| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 892, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 9.2246, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 9.1373, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 8.8311, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 9.2459, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 9.182, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 8.8163, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-07, | |
| "loss": 9.091, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 8.9724, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 8.9179, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-07, | |
| "loss": 9.1465, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2e-07, | |
| "loss": 8.807, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 8.769, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4e-07, | |
| "loss": 8.6194, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 8.9798, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.599999999999999e-07, | |
| "loss": 8.5752, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.2e-07, | |
| "loss": 8.2835, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.799999999999999e-07, | |
| "loss": 8.5145, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.4e-07, | |
| "loss": 8.1527, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-07, | |
| "loss": 8.2896, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-07, | |
| "loss": 8.4004, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 8.184, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0200000000000002e-06, | |
| "loss": 8.1589, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.08e-06, | |
| "loss": 8.0137, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.14e-06, | |
| "loss": 8.0527, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 8.1472, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.26e-06, | |
| "loss": 8.3813, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3199999999999999e-06, | |
| "loss": 8.0726, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.38e-06, | |
| "loss": 7.9304, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.44e-06, | |
| "loss": 7.8539, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5e-06, | |
| "loss": 7.997, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5599999999999999e-06, | |
| "loss": 7.9305, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.62e-06, | |
| "loss": 7.8678, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.68e-06, | |
| "loss": 7.8208, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.74e-06, | |
| "loss": 7.5448, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.8e-06, | |
| "loss": 7.707, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.86e-06, | |
| "loss": 7.6488, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 7.6675, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.98e-06, | |
| "loss": 7.6531, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.0400000000000004e-06, | |
| "loss": 7.8233, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 7.5342, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.16e-06, | |
| "loss": 7.4604, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.22e-06, | |
| "loss": 7.7178, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.28e-06, | |
| "loss": 7.4701, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.34e-06, | |
| "loss": 7.7004, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 7.7114, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.46e-06, | |
| "loss": 7.7806, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.52e-06, | |
| "loss": 7.5586, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.58e-06, | |
| "loss": 7.6923, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6399999999999997e-06, | |
| "loss": 7.7666, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.7e-06, | |
| "loss": 7.8198, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.76e-06, | |
| "loss": 7.3712, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.82e-06, | |
| "loss": 7.3484, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.88e-06, | |
| "loss": 7.2667, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 7.4204, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3e-06, | |
| "loss": 7.2713, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.06e-06, | |
| "loss": 7.3576, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.1199999999999998e-06, | |
| "loss": 7.3774, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.18e-06, | |
| "loss": 7.4335, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.24e-06, | |
| "loss": 7.2461, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.3e-06, | |
| "loss": 7.3461, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.36e-06, | |
| "loss": 7.3595, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.4200000000000003e-06, | |
| "loss": 7.2852, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.48e-06, | |
| "loss": 7.2491, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.54e-06, | |
| "loss": 7.3441, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.6e-06, | |
| "loss": 7.1761, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.66e-06, | |
| "loss": 7.4232, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.72e-06, | |
| "loss": 7.2364, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 7.1723, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 7.2477, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.9e-06, | |
| "loss": 7.1895, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.96e-06, | |
| "loss": 7.3421, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.0200000000000005e-06, | |
| "loss": 7.3882, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.080000000000001e-06, | |
| "loss": 7.1381, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.14e-06, | |
| "loss": 7.3217, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 7.1776, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.26e-06, | |
| "loss": 7.2761, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.32e-06, | |
| "loss": 7.167, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.3799999999999996e-06, | |
| "loss": 7.1506, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.44e-06, | |
| "loss": 7.1386, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5e-06, | |
| "loss": 7.1926, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.56e-06, | |
| "loss": 7.2166, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.62e-06, | |
| "loss": 7.1547, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.68e-06, | |
| "loss": 7.322, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.74e-06, | |
| "loss": 7.094, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 7.1213, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.86e-06, | |
| "loss": 7.2003, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.92e-06, | |
| "loss": 7.1351, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 7.1462, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.04e-06, | |
| "loss": 7.1968, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.1e-06, | |
| "loss": 7.2494, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.16e-06, | |
| "loss": 7.3753, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.22e-06, | |
| "loss": 7.2578, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.279999999999999e-06, | |
| "loss": 7.0413, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.34e-06, | |
| "loss": 7.1463, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.4e-06, | |
| "loss": 7.2888, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.46e-06, | |
| "loss": 7.3631, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.52e-06, | |
| "loss": 7.1825, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.58e-06, | |
| "loss": 7.2189, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.64e-06, | |
| "loss": 6.9729, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.7000000000000005e-06, | |
| "loss": 6.7578, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.76e-06, | |
| "loss": 7.2918, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5.82e-06, | |
| "loss": 7.3083, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 7.2555, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.940000000000001e-06, | |
| "loss": 7.1407, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6e-06, | |
| "loss": 7.0634, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 7.0999, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.12e-06, | |
| "loss": 7.2116, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.18e-06, | |
| "loss": 7.2409, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.2399999999999995e-06, | |
| "loss": 6.9523, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.3e-06, | |
| "loss": 7.0838, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.36e-06, | |
| "loss": 7.1254, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.42e-06, | |
| "loss": 7.2566, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.48e-06, | |
| "loss": 7.1117, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.54e-06, | |
| "loss": 7.0778, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.6e-06, | |
| "loss": 7.0946, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 6.9887, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.72e-06, | |
| "loss": 7.1303, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.78e-06, | |
| "loss": 7.0789, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.840000000000001e-06, | |
| "loss": 7.0958, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 6.900000000000001e-06, | |
| "loss": 7.1212, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 6.96e-06, | |
| "loss": 7.033, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.0200000000000006e-06, | |
| "loss": 7.1919, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.08e-06, | |
| "loss": 6.9572, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.14e-06, | |
