rate": 0.003, "loss": 4.041, "step": 23829 }, { "epoch": 0.2383, "grad_norm": 1.5641554627001137, "learning_rate": 0.003, "loss": 4.0686, "step": 23830 }, { "epoch": 0.23831, "grad_norm": 1.2992866981548012, "learning_rate": 0.003, "loss": 4.0372, "step": 23831 }, { "epoch": 0.23832, "grad_norm": 1.549147357978281, "learning_rate": 0.003, "loss": 4.0514, "step": 23832 }, { "epoch": 0.23833, "grad_norm": 1.0098718573679912, "learning_rate": 0.003, "loss": 4.0109, "step": 23833 }, { "epoch": 0.23834, "grad_norm": 1.4027929190896795, "learning_rate": 0.003, "loss": 4.0117, "step": 23834 }, { "epoch": 0.23835, "grad_norm": 1.4908354679872946, "learning_rate": 0.003, "loss": 4.0196, "step": 23835 }, { "epoch": 0.23836, "grad_norm": 1.2272859919591117, "learning_rate": 0.003, "loss": 4.0422, "step": 23836 }, { "epoch": 0.23837, "grad_norm": 1.4516299417369554, "learning_rate": 0.003, "loss": 4.0861, "step": 23837 }, { "epoch": 0.23838, "grad_norm": 1.160717661601082, "learning_rate": 0.003, "loss": 4.0479, "step": 23838 }, { "epoch": 0.23839, "grad_norm": 1.3644040754585327, "learning_rate": 0.003, "loss": 4.0497, "step": 23839 }, { "epoch": 0.2384, "grad_norm": 1.2573914287935082, "learning_rate": 0.003, "loss": 4.0222, "step": 23840 }, { "epoch": 0.23841, "grad_norm": 1.4239647317799968, "learning_rate": 0.003, "loss": 4.0472, "step": 23841 }, { "epoch": 0.23842, "grad_norm": 1.249961617385262, "learning_rate": 0.003, "loss": 4.0581, "step": 23842 }, { "epoch": 0.23843, "grad_norm": 1.2642193504748251, "learning_rate": 0.003, "loss": 4.0539, "step": 23843 }, { "epoch": 0.23844, "grad_norm": 1.2590280416023358, "learning_rate": 0.003, "loss": 4.0611, "step": 23844 }, { "epoch": 0.23845, "grad_norm": 1.2844352399921637, "learning_rate": 0.003, "loss": 4.0552, "step": 23845 }, { "epoch": 0.23846, "grad_norm": 1.3920554222875168, "learning_rate": 0.003, "loss": 4.0188, "step": 23846 }, { "epoch": 0.23847, "grad_norm": 1.124013501853405, "learning_rate": 0.003, "loss": 4.0419, "step": 23847 }, { "epoch": 0.23848, "grad_norm": 1.5373515416062635, "learning_rate": 0.003, "loss": 4.0226, "step": 23848 }, { "epoch": 0.23849, "grad_norm": 1.598751276652976, "learning_rate": 0.003, "loss": 4.0457, "step": 23849 }, { "epoch": 0.2385, "grad_norm": 1.377683284401099, "learning_rate": 0.003, "loss": 4.0477, "step": 23850 }, { "epoch": 0.23851, "grad_norm": 1.3489381886145224, "learning_rate": 0.003, "loss": 4.0629, "step": 23851 }, { "epoch": 0.23852, "grad_norm": 1.4180338550410077, "learning_rate": 0.003, "loss": 4.0477, "step": 23852 }, { "epoch": 0.23853, "grad_norm": 1.1771854326221283, "learning_rate": 0.003, "loss": 4.0567, "step": 23853 }, { "epoch": 0.23854, "grad_norm": 1.2383274412901384, "learning_rate": 0.003, "loss": 4.0337, "step": 23854 }, { "epoch": 0.23855, "grad_norm": 1.302489599276082, "learning_rate": 0.003, "loss": 4.0338, "step": 23855 }, { "epoch": 0.23856, "grad_norm": 1.1536083341454884, "learning_rate": 0.003, "loss": 4.0175, "step": 23856 }, { "epoch": 0.23857, "grad_norm": 1.3706374162366113, "learning_rate": 0.003, "loss": 4.0802, "step": 23857 }, { "epoch": 0.23858, "grad_norm": 1.0242977102554751, "learning_rate": 0.003, "loss": 4.0867, "step": 23858 }, { "epoch": 0.23859, "grad_norm": 1.7850463443733227, "learning_rate": 0.003, "loss": 4.0478, "step": 23859 }, { "epoch": 0.2386, "grad_norm": 1.0391871690177923, "learning_rate": 0.003, "loss": 4.051, "step": 23860 }, { "epoch": 0.23861, "grad_norm": 1.2922037975821985, "learning_rate": 0.003, "loss": 4.0198, "step": 23861 }, { "epoch": 0.23862, "grad_norm": 1.592220877943124, "learning_rate": 0.003, "loss": 4.0313, "step": 23862 }, { "epoch": 0.23863, "grad_norm": 1.4838739569698698, "learning_rate": 0.003, "loss": 4.0572, "step": 23863 }, { "epoch": 0.23864, "grad_norm": 1.0019969279981116, "learning_rate": 0.003, "loss": 4.0326, "step": 23864 }, { "epoch": 0.23865, "grad_norm": 1.504574838198865, "learning_rate": 0.003, "loss": 4.0521, "step": 23865 }, { "epoch": 0.23866, "grad_norm": 1.4502433564161168, "learning_rate": 0.003, "loss": 4.0493, "step": 23866 }, { "epoch": 0.23867, "grad_norm": 1.2194261114786873, "learning_rate": 0.003, "loss": 4.0735, "step": 23867 }, { "epoch": 0.23868, "grad_norm": 1.4199451992544048, "learning_rate": 0.003, "loss": 4.044, "step": 23868 }, { "epoch": 0.23869, "grad_norm": 0.9722460871001879, "learning_rate": 0.003, "loss": 4.0651, "step": 23869 }, { "epoch": 0.2387, "grad_norm": 1.3275709415085666, "learning_rate": 0.003, "loss": 4.081, "step": 23870 }, { "epoch": 0.23871, "grad_norm": 1.3124575739063766, "learning_rate": 0.003, "loss": 4.0367, "step": 23871 }, { "epoch": 0.23872, "grad_norm": 1.12416231680309, "learning_rate": 0.003, "loss": 4.0469, "step": 23872 }, { "epoch": 0.23873, "grad_norm": 1.3655021803586156, "learning_rate": 0.003, "loss": 4.0322, "step": 23873 }, { "epoch": 0.23874, "grad_norm": 1.3642139417615187, "learning_rate": 0.003, "loss": 4.0427, "step": 23874 }, { "epoch": 0.23875, "grad_norm": 1.4203298654199221, "learning_rate": 0.003, "loss": 4.0445, "step": 23875 }, { "epoch": 0.23876, "grad_norm": 1.3287913685715587, "learning_rate": 0.003, "loss": 4.0322, "step": 23876 }, { "epoch": 0.23877, "grad_norm": 1.1979706403322938, "learning_rate": 0.003, "loss": 4.0483, "step": 23877 }, { "epoch": 0.23878, "grad_norm": 1.398354538785704, "learning_rate": 0.003, "loss": 4.0205, "step": 23878 }, { "epoch": 0.23879, "grad_norm": 1.4483339112877203, "learning_rate": 0.003, "loss": 4.0316, "step": 23879 }, { "epoch": 0.2388, "grad_norm": 1.3617929015169337, "learning_rate": 0.003, "loss": 4.0452, "step": 23880 }, { "epoch": 0.23881, "grad_norm": 1.2376669318853526, "learning_rate": 0.003, "loss": 4.0521, "step": 23881 }, { "epoch": 0.23882, "grad_norm": 1.2989851904192538, "learning_rate": 0.003, "loss": 4.0282, "step": 23882 }, { "epoch": 0.23883, "grad_norm": 1.2192240097749598, "learning_rate": 0.003, "loss": 4.0681, "step": 23883 }, { "epoch": 0.23884, "grad_norm": 1.5565234952957556, "learning_rate": 0.003, "loss": 4.0451, "step": 23884 }, { "epoch": 0.23885, "grad_norm": 1.2852851775894476, "learning_rate": 0.003, "loss": 4.0566, "step": 23885 }, { "epoch": 0.23886, "grad_norm": 1.0855797188437504, "learning_rate": 0.003, "loss": 4.0126, "step": 23886 }, { "epoch": 0.23887, "grad_norm": 1.3733923882090262, "learning_rate": 0.003, "loss": 4.0391, "step": 23887 }, { "epoch": 0.23888, "grad_norm": 1.2477493118859189, "learning_rate": 0.003, "loss": 4.0268, "step": 23888 }, { "epoch": 0.23889, "grad_norm": 1.378770132035632, "learning_rate": 0.003, "loss": 4.0139, "step": 23889 }, { "epoch": 0.2389, "grad_norm": 1.1236372215200345, "learning_rate": 0.003, "loss": 4.0584, "step": 23890 }, { "epoch": 0.23891, "grad_norm": 1.4752441865290327, "learning_rate": 0.003, "loss": 4.0455, "step": 23891 }, { "epoch": 0.23892, "grad_norm": 1.491994115649753, "learning_rate": 0.003, "loss": 4.0241, "step": 23892 }, { "epoch": 0.23893, "grad_norm": 1.3611222759248516, "learning_rate": 0.003, "loss": 4.0527, "step": 23893 }, { "epoch": 0.23894, "grad_norm": 1.3772955454325002, "learning_rate": 0.003, "loss": 4.0696, "step": 23894 }, { "epoch": 0.23895, "grad_norm": 1.1042473469982372, "learning_rate": 0.003, "loss": 4.0396, "step": 23895 }, { "epoch": 0.23896, "grad_norm": 1.2933288487486887, "learning_rate": 0.003, "loss": 4.0191, "step": 23896 }, { "epoch": 0.23897, "grad_norm": 1.3700696460329007, "learning_rate": 0.003, "loss": 4.0438, "step": 23897 }, { "epoch": 0.23898, "grad_norm": 1.1711112000889272, "learning_rate": 0.003, "loss": 4.0514, "step": 23898 }, { "epoch": 0.23899, "grad_norm": 1.3390976906059344, "learning_rate": 0.003, "loss": 4.0173, "step": 23899 }, { "epoch": 0.239, "grad_norm": 1.343183119442978, "learning_rate": 0.003, "loss": 4.0315, "step": 23900 }, { "epoch": 0.23901, "grad_norm": 1.1599047190899325, "learning_rate": 0.003, "loss": 4.0459, "step": 23901 }, { "epoch": 0.23902, "grad_norm": 1.3154615208285156, "learning_rate": 0.003, "loss": 4.0454, "step": 23902 }, { "epoch": 0.23903, "grad_norm": 1.2701478756736824, "learning_rate": 0.003, "loss": 4.0745, "step": 23903 }, { "epoch": 0.23904, "grad_norm": 1.3176561557078468, "learning_rate": 0.003, "loss": 4.0422, "step": 23904 }, { "epoch": 0.23905, "grad_norm": 1.517375181933842, "learning_rate": 0.003, "loss": 4.0401, "step": 23905 }, { "epoch": 0.23906, "grad_norm": 1.2153736528093153, "learning_rate": 0.003, "loss": 4.0416, "step": 23906 }, { "epoch": 0.23907, "grad_norm": 1.2966550047377552, "learning_rate": 0.003, "loss": 4.0465, "step": 23907 }, { "epoch": 0.23908, "grad_norm": 1.42990230247806, "learning_rate": 0.003, "loss": 4.0505, "step": 23908 }, { "epoch": 0.23909, "grad_norm": 1.4349264854921642, "learning_rate": 0.003, "loss": 4.0389, "step": 23909 }, { "epoch": 0.2391, "grad_norm": 1.4429118568525954, "learning_rate": 0.003, "loss": 4.0555, "step": 23910 }, { "epoch": 0.23911, "grad_norm": 0.9919740989677, "learning_rate": 0.003, "loss": 4.039, "step": 23911 }, { "epoch": 0.23912, "grad_norm": 1.6443804619890663, "learning_rate": 0.003, "loss": 4.0541, "step": 23912 }, { "epoch": 0.23913, "grad_norm": 1.1913209952388295, "learning_rate": 0.003, "loss": 4.0314, "step": 23913 }, { "epoch": 0.23914, "grad_norm": 1.4260590770010235, "learning_rate": 0.003, "loss": 4.0567, "step": 23914 }, { "epoch": 0.23915, "grad_norm": 1.2864533400980362, "learning_rate": 0.003, "loss": 4.0383, "step": 23915 }, { "epoch": 0.23916, "grad_norm": 1.174091255374752, "learning_rate": 0.003, "loss": 4.0383, "step": 23916 }, { "epoch": 0.23917, "grad_norm": 1.245240603183376, "learning_rate": 0.003, "loss": 4.0247, "step": 23917 }, { "epoch": 0.23918, "grad_norm": 1.1735040253911069, "learning_rate": 0.003, "loss": 4.025, "step": 23918 }, { "epoch": 0.23919, "grad_norm": 1.3955068968486166, "learning_rate": 0.003, "loss": 4.082, "step": 23919 }, { "epoch": 0.2392, "grad_norm": 1.343259166379999, "learning_rate": 0.003, "loss": 4.0393, "step": 23920 }, { "epoch": 0.23921, "grad_norm": 1.1960584462807435, "learning_rate": 0.003, "loss": 4.0461, "step": 23921 }, { "epoch": 0.23922, "grad_norm": 1.4362051634626094, "learning_rate": 0.003, "loss": 4.0741, "step": 23922 }, { "epoch": 0.23923, "grad_norm": 1.0843870193553204, "learning_rate": 0.003, "loss": 4.0113, "step": 23923 }, { "epoch": 0.23924, "grad_norm": 1.8139929167710738, "learning_rate": 0.003, "loss": 4.0597, "step": 23924 }, { "epoch": 0.23925, "grad_norm": 1.102214726939177, "learning_rate": 0.003, "loss": 4.0368, "step": 23925 }, { "epoch": 0.23926, "grad_norm": 1.361268818889144, "learning_rate": 0.003, "loss": 4.07, "step": 23926 }, { "epoch": 0.23927, "grad_norm": 1.2781462400356522, "learning_rate": 0.003, "loss": 4.0592, "step": 23927 }, { "epoch": 0.23928, "grad_norm": 1.4085237853480144, "learning_rate": 0.003, "loss": 4.0405, "step": 23928 }, { "epoch": 0.23929, "grad_norm": 1.1299103698118826, "learning_rate": 0.003, "loss": 4.0458, "step": 23929 }, { "epoch": 0.2393, "grad_norm": 1.1193149770907784, "learning_rate": 0.003, "loss": 4.0159, "step": 23930 }, { "epoch": 0.23931, "grad_norm": 1.2000036179737923, "learning_rate": 0.003, "loss": 4.0753, "step": 23931 }, { "epoch": 0.23932, "grad_norm": 1.3428055454012973, "learning_rate": 0.003, "loss": 4.0443, "step": 23932 }, { "epoch": 0.23933, "grad_norm": 1.1231079789096867, "learning_rate": 0.003, "loss": 4.0228, "step": 23933 }, { "epoch": 0.23934, "grad_norm": 1.2047630901118993, "learning_rate": 0.003, "loss": 4.0157, "step": 23934 }, { "epoch": 0.23935, "grad_norm": 1.349759862704681, "learning_rate": 0.003, "loss": 4.0446, "step": 23935 }, { "epoch": 0.23936, "grad_norm": 1.1590498799121691, "learning_rate": 0.003, "loss": 4.0362, "step": 23936 }, { "epoch": 0.23937, "grad_norm": 1.684076940402808, "learning_rate": 0.003, "loss": 4.0384, "step": 23937 }, { "epoch": 0.23938, "grad_norm": 1.0512109044816607, "learning_rate": 0.003, "loss": 4.0708, "step": 23938 }, { "epoch": 0.23939, "grad_norm": 1.58449620312223, "learning_rate": 0.003, "loss": 4.0443, "step": 23939 }, { "epoch": 0.2394, "grad_norm": 1.1058617777341166, "learning_rate": 0.003, "loss": 4.0671, "step": 23940 }, { "epoch": 0.23941, "grad_norm": 1.4603807552559986, "learning_rate": 0.003, "loss": 4.0293, "step": 23941 }, { "epoch": 0.23942, "grad_norm": 1.3355246499353914, "learning_rate": 0.003, "loss": 4.0599, "step": 23942 }, { "epoch": 0.23943, "grad_norm": 1.130022072646399, "learning_rate": 0.003, "loss": 4.0194, "step": 23943 }, { "epoch": 0.23944, "grad_norm": 1.2861785264164831, "learning_rate": 0.003, "loss": 4.0416, "step": 23944 }, { "epoch": 0.23945, "grad_norm": 1.4141456332668252, "learning_rate": 0.003, "loss": 4.0416, "step": 23945 }, { "epoch": 0.23946, "grad_norm": 1.4075912134877817, "learning_rate": 0.003, "loss": 4.0271, "step": 23946 }, { "epoch": 0.23947, "grad_norm": 1.3455042193960427, "learning_rate": 0.003, "loss": 4.0511, "step": 23947 }, { "epoch": 0.23948, "grad_norm": 1.395223326067633, "learning_rate": 0.003, "loss": 4.043, "step": 23948 }, { "epoch": 0.23949, "grad_norm": 1.3483496449754433, "learning_rate": 0.003, "loss": 4.0403, "step": 23949 }, { "epoch": 0.2395, "grad_norm": 1.651674997521108, "learning_rate": 0.003, "loss": 4.0628, "step": 23950 }, { "epoch": 0.23951, "grad_norm": 0.9889639249406295, "learning_rate": 0.003, "loss": 4.0667, "step": 23951 }, { "epoch": 0.23952, "grad_norm": 1.3426795766213986, "learning_rate": 0.003, "loss": 4.0284, "step": 23952 }, { "epoch": 0.23953, "grad_norm": 1.3464411438419432, "learning_rate": 0.003, "loss": 4.0526, "step": 23953 }, { "epoch": 0.23954, "grad_norm": 1.3850160219657013, "learning_rate": 0.003, "loss": 4.0417, "step": 23954 }, { "epoch": 0.23955, "grad_norm": 1.3406011368608362, "learning_rate": 0.003, "loss": 4.068, "step": 23955 }, { "epoch": 0.23956, "grad_norm": 1.187828698337367, "learning_rate": 0.003, "loss": 4.0719, "step": 23956 }, { "epoch": 0.23957, "grad_norm": 1.2188035987237125, "learning_rate": 0.003, "loss": 4.0222, "step": 23957 }, { "epoch": 0.23958, "grad_norm": 1.4967539978718947, "learning_rate": 0.003, "loss": 4.0749, "step": 23958 }, { "epoch": 0.23959, "grad_norm": 1.1673118162064886, "learning_rate": 0.003, "loss": 4.0542, "step": 23959 }, { "epoch": 0.2396, "grad_norm": 1.3084845718881184, "learning_rate": 0.003, "loss": 4.0388, "step": 23960 }, { "epoch": 0.23961, "grad_norm": 1.3004831964065695, "learning_rate": 0.003, "loss": 4.0217, "step": 23961 }, { "epoch": 0.23962, "grad_norm": 1.2292601474880807, "learning_rate": 0.003, "loss": 4.0589, "step": 23962 }, { "epoch": 0.23963, "grad_norm": 1.2771172613658468, "learning_rate": 0.003, "loss": 4.0114, "step": 23963 }, { "epoch": 0.23964, "grad_norm": 1.3711370469195556, "learning_rate": 0.003, "loss": 4.0449, "step": 23964 }, { "epoch": 0.23965, "grad_norm": 1.5315316601779236, "learning_rate": 0.003, "loss": 4.0408, "step": 23965 }, { "epoch": 0.23966, "grad_norm": 1.2648013543316192, "learning_rate": 0.003, "loss": 4.0692, "step": 23966 }, { "epoch": 0.23967, "grad_norm": 1.2972828432340149, "learning_rate": 0.003, "loss": 4.0576, "step": 23967 }, { "epoch": 0.23968, "grad_norm": 1.242978208396597, "learning_rate": 0.003, "loss": 4.0765, "step": 23968 }, { "epoch": 0.23969, "grad_norm": 1.5614535313214628, "learning_rate": 0.003, "loss": 4.0476, "step": 23969 }, { "epoch": 0.2397, "grad_norm": 1.1211391780293622, "learning_rate": 0.003, "loss": 4.0318, "step": 23970 }, { "epoch": 0.23971, "grad_norm": 1.402249127312032, "learning_rate": 0.003, "loss": 4.067, "step": 23971 }, { "epoch": 0.23972, "grad_norm": 1.3028140291524055, "learning_rate": 0.003, "loss": 4.0312, "step": 23972 }, { "epoch": 0.23973, "grad_norm": 1.2924055010066422, "learning_rate": 0.003, "loss": 4.0659, "step": 23973 }, { "epoch": 0.23974, "grad_norm": 1.2035223173514789, "learning_rate": 0.003, "loss": 4.0552, "step": 23974 }, { "epoch": 0.23975, "grad_norm": 1.2502018478978507, "learning_rate": 0.003, "loss": 4.0349, "step": 23975 }, { "epoch": 