| "loss": 7.0968, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.2e-06, | |
| "loss": 7.1623, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.26e-06, | |
| "loss": 7.235, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.32e-06, | |
| "loss": 7.2622, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 7.1798, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.44e-06, | |
| "loss": 7.0186, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.5e-06, | |
| "loss": 7.1465, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 7.0823, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.62e-06, | |
| "loss": 7.1764, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 7.1002, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.74e-06, | |
| "loss": 7.2143, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.8e-06, | |
| "loss": 6.9495, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 7.0104, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.92e-06, | |
| "loss": 7.2685, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.98e-06, | |
| "loss": 7.1034, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.040000000000001e-06, | |
| "loss": 7.0242, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.1e-06, | |
| "loss": 7.0288, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.160000000000001e-06, | |
| "loss": 7.2486, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 7.2074, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.28e-06, | |
| "loss": 7.1155, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.340000000000001e-06, | |
| "loss": 6.9432, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 6.9311, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.459999999999999e-06, | |
| "loss": 6.8801, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.52e-06, | |
| "loss": 7.1537, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.58e-06, | |
| "loss": 6.6065, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.64e-06, | |
| "loss": 6.9624, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.7e-06, | |
| "loss": 6.4166, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.759999999999999e-06, | |
| "loss": 7.2044, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.82e-06, | |
| "loss": 7.1448, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.88e-06, | |
| "loss": 7.1486, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.939999999999999e-06, | |
| "loss": 6.9927, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9e-06, | |
| "loss": 7.0003, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.06e-06, | |
| "loss": 7.0689, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.12e-06, | |
| "loss": 7.2571, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.18e-06, | |
| "loss": 7.2219, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.24e-06, | |
| "loss": 7.1603, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.3e-06, | |
| "loss": 7.1569, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.36e-06, | |
| "loss": 7.0327, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.42e-06, | |
| "loss": 7.0264, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.48e-06, | |
| "loss": 7.0862, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.54e-06, | |
| "loss": 7.0314, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 7.0722, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.66e-06, | |
| "loss": 7.0334, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.72e-06, | |
| "loss": 6.9734, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 7.0998, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.84e-06, | |
| "loss": 7.1167, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.9e-06, | |
| "loss": 7.0092, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 7.202, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.002e-05, | |
| "loss": 7.1753, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.008e-05, | |
| "loss": 7.19, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.0140000000000001e-05, | |
| "loss": 7.0812, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.02e-05, | |
| "loss": 6.9601, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.0260000000000002e-05, | |
| "loss": 7.0665, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.032e-05, | |
| "loss": 7.0419, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.0379999999999999e-05, | |
| "loss": 6.9903, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.044e-05, | |
| "loss": 7.1019, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.05e-05, | |
| "loss": 7.1396, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.0559999999999999e-05, | |
| "loss": 7.0924, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.062e-05, | |
| "loss": 6.9487, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.068e-05, | |
| "loss": 7.0349, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.074e-05, | |
| "loss": 7.1434, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.08e-05, | |
| "loss": 7.0692, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.086e-05, | |
| "loss": 7.1847, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.092e-05, | |
| "loss": 7.1237, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.098e-05, | |
| "loss": 6.9167, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.104e-05, | |
| "loss": 6.9805, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.11e-05, | |
| "loss": 7.0209, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.116e-05, | |
| "loss": 7.0294, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1220000000000001e-05, | |
| "loss": 7.1835, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.128e-05, | |
| "loss": 7.1852, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.134e-05, | |
| "loss": 7.0206, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 6.768, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.146e-05, | |
| "loss": 7.1724, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.152e-05, | |
| "loss": 6.9272, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1580000000000001e-05, | |
| "loss": 6.8322, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.164e-05, | |
| "loss": 6.7014, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1700000000000001e-05, | |
| "loss": 6.4995, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.1760000000000001e-05, | |
| "loss": 7.1439, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.182e-05, | |
| "loss": 7.1787, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.1880000000000001e-05, | |
| "loss": 7.051, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.1940000000000001e-05, | |
| "loss": 7.0691, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.2e-05, | |
| "loss": 7.0265, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.2060000000000001e-05, | |
| "loss": 7.01, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.2120000000000001e-05, | |
| "loss": 7.0522, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.2180000000000002e-05, | |
| "loss": 6.9969, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.224e-05, | |
| "loss": 6.9861, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.2299999999999999e-05, | |
| "loss": 6.9958, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.236e-05, | |
| "loss": 6.8662, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.242e-05, | |
| "loss": 6.9427, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.2479999999999999e-05, | |
| "loss": 7.0203, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.254e-05, | |
| "loss": 7.0491, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.26e-05, | |
| "loss": 7.0124, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.2659999999999999e-05, | |
| "loss": 7.1293, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.272e-05, | |
| "loss": 6.9887, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.278e-05, | |
| "loss": 7.0182, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.284e-05, | |
| "loss": 6.9398, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.29e-05, | |
| "loss": 7.084, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.296e-05, | |
| "loss": 7.0237, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.302e-05, | |
| "loss": 6.9158, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.308e-05, | |
| "loss": 7.0453, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.314e-05, | |
| "loss": 6.9951, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.32e-05, | |
| "loss": 6.9467, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.326e-05, | |
| "loss": 7.1874, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.3320000000000001e-05, | |
| "loss": 6.9712, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.338e-05, | |
| "loss": 6.9711, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.344e-05, | |
| "loss": 7.0803, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 7.0759, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.356e-05, | |
| "loss": 6.9329, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.362e-05, | |
| "loss": 7.1158, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.3680000000000001e-05, | |
| "loss": 7.0195, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.374e-05, | |
| "loss": 7.026, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 7.0963, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.3860000000000001e-05, | |
| "loss": 6.8661, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.392e-05, | |
| "loss": 6.9358, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.3980000000000002e-05, | |
| "loss": 7.1396, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.4040000000000001e-05, | |
| "loss": 6.9051, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.4099999999999999e-05, | |
| "loss": 6.965, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.416e-05, | |
| "loss": 6.9123, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.422e-05, | |
| "loss": 7.0129, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.428e-05, | |
| "loss": 6.884, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.434e-05, | |
| "loss": 6.798, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.44e-05, | |
| "loss": 6.927, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.446e-05, | |
| "loss": 6.906, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.452e-05, | |
| "loss": 6.9235, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.458e-05, | |
| "loss": 7.0107, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.464e-05, | |
| "loss": 6.7736, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.47e-05, | |
| "loss": 6.5131, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.4760000000000001e-05, | |
| "loss": 7.2408, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.482e-05, | |
| "loss": 7.1705, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.488e-05, | |
| "loss": 7.0152, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.4940000000000001e-05, | |
| "loss": 7.0892, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.5e-05, | |
| "loss": 6.9197, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.506e-05, | |
| "loss": 7.091, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.5120000000000001e-05, | |
| "loss": 7.0727, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.518e-05, | |
| "loss": 6.9958, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.524e-05, | |
| "loss": 6.9301, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.53e-05, | |
| "loss": 7.0663, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.5360000000000002e-05, | |
| "loss": 6.9489, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.542e-05, | |
| "loss": 6.8822, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.548e-05, | |
| "loss": 7.0065, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.554e-05, | |
| "loss": 6.9719, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.56e-05, | |
| "loss": 6.9191, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5660000000000003e-05, | |
| "loss": 7.0878, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5720000000000002e-05, | |
| "loss": 7.0698, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.578e-05, | |
| "loss": 7.0582, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.584e-05, | |
| "loss": 6.9094, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.59e-05, | |
| "loss": 7.0062, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.596e-05, | |
| "loss": 6.8624, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6020000000000002e-05, | |
| "loss": 6.8755, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6080000000000002e-05, | |
| "loss": 6.923, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.614e-05, | |
| "loss": 6.7935, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.62e-05, | |
| "loss": 7.053, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.626e-05, | |
| "loss": 6.9004, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6320000000000003e-05, | |
| "loss": 7.0005, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6380000000000002e-05, | |
| "loss": 6.9169, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6440000000000002e-05, | |
| "loss": 6.9615, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.65e-05, | |
| "loss": 7.0077, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.656e-05, | |
| "loss": 6.8053, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6620000000000004e-05, | |
| "loss": 7.0004, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6680000000000003e-05, | |
| "loss": 6.7168, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6740000000000002e-05, | |
| "loss": 7.0215, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 6.9857, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.686e-05, | |
| "loss": 6.7984, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6919999999999997e-05, | |
| "loss": 7.0215, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.698e-05, | |
| "loss": 6.942, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.704e-05, | |
| "loss": 7.0736, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.71e-05, | |
| "loss": 7.1623, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.716e-05, | |
| "loss": 7.0506, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.7219999999999998e-05, | |
| "loss": 7.0028, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.728e-05, | |
| "loss": 6.8279, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.734e-05, | |
| "loss": 6.8636, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.74e-05, | |
| "loss": 6.7131, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.746e-05, | |
| "loss": 6.7668, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.7519999999999998e-05, | |
| "loss": 6.7466, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.758e-05, | |
| "loss": 6.557, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.764e-05, | |
| "loss": 6.5136, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.77e-05, | |
| "loss": 6.3274, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.776e-05, | |
| "loss": 7.0971, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.782e-05, | |
| "loss": 7.1435, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7879999999999998e-05, | |
| "loss": 6.9586, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.794e-05, | |
| "loss": 7.0416, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.8e-05, | |
| "loss": 6.8186, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.806e-05, | |
| "loss": 6.8905, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.812e-05, | |
| "loss": 6.9319, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.818e-05, | |
| "loss": 6.9557, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.824e-05, | |
| "loss": 6.8847, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.83e-05, | |
| "loss": 7.0108, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.836e-05, | |
| "loss": 7.0779, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.842e-05, | |
| "loss": 6.9719, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.848e-05, | |
| "loss": 6.9861, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.854e-05, | |
| "loss": 6.9209, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.86e-05, | |
| "loss": 6.8109, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.866e-05, | |
| "loss": 7.0507, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.872e-05, | |
| "loss": 6.9378, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.878e-05, | |
| "loss": 6.8479, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.884e-05, | |
| "loss": 6.7743, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 7.0167, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.896e-05, | |
| "loss": 6.9469, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.902e-05, | |
| "loss": 6.8474, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.908e-05, | |
| "loss": 6.9563, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.914e-05, | |
| "loss": 6.9896, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 6.9997, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9260000000000002e-05, | |
| "loss": 6.943, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.932e-05, | |
| "loss": 6.8517, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.938e-05, | |
| "loss": 6.9522, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.944e-05, | |
| "loss": 7.0001, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.95e-05, | |
| "loss": 6.9991, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9560000000000002e-05, | |
| "loss": 7.0175, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9620000000000002e-05, | |
| "loss": 6.739, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.968e-05, | |
| "loss": 6.873, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.974e-05, | |
| "loss": 6.9507, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.98e-05, | |
| "loss": 7.0429, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9860000000000003e-05, | |
| "loss": 6.8541, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9920000000000002e-05, | |
| "loss": 6.8273, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9980000000000002e-05, | |
| "loss": 6.9733, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.004e-05, | |
| "loss": 6.9591, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.01e-05, | |
| "loss": 6.9074, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.016e-05, | |
| "loss": 7.0047, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.0220000000000003e-05, | |
| "loss": 6.8685, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.0280000000000002e-05, | |
| "loss": 6.8809, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.0340000000000002e-05, | |
| "loss": 6.8344, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.04e-05, | |
| "loss": 6.7439, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.046e-05, | |
| "loss": 6.8392, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.0520000000000003e-05, | |
| "loss": 6.7882, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.0580000000000003e-05, | |
| "loss": 6.8275, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.064e-05, | |
| "loss": 6.7192, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.07e-05, | |
| "loss": 6.5946, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.0759999999999998e-05, | |
| "loss": 6.9426, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.082e-05, | |
| "loss": 6.9544, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.088e-05, | |
| "loss": 6.9189, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.094e-05, | |
| "loss": 6.8795, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.1e-05, | |
| "loss": 7.0007, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.1059999999999998e-05, | |
| "loss": 6.8779, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.1119999999999998e-05, | |
| "loss": 7.1244, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.118e-05, | |
| "loss": 6.8622, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.124e-05, | |