0.23976, "grad_norm": 1.1864741266835643, "learning_rate": 0.003, "loss": 4.0592, "step": 23976 }, { "epoch": 0.23977, "grad_norm": 1.4745737306346887, "learning_rate": 0.003, "loss": 4.0427, "step": 23977 }, { "epoch": 0.23978, "grad_norm": 1.3180764479988878, "learning_rate": 0.003, "loss": 4.0189, "step": 23978 }, { "epoch": 0.23979, "grad_norm": 1.2358016750429341, "learning_rate": 0.003, "loss": 4.0557, "step": 23979 }, { "epoch": 0.2398, "grad_norm": 1.6022903238570678, "learning_rate": 0.003, "loss": 4.0402, "step": 23980 }, { "epoch": 0.23981, "grad_norm": 1.1999729232944245, "learning_rate": 0.003, "loss": 4.0721, "step": 23981 }, { "epoch": 0.23982, "grad_norm": 1.563962617779391, "learning_rate": 0.003, "loss": 4.0405, "step": 23982 }, { "epoch": 0.23983, "grad_norm": 1.0149302408333853, "learning_rate": 0.003, "loss": 4.0259, "step": 23983 }, { "epoch": 0.23984, "grad_norm": 1.4591034645874765, "learning_rate": 0.003, "loss": 4.0524, "step": 23984 }, { "epoch": 0.23985, "grad_norm": 1.5435033214114402, "learning_rate": 0.003, "loss": 4.0239, "step": 23985 }, { "epoch": 0.23986, "grad_norm": 1.1173463970891755, "learning_rate": 0.003, "loss": 4.0205, "step": 23986 }, { "epoch": 0.23987, "grad_norm": 1.3489871220510115, "learning_rate": 0.003, "loss": 4.041, "step": 23987 }, { "epoch": 0.23988, "grad_norm": 1.2899771570825291, "learning_rate": 0.003, "loss": 4.0579, "step": 23988 }, { "epoch": 0.23989, "grad_norm": 1.4604472197547178, "learning_rate": 0.003, "loss": 4.0198, "step": 23989 }, { "epoch": 0.2399, "grad_norm": 1.0023798980935106, "learning_rate": 0.003, "loss": 4.0687, "step": 23990 }, { "epoch": 0.23991, "grad_norm": 1.6076021486754355, "learning_rate": 0.003, "loss": 4.0425, "step": 23991 }, { "epoch": 0.23992, "grad_norm": 1.1405577515065455, "learning_rate": 0.003, "loss": 4.0367, "step": 23992 }, { "epoch": 0.23993, "grad_norm": 1.5959655985346612, "learning_rate": 0.003, "loss": 4.0406, "step": 23993 }, { "epoch": 0.23994, "grad_norm": 1.04858574801776, "learning_rate": 0.003, "loss": 4.0511, "step": 23994 }, { "epoch": 0.23995, "grad_norm": 1.5945825663224402, "learning_rate": 0.003, "loss": 4.0369, "step": 23995 }, { "epoch": 0.23996, "grad_norm": 1.4518659707503745, "learning_rate": 0.003, "loss": 4.0556, "step": 23996 }, { "epoch": 0.23997, "grad_norm": 1.1267224429043454, "learning_rate": 0.003, "loss": 4.0559, "step": 23997 }, { "epoch": 0.23998, "grad_norm": 1.462903258157101, "learning_rate": 0.003, "loss": 4.0494, "step": 23998 }, { "epoch": 0.23999, "grad_norm": 1.1498469435385599, "learning_rate": 0.003, "loss": 4.027, "step": 23999 }, { "epoch": 0.24, "grad_norm": 1.3310701791888737, "learning_rate": 0.003, "loss": 4.0662, "step": 24000 }, { "epoch": 0.24001, "grad_norm": 1.1982137549497318, "learning_rate": 0.003, "loss": 4.0367, "step": 24001 }, { "epoch": 0.24002, "grad_norm": 1.2646313314859452, "learning_rate": 0.003, "loss": 4.04, "step": 24002 }, { "epoch": 0.24003, "grad_norm": 1.3246323138326386, "learning_rate": 0.003, "loss": 4.0626, "step": 24003 }, { "epoch": 0.24004, "grad_norm": 1.3850664593683781, "learning_rate": 0.003, "loss": 4.0578, "step": 24004 }, { "epoch": 0.24005, "grad_norm": 1.0703186089822372, "learning_rate": 0.003, "loss": 4.0317, "step": 24005 }, { "epoch": 0.24006, "grad_norm": 1.629681560304667, "learning_rate": 0.003, "loss": 4.054, "step": 24006 }, { "epoch": 0.24007, "grad_norm": 1.1777003924327751, "learning_rate": 0.003, "loss": 4.0339, "step": 24007 }, { "epoch": 0.24008, "grad_norm": 1.3143147464021314, "learning_rate": 0.003, "loss": 4.0627, "step": 24008 }, { "epoch": 0.24009, "grad_norm": 1.4833433096793711, "learning_rate": 0.003, "loss": 4.0667, "step": 24009 }, { "epoch": 0.2401, "grad_norm": 1.2583862993411046, "learning_rate": 0.003, "loss": 4.059, "step": 24010 }, { "epoch": 0.24011, "grad_norm": 1.1851522531391423, "learning_rate": 0.003, "loss": 4.0337, "step": 24011 }, { "epoch": 0.24012, "grad_norm": 1.5723383028490112, "learning_rate": 0.003, "loss": 4.0466, "step": 24012 }, { "epoch": 0.24013, "grad_norm": 1.158314434475222, "learning_rate": 0.003, "loss": 4.0514, "step": 24013 }, { "epoch": 0.24014, "grad_norm": 1.357320373169641, "learning_rate": 0.003, "loss": 4.0442, "step": 24014 }, { "epoch": 0.24015, "grad_norm": 1.1867233915244144, "learning_rate": 0.003, "loss": 4.0215, "step": 24015 }, { "epoch": 0.24016, "grad_norm": 1.3372652701141177, "learning_rate": 0.003, "loss": 4.0485, "step": 24016 }, { "epoch": 0.24017, "grad_norm": 1.5583296043937156, "learning_rate": 0.003, "loss": 4.0482, "step": 24017 }, { "epoch": 0.24018, "grad_norm": 1.4495285214185716, "learning_rate": 0.003, "loss": 4.0479, "step": 24018 }, { "epoch": 0.24019, "grad_norm": 1.2649948758687706, "learning_rate": 0.003, "loss": 4.0469, "step": 24019 }, { "epoch": 0.2402, "grad_norm": 1.2851584101425262, "learning_rate": 0.003, "loss": 4.0511, "step": 24020 }, { "epoch": 0.24021, "grad_norm": 1.1646020783663935, "learning_rate": 0.003, "loss": 4.0179, "step": 24021 }, { "epoch": 0.24022, "grad_norm": 1.2377366689250193, "learning_rate": 0.003, "loss": 4.0615, "step": 24022 }, { "epoch": 0.24023, "grad_norm": 1.1456376136850606, "learning_rate": 0.003, "loss": 4.0253, "step": 24023 }, { "epoch": 0.24024, "grad_norm": 1.3262440424356252, "learning_rate": 0.003, "loss": 4.0331, "step": 24024 }, { "epoch": 0.24025, "grad_norm": 1.1368202150579607, "learning_rate": 0.003, "loss": 4.0345, "step": 24025 }, { "epoch": 0.24026, "grad_norm": 1.5654731240234023, "learning_rate": 0.003, "loss": 4.0464, "step": 24026 }, { "epoch": 0.24027, "grad_norm": 1.2113056981443557, "learning_rate": 0.003, "loss": 4.0745, "step": 24027 }, { "epoch": 0.24028, "grad_norm": 1.2880896491858853, "learning_rate": 0.003, "loss": 4.0726, "step": 24028 }, { "epoch": 0.24029, "grad_norm": 1.2484307292541577, "learning_rate": 0.003, "loss": 4.0226, "step": 24029 }, { "epoch": 0.2403, "grad_norm": 1.3369505276842217, "learning_rate": 0.003, "loss": 4.0593, "step": 24030 }, { "epoch": 0.24031, "grad_norm": 1.2610814423552485, "learning_rate": 0.003, "loss": 4.0404, "step": 24031 }, { "epoch": 0.24032, "grad_norm": 1.5951341796225895, "learning_rate": 0.003, "loss": 4.0041, "step": 24032 }, { "epoch": 0.24033, "grad_norm": 1.0315160471783094, "learning_rate": 0.003, "loss": 4.045, "step": 24033 }, { "epoch": 0.24034, "grad_norm": 1.6147503672621737, "learning_rate": 0.003, "loss": 4.0775, "step": 24034 }, { "epoch": 0.24035, "grad_norm": 1.1054836851268373, "learning_rate": 0.003, "loss": 4.05, "step": 24035 }, { "epoch": 0.24036, "grad_norm": 1.3727561803944797, "learning_rate": 0.003, "loss": 4.0442, "step": 24036 }, { "epoch": 0.24037, "grad_norm": 1.2110746012597284, "learning_rate": 0.003, "loss": 4.0383, "step": 24037 }, { "epoch": 0.24038, "grad_norm": 1.3168542529178489, "learning_rate": 0.003, "loss": 4.0421, "step": 24038 }, { "epoch": 0.24039, "grad_norm": 1.2606419184589108, "learning_rate": 0.003, "loss": 4.0485, "step": 24039 }, { "epoch": 0.2404, "grad_norm": 1.5938366407875177, "learning_rate": 0.003, "loss": 4.0589, "step": 24040 }, { "epoch": 0.24041, "grad_norm": 1.1955604830900401, "learning_rate": 0.003, "loss": 4.0387, "step": 24041 }, { "epoch": 0.24042, "grad_norm": 1.3797002199196038, "learning_rate": 0.003, "loss": 4.0409, "step": 24042 }, { "epoch": 0.24043, "grad_norm": 1.304431247908543, "learning_rate": 0.003, "loss": 4.0723, "step": 24043 }, { "epoch": 0.24044, "grad_norm": 1.5420504565226043, "learning_rate": 0.003, "loss": 4.0332, "step": 24044 }, { "epoch": 0.24045, "grad_norm": 1.1785198375684904, "learning_rate": 0.003, "loss": 4.0465, "step": 24045 }, { "epoch": 0.24046, "grad_norm": 1.2507553757849583, "learning_rate": 0.003, "loss": 4.0396, "step": 24046 }, { "epoch": 0.24047, "grad_norm": 1.5860365861893448, "learning_rate": 0.003, "loss": 4.0673, "step": 24047 }, { "epoch": 0.24048, "grad_norm": 1.1769672222956735, "learning_rate": 0.003, "loss": 4.0462, "step": 24048 }, { "epoch": 0.24049, "grad_norm": 1.3754960823087479, "learning_rate": 0.003, "loss": 4.0476, "step": 24049 }, { "epoch": 0.2405, "grad_norm": 1.2091215555240222, "learning_rate": 0.003, "loss": 4.0419, "step": 24050 }, { "epoch": 0.24051, "grad_norm": 1.3007701248629797, "learning_rate": 0.003, "loss": 4.0507, "step": 24051 }, { "epoch": 0.24052, "grad_norm": 1.3936599313191824, "learning_rate": 0.003, "loss": 4.046, "step": 24052 }, { "epoch": 0.24053, "grad_norm": 1.216248177628059, "learning_rate": 0.003, "loss": 4.0412, "step": 24053 }, { "epoch": 0.24054, "grad_norm": 1.572958969016613, "learning_rate": 0.003, "loss": 4.0468, "step": 24054 }, { "epoch": 0.24055, "grad_norm": 1.2010040821767707, "learning_rate": 0.003, "loss": 4.0462, "step": 24055 }, { "epoch": 0.24056, "grad_norm": 1.287138029491045, "learning_rate": 0.003, "loss": 4.0275, "step": 24056 }, { "epoch": 0.24057, "grad_norm": 1.4102215201793016, "learning_rate": 0.003, "loss": 4.0174, "step": 24057 }, { "epoch": 0.24058, "grad_norm": 1.0508476097811532, "learning_rate": 0.003, "loss": 4.0296, "step": 24058 }, { "epoch": 0.24059, "grad_norm": 1.3301794685429869, "learning_rate": 0.003, "loss": 4.0558, "step": 24059 }, { "epoch": 0.2406, "grad_norm": 1.2957012304118594, "learning_rate": 0.003, "loss": 4.0438, "step": 24060 }, { "epoch": 0.24061, "grad_norm": 1.4571789840800085, "learning_rate": 0.003, "loss": 4.0413, "step": 24061 }, { "epoch": 0.24062, "grad_norm": 1.0925374024588101, "learning_rate": 0.003, "loss": 4.0321, "step": 24062 }, { "epoch": 0.24063, "grad_norm": 1.3884803422530934, "learning_rate": 0.003, "loss": 4.0546, "step": 24063 }, { "epoch": 0.24064, "grad_norm": 1.2080720749884977, "learning_rate": 0.003, "loss": 4.006, "step": 24064 }, { "epoch": 0.24065, "grad_norm": 1.2546024585843576, "learning_rate": 0.003, "loss": 4.0833, "step": 24065 }, { "epoch": 0.24066, "grad_norm": 1.3292052379495949, "learning_rate": 0.003, "loss": 4.051, "step": 24066 }, { "epoch": 0.24067, "grad_norm": 1.1706258732078991, "learning_rate": 0.003, "loss": 4.0458, "step": 24067 }, { "epoch": 0.24068, "grad_norm": 1.2431780350750468, "learning_rate": 0.003, "loss": 4.0172, "step": 24068 }, { "epoch": 0.24069, "grad_norm": 1.2778698583267054, "learning_rate": 0.003, "loss": 4.0736, "step": 24069 }, { "epoch": 0.2407, "grad_norm": 1.4366405045158976, "learning_rate": 0.003, "loss": 4.048, "step": 24070 }, { "epoch": 0.24071, "grad_norm": 1.2773845164410116, "learning_rate": 0.003, "loss": 4.0361, "step": 24071 }, { "epoch": 0.24072, "grad_norm": 1.3043612377841143, "learning_rate": 0.003, "loss": 4.0562, "step": 24072 }, { "epoch": 0.24073, "grad_norm": 1.0605411956497945, "learning_rate": 0.003, "loss": 4.0288, "step": 24073 }, { "epoch": 0.24074, "grad_norm": 1.576054650539737, "learning_rate": 0.003, "loss": 4.0926, "step": 24074 }, { "epoch": 0.24075, "grad_norm": 1.4201987062640198, "learning_rate": 0.003, "loss": 4.0359, "step": 24075 }, { "epoch": 0.24076, "grad_norm": 1.4596407788197026, "learning_rate": 0.003, "loss": 4.0977, "step": 24076 }, { "epoch": 0.24077, "grad_norm": 1.1723779214160353, "learning_rate": 0.003, "loss": 4.0347, "step": 24077 }, { "epoch": 0.24078, "grad_norm": 1.3307489772858394, "learning_rate": 0.003, "loss": 4.0637, "step": 24078 }, { "epoch": 0.24079, "grad_norm": 1.2614043198420335, "learning_rate": 0.003, "loss": 4.0568, "step": 24079 }, { "epoch": 0.2408, "grad_norm": 1.18445922083957, "learning_rate": 0.003, "loss": 4.048, "step": 24080 }, { "epoch": 0.24081, "grad_norm": 1.2861209160470866, "learning_rate": 0.003, "loss": 4.0384, "step": 24081 }, { "epoch": 0.24082, "grad_norm": 1.3215214659767252, "learning_rate": 0.003, "loss": 4.0335, "step": 24082 }, { "epoch": 0.24083, "grad_norm": 1.342771261033664, "learning_rate": 0.003, "loss": 4.0512, "step": 24083 }, { "epoch": 0.24084, "grad_norm": 1.177998771956405, "learning_rate": 0.003, "loss": 4.0813, "step": 24084 }, { "epoch": 0.24085, "grad_norm": 1.2560780848427364, "learning_rate": 0.003, "loss": 4.0597, "step": 24085 }, { "epoch": 0.24086, "grad_norm": 1.2655013123576726, "learning_rate": 0.003, "loss": 4.0222, "step": 24086 }, { "epoch": 0.24087, "grad_norm": 1.401608379446091, "learning_rate": 0.003, "loss": 4.0544, "step": 24087 }, { "epoch": 0.24088, "grad_norm": 1.412674505098607, "learning_rate": 0.003, "loss": 4.0557, "step": 24088 }, { "epoch": 0.24089, "grad_norm": 1.338184524837426, "learning_rate": 0.003, "loss": 4.0399, "step": 24089 }, { "epoch": 0.2409, "grad_norm": 1.4202003864751755, "learning_rate": 0.003, "loss": 4.0286, "step": 24090 }, { "epoch": 0.24091, "grad_norm": 1.2523241458205954, "learning_rate": 0.003, "loss": 4.0439, "step": 24091 }, { "epoch": 0.24092, "grad_norm": 1.6625972411922345, "learning_rate": 0.003, "loss": 4.0307, "step": 24092 }, { "epoch": 0.24093, "grad_norm": 1.158347401400497, "learning_rate": 0.003, "loss": 3.9999, "step": 24093 }, { "epoch": 0.24094, "grad_norm": 1.4852284057617953, "learning_rate": 0.003, "loss": 4.0185, "step": 24094 }, { "epoch": 0.24095, "grad_norm": 1.0070997358920182, "learning_rate": 0.003, "loss": 4.0516, "step": 24095 }, { "epoch": 0.24096, "grad_norm": 1.3903977487078547, "learning_rate": 0.003, "loss": 4.028, "step": 24096 }, { "epoch": 0.24097, "grad_norm": 1.4053241640400327, "learning_rate": 0.003, "loss": 4.0571, "step": 24097 }, { "epoch": 0.24098, "grad_norm": 1.3461881361065275, "learning_rate": 0.003, "loss": 4.0514, "step": 24098 }, { "epoch": 0.24099, "grad_norm": 1.2159455208325554, "learning_rate": 0.003, "loss": 4.0526, "step": 24099 }, { "epoch": 0.241, "grad_norm": 1.1785523608312385, "learning_rate": 0.003, "loss": 4.0469, "step": 24100 }, { "epoch": 0.24101, "grad_norm": 1.4863888449565132, "learning_rate": 0.003, "loss": 4.0341, "step": 24101 }, { "epoch": 0.24102, "grad_norm": 0.9853543881409829, "learning_rate": 0.003, "loss": 4.0272, "step": 24102 }, { "epoch": 0.24103, "grad_norm": 1.51006194624815, "learning_rate": 0.003, "loss": 4.068, "step": 24103 }, { "epoch": 0.24104, "grad_norm": 1.3078043330706701, "learning_rate": 0.003, "loss": 4.0107, "step": 24104 }, { "epoch": 0.24105, "grad_norm": 1.2549421001517347, "learning_rate": 0.003, "loss": 4.0366, "step": 24105 }, { "epoch": 0.24106, "grad_norm": 1.3438400649769986, "learning_rate": 0.003, "loss": 4.0821, "step": 24106 }, { "epoch": 0.24107, "grad_norm": 1.151860758649009, "learning_rate": 0.003, "loss": 4.0255, "step": 24107 }, { "epoch": 0.24108, "grad_norm": 1.316968475541291, "learning_rate": 0.003, "loss": 4.0545, "step": 24108 }, { "epoch": 0.24109, "grad_norm": 1.3956319483161022, "learning_rate": 0.003, "loss": 4.023, "step": 24109 }, { "epoch": 0.2411, "grad_norm": 1.395446475941159, "learning_rate": 0.003, "loss": 4.0597, "step": 24110 }, { "epoch": 0.24111, "grad_norm": 1.1877786251620326, "learning_rate": 0.003, "loss": 4.0501, "step": 24111 }, { "epoch": 0.24112, "grad_norm": 2.061453801897167, "learning_rate": 0.003, "loss": 4.0582, "step": 24112 }, { "epoch": 0.24113, "grad_norm": 1.3566006701899251, "learning_rate": 0.003, "loss": 4.0744, "step": 24113 }, { "epoch": 0.24114, "grad_norm": 1.267988731299872, "learning_rate": 0.003, "loss": 4.0307, "step": 24114 }, { "epoch": 0.24115, "grad_norm": 1.2913275469196175, "learning_rate": 0.003, "loss": 4.028, "step": 24115 }, { "epoch": 0.24116, "grad_norm": 1.484784690043986, "learning_rate": 0.003, "loss": 4.0467, "step": 24116 }, { "epoch": 0.24117, "grad_norm": 1.217173827574909, "learning_rate": 0.003, "loss": 4.0369, "step": 24117 }, { "epoch": 0.24118, "grad_norm": 1.3945843985752577, "learning_rate": 0.003, "loss": 4.0277, "step": 24118 }, { "epoch": 0.24119, "grad_norm": 1.1373727010916976, "learning_rate": 0.003, "loss": 4.0628, "step": 24119 }, { "epoch": 0.2412, "grad_norm": 1.383729558410073, "learning_rate": 0.003, "loss": 4.0381, "step": 24120 }, { "epoch": 0.24121, "grad_norm": 1.0536316602535394, "learning_rate": 0.003, "loss": 4.0311, "step": 24121 }, { "epoch": 0.24122, "grad_norm": 1.61885902593575, "learning_rate": 0.003, "loss": 