| "loss": 6.9796, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.13e-05, | |
| "loss": 6.7798, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.136e-05, | |
| "loss": 6.8914, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.1419999999999998e-05, | |
| "loss": 6.9833, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.148e-05, | |
| "loss": 6.8806, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.154e-05, | |
| "loss": 6.9408, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.16e-05, | |
| "loss": 6.8998, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.166e-05, | |
| "loss": 6.8415, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.172e-05, | |
| "loss": 6.8909, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.178e-05, | |
| "loss": 6.8517, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.184e-05, | |
| "loss": 6.7895, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.19e-05, | |
| "loss": 6.9884, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.196e-05, | |
| "loss": 6.8805, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.202e-05, | |
| "loss": 7.0732, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.208e-05, | |
| "loss": 6.9139, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.214e-05, | |
| "loss": 6.9828, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.22e-05, | |
| "loss": 6.7987, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.226e-05, | |
| "loss": 6.914, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.232e-05, | |
| "loss": 6.7596, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.238e-05, | |
| "loss": 6.8845, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.2440000000000002e-05, | |
| "loss": 6.8886, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.25e-05, | |
| "loss": 6.9432, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.256e-05, | |
| "loss": 6.969, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.262e-05, | |
| "loss": 6.8245, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.268e-05, | |
| "loss": 6.9876, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.274e-05, | |
| "loss": 6.9816, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 6.8357, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.286e-05, | |
| "loss": 6.8805, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.292e-05, | |
| "loss": 6.8696, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.298e-05, | |
| "loss": 6.8541, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.304e-05, | |
| "loss": 6.8526, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.3100000000000002e-05, | |
| "loss": 7.0966, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.3160000000000002e-05, | |
| "loss": 6.8956, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.322e-05, | |
| "loss": 6.8709, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.328e-05, | |
| "loss": 6.8491, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.334e-05, | |
| "loss": 7.0112, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 6.8404, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.3460000000000002e-05, | |
| "loss": 6.6512, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3520000000000002e-05, | |
| "loss": 6.5857, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.358e-05, | |
| "loss": 6.6665, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.364e-05, | |
| "loss": 6.6296, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.37e-05, | |
| "loss": 6.4654, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3760000000000003e-05, | |
| "loss": 7.1123, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3820000000000002e-05, | |
| "loss": 7.0724, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3880000000000002e-05, | |
| "loss": 6.9767, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.394e-05, | |
| "loss": 6.899, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.4e-05, | |
| "loss": 7.0785, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.4060000000000003e-05, | |
| "loss": 6.909, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.4120000000000003e-05, | |
| "loss": 6.9077, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.4180000000000002e-05, | |
| "loss": 6.8431, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.4240000000000002e-05, | |
| "loss": 7.0443, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.43e-05, | |
| "loss": 6.9833, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.4360000000000004e-05, | |
| "loss": 6.9221, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.442e-05, | |
| "loss": 6.9201, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.448e-05, | |
| "loss": 6.9065, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.454e-05, | |
| "loss": 6.7746, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 6.8324, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.4659999999999998e-05, | |
| "loss": 6.934, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.472e-05, | |
| "loss": 6.8337, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.478e-05, | |
| "loss": 6.9123, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.484e-05, | |
| "loss": 6.8812, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.49e-05, | |
| "loss": 6.9068, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.4959999999999998e-05, | |
| "loss": 6.9287, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.502e-05, | |
| "loss": 6.9944, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.508e-05, | |
| "loss": 6.9173, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.514e-05, | |
| "loss": 6.9294, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.52e-05, | |
| "loss": 6.9498, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.526e-05, | |
| "loss": 7.0747, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5319999999999998e-05, | |
| "loss": 6.8528, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.538e-05, | |
| "loss": 7.0238, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.544e-05, | |
| "loss": 6.8598, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.55e-05, | |
| "loss": 6.8975, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.556e-05, | |
| "loss": 7.0235, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.562e-05, | |
| "loss": 6.9912, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.568e-05, | |
| "loss": 6.6638, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.574e-05, | |
| "loss": 6.8904, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.58e-05, | |
| "loss": 6.8181, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.586e-05, | |
| "loss": 7.0223, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.592e-05, | |
| "loss": 6.9219, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5980000000000002e-05, | |
| "loss": 6.9923, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.604e-05, | |
| "loss": 6.7617, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.61e-05, | |
| "loss": 6.9326, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.616e-05, | |
| "loss": 6.9804, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.622e-05, | |
| "loss": 6.8346, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.628e-05, | |
| "loss": 6.8193, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.6340000000000002e-05, | |
| "loss": 6.9711, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.64e-05, | |
| "loss": 6.7013, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.646e-05, | |
| "loss": 6.7377, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.652e-05, | |
| "loss": 6.7358, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.658e-05, | |
| "loss": 6.7159, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.6640000000000002e-05, | |
| "loss": 6.4348, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.6700000000000002e-05, | |
| "loss": 6.4836, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.676e-05, | |
| "loss": 6.9618, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.682e-05, | |
| "loss": 7.0763, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.688e-05, | |
| "loss": 7.1267, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.6940000000000003e-05, | |
| "loss": 6.9226, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 7.0096, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.7060000000000002e-05, | |
| "loss": 7.0543, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.712e-05, | |
| "loss": 6.8422, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.718e-05, | |
| "loss": 6.9216, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.724e-05, | |
| "loss": 6.923, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7300000000000003e-05, | |
| "loss": 7.0043, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7360000000000002e-05, | |
| "loss": 6.8199, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7420000000000002e-05, | |
| "loss": 6.9491, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.748e-05, | |
| "loss": 6.8379, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.754e-05, | |
| "loss": 6.8074, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7600000000000003e-05, | |
| "loss": 6.9184, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7660000000000003e-05, | |
| "loss": 6.9016, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7720000000000002e-05, | |
| "loss": 6.8365, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.778e-05, | |
| "loss": 6.9663, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.784e-05, | |
| "loss": 6.8732, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.79e-05, | |
| "loss": 6.955, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.7960000000000003e-05, | |
| "loss": 6.8268, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.8020000000000003e-05, | |
| "loss": 6.8658, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.8080000000000002e-05, | |
| "loss": 6.8687, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.8139999999999998e-05, | |
| "loss": 6.7876, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 6.9435, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.826e-05, | |
| "loss": 6.9562, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.832e-05, | |
| "loss": 7.0299, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.838e-05, | |
| "loss": 6.8044, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.844e-05, | |
| "loss": 6.8591, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 6.8271, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.856e-05, | |
| "loss": 6.8259, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.862e-05, | |
| "loss": 6.8352, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.868e-05, | |
| "loss": 6.8316, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.874e-05, | |
| "loss": 6.9393, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.88e-05, | |
| "loss": 6.9123, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.8859999999999998e-05, | |
| "loss": 6.8544, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.892e-05, | |
| "loss": 6.9356, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.898e-05, | |
| "loss": 6.8117, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.904e-05, | |
| "loss": 7.0152, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.91e-05, | |
| "loss": 7.0389, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.916e-05, | |
| "loss": 6.8311, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.922e-05, | |
| "loss": 6.8967, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.928e-05, | |
| "loss": 6.8858, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.934e-05, | |
| "loss": 6.832, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.94e-05, | |
| "loss": 6.921, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.946e-05, | |
| "loss": 6.6614, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.9520000000000002e-05, | |
| "loss": 6.3714, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.958e-05, | |
| "loss": 6.9957, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.964e-05, | |
| "loss": 6.7645, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.97e-05, | |
| "loss": 6.3306, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "eval_loss": 6.914114475250244, | |
| "eval_runtime": 952.541, | |
| "eval_samples_per_second": 2.774, | |
| "eval_steps_per_second": 0.347, | |
| "eval_wer": 2.0110324027223845, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.976e-05, | |
| "loss": 7.2242, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.982e-05, | |
| "loss": 6.8443, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.9880000000000002e-05, | |
| "loss": 7.0131, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.994e-05, | |
| "loss": 7.0698, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3e-05, | |
| "loss": 7.0658, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.9923469387755103e-05, | |
| "loss": 7.0079, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.9846938775510208e-05, | |
| "loss": 6.9869, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.9770408163265306e-05, | |
| "loss": 6.8591, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.969387755102041e-05, | |
| "loss": 6.9314, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.961734693877551e-05, | |
| "loss": 6.9464, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.9540816326530616e-05, | |
| "loss": 6.8774, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.9464285714285714e-05, | |
| "loss": 6.9161, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9387755102040816e-05, | |
| "loss": 7.0163, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9311224489795918e-05, | |
| "loss": 6.9863, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9234693877551023e-05, | |
| "loss": 6.9584, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9158163265306125e-05, | |
| "loss": 6.7738, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9081632653061224e-05, | |
| "loss": 7.0052, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.9005102040816326e-05, | |
| "loss": 6.8515, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.892857142857143e-05, | |
| "loss": 7.0348, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.8852040816326533e-05, | |
| "loss": 7.0086, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.877551020408163e-05, | |
| "loss": 6.8237, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8698979591836733e-05, | |
| "loss": 6.8245, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.862244897959184e-05, | |
| "loss": 6.7518, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.854591836734694e-05, | |
| "loss": 6.8415, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8469387755102042e-05, | |
| "loss": 6.9844, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8392857142857144e-05, | |
| "loss": 6.7874, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8316326530612246e-05, | |
| "loss": 6.9059, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8239795918367348e-05, | |
| "loss": 6.938, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.816326530612245e-05, | |
| "loss": 6.9756, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.8086734693877552e-05, | |
| "loss": 7.0175, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.8010204081632654e-05, | |
| "loss": 6.8836, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.7933673469387756e-05, | |
| "loss": 6.8041, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.7857142857142858e-05, | |
| "loss": 6.8195, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.778061224489796e-05, | |
| "loss": 6.8403, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.770408163265306e-05, | |
| "loss": 6.8532, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.7627551020408163e-05, | |
| "loss": 7.0077, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.7551020408163265e-05, | |
| "loss": 6.8801, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.747448979591837e-05, | |
| "loss": 6.996, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.739795918367347e-05, | |
| "loss": 6.9724, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.732142857142857e-05, | |
| "loss": 6.9039, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7244897959183673e-05, | |
| "loss": 7.0277, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7168367346938778e-05, | |
| "loss": 6.8933, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7091836734693877e-05, | |
| "loss": 6.7796, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.701530612244898e-05, | |
| "loss": 6.6972, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6938775510204084e-05, | |
| "loss": 6.7981, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6862244897959186e-05, | |
| "loss": 6.7376, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6785714285714288e-05, | |
| "loss": 6.9709, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.6709183673469386e-05, | |
| "loss": 6.7878, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.663265306122449e-05, | |
| "loss": 6.3952, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6556122448979593e-05, | |
| "loss": 6.6438, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6479591836734695e-05, | |
| "loss": 6.9375, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6403061224489794e-05, | |
| "loss": 7.1262, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.63265306122449e-05, | |
| "loss": 6.9578, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.625e-05, | |
| "loss": 6.951, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6173469387755103e-05, | |
| "loss": 6.9577, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6096938775510205e-05, | |
| "loss": 6.9202, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6020408163265307e-05, | |
| "loss": 6.8923, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.594387755102041e-05, | |
| "loss": 6.9608, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.586734693877551e-05, | |
| "loss": 6.8879, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5790816326530613e-05, | |
| "loss": 6.8141, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5714285714285714e-05, | |
| "loss": 6.9607, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5637755102040816e-05, | |
| "loss": 6.7892, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5561224489795918e-05, | |
| "loss": 6.8645, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.548469387755102e-05, | |
| "loss": 6.8667, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5408163265306125e-05, | |
| "loss": 7.0323, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5331632653061224e-05, | |
| "loss": 6.7248, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.5255102040816326e-05, | |
| "loss": 6.9148, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.517857142857143e-05, | |
| "loss": 6.7584, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.5102040816326533e-05, | |
| "loss": 6.8439, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.502551020408163e-05, | |
| "loss": 6.9657, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.4948979591836734e-05, | |
| "loss": 6.8511, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.487244897959184e-05, | |
| "loss": 6.9668, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.479591836734694e-05, | |
| "loss": 6.823, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.4719387755102043e-05, | |
| "loss": 6.7057, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.464285714285714e-05, | |
| "loss": 6.9239, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.4566326530612246e-05, | |
| "loss": 6.7678, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.448979591836735e-05, | |
| "loss": 6.9136, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.441326530612245e-05, | |
| "loss": 6.961, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.433673469387755e-05, | |
| "loss": 6.8128, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.4260204081632654e-05, | |