4.0074, "step": 24122 }, { "epoch": 0.24123, "grad_norm": 1.1815668812768427, "learning_rate": 0.003, "loss": 4.0511, "step": 24123 }, { "epoch": 0.24124, "grad_norm": 1.4004992415250197, "learning_rate": 0.003, "loss": 4.0368, "step": 24124 }, { "epoch": 0.24125, "grad_norm": 1.3826544078435026, "learning_rate": 0.003, "loss": 4.0643, "step": 24125 }, { "epoch": 0.24126, "grad_norm": 1.3714547949108018, "learning_rate": 0.003, "loss": 4.0413, "step": 24126 }, { "epoch": 0.24127, "grad_norm": 1.1468589706740249, "learning_rate": 0.003, "loss": 4.0112, "step": 24127 }, { "epoch": 0.24128, "grad_norm": 1.609320363080084, "learning_rate": 0.003, "loss": 4.0733, "step": 24128 }, { "epoch": 0.24129, "grad_norm": 1.0848059707492403, "learning_rate": 0.003, "loss": 4.0557, "step": 24129 }, { "epoch": 0.2413, "grad_norm": 1.397503821827954, "learning_rate": 0.003, "loss": 4.0487, "step": 24130 }, { "epoch": 0.24131, "grad_norm": 1.1307636939554195, "learning_rate": 0.003, "loss": 4.0486, "step": 24131 }, { "epoch": 0.24132, "grad_norm": 1.4307423328710627, "learning_rate": 0.003, "loss": 4.0291, "step": 24132 }, { "epoch": 0.24133, "grad_norm": 1.2789014925741, "learning_rate": 0.003, "loss": 4.0243, "step": 24133 }, { "epoch": 0.24134, "grad_norm": 1.5011241911903856, "learning_rate": 0.003, "loss": 4.0413, "step": 24134 }, { "epoch": 0.24135, "grad_norm": 1.0176273730996737, "learning_rate": 0.003, "loss": 4.0448, "step": 24135 }, { "epoch": 0.24136, "grad_norm": 1.3978111834325413, "learning_rate": 0.003, "loss": 4.0476, "step": 24136 }, { "epoch": 0.24137, "grad_norm": 1.446912261572512, "learning_rate": 0.003, "loss": 4.0323, "step": 24137 }, { "epoch": 0.24138, "grad_norm": 1.1992773910946033, "learning_rate": 0.003, "loss": 4.0261, "step": 24138 }, { "epoch": 0.24139, "grad_norm": 1.3476554319462777, "learning_rate": 0.003, "loss": 4.0422, "step": 24139 }, { "epoch": 0.2414, "grad_norm": 1.24275737536399, "learning_rate": 0.003, "loss": 4.0331, "step": 24140 }, { "epoch": 0.24141, "grad_norm": 1.4448544422715612, "learning_rate": 0.003, "loss": 4.0479, "step": 24141 }, { "epoch": 0.24142, "grad_norm": 1.1972062783899364, "learning_rate": 0.003, "loss": 4.0517, "step": 24142 }, { "epoch": 0.24143, "grad_norm": 1.7132485479267714, "learning_rate": 0.003, "loss": 4.0611, "step": 24143 }, { "epoch": 0.24144, "grad_norm": 1.0388734757674982, "learning_rate": 0.003, "loss": 4.042, "step": 24144 }, { "epoch": 0.24145, "grad_norm": 1.599562870156419, "learning_rate": 0.003, "loss": 4.0468, "step": 24145 }, { "epoch": 0.24146, "grad_norm": 1.1054069167389455, "learning_rate": 0.003, "loss": 4.0465, "step": 24146 }, { "epoch": 0.24147, "grad_norm": 1.4286220053188623, "learning_rate": 0.003, "loss": 4.0655, "step": 24147 }, { "epoch": 0.24148, "grad_norm": 1.1585888574766916, "learning_rate": 0.003, "loss": 4.0311, "step": 24148 }, { "epoch": 0.24149, "grad_norm": 1.5220214127417249, "learning_rate": 0.003, "loss": 4.0491, "step": 24149 }, { "epoch": 0.2415, "grad_norm": 1.2399500095587201, "learning_rate": 0.003, "loss": 4.0368, "step": 24150 }, { "epoch": 0.24151, "grad_norm": 1.4139993181087314, "learning_rate": 0.003, "loss": 4.0388, "step": 24151 }, { "epoch": 0.24152, "grad_norm": 1.1853165761252153, "learning_rate": 0.003, "loss": 4.0429, "step": 24152 }, { "epoch": 0.24153, "grad_norm": 1.3849721384973632, "learning_rate": 0.003, "loss": 4.0364, "step": 24153 }, { "epoch": 0.24154, "grad_norm": 1.2048305559811268, "learning_rate": 0.003, "loss": 4.0454, "step": 24154 }, { "epoch": 0.24155, "grad_norm": 1.215891108060119, "learning_rate": 0.003, "loss": 4.0641, "step": 24155 }, { "epoch": 0.24156, "grad_norm": 1.3866339481387726, "learning_rate": 0.003, "loss": 4.0471, "step": 24156 }, { "epoch": 0.24157, "grad_norm": 1.2027140089421595, "learning_rate": 0.003, "loss": 4.0461, "step": 24157 }, { "epoch": 0.24158, "grad_norm": 1.2553627133674687, "learning_rate": 0.003, "loss": 4.0255, "step": 24158 }, { "epoch": 0.24159, "grad_norm": 1.4002793767433759, "learning_rate": 0.003, "loss": 4.0478, "step": 24159 }, { "epoch": 0.2416, "grad_norm": 1.2452213689519354, "learning_rate": 0.003, "loss": 4.0412, "step": 24160 }, { "epoch": 0.24161, "grad_norm": 1.377309109897054, "learning_rate": 0.003, "loss": 4.0368, "step": 24161 }, { "epoch": 0.24162, "grad_norm": 1.1850278520964292, "learning_rate": 0.003, "loss": 4.066, "step": 24162 }, { "epoch": 0.24163, "grad_norm": 1.2707558088155932, "learning_rate": 0.003, "loss": 4.0536, "step": 24163 }, { "epoch": 0.24164, "grad_norm": 1.158081770626405, "learning_rate": 0.003, "loss": 4.0077, "step": 24164 }, { "epoch": 0.24165, "grad_norm": 1.3424118307531159, "learning_rate": 0.003, "loss": 4.0544, "step": 24165 }, { "epoch": 0.24166, "grad_norm": 1.1320121788540976, "learning_rate": 0.003, "loss": 4.0502, "step": 24166 }, { "epoch": 0.24167, "grad_norm": 1.4212208024498993, "learning_rate": 0.003, "loss": 4.0394, "step": 24167 }, { "epoch": 0.24168, "grad_norm": 1.2979045883112956, "learning_rate": 0.003, "loss": 4.0867, "step": 24168 }, { "epoch": 0.24169, "grad_norm": 1.4194937361222688, "learning_rate": 0.003, "loss": 4.0308, "step": 24169 }, { "epoch": 0.2417, "grad_norm": 1.3487465387417823, "learning_rate": 0.003, "loss": 4.0417, "step": 24170 }, { "epoch": 0.24171, "grad_norm": 1.3690426466054055, "learning_rate": 0.003, "loss": 4.0333, "step": 24171 }, { "epoch": 0.24172, "grad_norm": 1.4580168425443805, "learning_rate": 0.003, "loss": 4.0284, "step": 24172 }, { "epoch": 0.24173, "grad_norm": 1.1467184413618816, "learning_rate": 0.003, "loss": 4.0396, "step": 24173 }, { "epoch": 0.24174, "grad_norm": 1.413430395678707, "learning_rate": 0.003, "loss": 4.0104, "step": 24174 }, { "epoch": 0.24175, "grad_norm": 1.1808060653876629, "learning_rate": 0.003, "loss": 4.0402, "step": 24175 }, { "epoch": 0.24176, "grad_norm": 1.4933671896603207, "learning_rate": 0.003, "loss": 4.067, "step": 24176 }, { "epoch": 0.24177, "grad_norm": 1.2332074232620434, "learning_rate": 0.003, "loss": 4.0433, "step": 24177 }, { "epoch": 0.24178, "grad_norm": 1.4811142454039032, "learning_rate": 0.003, "loss": 4.0521, "step": 24178 }, { "epoch": 0.24179, "grad_norm": 1.114262381817086, "learning_rate": 0.003, "loss": 4.0221, "step": 24179 }, { "epoch": 0.2418, "grad_norm": 1.4990236239677852, "learning_rate": 0.003, "loss": 4.0404, "step": 24180 }, { "epoch": 0.24181, "grad_norm": 1.2373042967318026, "learning_rate": 0.003, "loss": 4.0207, "step": 24181 }, { "epoch": 0.24182, "grad_norm": 1.1170506943862697, "learning_rate": 0.003, "loss": 4.0413, "step": 24182 }, { "epoch": 0.24183, "grad_norm": 1.448564075754916, "learning_rate": 0.003, "loss": 4.0268, "step": 24183 }, { "epoch": 0.24184, "grad_norm": 1.010696038721016, "learning_rate": 0.003, "loss": 4.0467, "step": 24184 }, { "epoch": 0.24185, "grad_norm": 1.4877647877060574, "learning_rate": 0.003, "loss": 4.0244, "step": 24185 }, { "epoch": 0.24186, "grad_norm": 1.2688204068011295, "learning_rate": 0.003, "loss": 4.0687, "step": 24186 }, { "epoch": 0.24187, "grad_norm": 1.4597897955671664, "learning_rate": 0.003, "loss": 4.0589, "step": 24187 }, { "epoch": 0.24188, "grad_norm": 1.2855935580104527, "learning_rate": 0.003, "loss": 4.0689, "step": 24188 }, { "epoch": 0.24189, "grad_norm": 1.4047741224409784, "learning_rate": 0.003, "loss": 4.0485, "step": 24189 }, { "epoch": 0.2419, "grad_norm": 1.1240708415452145, "learning_rate": 0.003, "loss": 4.0595, "step": 24190 }, { "epoch": 0.24191, "grad_norm": 1.366042198017698, "learning_rate": 0.003, "loss": 4.0483, "step": 24191 }, { "epoch": 0.24192, "grad_norm": 1.1332015181389599, "learning_rate": 0.003, "loss": 4.0641, "step": 24192 }, { "epoch": 0.24193, "grad_norm": 1.4331915469323964, "learning_rate": 0.003, "loss": 4.0431, "step": 24193 }, { "epoch": 0.24194, "grad_norm": 1.3090691943231467, "learning_rate": 0.003, "loss": 4.0349, "step": 24194 }, { "epoch": 0.24195, "grad_norm": 1.368338182236077, "learning_rate": 0.003, "loss": 4.0282, "step": 24195 }, { "epoch": 0.24196, "grad_norm": 1.162640578033725, "learning_rate": 0.003, "loss": 4.0869, "step": 24196 }, { "epoch": 0.24197, "grad_norm": 1.3403368656488615, "learning_rate": 0.003, "loss": 4.0311, "step": 24197 }, { "epoch": 0.24198, "grad_norm": 1.1680758027513731, "learning_rate": 0.003, "loss": 4.0282, "step": 24198 }, { "epoch": 0.24199, "grad_norm": 1.7607108783389123, "learning_rate": 0.003, "loss": 4.0486, "step": 24199 }, { "epoch": 0.242, "grad_norm": 1.2047350174285414, "learning_rate": 0.003, "loss": 4.0665, "step": 24200 }, { "epoch": 0.24201, "grad_norm": 1.0961437779727718, "learning_rate": 0.003, "loss": 4.0625, "step": 24201 }, { "epoch": 0.24202, "grad_norm": 1.434942644690441, "learning_rate": 0.003, "loss": 4.0363, "step": 24202 }, { "epoch": 0.24203, "grad_norm": 1.295186528501165, "learning_rate": 0.003, "loss": 4.0317, "step": 24203 }, { "epoch": 0.24204, "grad_norm": 1.5122603337371539, "learning_rate": 0.003, "loss": 4.05, "step": 24204 }, { "epoch": 0.24205, "grad_norm": 1.048177867933873, "learning_rate": 0.003, "loss": 4.0238, "step": 24205 }, { "epoch": 0.24206, "grad_norm": 1.467996421274092, "learning_rate": 0.003, "loss": 4.0162, "step": 24206 }, { "epoch": 0.24207, "grad_norm": 1.1646541773412056, "learning_rate": 0.003, "loss": 4.0642, "step": 24207 }, { "epoch": 0.24208, "grad_norm": 1.414036026722916, "learning_rate": 0.003, "loss": 4.0577, "step": 24208 }, { "epoch": 0.24209, "grad_norm": 1.1193202830813809, "learning_rate": 0.003, "loss": 4.0289, "step": 24209 }, { "epoch": 0.2421, "grad_norm": 1.3794354522643395, "learning_rate": 0.003, "loss": 4.0606, "step": 24210 }, { "epoch": 0.24211, "grad_norm": 1.4281462957787163, "learning_rate": 0.003, "loss": 4.0756, "step": 24211 }, { "epoch": 0.24212, "grad_norm": 1.4723415243212516, "learning_rate": 0.003, "loss": 4.0304, "step": 24212 }, { "epoch": 0.24213, "grad_norm": 1.5812721690714122, "learning_rate": 0.003, "loss": 4.0824, "step": 24213 }, { "epoch": 0.24214, "grad_norm": 1.2187648935517055, "learning_rate": 0.003, "loss": 4.0623, "step": 24214 }, { "epoch": 0.24215, "grad_norm": 1.3346064574693712, "learning_rate": 0.003, "loss": 4.047, "step": 24215 }, { "epoch": 0.24216, "grad_norm": 1.1344116079183073, "learning_rate": 0.003, "loss": 4.0247, "step": 24216 }, { "epoch": 0.24217, "grad_norm": 1.2778453683659157, "learning_rate": 0.003, "loss": 4.047, "step": 24217 }, { "epoch": 0.24218, "grad_norm": 1.2674101944841287, "learning_rate": 0.003, "loss": 4.0397, "step": 24218 }, { "epoch": 0.24219, "grad_norm": 1.1557672568613029, "learning_rate": 0.003, "loss": 4.0279, "step": 24219 }, { "epoch": 0.2422, "grad_norm": 1.481880075511149, "learning_rate": 0.003, "loss": 4.0482, "step": 24220 }, { "epoch": 0.24221, "grad_norm": 1.371758660476148, "learning_rate": 0.003, "loss": 4.0267, "step": 24221 }, { "epoch": 0.24222, "grad_norm": 1.308356176851615, "learning_rate": 0.003, "loss": 4.0178, "step": 24222 }, { "epoch": 0.24223, "grad_norm": 1.1225815741420113, "learning_rate": 0.003, "loss": 4.0129, "step": 24223 }, { "epoch": 0.24224, "grad_norm": 1.6620733678889827, "learning_rate": 0.003, "loss": 4.0406, "step": 24224 }, { "epoch": 0.24225, "grad_norm": 1.0717480896235967, "learning_rate": 0.003, "loss": 4.0464, "step": 24225 }, { "epoch": 0.24226, "grad_norm": 1.568252971844198, "learning_rate": 0.003, "loss": 4.0459, "step": 24226 }, { "epoch": 0.24227, "grad_norm": 1.047001777031136, "learning_rate": 0.003, "loss": 4.0708, "step": 24227 }, { "epoch": 0.24228, "grad_norm": 1.578879586049578, "learning_rate": 0.003, "loss": 4.0574, "step": 24228 }, { "epoch": 0.24229, "grad_norm": 1.255557463317058, "learning_rate": 0.003, "loss": 4.032, "step": 24229 }, { "epoch": 0.2423, "grad_norm": 1.3344929981570273, "learning_rate": 0.003, "loss": 4.05, "step": 24230 }, { "epoch": 0.24231, "grad_norm": 1.2765307928136662, "learning_rate": 0.003, "loss": 4.0597, "step": 24231 }, { "epoch": 0.24232, "grad_norm": 1.1689759846324472, "learning_rate": 0.003, "loss": 4.0144, "step": 24232 }, { "epoch": 0.24233, "grad_norm": 1.277077445032698, "learning_rate": 0.003, "loss": 4.0533, "step": 24233 }, { "epoch": 0.24234, "grad_norm": 1.3343097937856678, "learning_rate": 0.003, "loss": 4.0737, "step": 24234 }, { "epoch": 0.24235, "grad_norm": 1.2624879423626758, "learning_rate": 0.003, "loss": 4.0051, "step": 24235 }, { "epoch": 0.24236, "grad_norm": 1.5093815483667568, "learning_rate": 0.003, "loss": 4.0345, "step": 24236 }, { "epoch": 0.24237, "grad_norm": 1.1616988737126757, "learning_rate": 0.003, "loss": 4.0322, "step": 24237 }, { "epoch": 0.24238, "grad_norm": 1.32435415953714, "learning_rate": 0.003, "loss": 4.0503, "step": 24238 }, { "epoch": 0.24239, "grad_norm": 1.3531919944226536, "learning_rate": 0.003, "loss": 4.0537, "step": 24239 }, { "epoch": 0.2424, "grad_norm": 1.277315980957559, "learning_rate": 0.003, "loss": 4.0368, "step": 24240 }, { "epoch": 0.24241, "grad_norm": 1.249898718096146, "learning_rate": 0.003, "loss": 4.0383, "step": 24241 }, { "epoch": 0.24242, "grad_norm": 1.3099634113763017, "learning_rate": 0.003, "loss": 4.0407, "step": 24242 }, { "epoch": 0.24243, "grad_norm": 1.3078930431906108, "learning_rate": 0.003, "loss": 4.0339, "step": 24243 }, { "epoch": 0.24244, "grad_norm": 1.2572928772215564, "learning_rate": 0.003, "loss": 4.0296, "step": 24244 }, { "epoch": 0.24245, "grad_norm": 1.3101455000662532, "learning_rate": 0.003, "loss": 4.0274, "step": 24245 }, { "epoch": 0.24246, "grad_norm": 1.1508340042190612, "learning_rate": 0.003, "loss": 4.0577, "step": 24246 }, { "epoch": 0.24247, "grad_norm": 1.5441562857769022, "learning_rate": 0.003, "loss": 4.0587, "step": 24247 }, { "epoch": 0.24248, "grad_norm": 1.2496805076173356, "learning_rate": 0.003, "loss": 4.0663, "step": 24248 }, { "epoch": 0.24249, "grad_norm": 1.4245732948720276, "learning_rate": 0.003, "loss": 4.045, "step": 24249 }, { "epoch": 0.2425, "grad_norm": 1.3938018914946457, "learning_rate": 0.003, "loss": 4.0373, "step": 24250 }, { "epoch": 0.24251, "grad_norm": 1.2968316982519055, "learning_rate": 0.003, "loss": 4.0518, "step": 24251 }, { "epoch": 0.24252, "grad_norm": 1.3120068378531047, "learning_rate": 0.003, "loss": 4.0258, "step": 24252 }, { "epoch": 0.24253, "grad_norm": 1.2013476087596637, "learning_rate": 0.003, "loss": 4.0576, "step": 24253 }, { "epoch": 0.24254, "grad_norm": 1.2935208700558307, "learning_rate": 0.003, "loss": 4.0377, "step": 24254 }, { "epoch": 0.24255, "grad_norm": 1.1377021591754488, "learning_rate": 0.003, "loss": 4.0441, "step": 24255 }, { "epoch": 0.24256, "grad_norm": 1.412766070803767, "learning_rate": 0.003, "loss": 4.0356, "step": 24256 }, { "epoch": 0.24257, "grad_norm": 1.1053486006144808, "learning_rate": 0.003, "loss": 4.0242, "step": 24257 }, { "epoch": 0.24258, "grad_norm": 1.4926667315805335, "learning_rate": 0.003, "loss": 4.0379, "step": 24258 }, { "epoch": 0.24259, "grad_norm": 1.5672772468871516, "learning_rate": 0.003, "loss": 4.0064, "step": 24259 }, { "epoch": 0.2426, "grad_norm": 1.1536041794405874, "learning_rate": 0.003, "loss": 4.0413, "step": 24260 }, { "epoch": 0.24261, "grad_norm": 1.3519942069054107, "learning_rate": 0.003, "loss": 4.048, "step": 24261 }, { "epoch": 0.24262, "grad_norm": 1.3141326041698043, "learning_rate": 0.003, "loss": 4.0546, "step": 24262 }, { "epoch": 0.24263, "grad_norm": 1.198962458986583, "learning_rate": 0.003, "loss": 4.0556, "step": 24263 }, { "epoch": 0.24264, "grad_norm": 1.0927644248715185, "learning_rate": 0.003, "loss": 4.0792, "step": 24264 }, { "epoch": 0.24265, "grad_norm": 1.3462870878086504, "learning_rate": 0.003, "loss": 4.0354, "step": 24265 }, { "epoch": 0.24266, "grad_norm": 1.3672031064169925, "learning_rate": 0.003, "loss": 4.0243, "step": 24266 }, { "epoch": 0.24267, "grad_norm": 1.291376543037462, "learning_rate": 0.003, "loss": 4.054, "step": 24267 }, { "epoch": 0.24268, "grad_norm": 1.3418901765583673, "learning_rate": 0.003, "loss": 4.0647, "step": 24268 }, { "epoch": 0.24269, "grad_norm": 