| "loss": 6.9925, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.4183673469387756e-05, | |
| "loss": 7.0088, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.4107142857142858e-05, | |
| "loss": 6.8783, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.403061224489796e-05, | |
| "loss": 6.7756, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.3954081632653062e-05, | |
| "loss": 6.9015, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3877551020408164e-05, | |
| "loss": 6.8994, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3801020408163266e-05, | |
| "loss": 6.9457, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.372448979591837e-05, | |
| "loss": 6.9353, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.364795918367347e-05, | |
| "loss": 6.8825, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.357142857142857e-05, | |
| "loss": 6.783, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3494897959183673e-05, | |
| "loss": 6.8247, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.341836734693878e-05, | |
| "loss": 6.8881, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.3341836734693877e-05, | |
| "loss": 6.907, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.326530612244898e-05, | |
| "loss": 6.9815, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.318877551020408e-05, | |
| "loss": 6.6994, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3112244897959186e-05, | |
| "loss": 6.7467, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.3035714285714288e-05, | |
| "loss": 6.6612, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.2959183673469387e-05, | |
| "loss": 6.6789, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.288265306122449e-05, | |
| "loss": 6.5702, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.2806122448979594e-05, | |
| "loss": 6.4857, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.2729591836734696e-05, | |
| "loss": 6.2174, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.2653061224489794e-05, | |
| "loss": 6.9971, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.2576530612244896e-05, | |
| "loss": 6.8156, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.25e-05, | |
| "loss": 6.9443, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.2423469387755103e-05, | |
| "loss": 6.9254, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.2346938775510205e-05, | |
| "loss": 7.0202, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.2270408163265307e-05, | |
| "loss": 6.719, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.219387755102041e-05, | |
| "loss": 6.8925, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.211734693877551e-05, | |
| "loss": 6.9246, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.2040816326530613e-05, | |
| "loss": 6.8959, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.1964285714285715e-05, | |
| "loss": 6.8851, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.1887755102040817e-05, | |
| "loss": 6.8646, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.181122448979592e-05, | |
| "loss": 6.9031, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.173469387755102e-05, | |
| "loss": 6.8019, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1658163265306126e-05, | |
| "loss": 6.8787, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1581632653061224e-05, | |
| "loss": 7.0718, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1505102040816326e-05, | |
| "loss": 6.7524, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1428571428571428e-05, | |
| "loss": 6.8568, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1352040816326533e-05, | |
| "loss": 6.712, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1275510204081632e-05, | |
| "loss": 6.8248, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1198979591836734e-05, | |
| "loss": 6.8799, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1122448979591836e-05, | |
| "loss": 6.918, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.104591836734694e-05, | |
| "loss": 6.8227, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.0969387755102043e-05, | |
| "loss": 6.946, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.089285714285714e-05, | |
| "loss": 6.7692, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.0816326530612247e-05, | |
| "loss": 6.9808, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.073979591836735e-05, | |
| "loss": 6.8455, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.066326530612245e-05, | |
| "loss": 6.9412, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.058673469387755e-05, | |
| "loss": 6.9302, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0510204081632654e-05, | |
| "loss": 6.896, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0433673469387756e-05, | |
| "loss": 7.0522, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0357142857142858e-05, | |
| "loss": 6.8552, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0280612244897957e-05, | |
| "loss": 6.7395, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0204081632653062e-05, | |
| "loss": 6.9425, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0127551020408164e-05, | |
| "loss": 6.8416, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0051020408163266e-05, | |
| "loss": 6.956, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.9974489795918368e-05, | |
| "loss": 6.9566, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.989795918367347e-05, | |
| "loss": 6.9165, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.982142857142857e-05, | |
| "loss": 6.7259, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9744897959183674e-05, | |
| "loss": 7.0196, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9668367346938775e-05, | |
| "loss": 6.7428, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9591836734693877e-05, | |
| "loss": 6.8471, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.951530612244898e-05, | |
| "loss": 7.0086, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.943877551020408e-05, | |
| "loss": 6.897, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9362244897959186e-05, | |
| "loss": 6.6539, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.928571428571429e-05, | |
| "loss": 6.7148, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9209183673469387e-05, | |
| "loss": 6.8433, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.913265306122449e-05, | |
| "loss": 6.5996, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.9056122448979594e-05, | |
| "loss": 6.5503, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8979591836734696e-05, | |
| "loss": 6.7521, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8903061224489795e-05, | |
| "loss": 6.4819, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8826530612244896e-05, | |
| "loss": 7.0465, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 7.0229, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8673469387755104e-05, | |
| "loss": 7.0164, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8596938775510206e-05, | |
| "loss": 6.9124, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8520408163265304e-05, | |
| "loss": 6.9706, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.844387755102041e-05, | |
| "loss": 6.9488, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.836734693877551e-05, | |
| "loss": 6.7632, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8290816326530613e-05, | |
| "loss": 7.0554, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8214285714285712e-05, | |
| "loss": 6.8021, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8137755102040817e-05, | |
| "loss": 6.796, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.806122448979592e-05, | |
| "loss": 6.8337, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.798469387755102e-05, | |
| "loss": 6.8948, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7908163265306126e-05, | |
| "loss": 6.9559, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7831632653061225e-05, | |
| "loss": 6.7223, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7755102040816327e-05, | |
| "loss": 6.8449, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.767857142857143e-05, | |
| "loss": 6.9404, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7602040816326534e-05, | |
| "loss": 6.863, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7525510204081632e-05, | |
| "loss": 6.7984, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7448979591836734e-05, | |
| "loss": 6.9578, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7372448979591836e-05, | |
| "loss": 6.8244, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.729591836734694e-05, | |
| "loss": 6.7548, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7219387755102043e-05, | |
| "loss": 6.889, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 6.8286, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.7066326530612244e-05, | |
| "loss": 6.797, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.698979591836735e-05, | |
| "loss": 6.9046, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.691326530612245e-05, | |
| "loss": 6.9557, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.683673469387755e-05, | |
| "loss": 6.8563, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.676020408163265e-05, | |
| "loss": 6.6538, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.6683673469387757e-05, | |
| "loss": 6.8718, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.660714285714286e-05, | |
| "loss": 6.8163, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.6530612244897957e-05, | |
| "loss": 6.9295, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.6454081632653062e-05, | |
| "loss": 6.8405, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6377551020408164e-05, | |
| "loss": 6.8496, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6301020408163266e-05, | |
| "loss": 6.9388, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6224489795918368e-05, | |
| "loss": 6.9675, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.614795918367347e-05, | |
| "loss": 6.7805, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6071428571428572e-05, | |
| "loss": 6.7653, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.5994897959183674e-05, | |
| "loss": 6.999, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.5918367346938776e-05, | |
| "loss": 6.8595, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.5841836734693878e-05, | |
| "loss": 6.8792, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.576530612244898e-05, | |
| "loss": 6.7378, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.568877551020408e-05, | |
| "loss": 6.8505, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5612244897959183e-05, | |
| "loss": 6.6303, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.553571428571429e-05, | |
| "loss": 6.8162, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5459183673469387e-05, | |
| "loss": 6.8154, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.538265306122449e-05, | |
| "loss": 6.8108, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.530612244897959e-05, | |
| "loss": 6.685, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5229591836734695e-05, | |
| "loss": 6.4358, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5153061224489795e-05, | |
| "loss": 6.5531, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5076530612244897e-05, | |
| "loss": 6.2196, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.5e-05, | |
| "loss": 6.9518, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4923469387755104e-05, | |
| "loss": 6.9022, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4846938775510204e-05, | |
| "loss": 6.8897, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4770408163265308e-05, | |
| "loss": 6.7608, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4693877551020408e-05, | |
| "loss": 6.8291, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4617346938775512e-05, | |
| "loss": 6.9625, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4540816326530612e-05, | |
| "loss": 6.952, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4464285714285715e-05, | |
| "loss": 7.0419, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4387755102040816e-05, | |
| "loss": 6.8927, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.431122448979592e-05, | |
| "loss": 6.849, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4234693877551021e-05, | |
| "loss": 6.8324, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4158163265306123e-05, | |
| "loss": 6.727, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4081632653061225e-05, | |
| "loss": 6.7543, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4005102040816327e-05, | |
| "loss": 6.8509, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3928571428571429e-05, | |
| "loss": 6.9081, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.385204081632653e-05, | |
| "loss": 6.932, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3775510204081633e-05, | |
| "loss": 7.0868, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3698979591836735e-05, | |
| "loss": 6.8515, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3622448979591836e-05, | |
| "loss": 6.738, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3545918367346938e-05, | |
| "loss": 6.7979, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3469387755102042e-05, | |
| "loss": 6.875, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3392857142857144e-05, | |
| "loss": 6.7573, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3316326530612246e-05, | |
| "loss": 6.7725, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3239795918367348e-05, | |
| "loss": 6.9417, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.316326530612245e-05, | |
| "loss": 6.8765, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3086734693877551e-05, | |
| "loss": 6.9876, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3010204081632653e-05, | |
| "loss": 6.8864, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2933673469387755e-05, | |
| "loss": 6.7862, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2857142857142857e-05, | |
| "loss": 6.8214, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2780612244897959e-05, | |
| "loss": 6.9334, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2704081632653063e-05, | |
| "loss": 7.0888, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2627551020408163e-05, | |
| "loss": 6.9266, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2551020408163267e-05, | |
| "loss": 6.8692, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.2474489795918367e-05, | |
| "loss": 7.0387, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.239795918367347e-05, | |
| "loss": 6.7814, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.232142857142857e-05, | |
| "loss": 6.9312, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2244897959183674e-05, | |
| "loss": 6.8926, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2168367346938774e-05, | |
| "loss": 6.9204, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.2091836734693878e-05, | |
| "loss": 6.797, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.201530612244898e-05, | |
| "loss": 6.8846, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1938775510204082e-05, | |
| "loss": 6.9006, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1862244897959185e-05, | |
| "loss": 6.8803, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1785714285714286e-05, | |
| "loss": 6.7397, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.170918367346939e-05, | |
| "loss": 6.8819, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.163265306122449e-05, | |
| "loss": 6.7479, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1556122448979593e-05, | |
| "loss": 6.9675, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1479591836734693e-05, | |
| "loss": 6.6359, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1403061224489797e-05, | |
| "loss": 6.5651, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1326530612244897e-05, | |
| "loss": 6.5824, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.125e-05, | |
| "loss": 6.4211, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1173469387755103e-05, | |
| "loss": 6.8861, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1096938775510205e-05, | |
| "loss": 6.9163, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1020408163265306e-05, | |
| "loss": 6.9292, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0943877551020408e-05, | |
| "loss": 6.8178, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.086734693877551e-05, | |
| "loss": 6.8772, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0790816326530612e-05, | |
| "loss": 6.9699, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0714285714285714e-05, | |
| "loss": 6.8686, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0637755102040816e-05, | |
| "loss": 6.7655, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0561224489795918e-05, | |
| "loss": 6.8784, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0484693877551021e-05, | |
| "loss": 6.8053, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0408163265306123e-05, | |
| "loss": 6.8588, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.0331632653061225e-05, | |
| "loss": 6.8353, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0255102040816327e-05, | |
| "loss": 6.6755, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0178571428571429e-05, | |
| "loss": 6.692, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0102040816326531e-05, | |
| "loss": 6.8417, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0025510204081633e-05, | |
| "loss": 6.7841, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.948979591836735e-06, | |
| "loss": 6.7975, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.872448979591837e-06, | |
| "loss": 6.871, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.795918367346939e-06, | |
| "loss": 6.9317, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.71938775510204e-06, | |
| "loss": 6.7614, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.642857142857144e-06, | |
| "loss": 6.9313, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.566326530612244e-06, | |
| "loss": 6.9925, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.489795918367348e-06, | |
| "loss": 6.9193, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.413265306122448e-06, | |
| "loss": 6.7339, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.336734693877552e-06, | |
| "loss": 6.6202, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.260204081632652e-06, | |
| "loss": 6.7642, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.183673469387756e-06, | |
| "loss": 6.6974, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.107142857142856e-06, | |
| "loss": 6.8733, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.03061224489796e-06, | |
| "loss": 6.8635, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.954081632653063e-06, | |
| "loss": 6.8125, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.877551020408163e-06, | |
| "loss": 6.8385, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.801020408163267e-06, | |
| "loss": 6.7776, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.724489795918367e-06, | |
| "loss": 6.9029, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.64795918367347e-06, | |
| "loss": 6.7458, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 6.8837, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.494897959183675e-06, | |
| "loss": 6.8629, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.418367346938775e-06, | |
| "loss": 6.7897, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.341836734693878e-06, | |
| "loss": 6.8748, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.265306122448979e-06, | |
| "loss": 6.7965, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.188775510204082e-06, | |