1.3574717174686124, "learning_rate": 0.003, "loss": 4.0431, "step": 24269 }, { "epoch": 0.2427, "grad_norm": 1.3571988322609831, "learning_rate": 0.003, "loss": 4.0463, "step": 24270 }, { "epoch": 0.24271, "grad_norm": 1.2935055122759802, "learning_rate": 0.003, "loss": 4.0232, "step": 24271 }, { "epoch": 0.24272, "grad_norm": 1.3327856806884089, "learning_rate": 0.003, "loss": 4.0295, "step": 24272 }, { "epoch": 0.24273, "grad_norm": 1.198564049426773, "learning_rate": 0.003, "loss": 4.0294, "step": 24273 }, { "epoch": 0.24274, "grad_norm": 1.2504595965816272, "learning_rate": 0.003, "loss": 4.0535, "step": 24274 }, { "epoch": 0.24275, "grad_norm": 1.4428564297078146, "learning_rate": 0.003, "loss": 4.0271, "step": 24275 }, { "epoch": 0.24276, "grad_norm": 1.1674450737238984, "learning_rate": 0.003, "loss": 4.0295, "step": 24276 }, { "epoch": 0.24277, "grad_norm": 1.3185220628168617, "learning_rate": 0.003, "loss": 4.0482, "step": 24277 }, { "epoch": 0.24278, "grad_norm": 1.2340535805395114, "learning_rate": 0.003, "loss": 4.0341, "step": 24278 }, { "epoch": 0.24279, "grad_norm": 1.507975611775976, "learning_rate": 0.003, "loss": 4.0462, "step": 24279 }, { "epoch": 0.2428, "grad_norm": 1.2131537403595838, "learning_rate": 0.003, "loss": 4.0496, "step": 24280 }, { "epoch": 0.24281, "grad_norm": 1.2613812387359136, "learning_rate": 0.003, "loss": 4.0614, "step": 24281 }, { "epoch": 0.24282, "grad_norm": 1.264160759089061, "learning_rate": 0.003, "loss": 4.0209, "step": 24282 }, { "epoch": 0.24283, "grad_norm": 1.3453013059909218, "learning_rate": 0.003, "loss": 4.0359, "step": 24283 }, { "epoch": 0.24284, "grad_norm": 1.5025609629209615, "learning_rate": 0.003, "loss": 4.0571, "step": 24284 }, { "epoch": 0.24285, "grad_norm": 1.231970294944696, "learning_rate": 0.003, "loss": 4.0159, "step": 24285 }, { "epoch": 0.24286, "grad_norm": 1.4273096806634538, "learning_rate": 0.003, "loss": 4.0496, "step": 24286 }, { "epoch": 0.24287, "grad_norm": 1.1738704895508498, "learning_rate": 0.003, "loss": 4.0354, "step": 24287 }, { "epoch": 0.24288, "grad_norm": 1.3807691622228144, "learning_rate": 0.003, "loss": 4.0178, "step": 24288 }, { "epoch": 0.24289, "grad_norm": 1.446508532115049, "learning_rate": 0.003, "loss": 4.0403, "step": 24289 }, { "epoch": 0.2429, "grad_norm": 1.5243046036714811, "learning_rate": 0.003, "loss": 4.0504, "step": 24290 }, { "epoch": 0.24291, "grad_norm": 1.279020751005401, "learning_rate": 0.003, "loss": 4.0251, "step": 24291 }, { "epoch": 0.24292, "grad_norm": 1.344298576833801, "learning_rate": 0.003, "loss": 4.0502, "step": 24292 }, { "epoch": 0.24293, "grad_norm": 1.34572143624435, "learning_rate": 0.003, "loss": 3.9985, "step": 24293 }, { "epoch": 0.24294, "grad_norm": 1.2387541801867668, "learning_rate": 0.003, "loss": 4.03, "step": 24294 }, { "epoch": 0.24295, "grad_norm": 1.2741373636751783, "learning_rate": 0.003, "loss": 4.0398, "step": 24295 }, { "epoch": 0.24296, "grad_norm": 1.2745234807409727, "learning_rate": 0.003, "loss": 4.0249, "step": 24296 }, { "epoch": 0.24297, "grad_norm": 1.0891295685417062, "learning_rate": 0.003, "loss": 4.0311, "step": 24297 }, { "epoch": 0.24298, "grad_norm": 1.1796609971169227, "learning_rate": 0.003, "loss": 4.0324, "step": 24298 }, { "epoch": 0.24299, "grad_norm": 1.379088872647827, "learning_rate": 0.003, "loss": 4.027, "step": 24299 }, { "epoch": 0.243, "grad_norm": 1.3023287424810517, "learning_rate": 0.003, "loss": 4.0365, "step": 24300 }, { "epoch": 0.24301, "grad_norm": 1.3542299752803761, "learning_rate": 0.003, "loss": 4.01, "step": 24301 }, { "epoch": 0.24302, "grad_norm": 1.097581145180247, "learning_rate": 0.003, "loss": 4.0304, "step": 24302 }, { "epoch": 0.24303, "grad_norm": 1.4773640604740808, "learning_rate": 0.003, "loss": 4.0709, "step": 24303 }, { "epoch": 0.24304, "grad_norm": 1.0991500770619063, "learning_rate": 0.003, "loss": 4.0414, "step": 24304 }, { "epoch": 0.24305, "grad_norm": 1.608905695567478, "learning_rate": 0.003, "loss": 4.0668, "step": 24305 }, { "epoch": 0.24306, "grad_norm": 1.2088648847581218, "learning_rate": 0.003, "loss": 4.0356, "step": 24306 }, { "epoch": 0.24307, "grad_norm": 1.5473059889716545, "learning_rate": 0.003, "loss": 4.0878, "step": 24307 }, { "epoch": 0.24308, "grad_norm": 1.1063845682215177, "learning_rate": 0.003, "loss": 4.0289, "step": 24308 }, { "epoch": 0.24309, "grad_norm": 1.7032377624789472, "learning_rate": 0.003, "loss": 4.064, "step": 24309 }, { "epoch": 0.2431, "grad_norm": 1.2348758434532585, "learning_rate": 0.003, "loss": 4.0497, "step": 24310 }, { "epoch": 0.24311, "grad_norm": 1.675901933527108, "learning_rate": 0.003, "loss": 4.0464, "step": 24311 }, { "epoch": 0.24312, "grad_norm": 1.28701685329977, "learning_rate": 0.003, "loss": 4.0274, "step": 24312 }, { "epoch": 0.24313, "grad_norm": 1.2493169303194676, "learning_rate": 0.003, "loss": 4.0626, "step": 24313 }, { "epoch": 0.24314, "grad_norm": 1.3730708220398244, "learning_rate": 0.003, "loss": 4.0299, "step": 24314 }, { "epoch": 0.24315, "grad_norm": 1.3990743664317271, "learning_rate": 0.003, "loss": 4.055, "step": 24315 }, { "epoch": 0.24316, "grad_norm": 1.4150105767793484, "learning_rate": 0.003, "loss": 4.0455, "step": 24316 }, { "epoch": 0.24317, "grad_norm": 1.3229023075818864, "learning_rate": 0.003, "loss": 4.0858, "step": 24317 }, { "epoch": 0.24318, "grad_norm": 1.316400854388295, "learning_rate": 0.003, "loss": 4.0424, "step": 24318 }, { "epoch": 0.24319, "grad_norm": 1.2861302703731605, "learning_rate": 0.003, "loss": 4.0386, "step": 24319 }, { "epoch": 0.2432, "grad_norm": 1.2302203390549422, "learning_rate": 0.003, "loss": 4.0524, "step": 24320 }, { "epoch": 0.24321, "grad_norm": 1.1870239088361496, "learning_rate": 0.003, "loss": 4.0252, "step": 24321 }, { "epoch": 0.24322, "grad_norm": 1.1098744348245855, "learning_rate": 0.003, "loss": 4.0632, "step": 24322 }, { "epoch": 0.24323, "grad_norm": 1.4013358626080086, "learning_rate": 0.003, "loss": 4.0388, "step": 24323 }, { "epoch": 0.24324, "grad_norm": 1.2774390008585295, "learning_rate": 0.003, "loss": 4.045, "step": 24324 }, { "epoch": 0.24325, "grad_norm": 1.4232646560783073, "learning_rate": 0.003, "loss": 4.0348, "step": 24325 }, { "epoch": 0.24326, "grad_norm": 1.258223538648565, "learning_rate": 0.003, "loss": 4.0423, "step": 24326 }, { "epoch": 0.24327, "grad_norm": 1.2570427920975569, "learning_rate": 0.003, "loss": 4.0456, "step": 24327 }, { "epoch": 0.24328, "grad_norm": 1.4334107535479221, "learning_rate": 0.003, "loss": 4.0574, "step": 24328 }, { "epoch": 0.24329, "grad_norm": 1.2406233655124415, "learning_rate": 0.003, "loss": 4.0475, "step": 24329 }, { "epoch": 0.2433, "grad_norm": 1.4310929648549848, "learning_rate": 0.003, "loss": 4.036, "step": 24330 }, { "epoch": 0.24331, "grad_norm": 1.103687581279719, "learning_rate": 0.003, "loss": 4.014, "step": 24331 }, { "epoch": 0.24332, "grad_norm": 1.5151387176506648, "learning_rate": 0.003, "loss": 4.0726, "step": 24332 }, { "epoch": 0.24333, "grad_norm": 1.138491482282826, "learning_rate": 0.003, "loss": 4.0193, "step": 24333 }, { "epoch": 0.24334, "grad_norm": 1.3784484008845785, "learning_rate": 0.003, "loss": 4.0403, "step": 24334 }, { "epoch": 0.24335, "grad_norm": 1.399359208029288, "learning_rate": 0.003, "loss": 4.0468, "step": 24335 }, { "epoch": 0.24336, "grad_norm": 1.303246903348923, "learning_rate": 0.003, "loss": 4.0231, "step": 24336 }, { "epoch": 0.24337, "grad_norm": 1.3594896867625685, "learning_rate": 0.003, "loss": 4.0358, "step": 24337 }, { "epoch": 0.24338, "grad_norm": 1.3014414192233448, "learning_rate": 0.003, "loss": 4.0677, "step": 24338 }, { "epoch": 0.24339, "grad_norm": 1.1903845685813315, "learning_rate": 0.003, "loss": 4.0496, "step": 24339 }, { "epoch": 0.2434, "grad_norm": 1.4573698097913577, "learning_rate": 0.003, "loss": 4.0299, "step": 24340 }, { "epoch": 0.24341, "grad_norm": 1.1391009869645503, "learning_rate": 0.003, "loss": 4.0145, "step": 24341 }, { "epoch": 0.24342, "grad_norm": 1.5077851029104923, "learning_rate": 0.003, "loss": 4.0294, "step": 24342 }, { "epoch": 0.24343, "grad_norm": 1.4460109922666824, "learning_rate": 0.003, "loss": 4.0415, "step": 24343 }, { "epoch": 0.24344, "grad_norm": 1.2328575413037233, "learning_rate": 0.003, "loss": 4.0558, "step": 24344 }, { "epoch": 0.24345, "grad_norm": 1.4941659042148472, "learning_rate": 0.003, "loss": 4.0464, "step": 24345 }, { "epoch": 0.24346, "grad_norm": 1.266366569169842, "learning_rate": 0.003, "loss": 4.0528, "step": 24346 }, { "epoch": 0.24347, "grad_norm": 1.3315629415854084, "learning_rate": 0.003, "loss": 4.0449, "step": 24347 }, { "epoch": 0.24348, "grad_norm": 1.1091768802830682, "learning_rate": 0.003, "loss": 4.0453, "step": 24348 }, { "epoch": 0.24349, "grad_norm": 1.29103590953526, "learning_rate": 0.003, "loss": 4.0541, "step": 24349 }, { "epoch": 0.2435, "grad_norm": 1.331637533256418, "learning_rate": 0.003, "loss": 4.0534, "step": 24350 }, { "epoch": 0.24351, "grad_norm": 1.2626612531390127, "learning_rate": 0.003, "loss": 4.0304, "step": 24351 }, { "epoch": 0.24352, "grad_norm": 1.5421735794153124, "learning_rate": 0.003, "loss": 4.0649, "step": 24352 }, { "epoch": 0.24353, "grad_norm": 1.1070581822841798, "learning_rate": 0.003, "loss": 4.0421, "step": 24353 }, { "epoch": 0.24354, "grad_norm": 1.6211992093023875, "learning_rate": 0.003, "loss": 4.0576, "step": 24354 }, { "epoch": 0.24355, "grad_norm": 1.2349649328258079, "learning_rate": 0.003, "loss": 4.0519, "step": 24355 }, { "epoch": 0.24356, "grad_norm": 1.383760494906793, "learning_rate": 0.003, "loss": 4.0691, "step": 24356 }, { "epoch": 0.24357, "grad_norm": 1.2794547079780425, "learning_rate": 0.003, "loss": 4.0652, "step": 24357 }, { "epoch": 0.24358, "grad_norm": 1.3882579218338404, "learning_rate": 0.003, "loss": 4.035, "step": 24358 }, { "epoch": 0.24359, "grad_norm": 1.363947204059705, "learning_rate": 0.003, "loss": 4.0384, "step": 24359 }, { "epoch": 0.2436, "grad_norm": 1.3386789358283258, "learning_rate": 0.003, "loss": 4.0404, "step": 24360 }, { "epoch": 0.24361, "grad_norm": 1.3178675101574866, "learning_rate": 0.003, "loss": 4.0558, "step": 24361 }, { "epoch": 0.24362, "grad_norm": 1.3907757394468752, "learning_rate": 0.003, "loss": 4.0321, "step": 24362 }, { "epoch": 0.24363, "grad_norm": 1.2141064776142552, "learning_rate": 0.003, "loss": 4.0423, "step": 24363 }, { "epoch": 0.24364, "grad_norm": 1.2105623553777896, "learning_rate": 0.003, "loss": 4.0337, "step": 24364 }, { "epoch": 0.24365, "grad_norm": 1.1482711614982255, "learning_rate": 0.003, "loss": 4.0301, "step": 24365 }, { "epoch": 0.24366, "grad_norm": 1.5031399374710517, "learning_rate": 0.003, "loss": 4.0555, "step": 24366 }, { "epoch": 0.24367, "grad_norm": 1.03573165729022, "learning_rate": 0.003, "loss": 4.0419, "step": 24367 }, { "epoch": 0.24368, "grad_norm": 1.469894046856681, "learning_rate": 0.003, "loss": 4.0741, "step": 24368 }, { "epoch": 0.24369, "grad_norm": 1.1427960540261197, "learning_rate": 0.003, "loss": 4.0559, "step": 24369 }, { "epoch": 0.2437, "grad_norm": 1.3353205949902167, "learning_rate": 0.003, "loss": 4.022, "step": 24370 }, { "epoch": 0.24371, "grad_norm": 1.30374156245724, "learning_rate": 0.003, "loss": 4.0673, "step": 24371 }, { "epoch": 0.24372, "grad_norm": 1.3048593353312705, "learning_rate": 0.003, "loss": 4.0578, "step": 24372 }, { "epoch": 0.24373, "grad_norm": 1.277169693813719, "learning_rate": 0.003, "loss": 4.0604, "step": 24373 }, { "epoch": 0.24374, "grad_norm": 1.1710966085423475, "learning_rate": 0.003, "loss": 4.0574, "step": 24374 }, { "epoch": 0.24375, "grad_norm": 1.4131985407574053, "learning_rate": 0.003, "loss": 4.0399, "step": 24375 }, { "epoch": 0.24376, "grad_norm": 1.5634989100931784, "learning_rate": 0.003, "loss": 4.0278, "step": 24376 }, { "epoch": 0.24377, "grad_norm": 1.6439959070235346, "learning_rate": 0.003, "loss": 4.031, "step": 24377 }, { "epoch": 0.24378, "grad_norm": 1.2612910964210828, "learning_rate": 0.003, "loss": 4.0163, "step": 24378 }, { "epoch": 0.24379, "grad_norm": 1.2274938496146883, "learning_rate": 0.003, "loss": 4.0397, "step": 24379 }, { "epoch": 0.2438, "grad_norm": 1.5446903779722578, "learning_rate": 0.003, "loss": 4.0473, "step": 24380 }, { "epoch": 0.24381, "grad_norm": 1.2130595182196149, "learning_rate": 0.003, "loss": 4.0524, "step": 24381 }, { "epoch": 0.24382, "grad_norm": 1.501681768875987, "learning_rate": 0.003, "loss": 4.0522, "step": 24382 }, { "epoch": 0.24383, "grad_norm": 1.1199131797469297, "learning_rate": 0.003, "loss": 4.0277, "step": 24383 }, { "epoch": 0.24384, "grad_norm": 1.4750102998432133, "learning_rate": 0.003, "loss": 4.0551, "step": 24384 }, { "epoch": 0.24385, "grad_norm": 1.1469196242822652, "learning_rate": 0.003, "loss": 4.0285, "step": 24385 }, { "epoch": 0.24386, "grad_norm": 1.3636585490293693, "learning_rate": 0.003, "loss": 4.0581, "step": 24386 }, { "epoch": 0.24387, "grad_norm": 1.2484401742059057, "learning_rate": 0.003, "loss": 4.0499, "step": 24387 }, { "epoch": 0.24388, "grad_norm": 1.3543994028077897, "learning_rate": 0.003, "loss": 4.0293, "step": 24388 }, { "epoch": 0.24389, "grad_norm": 1.3625939017751754, "learning_rate": 0.003, "loss": 4.0601, "step": 24389 }, { "epoch": 0.2439, "grad_norm": 1.286243334650969, "learning_rate": 0.003, "loss": 4.0216, "step": 24390 }, { "epoch": 0.24391, "grad_norm": 1.4120548491492744, "learning_rate": 0.003, "loss": 4.0298, "step": 24391 }, { "epoch": 0.24392, "grad_norm": 1.2679482414437517, "learning_rate": 0.003, "loss": 4.0643, "step": 24392 }, { "epoch": 0.24393, "grad_norm": 1.5188804793319215, "learning_rate": 0.003, "loss": 4.0288, "step": 24393 }, { "epoch": 0.24394, "grad_norm": 1.1513339940591412, "learning_rate": 0.003, "loss": 4.0577, "step": 24394 }, { "epoch": 0.24395, "grad_norm": 1.4249037706512802, "learning_rate": 0.003, "loss": 4.0064, "step": 24395 }, { "epoch": 0.24396, "grad_norm": 1.2539243573385133, "learning_rate": 0.003, "loss": 4.0504, "step": 24396 }, { "epoch": 0.24397, "grad_norm": 1.4088073980669185, "learning_rate": 0.003, "loss": 4.0673, "step": 24397 }, { "epoch": 0.24398, "grad_norm": 1.1193242361233235, "learning_rate": 0.003, "loss": 4.0194, "step": 24398 }, { "epoch": 0.24399, "grad_norm": 1.5437511059858533, "learning_rate": 0.003, "loss": 4.0548, "step": 24399 }, { "epoch": 0.244, "grad_norm": 1.167029003402302, "learning_rate": 0.003, "loss": 4.0439, "step": 24400 }, { "epoch": 0.24401, "grad_norm": 1.4423006843544683, "learning_rate": 0.003, "loss": 4.0136, "step": 24401 }, { "epoch": 0.24402, "grad_norm": 1.19660625567266, "learning_rate": 0.003, "loss": 4.0443, "step": 24402 }, { "epoch": 0.24403, "grad_norm": 1.6566764224708204, "learning_rate": 0.003, "loss": 4.0685, "step": 24403 }, { "epoch": 0.24404, "grad_norm": 1.217797948898943, "learning_rate": 0.003, "loss": 4.0409, "step": 24404 }, { "epoch": 0.24405, "grad_norm": 1.2816964607024994, "learning_rate": 0.003, "loss": 4.0358, "step": 24405 }, { "epoch": 0.24406, "grad_norm": 1.160572080266186, "learning_rate": 0.003, "loss": 4.0488, "step": 24406 }, { "epoch": 0.24407, "grad_norm": 1.3570768974084684, "learning_rate": 0.003, "loss": 4.048, "step": 24407 }, { "epoch": 0.24408, "grad_norm": 1.3505746210838767, "learning_rate": 0.003, "loss": 4.047, "step": 24408 }, { "epoch": 0.24409, "grad_norm": 1.278392742908736, "learning_rate": 0.003, "loss": 4.032, "step": 24409 }, { "epoch": 0.2441, "grad_norm": 1.5412685559591057, "learning_rate": 0.003, "loss": 4.0656, "step": 24410 }, { "epoch": 0.24411, "grad_norm": 1.2919760468073036, "learning_rate": 0.003, "loss": 4.0253, "step": 24411 }, { "epoch": 0.24412, "grad_norm": 1.2053676679336764, "learning_rate": 0.003, "loss": 4.0306, "step": 24412 }, { "epoch": 0.24413, "grad_norm": 1.4097765389792798, "learning_rate": 0.003, "loss": 4.068, "step": 24413 }, { "epoch": 0.24414, "grad_norm": 1.1520224665585594, "learning_rate": 0.003, "loss": 4.0641, "step": 24414 }, { "epoch": 0.24415, "grad_norm": 1.3172573487463803, "learning_rate": 0.003, "loss": 4.0103, "step": 24415 }, { "epoch": 0.24416, "grad_norm": 1.181501462944302, "learning_rate": 0.003, "loss": 4.0533, "step": 24416 }, { "epoch": 0.24417, "grad_norm": 1.2850826135589895, "learning_rate": 0.003, "loss": 4.0304, "step": 24417 }, { "epoch": 0.24418, "grad_norm": 1.1813625743702172, "learning_rate": 0.003, "loss": 4.0257, "step": 24418 }, { "epoch": 0.24419, "grad_norm": 1.498848069287447, "learning_rate": 0.003, "loss": 4.0466, "step": 24419 }, { "epoch": 0.2442, "grad_norm": 1.3365506223278083, "learning_rate": 0.003, "loss": 4.0464, "step": 24420 }, { "epoch": 0.24421, "grad_norm": 1.507610616619639, "learning_rate": 0.003, "loss": 4.043, "step": 24421 }, { "epoch": 0.24422, "grad_norm": 1.375523461219169, "learning_rate": 0.003, "loss": 4.0607, "step": 24422 }, { "epoch": 0.24423, "grad_norm": 1.4578343728913137, "learning_rate": 0.003, "loss": 4.0343, "step": 24423 }, { "epoch": 0.24424, "grad_norm": 1.3188601759502132, "learning_rate": 0.003, "loss": 4.0615, "step": 24424 }, { "epoch": 0.24425, "grad_norm": 1.208672839410013, "learning_rate": 0.003, "loss": 4.0504, "step": 24425 }, { "epoch": 0.24426, "grad_norm": 1.2003951521769713, "learning_rate": 0.003, "loss": 4.0523, "step": 24426 }, { "epoch": 0.24427, "grad_norm": 1.421264008091024, "learning_rate": 0.003, "loss": 4.0547, "step": 24427 }, { "epoch": 0.24428, "grad_norm": 1.256893595211034, "learning_rate": 0.003, "loss": 4.0453, "step": 24428 }, { "epoch": 0.24429, "grad_norm": 1.3900440844068316, "learning_rate": 0.003, "loss": 4.0436, "step": 24429 }, { "epoch": 0.2443, "grad_norm": 1.2781165917986486, "learning_rate": 0.003, "loss": 4.0609, "step": 24430 }, { "epoch": 0.24431, "grad_norm": 1.3807708537073173, "learning_rate": 0.003, "loss": 4.0502, "step": 24431 }, { "epoch": 0.24432, "grad_norm": 1.257664141834115, "learning_rate": 0.003, "loss": 4.0431, "step": 24432 }, { "epoch": 0.24433, "grad_norm": 1.4134994521861892, "learning_rate": 0.003, "loss": 4.0619, "step": 24433 }, { "epoch": 0.24434, "grad_norm": 1.2339324484004306, "learning_rate": 0.003, "loss": 4.0371, "step": 24434 }, { "epoch": 0.24435, "grad_norm": 1.1599182271248694, "learning_rate": 0.003, "loss": 4.0289, "step": 24435 }, { "epoch": 0.24436, "grad_norm": 1.4577503314799183, "learning_rate": 0.003, "loss": 4.0453, "step": 24436 }, { "epoch": 0.24437, "grad_norm": 1.250495587068921, "learning_rate": 0.003, "loss": 4.019, "step": 24437 }, { "epoch": 0.24438, "grad_norm": 1.4686940998364177, "learning_rate": 0.003, "loss": 4.0264, "step": 24438 }, { "epoch": 0.24439, "grad_norm": 1.131835032873698, "learning_rate": 0.003, "loss": 4.0372, "step": 24439 }, { "epoch": 0.2444, "grad_norm": 1.6518414140033024, "learning_rate": 0.003, "loss": 4.0484, "step": 24440 }, { "epoch": 0.24441, "grad_norm": 1.0403182420996597, "learning_rate": 0.003, "loss": 4.0673, "step": 24441 }, { "epoch": 0.24442, "grad_norm": 1.5430143646665895, "learning_rate": 0.003, "loss": 4.0402, "step": 24442 }, { "epoch": 0.24443, "grad_norm": 1.1011653742817602, "learning_rate": 0.003, "loss": 4.0484, "step": 24443 }, { "epoch": 0.24444, "grad_norm": 1.4417584027895425, "learning_rate": 0.003, "loss": 4.0211, "step": 24444 }, { "epoch": 0.24445, "grad_norm": 1.2341631869518908, "learning_rate": 0.003, "loss": 4.0767, "step": 24445 }, { "epoch": 0.24446, "grad_norm": 1.3842106460037205, "learning_rate": 0.003, "loss": 4.0602, "step": 24446 }, { "epoch": 0.24447, "grad_norm": 1.4287571022327497, "learning_rate": 0.003, "loss": 4.0384, "step": 24447 }, { "epoch": 0.24448, "grad_norm": 1.0384178499521848, "learning_rate": 0.003, "loss": 4.0274, "step": 24448 }, { "epoch": 0.24449, "grad_norm": 1.3357146506276172, "learning_rate": 0.003, "loss": 4.0369, "step": 24449 }, { "epoch": 0.2445, "grad_norm": 1.2822850399810086, "learning_rate": 0.003, "loss": 4.02, "step": 24450 }, { "epoch": 0.24451, "grad_norm": 1.1737612358847367, "learning_rate": 0.003, "loss": 4.0512, "step": 24451 }, { "epoch": 0.24452, "grad_norm": 1.4190079707525498, "learning_rate": 0.003, "loss": 4.0529, "step": 24452 }, { "epoch": 0.24453, "grad_norm": 1.3956098662686127, "learning_rate": 0.003, "loss": 4.0689, "step": 24453 }, { "epoch": 0.24454, "grad_norm": 1.6616896674216715, "learning_rate": 0.003, "loss": 4.0488, "step": 24454 }, { "epoch": 0.24455, "grad_norm": 1.4288305035999362, "learning_rate": 0.003, "loss": 4.0398, "step": 24455 }, { "epoch": 0.24456, "grad_norm": 1.208736748260487, "learning_rate": 0.003, "loss": 4.0457, "step": 24456 }, { "epoch": 0.24457, "grad_norm": 1.4906421917853605, "learning_rate": 0.003, "loss": 4.0417, "step": 24457 }, { "epoch": 0.24458, "grad_norm": 1.219671554369932, "learning_rate": 0.003, "loss": 4.064, "step": 24458 }, { "epoch": 0.24459, "grad_norm": 1.311815167755257, "learning_rate": 0.003, "loss": 4.0284, "step": 24459 }, { "epoch": 0.2446, "grad_norm": 1.1017975530896578, "learning_rate": 0.003, "loss": 4.0294, "step": 24460 }, { "epoch": 0.24461, "grad_norm": 1.3433925817368302, "learning_rate": 0.003, "loss": 4.0468, "step": 24461 }, { "epoch": 0.24462, "grad_norm": 1.2182044458211283, "learning_rate": 0.003, "loss": 4.0389, "step": 24462 }, { "epoch": 0.24463, "grad_norm": 1.4343536697065558, "learning_rate": 0.003, "loss": 4.0456, "step": 24463 }, { "epoch": 0.24464, "grad_norm": 1.326231448343906, "learning_rate": 0.003, "loss": 4.061, "step": 24464 }, { "epoch": 0.24465, "grad_norm": 1.274985724216275, "learning_rate": 0.003, "loss": 4.0369, "step": 24465 }, { "epoch": 0.24466, "grad_norm": 1.5335393823798642, "learning_rate": 0.003, "loss": 4.066, "step": 24466 }, { "epoch": 0.24467, "grad_norm": 1.161642014877953, "learning_rate": 0.003, "loss": 4.0585, "step": 24467 }, { "epoch": 0.24468, "grad_norm": 1.613223118669463, "learning_rate": 0.003, "loss": 4.0419, "step": 24468 }, { "epoch": 0.24469, "grad_norm": 1.1301038545968003, "learning_rate": 0.003, "loss": 4.0867, "step": 24469 }, { "epoch": 0.2447, "grad_norm": 1.3914040131798775, "learning_rate": 0.003, "loss": 4.0239, "step": 24470 }, { "epoch": 0.24471, "grad_norm": 1.2760577866377025, "learning_rate": 0.003, "loss": 4.0301, "step": 24471 }, { "epoch": 0.24472, "grad_norm": 1.2836135686488033, "learning_rate": 0.003, "loss": 4.032, "step": 24472 }, { "epoch": 0.24473, "grad_norm": 1.1251757470155173, "learning_rate": 0.003, "loss": 4.0518, "step": 24473 }, { "epoch": 0.24474, "grad_norm": 1.3949337531023083, "learning_rate": 0.003, "loss": 4.0358, "step": 24474 }, { "epoch": 0.24475, "grad_norm": 1.1358461155418933, "learning_rate": 0.003, "loss": 4.043, "step": 24475 }, { "epoch": 0.24476, "grad_norm": 1.3852183903795379, "learning_rate": 0.003, "loss": 4.0353, "step": 24476 }, { "epoch": 0.24477, "grad_norm": 1.0656679158903217, "learning_rate": 0.003, "loss": 4.0231, "step": 24477 }, { "epoch": 0.24478, "grad_norm": 1.3576232048478207, "learning_rate": 0.003, "loss": 4.0497, "step": 24478 }, { "epoch": 0.24479, "grad_norm": 1.196936959037576, "learning_rate": 0.003, "loss": 4.0435, "step": 24479 }, { "epoch": 0.2448, "grad_norm": 1.2764074590267935, "learning_rate": 0.003, "loss": 4.0191, "step": 24480 }, { "epoch": 0.24481, "grad_norm": 1.5425521827525461, "learning_rate": 0.003, "loss": 4.0294, "step": 24481 }, { "epoch": 0.24482, "grad_norm": 1.4028500733814346, "learning_rate": 0.003, "loss": 4.0461, "step": 24482 }, { "epoch": 0.24483, "grad_norm": 1.9279301181966597, "learning_rate": 0.003, "loss": 4.0184, "step": 24483 }, { "epoch": 0.24484, "grad_norm": 1.0313661319034655, "learning_rate": 0.003, "loss": 4.0632, "step": 24484 }, { "epoch": 0.24485, "grad_norm": 1.2940843844400831, "learning_rate": 0.003, "loss": 4.0363, "step": 24485 }, { "epoch": 0.24486, "grad_norm": 1.2195266355531074, "learning_rate": 0.003, "loss": 4.0407, "step": 24486 }, { "epoch": 0.24487, "grad_norm": 1.4029352470606242, "learning_rate": 0.003, "loss": 4.0417, "step": 24487 }, { "epoch": 0.24488, "grad_norm": 1.404083273868884, "learning_rate": 0.003, "loss": 4.0809, "step": 24488 }, { "epoch": 0.24489, "grad_norm": 1.2174610046897985, "learning_rate": 0.003, "loss": 4.0665, "step": 24489 }, { "epoch": 0.2449, "grad_norm": 1.3279445169071793, "learning_rate": 0.003, "loss": 4.061, "step": 24490 }, { "epoch": 0.24491, "grad_norm": 1.239670336608579, "learning_rate": 0.003, "loss": 4.0593, "step": 24491 }, { "epoch": 0.24492, "grad_norm": 1.3742538228163146, "learning_rate": 0.003, "loss": 4.0449, "step": 24492 }, { "epoch": 0.24493, "grad_norm": 1.3652170532685641, "learning_rate": 0.003, "loss": 4.0164, "step": 24493 }, { "epoch": 0.24494, "grad_norm": 1.336665678580845, "learning_rate": 0.003, "loss": 4.0384, "step": 24494 }, { "epoch": 0.24495, "grad_norm": 1.4015942063806432, "learning_rate": 0.003, "loss": 4.0548, "step": 24495 }, { "epoch": 0.24496, "grad_norm": 1.0381947803290668, "learning_rate": 0.003, "loss": 4.0452, "step": 24496 }, { "epoch": 0.24497, "grad_norm": 1.6132895916769672, "learning_rate": 0.003, "loss": 4.0421, "step": 24497 }, { "epoch": 0.24498, "grad_norm": 1.1547106062903594, "learning_rate": 0.003, "loss": 4.0551, "step": 24498 }, { "epoch": 0.24499, "grad_norm": 1.5681254668710123, "learning_rate": 0.003, "loss": 4.0249, "step": 24499 }, { "epoch": 0.245, "grad_norm": 1.20962983361933, "learning_rate": 0.003, "loss": 4.0388, "step": 24500 }, { "epoch": 0.24501, "grad_norm": 1.2669359949920802, "learning_rate": 0.003, "loss": 4.0356, "step": 24501 }, { "epoch": 0.24502, "grad_norm": 1.247588521331606, "learning_rate": 0.003, "loss": 4.0763, "step": 24502 }, { "epoch": 0.24503, "grad_norm": 1.3477524654614856, "learning_rate": 0.003, "loss": 4.0304, "step": 24503 }, { "epoch": 0.24504, "grad_norm": 1.3385828413177792, "learning_rate": 0.003, "loss": 4.0369, "step": 24504 }, { "epoch": 0.24505, "grad_norm": 1.2603885391578664, "learning_rate": 0.003, "loss": 4.0248, "step": 24505 }, { "epoch": 0.24506, "grad_norm": 1.3773839120861833, "learning_rate": 0.003, "loss": 4.0528, "step": 24506 }, { "epoch": 0.24507, "grad_norm": 1.6678716711570647, "learning_rate": 0.003, "loss": 4.0534, "step": 24507 }, { "epoch": 0.24508, "grad_norm": 1.243321260104112, "learning_rate": 0.003, "loss": 4.0199, "step": 24508 }, { "epoch": 0.24509, "grad_norm": 1.4479103879536566, "learning_rate": 0.003, "loss": 4.0369, "step": 24509 }, { "epoch": 0.2451, "grad_norm": 1.2348195679519878, "learning_rate": 0.003, "loss": 4.0332, "step": 24510 }, { "epoch": 0.24511, "grad_norm": 1.2659594312271438, "learning_rate": 0.003, "loss": 4.0423, "step": 24511 }, { "epoch": 0.24512, "grad_norm": 1.1857350221228964, "learning_rate": 0.003, "loss": 4.0433, "step": 24512 }, { "epoch": 0.24513, "grad_norm": 1.3947144517267773, "learning_rate": 0.003, "loss": 4.0372, "step": 24513 }, { "epoch": 0.24514, "grad_norm": 1.1174055011501585, "learning_rate": 0.003, "loss": 4.0742, "step": 24514 }, { "epoch": 0.24515, "grad_norm": 1.294287122790688, "learning_rate": 0.003, "loss": 4.0442, "step": 24515 }, { "epoch": 0.24516, "grad_norm": 1.055639455236206, "learning_rate": 0.003, "loss": 4.0292, "step": 24516 }, { "epoch": 0.24517, "grad_norm": 1.4594500753428272, "learning_rate": 0.003, "loss": 4.039, "step": 24517 }, { "epoch": 0.24518, "grad_norm": 1.1671583501748672, "learning_rate": 0.003, "loss": 4.0414, "step": 24518 }, { "epoch": 0.24519, "grad_norm": 1.796210645180149, "learning_rate": 0.003, "loss": 4.0491, "step": 24519 }, { "epoch": 0.2452, "grad_norm": 1.1656828766903715, "learning_rate": 0.003, "loss": 4.0557, "step": 24520 }, { "epoch": 0.24521, "grad_norm": 1.3545401317287933, "learning_rate": 0.003, "loss": 4.05, "step": 24521 }, { "epoch": 0.24522, "grad_norm": 1.0884210364788287, "learning_rate": 0.003, "loss": 4.0698, "step": 24522 }, { "epoch": 0.24523, "grad_norm": 1.6566413709952723, "learning_rate": 0.003, "loss": 4.0424, "step": 24523 }, { "epoch": 0.24524, "grad_norm": 1.237552191869169, "learning_rate": 0.003, "loss": 4.065, "step": 24524 }, { "epoch": 0.24525, "grad_norm": 1.4451472074967542, "learning_rate": 0.003, "loss": 4.0566, "step": 24525 }, { "epoch": 0.24526, "grad_norm": 1.0565917543706052, "learning_rate": 0.003, "loss": 4.0381, "step": 24526 }, { "epoch": 0.24527, "grad_norm": 1.6121808891605443, "learning_rate": 0.003, "loss": 4.0833, "step": 24527 }, { "epoch": 0.24528, "grad_norm": 1.2307575216954734, "learning_rate": 0.003, "loss": 4.0561, "step": 24528 }, { "epoch": 0.24529, "grad_norm": 1.4754564653970716, "learning_rate": 0.003, "loss": 4.0611, "step": 24529 }, { "epoch": 0.2453, "grad_norm": 1.5447846233884956, "learning_rate": 0.003, "loss": 4.0489, "step": 24530 }, { "epoch": 0.24531, "grad_norm": 1.4122065983205467, "learning_rate": 0.003, "loss": 4.0183, "step": 24531 }, { "epoch": 0.24532, "grad_norm": 1.1605230042780361, "learning_rate": 0.003, "loss": 4.0531, "step": 24532 }, { "epoch": 0.24533, "grad_norm": 1.3733446178738185, "learning_rate": 0.003, "loss": 4.0308, "step": 24533 }, { "epoch": 0.24534, "grad_norm": 1.393069005317022, "learning_rate": 0.003, "loss": 4.0625, "step": 24534 }, { "epoch": 0.24535, "grad_norm": 1.1777037861050252, "learning_rate": 0.003, "loss": 4.0483, "step": 24535 }, { "epoch": 0.24536, "grad_norm": 1.4839888861445882, "learning_rate": 0.003, "loss": 4.0477, "step": 24536 }, { "epoch": 0.24537, "grad_norm": 1.135476649976957, "learning_rate": 0.003, "loss": 4.041, "step": 24537 }, { "epoch": 0.24538, "grad_norm": 1.6042579829009473, "learning_rate": 0.003, "loss": 4.0115, "step": 24538 }, { "epoch": 0.24539, "grad_norm": 1.0277540019348497, "learning_rate": 0.003, "loss": 4.0371, "step": 24539 }, { "epoch": 0.2454, "grad_norm": 1.4905150030061323, "learning_rate": 0.003, "loss": 4.0557, "step": 24540 }, { "epoch": 0.24541, "grad_norm": 1.0833401398421967, "learning_rate": 0.003, "loss": 4.0497, "step": 24541 }, { "epoch": 0.24542, "grad_norm": 1.6393310050095853, "learning_rate": 0.003, "loss": 4.059, "step": 24542 }, { "epoch": 0.24543, "grad_norm": 1.2489270071015557, "learning_rate": 0.003, "loss": 4.0664, "step": 24543 }, { "epoch": 0.24544, "grad_norm": 1.3765663950696332, "learning_rate": 0.003, "loss": 4.0737, "step": 24544 }, { "epoch": 0.24545, "grad_norm": 1.2498631490998917, "learning_rate": 0.003, "loss": 4.036, "step": 24545 }, { "epoch": 0.24546, "grad_norm": 1.222010572283642, "learning_rate": 0.003, "loss": 4.0594, "step": 24546 }, { "epoch": 0.24547, "grad_norm": 1.3206084983028674, "learning_rate": 0.003, "loss": 4.0758, "step": 24547 }, { "epoch": 0.24548, "grad_norm": 1.236253129593086, "learning_rate": 0.003, "loss": 4.0515, "step": 24548 }, { "epoch": 0.24549, "grad_norm": 1.4034262110781486, "learning_rate": 0.003, "loss": 4.0482, "step": 24549 }, { "epoch": 0.2455, "grad_norm": 1.1882870256342752, "learning_rate": 0.003, "loss": 4.0311, "step": 24550 }, { "epoch": 0.24551, "grad_norm": 1.5632220912390589, "learning_rate": 0.003, "loss": 4.0521, "step": 24551 }, { "epoch": 0.24552, "grad_norm": 1.346154084408589, "learning_rate": 0.003, "loss": 4.0441, "step": 24552 }, { "epoch": 0.24553, "grad_norm": 1.2091822152564482, "learning_rate": 0.003, "loss": 4.0509, "step": 24553 }, { "epoch": 0.24554, "grad_norm": 1.1585792156401415, "learning_rate": 0.003, "loss": 4.0724, "step": 24554 }, { "epoch": 0.24555, "grad_norm": 1.1888195496287441, "learning_rate": 0.003, "loss": 4.0343, "step": 24555 }, { "epoch": 0.24556, "grad_norm": 1.2994592204498774, "learning_rate": 0.003, "loss": 4.0433, "step": 24556 }, { "epoch": 0.24557, "grad_norm": 1.2953732094825978, "learning_rate": 0.003, "loss": 4.009, "step": 24557 }, { "epoch": 0.24558, "grad_norm": 1.4551132309697592, "learning_rate": 0.003, "loss": 4.0713, "step": 24558 }, { "epoch": 0.24559, "grad_norm": 1.2664372475090542, "learning_rate": 0.003, "loss": 4.0523, "step": 24559 }, { "epoch": 0.2456, "grad_norm": 1.5680081647814643, "learning_rate": 0.003, "loss": 4.0847, "step": 24560 }, { "epoch": 0.24561, "grad_norm": 1.4986508065676625, "learning_rate": 0.003, "loss": 4.0229, "step": 24561 }, { "epoch": 0.24562, "grad_norm": 1.3175557421315478, "learning_rate": 0.003, "loss": 4.0353, "step": 24562 }, { "epoch": 0.24563, "grad_norm": 1.487003322296809, "learning_rate": 0.003, "loss": 4.0706, "step": 24563 }, { "epoch": 0.24564, "grad_norm": 0.9274034468608103, "learning_rate": 0.003, "loss": 4.0239, "step": 24564 }, { "epoch": 0.24565, "grad_norm": 1.315593903885781, "learning_rate": 0.003, "loss": 4.0375, "step": 24565 }, { "epoch": 0.24566, "grad_norm": 1.1380597690493108, "learning_rate": 0.003, "loss": 4.0381, "step": 24566 }, { "epoch": 0.24567, "grad_norm": 1.5286179953409904, "learning_rate": 0.003, "loss": 4.0562, "step": 24567 }, { "epoch": 0.24568, "grad_norm": 1.2748392172647982, "learning_rate": 0.003, "loss": 4.0399, "step": 24568 }, { "epoch": 0.24569, "grad_norm": 1.5881659245536273, "learning_rate": 0.003, "loss": 4.0157, "step": 24569 }, { "epoch": 0.2457, "grad_norm": 1.1704966863521384, "learning_rate": 0.003, "loss": 4.0519, "step": 24570 }, { "epoch": 0.24571, "grad_norm": 1.363024976952242, "learning_rate": 0.003, "loss": 4.0729, "step": 24571 }, { "epoch": 0.24572, "grad_norm": 1.209518326800387, "learning_rate": 0.003, "loss": 4.0408, "step": 24572 }, { "epoch": 0.24573, "grad_norm": 1.5397765217767965, "learning_rate": 0.003, "loss": 4.0503, "step": 24573 }, { "epoch": 0.24574, "grad_norm": 1.134602315350047, "learning_rate": 0.003, "loss": 4.0659, "step": 24574 }, { "epoch": 0.24575, "grad_norm": 1.4678815697012149, "learning_rate": 0.003, "loss": 4.0337, "step": 24575 }, { "epoch": 0.24576, "grad_norm": 1.0790848780382025, "learning_rate": 0.003, "loss": 4.0397, "step": 24576 }, { "epoch": 0.24577, "grad_norm": 1.335404480150775, "learning_rate": 0.003, "loss": 4.0595, "step": 24577 }, { "epoch": 0.24578, "grad_norm": 1.3021074891682802, "learning_rate": 0.003, "loss": 4.0414, "step": 24578 }, { "epoch": 0.24579, "grad_norm": 1.2286809005795094, "learning_rate": 0.003, "loss": 4.0395, "step": 24579 }, { "epoch": 0.2458, "grad_norm": 1.4162283650362506, "learning_rate": 0.003, "loss": 4.0306, "step": 24580 }, { "epoch": 0.24581, "grad_norm": 1.2150062756548052, "learning_rate": 0.003, "loss": 4.036, "step": 24581 }, { "epoch": 0.24582, "grad_norm": 1.2836573080150377, "learning_rate": 0.003, "loss": 4.0877, "step": 24582 }, { "epoch": 0.24583, "grad_norm": 1.1362060948012536, "learning_rate": 0.003, "loss": 4.0499, "step": 24583 }, { "epoch": 0.24584, "grad_norm": 1.3166619351385562, "learning_rate": 0.003, "loss": 4.0673, "step": 24584 }, { "epoch": 0.24585, "grad_norm": 1.361701905821797, "learning_rate": 0.003, "loss": 4.0785, "step": 24585 }, { "epoch": 0.24586, "grad_norm": 1.174692030608153, "learning_rate": 0.003, "loss": 4.0491, "step": 24586 }, { "epoch": 0.24587, "grad_norm": 1.5193122545143058, "learning_rate": 0.003, "loss": 4.053, "step": 24587 }, { "epoch": 0.24588, "grad_norm": 1.1221785087614495, "learning_rate": 0.003, "loss": 4.028, "step": 24588 }, { "epoch": 0.24589, "grad_norm": 1.4140066228993404, "learning_rate": 0.003, "loss": 4.024, "step": 24589 }, { "epoch": 0.2459, "grad_norm": 1.2189373034345057, "learning_rate": 0.003, "loss": 4.0534, "step": 24590 }, { "epoch": 0.24591, "grad_norm": 1.3954985919171259, "learning_rate": 0.003, "loss": 4.0438, "step": 24591 }, { "epoch": 0.24592, "grad_norm": 1.2570197988187979, "learning_rate": 0.003, "loss": 4.058, "step": 24592 }, { "epoch": 0.24593, "grad_norm": 1.3830657924088505, "learning_rate": 0.003, "loss": 4.0647, "step": 24593 }, { "epoch": 0.24594, "grad_norm": 1.3109099888904479, "learning_rate": 0.003, "loss": 4.0476, "step": 24594 }, { "epoch": 0.24595, "grad_norm": 1.4752542833426738, "learning_rate": 0.003, "loss": 4.0445, "step": 24595 }, { "epoch": 0.24596, "grad_norm": 1.8093942114634118, "learning_rate": 0.003, "loss": 4.0507, "step": 24596 }, { "epoch": 0.24597, "grad_norm": 1.1875163494228915, "learning_rate": 0.003, "loss": 4.0748, "step": 24597 }, { "epoch": 0.24598, "grad_norm": 1.5295106400673966, "learning_rate": 0.003, "loss": 4.0514, "step": 24598 }, { "epoch": 0.24599, "grad_norm": 1.5553621730584248, "learning_rate": 0.003, "loss": 4.023, "step": 24599 }, { "epoch": 0.246, "grad_norm": 1.0943708649938018, "learning_rate": 0.003, "loss": 4.0025, "step": 24600 }, { "epoch": 0.24601, "grad_norm": 1.5324238663763916, "learning_rate": 0.003, "loss": 4.0545, "step": 24601 }, { "epoch": 0.24602, "grad_norm": 0.9865929654013261, "learning_rate": 0.003, "loss": 4.0511, "step": 24602 }, { "epoch": 0.24603, "grad_norm": 1.39837091860186, "learning_rate": 0.003, "loss": 4.0231, "step": 24603 }, { "epoch": 0.24604, "grad_norm": 1.1674379021235621, "learning_rate": 0.003, "loss": 4.0381, "step": 24604 }, { "epoch": 0.24605, "grad_norm": 1.3303618331297282, "learning_rate": 0.003, "loss": 4.0522, "step": 24605 }, { "epoch": 0.24606, "grad_norm": 1.3426334514107259, "learning_rate": 0.003, "loss": 4.0248, "step": 24606 }, { "epoch": 0.24607, "grad_norm": 1.1819994896526633, "learning_rate": 0.003, "loss": 4.0637, "step": 24607 }, { "epoch": 0.24608, "grad_norm": 1.225025403617744, "learning_rate": 0.003, "loss": 4.0574, "step": 24608 }, { "epoch": 0.24609, "grad_norm": 1.353660922034048, "learning_rate": 0.003, "loss": 4.0333, "step": 24609 }, { "epoch": 0.2461, "grad_norm": 1.4180331439397997, "learning_rate": 0.003, "loss": 4.0334, "step": 24610 }, { "epoch": 0.24611, "grad_norm": 1.391685917972673, "learning_rate": 0.003, "loss": 4.021, "step": 24611 }, { "epoch": 0.24612, "grad_norm": 1.1943647528055936, "learning_rate": 0.003, "loss": 4.0172, "step": 24612 }, { "epoch": 0.24613, "grad_norm": 1.3195618406423208, "learning_rate": 0.003, "loss": 4.0542, "step": 24613 }, { "epoch": 0.24614, "grad_norm": 1.270598821188231, "learning_rate": 0.003, "loss": 4.035, "step": 24614 }, { "epoch": 0.24615, "grad_norm": 1.3185533091717765, "learning_rate": 0.003, "loss": 4.0394, "step": 24615 }, { "epoch": 0.24616, "grad_norm": 1.1593932524997574, "learning_rate": 0.003, "loss": 4.0511, "step": 24616 }, { "epoch": 0.24617, "grad_norm": 1.5655085440115326, "learning_rate": 0.003, "loss": 4.0447, "step": 24617 }, { "epoch": 0.24618, "grad_norm": 1.075335990727989, "learning_rate": 0.003, "loss": 4.049, "step": 24618 }, { "epoch": 0.24619, "grad_norm": 1.420058777248943, "learning_rate": 0.003, "loss": 4.0574, "step": 24619 }, { "epoch": 0.2462, "grad_norm": 1.0918268751673923, "learning_rate": 0.003, "loss": 4.057, "step": 24620 }, { "epoch": 0.24621, "grad_norm": 1.976111999697177, "learning_rate": 0.003, "loss": 4.087, "step": 24621 }, { "epoch": 0.24622, "grad_norm": 1.0911559202404375, "learning_rate": 0.003, "loss": 4.037, "step": 24622 }, { "epoch": 0.24623, "grad_norm": 1.4037741843675444, "learning_rate": 0.003, "loss": 4.0516, "step": 24623 }, { "epoch": 0.24624, "grad_norm": 1.255028535780465, "learning_rate": 0.003, "loss": 4.0495, "step": 24624 }, { "epoch": 0.24625, "grad_norm": 1.4018604150067269, "learning_rate": 0.003, "loss": 4.0591, "step": 24625 }, { "epoch": 0.24626, "grad_norm": 1.5939892138353882, "learning_rate": 0.003, "loss": 4.0321, "step": 24626 }, { "epoch": 0.24627, "grad_norm": 1.4351595313608372, "learning_rate": 0.003, "loss": 4.0459, "step": 24627 }, { "epoch": 0.24628, "grad_norm": 1.4204216618273355, "learning_rate": 0.003, "loss": 4.0575, "step": 24628 }, { "epoch": 0.24629, "grad_norm": 1.1369933346609602, "learning_rate": 0.003, "loss": 4.0398, "step": 24629 }, { "epoch": 0.2463, "grad_norm": 1.5131769063727587, "learning_rate": 0.003, "loss": 4.0281, "step": 24630 }, { "epoch": 0.24631, "grad_norm": 1.0898521553812368, "learning_rate": 0.003, "loss": 4.0432, "step": 24631 }, { "epoch": 0.24632, "grad_norm": 1.3970642840223215, "learning_rate": 0.003, "loss": 4.0246, "step": 24632 }, { "epoch": 0.24633, "grad_norm": 1.0579311195392256, "learning_rate": 0.003, "loss": 4.0399, "step": 24633 }, { "epoch": 0.24634, "grad_norm": 1.3657986467401328, "learning_rate": 0.003, "loss": 4.0211, "step": 24634 }, { "epoch": 0.24635, "grad_norm": 1.109865105267184, "learning_rate": 0.003, "loss": 4.0633, "step": 24635 }, { "epoch": 0.24636, "grad_norm": 1.4600265981516272, "learning_rate": 0.003, "loss": 4.0567, "step": 24636 }, { "epoch": 0.24637, "grad_norm": 1.3169650620458038, "learning_rate": 0.003, "loss": 4.0353, "step": 24637 }, { "epoch": 0.24638, "grad_norm": 1.5109969429114762, "learning_rate": 0.003, "loss": 4.0373, "step": 24638 }, { "epoch": 0.24639, "grad_norm": 1.2036021203825946, "learning_rate": 0.003, "loss": 4.0237, "step": 24639 }, { "epoch": 0.2464, "grad_norm": 1.2882482932566781, "learning_rate": 0.003, "loss": 4.025, "step": 24640 }, { "epoch": 0.24641, "grad_norm": 1.6248550739869179, "learning_rate": 0.003, "loss": 4.0626, "step": 24641 }, { "epoch": 0.24642, "grad_norm": 1.0442128909171355, "learning_rate": 0.003, "loss": 4.0391, "step": 24642 }, { "epoch": 0.24643, "grad_norm": 1.393027503895784, "learning_rate": 0.003, "loss": 4.0736, "step": 24643 }, { "epoch": 0.24644, "grad_norm": 1.1759422390190883, "learning_rate": 0.003, "loss": 4.0347, "step": 24644 }, { "epoch": 0.24645, "grad_norm": 1.5185546369545304, "learning_rate": 0.003, "loss": 4.0521, "step": 24645 }, { "epoch": 0.24646, "grad_norm": 1.09837983903636, "learning_rate": 0.003, "loss": 4.036, "step": 24646 }, { "epoch": 0.24647, "grad_norm": 1.6195419550061216, "learning_rate": 0.003, "loss": 4.0492, "step": 24647 }, { "epoch": 0.24648, "grad_norm": 1.3013617301688172, "learning_rate": 0.003, "loss": 4.0523, "step": 24648 }, { "epoch": 0.24649, "grad_norm": 1.1932286571998207, "learning_rate": 0.003, "loss": 4.0371, "step": 24649 }, { "epoch": 0.2465, "grad_norm": 1.39804739285826, "learning_rate": 0.003, "loss": 4.0479, "step": 24650 }, { "epoch": 0.24651, "grad_norm": 1.291531026975577, "learning_rate": 0.003, "loss": 4.0242, "step": 24651 }, { "epoch": 0.24652, "grad_norm": 1.3042821706559988, "learning_rate": 0.003, "loss": 4.0258, "step": 24652 }, { "epoch": 0.24653, "grad_norm": 1.3248442248467729, "learning_rate": 0.003, "loss": 4.0257, "step": 24653 }, { "epoch": 0.24654, "grad_norm": 1.3082035128477907, "learning_rate": 0.003, "loss": 4.0693, "step": 24654 }, { "epoch": 0.24655, "grad_norm": 1.3693317044900468, "learning_rate": 0.003, "loss": 4.0863, "step": 24655 }, { "epoch": 0.24656, "grad_norm": 1.11185556858586, "learning_rate": 0.003, "loss": 4.0534, "step": 24656 }, { "epoch": 0.24657, "grad_norm": 1.6261556068405063, "learning_rate": 0.003, "loss": 4.0308, "step": 24657 }, { "epoch": 0.24658, "grad_norm": 1.2304751326220844, "learning_rate": 0.003, "loss": 4.0323, "step": 24658 }, { "epoch": 0.24659, "grad_norm": 1.4508123845516905, "learning_rate": 0.003, "loss": 4.0375, "step": 24659 }, { "epoch": 0.2466, "grad_norm": 1.2527596978161486, "learning_rate": 0.003, "loss": 4.0492, "step": 24660 }, { "epoch": 0.24661, "grad_norm": 1.4080833239751158, "learning_rate": 0.003, "loss": 4.0254, "step": 24661 }, { "epoch": 0.24662, "grad_norm": 1.1748121378591787, "learning_rate": 0.003, "loss": 4.0097, "step": 24662 }, { "epoch": 0.24663, "grad_norm": 1.148162313165084, "learning_rate": 0.003, "loss": 4.0368, "step": 24663 }, { "epoch": 0.24664, "grad_norm": 1.5627436868044455, "learning_rate": 0.003, "loss": 4.0434, "step": 24664 }, { "epoch": 0.24665, "grad_norm": 1.0266064427870762, "learning_rate": 0.003, "loss": 4.0281, "step": 24665 }, { "epoch": 0.24666, "grad_norm": 1.3119867924575164, "learning_rate": 0.003, "loss": 4.0399, "step": 24666 }, { "epoch": 0.24667, "grad_norm": 1.246639895630165, "learning_rate": 0.003, "loss": 4.0345, "step": 24667 }, { "epoch": 0.24668, "grad_norm": 1.8385608686779658, "learning_rate": 0.003, "loss": 4.0564, "step": 24668 }, { "epoch": 0.24669, "grad_norm": 1.1381394160669267, "learning_rate": 0.003, "loss": 4.0416, "step": 24669 }, { "epoch": 0.2467, "grad_norm": 1.2824021096641982, "learning_rate": 0.003, "loss": 4.02, "step": 24670 }, { "epoch": 0.24671, "grad_norm": 1.1762549218653497, "learning_rate": 0.003, "loss": 4.0365, "step": 24671 }, { "epoch": 0.24672, "grad_norm": 1.58865933635969, "learning_rate": 0.003, "loss": 4.0859, "step": 24672 }, { "epoch": 0.24673, "grad_norm": 1.1840540412335352, "learning_rate": 0.003, "loss": 4.0315, "step": 24673 }, { "epoch": 0.24674, "grad_norm": 1.2995494641211922, "learning_rate": 0.003, "loss": 4.0461, "step": 24674 }, { "epoch": 0.24675, "grad_norm": 1.3030218988693136, "learning_rate": 0.003, "loss": 4.0528, "step": 24675 }, { "epoch": 0.24676, "grad_norm": 1.2629691373253202, "learning_rate": 0.003, "loss": 4.0541, "step": 24676 }, { "epoch": 0.24677, "grad_norm": 1.3937666364863655, "learning_rate": 0.003, "loss": 4.0654, "step": 24677 }, { "epoch": 0.24678, "grad_norm": 1.3413538812072792, "learning_rate": 0.003, "loss": 4.038, "step": 24678 }, { "epoch": 0.24679, "grad_norm": 1.197981971325603, "learning_rate": 0.003, "loss": 4.0395, "step": 24679 }, { "epoch": 0.2468, "grad_norm": 1.3491724993718825, "learning_rate": 0.003, "loss": 4.0278, "step": 24680 }, { "epoch": 0.24681, "grad_norm": 1.426774027491637, "learning_rate": 0.003, "loss": 4.0292, "step": 24681 }, { "epoch": 0.24682, "grad_norm": 1.209217554220248, "learning_rate": 0.003, "loss": 4.0202, "step": 24682 }, { "epoch": 0.24683, "grad_norm": 1.4165296966409964, "learning_rate": 0.003, "loss": 4.0427, "step": 24683 }, { "epoch": 0.24684, "grad_norm": 1.1709265323177698, "learning_rate": 0.003, "loss": 4.0399, "step": 24684 }, { "epoch": 0.24685, "grad_norm": 1.075528142048488, "learning_rate": 0.003, "loss": 4.0123, "step": 24685 }, { "epoch": 0.24686, "grad_norm": 1.4081802136054868, "learning_rate": 0.003, "loss": 4.0565, "step": 24686 }, { "epoch": 0.24687, "grad_norm": 1.1185911634221184, "learning_rate": 0.003, "loss": 4.058, "step": 24687 }, { "epoch": 0.24688, "grad_norm": 1.4922467097586183, "learning_rate": 0.003, "loss": 4.0543, "step": 24688 }, { "epoch": 0.24689, "grad_norm": 1.303196131951756, "learning_rate": 0.003, "loss": 4.0133, "step": 24689 }, { "epoch": 0.2469, "grad_norm": 1.1429592682731438, "learning_rate": 0.003, "loss": 4.0681, "step": 24690 }, { "epoch": 0.24691, "grad_norm": 1.3209312908754482, "learning_rate": 0.003, "loss": 4.0675, "step": 24691 }, { "epoch": 0.24692, "grad_norm": 1.248526223449253, "learning_rate": 0.003, "loss": 4.0254, "step": 24692 }, { "epoch": 0.24693, "grad_norm": 1.3123696700134033, "learning_rate": 0.003, "loss": 4.0602, "step": 24693 }, { "epoch": 0.24694, "grad_norm": 1.5610281502167933, "learning_rate": 0.003, "loss": 4.0307, "step": 24694 }, { "epoch": 0.24695, "grad_norm": 1.0662173736404557, "learning_rate": 0.003, "loss": 4.0283, "step": 24695 }, { "epoch": 0.24696, "grad_norm": 1.4153850277726499, "learning_rate": 0.003, "loss": 4.0317, "step": 24696 }, { "epoch": 0.24697, "grad_norm": 1.4243857413173024, "learning_rate": 0.003, "loss": 4.0366, "step": 24697 }, { "epoch": 0.24698, "grad_norm": 1.6706664425847921, "learning_rate": 0.003, "loss": 4.0509, "step": 24698 }, { "epoch": 0.24699, "grad_norm": 1.1272008388777834, "learning_rate": 0.003, "loss": 4.062, "step": 24699 }, { "epoch": 0.247, "grad_norm": 1.315220519213806, "learning_rate": 0.003, "loss": 4.0459, "step": 24700 }, { "epoch": 0.24701, "grad_norm": 1.4036575572677676, "learning_rate": 0.003, "loss": 4.0278, "step": 24701 }, { "epoch": 0.24702, "grad_norm": 1.320689760719037, "learning_rate": 0.003, "loss": 4.0351, "step": 24702 }, { "epoch": 0.24703, "grad_norm": 1.3729646519522445, "learning_rate": 0.003, "loss": 4.0303, "step": 24703 }, { "epoch": 0.24704, "grad_norm": 1.4122543204723723, "learning_rate": 0.003, "loss": 4.0629, "step": 24704 }, { "epoch": 0.24705, "grad_norm": 1.1329785659810372, "learning_rate": 0.003, "loss": 4.056, "step": 24705 }, { "epoch": 0.24706, "grad_norm": 1.225864799648413, "learning_rate": 0.003, "loss": 4.0589, "step": 24706 }, { "epoch": 0.24707, "grad_norm": 1.3888954628970043, "learning_rate": 0.003, "loss": 4.0709, "step": 24707 }, { "epoch": 0.24708, "grad_norm": 1.2574468237656262, "learning_rate": 0.003, "loss": 4.0563, "step": 24708 }, { "epoch": 0.24709, "grad_norm": 1.3557089101752695, "learning_rate": 0.003, "loss": 4.0468, "step": 24709 }, { "epoch": 0.2471, "grad_norm": 1.5100492022687908, "learning_rate": 0.003, "loss": 4.0462, "step": 24710 }, { "epoch": 0.24711, "grad_norm": 1.1096014896358144, "learning_rate": 0.003, "loss": 4.0188, "step": 24711 }, { "epoch": 0.24712, "grad_norm": 1.3578662203165828, "learning_rate": 0.003, "loss": 4.0263, "step": 24712 }, { "epoch": 0.24713, "grad_norm": 1.2826864850854105, "learning_rate": 0.003, "loss": 4.0309, "step": 24713 }, { "epoch": 0.24714, "grad_norm": 1.2651306368814537, "learning_rate": 0.003, "loss": 4.0258, "step": 24714 }, { "epoch": 0.24715, "grad_norm": 1.3512547056840147, "learning_rate": 0.003, "loss": 4.0443, "step": 24715 }, { "epoch": 0.24716, "grad_norm": 1.1950572755603974, "learning_rate": 0.003, "loss": 4.0214, "step": 24716 }, { "epoch": 0.24717, "grad_norm": 1.430252233614018, "learning_rate": 0.003, "loss": 4.0271, "step": 24717 }, { "epoch": 0.24718, "grad_norm": 1.245590730430256, "learning_rate": 0.003, "loss": 4.057, "step": 24718 }, { "epoch": 0.24719, "grad_norm": 1.7860250976044554, "learning_rate": 0.003, "loss": 4.0361, "step": 24719 }, { "epoch": 0.2472, "grad_norm": 1.0308833507021609, "learning_rate": 0.003, "loss": 4.0299, "step": 24720 }, { "epoch": 0.24721, "grad_norm": 1.3425210027908412, "learning_rate": 0.003, "loss": 4.0634, "step": 24721 }, { "epoch": 0.24722, "grad_norm": 1.5274247626978565, "learning_rate": 0.003, "loss": 4.0501, "step": 24722 }, { "epoch": 0.24723, "grad_norm": 1.1421212200220046, "learning_rate": 0.003, "loss": 4.0098, "step": 24723 }, { "epoch": 0.24724, "grad_norm": 1.558193025652777, "learning_rate": 0.003, "loss": 4.0594, "step": 24724 }, { "epoch": 0.24725, "grad_norm": 1.0427942658773595, "learning_rate": 0.003, "loss": 4.0271, "step": 24725 }, { "epoch": 0.24726, "grad_norm": 1.5394102963297531, "learning_rate": 0.003, "loss": 4.0682, "step": 24726 }, { "epoch": 0.24727, "grad_norm": 1.260448997386808, "learning_rate": 0.003, "loss": 4.0323, "step": 24727 }, { "epoch": 0.24728, "grad_norm": 1.184134464797292, "learning_rate": 0.003, "loss": 4.0214, "step": 24728 }, { "epoch": 0.24729, "grad_norm": 1.2955163227837003, "learning_rate": 0.003, "loss": 4.072, "step": 24729 }, { "epoch": 0.2473, "grad_norm": 1.341876772559969, "learning_rate": 0.003, "loss": 4.0286, "step": 24730 }, { "epoch": 0.24731, "grad_norm": 1.2552013840497218, "learning_rate": 0.003, "loss": 4.0734, "step": 24731 }, { "epoch": 0.24732, "grad_norm": 1.6944801308915869, "learning_rate": 0.003, "loss": 4.0522, "step": 24732 }, { "epoch": 0.24733, "grad_norm": 1.1967212144215253, "learning_rate": 0.003, "loss": 4.0594, "step": 24733 }, { "epoch": 0.24734, "grad_norm": 1.165790288801221, "learning_rate": 0.003, "loss": 4.0083, "step": 24734 }, { "epoch": 0.24735, "grad_norm": 1.5622013523275582, "learning_rate": 0.003, "loss": 4.07, "step": 24735 }, { "epoch": 0.24736, "grad_norm": 1.2390457894957734, "learning_rate": 0.003, "loss": 4.0567, "step": 24736 }, { "epoch": 0.24737, "grad_norm": 1.30194934702204, "learning_rate": 0.003, "loss": 4.0354, "step": 24737 }, { "epoch": 0.24738, "grad_norm": 1.1536400015172708, "learning_rate": 0.003, "loss": 4.0471, "step": 24738 }, { "epoch": 0.24739, "grad_norm": 1.4124285335637157, "learning_rate": 0.003, "loss": 4.0813, "step": 24739 }, { "epoch": 0.2474, "grad_norm": 1.0967050518465338, "learning_rate": 0.003, "loss": 4.0523, "step": 24740 }, { "epoch": 0.24741, "grad_norm": 1.4710657972621985, "learning_rate": 0.003, "loss": 4.0437, "step": 24741 }, { "epoch": 0.24742, "grad_norm": 1.2096606346118244, "learning_rate": 0.003, "loss": 4.0472, "step": 24742 }, { "epoch": 0.24743, "grad_norm": 1.3679647516220717, "learning_rate": 0.003, "loss": 4.0333, "step": 24743 }, { "epoch": 0.24744, "grad_norm": 1.1742150836353038, "learning_rate": 0.003, "loss": 4.0255, "step": 24744 }, { "epoch": 0.24745, "grad_norm": 1.2732545294816122, "learning_rate": 0.003, "loss": 4.0392, "step": 24745 }, { "epoch": 0.24746, "grad_norm": 1.5917479743270604, "learning_rate": 0.003, "loss": 4.0588, "step": 24746 }, { "epoch": 0.24747, "grad_norm": 1.1161440284078028, "learning_rate": 0.003, "loss": 4.0633, "step": 24747 }, { "epoch": 0.24748, "grad_norm": 1.4639040731628656, "learning_rate": 0.003, "loss": 4.0556, "step": 24748 }, { "epoch": 0.24749, "grad_norm": 1.4738704838118968, "learning_rate": 0.003, "loss": 4.0474, "step": 24749 }, { "epoch": 0.2475, "grad_norm": 1.5126924294911215, "learning_rate": 0.003, "loss": 4.0203, "step": 24750 }, { "epoch": 0.24751, "grad_norm": 1.1082978514104664, "learning_rate": 0.003, "loss": 4.0574, "step": 24751 }, { "epoch": 0.24752, "grad_norm": 1.438038418817806, "learning_rate": 0.003, "loss": 4.0301, "step": 24752 }, { "epoch": 0.24753, "grad_norm": 1.2730727211030077, "learning_rate": 0.003, "loss": 4.0428, "step": 24753 }, { "epoch": 0.24754, "grad_norm": 1.397641301636969, "learning_rate": 0.003, "loss": 4.0731, "step": 24754 }, { "epoch": 0.24755, "grad_norm": 1.248110866539072, "learning_rate": 0.003, "loss": 4.0012, "step": 24755 }, { "epoch": 0.24756, "grad_norm": 1.434642856028332, "learning_rate": 0.003, "loss": 4.0522, "step": 24756 }, { "epoch": 0.24757, "grad_norm": 1.3081070512561468, "learning_rate": 0.003, "loss": 4.024, "step": 24757 }, { "epoch": 0.24758, "grad_norm": 1.3937514744875894, "learning_rate": 0.003, "loss": 4.0374, "step": 24758 }, { "epoch": 0.24759, "grad_norm": 1.162035572462083, "learning_rate": 0.003, "loss": 4.0531, "step": 24759 }, { "epoch": 0.2476, "grad_norm": 1.3917216881671628, "learning_rate": 0.003, "loss": 4.0528, "step": 24760 }, { "epoch": 0.24761, "grad_norm": 1.294718613942674, "learning_rate": 0.003, "loss": 4.0324, "step": 24761 }, { "epoch": 0.24762, "grad_norm": 1.186068602184914, "learning_rate": 0.003, "loss": 4.0707, "step": 24762 }, { "epoch": 0.24763, "grad_norm": 1.488225379369762, "learning_rate": 0.003, "loss": 4.072, "step": 24763 }, { "epoch": 0.24764, "grad_norm": 1.1990694990406738, "learning_rate": 0.003, "loss": 4.0653, "step": 24764 }, { "epoch": 0.24765, "grad_norm": 1.4359569549121276, "learning_rate": 0.003, "loss": 4.0467, "step": 24765 }, { "epoch": 0.24766, "grad_norm": 1.3549635171675725, "learning_rate": 0.003, "loss": 4.0489, "step": 24766 }, { "epoch": 0.24767, "grad_norm": 1.4808283938961393, "learning_rate": 0.003, "loss": 4.0823, "step": 24767 }, { "epoch": 0.24768, "grad_norm": 1.2438624114965242, "learning_rate": 0.003, "loss": 4.0451, "step": 24768 }, { "epoch": 0.24769, "grad_norm": 1.1912376400445561, "learning_rate": 0.003, "loss": 4.0171, "step": 24769 }, { "epoch": 0.2477, "grad_norm": 1.2325990814215235, "learning_rate": 0.003, "loss": 4.0593, "step": 24770 }, { "epoch": 0.24771, "grad_norm": 1.2736129763514705, "learning_rate": 0.003, "loss": 4.0466, "step": 24771 }, { "epoch": 0.24772, "grad_norm": 1.465094437512808, "learning_rate": 0.003, "loss": 4.02, "step": 24772 }, { "epoch": 0.24773, "grad_norm": 1.2135453610233906, "learning_rate": 0.003, "loss": 4.0282, "step": 24773 }, { "epoch": 0.24774, "grad_norm": 1.3624481935093014, "learning_rate": 0.003, "loss": 4.0468, "step": 24774 }, { "epoch": 0.24775, "grad_norm": 1.1536811584715754, "learning_rate": 0.003, "loss": 4.0254, "step": 24775 }, { "epoch": 0.24776, "grad_norm": 1.3187340249875852, "learning_rate": 0.003, "loss": 4.0569, "step": 24776 }, { "epoch": 0.24777, "grad_norm": 1.233953781610818, "learning_rate": 0.003, "loss": 4.049, "step": 24777 }, { "epoch": 0.24778, "grad_norm": 1.3960283235642599, "learning_rate": 0.003, "loss": 4.0684, "step": 24778 }, { "epoch": 0.24779, "grad_norm": 1.3948062730493118, "learning_rate": 0.003, "loss": 4.0335, "step": 24779 }, { "epoch": 0.2478, "grad_norm": 1.5227610942988106, "learning_rate": 0.003, "loss": 4.0557, "step": 24780 }, { "epoch": 0.24781, "grad_norm": 1.325066226081347, "learning_rate": 0.003, "loss": 4.026, "step": 24781 }, { "epoch": 0.24782, "grad_norm": 1.2162613188969975, "learning_rate": 0.003, "loss": 4.0511, "step": 24782 }, { "epoch": 0.24783, "grad_norm": 1.3143009857887438, "learning_rate": 0.003, "loss": 4.065, "step": 24783 }, { "epoch": 0.24784, "grad_norm": 1.1747425297920604, "learning_rate": 0.003, "loss": 4.047, "step": 24784 }, { "epoch": 0.24785, "grad_norm": 1.2020442552266992, "learning_rate": 0.003, "loss": 4.0338, "step": 24785 }, { "epoch": 0.24786, "grad_norm": 1.4193470845232725, "learning_rate": 0.003, "loss": 4.0581, "step": 24786 }, { "epoch": 0.24787, "grad_norm": 1.0744574878557753, "learning_rate": 0.003, "loss": 4.0408, "step": 24787 }, { "epoch": 0.24788, "grad_norm": 1.3867147598561682, "learning_rate": 0.003, "loss": 4.0273, "step": 24788 }, { "epoch": 0.24789, "grad_norm": 1.1033279944956433, "learning_rate": 0.003, "loss": 4.041, "step": 24789 }, { "epoch": 0.2479, "grad_norm": 1.514456470926197, "learning_rate": 0.003, "loss": 4.0456, "step": 24790 }, { "epoch": 0.24791, "grad_norm": 1.1661299733164954, "learning_rate": 0.003, "loss": 4.035, "step": 24791 }, { "epoch": 0.24792, "grad_norm": 1.3674813529224923, "learning_rate": 0.003, "loss": 4.0343, "step": 24792 }, { "epoch": 0.24793, "grad_norm": 1.3447654072368525, "learning_rate": 0.003, "loss": 4.03, "step": 24793 }, { "epoch": 0.24794, "grad_norm": 1.2024270036449556, "learning_rate": 0.003, "loss": 4.04, "step": 24794 }, { "epoch": 0.24795, "grad_norm": 1.189755279279851, "learning_rate": 0.003, "loss": 4.0484, "step": 24795 }, { "epoch": 0.24796, "grad_norm": 1.4394369789290944, "learning_rate": 0.003, "loss": 4.0596, "step": 24796 }, { "epoch": 0.24797, "grad_norm": 1.490508812769484, "learning_rate": 0.003, "loss": 4.0379, "step": 24797 }, { "epoch": 0.24798, "grad_norm": 1.1360240109791595, "learning_rate": 0.003, "loss": 4.0496, "step": 24798 }, { "epoch": 0.24799, "grad_norm": 1.3107246746499897, "learning_rate": 0.003, "loss": 4.0368, "step": 24799 }, { "epoch": 0.248, "grad_norm": 1.388135958393749, "learning_rate": 0.003, "loss": 4.0245, "step": 24800 }, { "epoch": 0.24801, "grad_norm": 1.4166670972679642, "learning_rate": 0.003, "loss": 4.0754, "step": 24801 }, { "epoch": 0.24802, "grad_norm": 1.3719663717740094, "learning_rate": 0.003, "loss": 4.0239, "step": 24802 }, { "epoch": 0.24803, "grad_norm": 1.315136630776163, "learning_rate": 0.003, "loss": 4.0367, "step": 24803 }, { "epoch": 0.24804, "grad_norm": 1.6254520770721155, "learning_rate": 0.003, "loss": 4.0716, "step": 24804 }, { "epoch": 0.24805, "grad_norm": 1.204621142480316, "learning_rate": 0.003, "loss": 4.017, "step": 24805 }, { "epoch": 0.24806, "grad_norm": 1.3101489215025623, "learning_rate": 0.003, "loss": 4.011, "step": 24806 }, { "epoch": 0.24807, "grad_norm": 1.2783580755497896, "learning_rate": 0.003, "loss": 4.038, "step": 24807 }, { "epoch": 0.24808, "grad_norm": 1.345337236584261, "learning_rate": 0.003, "loss": 4.0488, "step": 24808 }, { "epoch": 0.24809, "grad_norm": 1.2315383657301913, "learning_rate": 0.003, "loss": 4.0455, "step": 24809 }, { "epoch": 0.2481, "grad_norm": 1.3277653672294916, "learning_rate": 0.003, "loss": 4.0238, "step": 24810 }, { "epoch": 0.24811, "grad_norm": 1.1690599936909545, "learning_rate": 0.003, "loss": 4.0346, "step": 24811 }, { "epoch": 0.24812, "grad_norm": 1.5045631353919042, "learning_rate": 0.003, "loss": 4.0875, "step": 24812 }, { "epoch": 0.24813, "grad_norm": 1.0267103629545606, "learning_rate": 0.003, "loss": 4.03, "step": 24813 }, { "epoch": 0.24814, "grad_norm": 1.3509188671929264, "learning_rate": 0.003, "loss": 4.0512, "step": 24814 }, { "epoch": 0.24815, "grad_norm": 1.033809952506082, "learning_rate": 0.003, "loss": 4.0196, "step": 24815 }, { "epoch": 0.24816, "grad_norm": 1.7073427245155273, "learning_rate": 0.003, "loss": 4.0508, "step": 24816 }, { "epoch": 0.24817, "grad_norm": 1.1447934399893742, "learning_rate": 0.003, "loss": 4.0536, "step": 24817 }, { "epoch": 0.24818, "grad_norm": 1.7777331993370546, "learning_rate": 0.003, "loss": 4.0583, "step": 24818 }, { "epoch": 0.24819, "grad_norm": 1.2187031506523225, "learning_rate": 0.003, "loss": 4.0332, "step": 24819 }, { "epoch": 0.2482, "grad_norm": 1.2791381097948606, "learning_rate": 0.003, "loss": 4.0812, "step": 24820 }, { "epoch": 0.24821, "grad_norm": 1.3223645942575406, "learning_rate": 0.003, "loss": 4.0796, "step": 24821 }, { "epoch": 0.24822, "grad_norm": 1.3462386833969726, "learning_rate": 0.003, "loss": 4.0519, "step": 24822 }, { "epoch": 0.24823, "grad_norm": 1.3254529953182204, "learning_rate": 0.003, "loss": 4.0561, "step": 24823 }, { "epoch": 0.24824, "grad_norm": 1.3873215194984942, "learning_rate": 0.003, "loss": 4.014, "step": 24824 }, { "epoch": 0.24825, "grad_norm": 1.2892530199315535, "learning_rate": 0.003, "loss": 4.0469, "step": 24825 }, { "epoch": 0.24826, "grad_norm": 1.1606606625653142, "learning_rate": 0.003, "loss": 4.0678, "step": 24826 }, { "epoch": 0.24827, "grad_norm": 1.3272815149363186, "learning_rate": 0.003, "loss": 4.0362, "step": 24827 }, { "epoch": 0.24828, "grad_norm": 1.2169734939715162, "learning_rate": 0.003, "loss": 4.0401, "step": 24828 }, { "epoch": 0.24829, "grad_norm": 1.321292976815381, "learning_rate": 0.003, "loss": 4.0765, "step": 24829 }, { "epoch": 0.2483, "grad_norm": 1.2275146962478736, "learning_rate": 0.003, "loss": 4.0353, "step": 24830 }, { "epoch": 0.24831, "grad_norm": 1.2849327916822382, "learning_rate": 0.003, "loss": 4.0568, "step": 24831 }, { "epoch": 0.24832, "grad_norm": 1.2552342176496158, "learning_rate": 0.003, "loss": 4.0328, "step": 24832 }, { "epoch": 0.24833, "grad_norm": 1.1850142780196402, "learning_rate": 0.003, "loss": 4.0297, "step": 24833 }, { "epoch": 0.24834, "grad_norm": 1.4761905541094922, "learning_rate": 0.003, "loss": 4.0507, "step": 24834 }, { "epoch": 0.24835, "grad_norm": 0.9952552845547482, "learning_rate": 0.003, "loss": 4.0838, "step": 24835 }, { "epoch": 0.24836, "grad_norm": 1.64567562436426, "learning_rate": 0.003, "loss": 4.0759, "step": 24836 }, { "epoch": 0.24837, "grad_norm": 1.0335218112244615, "learning_rate": 0.003, "loss": 4.0591, "step": 24837 }, { "epoch": 0.24838, "grad_norm": 1.456751886992233, "learning_rate": 0.003, "loss": 4.068, "step": 24838 }, { "epoch": 0.24839, "grad_norm": 1.370694060841814, "learning_rate": 0.003, "loss": 4.0472, "step": 24839 }, { "epoch": 0.2484, "grad_norm": 1.6010495367342046, "learning_rate": 0.003, "loss": 4.0799, "step": 24840 }, { "epoch": 0.24841, "grad_norm": 1.2585243376857214, "learning_rate": 0.003, "loss": 4.061, "step": 24841 }, { "epoch": 0.24842, "grad_norm": 1.3955169206604339, "learning_rate": 0.003, "loss": 4.0374, "step": 24842 }, { "epoch": 0.24843, "grad_norm": 1.217558330392425, "learning_rate": 0.003, "loss": 4.0463, "step": 24843 }, { "epoch": 0.24844, "grad_norm": 1.3320087226292747, "learning_rate": 0.003, "loss": 4.0601, "step": 24844 }, { "epoch": 0.24845, "grad_norm": 1.2662195145008068, "learning_rate": 0.003, "loss": 4.0597, "step": 24845 }, { "epoch": 0.24846, "grad_norm": 1.6459588163912817, "learning_rate": 0.003, "loss": 4.0541, "step": 24846 }, { "epoch": 0.24847, "grad_norm": 1.0615972540018532, "learning_rate": 0.003, "loss": 4.034, "step": 24847 }, { "epoch": 0.24848, "grad_norm": 1.4247782708807473, "learning_rate": 0.003, "loss": 4.0545, "step": 24848 }, { "epoch": 0.24849, "grad_norm": 1.404094115799113, "learning_rate": 0.003, "loss": 4.0744, "step": 24849 }, { "epoch": 0.2485, "grad_norm": 1.1321383446283486, "learning_rate": 0.003, "loss": 4.038, "step": 24850 }, { "epoch": 0.24851, "grad_norm": 1.4268603954689754, "learning_rate": 0.003, "loss": 4.0403, "step": 24851 }, { "epoch": 0.24852, "grad_norm": 1.124554039505304, "learning_rate": 0.003, "loss": 4.0584, "step": 24852 }, { "epoch": 0.24853, "grad_norm": 1.5391706100709552, "learning_rate": 0.003, "loss": 4.0476, "step": 24853 }, { "epoch": 0.24854, "grad_norm": 1.345324769236744, "learning_rate": 0.003, "loss": 4.06, "step": 24854 }, { "epoch": 0.24855, "grad_norm": 1.2178827084696286, "learning_rate": 0.003, "loss": 4.0301, "step": 24855 }, { "epoch": 0.24856, "grad_norm": 1.3985756046679478, "learning_rate": 0.003, "loss": 4.0618, "step": 24856 }, { "epoch": 0.24857, "grad_norm": 1.282094905809099, "learning_rate": 0.003, "loss": 4.0323, "step": 24857 }, { "epoch": 0.24858, "grad_norm": 1.3971841829725187, "learning_rate": 0.003, "loss": 4.0444, "step": 24858 }, { "epoch": 0.24859, "grad_norm": 1.0167270092121614, "learning_rate": 0.003, "loss": 4.029, "step": 24859 }, { "epoch": 0.2486, "grad_norm": 1.6410253455301005, "learning_rate": 0.003, "loss": 4.0628, "step": 24860 }, { "epoch": 0.24861, "grad_norm": 1.0947343488654815, "learning_rate": 0.003, "loss": 4.0496, "step": 24861 }, { "epoch": 0.24862, "grad_norm": 1.615653705581511, "learning_rate": 0.003, "loss": 4.0453, "step": 24862 }, { "epoch": 0.24863, "grad_norm": 1.0480602943795017, "learning_rate": 0.003, "loss": 4.0575, "step": 24863 }, { "epoch": 0.24864, "grad_norm": 1.4340938238238872, "learning_rate": 0.003, "loss": 4.0439, "step": 24864 }, { "epoch": 0.24865, "grad_norm": 1.256553088722349, "learning_rate": 0.003, "loss": 4.0539, "step": 24865 }, { "epoch": 0.24866, "grad_norm": 1.2796771910035982, "learning_rate": 0.003, "loss": 4.0465, "step": 24866 }, { "epoch": 0.24867, "grad_norm": 1.4318073566017118, "learning_rate": 0.003, "loss": 4.0399, "step": 24867 }, { "epoch": 0.24868, "grad_norm": 1.3578843653158852, "learning_rate": 0.003, "loss": 4.0424, "step": 24868 }, { "epoch": 0.24869, "grad_norm": 1.2312214246564268, "learning_rate": 0.003, "loss": 4.027, "step": 24869 }, { "epoch": 0.2487, "grad_norm": 1.443613645179922, "learning_rate": 0.003, "loss": 4.0755, "step": 24870 }, { "epoch": 0.24871, "grad_norm": 1.1716001114947323, "learning_rate": 0.003, "loss": 4.053, "step": 24871 }, { "epoch": 0.24872, "grad_norm": 1.25469613093655, "learning_rate": 0.003, "loss": 4.0758, "step": 24872 }, { "epoch": 0.24873, "grad_norm": 1.1976399321957636, "learning_rate": 0.003, "loss": 4.0348, "step": 24873 }, { "epoch": 0.24874, "grad_norm": 1.4825872385087249, "learning_rate": 0.003, "loss": 4.0674, "step": 24874 }, { "epoch": 0.24875, "grad_norm": 1.199075633660232, "learning_rate": 0.003, "loss": 4.0442, "step": 24875 }, { "epoch": 0.24876, "grad_norm": 1.3294412459496947, "learning_rate": 0.003, "loss": 4.0491, "step": 24876 }, { "epoch": 0.24877, "grad_norm": 1.1899160828251503, "learning_rate": 0.003, "loss": 4.0858, "step": 24877 }, { "epoch": 0.24878, "grad_norm": 1.3507263116966315, "learning_rate": 0.003, "loss": 4.0269, "step": 24878 }, { "epoch": 0.24879, "grad_norm": 1.1752668344798443, "learning_rate": 0.003, "loss": 4.0554, "step": 24879 }, { "epoch": 0.2488, "grad_norm": 1.365871832238357, "learning_rate": 0.003, "loss": 4.0404, "step": 24880 }, { "epoch": 0.24881, "grad_norm": 1.1864351338647452, "learning_rate": 0.003, "loss": 4.0418, "step": 24881 }, { "epoch": 0.24882, "grad_norm": 1.8130414723760673, "learning_rate": 0.003, "loss": 4.0345, "step": 24882 }, { "epoch": 0.24883, "grad_norm": 1.1902322356464166, "learning_rate": 0.003, "loss": 4.0443, "step": 24883 }, { "epoch": 0.24884, "grad_norm": 1.3805386263469854, "learning_rate": 0.003, "loss": 4.0708, "step": 24884 }, { "epoch": 0.24885, "grad_norm": 1.393028605181782, "learning_rate": 0.003, "loss": 4.0701, "step": 24885 }, { "epoch": 0.24886, "grad_norm": 1.1816888359306001, "learning_rate": 0.003, "loss": 4.0475, "step": 24886 }, { "epoch": 0.24887, "grad_norm": 1.2593770053397053, "learning_rate": 0.003, "loss": 4.0282, "step": 24887 }, { "epoch": 0.24888, "grad_norm": 1.2623398076025991, "learning_rate": 0.003, "loss": 4.0369, "step": 24888 }, { "epoch": 0.24889, "grad_norm": 1.4081637875533586, "learning_rate": 0.003, "loss": 4.0476, "step": 24889 }, { "epoch": 0.2489, "grad_norm": 1.2347770438980452, "learning_rate": 0.003, "loss": 4.0259, "step": 24890 }, { "epoch": 0.24891, "grad_norm": 1.3143360459998792, "learning_rate": 0.003, "loss": 4.0535, "step": 24891 }, { "epoch": 0.24892, "grad_norm": 1.2505351261334239, "learning_rate": 0.003, "loss": 4.0871, "step": 24892 }, { "epoch": 0.24893, "grad_norm": 1.4881107895539392, "learning_rate": 0.003, "loss": 4.0542, "step": 24893 }, { "epoch": 0.24894, "grad_norm": 1.1980400185654467, "learning_rate": 0.003, "loss": 4.0272, "step": 24894 }, { "epoch": 0.24895, "grad_norm": 1.3963607941149216, "learning_rate": 0.003, "loss": 4.062, "step": 24895 }, { "epoch": 0.24896, "grad_norm": 1.1261507696117585, "learning_rate": 0.003, "loss": 4.0565, "step": 24896 }, { "epoch": 0.24897, "grad_norm": 1.4303821433547546, "learning_rate": 0.003, "loss": 4.0799, "step": 24897 }, { "epoch": 0.24898, "grad_norm": 1.106842255482435, "learning_rate": 0.003, "loss": 4.0562, "step": 24898 }, { "epoch": 0.24899, "grad_norm": 1.5781518443465556, "learning_rate": 0.003, "loss": 4.039, "step": 24899 }, { "epoch": 0.249, "grad_norm": 1.1324549165897573, "learning_rate": 0.003, "loss": 4.0539, "step": 24900 }, { "epoch": 0.24901, "grad_norm": 1.4005635818984645, "learning_rate": 0.003, "loss": 4.0265, "step": 24901 }, { "epoch": 0.24902, "grad_norm": 1.186576276095947, "learning_rate": 0.003, "loss": 4.0437, "step": 24902 }, { "epoch": 0.24903, "grad_norm": 1.4884892104138732, "learning_rate": 0.003, "loss": 4.0518, "step": 24903 }, { "epoch": 0.24904, "grad_norm": 1.1401012143422864, "learning_rate": 0.003, "loss": 4.0331, "step": 24904 }, { "epoch": 0.24905, "grad_norm": 1.5654247514891317, "learning_rate": 0.003, "loss": 4.078, "step": 24905 }, { "epoch": 0.24906, "grad_norm": 1.159083492034224, "learning_rate": 0.003, "loss": 4.0531, "step": 24906 }, { "epoch": 0.24907, "grad_norm": 1.3331641174505202, "learning_rate": 0.003, "loss": 4.0431, "step": 24907 }, { "epoch": 0.24908, "grad_norm": 1.3235734104050785, "learning_rate": 0.003, "loss": 4.0686, "step": 24908 }, { "epoch": 0.24909, "grad_norm": 1.266198004793703, "learning_rate": 0.003, "loss": 4.0586, "step": 24909 }, { "epoch": 0.2491, "grad_norm": 1.3622049963928764, "learning_rate": 0.003, "loss": 4.039, "step": 24910 }, { "epoch": 0.24911, "grad_norm": 1.2935211647872877, "learning_rate": 0.003, "loss": 4.0713, "step": 24911 }, { "epoch": 0.24912, "grad_norm": 1.2522251838013183, "learning_rate": 0.003, "loss": 4.0567, "step": 24912 }, { "epoch": 0.24913, "grad_norm": 1.3145815048689953, "learning_rate": 0.003, "loss": 4.0576, "step": 24913 }, { "epoch": 0.24914, "grad_norm": 1.3048849262888975, "learning_rate": 0.003, "loss": 4.0615, "step": 24914 }, { "epoch": 0.24915, "grad_norm": 1.3990876087624526, "learning_rate": 0.003, "loss": 4.0803, "step": 24915 }, { "epoch": 0.24916, "grad_norm": 1.125601860247761, "learning_rate": 0.003, "loss": 4.056, "step": 24916 }, { "epoch": 0.24917, "grad_norm": 1.4314018088372251, "learning_rate": 0.003, "loss": 4.0519, "step": 24917 }, { "epoch": 0.24918, "grad_norm": 1.2916457830616057, "learning_rate": 0.003, "loss": 4.0381, "step": 24918 }, { "epoch": 0.24919, "grad_norm": 1.383588221573143, "learning_rate": 0.003, "loss": 4.0387, "step": 24919 }, { "epoch": 0.2492, "grad_norm": 1.3577897857730004, "learning_rate": 0.003, "loss": 4.0461, "step": 24920 }, { "epoch": 0.24921, "grad_norm": 1.1757574360808953, "learning_rate": 0.003, "loss": 4.06, "step": 24921 }, { "epoch": 0.24922, "grad_norm": 1.7279469446662432, "learning_rate": 0.003, "loss": 4.0451, "step": 24922 }, { "epoch": 0.24923, "grad_norm": 1.2510772742299967, "learning_rate": 0.003, "loss": 4.0455, "step": 24923 }, { "epoch": 0.24924, "grad_norm": 1.316096105159205, "learning_rate": 0.003, "loss": 4.0404, "step": 24924 }, { "epoch": 0.24925, "grad_norm": 1.3815618594288337, "learning_rate": 0.003, "loss": 4.0421, "step": 24925 }, { "epoch": 0.24926, "grad_norm": 1.2221481336481006, "learning_rate": 0.003, "loss": 4.0414, "step": 24926 }, { "epoch": 0.24927, "grad_norm": 1.50187363948413, "learning_rate": 0.003, "loss": 4.0488, "step": 24927 }, { "epoch": 0.24928, "grad_norm": 1.275336787306216, "learning_rate": 0.003, "loss": 4.0564, "step": 24928 }, { "epoch": 0.24929, "grad_norm": 1.0147941597378278, "learning_rate": 0.003, "loss": 4.0606, "step": 24929 }, { "epoch": 0.2493, "grad_norm": 1.4789979646221407, "learning_rate": 0.003, "loss": 4.0262, "step": 24930 }, { "epoch": 0.24931, "grad_norm": 1.1891010042765089, "learning_rate": 0.003, "loss": 4.0722, "step": 24931 }, { "epoch": 0.24932, "grad_norm": 1.491651679136415, "learning_rate": 0.003, "loss": 4.0514, "step": 24932 }, { "epoch": 0.24933, "grad_norm": 1.265124354694425, "learning_rate": 0.003, "loss": 4.0612, "step": 24933 }, { "epoch": 0.24934, "grad_norm": 1.4084685637239533, "learning_rate": 0.003, "loss": 4.0656, "step": 24934 }, { "epoch": 0.24935, "grad_norm": 1.202236967976963, "learning_rate": 0.003, "loss": 4.0598, "step": 24935 }, { "epoch": 0.24936, "grad_norm": 1.2136212026763542, "learning_rate": 0.003, "loss": 4.0438, "step": 24936 }, { "epoch": 0.24937, "grad_norm": 1.559004736445429, "learning_rate": 0.003, "loss": 4.0571, "step": 24937 }, { "epoch": 0.24938, "grad_norm": 1.35275156518635, "learning_rate": 0.003, "loss": 4.0529, "step": 24938 }, { "epoch": 0.24939, "grad_norm": 1.5015669338305977, "learning_rate": 0.003, "loss": 4.0554, "step": 24939 }, { "epoch": 0.2494, "grad_norm": 1.275156545022829, "learning_rate": 0.003, "loss": 4.0497, "step": 24940 }, { "epoch": 0.24941, "grad_norm": 1.2182848171525187, "learning_rate": 0.003, "loss": 4.0356, "step": 24941 }, { "epoch": 0.24942, "grad_norm": 1.403902010995316, "learning_rate": 0.003, "loss": 4.0532, "step": 24942 }, { "epoch": 0.24943, "grad_norm": 1.2468410086238597, "learning_rate": 0.003, "loss": 4.0319, "step": 24943 }, { "epoch": 0.24944, "grad_norm": 1.2301364873403982, "learning_rate": 0.003, "loss": 4.019, "step": 24944 }, { "epoch": 0.24945, "grad_norm": 1.2033359033910367, "learning_rate": 0.003, "loss": 4.013, "step": 24945 }, { "epoch": 0.24946, "grad_norm": 1.4592038702804437, "learning_rate": 0.003, "loss": 4.0444, "step": 24946 }, { "epoch": 0.24947, "grad_norm": 1.1334231896485036, "learning_rate": 0.003, "loss": 4.0329, "step": 24947 }, { "epoch": 0.24948, "grad_norm": 1.4415079109978626, "learning_rate": 0.003, "loss": 4.0688, "step": 24948 }, { "epoch": 0.24949, "grad_norm": 1.17614349289862, "learning_rate": 0.003, "loss": 4.0143, "step": 24949 }, { "epoch": 0.2495, "grad_norm": 1.5593115471186048, "learning_rate": 0.003, "loss": 4.0692, "step": 24950 }, { "epoch": 0.24951, "grad_norm": 0.9283938653658199, "learning_rate": 0.003, "loss": 4.0428, "step": 24951 }, { "epoch": 0.24952, "grad_norm": 1.5458796456523693, "learning_rate": 0.003, "loss": 4.0307, "step": 24952 }, { "epoch": 0.24953, "grad_norm": 1.1037503801836566, "learning_rate": 0.003, "loss": 4.0273, "step": 24953 }, { "epoch": 0.24954, "grad_norm": 1.4507951213683754, "learning_rate": 0.003, "loss": 4.0333, "step": 24954 }, { "epoch": 0.24955, "grad_norm": 1.217547714662238, "learning_rate": 0.003, "loss": 4.06, "step": 24955 }, { "epoch": 0.24956, "grad_norm": 1.4072451690922527, "learning_rate": 0.003, "loss": 4.0538, "step": 24956 }, { "epoch": 0.24957, "grad_norm": 1.0918494147694893, "learning_rate": 0.003, "loss": 4.0649, "step": 24957 }, { "epoch": 0.24958, "grad_norm": 1.9321266883903285, "learning_rate": 0.003, "loss": 4.0669, "step": 24958 }, { "epoch": 0.24959, "grad_norm": 1.1152484527119353, "learning_rate": 0.003, "loss": 4.028, "step": 24959 }, { "epoch": 0.2496, "grad_norm": 1.3112365470042384, "learning_rate": 0.003, "loss": 4.0486, "step": 24960 }, { "epoch": 0.24961, "grad_norm": 1.3683398032590752, "learning_rate": 0.003, "loss": 4.0344, "step": 24961 }, { "epoch": 0.24962, "grad_norm": 1.3857230642207905, "learning_rate": 0.003, "loss": 4.0629, "step": 24962 }, { "epoch": 0.24963, "grad_norm": 1.2940292539319926, "learning_rate": 0.003, "loss": 4.0593, "step": 24963 }, { "epoch": 0.24964, "grad_norm": 1.1667678815727338, "learning_rate": 0.003, "loss": 4.0746, "step": 24964 }, { "epoch": 0.24965, "grad_norm": 1.3250197851198933, "learning_rate": 0.003, "loss": 4.062, "step": 24965 }, { "epoch": 0.24966, "grad_norm": 1.2424920919445346, "learning_rate": 0.003, "loss": 4.0703, "step": 24966 }, { "epoch": 0.24967, "grad_norm": 1.1263743998023235, "learning_rate": 0.003, "loss": 4.0711, "step": 24967 }, { "epoch": 0.24968, "grad_norm": 1.4848371770715436, "learning_rate": 0.003, "loss": 4.0387, "step": 24968 }, { "epoch": 0.24969, "grad_norm": 1.0306700634064905, "learning_rate": 0.003, "loss": 4.0461, "step": 24969 }, { "epoch": 0.2497, "grad_norm": 1.5267199933623845, "learning_rate": 0.003, "loss": 4.0301, "step": 24970 }, { "epoch": 0.24971, "grad_norm": 1.639320215013145, "learning_rate": 0.003, "loss": 4.0715, "step": 24971 }, { "epoch": 0.24972, "grad_norm": 1.1525468375552217, "learning_rate": 0.003, "loss": 4.0637, "step": 24972 }, { "epoch": 0.24973, "grad_norm": 1.7626246097754725, "learning_rate": 0.003, "loss": 4.0635, "step": 24973 }, { "epoch": 0.24974, "grad_norm": 1.3275200504636788, "learning_rate": 0.003, "loss": 4.0655, "step": 24974 }, { "epoch": 0.24975, "grad_norm": 1.3945586805684143, "learning_rate": 0.003, "loss": 4.043, "step": 24975 }, { "epoch": 0.24976, "grad_norm": 1.3400437810999533, "learning_rate": 0.003, "loss": 4.0694, "step": 24976 }, { "epoch": 0.24977, "grad_norm": 1.2333577154890487, "learning_rate": 0.003, "loss": 4.0517, "step": 24977 }, { "epoch": 0.24978, "grad_norm": 1.3397728435877232, "learning_rate": 0.003, "loss": 4.0395, "step": 24978 }, { "epoch": 0.24979, "grad_norm": 1.4886406015872247, "learning_rate": 0.003, "loss": 4.0762, "step": 24979 }, { "epoch": 0.2498, "grad_norm": 1.311555594354757, "learning_rate": 0.003, "loss": 4.0182, "step": 24980 }, { "epoch": 0.24981, "grad_norm": 1.4097119802725253, "learning_rate": 0.003, "loss": 4.0367, "step": 24981 }, { "epoch": 0.24982, "grad_norm": 1.257969179716512, "learning_rate": 0.003, "loss": 4.037, "step": 24982 }, { "epoch": 0.24983, "grad_norm": 1.528138306068266, "learning_rate": 0.003, "loss": 4.0239, "step": 24983 }, { "epoch": 0.24984, "grad_norm": 1.1447639613460119, "learning_rate": 0.003, "loss": 3.9934, "step": 24984 }, { "epoch": 0.24985, "grad_norm": 1.3336920987534133, "learning_rate": 0.003, "loss": 4.0609, "step": 24985 }, { "epoch": 0.24986, "grad_norm": 1.3190428493878157, "learning_rate": 0.003, "loss": 4.0399, "step": 24986 }, { "epoch": 0.24987, "grad_norm": 1.3301088479267904, "learning_rate": 0.003, "loss": 4.0463, "step": 24987 }, { "epoch": 0.24988, "grad_norm": 1.1570015010963344, "learning_rate": 0.003, "loss": 4.0594, "step": 24988 }, { "epoch": 0.24989, "grad_norm": 1.6213691303224982, "learning_rate": 0.003, "loss": 4.0505, "step": 24989 }, { "epoch": 0.2499, "grad_norm": 1.255825660263306, "learning_rate": 0.003, "loss": 4.0347, "step": 24990 }, { "epoch": 0.24991, "grad_norm": 1.588967416487042, "learning_rate": 0.003, "loss": 4.0475, "step": 24991 }, { "epoch": 0.24992, "grad_norm": 1.054545737037509, "learning_rate": 0.003, "loss": 4.0698, "step": 24992 }, { "epoch": 0.24993, "grad_norm": 1.433433725581734, "learning_rate": 0.003, "loss": 4.0524, "step": 24993 }, { "epoch": 0.24994, "grad_norm": 1.205648888491351, "learning_rate": 0.003, "loss": 4.0353, "step": 24994 }, { "epoch": 0.24995, "grad_norm": 1.4135515530513658, "learning_rate": 0.003, "loss": 4.0595, "step": 24995 }, { "epoch": 0.24996, "grad_norm": 1.3707149443458984, "learning_rate": 0.003, "loss": 4.0485, "step": 24996 }, { "epoch": 0.24997, "grad_norm": 1.3957772683443745, "learning_rate": 0.003, "loss": 4.0334, "step": 24997 }, { "epoch": 0.24998, "grad_norm": 1.4058385202482575, "learning_rate": 0.003, "loss": 4.0643, "step": 24998 }, { "epoch": 0.24999, "grad_norm": 1.1533823492711095, "learning_rate": 0.003, "loss": 4.0609, "step": 24999 }, { "epoch": 0.25, "grad_norm": 1.5495313955319767, "learning_rate": 0.003, "loss": 4.06, "step": 25000 } ], "logging_steps": 1, "max_steps": 100000, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 9.910910713856e+17, "train_batch_size": 1024, "trial_name": null, "trial_params": null }