| "loss": 6.7865, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.112244897959184e-06, | |
| "loss": 6.7277, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.035714285714286e-06, | |
| "loss": 6.6408, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.959183673469388e-06, | |
| "loss": 6.7423, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.88265306122449e-06, | |
| "loss": 6.7093, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.806122448979592e-06, | |
| "loss": 6.8157, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.729591836734694e-06, | |
| "loss": 6.7312, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.653061224489796e-06, | |
| "loss": 6.7111, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.5765306122448974e-06, | |
| "loss": 6.7191, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.5e-06, | |
| "loss": 6.8694, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.423469387755102e-06, | |
| "loss": 6.3313, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.346938775510204e-06, | |
| "loss": 6.91, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.270408163265306e-06, | |
| "loss": 6.849, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.193877551020408e-06, | |
| "loss": 6.9105, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.117346938775511e-06, | |
| "loss": 6.9051, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.0408163265306125e-06, | |
| "loss": 6.6685, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.964285714285714e-06, | |
| "loss": 6.9164, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.887755102040816e-06, | |
| "loss": 6.8981, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.811224489795918e-06, | |
| "loss": 6.7168, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.734693877551021e-06, | |
| "loss": 6.8595, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.658163265306123e-06, | |
| "loss": 6.8056, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.581632653061225e-06, | |
| "loss": 6.9164, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.505102040816327e-06, | |
| "loss": 6.9122, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.428571428571429e-06, | |
| "loss": 6.7648, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.352040816326531e-06, | |
| "loss": 6.845, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.275510204081633e-06, | |
| "loss": 6.8755, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.198979591836735e-06, | |
| "loss": 6.7565, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.122448979591837e-06, | |
| "loss": 6.9427, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.045918367346939e-06, | |
| "loss": 6.8183, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.969387755102041e-06, | |
| "loss": 6.7349, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.892857142857143e-06, | |
| "loss": 6.8932, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.816326530612245e-06, | |
| "loss": 6.9249, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.739795918367347e-06, | |
| "loss": 6.8097, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.6632653061224486e-06, | |
| "loss": 6.9243, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.586734693877551e-06, | |
| "loss": 6.7965, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.510204081632653e-06, | |
| "loss": 7.0408, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.433673469387755e-06, | |
| "loss": 6.9394, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.357142857142857e-06, | |
| "loss": 6.8263, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.280612244897959e-06, | |
| "loss": 6.7378, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.204081632653062e-06, | |
| "loss": 6.7529, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.127551020408164e-06, | |
| "loss": 6.8206, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.0510204081632655e-06, | |
| "loss": 6.7545, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.9744897959183674e-06, | |
| "loss": 6.7889, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.897959183673469e-06, | |
| "loss": 6.8035, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.821428571428572e-06, | |
| "loss": 6.8525, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.744897959183674e-06, | |
| "loss": 6.8054, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.668367346938776e-06, | |
| "loss": 6.9522, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.591836734693878e-06, | |
| "loss": 6.8324, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.51530612244898e-06, | |
| "loss": 6.9383, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.438775510204082e-06, | |
| "loss": 6.7394, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.3622448979591836e-06, | |
| "loss": 6.7162, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 6.8783, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.209183673469387e-06, | |
| "loss": 6.7736, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.132653061224489e-06, | |
| "loss": 6.976, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.056122448979592e-06, | |
| "loss": 6.7993, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.979591836734694e-06, | |
| "loss": 6.807, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.903061224489796e-06, | |
| "loss": 6.7076, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.826530612244898e-06, | |
| "loss": 6.5819, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.75e-06, | |
| "loss": 6.8109, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.673469387755102e-06, | |
| "loss": 6.5935, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.596938775510204e-06, | |
| "loss": 6.3915, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.5204081632653062e-06, | |
| "loss": 6.8856, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.443877551020408e-06, | |
| "loss": 6.8172, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.3673469387755105e-06, | |
| "loss": 7.0145, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.2908163265306124e-06, | |
| "loss": 6.8908, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.2142857142857143e-06, | |
| "loss": 6.8144, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.1377551020408166e-06, | |
| "loss": 6.9178, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.0612244897959185e-06, | |
| "loss": 6.9697, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.9846938775510205e-06, | |
| "loss": 6.902, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.9081632653061224e-06, | |
| "loss": 6.7325, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.8316326530612243e-06, | |
| "loss": 6.8723, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.7551020408163266e-06, | |
| "loss": 7.0073, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.6785714285714285e-06, | |
| "loss": 6.9034, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.602040816326531e-06, | |
| "loss": 6.8114, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.5255102040816328e-06, | |
| "loss": 7.0092, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.4489795918367347e-06, | |
| "loss": 6.9172, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.372448979591837e-06, | |
| "loss": 6.7697, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.295918367346939e-06, | |
| "loss": 6.9497, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.219387755102041e-06, | |
| "loss": 6.7357, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 6.853, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.0663265306122446e-06, | |
| "loss": 6.8345, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.989795918367347e-06, | |
| "loss": 6.7548, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.913265306122449e-06, | |
| "loss": 6.8532, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.836734693877551e-06, | |
| "loss": 6.8114, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7602040816326531e-06, | |
| "loss": 6.9239, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6836734693877552e-06, | |
| "loss": 6.9835, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6071428571428572e-06, | |
| "loss": 6.7284, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.5306122448979593e-06, | |
| "loss": 6.7125, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.4540816326530612e-06, | |
| "loss": 6.8183, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3775510204081633e-06, | |
| "loss": 6.8289, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3010204081632654e-06, | |
| "loss": 6.9183, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2244897959183673e-06, | |
| "loss": 6.6072, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1479591836734695e-06, | |
| "loss": 6.8833, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0714285714285714e-06, | |
| "loss": 6.7338, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.948979591836735e-07, | |
| "loss": 6.8157, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.183673469387755e-07, | |
| "loss": 6.8095, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.418367346938776e-07, | |
| "loss": 6.8044, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.653061224489796e-07, | |
| "loss": 6.5781, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.887755102040817e-07, | |
| "loss": 6.7909, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.122448979591837e-07, | |
| "loss": 6.6131, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.357142857142857e-07, | |
| "loss": 6.6244, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.5918367346938775e-07, | |
| "loss": 6.4836, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.826530612244898e-07, | |
| "loss": 6.2152, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 892, | |
| "total_flos": 0.0, | |
| "train_loss": 6.990358243074118, | |
| "train_runtime": 6205.6335, | |
| "train_samples_per_second": 4.599, | |
| "train_steps_per_second": 0.144 | |
| } | |
| ], | |
| "max_steps": 892, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |