{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5503329847542517,
  "eval_steps": 500,
  "global_step": 9999,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.503880235566074e-05,
      "grad_norm": 459.8753356933594,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 3.303,
      "step": 1
    },
    {
      "epoch": 0.00011007760471132149,
      "grad_norm": 314.2561950683594,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.8226,
      "step": 2
    },
    {
      "epoch": 0.0001651164070669822,
      "grad_norm": 314.1292419433594,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 2.8517,
      "step": 3
    },
    {
      "epoch": 0.00022015520942264297,
      "grad_norm": 312.4049072265625,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.6248,
      "step": 4
    },
    {
      "epoch": 0.0002751940117783037,
      "grad_norm": 353.7213134765625,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.7883,
      "step": 5
    },
    {
      "epoch": 0.0003302328141339644,
      "grad_norm": 278.41668701171875,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.5468,
      "step": 6
    },
    {
      "epoch": 0.0003852716164896252,
      "grad_norm": 336.14532470703125,
      "learning_rate": 7.000000000000001e-07,
      "loss": 2.7721,
      "step": 7
    },
    {
      "epoch": 0.00044031041884528595,
      "grad_norm": 201.19374084472656,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.4873,
      "step": 8
    },
    {
      "epoch": 0.0004953492212009466,
      "grad_norm": 184.7027587890625,
      "learning_rate": 9.000000000000001e-07,
      "loss": 2.6647,
      "step": 9
    },
    {
      "epoch": 0.0005503880235566074,
      "grad_norm": 154.597412109375,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 2.602,
      "step": 10
    },
    {
      "epoch": 0.0006054268259122681,
      "grad_norm": 40.47785568237305,
      "learning_rate": 1.1e-06,
      "loss": 2.6716,
      "step": 11
    },
    {
      "epoch": 0.0006604656282679288,
      "grad_norm": 25.338607788085938,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.2631,
      "step": 12
    },
    {
      "epoch": 0.0007155044306235897,
      "grad_norm": 24.976919174194336,
      "learning_rate": 1.3e-06,
      "loss": 2.3564,
      "step": 13
    },
    {
      "epoch": 0.0007705432329792504,
      "grad_norm": 15.239912033081055,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 2.3295,
      "step": 14
    },
    {
      "epoch": 0.0008255820353349112,
      "grad_norm": 14.125042915344238,
      "learning_rate": 1.5e-06,
      "loss": 2.307,
      "step": 15
    },
    {
      "epoch": 0.0008806208376905719,
      "grad_norm": 13.163726806640625,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 2.1493,
      "step": 16
    },
    {
      "epoch": 0.0009356596400462326,
      "grad_norm": 8.726515769958496,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 2.0333,
      "step": 17
    },
    {
      "epoch": 0.0009906984424018933,
      "grad_norm": 9.072502136230469,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 2.2046,
      "step": 18
    },
    {
      "epoch": 0.001045737244757554,
      "grad_norm": 9.412588119506836,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 2.2001,
      "step": 19
    },
    {
      "epoch": 0.0011007760471132147,
      "grad_norm": 8.67534065246582,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.7679,
      "step": 20
    },
    {
      "epoch": 0.0011558148494688755,
      "grad_norm": 14.015918731689453,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 1.9566,
      "step": 21
    },
    {
      "epoch": 0.0012108536518245362,
      "grad_norm": 7.9474687576293945,
      "learning_rate": 2.2e-06,
      "loss": 1.9085,
      "step": 22
    },
    {
      "epoch": 0.001265892454180197,
      "grad_norm": 6.806368350982666,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 1.7918,
      "step": 23
    },
    {
      "epoch": 0.0013209312565358577,
      "grad_norm": 5.3452582359313965,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.8321,
      "step": 24
    },
    {
      "epoch": 0.0013759700588915184,
      "grad_norm": 8.744244575500488,
      "learning_rate": 2.5e-06,
      "loss": 1.6317,
      "step": 25
    },
    {
      "epoch": 0.0014310088612471794,
      "grad_norm": 5.304683685302734,
      "learning_rate": 2.6e-06,
      "loss": 1.6846,
      "step": 26
    },
    {
      "epoch": 0.00148604766360284,
      "grad_norm": 5.650127410888672,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 1.7449,
      "step": 27
    },
    {
      "epoch": 0.0015410864659585008,
      "grad_norm": 5.479269504547119,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.8158,
      "step": 28
    },
    {
      "epoch": 0.0015961252683141616,
      "grad_norm": 4.873537063598633,
      "learning_rate": 2.9e-06,
      "loss": 1.8015,
      "step": 29
    },
    {
      "epoch": 0.0016511640706698223,
      "grad_norm": 4.971101760864258,
      "learning_rate": 3e-06,
      "loss": 1.9034,
      "step": 30
    },
    {
      "epoch": 0.001706202873025483,
      "grad_norm": 4.407571315765381,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 1.9037,
      "step": 31
    },
    {
      "epoch": 0.0017612416753811438,
      "grad_norm": 4.429073810577393,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.6812,
      "step": 32
    },
    {
      "epoch": 0.0018162804777368045,
      "grad_norm": 5.16085147857666,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 1.7627,
      "step": 33
    },
    {
      "epoch": 0.0018713192800924653,
      "grad_norm": 4.0805768966674805,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.6799,
      "step": 34
    },
    {
      "epoch": 0.001926358082448126,
      "grad_norm": 4.548702239990234,
      "learning_rate": 3.5e-06,
      "loss": 1.7799,
      "step": 35
    },
    {
      "epoch": 0.0019813968848037865,
      "grad_norm": 5.181888580322266,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.8235,
      "step": 36
    },
    {
      "epoch": 0.0020364356871594475,
      "grad_norm": 3.9876129627227783,
      "learning_rate": 3.7e-06,
      "loss": 1.5999,
      "step": 37
    },
    {
      "epoch": 0.002091474489515108,
      "grad_norm": 6.325051307678223,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 1.7499,
      "step": 38
    },
    {
      "epoch": 0.002146513291870769,
      "grad_norm": 6.199049949645996,
      "learning_rate": 3.900000000000001e-06,
      "loss": 1.784,
      "step": 39
    },
    {
      "epoch": 0.0022015520942264295,
      "grad_norm": 4.83912992477417,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.8895,
      "step": 40
    },
    {
      "epoch": 0.0022565908965820904,
      "grad_norm": 4.515626907348633,
      "learning_rate": 4.1e-06,
      "loss": 1.4887,
      "step": 41
    },
    {
      "epoch": 0.002311629698937751,
      "grad_norm": 5.032265663146973,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.7324,
      "step": 42
    },
    {
      "epoch": 0.002366668501293412,
      "grad_norm": 4.1879048347473145,
      "learning_rate": 4.3e-06,
      "loss": 1.4912,
      "step": 43
    },
    {
      "epoch": 0.0024217073036490724,
      "grad_norm": 4.128026485443115,
      "learning_rate": 4.4e-06,
      "loss": 1.554,
      "step": 44
    },
    {
      "epoch": 0.0024767461060047334,
      "grad_norm": 4.527958393096924,
      "learning_rate": 4.5e-06,
      "loss": 1.652,
      "step": 45
    },
    {
      "epoch": 0.002531784908360394,
      "grad_norm": 4.8388190269470215,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.6696,
      "step": 46
    },
    {
      "epoch": 0.002586823710716055,
      "grad_norm": 4.2088541984558105,
      "learning_rate": 4.7e-06,
      "loss": 1.568,
      "step": 47
    },
    {
      "epoch": 0.0026418625130717154,
      "grad_norm": 4.789997577667236,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.642,
      "step": 48
    },
    {
      "epoch": 0.0026969013154273763,
      "grad_norm": 4.408346652984619,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 1.5181,
      "step": 49
    },
    {
      "epoch": 0.002751940117783037,
      "grad_norm": 4.572340488433838,
      "learning_rate": 5e-06,
      "loss": 1.6698,
      "step": 50
    },
    {
      "epoch": 0.0028069789201386978,
      "grad_norm": 4.728564739227295,
      "learning_rate": 5.1e-06,
      "loss": 1.5785,
      "step": 51
    },
    {
      "epoch": 0.0028620177224943587,
      "grad_norm": 4.449855327606201,
      "learning_rate": 5.2e-06,
      "loss": 1.4624,
      "step": 52
    },
    {
      "epoch": 0.0029170565248500193,
      "grad_norm": 4.127189636230469,
      "learning_rate": 5.300000000000001e-06,
      "loss": 1.6061,
      "step": 53
    },
    {
      "epoch": 0.00297209532720568,
      "grad_norm": 4.244532108306885,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.491,
      "step": 54
    },
    {
      "epoch": 0.0030271341295613407,
      "grad_norm": 3.437682628631592,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.1967,
      "step": 55
    },
    {
      "epoch": 0.0030821729319170017,
      "grad_norm": 3.83516788482666,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.4731,
      "step": 56
    },
    {
      "epoch": 0.003137211734272662,
      "grad_norm": 3.9108972549438477,
      "learning_rate": 5.7e-06,
      "loss": 1.4393,
      "step": 57
    },
    {
      "epoch": 0.003192250536628323,
      "grad_norm": 3.5258419513702393,
      "learning_rate": 5.8e-06,
      "loss": 1.4206,
      "step": 58
    },
    {
      "epoch": 0.0032472893389839837,
      "grad_norm": 4.124903678894043,
      "learning_rate": 5.9e-06,
      "loss": 1.4747,
      "step": 59
    },
    {
      "epoch": 0.0033023281413396446,
      "grad_norm": 4.055769920349121,
      "learning_rate": 6e-06,
      "loss": 1.4655,
      "step": 60
    },
    {
      "epoch": 0.003357366943695305,
      "grad_norm": 3.904837131500244,
      "learning_rate": 6.1e-06,
      "loss": 1.5125,
      "step": 61
    },
    {
      "epoch": 0.003412405746050966,
      "grad_norm": 3.2904794216156006,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.4596,
      "step": 62
    },
    {
      "epoch": 0.0034674445484066266,
      "grad_norm": 3.24053692817688,
      "learning_rate": 6.300000000000001e-06,
      "loss": 1.3851,
      "step": 63
    },
    {
      "epoch": 0.0035224833507622876,
      "grad_norm": 3.457639217376709,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 1.4019,
      "step": 64
    },
    {
      "epoch": 0.003577522153117948,
      "grad_norm": 3.073054790496826,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.2872,
      "step": 65
    },
    {
      "epoch": 0.003632560955473609,
      "grad_norm": 2.6726694107055664,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.2361,
      "step": 66
    },
    {
      "epoch": 0.0036875997578292696,
      "grad_norm": 2.9378459453582764,
      "learning_rate": 6.700000000000001e-06,
      "loss": 1.4452,
      "step": 67
    },
    {
      "epoch": 0.0037426385601849305,
      "grad_norm": 2.81107234954834,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.4804,
      "step": 68
    },
    {
      "epoch": 0.003797677362540591,
      "grad_norm": 2.60062313079834,
      "learning_rate": 6.9e-06,
      "loss": 1.3263,
      "step": 69
    },
    {
      "epoch": 0.003852716164896252,
      "grad_norm": 2.5642921924591064,
      "learning_rate": 7e-06,
      "loss": 1.2751,
      "step": 70
    },
    {
      "epoch": 0.0039077549672519125,
      "grad_norm": 2.3608031272888184,
      "learning_rate": 7.100000000000001e-06,
      "loss": 1.2614,
      "step": 71
    },
    {
      "epoch": 0.003962793769607573,
      "grad_norm": 2.7201738357543945,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.5018,
      "step": 72
    },
    {
      "epoch": 0.004017832571963234,
      "grad_norm": 2.584726095199585,
      "learning_rate": 7.3e-06,
      "loss": 1.3519,
      "step": 73
    },
    {
      "epoch": 0.004072871374318895,
      "grad_norm": 1.9693044424057007,
      "learning_rate": 7.4e-06,
      "loss": 1.0934,
      "step": 74
    },
    {
      "epoch": 0.0041279101766745555,
      "grad_norm": 2.220736503601074,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.4687,
      "step": 75
    },
    {
      "epoch": 0.004182948979030216,
      "grad_norm": 2.2629456520080566,
      "learning_rate": 7.600000000000001e-06,
      "loss": 1.3328,
      "step": 76
    },
    {
      "epoch": 0.004237987781385877,
      "grad_norm": 2.051820993423462,
      "learning_rate": 7.7e-06,
      "loss": 1.3058,
      "step": 77
    },
    {
      "epoch": 0.004293026583741538,
      "grad_norm": 2.2451820373535156,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.3556,
      "step": 78
    },
    {
      "epoch": 0.004348065386097198,
      "grad_norm": 3.13584303855896,
      "learning_rate": 7.9e-06,
      "loss": 1.3262,
      "step": 79
    },
    {
      "epoch": 0.004403104188452859,
      "grad_norm": 5.024479866027832,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2103,
      "step": 80
    },
    {
      "epoch": 0.00445814299080852,
      "grad_norm": 2.070889711380005,
      "learning_rate": 8.1e-06,
      "loss": 1.1994,
      "step": 81
    },
    {
      "epoch": 0.004513181793164181,
      "grad_norm": 2.797286033630371,
      "learning_rate": 8.2e-06,
      "loss": 1.3075,
      "step": 82
    },
    {
      "epoch": 0.004568220595519841,
      "grad_norm": 2.11370849609375,
      "learning_rate": 8.3e-06,
      "loss": 1.36,
      "step": 83
    },
    {
      "epoch": 0.004623259397875502,
      "grad_norm": 2.5416152477264404,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.3484,
      "step": 84
    },
    {
      "epoch": 0.004678298200231163,
      "grad_norm": 2.4702343940734863,
      "learning_rate": 8.5e-06,
      "loss": 1.3677,
      "step": 85
    },
    {
      "epoch": 0.004733337002586824,
      "grad_norm": 3.670365333557129,
      "learning_rate": 8.6e-06,
      "loss": 1.2192,
      "step": 86
    },
    {
      "epoch": 0.004788375804942484,
      "grad_norm": 2.282954692840576,
      "learning_rate": 8.700000000000001e-06,
      "loss": 1.2982,
      "step": 87
    },
    {
      "epoch": 0.004843414607298145,
      "grad_norm": 2.3659238815307617,
      "learning_rate": 8.8e-06,
      "loss": 1.3206,
      "step": 88
    },
    {
      "epoch": 0.004898453409653806,
      "grad_norm": 4.939981460571289,
      "learning_rate": 8.900000000000001e-06,
      "loss": 1.4328,
      "step": 89
    },
    {
      "epoch": 0.004953492212009467,
      "grad_norm": 2.335858106613159,
      "learning_rate": 9e-06,
      "loss": 1.2603,
      "step": 90
    },
    {
      "epoch": 0.005008531014365127,
      "grad_norm": 2.2165043354034424,
      "learning_rate": 9.100000000000001e-06,
      "loss": 1.3141,
      "step": 91
    },
    {
      "epoch": 0.005063569816720788,
      "grad_norm": 2.7872185707092285,
      "learning_rate": 9.200000000000002e-06,
      "loss": 1.3314,
      "step": 92
    },
    {
      "epoch": 0.005118608619076449,
      "grad_norm": 2.6353912353515625,
      "learning_rate": 9.3e-06,
      "loss": 1.2027,
      "step": 93
    },
    {
      "epoch": 0.00517364742143211,
      "grad_norm": 3.2509102821350098,
      "learning_rate": 9.4e-06,
      "loss": 1.2316,
      "step": 94
    },
    {
      "epoch": 0.00522868622378777,
      "grad_norm": 2.4560611248016357,
      "learning_rate": 9.5e-06,
      "loss": 1.1848,
      "step": 95
    },
    {
      "epoch": 0.005283725026143431,
      "grad_norm": 2.338151216506958,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.2392,
      "step": 96
    },
    {
      "epoch": 0.005338763828499092,
      "grad_norm": 2.231065034866333,
      "learning_rate": 9.7e-06,
      "loss": 1.2089,
      "step": 97
    },
    {
      "epoch": 0.005393802630854753,
      "grad_norm": 2.278428077697754,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.2267,
      "step": 98
    },
    {
      "epoch": 0.005448841433210413,
      "grad_norm": 2.4422810077667236,
      "learning_rate": 9.9e-06,
      "loss": 1.2041,
      "step": 99
    },
    {
      "epoch": 0.005503880235566074,
      "grad_norm": 2.216248035430908,
      "learning_rate": 1e-05,
      "loss": 1.0798,
      "step": 100
    },
    {
      "epoch": 0.005558919037921735,
      "grad_norm": 2.3301615715026855,
      "learning_rate": 9.99999998121067e-06,
      "loss": 1.3069,
      "step": 101
    },
    {
      "epoch": 0.0056139578402773956,
      "grad_norm": 2.315436363220215,
      "learning_rate": 9.999999924842678e-06,
      "loss": 1.1589,
      "step": 102
    },
    {
      "epoch": 0.005668996642633056,
      "grad_norm": 2.3522140979766846,
      "learning_rate": 9.999999830896024e-06,
      "loss": 1.0978,
      "step": 103
    },
    {
      "epoch": 0.0057240354449887175,
      "grad_norm": 2.5798308849334717,
      "learning_rate": 9.99999969937071e-06,
      "loss": 1.0599,
      "step": 104
    },
    {
      "epoch": 0.005779074247344378,
      "grad_norm": 2.456644058227539,
      "learning_rate": 9.999999530266738e-06,
      "loss": 1.1682,
      "step": 105
    },
    {
      "epoch": 0.0058341130497000385,
      "grad_norm": 2.1559031009674072,
      "learning_rate": 9.999999323584106e-06,
      "loss": 1.0631,
      "step": 106
    },
    {
      "epoch": 0.005889151852055699,
      "grad_norm": 2.2985048294067383,
      "learning_rate": 9.99999907932282e-06,
      "loss": 1.1455,
      "step": 107
    },
    {
      "epoch": 0.00594419065441136,
      "grad_norm": 2.596167802810669,
      "learning_rate": 9.999998797482877e-06,
      "loss": 1.1686,
      "step": 108
    },
    {
      "epoch": 0.005999229456767021,
      "grad_norm": 2.378618001937866,
      "learning_rate": 9.999998478064283e-06,
      "loss": 1.2226,
      "step": 109
    },
    {
      "epoch": 0.0060542682591226814,
      "grad_norm": 2.228116750717163,
      "learning_rate": 9.999998121067038e-06,
      "loss": 1.1396,
      "step": 110
    },
    {
      "epoch": 0.006109307061478342,
      "grad_norm": 2.4419472217559814,
      "learning_rate": 9.999997726491146e-06,
      "loss": 1.1401,
      "step": 111
    },
    {
      "epoch": 0.006164345863834003,
      "grad_norm": 2.0695526599884033,
      "learning_rate": 9.999997294336608e-06,
      "loss": 1.1868,
      "step": 112
    },
    {
      "epoch": 0.006219384666189664,
      "grad_norm": 2.3170363903045654,
      "learning_rate": 9.99999682460343e-06,
      "loss": 1.1172,
      "step": 113
    },
    {
      "epoch": 0.006274423468545324,
      "grad_norm": 2.670466184616089,
      "learning_rate": 9.999996317291615e-06,
      "loss": 1.2481,
      "step": 114
    },
    {
      "epoch": 0.006329462270900985,
      "grad_norm": 2.1214540004730225,
      "learning_rate": 9.999995772401166e-06,
      "loss": 0.9994,
      "step": 115
    },
    {
      "epoch": 0.006384501073256646,
      "grad_norm": 1.9283969402313232,
      "learning_rate": 9.999995189932085e-06,
      "loss": 1.0692,
      "step": 116
    },
    {
      "epoch": 0.006439539875612307,
      "grad_norm": 2.2620882987976074,
      "learning_rate": 9.99999456988438e-06,
      "loss": 1.0725,
      "step": 117
    },
    {
      "epoch": 0.006494578677967967,
      "grad_norm": 2.2121341228485107,
      "learning_rate": 9.999993912258055e-06,
      "loss": 1.1328,
      "step": 118
    },
    {
      "epoch": 0.006549617480323628,
      "grad_norm": 2.298126220703125,
      "learning_rate": 9.999993217053113e-06,
      "loss": 1.1272,
      "step": 119
    },
    {
      "epoch": 0.006604656282679289,
      "grad_norm": 1.81593656539917,
      "learning_rate": 9.99999248426956e-06,
      "loss": 1.017,
      "step": 120
    },
    {
      "epoch": 0.00665969508503495,
      "grad_norm": 2.1174378395080566,
      "learning_rate": 9.999991713907403e-06,
      "loss": 1.0557,
      "step": 121
    },
    {
      "epoch": 0.00671473388739061,
      "grad_norm": 1.9061017036437988,
      "learning_rate": 9.999990905966647e-06,
      "loss": 1.0379,
      "step": 122
    },
    {
      "epoch": 0.006769772689746271,
      "grad_norm": 1.912500023841858,
      "learning_rate": 9.999990060447297e-06,
      "loss": 1.104,
      "step": 123
    },
    {
      "epoch": 0.006824811492101932,
      "grad_norm": 1.9249529838562012,
      "learning_rate": 9.99998917734936e-06,
      "loss": 1.0136,
      "step": 124
    },
    {
      "epoch": 0.006879850294457593,
      "grad_norm": 1.8504948616027832,
      "learning_rate": 9.999988256672843e-06,
      "loss": 0.99,
      "step": 125
    },
    {
      "epoch": 0.006934889096813253,
      "grad_norm": 1.720042109489441,
      "learning_rate": 9.999987298417753e-06,
      "loss": 1.0666,
      "step": 126
    },
    {
      "epoch": 0.006989927899168914,
      "grad_norm": 1.778251051902771,
      "learning_rate": 9.999986302584097e-06,
      "loss": 1.0424,
      "step": 127
    },
    {
      "epoch": 0.007044966701524575,
      "grad_norm": 1.9485961198806763,
      "learning_rate": 9.999985269171881e-06,
      "loss": 1.105,
      "step": 128
    },
    {
      "epoch": 0.007100005503880236,
      "grad_norm": 3.0802104473114014,
      "learning_rate": 9.999984198181114e-06,
      "loss": 1.1081,
      "step": 129
    },
    {
      "epoch": 0.007155044306235896,
      "grad_norm": 1.7476954460144043,
      "learning_rate": 9.999983089611806e-06,
      "loss": 0.9677,
      "step": 130
    },
    {
      "epoch": 0.007210083108591557,
      "grad_norm": 1.6127299070358276,
      "learning_rate": 9.999981943463963e-06,
      "loss": 0.9937,
      "step": 131
    },
    {
      "epoch": 0.007265121910947218,
      "grad_norm": 2.1477208137512207,
      "learning_rate": 9.999980759737594e-06,
      "loss": 1.0319,
      "step": 132
    },
    {
      "epoch": 0.007320160713302879,
      "grad_norm": 1.531163215637207,
      "learning_rate": 9.999979538432707e-06,
      "loss": 0.8696,
      "step": 133
    },
    {
      "epoch": 0.007375199515658539,
      "grad_norm": 1.8226820230484009,
      "learning_rate": 9.999978279549313e-06,
      "loss": 1.2061,
      "step": 134
    },
    {
      "epoch": 0.0074302383180142,
      "grad_norm": 1.481895923614502,
      "learning_rate": 9.99997698308742e-06,
      "loss": 0.949,
      "step": 135
    },
    {
      "epoch": 0.007485277120369861,
      "grad_norm": 1.6715927124023438,
      "learning_rate": 9.99997564904704e-06,
      "loss": 1.1579,
      "step": 136
    },
    {
      "epoch": 0.0075403159227255215,
      "grad_norm": 1.4235272407531738,
      "learning_rate": 9.999974277428179e-06,
      "loss": 1.064,
      "step": 137
    },
    {
      "epoch": 0.007595354725081182,
      "grad_norm": 1.3524872064590454,
      "learning_rate": 9.999972868230852e-06,
      "loss": 0.9141,
      "step": 138
    },
    {
      "epoch": 0.007650393527436843,
      "grad_norm": 1.3741765022277832,
      "learning_rate": 9.999971421455066e-06,
      "loss": 1.0256,
      "step": 139
    },
    {
      "epoch": 0.007705432329792504,
      "grad_norm": 1.9869598150253296,
      "learning_rate": 9.999969937100835e-06,
      "loss": 0.9489,
      "step": 140
    },
    {
      "epoch": 0.0077604711321481645,
      "grad_norm": 1.4785465002059937,
      "learning_rate": 9.999968415168166e-06,
      "loss": 0.9243,
      "step": 141
    },
    {
      "epoch": 0.007815509934503825,
      "grad_norm": 1.5476176738739014,
      "learning_rate": 9.999966855657074e-06,
      "loss": 1.178,
      "step": 142
    },
    {
      "epoch": 0.007870548736859486,
      "grad_norm": 1.500401258468628,
      "learning_rate": 9.99996525856757e-06,
      "loss": 0.9837,
      "step": 143
    },
    {
      "epoch": 0.007925587539215146,
      "grad_norm": 1.3777157068252563,
      "learning_rate": 9.999963623899664e-06,
      "loss": 1.0732,
      "step": 144
    },
    {
      "epoch": 0.007980626341570807,
      "grad_norm": 1.4466841220855713,
      "learning_rate": 9.99996195165337e-06,
      "loss": 0.9779,
      "step": 145
    },
    {
      "epoch": 0.008035665143926469,
      "grad_norm": 1.5304051637649536,
      "learning_rate": 9.9999602418287e-06,
      "loss": 1.196,
      "step": 146
    },
    {
      "epoch": 0.008090703946282128,
      "grad_norm": 1.9012362957000732,
      "learning_rate": 9.99995849442567e-06,
      "loss": 0.9797,
      "step": 147
    },
    {
      "epoch": 0.00814574274863779,
      "grad_norm": 1.430679202079773,
      "learning_rate": 9.999956709444289e-06,
      "loss": 0.9869,
      "step": 148
    },
    {
      "epoch": 0.00820078155099345,
      "grad_norm": 1.3489817380905151,
      "learning_rate": 9.99995488688457e-06,
      "loss": 1.0137,
      "step": 149
    },
    {
      "epoch": 0.008255820353349111,
      "grad_norm": 1.1878125667572021,
      "learning_rate": 9.999953026746531e-06,
      "loss": 0.9355,
      "step": 150
    },
    {
      "epoch": 0.008310859155704772,
      "grad_norm": 1.3481942415237427,
      "learning_rate": 9.999951129030182e-06,
      "loss": 1.1235,
      "step": 151
    },
    {
      "epoch": 0.008365897958060432,
      "grad_norm": 1.7335314750671387,
      "learning_rate": 9.999949193735539e-06,
      "loss": 0.9382,
      "step": 152
    },
    {
      "epoch": 0.008420936760416093,
      "grad_norm": 1.2029480934143066,
      "learning_rate": 9.999947220862615e-06,
      "loss": 0.9419,
      "step": 153
    },
    {
      "epoch": 0.008475975562771755,
      "grad_norm": 1.2104203701019287,
      "learning_rate": 9.999945210411428e-06,
      "loss": 0.9196,
      "step": 154
    },
    {
      "epoch": 0.008531014365127414,
      "grad_norm": 1.1857126951217651,
      "learning_rate": 9.999943162381991e-06,
      "loss": 0.9421,
      "step": 155
    },
    {
      "epoch": 0.008586053167483076,
      "grad_norm": 1.115027904510498,
      "learning_rate": 9.999941076774319e-06,
      "loss": 0.9634,
      "step": 156
    },
    {
      "epoch": 0.008641091969838737,
      "grad_norm": 1.4227553606033325,
      "learning_rate": 9.999938953588428e-06,
      "loss": 1.0036,
      "step": 157
    },
    {
      "epoch": 0.008696130772194397,
      "grad_norm": 1.2913776636123657,
      "learning_rate": 9.999936792824334e-06,
      "loss": 0.9232,
      "step": 158
    },
    {
      "epoch": 0.008751169574550058,
      "grad_norm": 1.2817318439483643,
      "learning_rate": 9.999934594482055e-06,
      "loss": 0.9691,
      "step": 159
    },
    {
      "epoch": 0.008806208376905718,
      "grad_norm": 1.5647841691970825,
      "learning_rate": 9.999932358561604e-06,
      "loss": 1.1842,
      "step": 160
    },
    {
      "epoch": 0.00886124717926138,
      "grad_norm": 1.368135929107666,
      "learning_rate": 9.999930085063002e-06,
      "loss": 1.0873,
      "step": 161
    },
    {
      "epoch": 0.00891628598161704,
      "grad_norm": 1.2297240495681763,
      "learning_rate": 9.999927773986262e-06,
      "loss": 1.0778,
      "step": 162
    },
    {
      "epoch": 0.0089713247839727,
      "grad_norm": 1.0658279657363892,
      "learning_rate": 9.999925425331405e-06,
      "loss": 0.9008,
      "step": 163
    },
    {
      "epoch": 0.009026363586328362,
      "grad_norm": 1.3484326601028442,
      "learning_rate": 9.999923039098445e-06,
      "loss": 1.0664,
      "step": 164
    },
    {
      "epoch": 0.009081402388684023,
      "grad_norm": 1.1839075088500977,
      "learning_rate": 9.999920615287401e-06,
      "loss": 0.9257,
      "step": 165
    },
    {
      "epoch": 0.009136441191039683,
      "grad_norm": 1.2757254838943481,
      "learning_rate": 9.999918153898295e-06,
      "loss": 0.9473,
      "step": 166
    },
    {
      "epoch": 0.009191479993395344,
      "grad_norm": 1.2414579391479492,
      "learning_rate": 9.99991565493114e-06,
      "loss": 1.1091,
      "step": 167
    },
    {
      "epoch": 0.009246518795751004,
      "grad_norm": 1.2802611589431763,
      "learning_rate": 9.999913118385959e-06,
      "loss": 1.063,
      "step": 168
    },
    {
      "epoch": 0.009301557598106665,
      "grad_norm": 1.2055327892303467,
      "learning_rate": 9.99991054426277e-06,
      "loss": 0.8,
      "step": 169
    },
    {
      "epoch": 0.009356596400462327,
      "grad_norm": 1.0391098260879517,
      "learning_rate": 9.99990793256159e-06,
      "loss": 0.8672,
      "step": 170
    },
    {
      "epoch": 0.009411635202817986,
      "grad_norm": 1.131536602973938,
      "learning_rate": 9.99990528328244e-06,
      "loss": 0.9569,
      "step": 171
    },
    {
      "epoch": 0.009466674005173648,
      "grad_norm": 1.164307951927185,
      "learning_rate": 9.999902596425342e-06,
      "loss": 0.9999,
      "step": 172
    },
    {
      "epoch": 0.009521712807529309,
      "grad_norm": 1.2099504470825195,
      "learning_rate": 9.999899871990313e-06,
      "loss": 0.9994,
      "step": 173
    },
    {
      "epoch": 0.009576751609884969,
      "grad_norm": 1.7294539213180542,
      "learning_rate": 9.999897109977376e-06,
      "loss": 1.0265,
      "step": 174
    },
    {
      "epoch": 0.00963179041224063,
      "grad_norm": 1.3009883165359497,
      "learning_rate": 9.99989431038655e-06,
      "loss": 0.9022,
      "step": 175
    },
    {
      "epoch": 0.00968682921459629,
      "grad_norm": 1.1014611721038818,
      "learning_rate": 9.999891473217857e-06,
      "loss": 0.8476,
      "step": 176
    },
    {
      "epoch": 0.009741868016951951,
      "grad_norm": 1.2410900592803955,
      "learning_rate": 9.99988859847132e-06,
      "loss": 1.0272,
      "step": 177
    },
    {
      "epoch": 0.009796906819307612,
      "grad_norm": 1.336348295211792,
      "learning_rate": 9.999885686146957e-06,
      "loss": 0.9456,
      "step": 178
    },
    {
      "epoch": 0.009851945621663272,
      "grad_norm": 1.2931095361709595,
      "learning_rate": 9.99988273624479e-06,
      "loss": 0.9554,
      "step": 179
    },
    {
      "epoch": 0.009906984424018933,
      "grad_norm": 1.2647838592529297,
      "learning_rate": 9.999879748764845e-06,
      "loss": 1.0394,
      "step": 180
    },
    {
      "epoch": 0.009962023226374595,
      "grad_norm": 1.3485127687454224,
      "learning_rate": 9.99987672370714e-06,
      "loss": 1.1016,
      "step": 181
    },
    {
      "epoch": 0.010017062028730254,
      "grad_norm": 1.110187292098999,
      "learning_rate": 9.999873661071702e-06,
      "loss": 0.946,
      "step": 182
    },
    {
      "epoch": 0.010072100831085916,
      "grad_norm": 1.0991623401641846,
      "learning_rate": 9.999870560858551e-06,
      "loss": 1.0084,
      "step": 183
    },
    {
      "epoch": 0.010127139633441576,
      "grad_norm": 1.049804449081421,
      "learning_rate": 9.999867423067713e-06,
      "loss": 0.8264,
      "step": 184
    },
    {
      "epoch": 0.010182178435797237,
      "grad_norm": 1.0947058200836182,
      "learning_rate": 9.999864247699207e-06,
      "loss": 0.8884,
      "step": 185
    },
    {
      "epoch": 0.010237217238152898,
      "grad_norm": 1.1147902011871338,
      "learning_rate": 9.999861034753061e-06,
      "loss": 0.9657,
      "step": 186
    },
    {
      "epoch": 0.010292256040508558,
      "grad_norm": 1.260027527809143,
      "learning_rate": 9.999857784229298e-06,
      "loss": 1.0102,
      "step": 187
    },
    {
      "epoch": 0.01034729484286422,
      "grad_norm": 1.1275582313537598,
      "learning_rate": 9.999854496127942e-06,
      "loss": 1.028,
      "step": 188
    },
    {
      "epoch": 0.01040233364521988,
      "grad_norm": 1.1377174854278564,
      "learning_rate": 9.999851170449018e-06,
      "loss": 1.032,
      "step": 189
    },
    {
      "epoch": 0.01045737244757554,
      "grad_norm": 1.1734225749969482,
      "learning_rate": 9.999847807192552e-06,
      "loss": 1.0009,
      "step": 190
    },
    {
      "epoch": 0.010512411249931202,
      "grad_norm": 1.1934596300125122,
      "learning_rate": 9.999844406358565e-06,
      "loss": 1.0432,
      "step": 191
    },
    {
      "epoch": 0.010567450052286861,
      "grad_norm": 1.0638024806976318,
      "learning_rate": 9.99984096794709e-06,
      "loss": 0.8651,
      "step": 192
    },
    {
      "epoch": 0.010622488854642523,
      "grad_norm": 1.2381829023361206,
      "learning_rate": 9.999837491958147e-06,
      "loss": 1.0088,
      "step": 193
    },
    {
      "epoch": 0.010677527656998184,
      "grad_norm": 1.030246615409851,
      "learning_rate": 9.999833978391763e-06,
      "loss": 0.9488,
      "step": 194
    },
    {
      "epoch": 0.010732566459353844,
      "grad_norm": 1.1640657186508179,
      "learning_rate": 9.999830427247965e-06,
      "loss": 1.0588,
      "step": 195
    },
    {
      "epoch": 0.010787605261709505,
      "grad_norm": 1.0431616306304932,
      "learning_rate": 9.99982683852678e-06,
      "loss": 0.8728,
      "step": 196
    },
    {
      "epoch": 0.010842644064065167,
      "grad_norm": 1.032263159751892,
      "learning_rate": 9.999823212228235e-06,
      "loss": 0.9498,
      "step": 197
    },
    {
      "epoch": 0.010897682866420826,
      "grad_norm": 1.1383745670318604,
      "learning_rate": 9.999819548352358e-06,
      "loss": 0.9498,
      "step": 198
    },
    {
      "epoch": 0.010952721668776488,
      "grad_norm": 1.1324639320373535,
      "learning_rate": 9.999815846899175e-06,
      "loss": 1.0432,
      "step": 199
    },
    {
      "epoch": 0.011007760471132147,
      "grad_norm": 1.188672661781311,
      "learning_rate": 9.999812107868714e-06,
      "loss": 0.982,
      "step": 200
    },
    {
      "epoch": 0.011062799273487809,
      "grad_norm": 1.1011098623275757,
      "learning_rate": 9.999808331261005e-06,
      "loss": 0.9587,
      "step": 201
    },
    {
      "epoch": 0.01111783807584347,
      "grad_norm": 1.1782938241958618,
      "learning_rate": 9.999804517076073e-06,
      "loss": 1.0659,
      "step": 202
    },
    {
      "epoch": 0.01117287687819913,
      "grad_norm": 1.0520117282867432,
      "learning_rate": 9.99980066531395e-06,
      "loss": 1.0056,
      "step": 203
    },
    {
      "epoch": 0.011227915680554791,
      "grad_norm": 1.1584919691085815,
      "learning_rate": 9.999796775974663e-06,
      "loss": 0.9435,
      "step": 204
    },
    {
      "epoch": 0.011282954482910452,
      "grad_norm": 1.2201849222183228,
      "learning_rate": 9.999792849058242e-06,
      "loss": 1.0562,
      "step": 205
    },
    {
      "epoch": 0.011337993285266112,
      "grad_norm": 1.2985976934432983,
      "learning_rate": 9.999788884564715e-06,
      "loss": 1.0126,
      "step": 206
    },
    {
      "epoch": 0.011393032087621774,
      "grad_norm": 0.9926307201385498,
      "learning_rate": 9.999784882494115e-06,
      "loss": 0.7875,
      "step": 207
    },
    {
      "epoch": 0.011448070889977435,
      "grad_norm": 1.103365182876587,
      "learning_rate": 9.99978084284647e-06,
      "loss": 0.9833,
      "step": 208
    },
    {
      "epoch": 0.011503109692333095,
      "grad_norm": 1.1798462867736816,
      "learning_rate": 9.99977676562181e-06,
      "loss": 0.8479,
      "step": 209
    },
    {
      "epoch": 0.011558148494688756,
      "grad_norm": 1.2887194156646729,
      "learning_rate": 9.999772650820168e-06,
      "loss": 0.9606,
      "step": 210
    },
    {
      "epoch": 0.011613187297044416,
      "grad_norm": 1.1120634078979492,
      "learning_rate": 9.99976849844157e-06,
      "loss": 0.9604,
      "step": 211
    },
    {
      "epoch": 0.011668226099400077,
      "grad_norm": 1.1248979568481445,
      "learning_rate": 9.999764308486052e-06,
      "loss": 0.9428,
      "step": 212
    },
    {
      "epoch": 0.011723264901755738,
      "grad_norm": 1.274610161781311,
      "learning_rate": 9.999760080953643e-06,
      "loss": 0.9044,
      "step": 213
    },
    {
      "epoch": 0.011778303704111398,
      "grad_norm": 1.1746865510940552,
      "learning_rate": 9.999755815844377e-06,
      "loss": 0.9114,
      "step": 214
    },
    {
      "epoch": 0.01183334250646706,
      "grad_norm": 1.2531086206436157,
      "learning_rate": 9.999751513158282e-06,
      "loss": 1.0785,
      "step": 215
    },
    {
      "epoch": 0.01188838130882272,
      "grad_norm": 1.0789539813995361,
      "learning_rate": 9.999747172895395e-06,
      "loss": 0.9794,
      "step": 216
    },
    {
      "epoch": 0.01194342011117838,
      "grad_norm": 1.1805329322814941,
      "learning_rate": 9.999742795055746e-06,
      "loss": 0.9602,
      "step": 217
    },
    {
      "epoch": 0.011998458913534042,
      "grad_norm": 2.309329032897949,
      "learning_rate": 9.99973837963937e-06,
      "loss": 0.9482,
      "step": 218
    },
    {
      "epoch": 0.012053497715889702,
      "grad_norm": 1.2379088401794434,
      "learning_rate": 9.999733926646296e-06,
      "loss": 1.0237,
      "step": 219
    },
    {
      "epoch": 0.012108536518245363,
      "grad_norm": 1.1581377983093262,
      "learning_rate": 9.999729436076562e-06,
      "loss": 1.0583,
      "step": 220
    },
    {
      "epoch": 0.012163575320601024,
      "grad_norm": 1.3006727695465088,
      "learning_rate": 9.999724907930199e-06,
      "loss": 0.9581,
      "step": 221
    },
    {
      "epoch": 0.012218614122956684,
      "grad_norm": 1.3215982913970947,
      "learning_rate": 9.999720342207243e-06,
      "loss": 0.9438,
      "step": 222
    },
    {
      "epoch": 0.012273652925312345,
      "grad_norm": 1.1107337474822998,
      "learning_rate": 9.999715738907727e-06,
      "loss": 0.9987,
      "step": 223
    },
    {
      "epoch": 0.012328691727668007,
      "grad_norm": 1.0745457410812378,
      "learning_rate": 9.999711098031685e-06,
      "loss": 0.9637,
      "step": 224
    },
    {
      "epoch": 0.012383730530023666,
      "grad_norm": 1.110861897468567,
      "learning_rate": 9.999706419579154e-06,
      "loss": 1.0225,
      "step": 225
    },
    {
      "epoch": 0.012438769332379328,
      "grad_norm": 1.0755527019500732,
      "learning_rate": 9.999701703550167e-06,
      "loss": 1.0204,
      "step": 226
    },
    {
      "epoch": 0.012493808134734987,
      "grad_norm": 1.1694976091384888,
      "learning_rate": 9.99969694994476e-06,
      "loss": 1.0566,
      "step": 227
    },
    {
      "epoch": 0.012548846937090649,
      "grad_norm": 1.455856442451477,
      "learning_rate": 9.99969215876297e-06,
      "loss": 0.9397,
      "step": 228
    },
    {
      "epoch": 0.01260388573944631,
      "grad_norm": 1.0707073211669922,
      "learning_rate": 9.99968733000483e-06,
      "loss": 0.8286,
      "step": 229
    },
    {
      "epoch": 0.01265892454180197,
      "grad_norm": 1.189548134803772,
      "learning_rate": 9.99968246367038e-06,
      "loss": 0.8762,
      "step": 230
    },
    {
      "epoch": 0.012713963344157631,
      "grad_norm": 1.1439214944839478,
      "learning_rate": 9.999677559759655e-06,
      "loss": 0.9187,
      "step": 231
    },
    {
      "epoch": 0.012769002146513293,
      "grad_norm": 1.2329761981964111,
      "learning_rate": 9.999672618272691e-06,
      "loss": 1.0374,
      "step": 232
    },
    {
      "epoch": 0.012824040948868952,
      "grad_norm": 1.1545134782791138,
      "learning_rate": 9.999667639209527e-06,
      "loss": 0.9343,
      "step": 233
    },
    {
      "epoch": 0.012879079751224614,
      "grad_norm": 1.0946775674819946,
      "learning_rate": 9.999662622570198e-06,
      "loss": 0.9568,
      "step": 234
    },
    {
      "epoch": 0.012934118553580273,
      "grad_norm": 1.2099589109420776,
      "learning_rate": 9.999657568354743e-06,
      "loss": 1.0364,
      "step": 235
    },
    {
      "epoch": 0.012989157355935935,
      "grad_norm": 1.09062922000885,
      "learning_rate": 9.999652476563202e-06,
      "loss": 1.0289,
      "step": 236
    },
    {
      "epoch": 0.013044196158291596,
      "grad_norm": 1.154557228088379,
      "learning_rate": 9.999647347195612e-06,
      "loss": 0.9925,
      "step": 237
    },
    {
      "epoch": 0.013099234960647256,
      "grad_norm": 1.025374174118042,
      "learning_rate": 9.999642180252008e-06,
      "loss": 0.9346,
      "step": 238
    },
    {
      "epoch": 0.013154273763002917,
      "grad_norm": 1.1473641395568848,
      "learning_rate": 9.999636975732433e-06,
      "loss": 1.0244,
      "step": 239
    },
    {
      "epoch": 0.013209312565358578,
      "grad_norm": 1.0421240329742432,
      "learning_rate": 9.999631733636923e-06,
      "loss": 0.9368,
      "step": 240
    },
    {
      "epoch": 0.013264351367714238,
      "grad_norm": 1.1076610088348389,
      "learning_rate": 9.99962645396552e-06,
      "loss": 1.0276,
      "step": 241
    },
    {
      "epoch": 0.0133193901700699,
      "grad_norm": 1.143559455871582,
      "learning_rate": 9.999621136718266e-06,
      "loss": 0.9626,
      "step": 242
    },
    {
      "epoch": 0.01337442897242556,
      "grad_norm": 1.0958378314971924,
      "learning_rate": 9.999615781895195e-06,
      "loss": 1.0254,
      "step": 243
    },
    {
      "epoch": 0.01342946777478122,
      "grad_norm": 1.117688536643982,
      "learning_rate": 9.99961038949635e-06,
      "loss": 0.9685,
      "step": 244
    },
    {
      "epoch": 0.013484506577136882,
      "grad_norm": 1.1645647287368774,
      "learning_rate": 9.999604959521771e-06,
      "loss": 1.0666,
      "step": 245
    },
    {
      "epoch": 0.013539545379492542,
      "grad_norm": 1.1238516569137573,
      "learning_rate": 9.999599491971502e-06,
      "loss": 1.0252,
      "step": 246
    },
    {
      "epoch": 0.013594584181848203,
      "grad_norm": 1.0196914672851562,
      "learning_rate": 9.999593986845579e-06,
      "loss": 0.9389,
      "step": 247
    },
    {
      "epoch": 0.013649622984203864,
      "grad_norm": 1.0231372117996216,
      "learning_rate": 9.999588444144049e-06,
      "loss": 0.8786,
      "step": 248
    },
    {
      "epoch": 0.013704661786559524,
      "grad_norm": 1.2504147291183472,
      "learning_rate": 9.999582863866947e-06,
      "loss": 1.0969,
      "step": 249
    },
    {
      "epoch": 0.013759700588915185,
      "grad_norm": 1.1123549938201904,
      "learning_rate": 9.99957724601432e-06,
      "loss": 0.8833,
      "step": 250
    },
    {
      "epoch": 0.013814739391270847,
      "grad_norm": 1.1068202257156372,
      "learning_rate": 9.999571590586208e-06,
      "loss": 0.9709,
      "step": 251
    },
    {
      "epoch": 0.013869778193626506,
      "grad_norm": 0.9891651272773743,
      "learning_rate": 9.999565897582655e-06,
      "loss": 0.8598,
      "step": 252
    },
    {
      "epoch": 0.013924816995982168,
      "grad_norm": 0.9866491556167603,
      "learning_rate": 9.999560167003703e-06,
      "loss": 0.8101,
      "step": 253
    },
    {
      "epoch": 0.013979855798337828,
      "grad_norm": 1.0862594842910767,
      "learning_rate": 9.999554398849396e-06,
      "loss": 0.9411,
      "step": 254
    },
    {
      "epoch": 0.014034894600693489,
      "grad_norm": 1.1898949146270752,
      "learning_rate": 9.999548593119774e-06,
      "loss": 0.9548,
      "step": 255
    },
    {
      "epoch": 0.01408993340304915,
      "grad_norm": 1.2167880535125732,
      "learning_rate": 9.999542749814886e-06,
      "loss": 1.0302,
      "step": 256
    },
    {
      "epoch": 0.01414497220540481,
      "grad_norm": 1.0784146785736084,
      "learning_rate": 9.999536868934771e-06,
      "loss": 0.8875,
      "step": 257
    },
    {
      "epoch": 0.014200011007760471,
      "grad_norm": 1.1128027439117432,
      "learning_rate": 9.999530950479475e-06,
      "loss": 0.9498,
      "step": 258
    },
    {
      "epoch": 0.014255049810116133,
      "grad_norm": 1.1311595439910889,
      "learning_rate": 9.999524994449044e-06,
      "loss": 0.9035,
      "step": 259
    },
    {
      "epoch": 0.014310088612471792,
      "grad_norm": 1.225615382194519,
      "learning_rate": 9.999519000843521e-06,
      "loss": 1.0104,
      "step": 260
    },
    {
      "epoch": 0.014365127414827454,
      "grad_norm": 1.2347793579101562,
      "learning_rate": 9.99951296966295e-06,
      "loss": 1.0288,
      "step": 261
    },
    {
      "epoch": 0.014420166217183113,
      "grad_norm": 1.1837103366851807,
      "learning_rate": 9.99950690090738e-06,
      "loss": 0.9553,
      "step": 262
    },
    {
      "epoch": 0.014475205019538775,
      "grad_norm": 1.1985397338867188,
      "learning_rate": 9.999500794576852e-06,
      "loss": 0.9561,
      "step": 263
    },
    {
      "epoch": 0.014530243821894436,
      "grad_norm": 1.036928415298462,
      "learning_rate": 9.999494650671418e-06,
      "loss": 0.8906,
      "step": 264
    },
    {
      "epoch": 0.014585282624250096,
      "grad_norm": 1.0797842741012573,
      "learning_rate": 9.999488469191116e-06,
      "loss": 0.8975,
      "step": 265
    },
    {
      "epoch": 0.014640321426605757,
      "grad_norm": 1.0571156740188599,
      "learning_rate": 9.999482250136e-06,
      "loss": 0.9334,
      "step": 266
    },
    {
      "epoch": 0.014695360228961419,
      "grad_norm": 1.2065023183822632,
      "learning_rate": 9.999475993506114e-06,
      "loss": 0.8986,
      "step": 267
    },
    {
      "epoch": 0.014750399031317078,
      "grad_norm": 1.201586127281189,
      "learning_rate": 9.999469699301502e-06,
      "loss": 0.9192,
      "step": 268
    },
    {
      "epoch": 0.01480543783367274,
      "grad_norm": 1.0470168590545654,
      "learning_rate": 9.999463367522216e-06,
      "loss": 0.8604,
      "step": 269
    },
    {
      "epoch": 0.0148604766360284,
      "grad_norm": 1.1142147779464722,
      "learning_rate": 9.9994569981683e-06,
      "loss": 0.9847,
      "step": 270
    },
    {
      "epoch": 0.01491551543838406,
      "grad_norm": 1.0352061986923218,
      "learning_rate": 9.999450591239805e-06,
      "loss": 0.8927,
      "step": 271
    },
    {
      "epoch": 0.014970554240739722,
      "grad_norm": 1.0353184938430786,
      "learning_rate": 9.999444146736779e-06,
      "loss": 0.8435,
      "step": 272
    },
    {
      "epoch": 0.015025593043095382,
      "grad_norm": 1.2091951370239258,
      "learning_rate": 9.999437664659267e-06,
      "loss": 0.8959,
      "step": 273
    },
    {
      "epoch": 0.015080631845451043,
      "grad_norm": 1.006361722946167,
      "learning_rate": 9.999431145007319e-06,
      "loss": 0.8579,
      "step": 274
    },
    {
      "epoch": 0.015135670647806704,
      "grad_norm": 1.1265509128570557,
      "learning_rate": 9.999424587780985e-06,
      "loss": 0.8808,
      "step": 275
    },
    {
      "epoch": 0.015190709450162364,
      "grad_norm": 1.060882568359375,
      "learning_rate": 9.999417992980317e-06,
      "loss": 1.044,
      "step": 276
    },
    {
      "epoch": 0.015245748252518026,
      "grad_norm": 1.0216747522354126,
      "learning_rate": 9.999411360605358e-06,
      "loss": 0.7773,
      "step": 277
    },
    {
      "epoch": 0.015300787054873685,
      "grad_norm": 1.1382462978363037,
      "learning_rate": 9.999404690656163e-06,
      "loss": 0.8954,
      "step": 278
    },
    {
      "epoch": 0.015355825857229347,
      "grad_norm": 1.113815188407898,
      "learning_rate": 9.99939798313278e-06,
      "loss": 0.8143,
      "step": 279
    },
    {
      "epoch": 0.015410864659585008,
      "grad_norm": 1.123530387878418,
      "learning_rate": 9.99939123803526e-06,
      "loss": 0.8872,
      "step": 280
    },
    {
      "epoch": 0.015465903461940668,
      "grad_norm": 1.0873669385910034,
      "learning_rate": 9.999384455363656e-06,
      "loss": 1.008,
      "step": 281
    },
    {
      "epoch": 0.015520942264296329,
      "grad_norm": 1.5956637859344482,
      "learning_rate": 9.999377635118014e-06,
      "loss": 0.9456,
      "step": 282
    },
    {
      "epoch": 0.01557598106665199,
      "grad_norm": 1.1471425294876099,
      "learning_rate": 9.999370777298389e-06,
      "loss": 0.9897,
      "step": 283
    },
    {
      "epoch": 0.01563101986900765,
      "grad_norm": 0.9960193634033203,
      "learning_rate": 9.999363881904831e-06,
      "loss": 0.8196,
      "step": 284
    },
    {
      "epoch": 0.01568605867136331,
      "grad_norm": 1.1033951044082642,
      "learning_rate": 9.999356948937393e-06,
      "loss": 0.879,
      "step": 285
    },
    {
      "epoch": 0.015741097473718973,
      "grad_norm": 1.157765507698059,
      "learning_rate": 9.999349978396126e-06,
      "loss": 1.0116,
      "step": 286
    },
    {
      "epoch": 0.015796136276074634,
      "grad_norm": 1.0472352504730225,
      "learning_rate": 9.999342970281084e-06,
      "loss": 0.8657,
      "step": 287
    },
    {
      "epoch": 0.015851175078430292,
      "grad_norm": 1.1346659660339355,
      "learning_rate": 9.999335924592315e-06,
      "loss": 0.8482,
      "step": 288
    },
    {
      "epoch": 0.015906213880785953,
      "grad_norm": 1.1164487600326538,
      "learning_rate": 9.999328841329879e-06,
      "loss": 1.0542,
      "step": 289
    },
    {
      "epoch": 0.015961252683141615,
      "grad_norm": 1.1890591382980347,
      "learning_rate": 9.999321720493825e-06,
      "loss": 0.9598,
      "step": 290
    },
    {
      "epoch": 0.016016291485497276,
      "grad_norm": 1.0419867038726807,
      "learning_rate": 9.999314562084205e-06,
      "loss": 0.9548,
      "step": 291
    },
    {
      "epoch": 0.016071330287852938,
      "grad_norm": 1.0652042627334595,
      "learning_rate": 9.999307366101077e-06,
      "loss": 0.9359,
      "step": 292
    },
    {
      "epoch": 0.016126369090208596,
      "grad_norm": 1.0166404247283936,
      "learning_rate": 9.999300132544492e-06,
      "loss": 0.9276,
      "step": 293
    },
    {
      "epoch": 0.016181407892564257,
      "grad_norm": 1.1638866662979126,
      "learning_rate": 9.999292861414507e-06,
      "loss": 0.957,
      "step": 294
    },
    {
      "epoch": 0.01623644669491992,
      "grad_norm": 1.5505993366241455,
      "learning_rate": 9.999285552711173e-06,
      "loss": 0.9878,
      "step": 295
    },
    {
      "epoch": 0.01629148549727558,
      "grad_norm": 1.177262783050537,
      "learning_rate": 9.999278206434549e-06,
      "loss": 0.8631,
      "step": 296
    },
    {
      "epoch": 0.01634652429963124,
      "grad_norm": 1.8578168153762817,
      "learning_rate": 9.999270822584687e-06,
      "loss": 0.9684,
      "step": 297
    },
    {
      "epoch": 0.0164015631019869,
      "grad_norm": 1.2617360353469849,
      "learning_rate": 9.999263401161643e-06,
      "loss": 1.014,
      "step": 298
    },
    {
      "epoch": 0.01645660190434256,
      "grad_norm": 0.9740132689476013,
      "learning_rate": 9.999255942165475e-06,
      "loss": 0.8606,
      "step": 299
    },
    {
      "epoch": 0.016511640706698222,
      "grad_norm": 0.9821745753288269,
      "learning_rate": 9.999248445596238e-06,
      "loss": 0.8241,
      "step": 300
    },
    {
      "epoch": 0.016566679509053883,
      "grad_norm": 1.0200445652008057,
      "learning_rate": 9.999240911453986e-06,
      "loss": 0.8256,
      "step": 301
    },
    {
      "epoch": 0.016621718311409545,
      "grad_norm": 1.4100390672683716,
      "learning_rate": 9.999233339738779e-06,
      "loss": 0.9057,
      "step": 302
    },
    {
      "epoch": 0.016676757113765206,
      "grad_norm": 1.056544303894043,
      "learning_rate": 9.99922573045067e-06,
      "loss": 1.0808,
      "step": 303
    },
    {
      "epoch": 0.016731795916120864,
      "grad_norm": 0.9271026253700256,
      "learning_rate": 9.99921808358972e-06,
      "loss": 0.878,
      "step": 304
    },
    {
      "epoch": 0.016786834718476525,
      "grad_norm": 0.9864157438278198,
      "learning_rate": 9.999210399155987e-06,
      "loss": 0.9198,
      "step": 305
    },
    {
      "epoch": 0.016841873520832187,
      "grad_norm": 1.093995451927185,
      "learning_rate": 9.999202677149525e-06,
      "loss": 0.9794,
      "step": 306
    },
    {
      "epoch": 0.016896912323187848,
      "grad_norm": 0.9717912077903748,
      "learning_rate": 9.999194917570395e-06,
      "loss": 0.8764,
      "step": 307
    },
    {
      "epoch": 0.01695195112554351,
      "grad_norm": 1.0026428699493408,
      "learning_rate": 9.999187120418653e-06,
      "loss": 0.8526,
      "step": 308
    },
    {
      "epoch": 0.017006989927899167,
      "grad_norm": 1.122870922088623,
      "learning_rate": 9.999179285694359e-06,
      "loss": 0.9773,
      "step": 309
    },
    {
      "epoch": 0.01706202873025483,
      "grad_norm": 1.0522836446762085,
      "learning_rate": 9.999171413397572e-06,
      "loss": 1.0183,
      "step": 310
    },
    {
      "epoch": 0.01711706753261049,
      "grad_norm": 0.9303658604621887,
      "learning_rate": 9.99916350352835e-06,
      "loss": 0.8402,
      "step": 311
    },
    {
      "epoch": 0.01717210633496615,
      "grad_norm": 0.9606096148490906,
      "learning_rate": 9.999155556086755e-06,
      "loss": 0.9692,
      "step": 312
    },
    {
      "epoch": 0.017227145137321813,
      "grad_norm": 1.176992416381836,
      "learning_rate": 9.999147571072844e-06,
      "loss": 0.8172,
      "step": 313
    },
    {
      "epoch": 0.017282183939677474,
      "grad_norm": 1.1948801279067993,
      "learning_rate": 9.999139548486678e-06,
      "loss": 1.0205,
      "step": 314
    },
    {
      "epoch": 0.017337222742033132,
      "grad_norm": 1.0064897537231445,
      "learning_rate": 9.999131488328318e-06,
      "loss": 0.9479,
      "step": 315
    },
    {
      "epoch": 0.017392261544388794,
      "grad_norm": 1.048242449760437,
      "learning_rate": 9.999123390597822e-06,
      "loss": 0.9862,
      "step": 316
    },
    {
      "epoch": 0.017447300346744455,
      "grad_norm": 1.12875497341156,
      "learning_rate": 9.999115255295256e-06,
      "loss": 0.9743,
      "step": 317
    },
    {
      "epoch": 0.017502339149100116,
      "grad_norm": 1.0607460737228394,
      "learning_rate": 9.999107082420674e-06,
      "loss": 0.8878,
      "step": 318
    },
    {
      "epoch": 0.017557377951455778,
      "grad_norm": 1.1480191946029663,
      "learning_rate": 9.999098871974144e-06,
      "loss": 0.8769,
      "step": 319
    },
    {
      "epoch": 0.017612416753811436,
      "grad_norm": 1.1150004863739014,
      "learning_rate": 9.999090623955724e-06,
      "loss": 0.8615,
      "step": 320
    },
    {
      "epoch": 0.017667455556167097,
      "grad_norm": 1.137839913368225,
      "learning_rate": 9.999082338365478e-06,
      "loss": 0.9703,
      "step": 321
    },
    {
      "epoch": 0.01772249435852276,
      "grad_norm": 1.0883489847183228,
      "learning_rate": 9.999074015203467e-06,
      "loss": 0.9273,
      "step": 322
    },
    {
      "epoch": 0.01777753316087842,
      "grad_norm": 1.0999557971954346,
      "learning_rate": 9.999065654469752e-06,
      "loss": 0.9605,
      "step": 323
    },
    {
      "epoch": 0.01783257196323408,
      "grad_norm": 0.9911689758300781,
      "learning_rate": 9.999057256164401e-06,
      "loss": 0.9117,
      "step": 324
    },
    {
      "epoch": 0.01788761076558974,
      "grad_norm": 1.040933609008789,
      "learning_rate": 9.999048820287472e-06,
      "loss": 0.9229,
      "step": 325
    },
    {
      "epoch": 0.0179426495679454,
      "grad_norm": 1.4341392517089844,
      "learning_rate": 9.999040346839031e-06,
      "loss": 1.0718,
      "step": 326
    },
    {
      "epoch": 0.017997688370301062,
      "grad_norm": 1.0246332883834839,
      "learning_rate": 9.99903183581914e-06,
      "loss": 0.9617,
      "step": 327
    },
    {
      "epoch": 0.018052727172656723,
      "grad_norm": 10.162322998046875,
      "learning_rate": 9.999023287227863e-06,
      "loss": 1.0391,
      "step": 328
    },
    {
      "epoch": 0.018107765975012385,
      "grad_norm": 1.3370027542114258,
      "learning_rate": 9.999014701065266e-06,
      "loss": 1.0211,
      "step": 329
    },
    {
      "epoch": 0.018162804777368046,
      "grad_norm": 1.0146219730377197,
      "learning_rate": 9.999006077331413e-06,
      "loss": 0.8611,
      "step": 330
    },
    {
      "epoch": 0.018217843579723704,
      "grad_norm": 1.0899269580841064,
      "learning_rate": 9.998997416026368e-06,
      "loss": 0.9209,
      "step": 331
    },
    {
      "epoch": 0.018272882382079365,
      "grad_norm": 1.1343204975128174,
      "learning_rate": 9.998988717150198e-06,
      "loss": 0.9405,
      "step": 332
    },
    {
      "epoch": 0.018327921184435027,
      "grad_norm": 1.2308380603790283,
      "learning_rate": 9.998979980702965e-06,
      "loss": 0.9579,
      "step": 333
    },
    {
      "epoch": 0.018382959986790688,
      "grad_norm": 1.1433519124984741,
      "learning_rate": 9.998971206684737e-06,
      "loss": 1.0045,
      "step": 334
    },
    {
      "epoch": 0.01843799878914635,
      "grad_norm": 1.0585781335830688,
      "learning_rate": 9.99896239509558e-06,
      "loss": 0.9171,
      "step": 335
    },
    {
      "epoch": 0.018493037591502007,
      "grad_norm": 1.2735164165496826,
      "learning_rate": 9.99895354593556e-06,
      "loss": 1.1001,
      "step": 336
    },
    {
      "epoch": 0.01854807639385767,
      "grad_norm": 1.2905755043029785,
      "learning_rate": 9.998944659204744e-06,
      "loss": 1.0294,
      "step": 337
    },
    {
      "epoch": 0.01860311519621333,
      "grad_norm": 1.1442075967788696,
      "learning_rate": 9.998935734903198e-06,
      "loss": 0.9385,
      "step": 338
    },
    {
      "epoch": 0.01865815399856899,
      "grad_norm": 1.1005232334136963,
      "learning_rate": 9.998926773030987e-06,
      "loss": 1.026,
      "step": 339
    },
    {
      "epoch": 0.018713192800924653,
      "grad_norm": 1.2770785093307495,
      "learning_rate": 9.998917773588182e-06,
      "loss": 1.0015,
      "step": 340
    },
    {
      "epoch": 0.01876823160328031,
      "grad_norm": 1.0963070392608643,
      "learning_rate": 9.998908736574849e-06,
      "loss": 0.9347,
      "step": 341
    },
    {
      "epoch": 0.018823270405635972,
      "grad_norm": 1.10364830493927,
      "learning_rate": 9.998899661991055e-06,
      "loss": 0.869,
      "step": 342
    },
    {
      "epoch": 0.018878309207991634,
      "grad_norm": 1.0364975929260254,
      "learning_rate": 9.99889054983687e-06,
      "loss": 0.9855,
      "step": 343
    },
    {
      "epoch": 0.018933348010347295,
      "grad_norm": 1.104702115058899,
      "learning_rate": 9.998881400112362e-06,
      "loss": 0.9555,
      "step": 344
    },
    {
      "epoch": 0.018988386812702956,
      "grad_norm": 0.9957441687583923,
      "learning_rate": 9.998872212817599e-06,
      "loss": 0.9634,
      "step": 345
    },
    {
      "epoch": 0.019043425615058618,
      "grad_norm": 1.262271523475647,
      "learning_rate": 9.998862987952651e-06,
      "loss": 1.0133,
      "step": 346
    },
    {
      "epoch": 0.019098464417414276,
      "grad_norm": 1.2075226306915283,
      "learning_rate": 9.998853725517587e-06,
      "loss": 1.0588,
      "step": 347
    },
    {
      "epoch": 0.019153503219769937,
      "grad_norm": 1.0609898567199707,
      "learning_rate": 9.998844425512477e-06,
      "loss": 0.9952,
      "step": 348
    },
    {
      "epoch": 0.0192085420221256,
      "grad_norm": 1.1930195093154907,
      "learning_rate": 9.998835087937389e-06,
      "loss": 0.9617,
      "step": 349
    },
    {
      "epoch": 0.01926358082448126,
      "grad_norm": 1.2359932661056519,
      "learning_rate": 9.998825712792396e-06,
      "loss": 0.8768,
      "step": 350
    },
    {
      "epoch": 0.01931861962683692,
      "grad_norm": 0.9984115362167358,
      "learning_rate": 9.998816300077566e-06,
      "loss": 0.8205,
      "step": 351
    },
    {
      "epoch": 0.01937365842919258,
      "grad_norm": 1.6853677034378052,
      "learning_rate": 9.998806849792972e-06,
      "loss": 0.9066,
      "step": 352
    },
    {
      "epoch": 0.01942869723154824,
      "grad_norm": 1.2869856357574463,
      "learning_rate": 9.998797361938683e-06,
      "loss": 1.0054,
      "step": 353
    },
    {
      "epoch": 0.019483736033903902,
      "grad_norm": 1.2791584730148315,
      "learning_rate": 9.99878783651477e-06,
      "loss": 0.7627,
      "step": 354
    },
    {
      "epoch": 0.019538774836259563,
      "grad_norm": 1.0795867443084717,
      "learning_rate": 9.998778273521307e-06,
      "loss": 0.9343,
      "step": 355
    },
    {
      "epoch": 0.019593813638615225,
      "grad_norm": 1.0926088094711304,
      "learning_rate": 9.998768672958365e-06,
      "loss": 0.943,
      "step": 356
    },
    {
      "epoch": 0.019648852440970886,
      "grad_norm": 1.0530847311019897,
      "learning_rate": 9.998759034826015e-06,
      "loss": 0.9656,
      "step": 357
    },
    {
      "epoch": 0.019703891243326544,
      "grad_norm": 1.1793400049209595,
      "learning_rate": 9.99874935912433e-06,
      "loss": 0.9799,
      "step": 358
    },
    {
      "epoch": 0.019758930045682205,
      "grad_norm": 1.0726191997528076,
      "learning_rate": 9.998739645853383e-06,
      "loss": 0.8739,
      "step": 359
    },
    {
      "epoch": 0.019813968848037867,
      "grad_norm": 1.0488981008529663,
      "learning_rate": 9.998729895013246e-06,
      "loss": 0.8986,
      "step": 360
    },
    {
      "epoch": 0.019869007650393528,
      "grad_norm": 1.8267477750778198,
      "learning_rate": 9.998720106603993e-06,
      "loss": 0.9175,
      "step": 361
    },
    {
      "epoch": 0.01992404645274919,
      "grad_norm": 0.9868306517601013,
      "learning_rate": 9.9987102806257e-06,
      "loss": 0.9609,
      "step": 362
    },
    {
      "epoch": 0.019979085255104848,
      "grad_norm": 1.0171183347702026,
      "learning_rate": 9.998700417078438e-06,
      "loss": 0.8904,
      "step": 363
    },
    {
      "epoch": 0.02003412405746051,
      "grad_norm": 0.9800812602043152,
      "learning_rate": 9.998690515962282e-06,
      "loss": 0.8344,
      "step": 364
    },
    {
      "epoch": 0.02008916285981617,
      "grad_norm": 1.024707317352295,
      "learning_rate": 9.998680577277304e-06,
      "loss": 0.9026,
      "step": 365
    },
    {
      "epoch": 0.02014420166217183,
      "grad_norm": 1.1056619882583618,
      "learning_rate": 9.998670601023584e-06,
      "loss": 1.017,
      "step": 366
    },
    {
      "epoch": 0.020199240464527493,
      "grad_norm": 1.0555908679962158,
      "learning_rate": 9.998660587201191e-06,
      "loss": 0.9627,
      "step": 367
    },
    {
      "epoch": 0.02025427926688315,
      "grad_norm": 0.9502031803131104,
      "learning_rate": 9.998650535810204e-06,
      "loss": 0.935,
      "step": 368
    },
    {
      "epoch": 0.020309318069238812,
      "grad_norm": 1.0355613231658936,
      "learning_rate": 9.998640446850699e-06,
      "loss": 0.9946,
      "step": 369
    },
    {
      "epoch": 0.020364356871594474,
      "grad_norm": 0.9906355142593384,
      "learning_rate": 9.99863032032275e-06,
      "loss": 0.9389,
      "step": 370
    },
    {
      "epoch": 0.020419395673950135,
      "grad_norm": 0.9483911395072937,
      "learning_rate": 9.99862015622643e-06,
      "loss": 0.979,
      "step": 371
    },
    {
      "epoch": 0.020474434476305797,
      "grad_norm": 0.9769986271858215,
      "learning_rate": 9.998609954561822e-06,
      "loss": 0.8972,
      "step": 372
    },
    {
      "epoch": 0.020529473278661458,
      "grad_norm": 1.1682699918746948,
      "learning_rate": 9.998599715329e-06,
      "loss": 0.943,
      "step": 373
    },
    {
      "epoch": 0.020584512081017116,
      "grad_norm": 1.007912516593933,
      "learning_rate": 9.99858943852804e-06,
      "loss": 0.8825,
      "step": 374
    },
    {
      "epoch": 0.020639550883372777,
      "grad_norm": 0.9788785576820374,
      "learning_rate": 9.99857912415902e-06,
      "loss": 0.9667,
      "step": 375
    },
    {
      "epoch": 0.02069458968572844,
      "grad_norm": 1.0804275274276733,
      "learning_rate": 9.998568772222017e-06,
      "loss": 1.0026,
      "step": 376
    },
    {
      "epoch": 0.0207496284880841,
      "grad_norm": 1.0859237909317017,
      "learning_rate": 9.998558382717109e-06,
      "loss": 0.9592,
      "step": 377
    },
    {
      "epoch": 0.02080466729043976,
      "grad_norm": 1.2925337553024292,
      "learning_rate": 9.998547955644373e-06,
      "loss": 0.9067,
      "step": 378
    },
    {
      "epoch": 0.02085970609279542,
      "grad_norm": 0.9853373765945435,
      "learning_rate": 9.99853749100389e-06,
      "loss": 0.9538,
      "step": 379
    },
    {
      "epoch": 0.02091474489515108,
      "grad_norm": 1.0461076498031616,
      "learning_rate": 9.998526988795738e-06,
      "loss": 0.9261,
      "step": 380
    },
    {
      "epoch": 0.020969783697506742,
      "grad_norm": 1.024559497833252,
      "learning_rate": 9.998516449019995e-06,
      "loss": 0.9117,
      "step": 381
    },
    {
      "epoch": 0.021024822499862404,
      "grad_norm": 1.1474825143814087,
      "learning_rate": 9.998505871676739e-06,
      "loss": 1.0177,
      "step": 382
    },
    {
      "epoch": 0.021079861302218065,
      "grad_norm": 0.9587596654891968,
      "learning_rate": 9.998495256766051e-06,
      "loss": 0.8809,
      "step": 383
    },
    {
      "epoch": 0.021134900104573723,
      "grad_norm": 0.9505122303962708,
      "learning_rate": 9.998484604288013e-06,
      "loss": 0.9266,
      "step": 384
    },
    {
      "epoch": 0.021189938906929384,
      "grad_norm": 0.9625647664070129,
      "learning_rate": 9.9984739142427e-06,
      "loss": 0.9073,
      "step": 385
    },
    {
      "epoch": 0.021244977709285046,
      "grad_norm": 0.9650934338569641,
      "learning_rate": 9.998463186630196e-06,
      "loss": 0.9042,
      "step": 386
    },
    {
      "epoch": 0.021300016511640707,
      "grad_norm": 1.0289491415023804,
      "learning_rate": 9.99845242145058e-06,
      "loss": 0.929,
      "step": 387
    },
    {
      "epoch": 0.02135505531399637,
      "grad_norm": 0.9543869495391846,
      "learning_rate": 9.998441618703935e-06,
      "loss": 0.9406,
      "step": 388
    },
    {
      "epoch": 0.02141009411635203,
      "grad_norm": 0.9276942610740662,
      "learning_rate": 9.99843077839034e-06,
      "loss": 0.8982,
      "step": 389
    },
    {
      "epoch": 0.021465132918707688,
      "grad_norm": 0.9264664053916931,
      "learning_rate": 9.998419900509877e-06,
      "loss": 0.7255,
      "step": 390
    },
    {
      "epoch": 0.02152017172106335,
      "grad_norm": 0.9961187243461609,
      "learning_rate": 9.998408985062628e-06,
      "loss": 0.9826,
      "step": 391
    },
    {
      "epoch": 0.02157521052341901,
      "grad_norm": 0.966596245765686,
      "learning_rate": 9.998398032048676e-06,
      "loss": 0.8159,
      "step": 392
    },
    {
      "epoch": 0.021630249325774672,
      "grad_norm": 1.1336095333099365,
      "learning_rate": 9.998387041468102e-06,
      "loss": 0.9289,
      "step": 393
    },
    {
      "epoch": 0.021685288128130333,
      "grad_norm": 1.0453619956970215,
      "learning_rate": 9.998376013320989e-06,
      "loss": 0.8816,
      "step": 394
    },
    {
      "epoch": 0.02174032693048599,
      "grad_norm": 0.8961821794509888,
      "learning_rate": 9.998364947607419e-06,
      "loss": 0.871,
      "step": 395
    },
    {
      "epoch": 0.021795365732841653,
      "grad_norm": 1.3420332670211792,
      "learning_rate": 9.998353844327477e-06,
      "loss": 0.9338,
      "step": 396
    },
    {
      "epoch": 0.021850404535197314,
      "grad_norm": 0.9635335206985474,
      "learning_rate": 9.998342703481246e-06,
      "loss": 0.9592,
      "step": 397
    },
    {
      "epoch": 0.021905443337552975,
      "grad_norm": 1.3322341442108154,
      "learning_rate": 9.998331525068807e-06,
      "loss": 1.0974,
      "step": 398
    },
    {
      "epoch": 0.021960482139908637,
      "grad_norm": 1.017220377922058,
      "learning_rate": 9.998320309090247e-06,
      "loss": 0.9827,
      "step": 399
    },
    {
      "epoch": 0.022015520942264295,
      "grad_norm": 1.0080329179763794,
      "learning_rate": 9.99830905554565e-06,
      "loss": 0.877,
      "step": 400
    },
    {
      "epoch": 0.022070559744619956,
      "grad_norm": 0.9883211255073547,
      "learning_rate": 9.998297764435101e-06,
      "loss": 0.9625,
      "step": 401
    },
    {
      "epoch": 0.022125598546975617,
      "grad_norm": 1.0948412418365479,
      "learning_rate": 9.998286435758684e-06,
      "loss": 0.9058,
      "step": 402
    },
    {
      "epoch": 0.02218063734933128,
      "grad_norm": 0.9402000308036804,
      "learning_rate": 9.998275069516482e-06,
      "loss": 0.8882,
      "step": 403
    },
    {
      "epoch": 0.02223567615168694,
      "grad_norm": 0.9858806133270264,
      "learning_rate": 9.998263665708583e-06,
      "loss": 0.9086,
      "step": 404
    },
    {
      "epoch": 0.0222907149540426,
      "grad_norm": 1.0556131601333618,
      "learning_rate": 9.998252224335073e-06,
      "loss": 0.9583,
      "step": 405
    },
    {
      "epoch": 0.02234575375639826,
      "grad_norm": 1.092766284942627,
      "learning_rate": 9.998240745396037e-06,
      "loss": 0.9124,
      "step": 406
    },
    {
      "epoch": 0.02240079255875392,
      "grad_norm": 1.1902250051498413,
      "learning_rate": 9.998229228891563e-06,
      "loss": 1.0566,
      "step": 407
    },
    {
      "epoch": 0.022455831361109582,
      "grad_norm": 1.067906141281128,
      "learning_rate": 9.998217674821734e-06,
      "loss": 0.9823,
      "step": 408
    },
    {
      "epoch": 0.022510870163465244,
      "grad_norm": 1.0051710605621338,
      "learning_rate": 9.998206083186638e-06,
      "loss": 0.9141,
      "step": 409
    },
    {
      "epoch": 0.022565908965820905,
      "grad_norm": 1.046412467956543,
      "learning_rate": 9.998194453986367e-06,
      "loss": 0.9439,
      "step": 410
    },
    {
      "epoch": 0.022620947768176563,
      "grad_norm": 1.1103553771972656,
      "learning_rate": 9.998182787221e-06,
      "loss": 0.9494,
      "step": 411
    },
    {
      "epoch": 0.022675986570532224,
      "grad_norm": 1.0508466958999634,
      "learning_rate": 9.998171082890632e-06,
      "loss": 0.9202,
      "step": 412
    },
    {
      "epoch": 0.022731025372887886,
      "grad_norm": 1.1364226341247559,
      "learning_rate": 9.998159340995347e-06,
      "loss": 0.9859,
      "step": 413
    },
    {
      "epoch": 0.022786064175243547,
      "grad_norm": 1.2073607444763184,
      "learning_rate": 9.998147561535234e-06,
      "loss": 0.8883,
      "step": 414
    },
    {
      "epoch": 0.02284110297759921,
      "grad_norm": 1.0657012462615967,
      "learning_rate": 9.998135744510384e-06,
      "loss": 0.8321,
      "step": 415
    },
    {
      "epoch": 0.02289614177995487,
      "grad_norm": 1.0101548433303833,
      "learning_rate": 9.998123889920881e-06,
      "loss": 0.9374,
      "step": 416
    },
    {
      "epoch": 0.022951180582310528,
      "grad_norm": 1.057455062866211,
      "learning_rate": 9.998111997766817e-06,
      "loss": 0.8831,
      "step": 417
    },
    {
      "epoch": 0.02300621938466619,
      "grad_norm": 1.206092357635498,
      "learning_rate": 9.998100068048282e-06,
      "loss": 0.8812,
      "step": 418
    },
    {
      "epoch": 0.02306125818702185,
      "grad_norm": 1.0709773302078247,
      "learning_rate": 9.998088100765366e-06,
      "loss": 0.9486,
      "step": 419
    },
    {
      "epoch": 0.023116296989377512,
      "grad_norm": 1.066469669342041,
      "learning_rate": 9.998076095918156e-06,
      "loss": 1.0229,
      "step": 420
    },
    {
      "epoch": 0.023171335791733173,
      "grad_norm": 1.0443583726882935,
      "learning_rate": 9.998064053506744e-06,
      "loss": 0.8615,
      "step": 421
    },
    {
      "epoch": 0.02322637459408883,
      "grad_norm": 1.103096842765808,
      "learning_rate": 9.99805197353122e-06,
      "loss": 0.9909,
      "step": 422
    },
    {
      "epoch": 0.023281413396444493,
      "grad_norm": 0.9804643392562866,
      "learning_rate": 9.998039855991677e-06,
      "loss": 0.9214,
      "step": 423
    },
    {
      "epoch": 0.023336452198800154,
      "grad_norm": 0.9880676865577698,
      "learning_rate": 9.998027700888202e-06,
      "loss": 0.9345,
      "step": 424
    },
    {
      "epoch": 0.023391491001155815,
      "grad_norm": 0.9633826017379761,
      "learning_rate": 9.99801550822089e-06,
      "loss": 0.9897,
      "step": 425
    },
    {
      "epoch": 0.023446529803511477,
      "grad_norm": 1.0159331560134888,
      "learning_rate": 9.998003277989831e-06,
      "loss": 0.9385,
      "step": 426
    },
    {
      "epoch": 0.023501568605867135,
      "grad_norm": 1.009667158126831,
      "learning_rate": 9.99799101019512e-06,
      "loss": 0.9013,
      "step": 427
    },
    {
      "epoch": 0.023556607408222796,
      "grad_norm": 0.9478578567504883,
      "learning_rate": 9.997978704836842e-06,
      "loss": 0.8775,
      "step": 428
    },
    {
      "epoch": 0.023611646210578457,
      "grad_norm": 1.013181447982788,
      "learning_rate": 9.997966361915096e-06,
      "loss": 0.8797,
      "step": 429
    },
    {
      "epoch": 0.02366668501293412,
      "grad_norm": 1.0337481498718262,
      "learning_rate": 9.997953981429974e-06,
      "loss": 1.0047,
      "step": 430
    },
    {
      "epoch": 0.02372172381528978,
      "grad_norm": 0.9423721432685852,
      "learning_rate": 9.997941563381566e-06,
      "loss": 0.8639,
      "step": 431
    },
    {
      "epoch": 0.02377676261764544,
      "grad_norm": 1.100492000579834,
      "learning_rate": 9.997929107769968e-06,
      "loss": 1.0022,
      "step": 432
    },
    {
      "epoch": 0.0238318014200011,
      "grad_norm": 1.1232364177703857,
      "learning_rate": 9.997916614595272e-06,
      "loss": 0.9145,
      "step": 433
    },
    {
      "epoch": 0.02388684022235676,
      "grad_norm": 0.9466833472251892,
      "learning_rate": 9.997904083857572e-06,
      "loss": 0.9397,
      "step": 434
    },
    {
      "epoch": 0.023941879024712422,
      "grad_norm": 0.9514566659927368,
      "learning_rate": 9.997891515556963e-06,
      "loss": 0.8025,
      "step": 435
    },
    {
      "epoch": 0.023996917827068084,
      "grad_norm": 0.9292222261428833,
      "learning_rate": 9.997878909693539e-06,
      "loss": 0.7739,
      "step": 436
    },
    {
      "epoch": 0.024051956629423745,
      "grad_norm": 1.1049963235855103,
      "learning_rate": 9.997866266267397e-06,
      "loss": 0.9439,
      "step": 437
    },
    {
      "epoch": 0.024106995431779403,
      "grad_norm": 1.0938019752502441,
      "learning_rate": 9.997853585278627e-06,
      "loss": 0.9479,
      "step": 438
    },
    {
      "epoch": 0.024162034234135064,
      "grad_norm": 1.0423611402511597,
      "learning_rate": 9.997840866727331e-06,
      "loss": 0.9309,
      "step": 439
    },
    {
      "epoch": 0.024217073036490726,
      "grad_norm": 1.0584756135940552,
      "learning_rate": 9.997828110613598e-06,
      "loss": 1.0218,
      "step": 440
    },
    {
      "epoch": 0.024272111838846387,
      "grad_norm": 0.9986408948898315,
      "learning_rate": 9.997815316937527e-06,
      "loss": 0.9734,
      "step": 441
    },
    {
      "epoch": 0.02432715064120205,
      "grad_norm": 0.9680983424186707,
      "learning_rate": 9.997802485699215e-06,
      "loss": 0.9286,
      "step": 442
    },
    {
      "epoch": 0.024382189443557706,
      "grad_norm": 1.2231700420379639,
      "learning_rate": 9.997789616898757e-06,
      "loss": 0.8083,
      "step": 443
    },
    {
      "epoch": 0.024437228245913368,
      "grad_norm": 1.0064021348953247,
      "learning_rate": 9.99777671053625e-06,
      "loss": 0.9161,
      "step": 444
    },
    {
      "epoch": 0.02449226704826903,
      "grad_norm": 0.9658541679382324,
      "learning_rate": 9.99776376661179e-06,
      "loss": 0.8027,
      "step": 445
    },
    {
      "epoch": 0.02454730585062469,
      "grad_norm": 0.9440343379974365,
      "learning_rate": 9.997750785125477e-06,
      "loss": 0.9124,
      "step": 446
    },
    {
      "epoch": 0.024602344652980352,
      "grad_norm": 0.998792827129364,
      "learning_rate": 9.997737766077404e-06,
      "loss": 0.8699,
      "step": 447
    },
    {
      "epoch": 0.024657383455336013,
      "grad_norm": 1.430880069732666,
      "learning_rate": 9.997724709467676e-06,
      "loss": 0.9158,
      "step": 448
    },
    {
      "epoch": 0.02471242225769167,
      "grad_norm": 0.9737820029258728,
      "learning_rate": 9.997711615296384e-06,
      "loss": 0.9496,
      "step": 449
    },
    {
      "epoch": 0.024767461060047333,
      "grad_norm": 0.9710075855255127,
      "learning_rate": 9.997698483563629e-06,
      "loss": 0.8714,
      "step": 450
    },
    {
      "epoch": 0.024822499862402994,
      "grad_norm": 1.5286253690719604,
      "learning_rate": 9.997685314269511e-06,
      "loss": 0.8421,
      "step": 451
    },
    {
      "epoch": 0.024877538664758655,
      "grad_norm": 1.0269445180892944,
      "learning_rate": 9.99767210741413e-06,
      "loss": 1.0131,
      "step": 452
    },
    {
      "epoch": 0.024932577467114317,
      "grad_norm": 0.9780508279800415,
      "learning_rate": 9.99765886299758e-06,
      "loss": 0.9897,
      "step": 453
    },
    {
      "epoch": 0.024987616269469975,
      "grad_norm": 0.998332679271698,
      "learning_rate": 9.997645581019965e-06,
      "loss": 0.9647,
      "step": 454
    },
    {
      "epoch": 0.025042655071825636,
      "grad_norm": 1.7062602043151855,
      "learning_rate": 9.997632261481383e-06,
      "loss": 1.0729,
      "step": 455
    },
    {
      "epoch": 0.025097693874181298,
      "grad_norm": 0.9793694615364075,
      "learning_rate": 9.997618904381936e-06,
      "loss": 0.9556,
      "step": 456
    },
    {
      "epoch": 0.02515273267653696,
      "grad_norm": 1.0183895826339722,
      "learning_rate": 9.997605509721721e-06,
      "loss": 0.9194,
      "step": 457
    },
    {
      "epoch": 0.02520777147889262,
      "grad_norm": 1.0288400650024414,
      "learning_rate": 9.997592077500844e-06,
      "loss": 0.955,
      "step": 458
    },
    {
      "epoch": 0.025262810281248282,
      "grad_norm": 0.9551253914833069,
      "learning_rate": 9.997578607719401e-06,
      "loss": 0.8498,
      "step": 459
    },
    {
      "epoch": 0.02531784908360394,
      "grad_norm": 0.9648008942604065,
      "learning_rate": 9.997565100377494e-06,
      "loss": 0.9306,
      "step": 460
    },
    {
      "epoch": 0.0253728878859596,
      "grad_norm": 0.9206677675247192,
      "learning_rate": 9.997551555475225e-06,
      "loss": 0.7874,
      "step": 461
    },
    {
      "epoch": 0.025427926688315262,
      "grad_norm": 1.0479545593261719,
      "learning_rate": 9.997537973012698e-06,
      "loss": 0.9201,
      "step": 462
    },
    {
      "epoch": 0.025482965490670924,
      "grad_norm": 1.0329946279525757,
      "learning_rate": 9.997524352990013e-06,
      "loss": 0.9577,
      "step": 463
    },
    {
      "epoch": 0.025538004293026585,
      "grad_norm": 1.1177828311920166,
      "learning_rate": 9.997510695407273e-06,
      "loss": 1.0041,
      "step": 464
    },
    {
      "epoch": 0.025593043095382243,
      "grad_norm": 1.0351577997207642,
      "learning_rate": 9.99749700026458e-06,
      "loss": 0.9952,
      "step": 465
    },
    {
      "epoch": 0.025648081897737905,
      "grad_norm": 0.905274510383606,
      "learning_rate": 9.997483267562035e-06,
      "loss": 0.8185,
      "step": 466
    },
    {
      "epoch": 0.025703120700093566,
      "grad_norm": 1.0749776363372803,
      "learning_rate": 9.997469497299747e-06,
      "loss": 1.0611,
      "step": 467
    },
    {
      "epoch": 0.025758159502449227,
      "grad_norm": 0.8972223401069641,
      "learning_rate": 9.997455689477815e-06,
      "loss": 0.8994,
      "step": 468
    },
    {
      "epoch": 0.02581319830480489,
      "grad_norm": 1.0669914484024048,
      "learning_rate": 9.997441844096342e-06,
      "loss": 1.06,
      "step": 469
    },
    {
      "epoch": 0.025868237107160547,
      "grad_norm": 1.0431914329528809,
      "learning_rate": 9.997427961155435e-06,
      "loss": 0.8657,
      "step": 470
    },
    {
      "epoch": 0.025923275909516208,
      "grad_norm": 0.9609962701797485,
      "learning_rate": 9.997414040655198e-06,
      "loss": 0.8864,
      "step": 471
    },
    {
      "epoch": 0.02597831471187187,
      "grad_norm": 1.0829721689224243,
      "learning_rate": 9.997400082595735e-06,
      "loss": 0.9221,
      "step": 472
    },
    {
      "epoch": 0.02603335351422753,
      "grad_norm": 0.992082953453064,
      "learning_rate": 9.99738608697715e-06,
      "loss": 0.8455,
      "step": 473
    },
    {
      "epoch": 0.026088392316583192,
      "grad_norm": 1.0486301183700562,
      "learning_rate": 9.997372053799547e-06,
      "loss": 0.8729,
      "step": 474
    },
    {
      "epoch": 0.026143431118938854,
      "grad_norm": 1.0328491926193237,
      "learning_rate": 9.997357983063036e-06,
      "loss": 0.8788,
      "step": 475
    },
    {
      "epoch": 0.02619846992129451,
      "grad_norm": 0.963333249092102,
      "learning_rate": 9.997343874767719e-06,
      "loss": 0.892,
      "step": 476
    },
    {
      "epoch": 0.026253508723650173,
      "grad_norm": 1.1606497764587402,
      "learning_rate": 9.997329728913704e-06,
      "loss": 0.9984,
      "step": 477
    },
    {
      "epoch": 0.026308547526005834,
      "grad_norm": 1.241650104522705,
      "learning_rate": 9.997315545501096e-06,
      "loss": 0.946,
      "step": 478
    },
    {
      "epoch": 0.026363586328361496,
      "grad_norm": 1.008004069328308,
      "learning_rate": 9.99730132453e-06,
      "loss": 0.849,
      "step": 479
    },
    {
      "epoch": 0.026418625130717157,
      "grad_norm": 0.9883478879928589,
      "learning_rate": 9.997287066000527e-06,
      "loss": 0.9478,
      "step": 480
    },
    {
      "epoch": 0.026473663933072815,
      "grad_norm": 1.0224446058273315,
      "learning_rate": 9.997272769912783e-06,
      "loss": 1.0318,
      "step": 481
    },
    {
      "epoch": 0.026528702735428476,
      "grad_norm": 0.9412569403648376,
      "learning_rate": 9.997258436266874e-06,
      "loss": 0.9119,
      "step": 482
    },
    {
      "epoch": 0.026583741537784138,
      "grad_norm": 0.9214537739753723,
      "learning_rate": 9.997244065062906e-06,
      "loss": 0.8785,
      "step": 483
    },
    {
      "epoch": 0.0266387803401398,
      "grad_norm": 1.0015628337860107,
      "learning_rate": 9.997229656300991e-06,
      "loss": 0.8869,
      "step": 484
    },
    {
      "epoch": 0.02669381914249546,
      "grad_norm": 0.8965190052986145,
      "learning_rate": 9.997215209981237e-06,
      "loss": 0.7009,
      "step": 485
    },
    {
      "epoch": 0.02674885794485112,
      "grad_norm": 1.1976135969161987,
      "learning_rate": 9.997200726103749e-06,
      "loss": 0.9795,
      "step": 486
    },
    {
      "epoch": 0.02680389674720678,
      "grad_norm": 0.864780843257904,
      "learning_rate": 9.997186204668639e-06,
      "loss": 0.7687,
      "step": 487
    },
    {
      "epoch": 0.02685893554956244,
      "grad_norm": 0.9946566820144653,
      "learning_rate": 9.997171645676013e-06,
      "loss": 0.9672,
      "step": 488
    },
    {
      "epoch": 0.026913974351918103,
      "grad_norm": 1.043835997581482,
      "learning_rate": 9.997157049125985e-06,
      "loss": 0.862,
      "step": 489
    },
    {
      "epoch": 0.026969013154273764,
      "grad_norm": 0.9697456955909729,
      "learning_rate": 9.99714241501866e-06,
      "loss": 0.8368,
      "step": 490
    },
    {
      "epoch": 0.027024051956629425,
      "grad_norm": 0.9975618124008179,
      "learning_rate": 9.997127743354153e-06,
      "loss": 0.8739,
      "step": 491
    },
    {
      "epoch": 0.027079090758985083,
      "grad_norm": 1.0055313110351562,
      "learning_rate": 9.99711303413257e-06,
      "loss": 0.9227,
      "step": 492
    },
    {
      "epoch": 0.027134129561340745,
      "grad_norm": 1.0418384075164795,
      "learning_rate": 9.997098287354024e-06,
      "loss": 0.9978,
      "step": 493
    },
    {
      "epoch": 0.027189168363696406,
      "grad_norm": 0.8648970723152161,
      "learning_rate": 9.997083503018625e-06,
      "loss": 0.8363,
      "step": 494
    },
    {
      "epoch": 0.027244207166052067,
      "grad_norm": 1.13506019115448,
      "learning_rate": 9.997068681126483e-06,
      "loss": 0.8851,
      "step": 495
    },
    {
      "epoch": 0.02729924596840773,
      "grad_norm": 0.974400520324707,
      "learning_rate": 9.997053821677712e-06,
      "loss": 0.8533,
      "step": 496
    },
    {
      "epoch": 0.027354284770763387,
      "grad_norm": 1.226507544517517,
      "learning_rate": 9.997038924672419e-06,
      "loss": 0.8586,
      "step": 497
    },
    {
      "epoch": 0.027409323573119048,
      "grad_norm": 1.004753589630127,
      "learning_rate": 9.997023990110721e-06,
      "loss": 0.8974,
      "step": 498
    },
    {
      "epoch": 0.02746436237547471,
      "grad_norm": 1.0492571592330933,
      "learning_rate": 9.997009017992729e-06,
      "loss": 0.8457,
      "step": 499
    },
    {
      "epoch": 0.02751940117783037,
      "grad_norm": 1.0068167448043823,
      "learning_rate": 9.996994008318554e-06,
      "loss": 0.9608,
      "step": 500
    },
    {
      "epoch": 0.027574439980186032,
      "grad_norm": 0.9686044454574585,
      "learning_rate": 9.996978961088311e-06,
      "loss": 0.9041,
      "step": 501
    },
    {
      "epoch": 0.027629478782541694,
      "grad_norm": 1.281728744506836,
      "learning_rate": 9.99696387630211e-06,
      "loss": 0.9739,
      "step": 502
    },
    {
      "epoch": 0.02768451758489735,
      "grad_norm": 0.9069758653640747,
      "learning_rate": 9.996948753960065e-06,
      "loss": 0.8467,
      "step": 503
    },
    {
      "epoch": 0.027739556387253013,
      "grad_norm": 1.0337222814559937,
      "learning_rate": 9.996933594062293e-06,
      "loss": 0.9638,
      "step": 504
    },
    {
      "epoch": 0.027794595189608674,
      "grad_norm": 0.9695359468460083,
      "learning_rate": 9.996918396608905e-06,
      "loss": 0.8986,
      "step": 505
    },
    {
      "epoch": 0.027849633991964336,
      "grad_norm": 0.9120615124702454,
      "learning_rate": 9.996903161600016e-06,
      "loss": 0.9103,
      "step": 506
    },
    {
      "epoch": 0.027904672794319997,
      "grad_norm": 0.9736546874046326,
      "learning_rate": 9.996887889035741e-06,
      "loss": 0.9308,
      "step": 507
    },
    {
      "epoch": 0.027959711596675655,
      "grad_norm": 1.0184897184371948,
      "learning_rate": 9.996872578916192e-06,
      "loss": 0.8978,
      "step": 508
    },
    {
      "epoch": 0.028014750399031316,
      "grad_norm": 0.9791838526725769,
      "learning_rate": 9.996857231241489e-06,
      "loss": 0.8639,
      "step": 509
    },
    {
      "epoch": 0.028069789201386978,
      "grad_norm": 1.2985681295394897,
      "learning_rate": 9.996841846011742e-06,
      "loss": 0.9581,
      "step": 510
    },
    {
      "epoch": 0.02812482800374264,
      "grad_norm": 1.0647368431091309,
      "learning_rate": 9.996826423227071e-06,
      "loss": 1.0565,
      "step": 511
    },
    {
      "epoch": 0.0281798668060983,
      "grad_norm": 1.0336421728134155,
      "learning_rate": 9.996810962887591e-06,
      "loss": 1.008,
      "step": 512
    },
    {
      "epoch": 0.02823490560845396,
      "grad_norm": 1.1838933229446411,
      "learning_rate": 9.996795464993416e-06,
      "loss": 0.8359,
      "step": 513
    },
    {
      "epoch": 0.02828994441080962,
      "grad_norm": 0.9898360371589661,
      "learning_rate": 9.996779929544663e-06,
      "loss": 0.8501,
      "step": 514
    },
    {
      "epoch": 0.02834498321316528,
      "grad_norm": 0.9836066365242004,
      "learning_rate": 9.99676435654145e-06,
      "loss": 0.8795,
      "step": 515
    },
    {
      "epoch": 0.028400022015520943,
      "grad_norm": 1.0621601343154907,
      "learning_rate": 9.996748745983895e-06,
      "loss": 0.8746,
      "step": 516
    },
    {
      "epoch": 0.028455060817876604,
      "grad_norm": 1.0082437992095947,
      "learning_rate": 9.996733097872113e-06,
      "loss": 0.9278,
      "step": 517
    },
    {
      "epoch": 0.028510099620232265,
      "grad_norm": 0.9903931617736816,
      "learning_rate": 9.996717412206222e-06,
      "loss": 0.8264,
      "step": 518
    },
    {
      "epoch": 0.028565138422587923,
      "grad_norm": 1.0797243118286133,
      "learning_rate": 9.996701688986342e-06,
      "loss": 1.0077,
      "step": 519
    },
    {
      "epoch": 0.028620177224943585,
      "grad_norm": 1.147133231163025,
      "learning_rate": 9.99668592821259e-06,
      "loss": 0.9374,
      "step": 520
    },
    {
      "epoch": 0.028675216027299246,
      "grad_norm": 0.9993947744369507,
      "learning_rate": 9.996670129885082e-06,
      "loss": 0.9562,
      "step": 521
    },
    {
      "epoch": 0.028730254829654907,
      "grad_norm": 0.8580895066261292,
      "learning_rate": 9.99665429400394e-06,
      "loss": 0.7985,
      "step": 522
    },
    {
      "epoch": 0.02878529363201057,
      "grad_norm": 0.9251388907432556,
      "learning_rate": 9.996638420569281e-06,
      "loss": 0.7323,
      "step": 523
    },
    {
      "epoch": 0.028840332434366227,
      "grad_norm": 1.0010193586349487,
      "learning_rate": 9.996622509581227e-06,
      "loss": 0.9316,
      "step": 524
    },
    {
      "epoch": 0.028895371236721888,
      "grad_norm": 0.9822579026222229,
      "learning_rate": 9.996606561039894e-06,
      "loss": 0.8978,
      "step": 525
    },
    {
      "epoch": 0.02895041003907755,
      "grad_norm": 1.0760595798492432,
      "learning_rate": 9.996590574945403e-06,
      "loss": 0.9125,
      "step": 526
    },
    {
      "epoch": 0.02900544884143321,
      "grad_norm": 1.138869285583496,
      "learning_rate": 9.996574551297876e-06,
      "loss": 0.8185,
      "step": 527
    },
    {
      "epoch": 0.029060487643788872,
      "grad_norm": 1.002994179725647,
      "learning_rate": 9.996558490097433e-06,
      "loss": 0.9404,
      "step": 528
    },
    {
      "epoch": 0.02911552644614453,
      "grad_norm": 0.9550611972808838,
      "learning_rate": 9.996542391344194e-06,
      "loss": 0.859,
      "step": 529
    },
    {
      "epoch": 0.02917056524850019,
      "grad_norm": 0.9236055612564087,
      "learning_rate": 9.996526255038277e-06,
      "loss": 0.7758,
      "step": 530
    },
    {
      "epoch": 0.029225604050855853,
      "grad_norm": 1.103966474533081,
      "learning_rate": 9.996510081179808e-06,
      "loss": 1.0147,
      "step": 531
    },
    {
      "epoch": 0.029280642853211514,
      "grad_norm": 0.9884665012359619,
      "learning_rate": 9.996493869768906e-06,
      "loss": 0.8784,
      "step": 532
    },
    {
      "epoch": 0.029335681655567176,
      "grad_norm": 0.9173223376274109,
      "learning_rate": 9.996477620805694e-06,
      "loss": 0.8741,
      "step": 533
    },
    {
      "epoch": 0.029390720457922837,
      "grad_norm": 0.965548574924469,
      "learning_rate": 9.996461334290294e-06,
      "loss": 0.8989,
      "step": 534
    },
    {
      "epoch": 0.029445759260278495,
      "grad_norm": 0.9939296245574951,
      "learning_rate": 9.996445010222828e-06,
      "loss": 0.8552,
      "step": 535
    },
    {
      "epoch": 0.029500798062634156,
      "grad_norm": 1.0081578493118286,
      "learning_rate": 9.996428648603417e-06,
      "loss": 0.9138,
      "step": 536
    },
    {
      "epoch": 0.029555836864989818,
      "grad_norm": 1.0139487981796265,
      "learning_rate": 9.996412249432188e-06,
      "loss": 0.9452,
      "step": 537
    },
    {
      "epoch": 0.02961087566734548,
      "grad_norm": 0.9463647603988647,
      "learning_rate": 9.996395812709262e-06,
      "loss": 0.8721,
      "step": 538
    },
    {
      "epoch": 0.02966591446970114,
      "grad_norm": 0.9981473684310913,
      "learning_rate": 9.99637933843476e-06,
      "loss": 0.7791,
      "step": 539
    },
    {
      "epoch": 0.0297209532720568,
      "grad_norm": 1.1637190580368042,
      "learning_rate": 9.996362826608812e-06,
      "loss": 0.8798,
      "step": 540
    },
    {
      "epoch": 0.02977599207441246,
      "grad_norm": 2.2887051105499268,
      "learning_rate": 9.996346277231536e-06,
      "loss": 0.9303,
      "step": 541
    },
    {
      "epoch": 0.02983103087676812,
      "grad_norm": 0.9173391461372375,
      "learning_rate": 9.99632969030306e-06,
      "loss": 0.8627,
      "step": 542
    },
    {
      "epoch": 0.029886069679123783,
      "grad_norm": 1.033355474472046,
      "learning_rate": 9.996313065823506e-06,
      "loss": 0.9906,
      "step": 543
    },
    {
      "epoch": 0.029941108481479444,
      "grad_norm": 0.9286639094352722,
      "learning_rate": 9.996296403793002e-06,
      "loss": 0.7043,
      "step": 544
    },
    {
      "epoch": 0.029996147283835102,
      "grad_norm": 0.963238000869751,
      "learning_rate": 9.996279704211671e-06,
      "loss": 1.0236,
      "step": 545
    },
    {
      "epoch": 0.030051186086190763,
      "grad_norm": 1.0275089740753174,
      "learning_rate": 9.99626296707964e-06,
      "loss": 0.976,
      "step": 546
    },
    {
      "epoch": 0.030106224888546425,
      "grad_norm": 1.0944674015045166,
      "learning_rate": 9.996246192397032e-06,
      "loss": 0.9209,
      "step": 547
    },
    {
      "epoch": 0.030161263690902086,
      "grad_norm": 0.9620945453643799,
      "learning_rate": 9.996229380163976e-06,
      "loss": 0.8973,
      "step": 548
    },
    {
      "epoch": 0.030216302493257748,
      "grad_norm": 1.032549500465393,
      "learning_rate": 9.996212530380597e-06,
      "loss": 0.892,
      "step": 549
    },
    {
      "epoch": 0.03027134129561341,
      "grad_norm": 1.0433719158172607,
      "learning_rate": 9.996195643047023e-06,
      "loss": 0.8428,
      "step": 550
    },
    {
      "epoch": 0.030326380097969067,
      "grad_norm": 1.1541085243225098,
      "learning_rate": 9.996178718163378e-06,
      "loss": 0.9084,
      "step": 551
    },
    {
      "epoch": 0.03038141890032473,
      "grad_norm": 0.9386873245239258,
      "learning_rate": 9.996161755729793e-06,
      "loss": 0.9246,
      "step": 552
    },
    {
      "epoch": 0.03043645770268039,
      "grad_norm": 1.092236042022705,
      "learning_rate": 9.996144755746393e-06,
      "loss": 0.8419,
      "step": 553
    },
    {
      "epoch": 0.03049149650503605,
      "grad_norm": 0.9517606496810913,
      "learning_rate": 9.996127718213306e-06,
      "loss": 0.9002,
      "step": 554
    },
    {
      "epoch": 0.030546535307391712,
      "grad_norm": 0.965972900390625,
      "learning_rate": 9.996110643130661e-06,
      "loss": 0.9197,
      "step": 555
    },
    {
      "epoch": 0.03060157410974737,
      "grad_norm": 0.9396095275878906,
      "learning_rate": 9.996093530498586e-06,
      "loss": 0.8686,
      "step": 556
    },
    {
      "epoch": 0.030656612912103032,
      "grad_norm": 1.0154120922088623,
      "learning_rate": 9.99607638031721e-06,
      "loss": 0.9773,
      "step": 557
    },
    {
      "epoch": 0.030711651714458693,
      "grad_norm": 1.3572301864624023,
      "learning_rate": 9.99605919258666e-06,
      "loss": 0.911,
      "step": 558
    },
    {
      "epoch": 0.030766690516814355,
      "grad_norm": 0.968278169631958,
      "learning_rate": 9.996041967307066e-06,
      "loss": 0.7704,
      "step": 559
    },
    {
      "epoch": 0.030821729319170016,
      "grad_norm": 0.9867869019508362,
      "learning_rate": 9.99602470447856e-06,
      "loss": 0.873,
      "step": 560
    },
    {
      "epoch": 0.030876768121525677,
      "grad_norm": 1.056450605392456,
      "learning_rate": 9.996007404101269e-06,
      "loss": 0.941,
      "step": 561
    },
    {
      "epoch": 0.030931806923881335,
      "grad_norm": 1.0419799089431763,
      "learning_rate": 9.995990066175321e-06,
      "loss": 0.957,
      "step": 562
    },
    {
      "epoch": 0.030986845726236997,
      "grad_norm": 0.9789314866065979,
      "learning_rate": 9.995972690700852e-06,
      "loss": 0.9229,
      "step": 563
    },
    {
      "epoch": 0.031041884528592658,
      "grad_norm": 0.917783796787262,
      "learning_rate": 9.995955277677989e-06,
      "loss": 0.8186,
      "step": 564
    },
    {
      "epoch": 0.03109692333094832,
      "grad_norm": 1.0231432914733887,
      "learning_rate": 9.995937827106863e-06,
      "loss": 0.8624,
      "step": 565
    },
    {
      "epoch": 0.03115196213330398,
      "grad_norm": 0.9552083015441895,
      "learning_rate": 9.995920338987605e-06,
      "loss": 0.7967,
      "step": 566
    },
    {
      "epoch": 0.03120700093565964,
      "grad_norm": 0.9441083669662476,
      "learning_rate": 9.995902813320349e-06,
      "loss": 0.8471,
      "step": 567
    },
    {
      "epoch": 0.0312620397380153,
      "grad_norm": 1.0025299787521362,
      "learning_rate": 9.995885250105223e-06,
      "loss": 0.8646,
      "step": 568
    },
    {
      "epoch": 0.03131707854037096,
      "grad_norm": 0.8997280597686768,
      "learning_rate": 9.99586764934236e-06,
      "loss": 0.8736,
      "step": 569
    },
    {
      "epoch": 0.03137211734272662,
      "grad_norm": 0.9090663194656372,
      "learning_rate": 9.995850011031896e-06,
      "loss": 0.8548,
      "step": 570
    },
    {
      "epoch": 0.031427156145082284,
      "grad_norm": 0.9641294479370117,
      "learning_rate": 9.995832335173959e-06,
      "loss": 0.8667,
      "step": 571
    },
    {
      "epoch": 0.031482194947437946,
      "grad_norm": 0.9165804982185364,
      "learning_rate": 9.995814621768682e-06,
      "loss": 0.803,
      "step": 572
    },
    {
      "epoch": 0.03153723374979361,
      "grad_norm": 0.9672492742538452,
      "learning_rate": 9.995796870816202e-06,
      "loss": 0.8335,
      "step": 573
    },
    {
      "epoch": 0.03159227255214927,
      "grad_norm": 0.9359404444694519,
      "learning_rate": 9.995779082316648e-06,
      "loss": 0.8294,
      "step": 574
    },
    {
      "epoch": 0.03164731135450492,
      "grad_norm": 0.926925003528595,
      "learning_rate": 9.995761256270157e-06,
      "loss": 0.7714,
      "step": 575
    },
    {
      "epoch": 0.031702350156860584,
      "grad_norm": 1.1848629713058472,
      "learning_rate": 9.995743392676862e-06,
      "loss": 0.8925,
      "step": 576
    },
    {
      "epoch": 0.031757388959216246,
      "grad_norm": 0.9624786972999573,
      "learning_rate": 9.995725491536897e-06,
      "loss": 0.9292,
      "step": 577
    },
    {
      "epoch": 0.03181242776157191,
      "grad_norm": 0.9479736089706421,
      "learning_rate": 9.995707552850396e-06,
      "loss": 0.8797,
      "step": 578
    },
    {
      "epoch": 0.03186746656392757,
      "grad_norm": 0.9551546573638916,
      "learning_rate": 9.995689576617494e-06,
      "loss": 0.8793,
      "step": 579
    },
    {
      "epoch": 0.03192250536628323,
      "grad_norm": 0.9210056662559509,
      "learning_rate": 9.995671562838325e-06,
      "loss": 0.9714,
      "step": 580
    },
    {
      "epoch": 0.03197754416863889,
      "grad_norm": 1.063117504119873,
      "learning_rate": 9.995653511513029e-06,
      "loss": 0.9608,
      "step": 581
    },
    {
      "epoch": 0.03203258297099455,
      "grad_norm": 0.9426459670066833,
      "learning_rate": 9.995635422641736e-06,
      "loss": 0.9102,
      "step": 582
    },
    {
      "epoch": 0.032087621773350214,
      "grad_norm": 1.0176693201065063,
      "learning_rate": 9.995617296224584e-06,
      "loss": 0.9109,
      "step": 583
    },
    {
      "epoch": 0.032142660575705875,
      "grad_norm": 0.9457042217254639,
      "learning_rate": 9.995599132261711e-06,
      "loss": 0.9017,
      "step": 584
    },
    {
      "epoch": 0.03219769937806154,
      "grad_norm": 1.5851638317108154,
      "learning_rate": 9.995580930753252e-06,
      "loss": 0.967,
      "step": 585
    },
    {
      "epoch": 0.03225273818041719,
      "grad_norm": 0.9961487054824829,
      "learning_rate": 9.995562691699345e-06,
      "loss": 0.9396,
      "step": 586
    },
    {
      "epoch": 0.03230777698277285,
      "grad_norm": 0.9892112016677856,
      "learning_rate": 9.995544415100125e-06,
      "loss": 0.9058,
      "step": 587
    },
    {
      "epoch": 0.032362815785128514,
      "grad_norm": 0.9052272439002991,
      "learning_rate": 9.99552610095573e-06,
      "loss": 0.9194,
      "step": 588
    },
    {
      "epoch": 0.032417854587484175,
      "grad_norm": 0.8381399512290955,
      "learning_rate": 9.995507749266297e-06,
      "loss": 0.7465,
      "step": 589
    },
    {
      "epoch": 0.03247289338983984,
      "grad_norm": 1.018964171409607,
      "learning_rate": 9.995489360031969e-06,
      "loss": 0.841,
      "step": 590
    },
    {
      "epoch": 0.0325279321921955,
      "grad_norm": 0.908311128616333,
      "learning_rate": 9.995470933252876e-06,
      "loss": 0.8592,
      "step": 591
    },
    {
      "epoch": 0.03258297099455116,
      "grad_norm": 1.2986040115356445,
      "learning_rate": 9.995452468929162e-06,
      "loss": 0.8341,
      "step": 592
    },
    {
      "epoch": 0.03263800979690682,
      "grad_norm": 1.6565190553665161,
      "learning_rate": 9.995433967060966e-06,
      "loss": 0.8681,
      "step": 593
    },
    {
      "epoch": 0.03269304859926248,
      "grad_norm": 0.9725674390792847,
      "learning_rate": 9.995415427648423e-06,
      "loss": 0.8449,
      "step": 594
    },
    {
      "epoch": 0.032748087401618144,
      "grad_norm": 0.8683852553367615,
      "learning_rate": 9.995396850691677e-06,
      "loss": 0.8478,
      "step": 595
    },
    {
      "epoch": 0.0328031262039738,
      "grad_norm": 0.9912856817245483,
      "learning_rate": 9.995378236190862e-06,
      "loss": 0.8912,
      "step": 596
    },
    {
      "epoch": 0.03285816500632946,
      "grad_norm": 0.9396800398826599,
      "learning_rate": 9.995359584146125e-06,
      "loss": 0.856,
      "step": 597
    },
    {
      "epoch": 0.03291320380868512,
      "grad_norm": 1.385006308555603,
      "learning_rate": 9.995340894557601e-06,
      "loss": 0.9633,
      "step": 598
    },
    {
      "epoch": 0.03296824261104078,
      "grad_norm": 0.8982875943183899,
      "learning_rate": 9.995322167425433e-06,
      "loss": 0.9244,
      "step": 599
    },
    {
      "epoch": 0.033023281413396444,
      "grad_norm": 0.8981022834777832,
      "learning_rate": 9.995303402749759e-06,
      "loss": 0.8854,
      "step": 600
    },
    {
      "epoch": 0.033078320215752105,
      "grad_norm": 0.9917197227478027,
      "learning_rate": 9.995284600530724e-06,
      "loss": 1.0086,
      "step": 601
    },
    {
      "epoch": 0.033133359018107766,
      "grad_norm": 1.0540626049041748,
      "learning_rate": 9.995265760768464e-06,
      "loss": 1.0022,
      "step": 602
    },
    {
      "epoch": 0.03318839782046343,
      "grad_norm": 0.9523479342460632,
      "learning_rate": 9.995246883463126e-06,
      "loss": 0.9893,
      "step": 603
    },
    {
      "epoch": 0.03324343662281909,
      "grad_norm": 0.9824770092964172,
      "learning_rate": 9.99522796861485e-06,
      "loss": 0.8385,
      "step": 604
    },
    {
      "epoch": 0.03329847542517475,
      "grad_norm": 1.0968893766403198,
      "learning_rate": 9.995209016223776e-06,
      "loss": 1.0109,
      "step": 605
    },
    {
      "epoch": 0.03335351422753041,
      "grad_norm": 0.9115625023841858,
      "learning_rate": 9.995190026290049e-06,
      "loss": 0.8656,
      "step": 606
    },
    {
      "epoch": 0.033408553029886066,
      "grad_norm": 0.9795814156532288,
      "learning_rate": 9.99517099881381e-06,
      "loss": 0.8941,
      "step": 607
    },
    {
      "epoch": 0.03346359183224173,
      "grad_norm": 0.9317291378974915,
      "learning_rate": 9.995151933795204e-06,
      "loss": 0.7819,
      "step": 608
    },
    {
      "epoch": 0.03351863063459739,
      "grad_norm": 0.9936283230781555,
      "learning_rate": 9.995132831234373e-06,
      "loss": 0.8674,
      "step": 609
    },
    {
      "epoch": 0.03357366943695305,
      "grad_norm": 0.9872812032699585,
      "learning_rate": 9.995113691131462e-06,
      "loss": 0.9038,
      "step": 610
    },
    {
      "epoch": 0.03362870823930871,
      "grad_norm": 0.9516895413398743,
      "learning_rate": 9.995094513486611e-06,
      "loss": 0.9038,
      "step": 611
    },
    {
      "epoch": 0.03368374704166437,
      "grad_norm": 1.090579867362976,
      "learning_rate": 9.995075298299968e-06,
      "loss": 0.9587,
      "step": 612
    },
    {
      "epoch": 0.033738785844020035,
      "grad_norm": 1.021398663520813,
      "learning_rate": 9.995056045571677e-06,
      "loss": 0.9569,
      "step": 613
    },
    {
      "epoch": 0.033793824646375696,
      "grad_norm": 1.009657382965088,
      "learning_rate": 9.99503675530188e-06,
      "loss": 0.8346,
      "step": 614
    },
    {
      "epoch": 0.03384886344873136,
      "grad_norm": 1.0478712320327759,
      "learning_rate": 9.995017427490725e-06,
      "loss": 1.0566,
      "step": 615
    },
    {
      "epoch": 0.03390390225108702,
      "grad_norm": 1.1391830444335938,
      "learning_rate": 9.994998062138355e-06,
      "loss": 1.0727,
      "step": 616
    },
    {
      "epoch": 0.03395894105344268,
      "grad_norm": 1.0172302722930908,
      "learning_rate": 9.994978659244918e-06,
      "loss": 0.7869,
      "step": 617
    },
    {
      "epoch": 0.034013979855798335,
      "grad_norm": 1.0532630681991577,
      "learning_rate": 9.994959218810558e-06,
      "loss": 0.8626,
      "step": 618
    },
    {
      "epoch": 0.034069018658153996,
      "grad_norm": 0.8300478458404541,
      "learning_rate": 9.99493974083542e-06,
      "loss": 0.8166,
      "step": 619
    },
    {
      "epoch": 0.03412405746050966,
      "grad_norm": 1.0613664388656616,
      "learning_rate": 9.994920225319656e-06,
      "loss": 0.8899,
      "step": 620
    },
    {
      "epoch": 0.03417909626286532,
      "grad_norm": 0.9827042818069458,
      "learning_rate": 9.994900672263406e-06,
      "loss": 0.8243,
      "step": 621
    },
    {
      "epoch": 0.03423413506522098,
      "grad_norm": 0.8790082931518555,
      "learning_rate": 9.994881081666818e-06,
      "loss": 0.8153,
      "step": 622
    },
    {
      "epoch": 0.03428917386757664,
      "grad_norm": 1.033378005027771,
      "learning_rate": 9.994861453530044e-06,
      "loss": 0.8916,
      "step": 623
    },
    {
      "epoch": 0.0343442126699323,
      "grad_norm": 0.9547238349914551,
      "learning_rate": 9.994841787853227e-06,
      "loss": 0.9141,
      "step": 624
    },
    {
      "epoch": 0.034399251472287964,
      "grad_norm": 0.9606438279151917,
      "learning_rate": 9.994822084636514e-06,
      "loss": 0.9435,
      "step": 625
    },
    {
      "epoch": 0.034454290274643626,
      "grad_norm": 0.8461503982543945,
      "learning_rate": 9.994802343880059e-06,
      "loss": 0.7914,
      "step": 626
    },
    {
      "epoch": 0.03450932907699929,
      "grad_norm": 1.144538402557373,
      "learning_rate": 9.994782565584004e-06,
      "loss": 0.8025,
      "step": 627
    },
    {
      "epoch": 0.03456436787935495,
      "grad_norm": 1.0099962949752808,
      "learning_rate": 9.994762749748502e-06,
      "loss": 0.9607,
      "step": 628
    },
    {
      "epoch": 0.0346194066817106,
      "grad_norm": 0.9822041988372803,
      "learning_rate": 9.9947428963737e-06,
      "loss": 0.9216,
      "step": 629
    },
    {
      "epoch": 0.034674445484066264,
      "grad_norm": 0.9056866765022278,
      "learning_rate": 9.994723005459746e-06,
      "loss": 0.7913,
      "step": 630
    },
    {
      "epoch": 0.034729484286421926,
      "grad_norm": 1.0099287033081055,
      "learning_rate": 9.994703077006792e-06,
      "loss": 0.9937,
      "step": 631
    },
    {
      "epoch": 0.03478452308877759,
      "grad_norm": 0.9559167623519897,
      "learning_rate": 9.994683111014984e-06,
      "loss": 0.9774,
      "step": 632
    },
    {
      "epoch": 0.03483956189113325,
      "grad_norm": 1.0359059572219849,
      "learning_rate": 9.994663107484478e-06,
      "loss": 0.9062,
      "step": 633
    },
    {
      "epoch": 0.03489460069348891,
      "grad_norm": 0.8803057074546814,
      "learning_rate": 9.99464306641542e-06,
      "loss": 0.9638,
      "step": 634
    },
    {
      "epoch": 0.03494963949584457,
      "grad_norm": 1.0926579236984253,
      "learning_rate": 9.994622987807962e-06,
      "loss": 1.0467,
      "step": 635
    },
    {
      "epoch": 0.03500467829820023,
      "grad_norm": 1.0051401853561401,
      "learning_rate": 9.994602871662253e-06,
      "loss": 0.8717,
      "step": 636
    },
    {
      "epoch": 0.035059717100555894,
      "grad_norm": 1.2007508277893066,
      "learning_rate": 9.994582717978448e-06,
      "loss": 0.8004,
      "step": 637
    },
    {
      "epoch": 0.035114755902911556,
      "grad_norm": 0.8826266527175903,
      "learning_rate": 9.994562526756695e-06,
      "loss": 0.8888,
      "step": 638
    },
    {
      "epoch": 0.03516979470526721,
      "grad_norm": 0.9953717589378357,
      "learning_rate": 9.994542297997147e-06,
      "loss": 0.8999,
      "step": 639
    },
    {
      "epoch": 0.03522483350762287,
      "grad_norm": 1.0203614234924316,
      "learning_rate": 9.994522031699958e-06,
      "loss": 0.8241,
      "step": 640
    },
    {
      "epoch": 0.03527987230997853,
      "grad_norm": 0.8760203719139099,
      "learning_rate": 9.994501727865276e-06,
      "loss": 0.7893,
      "step": 641
    },
    {
      "epoch": 0.035334911112334194,
      "grad_norm": 1.024888277053833,
      "learning_rate": 9.994481386493257e-06,
      "loss": 0.9865,
      "step": 642
    },
    {
      "epoch": 0.035389949914689856,
      "grad_norm": 0.907454788684845,
      "learning_rate": 9.994461007584052e-06,
      "loss": 0.891,
      "step": 643
    },
    {
      "epoch": 0.03544498871704552,
      "grad_norm": 1.0400965213775635,
      "learning_rate": 9.994440591137816e-06,
      "loss": 0.9345,
      "step": 644
    },
    {
      "epoch": 0.03550002751940118,
      "grad_norm": 0.9816616177558899,
      "learning_rate": 9.9944201371547e-06,
      "loss": 0.91,
      "step": 645
    },
    {
      "epoch": 0.03555506632175684,
      "grad_norm": 1.0528117418289185,
      "learning_rate": 9.99439964563486e-06,
      "loss": 0.952,
      "step": 646
    },
    {
      "epoch": 0.0356101051241125,
      "grad_norm": 0.9802080988883972,
      "learning_rate": 9.99437911657845e-06,
      "loss": 0.9392,
      "step": 647
    },
    {
      "epoch": 0.03566514392646816,
      "grad_norm": 0.9580393433570862,
      "learning_rate": 9.994358549985623e-06,
      "loss": 0.874,
      "step": 648
    },
    {
      "epoch": 0.035720182728823824,
      "grad_norm": 0.8935576677322388,
      "learning_rate": 9.994337945856533e-06,
      "loss": 0.8435,
      "step": 649
    },
    {
      "epoch": 0.03577522153117948,
      "grad_norm": 1.009699821472168,
      "learning_rate": 9.994317304191337e-06,
      "loss": 0.9436,
      "step": 650
    },
    {
      "epoch": 0.03583026033353514,
      "grad_norm": 0.9126121401786804,
      "learning_rate": 9.994296624990188e-06,
      "loss": 0.8424,
      "step": 651
    },
    {
      "epoch": 0.0358852991358908,
      "grad_norm": 0.9555553197860718,
      "learning_rate": 9.994275908253243e-06,
      "loss": 0.93,
      "step": 652
    },
    {
      "epoch": 0.03594033793824646,
      "grad_norm": 0.8359857797622681,
      "learning_rate": 9.994255153980658e-06,
      "loss": 0.6326,
      "step": 653
    },
    {
      "epoch": 0.035995376740602124,
      "grad_norm": 0.8918783664703369,
      "learning_rate": 9.994234362172587e-06,
      "loss": 0.8287,
      "step": 654
    },
    {
      "epoch": 0.036050415542957785,
      "grad_norm": 0.9878549575805664,
      "learning_rate": 9.994213532829188e-06,
      "loss": 0.8841,
      "step": 655
    },
    {
      "epoch": 0.03610545434531345,
      "grad_norm": 0.9504040479660034,
      "learning_rate": 9.994192665950617e-06,
      "loss": 1.0182,
      "step": 656
    },
    {
      "epoch": 0.03616049314766911,
      "grad_norm": 0.9531422257423401,
      "learning_rate": 9.99417176153703e-06,
      "loss": 0.8504,
      "step": 657
    },
    {
      "epoch": 0.03621553195002477,
      "grad_norm": 0.9580292105674744,
      "learning_rate": 9.994150819588587e-06,
      "loss": 0.8048,
      "step": 658
    },
    {
      "epoch": 0.03627057075238043,
      "grad_norm": 0.9786819815635681,
      "learning_rate": 9.99412984010544e-06,
      "loss": 0.9124,
      "step": 659
    },
    {
      "epoch": 0.03632560955473609,
      "grad_norm": 0.9733422994613647,
      "learning_rate": 9.994108823087751e-06,
      "loss": 0.8868,
      "step": 660
    },
    {
      "epoch": 0.03638064835709175,
      "grad_norm": 1.093173623085022,
      "learning_rate": 9.994087768535679e-06,
      "loss": 0.9428,
      "step": 661
    },
    {
      "epoch": 0.03643568715944741,
      "grad_norm": 0.9067148566246033,
      "learning_rate": 9.994066676449378e-06,
      "loss": 0.8838,
      "step": 662
    },
    {
      "epoch": 0.03649072596180307,
      "grad_norm": 0.9509521722793579,
      "learning_rate": 9.99404554682901e-06,
      "loss": 0.9034,
      "step": 663
    },
    {
      "epoch": 0.03654576476415873,
      "grad_norm": 0.9523824453353882,
      "learning_rate": 9.994024379674731e-06,
      "loss": 0.9623,
      "step": 664
    },
    {
      "epoch": 0.03660080356651439,
      "grad_norm": 0.987276554107666,
      "learning_rate": 9.994003174986703e-06,
      "loss": 0.8817,
      "step": 665
    },
    {
      "epoch": 0.036655842368870054,
      "grad_norm": 0.9500744342803955,
      "learning_rate": 9.993981932765083e-06,
      "loss": 0.9742,
      "step": 666
    },
    {
      "epoch": 0.036710881171225715,
      "grad_norm": 0.9420705437660217,
      "learning_rate": 9.993960653010034e-06,
      "loss": 0.9657,
      "step": 667
    },
    {
      "epoch": 0.036765919973581376,
      "grad_norm": 0.9443248510360718,
      "learning_rate": 9.99393933572171e-06,
      "loss": 0.8468,
      "step": 668
    },
    {
      "epoch": 0.03682095877593704,
      "grad_norm": 0.9666558504104614,
      "learning_rate": 9.993917980900276e-06,
      "loss": 0.9871,
      "step": 669
    },
    {
      "epoch": 0.0368759975782927,
      "grad_norm": 1.0236201286315918,
      "learning_rate": 9.993896588545892e-06,
      "loss": 0.9814,
      "step": 670
    },
    {
      "epoch": 0.03693103638064836,
      "grad_norm": 1.016190528869629,
      "learning_rate": 9.993875158658716e-06,
      "loss": 1.0156,
      "step": 671
    },
    {
      "epoch": 0.036986075183004015,
      "grad_norm": 0.9296661019325256,
      "learning_rate": 9.993853691238913e-06,
      "loss": 0.7956,
      "step": 672
    },
    {
      "epoch": 0.037041113985359676,
      "grad_norm": 0.9276684522628784,
      "learning_rate": 9.993832186286643e-06,
      "loss": 0.9253,
      "step": 673
    },
    {
      "epoch": 0.03709615278771534,
      "grad_norm": 0.8588787913322449,
      "learning_rate": 9.993810643802065e-06,
      "loss": 0.7878,
      "step": 674
    },
    {
      "epoch": 0.037151191590071,
      "grad_norm": 0.9955212473869324,
      "learning_rate": 9.993789063785344e-06,
      "loss": 0.8711,
      "step": 675
    },
    {
      "epoch": 0.03720623039242666,
      "grad_norm": 0.925578236579895,
      "learning_rate": 9.993767446236642e-06,
      "loss": 0.9431,
      "step": 676
    },
    {
      "epoch": 0.03726126919478232,
      "grad_norm": 0.9610552787780762,
      "learning_rate": 9.99374579115612e-06,
      "loss": 0.887,
      "step": 677
    },
    {
      "epoch": 0.03731630799713798,
      "grad_norm": 1.0052428245544434,
      "learning_rate": 9.99372409854394e-06,
      "loss": 0.8751,
      "step": 678
    },
    {
      "epoch": 0.037371346799493645,
      "grad_norm": 0.9503066539764404,
      "learning_rate": 9.99370236840027e-06,
      "loss": 0.8556,
      "step": 679
    },
    {
      "epoch": 0.037426385601849306,
      "grad_norm": 2.426232099533081,
      "learning_rate": 9.993680600725266e-06,
      "loss": 0.9077,
      "step": 680
    },
    {
      "epoch": 0.03748142440420497,
      "grad_norm": 0.9119723439216614,
      "learning_rate": 9.993658795519096e-06,
      "loss": 0.8575,
      "step": 681
    },
    {
      "epoch": 0.03753646320656062,
      "grad_norm": 0.9688286781311035,
      "learning_rate": 9.993636952781923e-06,
      "loss": 0.8921,
      "step": 682
    },
    {
      "epoch": 0.03759150200891628,
      "grad_norm": 1.030013084411621,
      "learning_rate": 9.993615072513913e-06,
      "loss": 0.8622,
      "step": 683
    },
    {
      "epoch": 0.037646540811271945,
      "grad_norm": 1.055187463760376,
      "learning_rate": 9.993593154715228e-06,
      "loss": 0.9251,
      "step": 684
    },
    {
      "epoch": 0.037701579613627606,
      "grad_norm": 1.0518591403961182,
      "learning_rate": 9.993571199386032e-06,
      "loss": 0.9575,
      "step": 685
    },
    {
      "epoch": 0.03775661841598327,
      "grad_norm": 0.9232666492462158,
      "learning_rate": 9.993549206526495e-06,
      "loss": 0.8522,
      "step": 686
    },
    {
      "epoch": 0.03781165721833893,
      "grad_norm": 1.0212332010269165,
      "learning_rate": 9.993527176136775e-06,
      "loss": 0.9358,
      "step": 687
    },
    {
      "epoch": 0.03786669602069459,
      "grad_norm": 0.9137141108512878,
      "learning_rate": 9.993505108217045e-06,
      "loss": 0.8561,
      "step": 688
    },
    {
      "epoch": 0.03792173482305025,
      "grad_norm": 1.0069375038146973,
      "learning_rate": 9.993483002767465e-06,
      "loss": 0.8274,
      "step": 689
    },
    {
      "epoch": 0.03797677362540591,
      "grad_norm": 0.9820672869682312,
      "learning_rate": 9.993460859788204e-06,
      "loss": 0.907,
      "step": 690
    },
    {
      "epoch": 0.038031812427761574,
      "grad_norm": 1.0042002201080322,
      "learning_rate": 9.993438679279428e-06,
      "loss": 0.9263,
      "step": 691
    },
    {
      "epoch": 0.038086851230117236,
      "grad_norm": 0.9733695983886719,
      "learning_rate": 9.993416461241304e-06,
      "loss": 0.8455,
      "step": 692
    },
    {
      "epoch": 0.03814189003247289,
      "grad_norm": 0.9106015563011169,
      "learning_rate": 9.993394205673996e-06,
      "loss": 0.8469,
      "step": 693
    },
    {
      "epoch": 0.03819692883482855,
      "grad_norm": 0.9802660346031189,
      "learning_rate": 9.993371912577677e-06,
      "loss": 0.8662,
      "step": 694
    },
    {
      "epoch": 0.03825196763718421,
      "grad_norm": 0.9183964729309082,
      "learning_rate": 9.99334958195251e-06,
      "loss": 0.8968,
      "step": 695
    },
    {
      "epoch": 0.038307006439539874,
      "grad_norm": 0.9572185277938843,
      "learning_rate": 9.993327213798663e-06,
      "loss": 0.953,
      "step": 696
    },
    {
      "epoch": 0.038362045241895536,
      "grad_norm": 1.4480071067810059,
      "learning_rate": 9.993304808116307e-06,
      "loss": 1.1131,
      "step": 697
    },
    {
      "epoch": 0.0384170840442512,
      "grad_norm": 0.9297361969947815,
      "learning_rate": 9.993282364905607e-06,
      "loss": 0.884,
      "step": 698
    },
    {
      "epoch": 0.03847212284660686,
      "grad_norm": 0.9400073885917664,
      "learning_rate": 9.993259884166735e-06,
      "loss": 0.932,
      "step": 699
    },
    {
      "epoch": 0.03852716164896252,
      "grad_norm": 0.9231798052787781,
      "learning_rate": 9.993237365899858e-06,
      "loss": 0.8981,
      "step": 700
    },
    {
      "epoch": 0.03858220045131818,
      "grad_norm": 0.8233712911605835,
      "learning_rate": 9.993214810105144e-06,
      "loss": 0.8218,
      "step": 701
    },
    {
      "epoch": 0.03863723925367384,
      "grad_norm": 1.0997854471206665,
      "learning_rate": 9.993192216782768e-06,
      "loss": 0.9298,
      "step": 702
    },
    {
      "epoch": 0.038692278056029504,
      "grad_norm": 0.9570802450180054,
      "learning_rate": 9.993169585932893e-06,
      "loss": 0.7815,
      "step": 703
    },
    {
      "epoch": 0.03874731685838516,
      "grad_norm": 0.9913730025291443,
      "learning_rate": 9.993146917555692e-06,
      "loss": 0.9621,
      "step": 704
    },
    {
      "epoch": 0.03880235566074082,
      "grad_norm": 1.088767409324646,
      "learning_rate": 9.993124211651334e-06,
      "loss": 0.9295,
      "step": 705
    },
    {
      "epoch": 0.03885739446309648,
      "grad_norm": 0.8199124336242676,
      "learning_rate": 9.993101468219995e-06,
      "loss": 0.7613,
      "step": 706
    },
    {
      "epoch": 0.03891243326545214,
      "grad_norm": 1.112566351890564,
      "learning_rate": 9.99307868726184e-06,
      "loss": 0.791,
      "step": 707
    },
    {
      "epoch": 0.038967472067807804,
      "grad_norm": 0.9372578859329224,
      "learning_rate": 9.99305586877704e-06,
      "loss": 0.8567,
      "step": 708
    },
    {
      "epoch": 0.039022510870163465,
      "grad_norm": 1.0167721509933472,
      "learning_rate": 9.99303301276577e-06,
      "loss": 0.9787,
      "step": 709
    },
    {
      "epoch": 0.03907754967251913,
      "grad_norm": 1.3526856899261475,
      "learning_rate": 9.993010119228202e-06,
      "loss": 1.2215,
      "step": 710
    },
    {
      "epoch": 0.03913258847487479,
      "grad_norm": 0.8819016814231873,
      "learning_rate": 9.992987188164505e-06,
      "loss": 0.7736,
      "step": 711
    },
    {
      "epoch": 0.03918762727723045,
      "grad_norm": 1.0033677816390991,
      "learning_rate": 9.992964219574852e-06,
      "loss": 0.9919,
      "step": 712
    },
    {
      "epoch": 0.03924266607958611,
      "grad_norm": 0.894926130771637,
      "learning_rate": 9.992941213459417e-06,
      "loss": 0.9058,
      "step": 713
    },
    {
      "epoch": 0.03929770488194177,
      "grad_norm": 0.9481377005577087,
      "learning_rate": 9.992918169818373e-06,
      "loss": 0.8436,
      "step": 714
    },
    {
      "epoch": 0.03935274368429743,
      "grad_norm": 0.9312933087348938,
      "learning_rate": 9.992895088651893e-06,
      "loss": 0.8869,
      "step": 715
    },
    {
      "epoch": 0.03940778248665309,
      "grad_norm": 0.9765705466270447,
      "learning_rate": 9.99287196996015e-06,
      "loss": 0.9512,
      "step": 716
    },
    {
      "epoch": 0.03946282128900875,
      "grad_norm": 0.9610235691070557,
      "learning_rate": 9.992848813743317e-06,
      "loss": 0.8005,
      "step": 717
    },
    {
      "epoch": 0.03951786009136441,
      "grad_norm": 1.102995753288269,
      "learning_rate": 9.99282562000157e-06,
      "loss": 0.8017,
      "step": 718
    },
    {
      "epoch": 0.03957289889372007,
      "grad_norm": 1.023317575454712,
      "learning_rate": 9.99280238873508e-06,
      "loss": 0.911,
      "step": 719
    },
    {
      "epoch": 0.039627937696075734,
      "grad_norm": 1.0531049966812134,
      "learning_rate": 9.992779119944025e-06,
      "loss": 0.8562,
      "step": 720
    },
    {
      "epoch": 0.039682976498431395,
      "grad_norm": 0.918250322341919,
      "learning_rate": 9.992755813628579e-06,
      "loss": 0.92,
      "step": 721
    },
    {
      "epoch": 0.039738015300787057,
      "grad_norm": 0.8508251309394836,
      "learning_rate": 9.992732469788915e-06,
      "loss": 0.7347,
      "step": 722
    },
    {
      "epoch": 0.03979305410314272,
      "grad_norm": 0.9184926152229309,
      "learning_rate": 9.992709088425211e-06,
      "loss": 0.8732,
      "step": 723
    },
    {
      "epoch": 0.03984809290549838,
      "grad_norm": 1.1613929271697998,
      "learning_rate": 9.992685669537643e-06,
      "loss": 0.9522,
      "step": 724
    },
    {
      "epoch": 0.039903131707854034,
      "grad_norm": 1.091513752937317,
      "learning_rate": 9.992662213126386e-06,
      "loss": 0.9646,
      "step": 725
    },
    {
      "epoch": 0.039958170510209695,
      "grad_norm": 1.057803750038147,
      "learning_rate": 9.992638719191615e-06,
      "loss": 0.7032,
      "step": 726
    },
    {
      "epoch": 0.040013209312565357,
      "grad_norm": 0.8771823644638062,
      "learning_rate": 9.992615187733508e-06,
      "loss": 0.8577,
      "step": 727
    },
    {
      "epoch": 0.04006824811492102,
      "grad_norm": 0.9471028447151184,
      "learning_rate": 9.992591618752244e-06,
      "loss": 0.9057,
      "step": 728
    },
    {
      "epoch": 0.04012328691727668,
      "grad_norm": 0.9547705054283142,
      "learning_rate": 9.992568012247995e-06,
      "loss": 0.9549,
      "step": 729
    },
    {
      "epoch": 0.04017832571963234,
      "grad_norm": 0.8862974047660828,
      "learning_rate": 9.992544368220941e-06,
      "loss": 0.8593,
      "step": 730
    },
    {
      "epoch": 0.040233364521988,
      "grad_norm": 0.906334400177002,
      "learning_rate": 9.992520686671261e-06,
      "loss": 0.8832,
      "step": 731
    },
    {
      "epoch": 0.04028840332434366,
      "grad_norm": 1.07270085811615,
      "learning_rate": 9.992496967599133e-06,
      "loss": 0.9409,
      "step": 732
    },
    {
      "epoch": 0.040343442126699325,
      "grad_norm": 0.9026005268096924,
      "learning_rate": 9.992473211004734e-06,
      "loss": 0.8326,
      "step": 733
    },
    {
      "epoch": 0.040398480929054986,
      "grad_norm": 0.9762942790985107,
      "learning_rate": 9.992449416888241e-06,
      "loss": 0.9048,
      "step": 734
    },
    {
      "epoch": 0.04045351973141065,
      "grad_norm": 0.9658033847808838,
      "learning_rate": 9.992425585249837e-06,
      "loss": 0.9219,
      "step": 735
    },
    {
      "epoch": 0.0405085585337663,
      "grad_norm": 0.8909044861793518,
      "learning_rate": 9.992401716089698e-06,
      "loss": 0.8564,
      "step": 736
    },
    {
      "epoch": 0.04056359733612196,
      "grad_norm": 1.0387929677963257,
      "learning_rate": 9.992377809408001e-06,
      "loss": 0.9533,
      "step": 737
    },
    {
      "epoch": 0.040618636138477625,
      "grad_norm": 0.9044275879859924,
      "learning_rate": 9.99235386520493e-06,
      "loss": 0.8508,
      "step": 738
    },
    {
      "epoch": 0.040673674940833286,
      "grad_norm": 1.019377589225769,
      "learning_rate": 9.992329883480667e-06,
      "loss": 0.8684,
      "step": 739
    },
    {
      "epoch": 0.04072871374318895,
      "grad_norm": 0.9394627213478088,
      "learning_rate": 9.992305864235385e-06,
      "loss": 0.7665,
      "step": 740
    },
    {
      "epoch": 0.04078375254554461,
      "grad_norm": 0.8652323484420776,
      "learning_rate": 9.99228180746927e-06,
      "loss": 0.8576,
      "step": 741
    },
    {
      "epoch": 0.04083879134790027,
      "grad_norm": 0.9347619414329529,
      "learning_rate": 9.992257713182502e-06,
      "loss": 0.9586,
      "step": 742
    },
    {
      "epoch": 0.04089383015025593,
      "grad_norm": 0.9510203003883362,
      "learning_rate": 9.99223358137526e-06,
      "loss": 0.9092,
      "step": 743
    },
    {
      "epoch": 0.04094886895261159,
      "grad_norm": 0.8242866396903992,
      "learning_rate": 9.992209412047729e-06,
      "loss": 0.6997,
      "step": 744
    },
    {
      "epoch": 0.041003907754967255,
      "grad_norm": 0.8842730522155762,
      "learning_rate": 9.992185205200087e-06,
      "loss": 0.8873,
      "step": 745
    },
    {
      "epoch": 0.041058946557322916,
      "grad_norm": 1.0813730955123901,
      "learning_rate": 9.992160960832518e-06,
      "loss": 1.0162,
      "step": 746
    },
    {
      "epoch": 0.04111398535967857,
      "grad_norm": 1.1276283264160156,
      "learning_rate": 9.9921366789452e-06,
      "loss": 1.0004,
      "step": 747
    },
    {
      "epoch": 0.04116902416203423,
      "grad_norm": 0.8810326457023621,
      "learning_rate": 9.992112359538323e-06,
      "loss": 0.7823,
      "step": 748
    },
    {
      "epoch": 0.04122406296438989,
      "grad_norm": 0.9939407110214233,
      "learning_rate": 9.992088002612066e-06,
      "loss": 1.0016,
      "step": 749
    },
    {
      "epoch": 0.041279101766745555,
      "grad_norm": 1.0963523387908936,
      "learning_rate": 9.99206360816661e-06,
      "loss": 0.9252,
      "step": 750
    },
    {
      "epoch": 0.041334140569101216,
      "grad_norm": 1.1346478462219238,
      "learning_rate": 9.99203917620214e-06,
      "loss": 0.9608,
      "step": 751
    },
    {
      "epoch": 0.04138917937145688,
      "grad_norm": 1.0108580589294434,
      "learning_rate": 9.992014706718841e-06,
      "loss": 0.9179,
      "step": 752
    },
    {
      "epoch": 0.04144421817381254,
      "grad_norm": 0.897293210029602,
      "learning_rate": 9.991990199716894e-06,
      "loss": 0.9295,
      "step": 753
    },
    {
      "epoch": 0.0414992569761682,
      "grad_norm": 1.0152363777160645,
      "learning_rate": 9.991965655196488e-06,
      "loss": 0.8467,
      "step": 754
    },
    {
      "epoch": 0.04155429577852386,
      "grad_norm": 0.8655388355255127,
      "learning_rate": 9.9919410731578e-06,
      "loss": 0.796,
      "step": 755
    },
    {
      "epoch": 0.04160933458087952,
      "grad_norm": 1.0140331983566284,
      "learning_rate": 9.991916453601023e-06,
      "loss": 0.8444,
      "step": 756
    },
    {
      "epoch": 0.041664373383235184,
      "grad_norm": 0.9387341141700745,
      "learning_rate": 9.991891796526338e-06,
      "loss": 0.8669,
      "step": 757
    },
    {
      "epoch": 0.04171941218559084,
      "grad_norm": 0.9395696520805359,
      "learning_rate": 9.991867101933928e-06,
      "loss": 0.8376,
      "step": 758
    },
    {
      "epoch": 0.0417744509879465,
      "grad_norm": 1.0856634378433228,
      "learning_rate": 9.991842369823983e-06,
      "loss": 0.9271,
      "step": 759
    },
    {
      "epoch": 0.04182948979030216,
      "grad_norm": 0.8777190446853638,
      "learning_rate": 9.991817600196687e-06,
      "loss": 0.9197,
      "step": 760
    },
    {
      "epoch": 0.04188452859265782,
      "grad_norm": 0.9639917016029358,
      "learning_rate": 9.991792793052225e-06,
      "loss": 0.8835,
      "step": 761
    },
    {
      "epoch": 0.041939567395013484,
      "grad_norm": 0.9384773969650269,
      "learning_rate": 9.991767948390785e-06,
      "loss": 0.8403,
      "step": 762
    },
    {
      "epoch": 0.041994606197369146,
      "grad_norm": 0.8987650275230408,
      "learning_rate": 9.991743066212554e-06,
      "loss": 0.7948,
      "step": 763
    },
    {
      "epoch": 0.04204964499972481,
      "grad_norm": 1.0545049905776978,
      "learning_rate": 9.991718146517717e-06,
      "loss": 0.9359,
      "step": 764
    },
    {
      "epoch": 0.04210468380208047,
      "grad_norm": 0.9840022325515747,
      "learning_rate": 9.991693189306463e-06,
      "loss": 0.9188,
      "step": 765
    },
    {
      "epoch": 0.04215972260443613,
      "grad_norm": 0.8769927620887756,
      "learning_rate": 9.991668194578981e-06,
      "loss": 0.8647,
      "step": 766
    },
    {
      "epoch": 0.04221476140679179,
      "grad_norm": 0.9268791675567627,
      "learning_rate": 9.991643162335455e-06,
      "loss": 0.897,
      "step": 767
    },
    {
      "epoch": 0.042269800209147446,
      "grad_norm": 0.9316747784614563,
      "learning_rate": 9.991618092576075e-06,
      "loss": 0.9341,
      "step": 768
    },
    {
      "epoch": 0.04232483901150311,
      "grad_norm": 0.8348364233970642,
      "learning_rate": 9.991592985301031e-06,
      "loss": 0.7528,
      "step": 769
    },
    {
      "epoch": 0.04237987781385877,
      "grad_norm": 0.9139068126678467,
      "learning_rate": 9.99156784051051e-06,
      "loss": 0.8596,
      "step": 770
    },
    {
      "epoch": 0.04243491661621443,
      "grad_norm": 0.9403928518295288,
      "learning_rate": 9.991542658204701e-06,
      "loss": 0.974,
      "step": 771
    },
    {
      "epoch": 0.04248995541857009,
      "grad_norm": 0.993549108505249,
      "learning_rate": 9.991517438383793e-06,
      "loss": 0.9479,
      "step": 772
    },
    {
      "epoch": 0.04254499422092575,
      "grad_norm": 0.8494916558265686,
      "learning_rate": 9.991492181047975e-06,
      "loss": 0.9149,
      "step": 773
    },
    {
      "epoch": 0.042600033023281414,
      "grad_norm": 1.0351910591125488,
      "learning_rate": 9.991466886197441e-06,
      "loss": 0.9552,
      "step": 774
    },
    {
      "epoch": 0.042655071825637075,
      "grad_norm": 0.916829526424408,
      "learning_rate": 9.991441553832375e-06,
      "loss": 0.8781,
      "step": 775
    },
    {
      "epoch": 0.04271011062799274,
      "grad_norm": 1.113476276397705,
      "learning_rate": 9.991416183952972e-06,
      "loss": 0.8137,
      "step": 776
    },
    {
      "epoch": 0.0427651494303484,
      "grad_norm": 1.1608171463012695,
      "learning_rate": 9.991390776559421e-06,
      "loss": 1.0045,
      "step": 777
    },
    {
      "epoch": 0.04282018823270406,
      "grad_norm": 1.0045493841171265,
      "learning_rate": 9.991365331651913e-06,
      "loss": 0.8813,
      "step": 778
    },
    {
      "epoch": 0.042875227035059714,
      "grad_norm": 0.918820858001709,
      "learning_rate": 9.991339849230639e-06,
      "loss": 0.9198,
      "step": 779
    },
    {
      "epoch": 0.042930265837415375,
      "grad_norm": 0.9875735640525818,
      "learning_rate": 9.991314329295792e-06,
      "loss": 0.8665,
      "step": 780
    },
    {
      "epoch": 0.04298530463977104,
      "grad_norm": 0.873768150806427,
      "learning_rate": 9.991288771847561e-06,
      "loss": 0.8606,
      "step": 781
    },
    {
      "epoch": 0.0430403434421267,
      "grad_norm": 0.8892746567726135,
      "learning_rate": 9.991263176886139e-06,
      "loss": 0.9011,
      "step": 782
    },
    {
      "epoch": 0.04309538224448236,
      "grad_norm": 1.097734808921814,
      "learning_rate": 9.99123754441172e-06,
      "loss": 1.009,
      "step": 783
    },
    {
      "epoch": 0.04315042104683802,
      "grad_norm": 1.0065964460372925,
      "learning_rate": 9.991211874424497e-06,
      "loss": 0.9492,
      "step": 784
    },
    {
      "epoch": 0.04320545984919368,
      "grad_norm": 1.0791678428649902,
      "learning_rate": 9.99118616692466e-06,
      "loss": 1.0142,
      "step": 785
    },
    {
      "epoch": 0.043260498651549344,
      "grad_norm": 0.9454777836799622,
      "learning_rate": 9.991160421912404e-06,
      "loss": 0.8058,
      "step": 786
    },
    {
      "epoch": 0.043315537453905005,
      "grad_norm": 0.9448156952857971,
      "learning_rate": 9.991134639387922e-06,
      "loss": 0.8184,
      "step": 787
    },
    {
      "epoch": 0.043370576256260666,
      "grad_norm": 0.9636550545692444,
      "learning_rate": 9.99110881935141e-06,
      "loss": 0.8606,
      "step": 788
    },
    {
      "epoch": 0.04342561505861633,
      "grad_norm": 0.9933613538742065,
      "learning_rate": 9.991082961803058e-06,
      "loss": 0.9449,
      "step": 789
    },
    {
      "epoch": 0.04348065386097198,
      "grad_norm": 0.8906797170639038,
      "learning_rate": 9.991057066743065e-06,
      "loss": 0.8053,
      "step": 790
    },
    {
      "epoch": 0.043535692663327644,
      "grad_norm": 1.0393906831741333,
      "learning_rate": 9.991031134171621e-06,
      "loss": 0.8487,
      "step": 791
    },
    {
      "epoch": 0.043590731465683305,
      "grad_norm": 1.0618231296539307,
      "learning_rate": 9.991005164088923e-06,
      "loss": 0.9847,
      "step": 792
    },
    {
      "epoch": 0.043645770268038966,
      "grad_norm": 0.9525149464607239,
      "learning_rate": 9.990979156495167e-06,
      "loss": 0.9318,
      "step": 793
    },
    {
      "epoch": 0.04370080907039463,
      "grad_norm": 0.9430851936340332,
      "learning_rate": 9.990953111390546e-06,
      "loss": 0.8483,
      "step": 794
    },
    {
      "epoch": 0.04375584787275029,
      "grad_norm": 0.9259672164916992,
      "learning_rate": 9.99092702877526e-06,
      "loss": 0.9365,
      "step": 795
    },
    {
      "epoch": 0.04381088667510595,
      "grad_norm": 0.942609965801239,
      "learning_rate": 9.9909009086495e-06,
      "loss": 0.8408,
      "step": 796
    },
    {
      "epoch": 0.04386592547746161,
      "grad_norm": 0.939255952835083,
      "learning_rate": 9.990874751013467e-06,
      "loss": 0.8749,
      "step": 797
    },
    {
      "epoch": 0.04392096427981727,
      "grad_norm": 1.1701711416244507,
      "learning_rate": 9.990848555867353e-06,
      "loss": 0.9312,
      "step": 798
    },
    {
      "epoch": 0.043976003082172935,
      "grad_norm": 1.0441124439239502,
      "learning_rate": 9.990822323211358e-06,
      "loss": 0.8618,
      "step": 799
    },
    {
      "epoch": 0.04403104188452859,
      "grad_norm": 0.9601489305496216,
      "learning_rate": 9.990796053045679e-06,
      "loss": 0.9569,
      "step": 800
    },
    {
      "epoch": 0.04408608068688425,
      "grad_norm": 0.9394032955169678,
      "learning_rate": 9.990769745370513e-06,
      "loss": 0.846,
      "step": 801
    },
    {
      "epoch": 0.04414111948923991,
      "grad_norm": 0.9631348252296448,
      "learning_rate": 9.990743400186056e-06,
      "loss": 0.8754,
      "step": 802
    },
    {
      "epoch": 0.04419615829159557,
      "grad_norm": 0.9234963059425354,
      "learning_rate": 9.990717017492508e-06,
      "loss": 0.8613,
      "step": 803
    },
    {
      "epoch": 0.044251197093951235,
      "grad_norm": 0.9169090390205383,
      "learning_rate": 9.990690597290069e-06,
      "loss": 0.8867,
      "step": 804
    },
    {
      "epoch": 0.044306235896306896,
      "grad_norm": 1.0194867849349976,
      "learning_rate": 9.990664139578933e-06,
      "loss": 0.8675,
      "step": 805
    },
    {
      "epoch": 0.04436127469866256,
      "grad_norm": 1.3226114511489868,
      "learning_rate": 9.990637644359302e-06,
      "loss": 0.997,
      "step": 806
    },
    {
      "epoch": 0.04441631350101822,
      "grad_norm": 0.8904317617416382,
      "learning_rate": 9.990611111631374e-06,
      "loss": 0.7274,
      "step": 807
    },
    {
      "epoch": 0.04447135230337388,
      "grad_norm": 0.8909007906913757,
      "learning_rate": 9.99058454139535e-06,
      "loss": 0.8141,
      "step": 808
    },
    {
      "epoch": 0.04452639110572954,
      "grad_norm": 1.004015564918518,
      "learning_rate": 9.990557933651429e-06,
      "loss": 0.9883,
      "step": 809
    },
    {
      "epoch": 0.0445814299080852,
      "grad_norm": 1.1215732097625732,
      "learning_rate": 9.990531288399807e-06,
      "loss": 0.9355,
      "step": 810
    },
    {
      "epoch": 0.04463646871044086,
      "grad_norm": 1.0545012950897217,
      "learning_rate": 9.99050460564069e-06,
      "loss": 0.9532,
      "step": 811
    },
    {
      "epoch": 0.04469150751279652,
      "grad_norm": 0.9608867168426514,
      "learning_rate": 9.990477885374277e-06,
      "loss": 0.9363,
      "step": 812
    },
    {
      "epoch": 0.04474654631515218,
      "grad_norm": 0.8750461935997009,
      "learning_rate": 9.990451127600766e-06,
      "loss": 0.7343,
      "step": 813
    },
    {
      "epoch": 0.04480158511750784,
      "grad_norm": 0.891740620136261,
      "learning_rate": 9.99042433232036e-06,
      "loss": 0.8541,
      "step": 814
    },
    {
      "epoch": 0.0448566239198635,
      "grad_norm": 1.1520029306411743,
      "learning_rate": 9.990397499533264e-06,
      "loss": 0.7696,
      "step": 815
    },
    {
      "epoch": 0.044911662722219164,
      "grad_norm": 0.9526278972625732,
      "learning_rate": 9.990370629239673e-06,
      "loss": 0.8953,
      "step": 816
    },
    {
      "epoch": 0.044966701524574826,
      "grad_norm": 0.9218434691429138,
      "learning_rate": 9.990343721439795e-06,
      "loss": 0.8198,
      "step": 817
    },
    {
      "epoch": 0.04502174032693049,
      "grad_norm": 0.8502745628356934,
      "learning_rate": 9.990316776133827e-06,
      "loss": 0.8035,
      "step": 818
    },
    {
      "epoch": 0.04507677912928615,
      "grad_norm": 0.8861565589904785,
      "learning_rate": 9.990289793321975e-06,
      "loss": 0.8626,
      "step": 819
    },
    {
      "epoch": 0.04513181793164181,
      "grad_norm": 1.1113256216049194,
      "learning_rate": 9.99026277300444e-06,
      "loss": 0.9363,
      "step": 820
    },
    {
      "epoch": 0.04518685673399747,
      "grad_norm": 0.9984708428382874,
      "learning_rate": 9.990235715181426e-06,
      "loss": 1.0376,
      "step": 821
    },
    {
      "epoch": 0.045241895536353126,
      "grad_norm": 0.9026711583137512,
      "learning_rate": 9.990208619853137e-06,
      "loss": 0.9079,
      "step": 822
    },
    {
      "epoch": 0.04529693433870879,
      "grad_norm": 0.8724965453147888,
      "learning_rate": 9.990181487019775e-06,
      "loss": 0.8665,
      "step": 823
    },
    {
      "epoch": 0.04535197314106445,
      "grad_norm": 0.8923047780990601,
      "learning_rate": 9.990154316681543e-06,
      "loss": 0.7779,
      "step": 824
    },
    {
      "epoch": 0.04540701194342011,
      "grad_norm": 0.9024640321731567,
      "learning_rate": 9.99012710883865e-06,
      "loss": 0.8859,
      "step": 825
    },
    {
      "epoch": 0.04546205074577577,
      "grad_norm": 0.9245888590812683,
      "learning_rate": 9.990099863491296e-06,
      "loss": 0.8501,
      "step": 826
    },
    {
      "epoch": 0.04551708954813143,
      "grad_norm": 0.9257050156593323,
      "learning_rate": 9.990072580639687e-06,
      "loss": 0.9561,
      "step": 827
    },
    {
      "epoch": 0.045572128350487094,
      "grad_norm": 0.995610773563385,
      "learning_rate": 9.99004526028403e-06,
      "loss": 0.917,
      "step": 828
    },
    {
      "epoch": 0.045627167152842756,
      "grad_norm": 0.9524009823799133,
      "learning_rate": 9.990017902424525e-06,
      "loss": 0.9184,
      "step": 829
    },
    {
      "epoch": 0.04568220595519842,
      "grad_norm": 0.9264503121376038,
      "learning_rate": 9.989990507061385e-06,
      "loss": 0.8615,
      "step": 830
    },
    {
      "epoch": 0.04573724475755408,
      "grad_norm": 1.0068570375442505,
      "learning_rate": 9.989963074194809e-06,
      "loss": 0.8331,
      "step": 831
    },
    {
      "epoch": 0.04579228355990974,
      "grad_norm": 0.9295952320098877,
      "learning_rate": 9.989935603825009e-06,
      "loss": 0.8387,
      "step": 832
    },
    {
      "epoch": 0.045847322362265394,
      "grad_norm": 1.0408827066421509,
      "learning_rate": 9.989908095952186e-06,
      "loss": 0.9686,
      "step": 833
    },
    {
      "epoch": 0.045902361164621056,
      "grad_norm": 0.8874136209487915,
      "learning_rate": 9.989880550576551e-06,
      "loss": 0.815,
      "step": 834
    },
    {
      "epoch": 0.04595739996697672,
      "grad_norm": 0.9898836016654968,
      "learning_rate": 9.989852967698311e-06,
      "loss": 0.9458,
      "step": 835
    },
    {
      "epoch": 0.04601243876933238,
      "grad_norm": 0.9828970432281494,
      "learning_rate": 9.989825347317668e-06,
      "loss": 0.7922,
      "step": 836
    },
    {
      "epoch": 0.04606747757168804,
      "grad_norm": 1.025447964668274,
      "learning_rate": 9.989797689434836e-06,
      "loss": 0.9349,
      "step": 837
    },
    {
      "epoch": 0.0461225163740437,
      "grad_norm": 0.8623831272125244,
      "learning_rate": 9.98976999405002e-06,
      "loss": 0.8786,
      "step": 838
    },
    {
      "epoch": 0.04617755517639936,
      "grad_norm": 0.9614997506141663,
      "learning_rate": 9.98974226116343e-06,
      "loss": 0.7885,
      "step": 839
    },
    {
      "epoch": 0.046232593978755024,
      "grad_norm": 1.0207616090774536,
      "learning_rate": 9.989714490775269e-06,
      "loss": 0.9786,
      "step": 840
    },
    {
      "epoch": 0.046287632781110685,
      "grad_norm": 0.8509595990180969,
      "learning_rate": 9.98968668288575e-06,
      "loss": 0.7312,
      "step": 841
    },
    {
      "epoch": 0.04634267158346635,
      "grad_norm": 0.9822607040405273,
      "learning_rate": 9.989658837495084e-06,
      "loss": 0.952,
      "step": 842
    },
    {
      "epoch": 0.046397710385822,
      "grad_norm": 1.0058252811431885,
      "learning_rate": 9.989630954603477e-06,
      "loss": 0.8811,
      "step": 843
    },
    {
      "epoch": 0.04645274918817766,
      "grad_norm": 1.0146985054016113,
      "learning_rate": 9.989603034211139e-06,
      "loss": 0.9051,
      "step": 844
    },
    {
      "epoch": 0.046507787990533324,
      "grad_norm": 0.8976503610610962,
      "learning_rate": 9.98957507631828e-06,
      "loss": 0.879,
      "step": 845
    },
    {
      "epoch": 0.046562826792888985,
      "grad_norm": 0.8791939616203308,
      "learning_rate": 9.989547080925111e-06,
      "loss": 0.8944,
      "step": 846
    },
    {
      "epoch": 0.04661786559524465,
      "grad_norm": 0.8530884981155396,
      "learning_rate": 9.989519048031842e-06,
      "loss": 0.9029,
      "step": 847
    },
    {
      "epoch": 0.04667290439760031,
      "grad_norm": 0.9621617197990417,
      "learning_rate": 9.989490977638683e-06,
      "loss": 0.8374,
      "step": 848
    },
    {
      "epoch": 0.04672794319995597,
      "grad_norm": 0.9629075527191162,
      "learning_rate": 9.989462869745845e-06,
      "loss": 0.9032,
      "step": 849
    },
    {
      "epoch": 0.04678298200231163,
      "grad_norm": 1.3256126642227173,
      "learning_rate": 9.989434724353541e-06,
      "loss": 0.9748,
      "step": 850
    },
    {
      "epoch": 0.04683802080466729,
      "grad_norm": 1.0230494737625122,
      "learning_rate": 9.989406541461979e-06,
      "loss": 0.9752,
      "step": 851
    },
    {
      "epoch": 0.046893059607022954,
      "grad_norm": 0.8454533219337463,
      "learning_rate": 9.989378321071375e-06,
      "loss": 0.8426,
      "step": 852
    },
    {
      "epoch": 0.046948098409378615,
      "grad_norm": 0.9995863437652588,
      "learning_rate": 9.989350063181939e-06,
      "loss": 0.9955,
      "step": 853
    },
    {
      "epoch": 0.04700313721173427,
      "grad_norm": 0.8956604599952698,
      "learning_rate": 9.989321767793883e-06,
      "loss": 0.9024,
      "step": 854
    },
    {
      "epoch": 0.04705817601408993,
      "grad_norm": 1.0123292207717896,
      "learning_rate": 9.989293434907419e-06,
      "loss": 0.7856,
      "step": 855
    },
    {
      "epoch": 0.04711321481644559,
      "grad_norm": 0.814577043056488,
      "learning_rate": 9.989265064522762e-06,
      "loss": 0.8377,
      "step": 856
    },
    {
      "epoch": 0.047168253618801254,
      "grad_norm": 1.1571552753448486,
      "learning_rate": 9.989236656640125e-06,
      "loss": 0.8562,
      "step": 857
    },
    {
      "epoch": 0.047223292421156915,
      "grad_norm": 0.9681577682495117,
      "learning_rate": 9.98920821125972e-06,
      "loss": 0.8473,
      "step": 858
    },
    {
      "epoch": 0.047278331223512576,
      "grad_norm": 0.9680121541023254,
      "learning_rate": 9.989179728381761e-06,
      "loss": 0.9811,
      "step": 859
    },
    {
      "epoch": 0.04733337002586824,
      "grad_norm": 0.985477089881897,
      "learning_rate": 9.989151208006464e-06,
      "loss": 0.6994,
      "step": 860
    },
    {
      "epoch": 0.0473884088282239,
      "grad_norm": 0.8612962365150452,
      "learning_rate": 9.98912265013404e-06,
      "loss": 0.7667,
      "step": 861
    },
    {
      "epoch": 0.04744344763057956,
      "grad_norm": 0.8884604573249817,
      "learning_rate": 9.989094054764708e-06,
      "loss": 0.8382,
      "step": 862
    },
    {
      "epoch": 0.04749848643293522,
      "grad_norm": 1.036881923675537,
      "learning_rate": 9.989065421898681e-06,
      "loss": 0.8748,
      "step": 863
    },
    {
      "epoch": 0.04755352523529088,
      "grad_norm": 0.9954493045806885,
      "learning_rate": 9.989036751536171e-06,
      "loss": 0.9174,
      "step": 864
    },
    {
      "epoch": 0.04760856403764654,
      "grad_norm": 0.9984694123268127,
      "learning_rate": 9.989008043677399e-06,
      "loss": 0.7636,
      "step": 865
    },
    {
      "epoch": 0.0476636028400022,
      "grad_norm": 1.0412588119506836,
      "learning_rate": 9.988979298322576e-06,
      "loss": 0.773,
      "step": 866
    },
    {
      "epoch": 0.04771864164235786,
      "grad_norm": 0.8034874796867371,
      "learning_rate": 9.98895051547192e-06,
      "loss": 0.7914,
      "step": 867
    },
    {
      "epoch": 0.04777368044471352,
      "grad_norm": 0.8983979225158691,
      "learning_rate": 9.988921695125648e-06,
      "loss": 0.7292,
      "step": 868
    },
    {
      "epoch": 0.04782871924706918,
      "grad_norm": 0.9445077776908875,
      "learning_rate": 9.988892837283976e-06,
      "loss": 0.8263,
      "step": 869
    },
    {
      "epoch": 0.047883758049424845,
      "grad_norm": 1.0753306150436401,
      "learning_rate": 9.988863941947121e-06,
      "loss": 1.1122,
      "step": 870
    },
    {
      "epoch": 0.047938796851780506,
      "grad_norm": 1.0091484785079956,
      "learning_rate": 9.9888350091153e-06,
      "loss": 0.9276,
      "step": 871
    },
    {
      "epoch": 0.04799383565413617,
      "grad_norm": 1.0977306365966797,
      "learning_rate": 9.988806038788732e-06,
      "loss": 0.854,
      "step": 872
    },
    {
      "epoch": 0.04804887445649183,
      "grad_norm": 1.0285007953643799,
      "learning_rate": 9.988777030967632e-06,
      "loss": 0.9441,
      "step": 873
    },
    {
      "epoch": 0.04810391325884749,
      "grad_norm": 0.8973976373672485,
      "learning_rate": 9.988747985652218e-06,
      "loss": 0.786,
      "step": 874
    },
    {
      "epoch": 0.04815895206120315,
      "grad_norm": 0.9809553623199463,
      "learning_rate": 9.98871890284271e-06,
      "loss": 0.9042,
      "step": 875
    },
    {
      "epoch": 0.048213990863558806,
      "grad_norm": 0.8514279723167419,
      "learning_rate": 9.988689782539326e-06,
      "loss": 0.7874,
      "step": 876
    },
    {
      "epoch": 0.04826902966591447,
      "grad_norm": 0.8299674391746521,
      "learning_rate": 9.988660624742286e-06,
      "loss": 0.8704,
      "step": 877
    },
    {
      "epoch": 0.04832406846827013,
      "grad_norm": 0.9862462282180786,
      "learning_rate": 9.988631429451809e-06,
      "loss": 0.9963,
      "step": 878
    },
    {
      "epoch": 0.04837910727062579,
      "grad_norm": 0.9041131734848022,
      "learning_rate": 9.988602196668111e-06,
      "loss": 0.9207,
      "step": 879
    },
    {
      "epoch": 0.04843414607298145,
      "grad_norm": 0.8597276210784912,
      "learning_rate": 9.988572926391416e-06,
      "loss": 0.8226,
      "step": 880
    },
    {
      "epoch": 0.04848918487533711,
      "grad_norm": 0.9494329690933228,
      "learning_rate": 9.988543618621941e-06,
      "loss": 0.8834,
      "step": 881
    },
    {
      "epoch": 0.048544223677692774,
      "grad_norm": 0.9129118323326111,
      "learning_rate": 9.98851427335991e-06,
      "loss": 0.7819,
      "step": 882
    },
    {
      "epoch": 0.048599262480048436,
      "grad_norm": 0.9145999550819397,
      "learning_rate": 9.988484890605539e-06,
      "loss": 0.885,
      "step": 883
    },
    {
      "epoch": 0.0486543012824041,
      "grad_norm": 1.0115307569503784,
      "learning_rate": 9.98845547035905e-06,
      "loss": 0.8347,
      "step": 884
    },
    {
      "epoch": 0.04870934008475976,
      "grad_norm": 1.1372706890106201,
      "learning_rate": 9.988426012620667e-06,
      "loss": 0.944,
      "step": 885
    },
    {
      "epoch": 0.04876437888711541,
      "grad_norm": 0.9502811431884766,
      "learning_rate": 9.98839651739061e-06,
      "loss": 0.9054,
      "step": 886
    },
    {
      "epoch": 0.048819417689471074,
      "grad_norm": 0.9612823128700256,
      "learning_rate": 9.988366984669097e-06,
      "loss": 0.8796,
      "step": 887
    },
    {
      "epoch": 0.048874456491826736,
      "grad_norm": 0.9551461935043335,
      "learning_rate": 9.988337414456355e-06,
      "loss": 0.8769,
      "step": 888
    },
    {
      "epoch": 0.0489294952941824,
      "grad_norm": 0.8554086089134216,
      "learning_rate": 9.988307806752603e-06,
      "loss": 0.892,
      "step": 889
    },
    {
      "epoch": 0.04898453409653806,
      "grad_norm": 0.8418886661529541,
      "learning_rate": 9.988278161558067e-06,
      "loss": 0.7568,
      "step": 890
    },
    {
      "epoch": 0.04903957289889372,
      "grad_norm": 1.4780360460281372,
      "learning_rate": 9.988248478872967e-06,
      "loss": 0.9126,
      "step": 891
    },
    {
      "epoch": 0.04909461170124938,
      "grad_norm": 0.8236714005470276,
      "learning_rate": 9.988218758697526e-06,
      "loss": 0.7317,
      "step": 892
    },
    {
      "epoch": 0.04914965050360504,
      "grad_norm": 0.8777141571044922,
      "learning_rate": 9.988189001031968e-06,
      "loss": 0.7989,
      "step": 893
    },
    {
      "epoch": 0.049204689305960704,
      "grad_norm": 1.0235031843185425,
      "learning_rate": 9.988159205876516e-06,
      "loss": 0.8335,
      "step": 894
    },
    {
      "epoch": 0.049259728108316365,
      "grad_norm": 0.9340357184410095,
      "learning_rate": 9.988129373231395e-06,
      "loss": 0.8129,
      "step": 895
    },
    {
      "epoch": 0.04931476691067203,
      "grad_norm": 1.7686667442321777,
      "learning_rate": 9.98809950309683e-06,
      "loss": 0.9792,
      "step": 896
    },
    {
      "epoch": 0.04936980571302768,
      "grad_norm": 0.9252369403839111,
      "learning_rate": 9.988069595473044e-06,
      "loss": 0.8671,
      "step": 897
    },
    {
      "epoch": 0.04942484451538334,
      "grad_norm": 0.9989960789680481,
      "learning_rate": 9.988039650360262e-06,
      "loss": 0.9245,
      "step": 898
    },
    {
      "epoch": 0.049479883317739004,
      "grad_norm": 1.062912106513977,
      "learning_rate": 9.98800966775871e-06,
      "loss": 0.9146,
      "step": 899
    },
    {
      "epoch": 0.049534922120094665,
      "grad_norm": 0.8698169589042664,
      "learning_rate": 9.98797964766861e-06,
      "loss": 0.8606,
      "step": 900
    },
    {
      "epoch": 0.04958996092245033,
      "grad_norm": 1.6754224300384521,
      "learning_rate": 9.98794959009019e-06,
      "loss": 0.9236,
      "step": 901
    },
    {
      "epoch": 0.04964499972480599,
      "grad_norm": 1.084174394607544,
      "learning_rate": 9.98791949502368e-06,
      "loss": 0.9252,
      "step": 902
    },
    {
      "epoch": 0.04970003852716165,
      "grad_norm": 0.9866724610328674,
      "learning_rate": 9.987889362469301e-06,
      "loss": 0.9096,
      "step": 903
    },
    {
      "epoch": 0.04975507732951731,
      "grad_norm": 0.8814040422439575,
      "learning_rate": 9.987859192427279e-06,
      "loss": 0.8475,
      "step": 904
    },
    {
      "epoch": 0.04981011613187297,
      "grad_norm": 0.8796457052230835,
      "learning_rate": 9.987828984897843e-06,
      "loss": 0.8478,
      "step": 905
    },
    {
      "epoch": 0.049865154934228634,
      "grad_norm": 1.0541884899139404,
      "learning_rate": 9.98779873988122e-06,
      "loss": 0.9799,
      "step": 906
    },
    {
      "epoch": 0.049920193736584295,
      "grad_norm": 0.91409832239151,
      "learning_rate": 9.987768457377636e-06,
      "loss": 0.8701,
      "step": 907
    },
    {
      "epoch": 0.04997523253893995,
      "grad_norm": 1.0120370388031006,
      "learning_rate": 9.98773813738732e-06,
      "loss": 0.8417,
      "step": 908
    },
    {
      "epoch": 0.05003027134129561,
      "grad_norm": 1.7744206190109253,
      "learning_rate": 9.987707779910499e-06,
      "loss": 0.9263,
      "step": 909
    },
    {
      "epoch": 0.05008531014365127,
      "grad_norm": 0.9423969388008118,
      "learning_rate": 9.987677384947402e-06,
      "loss": 0.9667,
      "step": 910
    },
    {
      "epoch": 0.050140348946006934,
      "grad_norm": 1.5940319299697876,
      "learning_rate": 9.987646952498256e-06,
      "loss": 0.9223,
      "step": 911
    },
    {
      "epoch": 0.050195387748362595,
      "grad_norm": 0.941792368888855,
      "learning_rate": 9.987616482563292e-06,
      "loss": 0.895,
      "step": 912
    },
    {
      "epoch": 0.05025042655071826,
      "grad_norm": 3.1945221424102783,
      "learning_rate": 9.987585975142738e-06,
      "loss": 0.837,
      "step": 913
    },
    {
      "epoch": 0.05030546535307392,
      "grad_norm": 2.0819199085235596,
      "learning_rate": 9.98755543023682e-06,
      "loss": 0.918,
      "step": 914
    },
    {
      "epoch": 0.05036050415542958,
      "grad_norm": 0.984282910823822,
      "learning_rate": 9.987524847845773e-06,
      "loss": 0.8589,
      "step": 915
    },
    {
      "epoch": 0.05041554295778524,
      "grad_norm": 0.9021026492118835,
      "learning_rate": 9.987494227969823e-06,
      "loss": 0.9053,
      "step": 916
    },
    {
      "epoch": 0.0504705817601409,
      "grad_norm": 2.6515462398529053,
      "learning_rate": 9.9874635706092e-06,
      "loss": 0.8874,
      "step": 917
    },
    {
      "epoch": 0.050525620562496563,
      "grad_norm": 0.8893095254898071,
      "learning_rate": 9.98743287576414e-06,
      "loss": 0.8259,
      "step": 918
    },
    {
      "epoch": 0.05058065936485222,
      "grad_norm": 0.9897775650024414,
      "learning_rate": 9.987402143434868e-06,
      "loss": 0.877,
      "step": 919
    },
    {
      "epoch": 0.05063569816720788,
      "grad_norm": 0.9391944408416748,
      "learning_rate": 9.987371373621614e-06,
      "loss": 0.9363,
      "step": 920
    },
    {
      "epoch": 0.05069073696956354,
      "grad_norm": 0.9585913419723511,
      "learning_rate": 9.987340566324615e-06,
      "loss": 0.8704,
      "step": 921
    },
    {
      "epoch": 0.0507457757719192,
      "grad_norm": 0.9210980534553528,
      "learning_rate": 9.987309721544098e-06,
      "loss": 0.9321,
      "step": 922
    },
    {
      "epoch": 0.05080081457427486,
      "grad_norm": 1.0713307857513428,
      "learning_rate": 9.987278839280295e-06,
      "loss": 0.9489,
      "step": 923
    },
    {
      "epoch": 0.050855853376630525,
      "grad_norm": 1.0178636312484741,
      "learning_rate": 9.98724791953344e-06,
      "loss": 0.853,
      "step": 924
    },
    {
      "epoch": 0.050910892178986186,
      "grad_norm": 0.9782636761665344,
      "learning_rate": 9.987216962303766e-06,
      "loss": 0.924,
      "step": 925
    },
    {
      "epoch": 0.05096593098134185,
      "grad_norm": 0.9474522471427917,
      "learning_rate": 9.987185967591503e-06,
      "loss": 0.8619,
      "step": 926
    },
    {
      "epoch": 0.05102096978369751,
      "grad_norm": 1.1875778436660767,
      "learning_rate": 9.987154935396885e-06,
      "loss": 1.012,
      "step": 927
    },
    {
      "epoch": 0.05107600858605317,
      "grad_norm": 1.0585243701934814,
      "learning_rate": 9.987123865720147e-06,
      "loss": 0.9019,
      "step": 928
    },
    {
      "epoch": 0.051131047388408825,
      "grad_norm": 0.9848800897598267,
      "learning_rate": 9.98709275856152e-06,
      "loss": 0.9665,
      "step": 929
    },
    {
      "epoch": 0.051186086190764486,
      "grad_norm": 1.04201078414917,
      "learning_rate": 9.987061613921238e-06,
      "loss": 0.9269,
      "step": 930
    },
    {
      "epoch": 0.05124112499312015,
      "grad_norm": 1.1748600006103516,
      "learning_rate": 9.987030431799537e-06,
      "loss": 0.8565,
      "step": 931
    },
    {
      "epoch": 0.05129616379547581,
      "grad_norm": 1.879232406616211,
      "learning_rate": 9.98699921219665e-06,
      "loss": 0.8535,
      "step": 932
    },
    {
      "epoch": 0.05135120259783147,
      "grad_norm": 0.9837847948074341,
      "learning_rate": 9.986967955112812e-06,
      "loss": 0.927,
      "step": 933
    },
    {
      "epoch": 0.05140624140018713,
      "grad_norm": 0.8637211918830872,
      "learning_rate": 9.986936660548257e-06,
      "loss": 0.7903,
      "step": 934
    },
    {
      "epoch": 0.05146128020254279,
      "grad_norm": 0.9078792929649353,
      "learning_rate": 9.986905328503222e-06,
      "loss": 0.9135,
      "step": 935
    },
    {
      "epoch": 0.051516319004898455,
      "grad_norm": 0.9763005971908569,
      "learning_rate": 9.98687395897794e-06,
      "loss": 0.9006,
      "step": 936
    },
    {
      "epoch": 0.051571357807254116,
      "grad_norm": 1.0174345970153809,
      "learning_rate": 9.98684255197265e-06,
      "loss": 0.9294,
      "step": 937
    },
    {
      "epoch": 0.05162639660960978,
      "grad_norm": 0.8709769248962402,
      "learning_rate": 9.986811107487584e-06,
      "loss": 0.7986,
      "step": 938
    },
    {
      "epoch": 0.05168143541196544,
      "grad_norm": 0.8717525601387024,
      "learning_rate": 9.986779625522983e-06,
      "loss": 0.8705,
      "step": 939
    },
    {
      "epoch": 0.05173647421432109,
      "grad_norm": 0.9682945013046265,
      "learning_rate": 9.98674810607908e-06,
      "loss": 0.8127,
      "step": 940
    },
    {
      "epoch": 0.051791513016676755,
      "grad_norm": 1.0248037576675415,
      "learning_rate": 9.986716549156113e-06,
      "loss": 0.9217,
      "step": 941
    },
    {
      "epoch": 0.051846551819032416,
      "grad_norm": 0.9883397221565247,
      "learning_rate": 9.98668495475432e-06,
      "loss": 0.853,
      "step": 942
    },
    {
      "epoch": 0.05190159062138808,
      "grad_norm": 0.9271108508110046,
      "learning_rate": 9.986653322873937e-06,
      "loss": 0.8807,
      "step": 943
    },
    {
      "epoch": 0.05195662942374374,
      "grad_norm": 0.9027101397514343,
      "learning_rate": 9.986621653515203e-06,
      "loss": 0.88,
      "step": 944
    },
    {
      "epoch": 0.0520116682260994,
      "grad_norm": 0.9807021617889404,
      "learning_rate": 9.986589946678354e-06,
      "loss": 0.8922,
      "step": 945
    },
    {
      "epoch": 0.05206670702845506,
      "grad_norm": 0.8779157400131226,
      "learning_rate": 9.98655820236363e-06,
      "loss": 0.8988,
      "step": 946
    },
    {
      "epoch": 0.05212174583081072,
      "grad_norm": 0.8182910680770874,
      "learning_rate": 9.986526420571272e-06,
      "loss": 0.7534,
      "step": 947
    },
    {
      "epoch": 0.052176784633166384,
      "grad_norm": 0.9205981492996216,
      "learning_rate": 9.986494601301513e-06,
      "loss": 0.7516,
      "step": 948
    },
    {
      "epoch": 0.052231823435522046,
      "grad_norm": 0.9829681515693665,
      "learning_rate": 9.986462744554598e-06,
      "loss": 0.9358,
      "step": 949
    },
    {
      "epoch": 0.05228686223787771,
      "grad_norm": 0.8869890570640564,
      "learning_rate": 9.986430850330762e-06,
      "loss": 0.7933,
      "step": 950
    },
    {
      "epoch": 0.05234190104023336,
      "grad_norm": 1.0226716995239258,
      "learning_rate": 9.986398918630248e-06,
      "loss": 0.9523,
      "step": 951
    },
    {
      "epoch": 0.05239693984258902,
      "grad_norm": 0.9549778699874878,
      "learning_rate": 9.986366949453293e-06,
      "loss": 0.9368,
      "step": 952
    },
    {
      "epoch": 0.052451978644944684,
      "grad_norm": 0.860454797744751,
      "learning_rate": 9.98633494280014e-06,
      "loss": 0.7618,
      "step": 953
    },
    {
      "epoch": 0.052507017447300346,
      "grad_norm": 0.9623841643333435,
      "learning_rate": 9.986302898671027e-06,
      "loss": 0.8356,
      "step": 954
    },
    {
      "epoch": 0.05256205624965601,
      "grad_norm": 0.9236606359481812,
      "learning_rate": 9.986270817066196e-06,
      "loss": 0.921,
      "step": 955
    },
    {
      "epoch": 0.05261709505201167,
      "grad_norm": 1.0599812269210815,
      "learning_rate": 9.98623869798589e-06,
      "loss": 0.8082,
      "step": 956
    },
    {
      "epoch": 0.05267213385436733,
      "grad_norm": 1.0321687459945679,
      "learning_rate": 9.986206541430347e-06,
      "loss": 0.9001,
      "step": 957
    },
    {
      "epoch": 0.05272717265672299,
      "grad_norm": 0.8884543776512146,
      "learning_rate": 9.986174347399813e-06,
      "loss": 0.8317,
      "step": 958
    },
    {
      "epoch": 0.05278221145907865,
      "grad_norm": 0.9592668414115906,
      "learning_rate": 9.986142115894526e-06,
      "loss": 0.9955,
      "step": 959
    },
    {
      "epoch": 0.052837250261434314,
      "grad_norm": 0.9604032039642334,
      "learning_rate": 9.986109846914729e-06,
      "loss": 0.876,
      "step": 960
    },
    {
      "epoch": 0.052892289063789975,
      "grad_norm": 0.9837536811828613,
      "learning_rate": 9.986077540460664e-06,
      "loss": 0.8247,
      "step": 961
    },
    {
      "epoch": 0.05294732786614563,
      "grad_norm": 0.8570861220359802,
      "learning_rate": 9.986045196532576e-06,
      "loss": 0.879,
      "step": 962
    },
    {
      "epoch": 0.05300236666850129,
      "grad_norm": 0.8441471457481384,
      "learning_rate": 9.986012815130708e-06,
      "loss": 0.8979,
      "step": 963
    },
    {
      "epoch": 0.05305740547085695,
      "grad_norm": 0.8976197838783264,
      "learning_rate": 9.985980396255302e-06,
      "loss": 0.9382,
      "step": 964
    },
    {
      "epoch": 0.053112444273212614,
      "grad_norm": 0.9685307741165161,
      "learning_rate": 9.985947939906599e-06,
      "loss": 0.8627,
      "step": 965
    },
    {
      "epoch": 0.053167483075568275,
      "grad_norm": 0.8939018249511719,
      "learning_rate": 9.98591544608485e-06,
      "loss": 0.9221,
      "step": 966
    },
    {
      "epoch": 0.05322252187792394,
      "grad_norm": 0.9218310713768005,
      "learning_rate": 9.985882914790292e-06,
      "loss": 0.8356,
      "step": 967
    },
    {
      "epoch": 0.0532775606802796,
      "grad_norm": 0.9342261552810669,
      "learning_rate": 9.985850346023174e-06,
      "loss": 0.971,
      "step": 968
    },
    {
      "epoch": 0.05333259948263526,
      "grad_norm": 1.0860705375671387,
      "learning_rate": 9.985817739783741e-06,
      "loss": 0.906,
      "step": 969
    },
    {
      "epoch": 0.05338763828499092,
      "grad_norm": 0.8675006031990051,
      "learning_rate": 9.985785096072234e-06,
      "loss": 0.906,
      "step": 970
    },
    {
      "epoch": 0.05344267708734658,
      "grad_norm": 0.8170626163482666,
      "learning_rate": 9.985752414888903e-06,
      "loss": 0.8109,
      "step": 971
    },
    {
      "epoch": 0.05349771588970224,
      "grad_norm": 0.936434805393219,
      "learning_rate": 9.98571969623399e-06,
      "loss": 0.9219,
      "step": 972
    },
    {
      "epoch": 0.0535527546920579,
      "grad_norm": 0.9316715002059937,
      "learning_rate": 9.985686940107741e-06,
      "loss": 0.8569,
      "step": 973
    },
    {
      "epoch": 0.05360779349441356,
      "grad_norm": 1.183008074760437,
      "learning_rate": 9.985654146510405e-06,
      "loss": 0.837,
      "step": 974
    },
    {
      "epoch": 0.05366283229676922,
      "grad_norm": 1.0788745880126953,
      "learning_rate": 9.98562131544223e-06,
      "loss": 0.8822,
      "step": 975
    },
    {
      "epoch": 0.05371787109912488,
      "grad_norm": 0.9285461902618408,
      "learning_rate": 9.985588446903455e-06,
      "loss": 0.9279,
      "step": 976
    },
    {
      "epoch": 0.053772909901480544,
      "grad_norm": 0.9389022588729858,
      "learning_rate": 9.985555540894334e-06,
      "loss": 0.9839,
      "step": 977
    },
    {
      "epoch": 0.053827948703836205,
      "grad_norm": 0.8920616507530212,
      "learning_rate": 9.985522597415112e-06,
      "loss": 0.9205,
      "step": 978
    },
    {
      "epoch": 0.053882987506191866,
      "grad_norm": 0.9755093455314636,
      "learning_rate": 9.985489616466035e-06,
      "loss": 0.8987,
      "step": 979
    },
    {
      "epoch": 0.05393802630854753,
      "grad_norm": 0.96027010679245,
      "learning_rate": 9.985456598047356e-06,
      "loss": 0.8543,
      "step": 980
    },
    {
      "epoch": 0.05399306511090319,
      "grad_norm": 1.0489718914031982,
      "learning_rate": 9.985423542159317e-06,
      "loss": 0.9179,
      "step": 981
    },
    {
      "epoch": 0.05404810391325885,
      "grad_norm": 0.8665526509284973,
      "learning_rate": 9.985390448802171e-06,
      "loss": 0.9047,
      "step": 982
    },
    {
      "epoch": 0.054103142715614505,
      "grad_norm": 0.8849464654922485,
      "learning_rate": 9.985357317976163e-06,
      "loss": 0.8892,
      "step": 983
    },
    {
      "epoch": 0.054158181517970166,
      "grad_norm": 1.0083115100860596,
      "learning_rate": 9.985324149681545e-06,
      "loss": 0.7713,
      "step": 984
    },
    {
      "epoch": 0.05421322032032583,
      "grad_norm": 0.8233863711357117,
      "learning_rate": 9.985290943918565e-06,
      "loss": 0.7967,
      "step": 985
    },
    {
      "epoch": 0.05426825912268149,
      "grad_norm": 0.9615303874015808,
      "learning_rate": 9.985257700687472e-06,
      "loss": 0.8576,
      "step": 986
    },
    {
      "epoch": 0.05432329792503715,
      "grad_norm": 0.8856416344642639,
      "learning_rate": 9.985224419988517e-06,
      "loss": 0.8614,
      "step": 987
    },
    {
      "epoch": 0.05437833672739281,
      "grad_norm": 0.968325674533844,
      "learning_rate": 9.98519110182195e-06,
      "loss": 0.8247,
      "step": 988
    },
    {
      "epoch": 0.05443337552974847,
      "grad_norm": 0.878402054309845,
      "learning_rate": 9.985157746188021e-06,
      "loss": 0.8661,
      "step": 989
    },
    {
      "epoch": 0.054488414332104135,
      "grad_norm": 0.8376438021659851,
      "learning_rate": 9.985124353086981e-06,
      "loss": 0.8554,
      "step": 990
    },
    {
      "epoch": 0.054543453134459796,
      "grad_norm": 1.0293036699295044,
      "learning_rate": 9.98509092251908e-06,
      "loss": 0.8049,
      "step": 991
    },
    {
      "epoch": 0.05459849193681546,
      "grad_norm": 1.2345234155654907,
      "learning_rate": 9.98505745448457e-06,
      "loss": 1.0358,
      "step": 992
    },
    {
      "epoch": 0.05465353073917112,
      "grad_norm": 0.9974482655525208,
      "learning_rate": 9.985023948983703e-06,
      "loss": 0.9329,
      "step": 993
    },
    {
      "epoch": 0.05470856954152677,
      "grad_norm": 1.383955478668213,
      "learning_rate": 9.984990406016732e-06,
      "loss": 0.8688,
      "step": 994
    },
    {
      "epoch": 0.054763608343882435,
      "grad_norm": 0.9369306564331055,
      "learning_rate": 9.984956825583906e-06,
      "loss": 0.8308,
      "step": 995
    },
    {
      "epoch": 0.054818647146238096,
      "grad_norm": 0.8676120042800903,
      "learning_rate": 9.984923207685478e-06,
      "loss": 0.8283,
      "step": 996
    },
    {
      "epoch": 0.05487368594859376,
      "grad_norm": 0.9218453168869019,
      "learning_rate": 9.984889552321704e-06,
      "loss": 0.7247,
      "step": 997
    },
    {
      "epoch": 0.05492872475094942,
      "grad_norm": 0.8575478196144104,
      "learning_rate": 9.984855859492833e-06,
      "loss": 0.8462,
      "step": 998
    },
    {
      "epoch": 0.05498376355330508,
      "grad_norm": 1.0042616128921509,
      "learning_rate": 9.98482212919912e-06,
      "loss": 0.9383,
      "step": 999
    },
    {
      "epoch": 0.05503880235566074,
      "grad_norm": 0.8642181158065796,
      "learning_rate": 9.984788361440817e-06,
      "loss": 0.8805,
      "step": 1000
    },
    {
      "epoch": 0.0550938411580164,
      "grad_norm": 0.8413823843002319,
      "learning_rate": 9.984754556218178e-06,
      "loss": 0.8161,
      "step": 1001
    },
    {
      "epoch": 0.055148879960372064,
      "grad_norm": 0.9473856091499329,
      "learning_rate": 9.984720713531462e-06,
      "loss": 0.8425,
      "step": 1002
    },
    {
      "epoch": 0.055203918762727726,
      "grad_norm": 0.7854379415512085,
      "learning_rate": 9.984686833380917e-06,
      "loss": 0.7506,
      "step": 1003
    },
    {
      "epoch": 0.05525895756508339,
      "grad_norm": 0.9481745958328247,
      "learning_rate": 9.984652915766801e-06,
      "loss": 0.954,
      "step": 1004
    },
    {
      "epoch": 0.05531399636743904,
      "grad_norm": 0.767803966999054,
      "learning_rate": 9.984618960689366e-06,
      "loss": 0.8113,
      "step": 1005
    },
    {
      "epoch": 0.0553690351697947,
      "grad_norm": 0.8957781195640564,
      "learning_rate": 9.984584968148871e-06,
      "loss": 0.9042,
      "step": 1006
    },
    {
      "epoch": 0.055424073972150364,
      "grad_norm": 1.116646409034729,
      "learning_rate": 9.98455093814557e-06,
      "loss": 0.8648,
      "step": 1007
    },
    {
      "epoch": 0.055479112774506026,
      "grad_norm": 0.9567018151283264,
      "learning_rate": 9.98451687067972e-06,
      "loss": 0.9446,
      "step": 1008
    },
    {
      "epoch": 0.05553415157686169,
      "grad_norm": 0.8470665812492371,
      "learning_rate": 9.98448276575157e-06,
      "loss": 0.8186,
      "step": 1009
    },
    {
      "epoch": 0.05558919037921735,
      "grad_norm": 0.9595193862915039,
      "learning_rate": 9.984448623361387e-06,
      "loss": 0.8406,
      "step": 1010
    },
    {
      "epoch": 0.05564422918157301,
      "grad_norm": 1.0579735040664673,
      "learning_rate": 9.98441444350942e-06,
      "loss": 0.9676,
      "step": 1011
    },
    {
      "epoch": 0.05569926798392867,
      "grad_norm": 0.8693701028823853,
      "learning_rate": 9.98438022619593e-06,
      "loss": 0.9451,
      "step": 1012
    },
    {
      "epoch": 0.05575430678628433,
      "grad_norm": 0.9251859784126282,
      "learning_rate": 9.98434597142117e-06,
      "loss": 0.7858,
      "step": 1013
    },
    {
      "epoch": 0.055809345588639994,
      "grad_norm": 0.8584280014038086,
      "learning_rate": 9.984311679185402e-06,
      "loss": 0.8481,
      "step": 1014
    },
    {
      "epoch": 0.05586438439099565,
      "grad_norm": 0.8903968334197998,
      "learning_rate": 9.98427734948888e-06,
      "loss": 0.7832,
      "step": 1015
    },
    {
      "epoch": 0.05591942319335131,
      "grad_norm": 0.905581533908844,
      "learning_rate": 9.984242982331864e-06,
      "loss": 0.8088,
      "step": 1016
    },
    {
      "epoch": 0.05597446199570697,
      "grad_norm": 0.9866476655006409,
      "learning_rate": 9.984208577714612e-06,
      "loss": 0.8366,
      "step": 1017
    },
    {
      "epoch": 0.05602950079806263,
      "grad_norm": 0.8843809962272644,
      "learning_rate": 9.984174135637384e-06,
      "loss": 0.8961,
      "step": 1018
    },
    {
      "epoch": 0.056084539600418294,
      "grad_norm": 0.9071753621101379,
      "learning_rate": 9.984139656100435e-06,
      "loss": 0.8671,
      "step": 1019
    },
    {
      "epoch": 0.056139578402773956,
      "grad_norm": 0.9894018173217773,
      "learning_rate": 9.984105139104028e-06,
      "loss": 0.9099,
      "step": 1020
    },
    {
      "epoch": 0.05619461720512962,
      "grad_norm": 0.8432741165161133,
      "learning_rate": 9.98407058464842e-06,
      "loss": 0.7817,
      "step": 1021
    },
    {
      "epoch": 0.05624965600748528,
      "grad_norm": 0.9538390040397644,
      "learning_rate": 9.984035992733873e-06,
      "loss": 0.8689,
      "step": 1022
    },
    {
      "epoch": 0.05630469480984094,
      "grad_norm": 0.9263421297073364,
      "learning_rate": 9.984001363360645e-06,
      "loss": 0.9066,
      "step": 1023
    },
    {
      "epoch": 0.0563597336121966,
      "grad_norm": 0.8921047449111938,
      "learning_rate": 9.983966696528996e-06,
      "loss": 0.8304,
      "step": 1024
    },
    {
      "epoch": 0.05641477241455226,
      "grad_norm": 0.8379812240600586,
      "learning_rate": 9.983931992239188e-06,
      "loss": 0.866,
      "step": 1025
    },
    {
      "epoch": 0.05646981121690792,
      "grad_norm": 0.9444219470024109,
      "learning_rate": 9.983897250491481e-06,
      "loss": 0.9456,
      "step": 1026
    },
    {
      "epoch": 0.05652485001926358,
      "grad_norm": 1.0268759727478027,
      "learning_rate": 9.983862471286137e-06,
      "loss": 0.8277,
      "step": 1027
    },
    {
      "epoch": 0.05657988882161924,
      "grad_norm": 1.3949217796325684,
      "learning_rate": 9.983827654623418e-06,
      "loss": 0.9721,
      "step": 1028
    },
    {
      "epoch": 0.0566349276239749,
      "grad_norm": 0.8899377584457397,
      "learning_rate": 9.983792800503582e-06,
      "loss": 0.8794,
      "step": 1029
    },
    {
      "epoch": 0.05668996642633056,
      "grad_norm": 0.989072322845459,
      "learning_rate": 9.983757908926895e-06,
      "loss": 0.8852,
      "step": 1030
    },
    {
      "epoch": 0.056745005228686224,
      "grad_norm": 0.9797759056091309,
      "learning_rate": 9.983722979893615e-06,
      "loss": 1.0405,
      "step": 1031
    },
    {
      "epoch": 0.056800044031041885,
      "grad_norm": 0.9044767618179321,
      "learning_rate": 9.98368801340401e-06,
      "loss": 0.7243,
      "step": 1032
    },
    {
      "epoch": 0.05685508283339755,
      "grad_norm": 1.116324782371521,
      "learning_rate": 9.983653009458338e-06,
      "loss": 0.9183,
      "step": 1033
    },
    {
      "epoch": 0.05691012163575321,
      "grad_norm": 0.9373337030410767,
      "learning_rate": 9.983617968056866e-06,
      "loss": 0.9417,
      "step": 1034
    },
    {
      "epoch": 0.05696516043810887,
      "grad_norm": 1.0587197542190552,
      "learning_rate": 9.983582889199855e-06,
      "loss": 0.896,
      "step": 1035
    },
    {
      "epoch": 0.05702019924046453,
      "grad_norm": 1.0080119371414185,
      "learning_rate": 9.983547772887568e-06,
      "loss": 0.924,
      "step": 1036
    },
    {
      "epoch": 0.057075238042820185,
      "grad_norm": 0.847091019153595,
      "learning_rate": 9.98351261912027e-06,
      "loss": 0.7443,
      "step": 1037
    },
    {
      "epoch": 0.05713027684517585,
      "grad_norm": 0.9876272082328796,
      "learning_rate": 9.983477427898225e-06,
      "loss": 0.9365,
      "step": 1038
    },
    {
      "epoch": 0.05718531564753151,
      "grad_norm": 0.9188169240951538,
      "learning_rate": 9.983442199221698e-06,
      "loss": 0.9213,
      "step": 1039
    },
    {
      "epoch": 0.05724035444988717,
      "grad_norm": 0.932399332523346,
      "learning_rate": 9.983406933090954e-06,
      "loss": 0.958,
      "step": 1040
    },
    {
      "epoch": 0.05729539325224283,
      "grad_norm": 0.9126465320587158,
      "learning_rate": 9.983371629506258e-06,
      "loss": 0.8913,
      "step": 1041
    },
    {
      "epoch": 0.05735043205459849,
      "grad_norm": 0.80904620885849,
      "learning_rate": 9.983336288467873e-06,
      "loss": 0.7719,
      "step": 1042
    },
    {
      "epoch": 0.057405470856954154,
      "grad_norm": 0.873833417892456,
      "learning_rate": 9.983300909976067e-06,
      "loss": 0.9201,
      "step": 1043
    },
    {
      "epoch": 0.057460509659309815,
      "grad_norm": 0.8331829309463501,
      "learning_rate": 9.983265494031107e-06,
      "loss": 0.8605,
      "step": 1044
    },
    {
      "epoch": 0.057515548461665476,
      "grad_norm": 0.8364768624305725,
      "learning_rate": 9.983230040633255e-06,
      "loss": 0.8627,
      "step": 1045
    },
    {
      "epoch": 0.05757058726402114,
      "grad_norm": 0.9226736426353455,
      "learning_rate": 9.98319454978278e-06,
      "loss": 0.9759,
      "step": 1046
    },
    {
      "epoch": 0.05762562606637679,
      "grad_norm": 0.8174427151679993,
      "learning_rate": 9.98315902147995e-06,
      "loss": 0.8066,
      "step": 1047
    },
    {
      "epoch": 0.057680664868732454,
      "grad_norm": 0.9154924750328064,
      "learning_rate": 9.98312345572503e-06,
      "loss": 0.9112,
      "step": 1048
    },
    {
      "epoch": 0.057735703671088115,
      "grad_norm": 0.8884655237197876,
      "learning_rate": 9.983087852518289e-06,
      "loss": 0.8699,
      "step": 1049
    },
    {
      "epoch": 0.057790742473443776,
      "grad_norm": 0.8849230408668518,
      "learning_rate": 9.983052211859992e-06,
      "loss": 0.8999,
      "step": 1050
    },
    {
      "epoch": 0.05784578127579944,
      "grad_norm": 1.025843858718872,
      "learning_rate": 9.98301653375041e-06,
      "loss": 0.7764,
      "step": 1051
    },
    {
      "epoch": 0.0579008200781551,
      "grad_norm": 0.900505006313324,
      "learning_rate": 9.98298081818981e-06,
      "loss": 0.9196,
      "step": 1052
    },
    {
      "epoch": 0.05795585888051076,
      "grad_norm": 0.9506704211235046,
      "learning_rate": 9.982945065178457e-06,
      "loss": 0.8319,
      "step": 1053
    },
    {
      "epoch": 0.05801089768286642,
      "grad_norm": 0.9439849853515625,
      "learning_rate": 9.982909274716626e-06,
      "loss": 0.8561,
      "step": 1054
    },
    {
      "epoch": 0.05806593648522208,
      "grad_norm": 0.8761240243911743,
      "learning_rate": 9.982873446804579e-06,
      "loss": 0.9681,
      "step": 1055
    },
    {
      "epoch": 0.058120975287577745,
      "grad_norm": 0.8756145238876343,
      "learning_rate": 9.982837581442592e-06,
      "loss": 0.8452,
      "step": 1056
    },
    {
      "epoch": 0.058176014089933406,
      "grad_norm": 0.8732383847236633,
      "learning_rate": 9.982801678630932e-06,
      "loss": 0.9018,
      "step": 1057
    },
    {
      "epoch": 0.05823105289228906,
      "grad_norm": 0.8338272571563721,
      "learning_rate": 9.982765738369867e-06,
      "loss": 0.9308,
      "step": 1058
    },
    {
      "epoch": 0.05828609169464472,
      "grad_norm": 0.843163013458252,
      "learning_rate": 9.982729760659669e-06,
      "loss": 0.7802,
      "step": 1059
    },
    {
      "epoch": 0.05834113049700038,
      "grad_norm": 1.2007580995559692,
      "learning_rate": 9.982693745500606e-06,
      "loss": 0.8406,
      "step": 1060
    },
    {
      "epoch": 0.058396169299356045,
      "grad_norm": 0.8760073781013489,
      "learning_rate": 9.982657692892954e-06,
      "loss": 0.8528,
      "step": 1061
    },
    {
      "epoch": 0.058451208101711706,
      "grad_norm": 0.925309419631958,
      "learning_rate": 9.982621602836978e-06,
      "loss": 0.9601,
      "step": 1062
    },
    {
      "epoch": 0.05850624690406737,
      "grad_norm": 0.9277135133743286,
      "learning_rate": 9.982585475332952e-06,
      "loss": 0.8405,
      "step": 1063
    },
    {
      "epoch": 0.05856128570642303,
      "grad_norm": 0.928044319152832,
      "learning_rate": 9.98254931038115e-06,
      "loss": 0.8259,
      "step": 1064
    },
    {
      "epoch": 0.05861632450877869,
      "grad_norm": 0.8363838195800781,
      "learning_rate": 9.982513107981837e-06,
      "loss": 0.8655,
      "step": 1065
    },
    {
      "epoch": 0.05867136331113435,
      "grad_norm": 0.9800984859466553,
      "learning_rate": 9.982476868135292e-06,
      "loss": 0.9285,
      "step": 1066
    },
    {
      "epoch": 0.05872640211349001,
      "grad_norm": 0.8062636256217957,
      "learning_rate": 9.982440590841785e-06,
      "loss": 0.754,
      "step": 1067
    },
    {
      "epoch": 0.058781440915845674,
      "grad_norm": 1.2010705471038818,
      "learning_rate": 9.982404276101586e-06,
      "loss": 0.9872,
      "step": 1068
    },
    {
      "epoch": 0.05883647971820133,
      "grad_norm": 1.0036406517028809,
      "learning_rate": 9.982367923914971e-06,
      "loss": 0.8724,
      "step": 1069
    },
    {
      "epoch": 0.05889151852055699,
      "grad_norm": 0.8768866658210754,
      "learning_rate": 9.982331534282212e-06,
      "loss": 0.838,
      "step": 1070
    },
    {
      "epoch": 0.05894655732291265,
      "grad_norm": 0.7892739176750183,
      "learning_rate": 9.982295107203584e-06,
      "loss": 0.6974,
      "step": 1071
    },
    {
      "epoch": 0.05900159612526831,
      "grad_norm": 0.863315999507904,
      "learning_rate": 9.982258642679358e-06,
      "loss": 0.9282,
      "step": 1072
    },
    {
      "epoch": 0.059056634927623974,
      "grad_norm": 0.8645132780075073,
      "learning_rate": 9.982222140709812e-06,
      "loss": 0.8504,
      "step": 1073
    },
    {
      "epoch": 0.059111673729979636,
      "grad_norm": 1.0003199577331543,
      "learning_rate": 9.982185601295216e-06,
      "loss": 1.0293,
      "step": 1074
    },
    {
      "epoch": 0.0591667125323353,
      "grad_norm": 0.8391831517219543,
      "learning_rate": 9.982149024435848e-06,
      "loss": 0.8609,
      "step": 1075
    },
    {
      "epoch": 0.05922175133469096,
      "grad_norm": 0.9940230846405029,
      "learning_rate": 9.982112410131981e-06,
      "loss": 0.9623,
      "step": 1076
    },
    {
      "epoch": 0.05927679013704662,
      "grad_norm": 1.0670262575149536,
      "learning_rate": 9.98207575838389e-06,
      "loss": 0.9952,
      "step": 1077
    },
    {
      "epoch": 0.05933182893940228,
      "grad_norm": 0.8506165742874146,
      "learning_rate": 9.982039069191853e-06,
      "loss": 0.8401,
      "step": 1078
    },
    {
      "epoch": 0.05938686774175794,
      "grad_norm": 0.8956409096717834,
      "learning_rate": 9.982002342556144e-06,
      "loss": 0.8779,
      "step": 1079
    },
    {
      "epoch": 0.0594419065441136,
      "grad_norm": 0.8955749273300171,
      "learning_rate": 9.981965578477038e-06,
      "loss": 0.8946,
      "step": 1080
    },
    {
      "epoch": 0.05949694534646926,
      "grad_norm": 0.9035234451293945,
      "learning_rate": 9.981928776954811e-06,
      "loss": 0.9352,
      "step": 1081
    },
    {
      "epoch": 0.05955198414882492,
      "grad_norm": 0.8748759627342224,
      "learning_rate": 9.981891937989743e-06,
      "loss": 0.8803,
      "step": 1082
    },
    {
      "epoch": 0.05960702295118058,
      "grad_norm": 0.9966281056404114,
      "learning_rate": 9.981855061582108e-06,
      "loss": 0.9304,
      "step": 1083
    },
    {
      "epoch": 0.05966206175353624,
      "grad_norm": 0.8696668148040771,
      "learning_rate": 9.981818147732183e-06,
      "loss": 0.8706,
      "step": 1084
    },
    {
      "epoch": 0.059717100555891904,
      "grad_norm": 0.9823188185691833,
      "learning_rate": 9.981781196440249e-06,
      "loss": 0.9431,
      "step": 1085
    },
    {
      "epoch": 0.059772139358247565,
      "grad_norm": 0.8401583433151245,
      "learning_rate": 9.981744207706577e-06,
      "loss": 0.8369,
      "step": 1086
    },
    {
      "epoch": 0.05982717816060323,
      "grad_norm": 0.8775757551193237,
      "learning_rate": 9.981707181531452e-06,
      "loss": 0.9516,
      "step": 1087
    },
    {
      "epoch": 0.05988221696295889,
      "grad_norm": 0.9153465628623962,
      "learning_rate": 9.981670117915148e-06,
      "loss": 0.8997,
      "step": 1088
    },
    {
      "epoch": 0.05993725576531455,
      "grad_norm": 0.9053078889846802,
      "learning_rate": 9.981633016857946e-06,
      "loss": 0.9452,
      "step": 1089
    },
    {
      "epoch": 0.059992294567670204,
      "grad_norm": 0.9154480695724487,
      "learning_rate": 9.981595878360123e-06,
      "loss": 0.8293,
      "step": 1090
    },
    {
      "epoch": 0.060047333370025865,
      "grad_norm": 0.85718834400177,
      "learning_rate": 9.981558702421958e-06,
      "loss": 0.876,
      "step": 1091
    },
    {
      "epoch": 0.06010237217238153,
      "grad_norm": 0.9437130689620972,
      "learning_rate": 9.981521489043734e-06,
      "loss": 0.9731,
      "step": 1092
    },
    {
      "epoch": 0.06015741097473719,
      "grad_norm": 0.9014891386032104,
      "learning_rate": 9.981484238225725e-06,
      "loss": 0.811,
      "step": 1093
    },
    {
      "epoch": 0.06021244977709285,
      "grad_norm": 0.8942846655845642,
      "learning_rate": 9.981446949968216e-06,
      "loss": 0.808,
      "step": 1094
    },
    {
      "epoch": 0.06026748857944851,
      "grad_norm": 0.855297863483429,
      "learning_rate": 9.981409624271483e-06,
      "loss": 0.8319,
      "step": 1095
    },
    {
      "epoch": 0.06032252738180417,
      "grad_norm": 0.9310913681983948,
      "learning_rate": 9.981372261135811e-06,
      "loss": 0.899,
      "step": 1096
    },
    {
      "epoch": 0.060377566184159834,
      "grad_norm": 0.8472979664802551,
      "learning_rate": 9.981334860561478e-06,
      "loss": 0.8818,
      "step": 1097
    },
    {
      "epoch": 0.060432604986515495,
      "grad_norm": 0.896617591381073,
      "learning_rate": 9.981297422548764e-06,
      "loss": 0.8991,
      "step": 1098
    },
    {
      "epoch": 0.06048764378887116,
      "grad_norm": 0.8543037176132202,
      "learning_rate": 9.981259947097954e-06,
      "loss": 0.8595,
      "step": 1099
    },
    {
      "epoch": 0.06054268259122682,
      "grad_norm": 0.8794904947280884,
      "learning_rate": 9.981222434209327e-06,
      "loss": 0.8561,
      "step": 1100
    },
    {
      "epoch": 0.06059772139358247,
      "grad_norm": 0.8882116675376892,
      "learning_rate": 9.981184883883165e-06,
      "loss": 0.8099,
      "step": 1101
    },
    {
      "epoch": 0.060652760195938134,
      "grad_norm": 1.0068262815475464,
      "learning_rate": 9.98114729611975e-06,
      "loss": 0.8104,
      "step": 1102
    },
    {
      "epoch": 0.060707798998293795,
      "grad_norm": 1.072316288948059,
      "learning_rate": 9.981109670919366e-06,
      "loss": 0.9877,
      "step": 1103
    },
    {
      "epoch": 0.06076283780064946,
      "grad_norm": 0.9959045052528381,
      "learning_rate": 9.981072008282298e-06,
      "loss": 0.906,
      "step": 1104
    },
    {
      "epoch": 0.06081787660300512,
      "grad_norm": 0.8712790608406067,
      "learning_rate": 9.981034308208823e-06,
      "loss": 0.8725,
      "step": 1105
    },
    {
      "epoch": 0.06087291540536078,
      "grad_norm": 0.9114679098129272,
      "learning_rate": 9.980996570699228e-06,
      "loss": 0.8385,
      "step": 1106
    },
    {
      "epoch": 0.06092795420771644,
      "grad_norm": 1.0024466514587402,
      "learning_rate": 9.980958795753796e-06,
      "loss": 0.8661,
      "step": 1107
    },
    {
      "epoch": 0.0609829930100721,
      "grad_norm": 0.9578461050987244,
      "learning_rate": 9.98092098337281e-06,
      "loss": 0.9358,
      "step": 1108
    },
    {
      "epoch": 0.061038031812427763,
      "grad_norm": 0.8677787780761719,
      "learning_rate": 9.980883133556557e-06,
      "loss": 0.8146,
      "step": 1109
    },
    {
      "epoch": 0.061093070614783425,
      "grad_norm": 0.9072276949882507,
      "learning_rate": 9.98084524630532e-06,
      "loss": 0.91,
      "step": 1110
    },
    {
      "epoch": 0.061148109417139086,
      "grad_norm": 0.8827292919158936,
      "learning_rate": 9.980807321619381e-06,
      "loss": 0.8854,
      "step": 1111
    },
    {
      "epoch": 0.06120314821949474,
      "grad_norm": 1.0012744665145874,
      "learning_rate": 9.98076935949903e-06,
      "loss": 0.8242,
      "step": 1112
    },
    {
      "epoch": 0.0612581870218504,
      "grad_norm": 0.9152620434761047,
      "learning_rate": 9.980731359944548e-06,
      "loss": 0.8832,
      "step": 1113
    },
    {
      "epoch": 0.061313225824206063,
      "grad_norm": 0.8986824750900269,
      "learning_rate": 9.980693322956222e-06,
      "loss": 0.7975,
      "step": 1114
    },
    {
      "epoch": 0.061368264626561725,
      "grad_norm": 0.9373019933700562,
      "learning_rate": 9.98065524853434e-06,
      "loss": 0.9541,
      "step": 1115
    },
    {
      "epoch": 0.061423303428917386,
      "grad_norm": 0.9875593781471252,
      "learning_rate": 9.980617136679185e-06,
      "loss": 1.0052,
      "step": 1116
    },
    {
      "epoch": 0.06147834223127305,
      "grad_norm": 1.0664819478988647,
      "learning_rate": 9.980578987391045e-06,
      "loss": 0.8584,
      "step": 1117
    },
    {
      "epoch": 0.06153338103362871,
      "grad_norm": 0.9149377942085266,
      "learning_rate": 9.980540800670207e-06,
      "loss": 0.8467,
      "step": 1118
    },
    {
      "epoch": 0.06158841983598437,
      "grad_norm": 0.9303194284439087,
      "learning_rate": 9.980502576516959e-06,
      "loss": 0.8219,
      "step": 1119
    },
    {
      "epoch": 0.06164345863834003,
      "grad_norm": 0.9059457778930664,
      "learning_rate": 9.980464314931583e-06,
      "loss": 0.8459,
      "step": 1120
    },
    {
      "epoch": 0.06169849744069569,
      "grad_norm": 0.9368849396705627,
      "learning_rate": 9.980426015914375e-06,
      "loss": 0.8933,
      "step": 1121
    },
    {
      "epoch": 0.061753536243051355,
      "grad_norm": 0.8188626766204834,
      "learning_rate": 9.980387679465615e-06,
      "loss": 0.807,
      "step": 1122
    },
    {
      "epoch": 0.06180857504540701,
      "grad_norm": 1.027171015739441,
      "learning_rate": 9.980349305585595e-06,
      "loss": 0.8919,
      "step": 1123
    },
    {
      "epoch": 0.06186361384776267,
      "grad_norm": 0.831649899482727,
      "learning_rate": 9.980310894274603e-06,
      "loss": 0.8109,
      "step": 1124
    },
    {
      "epoch": 0.06191865265011833,
      "grad_norm": 1.0170252323150635,
      "learning_rate": 9.980272445532928e-06,
      "loss": 0.9537,
      "step": 1125
    },
    {
      "epoch": 0.06197369145247399,
      "grad_norm": 0.97837233543396,
      "learning_rate": 9.980233959360858e-06,
      "loss": 0.9104,
      "step": 1126
    },
    {
      "epoch": 0.062028730254829655,
      "grad_norm": 0.9548324942588806,
      "learning_rate": 9.980195435758681e-06,
      "loss": 0.9473,
      "step": 1127
    },
    {
      "epoch": 0.062083769057185316,
      "grad_norm": 0.8675842881202698,
      "learning_rate": 9.980156874726692e-06,
      "loss": 0.8313,
      "step": 1128
    },
    {
      "epoch": 0.06213880785954098,
      "grad_norm": 0.8948968052864075,
      "learning_rate": 9.980118276265173e-06,
      "loss": 0.8008,
      "step": 1129
    },
    {
      "epoch": 0.06219384666189664,
      "grad_norm": 0.8914239406585693,
      "learning_rate": 9.98007964037442e-06,
      "loss": 0.7642,
      "step": 1130
    },
    {
      "epoch": 0.0622488854642523,
      "grad_norm": 0.9499951004981995,
      "learning_rate": 9.980040967054723e-06,
      "loss": 0.8669,
      "step": 1131
    },
    {
      "epoch": 0.06230392426660796,
      "grad_norm": 0.8959251642227173,
      "learning_rate": 9.980002256306369e-06,
      "loss": 0.9177,
      "step": 1132
    },
    {
      "epoch": 0.062358963068963616,
      "grad_norm": 0.8634380102157593,
      "learning_rate": 9.97996350812965e-06,
      "loss": 0.8252,
      "step": 1133
    },
    {
      "epoch": 0.06241400187131928,
      "grad_norm": 0.9380598068237305,
      "learning_rate": 9.97992472252486e-06,
      "loss": 0.9335,
      "step": 1134
    },
    {
      "epoch": 0.06246904067367494,
      "grad_norm": 0.8373183608055115,
      "learning_rate": 9.97988589949229e-06,
      "loss": 0.848,
      "step": 1135
    },
    {
      "epoch": 0.0625240794760306,
      "grad_norm": 0.9649023413658142,
      "learning_rate": 9.97984703903223e-06,
      "loss": 0.9648,
      "step": 1136
    },
    {
      "epoch": 0.06257911827838626,
      "grad_norm": 0.9972373843193054,
      "learning_rate": 9.979808141144972e-06,
      "loss": 0.9104,
      "step": 1137
    },
    {
      "epoch": 0.06263415708074192,
      "grad_norm": 0.8230985403060913,
      "learning_rate": 9.97976920583081e-06,
      "loss": 0.8393,
      "step": 1138
    },
    {
      "epoch": 0.06268919588309758,
      "grad_norm": 0.9775324463844299,
      "learning_rate": 9.979730233090034e-06,
      "loss": 0.8385,
      "step": 1139
    },
    {
      "epoch": 0.06274423468545325,
      "grad_norm": 0.8288110494613647,
      "learning_rate": 9.97969122292294e-06,
      "loss": 0.7308,
      "step": 1140
    },
    {
      "epoch": 0.06279927348780891,
      "grad_norm": 0.8980758786201477,
      "learning_rate": 9.979652175329819e-06,
      "loss": 0.863,
      "step": 1141
    },
    {
      "epoch": 0.06285431229016457,
      "grad_norm": 7.43889045715332,
      "learning_rate": 9.979613090310965e-06,
      "loss": 0.9412,
      "step": 1142
    },
    {
      "epoch": 0.06290935109252023,
      "grad_norm": 0.9758191704750061,
      "learning_rate": 9.97957396786667e-06,
      "loss": 0.8896,
      "step": 1143
    },
    {
      "epoch": 0.06296438989487589,
      "grad_norm": 0.8211693167686462,
      "learning_rate": 9.979534807997234e-06,
      "loss": 0.7352,
      "step": 1144
    },
    {
      "epoch": 0.06301942869723155,
      "grad_norm": 0.8643441796302795,
      "learning_rate": 9.979495610702945e-06,
      "loss": 0.8701,
      "step": 1145
    },
    {
      "epoch": 0.06307446749958721,
      "grad_norm": 1.0199437141418457,
      "learning_rate": 9.9794563759841e-06,
      "loss": 0.9025,
      "step": 1146
    },
    {
      "epoch": 0.06312950630194288,
      "grad_norm": 0.8367893695831299,
      "learning_rate": 9.979417103840994e-06,
      "loss": 0.8491,
      "step": 1147
    },
    {
      "epoch": 0.06318454510429854,
      "grad_norm": 0.9411819577217102,
      "learning_rate": 9.979377794273923e-06,
      "loss": 0.8501,
      "step": 1148
    },
    {
      "epoch": 0.06323958390665418,
      "grad_norm": 1.1497365236282349,
      "learning_rate": 9.97933844728318e-06,
      "loss": 1.0227,
      "step": 1149
    },
    {
      "epoch": 0.06329462270900985,
      "grad_norm": 0.9892984628677368,
      "learning_rate": 9.979299062869064e-06,
      "loss": 0.8942,
      "step": 1150
    },
    {
      "epoch": 0.06334966151136551,
      "grad_norm": 0.947952926158905,
      "learning_rate": 9.979259641031867e-06,
      "loss": 1.0149,
      "step": 1151
    },
    {
      "epoch": 0.06340470031372117,
      "grad_norm": 0.9060251712799072,
      "learning_rate": 9.979220181771889e-06,
      "loss": 0.8607,
      "step": 1152
    },
    {
      "epoch": 0.06345973911607683,
      "grad_norm": 0.8331984281539917,
      "learning_rate": 9.979180685089424e-06,
      "loss": 0.8777,
      "step": 1153
    },
    {
      "epoch": 0.06351477791843249,
      "grad_norm": 0.9133188724517822,
      "learning_rate": 9.97914115098477e-06,
      "loss": 0.7409,
      "step": 1154
    },
    {
      "epoch": 0.06356981672078815,
      "grad_norm": 0.9095513820648193,
      "learning_rate": 9.979101579458224e-06,
      "loss": 0.8938,
      "step": 1155
    },
    {
      "epoch": 0.06362485552314381,
      "grad_norm": 0.9584553241729736,
      "learning_rate": 9.979061970510082e-06,
      "loss": 0.8765,
      "step": 1156
    },
    {
      "epoch": 0.06367989432549948,
      "grad_norm": 0.8742124438285828,
      "learning_rate": 9.979022324140644e-06,
      "loss": 0.8564,
      "step": 1157
    },
    {
      "epoch": 0.06373493312785514,
      "grad_norm": 0.8776904344558716,
      "learning_rate": 9.978982640350208e-06,
      "loss": 0.8713,
      "step": 1158
    },
    {
      "epoch": 0.0637899719302108,
      "grad_norm": 0.8667464852333069,
      "learning_rate": 9.97894291913907e-06,
      "loss": 0.8705,
      "step": 1159
    },
    {
      "epoch": 0.06384501073256646,
      "grad_norm": 0.9028087854385376,
      "learning_rate": 9.978903160507531e-06,
      "loss": 0.8297,
      "step": 1160
    },
    {
      "epoch": 0.06390004953492212,
      "grad_norm": 0.900812029838562,
      "learning_rate": 9.978863364455887e-06,
      "loss": 0.8456,
      "step": 1161
    },
    {
      "epoch": 0.06395508833727778,
      "grad_norm": 0.9667207598686218,
      "learning_rate": 9.97882353098444e-06,
      "loss": 0.8081,
      "step": 1162
    },
    {
      "epoch": 0.06401012713963344,
      "grad_norm": 0.8959711194038391,
      "learning_rate": 9.978783660093488e-06,
      "loss": 0.8455,
      "step": 1163
    },
    {
      "epoch": 0.0640651659419891,
      "grad_norm": 0.8519117832183838,
      "learning_rate": 9.97874375178333e-06,
      "loss": 0.849,
      "step": 1164
    },
    {
      "epoch": 0.06412020474434477,
      "grad_norm": 1.0532654523849487,
      "learning_rate": 9.978703806054267e-06,
      "loss": 0.7356,
      "step": 1165
    },
    {
      "epoch": 0.06417524354670043,
      "grad_norm": 1.0313252210617065,
      "learning_rate": 9.9786638229066e-06,
      "loss": 1.024,
      "step": 1166
    },
    {
      "epoch": 0.06423028234905609,
      "grad_norm": 1.0567537546157837,
      "learning_rate": 9.978623802340627e-06,
      "loss": 0.9423,
      "step": 1167
    },
    {
      "epoch": 0.06428532115141175,
      "grad_norm": 0.8198097348213196,
      "learning_rate": 9.97858374435665e-06,
      "loss": 0.829,
      "step": 1168
    },
    {
      "epoch": 0.06434035995376741,
      "grad_norm": 0.8718193173408508,
      "learning_rate": 9.97854364895497e-06,
      "loss": 0.7184,
      "step": 1169
    },
    {
      "epoch": 0.06439539875612307,
      "grad_norm": 0.8037594556808472,
      "learning_rate": 9.978503516135892e-06,
      "loss": 0.7961,
      "step": 1170
    },
    {
      "epoch": 0.06445043755847872,
      "grad_norm": 0.9052229523658752,
      "learning_rate": 9.978463345899709e-06,
      "loss": 0.8016,
      "step": 1171
    },
    {
      "epoch": 0.06450547636083438,
      "grad_norm": 1.0194638967514038,
      "learning_rate": 9.978423138246731e-06,
      "loss": 0.9045,
      "step": 1172
    },
    {
      "epoch": 0.06456051516319004,
      "grad_norm": 0.953078031539917,
      "learning_rate": 9.978382893177259e-06,
      "loss": 0.9661,
      "step": 1173
    },
    {
      "epoch": 0.0646155539655457,
      "grad_norm": 0.8777341842651367,
      "learning_rate": 9.978342610691592e-06,
      "loss": 0.8685,
      "step": 1174
    },
    {
      "epoch": 0.06467059276790137,
      "grad_norm": 1.0178394317626953,
      "learning_rate": 9.978302290790034e-06,
      "loss": 0.9075,
      "step": 1175
    },
    {
      "epoch": 0.06472563157025703,
      "grad_norm": 0.935694694519043,
      "learning_rate": 9.978261933472889e-06,
      "loss": 0.8438,
      "step": 1176
    },
    {
      "epoch": 0.06478067037261269,
      "grad_norm": 1.0022411346435547,
      "learning_rate": 9.97822153874046e-06,
      "loss": 0.8701,
      "step": 1177
    },
    {
      "epoch": 0.06483570917496835,
      "grad_norm": 1.0371203422546387,
      "learning_rate": 9.97818110659305e-06,
      "loss": 0.9111,
      "step": 1178
    },
    {
      "epoch": 0.06489074797732401,
      "grad_norm": 0.7972478866577148,
      "learning_rate": 9.978140637030963e-06,
      "loss": 0.8602,
      "step": 1179
    },
    {
      "epoch": 0.06494578677967967,
      "grad_norm": 0.8556679487228394,
      "learning_rate": 9.978100130054505e-06,
      "loss": 0.9149,
      "step": 1180
    },
    {
      "epoch": 0.06500082558203533,
      "grad_norm": 0.92474365234375,
      "learning_rate": 9.978059585663979e-06,
      "loss": 0.8608,
      "step": 1181
    },
    {
      "epoch": 0.065055864384391,
      "grad_norm": 1.0170830488204956,
      "learning_rate": 9.978019003859687e-06,
      "loss": 0.9986,
      "step": 1182
    },
    {
      "epoch": 0.06511090318674666,
      "grad_norm": 0.9405049681663513,
      "learning_rate": 9.97797838464194e-06,
      "loss": 0.9023,
      "step": 1183
    },
    {
      "epoch": 0.06516594198910232,
      "grad_norm": 0.9351203441619873,
      "learning_rate": 9.977937728011038e-06,
      "loss": 0.8698,
      "step": 1184
    },
    {
      "epoch": 0.06522098079145798,
      "grad_norm": 0.8620241284370422,
      "learning_rate": 9.97789703396729e-06,
      "loss": 0.9393,
      "step": 1185
    },
    {
      "epoch": 0.06527601959381364,
      "grad_norm": 0.9440441131591797,
      "learning_rate": 9.977856302511e-06,
      "loss": 0.8249,
      "step": 1186
    },
    {
      "epoch": 0.0653310583961693,
      "grad_norm": 0.8311079144477844,
      "learning_rate": 9.977815533642474e-06,
      "loss": 0.8614,
      "step": 1187
    },
    {
      "epoch": 0.06538609719852496,
      "grad_norm": 0.8911672830581665,
      "learning_rate": 9.977774727362018e-06,
      "loss": 0.7909,
      "step": 1188
    },
    {
      "epoch": 0.06544113600088063,
      "grad_norm": 0.9237088561058044,
      "learning_rate": 9.97773388366994e-06,
      "loss": 0.7116,
      "step": 1189
    },
    {
      "epoch": 0.06549617480323629,
      "grad_norm": 1.1155747175216675,
      "learning_rate": 9.977693002566549e-06,
      "loss": 0.9248,
      "step": 1190
    },
    {
      "epoch": 0.06555121360559195,
      "grad_norm": 0.9386736750602722,
      "learning_rate": 9.977652084052148e-06,
      "loss": 0.8307,
      "step": 1191
    },
    {
      "epoch": 0.0656062524079476,
      "grad_norm": 1.1666040420532227,
      "learning_rate": 9.977611128127044e-06,
      "loss": 0.9723,
      "step": 1192
    },
    {
      "epoch": 0.06566129121030326,
      "grad_norm": 1.2366368770599365,
      "learning_rate": 9.977570134791552e-06,
      "loss": 0.8253,
      "step": 1193
    },
    {
      "epoch": 0.06571633001265892,
      "grad_norm": 0.823443591594696,
      "learning_rate": 9.977529104045971e-06,
      "loss": 0.7472,
      "step": 1194
    },
    {
      "epoch": 0.06577136881501458,
      "grad_norm": 0.9481683969497681,
      "learning_rate": 9.977488035890617e-06,
      "loss": 0.887,
      "step": 1195
    },
    {
      "epoch": 0.06582640761737024,
      "grad_norm": 0.9120422005653381,
      "learning_rate": 9.977446930325794e-06,
      "loss": 0.867,
      "step": 1196
    },
    {
      "epoch": 0.0658814464197259,
      "grad_norm": 0.8595587015151978,
      "learning_rate": 9.977405787351811e-06,
      "loss": 0.8532,
      "step": 1197
    },
    {
      "epoch": 0.06593648522208156,
      "grad_norm": 0.8590419888496399,
      "learning_rate": 9.97736460696898e-06,
      "loss": 0.8998,
      "step": 1198
    },
    {
      "epoch": 0.06599152402443723,
      "grad_norm": 0.9670939445495605,
      "learning_rate": 9.977323389177609e-06,
      "loss": 0.8964,
      "step": 1199
    },
    {
      "epoch": 0.06604656282679289,
      "grad_norm": 0.8870261907577515,
      "learning_rate": 9.977282133978006e-06,
      "loss": 0.9542,
      "step": 1200
    },
    {
      "epoch": 0.06610160162914855,
      "grad_norm": 0.942294180393219,
      "learning_rate": 9.977240841370484e-06,
      "loss": 0.8681,
      "step": 1201
    },
    {
      "epoch": 0.06615664043150421,
      "grad_norm": 0.9632517099380493,
      "learning_rate": 9.977199511355353e-06,
      "loss": 0.7327,
      "step": 1202
    },
    {
      "epoch": 0.06621167923385987,
      "grad_norm": 4.8085479736328125,
      "learning_rate": 9.97715814393292e-06,
      "loss": 0.8528,
      "step": 1203
    },
    {
      "epoch": 0.06626671803621553,
      "grad_norm": 0.9084093570709229,
      "learning_rate": 9.977116739103503e-06,
      "loss": 0.7836,
      "step": 1204
    },
    {
      "epoch": 0.0663217568385712,
      "grad_norm": 0.8961902260780334,
      "learning_rate": 9.977075296867406e-06,
      "loss": 0.854,
      "step": 1205
    },
    {
      "epoch": 0.06637679564092686,
      "grad_norm": 0.8727987408638,
      "learning_rate": 9.977033817224945e-06,
      "loss": 0.7931,
      "step": 1206
    },
    {
      "epoch": 0.06643183444328252,
      "grad_norm": 0.8263267874717712,
      "learning_rate": 9.976992300176428e-06,
      "loss": 0.852,
      "step": 1207
    },
    {
      "epoch": 0.06648687324563818,
      "grad_norm": 1.0499639511108398,
      "learning_rate": 9.97695074572217e-06,
      "loss": 1.0427,
      "step": 1208
    },
    {
      "epoch": 0.06654191204799384,
      "grad_norm": 0.9337313771247864,
      "learning_rate": 9.976909153862482e-06,
      "loss": 0.8035,
      "step": 1209
    },
    {
      "epoch": 0.0665969508503495,
      "grad_norm": 0.8795992732048035,
      "learning_rate": 9.976867524597678e-06,
      "loss": 0.9022,
      "step": 1210
    },
    {
      "epoch": 0.06665198965270516,
      "grad_norm": 0.9787294268608093,
      "learning_rate": 9.976825857928069e-06,
      "loss": 0.8259,
      "step": 1211
    },
    {
      "epoch": 0.06670702845506082,
      "grad_norm": 0.8570082187652588,
      "learning_rate": 9.976784153853969e-06,
      "loss": 0.8567,
      "step": 1212
    },
    {
      "epoch": 0.06676206725741649,
      "grad_norm": 1.0620380640029907,
      "learning_rate": 9.976742412375694e-06,
      "loss": 0.851,
      "step": 1213
    },
    {
      "epoch": 0.06681710605977213,
      "grad_norm": 0.8545439839363098,
      "learning_rate": 9.976700633493551e-06,
      "loss": 0.8827,
      "step": 1214
    },
    {
      "epoch": 0.0668721448621278,
      "grad_norm": 0.8543682098388672,
      "learning_rate": 9.97665881720786e-06,
      "loss": 0.8524,
      "step": 1215
    },
    {
      "epoch": 0.06692718366448346,
      "grad_norm": 0.7748527526855469,
      "learning_rate": 9.976616963518935e-06,
      "loss": 0.7459,
      "step": 1216
    },
    {
      "epoch": 0.06698222246683912,
      "grad_norm": 0.9876659512519836,
      "learning_rate": 9.976575072427087e-06,
      "loss": 0.8426,
      "step": 1217
    },
    {
      "epoch": 0.06703726126919478,
      "grad_norm": 0.8763901591300964,
      "learning_rate": 9.976533143932635e-06,
      "loss": 0.8561,
      "step": 1218
    },
    {
      "epoch": 0.06709230007155044,
      "grad_norm": 0.7816654443740845,
      "learning_rate": 9.97649117803589e-06,
      "loss": 0.8361,
      "step": 1219
    },
    {
      "epoch": 0.0671473388739061,
      "grad_norm": 0.8659802675247192,
      "learning_rate": 9.97644917473717e-06,
      "loss": 0.897,
      "step": 1220
    },
    {
      "epoch": 0.06720237767626176,
      "grad_norm": 0.9180877208709717,
      "learning_rate": 9.97640713403679e-06,
      "loss": 0.9516,
      "step": 1221
    },
    {
      "epoch": 0.06725741647861742,
      "grad_norm": 0.9624410271644592,
      "learning_rate": 9.976365055935067e-06,
      "loss": 0.9119,
      "step": 1222
    },
    {
      "epoch": 0.06731245528097309,
      "grad_norm": 0.8291105031967163,
      "learning_rate": 9.976322940432314e-06,
      "loss": 0.788,
      "step": 1223
    },
    {
      "epoch": 0.06736749408332875,
      "grad_norm": 0.9858983755111694,
      "learning_rate": 9.976280787528854e-06,
      "loss": 0.8794,
      "step": 1224
    },
    {
      "epoch": 0.06742253288568441,
      "grad_norm": 0.8283948302268982,
      "learning_rate": 9.976238597224996e-06,
      "loss": 0.8571,
      "step": 1225
    },
    {
      "epoch": 0.06747757168804007,
      "grad_norm": 0.8585363626480103,
      "learning_rate": 9.976196369521063e-06,
      "loss": 0.9005,
      "step": 1226
    },
    {
      "epoch": 0.06753261049039573,
      "grad_norm": 0.847882091999054,
      "learning_rate": 9.976154104417369e-06,
      "loss": 0.8058,
      "step": 1227
    },
    {
      "epoch": 0.06758764929275139,
      "grad_norm": 0.9045611023902893,
      "learning_rate": 9.976111801914232e-06,
      "loss": 0.7864,
      "step": 1228
    },
    {
      "epoch": 0.06764268809510705,
      "grad_norm": 0.805932879447937,
      "learning_rate": 9.976069462011972e-06,
      "loss": 0.8436,
      "step": 1229
    },
    {
      "epoch": 0.06769772689746271,
      "grad_norm": 0.8809003233909607,
      "learning_rate": 9.976027084710906e-06,
      "loss": 0.7876,
      "step": 1230
    },
    {
      "epoch": 0.06775276569981838,
      "grad_norm": 0.8681740760803223,
      "learning_rate": 9.975984670011352e-06,
      "loss": 0.877,
      "step": 1231
    },
    {
      "epoch": 0.06780780450217404,
      "grad_norm": 0.9909854531288147,
      "learning_rate": 9.975942217913627e-06,
      "loss": 0.8957,
      "step": 1232
    },
    {
      "epoch": 0.0678628433045297,
      "grad_norm": 0.9213934540748596,
      "learning_rate": 9.975899728418056e-06,
      "loss": 0.8344,
      "step": 1233
    },
    {
      "epoch": 0.06791788210688536,
      "grad_norm": 0.8289967179298401,
      "learning_rate": 9.975857201524952e-06,
      "loss": 0.876,
      "step": 1234
    },
    {
      "epoch": 0.06797292090924101,
      "grad_norm": 0.891812264919281,
      "learning_rate": 9.97581463723464e-06,
      "loss": 0.8611,
      "step": 1235
    },
    {
      "epoch": 0.06802795971159667,
      "grad_norm": 1.0301382541656494,
      "learning_rate": 9.975772035547435e-06,
      "loss": 0.8177,
      "step": 1236
    },
    {
      "epoch": 0.06808299851395233,
      "grad_norm": 0.8380662798881531,
      "learning_rate": 9.975729396463659e-06,
      "loss": 0.8631,
      "step": 1237
    },
    {
      "epoch": 0.06813803731630799,
      "grad_norm": 0.9226046204566956,
      "learning_rate": 9.975686719983633e-06,
      "loss": 0.8927,
      "step": 1238
    },
    {
      "epoch": 0.06819307611866365,
      "grad_norm": 0.8917136192321777,
      "learning_rate": 9.975644006107679e-06,
      "loss": 0.9048,
      "step": 1239
    },
    {
      "epoch": 0.06824811492101931,
      "grad_norm": 0.8559191226959229,
      "learning_rate": 9.975601254836114e-06,
      "loss": 0.8169,
      "step": 1240
    },
    {
      "epoch": 0.06830315372337498,
      "grad_norm": 0.9345341920852661,
      "learning_rate": 9.975558466169263e-06,
      "loss": 0.7929,
      "step": 1241
    },
    {
      "epoch": 0.06835819252573064,
      "grad_norm": 0.9155850410461426,
      "learning_rate": 9.975515640107447e-06,
      "loss": 0.8825,
      "step": 1242
    },
    {
      "epoch": 0.0684132313280863,
      "grad_norm": 0.899712860584259,
      "learning_rate": 9.975472776650987e-06,
      "loss": 0.825,
      "step": 1243
    },
    {
      "epoch": 0.06846827013044196,
      "grad_norm": 0.8280880451202393,
      "learning_rate": 9.975429875800206e-06,
      "loss": 0.8539,
      "step": 1244
    },
    {
      "epoch": 0.06852330893279762,
      "grad_norm": 0.9589636325836182,
      "learning_rate": 9.975386937555426e-06,
      "loss": 0.9465,
      "step": 1245
    },
    {
      "epoch": 0.06857834773515328,
      "grad_norm": 1.1027253866195679,
      "learning_rate": 9.97534396191697e-06,
      "loss": 0.87,
      "step": 1246
    },
    {
      "epoch": 0.06863338653750894,
      "grad_norm": 1.0510318279266357,
      "learning_rate": 9.975300948885158e-06,
      "loss": 0.8569,
      "step": 1247
    },
    {
      "epoch": 0.0686884253398646,
      "grad_norm": 0.8897958397865295,
      "learning_rate": 9.975257898460317e-06,
      "loss": 0.8431,
      "step": 1248
    },
    {
      "epoch": 0.06874346414222027,
      "grad_norm": 0.8827036619186401,
      "learning_rate": 9.975214810642771e-06,
      "loss": 0.922,
      "step": 1249
    },
    {
      "epoch": 0.06879850294457593,
      "grad_norm": 0.8798324465751648,
      "learning_rate": 9.97517168543284e-06,
      "loss": 0.7837,
      "step": 1250
    },
    {
      "epoch": 0.06885354174693159,
      "grad_norm": 0.9053803086280823,
      "learning_rate": 9.975128522830853e-06,
      "loss": 0.82,
      "step": 1251
    },
    {
      "epoch": 0.06890858054928725,
      "grad_norm": 0.8362607359886169,
      "learning_rate": 9.975085322837129e-06,
      "loss": 0.7684,
      "step": 1252
    },
    {
      "epoch": 0.06896361935164291,
      "grad_norm": 0.8898602724075317,
      "learning_rate": 9.975042085451997e-06,
      "loss": 0.8205,
      "step": 1253
    },
    {
      "epoch": 0.06901865815399857,
      "grad_norm": 0.9210274815559387,
      "learning_rate": 9.97499881067578e-06,
      "loss": 0.8364,
      "step": 1254
    },
    {
      "epoch": 0.06907369695635424,
      "grad_norm": 1.0881952047348022,
      "learning_rate": 9.974955498508804e-06,
      "loss": 0.8234,
      "step": 1255
    },
    {
      "epoch": 0.0691287357587099,
      "grad_norm": 0.8875024914741516,
      "learning_rate": 9.974912148951394e-06,
      "loss": 0.7974,
      "step": 1256
    },
    {
      "epoch": 0.06918377456106554,
      "grad_norm": 0.9065666794776917,
      "learning_rate": 9.974868762003876e-06,
      "loss": 0.7721,
      "step": 1257
    },
    {
      "epoch": 0.0692388133634212,
      "grad_norm": 0.8904553651809692,
      "learning_rate": 9.974825337666576e-06,
      "loss": 0.8551,
      "step": 1258
    },
    {
      "epoch": 0.06929385216577687,
      "grad_norm": 0.8586102724075317,
      "learning_rate": 9.974781875939821e-06,
      "loss": 0.8666,
      "step": 1259
    },
    {
      "epoch": 0.06934889096813253,
      "grad_norm": 0.9103402495384216,
      "learning_rate": 9.974738376823935e-06,
      "loss": 0.8361,
      "step": 1260
    },
    {
      "epoch": 0.06940392977048819,
      "grad_norm": 0.8657701015472412,
      "learning_rate": 9.974694840319249e-06,
      "loss": 0.8217,
      "step": 1261
    },
    {
      "epoch": 0.06945896857284385,
      "grad_norm": 0.865703821182251,
      "learning_rate": 9.974651266426088e-06,
      "loss": 0.8751,
      "step": 1262
    },
    {
      "epoch": 0.06951400737519951,
      "grad_norm": 0.8932577967643738,
      "learning_rate": 9.974607655144779e-06,
      "loss": 0.8709,
      "step": 1263
    },
    {
      "epoch": 0.06956904617755517,
      "grad_norm": 0.8417405486106873,
      "learning_rate": 9.97456400647565e-06,
      "loss": 0.8104,
      "step": 1264
    },
    {
      "epoch": 0.06962408497991084,
      "grad_norm": 0.8578035235404968,
      "learning_rate": 9.974520320419032e-06,
      "loss": 0.9173,
      "step": 1265
    },
    {
      "epoch": 0.0696791237822665,
      "grad_norm": 0.957539439201355,
      "learning_rate": 9.974476596975249e-06,
      "loss": 0.8955,
      "step": 1266
    },
    {
      "epoch": 0.06973416258462216,
      "grad_norm": 0.851222038269043,
      "learning_rate": 9.974432836144632e-06,
      "loss": 0.8696,
      "step": 1267
    },
    {
      "epoch": 0.06978920138697782,
      "grad_norm": 0.8178789615631104,
      "learning_rate": 9.974389037927508e-06,
      "loss": 0.7921,
      "step": 1268
    },
    {
      "epoch": 0.06984424018933348,
      "grad_norm": 0.954091489315033,
      "learning_rate": 9.97434520232421e-06,
      "loss": 0.9362,
      "step": 1269
    },
    {
      "epoch": 0.06989927899168914,
      "grad_norm": 0.8525053858757019,
      "learning_rate": 9.974301329335063e-06,
      "loss": 0.7996,
      "step": 1270
    },
    {
      "epoch": 0.0699543177940448,
      "grad_norm": 0.9340476393699646,
      "learning_rate": 9.9742574189604e-06,
      "loss": 0.9091,
      "step": 1271
    },
    {
      "epoch": 0.07000935659640047,
      "grad_norm": 0.7946187257766724,
      "learning_rate": 9.974213471200548e-06,
      "loss": 0.874,
      "step": 1272
    },
    {
      "epoch": 0.07006439539875613,
      "grad_norm": 0.8048381209373474,
      "learning_rate": 9.97416948605584e-06,
      "loss": 0.8557,
      "step": 1273
    },
    {
      "epoch": 0.07011943420111179,
      "grad_norm": 0.9849064946174622,
      "learning_rate": 9.974125463526607e-06,
      "loss": 0.8154,
      "step": 1274
    },
    {
      "epoch": 0.07017447300346745,
      "grad_norm": 0.9030239582061768,
      "learning_rate": 9.974081403613178e-06,
      "loss": 0.9411,
      "step": 1275
    },
    {
      "epoch": 0.07022951180582311,
      "grad_norm": 0.8869300484657288,
      "learning_rate": 9.974037306315882e-06,
      "loss": 0.8978,
      "step": 1276
    },
    {
      "epoch": 0.07028455060817877,
      "grad_norm": 0.8558536767959595,
      "learning_rate": 9.973993171635057e-06,
      "loss": 0.8937,
      "step": 1277
    },
    {
      "epoch": 0.07033958941053442,
      "grad_norm": 0.9005453586578369,
      "learning_rate": 9.973948999571029e-06,
      "loss": 0.9336,
      "step": 1278
    },
    {
      "epoch": 0.07039462821289008,
      "grad_norm": 0.8489978909492493,
      "learning_rate": 9.973904790124131e-06,
      "loss": 0.8267,
      "step": 1279
    },
    {
      "epoch": 0.07044966701524574,
      "grad_norm": 0.8295948505401611,
      "learning_rate": 9.973860543294696e-06,
      "loss": 0.8478,
      "step": 1280
    },
    {
      "epoch": 0.0705047058176014,
      "grad_norm": 0.8111379742622375,
      "learning_rate": 9.973816259083058e-06,
      "loss": 0.8333,
      "step": 1281
    },
    {
      "epoch": 0.07055974461995707,
      "grad_norm": 0.9380189776420593,
      "learning_rate": 9.973771937489547e-06,
      "loss": 0.9718,
      "step": 1282
    },
    {
      "epoch": 0.07061478342231273,
      "grad_norm": 1.251194953918457,
      "learning_rate": 9.973727578514499e-06,
      "loss": 0.9531,
      "step": 1283
    },
    {
      "epoch": 0.07066982222466839,
      "grad_norm": 0.9897224307060242,
      "learning_rate": 9.973683182158243e-06,
      "loss": 0.7853,
      "step": 1284
    },
    {
      "epoch": 0.07072486102702405,
      "grad_norm": 0.8409335017204285,
      "learning_rate": 9.973638748421119e-06,
      "loss": 0.7692,
      "step": 1285
    },
    {
      "epoch": 0.07077989982937971,
      "grad_norm": 0.9019681215286255,
      "learning_rate": 9.973594277303456e-06,
      "loss": 0.8135,
      "step": 1286
    },
    {
      "epoch": 0.07083493863173537,
      "grad_norm": 0.9236096739768982,
      "learning_rate": 9.973549768805588e-06,
      "loss": 0.9304,
      "step": 1287
    },
    {
      "epoch": 0.07088997743409103,
      "grad_norm": 0.9244743585586548,
      "learning_rate": 9.973505222927854e-06,
      "loss": 0.9056,
      "step": 1288
    },
    {
      "epoch": 0.0709450162364467,
      "grad_norm": 1.3418753147125244,
      "learning_rate": 9.973460639670585e-06,
      "loss": 0.8419,
      "step": 1289
    },
    {
      "epoch": 0.07100005503880236,
      "grad_norm": 0.8715767860412598,
      "learning_rate": 9.973416019034117e-06,
      "loss": 0.9704,
      "step": 1290
    },
    {
      "epoch": 0.07105509384115802,
      "grad_norm": 0.9609012007713318,
      "learning_rate": 9.973371361018787e-06,
      "loss": 0.8807,
      "step": 1291
    },
    {
      "epoch": 0.07111013264351368,
      "grad_norm": 0.8085873126983643,
      "learning_rate": 9.973326665624927e-06,
      "loss": 0.7947,
      "step": 1292
    },
    {
      "epoch": 0.07116517144586934,
      "grad_norm": 0.919280469417572,
      "learning_rate": 9.973281932852877e-06,
      "loss": 0.9743,
      "step": 1293
    },
    {
      "epoch": 0.071220210248225,
      "grad_norm": 1.0651074647903442,
      "learning_rate": 9.973237162702968e-06,
      "loss": 0.7164,
      "step": 1294
    },
    {
      "epoch": 0.07127524905058066,
      "grad_norm": 0.987251341342926,
      "learning_rate": 9.973192355175542e-06,
      "loss": 0.9286,
      "step": 1295
    },
    {
      "epoch": 0.07133028785293632,
      "grad_norm": 1.5507274866104126,
      "learning_rate": 9.973147510270935e-06,
      "loss": 0.9733,
      "step": 1296
    },
    {
      "epoch": 0.07138532665529199,
      "grad_norm": 0.8439416885375977,
      "learning_rate": 9.97310262798948e-06,
      "loss": 0.7462,
      "step": 1297
    },
    {
      "epoch": 0.07144036545764765,
      "grad_norm": 0.9604889750480652,
      "learning_rate": 9.973057708331519e-06,
      "loss": 1.0006,
      "step": 1298
    },
    {
      "epoch": 0.07149540426000331,
      "grad_norm": 0.8568960428237915,
      "learning_rate": 9.973012751297386e-06,
      "loss": 0.878,
      "step": 1299
    },
    {
      "epoch": 0.07155044306235896,
      "grad_norm": 0.8169522285461426,
      "learning_rate": 9.972967756887419e-06,
      "loss": 0.8241,
      "step": 1300
    },
    {
      "epoch": 0.07160548186471462,
      "grad_norm": 0.875738799571991,
      "learning_rate": 9.97292272510196e-06,
      "loss": 0.854,
      "step": 1301
    },
    {
      "epoch": 0.07166052066707028,
      "grad_norm": 0.7877739071846008,
      "learning_rate": 9.972877655941345e-06,
      "loss": 0.779,
      "step": 1302
    },
    {
      "epoch": 0.07171555946942594,
      "grad_norm": 0.8148574829101562,
      "learning_rate": 9.972832549405912e-06,
      "loss": 0.6965,
      "step": 1303
    },
    {
      "epoch": 0.0717705982717816,
      "grad_norm": 0.936720609664917,
      "learning_rate": 9.972787405495998e-06,
      "loss": 0.798,
      "step": 1304
    },
    {
      "epoch": 0.07182563707413726,
      "grad_norm": 0.8932886123657227,
      "learning_rate": 9.972742224211949e-06,
      "loss": 0.9196,
      "step": 1305
    },
    {
      "epoch": 0.07188067587649292,
      "grad_norm": 0.899246871471405,
      "learning_rate": 9.972697005554099e-06,
      "loss": 0.8081,
      "step": 1306
    },
    {
      "epoch": 0.07193571467884859,
      "grad_norm": 0.8789899349212646,
      "learning_rate": 9.972651749522788e-06,
      "loss": 0.89,
      "step": 1307
    },
    {
      "epoch": 0.07199075348120425,
      "grad_norm": 1.2412173748016357,
      "learning_rate": 9.97260645611836e-06,
      "loss": 0.9866,
      "step": 1308
    },
    {
      "epoch": 0.07204579228355991,
      "grad_norm": 0.8655833005905151,
      "learning_rate": 9.972561125341152e-06,
      "loss": 0.8144,
      "step": 1309
    },
    {
      "epoch": 0.07210083108591557,
      "grad_norm": 0.8705299496650696,
      "learning_rate": 9.972515757191506e-06,
      "loss": 0.8431,
      "step": 1310
    },
    {
      "epoch": 0.07215586988827123,
      "grad_norm": 0.8813188672065735,
      "learning_rate": 9.972470351669761e-06,
      "loss": 0.859,
      "step": 1311
    },
    {
      "epoch": 0.0722109086906269,
      "grad_norm": 2.043627977371216,
      "learning_rate": 9.972424908776262e-06,
      "loss": 0.9886,
      "step": 1312
    },
    {
      "epoch": 0.07226594749298255,
      "grad_norm": 0.9167500734329224,
      "learning_rate": 9.972379428511348e-06,
      "loss": 0.7203,
      "step": 1313
    },
    {
      "epoch": 0.07232098629533822,
      "grad_norm": 1.3145136833190918,
      "learning_rate": 9.972333910875358e-06,
      "loss": 0.9325,
      "step": 1314
    },
    {
      "epoch": 0.07237602509769388,
      "grad_norm": 0.834710419178009,
      "learning_rate": 9.972288355868641e-06,
      "loss": 0.9361,
      "step": 1315
    },
    {
      "epoch": 0.07243106390004954,
      "grad_norm": 0.9039230942726135,
      "learning_rate": 9.972242763491535e-06,
      "loss": 0.8027,
      "step": 1316
    },
    {
      "epoch": 0.0724861027024052,
      "grad_norm": 0.8911495208740234,
      "learning_rate": 9.972197133744384e-06,
      "loss": 0.951,
      "step": 1317
    },
    {
      "epoch": 0.07254114150476086,
      "grad_norm": 1.0752439498901367,
      "learning_rate": 9.972151466627529e-06,
      "loss": 0.8421,
      "step": 1318
    },
    {
      "epoch": 0.07259618030711652,
      "grad_norm": 0.926135778427124,
      "learning_rate": 9.972105762141314e-06,
      "loss": 0.8901,
      "step": 1319
    },
    {
      "epoch": 0.07265121910947218,
      "grad_norm": 0.8166295289993286,
      "learning_rate": 9.972060020286085e-06,
      "loss": 0.7845,
      "step": 1320
    },
    {
      "epoch": 0.07270625791182783,
      "grad_norm": 1.0000934600830078,
      "learning_rate": 9.972014241062182e-06,
      "loss": 0.8383,
      "step": 1321
    },
    {
      "epoch": 0.0727612967141835,
      "grad_norm": 1.2617899179458618,
      "learning_rate": 9.971968424469951e-06,
      "loss": 0.9826,
      "step": 1322
    },
    {
      "epoch": 0.07281633551653915,
      "grad_norm": 0.8451040983200073,
      "learning_rate": 9.971922570509738e-06,
      "loss": 0.8262,
      "step": 1323
    },
    {
      "epoch": 0.07287137431889482,
      "grad_norm": 0.8101939558982849,
      "learning_rate": 9.971876679181884e-06,
      "loss": 0.6904,
      "step": 1324
    },
    {
      "epoch": 0.07292641312125048,
      "grad_norm": 0.8805514574050903,
      "learning_rate": 9.971830750486736e-06,
      "loss": 0.8491,
      "step": 1325
    },
    {
      "epoch": 0.07298145192360614,
      "grad_norm": 0.8236901164054871,
      "learning_rate": 9.97178478442464e-06,
      "loss": 0.8462,
      "step": 1326
    },
    {
      "epoch": 0.0730364907259618,
      "grad_norm": 0.9183042645454407,
      "learning_rate": 9.971738780995938e-06,
      "loss": 0.7577,
      "step": 1327
    },
    {
      "epoch": 0.07309152952831746,
      "grad_norm": 0.8425934314727783,
      "learning_rate": 9.971692740200982e-06,
      "loss": 0.8462,
      "step": 1328
    },
    {
      "epoch": 0.07314656833067312,
      "grad_norm": 0.9114993214607239,
      "learning_rate": 9.971646662040112e-06,
      "loss": 0.9132,
      "step": 1329
    },
    {
      "epoch": 0.07320160713302878,
      "grad_norm": 0.8516649603843689,
      "learning_rate": 9.971600546513675e-06,
      "loss": 0.8819,
      "step": 1330
    },
    {
      "epoch": 0.07325664593538445,
      "grad_norm": 1.0859558582305908,
      "learning_rate": 9.971554393622023e-06,
      "loss": 0.9929,
      "step": 1331
    },
    {
      "epoch": 0.07331168473774011,
      "grad_norm": 0.8906900882720947,
      "learning_rate": 9.971508203365497e-06,
      "loss": 0.9166,
      "step": 1332
    },
    {
      "epoch": 0.07336672354009577,
      "grad_norm": 0.8931803703308105,
      "learning_rate": 9.971461975744445e-06,
      "loss": 0.864,
      "step": 1333
    },
    {
      "epoch": 0.07342176234245143,
      "grad_norm": 0.8404982686042786,
      "learning_rate": 9.971415710759216e-06,
      "loss": 0.8609,
      "step": 1334
    },
    {
      "epoch": 0.07347680114480709,
      "grad_norm": 0.8016490340232849,
      "learning_rate": 9.971369408410157e-06,
      "loss": 0.7694,
      "step": 1335
    },
    {
      "epoch": 0.07353183994716275,
      "grad_norm": 0.7700600028038025,
      "learning_rate": 9.971323068697618e-06,
      "loss": 0.7875,
      "step": 1336
    },
    {
      "epoch": 0.07358687874951841,
      "grad_norm": 0.8679799437522888,
      "learning_rate": 9.971276691621946e-06,
      "loss": 0.8409,
      "step": 1337
    },
    {
      "epoch": 0.07364191755187408,
      "grad_norm": 0.8329173922538757,
      "learning_rate": 9.971230277183486e-06,
      "loss": 0.8707,
      "step": 1338
    },
    {
      "epoch": 0.07369695635422974,
      "grad_norm": 0.8790140151977539,
      "learning_rate": 9.97118382538259e-06,
      "loss": 0.7631,
      "step": 1339
    },
    {
      "epoch": 0.0737519951565854,
      "grad_norm": 1.1895341873168945,
      "learning_rate": 9.97113733621961e-06,
      "loss": 0.8555,
      "step": 1340
    },
    {
      "epoch": 0.07380703395894106,
      "grad_norm": 0.8531593680381775,
      "learning_rate": 9.97109080969489e-06,
      "loss": 0.7192,
      "step": 1341
    },
    {
      "epoch": 0.07386207276129672,
      "grad_norm": 1.0388946533203125,
      "learning_rate": 9.971044245808784e-06,
      "loss": 0.8182,
      "step": 1342
    },
    {
      "epoch": 0.07391711156365237,
      "grad_norm": 0.8858556747436523,
      "learning_rate": 9.970997644561639e-06,
      "loss": 0.7981,
      "step": 1343
    },
    {
      "epoch": 0.07397215036600803,
      "grad_norm": 0.8710204362869263,
      "learning_rate": 9.970951005953807e-06,
      "loss": 0.7667,
      "step": 1344
    },
    {
      "epoch": 0.07402718916836369,
      "grad_norm": 0.9788708090782166,
      "learning_rate": 9.970904329985638e-06,
      "loss": 0.9693,
      "step": 1345
    },
    {
      "epoch": 0.07408222797071935,
      "grad_norm": 0.7805914878845215,
      "learning_rate": 9.970857616657482e-06,
      "loss": 0.6683,
      "step": 1346
    },
    {
      "epoch": 0.07413726677307501,
      "grad_norm": 0.9977933168411255,
      "learning_rate": 9.97081086596969e-06,
      "loss": 0.8288,
      "step": 1347
    },
    {
      "epoch": 0.07419230557543068,
      "grad_norm": 0.829115629196167,
      "learning_rate": 9.970764077922617e-06,
      "loss": 0.8361,
      "step": 1348
    },
    {
      "epoch": 0.07424734437778634,
      "grad_norm": 1.226120114326477,
      "learning_rate": 9.97071725251661e-06,
      "loss": 1.0008,
      "step": 1349
    },
    {
      "epoch": 0.074302383180142,
      "grad_norm": 0.8997750878334045,
      "learning_rate": 9.970670389752021e-06,
      "loss": 0.8048,
      "step": 1350
    },
    {
      "epoch": 0.07435742198249766,
      "grad_norm": 1.0885238647460938,
      "learning_rate": 9.970623489629205e-06,
      "loss": 0.9202,
      "step": 1351
    },
    {
      "epoch": 0.07441246078485332,
      "grad_norm": 0.8736100792884827,
      "learning_rate": 9.970576552148515e-06,
      "loss": 0.8515,
      "step": 1352
    },
    {
      "epoch": 0.07446749958720898,
      "grad_norm": 0.9211294651031494,
      "learning_rate": 9.970529577310301e-06,
      "loss": 0.9389,
      "step": 1353
    },
    {
      "epoch": 0.07452253838956464,
      "grad_norm": 0.9334765672683716,
      "learning_rate": 9.970482565114917e-06,
      "loss": 0.8165,
      "step": 1354
    },
    {
      "epoch": 0.0745775771919203,
      "grad_norm": 0.8307162523269653,
      "learning_rate": 9.970435515562717e-06,
      "loss": 0.7829,
      "step": 1355
    },
    {
      "epoch": 0.07463261599427597,
      "grad_norm": 0.987634003162384,
      "learning_rate": 9.970388428654055e-06,
      "loss": 0.848,
      "step": 1356
    },
    {
      "epoch": 0.07468765479663163,
      "grad_norm": 1.094752311706543,
      "learning_rate": 9.970341304389281e-06,
      "loss": 1.003,
      "step": 1357
    },
    {
      "epoch": 0.07474269359898729,
      "grad_norm": 0.9865909814834595,
      "learning_rate": 9.970294142768755e-06,
      "loss": 0.9116,
      "step": 1358
    },
    {
      "epoch": 0.07479773240134295,
      "grad_norm": 0.8404149413108826,
      "learning_rate": 9.970246943792828e-06,
      "loss": 0.8699,
      "step": 1359
    },
    {
      "epoch": 0.07485277120369861,
      "grad_norm": 0.9602416753768921,
      "learning_rate": 9.970199707461855e-06,
      "loss": 0.8166,
      "step": 1360
    },
    {
      "epoch": 0.07490781000605427,
      "grad_norm": 0.9748693704605103,
      "learning_rate": 9.970152433776193e-06,
      "loss": 0.8767,
      "step": 1361
    },
    {
      "epoch": 0.07496284880840993,
      "grad_norm": 0.8721657991409302,
      "learning_rate": 9.970105122736194e-06,
      "loss": 0.8825,
      "step": 1362
    },
    {
      "epoch": 0.0750178876107656,
      "grad_norm": 0.8683610558509827,
      "learning_rate": 9.970057774342215e-06,
      "loss": 0.7873,
      "step": 1363
    },
    {
      "epoch": 0.07507292641312124,
      "grad_norm": 0.856396496295929,
      "learning_rate": 9.970010388594613e-06,
      "loss": 0.8505,
      "step": 1364
    },
    {
      "epoch": 0.0751279652154769,
      "grad_norm": 1.0709880590438843,
      "learning_rate": 9.969962965493744e-06,
      "loss": 0.9519,
      "step": 1365
    },
    {
      "epoch": 0.07518300401783257,
      "grad_norm": 0.8839450478553772,
      "learning_rate": 9.969915505039963e-06,
      "loss": 0.8041,
      "step": 1366
    },
    {
      "epoch": 0.07523804282018823,
      "grad_norm": 0.89545738697052,
      "learning_rate": 9.969868007233627e-06,
      "loss": 0.8713,
      "step": 1367
    },
    {
      "epoch": 0.07529308162254389,
      "grad_norm": 0.9870849251747131,
      "learning_rate": 9.969820472075094e-06,
      "loss": 0.8655,
      "step": 1368
    },
    {
      "epoch": 0.07534812042489955,
      "grad_norm": 1.3123797178268433,
      "learning_rate": 9.96977289956472e-06,
      "loss": 1.0425,
      "step": 1369
    },
    {
      "epoch": 0.07540315922725521,
      "grad_norm": 0.8538400530815125,
      "learning_rate": 9.969725289702865e-06,
      "loss": 0.7052,
      "step": 1370
    },
    {
      "epoch": 0.07545819802961087,
      "grad_norm": 0.933397114276886,
      "learning_rate": 9.969677642489884e-06,
      "loss": 0.9819,
      "step": 1371
    },
    {
      "epoch": 0.07551323683196653,
      "grad_norm": 0.8428112268447876,
      "learning_rate": 9.969629957926134e-06,
      "loss": 0.7313,
      "step": 1372
    },
    {
      "epoch": 0.0755682756343222,
      "grad_norm": 0.9023239612579346,
      "learning_rate": 9.96958223601198e-06,
      "loss": 0.8297,
      "step": 1373
    },
    {
      "epoch": 0.07562331443667786,
      "grad_norm": 0.8971324563026428,
      "learning_rate": 9.969534476747771e-06,
      "loss": 0.8832,
      "step": 1374
    },
    {
      "epoch": 0.07567835323903352,
      "grad_norm": 0.8709388375282288,
      "learning_rate": 9.969486680133874e-06,
      "loss": 0.743,
      "step": 1375
    },
    {
      "epoch": 0.07573339204138918,
      "grad_norm": 0.9094591736793518,
      "learning_rate": 9.969438846170644e-06,
      "loss": 0.8294,
      "step": 1376
    },
    {
      "epoch": 0.07578843084374484,
      "grad_norm": 1.0753988027572632,
      "learning_rate": 9.969390974858444e-06,
      "loss": 0.7479,
      "step": 1377
    },
    {
      "epoch": 0.0758434696461005,
      "grad_norm": 0.933775007724762,
      "learning_rate": 9.96934306619763e-06,
      "loss": 0.8235,
      "step": 1378
    },
    {
      "epoch": 0.07589850844845616,
      "grad_norm": 0.8419735431671143,
      "learning_rate": 9.969295120188565e-06,
      "loss": 0.8103,
      "step": 1379
    },
    {
      "epoch": 0.07595354725081183,
      "grad_norm": 0.8912790417671204,
      "learning_rate": 9.969247136831606e-06,
      "loss": 0.911,
      "step": 1380
    },
    {
      "epoch": 0.07600858605316749,
      "grad_norm": 0.8780983090400696,
      "learning_rate": 9.969199116127118e-06,
      "loss": 0.8619,
      "step": 1381
    },
    {
      "epoch": 0.07606362485552315,
      "grad_norm": 0.8503809571266174,
      "learning_rate": 9.969151058075459e-06,
      "loss": 0.8093,
      "step": 1382
    },
    {
      "epoch": 0.07611866365787881,
      "grad_norm": 0.8633087277412415,
      "learning_rate": 9.96910296267699e-06,
      "loss": 0.7524,
      "step": 1383
    },
    {
      "epoch": 0.07617370246023447,
      "grad_norm": 1.1203595399856567,
      "learning_rate": 9.969054829932074e-06,
      "loss": 0.945,
      "step": 1384
    },
    {
      "epoch": 0.07622874126259013,
      "grad_norm": 0.8766878843307495,
      "learning_rate": 9.969006659841072e-06,
      "loss": 0.7537,
      "step": 1385
    },
    {
      "epoch": 0.07628378006494578,
      "grad_norm": 0.9795958399772644,
      "learning_rate": 9.968958452404345e-06,
      "loss": 0.7963,
      "step": 1386
    },
    {
      "epoch": 0.07633881886730144,
      "grad_norm": 0.9117506146430969,
      "learning_rate": 9.968910207622257e-06,
      "loss": 0.9469,
      "step": 1387
    },
    {
      "epoch": 0.0763938576696571,
      "grad_norm": 0.9731466770172119,
      "learning_rate": 9.96886192549517e-06,
      "loss": 0.9536,
      "step": 1388
    },
    {
      "epoch": 0.07644889647201276,
      "grad_norm": 0.8923571109771729,
      "learning_rate": 9.968813606023446e-06,
      "loss": 0.8362,
      "step": 1389
    },
    {
      "epoch": 0.07650393527436843,
      "grad_norm": 0.8819600343704224,
      "learning_rate": 9.96876524920745e-06,
      "loss": 0.6938,
      "step": 1390
    },
    {
      "epoch": 0.07655897407672409,
      "grad_norm": 0.9629887342453003,
      "learning_rate": 9.968716855047545e-06,
      "loss": 0.9104,
      "step": 1391
    },
    {
      "epoch": 0.07661401287907975,
      "grad_norm": 0.992770254611969,
      "learning_rate": 9.968668423544093e-06,
      "loss": 0.944,
      "step": 1392
    },
    {
      "epoch": 0.07666905168143541,
      "grad_norm": 0.8578491806983948,
      "learning_rate": 9.96861995469746e-06,
      "loss": 0.898,
      "step": 1393
    },
    {
      "epoch": 0.07672409048379107,
      "grad_norm": 1.1169229745864868,
      "learning_rate": 9.968571448508008e-06,
      "loss": 0.8324,
      "step": 1394
    },
    {
      "epoch": 0.07677912928614673,
      "grad_norm": 0.9600160121917725,
      "learning_rate": 9.968522904976106e-06,
      "loss": 0.9519,
      "step": 1395
    },
    {
      "epoch": 0.0768341680885024,
      "grad_norm": 0.8271373510360718,
      "learning_rate": 9.968474324102112e-06,
      "loss": 0.8576,
      "step": 1396
    },
    {
      "epoch": 0.07688920689085806,
      "grad_norm": 0.9437325596809387,
      "learning_rate": 9.968425705886397e-06,
      "loss": 0.9201,
      "step": 1397
    },
    {
      "epoch": 0.07694424569321372,
      "grad_norm": 0.8679039478302002,
      "learning_rate": 9.968377050329325e-06,
      "loss": 0.8893,
      "step": 1398
    },
    {
      "epoch": 0.07699928449556938,
      "grad_norm": 1.0178717374801636,
      "learning_rate": 9.96832835743126e-06,
      "loss": 0.9718,
      "step": 1399
    },
    {
      "epoch": 0.07705432329792504,
      "grad_norm": 0.8354432582855225,
      "learning_rate": 9.96827962719257e-06,
      "loss": 0.83,
      "step": 1400
    },
    {
      "epoch": 0.0771093621002807,
      "grad_norm": 1.2244631052017212,
      "learning_rate": 9.968230859613619e-06,
      "loss": 0.907,
      "step": 1401
    },
    {
      "epoch": 0.07716440090263636,
      "grad_norm": 0.9099625945091248,
      "learning_rate": 9.968182054694775e-06,
      "loss": 0.809,
      "step": 1402
    },
    {
      "epoch": 0.07721943970499202,
      "grad_norm": 0.8591424226760864,
      "learning_rate": 9.968133212436404e-06,
      "loss": 0.8869,
      "step": 1403
    },
    {
      "epoch": 0.07727447850734769,
      "grad_norm": 1.068003535270691,
      "learning_rate": 9.968084332838876e-06,
      "loss": 0.8747,
      "step": 1404
    },
    {
      "epoch": 0.07732951730970335,
      "grad_norm": 0.8503691554069519,
      "learning_rate": 9.968035415902555e-06,
      "loss": 0.7478,
      "step": 1405
    },
    {
      "epoch": 0.07738455611205901,
      "grad_norm": 0.9209537506103516,
      "learning_rate": 9.967986461627808e-06,
      "loss": 0.9052,
      "step": 1406
    },
    {
      "epoch": 0.07743959491441466,
      "grad_norm": 0.8447962999343872,
      "learning_rate": 9.967937470015006e-06,
      "loss": 0.7897,
      "step": 1407
    },
    {
      "epoch": 0.07749463371677032,
      "grad_norm": 0.8731846809387207,
      "learning_rate": 9.967888441064515e-06,
      "loss": 0.837,
      "step": 1408
    },
    {
      "epoch": 0.07754967251912598,
      "grad_norm": 0.9810444712638855,
      "learning_rate": 9.967839374776705e-06,
      "loss": 0.8236,
      "step": 1409
    },
    {
      "epoch": 0.07760471132148164,
      "grad_norm": 0.8283190131187439,
      "learning_rate": 9.967790271151944e-06,
      "loss": 0.8443,
      "step": 1410
    },
    {
      "epoch": 0.0776597501238373,
      "grad_norm": 0.7999932765960693,
      "learning_rate": 9.9677411301906e-06,
      "loss": 0.7945,
      "step": 1411
    },
    {
      "epoch": 0.07771478892619296,
      "grad_norm": 0.9435983300209045,
      "learning_rate": 9.967691951893044e-06,
      "loss": 0.9745,
      "step": 1412
    },
    {
      "epoch": 0.07776982772854862,
      "grad_norm": 0.8885984420776367,
      "learning_rate": 9.967642736259646e-06,
      "loss": 0.9163,
      "step": 1413
    },
    {
      "epoch": 0.07782486653090429,
      "grad_norm": 0.993928074836731,
      "learning_rate": 9.967593483290776e-06,
      "loss": 0.7797,
      "step": 1414
    },
    {
      "epoch": 0.07787990533325995,
      "grad_norm": 1.058830976486206,
      "learning_rate": 9.9675441929868e-06,
      "loss": 0.8671,
      "step": 1415
    },
    {
      "epoch": 0.07793494413561561,
      "grad_norm": 1.0469766855239868,
      "learning_rate": 9.967494865348093e-06,
      "loss": 0.8671,
      "step": 1416
    },
    {
      "epoch": 0.07798998293797127,
      "grad_norm": 0.902729868888855,
      "learning_rate": 9.967445500375025e-06,
      "loss": 0.8748,
      "step": 1417
    },
    {
      "epoch": 0.07804502174032693,
      "grad_norm": 0.90755295753479,
      "learning_rate": 9.967396098067965e-06,
      "loss": 0.8279,
      "step": 1418
    },
    {
      "epoch": 0.07810006054268259,
      "grad_norm": 0.8822374939918518,
      "learning_rate": 9.967346658427287e-06,
      "loss": 0.9386,
      "step": 1419
    },
    {
      "epoch": 0.07815509934503825,
      "grad_norm": 0.9201469421386719,
      "learning_rate": 9.96729718145336e-06,
      "loss": 0.8684,
      "step": 1420
    },
    {
      "epoch": 0.07821013814739392,
      "grad_norm": 0.9451109766960144,
      "learning_rate": 9.967247667146558e-06,
      "loss": 0.7854,
      "step": 1421
    },
    {
      "epoch": 0.07826517694974958,
      "grad_norm": 0.9146197438240051,
      "learning_rate": 9.96719811550725e-06,
      "loss": 0.8496,
      "step": 1422
    },
    {
      "epoch": 0.07832021575210524,
      "grad_norm": 0.9771224856376648,
      "learning_rate": 9.967148526535813e-06,
      "loss": 0.9657,
      "step": 1423
    },
    {
      "epoch": 0.0783752545544609,
      "grad_norm": 0.8437683582305908,
      "learning_rate": 9.967098900232616e-06,
      "loss": 0.8336,
      "step": 1424
    },
    {
      "epoch": 0.07843029335681656,
      "grad_norm": 0.8232185244560242,
      "learning_rate": 9.967049236598034e-06,
      "loss": 0.8878,
      "step": 1425
    },
    {
      "epoch": 0.07848533215917222,
      "grad_norm": 1.0200369358062744,
      "learning_rate": 9.96699953563244e-06,
      "loss": 0.8135,
      "step": 1426
    },
    {
      "epoch": 0.07854037096152788,
      "grad_norm": 0.8779187202453613,
      "learning_rate": 9.966949797336208e-06,
      "loss": 0.9124,
      "step": 1427
    },
    {
      "epoch": 0.07859540976388354,
      "grad_norm": 0.9557466506958008,
      "learning_rate": 9.966900021709708e-06,
      "loss": 0.9118,
      "step": 1428
    },
    {
      "epoch": 0.07865044856623919,
      "grad_norm": 0.8431050777435303,
      "learning_rate": 9.966850208753317e-06,
      "loss": 0.8361,
      "step": 1429
    },
    {
      "epoch": 0.07870548736859485,
      "grad_norm": 0.9269648194313049,
      "learning_rate": 9.966800358467412e-06,
      "loss": 0.9194,
      "step": 1430
    },
    {
      "epoch": 0.07876052617095052,
      "grad_norm": 0.818681538105011,
      "learning_rate": 9.966750470852363e-06,
      "loss": 0.7483,
      "step": 1431
    },
    {
      "epoch": 0.07881556497330618,
      "grad_norm": 0.8788284659385681,
      "learning_rate": 9.966700545908547e-06,
      "loss": 0.858,
      "step": 1432
    },
    {
      "epoch": 0.07887060377566184,
      "grad_norm": 0.7734160423278809,
      "learning_rate": 9.966650583636342e-06,
      "loss": 0.694,
      "step": 1433
    },
    {
      "epoch": 0.0789256425780175,
      "grad_norm": 0.8846608996391296,
      "learning_rate": 9.966600584036117e-06,
      "loss": 0.8144,
      "step": 1434
    },
    {
      "epoch": 0.07898068138037316,
      "grad_norm": 0.9740058183670044,
      "learning_rate": 9.966550547108254e-06,
      "loss": 0.9314,
      "step": 1435
    },
    {
      "epoch": 0.07903572018272882,
      "grad_norm": 0.8731759786605835,
      "learning_rate": 9.966500472853124e-06,
      "loss": 0.8475,
      "step": 1436
    },
    {
      "epoch": 0.07909075898508448,
      "grad_norm": 0.8984843492507935,
      "learning_rate": 9.966450361271109e-06,
      "loss": 0.7803,
      "step": 1437
    },
    {
      "epoch": 0.07914579778744014,
      "grad_norm": 0.8897966742515564,
      "learning_rate": 9.96640021236258e-06,
      "loss": 0.8879,
      "step": 1438
    },
    {
      "epoch": 0.0792008365897958,
      "grad_norm": 0.80704265832901,
      "learning_rate": 9.966350026127917e-06,
      "loss": 0.7585,
      "step": 1439
    },
    {
      "epoch": 0.07925587539215147,
      "grad_norm": 1.0807467699050903,
      "learning_rate": 9.966299802567499e-06,
      "loss": 1.078,
      "step": 1440
    },
    {
      "epoch": 0.07931091419450713,
      "grad_norm": 0.7994028925895691,
      "learning_rate": 9.966249541681697e-06,
      "loss": 0.8074,
      "step": 1441
    },
    {
      "epoch": 0.07936595299686279,
      "grad_norm": 0.877592921257019,
      "learning_rate": 9.966199243470895e-06,
      "loss": 0.8084,
      "step": 1442
    },
    {
      "epoch": 0.07942099179921845,
      "grad_norm": 0.7704572081565857,
      "learning_rate": 9.966148907935469e-06,
      "loss": 0.7206,
      "step": 1443
    },
    {
      "epoch": 0.07947603060157411,
      "grad_norm": 0.8222140669822693,
      "learning_rate": 9.966098535075797e-06,
      "loss": 0.7768,
      "step": 1444
    },
    {
      "epoch": 0.07953106940392977,
      "grad_norm": 1.389320731163025,
      "learning_rate": 9.966048124892257e-06,
      "loss": 1.0356,
      "step": 1445
    },
    {
      "epoch": 0.07958610820628544,
      "grad_norm": 0.9082457423210144,
      "learning_rate": 9.965997677385229e-06,
      "loss": 0.7379,
      "step": 1446
    },
    {
      "epoch": 0.0796411470086411,
      "grad_norm": 0.8029153943061829,
      "learning_rate": 9.965947192555093e-06,
      "loss": 0.7826,
      "step": 1447
    },
    {
      "epoch": 0.07969618581099676,
      "grad_norm": 0.8752758502960205,
      "learning_rate": 9.965896670402227e-06,
      "loss": 0.8526,
      "step": 1448
    },
    {
      "epoch": 0.07975122461335242,
      "grad_norm": 1.0665404796600342,
      "learning_rate": 9.965846110927009e-06,
      "loss": 0.858,
      "step": 1449
    },
    {
      "epoch": 0.07980626341570807,
      "grad_norm": 0.9468502402305603,
      "learning_rate": 9.96579551412982e-06,
      "loss": 0.9658,
      "step": 1450
    },
    {
      "epoch": 0.07986130221806373,
      "grad_norm": 1.0239403247833252,
      "learning_rate": 9.965744880011046e-06,
      "loss": 0.7995,
      "step": 1451
    },
    {
      "epoch": 0.07991634102041939,
      "grad_norm": 0.9808099865913391,
      "learning_rate": 9.965694208571059e-06,
      "loss": 1.0173,
      "step": 1452
    },
    {
      "epoch": 0.07997137982277505,
      "grad_norm": 0.9338780641555786,
      "learning_rate": 9.965643499810245e-06,
      "loss": 0.7917,
      "step": 1453
    },
    {
      "epoch": 0.08002641862513071,
      "grad_norm": 0.9294295310974121,
      "learning_rate": 9.965592753728981e-06,
      "loss": 0.88,
      "step": 1454
    },
    {
      "epoch": 0.08008145742748637,
      "grad_norm": 1.0261508226394653,
      "learning_rate": 9.965541970327654e-06,
      "loss": 0.8825,
      "step": 1455
    },
    {
      "epoch": 0.08013649622984204,
      "grad_norm": 0.8964946269989014,
      "learning_rate": 9.965491149606642e-06,
      "loss": 0.81,
      "step": 1456
    },
    {
      "epoch": 0.0801915350321977,
      "grad_norm": 0.9468267560005188,
      "learning_rate": 9.965440291566329e-06,
      "loss": 0.9453,
      "step": 1457
    },
    {
      "epoch": 0.08024657383455336,
      "grad_norm": 0.8289040327072144,
      "learning_rate": 9.965389396207092e-06,
      "loss": 0.7373,
      "step": 1458
    },
    {
      "epoch": 0.08030161263690902,
      "grad_norm": 0.8782384991645813,
      "learning_rate": 9.965338463529322e-06,
      "loss": 0.9199,
      "step": 1459
    },
    {
      "epoch": 0.08035665143926468,
      "grad_norm": 0.8613787293434143,
      "learning_rate": 9.965287493533395e-06,
      "loss": 0.8719,
      "step": 1460
    },
    {
      "epoch": 0.08041169024162034,
      "grad_norm": 0.8474903106689453,
      "learning_rate": 9.965236486219696e-06,
      "loss": 0.8033,
      "step": 1461
    },
    {
      "epoch": 0.080466729043976,
      "grad_norm": 1.1442681550979614,
      "learning_rate": 9.965185441588609e-06,
      "loss": 0.8996,
      "step": 1462
    },
    {
      "epoch": 0.08052176784633167,
      "grad_norm": 1.564138412475586,
      "learning_rate": 9.965134359640518e-06,
      "loss": 0.7451,
      "step": 1463
    },
    {
      "epoch": 0.08057680664868733,
      "grad_norm": 0.9211083054542542,
      "learning_rate": 9.965083240375806e-06,
      "loss": 0.8939,
      "step": 1464
    },
    {
      "epoch": 0.08063184545104299,
      "grad_norm": 0.9503418207168579,
      "learning_rate": 9.965032083794856e-06,
      "loss": 0.8544,
      "step": 1465
    },
    {
      "epoch": 0.08068688425339865,
      "grad_norm": 0.9304021596908569,
      "learning_rate": 9.964980889898055e-06,
      "loss": 0.9192,
      "step": 1466
    },
    {
      "epoch": 0.08074192305575431,
      "grad_norm": 0.8430425524711609,
      "learning_rate": 9.964929658685787e-06,
      "loss": 0.8586,
      "step": 1467
    },
    {
      "epoch": 0.08079696185810997,
      "grad_norm": 0.8671759366989136,
      "learning_rate": 9.964878390158437e-06,
      "loss": 0.8807,
      "step": 1468
    },
    {
      "epoch": 0.08085200066046563,
      "grad_norm": 0.9548830986022949,
      "learning_rate": 9.964827084316389e-06,
      "loss": 0.9033,
      "step": 1469
    },
    {
      "epoch": 0.0809070394628213,
      "grad_norm": 0.8736767768859863,
      "learning_rate": 9.964775741160029e-06,
      "loss": 0.8509,
      "step": 1470
    },
    {
      "epoch": 0.08096207826517696,
      "grad_norm": 0.8827025890350342,
      "learning_rate": 9.964724360689745e-06,
      "loss": 0.897,
      "step": 1471
    },
    {
      "epoch": 0.0810171170675326,
      "grad_norm": 1.02822744846344,
      "learning_rate": 9.964672942905921e-06,
      "loss": 1.0371,
      "step": 1472
    },
    {
      "epoch": 0.08107215586988827,
      "grad_norm": 0.8619557619094849,
      "learning_rate": 9.964621487808946e-06,
      "loss": 0.7654,
      "step": 1473
    },
    {
      "epoch": 0.08112719467224393,
      "grad_norm": 0.7855951189994812,
      "learning_rate": 9.9645699953992e-06,
      "loss": 0.7767,
      "step": 1474
    },
    {
      "epoch": 0.08118223347459959,
      "grad_norm": 0.8139809370040894,
      "learning_rate": 9.96451846567708e-06,
      "loss": 0.7535,
      "step": 1475
    },
    {
      "epoch": 0.08123727227695525,
      "grad_norm": 0.8491657376289368,
      "learning_rate": 9.964466898642966e-06,
      "loss": 0.854,
      "step": 1476
    },
    {
      "epoch": 0.08129231107931091,
      "grad_norm": 0.8968605399131775,
      "learning_rate": 9.964415294297247e-06,
      "loss": 0.8914,
      "step": 1477
    },
    {
      "epoch": 0.08134734988166657,
      "grad_norm": 0.8692505359649658,
      "learning_rate": 9.964363652640313e-06,
      "loss": 0.9245,
      "step": 1478
    },
    {
      "epoch": 0.08140238868402223,
      "grad_norm": 0.8916530013084412,
      "learning_rate": 9.964311973672549e-06,
      "loss": 0.7662,
      "step": 1479
    },
    {
      "epoch": 0.0814574274863779,
      "grad_norm": 0.8239215612411499,
      "learning_rate": 9.964260257394347e-06,
      "loss": 0.9191,
      "step": 1480
    },
    {
      "epoch": 0.08151246628873356,
      "grad_norm": 0.8672100901603699,
      "learning_rate": 9.964208503806092e-06,
      "loss": 0.7656,
      "step": 1481
    },
    {
      "epoch": 0.08156750509108922,
      "grad_norm": 0.9195712208747864,
      "learning_rate": 9.964156712908177e-06,
      "loss": 0.8656,
      "step": 1482
    },
    {
      "epoch": 0.08162254389344488,
      "grad_norm": 0.8282535672187805,
      "learning_rate": 9.964104884700986e-06,
      "loss": 0.8264,
      "step": 1483
    },
    {
      "epoch": 0.08167758269580054,
      "grad_norm": 0.8492032289505005,
      "learning_rate": 9.964053019184913e-06,
      "loss": 0.7816,
      "step": 1484
    },
    {
      "epoch": 0.0817326214981562,
      "grad_norm": 0.8491117358207703,
      "learning_rate": 9.964001116360347e-06,
      "loss": 0.7885,
      "step": 1485
    },
    {
      "epoch": 0.08178766030051186,
      "grad_norm": 0.9415153861045837,
      "learning_rate": 9.963949176227677e-06,
      "loss": 0.8165,
      "step": 1486
    },
    {
      "epoch": 0.08184269910286752,
      "grad_norm": 0.8462526202201843,
      "learning_rate": 9.963897198787294e-06,
      "loss": 0.8498,
      "step": 1487
    },
    {
      "epoch": 0.08189773790522319,
      "grad_norm": 0.8591959476470947,
      "learning_rate": 9.963845184039586e-06,
      "loss": 0.8906,
      "step": 1488
    },
    {
      "epoch": 0.08195277670757885,
      "grad_norm": 0.840761661529541,
      "learning_rate": 9.963793131984949e-06,
      "loss": 0.7831,
      "step": 1489
    },
    {
      "epoch": 0.08200781550993451,
      "grad_norm": 0.931404173374176,
      "learning_rate": 9.96374104262377e-06,
      "loss": 0.889,
      "step": 1490
    },
    {
      "epoch": 0.08206285431229017,
      "grad_norm": 0.9048783779144287,
      "learning_rate": 9.963688915956443e-06,
      "loss": 0.8321,
      "step": 1491
    },
    {
      "epoch": 0.08211789311464583,
      "grad_norm": 0.9145931601524353,
      "learning_rate": 9.96363675198336e-06,
      "loss": 0.9918,
      "step": 1492
    },
    {
      "epoch": 0.08217293191700148,
      "grad_norm": 0.9256643652915955,
      "learning_rate": 9.963584550704908e-06,
      "loss": 0.8731,
      "step": 1493
    },
    {
      "epoch": 0.08222797071935714,
      "grad_norm": 1.0212007761001587,
      "learning_rate": 9.963532312121486e-06,
      "loss": 0.9077,
      "step": 1494
    },
    {
      "epoch": 0.0822830095217128,
      "grad_norm": 0.9206242561340332,
      "learning_rate": 9.963480036233483e-06,
      "loss": 0.9076,
      "step": 1495
    },
    {
      "epoch": 0.08233804832406846,
      "grad_norm": 0.8846865296363831,
      "learning_rate": 9.963427723041294e-06,
      "loss": 0.6826,
      "step": 1496
    },
    {
      "epoch": 0.08239308712642412,
      "grad_norm": 0.8745351433753967,
      "learning_rate": 9.963375372545309e-06,
      "loss": 0.7935,
      "step": 1497
    },
    {
      "epoch": 0.08244812592877979,
      "grad_norm": 0.9019666314125061,
      "learning_rate": 9.963322984745924e-06,
      "loss": 0.8435,
      "step": 1498
    },
    {
      "epoch": 0.08250316473113545,
      "grad_norm": 0.8586859703063965,
      "learning_rate": 9.963270559643531e-06,
      "loss": 0.8118,
      "step": 1499
    },
    {
      "epoch": 0.08255820353349111,
      "grad_norm": 0.9192817807197571,
      "learning_rate": 9.963218097238528e-06,
      "loss": 0.824,
      "step": 1500
    },
    {
      "epoch": 0.08261324233584677,
      "grad_norm": 0.8972243070602417,
      "learning_rate": 9.963165597531304e-06,
      "loss": 0.8404,
      "step": 1501
    },
    {
      "epoch": 0.08266828113820243,
      "grad_norm": 0.8953961133956909,
      "learning_rate": 9.963113060522256e-06,
      "loss": 0.9031,
      "step": 1502
    },
    {
      "epoch": 0.0827233199405581,
      "grad_norm": 0.9551270604133606,
      "learning_rate": 9.963060486211779e-06,
      "loss": 0.9177,
      "step": 1503
    },
    {
      "epoch": 0.08277835874291375,
      "grad_norm": 0.8524616956710815,
      "learning_rate": 9.963007874600268e-06,
      "loss": 0.8582,
      "step": 1504
    },
    {
      "epoch": 0.08283339754526942,
      "grad_norm": 0.8148764371871948,
      "learning_rate": 9.962955225688118e-06,
      "loss": 0.6859,
      "step": 1505
    },
    {
      "epoch": 0.08288843634762508,
      "grad_norm": 0.9110590219497681,
      "learning_rate": 9.962902539475728e-06,
      "loss": 0.7189,
      "step": 1506
    },
    {
      "epoch": 0.08294347514998074,
      "grad_norm": 0.8700116872787476,
      "learning_rate": 9.962849815963487e-06,
      "loss": 0.9462,
      "step": 1507
    },
    {
      "epoch": 0.0829985139523364,
      "grad_norm": 0.877109706401825,
      "learning_rate": 9.962797055151797e-06,
      "loss": 0.8138,
      "step": 1508
    },
    {
      "epoch": 0.08305355275469206,
      "grad_norm": 0.7818365097045898,
      "learning_rate": 9.962744257041053e-06,
      "loss": 0.8474,
      "step": 1509
    },
    {
      "epoch": 0.08310859155704772,
      "grad_norm": 0.88360196352005,
      "learning_rate": 9.96269142163165e-06,
      "loss": 0.8724,
      "step": 1510
    },
    {
      "epoch": 0.08316363035940338,
      "grad_norm": 0.8982682228088379,
      "learning_rate": 9.962638548923988e-06,
      "loss": 0.9687,
      "step": 1511
    },
    {
      "epoch": 0.08321866916175905,
      "grad_norm": 0.7362002730369568,
      "learning_rate": 9.962585638918462e-06,
      "loss": 0.7666,
      "step": 1512
    },
    {
      "epoch": 0.08327370796411471,
      "grad_norm": 1.0993375778198242,
      "learning_rate": 9.962532691615472e-06,
      "loss": 0.8869,
      "step": 1513
    },
    {
      "epoch": 0.08332874676647037,
      "grad_norm": 0.8684842586517334,
      "learning_rate": 9.962479707015415e-06,
      "loss": 0.872,
      "step": 1514
    },
    {
      "epoch": 0.08338378556882602,
      "grad_norm": 1.0598478317260742,
      "learning_rate": 9.962426685118689e-06,
      "loss": 0.9102,
      "step": 1515
    },
    {
      "epoch": 0.08343882437118168,
      "grad_norm": 0.8492125272750854,
      "learning_rate": 9.96237362592569e-06,
      "loss": 0.7554,
      "step": 1516
    },
    {
      "epoch": 0.08349386317353734,
      "grad_norm": 0.8489052653312683,
      "learning_rate": 9.962320529436821e-06,
      "loss": 0.9139,
      "step": 1517
    },
    {
      "epoch": 0.083548901975893,
      "grad_norm": 0.8650774359703064,
      "learning_rate": 9.962267395652479e-06,
      "loss": 0.8717,
      "step": 1518
    },
    {
      "epoch": 0.08360394077824866,
      "grad_norm": 0.8393206596374512,
      "learning_rate": 9.962214224573064e-06,
      "loss": 0.8256,
      "step": 1519
    },
    {
      "epoch": 0.08365897958060432,
      "grad_norm": 0.8304896354675293,
      "learning_rate": 9.962161016198974e-06,
      "loss": 0.8232,
      "step": 1520
    },
    {
      "epoch": 0.08371401838295998,
      "grad_norm": 0.8718386292457581,
      "learning_rate": 9.962107770530612e-06,
      "loss": 0.8206,
      "step": 1521
    },
    {
      "epoch": 0.08376905718531565,
      "grad_norm": 0.9109341502189636,
      "learning_rate": 9.962054487568373e-06,
      "loss": 0.9576,
      "step": 1522
    },
    {
      "epoch": 0.08382409598767131,
      "grad_norm": 0.9543303847312927,
      "learning_rate": 9.962001167312663e-06,
      "loss": 0.8816,
      "step": 1523
    },
    {
      "epoch": 0.08387913479002697,
      "grad_norm": 0.9992844462394714,
      "learning_rate": 9.961947809763881e-06,
      "loss": 0.8682,
      "step": 1524
    },
    {
      "epoch": 0.08393417359238263,
      "grad_norm": 0.8092770576477051,
      "learning_rate": 9.961894414922425e-06,
      "loss": 0.6352,
      "step": 1525
    },
    {
      "epoch": 0.08398921239473829,
      "grad_norm": 0.9888653755187988,
      "learning_rate": 9.961840982788703e-06,
      "loss": 0.8721,
      "step": 1526
    },
    {
      "epoch": 0.08404425119709395,
      "grad_norm": 1.0092703104019165,
      "learning_rate": 9.961787513363108e-06,
      "loss": 0.7776,
      "step": 1527
    },
    {
      "epoch": 0.08409928999944961,
      "grad_norm": 0.8654646277427673,
      "learning_rate": 9.961734006646049e-06,
      "loss": 0.8835,
      "step": 1528
    },
    {
      "epoch": 0.08415432880180528,
      "grad_norm": 0.7630153298377991,
      "learning_rate": 9.961680462637924e-06,
      "loss": 0.7501,
      "step": 1529
    },
    {
      "epoch": 0.08420936760416094,
      "grad_norm": 1.1883158683776855,
      "learning_rate": 9.961626881339138e-06,
      "loss": 0.9476,
      "step": 1530
    },
    {
      "epoch": 0.0842644064065166,
      "grad_norm": 0.8710927963256836,
      "learning_rate": 9.96157326275009e-06,
      "loss": 0.749,
      "step": 1531
    },
    {
      "epoch": 0.08431944520887226,
      "grad_norm": 0.9500633478164673,
      "learning_rate": 9.961519606871188e-06,
      "loss": 0.8994,
      "step": 1532
    },
    {
      "epoch": 0.08437448401122792,
      "grad_norm": 0.873257577419281,
      "learning_rate": 9.961465913702833e-06,
      "loss": 0.816,
      "step": 1533
    },
    {
      "epoch": 0.08442952281358358,
      "grad_norm": 0.8007022142410278,
      "learning_rate": 9.961412183245426e-06,
      "loss": 0.787,
      "step": 1534
    },
    {
      "epoch": 0.08448456161593924,
      "grad_norm": 0.8998435139656067,
      "learning_rate": 9.961358415499374e-06,
      "loss": 0.8741,
      "step": 1535
    },
    {
      "epoch": 0.08453960041829489,
      "grad_norm": 0.9152502417564392,
      "learning_rate": 9.961304610465081e-06,
      "loss": 0.9749,
      "step": 1536
    },
    {
      "epoch": 0.08459463922065055,
      "grad_norm": 0.8961958289146423,
      "learning_rate": 9.961250768142949e-06,
      "loss": 0.8683,
      "step": 1537
    },
    {
      "epoch": 0.08464967802300621,
      "grad_norm": 0.8683995008468628,
      "learning_rate": 9.961196888533387e-06,
      "loss": 0.8347,
      "step": 1538
    },
    {
      "epoch": 0.08470471682536188,
      "grad_norm": 0.835221529006958,
      "learning_rate": 9.961142971636795e-06,
      "loss": 0.8936,
      "step": 1539
    },
    {
      "epoch": 0.08475975562771754,
      "grad_norm": 0.8666725158691406,
      "learning_rate": 9.96108901745358e-06,
      "loss": 0.7344,
      "step": 1540
    },
    {
      "epoch": 0.0848147944300732,
      "grad_norm": 0.9509082436561584,
      "learning_rate": 9.96103502598415e-06,
      "loss": 0.8965,
      "step": 1541
    },
    {
      "epoch": 0.08486983323242886,
      "grad_norm": 0.8134233951568604,
      "learning_rate": 9.960980997228908e-06,
      "loss": 0.797,
      "step": 1542
    },
    {
      "epoch": 0.08492487203478452,
      "grad_norm": 1.0432242155075073,
      "learning_rate": 9.96092693118826e-06,
      "loss": 0.8754,
      "step": 1543
    },
    {
      "epoch": 0.08497991083714018,
      "grad_norm": 0.9560218453407288,
      "learning_rate": 9.960872827862613e-06,
      "loss": 0.9238,
      "step": 1544
    },
    {
      "epoch": 0.08503494963949584,
      "grad_norm": 0.8471649885177612,
      "learning_rate": 9.960818687252374e-06,
      "loss": 0.8622,
      "step": 1545
    },
    {
      "epoch": 0.0850899884418515,
      "grad_norm": 1.2584747076034546,
      "learning_rate": 9.960764509357951e-06,
      "loss": 0.8007,
      "step": 1546
    },
    {
      "epoch": 0.08514502724420717,
      "grad_norm": 0.8730618953704834,
      "learning_rate": 9.960710294179748e-06,
      "loss": 0.7412,
      "step": 1547
    },
    {
      "epoch": 0.08520006604656283,
      "grad_norm": 0.8361592292785645,
      "learning_rate": 9.960656041718176e-06,
      "loss": 0.7018,
      "step": 1548
    },
    {
      "epoch": 0.08525510484891849,
      "grad_norm": 0.8351722359657288,
      "learning_rate": 9.96060175197364e-06,
      "loss": 0.843,
      "step": 1549
    },
    {
      "epoch": 0.08531014365127415,
      "grad_norm": 0.8665090203285217,
      "learning_rate": 9.960547424946549e-06,
      "loss": 0.8235,
      "step": 1550
    },
    {
      "epoch": 0.08536518245362981,
      "grad_norm": 0.9254478812217712,
      "learning_rate": 9.960493060637313e-06,
      "loss": 0.8122,
      "step": 1551
    },
    {
      "epoch": 0.08542022125598547,
      "grad_norm": 0.8712261319160461,
      "learning_rate": 9.960438659046337e-06,
      "loss": 0.823,
      "step": 1552
    },
    {
      "epoch": 0.08547526005834113,
      "grad_norm": 0.9027207493782043,
      "learning_rate": 9.960384220174033e-06,
      "loss": 0.7964,
      "step": 1553
    },
    {
      "epoch": 0.0855302988606968,
      "grad_norm": 0.854626476764679,
      "learning_rate": 9.960329744020808e-06,
      "loss": 0.755,
      "step": 1554
    },
    {
      "epoch": 0.08558533766305246,
      "grad_norm": 0.9398048520088196,
      "learning_rate": 9.960275230587073e-06,
      "loss": 0.8607,
      "step": 1555
    },
    {
      "epoch": 0.08564037646540812,
      "grad_norm": 1.008002758026123,
      "learning_rate": 9.960220679873238e-06,
      "loss": 0.9711,
      "step": 1556
    },
    {
      "epoch": 0.08569541526776378,
      "grad_norm": 0.8999453783035278,
      "learning_rate": 9.96016609187971e-06,
      "loss": 0.8233,
      "step": 1557
    },
    {
      "epoch": 0.08575045407011943,
      "grad_norm": 0.8912106156349182,
      "learning_rate": 9.960111466606903e-06,
      "loss": 0.8271,
      "step": 1558
    },
    {
      "epoch": 0.08580549287247509,
      "grad_norm": 0.9269998073577881,
      "learning_rate": 9.960056804055227e-06,
      "loss": 0.7959,
      "step": 1559
    },
    {
      "epoch": 0.08586053167483075,
      "grad_norm": 1.083815336227417,
      "learning_rate": 9.96000210422509e-06,
      "loss": 0.9436,
      "step": 1560
    },
    {
      "epoch": 0.08591557047718641,
      "grad_norm": 0.8906280398368835,
      "learning_rate": 9.959947367116905e-06,
      "loss": 0.9317,
      "step": 1561
    },
    {
      "epoch": 0.08597060927954207,
      "grad_norm": 1.211696743965149,
      "learning_rate": 9.959892592731084e-06,
      "loss": 0.9076,
      "step": 1562
    },
    {
      "epoch": 0.08602564808189773,
      "grad_norm": 0.9050534963607788,
      "learning_rate": 9.959837781068038e-06,
      "loss": 0.8728,
      "step": 1563
    },
    {
      "epoch": 0.0860806868842534,
      "grad_norm": 0.9384796619415283,
      "learning_rate": 9.959782932128178e-06,
      "loss": 0.9277,
      "step": 1564
    },
    {
      "epoch": 0.08613572568660906,
      "grad_norm": 0.795844316482544,
      "learning_rate": 9.959728045911915e-06,
      "loss": 0.7666,
      "step": 1565
    },
    {
      "epoch": 0.08619076448896472,
      "grad_norm": 0.925956666469574,
      "learning_rate": 9.959673122419668e-06,
      "loss": 0.815,
      "step": 1566
    },
    {
      "epoch": 0.08624580329132038,
      "grad_norm": 0.898047924041748,
      "learning_rate": 9.959618161651843e-06,
      "loss": 0.8131,
      "step": 1567
    },
    {
      "epoch": 0.08630084209367604,
      "grad_norm": 0.8656220436096191,
      "learning_rate": 9.959563163608856e-06,
      "loss": 0.9336,
      "step": 1568
    },
    {
      "epoch": 0.0863558808960317,
      "grad_norm": 0.9184645414352417,
      "learning_rate": 9.95950812829112e-06,
      "loss": 0.9557,
      "step": 1569
    },
    {
      "epoch": 0.08641091969838736,
      "grad_norm": 0.8607667684555054,
      "learning_rate": 9.959453055699048e-06,
      "loss": 0.8272,
      "step": 1570
    },
    {
      "epoch": 0.08646595850074303,
      "grad_norm": 0.9561272263526917,
      "learning_rate": 9.959397945833056e-06,
      "loss": 0.8876,
      "step": 1571
    },
    {
      "epoch": 0.08652099730309869,
      "grad_norm": 0.8562412261962891,
      "learning_rate": 9.959342798693556e-06,
      "loss": 0.8404,
      "step": 1572
    },
    {
      "epoch": 0.08657603610545435,
      "grad_norm": 0.8924610614776611,
      "learning_rate": 9.95928761428096e-06,
      "loss": 0.8779,
      "step": 1573
    },
    {
      "epoch": 0.08663107490781001,
      "grad_norm": 0.8343208432197571,
      "learning_rate": 9.95923239259569e-06,
      "loss": 0.8992,
      "step": 1574
    },
    {
      "epoch": 0.08668611371016567,
      "grad_norm": 0.8835015296936035,
      "learning_rate": 9.959177133638155e-06,
      "loss": 1.0026,
      "step": 1575
    },
    {
      "epoch": 0.08674115251252133,
      "grad_norm": 0.9540221095085144,
      "learning_rate": 9.959121837408771e-06,
      "loss": 0.8507,
      "step": 1576
    },
    {
      "epoch": 0.086796191314877,
      "grad_norm": 1.087817668914795,
      "learning_rate": 9.959066503907957e-06,
      "loss": 0.8607,
      "step": 1577
    },
    {
      "epoch": 0.08685123011723266,
      "grad_norm": 0.8072447180747986,
      "learning_rate": 9.959011133136124e-06,
      "loss": 0.882,
      "step": 1578
    },
    {
      "epoch": 0.0869062689195883,
      "grad_norm": 0.7646876573562622,
      "learning_rate": 9.958955725093694e-06,
      "loss": 0.7653,
      "step": 1579
    },
    {
      "epoch": 0.08696130772194396,
      "grad_norm": 0.8979537487030029,
      "learning_rate": 9.958900279781078e-06,
      "loss": 0.9033,
      "step": 1580
    },
    {
      "epoch": 0.08701634652429963,
      "grad_norm": 0.9445611834526062,
      "learning_rate": 9.958844797198696e-06,
      "loss": 0.9423,
      "step": 1581
    },
    {
      "epoch": 0.08707138532665529,
      "grad_norm": 0.8836671113967896,
      "learning_rate": 9.958789277346963e-06,
      "loss": 0.839,
      "step": 1582
    },
    {
      "epoch": 0.08712642412901095,
      "grad_norm": 1.0333542823791504,
      "learning_rate": 9.958733720226296e-06,
      "loss": 0.9211,
      "step": 1583
    },
    {
      "epoch": 0.08718146293136661,
      "grad_norm": 0.8084085583686829,
      "learning_rate": 9.958678125837117e-06,
      "loss": 0.8387,
      "step": 1584
    },
    {
      "epoch": 0.08723650173372227,
      "grad_norm": 0.7769419550895691,
      "learning_rate": 9.958622494179838e-06,
      "loss": 0.8307,
      "step": 1585
    },
    {
      "epoch": 0.08729154053607793,
      "grad_norm": 0.8387578129768372,
      "learning_rate": 9.95856682525488e-06,
      "loss": 0.8001,
      "step": 1586
    },
    {
      "epoch": 0.0873465793384336,
      "grad_norm": 0.8989812731742859,
      "learning_rate": 9.95851111906266e-06,
      "loss": 0.7752,
      "step": 1587
    },
    {
      "epoch": 0.08740161814078926,
      "grad_norm": 0.8558734655380249,
      "learning_rate": 9.958455375603602e-06,
      "loss": 0.8149,
      "step": 1588
    },
    {
      "epoch": 0.08745665694314492,
      "grad_norm": 0.8890896439552307,
      "learning_rate": 9.958399594878117e-06,
      "loss": 0.8232,
      "step": 1589
    },
    {
      "epoch": 0.08751169574550058,
      "grad_norm": 0.875912070274353,
      "learning_rate": 9.95834377688663e-06,
      "loss": 0.7458,
      "step": 1590
    },
    {
      "epoch": 0.08756673454785624,
      "grad_norm": 0.808355987071991,
      "learning_rate": 9.958287921629557e-06,
      "loss": 0.8296,
      "step": 1591
    },
    {
      "epoch": 0.0876217733502119,
      "grad_norm": 0.9637090563774109,
      "learning_rate": 9.958232029107318e-06,
      "loss": 0.8769,
      "step": 1592
    },
    {
      "epoch": 0.08767681215256756,
      "grad_norm": 0.8980715870857239,
      "learning_rate": 9.958176099320336e-06,
      "loss": 0.7995,
      "step": 1593
    },
    {
      "epoch": 0.08773185095492322,
      "grad_norm": 0.9369860291481018,
      "learning_rate": 9.95812013226903e-06,
      "loss": 0.8545,
      "step": 1594
    },
    {
      "epoch": 0.08778688975727889,
      "grad_norm": 0.8589349389076233,
      "learning_rate": 9.958064127953819e-06,
      "loss": 0.8693,
      "step": 1595
    },
    {
      "epoch": 0.08784192855963455,
      "grad_norm": 0.929207444190979,
      "learning_rate": 9.958008086375126e-06,
      "loss": 0.811,
      "step": 1596
    },
    {
      "epoch": 0.08789696736199021,
      "grad_norm": 1.0825661420822144,
      "learning_rate": 9.957952007533371e-06,
      "loss": 1.0145,
      "step": 1597
    },
    {
      "epoch": 0.08795200616434587,
      "grad_norm": 0.8818382024765015,
      "learning_rate": 9.957895891428978e-06,
      "loss": 0.7771,
      "step": 1598
    },
    {
      "epoch": 0.08800704496670153,
      "grad_norm": 0.882780909538269,
      "learning_rate": 9.957839738062363e-06,
      "loss": 0.8857,
      "step": 1599
    },
    {
      "epoch": 0.08806208376905718,
      "grad_norm": 0.9136924743652344,
      "learning_rate": 9.957783547433955e-06,
      "loss": 0.8873,
      "step": 1600
    },
    {
      "epoch": 0.08811712257141284,
      "grad_norm": 0.8896858096122742,
      "learning_rate": 9.95772731954417e-06,
      "loss": 0.8463,
      "step": 1601
    },
    {
      "epoch": 0.0881721613737685,
      "grad_norm": 0.8671631813049316,
      "learning_rate": 9.957671054393436e-06,
      "loss": 0.8333,
      "step": 1602
    },
    {
      "epoch": 0.08822720017612416,
      "grad_norm": 0.9442896246910095,
      "learning_rate": 9.957614751982172e-06,
      "loss": 0.9676,
      "step": 1603
    },
    {
      "epoch": 0.08828223897847982,
      "grad_norm": 0.8249240517616272,
      "learning_rate": 9.957558412310803e-06,
      "loss": 0.7746,
      "step": 1604
    },
    {
      "epoch": 0.08833727778083549,
      "grad_norm": 0.8125253319740295,
      "learning_rate": 9.957502035379751e-06,
      "loss": 0.7816,
      "step": 1605
    },
    {
      "epoch": 0.08839231658319115,
      "grad_norm": 0.8467233777046204,
      "learning_rate": 9.957445621189442e-06,
      "loss": 0.7697,
      "step": 1606
    },
    {
      "epoch": 0.08844735538554681,
      "grad_norm": 0.8322175145149231,
      "learning_rate": 9.957389169740299e-06,
      "loss": 0.7561,
      "step": 1607
    },
    {
      "epoch": 0.08850239418790247,
      "grad_norm": 0.869163453578949,
      "learning_rate": 9.957332681032746e-06,
      "loss": 0.8984,
      "step": 1608
    },
    {
      "epoch": 0.08855743299025813,
      "grad_norm": 0.8755944967269897,
      "learning_rate": 9.957276155067206e-06,
      "loss": 0.8016,
      "step": 1609
    },
    {
      "epoch": 0.08861247179261379,
      "grad_norm": 0.8152669668197632,
      "learning_rate": 9.957219591844108e-06,
      "loss": 0.7763,
      "step": 1610
    },
    {
      "epoch": 0.08866751059496945,
      "grad_norm": 0.979752779006958,
      "learning_rate": 9.957162991363871e-06,
      "loss": 0.7755,
      "step": 1611
    },
    {
      "epoch": 0.08872254939732512,
      "grad_norm": 1.0481054782867432,
      "learning_rate": 9.957106353626926e-06,
      "loss": 0.9395,
      "step": 1612
    },
    {
      "epoch": 0.08877758819968078,
      "grad_norm": 0.7773686647415161,
      "learning_rate": 9.957049678633697e-06,
      "loss": 0.7713,
      "step": 1613
    },
    {
      "epoch": 0.08883262700203644,
      "grad_norm": 0.838979959487915,
      "learning_rate": 9.956992966384609e-06,
      "loss": 0.7909,
      "step": 1614
    },
    {
      "epoch": 0.0888876658043921,
      "grad_norm": 0.9527049660682678,
      "learning_rate": 9.956936216880089e-06,
      "loss": 0.7944,
      "step": 1615
    },
    {
      "epoch": 0.08894270460674776,
      "grad_norm": 0.7967305183410645,
      "learning_rate": 9.956879430120561e-06,
      "loss": 0.7703,
      "step": 1616
    },
    {
      "epoch": 0.08899774340910342,
      "grad_norm": 0.9065802097320557,
      "learning_rate": 9.956822606106456e-06,
      "loss": 0.8188,
      "step": 1617
    },
    {
      "epoch": 0.08905278221145908,
      "grad_norm": 0.7329322099685669,
      "learning_rate": 9.956765744838199e-06,
      "loss": 0.8043,
      "step": 1618
    },
    {
      "epoch": 0.08910782101381474,
      "grad_norm": 0.864973247051239,
      "learning_rate": 9.95670884631622e-06,
      "loss": 0.8334,
      "step": 1619
    },
    {
      "epoch": 0.0891628598161704,
      "grad_norm": 1.073559045791626,
      "learning_rate": 9.95665191054094e-06,
      "loss": 0.7755,
      "step": 1620
    },
    {
      "epoch": 0.08921789861852607,
      "grad_norm": 0.7347918748855591,
      "learning_rate": 9.956594937512794e-06,
      "loss": 0.7556,
      "step": 1621
    },
    {
      "epoch": 0.08927293742088172,
      "grad_norm": 0.8756610751152039,
      "learning_rate": 9.956537927232205e-06,
      "loss": 0.8129,
      "step": 1622
    },
    {
      "epoch": 0.08932797622323738,
      "grad_norm": 0.9132435917854309,
      "learning_rate": 9.956480879699605e-06,
      "loss": 0.8221,
      "step": 1623
    },
    {
      "epoch": 0.08938301502559304,
      "grad_norm": 1.1978256702423096,
      "learning_rate": 9.956423794915421e-06,
      "loss": 0.8651,
      "step": 1624
    },
    {
      "epoch": 0.0894380538279487,
      "grad_norm": 0.8493894934654236,
      "learning_rate": 9.956366672880082e-06,
      "loss": 0.7267,
      "step": 1625
    },
    {
      "epoch": 0.08949309263030436,
      "grad_norm": 1.0971951484680176,
      "learning_rate": 9.956309513594019e-06,
      "loss": 0.7852,
      "step": 1626
    },
    {
      "epoch": 0.08954813143266002,
      "grad_norm": 0.899974524974823,
      "learning_rate": 9.95625231705766e-06,
      "loss": 0.8868,
      "step": 1627
    },
    {
      "epoch": 0.08960317023501568,
      "grad_norm": 0.8995566368103027,
      "learning_rate": 9.956195083271436e-06,
      "loss": 0.87,
      "step": 1628
    },
    {
      "epoch": 0.08965820903737134,
      "grad_norm": 0.8924218416213989,
      "learning_rate": 9.956137812235776e-06,
      "loss": 0.7885,
      "step": 1629
    },
    {
      "epoch": 0.089713247839727,
      "grad_norm": 0.9232820868492126,
      "learning_rate": 9.956080503951108e-06,
      "loss": 0.7923,
      "step": 1630
    },
    {
      "epoch": 0.08976828664208267,
      "grad_norm": 0.9298982620239258,
      "learning_rate": 9.956023158417869e-06,
      "loss": 0.8625,
      "step": 1631
    },
    {
      "epoch": 0.08982332544443833,
      "grad_norm": 0.86515212059021,
      "learning_rate": 9.955965775636488e-06,
      "loss": 0.7683,
      "step": 1632
    },
    {
      "epoch": 0.08987836424679399,
      "grad_norm": 0.8016952276229858,
      "learning_rate": 9.955908355607392e-06,
      "loss": 0.8122,
      "step": 1633
    },
    {
      "epoch": 0.08993340304914965,
      "grad_norm": 0.842703640460968,
      "learning_rate": 9.955850898331015e-06,
      "loss": 0.8487,
      "step": 1634
    },
    {
      "epoch": 0.08998844185150531,
      "grad_norm": 0.8239083886146545,
      "learning_rate": 9.95579340380779e-06,
      "loss": 0.8701,
      "step": 1635
    },
    {
      "epoch": 0.09004348065386097,
      "grad_norm": 0.8575418591499329,
      "learning_rate": 9.955735872038149e-06,
      "loss": 0.8263,
      "step": 1636
    },
    {
      "epoch": 0.09009851945621664,
      "grad_norm": 0.8884586095809937,
      "learning_rate": 9.955678303022522e-06,
      "loss": 0.8112,
      "step": 1637
    },
    {
      "epoch": 0.0901535582585723,
      "grad_norm": 0.9024681448936462,
      "learning_rate": 9.955620696761345e-06,
      "loss": 0.9174,
      "step": 1638
    },
    {
      "epoch": 0.09020859706092796,
      "grad_norm": 0.8151944875717163,
      "learning_rate": 9.955563053255049e-06,
      "loss": 0.806,
      "step": 1639
    },
    {
      "epoch": 0.09026363586328362,
      "grad_norm": 0.8292184472084045,
      "learning_rate": 9.955505372504069e-06,
      "loss": 0.8007,
      "step": 1640
    },
    {
      "epoch": 0.09031867466563928,
      "grad_norm": 0.9445936679840088,
      "learning_rate": 9.955447654508835e-06,
      "loss": 0.7089,
      "step": 1641
    },
    {
      "epoch": 0.09037371346799494,
      "grad_norm": 0.781579315662384,
      "learning_rate": 9.955389899269782e-06,
      "loss": 0.8224,
      "step": 1642
    },
    {
      "epoch": 0.09042875227035059,
      "grad_norm": 0.9028880596160889,
      "learning_rate": 9.955332106787348e-06,
      "loss": 0.7976,
      "step": 1643
    },
    {
      "epoch": 0.09048379107270625,
      "grad_norm": 1.0336887836456299,
      "learning_rate": 9.955274277061963e-06,
      "loss": 0.9296,
      "step": 1644
    },
    {
      "epoch": 0.09053882987506191,
      "grad_norm": 0.8894197940826416,
      "learning_rate": 9.955216410094062e-06,
      "loss": 0.815,
      "step": 1645
    },
    {
      "epoch": 0.09059386867741757,
      "grad_norm": 0.8955528140068054,
      "learning_rate": 9.955158505884083e-06,
      "loss": 0.8707,
      "step": 1646
    },
    {
      "epoch": 0.09064890747977324,
      "grad_norm": 0.8012683987617493,
      "learning_rate": 9.955100564432458e-06,
      "loss": 0.7467,
      "step": 1647
    },
    {
      "epoch": 0.0907039462821289,
      "grad_norm": 0.917969286441803,
      "learning_rate": 9.955042585739623e-06,
      "loss": 0.8835,
      "step": 1648
    },
    {
      "epoch": 0.09075898508448456,
      "grad_norm": 0.8066666722297668,
      "learning_rate": 9.954984569806014e-06,
      "loss": 0.8338,
      "step": 1649
    },
    {
      "epoch": 0.09081402388684022,
      "grad_norm": 1.1324070692062378,
      "learning_rate": 9.954926516632069e-06,
      "loss": 0.8245,
      "step": 1650
    },
    {
      "epoch": 0.09086906268919588,
      "grad_norm": 0.8196014761924744,
      "learning_rate": 9.954868426218222e-06,
      "loss": 0.7897,
      "step": 1651
    },
    {
      "epoch": 0.09092410149155154,
      "grad_norm": 0.8713478446006775,
      "learning_rate": 9.95481029856491e-06,
      "loss": 0.891,
      "step": 1652
    },
    {
      "epoch": 0.0909791402939072,
      "grad_norm": 0.8489059805870056,
      "learning_rate": 9.954752133672569e-06,
      "loss": 0.7748,
      "step": 1653
    },
    {
      "epoch": 0.09103417909626287,
      "grad_norm": 0.8914602994918823,
      "learning_rate": 9.954693931541638e-06,
      "loss": 0.8657,
      "step": 1654
    },
    {
      "epoch": 0.09108921789861853,
      "grad_norm": 0.9031614661216736,
      "learning_rate": 9.954635692172555e-06,
      "loss": 0.7409,
      "step": 1655
    },
    {
      "epoch": 0.09114425670097419,
      "grad_norm": 0.8680000305175781,
      "learning_rate": 9.954577415565756e-06,
      "loss": 0.8535,
      "step": 1656
    },
    {
      "epoch": 0.09119929550332985,
      "grad_norm": 0.830596923828125,
      "learning_rate": 9.954519101721679e-06,
      "loss": 0.8601,
      "step": 1657
    },
    {
      "epoch": 0.09125433430568551,
      "grad_norm": 0.9041332602500916,
      "learning_rate": 9.954460750640762e-06,
      "loss": 0.9104,
      "step": 1658
    },
    {
      "epoch": 0.09130937310804117,
      "grad_norm": 0.7786296606063843,
      "learning_rate": 9.954402362323445e-06,
      "loss": 0.7671,
      "step": 1659
    },
    {
      "epoch": 0.09136441191039683,
      "grad_norm": 1.0363564491271973,
      "learning_rate": 9.954343936770165e-06,
      "loss": 0.9339,
      "step": 1660
    },
    {
      "epoch": 0.0914194507127525,
      "grad_norm": 0.8049986958503723,
      "learning_rate": 9.954285473981363e-06,
      "loss": 0.8125,
      "step": 1661
    },
    {
      "epoch": 0.09147448951510816,
      "grad_norm": 0.7842011451721191,
      "learning_rate": 9.954226973957477e-06,
      "loss": 0.7153,
      "step": 1662
    },
    {
      "epoch": 0.09152952831746382,
      "grad_norm": 0.8929729461669922,
      "learning_rate": 9.954168436698948e-06,
      "loss": 0.9563,
      "step": 1663
    },
    {
      "epoch": 0.09158456711981948,
      "grad_norm": 0.8850226402282715,
      "learning_rate": 9.954109862206216e-06,
      "loss": 0.8257,
      "step": 1664
    },
    {
      "epoch": 0.09163960592217513,
      "grad_norm": 0.8673348426818848,
      "learning_rate": 9.954051250479719e-06,
      "loss": 0.9489,
      "step": 1665
    },
    {
      "epoch": 0.09169464472453079,
      "grad_norm": 0.8726119995117188,
      "learning_rate": 9.9539926015199e-06,
      "loss": 0.8222,
      "step": 1666
    },
    {
      "epoch": 0.09174968352688645,
      "grad_norm": 0.7609312534332275,
      "learning_rate": 9.953933915327196e-06,
      "loss": 0.7749,
      "step": 1667
    },
    {
      "epoch": 0.09180472232924211,
      "grad_norm": 0.857404887676239,
      "learning_rate": 9.953875191902055e-06,
      "loss": 0.8496,
      "step": 1668
    },
    {
      "epoch": 0.09185976113159777,
      "grad_norm": 0.7835526466369629,
      "learning_rate": 9.953816431244909e-06,
      "loss": 0.7258,
      "step": 1669
    },
    {
      "epoch": 0.09191479993395343,
      "grad_norm": 0.944984495639801,
      "learning_rate": 9.95375763335621e-06,
      "loss": 0.902,
      "step": 1670
    },
    {
      "epoch": 0.0919698387363091,
      "grad_norm": 0.9038936495780945,
      "learning_rate": 9.953698798236391e-06,
      "loss": 0.7559,
      "step": 1671
    },
    {
      "epoch": 0.09202487753866476,
      "grad_norm": 0.8450848460197449,
      "learning_rate": 9.953639925885898e-06,
      "loss": 0.8338,
      "step": 1672
    },
    {
      "epoch": 0.09207991634102042,
      "grad_norm": 0.827419102191925,
      "learning_rate": 9.953581016305175e-06,
      "loss": 0.8167,
      "step": 1673
    },
    {
      "epoch": 0.09213495514337608,
      "grad_norm": 0.8517075777053833,
      "learning_rate": 9.953522069494663e-06,
      "loss": 0.8681,
      "step": 1674
    },
    {
      "epoch": 0.09218999394573174,
      "grad_norm": 0.9504323601722717,
      "learning_rate": 9.953463085454804e-06,
      "loss": 0.8688,
      "step": 1675
    },
    {
      "epoch": 0.0922450327480874,
      "grad_norm": 0.8905719518661499,
      "learning_rate": 9.953404064186044e-06,
      "loss": 0.8818,
      "step": 1676
    },
    {
      "epoch": 0.09230007155044306,
      "grad_norm": 0.9223340153694153,
      "learning_rate": 9.953345005688822e-06,
      "loss": 0.8752,
      "step": 1677
    },
    {
      "epoch": 0.09235511035279872,
      "grad_norm": 1.0500547885894775,
      "learning_rate": 9.953285909963588e-06,
      "loss": 0.7816,
      "step": 1678
    },
    {
      "epoch": 0.09241014915515439,
      "grad_norm": 0.8407441973686218,
      "learning_rate": 9.953226777010781e-06,
      "loss": 0.745,
      "step": 1679
    },
    {
      "epoch": 0.09246518795751005,
      "grad_norm": 0.7997288107872009,
      "learning_rate": 9.953167606830847e-06,
      "loss": 0.8171,
      "step": 1680
    },
    {
      "epoch": 0.09252022675986571,
      "grad_norm": 0.9752318859100342,
      "learning_rate": 9.953108399424234e-06,
      "loss": 0.8719,
      "step": 1681
    },
    {
      "epoch": 0.09257526556222137,
      "grad_norm": 0.8524298667907715,
      "learning_rate": 9.953049154791382e-06,
      "loss": 0.8257,
      "step": 1682
    },
    {
      "epoch": 0.09263030436457703,
      "grad_norm": 0.9460529088973999,
      "learning_rate": 9.952989872932739e-06,
      "loss": 0.7278,
      "step": 1683
    },
    {
      "epoch": 0.0926853431669327,
      "grad_norm": 0.8959575891494751,
      "learning_rate": 9.95293055384875e-06,
      "loss": 0.903,
      "step": 1684
    },
    {
      "epoch": 0.09274038196928835,
      "grad_norm": 0.8764386177062988,
      "learning_rate": 9.95287119753986e-06,
      "loss": 0.7958,
      "step": 1685
    },
    {
      "epoch": 0.092795420771644,
      "grad_norm": 0.9611337184906006,
      "learning_rate": 9.952811804006517e-06,
      "loss": 0.8726,
      "step": 1686
    },
    {
      "epoch": 0.09285045957399966,
      "grad_norm": 0.8155574202537537,
      "learning_rate": 9.952752373249165e-06,
      "loss": 0.7882,
      "step": 1687
    },
    {
      "epoch": 0.09290549837635532,
      "grad_norm": 0.8789697289466858,
      "learning_rate": 9.952692905268253e-06,
      "loss": 0.8642,
      "step": 1688
    },
    {
      "epoch": 0.09296053717871099,
      "grad_norm": 0.7910027503967285,
      "learning_rate": 9.952633400064227e-06,
      "loss": 0.7852,
      "step": 1689
    },
    {
      "epoch": 0.09301557598106665,
      "grad_norm": 0.815819501876831,
      "learning_rate": 9.952573857637533e-06,
      "loss": 0.8606,
      "step": 1690
    },
    {
      "epoch": 0.09307061478342231,
      "grad_norm": 0.9840701818466187,
      "learning_rate": 9.95251427798862e-06,
      "loss": 0.9349,
      "step": 1691
    },
    {
      "epoch": 0.09312565358577797,
      "grad_norm": 0.8715788722038269,
      "learning_rate": 9.952454661117936e-06,
      "loss": 0.813,
      "step": 1692
    },
    {
      "epoch": 0.09318069238813363,
      "grad_norm": 0.8287779092788696,
      "learning_rate": 9.952395007025926e-06,
      "loss": 0.8346,
      "step": 1693
    },
    {
      "epoch": 0.0932357311904893,
      "grad_norm": 0.9375059008598328,
      "learning_rate": 9.952335315713044e-06,
      "loss": 0.8868,
      "step": 1694
    },
    {
      "epoch": 0.09329076999284495,
      "grad_norm": 0.9063667058944702,
      "learning_rate": 9.952275587179734e-06,
      "loss": 0.9562,
      "step": 1695
    },
    {
      "epoch": 0.09334580879520062,
      "grad_norm": 0.816643476486206,
      "learning_rate": 9.952215821426447e-06,
      "loss": 0.7456,
      "step": 1696
    },
    {
      "epoch": 0.09340084759755628,
      "grad_norm": 0.9004347324371338,
      "learning_rate": 9.95215601845363e-06,
      "loss": 0.8545,
      "step": 1697
    },
    {
      "epoch": 0.09345588639991194,
      "grad_norm": 0.919195830821991,
      "learning_rate": 9.952096178261736e-06,
      "loss": 0.9347,
      "step": 1698
    },
    {
      "epoch": 0.0935109252022676,
      "grad_norm": 0.8313261866569519,
      "learning_rate": 9.952036300851211e-06,
      "loss": 0.9169,
      "step": 1699
    },
    {
      "epoch": 0.09356596400462326,
      "grad_norm": 0.8674910664558411,
      "learning_rate": 9.951976386222507e-06,
      "loss": 0.7621,
      "step": 1700
    },
    {
      "epoch": 0.09362100280697892,
      "grad_norm": 0.8931052684783936,
      "learning_rate": 9.951916434376074e-06,
      "loss": 0.8702,
      "step": 1701
    },
    {
      "epoch": 0.09367604160933458,
      "grad_norm": 0.8748393058776855,
      "learning_rate": 9.951856445312364e-06,
      "loss": 0.7446,
      "step": 1702
    },
    {
      "epoch": 0.09373108041169025,
      "grad_norm": 1.005459189414978,
      "learning_rate": 9.951796419031825e-06,
      "loss": 0.9843,
      "step": 1703
    },
    {
      "epoch": 0.09378611921404591,
      "grad_norm": 1.0155184268951416,
      "learning_rate": 9.95173635553491e-06,
      "loss": 0.8868,
      "step": 1704
    },
    {
      "epoch": 0.09384115801640157,
      "grad_norm": 2.1387271881103516,
      "learning_rate": 9.951676254822072e-06,
      "loss": 0.8691,
      "step": 1705
    },
    {
      "epoch": 0.09389619681875723,
      "grad_norm": 0.9768403768539429,
      "learning_rate": 9.951616116893757e-06,
      "loss": 0.8409,
      "step": 1706
    },
    {
      "epoch": 0.09395123562111289,
      "grad_norm": 0.7994607090950012,
      "learning_rate": 9.951555941750424e-06,
      "loss": 0.7836,
      "step": 1707
    },
    {
      "epoch": 0.09400627442346854,
      "grad_norm": 0.8460201025009155,
      "learning_rate": 9.95149572939252e-06,
      "loss": 0.8216,
      "step": 1708
    },
    {
      "epoch": 0.0940613132258242,
      "grad_norm": 0.8904135227203369,
      "learning_rate": 9.951435479820499e-06,
      "loss": 0.9053,
      "step": 1709
    },
    {
      "epoch": 0.09411635202817986,
      "grad_norm": 0.9084494113922119,
      "learning_rate": 9.951375193034815e-06,
      "loss": 0.9308,
      "step": 1710
    },
    {
      "epoch": 0.09417139083053552,
      "grad_norm": 1.0826482772827148,
      "learning_rate": 9.951314869035921e-06,
      "loss": 0.8468,
      "step": 1711
    },
    {
      "epoch": 0.09422642963289118,
      "grad_norm": 0.8068915009498596,
      "learning_rate": 9.95125450782427e-06,
      "loss": 0.8253,
      "step": 1712
    },
    {
      "epoch": 0.09428146843524685,
      "grad_norm": 0.8445400595664978,
      "learning_rate": 9.951194109400316e-06,
      "loss": 0.8386,
      "step": 1713
    },
    {
      "epoch": 0.09433650723760251,
      "grad_norm": 0.8180645704269409,
      "learning_rate": 9.951133673764513e-06,
      "loss": 0.7907,
      "step": 1714
    },
    {
      "epoch": 0.09439154603995817,
      "grad_norm": 0.8111036419868469,
      "learning_rate": 9.951073200917311e-06,
      "loss": 0.7918,
      "step": 1715
    },
    {
      "epoch": 0.09444658484231383,
      "grad_norm": 0.862042248249054,
      "learning_rate": 9.951012690859172e-06,
      "loss": 0.783,
      "step": 1716
    },
    {
      "epoch": 0.09450162364466949,
      "grad_norm": 0.8189615607261658,
      "learning_rate": 9.950952143590544e-06,
      "loss": 0.8192,
      "step": 1717
    },
    {
      "epoch": 0.09455666244702515,
      "grad_norm": 0.9714062809944153,
      "learning_rate": 9.950891559111887e-06,
      "loss": 0.774,
      "step": 1718
    },
    {
      "epoch": 0.09461170124938081,
      "grad_norm": 0.9691846370697021,
      "learning_rate": 9.950830937423655e-06,
      "loss": 0.8347,
      "step": 1719
    },
    {
      "epoch": 0.09466674005173648,
      "grad_norm": 0.8488250970840454,
      "learning_rate": 9.950770278526301e-06,
      "loss": 0.8228,
      "step": 1720
    },
    {
      "epoch": 0.09472177885409214,
      "grad_norm": 0.8638359904289246,
      "learning_rate": 9.950709582420282e-06,
      "loss": 0.8973,
      "step": 1721
    },
    {
      "epoch": 0.0947768176564478,
      "grad_norm": 1.0148643255233765,
      "learning_rate": 9.950648849106058e-06,
      "loss": 0.9638,
      "step": 1722
    },
    {
      "epoch": 0.09483185645880346,
      "grad_norm": 0.8870131969451904,
      "learning_rate": 9.95058807858408e-06,
      "loss": 0.8259,
      "step": 1723
    },
    {
      "epoch": 0.09488689526115912,
      "grad_norm": 0.9134769439697266,
      "learning_rate": 9.950527270854807e-06,
      "loss": 0.865,
      "step": 1724
    },
    {
      "epoch": 0.09494193406351478,
      "grad_norm": 0.7221654653549194,
      "learning_rate": 9.950466425918697e-06,
      "loss": 0.7593,
      "step": 1725
    },
    {
      "epoch": 0.09499697286587044,
      "grad_norm": 0.9386674165725708,
      "learning_rate": 9.950405543776207e-06,
      "loss": 0.9508,
      "step": 1726
    },
    {
      "epoch": 0.0950520116682261,
      "grad_norm": 0.7850627899169922,
      "learning_rate": 9.950344624427795e-06,
      "loss": 0.7999,
      "step": 1727
    },
    {
      "epoch": 0.09510705047058177,
      "grad_norm": 0.921198308467865,
      "learning_rate": 9.950283667873916e-06,
      "loss": 0.8249,
      "step": 1728
    },
    {
      "epoch": 0.09516208927293741,
      "grad_norm": 0.9503389000892639,
      "learning_rate": 9.95022267411503e-06,
      "loss": 0.901,
      "step": 1729
    },
    {
      "epoch": 0.09521712807529308,
      "grad_norm": 0.7977343201637268,
      "learning_rate": 9.950161643151597e-06,
      "loss": 0.838,
      "step": 1730
    },
    {
      "epoch": 0.09527216687764874,
      "grad_norm": 0.9056238532066345,
      "learning_rate": 9.950100574984072e-06,
      "loss": 0.9756,
      "step": 1731
    },
    {
      "epoch": 0.0953272056800044,
      "grad_norm": 0.8092935681343079,
      "learning_rate": 9.950039469612918e-06,
      "loss": 0.8812,
      "step": 1732
    },
    {
      "epoch": 0.09538224448236006,
      "grad_norm": 0.823693573474884,
      "learning_rate": 9.949978327038592e-06,
      "loss": 0.7914,
      "step": 1733
    },
    {
      "epoch": 0.09543728328471572,
      "grad_norm": 0.9114876389503479,
      "learning_rate": 9.949917147261554e-06,
      "loss": 0.7944,
      "step": 1734
    },
    {
      "epoch": 0.09549232208707138,
      "grad_norm": 1.0084123611450195,
      "learning_rate": 9.949855930282262e-06,
      "loss": 0.8544,
      "step": 1735
    },
    {
      "epoch": 0.09554736088942704,
      "grad_norm": 0.842462956905365,
      "learning_rate": 9.949794676101181e-06,
      "loss": 0.7056,
      "step": 1736
    },
    {
      "epoch": 0.0956023996917827,
      "grad_norm": 1.00497305393219,
      "learning_rate": 9.949733384718766e-06,
      "loss": 0.8372,
      "step": 1737
    },
    {
      "epoch": 0.09565743849413837,
      "grad_norm": 1.0166410207748413,
      "learning_rate": 9.94967205613548e-06,
      "loss": 0.9316,
      "step": 1738
    },
    {
      "epoch": 0.09571247729649403,
      "grad_norm": 0.8520192503929138,
      "learning_rate": 9.949610690351784e-06,
      "loss": 0.786,
      "step": 1739
    },
    {
      "epoch": 0.09576751609884969,
      "grad_norm": 0.8003227114677429,
      "learning_rate": 9.949549287368139e-06,
      "loss": 0.8003,
      "step": 1740
    },
    {
      "epoch": 0.09582255490120535,
      "grad_norm": 0.8657151460647583,
      "learning_rate": 9.949487847185006e-06,
      "loss": 0.8407,
      "step": 1741
    },
    {
      "epoch": 0.09587759370356101,
      "grad_norm": 1.1119858026504517,
      "learning_rate": 9.949426369802848e-06,
      "loss": 0.8594,
      "step": 1742
    },
    {
      "epoch": 0.09593263250591667,
      "grad_norm": 0.8968474864959717,
      "learning_rate": 9.949364855222126e-06,
      "loss": 0.8254,
      "step": 1743
    },
    {
      "epoch": 0.09598767130827233,
      "grad_norm": 0.8740531206130981,
      "learning_rate": 9.949303303443304e-06,
      "loss": 0.8748,
      "step": 1744
    },
    {
      "epoch": 0.096042710110628,
      "grad_norm": 0.8833459615707397,
      "learning_rate": 9.94924171446684e-06,
      "loss": 0.838,
      "step": 1745
    },
    {
      "epoch": 0.09609774891298366,
      "grad_norm": 0.8783486485481262,
      "learning_rate": 9.949180088293201e-06,
      "loss": 0.7972,
      "step": 1746
    },
    {
      "epoch": 0.09615278771533932,
      "grad_norm": 0.9197877049446106,
      "learning_rate": 9.949118424922852e-06,
      "loss": 0.8669,
      "step": 1747
    },
    {
      "epoch": 0.09620782651769498,
      "grad_norm": 0.9771283864974976,
      "learning_rate": 9.949056724356251e-06,
      "loss": 0.8461,
      "step": 1748
    },
    {
      "epoch": 0.09626286532005064,
      "grad_norm": 0.8325022459030151,
      "learning_rate": 9.948994986593864e-06,
      "loss": 0.8482,
      "step": 1749
    },
    {
      "epoch": 0.0963179041224063,
      "grad_norm": 0.9732363224029541,
      "learning_rate": 9.948933211636158e-06,
      "loss": 0.8825,
      "step": 1750
    },
    {
      "epoch": 0.09637294292476195,
      "grad_norm": 0.8229798078536987,
      "learning_rate": 9.948871399483592e-06,
      "loss": 0.8079,
      "step": 1751
    },
    {
      "epoch": 0.09642798172711761,
      "grad_norm": 0.8861554265022278,
      "learning_rate": 9.948809550136635e-06,
      "loss": 0.8323,
      "step": 1752
    },
    {
      "epoch": 0.09648302052947327,
      "grad_norm": 1.0618904829025269,
      "learning_rate": 9.94874766359575e-06,
      "loss": 0.8519,
      "step": 1753
    },
    {
      "epoch": 0.09653805933182893,
      "grad_norm": 0.8494864702224731,
      "learning_rate": 9.948685739861403e-06,
      "loss": 0.961,
      "step": 1754
    },
    {
      "epoch": 0.0965930981341846,
      "grad_norm": 0.8872213959693909,
      "learning_rate": 9.948623778934058e-06,
      "loss": 0.9367,
      "step": 1755
    },
    {
      "epoch": 0.09664813693654026,
      "grad_norm": 0.8441230058670044,
      "learning_rate": 9.948561780814181e-06,
      "loss": 0.7654,
      "step": 1756
    },
    {
      "epoch": 0.09670317573889592,
      "grad_norm": 0.8072223663330078,
      "learning_rate": 9.948499745502239e-06,
      "loss": 0.7894,
      "step": 1757
    },
    {
      "epoch": 0.09675821454125158,
      "grad_norm": 0.8285261392593384,
      "learning_rate": 9.948437672998696e-06,
      "loss": 0.8351,
      "step": 1758
    },
    {
      "epoch": 0.09681325334360724,
      "grad_norm": 0.9272124767303467,
      "learning_rate": 9.94837556330402e-06,
      "loss": 0.8708,
      "step": 1759
    },
    {
      "epoch": 0.0968682921459629,
      "grad_norm": 0.8689375519752502,
      "learning_rate": 9.94831341641868e-06,
      "loss": 0.8478,
      "step": 1760
    },
    {
      "epoch": 0.09692333094831856,
      "grad_norm": 1.040784239768982,
      "learning_rate": 9.94825123234314e-06,
      "loss": 0.8915,
      "step": 1761
    },
    {
      "epoch": 0.09697836975067423,
      "grad_norm": 0.7819718718528748,
      "learning_rate": 9.948189011077867e-06,
      "loss": 0.7728,
      "step": 1762
    },
    {
      "epoch": 0.09703340855302989,
      "grad_norm": 0.7959379553794861,
      "learning_rate": 9.948126752623331e-06,
      "loss": 0.8248,
      "step": 1763
    },
    {
      "epoch": 0.09708844735538555,
      "grad_norm": 0.8844753503799438,
      "learning_rate": 9.94806445698e-06,
      "loss": 0.7742,
      "step": 1764
    },
    {
      "epoch": 0.09714348615774121,
      "grad_norm": 0.9168505668640137,
      "learning_rate": 9.948002124148339e-06,
      "loss": 0.9145,
      "step": 1765
    },
    {
      "epoch": 0.09719852496009687,
      "grad_norm": 0.7199662923812866,
      "learning_rate": 9.947939754128819e-06,
      "loss": 0.6652,
      "step": 1766
    },
    {
      "epoch": 0.09725356376245253,
      "grad_norm": 0.866470992565155,
      "learning_rate": 9.947877346921909e-06,
      "loss": 0.8293,
      "step": 1767
    },
    {
      "epoch": 0.0973086025648082,
      "grad_norm": 0.9124754667282104,
      "learning_rate": 9.947814902528078e-06,
      "loss": 0.8599,
      "step": 1768
    },
    {
      "epoch": 0.09736364136716386,
      "grad_norm": 0.9169870615005493,
      "learning_rate": 9.947752420947792e-06,
      "loss": 0.8382,
      "step": 1769
    },
    {
      "epoch": 0.09741868016951952,
      "grad_norm": 1.0147640705108643,
      "learning_rate": 9.947689902181526e-06,
      "loss": 0.8425,
      "step": 1770
    },
    {
      "epoch": 0.09747371897187518,
      "grad_norm": 0.778575599193573,
      "learning_rate": 9.947627346229745e-06,
      "loss": 0.6979,
      "step": 1771
    },
    {
      "epoch": 0.09752875777423083,
      "grad_norm": 0.815101146697998,
      "learning_rate": 9.947564753092922e-06,
      "loss": 0.8617,
      "step": 1772
    },
    {
      "epoch": 0.09758379657658649,
      "grad_norm": 0.9556358456611633,
      "learning_rate": 9.947502122771527e-06,
      "loss": 0.9009,
      "step": 1773
    },
    {
      "epoch": 0.09763883537894215,
      "grad_norm": 0.8603761196136475,
      "learning_rate": 9.94743945526603e-06,
      "loss": 0.9443,
      "step": 1774
    },
    {
      "epoch": 0.09769387418129781,
      "grad_norm": 0.8621761798858643,
      "learning_rate": 9.947376750576903e-06,
      "loss": 0.7537,
      "step": 1775
    },
    {
      "epoch": 0.09774891298365347,
      "grad_norm": 0.7399948835372925,
      "learning_rate": 9.947314008704616e-06,
      "loss": 0.7477,
      "step": 1776
    },
    {
      "epoch": 0.09780395178600913,
      "grad_norm": 0.8855582475662231,
      "learning_rate": 9.947251229649641e-06,
      "loss": 0.8745,
      "step": 1777
    },
    {
      "epoch": 0.0978589905883648,
      "grad_norm": 0.8718472719192505,
      "learning_rate": 9.947188413412452e-06,
      "loss": 0.9672,
      "step": 1778
    },
    {
      "epoch": 0.09791402939072046,
      "grad_norm": 0.8598514795303345,
      "learning_rate": 9.947125559993517e-06,
      "loss": 0.8278,
      "step": 1779
    },
    {
      "epoch": 0.09796906819307612,
      "grad_norm": 1.0373798608779907,
      "learning_rate": 9.947062669393312e-06,
      "loss": 0.8123,
      "step": 1780
    },
    {
      "epoch": 0.09802410699543178,
      "grad_norm": 1.0198705196380615,
      "learning_rate": 9.946999741612306e-06,
      "loss": 0.9039,
      "step": 1781
    },
    {
      "epoch": 0.09807914579778744,
      "grad_norm": 0.8770025968551636,
      "learning_rate": 9.946936776650977e-06,
      "loss": 0.8326,
      "step": 1782
    },
    {
      "epoch": 0.0981341846001431,
      "grad_norm": 0.7970215678215027,
      "learning_rate": 9.946873774509794e-06,
      "loss": 0.848,
      "step": 1783
    },
    {
      "epoch": 0.09818922340249876,
      "grad_norm": 0.90342777967453,
      "learning_rate": 9.946810735189231e-06,
      "loss": 0.7993,
      "step": 1784
    },
    {
      "epoch": 0.09824426220485442,
      "grad_norm": 1.2095681428909302,
      "learning_rate": 9.946747658689763e-06,
      "loss": 0.8544,
      "step": 1785
    },
    {
      "epoch": 0.09829930100721009,
      "grad_norm": 0.8500953316688538,
      "learning_rate": 9.946684545011866e-06,
      "loss": 0.8398,
      "step": 1786
    },
    {
      "epoch": 0.09835433980956575,
      "grad_norm": 0.8570724725723267,
      "learning_rate": 9.946621394156011e-06,
      "loss": 0.9255,
      "step": 1787
    },
    {
      "epoch": 0.09840937861192141,
      "grad_norm": 0.8314846158027649,
      "learning_rate": 9.946558206122672e-06,
      "loss": 0.8398,
      "step": 1788
    },
    {
      "epoch": 0.09846441741427707,
      "grad_norm": 0.8894716501235962,
      "learning_rate": 9.946494980912326e-06,
      "loss": 0.8612,
      "step": 1789
    },
    {
      "epoch": 0.09851945621663273,
      "grad_norm": 0.9555756449699402,
      "learning_rate": 9.94643171852545e-06,
      "loss": 0.9551,
      "step": 1790
    },
    {
      "epoch": 0.09857449501898839,
      "grad_norm": 0.9556692838668823,
      "learning_rate": 9.946368418962515e-06,
      "loss": 0.8175,
      "step": 1791
    },
    {
      "epoch": 0.09862953382134405,
      "grad_norm": 0.7288535833358765,
      "learning_rate": 9.946305082224e-06,
      "loss": 0.6162,
      "step": 1792
    },
    {
      "epoch": 0.09868457262369972,
      "grad_norm": 0.95478355884552,
      "learning_rate": 9.94624170831038e-06,
      "loss": 0.9089,
      "step": 1793
    },
    {
      "epoch": 0.09873961142605536,
      "grad_norm": 0.9080137610435486,
      "learning_rate": 9.946178297222133e-06,
      "loss": 0.9443,
      "step": 1794
    },
    {
      "epoch": 0.09879465022841102,
      "grad_norm": 0.8060124516487122,
      "learning_rate": 9.946114848959732e-06,
      "loss": 0.7412,
      "step": 1795
    },
    {
      "epoch": 0.09884968903076669,
      "grad_norm": 0.8487932085990906,
      "learning_rate": 9.946051363523655e-06,
      "loss": 0.7098,
      "step": 1796
    },
    {
      "epoch": 0.09890472783312235,
      "grad_norm": 0.8982037901878357,
      "learning_rate": 9.945987840914381e-06,
      "loss": 0.8304,
      "step": 1797
    },
    {
      "epoch": 0.09895976663547801,
      "grad_norm": 0.8124602437019348,
      "learning_rate": 9.945924281132386e-06,
      "loss": 0.8441,
      "step": 1798
    },
    {
      "epoch": 0.09901480543783367,
      "grad_norm": 0.8081663250923157,
      "learning_rate": 9.945860684178147e-06,
      "loss": 0.732,
      "step": 1799
    },
    {
      "epoch": 0.09906984424018933,
      "grad_norm": 0.7662907242774963,
      "learning_rate": 9.945797050052147e-06,
      "loss": 0.7538,
      "step": 1800
    },
    {
      "epoch": 0.09912488304254499,
      "grad_norm": 0.8418399095535278,
      "learning_rate": 9.945733378754856e-06,
      "loss": 0.8488,
      "step": 1801
    },
    {
      "epoch": 0.09917992184490065,
      "grad_norm": 0.7298988699913025,
      "learning_rate": 9.94566967028676e-06,
      "loss": 0.7822,
      "step": 1802
    },
    {
      "epoch": 0.09923496064725632,
      "grad_norm": 0.7788695693016052,
      "learning_rate": 9.945605924648332e-06,
      "loss": 0.8037,
      "step": 1803
    },
    {
      "epoch": 0.09928999944961198,
      "grad_norm": 0.939297080039978,
      "learning_rate": 9.945542141840054e-06,
      "loss": 0.8654,
      "step": 1804
    },
    {
      "epoch": 0.09934503825196764,
      "grad_norm": 0.9274358749389648,
      "learning_rate": 9.945478321862406e-06,
      "loss": 0.7712,
      "step": 1805
    },
    {
      "epoch": 0.0994000770543233,
      "grad_norm": 0.816561222076416,
      "learning_rate": 9.945414464715866e-06,
      "loss": 0.7676,
      "step": 1806
    },
    {
      "epoch": 0.09945511585667896,
      "grad_norm": 0.867915153503418,
      "learning_rate": 9.945350570400916e-06,
      "loss": 0.8343,
      "step": 1807
    },
    {
      "epoch": 0.09951015465903462,
      "grad_norm": 0.8446162939071655,
      "learning_rate": 9.945286638918034e-06,
      "loss": 0.8128,
      "step": 1808
    },
    {
      "epoch": 0.09956519346139028,
      "grad_norm": 0.8372986316680908,
      "learning_rate": 9.945222670267703e-06,
      "loss": 0.8611,
      "step": 1809
    },
    {
      "epoch": 0.09962023226374594,
      "grad_norm": 0.787836492061615,
      "learning_rate": 9.945158664450399e-06,
      "loss": 0.7286,
      "step": 1810
    },
    {
      "epoch": 0.0996752710661016,
      "grad_norm": 0.9293436408042908,
      "learning_rate": 9.945094621466609e-06,
      "loss": 0.8699,
      "step": 1811
    },
    {
      "epoch": 0.09973030986845727,
      "grad_norm": 0.8336932063102722,
      "learning_rate": 9.94503054131681e-06,
      "loss": 0.8222,
      "step": 1812
    },
    {
      "epoch": 0.09978534867081293,
      "grad_norm": 0.8310953378677368,
      "learning_rate": 9.944966424001486e-06,
      "loss": 0.8131,
      "step": 1813
    },
    {
      "epoch": 0.09984038747316859,
      "grad_norm": 0.7703443169593811,
      "learning_rate": 9.944902269521117e-06,
      "loss": 0.8135,
      "step": 1814
    },
    {
      "epoch": 0.09989542627552424,
      "grad_norm": 0.750990092754364,
      "learning_rate": 9.944838077876186e-06,
      "loss": 0.8137,
      "step": 1815
    },
    {
      "epoch": 0.0999504650778799,
      "grad_norm": 0.8502481579780579,
      "learning_rate": 9.944773849067178e-06,
      "loss": 0.8973,
      "step": 1816
    },
    {
      "epoch": 0.10000550388023556,
      "grad_norm": 0.8299791812896729,
      "learning_rate": 9.94470958309457e-06,
      "loss": 0.8341,
      "step": 1817
    },
    {
      "epoch": 0.10006054268259122,
      "grad_norm": 0.8519022464752197,
      "learning_rate": 9.94464527995885e-06,
      "loss": 0.8529,
      "step": 1818
    },
    {
      "epoch": 0.10011558148494688,
      "grad_norm": 0.9318063259124756,
      "learning_rate": 9.944580939660501e-06,
      "loss": 0.8978,
      "step": 1819
    },
    {
      "epoch": 0.10017062028730254,
      "grad_norm": 0.847023069858551,
      "learning_rate": 9.944516562200004e-06,
      "loss": 0.8007,
      "step": 1820
    },
    {
      "epoch": 0.1002256590896582,
      "grad_norm": 0.8817011117935181,
      "learning_rate": 9.944452147577844e-06,
      "loss": 0.8819,
      "step": 1821
    },
    {
      "epoch": 0.10028069789201387,
      "grad_norm": 0.8560144901275635,
      "learning_rate": 9.944387695794505e-06,
      "loss": 0.8219,
      "step": 1822
    },
    {
      "epoch": 0.10033573669436953,
      "grad_norm": 0.9358342885971069,
      "learning_rate": 9.944323206850472e-06,
      "loss": 0.8533,
      "step": 1823
    },
    {
      "epoch": 0.10039077549672519,
      "grad_norm": 0.8327087163925171,
      "learning_rate": 9.94425868074623e-06,
      "loss": 0.8359,
      "step": 1824
    },
    {
      "epoch": 0.10044581429908085,
      "grad_norm": 1.0590367317199707,
      "learning_rate": 9.944194117482263e-06,
      "loss": 0.9659,
      "step": 1825
    },
    {
      "epoch": 0.10050085310143651,
      "grad_norm": 0.8739829063415527,
      "learning_rate": 9.944129517059055e-06,
      "loss": 0.7868,
      "step": 1826
    },
    {
      "epoch": 0.10055589190379217,
      "grad_norm": 0.8465235233306885,
      "learning_rate": 9.944064879477093e-06,
      "loss": 0.8554,
      "step": 1827
    },
    {
      "epoch": 0.10061093070614784,
      "grad_norm": 0.9068321585655212,
      "learning_rate": 9.944000204736864e-06,
      "loss": 0.8648,
      "step": 1828
    },
    {
      "epoch": 0.1006659695085035,
      "grad_norm": 0.8308066725730896,
      "learning_rate": 9.943935492838853e-06,
      "loss": 0.8471,
      "step": 1829
    },
    {
      "epoch": 0.10072100831085916,
      "grad_norm": 0.9973901510238647,
      "learning_rate": 9.943870743783545e-06,
      "loss": 0.9398,
      "step": 1830
    },
    {
      "epoch": 0.10077604711321482,
      "grad_norm": 0.8532593250274658,
      "learning_rate": 9.94380595757143e-06,
      "loss": 0.9001,
      "step": 1831
    },
    {
      "epoch": 0.10083108591557048,
      "grad_norm": 0.8571139574050903,
      "learning_rate": 9.94374113420299e-06,
      "loss": 0.85,
      "step": 1832
    },
    {
      "epoch": 0.10088612471792614,
      "grad_norm": 0.905624508857727,
      "learning_rate": 9.943676273678717e-06,
      "loss": 0.9587,
      "step": 1833
    },
    {
      "epoch": 0.1009411635202818,
      "grad_norm": 1.0224663019180298,
      "learning_rate": 9.943611375999097e-06,
      "loss": 0.8236,
      "step": 1834
    },
    {
      "epoch": 0.10099620232263747,
      "grad_norm": 0.8900588154792786,
      "learning_rate": 9.943546441164615e-06,
      "loss": 0.877,
      "step": 1835
    },
    {
      "epoch": 0.10105124112499313,
      "grad_norm": 0.8852938413619995,
      "learning_rate": 9.943481469175765e-06,
      "loss": 0.9521,
      "step": 1836
    },
    {
      "epoch": 0.10110627992734877,
      "grad_norm": 0.9249371290206909,
      "learning_rate": 9.943416460033027e-06,
      "loss": 0.8541,
      "step": 1837
    },
    {
      "epoch": 0.10116131872970444,
      "grad_norm": 0.8533583283424377,
      "learning_rate": 9.943351413736897e-06,
      "loss": 0.8571,
      "step": 1838
    },
    {
      "epoch": 0.1012163575320601,
      "grad_norm": 0.743800699710846,
      "learning_rate": 9.94328633028786e-06,
      "loss": 0.749,
      "step": 1839
    },
    {
      "epoch": 0.10127139633441576,
      "grad_norm": 0.7836641669273376,
      "learning_rate": 9.943221209686407e-06,
      "loss": 0.8237,
      "step": 1840
    },
    {
      "epoch": 0.10132643513677142,
      "grad_norm": 0.800782322883606,
      "learning_rate": 9.943156051933024e-06,
      "loss": 0.8323,
      "step": 1841
    },
    {
      "epoch": 0.10138147393912708,
      "grad_norm": 0.7531478404998779,
      "learning_rate": 9.943090857028206e-06,
      "loss": 0.8041,
      "step": 1842
    },
    {
      "epoch": 0.10143651274148274,
      "grad_norm": 0.9837996959686279,
      "learning_rate": 9.94302562497244e-06,
      "loss": 0.8084,
      "step": 1843
    },
    {
      "epoch": 0.1014915515438384,
      "grad_norm": 0.8038331866264343,
      "learning_rate": 9.942960355766216e-06,
      "loss": 0.8454,
      "step": 1844
    },
    {
      "epoch": 0.10154659034619407,
      "grad_norm": 0.7822145819664001,
      "learning_rate": 9.942895049410024e-06,
      "loss": 0.8137,
      "step": 1845
    },
    {
      "epoch": 0.10160162914854973,
      "grad_norm": 0.8222663998603821,
      "learning_rate": 9.942829705904358e-06,
      "loss": 0.8981,
      "step": 1846
    },
    {
      "epoch": 0.10165666795090539,
      "grad_norm": 1.0095717906951904,
      "learning_rate": 9.942764325249707e-06,
      "loss": 0.9159,
      "step": 1847
    },
    {
      "epoch": 0.10171170675326105,
      "grad_norm": 0.8264054656028748,
      "learning_rate": 9.942698907446561e-06,
      "loss": 0.9233,
      "step": 1848
    },
    {
      "epoch": 0.10176674555561671,
      "grad_norm": 0.8244288563728333,
      "learning_rate": 9.942633452495414e-06,
      "loss": 0.8507,
      "step": 1849
    },
    {
      "epoch": 0.10182178435797237,
      "grad_norm": 0.8457715511322021,
      "learning_rate": 9.942567960396755e-06,
      "loss": 0.7897,
      "step": 1850
    },
    {
      "epoch": 0.10187682316032803,
      "grad_norm": 0.8356698155403137,
      "learning_rate": 9.94250243115108e-06,
      "loss": 0.7927,
      "step": 1851
    },
    {
      "epoch": 0.1019318619626837,
      "grad_norm": 0.8251230716705322,
      "learning_rate": 9.94243686475888e-06,
      "loss": 0.8977,
      "step": 1852
    },
    {
      "epoch": 0.10198690076503936,
      "grad_norm": 0.8370125889778137,
      "learning_rate": 9.942371261220647e-06,
      "loss": 0.8204,
      "step": 1853
    },
    {
      "epoch": 0.10204193956739502,
      "grad_norm": 1.6722066402435303,
      "learning_rate": 9.942305620536876e-06,
      "loss": 0.9284,
      "step": 1854
    },
    {
      "epoch": 0.10209697836975068,
      "grad_norm": 0.8424906730651855,
      "learning_rate": 9.942239942708057e-06,
      "loss": 0.833,
      "step": 1855
    },
    {
      "epoch": 0.10215201717210634,
      "grad_norm": 0.7475115656852722,
      "learning_rate": 9.942174227734686e-06,
      "loss": 0.6158,
      "step": 1856
    },
    {
      "epoch": 0.102207055974462,
      "grad_norm": 0.8652095198631287,
      "learning_rate": 9.942108475617256e-06,
      "loss": 0.8781,
      "step": 1857
    },
    {
      "epoch": 0.10226209477681765,
      "grad_norm": 1.0621691942214966,
      "learning_rate": 9.942042686356263e-06,
      "loss": 1.0276,
      "step": 1858
    },
    {
      "epoch": 0.10231713357917331,
      "grad_norm": 1.113357424736023,
      "learning_rate": 9.941976859952199e-06,
      "loss": 0.8799,
      "step": 1859
    },
    {
      "epoch": 0.10237217238152897,
      "grad_norm": 0.9153568148612976,
      "learning_rate": 9.94191099640556e-06,
      "loss": 0.7988,
      "step": 1860
    },
    {
      "epoch": 0.10242721118388463,
      "grad_norm": 0.9217341542243958,
      "learning_rate": 9.941845095716842e-06,
      "loss": 0.7785,
      "step": 1861
    },
    {
      "epoch": 0.1024822499862403,
      "grad_norm": 0.8702190518379211,
      "learning_rate": 9.941779157886538e-06,
      "loss": 0.7648,
      "step": 1862
    },
    {
      "epoch": 0.10253728878859596,
      "grad_norm": 0.8609822988510132,
      "learning_rate": 9.941713182915144e-06,
      "loss": 0.9095,
      "step": 1863
    },
    {
      "epoch": 0.10259232759095162,
      "grad_norm": 0.7766719460487366,
      "learning_rate": 9.941647170803157e-06,
      "loss": 0.6984,
      "step": 1864
    },
    {
      "epoch": 0.10264736639330728,
      "grad_norm": 0.8497375249862671,
      "learning_rate": 9.941581121551074e-06,
      "loss": 0.9161,
      "step": 1865
    },
    {
      "epoch": 0.10270240519566294,
      "grad_norm": 0.8007600903511047,
      "learning_rate": 9.941515035159388e-06,
      "loss": 0.8099,
      "step": 1866
    },
    {
      "epoch": 0.1027574439980186,
      "grad_norm": 0.7932959794998169,
      "learning_rate": 9.941448911628599e-06,
      "loss": 0.8049,
      "step": 1867
    },
    {
      "epoch": 0.10281248280037426,
      "grad_norm": 1.3169244527816772,
      "learning_rate": 9.941382750959203e-06,
      "loss": 0.8601,
      "step": 1868
    },
    {
      "epoch": 0.10286752160272992,
      "grad_norm": 0.8011140823364258,
      "learning_rate": 9.941316553151696e-06,
      "loss": 0.8397,
      "step": 1869
    },
    {
      "epoch": 0.10292256040508559,
      "grad_norm": 0.811210572719574,
      "learning_rate": 9.941250318206577e-06,
      "loss": 0.7863,
      "step": 1870
    },
    {
      "epoch": 0.10297759920744125,
      "grad_norm": 0.8172751665115356,
      "learning_rate": 9.941184046124342e-06,
      "loss": 0.8114,
      "step": 1871
    },
    {
      "epoch": 0.10303263800979691,
      "grad_norm": 0.8072887063026428,
      "learning_rate": 9.941117736905493e-06,
      "loss": 0.8928,
      "step": 1872
    },
    {
      "epoch": 0.10308767681215257,
      "grad_norm": 0.9111380577087402,
      "learning_rate": 9.941051390550524e-06,
      "loss": 0.866,
      "step": 1873
    },
    {
      "epoch": 0.10314271561450823,
      "grad_norm": 0.8158383369445801,
      "learning_rate": 9.940985007059936e-06,
      "loss": 0.7805,
      "step": 1874
    },
    {
      "epoch": 0.1031977544168639,
      "grad_norm": 0.8858961462974548,
      "learning_rate": 9.940918586434226e-06,
      "loss": 0.8424,
      "step": 1875
    },
    {
      "epoch": 0.10325279321921955,
      "grad_norm": 0.8835182189941406,
      "learning_rate": 9.940852128673895e-06,
      "loss": 0.7816,
      "step": 1876
    },
    {
      "epoch": 0.10330783202157522,
      "grad_norm": 1.044227123260498,
      "learning_rate": 9.940785633779444e-06,
      "loss": 0.8952,
      "step": 1877
    },
    {
      "epoch": 0.10336287082393088,
      "grad_norm": 0.8255050778388977,
      "learning_rate": 9.940719101751367e-06,
      "loss": 0.8215,
      "step": 1878
    },
    {
      "epoch": 0.10341790962628654,
      "grad_norm": 0.8561689257621765,
      "learning_rate": 9.940652532590172e-06,
      "loss": 0.9686,
      "step": 1879
    },
    {
      "epoch": 0.10347294842864219,
      "grad_norm": 0.8798959255218506,
      "learning_rate": 9.94058592629635e-06,
      "loss": 0.8993,
      "step": 1880
    },
    {
      "epoch": 0.10352798723099785,
      "grad_norm": 0.9292098879814148,
      "learning_rate": 9.940519282870411e-06,
      "loss": 0.8536,
      "step": 1881
    },
    {
      "epoch": 0.10358302603335351,
      "grad_norm": 0.8865400552749634,
      "learning_rate": 9.940452602312851e-06,
      "loss": 0.8024,
      "step": 1882
    },
    {
      "epoch": 0.10363806483570917,
      "grad_norm": 0.8985510468482971,
      "learning_rate": 9.94038588462417e-06,
      "loss": 0.7748,
      "step": 1883
    },
    {
      "epoch": 0.10369310363806483,
      "grad_norm": 0.9973617196083069,
      "learning_rate": 9.940319129804872e-06,
      "loss": 0.875,
      "step": 1884
    },
    {
      "epoch": 0.1037481424404205,
      "grad_norm": 0.8615350723266602,
      "learning_rate": 9.940252337855458e-06,
      "loss": 0.904,
      "step": 1885
    },
    {
      "epoch": 0.10380318124277615,
      "grad_norm": 0.8752412796020508,
      "learning_rate": 9.940185508776429e-06,
      "loss": 0.8735,
      "step": 1886
    },
    {
      "epoch": 0.10385822004513182,
      "grad_norm": 0.8639446496963501,
      "learning_rate": 9.94011864256829e-06,
      "loss": 0.7952,
      "step": 1887
    },
    {
      "epoch": 0.10391325884748748,
      "grad_norm": 0.7932116389274597,
      "learning_rate": 9.94005173923154e-06,
      "loss": 0.8721,
      "step": 1888
    },
    {
      "epoch": 0.10396829764984314,
      "grad_norm": 0.8573791980743408,
      "learning_rate": 9.939984798766685e-06,
      "loss": 0.9271,
      "step": 1889
    },
    {
      "epoch": 0.1040233364521988,
      "grad_norm": 0.9080122113227844,
      "learning_rate": 9.939917821174225e-06,
      "loss": 0.8991,
      "step": 1890
    },
    {
      "epoch": 0.10407837525455446,
      "grad_norm": 0.7883808612823486,
      "learning_rate": 9.939850806454664e-06,
      "loss": 0.6895,
      "step": 1891
    },
    {
      "epoch": 0.10413341405691012,
      "grad_norm": 0.8067768216133118,
      "learning_rate": 9.93978375460851e-06,
      "loss": 0.835,
      "step": 1892
    },
    {
      "epoch": 0.10418845285926578,
      "grad_norm": 0.8756459951400757,
      "learning_rate": 9.939716665636262e-06,
      "loss": 0.8144,
      "step": 1893
    },
    {
      "epoch": 0.10424349166162145,
      "grad_norm": 0.8056700825691223,
      "learning_rate": 9.939649539538425e-06,
      "loss": 0.7454,
      "step": 1894
    },
    {
      "epoch": 0.10429853046397711,
      "grad_norm": 1.0756300687789917,
      "learning_rate": 9.939582376315505e-06,
      "loss": 0.8096,
      "step": 1895
    },
    {
      "epoch": 0.10435356926633277,
      "grad_norm": 0.8938102126121521,
      "learning_rate": 9.939515175968006e-06,
      "loss": 0.7496,
      "step": 1896
    },
    {
      "epoch": 0.10440860806868843,
      "grad_norm": 0.9371656775474548,
      "learning_rate": 9.939447938496434e-06,
      "loss": 0.9817,
      "step": 1897
    },
    {
      "epoch": 0.10446364687104409,
      "grad_norm": 1.0216082334518433,
      "learning_rate": 9.939380663901292e-06,
      "loss": 0.8804,
      "step": 1898
    },
    {
      "epoch": 0.10451868567339975,
      "grad_norm": 0.8791126012802124,
      "learning_rate": 9.939313352183088e-06,
      "loss": 0.7811,
      "step": 1899
    },
    {
      "epoch": 0.10457372447575541,
      "grad_norm": 0.9925445914268494,
      "learning_rate": 9.939246003342326e-06,
      "loss": 0.8892,
      "step": 1900
    },
    {
      "epoch": 0.10462876327811106,
      "grad_norm": 1.0459916591644287,
      "learning_rate": 9.939178617379514e-06,
      "loss": 0.7938,
      "step": 1901
    },
    {
      "epoch": 0.10468380208046672,
      "grad_norm": 0.9103816747665405,
      "learning_rate": 9.93911119429516e-06,
      "loss": 0.8282,
      "step": 1902
    },
    {
      "epoch": 0.10473884088282238,
      "grad_norm": 0.9602296352386475,
      "learning_rate": 9.939043734089764e-06,
      "loss": 0.919,
      "step": 1903
    },
    {
      "epoch": 0.10479387968517805,
      "grad_norm": 0.9529246687889099,
      "learning_rate": 9.93897623676384e-06,
      "loss": 0.9469,
      "step": 1904
    },
    {
      "epoch": 0.10484891848753371,
      "grad_norm": 0.9619705080986023,
      "learning_rate": 9.938908702317893e-06,
      "loss": 0.9371,
      "step": 1905
    },
    {
      "epoch": 0.10490395728988937,
      "grad_norm": 1.0106935501098633,
      "learning_rate": 9.938841130752428e-06,
      "loss": 0.7502,
      "step": 1906
    },
    {
      "epoch": 0.10495899609224503,
      "grad_norm": 0.913985013961792,
      "learning_rate": 9.938773522067957e-06,
      "loss": 0.8172,
      "step": 1907
    },
    {
      "epoch": 0.10501403489460069,
      "grad_norm": 0.9474983215332031,
      "learning_rate": 9.938705876264985e-06,
      "loss": 0.8999,
      "step": 1908
    },
    {
      "epoch": 0.10506907369695635,
      "grad_norm": 0.9185097813606262,
      "learning_rate": 9.938638193344024e-06,
      "loss": 0.8976,
      "step": 1909
    },
    {
      "epoch": 0.10512411249931201,
      "grad_norm": 0.7633675932884216,
      "learning_rate": 9.938570473305578e-06,
      "loss": 0.7777,
      "step": 1910
    },
    {
      "epoch": 0.10517915130166768,
      "grad_norm": 0.9547691345214844,
      "learning_rate": 9.938502716150159e-06,
      "loss": 0.8154,
      "step": 1911
    },
    {
      "epoch": 0.10523419010402334,
      "grad_norm": 0.8556191921234131,
      "learning_rate": 9.938434921878275e-06,
      "loss": 0.828,
      "step": 1912
    },
    {
      "epoch": 0.105289228906379,
      "grad_norm": 0.9826140999794006,
      "learning_rate": 9.938367090490437e-06,
      "loss": 0.8085,
      "step": 1913
    },
    {
      "epoch": 0.10534426770873466,
      "grad_norm": 0.8610432744026184,
      "learning_rate": 9.938299221987154e-06,
      "loss": 0.9103,
      "step": 1914
    },
    {
      "epoch": 0.10539930651109032,
      "grad_norm": 0.8383543491363525,
      "learning_rate": 9.938231316368934e-06,
      "loss": 0.8182,
      "step": 1915
    },
    {
      "epoch": 0.10545434531344598,
      "grad_norm": 0.8552964925765991,
      "learning_rate": 9.93816337363629e-06,
      "loss": 0.8024,
      "step": 1916
    },
    {
      "epoch": 0.10550938411580164,
      "grad_norm": 0.9255730509757996,
      "learning_rate": 9.938095393789732e-06,
      "loss": 0.8566,
      "step": 1917
    },
    {
      "epoch": 0.1055644229181573,
      "grad_norm": 0.9882987141609192,
      "learning_rate": 9.938027376829774e-06,
      "loss": 0.7119,
      "step": 1918
    },
    {
      "epoch": 0.10561946172051297,
      "grad_norm": 1.139404535293579,
      "learning_rate": 9.93795932275692e-06,
      "loss": 0.8839,
      "step": 1919
    },
    {
      "epoch": 0.10567450052286863,
      "grad_norm": 1.004782795906067,
      "learning_rate": 9.937891231571686e-06,
      "loss": 0.904,
      "step": 1920
    },
    {
      "epoch": 0.10572953932522429,
      "grad_norm": 0.8437260389328003,
      "learning_rate": 9.937823103274585e-06,
      "loss": 0.7942,
      "step": 1921
    },
    {
      "epoch": 0.10578457812757995,
      "grad_norm": 1.1388722658157349,
      "learning_rate": 9.937754937866127e-06,
      "loss": 0.9491,
      "step": 1922
    },
    {
      "epoch": 0.1058396169299356,
      "grad_norm": 0.9266740083694458,
      "learning_rate": 9.937686735346823e-06,
      "loss": 0.9067,
      "step": 1923
    },
    {
      "epoch": 0.10589465573229126,
      "grad_norm": 0.7536123991012573,
      "learning_rate": 9.93761849571719e-06,
      "loss": 0.6533,
      "step": 1924
    },
    {
      "epoch": 0.10594969453464692,
      "grad_norm": 0.8781737089157104,
      "learning_rate": 9.937550218977737e-06,
      "loss": 0.8319,
      "step": 1925
    },
    {
      "epoch": 0.10600473333700258,
      "grad_norm": 0.8577924966812134,
      "learning_rate": 9.937481905128976e-06,
      "loss": 0.8604,
      "step": 1926
    },
    {
      "epoch": 0.10605977213935824,
      "grad_norm": 0.8351713418960571,
      "learning_rate": 9.937413554171424e-06,
      "loss": 0.946,
      "step": 1927
    },
    {
      "epoch": 0.1061148109417139,
      "grad_norm": 0.971491813659668,
      "learning_rate": 9.937345166105594e-06,
      "loss": 0.7383,
      "step": 1928
    },
    {
      "epoch": 0.10616984974406957,
      "grad_norm": 0.8020079731941223,
      "learning_rate": 9.937276740932001e-06,
      "loss": 0.7468,
      "step": 1929
    },
    {
      "epoch": 0.10622488854642523,
      "grad_norm": 0.9057347178459167,
      "learning_rate": 9.937208278651153e-06,
      "loss": 0.8223,
      "step": 1930
    },
    {
      "epoch": 0.10627992734878089,
      "grad_norm": 0.8384734392166138,
      "learning_rate": 9.937139779263574e-06,
      "loss": 0.8773,
      "step": 1931
    },
    {
      "epoch": 0.10633496615113655,
      "grad_norm": 0.8732065558433533,
      "learning_rate": 9.93707124276977e-06,
      "loss": 0.8265,
      "step": 1932
    },
    {
      "epoch": 0.10639000495349221,
      "grad_norm": 0.8744868040084839,
      "learning_rate": 9.937002669170264e-06,
      "loss": 0.8497,
      "step": 1933
    },
    {
      "epoch": 0.10644504375584787,
      "grad_norm": 0.8589879870414734,
      "learning_rate": 9.936934058465564e-06,
      "loss": 0.8116,
      "step": 1934
    },
    {
      "epoch": 0.10650008255820353,
      "grad_norm": 0.8614563941955566,
      "learning_rate": 9.936865410656192e-06,
      "loss": 0.7823,
      "step": 1935
    },
    {
      "epoch": 0.1065551213605592,
      "grad_norm": 0.8381434082984924,
      "learning_rate": 9.93679672574266e-06,
      "loss": 0.7889,
      "step": 1936
    },
    {
      "epoch": 0.10661016016291486,
      "grad_norm": 0.9834293127059937,
      "learning_rate": 9.936728003725484e-06,
      "loss": 0.8358,
      "step": 1937
    },
    {
      "epoch": 0.10666519896527052,
      "grad_norm": 0.8461851477622986,
      "learning_rate": 9.936659244605184e-06,
      "loss": 0.8408,
      "step": 1938
    },
    {
      "epoch": 0.10672023776762618,
      "grad_norm": 1.0186371803283691,
      "learning_rate": 9.936590448382273e-06,
      "loss": 0.8118,
      "step": 1939
    },
    {
      "epoch": 0.10677527656998184,
      "grad_norm": 0.866321325302124,
      "learning_rate": 9.93652161505727e-06,
      "loss": 0.8696,
      "step": 1940
    },
    {
      "epoch": 0.1068303153723375,
      "grad_norm": 0.9179622530937195,
      "learning_rate": 9.936452744630692e-06,
      "loss": 0.8419,
      "step": 1941
    },
    {
      "epoch": 0.10688535417469316,
      "grad_norm": 0.8250496983528137,
      "learning_rate": 9.936383837103057e-06,
      "loss": 0.8511,
      "step": 1942
    },
    {
      "epoch": 0.10694039297704883,
      "grad_norm": 0.8475700616836548,
      "learning_rate": 9.936314892474883e-06,
      "loss": 0.8404,
      "step": 1943
    },
    {
      "epoch": 0.10699543177940447,
      "grad_norm": 0.774334192276001,
      "learning_rate": 9.936245910746684e-06,
      "loss": 0.7461,
      "step": 1944
    },
    {
      "epoch": 0.10705047058176013,
      "grad_norm": 0.9313948154449463,
      "learning_rate": 9.936176891918986e-06,
      "loss": 0.8486,
      "step": 1945
    },
    {
      "epoch": 0.1071055093841158,
      "grad_norm": 0.8784124255180359,
      "learning_rate": 9.936107835992304e-06,
      "loss": 0.84,
      "step": 1946
    },
    {
      "epoch": 0.10716054818647146,
      "grad_norm": 0.9087465405464172,
      "learning_rate": 9.936038742967154e-06,
      "loss": 0.9012,
      "step": 1947
    },
    {
      "epoch": 0.10721558698882712,
      "grad_norm": 0.8462012410163879,
      "learning_rate": 9.93596961284406e-06,
      "loss": 0.9193,
      "step": 1948
    },
    {
      "epoch": 0.10727062579118278,
      "grad_norm": 0.8984553813934326,
      "learning_rate": 9.935900445623538e-06,
      "loss": 0.781,
      "step": 1949
    },
    {
      "epoch": 0.10732566459353844,
      "grad_norm": 0.9197295308113098,
      "learning_rate": 9.935831241306111e-06,
      "loss": 0.8861,
      "step": 1950
    },
    {
      "epoch": 0.1073807033958941,
      "grad_norm": 0.8452801704406738,
      "learning_rate": 9.935761999892296e-06,
      "loss": 0.8649,
      "step": 1951
    },
    {
      "epoch": 0.10743574219824976,
      "grad_norm": 0.8047192096710205,
      "learning_rate": 9.935692721382618e-06,
      "loss": 0.8704,
      "step": 1952
    },
    {
      "epoch": 0.10749078100060543,
      "grad_norm": 0.9536359906196594,
      "learning_rate": 9.935623405777593e-06,
      "loss": 0.7803,
      "step": 1953
    },
    {
      "epoch": 0.10754581980296109,
      "grad_norm": 0.8215291500091553,
      "learning_rate": 9.935554053077744e-06,
      "loss": 0.8247,
      "step": 1954
    },
    {
      "epoch": 0.10760085860531675,
      "grad_norm": 0.9261930584907532,
      "learning_rate": 9.93548466328359e-06,
      "loss": 0.8594,
      "step": 1955
    },
    {
      "epoch": 0.10765589740767241,
      "grad_norm": 0.7973492741584778,
      "learning_rate": 9.935415236395656e-06,
      "loss": 0.7464,
      "step": 1956
    },
    {
      "epoch": 0.10771093621002807,
      "grad_norm": 0.9328988790512085,
      "learning_rate": 9.935345772414463e-06,
      "loss": 0.8472,
      "step": 1957
    },
    {
      "epoch": 0.10776597501238373,
      "grad_norm": 0.9490759968757629,
      "learning_rate": 9.935276271340532e-06,
      "loss": 0.806,
      "step": 1958
    },
    {
      "epoch": 0.1078210138147394,
      "grad_norm": 0.9149925112724304,
      "learning_rate": 9.935206733174385e-06,
      "loss": 0.8741,
      "step": 1959
    },
    {
      "epoch": 0.10787605261709506,
      "grad_norm": 1.0074039697647095,
      "learning_rate": 9.935137157916546e-06,
      "loss": 0.8493,
      "step": 1960
    },
    {
      "epoch": 0.10793109141945072,
      "grad_norm": 0.8783678412437439,
      "learning_rate": 9.935067545567535e-06,
      "loss": 0.8132,
      "step": 1961
    },
    {
      "epoch": 0.10798613022180638,
      "grad_norm": 0.8273885250091553,
      "learning_rate": 9.934997896127879e-06,
      "loss": 0.7448,
      "step": 1962
    },
    {
      "epoch": 0.10804116902416204,
      "grad_norm": 0.761947512626648,
      "learning_rate": 9.9349282095981e-06,
      "loss": 0.7933,
      "step": 1963
    },
    {
      "epoch": 0.1080962078265177,
      "grad_norm": 0.814809262752533,
      "learning_rate": 9.934858485978722e-06,
      "loss": 0.7551,
      "step": 1964
    },
    {
      "epoch": 0.10815124662887336,
      "grad_norm": 0.8108895421028137,
      "learning_rate": 9.934788725270266e-06,
      "loss": 0.6787,
      "step": 1965
    },
    {
      "epoch": 0.10820628543122901,
      "grad_norm": 0.8669139742851257,
      "learning_rate": 9.934718927473262e-06,
      "loss": 0.8395,
      "step": 1966
    },
    {
      "epoch": 0.10826132423358467,
      "grad_norm": 0.9093756079673767,
      "learning_rate": 9.93464909258823e-06,
      "loss": 0.8341,
      "step": 1967
    },
    {
      "epoch": 0.10831636303594033,
      "grad_norm": 0.8923841714859009,
      "learning_rate": 9.934579220615697e-06,
      "loss": 0.9422,
      "step": 1968
    },
    {
      "epoch": 0.108371401838296,
      "grad_norm": 0.850429117679596,
      "learning_rate": 9.934509311556186e-06,
      "loss": 0.8446,
      "step": 1969
    },
    {
      "epoch": 0.10842644064065166,
      "grad_norm": 0.8762460350990295,
      "learning_rate": 9.934439365410224e-06,
      "loss": 0.7788,
      "step": 1970
    },
    {
      "epoch": 0.10848147944300732,
      "grad_norm": 0.9700387716293335,
      "learning_rate": 9.934369382178338e-06,
      "loss": 0.8455,
      "step": 1971
    },
    {
      "epoch": 0.10853651824536298,
      "grad_norm": 0.8003185987472534,
      "learning_rate": 9.934299361861053e-06,
      "loss": 0.8026,
      "step": 1972
    },
    {
      "epoch": 0.10859155704771864,
      "grad_norm": 0.9626984596252441,
      "learning_rate": 9.934229304458893e-06,
      "loss": 0.8219,
      "step": 1973
    },
    {
      "epoch": 0.1086465958500743,
      "grad_norm": 0.8722280859947205,
      "learning_rate": 9.934159209972386e-06,
      "loss": 0.8866,
      "step": 1974
    },
    {
      "epoch": 0.10870163465242996,
      "grad_norm": 0.838736355304718,
      "learning_rate": 9.934089078402061e-06,
      "loss": 0.7723,
      "step": 1975
    },
    {
      "epoch": 0.10875667345478562,
      "grad_norm": 0.8373032808303833,
      "learning_rate": 9.934018909748443e-06,
      "loss": 0.9003,
      "step": 1976
    },
    {
      "epoch": 0.10881171225714129,
      "grad_norm": 0.8704653978347778,
      "learning_rate": 9.93394870401206e-06,
      "loss": 0.8926,
      "step": 1977
    },
    {
      "epoch": 0.10886675105949695,
      "grad_norm": 0.8088163733482361,
      "learning_rate": 9.933878461193437e-06,
      "loss": 0.8059,
      "step": 1978
    },
    {
      "epoch": 0.10892178986185261,
      "grad_norm": 0.856421947479248,
      "learning_rate": 9.933808181293108e-06,
      "loss": 0.8447,
      "step": 1979
    },
    {
      "epoch": 0.10897682866420827,
      "grad_norm": 0.9676237106323242,
      "learning_rate": 9.933737864311595e-06,
      "loss": 0.9009,
      "step": 1980
    },
    {
      "epoch": 0.10903186746656393,
      "grad_norm": 0.7955103516578674,
      "learning_rate": 9.933667510249428e-06,
      "loss": 0.881,
      "step": 1981
    },
    {
      "epoch": 0.10908690626891959,
      "grad_norm": 0.7935854196548462,
      "learning_rate": 9.933597119107136e-06,
      "loss": 0.8773,
      "step": 1982
    },
    {
      "epoch": 0.10914194507127525,
      "grad_norm": 0.7726008296012878,
      "learning_rate": 9.933526690885251e-06,
      "loss": 0.8133,
      "step": 1983
    },
    {
      "epoch": 0.10919698387363092,
      "grad_norm": 0.8577712178230286,
      "learning_rate": 9.9334562255843e-06,
      "loss": 0.7455,
      "step": 1984
    },
    {
      "epoch": 0.10925202267598658,
      "grad_norm": 0.9996447563171387,
      "learning_rate": 9.933385723204812e-06,
      "loss": 0.7312,
      "step": 1985
    },
    {
      "epoch": 0.10930706147834224,
      "grad_norm": 0.9600629806518555,
      "learning_rate": 9.933315183747318e-06,
      "loss": 0.8792,
      "step": 1986
    },
    {
      "epoch": 0.10936210028069789,
      "grad_norm": 0.9126206636428833,
      "learning_rate": 9.933244607212347e-06,
      "loss": 1.0023,
      "step": 1987
    },
    {
      "epoch": 0.10941713908305355,
      "grad_norm": 0.774153470993042,
      "learning_rate": 9.93317399360043e-06,
      "loss": 0.7877,
      "step": 1988
    },
    {
      "epoch": 0.10947217788540921,
      "grad_norm": 0.848495364189148,
      "learning_rate": 9.933103342912096e-06,
      "loss": 0.8825,
      "step": 1989
    },
    {
      "epoch": 0.10952721668776487,
      "grad_norm": 0.806408166885376,
      "learning_rate": 9.933032655147881e-06,
      "loss": 0.7389,
      "step": 1990
    },
    {
      "epoch": 0.10958225549012053,
      "grad_norm": 0.8579222559928894,
      "learning_rate": 9.932961930308312e-06,
      "loss": 0.8283,
      "step": 1991
    },
    {
      "epoch": 0.10963729429247619,
      "grad_norm": 0.7548109292984009,
      "learning_rate": 9.93289116839392e-06,
      "loss": 0.7971,
      "step": 1992
    },
    {
      "epoch": 0.10969233309483185,
      "grad_norm": 0.7954711318016052,
      "learning_rate": 9.93282036940524e-06,
      "loss": 0.849,
      "step": 1993
    },
    {
      "epoch": 0.10974737189718752,
      "grad_norm": 0.7911425232887268,
      "learning_rate": 9.932749533342802e-06,
      "loss": 0.86,
      "step": 1994
    },
    {
      "epoch": 0.10980241069954318,
      "grad_norm": 0.8505094051361084,
      "learning_rate": 9.932678660207141e-06,
      "loss": 0.7871,
      "step": 1995
    },
    {
      "epoch": 0.10985744950189884,
      "grad_norm": 0.809612512588501,
      "learning_rate": 9.932607749998784e-06,
      "loss": 0.8337,
      "step": 1996
    },
    {
      "epoch": 0.1099124883042545,
      "grad_norm": 0.738523006439209,
      "learning_rate": 9.93253680271827e-06,
      "loss": 0.7634,
      "step": 1997
    },
    {
      "epoch": 0.10996752710661016,
      "grad_norm": 0.8434372544288635,
      "learning_rate": 9.932465818366128e-06,
      "loss": 0.7987,
      "step": 1998
    },
    {
      "epoch": 0.11002256590896582,
      "grad_norm": 0.8068081140518188,
      "learning_rate": 9.932394796942895e-06,
      "loss": 0.9496,
      "step": 1999
    },
    {
      "epoch": 0.11007760471132148,
      "grad_norm": 0.754342794418335,
      "learning_rate": 9.932323738449103e-06,
      "loss": 0.7355,
      "step": 2000
    },
    {
      "epoch": 0.11013264351367714,
      "grad_norm": 0.8830806612968445,
      "learning_rate": 9.932252642885285e-06,
      "loss": 0.8458,
      "step": 2001
    },
    {
      "epoch": 0.1101876823160328,
      "grad_norm": 0.9915485978126526,
      "learning_rate": 9.932181510251977e-06,
      "loss": 0.8116,
      "step": 2002
    },
    {
      "epoch": 0.11024272111838847,
      "grad_norm": 0.858368992805481,
      "learning_rate": 9.932110340549712e-06,
      "loss": 0.8354,
      "step": 2003
    },
    {
      "epoch": 0.11029775992074413,
      "grad_norm": 0.8591521382331848,
      "learning_rate": 9.932039133779028e-06,
      "loss": 0.8316,
      "step": 2004
    },
    {
      "epoch": 0.11035279872309979,
      "grad_norm": 0.8714838624000549,
      "learning_rate": 9.931967889940455e-06,
      "loss": 0.8106,
      "step": 2005
    },
    {
      "epoch": 0.11040783752545545,
      "grad_norm": 0.8082797527313232,
      "learning_rate": 9.931896609034534e-06,
      "loss": 0.7762,
      "step": 2006
    },
    {
      "epoch": 0.11046287632781111,
      "grad_norm": 0.9226199984550476,
      "learning_rate": 9.931825291061797e-06,
      "loss": 0.8641,
      "step": 2007
    },
    {
      "epoch": 0.11051791513016677,
      "grad_norm": 0.8883050680160522,
      "learning_rate": 9.931753936022783e-06,
      "loss": 0.9014,
      "step": 2008
    },
    {
      "epoch": 0.11057295393252242,
      "grad_norm": 0.9024807810783386,
      "learning_rate": 9.931682543918024e-06,
      "loss": 0.9085,
      "step": 2009
    },
    {
      "epoch": 0.11062799273487808,
      "grad_norm": 0.8381460905075073,
      "learning_rate": 9.931611114748062e-06,
      "loss": 0.8043,
      "step": 2010
    },
    {
      "epoch": 0.11068303153723374,
      "grad_norm": 1.1222339868545532,
      "learning_rate": 9.931539648513429e-06,
      "loss": 0.8388,
      "step": 2011
    },
    {
      "epoch": 0.1107380703395894,
      "grad_norm": 0.9710868000984192,
      "learning_rate": 9.931468145214665e-06,
      "loss": 0.8934,
      "step": 2012
    },
    {
      "epoch": 0.11079310914194507,
      "grad_norm": 0.9821141958236694,
      "learning_rate": 9.931396604852304e-06,
      "loss": 0.931,
      "step": 2013
    },
    {
      "epoch": 0.11084814794430073,
      "grad_norm": 1.0658717155456543,
      "learning_rate": 9.931325027426889e-06,
      "loss": 0.9032,
      "step": 2014
    },
    {
      "epoch": 0.11090318674665639,
      "grad_norm": 0.8836946487426758,
      "learning_rate": 9.931253412938956e-06,
      "loss": 0.9131,
      "step": 2015
    },
    {
      "epoch": 0.11095822554901205,
      "grad_norm": 0.8438361883163452,
      "learning_rate": 9.93118176138904e-06,
      "loss": 0.8674,
      "step": 2016
    },
    {
      "epoch": 0.11101326435136771,
      "grad_norm": 0.928142786026001,
      "learning_rate": 9.93111007277768e-06,
      "loss": 0.8882,
      "step": 2017
    },
    {
      "epoch": 0.11106830315372337,
      "grad_norm": 0.9176276922225952,
      "learning_rate": 9.93103834710542e-06,
      "loss": 0.8904,
      "step": 2018
    },
    {
      "epoch": 0.11112334195607904,
      "grad_norm": 1.0462889671325684,
      "learning_rate": 9.930966584372795e-06,
      "loss": 0.8029,
      "step": 2019
    },
    {
      "epoch": 0.1111783807584347,
      "grad_norm": 0.7627375721931458,
      "learning_rate": 9.930894784580344e-06,
      "loss": 0.8474,
      "step": 2020
    },
    {
      "epoch": 0.11123341956079036,
      "grad_norm": 1.0545588731765747,
      "learning_rate": 9.93082294772861e-06,
      "loss": 0.7985,
      "step": 2021
    },
    {
      "epoch": 0.11128845836314602,
      "grad_norm": 0.9752298593521118,
      "learning_rate": 9.93075107381813e-06,
      "loss": 0.8725,
      "step": 2022
    },
    {
      "epoch": 0.11134349716550168,
      "grad_norm": 0.8403159379959106,
      "learning_rate": 9.930679162849444e-06,
      "loss": 0.8854,
      "step": 2023
    },
    {
      "epoch": 0.11139853596785734,
      "grad_norm": 0.8879380226135254,
      "learning_rate": 9.930607214823094e-06,
      "loss": 0.7269,
      "step": 2024
    },
    {
      "epoch": 0.111453574770213,
      "grad_norm": 0.907256543636322,
      "learning_rate": 9.930535229739618e-06,
      "loss": 0.8145,
      "step": 2025
    },
    {
      "epoch": 0.11150861357256867,
      "grad_norm": 1.1066968441009521,
      "learning_rate": 9.93046320759956e-06,
      "loss": 0.9281,
      "step": 2026
    },
    {
      "epoch": 0.11156365237492433,
      "grad_norm": 0.9226258397102356,
      "learning_rate": 9.930391148403462e-06,
      "loss": 0.9048,
      "step": 2027
    },
    {
      "epoch": 0.11161869117727999,
      "grad_norm": 0.9652156829833984,
      "learning_rate": 9.930319052151862e-06,
      "loss": 0.9321,
      "step": 2028
    },
    {
      "epoch": 0.11167372997963565,
      "grad_norm": 0.9102638363838196,
      "learning_rate": 9.930246918845305e-06,
      "loss": 0.8169,
      "step": 2029
    },
    {
      "epoch": 0.1117287687819913,
      "grad_norm": 0.7765716314315796,
      "learning_rate": 9.93017474848433e-06,
      "loss": 0.7691,
      "step": 2030
    },
    {
      "epoch": 0.11178380758434696,
      "grad_norm": 0.9053775072097778,
      "learning_rate": 9.930102541069484e-06,
      "loss": 0.782,
      "step": 2031
    },
    {
      "epoch": 0.11183884638670262,
      "grad_norm": 0.8892827033996582,
      "learning_rate": 9.930030296601306e-06,
      "loss": 0.8575,
      "step": 2032
    },
    {
      "epoch": 0.11189388518905828,
      "grad_norm": 0.8947604894638062,
      "learning_rate": 9.929958015080339e-06,
      "loss": 0.8607,
      "step": 2033
    },
    {
      "epoch": 0.11194892399141394,
      "grad_norm": 0.8936871290206909,
      "learning_rate": 9.929885696507127e-06,
      "loss": 0.8111,
      "step": 2034
    },
    {
      "epoch": 0.1120039627937696,
      "grad_norm": 0.9579165577888489,
      "learning_rate": 9.929813340882214e-06,
      "loss": 0.911,
      "step": 2035
    },
    {
      "epoch": 0.11205900159612527,
      "grad_norm": 0.7885386347770691,
      "learning_rate": 9.929740948206146e-06,
      "loss": 0.8074,
      "step": 2036
    },
    {
      "epoch": 0.11211404039848093,
      "grad_norm": 0.817939281463623,
      "learning_rate": 9.929668518479462e-06,
      "loss": 0.8451,
      "step": 2037
    },
    {
      "epoch": 0.11216907920083659,
      "grad_norm": 0.8695761561393738,
      "learning_rate": 9.92959605170271e-06,
      "loss": 0.7158,
      "step": 2038
    },
    {
      "epoch": 0.11222411800319225,
      "grad_norm": 0.8569639325141907,
      "learning_rate": 9.929523547876433e-06,
      "loss": 0.8568,
      "step": 2039
    },
    {
      "epoch": 0.11227915680554791,
      "grad_norm": 0.8569897413253784,
      "learning_rate": 9.929451007001176e-06,
      "loss": 0.8971,
      "step": 2040
    },
    {
      "epoch": 0.11233419560790357,
      "grad_norm": 0.8520069718360901,
      "learning_rate": 9.929378429077487e-06,
      "loss": 0.9027,
      "step": 2041
    },
    {
      "epoch": 0.11238923441025923,
      "grad_norm": 0.9338961839675903,
      "learning_rate": 9.929305814105907e-06,
      "loss": 0.8646,
      "step": 2042
    },
    {
      "epoch": 0.1124442732126149,
      "grad_norm": 0.8497192859649658,
      "learning_rate": 9.929233162086985e-06,
      "loss": 0.9068,
      "step": 2043
    },
    {
      "epoch": 0.11249931201497056,
      "grad_norm": 0.8570863008499146,
      "learning_rate": 9.929160473021267e-06,
      "loss": 0.962,
      "step": 2044
    },
    {
      "epoch": 0.11255435081732622,
      "grad_norm": 0.9072359800338745,
      "learning_rate": 9.929087746909296e-06,
      "loss": 0.8454,
      "step": 2045
    },
    {
      "epoch": 0.11260938961968188,
      "grad_norm": 0.7920698523521423,
      "learning_rate": 9.929014983751623e-06,
      "loss": 0.8031,
      "step": 2046
    },
    {
      "epoch": 0.11266442842203754,
      "grad_norm": 1.0180169343948364,
      "learning_rate": 9.928942183548791e-06,
      "loss": 0.7759,
      "step": 2047
    },
    {
      "epoch": 0.1127194672243932,
      "grad_norm": 0.8746892809867859,
      "learning_rate": 9.928869346301351e-06,
      "loss": 0.9038,
      "step": 2048
    },
    {
      "epoch": 0.11277450602674886,
      "grad_norm": 0.8283438086509705,
      "learning_rate": 9.928796472009846e-06,
      "loss": 0.8883,
      "step": 2049
    },
    {
      "epoch": 0.11282954482910452,
      "grad_norm": 1.321917176246643,
      "learning_rate": 9.928723560674828e-06,
      "loss": 0.835,
      "step": 2050
    },
    {
      "epoch": 0.11288458363146017,
      "grad_norm": 0.9356202483177185,
      "learning_rate": 9.928650612296841e-06,
      "loss": 0.8077,
      "step": 2051
    },
    {
      "epoch": 0.11293962243381583,
      "grad_norm": 0.8493767380714417,
      "learning_rate": 9.928577626876439e-06,
      "loss": 0.8295,
      "step": 2052
    },
    {
      "epoch": 0.1129946612361715,
      "grad_norm": 0.784818708896637,
      "learning_rate": 9.928504604414164e-06,
      "loss": 0.8322,
      "step": 2053
    },
    {
      "epoch": 0.11304970003852716,
      "grad_norm": 0.9095364809036255,
      "learning_rate": 9.928431544910567e-06,
      "loss": 0.8757,
      "step": 2054
    },
    {
      "epoch": 0.11310473884088282,
      "grad_norm": 0.8889689445495605,
      "learning_rate": 9.9283584483662e-06,
      "loss": 0.8583,
      "step": 2055
    },
    {
      "epoch": 0.11315977764323848,
      "grad_norm": 0.8702652454376221,
      "learning_rate": 9.928285314781607e-06,
      "loss": 0.8414,
      "step": 2056
    },
    {
      "epoch": 0.11321481644559414,
      "grad_norm": 0.8531168699264526,
      "learning_rate": 9.928212144157342e-06,
      "loss": 0.7844,
      "step": 2057
    },
    {
      "epoch": 0.1132698552479498,
      "grad_norm": 1.0250271558761597,
      "learning_rate": 9.928138936493956e-06,
      "loss": 0.8766,
      "step": 2058
    },
    {
      "epoch": 0.11332489405030546,
      "grad_norm": 0.7963449358940125,
      "learning_rate": 9.928065691791996e-06,
      "loss": 0.8166,
      "step": 2059
    },
    {
      "epoch": 0.11337993285266112,
      "grad_norm": 1.1033011674880981,
      "learning_rate": 9.927992410052013e-06,
      "loss": 0.8748,
      "step": 2060
    },
    {
      "epoch": 0.11343497165501679,
      "grad_norm": 0.8760959506034851,
      "learning_rate": 9.927919091274558e-06,
      "loss": 0.8623,
      "step": 2061
    },
    {
      "epoch": 0.11349001045737245,
      "grad_norm": 1.1783028841018677,
      "learning_rate": 9.927845735460182e-06,
      "loss": 0.9144,
      "step": 2062
    },
    {
      "epoch": 0.11354504925972811,
      "grad_norm": 0.8868625164031982,
      "learning_rate": 9.927772342609437e-06,
      "loss": 0.8614,
      "step": 2063
    },
    {
      "epoch": 0.11360008806208377,
      "grad_norm": 0.8784704804420471,
      "learning_rate": 9.927698912722874e-06,
      "loss": 0.7802,
      "step": 2064
    },
    {
      "epoch": 0.11365512686443943,
      "grad_norm": 1.0090643167495728,
      "learning_rate": 9.927625445801046e-06,
      "loss": 0.8876,
      "step": 2065
    },
    {
      "epoch": 0.1137101656667951,
      "grad_norm": 0.7624390721321106,
      "learning_rate": 9.927551941844502e-06,
      "loss": 0.794,
      "step": 2066
    },
    {
      "epoch": 0.11376520446915075,
      "grad_norm": 0.7814189791679382,
      "learning_rate": 9.927478400853798e-06,
      "loss": 0.8176,
      "step": 2067
    },
    {
      "epoch": 0.11382024327150642,
      "grad_norm": 0.876338541507721,
      "learning_rate": 9.927404822829486e-06,
      "loss": 0.8634,
      "step": 2068
    },
    {
      "epoch": 0.11387528207386208,
      "grad_norm": 0.7931430339813232,
      "learning_rate": 9.927331207772117e-06,
      "loss": 0.8012,
      "step": 2069
    },
    {
      "epoch": 0.11393032087621774,
      "grad_norm": 1.0064504146575928,
      "learning_rate": 9.927257555682246e-06,
      "loss": 0.8321,
      "step": 2070
    },
    {
      "epoch": 0.1139853596785734,
      "grad_norm": 0.8233053684234619,
      "learning_rate": 9.927183866560425e-06,
      "loss": 0.8004,
      "step": 2071
    },
    {
      "epoch": 0.11404039848092906,
      "grad_norm": 1.0106632709503174,
      "learning_rate": 9.927110140407211e-06,
      "loss": 0.8627,
      "step": 2072
    },
    {
      "epoch": 0.11409543728328471,
      "grad_norm": 0.8262843489646912,
      "learning_rate": 9.927036377223155e-06,
      "loss": 0.737,
      "step": 2073
    },
    {
      "epoch": 0.11415047608564037,
      "grad_norm": 0.9349029660224915,
      "learning_rate": 9.926962577008813e-06,
      "loss": 0.9049,
      "step": 2074
    },
    {
      "epoch": 0.11420551488799603,
      "grad_norm": 0.8689929842948914,
      "learning_rate": 9.926888739764739e-06,
      "loss": 0.7858,
      "step": 2075
    },
    {
      "epoch": 0.1142605536903517,
      "grad_norm": 0.8442347645759583,
      "learning_rate": 9.926814865491487e-06,
      "loss": 0.8145,
      "step": 2076
    },
    {
      "epoch": 0.11431559249270735,
      "grad_norm": 0.9143397212028503,
      "learning_rate": 9.926740954189615e-06,
      "loss": 0.8025,
      "step": 2077
    },
    {
      "epoch": 0.11437063129506302,
      "grad_norm": 1.293251395225525,
      "learning_rate": 9.926667005859676e-06,
      "loss": 1.0256,
      "step": 2078
    },
    {
      "epoch": 0.11442567009741868,
      "grad_norm": 0.9661351442337036,
      "learning_rate": 9.926593020502226e-06,
      "loss": 0.991,
      "step": 2079
    },
    {
      "epoch": 0.11448070889977434,
      "grad_norm": 0.8110861778259277,
      "learning_rate": 9.926518998117823e-06,
      "loss": 0.7129,
      "step": 2080
    },
    {
      "epoch": 0.11453574770213,
      "grad_norm": 0.8351119160652161,
      "learning_rate": 9.92644493870702e-06,
      "loss": 0.8894,
      "step": 2081
    },
    {
      "epoch": 0.11459078650448566,
      "grad_norm": 0.8492733240127563,
      "learning_rate": 9.926370842270377e-06,
      "loss": 0.8039,
      "step": 2082
    },
    {
      "epoch": 0.11464582530684132,
      "grad_norm": 0.895353376865387,
      "learning_rate": 9.92629670880845e-06,
      "loss": 0.8743,
      "step": 2083
    },
    {
      "epoch": 0.11470086410919698,
      "grad_norm": 0.7871271967887878,
      "learning_rate": 9.926222538321795e-06,
      "loss": 0.8426,
      "step": 2084
    },
    {
      "epoch": 0.11475590291155265,
      "grad_norm": 0.8904643058776855,
      "learning_rate": 9.92614833081097e-06,
      "loss": 0.8454,
      "step": 2085
    },
    {
      "epoch": 0.11481094171390831,
      "grad_norm": 0.9166308641433716,
      "learning_rate": 9.926074086276532e-06,
      "loss": 0.9162,
      "step": 2086
    },
    {
      "epoch": 0.11486598051626397,
      "grad_norm": 0.8730728626251221,
      "learning_rate": 9.92599980471904e-06,
      "loss": 0.8524,
      "step": 2087
    },
    {
      "epoch": 0.11492101931861963,
      "grad_norm": 0.7932829260826111,
      "learning_rate": 9.925925486139052e-06,
      "loss": 0.7838,
      "step": 2088
    },
    {
      "epoch": 0.11497605812097529,
      "grad_norm": 1.0033760070800781,
      "learning_rate": 9.925851130537127e-06,
      "loss": 0.8746,
      "step": 2089
    },
    {
      "epoch": 0.11503109692333095,
      "grad_norm": 0.7783192992210388,
      "learning_rate": 9.925776737913823e-06,
      "loss": 0.7308,
      "step": 2090
    },
    {
      "epoch": 0.11508613572568661,
      "grad_norm": 0.8441587686538696,
      "learning_rate": 9.925702308269702e-06,
      "loss": 0.7933,
      "step": 2091
    },
    {
      "epoch": 0.11514117452804228,
      "grad_norm": 0.9433023929595947,
      "learning_rate": 9.925627841605319e-06,
      "loss": 0.7857,
      "step": 2092
    },
    {
      "epoch": 0.11519621333039794,
      "grad_norm": 0.8958256244659424,
      "learning_rate": 9.925553337921235e-06,
      "loss": 0.9116,
      "step": 2093
    },
    {
      "epoch": 0.11525125213275358,
      "grad_norm": 0.7610845565795898,
      "learning_rate": 9.925478797218011e-06,
      "loss": 0.8006,
      "step": 2094
    },
    {
      "epoch": 0.11530629093510925,
      "grad_norm": 0.7977023720741272,
      "learning_rate": 9.925404219496207e-06,
      "loss": 0.8068,
      "step": 2095
    },
    {
      "epoch": 0.11536132973746491,
      "grad_norm": 0.8087283372879028,
      "learning_rate": 9.925329604756383e-06,
      "loss": 0.7968,
      "step": 2096
    },
    {
      "epoch": 0.11541636853982057,
      "grad_norm": 1.1066477298736572,
      "learning_rate": 9.925254952999102e-06,
      "loss": 0.8167,
      "step": 2097
    },
    {
      "epoch": 0.11547140734217623,
      "grad_norm": 0.7806832194328308,
      "learning_rate": 9.925180264224921e-06,
      "loss": 0.8069,
      "step": 2098
    },
    {
      "epoch": 0.11552644614453189,
      "grad_norm": 0.7745190858840942,
      "learning_rate": 9.925105538434406e-06,
      "loss": 0.7968,
      "step": 2099
    },
    {
      "epoch": 0.11558148494688755,
      "grad_norm": 0.9045543074607849,
      "learning_rate": 9.925030775628113e-06,
      "loss": 0.8417,
      "step": 2100
    },
    {
      "epoch": 0.11563652374924321,
      "grad_norm": 1.2962623834609985,
      "learning_rate": 9.924955975806608e-06,
      "loss": 0.8162,
      "step": 2101
    },
    {
      "epoch": 0.11569156255159888,
      "grad_norm": 0.8571485877037048,
      "learning_rate": 9.924881138970453e-06,
      "loss": 0.8581,
      "step": 2102
    },
    {
      "epoch": 0.11574660135395454,
      "grad_norm": 0.8326650857925415,
      "learning_rate": 9.92480626512021e-06,
      "loss": 0.8438,
      "step": 2103
    },
    {
      "epoch": 0.1158016401563102,
      "grad_norm": 0.7973701357841492,
      "learning_rate": 9.924731354256441e-06,
      "loss": 0.8337,
      "step": 2104
    },
    {
      "epoch": 0.11585667895866586,
      "grad_norm": 0.8614075779914856,
      "learning_rate": 9.924656406379708e-06,
      "loss": 0.8275,
      "step": 2105
    },
    {
      "epoch": 0.11591171776102152,
      "grad_norm": 0.7911350131034851,
      "learning_rate": 9.924581421490577e-06,
      "loss": 0.8032,
      "step": 2106
    },
    {
      "epoch": 0.11596675656337718,
      "grad_norm": 0.8763116598129272,
      "learning_rate": 9.92450639958961e-06,
      "loss": 0.8725,
      "step": 2107
    },
    {
      "epoch": 0.11602179536573284,
      "grad_norm": 0.9754133224487305,
      "learning_rate": 9.92443134067737e-06,
      "loss": 0.9115,
      "step": 2108
    },
    {
      "epoch": 0.1160768341680885,
      "grad_norm": 0.7783731818199158,
      "learning_rate": 9.924356244754425e-06,
      "loss": 0.8223,
      "step": 2109
    },
    {
      "epoch": 0.11613187297044417,
      "grad_norm": 0.865301787853241,
      "learning_rate": 9.924281111821335e-06,
      "loss": 0.8053,
      "step": 2110
    },
    {
      "epoch": 0.11618691177279983,
      "grad_norm": 0.8654297590255737,
      "learning_rate": 9.924205941878666e-06,
      "loss": 0.716,
      "step": 2111
    },
    {
      "epoch": 0.11624195057515549,
      "grad_norm": 0.7646550536155701,
      "learning_rate": 9.924130734926982e-06,
      "loss": 0.8027,
      "step": 2112
    },
    {
      "epoch": 0.11629698937751115,
      "grad_norm": 0.810587465763092,
      "learning_rate": 9.924055490966851e-06,
      "loss": 0.7416,
      "step": 2113
    },
    {
      "epoch": 0.11635202817986681,
      "grad_norm": 0.8610082268714905,
      "learning_rate": 9.923980209998838e-06,
      "loss": 0.8527,
      "step": 2114
    },
    {
      "epoch": 0.11640706698222247,
      "grad_norm": 0.8409233689308167,
      "learning_rate": 9.923904892023506e-06,
      "loss": 0.8169,
      "step": 2115
    },
    {
      "epoch": 0.11646210578457812,
      "grad_norm": 0.7786587476730347,
      "learning_rate": 9.923829537041425e-06,
      "loss": 0.6897,
      "step": 2116
    },
    {
      "epoch": 0.11651714458693378,
      "grad_norm": 0.852908730506897,
      "learning_rate": 9.923754145053158e-06,
      "loss": 0.7821,
      "step": 2117
    },
    {
      "epoch": 0.11657218338928944,
      "grad_norm": 0.9130391478538513,
      "learning_rate": 9.923678716059273e-06,
      "loss": 1.0377,
      "step": 2118
    },
    {
      "epoch": 0.1166272221916451,
      "grad_norm": 0.8371701240539551,
      "learning_rate": 9.923603250060336e-06,
      "loss": 0.8312,
      "step": 2119
    },
    {
      "epoch": 0.11668226099400077,
      "grad_norm": 0.8045756220817566,
      "learning_rate": 9.923527747056916e-06,
      "loss": 0.7971,
      "step": 2120
    },
    {
      "epoch": 0.11673729979635643,
      "grad_norm": 0.8832160234451294,
      "learning_rate": 9.923452207049577e-06,
      "loss": 0.7362,
      "step": 2121
    },
    {
      "epoch": 0.11679233859871209,
      "grad_norm": 0.8253088593482971,
      "learning_rate": 9.923376630038893e-06,
      "loss": 0.8177,
      "step": 2122
    },
    {
      "epoch": 0.11684737740106775,
      "grad_norm": 0.7953168749809265,
      "learning_rate": 9.923301016025424e-06,
      "loss": 0.7053,
      "step": 2123
    },
    {
      "epoch": 0.11690241620342341,
      "grad_norm": 0.7256457805633545,
      "learning_rate": 9.923225365009745e-06,
      "loss": 0.7554,
      "step": 2124
    },
    {
      "epoch": 0.11695745500577907,
      "grad_norm": 0.9896693229675293,
      "learning_rate": 9.923149676992424e-06,
      "loss": 0.8285,
      "step": 2125
    },
    {
      "epoch": 0.11701249380813473,
      "grad_norm": 0.7846312522888184,
      "learning_rate": 9.923073951974023e-06,
      "loss": 0.7527,
      "step": 2126
    },
    {
      "epoch": 0.1170675326104904,
      "grad_norm": 0.8949825167655945,
      "learning_rate": 9.92299818995512e-06,
      "loss": 0.8545,
      "step": 2127
    },
    {
      "epoch": 0.11712257141284606,
      "grad_norm": 1.0023548603057861,
      "learning_rate": 9.922922390936278e-06,
      "loss": 0.7668,
      "step": 2128
    },
    {
      "epoch": 0.11717761021520172,
      "grad_norm": 0.8663881421089172,
      "learning_rate": 9.92284655491807e-06,
      "loss": 0.8073,
      "step": 2129
    },
    {
      "epoch": 0.11723264901755738,
      "grad_norm": 0.8274385929107666,
      "learning_rate": 9.922770681901064e-06,
      "loss": 0.9002,
      "step": 2130
    },
    {
      "epoch": 0.11728768781991304,
      "grad_norm": 0.8508959412574768,
      "learning_rate": 9.922694771885832e-06,
      "loss": 0.9325,
      "step": 2131
    },
    {
      "epoch": 0.1173427266222687,
      "grad_norm": 0.8176792860031128,
      "learning_rate": 9.922618824872946e-06,
      "loss": 0.8415,
      "step": 2132
    },
    {
      "epoch": 0.11739776542462436,
      "grad_norm": 0.770951509475708,
      "learning_rate": 9.922542840862971e-06,
      "loss": 0.8051,
      "step": 2133
    },
    {
      "epoch": 0.11745280422698003,
      "grad_norm": 0.8558167219161987,
      "learning_rate": 9.922466819856484e-06,
      "loss": 0.85,
      "step": 2134
    },
    {
      "epoch": 0.11750784302933569,
      "grad_norm": 0.8288151025772095,
      "learning_rate": 9.922390761854053e-06,
      "loss": 0.8141,
      "step": 2135
    },
    {
      "epoch": 0.11756288183169135,
      "grad_norm": 0.8220882415771484,
      "learning_rate": 9.922314666856252e-06,
      "loss": 0.8109,
      "step": 2136
    },
    {
      "epoch": 0.117617920634047,
      "grad_norm": 0.7875000238418579,
      "learning_rate": 9.92223853486365e-06,
      "loss": 0.9085,
      "step": 2137
    },
    {
      "epoch": 0.11767295943640266,
      "grad_norm": 0.8052374124526978,
      "learning_rate": 9.922162365876822e-06,
      "loss": 0.8785,
      "step": 2138
    },
    {
      "epoch": 0.11772799823875832,
      "grad_norm": 1.0311180353164673,
      "learning_rate": 9.922086159896338e-06,
      "loss": 0.9112,
      "step": 2139
    },
    {
      "epoch": 0.11778303704111398,
      "grad_norm": 0.943911075592041,
      "learning_rate": 9.922009916922773e-06,
      "loss": 0.8332,
      "step": 2140
    },
    {
      "epoch": 0.11783807584346964,
      "grad_norm": 0.8156648278236389,
      "learning_rate": 9.921933636956697e-06,
      "loss": 0.8837,
      "step": 2141
    },
    {
      "epoch": 0.1178931146458253,
      "grad_norm": 0.860292375087738,
      "learning_rate": 9.921857319998688e-06,
      "loss": 0.7963,
      "step": 2142
    },
    {
      "epoch": 0.11794815344818096,
      "grad_norm": 0.8861456513404846,
      "learning_rate": 9.921780966049315e-06,
      "loss": 0.8335,
      "step": 2143
    },
    {
      "epoch": 0.11800319225053663,
      "grad_norm": 0.793533205986023,
      "learning_rate": 9.921704575109155e-06,
      "loss": 0.7881,
      "step": 2144
    },
    {
      "epoch": 0.11805823105289229,
      "grad_norm": 0.8039320111274719,
      "learning_rate": 9.921628147178781e-06,
      "loss": 0.8369,
      "step": 2145
    },
    {
      "epoch": 0.11811326985524795,
      "grad_norm": 0.8785450458526611,
      "learning_rate": 9.921551682258765e-06,
      "loss": 0.7981,
      "step": 2146
    },
    {
      "epoch": 0.11816830865760361,
      "grad_norm": 0.810251772403717,
      "learning_rate": 9.921475180349687e-06,
      "loss": 0.7926,
      "step": 2147
    },
    {
      "epoch": 0.11822334745995927,
      "grad_norm": 0.8470801115036011,
      "learning_rate": 9.921398641452117e-06,
      "loss": 0.8061,
      "step": 2148
    },
    {
      "epoch": 0.11827838626231493,
      "grad_norm": 0.8147469162940979,
      "learning_rate": 9.921322065566633e-06,
      "loss": 0.7906,
      "step": 2149
    },
    {
      "epoch": 0.1183334250646706,
      "grad_norm": 0.8792327046394348,
      "learning_rate": 9.92124545269381e-06,
      "loss": 0.9025,
      "step": 2150
    },
    {
      "epoch": 0.11838846386702626,
      "grad_norm": 0.794607400894165,
      "learning_rate": 9.921168802834223e-06,
      "loss": 0.8284,
      "step": 2151
    },
    {
      "epoch": 0.11844350266938192,
      "grad_norm": 0.8601556420326233,
      "learning_rate": 9.921092115988447e-06,
      "loss": 0.8196,
      "step": 2152
    },
    {
      "epoch": 0.11849854147173758,
      "grad_norm": 0.786967933177948,
      "learning_rate": 9.921015392157062e-06,
      "loss": 0.8744,
      "step": 2153
    },
    {
      "epoch": 0.11855358027409324,
      "grad_norm": 0.8481432199478149,
      "learning_rate": 9.920938631340641e-06,
      "loss": 0.7206,
      "step": 2154
    },
    {
      "epoch": 0.1186086190764489,
      "grad_norm": 0.8025142550468445,
      "learning_rate": 9.920861833539765e-06,
      "loss": 0.8126,
      "step": 2155
    },
    {
      "epoch": 0.11866365787880456,
      "grad_norm": 0.9853057265281677,
      "learning_rate": 9.920784998755006e-06,
      "loss": 0.8883,
      "step": 2156
    },
    {
      "epoch": 0.11871869668116022,
      "grad_norm": 1.0008476972579956,
      "learning_rate": 9.920708126986947e-06,
      "loss": 0.9326,
      "step": 2157
    },
    {
      "epoch": 0.11877373548351589,
      "grad_norm": 0.837347686290741,
      "learning_rate": 9.920631218236161e-06,
      "loss": 0.9002,
      "step": 2158
    },
    {
      "epoch": 0.11882877428587153,
      "grad_norm": 0.7866735458374023,
      "learning_rate": 9.920554272503227e-06,
      "loss": 0.765,
      "step": 2159
    },
    {
      "epoch": 0.1188838130882272,
      "grad_norm": 0.8714935779571533,
      "learning_rate": 9.920477289788726e-06,
      "loss": 1.0294,
      "step": 2160
    },
    {
      "epoch": 0.11893885189058286,
      "grad_norm": 1.0671826601028442,
      "learning_rate": 9.920400270093234e-06,
      "loss": 0.8341,
      "step": 2161
    },
    {
      "epoch": 0.11899389069293852,
      "grad_norm": 0.8594604134559631,
      "learning_rate": 9.92032321341733e-06,
      "loss": 0.8731,
      "step": 2162
    },
    {
      "epoch": 0.11904892949529418,
      "grad_norm": 0.8387738466262817,
      "learning_rate": 9.920246119761597e-06,
      "loss": 0.7898,
      "step": 2163
    },
    {
      "epoch": 0.11910396829764984,
      "grad_norm": 0.8957195281982422,
      "learning_rate": 9.920168989126608e-06,
      "loss": 0.8475,
      "step": 2164
    },
    {
      "epoch": 0.1191590071000055,
      "grad_norm": 0.8224207162857056,
      "learning_rate": 9.920091821512948e-06,
      "loss": 0.7944,
      "step": 2165
    },
    {
      "epoch": 0.11921404590236116,
      "grad_norm": 1.0309031009674072,
      "learning_rate": 9.920014616921192e-06,
      "loss": 0.8992,
      "step": 2166
    },
    {
      "epoch": 0.11926908470471682,
      "grad_norm": 0.7300832271575928,
      "learning_rate": 9.919937375351925e-06,
      "loss": 0.7016,
      "step": 2167
    },
    {
      "epoch": 0.11932412350707249,
      "grad_norm": 0.7565537691116333,
      "learning_rate": 9.919860096805724e-06,
      "loss": 0.8113,
      "step": 2168
    },
    {
      "epoch": 0.11937916230942815,
      "grad_norm": 1.0101505517959595,
      "learning_rate": 9.919782781283174e-06,
      "loss": 0.8765,
      "step": 2169
    },
    {
      "epoch": 0.11943420111178381,
      "grad_norm": 0.8369461894035339,
      "learning_rate": 9.919705428784852e-06,
      "loss": 0.8248,
      "step": 2170
    },
    {
      "epoch": 0.11948923991413947,
      "grad_norm": 0.8106105327606201,
      "learning_rate": 9.919628039311342e-06,
      "loss": 0.8585,
      "step": 2171
    },
    {
      "epoch": 0.11954427871649513,
      "grad_norm": 0.7863745093345642,
      "learning_rate": 9.919550612863224e-06,
      "loss": 0.8393,
      "step": 2172
    },
    {
      "epoch": 0.11959931751885079,
      "grad_norm": 0.8664719462394714,
      "learning_rate": 9.919473149441081e-06,
      "loss": 0.8882,
      "step": 2173
    },
    {
      "epoch": 0.11965435632120645,
      "grad_norm": 0.6977574825286865,
      "learning_rate": 9.919395649045494e-06,
      "loss": 0.7264,
      "step": 2174
    },
    {
      "epoch": 0.11970939512356212,
      "grad_norm": 0.8000102639198303,
      "learning_rate": 9.919318111677045e-06,
      "loss": 0.7828,
      "step": 2175
    },
    {
      "epoch": 0.11976443392591778,
      "grad_norm": 0.868228018283844,
      "learning_rate": 9.91924053733632e-06,
      "loss": 0.7904,
      "step": 2176
    },
    {
      "epoch": 0.11981947272827344,
      "grad_norm": 0.839080274105072,
      "learning_rate": 9.9191629260239e-06,
      "loss": 0.7663,
      "step": 2177
    },
    {
      "epoch": 0.1198745115306291,
      "grad_norm": 0.8222747445106506,
      "learning_rate": 9.919085277740366e-06,
      "loss": 0.7208,
      "step": 2178
    },
    {
      "epoch": 0.11992955033298476,
      "grad_norm": 1.4550986289978027,
      "learning_rate": 9.919007592486304e-06,
      "loss": 0.8154,
      "step": 2179
    },
    {
      "epoch": 0.11998458913534041,
      "grad_norm": 0.9110257625579834,
      "learning_rate": 9.9189298702623e-06,
      "loss": 0.8134,
      "step": 2180
    },
    {
      "epoch": 0.12003962793769607,
      "grad_norm": 0.84796142578125,
      "learning_rate": 9.918852111068935e-06,
      "loss": 0.8074,
      "step": 2181
    },
    {
      "epoch": 0.12009466674005173,
      "grad_norm": 0.8134179711341858,
      "learning_rate": 9.918774314906793e-06,
      "loss": 0.6335,
      "step": 2182
    },
    {
      "epoch": 0.12014970554240739,
      "grad_norm": 0.8481448888778687,
      "learning_rate": 9.918696481776461e-06,
      "loss": 0.8804,
      "step": 2183
    },
    {
      "epoch": 0.12020474434476305,
      "grad_norm": 0.88057941198349,
      "learning_rate": 9.918618611678523e-06,
      "loss": 0.9326,
      "step": 2184
    },
    {
      "epoch": 0.12025978314711872,
      "grad_norm": 0.8435977697372437,
      "learning_rate": 9.918540704613564e-06,
      "loss": 0.8141,
      "step": 2185
    },
    {
      "epoch": 0.12031482194947438,
      "grad_norm": 0.8186982870101929,
      "learning_rate": 9.918462760582169e-06,
      "loss": 0.837,
      "step": 2186
    },
    {
      "epoch": 0.12036986075183004,
      "grad_norm": 0.887783944606781,
      "learning_rate": 9.918384779584924e-06,
      "loss": 0.8062,
      "step": 2187
    },
    {
      "epoch": 0.1204248995541857,
      "grad_norm": 0.9368415474891663,
      "learning_rate": 9.918306761622417e-06,
      "loss": 1.0098,
      "step": 2188
    },
    {
      "epoch": 0.12047993835654136,
      "grad_norm": 0.8443986773490906,
      "learning_rate": 9.918228706695232e-06,
      "loss": 0.8178,
      "step": 2189
    },
    {
      "epoch": 0.12053497715889702,
      "grad_norm": 0.7897284626960754,
      "learning_rate": 9.918150614803956e-06,
      "loss": 0.8013,
      "step": 2190
    },
    {
      "epoch": 0.12059001596125268,
      "grad_norm": 0.886012077331543,
      "learning_rate": 9.91807248594918e-06,
      "loss": 0.8141,
      "step": 2191
    },
    {
      "epoch": 0.12064505476360834,
      "grad_norm": 0.8585757613182068,
      "learning_rate": 9.917994320131484e-06,
      "loss": 0.8381,
      "step": 2192
    },
    {
      "epoch": 0.120700093565964,
      "grad_norm": 1.6192269325256348,
      "learning_rate": 9.917916117351459e-06,
      "loss": 0.9082,
      "step": 2193
    },
    {
      "epoch": 0.12075513236831967,
      "grad_norm": 1.160414457321167,
      "learning_rate": 9.917837877609695e-06,
      "loss": 0.8673,
      "step": 2194
    },
    {
      "epoch": 0.12081017117067533,
      "grad_norm": 0.8363412022590637,
      "learning_rate": 9.917759600906775e-06,
      "loss": 0.816,
      "step": 2195
    },
    {
      "epoch": 0.12086520997303099,
      "grad_norm": 0.8344097137451172,
      "learning_rate": 9.917681287243292e-06,
      "loss": 0.8629,
      "step": 2196
    },
    {
      "epoch": 0.12092024877538665,
      "grad_norm": 0.9817582368850708,
      "learning_rate": 9.917602936619834e-06,
      "loss": 0.8106,
      "step": 2197
    },
    {
      "epoch": 0.12097528757774231,
      "grad_norm": 0.8828088641166687,
      "learning_rate": 9.917524549036987e-06,
      "loss": 0.8465,
      "step": 2198
    },
    {
      "epoch": 0.12103032638009797,
      "grad_norm": 0.8428277969360352,
      "learning_rate": 9.917446124495344e-06,
      "loss": 0.7721,
      "step": 2199
    },
    {
      "epoch": 0.12108536518245364,
      "grad_norm": 0.8748664855957031,
      "learning_rate": 9.917367662995489e-06,
      "loss": 0.8679,
      "step": 2200
    },
    {
      "epoch": 0.1211404039848093,
      "grad_norm": 0.8652347922325134,
      "learning_rate": 9.917289164538018e-06,
      "loss": 0.8906,
      "step": 2201
    },
    {
      "epoch": 0.12119544278716494,
      "grad_norm": 1.157142162322998,
      "learning_rate": 9.917210629123518e-06,
      "loss": 0.9046,
      "step": 2202
    },
    {
      "epoch": 0.1212504815895206,
      "grad_norm": 0.8186333179473877,
      "learning_rate": 9.917132056752576e-06,
      "loss": 0.8494,
      "step": 2203
    },
    {
      "epoch": 0.12130552039187627,
      "grad_norm": 0.7769078612327576,
      "learning_rate": 9.917053447425788e-06,
      "loss": 0.8018,
      "step": 2204
    },
    {
      "epoch": 0.12136055919423193,
      "grad_norm": 0.9190469980239868,
      "learning_rate": 9.916974801143742e-06,
      "loss": 0.8206,
      "step": 2205
    },
    {
      "epoch": 0.12141559799658759,
      "grad_norm": 1.2200725078582764,
      "learning_rate": 9.91689611790703e-06,
      "loss": 0.9109,
      "step": 2206
    },
    {
      "epoch": 0.12147063679894325,
      "grad_norm": 0.7902093529701233,
      "learning_rate": 9.916817397716243e-06,
      "loss": 0.8314,
      "step": 2207
    },
    {
      "epoch": 0.12152567560129891,
      "grad_norm": 0.8160610198974609,
      "learning_rate": 9.91673864057197e-06,
      "loss": 0.8605,
      "step": 2208
    },
    {
      "epoch": 0.12158071440365457,
      "grad_norm": 0.833163857460022,
      "learning_rate": 9.916659846474807e-06,
      "loss": 0.8125,
      "step": 2209
    },
    {
      "epoch": 0.12163575320601024,
      "grad_norm": 0.776314377784729,
      "learning_rate": 9.916581015425346e-06,
      "loss": 0.8137,
      "step": 2210
    },
    {
      "epoch": 0.1216907920083659,
      "grad_norm": 0.8525915145874023,
      "learning_rate": 9.916502147424178e-06,
      "loss": 0.8703,
      "step": 2211
    },
    {
      "epoch": 0.12174583081072156,
      "grad_norm": 0.8268684148788452,
      "learning_rate": 9.916423242471895e-06,
      "loss": 0.7775,
      "step": 2212
    },
    {
      "epoch": 0.12180086961307722,
      "grad_norm": 0.8717706799507141,
      "learning_rate": 9.916344300569091e-06,
      "loss": 0.8002,
      "step": 2213
    },
    {
      "epoch": 0.12185590841543288,
      "grad_norm": 0.9499961137771606,
      "learning_rate": 9.91626532171636e-06,
      "loss": 0.8861,
      "step": 2214
    },
    {
      "epoch": 0.12191094721778854,
      "grad_norm": 0.9521885514259338,
      "learning_rate": 9.916186305914296e-06,
      "loss": 0.7602,
      "step": 2215
    },
    {
      "epoch": 0.1219659860201442,
      "grad_norm": 0.8945447206497192,
      "learning_rate": 9.916107253163488e-06,
      "loss": 0.8603,
      "step": 2216
    },
    {
      "epoch": 0.12202102482249987,
      "grad_norm": 0.8232392072677612,
      "learning_rate": 9.916028163464536e-06,
      "loss": 0.8419,
      "step": 2217
    },
    {
      "epoch": 0.12207606362485553,
      "grad_norm": 0.8183467984199524,
      "learning_rate": 9.915949036818032e-06,
      "loss": 0.9038,
      "step": 2218
    },
    {
      "epoch": 0.12213110242721119,
      "grad_norm": 0.7805467247962952,
      "learning_rate": 9.915869873224571e-06,
      "loss": 0.7313,
      "step": 2219
    },
    {
      "epoch": 0.12218614122956685,
      "grad_norm": 0.838101327419281,
      "learning_rate": 9.915790672684749e-06,
      "loss": 0.7973,
      "step": 2220
    },
    {
      "epoch": 0.12224118003192251,
      "grad_norm": 0.7795171141624451,
      "learning_rate": 9.915711435199158e-06,
      "loss": 0.7796,
      "step": 2221
    },
    {
      "epoch": 0.12229621883427817,
      "grad_norm": 0.7971234917640686,
      "learning_rate": 9.915632160768398e-06,
      "loss": 0.8309,
      "step": 2222
    },
    {
      "epoch": 0.12235125763663382,
      "grad_norm": 0.8543851375579834,
      "learning_rate": 9.915552849393061e-06,
      "loss": 0.7826,
      "step": 2223
    },
    {
      "epoch": 0.12240629643898948,
      "grad_norm": 0.9315086007118225,
      "learning_rate": 9.915473501073744e-06,
      "loss": 0.9294,
      "step": 2224
    },
    {
      "epoch": 0.12246133524134514,
      "grad_norm": 0.8794427514076233,
      "learning_rate": 9.915394115811046e-06,
      "loss": 0.8968,
      "step": 2225
    },
    {
      "epoch": 0.1225163740437008,
      "grad_norm": 0.9499204754829407,
      "learning_rate": 9.91531469360556e-06,
      "loss": 0.9841,
      "step": 2226
    },
    {
      "epoch": 0.12257141284605647,
      "grad_norm": 0.9233788251876831,
      "learning_rate": 9.915235234457885e-06,
      "loss": 0.7794,
      "step": 2227
    },
    {
      "epoch": 0.12262645164841213,
      "grad_norm": 0.8971870541572571,
      "learning_rate": 9.915155738368618e-06,
      "loss": 0.919,
      "step": 2228
    },
    {
      "epoch": 0.12268149045076779,
      "grad_norm": 0.8122105002403259,
      "learning_rate": 9.915076205338356e-06,
      "loss": 0.8227,
      "step": 2229
    },
    {
      "epoch": 0.12273652925312345,
      "grad_norm": 0.7878004908561707,
      "learning_rate": 9.914996635367696e-06,
      "loss": 0.7622,
      "step": 2230
    },
    {
      "epoch": 0.12279156805547911,
      "grad_norm": 0.8229606747627258,
      "learning_rate": 9.914917028457238e-06,
      "loss": 0.8265,
      "step": 2231
    },
    {
      "epoch": 0.12284660685783477,
      "grad_norm": 0.8972312808036804,
      "learning_rate": 9.914837384607578e-06,
      "loss": 0.8914,
      "step": 2232
    },
    {
      "epoch": 0.12290164566019043,
      "grad_norm": 0.762922465801239,
      "learning_rate": 9.914757703819318e-06,
      "loss": 0.6853,
      "step": 2233
    },
    {
      "epoch": 0.1229566844625461,
      "grad_norm": 0.8949442505836487,
      "learning_rate": 9.914677986093054e-06,
      "loss": 0.8303,
      "step": 2234
    },
    {
      "epoch": 0.12301172326490176,
      "grad_norm": 1.0220820903778076,
      "learning_rate": 9.914598231429384e-06,
      "loss": 1.0027,
      "step": 2235
    },
    {
      "epoch": 0.12306676206725742,
      "grad_norm": 0.8265436887741089,
      "learning_rate": 9.914518439828911e-06,
      "loss": 0.8317,
      "step": 2236
    },
    {
      "epoch": 0.12312180086961308,
      "grad_norm": 0.780444324016571,
      "learning_rate": 9.914438611292231e-06,
      "loss": 0.756,
      "step": 2237
    },
    {
      "epoch": 0.12317683967196874,
      "grad_norm": 0.8569482564926147,
      "learning_rate": 9.914358745819948e-06,
      "loss": 0.8126,
      "step": 2238
    },
    {
      "epoch": 0.1232318784743244,
      "grad_norm": 0.8167145848274231,
      "learning_rate": 9.91427884341266e-06,
      "loss": 0.8345,
      "step": 2239
    },
    {
      "epoch": 0.12328691727668006,
      "grad_norm": 0.7915990948677063,
      "learning_rate": 9.914198904070967e-06,
      "loss": 0.7416,
      "step": 2240
    },
    {
      "epoch": 0.12334195607903573,
      "grad_norm": 0.8568083047866821,
      "learning_rate": 9.91411892779547e-06,
      "loss": 0.8329,
      "step": 2241
    },
    {
      "epoch": 0.12339699488139139,
      "grad_norm": 1.1727303266525269,
      "learning_rate": 9.914038914586772e-06,
      "loss": 0.8421,
      "step": 2242
    },
    {
      "epoch": 0.12345203368374705,
      "grad_norm": 0.8706398010253906,
      "learning_rate": 9.913958864445472e-06,
      "loss": 0.9013,
      "step": 2243
    },
    {
      "epoch": 0.12350707248610271,
      "grad_norm": 0.8376144170761108,
      "learning_rate": 9.913878777372173e-06,
      "loss": 0.8456,
      "step": 2244
    },
    {
      "epoch": 0.12356211128845836,
      "grad_norm": 0.8388974070549011,
      "learning_rate": 9.913798653367478e-06,
      "loss": 0.787,
      "step": 2245
    },
    {
      "epoch": 0.12361715009081402,
      "grad_norm": 0.8625446557998657,
      "learning_rate": 9.913718492431984e-06,
      "loss": 0.7758,
      "step": 2246
    },
    {
      "epoch": 0.12367218889316968,
      "grad_norm": 0.8805570006370544,
      "learning_rate": 9.913638294566299e-06,
      "loss": 0.8755,
      "step": 2247
    },
    {
      "epoch": 0.12372722769552534,
      "grad_norm": 0.8102611899375916,
      "learning_rate": 9.913558059771025e-06,
      "loss": 0.8495,
      "step": 2248
    },
    {
      "epoch": 0.123782266497881,
      "grad_norm": 0.8506311774253845,
      "learning_rate": 9.913477788046762e-06,
      "loss": 0.7413,
      "step": 2249
    },
    {
      "epoch": 0.12383730530023666,
      "grad_norm": 1.0789196491241455,
      "learning_rate": 9.913397479394116e-06,
      "loss": 0.8993,
      "step": 2250
    },
    {
      "epoch": 0.12389234410259232,
      "grad_norm": 1.5664849281311035,
      "learning_rate": 9.91331713381369e-06,
      "loss": 0.8322,
      "step": 2251
    },
    {
      "epoch": 0.12394738290494799,
      "grad_norm": 1.1347390413284302,
      "learning_rate": 9.913236751306085e-06,
      "loss": 0.8756,
      "step": 2252
    },
    {
      "epoch": 0.12400242170730365,
      "grad_norm": 0.8111063241958618,
      "learning_rate": 9.913156331871911e-06,
      "loss": 0.831,
      "step": 2253
    },
    {
      "epoch": 0.12405746050965931,
      "grad_norm": 0.817812979221344,
      "learning_rate": 9.913075875511769e-06,
      "loss": 0.8531,
      "step": 2254
    },
    {
      "epoch": 0.12411249931201497,
      "grad_norm": 0.7678318619728088,
      "learning_rate": 9.912995382226263e-06,
      "loss": 0.8028,
      "step": 2255
    },
    {
      "epoch": 0.12416753811437063,
      "grad_norm": 0.8207805156707764,
      "learning_rate": 9.912914852015998e-06,
      "loss": 0.8856,
      "step": 2256
    },
    {
      "epoch": 0.1242225769167263,
      "grad_norm": 0.978484570980072,
      "learning_rate": 9.912834284881582e-06,
      "loss": 0.933,
      "step": 2257
    },
    {
      "epoch": 0.12427761571908195,
      "grad_norm": 0.9215858578681946,
      "learning_rate": 9.912753680823617e-06,
      "loss": 0.7771,
      "step": 2258
    },
    {
      "epoch": 0.12433265452143762,
      "grad_norm": 0.8542179465293884,
      "learning_rate": 9.91267303984271e-06,
      "loss": 0.8652,
      "step": 2259
    },
    {
      "epoch": 0.12438769332379328,
      "grad_norm": 0.7985575199127197,
      "learning_rate": 9.912592361939469e-06,
      "loss": 0.7011,
      "step": 2260
    },
    {
      "epoch": 0.12444273212614894,
      "grad_norm": 0.8868670463562012,
      "learning_rate": 9.912511647114498e-06,
      "loss": 0.8222,
      "step": 2261
    },
    {
      "epoch": 0.1244977709285046,
      "grad_norm": 0.7966209650039673,
      "learning_rate": 9.912430895368405e-06,
      "loss": 0.776,
      "step": 2262
    },
    {
      "epoch": 0.12455280973086026,
      "grad_norm": 0.7844830751419067,
      "learning_rate": 9.912350106701796e-06,
      "loss": 0.7513,
      "step": 2263
    },
    {
      "epoch": 0.12460784853321592,
      "grad_norm": 0.7788559794425964,
      "learning_rate": 9.912269281115278e-06,
      "loss": 0.8517,
      "step": 2264
    },
    {
      "epoch": 0.12466288733557158,
      "grad_norm": 0.778225839138031,
      "learning_rate": 9.912188418609461e-06,
      "loss": 0.7504,
      "step": 2265
    },
    {
      "epoch": 0.12471792613792723,
      "grad_norm": 0.7955968976020813,
      "learning_rate": 9.912107519184947e-06,
      "loss": 0.8152,
      "step": 2266
    },
    {
      "epoch": 0.1247729649402829,
      "grad_norm": 1.1202566623687744,
      "learning_rate": 9.912026582842352e-06,
      "loss": 0.9325,
      "step": 2267
    },
    {
      "epoch": 0.12482800374263855,
      "grad_norm": 0.9762749671936035,
      "learning_rate": 9.911945609582279e-06,
      "loss": 0.9027,
      "step": 2268
    },
    {
      "epoch": 0.12488304254499422,
      "grad_norm": 0.8311051726341248,
      "learning_rate": 9.911864599405336e-06,
      "loss": 0.838,
      "step": 2269
    },
    {
      "epoch": 0.12493808134734988,
      "grad_norm": 1.0136815309524536,
      "learning_rate": 9.911783552312134e-06,
      "loss": 0.9288,
      "step": 2270
    },
    {
      "epoch": 0.12499312014970554,
      "grad_norm": 0.7960494160652161,
      "learning_rate": 9.911702468303282e-06,
      "loss": 0.8007,
      "step": 2271
    },
    {
      "epoch": 0.1250481589520612,
      "grad_norm": 0.9980880618095398,
      "learning_rate": 9.911621347379388e-06,
      "loss": 0.8613,
      "step": 2272
    },
    {
      "epoch": 0.12510319775441686,
      "grad_norm": 0.8916807770729065,
      "learning_rate": 9.911540189541065e-06,
      "loss": 0.8783,
      "step": 2273
    },
    {
      "epoch": 0.12515823655677252,
      "grad_norm": 0.9455892443656921,
      "learning_rate": 9.911458994788919e-06,
      "loss": 0.8676,
      "step": 2274
    },
    {
      "epoch": 0.12521327535912818,
      "grad_norm": 0.7649906277656555,
      "learning_rate": 9.911377763123561e-06,
      "loss": 0.7763,
      "step": 2275
    },
    {
      "epoch": 0.12526831416148385,
      "grad_norm": 0.8971202373504639,
      "learning_rate": 9.911296494545604e-06,
      "loss": 0.9022,
      "step": 2276
    },
    {
      "epoch": 0.1253233529638395,
      "grad_norm": 0.833678126335144,
      "learning_rate": 9.911215189055657e-06,
      "loss": 0.8401,
      "step": 2277
    },
    {
      "epoch": 0.12537839176619517,
      "grad_norm": 0.8967958688735962,
      "learning_rate": 9.911133846654331e-06,
      "loss": 0.8678,
      "step": 2278
    },
    {
      "epoch": 0.12543343056855083,
      "grad_norm": 0.8195546865463257,
      "learning_rate": 9.911052467342239e-06,
      "loss": 0.842,
      "step": 2279
    },
    {
      "epoch": 0.1254884693709065,
      "grad_norm": 1.095815896987915,
      "learning_rate": 9.910971051119988e-06,
      "loss": 0.845,
      "step": 2280
    },
    {
      "epoch": 0.12554350817326215,
      "grad_norm": 0.9452629685401917,
      "learning_rate": 9.910889597988197e-06,
      "loss": 0.8971,
      "step": 2281
    },
    {
      "epoch": 0.12559854697561781,
      "grad_norm": 0.9872332215309143,
      "learning_rate": 9.910808107947471e-06,
      "loss": 0.7994,
      "step": 2282
    },
    {
      "epoch": 0.12565358577797348,
      "grad_norm": 0.7761966586112976,
      "learning_rate": 9.910726580998427e-06,
      "loss": 0.7791,
      "step": 2283
    },
    {
      "epoch": 0.12570862458032914,
      "grad_norm": 0.8950315713882446,
      "learning_rate": 9.910645017141678e-06,
      "loss": 0.8499,
      "step": 2284
    },
    {
      "epoch": 0.1257636633826848,
      "grad_norm": 0.8796371221542358,
      "learning_rate": 9.910563416377834e-06,
      "loss": 0.8587,
      "step": 2285
    },
    {
      "epoch": 0.12581870218504046,
      "grad_norm": 0.8291982412338257,
      "learning_rate": 9.91048177870751e-06,
      "loss": 0.9166,
      "step": 2286
    },
    {
      "epoch": 0.12587374098739612,
      "grad_norm": 0.758369505405426,
      "learning_rate": 9.91040010413132e-06,
      "loss": 0.8305,
      "step": 2287
    },
    {
      "epoch": 0.12592877978975178,
      "grad_norm": 0.8775640726089478,
      "learning_rate": 9.910318392649876e-06,
      "loss": 0.8513,
      "step": 2288
    },
    {
      "epoch": 0.12598381859210744,
      "grad_norm": 0.8581671118736267,
      "learning_rate": 9.910236644263796e-06,
      "loss": 0.8134,
      "step": 2289
    },
    {
      "epoch": 0.1260388573944631,
      "grad_norm": 0.8570736050605774,
      "learning_rate": 9.910154858973689e-06,
      "loss": 0.826,
      "step": 2290
    },
    {
      "epoch": 0.12609389619681877,
      "grad_norm": 0.8712487816810608,
      "learning_rate": 9.910073036780173e-06,
      "loss": 0.8042,
      "step": 2291
    },
    {
      "epoch": 0.12614893499917443,
      "grad_norm": 0.7584837675094604,
      "learning_rate": 9.909991177683862e-06,
      "loss": 0.7715,
      "step": 2292
    },
    {
      "epoch": 0.1262039738015301,
      "grad_norm": 0.8618917465209961,
      "learning_rate": 9.909909281685373e-06,
      "loss": 0.8755,
      "step": 2293
    },
    {
      "epoch": 0.12625901260388575,
      "grad_norm": 0.9530277848243713,
      "learning_rate": 9.90982734878532e-06,
      "loss": 0.8538,
      "step": 2294
    },
    {
      "epoch": 0.1263140514062414,
      "grad_norm": 0.8394436836242676,
      "learning_rate": 9.909745378984319e-06,
      "loss": 0.8401,
      "step": 2295
    },
    {
      "epoch": 0.12636909020859707,
      "grad_norm": 0.8224034309387207,
      "learning_rate": 9.909663372282984e-06,
      "loss": 0.7201,
      "step": 2296
    },
    {
      "epoch": 0.12642412901095273,
      "grad_norm": 0.8215349912643433,
      "learning_rate": 9.909581328681934e-06,
      "loss": 0.8824,
      "step": 2297
    },
    {
      "epoch": 0.12647916781330837,
      "grad_norm": 0.839389443397522,
      "learning_rate": 9.909499248181786e-06,
      "loss": 0.8056,
      "step": 2298
    },
    {
      "epoch": 0.12653420661566403,
      "grad_norm": 0.9440048933029175,
      "learning_rate": 9.909417130783156e-06,
      "loss": 0.908,
      "step": 2299
    },
    {
      "epoch": 0.1265892454180197,
      "grad_norm": 0.8336486220359802,
      "learning_rate": 9.90933497648666e-06,
      "loss": 0.8382,
      "step": 2300
    },
    {
      "epoch": 0.12664428422037535,
      "grad_norm": 1.1541366577148438,
      "learning_rate": 9.909252785292918e-06,
      "loss": 0.8782,
      "step": 2301
    },
    {
      "epoch": 0.12669932302273101,
      "grad_norm": 0.8730320334434509,
      "learning_rate": 9.909170557202545e-06,
      "loss": 0.7687,
      "step": 2302
    },
    {
      "epoch": 0.12675436182508668,
      "grad_norm": 0.9927527904510498,
      "learning_rate": 9.90908829221616e-06,
      "loss": 0.8134,
      "step": 2303
    },
    {
      "epoch": 0.12680940062744234,
      "grad_norm": 0.9521791338920593,
      "learning_rate": 9.909005990334381e-06,
      "loss": 0.9187,
      "step": 2304
    },
    {
      "epoch": 0.126864439429798,
      "grad_norm": 0.8012455701828003,
      "learning_rate": 9.908923651557828e-06,
      "loss": 0.8581,
      "step": 2305
    },
    {
      "epoch": 0.12691947823215366,
      "grad_norm": 0.8882689476013184,
      "learning_rate": 9.90884127588712e-06,
      "loss": 0.9317,
      "step": 2306
    },
    {
      "epoch": 0.12697451703450932,
      "grad_norm": 0.8408340215682983,
      "learning_rate": 9.908758863322872e-06,
      "loss": 0.8444,
      "step": 2307
    },
    {
      "epoch": 0.12702955583686498,
      "grad_norm": 0.7856307029724121,
      "learning_rate": 9.908676413865709e-06,
      "loss": 0.8457,
      "step": 2308
    },
    {
      "epoch": 0.12708459463922064,
      "grad_norm": 0.9459167718887329,
      "learning_rate": 9.908593927516247e-06,
      "loss": 0.8153,
      "step": 2309
    },
    {
      "epoch": 0.1271396334415763,
      "grad_norm": 0.8629655838012695,
      "learning_rate": 9.908511404275107e-06,
      "loss": 0.8279,
      "step": 2310
    },
    {
      "epoch": 0.12719467224393197,
      "grad_norm": 1.2012875080108643,
      "learning_rate": 9.90842884414291e-06,
      "loss": 1.4388,
      "step": 2311
    },
    {
      "epoch": 0.12724971104628763,
      "grad_norm": 1.20725417137146,
      "learning_rate": 9.908346247120274e-06,
      "loss": 0.8704,
      "step": 2312
    },
    {
      "epoch": 0.1273047498486433,
      "grad_norm": 0.8152929544448853,
      "learning_rate": 9.908263613207822e-06,
      "loss": 0.8618,
      "step": 2313
    },
    {
      "epoch": 0.12735978865099895,
      "grad_norm": 0.8400965332984924,
      "learning_rate": 9.908180942406175e-06,
      "loss": 0.7881,
      "step": 2314
    },
    {
      "epoch": 0.1274148274533546,
      "grad_norm": 0.8856974840164185,
      "learning_rate": 9.908098234715956e-06,
      "loss": 0.9073,
      "step": 2315
    },
    {
      "epoch": 0.12746986625571027,
      "grad_norm": 0.8708439469337463,
      "learning_rate": 9.908015490137782e-06,
      "loss": 0.8099,
      "step": 2316
    },
    {
      "epoch": 0.12752490505806593,
      "grad_norm": 0.8632444143295288,
      "learning_rate": 9.907932708672277e-06,
      "loss": 0.8472,
      "step": 2317
    },
    {
      "epoch": 0.1275799438604216,
      "grad_norm": 0.8977149128913879,
      "learning_rate": 9.907849890320062e-06,
      "loss": 0.8878,
      "step": 2318
    },
    {
      "epoch": 0.12763498266277726,
      "grad_norm": 0.8589425086975098,
      "learning_rate": 9.907767035081765e-06,
      "loss": 0.7905,
      "step": 2319
    },
    {
      "epoch": 0.12769002146513292,
      "grad_norm": 0.9873501062393188,
      "learning_rate": 9.907684142958002e-06,
      "loss": 0.9002,
      "step": 2320
    },
    {
      "epoch": 0.12774506026748858,
      "grad_norm": 0.8963840007781982,
      "learning_rate": 9.9076012139494e-06,
      "loss": 0.92,
      "step": 2321
    },
    {
      "epoch": 0.12780009906984424,
      "grad_norm": 0.7933574318885803,
      "learning_rate": 9.90751824805658e-06,
      "loss": 0.7664,
      "step": 2322
    },
    {
      "epoch": 0.1278551378721999,
      "grad_norm": 0.9660933017730713,
      "learning_rate": 9.907435245280167e-06,
      "loss": 0.9162,
      "step": 2323
    },
    {
      "epoch": 0.12791017667455556,
      "grad_norm": 0.8698949217796326,
      "learning_rate": 9.907352205620783e-06,
      "loss": 0.7988,
      "step": 2324
    },
    {
      "epoch": 0.12796521547691123,
      "grad_norm": 0.9077615141868591,
      "learning_rate": 9.907269129079055e-06,
      "loss": 0.8581,
      "step": 2325
    },
    {
      "epoch": 0.1280202542792669,
      "grad_norm": 0.9128179550170898,
      "learning_rate": 9.907186015655607e-06,
      "loss": 0.8552,
      "step": 2326
    },
    {
      "epoch": 0.12807529308162255,
      "grad_norm": 0.9321265816688538,
      "learning_rate": 9.907102865351062e-06,
      "loss": 0.889,
      "step": 2327
    },
    {
      "epoch": 0.1281303318839782,
      "grad_norm": 0.9687464833259583,
      "learning_rate": 9.907019678166044e-06,
      "loss": 0.7944,
      "step": 2328
    },
    {
      "epoch": 0.12818537068633387,
      "grad_norm": 0.862223207950592,
      "learning_rate": 9.90693645410118e-06,
      "loss": 0.7699,
      "step": 2329
    },
    {
      "epoch": 0.12824040948868953,
      "grad_norm": 0.9662127494812012,
      "learning_rate": 9.906853193157095e-06,
      "loss": 0.7818,
      "step": 2330
    },
    {
      "epoch": 0.1282954482910452,
      "grad_norm": 0.8008295297622681,
      "learning_rate": 9.906769895334413e-06,
      "loss": 0.8443,
      "step": 2331
    },
    {
      "epoch": 0.12835048709340086,
      "grad_norm": 0.8638464212417603,
      "learning_rate": 9.906686560633765e-06,
      "loss": 0.8438,
      "step": 2332
    },
    {
      "epoch": 0.12840552589575652,
      "grad_norm": 0.9215866327285767,
      "learning_rate": 9.906603189055773e-06,
      "loss": 0.7481,
      "step": 2333
    },
    {
      "epoch": 0.12846056469811218,
      "grad_norm": 0.7926739454269409,
      "learning_rate": 9.906519780601066e-06,
      "loss": 0.7404,
      "step": 2334
    },
    {
      "epoch": 0.12851560350046784,
      "grad_norm": 0.9590242505073547,
      "learning_rate": 9.906436335270268e-06,
      "loss": 0.8319,
      "step": 2335
    },
    {
      "epoch": 0.1285706423028235,
      "grad_norm": 1.0300076007843018,
      "learning_rate": 9.906352853064009e-06,
      "loss": 0.8635,
      "step": 2336
    },
    {
      "epoch": 0.12862568110517916,
      "grad_norm": 0.8401443958282471,
      "learning_rate": 9.906269333982915e-06,
      "loss": 0.9584,
      "step": 2337
    },
    {
      "epoch": 0.12868071990753482,
      "grad_norm": 0.8144069910049438,
      "learning_rate": 9.906185778027613e-06,
      "loss": 0.7375,
      "step": 2338
    },
    {
      "epoch": 0.12873575870989049,
      "grad_norm": 0.8513948917388916,
      "learning_rate": 9.906102185198733e-06,
      "loss": 0.8353,
      "step": 2339
    },
    {
      "epoch": 0.12879079751224615,
      "grad_norm": 0.8243077397346497,
      "learning_rate": 9.906018555496903e-06,
      "loss": 0.8665,
      "step": 2340
    },
    {
      "epoch": 0.12884583631460178,
      "grad_norm": 0.8699066042900085,
      "learning_rate": 9.905934888922749e-06,
      "loss": 0.8537,
      "step": 2341
    },
    {
      "epoch": 0.12890087511695744,
      "grad_norm": 1.0980210304260254,
      "learning_rate": 9.905851185476902e-06,
      "loss": 0.8887,
      "step": 2342
    },
    {
      "epoch": 0.1289559139193131,
      "grad_norm": 0.8189190030097961,
      "learning_rate": 9.905767445159992e-06,
      "loss": 0.8467,
      "step": 2343
    },
    {
      "epoch": 0.12901095272166876,
      "grad_norm": 0.8273541331291199,
      "learning_rate": 9.905683667972645e-06,
      "loss": 0.8701,
      "step": 2344
    },
    {
      "epoch": 0.12906599152402443,
      "grad_norm": 0.8987969160079956,
      "learning_rate": 9.905599853915496e-06,
      "loss": 0.909,
      "step": 2345
    },
    {
      "epoch": 0.1291210303263801,
      "grad_norm": 0.818268895149231,
      "learning_rate": 9.905516002989168e-06,
      "loss": 0.7946,
      "step": 2346
    },
    {
      "epoch": 0.12917606912873575,
      "grad_norm": 0.7401725053787231,
      "learning_rate": 9.905432115194296e-06,
      "loss": 0.7006,
      "step": 2347
    },
    {
      "epoch": 0.1292311079310914,
      "grad_norm": 0.8263179659843445,
      "learning_rate": 9.905348190531511e-06,
      "loss": 0.7768,
      "step": 2348
    },
    {
      "epoch": 0.12928614673344707,
      "grad_norm": 0.9241918921470642,
      "learning_rate": 9.90526422900144e-06,
      "loss": 0.8593,
      "step": 2349
    },
    {
      "epoch": 0.12934118553580273,
      "grad_norm": 0.7804501056671143,
      "learning_rate": 9.905180230604718e-06,
      "loss": 0.7607,
      "step": 2350
    },
    {
      "epoch": 0.1293962243381584,
      "grad_norm": 0.9408491253852844,
      "learning_rate": 9.905096195341973e-06,
      "loss": 0.8906,
      "step": 2351
    },
    {
      "epoch": 0.12945126314051406,
      "grad_norm": 1.0356301069259644,
      "learning_rate": 9.905012123213838e-06,
      "loss": 0.8051,
      "step": 2352
    },
    {
      "epoch": 0.12950630194286972,
      "grad_norm": 0.8546886444091797,
      "learning_rate": 9.904928014220945e-06,
      "loss": 0.7543,
      "step": 2353
    },
    {
      "epoch": 0.12956134074522538,
      "grad_norm": 0.9229897856712341,
      "learning_rate": 9.904843868363927e-06,
      "loss": 0.8823,
      "step": 2354
    },
    {
      "epoch": 0.12961637954758104,
      "grad_norm": 0.8364199995994568,
      "learning_rate": 9.904759685643414e-06,
      "loss": 0.8825,
      "step": 2355
    },
    {
      "epoch": 0.1296714183499367,
      "grad_norm": 0.9092077016830444,
      "learning_rate": 9.90467546606004e-06,
      "loss": 0.8721,
      "step": 2356
    },
    {
      "epoch": 0.12972645715229236,
      "grad_norm": 1.042973518371582,
      "learning_rate": 9.904591209614441e-06,
      "loss": 0.7984,
      "step": 2357
    },
    {
      "epoch": 0.12978149595464802,
      "grad_norm": 0.7262618541717529,
      "learning_rate": 9.904506916307243e-06,
      "loss": 0.6721,
      "step": 2358
    },
    {
      "epoch": 0.12983653475700369,
      "grad_norm": 0.7562826871871948,
      "learning_rate": 9.904422586139086e-06,
      "loss": 0.7702,
      "step": 2359
    },
    {
      "epoch": 0.12989157355935935,
      "grad_norm": 0.8821595907211304,
      "learning_rate": 9.904338219110603e-06,
      "loss": 0.8555,
      "step": 2360
    },
    {
      "epoch": 0.129946612361715,
      "grad_norm": 1.0340098142623901,
      "learning_rate": 9.904253815222424e-06,
      "loss": 0.9004,
      "step": 2361
    },
    {
      "epoch": 0.13000165116407067,
      "grad_norm": 0.8533693552017212,
      "learning_rate": 9.904169374475188e-06,
      "loss": 0.836,
      "step": 2362
    },
    {
      "epoch": 0.13005668996642633,
      "grad_norm": 0.8564199805259705,
      "learning_rate": 9.904084896869528e-06,
      "loss": 0.9281,
      "step": 2363
    },
    {
      "epoch": 0.130111728768782,
      "grad_norm": 0.7817538976669312,
      "learning_rate": 9.904000382406079e-06,
      "loss": 0.7444,
      "step": 2364
    },
    {
      "epoch": 0.13016676757113765,
      "grad_norm": 1.1420893669128418,
      "learning_rate": 9.903915831085473e-06,
      "loss": 0.9116,
      "step": 2365
    },
    {
      "epoch": 0.13022180637349332,
      "grad_norm": 0.9671920537948608,
      "learning_rate": 9.903831242908351e-06,
      "loss": 0.899,
      "step": 2366
    },
    {
      "epoch": 0.13027684517584898,
      "grad_norm": 0.8528717756271362,
      "learning_rate": 9.903746617875345e-06,
      "loss": 0.7231,
      "step": 2367
    },
    {
      "epoch": 0.13033188397820464,
      "grad_norm": 0.786960244178772,
      "learning_rate": 9.903661955987091e-06,
      "loss": 0.7997,
      "step": 2368
    },
    {
      "epoch": 0.1303869227805603,
      "grad_norm": 0.941683292388916,
      "learning_rate": 9.903577257244228e-06,
      "loss": 0.9127,
      "step": 2369
    },
    {
      "epoch": 0.13044196158291596,
      "grad_norm": 0.886900007724762,
      "learning_rate": 9.903492521647391e-06,
      "loss": 0.9086,
      "step": 2370
    },
    {
      "epoch": 0.13049700038527162,
      "grad_norm": 0.9924801588058472,
      "learning_rate": 9.903407749197216e-06,
      "loss": 0.9055,
      "step": 2371
    },
    {
      "epoch": 0.13055203918762728,
      "grad_norm": 0.6998724341392517,
      "learning_rate": 9.903322939894342e-06,
      "loss": 0.6972,
      "step": 2372
    },
    {
      "epoch": 0.13060707798998294,
      "grad_norm": 0.8448702096939087,
      "learning_rate": 9.903238093739404e-06,
      "loss": 0.7862,
      "step": 2373
    },
    {
      "epoch": 0.1306621167923386,
      "grad_norm": 0.8557441830635071,
      "learning_rate": 9.90315321073304e-06,
      "loss": 0.8364,
      "step": 2374
    },
    {
      "epoch": 0.13071715559469427,
      "grad_norm": 0.7978441119194031,
      "learning_rate": 9.903068290875892e-06,
      "loss": 0.7671,
      "step": 2375
    },
    {
      "epoch": 0.13077219439704993,
      "grad_norm": 0.781315803527832,
      "learning_rate": 9.902983334168594e-06,
      "loss": 0.7963,
      "step": 2376
    },
    {
      "epoch": 0.1308272331994056,
      "grad_norm": 0.7326155304908752,
      "learning_rate": 9.902898340611785e-06,
      "loss": 0.8,
      "step": 2377
    },
    {
      "epoch": 0.13088227200176125,
      "grad_norm": 0.7693139314651489,
      "learning_rate": 9.902813310206105e-06,
      "loss": 0.8459,
      "step": 2378
    },
    {
      "epoch": 0.1309373108041169,
      "grad_norm": 0.9441308975219727,
      "learning_rate": 9.902728242952191e-06,
      "loss": 0.8519,
      "step": 2379
    },
    {
      "epoch": 0.13099234960647257,
      "grad_norm": 0.8350616693496704,
      "learning_rate": 9.902643138850686e-06,
      "loss": 0.876,
      "step": 2380
    },
    {
      "epoch": 0.13104738840882824,
      "grad_norm": 0.8675554394721985,
      "learning_rate": 9.902557997902227e-06,
      "loss": 0.8172,
      "step": 2381
    },
    {
      "epoch": 0.1311024272111839,
      "grad_norm": 0.9618930220603943,
      "learning_rate": 9.902472820107454e-06,
      "loss": 0.8852,
      "step": 2382
    },
    {
      "epoch": 0.13115746601353956,
      "grad_norm": 0.862341046333313,
      "learning_rate": 9.902387605467007e-06,
      "loss": 0.9256,
      "step": 2383
    },
    {
      "epoch": 0.1312125048158952,
      "grad_norm": 0.8749859929084778,
      "learning_rate": 9.902302353981527e-06,
      "loss": 0.8809,
      "step": 2384
    },
    {
      "epoch": 0.13126754361825085,
      "grad_norm": 0.9061958193778992,
      "learning_rate": 9.902217065651657e-06,
      "loss": 0.779,
      "step": 2385
    },
    {
      "epoch": 0.13132258242060652,
      "grad_norm": 0.8909298777580261,
      "learning_rate": 9.902131740478033e-06,
      "loss": 0.8203,
      "step": 2386
    },
    {
      "epoch": 0.13137762122296218,
      "grad_norm": 0.8507269024848938,
      "learning_rate": 9.902046378461302e-06,
      "loss": 0.776,
      "step": 2387
    },
    {
      "epoch": 0.13143266002531784,
      "grad_norm": 0.9577299356460571,
      "learning_rate": 9.901960979602101e-06,
      "loss": 0.8104,
      "step": 2388
    },
    {
      "epoch": 0.1314876988276735,
      "grad_norm": 0.9244948625564575,
      "learning_rate": 9.901875543901074e-06,
      "loss": 0.9035,
      "step": 2389
    },
    {
      "epoch": 0.13154273763002916,
      "grad_norm": 0.7534334063529968,
      "learning_rate": 9.901790071358861e-06,
      "loss": 0.7262,
      "step": 2390
    },
    {
      "epoch": 0.13159777643238482,
      "grad_norm": 0.8920090198516846,
      "learning_rate": 9.901704561976106e-06,
      "loss": 0.932,
      "step": 2391
    },
    {
      "epoch": 0.13165281523474048,
      "grad_norm": 0.8524243235588074,
      "learning_rate": 9.901619015753455e-06,
      "loss": 0.8107,
      "step": 2392
    },
    {
      "epoch": 0.13170785403709614,
      "grad_norm": 0.8170381784439087,
      "learning_rate": 9.901533432691543e-06,
      "loss": 0.8814,
      "step": 2393
    },
    {
      "epoch": 0.1317628928394518,
      "grad_norm": 0.8281697034835815,
      "learning_rate": 9.90144781279102e-06,
      "loss": 0.8221,
      "step": 2394
    },
    {
      "epoch": 0.13181793164180747,
      "grad_norm": 0.9283351302146912,
      "learning_rate": 9.901362156052528e-06,
      "loss": 0.8346,
      "step": 2395
    },
    {
      "epoch": 0.13187297044416313,
      "grad_norm": 0.8331275582313538,
      "learning_rate": 9.901276462476708e-06,
      "loss": 0.7498,
      "step": 2396
    },
    {
      "epoch": 0.1319280092465188,
      "grad_norm": 0.8427191972732544,
      "learning_rate": 9.901190732064207e-06,
      "loss": 0.8265,
      "step": 2397
    },
    {
      "epoch": 0.13198304804887445,
      "grad_norm": 0.8510351777076721,
      "learning_rate": 9.901104964815669e-06,
      "loss": 0.8369,
      "step": 2398
    },
    {
      "epoch": 0.1320380868512301,
      "grad_norm": 0.8468914031982422,
      "learning_rate": 9.901019160731738e-06,
      "loss": 0.8585,
      "step": 2399
    },
    {
      "epoch": 0.13209312565358577,
      "grad_norm": 0.8302182555198669,
      "learning_rate": 9.900933319813058e-06,
      "loss": 0.8611,
      "step": 2400
    },
    {
      "epoch": 0.13214816445594144,
      "grad_norm": 0.8527448773384094,
      "learning_rate": 9.900847442060277e-06,
      "loss": 0.899,
      "step": 2401
    },
    {
      "epoch": 0.1322032032582971,
      "grad_norm": 0.8354688286781311,
      "learning_rate": 9.900761527474037e-06,
      "loss": 0.8083,
      "step": 2402
    },
    {
      "epoch": 0.13225824206065276,
      "grad_norm": 0.8612173795700073,
      "learning_rate": 9.900675576054986e-06,
      "loss": 0.8124,
      "step": 2403
    },
    {
      "epoch": 0.13231328086300842,
      "grad_norm": 0.7424876689910889,
      "learning_rate": 9.900589587803767e-06,
      "loss": 0.6884,
      "step": 2404
    },
    {
      "epoch": 0.13236831966536408,
      "grad_norm": 0.8431115746498108,
      "learning_rate": 9.90050356272103e-06,
      "loss": 0.9575,
      "step": 2405
    },
    {
      "epoch": 0.13242335846771974,
      "grad_norm": 0.7958092093467712,
      "learning_rate": 9.90041750080742e-06,
      "loss": 0.7608,
      "step": 2406
    },
    {
      "epoch": 0.1324783972700754,
      "grad_norm": 0.926258385181427,
      "learning_rate": 9.900331402063583e-06,
      "loss": 0.9072,
      "step": 2407
    },
    {
      "epoch": 0.13253343607243107,
      "grad_norm": 0.7952526807785034,
      "learning_rate": 9.900245266490169e-06,
      "loss": 0.8001,
      "step": 2408
    },
    {
      "epoch": 0.13258847487478673,
      "grad_norm": 0.8309933543205261,
      "learning_rate": 9.900159094087822e-06,
      "loss": 0.9154,
      "step": 2409
    },
    {
      "epoch": 0.1326435136771424,
      "grad_norm": 0.858007550239563,
      "learning_rate": 9.90007288485719e-06,
      "loss": 0.855,
      "step": 2410
    },
    {
      "epoch": 0.13269855247949805,
      "grad_norm": 0.9513822197914124,
      "learning_rate": 9.899986638798923e-06,
      "loss": 0.8162,
      "step": 2411
    },
    {
      "epoch": 0.1327535912818537,
      "grad_norm": 0.8387427926063538,
      "learning_rate": 9.899900355913668e-06,
      "loss": 0.8955,
      "step": 2412
    },
    {
      "epoch": 0.13280863008420937,
      "grad_norm": 0.7727940678596497,
      "learning_rate": 9.899814036202073e-06,
      "loss": 0.6765,
      "step": 2413
    },
    {
      "epoch": 0.13286366888656503,
      "grad_norm": 0.7760928869247437,
      "learning_rate": 9.899727679664788e-06,
      "loss": 0.7179,
      "step": 2414
    },
    {
      "epoch": 0.1329187076889207,
      "grad_norm": 0.7798073887825012,
      "learning_rate": 9.899641286302462e-06,
      "loss": 0.8541,
      "step": 2415
    },
    {
      "epoch": 0.13297374649127636,
      "grad_norm": 0.8302769660949707,
      "learning_rate": 9.899554856115743e-06,
      "loss": 0.8925,
      "step": 2416
    },
    {
      "epoch": 0.13302878529363202,
      "grad_norm": 0.8300751447677612,
      "learning_rate": 9.89946838910528e-06,
      "loss": 0.7489,
      "step": 2417
    },
    {
      "epoch": 0.13308382409598768,
      "grad_norm": 0.8032094240188599,
      "learning_rate": 9.899381885271725e-06,
      "loss": 0.811,
      "step": 2418
    },
    {
      "epoch": 0.13313886289834334,
      "grad_norm": 5.237870216369629,
      "learning_rate": 9.899295344615727e-06,
      "loss": 0.7609,
      "step": 2419
    },
    {
      "epoch": 0.133193901700699,
      "grad_norm": 0.8145740628242493,
      "learning_rate": 9.899208767137935e-06,
      "loss": 0.8435,
      "step": 2420
    },
    {
      "epoch": 0.13324894050305466,
      "grad_norm": 0.9716018438339233,
      "learning_rate": 9.899122152839004e-06,
      "loss": 0.7924,
      "step": 2421
    },
    {
      "epoch": 0.13330397930541033,
      "grad_norm": 0.7846183776855469,
      "learning_rate": 9.899035501719582e-06,
      "loss": 0.8941,
      "step": 2422
    },
    {
      "epoch": 0.133359018107766,
      "grad_norm": 0.7653689980506897,
      "learning_rate": 9.89894881378032e-06,
      "loss": 0.811,
      "step": 2423
    },
    {
      "epoch": 0.13341405691012165,
      "grad_norm": 0.8221875429153442,
      "learning_rate": 9.89886208902187e-06,
      "loss": 0.8131,
      "step": 2424
    },
    {
      "epoch": 0.1334690957124773,
      "grad_norm": 0.7422335147857666,
      "learning_rate": 9.898775327444885e-06,
      "loss": 0.6366,
      "step": 2425
    },
    {
      "epoch": 0.13352413451483297,
      "grad_norm": 0.8072695136070251,
      "learning_rate": 9.898688529050014e-06,
      "loss": 0.7989,
      "step": 2426
    },
    {
      "epoch": 0.1335791733171886,
      "grad_norm": 0.7717600464820862,
      "learning_rate": 9.898601693837911e-06,
      "loss": 0.7524,
      "step": 2427
    },
    {
      "epoch": 0.13363421211954427,
      "grad_norm": 0.8070919513702393,
      "learning_rate": 9.898514821809231e-06,
      "loss": 0.7724,
      "step": 2428
    },
    {
      "epoch": 0.13368925092189993,
      "grad_norm": 0.8184726238250732,
      "learning_rate": 9.898427912964624e-06,
      "loss": 0.845,
      "step": 2429
    },
    {
      "epoch": 0.1337442897242556,
      "grad_norm": 0.8168759346008301,
      "learning_rate": 9.898340967304744e-06,
      "loss": 0.8377,
      "step": 2430
    },
    {
      "epoch": 0.13379932852661125,
      "grad_norm": 0.8701872825622559,
      "learning_rate": 9.898253984830244e-06,
      "loss": 0.908,
      "step": 2431
    },
    {
      "epoch": 0.1338543673289669,
      "grad_norm": 0.8092133402824402,
      "learning_rate": 9.898166965541779e-06,
      "loss": 0.866,
      "step": 2432
    },
    {
      "epoch": 0.13390940613132257,
      "grad_norm": 0.8337095975875854,
      "learning_rate": 9.898079909440002e-06,
      "loss": 0.8622,
      "step": 2433
    },
    {
      "epoch": 0.13396444493367823,
      "grad_norm": 1.1016209125518799,
      "learning_rate": 9.897992816525567e-06,
      "loss": 0.8486,
      "step": 2434
    },
    {
      "epoch": 0.1340194837360339,
      "grad_norm": 0.8136518597602844,
      "learning_rate": 9.89790568679913e-06,
      "loss": 0.8681,
      "step": 2435
    },
    {
      "epoch": 0.13407452253838956,
      "grad_norm": 0.8202341794967651,
      "learning_rate": 9.897818520261344e-06,
      "loss": 0.9144,
      "step": 2436
    },
    {
      "epoch": 0.13412956134074522,
      "grad_norm": 0.8836861848831177,
      "learning_rate": 9.897731316912866e-06,
      "loss": 0.8643,
      "step": 2437
    },
    {
      "epoch": 0.13418460014310088,
      "grad_norm": 0.9040210247039795,
      "learning_rate": 9.89764407675435e-06,
      "loss": 0.7681,
      "step": 2438
    },
    {
      "epoch": 0.13423963894545654,
      "grad_norm": 0.8762359619140625,
      "learning_rate": 9.897556799786452e-06,
      "loss": 0.8765,
      "step": 2439
    },
    {
      "epoch": 0.1342946777478122,
      "grad_norm": 0.8859462738037109,
      "learning_rate": 9.897469486009827e-06,
      "loss": 0.9051,
      "step": 2440
    },
    {
      "epoch": 0.13434971655016786,
      "grad_norm": 0.7727539539337158,
      "learning_rate": 9.897382135425134e-06,
      "loss": 0.7397,
      "step": 2441
    },
    {
      "epoch": 0.13440475535252353,
      "grad_norm": 0.9018967151641846,
      "learning_rate": 9.897294748033028e-06,
      "loss": 0.8542,
      "step": 2442
    },
    {
      "epoch": 0.1344597941548792,
      "grad_norm": 0.8228337168693542,
      "learning_rate": 9.897207323834165e-06,
      "loss": 0.7585,
      "step": 2443
    },
    {
      "epoch": 0.13451483295723485,
      "grad_norm": 0.7509974241256714,
      "learning_rate": 9.897119862829203e-06,
      "loss": 0.7285,
      "step": 2444
    },
    {
      "epoch": 0.1345698717595905,
      "grad_norm": 0.9225835800170898,
      "learning_rate": 9.897032365018797e-06,
      "loss": 0.8352,
      "step": 2445
    },
    {
      "epoch": 0.13462491056194617,
      "grad_norm": 0.800981879234314,
      "learning_rate": 9.896944830403609e-06,
      "loss": 0.7352,
      "step": 2446
    },
    {
      "epoch": 0.13467994936430183,
      "grad_norm": 0.8263673186302185,
      "learning_rate": 9.896857258984294e-06,
      "loss": 0.8426,
      "step": 2447
    },
    {
      "epoch": 0.1347349881666575,
      "grad_norm": 0.8857110738754272,
      "learning_rate": 9.89676965076151e-06,
      "loss": 0.8078,
      "step": 2448
    },
    {
      "epoch": 0.13479002696901315,
      "grad_norm": 0.8637158274650574,
      "learning_rate": 9.896682005735916e-06,
      "loss": 0.8688,
      "step": 2449
    },
    {
      "epoch": 0.13484506577136882,
      "grad_norm": 0.9050095081329346,
      "learning_rate": 9.89659432390817e-06,
      "loss": 0.831,
      "step": 2450
    },
    {
      "epoch": 0.13490010457372448,
      "grad_norm": 0.829757034778595,
      "learning_rate": 9.896506605278933e-06,
      "loss": 0.8095,
      "step": 2451
    },
    {
      "epoch": 0.13495514337608014,
      "grad_norm": 0.8910449743270874,
      "learning_rate": 9.896418849848864e-06,
      "loss": 0.9134,
      "step": 2452
    },
    {
      "epoch": 0.1350101821784358,
      "grad_norm": 0.8856307864189148,
      "learning_rate": 9.89633105761862e-06,
      "loss": 0.8171,
      "step": 2453
    },
    {
      "epoch": 0.13506522098079146,
      "grad_norm": 0.8159938454627991,
      "learning_rate": 9.896243228588864e-06,
      "loss": 0.8205,
      "step": 2454
    },
    {
      "epoch": 0.13512025978314712,
      "grad_norm": 0.8200929760932922,
      "learning_rate": 9.896155362760254e-06,
      "loss": 0.7529,
      "step": 2455
    },
    {
      "epoch": 0.13517529858550278,
      "grad_norm": 0.7591279149055481,
      "learning_rate": 9.89606746013345e-06,
      "loss": 0.8205,
      "step": 2456
    },
    {
      "epoch": 0.13523033738785845,
      "grad_norm": 0.8598676323890686,
      "learning_rate": 9.895979520709114e-06,
      "loss": 0.8212,
      "step": 2457
    },
    {
      "epoch": 0.1352853761902141,
      "grad_norm": 0.7290365099906921,
      "learning_rate": 9.895891544487905e-06,
      "loss": 0.7893,
      "step": 2458
    },
    {
      "epoch": 0.13534041499256977,
      "grad_norm": 0.8040594458580017,
      "learning_rate": 9.895803531470487e-06,
      "loss": 0.8358,
      "step": 2459
    },
    {
      "epoch": 0.13539545379492543,
      "grad_norm": 0.9286525249481201,
      "learning_rate": 9.895715481657522e-06,
      "loss": 0.8104,
      "step": 2460
    },
    {
      "epoch": 0.1354504925972811,
      "grad_norm": 0.843054473400116,
      "learning_rate": 9.895627395049668e-06,
      "loss": 0.7872,
      "step": 2461
    },
    {
      "epoch": 0.13550553139963675,
      "grad_norm": 0.7894387245178223,
      "learning_rate": 9.895539271647588e-06,
      "loss": 0.8615,
      "step": 2462
    },
    {
      "epoch": 0.13556057020199241,
      "grad_norm": 0.9185294508934021,
      "learning_rate": 9.895451111451948e-06,
      "loss": 0.8732,
      "step": 2463
    },
    {
      "epoch": 0.13561560900434808,
      "grad_norm": 0.8586474657058716,
      "learning_rate": 9.895362914463405e-06,
      "loss": 0.9658,
      "step": 2464
    },
    {
      "epoch": 0.13567064780670374,
      "grad_norm": 0.8810474276542664,
      "learning_rate": 9.895274680682628e-06,
      "loss": 0.8622,
      "step": 2465
    },
    {
      "epoch": 0.1357256866090594,
      "grad_norm": 0.8862990736961365,
      "learning_rate": 9.895186410110273e-06,
      "loss": 0.916,
      "step": 2466
    },
    {
      "epoch": 0.13578072541141506,
      "grad_norm": 0.7916743159294128,
      "learning_rate": 9.89509810274701e-06,
      "loss": 0.837,
      "step": 2467
    },
    {
      "epoch": 0.13583576421377072,
      "grad_norm": 0.9063515663146973,
      "learning_rate": 9.8950097585935e-06,
      "loss": 0.8065,
      "step": 2468
    },
    {
      "epoch": 0.13589080301612638,
      "grad_norm": 0.7656043767929077,
      "learning_rate": 9.894921377650405e-06,
      "loss": 0.7064,
      "step": 2469
    },
    {
      "epoch": 0.13594584181848202,
      "grad_norm": 1.0630278587341309,
      "learning_rate": 9.894832959918392e-06,
      "loss": 0.8168,
      "step": 2470
    },
    {
      "epoch": 0.13600088062083768,
      "grad_norm": 0.9118956923484802,
      "learning_rate": 9.894744505398126e-06,
      "loss": 0.8972,
      "step": 2471
    },
    {
      "epoch": 0.13605591942319334,
      "grad_norm": 0.8989213705062866,
      "learning_rate": 9.89465601409027e-06,
      "loss": 0.8374,
      "step": 2472
    },
    {
      "epoch": 0.136110958225549,
      "grad_norm": 0.9398229718208313,
      "learning_rate": 9.894567485995489e-06,
      "loss": 0.8956,
      "step": 2473
    },
    {
      "epoch": 0.13616599702790466,
      "grad_norm": 0.7980280518531799,
      "learning_rate": 9.894478921114449e-06,
      "loss": 0.8055,
      "step": 2474
    },
    {
      "epoch": 0.13622103583026032,
      "grad_norm": 0.8910034894943237,
      "learning_rate": 9.894390319447816e-06,
      "loss": 0.8371,
      "step": 2475
    },
    {
      "epoch": 0.13627607463261598,
      "grad_norm": 0.7848070859909058,
      "learning_rate": 9.894301680996255e-06,
      "loss": 0.8024,
      "step": 2476
    },
    {
      "epoch": 0.13633111343497165,
      "grad_norm": 0.8538175821304321,
      "learning_rate": 9.894213005760434e-06,
      "loss": 0.8819,
      "step": 2477
    },
    {
      "epoch": 0.1363861522373273,
      "grad_norm": 0.7885367274284363,
      "learning_rate": 9.894124293741017e-06,
      "loss": 0.7916,
      "step": 2478
    },
    {
      "epoch": 0.13644119103968297,
      "grad_norm": 0.8555673956871033,
      "learning_rate": 9.894035544938672e-06,
      "loss": 0.8521,
      "step": 2479
    },
    {
      "epoch": 0.13649622984203863,
      "grad_norm": 0.8104771971702576,
      "learning_rate": 9.893946759354066e-06,
      "loss": 0.8437,
      "step": 2480
    },
    {
      "epoch": 0.1365512686443943,
      "grad_norm": 0.9131864309310913,
      "learning_rate": 9.893857936987866e-06,
      "loss": 0.8123,
      "step": 2481
    },
    {
      "epoch": 0.13660630744674995,
      "grad_norm": 0.9414293766021729,
      "learning_rate": 9.893769077840739e-06,
      "loss": 0.7897,
      "step": 2482
    },
    {
      "epoch": 0.13666134624910561,
      "grad_norm": 0.823265016078949,
      "learning_rate": 9.893680181913355e-06,
      "loss": 0.847,
      "step": 2483
    },
    {
      "epoch": 0.13671638505146128,
      "grad_norm": 0.82098788022995,
      "learning_rate": 9.89359124920638e-06,
      "loss": 0.7823,
      "step": 2484
    },
    {
      "epoch": 0.13677142385381694,
      "grad_norm": 0.817551851272583,
      "learning_rate": 9.893502279720483e-06,
      "loss": 0.8084,
      "step": 2485
    },
    {
      "epoch": 0.1368264626561726,
      "grad_norm": 1.0722150802612305,
      "learning_rate": 9.893413273456333e-06,
      "loss": 0.7394,
      "step": 2486
    },
    {
      "epoch": 0.13688150145852826,
      "grad_norm": 0.8045433759689331,
      "learning_rate": 9.893324230414598e-06,
      "loss": 0.7528,
      "step": 2487
    },
    {
      "epoch": 0.13693654026088392,
      "grad_norm": 0.8694071173667908,
      "learning_rate": 9.893235150595949e-06,
      "loss": 0.803,
      "step": 2488
    },
    {
      "epoch": 0.13699157906323958,
      "grad_norm": 0.8238615989685059,
      "learning_rate": 9.893146034001054e-06,
      "loss": 0.7909,
      "step": 2489
    },
    {
      "epoch": 0.13704661786559524,
      "grad_norm": 0.7782405018806458,
      "learning_rate": 9.893056880630583e-06,
      "loss": 0.6859,
      "step": 2490
    },
    {
      "epoch": 0.1371016566679509,
      "grad_norm": 0.7865599989891052,
      "learning_rate": 9.892967690485207e-06,
      "loss": 0.7982,
      "step": 2491
    },
    {
      "epoch": 0.13715669547030657,
      "grad_norm": 0.768120288848877,
      "learning_rate": 9.892878463565595e-06,
      "loss": 0.8234,
      "step": 2492
    },
    {
      "epoch": 0.13721173427266223,
      "grad_norm": 0.812493085861206,
      "learning_rate": 9.89278919987242e-06,
      "loss": 0.9152,
      "step": 2493
    },
    {
      "epoch": 0.1372667730750179,
      "grad_norm": 0.7256335616111755,
      "learning_rate": 9.892699899406348e-06,
      "loss": 0.6703,
      "step": 2494
    },
    {
      "epoch": 0.13732181187737355,
      "grad_norm": 0.8022804260253906,
      "learning_rate": 9.892610562168054e-06,
      "loss": 0.7918,
      "step": 2495
    },
    {
      "epoch": 0.1373768506797292,
      "grad_norm": 0.8204907774925232,
      "learning_rate": 9.89252118815821e-06,
      "loss": 0.9094,
      "step": 2496
    },
    {
      "epoch": 0.13743188948208487,
      "grad_norm": 0.9986788630485535,
      "learning_rate": 9.892431777377484e-06,
      "loss": 0.8921,
      "step": 2497
    },
    {
      "epoch": 0.13748692828444053,
      "grad_norm": 0.7937983870506287,
      "learning_rate": 9.892342329826554e-06,
      "loss": 0.8048,
      "step": 2498
    },
    {
      "epoch": 0.1375419670867962,
      "grad_norm": 0.9295744895935059,
      "learning_rate": 9.892252845506086e-06,
      "loss": 0.755,
      "step": 2499
    },
    {
      "epoch": 0.13759700588915186,
      "grad_norm": 0.7920984625816345,
      "learning_rate": 9.892163324416757e-06,
      "loss": 0.7603,
      "step": 2500
    },
    {
      "epoch": 0.13765204469150752,
      "grad_norm": 0.9229464530944824,
      "learning_rate": 9.892073766559236e-06,
      "loss": 0.8115,
      "step": 2501
    },
    {
      "epoch": 0.13770708349386318,
      "grad_norm": 0.8205353021621704,
      "learning_rate": 9.8919841719342e-06,
      "loss": 0.8357,
      "step": 2502
    },
    {
      "epoch": 0.13776212229621884,
      "grad_norm": 0.86461341381073,
      "learning_rate": 9.891894540542318e-06,
      "loss": 0.748,
      "step": 2503
    },
    {
      "epoch": 0.1378171610985745,
      "grad_norm": 0.767145574092865,
      "learning_rate": 9.891804872384267e-06,
      "loss": 0.7404,
      "step": 2504
    },
    {
      "epoch": 0.13787219990093016,
      "grad_norm": 0.7492040991783142,
      "learning_rate": 9.891715167460721e-06,
      "loss": 0.6958,
      "step": 2505
    },
    {
      "epoch": 0.13792723870328583,
      "grad_norm": 0.8643150329589844,
      "learning_rate": 9.891625425772353e-06,
      "loss": 0.8408,
      "step": 2506
    },
    {
      "epoch": 0.1379822775056415,
      "grad_norm": 0.8026981353759766,
      "learning_rate": 9.891535647319838e-06,
      "loss": 0.7895,
      "step": 2507
    },
    {
      "epoch": 0.13803731630799715,
      "grad_norm": 1.2780394554138184,
      "learning_rate": 9.89144583210385e-06,
      "loss": 0.9113,
      "step": 2508
    },
    {
      "epoch": 0.1380923551103528,
      "grad_norm": 0.8476191163063049,
      "learning_rate": 9.891355980125064e-06,
      "loss": 0.8224,
      "step": 2509
    },
    {
      "epoch": 0.13814739391270847,
      "grad_norm": 1.048682689666748,
      "learning_rate": 9.891266091384157e-06,
      "loss": 0.8913,
      "step": 2510
    },
    {
      "epoch": 0.13820243271506413,
      "grad_norm": 1.0314993858337402,
      "learning_rate": 9.891176165881801e-06,
      "loss": 0.8315,
      "step": 2511
    },
    {
      "epoch": 0.1382574715174198,
      "grad_norm": 0.9500058889389038,
      "learning_rate": 9.891086203618676e-06,
      "loss": 0.9185,
      "step": 2512
    },
    {
      "epoch": 0.13831251031977543,
      "grad_norm": 0.7860653400421143,
      "learning_rate": 9.890996204595457e-06,
      "loss": 0.804,
      "step": 2513
    },
    {
      "epoch": 0.1383675491221311,
      "grad_norm": 0.8354741930961609,
      "learning_rate": 9.89090616881282e-06,
      "loss": 0.8214,
      "step": 2514
    },
    {
      "epoch": 0.13842258792448675,
      "grad_norm": 0.9115905165672302,
      "learning_rate": 9.890816096271438e-06,
      "loss": 0.8801,
      "step": 2515
    },
    {
      "epoch": 0.1384776267268424,
      "grad_norm": 0.8852075338363647,
      "learning_rate": 9.890725986971994e-06,
      "loss": 0.8821,
      "step": 2516
    },
    {
      "epoch": 0.13853266552919807,
      "grad_norm": 0.804314374923706,
      "learning_rate": 9.890635840915164e-06,
      "loss": 0.8412,
      "step": 2517
    },
    {
      "epoch": 0.13858770433155373,
      "grad_norm": 0.8242805600166321,
      "learning_rate": 9.890545658101623e-06,
      "loss": 0.8447,
      "step": 2518
    },
    {
      "epoch": 0.1386427431339094,
      "grad_norm": 0.8385655879974365,
      "learning_rate": 9.890455438532048e-06,
      "loss": 0.8161,
      "step": 2519
    },
    {
      "epoch": 0.13869778193626506,
      "grad_norm": 0.7950524687767029,
      "learning_rate": 9.89036518220712e-06,
      "loss": 0.8024,
      "step": 2520
    },
    {
      "epoch": 0.13875282073862072,
      "grad_norm": 1.0031861066818237,
      "learning_rate": 9.890274889127518e-06,
      "loss": 0.8399,
      "step": 2521
    },
    {
      "epoch": 0.13880785954097638,
      "grad_norm": 0.8403242230415344,
      "learning_rate": 9.890184559293917e-06,
      "loss": 0.8115,
      "step": 2522
    },
    {
      "epoch": 0.13886289834333204,
      "grad_norm": 0.8389976024627686,
      "learning_rate": 9.890094192706998e-06,
      "loss": 0.9573,
      "step": 2523
    },
    {
      "epoch": 0.1389179371456877,
      "grad_norm": 0.8408516645431519,
      "learning_rate": 9.890003789367442e-06,
      "loss": 0.8572,
      "step": 2524
    },
    {
      "epoch": 0.13897297594804336,
      "grad_norm": 0.7607787251472473,
      "learning_rate": 9.889913349275925e-06,
      "loss": 0.8119,
      "step": 2525
    },
    {
      "epoch": 0.13902801475039903,
      "grad_norm": 0.7696373462677002,
      "learning_rate": 9.889822872433127e-06,
      "loss": 0.8287,
      "step": 2526
    },
    {
      "epoch": 0.1390830535527547,
      "grad_norm": 0.8518380522727966,
      "learning_rate": 9.889732358839732e-06,
      "loss": 0.9008,
      "step": 2527
    },
    {
      "epoch": 0.13913809235511035,
      "grad_norm": 0.8851314783096313,
      "learning_rate": 9.889641808496416e-06,
      "loss": 0.8148,
      "step": 2528
    },
    {
      "epoch": 0.139193131157466,
      "grad_norm": 0.9245797395706177,
      "learning_rate": 9.889551221403862e-06,
      "loss": 0.846,
      "step": 2529
    },
    {
      "epoch": 0.13924816995982167,
      "grad_norm": 0.8445762991905212,
      "learning_rate": 9.889460597562748e-06,
      "loss": 0.8306,
      "step": 2530
    },
    {
      "epoch": 0.13930320876217733,
      "grad_norm": 0.9149277806282043,
      "learning_rate": 9.88936993697376e-06,
      "loss": 0.8033,
      "step": 2531
    },
    {
      "epoch": 0.139358247564533,
      "grad_norm": 0.894666850566864,
      "learning_rate": 9.889279239637572e-06,
      "loss": 0.8299,
      "step": 2532
    },
    {
      "epoch": 0.13941328636688866,
      "grad_norm": 1.2897371053695679,
      "learning_rate": 9.889188505554871e-06,
      "loss": 0.7776,
      "step": 2533
    },
    {
      "epoch": 0.13946832516924432,
      "grad_norm": 0.8927022218704224,
      "learning_rate": 9.889097734726341e-06,
      "loss": 0.8706,
      "step": 2534
    },
    {
      "epoch": 0.13952336397159998,
      "grad_norm": 0.7688571214675903,
      "learning_rate": 9.889006927152658e-06,
      "loss": 0.8191,
      "step": 2535
    },
    {
      "epoch": 0.13957840277395564,
      "grad_norm": 0.926671028137207,
      "learning_rate": 9.88891608283451e-06,
      "loss": 0.7489,
      "step": 2536
    },
    {
      "epoch": 0.1396334415763113,
      "grad_norm": 0.8316965699195862,
      "learning_rate": 9.888825201772577e-06,
      "loss": 0.7783,
      "step": 2537
    },
    {
      "epoch": 0.13968848037866696,
      "grad_norm": 0.8619750738143921,
      "learning_rate": 9.88873428396754e-06,
      "loss": 0.8269,
      "step": 2538
    },
    {
      "epoch": 0.13974351918102262,
      "grad_norm": 0.8588540554046631,
      "learning_rate": 9.888643329420086e-06,
      "loss": 0.8133,
      "step": 2539
    },
    {
      "epoch": 0.13979855798337829,
      "grad_norm": 0.7947841882705688,
      "learning_rate": 9.8885523381309e-06,
      "loss": 0.8041,
      "step": 2540
    },
    {
      "epoch": 0.13985359678573395,
      "grad_norm": 0.8440257906913757,
      "learning_rate": 9.888461310100661e-06,
      "loss": 0.8324,
      "step": 2541
    },
    {
      "epoch": 0.1399086355880896,
      "grad_norm": 0.7842260003089905,
      "learning_rate": 9.888370245330055e-06,
      "loss": 0.8031,
      "step": 2542
    },
    {
      "epoch": 0.13996367439044527,
      "grad_norm": 0.8108223080635071,
      "learning_rate": 9.888279143819768e-06,
      "loss": 0.7998,
      "step": 2543
    },
    {
      "epoch": 0.14001871319280093,
      "grad_norm": 0.9748625159263611,
      "learning_rate": 9.888188005570482e-06,
      "loss": 0.9553,
      "step": 2544
    },
    {
      "epoch": 0.1400737519951566,
      "grad_norm": 0.8465562462806702,
      "learning_rate": 9.888096830582883e-06,
      "loss": 0.7884,
      "step": 2545
    },
    {
      "epoch": 0.14012879079751225,
      "grad_norm": 0.9339833855628967,
      "learning_rate": 9.88800561885766e-06,
      "loss": 0.8135,
      "step": 2546
    },
    {
      "epoch": 0.14018382959986792,
      "grad_norm": 0.7749297022819519,
      "learning_rate": 9.887914370395492e-06,
      "loss": 0.8411,
      "step": 2547
    },
    {
      "epoch": 0.14023886840222358,
      "grad_norm": 0.862606942653656,
      "learning_rate": 9.887823085197068e-06,
      "loss": 0.7631,
      "step": 2548
    },
    {
      "epoch": 0.14029390720457924,
      "grad_norm": 1.3383793830871582,
      "learning_rate": 9.887731763263076e-06,
      "loss": 0.7979,
      "step": 2549
    },
    {
      "epoch": 0.1403489460069349,
      "grad_norm": 0.8092008233070374,
      "learning_rate": 9.887640404594199e-06,
      "loss": 0.7566,
      "step": 2550
    },
    {
      "epoch": 0.14040398480929056,
      "grad_norm": 0.9233745336532593,
      "learning_rate": 9.887549009191126e-06,
      "loss": 0.8954,
      "step": 2551
    },
    {
      "epoch": 0.14045902361164622,
      "grad_norm": 0.8533664345741272,
      "learning_rate": 9.887457577054542e-06,
      "loss": 0.8311,
      "step": 2552
    },
    {
      "epoch": 0.14051406241400188,
      "grad_norm": 0.7679287791252136,
      "learning_rate": 9.887366108185135e-06,
      "loss": 0.7641,
      "step": 2553
    },
    {
      "epoch": 0.14056910121635754,
      "grad_norm": 0.7998354434967041,
      "learning_rate": 9.887274602583594e-06,
      "loss": 0.7759,
      "step": 2554
    },
    {
      "epoch": 0.1406241400187132,
      "grad_norm": 0.8877138495445251,
      "learning_rate": 9.887183060250605e-06,
      "loss": 0.8928,
      "step": 2555
    },
    {
      "epoch": 0.14067917882106884,
      "grad_norm": 0.8022066354751587,
      "learning_rate": 9.887091481186855e-06,
      "loss": 0.8233,
      "step": 2556
    },
    {
      "epoch": 0.1407342176234245,
      "grad_norm": 0.8419097065925598,
      "learning_rate": 9.886999865393035e-06,
      "loss": 0.8044,
      "step": 2557
    },
    {
      "epoch": 0.14078925642578016,
      "grad_norm": 0.9581286311149597,
      "learning_rate": 9.88690821286983e-06,
      "loss": 0.8531,
      "step": 2558
    },
    {
      "epoch": 0.14084429522813582,
      "grad_norm": 0.894851803779602,
      "learning_rate": 9.886816523617933e-06,
      "loss": 0.8594,
      "step": 2559
    },
    {
      "epoch": 0.14089933403049149,
      "grad_norm": 0.7813432812690735,
      "learning_rate": 9.886724797638032e-06,
      "loss": 0.7311,
      "step": 2560
    },
    {
      "epoch": 0.14095437283284715,
      "grad_norm": 0.8194118142127991,
      "learning_rate": 9.886633034930814e-06,
      "loss": 0.8067,
      "step": 2561
    },
    {
      "epoch": 0.1410094116352028,
      "grad_norm": 0.8091121912002563,
      "learning_rate": 9.88654123549697e-06,
      "loss": 0.7558,
      "step": 2562
    },
    {
      "epoch": 0.14106445043755847,
      "grad_norm": 0.8334764242172241,
      "learning_rate": 9.88644939933719e-06,
      "loss": 0.8375,
      "step": 2563
    },
    {
      "epoch": 0.14111948923991413,
      "grad_norm": 0.8283817768096924,
      "learning_rate": 9.886357526452166e-06,
      "loss": 0.7839,
      "step": 2564
    },
    {
      "epoch": 0.1411745280422698,
      "grad_norm": 0.8708772659301758,
      "learning_rate": 9.886265616842585e-06,
      "loss": 0.8193,
      "step": 2565
    },
    {
      "epoch": 0.14122956684462545,
      "grad_norm": 0.9883641600608826,
      "learning_rate": 9.886173670509141e-06,
      "loss": 0.9409,
      "step": 2566
    },
    {
      "epoch": 0.14128460564698112,
      "grad_norm": 0.8601766228675842,
      "learning_rate": 9.886081687452523e-06,
      "loss": 0.9391,
      "step": 2567
    },
    {
      "epoch": 0.14133964444933678,
      "grad_norm": 0.8729620575904846,
      "learning_rate": 9.885989667673422e-06,
      "loss": 0.8372,
      "step": 2568
    },
    {
      "epoch": 0.14139468325169244,
      "grad_norm": 0.7899564504623413,
      "learning_rate": 9.885897611172532e-06,
      "loss": 0.7773,
      "step": 2569
    },
    {
      "epoch": 0.1414497220540481,
      "grad_norm": 0.8120512962341309,
      "learning_rate": 9.885805517950542e-06,
      "loss": 0.887,
      "step": 2570
    },
    {
      "epoch": 0.14150476085640376,
      "grad_norm": 0.8475256562232971,
      "learning_rate": 9.885713388008148e-06,
      "loss": 0.7935,
      "step": 2571
    },
    {
      "epoch": 0.14155979965875942,
      "grad_norm": 0.7669919729232788,
      "learning_rate": 9.885621221346038e-06,
      "loss": 0.7728,
      "step": 2572
    },
    {
      "epoch": 0.14161483846111508,
      "grad_norm": 0.8298916220664978,
      "learning_rate": 9.885529017964906e-06,
      "loss": 0.7723,
      "step": 2573
    },
    {
      "epoch": 0.14166987726347074,
      "grad_norm": 0.8630721569061279,
      "learning_rate": 9.885436777865447e-06,
      "loss": 0.8395,
      "step": 2574
    },
    {
      "epoch": 0.1417249160658264,
      "grad_norm": 0.7566008567810059,
      "learning_rate": 9.885344501048352e-06,
      "loss": 0.806,
      "step": 2575
    },
    {
      "epoch": 0.14177995486818207,
      "grad_norm": 0.7870769500732422,
      "learning_rate": 9.885252187514316e-06,
      "loss": 0.7683,
      "step": 2576
    },
    {
      "epoch": 0.14183499367053773,
      "grad_norm": 0.879648745059967,
      "learning_rate": 9.885159837264033e-06,
      "loss": 0.8472,
      "step": 2577
    },
    {
      "epoch": 0.1418900324728934,
      "grad_norm": 0.76839280128479,
      "learning_rate": 9.885067450298196e-06,
      "loss": 0.8534,
      "step": 2578
    },
    {
      "epoch": 0.14194507127524905,
      "grad_norm": 0.8268701434135437,
      "learning_rate": 9.884975026617498e-06,
      "loss": 0.7799,
      "step": 2579
    },
    {
      "epoch": 0.1420001100776047,
      "grad_norm": 0.8226090669631958,
      "learning_rate": 9.884882566222638e-06,
      "loss": 0.6756,
      "step": 2580
    },
    {
      "epoch": 0.14205514887996037,
      "grad_norm": 0.8299756050109863,
      "learning_rate": 9.884790069114307e-06,
      "loss": 0.734,
      "step": 2581
    },
    {
      "epoch": 0.14211018768231604,
      "grad_norm": 0.8241812586784363,
      "learning_rate": 9.8846975352932e-06,
      "loss": 0.8335,
      "step": 2582
    },
    {
      "epoch": 0.1421652264846717,
      "grad_norm": 0.8458926677703857,
      "learning_rate": 9.884604964760016e-06,
      "loss": 0.7376,
      "step": 2583
    },
    {
      "epoch": 0.14222026528702736,
      "grad_norm": 0.876966655254364,
      "learning_rate": 9.884512357515447e-06,
      "loss": 0.9414,
      "step": 2584
    },
    {
      "epoch": 0.14227530408938302,
      "grad_norm": 0.770252525806427,
      "learning_rate": 9.88441971356019e-06,
      "loss": 0.8312,
      "step": 2585
    },
    {
      "epoch": 0.14233034289173868,
      "grad_norm": 0.7883023023605347,
      "learning_rate": 9.884327032894945e-06,
      "loss": 0.8568,
      "step": 2586
    },
    {
      "epoch": 0.14238538169409434,
      "grad_norm": 0.9092289209365845,
      "learning_rate": 9.884234315520405e-06,
      "loss": 0.9078,
      "step": 2587
    },
    {
      "epoch": 0.14244042049645,
      "grad_norm": 0.7946531176567078,
      "learning_rate": 9.884141561437266e-06,
      "loss": 0.6895,
      "step": 2588
    },
    {
      "epoch": 0.14249545929880567,
      "grad_norm": 0.7791070342063904,
      "learning_rate": 9.884048770646227e-06,
      "loss": 0.6984,
      "step": 2589
    },
    {
      "epoch": 0.14255049810116133,
      "grad_norm": 0.7775537371635437,
      "learning_rate": 9.883955943147982e-06,
      "loss": 0.7568,
      "step": 2590
    },
    {
      "epoch": 0.142605536903517,
      "grad_norm": 0.7735158801078796,
      "learning_rate": 9.883863078943234e-06,
      "loss": 0.8215,
      "step": 2591
    },
    {
      "epoch": 0.14266057570587265,
      "grad_norm": 0.881365180015564,
      "learning_rate": 9.88377017803268e-06,
      "loss": 0.8817,
      "step": 2592
    },
    {
      "epoch": 0.1427156145082283,
      "grad_norm": 0.8643443584442139,
      "learning_rate": 9.883677240417014e-06,
      "loss": 0.8024,
      "step": 2593
    },
    {
      "epoch": 0.14277065331058397,
      "grad_norm": 0.885713517665863,
      "learning_rate": 9.883584266096938e-06,
      "loss": 0.7612,
      "step": 2594
    },
    {
      "epoch": 0.14282569211293963,
      "grad_norm": 0.771340012550354,
      "learning_rate": 9.88349125507315e-06,
      "loss": 0.8293,
      "step": 2595
    },
    {
      "epoch": 0.1428807309152953,
      "grad_norm": 0.8284093737602234,
      "learning_rate": 9.88339820734635e-06,
      "loss": 0.8539,
      "step": 2596
    },
    {
      "epoch": 0.14293576971765096,
      "grad_norm": 0.9597725868225098,
      "learning_rate": 9.883305122917233e-06,
      "loss": 0.9054,
      "step": 2597
    },
    {
      "epoch": 0.14299080852000662,
      "grad_norm": 0.7552937269210815,
      "learning_rate": 9.883212001786504e-06,
      "loss": 0.8047,
      "step": 2598
    },
    {
      "epoch": 0.14304584732236225,
      "grad_norm": 0.8008492588996887,
      "learning_rate": 9.883118843954861e-06,
      "loss": 0.802,
      "step": 2599
    },
    {
      "epoch": 0.1431008861247179,
      "grad_norm": 0.8169753551483154,
      "learning_rate": 9.883025649423003e-06,
      "loss": 0.8837,
      "step": 2600
    },
    {
      "epoch": 0.14315592492707357,
      "grad_norm": 0.8521036505699158,
      "learning_rate": 9.882932418191632e-06,
      "loss": 0.8266,
      "step": 2601
    },
    {
      "epoch": 0.14321096372942924,
      "grad_norm": 0.8647341728210449,
      "learning_rate": 9.882839150261449e-06,
      "loss": 0.8949,
      "step": 2602
    },
    {
      "epoch": 0.1432660025317849,
      "grad_norm": 0.9236162304878235,
      "learning_rate": 9.882745845633153e-06,
      "loss": 0.8474,
      "step": 2603
    },
    {
      "epoch": 0.14332104133414056,
      "grad_norm": 0.8422677516937256,
      "learning_rate": 9.882652504307445e-06,
      "loss": 0.8396,
      "step": 2604
    },
    {
      "epoch": 0.14337608013649622,
      "grad_norm": 0.902036190032959,
      "learning_rate": 9.88255912628503e-06,
      "loss": 0.8075,
      "step": 2605
    },
    {
      "epoch": 0.14343111893885188,
      "grad_norm": 0.8972339630126953,
      "learning_rate": 9.882465711566605e-06,
      "loss": 0.8143,
      "step": 2606
    },
    {
      "epoch": 0.14348615774120754,
      "grad_norm": 0.8025243282318115,
      "learning_rate": 9.882372260152877e-06,
      "loss": 0.771,
      "step": 2607
    },
    {
      "epoch": 0.1435411965435632,
      "grad_norm": 0.8260911107063293,
      "learning_rate": 9.882278772044545e-06,
      "loss": 0.7679,
      "step": 2608
    },
    {
      "epoch": 0.14359623534591887,
      "grad_norm": 0.8069774508476257,
      "learning_rate": 9.882185247242313e-06,
      "loss": 0.8489,
      "step": 2609
    },
    {
      "epoch": 0.14365127414827453,
      "grad_norm": 0.8702567219734192,
      "learning_rate": 9.882091685746883e-06,
      "loss": 0.9258,
      "step": 2610
    },
    {
      "epoch": 0.1437063129506302,
      "grad_norm": 0.8841683268547058,
      "learning_rate": 9.881998087558959e-06,
      "loss": 0.7858,
      "step": 2611
    },
    {
      "epoch": 0.14376135175298585,
      "grad_norm": 0.7302986979484558,
      "learning_rate": 9.881904452679246e-06,
      "loss": 0.7339,
      "step": 2612
    },
    {
      "epoch": 0.1438163905553415,
      "grad_norm": 0.7852466106414795,
      "learning_rate": 9.881810781108442e-06,
      "loss": 0.8397,
      "step": 2613
    },
    {
      "epoch": 0.14387142935769717,
      "grad_norm": 0.7986249327659607,
      "learning_rate": 9.881717072847258e-06,
      "loss": 0.7573,
      "step": 2614
    },
    {
      "epoch": 0.14392646816005283,
      "grad_norm": 0.750000536441803,
      "learning_rate": 9.881623327896395e-06,
      "loss": 0.7128,
      "step": 2615
    },
    {
      "epoch": 0.1439815069624085,
      "grad_norm": 0.8796436786651611,
      "learning_rate": 9.881529546256557e-06,
      "loss": 0.9364,
      "step": 2616
    },
    {
      "epoch": 0.14403654576476416,
      "grad_norm": 0.8621297478675842,
      "learning_rate": 9.881435727928449e-06,
      "loss": 0.9323,
      "step": 2617
    },
    {
      "epoch": 0.14409158456711982,
      "grad_norm": 0.8213173151016235,
      "learning_rate": 9.881341872912777e-06,
      "loss": 0.7746,
      "step": 2618
    },
    {
      "epoch": 0.14414662336947548,
      "grad_norm": 0.7761938571929932,
      "learning_rate": 9.881247981210247e-06,
      "loss": 0.8065,
      "step": 2619
    },
    {
      "epoch": 0.14420166217183114,
      "grad_norm": 0.8333988785743713,
      "learning_rate": 9.881154052821564e-06,
      "loss": 0.8727,
      "step": 2620
    },
    {
      "epoch": 0.1442567009741868,
      "grad_norm": 0.7263909578323364,
      "learning_rate": 9.881060087747433e-06,
      "loss": 0.8194,
      "step": 2621
    },
    {
      "epoch": 0.14431173977654246,
      "grad_norm": 0.7472667098045349,
      "learning_rate": 9.880966085988562e-06,
      "loss": 0.77,
      "step": 2622
    },
    {
      "epoch": 0.14436677857889813,
      "grad_norm": 0.7999943494796753,
      "learning_rate": 9.880872047545656e-06,
      "loss": 0.7936,
      "step": 2623
    },
    {
      "epoch": 0.1444218173812538,
      "grad_norm": 0.8359610438346863,
      "learning_rate": 9.88077797241942e-06,
      "loss": 0.7946,
      "step": 2624
    },
    {
      "epoch": 0.14447685618360945,
      "grad_norm": 0.8666403889656067,
      "learning_rate": 9.880683860610566e-06,
      "loss": 0.8152,
      "step": 2625
    },
    {
      "epoch": 0.1445318949859651,
      "grad_norm": 0.7883741855621338,
      "learning_rate": 9.880589712119797e-06,
      "loss": 0.7972,
      "step": 2626
    },
    {
      "epoch": 0.14458693378832077,
      "grad_norm": 0.8048827648162842,
      "learning_rate": 9.880495526947824e-06,
      "loss": 0.8221,
      "step": 2627
    },
    {
      "epoch": 0.14464197259067643,
      "grad_norm": 0.718292236328125,
      "learning_rate": 9.88040130509535e-06,
      "loss": 0.7648,
      "step": 2628
    },
    {
      "epoch": 0.1446970113930321,
      "grad_norm": 0.7748421430587769,
      "learning_rate": 9.880307046563088e-06,
      "loss": 0.8146,
      "step": 2629
    },
    {
      "epoch": 0.14475205019538775,
      "grad_norm": 0.8015987873077393,
      "learning_rate": 9.880212751351745e-06,
      "loss": 0.7935,
      "step": 2630
    },
    {
      "epoch": 0.14480708899774342,
      "grad_norm": 0.7628459930419922,
      "learning_rate": 9.88011841946203e-06,
      "loss": 0.7469,
      "step": 2631
    },
    {
      "epoch": 0.14486212780009908,
      "grad_norm": 0.7152888774871826,
      "learning_rate": 9.88002405089465e-06,
      "loss": 0.7721,
      "step": 2632
    },
    {
      "epoch": 0.14491716660245474,
      "grad_norm": 0.8075545430183411,
      "learning_rate": 9.879929645650315e-06,
      "loss": 0.8799,
      "step": 2633
    },
    {
      "epoch": 0.1449722054048104,
      "grad_norm": 0.7981964945793152,
      "learning_rate": 9.879835203729736e-06,
      "loss": 0.8265,
      "step": 2634
    },
    {
      "epoch": 0.14502724420716606,
      "grad_norm": 0.7699866890907288,
      "learning_rate": 9.879740725133623e-06,
      "loss": 0.8489,
      "step": 2635
    },
    {
      "epoch": 0.14508228300952172,
      "grad_norm": 0.7991634011268616,
      "learning_rate": 9.879646209862682e-06,
      "loss": 0.8754,
      "step": 2636
    },
    {
      "epoch": 0.14513732181187738,
      "grad_norm": 0.8284991383552551,
      "learning_rate": 9.879551657917628e-06,
      "loss": 0.811,
      "step": 2637
    },
    {
      "epoch": 0.14519236061423305,
      "grad_norm": 0.9189227819442749,
      "learning_rate": 9.87945706929917e-06,
      "loss": 0.8486,
      "step": 2638
    },
    {
      "epoch": 0.1452473994165887,
      "grad_norm": 0.8599026799201965,
      "learning_rate": 9.879362444008018e-06,
      "loss": 0.8383,
      "step": 2639
    },
    {
      "epoch": 0.14530243821894437,
      "grad_norm": 0.8764603137969971,
      "learning_rate": 9.879267782044885e-06,
      "loss": 0.7918,
      "step": 2640
    },
    {
      "epoch": 0.14535747702130003,
      "grad_norm": 0.8061341047286987,
      "learning_rate": 9.87917308341048e-06,
      "loss": 0.8292,
      "step": 2641
    },
    {
      "epoch": 0.14541251582365566,
      "grad_norm": 1.031220555305481,
      "learning_rate": 9.879078348105518e-06,
      "loss": 0.6612,
      "step": 2642
    },
    {
      "epoch": 0.14546755462601133,
      "grad_norm": 1.014491319656372,
      "learning_rate": 9.878983576130708e-06,
      "loss": 0.8512,
      "step": 2643
    },
    {
      "epoch": 0.145522593428367,
      "grad_norm": 0.8365896940231323,
      "learning_rate": 9.878888767486764e-06,
      "loss": 0.7995,
      "step": 2644
    },
    {
      "epoch": 0.14557763223072265,
      "grad_norm": 0.8086197972297668,
      "learning_rate": 9.878793922174397e-06,
      "loss": 0.8069,
      "step": 2645
    },
    {
      "epoch": 0.1456326710330783,
      "grad_norm": 0.8075234889984131,
      "learning_rate": 9.878699040194322e-06,
      "loss": 0.8415,
      "step": 2646
    },
    {
      "epoch": 0.14568770983543397,
      "grad_norm": 0.9413748979568481,
      "learning_rate": 9.87860412154725e-06,
      "loss": 0.7811,
      "step": 2647
    },
    {
      "epoch": 0.14574274863778963,
      "grad_norm": 0.7744552493095398,
      "learning_rate": 9.878509166233895e-06,
      "loss": 0.7983,
      "step": 2648
    },
    {
      "epoch": 0.1457977874401453,
      "grad_norm": 0.8184664845466614,
      "learning_rate": 9.878414174254974e-06,
      "loss": 0.8052,
      "step": 2649
    },
    {
      "epoch": 0.14585282624250095,
      "grad_norm": 0.928814172744751,
      "learning_rate": 9.878319145611195e-06,
      "loss": 0.7695,
      "step": 2650
    },
    {
      "epoch": 0.14590786504485662,
      "grad_norm": 0.9623318314552307,
      "learning_rate": 9.878224080303276e-06,
      "loss": 0.9025,
      "step": 2651
    },
    {
      "epoch": 0.14596290384721228,
      "grad_norm": 0.866538405418396,
      "learning_rate": 9.87812897833193e-06,
      "loss": 0.7895,
      "step": 2652
    },
    {
      "epoch": 0.14601794264956794,
      "grad_norm": 0.9248599410057068,
      "learning_rate": 9.878033839697874e-06,
      "loss": 0.8532,
      "step": 2653
    },
    {
      "epoch": 0.1460729814519236,
      "grad_norm": 0.7866301536560059,
      "learning_rate": 9.87793866440182e-06,
      "loss": 0.8724,
      "step": 2654
    },
    {
      "epoch": 0.14612802025427926,
      "grad_norm": 0.8471634387969971,
      "learning_rate": 9.877843452444485e-06,
      "loss": 0.9184,
      "step": 2655
    },
    {
      "epoch": 0.14618305905663492,
      "grad_norm": 0.7367103695869446,
      "learning_rate": 9.877748203826585e-06,
      "loss": 0.7328,
      "step": 2656
    },
    {
      "epoch": 0.14623809785899058,
      "grad_norm": 0.95980304479599,
      "learning_rate": 9.877652918548834e-06,
      "loss": 0.9274,
      "step": 2657
    },
    {
      "epoch": 0.14629313666134625,
      "grad_norm": 1.0511064529418945,
      "learning_rate": 9.87755759661195e-06,
      "loss": 0.8223,
      "step": 2658
    },
    {
      "epoch": 0.1463481754637019,
      "grad_norm": 0.7616510391235352,
      "learning_rate": 9.877462238016649e-06,
      "loss": 0.7473,
      "step": 2659
    },
    {
      "epoch": 0.14640321426605757,
      "grad_norm": 0.7814056873321533,
      "learning_rate": 9.877366842763647e-06,
      "loss": 0.8898,
      "step": 2660
    },
    {
      "epoch": 0.14645825306841323,
      "grad_norm": 0.8707298636436462,
      "learning_rate": 9.877271410853662e-06,
      "loss": 0.8792,
      "step": 2661
    },
    {
      "epoch": 0.1465132918707689,
      "grad_norm": 0.8618701696395874,
      "learning_rate": 9.877175942287409e-06,
      "loss": 0.8761,
      "step": 2662
    },
    {
      "epoch": 0.14656833067312455,
      "grad_norm": 0.9437732100486755,
      "learning_rate": 9.877080437065609e-06,
      "loss": 0.7922,
      "step": 2663
    },
    {
      "epoch": 0.14662336947548021,
      "grad_norm": 0.9465780258178711,
      "learning_rate": 9.876984895188976e-06,
      "loss": 0.8449,
      "step": 2664
    },
    {
      "epoch": 0.14667840827783588,
      "grad_norm": 0.7149911522865295,
      "learning_rate": 9.876889316658233e-06,
      "loss": 0.6408,
      "step": 2665
    },
    {
      "epoch": 0.14673344708019154,
      "grad_norm": 0.9996811151504517,
      "learning_rate": 9.876793701474092e-06,
      "loss": 0.9324,
      "step": 2666
    },
    {
      "epoch": 0.1467884858825472,
      "grad_norm": 0.7941329479217529,
      "learning_rate": 9.876698049637277e-06,
      "loss": 0.8115,
      "step": 2667
    },
    {
      "epoch": 0.14684352468490286,
      "grad_norm": 0.754175066947937,
      "learning_rate": 9.876602361148504e-06,
      "loss": 0.7709,
      "step": 2668
    },
    {
      "epoch": 0.14689856348725852,
      "grad_norm": 0.7867946624755859,
      "learning_rate": 9.876506636008494e-06,
      "loss": 0.8578,
      "step": 2669
    },
    {
      "epoch": 0.14695360228961418,
      "grad_norm": 0.7441185116767883,
      "learning_rate": 9.876410874217965e-06,
      "loss": 0.8491,
      "step": 2670
    },
    {
      "epoch": 0.14700864109196984,
      "grad_norm": 0.8414027690887451,
      "learning_rate": 9.876315075777638e-06,
      "loss": 0.8404,
      "step": 2671
    },
    {
      "epoch": 0.1470636798943255,
      "grad_norm": 0.7911489009857178,
      "learning_rate": 9.876219240688231e-06,
      "loss": 0.8606,
      "step": 2672
    },
    {
      "epoch": 0.14711871869668117,
      "grad_norm": 0.8601381778717041,
      "learning_rate": 9.876123368950465e-06,
      "loss": 0.7753,
      "step": 2673
    },
    {
      "epoch": 0.14717375749903683,
      "grad_norm": 0.8672378659248352,
      "learning_rate": 9.876027460565062e-06,
      "loss": 0.7763,
      "step": 2674
    },
    {
      "epoch": 0.1472287963013925,
      "grad_norm": 0.7192933559417725,
      "learning_rate": 9.875931515532742e-06,
      "loss": 0.7681,
      "step": 2675
    },
    {
      "epoch": 0.14728383510374815,
      "grad_norm": 0.7483426332473755,
      "learning_rate": 9.875835533854226e-06,
      "loss": 0.8129,
      "step": 2676
    },
    {
      "epoch": 0.1473388739061038,
      "grad_norm": 0.8883694410324097,
      "learning_rate": 9.875739515530235e-06,
      "loss": 0.8912,
      "step": 2677
    },
    {
      "epoch": 0.14739391270845947,
      "grad_norm": 0.8440148234367371,
      "learning_rate": 9.87564346056149e-06,
      "loss": 0.8411,
      "step": 2678
    },
    {
      "epoch": 0.14744895151081513,
      "grad_norm": 0.8916668891906738,
      "learning_rate": 9.875547368948715e-06,
      "loss": 0.8484,
      "step": 2679
    },
    {
      "epoch": 0.1475039903131708,
      "grad_norm": 0.805258572101593,
      "learning_rate": 9.875451240692631e-06,
      "loss": 0.8172,
      "step": 2680
    },
    {
      "epoch": 0.14755902911552646,
      "grad_norm": 0.8322305679321289,
      "learning_rate": 9.87535507579396e-06,
      "loss": 0.809,
      "step": 2681
    },
    {
      "epoch": 0.14761406791788212,
      "grad_norm": 0.7320597767829895,
      "learning_rate": 9.875258874253424e-06,
      "loss": 0.7346,
      "step": 2682
    },
    {
      "epoch": 0.14766910672023778,
      "grad_norm": 1.018036127090454,
      "learning_rate": 9.875162636071749e-06,
      "loss": 0.931,
      "step": 2683
    },
    {
      "epoch": 0.14772414552259344,
      "grad_norm": 0.8601503968238831,
      "learning_rate": 9.875066361249657e-06,
      "loss": 0.7689,
      "step": 2684
    },
    {
      "epoch": 0.14777918432494908,
      "grad_norm": 0.8478472232818604,
      "learning_rate": 9.87497004978787e-06,
      "loss": 0.9545,
      "step": 2685
    },
    {
      "epoch": 0.14783422312730474,
      "grad_norm": 0.7510890364646912,
      "learning_rate": 9.874873701687115e-06,
      "loss": 0.7794,
      "step": 2686
    },
    {
      "epoch": 0.1478892619296604,
      "grad_norm": 0.8226999044418335,
      "learning_rate": 9.874777316948112e-06,
      "loss": 0.8477,
      "step": 2687
    },
    {
      "epoch": 0.14794430073201606,
      "grad_norm": 0.8284991979598999,
      "learning_rate": 9.874680895571588e-06,
      "loss": 0.8498,
      "step": 2688
    },
    {
      "epoch": 0.14799933953437172,
      "grad_norm": 0.9007356762886047,
      "learning_rate": 9.874584437558267e-06,
      "loss": 0.8526,
      "step": 2689
    },
    {
      "epoch": 0.14805437833672738,
      "grad_norm": 0.8770126104354858,
      "learning_rate": 9.874487942908877e-06,
      "loss": 0.844,
      "step": 2690
    },
    {
      "epoch": 0.14810941713908304,
      "grad_norm": 1.1561466455459595,
      "learning_rate": 9.874391411624138e-06,
      "loss": 0.976,
      "step": 2691
    },
    {
      "epoch": 0.1481644559414387,
      "grad_norm": 0.8162640929222107,
      "learning_rate": 9.874294843704777e-06,
      "loss": 0.8581,
      "step": 2692
    },
    {
      "epoch": 0.14821949474379437,
      "grad_norm": 0.8308132290840149,
      "learning_rate": 9.874198239151522e-06,
      "loss": 0.8303,
      "step": 2693
    },
    {
      "epoch": 0.14827453354615003,
      "grad_norm": 0.771024227142334,
      "learning_rate": 9.874101597965098e-06,
      "loss": 0.8351,
      "step": 2694
    },
    {
      "epoch": 0.1483295723485057,
      "grad_norm": 0.7588162422180176,
      "learning_rate": 9.874004920146232e-06,
      "loss": 0.7858,
      "step": 2695
    },
    {
      "epoch": 0.14838461115086135,
      "grad_norm": 0.8282446265220642,
      "learning_rate": 9.873908205695648e-06,
      "loss": 0.8465,
      "step": 2696
    },
    {
      "epoch": 0.148439649953217,
      "grad_norm": 0.8342786431312561,
      "learning_rate": 9.873811454614076e-06,
      "loss": 0.8688,
      "step": 2697
    },
    {
      "epoch": 0.14849468875557267,
      "grad_norm": 0.7957108020782471,
      "learning_rate": 9.87371466690224e-06,
      "loss": 0.8381,
      "step": 2698
    },
    {
      "epoch": 0.14854972755792833,
      "grad_norm": 0.8763726353645325,
      "learning_rate": 9.87361784256087e-06,
      "loss": 0.8922,
      "step": 2699
    },
    {
      "epoch": 0.148604766360284,
      "grad_norm": 0.7760055661201477,
      "learning_rate": 9.873520981590693e-06,
      "loss": 0.8384,
      "step": 2700
    },
    {
      "epoch": 0.14865980516263966,
      "grad_norm": 0.9691097736358643,
      "learning_rate": 9.873424083992436e-06,
      "loss": 0.8581,
      "step": 2701
    },
    {
      "epoch": 0.14871484396499532,
      "grad_norm": 0.9072558879852295,
      "learning_rate": 9.87332714976683e-06,
      "loss": 0.8942,
      "step": 2702
    },
    {
      "epoch": 0.14876988276735098,
      "grad_norm": 0.8961714506149292,
      "learning_rate": 9.8732301789146e-06,
      "loss": 0.8062,
      "step": 2703
    },
    {
      "epoch": 0.14882492156970664,
      "grad_norm": 1.4835050106048584,
      "learning_rate": 9.873133171436477e-06,
      "loss": 0.886,
      "step": 2704
    },
    {
      "epoch": 0.1488799603720623,
      "grad_norm": 0.8153702616691589,
      "learning_rate": 9.87303612733319e-06,
      "loss": 0.8369,
      "step": 2705
    },
    {
      "epoch": 0.14893499917441796,
      "grad_norm": 0.8755800724029541,
      "learning_rate": 9.872939046605467e-06,
      "loss": 0.7591,
      "step": 2706
    },
    {
      "epoch": 0.14899003797677363,
      "grad_norm": 0.8173243403434753,
      "learning_rate": 9.872841929254038e-06,
      "loss": 0.8626,
      "step": 2707
    },
    {
      "epoch": 0.1490450767791293,
      "grad_norm": 0.7891639471054077,
      "learning_rate": 9.872744775279634e-06,
      "loss": 0.737,
      "step": 2708
    },
    {
      "epoch": 0.14910011558148495,
      "grad_norm": 1.0270631313323975,
      "learning_rate": 9.872647584682985e-06,
      "loss": 0.9202,
      "step": 2709
    },
    {
      "epoch": 0.1491551543838406,
      "grad_norm": 0.7736123204231262,
      "learning_rate": 9.872550357464822e-06,
      "loss": 0.7835,
      "step": 2710
    },
    {
      "epoch": 0.14921019318619627,
      "grad_norm": 0.7791550159454346,
      "learning_rate": 9.872453093625873e-06,
      "loss": 0.8375,
      "step": 2711
    },
    {
      "epoch": 0.14926523198855193,
      "grad_norm": 0.8410583734512329,
      "learning_rate": 9.872355793166872e-06,
      "loss": 0.877,
      "step": 2712
    },
    {
      "epoch": 0.1493202707909076,
      "grad_norm": 0.8277738094329834,
      "learning_rate": 9.87225845608855e-06,
      "loss": 0.7255,
      "step": 2713
    },
    {
      "epoch": 0.14937530959326326,
      "grad_norm": 0.8617290258407593,
      "learning_rate": 9.872161082391635e-06,
      "loss": 0.7885,
      "step": 2714
    },
    {
      "epoch": 0.14943034839561892,
      "grad_norm": 0.8866406679153442,
      "learning_rate": 9.872063672076864e-06,
      "loss": 0.8621,
      "step": 2715
    },
    {
      "epoch": 0.14948538719797458,
      "grad_norm": 0.7581049799919128,
      "learning_rate": 9.871966225144964e-06,
      "loss": 0.8177,
      "step": 2716
    },
    {
      "epoch": 0.14954042600033024,
      "grad_norm": 0.833696722984314,
      "learning_rate": 9.871868741596673e-06,
      "loss": 0.8382,
      "step": 2717
    },
    {
      "epoch": 0.1495954648026859,
      "grad_norm": 1.0857365131378174,
      "learning_rate": 9.871771221432718e-06,
      "loss": 0.9254,
      "step": 2718
    },
    {
      "epoch": 0.14965050360504156,
      "grad_norm": 0.7622446417808533,
      "learning_rate": 9.871673664653837e-06,
      "loss": 0.832,
      "step": 2719
    },
    {
      "epoch": 0.14970554240739722,
      "grad_norm": 0.7436832785606384,
      "learning_rate": 9.871576071260758e-06,
      "loss": 0.7642,
      "step": 2720
    },
    {
      "epoch": 0.14976058120975289,
      "grad_norm": 0.8547641634941101,
      "learning_rate": 9.87147844125422e-06,
      "loss": 0.7584,
      "step": 2721
    },
    {
      "epoch": 0.14981562001210855,
      "grad_norm": 0.7634096145629883,
      "learning_rate": 9.871380774634953e-06,
      "loss": 0.8332,
      "step": 2722
    },
    {
      "epoch": 0.1498706588144642,
      "grad_norm": 0.7949081063270569,
      "learning_rate": 9.871283071403692e-06,
      "loss": 0.7812,
      "step": 2723
    },
    {
      "epoch": 0.14992569761681987,
      "grad_norm": 0.8089914321899414,
      "learning_rate": 9.871185331561171e-06,
      "loss": 0.8503,
      "step": 2724
    },
    {
      "epoch": 0.14998073641917553,
      "grad_norm": 0.8451627492904663,
      "learning_rate": 9.871087555108125e-06,
      "loss": 0.9101,
      "step": 2725
    },
    {
      "epoch": 0.1500357752215312,
      "grad_norm": 0.8399865627288818,
      "learning_rate": 9.87098974204529e-06,
      "loss": 0.8222,
      "step": 2726
    },
    {
      "epoch": 0.15009081402388685,
      "grad_norm": 0.7786773443222046,
      "learning_rate": 9.870891892373397e-06,
      "loss": 0.8069,
      "step": 2727
    },
    {
      "epoch": 0.1501458528262425,
      "grad_norm": 0.8530564308166504,
      "learning_rate": 9.870794006093188e-06,
      "loss": 0.9229,
      "step": 2728
    },
    {
      "epoch": 0.15020089162859815,
      "grad_norm": 0.7640067934989929,
      "learning_rate": 9.870696083205394e-06,
      "loss": 0.761,
      "step": 2729
    },
    {
      "epoch": 0.1502559304309538,
      "grad_norm": 0.8953121900558472,
      "learning_rate": 9.87059812371075e-06,
      "loss": 0.8537,
      "step": 2730
    },
    {
      "epoch": 0.15031096923330947,
      "grad_norm": 0.7779926657676697,
      "learning_rate": 9.870500127609996e-06,
      "loss": 0.8184,
      "step": 2731
    },
    {
      "epoch": 0.15036600803566513,
      "grad_norm": 0.9181544184684753,
      "learning_rate": 9.870402094903865e-06,
      "loss": 0.8583,
      "step": 2732
    },
    {
      "epoch": 0.1504210468380208,
      "grad_norm": 0.7629374861717224,
      "learning_rate": 9.870304025593097e-06,
      "loss": 0.6741,
      "step": 2733
    },
    {
      "epoch": 0.15047608564037646,
      "grad_norm": 1.1455601453781128,
      "learning_rate": 9.87020591967843e-06,
      "loss": 0.8602,
      "step": 2734
    },
    {
      "epoch": 0.15053112444273212,
      "grad_norm": 0.83924800157547,
      "learning_rate": 9.870107777160596e-06,
      "loss": 0.8847,
      "step": 2735
    },
    {
      "epoch": 0.15058616324508778,
      "grad_norm": 0.9293402433395386,
      "learning_rate": 9.870009598040336e-06,
      "loss": 0.9008,
      "step": 2736
    },
    {
      "epoch": 0.15064120204744344,
      "grad_norm": 0.8198057413101196,
      "learning_rate": 9.869911382318389e-06,
      "loss": 0.8004,
      "step": 2737
    },
    {
      "epoch": 0.1506962408497991,
      "grad_norm": 0.8139753341674805,
      "learning_rate": 9.86981312999549e-06,
      "loss": 0.7316,
      "step": 2738
    },
    {
      "epoch": 0.15075127965215476,
      "grad_norm": 0.854184091091156,
      "learning_rate": 9.86971484107238e-06,
      "loss": 0.9424,
      "step": 2739
    },
    {
      "epoch": 0.15080631845451042,
      "grad_norm": 0.8626797199249268,
      "learning_rate": 9.869616515549797e-06,
      "loss": 0.8882,
      "step": 2740
    },
    {
      "epoch": 0.15086135725686609,
      "grad_norm": 0.8447514176368713,
      "learning_rate": 9.869518153428479e-06,
      "loss": 0.7762,
      "step": 2741
    },
    {
      "epoch": 0.15091639605922175,
      "grad_norm": 1.1359349489212036,
      "learning_rate": 9.869419754709166e-06,
      "loss": 0.9233,
      "step": 2742
    },
    {
      "epoch": 0.1509714348615774,
      "grad_norm": 0.8095758557319641,
      "learning_rate": 9.869321319392597e-06,
      "loss": 0.8833,
      "step": 2743
    },
    {
      "epoch": 0.15102647366393307,
      "grad_norm": 0.8364169001579285,
      "learning_rate": 9.869222847479514e-06,
      "loss": 0.833,
      "step": 2744
    },
    {
      "epoch": 0.15108151246628873,
      "grad_norm": 0.7664803266525269,
      "learning_rate": 9.869124338970653e-06,
      "loss": 0.8125,
      "step": 2745
    },
    {
      "epoch": 0.1511365512686444,
      "grad_norm": 0.8129634857177734,
      "learning_rate": 9.86902579386676e-06,
      "loss": 0.8277,
      "step": 2746
    },
    {
      "epoch": 0.15119159007100005,
      "grad_norm": 0.8195592164993286,
      "learning_rate": 9.86892721216857e-06,
      "loss": 0.8489,
      "step": 2747
    },
    {
      "epoch": 0.15124662887335572,
      "grad_norm": 0.8116651177406311,
      "learning_rate": 9.868828593876827e-06,
      "loss": 0.7831,
      "step": 2748
    },
    {
      "epoch": 0.15130166767571138,
      "grad_norm": 0.8200114369392395,
      "learning_rate": 9.868729938992272e-06,
      "loss": 0.8956,
      "step": 2749
    },
    {
      "epoch": 0.15135670647806704,
      "grad_norm": 0.8521816730499268,
      "learning_rate": 9.868631247515645e-06,
      "loss": 0.804,
      "step": 2750
    },
    {
      "epoch": 0.1514117452804227,
      "grad_norm": 1.0386497974395752,
      "learning_rate": 9.868532519447691e-06,
      "loss": 0.8563,
      "step": 2751
    },
    {
      "epoch": 0.15146678408277836,
      "grad_norm": 0.8345486521720886,
      "learning_rate": 9.868433754789149e-06,
      "loss": 0.9838,
      "step": 2752
    },
    {
      "epoch": 0.15152182288513402,
      "grad_norm": 0.7207526564598083,
      "learning_rate": 9.868334953540762e-06,
      "loss": 0.6711,
      "step": 2753
    },
    {
      "epoch": 0.15157686168748968,
      "grad_norm": 0.8159164786338806,
      "learning_rate": 9.86823611570327e-06,
      "loss": 0.7591,
      "step": 2754
    },
    {
      "epoch": 0.15163190048984534,
      "grad_norm": 0.9062225818634033,
      "learning_rate": 9.868137241277422e-06,
      "loss": 0.8009,
      "step": 2755
    },
    {
      "epoch": 0.151686939292201,
      "grad_norm": 0.8136696219444275,
      "learning_rate": 9.868038330263957e-06,
      "loss": 0.7014,
      "step": 2756
    },
    {
      "epoch": 0.15174197809455667,
      "grad_norm": 0.7237691283226013,
      "learning_rate": 9.867939382663618e-06,
      "loss": 0.7766,
      "step": 2757
    },
    {
      "epoch": 0.15179701689691233,
      "grad_norm": 0.8913742303848267,
      "learning_rate": 9.86784039847715e-06,
      "loss": 0.9362,
      "step": 2758
    },
    {
      "epoch": 0.151852055699268,
      "grad_norm": 0.7763763070106506,
      "learning_rate": 9.867741377705296e-06,
      "loss": 0.7843,
      "step": 2759
    },
    {
      "epoch": 0.15190709450162365,
      "grad_norm": 0.8973854780197144,
      "learning_rate": 9.867642320348803e-06,
      "loss": 0.911,
      "step": 2760
    },
    {
      "epoch": 0.1519621333039793,
      "grad_norm": 0.7979685664176941,
      "learning_rate": 9.86754322640841e-06,
      "loss": 0.81,
      "step": 2761
    },
    {
      "epoch": 0.15201717210633497,
      "grad_norm": 0.7740911841392517,
      "learning_rate": 9.867444095884867e-06,
      "loss": 0.8197,
      "step": 2762
    },
    {
      "epoch": 0.15207221090869064,
      "grad_norm": 0.8400475978851318,
      "learning_rate": 9.867344928778916e-06,
      "loss": 0.8809,
      "step": 2763
    },
    {
      "epoch": 0.1521272497110463,
      "grad_norm": 0.8995040655136108,
      "learning_rate": 9.867245725091305e-06,
      "loss": 0.8382,
      "step": 2764
    },
    {
      "epoch": 0.15218228851340196,
      "grad_norm": 0.8162381052970886,
      "learning_rate": 9.867146484822779e-06,
      "loss": 0.9238,
      "step": 2765
    },
    {
      "epoch": 0.15223732731575762,
      "grad_norm": 0.7668827176094055,
      "learning_rate": 9.867047207974079e-06,
      "loss": 0.8345,
      "step": 2766
    },
    {
      "epoch": 0.15229236611811328,
      "grad_norm": 0.8719204664230347,
      "learning_rate": 9.866947894545957e-06,
      "loss": 0.7899,
      "step": 2767
    },
    {
      "epoch": 0.15234740492046894,
      "grad_norm": 0.9043570756912231,
      "learning_rate": 9.866848544539159e-06,
      "loss": 0.8783,
      "step": 2768
    },
    {
      "epoch": 0.1524024437228246,
      "grad_norm": 0.8859694004058838,
      "learning_rate": 9.866749157954428e-06,
      "loss": 0.862,
      "step": 2769
    },
    {
      "epoch": 0.15245748252518027,
      "grad_norm": 1.022719144821167,
      "learning_rate": 9.866649734792514e-06,
      "loss": 0.8943,
      "step": 2770
    },
    {
      "epoch": 0.1525125213275359,
      "grad_norm": 0.8710635900497437,
      "learning_rate": 9.866550275054163e-06,
      "loss": 0.7002,
      "step": 2771
    },
    {
      "epoch": 0.15256756012989156,
      "grad_norm": 0.8482942581176758,
      "learning_rate": 9.866450778740122e-06,
      "loss": 0.7529,
      "step": 2772
    },
    {
      "epoch": 0.15262259893224722,
      "grad_norm": 0.9637784361839294,
      "learning_rate": 9.866351245851142e-06,
      "loss": 0.8147,
      "step": 2773
    },
    {
      "epoch": 0.15267763773460288,
      "grad_norm": 1.0472246408462524,
      "learning_rate": 9.866251676387967e-06,
      "loss": 0.8019,
      "step": 2774
    },
    {
      "epoch": 0.15273267653695854,
      "grad_norm": 0.7916847467422485,
      "learning_rate": 9.866152070351347e-06,
      "loss": 0.7698,
      "step": 2775
    },
    {
      "epoch": 0.1527877153393142,
      "grad_norm": 0.8421853184700012,
      "learning_rate": 9.86605242774203e-06,
      "loss": 0.8085,
      "step": 2776
    },
    {
      "epoch": 0.15284275414166987,
      "grad_norm": 0.7990233898162842,
      "learning_rate": 9.865952748560768e-06,
      "loss": 0.8878,
      "step": 2777
    },
    {
      "epoch": 0.15289779294402553,
      "grad_norm": 0.8017451167106628,
      "learning_rate": 9.865853032808305e-06,
      "loss": 0.8707,
      "step": 2778
    },
    {
      "epoch": 0.1529528317463812,
      "grad_norm": 0.739850640296936,
      "learning_rate": 9.865753280485393e-06,
      "loss": 0.7884,
      "step": 2779
    },
    {
      "epoch": 0.15300787054873685,
      "grad_norm": 1.0682430267333984,
      "learning_rate": 9.865653491592784e-06,
      "loss": 0.8548,
      "step": 2780
    },
    {
      "epoch": 0.1530629093510925,
      "grad_norm": 0.7766296863555908,
      "learning_rate": 9.865553666131225e-06,
      "loss": 0.7786,
      "step": 2781
    },
    {
      "epoch": 0.15311794815344817,
      "grad_norm": 0.8903290629386902,
      "learning_rate": 9.865453804101466e-06,
      "loss": 0.8978,
      "step": 2782
    },
    {
      "epoch": 0.15317298695580384,
      "grad_norm": 0.8624514937400818,
      "learning_rate": 9.86535390550426e-06,
      "loss": 0.8472,
      "step": 2783
    },
    {
      "epoch": 0.1532280257581595,
      "grad_norm": 0.7765294909477234,
      "learning_rate": 9.865253970340356e-06,
      "loss": 0.7702,
      "step": 2784
    },
    {
      "epoch": 0.15328306456051516,
      "grad_norm": 0.9349095225334167,
      "learning_rate": 9.865153998610504e-06,
      "loss": 0.9154,
      "step": 2785
    },
    {
      "epoch": 0.15333810336287082,
      "grad_norm": 0.8435478210449219,
      "learning_rate": 9.865053990315458e-06,
      "loss": 0.8986,
      "step": 2786
    },
    {
      "epoch": 0.15339314216522648,
      "grad_norm": 0.8003486394882202,
      "learning_rate": 9.864953945455968e-06,
      "loss": 0.767,
      "step": 2787
    },
    {
      "epoch": 0.15344818096758214,
      "grad_norm": 0.8060823678970337,
      "learning_rate": 9.86485386403279e-06,
      "loss": 0.8332,
      "step": 2788
    },
    {
      "epoch": 0.1535032197699378,
      "grad_norm": 0.7914995551109314,
      "learning_rate": 9.864753746046668e-06,
      "loss": 0.6706,
      "step": 2789
    },
    {
      "epoch": 0.15355825857229347,
      "grad_norm": 0.7792215943336487,
      "learning_rate": 9.86465359149836e-06,
      "loss": 0.8721,
      "step": 2790
    },
    {
      "epoch": 0.15361329737464913,
      "grad_norm": 0.8572536110877991,
      "learning_rate": 9.864553400388619e-06,
      "loss": 0.8378,
      "step": 2791
    },
    {
      "epoch": 0.1536683361770048,
      "grad_norm": 0.7645615339279175,
      "learning_rate": 9.864453172718195e-06,
      "loss": 0.6909,
      "step": 2792
    },
    {
      "epoch": 0.15372337497936045,
      "grad_norm": 0.7627308964729309,
      "learning_rate": 9.864352908487846e-06,
      "loss": 0.7918,
      "step": 2793
    },
    {
      "epoch": 0.1537784137817161,
      "grad_norm": 1.0830100774765015,
      "learning_rate": 9.86425260769832e-06,
      "loss": 0.9007,
      "step": 2794
    },
    {
      "epoch": 0.15383345258407177,
      "grad_norm": 0.7667998671531677,
      "learning_rate": 9.864152270350374e-06,
      "loss": 0.832,
      "step": 2795
    },
    {
      "epoch": 0.15388849138642743,
      "grad_norm": 0.9967591762542725,
      "learning_rate": 9.864051896444764e-06,
      "loss": 0.8917,
      "step": 2796
    },
    {
      "epoch": 0.1539435301887831,
      "grad_norm": 0.8948462605476379,
      "learning_rate": 9.86395148598224e-06,
      "loss": 0.983,
      "step": 2797
    },
    {
      "epoch": 0.15399856899113876,
      "grad_norm": 0.7857423424720764,
      "learning_rate": 9.863851038963556e-06,
      "loss": 0.7826,
      "step": 2798
    },
    {
      "epoch": 0.15405360779349442,
      "grad_norm": 0.8821337223052979,
      "learning_rate": 9.863750555389473e-06,
      "loss": 0.8918,
      "step": 2799
    },
    {
      "epoch": 0.15410864659585008,
      "grad_norm": 0.7896875143051147,
      "learning_rate": 9.863650035260742e-06,
      "loss": 0.8199,
      "step": 2800
    },
    {
      "epoch": 0.15416368539820574,
      "grad_norm": 0.8046941161155701,
      "learning_rate": 9.86354947857812e-06,
      "loss": 0.8572,
      "step": 2801
    },
    {
      "epoch": 0.1542187242005614,
      "grad_norm": 0.7266830205917358,
      "learning_rate": 9.863448885342361e-06,
      "loss": 0.8315,
      "step": 2802
    },
    {
      "epoch": 0.15427376300291706,
      "grad_norm": 0.9009475708007812,
      "learning_rate": 9.863348255554222e-06,
      "loss": 0.7928,
      "step": 2803
    },
    {
      "epoch": 0.15432880180527273,
      "grad_norm": 0.963364839553833,
      "learning_rate": 9.863247589214459e-06,
      "loss": 0.918,
      "step": 2804
    },
    {
      "epoch": 0.1543838406076284,
      "grad_norm": 0.8278035521507263,
      "learning_rate": 9.863146886323829e-06,
      "loss": 0.8497,
      "step": 2805
    },
    {
      "epoch": 0.15443887940998405,
      "grad_norm": 0.7360561490058899,
      "learning_rate": 9.86304614688309e-06,
      "loss": 0.676,
      "step": 2806
    },
    {
      "epoch": 0.1544939182123397,
      "grad_norm": 0.7679837346076965,
      "learning_rate": 9.862945370892996e-06,
      "loss": 0.8114,
      "step": 2807
    },
    {
      "epoch": 0.15454895701469537,
      "grad_norm": 0.8550567030906677,
      "learning_rate": 9.862844558354309e-06,
      "loss": 0.8222,
      "step": 2808
    },
    {
      "epoch": 0.15460399581705103,
      "grad_norm": 0.7852397561073303,
      "learning_rate": 9.86274370926778e-06,
      "loss": 0.7449,
      "step": 2809
    },
    {
      "epoch": 0.1546590346194067,
      "grad_norm": 0.9120833277702332,
      "learning_rate": 9.862642823634175e-06,
      "loss": 0.8702,
      "step": 2810
    },
    {
      "epoch": 0.15471407342176235,
      "grad_norm": 0.8729703426361084,
      "learning_rate": 9.862541901454246e-06,
      "loss": 0.8064,
      "step": 2811
    },
    {
      "epoch": 0.15476911222411802,
      "grad_norm": 0.7935470342636108,
      "learning_rate": 9.862440942728754e-06,
      "loss": 0.8502,
      "step": 2812
    },
    {
      "epoch": 0.15482415102647368,
      "grad_norm": 0.8640689849853516,
      "learning_rate": 9.86233994745846e-06,
      "loss": 0.8159,
      "step": 2813
    },
    {
      "epoch": 0.1548791898288293,
      "grad_norm": 0.9959222078323364,
      "learning_rate": 9.862238915644116e-06,
      "loss": 0.7767,
      "step": 2814
    },
    {
      "epoch": 0.15493422863118497,
      "grad_norm": 0.7889506220817566,
      "learning_rate": 9.862137847286487e-06,
      "loss": 0.8293,
      "step": 2815
    },
    {
      "epoch": 0.15498926743354063,
      "grad_norm": 0.8764606714248657,
      "learning_rate": 9.862036742386335e-06,
      "loss": 0.856,
      "step": 2816
    },
    {
      "epoch": 0.1550443062358963,
      "grad_norm": 0.743727445602417,
      "learning_rate": 9.861935600944413e-06,
      "loss": 0.7099,
      "step": 2817
    },
    {
      "epoch": 0.15509934503825196,
      "grad_norm": 0.7866224050521851,
      "learning_rate": 9.861834422961485e-06,
      "loss": 0.8805,
      "step": 2818
    },
    {
      "epoch": 0.15515438384060762,
      "grad_norm": 0.8333723545074463,
      "learning_rate": 9.861733208438311e-06,
      "loss": 0.8486,
      "step": 2819
    },
    {
      "epoch": 0.15520942264296328,
      "grad_norm": 0.8261659741401672,
      "learning_rate": 9.861631957375652e-06,
      "loss": 0.8896,
      "step": 2820
    },
    {
      "epoch": 0.15526446144531894,
      "grad_norm": 0.8381538987159729,
      "learning_rate": 9.861530669774268e-06,
      "loss": 0.8686,
      "step": 2821
    },
    {
      "epoch": 0.1553195002476746,
      "grad_norm": 0.9184440970420837,
      "learning_rate": 9.861429345634923e-06,
      "loss": 0.9702,
      "step": 2822
    },
    {
      "epoch": 0.15537453905003026,
      "grad_norm": 0.8170294165611267,
      "learning_rate": 9.861327984958374e-06,
      "loss": 0.8298,
      "step": 2823
    },
    {
      "epoch": 0.15542957785238593,
      "grad_norm": 0.8361968398094177,
      "learning_rate": 9.861226587745385e-06,
      "loss": 0.8232,
      "step": 2824
    },
    {
      "epoch": 0.1554846166547416,
      "grad_norm": 0.7437820434570312,
      "learning_rate": 9.861125153996718e-06,
      "loss": 0.8271,
      "step": 2825
    },
    {
      "epoch": 0.15553965545709725,
      "grad_norm": 0.715887188911438,
      "learning_rate": 9.861023683713137e-06,
      "loss": 0.7726,
      "step": 2826
    },
    {
      "epoch": 0.1555946942594529,
      "grad_norm": 0.8358462452888489,
      "learning_rate": 9.860922176895403e-06,
      "loss": 0.8247,
      "step": 2827
    },
    {
      "epoch": 0.15564973306180857,
      "grad_norm": 0.8620158433914185,
      "learning_rate": 9.860820633544278e-06,
      "loss": 0.8804,
      "step": 2828
    },
    {
      "epoch": 0.15570477186416423,
      "grad_norm": 0.9035346508026123,
      "learning_rate": 9.860719053660527e-06,
      "loss": 0.7973,
      "step": 2829
    },
    {
      "epoch": 0.1557598106665199,
      "grad_norm": 0.8014782071113586,
      "learning_rate": 9.860617437244914e-06,
      "loss": 0.7914,
      "step": 2830
    },
    {
      "epoch": 0.15581484946887555,
      "grad_norm": 0.7788864970207214,
      "learning_rate": 9.8605157842982e-06,
      "loss": 0.7377,
      "step": 2831
    },
    {
      "epoch": 0.15586988827123122,
      "grad_norm": 0.7475222945213318,
      "learning_rate": 9.860414094821152e-06,
      "loss": 0.7173,
      "step": 2832
    },
    {
      "epoch": 0.15592492707358688,
      "grad_norm": 0.8866652846336365,
      "learning_rate": 9.86031236881453e-06,
      "loss": 0.8231,
      "step": 2833
    },
    {
      "epoch": 0.15597996587594254,
      "grad_norm": 0.8725677728652954,
      "learning_rate": 9.860210606279102e-06,
      "loss": 0.9025,
      "step": 2834
    },
    {
      "epoch": 0.1560350046782982,
      "grad_norm": 0.7608423233032227,
      "learning_rate": 9.860108807215634e-06,
      "loss": 0.8385,
      "step": 2835
    },
    {
      "epoch": 0.15609004348065386,
      "grad_norm": 0.8237566351890564,
      "learning_rate": 9.860006971624887e-06,
      "loss": 0.8635,
      "step": 2836
    },
    {
      "epoch": 0.15614508228300952,
      "grad_norm": 0.8078347444534302,
      "learning_rate": 9.859905099507629e-06,
      "loss": 0.7916,
      "step": 2837
    },
    {
      "epoch": 0.15620012108536518,
      "grad_norm": 0.8282070755958557,
      "learning_rate": 9.859803190864626e-06,
      "loss": 0.8141,
      "step": 2838
    },
    {
      "epoch": 0.15625515988772085,
      "grad_norm": 0.7639191150665283,
      "learning_rate": 9.859701245696642e-06,
      "loss": 0.7457,
      "step": 2839
    },
    {
      "epoch": 0.1563101986900765,
      "grad_norm": 0.8429144620895386,
      "learning_rate": 9.859599264004446e-06,
      "loss": 0.9176,
      "step": 2840
    },
    {
      "epoch": 0.15636523749243217,
      "grad_norm": 0.7792791724205017,
      "learning_rate": 9.859497245788801e-06,
      "loss": 0.8738,
      "step": 2841
    },
    {
      "epoch": 0.15642027629478783,
      "grad_norm": 0.9018417596817017,
      "learning_rate": 9.859395191050476e-06,
      "loss": 0.841,
      "step": 2842
    },
    {
      "epoch": 0.1564753150971435,
      "grad_norm": 0.7113705277442932,
      "learning_rate": 9.859293099790239e-06,
      "loss": 0.6576,
      "step": 2843
    },
    {
      "epoch": 0.15653035389949915,
      "grad_norm": 0.8376311659812927,
      "learning_rate": 9.859190972008853e-06,
      "loss": 0.8559,
      "step": 2844
    },
    {
      "epoch": 0.15658539270185481,
      "grad_norm": 0.7689141035079956,
      "learning_rate": 9.859088807707092e-06,
      "loss": 0.7844,
      "step": 2845
    },
    {
      "epoch": 0.15664043150421048,
      "grad_norm": 0.7559483647346497,
      "learning_rate": 9.858986606885717e-06,
      "loss": 0.8676,
      "step": 2846
    },
    {
      "epoch": 0.15669547030656614,
      "grad_norm": 0.7743827700614929,
      "learning_rate": 9.8588843695455e-06,
      "loss": 0.7995,
      "step": 2847
    },
    {
      "epoch": 0.1567505091089218,
      "grad_norm": 0.8631327152252197,
      "learning_rate": 9.85878209568721e-06,
      "loss": 0.801,
      "step": 2848
    },
    {
      "epoch": 0.15680554791127746,
      "grad_norm": 0.7454009056091309,
      "learning_rate": 9.858679785311613e-06,
      "loss": 0.8172,
      "step": 2849
    },
    {
      "epoch": 0.15686058671363312,
      "grad_norm": 0.7915313839912415,
      "learning_rate": 9.858577438419479e-06,
      "loss": 0.833,
      "step": 2850
    },
    {
      "epoch": 0.15691562551598878,
      "grad_norm": 0.8472526669502258,
      "learning_rate": 9.858475055011578e-06,
      "loss": 0.8249,
      "step": 2851
    },
    {
      "epoch": 0.15697066431834444,
      "grad_norm": 0.7967580556869507,
      "learning_rate": 9.85837263508868e-06,
      "loss": 0.7533,
      "step": 2852
    },
    {
      "epoch": 0.1570257031207001,
      "grad_norm": 0.7476257085800171,
      "learning_rate": 9.858270178651554e-06,
      "loss": 0.7918,
      "step": 2853
    },
    {
      "epoch": 0.15708074192305577,
      "grad_norm": 0.8736184239387512,
      "learning_rate": 9.858167685700968e-06,
      "loss": 0.8254,
      "step": 2854
    },
    {
      "epoch": 0.15713578072541143,
      "grad_norm": 0.8734819889068604,
      "learning_rate": 9.858065156237694e-06,
      "loss": 0.749,
      "step": 2855
    },
    {
      "epoch": 0.1571908195277671,
      "grad_norm": 1.0344874858856201,
      "learning_rate": 9.857962590262506e-06,
      "loss": 0.9578,
      "step": 2856
    },
    {
      "epoch": 0.15724585833012272,
      "grad_norm": 0.81183922290802,
      "learning_rate": 9.857859987776168e-06,
      "loss": 0.8845,
      "step": 2857
    },
    {
      "epoch": 0.15730089713247838,
      "grad_norm": 0.8252540230751038,
      "learning_rate": 9.857757348779456e-06,
      "loss": 0.7862,
      "step": 2858
    },
    {
      "epoch": 0.15735593593483405,
      "grad_norm": 0.7468119859695435,
      "learning_rate": 9.85765467327314e-06,
      "loss": 0.7587,
      "step": 2859
    },
    {
      "epoch": 0.1574109747371897,
      "grad_norm": 0.8095998167991638,
      "learning_rate": 9.857551961257993e-06,
      "loss": 0.7467,
      "step": 2860
    },
    {
      "epoch": 0.15746601353954537,
      "grad_norm": 0.8908564448356628,
      "learning_rate": 9.857449212734785e-06,
      "loss": 0.8199,
      "step": 2861
    },
    {
      "epoch": 0.15752105234190103,
      "grad_norm": 0.7605593204498291,
      "learning_rate": 9.857346427704288e-06,
      "loss": 0.7196,
      "step": 2862
    },
    {
      "epoch": 0.1575760911442567,
      "grad_norm": 0.9250784516334534,
      "learning_rate": 9.857243606167276e-06,
      "loss": 0.7366,
      "step": 2863
    },
    {
      "epoch": 0.15763112994661235,
      "grad_norm": 0.882796585559845,
      "learning_rate": 9.85714074812452e-06,
      "loss": 0.8422,
      "step": 2864
    },
    {
      "epoch": 0.15768616874896801,
      "grad_norm": 1.0014574527740479,
      "learning_rate": 9.857037853576797e-06,
      "loss": 0.8762,
      "step": 2865
    },
    {
      "epoch": 0.15774120755132368,
      "grad_norm": 0.86713045835495,
      "learning_rate": 9.856934922524877e-06,
      "loss": 0.9282,
      "step": 2866
    },
    {
      "epoch": 0.15779624635367934,
      "grad_norm": 1.1457390785217285,
      "learning_rate": 9.856831954969532e-06,
      "loss": 0.7947,
      "step": 2867
    },
    {
      "epoch": 0.157851285156035,
      "grad_norm": 0.8902556896209717,
      "learning_rate": 9.85672895091154e-06,
      "loss": 0.928,
      "step": 2868
    },
    {
      "epoch": 0.15790632395839066,
      "grad_norm": 0.7978467345237732,
      "learning_rate": 9.856625910351674e-06,
      "loss": 0.7382,
      "step": 2869
    },
    {
      "epoch": 0.15796136276074632,
      "grad_norm": 0.741457462310791,
      "learning_rate": 9.856522833290705e-06,
      "loss": 0.7736,
      "step": 2870
    },
    {
      "epoch": 0.15801640156310198,
      "grad_norm": 0.8330628871917725,
      "learning_rate": 9.856419719729413e-06,
      "loss": 0.8396,
      "step": 2871
    },
    {
      "epoch": 0.15807144036545764,
      "grad_norm": 0.8771876692771912,
      "learning_rate": 9.85631656966857e-06,
      "loss": 0.6669,
      "step": 2872
    },
    {
      "epoch": 0.1581264791678133,
      "grad_norm": 0.8073394298553467,
      "learning_rate": 9.85621338310895e-06,
      "loss": 0.8206,
      "step": 2873
    },
    {
      "epoch": 0.15818151797016897,
      "grad_norm": 1.1058349609375,
      "learning_rate": 9.85611016005133e-06,
      "loss": 0.9526,
      "step": 2874
    },
    {
      "epoch": 0.15823655677252463,
      "grad_norm": 0.7734992504119873,
      "learning_rate": 9.856006900496488e-06,
      "loss": 0.7477,
      "step": 2875
    },
    {
      "epoch": 0.1582915955748803,
      "grad_norm": 0.9053219556808472,
      "learning_rate": 9.855903604445196e-06,
      "loss": 0.8009,
      "step": 2876
    },
    {
      "epoch": 0.15834663437723595,
      "grad_norm": 0.8774041533470154,
      "learning_rate": 9.855800271898233e-06,
      "loss": 0.854,
      "step": 2877
    },
    {
      "epoch": 0.1584016731795916,
      "grad_norm": 0.8346550464630127,
      "learning_rate": 9.855696902856376e-06,
      "loss": 0.7976,
      "step": 2878
    },
    {
      "epoch": 0.15845671198194727,
      "grad_norm": 0.7781139016151428,
      "learning_rate": 9.855593497320401e-06,
      "loss": 0.7693,
      "step": 2879
    },
    {
      "epoch": 0.15851175078430293,
      "grad_norm": 0.8707864880561829,
      "learning_rate": 9.855490055291084e-06,
      "loss": 0.882,
      "step": 2880
    },
    {
      "epoch": 0.1585667895866586,
      "grad_norm": 0.7982275485992432,
      "learning_rate": 9.855386576769203e-06,
      "loss": 0.8457,
      "step": 2881
    },
    {
      "epoch": 0.15862182838901426,
      "grad_norm": 0.7577090263366699,
      "learning_rate": 9.855283061755536e-06,
      "loss": 0.754,
      "step": 2882
    },
    {
      "epoch": 0.15867686719136992,
      "grad_norm": 0.7422069311141968,
      "learning_rate": 9.855179510250863e-06,
      "loss": 0.673,
      "step": 2883
    },
    {
      "epoch": 0.15873190599372558,
      "grad_norm": 0.7730041742324829,
      "learning_rate": 9.85507592225596e-06,
      "loss": 0.7888,
      "step": 2884
    },
    {
      "epoch": 0.15878694479608124,
      "grad_norm": 0.7370560169219971,
      "learning_rate": 9.854972297771605e-06,
      "loss": 0.7762,
      "step": 2885
    },
    {
      "epoch": 0.1588419835984369,
      "grad_norm": 0.725074291229248,
      "learning_rate": 9.854868636798577e-06,
      "loss": 0.7951,
      "step": 2886
    },
    {
      "epoch": 0.15889702240079256,
      "grad_norm": 0.8088375926017761,
      "learning_rate": 9.854764939337657e-06,
      "loss": 0.8557,
      "step": 2887
    },
    {
      "epoch": 0.15895206120314823,
      "grad_norm": 0.8268256187438965,
      "learning_rate": 9.854661205389624e-06,
      "loss": 0.7641,
      "step": 2888
    },
    {
      "epoch": 0.1590071000055039,
      "grad_norm": 0.8079462051391602,
      "learning_rate": 9.854557434955257e-06,
      "loss": 0.7947,
      "step": 2889
    },
    {
      "epoch": 0.15906213880785955,
      "grad_norm": 0.8257912993431091,
      "learning_rate": 9.854453628035335e-06,
      "loss": 0.771,
      "step": 2890
    },
    {
      "epoch": 0.1591171776102152,
      "grad_norm": 0.8901774287223816,
      "learning_rate": 9.85434978463064e-06,
      "loss": 0.9415,
      "step": 2891
    },
    {
      "epoch": 0.15917221641257087,
      "grad_norm": 0.8283013105392456,
      "learning_rate": 9.854245904741948e-06,
      "loss": 0.7267,
      "step": 2892
    },
    {
      "epoch": 0.15922725521492653,
      "grad_norm": 0.8665382266044617,
      "learning_rate": 9.854141988370045e-06,
      "loss": 0.8681,
      "step": 2893
    },
    {
      "epoch": 0.1592822940172822,
      "grad_norm": 0.786494255065918,
      "learning_rate": 9.854038035515712e-06,
      "loss": 0.7614,
      "step": 2894
    },
    {
      "epoch": 0.15933733281963786,
      "grad_norm": 1.0548759698867798,
      "learning_rate": 9.853934046179727e-06,
      "loss": 0.861,
      "step": 2895
    },
    {
      "epoch": 0.15939237162199352,
      "grad_norm": 0.8565425276756287,
      "learning_rate": 9.853830020362873e-06,
      "loss": 0.7858,
      "step": 2896
    },
    {
      "epoch": 0.15944741042434918,
      "grad_norm": 0.7982691526412964,
      "learning_rate": 9.853725958065933e-06,
      "loss": 0.8797,
      "step": 2897
    },
    {
      "epoch": 0.15950244922670484,
      "grad_norm": 0.8613169193267822,
      "learning_rate": 9.853621859289686e-06,
      "loss": 0.9217,
      "step": 2898
    },
    {
      "epoch": 0.1595574880290605,
      "grad_norm": 0.950639009475708,
      "learning_rate": 9.853517724034918e-06,
      "loss": 0.8315,
      "step": 2899
    },
    {
      "epoch": 0.15961252683141613,
      "grad_norm": 0.7940176129341125,
      "learning_rate": 9.853413552302409e-06,
      "loss": 0.7713,
      "step": 2900
    },
    {
      "epoch": 0.1596675656337718,
      "grad_norm": 0.7716153264045715,
      "learning_rate": 9.853309344092944e-06,
      "loss": 0.7922,
      "step": 2901
    },
    {
      "epoch": 0.15972260443612746,
      "grad_norm": 0.7626190781593323,
      "learning_rate": 9.853205099407303e-06,
      "loss": 0.7278,
      "step": 2902
    },
    {
      "epoch": 0.15977764323848312,
      "grad_norm": 0.8523116707801819,
      "learning_rate": 9.853100818246274e-06,
      "loss": 0.8136,
      "step": 2903
    },
    {
      "epoch": 0.15983268204083878,
      "grad_norm": 0.7636643052101135,
      "learning_rate": 9.852996500610637e-06,
      "loss": 0.6984,
      "step": 2904
    },
    {
      "epoch": 0.15988772084319444,
      "grad_norm": 0.799201250076294,
      "learning_rate": 9.852892146501179e-06,
      "loss": 0.8319,
      "step": 2905
    },
    {
      "epoch": 0.1599427596455501,
      "grad_norm": 0.7743694186210632,
      "learning_rate": 9.85278775591868e-06,
      "loss": 0.81,
      "step": 2906
    },
    {
      "epoch": 0.15999779844790576,
      "grad_norm": 0.8964856863021851,
      "learning_rate": 9.85268332886393e-06,
      "loss": 0.9227,
      "step": 2907
    },
    {
      "epoch": 0.16005283725026143,
      "grad_norm": 0.8809369802474976,
      "learning_rate": 9.852578865337708e-06,
      "loss": 0.9285,
      "step": 2908
    },
    {
      "epoch": 0.1601078760526171,
      "grad_norm": 0.8960002064704895,
      "learning_rate": 9.852474365340806e-06,
      "loss": 0.8611,
      "step": 2909
    },
    {
      "epoch": 0.16016291485497275,
      "grad_norm": 0.7539754509925842,
      "learning_rate": 9.852369828874002e-06,
      "loss": 0.7455,
      "step": 2910
    },
    {
      "epoch": 0.1602179536573284,
      "grad_norm": 0.8189692497253418,
      "learning_rate": 9.852265255938088e-06,
      "loss": 0.8321,
      "step": 2911
    },
    {
      "epoch": 0.16027299245968407,
      "grad_norm": 0.8708549737930298,
      "learning_rate": 9.852160646533844e-06,
      "loss": 0.8373,
      "step": 2912
    },
    {
      "epoch": 0.16032803126203973,
      "grad_norm": 0.7701451778411865,
      "learning_rate": 9.852056000662063e-06,
      "loss": 0.805,
      "step": 2913
    },
    {
      "epoch": 0.1603830700643954,
      "grad_norm": 0.9111948609352112,
      "learning_rate": 9.851951318323526e-06,
      "loss": 0.8513,
      "step": 2914
    },
    {
      "epoch": 0.16043810886675106,
      "grad_norm": 0.7863909602165222,
      "learning_rate": 9.85184659951902e-06,
      "loss": 0.7856,
      "step": 2915
    },
    {
      "epoch": 0.16049314766910672,
      "grad_norm": 0.9000817537307739,
      "learning_rate": 9.851741844249336e-06,
      "loss": 0.9172,
      "step": 2916
    },
    {
      "epoch": 0.16054818647146238,
      "grad_norm": 1.0953118801116943,
      "learning_rate": 9.851637052515259e-06,
      "loss": 0.8564,
      "step": 2917
    },
    {
      "epoch": 0.16060322527381804,
      "grad_norm": 0.8405389785766602,
      "learning_rate": 9.851532224317575e-06,
      "loss": 0.8317,
      "step": 2918
    },
    {
      "epoch": 0.1606582640761737,
      "grad_norm": 0.8524565100669861,
      "learning_rate": 9.851427359657075e-06,
      "loss": 0.8765,
      "step": 2919
    },
    {
      "epoch": 0.16071330287852936,
      "grad_norm": 0.8234089016914368,
      "learning_rate": 9.851322458534546e-06,
      "loss": 0.7873,
      "step": 2920
    },
    {
      "epoch": 0.16076834168088502,
      "grad_norm": 0.7879638671875,
      "learning_rate": 9.851217520950775e-06,
      "loss": 0.8394,
      "step": 2921
    },
    {
      "epoch": 0.16082338048324069,
      "grad_norm": 0.8168820738792419,
      "learning_rate": 9.851112546906552e-06,
      "loss": 0.8223,
      "step": 2922
    },
    {
      "epoch": 0.16087841928559635,
      "grad_norm": 0.9423845410346985,
      "learning_rate": 9.851007536402666e-06,
      "loss": 0.9256,
      "step": 2923
    },
    {
      "epoch": 0.160933458087952,
      "grad_norm": 0.7875099778175354,
      "learning_rate": 9.850902489439906e-06,
      "loss": 0.8199,
      "step": 2924
    },
    {
      "epoch": 0.16098849689030767,
      "grad_norm": 0.6934793591499329,
      "learning_rate": 9.85079740601906e-06,
      "loss": 0.671,
      "step": 2925
    },
    {
      "epoch": 0.16104353569266333,
      "grad_norm": 0.8172206282615662,
      "learning_rate": 9.85069228614092e-06,
      "loss": 0.7633,
      "step": 2926
    },
    {
      "epoch": 0.161098574495019,
      "grad_norm": 0.72749263048172,
      "learning_rate": 9.850587129806274e-06,
      "loss": 0.8719,
      "step": 2927
    },
    {
      "epoch": 0.16115361329737465,
      "grad_norm": 0.8416743874549866,
      "learning_rate": 9.850481937015917e-06,
      "loss": 0.8438,
      "step": 2928
    },
    {
      "epoch": 0.16120865209973032,
      "grad_norm": 0.7415444850921631,
      "learning_rate": 9.850376707770633e-06,
      "loss": 0.7673,
      "step": 2929
    },
    {
      "epoch": 0.16126369090208598,
      "grad_norm": 0.9364289045333862,
      "learning_rate": 9.850271442071217e-06,
      "loss": 0.7224,
      "step": 2930
    },
    {
      "epoch": 0.16131872970444164,
      "grad_norm": 0.7314212918281555,
      "learning_rate": 9.85016613991846e-06,
      "loss": 0.7759,
      "step": 2931
    },
    {
      "epoch": 0.1613737685067973,
      "grad_norm": 0.8940219283103943,
      "learning_rate": 9.850060801313151e-06,
      "loss": 0.8432,
      "step": 2932
    },
    {
      "epoch": 0.16142880730915296,
      "grad_norm": 0.7499691843986511,
      "learning_rate": 9.849955426256084e-06,
      "loss": 0.8171,
      "step": 2933
    },
    {
      "epoch": 0.16148384611150862,
      "grad_norm": 0.8123053312301636,
      "learning_rate": 9.84985001474805e-06,
      "loss": 0.7839,
      "step": 2934
    },
    {
      "epoch": 0.16153888491386428,
      "grad_norm": 0.819618821144104,
      "learning_rate": 9.849744566789842e-06,
      "loss": 0.9123,
      "step": 2935
    },
    {
      "epoch": 0.16159392371621994,
      "grad_norm": 0.791088342666626,
      "learning_rate": 9.849639082382251e-06,
      "loss": 0.8347,
      "step": 2936
    },
    {
      "epoch": 0.1616489625185756,
      "grad_norm": 0.8166706562042236,
      "learning_rate": 9.849533561526072e-06,
      "loss": 0.8309,
      "step": 2937
    },
    {
      "epoch": 0.16170400132093127,
      "grad_norm": 0.7944774031639099,
      "learning_rate": 9.849428004222098e-06,
      "loss": 0.8387,
      "step": 2938
    },
    {
      "epoch": 0.16175904012328693,
      "grad_norm": 0.7414719462394714,
      "learning_rate": 9.849322410471119e-06,
      "loss": 0.71,
      "step": 2939
    },
    {
      "epoch": 0.1618140789256426,
      "grad_norm": 0.8983511924743652,
      "learning_rate": 9.849216780273931e-06,
      "loss": 0.8902,
      "step": 2940
    },
    {
      "epoch": 0.16186911772799825,
      "grad_norm": 0.9058687686920166,
      "learning_rate": 9.849111113631329e-06,
      "loss": 0.8804,
      "step": 2941
    },
    {
      "epoch": 0.1619241565303539,
      "grad_norm": 0.948871374130249,
      "learning_rate": 9.849005410544105e-06,
      "loss": 0.9871,
      "step": 2942
    },
    {
      "epoch": 0.16197919533270955,
      "grad_norm": 0.8240115642547607,
      "learning_rate": 9.848899671013055e-06,
      "loss": 0.8708,
      "step": 2943
    },
    {
      "epoch": 0.1620342341350652,
      "grad_norm": 0.879953145980835,
      "learning_rate": 9.848793895038972e-06,
      "loss": 0.9279,
      "step": 2944
    },
    {
      "epoch": 0.16208927293742087,
      "grad_norm": 0.8464690446853638,
      "learning_rate": 9.848688082622653e-06,
      "loss": 0.8418,
      "step": 2945
    },
    {
      "epoch": 0.16214431173977653,
      "grad_norm": 0.8990732431411743,
      "learning_rate": 9.848582233764891e-06,
      "loss": 0.8622,
      "step": 2946
    },
    {
      "epoch": 0.1621993505421322,
      "grad_norm": 0.8054911494255066,
      "learning_rate": 9.848476348466483e-06,
      "loss": 0.8295,
      "step": 2947
    },
    {
      "epoch": 0.16225438934448785,
      "grad_norm": 0.7904845476150513,
      "learning_rate": 9.848370426728226e-06,
      "loss": 0.7777,
      "step": 2948
    },
    {
      "epoch": 0.16230942814684352,
      "grad_norm": 1.0143954753875732,
      "learning_rate": 9.848264468550915e-06,
      "loss": 0.8556,
      "step": 2949
    },
    {
      "epoch": 0.16236446694919918,
      "grad_norm": 0.7201125621795654,
      "learning_rate": 9.848158473935344e-06,
      "loss": 0.7981,
      "step": 2950
    },
    {
      "epoch": 0.16241950575155484,
      "grad_norm": 0.8322157263755798,
      "learning_rate": 9.848052442882312e-06,
      "loss": 0.8323,
      "step": 2951
    },
    {
      "epoch": 0.1624745445539105,
      "grad_norm": 0.7740346193313599,
      "learning_rate": 9.847946375392617e-06,
      "loss": 0.8355,
      "step": 2952
    },
    {
      "epoch": 0.16252958335626616,
      "grad_norm": 0.8955645561218262,
      "learning_rate": 9.847840271467053e-06,
      "loss": 0.7161,
      "step": 2953
    },
    {
      "epoch": 0.16258462215862182,
      "grad_norm": 0.800364077091217,
      "learning_rate": 9.847734131106421e-06,
      "loss": 0.8165,
      "step": 2954
    },
    {
      "epoch": 0.16263966096097748,
      "grad_norm": 0.8305484056472778,
      "learning_rate": 9.847627954311516e-06,
      "loss": 0.7846,
      "step": 2955
    },
    {
      "epoch": 0.16269469976333314,
      "grad_norm": 0.7354590892791748,
      "learning_rate": 9.847521741083136e-06,
      "loss": 0.7743,
      "step": 2956
    },
    {
      "epoch": 0.1627497385656888,
      "grad_norm": 0.8173812627792358,
      "learning_rate": 9.847415491422083e-06,
      "loss": 0.8626,
      "step": 2957
    },
    {
      "epoch": 0.16280477736804447,
      "grad_norm": 0.7959356307983398,
      "learning_rate": 9.84730920532915e-06,
      "loss": 0.8016,
      "step": 2958
    },
    {
      "epoch": 0.16285981617040013,
      "grad_norm": 0.8256500363349915,
      "learning_rate": 9.84720288280514e-06,
      "loss": 0.7407,
      "step": 2959
    },
    {
      "epoch": 0.1629148549727558,
      "grad_norm": 0.8522148728370667,
      "learning_rate": 9.84709652385085e-06,
      "loss": 0.8342,
      "step": 2960
    },
    {
      "epoch": 0.16296989377511145,
      "grad_norm": 0.7791039943695068,
      "learning_rate": 9.84699012846708e-06,
      "loss": 0.7239,
      "step": 2961
    },
    {
      "epoch": 0.1630249325774671,
      "grad_norm": 0.84294193983078,
      "learning_rate": 9.84688369665463e-06,
      "loss": 0.7498,
      "step": 2962
    },
    {
      "epoch": 0.16307997137982277,
      "grad_norm": 0.7948899865150452,
      "learning_rate": 9.846777228414299e-06,
      "loss": 0.7713,
      "step": 2963
    },
    {
      "epoch": 0.16313501018217844,
      "grad_norm": 0.6673180460929871,
      "learning_rate": 9.846670723746888e-06,
      "loss": 0.6759,
      "step": 2964
    },
    {
      "epoch": 0.1631900489845341,
      "grad_norm": 0.8141015768051147,
      "learning_rate": 9.846564182653199e-06,
      "loss": 0.7928,
      "step": 2965
    },
    {
      "epoch": 0.16324508778688976,
      "grad_norm": 0.967830240726471,
      "learning_rate": 9.846457605134028e-06,
      "loss": 0.823,
      "step": 2966
    },
    {
      "epoch": 0.16330012658924542,
      "grad_norm": 0.8099361658096313,
      "learning_rate": 9.84635099119018e-06,
      "loss": 0.8724,
      "step": 2967
    },
    {
      "epoch": 0.16335516539160108,
      "grad_norm": 0.7913978099822998,
      "learning_rate": 9.846244340822456e-06,
      "loss": 0.7106,
      "step": 2968
    },
    {
      "epoch": 0.16341020419395674,
      "grad_norm": 0.8308563828468323,
      "learning_rate": 9.846137654031655e-06,
      "loss": 0.7631,
      "step": 2969
    },
    {
      "epoch": 0.1634652429963124,
      "grad_norm": 0.8634191751480103,
      "learning_rate": 9.846030930818582e-06,
      "loss": 0.7363,
      "step": 2970
    },
    {
      "epoch": 0.16352028179866807,
      "grad_norm": 0.8936432600021362,
      "learning_rate": 9.845924171184038e-06,
      "loss": 0.8714,
      "step": 2971
    },
    {
      "epoch": 0.16357532060102373,
      "grad_norm": 0.8776300549507141,
      "learning_rate": 9.845817375128825e-06,
      "loss": 0.914,
      "step": 2972
    },
    {
      "epoch": 0.1636303594033794,
      "grad_norm": 0.8793039321899414,
      "learning_rate": 9.845710542653745e-06,
      "loss": 0.7999,
      "step": 2973
    },
    {
      "epoch": 0.16368539820573505,
      "grad_norm": 0.8391743302345276,
      "learning_rate": 9.845603673759603e-06,
      "loss": 0.8124,
      "step": 2974
    },
    {
      "epoch": 0.1637404370080907,
      "grad_norm": 0.8487186431884766,
      "learning_rate": 9.845496768447199e-06,
      "loss": 0.8551,
      "step": 2975
    },
    {
      "epoch": 0.16379547581044637,
      "grad_norm": 0.7780638933181763,
      "learning_rate": 9.845389826717339e-06,
      "loss": 0.7281,
      "step": 2976
    },
    {
      "epoch": 0.16385051461280203,
      "grad_norm": 0.7209637761116028,
      "learning_rate": 9.845282848570825e-06,
      "loss": 0.6737,
      "step": 2977
    },
    {
      "epoch": 0.1639055534151577,
      "grad_norm": 0.8414756059646606,
      "learning_rate": 9.845175834008464e-06,
      "loss": 0.8003,
      "step": 2978
    },
    {
      "epoch": 0.16396059221751336,
      "grad_norm": 1.2730679512023926,
      "learning_rate": 9.845068783031057e-06,
      "loss": 0.8243,
      "step": 2979
    },
    {
      "epoch": 0.16401563101986902,
      "grad_norm": 0.8573475480079651,
      "learning_rate": 9.844961695639413e-06,
      "loss": 0.7844,
      "step": 2980
    },
    {
      "epoch": 0.16407066982222468,
      "grad_norm": 0.8029958605766296,
      "learning_rate": 9.84485457183433e-06,
      "loss": 0.7722,
      "step": 2981
    },
    {
      "epoch": 0.16412570862458034,
      "grad_norm": 0.7839805483818054,
      "learning_rate": 9.844747411616619e-06,
      "loss": 0.8146,
      "step": 2982
    },
    {
      "epoch": 0.164180747426936,
      "grad_norm": 0.7563499212265015,
      "learning_rate": 9.844640214987082e-06,
      "loss": 0.6909,
      "step": 2983
    },
    {
      "epoch": 0.16423578622929166,
      "grad_norm": 0.7199193239212036,
      "learning_rate": 9.844532981946527e-06,
      "loss": 0.702,
      "step": 2984
    },
    {
      "epoch": 0.16429082503164733,
      "grad_norm": 0.7519383430480957,
      "learning_rate": 9.844425712495758e-06,
      "loss": 0.6493,
      "step": 2985
    },
    {
      "epoch": 0.16434586383400296,
      "grad_norm": 0.7493193745613098,
      "learning_rate": 9.844318406635584e-06,
      "loss": 0.8318,
      "step": 2986
    },
    {
      "epoch": 0.16440090263635862,
      "grad_norm": 0.7951106429100037,
      "learning_rate": 9.84421106436681e-06,
      "loss": 0.923,
      "step": 2987
    },
    {
      "epoch": 0.16445594143871428,
      "grad_norm": 0.8350820541381836,
      "learning_rate": 9.844103685690238e-06,
      "loss": 0.8091,
      "step": 2988
    },
    {
      "epoch": 0.16451098024106994,
      "grad_norm": 0.773932695388794,
      "learning_rate": 9.843996270606683e-06,
      "loss": 0.8016,
      "step": 2989
    },
    {
      "epoch": 0.1645660190434256,
      "grad_norm": 0.8208432793617249,
      "learning_rate": 9.843888819116947e-06,
      "loss": 0.7704,
      "step": 2990
    },
    {
      "epoch": 0.16462105784578127,
      "grad_norm": 0.8552223443984985,
      "learning_rate": 9.84378133122184e-06,
      "loss": 0.8519,
      "step": 2991
    },
    {
      "epoch": 0.16467609664813693,
      "grad_norm": 0.8015661835670471,
      "learning_rate": 9.84367380692217e-06,
      "loss": 0.8389,
      "step": 2992
    },
    {
      "epoch": 0.1647311354504926,
      "grad_norm": 0.7828749418258667,
      "learning_rate": 9.843566246218743e-06,
      "loss": 0.7385,
      "step": 2993
    },
    {
      "epoch": 0.16478617425284825,
      "grad_norm": 0.7761647701263428,
      "learning_rate": 9.84345864911237e-06,
      "loss": 0.8419,
      "step": 2994
    },
    {
      "epoch": 0.1648412130552039,
      "grad_norm": 0.8839839100837708,
      "learning_rate": 9.843351015603857e-06,
      "loss": 0.8069,
      "step": 2995
    },
    {
      "epoch": 0.16489625185755957,
      "grad_norm": 0.8611735105514526,
      "learning_rate": 9.843243345694014e-06,
      "loss": 0.9406,
      "step": 2996
    },
    {
      "epoch": 0.16495129065991523,
      "grad_norm": 0.9042683839797974,
      "learning_rate": 9.84313563938365e-06,
      "loss": 0.821,
      "step": 2997
    },
    {
      "epoch": 0.1650063294622709,
      "grad_norm": 0.8333690762519836,
      "learning_rate": 9.843027896673577e-06,
      "loss": 0.781,
      "step": 2998
    },
    {
      "epoch": 0.16506136826462656,
      "grad_norm": 0.819922924041748,
      "learning_rate": 9.8429201175646e-06,
      "loss": 0.869,
      "step": 2999
    },
    {
      "epoch": 0.16511640706698222,
      "grad_norm": 0.8349948525428772,
      "learning_rate": 9.842812302057534e-06,
      "loss": 0.9271,
      "step": 3000
    },
    {
      "epoch": 0.16517144586933788,
      "grad_norm": 0.8981684446334839,
      "learning_rate": 9.842704450153187e-06,
      "loss": 0.7384,
      "step": 3001
    },
    {
      "epoch": 0.16522648467169354,
      "grad_norm": 0.839133083820343,
      "learning_rate": 9.842596561852369e-06,
      "loss": 0.9016,
      "step": 3002
    },
    {
      "epoch": 0.1652815234740492,
      "grad_norm": 0.8303349614143372,
      "learning_rate": 9.842488637155891e-06,
      "loss": 0.7488,
      "step": 3003
    },
    {
      "epoch": 0.16533656227640486,
      "grad_norm": 0.8748323917388916,
      "learning_rate": 9.842380676064566e-06,
      "loss": 0.8163,
      "step": 3004
    },
    {
      "epoch": 0.16539160107876053,
      "grad_norm": 0.782844603061676,
      "learning_rate": 9.842272678579203e-06,
      "loss": 0.8465,
      "step": 3005
    },
    {
      "epoch": 0.1654466398811162,
      "grad_norm": 0.8068844676017761,
      "learning_rate": 9.842164644700615e-06,
      "loss": 0.8856,
      "step": 3006
    },
    {
      "epoch": 0.16550167868347185,
      "grad_norm": 0.9174006581306458,
      "learning_rate": 9.842056574429615e-06,
      "loss": 0.7748,
      "step": 3007
    },
    {
      "epoch": 0.1655567174858275,
      "grad_norm": 0.7453809380531311,
      "learning_rate": 9.841948467767012e-06,
      "loss": 0.7565,
      "step": 3008
    },
    {
      "epoch": 0.16561175628818317,
      "grad_norm": 0.8408182859420776,
      "learning_rate": 9.841840324713622e-06,
      "loss": 0.7345,
      "step": 3009
    },
    {
      "epoch": 0.16566679509053883,
      "grad_norm": 0.8599638938903809,
      "learning_rate": 9.841732145270254e-06,
      "loss": 0.8163,
      "step": 3010
    },
    {
      "epoch": 0.1657218338928945,
      "grad_norm": 0.877616822719574,
      "learning_rate": 9.841623929437725e-06,
      "loss": 0.8685,
      "step": 3011
    },
    {
      "epoch": 0.16577687269525015,
      "grad_norm": 0.7765643000602722,
      "learning_rate": 9.841515677216846e-06,
      "loss": 0.7281,
      "step": 3012
    },
    {
      "epoch": 0.16583191149760582,
      "grad_norm": 0.7891712784767151,
      "learning_rate": 9.841407388608431e-06,
      "loss": 0.8618,
      "step": 3013
    },
    {
      "epoch": 0.16588695029996148,
      "grad_norm": 0.9215571284294128,
      "learning_rate": 9.841299063613295e-06,
      "loss": 0.8709,
      "step": 3014
    },
    {
      "epoch": 0.16594198910231714,
      "grad_norm": 0.8428288698196411,
      "learning_rate": 9.841190702232249e-06,
      "loss": 0.8227,
      "step": 3015
    },
    {
      "epoch": 0.1659970279046728,
      "grad_norm": 0.9294042587280273,
      "learning_rate": 9.841082304466112e-06,
      "loss": 0.8203,
      "step": 3016
    },
    {
      "epoch": 0.16605206670702846,
      "grad_norm": 0.7530880570411682,
      "learning_rate": 9.840973870315695e-06,
      "loss": 0.7681,
      "step": 3017
    },
    {
      "epoch": 0.16610710550938412,
      "grad_norm": 1.0149626731872559,
      "learning_rate": 9.840865399781814e-06,
      "loss": 0.9212,
      "step": 3018
    },
    {
      "epoch": 0.16616214431173978,
      "grad_norm": 0.7595353722572327,
      "learning_rate": 9.840756892865285e-06,
      "loss": 0.795,
      "step": 3019
    },
    {
      "epoch": 0.16621718311409545,
      "grad_norm": 0.7893253564834595,
      "learning_rate": 9.840648349566924e-06,
      "loss": 0.8147,
      "step": 3020
    },
    {
      "epoch": 0.1662722219164511,
      "grad_norm": 0.8190789222717285,
      "learning_rate": 9.840539769887543e-06,
      "loss": 0.7233,
      "step": 3021
    },
    {
      "epoch": 0.16632726071880677,
      "grad_norm": 0.7771229147911072,
      "learning_rate": 9.840431153827963e-06,
      "loss": 0.7172,
      "step": 3022
    },
    {
      "epoch": 0.16638229952116243,
      "grad_norm": 0.7379328012466431,
      "learning_rate": 9.840322501388998e-06,
      "loss": 0.7603,
      "step": 3023
    },
    {
      "epoch": 0.1664373383235181,
      "grad_norm": 0.9488499760627747,
      "learning_rate": 9.840213812571464e-06,
      "loss": 0.8025,
      "step": 3024
    },
    {
      "epoch": 0.16649237712587375,
      "grad_norm": 0.7135865092277527,
      "learning_rate": 9.84010508737618e-06,
      "loss": 0.7412,
      "step": 3025
    },
    {
      "epoch": 0.16654741592822941,
      "grad_norm": 1.6780112981796265,
      "learning_rate": 9.83999632580396e-06,
      "loss": 0.9231,
      "step": 3026
    },
    {
      "epoch": 0.16660245473058508,
      "grad_norm": 0.8815935850143433,
      "learning_rate": 9.839887527855623e-06,
      "loss": 0.7903,
      "step": 3027
    },
    {
      "epoch": 0.16665749353294074,
      "grad_norm": 0.8942261338233948,
      "learning_rate": 9.83977869353199e-06,
      "loss": 0.8328,
      "step": 3028
    },
    {
      "epoch": 0.16671253233529637,
      "grad_norm": 0.7866815328598022,
      "learning_rate": 9.839669822833873e-06,
      "loss": 0.8483,
      "step": 3029
    },
    {
      "epoch": 0.16676757113765203,
      "grad_norm": 0.8133070468902588,
      "learning_rate": 9.839560915762094e-06,
      "loss": 0.8665,
      "step": 3030
    },
    {
      "epoch": 0.1668226099400077,
      "grad_norm": 0.7442927360534668,
      "learning_rate": 9.839451972317469e-06,
      "loss": 0.6296,
      "step": 3031
    },
    {
      "epoch": 0.16687764874236335,
      "grad_norm": 0.7505021691322327,
      "learning_rate": 9.83934299250082e-06,
      "loss": 0.7976,
      "step": 3032
    },
    {
      "epoch": 0.16693268754471902,
      "grad_norm": 0.8310422897338867,
      "learning_rate": 9.839233976312964e-06,
      "loss": 0.9022,
      "step": 3033
    },
    {
      "epoch": 0.16698772634707468,
      "grad_norm": 0.8175413012504578,
      "learning_rate": 9.839124923754721e-06,
      "loss": 0.8653,
      "step": 3034
    },
    {
      "epoch": 0.16704276514943034,
      "grad_norm": 0.7963089346885681,
      "learning_rate": 9.839015834826912e-06,
      "loss": 0.7888,
      "step": 3035
    },
    {
      "epoch": 0.167097803951786,
      "grad_norm": 0.8923391699790955,
      "learning_rate": 9.838906709530353e-06,
      "loss": 0.9396,
      "step": 3036
    },
    {
      "epoch": 0.16715284275414166,
      "grad_norm": 0.7851678133010864,
      "learning_rate": 9.838797547865869e-06,
      "loss": 0.8163,
      "step": 3037
    },
    {
      "epoch": 0.16720788155649732,
      "grad_norm": 0.817877471446991,
      "learning_rate": 9.838688349834275e-06,
      "loss": 0.8928,
      "step": 3038
    },
    {
      "epoch": 0.16726292035885298,
      "grad_norm": 0.7603926062583923,
      "learning_rate": 9.838579115436395e-06,
      "loss": 0.7418,
      "step": 3039
    },
    {
      "epoch": 0.16731795916120865,
      "grad_norm": 0.8086446523666382,
      "learning_rate": 9.83846984467305e-06,
      "loss": 0.8017,
      "step": 3040
    },
    {
      "epoch": 0.1673729979635643,
      "grad_norm": 1.4745439291000366,
      "learning_rate": 9.838360537545061e-06,
      "loss": 0.7964,
      "step": 3041
    },
    {
      "epoch": 0.16742803676591997,
      "grad_norm": 0.778404176235199,
      "learning_rate": 9.83825119405325e-06,
      "loss": 0.7395,
      "step": 3042
    },
    {
      "epoch": 0.16748307556827563,
      "grad_norm": 0.8245886564254761,
      "learning_rate": 9.838141814198439e-06,
      "loss": 0.8697,
      "step": 3043
    },
    {
      "epoch": 0.1675381143706313,
      "grad_norm": 0.8395472764968872,
      "learning_rate": 9.838032397981448e-06,
      "loss": 0.8545,
      "step": 3044
    },
    {
      "epoch": 0.16759315317298695,
      "grad_norm": 0.8973744511604309,
      "learning_rate": 9.8379229454031e-06,
      "loss": 0.8999,
      "step": 3045
    },
    {
      "epoch": 0.16764819197534261,
      "grad_norm": 1.2034368515014648,
      "learning_rate": 9.837813456464219e-06,
      "loss": 0.9039,
      "step": 3046
    },
    {
      "epoch": 0.16770323077769828,
      "grad_norm": 0.862167477607727,
      "learning_rate": 9.837703931165625e-06,
      "loss": 0.889,
      "step": 3047
    },
    {
      "epoch": 0.16775826958005394,
      "grad_norm": 0.7624714970588684,
      "learning_rate": 9.837594369508146e-06,
      "loss": 0.7072,
      "step": 3048
    },
    {
      "epoch": 0.1678133083824096,
      "grad_norm": 0.7771621346473694,
      "learning_rate": 9.8374847714926e-06,
      "loss": 0.8769,
      "step": 3049
    },
    {
      "epoch": 0.16786834718476526,
      "grad_norm": 0.7834492921829224,
      "learning_rate": 9.837375137119816e-06,
      "loss": 0.841,
      "step": 3050
    },
    {
      "epoch": 0.16792338598712092,
      "grad_norm": 0.8175067901611328,
      "learning_rate": 9.837265466390612e-06,
      "loss": 0.8149,
      "step": 3051
    },
    {
      "epoch": 0.16797842478947658,
      "grad_norm": 0.7474493384361267,
      "learning_rate": 9.83715575930582e-06,
      "loss": 0.7716,
      "step": 3052
    },
    {
      "epoch": 0.16803346359183224,
      "grad_norm": 1.1263303756713867,
      "learning_rate": 9.837046015866257e-06,
      "loss": 0.8026,
      "step": 3053
    },
    {
      "epoch": 0.1680885023941879,
      "grad_norm": 0.8741740584373474,
      "learning_rate": 9.836936236072752e-06,
      "loss": 0.8795,
      "step": 3054
    },
    {
      "epoch": 0.16814354119654357,
      "grad_norm": 0.8108506798744202,
      "learning_rate": 9.83682641992613e-06,
      "loss": 0.7682,
      "step": 3055
    },
    {
      "epoch": 0.16819857999889923,
      "grad_norm": 0.9380543231964111,
      "learning_rate": 9.836716567427213e-06,
      "loss": 0.8739,
      "step": 3056
    },
    {
      "epoch": 0.1682536188012549,
      "grad_norm": 0.7755940556526184,
      "learning_rate": 9.83660667857683e-06,
      "loss": 0.7287,
      "step": 3057
    },
    {
      "epoch": 0.16830865760361055,
      "grad_norm": 0.808907151222229,
      "learning_rate": 9.836496753375807e-06,
      "loss": 0.7988,
      "step": 3058
    },
    {
      "epoch": 0.1683636964059662,
      "grad_norm": 1.1496524810791016,
      "learning_rate": 9.836386791824967e-06,
      "loss": 0.8621,
      "step": 3059
    },
    {
      "epoch": 0.16841873520832187,
      "grad_norm": 0.8550384640693665,
      "learning_rate": 9.83627679392514e-06,
      "loss": 0.913,
      "step": 3060
    },
    {
      "epoch": 0.16847377401067753,
      "grad_norm": 0.761142909526825,
      "learning_rate": 9.83616675967715e-06,
      "loss": 0.7271,
      "step": 3061
    },
    {
      "epoch": 0.1685288128130332,
      "grad_norm": 0.8496200442314148,
      "learning_rate": 9.836056689081828e-06,
      "loss": 0.7885,
      "step": 3062
    },
    {
      "epoch": 0.16858385161538886,
      "grad_norm": 0.8310382962226868,
      "learning_rate": 9.835946582139996e-06,
      "loss": 0.858,
      "step": 3063
    },
    {
      "epoch": 0.16863889041774452,
      "grad_norm": 0.7870821952819824,
      "learning_rate": 9.835836438852485e-06,
      "loss": 0.7791,
      "step": 3064
    },
    {
      "epoch": 0.16869392922010018,
      "grad_norm": 0.7170534729957581,
      "learning_rate": 9.83572625922012e-06,
      "loss": 0.6666,
      "step": 3065
    },
    {
      "epoch": 0.16874896802245584,
      "grad_norm": 0.9764187932014465,
      "learning_rate": 9.835616043243732e-06,
      "loss": 0.8341,
      "step": 3066
    },
    {
      "epoch": 0.1688040068248115,
      "grad_norm": 0.7453315258026123,
      "learning_rate": 9.83550579092415e-06,
      "loss": 0.7032,
      "step": 3067
    },
    {
      "epoch": 0.16885904562716716,
      "grad_norm": 0.9205759763717651,
      "learning_rate": 9.835395502262196e-06,
      "loss": 0.813,
      "step": 3068
    },
    {
      "epoch": 0.16891408442952283,
      "grad_norm": 0.8152205944061279,
      "learning_rate": 9.835285177258708e-06,
      "loss": 0.8275,
      "step": 3069
    },
    {
      "epoch": 0.1689691232318785,
      "grad_norm": 0.8065707087516785,
      "learning_rate": 9.83517481591451e-06,
      "loss": 0.8648,
      "step": 3070
    },
    {
      "epoch": 0.16902416203423415,
      "grad_norm": 0.7774410247802734,
      "learning_rate": 9.835064418230432e-06,
      "loss": 0.7818,
      "step": 3071
    },
    {
      "epoch": 0.16907920083658978,
      "grad_norm": 0.8591069579124451,
      "learning_rate": 9.834953984207305e-06,
      "loss": 0.8055,
      "step": 3072
    },
    {
      "epoch": 0.16913423963894544,
      "grad_norm": 0.7421612739562988,
      "learning_rate": 9.834843513845958e-06,
      "loss": 0.7543,
      "step": 3073
    },
    {
      "epoch": 0.1691892784413011,
      "grad_norm": 0.7855183482170105,
      "learning_rate": 9.83473300714722e-06,
      "loss": 0.7011,
      "step": 3074
    },
    {
      "epoch": 0.16924431724365677,
      "grad_norm": 0.8061636686325073,
      "learning_rate": 9.834622464111924e-06,
      "loss": 0.8096,
      "step": 3075
    },
    {
      "epoch": 0.16929935604601243,
      "grad_norm": 0.8048406839370728,
      "learning_rate": 9.834511884740898e-06,
      "loss": 0.8166,
      "step": 3076
    },
    {
      "epoch": 0.1693543948483681,
      "grad_norm": 0.8776549696922302,
      "learning_rate": 9.834401269034977e-06,
      "loss": 0.8169,
      "step": 3077
    },
    {
      "epoch": 0.16940943365072375,
      "grad_norm": 1.0208356380462646,
      "learning_rate": 9.83429061699499e-06,
      "loss": 0.6976,
      "step": 3078
    },
    {
      "epoch": 0.1694644724530794,
      "grad_norm": 0.7641016840934753,
      "learning_rate": 9.834179928621767e-06,
      "loss": 0.7109,
      "step": 3079
    },
    {
      "epoch": 0.16951951125543507,
      "grad_norm": 0.7648905515670776,
      "learning_rate": 9.834069203916143e-06,
      "loss": 0.7927,
      "step": 3080
    },
    {
      "epoch": 0.16957455005779073,
      "grad_norm": 0.7898744344711304,
      "learning_rate": 9.833958442878948e-06,
      "loss": 0.7911,
      "step": 3081
    },
    {
      "epoch": 0.1696295888601464,
      "grad_norm": 0.8812462687492371,
      "learning_rate": 9.833847645511016e-06,
      "loss": 0.8381,
      "step": 3082
    },
    {
      "epoch": 0.16968462766250206,
      "grad_norm": 0.8141197562217712,
      "learning_rate": 9.833736811813179e-06,
      "loss": 0.7422,
      "step": 3083
    },
    {
      "epoch": 0.16973966646485772,
      "grad_norm": 0.7860949635505676,
      "learning_rate": 9.83362594178627e-06,
      "loss": 0.7568,
      "step": 3084
    },
    {
      "epoch": 0.16979470526721338,
      "grad_norm": 0.6688396334648132,
      "learning_rate": 9.833515035431123e-06,
      "loss": 0.7143,
      "step": 3085
    },
    {
      "epoch": 0.16984974406956904,
      "grad_norm": 0.7525103092193604,
      "learning_rate": 9.833404092748569e-06,
      "loss": 0.8026,
      "step": 3086
    },
    {
      "epoch": 0.1699047828719247,
      "grad_norm": 0.8505181670188904,
      "learning_rate": 9.833293113739444e-06,
      "loss": 0.8894,
      "step": 3087
    },
    {
      "epoch": 0.16995982167428036,
      "grad_norm": 0.8432300090789795,
      "learning_rate": 9.833182098404583e-06,
      "loss": 0.7801,
      "step": 3088
    },
    {
      "epoch": 0.17001486047663603,
      "grad_norm": 0.7655903100967407,
      "learning_rate": 9.833071046744819e-06,
      "loss": 0.7838,
      "step": 3089
    },
    {
      "epoch": 0.1700698992789917,
      "grad_norm": 0.8436369895935059,
      "learning_rate": 9.832959958760986e-06,
      "loss": 0.8636,
      "step": 3090
    },
    {
      "epoch": 0.17012493808134735,
      "grad_norm": 0.7880234122276306,
      "learning_rate": 9.83284883445392e-06,
      "loss": 0.7701,
      "step": 3091
    },
    {
      "epoch": 0.170179976883703,
      "grad_norm": 0.7713757753372192,
      "learning_rate": 9.832737673824455e-06,
      "loss": 0.8652,
      "step": 3092
    },
    {
      "epoch": 0.17023501568605867,
      "grad_norm": 0.7905295491218567,
      "learning_rate": 9.832626476873428e-06,
      "loss": 0.8666,
      "step": 3093
    },
    {
      "epoch": 0.17029005448841433,
      "grad_norm": 0.7589883804321289,
      "learning_rate": 9.832515243601675e-06,
      "loss": 0.8051,
      "step": 3094
    },
    {
      "epoch": 0.17034509329077,
      "grad_norm": 0.9068838953971863,
      "learning_rate": 9.83240397401003e-06,
      "loss": 0.9037,
      "step": 3095
    },
    {
      "epoch": 0.17040013209312566,
      "grad_norm": 0.7465278506278992,
      "learning_rate": 9.83229266809933e-06,
      "loss": 0.7425,
      "step": 3096
    },
    {
      "epoch": 0.17045517089548132,
      "grad_norm": 0.8111177086830139,
      "learning_rate": 9.83218132587041e-06,
      "loss": 0.8034,
      "step": 3097
    },
    {
      "epoch": 0.17051020969783698,
      "grad_norm": 1.1007672548294067,
      "learning_rate": 9.832069947324112e-06,
      "loss": 0.9139,
      "step": 3098
    },
    {
      "epoch": 0.17056524850019264,
      "grad_norm": 0.881179690361023,
      "learning_rate": 9.831958532461269e-06,
      "loss": 0.9062,
      "step": 3099
    },
    {
      "epoch": 0.1706202873025483,
      "grad_norm": 0.8012413382530212,
      "learning_rate": 9.831847081282718e-06,
      "loss": 0.7956,
      "step": 3100
    },
    {
      "epoch": 0.17067532610490396,
      "grad_norm": 0.741731584072113,
      "learning_rate": 9.831735593789298e-06,
      "loss": 0.8754,
      "step": 3101
    },
    {
      "epoch": 0.17073036490725962,
      "grad_norm": 0.8945604562759399,
      "learning_rate": 9.831624069981848e-06,
      "loss": 0.8293,
      "step": 3102
    },
    {
      "epoch": 0.17078540370961529,
      "grad_norm": 0.7865545749664307,
      "learning_rate": 9.831512509861203e-06,
      "loss": 0.7812,
      "step": 3103
    },
    {
      "epoch": 0.17084044251197095,
      "grad_norm": 0.832847535610199,
      "learning_rate": 9.831400913428205e-06,
      "loss": 0.8925,
      "step": 3104
    },
    {
      "epoch": 0.1708954813143266,
      "grad_norm": 0.7374216914176941,
      "learning_rate": 9.83128928068369e-06,
      "loss": 0.8275,
      "step": 3105
    },
    {
      "epoch": 0.17095052011668227,
      "grad_norm": 0.748725414276123,
      "learning_rate": 9.831177611628497e-06,
      "loss": 0.8364,
      "step": 3106
    },
    {
      "epoch": 0.17100555891903793,
      "grad_norm": 0.810276448726654,
      "learning_rate": 9.831065906263468e-06,
      "loss": 0.861,
      "step": 3107
    },
    {
      "epoch": 0.1710605977213936,
      "grad_norm": 0.7607758641242981,
      "learning_rate": 9.83095416458944e-06,
      "loss": 0.7989,
      "step": 3108
    },
    {
      "epoch": 0.17111563652374925,
      "grad_norm": 0.7206127047538757,
      "learning_rate": 9.830842386607253e-06,
      "loss": 0.7187,
      "step": 3109
    },
    {
      "epoch": 0.17117067532610492,
      "grad_norm": 0.7775895595550537,
      "learning_rate": 9.83073057231775e-06,
      "loss": 0.8008,
      "step": 3110
    },
    {
      "epoch": 0.17122571412846058,
      "grad_norm": 0.8351094722747803,
      "learning_rate": 9.830618721721768e-06,
      "loss": 0.8025,
      "step": 3111
    },
    {
      "epoch": 0.17128075293081624,
      "grad_norm": 0.8090646266937256,
      "learning_rate": 9.830506834820148e-06,
      "loss": 0.8012,
      "step": 3112
    },
    {
      "epoch": 0.1713357917331719,
      "grad_norm": 0.7762801051139832,
      "learning_rate": 9.830394911613733e-06,
      "loss": 0.8428,
      "step": 3113
    },
    {
      "epoch": 0.17139083053552756,
      "grad_norm": 0.8117541074752808,
      "learning_rate": 9.83028295210336e-06,
      "loss": 0.8566,
      "step": 3114
    },
    {
      "epoch": 0.1714458693378832,
      "grad_norm": 0.8786184787750244,
      "learning_rate": 9.830170956289876e-06,
      "loss": 0.8386,
      "step": 3115
    },
    {
      "epoch": 0.17150090814023886,
      "grad_norm": 1.0181046724319458,
      "learning_rate": 9.83005892417412e-06,
      "loss": 0.8555,
      "step": 3116
    },
    {
      "epoch": 0.17155594694259452,
      "grad_norm": 0.8236173391342163,
      "learning_rate": 9.829946855756934e-06,
      "loss": 0.7933,
      "step": 3117
    },
    {
      "epoch": 0.17161098574495018,
      "grad_norm": 0.8058149814605713,
      "learning_rate": 9.829834751039157e-06,
      "loss": 0.842,
      "step": 3118
    },
    {
      "epoch": 0.17166602454730584,
      "grad_norm": 0.7419908046722412,
      "learning_rate": 9.82972261002164e-06,
      "loss": 0.8397,
      "step": 3119
    },
    {
      "epoch": 0.1717210633496615,
      "grad_norm": 0.7528164982795715,
      "learning_rate": 9.829610432705216e-06,
      "loss": 0.7931,
      "step": 3120
    },
    {
      "epoch": 0.17177610215201716,
      "grad_norm": 0.7357296943664551,
      "learning_rate": 9.829498219090736e-06,
      "loss": 0.8089,
      "step": 3121
    },
    {
      "epoch": 0.17183114095437282,
      "grad_norm": 0.7635773420333862,
      "learning_rate": 9.829385969179039e-06,
      "loss": 0.7442,
      "step": 3122
    },
    {
      "epoch": 0.17188617975672849,
      "grad_norm": 0.8200171589851379,
      "learning_rate": 9.82927368297097e-06,
      "loss": 0.757,
      "step": 3123
    },
    {
      "epoch": 0.17194121855908415,
      "grad_norm": 0.8367171287536621,
      "learning_rate": 9.829161360467374e-06,
      "loss": 0.915,
      "step": 3124
    },
    {
      "epoch": 0.1719962573614398,
      "grad_norm": 0.8460778594017029,
      "learning_rate": 9.829049001669091e-06,
      "loss": 0.8568,
      "step": 3125
    },
    {
      "epoch": 0.17205129616379547,
      "grad_norm": 0.7301799058914185,
      "learning_rate": 9.82893660657697e-06,
      "loss": 0.8041,
      "step": 3126
    },
    {
      "epoch": 0.17210633496615113,
      "grad_norm": 0.7858132123947144,
      "learning_rate": 9.828824175191854e-06,
      "loss": 0.8367,
      "step": 3127
    },
    {
      "epoch": 0.1721613737685068,
      "grad_norm": 0.8118360042572021,
      "learning_rate": 9.82871170751459e-06,
      "loss": 0.85,
      "step": 3128
    },
    {
      "epoch": 0.17221641257086245,
      "grad_norm": 0.9020261764526367,
      "learning_rate": 9.828599203546019e-06,
      "loss": 0.789,
      "step": 3129
    },
    {
      "epoch": 0.17227145137321812,
      "grad_norm": 0.8194546699523926,
      "learning_rate": 9.828486663286989e-06,
      "loss": 0.8644,
      "step": 3130
    },
    {
      "epoch": 0.17232649017557378,
      "grad_norm": 0.7764905095100403,
      "learning_rate": 9.828374086738345e-06,
      "loss": 0.7961,
      "step": 3131
    },
    {
      "epoch": 0.17238152897792944,
      "grad_norm": 0.7712632417678833,
      "learning_rate": 9.828261473900935e-06,
      "loss": 0.8082,
      "step": 3132
    },
    {
      "epoch": 0.1724365677802851,
      "grad_norm": 0.7100280523300171,
      "learning_rate": 9.828148824775604e-06,
      "loss": 0.7514,
      "step": 3133
    },
    {
      "epoch": 0.17249160658264076,
      "grad_norm": 0.7812890410423279,
      "learning_rate": 9.8280361393632e-06,
      "loss": 0.7125,
      "step": 3134
    },
    {
      "epoch": 0.17254664538499642,
      "grad_norm": 0.8772642612457275,
      "learning_rate": 9.827923417664568e-06,
      "loss": 0.8355,
      "step": 3135
    },
    {
      "epoch": 0.17260168418735208,
      "grad_norm": 0.9161205291748047,
      "learning_rate": 9.827810659680555e-06,
      "loss": 0.7511,
      "step": 3136
    },
    {
      "epoch": 0.17265672298970774,
      "grad_norm": 0.7628560662269592,
      "learning_rate": 9.82769786541201e-06,
      "loss": 0.882,
      "step": 3137
    },
    {
      "epoch": 0.1727117617920634,
      "grad_norm": 0.8203405737876892,
      "learning_rate": 9.827585034859781e-06,
      "loss": 0.8172,
      "step": 3138
    },
    {
      "epoch": 0.17276680059441907,
      "grad_norm": 0.8318095207214355,
      "learning_rate": 9.827472168024715e-06,
      "loss": 0.7784,
      "step": 3139
    },
    {
      "epoch": 0.17282183939677473,
      "grad_norm": 0.9137747287750244,
      "learning_rate": 9.827359264907658e-06,
      "loss": 0.8643,
      "step": 3140
    },
    {
      "epoch": 0.1728768781991304,
      "grad_norm": 0.9441068768501282,
      "learning_rate": 9.827246325509463e-06,
      "loss": 0.7936,
      "step": 3141
    },
    {
      "epoch": 0.17293191700148605,
      "grad_norm": 0.7402390837669373,
      "learning_rate": 9.827133349830977e-06,
      "loss": 0.7813,
      "step": 3142
    },
    {
      "epoch": 0.1729869558038417,
      "grad_norm": 0.8328836560249329,
      "learning_rate": 9.827020337873048e-06,
      "loss": 0.7676,
      "step": 3143
    },
    {
      "epoch": 0.17304199460619737,
      "grad_norm": 0.8106881380081177,
      "learning_rate": 9.826907289636526e-06,
      "loss": 0.9037,
      "step": 3144
    },
    {
      "epoch": 0.17309703340855304,
      "grad_norm": 0.8457425236701965,
      "learning_rate": 9.826794205122263e-06,
      "loss": 0.78,
      "step": 3145
    },
    {
      "epoch": 0.1731520722109087,
      "grad_norm": 0.9335517883300781,
      "learning_rate": 9.826681084331105e-06,
      "loss": 0.9197,
      "step": 3146
    },
    {
      "epoch": 0.17320711101326436,
      "grad_norm": 0.9098715782165527,
      "learning_rate": 9.826567927263904e-06,
      "loss": 0.932,
      "step": 3147
    },
    {
      "epoch": 0.17326214981562002,
      "grad_norm": 0.767234206199646,
      "learning_rate": 9.826454733921512e-06,
      "loss": 0.8717,
      "step": 3148
    },
    {
      "epoch": 0.17331718861797568,
      "grad_norm": 0.8114444017410278,
      "learning_rate": 9.826341504304775e-06,
      "loss": 0.8744,
      "step": 3149
    },
    {
      "epoch": 0.17337222742033134,
      "grad_norm": 0.7948976755142212,
      "learning_rate": 9.82622823841455e-06,
      "loss": 0.7947,
      "step": 3150
    },
    {
      "epoch": 0.173427266222687,
      "grad_norm": 0.7808204889297485,
      "learning_rate": 9.826114936251684e-06,
      "loss": 0.8151,
      "step": 3151
    },
    {
      "epoch": 0.17348230502504267,
      "grad_norm": 0.733860969543457,
      "learning_rate": 9.82600159781703e-06,
      "loss": 0.8018,
      "step": 3152
    },
    {
      "epoch": 0.17353734382739833,
      "grad_norm": 0.7630699276924133,
      "learning_rate": 9.825888223111442e-06,
      "loss": 0.7937,
      "step": 3153
    },
    {
      "epoch": 0.173592382629754,
      "grad_norm": 0.7892931699752808,
      "learning_rate": 9.825774812135766e-06,
      "loss": 0.782,
      "step": 3154
    },
    {
      "epoch": 0.17364742143210965,
      "grad_norm": 0.6642436385154724,
      "learning_rate": 9.825661364890862e-06,
      "loss": 0.6611,
      "step": 3155
    },
    {
      "epoch": 0.1737024602344653,
      "grad_norm": 0.7755968570709229,
      "learning_rate": 9.825547881377577e-06,
      "loss": 0.7835,
      "step": 3156
    },
    {
      "epoch": 0.17375749903682097,
      "grad_norm": 0.8406579494476318,
      "learning_rate": 9.825434361596766e-06,
      "loss": 0.9178,
      "step": 3157
    },
    {
      "epoch": 0.1738125378391766,
      "grad_norm": 0.8887308835983276,
      "learning_rate": 9.825320805549284e-06,
      "loss": 0.7951,
      "step": 3158
    },
    {
      "epoch": 0.17386757664153227,
      "grad_norm": 0.85418701171875,
      "learning_rate": 9.825207213235978e-06,
      "loss": 0.8671,
      "step": 3159
    },
    {
      "epoch": 0.17392261544388793,
      "grad_norm": 0.8831202983856201,
      "learning_rate": 9.82509358465771e-06,
      "loss": 0.8708,
      "step": 3160
    },
    {
      "epoch": 0.1739776542462436,
      "grad_norm": 0.9041616320610046,
      "learning_rate": 9.82497991981533e-06,
      "loss": 0.8981,
      "step": 3161
    },
    {
      "epoch": 0.17403269304859925,
      "grad_norm": 0.8169258832931519,
      "learning_rate": 9.824866218709692e-06,
      "loss": 0.8857,
      "step": 3162
    },
    {
      "epoch": 0.1740877318509549,
      "grad_norm": 0.8714475631713867,
      "learning_rate": 9.824752481341651e-06,
      "loss": 0.8552,
      "step": 3163
    },
    {
      "epoch": 0.17414277065331057,
      "grad_norm": 0.8261111378669739,
      "learning_rate": 9.824638707712061e-06,
      "loss": 0.808,
      "step": 3164
    },
    {
      "epoch": 0.17419780945566624,
      "grad_norm": 0.7542527914047241,
      "learning_rate": 9.82452489782178e-06,
      "loss": 0.8078,
      "step": 3165
    },
    {
      "epoch": 0.1742528482580219,
      "grad_norm": 1.309218168258667,
      "learning_rate": 9.824411051671658e-06,
      "loss": 0.9325,
      "step": 3166
    },
    {
      "epoch": 0.17430788706037756,
      "grad_norm": 0.8528563380241394,
      "learning_rate": 9.824297169262555e-06,
      "loss": 0.8493,
      "step": 3167
    },
    {
      "epoch": 0.17436292586273322,
      "grad_norm": 0.7777062058448792,
      "learning_rate": 9.824183250595328e-06,
      "loss": 0.7002,
      "step": 3168
    },
    {
      "epoch": 0.17441796466508888,
      "grad_norm": 0.7385506629943848,
      "learning_rate": 9.824069295670828e-06,
      "loss": 0.8396,
      "step": 3169
    },
    {
      "epoch": 0.17447300346744454,
      "grad_norm": 0.8316949605941772,
      "learning_rate": 9.823955304489918e-06,
      "loss": 0.8769,
      "step": 3170
    },
    {
      "epoch": 0.1745280422698002,
      "grad_norm": 0.8149139285087585,
      "learning_rate": 9.823841277053448e-06,
      "loss": 0.8009,
      "step": 3171
    },
    {
      "epoch": 0.17458308107215587,
      "grad_norm": 0.8761584162712097,
      "learning_rate": 9.82372721336228e-06,
      "loss": 0.7366,
      "step": 3172
    },
    {
      "epoch": 0.17463811987451153,
      "grad_norm": 0.7104084491729736,
      "learning_rate": 9.82361311341727e-06,
      "loss": 0.6704,
      "step": 3173
    },
    {
      "epoch": 0.1746931586768672,
      "grad_norm": 0.791806697845459,
      "learning_rate": 9.823498977219273e-06,
      "loss": 0.9054,
      "step": 3174
    },
    {
      "epoch": 0.17474819747922285,
      "grad_norm": 0.7675086855888367,
      "learning_rate": 9.82338480476915e-06,
      "loss": 0.751,
      "step": 3175
    },
    {
      "epoch": 0.1748032362815785,
      "grad_norm": 0.7380725145339966,
      "learning_rate": 9.823270596067759e-06,
      "loss": 0.7618,
      "step": 3176
    },
    {
      "epoch": 0.17485827508393417,
      "grad_norm": 0.7311519384384155,
      "learning_rate": 9.823156351115954e-06,
      "loss": 0.7424,
      "step": 3177
    },
    {
      "epoch": 0.17491331388628983,
      "grad_norm": 0.7888365387916565,
      "learning_rate": 9.8230420699146e-06,
      "loss": 0.7717,
      "step": 3178
    },
    {
      "epoch": 0.1749683526886455,
      "grad_norm": 0.9329265356063843,
      "learning_rate": 9.822927752464552e-06,
      "loss": 0.8256,
      "step": 3179
    },
    {
      "epoch": 0.17502339149100116,
      "grad_norm": 0.711794912815094,
      "learning_rate": 9.822813398766671e-06,
      "loss": 0.7373,
      "step": 3180
    },
    {
      "epoch": 0.17507843029335682,
      "grad_norm": 0.8713497519493103,
      "learning_rate": 9.822699008821813e-06,
      "loss": 0.8135,
      "step": 3181
    },
    {
      "epoch": 0.17513346909571248,
      "grad_norm": 0.6923471689224243,
      "learning_rate": 9.822584582630841e-06,
      "loss": 0.7589,
      "step": 3182
    },
    {
      "epoch": 0.17518850789806814,
      "grad_norm": 0.8648017048835754,
      "learning_rate": 9.822470120194616e-06,
      "loss": 0.7828,
      "step": 3183
    },
    {
      "epoch": 0.1752435467004238,
      "grad_norm": 0.8407077789306641,
      "learning_rate": 9.822355621513994e-06,
      "loss": 0.8537,
      "step": 3184
    },
    {
      "epoch": 0.17529858550277946,
      "grad_norm": 0.8076738119125366,
      "learning_rate": 9.822241086589841e-06,
      "loss": 0.7827,
      "step": 3185
    },
    {
      "epoch": 0.17535362430513513,
      "grad_norm": 0.8402661085128784,
      "learning_rate": 9.822126515423011e-06,
      "loss": 0.8247,
      "step": 3186
    },
    {
      "epoch": 0.1754086631074908,
      "grad_norm": 0.8911813497543335,
      "learning_rate": 9.822011908014373e-06,
      "loss": 0.8996,
      "step": 3187
    },
    {
      "epoch": 0.17546370190984645,
      "grad_norm": 0.8060111999511719,
      "learning_rate": 9.821897264364782e-06,
      "loss": 0.796,
      "step": 3188
    },
    {
      "epoch": 0.1755187407122021,
      "grad_norm": 0.8476423621177673,
      "learning_rate": 9.8217825844751e-06,
      "loss": 0.8657,
      "step": 3189
    },
    {
      "epoch": 0.17557377951455777,
      "grad_norm": 0.7614054083824158,
      "learning_rate": 9.821667868346194e-06,
      "loss": 0.8583,
      "step": 3190
    },
    {
      "epoch": 0.17562881831691343,
      "grad_norm": 0.8312287330627441,
      "learning_rate": 9.821553115978923e-06,
      "loss": 0.7718,
      "step": 3191
    },
    {
      "epoch": 0.1756838571192691,
      "grad_norm": 0.8199487328529358,
      "learning_rate": 9.82143832737415e-06,
      "loss": 0.7617,
      "step": 3192
    },
    {
      "epoch": 0.17573889592162475,
      "grad_norm": 0.7529115080833435,
      "learning_rate": 9.821323502532733e-06,
      "loss": 0.7587,
      "step": 3193
    },
    {
      "epoch": 0.17579393472398042,
      "grad_norm": 0.9205463528633118,
      "learning_rate": 9.821208641455542e-06,
      "loss": 0.7871,
      "step": 3194
    },
    {
      "epoch": 0.17584897352633608,
      "grad_norm": 0.8055161833763123,
      "learning_rate": 9.821093744143437e-06,
      "loss": 0.8133,
      "step": 3195
    },
    {
      "epoch": 0.17590401232869174,
      "grad_norm": 0.7322981953620911,
      "learning_rate": 9.82097881059728e-06,
      "loss": 0.7442,
      "step": 3196
    },
    {
      "epoch": 0.1759590511310474,
      "grad_norm": 1.0465941429138184,
      "learning_rate": 9.82086384081794e-06,
      "loss": 1.0073,
      "step": 3197
    },
    {
      "epoch": 0.17601408993340306,
      "grad_norm": 0.7607331275939941,
      "learning_rate": 9.820748834806278e-06,
      "loss": 0.8128,
      "step": 3198
    },
    {
      "epoch": 0.17606912873575872,
      "grad_norm": 0.7901879549026489,
      "learning_rate": 9.820633792563156e-06,
      "loss": 0.7928,
      "step": 3199
    },
    {
      "epoch": 0.17612416753811436,
      "grad_norm": 0.8010839223861694,
      "learning_rate": 9.820518714089442e-06,
      "loss": 0.7025,
      "step": 3200
    },
    {
      "epoch": 0.17617920634047002,
      "grad_norm": 0.8511317372322083,
      "learning_rate": 9.820403599385999e-06,
      "loss": 0.7947,
      "step": 3201
    },
    {
      "epoch": 0.17623424514282568,
      "grad_norm": 0.7978847026824951,
      "learning_rate": 9.820288448453693e-06,
      "loss": 0.7395,
      "step": 3202
    },
    {
      "epoch": 0.17628928394518134,
      "grad_norm": 0.6991232633590698,
      "learning_rate": 9.820173261293388e-06,
      "loss": 0.7113,
      "step": 3203
    },
    {
      "epoch": 0.176344322747537,
      "grad_norm": 0.8966444730758667,
      "learning_rate": 9.820058037905954e-06,
      "loss": 0.7399,
      "step": 3204
    },
    {
      "epoch": 0.17639936154989266,
      "grad_norm": 0.8042632341384888,
      "learning_rate": 9.819942778292253e-06,
      "loss": 0.8183,
      "step": 3205
    },
    {
      "epoch": 0.17645440035224833,
      "grad_norm": 0.8047537803649902,
      "learning_rate": 9.81982748245315e-06,
      "loss": 0.852,
      "step": 3206
    },
    {
      "epoch": 0.176509439154604,
      "grad_norm": 0.8277122378349304,
      "learning_rate": 9.819712150389517e-06,
      "loss": 0.8828,
      "step": 3207
    },
    {
      "epoch": 0.17656447795695965,
      "grad_norm": 0.8677185773849487,
      "learning_rate": 9.819596782102216e-06,
      "loss": 0.8416,
      "step": 3208
    },
    {
      "epoch": 0.1766195167593153,
      "grad_norm": 0.8750975728034973,
      "learning_rate": 9.819481377592115e-06,
      "loss": 0.9289,
      "step": 3209
    },
    {
      "epoch": 0.17667455556167097,
      "grad_norm": 0.7665122151374817,
      "learning_rate": 9.819365936860084e-06,
      "loss": 0.8653,
      "step": 3210
    },
    {
      "epoch": 0.17672959436402663,
      "grad_norm": 0.9341353178024292,
      "learning_rate": 9.819250459906989e-06,
      "loss": 0.7225,
      "step": 3211
    },
    {
      "epoch": 0.1767846331663823,
      "grad_norm": 0.7007241249084473,
      "learning_rate": 9.819134946733696e-06,
      "loss": 0.7429,
      "step": 3212
    },
    {
      "epoch": 0.17683967196873795,
      "grad_norm": 0.8001461029052734,
      "learning_rate": 9.819019397341074e-06,
      "loss": 0.759,
      "step": 3213
    },
    {
      "epoch": 0.17689471077109362,
      "grad_norm": 0.8936446905136108,
      "learning_rate": 9.818903811729993e-06,
      "loss": 0.8248,
      "step": 3214
    },
    {
      "epoch": 0.17694974957344928,
      "grad_norm": 0.805570125579834,
      "learning_rate": 9.818788189901321e-06,
      "loss": 0.9214,
      "step": 3215
    },
    {
      "epoch": 0.17700478837580494,
      "grad_norm": 0.7762455940246582,
      "learning_rate": 9.818672531855926e-06,
      "loss": 0.7848,
      "step": 3216
    },
    {
      "epoch": 0.1770598271781606,
      "grad_norm": 0.8391497731208801,
      "learning_rate": 9.81855683759468e-06,
      "loss": 0.7543,
      "step": 3217
    },
    {
      "epoch": 0.17711486598051626,
      "grad_norm": 0.8489046692848206,
      "learning_rate": 9.818441107118449e-06,
      "loss": 0.7908,
      "step": 3218
    },
    {
      "epoch": 0.17716990478287192,
      "grad_norm": 1.0949461460113525,
      "learning_rate": 9.818325340428105e-06,
      "loss": 0.8255,
      "step": 3219
    },
    {
      "epoch": 0.17722494358522758,
      "grad_norm": 0.8710842132568359,
      "learning_rate": 9.81820953752452e-06,
      "loss": 0.859,
      "step": 3220
    },
    {
      "epoch": 0.17727998238758325,
      "grad_norm": 0.7936064600944519,
      "learning_rate": 9.818093698408558e-06,
      "loss": 0.8475,
      "step": 3221
    },
    {
      "epoch": 0.1773350211899389,
      "grad_norm": 0.790341854095459,
      "learning_rate": 9.817977823081095e-06,
      "loss": 0.8137,
      "step": 3222
    },
    {
      "epoch": 0.17739005999229457,
      "grad_norm": 0.8154531717300415,
      "learning_rate": 9.817861911543002e-06,
      "loss": 0.8687,
      "step": 3223
    },
    {
      "epoch": 0.17744509879465023,
      "grad_norm": 0.8346067070960999,
      "learning_rate": 9.817745963795144e-06,
      "loss": 0.8905,
      "step": 3224
    },
    {
      "epoch": 0.1775001375970059,
      "grad_norm": 0.7137764096260071,
      "learning_rate": 9.817629979838401e-06,
      "loss": 0.7715,
      "step": 3225
    },
    {
      "epoch": 0.17755517639936155,
      "grad_norm": 0.7237628102302551,
      "learning_rate": 9.81751395967364e-06,
      "loss": 0.7824,
      "step": 3226
    },
    {
      "epoch": 0.17761021520171721,
      "grad_norm": 0.9481163024902344,
      "learning_rate": 9.817397903301733e-06,
      "loss": 0.7451,
      "step": 3227
    },
    {
      "epoch": 0.17766525400407288,
      "grad_norm": 0.9472424387931824,
      "learning_rate": 9.817281810723552e-06,
      "loss": 0.8774,
      "step": 3228
    },
    {
      "epoch": 0.17772029280642854,
      "grad_norm": 0.9295538663864136,
      "learning_rate": 9.81716568193997e-06,
      "loss": 0.8507,
      "step": 3229
    },
    {
      "epoch": 0.1777753316087842,
      "grad_norm": 0.7668172717094421,
      "learning_rate": 9.817049516951863e-06,
      "loss": 0.8547,
      "step": 3230
    },
    {
      "epoch": 0.17783037041113986,
      "grad_norm": 0.8640413880348206,
      "learning_rate": 9.8169333157601e-06,
      "loss": 0.8485,
      "step": 3231
    },
    {
      "epoch": 0.17788540921349552,
      "grad_norm": 0.9901431798934937,
      "learning_rate": 9.816817078365554e-06,
      "loss": 0.9236,
      "step": 3232
    },
    {
      "epoch": 0.17794044801585118,
      "grad_norm": 1.0242371559143066,
      "learning_rate": 9.816700804769104e-06,
      "loss": 0.8096,
      "step": 3233
    },
    {
      "epoch": 0.17799548681820684,
      "grad_norm": 0.910498857498169,
      "learning_rate": 9.816584494971617e-06,
      "loss": 0.829,
      "step": 3234
    },
    {
      "epoch": 0.1780505256205625,
      "grad_norm": 0.8254473805427551,
      "learning_rate": 9.816468148973972e-06,
      "loss": 0.7828,
      "step": 3235
    },
    {
      "epoch": 0.17810556442291817,
      "grad_norm": 0.7971221804618835,
      "learning_rate": 9.816351766777039e-06,
      "loss": 0.8057,
      "step": 3236
    },
    {
      "epoch": 0.17816060322527383,
      "grad_norm": 0.8151674270629883,
      "learning_rate": 9.816235348381697e-06,
      "loss": 0.7801,
      "step": 3237
    },
    {
      "epoch": 0.1782156420276295,
      "grad_norm": 0.7587556838989258,
      "learning_rate": 9.81611889378882e-06,
      "loss": 0.7814,
      "step": 3238
    },
    {
      "epoch": 0.17827068082998515,
      "grad_norm": 0.8843516111373901,
      "learning_rate": 9.816002402999283e-06,
      "loss": 0.8873,
      "step": 3239
    },
    {
      "epoch": 0.1783257196323408,
      "grad_norm": 0.917859673500061,
      "learning_rate": 9.81588587601396e-06,
      "loss": 0.8963,
      "step": 3240
    },
    {
      "epoch": 0.17838075843469647,
      "grad_norm": 0.8256439566612244,
      "learning_rate": 9.815769312833727e-06,
      "loss": 0.9157,
      "step": 3241
    },
    {
      "epoch": 0.17843579723705214,
      "grad_norm": 0.8364603519439697,
      "learning_rate": 9.815652713459462e-06,
      "loss": 0.8253,
      "step": 3242
    },
    {
      "epoch": 0.17849083603940777,
      "grad_norm": 0.7717131972312927,
      "learning_rate": 9.81553607789204e-06,
      "loss": 0.7211,
      "step": 3243
    },
    {
      "epoch": 0.17854587484176343,
      "grad_norm": 0.8069111704826355,
      "learning_rate": 9.815419406132338e-06,
      "loss": 0.8986,
      "step": 3244
    },
    {
      "epoch": 0.1786009136441191,
      "grad_norm": 0.9176943302154541,
      "learning_rate": 9.815302698181233e-06,
      "loss": 0.8084,
      "step": 3245
    },
    {
      "epoch": 0.17865595244647475,
      "grad_norm": 0.769183874130249,
      "learning_rate": 9.815185954039601e-06,
      "loss": 0.8084,
      "step": 3246
    },
    {
      "epoch": 0.17871099124883041,
      "grad_norm": 0.8070697784423828,
      "learning_rate": 9.815069173708321e-06,
      "loss": 0.8371,
      "step": 3247
    },
    {
      "epoch": 0.17876603005118608,
      "grad_norm": 0.7837347388267517,
      "learning_rate": 9.81495235718827e-06,
      "loss": 0.8015,
      "step": 3248
    },
    {
      "epoch": 0.17882106885354174,
      "grad_norm": 0.9248430728912354,
      "learning_rate": 9.814835504480327e-06,
      "loss": 0.8396,
      "step": 3249
    },
    {
      "epoch": 0.1788761076558974,
      "grad_norm": 0.7914367914199829,
      "learning_rate": 9.814718615585367e-06,
      "loss": 0.8068,
      "step": 3250
    },
    {
      "epoch": 0.17893114645825306,
      "grad_norm": 0.8612570762634277,
      "learning_rate": 9.814601690504273e-06,
      "loss": 0.8227,
      "step": 3251
    },
    {
      "epoch": 0.17898618526060872,
      "grad_norm": 0.7476248741149902,
      "learning_rate": 9.81448472923792e-06,
      "loss": 0.8609,
      "step": 3252
    },
    {
      "epoch": 0.17904122406296438,
      "grad_norm": 0.7455218434333801,
      "learning_rate": 9.81436773178719e-06,
      "loss": 0.7992,
      "step": 3253
    },
    {
      "epoch": 0.17909626286532004,
      "grad_norm": 0.7917896509170532,
      "learning_rate": 9.814250698152958e-06,
      "loss": 0.8383,
      "step": 3254
    },
    {
      "epoch": 0.1791513016676757,
      "grad_norm": 0.6926130652427673,
      "learning_rate": 9.81413362833611e-06,
      "loss": 0.709,
      "step": 3255
    },
    {
      "epoch": 0.17920634047003137,
      "grad_norm": 0.8219630718231201,
      "learning_rate": 9.814016522337519e-06,
      "loss": 0.9387,
      "step": 3256
    },
    {
      "epoch": 0.17926137927238703,
      "grad_norm": 0.8588619828224182,
      "learning_rate": 9.81389938015807e-06,
      "loss": 0.8354,
      "step": 3257
    },
    {
      "epoch": 0.1793164180747427,
      "grad_norm": 0.7868718504905701,
      "learning_rate": 9.81378220179864e-06,
      "loss": 0.8464,
      "step": 3258
    },
    {
      "epoch": 0.17937145687709835,
      "grad_norm": 0.789479672908783,
      "learning_rate": 9.813664987260114e-06,
      "loss": 0.8577,
      "step": 3259
    },
    {
      "epoch": 0.179426495679454,
      "grad_norm": 0.8280717730522156,
      "learning_rate": 9.81354773654337e-06,
      "loss": 0.765,
      "step": 3260
    },
    {
      "epoch": 0.17948153448180967,
      "grad_norm": 0.7660181522369385,
      "learning_rate": 9.813430449649289e-06,
      "loss": 0.7116,
      "step": 3261
    },
    {
      "epoch": 0.17953657328416534,
      "grad_norm": 0.8043892979621887,
      "learning_rate": 9.813313126578754e-06,
      "loss": 0.8398,
      "step": 3262
    },
    {
      "epoch": 0.179591612086521,
      "grad_norm": 0.8708420991897583,
      "learning_rate": 9.813195767332647e-06,
      "loss": 0.8246,
      "step": 3263
    },
    {
      "epoch": 0.17964665088887666,
      "grad_norm": 1.1456964015960693,
      "learning_rate": 9.813078371911846e-06,
      "loss": 0.8798,
      "step": 3264
    },
    {
      "epoch": 0.17970168969123232,
      "grad_norm": 0.9668154716491699,
      "learning_rate": 9.812960940317238e-06,
      "loss": 0.9645,
      "step": 3265
    },
    {
      "epoch": 0.17975672849358798,
      "grad_norm": 0.862050473690033,
      "learning_rate": 9.812843472549705e-06,
      "loss": 0.8675,
      "step": 3266
    },
    {
      "epoch": 0.17981176729594364,
      "grad_norm": 0.7776491641998291,
      "learning_rate": 9.812725968610126e-06,
      "loss": 0.7727,
      "step": 3267
    },
    {
      "epoch": 0.1798668060982993,
      "grad_norm": 0.7197048664093018,
      "learning_rate": 9.812608428499389e-06,
      "loss": 0.6877,
      "step": 3268
    },
    {
      "epoch": 0.17992184490065496,
      "grad_norm": 0.7995713353157043,
      "learning_rate": 9.812490852218375e-06,
      "loss": 0.8576,
      "step": 3269
    },
    {
      "epoch": 0.17997688370301063,
      "grad_norm": 0.8300820589065552,
      "learning_rate": 9.812373239767967e-06,
      "loss": 0.8119,
      "step": 3270
    },
    {
      "epoch": 0.1800319225053663,
      "grad_norm": 0.8625856041908264,
      "learning_rate": 9.812255591149052e-06,
      "loss": 0.7547,
      "step": 3271
    },
    {
      "epoch": 0.18008696130772195,
      "grad_norm": 1.016419768333435,
      "learning_rate": 9.812137906362511e-06,
      "loss": 0.8457,
      "step": 3272
    },
    {
      "epoch": 0.1801420001100776,
      "grad_norm": 0.7303110361099243,
      "learning_rate": 9.812020185409229e-06,
      "loss": 0.7954,
      "step": 3273
    },
    {
      "epoch": 0.18019703891243327,
      "grad_norm": 0.8632498383522034,
      "learning_rate": 9.811902428290093e-06,
      "loss": 0.8952,
      "step": 3274
    },
    {
      "epoch": 0.18025207771478893,
      "grad_norm": 0.7666932940483093,
      "learning_rate": 9.811784635005984e-06,
      "loss": 0.746,
      "step": 3275
    },
    {
      "epoch": 0.1803071165171446,
      "grad_norm": 0.8962032198905945,
      "learning_rate": 9.811666805557791e-06,
      "loss": 0.8654,
      "step": 3276
    },
    {
      "epoch": 0.18036215531950026,
      "grad_norm": 0.9399656057357788,
      "learning_rate": 9.811548939946397e-06,
      "loss": 0.8062,
      "step": 3277
    },
    {
      "epoch": 0.18041719412185592,
      "grad_norm": 0.7469807863235474,
      "learning_rate": 9.811431038172692e-06,
      "loss": 0.79,
      "step": 3278
    },
    {
      "epoch": 0.18047223292421158,
      "grad_norm": 0.7661105394363403,
      "learning_rate": 9.811313100237556e-06,
      "loss": 0.7768,
      "step": 3279
    },
    {
      "epoch": 0.18052727172656724,
      "grad_norm": 0.7567458748817444,
      "learning_rate": 9.811195126141881e-06,
      "loss": 0.7329,
      "step": 3280
    },
    {
      "epoch": 0.1805823105289229,
      "grad_norm": 0.7187278866767883,
      "learning_rate": 9.811077115886552e-06,
      "loss": 0.6511,
      "step": 3281
    },
    {
      "epoch": 0.18063734933127856,
      "grad_norm": 0.7641230821609497,
      "learning_rate": 9.810959069472452e-06,
      "loss": 0.7704,
      "step": 3282
    },
    {
      "epoch": 0.18069238813363422,
      "grad_norm": 0.7790704369544983,
      "learning_rate": 9.810840986900474e-06,
      "loss": 0.8142,
      "step": 3283
    },
    {
      "epoch": 0.18074742693598989,
      "grad_norm": 0.8102816343307495,
      "learning_rate": 9.810722868171502e-06,
      "loss": 0.765,
      "step": 3284
    },
    {
      "epoch": 0.18080246573834555,
      "grad_norm": 0.7251957058906555,
      "learning_rate": 9.810604713286424e-06,
      "loss": 0.7836,
      "step": 3285
    },
    {
      "epoch": 0.18085750454070118,
      "grad_norm": 0.845348060131073,
      "learning_rate": 9.81048652224613e-06,
      "loss": 0.8386,
      "step": 3286
    },
    {
      "epoch": 0.18091254334305684,
      "grad_norm": 0.8397864103317261,
      "learning_rate": 9.810368295051507e-06,
      "loss": 0.805,
      "step": 3287
    },
    {
      "epoch": 0.1809675821454125,
      "grad_norm": 1.0739909410476685,
      "learning_rate": 9.810250031703444e-06,
      "loss": 0.8735,
      "step": 3288
    },
    {
      "epoch": 0.18102262094776816,
      "grad_norm": 0.752091646194458,
      "learning_rate": 9.810131732202826e-06,
      "loss": 0.7814,
      "step": 3289
    },
    {
      "epoch": 0.18107765975012383,
      "grad_norm": 0.7826841473579407,
      "learning_rate": 9.810013396550548e-06,
      "loss": 0.7761,
      "step": 3290
    },
    {
      "epoch": 0.1811326985524795,
      "grad_norm": 0.6979131102561951,
      "learning_rate": 9.809895024747498e-06,
      "loss": 0.672,
      "step": 3291
    },
    {
      "epoch": 0.18118773735483515,
      "grad_norm": 0.8571394085884094,
      "learning_rate": 9.809776616794562e-06,
      "loss": 0.8795,
      "step": 3292
    },
    {
      "epoch": 0.1812427761571908,
      "grad_norm": 0.8287902474403381,
      "learning_rate": 9.809658172692634e-06,
      "loss": 0.9032,
      "step": 3293
    },
    {
      "epoch": 0.18129781495954647,
      "grad_norm": 0.7884420156478882,
      "learning_rate": 9.809539692442602e-06,
      "loss": 0.87,
      "step": 3294
    },
    {
      "epoch": 0.18135285376190213,
      "grad_norm": 0.8955305218696594,
      "learning_rate": 9.809421176045358e-06,
      "loss": 0.7982,
      "step": 3295
    },
    {
      "epoch": 0.1814078925642578,
      "grad_norm": 0.7893335819244385,
      "learning_rate": 9.809302623501791e-06,
      "loss": 0.7792,
      "step": 3296
    },
    {
      "epoch": 0.18146293136661346,
      "grad_norm": 0.8077870011329651,
      "learning_rate": 9.809184034812794e-06,
      "loss": 0.829,
      "step": 3297
    },
    {
      "epoch": 0.18151797016896912,
      "grad_norm": 0.8282631635665894,
      "learning_rate": 9.809065409979256e-06,
      "loss": 0.8502,
      "step": 3298
    },
    {
      "epoch": 0.18157300897132478,
      "grad_norm": 0.7988418936729431,
      "learning_rate": 9.808946749002068e-06,
      "loss": 0.7853,
      "step": 3299
    },
    {
      "epoch": 0.18162804777368044,
      "grad_norm": 0.7776056528091431,
      "learning_rate": 9.808828051882127e-06,
      "loss": 0.7843,
      "step": 3300
    },
    {
      "epoch": 0.1816830865760361,
      "grad_norm": 0.8772258758544922,
      "learning_rate": 9.80870931862032e-06,
      "loss": 0.896,
      "step": 3301
    },
    {
      "epoch": 0.18173812537839176,
      "grad_norm": 0.8080328702926636,
      "learning_rate": 9.80859054921754e-06,
      "loss": 0.8464,
      "step": 3302
    },
    {
      "epoch": 0.18179316418074742,
      "grad_norm": 0.862707257270813,
      "learning_rate": 9.808471743674682e-06,
      "loss": 0.8732,
      "step": 3303
    },
    {
      "epoch": 0.18184820298310309,
      "grad_norm": 1.1964820623397827,
      "learning_rate": 9.808352901992637e-06,
      "loss": 0.9911,
      "step": 3304
    },
    {
      "epoch": 0.18190324178545875,
      "grad_norm": 0.8597685694694519,
      "learning_rate": 9.808234024172298e-06,
      "loss": 0.8724,
      "step": 3305
    },
    {
      "epoch": 0.1819582805878144,
      "grad_norm": 0.8068556189537048,
      "learning_rate": 9.80811511021456e-06,
      "loss": 0.8116,
      "step": 3306
    },
    {
      "epoch": 0.18201331939017007,
      "grad_norm": 1.0014268159866333,
      "learning_rate": 9.807996160120317e-06,
      "loss": 0.8585,
      "step": 3307
    },
    {
      "epoch": 0.18206835819252573,
      "grad_norm": 0.8541132807731628,
      "learning_rate": 9.80787717389046e-06,
      "loss": 0.8505,
      "step": 3308
    },
    {
      "epoch": 0.1821233969948814,
      "grad_norm": 0.7973629832267761,
      "learning_rate": 9.807758151525886e-06,
      "loss": 0.8312,
      "step": 3309
    },
    {
      "epoch": 0.18217843579723705,
      "grad_norm": 0.82973712682724,
      "learning_rate": 9.807639093027488e-06,
      "loss": 0.894,
      "step": 3310
    },
    {
      "epoch": 0.18223347459959272,
      "grad_norm": 0.7729674577713013,
      "learning_rate": 9.807519998396162e-06,
      "loss": 0.7459,
      "step": 3311
    },
    {
      "epoch": 0.18228851340194838,
      "grad_norm": 0.8106189370155334,
      "learning_rate": 9.807400867632804e-06,
      "loss": 0.914,
      "step": 3312
    },
    {
      "epoch": 0.18234355220430404,
      "grad_norm": 0.7672377228736877,
      "learning_rate": 9.807281700738305e-06,
      "loss": 0.8475,
      "step": 3313
    },
    {
      "epoch": 0.1823985910066597,
      "grad_norm": 0.8776688575744629,
      "learning_rate": 9.807162497713566e-06,
      "loss": 0.7641,
      "step": 3314
    },
    {
      "epoch": 0.18245362980901536,
      "grad_norm": 0.8781917691230774,
      "learning_rate": 9.807043258559479e-06,
      "loss": 0.86,
      "step": 3315
    },
    {
      "epoch": 0.18250866861137102,
      "grad_norm": 0.819362998008728,
      "learning_rate": 9.806923983276942e-06,
      "loss": 0.8829,
      "step": 3316
    },
    {
      "epoch": 0.18256370741372668,
      "grad_norm": 0.8065270185470581,
      "learning_rate": 9.80680467186685e-06,
      "loss": 0.7335,
      "step": 3317
    },
    {
      "epoch": 0.18261874621608234,
      "grad_norm": 0.8692485690116882,
      "learning_rate": 9.806685324330102e-06,
      "loss": 0.8582,
      "step": 3318
    },
    {
      "epoch": 0.182673785018438,
      "grad_norm": 0.7910160422325134,
      "learning_rate": 9.806565940667594e-06,
      "loss": 0.8569,
      "step": 3319
    },
    {
      "epoch": 0.18272882382079367,
      "grad_norm": 0.8282253742218018,
      "learning_rate": 9.806446520880225e-06,
      "loss": 0.7791,
      "step": 3320
    },
    {
      "epoch": 0.18278386262314933,
      "grad_norm": 0.7513861060142517,
      "learning_rate": 9.806327064968887e-06,
      "loss": 0.7287,
      "step": 3321
    },
    {
      "epoch": 0.182838901425505,
      "grad_norm": 0.8141188621520996,
      "learning_rate": 9.806207572934483e-06,
      "loss": 0.7772,
      "step": 3322
    },
    {
      "epoch": 0.18289394022786065,
      "grad_norm": 0.7963125705718994,
      "learning_rate": 9.806088044777909e-06,
      "loss": 0.7993,
      "step": 3323
    },
    {
      "epoch": 0.1829489790302163,
      "grad_norm": 0.8527218103408813,
      "learning_rate": 9.805968480500063e-06,
      "loss": 0.822,
      "step": 3324
    },
    {
      "epoch": 0.18300401783257197,
      "grad_norm": 0.822467565536499,
      "learning_rate": 9.805848880101845e-06,
      "loss": 0.8606,
      "step": 3325
    },
    {
      "epoch": 0.18305905663492764,
      "grad_norm": 0.8197154402732849,
      "learning_rate": 9.805729243584154e-06,
      "loss": 0.9004,
      "step": 3326
    },
    {
      "epoch": 0.1831140954372833,
      "grad_norm": 0.8379594683647156,
      "learning_rate": 9.805609570947887e-06,
      "loss": 0.8467,
      "step": 3327
    },
    {
      "epoch": 0.18316913423963896,
      "grad_norm": 0.7787355184555054,
      "learning_rate": 9.805489862193947e-06,
      "loss": 0.8221,
      "step": 3328
    },
    {
      "epoch": 0.1832241730419946,
      "grad_norm": 0.8464100956916809,
      "learning_rate": 9.80537011732323e-06,
      "loss": 0.7722,
      "step": 3329
    },
    {
      "epoch": 0.18327921184435025,
      "grad_norm": 0.8351306319236755,
      "learning_rate": 9.805250336336637e-06,
      "loss": 0.7638,
      "step": 3330
    },
    {
      "epoch": 0.18333425064670592,
      "grad_norm": 0.8098864555358887,
      "learning_rate": 9.805130519235068e-06,
      "loss": 0.8448,
      "step": 3331
    },
    {
      "epoch": 0.18338928944906158,
      "grad_norm": 0.8290563821792603,
      "learning_rate": 9.805010666019427e-06,
      "loss": 0.6574,
      "step": 3332
    },
    {
      "epoch": 0.18344432825141724,
      "grad_norm": 0.7748262882232666,
      "learning_rate": 9.804890776690611e-06,
      "loss": 0.8002,
      "step": 3333
    },
    {
      "epoch": 0.1834993670537729,
      "grad_norm": 0.8422787189483643,
      "learning_rate": 9.80477085124952e-06,
      "loss": 0.8452,
      "step": 3334
    },
    {
      "epoch": 0.18355440585612856,
      "grad_norm": 0.7776510119438171,
      "learning_rate": 9.804650889697061e-06,
      "loss": 0.8774,
      "step": 3335
    },
    {
      "epoch": 0.18360944465848422,
      "grad_norm": 0.8449370861053467,
      "learning_rate": 9.80453089203413e-06,
      "loss": 0.8233,
      "step": 3336
    },
    {
      "epoch": 0.18366448346083988,
      "grad_norm": 0.8254217505455017,
      "learning_rate": 9.804410858261632e-06,
      "loss": 0.8778,
      "step": 3337
    },
    {
      "epoch": 0.18371952226319554,
      "grad_norm": 0.8673515915870667,
      "learning_rate": 9.804290788380466e-06,
      "loss": 0.8005,
      "step": 3338
    },
    {
      "epoch": 0.1837745610655512,
      "grad_norm": 0.8106067776679993,
      "learning_rate": 9.804170682391538e-06,
      "loss": 0.86,
      "step": 3339
    },
    {
      "epoch": 0.18382959986790687,
      "grad_norm": 0.8211669325828552,
      "learning_rate": 9.804050540295749e-06,
      "loss": 0.8013,
      "step": 3340
    },
    {
      "epoch": 0.18388463867026253,
      "grad_norm": 0.7866180539131165,
      "learning_rate": 9.803930362094003e-06,
      "loss": 0.8108,
      "step": 3341
    },
    {
      "epoch": 0.1839396774726182,
      "grad_norm": 0.8192055225372314,
      "learning_rate": 9.8038101477872e-06,
      "loss": 0.7586,
      "step": 3342
    },
    {
      "epoch": 0.18399471627497385,
      "grad_norm": 0.940910279750824,
      "learning_rate": 9.803689897376248e-06,
      "loss": 0.8174,
      "step": 3343
    },
    {
      "epoch": 0.1840497550773295,
      "grad_norm": 0.7979292869567871,
      "learning_rate": 9.803569610862048e-06,
      "loss": 0.8341,
      "step": 3344
    },
    {
      "epoch": 0.18410479387968517,
      "grad_norm": 0.7577546238899231,
      "learning_rate": 9.803449288245504e-06,
      "loss": 0.7775,
      "step": 3345
    },
    {
      "epoch": 0.18415983268204084,
      "grad_norm": 0.7255160212516785,
      "learning_rate": 9.80332892952752e-06,
      "loss": 0.7648,
      "step": 3346
    },
    {
      "epoch": 0.1842148714843965,
      "grad_norm": 0.8269388675689697,
      "learning_rate": 9.803208534709004e-06,
      "loss": 0.8902,
      "step": 3347
    },
    {
      "epoch": 0.18426991028675216,
      "grad_norm": 0.783867359161377,
      "learning_rate": 9.803088103790857e-06,
      "loss": 0.8191,
      "step": 3348
    },
    {
      "epoch": 0.18432494908910782,
      "grad_norm": 0.7658863663673401,
      "learning_rate": 9.802967636773986e-06,
      "loss": 0.7505,
      "step": 3349
    },
    {
      "epoch": 0.18437998789146348,
      "grad_norm": 0.701225757598877,
      "learning_rate": 9.802847133659294e-06,
      "loss": 0.7159,
      "step": 3350
    },
    {
      "epoch": 0.18443502669381914,
      "grad_norm": 0.9224311709403992,
      "learning_rate": 9.802726594447692e-06,
      "loss": 0.7766,
      "step": 3351
    },
    {
      "epoch": 0.1844900654961748,
      "grad_norm": 0.8835979700088501,
      "learning_rate": 9.80260601914008e-06,
      "loss": 0.9304,
      "step": 3352
    },
    {
      "epoch": 0.18454510429853047,
      "grad_norm": 0.7918481826782227,
      "learning_rate": 9.802485407737368e-06,
      "loss": 0.7691,
      "step": 3353
    },
    {
      "epoch": 0.18460014310088613,
      "grad_norm": 0.8855286240577698,
      "learning_rate": 9.80236476024046e-06,
      "loss": 0.9213,
      "step": 3354
    },
    {
      "epoch": 0.1846551819032418,
      "grad_norm": 0.7863314747810364,
      "learning_rate": 9.802244076650264e-06,
      "loss": 0.7675,
      "step": 3355
    },
    {
      "epoch": 0.18471022070559745,
      "grad_norm": 0.8230198621749878,
      "learning_rate": 9.802123356967687e-06,
      "loss": 0.7243,
      "step": 3356
    },
    {
      "epoch": 0.1847652595079531,
      "grad_norm": 0.8038737773895264,
      "learning_rate": 9.80200260119364e-06,
      "loss": 0.8094,
      "step": 3357
    },
    {
      "epoch": 0.18482029831030877,
      "grad_norm": 0.7656993269920349,
      "learning_rate": 9.801881809329022e-06,
      "loss": 0.7736,
      "step": 3358
    },
    {
      "epoch": 0.18487533711266443,
      "grad_norm": 0.8222082853317261,
      "learning_rate": 9.801760981374747e-06,
      "loss": 0.844,
      "step": 3359
    },
    {
      "epoch": 0.1849303759150201,
      "grad_norm": 0.7632889747619629,
      "learning_rate": 9.801640117331723e-06,
      "loss": 0.8354,
      "step": 3360
    },
    {
      "epoch": 0.18498541471737576,
      "grad_norm": 0.8308513760566711,
      "learning_rate": 9.801519217200857e-06,
      "loss": 0.8277,
      "step": 3361
    },
    {
      "epoch": 0.18504045351973142,
      "grad_norm": 0.7865434885025024,
      "learning_rate": 9.801398280983057e-06,
      "loss": 0.8614,
      "step": 3362
    },
    {
      "epoch": 0.18509549232208708,
      "grad_norm": 0.7249410152435303,
      "learning_rate": 9.801277308679232e-06,
      "loss": 0.7259,
      "step": 3363
    },
    {
      "epoch": 0.18515053112444274,
      "grad_norm": 0.7604461908340454,
      "learning_rate": 9.801156300290293e-06,
      "loss": 0.8507,
      "step": 3364
    },
    {
      "epoch": 0.1852055699267984,
      "grad_norm": 0.8725959062576294,
      "learning_rate": 9.801035255817149e-06,
      "loss": 0.7688,
      "step": 3365
    },
    {
      "epoch": 0.18526060872915406,
      "grad_norm": 0.7798827290534973,
      "learning_rate": 9.800914175260708e-06,
      "loss": 0.8788,
      "step": 3366
    },
    {
      "epoch": 0.18531564753150973,
      "grad_norm": 0.7060996890068054,
      "learning_rate": 9.800793058621882e-06,
      "loss": 0.8183,
      "step": 3367
    },
    {
      "epoch": 0.1853706863338654,
      "grad_norm": 0.7558063268661499,
      "learning_rate": 9.80067190590158e-06,
      "loss": 0.7834,
      "step": 3368
    },
    {
      "epoch": 0.18542572513622105,
      "grad_norm": 0.7411057353019714,
      "learning_rate": 9.800550717100714e-06,
      "loss": 0.8298,
      "step": 3369
    },
    {
      "epoch": 0.1854807639385767,
      "grad_norm": 0.8466144800186157,
      "learning_rate": 9.800429492220193e-06,
      "loss": 0.8297,
      "step": 3370
    },
    {
      "epoch": 0.18553580274093237,
      "grad_norm": 0.7302330136299133,
      "learning_rate": 9.800308231260928e-06,
      "loss": 0.72,
      "step": 3371
    },
    {
      "epoch": 0.185590841543288,
      "grad_norm": 0.8140530586242676,
      "learning_rate": 9.800186934223832e-06,
      "loss": 0.9287,
      "step": 3372
    },
    {
      "epoch": 0.18564588034564367,
      "grad_norm": 0.8246129751205444,
      "learning_rate": 9.800065601109817e-06,
      "loss": 0.7891,
      "step": 3373
    },
    {
      "epoch": 0.18570091914799933,
      "grad_norm": 0.8746623396873474,
      "learning_rate": 9.799944231919794e-06,
      "loss": 0.8549,
      "step": 3374
    },
    {
      "epoch": 0.185755957950355,
      "grad_norm": 0.9977195858955383,
      "learning_rate": 9.799822826654672e-06,
      "loss": 0.821,
      "step": 3375
    },
    {
      "epoch": 0.18581099675271065,
      "grad_norm": 0.8937395811080933,
      "learning_rate": 9.79970138531537e-06,
      "loss": 0.8639,
      "step": 3376
    },
    {
      "epoch": 0.1858660355550663,
      "grad_norm": 1.039695143699646,
      "learning_rate": 9.799579907902794e-06,
      "loss": 1.0425,
      "step": 3377
    },
    {
      "epoch": 0.18592107435742197,
      "grad_norm": 0.7847749590873718,
      "learning_rate": 9.799458394417863e-06,
      "loss": 0.8505,
      "step": 3378
    },
    {
      "epoch": 0.18597611315977763,
      "grad_norm": 0.760334312915802,
      "learning_rate": 9.799336844861486e-06,
      "loss": 0.7418,
      "step": 3379
    },
    {
      "epoch": 0.1860311519621333,
      "grad_norm": 0.7599604725837708,
      "learning_rate": 9.799215259234578e-06,
      "loss": 0.8305,
      "step": 3380
    },
    {
      "epoch": 0.18608619076448896,
      "grad_norm": 0.846767246723175,
      "learning_rate": 9.799093637538054e-06,
      "loss": 0.7526,
      "step": 3381
    },
    {
      "epoch": 0.18614122956684462,
      "grad_norm": 0.7840956449508667,
      "learning_rate": 9.798971979772825e-06,
      "loss": 0.8009,
      "step": 3382
    },
    {
      "epoch": 0.18619626836920028,
      "grad_norm": 0.7826499342918396,
      "learning_rate": 9.798850285939809e-06,
      "loss": 0.821,
      "step": 3383
    },
    {
      "epoch": 0.18625130717155594,
      "grad_norm": 0.7829813361167908,
      "learning_rate": 9.798728556039918e-06,
      "loss": 0.8053,
      "step": 3384
    },
    {
      "epoch": 0.1863063459739116,
      "grad_norm": 0.7267470359802246,
      "learning_rate": 9.798606790074067e-06,
      "loss": 0.6797,
      "step": 3385
    },
    {
      "epoch": 0.18636138477626726,
      "grad_norm": 0.8560196757316589,
      "learning_rate": 9.798484988043173e-06,
      "loss": 0.8476,
      "step": 3386
    },
    {
      "epoch": 0.18641642357862293,
      "grad_norm": 0.7920921444892883,
      "learning_rate": 9.798363149948148e-06,
      "loss": 0.8832,
      "step": 3387
    },
    {
      "epoch": 0.1864714623809786,
      "grad_norm": 0.8414384126663208,
      "learning_rate": 9.798241275789912e-06,
      "loss": 0.8607,
      "step": 3388
    },
    {
      "epoch": 0.18652650118333425,
      "grad_norm": 0.7255431413650513,
      "learning_rate": 9.798119365569378e-06,
      "loss": 0.6426,
      "step": 3389
    },
    {
      "epoch": 0.1865815399856899,
      "grad_norm": 0.8842852711677551,
      "learning_rate": 9.797997419287465e-06,
      "loss": 0.9058,
      "step": 3390
    },
    {
      "epoch": 0.18663657878804557,
      "grad_norm": 0.7178265452384949,
      "learning_rate": 9.797875436945086e-06,
      "loss": 0.8134,
      "step": 3391
    },
    {
      "epoch": 0.18669161759040123,
      "grad_norm": 0.7275096774101257,
      "learning_rate": 9.797753418543161e-06,
      "loss": 0.6858,
      "step": 3392
    },
    {
      "epoch": 0.1867466563927569,
      "grad_norm": 0.7587800025939941,
      "learning_rate": 9.797631364082605e-06,
      "loss": 0.7437,
      "step": 3393
    },
    {
      "epoch": 0.18680169519511255,
      "grad_norm": 0.9769744873046875,
      "learning_rate": 9.797509273564336e-06,
      "loss": 0.8024,
      "step": 3394
    },
    {
      "epoch": 0.18685673399746822,
      "grad_norm": 0.7662433385848999,
      "learning_rate": 9.79738714698927e-06,
      "loss": 0.8122,
      "step": 3395
    },
    {
      "epoch": 0.18691177279982388,
      "grad_norm": 0.8620306849479675,
      "learning_rate": 9.797264984358328e-06,
      "loss": 0.7952,
      "step": 3396
    },
    {
      "epoch": 0.18696681160217954,
      "grad_norm": 0.7542591094970703,
      "learning_rate": 9.797142785672427e-06,
      "loss": 0.8315,
      "step": 3397
    },
    {
      "epoch": 0.1870218504045352,
      "grad_norm": 0.7273713946342468,
      "learning_rate": 9.797020550932483e-06,
      "loss": 0.7316,
      "step": 3398
    },
    {
      "epoch": 0.18707688920689086,
      "grad_norm": 1.031592845916748,
      "learning_rate": 9.796898280139417e-06,
      "loss": 0.7478,
      "step": 3399
    },
    {
      "epoch": 0.18713192800924652,
      "grad_norm": 0.791407585144043,
      "learning_rate": 9.796775973294147e-06,
      "loss": 0.7742,
      "step": 3400
    },
    {
      "epoch": 0.18718696681160218,
      "grad_norm": 0.8311418294906616,
      "learning_rate": 9.796653630397595e-06,
      "loss": 0.8182,
      "step": 3401
    },
    {
      "epoch": 0.18724200561395785,
      "grad_norm": 0.7960993051528931,
      "learning_rate": 9.796531251450678e-06,
      "loss": 0.7606,
      "step": 3402
    },
    {
      "epoch": 0.1872970444163135,
      "grad_norm": 0.8671618103981018,
      "learning_rate": 9.796408836454316e-06,
      "loss": 0.7136,
      "step": 3403
    },
    {
      "epoch": 0.18735208321866917,
      "grad_norm": 1.1071348190307617,
      "learning_rate": 9.796286385409428e-06,
      "loss": 0.7729,
      "step": 3404
    },
    {
      "epoch": 0.18740712202102483,
      "grad_norm": 0.738217294216156,
      "learning_rate": 9.796163898316935e-06,
      "loss": 0.7425,
      "step": 3405
    },
    {
      "epoch": 0.1874621608233805,
      "grad_norm": 0.7567199468612671,
      "learning_rate": 9.796041375177758e-06,
      "loss": 0.8442,
      "step": 3406
    },
    {
      "epoch": 0.18751719962573615,
      "grad_norm": 0.7942413091659546,
      "learning_rate": 9.79591881599282e-06,
      "loss": 0.852,
      "step": 3407
    },
    {
      "epoch": 0.18757223842809181,
      "grad_norm": 0.7529355883598328,
      "learning_rate": 9.795796220763038e-06,
      "loss": 0.8086,
      "step": 3408
    },
    {
      "epoch": 0.18762727723044748,
      "grad_norm": 0.7645192742347717,
      "learning_rate": 9.795673589489337e-06,
      "loss": 0.831,
      "step": 3409
    },
    {
      "epoch": 0.18768231603280314,
      "grad_norm": 0.694791853427887,
      "learning_rate": 9.795550922172635e-06,
      "loss": 0.6919,
      "step": 3410
    },
    {
      "epoch": 0.1877373548351588,
      "grad_norm": 0.7041944265365601,
      "learning_rate": 9.795428218813858e-06,
      "loss": 0.7284,
      "step": 3411
    },
    {
      "epoch": 0.18779239363751446,
      "grad_norm": 0.8972276449203491,
      "learning_rate": 9.795305479413924e-06,
      "loss": 0.7156,
      "step": 3412
    },
    {
      "epoch": 0.18784743243987012,
      "grad_norm": 0.9730873107910156,
      "learning_rate": 9.795182703973758e-06,
      "loss": 0.8739,
      "step": 3413
    },
    {
      "epoch": 0.18790247124222578,
      "grad_norm": 0.8137956261634827,
      "learning_rate": 9.795059892494283e-06,
      "loss": 0.8189,
      "step": 3414
    },
    {
      "epoch": 0.18795751004458142,
      "grad_norm": 0.8171416521072388,
      "learning_rate": 9.794937044976422e-06,
      "loss": 0.9449,
      "step": 3415
    },
    {
      "epoch": 0.18801254884693708,
      "grad_norm": 0.7929911017417908,
      "learning_rate": 9.794814161421098e-06,
      "loss": 0.8034,
      "step": 3416
    },
    {
      "epoch": 0.18806758764929274,
      "grad_norm": 1.1045749187469482,
      "learning_rate": 9.794691241829233e-06,
      "loss": 0.875,
      "step": 3417
    },
    {
      "epoch": 0.1881226264516484,
      "grad_norm": 0.8141040205955505,
      "learning_rate": 9.794568286201752e-06,
      "loss": 0.787,
      "step": 3418
    },
    {
      "epoch": 0.18817766525400406,
      "grad_norm": 0.7615541815757751,
      "learning_rate": 9.79444529453958e-06,
      "loss": 0.8491,
      "step": 3419
    },
    {
      "epoch": 0.18823270405635972,
      "grad_norm": 0.848419189453125,
      "learning_rate": 9.79432226684364e-06,
      "loss": 0.7445,
      "step": 3420
    },
    {
      "epoch": 0.18828774285871538,
      "grad_norm": 0.8075067400932312,
      "learning_rate": 9.794199203114858e-06,
      "loss": 0.6581,
      "step": 3421
    },
    {
      "epoch": 0.18834278166107105,
      "grad_norm": 0.8473401069641113,
      "learning_rate": 9.794076103354158e-06,
      "loss": 0.839,
      "step": 3422
    },
    {
      "epoch": 0.1883978204634267,
      "grad_norm": 0.8211609721183777,
      "learning_rate": 9.793952967562463e-06,
      "loss": 0.7709,
      "step": 3423
    },
    {
      "epoch": 0.18845285926578237,
      "grad_norm": 0.7527804374694824,
      "learning_rate": 9.793829795740703e-06,
      "loss": 0.7315,
      "step": 3424
    },
    {
      "epoch": 0.18850789806813803,
      "grad_norm": 0.7971188426017761,
      "learning_rate": 9.793706587889802e-06,
      "loss": 0.7507,
      "step": 3425
    },
    {
      "epoch": 0.1885629368704937,
      "grad_norm": 1.024066686630249,
      "learning_rate": 9.793583344010684e-06,
      "loss": 0.9043,
      "step": 3426
    },
    {
      "epoch": 0.18861797567284935,
      "grad_norm": 0.7428625226020813,
      "learning_rate": 9.793460064104276e-06,
      "loss": 0.7435,
      "step": 3427
    },
    {
      "epoch": 0.18867301447520501,
      "grad_norm": 0.8438264727592468,
      "learning_rate": 9.793336748171507e-06,
      "loss": 0.8618,
      "step": 3428
    },
    {
      "epoch": 0.18872805327756068,
      "grad_norm": 0.7846877574920654,
      "learning_rate": 9.793213396213302e-06,
      "loss": 0.8064,
      "step": 3429
    },
    {
      "epoch": 0.18878309207991634,
      "grad_norm": 0.7527204751968384,
      "learning_rate": 9.793090008230587e-06,
      "loss": 0.7596,
      "step": 3430
    },
    {
      "epoch": 0.188838130882272,
      "grad_norm": 1.1236757040023804,
      "learning_rate": 9.792966584224292e-06,
      "loss": 0.8292,
      "step": 3431
    },
    {
      "epoch": 0.18889316968462766,
      "grad_norm": 0.8128102421760559,
      "learning_rate": 9.792843124195343e-06,
      "loss": 0.8073,
      "step": 3432
    },
    {
      "epoch": 0.18894820848698332,
      "grad_norm": 0.7668742537498474,
      "learning_rate": 9.792719628144667e-06,
      "loss": 0.7848,
      "step": 3433
    },
    {
      "epoch": 0.18900324728933898,
      "grad_norm": 1.8663485050201416,
      "learning_rate": 9.792596096073193e-06,
      "loss": 0.9388,
      "step": 3434
    },
    {
      "epoch": 0.18905828609169464,
      "grad_norm": 0.8066239356994629,
      "learning_rate": 9.792472527981852e-06,
      "loss": 0.6647,
      "step": 3435
    },
    {
      "epoch": 0.1891133248940503,
      "grad_norm": 0.8268817067146301,
      "learning_rate": 9.792348923871567e-06,
      "loss": 0.9676,
      "step": 3436
    },
    {
      "epoch": 0.18916836369640597,
      "grad_norm": 0.7165037393569946,
      "learning_rate": 9.792225283743272e-06,
      "loss": 0.6937,
      "step": 3437
    },
    {
      "epoch": 0.18922340249876163,
      "grad_norm": 0.7850403785705566,
      "learning_rate": 9.792101607597895e-06,
      "loss": 0.7782,
      "step": 3438
    },
    {
      "epoch": 0.1892784413011173,
      "grad_norm": 0.8839808702468872,
      "learning_rate": 9.791977895436365e-06,
      "loss": 0.7639,
      "step": 3439
    },
    {
      "epoch": 0.18933348010347295,
      "grad_norm": 0.8260362148284912,
      "learning_rate": 9.791854147259611e-06,
      "loss": 0.8201,
      "step": 3440
    },
    {
      "epoch": 0.1893885189058286,
      "grad_norm": 0.8792916536331177,
      "learning_rate": 9.791730363068564e-06,
      "loss": 0.8251,
      "step": 3441
    },
    {
      "epoch": 0.18944355770818427,
      "grad_norm": 0.8192774653434753,
      "learning_rate": 9.791606542864154e-06,
      "loss": 0.7944,
      "step": 3442
    },
    {
      "epoch": 0.18949859651053994,
      "grad_norm": 0.751470685005188,
      "learning_rate": 9.791482686647313e-06,
      "loss": 0.7563,
      "step": 3443
    },
    {
      "epoch": 0.1895536353128956,
      "grad_norm": 0.8902072906494141,
      "learning_rate": 9.79135879441897e-06,
      "loss": 0.7719,
      "step": 3444
    },
    {
      "epoch": 0.18960867411525126,
      "grad_norm": 0.7166435122489929,
      "learning_rate": 9.791234866180058e-06,
      "loss": 0.7871,
      "step": 3445
    },
    {
      "epoch": 0.18966371291760692,
      "grad_norm": 0.763416588306427,
      "learning_rate": 9.791110901931505e-06,
      "loss": 0.8226,
      "step": 3446
    },
    {
      "epoch": 0.18971875171996258,
      "grad_norm": 0.806633472442627,
      "learning_rate": 9.790986901674246e-06,
      "loss": 0.7828,
      "step": 3447
    },
    {
      "epoch": 0.18977379052231824,
      "grad_norm": 0.8139312863349915,
      "learning_rate": 9.790862865409213e-06,
      "loss": 0.8441,
      "step": 3448
    },
    {
      "epoch": 0.1898288293246739,
      "grad_norm": 0.8362452387809753,
      "learning_rate": 9.790738793137335e-06,
      "loss": 0.8765,
      "step": 3449
    },
    {
      "epoch": 0.18988386812702956,
      "grad_norm": 0.7736263871192932,
      "learning_rate": 9.790614684859549e-06,
      "loss": 0.8373,
      "step": 3450
    },
    {
      "epoch": 0.18993890692938523,
      "grad_norm": 0.8742800354957581,
      "learning_rate": 9.790490540576784e-06,
      "loss": 0.8976,
      "step": 3451
    },
    {
      "epoch": 0.1899939457317409,
      "grad_norm": 0.701505720615387,
      "learning_rate": 9.790366360289974e-06,
      "loss": 0.7799,
      "step": 3452
    },
    {
      "epoch": 0.19004898453409655,
      "grad_norm": 0.7771356701850891,
      "learning_rate": 9.790242144000055e-06,
      "loss": 0.7617,
      "step": 3453
    },
    {
      "epoch": 0.1901040233364522,
      "grad_norm": 0.897576093673706,
      "learning_rate": 9.790117891707955e-06,
      "loss": 0.7817,
      "step": 3454
    },
    {
      "epoch": 0.19015906213880787,
      "grad_norm": 0.7296561002731323,
      "learning_rate": 9.789993603414613e-06,
      "loss": 0.8344,
      "step": 3455
    },
    {
      "epoch": 0.19021410094116353,
      "grad_norm": 0.8099396228790283,
      "learning_rate": 9.789869279120962e-06,
      "loss": 0.7369,
      "step": 3456
    },
    {
      "epoch": 0.1902691397435192,
      "grad_norm": 0.7802554368972778,
      "learning_rate": 9.789744918827935e-06,
      "loss": 0.8383,
      "step": 3457
    },
    {
      "epoch": 0.19032417854587483,
      "grad_norm": 0.7508029341697693,
      "learning_rate": 9.789620522536467e-06,
      "loss": 0.825,
      "step": 3458
    },
    {
      "epoch": 0.1903792173482305,
      "grad_norm": 0.7782164216041565,
      "learning_rate": 9.789496090247494e-06,
      "loss": 0.7737,
      "step": 3459
    },
    {
      "epoch": 0.19043425615058615,
      "grad_norm": 0.7711489796638489,
      "learning_rate": 9.78937162196195e-06,
      "loss": 0.7694,
      "step": 3460
    },
    {
      "epoch": 0.1904892949529418,
      "grad_norm": 0.821579098701477,
      "learning_rate": 9.789247117680769e-06,
      "loss": 0.7493,
      "step": 3461
    },
    {
      "epoch": 0.19054433375529747,
      "grad_norm": 0.6700833439826965,
      "learning_rate": 9.789122577404892e-06,
      "loss": 0.7696,
      "step": 3462
    },
    {
      "epoch": 0.19059937255765314,
      "grad_norm": 0.854340136051178,
      "learning_rate": 9.78899800113525e-06,
      "loss": 0.9503,
      "step": 3463
    },
    {
      "epoch": 0.1906544113600088,
      "grad_norm": 0.8095537424087524,
      "learning_rate": 9.78887338887278e-06,
      "loss": 0.8435,
      "step": 3464
    },
    {
      "epoch": 0.19070945016236446,
      "grad_norm": 0.8156480193138123,
      "learning_rate": 9.78874874061842e-06,
      "loss": 0.8561,
      "step": 3465
    },
    {
      "epoch": 0.19076448896472012,
      "grad_norm": 0.8065482378005981,
      "learning_rate": 9.788624056373108e-06,
      "loss": 0.7793,
      "step": 3466
    },
    {
      "epoch": 0.19081952776707578,
      "grad_norm": 0.789601743221283,
      "learning_rate": 9.788499336137778e-06,
      "loss": 0.7523,
      "step": 3467
    },
    {
      "epoch": 0.19087456656943144,
      "grad_norm": 0.8322301506996155,
      "learning_rate": 9.788374579913369e-06,
      "loss": 0.9034,
      "step": 3468
    },
    {
      "epoch": 0.1909296053717871,
      "grad_norm": 0.8194506764411926,
      "learning_rate": 9.788249787700818e-06,
      "loss": 0.8601,
      "step": 3469
    },
    {
      "epoch": 0.19098464417414276,
      "grad_norm": 0.8419962525367737,
      "learning_rate": 9.788124959501065e-06,
      "loss": 0.869,
      "step": 3470
    },
    {
      "epoch": 0.19103968297649843,
      "grad_norm": 0.760637104511261,
      "learning_rate": 9.788000095315044e-06,
      "loss": 0.7293,
      "step": 3471
    },
    {
      "epoch": 0.1910947217788541,
      "grad_norm": 1.3964574337005615,
      "learning_rate": 9.787875195143697e-06,
      "loss": 0.8032,
      "step": 3472
    },
    {
      "epoch": 0.19114976058120975,
      "grad_norm": 0.8205012679100037,
      "learning_rate": 9.787750258987962e-06,
      "loss": 0.8868,
      "step": 3473
    },
    {
      "epoch": 0.1912047993835654,
      "grad_norm": 0.8183104991912842,
      "learning_rate": 9.78762528684878e-06,
      "loss": 0.7531,
      "step": 3474
    },
    {
      "epoch": 0.19125983818592107,
      "grad_norm": 0.7659775018692017,
      "learning_rate": 9.787500278727083e-06,
      "loss": 0.8081,
      "step": 3475
    },
    {
      "epoch": 0.19131487698827673,
      "grad_norm": 0.8262091279029846,
      "learning_rate": 9.787375234623819e-06,
      "loss": 0.82,
      "step": 3476
    },
    {
      "epoch": 0.1913699157906324,
      "grad_norm": 0.857761025428772,
      "learning_rate": 9.787250154539923e-06,
      "loss": 0.9133,
      "step": 3477
    },
    {
      "epoch": 0.19142495459298806,
      "grad_norm": 0.7551915645599365,
      "learning_rate": 9.787125038476334e-06,
      "loss": 0.7822,
      "step": 3478
    },
    {
      "epoch": 0.19147999339534372,
      "grad_norm": 0.7777357697486877,
      "learning_rate": 9.786999886433998e-06,
      "loss": 0.7676,
      "step": 3479
    },
    {
      "epoch": 0.19153503219769938,
      "grad_norm": 0.8389080166816711,
      "learning_rate": 9.786874698413852e-06,
      "loss": 0.7901,
      "step": 3480
    },
    {
      "epoch": 0.19159007100005504,
      "grad_norm": 0.7894837856292725,
      "learning_rate": 9.786749474416836e-06,
      "loss": 0.8393,
      "step": 3481
    },
    {
      "epoch": 0.1916451098024107,
      "grad_norm": 1.9752860069274902,
      "learning_rate": 9.786624214443893e-06,
      "loss": 0.7611,
      "step": 3482
    },
    {
      "epoch": 0.19170014860476636,
      "grad_norm": 0.8023802042007446,
      "learning_rate": 9.786498918495963e-06,
      "loss": 0.8426,
      "step": 3483
    },
    {
      "epoch": 0.19175518740712202,
      "grad_norm": 0.7232086658477783,
      "learning_rate": 9.78637358657399e-06,
      "loss": 0.6611,
      "step": 3484
    },
    {
      "epoch": 0.19181022620947769,
      "grad_norm": 0.8198665380477905,
      "learning_rate": 9.786248218678912e-06,
      "loss": 0.8795,
      "step": 3485
    },
    {
      "epoch": 0.19186526501183335,
      "grad_norm": 0.942404568195343,
      "learning_rate": 9.786122814811675e-06,
      "loss": 0.9146,
      "step": 3486
    },
    {
      "epoch": 0.191920303814189,
      "grad_norm": 0.7602691054344177,
      "learning_rate": 9.78599737497322e-06,
      "loss": 0.7514,
      "step": 3487
    },
    {
      "epoch": 0.19197534261654467,
      "grad_norm": 0.7981933951377869,
      "learning_rate": 9.785871899164489e-06,
      "loss": 0.7722,
      "step": 3488
    },
    {
      "epoch": 0.19203038141890033,
      "grad_norm": 0.8617631793022156,
      "learning_rate": 9.785746387386427e-06,
      "loss": 0.8989,
      "step": 3489
    },
    {
      "epoch": 0.192085420221256,
      "grad_norm": 0.7691803574562073,
      "learning_rate": 9.785620839639976e-06,
      "loss": 0.7929,
      "step": 3490
    },
    {
      "epoch": 0.19214045902361165,
      "grad_norm": 1.3053189516067505,
      "learning_rate": 9.785495255926078e-06,
      "loss": 0.8478,
      "step": 3491
    },
    {
      "epoch": 0.19219549782596732,
      "grad_norm": 0.807064950466156,
      "learning_rate": 9.785369636245681e-06,
      "loss": 0.7452,
      "step": 3492
    },
    {
      "epoch": 0.19225053662832298,
      "grad_norm": 0.8182778358459473,
      "learning_rate": 9.785243980599726e-06,
      "loss": 0.8371,
      "step": 3493
    },
    {
      "epoch": 0.19230557543067864,
      "grad_norm": 0.7654449343681335,
      "learning_rate": 9.785118288989157e-06,
      "loss": 0.8321,
      "step": 3494
    },
    {
      "epoch": 0.1923606142330343,
      "grad_norm": 0.7192448973655701,
      "learning_rate": 9.784992561414922e-06,
      "loss": 0.7451,
      "step": 3495
    },
    {
      "epoch": 0.19241565303538996,
      "grad_norm": 0.8639407753944397,
      "learning_rate": 9.784866797877964e-06,
      "loss": 0.9272,
      "step": 3496
    },
    {
      "epoch": 0.19247069183774562,
      "grad_norm": 0.8329927921295166,
      "learning_rate": 9.784740998379225e-06,
      "loss": 0.8034,
      "step": 3497
    },
    {
      "epoch": 0.19252573064010128,
      "grad_norm": 0.7975476980209351,
      "learning_rate": 9.784615162919656e-06,
      "loss": 0.6885,
      "step": 3498
    },
    {
      "epoch": 0.19258076944245694,
      "grad_norm": 0.8077559471130371,
      "learning_rate": 9.7844892915002e-06,
      "loss": 0.8745,
      "step": 3499
    },
    {
      "epoch": 0.1926358082448126,
      "grad_norm": 0.7957825660705566,
      "learning_rate": 9.7843633841218e-06,
      "loss": 0.7612,
      "step": 3500
    },
    {
      "epoch": 0.19269084704716824,
      "grad_norm": 0.8478250503540039,
      "learning_rate": 9.784237440785408e-06,
      "loss": 0.8675,
      "step": 3501
    },
    {
      "epoch": 0.1927458858495239,
      "grad_norm": 0.7289726138114929,
      "learning_rate": 9.78411146149197e-06,
      "loss": 0.7126,
      "step": 3502
    },
    {
      "epoch": 0.19280092465187956,
      "grad_norm": 0.7608509063720703,
      "learning_rate": 9.783985446242427e-06,
      "loss": 0.7049,
      "step": 3503
    },
    {
      "epoch": 0.19285596345423522,
      "grad_norm": 0.8985201120376587,
      "learning_rate": 9.783859395037733e-06,
      "loss": 0.8067,
      "step": 3504
    },
    {
      "epoch": 0.19291100225659089,
      "grad_norm": 0.7563273906707764,
      "learning_rate": 9.78373330787883e-06,
      "loss": 0.7018,
      "step": 3505
    },
    {
      "epoch": 0.19296604105894655,
      "grad_norm": 0.8022900223731995,
      "learning_rate": 9.78360718476667e-06,
      "loss": 0.8346,
      "step": 3506
    },
    {
      "epoch": 0.1930210798613022,
      "grad_norm": 0.897566020488739,
      "learning_rate": 9.783481025702197e-06,
      "loss": 0.9465,
      "step": 3507
    },
    {
      "epoch": 0.19307611866365787,
      "grad_norm": 0.9550303220748901,
      "learning_rate": 9.783354830686363e-06,
      "loss": 0.8904,
      "step": 3508
    },
    {
      "epoch": 0.19313115746601353,
      "grad_norm": 0.8152582049369812,
      "learning_rate": 9.783228599720114e-06,
      "loss": 0.7776,
      "step": 3509
    },
    {
      "epoch": 0.1931861962683692,
      "grad_norm": 0.7421940565109253,
      "learning_rate": 9.783102332804398e-06,
      "loss": 0.6847,
      "step": 3510
    },
    {
      "epoch": 0.19324123507072485,
      "grad_norm": 0.7414368391036987,
      "learning_rate": 9.782976029940167e-06,
      "loss": 0.8435,
      "step": 3511
    },
    {
      "epoch": 0.19329627387308052,
      "grad_norm": 0.7845529317855835,
      "learning_rate": 9.782849691128366e-06,
      "loss": 0.8255,
      "step": 3512
    },
    {
      "epoch": 0.19335131267543618,
      "grad_norm": 0.7779788970947266,
      "learning_rate": 9.78272331636995e-06,
      "loss": 0.7801,
      "step": 3513
    },
    {
      "epoch": 0.19340635147779184,
      "grad_norm": 0.7537885904312134,
      "learning_rate": 9.782596905665865e-06,
      "loss": 0.7501,
      "step": 3514
    },
    {
      "epoch": 0.1934613902801475,
      "grad_norm": 0.7585812211036682,
      "learning_rate": 9.782470459017059e-06,
      "loss": 0.8425,
      "step": 3515
    },
    {
      "epoch": 0.19351642908250316,
      "grad_norm": 0.7923589944839478,
      "learning_rate": 9.78234397642449e-06,
      "loss": 0.8412,
      "step": 3516
    },
    {
      "epoch": 0.19357146788485882,
      "grad_norm": 0.8710628151893616,
      "learning_rate": 9.7822174578891e-06,
      "loss": 0.8014,
      "step": 3517
    },
    {
      "epoch": 0.19362650668721448,
      "grad_norm": 0.7646920084953308,
      "learning_rate": 9.782090903411845e-06,
      "loss": 0.8256,
      "step": 3518
    },
    {
      "epoch": 0.19368154548957014,
      "grad_norm": 0.7560480833053589,
      "learning_rate": 9.781964312993675e-06,
      "loss": 0.7816,
      "step": 3519
    },
    {
      "epoch": 0.1937365842919258,
      "grad_norm": 0.7438123226165771,
      "learning_rate": 9.78183768663554e-06,
      "loss": 0.8319,
      "step": 3520
    },
    {
      "epoch": 0.19379162309428147,
      "grad_norm": 0.7239874601364136,
      "learning_rate": 9.781711024338394e-06,
      "loss": 0.6968,
      "step": 3521
    },
    {
      "epoch": 0.19384666189663713,
      "grad_norm": 0.881197988986969,
      "learning_rate": 9.781584326103188e-06,
      "loss": 0.9493,
      "step": 3522
    },
    {
      "epoch": 0.1939017006989928,
      "grad_norm": 0.7903854846954346,
      "learning_rate": 9.781457591930874e-06,
      "loss": 0.8312,
      "step": 3523
    },
    {
      "epoch": 0.19395673950134845,
      "grad_norm": 0.7375456094741821,
      "learning_rate": 9.781330821822405e-06,
      "loss": 0.7434,
      "step": 3524
    },
    {
      "epoch": 0.1940117783037041,
      "grad_norm": 0.7101724743843079,
      "learning_rate": 9.781204015778733e-06,
      "loss": 0.75,
      "step": 3525
    },
    {
      "epoch": 0.19406681710605977,
      "grad_norm": 0.8267471194267273,
      "learning_rate": 9.781077173800812e-06,
      "loss": 0.8807,
      "step": 3526
    },
    {
      "epoch": 0.19412185590841544,
      "grad_norm": 0.9014178514480591,
      "learning_rate": 9.780950295889594e-06,
      "loss": 0.7836,
      "step": 3527
    },
    {
      "epoch": 0.1941768947107711,
      "grad_norm": 0.7579739689826965,
      "learning_rate": 9.780823382046034e-06,
      "loss": 0.8331,
      "step": 3528
    },
    {
      "epoch": 0.19423193351312676,
      "grad_norm": 0.8308925032615662,
      "learning_rate": 9.780696432271084e-06,
      "loss": 0.794,
      "step": 3529
    },
    {
      "epoch": 0.19428697231548242,
      "grad_norm": 0.7461574673652649,
      "learning_rate": 9.780569446565701e-06,
      "loss": 0.8155,
      "step": 3530
    },
    {
      "epoch": 0.19434201111783808,
      "grad_norm": 0.8658885359764099,
      "learning_rate": 9.780442424930836e-06,
      "loss": 0.7907,
      "step": 3531
    },
    {
      "epoch": 0.19439704992019374,
      "grad_norm": 0.7243279218673706,
      "learning_rate": 9.780315367367449e-06,
      "loss": 0.7985,
      "step": 3532
    },
    {
      "epoch": 0.1944520887225494,
      "grad_norm": 0.8482224345207214,
      "learning_rate": 9.780188273876486e-06,
      "loss": 0.9095,
      "step": 3533
    },
    {
      "epoch": 0.19450712752490507,
      "grad_norm": 0.8675364255905151,
      "learning_rate": 9.78006114445891e-06,
      "loss": 0.759,
      "step": 3534
    },
    {
      "epoch": 0.19456216632726073,
      "grad_norm": 0.8388474583625793,
      "learning_rate": 9.779933979115675e-06,
      "loss": 0.8331,
      "step": 3535
    },
    {
      "epoch": 0.1946172051296164,
      "grad_norm": 0.8050872683525085,
      "learning_rate": 9.779806777847735e-06,
      "loss": 0.861,
      "step": 3536
    },
    {
      "epoch": 0.19467224393197205,
      "grad_norm": 0.8401390910148621,
      "learning_rate": 9.779679540656046e-06,
      "loss": 0.755,
      "step": 3537
    },
    {
      "epoch": 0.1947272827343277,
      "grad_norm": 0.865160346031189,
      "learning_rate": 9.779552267541566e-06,
      "loss": 0.7515,
      "step": 3538
    },
    {
      "epoch": 0.19478232153668337,
      "grad_norm": 0.923086941242218,
      "learning_rate": 9.77942495850525e-06,
      "loss": 0.8032,
      "step": 3539
    },
    {
      "epoch": 0.19483736033903903,
      "grad_norm": 0.8402467966079712,
      "learning_rate": 9.779297613548056e-06,
      "loss": 0.9198,
      "step": 3540
    },
    {
      "epoch": 0.1948923991413947,
      "grad_norm": 0.7875306606292725,
      "learning_rate": 9.779170232670939e-06,
      "loss": 0.712,
      "step": 3541
    },
    {
      "epoch": 0.19494743794375036,
      "grad_norm": 0.7996379137039185,
      "learning_rate": 9.779042815874858e-06,
      "loss": 0.8126,
      "step": 3542
    },
    {
      "epoch": 0.19500247674610602,
      "grad_norm": 0.7644525766372681,
      "learning_rate": 9.778915363160773e-06,
      "loss": 0.8602,
      "step": 3543
    },
    {
      "epoch": 0.19505751554846165,
      "grad_norm": 0.8068630695343018,
      "learning_rate": 9.778787874529635e-06,
      "loss": 0.736,
      "step": 3544
    },
    {
      "epoch": 0.1951125543508173,
      "grad_norm": 0.7889519929885864,
      "learning_rate": 9.77866034998241e-06,
      "loss": 0.755,
      "step": 3545
    },
    {
      "epoch": 0.19516759315317297,
      "grad_norm": 0.7895978689193726,
      "learning_rate": 9.778532789520053e-06,
      "loss": 0.8213,
      "step": 3546
    },
    {
      "epoch": 0.19522263195552864,
      "grad_norm": 0.8571796417236328,
      "learning_rate": 9.77840519314352e-06,
      "loss": 0.8193,
      "step": 3547
    },
    {
      "epoch": 0.1952776707578843,
      "grad_norm": 0.6880007982254028,
      "learning_rate": 9.778277560853775e-06,
      "loss": 0.6354,
      "step": 3548
    },
    {
      "epoch": 0.19533270956023996,
      "grad_norm": 0.8155353665351868,
      "learning_rate": 9.778149892651775e-06,
      "loss": 0.8518,
      "step": 3549
    },
    {
      "epoch": 0.19538774836259562,
      "grad_norm": 0.851021945476532,
      "learning_rate": 9.778022188538479e-06,
      "loss": 0.8506,
      "step": 3550
    },
    {
      "epoch": 0.19544278716495128,
      "grad_norm": 0.8910510540008545,
      "learning_rate": 9.777894448514847e-06,
      "loss": 0.8825,
      "step": 3551
    },
    {
      "epoch": 0.19549782596730694,
      "grad_norm": 0.8156018853187561,
      "learning_rate": 9.777766672581838e-06,
      "loss": 0.8262,
      "step": 3552
    },
    {
      "epoch": 0.1955528647696626,
      "grad_norm": 0.756340503692627,
      "learning_rate": 9.777638860740415e-06,
      "loss": 0.7094,
      "step": 3553
    },
    {
      "epoch": 0.19560790357201827,
      "grad_norm": 0.7604243159294128,
      "learning_rate": 9.777511012991538e-06,
      "loss": 0.8089,
      "step": 3554
    },
    {
      "epoch": 0.19566294237437393,
      "grad_norm": 0.7609277963638306,
      "learning_rate": 9.777383129336167e-06,
      "loss": 0.7853,
      "step": 3555
    },
    {
      "epoch": 0.1957179811767296,
      "grad_norm": 1.3562177419662476,
      "learning_rate": 9.77725520977526e-06,
      "loss": 0.7051,
      "step": 3556
    },
    {
      "epoch": 0.19577301997908525,
      "grad_norm": 0.7428582310676575,
      "learning_rate": 9.777127254309784e-06,
      "loss": 0.734,
      "step": 3557
    },
    {
      "epoch": 0.1958280587814409,
      "grad_norm": 0.6941032409667969,
      "learning_rate": 9.776999262940698e-06,
      "loss": 0.7862,
      "step": 3558
    },
    {
      "epoch": 0.19588309758379657,
      "grad_norm": 0.8249906301498413,
      "learning_rate": 9.776871235668966e-06,
      "loss": 0.8324,
      "step": 3559
    },
    {
      "epoch": 0.19593813638615223,
      "grad_norm": 0.6778795719146729,
      "learning_rate": 9.776743172495546e-06,
      "loss": 0.743,
      "step": 3560
    },
    {
      "epoch": 0.1959931751885079,
      "grad_norm": 0.8454411625862122,
      "learning_rate": 9.776615073421405e-06,
      "loss": 0.8625,
      "step": 3561
    },
    {
      "epoch": 0.19604821399086356,
      "grad_norm": 0.8303809762001038,
      "learning_rate": 9.776486938447503e-06,
      "loss": 0.8806,
      "step": 3562
    },
    {
      "epoch": 0.19610325279321922,
      "grad_norm": 0.8814080357551575,
      "learning_rate": 9.776358767574803e-06,
      "loss": 0.9096,
      "step": 3563
    },
    {
      "epoch": 0.19615829159557488,
      "grad_norm": 0.7860022187232971,
      "learning_rate": 9.77623056080427e-06,
      "loss": 0.8101,
      "step": 3564
    },
    {
      "epoch": 0.19621333039793054,
      "grad_norm": 0.7604898810386658,
      "learning_rate": 9.776102318136866e-06,
      "loss": 0.8121,
      "step": 3565
    },
    {
      "epoch": 0.1962683692002862,
      "grad_norm": 0.810708224773407,
      "learning_rate": 9.775974039573555e-06,
      "loss": 0.8334,
      "step": 3566
    },
    {
      "epoch": 0.19632340800264186,
      "grad_norm": 1.0174707174301147,
      "learning_rate": 9.775845725115301e-06,
      "loss": 0.8147,
      "step": 3567
    },
    {
      "epoch": 0.19637844680499753,
      "grad_norm": 0.825137734413147,
      "learning_rate": 9.77571737476307e-06,
      "loss": 0.816,
      "step": 3568
    },
    {
      "epoch": 0.1964334856073532,
      "grad_norm": 0.9023691415786743,
      "learning_rate": 9.775588988517826e-06,
      "loss": 0.9157,
      "step": 3569
    },
    {
      "epoch": 0.19648852440970885,
      "grad_norm": 0.7287655472755432,
      "learning_rate": 9.775460566380534e-06,
      "loss": 0.7414,
      "step": 3570
    },
    {
      "epoch": 0.1965435632120645,
      "grad_norm": 0.8675361275672913,
      "learning_rate": 9.775332108352158e-06,
      "loss": 0.7212,
      "step": 3571
    },
    {
      "epoch": 0.19659860201442017,
      "grad_norm": 0.8633139729499817,
      "learning_rate": 9.775203614433664e-06,
      "loss": 0.7254,
      "step": 3572
    },
    {
      "epoch": 0.19665364081677583,
      "grad_norm": 0.8628275394439697,
      "learning_rate": 9.775075084626017e-06,
      "loss": 0.7403,
      "step": 3573
    },
    {
      "epoch": 0.1967086796191315,
      "grad_norm": 0.86918044090271,
      "learning_rate": 9.774946518930184e-06,
      "loss": 0.8208,
      "step": 3574
    },
    {
      "epoch": 0.19676371842148715,
      "grad_norm": 1.3616218566894531,
      "learning_rate": 9.774817917347132e-06,
      "loss": 0.7432,
      "step": 3575
    },
    {
      "epoch": 0.19681875722384282,
      "grad_norm": 0.929084062576294,
      "learning_rate": 9.774689279877827e-06,
      "loss": 0.9567,
      "step": 3576
    },
    {
      "epoch": 0.19687379602619848,
      "grad_norm": 0.7732542753219604,
      "learning_rate": 9.774560606523234e-06,
      "loss": 0.8682,
      "step": 3577
    },
    {
      "epoch": 0.19692883482855414,
      "grad_norm": 0.7933471202850342,
      "learning_rate": 9.774431897284323e-06,
      "loss": 0.7112,
      "step": 3578
    },
    {
      "epoch": 0.1969838736309098,
      "grad_norm": 0.8229583501815796,
      "learning_rate": 9.77430315216206e-06,
      "loss": 0.762,
      "step": 3579
    },
    {
      "epoch": 0.19703891243326546,
      "grad_norm": 0.7571341395378113,
      "learning_rate": 9.774174371157412e-06,
      "loss": 0.7627,
      "step": 3580
    },
    {
      "epoch": 0.19709395123562112,
      "grad_norm": 1.1551839113235474,
      "learning_rate": 9.774045554271347e-06,
      "loss": 0.8621,
      "step": 3581
    },
    {
      "epoch": 0.19714899003797678,
      "grad_norm": 0.8546237349510193,
      "learning_rate": 9.773916701504833e-06,
      "loss": 0.8183,
      "step": 3582
    },
    {
      "epoch": 0.19720402884033245,
      "grad_norm": 0.7297555804252625,
      "learning_rate": 9.773787812858841e-06,
      "loss": 0.8098,
      "step": 3583
    },
    {
      "epoch": 0.1972590676426881,
      "grad_norm": 0.7846053838729858,
      "learning_rate": 9.773658888334336e-06,
      "loss": 0.7874,
      "step": 3584
    },
    {
      "epoch": 0.19731410644504377,
      "grad_norm": 0.8949562907218933,
      "learning_rate": 9.773529927932288e-06,
      "loss": 0.8651,
      "step": 3585
    },
    {
      "epoch": 0.19736914524739943,
      "grad_norm": 0.8041829466819763,
      "learning_rate": 9.773400931653668e-06,
      "loss": 0.7519,
      "step": 3586
    },
    {
      "epoch": 0.19742418404975506,
      "grad_norm": 0.8090983033180237,
      "learning_rate": 9.773271899499444e-06,
      "loss": 0.8606,
      "step": 3587
    },
    {
      "epoch": 0.19747922285211073,
      "grad_norm": 0.7954100966453552,
      "learning_rate": 9.773142831470587e-06,
      "loss": 0.9028,
      "step": 3588
    },
    {
      "epoch": 0.1975342616544664,
      "grad_norm": 0.6865562796592712,
      "learning_rate": 9.773013727568066e-06,
      "loss": 0.7323,
      "step": 3589
    },
    {
      "epoch": 0.19758930045682205,
      "grad_norm": 0.9144858717918396,
      "learning_rate": 9.772884587792851e-06,
      "loss": 0.8178,
      "step": 3590
    },
    {
      "epoch": 0.1976443392591777,
      "grad_norm": 0.8096563220024109,
      "learning_rate": 9.772755412145913e-06,
      "loss": 0.7749,
      "step": 3591
    },
    {
      "epoch": 0.19769937806153337,
      "grad_norm": 1.4496957063674927,
      "learning_rate": 9.772626200628222e-06,
      "loss": 0.7981,
      "step": 3592
    },
    {
      "epoch": 0.19775441686388903,
      "grad_norm": 0.7699438333511353,
      "learning_rate": 9.77249695324075e-06,
      "loss": 0.7683,
      "step": 3593
    },
    {
      "epoch": 0.1978094556662447,
      "grad_norm": 0.7883017063140869,
      "learning_rate": 9.77236766998447e-06,
      "loss": 0.7668,
      "step": 3594
    },
    {
      "epoch": 0.19786449446860035,
      "grad_norm": 0.7552568912506104,
      "learning_rate": 9.772238350860352e-06,
      "loss": 0.7914,
      "step": 3595
    },
    {
      "epoch": 0.19791953327095602,
      "grad_norm": 0.8585009574890137,
      "learning_rate": 9.772108995869366e-06,
      "loss": 0.9888,
      "step": 3596
    },
    {
      "epoch": 0.19797457207331168,
      "grad_norm": 0.9459839463233948,
      "learning_rate": 9.77197960501249e-06,
      "loss": 0.9923,
      "step": 3597
    },
    {
      "epoch": 0.19802961087566734,
      "grad_norm": 0.844771683216095,
      "learning_rate": 9.77185017829069e-06,
      "loss": 0.8427,
      "step": 3598
    },
    {
      "epoch": 0.198084649678023,
      "grad_norm": 0.749700665473938,
      "learning_rate": 9.77172071570494e-06,
      "loss": 0.8111,
      "step": 3599
    },
    {
      "epoch": 0.19813968848037866,
      "grad_norm": 0.7297450304031372,
      "learning_rate": 9.771591217256216e-06,
      "loss": 0.7783,
      "step": 3600
    },
    {
      "epoch": 0.19819472728273432,
      "grad_norm": 0.7928450703620911,
      "learning_rate": 9.77146168294549e-06,
      "loss": 0.8755,
      "step": 3601
    },
    {
      "epoch": 0.19824976608508998,
      "grad_norm": 0.7236143946647644,
      "learning_rate": 9.771332112773734e-06,
      "loss": 0.7159,
      "step": 3602
    },
    {
      "epoch": 0.19830480488744565,
      "grad_norm": 0.8170965313911438,
      "learning_rate": 9.771202506741926e-06,
      "loss": 0.9093,
      "step": 3603
    },
    {
      "epoch": 0.1983598436898013,
      "grad_norm": 0.8834578990936279,
      "learning_rate": 9.771072864851035e-06,
      "loss": 0.8961,
      "step": 3604
    },
    {
      "epoch": 0.19841488249215697,
      "grad_norm": 1.3750289678573608,
      "learning_rate": 9.770943187102037e-06,
      "loss": 0.8175,
      "step": 3605
    },
    {
      "epoch": 0.19846992129451263,
      "grad_norm": 0.7016286253929138,
      "learning_rate": 9.770813473495909e-06,
      "loss": 0.7171,
      "step": 3606
    },
    {
      "epoch": 0.1985249600968683,
      "grad_norm": 0.7792307734489441,
      "learning_rate": 9.770683724033622e-06,
      "loss": 0.6892,
      "step": 3607
    },
    {
      "epoch": 0.19857999889922395,
      "grad_norm": 0.789820671081543,
      "learning_rate": 9.770553938716153e-06,
      "loss": 0.8531,
      "step": 3608
    },
    {
      "epoch": 0.19863503770157961,
      "grad_norm": 0.7585997581481934,
      "learning_rate": 9.77042411754448e-06,
      "loss": 0.8195,
      "step": 3609
    },
    {
      "epoch": 0.19869007650393528,
      "grad_norm": 0.8989273905754089,
      "learning_rate": 9.770294260519573e-06,
      "loss": 0.891,
      "step": 3610
    },
    {
      "epoch": 0.19874511530629094,
      "grad_norm": 0.8044012188911438,
      "learning_rate": 9.770164367642414e-06,
      "loss": 0.8428,
      "step": 3611
    },
    {
      "epoch": 0.1988001541086466,
      "grad_norm": 0.7847021222114563,
      "learning_rate": 9.770034438913975e-06,
      "loss": 0.8302,
      "step": 3612
    },
    {
      "epoch": 0.19885519291100226,
      "grad_norm": 0.9260531663894653,
      "learning_rate": 9.769904474335234e-06,
      "loss": 0.8187,
      "step": 3613
    },
    {
      "epoch": 0.19891023171335792,
      "grad_norm": 0.7491805553436279,
      "learning_rate": 9.769774473907168e-06,
      "loss": 0.8374,
      "step": 3614
    },
    {
      "epoch": 0.19896527051571358,
      "grad_norm": 1.1665992736816406,
      "learning_rate": 9.769644437630754e-06,
      "loss": 0.8154,
      "step": 3615
    },
    {
      "epoch": 0.19902030931806924,
      "grad_norm": 0.9162279963493347,
      "learning_rate": 9.769514365506968e-06,
      "loss": 0.8883,
      "step": 3616
    },
    {
      "epoch": 0.1990753481204249,
      "grad_norm": 0.8980437517166138,
      "learning_rate": 9.769384257536791e-06,
      "loss": 0.8948,
      "step": 3617
    },
    {
      "epoch": 0.19913038692278057,
      "grad_norm": 0.7544137835502625,
      "learning_rate": 9.769254113721197e-06,
      "loss": 0.7763,
      "step": 3618
    },
    {
      "epoch": 0.19918542572513623,
      "grad_norm": 0.8393334746360779,
      "learning_rate": 9.769123934061168e-06,
      "loss": 0.8361,
      "step": 3619
    },
    {
      "epoch": 0.1992404645274919,
      "grad_norm": 0.8184031248092651,
      "learning_rate": 9.768993718557678e-06,
      "loss": 0.8104,
      "step": 3620
    },
    {
      "epoch": 0.19929550332984755,
      "grad_norm": 0.8023706674575806,
      "learning_rate": 9.76886346721171e-06,
      "loss": 0.7824,
      "step": 3621
    },
    {
      "epoch": 0.1993505421322032,
      "grad_norm": 0.9354264736175537,
      "learning_rate": 9.768733180024238e-06,
      "loss": 0.7782,
      "step": 3622
    },
    {
      "epoch": 0.19940558093455887,
      "grad_norm": 0.7037177681922913,
      "learning_rate": 9.768602856996244e-06,
      "loss": 0.8054,
      "step": 3623
    },
    {
      "epoch": 0.19946061973691454,
      "grad_norm": 0.7926928997039795,
      "learning_rate": 9.768472498128709e-06,
      "loss": 0.8864,
      "step": 3624
    },
    {
      "epoch": 0.1995156585392702,
      "grad_norm": 0.7963769435882568,
      "learning_rate": 9.76834210342261e-06,
      "loss": 0.8505,
      "step": 3625
    },
    {
      "epoch": 0.19957069734162586,
      "grad_norm": 0.8553926944732666,
      "learning_rate": 9.768211672878929e-06,
      "loss": 0.8519,
      "step": 3626
    },
    {
      "epoch": 0.19962573614398152,
      "grad_norm": 0.8147156834602356,
      "learning_rate": 9.768081206498644e-06,
      "loss": 0.8091,
      "step": 3627
    },
    {
      "epoch": 0.19968077494633718,
      "grad_norm": 0.8226443529129028,
      "learning_rate": 9.767950704282739e-06,
      "loss": 0.8561,
      "step": 3628
    },
    {
      "epoch": 0.19973581374869284,
      "grad_norm": 0.7246909141540527,
      "learning_rate": 9.76782016623219e-06,
      "loss": 0.7318,
      "step": 3629
    },
    {
      "epoch": 0.19979085255104848,
      "grad_norm": 1.0527293682098389,
      "learning_rate": 9.767689592347983e-06,
      "loss": 0.7699,
      "step": 3630
    },
    {
      "epoch": 0.19984589135340414,
      "grad_norm": 0.7433847188949585,
      "learning_rate": 9.767558982631097e-06,
      "loss": 0.8619,
      "step": 3631
    },
    {
      "epoch": 0.1999009301557598,
      "grad_norm": 0.7901468873023987,
      "learning_rate": 9.767428337082513e-06,
      "loss": 0.8365,
      "step": 3632
    },
    {
      "epoch": 0.19995596895811546,
      "grad_norm": 0.7766845226287842,
      "learning_rate": 9.767297655703215e-06,
      "loss": 0.7767,
      "step": 3633
    },
    {
      "epoch": 0.20001100776047112,
      "grad_norm": 0.7785109281539917,
      "learning_rate": 9.767166938494183e-06,
      "loss": 0.7114,
      "step": 3634
    },
    {
      "epoch": 0.20006604656282678,
      "grad_norm": 0.8068187832832336,
      "learning_rate": 9.767036185456402e-06,
      "loss": 0.8142,
      "step": 3635
    },
    {
      "epoch": 0.20012108536518244,
      "grad_norm": 0.7893292307853699,
      "learning_rate": 9.766905396590851e-06,
      "loss": 0.8658,
      "step": 3636
    },
    {
      "epoch": 0.2001761241675381,
      "grad_norm": 0.8647506237030029,
      "learning_rate": 9.766774571898516e-06,
      "loss": 0.84,
      "step": 3637
    },
    {
      "epoch": 0.20023116296989377,
      "grad_norm": 0.8545078635215759,
      "learning_rate": 9.766643711380378e-06,
      "loss": 0.8455,
      "step": 3638
    },
    {
      "epoch": 0.20028620177224943,
      "grad_norm": 0.924404501914978,
      "learning_rate": 9.766512815037424e-06,
      "loss": 0.6954,
      "step": 3639
    },
    {
      "epoch": 0.2003412405746051,
      "grad_norm": 0.8077614903450012,
      "learning_rate": 9.766381882870635e-06,
      "loss": 0.7724,
      "step": 3640
    },
    {
      "epoch": 0.20039627937696075,
      "grad_norm": 0.8886739015579224,
      "learning_rate": 9.766250914880994e-06,
      "loss": 0.8318,
      "step": 3641
    },
    {
      "epoch": 0.2004513181793164,
      "grad_norm": 0.8086267113685608,
      "learning_rate": 9.76611991106949e-06,
      "loss": 0.8494,
      "step": 3642
    },
    {
      "epoch": 0.20050635698167207,
      "grad_norm": 0.8606873750686646,
      "learning_rate": 9.765988871437101e-06,
      "loss": 0.8488,
      "step": 3643
    },
    {
      "epoch": 0.20056139578402774,
      "grad_norm": 0.6966355443000793,
      "learning_rate": 9.76585779598482e-06,
      "loss": 0.7361,
      "step": 3644
    },
    {
      "epoch": 0.2006164345863834,
      "grad_norm": 0.8474385738372803,
      "learning_rate": 9.765726684713623e-06,
      "loss": 0.8354,
      "step": 3645
    },
    {
      "epoch": 0.20067147338873906,
      "grad_norm": 0.7609736919403076,
      "learning_rate": 9.765595537624502e-06,
      "loss": 0.7297,
      "step": 3646
    },
    {
      "epoch": 0.20072651219109472,
      "grad_norm": 1.08648681640625,
      "learning_rate": 9.76546435471844e-06,
      "loss": 0.7534,
      "step": 3647
    },
    {
      "epoch": 0.20078155099345038,
      "grad_norm": 0.7437332272529602,
      "learning_rate": 9.765333135996425e-06,
      "loss": 0.8532,
      "step": 3648
    },
    {
      "epoch": 0.20083658979580604,
      "grad_norm": 0.9016552567481995,
      "learning_rate": 9.76520188145944e-06,
      "loss": 0.7968,
      "step": 3649
    },
    {
      "epoch": 0.2008916285981617,
      "grad_norm": 0.8916428089141846,
      "learning_rate": 9.765070591108473e-06,
      "loss": 0.9601,
      "step": 3650
    },
    {
      "epoch": 0.20094666740051736,
      "grad_norm": 0.7679058313369751,
      "learning_rate": 9.764939264944512e-06,
      "loss": 0.816,
      "step": 3651
    },
    {
      "epoch": 0.20100170620287303,
      "grad_norm": 0.7716549634933472,
      "learning_rate": 9.764807902968543e-06,
      "loss": 0.876,
      "step": 3652
    },
    {
      "epoch": 0.2010567450052287,
      "grad_norm": 0.8288074731826782,
      "learning_rate": 9.764676505181554e-06,
      "loss": 0.8054,
      "step": 3653
    },
    {
      "epoch": 0.20111178380758435,
      "grad_norm": 0.7906842827796936,
      "learning_rate": 9.76454507158453e-06,
      "loss": 0.8026,
      "step": 3654
    },
    {
      "epoch": 0.20116682260994,
      "grad_norm": 0.8093311190605164,
      "learning_rate": 9.764413602178461e-06,
      "loss": 0.8093,
      "step": 3655
    },
    {
      "epoch": 0.20122186141229567,
      "grad_norm": 0.7234730124473572,
      "learning_rate": 9.764282096964335e-06,
      "loss": 0.7194,
      "step": 3656
    },
    {
      "epoch": 0.20127690021465133,
      "grad_norm": 0.9048555493354797,
      "learning_rate": 9.76415055594314e-06,
      "loss": 0.8996,
      "step": 3657
    },
    {
      "epoch": 0.201331939017007,
      "grad_norm": 0.7630691528320312,
      "learning_rate": 9.764018979115864e-06,
      "loss": 0.7876,
      "step": 3658
    },
    {
      "epoch": 0.20138697781936266,
      "grad_norm": 0.9551032781600952,
      "learning_rate": 9.763887366483498e-06,
      "loss": 0.8249,
      "step": 3659
    },
    {
      "epoch": 0.20144201662171832,
      "grad_norm": 0.6988314986228943,
      "learning_rate": 9.76375571804703e-06,
      "loss": 0.8011,
      "step": 3660
    },
    {
      "epoch": 0.20149705542407398,
      "grad_norm": 0.7790704369544983,
      "learning_rate": 9.763624033807448e-06,
      "loss": 0.8287,
      "step": 3661
    },
    {
      "epoch": 0.20155209422642964,
      "grad_norm": 0.7201293706893921,
      "learning_rate": 9.763492313765743e-06,
      "loss": 0.7854,
      "step": 3662
    },
    {
      "epoch": 0.2016071330287853,
      "grad_norm": 0.8691730499267578,
      "learning_rate": 9.763360557922905e-06,
      "loss": 0.8348,
      "step": 3663
    },
    {
      "epoch": 0.20166217183114096,
      "grad_norm": 0.7660881876945496,
      "learning_rate": 9.763228766279924e-06,
      "loss": 0.7686,
      "step": 3664
    },
    {
      "epoch": 0.20171721063349662,
      "grad_norm": 1.083796501159668,
      "learning_rate": 9.76309693883779e-06,
      "loss": 0.8848,
      "step": 3665
    },
    {
      "epoch": 0.20177224943585229,
      "grad_norm": 0.7892678380012512,
      "learning_rate": 9.762965075597496e-06,
      "loss": 0.7804,
      "step": 3666
    },
    {
      "epoch": 0.20182728823820795,
      "grad_norm": 0.7166122198104858,
      "learning_rate": 9.762833176560031e-06,
      "loss": 0.761,
      "step": 3667
    },
    {
      "epoch": 0.2018823270405636,
      "grad_norm": 0.8187084794044495,
      "learning_rate": 9.762701241726386e-06,
      "loss": 0.8251,
      "step": 3668
    },
    {
      "epoch": 0.20193736584291927,
      "grad_norm": 0.6930577158927917,
      "learning_rate": 9.762569271097556e-06,
      "loss": 0.6795,
      "step": 3669
    },
    {
      "epoch": 0.20199240464527493,
      "grad_norm": 0.8085465431213379,
      "learning_rate": 9.762437264674527e-06,
      "loss": 0.8415,
      "step": 3670
    },
    {
      "epoch": 0.2020474434476306,
      "grad_norm": 0.8111084699630737,
      "learning_rate": 9.762305222458294e-06,
      "loss": 0.792,
      "step": 3671
    },
    {
      "epoch": 0.20210248224998625,
      "grad_norm": 0.8200401067733765,
      "learning_rate": 9.762173144449852e-06,
      "loss": 0.8224,
      "step": 3672
    },
    {
      "epoch": 0.2021575210523419,
      "grad_norm": 0.8460109233856201,
      "learning_rate": 9.762041030650192e-06,
      "loss": 0.9025,
      "step": 3673
    },
    {
      "epoch": 0.20221255985469755,
      "grad_norm": 0.8152671456336975,
      "learning_rate": 9.761908881060303e-06,
      "loss": 0.9002,
      "step": 3674
    },
    {
      "epoch": 0.2022675986570532,
      "grad_norm": 0.8204773664474487,
      "learning_rate": 9.761776695681185e-06,
      "loss": 0.8324,
      "step": 3675
    },
    {
      "epoch": 0.20232263745940887,
      "grad_norm": 0.8121044039726257,
      "learning_rate": 9.761644474513825e-06,
      "loss": 0.855,
      "step": 3676
    },
    {
      "epoch": 0.20237767626176453,
      "grad_norm": 0.79920494556427,
      "learning_rate": 9.76151221755922e-06,
      "loss": 0.7837,
      "step": 3677
    },
    {
      "epoch": 0.2024327150641202,
      "grad_norm": 0.862808346748352,
      "learning_rate": 9.761379924818367e-06,
      "loss": 0.8714,
      "step": 3678
    },
    {
      "epoch": 0.20248775386647586,
      "grad_norm": 0.7135004997253418,
      "learning_rate": 9.761247596292254e-06,
      "loss": 0.774,
      "step": 3679
    },
    {
      "epoch": 0.20254279266883152,
      "grad_norm": 0.7967603802680969,
      "learning_rate": 9.761115231981878e-06,
      "loss": 0.919,
      "step": 3680
    },
    {
      "epoch": 0.20259783147118718,
      "grad_norm": 0.7425099611282349,
      "learning_rate": 9.760982831888236e-06,
      "loss": 0.819,
      "step": 3681
    },
    {
      "epoch": 0.20265287027354284,
      "grad_norm": 0.7631763815879822,
      "learning_rate": 9.760850396012323e-06,
      "loss": 0.816,
      "step": 3682
    },
    {
      "epoch": 0.2027079090758985,
      "grad_norm": 0.7931755185127258,
      "learning_rate": 9.76071792435513e-06,
      "loss": 0.8299,
      "step": 3683
    },
    {
      "epoch": 0.20276294787825416,
      "grad_norm": 0.8409438729286194,
      "learning_rate": 9.760585416917657e-06,
      "loss": 0.8503,
      "step": 3684
    },
    {
      "epoch": 0.20281798668060982,
      "grad_norm": 0.7632728815078735,
      "learning_rate": 9.760452873700898e-06,
      "loss": 0.8394,
      "step": 3685
    },
    {
      "epoch": 0.20287302548296549,
      "grad_norm": 0.7765083312988281,
      "learning_rate": 9.76032029470585e-06,
      "loss": 0.8879,
      "step": 3686
    },
    {
      "epoch": 0.20292806428532115,
      "grad_norm": 0.7736936807632446,
      "learning_rate": 9.760187679933507e-06,
      "loss": 0.7987,
      "step": 3687
    },
    {
      "epoch": 0.2029831030876768,
      "grad_norm": 0.8270270824432373,
      "learning_rate": 9.760055029384869e-06,
      "loss": 0.8267,
      "step": 3688
    },
    {
      "epoch": 0.20303814189003247,
      "grad_norm": 0.7742369174957275,
      "learning_rate": 9.759922343060932e-06,
      "loss": 0.8447,
      "step": 3689
    },
    {
      "epoch": 0.20309318069238813,
      "grad_norm": 0.7543869018554688,
      "learning_rate": 9.759789620962692e-06,
      "loss": 0.7325,
      "step": 3690
    },
    {
      "epoch": 0.2031482194947438,
      "grad_norm": 0.7913174033164978,
      "learning_rate": 9.759656863091147e-06,
      "loss": 0.8622,
      "step": 3691
    },
    {
      "epoch": 0.20320325829709945,
      "grad_norm": 0.7445376515388489,
      "learning_rate": 9.759524069447296e-06,
      "loss": 0.7115,
      "step": 3692
    },
    {
      "epoch": 0.20325829709945512,
      "grad_norm": 0.7744696140289307,
      "learning_rate": 9.759391240032136e-06,
      "loss": 0.8437,
      "step": 3693
    },
    {
      "epoch": 0.20331333590181078,
      "grad_norm": 0.6984724998474121,
      "learning_rate": 9.759258374846665e-06,
      "loss": 0.7415,
      "step": 3694
    },
    {
      "epoch": 0.20336837470416644,
      "grad_norm": 0.7453249096870422,
      "learning_rate": 9.759125473891882e-06,
      "loss": 0.7708,
      "step": 3695
    },
    {
      "epoch": 0.2034234135065221,
      "grad_norm": 0.7459438443183899,
      "learning_rate": 9.758992537168787e-06,
      "loss": 0.7961,
      "step": 3696
    },
    {
      "epoch": 0.20347845230887776,
      "grad_norm": 0.808944582939148,
      "learning_rate": 9.758859564678377e-06,
      "loss": 0.8875,
      "step": 3697
    },
    {
      "epoch": 0.20353349111123342,
      "grad_norm": 0.7202889323234558,
      "learning_rate": 9.758726556421652e-06,
      "loss": 0.8064,
      "step": 3698
    },
    {
      "epoch": 0.20358852991358908,
      "grad_norm": 0.7874952554702759,
      "learning_rate": 9.758593512399613e-06,
      "loss": 0.7881,
      "step": 3699
    },
    {
      "epoch": 0.20364356871594474,
      "grad_norm": 0.771300733089447,
      "learning_rate": 9.758460432613259e-06,
      "loss": 0.8938,
      "step": 3700
    },
    {
      "epoch": 0.2036986075183004,
      "grad_norm": 0.7332000136375427,
      "learning_rate": 9.758327317063589e-06,
      "loss": 0.7369,
      "step": 3701
    },
    {
      "epoch": 0.20375364632065607,
      "grad_norm": 0.8206236958503723,
      "learning_rate": 9.758194165751604e-06,
      "loss": 0.8727,
      "step": 3702
    },
    {
      "epoch": 0.20380868512301173,
      "grad_norm": 0.7750238180160522,
      "learning_rate": 9.758060978678308e-06,
      "loss": 0.8013,
      "step": 3703
    },
    {
      "epoch": 0.2038637239253674,
      "grad_norm": 0.7213704586029053,
      "learning_rate": 9.757927755844698e-06,
      "loss": 0.7413,
      "step": 3704
    },
    {
      "epoch": 0.20391876272772305,
      "grad_norm": 0.8982640504837036,
      "learning_rate": 9.757794497251776e-06,
      "loss": 0.9077,
      "step": 3705
    },
    {
      "epoch": 0.2039738015300787,
      "grad_norm": 0.8439363241195679,
      "learning_rate": 9.757661202900544e-06,
      "loss": 0.7887,
      "step": 3706
    },
    {
      "epoch": 0.20402884033243437,
      "grad_norm": 0.7700560688972473,
      "learning_rate": 9.757527872792005e-06,
      "loss": 0.8677,
      "step": 3707
    },
    {
      "epoch": 0.20408387913479004,
      "grad_norm": 0.7462438941001892,
      "learning_rate": 9.75739450692716e-06,
      "loss": 0.7937,
      "step": 3708
    },
    {
      "epoch": 0.2041389179371457,
      "grad_norm": 0.9125999808311462,
      "learning_rate": 9.75726110530701e-06,
      "loss": 0.9374,
      "step": 3709
    },
    {
      "epoch": 0.20419395673950136,
      "grad_norm": 0.8949875831604004,
      "learning_rate": 9.75712766793256e-06,
      "loss": 0.8586,
      "step": 3710
    },
    {
      "epoch": 0.20424899554185702,
      "grad_norm": 0.9042442440986633,
      "learning_rate": 9.756994194804812e-06,
      "loss": 0.9411,
      "step": 3711
    },
    {
      "epoch": 0.20430403434421268,
      "grad_norm": 0.7646238207817078,
      "learning_rate": 9.756860685924769e-06,
      "loss": 0.8353,
      "step": 3712
    },
    {
      "epoch": 0.20435907314656834,
      "grad_norm": 0.7551934123039246,
      "learning_rate": 9.756727141293434e-06,
      "loss": 0.8109,
      "step": 3713
    },
    {
      "epoch": 0.204414111948924,
      "grad_norm": 0.7526532411575317,
      "learning_rate": 9.756593560911811e-06,
      "loss": 0.8509,
      "step": 3714
    },
    {
      "epoch": 0.20446915075127967,
      "grad_norm": 0.8423319458961487,
      "learning_rate": 9.756459944780903e-06,
      "loss": 0.9003,
      "step": 3715
    },
    {
      "epoch": 0.2045241895536353,
      "grad_norm": 0.7966015934944153,
      "learning_rate": 9.756326292901716e-06,
      "loss": 0.7606,
      "step": 3716
    },
    {
      "epoch": 0.20457922835599096,
      "grad_norm": 0.7642805576324463,
      "learning_rate": 9.756192605275256e-06,
      "loss": 0.8321,
      "step": 3717
    },
    {
      "epoch": 0.20463426715834662,
      "grad_norm": 0.7285729646682739,
      "learning_rate": 9.756058881902524e-06,
      "loss": 0.7375,
      "step": 3718
    },
    {
      "epoch": 0.20468930596070228,
      "grad_norm": 0.852020263671875,
      "learning_rate": 9.755925122784525e-06,
      "loss": 0.8207,
      "step": 3719
    },
    {
      "epoch": 0.20474434476305794,
      "grad_norm": 0.8227072358131409,
      "learning_rate": 9.755791327922268e-06,
      "loss": 0.872,
      "step": 3720
    },
    {
      "epoch": 0.2047993835654136,
      "grad_norm": 1.0128127336502075,
      "learning_rate": 9.755657497316755e-06,
      "loss": 0.9186,
      "step": 3721
    },
    {
      "epoch": 0.20485442236776927,
      "grad_norm": 0.8208017349243164,
      "learning_rate": 9.755523630968994e-06,
      "loss": 0.6968,
      "step": 3722
    },
    {
      "epoch": 0.20490946117012493,
      "grad_norm": 0.7716407179832458,
      "learning_rate": 9.75538972887999e-06,
      "loss": 0.8068,
      "step": 3723
    },
    {
      "epoch": 0.2049644999724806,
      "grad_norm": 0.779608964920044,
      "learning_rate": 9.75525579105075e-06,
      "loss": 0.6968,
      "step": 3724
    },
    {
      "epoch": 0.20501953877483625,
      "grad_norm": 0.7463479042053223,
      "learning_rate": 9.75512181748228e-06,
      "loss": 0.7581,
      "step": 3725
    },
    {
      "epoch": 0.2050745775771919,
      "grad_norm": 0.8104956150054932,
      "learning_rate": 9.754987808175587e-06,
      "loss": 0.7838,
      "step": 3726
    },
    {
      "epoch": 0.20512961637954757,
      "grad_norm": 0.7911564707756042,
      "learning_rate": 9.75485376313168e-06,
      "loss": 0.848,
      "step": 3727
    },
    {
      "epoch": 0.20518465518190324,
      "grad_norm": 0.8340871334075928,
      "learning_rate": 9.754719682351564e-06,
      "loss": 0.7879,
      "step": 3728
    },
    {
      "epoch": 0.2052396939842589,
      "grad_norm": 1.5543067455291748,
      "learning_rate": 9.754585565836247e-06,
      "loss": 0.8091,
      "step": 3729
    },
    {
      "epoch": 0.20529473278661456,
      "grad_norm": 0.8262580633163452,
      "learning_rate": 9.754451413586739e-06,
      "loss": 0.9076,
      "step": 3730
    },
    {
      "epoch": 0.20534977158897022,
      "grad_norm": 0.7558280825614929,
      "learning_rate": 9.754317225604045e-06,
      "loss": 0.7781,
      "step": 3731
    },
    {
      "epoch": 0.20540481039132588,
      "grad_norm": 0.7197710275650024,
      "learning_rate": 9.754183001889177e-06,
      "loss": 0.765,
      "step": 3732
    },
    {
      "epoch": 0.20545984919368154,
      "grad_norm": 0.8053440451622009,
      "learning_rate": 9.754048742443141e-06,
      "loss": 0.7986,
      "step": 3733
    },
    {
      "epoch": 0.2055148879960372,
      "grad_norm": 0.9183983206748962,
      "learning_rate": 9.753914447266947e-06,
      "loss": 0.8522,
      "step": 3734
    },
    {
      "epoch": 0.20556992679839287,
      "grad_norm": 0.8095504641532898,
      "learning_rate": 9.753780116361607e-06,
      "loss": 0.7243,
      "step": 3735
    },
    {
      "epoch": 0.20562496560074853,
      "grad_norm": 0.816818356513977,
      "learning_rate": 9.753645749728127e-06,
      "loss": 0.8262,
      "step": 3736
    },
    {
      "epoch": 0.2056800044031042,
      "grad_norm": 0.8425988554954529,
      "learning_rate": 9.753511347367516e-06,
      "loss": 0.8142,
      "step": 3737
    },
    {
      "epoch": 0.20573504320545985,
      "grad_norm": 0.7719724178314209,
      "learning_rate": 9.753376909280789e-06,
      "loss": 0.8444,
      "step": 3738
    },
    {
      "epoch": 0.2057900820078155,
      "grad_norm": 0.877646803855896,
      "learning_rate": 9.753242435468952e-06,
      "loss": 0.8515,
      "step": 3739
    },
    {
      "epoch": 0.20584512081017117,
      "grad_norm": 0.9261211156845093,
      "learning_rate": 9.753107925933017e-06,
      "loss": 0.7605,
      "step": 3740
    },
    {
      "epoch": 0.20590015961252683,
      "grad_norm": 0.7790889739990234,
      "learning_rate": 9.752973380673995e-06,
      "loss": 0.7911,
      "step": 3741
    },
    {
      "epoch": 0.2059551984148825,
      "grad_norm": 0.7112367153167725,
      "learning_rate": 9.752838799692899e-06,
      "loss": 0.8212,
      "step": 3742
    },
    {
      "epoch": 0.20601023721723816,
      "grad_norm": 0.7568365335464478,
      "learning_rate": 9.752704182990736e-06,
      "loss": 0.8505,
      "step": 3743
    },
    {
      "epoch": 0.20606527601959382,
      "grad_norm": 0.7501981258392334,
      "learning_rate": 9.752569530568523e-06,
      "loss": 0.8191,
      "step": 3744
    },
    {
      "epoch": 0.20612031482194948,
      "grad_norm": 0.7822220325469971,
      "learning_rate": 9.752434842427268e-06,
      "loss": 0.8032,
      "step": 3745
    },
    {
      "epoch": 0.20617535362430514,
      "grad_norm": 0.810197114944458,
      "learning_rate": 9.752300118567987e-06,
      "loss": 0.7789,
      "step": 3746
    },
    {
      "epoch": 0.2062303924266608,
      "grad_norm": 0.7386943101882935,
      "learning_rate": 9.752165358991688e-06,
      "loss": 0.7733,
      "step": 3747
    },
    {
      "epoch": 0.20628543122901646,
      "grad_norm": 0.7086807489395142,
      "learning_rate": 9.75203056369939e-06,
      "loss": 0.6328,
      "step": 3748
    },
    {
      "epoch": 0.20634047003137213,
      "grad_norm": 0.9881154894828796,
      "learning_rate": 9.751895732692099e-06,
      "loss": 0.8515,
      "step": 3749
    },
    {
      "epoch": 0.2063955088337278,
      "grad_norm": 0.813521683216095,
      "learning_rate": 9.751760865970831e-06,
      "loss": 0.8438,
      "step": 3750
    },
    {
      "epoch": 0.20645054763608345,
      "grad_norm": 0.8357470631599426,
      "learning_rate": 9.751625963536602e-06,
      "loss": 0.7635,
      "step": 3751
    },
    {
      "epoch": 0.2065055864384391,
      "grad_norm": 0.8629693388938904,
      "learning_rate": 9.751491025390423e-06,
      "loss": 0.888,
      "step": 3752
    },
    {
      "epoch": 0.20656062524079477,
      "grad_norm": 0.8844664096832275,
      "learning_rate": 9.751356051533311e-06,
      "loss": 0.7654,
      "step": 3753
    },
    {
      "epoch": 0.20661566404315043,
      "grad_norm": 0.7006319165229797,
      "learning_rate": 9.751221041966276e-06,
      "loss": 0.7618,
      "step": 3754
    },
    {
      "epoch": 0.2066707028455061,
      "grad_norm": 0.9291046261787415,
      "learning_rate": 9.75108599669034e-06,
      "loss": 0.8485,
      "step": 3755
    },
    {
      "epoch": 0.20672574164786175,
      "grad_norm": 0.7670828700065613,
      "learning_rate": 9.75095091570651e-06,
      "loss": 0.7856,
      "step": 3756
    },
    {
      "epoch": 0.20678078045021742,
      "grad_norm": 0.8709883689880371,
      "learning_rate": 9.750815799015804e-06,
      "loss": 0.7983,
      "step": 3757
    },
    {
      "epoch": 0.20683581925257308,
      "grad_norm": 0.7688055634498596,
      "learning_rate": 9.750680646619241e-06,
      "loss": 0.8064,
      "step": 3758
    },
    {
      "epoch": 0.2068908580549287,
      "grad_norm": 0.9492738246917725,
      "learning_rate": 9.750545458517832e-06,
      "loss": 0.8256,
      "step": 3759
    },
    {
      "epoch": 0.20694589685728437,
      "grad_norm": 0.9685352444648743,
      "learning_rate": 9.750410234712596e-06,
      "loss": 0.839,
      "step": 3760
    },
    {
      "epoch": 0.20700093565964003,
      "grad_norm": 0.788577139377594,
      "learning_rate": 9.750274975204547e-06,
      "loss": 0.8743,
      "step": 3761
    },
    {
      "epoch": 0.2070559744619957,
      "grad_norm": 0.8496370315551758,
      "learning_rate": 9.750139679994703e-06,
      "loss": 0.9286,
      "step": 3762
    },
    {
      "epoch": 0.20711101326435136,
      "grad_norm": 0.9539788961410522,
      "learning_rate": 9.750004349084083e-06,
      "loss": 0.7568,
      "step": 3763
    },
    {
      "epoch": 0.20716605206670702,
      "grad_norm": 0.8825643062591553,
      "learning_rate": 9.7498689824737e-06,
      "loss": 0.9339,
      "step": 3764
    },
    {
      "epoch": 0.20722109086906268,
      "grad_norm": 0.7771373391151428,
      "learning_rate": 9.749733580164573e-06,
      "loss": 0.851,
      "step": 3765
    },
    {
      "epoch": 0.20727612967141834,
      "grad_norm": 0.7460281252861023,
      "learning_rate": 9.749598142157721e-06,
      "loss": 0.8208,
      "step": 3766
    },
    {
      "epoch": 0.207331168473774,
      "grad_norm": 0.8370739817619324,
      "learning_rate": 9.74946266845416e-06,
      "loss": 0.8634,
      "step": 3767
    },
    {
      "epoch": 0.20738620727612966,
      "grad_norm": 0.7770463228225708,
      "learning_rate": 9.749327159054907e-06,
      "loss": 0.7955,
      "step": 3768
    },
    {
      "epoch": 0.20744124607848533,
      "grad_norm": 0.8048208355903625,
      "learning_rate": 9.749191613960985e-06,
      "loss": 0.7736,
      "step": 3769
    },
    {
      "epoch": 0.207496284880841,
      "grad_norm": 0.9187547564506531,
      "learning_rate": 9.74905603317341e-06,
      "loss": 0.8534,
      "step": 3770
    },
    {
      "epoch": 0.20755132368319665,
      "grad_norm": 0.7304024696350098,
      "learning_rate": 9.7489204166932e-06,
      "loss": 0.72,
      "step": 3771
    },
    {
      "epoch": 0.2076063624855523,
      "grad_norm": 0.86177659034729,
      "learning_rate": 9.748784764521376e-06,
      "loss": 0.7838,
      "step": 3772
    },
    {
      "epoch": 0.20766140128790797,
      "grad_norm": 0.7988011837005615,
      "learning_rate": 9.748649076658956e-06,
      "loss": 0.7776,
      "step": 3773
    },
    {
      "epoch": 0.20771644009026363,
      "grad_norm": 0.706099808216095,
      "learning_rate": 9.74851335310696e-06,
      "loss": 0.759,
      "step": 3774
    },
    {
      "epoch": 0.2077714788926193,
      "grad_norm": 0.8125914931297302,
      "learning_rate": 9.748377593866412e-06,
      "loss": 0.8155,
      "step": 3775
    },
    {
      "epoch": 0.20782651769497495,
      "grad_norm": 0.8603429794311523,
      "learning_rate": 9.748241798938326e-06,
      "loss": 0.8018,
      "step": 3776
    },
    {
      "epoch": 0.20788155649733062,
      "grad_norm": 0.7735254764556885,
      "learning_rate": 9.748105968323726e-06,
      "loss": 0.7788,
      "step": 3777
    },
    {
      "epoch": 0.20793659529968628,
      "grad_norm": 0.9037501811981201,
      "learning_rate": 9.747970102023635e-06,
      "loss": 0.8907,
      "step": 3778
    },
    {
      "epoch": 0.20799163410204194,
      "grad_norm": 0.8781846761703491,
      "learning_rate": 9.74783420003907e-06,
      "loss": 0.867,
      "step": 3779
    },
    {
      "epoch": 0.2080466729043976,
      "grad_norm": 0.8486423492431641,
      "learning_rate": 9.747698262371052e-06,
      "loss": 0.817,
      "step": 3780
    },
    {
      "epoch": 0.20810171170675326,
      "grad_norm": 0.8242751359939575,
      "learning_rate": 9.747562289020607e-06,
      "loss": 0.7385,
      "step": 3781
    },
    {
      "epoch": 0.20815675050910892,
      "grad_norm": 0.8776529431343079,
      "learning_rate": 9.747426279988754e-06,
      "loss": 0.8222,
      "step": 3782
    },
    {
      "epoch": 0.20821178931146458,
      "grad_norm": 0.7428975105285645,
      "learning_rate": 9.747290235276517e-06,
      "loss": 0.6954,
      "step": 3783
    },
    {
      "epoch": 0.20826682811382025,
      "grad_norm": 0.8631997108459473,
      "learning_rate": 9.747154154884917e-06,
      "loss": 0.7956,
      "step": 3784
    },
    {
      "epoch": 0.2083218669161759,
      "grad_norm": 0.7819229364395142,
      "learning_rate": 9.747018038814976e-06,
      "loss": 0.778,
      "step": 3785
    },
    {
      "epoch": 0.20837690571853157,
      "grad_norm": 0.7770963311195374,
      "learning_rate": 9.746881887067718e-06,
      "loss": 0.8055,
      "step": 3786
    },
    {
      "epoch": 0.20843194452088723,
      "grad_norm": 0.7168729305267334,
      "learning_rate": 9.746745699644169e-06,
      "loss": 0.7476,
      "step": 3787
    },
    {
      "epoch": 0.2084869833232429,
      "grad_norm": 0.7963632941246033,
      "learning_rate": 9.746609476545348e-06,
      "loss": 0.8083,
      "step": 3788
    },
    {
      "epoch": 0.20854202212559855,
      "grad_norm": 0.6689679026603699,
      "learning_rate": 9.746473217772281e-06,
      "loss": 0.6687,
      "step": 3789
    },
    {
      "epoch": 0.20859706092795421,
      "grad_norm": 0.8085560202598572,
      "learning_rate": 9.746336923325991e-06,
      "loss": 0.8221,
      "step": 3790
    },
    {
      "epoch": 0.20865209973030988,
      "grad_norm": 0.7215744256973267,
      "learning_rate": 9.746200593207505e-06,
      "loss": 0.7261,
      "step": 3791
    },
    {
      "epoch": 0.20870713853266554,
      "grad_norm": 0.7821729779243469,
      "learning_rate": 9.746064227417844e-06,
      "loss": 0.7683,
      "step": 3792
    },
    {
      "epoch": 0.2087621773350212,
      "grad_norm": 1.0014925003051758,
      "learning_rate": 9.745927825958036e-06,
      "loss": 0.7485,
      "step": 3793
    },
    {
      "epoch": 0.20881721613737686,
      "grad_norm": 0.9447367787361145,
      "learning_rate": 9.745791388829102e-06,
      "loss": 0.835,
      "step": 3794
    },
    {
      "epoch": 0.20887225493973252,
      "grad_norm": 0.7333751916885376,
      "learning_rate": 9.745654916032073e-06,
      "loss": 0.811,
      "step": 3795
    },
    {
      "epoch": 0.20892729374208818,
      "grad_norm": 0.7516912221908569,
      "learning_rate": 9.745518407567973e-06,
      "loss": 0.7669,
      "step": 3796
    },
    {
      "epoch": 0.20898233254444384,
      "grad_norm": 0.7826053500175476,
      "learning_rate": 9.745381863437824e-06,
      "loss": 0.7963,
      "step": 3797
    },
    {
      "epoch": 0.2090373713467995,
      "grad_norm": 0.8258751630783081,
      "learning_rate": 9.745245283642658e-06,
      "loss": 0.7929,
      "step": 3798
    },
    {
      "epoch": 0.20909241014915517,
      "grad_norm": 0.7990522980690002,
      "learning_rate": 9.745108668183497e-06,
      "loss": 0.8518,
      "step": 3799
    },
    {
      "epoch": 0.20914744895151083,
      "grad_norm": 1.3855403661727905,
      "learning_rate": 9.744972017061369e-06,
      "loss": 0.7768,
      "step": 3800
    },
    {
      "epoch": 0.2092024877538665,
      "grad_norm": 0.8456707000732422,
      "learning_rate": 9.744835330277302e-06,
      "loss": 0.7629,
      "step": 3801
    },
    {
      "epoch": 0.20925752655622212,
      "grad_norm": 0.8992564678192139,
      "learning_rate": 9.744698607832323e-06,
      "loss": 0.8991,
      "step": 3802
    },
    {
      "epoch": 0.20931256535857778,
      "grad_norm": 0.8533509969711304,
      "learning_rate": 9.744561849727459e-06,
      "loss": 0.8883,
      "step": 3803
    },
    {
      "epoch": 0.20936760416093345,
      "grad_norm": 0.8363122940063477,
      "learning_rate": 9.744425055963739e-06,
      "loss": 0.8537,
      "step": 3804
    },
    {
      "epoch": 0.2094226429632891,
      "grad_norm": 0.7462213039398193,
      "learning_rate": 9.744288226542189e-06,
      "loss": 0.7713,
      "step": 3805
    },
    {
      "epoch": 0.20947768176564477,
      "grad_norm": 0.8148539066314697,
      "learning_rate": 9.744151361463841e-06,
      "loss": 0.7887,
      "step": 3806
    },
    {
      "epoch": 0.20953272056800043,
      "grad_norm": 0.7504319548606873,
      "learning_rate": 9.744014460729718e-06,
      "loss": 0.7385,
      "step": 3807
    },
    {
      "epoch": 0.2095877593703561,
      "grad_norm": 0.9291114807128906,
      "learning_rate": 9.743877524340854e-06,
      "loss": 0.9886,
      "step": 3808
    },
    {
      "epoch": 0.20964279817271175,
      "grad_norm": 0.7747925519943237,
      "learning_rate": 9.743740552298276e-06,
      "loss": 0.8772,
      "step": 3809
    },
    {
      "epoch": 0.20969783697506741,
      "grad_norm": 0.7283097505569458,
      "learning_rate": 9.743603544603016e-06,
      "loss": 0.7403,
      "step": 3810
    },
    {
      "epoch": 0.20975287577742308,
      "grad_norm": 0.8403457999229431,
      "learning_rate": 9.743466501256098e-06,
      "loss": 0.7998,
      "step": 3811
    },
    {
      "epoch": 0.20980791457977874,
      "grad_norm": 0.8218665719032288,
      "learning_rate": 9.743329422258557e-06,
      "loss": 0.8019,
      "step": 3812
    },
    {
      "epoch": 0.2098629533821344,
      "grad_norm": 0.6991317868232727,
      "learning_rate": 9.743192307611423e-06,
      "loss": 0.743,
      "step": 3813
    },
    {
      "epoch": 0.20991799218449006,
      "grad_norm": 0.767295241355896,
      "learning_rate": 9.743055157315725e-06,
      "loss": 0.8003,
      "step": 3814
    },
    {
      "epoch": 0.20997303098684572,
      "grad_norm": 0.9457303285598755,
      "learning_rate": 9.742917971372492e-06,
      "loss": 0.8448,
      "step": 3815
    },
    {
      "epoch": 0.21002806978920138,
      "grad_norm": 0.7839058637619019,
      "learning_rate": 9.742780749782758e-06,
      "loss": 0.8828,
      "step": 3816
    },
    {
      "epoch": 0.21008310859155704,
      "grad_norm": 0.7831344604492188,
      "learning_rate": 9.742643492547553e-06,
      "loss": 0.7714,
      "step": 3817
    },
    {
      "epoch": 0.2101381473939127,
      "grad_norm": 0.7637175917625427,
      "learning_rate": 9.74250619966791e-06,
      "loss": 0.7508,
      "step": 3818
    },
    {
      "epoch": 0.21019318619626837,
      "grad_norm": 0.8830221891403198,
      "learning_rate": 9.74236887114486e-06,
      "loss": 0.8508,
      "step": 3819
    },
    {
      "epoch": 0.21024822499862403,
      "grad_norm": 0.7803365588188171,
      "learning_rate": 9.742231506979434e-06,
      "loss": 0.8094,
      "step": 3820
    },
    {
      "epoch": 0.2103032638009797,
      "grad_norm": 0.7701493501663208,
      "learning_rate": 9.742094107172666e-06,
      "loss": 0.8851,
      "step": 3821
    },
    {
      "epoch": 0.21035830260333535,
      "grad_norm": 0.6434544324874878,
      "learning_rate": 9.741956671725588e-06,
      "loss": 0.7015,
      "step": 3822
    },
    {
      "epoch": 0.210413341405691,
      "grad_norm": 0.7294684052467346,
      "learning_rate": 9.741819200639233e-06,
      "loss": 0.7357,
      "step": 3823
    },
    {
      "epoch": 0.21046838020804667,
      "grad_norm": 0.702367901802063,
      "learning_rate": 9.741681693914635e-06,
      "loss": 0.7518,
      "step": 3824
    },
    {
      "epoch": 0.21052341901040234,
      "grad_norm": 0.7567246556282043,
      "learning_rate": 9.741544151552826e-06,
      "loss": 0.8259,
      "step": 3825
    },
    {
      "epoch": 0.210578457812758,
      "grad_norm": 0.7478607892990112,
      "learning_rate": 9.741406573554841e-06,
      "loss": 0.81,
      "step": 3826
    },
    {
      "epoch": 0.21063349661511366,
      "grad_norm": 0.7270129323005676,
      "learning_rate": 9.741268959921712e-06,
      "loss": 0.8201,
      "step": 3827
    },
    {
      "epoch": 0.21068853541746932,
      "grad_norm": 0.8108176589012146,
      "learning_rate": 9.741131310654475e-06,
      "loss": 0.8425,
      "step": 3828
    },
    {
      "epoch": 0.21074357421982498,
      "grad_norm": 0.7773691415786743,
      "learning_rate": 9.740993625754165e-06,
      "loss": 0.8372,
      "step": 3829
    },
    {
      "epoch": 0.21079861302218064,
      "grad_norm": 0.8988421559333801,
      "learning_rate": 9.740855905221816e-06,
      "loss": 0.8285,
      "step": 3830
    },
    {
      "epoch": 0.2108536518245363,
      "grad_norm": 0.7339534759521484,
      "learning_rate": 9.740718149058462e-06,
      "loss": 0.7567,
      "step": 3831
    },
    {
      "epoch": 0.21090869062689196,
      "grad_norm": 0.8465108275413513,
      "learning_rate": 9.740580357265141e-06,
      "loss": 0.8747,
      "step": 3832
    },
    {
      "epoch": 0.21096372942924763,
      "grad_norm": 0.7956714034080505,
      "learning_rate": 9.740442529842885e-06,
      "loss": 0.7665,
      "step": 3833
    },
    {
      "epoch": 0.2110187682316033,
      "grad_norm": 0.96270751953125,
      "learning_rate": 9.740304666792733e-06,
      "loss": 0.8338,
      "step": 3834
    },
    {
      "epoch": 0.21107380703395895,
      "grad_norm": 0.812329113483429,
      "learning_rate": 9.74016676811572e-06,
      "loss": 0.8407,
      "step": 3835
    },
    {
      "epoch": 0.2111288458363146,
      "grad_norm": 0.7975192070007324,
      "learning_rate": 9.740028833812882e-06,
      "loss": 0.7836,
      "step": 3836
    },
    {
      "epoch": 0.21118388463867027,
      "grad_norm": 0.826621949672699,
      "learning_rate": 9.739890863885258e-06,
      "loss": 0.732,
      "step": 3837
    },
    {
      "epoch": 0.21123892344102593,
      "grad_norm": 0.9015662670135498,
      "learning_rate": 9.73975285833388e-06,
      "loss": 0.8837,
      "step": 3838
    },
    {
      "epoch": 0.2112939622433816,
      "grad_norm": 0.7641518712043762,
      "learning_rate": 9.73961481715979e-06,
      "loss": 0.7334,
      "step": 3839
    },
    {
      "epoch": 0.21134900104573726,
      "grad_norm": 0.8062206506729126,
      "learning_rate": 9.739476740364023e-06,
      "loss": 0.8381,
      "step": 3840
    },
    {
      "epoch": 0.21140403984809292,
      "grad_norm": 0.7301875352859497,
      "learning_rate": 9.739338627947619e-06,
      "loss": 0.7389,
      "step": 3841
    },
    {
      "epoch": 0.21145907865044858,
      "grad_norm": 0.8995181322097778,
      "learning_rate": 9.739200479911612e-06,
      "loss": 0.8111,
      "step": 3842
    },
    {
      "epoch": 0.21151411745280424,
      "grad_norm": 0.7154433131217957,
      "learning_rate": 9.739062296257045e-06,
      "loss": 0.7501,
      "step": 3843
    },
    {
      "epoch": 0.2115691562551599,
      "grad_norm": 0.8403087854385376,
      "learning_rate": 9.738924076984954e-06,
      "loss": 0.8212,
      "step": 3844
    },
    {
      "epoch": 0.21162419505751554,
      "grad_norm": 0.7616639137268066,
      "learning_rate": 9.738785822096377e-06,
      "loss": 0.82,
      "step": 3845
    },
    {
      "epoch": 0.2116792338598712,
      "grad_norm": 0.7897970080375671,
      "learning_rate": 9.738647531592356e-06,
      "loss": 0.7972,
      "step": 3846
    },
    {
      "epoch": 0.21173427266222686,
      "grad_norm": 0.7909015417098999,
      "learning_rate": 9.738509205473928e-06,
      "loss": 0.7939,
      "step": 3847
    },
    {
      "epoch": 0.21178931146458252,
      "grad_norm": 0.9553212523460388,
      "learning_rate": 9.73837084374213e-06,
      "loss": 0.8672,
      "step": 3848
    },
    {
      "epoch": 0.21184435026693818,
      "grad_norm": 0.9558283686637878,
      "learning_rate": 9.73823244639801e-06,
      "loss": 0.897,
      "step": 3849
    },
    {
      "epoch": 0.21189938906929384,
      "grad_norm": 0.819530725479126,
      "learning_rate": 9.7380940134426e-06,
      "loss": 0.86,
      "step": 3850
    },
    {
      "epoch": 0.2119544278716495,
      "grad_norm": 0.7301751971244812,
      "learning_rate": 9.737955544876945e-06,
      "loss": 0.8265,
      "step": 3851
    },
    {
      "epoch": 0.21200946667400516,
      "grad_norm": 0.8564972281455994,
      "learning_rate": 9.737817040702085e-06,
      "loss": 0.8253,
      "step": 3852
    },
    {
      "epoch": 0.21206450547636083,
      "grad_norm": 0.7715204358100891,
      "learning_rate": 9.737678500919059e-06,
      "loss": 0.7779,
      "step": 3853
    },
    {
      "epoch": 0.2121195442787165,
      "grad_norm": 0.7296929955482483,
      "learning_rate": 9.73753992552891e-06,
      "loss": 0.787,
      "step": 3854
    },
    {
      "epoch": 0.21217458308107215,
      "grad_norm": 0.8574217557907104,
      "learning_rate": 9.73740131453268e-06,
      "loss": 0.797,
      "step": 3855
    },
    {
      "epoch": 0.2122296218834278,
      "grad_norm": 0.8320643901824951,
      "learning_rate": 9.737262667931409e-06,
      "loss": 0.876,
      "step": 3856
    },
    {
      "epoch": 0.21228466068578347,
      "grad_norm": 0.7313587069511414,
      "learning_rate": 9.73712398572614e-06,
      "loss": 0.7151,
      "step": 3857
    },
    {
      "epoch": 0.21233969948813913,
      "grad_norm": 0.7039312720298767,
      "learning_rate": 9.736985267917916e-06,
      "loss": 0.7353,
      "step": 3858
    },
    {
      "epoch": 0.2123947382904948,
      "grad_norm": 0.7893409132957458,
      "learning_rate": 9.736846514507776e-06,
      "loss": 0.8383,
      "step": 3859
    },
    {
      "epoch": 0.21244977709285046,
      "grad_norm": 0.8771371245384216,
      "learning_rate": 9.736707725496767e-06,
      "loss": 0.7543,
      "step": 3860
    },
    {
      "epoch": 0.21250481589520612,
      "grad_norm": 1.0067707300186157,
      "learning_rate": 9.736568900885932e-06,
      "loss": 0.796,
      "step": 3861
    },
    {
      "epoch": 0.21255985469756178,
      "grad_norm": 0.9171931743621826,
      "learning_rate": 9.736430040676312e-06,
      "loss": 0.8174,
      "step": 3862
    },
    {
      "epoch": 0.21261489349991744,
      "grad_norm": 0.7616068720817566,
      "learning_rate": 9.736291144868952e-06,
      "loss": 0.7762,
      "step": 3863
    },
    {
      "epoch": 0.2126699323022731,
      "grad_norm": 0.789010226726532,
      "learning_rate": 9.736152213464895e-06,
      "loss": 0.7749,
      "step": 3864
    },
    {
      "epoch": 0.21272497110462876,
      "grad_norm": 0.7943348288536072,
      "learning_rate": 9.736013246465187e-06,
      "loss": 0.6687,
      "step": 3865
    },
    {
      "epoch": 0.21278000990698442,
      "grad_norm": 0.8351758718490601,
      "learning_rate": 9.73587424387087e-06,
      "loss": 0.9201,
      "step": 3866
    },
    {
      "epoch": 0.21283504870934009,
      "grad_norm": 0.7710975408554077,
      "learning_rate": 9.735735205682991e-06,
      "loss": 0.8357,
      "step": 3867
    },
    {
      "epoch": 0.21289008751169575,
      "grad_norm": 0.8955768942832947,
      "learning_rate": 9.73559613190259e-06,
      "loss": 0.8396,
      "step": 3868
    },
    {
      "epoch": 0.2129451263140514,
      "grad_norm": 0.8664666414260864,
      "learning_rate": 9.735457022530722e-06,
      "loss": 0.8176,
      "step": 3869
    },
    {
      "epoch": 0.21300016511640707,
      "grad_norm": 0.7955949902534485,
      "learning_rate": 9.735317877568424e-06,
      "loss": 0.8421,
      "step": 3870
    },
    {
      "epoch": 0.21305520391876273,
      "grad_norm": 0.849866509437561,
      "learning_rate": 9.735178697016742e-06,
      "loss": 0.7677,
      "step": 3871
    },
    {
      "epoch": 0.2131102427211184,
      "grad_norm": 0.7782625555992126,
      "learning_rate": 9.735039480876727e-06,
      "loss": 0.7838,
      "step": 3872
    },
    {
      "epoch": 0.21316528152347405,
      "grad_norm": 0.7734919190406799,
      "learning_rate": 9.734900229149423e-06,
      "loss": 0.757,
      "step": 3873
    },
    {
      "epoch": 0.21322032032582972,
      "grad_norm": 0.8462040424346924,
      "learning_rate": 9.734760941835876e-06,
      "loss": 0.8841,
      "step": 3874
    },
    {
      "epoch": 0.21327535912818538,
      "grad_norm": 0.7219869494438171,
      "learning_rate": 9.734621618937133e-06,
      "loss": 0.7651,
      "step": 3875
    },
    {
      "epoch": 0.21333039793054104,
      "grad_norm": 0.7550874352455139,
      "learning_rate": 9.734482260454241e-06,
      "loss": 0.8032,
      "step": 3876
    },
    {
      "epoch": 0.2133854367328967,
      "grad_norm": 0.7504588961601257,
      "learning_rate": 9.734342866388247e-06,
      "loss": 0.7923,
      "step": 3877
    },
    {
      "epoch": 0.21344047553525236,
      "grad_norm": 0.7407390475273132,
      "learning_rate": 9.7342034367402e-06,
      "loss": 0.7569,
      "step": 3878
    },
    {
      "epoch": 0.21349551433760802,
      "grad_norm": 0.7911562323570251,
      "learning_rate": 9.734063971511147e-06,
      "loss": 0.8726,
      "step": 3879
    },
    {
      "epoch": 0.21355055313996368,
      "grad_norm": 0.9132450819015503,
      "learning_rate": 9.733924470702139e-06,
      "loss": 0.9445,
      "step": 3880
    },
    {
      "epoch": 0.21360559194231934,
      "grad_norm": 0.9639442563056946,
      "learning_rate": 9.733784934314218e-06,
      "loss": 0.7307,
      "step": 3881
    },
    {
      "epoch": 0.213660630744675,
      "grad_norm": 0.7724352478981018,
      "learning_rate": 9.73364536234844e-06,
      "loss": 0.8337,
      "step": 3882
    },
    {
      "epoch": 0.21371566954703067,
      "grad_norm": 0.9643296599388123,
      "learning_rate": 9.733505754805848e-06,
      "loss": 0.8277,
      "step": 3883
    },
    {
      "epoch": 0.21377070834938633,
      "grad_norm": 0.8135218620300293,
      "learning_rate": 9.733366111687494e-06,
      "loss": 0.7933,
      "step": 3884
    },
    {
      "epoch": 0.213825747151742,
      "grad_norm": 0.7527105212211609,
      "learning_rate": 9.733226432994426e-06,
      "loss": 0.7302,
      "step": 3885
    },
    {
      "epoch": 0.21388078595409765,
      "grad_norm": 1.090550184249878,
      "learning_rate": 9.733086718727698e-06,
      "loss": 0.8646,
      "step": 3886
    },
    {
      "epoch": 0.2139358247564533,
      "grad_norm": 0.9227491617202759,
      "learning_rate": 9.732946968888358e-06,
      "loss": 0.8525,
      "step": 3887
    },
    {
      "epoch": 0.21399086355880895,
      "grad_norm": 0.7781830430030823,
      "learning_rate": 9.732807183477454e-06,
      "loss": 0.8757,
      "step": 3888
    },
    {
      "epoch": 0.2140459023611646,
      "grad_norm": 0.7740090489387512,
      "learning_rate": 9.732667362496036e-06,
      "loss": 0.7557,
      "step": 3889
    },
    {
      "epoch": 0.21410094116352027,
      "grad_norm": 0.7341694831848145,
      "learning_rate": 9.732527505945159e-06,
      "loss": 0.7481,
      "step": 3890
    },
    {
      "epoch": 0.21415597996587593,
      "grad_norm": 0.8691402673721313,
      "learning_rate": 9.732387613825872e-06,
      "loss": 0.8395,
      "step": 3891
    },
    {
      "epoch": 0.2142110187682316,
      "grad_norm": 0.7845497131347656,
      "learning_rate": 9.732247686139227e-06,
      "loss": 0.6999,
      "step": 3892
    },
    {
      "epoch": 0.21426605757058725,
      "grad_norm": 0.7944281697273254,
      "learning_rate": 9.732107722886275e-06,
      "loss": 0.7677,
      "step": 3893
    },
    {
      "epoch": 0.21432109637294292,
      "grad_norm": 0.904195249080658,
      "learning_rate": 9.731967724068065e-06,
      "loss": 0.8429,
      "step": 3894
    },
    {
      "epoch": 0.21437613517529858,
      "grad_norm": 0.7968988418579102,
      "learning_rate": 9.731827689685655e-06,
      "loss": 0.8224,
      "step": 3895
    },
    {
      "epoch": 0.21443117397765424,
      "grad_norm": 0.773674726486206,
      "learning_rate": 9.731687619740095e-06,
      "loss": 0.7743,
      "step": 3896
    },
    {
      "epoch": 0.2144862127800099,
      "grad_norm": 0.7873631715774536,
      "learning_rate": 9.731547514232439e-06,
      "loss": 0.8581,
      "step": 3897
    },
    {
      "epoch": 0.21454125158236556,
      "grad_norm": 0.7989653944969177,
      "learning_rate": 9.731407373163735e-06,
      "loss": 0.8447,
      "step": 3898
    },
    {
      "epoch": 0.21459629038472122,
      "grad_norm": 0.74820876121521,
      "learning_rate": 9.73126719653504e-06,
      "loss": 0.8745,
      "step": 3899
    },
    {
      "epoch": 0.21465132918707688,
      "grad_norm": 0.7191246747970581,
      "learning_rate": 9.731126984347408e-06,
      "loss": 0.7533,
      "step": 3900
    },
    {
      "epoch": 0.21470636798943254,
      "grad_norm": 0.7718465328216553,
      "learning_rate": 9.730986736601893e-06,
      "loss": 0.8184,
      "step": 3901
    },
    {
      "epoch": 0.2147614067917882,
      "grad_norm": 0.7055066823959351,
      "learning_rate": 9.730846453299547e-06,
      "loss": 0.7352,
      "step": 3902
    },
    {
      "epoch": 0.21481644559414387,
      "grad_norm": 0.7500855326652527,
      "learning_rate": 9.730706134441425e-06,
      "loss": 0.8111,
      "step": 3903
    },
    {
      "epoch": 0.21487148439649953,
      "grad_norm": 0.7568232417106628,
      "learning_rate": 9.730565780028583e-06,
      "loss": 0.8126,
      "step": 3904
    },
    {
      "epoch": 0.2149265231988552,
      "grad_norm": 0.7418738007545471,
      "learning_rate": 9.730425390062075e-06,
      "loss": 0.8014,
      "step": 3905
    },
    {
      "epoch": 0.21498156200121085,
      "grad_norm": 0.7967441082000732,
      "learning_rate": 9.730284964542955e-06,
      "loss": 0.7965,
      "step": 3906
    },
    {
      "epoch": 0.2150366008035665,
      "grad_norm": 0.7444791197776794,
      "learning_rate": 9.730144503472281e-06,
      "loss": 0.7113,
      "step": 3907
    },
    {
      "epoch": 0.21509163960592217,
      "grad_norm": 0.8372869491577148,
      "learning_rate": 9.730004006851107e-06,
      "loss": 0.838,
      "step": 3908
    },
    {
      "epoch": 0.21514667840827784,
      "grad_norm": 0.7984300851821899,
      "learning_rate": 9.729863474680488e-06,
      "loss": 0.856,
      "step": 3909
    },
    {
      "epoch": 0.2152017172106335,
      "grad_norm": 0.7508612871170044,
      "learning_rate": 9.72972290696148e-06,
      "loss": 0.7947,
      "step": 3910
    },
    {
      "epoch": 0.21525675601298916,
      "grad_norm": 0.7559992074966431,
      "learning_rate": 9.729582303695142e-06,
      "loss": 0.785,
      "step": 3911
    },
    {
      "epoch": 0.21531179481534482,
      "grad_norm": 0.7764164209365845,
      "learning_rate": 9.729441664882531e-06,
      "loss": 0.8297,
      "step": 3912
    },
    {
      "epoch": 0.21536683361770048,
      "grad_norm": 0.8112726211547852,
      "learning_rate": 9.7293009905247e-06,
      "loss": 0.8073,
      "step": 3913
    },
    {
      "epoch": 0.21542187242005614,
      "grad_norm": 0.9748952388763428,
      "learning_rate": 9.729160280622709e-06,
      "loss": 0.7584,
      "step": 3914
    },
    {
      "epoch": 0.2154769112224118,
      "grad_norm": 0.789191484451294,
      "learning_rate": 9.729019535177617e-06,
      "loss": 0.7568,
      "step": 3915
    },
    {
      "epoch": 0.21553195002476747,
      "grad_norm": 0.7300963401794434,
      "learning_rate": 9.728878754190478e-06,
      "loss": 0.8029,
      "step": 3916
    },
    {
      "epoch": 0.21558698882712313,
      "grad_norm": 0.9201067686080933,
      "learning_rate": 9.728737937662354e-06,
      "loss": 0.8665,
      "step": 3917
    },
    {
      "epoch": 0.2156420276294788,
      "grad_norm": 0.8820425271987915,
      "learning_rate": 9.728597085594301e-06,
      "loss": 0.8378,
      "step": 3918
    },
    {
      "epoch": 0.21569706643183445,
      "grad_norm": 0.7762684226036072,
      "learning_rate": 9.728456197987376e-06,
      "loss": 0.8005,
      "step": 3919
    },
    {
      "epoch": 0.2157521052341901,
      "grad_norm": 0.8429732918739319,
      "learning_rate": 9.728315274842641e-06,
      "loss": 0.8337,
      "step": 3920
    },
    {
      "epoch": 0.21580714403654577,
      "grad_norm": 0.7820748090744019,
      "learning_rate": 9.728174316161156e-06,
      "loss": 0.8085,
      "step": 3921
    },
    {
      "epoch": 0.21586218283890143,
      "grad_norm": 0.8748064637184143,
      "learning_rate": 9.728033321943977e-06,
      "loss": 0.7734,
      "step": 3922
    },
    {
      "epoch": 0.2159172216412571,
      "grad_norm": 0.8878050446510315,
      "learning_rate": 9.727892292192166e-06,
      "loss": 0.9226,
      "step": 3923
    },
    {
      "epoch": 0.21597226044361276,
      "grad_norm": 0.8156047463417053,
      "learning_rate": 9.72775122690678e-06,
      "loss": 0.8111,
      "step": 3924
    },
    {
      "epoch": 0.21602729924596842,
      "grad_norm": 0.7392945885658264,
      "learning_rate": 9.727610126088883e-06,
      "loss": 0.758,
      "step": 3925
    },
    {
      "epoch": 0.21608233804832408,
      "grad_norm": 0.7573148608207703,
      "learning_rate": 9.727468989739532e-06,
      "loss": 0.8142,
      "step": 3926
    },
    {
      "epoch": 0.21613737685067974,
      "grad_norm": 0.831847608089447,
      "learning_rate": 9.727327817859792e-06,
      "loss": 0.7337,
      "step": 3927
    },
    {
      "epoch": 0.2161924156530354,
      "grad_norm": 0.8012371063232422,
      "learning_rate": 9.72718661045072e-06,
      "loss": 0.8128,
      "step": 3928
    },
    {
      "epoch": 0.21624745445539106,
      "grad_norm": 0.7985890507698059,
      "learning_rate": 9.72704536751338e-06,
      "loss": 0.8549,
      "step": 3929
    },
    {
      "epoch": 0.21630249325774673,
      "grad_norm": 0.7194695472717285,
      "learning_rate": 9.726904089048832e-06,
      "loss": 0.775,
      "step": 3930
    },
    {
      "epoch": 0.21635753206010236,
      "grad_norm": 0.8029330968856812,
      "learning_rate": 9.726762775058138e-06,
      "loss": 0.9167,
      "step": 3931
    },
    {
      "epoch": 0.21641257086245802,
      "grad_norm": 0.7388954162597656,
      "learning_rate": 9.72662142554236e-06,
      "loss": 0.7295,
      "step": 3932
    },
    {
      "epoch": 0.21646760966481368,
      "grad_norm": 0.798796534538269,
      "learning_rate": 9.726480040502559e-06,
      "loss": 0.8686,
      "step": 3933
    },
    {
      "epoch": 0.21652264846716934,
      "grad_norm": 0.9977202415466309,
      "learning_rate": 9.726338619939802e-06,
      "loss": 0.8387,
      "step": 3934
    },
    {
      "epoch": 0.216577687269525,
      "grad_norm": 0.8173295855522156,
      "learning_rate": 9.726197163855148e-06,
      "loss": 0.7773,
      "step": 3935
    },
    {
      "epoch": 0.21663272607188067,
      "grad_norm": 0.6519538760185242,
      "learning_rate": 9.72605567224966e-06,
      "loss": 0.6319,
      "step": 3936
    },
    {
      "epoch": 0.21668776487423633,
      "grad_norm": 0.8004894852638245,
      "learning_rate": 9.725914145124404e-06,
      "loss": 0.8281,
      "step": 3937
    },
    {
      "epoch": 0.216742803676592,
      "grad_norm": 0.7327558398246765,
      "learning_rate": 9.725772582480442e-06,
      "loss": 0.7105,
      "step": 3938
    },
    {
      "epoch": 0.21679784247894765,
      "grad_norm": 0.7624199986457825,
      "learning_rate": 9.725630984318839e-06,
      "loss": 0.7823,
      "step": 3939
    },
    {
      "epoch": 0.2168528812813033,
      "grad_norm": 0.7750238180160522,
      "learning_rate": 9.725489350640658e-06,
      "loss": 0.8147,
      "step": 3940
    },
    {
      "epoch": 0.21690792008365897,
      "grad_norm": 0.6886566877365112,
      "learning_rate": 9.725347681446964e-06,
      "loss": 0.7263,
      "step": 3941
    },
    {
      "epoch": 0.21696295888601463,
      "grad_norm": 0.882060170173645,
      "learning_rate": 9.725205976738821e-06,
      "loss": 0.8931,
      "step": 3942
    },
    {
      "epoch": 0.2170179976883703,
      "grad_norm": 0.7946881055831909,
      "learning_rate": 9.725064236517297e-06,
      "loss": 0.8036,
      "step": 3943
    },
    {
      "epoch": 0.21707303649072596,
      "grad_norm": 0.7062187194824219,
      "learning_rate": 9.724922460783453e-06,
      "loss": 0.6915,
      "step": 3944
    },
    {
      "epoch": 0.21712807529308162,
      "grad_norm": 0.7978640794754028,
      "learning_rate": 9.724780649538356e-06,
      "loss": 0.8873,
      "step": 3945
    },
    {
      "epoch": 0.21718311409543728,
      "grad_norm": 0.8828096389770508,
      "learning_rate": 9.724638802783073e-06,
      "loss": 0.7114,
      "step": 3946
    },
    {
      "epoch": 0.21723815289779294,
      "grad_norm": 0.7301073670387268,
      "learning_rate": 9.724496920518672e-06,
      "loss": 0.8107,
      "step": 3947
    },
    {
      "epoch": 0.2172931917001486,
      "grad_norm": 0.7944212555885315,
      "learning_rate": 9.724355002746213e-06,
      "loss": 0.8135,
      "step": 3948
    },
    {
      "epoch": 0.21734823050250426,
      "grad_norm": 0.7988898754119873,
      "learning_rate": 9.724213049466768e-06,
      "loss": 0.7173,
      "step": 3949
    },
    {
      "epoch": 0.21740326930485993,
      "grad_norm": 0.7734915018081665,
      "learning_rate": 9.724071060681401e-06,
      "loss": 0.8131,
      "step": 3950
    },
    {
      "epoch": 0.2174583081072156,
      "grad_norm": 0.6856646537780762,
      "learning_rate": 9.723929036391183e-06,
      "loss": 0.6873,
      "step": 3951
    },
    {
      "epoch": 0.21751334690957125,
      "grad_norm": 0.8652976751327515,
      "learning_rate": 9.723786976597179e-06,
      "loss": 0.7908,
      "step": 3952
    },
    {
      "epoch": 0.2175683857119269,
      "grad_norm": 0.7325445413589478,
      "learning_rate": 9.723644881300453e-06,
      "loss": 0.7389,
      "step": 3953
    },
    {
      "epoch": 0.21762342451428257,
      "grad_norm": 0.8596270084381104,
      "learning_rate": 9.723502750502079e-06,
      "loss": 0.7785,
      "step": 3954
    },
    {
      "epoch": 0.21767846331663823,
      "grad_norm": 0.739248514175415,
      "learning_rate": 9.723360584203123e-06,
      "loss": 0.8125,
      "step": 3955
    },
    {
      "epoch": 0.2177335021189939,
      "grad_norm": 0.815617561340332,
      "learning_rate": 9.723218382404652e-06,
      "loss": 0.8682,
      "step": 3956
    },
    {
      "epoch": 0.21778854092134955,
      "grad_norm": 0.758756160736084,
      "learning_rate": 9.723076145107738e-06,
      "loss": 0.7717,
      "step": 3957
    },
    {
      "epoch": 0.21784357972370522,
      "grad_norm": 0.9007643461227417,
      "learning_rate": 9.722933872313445e-06,
      "loss": 0.7901,
      "step": 3958
    },
    {
      "epoch": 0.21789861852606088,
      "grad_norm": 0.781548023223877,
      "learning_rate": 9.722791564022846e-06,
      "loss": 0.8338,
      "step": 3959
    },
    {
      "epoch": 0.21795365732841654,
      "grad_norm": 0.7730190753936768,
      "learning_rate": 9.722649220237011e-06,
      "loss": 0.8032,
      "step": 3960
    },
    {
      "epoch": 0.2180086961307722,
      "grad_norm": 0.8737791776657104,
      "learning_rate": 9.722506840957009e-06,
      "loss": 0.8436,
      "step": 3961
    },
    {
      "epoch": 0.21806373493312786,
      "grad_norm": 0.8151329159736633,
      "learning_rate": 9.722364426183908e-06,
      "loss": 0.8115,
      "step": 3962
    },
    {
      "epoch": 0.21811877373548352,
      "grad_norm": 0.7852860689163208,
      "learning_rate": 9.722221975918782e-06,
      "loss": 0.7977,
      "step": 3963
    },
    {
      "epoch": 0.21817381253783918,
      "grad_norm": 0.9064140319824219,
      "learning_rate": 9.722079490162698e-06,
      "loss": 0.8799,
      "step": 3964
    },
    {
      "epoch": 0.21822885134019485,
      "grad_norm": 0.8579906821250916,
      "learning_rate": 9.72193696891673e-06,
      "loss": 0.7825,
      "step": 3965
    },
    {
      "epoch": 0.2182838901425505,
      "grad_norm": 0.8005900382995605,
      "learning_rate": 9.721794412181946e-06,
      "loss": 0.8601,
      "step": 3966
    },
    {
      "epoch": 0.21833892894490617,
      "grad_norm": 0.7661529183387756,
      "learning_rate": 9.721651819959421e-06,
      "loss": 0.7446,
      "step": 3967
    },
    {
      "epoch": 0.21839396774726183,
      "grad_norm": 0.7558436989784241,
      "learning_rate": 9.721509192250224e-06,
      "loss": 0.7484,
      "step": 3968
    },
    {
      "epoch": 0.2184490065496175,
      "grad_norm": 0.765446126461029,
      "learning_rate": 9.721366529055427e-06,
      "loss": 0.7727,
      "step": 3969
    },
    {
      "epoch": 0.21850404535197315,
      "grad_norm": 0.7329973578453064,
      "learning_rate": 9.721223830376103e-06,
      "loss": 0.797,
      "step": 3970
    },
    {
      "epoch": 0.21855908415432881,
      "grad_norm": 0.8881974220275879,
      "learning_rate": 9.721081096213324e-06,
      "loss": 0.9199,
      "step": 3971
    },
    {
      "epoch": 0.21861412295668448,
      "grad_norm": 0.8246786594390869,
      "learning_rate": 9.720938326568165e-06,
      "loss": 0.9108,
      "step": 3972
    },
    {
      "epoch": 0.21866916175904014,
      "grad_norm": 0.7187291979789734,
      "learning_rate": 9.720795521441697e-06,
      "loss": 0.7756,
      "step": 3973
    },
    {
      "epoch": 0.21872420056139577,
      "grad_norm": 0.7880695462226868,
      "learning_rate": 9.720652680834995e-06,
      "loss": 0.8548,
      "step": 3974
    },
    {
      "epoch": 0.21877923936375143,
      "grad_norm": 0.8841108679771423,
      "learning_rate": 9.720509804749128e-06,
      "loss": 0.8477,
      "step": 3975
    },
    {
      "epoch": 0.2188342781661071,
      "grad_norm": 0.9061402678489685,
      "learning_rate": 9.720366893185173e-06,
      "loss": 0.8235,
      "step": 3976
    },
    {
      "epoch": 0.21888931696846275,
      "grad_norm": 0.8342392444610596,
      "learning_rate": 9.720223946144206e-06,
      "loss": 0.7777,
      "step": 3977
    },
    {
      "epoch": 0.21894435577081842,
      "grad_norm": 0.7933762073516846,
      "learning_rate": 9.720080963627299e-06,
      "loss": 0.7943,
      "step": 3978
    },
    {
      "epoch": 0.21899939457317408,
      "grad_norm": 0.8358896374702454,
      "learning_rate": 9.719937945635527e-06,
      "loss": 0.8932,
      "step": 3979
    },
    {
      "epoch": 0.21905443337552974,
      "grad_norm": 0.7479808926582336,
      "learning_rate": 9.719794892169964e-06,
      "loss": 0.7446,
      "step": 3980
    },
    {
      "epoch": 0.2191094721778854,
      "grad_norm": 0.7920958399772644,
      "learning_rate": 9.719651803231685e-06,
      "loss": 0.7489,
      "step": 3981
    },
    {
      "epoch": 0.21916451098024106,
      "grad_norm": 0.7098824977874756,
      "learning_rate": 9.719508678821768e-06,
      "loss": 0.7763,
      "step": 3982
    },
    {
      "epoch": 0.21921954978259672,
      "grad_norm": 0.8733491897583008,
      "learning_rate": 9.719365518941288e-06,
      "loss": 0.7325,
      "step": 3983
    },
    {
      "epoch": 0.21927458858495238,
      "grad_norm": 0.8328796029090881,
      "learning_rate": 9.719222323591318e-06,
      "loss": 0.9097,
      "step": 3984
    },
    {
      "epoch": 0.21932962738730805,
      "grad_norm": 0.7869352698326111,
      "learning_rate": 9.719079092772936e-06,
      "loss": 0.759,
      "step": 3985
    },
    {
      "epoch": 0.2193846661896637,
      "grad_norm": 0.8278539180755615,
      "learning_rate": 9.718935826487221e-06,
      "loss": 0.8545,
      "step": 3986
    },
    {
      "epoch": 0.21943970499201937,
      "grad_norm": 0.8122449517250061,
      "learning_rate": 9.718792524735246e-06,
      "loss": 0.7646,
      "step": 3987
    },
    {
      "epoch": 0.21949474379437503,
      "grad_norm": 1.072253942489624,
      "learning_rate": 9.71864918751809e-06,
      "loss": 0.915,
      "step": 3988
    },
    {
      "epoch": 0.2195497825967307,
      "grad_norm": 0.7770013213157654,
      "learning_rate": 9.718505814836829e-06,
      "loss": 0.7561,
      "step": 3989
    },
    {
      "epoch": 0.21960482139908635,
      "grad_norm": 0.9011678695678711,
      "learning_rate": 9.718362406692544e-06,
      "loss": 0.7532,
      "step": 3990
    },
    {
      "epoch": 0.21965986020144201,
      "grad_norm": 0.8867584466934204,
      "learning_rate": 9.718218963086307e-06,
      "loss": 0.8732,
      "step": 3991
    },
    {
      "epoch": 0.21971489900379768,
      "grad_norm": 0.8884773850440979,
      "learning_rate": 9.718075484019201e-06,
      "loss": 0.7403,
      "step": 3992
    },
    {
      "epoch": 0.21976993780615334,
      "grad_norm": 0.8995673060417175,
      "learning_rate": 9.7179319694923e-06,
      "loss": 0.9283,
      "step": 3993
    },
    {
      "epoch": 0.219824976608509,
      "grad_norm": 0.7875818014144897,
      "learning_rate": 9.717788419506688e-06,
      "loss": 0.8633,
      "step": 3994
    },
    {
      "epoch": 0.21988001541086466,
      "grad_norm": 0.7693219184875488,
      "learning_rate": 9.71764483406344e-06,
      "loss": 0.8073,
      "step": 3995
    },
    {
      "epoch": 0.21993505421322032,
      "grad_norm": 0.7932817339897156,
      "learning_rate": 9.717501213163636e-06,
      "loss": 0.7537,
      "step": 3996
    },
    {
      "epoch": 0.21999009301557598,
      "grad_norm": 0.8274912238121033,
      "learning_rate": 9.717357556808358e-06,
      "loss": 0.7715,
      "step": 3997
    },
    {
      "epoch": 0.22004513181793164,
      "grad_norm": 0.7533993124961853,
      "learning_rate": 9.71721386499868e-06,
      "loss": 0.7482,
      "step": 3998
    },
    {
      "epoch": 0.2201001706202873,
      "grad_norm": 1.028228759765625,
      "learning_rate": 9.717070137735687e-06,
      "loss": 0.9897,
      "step": 3999
    },
    {
      "epoch": 0.22015520942264297,
      "grad_norm": 1.1093978881835938,
      "learning_rate": 9.716926375020457e-06,
      "loss": 0.8701,
      "step": 4000
    },
    {
      "epoch": 0.22021024822499863,
      "grad_norm": 0.7891124486923218,
      "learning_rate": 9.716782576854073e-06,
      "loss": 0.8533,
      "step": 4001
    },
    {
      "epoch": 0.2202652870273543,
      "grad_norm": 1.1783788204193115,
      "learning_rate": 9.716638743237611e-06,
      "loss": 0.8088,
      "step": 4002
    },
    {
      "epoch": 0.22032032582970995,
      "grad_norm": 0.8713383078575134,
      "learning_rate": 9.716494874172157e-06,
      "loss": 0.8382,
      "step": 4003
    },
    {
      "epoch": 0.2203753646320656,
      "grad_norm": 0.7821565270423889,
      "learning_rate": 9.716350969658787e-06,
      "loss": 0.8168,
      "step": 4004
    },
    {
      "epoch": 0.22043040343442127,
      "grad_norm": 0.7642589211463928,
      "learning_rate": 9.716207029698589e-06,
      "loss": 0.7209,
      "step": 4005
    },
    {
      "epoch": 0.22048544223677694,
      "grad_norm": 0.935625433921814,
      "learning_rate": 9.716063054292639e-06,
      "loss": 0.8436,
      "step": 4006
    },
    {
      "epoch": 0.2205404810391326,
      "grad_norm": 0.7064627408981323,
      "learning_rate": 9.715919043442024e-06,
      "loss": 0.7651,
      "step": 4007
    },
    {
      "epoch": 0.22059551984148826,
      "grad_norm": 0.6980876326560974,
      "learning_rate": 9.715774997147823e-06,
      "loss": 0.7842,
      "step": 4008
    },
    {
      "epoch": 0.22065055864384392,
      "grad_norm": 0.7691119313240051,
      "learning_rate": 9.715630915411118e-06,
      "loss": 0.7345,
      "step": 4009
    },
    {
      "epoch": 0.22070559744619958,
      "grad_norm": 0.8870186805725098,
      "learning_rate": 9.715486798232994e-06,
      "loss": 0.7531,
      "step": 4010
    },
    {
      "epoch": 0.22076063624855524,
      "grad_norm": 0.7225383520126343,
      "learning_rate": 9.715342645614533e-06,
      "loss": 0.8543,
      "step": 4011
    },
    {
      "epoch": 0.2208156750509109,
      "grad_norm": 0.7517428994178772,
      "learning_rate": 9.71519845755682e-06,
      "loss": 0.84,
      "step": 4012
    },
    {
      "epoch": 0.22087071385326656,
      "grad_norm": 0.8115549087524414,
      "learning_rate": 9.715054234060937e-06,
      "loss": 0.7823,
      "step": 4013
    },
    {
      "epoch": 0.22092575265562223,
      "grad_norm": 1.6656148433685303,
      "learning_rate": 9.714909975127968e-06,
      "loss": 0.8951,
      "step": 4014
    },
    {
      "epoch": 0.2209807914579779,
      "grad_norm": 0.906508207321167,
      "learning_rate": 9.714765680758997e-06,
      "loss": 0.8599,
      "step": 4015
    },
    {
      "epoch": 0.22103583026033355,
      "grad_norm": 0.8274093866348267,
      "learning_rate": 9.71462135095511e-06,
      "loss": 0.9568,
      "step": 4016
    },
    {
      "epoch": 0.22109086906268918,
      "grad_norm": 0.7745386958122253,
      "learning_rate": 9.714476985717393e-06,
      "loss": 0.8641,
      "step": 4017
    },
    {
      "epoch": 0.22114590786504484,
      "grad_norm": 0.8112689256668091,
      "learning_rate": 9.714332585046928e-06,
      "loss": 0.834,
      "step": 4018
    },
    {
      "epoch": 0.2212009466674005,
      "grad_norm": 0.916847825050354,
      "learning_rate": 9.714188148944799e-06,
      "loss": 0.8546,
      "step": 4019
    },
    {
      "epoch": 0.22125598546975617,
      "grad_norm": 0.8595414161682129,
      "learning_rate": 9.714043677412096e-06,
      "loss": 0.9388,
      "step": 4020
    },
    {
      "epoch": 0.22131102427211183,
      "grad_norm": 0.8672438263893127,
      "learning_rate": 9.713899170449901e-06,
      "loss": 0.8151,
      "step": 4021
    },
    {
      "epoch": 0.2213660630744675,
      "grad_norm": 0.699749767780304,
      "learning_rate": 9.713754628059304e-06,
      "loss": 0.7433,
      "step": 4022
    },
    {
      "epoch": 0.22142110187682315,
      "grad_norm": 0.8071898818016052,
      "learning_rate": 9.713610050241387e-06,
      "loss": 0.7663,
      "step": 4023
    },
    {
      "epoch": 0.2214761406791788,
      "grad_norm": 0.745030403137207,
      "learning_rate": 9.713465436997239e-06,
      "loss": 0.7733,
      "step": 4024
    },
    {
      "epoch": 0.22153117948153447,
      "grad_norm": 0.8034930229187012,
      "learning_rate": 9.713320788327947e-06,
      "loss": 0.9015,
      "step": 4025
    },
    {
      "epoch": 0.22158621828389014,
      "grad_norm": 0.8549708724021912,
      "learning_rate": 9.713176104234597e-06,
      "loss": 0.7127,
      "step": 4026
    },
    {
      "epoch": 0.2216412570862458,
      "grad_norm": 0.8432256579399109,
      "learning_rate": 9.713031384718277e-06,
      "loss": 0.8163,
      "step": 4027
    },
    {
      "epoch": 0.22169629588860146,
      "grad_norm": 0.7623703479766846,
      "learning_rate": 9.712886629780075e-06,
      "loss": 0.8272,
      "step": 4028
    },
    {
      "epoch": 0.22175133469095712,
      "grad_norm": 0.8425806760787964,
      "learning_rate": 9.712741839421079e-06,
      "loss": 0.7907,
      "step": 4029
    },
    {
      "epoch": 0.22180637349331278,
      "grad_norm": 0.7477750778198242,
      "learning_rate": 9.712597013642376e-06,
      "loss": 0.7662,
      "step": 4030
    },
    {
      "epoch": 0.22186141229566844,
      "grad_norm": 0.7761805057525635,
      "learning_rate": 9.712452152445056e-06,
      "loss": 0.7999,
      "step": 4031
    },
    {
      "epoch": 0.2219164510980241,
      "grad_norm": 0.8604531288146973,
      "learning_rate": 9.712307255830207e-06,
      "loss": 0.812,
      "step": 4032
    },
    {
      "epoch": 0.22197148990037976,
      "grad_norm": 0.8113332986831665,
      "learning_rate": 9.712162323798918e-06,
      "loss": 0.8092,
      "step": 4033
    },
    {
      "epoch": 0.22202652870273543,
      "grad_norm": 0.7980128526687622,
      "learning_rate": 9.71201735635228e-06,
      "loss": 0.6934,
      "step": 4034
    },
    {
      "epoch": 0.2220815675050911,
      "grad_norm": 0.7819470763206482,
      "learning_rate": 9.711872353491377e-06,
      "loss": 0.8531,
      "step": 4035
    },
    {
      "epoch": 0.22213660630744675,
      "grad_norm": 0.8283445835113525,
      "learning_rate": 9.711727315217305e-06,
      "loss": 0.8594,
      "step": 4036
    },
    {
      "epoch": 0.2221916451098024,
      "grad_norm": 0.7282612919807434,
      "learning_rate": 9.711582241531153e-06,
      "loss": 0.7374,
      "step": 4037
    },
    {
      "epoch": 0.22224668391215807,
      "grad_norm": 0.9564353823661804,
      "learning_rate": 9.711437132434007e-06,
      "loss": 0.7996,
      "step": 4038
    },
    {
      "epoch": 0.22230172271451373,
      "grad_norm": 0.8559701442718506,
      "learning_rate": 9.711291987926963e-06,
      "loss": 0.949,
      "step": 4039
    },
    {
      "epoch": 0.2223567615168694,
      "grad_norm": 0.7515334486961365,
      "learning_rate": 9.71114680801111e-06,
      "loss": 0.7188,
      "step": 4040
    },
    {
      "epoch": 0.22241180031922506,
      "grad_norm": 0.7685608863830566,
      "learning_rate": 9.711001592687537e-06,
      "loss": 0.7679,
      "step": 4041
    },
    {
      "epoch": 0.22246683912158072,
      "grad_norm": 0.6848913431167603,
      "learning_rate": 9.710856341957337e-06,
      "loss": 0.7666,
      "step": 4042
    },
    {
      "epoch": 0.22252187792393638,
      "grad_norm": 0.7270542979240417,
      "learning_rate": 9.710711055821602e-06,
      "loss": 0.7563,
      "step": 4043
    },
    {
      "epoch": 0.22257691672629204,
      "grad_norm": 0.7965164184570312,
      "learning_rate": 9.710565734281424e-06,
      "loss": 0.7586,
      "step": 4044
    },
    {
      "epoch": 0.2226319555286477,
      "grad_norm": 0.7872949242591858,
      "learning_rate": 9.710420377337895e-06,
      "loss": 0.8423,
      "step": 4045
    },
    {
      "epoch": 0.22268699433100336,
      "grad_norm": 0.7466526627540588,
      "learning_rate": 9.710274984992107e-06,
      "loss": 0.7578,
      "step": 4046
    },
    {
      "epoch": 0.22274203313335902,
      "grad_norm": 0.7208731770515442,
      "learning_rate": 9.710129557245154e-06,
      "loss": 0.7019,
      "step": 4047
    },
    {
      "epoch": 0.22279707193571469,
      "grad_norm": 0.6953400373458862,
      "learning_rate": 9.709984094098127e-06,
      "loss": 0.7234,
      "step": 4048
    },
    {
      "epoch": 0.22285211073807035,
      "grad_norm": 0.7866283059120178,
      "learning_rate": 9.709838595552122e-06,
      "loss": 0.785,
      "step": 4049
    },
    {
      "epoch": 0.222907149540426,
      "grad_norm": 0.7404114007949829,
      "learning_rate": 9.709693061608227e-06,
      "loss": 0.7706,
      "step": 4050
    },
    {
      "epoch": 0.22296218834278167,
      "grad_norm": 0.8788254857063293,
      "learning_rate": 9.709547492267544e-06,
      "loss": 0.8392,
      "step": 4051
    },
    {
      "epoch": 0.22301722714513733,
      "grad_norm": 0.7493161559104919,
      "learning_rate": 9.70940188753116e-06,
      "loss": 0.8346,
      "step": 4052
    },
    {
      "epoch": 0.223072265947493,
      "grad_norm": 0.7340379357337952,
      "learning_rate": 9.709256247400174e-06,
      "loss": 0.7715,
      "step": 4053
    },
    {
      "epoch": 0.22312730474984865,
      "grad_norm": 0.7291178107261658,
      "learning_rate": 9.709110571875677e-06,
      "loss": 0.866,
      "step": 4054
    },
    {
      "epoch": 0.22318234355220432,
      "grad_norm": 0.8046013712882996,
      "learning_rate": 9.708964860958765e-06,
      "loss": 0.7885,
      "step": 4055
    },
    {
      "epoch": 0.22323738235455998,
      "grad_norm": 0.832941472530365,
      "learning_rate": 9.708819114650535e-06,
      "loss": 0.873,
      "step": 4056
    },
    {
      "epoch": 0.22329242115691564,
      "grad_norm": 0.6933377981185913,
      "learning_rate": 9.70867333295208e-06,
      "loss": 0.7944,
      "step": 4057
    },
    {
      "epoch": 0.2233474599592713,
      "grad_norm": 0.7976044416427612,
      "learning_rate": 9.708527515864499e-06,
      "loss": 0.72,
      "step": 4058
    },
    {
      "epoch": 0.22340249876162696,
      "grad_norm": 0.7698904871940613,
      "learning_rate": 9.708381663388884e-06,
      "loss": 0.7603,
      "step": 4059
    },
    {
      "epoch": 0.2234575375639826,
      "grad_norm": 0.7554401159286499,
      "learning_rate": 9.708235775526331e-06,
      "loss": 0.7488,
      "step": 4060
    },
    {
      "epoch": 0.22351257636633826,
      "grad_norm": 0.7382954359054565,
      "learning_rate": 9.70808985227794e-06,
      "loss": 0.7418,
      "step": 4061
    },
    {
      "epoch": 0.22356761516869392,
      "grad_norm": 0.7220499515533447,
      "learning_rate": 9.707943893644806e-06,
      "loss": 0.7691,
      "step": 4062
    },
    {
      "epoch": 0.22362265397104958,
      "grad_norm": 0.727542519569397,
      "learning_rate": 9.707797899628027e-06,
      "loss": 0.7603,
      "step": 4063
    },
    {
      "epoch": 0.22367769277340524,
      "grad_norm": 0.7857500910758972,
      "learning_rate": 9.707651870228697e-06,
      "loss": 0.8633,
      "step": 4064
    },
    {
      "epoch": 0.2237327315757609,
      "grad_norm": 0.7975600361824036,
      "learning_rate": 9.707505805447917e-06,
      "loss": 0.8591,
      "step": 4065
    },
    {
      "epoch": 0.22378777037811656,
      "grad_norm": 1.0063475370407104,
      "learning_rate": 9.707359705286784e-06,
      "loss": 0.7935,
      "step": 4066
    },
    {
      "epoch": 0.22384280918047222,
      "grad_norm": 0.7307062745094299,
      "learning_rate": 9.707213569746393e-06,
      "loss": 0.797,
      "step": 4067
    },
    {
      "epoch": 0.22389784798282789,
      "grad_norm": 0.7891914248466492,
      "learning_rate": 9.707067398827847e-06,
      "loss": 0.853,
      "step": 4068
    },
    {
      "epoch": 0.22395288678518355,
      "grad_norm": 0.7479422092437744,
      "learning_rate": 9.706921192532242e-06,
      "loss": 0.7359,
      "step": 4069
    },
    {
      "epoch": 0.2240079255875392,
      "grad_norm": 0.8436065912246704,
      "learning_rate": 9.706774950860676e-06,
      "loss": 0.7916,
      "step": 4070
    },
    {
      "epoch": 0.22406296438989487,
      "grad_norm": 0.7586960196495056,
      "learning_rate": 9.706628673814252e-06,
      "loss": 0.7871,
      "step": 4071
    },
    {
      "epoch": 0.22411800319225053,
      "grad_norm": 0.8181111812591553,
      "learning_rate": 9.706482361394064e-06,
      "loss": 0.7782,
      "step": 4072
    },
    {
      "epoch": 0.2241730419946062,
      "grad_norm": 0.7205253839492798,
      "learning_rate": 9.706336013601217e-06,
      "loss": 0.7912,
      "step": 4073
    },
    {
      "epoch": 0.22422808079696185,
      "grad_norm": 0.9823397397994995,
      "learning_rate": 9.706189630436806e-06,
      "loss": 0.8393,
      "step": 4074
    },
    {
      "epoch": 0.22428311959931752,
      "grad_norm": 0.7360854148864746,
      "learning_rate": 9.706043211901935e-06,
      "loss": 0.8239,
      "step": 4075
    },
    {
      "epoch": 0.22433815840167318,
      "grad_norm": 0.7590144872665405,
      "learning_rate": 9.705896757997701e-06,
      "loss": 0.7177,
      "step": 4076
    },
    {
      "epoch": 0.22439319720402884,
      "grad_norm": 0.7691343426704407,
      "learning_rate": 9.70575026872521e-06,
      "loss": 0.7731,
      "step": 4077
    },
    {
      "epoch": 0.2244482360063845,
      "grad_norm": 0.7057286500930786,
      "learning_rate": 9.705603744085556e-06,
      "loss": 0.7746,
      "step": 4078
    },
    {
      "epoch": 0.22450327480874016,
      "grad_norm": 0.7954769134521484,
      "learning_rate": 9.705457184079847e-06,
      "loss": 0.8215,
      "step": 4079
    },
    {
      "epoch": 0.22455831361109582,
      "grad_norm": 0.7089072465896606,
      "learning_rate": 9.70531058870918e-06,
      "loss": 0.7263,
      "step": 4080
    },
    {
      "epoch": 0.22461335241345148,
      "grad_norm": 0.9847552180290222,
      "learning_rate": 9.705163957974657e-06,
      "loss": 0.8948,
      "step": 4081
    },
    {
      "epoch": 0.22466839121580715,
      "grad_norm": 0.7977012395858765,
      "learning_rate": 9.705017291877383e-06,
      "loss": 0.7518,
      "step": 4082
    },
    {
      "epoch": 0.2247234300181628,
      "grad_norm": 0.8084518909454346,
      "learning_rate": 9.704870590418458e-06,
      "loss": 0.8711,
      "step": 4083
    },
    {
      "epoch": 0.22477846882051847,
      "grad_norm": 0.9151536822319031,
      "learning_rate": 9.704723853598986e-06,
      "loss": 0.8217,
      "step": 4084
    },
    {
      "epoch": 0.22483350762287413,
      "grad_norm": 0.908136248588562,
      "learning_rate": 9.704577081420065e-06,
      "loss": 0.6961,
      "step": 4085
    },
    {
      "epoch": 0.2248885464252298,
      "grad_norm": 0.8569996953010559,
      "learning_rate": 9.704430273882806e-06,
      "loss": 0.8405,
      "step": 4086
    },
    {
      "epoch": 0.22494358522758545,
      "grad_norm": 0.7687774300575256,
      "learning_rate": 9.704283430988307e-06,
      "loss": 0.6903,
      "step": 4087
    },
    {
      "epoch": 0.2249986240299411,
      "grad_norm": 0.863203763961792,
      "learning_rate": 9.704136552737673e-06,
      "loss": 0.8927,
      "step": 4088
    },
    {
      "epoch": 0.22505366283229677,
      "grad_norm": 1.252581238746643,
      "learning_rate": 9.703989639132008e-06,
      "loss": 0.8792,
      "step": 4089
    },
    {
      "epoch": 0.22510870163465244,
      "grad_norm": 0.7844160795211792,
      "learning_rate": 9.703842690172415e-06,
      "loss": 0.844,
      "step": 4090
    },
    {
      "epoch": 0.2251637404370081,
      "grad_norm": 0.8669766187667847,
      "learning_rate": 9.703695705860002e-06,
      "loss": 0.7008,
      "step": 4091
    },
    {
      "epoch": 0.22521877923936376,
      "grad_norm": 0.7180137634277344,
      "learning_rate": 9.703548686195869e-06,
      "loss": 0.8242,
      "step": 4092
    },
    {
      "epoch": 0.22527381804171942,
      "grad_norm": 0.7225000858306885,
      "learning_rate": 9.703401631181124e-06,
      "loss": 0.724,
      "step": 4093
    },
    {
      "epoch": 0.22532885684407508,
      "grad_norm": 0.8348065614700317,
      "learning_rate": 9.70325454081687e-06,
      "loss": 0.7996,
      "step": 4094
    },
    {
      "epoch": 0.22538389564643074,
      "grad_norm": 0.8099488019943237,
      "learning_rate": 9.703107415104216e-06,
      "loss": 0.7498,
      "step": 4095
    },
    {
      "epoch": 0.2254389344487864,
      "grad_norm": 0.7051188945770264,
      "learning_rate": 9.702960254044264e-06,
      "loss": 0.7322,
      "step": 4096
    },
    {
      "epoch": 0.22549397325114207,
      "grad_norm": 0.742859423160553,
      "learning_rate": 9.702813057638122e-06,
      "loss": 0.746,
      "step": 4097
    },
    {
      "epoch": 0.22554901205349773,
      "grad_norm": 0.7981536984443665,
      "learning_rate": 9.702665825886897e-06,
      "loss": 0.8705,
      "step": 4098
    },
    {
      "epoch": 0.2256040508558534,
      "grad_norm": 1.0317178964614868,
      "learning_rate": 9.702518558791693e-06,
      "loss": 0.8261,
      "step": 4099
    },
    {
      "epoch": 0.22565908965820905,
      "grad_norm": 0.7811983823776245,
      "learning_rate": 9.702371256353618e-06,
      "loss": 0.7633,
      "step": 4100
    },
    {
      "epoch": 0.2257141284605647,
      "grad_norm": 0.8288078308105469,
      "learning_rate": 9.702223918573782e-06,
      "loss": 0.7974,
      "step": 4101
    },
    {
      "epoch": 0.22576916726292034,
      "grad_norm": 0.8932577967643738,
      "learning_rate": 9.702076545453286e-06,
      "loss": 0.7517,
      "step": 4102
    },
    {
      "epoch": 0.225824206065276,
      "grad_norm": 0.8342248201370239,
      "learning_rate": 9.701929136993243e-06,
      "loss": 0.8634,
      "step": 4103
    },
    {
      "epoch": 0.22587924486763167,
      "grad_norm": 0.790392279624939,
      "learning_rate": 9.701781693194761e-06,
      "loss": 0.7705,
      "step": 4104
    },
    {
      "epoch": 0.22593428366998733,
      "grad_norm": 0.824691891670227,
      "learning_rate": 9.701634214058944e-06,
      "loss": 0.877,
      "step": 4105
    },
    {
      "epoch": 0.225989322472343,
      "grad_norm": 0.9237051010131836,
      "learning_rate": 9.701486699586904e-06,
      "loss": 0.842,
      "step": 4106
    },
    {
      "epoch": 0.22604436127469865,
      "grad_norm": 0.7453535199165344,
      "learning_rate": 9.701339149779747e-06,
      "loss": 0.8217,
      "step": 4107
    },
    {
      "epoch": 0.2260994000770543,
      "grad_norm": 0.727872371673584,
      "learning_rate": 9.701191564638586e-06,
      "loss": 0.849,
      "step": 4108
    },
    {
      "epoch": 0.22615443887940997,
      "grad_norm": 0.966585636138916,
      "learning_rate": 9.701043944164526e-06,
      "loss": 0.7742,
      "step": 4109
    },
    {
      "epoch": 0.22620947768176564,
      "grad_norm": 0.7556117177009583,
      "learning_rate": 9.700896288358678e-06,
      "loss": 0.7498,
      "step": 4110
    },
    {
      "epoch": 0.2262645164841213,
      "grad_norm": 0.848143458366394,
      "learning_rate": 9.700748597222151e-06,
      "loss": 0.7237,
      "step": 4111
    },
    {
      "epoch": 0.22631955528647696,
      "grad_norm": 0.9046787619590759,
      "learning_rate": 9.700600870756056e-06,
      "loss": 0.8066,
      "step": 4112
    },
    {
      "epoch": 0.22637459408883262,
      "grad_norm": 0.923159658908844,
      "learning_rate": 9.700453108961505e-06,
      "loss": 0.8404,
      "step": 4113
    },
    {
      "epoch": 0.22642963289118828,
      "grad_norm": 0.8697664737701416,
      "learning_rate": 9.700305311839606e-06,
      "loss": 0.7269,
      "step": 4114
    },
    {
      "epoch": 0.22648467169354394,
      "grad_norm": 0.8179994821548462,
      "learning_rate": 9.70015747939147e-06,
      "loss": 0.8083,
      "step": 4115
    },
    {
      "epoch": 0.2265397104958996,
      "grad_norm": 0.7961694002151489,
      "learning_rate": 9.700009611618208e-06,
      "loss": 0.7327,
      "step": 4116
    },
    {
      "epoch": 0.22659474929825527,
      "grad_norm": 0.7317802309989929,
      "learning_rate": 9.699861708520934e-06,
      "loss": 0.8273,
      "step": 4117
    },
    {
      "epoch": 0.22664978810061093,
      "grad_norm": 0.9190557599067688,
      "learning_rate": 9.699713770100757e-06,
      "loss": 0.8027,
      "step": 4118
    },
    {
      "epoch": 0.2267048269029666,
      "grad_norm": 0.7618072628974915,
      "learning_rate": 9.699565796358788e-06,
      "loss": 0.7669,
      "step": 4119
    },
    {
      "epoch": 0.22675986570532225,
      "grad_norm": 1.0236154794692993,
      "learning_rate": 9.699417787296139e-06,
      "loss": 0.7511,
      "step": 4120
    },
    {
      "epoch": 0.2268149045076779,
      "grad_norm": 0.8011670708656311,
      "learning_rate": 9.699269742913927e-06,
      "loss": 0.7644,
      "step": 4121
    },
    {
      "epoch": 0.22686994331003357,
      "grad_norm": 0.7808024287223816,
      "learning_rate": 9.69912166321326e-06,
      "loss": 0.7894,
      "step": 4122
    },
    {
      "epoch": 0.22692498211238923,
      "grad_norm": 0.8645655512809753,
      "learning_rate": 9.698973548195252e-06,
      "loss": 0.7989,
      "step": 4123
    },
    {
      "epoch": 0.2269800209147449,
      "grad_norm": 0.7478770613670349,
      "learning_rate": 9.698825397861017e-06,
      "loss": 0.7758,
      "step": 4124
    },
    {
      "epoch": 0.22703505971710056,
      "grad_norm": 0.8988361954689026,
      "learning_rate": 9.698677212211668e-06,
      "loss": 0.8312,
      "step": 4125
    },
    {
      "epoch": 0.22709009851945622,
      "grad_norm": 0.773028552532196,
      "learning_rate": 9.69852899124832e-06,
      "loss": 0.7415,
      "step": 4126
    },
    {
      "epoch": 0.22714513732181188,
      "grad_norm": 0.8173778653144836,
      "learning_rate": 9.698380734972085e-06,
      "loss": 0.8241,
      "step": 4127
    },
    {
      "epoch": 0.22720017612416754,
      "grad_norm": 0.7868672013282776,
      "learning_rate": 9.698232443384078e-06,
      "loss": 0.7294,
      "step": 4128
    },
    {
      "epoch": 0.2272552149265232,
      "grad_norm": 0.8662189841270447,
      "learning_rate": 9.698084116485413e-06,
      "loss": 0.9307,
      "step": 4129
    },
    {
      "epoch": 0.22731025372887886,
      "grad_norm": 0.7571321129798889,
      "learning_rate": 9.697935754277207e-06,
      "loss": 0.7756,
      "step": 4130
    },
    {
      "epoch": 0.22736529253123453,
      "grad_norm": 0.8222649097442627,
      "learning_rate": 9.697787356760574e-06,
      "loss": 0.8689,
      "step": 4131
    },
    {
      "epoch": 0.2274203313335902,
      "grad_norm": 0.8302241563796997,
      "learning_rate": 9.697638923936626e-06,
      "loss": 0.8139,
      "step": 4132
    },
    {
      "epoch": 0.22747537013594585,
      "grad_norm": 0.779951274394989,
      "learning_rate": 9.697490455806482e-06,
      "loss": 0.7493,
      "step": 4133
    },
    {
      "epoch": 0.2275304089383015,
      "grad_norm": 0.8409813046455383,
      "learning_rate": 9.697341952371257e-06,
      "loss": 0.777,
      "step": 4134
    },
    {
      "epoch": 0.22758544774065717,
      "grad_norm": 0.8599729537963867,
      "learning_rate": 9.697193413632068e-06,
      "loss": 0.7678,
      "step": 4135
    },
    {
      "epoch": 0.22764048654301283,
      "grad_norm": 0.7505115270614624,
      "learning_rate": 9.69704483959003e-06,
      "loss": 0.787,
      "step": 4136
    },
    {
      "epoch": 0.2276955253453685,
      "grad_norm": 0.7326868176460266,
      "learning_rate": 9.696896230246262e-06,
      "loss": 0.7066,
      "step": 4137
    },
    {
      "epoch": 0.22775056414772415,
      "grad_norm": 0.8269753456115723,
      "learning_rate": 9.696747585601878e-06,
      "loss": 0.7379,
      "step": 4138
    },
    {
      "epoch": 0.22780560295007982,
      "grad_norm": 0.7841970324516296,
      "learning_rate": 9.696598905657997e-06,
      "loss": 0.764,
      "step": 4139
    },
    {
      "epoch": 0.22786064175243548,
      "grad_norm": 0.7131417989730835,
      "learning_rate": 9.696450190415735e-06,
      "loss": 0.7629,
      "step": 4140
    },
    {
      "epoch": 0.22791568055479114,
      "grad_norm": 0.7922703623771667,
      "learning_rate": 9.69630143987621e-06,
      "loss": 0.8354,
      "step": 4141
    },
    {
      "epoch": 0.2279707193571468,
      "grad_norm": 0.9628629684448242,
      "learning_rate": 9.696152654040543e-06,
      "loss": 0.8077,
      "step": 4142
    },
    {
      "epoch": 0.22802575815950246,
      "grad_norm": 0.8566663265228271,
      "learning_rate": 9.696003832909847e-06,
      "loss": 0.685,
      "step": 4143
    },
    {
      "epoch": 0.22808079696185812,
      "grad_norm": 0.7181339859962463,
      "learning_rate": 9.695854976485244e-06,
      "loss": 0.8135,
      "step": 4144
    },
    {
      "epoch": 0.22813583576421376,
      "grad_norm": 0.9119813442230225,
      "learning_rate": 9.695706084767853e-06,
      "loss": 0.7276,
      "step": 4145
    },
    {
      "epoch": 0.22819087456656942,
      "grad_norm": 0.8547400832176208,
      "learning_rate": 9.69555715775879e-06,
      "loss": 0.8656,
      "step": 4146
    },
    {
      "epoch": 0.22824591336892508,
      "grad_norm": 0.77585768699646,
      "learning_rate": 9.695408195459179e-06,
      "loss": 0.8218,
      "step": 4147
    },
    {
      "epoch": 0.22830095217128074,
      "grad_norm": 0.7832447290420532,
      "learning_rate": 9.695259197870135e-06,
      "loss": 0.8002,
      "step": 4148
    },
    {
      "epoch": 0.2283559909736364,
      "grad_norm": 0.9184865355491638,
      "learning_rate": 9.69511016499278e-06,
      "loss": 0.8651,
      "step": 4149
    },
    {
      "epoch": 0.22841102977599206,
      "grad_norm": 0.8663797974586487,
      "learning_rate": 9.694961096828235e-06,
      "loss": 0.7381,
      "step": 4150
    },
    {
      "epoch": 0.22846606857834773,
      "grad_norm": 0.843265950679779,
      "learning_rate": 9.694811993377617e-06,
      "loss": 0.8546,
      "step": 4151
    },
    {
      "epoch": 0.2285211073807034,
      "grad_norm": 0.8021818399429321,
      "learning_rate": 9.694662854642049e-06,
      "loss": 0.9166,
      "step": 4152
    },
    {
      "epoch": 0.22857614618305905,
      "grad_norm": 0.7762879729270935,
      "learning_rate": 9.694513680622653e-06,
      "loss": 0.7055,
      "step": 4153
    },
    {
      "epoch": 0.2286311849854147,
      "grad_norm": 0.809352457523346,
      "learning_rate": 9.694364471320548e-06,
      "loss": 0.7988,
      "step": 4154
    },
    {
      "epoch": 0.22868622378777037,
      "grad_norm": 0.7239902019500732,
      "learning_rate": 9.694215226736858e-06,
      "loss": 0.7783,
      "step": 4155
    },
    {
      "epoch": 0.22874126259012603,
      "grad_norm": 0.7072625160217285,
      "learning_rate": 9.694065946872702e-06,
      "loss": 0.7607,
      "step": 4156
    },
    {
      "epoch": 0.2287963013924817,
      "grad_norm": 0.7696169018745422,
      "learning_rate": 9.693916631729201e-06,
      "loss": 0.7519,
      "step": 4157
    },
    {
      "epoch": 0.22885134019483735,
      "grad_norm": 0.9198557734489441,
      "learning_rate": 9.69376728130748e-06,
      "loss": 0.7754,
      "step": 4158
    },
    {
      "epoch": 0.22890637899719302,
      "grad_norm": 0.7589097619056702,
      "learning_rate": 9.693617895608662e-06,
      "loss": 0.7258,
      "step": 4159
    },
    {
      "epoch": 0.22896141779954868,
      "grad_norm": 0.8351333141326904,
      "learning_rate": 9.693468474633867e-06,
      "loss": 0.8633,
      "step": 4160
    },
    {
      "epoch": 0.22901645660190434,
      "grad_norm": 0.8331828713417053,
      "learning_rate": 9.69331901838422e-06,
      "loss": 0.7361,
      "step": 4161
    },
    {
      "epoch": 0.22907149540426,
      "grad_norm": 0.8810774087905884,
      "learning_rate": 9.693169526860843e-06,
      "loss": 0.7651,
      "step": 4162
    },
    {
      "epoch": 0.22912653420661566,
      "grad_norm": 0.8151684999465942,
      "learning_rate": 9.69302000006486e-06,
      "loss": 0.8533,
      "step": 4163
    },
    {
      "epoch": 0.22918157300897132,
      "grad_norm": 0.8683320879936218,
      "learning_rate": 9.692870437997394e-06,
      "loss": 0.8323,
      "step": 4164
    },
    {
      "epoch": 0.22923661181132698,
      "grad_norm": 0.7488875389099121,
      "learning_rate": 9.692720840659572e-06,
      "loss": 0.8414,
      "step": 4165
    },
    {
      "epoch": 0.22929165061368265,
      "grad_norm": 0.7916452288627625,
      "learning_rate": 9.692571208052515e-06,
      "loss": 0.7058,
      "step": 4166
    },
    {
      "epoch": 0.2293466894160383,
      "grad_norm": 0.8228384256362915,
      "learning_rate": 9.69242154017735e-06,
      "loss": 0.7667,
      "step": 4167
    },
    {
      "epoch": 0.22940172821839397,
      "grad_norm": 0.7395613789558411,
      "learning_rate": 9.692271837035202e-06,
      "loss": 0.7649,
      "step": 4168
    },
    {
      "epoch": 0.22945676702074963,
      "grad_norm": 0.7187666893005371,
      "learning_rate": 9.692122098627192e-06,
      "loss": 0.7575,
      "step": 4169
    },
    {
      "epoch": 0.2295118058231053,
      "grad_norm": 0.7060030102729797,
      "learning_rate": 9.691972324954449e-06,
      "loss": 0.8309,
      "step": 4170
    },
    {
      "epoch": 0.22956684462546095,
      "grad_norm": 0.7807210087776184,
      "learning_rate": 9.691822516018099e-06,
      "loss": 0.8185,
      "step": 4171
    },
    {
      "epoch": 0.22962188342781661,
      "grad_norm": 0.6918593645095825,
      "learning_rate": 9.691672671819265e-06,
      "loss": 0.6983,
      "step": 4172
    },
    {
      "epoch": 0.22967692223017228,
      "grad_norm": 0.7947858572006226,
      "learning_rate": 9.691522792359077e-06,
      "loss": 0.8098,
      "step": 4173
    },
    {
      "epoch": 0.22973196103252794,
      "grad_norm": 0.7907306551933289,
      "learning_rate": 9.691372877638658e-06,
      "loss": 0.8,
      "step": 4174
    },
    {
      "epoch": 0.2297869998348836,
      "grad_norm": 0.7669435739517212,
      "learning_rate": 9.691222927659137e-06,
      "loss": 0.8121,
      "step": 4175
    },
    {
      "epoch": 0.22984203863723926,
      "grad_norm": 0.8128299117088318,
      "learning_rate": 9.691072942421642e-06,
      "loss": 0.7554,
      "step": 4176
    },
    {
      "epoch": 0.22989707743959492,
      "grad_norm": 0.9043960571289062,
      "learning_rate": 9.690922921927295e-06,
      "loss": 0.8601,
      "step": 4177
    },
    {
      "epoch": 0.22995211624195058,
      "grad_norm": 0.835445761680603,
      "learning_rate": 9.690772866177229e-06,
      "loss": 0.8185,
      "step": 4178
    },
    {
      "epoch": 0.23000715504430624,
      "grad_norm": 0.734601616859436,
      "learning_rate": 9.69062277517257e-06,
      "loss": 0.6486,
      "step": 4179
    },
    {
      "epoch": 0.2300621938466619,
      "grad_norm": 0.8252671957015991,
      "learning_rate": 9.690472648914445e-06,
      "loss": 0.8455,
      "step": 4180
    },
    {
      "epoch": 0.23011723264901757,
      "grad_norm": 0.8266329169273376,
      "learning_rate": 9.690322487403984e-06,
      "loss": 0.7348,
      "step": 4181
    },
    {
      "epoch": 0.23017227145137323,
      "grad_norm": 0.8280256390571594,
      "learning_rate": 9.690172290642314e-06,
      "loss": 0.8191,
      "step": 4182
    },
    {
      "epoch": 0.2302273102537289,
      "grad_norm": 0.8854276537895203,
      "learning_rate": 9.690022058630564e-06,
      "loss": 0.9327,
      "step": 4183
    },
    {
      "epoch": 0.23028234905608455,
      "grad_norm": 0.7308807969093323,
      "learning_rate": 9.689871791369865e-06,
      "loss": 0.8144,
      "step": 4184
    },
    {
      "epoch": 0.2303373878584402,
      "grad_norm": 0.7171719670295715,
      "learning_rate": 9.689721488861344e-06,
      "loss": 0.8265,
      "step": 4185
    },
    {
      "epoch": 0.23039242666079587,
      "grad_norm": 0.7955548763275146,
      "learning_rate": 9.689571151106131e-06,
      "loss": 0.7313,
      "step": 4186
    },
    {
      "epoch": 0.23044746546315154,
      "grad_norm": 0.8218876123428345,
      "learning_rate": 9.689420778105359e-06,
      "loss": 0.883,
      "step": 4187
    },
    {
      "epoch": 0.23050250426550717,
      "grad_norm": 0.79570072889328,
      "learning_rate": 9.689270369860154e-06,
      "loss": 0.8898,
      "step": 4188
    },
    {
      "epoch": 0.23055754306786283,
      "grad_norm": 0.8163344264030457,
      "learning_rate": 9.689119926371649e-06,
      "loss": 0.8638,
      "step": 4189
    },
    {
      "epoch": 0.2306125818702185,
      "grad_norm": 0.7767764329910278,
      "learning_rate": 9.688969447640972e-06,
      "loss": 0.7822,
      "step": 4190
    },
    {
      "epoch": 0.23066762067257415,
      "grad_norm": 0.9357114434242249,
      "learning_rate": 9.688818933669258e-06,
      "loss": 0.8031,
      "step": 4191
    },
    {
      "epoch": 0.23072265947492981,
      "grad_norm": 0.8340080380439758,
      "learning_rate": 9.688668384457635e-06,
      "loss": 0.8947,
      "step": 4192
    },
    {
      "epoch": 0.23077769827728548,
      "grad_norm": 0.8187471628189087,
      "learning_rate": 9.688517800007235e-06,
      "loss": 0.7989,
      "step": 4193
    },
    {
      "epoch": 0.23083273707964114,
      "grad_norm": 0.8131871819496155,
      "learning_rate": 9.688367180319191e-06,
      "loss": 0.8377,
      "step": 4194
    },
    {
      "epoch": 0.2308877758819968,
      "grad_norm": 0.7933448553085327,
      "learning_rate": 9.688216525394634e-06,
      "loss": 0.8723,
      "step": 4195
    },
    {
      "epoch": 0.23094281468435246,
      "grad_norm": 0.7262325286865234,
      "learning_rate": 9.688065835234695e-06,
      "loss": 0.7802,
      "step": 4196
    },
    {
      "epoch": 0.23099785348670812,
      "grad_norm": 0.8289293050765991,
      "learning_rate": 9.68791510984051e-06,
      "loss": 0.642,
      "step": 4197
    },
    {
      "epoch": 0.23105289228906378,
      "grad_norm": 0.8835988640785217,
      "learning_rate": 9.687764349213211e-06,
      "loss": 0.9002,
      "step": 4198
    },
    {
      "epoch": 0.23110793109141944,
      "grad_norm": 0.9478649497032166,
      "learning_rate": 9.687613553353927e-06,
      "loss": 0.8668,
      "step": 4199
    },
    {
      "epoch": 0.2311629698937751,
      "grad_norm": 0.872936487197876,
      "learning_rate": 9.687462722263796e-06,
      "loss": 0.8312,
      "step": 4200
    },
    {
      "epoch": 0.23121800869613077,
      "grad_norm": 0.7073879241943359,
      "learning_rate": 9.68731185594395e-06,
      "loss": 0.776,
      "step": 4201
    },
    {
      "epoch": 0.23127304749848643,
      "grad_norm": 0.8265218734741211,
      "learning_rate": 9.687160954395522e-06,
      "loss": 0.8152,
      "step": 4202
    },
    {
      "epoch": 0.2313280863008421,
      "grad_norm": 0.8027207255363464,
      "learning_rate": 9.687010017619649e-06,
      "loss": 0.9514,
      "step": 4203
    },
    {
      "epoch": 0.23138312510319775,
      "grad_norm": 0.7416790127754211,
      "learning_rate": 9.68685904561746e-06,
      "loss": 0.7708,
      "step": 4204
    },
    {
      "epoch": 0.2314381639055534,
      "grad_norm": 0.7916150689125061,
      "learning_rate": 9.686708038390096e-06,
      "loss": 0.7753,
      "step": 4205
    },
    {
      "epoch": 0.23149320270790907,
      "grad_norm": 0.7213300466537476,
      "learning_rate": 9.686556995938688e-06,
      "loss": 0.83,
      "step": 4206
    },
    {
      "epoch": 0.23154824151026474,
      "grad_norm": 0.7595892548561096,
      "learning_rate": 9.68640591826437e-06,
      "loss": 0.8186,
      "step": 4207
    },
    {
      "epoch": 0.2316032803126204,
      "grad_norm": 0.7042104601860046,
      "learning_rate": 9.686254805368282e-06,
      "loss": 0.7126,
      "step": 4208
    },
    {
      "epoch": 0.23165831911497606,
      "grad_norm": 0.7416805028915405,
      "learning_rate": 9.686103657251558e-06,
      "loss": 0.7791,
      "step": 4209
    },
    {
      "epoch": 0.23171335791733172,
      "grad_norm": 0.9868568181991577,
      "learning_rate": 9.685952473915333e-06,
      "loss": 0.8453,
      "step": 4210
    },
    {
      "epoch": 0.23176839671968738,
      "grad_norm": 0.7133191823959351,
      "learning_rate": 9.68580125536074e-06,
      "loss": 0.6061,
      "step": 4211
    },
    {
      "epoch": 0.23182343552204304,
      "grad_norm": 0.8307366967201233,
      "learning_rate": 9.685650001588921e-06,
      "loss": 0.8403,
      "step": 4212
    },
    {
      "epoch": 0.2318784743243987,
      "grad_norm": 0.8395226001739502,
      "learning_rate": 9.685498712601014e-06,
      "loss": 0.7945,
      "step": 4213
    },
    {
      "epoch": 0.23193351312675436,
      "grad_norm": 0.7557219862937927,
      "learning_rate": 9.68534738839815e-06,
      "loss": 0.7765,
      "step": 4214
    },
    {
      "epoch": 0.23198855192911003,
      "grad_norm": 0.7003554105758667,
      "learning_rate": 9.68519602898147e-06,
      "loss": 0.7228,
      "step": 4215
    },
    {
      "epoch": 0.2320435907314657,
      "grad_norm": 0.8422999382019043,
      "learning_rate": 9.68504463435211e-06,
      "loss": 0.8524,
      "step": 4216
    },
    {
      "epoch": 0.23209862953382135,
      "grad_norm": 0.9369016289710999,
      "learning_rate": 9.68489320451121e-06,
      "loss": 0.7646,
      "step": 4217
    },
    {
      "epoch": 0.232153668336177,
      "grad_norm": 0.8456607460975647,
      "learning_rate": 9.684741739459905e-06,
      "loss": 0.7481,
      "step": 4218
    },
    {
      "epoch": 0.23220870713853267,
      "grad_norm": 0.9284812211990356,
      "learning_rate": 9.684590239199336e-06,
      "loss": 0.8192,
      "step": 4219
    },
    {
      "epoch": 0.23226374594088833,
      "grad_norm": 0.8474242687225342,
      "learning_rate": 9.68443870373064e-06,
      "loss": 0.7143,
      "step": 4220
    },
    {
      "epoch": 0.232318784743244,
      "grad_norm": 0.8259334564208984,
      "learning_rate": 9.684287133054957e-06,
      "loss": 0.8667,
      "step": 4221
    },
    {
      "epoch": 0.23237382354559966,
      "grad_norm": 0.8016416430473328,
      "learning_rate": 9.684135527173427e-06,
      "loss": 0.8694,
      "step": 4222
    },
    {
      "epoch": 0.23242886234795532,
      "grad_norm": 0.7575937509536743,
      "learning_rate": 9.683983886087186e-06,
      "loss": 0.7591,
      "step": 4223
    },
    {
      "epoch": 0.23248390115031098,
      "grad_norm": 0.7004683613777161,
      "learning_rate": 9.683832209797377e-06,
      "loss": 0.739,
      "step": 4224
    },
    {
      "epoch": 0.23253893995266664,
      "grad_norm": 0.8265832662582397,
      "learning_rate": 9.68368049830514e-06,
      "loss": 0.7705,
      "step": 4225
    },
    {
      "epoch": 0.2325939787550223,
      "grad_norm": 0.7705711722373962,
      "learning_rate": 9.683528751611612e-06,
      "loss": 0.7896,
      "step": 4226
    },
    {
      "epoch": 0.23264901755737796,
      "grad_norm": 0.7426978349685669,
      "learning_rate": 9.683376969717937e-06,
      "loss": 0.8217,
      "step": 4227
    },
    {
      "epoch": 0.23270405635973362,
      "grad_norm": 0.7425839304924011,
      "learning_rate": 9.683225152625255e-06,
      "loss": 0.7426,
      "step": 4228
    },
    {
      "epoch": 0.23275909516208929,
      "grad_norm": 1.0415440797805786,
      "learning_rate": 9.683073300334705e-06,
      "loss": 0.8585,
      "step": 4229
    },
    {
      "epoch": 0.23281413396444495,
      "grad_norm": 0.7706055045127869,
      "learning_rate": 9.68292141284743e-06,
      "loss": 0.8349,
      "step": 4230
    },
    {
      "epoch": 0.23286917276680058,
      "grad_norm": 0.8407607674598694,
      "learning_rate": 9.682769490164572e-06,
      "loss": 0.8592,
      "step": 4231
    },
    {
      "epoch": 0.23292421156915624,
      "grad_norm": 0.6830767393112183,
      "learning_rate": 9.68261753228727e-06,
      "loss": 0.6773,
      "step": 4232
    },
    {
      "epoch": 0.2329792503715119,
      "grad_norm": 1.6661429405212402,
      "learning_rate": 9.68246553921667e-06,
      "loss": 1.005,
      "step": 4233
    },
    {
      "epoch": 0.23303428917386756,
      "grad_norm": 0.7677092552185059,
      "learning_rate": 9.682313510953912e-06,
      "loss": 0.7689,
      "step": 4234
    },
    {
      "epoch": 0.23308932797622323,
      "grad_norm": 0.7232248187065125,
      "learning_rate": 9.682161447500139e-06,
      "loss": 0.7765,
      "step": 4235
    },
    {
      "epoch": 0.2331443667785789,
      "grad_norm": 0.8667388558387756,
      "learning_rate": 9.682009348856494e-06,
      "loss": 0.8099,
      "step": 4236
    },
    {
      "epoch": 0.23319940558093455,
      "grad_norm": 0.8220446705818176,
      "learning_rate": 9.68185721502412e-06,
      "loss": 0.8078,
      "step": 4237
    },
    {
      "epoch": 0.2332544443832902,
      "grad_norm": 0.9670323133468628,
      "learning_rate": 9.68170504600416e-06,
      "loss": 0.8912,
      "step": 4238
    },
    {
      "epoch": 0.23330948318564587,
      "grad_norm": 0.7950771450996399,
      "learning_rate": 9.68155284179776e-06,
      "loss": 0.8165,
      "step": 4239
    },
    {
      "epoch": 0.23336452198800153,
      "grad_norm": 0.7606233358383179,
      "learning_rate": 9.68140060240606e-06,
      "loss": 0.7795,
      "step": 4240
    },
    {
      "epoch": 0.2334195607903572,
      "grad_norm": 0.9580656886100769,
      "learning_rate": 9.681248327830205e-06,
      "loss": 0.7949,
      "step": 4241
    },
    {
      "epoch": 0.23347459959271286,
      "grad_norm": 0.6878347992897034,
      "learning_rate": 9.681096018071341e-06,
      "loss": 0.7776,
      "step": 4242
    },
    {
      "epoch": 0.23352963839506852,
      "grad_norm": 0.8449816107749939,
      "learning_rate": 9.680943673130614e-06,
      "loss": 0.8456,
      "step": 4243
    },
    {
      "epoch": 0.23358467719742418,
      "grad_norm": 0.77314692735672,
      "learning_rate": 9.680791293009167e-06,
      "loss": 0.7915,
      "step": 4244
    },
    {
      "epoch": 0.23363971599977984,
      "grad_norm": 0.8034142255783081,
      "learning_rate": 9.680638877708146e-06,
      "loss": 0.7377,
      "step": 4245
    },
    {
      "epoch": 0.2336947548021355,
      "grad_norm": 0.8754952549934387,
      "learning_rate": 9.680486427228695e-06,
      "loss": 0.8072,
      "step": 4246
    },
    {
      "epoch": 0.23374979360449116,
      "grad_norm": 0.8169820308685303,
      "learning_rate": 9.680333941571963e-06,
      "loss": 0.8253,
      "step": 4247
    },
    {
      "epoch": 0.23380483240684682,
      "grad_norm": 0.7848341464996338,
      "learning_rate": 9.680181420739092e-06,
      "loss": 0.8243,
      "step": 4248
    },
    {
      "epoch": 0.23385987120920249,
      "grad_norm": 0.7599799036979675,
      "learning_rate": 9.68002886473123e-06,
      "loss": 0.781,
      "step": 4249
    },
    {
      "epoch": 0.23391491001155815,
      "grad_norm": 0.8920254707336426,
      "learning_rate": 9.679876273549524e-06,
      "loss": 0.8199,
      "step": 4250
    },
    {
      "epoch": 0.2339699488139138,
      "grad_norm": 0.7813586592674255,
      "learning_rate": 9.679723647195121e-06,
      "loss": 0.7758,
      "step": 4251
    },
    {
      "epoch": 0.23402498761626947,
      "grad_norm": 0.735282838344574,
      "learning_rate": 9.679570985669168e-06,
      "loss": 0.7651,
      "step": 4252
    },
    {
      "epoch": 0.23408002641862513,
      "grad_norm": 0.7305853962898254,
      "learning_rate": 9.679418288972813e-06,
      "loss": 0.8202,
      "step": 4253
    },
    {
      "epoch": 0.2341350652209808,
      "grad_norm": 0.8331005573272705,
      "learning_rate": 9.6792655571072e-06,
      "loss": 0.8784,
      "step": 4254
    },
    {
      "epoch": 0.23419010402333645,
      "grad_norm": 0.8526305556297302,
      "learning_rate": 9.679112790073481e-06,
      "loss": 0.8116,
      "step": 4255
    },
    {
      "epoch": 0.23424514282569212,
      "grad_norm": 0.741073489189148,
      "learning_rate": 9.678959987872805e-06,
      "loss": 0.6928,
      "step": 4256
    },
    {
      "epoch": 0.23430018162804778,
      "grad_norm": 0.727859616279602,
      "learning_rate": 9.678807150506315e-06,
      "loss": 0.7571,
      "step": 4257
    },
    {
      "epoch": 0.23435522043040344,
      "grad_norm": 0.8890698552131653,
      "learning_rate": 9.678654277975165e-06,
      "loss": 0.8145,
      "step": 4258
    },
    {
      "epoch": 0.2344102592327591,
      "grad_norm": 0.7372937798500061,
      "learning_rate": 9.6785013702805e-06,
      "loss": 0.7104,
      "step": 4259
    },
    {
      "epoch": 0.23446529803511476,
      "grad_norm": 0.7205008268356323,
      "learning_rate": 9.678348427423472e-06,
      "loss": 0.7498,
      "step": 4260
    },
    {
      "epoch": 0.23452033683747042,
      "grad_norm": 0.7766392230987549,
      "learning_rate": 9.67819544940523e-06,
      "loss": 0.7814,
      "step": 4261
    },
    {
      "epoch": 0.23457537563982608,
      "grad_norm": 0.7441498637199402,
      "learning_rate": 9.678042436226922e-06,
      "loss": 0.7429,
      "step": 4262
    },
    {
      "epoch": 0.23463041444218175,
      "grad_norm": 0.8838522434234619,
      "learning_rate": 9.677889387889701e-06,
      "loss": 0.8719,
      "step": 4263
    },
    {
      "epoch": 0.2346854532445374,
      "grad_norm": 1.2349655628204346,
      "learning_rate": 9.677736304394716e-06,
      "loss": 0.8491,
      "step": 4264
    },
    {
      "epoch": 0.23474049204689307,
      "grad_norm": 0.8050087690353394,
      "learning_rate": 9.677583185743116e-06,
      "loss": 0.795,
      "step": 4265
    },
    {
      "epoch": 0.23479553084924873,
      "grad_norm": 0.7885709404945374,
      "learning_rate": 9.677430031936051e-06,
      "loss": 0.8594,
      "step": 4266
    },
    {
      "epoch": 0.2348505696516044,
      "grad_norm": 0.7753557562828064,
      "learning_rate": 9.677276842974676e-06,
      "loss": 0.8196,
      "step": 4267
    },
    {
      "epoch": 0.23490560845396005,
      "grad_norm": 0.7325392961502075,
      "learning_rate": 9.67712361886014e-06,
      "loss": 0.7905,
      "step": 4268
    },
    {
      "epoch": 0.2349606472563157,
      "grad_norm": 0.7925617694854736,
      "learning_rate": 9.676970359593594e-06,
      "loss": 0.7416,
      "step": 4269
    },
    {
      "epoch": 0.23501568605867137,
      "grad_norm": 0.7981371283531189,
      "learning_rate": 9.676817065176192e-06,
      "loss": 0.81,
      "step": 4270
    },
    {
      "epoch": 0.23507072486102704,
      "grad_norm": 0.7490524053573608,
      "learning_rate": 9.676663735609084e-06,
      "loss": 0.8347,
      "step": 4271
    },
    {
      "epoch": 0.2351257636633827,
      "grad_norm": 1.000349521636963,
      "learning_rate": 9.676510370893424e-06,
      "loss": 0.7469,
      "step": 4272
    },
    {
      "epoch": 0.23518080246573836,
      "grad_norm": 0.9310774207115173,
      "learning_rate": 9.676356971030364e-06,
      "loss": 0.8088,
      "step": 4273
    },
    {
      "epoch": 0.235235841268094,
      "grad_norm": 0.8868544101715088,
      "learning_rate": 9.676203536021055e-06,
      "loss": 0.7472,
      "step": 4274
    },
    {
      "epoch": 0.23529088007044965,
      "grad_norm": 0.7702255845069885,
      "learning_rate": 9.676050065866653e-06,
      "loss": 0.8395,
      "step": 4275
    },
    {
      "epoch": 0.23534591887280532,
      "grad_norm": 0.7138833999633789,
      "learning_rate": 9.675896560568311e-06,
      "loss": 0.8529,
      "step": 4276
    },
    {
      "epoch": 0.23540095767516098,
      "grad_norm": 0.8399729132652283,
      "learning_rate": 9.675743020127182e-06,
      "loss": 0.7844,
      "step": 4277
    },
    {
      "epoch": 0.23545599647751664,
      "grad_norm": 0.8500726819038391,
      "learning_rate": 9.67558944454442e-06,
      "loss": 0.8209,
      "step": 4278
    },
    {
      "epoch": 0.2355110352798723,
      "grad_norm": 0.766638994216919,
      "learning_rate": 9.675435833821178e-06,
      "loss": 0.7834,
      "step": 4279
    },
    {
      "epoch": 0.23556607408222796,
      "grad_norm": 0.9121370315551758,
      "learning_rate": 9.675282187958613e-06,
      "loss": 0.8697,
      "step": 4280
    },
    {
      "epoch": 0.23562111288458362,
      "grad_norm": 0.7862319946289062,
      "learning_rate": 9.675128506957879e-06,
      "loss": 0.8262,
      "step": 4281
    },
    {
      "epoch": 0.23567615168693928,
      "grad_norm": 1.072777509689331,
      "learning_rate": 9.67497479082013e-06,
      "loss": 0.7963,
      "step": 4282
    },
    {
      "epoch": 0.23573119048929495,
      "grad_norm": 0.7574695944786072,
      "learning_rate": 9.67482103954652e-06,
      "loss": 0.8178,
      "step": 4283
    },
    {
      "epoch": 0.2357862292916506,
      "grad_norm": 0.7996877431869507,
      "learning_rate": 9.674667253138209e-06,
      "loss": 0.8465,
      "step": 4284
    },
    {
      "epoch": 0.23584126809400627,
      "grad_norm": 0.711513340473175,
      "learning_rate": 9.674513431596349e-06,
      "loss": 0.7445,
      "step": 4285
    },
    {
      "epoch": 0.23589630689636193,
      "grad_norm": 0.7431296706199646,
      "learning_rate": 9.674359574922098e-06,
      "loss": 0.8102,
      "step": 4286
    },
    {
      "epoch": 0.2359513456987176,
      "grad_norm": 0.7745676040649414,
      "learning_rate": 9.674205683116612e-06,
      "loss": 0.8733,
      "step": 4287
    },
    {
      "epoch": 0.23600638450107325,
      "grad_norm": 1.0117937326431274,
      "learning_rate": 9.674051756181046e-06,
      "loss": 0.9035,
      "step": 4288
    },
    {
      "epoch": 0.2360614233034289,
      "grad_norm": 0.7848078608512878,
      "learning_rate": 9.67389779411656e-06,
      "loss": 0.8486,
      "step": 4289
    },
    {
      "epoch": 0.23611646210578457,
      "grad_norm": 0.8439378142356873,
      "learning_rate": 9.673743796924307e-06,
      "loss": 0.8032,
      "step": 4290
    },
    {
      "epoch": 0.23617150090814024,
      "grad_norm": 0.8268104791641235,
      "learning_rate": 9.673589764605449e-06,
      "loss": 0.8182,
      "step": 4291
    },
    {
      "epoch": 0.2362265397104959,
      "grad_norm": 0.8896234631538391,
      "learning_rate": 9.67343569716114e-06,
      "loss": 0.8081,
      "step": 4292
    },
    {
      "epoch": 0.23628157851285156,
      "grad_norm": 0.8515019416809082,
      "learning_rate": 9.67328159459254e-06,
      "loss": 0.8239,
      "step": 4293
    },
    {
      "epoch": 0.23633661731520722,
      "grad_norm": 0.7779792547225952,
      "learning_rate": 9.673127456900806e-06,
      "loss": 0.8437,
      "step": 4294
    },
    {
      "epoch": 0.23639165611756288,
      "grad_norm": 0.7782402634620667,
      "learning_rate": 9.672973284087097e-06,
      "loss": 0.8498,
      "step": 4295
    },
    {
      "epoch": 0.23644669491991854,
      "grad_norm": 0.7588973641395569,
      "learning_rate": 9.67281907615257e-06,
      "loss": 0.7034,
      "step": 4296
    },
    {
      "epoch": 0.2365017337222742,
      "grad_norm": 0.8426640629768372,
      "learning_rate": 9.67266483309839e-06,
      "loss": 0.803,
      "step": 4297
    },
    {
      "epoch": 0.23655677252462987,
      "grad_norm": 0.8945889472961426,
      "learning_rate": 9.672510554925707e-06,
      "loss": 0.8971,
      "step": 4298
    },
    {
      "epoch": 0.23661181132698553,
      "grad_norm": 0.8604227304458618,
      "learning_rate": 9.672356241635688e-06,
      "loss": 0.7548,
      "step": 4299
    },
    {
      "epoch": 0.2366668501293412,
      "grad_norm": 0.7277490496635437,
      "learning_rate": 9.672201893229489e-06,
      "loss": 0.8083,
      "step": 4300
    },
    {
      "epoch": 0.23672188893169685,
      "grad_norm": 0.9089379906654358,
      "learning_rate": 9.672047509708273e-06,
      "loss": 0.9717,
      "step": 4301
    },
    {
      "epoch": 0.2367769277340525,
      "grad_norm": 0.7207155823707581,
      "learning_rate": 9.671893091073198e-06,
      "loss": 0.6794,
      "step": 4302
    },
    {
      "epoch": 0.23683196653640817,
      "grad_norm": 0.7319806814193726,
      "learning_rate": 9.671738637325425e-06,
      "loss": 0.6821,
      "step": 4303
    },
    {
      "epoch": 0.23688700533876383,
      "grad_norm": 0.7339589595794678,
      "learning_rate": 9.671584148466112e-06,
      "loss": 0.7895,
      "step": 4304
    },
    {
      "epoch": 0.2369420441411195,
      "grad_norm": 0.7725476622581482,
      "learning_rate": 9.671429624496428e-06,
      "loss": 0.7414,
      "step": 4305
    },
    {
      "epoch": 0.23699708294347516,
      "grad_norm": 0.7040137648582458,
      "learning_rate": 9.671275065417527e-06,
      "loss": 0.696,
      "step": 4306
    },
    {
      "epoch": 0.23705212174583082,
      "grad_norm": 0.8804189562797546,
      "learning_rate": 9.671120471230572e-06,
      "loss": 0.8184,
      "step": 4307
    },
    {
      "epoch": 0.23710716054818648,
      "grad_norm": 0.8062872886657715,
      "learning_rate": 9.670965841936728e-06,
      "loss": 0.7856,
      "step": 4308
    },
    {
      "epoch": 0.23716219935054214,
      "grad_norm": 0.7537097930908203,
      "learning_rate": 9.670811177537154e-06,
      "loss": 0.7562,
      "step": 4309
    },
    {
      "epoch": 0.2372172381528978,
      "grad_norm": 0.8168618083000183,
      "learning_rate": 9.670656478033013e-06,
      "loss": 0.7416,
      "step": 4310
    },
    {
      "epoch": 0.23727227695525346,
      "grad_norm": 0.8367040157318115,
      "learning_rate": 9.670501743425469e-06,
      "loss": 0.7759,
      "step": 4311
    },
    {
      "epoch": 0.23732731575760913,
      "grad_norm": 0.860418975353241,
      "learning_rate": 9.670346973715683e-06,
      "loss": 0.9013,
      "step": 4312
    },
    {
      "epoch": 0.2373823545599648,
      "grad_norm": 0.8736678957939148,
      "learning_rate": 9.67019216890482e-06,
      "loss": 0.8677,
      "step": 4313
    },
    {
      "epoch": 0.23743739336232045,
      "grad_norm": 0.8258964419364929,
      "learning_rate": 9.670037328994044e-06,
      "loss": 0.8208,
      "step": 4314
    },
    {
      "epoch": 0.2374924321646761,
      "grad_norm": 0.7936292886734009,
      "learning_rate": 9.669882453984516e-06,
      "loss": 0.8643,
      "step": 4315
    },
    {
      "epoch": 0.23754747096703177,
      "grad_norm": 0.805500864982605,
      "learning_rate": 9.669727543877401e-06,
      "loss": 0.779,
      "step": 4316
    },
    {
      "epoch": 0.2376025097693874,
      "grad_norm": 0.8072311282157898,
      "learning_rate": 9.669572598673866e-06,
      "loss": 0.8258,
      "step": 4317
    },
    {
      "epoch": 0.23765754857174307,
      "grad_norm": 0.8917607665061951,
      "learning_rate": 9.669417618375072e-06,
      "loss": 0.7528,
      "step": 4318
    },
    {
      "epoch": 0.23771258737409873,
      "grad_norm": 0.7054246068000793,
      "learning_rate": 9.669262602982186e-06,
      "loss": 0.86,
      "step": 4319
    },
    {
      "epoch": 0.2377676261764544,
      "grad_norm": 0.8600299954414368,
      "learning_rate": 9.66910755249637e-06,
      "loss": 0.8165,
      "step": 4320
    },
    {
      "epoch": 0.23782266497881005,
      "grad_norm": 0.8685561418533325,
      "learning_rate": 9.668952466918793e-06,
      "loss": 0.8129,
      "step": 4321
    },
    {
      "epoch": 0.2378777037811657,
      "grad_norm": 0.7859770655632019,
      "learning_rate": 9.668797346250618e-06,
      "loss": 0.8703,
      "step": 4322
    },
    {
      "epoch": 0.23793274258352137,
      "grad_norm": 0.8128730058670044,
      "learning_rate": 9.668642190493015e-06,
      "loss": 0.7595,
      "step": 4323
    },
    {
      "epoch": 0.23798778138587703,
      "grad_norm": 0.8223204612731934,
      "learning_rate": 9.668486999647143e-06,
      "loss": 0.825,
      "step": 4324
    },
    {
      "epoch": 0.2380428201882327,
      "grad_norm": 0.859619677066803,
      "learning_rate": 9.668331773714175e-06,
      "loss": 0.8239,
      "step": 4325
    },
    {
      "epoch": 0.23809785899058836,
      "grad_norm": 0.9861679673194885,
      "learning_rate": 9.668176512695273e-06,
      "loss": 0.8409,
      "step": 4326
    },
    {
      "epoch": 0.23815289779294402,
      "grad_norm": 0.7178627252578735,
      "learning_rate": 9.668021216591607e-06,
      "loss": 0.818,
      "step": 4327
    },
    {
      "epoch": 0.23820793659529968,
      "grad_norm": 0.9160923957824707,
      "learning_rate": 9.667865885404343e-06,
      "loss": 0.8703,
      "step": 4328
    },
    {
      "epoch": 0.23826297539765534,
      "grad_norm": 0.7043942213058472,
      "learning_rate": 9.667710519134648e-06,
      "loss": 0.6884,
      "step": 4329
    },
    {
      "epoch": 0.238318014200011,
      "grad_norm": 1.213121771812439,
      "learning_rate": 9.667555117783691e-06,
      "loss": 0.7843,
      "step": 4330
    },
    {
      "epoch": 0.23837305300236666,
      "grad_norm": 0.8008033037185669,
      "learning_rate": 9.66739968135264e-06,
      "loss": 0.9312,
      "step": 4331
    },
    {
      "epoch": 0.23842809180472233,
      "grad_norm": 0.7862009406089783,
      "learning_rate": 9.667244209842662e-06,
      "loss": 0.6965,
      "step": 4332
    },
    {
      "epoch": 0.238483130607078,
      "grad_norm": 1.081398844718933,
      "learning_rate": 9.667088703254923e-06,
      "loss": 0.9793,
      "step": 4333
    },
    {
      "epoch": 0.23853816940943365,
      "grad_norm": 0.7672395706176758,
      "learning_rate": 9.666933161590597e-06,
      "loss": 0.813,
      "step": 4334
    },
    {
      "epoch": 0.2385932082117893,
      "grad_norm": 0.6955493092536926,
      "learning_rate": 9.66677758485085e-06,
      "loss": 0.7778,
      "step": 4335
    },
    {
      "epoch": 0.23864824701414497,
      "grad_norm": 0.8609682321548462,
      "learning_rate": 9.666621973036854e-06,
      "loss": 0.7817,
      "step": 4336
    },
    {
      "epoch": 0.23870328581650063,
      "grad_norm": 0.7312196493148804,
      "learning_rate": 9.666466326149774e-06,
      "loss": 0.7368,
      "step": 4337
    },
    {
      "epoch": 0.2387583246188563,
      "grad_norm": 0.7964538931846619,
      "learning_rate": 9.666310644190782e-06,
      "loss": 0.8124,
      "step": 4338
    },
    {
      "epoch": 0.23881336342121195,
      "grad_norm": 1.1138910055160522,
      "learning_rate": 9.66615492716105e-06,
      "loss": 0.8886,
      "step": 4339
    },
    {
      "epoch": 0.23886840222356762,
      "grad_norm": 0.8789949417114258,
      "learning_rate": 9.665999175061747e-06,
      "loss": 0.7854,
      "step": 4340
    },
    {
      "epoch": 0.23892344102592328,
      "grad_norm": 0.7761380076408386,
      "learning_rate": 9.665843387894041e-06,
      "loss": 0.7915,
      "step": 4341
    },
    {
      "epoch": 0.23897847982827894,
      "grad_norm": 0.888482928276062,
      "learning_rate": 9.665687565659106e-06,
      "loss": 0.8799,
      "step": 4342
    },
    {
      "epoch": 0.2390335186306346,
      "grad_norm": 0.7799200415611267,
      "learning_rate": 9.665531708358111e-06,
      "loss": 0.8519,
      "step": 4343
    },
    {
      "epoch": 0.23908855743299026,
      "grad_norm": 0.7407697439193726,
      "learning_rate": 9.665375815992231e-06,
      "loss": 0.7637,
      "step": 4344
    },
    {
      "epoch": 0.23914359623534592,
      "grad_norm": 0.8098278045654297,
      "learning_rate": 9.665219888562634e-06,
      "loss": 0.7991,
      "step": 4345
    },
    {
      "epoch": 0.23919863503770158,
      "grad_norm": 0.7585136294364929,
      "learning_rate": 9.665063926070493e-06,
      "loss": 0.8478,
      "step": 4346
    },
    {
      "epoch": 0.23925367384005725,
      "grad_norm": 0.7294817566871643,
      "learning_rate": 9.66490792851698e-06,
      "loss": 0.8312,
      "step": 4347
    },
    {
      "epoch": 0.2393087126424129,
      "grad_norm": 0.8325762748718262,
      "learning_rate": 9.664751895903269e-06,
      "loss": 0.9365,
      "step": 4348
    },
    {
      "epoch": 0.23936375144476857,
      "grad_norm": 0.9992470741271973,
      "learning_rate": 9.66459582823053e-06,
      "loss": 0.8649,
      "step": 4349
    },
    {
      "epoch": 0.23941879024712423,
      "grad_norm": 0.7206875681877136,
      "learning_rate": 9.664439725499938e-06,
      "loss": 0.7013,
      "step": 4350
    },
    {
      "epoch": 0.2394738290494799,
      "grad_norm": 0.946657657623291,
      "learning_rate": 9.664283587712665e-06,
      "loss": 0.7953,
      "step": 4351
    },
    {
      "epoch": 0.23952886785183555,
      "grad_norm": 0.7684911489486694,
      "learning_rate": 9.664127414869887e-06,
      "loss": 0.8403,
      "step": 4352
    },
    {
      "epoch": 0.23958390665419121,
      "grad_norm": 0.7875770926475525,
      "learning_rate": 9.663971206972773e-06,
      "loss": 0.7961,
      "step": 4353
    },
    {
      "epoch": 0.23963894545654688,
      "grad_norm": 0.7387273907661438,
      "learning_rate": 9.663814964022502e-06,
      "loss": 0.8265,
      "step": 4354
    },
    {
      "epoch": 0.23969398425890254,
      "grad_norm": 0.7413492202758789,
      "learning_rate": 9.663658686020245e-06,
      "loss": 0.8458,
      "step": 4355
    },
    {
      "epoch": 0.2397490230612582,
      "grad_norm": 0.7563235759735107,
      "learning_rate": 9.663502372967177e-06,
      "loss": 0.8498,
      "step": 4356
    },
    {
      "epoch": 0.23980406186361386,
      "grad_norm": 0.7529472708702087,
      "learning_rate": 9.663346024864475e-06,
      "loss": 0.7597,
      "step": 4357
    },
    {
      "epoch": 0.23985910066596952,
      "grad_norm": 0.7582191824913025,
      "learning_rate": 9.663189641713314e-06,
      "loss": 0.804,
      "step": 4358
    },
    {
      "epoch": 0.23991413946832518,
      "grad_norm": 0.8394485712051392,
      "learning_rate": 9.663033223514865e-06,
      "loss": 0.8329,
      "step": 4359
    },
    {
      "epoch": 0.23996917827068082,
      "grad_norm": 0.7088292241096497,
      "learning_rate": 9.662876770270308e-06,
      "loss": 0.7131,
      "step": 4360
    },
    {
      "epoch": 0.24002421707303648,
      "grad_norm": 0.8548080325126648,
      "learning_rate": 9.662720281980817e-06,
      "loss": 0.8925,
      "step": 4361
    },
    {
      "epoch": 0.24007925587539214,
      "grad_norm": 0.8027567267417908,
      "learning_rate": 9.662563758647568e-06,
      "loss": 0.8652,
      "step": 4362
    },
    {
      "epoch": 0.2401342946777478,
      "grad_norm": 0.7471736669540405,
      "learning_rate": 9.662407200271738e-06,
      "loss": 0.7722,
      "step": 4363
    },
    {
      "epoch": 0.24018933348010346,
      "grad_norm": 0.7358804941177368,
      "learning_rate": 9.662250606854504e-06,
      "loss": 0.767,
      "step": 4364
    },
    {
      "epoch": 0.24024437228245912,
      "grad_norm": 0.7948476672172546,
      "learning_rate": 9.662093978397042e-06,
      "loss": 0.961,
      "step": 4365
    },
    {
      "epoch": 0.24029941108481478,
      "grad_norm": 0.7030961513519287,
      "learning_rate": 9.66193731490053e-06,
      "loss": 0.7826,
      "step": 4366
    },
    {
      "epoch": 0.24035444988717045,
      "grad_norm": 0.8376098871231079,
      "learning_rate": 9.661780616366145e-06,
      "loss": 0.7697,
      "step": 4367
    },
    {
      "epoch": 0.2404094886895261,
      "grad_norm": 0.7449594140052795,
      "learning_rate": 9.661623882795065e-06,
      "loss": 0.7944,
      "step": 4368
    },
    {
      "epoch": 0.24046452749188177,
      "grad_norm": 0.7317184805870056,
      "learning_rate": 9.661467114188468e-06,
      "loss": 0.7059,
      "step": 4369
    },
    {
      "epoch": 0.24051956629423743,
      "grad_norm": 0.843912661075592,
      "learning_rate": 9.661310310547531e-06,
      "loss": 0.7889,
      "step": 4370
    },
    {
      "epoch": 0.2405746050965931,
      "grad_norm": 0.8673211336135864,
      "learning_rate": 9.661153471873435e-06,
      "loss": 0.7234,
      "step": 4371
    },
    {
      "epoch": 0.24062964389894875,
      "grad_norm": 0.8179688453674316,
      "learning_rate": 9.660996598167354e-06,
      "loss": 0.8937,
      "step": 4372
    },
    {
      "epoch": 0.24068468270130441,
      "grad_norm": 0.7800211906433105,
      "learning_rate": 9.660839689430473e-06,
      "loss": 0.8596,
      "step": 4373
    },
    {
      "epoch": 0.24073972150366008,
      "grad_norm": 0.8781671524047852,
      "learning_rate": 9.660682745663967e-06,
      "loss": 0.8507,
      "step": 4374
    },
    {
      "epoch": 0.24079476030601574,
      "grad_norm": 0.7701708674430847,
      "learning_rate": 9.660525766869019e-06,
      "loss": 0.8212,
      "step": 4375
    },
    {
      "epoch": 0.2408497991083714,
      "grad_norm": 0.7721084356307983,
      "learning_rate": 9.660368753046806e-06,
      "loss": 0.7493,
      "step": 4376
    },
    {
      "epoch": 0.24090483791072706,
      "grad_norm": 0.8126489520072937,
      "learning_rate": 9.660211704198508e-06,
      "loss": 0.8527,
      "step": 4377
    },
    {
      "epoch": 0.24095987671308272,
      "grad_norm": 0.8172717690467834,
      "learning_rate": 9.660054620325307e-06,
      "loss": 0.8448,
      "step": 4378
    },
    {
      "epoch": 0.24101491551543838,
      "grad_norm": 0.8293611407279968,
      "learning_rate": 9.659897501428384e-06,
      "loss": 0.9318,
      "step": 4379
    },
    {
      "epoch": 0.24106995431779404,
      "grad_norm": 0.7445098161697388,
      "learning_rate": 9.659740347508917e-06,
      "loss": 0.7358,
      "step": 4380
    },
    {
      "epoch": 0.2411249931201497,
      "grad_norm": 0.7778907418251038,
      "learning_rate": 9.659583158568088e-06,
      "loss": 0.7671,
      "step": 4381
    },
    {
      "epoch": 0.24118003192250537,
      "grad_norm": 0.7828608751296997,
      "learning_rate": 9.659425934607082e-06,
      "loss": 0.8141,
      "step": 4382
    },
    {
      "epoch": 0.24123507072486103,
      "grad_norm": 0.9433113932609558,
      "learning_rate": 9.659268675627075e-06,
      "loss": 0.7904,
      "step": 4383
    },
    {
      "epoch": 0.2412901095272167,
      "grad_norm": 0.7097491025924683,
      "learning_rate": 9.659111381629255e-06,
      "loss": 0.7445,
      "step": 4384
    },
    {
      "epoch": 0.24134514832957235,
      "grad_norm": 0.7450230717658997,
      "learning_rate": 9.6589540526148e-06,
      "loss": 0.6869,
      "step": 4385
    },
    {
      "epoch": 0.241400187131928,
      "grad_norm": 0.7429760694503784,
      "learning_rate": 9.658796688584893e-06,
      "loss": 0.7367,
      "step": 4386
    },
    {
      "epoch": 0.24145522593428367,
      "grad_norm": 0.7250030040740967,
      "learning_rate": 9.658639289540716e-06,
      "loss": 0.7502,
      "step": 4387
    },
    {
      "epoch": 0.24151026473663934,
      "grad_norm": 0.6577159762382507,
      "learning_rate": 9.658481855483455e-06,
      "loss": 0.5785,
      "step": 4388
    },
    {
      "epoch": 0.241565303538995,
      "grad_norm": 0.7846524119377136,
      "learning_rate": 9.65832438641429e-06,
      "loss": 0.7435,
      "step": 4389
    },
    {
      "epoch": 0.24162034234135066,
      "grad_norm": 0.8370404839515686,
      "learning_rate": 9.658166882334408e-06,
      "loss": 0.8536,
      "step": 4390
    },
    {
      "epoch": 0.24167538114370632,
      "grad_norm": 0.7451018691062927,
      "learning_rate": 9.658009343244987e-06,
      "loss": 0.8443,
      "step": 4391
    },
    {
      "epoch": 0.24173041994606198,
      "grad_norm": 0.7629074454307556,
      "learning_rate": 9.657851769147218e-06,
      "loss": 0.7394,
      "step": 4392
    },
    {
      "epoch": 0.24178545874841764,
      "grad_norm": 0.7767705321311951,
      "learning_rate": 9.657694160042282e-06,
      "loss": 0.8497,
      "step": 4393
    },
    {
      "epoch": 0.2418404975507733,
      "grad_norm": 0.8635357022285461,
      "learning_rate": 9.65753651593136e-06,
      "loss": 0.8495,
      "step": 4394
    },
    {
      "epoch": 0.24189553635312896,
      "grad_norm": 0.7652365565299988,
      "learning_rate": 9.657378836815643e-06,
      "loss": 0.7967,
      "step": 4395
    },
    {
      "epoch": 0.24195057515548463,
      "grad_norm": 0.7721680402755737,
      "learning_rate": 9.657221122696313e-06,
      "loss": 0.8227,
      "step": 4396
    },
    {
      "epoch": 0.2420056139578403,
      "grad_norm": 1.016366720199585,
      "learning_rate": 9.657063373574555e-06,
      "loss": 0.8291,
      "step": 4397
    },
    {
      "epoch": 0.24206065276019595,
      "grad_norm": 0.7770145535469055,
      "learning_rate": 9.656905589451555e-06,
      "loss": 0.8335,
      "step": 4398
    },
    {
      "epoch": 0.2421156915625516,
      "grad_norm": 0.812882125377655,
      "learning_rate": 9.6567477703285e-06,
      "loss": 0.8189,
      "step": 4399
    },
    {
      "epoch": 0.24217073036490727,
      "grad_norm": 0.7253247499465942,
      "learning_rate": 9.656589916206576e-06,
      "loss": 0.8418,
      "step": 4400
    },
    {
      "epoch": 0.24222576916726293,
      "grad_norm": 0.7784958481788635,
      "learning_rate": 9.656432027086969e-06,
      "loss": 0.8541,
      "step": 4401
    },
    {
      "epoch": 0.2422808079696186,
      "grad_norm": 0.8001978397369385,
      "learning_rate": 9.656274102970865e-06,
      "loss": 0.8888,
      "step": 4402
    },
    {
      "epoch": 0.24233584677197423,
      "grad_norm": 0.7535765767097473,
      "learning_rate": 9.656116143859448e-06,
      "loss": 0.7691,
      "step": 4403
    },
    {
      "epoch": 0.2423908855743299,
      "grad_norm": 0.6554346680641174,
      "learning_rate": 9.655958149753913e-06,
      "loss": 0.7592,
      "step": 4404
    },
    {
      "epoch": 0.24244592437668555,
      "grad_norm": 0.8599995374679565,
      "learning_rate": 9.655800120655439e-06,
      "loss": 0.8396,
      "step": 4405
    },
    {
      "epoch": 0.2425009631790412,
      "grad_norm": 0.8172232508659363,
      "learning_rate": 9.65564205656522e-06,
      "loss": 0.6931,
      "step": 4406
    },
    {
      "epoch": 0.24255600198139687,
      "grad_norm": 0.8005852699279785,
      "learning_rate": 9.65548395748444e-06,
      "loss": 0.8344,
      "step": 4407
    },
    {
      "epoch": 0.24261104078375254,
      "grad_norm": 0.7823762893676758,
      "learning_rate": 9.65532582341429e-06,
      "loss": 0.7991,
      "step": 4408
    },
    {
      "epoch": 0.2426660795861082,
      "grad_norm": 0.7743250727653503,
      "learning_rate": 9.655167654355957e-06,
      "loss": 0.9048,
      "step": 4409
    },
    {
      "epoch": 0.24272111838846386,
      "grad_norm": 0.9825221300125122,
      "learning_rate": 9.655009450310629e-06,
      "loss": 0.7491,
      "step": 4410
    },
    {
      "epoch": 0.24277615719081952,
      "grad_norm": 1.2921068668365479,
      "learning_rate": 9.654851211279496e-06,
      "loss": 0.8175,
      "step": 4411
    },
    {
      "epoch": 0.24283119599317518,
      "grad_norm": 0.8267684578895569,
      "learning_rate": 9.65469293726375e-06,
      "loss": 0.8896,
      "step": 4412
    },
    {
      "epoch": 0.24288623479553084,
      "grad_norm": 0.8020186424255371,
      "learning_rate": 9.654534628264576e-06,
      "loss": 0.7145,
      "step": 4413
    },
    {
      "epoch": 0.2429412735978865,
      "grad_norm": 0.8192574977874756,
      "learning_rate": 9.654376284283166e-06,
      "loss": 0.7451,
      "step": 4414
    },
    {
      "epoch": 0.24299631240024216,
      "grad_norm": 0.7733662128448486,
      "learning_rate": 9.65421790532071e-06,
      "loss": 0.768,
      "step": 4415
    },
    {
      "epoch": 0.24305135120259783,
      "grad_norm": 0.8342406153678894,
      "learning_rate": 9.654059491378396e-06,
      "loss": 0.8137,
      "step": 4416
    },
    {
      "epoch": 0.2431063900049535,
      "grad_norm": 1.014755368232727,
      "learning_rate": 9.653901042457418e-06,
      "loss": 0.8922,
      "step": 4417
    },
    {
      "epoch": 0.24316142880730915,
      "grad_norm": 0.864608645439148,
      "learning_rate": 9.653742558558967e-06,
      "loss": 0.9412,
      "step": 4418
    },
    {
      "epoch": 0.2432164676096648,
      "grad_norm": 0.7383908033370972,
      "learning_rate": 9.65358403968423e-06,
      "loss": 0.8261,
      "step": 4419
    },
    {
      "epoch": 0.24327150641202047,
      "grad_norm": 0.7464672923088074,
      "learning_rate": 9.653425485834403e-06,
      "loss": 0.7074,
      "step": 4420
    },
    {
      "epoch": 0.24332654521437613,
      "grad_norm": 0.7010141611099243,
      "learning_rate": 9.653266897010676e-06,
      "loss": 0.6849,
      "step": 4421
    },
    {
      "epoch": 0.2433815840167318,
      "grad_norm": 0.7135268449783325,
      "learning_rate": 9.653108273214239e-06,
      "loss": 0.8228,
      "step": 4422
    },
    {
      "epoch": 0.24343662281908746,
      "grad_norm": 0.8061006665229797,
      "learning_rate": 9.652949614446287e-06,
      "loss": 0.8345,
      "step": 4423
    },
    {
      "epoch": 0.24349166162144312,
      "grad_norm": 0.6954759955406189,
      "learning_rate": 9.652790920708011e-06,
      "loss": 0.7189,
      "step": 4424
    },
    {
      "epoch": 0.24354670042379878,
      "grad_norm": 0.8669333457946777,
      "learning_rate": 9.652632192000603e-06,
      "loss": 0.8872,
      "step": 4425
    },
    {
      "epoch": 0.24360173922615444,
      "grad_norm": 0.7445051670074463,
      "learning_rate": 9.652473428325258e-06,
      "loss": 0.826,
      "step": 4426
    },
    {
      "epoch": 0.2436567780285101,
      "grad_norm": 0.7444632649421692,
      "learning_rate": 9.652314629683165e-06,
      "loss": 0.8568,
      "step": 4427
    },
    {
      "epoch": 0.24371181683086576,
      "grad_norm": 0.7160165309906006,
      "learning_rate": 9.652155796075524e-06,
      "loss": 0.799,
      "step": 4428
    },
    {
      "epoch": 0.24376685563322142,
      "grad_norm": 0.7098904252052307,
      "learning_rate": 9.651996927503526e-06,
      "loss": 0.8148,
      "step": 4429
    },
    {
      "epoch": 0.24382189443557709,
      "grad_norm": 0.7911115288734436,
      "learning_rate": 9.651838023968363e-06,
      "loss": 0.8279,
      "step": 4430
    },
    {
      "epoch": 0.24387693323793275,
      "grad_norm": 0.8887501955032349,
      "learning_rate": 9.651679085471229e-06,
      "loss": 0.8464,
      "step": 4431
    },
    {
      "epoch": 0.2439319720402884,
      "grad_norm": 0.8343196511268616,
      "learning_rate": 9.651520112013321e-06,
      "loss": 0.7364,
      "step": 4432
    },
    {
      "epoch": 0.24398701084264407,
      "grad_norm": 0.7279361486434937,
      "learning_rate": 9.651361103595835e-06,
      "loss": 0.7958,
      "step": 4433
    },
    {
      "epoch": 0.24404204964499973,
      "grad_norm": 0.8221089243888855,
      "learning_rate": 9.651202060219962e-06,
      "loss": 0.7753,
      "step": 4434
    },
    {
      "epoch": 0.2440970884473554,
      "grad_norm": 0.7205086350440979,
      "learning_rate": 9.6510429818869e-06,
      "loss": 0.7411,
      "step": 4435
    },
    {
      "epoch": 0.24415212724971105,
      "grad_norm": 0.854967474937439,
      "learning_rate": 9.650883868597845e-06,
      "loss": 0.8192,
      "step": 4436
    },
    {
      "epoch": 0.24420716605206672,
      "grad_norm": 0.7622473835945129,
      "learning_rate": 9.65072472035399e-06,
      "loss": 0.7645,
      "step": 4437
    },
    {
      "epoch": 0.24426220485442238,
      "grad_norm": 0.7430302500724792,
      "learning_rate": 9.650565537156533e-06,
      "loss": 0.7817,
      "step": 4438
    },
    {
      "epoch": 0.24431724365677804,
      "grad_norm": 0.8022677898406982,
      "learning_rate": 9.650406319006672e-06,
      "loss": 0.8035,
      "step": 4439
    },
    {
      "epoch": 0.2443722824591337,
      "grad_norm": 0.7346476912498474,
      "learning_rate": 9.6502470659056e-06,
      "loss": 0.826,
      "step": 4440
    },
    {
      "epoch": 0.24442732126148936,
      "grad_norm": 0.8393376469612122,
      "learning_rate": 9.650087777854517e-06,
      "loss": 0.8073,
      "step": 4441
    },
    {
      "epoch": 0.24448236006384502,
      "grad_norm": 0.7920215129852295,
      "learning_rate": 9.649928454854618e-06,
      "loss": 0.7774,
      "step": 4442
    },
    {
      "epoch": 0.24453739886620068,
      "grad_norm": 0.8192804455757141,
      "learning_rate": 9.649769096907102e-06,
      "loss": 0.7817,
      "step": 4443
    },
    {
      "epoch": 0.24459243766855635,
      "grad_norm": 0.7727654576301575,
      "learning_rate": 9.649609704013167e-06,
      "loss": 0.8201,
      "step": 4444
    },
    {
      "epoch": 0.244647476470912,
      "grad_norm": 0.8005746603012085,
      "learning_rate": 9.649450276174008e-06,
      "loss": 0.8893,
      "step": 4445
    },
    {
      "epoch": 0.24470251527326764,
      "grad_norm": 0.9029125571250916,
      "learning_rate": 9.649290813390828e-06,
      "loss": 0.7735,
      "step": 4446
    },
    {
      "epoch": 0.2447575540756233,
      "grad_norm": 0.8336170315742493,
      "learning_rate": 9.64913131566482e-06,
      "loss": 0.7505,
      "step": 4447
    },
    {
      "epoch": 0.24481259287797896,
      "grad_norm": 1.0272265672683716,
      "learning_rate": 9.648971782997188e-06,
      "loss": 0.8371,
      "step": 4448
    },
    {
      "epoch": 0.24486763168033462,
      "grad_norm": 0.8095843195915222,
      "learning_rate": 9.648812215389128e-06,
      "loss": 0.7599,
      "step": 4449
    },
    {
      "epoch": 0.24492267048269029,
      "grad_norm": 0.7690166234970093,
      "learning_rate": 9.648652612841837e-06,
      "loss": 0.8172,
      "step": 4450
    },
    {
      "epoch": 0.24497770928504595,
      "grad_norm": 0.8282617926597595,
      "learning_rate": 9.64849297535652e-06,
      "loss": 0.8477,
      "step": 4451
    },
    {
      "epoch": 0.2450327480874016,
      "grad_norm": 0.8307822346687317,
      "learning_rate": 9.648333302934373e-06,
      "loss": 0.7744,
      "step": 4452
    },
    {
      "epoch": 0.24508778688975727,
      "grad_norm": 0.7619080543518066,
      "learning_rate": 9.6481735955766e-06,
      "loss": 0.8417,
      "step": 4453
    },
    {
      "epoch": 0.24514282569211293,
      "grad_norm": 0.7879447937011719,
      "learning_rate": 9.648013853284396e-06,
      "loss": 0.7799,
      "step": 4454
    },
    {
      "epoch": 0.2451978644944686,
      "grad_norm": 0.7352256774902344,
      "learning_rate": 9.647854076058965e-06,
      "loss": 0.8386,
      "step": 4455
    },
    {
      "epoch": 0.24525290329682425,
      "grad_norm": 0.8318933248519897,
      "learning_rate": 9.647694263901507e-06,
      "loss": 0.7631,
      "step": 4456
    },
    {
      "epoch": 0.24530794209917992,
      "grad_norm": 0.8609912395477295,
      "learning_rate": 9.647534416813221e-06,
      "loss": 0.7479,
      "step": 4457
    },
    {
      "epoch": 0.24536298090153558,
      "grad_norm": 0.9590480327606201,
      "learning_rate": 9.647374534795311e-06,
      "loss": 0.8543,
      "step": 4458
    },
    {
      "epoch": 0.24541801970389124,
      "grad_norm": 0.7902723550796509,
      "learning_rate": 9.647214617848979e-06,
      "loss": 0.6796,
      "step": 4459
    },
    {
      "epoch": 0.2454730585062469,
      "grad_norm": 0.7725642919540405,
      "learning_rate": 9.647054665975427e-06,
      "loss": 0.7563,
      "step": 4460
    },
    {
      "epoch": 0.24552809730860256,
      "grad_norm": 0.8387014269828796,
      "learning_rate": 9.646894679175853e-06,
      "loss": 0.8184,
      "step": 4461
    },
    {
      "epoch": 0.24558313611095822,
      "grad_norm": 0.9200852513313293,
      "learning_rate": 9.646734657451464e-06,
      "loss": 0.8436,
      "step": 4462
    },
    {
      "epoch": 0.24563817491331388,
      "grad_norm": 0.7565840482711792,
      "learning_rate": 9.646574600803462e-06,
      "loss": 0.7393,
      "step": 4463
    },
    {
      "epoch": 0.24569321371566955,
      "grad_norm": 0.7685559988021851,
      "learning_rate": 9.646414509233048e-06,
      "loss": 0.7836,
      "step": 4464
    },
    {
      "epoch": 0.2457482525180252,
      "grad_norm": 0.8172003030776978,
      "learning_rate": 9.646254382741428e-06,
      "loss": 0.787,
      "step": 4465
    },
    {
      "epoch": 0.24580329132038087,
      "grad_norm": 0.902632474899292,
      "learning_rate": 9.646094221329802e-06,
      "loss": 0.7139,
      "step": 4466
    },
    {
      "epoch": 0.24585833012273653,
      "grad_norm": 0.7810692191123962,
      "learning_rate": 9.645934024999374e-06,
      "loss": 0.6904,
      "step": 4467
    },
    {
      "epoch": 0.2459133689250922,
      "grad_norm": 0.7242134213447571,
      "learning_rate": 9.645773793751352e-06,
      "loss": 0.7035,
      "step": 4468
    },
    {
      "epoch": 0.24596840772744785,
      "grad_norm": 0.7192920446395874,
      "learning_rate": 9.645613527586938e-06,
      "loss": 0.7081,
      "step": 4469
    },
    {
      "epoch": 0.2460234465298035,
      "grad_norm": 0.7613840103149414,
      "learning_rate": 9.645453226507336e-06,
      "loss": 0.8066,
      "step": 4470
    },
    {
      "epoch": 0.24607848533215917,
      "grad_norm": 0.8154922127723694,
      "learning_rate": 9.64529289051375e-06,
      "loss": 0.812,
      "step": 4471
    },
    {
      "epoch": 0.24613352413451484,
      "grad_norm": 0.9521573185920715,
      "learning_rate": 9.645132519607387e-06,
      "loss": 0.7456,
      "step": 4472
    },
    {
      "epoch": 0.2461885629368705,
      "grad_norm": 0.785943329334259,
      "learning_rate": 9.64497211378945e-06,
      "loss": 0.832,
      "step": 4473
    },
    {
      "epoch": 0.24624360173922616,
      "grad_norm": 0.7675127983093262,
      "learning_rate": 9.644811673061148e-06,
      "loss": 0.7984,
      "step": 4474
    },
    {
      "epoch": 0.24629864054158182,
      "grad_norm": 0.7317580580711365,
      "learning_rate": 9.644651197423683e-06,
      "loss": 0.7634,
      "step": 4475
    },
    {
      "epoch": 0.24635367934393748,
      "grad_norm": 0.744937539100647,
      "learning_rate": 9.644490686878265e-06,
      "loss": 0.729,
      "step": 4476
    },
    {
      "epoch": 0.24640871814629314,
      "grad_norm": 0.7472458481788635,
      "learning_rate": 9.644330141426097e-06,
      "loss": 0.7517,
      "step": 4477
    },
    {
      "epoch": 0.2464637569486488,
      "grad_norm": 0.8379414677619934,
      "learning_rate": 9.644169561068387e-06,
      "loss": 0.8008,
      "step": 4478
    },
    {
      "epoch": 0.24651879575100447,
      "grad_norm": 0.8845154047012329,
      "learning_rate": 9.64400894580634e-06,
      "loss": 0.8135,
      "step": 4479
    },
    {
      "epoch": 0.24657383455336013,
      "grad_norm": 0.7394443154335022,
      "learning_rate": 9.643848295641167e-06,
      "loss": 0.7697,
      "step": 4480
    },
    {
      "epoch": 0.2466288733557158,
      "grad_norm": 0.8840840458869934,
      "learning_rate": 9.643687610574073e-06,
      "loss": 0.825,
      "step": 4481
    },
    {
      "epoch": 0.24668391215807145,
      "grad_norm": 0.7924874424934387,
      "learning_rate": 9.643526890606265e-06,
      "loss": 0.793,
      "step": 4482
    },
    {
      "epoch": 0.2467389509604271,
      "grad_norm": 0.7966769933700562,
      "learning_rate": 9.643366135738951e-06,
      "loss": 0.8042,
      "step": 4483
    },
    {
      "epoch": 0.24679398976278277,
      "grad_norm": 0.911756694316864,
      "learning_rate": 9.643205345973343e-06,
      "loss": 0.7801,
      "step": 4484
    },
    {
      "epoch": 0.24684902856513843,
      "grad_norm": 0.903378963470459,
      "learning_rate": 9.643044521310645e-06,
      "loss": 0.7863,
      "step": 4485
    },
    {
      "epoch": 0.2469040673674941,
      "grad_norm": 0.9021226167678833,
      "learning_rate": 9.642883661752067e-06,
      "loss": 0.8005,
      "step": 4486
    },
    {
      "epoch": 0.24695910616984976,
      "grad_norm": 0.8853413462638855,
      "learning_rate": 9.64272276729882e-06,
      "loss": 0.8371,
      "step": 4487
    },
    {
      "epoch": 0.24701414497220542,
      "grad_norm": 1.0654630661010742,
      "learning_rate": 9.642561837952108e-06,
      "loss": 0.92,
      "step": 4488
    },
    {
      "epoch": 0.24706918377456105,
      "grad_norm": 0.8663573265075684,
      "learning_rate": 9.642400873713146e-06,
      "loss": 0.8066,
      "step": 4489
    },
    {
      "epoch": 0.2471242225769167,
      "grad_norm": 0.7483134269714355,
      "learning_rate": 9.642239874583143e-06,
      "loss": 0.9013,
      "step": 4490
    },
    {
      "epoch": 0.24717926137927237,
      "grad_norm": 0.7582293748855591,
      "learning_rate": 9.642078840563306e-06,
      "loss": 0.7795,
      "step": 4491
    },
    {
      "epoch": 0.24723430018162804,
      "grad_norm": 0.8276637196540833,
      "learning_rate": 9.641917771654848e-06,
      "loss": 0.7756,
      "step": 4492
    },
    {
      "epoch": 0.2472893389839837,
      "grad_norm": 0.697088360786438,
      "learning_rate": 9.641756667858976e-06,
      "loss": 0.7092,
      "step": 4493
    },
    {
      "epoch": 0.24734437778633936,
      "grad_norm": 0.8960816860198975,
      "learning_rate": 9.641595529176907e-06,
      "loss": 0.8835,
      "step": 4494
    },
    {
      "epoch": 0.24739941658869502,
      "grad_norm": 0.9210898280143738,
      "learning_rate": 9.641434355609846e-06,
      "loss": 0.7881,
      "step": 4495
    },
    {
      "epoch": 0.24745445539105068,
      "grad_norm": 0.7205467820167542,
      "learning_rate": 9.64127314715901e-06,
      "loss": 0.7204,
      "step": 4496
    },
    {
      "epoch": 0.24750949419340634,
      "grad_norm": 0.7313701510429382,
      "learning_rate": 9.641111903825603e-06,
      "loss": 0.8296,
      "step": 4497
    },
    {
      "epoch": 0.247564532995762,
      "grad_norm": 0.771159827709198,
      "learning_rate": 9.640950625610845e-06,
      "loss": 0.7974,
      "step": 4498
    },
    {
      "epoch": 0.24761957179811767,
      "grad_norm": 0.9227705597877502,
      "learning_rate": 9.64078931251594e-06,
      "loss": 0.9215,
      "step": 4499
    },
    {
      "epoch": 0.24767461060047333,
      "grad_norm": 0.7569915652275085,
      "learning_rate": 9.64062796454211e-06,
      "loss": 0.83,
      "step": 4500
    },
    {
      "epoch": 0.247729649402829,
      "grad_norm": 0.7453131675720215,
      "learning_rate": 9.64046658169056e-06,
      "loss": 0.6747,
      "step": 4501
    },
    {
      "epoch": 0.24778468820518465,
      "grad_norm": 0.7228132486343384,
      "learning_rate": 9.640305163962504e-06,
      "loss": 0.7535,
      "step": 4502
    },
    {
      "epoch": 0.2478397270075403,
      "grad_norm": 0.8160690069198608,
      "learning_rate": 9.640143711359159e-06,
      "loss": 0.8655,
      "step": 4503
    },
    {
      "epoch": 0.24789476580989597,
      "grad_norm": 0.7641691565513611,
      "learning_rate": 9.639982223881735e-06,
      "loss": 0.8353,
      "step": 4504
    },
    {
      "epoch": 0.24794980461225163,
      "grad_norm": 0.8669107556343079,
      "learning_rate": 9.639820701531445e-06,
      "loss": 0.8614,
      "step": 4505
    },
    {
      "epoch": 0.2480048434146073,
      "grad_norm": 0.7433111667633057,
      "learning_rate": 9.639659144309508e-06,
      "loss": 0.6891,
      "step": 4506
    },
    {
      "epoch": 0.24805988221696296,
      "grad_norm": 1.4303346872329712,
      "learning_rate": 9.639497552217131e-06,
      "loss": 0.8016,
      "step": 4507
    },
    {
      "epoch": 0.24811492101931862,
      "grad_norm": 0.8684772253036499,
      "learning_rate": 9.639335925255535e-06,
      "loss": 0.8324,
      "step": 4508
    },
    {
      "epoch": 0.24816995982167428,
      "grad_norm": 0.9222162365913391,
      "learning_rate": 9.639174263425932e-06,
      "loss": 0.8715,
      "step": 4509
    },
    {
      "epoch": 0.24822499862402994,
      "grad_norm": 0.9789180755615234,
      "learning_rate": 9.639012566729535e-06,
      "loss": 0.823,
      "step": 4510
    },
    {
      "epoch": 0.2482800374263856,
      "grad_norm": 0.8475140333175659,
      "learning_rate": 9.638850835167564e-06,
      "loss": 0.768,
      "step": 4511
    },
    {
      "epoch": 0.24833507622874126,
      "grad_norm": 0.7943722605705261,
      "learning_rate": 9.63868906874123e-06,
      "loss": 0.788,
      "step": 4512
    },
    {
      "epoch": 0.24839011503109693,
      "grad_norm": 0.8723915815353394,
      "learning_rate": 9.63852726745175e-06,
      "loss": 0.7865,
      "step": 4513
    },
    {
      "epoch": 0.2484451538334526,
      "grad_norm": 0.837001383304596,
      "learning_rate": 9.638365431300342e-06,
      "loss": 0.7799,
      "step": 4514
    },
    {
      "epoch": 0.24850019263580825,
      "grad_norm": 0.7992665767669678,
      "learning_rate": 9.638203560288222e-06,
      "loss": 0.8951,
      "step": 4515
    },
    {
      "epoch": 0.2485552314381639,
      "grad_norm": 0.8712993264198303,
      "learning_rate": 9.638041654416603e-06,
      "loss": 0.8157,
      "step": 4516
    },
    {
      "epoch": 0.24861027024051957,
      "grad_norm": 0.7176356911659241,
      "learning_rate": 9.637879713686706e-06,
      "loss": 0.8197,
      "step": 4517
    },
    {
      "epoch": 0.24866530904287523,
      "grad_norm": 0.7624368071556091,
      "learning_rate": 9.637717738099747e-06,
      "loss": 0.7545,
      "step": 4518
    },
    {
      "epoch": 0.2487203478452309,
      "grad_norm": 0.857222318649292,
      "learning_rate": 9.637555727656943e-06,
      "loss": 0.8146,
      "step": 4519
    },
    {
      "epoch": 0.24877538664758655,
      "grad_norm": 0.7461313605308533,
      "learning_rate": 9.637393682359511e-06,
      "loss": 0.8569,
      "step": 4520
    },
    {
      "epoch": 0.24883042544994222,
      "grad_norm": 0.8491896986961365,
      "learning_rate": 9.637231602208668e-06,
      "loss": 0.863,
      "step": 4521
    },
    {
      "epoch": 0.24888546425229788,
      "grad_norm": 0.8139386177062988,
      "learning_rate": 9.637069487205635e-06,
      "loss": 0.7105,
      "step": 4522
    },
    {
      "epoch": 0.24894050305465354,
      "grad_norm": 0.7782894968986511,
      "learning_rate": 9.636907337351629e-06,
      "loss": 0.8044,
      "step": 4523
    },
    {
      "epoch": 0.2489955418570092,
      "grad_norm": 0.8225486874580383,
      "learning_rate": 9.636745152647868e-06,
      "loss": 0.7877,
      "step": 4524
    },
    {
      "epoch": 0.24905058065936486,
      "grad_norm": 0.9087927341461182,
      "learning_rate": 9.636582933095573e-06,
      "loss": 0.8017,
      "step": 4525
    },
    {
      "epoch": 0.24910561946172052,
      "grad_norm": 0.7392508387565613,
      "learning_rate": 9.636420678695962e-06,
      "loss": 0.7953,
      "step": 4526
    },
    {
      "epoch": 0.24916065826407618,
      "grad_norm": 0.7906273007392883,
      "learning_rate": 9.636258389450253e-06,
      "loss": 0.9491,
      "step": 4527
    },
    {
      "epoch": 0.24921569706643185,
      "grad_norm": 0.840394139289856,
      "learning_rate": 9.636096065359666e-06,
      "loss": 0.8621,
      "step": 4528
    },
    {
      "epoch": 0.2492707358687875,
      "grad_norm": 0.7923862934112549,
      "learning_rate": 9.635933706425424e-06,
      "loss": 0.8215,
      "step": 4529
    },
    {
      "epoch": 0.24932577467114317,
      "grad_norm": 0.8372805714607239,
      "learning_rate": 9.635771312648744e-06,
      "loss": 0.8845,
      "step": 4530
    },
    {
      "epoch": 0.24938081347349883,
      "grad_norm": 0.7569165229797363,
      "learning_rate": 9.635608884030848e-06,
      "loss": 0.8406,
      "step": 4531
    },
    {
      "epoch": 0.24943585227585446,
      "grad_norm": 0.8260865807533264,
      "learning_rate": 9.635446420572956e-06,
      "loss": 0.8418,
      "step": 4532
    },
    {
      "epoch": 0.24949089107821013,
      "grad_norm": 0.6841318607330322,
      "learning_rate": 9.635283922276291e-06,
      "loss": 0.6732,
      "step": 4533
    },
    {
      "epoch": 0.2495459298805658,
      "grad_norm": 0.7055326104164124,
      "learning_rate": 9.635121389142072e-06,
      "loss": 0.7702,
      "step": 4534
    },
    {
      "epoch": 0.24960096868292145,
      "grad_norm": 0.7293457388877869,
      "learning_rate": 9.63495882117152e-06,
      "loss": 0.6836,
      "step": 4535
    },
    {
      "epoch": 0.2496560074852771,
      "grad_norm": 0.7411924004554749,
      "learning_rate": 9.63479621836586e-06,
      "loss": 0.8686,
      "step": 4536
    },
    {
      "epoch": 0.24971104628763277,
      "grad_norm": 0.7864643931388855,
      "learning_rate": 9.634633580726313e-06,
      "loss": 0.7801,
      "step": 4537
    },
    {
      "epoch": 0.24976608508998843,
      "grad_norm": 0.9730797410011292,
      "learning_rate": 9.634470908254099e-06,
      "loss": 0.8362,
      "step": 4538
    },
    {
      "epoch": 0.2498211238923441,
      "grad_norm": 0.8390370011329651,
      "learning_rate": 9.634308200950442e-06,
      "loss": 0.8079,
      "step": 4539
    },
    {
      "epoch": 0.24987616269469975,
      "grad_norm": 0.8951246738433838,
      "learning_rate": 9.634145458816566e-06,
      "loss": 0.7662,
      "step": 4540
    },
    {
      "epoch": 0.24993120149705542,
      "grad_norm": 0.7654157280921936,
      "learning_rate": 9.633982681853693e-06,
      "loss": 0.8699,
      "step": 4541
    },
    {
      "epoch": 0.24998624029941108,
      "grad_norm": 0.8152109980583191,
      "learning_rate": 9.633819870063046e-06,
      "loss": 0.7875,
      "step": 4542
    },
    {
      "epoch": 0.25004127910176677,
      "grad_norm": 0.9407321214675903,
      "learning_rate": 9.63365702344585e-06,
      "loss": 0.7708,
      "step": 4543
    },
    {
      "epoch": 0.2500963179041224,
      "grad_norm": 0.8169927597045898,
      "learning_rate": 9.633494142003327e-06,
      "loss": 0.8078,
      "step": 4544
    },
    {
      "epoch": 0.2501513567064781,
      "grad_norm": 0.7380755543708801,
      "learning_rate": 9.633331225736704e-06,
      "loss": 0.7818,
      "step": 4545
    },
    {
      "epoch": 0.2502063955088337,
      "grad_norm": 0.8124812841415405,
      "learning_rate": 9.633168274647203e-06,
      "loss": 0.8133,
      "step": 4546
    },
    {
      "epoch": 0.2502614343111894,
      "grad_norm": 0.8511367440223694,
      "learning_rate": 9.63300528873605e-06,
      "loss": 0.7747,
      "step": 4547
    },
    {
      "epoch": 0.25031647311354505,
      "grad_norm": 0.7305121421813965,
      "learning_rate": 9.632842268004469e-06,
      "loss": 0.8479,
      "step": 4548
    },
    {
      "epoch": 0.25037151191590074,
      "grad_norm": 0.7127692103385925,
      "learning_rate": 9.632679212453686e-06,
      "loss": 0.8514,
      "step": 4549
    },
    {
      "epoch": 0.25042655071825637,
      "grad_norm": 0.8251872062683105,
      "learning_rate": 9.632516122084926e-06,
      "loss": 0.7686,
      "step": 4550
    },
    {
      "epoch": 0.25048158952061206,
      "grad_norm": 0.6756613850593567,
      "learning_rate": 9.632352996899413e-06,
      "loss": 0.5959,
      "step": 4551
    },
    {
      "epoch": 0.2505366283229677,
      "grad_norm": 0.9266120791435242,
      "learning_rate": 9.632189836898377e-06,
      "loss": 0.7889,
      "step": 4552
    },
    {
      "epoch": 0.2505916671253233,
      "grad_norm": 0.769890546798706,
      "learning_rate": 9.63202664208304e-06,
      "loss": 0.7864,
      "step": 4553
    },
    {
      "epoch": 0.250646705927679,
      "grad_norm": 0.7314025163650513,
      "learning_rate": 9.631863412454634e-06,
      "loss": 0.8088,
      "step": 4554
    },
    {
      "epoch": 0.25070174473003465,
      "grad_norm": 0.818317711353302,
      "learning_rate": 9.63170014801438e-06,
      "loss": 0.7096,
      "step": 4555
    },
    {
      "epoch": 0.25075678353239034,
      "grad_norm": 0.7538807392120361,
      "learning_rate": 9.631536848763508e-06,
      "loss": 0.7779,
      "step": 4556
    },
    {
      "epoch": 0.25081182233474597,
      "grad_norm": 0.7658100128173828,
      "learning_rate": 9.631373514703247e-06,
      "loss": 0.8535,
      "step": 4557
    },
    {
      "epoch": 0.25086686113710166,
      "grad_norm": 0.8019290566444397,
      "learning_rate": 9.631210145834819e-06,
      "loss": 0.8141,
      "step": 4558
    },
    {
      "epoch": 0.2509218999394573,
      "grad_norm": 0.7257653474807739,
      "learning_rate": 9.631046742159456e-06,
      "loss": 0.7451,
      "step": 4559
    },
    {
      "epoch": 0.250976938741813,
      "grad_norm": 0.7546024918556213,
      "learning_rate": 9.630883303678386e-06,
      "loss": 0.7707,
      "step": 4560
    },
    {
      "epoch": 0.2510319775441686,
      "grad_norm": 0.7288938760757446,
      "learning_rate": 9.630719830392835e-06,
      "loss": 0.7362,
      "step": 4561
    },
    {
      "epoch": 0.2510870163465243,
      "grad_norm": 0.7814223170280457,
      "learning_rate": 9.630556322304036e-06,
      "loss": 0.8514,
      "step": 4562
    },
    {
      "epoch": 0.25114205514887994,
      "grad_norm": 0.7561381459236145,
      "learning_rate": 9.630392779413214e-06,
      "loss": 0.7659,
      "step": 4563
    },
    {
      "epoch": 0.25119709395123563,
      "grad_norm": 0.750641942024231,
      "learning_rate": 9.6302292017216e-06,
      "loss": 0.8496,
      "step": 4564
    },
    {
      "epoch": 0.25125213275359126,
      "grad_norm": 0.832155704498291,
      "learning_rate": 9.630065589230422e-06,
      "loss": 0.7778,
      "step": 4565
    },
    {
      "epoch": 0.25130717155594695,
      "grad_norm": 0.8202440142631531,
      "learning_rate": 9.62990194194091e-06,
      "loss": 0.8962,
      "step": 4566
    },
    {
      "epoch": 0.2513622103583026,
      "grad_norm": 0.8777977824211121,
      "learning_rate": 9.629738259854295e-06,
      "loss": 0.7215,
      "step": 4567
    },
    {
      "epoch": 0.2514172491606583,
      "grad_norm": 1.1868599653244019,
      "learning_rate": 9.629574542971806e-06,
      "loss": 0.8238,
      "step": 4568
    },
    {
      "epoch": 0.2514722879630139,
      "grad_norm": 0.9128753542900085,
      "learning_rate": 9.629410791294675e-06,
      "loss": 0.7638,
      "step": 4569
    },
    {
      "epoch": 0.2515273267653696,
      "grad_norm": 0.7350082993507385,
      "learning_rate": 9.629247004824132e-06,
      "loss": 0.8041,
      "step": 4570
    },
    {
      "epoch": 0.25158236556772523,
      "grad_norm": 0.7279660701751709,
      "learning_rate": 9.629083183561407e-06,
      "loss": 0.7377,
      "step": 4571
    },
    {
      "epoch": 0.2516374043700809,
      "grad_norm": 0.8570461273193359,
      "learning_rate": 9.628919327507732e-06,
      "loss": 0.8106,
      "step": 4572
    },
    {
      "epoch": 0.25169244317243655,
      "grad_norm": 0.8998312950134277,
      "learning_rate": 9.62875543666434e-06,
      "loss": 0.8171,
      "step": 4573
    },
    {
      "epoch": 0.25174748197479224,
      "grad_norm": 0.7631624937057495,
      "learning_rate": 9.628591511032456e-06,
      "loss": 0.7871,
      "step": 4574
    },
    {
      "epoch": 0.2518025207771479,
      "grad_norm": 0.7752320766448975,
      "learning_rate": 9.628427550613322e-06,
      "loss": 0.8241,
      "step": 4575
    },
    {
      "epoch": 0.25185755957950356,
      "grad_norm": 0.8741563558578491,
      "learning_rate": 9.628263555408163e-06,
      "loss": 0.7312,
      "step": 4576
    },
    {
      "epoch": 0.2519125983818592,
      "grad_norm": 0.8615008592605591,
      "learning_rate": 9.628099525418216e-06,
      "loss": 0.8586,
      "step": 4577
    },
    {
      "epoch": 0.2519676371842149,
      "grad_norm": 0.8273662328720093,
      "learning_rate": 9.62793546064471e-06,
      "loss": 0.7838,
      "step": 4578
    },
    {
      "epoch": 0.2520226759865705,
      "grad_norm": 0.7454090118408203,
      "learning_rate": 9.627771361088882e-06,
      "loss": 0.8461,
      "step": 4579
    },
    {
      "epoch": 0.2520777147889262,
      "grad_norm": 0.8225379586219788,
      "learning_rate": 9.627607226751962e-06,
      "loss": 0.7792,
      "step": 4580
    },
    {
      "epoch": 0.25213275359128184,
      "grad_norm": 0.8655416369438171,
      "learning_rate": 9.627443057635184e-06,
      "loss": 0.8165,
      "step": 4581
    },
    {
      "epoch": 0.25218779239363753,
      "grad_norm": 0.7735984921455383,
      "learning_rate": 9.627278853739783e-06,
      "loss": 0.8208,
      "step": 4582
    },
    {
      "epoch": 0.25224283119599317,
      "grad_norm": 0.8293350338935852,
      "learning_rate": 9.627114615066994e-06,
      "loss": 0.7394,
      "step": 4583
    },
    {
      "epoch": 0.25229786999834886,
      "grad_norm": 0.7840214371681213,
      "learning_rate": 9.626950341618048e-06,
      "loss": 0.8522,
      "step": 4584
    },
    {
      "epoch": 0.2523529088007045,
      "grad_norm": 0.7724186182022095,
      "learning_rate": 9.626786033394185e-06,
      "loss": 0.8175,
      "step": 4585
    },
    {
      "epoch": 0.2524079476030602,
      "grad_norm": 1.0751588344573975,
      "learning_rate": 9.626621690396634e-06,
      "loss": 0.9229,
      "step": 4586
    },
    {
      "epoch": 0.2524629864054158,
      "grad_norm": 0.7016913294792175,
      "learning_rate": 9.626457312626634e-06,
      "loss": 0.6883,
      "step": 4587
    },
    {
      "epoch": 0.2525180252077715,
      "grad_norm": 0.918377697467804,
      "learning_rate": 9.626292900085419e-06,
      "loss": 0.7889,
      "step": 4588
    },
    {
      "epoch": 0.25257306401012714,
      "grad_norm": 1.006564736366272,
      "learning_rate": 9.626128452774226e-06,
      "loss": 0.7888,
      "step": 4589
    },
    {
      "epoch": 0.2526281028124828,
      "grad_norm": 1.0214998722076416,
      "learning_rate": 9.625963970694287e-06,
      "loss": 0.768,
      "step": 4590
    },
    {
      "epoch": 0.25268314161483846,
      "grad_norm": 0.7980843186378479,
      "learning_rate": 9.625799453846844e-06,
      "loss": 0.8662,
      "step": 4591
    },
    {
      "epoch": 0.25273818041719415,
      "grad_norm": 0.734582245349884,
      "learning_rate": 9.625634902233128e-06,
      "loss": 0.759,
      "step": 4592
    },
    {
      "epoch": 0.2527932192195498,
      "grad_norm": 0.7185904383659363,
      "learning_rate": 9.62547031585438e-06,
      "loss": 0.774,
      "step": 4593
    },
    {
      "epoch": 0.25284825802190547,
      "grad_norm": 0.7356622219085693,
      "learning_rate": 9.625305694711835e-06,
      "loss": 0.7435,
      "step": 4594
    },
    {
      "epoch": 0.2529032968242611,
      "grad_norm": 0.7589355707168579,
      "learning_rate": 9.62514103880673e-06,
      "loss": 0.807,
      "step": 4595
    },
    {
      "epoch": 0.25295833562661674,
      "grad_norm": 0.889228880405426,
      "learning_rate": 9.624976348140305e-06,
      "loss": 0.8609,
      "step": 4596
    },
    {
      "epoch": 0.2530133744289724,
      "grad_norm": 0.7546125650405884,
      "learning_rate": 9.624811622713793e-06,
      "loss": 0.8379,
      "step": 4597
    },
    {
      "epoch": 0.25306841323132806,
      "grad_norm": 0.8262770175933838,
      "learning_rate": 9.624646862528436e-06,
      "loss": 0.7611,
      "step": 4598
    },
    {
      "epoch": 0.25312345203368375,
      "grad_norm": 0.8876076936721802,
      "learning_rate": 9.624482067585472e-06,
      "loss": 0.8106,
      "step": 4599
    },
    {
      "epoch": 0.2531784908360394,
      "grad_norm": 0.7045544981956482,
      "learning_rate": 9.624317237886137e-06,
      "loss": 0.7121,
      "step": 4600
    },
    {
      "epoch": 0.25323352963839507,
      "grad_norm": 0.7693355083465576,
      "learning_rate": 9.624152373431672e-06,
      "loss": 0.8052,
      "step": 4601
    },
    {
      "epoch": 0.2532885684407507,
      "grad_norm": 0.8072683811187744,
      "learning_rate": 9.623987474223316e-06,
      "loss": 0.8543,
      "step": 4602
    },
    {
      "epoch": 0.2533436072431064,
      "grad_norm": 0.8158687949180603,
      "learning_rate": 9.62382254026231e-06,
      "loss": 0.6922,
      "step": 4603
    },
    {
      "epoch": 0.25339864604546203,
      "grad_norm": 0.7688641548156738,
      "learning_rate": 9.623657571549887e-06,
      "loss": 0.7198,
      "step": 4604
    },
    {
      "epoch": 0.2534536848478177,
      "grad_norm": 0.7806578278541565,
      "learning_rate": 9.623492568087293e-06,
      "loss": 0.8539,
      "step": 4605
    },
    {
      "epoch": 0.25350872365017335,
      "grad_norm": 0.9557347893714905,
      "learning_rate": 9.623327529875769e-06,
      "loss": 0.6996,
      "step": 4606
    },
    {
      "epoch": 0.25356376245252904,
      "grad_norm": 0.9465067386627197,
      "learning_rate": 9.62316245691655e-06,
      "loss": 0.8756,
      "step": 4607
    },
    {
      "epoch": 0.2536188012548847,
      "grad_norm": 0.8029165863990784,
      "learning_rate": 9.62299734921088e-06,
      "loss": 0.8573,
      "step": 4608
    },
    {
      "epoch": 0.25367384005724036,
      "grad_norm": 0.7530128955841064,
      "learning_rate": 9.62283220676e-06,
      "loss": 0.7466,
      "step": 4609
    },
    {
      "epoch": 0.253728878859596,
      "grad_norm": 0.6704453825950623,
      "learning_rate": 9.622667029565151e-06,
      "loss": 0.6512,
      "step": 4610
    },
    {
      "epoch": 0.2537839176619517,
      "grad_norm": 0.7162728309631348,
      "learning_rate": 9.622501817627574e-06,
      "loss": 0.7615,
      "step": 4611
    },
    {
      "epoch": 0.2538389564643073,
      "grad_norm": 0.7599188089370728,
      "learning_rate": 9.622336570948509e-06,
      "loss": 0.8463,
      "step": 4612
    },
    {
      "epoch": 0.253893995266663,
      "grad_norm": 0.7922326922416687,
      "learning_rate": 9.6221712895292e-06,
      "loss": 0.9221,
      "step": 4613
    },
    {
      "epoch": 0.25394903406901864,
      "grad_norm": 1.4635218381881714,
      "learning_rate": 9.622005973370892e-06,
      "loss": 0.9159,
      "step": 4614
    },
    {
      "epoch": 0.25400407287137433,
      "grad_norm": 0.8695057034492493,
      "learning_rate": 9.62184062247482e-06,
      "loss": 0.6792,
      "step": 4615
    },
    {
      "epoch": 0.25405911167372996,
      "grad_norm": 0.8070930242538452,
      "learning_rate": 9.621675236842235e-06,
      "loss": 0.8257,
      "step": 4616
    },
    {
      "epoch": 0.25411415047608565,
      "grad_norm": 0.8642075061798096,
      "learning_rate": 9.621509816474372e-06,
      "loss": 0.8223,
      "step": 4617
    },
    {
      "epoch": 0.2541691892784413,
      "grad_norm": 0.7131080031394958,
      "learning_rate": 9.621344361372483e-06,
      "loss": 0.6831,
      "step": 4618
    },
    {
      "epoch": 0.254224228080797,
      "grad_norm": 0.7582216262817383,
      "learning_rate": 9.621178871537804e-06,
      "loss": 0.8091,
      "step": 4619
    },
    {
      "epoch": 0.2542792668831526,
      "grad_norm": 0.7705016732215881,
      "learning_rate": 9.62101334697158e-06,
      "loss": 0.7537,
      "step": 4620
    },
    {
      "epoch": 0.2543343056855083,
      "grad_norm": 0.7638342976570129,
      "learning_rate": 9.62084778767506e-06,
      "loss": 0.7661,
      "step": 4621
    },
    {
      "epoch": 0.25438934448786393,
      "grad_norm": 0.9296607971191406,
      "learning_rate": 9.620682193649482e-06,
      "loss": 0.8875,
      "step": 4622
    },
    {
      "epoch": 0.2544443832902196,
      "grad_norm": 0.795394778251648,
      "learning_rate": 9.620516564896096e-06,
      "loss": 0.6884,
      "step": 4623
    },
    {
      "epoch": 0.25449942209257526,
      "grad_norm": 0.9164957404136658,
      "learning_rate": 9.620350901416142e-06,
      "loss": 0.8693,
      "step": 4624
    },
    {
      "epoch": 0.25455446089493095,
      "grad_norm": 0.8306281566619873,
      "learning_rate": 9.62018520321087e-06,
      "loss": 0.8972,
      "step": 4625
    },
    {
      "epoch": 0.2546094996972866,
      "grad_norm": 0.778831422328949,
      "learning_rate": 9.620019470281521e-06,
      "loss": 0.7574,
      "step": 4626
    },
    {
      "epoch": 0.25466453849964227,
      "grad_norm": 0.9326225519180298,
      "learning_rate": 9.619853702629343e-06,
      "loss": 0.7712,
      "step": 4627
    },
    {
      "epoch": 0.2547195773019979,
      "grad_norm": 0.8772255182266235,
      "learning_rate": 9.619687900255581e-06,
      "loss": 0.8241,
      "step": 4628
    },
    {
      "epoch": 0.2547746161043536,
      "grad_norm": 0.8777550458908081,
      "learning_rate": 9.619522063161482e-06,
      "loss": 0.8724,
      "step": 4629
    },
    {
      "epoch": 0.2548296549067092,
      "grad_norm": 0.8332602381706238,
      "learning_rate": 9.61935619134829e-06,
      "loss": 0.8716,
      "step": 4630
    },
    {
      "epoch": 0.2548846937090649,
      "grad_norm": 0.8246355056762695,
      "learning_rate": 9.619190284817255e-06,
      "loss": 0.7789,
      "step": 4631
    },
    {
      "epoch": 0.25493973251142055,
      "grad_norm": 0.7200644612312317,
      "learning_rate": 9.61902434356962e-06,
      "loss": 0.7956,
      "step": 4632
    },
    {
      "epoch": 0.25499477131377624,
      "grad_norm": 0.827756404876709,
      "learning_rate": 9.618858367606638e-06,
      "loss": 0.7925,
      "step": 4633
    },
    {
      "epoch": 0.25504981011613187,
      "grad_norm": 0.7749341726303101,
      "learning_rate": 9.618692356929551e-06,
      "loss": 0.8706,
      "step": 4634
    },
    {
      "epoch": 0.25510484891848756,
      "grad_norm": 0.7233432531356812,
      "learning_rate": 9.618526311539608e-06,
      "loss": 0.7725,
      "step": 4635
    },
    {
      "epoch": 0.2551598877208432,
      "grad_norm": 0.846340537071228,
      "learning_rate": 9.618360231438058e-06,
      "loss": 0.8758,
      "step": 4636
    },
    {
      "epoch": 0.2552149265231989,
      "grad_norm": 0.8262908458709717,
      "learning_rate": 9.61819411662615e-06,
      "loss": 0.7758,
      "step": 4637
    },
    {
      "epoch": 0.2552699653255545,
      "grad_norm": 0.7829110026359558,
      "learning_rate": 9.61802796710513e-06,
      "loss": 0.8494,
      "step": 4638
    },
    {
      "epoch": 0.25532500412791015,
      "grad_norm": 0.7480815649032593,
      "learning_rate": 9.617861782876247e-06,
      "loss": 0.7639,
      "step": 4639
    },
    {
      "epoch": 0.25538004293026584,
      "grad_norm": 0.8782994747161865,
      "learning_rate": 9.617695563940752e-06,
      "loss": 0.9651,
      "step": 4640
    },
    {
      "epoch": 0.25543508173262147,
      "grad_norm": 0.7215868234634399,
      "learning_rate": 9.617529310299895e-06,
      "loss": 0.7833,
      "step": 4641
    },
    {
      "epoch": 0.25549012053497716,
      "grad_norm": 0.8287535905838013,
      "learning_rate": 9.617363021954922e-06,
      "loss": 0.901,
      "step": 4642
    },
    {
      "epoch": 0.2555451593373328,
      "grad_norm": 0.7679935097694397,
      "learning_rate": 9.617196698907084e-06,
      "loss": 0.761,
      "step": 4643
    },
    {
      "epoch": 0.2556001981396885,
      "grad_norm": 0.7765942811965942,
      "learning_rate": 9.617030341157632e-06,
      "loss": 0.7356,
      "step": 4644
    },
    {
      "epoch": 0.2556552369420441,
      "grad_norm": 0.6964583396911621,
      "learning_rate": 9.616863948707816e-06,
      "loss": 0.7683,
      "step": 4645
    },
    {
      "epoch": 0.2557102757443998,
      "grad_norm": 0.8031953573226929,
      "learning_rate": 9.616697521558886e-06,
      "loss": 0.7875,
      "step": 4646
    },
    {
      "epoch": 0.25576531454675544,
      "grad_norm": 0.7155965566635132,
      "learning_rate": 9.616531059712094e-06,
      "loss": 0.6516,
      "step": 4647
    },
    {
      "epoch": 0.25582035334911113,
      "grad_norm": 0.6870070099830627,
      "learning_rate": 9.61636456316869e-06,
      "loss": 0.7217,
      "step": 4648
    },
    {
      "epoch": 0.25587539215146676,
      "grad_norm": 0.7686315774917603,
      "learning_rate": 9.616198031929926e-06,
      "loss": 0.8136,
      "step": 4649
    },
    {
      "epoch": 0.25593043095382245,
      "grad_norm": 0.7532772421836853,
      "learning_rate": 9.616031465997054e-06,
      "loss": 0.696,
      "step": 4650
    },
    {
      "epoch": 0.2559854697561781,
      "grad_norm": 0.8111574053764343,
      "learning_rate": 9.615864865371323e-06,
      "loss": 0.8501,
      "step": 4651
    },
    {
      "epoch": 0.2560405085585338,
      "grad_norm": 0.771065890789032,
      "learning_rate": 9.615698230053989e-06,
      "loss": 0.7417,
      "step": 4652
    },
    {
      "epoch": 0.2560955473608894,
      "grad_norm": 0.7468003034591675,
      "learning_rate": 9.6155315600463e-06,
      "loss": 0.7303,
      "step": 4653
    },
    {
      "epoch": 0.2561505861632451,
      "grad_norm": 0.8041057586669922,
      "learning_rate": 9.615364855349514e-06,
      "loss": 0.8689,
      "step": 4654
    },
    {
      "epoch": 0.25620562496560073,
      "grad_norm": 0.8439033627510071,
      "learning_rate": 9.61519811596488e-06,
      "loss": 0.8654,
      "step": 4655
    },
    {
      "epoch": 0.2562606637679564,
      "grad_norm": 0.7768430113792419,
      "learning_rate": 9.615031341893653e-06,
      "loss": 0.8789,
      "step": 4656
    },
    {
      "epoch": 0.25631570257031205,
      "grad_norm": 0.712876558303833,
      "learning_rate": 9.614864533137086e-06,
      "loss": 0.7497,
      "step": 4657
    },
    {
      "epoch": 0.25637074137266774,
      "grad_norm": 0.7586949467658997,
      "learning_rate": 9.614697689696431e-06,
      "loss": 0.81,
      "step": 4658
    },
    {
      "epoch": 0.2564257801750234,
      "grad_norm": 0.717078447341919,
      "learning_rate": 9.614530811572946e-06,
      "loss": 0.8023,
      "step": 4659
    },
    {
      "epoch": 0.25648081897737907,
      "grad_norm": 0.7369407415390015,
      "learning_rate": 9.61436389876788e-06,
      "loss": 0.784,
      "step": 4660
    },
    {
      "epoch": 0.2565358577797347,
      "grad_norm": 0.7536265850067139,
      "learning_rate": 9.61419695128249e-06,
      "loss": 0.7687,
      "step": 4661
    },
    {
      "epoch": 0.2565908965820904,
      "grad_norm": 0.9718124866485596,
      "learning_rate": 9.614029969118033e-06,
      "loss": 0.8495,
      "step": 4662
    },
    {
      "epoch": 0.256645935384446,
      "grad_norm": 1.1578630208969116,
      "learning_rate": 9.613862952275762e-06,
      "loss": 0.9189,
      "step": 4663
    },
    {
      "epoch": 0.2567009741868017,
      "grad_norm": 0.7752498984336853,
      "learning_rate": 9.613695900756929e-06,
      "loss": 0.7677,
      "step": 4664
    },
    {
      "epoch": 0.25675601298915735,
      "grad_norm": 0.9640393257141113,
      "learning_rate": 9.613528814562795e-06,
      "loss": 0.719,
      "step": 4665
    },
    {
      "epoch": 0.25681105179151303,
      "grad_norm": 0.7690972089767456,
      "learning_rate": 9.613361693694614e-06,
      "loss": 0.7977,
      "step": 4666
    },
    {
      "epoch": 0.25686609059386867,
      "grad_norm": 0.8390190601348877,
      "learning_rate": 9.61319453815364e-06,
      "loss": 0.8032,
      "step": 4667
    },
    {
      "epoch": 0.25692112939622436,
      "grad_norm": 0.8293220400810242,
      "learning_rate": 9.613027347941131e-06,
      "loss": 0.8645,
      "step": 4668
    },
    {
      "epoch": 0.25697616819858,
      "grad_norm": 0.8020731210708618,
      "learning_rate": 9.612860123058344e-06,
      "loss": 0.8374,
      "step": 4669
    },
    {
      "epoch": 0.2570312070009357,
      "grad_norm": 0.7756736278533936,
      "learning_rate": 9.612692863506534e-06,
      "loss": 0.7318,
      "step": 4670
    },
    {
      "epoch": 0.2570862458032913,
      "grad_norm": 0.895416259765625,
      "learning_rate": 9.61252556928696e-06,
      "loss": 0.9654,
      "step": 4671
    },
    {
      "epoch": 0.257141284605647,
      "grad_norm": 0.8647375106811523,
      "learning_rate": 9.61235824040088e-06,
      "loss": 0.7411,
      "step": 4672
    },
    {
      "epoch": 0.25719632340800264,
      "grad_norm": 0.6927250623703003,
      "learning_rate": 9.612190876849546e-06,
      "loss": 0.7558,
      "step": 4673
    },
    {
      "epoch": 0.2572513622103583,
      "grad_norm": 0.7614898085594177,
      "learning_rate": 9.612023478634222e-06,
      "loss": 0.7696,
      "step": 4674
    },
    {
      "epoch": 0.25730640101271396,
      "grad_norm": 0.7910586595535278,
      "learning_rate": 9.611856045756166e-06,
      "loss": 0.8207,
      "step": 4675
    },
    {
      "epoch": 0.25736143981506965,
      "grad_norm": 0.7330125570297241,
      "learning_rate": 9.611688578216632e-06,
      "loss": 0.8615,
      "step": 4676
    },
    {
      "epoch": 0.2574164786174253,
      "grad_norm": 0.7703417539596558,
      "learning_rate": 9.611521076016882e-06,
      "loss": 0.8321,
      "step": 4677
    },
    {
      "epoch": 0.25747151741978097,
      "grad_norm": 0.7121796607971191,
      "learning_rate": 9.611353539158174e-06,
      "loss": 0.8228,
      "step": 4678
    },
    {
      "epoch": 0.2575265562221366,
      "grad_norm": 0.8313117027282715,
      "learning_rate": 9.611185967641768e-06,
      "loss": 0.9012,
      "step": 4679
    },
    {
      "epoch": 0.2575815950244923,
      "grad_norm": 0.806776225566864,
      "learning_rate": 9.61101836146892e-06,
      "loss": 0.769,
      "step": 4680
    },
    {
      "epoch": 0.2576366338268479,
      "grad_norm": 0.7049515843391418,
      "learning_rate": 9.610850720640894e-06,
      "loss": 0.7938,
      "step": 4681
    },
    {
      "epoch": 0.25769167262920356,
      "grad_norm": 0.7286638021469116,
      "learning_rate": 9.610683045158948e-06,
      "loss": 0.8168,
      "step": 4682
    },
    {
      "epoch": 0.25774671143155925,
      "grad_norm": 0.7916898727416992,
      "learning_rate": 9.610515335024345e-06,
      "loss": 0.7681,
      "step": 4683
    },
    {
      "epoch": 0.2578017502339149,
      "grad_norm": 0.7649673819541931,
      "learning_rate": 9.61034759023834e-06,
      "loss": 0.7273,
      "step": 4684
    },
    {
      "epoch": 0.2578567890362706,
      "grad_norm": 0.8280686736106873,
      "learning_rate": 9.610179810802196e-06,
      "loss": 0.7968,
      "step": 4685
    },
    {
      "epoch": 0.2579118278386262,
      "grad_norm": 0.7206569910049438,
      "learning_rate": 9.610011996717175e-06,
      "loss": 0.7359,
      "step": 4686
    },
    {
      "epoch": 0.2579668666409819,
      "grad_norm": 0.7365424036979675,
      "learning_rate": 9.60984414798454e-06,
      "loss": 0.7962,
      "step": 4687
    },
    {
      "epoch": 0.25802190544333753,
      "grad_norm": 0.8030344247817993,
      "learning_rate": 9.609676264605549e-06,
      "loss": 0.7931,
      "step": 4688
    },
    {
      "epoch": 0.2580769442456932,
      "grad_norm": 0.8812693357467651,
      "learning_rate": 9.609508346581464e-06,
      "loss": 0.8493,
      "step": 4689
    },
    {
      "epoch": 0.25813198304804885,
      "grad_norm": 0.8026734590530396,
      "learning_rate": 9.60934039391355e-06,
      "loss": 0.8368,
      "step": 4690
    },
    {
      "epoch": 0.25818702185040454,
      "grad_norm": 0.8270768523216248,
      "learning_rate": 9.609172406603067e-06,
      "loss": 0.9077,
      "step": 4691
    },
    {
      "epoch": 0.2582420606527602,
      "grad_norm": 0.7362856864929199,
      "learning_rate": 9.609004384651276e-06,
      "loss": 0.7384,
      "step": 4692
    },
    {
      "epoch": 0.25829709945511586,
      "grad_norm": 0.7195929288864136,
      "learning_rate": 9.608836328059444e-06,
      "loss": 0.8475,
      "step": 4693
    },
    {
      "epoch": 0.2583521382574715,
      "grad_norm": 0.7653167843818665,
      "learning_rate": 9.60866823682883e-06,
      "loss": 0.7704,
      "step": 4694
    },
    {
      "epoch": 0.2584071770598272,
      "grad_norm": 0.7056792974472046,
      "learning_rate": 9.6085001109607e-06,
      "loss": 0.7835,
      "step": 4695
    },
    {
      "epoch": 0.2584622158621828,
      "grad_norm": 0.7299804091453552,
      "learning_rate": 9.60833195045632e-06,
      "loss": 0.7894,
      "step": 4696
    },
    {
      "epoch": 0.2585172546645385,
      "grad_norm": 0.7235645055770874,
      "learning_rate": 9.608163755316948e-06,
      "loss": 0.8113,
      "step": 4697
    },
    {
      "epoch": 0.25857229346689414,
      "grad_norm": 0.7066782116889954,
      "learning_rate": 9.60799552554385e-06,
      "loss": 0.739,
      "step": 4698
    },
    {
      "epoch": 0.25862733226924983,
      "grad_norm": 0.769930362701416,
      "learning_rate": 9.607827261138291e-06,
      "loss": 0.7565,
      "step": 4699
    },
    {
      "epoch": 0.25868237107160547,
      "grad_norm": 0.8875935077667236,
      "learning_rate": 9.607658962101538e-06,
      "loss": 0.849,
      "step": 4700
    },
    {
      "epoch": 0.25873740987396115,
      "grad_norm": 0.7887380123138428,
      "learning_rate": 9.60749062843485e-06,
      "loss": 0.8795,
      "step": 4701
    },
    {
      "epoch": 0.2587924486763168,
      "grad_norm": 0.7600420117378235,
      "learning_rate": 9.607322260139499e-06,
      "loss": 0.7581,
      "step": 4702
    },
    {
      "epoch": 0.2588474874786725,
      "grad_norm": 0.7431491017341614,
      "learning_rate": 9.607153857216746e-06,
      "loss": 0.7119,
      "step": 4703
    },
    {
      "epoch": 0.2589025262810281,
      "grad_norm": 0.7444193363189697,
      "learning_rate": 9.606985419667858e-06,
      "loss": 0.7492,
      "step": 4704
    },
    {
      "epoch": 0.2589575650833838,
      "grad_norm": 0.8348917365074158,
      "learning_rate": 9.6068169474941e-06,
      "loss": 0.7656,
      "step": 4705
    },
    {
      "epoch": 0.25901260388573943,
      "grad_norm": 0.6790240406990051,
      "learning_rate": 9.60664844069674e-06,
      "loss": 0.6354,
      "step": 4706
    },
    {
      "epoch": 0.2590676426880951,
      "grad_norm": 0.8425769805908203,
      "learning_rate": 9.606479899277044e-06,
      "loss": 0.7927,
      "step": 4707
    },
    {
      "epoch": 0.25912268149045076,
      "grad_norm": 0.7234740853309631,
      "learning_rate": 9.606311323236277e-06,
      "loss": 0.8122,
      "step": 4708
    },
    {
      "epoch": 0.25917772029280645,
      "grad_norm": 0.839507520198822,
      "learning_rate": 9.606142712575707e-06,
      "loss": 0.8807,
      "step": 4709
    },
    {
      "epoch": 0.2592327590951621,
      "grad_norm": 0.7155291438102722,
      "learning_rate": 9.605974067296601e-06,
      "loss": 0.7852,
      "step": 4710
    },
    {
      "epoch": 0.25928779789751777,
      "grad_norm": 0.7222152352333069,
      "learning_rate": 9.605805387400228e-06,
      "loss": 0.7362,
      "step": 4711
    },
    {
      "epoch": 0.2593428366998734,
      "grad_norm": 0.8350114226341248,
      "learning_rate": 9.605636672887854e-06,
      "loss": 0.7201,
      "step": 4712
    },
    {
      "epoch": 0.2593978755022291,
      "grad_norm": 0.6805943250656128,
      "learning_rate": 9.605467923760747e-06,
      "loss": 0.6936,
      "step": 4713
    },
    {
      "epoch": 0.2594529143045847,
      "grad_norm": 0.7863980531692505,
      "learning_rate": 9.605299140020177e-06,
      "loss": 0.9079,
      "step": 4714
    },
    {
      "epoch": 0.2595079531069404,
      "grad_norm": 0.838843584060669,
      "learning_rate": 9.60513032166741e-06,
      "loss": 0.839,
      "step": 4715
    },
    {
      "epoch": 0.25956299190929605,
      "grad_norm": 0.7872797250747681,
      "learning_rate": 9.60496146870372e-06,
      "loss": 0.9164,
      "step": 4716
    },
    {
      "epoch": 0.25961803071165174,
      "grad_norm": 0.7300794720649719,
      "learning_rate": 9.604792581130369e-06,
      "loss": 0.8227,
      "step": 4717
    },
    {
      "epoch": 0.25967306951400737,
      "grad_norm": 0.8420879244804382,
      "learning_rate": 9.60462365894863e-06,
      "loss": 0.7865,
      "step": 4718
    },
    {
      "epoch": 0.25972810831636306,
      "grad_norm": 0.807697057723999,
      "learning_rate": 9.604454702159771e-06,
      "loss": 0.9081,
      "step": 4719
    },
    {
      "epoch": 0.2597831471187187,
      "grad_norm": 0.9041245579719543,
      "learning_rate": 9.604285710765064e-06,
      "loss": 0.8102,
      "step": 4720
    },
    {
      "epoch": 0.2598381859210744,
      "grad_norm": 0.7061690092086792,
      "learning_rate": 9.604116684765779e-06,
      "loss": 0.762,
      "step": 4721
    },
    {
      "epoch": 0.25989322472343,
      "grad_norm": 0.7790346741676331,
      "learning_rate": 9.603947624163186e-06,
      "loss": 0.8038,
      "step": 4722
    },
    {
      "epoch": 0.2599482635257857,
      "grad_norm": 0.8109704256057739,
      "learning_rate": 9.603778528958553e-06,
      "loss": 0.9105,
      "step": 4723
    },
    {
      "epoch": 0.26000330232814134,
      "grad_norm": 0.7396997213363647,
      "learning_rate": 9.603609399153153e-06,
      "loss": 0.8384,
      "step": 4724
    },
    {
      "epoch": 0.260058341130497,
      "grad_norm": 0.8594317436218262,
      "learning_rate": 9.603440234748257e-06,
      "loss": 0.8301,
      "step": 4725
    },
    {
      "epoch": 0.26011337993285266,
      "grad_norm": 0.7087241411209106,
      "learning_rate": 9.603271035745138e-06,
      "loss": 0.6652,
      "step": 4726
    },
    {
      "epoch": 0.2601684187352083,
      "grad_norm": 0.7405440211296082,
      "learning_rate": 9.603101802145065e-06,
      "loss": 0.7804,
      "step": 4727
    },
    {
      "epoch": 0.260223457537564,
      "grad_norm": 0.8637508749961853,
      "learning_rate": 9.602932533949312e-06,
      "loss": 0.8509,
      "step": 4728
    },
    {
      "epoch": 0.2602784963399196,
      "grad_norm": 0.7040451765060425,
      "learning_rate": 9.60276323115915e-06,
      "loss": 0.7842,
      "step": 4729
    },
    {
      "epoch": 0.2603335351422753,
      "grad_norm": 0.7743955254554749,
      "learning_rate": 9.602593893775852e-06,
      "loss": 0.8492,
      "step": 4730
    },
    {
      "epoch": 0.26038857394463094,
      "grad_norm": 0.7110480070114136,
      "learning_rate": 9.602424521800688e-06,
      "loss": 0.7227,
      "step": 4731
    },
    {
      "epoch": 0.26044361274698663,
      "grad_norm": 1.0066583156585693,
      "learning_rate": 9.602255115234936e-06,
      "loss": 0.8825,
      "step": 4732
    },
    {
      "epoch": 0.26049865154934226,
      "grad_norm": 0.7746492624282837,
      "learning_rate": 9.602085674079864e-06,
      "loss": 0.8316,
      "step": 4733
    },
    {
      "epoch": 0.26055369035169795,
      "grad_norm": 0.7394356727600098,
      "learning_rate": 9.60191619833675e-06,
      "loss": 0.746,
      "step": 4734
    },
    {
      "epoch": 0.2606087291540536,
      "grad_norm": 0.7140582203865051,
      "learning_rate": 9.601746688006866e-06,
      "loss": 0.7204,
      "step": 4735
    },
    {
      "epoch": 0.2606637679564093,
      "grad_norm": 0.753399133682251,
      "learning_rate": 9.601577143091483e-06,
      "loss": 0.8157,
      "step": 4736
    },
    {
      "epoch": 0.2607188067587649,
      "grad_norm": 0.674320638179779,
      "learning_rate": 9.601407563591881e-06,
      "loss": 0.7279,
      "step": 4737
    },
    {
      "epoch": 0.2607738455611206,
      "grad_norm": 0.855944037437439,
      "learning_rate": 9.60123794950933e-06,
      "loss": 0.804,
      "step": 4738
    },
    {
      "epoch": 0.26082888436347623,
      "grad_norm": 0.6833948493003845,
      "learning_rate": 9.601068300845106e-06,
      "loss": 0.701,
      "step": 4739
    },
    {
      "epoch": 0.2608839231658319,
      "grad_norm": 0.8085536360740662,
      "learning_rate": 9.600898617600485e-06,
      "loss": 0.8435,
      "step": 4740
    },
    {
      "epoch": 0.26093896196818755,
      "grad_norm": 0.752849817276001,
      "learning_rate": 9.600728899776741e-06,
      "loss": 0.7205,
      "step": 4741
    },
    {
      "epoch": 0.26099400077054324,
      "grad_norm": 0.7320554852485657,
      "learning_rate": 9.600559147375151e-06,
      "loss": 0.7556,
      "step": 4742
    },
    {
      "epoch": 0.2610490395728989,
      "grad_norm": 0.7789202928543091,
      "learning_rate": 9.600389360396988e-06,
      "loss": 0.8467,
      "step": 4743
    },
    {
      "epoch": 0.26110407837525457,
      "grad_norm": 0.8480898141860962,
      "learning_rate": 9.600219538843532e-06,
      "loss": 0.7762,
      "step": 4744
    },
    {
      "epoch": 0.2611591171776102,
      "grad_norm": 0.8382542133331299,
      "learning_rate": 9.600049682716055e-06,
      "loss": 0.9051,
      "step": 4745
    },
    {
      "epoch": 0.2612141559799659,
      "grad_norm": 0.8319274187088013,
      "learning_rate": 9.599879792015838e-06,
      "loss": 0.8221,
      "step": 4746
    },
    {
      "epoch": 0.2612691947823215,
      "grad_norm": 0.7325875163078308,
      "learning_rate": 9.599709866744156e-06,
      "loss": 0.7968,
      "step": 4747
    },
    {
      "epoch": 0.2613242335846772,
      "grad_norm": 0.7053360342979431,
      "learning_rate": 9.599539906902285e-06,
      "loss": 0.7073,
      "step": 4748
    },
    {
      "epoch": 0.26137927238703285,
      "grad_norm": 0.763017475605011,
      "learning_rate": 9.599369912491503e-06,
      "loss": 0.7031,
      "step": 4749
    },
    {
      "epoch": 0.26143431118938854,
      "grad_norm": 0.6816151738166809,
      "learning_rate": 9.599199883513088e-06,
      "loss": 0.7295,
      "step": 4750
    },
    {
      "epoch": 0.26148934999174417,
      "grad_norm": 0.8143941164016724,
      "learning_rate": 9.599029819968319e-06,
      "loss": 0.8449,
      "step": 4751
    },
    {
      "epoch": 0.26154438879409986,
      "grad_norm": 0.8093858361244202,
      "learning_rate": 9.598859721858471e-06,
      "loss": 0.8397,
      "step": 4752
    },
    {
      "epoch": 0.2615994275964555,
      "grad_norm": 0.7431835532188416,
      "learning_rate": 9.598689589184827e-06,
      "loss": 0.7299,
      "step": 4753
    },
    {
      "epoch": 0.2616544663988112,
      "grad_norm": 0.9871510863304138,
      "learning_rate": 9.59851942194866e-06,
      "loss": 0.7992,
      "step": 4754
    },
    {
      "epoch": 0.2617095052011668,
      "grad_norm": 0.9304273724555969,
      "learning_rate": 9.598349220151254e-06,
      "loss": 0.7519,
      "step": 4755
    },
    {
      "epoch": 0.2617645440035225,
      "grad_norm": 0.9361812472343445,
      "learning_rate": 9.598178983793886e-06,
      "loss": 0.8131,
      "step": 4756
    },
    {
      "epoch": 0.26181958280587814,
      "grad_norm": 0.7783429622650146,
      "learning_rate": 9.598008712877835e-06,
      "loss": 0.7351,
      "step": 4757
    },
    {
      "epoch": 0.2618746216082338,
      "grad_norm": 0.8739376068115234,
      "learning_rate": 9.597838407404381e-06,
      "loss": 0.9458,
      "step": 4758
    },
    {
      "epoch": 0.26192966041058946,
      "grad_norm": 0.7076277732849121,
      "learning_rate": 9.597668067374805e-06,
      "loss": 0.7632,
      "step": 4759
    },
    {
      "epoch": 0.26198469921294515,
      "grad_norm": 0.7652345299720764,
      "learning_rate": 9.597497692790386e-06,
      "loss": 0.8018,
      "step": 4760
    },
    {
      "epoch": 0.2620397380153008,
      "grad_norm": 0.7332149147987366,
      "learning_rate": 9.597327283652405e-06,
      "loss": 0.8223,
      "step": 4761
    },
    {
      "epoch": 0.26209477681765647,
      "grad_norm": 0.8361638784408569,
      "learning_rate": 9.597156839962145e-06,
      "loss": 0.8784,
      "step": 4762
    },
    {
      "epoch": 0.2621498156200121,
      "grad_norm": 1.183772325515747,
      "learning_rate": 9.596986361720882e-06,
      "loss": 0.8768,
      "step": 4763
    },
    {
      "epoch": 0.2622048544223678,
      "grad_norm": 0.9895418882369995,
      "learning_rate": 9.596815848929902e-06,
      "loss": 0.714,
      "step": 4764
    },
    {
      "epoch": 0.26225989322472343,
      "grad_norm": 0.8210558295249939,
      "learning_rate": 9.59664530159048e-06,
      "loss": 0.7246,
      "step": 4765
    },
    {
      "epoch": 0.2623149320270791,
      "grad_norm": 0.8003455996513367,
      "learning_rate": 9.596474719703908e-06,
      "loss": 0.8385,
      "step": 4766
    },
    {
      "epoch": 0.26236997082943475,
      "grad_norm": 0.7555826306343079,
      "learning_rate": 9.59630410327146e-06,
      "loss": 0.7243,
      "step": 4767
    },
    {
      "epoch": 0.2624250096317904,
      "grad_norm": 0.7746273279190063,
      "learning_rate": 9.596133452294421e-06,
      "loss": 0.8763,
      "step": 4768
    },
    {
      "epoch": 0.2624800484341461,
      "grad_norm": 0.7238507866859436,
      "learning_rate": 9.595962766774074e-06,
      "loss": 0.8302,
      "step": 4769
    },
    {
      "epoch": 0.2625350872365017,
      "grad_norm": 0.7874132394790649,
      "learning_rate": 9.595792046711699e-06,
      "loss": 0.7979,
      "step": 4770
    },
    {
      "epoch": 0.2625901260388574,
      "grad_norm": 0.8792033791542053,
      "learning_rate": 9.595621292108583e-06,
      "loss": 0.8555,
      "step": 4771
    },
    {
      "epoch": 0.26264516484121303,
      "grad_norm": 0.7026945948600769,
      "learning_rate": 9.595450502966006e-06,
      "loss": 0.718,
      "step": 4772
    },
    {
      "epoch": 0.2627002036435687,
      "grad_norm": 0.7747959494590759,
      "learning_rate": 9.595279679285254e-06,
      "loss": 0.8329,
      "step": 4773
    },
    {
      "epoch": 0.26275524244592435,
      "grad_norm": 0.697979748249054,
      "learning_rate": 9.59510882106761e-06,
      "loss": 0.7456,
      "step": 4774
    },
    {
      "epoch": 0.26281028124828004,
      "grad_norm": 0.7600447535514832,
      "learning_rate": 9.594937928314359e-06,
      "loss": 0.875,
      "step": 4775
    },
    {
      "epoch": 0.2628653200506357,
      "grad_norm": 0.7591384649276733,
      "learning_rate": 9.594767001026783e-06,
      "loss": 0.7607,
      "step": 4776
    },
    {
      "epoch": 0.26292035885299136,
      "grad_norm": 0.9267380833625793,
      "learning_rate": 9.59459603920617e-06,
      "loss": 0.8926,
      "step": 4777
    },
    {
      "epoch": 0.262975397655347,
      "grad_norm": 0.7751328349113464,
      "learning_rate": 9.594425042853802e-06,
      "loss": 0.7449,
      "step": 4778
    },
    {
      "epoch": 0.2630304364577027,
      "grad_norm": 0.7066012620925903,
      "learning_rate": 9.594254011970966e-06,
      "loss": 0.8374,
      "step": 4779
    },
    {
      "epoch": 0.2630854752600583,
      "grad_norm": 0.7564317584037781,
      "learning_rate": 9.594082946558945e-06,
      "loss": 0.735,
      "step": 4780
    },
    {
      "epoch": 0.263140514062414,
      "grad_norm": 0.8151416182518005,
      "learning_rate": 9.593911846619027e-06,
      "loss": 0.8575,
      "step": 4781
    },
    {
      "epoch": 0.26319555286476964,
      "grad_norm": 0.719261646270752,
      "learning_rate": 9.593740712152497e-06,
      "loss": 0.7981,
      "step": 4782
    },
    {
      "epoch": 0.26325059166712533,
      "grad_norm": 0.8627344369888306,
      "learning_rate": 9.593569543160642e-06,
      "loss": 0.895,
      "step": 4783
    },
    {
      "epoch": 0.26330563046948097,
      "grad_norm": 1.293272614479065,
      "learning_rate": 9.593398339644748e-06,
      "loss": 0.7531,
      "step": 4784
    },
    {
      "epoch": 0.26336066927183666,
      "grad_norm": 0.8475207686424255,
      "learning_rate": 9.593227101606102e-06,
      "loss": 0.9091,
      "step": 4785
    },
    {
      "epoch": 0.2634157080741923,
      "grad_norm": 0.78054279088974,
      "learning_rate": 9.593055829045989e-06,
      "loss": 0.7692,
      "step": 4786
    },
    {
      "epoch": 0.263470746876548,
      "grad_norm": 0.7677399516105652,
      "learning_rate": 9.592884521965699e-06,
      "loss": 0.6232,
      "step": 4787
    },
    {
      "epoch": 0.2635257856789036,
      "grad_norm": 0.7232677340507507,
      "learning_rate": 9.59271318036652e-06,
      "loss": 0.8087,
      "step": 4788
    },
    {
      "epoch": 0.2635808244812593,
      "grad_norm": 0.8728463649749756,
      "learning_rate": 9.592541804249735e-06,
      "loss": 0.7824,
      "step": 4789
    },
    {
      "epoch": 0.26363586328361494,
      "grad_norm": 0.7569910883903503,
      "learning_rate": 9.592370393616637e-06,
      "loss": 0.7418,
      "step": 4790
    },
    {
      "epoch": 0.2636909020859706,
      "grad_norm": 0.7631934285163879,
      "learning_rate": 9.592198948468511e-06,
      "loss": 0.7929,
      "step": 4791
    },
    {
      "epoch": 0.26374594088832626,
      "grad_norm": 0.8021631240844727,
      "learning_rate": 9.592027468806649e-06,
      "loss": 0.8111,
      "step": 4792
    },
    {
      "epoch": 0.26380097969068195,
      "grad_norm": 0.9454651474952698,
      "learning_rate": 9.591855954632336e-06,
      "loss": 0.8239,
      "step": 4793
    },
    {
      "epoch": 0.2638560184930376,
      "grad_norm": 0.672924280166626,
      "learning_rate": 9.591684405946863e-06,
      "loss": 0.6877,
      "step": 4794
    },
    {
      "epoch": 0.26391105729539327,
      "grad_norm": 0.7942802906036377,
      "learning_rate": 9.59151282275152e-06,
      "loss": 0.9002,
      "step": 4795
    },
    {
      "epoch": 0.2639660960977489,
      "grad_norm": 0.7131155133247375,
      "learning_rate": 9.591341205047596e-06,
      "loss": 0.7692,
      "step": 4796
    },
    {
      "epoch": 0.2640211349001046,
      "grad_norm": 1.0395869016647339,
      "learning_rate": 9.59116955283638e-06,
      "loss": 0.8352,
      "step": 4797
    },
    {
      "epoch": 0.2640761737024602,
      "grad_norm": 0.9503256678581238,
      "learning_rate": 9.590997866119163e-06,
      "loss": 1.0287,
      "step": 4798
    },
    {
      "epoch": 0.2641312125048159,
      "grad_norm": 0.7539612054824829,
      "learning_rate": 9.590826144897235e-06,
      "loss": 0.872,
      "step": 4799
    },
    {
      "epoch": 0.26418625130717155,
      "grad_norm": 0.7067893743515015,
      "learning_rate": 9.590654389171885e-06,
      "loss": 0.7636,
      "step": 4800
    },
    {
      "epoch": 0.26424129010952724,
      "grad_norm": 0.7355281710624695,
      "learning_rate": 9.590482598944407e-06,
      "loss": 0.7715,
      "step": 4801
    },
    {
      "epoch": 0.26429632891188287,
      "grad_norm": 0.7589674592018127,
      "learning_rate": 9.590310774216089e-06,
      "loss": 0.7451,
      "step": 4802
    },
    {
      "epoch": 0.26435136771423856,
      "grad_norm": 0.701386034488678,
      "learning_rate": 9.590138914988226e-06,
      "loss": 0.7317,
      "step": 4803
    },
    {
      "epoch": 0.2644064065165942,
      "grad_norm": 0.7663118243217468,
      "learning_rate": 9.589967021262105e-06,
      "loss": 0.8227,
      "step": 4804
    },
    {
      "epoch": 0.2644614453189499,
      "grad_norm": 0.7059655785560608,
      "learning_rate": 9.589795093039023e-06,
      "loss": 0.7829,
      "step": 4805
    },
    {
      "epoch": 0.2645164841213055,
      "grad_norm": 0.7377020120620728,
      "learning_rate": 9.58962313032027e-06,
      "loss": 0.8308,
      "step": 4806
    },
    {
      "epoch": 0.2645715229236612,
      "grad_norm": 0.8635388612747192,
      "learning_rate": 9.589451133107134e-06,
      "loss": 0.7882,
      "step": 4807
    },
    {
      "epoch": 0.26462656172601684,
      "grad_norm": 0.8282824754714966,
      "learning_rate": 9.589279101400915e-06,
      "loss": 0.8055,
      "step": 4808
    },
    {
      "epoch": 0.26468160052837253,
      "grad_norm": 0.7026814818382263,
      "learning_rate": 9.589107035202903e-06,
      "loss": 0.7567,
      "step": 4809
    },
    {
      "epoch": 0.26473663933072816,
      "grad_norm": 0.7575708031654358,
      "learning_rate": 9.588934934514392e-06,
      "loss": 0.7456,
      "step": 4810
    },
    {
      "epoch": 0.2647916781330838,
      "grad_norm": 0.9732069969177246,
      "learning_rate": 9.588762799336671e-06,
      "loss": 0.8217,
      "step": 4811
    },
    {
      "epoch": 0.2648467169354395,
      "grad_norm": 0.786803126335144,
      "learning_rate": 9.58859062967104e-06,
      "loss": 0.729,
      "step": 4812
    },
    {
      "epoch": 0.2649017557377951,
      "grad_norm": 0.8068973422050476,
      "learning_rate": 9.588418425518789e-06,
      "loss": 0.8204,
      "step": 4813
    },
    {
      "epoch": 0.2649567945401508,
      "grad_norm": 0.8222702145576477,
      "learning_rate": 9.588246186881213e-06,
      "loss": 0.8349,
      "step": 4814
    },
    {
      "epoch": 0.26501183334250644,
      "grad_norm": 0.7560802698135376,
      "learning_rate": 9.588073913759608e-06,
      "loss": 0.7601,
      "step": 4815
    },
    {
      "epoch": 0.26506687214486213,
      "grad_norm": 0.9221365451812744,
      "learning_rate": 9.587901606155266e-06,
      "loss": 0.7725,
      "step": 4816
    },
    {
      "epoch": 0.26512191094721776,
      "grad_norm": 0.8092262744903564,
      "learning_rate": 9.587729264069485e-06,
      "loss": 0.9074,
      "step": 4817
    },
    {
      "epoch": 0.26517694974957345,
      "grad_norm": 0.8183920979499817,
      "learning_rate": 9.587556887503557e-06,
      "loss": 0.8321,
      "step": 4818
    },
    {
      "epoch": 0.2652319885519291,
      "grad_norm": 0.7023420929908752,
      "learning_rate": 9.587384476458781e-06,
      "loss": 0.7842,
      "step": 4819
    },
    {
      "epoch": 0.2652870273542848,
      "grad_norm": 1.2864880561828613,
      "learning_rate": 9.58721203093645e-06,
      "loss": 0.7519,
      "step": 4820
    },
    {
      "epoch": 0.2653420661566404,
      "grad_norm": 0.8133784532546997,
      "learning_rate": 9.587039550937864e-06,
      "loss": 0.8208,
      "step": 4821
    },
    {
      "epoch": 0.2653971049589961,
      "grad_norm": 0.739732027053833,
      "learning_rate": 9.586867036464314e-06,
      "loss": 0.8553,
      "step": 4822
    },
    {
      "epoch": 0.26545214376135173,
      "grad_norm": 0.7539162635803223,
      "learning_rate": 9.5866944875171e-06,
      "loss": 0.7385,
      "step": 4823
    },
    {
      "epoch": 0.2655071825637074,
      "grad_norm": 0.8012336492538452,
      "learning_rate": 9.58652190409752e-06,
      "loss": 0.8343,
      "step": 4824
    },
    {
      "epoch": 0.26556222136606306,
      "grad_norm": 0.7972521185874939,
      "learning_rate": 9.586349286206865e-06,
      "loss": 0.8481,
      "step": 4825
    },
    {
      "epoch": 0.26561726016841875,
      "grad_norm": 0.7772900462150574,
      "learning_rate": 9.58617663384644e-06,
      "loss": 0.7655,
      "step": 4826
    },
    {
      "epoch": 0.2656722989707744,
      "grad_norm": 0.677916944026947,
      "learning_rate": 9.586003947017537e-06,
      "loss": 0.696,
      "step": 4827
    },
    {
      "epoch": 0.26572733777313007,
      "grad_norm": 0.8254117369651794,
      "learning_rate": 9.585831225721455e-06,
      "loss": 0.7841,
      "step": 4828
    },
    {
      "epoch": 0.2657823765754857,
      "grad_norm": 0.7256904244422913,
      "learning_rate": 9.585658469959496e-06,
      "loss": 0.8057,
      "step": 4829
    },
    {
      "epoch": 0.2658374153778414,
      "grad_norm": 0.7651757001876831,
      "learning_rate": 9.585485679732953e-06,
      "loss": 0.7918,
      "step": 4830
    },
    {
      "epoch": 0.265892454180197,
      "grad_norm": 0.7581052184104919,
      "learning_rate": 9.58531285504313e-06,
      "loss": 0.759,
      "step": 4831
    },
    {
      "epoch": 0.2659474929825527,
      "grad_norm": 0.7190486192703247,
      "learning_rate": 9.58513999589132e-06,
      "loss": 0.7403,
      "step": 4832
    },
    {
      "epoch": 0.26600253178490835,
      "grad_norm": 0.8603141903877258,
      "learning_rate": 9.584967102278825e-06,
      "loss": 0.8944,
      "step": 4833
    },
    {
      "epoch": 0.26605757058726404,
      "grad_norm": 0.806297779083252,
      "learning_rate": 9.584794174206947e-06,
      "loss": 0.7039,
      "step": 4834
    },
    {
      "epoch": 0.26611260938961967,
      "grad_norm": 0.7604451775550842,
      "learning_rate": 9.584621211676981e-06,
      "loss": 0.8076,
      "step": 4835
    },
    {
      "epoch": 0.26616764819197536,
      "grad_norm": 0.7276773452758789,
      "learning_rate": 9.584448214690232e-06,
      "loss": 0.786,
      "step": 4836
    },
    {
      "epoch": 0.266222686994331,
      "grad_norm": 0.8737080693244934,
      "learning_rate": 9.584275183247994e-06,
      "loss": 0.8071,
      "step": 4837
    },
    {
      "epoch": 0.2662777257966867,
      "grad_norm": 0.8447219133377075,
      "learning_rate": 9.584102117351574e-06,
      "loss": 0.7682,
      "step": 4838
    },
    {
      "epoch": 0.2663327645990423,
      "grad_norm": 0.7001703381538391,
      "learning_rate": 9.583929017002268e-06,
      "loss": 0.7077,
      "step": 4839
    },
    {
      "epoch": 0.266387803401398,
      "grad_norm": 0.7935730218887329,
      "learning_rate": 9.583755882201377e-06,
      "loss": 0.8122,
      "step": 4840
    },
    {
      "epoch": 0.26644284220375364,
      "grad_norm": 0.8763312697410583,
      "learning_rate": 9.583582712950207e-06,
      "loss": 0.8241,
      "step": 4841
    },
    {
      "epoch": 0.2664978810061093,
      "grad_norm": 0.7910245656967163,
      "learning_rate": 9.583409509250055e-06,
      "loss": 0.7717,
      "step": 4842
    },
    {
      "epoch": 0.26655291980846496,
      "grad_norm": 0.7975226640701294,
      "learning_rate": 9.583236271102222e-06,
      "loss": 0.7165,
      "step": 4843
    },
    {
      "epoch": 0.26660795861082065,
      "grad_norm": 0.8060342073440552,
      "learning_rate": 9.583062998508014e-06,
      "loss": 0.7659,
      "step": 4844
    },
    {
      "epoch": 0.2666629974131763,
      "grad_norm": 0.8779375553131104,
      "learning_rate": 9.582889691468732e-06,
      "loss": 0.8207,
      "step": 4845
    },
    {
      "epoch": 0.266718036215532,
      "grad_norm": 0.7409310936927795,
      "learning_rate": 9.582716349985677e-06,
      "loss": 0.8439,
      "step": 4846
    },
    {
      "epoch": 0.2667730750178876,
      "grad_norm": 0.8871899843215942,
      "learning_rate": 9.582542974060152e-06,
      "loss": 0.8305,
      "step": 4847
    },
    {
      "epoch": 0.2668281138202433,
      "grad_norm": 0.9003115296363831,
      "learning_rate": 9.58236956369346e-06,
      "loss": 0.8334,
      "step": 4848
    },
    {
      "epoch": 0.26688315262259893,
      "grad_norm": 1.0149577856063843,
      "learning_rate": 9.582196118886909e-06,
      "loss": 0.7962,
      "step": 4849
    },
    {
      "epoch": 0.2669381914249546,
      "grad_norm": 0.785214900970459,
      "learning_rate": 9.582022639641795e-06,
      "loss": 0.7806,
      "step": 4850
    },
    {
      "epoch": 0.26699323022731025,
      "grad_norm": 0.9833952188491821,
      "learning_rate": 9.581849125959426e-06,
      "loss": 0.7607,
      "step": 4851
    },
    {
      "epoch": 0.26704826902966594,
      "grad_norm": 1.404751181602478,
      "learning_rate": 9.581675577841104e-06,
      "loss": 0.9046,
      "step": 4852
    },
    {
      "epoch": 0.2671033078320216,
      "grad_norm": 0.791159451007843,
      "learning_rate": 9.581501995288137e-06,
      "loss": 0.6582,
      "step": 4853
    },
    {
      "epoch": 0.2671583466343772,
      "grad_norm": 0.8507272005081177,
      "learning_rate": 9.581328378301827e-06,
      "loss": 0.8946,
      "step": 4854
    },
    {
      "epoch": 0.2672133854367329,
      "grad_norm": 0.7372786998748779,
      "learning_rate": 9.58115472688348e-06,
      "loss": 0.7865,
      "step": 4855
    },
    {
      "epoch": 0.26726842423908853,
      "grad_norm": 0.8293853998184204,
      "learning_rate": 9.580981041034398e-06,
      "loss": 0.9113,
      "step": 4856
    },
    {
      "epoch": 0.2673234630414442,
      "grad_norm": 0.7212402820587158,
      "learning_rate": 9.580807320755889e-06,
      "loss": 0.7149,
      "step": 4857
    },
    {
      "epoch": 0.26737850184379985,
      "grad_norm": 0.7885197401046753,
      "learning_rate": 9.58063356604926e-06,
      "loss": 0.8651,
      "step": 4858
    },
    {
      "epoch": 0.26743354064615554,
      "grad_norm": 0.8444308042526245,
      "learning_rate": 9.580459776915814e-06,
      "loss": 0.7968,
      "step": 4859
    },
    {
      "epoch": 0.2674885794485112,
      "grad_norm": 0.7974254488945007,
      "learning_rate": 9.58028595335686e-06,
      "loss": 0.8499,
      "step": 4860
    },
    {
      "epoch": 0.26754361825086687,
      "grad_norm": 0.7491242289543152,
      "learning_rate": 9.580112095373702e-06,
      "loss": 0.8278,
      "step": 4861
    },
    {
      "epoch": 0.2675986570532225,
      "grad_norm": 0.6856499314308167,
      "learning_rate": 9.579938202967646e-06,
      "loss": 0.7466,
      "step": 4862
    },
    {
      "epoch": 0.2676536958555782,
      "grad_norm": 0.7347447872161865,
      "learning_rate": 9.579764276140002e-06,
      "loss": 0.8046,
      "step": 4863
    },
    {
      "epoch": 0.2677087346579338,
      "grad_norm": 0.6797083020210266,
      "learning_rate": 9.579590314892077e-06,
      "loss": 0.7012,
      "step": 4864
    },
    {
      "epoch": 0.2677637734602895,
      "grad_norm": 0.8219562768936157,
      "learning_rate": 9.579416319225175e-06,
      "loss": 0.7592,
      "step": 4865
    },
    {
      "epoch": 0.26781881226264515,
      "grad_norm": 0.7388357520103455,
      "learning_rate": 9.579242289140607e-06,
      "loss": 0.8179,
      "step": 4866
    },
    {
      "epoch": 0.26787385106500083,
      "grad_norm": 0.7394490838050842,
      "learning_rate": 9.579068224639679e-06,
      "loss": 0.694,
      "step": 4867
    },
    {
      "epoch": 0.26792888986735647,
      "grad_norm": 0.7309017181396484,
      "learning_rate": 9.578894125723699e-06,
      "loss": 0.7882,
      "step": 4868
    },
    {
      "epoch": 0.26798392866971216,
      "grad_norm": 0.7785035967826843,
      "learning_rate": 9.578719992393978e-06,
      "loss": 0.8142,
      "step": 4869
    },
    {
      "epoch": 0.2680389674720678,
      "grad_norm": 0.8983079195022583,
      "learning_rate": 9.57854582465182e-06,
      "loss": 0.7809,
      "step": 4870
    },
    {
      "epoch": 0.2680940062744235,
      "grad_norm": 0.7433765530586243,
      "learning_rate": 9.578371622498542e-06,
      "loss": 0.8937,
      "step": 4871
    },
    {
      "epoch": 0.2681490450767791,
      "grad_norm": 0.8808667659759521,
      "learning_rate": 9.578197385935446e-06,
      "loss": 0.7821,
      "step": 4872
    },
    {
      "epoch": 0.2682040838791348,
      "grad_norm": 0.825794517993927,
      "learning_rate": 9.578023114963843e-06,
      "loss": 0.8228,
      "step": 4873
    },
    {
      "epoch": 0.26825912268149044,
      "grad_norm": 1.0165129899978638,
      "learning_rate": 9.577848809585046e-06,
      "loss": 0.7964,
      "step": 4874
    },
    {
      "epoch": 0.2683141614838461,
      "grad_norm": 0.742028534412384,
      "learning_rate": 9.577674469800362e-06,
      "loss": 0.9126,
      "step": 4875
    },
    {
      "epoch": 0.26836920028620176,
      "grad_norm": 0.7571890354156494,
      "learning_rate": 9.577500095611101e-06,
      "loss": 0.879,
      "step": 4876
    },
    {
      "epoch": 0.26842423908855745,
      "grad_norm": 0.7577160596847534,
      "learning_rate": 9.577325687018575e-06,
      "loss": 0.8048,
      "step": 4877
    },
    {
      "epoch": 0.2684792778909131,
      "grad_norm": 0.7704411745071411,
      "learning_rate": 9.577151244024095e-06,
      "loss": 0.7451,
      "step": 4878
    },
    {
      "epoch": 0.26853431669326877,
      "grad_norm": 0.8323166966438293,
      "learning_rate": 9.57697676662897e-06,
      "loss": 0.7591,
      "step": 4879
    },
    {
      "epoch": 0.2685893554956244,
      "grad_norm": 0.7257028222084045,
      "learning_rate": 9.576802254834516e-06,
      "loss": 0.7941,
      "step": 4880
    },
    {
      "epoch": 0.2686443942979801,
      "grad_norm": 0.8170442581176758,
      "learning_rate": 9.57662770864204e-06,
      "loss": 0.8617,
      "step": 4881
    },
    {
      "epoch": 0.2686994331003357,
      "grad_norm": 0.7435339689254761,
      "learning_rate": 9.576453128052852e-06,
      "loss": 0.7683,
      "step": 4882
    },
    {
      "epoch": 0.2687544719026914,
      "grad_norm": 0.7932955026626587,
      "learning_rate": 9.576278513068271e-06,
      "loss": 0.7103,
      "step": 4883
    },
    {
      "epoch": 0.26880951070504705,
      "grad_norm": 0.8008469939231873,
      "learning_rate": 9.576103863689604e-06,
      "loss": 0.8144,
      "step": 4884
    },
    {
      "epoch": 0.26886454950740274,
      "grad_norm": 0.8573774695396423,
      "learning_rate": 9.575929179918167e-06,
      "loss": 0.8992,
      "step": 4885
    },
    {
      "epoch": 0.2689195883097584,
      "grad_norm": 0.7326993942260742,
      "learning_rate": 9.57575446175527e-06,
      "loss": 0.699,
      "step": 4886
    },
    {
      "epoch": 0.26897462711211406,
      "grad_norm": 0.8249791264533997,
      "learning_rate": 9.575579709202228e-06,
      "loss": 0.7445,
      "step": 4887
    },
    {
      "epoch": 0.2690296659144697,
      "grad_norm": 0.7136644124984741,
      "learning_rate": 9.575404922260351e-06,
      "loss": 0.779,
      "step": 4888
    },
    {
      "epoch": 0.2690847047168254,
      "grad_norm": 1.0130438804626465,
      "learning_rate": 9.575230100930958e-06,
      "loss": 0.8535,
      "step": 4889
    },
    {
      "epoch": 0.269139743519181,
      "grad_norm": 0.6784926652908325,
      "learning_rate": 9.575055245215358e-06,
      "loss": 0.6745,
      "step": 4890
    },
    {
      "epoch": 0.2691947823215367,
      "grad_norm": 0.7492508888244629,
      "learning_rate": 9.57488035511487e-06,
      "loss": 0.6748,
      "step": 4891
    },
    {
      "epoch": 0.26924982112389234,
      "grad_norm": 0.7951217889785767,
      "learning_rate": 9.574705430630807e-06,
      "loss": 0.8119,
      "step": 4892
    },
    {
      "epoch": 0.26930485992624803,
      "grad_norm": 0.9756677746772766,
      "learning_rate": 9.574530471764478e-06,
      "loss": 0.855,
      "step": 4893
    },
    {
      "epoch": 0.26935989872860366,
      "grad_norm": 0.7806811928749084,
      "learning_rate": 9.574355478517206e-06,
      "loss": 0.8432,
      "step": 4894
    },
    {
      "epoch": 0.26941493753095935,
      "grad_norm": 0.7814774513244629,
      "learning_rate": 9.574180450890301e-06,
      "loss": 0.8226,
      "step": 4895
    },
    {
      "epoch": 0.269469976333315,
      "grad_norm": 0.7745325565338135,
      "learning_rate": 9.574005388885081e-06,
      "loss": 0.7722,
      "step": 4896
    },
    {
      "epoch": 0.2695250151356706,
      "grad_norm": 0.7805666327476501,
      "learning_rate": 9.573830292502862e-06,
      "loss": 0.8357,
      "step": 4897
    },
    {
      "epoch": 0.2695800539380263,
      "grad_norm": 0.8428031802177429,
      "learning_rate": 9.573655161744958e-06,
      "loss": 0.8056,
      "step": 4898
    },
    {
      "epoch": 0.26963509274038194,
      "grad_norm": 0.7896600961685181,
      "learning_rate": 9.573479996612684e-06,
      "loss": 0.7984,
      "step": 4899
    },
    {
      "epoch": 0.26969013154273763,
      "grad_norm": 0.7718683481216431,
      "learning_rate": 9.57330479710736e-06,
      "loss": 0.7527,
      "step": 4900
    },
    {
      "epoch": 0.26974517034509327,
      "grad_norm": 0.7868129014968872,
      "learning_rate": 9.573129563230302e-06,
      "loss": 0.7876,
      "step": 4901
    },
    {
      "epoch": 0.26980020914744895,
      "grad_norm": 0.8493777513504028,
      "learning_rate": 9.572954294982826e-06,
      "loss": 0.864,
      "step": 4902
    },
    {
      "epoch": 0.2698552479498046,
      "grad_norm": 0.7492502331733704,
      "learning_rate": 9.57277899236625e-06,
      "loss": 0.8236,
      "step": 4903
    },
    {
      "epoch": 0.2699102867521603,
      "grad_norm": 1.0534250736236572,
      "learning_rate": 9.57260365538189e-06,
      "loss": 0.8012,
      "step": 4904
    },
    {
      "epoch": 0.2699653255545159,
      "grad_norm": 0.7557470202445984,
      "learning_rate": 9.572428284031065e-06,
      "loss": 0.9084,
      "step": 4905
    },
    {
      "epoch": 0.2700203643568716,
      "grad_norm": 0.8055123686790466,
      "learning_rate": 9.572252878315094e-06,
      "loss": 0.7468,
      "step": 4906
    },
    {
      "epoch": 0.27007540315922723,
      "grad_norm": 0.8399039506912231,
      "learning_rate": 9.572077438235294e-06,
      "loss": 0.9293,
      "step": 4907
    },
    {
      "epoch": 0.2701304419615829,
      "grad_norm": 0.9800041317939758,
      "learning_rate": 9.571901963792983e-06,
      "loss": 0.8664,
      "step": 4908
    },
    {
      "epoch": 0.27018548076393856,
      "grad_norm": 0.7732129096984863,
      "learning_rate": 9.571726454989482e-06,
      "loss": 0.7227,
      "step": 4909
    },
    {
      "epoch": 0.27024051956629425,
      "grad_norm": 0.730754017829895,
      "learning_rate": 9.571550911826109e-06,
      "loss": 0.6467,
      "step": 4910
    },
    {
      "epoch": 0.2702955583686499,
      "grad_norm": 0.8245325684547424,
      "learning_rate": 9.57137533430418e-06,
      "loss": 0.7847,
      "step": 4911
    },
    {
      "epoch": 0.27035059717100557,
      "grad_norm": 0.8606786131858826,
      "learning_rate": 9.57119972242502e-06,
      "loss": 0.9556,
      "step": 4912
    },
    {
      "epoch": 0.2704056359733612,
      "grad_norm": 0.7480195164680481,
      "learning_rate": 9.571024076189947e-06,
      "loss": 0.8504,
      "step": 4913
    },
    {
      "epoch": 0.2704606747757169,
      "grad_norm": 0.718913197517395,
      "learning_rate": 9.57084839560028e-06,
      "loss": 0.7869,
      "step": 4914
    },
    {
      "epoch": 0.2705157135780725,
      "grad_norm": 0.9778180122375488,
      "learning_rate": 9.57067268065734e-06,
      "loss": 0.8514,
      "step": 4915
    },
    {
      "epoch": 0.2705707523804282,
      "grad_norm": 0.7394844889640808,
      "learning_rate": 9.570496931362448e-06,
      "loss": 0.7906,
      "step": 4916
    },
    {
      "epoch": 0.27062579118278385,
      "grad_norm": 0.7648600339889526,
      "learning_rate": 9.570321147716923e-06,
      "loss": 0.8194,
      "step": 4917
    },
    {
      "epoch": 0.27068082998513954,
      "grad_norm": 0.8002632260322571,
      "learning_rate": 9.57014532972209e-06,
      "loss": 0.8079,
      "step": 4918
    },
    {
      "epoch": 0.27073586878749517,
      "grad_norm": 0.8668341040611267,
      "learning_rate": 9.569969477379267e-06,
      "loss": 0.8954,
      "step": 4919
    },
    {
      "epoch": 0.27079090758985086,
      "grad_norm": 0.7403327226638794,
      "learning_rate": 9.569793590689775e-06,
      "loss": 0.7755,
      "step": 4920
    },
    {
      "epoch": 0.2708459463922065,
      "grad_norm": 0.7399682998657227,
      "learning_rate": 9.569617669654938e-06,
      "loss": 0.8203,
      "step": 4921
    },
    {
      "epoch": 0.2709009851945622,
      "grad_norm": 0.788600504398346,
      "learning_rate": 9.56944171427608e-06,
      "loss": 0.7565,
      "step": 4922
    },
    {
      "epoch": 0.2709560239969178,
      "grad_norm": 0.7044861912727356,
      "learning_rate": 9.56926572455452e-06,
      "loss": 0.7073,
      "step": 4923
    },
    {
      "epoch": 0.2710110627992735,
      "grad_norm": 0.8195114135742188,
      "learning_rate": 9.569089700491581e-06,
      "loss": 0.8658,
      "step": 4924
    },
    {
      "epoch": 0.27106610160162914,
      "grad_norm": 0.7792258858680725,
      "learning_rate": 9.568913642088589e-06,
      "loss": 0.8628,
      "step": 4925
    },
    {
      "epoch": 0.27112114040398483,
      "grad_norm": 0.764930248260498,
      "learning_rate": 9.568737549346862e-06,
      "loss": 0.7761,
      "step": 4926
    },
    {
      "epoch": 0.27117617920634046,
      "grad_norm": 0.7226328253746033,
      "learning_rate": 9.56856142226773e-06,
      "loss": 0.7208,
      "step": 4927
    },
    {
      "epoch": 0.27123121800869615,
      "grad_norm": 0.8726598620414734,
      "learning_rate": 9.568385260852512e-06,
      "loss": 0.8599,
      "step": 4928
    },
    {
      "epoch": 0.2712862568110518,
      "grad_norm": 1.0126571655273438,
      "learning_rate": 9.568209065102533e-06,
      "loss": 0.8145,
      "step": 4929
    },
    {
      "epoch": 0.2713412956134075,
      "grad_norm": 0.7764692306518555,
      "learning_rate": 9.568032835019116e-06,
      "loss": 0.6758,
      "step": 4930
    },
    {
      "epoch": 0.2713963344157631,
      "grad_norm": 0.6955474019050598,
      "learning_rate": 9.567856570603589e-06,
      "loss": 0.7461,
      "step": 4931
    },
    {
      "epoch": 0.2714513732181188,
      "grad_norm": 0.7136832475662231,
      "learning_rate": 9.567680271857274e-06,
      "loss": 0.7692,
      "step": 4932
    },
    {
      "epoch": 0.27150641202047443,
      "grad_norm": 1.2288198471069336,
      "learning_rate": 9.567503938781497e-06,
      "loss": 0.7815,
      "step": 4933
    },
    {
      "epoch": 0.2715614508228301,
      "grad_norm": 0.9182234406471252,
      "learning_rate": 9.567327571377584e-06,
      "loss": 0.8822,
      "step": 4934
    },
    {
      "epoch": 0.27161648962518575,
      "grad_norm": 0.7684763669967651,
      "learning_rate": 9.567151169646859e-06,
      "loss": 0.7618,
      "step": 4935
    },
    {
      "epoch": 0.27167152842754144,
      "grad_norm": 0.872360348701477,
      "learning_rate": 9.566974733590647e-06,
      "loss": 0.7975,
      "step": 4936
    },
    {
      "epoch": 0.2717265672298971,
      "grad_norm": 0.9010463356971741,
      "learning_rate": 9.566798263210277e-06,
      "loss": 0.7159,
      "step": 4937
    },
    {
      "epoch": 0.27178160603225276,
      "grad_norm": 0.7254281044006348,
      "learning_rate": 9.566621758507072e-06,
      "loss": 0.6724,
      "step": 4938
    },
    {
      "epoch": 0.2718366448346084,
      "grad_norm": 0.8478212356567383,
      "learning_rate": 9.566445219482363e-06,
      "loss": 0.659,
      "step": 4939
    },
    {
      "epoch": 0.27189168363696403,
      "grad_norm": 0.9038714170455933,
      "learning_rate": 9.56626864613747e-06,
      "loss": 0.8766,
      "step": 4940
    },
    {
      "epoch": 0.2719467224393197,
      "grad_norm": 0.9704582691192627,
      "learning_rate": 9.566092038473728e-06,
      "loss": 0.8972,
      "step": 4941
    },
    {
      "epoch": 0.27200176124167535,
      "grad_norm": 0.7069430947303772,
      "learning_rate": 9.565915396492459e-06,
      "loss": 0.8116,
      "step": 4942
    },
    {
      "epoch": 0.27205680004403104,
      "grad_norm": 0.7432642579078674,
      "learning_rate": 9.565738720194993e-06,
      "loss": 0.847,
      "step": 4943
    },
    {
      "epoch": 0.2721118388463867,
      "grad_norm": 0.6813814043998718,
      "learning_rate": 9.565562009582655e-06,
      "loss": 0.7146,
      "step": 4944
    },
    {
      "epoch": 0.27216687764874237,
      "grad_norm": 0.7447707056999207,
      "learning_rate": 9.565385264656776e-06,
      "loss": 0.7696,
      "step": 4945
    },
    {
      "epoch": 0.272221916451098,
      "grad_norm": 0.875073254108429,
      "learning_rate": 9.565208485418685e-06,
      "loss": 0.8714,
      "step": 4946
    },
    {
      "epoch": 0.2722769552534537,
      "grad_norm": 0.7753880620002747,
      "learning_rate": 9.565031671869707e-06,
      "loss": 0.739,
      "step": 4947
    },
    {
      "epoch": 0.2723319940558093,
      "grad_norm": 0.749264121055603,
      "learning_rate": 9.564854824011172e-06,
      "loss": 0.7957,
      "step": 4948
    },
    {
      "epoch": 0.272387032858165,
      "grad_norm": 0.6733991503715515,
      "learning_rate": 9.564677941844412e-06,
      "loss": 0.7402,
      "step": 4949
    },
    {
      "epoch": 0.27244207166052065,
      "grad_norm": 0.7426447868347168,
      "learning_rate": 9.564501025370753e-06,
      "loss": 0.7977,
      "step": 4950
    },
    {
      "epoch": 0.27249711046287634,
      "grad_norm": 0.7930514812469482,
      "learning_rate": 9.564324074591529e-06,
      "loss": 0.8485,
      "step": 4951
    },
    {
      "epoch": 0.27255214926523197,
      "grad_norm": 0.8087072968482971,
      "learning_rate": 9.564147089508064e-06,
      "loss": 0.9215,
      "step": 4952
    },
    {
      "epoch": 0.27260718806758766,
      "grad_norm": 0.7560327053070068,
      "learning_rate": 9.563970070121694e-06,
      "loss": 0.7966,
      "step": 4953
    },
    {
      "epoch": 0.2726622268699433,
      "grad_norm": 0.735573947429657,
      "learning_rate": 9.563793016433744e-06,
      "loss": 0.7737,
      "step": 4954
    },
    {
      "epoch": 0.272717265672299,
      "grad_norm": 0.7603545784950256,
      "learning_rate": 9.563615928445548e-06,
      "loss": 0.7717,
      "step": 4955
    },
    {
      "epoch": 0.2727723044746546,
      "grad_norm": 0.7185375094413757,
      "learning_rate": 9.563438806158437e-06,
      "loss": 0.8057,
      "step": 4956
    },
    {
      "epoch": 0.2728273432770103,
      "grad_norm": 0.7619272470474243,
      "learning_rate": 9.56326164957374e-06,
      "loss": 0.8173,
      "step": 4957
    },
    {
      "epoch": 0.27288238207936594,
      "grad_norm": 0.7868000864982605,
      "learning_rate": 9.563084458692793e-06,
      "loss": 0.6855,
      "step": 4958
    },
    {
      "epoch": 0.2729374208817216,
      "grad_norm": 0.7949535846710205,
      "learning_rate": 9.562907233516923e-06,
      "loss": 0.7754,
      "step": 4959
    },
    {
      "epoch": 0.27299245968407726,
      "grad_norm": 0.7037919163703918,
      "learning_rate": 9.562729974047462e-06,
      "loss": 0.7419,
      "step": 4960
    },
    {
      "epoch": 0.27304749848643295,
      "grad_norm": 0.7236568927764893,
      "learning_rate": 9.562552680285746e-06,
      "loss": 0.7135,
      "step": 4961
    },
    {
      "epoch": 0.2731025372887886,
      "grad_norm": 0.8410467505455017,
      "learning_rate": 9.562375352233105e-06,
      "loss": 0.8507,
      "step": 4962
    },
    {
      "epoch": 0.27315757609114427,
      "grad_norm": 0.8043560981750488,
      "learning_rate": 9.562197989890871e-06,
      "loss": 0.8484,
      "step": 4963
    },
    {
      "epoch": 0.2732126148934999,
      "grad_norm": 0.6926127672195435,
      "learning_rate": 9.56202059326038e-06,
      "loss": 0.8087,
      "step": 4964
    },
    {
      "epoch": 0.2732676536958556,
      "grad_norm": 0.7149024605751038,
      "learning_rate": 9.561843162342961e-06,
      "loss": 0.7349,
      "step": 4965
    },
    {
      "epoch": 0.27332269249821123,
      "grad_norm": 0.7165781855583191,
      "learning_rate": 9.561665697139952e-06,
      "loss": 0.8139,
      "step": 4966
    },
    {
      "epoch": 0.2733777313005669,
      "grad_norm": 0.7481133341789246,
      "learning_rate": 9.561488197652684e-06,
      "loss": 0.7712,
      "step": 4967
    },
    {
      "epoch": 0.27343277010292255,
      "grad_norm": 0.6928209066390991,
      "learning_rate": 9.561310663882491e-06,
      "loss": 0.7524,
      "step": 4968
    },
    {
      "epoch": 0.27348780890527824,
      "grad_norm": 0.7397856116294861,
      "learning_rate": 9.561133095830708e-06,
      "loss": 0.718,
      "step": 4969
    },
    {
      "epoch": 0.2735428477076339,
      "grad_norm": 0.7712383270263672,
      "learning_rate": 9.560955493498672e-06,
      "loss": 0.8201,
      "step": 4970
    },
    {
      "epoch": 0.27359788650998956,
      "grad_norm": 0.96076899766922,
      "learning_rate": 9.560777856887714e-06,
      "loss": 0.8555,
      "step": 4971
    },
    {
      "epoch": 0.2736529253123452,
      "grad_norm": 0.7331019639968872,
      "learning_rate": 9.56060018599917e-06,
      "loss": 0.8315,
      "step": 4972
    },
    {
      "epoch": 0.2737079641147009,
      "grad_norm": 0.7157140970230103,
      "learning_rate": 9.560422480834374e-06,
      "loss": 0.7177,
      "step": 4973
    },
    {
      "epoch": 0.2737630029170565,
      "grad_norm": 0.807614266872406,
      "learning_rate": 9.560244741394666e-06,
      "loss": 0.8413,
      "step": 4974
    },
    {
      "epoch": 0.2738180417194122,
      "grad_norm": 0.7618574500083923,
      "learning_rate": 9.560066967681378e-06,
      "loss": 0.8248,
      "step": 4975
    },
    {
      "epoch": 0.27387308052176784,
      "grad_norm": 0.7886885404586792,
      "learning_rate": 9.559889159695848e-06,
      "loss": 0.8793,
      "step": 4976
    },
    {
      "epoch": 0.27392811932412353,
      "grad_norm": 1.0090755224227905,
      "learning_rate": 9.559711317439411e-06,
      "loss": 0.9255,
      "step": 4977
    },
    {
      "epoch": 0.27398315812647916,
      "grad_norm": 0.7855443358421326,
      "learning_rate": 9.559533440913405e-06,
      "loss": 0.8001,
      "step": 4978
    },
    {
      "epoch": 0.27403819692883485,
      "grad_norm": 0.768741250038147,
      "learning_rate": 9.559355530119165e-06,
      "loss": 0.8109,
      "step": 4979
    },
    {
      "epoch": 0.2740932357311905,
      "grad_norm": 0.759589672088623,
      "learning_rate": 9.55917758505803e-06,
      "loss": 0.8001,
      "step": 4980
    },
    {
      "epoch": 0.2741482745335462,
      "grad_norm": 0.7937445640563965,
      "learning_rate": 9.558999605731338e-06,
      "loss": 0.8924,
      "step": 4981
    },
    {
      "epoch": 0.2742033133359018,
      "grad_norm": 0.9041592478752136,
      "learning_rate": 9.558821592140423e-06,
      "loss": 0.9167,
      "step": 4982
    },
    {
      "epoch": 0.27425835213825744,
      "grad_norm": 0.6971380710601807,
      "learning_rate": 9.558643544286627e-06,
      "loss": 0.7589,
      "step": 4983
    },
    {
      "epoch": 0.27431339094061313,
      "grad_norm": 0.9292929172515869,
      "learning_rate": 9.558465462171287e-06,
      "loss": 0.9566,
      "step": 4984
    },
    {
      "epoch": 0.27436842974296877,
      "grad_norm": 0.8320629000663757,
      "learning_rate": 9.558287345795738e-06,
      "loss": 0.8854,
      "step": 4985
    },
    {
      "epoch": 0.27442346854532446,
      "grad_norm": 0.797272801399231,
      "learning_rate": 9.558109195161325e-06,
      "loss": 0.7838,
      "step": 4986
    },
    {
      "epoch": 0.2744785073476801,
      "grad_norm": 0.9702700972557068,
      "learning_rate": 9.557931010269382e-06,
      "loss": 0.8593,
      "step": 4987
    },
    {
      "epoch": 0.2745335461500358,
      "grad_norm": 0.8309103846549988,
      "learning_rate": 9.557752791121248e-06,
      "loss": 0.8902,
      "step": 4988
    },
    {
      "epoch": 0.2745885849523914,
      "grad_norm": 0.706667959690094,
      "learning_rate": 9.557574537718265e-06,
      "loss": 0.7259,
      "step": 4989
    },
    {
      "epoch": 0.2746436237547471,
      "grad_norm": 0.770239531993866,
      "learning_rate": 9.557396250061771e-06,
      "loss": 0.8644,
      "step": 4990
    },
    {
      "epoch": 0.27469866255710274,
      "grad_norm": 0.8695803880691528,
      "learning_rate": 9.557217928153108e-06,
      "loss": 0.895,
      "step": 4991
    },
    {
      "epoch": 0.2747537013594584,
      "grad_norm": 0.7525948286056519,
      "learning_rate": 9.557039571993614e-06,
      "loss": 0.7029,
      "step": 4992
    },
    {
      "epoch": 0.27480874016181406,
      "grad_norm": 0.7616680264472961,
      "learning_rate": 9.556861181584631e-06,
      "loss": 0.8025,
      "step": 4993
    },
    {
      "epoch": 0.27486377896416975,
      "grad_norm": 0.7216167449951172,
      "learning_rate": 9.5566827569275e-06,
      "loss": 0.8314,
      "step": 4994
    },
    {
      "epoch": 0.2749188177665254,
      "grad_norm": 0.7412614226341248,
      "learning_rate": 9.55650429802356e-06,
      "loss": 0.7877,
      "step": 4995
    },
    {
      "epoch": 0.27497385656888107,
      "grad_norm": 0.7176525592803955,
      "learning_rate": 9.556325804874154e-06,
      "loss": 0.7615,
      "step": 4996
    },
    {
      "epoch": 0.2750288953712367,
      "grad_norm": 0.7544515132904053,
      "learning_rate": 9.556147277480623e-06,
      "loss": 0.8352,
      "step": 4997
    },
    {
      "epoch": 0.2750839341735924,
      "grad_norm": 0.7318205833435059,
      "learning_rate": 9.555968715844309e-06,
      "loss": 0.7403,
      "step": 4998
    },
    {
      "epoch": 0.275138972975948,
      "grad_norm": 0.7495027780532837,
      "learning_rate": 9.555790119966552e-06,
      "loss": 0.7611,
      "step": 4999
    },
    {
      "epoch": 0.2751940117783037,
      "grad_norm": 0.7544401288032532,
      "learning_rate": 9.555611489848697e-06,
      "loss": 0.8594,
      "step": 5000
    },
    {
      "epoch": 0.27524905058065935,
      "grad_norm": 0.7698250412940979,
      "learning_rate": 9.555432825492084e-06,
      "loss": 0.8438,
      "step": 5001
    },
    {
      "epoch": 0.27530408938301504,
      "grad_norm": 0.7668892741203308,
      "learning_rate": 9.555254126898059e-06,
      "loss": 0.8082,
      "step": 5002
    },
    {
      "epoch": 0.27535912818537067,
      "grad_norm": 0.9170669317245483,
      "learning_rate": 9.555075394067963e-06,
      "loss": 0.7443,
      "step": 5003
    },
    {
      "epoch": 0.27541416698772636,
      "grad_norm": 0.7890255451202393,
      "learning_rate": 9.55489662700314e-06,
      "loss": 0.8269,
      "step": 5004
    },
    {
      "epoch": 0.275469205790082,
      "grad_norm": 0.6740512847900391,
      "learning_rate": 9.554717825704932e-06,
      "loss": 0.6906,
      "step": 5005
    },
    {
      "epoch": 0.2755242445924377,
      "grad_norm": 0.8032376170158386,
      "learning_rate": 9.554538990174685e-06,
      "loss": 0.812,
      "step": 5006
    },
    {
      "epoch": 0.2755792833947933,
      "grad_norm": 0.6932135224342346,
      "learning_rate": 9.554360120413741e-06,
      "loss": 0.7823,
      "step": 5007
    },
    {
      "epoch": 0.275634322197149,
      "grad_norm": 0.7447643876075745,
      "learning_rate": 9.554181216423447e-06,
      "loss": 0.8753,
      "step": 5008
    },
    {
      "epoch": 0.27568936099950464,
      "grad_norm": 0.8035081624984741,
      "learning_rate": 9.554002278205145e-06,
      "loss": 0.7135,
      "step": 5009
    },
    {
      "epoch": 0.27574439980186033,
      "grad_norm": 0.7544171214103699,
      "learning_rate": 9.553823305760182e-06,
      "loss": 0.7574,
      "step": 5010
    },
    {
      "epoch": 0.27579943860421596,
      "grad_norm": 0.6648419499397278,
      "learning_rate": 9.553644299089902e-06,
      "loss": 0.7566,
      "step": 5011
    },
    {
      "epoch": 0.27585447740657165,
      "grad_norm": 0.7481752038002014,
      "learning_rate": 9.55346525819565e-06,
      "loss": 0.7862,
      "step": 5012
    },
    {
      "epoch": 0.2759095162089273,
      "grad_norm": 0.7000668048858643,
      "learning_rate": 9.55328618307877e-06,
      "loss": 0.7767,
      "step": 5013
    },
    {
      "epoch": 0.275964555011283,
      "grad_norm": 0.7435166239738464,
      "learning_rate": 9.553107073740612e-06,
      "loss": 0.6888,
      "step": 5014
    },
    {
      "epoch": 0.2760195938136386,
      "grad_norm": 0.7593170404434204,
      "learning_rate": 9.552927930182521e-06,
      "loss": 0.7272,
      "step": 5015
    },
    {
      "epoch": 0.2760746326159943,
      "grad_norm": 0.870079755783081,
      "learning_rate": 9.55274875240584e-06,
      "loss": 0.8692,
      "step": 5016
    },
    {
      "epoch": 0.27612967141834993,
      "grad_norm": 0.8550307750701904,
      "learning_rate": 9.55256954041192e-06,
      "loss": 0.8729,
      "step": 5017
    },
    {
      "epoch": 0.2761847102207056,
      "grad_norm": 0.888830304145813,
      "learning_rate": 9.552390294202105e-06,
      "loss": 0.8607,
      "step": 5018
    },
    {
      "epoch": 0.27623974902306125,
      "grad_norm": 0.8295729160308838,
      "learning_rate": 9.552211013777743e-06,
      "loss": 0.8722,
      "step": 5019
    },
    {
      "epoch": 0.27629478782541694,
      "grad_norm": 0.7732356190681458,
      "learning_rate": 9.552031699140182e-06,
      "loss": 0.8332,
      "step": 5020
    },
    {
      "epoch": 0.2763498266277726,
      "grad_norm": 0.9132987856864929,
      "learning_rate": 9.55185235029077e-06,
      "loss": 0.769,
      "step": 5021
    },
    {
      "epoch": 0.27640486543012827,
      "grad_norm": 0.7221076488494873,
      "learning_rate": 9.551672967230851e-06,
      "loss": 0.8505,
      "step": 5022
    },
    {
      "epoch": 0.2764599042324839,
      "grad_norm": 0.8526949882507324,
      "learning_rate": 9.551493549961778e-06,
      "loss": 0.8002,
      "step": 5023
    },
    {
      "epoch": 0.2765149430348396,
      "grad_norm": 0.9513188004493713,
      "learning_rate": 9.551314098484901e-06,
      "loss": 0.8558,
      "step": 5024
    },
    {
      "epoch": 0.2765699818371952,
      "grad_norm": 0.7543003559112549,
      "learning_rate": 9.551134612801563e-06,
      "loss": 0.8292,
      "step": 5025
    },
    {
      "epoch": 0.27662502063955086,
      "grad_norm": 0.7531017065048218,
      "learning_rate": 9.550955092913115e-06,
      "loss": 0.7837,
      "step": 5026
    },
    {
      "epoch": 0.27668005944190655,
      "grad_norm": 0.8725717663764954,
      "learning_rate": 9.550775538820907e-06,
      "loss": 0.8362,
      "step": 5027
    },
    {
      "epoch": 0.2767350982442622,
      "grad_norm": 0.8122721910476685,
      "learning_rate": 9.550595950526288e-06,
      "loss": 0.8539,
      "step": 5028
    },
    {
      "epoch": 0.27679013704661787,
      "grad_norm": 0.7756829261779785,
      "learning_rate": 9.550416328030608e-06,
      "loss": 0.787,
      "step": 5029
    },
    {
      "epoch": 0.2768451758489735,
      "grad_norm": 0.9086001515388489,
      "learning_rate": 9.550236671335218e-06,
      "loss": 0.7972,
      "step": 5030
    },
    {
      "epoch": 0.2769002146513292,
      "grad_norm": 0.7857060432434082,
      "learning_rate": 9.550056980441466e-06,
      "loss": 0.7577,
      "step": 5031
    },
    {
      "epoch": 0.2769552534536848,
      "grad_norm": 0.8190392851829529,
      "learning_rate": 9.549877255350703e-06,
      "loss": 0.81,
      "step": 5032
    },
    {
      "epoch": 0.2770102922560405,
      "grad_norm": 0.7714588642120361,
      "learning_rate": 9.549697496064283e-06,
      "loss": 0.7916,
      "step": 5033
    },
    {
      "epoch": 0.27706533105839615,
      "grad_norm": 0.7178533673286438,
      "learning_rate": 9.549517702583552e-06,
      "loss": 0.8001,
      "step": 5034
    },
    {
      "epoch": 0.27712036986075184,
      "grad_norm": 0.7552955150604248,
      "learning_rate": 9.549337874909865e-06,
      "loss": 0.8361,
      "step": 5035
    },
    {
      "epoch": 0.27717540866310747,
      "grad_norm": 0.7823992371559143,
      "learning_rate": 9.549158013044573e-06,
      "loss": 0.7033,
      "step": 5036
    },
    {
      "epoch": 0.27723044746546316,
      "grad_norm": 0.731504499912262,
      "learning_rate": 9.548978116989026e-06,
      "loss": 0.73,
      "step": 5037
    },
    {
      "epoch": 0.2772854862678188,
      "grad_norm": 0.7455994486808777,
      "learning_rate": 9.548798186744578e-06,
      "loss": 0.8005,
      "step": 5038
    },
    {
      "epoch": 0.2773405250701745,
      "grad_norm": 0.7020164728164673,
      "learning_rate": 9.54861822231258e-06,
      "loss": 0.6707,
      "step": 5039
    },
    {
      "epoch": 0.2773955638725301,
      "grad_norm": 0.7526360750198364,
      "learning_rate": 9.548438223694385e-06,
      "loss": 0.7686,
      "step": 5040
    },
    {
      "epoch": 0.2774506026748858,
      "grad_norm": 0.7268579006195068,
      "learning_rate": 9.548258190891344e-06,
      "loss": 0.7039,
      "step": 5041
    },
    {
      "epoch": 0.27750564147724144,
      "grad_norm": 0.9361631274223328,
      "learning_rate": 9.548078123904815e-06,
      "loss": 0.8023,
      "step": 5042
    },
    {
      "epoch": 0.2775606802795971,
      "grad_norm": 0.7786710262298584,
      "learning_rate": 9.547898022736147e-06,
      "loss": 0.6866,
      "step": 5043
    },
    {
      "epoch": 0.27761571908195276,
      "grad_norm": 0.7175624370574951,
      "learning_rate": 9.547717887386695e-06,
      "loss": 0.7554,
      "step": 5044
    },
    {
      "epoch": 0.27767075788430845,
      "grad_norm": 0.9157657623291016,
      "learning_rate": 9.547537717857813e-06,
      "loss": 0.7936,
      "step": 5045
    },
    {
      "epoch": 0.2777257966866641,
      "grad_norm": 0.7881377935409546,
      "learning_rate": 9.547357514150854e-06,
      "loss": 0.8198,
      "step": 5046
    },
    {
      "epoch": 0.2777808354890198,
      "grad_norm": 1.0444039106369019,
      "learning_rate": 9.547177276267173e-06,
      "loss": 0.7954,
      "step": 5047
    },
    {
      "epoch": 0.2778358742913754,
      "grad_norm": 0.7889506220817566,
      "learning_rate": 9.546997004208124e-06,
      "loss": 0.7697,
      "step": 5048
    },
    {
      "epoch": 0.2778909130937311,
      "grad_norm": 0.7304134368896484,
      "learning_rate": 9.546816697975066e-06,
      "loss": 0.7034,
      "step": 5049
    },
    {
      "epoch": 0.27794595189608673,
      "grad_norm": 0.7783082723617554,
      "learning_rate": 9.546636357569347e-06,
      "loss": 0.8185,
      "step": 5050
    },
    {
      "epoch": 0.2780009906984424,
      "grad_norm": 0.750712513923645,
      "learning_rate": 9.54645598299233e-06,
      "loss": 0.7336,
      "step": 5051
    },
    {
      "epoch": 0.27805602950079805,
      "grad_norm": 0.7849590182304382,
      "learning_rate": 9.546275574245364e-06,
      "loss": 0.8088,
      "step": 5052
    },
    {
      "epoch": 0.27811106830315374,
      "grad_norm": 0.8490208983421326,
      "learning_rate": 9.546095131329809e-06,
      "loss": 0.8507,
      "step": 5053
    },
    {
      "epoch": 0.2781661071055094,
      "grad_norm": 0.8107250928878784,
      "learning_rate": 9.54591465424702e-06,
      "loss": 0.7787,
      "step": 5054
    },
    {
      "epoch": 0.27822114590786506,
      "grad_norm": 0.8278594613075256,
      "learning_rate": 9.54573414299835e-06,
      "loss": 0.7836,
      "step": 5055
    },
    {
      "epoch": 0.2782761847102207,
      "grad_norm": 0.7982015013694763,
      "learning_rate": 9.545553597585163e-06,
      "loss": 0.7672,
      "step": 5056
    },
    {
      "epoch": 0.2783312235125764,
      "grad_norm": 0.7311522364616394,
      "learning_rate": 9.54537301800881e-06,
      "loss": 0.7571,
      "step": 5057
    },
    {
      "epoch": 0.278386262314932,
      "grad_norm": 0.8039999604225159,
      "learning_rate": 9.545192404270651e-06,
      "loss": 0.764,
      "step": 5058
    },
    {
      "epoch": 0.2784413011172877,
      "grad_norm": 0.7810946702957153,
      "learning_rate": 9.545011756372042e-06,
      "loss": 0.9217,
      "step": 5059
    },
    {
      "epoch": 0.27849633991964334,
      "grad_norm": 0.7092248797416687,
      "learning_rate": 9.544831074314343e-06,
      "loss": 0.7599,
      "step": 5060
    },
    {
      "epoch": 0.27855137872199903,
      "grad_norm": 0.831550657749176,
      "learning_rate": 9.544650358098908e-06,
      "loss": 0.7278,
      "step": 5061
    },
    {
      "epoch": 0.27860641752435467,
      "grad_norm": 0.7645474076271057,
      "learning_rate": 9.544469607727098e-06,
      "loss": 0.7945,
      "step": 5062
    },
    {
      "epoch": 0.27866145632671036,
      "grad_norm": 0.6956788301467896,
      "learning_rate": 9.544288823200273e-06,
      "loss": 0.749,
      "step": 5063
    },
    {
      "epoch": 0.278716495129066,
      "grad_norm": 0.7262974381446838,
      "learning_rate": 9.544108004519786e-06,
      "loss": 0.8074,
      "step": 5064
    },
    {
      "epoch": 0.2787715339314217,
      "grad_norm": 0.7439202666282654,
      "learning_rate": 9.543927151687001e-06,
      "loss": 0.9403,
      "step": 5065
    },
    {
      "epoch": 0.2788265727337773,
      "grad_norm": 0.8468778133392334,
      "learning_rate": 9.543746264703277e-06,
      "loss": 0.8182,
      "step": 5066
    },
    {
      "epoch": 0.278881611536133,
      "grad_norm": 0.8396204113960266,
      "learning_rate": 9.54356534356997e-06,
      "loss": 0.8067,
      "step": 5067
    },
    {
      "epoch": 0.27893665033848863,
      "grad_norm": 0.718758225440979,
      "learning_rate": 9.543384388288445e-06,
      "loss": 0.8172,
      "step": 5068
    },
    {
      "epoch": 0.27899168914084427,
      "grad_norm": 0.7562685012817383,
      "learning_rate": 9.543203398860056e-06,
      "loss": 0.9053,
      "step": 5069
    },
    {
      "epoch": 0.27904672794319996,
      "grad_norm": 0.9592792987823486,
      "learning_rate": 9.543022375286169e-06,
      "loss": 0.9375,
      "step": 5070
    },
    {
      "epoch": 0.2791017667455556,
      "grad_norm": 0.7162739634513855,
      "learning_rate": 9.54284131756814e-06,
      "loss": 0.7297,
      "step": 5071
    },
    {
      "epoch": 0.2791568055479113,
      "grad_norm": 0.7703517079353333,
      "learning_rate": 9.542660225707335e-06,
      "loss": 0.8863,
      "step": 5072
    },
    {
      "epoch": 0.2792118443502669,
      "grad_norm": 0.7860418558120728,
      "learning_rate": 9.542479099705109e-06,
      "loss": 0.8335,
      "step": 5073
    },
    {
      "epoch": 0.2792668831526226,
      "grad_norm": 0.8880825042724609,
      "learning_rate": 9.542297939562825e-06,
      "loss": 0.8344,
      "step": 5074
    },
    {
      "epoch": 0.27932192195497824,
      "grad_norm": 0.7900505661964417,
      "learning_rate": 9.542116745281849e-06,
      "loss": 0.7613,
      "step": 5075
    },
    {
      "epoch": 0.2793769607573339,
      "grad_norm": 0.7446081042289734,
      "learning_rate": 9.541935516863536e-06,
      "loss": 0.6615,
      "step": 5076
    },
    {
      "epoch": 0.27943199955968956,
      "grad_norm": 0.7831308245658875,
      "learning_rate": 9.541754254309254e-06,
      "loss": 0.779,
      "step": 5077
    },
    {
      "epoch": 0.27948703836204525,
      "grad_norm": 0.9007606506347656,
      "learning_rate": 9.541572957620361e-06,
      "loss": 0.8883,
      "step": 5078
    },
    {
      "epoch": 0.2795420771644009,
      "grad_norm": 0.8033407330513,
      "learning_rate": 9.541391626798222e-06,
      "loss": 0.7354,
      "step": 5079
    },
    {
      "epoch": 0.27959711596675657,
      "grad_norm": 0.9259470105171204,
      "learning_rate": 9.5412102618442e-06,
      "loss": 0.7602,
      "step": 5080
    },
    {
      "epoch": 0.2796521547691122,
      "grad_norm": 0.786523163318634,
      "learning_rate": 9.541028862759656e-06,
      "loss": 0.7402,
      "step": 5081
    },
    {
      "epoch": 0.2797071935714679,
      "grad_norm": 0.8053372502326965,
      "learning_rate": 9.540847429545954e-06,
      "loss": 0.825,
      "step": 5082
    },
    {
      "epoch": 0.2797622323738235,
      "grad_norm": 0.8578022122383118,
      "learning_rate": 9.54066596220446e-06,
      "loss": 0.7866,
      "step": 5083
    },
    {
      "epoch": 0.2798172711761792,
      "grad_norm": 0.916161835193634,
      "learning_rate": 9.540484460736535e-06,
      "loss": 0.5961,
      "step": 5084
    },
    {
      "epoch": 0.27987230997853485,
      "grad_norm": 0.7843562960624695,
      "learning_rate": 9.540302925143545e-06,
      "loss": 0.764,
      "step": 5085
    },
    {
      "epoch": 0.27992734878089054,
      "grad_norm": 0.7392510771751404,
      "learning_rate": 9.540121355426852e-06,
      "loss": 0.8038,
      "step": 5086
    },
    {
      "epoch": 0.2799823875832462,
      "grad_norm": 0.7406296133995056,
      "learning_rate": 9.539939751587825e-06,
      "loss": 0.8202,
      "step": 5087
    },
    {
      "epoch": 0.28003742638560186,
      "grad_norm": 0.7274924516677856,
      "learning_rate": 9.539758113627823e-06,
      "loss": 0.7691,
      "step": 5088
    },
    {
      "epoch": 0.2800924651879575,
      "grad_norm": 0.8563184142112732,
      "learning_rate": 9.539576441548218e-06,
      "loss": 0.8341,
      "step": 5089
    },
    {
      "epoch": 0.2801475039903132,
      "grad_norm": 0.7708351016044617,
      "learning_rate": 9.539394735350366e-06,
      "loss": 0.7126,
      "step": 5090
    },
    {
      "epoch": 0.2802025427926688,
      "grad_norm": 0.7314836382865906,
      "learning_rate": 9.539212995035642e-06,
      "loss": 0.7465,
      "step": 5091
    },
    {
      "epoch": 0.2802575815950245,
      "grad_norm": 0.7594754695892334,
      "learning_rate": 9.539031220605409e-06,
      "loss": 0.7563,
      "step": 5092
    },
    {
      "epoch": 0.28031262039738014,
      "grad_norm": 0.699414074420929,
      "learning_rate": 9.53884941206103e-06,
      "loss": 0.7847,
      "step": 5093
    },
    {
      "epoch": 0.28036765919973583,
      "grad_norm": 0.8013063073158264,
      "learning_rate": 9.538667569403877e-06,
      "loss": 0.7769,
      "step": 5094
    },
    {
      "epoch": 0.28042269800209146,
      "grad_norm": 0.7778805494308472,
      "learning_rate": 9.538485692635312e-06,
      "loss": 0.7646,
      "step": 5095
    },
    {
      "epoch": 0.28047773680444715,
      "grad_norm": 0.785649299621582,
      "learning_rate": 9.538303781756702e-06,
      "loss": 0.8162,
      "step": 5096
    },
    {
      "epoch": 0.2805327756068028,
      "grad_norm": 0.7073212265968323,
      "learning_rate": 9.538121836769417e-06,
      "loss": 0.7208,
      "step": 5097
    },
    {
      "epoch": 0.2805878144091585,
      "grad_norm": 0.7545642852783203,
      "learning_rate": 9.53793985767482e-06,
      "loss": 0.8673,
      "step": 5098
    },
    {
      "epoch": 0.2806428532115141,
      "grad_norm": 0.6818416118621826,
      "learning_rate": 9.537757844474285e-06,
      "loss": 0.7576,
      "step": 5099
    },
    {
      "epoch": 0.2806978920138698,
      "grad_norm": 0.6718038320541382,
      "learning_rate": 9.537575797169176e-06,
      "loss": 0.6683,
      "step": 5100
    },
    {
      "epoch": 0.28075293081622543,
      "grad_norm": 0.7851004600524902,
      "learning_rate": 9.53739371576086e-06,
      "loss": 0.8871,
      "step": 5101
    },
    {
      "epoch": 0.2808079696185811,
      "grad_norm": 0.7565650343894958,
      "learning_rate": 9.53721160025071e-06,
      "loss": 0.8799,
      "step": 5102
    },
    {
      "epoch": 0.28086300842093676,
      "grad_norm": 0.7522932887077332,
      "learning_rate": 9.537029450640091e-06,
      "loss": 0.838,
      "step": 5103
    },
    {
      "epoch": 0.28091804722329244,
      "grad_norm": 0.929634690284729,
      "learning_rate": 9.536847266930375e-06,
      "loss": 0.7997,
      "step": 5104
    },
    {
      "epoch": 0.2809730860256481,
      "grad_norm": 0.8050084710121155,
      "learning_rate": 9.536665049122928e-06,
      "loss": 0.7652,
      "step": 5105
    },
    {
      "epoch": 0.28102812482800377,
      "grad_norm": 0.7401233315467834,
      "learning_rate": 9.53648279721912e-06,
      "loss": 0.7904,
      "step": 5106
    },
    {
      "epoch": 0.2810831636303594,
      "grad_norm": 0.7125453948974609,
      "learning_rate": 9.536300511220322e-06,
      "loss": 0.7349,
      "step": 5107
    },
    {
      "epoch": 0.2811382024327151,
      "grad_norm": 0.7165758609771729,
      "learning_rate": 9.536118191127905e-06,
      "loss": 0.7314,
      "step": 5108
    },
    {
      "epoch": 0.2811932412350707,
      "grad_norm": 0.7507439851760864,
      "learning_rate": 9.535935836943237e-06,
      "loss": 0.7603,
      "step": 5109
    },
    {
      "epoch": 0.2812482800374264,
      "grad_norm": 0.7832109332084656,
      "learning_rate": 9.535753448667688e-06,
      "loss": 0.7279,
      "step": 5110
    },
    {
      "epoch": 0.28130331883978205,
      "grad_norm": 0.7346609234809875,
      "learning_rate": 9.535571026302633e-06,
      "loss": 0.6882,
      "step": 5111
    },
    {
      "epoch": 0.2813583576421377,
      "grad_norm": 0.7569608688354492,
      "learning_rate": 9.535388569849437e-06,
      "loss": 0.8451,
      "step": 5112
    },
    {
      "epoch": 0.28141339644449337,
      "grad_norm": 0.7319865822792053,
      "learning_rate": 9.535206079309478e-06,
      "loss": 0.8161,
      "step": 5113
    },
    {
      "epoch": 0.281468435246849,
      "grad_norm": 0.7744631171226501,
      "learning_rate": 9.535023554684122e-06,
      "loss": 0.8025,
      "step": 5114
    },
    {
      "epoch": 0.2815234740492047,
      "grad_norm": 0.6867525577545166,
      "learning_rate": 9.534840995974743e-06,
      "loss": 0.7693,
      "step": 5115
    },
    {
      "epoch": 0.2815785128515603,
      "grad_norm": 0.7625848054885864,
      "learning_rate": 9.534658403182715e-06,
      "loss": 0.8034,
      "step": 5116
    },
    {
      "epoch": 0.281633551653916,
      "grad_norm": 0.7369832992553711,
      "learning_rate": 9.534475776309406e-06,
      "loss": 0.873,
      "step": 5117
    },
    {
      "epoch": 0.28168859045627165,
      "grad_norm": 0.7267127633094788,
      "learning_rate": 9.534293115356191e-06,
      "loss": 0.7954,
      "step": 5118
    },
    {
      "epoch": 0.28174362925862734,
      "grad_norm": 0.7244247794151306,
      "learning_rate": 9.534110420324443e-06,
      "loss": 0.7784,
      "step": 5119
    },
    {
      "epoch": 0.28179866806098297,
      "grad_norm": 0.8207812905311584,
      "learning_rate": 9.533927691215534e-06,
      "loss": 0.8696,
      "step": 5120
    },
    {
      "epoch": 0.28185370686333866,
      "grad_norm": 0.8669891357421875,
      "learning_rate": 9.53374492803084e-06,
      "loss": 0.8203,
      "step": 5121
    },
    {
      "epoch": 0.2819087456656943,
      "grad_norm": 0.7650816440582275,
      "learning_rate": 9.533562130771732e-06,
      "loss": 0.77,
      "step": 5122
    },
    {
      "epoch": 0.28196378446805,
      "grad_norm": 0.7664972543716431,
      "learning_rate": 9.533379299439584e-06,
      "loss": 0.7187,
      "step": 5123
    },
    {
      "epoch": 0.2820188232704056,
      "grad_norm": 0.7921896576881409,
      "learning_rate": 9.533196434035772e-06,
      "loss": 0.8669,
      "step": 5124
    },
    {
      "epoch": 0.2820738620727613,
      "grad_norm": 0.7714456915855408,
      "learning_rate": 9.533013534561669e-06,
      "loss": 0.8783,
      "step": 5125
    },
    {
      "epoch": 0.28212890087511694,
      "grad_norm": 0.7222065329551697,
      "learning_rate": 9.532830601018648e-06,
      "loss": 0.7449,
      "step": 5126
    },
    {
      "epoch": 0.28218393967747263,
      "grad_norm": 0.718142569065094,
      "learning_rate": 9.532647633408085e-06,
      "loss": 0.8226,
      "step": 5127
    },
    {
      "epoch": 0.28223897847982826,
      "grad_norm": 0.730592668056488,
      "learning_rate": 9.532464631731357e-06,
      "loss": 0.7878,
      "step": 5128
    },
    {
      "epoch": 0.28229401728218395,
      "grad_norm": 0.7841802835464478,
      "learning_rate": 9.532281595989839e-06,
      "loss": 0.8262,
      "step": 5129
    },
    {
      "epoch": 0.2823490560845396,
      "grad_norm": 0.8617212772369385,
      "learning_rate": 9.532098526184904e-06,
      "loss": 0.8368,
      "step": 5130
    },
    {
      "epoch": 0.2824040948868953,
      "grad_norm": 0.6968556642532349,
      "learning_rate": 9.53191542231793e-06,
      "loss": 0.6848,
      "step": 5131
    },
    {
      "epoch": 0.2824591336892509,
      "grad_norm": 0.7872157096862793,
      "learning_rate": 9.531732284390294e-06,
      "loss": 0.7898,
      "step": 5132
    },
    {
      "epoch": 0.2825141724916066,
      "grad_norm": 0.7727276086807251,
      "learning_rate": 9.53154911240337e-06,
      "loss": 0.8506,
      "step": 5133
    },
    {
      "epoch": 0.28256921129396223,
      "grad_norm": 0.7279896140098572,
      "learning_rate": 9.531365906358536e-06,
      "loss": 0.7415,
      "step": 5134
    },
    {
      "epoch": 0.2826242500963179,
      "grad_norm": 0.7457457780838013,
      "learning_rate": 9.53118266625717e-06,
      "loss": 0.7652,
      "step": 5135
    },
    {
      "epoch": 0.28267928889867355,
      "grad_norm": 0.8989270329475403,
      "learning_rate": 9.530999392100646e-06,
      "loss": 0.9085,
      "step": 5136
    },
    {
      "epoch": 0.28273432770102924,
      "grad_norm": 0.9622626304626465,
      "learning_rate": 9.530816083890347e-06,
      "loss": 0.8726,
      "step": 5137
    },
    {
      "epoch": 0.2827893665033849,
      "grad_norm": 0.7712846994400024,
      "learning_rate": 9.530632741627643e-06,
      "loss": 0.765,
      "step": 5138
    },
    {
      "epoch": 0.28284440530574056,
      "grad_norm": 0.8320727348327637,
      "learning_rate": 9.530449365313918e-06,
      "loss": 0.7828,
      "step": 5139
    },
    {
      "epoch": 0.2828994441080962,
      "grad_norm": 0.9310963153839111,
      "learning_rate": 9.530265954950549e-06,
      "loss": 0.8482,
      "step": 5140
    },
    {
      "epoch": 0.2829544829104519,
      "grad_norm": 0.9984502792358398,
      "learning_rate": 9.530082510538914e-06,
      "loss": 0.8673,
      "step": 5141
    },
    {
      "epoch": 0.2830095217128075,
      "grad_norm": 0.8300992250442505,
      "learning_rate": 9.52989903208039e-06,
      "loss": 0.8232,
      "step": 5142
    },
    {
      "epoch": 0.2830645605151632,
      "grad_norm": 0.930052638053894,
      "learning_rate": 9.529715519576356e-06,
      "loss": 0.7766,
      "step": 5143
    },
    {
      "epoch": 0.28311959931751884,
      "grad_norm": 0.8038359880447388,
      "learning_rate": 9.529531973028194e-06,
      "loss": 0.712,
      "step": 5144
    },
    {
      "epoch": 0.28317463811987453,
      "grad_norm": 0.856250524520874,
      "learning_rate": 9.529348392437283e-06,
      "loss": 0.8578,
      "step": 5145
    },
    {
      "epoch": 0.28322967692223017,
      "grad_norm": 0.7602483630180359,
      "learning_rate": 9.529164777805002e-06,
      "loss": 0.749,
      "step": 5146
    },
    {
      "epoch": 0.28328471572458586,
      "grad_norm": 0.8946549892425537,
      "learning_rate": 9.52898112913273e-06,
      "loss": 0.8101,
      "step": 5147
    },
    {
      "epoch": 0.2833397545269415,
      "grad_norm": 0.8015615344047546,
      "learning_rate": 9.52879744642185e-06,
      "loss": 0.8203,
      "step": 5148
    },
    {
      "epoch": 0.2833947933292972,
      "grad_norm": 0.7767183780670166,
      "learning_rate": 9.528613729673738e-06,
      "loss": 0.8409,
      "step": 5149
    },
    {
      "epoch": 0.2834498321316528,
      "grad_norm": 0.7604000568389893,
      "learning_rate": 9.52842997888978e-06,
      "loss": 0.8853,
      "step": 5150
    },
    {
      "epoch": 0.2835048709340085,
      "grad_norm": 0.7079401016235352,
      "learning_rate": 9.528246194071353e-06,
      "loss": 0.6855,
      "step": 5151
    },
    {
      "epoch": 0.28355990973636414,
      "grad_norm": 0.7616782188415527,
      "learning_rate": 9.52806237521984e-06,
      "loss": 0.785,
      "step": 5152
    },
    {
      "epoch": 0.2836149485387198,
      "grad_norm": 0.7408583760261536,
      "learning_rate": 9.527878522336622e-06,
      "loss": 0.7105,
      "step": 5153
    },
    {
      "epoch": 0.28366998734107546,
      "grad_norm": 0.694821834564209,
      "learning_rate": 9.52769463542308e-06,
      "loss": 0.6552,
      "step": 5154
    },
    {
      "epoch": 0.2837250261434311,
      "grad_norm": 0.796925961971283,
      "learning_rate": 9.5275107144806e-06,
      "loss": 0.7122,
      "step": 5155
    },
    {
      "epoch": 0.2837800649457868,
      "grad_norm": 0.8001971244812012,
      "learning_rate": 9.527326759510558e-06,
      "loss": 0.8528,
      "step": 5156
    },
    {
      "epoch": 0.2838351037481424,
      "grad_norm": 0.8605831265449524,
      "learning_rate": 9.527142770514341e-06,
      "loss": 0.7948,
      "step": 5157
    },
    {
      "epoch": 0.2838901425504981,
      "grad_norm": 0.8380078077316284,
      "learning_rate": 9.526958747493334e-06,
      "loss": 0.8184,
      "step": 5158
    },
    {
      "epoch": 0.28394518135285374,
      "grad_norm": 0.8758485317230225,
      "learning_rate": 9.526774690448913e-06,
      "loss": 0.7625,
      "step": 5159
    },
    {
      "epoch": 0.2840002201552094,
      "grad_norm": 0.7078989744186401,
      "learning_rate": 9.526590599382466e-06,
      "loss": 0.8179,
      "step": 5160
    },
    {
      "epoch": 0.28405525895756506,
      "grad_norm": 0.6668990850448608,
      "learning_rate": 9.526406474295376e-06,
      "loss": 0.7169,
      "step": 5161
    },
    {
      "epoch": 0.28411029775992075,
      "grad_norm": 0.7666084170341492,
      "learning_rate": 9.526222315189026e-06,
      "loss": 0.8511,
      "step": 5162
    },
    {
      "epoch": 0.2841653365622764,
      "grad_norm": 0.7390545606613159,
      "learning_rate": 9.526038122064802e-06,
      "loss": 0.7926,
      "step": 5163
    },
    {
      "epoch": 0.28422037536463207,
      "grad_norm": 0.7972092032432556,
      "learning_rate": 9.525853894924086e-06,
      "loss": 0.9166,
      "step": 5164
    },
    {
      "epoch": 0.2842754141669877,
      "grad_norm": 0.8988455533981323,
      "learning_rate": 9.525669633768265e-06,
      "loss": 0.9497,
      "step": 5165
    },
    {
      "epoch": 0.2843304529693434,
      "grad_norm": 0.7092710137367249,
      "learning_rate": 9.525485338598722e-06,
      "loss": 0.7241,
      "step": 5166
    },
    {
      "epoch": 0.28438549177169903,
      "grad_norm": 0.8630063533782959,
      "learning_rate": 9.525301009416843e-06,
      "loss": 0.8318,
      "step": 5167
    },
    {
      "epoch": 0.2844405305740547,
      "grad_norm": 0.7336890697479248,
      "learning_rate": 9.52511664622401e-06,
      "loss": 0.7077,
      "step": 5168
    },
    {
      "epoch": 0.28449556937641035,
      "grad_norm": 0.8156722784042358,
      "learning_rate": 9.524932249021615e-06,
      "loss": 0.8573,
      "step": 5169
    },
    {
      "epoch": 0.28455060817876604,
      "grad_norm": 0.7061388492584229,
      "learning_rate": 9.524747817811038e-06,
      "loss": 0.7432,
      "step": 5170
    },
    {
      "epoch": 0.2846056469811217,
      "grad_norm": 0.7948413491249084,
      "learning_rate": 9.52456335259367e-06,
      "loss": 0.8082,
      "step": 5171
    },
    {
      "epoch": 0.28466068578347736,
      "grad_norm": 0.7208091020584106,
      "learning_rate": 9.524378853370893e-06,
      "loss": 0.7027,
      "step": 5172
    },
    {
      "epoch": 0.284715724585833,
      "grad_norm": 0.8377540111541748,
      "learning_rate": 9.524194320144096e-06,
      "loss": 0.7093,
      "step": 5173
    },
    {
      "epoch": 0.2847707633881887,
      "grad_norm": 0.8734563589096069,
      "learning_rate": 9.524009752914666e-06,
      "loss": 0.8422,
      "step": 5174
    },
    {
      "epoch": 0.2848258021905443,
      "grad_norm": 0.7303940653800964,
      "learning_rate": 9.523825151683989e-06,
      "loss": 0.811,
      "step": 5175
    },
    {
      "epoch": 0.2848808409929,
      "grad_norm": 0.7653842568397522,
      "learning_rate": 9.523640516453455e-06,
      "loss": 0.8595,
      "step": 5176
    },
    {
      "epoch": 0.28493587979525564,
      "grad_norm": 0.7366930246353149,
      "learning_rate": 9.523455847224448e-06,
      "loss": 0.7832,
      "step": 5177
    },
    {
      "epoch": 0.28499091859761133,
      "grad_norm": 0.7908505797386169,
      "learning_rate": 9.523271143998357e-06,
      "loss": 0.8115,
      "step": 5178
    },
    {
      "epoch": 0.28504595739996696,
      "grad_norm": 0.8176048398017883,
      "learning_rate": 9.523086406776572e-06,
      "loss": 0.8377,
      "step": 5179
    },
    {
      "epoch": 0.28510099620232265,
      "grad_norm": 0.724086344242096,
      "learning_rate": 9.52290163556048e-06,
      "loss": 0.7804,
      "step": 5180
    },
    {
      "epoch": 0.2851560350046783,
      "grad_norm": 0.6461299657821655,
      "learning_rate": 9.52271683035147e-06,
      "loss": 0.5727,
      "step": 5181
    },
    {
      "epoch": 0.285211073807034,
      "grad_norm": 0.7275353074073792,
      "learning_rate": 9.522531991150932e-06,
      "loss": 0.8345,
      "step": 5182
    },
    {
      "epoch": 0.2852661126093896,
      "grad_norm": 0.7321951985359192,
      "learning_rate": 9.522347117960253e-06,
      "loss": 0.8832,
      "step": 5183
    },
    {
      "epoch": 0.2853211514117453,
      "grad_norm": 0.7526552677154541,
      "learning_rate": 9.522162210780825e-06,
      "loss": 0.831,
      "step": 5184
    },
    {
      "epoch": 0.28537619021410093,
      "grad_norm": 0.7592381238937378,
      "learning_rate": 9.521977269614036e-06,
      "loss": 0.7293,
      "step": 5185
    },
    {
      "epoch": 0.2854312290164566,
      "grad_norm": 0.8060448169708252,
      "learning_rate": 9.521792294461274e-06,
      "loss": 0.819,
      "step": 5186
    },
    {
      "epoch": 0.28548626781881226,
      "grad_norm": 0.7178553342819214,
      "learning_rate": 9.521607285323932e-06,
      "loss": 0.7526,
      "step": 5187
    },
    {
      "epoch": 0.28554130662116795,
      "grad_norm": 0.8186969757080078,
      "learning_rate": 9.521422242203401e-06,
      "loss": 0.8526,
      "step": 5188
    },
    {
      "epoch": 0.2855963454235236,
      "grad_norm": 0.8480883240699768,
      "learning_rate": 9.521237165101071e-06,
      "loss": 0.8088,
      "step": 5189
    },
    {
      "epoch": 0.28565138422587927,
      "grad_norm": 0.8053719401359558,
      "learning_rate": 9.521052054018333e-06,
      "loss": 0.928,
      "step": 5190
    },
    {
      "epoch": 0.2857064230282349,
      "grad_norm": 0.6937163472175598,
      "learning_rate": 9.52086690895658e-06,
      "loss": 0.7418,
      "step": 5191
    },
    {
      "epoch": 0.2857614618305906,
      "grad_norm": 1.0616179704666138,
      "learning_rate": 9.520681729917196e-06,
      "loss": 0.8726,
      "step": 5192
    },
    {
      "epoch": 0.2858165006329462,
      "grad_norm": 0.7504106163978577,
      "learning_rate": 9.520496516901582e-06,
      "loss": 0.844,
      "step": 5193
    },
    {
      "epoch": 0.2858715394353019,
      "grad_norm": 0.7634509205818176,
      "learning_rate": 9.520311269911127e-06,
      "loss": 0.7595,
      "step": 5194
    },
    {
      "epoch": 0.28592657823765755,
      "grad_norm": 0.7069799900054932,
      "learning_rate": 9.52012598894722e-06,
      "loss": 0.7566,
      "step": 5195
    },
    {
      "epoch": 0.28598161704001324,
      "grad_norm": 0.695737361907959,
      "learning_rate": 9.519940674011256e-06,
      "loss": 0.7534,
      "step": 5196
    },
    {
      "epoch": 0.28603665584236887,
      "grad_norm": 0.7212124466896057,
      "learning_rate": 9.51975532510463e-06,
      "loss": 0.8237,
      "step": 5197
    },
    {
      "epoch": 0.2860916946447245,
      "grad_norm": 0.7274062633514404,
      "learning_rate": 9.519569942228732e-06,
      "loss": 0.756,
      "step": 5198
    },
    {
      "epoch": 0.2861467334470802,
      "grad_norm": 0.7038697600364685,
      "learning_rate": 9.519384525384956e-06,
      "loss": 0.7308,
      "step": 5199
    },
    {
      "epoch": 0.2862017722494358,
      "grad_norm": 0.6897109150886536,
      "learning_rate": 9.519199074574694e-06,
      "loss": 0.7858,
      "step": 5200
    },
    {
      "epoch": 0.2862568110517915,
      "grad_norm": 0.8471527099609375,
      "learning_rate": 9.519013589799343e-06,
      "loss": 0.8198,
      "step": 5201
    },
    {
      "epoch": 0.28631184985414715,
      "grad_norm": 0.6828129291534424,
      "learning_rate": 9.518828071060295e-06,
      "loss": 0.7734,
      "step": 5202
    },
    {
      "epoch": 0.28636688865650284,
      "grad_norm": 0.7437755465507507,
      "learning_rate": 9.518642518358946e-06,
      "loss": 0.7669,
      "step": 5203
    },
    {
      "epoch": 0.28642192745885847,
      "grad_norm": 0.8841923475265503,
      "learning_rate": 9.518456931696689e-06,
      "loss": 0.8201,
      "step": 5204
    },
    {
      "epoch": 0.28647696626121416,
      "grad_norm": 0.9514154195785522,
      "learning_rate": 9.518271311074917e-06,
      "loss": 0.7864,
      "step": 5205
    },
    {
      "epoch": 0.2865320050635698,
      "grad_norm": 0.830795407295227,
      "learning_rate": 9.51808565649503e-06,
      "loss": 0.8024,
      "step": 5206
    },
    {
      "epoch": 0.2865870438659255,
      "grad_norm": 0.7274934649467468,
      "learning_rate": 9.51789996795842e-06,
      "loss": 0.7631,
      "step": 5207
    },
    {
      "epoch": 0.2866420826682811,
      "grad_norm": 0.7004290223121643,
      "learning_rate": 9.517714245466482e-06,
      "loss": 0.7344,
      "step": 5208
    },
    {
      "epoch": 0.2866971214706368,
      "grad_norm": 0.8559010624885559,
      "learning_rate": 9.517528489020614e-06,
      "loss": 0.7502,
      "step": 5209
    },
    {
      "epoch": 0.28675216027299244,
      "grad_norm": 0.8913494348526001,
      "learning_rate": 9.517342698622212e-06,
      "loss": 0.8908,
      "step": 5210
    },
    {
      "epoch": 0.28680719907534813,
      "grad_norm": 0.8375207781791687,
      "learning_rate": 9.51715687427267e-06,
      "loss": 0.7701,
      "step": 5211
    },
    {
      "epoch": 0.28686223787770376,
      "grad_norm": 1.1804776191711426,
      "learning_rate": 9.516971015973386e-06,
      "loss": 0.8449,
      "step": 5212
    },
    {
      "epoch": 0.28691727668005945,
      "grad_norm": 0.7260473370552063,
      "learning_rate": 9.516785123725758e-06,
      "loss": 0.7978,
      "step": 5213
    },
    {
      "epoch": 0.2869723154824151,
      "grad_norm": 0.8159041404724121,
      "learning_rate": 9.516599197531182e-06,
      "loss": 0.7454,
      "step": 5214
    },
    {
      "epoch": 0.2870273542847708,
      "grad_norm": 0.7850227952003479,
      "learning_rate": 9.516413237391056e-06,
      "loss": 0.8082,
      "step": 5215
    },
    {
      "epoch": 0.2870823930871264,
      "grad_norm": 0.7596960067749023,
      "learning_rate": 9.516227243306774e-06,
      "loss": 0.7286,
      "step": 5216
    },
    {
      "epoch": 0.2871374318894821,
      "grad_norm": 0.8763321042060852,
      "learning_rate": 9.516041215279741e-06,
      "loss": 0.8685,
      "step": 5217
    },
    {
      "epoch": 0.28719247069183773,
      "grad_norm": 1.2130110263824463,
      "learning_rate": 9.515855153311349e-06,
      "loss": 0.8374,
      "step": 5218
    },
    {
      "epoch": 0.2872475094941934,
      "grad_norm": 0.7578628063201904,
      "learning_rate": 9.515669057402999e-06,
      "loss": 0.793,
      "step": 5219
    },
    {
      "epoch": 0.28730254829654905,
      "grad_norm": 0.9085225462913513,
      "learning_rate": 9.515482927556088e-06,
      "loss": 0.8366,
      "step": 5220
    },
    {
      "epoch": 0.28735758709890474,
      "grad_norm": 0.7107900977134705,
      "learning_rate": 9.515296763772017e-06,
      "loss": 0.6571,
      "step": 5221
    },
    {
      "epoch": 0.2874126259012604,
      "grad_norm": 0.7742018699645996,
      "learning_rate": 9.515110566052183e-06,
      "loss": 0.8387,
      "step": 5222
    },
    {
      "epoch": 0.28746766470361607,
      "grad_norm": 0.8934319615364075,
      "learning_rate": 9.514924334397987e-06,
      "loss": 0.8546,
      "step": 5223
    },
    {
      "epoch": 0.2875227035059717,
      "grad_norm": 0.720245897769928,
      "learning_rate": 9.51473806881083e-06,
      "loss": 0.7459,
      "step": 5224
    },
    {
      "epoch": 0.2875777423083274,
      "grad_norm": 0.7074370384216309,
      "learning_rate": 9.514551769292109e-06,
      "loss": 0.8598,
      "step": 5225
    },
    {
      "epoch": 0.287632781110683,
      "grad_norm": 0.7608621120452881,
      "learning_rate": 9.514365435843226e-06,
      "loss": 0.7263,
      "step": 5226
    },
    {
      "epoch": 0.2876878199130387,
      "grad_norm": 0.7581011652946472,
      "learning_rate": 9.51417906846558e-06,
      "loss": 0.7498,
      "step": 5227
    },
    {
      "epoch": 0.28774285871539435,
      "grad_norm": 0.8184412121772766,
      "learning_rate": 9.513992667160572e-06,
      "loss": 0.6889,
      "step": 5228
    },
    {
      "epoch": 0.28779789751775003,
      "grad_norm": 0.6835145354270935,
      "learning_rate": 9.513806231929605e-06,
      "loss": 0.7399,
      "step": 5229
    },
    {
      "epoch": 0.28785293632010567,
      "grad_norm": 0.7601536512374878,
      "learning_rate": 9.513619762774077e-06,
      "loss": 0.846,
      "step": 5230
    },
    {
      "epoch": 0.28790797512246136,
      "grad_norm": 0.781491219997406,
      "learning_rate": 9.513433259695392e-06,
      "loss": 0.8326,
      "step": 5231
    },
    {
      "epoch": 0.287963013924817,
      "grad_norm": 0.7978106141090393,
      "learning_rate": 9.513246722694951e-06,
      "loss": 0.7917,
      "step": 5232
    },
    {
      "epoch": 0.2880180527271727,
      "grad_norm": 0.8071381449699402,
      "learning_rate": 9.513060151774156e-06,
      "loss": 0.8054,
      "step": 5233
    },
    {
      "epoch": 0.2880730915295283,
      "grad_norm": 0.815567135810852,
      "learning_rate": 9.512873546934406e-06,
      "loss": 0.8647,
      "step": 5234
    },
    {
      "epoch": 0.288128130331884,
      "grad_norm": 0.8255048990249634,
      "learning_rate": 9.512686908177111e-06,
      "loss": 0.9011,
      "step": 5235
    },
    {
      "epoch": 0.28818316913423964,
      "grad_norm": 0.8392062187194824,
      "learning_rate": 9.512500235503666e-06,
      "loss": 0.8778,
      "step": 5236
    },
    {
      "epoch": 0.2882382079365953,
      "grad_norm": 0.7256191372871399,
      "learning_rate": 9.512313528915478e-06,
      "loss": 0.7231,
      "step": 5237
    },
    {
      "epoch": 0.28829324673895096,
      "grad_norm": 0.9041032195091248,
      "learning_rate": 9.51212678841395e-06,
      "loss": 0.8469,
      "step": 5238
    },
    {
      "epoch": 0.28834828554130665,
      "grad_norm": 0.7857525944709778,
      "learning_rate": 9.511940014000485e-06,
      "loss": 0.7447,
      "step": 5239
    },
    {
      "epoch": 0.2884033243436623,
      "grad_norm": 0.6925225257873535,
      "learning_rate": 9.511753205676485e-06,
      "loss": 0.8302,
      "step": 5240
    },
    {
      "epoch": 0.2884583631460179,
      "grad_norm": 0.7253623008728027,
      "learning_rate": 9.511566363443356e-06,
      "loss": 0.8373,
      "step": 5241
    },
    {
      "epoch": 0.2885134019483736,
      "grad_norm": 0.7198607921600342,
      "learning_rate": 9.511379487302504e-06,
      "loss": 0.79,
      "step": 5242
    },
    {
      "epoch": 0.28856844075072924,
      "grad_norm": 0.7966421246528625,
      "learning_rate": 9.511192577255328e-06,
      "loss": 0.7933,
      "step": 5243
    },
    {
      "epoch": 0.2886234795530849,
      "grad_norm": 0.9159359931945801,
      "learning_rate": 9.511005633303239e-06,
      "loss": 0.7254,
      "step": 5244
    },
    {
      "epoch": 0.28867851835544056,
      "grad_norm": 0.9514481425285339,
      "learning_rate": 9.510818655447638e-06,
      "loss": 0.8916,
      "step": 5245
    },
    {
      "epoch": 0.28873355715779625,
      "grad_norm": 0.7505099773406982,
      "learning_rate": 9.510631643689932e-06,
      "loss": 0.765,
      "step": 5246
    },
    {
      "epoch": 0.2887885959601519,
      "grad_norm": 0.7824658751487732,
      "learning_rate": 9.510444598031526e-06,
      "loss": 0.6972,
      "step": 5247
    },
    {
      "epoch": 0.2888436347625076,
      "grad_norm": 0.7778681516647339,
      "learning_rate": 9.510257518473824e-06,
      "loss": 0.8705,
      "step": 5248
    },
    {
      "epoch": 0.2888986735648632,
      "grad_norm": 0.6785199642181396,
      "learning_rate": 9.510070405018235e-06,
      "loss": 0.6889,
      "step": 5249
    },
    {
      "epoch": 0.2889537123672189,
      "grad_norm": 0.7045316100120544,
      "learning_rate": 9.509883257666164e-06,
      "loss": 0.7979,
      "step": 5250
    },
    {
      "epoch": 0.28900875116957453,
      "grad_norm": 1.3174562454223633,
      "learning_rate": 9.509696076419018e-06,
      "loss": 0.8802,
      "step": 5251
    },
    {
      "epoch": 0.2890637899719302,
      "grad_norm": 1.1800767183303833,
      "learning_rate": 9.509508861278205e-06,
      "loss": 0.9246,
      "step": 5252
    },
    {
      "epoch": 0.28911882877428585,
      "grad_norm": 0.7057580947875977,
      "learning_rate": 9.509321612245128e-06,
      "loss": 0.7565,
      "step": 5253
    },
    {
      "epoch": 0.28917386757664154,
      "grad_norm": 0.7681905031204224,
      "learning_rate": 9.509134329321197e-06,
      "loss": 0.8678,
      "step": 5254
    },
    {
      "epoch": 0.2892289063789972,
      "grad_norm": 0.96025550365448,
      "learning_rate": 9.50894701250782e-06,
      "loss": 0.9108,
      "step": 5255
    },
    {
      "epoch": 0.28928394518135286,
      "grad_norm": 0.7786841988563538,
      "learning_rate": 9.508759661806405e-06,
      "loss": 0.7747,
      "step": 5256
    },
    {
      "epoch": 0.2893389839837085,
      "grad_norm": 0.7073540091514587,
      "learning_rate": 9.508572277218358e-06,
      "loss": 0.7573,
      "step": 5257
    },
    {
      "epoch": 0.2893940227860642,
      "grad_norm": 0.6648856401443481,
      "learning_rate": 9.50838485874509e-06,
      "loss": 0.7294,
      "step": 5258
    },
    {
      "epoch": 0.2894490615884198,
      "grad_norm": 0.6794270873069763,
      "learning_rate": 9.508197406388007e-06,
      "loss": 0.7001,
      "step": 5259
    },
    {
      "epoch": 0.2895041003907755,
      "grad_norm": 0.6819350123405457,
      "learning_rate": 9.50800992014852e-06,
      "loss": 0.7114,
      "step": 5260
    },
    {
      "epoch": 0.28955913919313114,
      "grad_norm": 0.6616997122764587,
      "learning_rate": 9.507822400028036e-06,
      "loss": 0.7108,
      "step": 5261
    },
    {
      "epoch": 0.28961417799548683,
      "grad_norm": 0.7447230219841003,
      "learning_rate": 9.507634846027966e-06,
      "loss": 0.7865,
      "step": 5262
    },
    {
      "epoch": 0.28966921679784247,
      "grad_norm": 0.7826278209686279,
      "learning_rate": 9.50744725814972e-06,
      "loss": 0.7922,
      "step": 5263
    },
    {
      "epoch": 0.28972425560019816,
      "grad_norm": 0.8054459095001221,
      "learning_rate": 9.507259636394706e-06,
      "loss": 0.795,
      "step": 5264
    },
    {
      "epoch": 0.2897792944025538,
      "grad_norm": 0.9539191722869873,
      "learning_rate": 9.507071980764335e-06,
      "loss": 0.9495,
      "step": 5265
    },
    {
      "epoch": 0.2898343332049095,
      "grad_norm": 0.8877993226051331,
      "learning_rate": 9.506884291260017e-06,
      "loss": 0.8418,
      "step": 5266
    },
    {
      "epoch": 0.2898893720072651,
      "grad_norm": 0.6620327234268188,
      "learning_rate": 9.506696567883164e-06,
      "loss": 0.6285,
      "step": 5267
    },
    {
      "epoch": 0.2899444108096208,
      "grad_norm": 0.7604434490203857,
      "learning_rate": 9.506508810635187e-06,
      "loss": 0.8562,
      "step": 5268
    },
    {
      "epoch": 0.28999944961197643,
      "grad_norm": 0.8181812763214111,
      "learning_rate": 9.506321019517494e-06,
      "loss": 0.905,
      "step": 5269
    },
    {
      "epoch": 0.2900544884143321,
      "grad_norm": 0.7776391506195068,
      "learning_rate": 9.5061331945315e-06,
      "loss": 0.8871,
      "step": 5270
    },
    {
      "epoch": 0.29010952721668776,
      "grad_norm": 0.8125039339065552,
      "learning_rate": 9.505945335678613e-06,
      "loss": 0.7254,
      "step": 5271
    },
    {
      "epoch": 0.29016456601904345,
      "grad_norm": 0.7229846715927124,
      "learning_rate": 9.50575744296025e-06,
      "loss": 0.8192,
      "step": 5272
    },
    {
      "epoch": 0.2902196048213991,
      "grad_norm": 0.72443026304245,
      "learning_rate": 9.505569516377817e-06,
      "loss": 0.7813,
      "step": 5273
    },
    {
      "epoch": 0.29027464362375477,
      "grad_norm": 0.6798073053359985,
      "learning_rate": 9.505381555932731e-06,
      "loss": 0.7655,
      "step": 5274
    },
    {
      "epoch": 0.2903296824261104,
      "grad_norm": 1.0805624723434448,
      "learning_rate": 9.505193561626404e-06,
      "loss": 0.9035,
      "step": 5275
    },
    {
      "epoch": 0.2903847212284661,
      "grad_norm": 0.7579694986343384,
      "learning_rate": 9.505005533460247e-06,
      "loss": 0.8612,
      "step": 5276
    },
    {
      "epoch": 0.2904397600308217,
      "grad_norm": 1.2496099472045898,
      "learning_rate": 9.504817471435676e-06,
      "loss": 0.813,
      "step": 5277
    },
    {
      "epoch": 0.2904947988331774,
      "grad_norm": 0.6915673017501831,
      "learning_rate": 9.504629375554102e-06,
      "loss": 0.6891,
      "step": 5278
    },
    {
      "epoch": 0.29054983763553305,
      "grad_norm": 0.8581767082214355,
      "learning_rate": 9.504441245816937e-06,
      "loss": 0.7137,
      "step": 5279
    },
    {
      "epoch": 0.29060487643788874,
      "grad_norm": 0.7469545006752014,
      "learning_rate": 9.504253082225601e-06,
      "loss": 0.7621,
      "step": 5280
    },
    {
      "epoch": 0.29065991524024437,
      "grad_norm": 0.7725615501403809,
      "learning_rate": 9.504064884781503e-06,
      "loss": 0.7988,
      "step": 5281
    },
    {
      "epoch": 0.29071495404260006,
      "grad_norm": 1.0187722444534302,
      "learning_rate": 9.503876653486058e-06,
      "loss": 0.7772,
      "step": 5282
    },
    {
      "epoch": 0.2907699928449557,
      "grad_norm": 0.675574779510498,
      "learning_rate": 9.503688388340683e-06,
      "loss": 0.7096,
      "step": 5283
    },
    {
      "epoch": 0.2908250316473113,
      "grad_norm": 0.7980207800865173,
      "learning_rate": 9.503500089346792e-06,
      "loss": 0.8291,
      "step": 5284
    },
    {
      "epoch": 0.290880070449667,
      "grad_norm": 0.6891655325889587,
      "learning_rate": 9.503311756505797e-06,
      "loss": 0.7186,
      "step": 5285
    },
    {
      "epoch": 0.29093510925202265,
      "grad_norm": 0.7273408770561218,
      "learning_rate": 9.50312338981912e-06,
      "loss": 0.7483,
      "step": 5286
    },
    {
      "epoch": 0.29099014805437834,
      "grad_norm": 0.7346869111061096,
      "learning_rate": 9.50293498928817e-06,
      "loss": 0.766,
      "step": 5287
    },
    {
      "epoch": 0.291045186856734,
      "grad_norm": 0.7627394795417786,
      "learning_rate": 9.502746554914368e-06,
      "loss": 0.867,
      "step": 5288
    },
    {
      "epoch": 0.29110022565908966,
      "grad_norm": 0.8477200865745544,
      "learning_rate": 9.502558086699128e-06,
      "loss": 0.8317,
      "step": 5289
    },
    {
      "epoch": 0.2911552644614453,
      "grad_norm": 0.7696006894111633,
      "learning_rate": 9.502369584643867e-06,
      "loss": 0.7814,
      "step": 5290
    },
    {
      "epoch": 0.291210303263801,
      "grad_norm": 0.7614455819129944,
      "learning_rate": 9.502181048749999e-06,
      "loss": 0.7398,
      "step": 5291
    },
    {
      "epoch": 0.2912653420661566,
      "grad_norm": 0.7877628207206726,
      "learning_rate": 9.501992479018946e-06,
      "loss": 0.8731,
      "step": 5292
    },
    {
      "epoch": 0.2913203808685123,
      "grad_norm": 0.7455846667289734,
      "learning_rate": 9.50180387545212e-06,
      "loss": 0.7059,
      "step": 5293
    },
    {
      "epoch": 0.29137541967086794,
      "grad_norm": 1.145520567893982,
      "learning_rate": 9.501615238050944e-06,
      "loss": 0.6968,
      "step": 5294
    },
    {
      "epoch": 0.29143045847322363,
      "grad_norm": 0.8100234866142273,
      "learning_rate": 9.501426566816831e-06,
      "loss": 0.8122,
      "step": 5295
    },
    {
      "epoch": 0.29148549727557926,
      "grad_norm": 0.6813066005706787,
      "learning_rate": 9.501237861751203e-06,
      "loss": 0.6718,
      "step": 5296
    },
    {
      "epoch": 0.29154053607793495,
      "grad_norm": 0.7400195002555847,
      "learning_rate": 9.501049122855473e-06,
      "loss": 0.802,
      "step": 5297
    },
    {
      "epoch": 0.2915955748802906,
      "grad_norm": 0.7948681712150574,
      "learning_rate": 9.500860350131065e-06,
      "loss": 0.8237,
      "step": 5298
    },
    {
      "epoch": 0.2916506136826463,
      "grad_norm": 0.772093653678894,
      "learning_rate": 9.500671543579394e-06,
      "loss": 0.7687,
      "step": 5299
    },
    {
      "epoch": 0.2917056524850019,
      "grad_norm": 0.7468486428260803,
      "learning_rate": 9.500482703201881e-06,
      "loss": 0.7827,
      "step": 5300
    },
    {
      "epoch": 0.2917606912873576,
      "grad_norm": 0.7284440398216248,
      "learning_rate": 9.500293828999945e-06,
      "loss": 0.8086,
      "step": 5301
    },
    {
      "epoch": 0.29181573008971323,
      "grad_norm": 0.8014211654663086,
      "learning_rate": 9.500104920975005e-06,
      "loss": 0.8409,
      "step": 5302
    },
    {
      "epoch": 0.2918707688920689,
      "grad_norm": 0.7588346004486084,
      "learning_rate": 9.49991597912848e-06,
      "loss": 0.7149,
      "step": 5303
    },
    {
      "epoch": 0.29192580769442456,
      "grad_norm": 0.8098518252372742,
      "learning_rate": 9.499727003461794e-06,
      "loss": 0.8375,
      "step": 5304
    },
    {
      "epoch": 0.29198084649678024,
      "grad_norm": 0.8502426743507385,
      "learning_rate": 9.499537993976363e-06,
      "loss": 0.8177,
      "step": 5305
    },
    {
      "epoch": 0.2920358852991359,
      "grad_norm": 0.8010903596878052,
      "learning_rate": 9.499348950673607e-06,
      "loss": 0.8457,
      "step": 5306
    },
    {
      "epoch": 0.29209092410149157,
      "grad_norm": 0.6628156304359436,
      "learning_rate": 9.49915987355495e-06,
      "loss": 0.7327,
      "step": 5307
    },
    {
      "epoch": 0.2921459629038472,
      "grad_norm": 0.7414939999580383,
      "learning_rate": 9.49897076262181e-06,
      "loss": 0.8271,
      "step": 5308
    },
    {
      "epoch": 0.2922010017062029,
      "grad_norm": 0.7490847706794739,
      "learning_rate": 9.498781617875613e-06,
      "loss": 0.7689,
      "step": 5309
    },
    {
      "epoch": 0.2922560405085585,
      "grad_norm": 0.7913424968719482,
      "learning_rate": 9.498592439317777e-06,
      "loss": 0.8571,
      "step": 5310
    },
    {
      "epoch": 0.2923110793109142,
      "grad_norm": 0.6903867125511169,
      "learning_rate": 9.498403226949724e-06,
      "loss": 0.7325,
      "step": 5311
    },
    {
      "epoch": 0.29236611811326985,
      "grad_norm": 0.8087130188941956,
      "learning_rate": 9.498213980772875e-06,
      "loss": 0.8167,
      "step": 5312
    },
    {
      "epoch": 0.29242115691562554,
      "grad_norm": 1.1316752433776855,
      "learning_rate": 9.498024700788655e-06,
      "loss": 0.912,
      "step": 5313
    },
    {
      "epoch": 0.29247619571798117,
      "grad_norm": 0.8701719045639038,
      "learning_rate": 9.497835386998486e-06,
      "loss": 0.8728,
      "step": 5314
    },
    {
      "epoch": 0.29253123452033686,
      "grad_norm": 0.6688953638076782,
      "learning_rate": 9.49764603940379e-06,
      "loss": 0.6561,
      "step": 5315
    },
    {
      "epoch": 0.2925862733226925,
      "grad_norm": 0.8067505359649658,
      "learning_rate": 9.49745665800599e-06,
      "loss": 0.8419,
      "step": 5316
    },
    {
      "epoch": 0.2926413121250482,
      "grad_norm": 0.7157390117645264,
      "learning_rate": 9.49726724280651e-06,
      "loss": 0.7964,
      "step": 5317
    },
    {
      "epoch": 0.2926963509274038,
      "grad_norm": 0.7038627862930298,
      "learning_rate": 9.497077793806772e-06,
      "loss": 0.7343,
      "step": 5318
    },
    {
      "epoch": 0.2927513897297595,
      "grad_norm": 0.7674478888511658,
      "learning_rate": 9.4968883110082e-06,
      "loss": 0.7624,
      "step": 5319
    },
    {
      "epoch": 0.29280642853211514,
      "grad_norm": 0.6708847284317017,
      "learning_rate": 9.496698794412223e-06,
      "loss": 0.6554,
      "step": 5320
    },
    {
      "epoch": 0.2928614673344708,
      "grad_norm": 0.8332329392433167,
      "learning_rate": 9.49650924402026e-06,
      "loss": 0.9357,
      "step": 5321
    },
    {
      "epoch": 0.29291650613682646,
      "grad_norm": 0.7601341605186462,
      "learning_rate": 9.496319659833737e-06,
      "loss": 0.8208,
      "step": 5322
    },
    {
      "epoch": 0.29297154493918215,
      "grad_norm": 0.8320396542549133,
      "learning_rate": 9.496130041854077e-06,
      "loss": 0.8423,
      "step": 5323
    },
    {
      "epoch": 0.2930265837415378,
      "grad_norm": 0.8242839574813843,
      "learning_rate": 9.49594039008271e-06,
      "loss": 0.9101,
      "step": 5324
    },
    {
      "epoch": 0.29308162254389347,
      "grad_norm": 0.8906320333480835,
      "learning_rate": 9.495750704521058e-06,
      "loss": 0.7343,
      "step": 5325
    },
    {
      "epoch": 0.2931366613462491,
      "grad_norm": 0.7964318990707397,
      "learning_rate": 9.495560985170546e-06,
      "loss": 0.7789,
      "step": 5326
    },
    {
      "epoch": 0.29319170014860474,
      "grad_norm": 0.8267771601676941,
      "learning_rate": 9.495371232032602e-06,
      "loss": 0.7447,
      "step": 5327
    },
    {
      "epoch": 0.29324673895096043,
      "grad_norm": 0.8120046257972717,
      "learning_rate": 9.49518144510865e-06,
      "loss": 0.7803,
      "step": 5328
    },
    {
      "epoch": 0.29330177775331606,
      "grad_norm": 0.7314801812171936,
      "learning_rate": 9.494991624400119e-06,
      "loss": 0.6758,
      "step": 5329
    },
    {
      "epoch": 0.29335681655567175,
      "grad_norm": 0.6989930272102356,
      "learning_rate": 9.494801769908433e-06,
      "loss": 0.7945,
      "step": 5330
    },
    {
      "epoch": 0.2934118553580274,
      "grad_norm": 0.7804785966873169,
      "learning_rate": 9.494611881635021e-06,
      "loss": 0.7977,
      "step": 5331
    },
    {
      "epoch": 0.2934668941603831,
      "grad_norm": 0.8377045392990112,
      "learning_rate": 9.494421959581308e-06,
      "loss": 0.8077,
      "step": 5332
    },
    {
      "epoch": 0.2935219329627387,
      "grad_norm": 0.7463418245315552,
      "learning_rate": 9.494232003748724e-06,
      "loss": 0.783,
      "step": 5333
    },
    {
      "epoch": 0.2935769717650944,
      "grad_norm": 0.7598912715911865,
      "learning_rate": 9.494042014138695e-06,
      "loss": 0.7869,
      "step": 5334
    },
    {
      "epoch": 0.29363201056745003,
      "grad_norm": 0.7634113430976868,
      "learning_rate": 9.493851990752648e-06,
      "loss": 0.8108,
      "step": 5335
    },
    {
      "epoch": 0.2936870493698057,
      "grad_norm": 0.8056474328041077,
      "learning_rate": 9.493661933592013e-06,
      "loss": 0.7921,
      "step": 5336
    },
    {
      "epoch": 0.29374208817216135,
      "grad_norm": 0.8699371218681335,
      "learning_rate": 9.493471842658219e-06,
      "loss": 0.8833,
      "step": 5337
    },
    {
      "epoch": 0.29379712697451704,
      "grad_norm": 0.8803261518478394,
      "learning_rate": 9.493281717952691e-06,
      "loss": 0.7848,
      "step": 5338
    },
    {
      "epoch": 0.2938521657768727,
      "grad_norm": 0.7678453922271729,
      "learning_rate": 9.493091559476864e-06,
      "loss": 0.836,
      "step": 5339
    },
    {
      "epoch": 0.29390720457922836,
      "grad_norm": 0.7653701305389404,
      "learning_rate": 9.49290136723216e-06,
      "loss": 0.8215,
      "step": 5340
    },
    {
      "epoch": 0.293962243381584,
      "grad_norm": 0.768120527267456,
      "learning_rate": 9.492711141220013e-06,
      "loss": 0.7498,
      "step": 5341
    },
    {
      "epoch": 0.2940172821839397,
      "grad_norm": 0.7665749788284302,
      "learning_rate": 9.492520881441854e-06,
      "loss": 0.7883,
      "step": 5342
    },
    {
      "epoch": 0.2940723209862953,
      "grad_norm": 0.7405015230178833,
      "learning_rate": 9.492330587899108e-06,
      "loss": 0.8112,
      "step": 5343
    },
    {
      "epoch": 0.294127359788651,
      "grad_norm": 0.7183459997177124,
      "learning_rate": 9.492140260593208e-06,
      "loss": 0.8227,
      "step": 5344
    },
    {
      "epoch": 0.29418239859100664,
      "grad_norm": 0.7453572154045105,
      "learning_rate": 9.491949899525585e-06,
      "loss": 0.8148,
      "step": 5345
    },
    {
      "epoch": 0.29423743739336233,
      "grad_norm": 0.8963750600814819,
      "learning_rate": 9.491759504697669e-06,
      "loss": 0.9261,
      "step": 5346
    },
    {
      "epoch": 0.29429247619571797,
      "grad_norm": 0.7631667256355286,
      "learning_rate": 9.49156907611089e-06,
      "loss": 0.7708,
      "step": 5347
    },
    {
      "epoch": 0.29434751499807366,
      "grad_norm": 0.6324381232261658,
      "learning_rate": 9.49137861376668e-06,
      "loss": 0.6688,
      "step": 5348
    },
    {
      "epoch": 0.2944025538004293,
      "grad_norm": 0.6969807147979736,
      "learning_rate": 9.491188117666472e-06,
      "loss": 0.7516,
      "step": 5349
    },
    {
      "epoch": 0.294457592602785,
      "grad_norm": 1.633340835571289,
      "learning_rate": 9.490997587811697e-06,
      "loss": 0.8111,
      "step": 5350
    },
    {
      "epoch": 0.2945126314051406,
      "grad_norm": 0.7084371447563171,
      "learning_rate": 9.490807024203785e-06,
      "loss": 0.8375,
      "step": 5351
    },
    {
      "epoch": 0.2945676702074963,
      "grad_norm": 0.7335958480834961,
      "learning_rate": 9.490616426844169e-06,
      "loss": 0.7884,
      "step": 5352
    },
    {
      "epoch": 0.29462270900985194,
      "grad_norm": 0.7560276985168457,
      "learning_rate": 9.490425795734282e-06,
      "loss": 0.8918,
      "step": 5353
    },
    {
      "epoch": 0.2946777478122076,
      "grad_norm": 0.9185894727706909,
      "learning_rate": 9.490235130875557e-06,
      "loss": 0.7976,
      "step": 5354
    },
    {
      "epoch": 0.29473278661456326,
      "grad_norm": 0.7871553897857666,
      "learning_rate": 9.490044432269427e-06,
      "loss": 0.8564,
      "step": 5355
    },
    {
      "epoch": 0.29478782541691895,
      "grad_norm": 0.8736812472343445,
      "learning_rate": 9.489853699917326e-06,
      "loss": 0.8114,
      "step": 5356
    },
    {
      "epoch": 0.2948428642192746,
      "grad_norm": 0.8068968653678894,
      "learning_rate": 9.489662933820684e-06,
      "loss": 0.9198,
      "step": 5357
    },
    {
      "epoch": 0.29489790302163027,
      "grad_norm": 0.7816325426101685,
      "learning_rate": 9.489472133980939e-06,
      "loss": 0.8012,
      "step": 5358
    },
    {
      "epoch": 0.2949529418239859,
      "grad_norm": 0.7248200178146362,
      "learning_rate": 9.489281300399522e-06,
      "loss": 0.8099,
      "step": 5359
    },
    {
      "epoch": 0.2950079806263416,
      "grad_norm": 0.7887724041938782,
      "learning_rate": 9.48909043307787e-06,
      "loss": 0.884,
      "step": 5360
    },
    {
      "epoch": 0.2950630194286972,
      "grad_norm": 0.765163004398346,
      "learning_rate": 9.488899532017415e-06,
      "loss": 0.8563,
      "step": 5361
    },
    {
      "epoch": 0.2951180582310529,
      "grad_norm": 0.7658557295799255,
      "learning_rate": 9.488708597219592e-06,
      "loss": 0.8897,
      "step": 5362
    },
    {
      "epoch": 0.29517309703340855,
      "grad_norm": 0.6653227806091309,
      "learning_rate": 9.488517628685838e-06,
      "loss": 0.7107,
      "step": 5363
    },
    {
      "epoch": 0.29522813583576424,
      "grad_norm": 0.787739098072052,
      "learning_rate": 9.488326626417586e-06,
      "loss": 0.8181,
      "step": 5364
    },
    {
      "epoch": 0.29528317463811987,
      "grad_norm": 0.7822532057762146,
      "learning_rate": 9.488135590416275e-06,
      "loss": 0.8238,
      "step": 5365
    },
    {
      "epoch": 0.29533821344047556,
      "grad_norm": 0.7797419428825378,
      "learning_rate": 9.487944520683334e-06,
      "loss": 0.8484,
      "step": 5366
    },
    {
      "epoch": 0.2953932522428312,
      "grad_norm": 0.7230222225189209,
      "learning_rate": 9.487753417220207e-06,
      "loss": 0.8193,
      "step": 5367
    },
    {
      "epoch": 0.2954482910451869,
      "grad_norm": 0.8256810307502747,
      "learning_rate": 9.487562280028325e-06,
      "loss": 0.7691,
      "step": 5368
    },
    {
      "epoch": 0.2955033298475425,
      "grad_norm": 0.7704648375511169,
      "learning_rate": 9.487371109109127e-06,
      "loss": 0.8235,
      "step": 5369
    },
    {
      "epoch": 0.29555836864989815,
      "grad_norm": 0.7580391764640808,
      "learning_rate": 9.487179904464048e-06,
      "loss": 0.7911,
      "step": 5370
    },
    {
      "epoch": 0.29561340745225384,
      "grad_norm": 0.7211806774139404,
      "learning_rate": 9.486988666094526e-06,
      "loss": 0.7188,
      "step": 5371
    },
    {
      "epoch": 0.2956684462546095,
      "grad_norm": 0.8375828862190247,
      "learning_rate": 9.486797394001999e-06,
      "loss": 0.881,
      "step": 5372
    },
    {
      "epoch": 0.29572348505696516,
      "grad_norm": 0.8500093221664429,
      "learning_rate": 9.486606088187903e-06,
      "loss": 0.8632,
      "step": 5373
    },
    {
      "epoch": 0.2957785238593208,
      "grad_norm": 0.7754727005958557,
      "learning_rate": 9.486414748653677e-06,
      "loss": 0.8124,
      "step": 5374
    },
    {
      "epoch": 0.2958335626616765,
      "grad_norm": 0.9395208954811096,
      "learning_rate": 9.486223375400759e-06,
      "loss": 0.8046,
      "step": 5375
    },
    {
      "epoch": 0.2958886014640321,
      "grad_norm": 0.7587517499923706,
      "learning_rate": 9.486031968430587e-06,
      "loss": 0.7852,
      "step": 5376
    },
    {
      "epoch": 0.2959436402663878,
      "grad_norm": 0.6921781301498413,
      "learning_rate": 9.485840527744599e-06,
      "loss": 0.7392,
      "step": 5377
    },
    {
      "epoch": 0.29599867906874344,
      "grad_norm": 0.8768522143363953,
      "learning_rate": 9.485649053344233e-06,
      "loss": 0.7819,
      "step": 5378
    },
    {
      "epoch": 0.29605371787109913,
      "grad_norm": 0.7565680146217346,
      "learning_rate": 9.485457545230932e-06,
      "loss": 0.7489,
      "step": 5379
    },
    {
      "epoch": 0.29610875667345476,
      "grad_norm": 0.7760992050170898,
      "learning_rate": 9.485266003406132e-06,
      "loss": 0.8129,
      "step": 5380
    },
    {
      "epoch": 0.29616379547581045,
      "grad_norm": 0.7726097106933594,
      "learning_rate": 9.485074427871272e-06,
      "loss": 0.725,
      "step": 5381
    },
    {
      "epoch": 0.2962188342781661,
      "grad_norm": 0.6885473728179932,
      "learning_rate": 9.484882818627796e-06,
      "loss": 0.685,
      "step": 5382
    },
    {
      "epoch": 0.2962738730805218,
      "grad_norm": 0.776509702205658,
      "learning_rate": 9.484691175677138e-06,
      "loss": 0.8077,
      "step": 5383
    },
    {
      "epoch": 0.2963289118828774,
      "grad_norm": 0.7436297535896301,
      "learning_rate": 9.484499499020744e-06,
      "loss": 0.8161,
      "step": 5384
    },
    {
      "epoch": 0.2963839506852331,
      "grad_norm": 0.7604314088821411,
      "learning_rate": 9.484307788660052e-06,
      "loss": 0.825,
      "step": 5385
    },
    {
      "epoch": 0.29643898948758873,
      "grad_norm": 0.7230789065361023,
      "learning_rate": 9.484116044596501e-06,
      "loss": 0.8005,
      "step": 5386
    },
    {
      "epoch": 0.2964940282899444,
      "grad_norm": 0.820442259311676,
      "learning_rate": 9.483924266831536e-06,
      "loss": 0.789,
      "step": 5387
    },
    {
      "epoch": 0.29654906709230006,
      "grad_norm": 0.7514582276344299,
      "learning_rate": 9.483732455366596e-06,
      "loss": 0.8531,
      "step": 5388
    },
    {
      "epoch": 0.29660410589465575,
      "grad_norm": 0.6671503782272339,
      "learning_rate": 9.483540610203124e-06,
      "loss": 0.7627,
      "step": 5389
    },
    {
      "epoch": 0.2966591446970114,
      "grad_norm": 0.6955942511558533,
      "learning_rate": 9.483348731342559e-06,
      "loss": 0.726,
      "step": 5390
    },
    {
      "epoch": 0.29671418349936707,
      "grad_norm": 0.769781768321991,
      "learning_rate": 9.483156818786347e-06,
      "loss": 0.8064,
      "step": 5391
    },
    {
      "epoch": 0.2967692223017227,
      "grad_norm": 1.0764707326889038,
      "learning_rate": 9.482964872535927e-06,
      "loss": 0.8249,
      "step": 5392
    },
    {
      "epoch": 0.2968242611040784,
      "grad_norm": 1.0508921146392822,
      "learning_rate": 9.482772892592744e-06,
      "loss": 0.706,
      "step": 5393
    },
    {
      "epoch": 0.296879299906434,
      "grad_norm": 0.6442564129829407,
      "learning_rate": 9.482580878958239e-06,
      "loss": 0.6025,
      "step": 5394
    },
    {
      "epoch": 0.2969343387087897,
      "grad_norm": 0.7622735500335693,
      "learning_rate": 9.482388831633856e-06,
      "loss": 0.7639,
      "step": 5395
    },
    {
      "epoch": 0.29698937751114535,
      "grad_norm": 0.8179057240486145,
      "learning_rate": 9.482196750621038e-06,
      "loss": 0.7641,
      "step": 5396
    },
    {
      "epoch": 0.29704441631350104,
      "grad_norm": 0.7955192923545837,
      "learning_rate": 9.48200463592123e-06,
      "loss": 0.8407,
      "step": 5397
    },
    {
      "epoch": 0.29709945511585667,
      "grad_norm": 0.7909773588180542,
      "learning_rate": 9.481812487535875e-06,
      "loss": 0.7833,
      "step": 5398
    },
    {
      "epoch": 0.29715449391821236,
      "grad_norm": 0.8409042954444885,
      "learning_rate": 9.481620305466417e-06,
      "loss": 0.7788,
      "step": 5399
    },
    {
      "epoch": 0.297209532720568,
      "grad_norm": 0.7521414160728455,
      "learning_rate": 9.4814280897143e-06,
      "loss": 0.7192,
      "step": 5400
    },
    {
      "epoch": 0.2972645715229237,
      "grad_norm": 0.7016280889511108,
      "learning_rate": 9.481235840280969e-06,
      "loss": 0.7181,
      "step": 5401
    },
    {
      "epoch": 0.2973196103252793,
      "grad_norm": 0.7257362604141235,
      "learning_rate": 9.48104355716787e-06,
      "loss": 0.7845,
      "step": 5402
    },
    {
      "epoch": 0.297374649127635,
      "grad_norm": 0.8048765659332275,
      "learning_rate": 9.480851240376445e-06,
      "loss": 0.7921,
      "step": 5403
    },
    {
      "epoch": 0.29742968792999064,
      "grad_norm": 0.8715546131134033,
      "learning_rate": 9.480658889908143e-06,
      "loss": 0.856,
      "step": 5404
    },
    {
      "epoch": 0.2974847267323463,
      "grad_norm": 0.7211160063743591,
      "learning_rate": 9.480466505764408e-06,
      "loss": 0.7687,
      "step": 5405
    },
    {
      "epoch": 0.29753976553470196,
      "grad_norm": 0.8749645352363586,
      "learning_rate": 9.480274087946686e-06,
      "loss": 0.8419,
      "step": 5406
    },
    {
      "epoch": 0.29759480433705765,
      "grad_norm": 0.7986398935317993,
      "learning_rate": 9.480081636456424e-06,
      "loss": 0.8309,
      "step": 5407
    },
    {
      "epoch": 0.2976498431394133,
      "grad_norm": 0.8435508012771606,
      "learning_rate": 9.479889151295067e-06,
      "loss": 0.7457,
      "step": 5408
    },
    {
      "epoch": 0.297704881941769,
      "grad_norm": 0.8725010752677917,
      "learning_rate": 9.479696632464063e-06,
      "loss": 0.8069,
      "step": 5409
    },
    {
      "epoch": 0.2977599207441246,
      "grad_norm": 0.7364320158958435,
      "learning_rate": 9.479504079964856e-06,
      "loss": 0.8316,
      "step": 5410
    },
    {
      "epoch": 0.2978149595464803,
      "grad_norm": 0.7967824935913086,
      "learning_rate": 9.479311493798898e-06,
      "loss": 0.7689,
      "step": 5411
    },
    {
      "epoch": 0.29786999834883593,
      "grad_norm": 0.8415414094924927,
      "learning_rate": 9.479118873967632e-06,
      "loss": 0.8288,
      "step": 5412
    },
    {
      "epoch": 0.29792503715119156,
      "grad_norm": 0.9723265767097473,
      "learning_rate": 9.478926220472508e-06,
      "loss": 0.7422,
      "step": 5413
    },
    {
      "epoch": 0.29798007595354725,
      "grad_norm": 0.7203155159950256,
      "learning_rate": 9.478733533314974e-06,
      "loss": 0.707,
      "step": 5414
    },
    {
      "epoch": 0.2980351147559029,
      "grad_norm": 0.7643926739692688,
      "learning_rate": 9.478540812496478e-06,
      "loss": 0.7793,
      "step": 5415
    },
    {
      "epoch": 0.2980901535582586,
      "grad_norm": 0.9177087545394897,
      "learning_rate": 9.478348058018467e-06,
      "loss": 0.865,
      "step": 5416
    },
    {
      "epoch": 0.2981451923606142,
      "grad_norm": 0.678931713104248,
      "learning_rate": 9.478155269882392e-06,
      "loss": 0.7716,
      "step": 5417
    },
    {
      "epoch": 0.2982002311629699,
      "grad_norm": 0.8440513610839844,
      "learning_rate": 9.4779624480897e-06,
      "loss": 0.8904,
      "step": 5418
    },
    {
      "epoch": 0.29825526996532553,
      "grad_norm": 0.8508756756782532,
      "learning_rate": 9.47776959264184e-06,
      "loss": 0.7994,
      "step": 5419
    },
    {
      "epoch": 0.2983103087676812,
      "grad_norm": 0.8736951947212219,
      "learning_rate": 9.477576703540265e-06,
      "loss": 0.8374,
      "step": 5420
    },
    {
      "epoch": 0.29836534757003685,
      "grad_norm": 0.8063240051269531,
      "learning_rate": 9.47738378078642e-06,
      "loss": 0.7217,
      "step": 5421
    },
    {
      "epoch": 0.29842038637239254,
      "grad_norm": 1.1495088338851929,
      "learning_rate": 9.477190824381757e-06,
      "loss": 0.8902,
      "step": 5422
    },
    {
      "epoch": 0.2984754251747482,
      "grad_norm": 1.0241554975509644,
      "learning_rate": 9.476997834327725e-06,
      "loss": 0.9354,
      "step": 5423
    },
    {
      "epoch": 0.29853046397710387,
      "grad_norm": 0.939950168132782,
      "learning_rate": 9.476804810625779e-06,
      "loss": 0.8714,
      "step": 5424
    },
    {
      "epoch": 0.2985855027794595,
      "grad_norm": 0.7592660188674927,
      "learning_rate": 9.476611753277364e-06,
      "loss": 0.7513,
      "step": 5425
    },
    {
      "epoch": 0.2986405415818152,
      "grad_norm": 0.776153028011322,
      "learning_rate": 9.476418662283935e-06,
      "loss": 0.7828,
      "step": 5426
    },
    {
      "epoch": 0.2986955803841708,
      "grad_norm": 0.9317814707756042,
      "learning_rate": 9.47622553764694e-06,
      "loss": 0.865,
      "step": 5427
    },
    {
      "epoch": 0.2987506191865265,
      "grad_norm": 0.7770501971244812,
      "learning_rate": 9.476032379367832e-06,
      "loss": 0.7281,
      "step": 5428
    },
    {
      "epoch": 0.29880565798888215,
      "grad_norm": 0.7815201282501221,
      "learning_rate": 9.475839187448064e-06,
      "loss": 0.7565,
      "step": 5429
    },
    {
      "epoch": 0.29886069679123783,
      "grad_norm": 0.7992607951164246,
      "learning_rate": 9.475645961889086e-06,
      "loss": 0.8109,
      "step": 5430
    },
    {
      "epoch": 0.29891573559359347,
      "grad_norm": 0.7780614495277405,
      "learning_rate": 9.475452702692351e-06,
      "loss": 0.7814,
      "step": 5431
    },
    {
      "epoch": 0.29897077439594916,
      "grad_norm": 0.7409062385559082,
      "learning_rate": 9.475259409859313e-06,
      "loss": 0.7712,
      "step": 5432
    },
    {
      "epoch": 0.2990258131983048,
      "grad_norm": 0.7935584187507629,
      "learning_rate": 9.47506608339142e-06,
      "loss": 0.8301,
      "step": 5433
    },
    {
      "epoch": 0.2990808520006605,
      "grad_norm": 0.6931030750274658,
      "learning_rate": 9.474872723290132e-06,
      "loss": 0.7471,
      "step": 5434
    },
    {
      "epoch": 0.2991358908030161,
      "grad_norm": 0.7622918486595154,
      "learning_rate": 9.474679329556894e-06,
      "loss": 0.7727,
      "step": 5435
    },
    {
      "epoch": 0.2991909296053718,
      "grad_norm": 0.7957701086997986,
      "learning_rate": 9.474485902193169e-06,
      "loss": 0.7663,
      "step": 5436
    },
    {
      "epoch": 0.29924596840772744,
      "grad_norm": 1.0600612163543701,
      "learning_rate": 9.474292441200404e-06,
      "loss": 0.7861,
      "step": 5437
    },
    {
      "epoch": 0.2993010072100831,
      "grad_norm": 0.7343600392341614,
      "learning_rate": 9.474098946580053e-06,
      "loss": 0.8609,
      "step": 5438
    },
    {
      "epoch": 0.29935604601243876,
      "grad_norm": 0.7477726340293884,
      "learning_rate": 9.473905418333573e-06,
      "loss": 0.7683,
      "step": 5439
    },
    {
      "epoch": 0.29941108481479445,
      "grad_norm": 0.7955546379089355,
      "learning_rate": 9.473711856462417e-06,
      "loss": 0.8406,
      "step": 5440
    },
    {
      "epoch": 0.2994661236171501,
      "grad_norm": 0.8291183114051819,
      "learning_rate": 9.47351826096804e-06,
      "loss": 0.6919,
      "step": 5441
    },
    {
      "epoch": 0.29952116241950577,
      "grad_norm": 0.8899849057197571,
      "learning_rate": 9.473324631851898e-06,
      "loss": 0.9403,
      "step": 5442
    },
    {
      "epoch": 0.2995762012218614,
      "grad_norm": 0.837066650390625,
      "learning_rate": 9.473130969115445e-06,
      "loss": 0.8676,
      "step": 5443
    },
    {
      "epoch": 0.2996312400242171,
      "grad_norm": 0.8385708928108215,
      "learning_rate": 9.472937272760138e-06,
      "loss": 0.7588,
      "step": 5444
    },
    {
      "epoch": 0.2996862788265727,
      "grad_norm": 0.6990595459938049,
      "learning_rate": 9.472743542787431e-06,
      "loss": 0.6769,
      "step": 5445
    },
    {
      "epoch": 0.2997413176289284,
      "grad_norm": 0.789165735244751,
      "learning_rate": 9.472549779198781e-06,
      "loss": 0.8084,
      "step": 5446
    },
    {
      "epoch": 0.29979635643128405,
      "grad_norm": 0.8820298314094543,
      "learning_rate": 9.472355981995643e-06,
      "loss": 0.8262,
      "step": 5447
    },
    {
      "epoch": 0.29985139523363974,
      "grad_norm": 0.8928382992744446,
      "learning_rate": 9.472162151179475e-06,
      "loss": 0.8123,
      "step": 5448
    },
    {
      "epoch": 0.2999064340359954,
      "grad_norm": 0.7688086032867432,
      "learning_rate": 9.471968286751735e-06,
      "loss": 0.6846,
      "step": 5449
    },
    {
      "epoch": 0.29996147283835106,
      "grad_norm": 0.6962918043136597,
      "learning_rate": 9.471774388713877e-06,
      "loss": 0.7872,
      "step": 5450
    },
    {
      "epoch": 0.3000165116407067,
      "grad_norm": 0.7467569708824158,
      "learning_rate": 9.47158045706736e-06,
      "loss": 0.8201,
      "step": 5451
    },
    {
      "epoch": 0.3000715504430624,
      "grad_norm": 0.7651814222335815,
      "learning_rate": 9.471386491813642e-06,
      "loss": 0.7734,
      "step": 5452
    },
    {
      "epoch": 0.300126589245418,
      "grad_norm": 0.8001144528388977,
      "learning_rate": 9.47119249295418e-06,
      "loss": 0.8266,
      "step": 5453
    },
    {
      "epoch": 0.3001816280477737,
      "grad_norm": 0.7937704920768738,
      "learning_rate": 9.47099846049043e-06,
      "loss": 0.8025,
      "step": 5454
    },
    {
      "epoch": 0.30023666685012934,
      "grad_norm": 0.7353448867797852,
      "learning_rate": 9.470804394423853e-06,
      "loss": 0.7926,
      "step": 5455
    },
    {
      "epoch": 0.300291705652485,
      "grad_norm": 0.9116304516792297,
      "learning_rate": 9.470610294755908e-06,
      "loss": 0.8295,
      "step": 5456
    },
    {
      "epoch": 0.30034674445484066,
      "grad_norm": 0.7169163823127747,
      "learning_rate": 9.470416161488053e-06,
      "loss": 0.822,
      "step": 5457
    },
    {
      "epoch": 0.3004017832571963,
      "grad_norm": 1.0421968698501587,
      "learning_rate": 9.470221994621747e-06,
      "loss": 0.9273,
      "step": 5458
    },
    {
      "epoch": 0.300456822059552,
      "grad_norm": 0.9064405560493469,
      "learning_rate": 9.470027794158447e-06,
      "loss": 0.7087,
      "step": 5459
    },
    {
      "epoch": 0.3005118608619076,
      "grad_norm": 0.6766010522842407,
      "learning_rate": 9.469833560099617e-06,
      "loss": 0.7063,
      "step": 5460
    },
    {
      "epoch": 0.3005668996642633,
      "grad_norm": 0.7987816333770752,
      "learning_rate": 9.469639292446712e-06,
      "loss": 0.8216,
      "step": 5461
    },
    {
      "epoch": 0.30062193846661894,
      "grad_norm": 0.776792049407959,
      "learning_rate": 9.469444991201197e-06,
      "loss": 0.8598,
      "step": 5462
    },
    {
      "epoch": 0.30067697726897463,
      "grad_norm": 0.8048756718635559,
      "learning_rate": 9.469250656364529e-06,
      "loss": 0.8645,
      "step": 5463
    },
    {
      "epoch": 0.30073201607133027,
      "grad_norm": 1.0650218725204468,
      "learning_rate": 9.46905628793817e-06,
      "loss": 0.8918,
      "step": 5464
    },
    {
      "epoch": 0.30078705487368596,
      "grad_norm": 0.7378712296485901,
      "learning_rate": 9.468861885923577e-06,
      "loss": 0.6866,
      "step": 5465
    },
    {
      "epoch": 0.3008420936760416,
      "grad_norm": 0.7382808327674866,
      "learning_rate": 9.468667450322218e-06,
      "loss": 0.8413,
      "step": 5466
    },
    {
      "epoch": 0.3008971324783973,
      "grad_norm": 0.8390250205993652,
      "learning_rate": 9.468472981135548e-06,
      "loss": 0.8275,
      "step": 5467
    },
    {
      "epoch": 0.3009521712807529,
      "grad_norm": 0.9169766902923584,
      "learning_rate": 9.468278478365034e-06,
      "loss": 0.8274,
      "step": 5468
    },
    {
      "epoch": 0.3010072100831086,
      "grad_norm": 0.7487995028495789,
      "learning_rate": 9.468083942012134e-06,
      "loss": 0.7729,
      "step": 5469
    },
    {
      "epoch": 0.30106224888546423,
      "grad_norm": 0.7457556128501892,
      "learning_rate": 9.467889372078309e-06,
      "loss": 0.7435,
      "step": 5470
    },
    {
      "epoch": 0.3011172876878199,
      "grad_norm": 0.7085639834403992,
      "learning_rate": 9.467694768565026e-06,
      "loss": 0.7686,
      "step": 5471
    },
    {
      "epoch": 0.30117232649017556,
      "grad_norm": 0.7396196722984314,
      "learning_rate": 9.467500131473744e-06,
      "loss": 0.7496,
      "step": 5472
    },
    {
      "epoch": 0.30122736529253125,
      "grad_norm": 0.7906790971755981,
      "learning_rate": 9.467305460805927e-06,
      "loss": 0.8341,
      "step": 5473
    },
    {
      "epoch": 0.3012824040948869,
      "grad_norm": 0.673541247844696,
      "learning_rate": 9.467110756563039e-06,
      "loss": 0.8041,
      "step": 5474
    },
    {
      "epoch": 0.30133744289724257,
      "grad_norm": 0.8247049450874329,
      "learning_rate": 9.46691601874654e-06,
      "loss": 0.8227,
      "step": 5475
    },
    {
      "epoch": 0.3013924816995982,
      "grad_norm": 0.7564057111740112,
      "learning_rate": 9.466721247357898e-06,
      "loss": 0.8181,
      "step": 5476
    },
    {
      "epoch": 0.3014475205019539,
      "grad_norm": 0.7533192038536072,
      "learning_rate": 9.466526442398574e-06,
      "loss": 0.782,
      "step": 5477
    },
    {
      "epoch": 0.3015025593043095,
      "grad_norm": 0.6934120059013367,
      "learning_rate": 9.466331603870033e-06,
      "loss": 0.7153,
      "step": 5478
    },
    {
      "epoch": 0.3015575981066652,
      "grad_norm": 0.7417232990264893,
      "learning_rate": 9.466136731773738e-06,
      "loss": 0.753,
      "step": 5479
    },
    {
      "epoch": 0.30161263690902085,
      "grad_norm": 0.7421486973762512,
      "learning_rate": 9.465941826111156e-06,
      "loss": 0.7668,
      "step": 5480
    },
    {
      "epoch": 0.30166767571137654,
      "grad_norm": 1.0851647853851318,
      "learning_rate": 9.465746886883751e-06,
      "loss": 0.8019,
      "step": 5481
    },
    {
      "epoch": 0.30172271451373217,
      "grad_norm": 0.9209244847297668,
      "learning_rate": 9.465551914092987e-06,
      "loss": 0.7912,
      "step": 5482
    },
    {
      "epoch": 0.30177775331608786,
      "grad_norm": 0.6915135383605957,
      "learning_rate": 9.465356907740331e-06,
      "loss": 0.8112,
      "step": 5483
    },
    {
      "epoch": 0.3018327921184435,
      "grad_norm": 0.824593722820282,
      "learning_rate": 9.465161867827247e-06,
      "loss": 0.7969,
      "step": 5484
    },
    {
      "epoch": 0.3018878309207992,
      "grad_norm": 0.7985100746154785,
      "learning_rate": 9.464966794355201e-06,
      "loss": 0.8258,
      "step": 5485
    },
    {
      "epoch": 0.3019428697231548,
      "grad_norm": 0.8471764326095581,
      "learning_rate": 9.464771687325663e-06,
      "loss": 0.8241,
      "step": 5486
    },
    {
      "epoch": 0.3019979085255105,
      "grad_norm": 0.8133455514907837,
      "learning_rate": 9.464576546740093e-06,
      "loss": 0.7809,
      "step": 5487
    },
    {
      "epoch": 0.30205294732786614,
      "grad_norm": 0.7684013843536377,
      "learning_rate": 9.464381372599961e-06,
      "loss": 0.9023,
      "step": 5488
    },
    {
      "epoch": 0.30210798613022183,
      "grad_norm": 0.7818747758865356,
      "learning_rate": 9.464186164906735e-06,
      "loss": 0.7152,
      "step": 5489
    },
    {
      "epoch": 0.30216302493257746,
      "grad_norm": 0.7524297833442688,
      "learning_rate": 9.46399092366188e-06,
      "loss": 0.782,
      "step": 5490
    },
    {
      "epoch": 0.30221806373493315,
      "grad_norm": 0.6550590991973877,
      "learning_rate": 9.463795648866864e-06,
      "loss": 0.7696,
      "step": 5491
    },
    {
      "epoch": 0.3022731025372888,
      "grad_norm": 0.8679335117340088,
      "learning_rate": 9.463600340523154e-06,
      "loss": 0.8115,
      "step": 5492
    },
    {
      "epoch": 0.3023281413396445,
      "grad_norm": 0.692500114440918,
      "learning_rate": 9.46340499863222e-06,
      "loss": 0.7692,
      "step": 5493
    },
    {
      "epoch": 0.3023831801420001,
      "grad_norm": 0.8604017496109009,
      "learning_rate": 9.463209623195528e-06,
      "loss": 0.8547,
      "step": 5494
    },
    {
      "epoch": 0.3024382189443558,
      "grad_norm": 0.6715821623802185,
      "learning_rate": 9.463014214214548e-06,
      "loss": 0.7638,
      "step": 5495
    },
    {
      "epoch": 0.30249325774671143,
      "grad_norm": 0.7803179025650024,
      "learning_rate": 9.462818771690747e-06,
      "loss": 0.7795,
      "step": 5496
    },
    {
      "epoch": 0.3025482965490671,
      "grad_norm": 0.787323534488678,
      "learning_rate": 9.462623295625596e-06,
      "loss": 0.735,
      "step": 5497
    },
    {
      "epoch": 0.30260333535142275,
      "grad_norm": 0.9943159222602844,
      "learning_rate": 9.462427786020563e-06,
      "loss": 0.7451,
      "step": 5498
    },
    {
      "epoch": 0.3026583741537784,
      "grad_norm": 0.772524893283844,
      "learning_rate": 9.462232242877116e-06,
      "loss": 0.9167,
      "step": 5499
    },
    {
      "epoch": 0.3027134129561341,
      "grad_norm": 0.7204643487930298,
      "learning_rate": 9.462036666196726e-06,
      "loss": 0.7442,
      "step": 5500
    },
    {
      "epoch": 0.3027684517584897,
      "grad_norm": 0.7450547218322754,
      "learning_rate": 9.461841055980863e-06,
      "loss": 0.8002,
      "step": 5501
    },
    {
      "epoch": 0.3028234905608454,
      "grad_norm": 0.8096264004707336,
      "learning_rate": 9.461645412230997e-06,
      "loss": 0.8601,
      "step": 5502
    },
    {
      "epoch": 0.30287852936320103,
      "grad_norm": 0.684968888759613,
      "learning_rate": 9.461449734948597e-06,
      "loss": 0.7251,
      "step": 5503
    },
    {
      "epoch": 0.3029335681655567,
      "grad_norm": 0.7727203369140625,
      "learning_rate": 9.461254024135138e-06,
      "loss": 0.7797,
      "step": 5504
    },
    {
      "epoch": 0.30298860696791236,
      "grad_norm": 0.9292891025543213,
      "learning_rate": 9.461058279792086e-06,
      "loss": 0.7519,
      "step": 5505
    },
    {
      "epoch": 0.30304364577026804,
      "grad_norm": 0.7836466431617737,
      "learning_rate": 9.460862501920915e-06,
      "loss": 0.8201,
      "step": 5506
    },
    {
      "epoch": 0.3030986845726237,
      "grad_norm": 0.9043576121330261,
      "learning_rate": 9.460666690523094e-06,
      "loss": 0.79,
      "step": 5507
    },
    {
      "epoch": 0.30315372337497937,
      "grad_norm": 0.8339952230453491,
      "learning_rate": 9.460470845600098e-06,
      "loss": 0.8392,
      "step": 5508
    },
    {
      "epoch": 0.303208762177335,
      "grad_norm": 0.7603133320808411,
      "learning_rate": 9.460274967153395e-06,
      "loss": 0.7168,
      "step": 5509
    },
    {
      "epoch": 0.3032638009796907,
      "grad_norm": 0.7287996411323547,
      "learning_rate": 9.460079055184461e-06,
      "loss": 0.7452,
      "step": 5510
    },
    {
      "epoch": 0.3033188397820463,
      "grad_norm": 0.707953691482544,
      "learning_rate": 9.459883109694767e-06,
      "loss": 0.8081,
      "step": 5511
    },
    {
      "epoch": 0.303373878584402,
      "grad_norm": 0.7556451559066772,
      "learning_rate": 9.459687130685784e-06,
      "loss": 0.8145,
      "step": 5512
    },
    {
      "epoch": 0.30342891738675765,
      "grad_norm": 0.8076426386833191,
      "learning_rate": 9.459491118158987e-06,
      "loss": 0.8006,
      "step": 5513
    },
    {
      "epoch": 0.30348395618911334,
      "grad_norm": 0.7343682646751404,
      "learning_rate": 9.459295072115849e-06,
      "loss": 0.7574,
      "step": 5514
    },
    {
      "epoch": 0.30353899499146897,
      "grad_norm": 0.68440181016922,
      "learning_rate": 9.459098992557843e-06,
      "loss": 0.7432,
      "step": 5515
    },
    {
      "epoch": 0.30359403379382466,
      "grad_norm": 0.8278071880340576,
      "learning_rate": 9.458902879486441e-06,
      "loss": 0.8357,
      "step": 5516
    },
    {
      "epoch": 0.3036490725961803,
      "grad_norm": 0.8377245664596558,
      "learning_rate": 9.458706732903121e-06,
      "loss": 0.7552,
      "step": 5517
    },
    {
      "epoch": 0.303704111398536,
      "grad_norm": 0.7354543805122375,
      "learning_rate": 9.458510552809353e-06,
      "loss": 0.7862,
      "step": 5518
    },
    {
      "epoch": 0.3037591502008916,
      "grad_norm": 0.8071799874305725,
      "learning_rate": 9.458314339206611e-06,
      "loss": 0.8428,
      "step": 5519
    },
    {
      "epoch": 0.3038141890032473,
      "grad_norm": 0.7452389597892761,
      "learning_rate": 9.458118092096376e-06,
      "loss": 0.8252,
      "step": 5520
    },
    {
      "epoch": 0.30386922780560294,
      "grad_norm": 0.7370620965957642,
      "learning_rate": 9.457921811480115e-06,
      "loss": 0.8143,
      "step": 5521
    },
    {
      "epoch": 0.3039242666079586,
      "grad_norm": 0.8816156387329102,
      "learning_rate": 9.45772549735931e-06,
      "loss": 0.7163,
      "step": 5522
    },
    {
      "epoch": 0.30397930541031426,
      "grad_norm": 0.7208901643753052,
      "learning_rate": 9.457529149735432e-06,
      "loss": 0.7877,
      "step": 5523
    },
    {
      "epoch": 0.30403434421266995,
      "grad_norm": 0.820792019367218,
      "learning_rate": 9.457332768609959e-06,
      "loss": 0.8275,
      "step": 5524
    },
    {
      "epoch": 0.3040893830150256,
      "grad_norm": 0.8471686244010925,
      "learning_rate": 9.457136353984365e-06,
      "loss": 0.8127,
      "step": 5525
    },
    {
      "epoch": 0.30414442181738127,
      "grad_norm": 0.9448342323303223,
      "learning_rate": 9.456939905860127e-06,
      "loss": 0.8157,
      "step": 5526
    },
    {
      "epoch": 0.3041994606197369,
      "grad_norm": 0.7835188508033752,
      "learning_rate": 9.456743424238723e-06,
      "loss": 0.7116,
      "step": 5527
    },
    {
      "epoch": 0.3042544994220926,
      "grad_norm": 0.8884950876235962,
      "learning_rate": 9.456546909121629e-06,
      "loss": 0.8514,
      "step": 5528
    },
    {
      "epoch": 0.30430953822444823,
      "grad_norm": 0.7400928735733032,
      "learning_rate": 9.45635036051032e-06,
      "loss": 0.8207,
      "step": 5529
    },
    {
      "epoch": 0.3043645770268039,
      "grad_norm": 0.8278732299804688,
      "learning_rate": 9.456153778406274e-06,
      "loss": 0.8269,
      "step": 5530
    },
    {
      "epoch": 0.30441961582915955,
      "grad_norm": 0.7423332929611206,
      "learning_rate": 9.45595716281097e-06,
      "loss": 0.7937,
      "step": 5531
    },
    {
      "epoch": 0.30447465463151524,
      "grad_norm": 1.5018088817596436,
      "learning_rate": 9.455760513725885e-06,
      "loss": 0.7935,
      "step": 5532
    },
    {
      "epoch": 0.3045296934338709,
      "grad_norm": 0.8105388283729553,
      "learning_rate": 9.455563831152496e-06,
      "loss": 0.8225,
      "step": 5533
    },
    {
      "epoch": 0.30458473223622656,
      "grad_norm": 0.6874535083770752,
      "learning_rate": 9.455367115092283e-06,
      "loss": 0.7301,
      "step": 5534
    },
    {
      "epoch": 0.3046397710385822,
      "grad_norm": 0.8085837960243225,
      "learning_rate": 9.455170365546721e-06,
      "loss": 0.83,
      "step": 5535
    },
    {
      "epoch": 0.3046948098409379,
      "grad_norm": 0.810773491859436,
      "learning_rate": 9.454973582517293e-06,
      "loss": 0.7186,
      "step": 5536
    },
    {
      "epoch": 0.3047498486432935,
      "grad_norm": 0.7290367484092712,
      "learning_rate": 9.454776766005476e-06,
      "loss": 0.8181,
      "step": 5537
    },
    {
      "epoch": 0.3048048874456492,
      "grad_norm": 0.773728609085083,
      "learning_rate": 9.45457991601275e-06,
      "loss": 0.8454,
      "step": 5538
    },
    {
      "epoch": 0.30485992624800484,
      "grad_norm": 0.792169451713562,
      "learning_rate": 9.454383032540592e-06,
      "loss": 0.8797,
      "step": 5539
    },
    {
      "epoch": 0.30491496505036053,
      "grad_norm": 0.7478733658790588,
      "learning_rate": 9.454186115590485e-06,
      "loss": 0.7544,
      "step": 5540
    },
    {
      "epoch": 0.30497000385271616,
      "grad_norm": 0.8527306318283081,
      "learning_rate": 9.453989165163906e-06,
      "loss": 0.8379,
      "step": 5541
    },
    {
      "epoch": 0.3050250426550718,
      "grad_norm": 0.8829329013824463,
      "learning_rate": 9.453792181262337e-06,
      "loss": 0.7643,
      "step": 5542
    },
    {
      "epoch": 0.3050800814574275,
      "grad_norm": 0.9477338790893555,
      "learning_rate": 9.453595163887258e-06,
      "loss": 0.7414,
      "step": 5543
    },
    {
      "epoch": 0.3051351202597831,
      "grad_norm": 0.8311536312103271,
      "learning_rate": 9.453398113040151e-06,
      "loss": 0.8133,
      "step": 5544
    },
    {
      "epoch": 0.3051901590621388,
      "grad_norm": 0.8035525679588318,
      "learning_rate": 9.453201028722497e-06,
      "loss": 0.7841,
      "step": 5545
    },
    {
      "epoch": 0.30524519786449444,
      "grad_norm": 0.7779183983802795,
      "learning_rate": 9.453003910935775e-06,
      "loss": 0.7696,
      "step": 5546
    },
    {
      "epoch": 0.30530023666685013,
      "grad_norm": 0.7843946218490601,
      "learning_rate": 9.452806759681465e-06,
      "loss": 0.6018,
      "step": 5547
    },
    {
      "epoch": 0.30535527546920577,
      "grad_norm": 0.7215032577514648,
      "learning_rate": 9.452609574961053e-06,
      "loss": 0.7457,
      "step": 5548
    },
    {
      "epoch": 0.30541031427156146,
      "grad_norm": 0.9628198742866516,
      "learning_rate": 9.452412356776021e-06,
      "loss": 0.8061,
      "step": 5549
    },
    {
      "epoch": 0.3054653530739171,
      "grad_norm": 0.9468308687210083,
      "learning_rate": 9.452215105127848e-06,
      "loss": 0.7909,
      "step": 5550
    },
    {
      "epoch": 0.3055203918762728,
      "grad_norm": 0.876402735710144,
      "learning_rate": 9.452017820018017e-06,
      "loss": 0.69,
      "step": 5551
    },
    {
      "epoch": 0.3055754306786284,
      "grad_norm": 1.03409743309021,
      "learning_rate": 9.451820501448014e-06,
      "loss": 0.8375,
      "step": 5552
    },
    {
      "epoch": 0.3056304694809841,
      "grad_norm": 0.8057541847229004,
      "learning_rate": 9.45162314941932e-06,
      "loss": 0.7704,
      "step": 5553
    },
    {
      "epoch": 0.30568550828333974,
      "grad_norm": 0.7256304025650024,
      "learning_rate": 9.451425763933417e-06,
      "loss": 0.7819,
      "step": 5554
    },
    {
      "epoch": 0.3057405470856954,
      "grad_norm": 0.7982180118560791,
      "learning_rate": 9.451228344991788e-06,
      "loss": 0.8094,
      "step": 5555
    },
    {
      "epoch": 0.30579558588805106,
      "grad_norm": 1.0314620733261108,
      "learning_rate": 9.45103089259592e-06,
      "loss": 0.7777,
      "step": 5556
    },
    {
      "epoch": 0.30585062469040675,
      "grad_norm": 0.6948755383491516,
      "learning_rate": 9.450833406747294e-06,
      "loss": 0.7189,
      "step": 5557
    },
    {
      "epoch": 0.3059056634927624,
      "grad_norm": 0.7412117719650269,
      "learning_rate": 9.450635887447396e-06,
      "loss": 0.783,
      "step": 5558
    },
    {
      "epoch": 0.30596070229511807,
      "grad_norm": 0.7394647002220154,
      "learning_rate": 9.450438334697711e-06,
      "loss": 0.7888,
      "step": 5559
    },
    {
      "epoch": 0.3060157410974737,
      "grad_norm": 0.692701518535614,
      "learning_rate": 9.450240748499725e-06,
      "loss": 0.7427,
      "step": 5560
    },
    {
      "epoch": 0.3060707798998294,
      "grad_norm": 0.6854925751686096,
      "learning_rate": 9.450043128854916e-06,
      "loss": 0.7877,
      "step": 5561
    },
    {
      "epoch": 0.306125818702185,
      "grad_norm": 0.8073517680168152,
      "learning_rate": 9.449845475764776e-06,
      "loss": 0.8715,
      "step": 5562
    },
    {
      "epoch": 0.3061808575045407,
      "grad_norm": 0.9672908186912537,
      "learning_rate": 9.449647789230789e-06,
      "loss": 0.782,
      "step": 5563
    },
    {
      "epoch": 0.30623589630689635,
      "grad_norm": 0.7409735918045044,
      "learning_rate": 9.44945006925444e-06,
      "loss": 0.7956,
      "step": 5564
    },
    {
      "epoch": 0.30629093510925204,
      "grad_norm": 0.7839213609695435,
      "learning_rate": 9.449252315837215e-06,
      "loss": 0.7559,
      "step": 5565
    },
    {
      "epoch": 0.30634597391160767,
      "grad_norm": 0.668393075466156,
      "learning_rate": 9.449054528980602e-06,
      "loss": 0.717,
      "step": 5566
    },
    {
      "epoch": 0.30640101271396336,
      "grad_norm": 0.8818438053131104,
      "learning_rate": 9.448856708686084e-06,
      "loss": 0.7801,
      "step": 5567
    },
    {
      "epoch": 0.306456051516319,
      "grad_norm": 0.7331361770629883,
      "learning_rate": 9.44865885495515e-06,
      "loss": 0.6999,
      "step": 5568
    },
    {
      "epoch": 0.3065110903186747,
      "grad_norm": 0.7818138599395752,
      "learning_rate": 9.448460967789288e-06,
      "loss": 0.7437,
      "step": 5569
    },
    {
      "epoch": 0.3065661291210303,
      "grad_norm": 0.7713417410850525,
      "learning_rate": 9.448263047189985e-06,
      "loss": 0.8523,
      "step": 5570
    },
    {
      "epoch": 0.306621167923386,
      "grad_norm": 0.7152866125106812,
      "learning_rate": 9.448065093158726e-06,
      "loss": 0.7706,
      "step": 5571
    },
    {
      "epoch": 0.30667620672574164,
      "grad_norm": 0.7486638426780701,
      "learning_rate": 9.447867105697e-06,
      "loss": 0.7738,
      "step": 5572
    },
    {
      "epoch": 0.30673124552809733,
      "grad_norm": 0.7014918923377991,
      "learning_rate": 9.447669084806297e-06,
      "loss": 0.7013,
      "step": 5573
    },
    {
      "epoch": 0.30678628433045296,
      "grad_norm": 0.8328303694725037,
      "learning_rate": 9.447471030488102e-06,
      "loss": 0.8113,
      "step": 5574
    },
    {
      "epoch": 0.30684132313280865,
      "grad_norm": 0.6800024509429932,
      "learning_rate": 9.447272942743906e-06,
      "loss": 0.6786,
      "step": 5575
    },
    {
      "epoch": 0.3068963619351643,
      "grad_norm": 0.6827595829963684,
      "learning_rate": 9.447074821575198e-06,
      "loss": 0.812,
      "step": 5576
    },
    {
      "epoch": 0.30695140073752,
      "grad_norm": 0.8775614500045776,
      "learning_rate": 9.446876666983465e-06,
      "loss": 0.7683,
      "step": 5577
    },
    {
      "epoch": 0.3070064395398756,
      "grad_norm": 0.7440332174301147,
      "learning_rate": 9.446678478970198e-06,
      "loss": 0.7152,
      "step": 5578
    },
    {
      "epoch": 0.3070614783422313,
      "grad_norm": 0.7031408548355103,
      "learning_rate": 9.446480257536885e-06,
      "loss": 0.7603,
      "step": 5579
    },
    {
      "epoch": 0.30711651714458693,
      "grad_norm": 0.8419817090034485,
      "learning_rate": 9.446282002685019e-06,
      "loss": 0.9939,
      "step": 5580
    },
    {
      "epoch": 0.3071715559469426,
      "grad_norm": 0.7622908353805542,
      "learning_rate": 9.446083714416085e-06,
      "loss": 0.8682,
      "step": 5581
    },
    {
      "epoch": 0.30722659474929825,
      "grad_norm": 0.7341362833976746,
      "learning_rate": 9.445885392731576e-06,
      "loss": 0.848,
      "step": 5582
    },
    {
      "epoch": 0.30728163355165394,
      "grad_norm": 0.7248286604881287,
      "learning_rate": 9.445687037632984e-06,
      "loss": 0.7699,
      "step": 5583
    },
    {
      "epoch": 0.3073366723540096,
      "grad_norm": 0.9409947991371155,
      "learning_rate": 9.445488649121797e-06,
      "loss": 1.0051,
      "step": 5584
    },
    {
      "epoch": 0.3073917111563652,
      "grad_norm": 0.7279968857765198,
      "learning_rate": 9.445290227199509e-06,
      "loss": 0.8001,
      "step": 5585
    },
    {
      "epoch": 0.3074467499587209,
      "grad_norm": 0.7904797196388245,
      "learning_rate": 9.445091771867607e-06,
      "loss": 0.8892,
      "step": 5586
    },
    {
      "epoch": 0.30750178876107653,
      "grad_norm": 0.7090430855751038,
      "learning_rate": 9.444893283127587e-06,
      "loss": 0.5983,
      "step": 5587
    },
    {
      "epoch": 0.3075568275634322,
      "grad_norm": 0.8363901376724243,
      "learning_rate": 9.444694760980939e-06,
      "loss": 0.7688,
      "step": 5588
    },
    {
      "epoch": 0.30761186636578786,
      "grad_norm": 0.7487169504165649,
      "learning_rate": 9.444496205429152e-06,
      "loss": 0.7585,
      "step": 5589
    },
    {
      "epoch": 0.30766690516814355,
      "grad_norm": 0.750801146030426,
      "learning_rate": 9.444297616473724e-06,
      "loss": 0.6493,
      "step": 5590
    },
    {
      "epoch": 0.3077219439704992,
      "grad_norm": 0.754846453666687,
      "learning_rate": 9.444098994116144e-06,
      "loss": 0.8528,
      "step": 5591
    },
    {
      "epoch": 0.30777698277285487,
      "grad_norm": 0.7088152766227722,
      "learning_rate": 9.443900338357907e-06,
      "loss": 0.7927,
      "step": 5592
    },
    {
      "epoch": 0.3078320215752105,
      "grad_norm": 0.7077113389968872,
      "learning_rate": 9.443701649200503e-06,
      "loss": 0.7996,
      "step": 5593
    },
    {
      "epoch": 0.3078870603775662,
      "grad_norm": 0.732982873916626,
      "learning_rate": 9.443502926645427e-06,
      "loss": 0.7473,
      "step": 5594
    },
    {
      "epoch": 0.3079420991799218,
      "grad_norm": 0.7068434357643127,
      "learning_rate": 9.443304170694174e-06,
      "loss": 0.7575,
      "step": 5595
    },
    {
      "epoch": 0.3079971379822775,
      "grad_norm": 0.7703887224197388,
      "learning_rate": 9.443105381348234e-06,
      "loss": 0.8157,
      "step": 5596
    },
    {
      "epoch": 0.30805217678463315,
      "grad_norm": 0.806924045085907,
      "learning_rate": 9.442906558609103e-06,
      "loss": 0.7572,
      "step": 5597
    },
    {
      "epoch": 0.30810721558698884,
      "grad_norm": 0.8364617824554443,
      "learning_rate": 9.442707702478278e-06,
      "loss": 0.7491,
      "step": 5598
    },
    {
      "epoch": 0.30816225438934447,
      "grad_norm": 0.9269624352455139,
      "learning_rate": 9.442508812957249e-06,
      "loss": 0.8746,
      "step": 5599
    },
    {
      "epoch": 0.30821729319170016,
      "grad_norm": 0.7308455109596252,
      "learning_rate": 9.442309890047515e-06,
      "loss": 0.8068,
      "step": 5600
    },
    {
      "epoch": 0.3082723319940558,
      "grad_norm": 0.812622606754303,
      "learning_rate": 9.442110933750567e-06,
      "loss": 0.9137,
      "step": 5601
    },
    {
      "epoch": 0.3083273707964115,
      "grad_norm": 0.7100754976272583,
      "learning_rate": 9.441911944067905e-06,
      "loss": 0.7471,
      "step": 5602
    },
    {
      "epoch": 0.3083824095987671,
      "grad_norm": 0.760208010673523,
      "learning_rate": 9.44171292100102e-06,
      "loss": 0.8243,
      "step": 5603
    },
    {
      "epoch": 0.3084374484011228,
      "grad_norm": 0.6931812763214111,
      "learning_rate": 9.44151386455141e-06,
      "loss": 0.7523,
      "step": 5604
    },
    {
      "epoch": 0.30849248720347844,
      "grad_norm": 0.6584734916687012,
      "learning_rate": 9.44131477472057e-06,
      "loss": 0.6929,
      "step": 5605
    },
    {
      "epoch": 0.3085475260058341,
      "grad_norm": 0.977661669254303,
      "learning_rate": 9.441115651509997e-06,
      "loss": 0.8003,
      "step": 5606
    },
    {
      "epoch": 0.30860256480818976,
      "grad_norm": 0.650434672832489,
      "learning_rate": 9.440916494921189e-06,
      "loss": 0.6629,
      "step": 5607
    },
    {
      "epoch": 0.30865760361054545,
      "grad_norm": 0.6804447770118713,
      "learning_rate": 9.44071730495564e-06,
      "loss": 0.7216,
      "step": 5608
    },
    {
      "epoch": 0.3087126424129011,
      "grad_norm": 0.7942929267883301,
      "learning_rate": 9.44051808161485e-06,
      "loss": 0.7593,
      "step": 5609
    },
    {
      "epoch": 0.3087676812152568,
      "grad_norm": 0.7069621086120605,
      "learning_rate": 9.440318824900313e-06,
      "loss": 0.7453,
      "step": 5610
    },
    {
      "epoch": 0.3088227200176124,
      "grad_norm": 0.7903168797492981,
      "learning_rate": 9.440119534813528e-06,
      "loss": 0.8084,
      "step": 5611
    },
    {
      "epoch": 0.3088777588199681,
      "grad_norm": 0.7828298807144165,
      "learning_rate": 9.439920211355993e-06,
      "loss": 0.7556,
      "step": 5612
    },
    {
      "epoch": 0.30893279762232373,
      "grad_norm": 0.8118648529052734,
      "learning_rate": 9.43972085452921e-06,
      "loss": 0.8548,
      "step": 5613
    },
    {
      "epoch": 0.3089878364246794,
      "grad_norm": 0.9169642329216003,
      "learning_rate": 9.439521464334669e-06,
      "loss": 0.833,
      "step": 5614
    },
    {
      "epoch": 0.30904287522703505,
      "grad_norm": 0.7844422459602356,
      "learning_rate": 9.439322040773875e-06,
      "loss": 0.8363,
      "step": 5615
    },
    {
      "epoch": 0.30909791402939074,
      "grad_norm": 1.4801305532455444,
      "learning_rate": 9.439122583848324e-06,
      "loss": 0.7617,
      "step": 5616
    },
    {
      "epoch": 0.3091529528317464,
      "grad_norm": 0.7737647891044617,
      "learning_rate": 9.438923093559517e-06,
      "loss": 0.7224,
      "step": 5617
    },
    {
      "epoch": 0.30920799163410206,
      "grad_norm": 0.7279127836227417,
      "learning_rate": 9.438723569908952e-06,
      "loss": 0.7783,
      "step": 5618
    },
    {
      "epoch": 0.3092630304364577,
      "grad_norm": 0.7635996341705322,
      "learning_rate": 9.438524012898127e-06,
      "loss": 0.8408,
      "step": 5619
    },
    {
      "epoch": 0.3093180692388134,
      "grad_norm": 0.818445086479187,
      "learning_rate": 9.438324422528547e-06,
      "loss": 0.8836,
      "step": 5620
    },
    {
      "epoch": 0.309373108041169,
      "grad_norm": 0.8620640635490417,
      "learning_rate": 9.438124798801706e-06,
      "loss": 0.925,
      "step": 5621
    },
    {
      "epoch": 0.3094281468435247,
      "grad_norm": 0.7294883728027344,
      "learning_rate": 9.437925141719108e-06,
      "loss": 0.8387,
      "step": 5622
    },
    {
      "epoch": 0.30948318564588034,
      "grad_norm": 0.6696046590805054,
      "learning_rate": 9.437725451282252e-06,
      "loss": 0.6712,
      "step": 5623
    },
    {
      "epoch": 0.30953822444823603,
      "grad_norm": 0.8200504779815674,
      "learning_rate": 9.43752572749264e-06,
      "loss": 0.8191,
      "step": 5624
    },
    {
      "epoch": 0.30959326325059167,
      "grad_norm": 0.8440756797790527,
      "learning_rate": 9.437325970351773e-06,
      "loss": 0.7412,
      "step": 5625
    },
    {
      "epoch": 0.30964830205294736,
      "grad_norm": 0.8550771474838257,
      "learning_rate": 9.43712617986115e-06,
      "loss": 0.7842,
      "step": 5626
    },
    {
      "epoch": 0.309703340855303,
      "grad_norm": 0.8203451037406921,
      "learning_rate": 9.436926356022275e-06,
      "loss": 0.8298,
      "step": 5627
    },
    {
      "epoch": 0.3097583796576586,
      "grad_norm": 1.0105336904525757,
      "learning_rate": 9.436726498836651e-06,
      "loss": 0.8416,
      "step": 5628
    },
    {
      "epoch": 0.3098134184600143,
      "grad_norm": 0.7684324383735657,
      "learning_rate": 9.436526608305777e-06,
      "loss": 0.7051,
      "step": 5629
    },
    {
      "epoch": 0.30986845726236995,
      "grad_norm": 0.7284610867500305,
      "learning_rate": 9.436326684431157e-06,
      "loss": 0.755,
      "step": 5630
    },
    {
      "epoch": 0.30992349606472563,
      "grad_norm": 0.7125874161720276,
      "learning_rate": 9.436126727214293e-06,
      "loss": 0.7336,
      "step": 5631
    },
    {
      "epoch": 0.30997853486708127,
      "grad_norm": 0.7008525729179382,
      "learning_rate": 9.435926736656687e-06,
      "loss": 0.7185,
      "step": 5632
    },
    {
      "epoch": 0.31003357366943696,
      "grad_norm": 0.7087175250053406,
      "learning_rate": 9.435726712759844e-06,
      "loss": 0.717,
      "step": 5633
    },
    {
      "epoch": 0.3100886124717926,
      "grad_norm": 0.7892497777938843,
      "learning_rate": 9.435526655525267e-06,
      "loss": 0.8308,
      "step": 5634
    },
    {
      "epoch": 0.3101436512741483,
      "grad_norm": 0.733906626701355,
      "learning_rate": 9.435326564954457e-06,
      "loss": 0.7421,
      "step": 5635
    },
    {
      "epoch": 0.3101986900765039,
      "grad_norm": 0.7874915599822998,
      "learning_rate": 9.43512644104892e-06,
      "loss": 0.8808,
      "step": 5636
    },
    {
      "epoch": 0.3102537288788596,
      "grad_norm": 0.6849297881126404,
      "learning_rate": 9.434926283810162e-06,
      "loss": 0.7297,
      "step": 5637
    },
    {
      "epoch": 0.31030876768121524,
      "grad_norm": 0.7847834825515747,
      "learning_rate": 9.434726093239685e-06,
      "loss": 0.7873,
      "step": 5638
    },
    {
      "epoch": 0.3103638064835709,
      "grad_norm": 0.6999106407165527,
      "learning_rate": 9.434525869338992e-06,
      "loss": 0.7699,
      "step": 5639
    },
    {
      "epoch": 0.31041884528592656,
      "grad_norm": 0.7662788033485413,
      "learning_rate": 9.43432561210959e-06,
      "loss": 0.7583,
      "step": 5640
    },
    {
      "epoch": 0.31047388408828225,
      "grad_norm": 0.8336607217788696,
      "learning_rate": 9.434125321552985e-06,
      "loss": 0.7297,
      "step": 5641
    },
    {
      "epoch": 0.3105289228906379,
      "grad_norm": 0.8038349151611328,
      "learning_rate": 9.433924997670681e-06,
      "loss": 0.798,
      "step": 5642
    },
    {
      "epoch": 0.31058396169299357,
      "grad_norm": 0.6819794178009033,
      "learning_rate": 9.433724640464181e-06,
      "loss": 0.7951,
      "step": 5643
    },
    {
      "epoch": 0.3106390004953492,
      "grad_norm": 0.916238009929657,
      "learning_rate": 9.433524249934995e-06,
      "loss": 0.7371,
      "step": 5644
    },
    {
      "epoch": 0.3106940392977049,
      "grad_norm": 0.8390263915061951,
      "learning_rate": 9.433323826084628e-06,
      "loss": 0.8211,
      "step": 5645
    },
    {
      "epoch": 0.3107490781000605,
      "grad_norm": 0.7957239747047424,
      "learning_rate": 9.433123368914586e-06,
      "loss": 0.8406,
      "step": 5646
    },
    {
      "epoch": 0.3108041169024162,
      "grad_norm": 0.6771933436393738,
      "learning_rate": 9.432922878426374e-06,
      "loss": 0.7664,
      "step": 5647
    },
    {
      "epoch": 0.31085915570477185,
      "grad_norm": 0.7874065041542053,
      "learning_rate": 9.432722354621503e-06,
      "loss": 0.7445,
      "step": 5648
    },
    {
      "epoch": 0.31091419450712754,
      "grad_norm": 0.674749493598938,
      "learning_rate": 9.432521797501475e-06,
      "loss": 0.745,
      "step": 5649
    },
    {
      "epoch": 0.3109692333094832,
      "grad_norm": 0.7695828676223755,
      "learning_rate": 9.432321207067799e-06,
      "loss": 0.7555,
      "step": 5650
    },
    {
      "epoch": 0.31102427211183886,
      "grad_norm": 0.8050221800804138,
      "learning_rate": 9.432120583321984e-06,
      "loss": 0.8464,
      "step": 5651
    },
    {
      "epoch": 0.3110793109141945,
      "grad_norm": 0.7242713570594788,
      "learning_rate": 9.431919926265538e-06,
      "loss": 0.7439,
      "step": 5652
    },
    {
      "epoch": 0.3111343497165502,
      "grad_norm": 0.7372434735298157,
      "learning_rate": 9.431719235899967e-06,
      "loss": 0.7973,
      "step": 5653
    },
    {
      "epoch": 0.3111893885189058,
      "grad_norm": 0.7573439478874207,
      "learning_rate": 9.431518512226783e-06,
      "loss": 0.8259,
      "step": 5654
    },
    {
      "epoch": 0.3112444273212615,
      "grad_norm": 0.7098552584648132,
      "learning_rate": 9.43131775524749e-06,
      "loss": 0.8159,
      "step": 5655
    },
    {
      "epoch": 0.31129946612361714,
      "grad_norm": 0.7804632186889648,
      "learning_rate": 9.431116964963599e-06,
      "loss": 0.7795,
      "step": 5656
    },
    {
      "epoch": 0.31135450492597283,
      "grad_norm": 1.0158027410507202,
      "learning_rate": 9.43091614137662e-06,
      "loss": 0.7935,
      "step": 5657
    },
    {
      "epoch": 0.31140954372832846,
      "grad_norm": 0.708238422870636,
      "learning_rate": 9.430715284488059e-06,
      "loss": 0.7592,
      "step": 5658
    },
    {
      "epoch": 0.31146458253068415,
      "grad_norm": 0.7086984515190125,
      "learning_rate": 9.43051439429943e-06,
      "loss": 0.7303,
      "step": 5659
    },
    {
      "epoch": 0.3115196213330398,
      "grad_norm": 0.7620081305503845,
      "learning_rate": 9.43031347081224e-06,
      "loss": 0.7429,
      "step": 5660
    },
    {
      "epoch": 0.3115746601353955,
      "grad_norm": 0.746126115322113,
      "learning_rate": 9.430112514028e-06,
      "loss": 0.8836,
      "step": 5661
    },
    {
      "epoch": 0.3116296989377511,
      "grad_norm": 0.9113686680793762,
      "learning_rate": 9.429911523948221e-06,
      "loss": 0.6343,
      "step": 5662
    },
    {
      "epoch": 0.3116847377401068,
      "grad_norm": 0.700890839099884,
      "learning_rate": 9.429710500574413e-06,
      "loss": 0.8201,
      "step": 5663
    },
    {
      "epoch": 0.31173977654246243,
      "grad_norm": 0.7428706288337708,
      "learning_rate": 9.429509443908085e-06,
      "loss": 0.6838,
      "step": 5664
    },
    {
      "epoch": 0.3117948153448181,
      "grad_norm": 0.851725697517395,
      "learning_rate": 9.429308353950752e-06,
      "loss": 0.7151,
      "step": 5665
    },
    {
      "epoch": 0.31184985414717376,
      "grad_norm": 0.8555309176445007,
      "learning_rate": 9.42910723070392e-06,
      "loss": 0.7384,
      "step": 5666
    },
    {
      "epoch": 0.31190489294952944,
      "grad_norm": 0.735927939414978,
      "learning_rate": 9.428906074169107e-06,
      "loss": 0.6911,
      "step": 5667
    },
    {
      "epoch": 0.3119599317518851,
      "grad_norm": 0.8007609844207764,
      "learning_rate": 9.42870488434782e-06,
      "loss": 0.869,
      "step": 5668
    },
    {
      "epoch": 0.31201497055424077,
      "grad_norm": 0.7604133486747742,
      "learning_rate": 9.42850366124157e-06,
      "loss": 0.7633,
      "step": 5669
    },
    {
      "epoch": 0.3120700093565964,
      "grad_norm": 0.8181144595146179,
      "learning_rate": 9.428302404851875e-06,
      "loss": 0.7631,
      "step": 5670
    },
    {
      "epoch": 0.31212504815895203,
      "grad_norm": 0.7115523219108582,
      "learning_rate": 9.428101115180243e-06,
      "loss": 0.734,
      "step": 5671
    },
    {
      "epoch": 0.3121800869613077,
      "grad_norm": 0.7165855765342712,
      "learning_rate": 9.42789979222819e-06,
      "loss": 0.8068,
      "step": 5672
    },
    {
      "epoch": 0.31223512576366336,
      "grad_norm": 0.6515665650367737,
      "learning_rate": 9.427698435997225e-06,
      "loss": 0.6946,
      "step": 5673
    },
    {
      "epoch": 0.31229016456601905,
      "grad_norm": 0.7692676186561584,
      "learning_rate": 9.427497046488867e-06,
      "loss": 0.7387,
      "step": 5674
    },
    {
      "epoch": 0.3123452033683747,
      "grad_norm": 0.70064777135849,
      "learning_rate": 9.427295623704625e-06,
      "loss": 0.7976,
      "step": 5675
    },
    {
      "epoch": 0.31240024217073037,
      "grad_norm": 0.7464852333068848,
      "learning_rate": 9.427094167646013e-06,
      "loss": 0.7574,
      "step": 5676
    },
    {
      "epoch": 0.312455280973086,
      "grad_norm": 0.7721675634384155,
      "learning_rate": 9.426892678314548e-06,
      "loss": 0.7405,
      "step": 5677
    },
    {
      "epoch": 0.3125103197754417,
      "grad_norm": 0.6581596732139587,
      "learning_rate": 9.42669115571174e-06,
      "loss": 0.6972,
      "step": 5678
    },
    {
      "epoch": 0.3125653585777973,
      "grad_norm": 0.8722662329673767,
      "learning_rate": 9.426489599839108e-06,
      "loss": 0.8073,
      "step": 5679
    },
    {
      "epoch": 0.312620397380153,
      "grad_norm": 0.6800306439399719,
      "learning_rate": 9.426288010698165e-06,
      "loss": 0.7721,
      "step": 5680
    },
    {
      "epoch": 0.31267543618250865,
      "grad_norm": 0.7443979382514954,
      "learning_rate": 9.426086388290428e-06,
      "loss": 0.7719,
      "step": 5681
    },
    {
      "epoch": 0.31273047498486434,
      "grad_norm": 0.7818729877471924,
      "learning_rate": 9.425884732617407e-06,
      "loss": 0.7815,
      "step": 5682
    },
    {
      "epoch": 0.31278551378721997,
      "grad_norm": 0.7640877366065979,
      "learning_rate": 9.425683043680624e-06,
      "loss": 0.8315,
      "step": 5683
    },
    {
      "epoch": 0.31284055258957566,
      "grad_norm": 0.6871064305305481,
      "learning_rate": 9.42548132148159e-06,
      "loss": 0.8017,
      "step": 5684
    },
    {
      "epoch": 0.3128955913919313,
      "grad_norm": 0.8394801616668701,
      "learning_rate": 9.425279566021824e-06,
      "loss": 0.763,
      "step": 5685
    },
    {
      "epoch": 0.312950630194287,
      "grad_norm": 0.7104960083961487,
      "learning_rate": 9.42507777730284e-06,
      "loss": 0.7991,
      "step": 5686
    },
    {
      "epoch": 0.3130056689966426,
      "grad_norm": 0.7820347547531128,
      "learning_rate": 9.424875955326159e-06,
      "loss": 0.825,
      "step": 5687
    },
    {
      "epoch": 0.3130607077989983,
      "grad_norm": 0.783343493938446,
      "learning_rate": 9.424674100093292e-06,
      "loss": 0.8189,
      "step": 5688
    },
    {
      "epoch": 0.31311574660135394,
      "grad_norm": 0.7998474836349487,
      "learning_rate": 9.42447221160576e-06,
      "loss": 0.7382,
      "step": 5689
    },
    {
      "epoch": 0.31317078540370963,
      "grad_norm": 0.7232120633125305,
      "learning_rate": 9.424270289865078e-06,
      "loss": 0.8556,
      "step": 5690
    },
    {
      "epoch": 0.31322582420606526,
      "grad_norm": 0.7944191694259644,
      "learning_rate": 9.424068334872764e-06,
      "loss": 0.8272,
      "step": 5691
    },
    {
      "epoch": 0.31328086300842095,
      "grad_norm": 0.7951859831809998,
      "learning_rate": 9.42386634663034e-06,
      "loss": 0.7612,
      "step": 5692
    },
    {
      "epoch": 0.3133359018107766,
      "grad_norm": 1.394667387008667,
      "learning_rate": 9.423664325139318e-06,
      "loss": 0.8108,
      "step": 5693
    },
    {
      "epoch": 0.3133909406131323,
      "grad_norm": 0.868886411190033,
      "learning_rate": 9.42346227040122e-06,
      "loss": 0.8308,
      "step": 5694
    },
    {
      "epoch": 0.3134459794154879,
      "grad_norm": 0.9442586302757263,
      "learning_rate": 9.423260182417563e-06,
      "loss": 0.9145,
      "step": 5695
    },
    {
      "epoch": 0.3135010182178436,
      "grad_norm": 0.7432793974876404,
      "learning_rate": 9.423058061189868e-06,
      "loss": 0.7715,
      "step": 5696
    },
    {
      "epoch": 0.31355605702019923,
      "grad_norm": 0.7221946120262146,
      "learning_rate": 9.422855906719652e-06,
      "loss": 0.7588,
      "step": 5697
    },
    {
      "epoch": 0.3136110958225549,
      "grad_norm": 0.7459834814071655,
      "learning_rate": 9.422653719008434e-06,
      "loss": 0.7834,
      "step": 5698
    },
    {
      "epoch": 0.31366613462491055,
      "grad_norm": 0.8562330007553101,
      "learning_rate": 9.422451498057737e-06,
      "loss": 0.6994,
      "step": 5699
    },
    {
      "epoch": 0.31372117342726624,
      "grad_norm": 0.672696053981781,
      "learning_rate": 9.422249243869075e-06,
      "loss": 0.7201,
      "step": 5700
    },
    {
      "epoch": 0.3137762122296219,
      "grad_norm": 0.7459990382194519,
      "learning_rate": 9.422046956443973e-06,
      "loss": 0.7663,
      "step": 5701
    },
    {
      "epoch": 0.31383125103197757,
      "grad_norm": 0.9653169512748718,
      "learning_rate": 9.42184463578395e-06,
      "loss": 0.8899,
      "step": 5702
    },
    {
      "epoch": 0.3138862898343332,
      "grad_norm": 0.7137778997421265,
      "learning_rate": 9.421642281890526e-06,
      "loss": 0.74,
      "step": 5703
    },
    {
      "epoch": 0.3139413286366889,
      "grad_norm": 0.6961745619773865,
      "learning_rate": 9.421439894765222e-06,
      "loss": 0.7309,
      "step": 5704
    },
    {
      "epoch": 0.3139963674390445,
      "grad_norm": 0.7843212485313416,
      "learning_rate": 9.421237474409559e-06,
      "loss": 0.8654,
      "step": 5705
    },
    {
      "epoch": 0.3140514062414002,
      "grad_norm": 0.7560604810714722,
      "learning_rate": 9.42103502082506e-06,
      "loss": 0.7949,
      "step": 5706
    },
    {
      "epoch": 0.31410644504375584,
      "grad_norm": 0.756200909614563,
      "learning_rate": 9.420832534013245e-06,
      "loss": 0.7315,
      "step": 5707
    },
    {
      "epoch": 0.31416148384611153,
      "grad_norm": 0.7857967615127563,
      "learning_rate": 9.420630013975635e-06,
      "loss": 0.7698,
      "step": 5708
    },
    {
      "epoch": 0.31421652264846717,
      "grad_norm": 0.6943809986114502,
      "learning_rate": 9.420427460713754e-06,
      "loss": 0.7691,
      "step": 5709
    },
    {
      "epoch": 0.31427156145082286,
      "grad_norm": 0.7460532188415527,
      "learning_rate": 9.420224874229123e-06,
      "loss": 0.7679,
      "step": 5710
    },
    {
      "epoch": 0.3143266002531785,
      "grad_norm": 0.764406144618988,
      "learning_rate": 9.420022254523265e-06,
      "loss": 0.9545,
      "step": 5711
    },
    {
      "epoch": 0.3143816390555342,
      "grad_norm": 0.7191083431243896,
      "learning_rate": 9.419819601597703e-06,
      "loss": 0.728,
      "step": 5712
    },
    {
      "epoch": 0.3144366778578898,
      "grad_norm": 0.8799699544906616,
      "learning_rate": 9.419616915453959e-06,
      "loss": 0.6911,
      "step": 5713
    },
    {
      "epoch": 0.31449171666024545,
      "grad_norm": 0.7505975365638733,
      "learning_rate": 9.419414196093558e-06,
      "loss": 0.7953,
      "step": 5714
    },
    {
      "epoch": 0.31454675546260114,
      "grad_norm": 0.7575502395629883,
      "learning_rate": 9.419211443518023e-06,
      "loss": 0.7752,
      "step": 5715
    },
    {
      "epoch": 0.31460179426495677,
      "grad_norm": 0.7220337986946106,
      "learning_rate": 9.419008657728879e-06,
      "loss": 0.7894,
      "step": 5716
    },
    {
      "epoch": 0.31465683306731246,
      "grad_norm": 0.7797306776046753,
      "learning_rate": 9.418805838727648e-06,
      "loss": 0.7582,
      "step": 5717
    },
    {
      "epoch": 0.3147118718696681,
      "grad_norm": 0.9011242985725403,
      "learning_rate": 9.418602986515855e-06,
      "loss": 0.7379,
      "step": 5718
    },
    {
      "epoch": 0.3147669106720238,
      "grad_norm": 0.7568445801734924,
      "learning_rate": 9.418400101095025e-06,
      "loss": 0.8003,
      "step": 5719
    },
    {
      "epoch": 0.3148219494743794,
      "grad_norm": 0.6810547709465027,
      "learning_rate": 9.418197182466681e-06,
      "loss": 0.7186,
      "step": 5720
    },
    {
      "epoch": 0.3148769882767351,
      "grad_norm": 0.7390284538269043,
      "learning_rate": 9.417994230632352e-06,
      "loss": 0.7478,
      "step": 5721
    },
    {
      "epoch": 0.31493202707909074,
      "grad_norm": 0.695286214351654,
      "learning_rate": 9.41779124559356e-06,
      "loss": 0.7467,
      "step": 5722
    },
    {
      "epoch": 0.3149870658814464,
      "grad_norm": 0.7783445715904236,
      "learning_rate": 9.41758822735183e-06,
      "loss": 0.824,
      "step": 5723
    },
    {
      "epoch": 0.31504210468380206,
      "grad_norm": 0.7176268696784973,
      "learning_rate": 9.41738517590869e-06,
      "loss": 0.7596,
      "step": 5724
    },
    {
      "epoch": 0.31509714348615775,
      "grad_norm": 0.7829678058624268,
      "learning_rate": 9.417182091265668e-06,
      "loss": 0.8184,
      "step": 5725
    },
    {
      "epoch": 0.3151521822885134,
      "grad_norm": 0.7461703419685364,
      "learning_rate": 9.416978973424286e-06,
      "loss": 0.8732,
      "step": 5726
    },
    {
      "epoch": 0.31520722109086907,
      "grad_norm": 0.7186999320983887,
      "learning_rate": 9.416775822386073e-06,
      "loss": 0.6878,
      "step": 5727
    },
    {
      "epoch": 0.3152622598932247,
      "grad_norm": 0.6775033473968506,
      "learning_rate": 9.416572638152553e-06,
      "loss": 0.7211,
      "step": 5728
    },
    {
      "epoch": 0.3153172986955804,
      "grad_norm": 0.6845641732215881,
      "learning_rate": 9.416369420725258e-06,
      "loss": 0.7282,
      "step": 5729
    },
    {
      "epoch": 0.31537233749793603,
      "grad_norm": 0.8301281929016113,
      "learning_rate": 9.416166170105712e-06,
      "loss": 0.7999,
      "step": 5730
    },
    {
      "epoch": 0.3154273763002917,
      "grad_norm": 0.8487183451652527,
      "learning_rate": 9.415962886295442e-06,
      "loss": 0.8202,
      "step": 5731
    },
    {
      "epoch": 0.31548241510264735,
      "grad_norm": 0.74607914686203,
      "learning_rate": 9.415759569295979e-06,
      "loss": 0.7552,
      "step": 5732
    },
    {
      "epoch": 0.31553745390500304,
      "grad_norm": 0.7774194478988647,
      "learning_rate": 9.415556219108846e-06,
      "loss": 0.7847,
      "step": 5733
    },
    {
      "epoch": 0.3155924927073587,
      "grad_norm": 0.7782126069068909,
      "learning_rate": 9.415352835735576e-06,
      "loss": 0.8001,
      "step": 5734
    },
    {
      "epoch": 0.31564753150971436,
      "grad_norm": 0.7577764987945557,
      "learning_rate": 9.415149419177698e-06,
      "loss": 0.8262,
      "step": 5735
    },
    {
      "epoch": 0.31570257031207,
      "grad_norm": 0.7949855327606201,
      "learning_rate": 9.414945969436737e-06,
      "loss": 0.8259,
      "step": 5736
    },
    {
      "epoch": 0.3157576091144257,
      "grad_norm": 0.7670153379440308,
      "learning_rate": 9.414742486514224e-06,
      "loss": 0.7181,
      "step": 5737
    },
    {
      "epoch": 0.3158126479167813,
      "grad_norm": 0.7852359414100647,
      "learning_rate": 9.414538970411687e-06,
      "loss": 0.8802,
      "step": 5738
    },
    {
      "epoch": 0.315867686719137,
      "grad_norm": 0.8300517201423645,
      "learning_rate": 9.414335421130658e-06,
      "loss": 0.7665,
      "step": 5739
    },
    {
      "epoch": 0.31592272552149264,
      "grad_norm": 0.7631614804267883,
      "learning_rate": 9.414131838672666e-06,
      "loss": 0.8864,
      "step": 5740
    },
    {
      "epoch": 0.31597776432384833,
      "grad_norm": 0.7946471571922302,
      "learning_rate": 9.41392822303924e-06,
      "loss": 0.7587,
      "step": 5741
    },
    {
      "epoch": 0.31603280312620396,
      "grad_norm": 0.7043818235397339,
      "learning_rate": 9.413724574231912e-06,
      "loss": 0.7793,
      "step": 5742
    },
    {
      "epoch": 0.31608784192855965,
      "grad_norm": 0.7276063561439514,
      "learning_rate": 9.41352089225221e-06,
      "loss": 0.8064,
      "step": 5743
    },
    {
      "epoch": 0.3161428807309153,
      "grad_norm": 0.7141419053077698,
      "learning_rate": 9.413317177101667e-06,
      "loss": 0.7251,
      "step": 5744
    },
    {
      "epoch": 0.316197919533271,
      "grad_norm": 0.7961493730545044,
      "learning_rate": 9.413113428781815e-06,
      "loss": 0.8438,
      "step": 5745
    },
    {
      "epoch": 0.3162529583356266,
      "grad_norm": 0.7046970129013062,
      "learning_rate": 9.412909647294181e-06,
      "loss": 0.8319,
      "step": 5746
    },
    {
      "epoch": 0.3163079971379823,
      "grad_norm": 0.8231918215751648,
      "learning_rate": 9.412705832640302e-06,
      "loss": 0.7707,
      "step": 5747
    },
    {
      "epoch": 0.31636303594033793,
      "grad_norm": 0.769840657711029,
      "learning_rate": 9.412501984821705e-06,
      "loss": 0.6819,
      "step": 5748
    },
    {
      "epoch": 0.3164180747426936,
      "grad_norm": 0.7526834607124329,
      "learning_rate": 9.412298103839925e-06,
      "loss": 0.8106,
      "step": 5749
    },
    {
      "epoch": 0.31647311354504926,
      "grad_norm": 0.6763152480125427,
      "learning_rate": 9.412094189696494e-06,
      "loss": 0.7577,
      "step": 5750
    },
    {
      "epoch": 0.31652815234740495,
      "grad_norm": 0.8460820317268372,
      "learning_rate": 9.411890242392945e-06,
      "loss": 0.752,
      "step": 5751
    },
    {
      "epoch": 0.3165831911497606,
      "grad_norm": 0.7610277533531189,
      "learning_rate": 9.411686261930809e-06,
      "loss": 0.7284,
      "step": 5752
    },
    {
      "epoch": 0.31663822995211627,
      "grad_norm": 0.7596566081047058,
      "learning_rate": 9.411482248311619e-06,
      "loss": 0.8518,
      "step": 5753
    },
    {
      "epoch": 0.3166932687544719,
      "grad_norm": 0.7615048885345459,
      "learning_rate": 9.41127820153691e-06,
      "loss": 0.8232,
      "step": 5754
    },
    {
      "epoch": 0.3167483075568276,
      "grad_norm": 0.7882834672927856,
      "learning_rate": 9.411074121608215e-06,
      "loss": 0.8682,
      "step": 5755
    },
    {
      "epoch": 0.3168033463591832,
      "grad_norm": 0.748002827167511,
      "learning_rate": 9.410870008527067e-06,
      "loss": 0.7934,
      "step": 5756
    },
    {
      "epoch": 0.31685838516153886,
      "grad_norm": 0.7677696943283081,
      "learning_rate": 9.410665862295003e-06,
      "loss": 0.8114,
      "step": 5757
    },
    {
      "epoch": 0.31691342396389455,
      "grad_norm": 0.8966217041015625,
      "learning_rate": 9.410461682913552e-06,
      "loss": 0.8005,
      "step": 5758
    },
    {
      "epoch": 0.3169684627662502,
      "grad_norm": 0.8769435286521912,
      "learning_rate": 9.410257470384253e-06,
      "loss": 0.7935,
      "step": 5759
    },
    {
      "epoch": 0.31702350156860587,
      "grad_norm": 0.9828680753707886,
      "learning_rate": 9.41005322470864e-06,
      "loss": 0.8182,
      "step": 5760
    },
    {
      "epoch": 0.3170785403709615,
      "grad_norm": 0.7340976595878601,
      "learning_rate": 9.409848945888245e-06,
      "loss": 0.7832,
      "step": 5761
    },
    {
      "epoch": 0.3171335791733172,
      "grad_norm": 0.7516821622848511,
      "learning_rate": 9.409644633924609e-06,
      "loss": 0.8223,
      "step": 5762
    },
    {
      "epoch": 0.3171886179756728,
      "grad_norm": 0.7556331157684326,
      "learning_rate": 9.409440288819263e-06,
      "loss": 0.7631,
      "step": 5763
    },
    {
      "epoch": 0.3172436567780285,
      "grad_norm": 0.6182114481925964,
      "learning_rate": 9.409235910573743e-06,
      "loss": 0.558,
      "step": 5764
    },
    {
      "epoch": 0.31729869558038415,
      "grad_norm": 0.7854578495025635,
      "learning_rate": 9.409031499189586e-06,
      "loss": 0.8496,
      "step": 5765
    },
    {
      "epoch": 0.31735373438273984,
      "grad_norm": 0.7246551513671875,
      "learning_rate": 9.40882705466833e-06,
      "loss": 0.8407,
      "step": 5766
    },
    {
      "epoch": 0.31740877318509547,
      "grad_norm": 1.089107632637024,
      "learning_rate": 9.40862257701151e-06,
      "loss": 0.8363,
      "step": 5767
    },
    {
      "epoch": 0.31746381198745116,
      "grad_norm": 0.9886558055877686,
      "learning_rate": 9.408418066220664e-06,
      "loss": 0.6888,
      "step": 5768
    },
    {
      "epoch": 0.3175188507898068,
      "grad_norm": 0.8724960088729858,
      "learning_rate": 9.408213522297325e-06,
      "loss": 0.7717,
      "step": 5769
    },
    {
      "epoch": 0.3175738895921625,
      "grad_norm": 0.7453228831291199,
      "learning_rate": 9.408008945243035e-06,
      "loss": 0.7081,
      "step": 5770
    },
    {
      "epoch": 0.3176289283945181,
      "grad_norm": 0.7601909637451172,
      "learning_rate": 9.40780433505933e-06,
      "loss": 0.7974,
      "step": 5771
    },
    {
      "epoch": 0.3176839671968738,
      "grad_norm": 0.7704907655715942,
      "learning_rate": 9.407599691747746e-06,
      "loss": 0.7521,
      "step": 5772
    },
    {
      "epoch": 0.31773900599922944,
      "grad_norm": 0.7639214396476746,
      "learning_rate": 9.407395015309824e-06,
      "loss": 0.7888,
      "step": 5773
    },
    {
      "epoch": 0.31779404480158513,
      "grad_norm": 0.711355984210968,
      "learning_rate": 9.4071903057471e-06,
      "loss": 0.7482,
      "step": 5774
    },
    {
      "epoch": 0.31784908360394076,
      "grad_norm": 0.6097242832183838,
      "learning_rate": 9.406985563061114e-06,
      "loss": 0.6533,
      "step": 5775
    },
    {
      "epoch": 0.31790412240629645,
      "grad_norm": 0.807133138179779,
      "learning_rate": 9.406780787253402e-06,
      "loss": 0.7788,
      "step": 5776
    },
    {
      "epoch": 0.3179591612086521,
      "grad_norm": 0.6938545107841492,
      "learning_rate": 9.406575978325508e-06,
      "loss": 0.8046,
      "step": 5777
    },
    {
      "epoch": 0.3180142000110078,
      "grad_norm": 0.848858118057251,
      "learning_rate": 9.406371136278968e-06,
      "loss": 0.8481,
      "step": 5778
    },
    {
      "epoch": 0.3180692388133634,
      "grad_norm": 0.8496920466423035,
      "learning_rate": 9.40616626111532e-06,
      "loss": 0.8172,
      "step": 5779
    },
    {
      "epoch": 0.3181242776157191,
      "grad_norm": 0.8169928193092346,
      "learning_rate": 9.405961352836107e-06,
      "loss": 0.792,
      "step": 5780
    },
    {
      "epoch": 0.31817931641807473,
      "grad_norm": 0.9380607604980469,
      "learning_rate": 9.405756411442868e-06,
      "loss": 0.8371,
      "step": 5781
    },
    {
      "epoch": 0.3182343552204304,
      "grad_norm": 0.6938190460205078,
      "learning_rate": 9.405551436937144e-06,
      "loss": 0.7825,
      "step": 5782
    },
    {
      "epoch": 0.31828939402278605,
      "grad_norm": 0.7726871371269226,
      "learning_rate": 9.405346429320473e-06,
      "loss": 0.6481,
      "step": 5783
    },
    {
      "epoch": 0.31834443282514174,
      "grad_norm": 0.77762770652771,
      "learning_rate": 9.4051413885944e-06,
      "loss": 0.6916,
      "step": 5784
    },
    {
      "epoch": 0.3183994716274974,
      "grad_norm": 0.7580817341804504,
      "learning_rate": 9.404936314760459e-06,
      "loss": 0.8222,
      "step": 5785
    },
    {
      "epoch": 0.31845451042985307,
      "grad_norm": 0.6984102725982666,
      "learning_rate": 9.4047312078202e-06,
      "loss": 0.707,
      "step": 5786
    },
    {
      "epoch": 0.3185095492322087,
      "grad_norm": 0.6887965202331543,
      "learning_rate": 9.404526067775159e-06,
      "loss": 0.7289,
      "step": 5787
    },
    {
      "epoch": 0.3185645880345644,
      "grad_norm": 0.7022155523300171,
      "learning_rate": 9.404320894626879e-06,
      "loss": 0.741,
      "step": 5788
    },
    {
      "epoch": 0.31861962683692,
      "grad_norm": 0.8007381558418274,
      "learning_rate": 9.404115688376903e-06,
      "loss": 0.8332,
      "step": 5789
    },
    {
      "epoch": 0.3186746656392757,
      "grad_norm": 0.6985924243927002,
      "learning_rate": 9.40391044902677e-06,
      "loss": 0.7849,
      "step": 5790
    },
    {
      "epoch": 0.31872970444163135,
      "grad_norm": 0.771060585975647,
      "learning_rate": 9.403705176578028e-06,
      "loss": 0.8728,
      "step": 5791
    },
    {
      "epoch": 0.31878474324398703,
      "grad_norm": 0.6976794600486755,
      "learning_rate": 9.403499871032214e-06,
      "loss": 0.7621,
      "step": 5792
    },
    {
      "epoch": 0.31883978204634267,
      "grad_norm": 0.7552126049995422,
      "learning_rate": 9.403294532390876e-06,
      "loss": 0.7641,
      "step": 5793
    },
    {
      "epoch": 0.31889482084869836,
      "grad_norm": 1.0032007694244385,
      "learning_rate": 9.403089160655553e-06,
      "loss": 0.8497,
      "step": 5794
    },
    {
      "epoch": 0.318949859651054,
      "grad_norm": 0.7193583250045776,
      "learning_rate": 9.402883755827792e-06,
      "loss": 0.7991,
      "step": 5795
    },
    {
      "epoch": 0.3190048984534097,
      "grad_norm": 0.7665852308273315,
      "learning_rate": 9.402678317909135e-06,
      "loss": 0.7692,
      "step": 5796
    },
    {
      "epoch": 0.3190599372557653,
      "grad_norm": 0.7514237761497498,
      "learning_rate": 9.402472846901125e-06,
      "loss": 0.7388,
      "step": 5797
    },
    {
      "epoch": 0.319114976058121,
      "grad_norm": 0.6817325353622437,
      "learning_rate": 9.402267342805309e-06,
      "loss": 0.7249,
      "step": 5798
    },
    {
      "epoch": 0.31917001486047664,
      "grad_norm": 0.7659624218940735,
      "learning_rate": 9.402061805623229e-06,
      "loss": 0.755,
      "step": 5799
    },
    {
      "epoch": 0.31922505366283227,
      "grad_norm": 0.7860668301582336,
      "learning_rate": 9.401856235356431e-06,
      "loss": 0.8175,
      "step": 5800
    },
    {
      "epoch": 0.31928009246518796,
      "grad_norm": 0.714030921459198,
      "learning_rate": 9.401650632006461e-06,
      "loss": 0.7359,
      "step": 5801
    },
    {
      "epoch": 0.3193351312675436,
      "grad_norm": 0.6052672266960144,
      "learning_rate": 9.401444995574862e-06,
      "loss": 0.6167,
      "step": 5802
    },
    {
      "epoch": 0.3193901700698993,
      "grad_norm": 0.7960858941078186,
      "learning_rate": 9.40123932606318e-06,
      "loss": 0.7542,
      "step": 5803
    },
    {
      "epoch": 0.3194452088722549,
      "grad_norm": 0.7926718592643738,
      "learning_rate": 9.401033623472962e-06,
      "loss": 0.8292,
      "step": 5804
    },
    {
      "epoch": 0.3195002476746106,
      "grad_norm": 0.7950098514556885,
      "learning_rate": 9.400827887805754e-06,
      "loss": 0.9332,
      "step": 5805
    },
    {
      "epoch": 0.31955528647696624,
      "grad_norm": 0.7564939260482788,
      "learning_rate": 9.400622119063101e-06,
      "loss": 0.7217,
      "step": 5806
    },
    {
      "epoch": 0.3196103252793219,
      "grad_norm": 0.7582511901855469,
      "learning_rate": 9.40041631724655e-06,
      "loss": 0.723,
      "step": 5807
    },
    {
      "epoch": 0.31966536408167756,
      "grad_norm": 0.8826366066932678,
      "learning_rate": 9.400210482357648e-06,
      "loss": 0.6977,
      "step": 5808
    },
    {
      "epoch": 0.31972040288403325,
      "grad_norm": 0.7029523253440857,
      "learning_rate": 9.400004614397941e-06,
      "loss": 0.6949,
      "step": 5809
    },
    {
      "epoch": 0.3197754416863889,
      "grad_norm": 0.7651532888412476,
      "learning_rate": 9.399798713368979e-06,
      "loss": 0.7158,
      "step": 5810
    },
    {
      "epoch": 0.3198304804887446,
      "grad_norm": 0.9379491806030273,
      "learning_rate": 9.399592779272307e-06,
      "loss": 0.7639,
      "step": 5811
    },
    {
      "epoch": 0.3198855192911002,
      "grad_norm": 0.7945839762687683,
      "learning_rate": 9.399386812109474e-06,
      "loss": 0.8175,
      "step": 5812
    },
    {
      "epoch": 0.3199405580934559,
      "grad_norm": 0.9462345242500305,
      "learning_rate": 9.399180811882025e-06,
      "loss": 0.6635,
      "step": 5813
    },
    {
      "epoch": 0.31999559689581153,
      "grad_norm": 1.0449726581573486,
      "learning_rate": 9.398974778591513e-06,
      "loss": 0.789,
      "step": 5814
    },
    {
      "epoch": 0.3200506356981672,
      "grad_norm": 0.8295683860778809,
      "learning_rate": 9.398768712239483e-06,
      "loss": 0.7937,
      "step": 5815
    },
    {
      "epoch": 0.32010567450052285,
      "grad_norm": 0.7578030228614807,
      "learning_rate": 9.398562612827485e-06,
      "loss": 0.8291,
      "step": 5816
    },
    {
      "epoch": 0.32016071330287854,
      "grad_norm": 0.804563581943512,
      "learning_rate": 9.398356480357068e-06,
      "loss": 0.7604,
      "step": 5817
    },
    {
      "epoch": 0.3202157521052342,
      "grad_norm": 0.8073337078094482,
      "learning_rate": 9.39815031482978e-06,
      "loss": 0.8288,
      "step": 5818
    },
    {
      "epoch": 0.32027079090758986,
      "grad_norm": 0.8054978251457214,
      "learning_rate": 9.397944116247173e-06,
      "loss": 0.819,
      "step": 5819
    },
    {
      "epoch": 0.3203258297099455,
      "grad_norm": 0.8304697871208191,
      "learning_rate": 9.397737884610794e-06,
      "loss": 0.7991,
      "step": 5820
    },
    {
      "epoch": 0.3203808685123012,
      "grad_norm": 0.784662663936615,
      "learning_rate": 9.397531619922195e-06,
      "loss": 0.763,
      "step": 5821
    },
    {
      "epoch": 0.3204359073146568,
      "grad_norm": 0.726046085357666,
      "learning_rate": 9.397325322182926e-06,
      "loss": 0.7926,
      "step": 5822
    },
    {
      "epoch": 0.3204909461170125,
      "grad_norm": 0.7291107773780823,
      "learning_rate": 9.397118991394535e-06,
      "loss": 0.6871,
      "step": 5823
    },
    {
      "epoch": 0.32054598491936814,
      "grad_norm": 0.7870203256607056,
      "learning_rate": 9.396912627558577e-06,
      "loss": 0.7827,
      "step": 5824
    },
    {
      "epoch": 0.32060102372172383,
      "grad_norm": 0.8665844798088074,
      "learning_rate": 9.3967062306766e-06,
      "loss": 0.8098,
      "step": 5825
    },
    {
      "epoch": 0.32065606252407947,
      "grad_norm": 0.7743843793869019,
      "learning_rate": 9.396499800750157e-06,
      "loss": 0.835,
      "step": 5826
    },
    {
      "epoch": 0.32071110132643516,
      "grad_norm": 0.7724023461341858,
      "learning_rate": 9.396293337780796e-06,
      "loss": 0.8928,
      "step": 5827
    },
    {
      "epoch": 0.3207661401287908,
      "grad_norm": 0.7497217655181885,
      "learning_rate": 9.39608684177007e-06,
      "loss": 0.8035,
      "step": 5828
    },
    {
      "epoch": 0.3208211789311465,
      "grad_norm": 0.8346971869468689,
      "learning_rate": 9.395880312719536e-06,
      "loss": 0.8879,
      "step": 5829
    },
    {
      "epoch": 0.3208762177335021,
      "grad_norm": 0.836626410484314,
      "learning_rate": 9.39567375063074e-06,
      "loss": 0.8523,
      "step": 5830
    },
    {
      "epoch": 0.3209312565358578,
      "grad_norm": 0.734428346157074,
      "learning_rate": 9.395467155505237e-06,
      "loss": 0.7568,
      "step": 5831
    },
    {
      "epoch": 0.32098629533821343,
      "grad_norm": 0.6620383858680725,
      "learning_rate": 9.39526052734458e-06,
      "loss": 0.7296,
      "step": 5832
    },
    {
      "epoch": 0.3210413341405691,
      "grad_norm": 0.9356484413146973,
      "learning_rate": 9.39505386615032e-06,
      "loss": 0.8233,
      "step": 5833
    },
    {
      "epoch": 0.32109637294292476,
      "grad_norm": 0.9238032698631287,
      "learning_rate": 9.394847171924013e-06,
      "loss": 0.7397,
      "step": 5834
    },
    {
      "epoch": 0.32115141174528045,
      "grad_norm": 0.7161185145378113,
      "learning_rate": 9.39464044466721e-06,
      "loss": 0.7541,
      "step": 5835
    },
    {
      "epoch": 0.3212064505476361,
      "grad_norm": 0.8381507396697998,
      "learning_rate": 9.394433684381467e-06,
      "loss": 0.7839,
      "step": 5836
    },
    {
      "epoch": 0.32126148934999177,
      "grad_norm": 0.8299819231033325,
      "learning_rate": 9.394226891068337e-06,
      "loss": 0.871,
      "step": 5837
    },
    {
      "epoch": 0.3213165281523474,
      "grad_norm": 0.7443987131118774,
      "learning_rate": 9.394020064729372e-06,
      "loss": 0.7661,
      "step": 5838
    },
    {
      "epoch": 0.3213715669547031,
      "grad_norm": 0.7084206938743591,
      "learning_rate": 9.393813205366128e-06,
      "loss": 0.7609,
      "step": 5839
    },
    {
      "epoch": 0.3214266057570587,
      "grad_norm": 0.7443114519119263,
      "learning_rate": 9.393606312980164e-06,
      "loss": 0.8189,
      "step": 5840
    },
    {
      "epoch": 0.3214816445594144,
      "grad_norm": 0.7157652974128723,
      "learning_rate": 9.393399387573028e-06,
      "loss": 0.8369,
      "step": 5841
    },
    {
      "epoch": 0.32153668336177005,
      "grad_norm": 0.709507942199707,
      "learning_rate": 9.393192429146278e-06,
      "loss": 0.7314,
      "step": 5842
    },
    {
      "epoch": 0.3215917221641257,
      "grad_norm": 0.7704687714576721,
      "learning_rate": 9.39298543770147e-06,
      "loss": 0.8793,
      "step": 5843
    },
    {
      "epoch": 0.32164676096648137,
      "grad_norm": 0.8123828172683716,
      "learning_rate": 9.39277841324016e-06,
      "loss": 0.8748,
      "step": 5844
    },
    {
      "epoch": 0.321701799768837,
      "grad_norm": 0.6951777338981628,
      "learning_rate": 9.392571355763903e-06,
      "loss": 0.7883,
      "step": 5845
    },
    {
      "epoch": 0.3217568385711927,
      "grad_norm": 0.6753274202346802,
      "learning_rate": 9.392364265274256e-06,
      "loss": 0.7292,
      "step": 5846
    },
    {
      "epoch": 0.3218118773735483,
      "grad_norm": 0.7940227389335632,
      "learning_rate": 9.392157141772775e-06,
      "loss": 0.7919,
      "step": 5847
    },
    {
      "epoch": 0.321866916175904,
      "grad_norm": 0.6706317067146301,
      "learning_rate": 9.391949985261016e-06,
      "loss": 0.6791,
      "step": 5848
    },
    {
      "epoch": 0.32192195497825965,
      "grad_norm": 0.7898741960525513,
      "learning_rate": 9.391742795740537e-06,
      "loss": 0.7539,
      "step": 5849
    },
    {
      "epoch": 0.32197699378061534,
      "grad_norm": 0.7623887658119202,
      "learning_rate": 9.391535573212895e-06,
      "loss": 0.7891,
      "step": 5850
    },
    {
      "epoch": 0.322032032582971,
      "grad_norm": 0.6852909326553345,
      "learning_rate": 9.391328317679647e-06,
      "loss": 0.6587,
      "step": 5851
    },
    {
      "epoch": 0.32208707138532666,
      "grad_norm": 0.7944231033325195,
      "learning_rate": 9.39112102914235e-06,
      "loss": 0.8316,
      "step": 5852
    },
    {
      "epoch": 0.3221421101876823,
      "grad_norm": 0.6720889806747437,
      "learning_rate": 9.390913707602563e-06,
      "loss": 0.7791,
      "step": 5853
    },
    {
      "epoch": 0.322197148990038,
      "grad_norm": 0.7482234239578247,
      "learning_rate": 9.390706353061845e-06,
      "loss": 0.826,
      "step": 5854
    },
    {
      "epoch": 0.3222521877923936,
      "grad_norm": 0.6821579933166504,
      "learning_rate": 9.390498965521752e-06,
      "loss": 0.7183,
      "step": 5855
    },
    {
      "epoch": 0.3223072265947493,
      "grad_norm": 0.755171537399292,
      "learning_rate": 9.390291544983845e-06,
      "loss": 0.6887,
      "step": 5856
    },
    {
      "epoch": 0.32236226539710494,
      "grad_norm": 0.748824417591095,
      "learning_rate": 9.39008409144968e-06,
      "loss": 0.7169,
      "step": 5857
    },
    {
      "epoch": 0.32241730419946063,
      "grad_norm": 0.7479343414306641,
      "learning_rate": 9.38987660492082e-06,
      "loss": 0.8122,
      "step": 5858
    },
    {
      "epoch": 0.32247234300181626,
      "grad_norm": 0.7459376454353333,
      "learning_rate": 9.389669085398823e-06,
      "loss": 0.7782,
      "step": 5859
    },
    {
      "epoch": 0.32252738180417195,
      "grad_norm": 0.7016253471374512,
      "learning_rate": 9.389461532885246e-06,
      "loss": 0.7866,
      "step": 5860
    },
    {
      "epoch": 0.3225824206065276,
      "grad_norm": 0.6711822152137756,
      "learning_rate": 9.389253947381654e-06,
      "loss": 0.7223,
      "step": 5861
    },
    {
      "epoch": 0.3226374594088833,
      "grad_norm": 0.855045735836029,
      "learning_rate": 9.389046328889602e-06,
      "loss": 0.7327,
      "step": 5862
    },
    {
      "epoch": 0.3226924982112389,
      "grad_norm": 0.7309823632240295,
      "learning_rate": 9.388838677410654e-06,
      "loss": 0.7737,
      "step": 5863
    },
    {
      "epoch": 0.3227475370135946,
      "grad_norm": 0.7737841010093689,
      "learning_rate": 9.388630992946369e-06,
      "loss": 0.7061,
      "step": 5864
    },
    {
      "epoch": 0.32280257581595023,
      "grad_norm": 0.9448195099830627,
      "learning_rate": 9.388423275498307e-06,
      "loss": 0.8382,
      "step": 5865
    },
    {
      "epoch": 0.3228576146183059,
      "grad_norm": 0.7348229885101318,
      "learning_rate": 9.388215525068032e-06,
      "loss": 0.8317,
      "step": 5866
    },
    {
      "epoch": 0.32291265342066156,
      "grad_norm": 1.2628185749053955,
      "learning_rate": 9.388007741657103e-06,
      "loss": 0.7959,
      "step": 5867
    },
    {
      "epoch": 0.32296769222301724,
      "grad_norm": 0.7730327844619751,
      "learning_rate": 9.387799925267083e-06,
      "loss": 0.7455,
      "step": 5868
    },
    {
      "epoch": 0.3230227310253729,
      "grad_norm": 0.8273047804832458,
      "learning_rate": 9.387592075899532e-06,
      "loss": 0.877,
      "step": 5869
    },
    {
      "epoch": 0.32307776982772857,
      "grad_norm": 0.7413405776023865,
      "learning_rate": 9.387384193556014e-06,
      "loss": 0.7734,
      "step": 5870
    },
    {
      "epoch": 0.3231328086300842,
      "grad_norm": 1.0173207521438599,
      "learning_rate": 9.387176278238092e-06,
      "loss": 0.8674,
      "step": 5871
    },
    {
      "epoch": 0.3231878474324399,
      "grad_norm": 0.7741677761077881,
      "learning_rate": 9.386968329947327e-06,
      "loss": 0.8226,
      "step": 5872
    },
    {
      "epoch": 0.3232428862347955,
      "grad_norm": 0.8912034034729004,
      "learning_rate": 9.38676034868528e-06,
      "loss": 0.7977,
      "step": 5873
    },
    {
      "epoch": 0.3232979250371512,
      "grad_norm": 0.7343642711639404,
      "learning_rate": 9.386552334453519e-06,
      "loss": 0.7639,
      "step": 5874
    },
    {
      "epoch": 0.32335296383950685,
      "grad_norm": 0.697225034236908,
      "learning_rate": 9.386344287253603e-06,
      "loss": 0.6801,
      "step": 5875
    },
    {
      "epoch": 0.32340800264186254,
      "grad_norm": 0.7082511186599731,
      "learning_rate": 9.386136207087099e-06,
      "loss": 0.746,
      "step": 5876
    },
    {
      "epoch": 0.32346304144421817,
      "grad_norm": 0.671419620513916,
      "learning_rate": 9.38592809395557e-06,
      "loss": 0.7023,
      "step": 5877
    },
    {
      "epoch": 0.32351808024657386,
      "grad_norm": 0.775834321975708,
      "learning_rate": 9.385719947860579e-06,
      "loss": 0.7797,
      "step": 5878
    },
    {
      "epoch": 0.3235731190489295,
      "grad_norm": 0.7867023348808289,
      "learning_rate": 9.38551176880369e-06,
      "loss": 0.8165,
      "step": 5879
    },
    {
      "epoch": 0.3236281578512852,
      "grad_norm": 0.7099916934967041,
      "learning_rate": 9.385303556786469e-06,
      "loss": 0.7598,
      "step": 5880
    },
    {
      "epoch": 0.3236831966536408,
      "grad_norm": 0.7362176179885864,
      "learning_rate": 9.385095311810479e-06,
      "loss": 0.8002,
      "step": 5881
    },
    {
      "epoch": 0.3237382354559965,
      "grad_norm": 0.7310882806777954,
      "learning_rate": 9.384887033877288e-06,
      "loss": 0.7641,
      "step": 5882
    },
    {
      "epoch": 0.32379327425835214,
      "grad_norm": 0.7769907116889954,
      "learning_rate": 9.384678722988458e-06,
      "loss": 0.7938,
      "step": 5883
    },
    {
      "epoch": 0.3238483130607078,
      "grad_norm": 0.9913623929023743,
      "learning_rate": 9.384470379145558e-06,
      "loss": 0.8203,
      "step": 5884
    },
    {
      "epoch": 0.32390335186306346,
      "grad_norm": 0.8765702247619629,
      "learning_rate": 9.384262002350153e-06,
      "loss": 0.9343,
      "step": 5885
    },
    {
      "epoch": 0.3239583906654191,
      "grad_norm": 0.8122400641441345,
      "learning_rate": 9.384053592603808e-06,
      "loss": 0.8325,
      "step": 5886
    },
    {
      "epoch": 0.3240134294677748,
      "grad_norm": 0.7600317597389221,
      "learning_rate": 9.383845149908089e-06,
      "loss": 0.8335,
      "step": 5887
    },
    {
      "epoch": 0.3240684682701304,
      "grad_norm": 0.9472025632858276,
      "learning_rate": 9.383636674264563e-06,
      "loss": 0.7265,
      "step": 5888
    },
    {
      "epoch": 0.3241235070724861,
      "grad_norm": 0.6961854100227356,
      "learning_rate": 9.383428165674797e-06,
      "loss": 0.6962,
      "step": 5889
    },
    {
      "epoch": 0.32417854587484174,
      "grad_norm": 0.7032504081726074,
      "learning_rate": 9.38321962414036e-06,
      "loss": 0.7627,
      "step": 5890
    },
    {
      "epoch": 0.32423358467719743,
      "grad_norm": 0.7727648019790649,
      "learning_rate": 9.383011049662816e-06,
      "loss": 0.757,
      "step": 5891
    },
    {
      "epoch": 0.32428862347955306,
      "grad_norm": 0.7263824343681335,
      "learning_rate": 9.382802442243735e-06,
      "loss": 0.8057,
      "step": 5892
    },
    {
      "epoch": 0.32434366228190875,
      "grad_norm": 0.7576926350593567,
      "learning_rate": 9.382593801884683e-06,
      "loss": 0.763,
      "step": 5893
    },
    {
      "epoch": 0.3243987010842644,
      "grad_norm": 0.7468064427375793,
      "learning_rate": 9.38238512858723e-06,
      "loss": 0.731,
      "step": 5894
    },
    {
      "epoch": 0.3244537398866201,
      "grad_norm": 0.9570005536079407,
      "learning_rate": 9.382176422352944e-06,
      "loss": 0.7985,
      "step": 5895
    },
    {
      "epoch": 0.3245087786889757,
      "grad_norm": 0.7296027541160583,
      "learning_rate": 9.381967683183393e-06,
      "loss": 0.8117,
      "step": 5896
    },
    {
      "epoch": 0.3245638174913314,
      "grad_norm": 0.7330880165100098,
      "learning_rate": 9.381758911080145e-06,
      "loss": 0.7229,
      "step": 5897
    },
    {
      "epoch": 0.32461885629368703,
      "grad_norm": 0.7247695922851562,
      "learning_rate": 9.38155010604477e-06,
      "loss": 0.7704,
      "step": 5898
    },
    {
      "epoch": 0.3246738950960427,
      "grad_norm": 0.8011599779129028,
      "learning_rate": 9.381341268078836e-06,
      "loss": 0.6982,
      "step": 5899
    },
    {
      "epoch": 0.32472893389839835,
      "grad_norm": 0.7931570410728455,
      "learning_rate": 9.381132397183917e-06,
      "loss": 0.8188,
      "step": 5900
    },
    {
      "epoch": 0.32478397270075404,
      "grad_norm": 0.7469003200531006,
      "learning_rate": 9.380923493361577e-06,
      "loss": 0.7638,
      "step": 5901
    },
    {
      "epoch": 0.3248390115031097,
      "grad_norm": 0.7442750334739685,
      "learning_rate": 9.380714556613391e-06,
      "loss": 0.8134,
      "step": 5902
    },
    {
      "epoch": 0.32489405030546537,
      "grad_norm": 0.8014402985572815,
      "learning_rate": 9.380505586940925e-06,
      "loss": 0.838,
      "step": 5903
    },
    {
      "epoch": 0.324949089107821,
      "grad_norm": 0.7287543416023254,
      "learning_rate": 9.380296584345751e-06,
      "loss": 0.7317,
      "step": 5904
    },
    {
      "epoch": 0.3250041279101767,
      "grad_norm": 0.7754266262054443,
      "learning_rate": 9.380087548829441e-06,
      "loss": 0.7205,
      "step": 5905
    },
    {
      "epoch": 0.3250591667125323,
      "grad_norm": 0.7439714074134827,
      "learning_rate": 9.379878480393567e-06,
      "loss": 0.821,
      "step": 5906
    },
    {
      "epoch": 0.325114205514888,
      "grad_norm": 0.7142870426177979,
      "learning_rate": 9.379669379039698e-06,
      "loss": 0.7462,
      "step": 5907
    },
    {
      "epoch": 0.32516924431724364,
      "grad_norm": 0.6522948145866394,
      "learning_rate": 9.379460244769407e-06,
      "loss": 0.739,
      "step": 5908
    },
    {
      "epoch": 0.32522428311959933,
      "grad_norm": 0.7879271507263184,
      "learning_rate": 9.379251077584263e-06,
      "loss": 0.719,
      "step": 5909
    },
    {
      "epoch": 0.32527932192195497,
      "grad_norm": 0.6969109773635864,
      "learning_rate": 9.379041877485842e-06,
      "loss": 0.7517,
      "step": 5910
    },
    {
      "epoch": 0.32533436072431066,
      "grad_norm": 0.736890971660614,
      "learning_rate": 9.378832644475714e-06,
      "loss": 0.7797,
      "step": 5911
    },
    {
      "epoch": 0.3253893995266663,
      "grad_norm": 0.7504066824913025,
      "learning_rate": 9.378623378555451e-06,
      "loss": 0.7502,
      "step": 5912
    },
    {
      "epoch": 0.325444438329022,
      "grad_norm": 0.9339223504066467,
      "learning_rate": 9.378414079726629e-06,
      "loss": 0.8842,
      "step": 5913
    },
    {
      "epoch": 0.3254994771313776,
      "grad_norm": 1.08317232131958,
      "learning_rate": 9.378204747990818e-06,
      "loss": 0.7503,
      "step": 5914
    },
    {
      "epoch": 0.3255545159337333,
      "grad_norm": 0.722665011882782,
      "learning_rate": 9.37799538334959e-06,
      "loss": 0.7825,
      "step": 5915
    },
    {
      "epoch": 0.32560955473608894,
      "grad_norm": 0.7969509959220886,
      "learning_rate": 9.377785985804521e-06,
      "loss": 0.8678,
      "step": 5916
    },
    {
      "epoch": 0.3256645935384446,
      "grad_norm": 0.7944697141647339,
      "learning_rate": 9.377576555357187e-06,
      "loss": 0.8067,
      "step": 5917
    },
    {
      "epoch": 0.32571963234080026,
      "grad_norm": 0.905580461025238,
      "learning_rate": 9.377367092009158e-06,
      "loss": 0.7689,
      "step": 5918
    },
    {
      "epoch": 0.32577467114315595,
      "grad_norm": 0.7428018450737,
      "learning_rate": 9.37715759576201e-06,
      "loss": 0.7748,
      "step": 5919
    },
    {
      "epoch": 0.3258297099455116,
      "grad_norm": 0.7746098041534424,
      "learning_rate": 9.376948066617316e-06,
      "loss": 0.7235,
      "step": 5920
    },
    {
      "epoch": 0.32588474874786727,
      "grad_norm": 0.6842886805534363,
      "learning_rate": 9.376738504576653e-06,
      "loss": 0.7697,
      "step": 5921
    },
    {
      "epoch": 0.3259397875502229,
      "grad_norm": 0.7858961224555969,
      "learning_rate": 9.376528909641595e-06,
      "loss": 0.7746,
      "step": 5922
    },
    {
      "epoch": 0.3259948263525786,
      "grad_norm": 0.7534621357917786,
      "learning_rate": 9.376319281813717e-06,
      "loss": 0.8183,
      "step": 5923
    },
    {
      "epoch": 0.3260498651549342,
      "grad_norm": 1.2406045198440552,
      "learning_rate": 9.376109621094594e-06,
      "loss": 0.8173,
      "step": 5924
    },
    {
      "epoch": 0.3261049039572899,
      "grad_norm": 0.740075945854187,
      "learning_rate": 9.375899927485804e-06,
      "loss": 0.725,
      "step": 5925
    },
    {
      "epoch": 0.32615994275964555,
      "grad_norm": 0.8432604074478149,
      "learning_rate": 9.375690200988921e-06,
      "loss": 0.7805,
      "step": 5926
    },
    {
      "epoch": 0.32621498156200124,
      "grad_norm": 0.7652943134307861,
      "learning_rate": 9.37548044160552e-06,
      "loss": 0.8609,
      "step": 5927
    },
    {
      "epoch": 0.32627002036435687,
      "grad_norm": 0.7629607915878296,
      "learning_rate": 9.37527064933718e-06,
      "loss": 0.8776,
      "step": 5928
    },
    {
      "epoch": 0.3263250591667125,
      "grad_norm": 0.8648995757102966,
      "learning_rate": 9.375060824185479e-06,
      "loss": 0.7543,
      "step": 5929
    },
    {
      "epoch": 0.3263800979690682,
      "grad_norm": 0.8069457411766052,
      "learning_rate": 9.374850966151989e-06,
      "loss": 0.7995,
      "step": 5930
    },
    {
      "epoch": 0.32643513677142383,
      "grad_norm": 0.7948445677757263,
      "learning_rate": 9.374641075238293e-06,
      "loss": 0.8312,
      "step": 5931
    },
    {
      "epoch": 0.3264901755737795,
      "grad_norm": 0.7739841341972351,
      "learning_rate": 9.374431151445963e-06,
      "loss": 0.8442,
      "step": 5932
    },
    {
      "epoch": 0.32654521437613515,
      "grad_norm": 0.7382220029830933,
      "learning_rate": 9.374221194776583e-06,
      "loss": 0.7519,
      "step": 5933
    },
    {
      "epoch": 0.32660025317849084,
      "grad_norm": 0.7876916527748108,
      "learning_rate": 9.374011205231725e-06,
      "loss": 0.817,
      "step": 5934
    },
    {
      "epoch": 0.3266552919808465,
      "grad_norm": 0.7175565958023071,
      "learning_rate": 9.373801182812969e-06,
      "loss": 0.7317,
      "step": 5935
    },
    {
      "epoch": 0.32671033078320216,
      "grad_norm": 0.7739143967628479,
      "learning_rate": 9.373591127521894e-06,
      "loss": 0.8134,
      "step": 5936
    },
    {
      "epoch": 0.3267653695855578,
      "grad_norm": 0.7388991713523865,
      "learning_rate": 9.373381039360082e-06,
      "loss": 0.8758,
      "step": 5937
    },
    {
      "epoch": 0.3268204083879135,
      "grad_norm": 0.7393535375595093,
      "learning_rate": 9.373170918329105e-06,
      "loss": 0.7453,
      "step": 5938
    },
    {
      "epoch": 0.3268754471902691,
      "grad_norm": 0.7168294191360474,
      "learning_rate": 9.372960764430547e-06,
      "loss": 0.6535,
      "step": 5939
    },
    {
      "epoch": 0.3269304859926248,
      "grad_norm": 0.7472337484359741,
      "learning_rate": 9.372750577665988e-06,
      "loss": 0.8065,
      "step": 5940
    },
    {
      "epoch": 0.32698552479498044,
      "grad_norm": 0.7211272120475769,
      "learning_rate": 9.372540358037005e-06,
      "loss": 0.7389,
      "step": 5941
    },
    {
      "epoch": 0.32704056359733613,
      "grad_norm": 0.8097178339958191,
      "learning_rate": 9.37233010554518e-06,
      "loss": 0.8034,
      "step": 5942
    },
    {
      "epoch": 0.32709560239969176,
      "grad_norm": 0.7929103970527649,
      "learning_rate": 9.372119820192091e-06,
      "loss": 0.796,
      "step": 5943
    },
    {
      "epoch": 0.32715064120204745,
      "grad_norm": 0.701171875,
      "learning_rate": 9.37190950197932e-06,
      "loss": 0.7092,
      "step": 5944
    },
    {
      "epoch": 0.3272056800044031,
      "grad_norm": 0.679142951965332,
      "learning_rate": 9.371699150908448e-06,
      "loss": 0.6995,
      "step": 5945
    },
    {
      "epoch": 0.3272607188067588,
      "grad_norm": 0.7757906913757324,
      "learning_rate": 9.371488766981057e-06,
      "loss": 0.8662,
      "step": 5946
    },
    {
      "epoch": 0.3273157576091144,
      "grad_norm": 0.8086597323417664,
      "learning_rate": 9.371278350198724e-06,
      "loss": 0.7455,
      "step": 5947
    },
    {
      "epoch": 0.3273707964114701,
      "grad_norm": 0.6443416476249695,
      "learning_rate": 9.371067900563033e-06,
      "loss": 0.7262,
      "step": 5948
    },
    {
      "epoch": 0.32742583521382573,
      "grad_norm": 0.8132354021072388,
      "learning_rate": 9.370857418075567e-06,
      "loss": 0.7841,
      "step": 5949
    },
    {
      "epoch": 0.3274808740161814,
      "grad_norm": 0.6811150908470154,
      "learning_rate": 9.370646902737907e-06,
      "loss": 0.6955,
      "step": 5950
    },
    {
      "epoch": 0.32753591281853706,
      "grad_norm": 0.8956614136695862,
      "learning_rate": 9.370436354551633e-06,
      "loss": 0.8218,
      "step": 5951
    },
    {
      "epoch": 0.32759095162089275,
      "grad_norm": 0.6807655692100525,
      "learning_rate": 9.370225773518332e-06,
      "loss": 0.7869,
      "step": 5952
    },
    {
      "epoch": 0.3276459904232484,
      "grad_norm": 0.7506592869758606,
      "learning_rate": 9.37001515963958e-06,
      "loss": 0.7975,
      "step": 5953
    },
    {
      "epoch": 0.32770102922560407,
      "grad_norm": 0.7488718032836914,
      "learning_rate": 9.369804512916966e-06,
      "loss": 0.7611,
      "step": 5954
    },
    {
      "epoch": 0.3277560680279597,
      "grad_norm": 0.734569251537323,
      "learning_rate": 9.369593833352073e-06,
      "loss": 0.8532,
      "step": 5955
    },
    {
      "epoch": 0.3278111068303154,
      "grad_norm": 0.780170738697052,
      "learning_rate": 9.36938312094648e-06,
      "loss": 0.7766,
      "step": 5956
    },
    {
      "epoch": 0.327866145632671,
      "grad_norm": 0.6329935193061829,
      "learning_rate": 9.369172375701774e-06,
      "loss": 0.6789,
      "step": 5957
    },
    {
      "epoch": 0.3279211844350267,
      "grad_norm": 1.0177193880081177,
      "learning_rate": 9.368961597619537e-06,
      "loss": 0.8362,
      "step": 5958
    },
    {
      "epoch": 0.32797622323738235,
      "grad_norm": 0.730696439743042,
      "learning_rate": 9.368750786701354e-06,
      "loss": 0.7696,
      "step": 5959
    },
    {
      "epoch": 0.32803126203973804,
      "grad_norm": 0.7946468591690063,
      "learning_rate": 9.36853994294881e-06,
      "loss": 0.8559,
      "step": 5960
    },
    {
      "epoch": 0.32808630084209367,
      "grad_norm": 0.9353142976760864,
      "learning_rate": 9.368329066363489e-06,
      "loss": 0.9041,
      "step": 5961
    },
    {
      "epoch": 0.32814133964444936,
      "grad_norm": 0.7256187796592712,
      "learning_rate": 9.368118156946977e-06,
      "loss": 0.787,
      "step": 5962
    },
    {
      "epoch": 0.328196378446805,
      "grad_norm": 0.7454268336296082,
      "learning_rate": 9.367907214700858e-06,
      "loss": 0.7255,
      "step": 5963
    },
    {
      "epoch": 0.3282514172491607,
      "grad_norm": 0.7087902426719666,
      "learning_rate": 9.367696239626716e-06,
      "loss": 0.7166,
      "step": 5964
    },
    {
      "epoch": 0.3283064560515163,
      "grad_norm": 0.8217566609382629,
      "learning_rate": 9.36748523172614e-06,
      "loss": 0.8351,
      "step": 5965
    },
    {
      "epoch": 0.328361494853872,
      "grad_norm": 0.7712824940681458,
      "learning_rate": 9.367274191000713e-06,
      "loss": 0.7561,
      "step": 5966
    },
    {
      "epoch": 0.32841653365622764,
      "grad_norm": 0.6798166036605835,
      "learning_rate": 9.367063117452024e-06,
      "loss": 0.7447,
      "step": 5967
    },
    {
      "epoch": 0.3284715724585833,
      "grad_norm": 0.7139115929603577,
      "learning_rate": 9.366852011081655e-06,
      "loss": 0.7728,
      "step": 5968
    },
    {
      "epoch": 0.32852661126093896,
      "grad_norm": 1.0488213300704956,
      "learning_rate": 9.366640871891196e-06,
      "loss": 0.8283,
      "step": 5969
    },
    {
      "epoch": 0.32858165006329465,
      "grad_norm": 0.7939574122428894,
      "learning_rate": 9.366429699882233e-06,
      "loss": 0.849,
      "step": 5970
    },
    {
      "epoch": 0.3286366888656503,
      "grad_norm": 0.7959052324295044,
      "learning_rate": 9.366218495056356e-06,
      "loss": 0.7469,
      "step": 5971
    },
    {
      "epoch": 0.3286917276680059,
      "grad_norm": 0.7293235063552856,
      "learning_rate": 9.366007257415146e-06,
      "loss": 0.8537,
      "step": 5972
    },
    {
      "epoch": 0.3287467664703616,
      "grad_norm": 0.7490390539169312,
      "learning_rate": 9.365795986960196e-06,
      "loss": 0.8166,
      "step": 5973
    },
    {
      "epoch": 0.32880180527271724,
      "grad_norm": 0.6572316884994507,
      "learning_rate": 9.365584683693093e-06,
      "loss": 0.6919,
      "step": 5974
    },
    {
      "epoch": 0.32885684407507293,
      "grad_norm": 0.7286609411239624,
      "learning_rate": 9.365373347615421e-06,
      "loss": 0.768,
      "step": 5975
    },
    {
      "epoch": 0.32891188287742856,
      "grad_norm": 0.7798202037811279,
      "learning_rate": 9.365161978728772e-06,
      "loss": 0.788,
      "step": 5976
    },
    {
      "epoch": 0.32896692167978425,
      "grad_norm": 0.7224245071411133,
      "learning_rate": 9.364950577034737e-06,
      "loss": 0.7551,
      "step": 5977
    },
    {
      "epoch": 0.3290219604821399,
      "grad_norm": 0.7238701581954956,
      "learning_rate": 9.364739142534898e-06,
      "loss": 0.6663,
      "step": 5978
    },
    {
      "epoch": 0.3290769992844956,
      "grad_norm": 0.8947147727012634,
      "learning_rate": 9.36452767523085e-06,
      "loss": 0.8559,
      "step": 5979
    },
    {
      "epoch": 0.3291320380868512,
      "grad_norm": 0.7346563935279846,
      "learning_rate": 9.36431617512418e-06,
      "loss": 0.7915,
      "step": 5980
    },
    {
      "epoch": 0.3291870768892069,
      "grad_norm": 0.7674046158790588,
      "learning_rate": 9.364104642216479e-06,
      "loss": 0.7643,
      "step": 5981
    },
    {
      "epoch": 0.32924211569156253,
      "grad_norm": 0.7288179397583008,
      "learning_rate": 9.363893076509335e-06,
      "loss": 0.7796,
      "step": 5982
    },
    {
      "epoch": 0.3292971544939182,
      "grad_norm": 0.6603766083717346,
      "learning_rate": 9.363681478004339e-06,
      "loss": 0.7035,
      "step": 5983
    },
    {
      "epoch": 0.32935219329627385,
      "grad_norm": 0.7523066997528076,
      "learning_rate": 9.36346984670308e-06,
      "loss": 0.8196,
      "step": 5984
    },
    {
      "epoch": 0.32940723209862954,
      "grad_norm": 0.730312168598175,
      "learning_rate": 9.36325818260715e-06,
      "loss": 0.7967,
      "step": 5985
    },
    {
      "epoch": 0.3294622709009852,
      "grad_norm": 0.7341319918632507,
      "learning_rate": 9.363046485718139e-06,
      "loss": 0.8361,
      "step": 5986
    },
    {
      "epoch": 0.32951730970334087,
      "grad_norm": 0.839894711971283,
      "learning_rate": 9.36283475603764e-06,
      "loss": 0.862,
      "step": 5987
    },
    {
      "epoch": 0.3295723485056965,
      "grad_norm": 0.7794893980026245,
      "learning_rate": 9.362622993567243e-06,
      "loss": 0.8521,
      "step": 5988
    },
    {
      "epoch": 0.3296273873080522,
      "grad_norm": 0.929410457611084,
      "learning_rate": 9.362411198308538e-06,
      "loss": 0.7644,
      "step": 5989
    },
    {
      "epoch": 0.3296824261104078,
      "grad_norm": 0.7687333226203918,
      "learning_rate": 9.362199370263118e-06,
      "loss": 0.8047,
      "step": 5990
    },
    {
      "epoch": 0.3297374649127635,
      "grad_norm": 0.8040616512298584,
      "learning_rate": 9.361987509432576e-06,
      "loss": 0.7574,
      "step": 5991
    },
    {
      "epoch": 0.32979250371511915,
      "grad_norm": 0.7743237614631653,
      "learning_rate": 9.361775615818503e-06,
      "loss": 0.8491,
      "step": 5992
    },
    {
      "epoch": 0.32984754251747483,
      "grad_norm": 1.2796664237976074,
      "learning_rate": 9.361563689422493e-06,
      "loss": 0.7975,
      "step": 5993
    },
    {
      "epoch": 0.32990258131983047,
      "grad_norm": 0.9493466019630432,
      "learning_rate": 9.361351730246136e-06,
      "loss": 1.0258,
      "step": 5994
    },
    {
      "epoch": 0.32995762012218616,
      "grad_norm": 0.7148050665855408,
      "learning_rate": 9.36113973829103e-06,
      "loss": 0.805,
      "step": 5995
    },
    {
      "epoch": 0.3300126589245418,
      "grad_norm": 0.723426342010498,
      "learning_rate": 9.360927713558762e-06,
      "loss": 0.6886,
      "step": 5996
    },
    {
      "epoch": 0.3300676977268975,
      "grad_norm": 0.8274679183959961,
      "learning_rate": 9.360715656050929e-06,
      "loss": 0.8559,
      "step": 5997
    },
    {
      "epoch": 0.3301227365292531,
      "grad_norm": 0.7493795156478882,
      "learning_rate": 9.360503565769126e-06,
      "loss": 0.8266,
      "step": 5998
    },
    {
      "epoch": 0.3301777753316088,
      "grad_norm": 0.7690125703811646,
      "learning_rate": 9.360291442714944e-06,
      "loss": 0.783,
      "step": 5999
    },
    {
      "epoch": 0.33023281413396444,
      "grad_norm": 0.8740219473838806,
      "learning_rate": 9.360079286889981e-06,
      "loss": 0.8409,
      "step": 6000
    },
    {
      "epoch": 0.3302878529363201,
      "grad_norm": 0.6931017637252808,
      "learning_rate": 9.359867098295827e-06,
      "loss": 0.7985,
      "step": 6001
    },
    {
      "epoch": 0.33034289173867576,
      "grad_norm": 0.915532112121582,
      "learning_rate": 9.35965487693408e-06,
      "loss": 0.8718,
      "step": 6002
    },
    {
      "epoch": 0.33039793054103145,
      "grad_norm": 0.7898837924003601,
      "learning_rate": 9.359442622806332e-06,
      "loss": 0.8571,
      "step": 6003
    },
    {
      "epoch": 0.3304529693433871,
      "grad_norm": 0.8661002516746521,
      "learning_rate": 9.359230335914182e-06,
      "loss": 0.7963,
      "step": 6004
    },
    {
      "epoch": 0.33050800814574277,
      "grad_norm": 0.7188493013381958,
      "learning_rate": 9.359018016259223e-06,
      "loss": 0.8188,
      "step": 6005
    },
    {
      "epoch": 0.3305630469480984,
      "grad_norm": 0.8648282289505005,
      "learning_rate": 9.358805663843051e-06,
      "loss": 0.9136,
      "step": 6006
    },
    {
      "epoch": 0.3306180857504541,
      "grad_norm": 0.8010255098342896,
      "learning_rate": 9.358593278667265e-06,
      "loss": 0.849,
      "step": 6007
    },
    {
      "epoch": 0.3306731245528097,
      "grad_norm": 0.8128451108932495,
      "learning_rate": 9.358380860733456e-06,
      "loss": 0.8082,
      "step": 6008
    },
    {
      "epoch": 0.3307281633551654,
      "grad_norm": 1.0003761053085327,
      "learning_rate": 9.358168410043224e-06,
      "loss": 0.9064,
      "step": 6009
    },
    {
      "epoch": 0.33078320215752105,
      "grad_norm": 0.7412391901016235,
      "learning_rate": 9.357955926598163e-06,
      "loss": 0.8049,
      "step": 6010
    },
    {
      "epoch": 0.33083824095987674,
      "grad_norm": 0.795615553855896,
      "learning_rate": 9.357743410399875e-06,
      "loss": 0.7923,
      "step": 6011
    },
    {
      "epoch": 0.3308932797622324,
      "grad_norm": 0.8696123957633972,
      "learning_rate": 9.357530861449953e-06,
      "loss": 0.8543,
      "step": 6012
    },
    {
      "epoch": 0.33094831856458806,
      "grad_norm": 0.8909900784492493,
      "learning_rate": 9.357318279749994e-06,
      "loss": 0.6157,
      "step": 6013
    },
    {
      "epoch": 0.3310033573669437,
      "grad_norm": 0.7326250672340393,
      "learning_rate": 9.357105665301597e-06,
      "loss": 0.7647,
      "step": 6014
    },
    {
      "epoch": 0.33105839616929933,
      "grad_norm": 0.8425576090812683,
      "learning_rate": 9.356893018106364e-06,
      "loss": 0.7832,
      "step": 6015
    },
    {
      "epoch": 0.331113434971655,
      "grad_norm": 0.7404599785804749,
      "learning_rate": 9.356680338165885e-06,
      "loss": 0.7759,
      "step": 6016
    },
    {
      "epoch": 0.33116847377401065,
      "grad_norm": 0.6935396790504456,
      "learning_rate": 9.356467625481765e-06,
      "loss": 0.7488,
      "step": 6017
    },
    {
      "epoch": 0.33122351257636634,
      "grad_norm": 0.7799031138420105,
      "learning_rate": 9.3562548800556e-06,
      "loss": 0.7617,
      "step": 6018
    },
    {
      "epoch": 0.331278551378722,
      "grad_norm": 0.7824636101722717,
      "learning_rate": 9.35604210188899e-06,
      "loss": 0.7936,
      "step": 6019
    },
    {
      "epoch": 0.33133359018107766,
      "grad_norm": 0.7051861882209778,
      "learning_rate": 9.355829290983531e-06,
      "loss": 0.7869,
      "step": 6020
    },
    {
      "epoch": 0.3313886289834333,
      "grad_norm": 0.8172006607055664,
      "learning_rate": 9.355616447340826e-06,
      "loss": 0.8888,
      "step": 6021
    },
    {
      "epoch": 0.331443667785789,
      "grad_norm": 0.7263272404670715,
      "learning_rate": 9.355403570962475e-06,
      "loss": 0.8393,
      "step": 6022
    },
    {
      "epoch": 0.3314987065881446,
      "grad_norm": 0.7143926620483398,
      "learning_rate": 9.355190661850077e-06,
      "loss": 0.6693,
      "step": 6023
    },
    {
      "epoch": 0.3315537453905003,
      "grad_norm": 0.7294363975524902,
      "learning_rate": 9.354977720005232e-06,
      "loss": 0.8035,
      "step": 6024
    },
    {
      "epoch": 0.33160878419285594,
      "grad_norm": 0.7072308659553528,
      "learning_rate": 9.354764745429538e-06,
      "loss": 0.761,
      "step": 6025
    },
    {
      "epoch": 0.33166382299521163,
      "grad_norm": 0.6945865154266357,
      "learning_rate": 9.3545517381246e-06,
      "loss": 0.7212,
      "step": 6026
    },
    {
      "epoch": 0.33171886179756727,
      "grad_norm": 0.7645060420036316,
      "learning_rate": 9.354338698092016e-06,
      "loss": 0.812,
      "step": 6027
    },
    {
      "epoch": 0.33177390059992296,
      "grad_norm": 0.9494503140449524,
      "learning_rate": 9.354125625333387e-06,
      "loss": 0.9037,
      "step": 6028
    },
    {
      "epoch": 0.3318289394022786,
      "grad_norm": 0.7311872243881226,
      "learning_rate": 9.353912519850317e-06,
      "loss": 0.7137,
      "step": 6029
    },
    {
      "epoch": 0.3318839782046343,
      "grad_norm": 0.658562958240509,
      "learning_rate": 9.353699381644405e-06,
      "loss": 0.7048,
      "step": 6030
    },
    {
      "epoch": 0.3319390170069899,
      "grad_norm": 0.8106339573860168,
      "learning_rate": 9.353486210717253e-06,
      "loss": 0.8905,
      "step": 6031
    },
    {
      "epoch": 0.3319940558093456,
      "grad_norm": 0.8166239261627197,
      "learning_rate": 9.353273007070465e-06,
      "loss": 0.7011,
      "step": 6032
    },
    {
      "epoch": 0.33204909461170123,
      "grad_norm": 0.730172872543335,
      "learning_rate": 9.353059770705643e-06,
      "loss": 0.6934,
      "step": 6033
    },
    {
      "epoch": 0.3321041334140569,
      "grad_norm": 0.7633965611457825,
      "learning_rate": 9.352846501624387e-06,
      "loss": 0.7379,
      "step": 6034
    },
    {
      "epoch": 0.33215917221641256,
      "grad_norm": 0.7786447405815125,
      "learning_rate": 9.352633199828304e-06,
      "loss": 0.8533,
      "step": 6035
    },
    {
      "epoch": 0.33221421101876825,
      "grad_norm": 0.7211753726005554,
      "learning_rate": 9.352419865318993e-06,
      "loss": 0.815,
      "step": 6036
    },
    {
      "epoch": 0.3322692498211239,
      "grad_norm": 0.6861024498939514,
      "learning_rate": 9.352206498098062e-06,
      "loss": 0.7678,
      "step": 6037
    },
    {
      "epoch": 0.33232428862347957,
      "grad_norm": 0.7702088952064514,
      "learning_rate": 9.35199309816711e-06,
      "loss": 0.8463,
      "step": 6038
    },
    {
      "epoch": 0.3323793274258352,
      "grad_norm": 0.7179547548294067,
      "learning_rate": 9.351779665527742e-06,
      "loss": 0.8315,
      "step": 6039
    },
    {
      "epoch": 0.3324343662281909,
      "grad_norm": 0.8686990737915039,
      "learning_rate": 9.351566200181565e-06,
      "loss": 0.8396,
      "step": 6040
    },
    {
      "epoch": 0.3324894050305465,
      "grad_norm": 0.7269062995910645,
      "learning_rate": 9.351352702130181e-06,
      "loss": 0.7126,
      "step": 6041
    },
    {
      "epoch": 0.3325444438329022,
      "grad_norm": 0.7759222984313965,
      "learning_rate": 9.351139171375195e-06,
      "loss": 0.8383,
      "step": 6042
    },
    {
      "epoch": 0.33259948263525785,
      "grad_norm": 0.6882128119468689,
      "learning_rate": 9.350925607918212e-06,
      "loss": 0.6371,
      "step": 6043
    },
    {
      "epoch": 0.33265452143761354,
      "grad_norm": 0.7552365660667419,
      "learning_rate": 9.350712011760834e-06,
      "loss": 0.8018,
      "step": 6044
    },
    {
      "epoch": 0.33270956023996917,
      "grad_norm": 0.8320692181587219,
      "learning_rate": 9.350498382904672e-06,
      "loss": 0.8556,
      "step": 6045
    },
    {
      "epoch": 0.33276459904232486,
      "grad_norm": 0.7542223334312439,
      "learning_rate": 9.350284721351326e-06,
      "loss": 0.8006,
      "step": 6046
    },
    {
      "epoch": 0.3328196378446805,
      "grad_norm": 1.2724859714508057,
      "learning_rate": 9.350071027102406e-06,
      "loss": 0.9253,
      "step": 6047
    },
    {
      "epoch": 0.3328746766470362,
      "grad_norm": 0.731383204460144,
      "learning_rate": 9.349857300159517e-06,
      "loss": 0.83,
      "step": 6048
    },
    {
      "epoch": 0.3329297154493918,
      "grad_norm": 0.731419026851654,
      "learning_rate": 9.349643540524265e-06,
      "loss": 0.779,
      "step": 6049
    },
    {
      "epoch": 0.3329847542517475,
      "grad_norm": 0.8462278842926025,
      "learning_rate": 9.349429748198256e-06,
      "loss": 0.84,
      "step": 6050
    },
    {
      "epoch": 0.33303979305410314,
      "grad_norm": 0.8199888467788696,
      "learning_rate": 9.349215923183098e-06,
      "loss": 0.844,
      "step": 6051
    },
    {
      "epoch": 0.33309483185645883,
      "grad_norm": 0.8696722984313965,
      "learning_rate": 9.349002065480397e-06,
      "loss": 0.709,
      "step": 6052
    },
    {
      "epoch": 0.33314987065881446,
      "grad_norm": 0.8484870195388794,
      "learning_rate": 9.34878817509176e-06,
      "loss": 0.7434,
      "step": 6053
    },
    {
      "epoch": 0.33320490946117015,
      "grad_norm": 0.8392589688301086,
      "learning_rate": 9.348574252018796e-06,
      "loss": 0.8972,
      "step": 6054
    },
    {
      "epoch": 0.3332599482635258,
      "grad_norm": 0.673829972743988,
      "learning_rate": 9.34836029626311e-06,
      "loss": 0.6789,
      "step": 6055
    },
    {
      "epoch": 0.3333149870658815,
      "grad_norm": 0.6693649888038635,
      "learning_rate": 9.348146307826315e-06,
      "loss": 0.68,
      "step": 6056
    },
    {
      "epoch": 0.3333700258682371,
      "grad_norm": 0.8516272306442261,
      "learning_rate": 9.347932286710014e-06,
      "loss": 0.8585,
      "step": 6057
    },
    {
      "epoch": 0.33342506467059274,
      "grad_norm": 0.7431588768959045,
      "learning_rate": 9.347718232915818e-06,
      "loss": 0.8239,
      "step": 6058
    },
    {
      "epoch": 0.33348010347294843,
      "grad_norm": 0.8823427557945251,
      "learning_rate": 9.347504146445336e-06,
      "loss": 0.845,
      "step": 6059
    },
    {
      "epoch": 0.33353514227530406,
      "grad_norm": 0.7884035110473633,
      "learning_rate": 9.347290027300177e-06,
      "loss": 0.8503,
      "step": 6060
    },
    {
      "epoch": 0.33359018107765975,
      "grad_norm": 0.841397225856781,
      "learning_rate": 9.34707587548195e-06,
      "loss": 0.7551,
      "step": 6061
    },
    {
      "epoch": 0.3336452198800154,
      "grad_norm": 0.7592034935951233,
      "learning_rate": 9.346861690992263e-06,
      "loss": 0.8516,
      "step": 6062
    },
    {
      "epoch": 0.3337002586823711,
      "grad_norm": 0.6925262212753296,
      "learning_rate": 9.346647473832728e-06,
      "loss": 0.7351,
      "step": 6063
    },
    {
      "epoch": 0.3337552974847267,
      "grad_norm": 0.8152759075164795,
      "learning_rate": 9.346433224004955e-06,
      "loss": 0.7673,
      "step": 6064
    },
    {
      "epoch": 0.3338103362870824,
      "grad_norm": 0.7383455038070679,
      "learning_rate": 9.346218941510551e-06,
      "loss": 0.7312,
      "step": 6065
    },
    {
      "epoch": 0.33386537508943803,
      "grad_norm": 0.7905310392379761,
      "learning_rate": 9.346004626351131e-06,
      "loss": 0.7891,
      "step": 6066
    },
    {
      "epoch": 0.3339204138917937,
      "grad_norm": 0.7032167315483093,
      "learning_rate": 9.345790278528305e-06,
      "loss": 0.8358,
      "step": 6067
    },
    {
      "epoch": 0.33397545269414936,
      "grad_norm": 0.6415952444076538,
      "learning_rate": 9.34557589804368e-06,
      "loss": 0.6716,
      "step": 6068
    },
    {
      "epoch": 0.33403049149650504,
      "grad_norm": 0.7558899521827698,
      "learning_rate": 9.34536148489887e-06,
      "loss": 0.781,
      "step": 6069
    },
    {
      "epoch": 0.3340855302988607,
      "grad_norm": 0.8913301825523376,
      "learning_rate": 9.345147039095485e-06,
      "loss": 0.8482,
      "step": 6070
    },
    {
      "epoch": 0.33414056910121637,
      "grad_norm": 0.768984854221344,
      "learning_rate": 9.34493256063514e-06,
      "loss": 0.7578,
      "step": 6071
    },
    {
      "epoch": 0.334195607903572,
      "grad_norm": 0.7428637742996216,
      "learning_rate": 9.344718049519445e-06,
      "loss": 0.7812,
      "step": 6072
    },
    {
      "epoch": 0.3342506467059277,
      "grad_norm": 0.7290430665016174,
      "learning_rate": 9.344503505750012e-06,
      "loss": 0.7536,
      "step": 6073
    },
    {
      "epoch": 0.3343056855082833,
      "grad_norm": 0.7637680172920227,
      "learning_rate": 9.344288929328453e-06,
      "loss": 0.8576,
      "step": 6074
    },
    {
      "epoch": 0.334360724310639,
      "grad_norm": 0.9568214416503906,
      "learning_rate": 9.344074320256379e-06,
      "loss": 0.897,
      "step": 6075
    },
    {
      "epoch": 0.33441576311299465,
      "grad_norm": 0.7516217827796936,
      "learning_rate": 9.34385967853541e-06,
      "loss": 0.7853,
      "step": 6076
    },
    {
      "epoch": 0.33447080191535034,
      "grad_norm": 0.833039402961731,
      "learning_rate": 9.34364500416715e-06,
      "loss": 0.702,
      "step": 6077
    },
    {
      "epoch": 0.33452584071770597,
      "grad_norm": 0.8080580830574036,
      "learning_rate": 9.34343029715322e-06,
      "loss": 0.7867,
      "step": 6078
    },
    {
      "epoch": 0.33458087952006166,
      "grad_norm": 0.8039596080780029,
      "learning_rate": 9.343215557495229e-06,
      "loss": 0.8221,
      "step": 6079
    },
    {
      "epoch": 0.3346359183224173,
      "grad_norm": 0.7003986835479736,
      "learning_rate": 9.343000785194794e-06,
      "loss": 0.746,
      "step": 6080
    },
    {
      "epoch": 0.334690957124773,
      "grad_norm": 0.6623722314834595,
      "learning_rate": 9.342785980253526e-06,
      "loss": 0.6998,
      "step": 6081
    },
    {
      "epoch": 0.3347459959271286,
      "grad_norm": 0.8425901532173157,
      "learning_rate": 9.342571142673042e-06,
      "loss": 0.8789,
      "step": 6082
    },
    {
      "epoch": 0.3348010347294843,
      "grad_norm": 0.7263861894607544,
      "learning_rate": 9.342356272454954e-06,
      "loss": 0.7299,
      "step": 6083
    },
    {
      "epoch": 0.33485607353183994,
      "grad_norm": 0.8420364260673523,
      "learning_rate": 9.34214136960088e-06,
      "loss": 0.8073,
      "step": 6084
    },
    {
      "epoch": 0.3349111123341956,
      "grad_norm": 0.950019359588623,
      "learning_rate": 9.341926434112435e-06,
      "loss": 0.9288,
      "step": 6085
    },
    {
      "epoch": 0.33496615113655126,
      "grad_norm": 0.7583657503128052,
      "learning_rate": 9.341711465991231e-06,
      "loss": 0.8079,
      "step": 6086
    },
    {
      "epoch": 0.33502118993890695,
      "grad_norm": 0.7623111605644226,
      "learning_rate": 9.341496465238887e-06,
      "loss": 0.879,
      "step": 6087
    },
    {
      "epoch": 0.3350762287412626,
      "grad_norm": 0.8934749960899353,
      "learning_rate": 9.341281431857017e-06,
      "loss": 0.9348,
      "step": 6088
    },
    {
      "epoch": 0.33513126754361827,
      "grad_norm": 0.7363337874412537,
      "learning_rate": 9.341066365847238e-06,
      "loss": 0.8284,
      "step": 6089
    },
    {
      "epoch": 0.3351863063459739,
      "grad_norm": 0.6408932209014893,
      "learning_rate": 9.340851267211166e-06,
      "loss": 0.6019,
      "step": 6090
    },
    {
      "epoch": 0.3352413451483296,
      "grad_norm": 0.8491614460945129,
      "learning_rate": 9.34063613595042e-06,
      "loss": 0.7287,
      "step": 6091
    },
    {
      "epoch": 0.33529638395068523,
      "grad_norm": 0.6922628879547119,
      "learning_rate": 9.340420972066612e-06,
      "loss": 0.6649,
      "step": 6092
    },
    {
      "epoch": 0.3353514227530409,
      "grad_norm": 0.7304210662841797,
      "learning_rate": 9.340205775561364e-06,
      "loss": 0.7373,
      "step": 6093
    },
    {
      "epoch": 0.33540646155539655,
      "grad_norm": 0.8924282193183899,
      "learning_rate": 9.339990546436289e-06,
      "loss": 0.8337,
      "step": 6094
    },
    {
      "epoch": 0.33546150035775224,
      "grad_norm": 0.7671791315078735,
      "learning_rate": 9.339775284693008e-06,
      "loss": 0.856,
      "step": 6095
    },
    {
      "epoch": 0.3355165391601079,
      "grad_norm": 0.830427348613739,
      "learning_rate": 9.339559990333138e-06,
      "loss": 0.7204,
      "step": 6096
    },
    {
      "epoch": 0.33557157796246356,
      "grad_norm": 0.7064357399940491,
      "learning_rate": 9.339344663358297e-06,
      "loss": 0.8533,
      "step": 6097
    },
    {
      "epoch": 0.3356266167648192,
      "grad_norm": 0.7828566431999207,
      "learning_rate": 9.3391293037701e-06,
      "loss": 0.7203,
      "step": 6098
    },
    {
      "epoch": 0.3356816555671749,
      "grad_norm": 0.7686871886253357,
      "learning_rate": 9.338913911570172e-06,
      "loss": 0.7813,
      "step": 6099
    },
    {
      "epoch": 0.3357366943695305,
      "grad_norm": 0.7536553740501404,
      "learning_rate": 9.338698486760126e-06,
      "loss": 0.7581,
      "step": 6100
    },
    {
      "epoch": 0.33579173317188615,
      "grad_norm": 0.7240094542503357,
      "learning_rate": 9.338483029341586e-06,
      "loss": 0.7513,
      "step": 6101
    },
    {
      "epoch": 0.33584677197424184,
      "grad_norm": 0.7519696354866028,
      "learning_rate": 9.338267539316169e-06,
      "loss": 0.8139,
      "step": 6102
    },
    {
      "epoch": 0.3359018107765975,
      "grad_norm": 0.7267377376556396,
      "learning_rate": 9.338052016685492e-06,
      "loss": 0.7807,
      "step": 6103
    },
    {
      "epoch": 0.33595684957895317,
      "grad_norm": 0.6925491094589233,
      "learning_rate": 9.33783646145118e-06,
      "loss": 0.8124,
      "step": 6104
    },
    {
      "epoch": 0.3360118883813088,
      "grad_norm": 0.6896460652351379,
      "learning_rate": 9.337620873614848e-06,
      "loss": 0.7459,
      "step": 6105
    },
    {
      "epoch": 0.3360669271836645,
      "grad_norm": 0.8631082773208618,
      "learning_rate": 9.337405253178121e-06,
      "loss": 0.7662,
      "step": 6106
    },
    {
      "epoch": 0.3361219659860201,
      "grad_norm": 0.76750248670578,
      "learning_rate": 9.337189600142614e-06,
      "loss": 0.9016,
      "step": 6107
    },
    {
      "epoch": 0.3361770047883758,
      "grad_norm": 0.9230479001998901,
      "learning_rate": 9.336973914509952e-06,
      "loss": 0.7631,
      "step": 6108
    },
    {
      "epoch": 0.33623204359073144,
      "grad_norm": 0.746776282787323,
      "learning_rate": 9.336758196281756e-06,
      "loss": 0.6934,
      "step": 6109
    },
    {
      "epoch": 0.33628708239308713,
      "grad_norm": 0.7631211280822754,
      "learning_rate": 9.336542445459646e-06,
      "loss": 0.7957,
      "step": 6110
    },
    {
      "epoch": 0.33634212119544277,
      "grad_norm": 0.7460417151451111,
      "learning_rate": 9.336326662045243e-06,
      "loss": 0.7979,
      "step": 6111
    },
    {
      "epoch": 0.33639715999779846,
      "grad_norm": 0.7072319388389587,
      "learning_rate": 9.336110846040171e-06,
      "loss": 0.763,
      "step": 6112
    },
    {
      "epoch": 0.3364521988001541,
      "grad_norm": 0.822266697883606,
      "learning_rate": 9.33589499744605e-06,
      "loss": 0.7719,
      "step": 6113
    },
    {
      "epoch": 0.3365072376025098,
      "grad_norm": 0.778685986995697,
      "learning_rate": 9.335679116264502e-06,
      "loss": 0.896,
      "step": 6114
    },
    {
      "epoch": 0.3365622764048654,
      "grad_norm": 0.9335552453994751,
      "learning_rate": 9.33546320249715e-06,
      "loss": 0.7317,
      "step": 6115
    },
    {
      "epoch": 0.3366173152072211,
      "grad_norm": 0.755109965801239,
      "learning_rate": 9.33524725614562e-06,
      "loss": 0.8184,
      "step": 6116
    },
    {
      "epoch": 0.33667235400957674,
      "grad_norm": 0.7963696122169495,
      "learning_rate": 9.33503127721153e-06,
      "loss": 0.7835,
      "step": 6117
    },
    {
      "epoch": 0.3367273928119324,
      "grad_norm": 0.8298614621162415,
      "learning_rate": 9.334815265696506e-06,
      "loss": 0.7946,
      "step": 6118
    },
    {
      "epoch": 0.33678243161428806,
      "grad_norm": 0.728638768196106,
      "learning_rate": 9.33459922160217e-06,
      "loss": 0.801,
      "step": 6119
    },
    {
      "epoch": 0.33683747041664375,
      "grad_norm": 0.7275198698043823,
      "learning_rate": 9.334383144930146e-06,
      "loss": 0.7721,
      "step": 6120
    },
    {
      "epoch": 0.3368925092189994,
      "grad_norm": 0.7146986722946167,
      "learning_rate": 9.33416703568206e-06,
      "loss": 0.7573,
      "step": 6121
    },
    {
      "epoch": 0.33694754802135507,
      "grad_norm": 0.7875215411186218,
      "learning_rate": 9.333950893859533e-06,
      "loss": 0.8223,
      "step": 6122
    },
    {
      "epoch": 0.3370025868237107,
      "grad_norm": 0.7636967301368713,
      "learning_rate": 9.333734719464193e-06,
      "loss": 0.7596,
      "step": 6123
    },
    {
      "epoch": 0.3370576256260664,
      "grad_norm": 0.8068925142288208,
      "learning_rate": 9.333518512497663e-06,
      "loss": 0.834,
      "step": 6124
    },
    {
      "epoch": 0.337112664428422,
      "grad_norm": 0.7153680920600891,
      "learning_rate": 9.333302272961566e-06,
      "loss": 0.703,
      "step": 6125
    },
    {
      "epoch": 0.3371677032307777,
      "grad_norm": 0.7429617047309875,
      "learning_rate": 9.33308600085753e-06,
      "loss": 0.7327,
      "step": 6126
    },
    {
      "epoch": 0.33722274203313335,
      "grad_norm": 0.6937283873558044,
      "learning_rate": 9.33286969618718e-06,
      "loss": 0.6494,
      "step": 6127
    },
    {
      "epoch": 0.33727778083548904,
      "grad_norm": 0.7775923609733582,
      "learning_rate": 9.33265335895214e-06,
      "loss": 0.8668,
      "step": 6128
    },
    {
      "epoch": 0.33733281963784467,
      "grad_norm": 0.6911064386367798,
      "learning_rate": 9.33243698915404e-06,
      "loss": 0.6462,
      "step": 6129
    },
    {
      "epoch": 0.33738785844020036,
      "grad_norm": 0.8951280117034912,
      "learning_rate": 9.3322205867945e-06,
      "loss": 0.825,
      "step": 6130
    },
    {
      "epoch": 0.337442897242556,
      "grad_norm": 0.9521064758300781,
      "learning_rate": 9.332004151875151e-06,
      "loss": 0.641,
      "step": 6131
    },
    {
      "epoch": 0.3374979360449117,
      "grad_norm": 0.7036865949630737,
      "learning_rate": 9.33178768439762e-06,
      "loss": 0.804,
      "step": 6132
    },
    {
      "epoch": 0.3375529748472673,
      "grad_norm": 1.0232574939727783,
      "learning_rate": 9.331571184363529e-06,
      "loss": 0.8577,
      "step": 6133
    },
    {
      "epoch": 0.337608013649623,
      "grad_norm": 0.9680090546607971,
      "learning_rate": 9.33135465177451e-06,
      "loss": 0.7725,
      "step": 6134
    },
    {
      "epoch": 0.33766305245197864,
      "grad_norm": 0.7664901614189148,
      "learning_rate": 9.33113808663219e-06,
      "loss": 0.8406,
      "step": 6135
    },
    {
      "epoch": 0.33771809125433433,
      "grad_norm": 0.6703250408172607,
      "learning_rate": 9.330921488938193e-06,
      "loss": 0.7311,
      "step": 6136
    },
    {
      "epoch": 0.33777313005668996,
      "grad_norm": 0.7364899516105652,
      "learning_rate": 9.330704858694151e-06,
      "loss": 0.8571,
      "step": 6137
    },
    {
      "epoch": 0.33782816885904565,
      "grad_norm": 0.7167731523513794,
      "learning_rate": 9.33048819590169e-06,
      "loss": 0.7597,
      "step": 6138
    },
    {
      "epoch": 0.3378832076614013,
      "grad_norm": 0.7761037945747375,
      "learning_rate": 9.33027150056244e-06,
      "loss": 0.8112,
      "step": 6139
    },
    {
      "epoch": 0.337938246463757,
      "grad_norm": 0.8143900632858276,
      "learning_rate": 9.330054772678028e-06,
      "loss": 0.8213,
      "step": 6140
    },
    {
      "epoch": 0.3379932852661126,
      "grad_norm": 0.7181026339530945,
      "learning_rate": 9.329838012250083e-06,
      "loss": 0.8228,
      "step": 6141
    },
    {
      "epoch": 0.3380483240684683,
      "grad_norm": 0.7229815721511841,
      "learning_rate": 9.329621219280235e-06,
      "loss": 0.8205,
      "step": 6142
    },
    {
      "epoch": 0.33810336287082393,
      "grad_norm": 0.7120887637138367,
      "learning_rate": 9.329404393770113e-06,
      "loss": 0.8012,
      "step": 6143
    },
    {
      "epoch": 0.33815840167317956,
      "grad_norm": 0.7859634757041931,
      "learning_rate": 9.329187535721346e-06,
      "loss": 0.7583,
      "step": 6144
    },
    {
      "epoch": 0.33821344047553525,
      "grad_norm": 0.7630401253700256,
      "learning_rate": 9.328970645135564e-06,
      "loss": 0.9087,
      "step": 6145
    },
    {
      "epoch": 0.3382684792778909,
      "grad_norm": 0.7028466463088989,
      "learning_rate": 9.328753722014399e-06,
      "loss": 0.7253,
      "step": 6146
    },
    {
      "epoch": 0.3383235180802466,
      "grad_norm": 0.8910240530967712,
      "learning_rate": 9.328536766359477e-06,
      "loss": 0.9048,
      "step": 6147
    },
    {
      "epoch": 0.3383785568826022,
      "grad_norm": 0.6695914268493652,
      "learning_rate": 9.328319778172435e-06,
      "loss": 0.6817,
      "step": 6148
    },
    {
      "epoch": 0.3384335956849579,
      "grad_norm": 0.9667700529098511,
      "learning_rate": 9.328102757454898e-06,
      "loss": 0.7721,
      "step": 6149
    },
    {
      "epoch": 0.33848863448731353,
      "grad_norm": 0.7267603874206543,
      "learning_rate": 9.3278857042085e-06,
      "loss": 0.7263,
      "step": 6150
    },
    {
      "epoch": 0.3385436732896692,
      "grad_norm": 0.7603437900543213,
      "learning_rate": 9.32766861843487e-06,
      "loss": 0.7856,
      "step": 6151
    },
    {
      "epoch": 0.33859871209202486,
      "grad_norm": 0.7355918288230896,
      "learning_rate": 9.327451500135641e-06,
      "loss": 0.7687,
      "step": 6152
    },
    {
      "epoch": 0.33865375089438055,
      "grad_norm": 0.712210476398468,
      "learning_rate": 9.327234349312446e-06,
      "loss": 0.7689,
      "step": 6153
    },
    {
      "epoch": 0.3387087896967362,
      "grad_norm": 0.9011964797973633,
      "learning_rate": 9.327017165966916e-06,
      "loss": 0.888,
      "step": 6154
    },
    {
      "epoch": 0.33876382849909187,
      "grad_norm": 0.7334766387939453,
      "learning_rate": 9.326799950100683e-06,
      "loss": 0.7577,
      "step": 6155
    },
    {
      "epoch": 0.3388188673014475,
      "grad_norm": 0.711370587348938,
      "learning_rate": 9.32658270171538e-06,
      "loss": 0.7653,
      "step": 6156
    },
    {
      "epoch": 0.3388739061038032,
      "grad_norm": 0.8465714454650879,
      "learning_rate": 9.32636542081264e-06,
      "loss": 0.7252,
      "step": 6157
    },
    {
      "epoch": 0.3389289449061588,
      "grad_norm": 0.8105099201202393,
      "learning_rate": 9.326148107394094e-06,
      "loss": 0.7886,
      "step": 6158
    },
    {
      "epoch": 0.3389839837085145,
      "grad_norm": 0.8082063794136047,
      "learning_rate": 9.32593076146138e-06,
      "loss": 0.8968,
      "step": 6159
    },
    {
      "epoch": 0.33903902251087015,
      "grad_norm": 0.7451661229133606,
      "learning_rate": 9.325713383016125e-06,
      "loss": 0.762,
      "step": 6160
    },
    {
      "epoch": 0.33909406131322584,
      "grad_norm": 0.8174484372138977,
      "learning_rate": 9.325495972059968e-06,
      "loss": 0.8285,
      "step": 6161
    },
    {
      "epoch": 0.33914910011558147,
      "grad_norm": 0.7690935134887695,
      "learning_rate": 9.32527852859454e-06,
      "loss": 0.8908,
      "step": 6162
    },
    {
      "epoch": 0.33920413891793716,
      "grad_norm": 0.7730095386505127,
      "learning_rate": 9.325061052621476e-06,
      "loss": 0.8571,
      "step": 6163
    },
    {
      "epoch": 0.3392591777202928,
      "grad_norm": 0.7750043869018555,
      "learning_rate": 9.324843544142412e-06,
      "loss": 0.8314,
      "step": 6164
    },
    {
      "epoch": 0.3393142165226485,
      "grad_norm": 0.8184822797775269,
      "learning_rate": 9.32462600315898e-06,
      "loss": 0.8783,
      "step": 6165
    },
    {
      "epoch": 0.3393692553250041,
      "grad_norm": 0.8553629517555237,
      "learning_rate": 9.32440842967282e-06,
      "loss": 0.7116,
      "step": 6166
    },
    {
      "epoch": 0.3394242941273598,
      "grad_norm": 0.8072115778923035,
      "learning_rate": 9.324190823685562e-06,
      "loss": 0.7498,
      "step": 6167
    },
    {
      "epoch": 0.33947933292971544,
      "grad_norm": 0.7787594795227051,
      "learning_rate": 9.323973185198843e-06,
      "loss": 0.7567,
      "step": 6168
    },
    {
      "epoch": 0.3395343717320711,
      "grad_norm": 0.7571421265602112,
      "learning_rate": 9.323755514214299e-06,
      "loss": 0.8349,
      "step": 6169
    },
    {
      "epoch": 0.33958941053442676,
      "grad_norm": 0.6768494248390198,
      "learning_rate": 9.323537810733565e-06,
      "loss": 0.7382,
      "step": 6170
    },
    {
      "epoch": 0.33964444933678245,
      "grad_norm": 0.7091678380966187,
      "learning_rate": 9.32332007475828e-06,
      "loss": 0.8107,
      "step": 6171
    },
    {
      "epoch": 0.3396994881391381,
      "grad_norm": 0.6896559596061707,
      "learning_rate": 9.323102306290078e-06,
      "loss": 0.7973,
      "step": 6172
    },
    {
      "epoch": 0.3397545269414938,
      "grad_norm": 0.7383756637573242,
      "learning_rate": 9.322884505330595e-06,
      "loss": 0.7998,
      "step": 6173
    },
    {
      "epoch": 0.3398095657438494,
      "grad_norm": 0.7487883567810059,
      "learning_rate": 9.32266667188147e-06,
      "loss": 0.7928,
      "step": 6174
    },
    {
      "epoch": 0.3398646045462051,
      "grad_norm": 0.7935298681259155,
      "learning_rate": 9.32244880594434e-06,
      "loss": 0.8457,
      "step": 6175
    },
    {
      "epoch": 0.33991964334856073,
      "grad_norm": 0.6571856737136841,
      "learning_rate": 9.322230907520841e-06,
      "loss": 0.7177,
      "step": 6176
    },
    {
      "epoch": 0.3399746821509164,
      "grad_norm": 0.7694165706634521,
      "learning_rate": 9.322012976612613e-06,
      "loss": 0.7124,
      "step": 6177
    },
    {
      "epoch": 0.34002972095327205,
      "grad_norm": 0.8665503263473511,
      "learning_rate": 9.32179501322129e-06,
      "loss": 0.8054,
      "step": 6178
    },
    {
      "epoch": 0.34008475975562774,
      "grad_norm": 0.6794337034225464,
      "learning_rate": 9.321577017348515e-06,
      "loss": 0.6468,
      "step": 6179
    },
    {
      "epoch": 0.3401397985579834,
      "grad_norm": 0.7875672578811646,
      "learning_rate": 9.32135898899592e-06,
      "loss": 0.8384,
      "step": 6180
    },
    {
      "epoch": 0.34019483736033906,
      "grad_norm": 0.8050880432128906,
      "learning_rate": 9.321140928165152e-06,
      "loss": 0.7261,
      "step": 6181
    },
    {
      "epoch": 0.3402498761626947,
      "grad_norm": 0.7489742040634155,
      "learning_rate": 9.320922834857844e-06,
      "loss": 0.8252,
      "step": 6182
    },
    {
      "epoch": 0.3403049149650504,
      "grad_norm": 0.7785589098930359,
      "learning_rate": 9.320704709075637e-06,
      "loss": 0.7123,
      "step": 6183
    },
    {
      "epoch": 0.340359953767406,
      "grad_norm": 0.7698208689689636,
      "learning_rate": 9.320486550820169e-06,
      "loss": 0.704,
      "step": 6184
    },
    {
      "epoch": 0.3404149925697617,
      "grad_norm": 0.78490149974823,
      "learning_rate": 9.320268360093081e-06,
      "loss": 0.8446,
      "step": 6185
    },
    {
      "epoch": 0.34047003137211734,
      "grad_norm": 0.6684672236442566,
      "learning_rate": 9.320050136896012e-06,
      "loss": 0.6728,
      "step": 6186
    },
    {
      "epoch": 0.340525070174473,
      "grad_norm": 0.818122386932373,
      "learning_rate": 9.319831881230603e-06,
      "loss": 0.7744,
      "step": 6187
    },
    {
      "epoch": 0.34058010897682867,
      "grad_norm": 0.83867347240448,
      "learning_rate": 9.319613593098494e-06,
      "loss": 0.7423,
      "step": 6188
    },
    {
      "epoch": 0.3406351477791843,
      "grad_norm": 0.7800338268280029,
      "learning_rate": 9.319395272501326e-06,
      "loss": 0.8189,
      "step": 6189
    },
    {
      "epoch": 0.34069018658154,
      "grad_norm": 0.7530137300491333,
      "learning_rate": 9.319176919440737e-06,
      "loss": 0.7978,
      "step": 6190
    },
    {
      "epoch": 0.3407452253838956,
      "grad_norm": 0.8916274309158325,
      "learning_rate": 9.318958533918374e-06,
      "loss": 0.8828,
      "step": 6191
    },
    {
      "epoch": 0.3408002641862513,
      "grad_norm": 0.76950603723526,
      "learning_rate": 9.318740115935873e-06,
      "loss": 0.7691,
      "step": 6192
    },
    {
      "epoch": 0.34085530298860695,
      "grad_norm": 0.8348222970962524,
      "learning_rate": 9.318521665494877e-06,
      "loss": 0.8022,
      "step": 6193
    },
    {
      "epoch": 0.34091034179096263,
      "grad_norm": 0.6879388689994812,
      "learning_rate": 9.318303182597029e-06,
      "loss": 0.747,
      "step": 6194
    },
    {
      "epoch": 0.34096538059331827,
      "grad_norm": 0.8032572269439697,
      "learning_rate": 9.31808466724397e-06,
      "loss": 0.7621,
      "step": 6195
    },
    {
      "epoch": 0.34102041939567396,
      "grad_norm": 0.6842368841171265,
      "learning_rate": 9.317866119437342e-06,
      "loss": 0.6867,
      "step": 6196
    },
    {
      "epoch": 0.3410754581980296,
      "grad_norm": 0.7797672152519226,
      "learning_rate": 9.317647539178788e-06,
      "loss": 0.8329,
      "step": 6197
    },
    {
      "epoch": 0.3411304970003853,
      "grad_norm": 0.6865420341491699,
      "learning_rate": 9.317428926469952e-06,
      "loss": 0.7544,
      "step": 6198
    },
    {
      "epoch": 0.3411855358027409,
      "grad_norm": 0.818217396736145,
      "learning_rate": 9.317210281312475e-06,
      "loss": 0.8853,
      "step": 6199
    },
    {
      "epoch": 0.3412405746050966,
      "grad_norm": 0.7531415224075317,
      "learning_rate": 9.316991603708001e-06,
      "loss": 0.8225,
      "step": 6200
    },
    {
      "epoch": 0.34129561340745224,
      "grad_norm": 0.7347036600112915,
      "learning_rate": 9.316772893658173e-06,
      "loss": 0.7817,
      "step": 6201
    },
    {
      "epoch": 0.3413506522098079,
      "grad_norm": 0.7162033915519714,
      "learning_rate": 9.316554151164636e-06,
      "loss": 0.7836,
      "step": 6202
    },
    {
      "epoch": 0.34140569101216356,
      "grad_norm": 0.7421988248825073,
      "learning_rate": 9.316335376229035e-06,
      "loss": 0.7782,
      "step": 6203
    },
    {
      "epoch": 0.34146072981451925,
      "grad_norm": 0.7672573328018188,
      "learning_rate": 9.31611656885301e-06,
      "loss": 0.8585,
      "step": 6204
    },
    {
      "epoch": 0.3415157686168749,
      "grad_norm": 0.6898330450057983,
      "learning_rate": 9.31589772903821e-06,
      "loss": 0.7719,
      "step": 6205
    },
    {
      "epoch": 0.34157080741923057,
      "grad_norm": 0.7700635194778442,
      "learning_rate": 9.315678856786279e-06,
      "loss": 0.7345,
      "step": 6206
    },
    {
      "epoch": 0.3416258462215862,
      "grad_norm": 0.6982038617134094,
      "learning_rate": 9.315459952098858e-06,
      "loss": 0.8332,
      "step": 6207
    },
    {
      "epoch": 0.3416808850239419,
      "grad_norm": 0.8882858753204346,
      "learning_rate": 9.315241014977598e-06,
      "loss": 0.9029,
      "step": 6208
    },
    {
      "epoch": 0.3417359238262975,
      "grad_norm": 0.7313854098320007,
      "learning_rate": 9.31502204542414e-06,
      "loss": 0.8061,
      "step": 6209
    },
    {
      "epoch": 0.3417909626286532,
      "grad_norm": 0.7324157953262329,
      "learning_rate": 9.314803043440131e-06,
      "loss": 0.7889,
      "step": 6210
    },
    {
      "epoch": 0.34184600143100885,
      "grad_norm": 0.7498225569725037,
      "learning_rate": 9.314584009027218e-06,
      "loss": 0.7937,
      "step": 6211
    },
    {
      "epoch": 0.34190104023336454,
      "grad_norm": 0.7093212008476257,
      "learning_rate": 9.314364942187048e-06,
      "loss": 0.8404,
      "step": 6212
    },
    {
      "epoch": 0.3419560790357202,
      "grad_norm": 0.7008668780326843,
      "learning_rate": 9.314145842921264e-06,
      "loss": 0.8175,
      "step": 6213
    },
    {
      "epoch": 0.34201111783807586,
      "grad_norm": 0.8049909472465515,
      "learning_rate": 9.313926711231516e-06,
      "loss": 0.78,
      "step": 6214
    },
    {
      "epoch": 0.3420661566404315,
      "grad_norm": 0.7777613997459412,
      "learning_rate": 9.313707547119448e-06,
      "loss": 0.9566,
      "step": 6215
    },
    {
      "epoch": 0.3421211954427872,
      "grad_norm": 0.7787579894065857,
      "learning_rate": 9.31348835058671e-06,
      "loss": 0.7698,
      "step": 6216
    },
    {
      "epoch": 0.3421762342451428,
      "grad_norm": 0.7779031991958618,
      "learning_rate": 9.313269121634947e-06,
      "loss": 0.8853,
      "step": 6217
    },
    {
      "epoch": 0.3422312730474985,
      "grad_norm": 0.7194382548332214,
      "learning_rate": 9.313049860265809e-06,
      "loss": 0.8399,
      "step": 6218
    },
    {
      "epoch": 0.34228631184985414,
      "grad_norm": 0.6513093709945679,
      "learning_rate": 9.312830566480943e-06,
      "loss": 0.7156,
      "step": 6219
    },
    {
      "epoch": 0.34234135065220983,
      "grad_norm": 0.935325026512146,
      "learning_rate": 9.312611240281996e-06,
      "loss": 0.7525,
      "step": 6220
    },
    {
      "epoch": 0.34239638945456546,
      "grad_norm": 0.7539558410644531,
      "learning_rate": 9.312391881670618e-06,
      "loss": 0.7716,
      "step": 6221
    },
    {
      "epoch": 0.34245142825692115,
      "grad_norm": 0.7239616513252258,
      "learning_rate": 9.312172490648457e-06,
      "loss": 0.7272,
      "step": 6222
    },
    {
      "epoch": 0.3425064670592768,
      "grad_norm": 0.7742316126823425,
      "learning_rate": 9.311953067217162e-06,
      "loss": 0.7657,
      "step": 6223
    },
    {
      "epoch": 0.3425615058616325,
      "grad_norm": 0.782691240310669,
      "learning_rate": 9.311733611378379e-06,
      "loss": 0.813,
      "step": 6224
    },
    {
      "epoch": 0.3426165446639881,
      "grad_norm": 0.7448118329048157,
      "learning_rate": 9.311514123133765e-06,
      "loss": 0.8298,
      "step": 6225
    },
    {
      "epoch": 0.3426715834663438,
      "grad_norm": 0.8201695680618286,
      "learning_rate": 9.311294602484961e-06,
      "loss": 0.7738,
      "step": 6226
    },
    {
      "epoch": 0.34272662226869943,
      "grad_norm": 0.6928383111953735,
      "learning_rate": 9.311075049433625e-06,
      "loss": 0.6829,
      "step": 6227
    },
    {
      "epoch": 0.3427816610710551,
      "grad_norm": 0.7509302496910095,
      "learning_rate": 9.310855463981399e-06,
      "loss": 0.6265,
      "step": 6228
    },
    {
      "epoch": 0.34283669987341076,
      "grad_norm": 0.7012569308280945,
      "learning_rate": 9.310635846129938e-06,
      "loss": 0.7478,
      "step": 6229
    },
    {
      "epoch": 0.3428917386757664,
      "grad_norm": 0.7428532242774963,
      "learning_rate": 9.310416195880894e-06,
      "loss": 0.7434,
      "step": 6230
    },
    {
      "epoch": 0.3429467774781221,
      "grad_norm": 0.9089111685752869,
      "learning_rate": 9.310196513235915e-06,
      "loss": 0.6991,
      "step": 6231
    },
    {
      "epoch": 0.3430018162804777,
      "grad_norm": 0.7633285522460938,
      "learning_rate": 9.309976798196651e-06,
      "loss": 0.7789,
      "step": 6232
    },
    {
      "epoch": 0.3430568550828334,
      "grad_norm": 0.7035595178604126,
      "learning_rate": 9.309757050764756e-06,
      "loss": 0.6784,
      "step": 6233
    },
    {
      "epoch": 0.34311189388518903,
      "grad_norm": 0.8782615661621094,
      "learning_rate": 9.309537270941881e-06,
      "loss": 0.8861,
      "step": 6234
    },
    {
      "epoch": 0.3431669326875447,
      "grad_norm": 0.7690381407737732,
      "learning_rate": 9.309317458729677e-06,
      "loss": 0.7701,
      "step": 6235
    },
    {
      "epoch": 0.34322197148990036,
      "grad_norm": 0.7730939388275146,
      "learning_rate": 9.309097614129797e-06,
      "loss": 0.8004,
      "step": 6236
    },
    {
      "epoch": 0.34327701029225605,
      "grad_norm": 0.9295101761817932,
      "learning_rate": 9.308877737143894e-06,
      "loss": 0.6964,
      "step": 6237
    },
    {
      "epoch": 0.3433320490946117,
      "grad_norm": 0.7496231198310852,
      "learning_rate": 9.308657827773617e-06,
      "loss": 0.8107,
      "step": 6238
    },
    {
      "epoch": 0.34338708789696737,
      "grad_norm": 0.7656146287918091,
      "learning_rate": 9.308437886020622e-06,
      "loss": 0.8016,
      "step": 6239
    },
    {
      "epoch": 0.343442126699323,
      "grad_norm": 0.8925992846488953,
      "learning_rate": 9.308217911886562e-06,
      "loss": 0.7136,
      "step": 6240
    },
    {
      "epoch": 0.3434971655016787,
      "grad_norm": 0.7669470906257629,
      "learning_rate": 9.307997905373087e-06,
      "loss": 0.8284,
      "step": 6241
    },
    {
      "epoch": 0.3435522043040343,
      "grad_norm": 0.6964572072029114,
      "learning_rate": 9.307777866481855e-06,
      "loss": 0.7926,
      "step": 6242
    },
    {
      "epoch": 0.34360724310639,
      "grad_norm": 0.8405120968818665,
      "learning_rate": 9.307557795214517e-06,
      "loss": 0.9398,
      "step": 6243
    },
    {
      "epoch": 0.34366228190874565,
      "grad_norm": 0.7517451643943787,
      "learning_rate": 9.30733769157273e-06,
      "loss": 0.8315,
      "step": 6244
    },
    {
      "epoch": 0.34371732071110134,
      "grad_norm": 0.7740843892097473,
      "learning_rate": 9.307117555558144e-06,
      "loss": 0.8287,
      "step": 6245
    },
    {
      "epoch": 0.34377235951345697,
      "grad_norm": 0.7214275598526001,
      "learning_rate": 9.306897387172413e-06,
      "loss": 0.7416,
      "step": 6246
    },
    {
      "epoch": 0.34382739831581266,
      "grad_norm": 0.8217877745628357,
      "learning_rate": 9.306677186417197e-06,
      "loss": 0.8365,
      "step": 6247
    },
    {
      "epoch": 0.3438824371181683,
      "grad_norm": 0.7397332191467285,
      "learning_rate": 9.306456953294148e-06,
      "loss": 0.7284,
      "step": 6248
    },
    {
      "epoch": 0.343937475920524,
      "grad_norm": 0.8141350746154785,
      "learning_rate": 9.30623668780492e-06,
      "loss": 0.8976,
      "step": 6249
    },
    {
      "epoch": 0.3439925147228796,
      "grad_norm": 0.7078670263290405,
      "learning_rate": 9.306016389951171e-06,
      "loss": 0.8167,
      "step": 6250
    },
    {
      "epoch": 0.3440475535252353,
      "grad_norm": 0.7136256098747253,
      "learning_rate": 9.305796059734553e-06,
      "loss": 0.7916,
      "step": 6251
    },
    {
      "epoch": 0.34410259232759094,
      "grad_norm": 1.6186310052871704,
      "learning_rate": 9.305575697156726e-06,
      "loss": 0.8148,
      "step": 6252
    },
    {
      "epoch": 0.34415763112994663,
      "grad_norm": 0.7567281126976013,
      "learning_rate": 9.305355302219346e-06,
      "loss": 0.8676,
      "step": 6253
    },
    {
      "epoch": 0.34421266993230226,
      "grad_norm": 0.9036027193069458,
      "learning_rate": 9.305134874924067e-06,
      "loss": 0.8111,
      "step": 6254
    },
    {
      "epoch": 0.34426770873465795,
      "grad_norm": 0.9375718235969543,
      "learning_rate": 9.304914415272547e-06,
      "loss": 0.6176,
      "step": 6255
    },
    {
      "epoch": 0.3443227475370136,
      "grad_norm": 0.7309718132019043,
      "learning_rate": 9.304693923266441e-06,
      "loss": 0.7313,
      "step": 6256
    },
    {
      "epoch": 0.3443777863393693,
      "grad_norm": 0.7499229311943054,
      "learning_rate": 9.30447339890741e-06,
      "loss": 0.6704,
      "step": 6257
    },
    {
      "epoch": 0.3444328251417249,
      "grad_norm": 0.7553356289863586,
      "learning_rate": 9.304252842197108e-06,
      "loss": 0.8671,
      "step": 6258
    },
    {
      "epoch": 0.3444878639440806,
      "grad_norm": 0.7144323587417603,
      "learning_rate": 9.304032253137194e-06,
      "loss": 0.7684,
      "step": 6259
    },
    {
      "epoch": 0.34454290274643623,
      "grad_norm": 0.7566905617713928,
      "learning_rate": 9.303811631729324e-06,
      "loss": 0.8381,
      "step": 6260
    },
    {
      "epoch": 0.3445979415487919,
      "grad_norm": 0.7300242185592651,
      "learning_rate": 9.30359097797516e-06,
      "loss": 0.7044,
      "step": 6261
    },
    {
      "epoch": 0.34465298035114755,
      "grad_norm": 0.6504725813865662,
      "learning_rate": 9.303370291876359e-06,
      "loss": 0.6693,
      "step": 6262
    },
    {
      "epoch": 0.34470801915350324,
      "grad_norm": 0.7010672688484192,
      "learning_rate": 9.303149573434576e-06,
      "loss": 0.6635,
      "step": 6263
    },
    {
      "epoch": 0.3447630579558589,
      "grad_norm": 0.8416483998298645,
      "learning_rate": 9.302928822651473e-06,
      "loss": 0.8408,
      "step": 6264
    },
    {
      "epoch": 0.34481809675821457,
      "grad_norm": 0.7011786699295044,
      "learning_rate": 9.302708039528712e-06,
      "loss": 0.7636,
      "step": 6265
    },
    {
      "epoch": 0.3448731355605702,
      "grad_norm": 0.7361586689949036,
      "learning_rate": 9.302487224067947e-06,
      "loss": 0.824,
      "step": 6266
    },
    {
      "epoch": 0.3449281743629259,
      "grad_norm": 0.7747073173522949,
      "learning_rate": 9.302266376270839e-06,
      "loss": 0.8012,
      "step": 6267
    },
    {
      "epoch": 0.3449832131652815,
      "grad_norm": 0.9407958388328552,
      "learning_rate": 9.302045496139049e-06,
      "loss": 0.8664,
      "step": 6268
    },
    {
      "epoch": 0.3450382519676372,
      "grad_norm": 0.8674719929695129,
      "learning_rate": 9.301824583674238e-06,
      "loss": 0.8842,
      "step": 6269
    },
    {
      "epoch": 0.34509329076999284,
      "grad_norm": 0.7697336673736572,
      "learning_rate": 9.301603638878062e-06,
      "loss": 0.7148,
      "step": 6270
    },
    {
      "epoch": 0.34514832957234853,
      "grad_norm": 0.7220168709754944,
      "learning_rate": 9.301382661752187e-06,
      "loss": 0.7199,
      "step": 6271
    },
    {
      "epoch": 0.34520336837470417,
      "grad_norm": 0.6745235919952393,
      "learning_rate": 9.301161652298272e-06,
      "loss": 0.708,
      "step": 6272
    },
    {
      "epoch": 0.3452584071770598,
      "grad_norm": 0.7062309980392456,
      "learning_rate": 9.300940610517974e-06,
      "loss": 0.863,
      "step": 6273
    },
    {
      "epoch": 0.3453134459794155,
      "grad_norm": 0.7499971985816956,
      "learning_rate": 9.300719536412961e-06,
      "loss": 0.7976,
      "step": 6274
    },
    {
      "epoch": 0.3453684847817711,
      "grad_norm": 0.8304464221000671,
      "learning_rate": 9.30049842998489e-06,
      "loss": 0.8689,
      "step": 6275
    },
    {
      "epoch": 0.3454235235841268,
      "grad_norm": 0.7460494041442871,
      "learning_rate": 9.300277291235423e-06,
      "loss": 0.7499,
      "step": 6276
    },
    {
      "epoch": 0.34547856238648245,
      "grad_norm": 0.758788526058197,
      "learning_rate": 9.300056120166225e-06,
      "loss": 0.7501,
      "step": 6277
    },
    {
      "epoch": 0.34553360118883814,
      "grad_norm": 0.7204456925392151,
      "learning_rate": 9.299834916778955e-06,
      "loss": 0.8234,
      "step": 6278
    },
    {
      "epoch": 0.34558863999119377,
      "grad_norm": 0.7647501826286316,
      "learning_rate": 9.299613681075277e-06,
      "loss": 0.8653,
      "step": 6279
    },
    {
      "epoch": 0.34564367879354946,
      "grad_norm": 0.7543594837188721,
      "learning_rate": 9.299392413056853e-06,
      "loss": 0.7915,
      "step": 6280
    },
    {
      "epoch": 0.3456987175959051,
      "grad_norm": 0.7691700458526611,
      "learning_rate": 9.299171112725347e-06,
      "loss": 0.7429,
      "step": 6281
    },
    {
      "epoch": 0.3457537563982608,
      "grad_norm": 0.7703940272331238,
      "learning_rate": 9.29894978008242e-06,
      "loss": 0.7424,
      "step": 6282
    },
    {
      "epoch": 0.3458087952006164,
      "grad_norm": 0.8482547402381897,
      "learning_rate": 9.29872841512974e-06,
      "loss": 0.8971,
      "step": 6283
    },
    {
      "epoch": 0.3458638340029721,
      "grad_norm": 0.755224883556366,
      "learning_rate": 9.298507017868966e-06,
      "loss": 0.7984,
      "step": 6284
    },
    {
      "epoch": 0.34591887280532774,
      "grad_norm": 1.079891324043274,
      "learning_rate": 9.298285588301766e-06,
      "loss": 0.8301,
      "step": 6285
    },
    {
      "epoch": 0.3459739116076834,
      "grad_norm": 0.7357321381568909,
      "learning_rate": 9.2980641264298e-06,
      "loss": 0.9018,
      "step": 6286
    },
    {
      "epoch": 0.34602895041003906,
      "grad_norm": 0.7541963458061218,
      "learning_rate": 9.297842632254734e-06,
      "loss": 0.8716,
      "step": 6287
    },
    {
      "epoch": 0.34608398921239475,
      "grad_norm": 1.1570138931274414,
      "learning_rate": 9.297621105778235e-06,
      "loss": 0.9163,
      "step": 6288
    },
    {
      "epoch": 0.3461390280147504,
      "grad_norm": 0.7626895904541016,
      "learning_rate": 9.297399547001965e-06,
      "loss": 0.8162,
      "step": 6289
    },
    {
      "epoch": 0.34619406681710607,
      "grad_norm": 0.758469820022583,
      "learning_rate": 9.297177955927593e-06,
      "loss": 0.8966,
      "step": 6290
    },
    {
      "epoch": 0.3462491056194617,
      "grad_norm": 0.8998799324035645,
      "learning_rate": 9.296956332556779e-06,
      "loss": 0.8127,
      "step": 6291
    },
    {
      "epoch": 0.3463041444218174,
      "grad_norm": 0.7470666170120239,
      "learning_rate": 9.29673467689119e-06,
      "loss": 0.7738,
      "step": 6292
    },
    {
      "epoch": 0.34635918322417303,
      "grad_norm": 0.8066977858543396,
      "learning_rate": 9.296512988932497e-06,
      "loss": 0.8958,
      "step": 6293
    },
    {
      "epoch": 0.3464142220265287,
      "grad_norm": 0.8394894003868103,
      "learning_rate": 9.29629126868236e-06,
      "loss": 0.8023,
      "step": 6294
    },
    {
      "epoch": 0.34646926082888435,
      "grad_norm": 0.9053472876548767,
      "learning_rate": 9.29606951614245e-06,
      "loss": 0.8244,
      "step": 6295
    },
    {
      "epoch": 0.34652429963124004,
      "grad_norm": 0.6996710896492004,
      "learning_rate": 9.295847731314428e-06,
      "loss": 0.8203,
      "step": 6296
    },
    {
      "epoch": 0.3465793384335957,
      "grad_norm": 0.7236999273300171,
      "learning_rate": 9.295625914199968e-06,
      "loss": 0.6982,
      "step": 6297
    },
    {
      "epoch": 0.34663437723595136,
      "grad_norm": 0.7006070017814636,
      "learning_rate": 9.295404064800733e-06,
      "loss": 0.7881,
      "step": 6298
    },
    {
      "epoch": 0.346689416038307,
      "grad_norm": 0.8188902735710144,
      "learning_rate": 9.29518218311839e-06,
      "loss": 0.7472,
      "step": 6299
    },
    {
      "epoch": 0.3467444548406627,
      "grad_norm": 0.7708863019943237,
      "learning_rate": 9.294960269154608e-06,
      "loss": 0.7572,
      "step": 6300
    },
    {
      "epoch": 0.3467994936430183,
      "grad_norm": 0.7819802761077881,
      "learning_rate": 9.294738322911052e-06,
      "loss": 0.8486,
      "step": 6301
    },
    {
      "epoch": 0.346854532445374,
      "grad_norm": 0.7160501480102539,
      "learning_rate": 9.294516344389394e-06,
      "loss": 0.8104,
      "step": 6302
    },
    {
      "epoch": 0.34690957124772964,
      "grad_norm": 0.7426022887229919,
      "learning_rate": 9.294294333591302e-06,
      "loss": 0.7158,
      "step": 6303
    },
    {
      "epoch": 0.34696461005008533,
      "grad_norm": 0.8397019505500793,
      "learning_rate": 9.294072290518441e-06,
      "loss": 0.8466,
      "step": 6304
    },
    {
      "epoch": 0.34701964885244097,
      "grad_norm": 0.7220905423164368,
      "learning_rate": 9.293850215172483e-06,
      "loss": 0.7619,
      "step": 6305
    },
    {
      "epoch": 0.34707468765479665,
      "grad_norm": 0.7401862740516663,
      "learning_rate": 9.293628107555097e-06,
      "loss": 0.7873,
      "step": 6306
    },
    {
      "epoch": 0.3471297264571523,
      "grad_norm": 0.6764525175094604,
      "learning_rate": 9.29340596766795e-06,
      "loss": 0.7278,
      "step": 6307
    },
    {
      "epoch": 0.347184765259508,
      "grad_norm": 0.8553194403648376,
      "learning_rate": 9.293183795512715e-06,
      "loss": 0.9074,
      "step": 6308
    },
    {
      "epoch": 0.3472398040618636,
      "grad_norm": 0.6796454191207886,
      "learning_rate": 9.292961591091058e-06,
      "loss": 0.7179,
      "step": 6309
    },
    {
      "epoch": 0.3472948428642193,
      "grad_norm": 0.6075254082679749,
      "learning_rate": 9.292739354404652e-06,
      "loss": 0.7228,
      "step": 6310
    },
    {
      "epoch": 0.34734988166657493,
      "grad_norm": 0.7366840243339539,
      "learning_rate": 9.292517085455166e-06,
      "loss": 0.7934,
      "step": 6311
    },
    {
      "epoch": 0.3474049204689306,
      "grad_norm": 0.6820569038391113,
      "learning_rate": 9.29229478424427e-06,
      "loss": 0.7315,
      "step": 6312
    },
    {
      "epoch": 0.34745995927128626,
      "grad_norm": 0.8356956243515015,
      "learning_rate": 9.292072450773635e-06,
      "loss": 0.7787,
      "step": 6313
    },
    {
      "epoch": 0.34751499807364195,
      "grad_norm": 0.70506352186203,
      "learning_rate": 9.291850085044933e-06,
      "loss": 0.7411,
      "step": 6314
    },
    {
      "epoch": 0.3475700368759976,
      "grad_norm": 0.9074786901473999,
      "learning_rate": 9.291627687059835e-06,
      "loss": 0.7352,
      "step": 6315
    },
    {
      "epoch": 0.3476250756783532,
      "grad_norm": 0.7858747839927673,
      "learning_rate": 9.291405256820013e-06,
      "loss": 0.7816,
      "step": 6316
    },
    {
      "epoch": 0.3476801144807089,
      "grad_norm": 0.8576731085777283,
      "learning_rate": 9.291182794327134e-06,
      "loss": 0.7861,
      "step": 6317
    },
    {
      "epoch": 0.34773515328306454,
      "grad_norm": 0.7500558495521545,
      "learning_rate": 9.290960299582877e-06,
      "loss": 0.8028,
      "step": 6318
    },
    {
      "epoch": 0.3477901920854202,
      "grad_norm": 0.6577744483947754,
      "learning_rate": 9.29073777258891e-06,
      "loss": 0.7458,
      "step": 6319
    },
    {
      "epoch": 0.34784523088777586,
      "grad_norm": 0.742855429649353,
      "learning_rate": 9.290515213346906e-06,
      "loss": 0.755,
      "step": 6320
    },
    {
      "epoch": 0.34790026969013155,
      "grad_norm": 0.7626619338989258,
      "learning_rate": 9.290292621858542e-06,
      "loss": 0.6671,
      "step": 6321
    },
    {
      "epoch": 0.3479553084924872,
      "grad_norm": 0.7139305472373962,
      "learning_rate": 9.290069998125481e-06,
      "loss": 0.7981,
      "step": 6322
    },
    {
      "epoch": 0.34801034729484287,
      "grad_norm": 0.9249686002731323,
      "learning_rate": 9.289847342149407e-06,
      "loss": 0.7243,
      "step": 6323
    },
    {
      "epoch": 0.3480653860971985,
      "grad_norm": 0.8090649843215942,
      "learning_rate": 9.289624653931986e-06,
      "loss": 0.7892,
      "step": 6324
    },
    {
      "epoch": 0.3481204248995542,
      "grad_norm": 0.6845510005950928,
      "learning_rate": 9.289401933474895e-06,
      "loss": 0.7427,
      "step": 6325
    },
    {
      "epoch": 0.3481754637019098,
      "grad_norm": 0.7620648741722107,
      "learning_rate": 9.289179180779808e-06,
      "loss": 0.7715,
      "step": 6326
    },
    {
      "epoch": 0.3482305025042655,
      "grad_norm": 0.7441076040267944,
      "learning_rate": 9.288956395848398e-06,
      "loss": 0.7814,
      "step": 6327
    },
    {
      "epoch": 0.34828554130662115,
      "grad_norm": 0.6777048707008362,
      "learning_rate": 9.28873357868234e-06,
      "loss": 0.759,
      "step": 6328
    },
    {
      "epoch": 0.34834058010897684,
      "grad_norm": 0.6534250974655151,
      "learning_rate": 9.288510729283307e-06,
      "loss": 0.6777,
      "step": 6329
    },
    {
      "epoch": 0.34839561891133247,
      "grad_norm": 0.8205152153968811,
      "learning_rate": 9.288287847652977e-06,
      "loss": 0.8027,
      "step": 6330
    },
    {
      "epoch": 0.34845065771368816,
      "grad_norm": 0.7152554392814636,
      "learning_rate": 9.288064933793024e-06,
      "loss": 0.7956,
      "step": 6331
    },
    {
      "epoch": 0.3485056965160438,
      "grad_norm": 0.9816664457321167,
      "learning_rate": 9.287841987705121e-06,
      "loss": 0.828,
      "step": 6332
    },
    {
      "epoch": 0.3485607353183995,
      "grad_norm": 0.826554000377655,
      "learning_rate": 9.287619009390945e-06,
      "loss": 0.8544,
      "step": 6333
    },
    {
      "epoch": 0.3486157741207551,
      "grad_norm": 0.7255695462226868,
      "learning_rate": 9.287395998852175e-06,
      "loss": 0.7749,
      "step": 6334
    },
    {
      "epoch": 0.3486708129231108,
      "grad_norm": 0.7161709070205688,
      "learning_rate": 9.287172956090482e-06,
      "loss": 0.7114,
      "step": 6335
    },
    {
      "epoch": 0.34872585172546644,
      "grad_norm": 0.7219997644424438,
      "learning_rate": 9.286949881107546e-06,
      "loss": 0.8309,
      "step": 6336
    },
    {
      "epoch": 0.34878089052782213,
      "grad_norm": 0.7269770503044128,
      "learning_rate": 9.286726773905042e-06,
      "loss": 0.8039,
      "step": 6337
    },
    {
      "epoch": 0.34883592933017776,
      "grad_norm": 0.8142165541648865,
      "learning_rate": 9.286503634484645e-06,
      "loss": 0.7673,
      "step": 6338
    },
    {
      "epoch": 0.34889096813253345,
      "grad_norm": 0.7568639516830444,
      "learning_rate": 9.286280462848037e-06,
      "loss": 0.8471,
      "step": 6339
    },
    {
      "epoch": 0.3489460069348891,
      "grad_norm": 0.7927737236022949,
      "learning_rate": 9.28605725899689e-06,
      "loss": 0.8828,
      "step": 6340
    },
    {
      "epoch": 0.3490010457372448,
      "grad_norm": 0.9755893349647522,
      "learning_rate": 9.285834022932885e-06,
      "loss": 0.837,
      "step": 6341
    },
    {
      "epoch": 0.3490560845396004,
      "grad_norm": 0.6831560730934143,
      "learning_rate": 9.2856107546577e-06,
      "loss": 0.7169,
      "step": 6342
    },
    {
      "epoch": 0.3491111233419561,
      "grad_norm": 0.728239119052887,
      "learning_rate": 9.285387454173009e-06,
      "loss": 0.7805,
      "step": 6343
    },
    {
      "epoch": 0.34916616214431173,
      "grad_norm": 0.6979145407676697,
      "learning_rate": 9.285164121480495e-06,
      "loss": 0.7794,
      "step": 6344
    },
    {
      "epoch": 0.3492212009466674,
      "grad_norm": 0.7206674218177795,
      "learning_rate": 9.284940756581834e-06,
      "loss": 0.7198,
      "step": 6345
    },
    {
      "epoch": 0.34927623974902305,
      "grad_norm": 0.8156035542488098,
      "learning_rate": 9.284717359478705e-06,
      "loss": 0.884,
      "step": 6346
    },
    {
      "epoch": 0.34933127855137874,
      "grad_norm": 0.6876983046531677,
      "learning_rate": 9.284493930172788e-06,
      "loss": 0.7426,
      "step": 6347
    },
    {
      "epoch": 0.3493863173537344,
      "grad_norm": 0.6856677532196045,
      "learning_rate": 9.284270468665762e-06,
      "loss": 0.7085,
      "step": 6348
    },
    {
      "epoch": 0.34944135615609007,
      "grad_norm": 0.8378047943115234,
      "learning_rate": 9.284046974959304e-06,
      "loss": 0.725,
      "step": 6349
    },
    {
      "epoch": 0.3494963949584457,
      "grad_norm": 0.7410693764686584,
      "learning_rate": 9.283823449055097e-06,
      "loss": 0.7953,
      "step": 6350
    },
    {
      "epoch": 0.3495514337608014,
      "grad_norm": 0.7558375000953674,
      "learning_rate": 9.28359989095482e-06,
      "loss": 0.8052,
      "step": 6351
    },
    {
      "epoch": 0.349606472563157,
      "grad_norm": 0.7176862955093384,
      "learning_rate": 9.283376300660151e-06,
      "loss": 0.7077,
      "step": 6352
    },
    {
      "epoch": 0.3496615113655127,
      "grad_norm": 0.7443307042121887,
      "learning_rate": 9.283152678172774e-06,
      "loss": 0.7557,
      "step": 6353
    },
    {
      "epoch": 0.34971655016786835,
      "grad_norm": 0.6653748750686646,
      "learning_rate": 9.282929023494368e-06,
      "loss": 0.7558,
      "step": 6354
    },
    {
      "epoch": 0.34977158897022403,
      "grad_norm": 0.8139400482177734,
      "learning_rate": 9.282705336626615e-06,
      "loss": 0.847,
      "step": 6355
    },
    {
      "epoch": 0.34982662777257967,
      "grad_norm": 1.012450933456421,
      "learning_rate": 9.282481617571193e-06,
      "loss": 0.744,
      "step": 6356
    },
    {
      "epoch": 0.34988166657493536,
      "grad_norm": 0.7877402305603027,
      "learning_rate": 9.282257866329784e-06,
      "loss": 0.7475,
      "step": 6357
    },
    {
      "epoch": 0.349936705377291,
      "grad_norm": 0.7989935874938965,
      "learning_rate": 9.282034082904075e-06,
      "loss": 0.7379,
      "step": 6358
    },
    {
      "epoch": 0.3499917441796466,
      "grad_norm": 0.6665796637535095,
      "learning_rate": 9.281810267295741e-06,
      "loss": 0.7253,
      "step": 6359
    },
    {
      "epoch": 0.3500467829820023,
      "grad_norm": 0.8344665765762329,
      "learning_rate": 9.28158641950647e-06,
      "loss": 0.8095,
      "step": 6360
    },
    {
      "epoch": 0.35010182178435795,
      "grad_norm": 0.8312307596206665,
      "learning_rate": 9.281362539537939e-06,
      "loss": 0.8452,
      "step": 6361
    },
    {
      "epoch": 0.35015686058671364,
      "grad_norm": 0.7423825263977051,
      "learning_rate": 9.281138627391834e-06,
      "loss": 0.8291,
      "step": 6362
    },
    {
      "epoch": 0.35021189938906927,
      "grad_norm": 0.7594212293624878,
      "learning_rate": 9.280914683069837e-06,
      "loss": 0.8314,
      "step": 6363
    },
    {
      "epoch": 0.35026693819142496,
      "grad_norm": 0.8059762716293335,
      "learning_rate": 9.280690706573633e-06,
      "loss": 0.7695,
      "step": 6364
    },
    {
      "epoch": 0.3503219769937806,
      "grad_norm": 0.8053386807441711,
      "learning_rate": 9.280466697904902e-06,
      "loss": 0.8941,
      "step": 6365
    },
    {
      "epoch": 0.3503770157961363,
      "grad_norm": 0.6703817248344421,
      "learning_rate": 9.280242657065329e-06,
      "loss": 0.5978,
      "step": 6366
    },
    {
      "epoch": 0.3504320545984919,
      "grad_norm": 0.9359784722328186,
      "learning_rate": 9.280018584056598e-06,
      "loss": 0.8479,
      "step": 6367
    },
    {
      "epoch": 0.3504870934008476,
      "grad_norm": 0.7692418098449707,
      "learning_rate": 9.279794478880393e-06,
      "loss": 0.7254,
      "step": 6368
    },
    {
      "epoch": 0.35054213220320324,
      "grad_norm": 0.7992031574249268,
      "learning_rate": 9.279570341538397e-06,
      "loss": 0.6749,
      "step": 6369
    },
    {
      "epoch": 0.3505971710055589,
      "grad_norm": 0.7735288739204407,
      "learning_rate": 9.279346172032297e-06,
      "loss": 0.8545,
      "step": 6370
    },
    {
      "epoch": 0.35065220980791456,
      "grad_norm": 0.7124339938163757,
      "learning_rate": 9.279121970363778e-06,
      "loss": 0.8066,
      "step": 6371
    },
    {
      "epoch": 0.35070724861027025,
      "grad_norm": 0.8116535544395447,
      "learning_rate": 9.278897736534521e-06,
      "loss": 0.8197,
      "step": 6372
    },
    {
      "epoch": 0.3507622874126259,
      "grad_norm": 0.9377869963645935,
      "learning_rate": 9.278673470546217e-06,
      "loss": 0.74,
      "step": 6373
    },
    {
      "epoch": 0.3508173262149816,
      "grad_norm": 0.6726253628730774,
      "learning_rate": 9.278449172400548e-06,
      "loss": 0.6389,
      "step": 6374
    },
    {
      "epoch": 0.3508723650173372,
      "grad_norm": 0.8470593094825745,
      "learning_rate": 9.278224842099198e-06,
      "loss": 0.8059,
      "step": 6375
    },
    {
      "epoch": 0.3509274038196929,
      "grad_norm": 0.7041867971420288,
      "learning_rate": 9.278000479643857e-06,
      "loss": 0.7409,
      "step": 6376
    },
    {
      "epoch": 0.35098244262204853,
      "grad_norm": 0.7467322945594788,
      "learning_rate": 9.27777608503621e-06,
      "loss": 0.823,
      "step": 6377
    },
    {
      "epoch": 0.3510374814244042,
      "grad_norm": 0.7211065888404846,
      "learning_rate": 9.277551658277942e-06,
      "loss": 0.7655,
      "step": 6378
    },
    {
      "epoch": 0.35109252022675985,
      "grad_norm": 0.7709450125694275,
      "learning_rate": 9.27732719937074e-06,
      "loss": 0.8938,
      "step": 6379
    },
    {
      "epoch": 0.35114755902911554,
      "grad_norm": 0.7672929167747498,
      "learning_rate": 9.277102708316293e-06,
      "loss": 0.6814,
      "step": 6380
    },
    {
      "epoch": 0.3512025978314712,
      "grad_norm": 0.7334907650947571,
      "learning_rate": 9.276878185116287e-06,
      "loss": 0.6608,
      "step": 6381
    },
    {
      "epoch": 0.35125763663382686,
      "grad_norm": 0.7011460065841675,
      "learning_rate": 9.27665362977241e-06,
      "loss": 0.8196,
      "step": 6382
    },
    {
      "epoch": 0.3513126754361825,
      "grad_norm": 0.7388820052146912,
      "learning_rate": 9.276429042286349e-06,
      "loss": 0.8793,
      "step": 6383
    },
    {
      "epoch": 0.3513677142385382,
      "grad_norm": 0.809725821018219,
      "learning_rate": 9.27620442265979e-06,
      "loss": 0.6976,
      "step": 6384
    },
    {
      "epoch": 0.3514227530408938,
      "grad_norm": 0.6933012008666992,
      "learning_rate": 9.275979770894424e-06,
      "loss": 0.759,
      "step": 6385
    },
    {
      "epoch": 0.3514777918432495,
      "grad_norm": 0.7928480505943298,
      "learning_rate": 9.27575508699194e-06,
      "loss": 0.7462,
      "step": 6386
    },
    {
      "epoch": 0.35153283064560514,
      "grad_norm": 0.8461304903030396,
      "learning_rate": 9.275530370954024e-06,
      "loss": 0.8184,
      "step": 6387
    },
    {
      "epoch": 0.35158786944796083,
      "grad_norm": 0.7624425292015076,
      "learning_rate": 9.275305622782366e-06,
      "loss": 0.7913,
      "step": 6388
    },
    {
      "epoch": 0.35164290825031647,
      "grad_norm": 0.7103675007820129,
      "learning_rate": 9.275080842478657e-06,
      "loss": 0.7633,
      "step": 6389
    },
    {
      "epoch": 0.35169794705267216,
      "grad_norm": 0.9002664089202881,
      "learning_rate": 9.274856030044583e-06,
      "loss": 0.7643,
      "step": 6390
    },
    {
      "epoch": 0.3517529858550278,
      "grad_norm": 0.7658692002296448,
      "learning_rate": 9.274631185481836e-06,
      "loss": 0.8028,
      "step": 6391
    },
    {
      "epoch": 0.3518080246573835,
      "grad_norm": 0.6747875809669495,
      "learning_rate": 9.274406308792106e-06,
      "loss": 0.695,
      "step": 6392
    },
    {
      "epoch": 0.3518630634597391,
      "grad_norm": 0.8197165131568909,
      "learning_rate": 9.27418139997708e-06,
      "loss": 0.7218,
      "step": 6393
    },
    {
      "epoch": 0.3519181022620948,
      "grad_norm": 0.7597750425338745,
      "learning_rate": 9.273956459038453e-06,
      "loss": 0.7738,
      "step": 6394
    },
    {
      "epoch": 0.35197314106445043,
      "grad_norm": 0.7365928888320923,
      "learning_rate": 9.273731485977912e-06,
      "loss": 0.7906,
      "step": 6395
    },
    {
      "epoch": 0.3520281798668061,
      "grad_norm": 0.7313928604125977,
      "learning_rate": 9.273506480797151e-06,
      "loss": 0.834,
      "step": 6396
    },
    {
      "epoch": 0.35208321866916176,
      "grad_norm": 0.758886456489563,
      "learning_rate": 9.273281443497858e-06,
      "loss": 0.8883,
      "step": 6397
    },
    {
      "epoch": 0.35213825747151745,
      "grad_norm": 0.7318256497383118,
      "learning_rate": 9.273056374081726e-06,
      "loss": 0.7463,
      "step": 6398
    },
    {
      "epoch": 0.3521932962738731,
      "grad_norm": 0.778448224067688,
      "learning_rate": 9.272831272550446e-06,
      "loss": 0.6838,
      "step": 6399
    },
    {
      "epoch": 0.3522483350762287,
      "grad_norm": 0.7392274141311646,
      "learning_rate": 9.272606138905709e-06,
      "loss": 0.7237,
      "step": 6400
    },
    {
      "epoch": 0.3523033738785844,
      "grad_norm": 0.8803032040596008,
      "learning_rate": 9.272380973149209e-06,
      "loss": 0.7839,
      "step": 6401
    },
    {
      "epoch": 0.35235841268094004,
      "grad_norm": 0.7506754994392395,
      "learning_rate": 9.272155775282636e-06,
      "loss": 0.7665,
      "step": 6402
    },
    {
      "epoch": 0.3524134514832957,
      "grad_norm": 0.8136595487594604,
      "learning_rate": 9.271930545307686e-06,
      "loss": 0.9111,
      "step": 6403
    },
    {
      "epoch": 0.35246849028565136,
      "grad_norm": 0.7976880073547363,
      "learning_rate": 9.271705283226047e-06,
      "loss": 0.735,
      "step": 6404
    },
    {
      "epoch": 0.35252352908800705,
      "grad_norm": 0.89708411693573,
      "learning_rate": 9.271479989039415e-06,
      "loss": 0.7698,
      "step": 6405
    },
    {
      "epoch": 0.3525785678903627,
      "grad_norm": 0.8618703484535217,
      "learning_rate": 9.271254662749484e-06,
      "loss": 0.9001,
      "step": 6406
    },
    {
      "epoch": 0.35263360669271837,
      "grad_norm": 0.7143027186393738,
      "learning_rate": 9.271029304357946e-06,
      "loss": 0.8188,
      "step": 6407
    },
    {
      "epoch": 0.352688645495074,
      "grad_norm": 0.795365571975708,
      "learning_rate": 9.270803913866496e-06,
      "loss": 0.7389,
      "step": 6408
    },
    {
      "epoch": 0.3527436842974297,
      "grad_norm": 0.6947643756866455,
      "learning_rate": 9.270578491276825e-06,
      "loss": 0.7278,
      "step": 6409
    },
    {
      "epoch": 0.3527987230997853,
      "grad_norm": 0.7806137204170227,
      "learning_rate": 9.27035303659063e-06,
      "loss": 0.808,
      "step": 6410
    },
    {
      "epoch": 0.352853761902141,
      "grad_norm": 0.8908704519271851,
      "learning_rate": 9.270127549809606e-06,
      "loss": 0.8659,
      "step": 6411
    },
    {
      "epoch": 0.35290880070449665,
      "grad_norm": 0.8171417713165283,
      "learning_rate": 9.269902030935445e-06,
      "loss": 0.7918,
      "step": 6412
    },
    {
      "epoch": 0.35296383950685234,
      "grad_norm": 0.7556712627410889,
      "learning_rate": 9.269676479969842e-06,
      "loss": 0.7121,
      "step": 6413
    },
    {
      "epoch": 0.353018878309208,
      "grad_norm": 0.8080483675003052,
      "learning_rate": 9.269450896914495e-06,
      "loss": 0.8185,
      "step": 6414
    },
    {
      "epoch": 0.35307391711156366,
      "grad_norm": 0.8514583706855774,
      "learning_rate": 9.2692252817711e-06,
      "loss": 0.8055,
      "step": 6415
    },
    {
      "epoch": 0.3531289559139193,
      "grad_norm": 0.7914162278175354,
      "learning_rate": 9.268999634541347e-06,
      "loss": 0.759,
      "step": 6416
    },
    {
      "epoch": 0.353183994716275,
      "grad_norm": 0.6452118754386902,
      "learning_rate": 9.268773955226937e-06,
      "loss": 0.6797,
      "step": 6417
    },
    {
      "epoch": 0.3532390335186306,
      "grad_norm": 0.6876220107078552,
      "learning_rate": 9.268548243829565e-06,
      "loss": 0.7365,
      "step": 6418
    },
    {
      "epoch": 0.3532940723209863,
      "grad_norm": 0.758550226688385,
      "learning_rate": 9.268322500350926e-06,
      "loss": 0.7069,
      "step": 6419
    },
    {
      "epoch": 0.35334911112334194,
      "grad_norm": 0.7905879020690918,
      "learning_rate": 9.268096724792718e-06,
      "loss": 0.8024,
      "step": 6420
    },
    {
      "epoch": 0.35340414992569763,
      "grad_norm": 0.755253255367279,
      "learning_rate": 9.267870917156638e-06,
      "loss": 0.8018,
      "step": 6421
    },
    {
      "epoch": 0.35345918872805326,
      "grad_norm": 0.6879923343658447,
      "learning_rate": 9.267645077444382e-06,
      "loss": 0.7267,
      "step": 6422
    },
    {
      "epoch": 0.35351422753040895,
      "grad_norm": 0.766214907169342,
      "learning_rate": 9.267419205657649e-06,
      "loss": 0.7801,
      "step": 6423
    },
    {
      "epoch": 0.3535692663327646,
      "grad_norm": 0.868776798248291,
      "learning_rate": 9.267193301798135e-06,
      "loss": 0.9234,
      "step": 6424
    },
    {
      "epoch": 0.3536243051351203,
      "grad_norm": 1.2007492780685425,
      "learning_rate": 9.266967365867536e-06,
      "loss": 0.7743,
      "step": 6425
    },
    {
      "epoch": 0.3536793439374759,
      "grad_norm": 0.7445551156997681,
      "learning_rate": 9.266741397867556e-06,
      "loss": 0.6755,
      "step": 6426
    },
    {
      "epoch": 0.3537343827398316,
      "grad_norm": 0.7493785619735718,
      "learning_rate": 9.266515397799889e-06,
      "loss": 0.7891,
      "step": 6427
    },
    {
      "epoch": 0.35378942154218723,
      "grad_norm": 0.6718230843544006,
      "learning_rate": 9.266289365666234e-06,
      "loss": 0.6908,
      "step": 6428
    },
    {
      "epoch": 0.3538444603445429,
      "grad_norm": 0.7783547639846802,
      "learning_rate": 9.266063301468289e-06,
      "loss": 0.7115,
      "step": 6429
    },
    {
      "epoch": 0.35389949914689856,
      "grad_norm": 0.745627224445343,
      "learning_rate": 9.265837205207755e-06,
      "loss": 0.8421,
      "step": 6430
    },
    {
      "epoch": 0.35395453794925424,
      "grad_norm": 0.7314152717590332,
      "learning_rate": 9.26561107688633e-06,
      "loss": 0.807,
      "step": 6431
    },
    {
      "epoch": 0.3540095767516099,
      "grad_norm": 0.6975863575935364,
      "learning_rate": 9.265384916505714e-06,
      "loss": 0.7787,
      "step": 6432
    },
    {
      "epoch": 0.35406461555396557,
      "grad_norm": 0.9758319854736328,
      "learning_rate": 9.265158724067608e-06,
      "loss": 0.8668,
      "step": 6433
    },
    {
      "epoch": 0.3541196543563212,
      "grad_norm": 0.7686764001846313,
      "learning_rate": 9.264932499573711e-06,
      "loss": 0.7428,
      "step": 6434
    },
    {
      "epoch": 0.3541746931586769,
      "grad_norm": 0.8761935830116272,
      "learning_rate": 9.26470624302572e-06,
      "loss": 0.8022,
      "step": 6435
    },
    {
      "epoch": 0.3542297319610325,
      "grad_norm": 0.9145118594169617,
      "learning_rate": 9.264479954425341e-06,
      "loss": 0.7994,
      "step": 6436
    },
    {
      "epoch": 0.3542847707633882,
      "grad_norm": 0.8217951655387878,
      "learning_rate": 9.264253633774271e-06,
      "loss": 0.7235,
      "step": 6437
    },
    {
      "epoch": 0.35433980956574385,
      "grad_norm": 0.7624716758728027,
      "learning_rate": 9.264027281074214e-06,
      "loss": 0.8238,
      "step": 6438
    },
    {
      "epoch": 0.35439484836809954,
      "grad_norm": 0.7772085070610046,
      "learning_rate": 9.26380089632687e-06,
      "loss": 0.7941,
      "step": 6439
    },
    {
      "epoch": 0.35444988717045517,
      "grad_norm": 1.0462371110916138,
      "learning_rate": 9.263574479533937e-06,
      "loss": 0.8255,
      "step": 6440
    },
    {
      "epoch": 0.35450492597281086,
      "grad_norm": 0.8523101210594177,
      "learning_rate": 9.263348030697119e-06,
      "loss": 0.8489,
      "step": 6441
    },
    {
      "epoch": 0.3545599647751665,
      "grad_norm": 1.0292255878448486,
      "learning_rate": 9.26312154981812e-06,
      "loss": 0.7989,
      "step": 6442
    },
    {
      "epoch": 0.3546150035775221,
      "grad_norm": 0.7621143460273743,
      "learning_rate": 9.262895036898641e-06,
      "loss": 0.8154,
      "step": 6443
    },
    {
      "epoch": 0.3546700423798778,
      "grad_norm": 0.7158074378967285,
      "learning_rate": 9.262668491940382e-06,
      "loss": 0.7821,
      "step": 6444
    },
    {
      "epoch": 0.35472508118223345,
      "grad_norm": 0.7969478964805603,
      "learning_rate": 9.26244191494505e-06,
      "loss": 0.8535,
      "step": 6445
    },
    {
      "epoch": 0.35478011998458914,
      "grad_norm": 0.9244762063026428,
      "learning_rate": 9.262215305914345e-06,
      "loss": 0.7585,
      "step": 6446
    },
    {
      "epoch": 0.35483515878694477,
      "grad_norm": 0.6862454414367676,
      "learning_rate": 9.26198866484997e-06,
      "loss": 0.7294,
      "step": 6447
    },
    {
      "epoch": 0.35489019758930046,
      "grad_norm": 0.6816834211349487,
      "learning_rate": 9.261761991753629e-06,
      "loss": 0.7763,
      "step": 6448
    },
    {
      "epoch": 0.3549452363916561,
      "grad_norm": 0.792539119720459,
      "learning_rate": 9.261535286627025e-06,
      "loss": 0.7829,
      "step": 6449
    },
    {
      "epoch": 0.3550002751940118,
      "grad_norm": 0.8563211560249329,
      "learning_rate": 9.261308549471866e-06,
      "loss": 0.8945,
      "step": 6450
    },
    {
      "epoch": 0.3550553139963674,
      "grad_norm": 0.7241078019142151,
      "learning_rate": 9.26108178028985e-06,
      "loss": 0.6936,
      "step": 6451
    },
    {
      "epoch": 0.3551103527987231,
      "grad_norm": 0.7150034308433533,
      "learning_rate": 9.260854979082682e-06,
      "loss": 0.7689,
      "step": 6452
    },
    {
      "epoch": 0.35516539160107874,
      "grad_norm": 0.8630193471908569,
      "learning_rate": 9.260628145852073e-06,
      "loss": 0.8506,
      "step": 6453
    },
    {
      "epoch": 0.35522043040343443,
      "grad_norm": 0.7133893370628357,
      "learning_rate": 9.26040128059972e-06,
      "loss": 0.7976,
      "step": 6454
    },
    {
      "epoch": 0.35527546920579006,
      "grad_norm": 0.6984630823135376,
      "learning_rate": 9.260174383327332e-06,
      "loss": 0.7442,
      "step": 6455
    },
    {
      "epoch": 0.35533050800814575,
      "grad_norm": 0.7166933417320251,
      "learning_rate": 9.259947454036613e-06,
      "loss": 0.813,
      "step": 6456
    },
    {
      "epoch": 0.3553855468105014,
      "grad_norm": 0.7353581190109253,
      "learning_rate": 9.259720492729272e-06,
      "loss": 0.8157,
      "step": 6457
    },
    {
      "epoch": 0.3554405856128571,
      "grad_norm": 0.6810038089752197,
      "learning_rate": 9.259493499407011e-06,
      "loss": 0.7423,
      "step": 6458
    },
    {
      "epoch": 0.3554956244152127,
      "grad_norm": 1.1599586009979248,
      "learning_rate": 9.259266474071535e-06,
      "loss": 0.7159,
      "step": 6459
    },
    {
      "epoch": 0.3555506632175684,
      "grad_norm": 0.7857629060745239,
      "learning_rate": 9.259039416724554e-06,
      "loss": 0.7846,
      "step": 6460
    },
    {
      "epoch": 0.35560570201992403,
      "grad_norm": 0.705333948135376,
      "learning_rate": 9.258812327367773e-06,
      "loss": 0.751,
      "step": 6461
    },
    {
      "epoch": 0.3556607408222797,
      "grad_norm": 0.6899998188018799,
      "learning_rate": 9.258585206002897e-06,
      "loss": 0.7303,
      "step": 6462
    },
    {
      "epoch": 0.35571577962463535,
      "grad_norm": 0.8007912039756775,
      "learning_rate": 9.258358052631637e-06,
      "loss": 0.7363,
      "step": 6463
    },
    {
      "epoch": 0.35577081842699104,
      "grad_norm": 0.9403146505355835,
      "learning_rate": 9.258130867255695e-06,
      "loss": 0.9096,
      "step": 6464
    },
    {
      "epoch": 0.3558258572293467,
      "grad_norm": 0.7069174647331238,
      "learning_rate": 9.257903649876782e-06,
      "loss": 0.7362,
      "step": 6465
    },
    {
      "epoch": 0.35588089603170237,
      "grad_norm": 0.770807683467865,
      "learning_rate": 9.257676400496607e-06,
      "loss": 0.7904,
      "step": 6466
    },
    {
      "epoch": 0.355935934834058,
      "grad_norm": 0.8586871027946472,
      "learning_rate": 9.257449119116874e-06,
      "loss": 0.7596,
      "step": 6467
    },
    {
      "epoch": 0.3559909736364137,
      "grad_norm": 0.6934101581573486,
      "learning_rate": 9.257221805739294e-06,
      "loss": 0.6655,
      "step": 6468
    },
    {
      "epoch": 0.3560460124387693,
      "grad_norm": 0.9494497179985046,
      "learning_rate": 9.256994460365573e-06,
      "loss": 0.7923,
      "step": 6469
    },
    {
      "epoch": 0.356101051241125,
      "grad_norm": 0.7131130695343018,
      "learning_rate": 9.256767082997422e-06,
      "loss": 0.819,
      "step": 6470
    },
    {
      "epoch": 0.35615609004348064,
      "grad_norm": 0.8641398549079895,
      "learning_rate": 9.25653967363655e-06,
      "loss": 0.8275,
      "step": 6471
    },
    {
      "epoch": 0.35621112884583633,
      "grad_norm": 0.7350367307662964,
      "learning_rate": 9.256312232284665e-06,
      "loss": 0.7991,
      "step": 6472
    },
    {
      "epoch": 0.35626616764819197,
      "grad_norm": 0.8174671530723572,
      "learning_rate": 9.256084758943476e-06,
      "loss": 0.7147,
      "step": 6473
    },
    {
      "epoch": 0.35632120645054766,
      "grad_norm": 0.7560263872146606,
      "learning_rate": 9.255857253614693e-06,
      "loss": 0.7435,
      "step": 6474
    },
    {
      "epoch": 0.3563762452529033,
      "grad_norm": 0.7465197443962097,
      "learning_rate": 9.255629716300025e-06,
      "loss": 0.8228,
      "step": 6475
    },
    {
      "epoch": 0.356431284055259,
      "grad_norm": 0.7130733728408813,
      "learning_rate": 9.255402147001184e-06,
      "loss": 0.8361,
      "step": 6476
    },
    {
      "epoch": 0.3564863228576146,
      "grad_norm": 0.7200759053230286,
      "learning_rate": 9.255174545719882e-06,
      "loss": 0.7387,
      "step": 6477
    },
    {
      "epoch": 0.3565413616599703,
      "grad_norm": 0.8387622237205505,
      "learning_rate": 9.254946912457826e-06,
      "loss": 0.8427,
      "step": 6478
    },
    {
      "epoch": 0.35659640046232594,
      "grad_norm": 0.7263510823249817,
      "learning_rate": 9.254719247216725e-06,
      "loss": 0.712,
      "step": 6479
    },
    {
      "epoch": 0.3566514392646816,
      "grad_norm": 0.7393862009048462,
      "learning_rate": 9.254491549998296e-06,
      "loss": 0.6916,
      "step": 6480
    },
    {
      "epoch": 0.35670647806703726,
      "grad_norm": 0.7289569973945618,
      "learning_rate": 9.254263820804246e-06,
      "loss": 0.7561,
      "step": 6481
    },
    {
      "epoch": 0.35676151686939295,
      "grad_norm": 0.7597448825836182,
      "learning_rate": 9.254036059636288e-06,
      "loss": 0.853,
      "step": 6482
    },
    {
      "epoch": 0.3568165556717486,
      "grad_norm": 0.7652063369750977,
      "learning_rate": 9.253808266496136e-06,
      "loss": 0.7652,
      "step": 6483
    },
    {
      "epoch": 0.35687159447410427,
      "grad_norm": 1.193938136100769,
      "learning_rate": 9.253580441385497e-06,
      "loss": 0.8288,
      "step": 6484
    },
    {
      "epoch": 0.3569266332764599,
      "grad_norm": 0.9258719086647034,
      "learning_rate": 9.253352584306087e-06,
      "loss": 0.807,
      "step": 6485
    },
    {
      "epoch": 0.35698167207881554,
      "grad_norm": 0.78384929895401,
      "learning_rate": 9.253124695259617e-06,
      "loss": 0.7785,
      "step": 6486
    },
    {
      "epoch": 0.3570367108811712,
      "grad_norm": 0.801403284072876,
      "learning_rate": 9.252896774247802e-06,
      "loss": 0.8382,
      "step": 6487
    },
    {
      "epoch": 0.35709174968352686,
      "grad_norm": 0.9472376108169556,
      "learning_rate": 9.25266882127235e-06,
      "loss": 0.8661,
      "step": 6488
    },
    {
      "epoch": 0.35714678848588255,
      "grad_norm": 0.7575686573982239,
      "learning_rate": 9.252440836334981e-06,
      "loss": 0.8428,
      "step": 6489
    },
    {
      "epoch": 0.3572018272882382,
      "grad_norm": 0.736282467842102,
      "learning_rate": 9.252212819437402e-06,
      "loss": 0.801,
      "step": 6490
    },
    {
      "epoch": 0.35725686609059387,
      "grad_norm": 0.7420864701271057,
      "learning_rate": 9.251984770581332e-06,
      "loss": 0.8849,
      "step": 6491
    },
    {
      "epoch": 0.3573119048929495,
      "grad_norm": 0.7129189372062683,
      "learning_rate": 9.251756689768482e-06,
      "loss": 0.7716,
      "step": 6492
    },
    {
      "epoch": 0.3573669436953052,
      "grad_norm": 0.7777297496795654,
      "learning_rate": 9.251528577000566e-06,
      "loss": 0.8183,
      "step": 6493
    },
    {
      "epoch": 0.35742198249766083,
      "grad_norm": 0.7644590139389038,
      "learning_rate": 9.2513004322793e-06,
      "loss": 0.6319,
      "step": 6494
    },
    {
      "epoch": 0.3574770213000165,
      "grad_norm": 0.7112484574317932,
      "learning_rate": 9.251072255606399e-06,
      "loss": 0.8012,
      "step": 6495
    },
    {
      "epoch": 0.35753206010237215,
      "grad_norm": 0.7772265076637268,
      "learning_rate": 9.250844046983576e-06,
      "loss": 0.8372,
      "step": 6496
    },
    {
      "epoch": 0.35758709890472784,
      "grad_norm": 0.9530157446861267,
      "learning_rate": 9.250615806412546e-06,
      "loss": 0.8683,
      "step": 6497
    },
    {
      "epoch": 0.3576421377070835,
      "grad_norm": 0.7249575257301331,
      "learning_rate": 9.250387533895026e-06,
      "loss": 0.7091,
      "step": 6498
    },
    {
      "epoch": 0.35769717650943916,
      "grad_norm": 0.8549422025680542,
      "learning_rate": 9.25015922943273e-06,
      "loss": 0.8376,
      "step": 6499
    },
    {
      "epoch": 0.3577522153117948,
      "grad_norm": 0.74477618932724,
      "learning_rate": 9.249930893027376e-06,
      "loss": 0.7594,
      "step": 6500
    },
    {
      "epoch": 0.3578072541141505,
      "grad_norm": 0.8269739151000977,
      "learning_rate": 9.24970252468068e-06,
      "loss": 0.6473,
      "step": 6501
    },
    {
      "epoch": 0.3578622929165061,
      "grad_norm": 0.8375437259674072,
      "learning_rate": 9.249474124394358e-06,
      "loss": 0.7631,
      "step": 6502
    },
    {
      "epoch": 0.3579173317188618,
      "grad_norm": 0.8680340051651001,
      "learning_rate": 9.249245692170123e-06,
      "loss": 0.7863,
      "step": 6503
    },
    {
      "epoch": 0.35797237052121744,
      "grad_norm": 0.7179692983627319,
      "learning_rate": 9.249017228009696e-06,
      "loss": 0.8022,
      "step": 6504
    },
    {
      "epoch": 0.35802740932357313,
      "grad_norm": 0.7797464728355408,
      "learning_rate": 9.248788731914794e-06,
      "loss": 0.8067,
      "step": 6505
    },
    {
      "epoch": 0.35808244812592877,
      "grad_norm": 0.8032993674278259,
      "learning_rate": 9.248560203887133e-06,
      "loss": 0.7383,
      "step": 6506
    },
    {
      "epoch": 0.35813748692828445,
      "grad_norm": 0.7714722156524658,
      "learning_rate": 9.24833164392843e-06,
      "loss": 0.7149,
      "step": 6507
    },
    {
      "epoch": 0.3581925257306401,
      "grad_norm": 0.7492430210113525,
      "learning_rate": 9.248103052040404e-06,
      "loss": 0.7645,
      "step": 6508
    },
    {
      "epoch": 0.3582475645329958,
      "grad_norm": 0.6843901872634888,
      "learning_rate": 9.247874428224773e-06,
      "loss": 0.7183,
      "step": 6509
    },
    {
      "epoch": 0.3583026033353514,
      "grad_norm": 0.8370186686515808,
      "learning_rate": 9.247645772483254e-06,
      "loss": 0.7832,
      "step": 6510
    },
    {
      "epoch": 0.3583576421377071,
      "grad_norm": 0.7907791137695312,
      "learning_rate": 9.247417084817567e-06,
      "loss": 0.8742,
      "step": 6511
    },
    {
      "epoch": 0.35841268094006273,
      "grad_norm": 0.7950869798660278,
      "learning_rate": 9.247188365229428e-06,
      "loss": 0.8705,
      "step": 6512
    },
    {
      "epoch": 0.3584677197424184,
      "grad_norm": 0.7276936173439026,
      "learning_rate": 9.24695961372056e-06,
      "loss": 0.7629,
      "step": 6513
    },
    {
      "epoch": 0.35852275854477406,
      "grad_norm": 0.7761141657829285,
      "learning_rate": 9.24673083029268e-06,
      "loss": 0.8813,
      "step": 6514
    },
    {
      "epoch": 0.35857779734712975,
      "grad_norm": 0.7528283596038818,
      "learning_rate": 9.24650201494751e-06,
      "loss": 0.7885,
      "step": 6515
    },
    {
      "epoch": 0.3586328361494854,
      "grad_norm": 0.8972534537315369,
      "learning_rate": 9.246273167686765e-06,
      "loss": 0.9081,
      "step": 6516
    },
    {
      "epoch": 0.35868787495184107,
      "grad_norm": 0.7658557891845703,
      "learning_rate": 9.246044288512168e-06,
      "loss": 0.8451,
      "step": 6517
    },
    {
      "epoch": 0.3587429137541967,
      "grad_norm": 0.8013193607330322,
      "learning_rate": 9.245815377425438e-06,
      "loss": 0.7236,
      "step": 6518
    },
    {
      "epoch": 0.3587979525565524,
      "grad_norm": 0.8134163022041321,
      "learning_rate": 9.245586434428298e-06,
      "loss": 0.908,
      "step": 6519
    },
    {
      "epoch": 0.358852991358908,
      "grad_norm": 0.6479801535606384,
      "learning_rate": 9.245357459522466e-06,
      "loss": 0.7397,
      "step": 6520
    },
    {
      "epoch": 0.3589080301612637,
      "grad_norm": 0.70014488697052,
      "learning_rate": 9.245128452709665e-06,
      "loss": 0.6898,
      "step": 6521
    },
    {
      "epoch": 0.35896306896361935,
      "grad_norm": 0.7645437717437744,
      "learning_rate": 9.244899413991613e-06,
      "loss": 0.8319,
      "step": 6522
    },
    {
      "epoch": 0.35901810776597504,
      "grad_norm": 0.6812799572944641,
      "learning_rate": 9.244670343370033e-06,
      "loss": 0.7359,
      "step": 6523
    },
    {
      "epoch": 0.35907314656833067,
      "grad_norm": 0.6573774218559265,
      "learning_rate": 9.244441240846647e-06,
      "loss": 0.742,
      "step": 6524
    },
    {
      "epoch": 0.35912818537068636,
      "grad_norm": 0.7870661020278931,
      "learning_rate": 9.244212106423178e-06,
      "loss": 0.7307,
      "step": 6525
    },
    {
      "epoch": 0.359183224173042,
      "grad_norm": 0.9163166284561157,
      "learning_rate": 9.243982940101347e-06,
      "loss": 0.8584,
      "step": 6526
    },
    {
      "epoch": 0.3592382629753977,
      "grad_norm": 0.766888439655304,
      "learning_rate": 9.243753741882874e-06,
      "loss": 0.8093,
      "step": 6527
    },
    {
      "epoch": 0.3592933017777533,
      "grad_norm": 0.7831236124038696,
      "learning_rate": 9.243524511769486e-06,
      "loss": 0.8665,
      "step": 6528
    },
    {
      "epoch": 0.35934834058010895,
      "grad_norm": 0.7485133409500122,
      "learning_rate": 9.243295249762904e-06,
      "loss": 0.7336,
      "step": 6529
    },
    {
      "epoch": 0.35940337938246464,
      "grad_norm": 0.7231502532958984,
      "learning_rate": 9.24306595586485e-06,
      "loss": 0.8095,
      "step": 6530
    },
    {
      "epoch": 0.35945841818482027,
      "grad_norm": 0.821898877620697,
      "learning_rate": 9.242836630077048e-06,
      "loss": 0.831,
      "step": 6531
    },
    {
      "epoch": 0.35951345698717596,
      "grad_norm": 0.6792737245559692,
      "learning_rate": 9.242607272401223e-06,
      "loss": 0.7183,
      "step": 6532
    },
    {
      "epoch": 0.3595684957895316,
      "grad_norm": 0.7200430631637573,
      "learning_rate": 9.242377882839095e-06,
      "loss": 0.7256,
      "step": 6533
    },
    {
      "epoch": 0.3596235345918873,
      "grad_norm": 0.6713700890541077,
      "learning_rate": 9.242148461392393e-06,
      "loss": 0.7416,
      "step": 6534
    },
    {
      "epoch": 0.3596785733942429,
      "grad_norm": 0.7054564356803894,
      "learning_rate": 9.241919008062836e-06,
      "loss": 0.6856,
      "step": 6535
    },
    {
      "epoch": 0.3597336121965986,
      "grad_norm": 0.7516196966171265,
      "learning_rate": 9.241689522852152e-06,
      "loss": 0.7149,
      "step": 6536
    },
    {
      "epoch": 0.35978865099895424,
      "grad_norm": 0.8547651767730713,
      "learning_rate": 9.241460005762067e-06,
      "loss": 0.7075,
      "step": 6537
    },
    {
      "epoch": 0.35984368980130993,
      "grad_norm": 0.6791819334030151,
      "learning_rate": 9.241230456794302e-06,
      "loss": 0.6449,
      "step": 6538
    },
    {
      "epoch": 0.35989872860366556,
      "grad_norm": 0.8365122079849243,
      "learning_rate": 9.241000875950583e-06,
      "loss": 0.7619,
      "step": 6539
    },
    {
      "epoch": 0.35995376740602125,
      "grad_norm": 0.763829231262207,
      "learning_rate": 9.24077126323264e-06,
      "loss": 0.71,
      "step": 6540
    },
    {
      "epoch": 0.3600088062083769,
      "grad_norm": 0.7698483467102051,
      "learning_rate": 9.240541618642193e-06,
      "loss": 0.7949,
      "step": 6541
    },
    {
      "epoch": 0.3600638450107326,
      "grad_norm": 0.7331508994102478,
      "learning_rate": 9.24031194218097e-06,
      "loss": 0.8292,
      "step": 6542
    },
    {
      "epoch": 0.3601188838130882,
      "grad_norm": 0.7507451772689819,
      "learning_rate": 9.2400822338507e-06,
      "loss": 0.8651,
      "step": 6543
    },
    {
      "epoch": 0.3601739226154439,
      "grad_norm": 0.8537001609802246,
      "learning_rate": 9.239852493653104e-06,
      "loss": 0.848,
      "step": 6544
    },
    {
      "epoch": 0.36022896141779953,
      "grad_norm": 0.683311939239502,
      "learning_rate": 9.239622721589913e-06,
      "loss": 0.803,
      "step": 6545
    },
    {
      "epoch": 0.3602840002201552,
      "grad_norm": 0.6916974186897278,
      "learning_rate": 9.239392917662852e-06,
      "loss": 0.8037,
      "step": 6546
    },
    {
      "epoch": 0.36033903902251085,
      "grad_norm": 0.798795223236084,
      "learning_rate": 9.23916308187365e-06,
      "loss": 0.8037,
      "step": 6547
    },
    {
      "epoch": 0.36039407782486654,
      "grad_norm": 0.7284069657325745,
      "learning_rate": 9.238933214224032e-06,
      "loss": 0.7365,
      "step": 6548
    },
    {
      "epoch": 0.3604491166272222,
      "grad_norm": 0.7789250016212463,
      "learning_rate": 9.238703314715727e-06,
      "loss": 0.788,
      "step": 6549
    },
    {
      "epoch": 0.36050415542957787,
      "grad_norm": 0.7029675841331482,
      "learning_rate": 9.238473383350462e-06,
      "loss": 0.7796,
      "step": 6550
    },
    {
      "epoch": 0.3605591942319335,
      "grad_norm": 0.9094457626342773,
      "learning_rate": 9.238243420129965e-06,
      "loss": 0.7884,
      "step": 6551
    },
    {
      "epoch": 0.3606142330342892,
      "grad_norm": 0.8253848552703857,
      "learning_rate": 9.238013425055965e-06,
      "loss": 0.7671,
      "step": 6552
    },
    {
      "epoch": 0.3606692718366448,
      "grad_norm": 0.7052987813949585,
      "learning_rate": 9.237783398130193e-06,
      "loss": 0.7511,
      "step": 6553
    },
    {
      "epoch": 0.3607243106390005,
      "grad_norm": 0.7506607174873352,
      "learning_rate": 9.237553339354373e-06,
      "loss": 0.6804,
      "step": 6554
    },
    {
      "epoch": 0.36077934944135615,
      "grad_norm": 0.725106418132782,
      "learning_rate": 9.237323248730237e-06,
      "loss": 0.7658,
      "step": 6555
    },
    {
      "epoch": 0.36083438824371183,
      "grad_norm": 0.8164945244789124,
      "learning_rate": 9.237093126259515e-06,
      "loss": 0.7857,
      "step": 6556
    },
    {
      "epoch": 0.36088942704606747,
      "grad_norm": 0.6937377452850342,
      "learning_rate": 9.236862971943934e-06,
      "loss": 0.6985,
      "step": 6557
    },
    {
      "epoch": 0.36094446584842316,
      "grad_norm": 0.7511105537414551,
      "learning_rate": 9.236632785785225e-06,
      "loss": 0.7891,
      "step": 6558
    },
    {
      "epoch": 0.3609995046507788,
      "grad_norm": 0.7217637896537781,
      "learning_rate": 9.236402567785118e-06,
      "loss": 0.7942,
      "step": 6559
    },
    {
      "epoch": 0.3610545434531345,
      "grad_norm": 1.1438478231430054,
      "learning_rate": 9.236172317945343e-06,
      "loss": 0.8311,
      "step": 6560
    },
    {
      "epoch": 0.3611095822554901,
      "grad_norm": 0.7414245009422302,
      "learning_rate": 9.23594203626763e-06,
      "loss": 0.7726,
      "step": 6561
    },
    {
      "epoch": 0.3611646210578458,
      "grad_norm": 0.7762154340744019,
      "learning_rate": 9.235711722753712e-06,
      "loss": 0.7891,
      "step": 6562
    },
    {
      "epoch": 0.36121965986020144,
      "grad_norm": 0.7368801832199097,
      "learning_rate": 9.23548137740532e-06,
      "loss": 0.7656,
      "step": 6563
    },
    {
      "epoch": 0.3612746986625571,
      "grad_norm": 0.7571502923965454,
      "learning_rate": 9.235251000224181e-06,
      "loss": 0.7845,
      "step": 6564
    },
    {
      "epoch": 0.36132973746491276,
      "grad_norm": 0.8078309297561646,
      "learning_rate": 9.235020591212031e-06,
      "loss": 0.7969,
      "step": 6565
    },
    {
      "epoch": 0.36138477626726845,
      "grad_norm": 0.6897913813591003,
      "learning_rate": 9.234790150370599e-06,
      "loss": 0.6922,
      "step": 6566
    },
    {
      "epoch": 0.3614398150696241,
      "grad_norm": 0.8053449988365173,
      "learning_rate": 9.234559677701618e-06,
      "loss": 0.9126,
      "step": 6567
    },
    {
      "epoch": 0.36149485387197977,
      "grad_norm": 0.8400903940200806,
      "learning_rate": 9.23432917320682e-06,
      "loss": 0.8144,
      "step": 6568
    },
    {
      "epoch": 0.3615498926743354,
      "grad_norm": 0.7753110527992249,
      "learning_rate": 9.234098636887935e-06,
      "loss": 0.7025,
      "step": 6569
    },
    {
      "epoch": 0.3616049314766911,
      "grad_norm": 0.7901243567466736,
      "learning_rate": 9.233868068746702e-06,
      "loss": 0.783,
      "step": 6570
    },
    {
      "epoch": 0.3616599702790467,
      "grad_norm": 1.2297497987747192,
      "learning_rate": 9.233637468784849e-06,
      "loss": 0.8541,
      "step": 6571
    },
    {
      "epoch": 0.36171500908140236,
      "grad_norm": 0.7590478658676147,
      "learning_rate": 9.233406837004108e-06,
      "loss": 0.7856,
      "step": 6572
    },
    {
      "epoch": 0.36177004788375805,
      "grad_norm": 0.6651493310928345,
      "learning_rate": 9.233176173406216e-06,
      "loss": 0.6822,
      "step": 6573
    },
    {
      "epoch": 0.3618250866861137,
      "grad_norm": 0.7760787010192871,
      "learning_rate": 9.232945477992905e-06,
      "loss": 0.8017,
      "step": 6574
    },
    {
      "epoch": 0.3618801254884694,
      "grad_norm": 0.8788009285926819,
      "learning_rate": 9.232714750765908e-06,
      "loss": 0.7812,
      "step": 6575
    },
    {
      "epoch": 0.361935164290825,
      "grad_norm": 0.7014517188072205,
      "learning_rate": 9.232483991726961e-06,
      "loss": 0.7293,
      "step": 6576
    },
    {
      "epoch": 0.3619902030931807,
      "grad_norm": 0.7586061954498291,
      "learning_rate": 9.232253200877797e-06,
      "loss": 0.7953,
      "step": 6577
    },
    {
      "epoch": 0.36204524189553633,
      "grad_norm": 0.8202564120292664,
      "learning_rate": 9.232022378220151e-06,
      "loss": 0.8545,
      "step": 6578
    },
    {
      "epoch": 0.362100280697892,
      "grad_norm": 0.7816846966743469,
      "learning_rate": 9.231791523755758e-06,
      "loss": 0.8573,
      "step": 6579
    },
    {
      "epoch": 0.36215531950024765,
      "grad_norm": 0.883222222328186,
      "learning_rate": 9.23156063748635e-06,
      "loss": 0.7733,
      "step": 6580
    },
    {
      "epoch": 0.36221035830260334,
      "grad_norm": 0.8472830057144165,
      "learning_rate": 9.231329719413668e-06,
      "loss": 0.8931,
      "step": 6581
    },
    {
      "epoch": 0.362265397104959,
      "grad_norm": 0.7916087508201599,
      "learning_rate": 9.231098769539443e-06,
      "loss": 0.8806,
      "step": 6582
    },
    {
      "epoch": 0.36232043590731466,
      "grad_norm": 0.815339982509613,
      "learning_rate": 9.230867787865414e-06,
      "loss": 0.9081,
      "step": 6583
    },
    {
      "epoch": 0.3623754747096703,
      "grad_norm": 1.2352560758590698,
      "learning_rate": 9.230636774393312e-06,
      "loss": 0.726,
      "step": 6584
    },
    {
      "epoch": 0.362430513512026,
      "grad_norm": 0.759308397769928,
      "learning_rate": 9.230405729124878e-06,
      "loss": 0.7648,
      "step": 6585
    },
    {
      "epoch": 0.3624855523143816,
      "grad_norm": 0.8285754323005676,
      "learning_rate": 9.230174652061847e-06,
      "loss": 0.7972,
      "step": 6586
    },
    {
      "epoch": 0.3625405911167373,
      "grad_norm": 0.7393043041229248,
      "learning_rate": 9.229943543205956e-06,
      "loss": 0.7859,
      "step": 6587
    },
    {
      "epoch": 0.36259562991909294,
      "grad_norm": 0.7354594469070435,
      "learning_rate": 9.229712402558942e-06,
      "loss": 0.6683,
      "step": 6588
    },
    {
      "epoch": 0.36265066872144863,
      "grad_norm": 0.8244406580924988,
      "learning_rate": 9.229481230122543e-06,
      "loss": 0.6977,
      "step": 6589
    },
    {
      "epoch": 0.36270570752380427,
      "grad_norm": 0.810565173625946,
      "learning_rate": 9.229250025898493e-06,
      "loss": 0.7278,
      "step": 6590
    },
    {
      "epoch": 0.36276074632615996,
      "grad_norm": 0.7443352937698364,
      "learning_rate": 9.229018789888532e-06,
      "loss": 0.7821,
      "step": 6591
    },
    {
      "epoch": 0.3628157851285156,
      "grad_norm": 0.9211748838424683,
      "learning_rate": 9.228787522094398e-06,
      "loss": 0.9174,
      "step": 6592
    },
    {
      "epoch": 0.3628708239308713,
      "grad_norm": 0.7099255919456482,
      "learning_rate": 9.22855622251783e-06,
      "loss": 0.74,
      "step": 6593
    },
    {
      "epoch": 0.3629258627332269,
      "grad_norm": 0.7373029589653015,
      "learning_rate": 9.228324891160564e-06,
      "loss": 0.7909,
      "step": 6594
    },
    {
      "epoch": 0.3629809015355826,
      "grad_norm": 0.8774755001068115,
      "learning_rate": 9.22809352802434e-06,
      "loss": 0.8354,
      "step": 6595
    },
    {
      "epoch": 0.36303594033793823,
      "grad_norm": 0.7547696232795715,
      "learning_rate": 9.227862133110899e-06,
      "loss": 0.6942,
      "step": 6596
    },
    {
      "epoch": 0.3630909791402939,
      "grad_norm": 0.7868191003799438,
      "learning_rate": 9.227630706421975e-06,
      "loss": 0.7575,
      "step": 6597
    },
    {
      "epoch": 0.36314601794264956,
      "grad_norm": 0.6753721237182617,
      "learning_rate": 9.227399247959312e-06,
      "loss": 0.7092,
      "step": 6598
    },
    {
      "epoch": 0.36320105674500525,
      "grad_norm": 0.7317304611206055,
      "learning_rate": 9.227167757724646e-06,
      "loss": 0.8372,
      "step": 6599
    },
    {
      "epoch": 0.3632560955473609,
      "grad_norm": 0.8928040266036987,
      "learning_rate": 9.226936235719721e-06,
      "loss": 0.8536,
      "step": 6600
    },
    {
      "epoch": 0.36331113434971657,
      "grad_norm": 0.7178280353546143,
      "learning_rate": 9.226704681946275e-06,
      "loss": 0.7648,
      "step": 6601
    },
    {
      "epoch": 0.3633661731520722,
      "grad_norm": 0.7439851760864258,
      "learning_rate": 9.226473096406046e-06,
      "loss": 0.8284,
      "step": 6602
    },
    {
      "epoch": 0.3634212119544279,
      "grad_norm": 0.7000887989997864,
      "learning_rate": 9.226241479100777e-06,
      "loss": 0.7797,
      "step": 6603
    },
    {
      "epoch": 0.3634762507567835,
      "grad_norm": 0.7882626056671143,
      "learning_rate": 9.226009830032209e-06,
      "loss": 0.72,
      "step": 6604
    },
    {
      "epoch": 0.3635312895591392,
      "grad_norm": 0.6445927619934082,
      "learning_rate": 9.225778149202081e-06,
      "loss": 0.6785,
      "step": 6605
    },
    {
      "epoch": 0.36358632836149485,
      "grad_norm": 0.7348469495773315,
      "learning_rate": 9.225546436612137e-06,
      "loss": 0.8117,
      "step": 6606
    },
    {
      "epoch": 0.36364136716385054,
      "grad_norm": 0.7455001473426819,
      "learning_rate": 9.225314692264118e-06,
      "loss": 0.8196,
      "step": 6607
    },
    {
      "epoch": 0.36369640596620617,
      "grad_norm": 0.7149390578269958,
      "learning_rate": 9.225082916159762e-06,
      "loss": 0.8841,
      "step": 6608
    },
    {
      "epoch": 0.36375144476856186,
      "grad_norm": 0.7095748782157898,
      "learning_rate": 9.224851108300816e-06,
      "loss": 0.7336,
      "step": 6609
    },
    {
      "epoch": 0.3638064835709175,
      "grad_norm": 0.7112231850624084,
      "learning_rate": 9.224619268689019e-06,
      "loss": 0.8606,
      "step": 6610
    },
    {
      "epoch": 0.3638615223732732,
      "grad_norm": 0.8052846789360046,
      "learning_rate": 9.224387397326115e-06,
      "loss": 0.7838,
      "step": 6611
    },
    {
      "epoch": 0.3639165611756288,
      "grad_norm": 0.7538836002349854,
      "learning_rate": 9.224155494213846e-06,
      "loss": 0.8252,
      "step": 6612
    },
    {
      "epoch": 0.3639715999779845,
      "grad_norm": 0.6968722343444824,
      "learning_rate": 9.223923559353956e-06,
      "loss": 0.759,
      "step": 6613
    },
    {
      "epoch": 0.36402663878034014,
      "grad_norm": 0.7797368168830872,
      "learning_rate": 9.223691592748185e-06,
      "loss": 0.8452,
      "step": 6614
    },
    {
      "epoch": 0.3640816775826958,
      "grad_norm": 0.7738572955131531,
      "learning_rate": 9.223459594398278e-06,
      "loss": 0.806,
      "step": 6615
    },
    {
      "epoch": 0.36413671638505146,
      "grad_norm": 0.7998547554016113,
      "learning_rate": 9.223227564305983e-06,
      "loss": 0.748,
      "step": 6616
    },
    {
      "epoch": 0.3641917551874071,
      "grad_norm": 0.838666558265686,
      "learning_rate": 9.222995502473037e-06,
      "loss": 0.8252,
      "step": 6617
    },
    {
      "epoch": 0.3642467939897628,
      "grad_norm": 1.1672697067260742,
      "learning_rate": 9.222763408901189e-06,
      "loss": 0.806,
      "step": 6618
    },
    {
      "epoch": 0.3643018327921184,
      "grad_norm": 0.6721193194389343,
      "learning_rate": 9.22253128359218e-06,
      "loss": 0.6897,
      "step": 6619
    },
    {
      "epoch": 0.3643568715944741,
      "grad_norm": 0.8152795433998108,
      "learning_rate": 9.222299126547758e-06,
      "loss": 0.8377,
      "step": 6620
    },
    {
      "epoch": 0.36441191039682974,
      "grad_norm": 0.7959492206573486,
      "learning_rate": 9.222066937769664e-06,
      "loss": 0.8496,
      "step": 6621
    },
    {
      "epoch": 0.36446694919918543,
      "grad_norm": 0.7759784460067749,
      "learning_rate": 9.221834717259646e-06,
      "loss": 0.7736,
      "step": 6622
    },
    {
      "epoch": 0.36452198800154106,
      "grad_norm": 0.6929076313972473,
      "learning_rate": 9.221602465019449e-06,
      "loss": 0.7759,
      "step": 6623
    },
    {
      "epoch": 0.36457702680389675,
      "grad_norm": 0.7323315143585205,
      "learning_rate": 9.221370181050817e-06,
      "loss": 0.7958,
      "step": 6624
    },
    {
      "epoch": 0.3646320656062524,
      "grad_norm": 0.7177294492721558,
      "learning_rate": 9.221137865355496e-06,
      "loss": 0.8405,
      "step": 6625
    },
    {
      "epoch": 0.3646871044086081,
      "grad_norm": 0.7425093650817871,
      "learning_rate": 9.220905517935235e-06,
      "loss": 0.7722,
      "step": 6626
    },
    {
      "epoch": 0.3647421432109637,
      "grad_norm": 0.8761040568351746,
      "learning_rate": 9.220673138791775e-06,
      "loss": 0.8617,
      "step": 6627
    },
    {
      "epoch": 0.3647971820133194,
      "grad_norm": 0.927509069442749,
      "learning_rate": 9.220440727926869e-06,
      "loss": 0.7839,
      "step": 6628
    },
    {
      "epoch": 0.36485222081567503,
      "grad_norm": 0.874399721622467,
      "learning_rate": 9.220208285342258e-06,
      "loss": 0.9697,
      "step": 6629
    },
    {
      "epoch": 0.3649072596180307,
      "grad_norm": 0.931384801864624,
      "learning_rate": 9.219975811039691e-06,
      "loss": 0.8142,
      "step": 6630
    },
    {
      "epoch": 0.36496229842038636,
      "grad_norm": 0.8567885160446167,
      "learning_rate": 9.219743305020916e-06,
      "loss": 0.7623,
      "step": 6631
    },
    {
      "epoch": 0.36501733722274204,
      "grad_norm": 0.7287514209747314,
      "learning_rate": 9.21951076728768e-06,
      "loss": 0.8044,
      "step": 6632
    },
    {
      "epoch": 0.3650723760250977,
      "grad_norm": 0.7234703302383423,
      "learning_rate": 9.21927819784173e-06,
      "loss": 0.7736,
      "step": 6633
    },
    {
      "epoch": 0.36512741482745337,
      "grad_norm": 0.7174978256225586,
      "learning_rate": 9.219045596684815e-06,
      "loss": 0.7658,
      "step": 6634
    },
    {
      "epoch": 0.365182453629809,
      "grad_norm": 0.751075804233551,
      "learning_rate": 9.218812963818682e-06,
      "loss": 0.7586,
      "step": 6635
    },
    {
      "epoch": 0.3652374924321647,
      "grad_norm": 0.755283534526825,
      "learning_rate": 9.21858029924508e-06,
      "loss": 0.8904,
      "step": 6636
    },
    {
      "epoch": 0.3652925312345203,
      "grad_norm": 0.6439716815948486,
      "learning_rate": 9.21834760296576e-06,
      "loss": 0.7335,
      "step": 6637
    },
    {
      "epoch": 0.365347570036876,
      "grad_norm": 0.735285758972168,
      "learning_rate": 9.218114874982467e-06,
      "loss": 0.7193,
      "step": 6638
    },
    {
      "epoch": 0.36540260883923165,
      "grad_norm": 0.7724307775497437,
      "learning_rate": 9.217882115296952e-06,
      "loss": 0.8322,
      "step": 6639
    },
    {
      "epoch": 0.36545764764158734,
      "grad_norm": 0.7771303653717041,
      "learning_rate": 9.217649323910964e-06,
      "loss": 0.7952,
      "step": 6640
    },
    {
      "epoch": 0.36551268644394297,
      "grad_norm": 0.7753337621688843,
      "learning_rate": 9.217416500826251e-06,
      "loss": 0.8501,
      "step": 6641
    },
    {
      "epoch": 0.36556772524629866,
      "grad_norm": 0.8104514479637146,
      "learning_rate": 9.217183646044567e-06,
      "loss": 0.8503,
      "step": 6642
    },
    {
      "epoch": 0.3656227640486543,
      "grad_norm": 0.7191929221153259,
      "learning_rate": 9.21695075956766e-06,
      "loss": 0.7578,
      "step": 6643
    },
    {
      "epoch": 0.36567780285101,
      "grad_norm": 0.745837926864624,
      "learning_rate": 9.216717841397277e-06,
      "loss": 0.819,
      "step": 6644
    },
    {
      "epoch": 0.3657328416533656,
      "grad_norm": 0.7019662261009216,
      "learning_rate": 9.216484891535174e-06,
      "loss": 0.8024,
      "step": 6645
    },
    {
      "epoch": 0.3657878804557213,
      "grad_norm": 0.9709738492965698,
      "learning_rate": 9.216251909983095e-06,
      "loss": 0.7653,
      "step": 6646
    },
    {
      "epoch": 0.36584291925807694,
      "grad_norm": 0.7973032593727112,
      "learning_rate": 9.2160188967428e-06,
      "loss": 0.8071,
      "step": 6647
    },
    {
      "epoch": 0.3658979580604326,
      "grad_norm": 0.6945796012878418,
      "learning_rate": 9.215785851816034e-06,
      "loss": 0.6831,
      "step": 6648
    },
    {
      "epoch": 0.36595299686278826,
      "grad_norm": 0.8685100674629211,
      "learning_rate": 9.21555277520455e-06,
      "loss": 0.821,
      "step": 6649
    },
    {
      "epoch": 0.36600803566514395,
      "grad_norm": 1.0164310932159424,
      "learning_rate": 9.2153196669101e-06,
      "loss": 0.7861,
      "step": 6650
    },
    {
      "epoch": 0.3660630744674996,
      "grad_norm": 0.8572850227355957,
      "learning_rate": 9.215086526934435e-06,
      "loss": 0.7982,
      "step": 6651
    },
    {
      "epoch": 0.36611811326985527,
      "grad_norm": 0.7481987476348877,
      "learning_rate": 9.214853355279307e-06,
      "loss": 0.8258,
      "step": 6652
    },
    {
      "epoch": 0.3661731520722109,
      "grad_norm": 0.750344455242157,
      "learning_rate": 9.214620151946472e-06,
      "loss": 0.7842,
      "step": 6653
    },
    {
      "epoch": 0.3662281908745666,
      "grad_norm": 1.0266414880752563,
      "learning_rate": 9.214386916937678e-06,
      "loss": 0.7313,
      "step": 6654
    },
    {
      "epoch": 0.36628322967692223,
      "grad_norm": 0.7913589477539062,
      "learning_rate": 9.214153650254682e-06,
      "loss": 0.8251,
      "step": 6655
    },
    {
      "epoch": 0.3663382684792779,
      "grad_norm": 0.7185465693473816,
      "learning_rate": 9.213920351899235e-06,
      "loss": 0.7145,
      "step": 6656
    },
    {
      "epoch": 0.36639330728163355,
      "grad_norm": 0.7185063362121582,
      "learning_rate": 9.213687021873088e-06,
      "loss": 0.8321,
      "step": 6657
    },
    {
      "epoch": 0.3664483460839892,
      "grad_norm": 0.8380091190338135,
      "learning_rate": 9.213453660178e-06,
      "loss": 0.8293,
      "step": 6658
    },
    {
      "epoch": 0.3665033848863449,
      "grad_norm": 0.7569485306739807,
      "learning_rate": 9.21322026681572e-06,
      "loss": 0.7201,
      "step": 6659
    },
    {
      "epoch": 0.3665584236887005,
      "grad_norm": 0.7212445735931396,
      "learning_rate": 9.212986841788005e-06,
      "loss": 0.7869,
      "step": 6660
    },
    {
      "epoch": 0.3666134624910562,
      "grad_norm": 0.9435489773750305,
      "learning_rate": 9.212753385096612e-06,
      "loss": 0.8469,
      "step": 6661
    },
    {
      "epoch": 0.36666850129341183,
      "grad_norm": 0.6609265208244324,
      "learning_rate": 9.212519896743289e-06,
      "loss": 0.6446,
      "step": 6662
    },
    {
      "epoch": 0.3667235400957675,
      "grad_norm": 0.7232604026794434,
      "learning_rate": 9.212286376729794e-06,
      "loss": 0.7138,
      "step": 6663
    },
    {
      "epoch": 0.36677857889812315,
      "grad_norm": 0.7276197075843811,
      "learning_rate": 9.212052825057882e-06,
      "loss": 0.725,
      "step": 6664
    },
    {
      "epoch": 0.36683361770047884,
      "grad_norm": 0.7029727101325989,
      "learning_rate": 9.21181924172931e-06,
      "loss": 0.6973,
      "step": 6665
    },
    {
      "epoch": 0.3668886565028345,
      "grad_norm": 0.7292968034744263,
      "learning_rate": 9.21158562674583e-06,
      "loss": 0.6984,
      "step": 6666
    },
    {
      "epoch": 0.36694369530519017,
      "grad_norm": 0.6977009177207947,
      "learning_rate": 9.2113519801092e-06,
      "loss": 0.7752,
      "step": 6667
    },
    {
      "epoch": 0.3669987341075458,
      "grad_norm": 0.8019471764564514,
      "learning_rate": 9.211118301821176e-06,
      "loss": 0.7481,
      "step": 6668
    },
    {
      "epoch": 0.3670537729099015,
      "grad_norm": 0.8097867965698242,
      "learning_rate": 9.210884591883516e-06,
      "loss": 0.8077,
      "step": 6669
    },
    {
      "epoch": 0.3671088117122571,
      "grad_norm": 1.1622828245162964,
      "learning_rate": 9.210650850297973e-06,
      "loss": 0.8053,
      "step": 6670
    },
    {
      "epoch": 0.3671638505146128,
      "grad_norm": 0.8188957571983337,
      "learning_rate": 9.210417077066304e-06,
      "loss": 0.7731,
      "step": 6671
    },
    {
      "epoch": 0.36721888931696844,
      "grad_norm": 0.8531584739685059,
      "learning_rate": 9.210183272190269e-06,
      "loss": 0.8183,
      "step": 6672
    },
    {
      "epoch": 0.36727392811932413,
      "grad_norm": 0.8007203936576843,
      "learning_rate": 9.209949435671624e-06,
      "loss": 0.7906,
      "step": 6673
    },
    {
      "epoch": 0.36732896692167977,
      "grad_norm": 0.8284860253334045,
      "learning_rate": 9.209715567512126e-06,
      "loss": 0.7845,
      "step": 6674
    },
    {
      "epoch": 0.36738400572403546,
      "grad_norm": 0.7735304236412048,
      "learning_rate": 9.209481667713533e-06,
      "loss": 0.7333,
      "step": 6675
    },
    {
      "epoch": 0.3674390445263911,
      "grad_norm": 0.7390912771224976,
      "learning_rate": 9.209247736277601e-06,
      "loss": 0.7992,
      "step": 6676
    },
    {
      "epoch": 0.3674940833287468,
      "grad_norm": 0.6871926784515381,
      "learning_rate": 9.209013773206091e-06,
      "loss": 0.7765,
      "step": 6677
    },
    {
      "epoch": 0.3675491221311024,
      "grad_norm": 0.7241746187210083,
      "learning_rate": 9.208779778500758e-06,
      "loss": 0.7124,
      "step": 6678
    },
    {
      "epoch": 0.3676041609334581,
      "grad_norm": 0.7362630367279053,
      "learning_rate": 9.208545752163365e-06,
      "loss": 0.7695,
      "step": 6679
    },
    {
      "epoch": 0.36765919973581374,
      "grad_norm": 0.7577944993972778,
      "learning_rate": 9.208311694195669e-06,
      "loss": 0.8302,
      "step": 6680
    },
    {
      "epoch": 0.3677142385381694,
      "grad_norm": 0.7182355523109436,
      "learning_rate": 9.208077604599427e-06,
      "loss": 0.8182,
      "step": 6681
    },
    {
      "epoch": 0.36776927734052506,
      "grad_norm": 0.7636679410934448,
      "learning_rate": 9.207843483376402e-06,
      "loss": 0.7266,
      "step": 6682
    },
    {
      "epoch": 0.36782431614288075,
      "grad_norm": 0.7325936555862427,
      "learning_rate": 9.207609330528349e-06,
      "loss": 0.735,
      "step": 6683
    },
    {
      "epoch": 0.3678793549452364,
      "grad_norm": 1.1119143962860107,
      "learning_rate": 9.207375146057033e-06,
      "loss": 1.0124,
      "step": 6684
    },
    {
      "epoch": 0.36793439374759207,
      "grad_norm": 0.7694228291511536,
      "learning_rate": 9.207140929964212e-06,
      "loss": 0.7803,
      "step": 6685
    },
    {
      "epoch": 0.3679894325499477,
      "grad_norm": 0.7628658413887024,
      "learning_rate": 9.206906682251644e-06,
      "loss": 0.8057,
      "step": 6686
    },
    {
      "epoch": 0.3680444713523034,
      "grad_norm": 0.766266942024231,
      "learning_rate": 9.206672402921092e-06,
      "loss": 0.7827,
      "step": 6687
    },
    {
      "epoch": 0.368099510154659,
      "grad_norm": 0.7355746626853943,
      "learning_rate": 9.206438091974316e-06,
      "loss": 0.8146,
      "step": 6688
    },
    {
      "epoch": 0.3681545489570147,
      "grad_norm": 0.8464547395706177,
      "learning_rate": 9.20620374941308e-06,
      "loss": 0.8296,
      "step": 6689
    },
    {
      "epoch": 0.36820958775937035,
      "grad_norm": 0.7113955616950989,
      "learning_rate": 9.20596937523914e-06,
      "loss": 0.7621,
      "step": 6690
    },
    {
      "epoch": 0.36826462656172604,
      "grad_norm": 0.7141324877738953,
      "learning_rate": 9.205734969454259e-06,
      "loss": 0.738,
      "step": 6691
    },
    {
      "epoch": 0.36831966536408167,
      "grad_norm": 0.7576237320899963,
      "learning_rate": 9.2055005320602e-06,
      "loss": 0.7727,
      "step": 6692
    },
    {
      "epoch": 0.36837470416643736,
      "grad_norm": 0.7448444962501526,
      "learning_rate": 9.205266063058727e-06,
      "loss": 0.8238,
      "step": 6693
    },
    {
      "epoch": 0.368429742968793,
      "grad_norm": 0.7441811561584473,
      "learning_rate": 9.205031562451599e-06,
      "loss": 0.7518,
      "step": 6694
    },
    {
      "epoch": 0.3684847817711487,
      "grad_norm": 0.9284115433692932,
      "learning_rate": 9.20479703024058e-06,
      "loss": 0.817,
      "step": 6695
    },
    {
      "epoch": 0.3685398205735043,
      "grad_norm": 0.7019243836402893,
      "learning_rate": 9.204562466427431e-06,
      "loss": 0.7403,
      "step": 6696
    },
    {
      "epoch": 0.36859485937586,
      "grad_norm": 0.6345306634902954,
      "learning_rate": 9.204327871013917e-06,
      "loss": 0.7058,
      "step": 6697
    },
    {
      "epoch": 0.36864989817821564,
      "grad_norm": 0.7375063300132751,
      "learning_rate": 9.2040932440018e-06,
      "loss": 0.831,
      "step": 6698
    },
    {
      "epoch": 0.36870493698057133,
      "grad_norm": 0.8213731050491333,
      "learning_rate": 9.203858585392842e-06,
      "loss": 0.7677,
      "step": 6699
    },
    {
      "epoch": 0.36875997578292696,
      "grad_norm": 0.7114601731300354,
      "learning_rate": 9.203623895188809e-06,
      "loss": 0.8015,
      "step": 6700
    },
    {
      "epoch": 0.3688150145852826,
      "grad_norm": 0.7707667350769043,
      "learning_rate": 9.203389173391463e-06,
      "loss": 0.7758,
      "step": 6701
    },
    {
      "epoch": 0.3688700533876383,
      "grad_norm": 0.7374396920204163,
      "learning_rate": 9.203154420002572e-06,
      "loss": 0.7583,
      "step": 6702
    },
    {
      "epoch": 0.3689250921899939,
      "grad_norm": 0.7156866192817688,
      "learning_rate": 9.202919635023895e-06,
      "loss": 0.8173,
      "step": 6703
    },
    {
      "epoch": 0.3689801309923496,
      "grad_norm": 0.6811904311180115,
      "learning_rate": 9.2026848184572e-06,
      "loss": 0.7441,
      "step": 6704
    },
    {
      "epoch": 0.36903516979470524,
      "grad_norm": 0.7515163421630859,
      "learning_rate": 9.20244997030425e-06,
      "loss": 0.7927,
      "step": 6705
    },
    {
      "epoch": 0.36909020859706093,
      "grad_norm": 0.761116087436676,
      "learning_rate": 9.202215090566813e-06,
      "loss": 0.7686,
      "step": 6706
    },
    {
      "epoch": 0.36914524739941657,
      "grad_norm": 0.8726711869239807,
      "learning_rate": 9.20198017924665e-06,
      "loss": 0.7831,
      "step": 6707
    },
    {
      "epoch": 0.36920028620177225,
      "grad_norm": 0.6868153810501099,
      "learning_rate": 9.20174523634553e-06,
      "loss": 0.7855,
      "step": 6708
    },
    {
      "epoch": 0.3692553250041279,
      "grad_norm": 0.7140498757362366,
      "learning_rate": 9.201510261865218e-06,
      "loss": 0.8144,
      "step": 6709
    },
    {
      "epoch": 0.3693103638064836,
      "grad_norm": 0.8745181560516357,
      "learning_rate": 9.201275255807478e-06,
      "loss": 0.9204,
      "step": 6710
    },
    {
      "epoch": 0.3693654026088392,
      "grad_norm": 0.6535945534706116,
      "learning_rate": 9.20104021817408e-06,
      "loss": 0.7729,
      "step": 6711
    },
    {
      "epoch": 0.3694204414111949,
      "grad_norm": 0.655857503414154,
      "learning_rate": 9.200805148966785e-06,
      "loss": 0.8373,
      "step": 6712
    },
    {
      "epoch": 0.36947548021355053,
      "grad_norm": 0.8393271565437317,
      "learning_rate": 9.200570048187365e-06,
      "loss": 0.8532,
      "step": 6713
    },
    {
      "epoch": 0.3695305190159062,
      "grad_norm": 0.7484574913978577,
      "learning_rate": 9.200334915837585e-06,
      "loss": 0.8411,
      "step": 6714
    },
    {
      "epoch": 0.36958555781826186,
      "grad_norm": 0.9913665652275085,
      "learning_rate": 9.200099751919212e-06,
      "loss": 0.9011,
      "step": 6715
    },
    {
      "epoch": 0.36964059662061755,
      "grad_norm": 0.7314063310623169,
      "learning_rate": 9.199864556434013e-06,
      "loss": 0.7184,
      "step": 6716
    },
    {
      "epoch": 0.3696956354229732,
      "grad_norm": 0.7881553173065186,
      "learning_rate": 9.199629329383758e-06,
      "loss": 0.796,
      "step": 6717
    },
    {
      "epoch": 0.36975067422532887,
      "grad_norm": 0.7440283298492432,
      "learning_rate": 9.199394070770212e-06,
      "loss": 0.7472,
      "step": 6718
    },
    {
      "epoch": 0.3698057130276845,
      "grad_norm": 0.6916326880455017,
      "learning_rate": 9.199158780595144e-06,
      "loss": 0.6808,
      "step": 6719
    },
    {
      "epoch": 0.3698607518300402,
      "grad_norm": 0.8482714295387268,
      "learning_rate": 9.198923458860323e-06,
      "loss": 0.7795,
      "step": 6720
    },
    {
      "epoch": 0.3699157906323958,
      "grad_norm": 0.7541999816894531,
      "learning_rate": 9.198688105567516e-06,
      "loss": 0.7917,
      "step": 6721
    },
    {
      "epoch": 0.3699708294347515,
      "grad_norm": 0.794335126876831,
      "learning_rate": 9.198452720718494e-06,
      "loss": 0.8463,
      "step": 6722
    },
    {
      "epoch": 0.37002586823710715,
      "grad_norm": 0.7866827845573425,
      "learning_rate": 9.198217304315025e-06,
      "loss": 0.7938,
      "step": 6723
    },
    {
      "epoch": 0.37008090703946284,
      "grad_norm": 0.7393556833267212,
      "learning_rate": 9.19798185635888e-06,
      "loss": 0.7825,
      "step": 6724
    },
    {
      "epoch": 0.37013594584181847,
      "grad_norm": 0.7131090760231018,
      "learning_rate": 9.197746376851825e-06,
      "loss": 0.7184,
      "step": 6725
    },
    {
      "epoch": 0.37019098464417416,
      "grad_norm": 0.7054039239883423,
      "learning_rate": 9.197510865795634e-06,
      "loss": 0.7458,
      "step": 6726
    },
    {
      "epoch": 0.3702460234465298,
      "grad_norm": 0.7437009811401367,
      "learning_rate": 9.197275323192073e-06,
      "loss": 0.7921,
      "step": 6727
    },
    {
      "epoch": 0.3703010622488855,
      "grad_norm": 1.0703076124191284,
      "learning_rate": 9.197039749042916e-06,
      "loss": 0.771,
      "step": 6728
    },
    {
      "epoch": 0.3703561010512411,
      "grad_norm": 0.8278045654296875,
      "learning_rate": 9.196804143349929e-06,
      "loss": 0.8984,
      "step": 6729
    },
    {
      "epoch": 0.3704111398535968,
      "grad_norm": 0.7713067531585693,
      "learning_rate": 9.196568506114887e-06,
      "loss": 0.7702,
      "step": 6730
    },
    {
      "epoch": 0.37046617865595244,
      "grad_norm": 0.9040505290031433,
      "learning_rate": 9.19633283733956e-06,
      "loss": 0.7113,
      "step": 6731
    },
    {
      "epoch": 0.3705212174583081,
      "grad_norm": 0.8853700757026672,
      "learning_rate": 9.196097137025718e-06,
      "loss": 0.8445,
      "step": 6732
    },
    {
      "epoch": 0.37057625626066376,
      "grad_norm": 0.6870817542076111,
      "learning_rate": 9.195861405175133e-06,
      "loss": 0.7613,
      "step": 6733
    },
    {
      "epoch": 0.37063129506301945,
      "grad_norm": 0.7539152503013611,
      "learning_rate": 9.195625641789579e-06,
      "loss": 0.7478,
      "step": 6734
    },
    {
      "epoch": 0.3706863338653751,
      "grad_norm": 0.7084356546401978,
      "learning_rate": 9.195389846870822e-06,
      "loss": 0.7803,
      "step": 6735
    },
    {
      "epoch": 0.3707413726677308,
      "grad_norm": 0.7883948087692261,
      "learning_rate": 9.19515402042064e-06,
      "loss": 0.8606,
      "step": 6736
    },
    {
      "epoch": 0.3707964114700864,
      "grad_norm": 0.714948296546936,
      "learning_rate": 9.194918162440804e-06,
      "loss": 0.8066,
      "step": 6737
    },
    {
      "epoch": 0.3708514502724421,
      "grad_norm": 0.7110786437988281,
      "learning_rate": 9.194682272933085e-06,
      "loss": 0.7439,
      "step": 6738
    },
    {
      "epoch": 0.37090648907479773,
      "grad_norm": 0.7281045317649841,
      "learning_rate": 9.194446351899257e-06,
      "loss": 0.7772,
      "step": 6739
    },
    {
      "epoch": 0.3709615278771534,
      "grad_norm": 0.7351245880126953,
      "learning_rate": 9.194210399341093e-06,
      "loss": 0.8777,
      "step": 6740
    },
    {
      "epoch": 0.37101656667950905,
      "grad_norm": 0.8028532266616821,
      "learning_rate": 9.193974415260367e-06,
      "loss": 0.7461,
      "step": 6741
    },
    {
      "epoch": 0.37107160548186474,
      "grad_norm": 0.8015451431274414,
      "learning_rate": 9.19373839965885e-06,
      "loss": 0.8006,
      "step": 6742
    },
    {
      "epoch": 0.3711266442842204,
      "grad_norm": 0.9567442536354065,
      "learning_rate": 9.193502352538321e-06,
      "loss": 0.8636,
      "step": 6743
    },
    {
      "epoch": 0.371181683086576,
      "grad_norm": 1.1413114070892334,
      "learning_rate": 9.193266273900547e-06,
      "loss": 0.8976,
      "step": 6744
    },
    {
      "epoch": 0.3712367218889317,
      "grad_norm": 0.6971789002418518,
      "learning_rate": 9.19303016374731e-06,
      "loss": 0.7419,
      "step": 6745
    },
    {
      "epoch": 0.37129176069128733,
      "grad_norm": 0.8117435574531555,
      "learning_rate": 9.192794022080378e-06,
      "loss": 0.8166,
      "step": 6746
    },
    {
      "epoch": 0.371346799493643,
      "grad_norm": 0.7748119831085205,
      "learning_rate": 9.19255784890153e-06,
      "loss": 0.8073,
      "step": 6747
    },
    {
      "epoch": 0.37140183829599865,
      "grad_norm": 0.6550068259239197,
      "learning_rate": 9.192321644212539e-06,
      "loss": 0.6976,
      "step": 6748
    },
    {
      "epoch": 0.37145687709835434,
      "grad_norm": 0.7931404709815979,
      "learning_rate": 9.19208540801518e-06,
      "loss": 0.7153,
      "step": 6749
    },
    {
      "epoch": 0.37151191590071,
      "grad_norm": 0.7107539176940918,
      "learning_rate": 9.19184914031123e-06,
      "loss": 0.7616,
      "step": 6750
    },
    {
      "epoch": 0.37156695470306567,
      "grad_norm": 0.6983848810195923,
      "learning_rate": 9.191612841102463e-06,
      "loss": 0.6507,
      "step": 6751
    },
    {
      "epoch": 0.3716219935054213,
      "grad_norm": 0.7653477787971497,
      "learning_rate": 9.191376510390657e-06,
      "loss": 0.708,
      "step": 6752
    },
    {
      "epoch": 0.371677032307777,
      "grad_norm": 0.8903954029083252,
      "learning_rate": 9.191140148177586e-06,
      "loss": 0.8131,
      "step": 6753
    },
    {
      "epoch": 0.3717320711101326,
      "grad_norm": 0.7584933042526245,
      "learning_rate": 9.190903754465028e-06,
      "loss": 0.8178,
      "step": 6754
    },
    {
      "epoch": 0.3717871099124883,
      "grad_norm": 0.7338405847549438,
      "learning_rate": 9.19066732925476e-06,
      "loss": 0.7717,
      "step": 6755
    },
    {
      "epoch": 0.37184214871484395,
      "grad_norm": 0.764944851398468,
      "learning_rate": 9.190430872548557e-06,
      "loss": 0.7762,
      "step": 6756
    },
    {
      "epoch": 0.37189718751719963,
      "grad_norm": 0.7362231612205505,
      "learning_rate": 9.190194384348199e-06,
      "loss": 0.8277,
      "step": 6757
    },
    {
      "epoch": 0.37195222631955527,
      "grad_norm": 0.7462226748466492,
      "learning_rate": 9.18995786465546e-06,
      "loss": 0.7362,
      "step": 6758
    },
    {
      "epoch": 0.37200726512191096,
      "grad_norm": 0.7769725322723389,
      "learning_rate": 9.18972131347212e-06,
      "loss": 0.8217,
      "step": 6759
    },
    {
      "epoch": 0.3720623039242666,
      "grad_norm": 0.7263969779014587,
      "learning_rate": 9.189484730799956e-06,
      "loss": 0.7719,
      "step": 6760
    },
    {
      "epoch": 0.3721173427266223,
      "grad_norm": 0.7612473964691162,
      "learning_rate": 9.189248116640746e-06,
      "loss": 0.7149,
      "step": 6761
    },
    {
      "epoch": 0.3721723815289779,
      "grad_norm": 0.6813042759895325,
      "learning_rate": 9.189011470996268e-06,
      "loss": 0.7119,
      "step": 6762
    },
    {
      "epoch": 0.3722274203313336,
      "grad_norm": 0.7376571297645569,
      "learning_rate": 9.188774793868302e-06,
      "loss": 0.7998,
      "step": 6763
    },
    {
      "epoch": 0.37228245913368924,
      "grad_norm": 0.8592102527618408,
      "learning_rate": 9.188538085258626e-06,
      "loss": 0.8026,
      "step": 6764
    },
    {
      "epoch": 0.3723374979360449,
      "grad_norm": 0.7666613459587097,
      "learning_rate": 9.188301345169017e-06,
      "loss": 0.8571,
      "step": 6765
    },
    {
      "epoch": 0.37239253673840056,
      "grad_norm": 0.7118985652923584,
      "learning_rate": 9.188064573601258e-06,
      "loss": 0.7637,
      "step": 6766
    },
    {
      "epoch": 0.37244757554075625,
      "grad_norm": 0.8247082233428955,
      "learning_rate": 9.187827770557127e-06,
      "loss": 0.8209,
      "step": 6767
    },
    {
      "epoch": 0.3725026143431119,
      "grad_norm": 0.7259567975997925,
      "learning_rate": 9.187590936038403e-06,
      "loss": 0.7918,
      "step": 6768
    },
    {
      "epoch": 0.37255765314546757,
      "grad_norm": 0.7409893274307251,
      "learning_rate": 9.187354070046867e-06,
      "loss": 0.8004,
      "step": 6769
    },
    {
      "epoch": 0.3726126919478232,
      "grad_norm": 0.8163084387779236,
      "learning_rate": 9.187117172584298e-06,
      "loss": 0.8452,
      "step": 6770
    },
    {
      "epoch": 0.3726677307501789,
      "grad_norm": 0.9241586923599243,
      "learning_rate": 9.186880243652477e-06,
      "loss": 0.8939,
      "step": 6771
    },
    {
      "epoch": 0.3727227695525345,
      "grad_norm": 0.710434079170227,
      "learning_rate": 9.186643283253185e-06,
      "loss": 0.7337,
      "step": 6772
    },
    {
      "epoch": 0.3727778083548902,
      "grad_norm": 0.7850505709648132,
      "learning_rate": 9.186406291388203e-06,
      "loss": 0.7892,
      "step": 6773
    },
    {
      "epoch": 0.37283284715724585,
      "grad_norm": 0.813979983329773,
      "learning_rate": 9.186169268059311e-06,
      "loss": 0.7993,
      "step": 6774
    },
    {
      "epoch": 0.37288788595960154,
      "grad_norm": 0.7923213243484497,
      "learning_rate": 9.185932213268292e-06,
      "loss": 0.7501,
      "step": 6775
    },
    {
      "epoch": 0.3729429247619572,
      "grad_norm": 0.7923155426979065,
      "learning_rate": 9.185695127016928e-06,
      "loss": 0.8435,
      "step": 6776
    },
    {
      "epoch": 0.37299796356431286,
      "grad_norm": 0.69893479347229,
      "learning_rate": 9.185458009306999e-06,
      "loss": 0.7155,
      "step": 6777
    },
    {
      "epoch": 0.3730530023666685,
      "grad_norm": 0.7848305106163025,
      "learning_rate": 9.185220860140289e-06,
      "loss": 0.7971,
      "step": 6778
    },
    {
      "epoch": 0.3731080411690242,
      "grad_norm": 0.6707655787467957,
      "learning_rate": 9.184983679518578e-06,
      "loss": 0.6939,
      "step": 6779
    },
    {
      "epoch": 0.3731630799713798,
      "grad_norm": 0.6612532734870911,
      "learning_rate": 9.18474646744365e-06,
      "loss": 0.7361,
      "step": 6780
    },
    {
      "epoch": 0.3732181187737355,
      "grad_norm": 0.7753985524177551,
      "learning_rate": 9.184509223917288e-06,
      "loss": 0.7263,
      "step": 6781
    },
    {
      "epoch": 0.37327315757609114,
      "grad_norm": 0.6856646537780762,
      "learning_rate": 9.184271948941275e-06,
      "loss": 0.6923,
      "step": 6782
    },
    {
      "epoch": 0.37332819637844683,
      "grad_norm": 0.7223647832870483,
      "learning_rate": 9.184034642517393e-06,
      "loss": 0.793,
      "step": 6783
    },
    {
      "epoch": 0.37338323518080246,
      "grad_norm": 0.7428838014602661,
      "learning_rate": 9.183797304647428e-06,
      "loss": 0.7781,
      "step": 6784
    },
    {
      "epoch": 0.37343827398315815,
      "grad_norm": 0.7301773428916931,
      "learning_rate": 9.183559935333161e-06,
      "loss": 0.7964,
      "step": 6785
    },
    {
      "epoch": 0.3734933127855138,
      "grad_norm": 0.7883384823799133,
      "learning_rate": 9.183322534576378e-06,
      "loss": 0.8904,
      "step": 6786
    },
    {
      "epoch": 0.3735483515878694,
      "grad_norm": 0.7943564653396606,
      "learning_rate": 9.183085102378864e-06,
      "loss": 0.7229,
      "step": 6787
    },
    {
      "epoch": 0.3736033903902251,
      "grad_norm": 0.7385129928588867,
      "learning_rate": 9.1828476387424e-06,
      "loss": 0.7967,
      "step": 6788
    },
    {
      "epoch": 0.37365842919258074,
      "grad_norm": 0.7968102097511292,
      "learning_rate": 9.182610143668775e-06,
      "loss": 0.8016,
      "step": 6789
    },
    {
      "epoch": 0.37371346799493643,
      "grad_norm": 0.7810283303260803,
      "learning_rate": 9.18237261715977e-06,
      "loss": 0.8956,
      "step": 6790
    },
    {
      "epoch": 0.37376850679729207,
      "grad_norm": 0.7110065221786499,
      "learning_rate": 9.182135059217172e-06,
      "loss": 0.7808,
      "step": 6791
    },
    {
      "epoch": 0.37382354559964776,
      "grad_norm": 0.7513633370399475,
      "learning_rate": 9.181897469842767e-06,
      "loss": 0.8236,
      "step": 6792
    },
    {
      "epoch": 0.3738785844020034,
      "grad_norm": 0.7850426435470581,
      "learning_rate": 9.18165984903834e-06,
      "loss": 0.8642,
      "step": 6793
    },
    {
      "epoch": 0.3739336232043591,
      "grad_norm": 1.4948225021362305,
      "learning_rate": 9.181422196805676e-06,
      "loss": 0.8765,
      "step": 6794
    },
    {
      "epoch": 0.3739886620067147,
      "grad_norm": 0.8242343068122864,
      "learning_rate": 9.181184513146563e-06,
      "loss": 0.7213,
      "step": 6795
    },
    {
      "epoch": 0.3740437008090704,
      "grad_norm": 0.8017476797103882,
      "learning_rate": 9.180946798062786e-06,
      "loss": 0.655,
      "step": 6796
    },
    {
      "epoch": 0.37409873961142603,
      "grad_norm": 0.9573387503623962,
      "learning_rate": 9.180709051556132e-06,
      "loss": 0.8674,
      "step": 6797
    },
    {
      "epoch": 0.3741537784137817,
      "grad_norm": 0.7575511932373047,
      "learning_rate": 9.180471273628388e-06,
      "loss": 0.8672,
      "step": 6798
    },
    {
      "epoch": 0.37420881721613736,
      "grad_norm": 0.7723323702812195,
      "learning_rate": 9.180233464281343e-06,
      "loss": 0.7698,
      "step": 6799
    },
    {
      "epoch": 0.37426385601849305,
      "grad_norm": 0.8352731466293335,
      "learning_rate": 9.17999562351678e-06,
      "loss": 0.9248,
      "step": 6800
    },
    {
      "epoch": 0.3743188948208487,
      "grad_norm": 0.7459322214126587,
      "learning_rate": 9.179757751336488e-06,
      "loss": 0.7561,
      "step": 6801
    },
    {
      "epoch": 0.37437393362320437,
      "grad_norm": 0.8053051829338074,
      "learning_rate": 9.179519847742257e-06,
      "loss": 0.8743,
      "step": 6802
    },
    {
      "epoch": 0.37442897242556,
      "grad_norm": 0.7781768441200256,
      "learning_rate": 9.179281912735873e-06,
      "loss": 0.7426,
      "step": 6803
    },
    {
      "epoch": 0.3744840112279157,
      "grad_norm": 0.6812007427215576,
      "learning_rate": 9.179043946319126e-06,
      "loss": 0.761,
      "step": 6804
    },
    {
      "epoch": 0.3745390500302713,
      "grad_norm": 0.8327108025550842,
      "learning_rate": 9.178805948493803e-06,
      "loss": 0.7633,
      "step": 6805
    },
    {
      "epoch": 0.374594088832627,
      "grad_norm": 0.7519007921218872,
      "learning_rate": 9.178567919261692e-06,
      "loss": 0.8268,
      "step": 6806
    },
    {
      "epoch": 0.37464912763498265,
      "grad_norm": 0.7507897019386292,
      "learning_rate": 9.178329858624584e-06,
      "loss": 0.8734,
      "step": 6807
    },
    {
      "epoch": 0.37470416643733834,
      "grad_norm": 0.6874666213989258,
      "learning_rate": 9.178091766584267e-06,
      "loss": 0.6669,
      "step": 6808
    },
    {
      "epoch": 0.37475920523969397,
      "grad_norm": 0.6987403631210327,
      "learning_rate": 9.17785364314253e-06,
      "loss": 0.7627,
      "step": 6809
    },
    {
      "epoch": 0.37481424404204966,
      "grad_norm": 0.7777343392372131,
      "learning_rate": 9.177615488301163e-06,
      "loss": 0.7637,
      "step": 6810
    },
    {
      "epoch": 0.3748692828444053,
      "grad_norm": 0.71980881690979,
      "learning_rate": 9.177377302061958e-06,
      "loss": 0.7964,
      "step": 6811
    },
    {
      "epoch": 0.374924321646761,
      "grad_norm": 0.627328634262085,
      "learning_rate": 9.177139084426704e-06,
      "loss": 0.6862,
      "step": 6812
    },
    {
      "epoch": 0.3749793604491166,
      "grad_norm": 0.7099852561950684,
      "learning_rate": 9.176900835397188e-06,
      "loss": 0.7592,
      "step": 6813
    },
    {
      "epoch": 0.3750343992514723,
      "grad_norm": 0.7880212664604187,
      "learning_rate": 9.176662554975205e-06,
      "loss": 0.756,
      "step": 6814
    },
    {
      "epoch": 0.37508943805382794,
      "grad_norm": 0.7347460389137268,
      "learning_rate": 9.176424243162546e-06,
      "loss": 0.8537,
      "step": 6815
    },
    {
      "epoch": 0.37514447685618363,
      "grad_norm": 0.7020999789237976,
      "learning_rate": 9.176185899960996e-06,
      "loss": 0.7844,
      "step": 6816
    },
    {
      "epoch": 0.37519951565853926,
      "grad_norm": 0.6857696175575256,
      "learning_rate": 9.175947525372355e-06,
      "loss": 0.8491,
      "step": 6817
    },
    {
      "epoch": 0.37525455446089495,
      "grad_norm": 0.6882391571998596,
      "learning_rate": 9.175709119398409e-06,
      "loss": 0.7797,
      "step": 6818
    },
    {
      "epoch": 0.3753095932632506,
      "grad_norm": 0.7788485288619995,
      "learning_rate": 9.17547068204095e-06,
      "loss": 0.6898,
      "step": 6819
    },
    {
      "epoch": 0.3753646320656063,
      "grad_norm": 0.8529300093650818,
      "learning_rate": 9.17523221330177e-06,
      "loss": 0.8113,
      "step": 6820
    },
    {
      "epoch": 0.3754196708679619,
      "grad_norm": 0.6297540068626404,
      "learning_rate": 9.174993713182663e-06,
      "loss": 0.7133,
      "step": 6821
    },
    {
      "epoch": 0.3754747096703176,
      "grad_norm": 0.8225051760673523,
      "learning_rate": 9.174755181685422e-06,
      "loss": 0.83,
      "step": 6822
    },
    {
      "epoch": 0.37552974847267323,
      "grad_norm": 0.7445290684700012,
      "learning_rate": 9.174516618811838e-06,
      "loss": 0.8597,
      "step": 6823
    },
    {
      "epoch": 0.3755847872750289,
      "grad_norm": 0.7890744209289551,
      "learning_rate": 9.174278024563706e-06,
      "loss": 0.8021,
      "step": 6824
    },
    {
      "epoch": 0.37563982607738455,
      "grad_norm": 0.644434928894043,
      "learning_rate": 9.174039398942815e-06,
      "loss": 0.7154,
      "step": 6825
    },
    {
      "epoch": 0.37569486487974024,
      "grad_norm": 0.7664980292320251,
      "learning_rate": 9.173800741950962e-06,
      "loss": 0.8496,
      "step": 6826
    },
    {
      "epoch": 0.3757499036820959,
      "grad_norm": 0.8062339425086975,
      "learning_rate": 9.173562053589942e-06,
      "loss": 0.7736,
      "step": 6827
    },
    {
      "epoch": 0.37580494248445157,
      "grad_norm": 0.6334213018417358,
      "learning_rate": 9.173323333861543e-06,
      "loss": 0.6513,
      "step": 6828
    },
    {
      "epoch": 0.3758599812868072,
      "grad_norm": 0.6825501322746277,
      "learning_rate": 9.173084582767567e-06,
      "loss": 0.755,
      "step": 6829
    },
    {
      "epoch": 0.37591502008916283,
      "grad_norm": 0.7353835105895996,
      "learning_rate": 9.172845800309801e-06,
      "loss": 0.7783,
      "step": 6830
    },
    {
      "epoch": 0.3759700588915185,
      "grad_norm": 0.7830193638801575,
      "learning_rate": 9.172606986490046e-06,
      "loss": 0.7352,
      "step": 6831
    },
    {
      "epoch": 0.37602509769387416,
      "grad_norm": 0.7464943528175354,
      "learning_rate": 9.172368141310091e-06,
      "loss": 0.6454,
      "step": 6832
    },
    {
      "epoch": 0.37608013649622984,
      "grad_norm": 0.7171493172645569,
      "learning_rate": 9.172129264771736e-06,
      "loss": 0.7978,
      "step": 6833
    },
    {
      "epoch": 0.3761351752985855,
      "grad_norm": 0.6929624676704407,
      "learning_rate": 9.171890356876774e-06,
      "loss": 0.8026,
      "step": 6834
    },
    {
      "epoch": 0.37619021410094117,
      "grad_norm": 0.7240758538246155,
      "learning_rate": 9.171651417627e-06,
      "loss": 0.8469,
      "step": 6835
    },
    {
      "epoch": 0.3762452529032968,
      "grad_norm": 0.7713736891746521,
      "learning_rate": 9.17141244702421e-06,
      "loss": 0.8307,
      "step": 6836
    },
    {
      "epoch": 0.3763002917056525,
      "grad_norm": 0.7417639493942261,
      "learning_rate": 9.171173445070203e-06,
      "loss": 0.8165,
      "step": 6837
    },
    {
      "epoch": 0.3763553305080081,
      "grad_norm": 0.811005711555481,
      "learning_rate": 9.17093441176677e-06,
      "loss": 0.8418,
      "step": 6838
    },
    {
      "epoch": 0.3764103693103638,
      "grad_norm": 0.9996818900108337,
      "learning_rate": 9.170695347115713e-06,
      "loss": 0.851,
      "step": 6839
    },
    {
      "epoch": 0.37646540811271945,
      "grad_norm": 0.7703381776809692,
      "learning_rate": 9.170456251118824e-06,
      "loss": 0.8308,
      "step": 6840
    },
    {
      "epoch": 0.37652044691507514,
      "grad_norm": 0.7194466590881348,
      "learning_rate": 9.170217123777904e-06,
      "loss": 0.699,
      "step": 6841
    },
    {
      "epoch": 0.37657548571743077,
      "grad_norm": 0.7146462202072144,
      "learning_rate": 9.169977965094748e-06,
      "loss": 0.8247,
      "step": 6842
    },
    {
      "epoch": 0.37663052451978646,
      "grad_norm": 0.7490555047988892,
      "learning_rate": 9.169738775071153e-06,
      "loss": 0.8627,
      "step": 6843
    },
    {
      "epoch": 0.3766855633221421,
      "grad_norm": 0.827996015548706,
      "learning_rate": 9.169499553708919e-06,
      "loss": 0.7454,
      "step": 6844
    },
    {
      "epoch": 0.3767406021244978,
      "grad_norm": 0.7185913324356079,
      "learning_rate": 9.16926030100984e-06,
      "loss": 0.7018,
      "step": 6845
    },
    {
      "epoch": 0.3767956409268534,
      "grad_norm": 0.7879654169082642,
      "learning_rate": 9.169021016975718e-06,
      "loss": 0.8144,
      "step": 6846
    },
    {
      "epoch": 0.3768506797292091,
      "grad_norm": 0.7072417736053467,
      "learning_rate": 9.168781701608352e-06,
      "loss": 0.7572,
      "step": 6847
    },
    {
      "epoch": 0.37690571853156474,
      "grad_norm": 0.7359803915023804,
      "learning_rate": 9.168542354909536e-06,
      "loss": 0.7712,
      "step": 6848
    },
    {
      "epoch": 0.3769607573339204,
      "grad_norm": 0.7672479748725891,
      "learning_rate": 9.168302976881072e-06,
      "loss": 0.7696,
      "step": 6849
    },
    {
      "epoch": 0.37701579613627606,
      "grad_norm": 0.7276006937026978,
      "learning_rate": 9.168063567524758e-06,
      "loss": 0.8235,
      "step": 6850
    },
    {
      "epoch": 0.37707083493863175,
      "grad_norm": 0.673577606678009,
      "learning_rate": 9.167824126842396e-06,
      "loss": 0.6515,
      "step": 6851
    },
    {
      "epoch": 0.3771258737409874,
      "grad_norm": 0.7257997989654541,
      "learning_rate": 9.167584654835782e-06,
      "loss": 0.729,
      "step": 6852
    },
    {
      "epoch": 0.37718091254334307,
      "grad_norm": 0.6655071377754211,
      "learning_rate": 9.167345151506717e-06,
      "loss": 0.7917,
      "step": 6853
    },
    {
      "epoch": 0.3772359513456987,
      "grad_norm": 0.7603726983070374,
      "learning_rate": 9.167105616857002e-06,
      "loss": 0.8383,
      "step": 6854
    },
    {
      "epoch": 0.3772909901480544,
      "grad_norm": 0.7066939473152161,
      "learning_rate": 9.166866050888437e-06,
      "loss": 0.7589,
      "step": 6855
    },
    {
      "epoch": 0.37734602895041003,
      "grad_norm": 0.7002355456352234,
      "learning_rate": 9.16662645360282e-06,
      "loss": 0.8305,
      "step": 6856
    },
    {
      "epoch": 0.3774010677527657,
      "grad_norm": 0.9499780535697937,
      "learning_rate": 9.166386825001957e-06,
      "loss": 0.78,
      "step": 6857
    },
    {
      "epoch": 0.37745610655512135,
      "grad_norm": 0.7136938571929932,
      "learning_rate": 9.166147165087645e-06,
      "loss": 0.7449,
      "step": 6858
    },
    {
      "epoch": 0.37751114535747704,
      "grad_norm": 0.740443766117096,
      "learning_rate": 9.165907473861687e-06,
      "loss": 0.8228,
      "step": 6859
    },
    {
      "epoch": 0.3775661841598327,
      "grad_norm": 0.7649856209754944,
      "learning_rate": 9.165667751325879e-06,
      "loss": 0.7762,
      "step": 6860
    },
    {
      "epoch": 0.37762122296218836,
      "grad_norm": 0.743251383304596,
      "learning_rate": 9.165427997482032e-06,
      "loss": 0.7536,
      "step": 6861
    },
    {
      "epoch": 0.377676261764544,
      "grad_norm": 0.7023851871490479,
      "learning_rate": 9.165188212331941e-06,
      "loss": 0.7327,
      "step": 6862
    },
    {
      "epoch": 0.3777313005668997,
      "grad_norm": 0.7304333448410034,
      "learning_rate": 9.164948395877411e-06,
      "loss": 0.8816,
      "step": 6863
    },
    {
      "epoch": 0.3777863393692553,
      "grad_norm": 0.6666659116744995,
      "learning_rate": 9.164708548120244e-06,
      "loss": 0.7821,
      "step": 6864
    },
    {
      "epoch": 0.377841378171611,
      "grad_norm": 0.6542865037918091,
      "learning_rate": 9.164468669062242e-06,
      "loss": 0.7044,
      "step": 6865
    },
    {
      "epoch": 0.37789641697396664,
      "grad_norm": 0.7436043620109558,
      "learning_rate": 9.16422875870521e-06,
      "loss": 0.8492,
      "step": 6866
    },
    {
      "epoch": 0.37795145577632233,
      "grad_norm": 0.7660424709320068,
      "learning_rate": 9.163988817050947e-06,
      "loss": 0.7236,
      "step": 6867
    },
    {
      "epoch": 0.37800649457867797,
      "grad_norm": 0.7288914918899536,
      "learning_rate": 9.16374884410126e-06,
      "loss": 0.6361,
      "step": 6868
    },
    {
      "epoch": 0.37806153338103365,
      "grad_norm": 0.884832501411438,
      "learning_rate": 9.163508839857948e-06,
      "loss": 0.8112,
      "step": 6869
    },
    {
      "epoch": 0.3781165721833893,
      "grad_norm": 0.937660813331604,
      "learning_rate": 9.163268804322822e-06,
      "loss": 0.6405,
      "step": 6870
    },
    {
      "epoch": 0.378171610985745,
      "grad_norm": 0.8295212388038635,
      "learning_rate": 9.16302873749768e-06,
      "loss": 0.8107,
      "step": 6871
    },
    {
      "epoch": 0.3782266497881006,
      "grad_norm": 1.0573647022247314,
      "learning_rate": 9.16278863938433e-06,
      "loss": 0.7792,
      "step": 6872
    },
    {
      "epoch": 0.37828168859045624,
      "grad_norm": 0.8450027108192444,
      "learning_rate": 9.162548509984574e-06,
      "loss": 0.8103,
      "step": 6873
    },
    {
      "epoch": 0.37833672739281193,
      "grad_norm": 0.7372947931289673,
      "learning_rate": 9.162308349300218e-06,
      "loss": 0.8232,
      "step": 6874
    },
    {
      "epoch": 0.37839176619516757,
      "grad_norm": 0.7573776841163635,
      "learning_rate": 9.162068157333066e-06,
      "loss": 0.773,
      "step": 6875
    },
    {
      "epoch": 0.37844680499752326,
      "grad_norm": 0.7883201241493225,
      "learning_rate": 9.161827934084924e-06,
      "loss": 0.7561,
      "step": 6876
    },
    {
      "epoch": 0.3785018437998789,
      "grad_norm": 0.7195025086402893,
      "learning_rate": 9.161587679557598e-06,
      "loss": 0.798,
      "step": 6877
    },
    {
      "epoch": 0.3785568826022346,
      "grad_norm": 0.7047843337059021,
      "learning_rate": 9.161347393752891e-06,
      "loss": 0.8122,
      "step": 6878
    },
    {
      "epoch": 0.3786119214045902,
      "grad_norm": 0.7354363203048706,
      "learning_rate": 9.161107076672613e-06,
      "loss": 0.7296,
      "step": 6879
    },
    {
      "epoch": 0.3786669602069459,
      "grad_norm": 0.7748313546180725,
      "learning_rate": 9.160866728318567e-06,
      "loss": 0.9576,
      "step": 6880
    },
    {
      "epoch": 0.37872199900930154,
      "grad_norm": 0.7197638750076294,
      "learning_rate": 9.16062634869256e-06,
      "loss": 0.8054,
      "step": 6881
    },
    {
      "epoch": 0.3787770378116572,
      "grad_norm": 0.7086492776870728,
      "learning_rate": 9.1603859377964e-06,
      "loss": 0.8938,
      "step": 6882
    },
    {
      "epoch": 0.37883207661401286,
      "grad_norm": 0.7764425873756409,
      "learning_rate": 9.160145495631894e-06,
      "loss": 0.7562,
      "step": 6883
    },
    {
      "epoch": 0.37888711541636855,
      "grad_norm": 0.7673479914665222,
      "learning_rate": 9.159905022200846e-06,
      "loss": 0.6783,
      "step": 6884
    },
    {
      "epoch": 0.3789421542187242,
      "grad_norm": 0.7323669195175171,
      "learning_rate": 9.159664517505067e-06,
      "loss": 0.8274,
      "step": 6885
    },
    {
      "epoch": 0.37899719302107987,
      "grad_norm": 0.8283136487007141,
      "learning_rate": 9.159423981546362e-06,
      "loss": 0.7184,
      "step": 6886
    },
    {
      "epoch": 0.3790522318234355,
      "grad_norm": 0.6949145793914795,
      "learning_rate": 9.15918341432654e-06,
      "loss": 0.7843,
      "step": 6887
    },
    {
      "epoch": 0.3791072706257912,
      "grad_norm": 0.8584639430046082,
      "learning_rate": 9.158942815847408e-06,
      "loss": 0.71,
      "step": 6888
    },
    {
      "epoch": 0.3791623094281468,
      "grad_norm": 0.7125271558761597,
      "learning_rate": 9.158702186110777e-06,
      "loss": 0.7432,
      "step": 6889
    },
    {
      "epoch": 0.3792173482305025,
      "grad_norm": 0.6657430529594421,
      "learning_rate": 9.158461525118452e-06,
      "loss": 0.6715,
      "step": 6890
    },
    {
      "epoch": 0.37927238703285815,
      "grad_norm": 0.770226240158081,
      "learning_rate": 9.158220832872243e-06,
      "loss": 0.7029,
      "step": 6891
    },
    {
      "epoch": 0.37932742583521384,
      "grad_norm": 0.7697272300720215,
      "learning_rate": 9.15798010937396e-06,
      "loss": 0.686,
      "step": 6892
    },
    {
      "epoch": 0.37938246463756947,
      "grad_norm": 0.7693290710449219,
      "learning_rate": 9.157739354625413e-06,
      "loss": 0.7669,
      "step": 6893
    },
    {
      "epoch": 0.37943750343992516,
      "grad_norm": 0.8365996479988098,
      "learning_rate": 9.157498568628406e-06,
      "loss": 0.8254,
      "step": 6894
    },
    {
      "epoch": 0.3794925422422808,
      "grad_norm": 0.8075883388519287,
      "learning_rate": 9.157257751384756e-06,
      "loss": 0.8311,
      "step": 6895
    },
    {
      "epoch": 0.3795475810446365,
      "grad_norm": 0.8422812819480896,
      "learning_rate": 9.15701690289627e-06,
      "loss": 0.9173,
      "step": 6896
    },
    {
      "epoch": 0.3796026198469921,
      "grad_norm": 0.7930355072021484,
      "learning_rate": 9.156776023164755e-06,
      "loss": 0.9376,
      "step": 6897
    },
    {
      "epoch": 0.3796576586493478,
      "grad_norm": 0.7877563238143921,
      "learning_rate": 9.156535112192026e-06,
      "loss": 0.8358,
      "step": 6898
    },
    {
      "epoch": 0.37971269745170344,
      "grad_norm": 0.7712885141372681,
      "learning_rate": 9.156294169979891e-06,
      "loss": 0.8781,
      "step": 6899
    },
    {
      "epoch": 0.37976773625405913,
      "grad_norm": 0.6953728199005127,
      "learning_rate": 9.156053196530162e-06,
      "loss": 0.7861,
      "step": 6900
    },
    {
      "epoch": 0.37982277505641476,
      "grad_norm": 0.9581564664840698,
      "learning_rate": 9.155812191844649e-06,
      "loss": 0.8294,
      "step": 6901
    },
    {
      "epoch": 0.37987781385877045,
      "grad_norm": 0.738571286201477,
      "learning_rate": 9.155571155925166e-06,
      "loss": 0.7998,
      "step": 6902
    },
    {
      "epoch": 0.3799328526611261,
      "grad_norm": 0.7059765458106995,
      "learning_rate": 9.155330088773519e-06,
      "loss": 0.7877,
      "step": 6903
    },
    {
      "epoch": 0.3799878914634818,
      "grad_norm": 0.8572642207145691,
      "learning_rate": 9.155088990391527e-06,
      "loss": 0.7333,
      "step": 6904
    },
    {
      "epoch": 0.3800429302658374,
      "grad_norm": 0.7442637085914612,
      "learning_rate": 9.154847860780996e-06,
      "loss": 0.685,
      "step": 6905
    },
    {
      "epoch": 0.3800979690681931,
      "grad_norm": 0.7787682414054871,
      "learning_rate": 9.154606699943741e-06,
      "loss": 0.7893,
      "step": 6906
    },
    {
      "epoch": 0.38015300787054873,
      "grad_norm": 0.8973822593688965,
      "learning_rate": 9.154365507881574e-06,
      "loss": 0.8297,
      "step": 6907
    },
    {
      "epoch": 0.3802080466729044,
      "grad_norm": 0.7759919166564941,
      "learning_rate": 9.154124284596311e-06,
      "loss": 0.8257,
      "step": 6908
    },
    {
      "epoch": 0.38026308547526005,
      "grad_norm": 0.8042850494384766,
      "learning_rate": 9.153883030089759e-06,
      "loss": 0.8024,
      "step": 6909
    },
    {
      "epoch": 0.38031812427761574,
      "grad_norm": 0.8285790085792542,
      "learning_rate": 9.153641744363733e-06,
      "loss": 0.7824,
      "step": 6910
    },
    {
      "epoch": 0.3803731630799714,
      "grad_norm": 0.7225445508956909,
      "learning_rate": 9.15340042742005e-06,
      "loss": 0.8065,
      "step": 6911
    },
    {
      "epoch": 0.38042820188232707,
      "grad_norm": 0.7685298919677734,
      "learning_rate": 9.15315907926052e-06,
      "loss": 0.8151,
      "step": 6912
    },
    {
      "epoch": 0.3804832406846827,
      "grad_norm": 0.9005589485168457,
      "learning_rate": 9.152917699886958e-06,
      "loss": 0.8413,
      "step": 6913
    },
    {
      "epoch": 0.3805382794870384,
      "grad_norm": 0.8715279698371887,
      "learning_rate": 9.152676289301178e-06,
      "loss": 0.7233,
      "step": 6914
    },
    {
      "epoch": 0.380593318289394,
      "grad_norm": 0.8764133453369141,
      "learning_rate": 9.152434847504996e-06,
      "loss": 0.783,
      "step": 6915
    },
    {
      "epoch": 0.38064835709174966,
      "grad_norm": 0.6847019195556641,
      "learning_rate": 9.152193374500225e-06,
      "loss": 0.7133,
      "step": 6916
    },
    {
      "epoch": 0.38070339589410535,
      "grad_norm": 0.7562721371650696,
      "learning_rate": 9.151951870288678e-06,
      "loss": 0.8155,
      "step": 6917
    },
    {
      "epoch": 0.380758434696461,
      "grad_norm": 0.6888439059257507,
      "learning_rate": 9.151710334872173e-06,
      "loss": 0.6395,
      "step": 6918
    },
    {
      "epoch": 0.38081347349881667,
      "grad_norm": 1.0951511859893799,
      "learning_rate": 9.151468768252525e-06,
      "loss": 0.8936,
      "step": 6919
    },
    {
      "epoch": 0.3808685123011723,
      "grad_norm": 0.7261115908622742,
      "learning_rate": 9.151227170431549e-06,
      "loss": 0.7864,
      "step": 6920
    },
    {
      "epoch": 0.380923551103528,
      "grad_norm": 1.2851859331130981,
      "learning_rate": 9.150985541411061e-06,
      "loss": 0.9419,
      "step": 6921
    },
    {
      "epoch": 0.3809785899058836,
      "grad_norm": 0.7621721625328064,
      "learning_rate": 9.150743881192876e-06,
      "loss": 0.7773,
      "step": 6922
    },
    {
      "epoch": 0.3810336287082393,
      "grad_norm": 0.7605605721473694,
      "learning_rate": 9.150502189778811e-06,
      "loss": 0.8752,
      "step": 6923
    },
    {
      "epoch": 0.38108866751059495,
      "grad_norm": 0.8422327041625977,
      "learning_rate": 9.150260467170683e-06,
      "loss": 0.8555,
      "step": 6924
    },
    {
      "epoch": 0.38114370631295064,
      "grad_norm": 0.7227829098701477,
      "learning_rate": 9.15001871337031e-06,
      "loss": 0.7637,
      "step": 6925
    },
    {
      "epoch": 0.38119874511530627,
      "grad_norm": 0.6568942666053772,
      "learning_rate": 9.149776928379506e-06,
      "loss": 0.6944,
      "step": 6926
    },
    {
      "epoch": 0.38125378391766196,
      "grad_norm": 0.9317567944526672,
      "learning_rate": 9.149535112200087e-06,
      "loss": 0.8098,
      "step": 6927
    },
    {
      "epoch": 0.3813088227200176,
      "grad_norm": 0.6374759674072266,
      "learning_rate": 9.149293264833877e-06,
      "loss": 0.6654,
      "step": 6928
    },
    {
      "epoch": 0.3813638615223733,
      "grad_norm": 0.7276837825775146,
      "learning_rate": 9.149051386282685e-06,
      "loss": 0.7728,
      "step": 6929
    },
    {
      "epoch": 0.3814189003247289,
      "grad_norm": 0.7573683261871338,
      "learning_rate": 9.148809476548337e-06,
      "loss": 0.7681,
      "step": 6930
    },
    {
      "epoch": 0.3814739391270846,
      "grad_norm": 0.7535703778266907,
      "learning_rate": 9.148567535632647e-06,
      "loss": 0.8498,
      "step": 6931
    },
    {
      "epoch": 0.38152897792944024,
      "grad_norm": 0.7510126233100891,
      "learning_rate": 9.148325563537432e-06,
      "loss": 0.7874,
      "step": 6932
    },
    {
      "epoch": 0.3815840167317959,
      "grad_norm": 0.7809224724769592,
      "learning_rate": 9.148083560264515e-06,
      "loss": 0.7223,
      "step": 6933
    },
    {
      "epoch": 0.38163905553415156,
      "grad_norm": 0.7433155179023743,
      "learning_rate": 9.14784152581571e-06,
      "loss": 0.7914,
      "step": 6934
    },
    {
      "epoch": 0.38169409433650725,
      "grad_norm": 0.7142858505249023,
      "learning_rate": 9.14759946019284e-06,
      "loss": 0.781,
      "step": 6935
    },
    {
      "epoch": 0.3817491331388629,
      "grad_norm": 0.7910202741622925,
      "learning_rate": 9.147357363397721e-06,
      "loss": 0.755,
      "step": 6936
    },
    {
      "epoch": 0.3818041719412186,
      "grad_norm": 1.007727026939392,
      "learning_rate": 9.147115235432176e-06,
      "loss": 0.7809,
      "step": 6937
    },
    {
      "epoch": 0.3818592107435742,
      "grad_norm": 0.7227005362510681,
      "learning_rate": 9.146873076298024e-06,
      "loss": 0.7276,
      "step": 6938
    },
    {
      "epoch": 0.3819142495459299,
      "grad_norm": 0.6945967674255371,
      "learning_rate": 9.146630885997081e-06,
      "loss": 0.825,
      "step": 6939
    },
    {
      "epoch": 0.38196928834828553,
      "grad_norm": 0.6719669103622437,
      "learning_rate": 9.146388664531172e-06,
      "loss": 0.6486,
      "step": 6940
    },
    {
      "epoch": 0.3820243271506412,
      "grad_norm": 0.7528467178344727,
      "learning_rate": 9.146146411902115e-06,
      "loss": 0.8143,
      "step": 6941
    },
    {
      "epoch": 0.38207936595299685,
      "grad_norm": 0.6835548877716064,
      "learning_rate": 9.145904128111732e-06,
      "loss": 0.7742,
      "step": 6942
    },
    {
      "epoch": 0.38213440475535254,
      "grad_norm": 0.7829870581626892,
      "learning_rate": 9.145661813161844e-06,
      "loss": 0.8147,
      "step": 6943
    },
    {
      "epoch": 0.3821894435577082,
      "grad_norm": 0.6833155155181885,
      "learning_rate": 9.145419467054271e-06,
      "loss": 0.7615,
      "step": 6944
    },
    {
      "epoch": 0.38224448236006386,
      "grad_norm": 0.7577275037765503,
      "learning_rate": 9.145177089790833e-06,
      "loss": 0.8611,
      "step": 6945
    },
    {
      "epoch": 0.3822995211624195,
      "grad_norm": 0.7102984189987183,
      "learning_rate": 9.144934681373356e-06,
      "loss": 0.8373,
      "step": 6946
    },
    {
      "epoch": 0.3823545599647752,
      "grad_norm": 0.6906121373176575,
      "learning_rate": 9.144692241803658e-06,
      "loss": 0.8314,
      "step": 6947
    },
    {
      "epoch": 0.3824095987671308,
      "grad_norm": 0.7790967226028442,
      "learning_rate": 9.144449771083563e-06,
      "loss": 0.8285,
      "step": 6948
    },
    {
      "epoch": 0.3824646375694865,
      "grad_norm": 0.8420237898826599,
      "learning_rate": 9.144207269214893e-06,
      "loss": 0.8159,
      "step": 6949
    },
    {
      "epoch": 0.38251967637184214,
      "grad_norm": 0.7944310307502747,
      "learning_rate": 9.143964736199471e-06,
      "loss": 0.7981,
      "step": 6950
    },
    {
      "epoch": 0.38257471517419783,
      "grad_norm": 0.7610076069831848,
      "learning_rate": 9.14372217203912e-06,
      "loss": 0.8011,
      "step": 6951
    },
    {
      "epoch": 0.38262975397655347,
      "grad_norm": 0.7183333039283752,
      "learning_rate": 9.143479576735661e-06,
      "loss": 0.7504,
      "step": 6952
    },
    {
      "epoch": 0.38268479277890916,
      "grad_norm": 0.7363573312759399,
      "learning_rate": 9.14323695029092e-06,
      "loss": 0.7561,
      "step": 6953
    },
    {
      "epoch": 0.3827398315812648,
      "grad_norm": 0.7330427765846252,
      "learning_rate": 9.142994292706716e-06,
      "loss": 0.754,
      "step": 6954
    },
    {
      "epoch": 0.3827948703836205,
      "grad_norm": 0.8307509422302246,
      "learning_rate": 9.142751603984879e-06,
      "loss": 0.8059,
      "step": 6955
    },
    {
      "epoch": 0.3828499091859761,
      "grad_norm": 0.7340347766876221,
      "learning_rate": 9.142508884127228e-06,
      "loss": 0.8636,
      "step": 6956
    },
    {
      "epoch": 0.3829049479883318,
      "grad_norm": 0.7032678127288818,
      "learning_rate": 9.14226613313559e-06,
      "loss": 0.8237,
      "step": 6957
    },
    {
      "epoch": 0.38295998679068743,
      "grad_norm": 0.769809365272522,
      "learning_rate": 9.142023351011788e-06,
      "loss": 0.7523,
      "step": 6958
    },
    {
      "epoch": 0.38301502559304307,
      "grad_norm": 0.7446833252906799,
      "learning_rate": 9.141780537757647e-06,
      "loss": 0.8382,
      "step": 6959
    },
    {
      "epoch": 0.38307006439539876,
      "grad_norm": 0.6926285028457642,
      "learning_rate": 9.141537693374994e-06,
      "loss": 0.7997,
      "step": 6960
    },
    {
      "epoch": 0.3831251031977544,
      "grad_norm": 0.7303034067153931,
      "learning_rate": 9.141294817865651e-06,
      "loss": 0.794,
      "step": 6961
    },
    {
      "epoch": 0.3831801420001101,
      "grad_norm": 0.7453297972679138,
      "learning_rate": 9.141051911231445e-06,
      "loss": 0.7031,
      "step": 6962
    },
    {
      "epoch": 0.3832351808024657,
      "grad_norm": 0.8503912091255188,
      "learning_rate": 9.140808973474201e-06,
      "loss": 0.7855,
      "step": 6963
    },
    {
      "epoch": 0.3832902196048214,
      "grad_norm": 0.7304036617279053,
      "learning_rate": 9.140566004595746e-06,
      "loss": 0.7062,
      "step": 6964
    },
    {
      "epoch": 0.38334525840717704,
      "grad_norm": 0.7534968852996826,
      "learning_rate": 9.140323004597904e-06,
      "loss": 0.8138,
      "step": 6965
    },
    {
      "epoch": 0.3834002972095327,
      "grad_norm": 0.8122013807296753,
      "learning_rate": 9.140079973482503e-06,
      "loss": 0.7769,
      "step": 6966
    },
    {
      "epoch": 0.38345533601188836,
      "grad_norm": 0.7345744967460632,
      "learning_rate": 9.13983691125137e-06,
      "loss": 0.7588,
      "step": 6967
    },
    {
      "epoch": 0.38351037481424405,
      "grad_norm": 0.7251620292663574,
      "learning_rate": 9.13959381790633e-06,
      "loss": 0.8027,
      "step": 6968
    },
    {
      "epoch": 0.3835654136165997,
      "grad_norm": 0.7157594561576843,
      "learning_rate": 9.139350693449212e-06,
      "loss": 0.7233,
      "step": 6969
    },
    {
      "epoch": 0.38362045241895537,
      "grad_norm": 0.8076621890068054,
      "learning_rate": 9.139107537881842e-06,
      "loss": 0.7256,
      "step": 6970
    },
    {
      "epoch": 0.383675491221311,
      "grad_norm": 0.717182993888855,
      "learning_rate": 9.138864351206047e-06,
      "loss": 0.7003,
      "step": 6971
    },
    {
      "epoch": 0.3837305300236667,
      "grad_norm": 0.7534194588661194,
      "learning_rate": 9.138621133423656e-06,
      "loss": 0.7315,
      "step": 6972
    },
    {
      "epoch": 0.3837855688260223,
      "grad_norm": 0.6400160193443298,
      "learning_rate": 9.138377884536494e-06,
      "loss": 0.6814,
      "step": 6973
    },
    {
      "epoch": 0.383840607628378,
      "grad_norm": 0.7319507002830505,
      "learning_rate": 9.138134604546394e-06,
      "loss": 0.7942,
      "step": 6974
    },
    {
      "epoch": 0.38389564643073365,
      "grad_norm": 0.7109829783439636,
      "learning_rate": 9.137891293455181e-06,
      "loss": 0.7528,
      "step": 6975
    },
    {
      "epoch": 0.38395068523308934,
      "grad_norm": 1.006724238395691,
      "learning_rate": 9.137647951264685e-06,
      "loss": 0.7652,
      "step": 6976
    },
    {
      "epoch": 0.384005724035445,
      "grad_norm": 0.7080540060997009,
      "learning_rate": 9.137404577976736e-06,
      "loss": 0.7706,
      "step": 6977
    },
    {
      "epoch": 0.38406076283780066,
      "grad_norm": 0.7551368474960327,
      "learning_rate": 9.137161173593161e-06,
      "loss": 0.8202,
      "step": 6978
    },
    {
      "epoch": 0.3841158016401563,
      "grad_norm": 0.6624314785003662,
      "learning_rate": 9.13691773811579e-06,
      "loss": 0.7258,
      "step": 6979
    },
    {
      "epoch": 0.384170840442512,
      "grad_norm": 0.9603848457336426,
      "learning_rate": 9.136674271546451e-06,
      "loss": 0.9415,
      "step": 6980
    },
    {
      "epoch": 0.3842258792448676,
      "grad_norm": 0.6964829564094543,
      "learning_rate": 9.136430773886977e-06,
      "loss": 0.7604,
      "step": 6981
    },
    {
      "epoch": 0.3842809180472233,
      "grad_norm": 0.6503588557243347,
      "learning_rate": 9.136187245139197e-06,
      "loss": 0.7141,
      "step": 6982
    },
    {
      "epoch": 0.38433595684957894,
      "grad_norm": 0.9179829359054565,
      "learning_rate": 9.13594368530494e-06,
      "loss": 0.7619,
      "step": 6983
    },
    {
      "epoch": 0.38439099565193463,
      "grad_norm": 0.7993278503417969,
      "learning_rate": 9.135700094386038e-06,
      "loss": 0.832,
      "step": 6984
    },
    {
      "epoch": 0.38444603445429026,
      "grad_norm": 0.8136988282203674,
      "learning_rate": 9.13545647238432e-06,
      "loss": 0.8127,
      "step": 6985
    },
    {
      "epoch": 0.38450107325664595,
      "grad_norm": 0.9918104410171509,
      "learning_rate": 9.135212819301619e-06,
      "loss": 0.836,
      "step": 6986
    },
    {
      "epoch": 0.3845561120590016,
      "grad_norm": 0.7767511010169983,
      "learning_rate": 9.134969135139765e-06,
      "loss": 0.8391,
      "step": 6987
    },
    {
      "epoch": 0.3846111508613573,
      "grad_norm": 0.6889285445213318,
      "learning_rate": 9.134725419900589e-06,
      "loss": 0.7639,
      "step": 6988
    },
    {
      "epoch": 0.3846661896637129,
      "grad_norm": 1.803467035293579,
      "learning_rate": 9.134481673585924e-06,
      "loss": 0.7629,
      "step": 6989
    },
    {
      "epoch": 0.3847212284660686,
      "grad_norm": 0.721581757068634,
      "learning_rate": 9.134237896197603e-06,
      "loss": 0.8194,
      "step": 6990
    },
    {
      "epoch": 0.38477626726842423,
      "grad_norm": 0.8163189888000488,
      "learning_rate": 9.133994087737456e-06,
      "loss": 0.7789,
      "step": 6991
    },
    {
      "epoch": 0.3848313060707799,
      "grad_norm": 0.7518420815467834,
      "learning_rate": 9.133750248207315e-06,
      "loss": 0.7529,
      "step": 6992
    },
    {
      "epoch": 0.38488634487313556,
      "grad_norm": 0.7318000197410583,
      "learning_rate": 9.133506377609015e-06,
      "loss": 0.7829,
      "step": 6993
    },
    {
      "epoch": 0.38494138367549124,
      "grad_norm": 0.7765058875083923,
      "learning_rate": 9.133262475944386e-06,
      "loss": 0.7902,
      "step": 6994
    },
    {
      "epoch": 0.3849964224778469,
      "grad_norm": 0.845567524433136,
      "learning_rate": 9.133018543215265e-06,
      "loss": 0.8117,
      "step": 6995
    },
    {
      "epoch": 0.38505146128020257,
      "grad_norm": 0.7081887125968933,
      "learning_rate": 9.13277457942348e-06,
      "loss": 0.8131,
      "step": 6996
    },
    {
      "epoch": 0.3851065000825582,
      "grad_norm": 0.7447869777679443,
      "learning_rate": 9.132530584570869e-06,
      "loss": 0.7765,
      "step": 6997
    },
    {
      "epoch": 0.3851615388849139,
      "grad_norm": 0.8554795384407043,
      "learning_rate": 9.132286558659265e-06,
      "loss": 0.8966,
      "step": 6998
    },
    {
      "epoch": 0.3852165776872695,
      "grad_norm": 0.7117023468017578,
      "learning_rate": 9.1320425016905e-06,
      "loss": 0.7461,
      "step": 6999
    },
    {
      "epoch": 0.3852716164896252,
      "grad_norm": 0.6965934038162231,
      "learning_rate": 9.131798413666411e-06,
      "loss": 0.6827,
      "step": 7000
    },
    {
      "epoch": 0.38532665529198085,
      "grad_norm": 0.7449018359184265,
      "learning_rate": 9.13155429458883e-06,
      "loss": 0.7562,
      "step": 7001
    },
    {
      "epoch": 0.3853816940943365,
      "grad_norm": 0.7764221429824829,
      "learning_rate": 9.131310144459593e-06,
      "loss": 0.7842,
      "step": 7002
    },
    {
      "epoch": 0.38543673289669217,
      "grad_norm": 0.9788658618927002,
      "learning_rate": 9.131065963280536e-06,
      "loss": 0.7857,
      "step": 7003
    },
    {
      "epoch": 0.3854917716990478,
      "grad_norm": 0.7900908589363098,
      "learning_rate": 9.13082175105349e-06,
      "loss": 0.8733,
      "step": 7004
    },
    {
      "epoch": 0.3855468105014035,
      "grad_norm": 0.814822793006897,
      "learning_rate": 9.130577507780298e-06,
      "loss": 0.8032,
      "step": 7005
    },
    {
      "epoch": 0.3856018493037591,
      "grad_norm": 1.0648475885391235,
      "learning_rate": 9.130333233462789e-06,
      "loss": 0.8078,
      "step": 7006
    },
    {
      "epoch": 0.3856568881061148,
      "grad_norm": 0.7359917163848877,
      "learning_rate": 9.130088928102799e-06,
      "loss": 0.6491,
      "step": 7007
    },
    {
      "epoch": 0.38571192690847045,
      "grad_norm": 0.7321771383285522,
      "learning_rate": 9.129844591702169e-06,
      "loss": 0.7663,
      "step": 7008
    },
    {
      "epoch": 0.38576696571082614,
      "grad_norm": 0.6937146186828613,
      "learning_rate": 9.129600224262732e-06,
      "loss": 0.7835,
      "step": 7009
    },
    {
      "epoch": 0.38582200451318177,
      "grad_norm": 0.7330107688903809,
      "learning_rate": 9.129355825786323e-06,
      "loss": 0.7626,
      "step": 7010
    },
    {
      "epoch": 0.38587704331553746,
      "grad_norm": 0.7021715044975281,
      "learning_rate": 9.129111396274783e-06,
      "loss": 0.7115,
      "step": 7011
    },
    {
      "epoch": 0.3859320821178931,
      "grad_norm": 0.6599563360214233,
      "learning_rate": 9.128866935729947e-06,
      "loss": 0.6554,
      "step": 7012
    },
    {
      "epoch": 0.3859871209202488,
      "grad_norm": 0.7323513031005859,
      "learning_rate": 9.128622444153652e-06,
      "loss": 0.7392,
      "step": 7013
    },
    {
      "epoch": 0.3860421597226044,
      "grad_norm": 0.681888222694397,
      "learning_rate": 9.128377921547736e-06,
      "loss": 0.7474,
      "step": 7014
    },
    {
      "epoch": 0.3860971985249601,
      "grad_norm": 0.8454889059066772,
      "learning_rate": 9.128133367914036e-06,
      "loss": 0.8355,
      "step": 7015
    },
    {
      "epoch": 0.38615223732731574,
      "grad_norm": 0.7514123916625977,
      "learning_rate": 9.12788878325439e-06,
      "loss": 0.7683,
      "step": 7016
    },
    {
      "epoch": 0.38620727612967143,
      "grad_norm": 0.7317092418670654,
      "learning_rate": 9.12764416757064e-06,
      "loss": 0.7201,
      "step": 7017
    },
    {
      "epoch": 0.38626231493202706,
      "grad_norm": 0.7626729011535645,
      "learning_rate": 9.127399520864619e-06,
      "loss": 0.7701,
      "step": 7018
    },
    {
      "epoch": 0.38631735373438275,
      "grad_norm": 0.9790363311767578,
      "learning_rate": 9.127154843138168e-06,
      "loss": 0.8034,
      "step": 7019
    },
    {
      "epoch": 0.3863723925367384,
      "grad_norm": 0.663593590259552,
      "learning_rate": 9.126910134393125e-06,
      "loss": 0.661,
      "step": 7020
    },
    {
      "epoch": 0.3864274313390941,
      "grad_norm": 0.6599924564361572,
      "learning_rate": 9.126665394631332e-06,
      "loss": 0.7395,
      "step": 7021
    },
    {
      "epoch": 0.3864824701414497,
      "grad_norm": 0.8493411540985107,
      "learning_rate": 9.126420623854625e-06,
      "loss": 0.8008,
      "step": 7022
    },
    {
      "epoch": 0.3865375089438054,
      "grad_norm": 0.7587194442749023,
      "learning_rate": 9.126175822064846e-06,
      "loss": 0.7533,
      "step": 7023
    },
    {
      "epoch": 0.38659254774616103,
      "grad_norm": 0.773764431476593,
      "learning_rate": 9.125930989263835e-06,
      "loss": 0.75,
      "step": 7024
    },
    {
      "epoch": 0.3866475865485167,
      "grad_norm": 0.7126749753952026,
      "learning_rate": 9.12568612545343e-06,
      "loss": 0.7794,
      "step": 7025
    },
    {
      "epoch": 0.38670262535087235,
      "grad_norm": 0.7404584884643555,
      "learning_rate": 9.125441230635472e-06,
      "loss": 0.7264,
      "step": 7026
    },
    {
      "epoch": 0.38675766415322804,
      "grad_norm": 0.8057644367218018,
      "learning_rate": 9.125196304811804e-06,
      "loss": 0.8058,
      "step": 7027
    },
    {
      "epoch": 0.3868127029555837,
      "grad_norm": 0.9586995840072632,
      "learning_rate": 9.124951347984263e-06,
      "loss": 0.7659,
      "step": 7028
    },
    {
      "epoch": 0.38686774175793937,
      "grad_norm": 0.7567793726921082,
      "learning_rate": 9.124706360154693e-06,
      "loss": 0.8961,
      "step": 7029
    },
    {
      "epoch": 0.386922780560295,
      "grad_norm": 0.8523182272911072,
      "learning_rate": 9.124461341324934e-06,
      "loss": 0.8815,
      "step": 7030
    },
    {
      "epoch": 0.3869778193626507,
      "grad_norm": 0.7466379404067993,
      "learning_rate": 9.124216291496826e-06,
      "loss": 0.7817,
      "step": 7031
    },
    {
      "epoch": 0.3870328581650063,
      "grad_norm": 0.6721325516700745,
      "learning_rate": 9.123971210672214e-06,
      "loss": 0.7637,
      "step": 7032
    },
    {
      "epoch": 0.387087896967362,
      "grad_norm": 0.6620928049087524,
      "learning_rate": 9.123726098852936e-06,
      "loss": 0.6956,
      "step": 7033
    },
    {
      "epoch": 0.38714293576971764,
      "grad_norm": 0.6784290671348572,
      "learning_rate": 9.12348095604084e-06,
      "loss": 0.7034,
      "step": 7034
    },
    {
      "epoch": 0.38719797457207333,
      "grad_norm": 0.7138848304748535,
      "learning_rate": 9.123235782237763e-06,
      "loss": 0.6037,
      "step": 7035
    },
    {
      "epoch": 0.38725301337442897,
      "grad_norm": 0.8473613858222961,
      "learning_rate": 9.122990577445548e-06,
      "loss": 0.8157,
      "step": 7036
    },
    {
      "epoch": 0.38730805217678466,
      "grad_norm": 0.835381031036377,
      "learning_rate": 9.122745341666041e-06,
      "loss": 0.8736,
      "step": 7037
    },
    {
      "epoch": 0.3873630909791403,
      "grad_norm": 0.8823271989822388,
      "learning_rate": 9.122500074901083e-06,
      "loss": 0.7448,
      "step": 7038
    },
    {
      "epoch": 0.387418129781496,
      "grad_norm": 0.6494244933128357,
      "learning_rate": 9.122254777152519e-06,
      "loss": 0.7423,
      "step": 7039
    },
    {
      "epoch": 0.3874731685838516,
      "grad_norm": 0.7232181429862976,
      "learning_rate": 9.122009448422191e-06,
      "loss": 0.8489,
      "step": 7040
    },
    {
      "epoch": 0.3875282073862073,
      "grad_norm": 0.7357699275016785,
      "learning_rate": 9.121764088711945e-06,
      "loss": 0.8799,
      "step": 7041
    },
    {
      "epoch": 0.38758324618856294,
      "grad_norm": 0.7638574838638306,
      "learning_rate": 9.121518698023621e-06,
      "loss": 0.8539,
      "step": 7042
    },
    {
      "epoch": 0.3876382849909186,
      "grad_norm": 0.7407062649726868,
      "learning_rate": 9.121273276359068e-06,
      "loss": 0.7152,
      "step": 7043
    },
    {
      "epoch": 0.38769332379327426,
      "grad_norm": 0.6945983171463013,
      "learning_rate": 9.121027823720126e-06,
      "loss": 0.8224,
      "step": 7044
    },
    {
      "epoch": 0.3877483625956299,
      "grad_norm": 0.7163639068603516,
      "learning_rate": 9.120782340108643e-06,
      "loss": 0.808,
      "step": 7045
    },
    {
      "epoch": 0.3878034013979856,
      "grad_norm": 0.7062035799026489,
      "learning_rate": 9.120536825526463e-06,
      "loss": 0.783,
      "step": 7046
    },
    {
      "epoch": 0.3878584402003412,
      "grad_norm": 0.7459971308708191,
      "learning_rate": 9.120291279975431e-06,
      "loss": 0.8219,
      "step": 7047
    },
    {
      "epoch": 0.3879134790026969,
      "grad_norm": 0.9016150236129761,
      "learning_rate": 9.120045703457394e-06,
      "loss": 0.8605,
      "step": 7048
    },
    {
      "epoch": 0.38796851780505254,
      "grad_norm": 0.78440922498703,
      "learning_rate": 9.119800095974193e-06,
      "loss": 0.8424,
      "step": 7049
    },
    {
      "epoch": 0.3880235566074082,
      "grad_norm": 0.751504123210907,
      "learning_rate": 9.119554457527681e-06,
      "loss": 0.701,
      "step": 7050
    },
    {
      "epoch": 0.38807859540976386,
      "grad_norm": 0.7540284991264343,
      "learning_rate": 9.119308788119698e-06,
      "loss": 0.7912,
      "step": 7051
    },
    {
      "epoch": 0.38813363421211955,
      "grad_norm": 0.7977007627487183,
      "learning_rate": 9.119063087752094e-06,
      "loss": 0.9297,
      "step": 7052
    },
    {
      "epoch": 0.3881886730144752,
      "grad_norm": 0.6923508644104004,
      "learning_rate": 9.118817356426715e-06,
      "loss": 0.7458,
      "step": 7053
    },
    {
      "epoch": 0.38824371181683087,
      "grad_norm": 0.7170272469520569,
      "learning_rate": 9.118571594145406e-06,
      "loss": 0.733,
      "step": 7054
    },
    {
      "epoch": 0.3882987506191865,
      "grad_norm": 0.7547701001167297,
      "learning_rate": 9.118325800910015e-06,
      "loss": 0.7758,
      "step": 7055
    },
    {
      "epoch": 0.3883537894215422,
      "grad_norm": 0.7921421527862549,
      "learning_rate": 9.118079976722391e-06,
      "loss": 0.8262,
      "step": 7056
    },
    {
      "epoch": 0.38840882822389783,
      "grad_norm": 0.734470784664154,
      "learning_rate": 9.117834121584379e-06,
      "loss": 0.817,
      "step": 7057
    },
    {
      "epoch": 0.3884638670262535,
      "grad_norm": 0.8106420040130615,
      "learning_rate": 9.117588235497829e-06,
      "loss": 0.8203,
      "step": 7058
    },
    {
      "epoch": 0.38851890582860915,
      "grad_norm": 0.7355543375015259,
      "learning_rate": 9.117342318464588e-06,
      "loss": 0.8076,
      "step": 7059
    },
    {
      "epoch": 0.38857394463096484,
      "grad_norm": 0.7665252685546875,
      "learning_rate": 9.117096370486504e-06,
      "loss": 0.7611,
      "step": 7060
    },
    {
      "epoch": 0.3886289834333205,
      "grad_norm": 0.7968598008155823,
      "learning_rate": 9.116850391565426e-06,
      "loss": 0.6461,
      "step": 7061
    },
    {
      "epoch": 0.38868402223567616,
      "grad_norm": 0.7187741994857788,
      "learning_rate": 9.116604381703203e-06,
      "loss": 0.7982,
      "step": 7062
    },
    {
      "epoch": 0.3887390610380318,
      "grad_norm": 0.8566913604736328,
      "learning_rate": 9.11635834090168e-06,
      "loss": 0.9072,
      "step": 7063
    },
    {
      "epoch": 0.3887940998403875,
      "grad_norm": 0.7120797038078308,
      "learning_rate": 9.116112269162714e-06,
      "loss": 0.7353,
      "step": 7064
    },
    {
      "epoch": 0.3888491386427431,
      "grad_norm": 0.7230019569396973,
      "learning_rate": 9.115866166488148e-06,
      "loss": 0.7717,
      "step": 7065
    },
    {
      "epoch": 0.3889041774450988,
      "grad_norm": 0.6650584936141968,
      "learning_rate": 9.115620032879833e-06,
      "loss": 0.7384,
      "step": 7066
    },
    {
      "epoch": 0.38895921624745444,
      "grad_norm": 0.970750629901886,
      "learning_rate": 9.115373868339621e-06,
      "loss": 0.8478,
      "step": 7067
    },
    {
      "epoch": 0.38901425504981013,
      "grad_norm": 0.7066280245780945,
      "learning_rate": 9.115127672869359e-06,
      "loss": 0.7638,
      "step": 7068
    },
    {
      "epoch": 0.38906929385216577,
      "grad_norm": 0.6952232718467712,
      "learning_rate": 9.1148814464709e-06,
      "loss": 0.7869,
      "step": 7069
    },
    {
      "epoch": 0.38912433265452145,
      "grad_norm": 0.804489254951477,
      "learning_rate": 9.114635189146094e-06,
      "loss": 0.7905,
      "step": 7070
    },
    {
      "epoch": 0.3891793714568771,
      "grad_norm": 0.6988457441329956,
      "learning_rate": 9.114388900896791e-06,
      "loss": 0.7107,
      "step": 7071
    },
    {
      "epoch": 0.3892344102592328,
      "grad_norm": 0.6379980444908142,
      "learning_rate": 9.114142581724842e-06,
      "loss": 0.733,
      "step": 7072
    },
    {
      "epoch": 0.3892894490615884,
      "grad_norm": 0.7238649129867554,
      "learning_rate": 9.113896231632098e-06,
      "loss": 0.8252,
      "step": 7073
    },
    {
      "epoch": 0.3893444878639441,
      "grad_norm": 0.7168585062026978,
      "learning_rate": 9.113649850620412e-06,
      "loss": 0.6459,
      "step": 7074
    },
    {
      "epoch": 0.38939952666629973,
      "grad_norm": 0.7315915822982788,
      "learning_rate": 9.113403438691634e-06,
      "loss": 0.7557,
      "step": 7075
    },
    {
      "epoch": 0.3894545654686554,
      "grad_norm": 0.7438754439353943,
      "learning_rate": 9.11315699584762e-06,
      "loss": 0.7938,
      "step": 7076
    },
    {
      "epoch": 0.38950960427101106,
      "grad_norm": 0.7497848272323608,
      "learning_rate": 9.112910522090215e-06,
      "loss": 0.8232,
      "step": 7077
    },
    {
      "epoch": 0.38956464307336675,
      "grad_norm": 0.8072896003723145,
      "learning_rate": 9.112664017421277e-06,
      "loss": 0.7974,
      "step": 7078
    },
    {
      "epoch": 0.3896196818757224,
      "grad_norm": 0.7255920767784119,
      "learning_rate": 9.112417481842657e-06,
      "loss": 0.7658,
      "step": 7079
    },
    {
      "epoch": 0.38967472067807807,
      "grad_norm": 0.6263132095336914,
      "learning_rate": 9.112170915356209e-06,
      "loss": 0.7188,
      "step": 7080
    },
    {
      "epoch": 0.3897297594804337,
      "grad_norm": 0.6817660927772522,
      "learning_rate": 9.111924317963785e-06,
      "loss": 0.7406,
      "step": 7081
    },
    {
      "epoch": 0.3897847982827894,
      "grad_norm": 0.7829134464263916,
      "learning_rate": 9.111677689667238e-06,
      "loss": 0.8406,
      "step": 7082
    },
    {
      "epoch": 0.389839837085145,
      "grad_norm": 0.7122843861579895,
      "learning_rate": 9.111431030468421e-06,
      "loss": 0.7722,
      "step": 7083
    },
    {
      "epoch": 0.3898948758875007,
      "grad_norm": 0.7041764259338379,
      "learning_rate": 9.11118434036919e-06,
      "loss": 0.8307,
      "step": 7084
    },
    {
      "epoch": 0.38994991468985635,
      "grad_norm": 0.7582009434700012,
      "learning_rate": 9.110937619371398e-06,
      "loss": 0.7461,
      "step": 7085
    },
    {
      "epoch": 0.39000495349221204,
      "grad_norm": 0.7156100273132324,
      "learning_rate": 9.110690867476899e-06,
      "loss": 0.7294,
      "step": 7086
    },
    {
      "epoch": 0.39005999229456767,
      "grad_norm": 0.79449063539505,
      "learning_rate": 9.110444084687549e-06,
      "loss": 0.8652,
      "step": 7087
    },
    {
      "epoch": 0.3901150310969233,
      "grad_norm": 0.7692831754684448,
      "learning_rate": 9.1101972710052e-06,
      "loss": 0.7899,
      "step": 7088
    },
    {
      "epoch": 0.390170069899279,
      "grad_norm": 0.7189639806747437,
      "learning_rate": 9.109950426431708e-06,
      "loss": 0.726,
      "step": 7089
    },
    {
      "epoch": 0.3902251087016346,
      "grad_norm": 0.7491177916526794,
      "learning_rate": 9.10970355096893e-06,
      "loss": 0.8881,
      "step": 7090
    },
    {
      "epoch": 0.3902801475039903,
      "grad_norm": 0.783027172088623,
      "learning_rate": 9.10945664461872e-06,
      "loss": 0.7728,
      "step": 7091
    },
    {
      "epoch": 0.39033518630634595,
      "grad_norm": 1.0871556997299194,
      "learning_rate": 9.109209707382934e-06,
      "loss": 0.8059,
      "step": 7092
    },
    {
      "epoch": 0.39039022510870164,
      "grad_norm": 0.7287113666534424,
      "learning_rate": 9.108962739263429e-06,
      "loss": 0.7896,
      "step": 7093
    },
    {
      "epoch": 0.39044526391105727,
      "grad_norm": 0.7801700234413147,
      "learning_rate": 9.108715740262058e-06,
      "loss": 0.8012,
      "step": 7094
    },
    {
      "epoch": 0.39050030271341296,
      "grad_norm": 0.846709132194519,
      "learning_rate": 9.10846871038068e-06,
      "loss": 0.8392,
      "step": 7095
    },
    {
      "epoch": 0.3905553415157686,
      "grad_norm": 0.7408092617988586,
      "learning_rate": 9.10822164962115e-06,
      "loss": 0.8657,
      "step": 7096
    },
    {
      "epoch": 0.3906103803181243,
      "grad_norm": 0.6748743057250977,
      "learning_rate": 9.107974557985328e-06,
      "loss": 0.7659,
      "step": 7097
    },
    {
      "epoch": 0.3906654191204799,
      "grad_norm": 0.7512170672416687,
      "learning_rate": 9.107727435475067e-06,
      "loss": 0.7704,
      "step": 7098
    },
    {
      "epoch": 0.3907204579228356,
      "grad_norm": 0.9039596319198608,
      "learning_rate": 9.107480282092227e-06,
      "loss": 0.8412,
      "step": 7099
    },
    {
      "epoch": 0.39077549672519124,
      "grad_norm": 0.829785943031311,
      "learning_rate": 9.107233097838663e-06,
      "loss": 0.8229,
      "step": 7100
    },
    {
      "epoch": 0.39083053552754693,
      "grad_norm": 0.7597842812538147,
      "learning_rate": 9.106985882716238e-06,
      "loss": 0.7798,
      "step": 7101
    },
    {
      "epoch": 0.39088557432990256,
      "grad_norm": 0.7619945406913757,
      "learning_rate": 9.106738636726802e-06,
      "loss": 0.7504,
      "step": 7102
    },
    {
      "epoch": 0.39094061313225825,
      "grad_norm": 0.6791092157363892,
      "learning_rate": 9.10649135987222e-06,
      "loss": 0.8167,
      "step": 7103
    },
    {
      "epoch": 0.3909956519346139,
      "grad_norm": 0.7977412343025208,
      "learning_rate": 9.10624405215435e-06,
      "loss": 0.8252,
      "step": 7104
    },
    {
      "epoch": 0.3910506907369696,
      "grad_norm": 0.7329283356666565,
      "learning_rate": 9.105996713575047e-06,
      "loss": 0.7084,
      "step": 7105
    },
    {
      "epoch": 0.3911057295393252,
      "grad_norm": 0.7125133872032166,
      "learning_rate": 9.105749344136172e-06,
      "loss": 0.6672,
      "step": 7106
    },
    {
      "epoch": 0.3911607683416809,
      "grad_norm": 0.6974679827690125,
      "learning_rate": 9.105501943839583e-06,
      "loss": 0.7354,
      "step": 7107
    },
    {
      "epoch": 0.39121580714403653,
      "grad_norm": 0.7191265225410461,
      "learning_rate": 9.10525451268714e-06,
      "loss": 0.8133,
      "step": 7108
    },
    {
      "epoch": 0.3912708459463922,
      "grad_norm": 0.7188206911087036,
      "learning_rate": 9.105007050680704e-06,
      "loss": 0.7947,
      "step": 7109
    },
    {
      "epoch": 0.39132588474874785,
      "grad_norm": 0.9017364382743835,
      "learning_rate": 9.104759557822135e-06,
      "loss": 0.7848,
      "step": 7110
    },
    {
      "epoch": 0.39138092355110354,
      "grad_norm": 0.7551164031028748,
      "learning_rate": 9.104512034113292e-06,
      "loss": 0.8266,
      "step": 7111
    },
    {
      "epoch": 0.3914359623534592,
      "grad_norm": 0.7810001969337463,
      "learning_rate": 9.104264479556033e-06,
      "loss": 0.7731,
      "step": 7112
    },
    {
      "epoch": 0.39149100115581487,
      "grad_norm": 0.787723183631897,
      "learning_rate": 9.104016894152223e-06,
      "loss": 0.8008,
      "step": 7113
    },
    {
      "epoch": 0.3915460399581705,
      "grad_norm": 0.7303524017333984,
      "learning_rate": 9.103769277903718e-06,
      "loss": 0.826,
      "step": 7114
    },
    {
      "epoch": 0.3916010787605262,
      "grad_norm": 0.707759439945221,
      "learning_rate": 9.103521630812384e-06,
      "loss": 0.6303,
      "step": 7115
    },
    {
      "epoch": 0.3916561175628818,
      "grad_norm": 0.6929940581321716,
      "learning_rate": 9.10327395288008e-06,
      "loss": 0.733,
      "step": 7116
    },
    {
      "epoch": 0.3917111563652375,
      "grad_norm": 0.7133205533027649,
      "learning_rate": 9.103026244108667e-06,
      "loss": 0.8421,
      "step": 7117
    },
    {
      "epoch": 0.39176619516759315,
      "grad_norm": 1.2049434185028076,
      "learning_rate": 9.102778504500005e-06,
      "loss": 0.8618,
      "step": 7118
    },
    {
      "epoch": 0.39182123396994883,
      "grad_norm": 0.7792720198631287,
      "learning_rate": 9.10253073405596e-06,
      "loss": 0.717,
      "step": 7119
    },
    {
      "epoch": 0.39187627277230447,
      "grad_norm": 0.7234412431716919,
      "learning_rate": 9.10228293277839e-06,
      "loss": 0.7547,
      "step": 7120
    },
    {
      "epoch": 0.39193131157466016,
      "grad_norm": 0.6845420002937317,
      "learning_rate": 9.102035100669162e-06,
      "loss": 0.7255,
      "step": 7121
    },
    {
      "epoch": 0.3919863503770158,
      "grad_norm": 0.7446799874305725,
      "learning_rate": 9.101787237730135e-06,
      "loss": 0.7947,
      "step": 7122
    },
    {
      "epoch": 0.3920413891793715,
      "grad_norm": 0.812924325466156,
      "learning_rate": 9.101539343963176e-06,
      "loss": 0.843,
      "step": 7123
    },
    {
      "epoch": 0.3920964279817271,
      "grad_norm": 0.7373847365379333,
      "learning_rate": 9.101291419370141e-06,
      "loss": 0.7703,
      "step": 7124
    },
    {
      "epoch": 0.3921514667840828,
      "grad_norm": 0.8305120468139648,
      "learning_rate": 9.101043463952899e-06,
      "loss": 0.8904,
      "step": 7125
    },
    {
      "epoch": 0.39220650558643844,
      "grad_norm": 0.7263030409812927,
      "learning_rate": 9.100795477713313e-06,
      "loss": 0.8319,
      "step": 7126
    },
    {
      "epoch": 0.3922615443887941,
      "grad_norm": 0.8358581066131592,
      "learning_rate": 9.100547460653245e-06,
      "loss": 0.8305,
      "step": 7127
    },
    {
      "epoch": 0.39231658319114976,
      "grad_norm": 0.6608800292015076,
      "learning_rate": 9.10029941277456e-06,
      "loss": 0.7815,
      "step": 7128
    },
    {
      "epoch": 0.39237162199350545,
      "grad_norm": 0.8590257167816162,
      "learning_rate": 9.100051334079122e-06,
      "loss": 0.8292,
      "step": 7129
    },
    {
      "epoch": 0.3924266607958611,
      "grad_norm": 0.6241755485534668,
      "learning_rate": 9.099803224568797e-06,
      "loss": 0.6568,
      "step": 7130
    },
    {
      "epoch": 0.3924816995982167,
      "grad_norm": 0.7298059463500977,
      "learning_rate": 9.099555084245447e-06,
      "loss": 0.727,
      "step": 7131
    },
    {
      "epoch": 0.3925367384005724,
      "grad_norm": 0.7741055488586426,
      "learning_rate": 9.099306913110939e-06,
      "loss": 0.8481,
      "step": 7132
    },
    {
      "epoch": 0.39259177720292804,
      "grad_norm": 0.9674170613288879,
      "learning_rate": 9.099058711167137e-06,
      "loss": 0.8507,
      "step": 7133
    },
    {
      "epoch": 0.3926468160052837,
      "grad_norm": 0.7285159826278687,
      "learning_rate": 9.098810478415907e-06,
      "loss": 0.766,
      "step": 7134
    },
    {
      "epoch": 0.39270185480763936,
      "grad_norm": 0.7215660810470581,
      "learning_rate": 9.098562214859115e-06,
      "loss": 0.794,
      "step": 7135
    },
    {
      "epoch": 0.39275689360999505,
      "grad_norm": 0.764437735080719,
      "learning_rate": 9.098313920498627e-06,
      "loss": 0.8228,
      "step": 7136
    },
    {
      "epoch": 0.3928119324123507,
      "grad_norm": 0.7222796082496643,
      "learning_rate": 9.098065595336309e-06,
      "loss": 0.8064,
      "step": 7137
    },
    {
      "epoch": 0.3928669712147064,
      "grad_norm": 0.7044625878334045,
      "learning_rate": 9.097817239374024e-06,
      "loss": 0.8017,
      "step": 7138
    },
    {
      "epoch": 0.392922010017062,
      "grad_norm": 0.7929979562759399,
      "learning_rate": 9.097568852613646e-06,
      "loss": 0.7527,
      "step": 7139
    },
    {
      "epoch": 0.3929770488194177,
      "grad_norm": 0.7833721041679382,
      "learning_rate": 9.097320435057033e-06,
      "loss": 0.8335,
      "step": 7140
    },
    {
      "epoch": 0.39303208762177333,
      "grad_norm": 0.8365728259086609,
      "learning_rate": 9.097071986706058e-06,
      "loss": 0.6439,
      "step": 7141
    },
    {
      "epoch": 0.393087126424129,
      "grad_norm": 0.7547842264175415,
      "learning_rate": 9.096823507562588e-06,
      "loss": 0.8316,
      "step": 7142
    },
    {
      "epoch": 0.39314216522648465,
      "grad_norm": 0.6598891019821167,
      "learning_rate": 9.09657499762849e-06,
      "loss": 0.6547,
      "step": 7143
    },
    {
      "epoch": 0.39319720402884034,
      "grad_norm": 0.7913638949394226,
      "learning_rate": 9.096326456905627e-06,
      "loss": 0.7964,
      "step": 7144
    },
    {
      "epoch": 0.393252242831196,
      "grad_norm": 0.6927905082702637,
      "learning_rate": 9.096077885395874e-06,
      "loss": 0.7836,
      "step": 7145
    },
    {
      "epoch": 0.39330728163355166,
      "grad_norm": 0.7505417466163635,
      "learning_rate": 9.095829283101094e-06,
      "loss": 0.7707,
      "step": 7146
    },
    {
      "epoch": 0.3933623204359073,
      "grad_norm": 0.8797083497047424,
      "learning_rate": 9.095580650023158e-06,
      "loss": 0.866,
      "step": 7147
    },
    {
      "epoch": 0.393417359238263,
      "grad_norm": 0.7023645639419556,
      "learning_rate": 9.095331986163935e-06,
      "loss": 0.7013,
      "step": 7148
    },
    {
      "epoch": 0.3934723980406186,
      "grad_norm": 0.697354793548584,
      "learning_rate": 9.095083291525293e-06,
      "loss": 0.7691,
      "step": 7149
    },
    {
      "epoch": 0.3935274368429743,
      "grad_norm": 0.7211105227470398,
      "learning_rate": 9.094834566109101e-06,
      "loss": 0.6816,
      "step": 7150
    },
    {
      "epoch": 0.39358247564532994,
      "grad_norm": 0.8593278527259827,
      "learning_rate": 9.094585809917227e-06,
      "loss": 0.915,
      "step": 7151
    },
    {
      "epoch": 0.39363751444768563,
      "grad_norm": 0.7406070828437805,
      "learning_rate": 9.094337022951545e-06,
      "loss": 0.7825,
      "step": 7152
    },
    {
      "epoch": 0.39369255325004127,
      "grad_norm": 0.7644504308700562,
      "learning_rate": 9.09408820521392e-06,
      "loss": 0.6796,
      "step": 7153
    },
    {
      "epoch": 0.39374759205239696,
      "grad_norm": 0.8239033222198486,
      "learning_rate": 9.093839356706224e-06,
      "loss": 0.8396,
      "step": 7154
    },
    {
      "epoch": 0.3938026308547526,
      "grad_norm": 0.6433991193771362,
      "learning_rate": 9.093590477430327e-06,
      "loss": 0.6941,
      "step": 7155
    },
    {
      "epoch": 0.3938576696571083,
      "grad_norm": 0.6979972124099731,
      "learning_rate": 9.093341567388102e-06,
      "loss": 0.8142,
      "step": 7156
    },
    {
      "epoch": 0.3939127084594639,
      "grad_norm": 0.7062026262283325,
      "learning_rate": 9.093092626581414e-06,
      "loss": 0.804,
      "step": 7157
    },
    {
      "epoch": 0.3939677472618196,
      "grad_norm": 0.7070814967155457,
      "learning_rate": 9.09284365501214e-06,
      "loss": 0.765,
      "step": 7158
    },
    {
      "epoch": 0.39402278606417523,
      "grad_norm": 0.8577908873558044,
      "learning_rate": 9.092594652682147e-06,
      "loss": 0.7074,
      "step": 7159
    },
    {
      "epoch": 0.3940778248665309,
      "grad_norm": 0.7386197447776794,
      "learning_rate": 9.092345619593309e-06,
      "loss": 0.7629,
      "step": 7160
    },
    {
      "epoch": 0.39413286366888656,
      "grad_norm": 0.8048123121261597,
      "learning_rate": 9.092096555747496e-06,
      "loss": 0.9225,
      "step": 7161
    },
    {
      "epoch": 0.39418790247124225,
      "grad_norm": 0.7479888200759888,
      "learning_rate": 9.091847461146582e-06,
      "loss": 0.7284,
      "step": 7162
    },
    {
      "epoch": 0.3942429412735979,
      "grad_norm": 0.7448734045028687,
      "learning_rate": 9.091598335792438e-06,
      "loss": 0.8694,
      "step": 7163
    },
    {
      "epoch": 0.39429798007595357,
      "grad_norm": 0.7511261701583862,
      "learning_rate": 9.091349179686935e-06,
      "loss": 0.7822,
      "step": 7164
    },
    {
      "epoch": 0.3943530188783092,
      "grad_norm": 0.7079344391822815,
      "learning_rate": 9.091099992831946e-06,
      "loss": 0.7238,
      "step": 7165
    },
    {
      "epoch": 0.3944080576806649,
      "grad_norm": 0.7007229328155518,
      "learning_rate": 9.090850775229347e-06,
      "loss": 0.7269,
      "step": 7166
    },
    {
      "epoch": 0.3944630964830205,
      "grad_norm": 0.769800066947937,
      "learning_rate": 9.090601526881007e-06,
      "loss": 0.7894,
      "step": 7167
    },
    {
      "epoch": 0.3945181352853762,
      "grad_norm": 0.7211676836013794,
      "learning_rate": 9.090352247788801e-06,
      "loss": 0.7998,
      "step": 7168
    },
    {
      "epoch": 0.39457317408773185,
      "grad_norm": 0.6784254312515259,
      "learning_rate": 9.090102937954602e-06,
      "loss": 0.7576,
      "step": 7169
    },
    {
      "epoch": 0.39462821289008754,
      "grad_norm": 0.7696946859359741,
      "learning_rate": 9.089853597380285e-06,
      "loss": 0.8395,
      "step": 7170
    },
    {
      "epoch": 0.39468325169244317,
      "grad_norm": 0.8720405697822571,
      "learning_rate": 9.089604226067723e-06,
      "loss": 0.8971,
      "step": 7171
    },
    {
      "epoch": 0.39473829049479886,
      "grad_norm": 0.8457947373390198,
      "learning_rate": 9.08935482401879e-06,
      "loss": 0.7002,
      "step": 7172
    },
    {
      "epoch": 0.3947933292971545,
      "grad_norm": 0.8181997537612915,
      "learning_rate": 9.089105391235361e-06,
      "loss": 0.8949,
      "step": 7173
    },
    {
      "epoch": 0.3948483680995101,
      "grad_norm": 0.7717136144638062,
      "learning_rate": 9.08885592771931e-06,
      "loss": 0.829,
      "step": 7174
    },
    {
      "epoch": 0.3949034069018658,
      "grad_norm": 0.6941567063331604,
      "learning_rate": 9.088606433472514e-06,
      "loss": 0.7592,
      "step": 7175
    },
    {
      "epoch": 0.39495844570422145,
      "grad_norm": 0.7358599901199341,
      "learning_rate": 9.088356908496845e-06,
      "loss": 0.8657,
      "step": 7176
    },
    {
      "epoch": 0.39501348450657714,
      "grad_norm": 1.1329307556152344,
      "learning_rate": 9.08810735279418e-06,
      "loss": 0.8307,
      "step": 7177
    },
    {
      "epoch": 0.3950685233089328,
      "grad_norm": 0.7011532187461853,
      "learning_rate": 9.087857766366395e-06,
      "loss": 0.7487,
      "step": 7178
    },
    {
      "epoch": 0.39512356211128846,
      "grad_norm": 0.7390572428703308,
      "learning_rate": 9.087608149215366e-06,
      "loss": 0.8244,
      "step": 7179
    },
    {
      "epoch": 0.3951786009136441,
      "grad_norm": 0.6907634735107422,
      "learning_rate": 9.087358501342966e-06,
      "loss": 0.751,
      "step": 7180
    },
    {
      "epoch": 0.3952336397159998,
      "grad_norm": 0.7467379570007324,
      "learning_rate": 9.087108822751076e-06,
      "loss": 0.8549,
      "step": 7181
    },
    {
      "epoch": 0.3952886785183554,
      "grad_norm": 0.7493302226066589,
      "learning_rate": 9.086859113441568e-06,
      "loss": 0.8332,
      "step": 7182
    },
    {
      "epoch": 0.3953437173207111,
      "grad_norm": 0.8364959955215454,
      "learning_rate": 9.086609373416321e-06,
      "loss": 0.7873,
      "step": 7183
    },
    {
      "epoch": 0.39539875612306674,
      "grad_norm": 0.7330418825149536,
      "learning_rate": 9.086359602677214e-06,
      "loss": 0.7861,
      "step": 7184
    },
    {
      "epoch": 0.39545379492542243,
      "grad_norm": 0.7296311855316162,
      "learning_rate": 9.086109801226121e-06,
      "loss": 0.7946,
      "step": 7185
    },
    {
      "epoch": 0.39550883372777806,
      "grad_norm": 0.7884660363197327,
      "learning_rate": 9.085859969064921e-06,
      "loss": 0.7851,
      "step": 7186
    },
    {
      "epoch": 0.39556387253013375,
      "grad_norm": 0.7311955690383911,
      "learning_rate": 9.08561010619549e-06,
      "loss": 0.7645,
      "step": 7187
    },
    {
      "epoch": 0.3956189113324894,
      "grad_norm": 0.7447296977043152,
      "learning_rate": 9.085360212619707e-06,
      "loss": 0.7446,
      "step": 7188
    },
    {
      "epoch": 0.3956739501348451,
      "grad_norm": 0.755628228187561,
      "learning_rate": 9.08511028833945e-06,
      "loss": 0.8107,
      "step": 7189
    },
    {
      "epoch": 0.3957289889372007,
      "grad_norm": 0.6800833940505981,
      "learning_rate": 9.0848603333566e-06,
      "loss": 0.7471,
      "step": 7190
    },
    {
      "epoch": 0.3957840277395564,
      "grad_norm": 0.6396341919898987,
      "learning_rate": 9.08461034767303e-06,
      "loss": 0.6797,
      "step": 7191
    },
    {
      "epoch": 0.39583906654191203,
      "grad_norm": 0.729680597782135,
      "learning_rate": 9.084360331290625e-06,
      "loss": 0.7224,
      "step": 7192
    },
    {
      "epoch": 0.3958941053442677,
      "grad_norm": 0.7630584239959717,
      "learning_rate": 9.084110284211259e-06,
      "loss": 0.8203,
      "step": 7193
    },
    {
      "epoch": 0.39594914414662336,
      "grad_norm": 0.8799235820770264,
      "learning_rate": 9.083860206436813e-06,
      "loss": 0.8312,
      "step": 7194
    },
    {
      "epoch": 0.39600418294897904,
      "grad_norm": 0.797081708908081,
      "learning_rate": 9.083610097969169e-06,
      "loss": 0.7561,
      "step": 7195
    },
    {
      "epoch": 0.3960592217513347,
      "grad_norm": 0.7408759593963623,
      "learning_rate": 9.083359958810203e-06,
      "loss": 0.7854,
      "step": 7196
    },
    {
      "epoch": 0.39611426055369037,
      "grad_norm": 0.7552130222320557,
      "learning_rate": 9.083109788961797e-06,
      "loss": 0.8145,
      "step": 7197
    },
    {
      "epoch": 0.396169299356046,
      "grad_norm": 0.7147447466850281,
      "learning_rate": 9.08285958842583e-06,
      "loss": 0.792,
      "step": 7198
    },
    {
      "epoch": 0.3962243381584017,
      "grad_norm": 0.7416259050369263,
      "learning_rate": 9.082609357204183e-06,
      "loss": 0.7801,
      "step": 7199
    },
    {
      "epoch": 0.3962793769607573,
      "grad_norm": 0.7551109194755554,
      "learning_rate": 9.082359095298741e-06,
      "loss": 0.8841,
      "step": 7200
    },
    {
      "epoch": 0.396334415763113,
      "grad_norm": 0.761472225189209,
      "learning_rate": 9.082108802711377e-06,
      "loss": 0.9061,
      "step": 7201
    },
    {
      "epoch": 0.39638945456546865,
      "grad_norm": 0.7234126329421997,
      "learning_rate": 9.081858479443977e-06,
      "loss": 0.8308,
      "step": 7202
    },
    {
      "epoch": 0.39644449336782434,
      "grad_norm": 0.7204816341400146,
      "learning_rate": 9.08160812549842e-06,
      "loss": 0.7481,
      "step": 7203
    },
    {
      "epoch": 0.39649953217017997,
      "grad_norm": 0.7207956910133362,
      "learning_rate": 9.081357740876591e-06,
      "loss": 0.762,
      "step": 7204
    },
    {
      "epoch": 0.39655457097253566,
      "grad_norm": 0.7967123985290527,
      "learning_rate": 9.081107325580367e-06,
      "loss": 0.7931,
      "step": 7205
    },
    {
      "epoch": 0.3966096097748913,
      "grad_norm": 0.9839354753494263,
      "learning_rate": 9.080856879611635e-06,
      "loss": 0.8182,
      "step": 7206
    },
    {
      "epoch": 0.396664648577247,
      "grad_norm": 0.8468357920646667,
      "learning_rate": 9.080606402972274e-06,
      "loss": 0.7056,
      "step": 7207
    },
    {
      "epoch": 0.3967196873796026,
      "grad_norm": 0.6549574136734009,
      "learning_rate": 9.080355895664169e-06,
      "loss": 0.7604,
      "step": 7208
    },
    {
      "epoch": 0.3967747261819583,
      "grad_norm": 0.7475417256355286,
      "learning_rate": 9.080105357689201e-06,
      "loss": 0.7107,
      "step": 7209
    },
    {
      "epoch": 0.39682976498431394,
      "grad_norm": 0.7464179992675781,
      "learning_rate": 9.079854789049251e-06,
      "loss": 0.793,
      "step": 7210
    },
    {
      "epoch": 0.3968848037866696,
      "grad_norm": 0.8332071900367737,
      "learning_rate": 9.079604189746207e-06,
      "loss": 0.8383,
      "step": 7211
    },
    {
      "epoch": 0.39693984258902526,
      "grad_norm": 0.722055196762085,
      "learning_rate": 9.07935355978195e-06,
      "loss": 0.8569,
      "step": 7212
    },
    {
      "epoch": 0.39699488139138095,
      "grad_norm": 0.7442018389701843,
      "learning_rate": 9.079102899158363e-06,
      "loss": 0.8165,
      "step": 7213
    },
    {
      "epoch": 0.3970499201937366,
      "grad_norm": 0.6986141204833984,
      "learning_rate": 9.07885220787733e-06,
      "loss": 0.7562,
      "step": 7214
    },
    {
      "epoch": 0.39710495899609227,
      "grad_norm": 0.7718464732170105,
      "learning_rate": 9.078601485940736e-06,
      "loss": 0.8529,
      "step": 7215
    },
    {
      "epoch": 0.3971599977984479,
      "grad_norm": 0.7583653330802917,
      "learning_rate": 9.078350733350464e-06,
      "loss": 0.7855,
      "step": 7216
    },
    {
      "epoch": 0.39721503660080354,
      "grad_norm": 0.7699223160743713,
      "learning_rate": 9.078099950108401e-06,
      "loss": 0.8061,
      "step": 7217
    },
    {
      "epoch": 0.39727007540315923,
      "grad_norm": 0.7374141812324524,
      "learning_rate": 9.07784913621643e-06,
      "loss": 0.789,
      "step": 7218
    },
    {
      "epoch": 0.39732511420551486,
      "grad_norm": 0.7446104884147644,
      "learning_rate": 9.077598291676436e-06,
      "loss": 0.8381,
      "step": 7219
    },
    {
      "epoch": 0.39738015300787055,
      "grad_norm": 0.7017301917076111,
      "learning_rate": 9.077347416490305e-06,
      "loss": 0.7153,
      "step": 7220
    },
    {
      "epoch": 0.3974351918102262,
      "grad_norm": 0.7676172852516174,
      "learning_rate": 9.077096510659922e-06,
      "loss": 0.8029,
      "step": 7221
    },
    {
      "epoch": 0.3974902306125819,
      "grad_norm": 0.9340602159500122,
      "learning_rate": 9.076845574187174e-06,
      "loss": 0.7865,
      "step": 7222
    },
    {
      "epoch": 0.3975452694149375,
      "grad_norm": 0.8634235262870789,
      "learning_rate": 9.076594607073945e-06,
      "loss": 0.7606,
      "step": 7223
    },
    {
      "epoch": 0.3976003082172932,
      "grad_norm": 0.8967369198799133,
      "learning_rate": 9.076343609322123e-06,
      "loss": 0.7011,
      "step": 7224
    },
    {
      "epoch": 0.39765534701964883,
      "grad_norm": 0.7269352078437805,
      "learning_rate": 9.076092580933594e-06,
      "loss": 0.8043,
      "step": 7225
    },
    {
      "epoch": 0.3977103858220045,
      "grad_norm": 0.7550628781318665,
      "learning_rate": 9.075841521910243e-06,
      "loss": 0.7344,
      "step": 7226
    },
    {
      "epoch": 0.39776542462436015,
      "grad_norm": 0.6973844766616821,
      "learning_rate": 9.075590432253958e-06,
      "loss": 0.6995,
      "step": 7227
    },
    {
      "epoch": 0.39782046342671584,
      "grad_norm": 0.648560643196106,
      "learning_rate": 9.075339311966627e-06,
      "loss": 0.6997,
      "step": 7228
    },
    {
      "epoch": 0.3978755022290715,
      "grad_norm": 0.8457548022270203,
      "learning_rate": 9.075088161050134e-06,
      "loss": 0.8548,
      "step": 7229
    },
    {
      "epoch": 0.39793054103142717,
      "grad_norm": 0.7644637823104858,
      "learning_rate": 9.074836979506373e-06,
      "loss": 0.6966,
      "step": 7230
    },
    {
      "epoch": 0.3979855798337828,
      "grad_norm": 0.7146210670471191,
      "learning_rate": 9.074585767337227e-06,
      "loss": 0.7673,
      "step": 7231
    },
    {
      "epoch": 0.3980406186361385,
      "grad_norm": 0.8570694327354431,
      "learning_rate": 9.074334524544585e-06,
      "loss": 0.8233,
      "step": 7232
    },
    {
      "epoch": 0.3980956574384941,
      "grad_norm": 0.7257633805274963,
      "learning_rate": 9.074083251130334e-06,
      "loss": 0.7464,
      "step": 7233
    },
    {
      "epoch": 0.3981506962408498,
      "grad_norm": 0.9377032518386841,
      "learning_rate": 9.073831947096365e-06,
      "loss": 0.7814,
      "step": 7234
    },
    {
      "epoch": 0.39820573504320544,
      "grad_norm": 0.8105629086494446,
      "learning_rate": 9.073580612444566e-06,
      "loss": 0.8069,
      "step": 7235
    },
    {
      "epoch": 0.39826077384556113,
      "grad_norm": 0.7874456644058228,
      "learning_rate": 9.073329247176824e-06,
      "loss": 0.8414,
      "step": 7236
    },
    {
      "epoch": 0.39831581264791677,
      "grad_norm": 0.6829617023468018,
      "learning_rate": 9.07307785129503e-06,
      "loss": 0.7633,
      "step": 7237
    },
    {
      "epoch": 0.39837085145027246,
      "grad_norm": 0.6838501691818237,
      "learning_rate": 9.072826424801075e-06,
      "loss": 0.6972,
      "step": 7238
    },
    {
      "epoch": 0.3984258902526281,
      "grad_norm": 0.7054216861724854,
      "learning_rate": 9.072574967696845e-06,
      "loss": 0.8049,
      "step": 7239
    },
    {
      "epoch": 0.3984809290549838,
      "grad_norm": 0.9462615847587585,
      "learning_rate": 9.072323479984232e-06,
      "loss": 0.7988,
      "step": 7240
    },
    {
      "epoch": 0.3985359678573394,
      "grad_norm": 0.7334465980529785,
      "learning_rate": 9.072071961665128e-06,
      "loss": 0.7538,
      "step": 7241
    },
    {
      "epoch": 0.3985910066596951,
      "grad_norm": 0.7506609559059143,
      "learning_rate": 9.071820412741418e-06,
      "loss": 0.7991,
      "step": 7242
    },
    {
      "epoch": 0.39864604546205074,
      "grad_norm": 0.6858688592910767,
      "learning_rate": 9.071568833214998e-06,
      "loss": 0.7258,
      "step": 7243
    },
    {
      "epoch": 0.3987010842644064,
      "grad_norm": 0.8117396235466003,
      "learning_rate": 9.071317223087754e-06,
      "loss": 0.752,
      "step": 7244
    },
    {
      "epoch": 0.39875612306676206,
      "grad_norm": 0.7772389054298401,
      "learning_rate": 9.071065582361582e-06,
      "loss": 0.7444,
      "step": 7245
    },
    {
      "epoch": 0.39881116186911775,
      "grad_norm": 0.7221882939338684,
      "learning_rate": 9.07081391103837e-06,
      "loss": 0.8035,
      "step": 7246
    },
    {
      "epoch": 0.3988662006714734,
      "grad_norm": 0.8113289475440979,
      "learning_rate": 9.07056220912001e-06,
      "loss": 0.7623,
      "step": 7247
    },
    {
      "epoch": 0.39892123947382907,
      "grad_norm": 0.730823278427124,
      "learning_rate": 9.070310476608395e-06,
      "loss": 0.7872,
      "step": 7248
    },
    {
      "epoch": 0.3989762782761847,
      "grad_norm": 0.7690893411636353,
      "learning_rate": 9.070058713505415e-06,
      "loss": 0.7402,
      "step": 7249
    },
    {
      "epoch": 0.3990313170785404,
      "grad_norm": 0.6768597364425659,
      "learning_rate": 9.069806919812963e-06,
      "loss": 0.7283,
      "step": 7250
    },
    {
      "epoch": 0.399086355880896,
      "grad_norm": 0.6938686370849609,
      "learning_rate": 9.069555095532932e-06,
      "loss": 0.7209,
      "step": 7251
    },
    {
      "epoch": 0.3991413946832517,
      "grad_norm": 0.7162025570869446,
      "learning_rate": 9.069303240667215e-06,
      "loss": 0.7915,
      "step": 7252
    },
    {
      "epoch": 0.39919643348560735,
      "grad_norm": 0.9170399308204651,
      "learning_rate": 9.069051355217704e-06,
      "loss": 0.8399,
      "step": 7253
    },
    {
      "epoch": 0.39925147228796304,
      "grad_norm": 0.7080186009407043,
      "learning_rate": 9.068799439186291e-06,
      "loss": 0.8678,
      "step": 7254
    },
    {
      "epoch": 0.39930651109031867,
      "grad_norm": 1.013613224029541,
      "learning_rate": 9.068547492574872e-06,
      "loss": 0.817,
      "step": 7255
    },
    {
      "epoch": 0.39936154989267436,
      "grad_norm": 0.6911013722419739,
      "learning_rate": 9.068295515385337e-06,
      "loss": 0.7048,
      "step": 7256
    },
    {
      "epoch": 0.39941658869503,
      "grad_norm": 0.748219907283783,
      "learning_rate": 9.068043507619584e-06,
      "loss": 0.8115,
      "step": 7257
    },
    {
      "epoch": 0.3994716274973857,
      "grad_norm": 0.6763347387313843,
      "learning_rate": 9.067791469279504e-06,
      "loss": 0.763,
      "step": 7258
    },
    {
      "epoch": 0.3995266662997413,
      "grad_norm": 0.7291030287742615,
      "learning_rate": 9.067539400366993e-06,
      "loss": 0.7319,
      "step": 7259
    },
    {
      "epoch": 0.39958170510209695,
      "grad_norm": 0.6515628695487976,
      "learning_rate": 9.067287300883945e-06,
      "loss": 0.7903,
      "step": 7260
    },
    {
      "epoch": 0.39963674390445264,
      "grad_norm": 0.7815985679626465,
      "learning_rate": 9.067035170832253e-06,
      "loss": 0.8241,
      "step": 7261
    },
    {
      "epoch": 0.3996917827068083,
      "grad_norm": 0.6747417449951172,
      "learning_rate": 9.066783010213812e-06,
      "loss": 0.7544,
      "step": 7262
    },
    {
      "epoch": 0.39974682150916396,
      "grad_norm": 0.6568340063095093,
      "learning_rate": 9.066530819030522e-06,
      "loss": 0.7754,
      "step": 7263
    },
    {
      "epoch": 0.3998018603115196,
      "grad_norm": 0.6703339219093323,
      "learning_rate": 9.066278597284273e-06,
      "loss": 0.7581,
      "step": 7264
    },
    {
      "epoch": 0.3998568991138753,
      "grad_norm": 0.7421279549598694,
      "learning_rate": 9.066026344976962e-06,
      "loss": 0.7974,
      "step": 7265
    },
    {
      "epoch": 0.3999119379162309,
      "grad_norm": 0.7226015329360962,
      "learning_rate": 9.065774062110486e-06,
      "loss": 0.7777,
      "step": 7266
    },
    {
      "epoch": 0.3999669767185866,
      "grad_norm": 0.7092894911766052,
      "learning_rate": 9.06552174868674e-06,
      "loss": 0.7885,
      "step": 7267
    },
    {
      "epoch": 0.40002201552094224,
      "grad_norm": 0.837902307510376,
      "learning_rate": 9.065269404707622e-06,
      "loss": 0.7425,
      "step": 7268
    },
    {
      "epoch": 0.40007705432329793,
      "grad_norm": 0.803811252117157,
      "learning_rate": 9.065017030175027e-06,
      "loss": 0.8418,
      "step": 7269
    },
    {
      "epoch": 0.40013209312565357,
      "grad_norm": 0.8110278248786926,
      "learning_rate": 9.064764625090854e-06,
      "loss": 0.7724,
      "step": 7270
    },
    {
      "epoch": 0.40018713192800925,
      "grad_norm": 0.7305173277854919,
      "learning_rate": 9.064512189456995e-06,
      "loss": 0.7465,
      "step": 7271
    },
    {
      "epoch": 0.4002421707303649,
      "grad_norm": 0.7312467694282532,
      "learning_rate": 9.06425972327535e-06,
      "loss": 0.8406,
      "step": 7272
    },
    {
      "epoch": 0.4002972095327206,
      "grad_norm": 0.7348741292953491,
      "learning_rate": 9.064007226547819e-06,
      "loss": 0.8103,
      "step": 7273
    },
    {
      "epoch": 0.4003522483350762,
      "grad_norm": 0.6561787724494934,
      "learning_rate": 9.063754699276297e-06,
      "loss": 0.6634,
      "step": 7274
    },
    {
      "epoch": 0.4004072871374319,
      "grad_norm": 0.7924866080284119,
      "learning_rate": 9.063502141462682e-06,
      "loss": 0.6592,
      "step": 7275
    },
    {
      "epoch": 0.40046232593978753,
      "grad_norm": 0.6873973608016968,
      "learning_rate": 9.063249553108873e-06,
      "loss": 0.7912,
      "step": 7276
    },
    {
      "epoch": 0.4005173647421432,
      "grad_norm": 0.6872708797454834,
      "learning_rate": 9.062996934216768e-06,
      "loss": 0.732,
      "step": 7277
    },
    {
      "epoch": 0.40057240354449886,
      "grad_norm": 0.7381585836410522,
      "learning_rate": 9.062744284788265e-06,
      "loss": 0.84,
      "step": 7278
    },
    {
      "epoch": 0.40062744234685455,
      "grad_norm": 0.7885964512825012,
      "learning_rate": 9.062491604825266e-06,
      "loss": 0.8229,
      "step": 7279
    },
    {
      "epoch": 0.4006824811492102,
      "grad_norm": 0.9066407680511475,
      "learning_rate": 9.062238894329664e-06,
      "loss": 0.7299,
      "step": 7280
    },
    {
      "epoch": 0.40073751995156587,
      "grad_norm": 0.7694007754325867,
      "learning_rate": 9.061986153303364e-06,
      "loss": 0.8033,
      "step": 7281
    },
    {
      "epoch": 0.4007925587539215,
      "grad_norm": 1.021766185760498,
      "learning_rate": 9.061733381748263e-06,
      "loss": 0.79,
      "step": 7282
    },
    {
      "epoch": 0.4008475975562772,
      "grad_norm": 0.7776662111282349,
      "learning_rate": 9.06148057966626e-06,
      "loss": 0.8484,
      "step": 7283
    },
    {
      "epoch": 0.4009026363586328,
      "grad_norm": 0.8646043539047241,
      "learning_rate": 9.061227747059257e-06,
      "loss": 0.8223,
      "step": 7284
    },
    {
      "epoch": 0.4009576751609885,
      "grad_norm": 0.7347257733345032,
      "learning_rate": 9.060974883929154e-06,
      "loss": 0.8062,
      "step": 7285
    },
    {
      "epoch": 0.40101271396334415,
      "grad_norm": 0.8233902454376221,
      "learning_rate": 9.06072199027785e-06,
      "loss": 0.8922,
      "step": 7286
    },
    {
      "epoch": 0.40106775276569984,
      "grad_norm": 0.7099601030349731,
      "learning_rate": 9.060469066107246e-06,
      "loss": 0.7125,
      "step": 7287
    },
    {
      "epoch": 0.40112279156805547,
      "grad_norm": 0.7549998164176941,
      "learning_rate": 9.060216111419246e-06,
      "loss": 0.7851,
      "step": 7288
    },
    {
      "epoch": 0.40117783037041116,
      "grad_norm": 0.753516435623169,
      "learning_rate": 9.059963126215748e-06,
      "loss": 0.7831,
      "step": 7289
    },
    {
      "epoch": 0.4012328691727668,
      "grad_norm": 0.6718429327011108,
      "learning_rate": 9.059710110498651e-06,
      "loss": 0.7305,
      "step": 7290
    },
    {
      "epoch": 0.4012879079751225,
      "grad_norm": 0.6796036958694458,
      "learning_rate": 9.05945706426986e-06,
      "loss": 0.802,
      "step": 7291
    },
    {
      "epoch": 0.4013429467774781,
      "grad_norm": 0.8046827912330627,
      "learning_rate": 9.05920398753128e-06,
      "loss": 0.7286,
      "step": 7292
    },
    {
      "epoch": 0.4013979855798338,
      "grad_norm": 0.7518643140792847,
      "learning_rate": 9.058950880284807e-06,
      "loss": 0.7287,
      "step": 7293
    },
    {
      "epoch": 0.40145302438218944,
      "grad_norm": 0.8386855125427246,
      "learning_rate": 9.058697742532345e-06,
      "loss": 0.8201,
      "step": 7294
    },
    {
      "epoch": 0.4015080631845451,
      "grad_norm": 0.7780192494392395,
      "learning_rate": 9.058444574275797e-06,
      "loss": 0.7999,
      "step": 7295
    },
    {
      "epoch": 0.40156310198690076,
      "grad_norm": 0.7715566754341125,
      "learning_rate": 9.058191375517068e-06,
      "loss": 0.732,
      "step": 7296
    },
    {
      "epoch": 0.40161814078925645,
      "grad_norm": 0.9940280914306641,
      "learning_rate": 9.057938146258057e-06,
      "loss": 0.8247,
      "step": 7297
    },
    {
      "epoch": 0.4016731795916121,
      "grad_norm": 0.7567923069000244,
      "learning_rate": 9.05768488650067e-06,
      "loss": 0.8254,
      "step": 7298
    },
    {
      "epoch": 0.4017282183939678,
      "grad_norm": 0.7544496655464172,
      "learning_rate": 9.05743159624681e-06,
      "loss": 0.811,
      "step": 7299
    },
    {
      "epoch": 0.4017832571963234,
      "grad_norm": 0.63368821144104,
      "learning_rate": 9.05717827549838e-06,
      "loss": 0.6498,
      "step": 7300
    },
    {
      "epoch": 0.4018382959986791,
      "grad_norm": 0.7077621221542358,
      "learning_rate": 9.056924924257284e-06,
      "loss": 0.7401,
      "step": 7301
    },
    {
      "epoch": 0.40189333480103473,
      "grad_norm": 0.6782366037368774,
      "learning_rate": 9.056671542525426e-06,
      "loss": 0.8013,
      "step": 7302
    },
    {
      "epoch": 0.40194837360339036,
      "grad_norm": 0.6605678200721741,
      "learning_rate": 9.056418130304709e-06,
      "loss": 0.8038,
      "step": 7303
    },
    {
      "epoch": 0.40200341240574605,
      "grad_norm": 0.8716840147972107,
      "learning_rate": 9.056164687597041e-06,
      "loss": 0.7652,
      "step": 7304
    },
    {
      "epoch": 0.4020584512081017,
      "grad_norm": 0.8464542031288147,
      "learning_rate": 9.055911214404325e-06,
      "loss": 0.8663,
      "step": 7305
    },
    {
      "epoch": 0.4021134900104574,
      "grad_norm": 0.7165409326553345,
      "learning_rate": 9.055657710728466e-06,
      "loss": 0.8028,
      "step": 7306
    },
    {
      "epoch": 0.402168528812813,
      "grad_norm": 0.7313430309295654,
      "learning_rate": 9.055404176571369e-06,
      "loss": 0.7538,
      "step": 7307
    },
    {
      "epoch": 0.4022235676151687,
      "grad_norm": 0.7757230401039124,
      "learning_rate": 9.05515061193494e-06,
      "loss": 0.9096,
      "step": 7308
    },
    {
      "epoch": 0.40227860641752433,
      "grad_norm": 0.7178354859352112,
      "learning_rate": 9.054897016821085e-06,
      "loss": 0.7186,
      "step": 7309
    },
    {
      "epoch": 0.40233364521988,
      "grad_norm": 0.8331356048583984,
      "learning_rate": 9.054643391231708e-06,
      "loss": 0.8724,
      "step": 7310
    },
    {
      "epoch": 0.40238868402223565,
      "grad_norm": 0.7709757685661316,
      "learning_rate": 9.054389735168717e-06,
      "loss": 0.692,
      "step": 7311
    },
    {
      "epoch": 0.40244372282459134,
      "grad_norm": 0.7393380999565125,
      "learning_rate": 9.054136048634018e-06,
      "loss": 0.7863,
      "step": 7312
    },
    {
      "epoch": 0.402498761626947,
      "grad_norm": 0.7372385859489441,
      "learning_rate": 9.053882331629518e-06,
      "loss": 0.781,
      "step": 7313
    },
    {
      "epoch": 0.40255380042930267,
      "grad_norm": 0.7076019048690796,
      "learning_rate": 9.053628584157123e-06,
      "loss": 0.7598,
      "step": 7314
    },
    {
      "epoch": 0.4026088392316583,
      "grad_norm": 0.7465673685073853,
      "learning_rate": 9.053374806218742e-06,
      "loss": 0.7454,
      "step": 7315
    },
    {
      "epoch": 0.402663878034014,
      "grad_norm": 0.7414120435714722,
      "learning_rate": 9.05312099781628e-06,
      "loss": 0.7135,
      "step": 7316
    },
    {
      "epoch": 0.4027189168363696,
      "grad_norm": 0.7490748167037964,
      "learning_rate": 9.052867158951646e-06,
      "loss": 0.6833,
      "step": 7317
    },
    {
      "epoch": 0.4027739556387253,
      "grad_norm": 0.8027878999710083,
      "learning_rate": 9.052613289626747e-06,
      "loss": 0.7466,
      "step": 7318
    },
    {
      "epoch": 0.40282899444108095,
      "grad_norm": 0.6777862310409546,
      "learning_rate": 9.052359389843493e-06,
      "loss": 0.7446,
      "step": 7319
    },
    {
      "epoch": 0.40288403324343663,
      "grad_norm": 0.9240381717681885,
      "learning_rate": 9.052105459603787e-06,
      "loss": 0.7801,
      "step": 7320
    },
    {
      "epoch": 0.40293907204579227,
      "grad_norm": 0.9592602252960205,
      "learning_rate": 9.051851498909543e-06,
      "loss": 0.9648,
      "step": 7321
    },
    {
      "epoch": 0.40299411084814796,
      "grad_norm": 0.8469638228416443,
      "learning_rate": 9.051597507762669e-06,
      "loss": 0.8303,
      "step": 7322
    },
    {
      "epoch": 0.4030491496505036,
      "grad_norm": 0.6981443166732788,
      "learning_rate": 9.05134348616507e-06,
      "loss": 0.7245,
      "step": 7323
    },
    {
      "epoch": 0.4031041884528593,
      "grad_norm": 0.7133469581604004,
      "learning_rate": 9.05108943411866e-06,
      "loss": 0.7763,
      "step": 7324
    },
    {
      "epoch": 0.4031592272552149,
      "grad_norm": 0.7043703198432922,
      "learning_rate": 9.050835351625344e-06,
      "loss": 0.8247,
      "step": 7325
    },
    {
      "epoch": 0.4032142660575706,
      "grad_norm": 0.6662501692771912,
      "learning_rate": 9.050581238687036e-06,
      "loss": 0.7669,
      "step": 7326
    },
    {
      "epoch": 0.40326930485992624,
      "grad_norm": 0.6482356786727905,
      "learning_rate": 9.050327095305643e-06,
      "loss": 0.6477,
      "step": 7327
    },
    {
      "epoch": 0.4033243436622819,
      "grad_norm": 0.7465450167655945,
      "learning_rate": 9.050072921483076e-06,
      "loss": 0.8053,
      "step": 7328
    },
    {
      "epoch": 0.40337938246463756,
      "grad_norm": 0.6765472292900085,
      "learning_rate": 9.049818717221245e-06,
      "loss": 0.765,
      "step": 7329
    },
    {
      "epoch": 0.40343442126699325,
      "grad_norm": 0.7098689675331116,
      "learning_rate": 9.04956448252206e-06,
      "loss": 0.8059,
      "step": 7330
    },
    {
      "epoch": 0.4034894600693489,
      "grad_norm": 0.6773823499679565,
      "learning_rate": 9.049310217387432e-06,
      "loss": 0.6848,
      "step": 7331
    },
    {
      "epoch": 0.40354449887170457,
      "grad_norm": 0.6884829998016357,
      "learning_rate": 9.049055921819275e-06,
      "loss": 0.696,
      "step": 7332
    },
    {
      "epoch": 0.4035995376740602,
      "grad_norm": 0.662545919418335,
      "learning_rate": 9.048801595819494e-06,
      "loss": 0.8286,
      "step": 7333
    },
    {
      "epoch": 0.4036545764764159,
      "grad_norm": 0.6863077878952026,
      "learning_rate": 9.048547239390007e-06,
      "loss": 0.7215,
      "step": 7334
    },
    {
      "epoch": 0.4037096152787715,
      "grad_norm": 0.6982632875442505,
      "learning_rate": 9.048292852532721e-06,
      "loss": 0.7635,
      "step": 7335
    },
    {
      "epoch": 0.4037646540811272,
      "grad_norm": 0.8512400984764099,
      "learning_rate": 9.048038435249548e-06,
      "loss": 0.6226,
      "step": 7336
    },
    {
      "epoch": 0.40381969288348285,
      "grad_norm": 0.6952843070030212,
      "learning_rate": 9.047783987542405e-06,
      "loss": 0.8317,
      "step": 7337
    },
    {
      "epoch": 0.40387473168583854,
      "grad_norm": 0.7802778482437134,
      "learning_rate": 9.0475295094132e-06,
      "loss": 0.8615,
      "step": 7338
    },
    {
      "epoch": 0.4039297704881942,
      "grad_norm": 0.8783930540084839,
      "learning_rate": 9.047275000863844e-06,
      "loss": 0.743,
      "step": 7339
    },
    {
      "epoch": 0.40398480929054986,
      "grad_norm": 0.7205806970596313,
      "learning_rate": 9.047020461896256e-06,
      "loss": 0.7953,
      "step": 7340
    },
    {
      "epoch": 0.4040398480929055,
      "grad_norm": 0.8438451290130615,
      "learning_rate": 9.046765892512344e-06,
      "loss": 0.7613,
      "step": 7341
    },
    {
      "epoch": 0.4040948868952612,
      "grad_norm": 0.7300973534584045,
      "learning_rate": 9.046511292714021e-06,
      "loss": 0.7856,
      "step": 7342
    },
    {
      "epoch": 0.4041499256976168,
      "grad_norm": 0.8472041487693787,
      "learning_rate": 9.046256662503206e-06,
      "loss": 0.8526,
      "step": 7343
    },
    {
      "epoch": 0.4042049644999725,
      "grad_norm": 0.789465606212616,
      "learning_rate": 9.046002001881807e-06,
      "loss": 0.7792,
      "step": 7344
    },
    {
      "epoch": 0.40426000330232814,
      "grad_norm": 0.7720938920974731,
      "learning_rate": 9.04574731085174e-06,
      "loss": 0.8065,
      "step": 7345
    },
    {
      "epoch": 0.4043150421046838,
      "grad_norm": 0.6968526840209961,
      "learning_rate": 9.04549258941492e-06,
      "loss": 0.8135,
      "step": 7346
    },
    {
      "epoch": 0.40437008090703946,
      "grad_norm": 0.746865451335907,
      "learning_rate": 9.04523783757326e-06,
      "loss": 0.8216,
      "step": 7347
    },
    {
      "epoch": 0.4044251197093951,
      "grad_norm": 0.6750560998916626,
      "learning_rate": 9.044983055328676e-06,
      "loss": 0.7883,
      "step": 7348
    },
    {
      "epoch": 0.4044801585117508,
      "grad_norm": 0.6791195273399353,
      "learning_rate": 9.044728242683081e-06,
      "loss": 0.7721,
      "step": 7349
    },
    {
      "epoch": 0.4045351973141064,
      "grad_norm": 0.7238358855247498,
      "learning_rate": 9.044473399638392e-06,
      "loss": 0.739,
      "step": 7350
    },
    {
      "epoch": 0.4045902361164621,
      "grad_norm": 0.6793557405471802,
      "learning_rate": 9.044218526196523e-06,
      "loss": 0.7853,
      "step": 7351
    },
    {
      "epoch": 0.40464527491881774,
      "grad_norm": 0.767564058303833,
      "learning_rate": 9.043963622359392e-06,
      "loss": 0.8158,
      "step": 7352
    },
    {
      "epoch": 0.40470031372117343,
      "grad_norm": 0.6800708770751953,
      "learning_rate": 9.043708688128909e-06,
      "loss": 0.7493,
      "step": 7353
    },
    {
      "epoch": 0.40475535252352907,
      "grad_norm": 0.75978022813797,
      "learning_rate": 9.043453723506996e-06,
      "loss": 0.7066,
      "step": 7354
    },
    {
      "epoch": 0.40481039132588476,
      "grad_norm": 1.0194984674453735,
      "learning_rate": 9.043198728495568e-06,
      "loss": 0.6238,
      "step": 7355
    },
    {
      "epoch": 0.4048654301282404,
      "grad_norm": 0.7102386355400085,
      "learning_rate": 9.04294370309654e-06,
      "loss": 0.75,
      "step": 7356
    },
    {
      "epoch": 0.4049204689305961,
      "grad_norm": 0.8468191623687744,
      "learning_rate": 9.04268864731183e-06,
      "loss": 0.8095,
      "step": 7357
    },
    {
      "epoch": 0.4049755077329517,
      "grad_norm": 0.7022871971130371,
      "learning_rate": 9.042433561143353e-06,
      "loss": 0.8394,
      "step": 7358
    },
    {
      "epoch": 0.4050305465353074,
      "grad_norm": 1.1873482465744019,
      "learning_rate": 9.042178444593028e-06,
      "loss": 0.7863,
      "step": 7359
    },
    {
      "epoch": 0.40508558533766303,
      "grad_norm": 0.7074940204620361,
      "learning_rate": 9.041923297662772e-06,
      "loss": 0.7067,
      "step": 7360
    },
    {
      "epoch": 0.4051406241400187,
      "grad_norm": 0.7602211833000183,
      "learning_rate": 9.041668120354503e-06,
      "loss": 0.6594,
      "step": 7361
    },
    {
      "epoch": 0.40519566294237436,
      "grad_norm": 0.7903324365615845,
      "learning_rate": 9.041412912670138e-06,
      "loss": 0.7978,
      "step": 7362
    },
    {
      "epoch": 0.40525070174473005,
      "grad_norm": 0.7422891855239868,
      "learning_rate": 9.041157674611595e-06,
      "loss": 0.8162,
      "step": 7363
    },
    {
      "epoch": 0.4053057405470857,
      "grad_norm": 0.7978767156600952,
      "learning_rate": 9.040902406180791e-06,
      "loss": 0.762,
      "step": 7364
    },
    {
      "epoch": 0.40536077934944137,
      "grad_norm": 0.7719776630401611,
      "learning_rate": 9.04064710737965e-06,
      "loss": 0.8098,
      "step": 7365
    },
    {
      "epoch": 0.405415818151797,
      "grad_norm": 0.8646591305732727,
      "learning_rate": 9.040391778210083e-06,
      "loss": 0.9372,
      "step": 7366
    },
    {
      "epoch": 0.4054708569541527,
      "grad_norm": 0.6616937518119812,
      "learning_rate": 9.040136418674015e-06,
      "loss": 0.7424,
      "step": 7367
    },
    {
      "epoch": 0.4055258957565083,
      "grad_norm": 0.7676553130149841,
      "learning_rate": 9.039881028773363e-06,
      "loss": 0.6327,
      "step": 7368
    },
    {
      "epoch": 0.405580934558864,
      "grad_norm": 0.6838239431381226,
      "learning_rate": 9.039625608510047e-06,
      "loss": 0.7548,
      "step": 7369
    },
    {
      "epoch": 0.40563597336121965,
      "grad_norm": 0.7476304769515991,
      "learning_rate": 9.039370157885986e-06,
      "loss": 0.7262,
      "step": 7370
    },
    {
      "epoch": 0.40569101216357534,
      "grad_norm": 0.8985139727592468,
      "learning_rate": 9.0391146769031e-06,
      "loss": 0.7729,
      "step": 7371
    },
    {
      "epoch": 0.40574605096593097,
      "grad_norm": 0.7840422987937927,
      "learning_rate": 9.038859165563308e-06,
      "loss": 0.7855,
      "step": 7372
    },
    {
      "epoch": 0.40580108976828666,
      "grad_norm": 0.6777672171592712,
      "learning_rate": 9.038603623868534e-06,
      "loss": 0.7379,
      "step": 7373
    },
    {
      "epoch": 0.4058561285706423,
      "grad_norm": 0.7226746678352356,
      "learning_rate": 9.038348051820694e-06,
      "loss": 0.7686,
      "step": 7374
    },
    {
      "epoch": 0.405911167372998,
      "grad_norm": 0.7647444605827332,
      "learning_rate": 9.038092449421713e-06,
      "loss": 0.8859,
      "step": 7375
    },
    {
      "epoch": 0.4059662061753536,
      "grad_norm": 0.6524979472160339,
      "learning_rate": 9.037836816673508e-06,
      "loss": 0.6982,
      "step": 7376
    },
    {
      "epoch": 0.4060212449777093,
      "grad_norm": 0.7842861413955688,
      "learning_rate": 9.037581153578004e-06,
      "loss": 0.8099,
      "step": 7377
    },
    {
      "epoch": 0.40607628378006494,
      "grad_norm": 0.6424387693405151,
      "learning_rate": 9.03732546013712e-06,
      "loss": 0.7387,
      "step": 7378
    },
    {
      "epoch": 0.40613132258242063,
      "grad_norm": 0.8444356918334961,
      "learning_rate": 9.037069736352779e-06,
      "loss": 0.8813,
      "step": 7379
    },
    {
      "epoch": 0.40618636138477626,
      "grad_norm": 0.6487529277801514,
      "learning_rate": 9.036813982226904e-06,
      "loss": 0.7609,
      "step": 7380
    },
    {
      "epoch": 0.40624140018713195,
      "grad_norm": 0.7891185879707336,
      "learning_rate": 9.036558197761413e-06,
      "loss": 0.8589,
      "step": 7381
    },
    {
      "epoch": 0.4062964389894876,
      "grad_norm": 0.7183120250701904,
      "learning_rate": 9.036302382958233e-06,
      "loss": 0.8429,
      "step": 7382
    },
    {
      "epoch": 0.4063514777918433,
      "grad_norm": 0.6386578679084778,
      "learning_rate": 9.036046537819283e-06,
      "loss": 0.6955,
      "step": 7383
    },
    {
      "epoch": 0.4064065165941989,
      "grad_norm": 0.7572369575500488,
      "learning_rate": 9.035790662346488e-06,
      "loss": 0.8018,
      "step": 7384
    },
    {
      "epoch": 0.4064615553965546,
      "grad_norm": 0.7105650305747986,
      "learning_rate": 9.035534756541771e-06,
      "loss": 0.8527,
      "step": 7385
    },
    {
      "epoch": 0.40651659419891023,
      "grad_norm": 0.7031856179237366,
      "learning_rate": 9.035278820407056e-06,
      "loss": 0.6991,
      "step": 7386
    },
    {
      "epoch": 0.4065716330012659,
      "grad_norm": 0.7407381534576416,
      "learning_rate": 9.035022853944266e-06,
      "loss": 0.708,
      "step": 7387
    },
    {
      "epoch": 0.40662667180362155,
      "grad_norm": 0.7078498601913452,
      "learning_rate": 9.034766857155322e-06,
      "loss": 0.7584,
      "step": 7388
    },
    {
      "epoch": 0.4066817106059772,
      "grad_norm": 0.7643301486968994,
      "learning_rate": 9.034510830042151e-06,
      "loss": 0.7836,
      "step": 7389
    },
    {
      "epoch": 0.4067367494083329,
      "grad_norm": 0.7165302038192749,
      "learning_rate": 9.034254772606676e-06,
      "loss": 0.7769,
      "step": 7390
    },
    {
      "epoch": 0.4067917882106885,
      "grad_norm": 0.7442395091056824,
      "learning_rate": 9.033998684850824e-06,
      "loss": 0.7231,
      "step": 7391
    },
    {
      "epoch": 0.4068468270130442,
      "grad_norm": 0.7425046563148499,
      "learning_rate": 9.033742566776517e-06,
      "loss": 0.7709,
      "step": 7392
    },
    {
      "epoch": 0.40690186581539983,
      "grad_norm": 0.768419086933136,
      "learning_rate": 9.03348641838568e-06,
      "loss": 0.7768,
      "step": 7393
    },
    {
      "epoch": 0.4069569046177555,
      "grad_norm": 0.6785634160041809,
      "learning_rate": 9.03323023968024e-06,
      "loss": 0.7468,
      "step": 7394
    },
    {
      "epoch": 0.40701194342011116,
      "grad_norm": 0.7075444459915161,
      "learning_rate": 9.03297403066212e-06,
      "loss": 0.7757,
      "step": 7395
    },
    {
      "epoch": 0.40706698222246684,
      "grad_norm": 0.7580223679542542,
      "learning_rate": 9.032717791333247e-06,
      "loss": 0.7311,
      "step": 7396
    },
    {
      "epoch": 0.4071220210248225,
      "grad_norm": 0.8110041618347168,
      "learning_rate": 9.032461521695546e-06,
      "loss": 0.7923,
      "step": 7397
    },
    {
      "epoch": 0.40717705982717817,
      "grad_norm": 0.7204881310462952,
      "learning_rate": 9.032205221750945e-06,
      "loss": 0.759,
      "step": 7398
    },
    {
      "epoch": 0.4072320986295338,
      "grad_norm": 0.8392491340637207,
      "learning_rate": 9.031948891501368e-06,
      "loss": 0.8292,
      "step": 7399
    },
    {
      "epoch": 0.4072871374318895,
      "grad_norm": 0.7134600281715393,
      "learning_rate": 9.031692530948742e-06,
      "loss": 0.7,
      "step": 7400
    },
    {
      "epoch": 0.4073421762342451,
      "grad_norm": 0.6324336528778076,
      "learning_rate": 9.031436140094995e-06,
      "loss": 0.6964,
      "step": 7401
    },
    {
      "epoch": 0.4073972150366008,
      "grad_norm": 0.7281947731971741,
      "learning_rate": 9.031179718942052e-06,
      "loss": 0.7567,
      "step": 7402
    },
    {
      "epoch": 0.40745225383895645,
      "grad_norm": 0.8828619718551636,
      "learning_rate": 9.030923267491842e-06,
      "loss": 0.8139,
      "step": 7403
    },
    {
      "epoch": 0.40750729264131214,
      "grad_norm": 0.7039986252784729,
      "learning_rate": 9.030666785746292e-06,
      "loss": 0.7339,
      "step": 7404
    },
    {
      "epoch": 0.40756233144366777,
      "grad_norm": 0.7049984931945801,
      "learning_rate": 9.030410273707331e-06,
      "loss": 0.6842,
      "step": 7405
    },
    {
      "epoch": 0.40761737024602346,
      "grad_norm": 0.7149737477302551,
      "learning_rate": 9.030153731376883e-06,
      "loss": 0.6837,
      "step": 7406
    },
    {
      "epoch": 0.4076724090483791,
      "grad_norm": 1.0804089307785034,
      "learning_rate": 9.029897158756878e-06,
      "loss": 0.7726,
      "step": 7407
    },
    {
      "epoch": 0.4077274478507348,
      "grad_norm": 0.8354909420013428,
      "learning_rate": 9.029640555849244e-06,
      "loss": 0.8058,
      "step": 7408
    },
    {
      "epoch": 0.4077824866530904,
      "grad_norm": 0.7091527581214905,
      "learning_rate": 9.029383922655914e-06,
      "loss": 0.7636,
      "step": 7409
    },
    {
      "epoch": 0.4078375254554461,
      "grad_norm": 0.6720988750457764,
      "learning_rate": 9.029127259178809e-06,
      "loss": 0.7179,
      "step": 7410
    },
    {
      "epoch": 0.40789256425780174,
      "grad_norm": 0.685858964920044,
      "learning_rate": 9.028870565419865e-06,
      "loss": 0.7637,
      "step": 7411
    },
    {
      "epoch": 0.4079476030601574,
      "grad_norm": 0.7505033016204834,
      "learning_rate": 9.028613841381007e-06,
      "loss": 0.7463,
      "step": 7412
    },
    {
      "epoch": 0.40800264186251306,
      "grad_norm": 0.8801671862602234,
      "learning_rate": 9.028357087064166e-06,
      "loss": 0.8399,
      "step": 7413
    },
    {
      "epoch": 0.40805768066486875,
      "grad_norm": 0.7441918849945068,
      "learning_rate": 9.02810030247127e-06,
      "loss": 0.7689,
      "step": 7414
    },
    {
      "epoch": 0.4081127194672244,
      "grad_norm": 0.7410128712654114,
      "learning_rate": 9.027843487604251e-06,
      "loss": 0.8013,
      "step": 7415
    },
    {
      "epoch": 0.40816775826958007,
      "grad_norm": 0.8075226545333862,
      "learning_rate": 9.02758664246504e-06,
      "loss": 0.7717,
      "step": 7416
    },
    {
      "epoch": 0.4082227970719357,
      "grad_norm": 0.7985545992851257,
      "learning_rate": 9.027329767055566e-06,
      "loss": 0.8459,
      "step": 7417
    },
    {
      "epoch": 0.4082778358742914,
      "grad_norm": 0.7887235283851624,
      "learning_rate": 9.027072861377757e-06,
      "loss": 0.8201,
      "step": 7418
    },
    {
      "epoch": 0.40833287467664703,
      "grad_norm": 0.7876266241073608,
      "learning_rate": 9.02681592543355e-06,
      "loss": 0.8205,
      "step": 7419
    },
    {
      "epoch": 0.4083879134790027,
      "grad_norm": 0.758168637752533,
      "learning_rate": 9.02655895922487e-06,
      "loss": 0.6619,
      "step": 7420
    },
    {
      "epoch": 0.40844295228135835,
      "grad_norm": 0.7279811501502991,
      "learning_rate": 9.02630196275365e-06,
      "loss": 0.7634,
      "step": 7421
    },
    {
      "epoch": 0.40849799108371404,
      "grad_norm": 0.7540523409843445,
      "learning_rate": 9.026044936021822e-06,
      "loss": 0.7819,
      "step": 7422
    },
    {
      "epoch": 0.4085530298860697,
      "grad_norm": 0.8091018795967102,
      "learning_rate": 9.02578787903132e-06,
      "loss": 0.7749,
      "step": 7423
    },
    {
      "epoch": 0.40860806868842536,
      "grad_norm": 0.7625396847724915,
      "learning_rate": 9.025530791784074e-06,
      "loss": 0.7635,
      "step": 7424
    },
    {
      "epoch": 0.408663107490781,
      "grad_norm": 0.7663947939872742,
      "learning_rate": 9.025273674282015e-06,
      "loss": 0.8281,
      "step": 7425
    },
    {
      "epoch": 0.4087181462931367,
      "grad_norm": 0.6672662496566772,
      "learning_rate": 9.025016526527077e-06,
      "loss": 0.641,
      "step": 7426
    },
    {
      "epoch": 0.4087731850954923,
      "grad_norm": 0.7649143934249878,
      "learning_rate": 9.024759348521193e-06,
      "loss": 0.7462,
      "step": 7427
    },
    {
      "epoch": 0.408828223897848,
      "grad_norm": 0.7540067434310913,
      "learning_rate": 9.024502140266293e-06,
      "loss": 0.8756,
      "step": 7428
    },
    {
      "epoch": 0.40888326270020364,
      "grad_norm": 0.721615731716156,
      "learning_rate": 9.024244901764314e-06,
      "loss": 0.8507,
      "step": 7429
    },
    {
      "epoch": 0.40893830150255933,
      "grad_norm": 0.6949496269226074,
      "learning_rate": 9.023987633017186e-06,
      "loss": 0.7021,
      "step": 7430
    },
    {
      "epoch": 0.40899334030491497,
      "grad_norm": 0.7108990550041199,
      "learning_rate": 9.023730334026845e-06,
      "loss": 0.807,
      "step": 7431
    },
    {
      "epoch": 0.4090483791072706,
      "grad_norm": 0.7606124877929688,
      "learning_rate": 9.023473004795225e-06,
      "loss": 0.7769,
      "step": 7432
    },
    {
      "epoch": 0.4091034179096263,
      "grad_norm": 0.7792031764984131,
      "learning_rate": 9.023215645324256e-06,
      "loss": 0.728,
      "step": 7433
    },
    {
      "epoch": 0.4091584567119819,
      "grad_norm": 0.728884756565094,
      "learning_rate": 9.022958255615877e-06,
      "loss": 0.7831,
      "step": 7434
    },
    {
      "epoch": 0.4092134955143376,
      "grad_norm": 0.8196625709533691,
      "learning_rate": 9.022700835672022e-06,
      "loss": 0.8265,
      "step": 7435
    },
    {
      "epoch": 0.40926853431669324,
      "grad_norm": 0.762734055519104,
      "learning_rate": 9.022443385494621e-06,
      "loss": 0.8028,
      "step": 7436
    },
    {
      "epoch": 0.40932357311904893,
      "grad_norm": 0.7259558439254761,
      "learning_rate": 9.022185905085614e-06,
      "loss": 0.789,
      "step": 7437
    },
    {
      "epoch": 0.40937861192140457,
      "grad_norm": 0.7402371764183044,
      "learning_rate": 9.021928394446936e-06,
      "loss": 0.7667,
      "step": 7438
    },
    {
      "epoch": 0.40943365072376026,
      "grad_norm": 0.8399797677993774,
      "learning_rate": 9.021670853580519e-06,
      "loss": 0.8451,
      "step": 7439
    },
    {
      "epoch": 0.4094886895261159,
      "grad_norm": 0.6439585089683533,
      "learning_rate": 9.0214132824883e-06,
      "loss": 0.776,
      "step": 7440
    },
    {
      "epoch": 0.4095437283284716,
      "grad_norm": 0.6956612467765808,
      "learning_rate": 9.021155681172215e-06,
      "loss": 0.6921,
      "step": 7441
    },
    {
      "epoch": 0.4095987671308272,
      "grad_norm": 0.855413556098938,
      "learning_rate": 9.020898049634203e-06,
      "loss": 0.8552,
      "step": 7442
    },
    {
      "epoch": 0.4096538059331829,
      "grad_norm": 0.6690535545349121,
      "learning_rate": 9.020640387876194e-06,
      "loss": 0.7552,
      "step": 7443
    },
    {
      "epoch": 0.40970884473553854,
      "grad_norm": 0.6615462899208069,
      "learning_rate": 9.020382695900131e-06,
      "loss": 0.8216,
      "step": 7444
    },
    {
      "epoch": 0.4097638835378942,
      "grad_norm": 0.6975858211517334,
      "learning_rate": 9.020124973707947e-06,
      "loss": 0.7453,
      "step": 7445
    },
    {
      "epoch": 0.40981892234024986,
      "grad_norm": 0.6461964249610901,
      "learning_rate": 9.019867221301579e-06,
      "loss": 0.656,
      "step": 7446
    },
    {
      "epoch": 0.40987396114260555,
      "grad_norm": 0.7221645712852478,
      "learning_rate": 9.019609438682967e-06,
      "loss": 0.661,
      "step": 7447
    },
    {
      "epoch": 0.4099289999449612,
      "grad_norm": 0.6785755753517151,
      "learning_rate": 9.019351625854044e-06,
      "loss": 0.7294,
      "step": 7448
    },
    {
      "epoch": 0.40998403874731687,
      "grad_norm": 0.7040538787841797,
      "learning_rate": 9.019093782816751e-06,
      "loss": 0.8546,
      "step": 7449
    },
    {
      "epoch": 0.4100390775496725,
      "grad_norm": 0.737922191619873,
      "learning_rate": 9.018835909573025e-06,
      "loss": 0.8144,
      "step": 7450
    },
    {
      "epoch": 0.4100941163520282,
      "grad_norm": 0.6705496311187744,
      "learning_rate": 9.018578006124802e-06,
      "loss": 0.6937,
      "step": 7451
    },
    {
      "epoch": 0.4101491551543838,
      "grad_norm": 0.7347431182861328,
      "learning_rate": 9.018320072474026e-06,
      "loss": 0.7716,
      "step": 7452
    },
    {
      "epoch": 0.4102041939567395,
      "grad_norm": 0.7023493647575378,
      "learning_rate": 9.018062108622631e-06,
      "loss": 0.7295,
      "step": 7453
    },
    {
      "epoch": 0.41025923275909515,
      "grad_norm": 0.8017870187759399,
      "learning_rate": 9.017804114572556e-06,
      "loss": 0.7471,
      "step": 7454
    },
    {
      "epoch": 0.41031427156145084,
      "grad_norm": 0.9171211123466492,
      "learning_rate": 9.01754609032574e-06,
      "loss": 0.8262,
      "step": 7455
    },
    {
      "epoch": 0.41036931036380647,
      "grad_norm": 0.6682952046394348,
      "learning_rate": 9.017288035884124e-06,
      "loss": 0.7165,
      "step": 7456
    },
    {
      "epoch": 0.41042434916616216,
      "grad_norm": 0.9339122772216797,
      "learning_rate": 9.017029951249648e-06,
      "loss": 0.8618,
      "step": 7457
    },
    {
      "epoch": 0.4104793879685178,
      "grad_norm": 0.7063136696815491,
      "learning_rate": 9.016771836424248e-06,
      "loss": 0.8068,
      "step": 7458
    },
    {
      "epoch": 0.4105344267708735,
      "grad_norm": 0.6717063784599304,
      "learning_rate": 9.016513691409867e-06,
      "loss": 0.738,
      "step": 7459
    },
    {
      "epoch": 0.4105894655732291,
      "grad_norm": 0.6807749271392822,
      "learning_rate": 9.016255516208443e-06,
      "loss": 0.7842,
      "step": 7460
    },
    {
      "epoch": 0.4106445043755848,
      "grad_norm": 0.6990453600883484,
      "learning_rate": 9.01599731082192e-06,
      "loss": 0.7726,
      "step": 7461
    },
    {
      "epoch": 0.41069954317794044,
      "grad_norm": 0.6704931259155273,
      "learning_rate": 9.015739075252234e-06,
      "loss": 0.7006,
      "step": 7462
    },
    {
      "epoch": 0.41075458198029613,
      "grad_norm": 0.7162300944328308,
      "learning_rate": 9.01548080950133e-06,
      "loss": 0.8462,
      "step": 7463
    },
    {
      "epoch": 0.41080962078265176,
      "grad_norm": 0.6845411658287048,
      "learning_rate": 9.015222513571144e-06,
      "loss": 0.7466,
      "step": 7464
    },
    {
      "epoch": 0.41086465958500745,
      "grad_norm": 0.7146134376525879,
      "learning_rate": 9.014964187463623e-06,
      "loss": 0.7594,
      "step": 7465
    },
    {
      "epoch": 0.4109196983873631,
      "grad_norm": 0.7664906978607178,
      "learning_rate": 9.014705831180706e-06,
      "loss": 0.8376,
      "step": 7466
    },
    {
      "epoch": 0.4109747371897188,
      "grad_norm": 0.7319341897964478,
      "learning_rate": 9.014447444724332e-06,
      "loss": 0.7748,
      "step": 7467
    },
    {
      "epoch": 0.4110297759920744,
      "grad_norm": 0.7269605398178101,
      "learning_rate": 9.014189028096448e-06,
      "loss": 0.6941,
      "step": 7468
    },
    {
      "epoch": 0.4110848147944301,
      "grad_norm": 0.72607421875,
      "learning_rate": 9.013930581298993e-06,
      "loss": 0.7174,
      "step": 7469
    },
    {
      "epoch": 0.41113985359678573,
      "grad_norm": 0.7385421991348267,
      "learning_rate": 9.01367210433391e-06,
      "loss": 0.7761,
      "step": 7470
    },
    {
      "epoch": 0.4111948923991414,
      "grad_norm": 0.8392042517662048,
      "learning_rate": 9.013413597203144e-06,
      "loss": 0.7417,
      "step": 7471
    },
    {
      "epoch": 0.41124993120149705,
      "grad_norm": 0.7454584836959839,
      "learning_rate": 9.013155059908634e-06,
      "loss": 0.8976,
      "step": 7472
    },
    {
      "epoch": 0.41130497000385274,
      "grad_norm": 0.7358037829399109,
      "learning_rate": 9.012896492452325e-06,
      "loss": 0.7706,
      "step": 7473
    },
    {
      "epoch": 0.4113600088062084,
      "grad_norm": 0.7454121708869934,
      "learning_rate": 9.01263789483616e-06,
      "loss": 0.7425,
      "step": 7474
    },
    {
      "epoch": 0.411415047608564,
      "grad_norm": 0.7842294573783875,
      "learning_rate": 9.012379267062081e-06,
      "loss": 0.7739,
      "step": 7475
    },
    {
      "epoch": 0.4114700864109197,
      "grad_norm": 0.7181714773178101,
      "learning_rate": 9.012120609132036e-06,
      "loss": 0.8466,
      "step": 7476
    },
    {
      "epoch": 0.41152512521327533,
      "grad_norm": 0.7239206433296204,
      "learning_rate": 9.011861921047966e-06,
      "loss": 0.7493,
      "step": 7477
    },
    {
      "epoch": 0.411580164015631,
      "grad_norm": 0.6773414611816406,
      "learning_rate": 9.011603202811816e-06,
      "loss": 0.7433,
      "step": 7478
    },
    {
      "epoch": 0.41163520281798666,
      "grad_norm": 0.7770900130271912,
      "learning_rate": 9.011344454425527e-06,
      "loss": 0.7488,
      "step": 7479
    },
    {
      "epoch": 0.41169024162034235,
      "grad_norm": 0.7305957674980164,
      "learning_rate": 9.011085675891051e-06,
      "loss": 0.7989,
      "step": 7480
    },
    {
      "epoch": 0.411745280422698,
      "grad_norm": 0.734603762626648,
      "learning_rate": 9.010826867210327e-06,
      "loss": 0.805,
      "step": 7481
    },
    {
      "epoch": 0.41180031922505367,
      "grad_norm": 0.7438979148864746,
      "learning_rate": 9.010568028385303e-06,
      "loss": 0.8407,
      "step": 7482
    },
    {
      "epoch": 0.4118553580274093,
      "grad_norm": 0.6718543767929077,
      "learning_rate": 9.01030915941792e-06,
      "loss": 0.7575,
      "step": 7483
    },
    {
      "epoch": 0.411910396829765,
      "grad_norm": 0.8157614469528198,
      "learning_rate": 9.01005026031013e-06,
      "loss": 0.8231,
      "step": 7484
    },
    {
      "epoch": 0.4119654356321206,
      "grad_norm": 0.8927714824676514,
      "learning_rate": 9.009791331063874e-06,
      "loss": 0.808,
      "step": 7485
    },
    {
      "epoch": 0.4120204744344763,
      "grad_norm": 0.7604075074195862,
      "learning_rate": 9.009532371681101e-06,
      "loss": 0.7505,
      "step": 7486
    },
    {
      "epoch": 0.41207551323683195,
      "grad_norm": 0.6861944794654846,
      "learning_rate": 9.009273382163754e-06,
      "loss": 0.719,
      "step": 7487
    },
    {
      "epoch": 0.41213055203918764,
      "grad_norm": 0.7043709754943848,
      "learning_rate": 9.009014362513784e-06,
      "loss": 0.8193,
      "step": 7488
    },
    {
      "epoch": 0.41218559084154327,
      "grad_norm": 0.7459648847579956,
      "learning_rate": 9.008755312733136e-06,
      "loss": 0.8617,
      "step": 7489
    },
    {
      "epoch": 0.41224062964389896,
      "grad_norm": 0.7272594571113586,
      "learning_rate": 9.008496232823754e-06,
      "loss": 0.7255,
      "step": 7490
    },
    {
      "epoch": 0.4122956684462546,
      "grad_norm": 0.7486668229103088,
      "learning_rate": 9.008237122787586e-06,
      "loss": 0.6479,
      "step": 7491
    },
    {
      "epoch": 0.4123507072486103,
      "grad_norm": 0.8149027228355408,
      "learning_rate": 9.007977982626582e-06,
      "loss": 0.8052,
      "step": 7492
    },
    {
      "epoch": 0.4124057460509659,
      "grad_norm": 0.7054859399795532,
      "learning_rate": 9.00771881234269e-06,
      "loss": 0.8215,
      "step": 7493
    },
    {
      "epoch": 0.4124607848533216,
      "grad_norm": 0.6840499639511108,
      "learning_rate": 9.007459611937854e-06,
      "loss": 0.776,
      "step": 7494
    },
    {
      "epoch": 0.41251582365567724,
      "grad_norm": 0.7340932488441467,
      "learning_rate": 9.007200381414026e-06,
      "loss": 0.713,
      "step": 7495
    },
    {
      "epoch": 0.4125708624580329,
      "grad_norm": 0.8282599449157715,
      "learning_rate": 9.00694112077315e-06,
      "loss": 0.7037,
      "step": 7496
    },
    {
      "epoch": 0.41262590126038856,
      "grad_norm": 0.849588930606842,
      "learning_rate": 9.00668183001718e-06,
      "loss": 0.7845,
      "step": 7497
    },
    {
      "epoch": 0.41268094006274425,
      "grad_norm": 0.8330783843994141,
      "learning_rate": 9.00642250914806e-06,
      "loss": 0.9049,
      "step": 7498
    },
    {
      "epoch": 0.4127359788650999,
      "grad_norm": 0.7020101547241211,
      "learning_rate": 9.00616315816774e-06,
      "loss": 0.8146,
      "step": 7499
    },
    {
      "epoch": 0.4127910176674556,
      "grad_norm": 0.7632037997245789,
      "learning_rate": 9.005903777078173e-06,
      "loss": 0.6629,
      "step": 7500
    },
    {
      "epoch": 0.4128460564698112,
      "grad_norm": 0.7286840081214905,
      "learning_rate": 9.005644365881304e-06,
      "loss": 0.7795,
      "step": 7501
    },
    {
      "epoch": 0.4129010952721669,
      "grad_norm": 0.710451066493988,
      "learning_rate": 9.005384924579084e-06,
      "loss": 0.7615,
      "step": 7502
    },
    {
      "epoch": 0.41295613407452253,
      "grad_norm": 0.7657510042190552,
      "learning_rate": 9.005125453173463e-06,
      "loss": 0.8938,
      "step": 7503
    },
    {
      "epoch": 0.4130111728768782,
      "grad_norm": 0.6978467702865601,
      "learning_rate": 9.004865951666392e-06,
      "loss": 0.7464,
      "step": 7504
    },
    {
      "epoch": 0.41306621167923385,
      "grad_norm": 0.7028319835662842,
      "learning_rate": 9.00460642005982e-06,
      "loss": 0.7899,
      "step": 7505
    },
    {
      "epoch": 0.41312125048158954,
      "grad_norm": 0.923951268196106,
      "learning_rate": 9.004346858355698e-06,
      "loss": 0.8851,
      "step": 7506
    },
    {
      "epoch": 0.4131762892839452,
      "grad_norm": 0.7293704748153687,
      "learning_rate": 9.004087266555978e-06,
      "loss": 0.7594,
      "step": 7507
    },
    {
      "epoch": 0.41323132808630086,
      "grad_norm": 0.7458868622779846,
      "learning_rate": 9.003827644662608e-06,
      "loss": 0.7538,
      "step": 7508
    },
    {
      "epoch": 0.4132863668886565,
      "grad_norm": 0.6764113306999207,
      "learning_rate": 9.003567992677543e-06,
      "loss": 0.7303,
      "step": 7509
    },
    {
      "epoch": 0.4133414056910122,
      "grad_norm": 0.7827350497245789,
      "learning_rate": 9.003308310602732e-06,
      "loss": 0.7708,
      "step": 7510
    },
    {
      "epoch": 0.4133964444933678,
      "grad_norm": 0.7683281302452087,
      "learning_rate": 9.003048598440127e-06,
      "loss": 0.7971,
      "step": 7511
    },
    {
      "epoch": 0.4134514832957235,
      "grad_norm": 0.8793813586235046,
      "learning_rate": 9.002788856191679e-06,
      "loss": 0.7434,
      "step": 7512
    },
    {
      "epoch": 0.41350652209807914,
      "grad_norm": 0.6598063111305237,
      "learning_rate": 9.002529083859343e-06,
      "loss": 0.7082,
      "step": 7513
    },
    {
      "epoch": 0.41356156090043483,
      "grad_norm": 0.8239839673042297,
      "learning_rate": 9.002269281445071e-06,
      "loss": 0.8457,
      "step": 7514
    },
    {
      "epoch": 0.41361659970279047,
      "grad_norm": 0.7433123588562012,
      "learning_rate": 9.002009448950812e-06,
      "loss": 0.7399,
      "step": 7515
    },
    {
      "epoch": 0.41367163850514616,
      "grad_norm": 0.8310487866401672,
      "learning_rate": 9.001749586378524e-06,
      "loss": 0.7482,
      "step": 7516
    },
    {
      "epoch": 0.4137266773075018,
      "grad_norm": 0.7170824408531189,
      "learning_rate": 9.001489693730155e-06,
      "loss": 0.7856,
      "step": 7517
    },
    {
      "epoch": 0.4137817161098574,
      "grad_norm": 0.9063520431518555,
      "learning_rate": 9.00122977100766e-06,
      "loss": 0.8623,
      "step": 7518
    },
    {
      "epoch": 0.4138367549122131,
      "grad_norm": 0.8753733038902283,
      "learning_rate": 9.000969818212996e-06,
      "loss": 0.7875,
      "step": 7519
    },
    {
      "epoch": 0.41389179371456875,
      "grad_norm": 0.7013519406318665,
      "learning_rate": 9.000709835348112e-06,
      "loss": 0.724,
      "step": 7520
    },
    {
      "epoch": 0.41394683251692443,
      "grad_norm": 0.7385973334312439,
      "learning_rate": 9.000449822414963e-06,
      "loss": 0.7286,
      "step": 7521
    },
    {
      "epoch": 0.41400187131928007,
      "grad_norm": 0.7605431079864502,
      "learning_rate": 9.000189779415505e-06,
      "loss": 0.728,
      "step": 7522
    },
    {
      "epoch": 0.41405691012163576,
      "grad_norm": 0.7631710767745972,
      "learning_rate": 8.99992970635169e-06,
      "loss": 0.8276,
      "step": 7523
    },
    {
      "epoch": 0.4141119489239914,
      "grad_norm": 0.8066657185554504,
      "learning_rate": 8.999669603225477e-06,
      "loss": 0.8319,
      "step": 7524
    },
    {
      "epoch": 0.4141669877263471,
      "grad_norm": 0.689407229423523,
      "learning_rate": 8.999409470038815e-06,
      "loss": 0.6675,
      "step": 7525
    },
    {
      "epoch": 0.4142220265287027,
      "grad_norm": 0.7391255497932434,
      "learning_rate": 8.999149306793664e-06,
      "loss": 0.8228,
      "step": 7526
    },
    {
      "epoch": 0.4142770653310584,
      "grad_norm": 0.7208844423294067,
      "learning_rate": 8.998889113491977e-06,
      "loss": 0.7689,
      "step": 7527
    },
    {
      "epoch": 0.41433210413341404,
      "grad_norm": 0.8278803825378418,
      "learning_rate": 8.99862889013571e-06,
      "loss": 0.7964,
      "step": 7528
    },
    {
      "epoch": 0.4143871429357697,
      "grad_norm": 0.7287253141403198,
      "learning_rate": 8.998368636726817e-06,
      "loss": 0.7689,
      "step": 7529
    },
    {
      "epoch": 0.41444218173812536,
      "grad_norm": 0.7159145474433899,
      "learning_rate": 8.998108353267257e-06,
      "loss": 0.7537,
      "step": 7530
    },
    {
      "epoch": 0.41449722054048105,
      "grad_norm": 0.7605739235877991,
      "learning_rate": 8.997848039758985e-06,
      "loss": 0.7327,
      "step": 7531
    },
    {
      "epoch": 0.4145522593428367,
      "grad_norm": 0.7290406227111816,
      "learning_rate": 8.997587696203958e-06,
      "loss": 0.6804,
      "step": 7532
    },
    {
      "epoch": 0.41460729814519237,
      "grad_norm": 0.7613189816474915,
      "learning_rate": 8.997327322604131e-06,
      "loss": 0.7465,
      "step": 7533
    },
    {
      "epoch": 0.414662336947548,
      "grad_norm": 0.7796703577041626,
      "learning_rate": 8.99706691896146e-06,
      "loss": 0.7444,
      "step": 7534
    },
    {
      "epoch": 0.4147173757499037,
      "grad_norm": 0.8758549094200134,
      "learning_rate": 8.996806485277904e-06,
      "loss": 0.8586,
      "step": 7535
    },
    {
      "epoch": 0.4147724145522593,
      "grad_norm": 0.9599420428276062,
      "learning_rate": 8.996546021555423e-06,
      "loss": 0.7554,
      "step": 7536
    },
    {
      "epoch": 0.414827453354615,
      "grad_norm": 0.8216326236724854,
      "learning_rate": 8.996285527795972e-06,
      "loss": 0.7995,
      "step": 7537
    },
    {
      "epoch": 0.41488249215697065,
      "grad_norm": 0.6777452230453491,
      "learning_rate": 8.996025004001507e-06,
      "loss": 0.7809,
      "step": 7538
    },
    {
      "epoch": 0.41493753095932634,
      "grad_norm": 0.7354100942611694,
      "learning_rate": 8.995764450173989e-06,
      "loss": 0.6548,
      "step": 7539
    },
    {
      "epoch": 0.414992569761682,
      "grad_norm": 0.7548280358314514,
      "learning_rate": 8.995503866315373e-06,
      "loss": 0.8308,
      "step": 7540
    },
    {
      "epoch": 0.41504760856403766,
      "grad_norm": 0.6891447901725769,
      "learning_rate": 8.995243252427622e-06,
      "loss": 0.8386,
      "step": 7541
    },
    {
      "epoch": 0.4151026473663933,
      "grad_norm": 0.6848340034484863,
      "learning_rate": 8.99498260851269e-06,
      "loss": 0.7587,
      "step": 7542
    },
    {
      "epoch": 0.415157686168749,
      "grad_norm": 0.7109090685844421,
      "learning_rate": 8.994721934572538e-06,
      "loss": 0.6847,
      "step": 7543
    },
    {
      "epoch": 0.4152127249711046,
      "grad_norm": 0.6708144545555115,
      "learning_rate": 8.994461230609128e-06,
      "loss": 0.7266,
      "step": 7544
    },
    {
      "epoch": 0.4152677637734603,
      "grad_norm": 0.6985414028167725,
      "learning_rate": 8.994200496624415e-06,
      "loss": 0.7696,
      "step": 7545
    },
    {
      "epoch": 0.41532280257581594,
      "grad_norm": 0.6989198923110962,
      "learning_rate": 8.993939732620359e-06,
      "loss": 0.7894,
      "step": 7546
    },
    {
      "epoch": 0.41537784137817163,
      "grad_norm": 0.6667589545249939,
      "learning_rate": 8.993678938598921e-06,
      "loss": 0.7417,
      "step": 7547
    },
    {
      "epoch": 0.41543288018052726,
      "grad_norm": 1.0692487955093384,
      "learning_rate": 8.993418114562064e-06,
      "loss": 0.7147,
      "step": 7548
    },
    {
      "epoch": 0.41548791898288295,
      "grad_norm": 0.6709207892417908,
      "learning_rate": 8.993157260511742e-06,
      "loss": 0.7694,
      "step": 7549
    },
    {
      "epoch": 0.4155429577852386,
      "grad_norm": 0.6714604496955872,
      "learning_rate": 8.992896376449923e-06,
      "loss": 0.6969,
      "step": 7550
    },
    {
      "epoch": 0.4155979965875943,
      "grad_norm": 0.8266897201538086,
      "learning_rate": 8.99263546237856e-06,
      "loss": 0.8392,
      "step": 7551
    },
    {
      "epoch": 0.4156530353899499,
      "grad_norm": 0.675188422203064,
      "learning_rate": 8.992374518299619e-06,
      "loss": 0.7525,
      "step": 7552
    },
    {
      "epoch": 0.4157080741923056,
      "grad_norm": 0.7406265139579773,
      "learning_rate": 8.992113544215059e-06,
      "loss": 0.7895,
      "step": 7553
    },
    {
      "epoch": 0.41576311299466123,
      "grad_norm": 0.837336003780365,
      "learning_rate": 8.991852540126844e-06,
      "loss": 0.7376,
      "step": 7554
    },
    {
      "epoch": 0.4158181517970169,
      "grad_norm": 0.6774994730949402,
      "learning_rate": 8.991591506036931e-06,
      "loss": 0.7231,
      "step": 7555
    },
    {
      "epoch": 0.41587319059937256,
      "grad_norm": 0.6941245794296265,
      "learning_rate": 8.991330441947287e-06,
      "loss": 0.7213,
      "step": 7556
    },
    {
      "epoch": 0.41592822940172824,
      "grad_norm": 0.7588210105895996,
      "learning_rate": 8.991069347859871e-06,
      "loss": 0.7829,
      "step": 7557
    },
    {
      "epoch": 0.4159832682040839,
      "grad_norm": 0.7580196857452393,
      "learning_rate": 8.990808223776647e-06,
      "loss": 0.7782,
      "step": 7558
    },
    {
      "epoch": 0.41603830700643957,
      "grad_norm": 0.7597478032112122,
      "learning_rate": 8.990547069699576e-06,
      "loss": 0.7764,
      "step": 7559
    },
    {
      "epoch": 0.4160933458087952,
      "grad_norm": 0.7950314283370972,
      "learning_rate": 8.990285885630622e-06,
      "loss": 0.7263,
      "step": 7560
    },
    {
      "epoch": 0.41614838461115083,
      "grad_norm": 0.6962432265281677,
      "learning_rate": 8.990024671571747e-06,
      "loss": 0.6616,
      "step": 7561
    },
    {
      "epoch": 0.4162034234135065,
      "grad_norm": 0.682816207408905,
      "learning_rate": 8.989763427524915e-06,
      "loss": 0.7862,
      "step": 7562
    },
    {
      "epoch": 0.41625846221586216,
      "grad_norm": 0.686673104763031,
      "learning_rate": 8.989502153492089e-06,
      "loss": 0.8199,
      "step": 7563
    },
    {
      "epoch": 0.41631350101821785,
      "grad_norm": 0.7954965233802795,
      "learning_rate": 8.989240849475231e-06,
      "loss": 0.8021,
      "step": 7564
    },
    {
      "epoch": 0.4163685398205735,
      "grad_norm": 0.7516284584999084,
      "learning_rate": 8.988979515476309e-06,
      "loss": 0.7803,
      "step": 7565
    },
    {
      "epoch": 0.41642357862292917,
      "grad_norm": 0.7148317694664001,
      "learning_rate": 8.988718151497284e-06,
      "loss": 0.7407,
      "step": 7566
    },
    {
      "epoch": 0.4164786174252848,
      "grad_norm": 0.7898986339569092,
      "learning_rate": 8.98845675754012e-06,
      "loss": 0.8382,
      "step": 7567
    },
    {
      "epoch": 0.4165336562276405,
      "grad_norm": 0.7014235854148865,
      "learning_rate": 8.988195333606784e-06,
      "loss": 0.7205,
      "step": 7568
    },
    {
      "epoch": 0.4165886950299961,
      "grad_norm": 0.6520957350730896,
      "learning_rate": 8.987933879699238e-06,
      "loss": 0.7452,
      "step": 7569
    },
    {
      "epoch": 0.4166437338323518,
      "grad_norm": 0.7462863922119141,
      "learning_rate": 8.987672395819449e-06,
      "loss": 0.7787,
      "step": 7570
    },
    {
      "epoch": 0.41669877263470745,
      "grad_norm": 0.7366049885749817,
      "learning_rate": 8.987410881969382e-06,
      "loss": 0.7662,
      "step": 7571
    },
    {
      "epoch": 0.41675381143706314,
      "grad_norm": 0.7732293009757996,
      "learning_rate": 8.987149338151002e-06,
      "loss": 0.8258,
      "step": 7572
    },
    {
      "epoch": 0.41680885023941877,
      "grad_norm": 0.9309358596801758,
      "learning_rate": 8.986887764366275e-06,
      "loss": 0.6538,
      "step": 7573
    },
    {
      "epoch": 0.41686388904177446,
      "grad_norm": 0.6976680755615234,
      "learning_rate": 8.986626160617167e-06,
      "loss": 0.7175,
      "step": 7574
    },
    {
      "epoch": 0.4169189278441301,
      "grad_norm": 0.7541783452033997,
      "learning_rate": 8.986364526905645e-06,
      "loss": 0.8153,
      "step": 7575
    },
    {
      "epoch": 0.4169739666464858,
      "grad_norm": 0.8968943357467651,
      "learning_rate": 8.986102863233673e-06,
      "loss": 0.7859,
      "step": 7576
    },
    {
      "epoch": 0.4170290054488414,
      "grad_norm": 0.6910044550895691,
      "learning_rate": 8.985841169603218e-06,
      "loss": 0.8381,
      "step": 7577
    },
    {
      "epoch": 0.4170840442511971,
      "grad_norm": 0.8944257497787476,
      "learning_rate": 8.985579446016249e-06,
      "loss": 0.7062,
      "step": 7578
    },
    {
      "epoch": 0.41713908305355274,
      "grad_norm": 0.6665629744529724,
      "learning_rate": 8.98531769247473e-06,
      "loss": 0.7928,
      "step": 7579
    },
    {
      "epoch": 0.41719412185590843,
      "grad_norm": 0.7642979621887207,
      "learning_rate": 8.985055908980634e-06,
      "loss": 0.8442,
      "step": 7580
    },
    {
      "epoch": 0.41724916065826406,
      "grad_norm": 0.7575559020042419,
      "learning_rate": 8.98479409553592e-06,
      "loss": 0.795,
      "step": 7581
    },
    {
      "epoch": 0.41730419946061975,
      "grad_norm": 0.6567206978797913,
      "learning_rate": 8.984532252142563e-06,
      "loss": 0.713,
      "step": 7582
    },
    {
      "epoch": 0.4173592382629754,
      "grad_norm": 0.6677179336547852,
      "learning_rate": 8.984270378802527e-06,
      "loss": 0.8173,
      "step": 7583
    },
    {
      "epoch": 0.4174142770653311,
      "grad_norm": 0.6846007704734802,
      "learning_rate": 8.984008475517782e-06,
      "loss": 0.7154,
      "step": 7584
    },
    {
      "epoch": 0.4174693158676867,
      "grad_norm": 0.7758762836456299,
      "learning_rate": 8.983746542290294e-06,
      "loss": 0.8686,
      "step": 7585
    },
    {
      "epoch": 0.4175243546700424,
      "grad_norm": 0.6850305199623108,
      "learning_rate": 8.983484579122036e-06,
      "loss": 0.7568,
      "step": 7586
    },
    {
      "epoch": 0.41757939347239803,
      "grad_norm": 0.7165307998657227,
      "learning_rate": 8.983222586014973e-06,
      "loss": 0.7856,
      "step": 7587
    },
    {
      "epoch": 0.4176344322747537,
      "grad_norm": 0.7747449278831482,
      "learning_rate": 8.982960562971074e-06,
      "loss": 0.8148,
      "step": 7588
    },
    {
      "epoch": 0.41768947107710935,
      "grad_norm": 0.789235532283783,
      "learning_rate": 8.982698509992311e-06,
      "loss": 0.8021,
      "step": 7589
    },
    {
      "epoch": 0.41774450987946504,
      "grad_norm": 0.664186954498291,
      "learning_rate": 8.982436427080652e-06,
      "loss": 0.7394,
      "step": 7590
    },
    {
      "epoch": 0.4177995486818207,
      "grad_norm": 0.7045899033546448,
      "learning_rate": 8.982174314238069e-06,
      "loss": 0.7029,
      "step": 7591
    },
    {
      "epoch": 0.41785458748417637,
      "grad_norm": 0.7569751739501953,
      "learning_rate": 8.981912171466525e-06,
      "loss": 0.6106,
      "step": 7592
    },
    {
      "epoch": 0.417909626286532,
      "grad_norm": 0.7383938431739807,
      "learning_rate": 8.981649998767998e-06,
      "loss": 0.8163,
      "step": 7593
    },
    {
      "epoch": 0.4179646650888877,
      "grad_norm": 0.7314342856407166,
      "learning_rate": 8.981387796144456e-06,
      "loss": 0.6847,
      "step": 7594
    },
    {
      "epoch": 0.4180197038912433,
      "grad_norm": 0.7249840497970581,
      "learning_rate": 8.981125563597867e-06,
      "loss": 0.8025,
      "step": 7595
    },
    {
      "epoch": 0.418074742693599,
      "grad_norm": 0.7260022759437561,
      "learning_rate": 8.980863301130206e-06,
      "loss": 0.7807,
      "step": 7596
    },
    {
      "epoch": 0.41812978149595464,
      "grad_norm": 0.6249421834945679,
      "learning_rate": 8.980601008743441e-06,
      "loss": 0.6744,
      "step": 7597
    },
    {
      "epoch": 0.41818482029831033,
      "grad_norm": 0.8132835626602173,
      "learning_rate": 8.980338686439544e-06,
      "loss": 0.7992,
      "step": 7598
    },
    {
      "epoch": 0.41823985910066597,
      "grad_norm": 0.7279506921768188,
      "learning_rate": 8.980076334220487e-06,
      "loss": 0.8402,
      "step": 7599
    },
    {
      "epoch": 0.41829489790302166,
      "grad_norm": 0.7168325781822205,
      "learning_rate": 8.979813952088242e-06,
      "loss": 0.9107,
      "step": 7600
    },
    {
      "epoch": 0.4183499367053773,
      "grad_norm": 0.633661150932312,
      "learning_rate": 8.97955154004478e-06,
      "loss": 0.6328,
      "step": 7601
    },
    {
      "epoch": 0.418404975507733,
      "grad_norm": 0.6770638227462769,
      "learning_rate": 8.979289098092074e-06,
      "loss": 0.7604,
      "step": 7602
    },
    {
      "epoch": 0.4184600143100886,
      "grad_norm": 0.7589067816734314,
      "learning_rate": 8.979026626232098e-06,
      "loss": 0.7774,
      "step": 7603
    },
    {
      "epoch": 0.41851505311244425,
      "grad_norm": 0.7116312980651855,
      "learning_rate": 8.97876412446682e-06,
      "loss": 0.8186,
      "step": 7604
    },
    {
      "epoch": 0.41857009191479994,
      "grad_norm": 0.7369259595870972,
      "learning_rate": 8.978501592798219e-06,
      "loss": 0.6705,
      "step": 7605
    },
    {
      "epoch": 0.41862513071715557,
      "grad_norm": 0.6201806664466858,
      "learning_rate": 8.978239031228265e-06,
      "loss": 0.7011,
      "step": 7606
    },
    {
      "epoch": 0.41868016951951126,
      "grad_norm": 0.7652842998504639,
      "learning_rate": 8.977976439758929e-06,
      "loss": 0.8112,
      "step": 7607
    },
    {
      "epoch": 0.4187352083218669,
      "grad_norm": 0.7214640974998474,
      "learning_rate": 8.97771381839219e-06,
      "loss": 0.767,
      "step": 7608
    },
    {
      "epoch": 0.4187902471242226,
      "grad_norm": 0.8093706369400024,
      "learning_rate": 8.977451167130015e-06,
      "loss": 0.8112,
      "step": 7609
    },
    {
      "epoch": 0.4188452859265782,
      "grad_norm": 0.7023005485534668,
      "learning_rate": 8.977188485974382e-06,
      "loss": 0.7678,
      "step": 7610
    },
    {
      "epoch": 0.4189003247289339,
      "grad_norm": 0.8126183748245239,
      "learning_rate": 8.976925774927267e-06,
      "loss": 0.8207,
      "step": 7611
    },
    {
      "epoch": 0.41895536353128954,
      "grad_norm": 0.9624595642089844,
      "learning_rate": 8.976663033990643e-06,
      "loss": 0.7853,
      "step": 7612
    },
    {
      "epoch": 0.4190104023336452,
      "grad_norm": 0.7866421937942505,
      "learning_rate": 8.976400263166483e-06,
      "loss": 0.6319,
      "step": 7613
    },
    {
      "epoch": 0.41906544113600086,
      "grad_norm": 0.7555810213088989,
      "learning_rate": 8.976137462456762e-06,
      "loss": 0.7781,
      "step": 7614
    },
    {
      "epoch": 0.41912047993835655,
      "grad_norm": 0.7383303046226501,
      "learning_rate": 8.975874631863457e-06,
      "loss": 0.8152,
      "step": 7615
    },
    {
      "epoch": 0.4191755187407122,
      "grad_norm": 0.7873355746269226,
      "learning_rate": 8.975611771388542e-06,
      "loss": 0.723,
      "step": 7616
    },
    {
      "epoch": 0.41923055754306787,
      "grad_norm": 0.7265962362289429,
      "learning_rate": 8.975348881033993e-06,
      "loss": 0.8016,
      "step": 7617
    },
    {
      "epoch": 0.4192855963454235,
      "grad_norm": 0.7074393033981323,
      "learning_rate": 8.975085960801788e-06,
      "loss": 0.7453,
      "step": 7618
    },
    {
      "epoch": 0.4193406351477792,
      "grad_norm": 0.6975581049919128,
      "learning_rate": 8.9748230106939e-06,
      "loss": 0.6516,
      "step": 7619
    },
    {
      "epoch": 0.41939567395013483,
      "grad_norm": 0.7730469107627869,
      "learning_rate": 8.974560030712304e-06,
      "loss": 0.7297,
      "step": 7620
    },
    {
      "epoch": 0.4194507127524905,
      "grad_norm": 0.7289026379585266,
      "learning_rate": 8.974297020858982e-06,
      "loss": 0.7087,
      "step": 7621
    },
    {
      "epoch": 0.41950575155484615,
      "grad_norm": 0.8029256463050842,
      "learning_rate": 8.974033981135906e-06,
      "loss": 0.7923,
      "step": 7622
    },
    {
      "epoch": 0.41956079035720184,
      "grad_norm": 0.765312135219574,
      "learning_rate": 8.973770911545055e-06,
      "loss": 0.7824,
      "step": 7623
    },
    {
      "epoch": 0.4196158291595575,
      "grad_norm": 0.7903861403465271,
      "learning_rate": 8.973507812088404e-06,
      "loss": 0.8207,
      "step": 7624
    },
    {
      "epoch": 0.41967086796191316,
      "grad_norm": 0.6875497698783875,
      "learning_rate": 8.973244682767934e-06,
      "loss": 0.7972,
      "step": 7625
    },
    {
      "epoch": 0.4197259067642688,
      "grad_norm": 0.7781878709793091,
      "learning_rate": 8.972981523585617e-06,
      "loss": 0.754,
      "step": 7626
    },
    {
      "epoch": 0.4197809455666245,
      "grad_norm": 0.6495640873908997,
      "learning_rate": 8.972718334543437e-06,
      "loss": 0.6851,
      "step": 7627
    },
    {
      "epoch": 0.4198359843689801,
      "grad_norm": 0.7610780596733093,
      "learning_rate": 8.97245511564337e-06,
      "loss": 0.8161,
      "step": 7628
    },
    {
      "epoch": 0.4198910231713358,
      "grad_norm": 0.7764771580696106,
      "learning_rate": 8.972191866887393e-06,
      "loss": 0.8341,
      "step": 7629
    },
    {
      "epoch": 0.41994606197369144,
      "grad_norm": 0.7709774374961853,
      "learning_rate": 8.971928588277485e-06,
      "loss": 0.765,
      "step": 7630
    },
    {
      "epoch": 0.42000110077604713,
      "grad_norm": 0.8213009238243103,
      "learning_rate": 8.971665279815625e-06,
      "loss": 0.8971,
      "step": 7631
    },
    {
      "epoch": 0.42005613957840277,
      "grad_norm": 0.7232406735420227,
      "learning_rate": 8.971401941503792e-06,
      "loss": 0.7919,
      "step": 7632
    },
    {
      "epoch": 0.42011117838075845,
      "grad_norm": 0.7322028279304504,
      "learning_rate": 8.971138573343964e-06,
      "loss": 0.8167,
      "step": 7633
    },
    {
      "epoch": 0.4201662171831141,
      "grad_norm": 0.7204442024230957,
      "learning_rate": 8.970875175338123e-06,
      "loss": 0.8152,
      "step": 7634
    },
    {
      "epoch": 0.4202212559854698,
      "grad_norm": 0.7385342121124268,
      "learning_rate": 8.970611747488246e-06,
      "loss": 0.8204,
      "step": 7635
    },
    {
      "epoch": 0.4202762947878254,
      "grad_norm": 0.758941113948822,
      "learning_rate": 8.970348289796316e-06,
      "loss": 0.8402,
      "step": 7636
    },
    {
      "epoch": 0.4203313335901811,
      "grad_norm": 0.7331902384757996,
      "learning_rate": 8.970084802264309e-06,
      "loss": 0.7305,
      "step": 7637
    },
    {
      "epoch": 0.42038637239253673,
      "grad_norm": 0.7822885513305664,
      "learning_rate": 8.969821284894208e-06,
      "loss": 0.8708,
      "step": 7638
    },
    {
      "epoch": 0.4204414111948924,
      "grad_norm": 0.6625984311103821,
      "learning_rate": 8.969557737687992e-06,
      "loss": 0.7806,
      "step": 7639
    },
    {
      "epoch": 0.42049644999724806,
      "grad_norm": 1.02848482131958,
      "learning_rate": 8.969294160647645e-06,
      "loss": 0.7176,
      "step": 7640
    },
    {
      "epoch": 0.42055148879960375,
      "grad_norm": 0.7888724207878113,
      "learning_rate": 8.969030553775144e-06,
      "loss": 0.8326,
      "step": 7641
    },
    {
      "epoch": 0.4206065276019594,
      "grad_norm": 0.7148883938789368,
      "learning_rate": 8.968766917072472e-06,
      "loss": 0.7405,
      "step": 7642
    },
    {
      "epoch": 0.42066156640431507,
      "grad_norm": 0.6629698872566223,
      "learning_rate": 8.96850325054161e-06,
      "loss": 0.845,
      "step": 7643
    },
    {
      "epoch": 0.4207166052066707,
      "grad_norm": 0.8414682149887085,
      "learning_rate": 8.96823955418454e-06,
      "loss": 1.3631,
      "step": 7644
    },
    {
      "epoch": 0.4207716440090264,
      "grad_norm": 0.7105298638343811,
      "learning_rate": 8.967975828003244e-06,
      "loss": 0.6808,
      "step": 7645
    },
    {
      "epoch": 0.420826682811382,
      "grad_norm": 0.7324852347373962,
      "learning_rate": 8.967712071999703e-06,
      "loss": 0.8237,
      "step": 7646
    },
    {
      "epoch": 0.42088172161373766,
      "grad_norm": 0.737324595451355,
      "learning_rate": 8.9674482861759e-06,
      "loss": 0.8486,
      "step": 7647
    },
    {
      "epoch": 0.42093676041609335,
      "grad_norm": 0.6763800382614136,
      "learning_rate": 8.967184470533818e-06,
      "loss": 0.72,
      "step": 7648
    },
    {
      "epoch": 0.420991799218449,
      "grad_norm": 0.7560757994651794,
      "learning_rate": 8.96692062507544e-06,
      "loss": 0.7704,
      "step": 7649
    },
    {
      "epoch": 0.42104683802080467,
      "grad_norm": 0.7289260029792786,
      "learning_rate": 8.966656749802748e-06,
      "loss": 0.7411,
      "step": 7650
    },
    {
      "epoch": 0.4211018768231603,
      "grad_norm": 0.6935442686080933,
      "learning_rate": 8.966392844717726e-06,
      "loss": 0.7848,
      "step": 7651
    },
    {
      "epoch": 0.421156915625516,
      "grad_norm": 0.7111918330192566,
      "learning_rate": 8.966128909822356e-06,
      "loss": 0.8377,
      "step": 7652
    },
    {
      "epoch": 0.4212119544278716,
      "grad_norm": 0.8594884872436523,
      "learning_rate": 8.965864945118625e-06,
      "loss": 0.8227,
      "step": 7653
    },
    {
      "epoch": 0.4212669932302273,
      "grad_norm": 0.6521008014678955,
      "learning_rate": 8.965600950608513e-06,
      "loss": 0.7034,
      "step": 7654
    },
    {
      "epoch": 0.42132203203258295,
      "grad_norm": 0.6362404823303223,
      "learning_rate": 8.965336926294007e-06,
      "loss": 0.6712,
      "step": 7655
    },
    {
      "epoch": 0.42137707083493864,
      "grad_norm": 0.6955040097236633,
      "learning_rate": 8.965072872177088e-06,
      "loss": 0.7789,
      "step": 7656
    },
    {
      "epoch": 0.42143210963729427,
      "grad_norm": 0.7311720252037048,
      "learning_rate": 8.964808788259745e-06,
      "loss": 0.7522,
      "step": 7657
    },
    {
      "epoch": 0.42148714843964996,
      "grad_norm": 0.781131386756897,
      "learning_rate": 8.96454467454396e-06,
      "loss": 0.7831,
      "step": 7658
    },
    {
      "epoch": 0.4215421872420056,
      "grad_norm": 0.6740639805793762,
      "learning_rate": 8.964280531031718e-06,
      "loss": 0.7102,
      "step": 7659
    },
    {
      "epoch": 0.4215972260443613,
      "grad_norm": 0.7843424677848816,
      "learning_rate": 8.964016357725003e-06,
      "loss": 0.8325,
      "step": 7660
    },
    {
      "epoch": 0.4216522648467169,
      "grad_norm": 0.7833517789840698,
      "learning_rate": 8.963752154625804e-06,
      "loss": 0.8603,
      "step": 7661
    },
    {
      "epoch": 0.4217073036490726,
      "grad_norm": 0.7270992994308472,
      "learning_rate": 8.963487921736104e-06,
      "loss": 0.745,
      "step": 7662
    },
    {
      "epoch": 0.42176234245142824,
      "grad_norm": 0.6517582535743713,
      "learning_rate": 8.963223659057892e-06,
      "loss": 0.6983,
      "step": 7663
    },
    {
      "epoch": 0.42181738125378393,
      "grad_norm": 0.6974934935569763,
      "learning_rate": 8.962959366593149e-06,
      "loss": 0.733,
      "step": 7664
    },
    {
      "epoch": 0.42187242005613956,
      "grad_norm": 0.712045431137085,
      "learning_rate": 8.962695044343865e-06,
      "loss": 0.725,
      "step": 7665
    },
    {
      "epoch": 0.42192745885849525,
      "grad_norm": 0.7311459183692932,
      "learning_rate": 8.962430692312028e-06,
      "loss": 0.8025,
      "step": 7666
    },
    {
      "epoch": 0.4219824976608509,
      "grad_norm": 0.7439966201782227,
      "learning_rate": 8.962166310499621e-06,
      "loss": 0.7711,
      "step": 7667
    },
    {
      "epoch": 0.4220375364632066,
      "grad_norm": 0.690832257270813,
      "learning_rate": 8.961901898908632e-06,
      "loss": 0.8414,
      "step": 7668
    },
    {
      "epoch": 0.4220925752655622,
      "grad_norm": 0.8437964916229248,
      "learning_rate": 8.961637457541049e-06,
      "loss": 0.8253,
      "step": 7669
    },
    {
      "epoch": 0.4221476140679179,
      "grad_norm": 0.7876344323158264,
      "learning_rate": 8.96137298639886e-06,
      "loss": 0.754,
      "step": 7670
    },
    {
      "epoch": 0.42220265287027353,
      "grad_norm": 0.7551780343055725,
      "learning_rate": 8.961108485484052e-06,
      "loss": 0.8555,
      "step": 7671
    },
    {
      "epoch": 0.4222576916726292,
      "grad_norm": 0.6867276430130005,
      "learning_rate": 8.96084395479861e-06,
      "loss": 0.7216,
      "step": 7672
    },
    {
      "epoch": 0.42231273047498485,
      "grad_norm": 0.9052873849868774,
      "learning_rate": 8.960579394344528e-06,
      "loss": 0.7945,
      "step": 7673
    },
    {
      "epoch": 0.42236776927734054,
      "grad_norm": 0.6731994152069092,
      "learning_rate": 8.96031480412379e-06,
      "loss": 0.7691,
      "step": 7674
    },
    {
      "epoch": 0.4224228080796962,
      "grad_norm": 0.7074670195579529,
      "learning_rate": 8.960050184138389e-06,
      "loss": 0.8008,
      "step": 7675
    },
    {
      "epoch": 0.42247784688205187,
      "grad_norm": 0.9482604265213013,
      "learning_rate": 8.959785534390309e-06,
      "loss": 0.7095,
      "step": 7676
    },
    {
      "epoch": 0.4225328856844075,
      "grad_norm": 0.6915413737297058,
      "learning_rate": 8.95952085488154e-06,
      "loss": 0.6717,
      "step": 7677
    },
    {
      "epoch": 0.4225879244867632,
      "grad_norm": 0.7565900087356567,
      "learning_rate": 8.959256145614073e-06,
      "loss": 0.8311,
      "step": 7678
    },
    {
      "epoch": 0.4226429632891188,
      "grad_norm": 0.8307167887687683,
      "learning_rate": 8.958991406589896e-06,
      "loss": 0.8585,
      "step": 7679
    },
    {
      "epoch": 0.4226980020914745,
      "grad_norm": 0.7955091595649719,
      "learning_rate": 8.958726637811e-06,
      "loss": 0.8154,
      "step": 7680
    },
    {
      "epoch": 0.42275304089383015,
      "grad_norm": 0.7692292332649231,
      "learning_rate": 8.958461839279376e-06,
      "loss": 0.7965,
      "step": 7681
    },
    {
      "epoch": 0.42280807969618583,
      "grad_norm": 0.7355942726135254,
      "learning_rate": 8.95819701099701e-06,
      "loss": 0.7557,
      "step": 7682
    },
    {
      "epoch": 0.42286311849854147,
      "grad_norm": 0.8781518936157227,
      "learning_rate": 8.957932152965895e-06,
      "loss": 0.8033,
      "step": 7683
    },
    {
      "epoch": 0.42291815730089716,
      "grad_norm": 0.7180802226066589,
      "learning_rate": 8.957667265188022e-06,
      "loss": 0.7283,
      "step": 7684
    },
    {
      "epoch": 0.4229731961032528,
      "grad_norm": 0.6967236995697021,
      "learning_rate": 8.95740234766538e-06,
      "loss": 0.769,
      "step": 7685
    },
    {
      "epoch": 0.4230282349056085,
      "grad_norm": 0.7462503910064697,
      "learning_rate": 8.957137400399963e-06,
      "loss": 0.8179,
      "step": 7686
    },
    {
      "epoch": 0.4230832737079641,
      "grad_norm": 0.67714524269104,
      "learning_rate": 8.956872423393761e-06,
      "loss": 0.7976,
      "step": 7687
    },
    {
      "epoch": 0.4231383125103198,
      "grad_norm": 0.8239946365356445,
      "learning_rate": 8.956607416648763e-06,
      "loss": 0.7946,
      "step": 7688
    },
    {
      "epoch": 0.42319335131267544,
      "grad_norm": 0.6724610924720764,
      "learning_rate": 8.956342380166963e-06,
      "loss": 0.7633,
      "step": 7689
    },
    {
      "epoch": 0.42324839011503107,
      "grad_norm": 0.744987964630127,
      "learning_rate": 8.956077313950354e-06,
      "loss": 0.9028,
      "step": 7690
    },
    {
      "epoch": 0.42330342891738676,
      "grad_norm": 0.7700596451759338,
      "learning_rate": 8.955812218000925e-06,
      "loss": 0.8954,
      "step": 7691
    },
    {
      "epoch": 0.4233584677197424,
      "grad_norm": 0.6952996253967285,
      "learning_rate": 8.955547092320673e-06,
      "loss": 0.8094,
      "step": 7692
    },
    {
      "epoch": 0.4234135065220981,
      "grad_norm": 0.6410536766052246,
      "learning_rate": 8.955281936911586e-06,
      "loss": 0.6281,
      "step": 7693
    },
    {
      "epoch": 0.4234685453244537,
      "grad_norm": 1.0939754247665405,
      "learning_rate": 8.95501675177566e-06,
      "loss": 0.8239,
      "step": 7694
    },
    {
      "epoch": 0.4235235841268094,
      "grad_norm": 0.7419464588165283,
      "learning_rate": 8.954751536914885e-06,
      "loss": 0.8015,
      "step": 7695
    },
    {
      "epoch": 0.42357862292916504,
      "grad_norm": 0.8171356320381165,
      "learning_rate": 8.954486292331257e-06,
      "loss": 0.8183,
      "step": 7696
    },
    {
      "epoch": 0.4236336617315207,
      "grad_norm": 0.745884358882904,
      "learning_rate": 8.95422101802677e-06,
      "loss": 0.7457,
      "step": 7697
    },
    {
      "epoch": 0.42368870053387636,
      "grad_norm": 0.7355740070343018,
      "learning_rate": 8.953955714003414e-06,
      "loss": 0.7517,
      "step": 7698
    },
    {
      "epoch": 0.42374373933623205,
      "grad_norm": 0.7103458642959595,
      "learning_rate": 8.953690380263186e-06,
      "loss": 0.7306,
      "step": 7699
    },
    {
      "epoch": 0.4237987781385877,
      "grad_norm": 0.7453970909118652,
      "learning_rate": 8.95342501680808e-06,
      "loss": 0.8396,
      "step": 7700
    },
    {
      "epoch": 0.4238538169409434,
      "grad_norm": 0.7132760286331177,
      "learning_rate": 8.953159623640088e-06,
      "loss": 0.7861,
      "step": 7701
    },
    {
      "epoch": 0.423908855743299,
      "grad_norm": 0.785827100276947,
      "learning_rate": 8.952894200761209e-06,
      "loss": 0.8681,
      "step": 7702
    },
    {
      "epoch": 0.4239638945456547,
      "grad_norm": 0.7075281143188477,
      "learning_rate": 8.952628748173433e-06,
      "loss": 0.7257,
      "step": 7703
    },
    {
      "epoch": 0.42401893334801033,
      "grad_norm": 0.8205186724662781,
      "learning_rate": 8.952363265878758e-06,
      "loss": 0.7361,
      "step": 7704
    },
    {
      "epoch": 0.424073972150366,
      "grad_norm": 0.6517061591148376,
      "learning_rate": 8.952097753879181e-06,
      "loss": 0.7127,
      "step": 7705
    },
    {
      "epoch": 0.42412901095272165,
      "grad_norm": 0.7252761125564575,
      "learning_rate": 8.951832212176692e-06,
      "loss": 0.796,
      "step": 7706
    },
    {
      "epoch": 0.42418404975507734,
      "grad_norm": 0.6688609719276428,
      "learning_rate": 8.951566640773292e-06,
      "loss": 0.7698,
      "step": 7707
    },
    {
      "epoch": 0.424239088557433,
      "grad_norm": 0.7163566946983337,
      "learning_rate": 8.951301039670974e-06,
      "loss": 0.8069,
      "step": 7708
    },
    {
      "epoch": 0.42429412735978866,
      "grad_norm": 0.7027623057365417,
      "learning_rate": 8.951035408871735e-06,
      "loss": 0.7061,
      "step": 7709
    },
    {
      "epoch": 0.4243491661621443,
      "grad_norm": 0.9558683037757874,
      "learning_rate": 8.950769748377572e-06,
      "loss": 0.926,
      "step": 7710
    },
    {
      "epoch": 0.4244042049645,
      "grad_norm": 0.7173893451690674,
      "learning_rate": 8.950504058190482e-06,
      "loss": 0.7519,
      "step": 7711
    },
    {
      "epoch": 0.4244592437668556,
      "grad_norm": 0.8481128811836243,
      "learning_rate": 8.950238338312459e-06,
      "loss": 0.7804,
      "step": 7712
    },
    {
      "epoch": 0.4245142825692113,
      "grad_norm": 0.6957072615623474,
      "learning_rate": 8.949972588745502e-06,
      "loss": 0.611,
      "step": 7713
    },
    {
      "epoch": 0.42456932137156694,
      "grad_norm": 0.7910122871398926,
      "learning_rate": 8.94970680949161e-06,
      "loss": 0.8435,
      "step": 7714
    },
    {
      "epoch": 0.42462436017392263,
      "grad_norm": 0.8068616986274719,
      "learning_rate": 8.949441000552777e-06,
      "loss": 0.8658,
      "step": 7715
    },
    {
      "epoch": 0.42467939897627827,
      "grad_norm": 0.718110978603363,
      "learning_rate": 8.949175161931006e-06,
      "loss": 0.7908,
      "step": 7716
    },
    {
      "epoch": 0.42473443777863396,
      "grad_norm": 0.7329656481742859,
      "learning_rate": 8.948909293628289e-06,
      "loss": 0.7477,
      "step": 7717
    },
    {
      "epoch": 0.4247894765809896,
      "grad_norm": 0.7046940326690674,
      "learning_rate": 8.948643395646625e-06,
      "loss": 0.7985,
      "step": 7718
    },
    {
      "epoch": 0.4248445153833453,
      "grad_norm": 0.6699581742286682,
      "learning_rate": 8.948377467988017e-06,
      "loss": 0.6575,
      "step": 7719
    },
    {
      "epoch": 0.4248995541857009,
      "grad_norm": 0.8055217266082764,
      "learning_rate": 8.94811151065446e-06,
      "loss": 0.7008,
      "step": 7720
    },
    {
      "epoch": 0.4249545929880566,
      "grad_norm": 0.8374543190002441,
      "learning_rate": 8.947845523647954e-06,
      "loss": 0.8918,
      "step": 7721
    },
    {
      "epoch": 0.42500963179041223,
      "grad_norm": 0.6974833607673645,
      "learning_rate": 8.947579506970498e-06,
      "loss": 0.8594,
      "step": 7722
    },
    {
      "epoch": 0.4250646705927679,
      "grad_norm": 0.7466567754745483,
      "learning_rate": 8.947313460624091e-06,
      "loss": 0.6935,
      "step": 7723
    },
    {
      "epoch": 0.42511970939512356,
      "grad_norm": 0.8118101358413696,
      "learning_rate": 8.947047384610734e-06,
      "loss": 0.8432,
      "step": 7724
    },
    {
      "epoch": 0.42517474819747925,
      "grad_norm": 0.6885644197463989,
      "learning_rate": 8.946781278932422e-06,
      "loss": 0.8059,
      "step": 7725
    },
    {
      "epoch": 0.4252297869998349,
      "grad_norm": 0.7257012128829956,
      "learning_rate": 8.94651514359116e-06,
      "loss": 0.8239,
      "step": 7726
    },
    {
      "epoch": 0.42528482580219057,
      "grad_norm": 1.311591386795044,
      "learning_rate": 8.946248978588947e-06,
      "loss": 0.8207,
      "step": 7727
    },
    {
      "epoch": 0.4253398646045462,
      "grad_norm": 0.7694151997566223,
      "learning_rate": 8.945982783927784e-06,
      "loss": 0.8948,
      "step": 7728
    },
    {
      "epoch": 0.4253949034069019,
      "grad_norm": 0.6922980546951294,
      "learning_rate": 8.945716559609669e-06,
      "loss": 0.7883,
      "step": 7729
    },
    {
      "epoch": 0.4254499422092575,
      "grad_norm": 0.7803757786750793,
      "learning_rate": 8.945450305636605e-06,
      "loss": 0.9166,
      "step": 7730
    },
    {
      "epoch": 0.4255049810116132,
      "grad_norm": 0.6775311827659607,
      "learning_rate": 8.945184022010593e-06,
      "loss": 0.6976,
      "step": 7731
    },
    {
      "epoch": 0.42556001981396885,
      "grad_norm": 0.7108052968978882,
      "learning_rate": 8.944917708733634e-06,
      "loss": 0.7763,
      "step": 7732
    },
    {
      "epoch": 0.4256150586163245,
      "grad_norm": 0.7215770483016968,
      "learning_rate": 8.94465136580773e-06,
      "loss": 0.7907,
      "step": 7733
    },
    {
      "epoch": 0.42567009741868017,
      "grad_norm": 0.6690788865089417,
      "learning_rate": 8.944384993234881e-06,
      "loss": 0.8403,
      "step": 7734
    },
    {
      "epoch": 0.4257251362210358,
      "grad_norm": 0.7372478246688843,
      "learning_rate": 8.94411859101709e-06,
      "loss": 0.7618,
      "step": 7735
    },
    {
      "epoch": 0.4257801750233915,
      "grad_norm": 0.9398306608200073,
      "learning_rate": 8.94385215915636e-06,
      "loss": 0.9043,
      "step": 7736
    },
    {
      "epoch": 0.4258352138257471,
      "grad_norm": 0.8790311217308044,
      "learning_rate": 8.943585697654693e-06,
      "loss": 0.9378,
      "step": 7737
    },
    {
      "epoch": 0.4258902526281028,
      "grad_norm": 0.7579166889190674,
      "learning_rate": 8.943319206514091e-06,
      "loss": 0.7913,
      "step": 7738
    },
    {
      "epoch": 0.42594529143045845,
      "grad_norm": 0.6426860690116882,
      "learning_rate": 8.943052685736559e-06,
      "loss": 0.744,
      "step": 7739
    },
    {
      "epoch": 0.42600033023281414,
      "grad_norm": 0.688117265701294,
      "learning_rate": 8.942786135324098e-06,
      "loss": 0.8386,
      "step": 7740
    },
    {
      "epoch": 0.4260553690351698,
      "grad_norm": 0.7178692817687988,
      "learning_rate": 8.94251955527871e-06,
      "loss": 0.7937,
      "step": 7741
    },
    {
      "epoch": 0.42611040783752546,
      "grad_norm": 0.7980415225028992,
      "learning_rate": 8.942252945602403e-06,
      "loss": 0.76,
      "step": 7742
    },
    {
      "epoch": 0.4261654466398811,
      "grad_norm": 0.6858333349227905,
      "learning_rate": 8.941986306297175e-06,
      "loss": 0.8155,
      "step": 7743
    },
    {
      "epoch": 0.4262204854422368,
      "grad_norm": 0.763297975063324,
      "learning_rate": 8.941719637365037e-06,
      "loss": 0.8003,
      "step": 7744
    },
    {
      "epoch": 0.4262755242445924,
      "grad_norm": 0.661016047000885,
      "learning_rate": 8.941452938807986e-06,
      "loss": 0.6788,
      "step": 7745
    },
    {
      "epoch": 0.4263305630469481,
      "grad_norm": 0.7168089151382446,
      "learning_rate": 8.94118621062803e-06,
      "loss": 0.7791,
      "step": 7746
    },
    {
      "epoch": 0.42638560184930374,
      "grad_norm": 0.6879743337631226,
      "learning_rate": 8.940919452827174e-06,
      "loss": 0.7978,
      "step": 7747
    },
    {
      "epoch": 0.42644064065165943,
      "grad_norm": 0.672298014163971,
      "learning_rate": 8.940652665407424e-06,
      "loss": 0.7569,
      "step": 7748
    },
    {
      "epoch": 0.42649567945401506,
      "grad_norm": 0.7237414717674255,
      "learning_rate": 8.940385848370782e-06,
      "loss": 0.6788,
      "step": 7749
    },
    {
      "epoch": 0.42655071825637075,
      "grad_norm": 0.6793895363807678,
      "learning_rate": 8.940119001719255e-06,
      "loss": 0.749,
      "step": 7750
    },
    {
      "epoch": 0.4266057570587264,
      "grad_norm": 1.1172789335250854,
      "learning_rate": 8.939852125454847e-06,
      "loss": 0.9017,
      "step": 7751
    },
    {
      "epoch": 0.4266607958610821,
      "grad_norm": 0.7138717770576477,
      "learning_rate": 8.939585219579567e-06,
      "loss": 0.8586,
      "step": 7752
    },
    {
      "epoch": 0.4267158346634377,
      "grad_norm": 0.8678629398345947,
      "learning_rate": 8.939318284095417e-06,
      "loss": 0.7333,
      "step": 7753
    },
    {
      "epoch": 0.4267708734657934,
      "grad_norm": 0.7274941802024841,
      "learning_rate": 8.939051319004407e-06,
      "loss": 0.8426,
      "step": 7754
    },
    {
      "epoch": 0.42682591226814903,
      "grad_norm": 0.6845358610153198,
      "learning_rate": 8.93878432430854e-06,
      "loss": 0.7731,
      "step": 7755
    },
    {
      "epoch": 0.4268809510705047,
      "grad_norm": 0.7042781710624695,
      "learning_rate": 8.938517300009826e-06,
      "loss": 0.6703,
      "step": 7756
    },
    {
      "epoch": 0.42693598987286036,
      "grad_norm": 0.7147190570831299,
      "learning_rate": 8.93825024611027e-06,
      "loss": 0.7977,
      "step": 7757
    },
    {
      "epoch": 0.42699102867521604,
      "grad_norm": 0.6584187150001526,
      "learning_rate": 8.93798316261188e-06,
      "loss": 0.716,
      "step": 7758
    },
    {
      "epoch": 0.4270460674775717,
      "grad_norm": 0.8061439990997314,
      "learning_rate": 8.93771604951666e-06,
      "loss": 0.9075,
      "step": 7759
    },
    {
      "epoch": 0.42710110627992737,
      "grad_norm": 0.6741406917572021,
      "learning_rate": 8.937448906826622e-06,
      "loss": 0.7828,
      "step": 7760
    },
    {
      "epoch": 0.427156145082283,
      "grad_norm": 0.8791692852973938,
      "learning_rate": 8.937181734543773e-06,
      "loss": 0.7685,
      "step": 7761
    },
    {
      "epoch": 0.4272111838846387,
      "grad_norm": 0.6804112195968628,
      "learning_rate": 8.936914532670119e-06,
      "loss": 0.7672,
      "step": 7762
    },
    {
      "epoch": 0.4272662226869943,
      "grad_norm": 0.6983451843261719,
      "learning_rate": 8.936647301207668e-06,
      "loss": 0.8228,
      "step": 7763
    },
    {
      "epoch": 0.42732126148935,
      "grad_norm": 0.8248929977416992,
      "learning_rate": 8.936380040158432e-06,
      "loss": 0.7628,
      "step": 7764
    },
    {
      "epoch": 0.42737630029170565,
      "grad_norm": 0.8324941992759705,
      "learning_rate": 8.936112749524415e-06,
      "loss": 0.8125,
      "step": 7765
    },
    {
      "epoch": 0.42743133909406134,
      "grad_norm": 0.7489150762557983,
      "learning_rate": 8.935845429307631e-06,
      "loss": 0.8766,
      "step": 7766
    },
    {
      "epoch": 0.42748637789641697,
      "grad_norm": 0.7323104739189148,
      "learning_rate": 8.935578079510083e-06,
      "loss": 0.8607,
      "step": 7767
    },
    {
      "epoch": 0.42754141669877266,
      "grad_norm": 0.6825152635574341,
      "learning_rate": 8.935310700133786e-06,
      "loss": 0.7817,
      "step": 7768
    },
    {
      "epoch": 0.4275964555011283,
      "grad_norm": 0.8928677439689636,
      "learning_rate": 8.935043291180748e-06,
      "loss": 0.7621,
      "step": 7769
    },
    {
      "epoch": 0.427651494303484,
      "grad_norm": 0.7071405649185181,
      "learning_rate": 8.934775852652975e-06,
      "loss": 0.7798,
      "step": 7770
    },
    {
      "epoch": 0.4277065331058396,
      "grad_norm": 0.8225427269935608,
      "learning_rate": 8.934508384552481e-06,
      "loss": 0.7212,
      "step": 7771
    },
    {
      "epoch": 0.4277615719081953,
      "grad_norm": 0.6931234002113342,
      "learning_rate": 8.934240886881276e-06,
      "loss": 0.7301,
      "step": 7772
    },
    {
      "epoch": 0.42781661071055094,
      "grad_norm": 0.6901859641075134,
      "learning_rate": 8.933973359641369e-06,
      "loss": 0.6974,
      "step": 7773
    },
    {
      "epoch": 0.4278716495129066,
      "grad_norm": 0.7736960649490356,
      "learning_rate": 8.93370580283477e-06,
      "loss": 0.6562,
      "step": 7774
    },
    {
      "epoch": 0.42792668831526226,
      "grad_norm": 0.7363499999046326,
      "learning_rate": 8.933438216463495e-06,
      "loss": 0.8274,
      "step": 7775
    },
    {
      "epoch": 0.4279817271176179,
      "grad_norm": 0.6855602860450745,
      "learning_rate": 8.933170600529548e-06,
      "loss": 0.7576,
      "step": 7776
    },
    {
      "epoch": 0.4280367659199736,
      "grad_norm": 0.7641676664352417,
      "learning_rate": 8.932902955034945e-06,
      "loss": 0.7837,
      "step": 7777
    },
    {
      "epoch": 0.4280918047223292,
      "grad_norm": 0.74812251329422,
      "learning_rate": 8.932635279981695e-06,
      "loss": 0.8402,
      "step": 7778
    },
    {
      "epoch": 0.4281468435246849,
      "grad_norm": 0.7445259094238281,
      "learning_rate": 8.932367575371813e-06,
      "loss": 0.862,
      "step": 7779
    },
    {
      "epoch": 0.42820188232704054,
      "grad_norm": 0.8977177739143372,
      "learning_rate": 8.932099841207306e-06,
      "loss": 0.7735,
      "step": 7780
    },
    {
      "epoch": 0.42825692112939623,
      "grad_norm": 0.74172043800354,
      "learning_rate": 8.93183207749019e-06,
      "loss": 0.7053,
      "step": 7781
    },
    {
      "epoch": 0.42831195993175186,
      "grad_norm": 0.6670083999633789,
      "learning_rate": 8.931564284222479e-06,
      "loss": 0.6348,
      "step": 7782
    },
    {
      "epoch": 0.42836699873410755,
      "grad_norm": 0.7575422525405884,
      "learning_rate": 8.93129646140618e-06,
      "loss": 0.9354,
      "step": 7783
    },
    {
      "epoch": 0.4284220375364632,
      "grad_norm": 0.7436977624893188,
      "learning_rate": 8.931028609043311e-06,
      "loss": 0.7461,
      "step": 7784
    },
    {
      "epoch": 0.4284770763388189,
      "grad_norm": 0.7383070588111877,
      "learning_rate": 8.930760727135882e-06,
      "loss": 0.7629,
      "step": 7785
    },
    {
      "epoch": 0.4285321151411745,
      "grad_norm": 0.6926067471504211,
      "learning_rate": 8.93049281568591e-06,
      "loss": 0.6788,
      "step": 7786
    },
    {
      "epoch": 0.4285871539435302,
      "grad_norm": 0.7680530548095703,
      "learning_rate": 8.930224874695404e-06,
      "loss": 0.722,
      "step": 7787
    },
    {
      "epoch": 0.42864219274588583,
      "grad_norm": 0.9880867004394531,
      "learning_rate": 8.92995690416638e-06,
      "loss": 0.833,
      "step": 7788
    },
    {
      "epoch": 0.4286972315482415,
      "grad_norm": 0.7915430068969727,
      "learning_rate": 8.929688904100853e-06,
      "loss": 0.7643,
      "step": 7789
    },
    {
      "epoch": 0.42875227035059715,
      "grad_norm": 0.6972275376319885,
      "learning_rate": 8.929420874500836e-06,
      "loss": 0.7697,
      "step": 7790
    },
    {
      "epoch": 0.42880730915295284,
      "grad_norm": 0.9583331346511841,
      "learning_rate": 8.929152815368343e-06,
      "loss": 0.7591,
      "step": 7791
    },
    {
      "epoch": 0.4288623479553085,
      "grad_norm": 0.7254299521446228,
      "learning_rate": 8.928884726705388e-06,
      "loss": 0.7913,
      "step": 7792
    },
    {
      "epoch": 0.42891738675766417,
      "grad_norm": 0.7925865054130554,
      "learning_rate": 8.928616608513989e-06,
      "loss": 0.8248,
      "step": 7793
    },
    {
      "epoch": 0.4289724255600198,
      "grad_norm": 0.9367457628250122,
      "learning_rate": 8.928348460796157e-06,
      "loss": 0.7767,
      "step": 7794
    },
    {
      "epoch": 0.4290274643623755,
      "grad_norm": 0.8511868119239807,
      "learning_rate": 8.928080283553912e-06,
      "loss": 0.841,
      "step": 7795
    },
    {
      "epoch": 0.4290825031647311,
      "grad_norm": 0.8518061637878418,
      "learning_rate": 8.927812076789267e-06,
      "loss": 0.7907,
      "step": 7796
    },
    {
      "epoch": 0.4291375419670868,
      "grad_norm": 0.7208365797996521,
      "learning_rate": 8.927543840504236e-06,
      "loss": 0.7344,
      "step": 7797
    },
    {
      "epoch": 0.42919258076944244,
      "grad_norm": 0.7541850209236145,
      "learning_rate": 8.927275574700838e-06,
      "loss": 0.7724,
      "step": 7798
    },
    {
      "epoch": 0.42924761957179813,
      "grad_norm": 0.7378629446029663,
      "learning_rate": 8.927007279381087e-06,
      "loss": 0.7614,
      "step": 7799
    },
    {
      "epoch": 0.42930265837415377,
      "grad_norm": 0.7358561158180237,
      "learning_rate": 8.926738954547001e-06,
      "loss": 0.7288,
      "step": 7800
    },
    {
      "epoch": 0.42935769717650946,
      "grad_norm": 0.7385967969894409,
      "learning_rate": 8.926470600200597e-06,
      "loss": 0.7562,
      "step": 7801
    },
    {
      "epoch": 0.4294127359788651,
      "grad_norm": 0.6904877424240112,
      "learning_rate": 8.92620221634389e-06,
      "loss": 0.6507,
      "step": 7802
    },
    {
      "epoch": 0.4294677747812208,
      "grad_norm": 0.7205148935317993,
      "learning_rate": 8.925933802978898e-06,
      "loss": 0.7683,
      "step": 7803
    },
    {
      "epoch": 0.4295228135835764,
      "grad_norm": 0.6830344200134277,
      "learning_rate": 8.925665360107639e-06,
      "loss": 0.6886,
      "step": 7804
    },
    {
      "epoch": 0.4295778523859321,
      "grad_norm": 0.7648812532424927,
      "learning_rate": 8.92539688773213e-06,
      "loss": 0.7559,
      "step": 7805
    },
    {
      "epoch": 0.42963289118828774,
      "grad_norm": 0.7819112539291382,
      "learning_rate": 8.925128385854389e-06,
      "loss": 0.7443,
      "step": 7806
    },
    {
      "epoch": 0.4296879299906434,
      "grad_norm": 0.6742433309555054,
      "learning_rate": 8.924859854476433e-06,
      "loss": 0.7191,
      "step": 7807
    },
    {
      "epoch": 0.42974296879299906,
      "grad_norm": 0.7368177771568298,
      "learning_rate": 8.924591293600281e-06,
      "loss": 0.6946,
      "step": 7808
    },
    {
      "epoch": 0.42979800759535475,
      "grad_norm": 0.663112998008728,
      "learning_rate": 8.924322703227953e-06,
      "loss": 0.7405,
      "step": 7809
    },
    {
      "epoch": 0.4298530463977104,
      "grad_norm": 0.6735410690307617,
      "learning_rate": 8.924054083361465e-06,
      "loss": 0.7982,
      "step": 7810
    },
    {
      "epoch": 0.42990808520006607,
      "grad_norm": 0.7770369648933411,
      "learning_rate": 8.923785434002834e-06,
      "loss": 0.9179,
      "step": 7811
    },
    {
      "epoch": 0.4299631240024217,
      "grad_norm": 0.7464482188224792,
      "learning_rate": 8.923516755154085e-06,
      "loss": 0.8514,
      "step": 7812
    },
    {
      "epoch": 0.4300181628047774,
      "grad_norm": 0.9249551892280579,
      "learning_rate": 8.923248046817235e-06,
      "loss": 0.8287,
      "step": 7813
    },
    {
      "epoch": 0.430073201607133,
      "grad_norm": 0.7071338891983032,
      "learning_rate": 8.922979308994302e-06,
      "loss": 0.7509,
      "step": 7814
    },
    {
      "epoch": 0.4301282404094887,
      "grad_norm": 0.6910794377326965,
      "learning_rate": 8.922710541687305e-06,
      "loss": 0.7373,
      "step": 7815
    },
    {
      "epoch": 0.43018327921184435,
      "grad_norm": 0.8424028158187866,
      "learning_rate": 8.922441744898267e-06,
      "loss": 0.741,
      "step": 7816
    },
    {
      "epoch": 0.43023831801420004,
      "grad_norm": 0.8162125945091248,
      "learning_rate": 8.922172918629208e-06,
      "loss": 0.8044,
      "step": 7817
    },
    {
      "epoch": 0.43029335681655567,
      "grad_norm": 0.7415170669555664,
      "learning_rate": 8.921904062882145e-06,
      "loss": 0.7427,
      "step": 7818
    },
    {
      "epoch": 0.4303483956189113,
      "grad_norm": 1.1357808113098145,
      "learning_rate": 8.921635177659103e-06,
      "loss": 0.7802,
      "step": 7819
    },
    {
      "epoch": 0.430403434421267,
      "grad_norm": 0.7039839625358582,
      "learning_rate": 8.9213662629621e-06,
      "loss": 0.7368,
      "step": 7820
    },
    {
      "epoch": 0.43045847322362263,
      "grad_norm": 0.721077024936676,
      "learning_rate": 8.921097318793157e-06,
      "loss": 0.6575,
      "step": 7821
    },
    {
      "epoch": 0.4305135120259783,
      "grad_norm": 0.7823510766029358,
      "learning_rate": 8.920828345154297e-06,
      "loss": 0.7499,
      "step": 7822
    },
    {
      "epoch": 0.43056855082833395,
      "grad_norm": 0.6400569677352905,
      "learning_rate": 8.920559342047539e-06,
      "loss": 0.7091,
      "step": 7823
    },
    {
      "epoch": 0.43062358963068964,
      "grad_norm": 0.8974951505661011,
      "learning_rate": 8.920290309474908e-06,
      "loss": 0.7228,
      "step": 7824
    },
    {
      "epoch": 0.4306786284330453,
      "grad_norm": 0.8176010847091675,
      "learning_rate": 8.920021247438426e-06,
      "loss": 0.8852,
      "step": 7825
    },
    {
      "epoch": 0.43073366723540096,
      "grad_norm": 0.7591422200202942,
      "learning_rate": 8.919752155940112e-06,
      "loss": 0.8382,
      "step": 7826
    },
    {
      "epoch": 0.4307887060377566,
      "grad_norm": 0.7089776396751404,
      "learning_rate": 8.919483034981988e-06,
      "loss": 0.7188,
      "step": 7827
    },
    {
      "epoch": 0.4308437448401123,
      "grad_norm": 0.7328840494155884,
      "learning_rate": 8.919213884566081e-06,
      "loss": 0.7609,
      "step": 7828
    },
    {
      "epoch": 0.4308987836424679,
      "grad_norm": 0.6473509669303894,
      "learning_rate": 8.918944704694411e-06,
      "loss": 0.7027,
      "step": 7829
    },
    {
      "epoch": 0.4309538224448236,
      "grad_norm": 0.6585624814033508,
      "learning_rate": 8.918675495369003e-06,
      "loss": 0.7133,
      "step": 7830
    },
    {
      "epoch": 0.43100886124717924,
      "grad_norm": 0.7232397794723511,
      "learning_rate": 8.918406256591876e-06,
      "loss": 0.7458,
      "step": 7831
    },
    {
      "epoch": 0.43106390004953493,
      "grad_norm": 0.8752645254135132,
      "learning_rate": 8.918136988365059e-06,
      "loss": 0.671,
      "step": 7832
    },
    {
      "epoch": 0.43111893885189057,
      "grad_norm": 0.7890885472297668,
      "learning_rate": 8.917867690690573e-06,
      "loss": 0.7674,
      "step": 7833
    },
    {
      "epoch": 0.43117397765424625,
      "grad_norm": 0.6725128293037415,
      "learning_rate": 8.917598363570441e-06,
      "loss": 0.7373,
      "step": 7834
    },
    {
      "epoch": 0.4312290164566019,
      "grad_norm": 0.808897852897644,
      "learning_rate": 8.917329007006688e-06,
      "loss": 0.8397,
      "step": 7835
    },
    {
      "epoch": 0.4312840552589576,
      "grad_norm": 0.7268605828285217,
      "learning_rate": 8.91705962100134e-06,
      "loss": 0.7957,
      "step": 7836
    },
    {
      "epoch": 0.4313390940613132,
      "grad_norm": 0.7336069345474243,
      "learning_rate": 8.916790205556421e-06,
      "loss": 0.746,
      "step": 7837
    },
    {
      "epoch": 0.4313941328636689,
      "grad_norm": 0.7380902171134949,
      "learning_rate": 8.916520760673955e-06,
      "loss": 0.674,
      "step": 7838
    },
    {
      "epoch": 0.43144917166602453,
      "grad_norm": 0.8041831851005554,
      "learning_rate": 8.916251286355967e-06,
      "loss": 0.8392,
      "step": 7839
    },
    {
      "epoch": 0.4315042104683802,
      "grad_norm": 0.6745681166648865,
      "learning_rate": 8.915981782604481e-06,
      "loss": 0.7676,
      "step": 7840
    },
    {
      "epoch": 0.43155924927073586,
      "grad_norm": 0.6572039127349854,
      "learning_rate": 8.915712249421526e-06,
      "loss": 0.7471,
      "step": 7841
    },
    {
      "epoch": 0.43161428807309155,
      "grad_norm": 0.7250062227249146,
      "learning_rate": 8.915442686809124e-06,
      "loss": 0.8566,
      "step": 7842
    },
    {
      "epoch": 0.4316693268754472,
      "grad_norm": 0.7008941769599915,
      "learning_rate": 8.915173094769306e-06,
      "loss": 0.7876,
      "step": 7843
    },
    {
      "epoch": 0.43172436567780287,
      "grad_norm": 0.7078337073326111,
      "learning_rate": 8.914903473304093e-06,
      "loss": 0.756,
      "step": 7844
    },
    {
      "epoch": 0.4317794044801585,
      "grad_norm": 0.7822949886322021,
      "learning_rate": 8.914633822415513e-06,
      "loss": 0.9423,
      "step": 7845
    },
    {
      "epoch": 0.4318344432825142,
      "grad_norm": 0.6707580089569092,
      "learning_rate": 8.914364142105593e-06,
      "loss": 0.639,
      "step": 7846
    },
    {
      "epoch": 0.4318894820848698,
      "grad_norm": 0.7868423461914062,
      "learning_rate": 8.914094432376362e-06,
      "loss": 0.7768,
      "step": 7847
    },
    {
      "epoch": 0.4319445208872255,
      "grad_norm": 0.6147592067718506,
      "learning_rate": 8.913824693229845e-06,
      "loss": 0.6693,
      "step": 7848
    },
    {
      "epoch": 0.43199955968958115,
      "grad_norm": 0.6901249885559082,
      "learning_rate": 8.913554924668067e-06,
      "loss": 0.7779,
      "step": 7849
    },
    {
      "epoch": 0.43205459849193684,
      "grad_norm": 0.7062137126922607,
      "learning_rate": 8.913285126693058e-06,
      "loss": 0.7951,
      "step": 7850
    },
    {
      "epoch": 0.43210963729429247,
      "grad_norm": 0.6363390684127808,
      "learning_rate": 8.913015299306846e-06,
      "loss": 0.6723,
      "step": 7851
    },
    {
      "epoch": 0.43216467609664816,
      "grad_norm": 0.7168677449226379,
      "learning_rate": 8.912745442511459e-06,
      "loss": 0.7442,
      "step": 7852
    },
    {
      "epoch": 0.4322197148990038,
      "grad_norm": 0.7347995042800903,
      "learning_rate": 8.912475556308925e-06,
      "loss": 0.8361,
      "step": 7853
    },
    {
      "epoch": 0.4322747537013595,
      "grad_norm": 0.683777391910553,
      "learning_rate": 8.91220564070127e-06,
      "loss": 0.7583,
      "step": 7854
    },
    {
      "epoch": 0.4323297925037151,
      "grad_norm": 0.7436330914497375,
      "learning_rate": 8.911935695690527e-06,
      "loss": 0.8414,
      "step": 7855
    },
    {
      "epoch": 0.4323848313060708,
      "grad_norm": 0.7748109102249146,
      "learning_rate": 8.911665721278721e-06,
      "loss": 0.7812,
      "step": 7856
    },
    {
      "epoch": 0.43243987010842644,
      "grad_norm": 0.7984411120414734,
      "learning_rate": 8.911395717467883e-06,
      "loss": 0.6845,
      "step": 7857
    },
    {
      "epoch": 0.4324949089107821,
      "grad_norm": 0.680144727230072,
      "learning_rate": 8.911125684260042e-06,
      "loss": 0.7156,
      "step": 7858
    },
    {
      "epoch": 0.43254994771313776,
      "grad_norm": 0.7738325595855713,
      "learning_rate": 8.910855621657228e-06,
      "loss": 0.7295,
      "step": 7859
    },
    {
      "epoch": 0.43260498651549345,
      "grad_norm": 0.7276971340179443,
      "learning_rate": 8.910585529661469e-06,
      "loss": 0.7982,
      "step": 7860
    },
    {
      "epoch": 0.4326600253178491,
      "grad_norm": 0.7655037641525269,
      "learning_rate": 8.910315408274796e-06,
      "loss": 0.8416,
      "step": 7861
    },
    {
      "epoch": 0.4327150641202047,
      "grad_norm": 0.7220892906188965,
      "learning_rate": 8.910045257499238e-06,
      "loss": 0.8002,
      "step": 7862
    },
    {
      "epoch": 0.4327701029225604,
      "grad_norm": 0.6255655884742737,
      "learning_rate": 8.90977507733683e-06,
      "loss": 0.6477,
      "step": 7863
    },
    {
      "epoch": 0.43282514172491604,
      "grad_norm": 0.649472713470459,
      "learning_rate": 8.909504867789594e-06,
      "loss": 0.6838,
      "step": 7864
    },
    {
      "epoch": 0.43288018052727173,
      "grad_norm": 0.6915234923362732,
      "learning_rate": 8.909234628859568e-06,
      "loss": 0.7146,
      "step": 7865
    },
    {
      "epoch": 0.43293521932962736,
      "grad_norm": 0.7120145559310913,
      "learning_rate": 8.908964360548783e-06,
      "loss": 0.7782,
      "step": 7866
    },
    {
      "epoch": 0.43299025813198305,
      "grad_norm": 0.8125410079956055,
      "learning_rate": 8.908694062859267e-06,
      "loss": 0.7514,
      "step": 7867
    },
    {
      "epoch": 0.4330452969343387,
      "grad_norm": 0.6821436882019043,
      "learning_rate": 8.908423735793053e-06,
      "loss": 0.8074,
      "step": 7868
    },
    {
      "epoch": 0.4331003357366944,
      "grad_norm": 0.8079590201377869,
      "learning_rate": 8.908153379352171e-06,
      "loss": 0.7932,
      "step": 7869
    },
    {
      "epoch": 0.43315537453905,
      "grad_norm": 0.676013708114624,
      "learning_rate": 8.907882993538655e-06,
      "loss": 0.6611,
      "step": 7870
    },
    {
      "epoch": 0.4332104133414057,
      "grad_norm": 0.706624448299408,
      "learning_rate": 8.907612578354537e-06,
      "loss": 0.8241,
      "step": 7871
    },
    {
      "epoch": 0.43326545214376133,
      "grad_norm": 0.6533300876617432,
      "learning_rate": 8.907342133801848e-06,
      "loss": 0.6969,
      "step": 7872
    },
    {
      "epoch": 0.433320490946117,
      "grad_norm": 0.6778282523155212,
      "learning_rate": 8.907071659882622e-06,
      "loss": 0.6877,
      "step": 7873
    },
    {
      "epoch": 0.43337552974847265,
      "grad_norm": 0.7068879008293152,
      "learning_rate": 8.906801156598892e-06,
      "loss": 0.7912,
      "step": 7874
    },
    {
      "epoch": 0.43343056855082834,
      "grad_norm": 0.6620263457298279,
      "learning_rate": 8.90653062395269e-06,
      "loss": 0.7317,
      "step": 7875
    },
    {
      "epoch": 0.433485607353184,
      "grad_norm": 0.7084807753562927,
      "learning_rate": 8.906260061946049e-06,
      "loss": 0.7268,
      "step": 7876
    },
    {
      "epoch": 0.43354064615553967,
      "grad_norm": 0.7899147272109985,
      "learning_rate": 8.905989470581003e-06,
      "loss": 0.8258,
      "step": 7877
    },
    {
      "epoch": 0.4335956849578953,
      "grad_norm": 0.6657128930091858,
      "learning_rate": 8.905718849859585e-06,
      "loss": 0.6564,
      "step": 7878
    },
    {
      "epoch": 0.433650723760251,
      "grad_norm": 0.8737723231315613,
      "learning_rate": 8.905448199783831e-06,
      "loss": 0.8646,
      "step": 7879
    },
    {
      "epoch": 0.4337057625626066,
      "grad_norm": 0.7517673969268799,
      "learning_rate": 8.905177520355775e-06,
      "loss": 0.7658,
      "step": 7880
    },
    {
      "epoch": 0.4337608013649623,
      "grad_norm": 0.6724270582199097,
      "learning_rate": 8.904906811577447e-06,
      "loss": 0.7509,
      "step": 7881
    },
    {
      "epoch": 0.43381584016731795,
      "grad_norm": 0.6490511894226074,
      "learning_rate": 8.904636073450885e-06,
      "loss": 0.7282,
      "step": 7882
    },
    {
      "epoch": 0.43387087896967363,
      "grad_norm": 0.73885178565979,
      "learning_rate": 8.904365305978126e-06,
      "loss": 0.7575,
      "step": 7883
    },
    {
      "epoch": 0.43392591777202927,
      "grad_norm": 0.6823462843894958,
      "learning_rate": 8.9040945091612e-06,
      "loss": 0.7566,
      "step": 7884
    },
    {
      "epoch": 0.43398095657438496,
      "grad_norm": 0.6705971956253052,
      "learning_rate": 8.903823683002146e-06,
      "loss": 0.7726,
      "step": 7885
    },
    {
      "epoch": 0.4340359953767406,
      "grad_norm": 0.6898428201675415,
      "learning_rate": 8.903552827502998e-06,
      "loss": 0.7545,
      "step": 7886
    },
    {
      "epoch": 0.4340910341790963,
      "grad_norm": 0.810357928276062,
      "learning_rate": 8.90328194266579e-06,
      "loss": 0.8883,
      "step": 7887
    },
    {
      "epoch": 0.4341460729814519,
      "grad_norm": 0.6505162119865417,
      "learning_rate": 8.903011028492563e-06,
      "loss": 0.7205,
      "step": 7888
    },
    {
      "epoch": 0.4342011117838076,
      "grad_norm": 0.8401693105697632,
      "learning_rate": 8.902740084985348e-06,
      "loss": 0.8105,
      "step": 7889
    },
    {
      "epoch": 0.43425615058616324,
      "grad_norm": 0.7151880860328674,
      "learning_rate": 8.902469112146183e-06,
      "loss": 0.7748,
      "step": 7890
    },
    {
      "epoch": 0.4343111893885189,
      "grad_norm": 0.7257007956504822,
      "learning_rate": 8.902198109977107e-06,
      "loss": 0.7818,
      "step": 7891
    },
    {
      "epoch": 0.43436622819087456,
      "grad_norm": 0.786691427230835,
      "learning_rate": 8.901927078480153e-06,
      "loss": 0.8527,
      "step": 7892
    },
    {
      "epoch": 0.43442126699323025,
      "grad_norm": 0.7420910596847534,
      "learning_rate": 8.901656017657358e-06,
      "loss": 0.7087,
      "step": 7893
    },
    {
      "epoch": 0.4344763057955859,
      "grad_norm": 0.6713958978652954,
      "learning_rate": 8.901384927510763e-06,
      "loss": 0.7366,
      "step": 7894
    },
    {
      "epoch": 0.43453134459794157,
      "grad_norm": 1.0276658535003662,
      "learning_rate": 8.901113808042402e-06,
      "loss": 0.7462,
      "step": 7895
    },
    {
      "epoch": 0.4345863834002972,
      "grad_norm": 0.7207444906234741,
      "learning_rate": 8.900842659254314e-06,
      "loss": 0.6777,
      "step": 7896
    },
    {
      "epoch": 0.4346414222026529,
      "grad_norm": 0.7581979036331177,
      "learning_rate": 8.900571481148538e-06,
      "loss": 0.8081,
      "step": 7897
    },
    {
      "epoch": 0.4346964610050085,
      "grad_norm": 0.9224075675010681,
      "learning_rate": 8.90030027372711e-06,
      "loss": 0.892,
      "step": 7898
    },
    {
      "epoch": 0.4347514998073642,
      "grad_norm": 0.6844260096549988,
      "learning_rate": 8.900029036992069e-06,
      "loss": 0.8063,
      "step": 7899
    },
    {
      "epoch": 0.43480653860971985,
      "grad_norm": 0.7008691430091858,
      "learning_rate": 8.899757770945453e-06,
      "loss": 0.6998,
      "step": 7900
    },
    {
      "epoch": 0.43486157741207554,
      "grad_norm": 0.7311949729919434,
      "learning_rate": 8.899486475589303e-06,
      "loss": 0.7724,
      "step": 7901
    },
    {
      "epoch": 0.4349166162144312,
      "grad_norm": 0.7441468238830566,
      "learning_rate": 8.899215150925656e-06,
      "loss": 0.7728,
      "step": 7902
    },
    {
      "epoch": 0.43497165501678686,
      "grad_norm": 0.7405179142951965,
      "learning_rate": 8.89894379695655e-06,
      "loss": 0.8267,
      "step": 7903
    },
    {
      "epoch": 0.4350266938191425,
      "grad_norm": 0.6967620253562927,
      "learning_rate": 8.898672413684029e-06,
      "loss": 0.7284,
      "step": 7904
    },
    {
      "epoch": 0.43508173262149813,
      "grad_norm": 0.8979219794273376,
      "learning_rate": 8.898401001110127e-06,
      "loss": 0.8267,
      "step": 7905
    },
    {
      "epoch": 0.4351367714238538,
      "grad_norm": 0.7905356884002686,
      "learning_rate": 8.898129559236888e-06,
      "loss": 0.8011,
      "step": 7906
    },
    {
      "epoch": 0.43519181022620945,
      "grad_norm": 0.6740859150886536,
      "learning_rate": 8.897858088066351e-06,
      "loss": 0.6597,
      "step": 7907
    },
    {
      "epoch": 0.43524684902856514,
      "grad_norm": 0.7451572418212891,
      "learning_rate": 8.897586587600555e-06,
      "loss": 0.7466,
      "step": 7908
    },
    {
      "epoch": 0.4353018878309208,
      "grad_norm": 0.7726565003395081,
      "learning_rate": 8.897315057841542e-06,
      "loss": 0.7873,
      "step": 7909
    },
    {
      "epoch": 0.43535692663327646,
      "grad_norm": 0.8348171710968018,
      "learning_rate": 8.897043498791354e-06,
      "loss": 0.7583,
      "step": 7910
    },
    {
      "epoch": 0.4354119654356321,
      "grad_norm": 0.6714087724685669,
      "learning_rate": 8.896771910452027e-06,
      "loss": 0.7909,
      "step": 7911
    },
    {
      "epoch": 0.4354670042379878,
      "grad_norm": 0.7397969365119934,
      "learning_rate": 8.896500292825607e-06,
      "loss": 0.7734,
      "step": 7912
    },
    {
      "epoch": 0.4355220430403434,
      "grad_norm": 0.6806391477584839,
      "learning_rate": 8.896228645914133e-06,
      "loss": 0.7898,
      "step": 7913
    },
    {
      "epoch": 0.4355770818426991,
      "grad_norm": 0.7135224342346191,
      "learning_rate": 8.89595696971965e-06,
      "loss": 0.7453,
      "step": 7914
    },
    {
      "epoch": 0.43563212064505474,
      "grad_norm": 0.8275992274284363,
      "learning_rate": 8.895685264244195e-06,
      "loss": 0.7326,
      "step": 7915
    },
    {
      "epoch": 0.43568715944741043,
      "grad_norm": 0.7254159450531006,
      "learning_rate": 8.895413529489813e-06,
      "loss": 0.7523,
      "step": 7916
    },
    {
      "epoch": 0.43574219824976607,
      "grad_norm": 0.8060647249221802,
      "learning_rate": 8.895141765458546e-06,
      "loss": 0.7878,
      "step": 7917
    },
    {
      "epoch": 0.43579723705212176,
      "grad_norm": 0.7007316946983337,
      "learning_rate": 8.894869972152435e-06,
      "loss": 0.7837,
      "step": 7918
    },
    {
      "epoch": 0.4358522758544774,
      "grad_norm": 0.6874841451644897,
      "learning_rate": 8.894598149573524e-06,
      "loss": 0.7773,
      "step": 7919
    },
    {
      "epoch": 0.4359073146568331,
      "grad_norm": 0.7557696104049683,
      "learning_rate": 8.894326297723856e-06,
      "loss": 0.6905,
      "step": 7920
    },
    {
      "epoch": 0.4359623534591887,
      "grad_norm": 0.7589512467384338,
      "learning_rate": 8.894054416605475e-06,
      "loss": 0.8292,
      "step": 7921
    },
    {
      "epoch": 0.4360173922615444,
      "grad_norm": 0.9062818884849548,
      "learning_rate": 8.893782506220424e-06,
      "loss": 0.9149,
      "step": 7922
    },
    {
      "epoch": 0.43607243106390003,
      "grad_norm": 0.7553420662879944,
      "learning_rate": 8.893510566570744e-06,
      "loss": 0.7256,
      "step": 7923
    },
    {
      "epoch": 0.4361274698662557,
      "grad_norm": 0.7130489349365234,
      "learning_rate": 8.89323859765848e-06,
      "loss": 0.7375,
      "step": 7924
    },
    {
      "epoch": 0.43618250866861136,
      "grad_norm": 0.6234793066978455,
      "learning_rate": 8.89296659948568e-06,
      "loss": 0.716,
      "step": 7925
    },
    {
      "epoch": 0.43623754747096705,
      "grad_norm": 0.7527539134025574,
      "learning_rate": 8.892694572054383e-06,
      "loss": 0.7884,
      "step": 7926
    },
    {
      "epoch": 0.4362925862733227,
      "grad_norm": 0.7677647471427917,
      "learning_rate": 8.892422515366636e-06,
      "loss": 0.7136,
      "step": 7927
    },
    {
      "epoch": 0.43634762507567837,
      "grad_norm": 0.7212143540382385,
      "learning_rate": 8.892150429424484e-06,
      "loss": 0.8113,
      "step": 7928
    },
    {
      "epoch": 0.436402663878034,
      "grad_norm": 0.6735568046569824,
      "learning_rate": 8.89187831422997e-06,
      "loss": 0.6472,
      "step": 7929
    },
    {
      "epoch": 0.4364577026803897,
      "grad_norm": 0.7120702862739563,
      "learning_rate": 8.891606169785141e-06,
      "loss": 0.8032,
      "step": 7930
    },
    {
      "epoch": 0.4365127414827453,
      "grad_norm": 0.679499089717865,
      "learning_rate": 8.891333996092041e-06,
      "loss": 0.7366,
      "step": 7931
    },
    {
      "epoch": 0.436567780285101,
      "grad_norm": 0.7774114012718201,
      "learning_rate": 8.891061793152718e-06,
      "loss": 0.7917,
      "step": 7932
    },
    {
      "epoch": 0.43662281908745665,
      "grad_norm": 0.6951174139976501,
      "learning_rate": 8.890789560969216e-06,
      "loss": 0.7518,
      "step": 7933
    },
    {
      "epoch": 0.43667785788981234,
      "grad_norm": 0.7645227909088135,
      "learning_rate": 8.89051729954358e-06,
      "loss": 0.7787,
      "step": 7934
    },
    {
      "epoch": 0.43673289669216797,
      "grad_norm": 0.7127084732055664,
      "learning_rate": 8.890245008877857e-06,
      "loss": 0.8137,
      "step": 7935
    },
    {
      "epoch": 0.43678793549452366,
      "grad_norm": 0.7541413903236389,
      "learning_rate": 8.889972688974095e-06,
      "loss": 0.776,
      "step": 7936
    },
    {
      "epoch": 0.4368429742968793,
      "grad_norm": 0.690963625907898,
      "learning_rate": 8.889700339834339e-06,
      "loss": 0.7691,
      "step": 7937
    },
    {
      "epoch": 0.436898013099235,
      "grad_norm": 0.750221848487854,
      "learning_rate": 8.889427961460636e-06,
      "loss": 0.7831,
      "step": 7938
    },
    {
      "epoch": 0.4369530519015906,
      "grad_norm": 0.7255545854568481,
      "learning_rate": 8.889155553855035e-06,
      "loss": 0.7831,
      "step": 7939
    },
    {
      "epoch": 0.4370080907039463,
      "grad_norm": 0.7187026143074036,
      "learning_rate": 8.88888311701958e-06,
      "loss": 0.792,
      "step": 7940
    },
    {
      "epoch": 0.43706312950630194,
      "grad_norm": 0.8313350081443787,
      "learning_rate": 8.888610650956322e-06,
      "loss": 0.706,
      "step": 7941
    },
    {
      "epoch": 0.43711816830865763,
      "grad_norm": 0.8083454370498657,
      "learning_rate": 8.888338155667307e-06,
      "loss": 0.7857,
      "step": 7942
    },
    {
      "epoch": 0.43717320711101326,
      "grad_norm": 0.8200840353965759,
      "learning_rate": 8.888065631154583e-06,
      "loss": 0.8601,
      "step": 7943
    },
    {
      "epoch": 0.43722824591336895,
      "grad_norm": 0.7503816485404968,
      "learning_rate": 8.887793077420198e-06,
      "loss": 0.7744,
      "step": 7944
    },
    {
      "epoch": 0.4372832847157246,
      "grad_norm": 0.7466493248939514,
      "learning_rate": 8.887520494466202e-06,
      "loss": 0.7818,
      "step": 7945
    },
    {
      "epoch": 0.4373383235180803,
      "grad_norm": 0.728118360042572,
      "learning_rate": 8.887247882294641e-06,
      "loss": 0.7157,
      "step": 7946
    },
    {
      "epoch": 0.4373933623204359,
      "grad_norm": 0.9199670553207397,
      "learning_rate": 8.886975240907568e-06,
      "loss": 0.8283,
      "step": 7947
    },
    {
      "epoch": 0.43744840112279154,
      "grad_norm": 0.735584557056427,
      "learning_rate": 8.886702570307027e-06,
      "loss": 0.6588,
      "step": 7948
    },
    {
      "epoch": 0.43750343992514723,
      "grad_norm": 0.8619036674499512,
      "learning_rate": 8.886429870495072e-06,
      "loss": 0.7269,
      "step": 7949
    },
    {
      "epoch": 0.43755847872750286,
      "grad_norm": 0.7304830551147461,
      "learning_rate": 8.886157141473747e-06,
      "loss": 0.6725,
      "step": 7950
    },
    {
      "epoch": 0.43761351752985855,
      "grad_norm": 0.7669086456298828,
      "learning_rate": 8.885884383245109e-06,
      "loss": 0.6957,
      "step": 7951
    },
    {
      "epoch": 0.4376685563322142,
      "grad_norm": 0.7558299899101257,
      "learning_rate": 8.885611595811203e-06,
      "loss": 0.8159,
      "step": 7952
    },
    {
      "epoch": 0.4377235951345699,
      "grad_norm": 0.7661786079406738,
      "learning_rate": 8.88533877917408e-06,
      "loss": 0.764,
      "step": 7953
    },
    {
      "epoch": 0.4377786339369255,
      "grad_norm": 0.7461101412773132,
      "learning_rate": 8.88506593333579e-06,
      "loss": 0.7544,
      "step": 7954
    },
    {
      "epoch": 0.4378336727392812,
      "grad_norm": 0.7989180088043213,
      "learning_rate": 8.884793058298387e-06,
      "loss": 0.6913,
      "step": 7955
    },
    {
      "epoch": 0.43788871154163683,
      "grad_norm": 0.7964022755622864,
      "learning_rate": 8.884520154063917e-06,
      "loss": 0.7339,
      "step": 7956
    },
    {
      "epoch": 0.4379437503439925,
      "grad_norm": 0.7278034687042236,
      "learning_rate": 8.884247220634433e-06,
      "loss": 0.8477,
      "step": 7957
    },
    {
      "epoch": 0.43799878914634816,
      "grad_norm": 0.7294753789901733,
      "learning_rate": 8.883974258011988e-06,
      "loss": 0.8412,
      "step": 7958
    },
    {
      "epoch": 0.43805382794870384,
      "grad_norm": 0.665734589099884,
      "learning_rate": 8.88370126619863e-06,
      "loss": 0.7838,
      "step": 7959
    },
    {
      "epoch": 0.4381088667510595,
      "grad_norm": 0.6984216570854187,
      "learning_rate": 8.883428245196414e-06,
      "loss": 0.7657,
      "step": 7960
    },
    {
      "epoch": 0.43816390555341517,
      "grad_norm": 0.8048402070999146,
      "learning_rate": 8.883155195007393e-06,
      "loss": 0.7553,
      "step": 7961
    },
    {
      "epoch": 0.4382189443557708,
      "grad_norm": 0.7145794630050659,
      "learning_rate": 8.882882115633616e-06,
      "loss": 0.6583,
      "step": 7962
    },
    {
      "epoch": 0.4382739831581265,
      "grad_norm": 0.7073546648025513,
      "learning_rate": 8.882609007077135e-06,
      "loss": 0.7869,
      "step": 7963
    },
    {
      "epoch": 0.4383290219604821,
      "grad_norm": 0.8300859928131104,
      "learning_rate": 8.882335869340004e-06,
      "loss": 0.773,
      "step": 7964
    },
    {
      "epoch": 0.4383840607628378,
      "grad_norm": 0.8343188762664795,
      "learning_rate": 8.882062702424276e-06,
      "loss": 0.6743,
      "step": 7965
    },
    {
      "epoch": 0.43843909956519345,
      "grad_norm": 0.7106530666351318,
      "learning_rate": 8.881789506332007e-06,
      "loss": 0.7414,
      "step": 7966
    },
    {
      "epoch": 0.43849413836754914,
      "grad_norm": 0.7015630602836609,
      "learning_rate": 8.881516281065244e-06,
      "loss": 0.7434,
      "step": 7967
    },
    {
      "epoch": 0.43854917716990477,
      "grad_norm": 0.8106673955917358,
      "learning_rate": 8.881243026626044e-06,
      "loss": 0.7741,
      "step": 7968
    },
    {
      "epoch": 0.43860421597226046,
      "grad_norm": 0.8181495070457458,
      "learning_rate": 8.88096974301646e-06,
      "loss": 0.8046,
      "step": 7969
    },
    {
      "epoch": 0.4386592547746161,
      "grad_norm": 0.7767857313156128,
      "learning_rate": 8.880696430238546e-06,
      "loss": 0.8586,
      "step": 7970
    },
    {
      "epoch": 0.4387142935769718,
      "grad_norm": 0.7257522940635681,
      "learning_rate": 8.880423088294359e-06,
      "loss": 0.7799,
      "step": 7971
    },
    {
      "epoch": 0.4387693323793274,
      "grad_norm": 0.6896021366119385,
      "learning_rate": 8.880149717185948e-06,
      "loss": 0.8178,
      "step": 7972
    },
    {
      "epoch": 0.4388243711816831,
      "grad_norm": 0.7646406292915344,
      "learning_rate": 8.879876316915372e-06,
      "loss": 0.8754,
      "step": 7973
    },
    {
      "epoch": 0.43887940998403874,
      "grad_norm": 0.8043848872184753,
      "learning_rate": 8.879602887484684e-06,
      "loss": 0.8562,
      "step": 7974
    },
    {
      "epoch": 0.4389344487863944,
      "grad_norm": 0.6727305054664612,
      "learning_rate": 8.879329428895937e-06,
      "loss": 0.6168,
      "step": 7975
    },
    {
      "epoch": 0.43898948758875006,
      "grad_norm": 0.7634731531143188,
      "learning_rate": 8.87905594115119e-06,
      "loss": 0.857,
      "step": 7976
    },
    {
      "epoch": 0.43904452639110575,
      "grad_norm": 0.6544492244720459,
      "learning_rate": 8.878782424252497e-06,
      "loss": 0.6302,
      "step": 7977
    },
    {
      "epoch": 0.4390995651934614,
      "grad_norm": 0.8126636743545532,
      "learning_rate": 8.878508878201915e-06,
      "loss": 0.7823,
      "step": 7978
    },
    {
      "epoch": 0.43915460399581707,
      "grad_norm": 0.7235779166221619,
      "learning_rate": 8.878235303001497e-06,
      "loss": 0.7527,
      "step": 7979
    },
    {
      "epoch": 0.4392096427981727,
      "grad_norm": 0.6961055397987366,
      "learning_rate": 8.8779616986533e-06,
      "loss": 0.7383,
      "step": 7980
    },
    {
      "epoch": 0.4392646816005284,
      "grad_norm": 0.7684490084648132,
      "learning_rate": 8.877688065159382e-06,
      "loss": 0.8009,
      "step": 7981
    },
    {
      "epoch": 0.43931972040288403,
      "grad_norm": 0.7897803783416748,
      "learning_rate": 8.877414402521797e-06,
      "loss": 0.7561,
      "step": 7982
    },
    {
      "epoch": 0.4393747592052397,
      "grad_norm": 0.7877688407897949,
      "learning_rate": 8.877140710742606e-06,
      "loss": 0.7949,
      "step": 7983
    },
    {
      "epoch": 0.43942979800759535,
      "grad_norm": 0.8341611623764038,
      "learning_rate": 8.876866989823862e-06,
      "loss": 0.7585,
      "step": 7984
    },
    {
      "epoch": 0.43948483680995104,
      "grad_norm": 0.7663636207580566,
      "learning_rate": 8.876593239767622e-06,
      "loss": 0.771,
      "step": 7985
    },
    {
      "epoch": 0.4395398756123067,
      "grad_norm": 0.6824129223823547,
      "learning_rate": 8.876319460575946e-06,
      "loss": 0.7852,
      "step": 7986
    },
    {
      "epoch": 0.43959491441466236,
      "grad_norm": 0.6533854007720947,
      "learning_rate": 8.876045652250891e-06,
      "loss": 0.723,
      "step": 7987
    },
    {
      "epoch": 0.439649953217018,
      "grad_norm": 0.7174259424209595,
      "learning_rate": 8.875771814794515e-06,
      "loss": 0.749,
      "step": 7988
    },
    {
      "epoch": 0.4397049920193737,
      "grad_norm": 0.8585928678512573,
      "learning_rate": 8.875497948208875e-06,
      "loss": 0.6727,
      "step": 7989
    },
    {
      "epoch": 0.4397600308217293,
      "grad_norm": 0.7558062672615051,
      "learning_rate": 8.875224052496029e-06,
      "loss": 0.7929,
      "step": 7990
    },
    {
      "epoch": 0.43981506962408495,
      "grad_norm": 0.7063853144645691,
      "learning_rate": 8.874950127658037e-06,
      "loss": 0.7397,
      "step": 7991
    },
    {
      "epoch": 0.43987010842644064,
      "grad_norm": 0.7165526747703552,
      "learning_rate": 8.874676173696956e-06,
      "loss": 0.7678,
      "step": 7992
    },
    {
      "epoch": 0.4399251472287963,
      "grad_norm": 0.7657830715179443,
      "learning_rate": 8.874402190614847e-06,
      "loss": 0.8318,
      "step": 7993
    },
    {
      "epoch": 0.43998018603115197,
      "grad_norm": 0.7776834964752197,
      "learning_rate": 8.874128178413769e-06,
      "loss": 0.8589,
      "step": 7994
    },
    {
      "epoch": 0.4400352248335076,
      "grad_norm": 0.6805633306503296,
      "learning_rate": 8.873854137095778e-06,
      "loss": 0.7009,
      "step": 7995
    },
    {
      "epoch": 0.4400902636358633,
      "grad_norm": 0.6962490677833557,
      "learning_rate": 8.87358006666294e-06,
      "loss": 0.7896,
      "step": 7996
    },
    {
      "epoch": 0.4401453024382189,
      "grad_norm": 0.611610472202301,
      "learning_rate": 8.873305967117307e-06,
      "loss": 0.5993,
      "step": 7997
    },
    {
      "epoch": 0.4402003412405746,
      "grad_norm": 0.7442964911460876,
      "learning_rate": 8.873031838460946e-06,
      "loss": 0.8277,
      "step": 7998
    },
    {
      "epoch": 0.44025538004293024,
      "grad_norm": 0.6858734488487244,
      "learning_rate": 8.872757680695914e-06,
      "loss": 0.8064,
      "step": 7999
    },
    {
      "epoch": 0.44031041884528593,
      "grad_norm": 0.6654849052429199,
      "learning_rate": 8.872483493824273e-06,
      "loss": 0.7408,
      "step": 8000
    },
    {
      "epoch": 0.44036545764764157,
      "grad_norm": 0.8241575956344604,
      "learning_rate": 8.87220927784808e-06,
      "loss": 0.8819,
      "step": 8001
    },
    {
      "epoch": 0.44042049644999726,
      "grad_norm": 0.7078573107719421,
      "learning_rate": 8.8719350327694e-06,
      "loss": 0.7709,
      "step": 8002
    },
    {
      "epoch": 0.4404755352523529,
      "grad_norm": 0.7369210720062256,
      "learning_rate": 8.871660758590292e-06,
      "loss": 0.7867,
      "step": 8003
    },
    {
      "epoch": 0.4405305740547086,
      "grad_norm": 0.7206673622131348,
      "learning_rate": 8.87138645531282e-06,
      "loss": 0.8697,
      "step": 8004
    },
    {
      "epoch": 0.4405856128570642,
      "grad_norm": 0.8370183706283569,
      "learning_rate": 8.871112122939041e-06,
      "loss": 0.7201,
      "step": 8005
    },
    {
      "epoch": 0.4406406516594199,
      "grad_norm": 0.8015196323394775,
      "learning_rate": 8.870837761471023e-06,
      "loss": 0.774,
      "step": 8006
    },
    {
      "epoch": 0.44069569046177554,
      "grad_norm": 0.730185329914093,
      "learning_rate": 8.870563370910821e-06,
      "loss": 0.7153,
      "step": 8007
    },
    {
      "epoch": 0.4407507292641312,
      "grad_norm": 0.6719930768013,
      "learning_rate": 8.870288951260503e-06,
      "loss": 0.7949,
      "step": 8008
    },
    {
      "epoch": 0.44080576806648686,
      "grad_norm": 0.7614291906356812,
      "learning_rate": 8.870014502522128e-06,
      "loss": 0.7143,
      "step": 8009
    },
    {
      "epoch": 0.44086080686884255,
      "grad_norm": 0.7438056468963623,
      "learning_rate": 8.86974002469776e-06,
      "loss": 0.6859,
      "step": 8010
    },
    {
      "epoch": 0.4409158456711982,
      "grad_norm": 0.759903073310852,
      "learning_rate": 8.869465517789463e-06,
      "loss": 0.8095,
      "step": 8011
    },
    {
      "epoch": 0.44097088447355387,
      "grad_norm": 0.7622823119163513,
      "learning_rate": 8.869190981799298e-06,
      "loss": 0.786,
      "step": 8012
    },
    {
      "epoch": 0.4410259232759095,
      "grad_norm": 0.677003800868988,
      "learning_rate": 8.86891641672933e-06,
      "loss": 0.7074,
      "step": 8013
    },
    {
      "epoch": 0.4410809620782652,
      "grad_norm": 0.9258451461791992,
      "learning_rate": 8.86864182258162e-06,
      "loss": 0.7218,
      "step": 8014
    },
    {
      "epoch": 0.4411360008806208,
      "grad_norm": 0.7027828693389893,
      "learning_rate": 8.868367199358236e-06,
      "loss": 0.7654,
      "step": 8015
    },
    {
      "epoch": 0.4411910396829765,
      "grad_norm": 0.8279967308044434,
      "learning_rate": 8.868092547061239e-06,
      "loss": 0.8969,
      "step": 8016
    },
    {
      "epoch": 0.44124607848533215,
      "grad_norm": 0.7366079688072205,
      "learning_rate": 8.867817865692693e-06,
      "loss": 0.8421,
      "step": 8017
    },
    {
      "epoch": 0.44130111728768784,
      "grad_norm": 0.7548787593841553,
      "learning_rate": 8.867543155254665e-06,
      "loss": 0.79,
      "step": 8018
    },
    {
      "epoch": 0.44135615609004347,
      "grad_norm": 0.7558487057685852,
      "learning_rate": 8.867268415749215e-06,
      "loss": 0.8461,
      "step": 8019
    },
    {
      "epoch": 0.44141119489239916,
      "grad_norm": 0.6413403153419495,
      "learning_rate": 8.866993647178413e-06,
      "loss": 0.6811,
      "step": 8020
    },
    {
      "epoch": 0.4414662336947548,
      "grad_norm": 0.9251089692115784,
      "learning_rate": 8.86671884954432e-06,
      "loss": 0.868,
      "step": 8021
    },
    {
      "epoch": 0.4415212724971105,
      "grad_norm": 0.7920099496841431,
      "learning_rate": 8.866444022849006e-06,
      "loss": 0.8131,
      "step": 8022
    },
    {
      "epoch": 0.4415763112994661,
      "grad_norm": 0.8738380670547485,
      "learning_rate": 8.866169167094532e-06,
      "loss": 0.857,
      "step": 8023
    },
    {
      "epoch": 0.4416313501018218,
      "grad_norm": 0.7181336283683777,
      "learning_rate": 8.865894282282965e-06,
      "loss": 0.7869,
      "step": 8024
    },
    {
      "epoch": 0.44168638890417744,
      "grad_norm": 0.8003776669502258,
      "learning_rate": 8.865619368416373e-06,
      "loss": 0.8874,
      "step": 8025
    },
    {
      "epoch": 0.44174142770653313,
      "grad_norm": 0.7186623215675354,
      "learning_rate": 8.86534442549682e-06,
      "loss": 0.7931,
      "step": 8026
    },
    {
      "epoch": 0.44179646650888876,
      "grad_norm": 0.7006831765174866,
      "learning_rate": 8.865069453526371e-06,
      "loss": 0.7046,
      "step": 8027
    },
    {
      "epoch": 0.44185150531124445,
      "grad_norm": 0.7394786477088928,
      "learning_rate": 8.864794452507097e-06,
      "loss": 0.685,
      "step": 8028
    },
    {
      "epoch": 0.4419065441136001,
      "grad_norm": 0.7512097358703613,
      "learning_rate": 8.864519422441062e-06,
      "loss": 0.8047,
      "step": 8029
    },
    {
      "epoch": 0.4419615829159558,
      "grad_norm": 0.6866902709007263,
      "learning_rate": 8.864244363330333e-06,
      "loss": 0.7099,
      "step": 8030
    },
    {
      "epoch": 0.4420166217183114,
      "grad_norm": 0.7316723465919495,
      "learning_rate": 8.863969275176978e-06,
      "loss": 0.7767,
      "step": 8031
    },
    {
      "epoch": 0.4420716605206671,
      "grad_norm": 0.7103593349456787,
      "learning_rate": 8.863694157983064e-06,
      "loss": 0.7832,
      "step": 8032
    },
    {
      "epoch": 0.44212669932302273,
      "grad_norm": 0.6922749876976013,
      "learning_rate": 8.863419011750659e-06,
      "loss": 0.7833,
      "step": 8033
    },
    {
      "epoch": 0.44218173812537837,
      "grad_norm": 0.7989425659179688,
      "learning_rate": 8.863143836481831e-06,
      "loss": 0.8651,
      "step": 8034
    },
    {
      "epoch": 0.44223677692773405,
      "grad_norm": 0.6765440702438354,
      "learning_rate": 8.862868632178648e-06,
      "loss": 0.7858,
      "step": 8035
    },
    {
      "epoch": 0.4422918157300897,
      "grad_norm": 0.670767068862915,
      "learning_rate": 8.862593398843178e-06,
      "loss": 0.6789,
      "step": 8036
    },
    {
      "epoch": 0.4423468545324454,
      "grad_norm": 0.7556853294372559,
      "learning_rate": 8.86231813647749e-06,
      "loss": 0.8036,
      "step": 8037
    },
    {
      "epoch": 0.442401893334801,
      "grad_norm": 0.788690984249115,
      "learning_rate": 8.862042845083654e-06,
      "loss": 0.8355,
      "step": 8038
    },
    {
      "epoch": 0.4424569321371567,
      "grad_norm": 0.8439056873321533,
      "learning_rate": 8.861767524663736e-06,
      "loss": 0.7327,
      "step": 8039
    },
    {
      "epoch": 0.44251197093951233,
      "grad_norm": 0.7101821899414062,
      "learning_rate": 8.861492175219808e-06,
      "loss": 0.8303,
      "step": 8040
    },
    {
      "epoch": 0.442567009741868,
      "grad_norm": 0.741680383682251,
      "learning_rate": 8.861216796753937e-06,
      "loss": 0.7377,
      "step": 8041
    },
    {
      "epoch": 0.44262204854422366,
      "grad_norm": 0.7588099837303162,
      "learning_rate": 8.860941389268196e-06,
      "loss": 0.8217,
      "step": 8042
    },
    {
      "epoch": 0.44267708734657935,
      "grad_norm": 0.7654829025268555,
      "learning_rate": 8.860665952764654e-06,
      "loss": 0.8416,
      "step": 8043
    },
    {
      "epoch": 0.442732126148935,
      "grad_norm": 0.7025987505912781,
      "learning_rate": 8.860390487245378e-06,
      "loss": 0.7312,
      "step": 8044
    },
    {
      "epoch": 0.44278716495129067,
      "grad_norm": 0.7206251621246338,
      "learning_rate": 8.860114992712441e-06,
      "loss": 0.7522,
      "step": 8045
    },
    {
      "epoch": 0.4428422037536463,
      "grad_norm": 0.7041749954223633,
      "learning_rate": 8.859839469167912e-06,
      "loss": 0.746,
      "step": 8046
    },
    {
      "epoch": 0.442897242556002,
      "grad_norm": 0.6941862106323242,
      "learning_rate": 8.859563916613864e-06,
      "loss": 0.7692,
      "step": 8047
    },
    {
      "epoch": 0.4429522813583576,
      "grad_norm": 0.6897740364074707,
      "learning_rate": 8.859288335052367e-06,
      "loss": 0.7963,
      "step": 8048
    },
    {
      "epoch": 0.4430073201607133,
      "grad_norm": 0.6744545698165894,
      "learning_rate": 8.859012724485492e-06,
      "loss": 0.7647,
      "step": 8049
    },
    {
      "epoch": 0.44306235896306895,
      "grad_norm": 0.7899364829063416,
      "learning_rate": 8.858737084915309e-06,
      "loss": 0.8373,
      "step": 8050
    },
    {
      "epoch": 0.44311739776542464,
      "grad_norm": 0.806016743183136,
      "learning_rate": 8.85846141634389e-06,
      "loss": 0.7871,
      "step": 8051
    },
    {
      "epoch": 0.44317243656778027,
      "grad_norm": 0.7444993257522583,
      "learning_rate": 8.85818571877331e-06,
      "loss": 0.8099,
      "step": 8052
    },
    {
      "epoch": 0.44322747537013596,
      "grad_norm": 0.772735059261322,
      "learning_rate": 8.85790999220564e-06,
      "loss": 0.7113,
      "step": 8053
    },
    {
      "epoch": 0.4432825141724916,
      "grad_norm": 0.7743984460830688,
      "learning_rate": 8.85763423664295e-06,
      "loss": 0.8935,
      "step": 8054
    },
    {
      "epoch": 0.4433375529748473,
      "grad_norm": 0.6751214265823364,
      "learning_rate": 8.857358452087313e-06,
      "loss": 0.6769,
      "step": 8055
    },
    {
      "epoch": 0.4433925917772029,
      "grad_norm": 0.6921005845069885,
      "learning_rate": 8.857082638540803e-06,
      "loss": 0.7071,
      "step": 8056
    },
    {
      "epoch": 0.4434476305795586,
      "grad_norm": 0.7884092330932617,
      "learning_rate": 8.856806796005491e-06,
      "loss": 0.7919,
      "step": 8057
    },
    {
      "epoch": 0.44350266938191424,
      "grad_norm": 0.6522679924964905,
      "learning_rate": 8.856530924483452e-06,
      "loss": 0.7449,
      "step": 8058
    },
    {
      "epoch": 0.4435577081842699,
      "grad_norm": 0.7172590494155884,
      "learning_rate": 8.85625502397676e-06,
      "loss": 0.7306,
      "step": 8059
    },
    {
      "epoch": 0.44361274698662556,
      "grad_norm": 0.698658287525177,
      "learning_rate": 8.855979094487488e-06,
      "loss": 0.803,
      "step": 8060
    },
    {
      "epoch": 0.44366778578898125,
      "grad_norm": 0.685589075088501,
      "learning_rate": 8.855703136017708e-06,
      "loss": 0.763,
      "step": 8061
    },
    {
      "epoch": 0.4437228245913369,
      "grad_norm": 0.8259774446487427,
      "learning_rate": 8.855427148569495e-06,
      "loss": 0.811,
      "step": 8062
    },
    {
      "epoch": 0.4437778633936926,
      "grad_norm": 0.6976660490036011,
      "learning_rate": 8.855151132144926e-06,
      "loss": 0.7345,
      "step": 8063
    },
    {
      "epoch": 0.4438329021960482,
      "grad_norm": 0.7696738243103027,
      "learning_rate": 8.854875086746071e-06,
      "loss": 0.823,
      "step": 8064
    },
    {
      "epoch": 0.4438879409984039,
      "grad_norm": 0.6627930998802185,
      "learning_rate": 8.854599012375006e-06,
      "loss": 0.7455,
      "step": 8065
    },
    {
      "epoch": 0.44394297980075953,
      "grad_norm": 0.7492700815200806,
      "learning_rate": 8.854322909033809e-06,
      "loss": 0.8195,
      "step": 8066
    },
    {
      "epoch": 0.4439980186031152,
      "grad_norm": 0.8335888981819153,
      "learning_rate": 8.85404677672455e-06,
      "loss": 0.7683,
      "step": 8067
    },
    {
      "epoch": 0.44405305740547085,
      "grad_norm": 0.7448242902755737,
      "learning_rate": 8.853770615449309e-06,
      "loss": 0.8352,
      "step": 8068
    },
    {
      "epoch": 0.44410809620782654,
      "grad_norm": 0.700616180896759,
      "learning_rate": 8.853494425210158e-06,
      "loss": 0.7892,
      "step": 8069
    },
    {
      "epoch": 0.4441631350101822,
      "grad_norm": 0.6959284543991089,
      "learning_rate": 8.853218206009176e-06,
      "loss": 0.6944,
      "step": 8070
    },
    {
      "epoch": 0.44421817381253786,
      "grad_norm": 0.7507375478744507,
      "learning_rate": 8.852941957848438e-06,
      "loss": 0.8921,
      "step": 8071
    },
    {
      "epoch": 0.4442732126148935,
      "grad_norm": 0.7843918204307556,
      "learning_rate": 8.852665680730019e-06,
      "loss": 0.816,
      "step": 8072
    },
    {
      "epoch": 0.4443282514172492,
      "grad_norm": 0.8702702522277832,
      "learning_rate": 8.852389374655995e-06,
      "loss": 0.8191,
      "step": 8073
    },
    {
      "epoch": 0.4443832902196048,
      "grad_norm": 0.6784317493438721,
      "learning_rate": 8.852113039628445e-06,
      "loss": 0.7726,
      "step": 8074
    },
    {
      "epoch": 0.4444383290219605,
      "grad_norm": 0.724530041217804,
      "learning_rate": 8.851836675649443e-06,
      "loss": 0.8214,
      "step": 8075
    },
    {
      "epoch": 0.44449336782431614,
      "grad_norm": 0.9814287424087524,
      "learning_rate": 8.851560282721067e-06,
      "loss": 0.8368,
      "step": 8076
    },
    {
      "epoch": 0.4445484066266718,
      "grad_norm": 0.6606815457344055,
      "learning_rate": 8.851283860845398e-06,
      "loss": 0.7772,
      "step": 8077
    },
    {
      "epoch": 0.44460344542902747,
      "grad_norm": 0.6910951137542725,
      "learning_rate": 8.851007410024507e-06,
      "loss": 0.7007,
      "step": 8078
    },
    {
      "epoch": 0.4446584842313831,
      "grad_norm": 0.6764300465583801,
      "learning_rate": 8.850730930260479e-06,
      "loss": 0.7265,
      "step": 8079
    },
    {
      "epoch": 0.4447135230337388,
      "grad_norm": 0.669622004032135,
      "learning_rate": 8.850454421555386e-06,
      "loss": 0.7551,
      "step": 8080
    },
    {
      "epoch": 0.4447685618360944,
      "grad_norm": 0.7068240642547607,
      "learning_rate": 8.850177883911307e-06,
      "loss": 0.8358,
      "step": 8081
    },
    {
      "epoch": 0.4448236006384501,
      "grad_norm": 0.7100360989570618,
      "learning_rate": 8.849901317330324e-06,
      "loss": 0.7074,
      "step": 8082
    },
    {
      "epoch": 0.44487863944080575,
      "grad_norm": 0.7510328888893127,
      "learning_rate": 8.849624721814511e-06,
      "loss": 0.6654,
      "step": 8083
    },
    {
      "epoch": 0.44493367824316143,
      "grad_norm": 0.8106432557106018,
      "learning_rate": 8.849348097365951e-06,
      "loss": 0.6944,
      "step": 8084
    },
    {
      "epoch": 0.44498871704551707,
      "grad_norm": 0.6852346062660217,
      "learning_rate": 8.84907144398672e-06,
      "loss": 0.7203,
      "step": 8085
    },
    {
      "epoch": 0.44504375584787276,
      "grad_norm": 0.8495593667030334,
      "learning_rate": 8.848794761678898e-06,
      "loss": 0.7918,
      "step": 8086
    },
    {
      "epoch": 0.4450987946502284,
      "grad_norm": 0.7110981941223145,
      "learning_rate": 8.848518050444565e-06,
      "loss": 0.8176,
      "step": 8087
    },
    {
      "epoch": 0.4451538334525841,
      "grad_norm": 0.7740922570228577,
      "learning_rate": 8.8482413102858e-06,
      "loss": 0.7573,
      "step": 8088
    },
    {
      "epoch": 0.4452088722549397,
      "grad_norm": 0.9645134806632996,
      "learning_rate": 8.847964541204685e-06,
      "loss": 0.7842,
      "step": 8089
    },
    {
      "epoch": 0.4452639110572954,
      "grad_norm": 0.767621636390686,
      "learning_rate": 8.847687743203299e-06,
      "loss": 0.8182,
      "step": 8090
    },
    {
      "epoch": 0.44531894985965104,
      "grad_norm": 0.6842975616455078,
      "learning_rate": 8.84741091628372e-06,
      "loss": 0.7795,
      "step": 8091
    },
    {
      "epoch": 0.4453739886620067,
      "grad_norm": 0.768644392490387,
      "learning_rate": 8.847134060448032e-06,
      "loss": 0.7363,
      "step": 8092
    },
    {
      "epoch": 0.44542902746436236,
      "grad_norm": 0.6813824772834778,
      "learning_rate": 8.846857175698314e-06,
      "loss": 0.7601,
      "step": 8093
    },
    {
      "epoch": 0.44548406626671805,
      "grad_norm": 0.8608306646347046,
      "learning_rate": 8.846580262036645e-06,
      "loss": 0.8205,
      "step": 8094
    },
    {
      "epoch": 0.4455391050690737,
      "grad_norm": 0.6917694807052612,
      "learning_rate": 8.84630331946511e-06,
      "loss": 0.7207,
      "step": 8095
    },
    {
      "epoch": 0.44559414387142937,
      "grad_norm": 0.6777203679084778,
      "learning_rate": 8.84602634798579e-06,
      "loss": 0.6939,
      "step": 8096
    },
    {
      "epoch": 0.445649182673785,
      "grad_norm": 0.7249894142150879,
      "learning_rate": 8.845749347600764e-06,
      "loss": 0.7918,
      "step": 8097
    },
    {
      "epoch": 0.4457042214761407,
      "grad_norm": 0.7446995973587036,
      "learning_rate": 8.845472318312116e-06,
      "loss": 0.7379,
      "step": 8098
    },
    {
      "epoch": 0.4457592602784963,
      "grad_norm": 0.8245479464530945,
      "learning_rate": 8.845195260121927e-06,
      "loss": 0.8532,
      "step": 8099
    },
    {
      "epoch": 0.445814299080852,
      "grad_norm": 0.7160329818725586,
      "learning_rate": 8.84491817303228e-06,
      "loss": 0.7042,
      "step": 8100
    },
    {
      "epoch": 0.44586933788320765,
      "grad_norm": 0.8056026101112366,
      "learning_rate": 8.844641057045257e-06,
      "loss": 0.8581,
      "step": 8101
    },
    {
      "epoch": 0.44592437668556334,
      "grad_norm": 0.7257886528968811,
      "learning_rate": 8.84436391216294e-06,
      "loss": 0.7297,
      "step": 8102
    },
    {
      "epoch": 0.445979415487919,
      "grad_norm": 0.7400404810905457,
      "learning_rate": 8.844086738387415e-06,
      "loss": 0.7703,
      "step": 8103
    },
    {
      "epoch": 0.44603445429027466,
      "grad_norm": 0.665271520614624,
      "learning_rate": 8.843809535720763e-06,
      "loss": 0.7769,
      "step": 8104
    },
    {
      "epoch": 0.4460894930926303,
      "grad_norm": 0.7041043639183044,
      "learning_rate": 8.843532304165066e-06,
      "loss": 0.7995,
      "step": 8105
    },
    {
      "epoch": 0.446144531894986,
      "grad_norm": 0.8517841100692749,
      "learning_rate": 8.84325504372241e-06,
      "loss": 0.8239,
      "step": 8106
    },
    {
      "epoch": 0.4461995706973416,
      "grad_norm": 0.7045741677284241,
      "learning_rate": 8.842977754394877e-06,
      "loss": 0.7982,
      "step": 8107
    },
    {
      "epoch": 0.4462546094996973,
      "grad_norm": 0.7056185007095337,
      "learning_rate": 8.842700436184552e-06,
      "loss": 0.8003,
      "step": 8108
    },
    {
      "epoch": 0.44630964830205294,
      "grad_norm": 0.9042232632637024,
      "learning_rate": 8.842423089093519e-06,
      "loss": 0.7534,
      "step": 8109
    },
    {
      "epoch": 0.44636468710440863,
      "grad_norm": 0.8584854602813721,
      "learning_rate": 8.842145713123863e-06,
      "loss": 0.7759,
      "step": 8110
    },
    {
      "epoch": 0.44641972590676426,
      "grad_norm": 0.7333530187606812,
      "learning_rate": 8.841868308277668e-06,
      "loss": 0.7218,
      "step": 8111
    },
    {
      "epoch": 0.44647476470911995,
      "grad_norm": 0.7866941094398499,
      "learning_rate": 8.84159087455702e-06,
      "loss": 0.7016,
      "step": 8112
    },
    {
      "epoch": 0.4465298035114756,
      "grad_norm": 0.7785252928733826,
      "learning_rate": 8.841313411964001e-06,
      "loss": 0.8232,
      "step": 8113
    },
    {
      "epoch": 0.4465848423138313,
      "grad_norm": 0.7060698866844177,
      "learning_rate": 8.841035920500702e-06,
      "loss": 0.6987,
      "step": 8114
    },
    {
      "epoch": 0.4466398811161869,
      "grad_norm": 0.7211717963218689,
      "learning_rate": 8.840758400169203e-06,
      "loss": 0.8604,
      "step": 8115
    },
    {
      "epoch": 0.4466949199185426,
      "grad_norm": 0.979678213596344,
      "learning_rate": 8.840480850971593e-06,
      "loss": 0.9028,
      "step": 8116
    },
    {
      "epoch": 0.44674995872089823,
      "grad_norm": 0.6595104336738586,
      "learning_rate": 8.840203272909957e-06,
      "loss": 0.6899,
      "step": 8117
    },
    {
      "epoch": 0.4468049975232539,
      "grad_norm": 0.6392405033111572,
      "learning_rate": 8.83992566598638e-06,
      "loss": 0.7729,
      "step": 8118
    },
    {
      "epoch": 0.44686003632560956,
      "grad_norm": 1.1084040403366089,
      "learning_rate": 8.839648030202949e-06,
      "loss": 0.822,
      "step": 8119
    },
    {
      "epoch": 0.4469150751279652,
      "grad_norm": 0.7024106383323669,
      "learning_rate": 8.839370365561754e-06,
      "loss": 0.7615,
      "step": 8120
    },
    {
      "epoch": 0.4469701139303209,
      "grad_norm": 0.7204060554504395,
      "learning_rate": 8.839092672064878e-06,
      "loss": 0.7527,
      "step": 8121
    },
    {
      "epoch": 0.4470251527326765,
      "grad_norm": 0.7307723760604858,
      "learning_rate": 8.838814949714407e-06,
      "loss": 0.8139,
      "step": 8122
    },
    {
      "epoch": 0.4470801915350322,
      "grad_norm": 0.824034571647644,
      "learning_rate": 8.838537198512434e-06,
      "loss": 0.8299,
      "step": 8123
    },
    {
      "epoch": 0.44713523033738783,
      "grad_norm": 0.6603747606277466,
      "learning_rate": 8.83825941846104e-06,
      "loss": 0.6762,
      "step": 8124
    },
    {
      "epoch": 0.4471902691397435,
      "grad_norm": 0.7403088808059692,
      "learning_rate": 8.837981609562316e-06,
      "loss": 0.716,
      "step": 8125
    },
    {
      "epoch": 0.44724530794209916,
      "grad_norm": 0.742173969745636,
      "learning_rate": 8.837703771818351e-06,
      "loss": 0.7672,
      "step": 8126
    },
    {
      "epoch": 0.44730034674445485,
      "grad_norm": 0.7158839106559753,
      "learning_rate": 8.837425905231232e-06,
      "loss": 0.6941,
      "step": 8127
    },
    {
      "epoch": 0.4473553855468105,
      "grad_norm": 0.7659464478492737,
      "learning_rate": 8.837148009803044e-06,
      "loss": 0.7293,
      "step": 8128
    },
    {
      "epoch": 0.44741042434916617,
      "grad_norm": 0.8681113719940186,
      "learning_rate": 8.836870085535882e-06,
      "loss": 0.8647,
      "step": 8129
    },
    {
      "epoch": 0.4474654631515218,
      "grad_norm": 0.7117272615432739,
      "learning_rate": 8.83659213243183e-06,
      "loss": 0.8035,
      "step": 8130
    },
    {
      "epoch": 0.4475205019538775,
      "grad_norm": 0.8220957517623901,
      "learning_rate": 8.836314150492978e-06,
      "loss": 0.6978,
      "step": 8131
    },
    {
      "epoch": 0.4475755407562331,
      "grad_norm": 0.7045003175735474,
      "learning_rate": 8.836036139721418e-06,
      "loss": 0.747,
      "step": 8132
    },
    {
      "epoch": 0.4476305795585888,
      "grad_norm": 0.6833191514015198,
      "learning_rate": 8.835758100119235e-06,
      "loss": 0.7604,
      "step": 8133
    },
    {
      "epoch": 0.44768561836094445,
      "grad_norm": 0.7305697798728943,
      "learning_rate": 8.835480031688521e-06,
      "loss": 0.7301,
      "step": 8134
    },
    {
      "epoch": 0.44774065716330014,
      "grad_norm": 0.7266964912414551,
      "learning_rate": 8.835201934431366e-06,
      "loss": 0.7675,
      "step": 8135
    },
    {
      "epoch": 0.44779569596565577,
      "grad_norm": 0.6822015047073364,
      "learning_rate": 8.834923808349861e-06,
      "loss": 0.8226,
      "step": 8136
    },
    {
      "epoch": 0.44785073476801146,
      "grad_norm": 0.7443515062332153,
      "learning_rate": 8.834645653446095e-06,
      "loss": 0.9289,
      "step": 8137
    },
    {
      "epoch": 0.4479057735703671,
      "grad_norm": 0.7337210178375244,
      "learning_rate": 8.834367469722158e-06,
      "loss": 0.7758,
      "step": 8138
    },
    {
      "epoch": 0.4479608123727228,
      "grad_norm": 0.6794925332069397,
      "learning_rate": 8.83408925718014e-06,
      "loss": 0.8426,
      "step": 8139
    },
    {
      "epoch": 0.4480158511750784,
      "grad_norm": 0.7808265089988708,
      "learning_rate": 8.833811015822135e-06,
      "loss": 0.8464,
      "step": 8140
    },
    {
      "epoch": 0.4480708899774341,
      "grad_norm": 0.7837018370628357,
      "learning_rate": 8.833532745650234e-06,
      "loss": 0.8722,
      "step": 8141
    },
    {
      "epoch": 0.44812592877978974,
      "grad_norm": 0.9218140840530396,
      "learning_rate": 8.833254446666526e-06,
      "loss": 0.7981,
      "step": 8142
    },
    {
      "epoch": 0.44818096758214543,
      "grad_norm": 0.7980387806892395,
      "learning_rate": 8.832976118873103e-06,
      "loss": 0.7705,
      "step": 8143
    },
    {
      "epoch": 0.44823600638450106,
      "grad_norm": 0.7354007363319397,
      "learning_rate": 8.832697762272057e-06,
      "loss": 0.8286,
      "step": 8144
    },
    {
      "epoch": 0.44829104518685675,
      "grad_norm": 0.7006223201751709,
      "learning_rate": 8.832419376865482e-06,
      "loss": 0.7107,
      "step": 8145
    },
    {
      "epoch": 0.4483460839892124,
      "grad_norm": 0.7838212847709656,
      "learning_rate": 8.83214096265547e-06,
      "loss": 0.7676,
      "step": 8146
    },
    {
      "epoch": 0.4484011227915681,
      "grad_norm": 0.7768213748931885,
      "learning_rate": 8.83186251964411e-06,
      "loss": 0.8689,
      "step": 8147
    },
    {
      "epoch": 0.4484561615939237,
      "grad_norm": 0.7451630234718323,
      "learning_rate": 8.831584047833497e-06,
      "loss": 0.8625,
      "step": 8148
    },
    {
      "epoch": 0.4485112003962794,
      "grad_norm": 0.7573269605636597,
      "learning_rate": 8.831305547225725e-06,
      "loss": 0.7357,
      "step": 8149
    },
    {
      "epoch": 0.44856623919863503,
      "grad_norm": 0.6884848475456238,
      "learning_rate": 8.831027017822886e-06,
      "loss": 0.7306,
      "step": 8150
    },
    {
      "epoch": 0.4486212780009907,
      "grad_norm": 0.7715907096862793,
      "learning_rate": 8.830748459627073e-06,
      "loss": 0.8311,
      "step": 8151
    },
    {
      "epoch": 0.44867631680334635,
      "grad_norm": 0.6919859647750854,
      "learning_rate": 8.83046987264038e-06,
      "loss": 0.845,
      "step": 8152
    },
    {
      "epoch": 0.44873135560570204,
      "grad_norm": 0.7066411972045898,
      "learning_rate": 8.830191256864902e-06,
      "loss": 0.7554,
      "step": 8153
    },
    {
      "epoch": 0.4487863944080577,
      "grad_norm": 0.754196047782898,
      "learning_rate": 8.829912612302729e-06,
      "loss": 0.7396,
      "step": 8154
    },
    {
      "epoch": 0.44884143321041337,
      "grad_norm": 0.7612286806106567,
      "learning_rate": 8.82963393895596e-06,
      "loss": 0.8154,
      "step": 8155
    },
    {
      "epoch": 0.448896472012769,
      "grad_norm": 0.8576892614364624,
      "learning_rate": 8.829355236826688e-06,
      "loss": 0.7395,
      "step": 8156
    },
    {
      "epoch": 0.4489515108151247,
      "grad_norm": 0.6813738346099854,
      "learning_rate": 8.829076505917005e-06,
      "loss": 0.7661,
      "step": 8157
    },
    {
      "epoch": 0.4490065496174803,
      "grad_norm": 0.7453964948654175,
      "learning_rate": 8.828797746229009e-06,
      "loss": 0.8221,
      "step": 8158
    },
    {
      "epoch": 0.449061588419836,
      "grad_norm": 0.7546728849411011,
      "learning_rate": 8.828518957764795e-06,
      "loss": 0.7717,
      "step": 8159
    },
    {
      "epoch": 0.44911662722219164,
      "grad_norm": 0.8270652890205383,
      "learning_rate": 8.828240140526456e-06,
      "loss": 0.7582,
      "step": 8160
    },
    {
      "epoch": 0.44917166602454733,
      "grad_norm": 0.8188696503639221,
      "learning_rate": 8.827961294516089e-06,
      "loss": 0.8841,
      "step": 8161
    },
    {
      "epoch": 0.44922670482690297,
      "grad_norm": 0.9101365208625793,
      "learning_rate": 8.82768241973579e-06,
      "loss": 0.7099,
      "step": 8162
    },
    {
      "epoch": 0.4492817436292586,
      "grad_norm": 0.6749762892723083,
      "learning_rate": 8.827403516187656e-06,
      "loss": 0.7766,
      "step": 8163
    },
    {
      "epoch": 0.4493367824316143,
      "grad_norm": 1.1351534128189087,
      "learning_rate": 8.827124583873781e-06,
      "loss": 0.7536,
      "step": 8164
    },
    {
      "epoch": 0.4493918212339699,
      "grad_norm": 0.8729487061500549,
      "learning_rate": 8.826845622796261e-06,
      "loss": 0.8613,
      "step": 8165
    },
    {
      "epoch": 0.4494468600363256,
      "grad_norm": 0.7495871782302856,
      "learning_rate": 8.826566632957193e-06,
      "loss": 0.8365,
      "step": 8166
    },
    {
      "epoch": 0.44950189883868125,
      "grad_norm": 0.6414516568183899,
      "learning_rate": 8.826287614358677e-06,
      "loss": 0.6574,
      "step": 8167
    },
    {
      "epoch": 0.44955693764103694,
      "grad_norm": 0.6954017281532288,
      "learning_rate": 8.826008567002805e-06,
      "loss": 0.7857,
      "step": 8168
    },
    {
      "epoch": 0.44961197644339257,
      "grad_norm": 0.7199459075927734,
      "learning_rate": 8.825729490891678e-06,
      "loss": 0.8585,
      "step": 8169
    },
    {
      "epoch": 0.44966701524574826,
      "grad_norm": 0.8245406746864319,
      "learning_rate": 8.825450386027392e-06,
      "loss": 0.7238,
      "step": 8170
    },
    {
      "epoch": 0.4497220540481039,
      "grad_norm": 0.6348667740821838,
      "learning_rate": 8.825171252412044e-06,
      "loss": 0.6991,
      "step": 8171
    },
    {
      "epoch": 0.4497770928504596,
      "grad_norm": 0.6304741501808167,
      "learning_rate": 8.824892090047734e-06,
      "loss": 0.7101,
      "step": 8172
    },
    {
      "epoch": 0.4498321316528152,
      "grad_norm": 0.7088820338249207,
      "learning_rate": 8.82461289893656e-06,
      "loss": 0.8217,
      "step": 8173
    },
    {
      "epoch": 0.4498871704551709,
      "grad_norm": 0.7570851445198059,
      "learning_rate": 8.824333679080617e-06,
      "loss": 0.8029,
      "step": 8174
    },
    {
      "epoch": 0.44994220925752654,
      "grad_norm": 0.7544378042221069,
      "learning_rate": 8.824054430482007e-06,
      "loss": 0.777,
      "step": 8175
    },
    {
      "epoch": 0.4499972480598822,
      "grad_norm": 0.8226260542869568,
      "learning_rate": 8.823775153142827e-06,
      "loss": 0.8391,
      "step": 8176
    },
    {
      "epoch": 0.45005228686223786,
      "grad_norm": 0.6861422061920166,
      "learning_rate": 8.823495847065176e-06,
      "loss": 0.7491,
      "step": 8177
    },
    {
      "epoch": 0.45010732566459355,
      "grad_norm": 0.6643275618553162,
      "learning_rate": 8.823216512251153e-06,
      "loss": 0.6773,
      "step": 8178
    },
    {
      "epoch": 0.4501623644669492,
      "grad_norm": 0.8201391100883484,
      "learning_rate": 8.82293714870286e-06,
      "loss": 0.8065,
      "step": 8179
    },
    {
      "epoch": 0.45021740326930487,
      "grad_norm": 0.7783405780792236,
      "learning_rate": 8.822657756422394e-06,
      "loss": 0.7884,
      "step": 8180
    },
    {
      "epoch": 0.4502724420716605,
      "grad_norm": 0.720745861530304,
      "learning_rate": 8.822378335411856e-06,
      "loss": 0.765,
      "step": 8181
    },
    {
      "epoch": 0.4503274808740162,
      "grad_norm": 0.740364670753479,
      "learning_rate": 8.822098885673346e-06,
      "loss": 0.6354,
      "step": 8182
    },
    {
      "epoch": 0.45038251967637183,
      "grad_norm": 0.8049225807189941,
      "learning_rate": 8.821819407208963e-06,
      "loss": 0.7023,
      "step": 8183
    },
    {
      "epoch": 0.4504375584787275,
      "grad_norm": 0.7320911288261414,
      "learning_rate": 8.821539900020808e-06,
      "loss": 0.8429,
      "step": 8184
    },
    {
      "epoch": 0.45049259728108315,
      "grad_norm": 0.7065376043319702,
      "learning_rate": 8.821260364110984e-06,
      "loss": 0.7283,
      "step": 8185
    },
    {
      "epoch": 0.45054763608343884,
      "grad_norm": 0.7172972559928894,
      "learning_rate": 8.820980799481588e-06,
      "loss": 0.7673,
      "step": 8186
    },
    {
      "epoch": 0.4506026748857945,
      "grad_norm": 0.712273895740509,
      "learning_rate": 8.820701206134724e-06,
      "loss": 0.7317,
      "step": 8187
    },
    {
      "epoch": 0.45065771368815016,
      "grad_norm": 0.6954227685928345,
      "learning_rate": 8.820421584072492e-06,
      "loss": 0.7037,
      "step": 8188
    },
    {
      "epoch": 0.4507127524905058,
      "grad_norm": 0.6790304780006409,
      "learning_rate": 8.820141933296994e-06,
      "loss": 0.7544,
      "step": 8189
    },
    {
      "epoch": 0.4507677912928615,
      "grad_norm": 0.7483745813369751,
      "learning_rate": 8.819862253810332e-06,
      "loss": 0.7894,
      "step": 8190
    },
    {
      "epoch": 0.4508228300952171,
      "grad_norm": 0.7926133871078491,
      "learning_rate": 8.819582545614608e-06,
      "loss": 0.8085,
      "step": 8191
    },
    {
      "epoch": 0.4508778688975728,
      "grad_norm": 0.8442840576171875,
      "learning_rate": 8.819302808711924e-06,
      "loss": 0.8252,
      "step": 8192
    },
    {
      "epoch": 0.45093290769992844,
      "grad_norm": 0.8359581232070923,
      "learning_rate": 8.819023043104383e-06,
      "loss": 0.8187,
      "step": 8193
    },
    {
      "epoch": 0.45098794650228413,
      "grad_norm": 0.7793936133384705,
      "learning_rate": 8.818743248794085e-06,
      "loss": 0.8425,
      "step": 8194
    },
    {
      "epoch": 0.45104298530463977,
      "grad_norm": 0.735509991645813,
      "learning_rate": 8.818463425783136e-06,
      "loss": 0.7781,
      "step": 8195
    },
    {
      "epoch": 0.45109802410699545,
      "grad_norm": 0.6735361814498901,
      "learning_rate": 8.818183574073639e-06,
      "loss": 0.6987,
      "step": 8196
    },
    {
      "epoch": 0.4511530629093511,
      "grad_norm": 0.7780157923698425,
      "learning_rate": 8.817903693667695e-06,
      "loss": 0.8474,
      "step": 8197
    },
    {
      "epoch": 0.4512081017117068,
      "grad_norm": 0.6714445948600769,
      "learning_rate": 8.817623784567411e-06,
      "loss": 0.7216,
      "step": 8198
    },
    {
      "epoch": 0.4512631405140624,
      "grad_norm": 0.6311395168304443,
      "learning_rate": 8.817343846774886e-06,
      "loss": 0.5724,
      "step": 8199
    },
    {
      "epoch": 0.4513181793164181,
      "grad_norm": 0.7446333169937134,
      "learning_rate": 8.817063880292227e-06,
      "loss": 0.7867,
      "step": 8200
    },
    {
      "epoch": 0.45137321811877373,
      "grad_norm": 0.7684246301651001,
      "learning_rate": 8.816783885121539e-06,
      "loss": 0.8141,
      "step": 8201
    },
    {
      "epoch": 0.4514282569211294,
      "grad_norm": 0.754781186580658,
      "learning_rate": 8.816503861264925e-06,
      "loss": 0.8438,
      "step": 8202
    },
    {
      "epoch": 0.45148329572348506,
      "grad_norm": 0.7705762982368469,
      "learning_rate": 8.816223808724488e-06,
      "loss": 0.8948,
      "step": 8203
    },
    {
      "epoch": 0.4515383345258407,
      "grad_norm": 0.7731552720069885,
      "learning_rate": 8.815943727502333e-06,
      "loss": 0.7462,
      "step": 8204
    },
    {
      "epoch": 0.4515933733281964,
      "grad_norm": 0.6615393757820129,
      "learning_rate": 8.81566361760057e-06,
      "loss": 0.7499,
      "step": 8205
    },
    {
      "epoch": 0.451648412130552,
      "grad_norm": 0.724453866481781,
      "learning_rate": 8.8153834790213e-06,
      "loss": 0.7382,
      "step": 8206
    },
    {
      "epoch": 0.4517034509329077,
      "grad_norm": 0.6369735598564148,
      "learning_rate": 8.815103311766629e-06,
      "loss": 0.7452,
      "step": 8207
    },
    {
      "epoch": 0.45175848973526334,
      "grad_norm": 0.686000406742096,
      "learning_rate": 8.814823115838659e-06,
      "loss": 0.6971,
      "step": 8208
    },
    {
      "epoch": 0.451813528537619,
      "grad_norm": 0.7372714281082153,
      "learning_rate": 8.814542891239505e-06,
      "loss": 0.8553,
      "step": 8209
    },
    {
      "epoch": 0.45186856733997466,
      "grad_norm": 0.8348672986030579,
      "learning_rate": 8.814262637971264e-06,
      "loss": 0.7135,
      "step": 8210
    },
    {
      "epoch": 0.45192360614233035,
      "grad_norm": 0.7829258441925049,
      "learning_rate": 8.813982356036049e-06,
      "loss": 0.7974,
      "step": 8211
    },
    {
      "epoch": 0.451978644944686,
      "grad_norm": 0.7013983726501465,
      "learning_rate": 8.81370204543596e-06,
      "loss": 0.7531,
      "step": 8212
    },
    {
      "epoch": 0.45203368374704167,
      "grad_norm": 0.8424196243286133,
      "learning_rate": 8.81342170617311e-06,
      "loss": 0.8217,
      "step": 8213
    },
    {
      "epoch": 0.4520887225493973,
      "grad_norm": 0.7113365530967712,
      "learning_rate": 8.813141338249603e-06,
      "loss": 0.7728,
      "step": 8214
    },
    {
      "epoch": 0.452143761351753,
      "grad_norm": 0.958642303943634,
      "learning_rate": 8.812860941667545e-06,
      "loss": 0.7234,
      "step": 8215
    },
    {
      "epoch": 0.4521988001541086,
      "grad_norm": 0.6712706685066223,
      "learning_rate": 8.812580516429045e-06,
      "loss": 0.6998,
      "step": 8216
    },
    {
      "epoch": 0.4522538389564643,
      "grad_norm": 0.7258469462394714,
      "learning_rate": 8.812300062536212e-06,
      "loss": 0.6758,
      "step": 8217
    },
    {
      "epoch": 0.45230887775881995,
      "grad_norm": 0.735047459602356,
      "learning_rate": 8.812019579991152e-06,
      "loss": 0.7045,
      "step": 8218
    },
    {
      "epoch": 0.45236391656117564,
      "grad_norm": 0.8339886665344238,
      "learning_rate": 8.811739068795971e-06,
      "loss": 0.8069,
      "step": 8219
    },
    {
      "epoch": 0.45241895536353127,
      "grad_norm": 0.7170082926750183,
      "learning_rate": 8.81145852895278e-06,
      "loss": 0.6345,
      "step": 8220
    },
    {
      "epoch": 0.45247399416588696,
      "grad_norm": 0.6892569661140442,
      "learning_rate": 8.81117796046369e-06,
      "loss": 0.712,
      "step": 8221
    },
    {
      "epoch": 0.4525290329682426,
      "grad_norm": 0.6837140321731567,
      "learning_rate": 8.810897363330804e-06,
      "loss": 0.7184,
      "step": 8222
    },
    {
      "epoch": 0.4525840717705983,
      "grad_norm": 0.7410069108009338,
      "learning_rate": 8.810616737556235e-06,
      "loss": 0.8265,
      "step": 8223
    },
    {
      "epoch": 0.4526391105729539,
      "grad_norm": 0.6945875883102417,
      "learning_rate": 8.810336083142089e-06,
      "loss": 0.7163,
      "step": 8224
    },
    {
      "epoch": 0.4526941493753096,
      "grad_norm": 0.6978884339332581,
      "learning_rate": 8.810055400090477e-06,
      "loss": 0.795,
      "step": 8225
    },
    {
      "epoch": 0.45274918817766524,
      "grad_norm": 0.7209095358848572,
      "learning_rate": 8.809774688403509e-06,
      "loss": 0.7317,
      "step": 8226
    },
    {
      "epoch": 0.45280422698002093,
      "grad_norm": 0.7279626727104187,
      "learning_rate": 8.809493948083294e-06,
      "loss": 0.7699,
      "step": 8227
    },
    {
      "epoch": 0.45285926578237656,
      "grad_norm": 0.7642556428909302,
      "learning_rate": 8.809213179131943e-06,
      "loss": 0.8518,
      "step": 8228
    },
    {
      "epoch": 0.45291430458473225,
      "grad_norm": 0.6868709325790405,
      "learning_rate": 8.808932381551565e-06,
      "loss": 0.737,
      "step": 8229
    },
    {
      "epoch": 0.4529693433870879,
      "grad_norm": 0.7012789845466614,
      "learning_rate": 8.80865155534427e-06,
      "loss": 0.8146,
      "step": 8230
    },
    {
      "epoch": 0.4530243821894436,
      "grad_norm": 0.678683340549469,
      "learning_rate": 8.808370700512171e-06,
      "loss": 0.7531,
      "step": 8231
    },
    {
      "epoch": 0.4530794209917992,
      "grad_norm": 0.690559983253479,
      "learning_rate": 8.808089817057377e-06,
      "loss": 0.6779,
      "step": 8232
    },
    {
      "epoch": 0.4531344597941549,
      "grad_norm": 0.7179763317108154,
      "learning_rate": 8.807808904981997e-06,
      "loss": 0.8815,
      "step": 8233
    },
    {
      "epoch": 0.45318949859651053,
      "grad_norm": 0.7708277702331543,
      "learning_rate": 8.807527964288147e-06,
      "loss": 0.8084,
      "step": 8234
    },
    {
      "epoch": 0.4532445373988662,
      "grad_norm": 0.6828494071960449,
      "learning_rate": 8.807246994977936e-06,
      "loss": 0.7587,
      "step": 8235
    },
    {
      "epoch": 0.45329957620122185,
      "grad_norm": 0.7085250616073608,
      "learning_rate": 8.806965997053475e-06,
      "loss": 0.7894,
      "step": 8236
    },
    {
      "epoch": 0.45335461500357754,
      "grad_norm": 0.7723467946052551,
      "learning_rate": 8.806684970516876e-06,
      "loss": 0.7408,
      "step": 8237
    },
    {
      "epoch": 0.4534096538059332,
      "grad_norm": 0.8887566328048706,
      "learning_rate": 8.806403915370253e-06,
      "loss": 0.9022,
      "step": 8238
    },
    {
      "epoch": 0.45346469260828887,
      "grad_norm": 0.7379833459854126,
      "learning_rate": 8.806122831615718e-06,
      "loss": 0.8264,
      "step": 8239
    },
    {
      "epoch": 0.4535197314106445,
      "grad_norm": 0.903279721736908,
      "learning_rate": 8.80584171925538e-06,
      "loss": 0.7432,
      "step": 8240
    },
    {
      "epoch": 0.4535747702130002,
      "grad_norm": 0.7671363353729248,
      "learning_rate": 8.805560578291356e-06,
      "loss": 0.8109,
      "step": 8241
    },
    {
      "epoch": 0.4536298090153558,
      "grad_norm": 0.6047827005386353,
      "learning_rate": 8.805279408725755e-06,
      "loss": 0.6628,
      "step": 8242
    },
    {
      "epoch": 0.4536848478177115,
      "grad_norm": 1.0570796728134155,
      "learning_rate": 8.804998210560696e-06,
      "loss": 0.7981,
      "step": 8243
    },
    {
      "epoch": 0.45373988662006715,
      "grad_norm": 0.7116600871086121,
      "learning_rate": 8.804716983798288e-06,
      "loss": 0.7601,
      "step": 8244
    },
    {
      "epoch": 0.45379492542242283,
      "grad_norm": 0.7162767648696899,
      "learning_rate": 8.804435728440644e-06,
      "loss": 0.8389,
      "step": 8245
    },
    {
      "epoch": 0.45384996422477847,
      "grad_norm": 0.6715626120567322,
      "learning_rate": 8.80415444448988e-06,
      "loss": 0.6377,
      "step": 8246
    },
    {
      "epoch": 0.4539050030271341,
      "grad_norm": 0.7168908715248108,
      "learning_rate": 8.80387313194811e-06,
      "loss": 0.7946,
      "step": 8247
    },
    {
      "epoch": 0.4539600418294898,
      "grad_norm": 0.7497992515563965,
      "learning_rate": 8.803591790817448e-06,
      "loss": 0.8026,
      "step": 8248
    },
    {
      "epoch": 0.4540150806318454,
      "grad_norm": 0.6665049195289612,
      "learning_rate": 8.803310421100009e-06,
      "loss": 0.779,
      "step": 8249
    },
    {
      "epoch": 0.4540701194342011,
      "grad_norm": 0.766674280166626,
      "learning_rate": 8.803029022797905e-06,
      "loss": 0.7467,
      "step": 8250
    },
    {
      "epoch": 0.45412515823655675,
      "grad_norm": 0.7306104302406311,
      "learning_rate": 8.802747595913255e-06,
      "loss": 0.8323,
      "step": 8251
    },
    {
      "epoch": 0.45418019703891244,
      "grad_norm": 0.6425766944885254,
      "learning_rate": 8.802466140448169e-06,
      "loss": 0.7226,
      "step": 8252
    },
    {
      "epoch": 0.45423523584126807,
      "grad_norm": 0.7992560267448425,
      "learning_rate": 8.802184656404769e-06,
      "loss": 0.7285,
      "step": 8253
    },
    {
      "epoch": 0.45429027464362376,
      "grad_norm": 0.6935924887657166,
      "learning_rate": 8.801903143785164e-06,
      "loss": 0.5757,
      "step": 8254
    },
    {
      "epoch": 0.4543453134459794,
      "grad_norm": 0.7091512084007263,
      "learning_rate": 8.801621602591473e-06,
      "loss": 0.7719,
      "step": 8255
    },
    {
      "epoch": 0.4544003522483351,
      "grad_norm": 0.851231038570404,
      "learning_rate": 8.801340032825814e-06,
      "loss": 0.7804,
      "step": 8256
    },
    {
      "epoch": 0.4544553910506907,
      "grad_norm": 0.7443445920944214,
      "learning_rate": 8.801058434490298e-06,
      "loss": 0.7172,
      "step": 8257
    },
    {
      "epoch": 0.4545104298530464,
      "grad_norm": 0.7156546115875244,
      "learning_rate": 8.800776807587046e-06,
      "loss": 0.7756,
      "step": 8258
    },
    {
      "epoch": 0.45456546865540204,
      "grad_norm": 0.8027580380439758,
      "learning_rate": 8.800495152118172e-06,
      "loss": 0.8035,
      "step": 8259
    },
    {
      "epoch": 0.4546205074577577,
      "grad_norm": 0.6868240833282471,
      "learning_rate": 8.800213468085794e-06,
      "loss": 0.7159,
      "step": 8260
    },
    {
      "epoch": 0.45467554626011336,
      "grad_norm": 0.9127504229545593,
      "learning_rate": 8.79993175549203e-06,
      "loss": 0.7705,
      "step": 8261
    },
    {
      "epoch": 0.45473058506246905,
      "grad_norm": 0.7074575424194336,
      "learning_rate": 8.799650014338994e-06,
      "loss": 0.7841,
      "step": 8262
    },
    {
      "epoch": 0.4547856238648247,
      "grad_norm": 0.7462378740310669,
      "learning_rate": 8.799368244628807e-06,
      "loss": 0.8125,
      "step": 8263
    },
    {
      "epoch": 0.4548406626671804,
      "grad_norm": 0.7510300874710083,
      "learning_rate": 8.799086446363585e-06,
      "loss": 0.8354,
      "step": 8264
    },
    {
      "epoch": 0.454895701469536,
      "grad_norm": 0.7134591937065125,
      "learning_rate": 8.798804619545446e-06,
      "loss": 0.7968,
      "step": 8265
    },
    {
      "epoch": 0.4549507402718917,
      "grad_norm": 1.0424071550369263,
      "learning_rate": 8.798522764176509e-06,
      "loss": 0.8638,
      "step": 8266
    },
    {
      "epoch": 0.45500577907424733,
      "grad_norm": 0.6805267930030823,
      "learning_rate": 8.79824088025889e-06,
      "loss": 0.757,
      "step": 8267
    },
    {
      "epoch": 0.455060817876603,
      "grad_norm": 0.8145313262939453,
      "learning_rate": 8.79795896779471e-06,
      "loss": 0.7589,
      "step": 8268
    },
    {
      "epoch": 0.45511585667895865,
      "grad_norm": 0.7611781358718872,
      "learning_rate": 8.79767702678609e-06,
      "loss": 0.8426,
      "step": 8269
    },
    {
      "epoch": 0.45517089548131434,
      "grad_norm": 0.7639568448066711,
      "learning_rate": 8.797395057235142e-06,
      "loss": 0.6609,
      "step": 8270
    },
    {
      "epoch": 0.45522593428367,
      "grad_norm": 0.8577544093132019,
      "learning_rate": 8.79711305914399e-06,
      "loss": 0.8085,
      "step": 8271
    },
    {
      "epoch": 0.45528097308602566,
      "grad_norm": 0.7740383148193359,
      "learning_rate": 8.796831032514754e-06,
      "loss": 0.8689,
      "step": 8272
    },
    {
      "epoch": 0.4553360118883813,
      "grad_norm": 0.7300885915756226,
      "learning_rate": 8.796548977349553e-06,
      "loss": 0.8303,
      "step": 8273
    },
    {
      "epoch": 0.455391050690737,
      "grad_norm": 0.6677057147026062,
      "learning_rate": 8.796266893650504e-06,
      "loss": 0.7449,
      "step": 8274
    },
    {
      "epoch": 0.4554460894930926,
      "grad_norm": 0.7269144058227539,
      "learning_rate": 8.79598478141973e-06,
      "loss": 0.8744,
      "step": 8275
    },
    {
      "epoch": 0.4555011282954483,
      "grad_norm": 0.7458559274673462,
      "learning_rate": 8.795702640659351e-06,
      "loss": 0.8036,
      "step": 8276
    },
    {
      "epoch": 0.45555616709780394,
      "grad_norm": 0.7693114280700684,
      "learning_rate": 8.795420471371487e-06,
      "loss": 0.7617,
      "step": 8277
    },
    {
      "epoch": 0.45561120590015963,
      "grad_norm": 0.7594510316848755,
      "learning_rate": 8.79513827355826e-06,
      "loss": 0.7049,
      "step": 8278
    },
    {
      "epoch": 0.45566624470251527,
      "grad_norm": 0.7481217980384827,
      "learning_rate": 8.794856047221786e-06,
      "loss": 0.804,
      "step": 8279
    },
    {
      "epoch": 0.45572128350487096,
      "grad_norm": 0.726859986782074,
      "learning_rate": 8.794573792364192e-06,
      "loss": 0.7322,
      "step": 8280
    },
    {
      "epoch": 0.4557763223072266,
      "grad_norm": 0.6800149083137512,
      "learning_rate": 8.794291508987597e-06,
      "loss": 0.8467,
      "step": 8281
    },
    {
      "epoch": 0.4558313611095823,
      "grad_norm": 0.6264217495918274,
      "learning_rate": 8.794009197094122e-06,
      "loss": 0.6203,
      "step": 8282
    },
    {
      "epoch": 0.4558863999119379,
      "grad_norm": 0.6973850131034851,
      "learning_rate": 8.79372685668589e-06,
      "loss": 0.8211,
      "step": 8283
    },
    {
      "epoch": 0.4559414387142936,
      "grad_norm": 0.6992879509925842,
      "learning_rate": 8.793444487765022e-06,
      "loss": 0.7831,
      "step": 8284
    },
    {
      "epoch": 0.45599647751664923,
      "grad_norm": 0.7641519904136658,
      "learning_rate": 8.793162090333643e-06,
      "loss": 0.7519,
      "step": 8285
    },
    {
      "epoch": 0.4560515163190049,
      "grad_norm": 0.7296152710914612,
      "learning_rate": 8.79287966439387e-06,
      "loss": 0.8738,
      "step": 8286
    },
    {
      "epoch": 0.45610655512136056,
      "grad_norm": 0.7549383044242859,
      "learning_rate": 8.79259720994783e-06,
      "loss": 0.7868,
      "step": 8287
    },
    {
      "epoch": 0.45616159392371625,
      "grad_norm": 0.7932083606719971,
      "learning_rate": 8.792314726997644e-06,
      "loss": 0.8443,
      "step": 8288
    },
    {
      "epoch": 0.4562166327260719,
      "grad_norm": 0.7999894022941589,
      "learning_rate": 8.792032215545437e-06,
      "loss": 0.852,
      "step": 8289
    },
    {
      "epoch": 0.4562716715284275,
      "grad_norm": 0.8092383742332458,
      "learning_rate": 8.79174967559333e-06,
      "loss": 0.7922,
      "step": 8290
    },
    {
      "epoch": 0.4563267103307832,
      "grad_norm": 0.7481340169906616,
      "learning_rate": 8.791467107143447e-06,
      "loss": 0.7086,
      "step": 8291
    },
    {
      "epoch": 0.45638174913313884,
      "grad_norm": 0.8096129298210144,
      "learning_rate": 8.791184510197912e-06,
      "loss": 0.6645,
      "step": 8292
    },
    {
      "epoch": 0.4564367879354945,
      "grad_norm": 0.7276492118835449,
      "learning_rate": 8.79090188475885e-06,
      "loss": 0.7174,
      "step": 8293
    },
    {
      "epoch": 0.45649182673785016,
      "grad_norm": 0.815535843372345,
      "learning_rate": 8.790619230828385e-06,
      "loss": 0.8622,
      "step": 8294
    },
    {
      "epoch": 0.45654686554020585,
      "grad_norm": 0.8191169500350952,
      "learning_rate": 8.790336548408637e-06,
      "loss": 0.8666,
      "step": 8295
    },
    {
      "epoch": 0.4566019043425615,
      "grad_norm": 0.7449167966842651,
      "learning_rate": 8.790053837501737e-06,
      "loss": 0.7728,
      "step": 8296
    },
    {
      "epoch": 0.45665694314491717,
      "grad_norm": 0.7311065196990967,
      "learning_rate": 8.789771098109808e-06,
      "loss": 0.8059,
      "step": 8297
    },
    {
      "epoch": 0.4567119819472728,
      "grad_norm": 0.7381907105445862,
      "learning_rate": 8.789488330234971e-06,
      "loss": 0.7722,
      "step": 8298
    },
    {
      "epoch": 0.4567670207496285,
      "grad_norm": 0.8180661201477051,
      "learning_rate": 8.789205533879355e-06,
      "loss": 0.9032,
      "step": 8299
    },
    {
      "epoch": 0.4568220595519841,
      "grad_norm": 0.7993118762969971,
      "learning_rate": 8.788922709045087e-06,
      "loss": 0.8065,
      "step": 8300
    },
    {
      "epoch": 0.4568770983543398,
      "grad_norm": 0.8449206948280334,
      "learning_rate": 8.788639855734287e-06,
      "loss": 0.7895,
      "step": 8301
    },
    {
      "epoch": 0.45693213715669545,
      "grad_norm": 0.9224583506584167,
      "learning_rate": 8.788356973949084e-06,
      "loss": 0.78,
      "step": 8302
    },
    {
      "epoch": 0.45698717595905114,
      "grad_norm": 0.7109915614128113,
      "learning_rate": 8.788074063691604e-06,
      "loss": 0.8029,
      "step": 8303
    },
    {
      "epoch": 0.4570422147614068,
      "grad_norm": 0.7372310757637024,
      "learning_rate": 8.787791124963976e-06,
      "loss": 0.8118,
      "step": 8304
    },
    {
      "epoch": 0.45709725356376246,
      "grad_norm": 0.8127168416976929,
      "learning_rate": 8.787508157768323e-06,
      "loss": 0.8665,
      "step": 8305
    },
    {
      "epoch": 0.4571522923661181,
      "grad_norm": 0.7193050980567932,
      "learning_rate": 8.787225162106771e-06,
      "loss": 0.749,
      "step": 8306
    },
    {
      "epoch": 0.4572073311684738,
      "grad_norm": 0.8825041651725769,
      "learning_rate": 8.786942137981449e-06,
      "loss": 0.9651,
      "step": 8307
    },
    {
      "epoch": 0.4572623699708294,
      "grad_norm": 0.6854885816574097,
      "learning_rate": 8.786659085394485e-06,
      "loss": 0.8259,
      "step": 8308
    },
    {
      "epoch": 0.4573174087731851,
      "grad_norm": 0.6698010563850403,
      "learning_rate": 8.786376004348004e-06,
      "loss": 0.7212,
      "step": 8309
    },
    {
      "epoch": 0.45737244757554074,
      "grad_norm": 0.7706398963928223,
      "learning_rate": 8.786092894844132e-06,
      "loss": 0.719,
      "step": 8310
    },
    {
      "epoch": 0.45742748637789643,
      "grad_norm": 0.8905620574951172,
      "learning_rate": 8.785809756885002e-06,
      "loss": 0.7518,
      "step": 8311
    },
    {
      "epoch": 0.45748252518025206,
      "grad_norm": 0.7537117004394531,
      "learning_rate": 8.78552659047274e-06,
      "loss": 0.8267,
      "step": 8312
    },
    {
      "epoch": 0.45753756398260775,
      "grad_norm": 0.7840754985809326,
      "learning_rate": 8.78524339560947e-06,
      "loss": 0.8417,
      "step": 8313
    },
    {
      "epoch": 0.4575926027849634,
      "grad_norm": 0.7373713254928589,
      "learning_rate": 8.784960172297327e-06,
      "loss": 0.784,
      "step": 8314
    },
    {
      "epoch": 0.4576476415873191,
      "grad_norm": 0.6648432016372681,
      "learning_rate": 8.784676920538436e-06,
      "loss": 0.7252,
      "step": 8315
    },
    {
      "epoch": 0.4577026803896747,
      "grad_norm": 0.7904912829399109,
      "learning_rate": 8.784393640334925e-06,
      "loss": 0.7777,
      "step": 8316
    },
    {
      "epoch": 0.4577577191920304,
      "grad_norm": 0.7691501379013062,
      "learning_rate": 8.784110331688927e-06,
      "loss": 0.733,
      "step": 8317
    },
    {
      "epoch": 0.45781275799438603,
      "grad_norm": 0.6054617762565613,
      "learning_rate": 8.783826994602566e-06,
      "loss": 0.6367,
      "step": 8318
    },
    {
      "epoch": 0.4578677967967417,
      "grad_norm": 0.7495457530021667,
      "learning_rate": 8.783543629077976e-06,
      "loss": 0.8672,
      "step": 8319
    },
    {
      "epoch": 0.45792283559909736,
      "grad_norm": 0.6979867815971375,
      "learning_rate": 8.783260235117283e-06,
      "loss": 0.7338,
      "step": 8320
    },
    {
      "epoch": 0.45797787440145304,
      "grad_norm": 0.6927759647369385,
      "learning_rate": 8.78297681272262e-06,
      "loss": 0.6925,
      "step": 8321
    },
    {
      "epoch": 0.4580329132038087,
      "grad_norm": 0.9076687097549438,
      "learning_rate": 8.782693361896115e-06,
      "loss": 0.8225,
      "step": 8322
    },
    {
      "epoch": 0.45808795200616437,
      "grad_norm": 0.7990893721580505,
      "learning_rate": 8.782409882639902e-06,
      "loss": 0.8144,
      "step": 8323
    },
    {
      "epoch": 0.45814299080852,
      "grad_norm": 0.7958230376243591,
      "learning_rate": 8.782126374956107e-06,
      "loss": 0.7717,
      "step": 8324
    },
    {
      "epoch": 0.4581980296108757,
      "grad_norm": 0.7694645524024963,
      "learning_rate": 8.781842838846861e-06,
      "loss": 0.8314,
      "step": 8325
    },
    {
      "epoch": 0.4582530684132313,
      "grad_norm": 0.8653621077537537,
      "learning_rate": 8.781559274314297e-06,
      "loss": 0.7567,
      "step": 8326
    },
    {
      "epoch": 0.458308107215587,
      "grad_norm": 0.7834668755531311,
      "learning_rate": 8.781275681360548e-06,
      "loss": 0.7431,
      "step": 8327
    },
    {
      "epoch": 0.45836314601794265,
      "grad_norm": 0.6800104975700378,
      "learning_rate": 8.780992059987742e-06,
      "loss": 0.8266,
      "step": 8328
    },
    {
      "epoch": 0.45841818482029834,
      "grad_norm": 0.7274910807609558,
      "learning_rate": 8.780708410198011e-06,
      "loss": 0.7358,
      "step": 8329
    },
    {
      "epoch": 0.45847322362265397,
      "grad_norm": 0.8102344870567322,
      "learning_rate": 8.780424731993488e-06,
      "loss": 0.7397,
      "step": 8330
    },
    {
      "epoch": 0.45852826242500966,
      "grad_norm": 0.7536956071853638,
      "learning_rate": 8.780141025376305e-06,
      "loss": 0.7053,
      "step": 8331
    },
    {
      "epoch": 0.4585833012273653,
      "grad_norm": 0.678535521030426,
      "learning_rate": 8.779857290348594e-06,
      "loss": 0.792,
      "step": 8332
    },
    {
      "epoch": 0.4586383400297209,
      "grad_norm": 0.8847216963768005,
      "learning_rate": 8.779573526912487e-06,
      "loss": 0.8117,
      "step": 8333
    },
    {
      "epoch": 0.4586933788320766,
      "grad_norm": 0.6997288465499878,
      "learning_rate": 8.779289735070117e-06,
      "loss": 0.7797,
      "step": 8334
    },
    {
      "epoch": 0.45874841763443225,
      "grad_norm": 0.7445441484451294,
      "learning_rate": 8.779005914823617e-06,
      "loss": 0.7505,
      "step": 8335
    },
    {
      "epoch": 0.45880345643678794,
      "grad_norm": 0.618844211101532,
      "learning_rate": 8.778722066175121e-06,
      "loss": 0.661,
      "step": 8336
    },
    {
      "epoch": 0.45885849523914357,
      "grad_norm": 0.6810492873191833,
      "learning_rate": 8.778438189126761e-06,
      "loss": 0.6819,
      "step": 8337
    },
    {
      "epoch": 0.45891353404149926,
      "grad_norm": 0.6785591244697571,
      "learning_rate": 8.778154283680671e-06,
      "loss": 0.7808,
      "step": 8338
    },
    {
      "epoch": 0.4589685728438549,
      "grad_norm": 0.7461212873458862,
      "learning_rate": 8.777870349838984e-06,
      "loss": 0.8566,
      "step": 8339
    },
    {
      "epoch": 0.4590236116462106,
      "grad_norm": 0.6731496453285217,
      "learning_rate": 8.777586387603836e-06,
      "loss": 0.823,
      "step": 8340
    },
    {
      "epoch": 0.4590786504485662,
      "grad_norm": 0.7295553684234619,
      "learning_rate": 8.77730239697736e-06,
      "loss": 0.9229,
      "step": 8341
    },
    {
      "epoch": 0.4591336892509219,
      "grad_norm": 0.783275842666626,
      "learning_rate": 8.77701837796169e-06,
      "loss": 0.782,
      "step": 8342
    },
    {
      "epoch": 0.45918872805327754,
      "grad_norm": 0.6952852606773376,
      "learning_rate": 8.77673433055896e-06,
      "loss": 0.7977,
      "step": 8343
    },
    {
      "epoch": 0.45924376685563323,
      "grad_norm": 0.7381969094276428,
      "learning_rate": 8.776450254771305e-06,
      "loss": 0.768,
      "step": 8344
    },
    {
      "epoch": 0.45929880565798886,
      "grad_norm": 0.7911093831062317,
      "learning_rate": 8.776166150600862e-06,
      "loss": 0.8284,
      "step": 8345
    },
    {
      "epoch": 0.45935384446034455,
      "grad_norm": 0.7319246530532837,
      "learning_rate": 8.775882018049765e-06,
      "loss": 0.8135,
      "step": 8346
    },
    {
      "epoch": 0.4594088832627002,
      "grad_norm": 0.7888429760932922,
      "learning_rate": 8.77559785712015e-06,
      "loss": 0.9001,
      "step": 8347
    },
    {
      "epoch": 0.4594639220650559,
      "grad_norm": 0.6983326077461243,
      "learning_rate": 8.775313667814151e-06,
      "loss": 0.7537,
      "step": 8348
    },
    {
      "epoch": 0.4595189608674115,
      "grad_norm": 0.7532416582107544,
      "learning_rate": 8.775029450133905e-06,
      "loss": 0.8307,
      "step": 8349
    },
    {
      "epoch": 0.4595739996697672,
      "grad_norm": 0.7159993052482605,
      "learning_rate": 8.774745204081549e-06,
      "loss": 0.7874,
      "step": 8350
    },
    {
      "epoch": 0.45962903847212283,
      "grad_norm": 0.6898767352104187,
      "learning_rate": 8.774460929659218e-06,
      "loss": 0.7453,
      "step": 8351
    },
    {
      "epoch": 0.4596840772744785,
      "grad_norm": 0.6833236813545227,
      "learning_rate": 8.774176626869051e-06,
      "loss": 0.7281,
      "step": 8352
    },
    {
      "epoch": 0.45973911607683415,
      "grad_norm": 0.7840244770050049,
      "learning_rate": 8.77389229571318e-06,
      "loss": 0.7194,
      "step": 8353
    },
    {
      "epoch": 0.45979415487918984,
      "grad_norm": 0.7920441627502441,
      "learning_rate": 8.773607936193747e-06,
      "loss": 0.7135,
      "step": 8354
    },
    {
      "epoch": 0.4598491936815455,
      "grad_norm": 0.7395668625831604,
      "learning_rate": 8.773323548312884e-06,
      "loss": 0.8162,
      "step": 8355
    },
    {
      "epoch": 0.45990423248390117,
      "grad_norm": 0.7854128479957581,
      "learning_rate": 8.773039132072734e-06,
      "loss": 0.8252,
      "step": 8356
    },
    {
      "epoch": 0.4599592712862568,
      "grad_norm": 0.694997251033783,
      "learning_rate": 8.772754687475431e-06,
      "loss": 0.6627,
      "step": 8357
    },
    {
      "epoch": 0.4600143100886125,
      "grad_norm": 0.7698866724967957,
      "learning_rate": 8.772470214523112e-06,
      "loss": 0.8814,
      "step": 8358
    },
    {
      "epoch": 0.4600693488909681,
      "grad_norm": 0.7323407530784607,
      "learning_rate": 8.77218571321792e-06,
      "loss": 0.7769,
      "step": 8359
    },
    {
      "epoch": 0.4601243876933238,
      "grad_norm": 0.6637027263641357,
      "learning_rate": 8.771901183561986e-06,
      "loss": 0.6741,
      "step": 8360
    },
    {
      "epoch": 0.46017942649567944,
      "grad_norm": 0.7423702478408813,
      "learning_rate": 8.771616625557455e-06,
      "loss": 0.7303,
      "step": 8361
    },
    {
      "epoch": 0.46023446529803513,
      "grad_norm": 0.7599568367004395,
      "learning_rate": 8.771332039206463e-06,
      "loss": 0.8161,
      "step": 8362
    },
    {
      "epoch": 0.46028950410039077,
      "grad_norm": 0.9063183069229126,
      "learning_rate": 8.771047424511148e-06,
      "loss": 0.8098,
      "step": 8363
    },
    {
      "epoch": 0.46034454290274646,
      "grad_norm": 0.658210813999176,
      "learning_rate": 8.770762781473651e-06,
      "loss": 0.7097,
      "step": 8364
    },
    {
      "epoch": 0.4603995817051021,
      "grad_norm": 0.8396975994110107,
      "learning_rate": 8.770478110096111e-06,
      "loss": 0.8731,
      "step": 8365
    },
    {
      "epoch": 0.4604546205074578,
      "grad_norm": 0.7334815263748169,
      "learning_rate": 8.770193410380663e-06,
      "loss": 0.7689,
      "step": 8366
    },
    {
      "epoch": 0.4605096593098134,
      "grad_norm": 0.8220386505126953,
      "learning_rate": 8.769908682329453e-06,
      "loss": 0.8139,
      "step": 8367
    },
    {
      "epoch": 0.4605646981121691,
      "grad_norm": 0.8077995181083679,
      "learning_rate": 8.76962392594462e-06,
      "loss": 0.7379,
      "step": 8368
    },
    {
      "epoch": 0.46061973691452474,
      "grad_norm": 0.8007730841636658,
      "learning_rate": 8.7693391412283e-06,
      "loss": 0.7835,
      "step": 8369
    },
    {
      "epoch": 0.4606747757168804,
      "grad_norm": 0.7108187079429626,
      "learning_rate": 8.769054328182637e-06,
      "loss": 0.6787,
      "step": 8370
    },
    {
      "epoch": 0.46072981451923606,
      "grad_norm": 0.7623056173324585,
      "learning_rate": 8.768769486809772e-06,
      "loss": 0.8056,
      "step": 8371
    },
    {
      "epoch": 0.46078485332159175,
      "grad_norm": 0.6991614103317261,
      "learning_rate": 8.768484617111843e-06,
      "loss": 0.7404,
      "step": 8372
    },
    {
      "epoch": 0.4608398921239474,
      "grad_norm": 0.7531471848487854,
      "learning_rate": 8.768199719090991e-06,
      "loss": 0.8104,
      "step": 8373
    },
    {
      "epoch": 0.46089493092630307,
      "grad_norm": 1.0271111726760864,
      "learning_rate": 8.76791479274936e-06,
      "loss": 0.9028,
      "step": 8374
    },
    {
      "epoch": 0.4609499697286587,
      "grad_norm": 0.7346897125244141,
      "learning_rate": 8.76762983808909e-06,
      "loss": 0.8179,
      "step": 8375
    },
    {
      "epoch": 0.46100500853101434,
      "grad_norm": 0.6413559913635254,
      "learning_rate": 8.767344855112324e-06,
      "loss": 0.7995,
      "step": 8376
    },
    {
      "epoch": 0.46106004733337,
      "grad_norm": 0.7187537550926208,
      "learning_rate": 8.767059843821199e-06,
      "loss": 0.7973,
      "step": 8377
    },
    {
      "epoch": 0.46111508613572566,
      "grad_norm": 0.6819092035293579,
      "learning_rate": 8.766774804217864e-06,
      "loss": 0.8255,
      "step": 8378
    },
    {
      "epoch": 0.46117012493808135,
      "grad_norm": 0.683318018913269,
      "learning_rate": 8.766489736304457e-06,
      "loss": 0.6794,
      "step": 8379
    },
    {
      "epoch": 0.461225163740437,
      "grad_norm": 0.7345470786094666,
      "learning_rate": 8.76620464008312e-06,
      "loss": 0.8741,
      "step": 8380
    },
    {
      "epoch": 0.46128020254279267,
      "grad_norm": 0.7369397282600403,
      "learning_rate": 8.765919515556e-06,
      "loss": 0.8301,
      "step": 8381
    },
    {
      "epoch": 0.4613352413451483,
      "grad_norm": 0.7304979562759399,
      "learning_rate": 8.765634362725233e-06,
      "loss": 0.7507,
      "step": 8382
    },
    {
      "epoch": 0.461390280147504,
      "grad_norm": 0.7968454957008362,
      "learning_rate": 8.765349181592969e-06,
      "loss": 0.7396,
      "step": 8383
    },
    {
      "epoch": 0.46144531894985963,
      "grad_norm": 0.691439151763916,
      "learning_rate": 8.765063972161347e-06,
      "loss": 0.7199,
      "step": 8384
    },
    {
      "epoch": 0.4615003577522153,
      "grad_norm": 0.8355879187583923,
      "learning_rate": 8.764778734432513e-06,
      "loss": 0.7369,
      "step": 8385
    },
    {
      "epoch": 0.46155539655457095,
      "grad_norm": 0.908017098903656,
      "learning_rate": 8.76449346840861e-06,
      "loss": 0.8271,
      "step": 8386
    },
    {
      "epoch": 0.46161043535692664,
      "grad_norm": 0.6426172852516174,
      "learning_rate": 8.764208174091781e-06,
      "loss": 0.6646,
      "step": 8387
    },
    {
      "epoch": 0.4616654741592823,
      "grad_norm": 0.7003652453422546,
      "learning_rate": 8.763922851484171e-06,
      "loss": 0.7272,
      "step": 8388
    },
    {
      "epoch": 0.46172051296163796,
      "grad_norm": 0.7470494508743286,
      "learning_rate": 8.763637500587925e-06,
      "loss": 0.8333,
      "step": 8389
    },
    {
      "epoch": 0.4617755517639936,
      "grad_norm": 0.6974903345108032,
      "learning_rate": 8.763352121405187e-06,
      "loss": 0.834,
      "step": 8390
    },
    {
      "epoch": 0.4618305905663493,
      "grad_norm": 0.8146659135818481,
      "learning_rate": 8.7630667139381e-06,
      "loss": 0.724,
      "step": 8391
    },
    {
      "epoch": 0.4618856293687049,
      "grad_norm": 0.6614096164703369,
      "learning_rate": 8.762781278188813e-06,
      "loss": 0.6822,
      "step": 8392
    },
    {
      "epoch": 0.4619406681710606,
      "grad_norm": 0.712944746017456,
      "learning_rate": 8.762495814159469e-06,
      "loss": 0.7864,
      "step": 8393
    },
    {
      "epoch": 0.46199570697341624,
      "grad_norm": 0.7531552910804749,
      "learning_rate": 8.762210321852213e-06,
      "loss": 0.7494,
      "step": 8394
    },
    {
      "epoch": 0.46205074577577193,
      "grad_norm": 0.8150199055671692,
      "learning_rate": 8.761924801269191e-06,
      "loss": 0.7869,
      "step": 8395
    },
    {
      "epoch": 0.46210578457812757,
      "grad_norm": 0.8586462736129761,
      "learning_rate": 8.76163925241255e-06,
      "loss": 0.7647,
      "step": 8396
    },
    {
      "epoch": 0.46216082338048325,
      "grad_norm": 0.7258061766624451,
      "learning_rate": 8.761353675284434e-06,
      "loss": 0.7672,
      "step": 8397
    },
    {
      "epoch": 0.4622158621828389,
      "grad_norm": 0.6592851281166077,
      "learning_rate": 8.761068069886992e-06,
      "loss": 0.7488,
      "step": 8398
    },
    {
      "epoch": 0.4622709009851946,
      "grad_norm": 0.7410836219787598,
      "learning_rate": 8.760782436222368e-06,
      "loss": 0.6669,
      "step": 8399
    },
    {
      "epoch": 0.4623259397875502,
      "grad_norm": 0.7121642231941223,
      "learning_rate": 8.76049677429271e-06,
      "loss": 0.7005,
      "step": 8400
    },
    {
      "epoch": 0.4623809785899059,
      "grad_norm": 0.7170663475990295,
      "learning_rate": 8.760211084100166e-06,
      "loss": 0.8154,
      "step": 8401
    },
    {
      "epoch": 0.46243601739226153,
      "grad_norm": 0.6851769685745239,
      "learning_rate": 8.759925365646882e-06,
      "loss": 0.7948,
      "step": 8402
    },
    {
      "epoch": 0.4624910561946172,
      "grad_norm": 0.7728533744812012,
      "learning_rate": 8.759639618935006e-06,
      "loss": 0.8263,
      "step": 8403
    },
    {
      "epoch": 0.46254609499697286,
      "grad_norm": 0.7276784777641296,
      "learning_rate": 8.759353843966682e-06,
      "loss": 0.6992,
      "step": 8404
    },
    {
      "epoch": 0.46260113379932855,
      "grad_norm": 0.7533649802207947,
      "learning_rate": 8.759068040744063e-06,
      "loss": 0.7744,
      "step": 8405
    },
    {
      "epoch": 0.4626561726016842,
      "grad_norm": 0.6911979913711548,
      "learning_rate": 8.758782209269294e-06,
      "loss": 0.6977,
      "step": 8406
    },
    {
      "epoch": 0.46271121140403987,
      "grad_norm": 0.6723766922950745,
      "learning_rate": 8.758496349544526e-06,
      "loss": 0.7286,
      "step": 8407
    },
    {
      "epoch": 0.4627662502063955,
      "grad_norm": 0.7327921390533447,
      "learning_rate": 8.758210461571903e-06,
      "loss": 0.7708,
      "step": 8408
    },
    {
      "epoch": 0.4628212890087512,
      "grad_norm": 0.7498626708984375,
      "learning_rate": 8.757924545353578e-06,
      "loss": 0.7476,
      "step": 8409
    },
    {
      "epoch": 0.4628763278111068,
      "grad_norm": 0.8944914937019348,
      "learning_rate": 8.757638600891696e-06,
      "loss": 0.7814,
      "step": 8410
    },
    {
      "epoch": 0.4629313666134625,
      "grad_norm": 0.7242841124534607,
      "learning_rate": 8.757352628188411e-06,
      "loss": 0.7564,
      "step": 8411
    },
    {
      "epoch": 0.46298640541581815,
      "grad_norm": 0.6706324815750122,
      "learning_rate": 8.757066627245866e-06,
      "loss": 0.7792,
      "step": 8412
    },
    {
      "epoch": 0.46304144421817384,
      "grad_norm": 0.8044155836105347,
      "learning_rate": 8.756780598066218e-06,
      "loss": 0.7873,
      "step": 8413
    },
    {
      "epoch": 0.46309648302052947,
      "grad_norm": 0.9265295267105103,
      "learning_rate": 8.75649454065161e-06,
      "loss": 0.878,
      "step": 8414
    },
    {
      "epoch": 0.46315152182288516,
      "grad_norm": 0.8162378668785095,
      "learning_rate": 8.756208455004194e-06,
      "loss": 0.8758,
      "step": 8415
    },
    {
      "epoch": 0.4632065606252408,
      "grad_norm": 0.7081401348114014,
      "learning_rate": 8.755922341126121e-06,
      "loss": 0.8053,
      "step": 8416
    },
    {
      "epoch": 0.4632615994275965,
      "grad_norm": 0.663885235786438,
      "learning_rate": 8.755636199019544e-06,
      "loss": 0.7456,
      "step": 8417
    },
    {
      "epoch": 0.4633166382299521,
      "grad_norm": 0.6934974193572998,
      "learning_rate": 8.755350028686608e-06,
      "loss": 0.7316,
      "step": 8418
    },
    {
      "epoch": 0.46337167703230775,
      "grad_norm": 0.7162168025970459,
      "learning_rate": 8.755063830129467e-06,
      "loss": 0.8566,
      "step": 8419
    },
    {
      "epoch": 0.46342671583466344,
      "grad_norm": 0.7507640719413757,
      "learning_rate": 8.75477760335027e-06,
      "loss": 0.8141,
      "step": 8420
    },
    {
      "epoch": 0.46348175463701907,
      "grad_norm": 0.6853382587432861,
      "learning_rate": 8.754491348351172e-06,
      "loss": 0.6995,
      "step": 8421
    },
    {
      "epoch": 0.46353679343937476,
      "grad_norm": 0.6421381831169128,
      "learning_rate": 8.75420506513432e-06,
      "loss": 0.6344,
      "step": 8422
    },
    {
      "epoch": 0.4635918322417304,
      "grad_norm": 0.8042624592781067,
      "learning_rate": 8.753918753701868e-06,
      "loss": 0.7506,
      "step": 8423
    },
    {
      "epoch": 0.4636468710440861,
      "grad_norm": 0.7184088230133057,
      "learning_rate": 8.753632414055969e-06,
      "loss": 0.7997,
      "step": 8424
    },
    {
      "epoch": 0.4637019098464417,
      "grad_norm": 0.749919593334198,
      "learning_rate": 8.753346046198773e-06,
      "loss": 0.8168,
      "step": 8425
    },
    {
      "epoch": 0.4637569486487974,
      "grad_norm": 0.6583670973777771,
      "learning_rate": 8.753059650132433e-06,
      "loss": 0.6615,
      "step": 8426
    },
    {
      "epoch": 0.46381198745115304,
      "grad_norm": 0.7560496926307678,
      "learning_rate": 8.7527732258591e-06,
      "loss": 0.7221,
      "step": 8427
    },
    {
      "epoch": 0.46386702625350873,
      "grad_norm": 0.7031972408294678,
      "learning_rate": 8.752486773380928e-06,
      "loss": 0.8124,
      "step": 8428
    },
    {
      "epoch": 0.46392206505586436,
      "grad_norm": 0.684124767780304,
      "learning_rate": 8.752200292700072e-06,
      "loss": 0.6862,
      "step": 8429
    },
    {
      "epoch": 0.46397710385822005,
      "grad_norm": 0.8015589118003845,
      "learning_rate": 8.751913783818682e-06,
      "loss": 0.7863,
      "step": 8430
    },
    {
      "epoch": 0.4640321426605757,
      "grad_norm": 0.6815705299377441,
      "learning_rate": 8.751627246738912e-06,
      "loss": 0.8116,
      "step": 8431
    },
    {
      "epoch": 0.4640871814629314,
      "grad_norm": 0.7402058839797974,
      "learning_rate": 8.751340681462914e-06,
      "loss": 0.7341,
      "step": 8432
    },
    {
      "epoch": 0.464142220265287,
      "grad_norm": 0.7484470009803772,
      "learning_rate": 8.751054087992848e-06,
      "loss": 0.8103,
      "step": 8433
    },
    {
      "epoch": 0.4641972590676427,
      "grad_norm": 0.8148707151412964,
      "learning_rate": 8.75076746633086e-06,
      "loss": 0.8995,
      "step": 8434
    },
    {
      "epoch": 0.46425229786999833,
      "grad_norm": 0.6403086185455322,
      "learning_rate": 8.750480816479107e-06,
      "loss": 0.6705,
      "step": 8435
    },
    {
      "epoch": 0.464307336672354,
      "grad_norm": 0.7787690758705139,
      "learning_rate": 8.750194138439748e-06,
      "loss": 0.854,
      "step": 8436
    },
    {
      "epoch": 0.46436237547470965,
      "grad_norm": 0.6975393891334534,
      "learning_rate": 8.749907432214931e-06,
      "loss": 0.7588,
      "step": 8437
    },
    {
      "epoch": 0.46441741427706534,
      "grad_norm": 0.8002430200576782,
      "learning_rate": 8.749620697806812e-06,
      "loss": 0.8244,
      "step": 8438
    },
    {
      "epoch": 0.464472453079421,
      "grad_norm": 0.8049100637435913,
      "learning_rate": 8.74933393521755e-06,
      "loss": 0.7686,
      "step": 8439
    },
    {
      "epoch": 0.46452749188177667,
      "grad_norm": 0.6716971397399902,
      "learning_rate": 8.749047144449298e-06,
      "loss": 0.7823,
      "step": 8440
    },
    {
      "epoch": 0.4645825306841323,
      "grad_norm": 0.7292011380195618,
      "learning_rate": 8.748760325504212e-06,
      "loss": 0.7643,
      "step": 8441
    },
    {
      "epoch": 0.464637569486488,
      "grad_norm": 0.6823335886001587,
      "learning_rate": 8.748473478384444e-06,
      "loss": 0.7539,
      "step": 8442
    },
    {
      "epoch": 0.4646926082888436,
      "grad_norm": 0.761730968952179,
      "learning_rate": 8.748186603092155e-06,
      "loss": 0.7279,
      "step": 8443
    },
    {
      "epoch": 0.4647476470911993,
      "grad_norm": 0.694007933139801,
      "learning_rate": 8.747899699629498e-06,
      "loss": 0.7907,
      "step": 8444
    },
    {
      "epoch": 0.46480268589355495,
      "grad_norm": 0.7638683319091797,
      "learning_rate": 8.74761276799863e-06,
      "loss": 0.7278,
      "step": 8445
    },
    {
      "epoch": 0.46485772469591063,
      "grad_norm": 0.6281229853630066,
      "learning_rate": 8.747325808201708e-06,
      "loss": 0.6609,
      "step": 8446
    },
    {
      "epoch": 0.46491276349826627,
      "grad_norm": 0.7273259162902832,
      "learning_rate": 8.747038820240887e-06,
      "loss": 0.7553,
      "step": 8447
    },
    {
      "epoch": 0.46496780230062196,
      "grad_norm": 0.807482898235321,
      "learning_rate": 8.746751804118326e-06,
      "loss": 0.7783,
      "step": 8448
    },
    {
      "epoch": 0.4650228411029776,
      "grad_norm": 0.7088230848312378,
      "learning_rate": 8.746464759836182e-06,
      "loss": 0.762,
      "step": 8449
    },
    {
      "epoch": 0.4650778799053333,
      "grad_norm": 0.7039850354194641,
      "learning_rate": 8.746177687396612e-06,
      "loss": 0.7811,
      "step": 8450
    },
    {
      "epoch": 0.4651329187076889,
      "grad_norm": 0.7154161334037781,
      "learning_rate": 8.745890586801773e-06,
      "loss": 0.76,
      "step": 8451
    },
    {
      "epoch": 0.4651879575100446,
      "grad_norm": 0.6738846302032471,
      "learning_rate": 8.745603458053822e-06,
      "loss": 0.7119,
      "step": 8452
    },
    {
      "epoch": 0.46524299631240024,
      "grad_norm": 0.6615753173828125,
      "learning_rate": 8.745316301154919e-06,
      "loss": 0.8061,
      "step": 8453
    },
    {
      "epoch": 0.4652980351147559,
      "grad_norm": 0.7285076379776001,
      "learning_rate": 8.74502911610722e-06,
      "loss": 0.7522,
      "step": 8454
    },
    {
      "epoch": 0.46535307391711156,
      "grad_norm": 0.7100732922554016,
      "learning_rate": 8.744741902912886e-06,
      "loss": 0.7665,
      "step": 8455
    },
    {
      "epoch": 0.46540811271946725,
      "grad_norm": 0.6564487814903259,
      "learning_rate": 8.744454661574074e-06,
      "loss": 0.7352,
      "step": 8456
    },
    {
      "epoch": 0.4654631515218229,
      "grad_norm": 0.689549446105957,
      "learning_rate": 8.744167392092944e-06,
      "loss": 0.7011,
      "step": 8457
    },
    {
      "epoch": 0.46551819032417857,
      "grad_norm": 0.6660958528518677,
      "learning_rate": 8.743880094471651e-06,
      "loss": 0.7074,
      "step": 8458
    },
    {
      "epoch": 0.4655732291265342,
      "grad_norm": 0.7470804452896118,
      "learning_rate": 8.743592768712361e-06,
      "loss": 0.6684,
      "step": 8459
    },
    {
      "epoch": 0.4656282679288899,
      "grad_norm": 0.8058002591133118,
      "learning_rate": 8.743305414817227e-06,
      "loss": 0.7945,
      "step": 8460
    },
    {
      "epoch": 0.4656833067312455,
      "grad_norm": 0.7756261825561523,
      "learning_rate": 8.743018032788413e-06,
      "loss": 0.8442,
      "step": 8461
    },
    {
      "epoch": 0.46573834553360116,
      "grad_norm": 0.9267478585243225,
      "learning_rate": 8.742730622628077e-06,
      "loss": 0.8721,
      "step": 8462
    },
    {
      "epoch": 0.46579338433595685,
      "grad_norm": 0.8684219121932983,
      "learning_rate": 8.74244318433838e-06,
      "loss": 0.7833,
      "step": 8463
    },
    {
      "epoch": 0.4658484231383125,
      "grad_norm": 0.7060475945472717,
      "learning_rate": 8.742155717921481e-06,
      "loss": 0.7724,
      "step": 8464
    },
    {
      "epoch": 0.4659034619406682,
      "grad_norm": 0.7316318154335022,
      "learning_rate": 8.741868223379543e-06,
      "loss": 0.7489,
      "step": 8465
    },
    {
      "epoch": 0.4659585007430238,
      "grad_norm": 0.8131282925605774,
      "learning_rate": 8.741580700714724e-06,
      "loss": 0.7453,
      "step": 8466
    },
    {
      "epoch": 0.4660135395453795,
      "grad_norm": 0.6985850930213928,
      "learning_rate": 8.741293149929187e-06,
      "loss": 0.7083,
      "step": 8467
    },
    {
      "epoch": 0.46606857834773513,
      "grad_norm": 0.7512301206588745,
      "learning_rate": 8.74100557102509e-06,
      "loss": 0.7343,
      "step": 8468
    },
    {
      "epoch": 0.4661236171500908,
      "grad_norm": 0.7547290921211243,
      "learning_rate": 8.740717964004596e-06,
      "loss": 0.8358,
      "step": 8469
    },
    {
      "epoch": 0.46617865595244645,
      "grad_norm": 0.9091271758079529,
      "learning_rate": 8.740430328869868e-06,
      "loss": 0.762,
      "step": 8470
    },
    {
      "epoch": 0.46623369475480214,
      "grad_norm": 0.6960130333900452,
      "learning_rate": 8.740142665623069e-06,
      "loss": 0.7317,
      "step": 8471
    },
    {
      "epoch": 0.4662887335571578,
      "grad_norm": 0.684309184551239,
      "learning_rate": 8.739854974266357e-06,
      "loss": 0.7653,
      "step": 8472
    },
    {
      "epoch": 0.46634377235951346,
      "grad_norm": 0.7669411301612854,
      "learning_rate": 8.739567254801898e-06,
      "loss": 0.7152,
      "step": 8473
    },
    {
      "epoch": 0.4663988111618691,
      "grad_norm": 0.7072784900665283,
      "learning_rate": 8.73927950723185e-06,
      "loss": 0.7508,
      "step": 8474
    },
    {
      "epoch": 0.4664538499642248,
      "grad_norm": 0.7249277234077454,
      "learning_rate": 8.73899173155838e-06,
      "loss": 0.7469,
      "step": 8475
    },
    {
      "epoch": 0.4665088887665804,
      "grad_norm": 0.7664750218391418,
      "learning_rate": 8.738703927783647e-06,
      "loss": 0.8692,
      "step": 8476
    },
    {
      "epoch": 0.4665639275689361,
      "grad_norm": 0.7579765319824219,
      "learning_rate": 8.738416095909818e-06,
      "loss": 0.8283,
      "step": 8477
    },
    {
      "epoch": 0.46661896637129174,
      "grad_norm": 0.7066456079483032,
      "learning_rate": 8.738128235939054e-06,
      "loss": 0.7125,
      "step": 8478
    },
    {
      "epoch": 0.46667400517364743,
      "grad_norm": 0.766106367111206,
      "learning_rate": 8.737840347873518e-06,
      "loss": 0.7683,
      "step": 8479
    },
    {
      "epoch": 0.46672904397600307,
      "grad_norm": 0.7599226236343384,
      "learning_rate": 8.737552431715374e-06,
      "loss": 0.8375,
      "step": 8480
    },
    {
      "epoch": 0.46678408277835876,
      "grad_norm": 0.6955341100692749,
      "learning_rate": 8.737264487466789e-06,
      "loss": 0.7012,
      "step": 8481
    },
    {
      "epoch": 0.4668391215807144,
      "grad_norm": 0.6096246242523193,
      "learning_rate": 8.736976515129923e-06,
      "loss": 0.6126,
      "step": 8482
    },
    {
      "epoch": 0.4668941603830701,
      "grad_norm": 0.7469536066055298,
      "learning_rate": 8.73668851470694e-06,
      "loss": 0.7675,
      "step": 8483
    },
    {
      "epoch": 0.4669491991854257,
      "grad_norm": 0.8018775582313538,
      "learning_rate": 8.73640048620001e-06,
      "loss": 0.7372,
      "step": 8484
    },
    {
      "epoch": 0.4670042379877814,
      "grad_norm": 0.7446827292442322,
      "learning_rate": 8.736112429611293e-06,
      "loss": 0.7277,
      "step": 8485
    },
    {
      "epoch": 0.46705927679013703,
      "grad_norm": 0.6292026042938232,
      "learning_rate": 8.735824344942954e-06,
      "loss": 0.6172,
      "step": 8486
    },
    {
      "epoch": 0.4671143155924927,
      "grad_norm": 0.7207980751991272,
      "learning_rate": 8.735536232197159e-06,
      "loss": 0.8363,
      "step": 8487
    },
    {
      "epoch": 0.46716935439484836,
      "grad_norm": 0.8585891127586365,
      "learning_rate": 8.735248091376073e-06,
      "loss": 0.8006,
      "step": 8488
    },
    {
      "epoch": 0.46722439319720405,
      "grad_norm": 0.8149702548980713,
      "learning_rate": 8.734959922481863e-06,
      "loss": 0.7869,
      "step": 8489
    },
    {
      "epoch": 0.4672794319995597,
      "grad_norm": 0.7113268971443176,
      "learning_rate": 8.734671725516695e-06,
      "loss": 0.7774,
      "step": 8490
    },
    {
      "epoch": 0.46733447080191537,
      "grad_norm": 0.6940683722496033,
      "learning_rate": 8.734383500482733e-06,
      "loss": 0.7157,
      "step": 8491
    },
    {
      "epoch": 0.467389509604271,
      "grad_norm": 0.7823536396026611,
      "learning_rate": 8.734095247382145e-06,
      "loss": 0.8161,
      "step": 8492
    },
    {
      "epoch": 0.4674445484066267,
      "grad_norm": 0.7094922065734863,
      "learning_rate": 8.733806966217096e-06,
      "loss": 0.7593,
      "step": 8493
    },
    {
      "epoch": 0.4674995872089823,
      "grad_norm": 0.656432569026947,
      "learning_rate": 8.733518656989753e-06,
      "loss": 0.7853,
      "step": 8494
    },
    {
      "epoch": 0.467554626011338,
      "grad_norm": 0.6715715527534485,
      "learning_rate": 8.733230319702284e-06,
      "loss": 0.839,
      "step": 8495
    },
    {
      "epoch": 0.46760966481369365,
      "grad_norm": 0.7496705055236816,
      "learning_rate": 8.732941954356854e-06,
      "loss": 0.8231,
      "step": 8496
    },
    {
      "epoch": 0.46766470361604934,
      "grad_norm": 0.7728047370910645,
      "learning_rate": 8.732653560955635e-06,
      "loss": 0.7852,
      "step": 8497
    },
    {
      "epoch": 0.46771974241840497,
      "grad_norm": 1.5637458562850952,
      "learning_rate": 8.732365139500787e-06,
      "loss": 0.7749,
      "step": 8498
    },
    {
      "epoch": 0.46777478122076066,
      "grad_norm": 0.6603190898895264,
      "learning_rate": 8.732076689994484e-06,
      "loss": 0.6628,
      "step": 8499
    },
    {
      "epoch": 0.4678298200231163,
      "grad_norm": 0.7170974612236023,
      "learning_rate": 8.73178821243889e-06,
      "loss": 0.7855,
      "step": 8500
    },
    {
      "epoch": 0.467884858825472,
      "grad_norm": 0.7220103740692139,
      "learning_rate": 8.731499706836175e-06,
      "loss": 0.7035,
      "step": 8501
    },
    {
      "epoch": 0.4679398976278276,
      "grad_norm": 0.6940942406654358,
      "learning_rate": 8.731211173188507e-06,
      "loss": 0.7857,
      "step": 8502
    },
    {
      "epoch": 0.4679949364301833,
      "grad_norm": 2.441596508026123,
      "learning_rate": 8.730922611498057e-06,
      "loss": 0.695,
      "step": 8503
    },
    {
      "epoch": 0.46804997523253894,
      "grad_norm": 0.7654910087585449,
      "learning_rate": 8.730634021766989e-06,
      "loss": 0.788,
      "step": 8504
    },
    {
      "epoch": 0.4681050140348946,
      "grad_norm": 0.791824996471405,
      "learning_rate": 8.730345403997475e-06,
      "loss": 0.7899,
      "step": 8505
    },
    {
      "epoch": 0.46816005283725026,
      "grad_norm": 0.6863934993743896,
      "learning_rate": 8.730056758191682e-06,
      "loss": 0.7402,
      "step": 8506
    },
    {
      "epoch": 0.4682150916396059,
      "grad_norm": 0.7920359373092651,
      "learning_rate": 8.729768084351783e-06,
      "loss": 0.7835,
      "step": 8507
    },
    {
      "epoch": 0.4682701304419616,
      "grad_norm": 0.7077129483222961,
      "learning_rate": 8.729479382479944e-06,
      "loss": 0.7761,
      "step": 8508
    },
    {
      "epoch": 0.4683251692443172,
      "grad_norm": 0.6870049238204956,
      "learning_rate": 8.729190652578337e-06,
      "loss": 0.8169,
      "step": 8509
    },
    {
      "epoch": 0.4683802080466729,
      "grad_norm": 0.6802713871002197,
      "learning_rate": 8.728901894649131e-06,
      "loss": 0.7914,
      "step": 8510
    },
    {
      "epoch": 0.46843524684902854,
      "grad_norm": 0.6645112633705139,
      "learning_rate": 8.728613108694497e-06,
      "loss": 0.7543,
      "step": 8511
    },
    {
      "epoch": 0.46849028565138423,
      "grad_norm": 0.708292543888092,
      "learning_rate": 8.728324294716604e-06,
      "loss": 0.7015,
      "step": 8512
    },
    {
      "epoch": 0.46854532445373986,
      "grad_norm": 0.7444465160369873,
      "learning_rate": 8.728035452717625e-06,
      "loss": 0.7999,
      "step": 8513
    },
    {
      "epoch": 0.46860036325609555,
      "grad_norm": 0.7028616666793823,
      "learning_rate": 8.727746582699728e-06,
      "loss": 0.8094,
      "step": 8514
    },
    {
      "epoch": 0.4686554020584512,
      "grad_norm": 0.7063208222389221,
      "learning_rate": 8.727457684665088e-06,
      "loss": 0.8028,
      "step": 8515
    },
    {
      "epoch": 0.4687104408608069,
      "grad_norm": 0.8455138802528381,
      "learning_rate": 8.727168758615871e-06,
      "loss": 0.7691,
      "step": 8516
    },
    {
      "epoch": 0.4687654796631625,
      "grad_norm": 1.0325778722763062,
      "learning_rate": 8.726879804554252e-06,
      "loss": 0.7042,
      "step": 8517
    },
    {
      "epoch": 0.4688205184655182,
      "grad_norm": 0.7352754473686218,
      "learning_rate": 8.726590822482402e-06,
      "loss": 0.8467,
      "step": 8518
    },
    {
      "epoch": 0.46887555726787383,
      "grad_norm": 0.7247193455696106,
      "learning_rate": 8.726301812402494e-06,
      "loss": 0.8034,
      "step": 8519
    },
    {
      "epoch": 0.4689305960702295,
      "grad_norm": 0.6876820921897888,
      "learning_rate": 8.726012774316699e-06,
      "loss": 0.7308,
      "step": 8520
    },
    {
      "epoch": 0.46898563487258516,
      "grad_norm": 0.6987231969833374,
      "learning_rate": 8.725723708227188e-06,
      "loss": 0.7655,
      "step": 8521
    },
    {
      "epoch": 0.46904067367494084,
      "grad_norm": 0.7471843361854553,
      "learning_rate": 8.725434614136135e-06,
      "loss": 0.7271,
      "step": 8522
    },
    {
      "epoch": 0.4690957124772965,
      "grad_norm": 0.7564642429351807,
      "learning_rate": 8.725145492045715e-06,
      "loss": 0.7335,
      "step": 8523
    },
    {
      "epoch": 0.46915075127965217,
      "grad_norm": 0.7488992214202881,
      "learning_rate": 8.724856341958095e-06,
      "loss": 0.8815,
      "step": 8524
    },
    {
      "epoch": 0.4692057900820078,
      "grad_norm": 0.6776759028434753,
      "learning_rate": 8.724567163875455e-06,
      "loss": 0.7452,
      "step": 8525
    },
    {
      "epoch": 0.4692608288843635,
      "grad_norm": 0.6905981302261353,
      "learning_rate": 8.724277957799963e-06,
      "loss": 0.6815,
      "step": 8526
    },
    {
      "epoch": 0.4693158676867191,
      "grad_norm": 0.7392297983169556,
      "learning_rate": 8.723988723733795e-06,
      "loss": 0.7546,
      "step": 8527
    },
    {
      "epoch": 0.4693709064890748,
      "grad_norm": 0.7479110360145569,
      "learning_rate": 8.723699461679128e-06,
      "loss": 0.7455,
      "step": 8528
    },
    {
      "epoch": 0.46942594529143045,
      "grad_norm": 0.7231360673904419,
      "learning_rate": 8.723410171638129e-06,
      "loss": 0.7611,
      "step": 8529
    },
    {
      "epoch": 0.46948098409378614,
      "grad_norm": 0.7493714690208435,
      "learning_rate": 8.723120853612976e-06,
      "loss": 0.6997,
      "step": 8530
    },
    {
      "epoch": 0.46953602289614177,
      "grad_norm": 0.8056793808937073,
      "learning_rate": 8.722831507605844e-06,
      "loss": 0.7431,
      "step": 8531
    },
    {
      "epoch": 0.46959106169849746,
      "grad_norm": 0.7528547048568726,
      "learning_rate": 8.722542133618907e-06,
      "loss": 0.8798,
      "step": 8532
    },
    {
      "epoch": 0.4696461005008531,
      "grad_norm": 0.6964863538742065,
      "learning_rate": 8.72225273165434e-06,
      "loss": 0.8462,
      "step": 8533
    },
    {
      "epoch": 0.4697011393032088,
      "grad_norm": 0.7354302406311035,
      "learning_rate": 8.721963301714318e-06,
      "loss": 0.7882,
      "step": 8534
    },
    {
      "epoch": 0.4697561781055644,
      "grad_norm": 0.7365205883979797,
      "learning_rate": 8.721673843801014e-06,
      "loss": 0.7483,
      "step": 8535
    },
    {
      "epoch": 0.4698112169079201,
      "grad_norm": 0.7485378384590149,
      "learning_rate": 8.72138435791661e-06,
      "loss": 0.8539,
      "step": 8536
    },
    {
      "epoch": 0.46986625571027574,
      "grad_norm": 0.7674353718757629,
      "learning_rate": 8.721094844063274e-06,
      "loss": 0.834,
      "step": 8537
    },
    {
      "epoch": 0.4699212945126314,
      "grad_norm": 0.7054184079170227,
      "learning_rate": 8.720805302243185e-06,
      "loss": 0.7938,
      "step": 8538
    },
    {
      "epoch": 0.46997633331498706,
      "grad_norm": 0.7414574027061462,
      "learning_rate": 8.72051573245852e-06,
      "loss": 0.7932,
      "step": 8539
    },
    {
      "epoch": 0.47003137211734275,
      "grad_norm": 0.6734428405761719,
      "learning_rate": 8.720226134711455e-06,
      "loss": 0.8775,
      "step": 8540
    },
    {
      "epoch": 0.4700864109196984,
      "grad_norm": 0.6588559150695801,
      "learning_rate": 8.719936509004166e-06,
      "loss": 0.6985,
      "step": 8541
    },
    {
      "epoch": 0.4701414497220541,
      "grad_norm": 0.6557223200798035,
      "learning_rate": 8.71964685533883e-06,
      "loss": 0.7243,
      "step": 8542
    },
    {
      "epoch": 0.4701964885244097,
      "grad_norm": 0.7876269221305847,
      "learning_rate": 8.719357173717624e-06,
      "loss": 0.8075,
      "step": 8543
    },
    {
      "epoch": 0.4702515273267654,
      "grad_norm": 0.8346554040908813,
      "learning_rate": 8.719067464142726e-06,
      "loss": 0.8427,
      "step": 8544
    },
    {
      "epoch": 0.47030656612912103,
      "grad_norm": 0.7190483808517456,
      "learning_rate": 8.718777726616311e-06,
      "loss": 0.7689,
      "step": 8545
    },
    {
      "epoch": 0.4703616049314767,
      "grad_norm": 1.303118109703064,
      "learning_rate": 8.718487961140558e-06,
      "loss": 0.7537,
      "step": 8546
    },
    {
      "epoch": 0.47041664373383235,
      "grad_norm": 0.7733024954795837,
      "learning_rate": 8.718198167717647e-06,
      "loss": 0.747,
      "step": 8547
    },
    {
      "epoch": 0.470471682536188,
      "grad_norm": 0.6692484617233276,
      "learning_rate": 8.717908346349751e-06,
      "loss": 0.725,
      "step": 8548
    },
    {
      "epoch": 0.4705267213385437,
      "grad_norm": 0.9639461636543274,
      "learning_rate": 8.717618497039054e-06,
      "loss": 0.8642,
      "step": 8549
    },
    {
      "epoch": 0.4705817601408993,
      "grad_norm": 0.7584646344184875,
      "learning_rate": 8.717328619787728e-06,
      "loss": 0.8174,
      "step": 8550
    },
    {
      "epoch": 0.470636798943255,
      "grad_norm": 0.7051709890365601,
      "learning_rate": 8.717038714597957e-06,
      "loss": 0.7962,
      "step": 8551
    },
    {
      "epoch": 0.47069183774561063,
      "grad_norm": 0.738913893699646,
      "learning_rate": 8.716748781471918e-06,
      "loss": 0.7367,
      "step": 8552
    },
    {
      "epoch": 0.4707468765479663,
      "grad_norm": 0.7027214169502258,
      "learning_rate": 8.716458820411791e-06,
      "loss": 0.7613,
      "step": 8553
    },
    {
      "epoch": 0.47080191535032195,
      "grad_norm": 0.6701993346214294,
      "learning_rate": 8.716168831419754e-06,
      "loss": 0.638,
      "step": 8554
    },
    {
      "epoch": 0.47085695415267764,
      "grad_norm": 0.7422072887420654,
      "learning_rate": 8.715878814497984e-06,
      "loss": 0.8338,
      "step": 8555
    },
    {
      "epoch": 0.4709119929550333,
      "grad_norm": 0.985992968082428,
      "learning_rate": 8.715588769648667e-06,
      "loss": 0.7765,
      "step": 8556
    },
    {
      "epoch": 0.47096703175738897,
      "grad_norm": 0.6937553882598877,
      "learning_rate": 8.715298696873978e-06,
      "loss": 0.7306,
      "step": 8557
    },
    {
      "epoch": 0.4710220705597446,
      "grad_norm": 1.1683214902877808,
      "learning_rate": 8.715008596176099e-06,
      "loss": 0.7782,
      "step": 8558
    },
    {
      "epoch": 0.4710771093621003,
      "grad_norm": 0.7493681907653809,
      "learning_rate": 8.714718467557209e-06,
      "loss": 0.9166,
      "step": 8559
    },
    {
      "epoch": 0.4711321481644559,
      "grad_norm": 0.7562084794044495,
      "learning_rate": 8.71442831101949e-06,
      "loss": 0.7999,
      "step": 8560
    },
    {
      "epoch": 0.4711871869668116,
      "grad_norm": 0.7950266003608704,
      "learning_rate": 8.71413812656512e-06,
      "loss": 0.8094,
      "step": 8561
    },
    {
      "epoch": 0.47124222576916724,
      "grad_norm": 1.1411044597625732,
      "learning_rate": 8.713847914196287e-06,
      "loss": 0.7631,
      "step": 8562
    },
    {
      "epoch": 0.47129726457152293,
      "grad_norm": 0.7270122170448303,
      "learning_rate": 8.713557673915162e-06,
      "loss": 0.7529,
      "step": 8563
    },
    {
      "epoch": 0.47135230337387857,
      "grad_norm": 0.8138573169708252,
      "learning_rate": 8.713267405723935e-06,
      "loss": 0.8215,
      "step": 8564
    },
    {
      "epoch": 0.47140734217623426,
      "grad_norm": 0.732982873916626,
      "learning_rate": 8.712977109624783e-06,
      "loss": 0.7099,
      "step": 8565
    },
    {
      "epoch": 0.4714623809785899,
      "grad_norm": 0.7307591438293457,
      "learning_rate": 8.712686785619888e-06,
      "loss": 0.7035,
      "step": 8566
    },
    {
      "epoch": 0.4715174197809456,
      "grad_norm": 0.8684857487678528,
      "learning_rate": 8.712396433711434e-06,
      "loss": 0.8605,
      "step": 8567
    },
    {
      "epoch": 0.4715724585833012,
      "grad_norm": 0.7490718364715576,
      "learning_rate": 8.712106053901603e-06,
      "loss": 0.7439,
      "step": 8568
    },
    {
      "epoch": 0.4716274973856569,
      "grad_norm": 0.8572973012924194,
      "learning_rate": 8.711815646192575e-06,
      "loss": 0.8187,
      "step": 8569
    },
    {
      "epoch": 0.47168253618801254,
      "grad_norm": 0.785270094871521,
      "learning_rate": 8.711525210586536e-06,
      "loss": 0.7812,
      "step": 8570
    },
    {
      "epoch": 0.4717375749903682,
      "grad_norm": 0.683651864528656,
      "learning_rate": 8.711234747085663e-06,
      "loss": 0.7682,
      "step": 8571
    },
    {
      "epoch": 0.47179261379272386,
      "grad_norm": 0.7990714907646179,
      "learning_rate": 8.710944255692147e-06,
      "loss": 0.8114,
      "step": 8572
    },
    {
      "epoch": 0.47184765259507955,
      "grad_norm": 0.9354856610298157,
      "learning_rate": 8.710653736408165e-06,
      "loss": 0.7353,
      "step": 8573
    },
    {
      "epoch": 0.4719026913974352,
      "grad_norm": 0.8309356570243835,
      "learning_rate": 8.710363189235904e-06,
      "loss": 0.8635,
      "step": 8574
    },
    {
      "epoch": 0.47195773019979087,
      "grad_norm": 0.7018463015556335,
      "learning_rate": 8.710072614177547e-06,
      "loss": 0.6372,
      "step": 8575
    },
    {
      "epoch": 0.4720127690021465,
      "grad_norm": 0.7626469135284424,
      "learning_rate": 8.709782011235277e-06,
      "loss": 0.7684,
      "step": 8576
    },
    {
      "epoch": 0.4720678078045022,
      "grad_norm": 0.6995826959609985,
      "learning_rate": 8.70949138041128e-06,
      "loss": 0.7301,
      "step": 8577
    },
    {
      "epoch": 0.4721228466068578,
      "grad_norm": 0.719307541847229,
      "learning_rate": 8.709200721707736e-06,
      "loss": 0.7437,
      "step": 8578
    },
    {
      "epoch": 0.4721778854092135,
      "grad_norm": 0.7355539202690125,
      "learning_rate": 8.708910035126832e-06,
      "loss": 0.7926,
      "step": 8579
    },
    {
      "epoch": 0.47223292421156915,
      "grad_norm": 0.7262680530548096,
      "learning_rate": 8.708619320670755e-06,
      "loss": 0.7641,
      "step": 8580
    },
    {
      "epoch": 0.47228796301392484,
      "grad_norm": 0.844745934009552,
      "learning_rate": 8.708328578341687e-06,
      "loss": 0.7228,
      "step": 8581
    },
    {
      "epoch": 0.47234300181628047,
      "grad_norm": 0.8169287443161011,
      "learning_rate": 8.708037808141814e-06,
      "loss": 0.7076,
      "step": 8582
    },
    {
      "epoch": 0.47239804061863616,
      "grad_norm": 0.7342209219932556,
      "learning_rate": 8.707747010073322e-06,
      "loss": 0.7997,
      "step": 8583
    },
    {
      "epoch": 0.4724530794209918,
      "grad_norm": 0.7138200402259827,
      "learning_rate": 8.707456184138394e-06,
      "loss": 0.7796,
      "step": 8584
    },
    {
      "epoch": 0.4725081182233475,
      "grad_norm": 0.7168061137199402,
      "learning_rate": 8.70716533033922e-06,
      "loss": 0.6876,
      "step": 8585
    },
    {
      "epoch": 0.4725631570257031,
      "grad_norm": 0.7256397604942322,
      "learning_rate": 8.706874448677982e-06,
      "loss": 0.8296,
      "step": 8586
    },
    {
      "epoch": 0.4726181958280588,
      "grad_norm": 0.8232730627059937,
      "learning_rate": 8.70658353915687e-06,
      "loss": 0.8001,
      "step": 8587
    },
    {
      "epoch": 0.47267323463041444,
      "grad_norm": 0.7110162973403931,
      "learning_rate": 8.706292601778067e-06,
      "loss": 0.7061,
      "step": 8588
    },
    {
      "epoch": 0.47272827343277013,
      "grad_norm": 0.9466721415519714,
      "learning_rate": 8.706001636543761e-06,
      "loss": 0.8713,
      "step": 8589
    },
    {
      "epoch": 0.47278331223512576,
      "grad_norm": 0.7017776370048523,
      "learning_rate": 8.705710643456138e-06,
      "loss": 0.759,
      "step": 8590
    },
    {
      "epoch": 0.4728383510374814,
      "grad_norm": 0.7140772938728333,
      "learning_rate": 8.705419622517386e-06,
      "loss": 0.6962,
      "step": 8591
    },
    {
      "epoch": 0.4728933898398371,
      "grad_norm": 1.1076452732086182,
      "learning_rate": 8.705128573729694e-06,
      "loss": 0.8264,
      "step": 8592
    },
    {
      "epoch": 0.4729484286421927,
      "grad_norm": 0.7308200597763062,
      "learning_rate": 8.704837497095247e-06,
      "loss": 0.6243,
      "step": 8593
    },
    {
      "epoch": 0.4730034674445484,
      "grad_norm": 0.9445781111717224,
      "learning_rate": 8.704546392616231e-06,
      "loss": 0.6676,
      "step": 8594
    },
    {
      "epoch": 0.47305850624690404,
      "grad_norm": 0.6527873277664185,
      "learning_rate": 8.704255260294837e-06,
      "loss": 0.6979,
      "step": 8595
    },
    {
      "epoch": 0.47311354504925973,
      "grad_norm": 0.6732963919639587,
      "learning_rate": 8.703964100133252e-06,
      "loss": 0.7724,
      "step": 8596
    },
    {
      "epoch": 0.47316858385161537,
      "grad_norm": 0.7661726474761963,
      "learning_rate": 8.703672912133665e-06,
      "loss": 0.7988,
      "step": 8597
    },
    {
      "epoch": 0.47322362265397105,
      "grad_norm": 0.7006877660751343,
      "learning_rate": 8.703381696298262e-06,
      "loss": 0.6765,
      "step": 8598
    },
    {
      "epoch": 0.4732786614563267,
      "grad_norm": 0.7195086479187012,
      "learning_rate": 8.703090452629236e-06,
      "loss": 0.6676,
      "step": 8599
    },
    {
      "epoch": 0.4733337002586824,
      "grad_norm": 0.6692042350769043,
      "learning_rate": 8.702799181128771e-06,
      "loss": 0.7882,
      "step": 8600
    },
    {
      "epoch": 0.473388739061038,
      "grad_norm": 0.7736524343490601,
      "learning_rate": 8.70250788179906e-06,
      "loss": 0.7977,
      "step": 8601
    },
    {
      "epoch": 0.4734437778633937,
      "grad_norm": 0.8821607828140259,
      "learning_rate": 8.70221655464229e-06,
      "loss": 0.7465,
      "step": 8602
    },
    {
      "epoch": 0.47349881666574933,
      "grad_norm": 0.7565156817436218,
      "learning_rate": 8.701925199660652e-06,
      "loss": 0.831,
      "step": 8603
    },
    {
      "epoch": 0.473553855468105,
      "grad_norm": 0.8542304039001465,
      "learning_rate": 8.701633816856335e-06,
      "loss": 0.7538,
      "step": 8604
    },
    {
      "epoch": 0.47360889427046066,
      "grad_norm": 0.6891050338745117,
      "learning_rate": 8.701342406231529e-06,
      "loss": 0.7687,
      "step": 8605
    },
    {
      "epoch": 0.47366393307281635,
      "grad_norm": 0.8570719361305237,
      "learning_rate": 8.701050967788424e-06,
      "loss": 0.7236,
      "step": 8606
    },
    {
      "epoch": 0.473718971875172,
      "grad_norm": 0.7921456098556519,
      "learning_rate": 8.700759501529212e-06,
      "loss": 0.8214,
      "step": 8607
    },
    {
      "epoch": 0.47377401067752767,
      "grad_norm": 0.7584527730941772,
      "learning_rate": 8.70046800745608e-06,
      "loss": 0.8204,
      "step": 8608
    },
    {
      "epoch": 0.4738290494798833,
      "grad_norm": 0.8033978343009949,
      "learning_rate": 8.700176485571222e-06,
      "loss": 0.8278,
      "step": 8609
    },
    {
      "epoch": 0.473884088282239,
      "grad_norm": 0.9950750470161438,
      "learning_rate": 8.699884935876828e-06,
      "loss": 0.8181,
      "step": 8610
    },
    {
      "epoch": 0.4739391270845946,
      "grad_norm": 0.7213684916496277,
      "learning_rate": 8.69959335837509e-06,
      "loss": 0.7099,
      "step": 8611
    },
    {
      "epoch": 0.4739941658869503,
      "grad_norm": 0.7847200632095337,
      "learning_rate": 8.699301753068199e-06,
      "loss": 0.8272,
      "step": 8612
    },
    {
      "epoch": 0.47404920468930595,
      "grad_norm": 0.7075058221817017,
      "learning_rate": 8.699010119958344e-06,
      "loss": 0.7127,
      "step": 8613
    },
    {
      "epoch": 0.47410424349166164,
      "grad_norm": 0.682741641998291,
      "learning_rate": 8.69871845904772e-06,
      "loss": 0.8446,
      "step": 8614
    },
    {
      "epoch": 0.47415928229401727,
      "grad_norm": 0.7120605111122131,
      "learning_rate": 8.69842677033852e-06,
      "loss": 0.7776,
      "step": 8615
    },
    {
      "epoch": 0.47421432109637296,
      "grad_norm": 0.822405219078064,
      "learning_rate": 8.698135053832933e-06,
      "loss": 0.8018,
      "step": 8616
    },
    {
      "epoch": 0.4742693598987286,
      "grad_norm": 0.6815186738967896,
      "learning_rate": 8.697843309533152e-06,
      "loss": 0.7413,
      "step": 8617
    },
    {
      "epoch": 0.4743243987010843,
      "grad_norm": 0.7587849497795105,
      "learning_rate": 8.69755153744137e-06,
      "loss": 0.7809,
      "step": 8618
    },
    {
      "epoch": 0.4743794375034399,
      "grad_norm": 0.7092488408088684,
      "learning_rate": 8.697259737559782e-06,
      "loss": 0.7921,
      "step": 8619
    },
    {
      "epoch": 0.4744344763057956,
      "grad_norm": 0.7396836280822754,
      "learning_rate": 8.69696790989058e-06,
      "loss": 0.7946,
      "step": 8620
    },
    {
      "epoch": 0.47448951510815124,
      "grad_norm": 0.6760729551315308,
      "learning_rate": 8.696676054435955e-06,
      "loss": 0.7389,
      "step": 8621
    },
    {
      "epoch": 0.4745445539105069,
      "grad_norm": 1.1640692949295044,
      "learning_rate": 8.696384171198105e-06,
      "loss": 0.8291,
      "step": 8622
    },
    {
      "epoch": 0.47459959271286256,
      "grad_norm": 0.7415158152580261,
      "learning_rate": 8.696092260179219e-06,
      "loss": 0.7534,
      "step": 8623
    },
    {
      "epoch": 0.47465463151521825,
      "grad_norm": 0.7730052471160889,
      "learning_rate": 8.695800321381492e-06,
      "loss": 0.8447,
      "step": 8624
    },
    {
      "epoch": 0.4747096703175739,
      "grad_norm": 0.811522364616394,
      "learning_rate": 8.695508354807121e-06,
      "loss": 0.7466,
      "step": 8625
    },
    {
      "epoch": 0.4747647091199296,
      "grad_norm": 0.7908332347869873,
      "learning_rate": 8.695216360458298e-06,
      "loss": 0.7769,
      "step": 8626
    },
    {
      "epoch": 0.4748197479222852,
      "grad_norm": 0.744971752166748,
      "learning_rate": 8.694924338337217e-06,
      "loss": 0.7651,
      "step": 8627
    },
    {
      "epoch": 0.4748747867246409,
      "grad_norm": 0.705565869808197,
      "learning_rate": 8.694632288446075e-06,
      "loss": 0.8258,
      "step": 8628
    },
    {
      "epoch": 0.47492982552699653,
      "grad_norm": 0.8199328780174255,
      "learning_rate": 8.694340210787065e-06,
      "loss": 0.733,
      "step": 8629
    },
    {
      "epoch": 0.4749848643293522,
      "grad_norm": 0.6965511441230774,
      "learning_rate": 8.694048105362382e-06,
      "loss": 0.7548,
      "step": 8630
    },
    {
      "epoch": 0.47503990313170785,
      "grad_norm": 0.7943055629730225,
      "learning_rate": 8.693755972174225e-06,
      "loss": 0.7518,
      "step": 8631
    },
    {
      "epoch": 0.47509494193406354,
      "grad_norm": 0.6277437806129456,
      "learning_rate": 8.693463811224785e-06,
      "loss": 0.6941,
      "step": 8632
    },
    {
      "epoch": 0.4751499807364192,
      "grad_norm": 1.0745574235916138,
      "learning_rate": 8.693171622516259e-06,
      "loss": 0.8056,
      "step": 8633
    },
    {
      "epoch": 0.4752050195387748,
      "grad_norm": 0.7005153894424438,
      "learning_rate": 8.692879406050844e-06,
      "loss": 0.757,
      "step": 8634
    },
    {
      "epoch": 0.4752600583411305,
      "grad_norm": 0.6971127986907959,
      "learning_rate": 8.692587161830737e-06,
      "loss": 0.7509,
      "step": 8635
    },
    {
      "epoch": 0.47531509714348613,
      "grad_norm": 0.7583497762680054,
      "learning_rate": 8.692294889858133e-06,
      "loss": 0.7895,
      "step": 8636
    },
    {
      "epoch": 0.4753701359458418,
      "grad_norm": 0.719932496547699,
      "learning_rate": 8.692002590135228e-06,
      "loss": 0.762,
      "step": 8637
    },
    {
      "epoch": 0.47542517474819745,
      "grad_norm": 0.7041804790496826,
      "learning_rate": 8.691710262664222e-06,
      "loss": 0.7101,
      "step": 8638
    },
    {
      "epoch": 0.47548021355055314,
      "grad_norm": 0.7395016551017761,
      "learning_rate": 8.691417907447309e-06,
      "loss": 0.723,
      "step": 8639
    },
    {
      "epoch": 0.4755352523529088,
      "grad_norm": 0.6605637073516846,
      "learning_rate": 8.691125524486686e-06,
      "loss": 0.644,
      "step": 8640
    },
    {
      "epoch": 0.47559029115526447,
      "grad_norm": 0.694732129573822,
      "learning_rate": 8.690833113784552e-06,
      "loss": 0.7162,
      "step": 8641
    },
    {
      "epoch": 0.4756453299576201,
      "grad_norm": 0.7622451186180115,
      "learning_rate": 8.690540675343105e-06,
      "loss": 0.6995,
      "step": 8642
    },
    {
      "epoch": 0.4757003687599758,
      "grad_norm": 0.6961628794670105,
      "learning_rate": 8.69024820916454e-06,
      "loss": 0.7955,
      "step": 8643
    },
    {
      "epoch": 0.4757554075623314,
      "grad_norm": 0.706266462802887,
      "learning_rate": 8.68995571525106e-06,
      "loss": 0.7237,
      "step": 8644
    },
    {
      "epoch": 0.4758104463646871,
      "grad_norm": 0.7727495431900024,
      "learning_rate": 8.689663193604858e-06,
      "loss": 0.7215,
      "step": 8645
    },
    {
      "epoch": 0.47586548516704275,
      "grad_norm": 0.7320648431777954,
      "learning_rate": 8.689370644228136e-06,
      "loss": 0.7592,
      "step": 8646
    },
    {
      "epoch": 0.47592052396939843,
      "grad_norm": 0.8149487376213074,
      "learning_rate": 8.689078067123093e-06,
      "loss": 0.7666,
      "step": 8647
    },
    {
      "epoch": 0.47597556277175407,
      "grad_norm": 0.6584552526473999,
      "learning_rate": 8.688785462291927e-06,
      "loss": 0.7497,
      "step": 8648
    },
    {
      "epoch": 0.47603060157410976,
      "grad_norm": 0.7197825312614441,
      "learning_rate": 8.688492829736836e-06,
      "loss": 0.7559,
      "step": 8649
    },
    {
      "epoch": 0.4760856403764654,
      "grad_norm": 0.8116913437843323,
      "learning_rate": 8.68820016946002e-06,
      "loss": 0.7029,
      "step": 8650
    },
    {
      "epoch": 0.4761406791788211,
      "grad_norm": 0.6733378171920776,
      "learning_rate": 8.68790748146368e-06,
      "loss": 0.7242,
      "step": 8651
    },
    {
      "epoch": 0.4761957179811767,
      "grad_norm": 0.690464437007904,
      "learning_rate": 8.687614765750012e-06,
      "loss": 0.6668,
      "step": 8652
    },
    {
      "epoch": 0.4762507567835324,
      "grad_norm": 0.7901185154914856,
      "learning_rate": 8.687322022321221e-06,
      "loss": 0.7436,
      "step": 8653
    },
    {
      "epoch": 0.47630579558588804,
      "grad_norm": 0.7608267068862915,
      "learning_rate": 8.687029251179504e-06,
      "loss": 0.8292,
      "step": 8654
    },
    {
      "epoch": 0.4763608343882437,
      "grad_norm": 0.6851119995117188,
      "learning_rate": 8.686736452327062e-06,
      "loss": 0.7974,
      "step": 8655
    },
    {
      "epoch": 0.47641587319059936,
      "grad_norm": 0.6946395635604858,
      "learning_rate": 8.686443625766094e-06,
      "loss": 0.6745,
      "step": 8656
    },
    {
      "epoch": 0.47647091199295505,
      "grad_norm": 0.7403521537780762,
      "learning_rate": 8.686150771498804e-06,
      "loss": 0.7759,
      "step": 8657
    },
    {
      "epoch": 0.4765259507953107,
      "grad_norm": 0.8415689468383789,
      "learning_rate": 8.685857889527393e-06,
      "loss": 0.7911,
      "step": 8658
    },
    {
      "epoch": 0.47658098959766637,
      "grad_norm": 0.6947778463363647,
      "learning_rate": 8.68556497985406e-06,
      "loss": 0.8026,
      "step": 8659
    },
    {
      "epoch": 0.476636028400022,
      "grad_norm": 0.6807059645652771,
      "learning_rate": 8.685272042481006e-06,
      "loss": 0.7194,
      "step": 8660
    },
    {
      "epoch": 0.4766910672023777,
      "grad_norm": 0.8948639631271362,
      "learning_rate": 8.684979077410434e-06,
      "loss": 0.8017,
      "step": 8661
    },
    {
      "epoch": 0.4767461060047333,
      "grad_norm": 0.6697849035263062,
      "learning_rate": 8.684686084644546e-06,
      "loss": 0.7653,
      "step": 8662
    },
    {
      "epoch": 0.476801144807089,
      "grad_norm": 0.7303311228752136,
      "learning_rate": 8.684393064185543e-06,
      "loss": 0.8287,
      "step": 8663
    },
    {
      "epoch": 0.47685618360944465,
      "grad_norm": 0.6545100808143616,
      "learning_rate": 8.68410001603563e-06,
      "loss": 0.7438,
      "step": 8664
    },
    {
      "epoch": 0.47691122241180034,
      "grad_norm": 0.8757766485214233,
      "learning_rate": 8.683806940197006e-06,
      "loss": 0.8343,
      "step": 8665
    },
    {
      "epoch": 0.476966261214156,
      "grad_norm": 0.6414330005645752,
      "learning_rate": 8.683513836671876e-06,
      "loss": 0.7201,
      "step": 8666
    },
    {
      "epoch": 0.47702130001651166,
      "grad_norm": 0.6736441850662231,
      "learning_rate": 8.68322070546244e-06,
      "loss": 0.7365,
      "step": 8667
    },
    {
      "epoch": 0.4770763388188673,
      "grad_norm": 0.780491054058075,
      "learning_rate": 8.682927546570905e-06,
      "loss": 0.924,
      "step": 8668
    },
    {
      "epoch": 0.477131377621223,
      "grad_norm": 0.6913807988166809,
      "learning_rate": 8.68263435999947e-06,
      "loss": 0.8269,
      "step": 8669
    },
    {
      "epoch": 0.4771864164235786,
      "grad_norm": 0.7264360189437866,
      "learning_rate": 8.682341145750344e-06,
      "loss": 0.788,
      "step": 8670
    },
    {
      "epoch": 0.4772414552259343,
      "grad_norm": 0.7777243852615356,
      "learning_rate": 8.682047903825725e-06,
      "loss": 0.8691,
      "step": 8671
    },
    {
      "epoch": 0.47729649402828994,
      "grad_norm": 0.7590457797050476,
      "learning_rate": 8.681754634227821e-06,
      "loss": 0.8249,
      "step": 8672
    },
    {
      "epoch": 0.47735153283064563,
      "grad_norm": 0.7672324776649475,
      "learning_rate": 8.681461336958836e-06,
      "loss": 0.8334,
      "step": 8673
    },
    {
      "epoch": 0.47740657163300126,
      "grad_norm": 0.7181395888328552,
      "learning_rate": 8.681168012020971e-06,
      "loss": 0.8089,
      "step": 8674
    },
    {
      "epoch": 0.47746161043535695,
      "grad_norm": 0.7671428918838501,
      "learning_rate": 8.680874659416433e-06,
      "loss": 0.7634,
      "step": 8675
    },
    {
      "epoch": 0.4775166492377126,
      "grad_norm": 0.73219895362854,
      "learning_rate": 8.680581279147427e-06,
      "loss": 0.7013,
      "step": 8676
    },
    {
      "epoch": 0.4775716880400682,
      "grad_norm": 0.8050867319107056,
      "learning_rate": 8.680287871216158e-06,
      "loss": 0.7524,
      "step": 8677
    },
    {
      "epoch": 0.4776267268424239,
      "grad_norm": 0.7154340744018555,
      "learning_rate": 8.679994435624828e-06,
      "loss": 0.802,
      "step": 8678
    },
    {
      "epoch": 0.47768176564477954,
      "grad_norm": 0.7005884051322937,
      "learning_rate": 8.679700972375647e-06,
      "loss": 0.7633,
      "step": 8679
    },
    {
      "epoch": 0.47773680444713523,
      "grad_norm": 0.8203871846199036,
      "learning_rate": 8.679407481470818e-06,
      "loss": 0.7782,
      "step": 8680
    },
    {
      "epoch": 0.47779184324949087,
      "grad_norm": 0.6582844853401184,
      "learning_rate": 8.679113962912547e-06,
      "loss": 0.6799,
      "step": 8681
    },
    {
      "epoch": 0.47784688205184656,
      "grad_norm": 0.7052889466285706,
      "learning_rate": 8.67882041670304e-06,
      "loss": 0.7814,
      "step": 8682
    },
    {
      "epoch": 0.4779019208542022,
      "grad_norm": 0.7533165812492371,
      "learning_rate": 8.678526842844504e-06,
      "loss": 0.7983,
      "step": 8683
    },
    {
      "epoch": 0.4779569596565579,
      "grad_norm": 0.7335212230682373,
      "learning_rate": 8.678233241339144e-06,
      "loss": 0.8023,
      "step": 8684
    },
    {
      "epoch": 0.4780119984589135,
      "grad_norm": 0.7824274897575378,
      "learning_rate": 8.67793961218917e-06,
      "loss": 0.8219,
      "step": 8685
    },
    {
      "epoch": 0.4780670372612692,
      "grad_norm": 0.6547996401786804,
      "learning_rate": 8.677645955396784e-06,
      "loss": 0.715,
      "step": 8686
    },
    {
      "epoch": 0.47812207606362483,
      "grad_norm": 0.7507368326187134,
      "learning_rate": 8.677352270964196e-06,
      "loss": 0.9379,
      "step": 8687
    },
    {
      "epoch": 0.4781771148659805,
      "grad_norm": 0.6403020620346069,
      "learning_rate": 8.677058558893613e-06,
      "loss": 0.659,
      "step": 8688
    },
    {
      "epoch": 0.47823215366833616,
      "grad_norm": 0.7075803279876709,
      "learning_rate": 8.676764819187242e-06,
      "loss": 0.7515,
      "step": 8689
    },
    {
      "epoch": 0.47828719247069185,
      "grad_norm": 0.6899601817131042,
      "learning_rate": 8.676471051847291e-06,
      "loss": 0.8398,
      "step": 8690
    },
    {
      "epoch": 0.4783422312730475,
      "grad_norm": 0.7145645618438721,
      "learning_rate": 8.676177256875969e-06,
      "loss": 0.7711,
      "step": 8691
    },
    {
      "epoch": 0.47839727007540317,
      "grad_norm": 0.7139655351638794,
      "learning_rate": 8.675883434275479e-06,
      "loss": 0.8664,
      "step": 8692
    },
    {
      "epoch": 0.4784523088777588,
      "grad_norm": 0.7100433111190796,
      "learning_rate": 8.675589584048037e-06,
      "loss": 0.7812,
      "step": 8693
    },
    {
      "epoch": 0.4785073476801145,
      "grad_norm": 0.6103882789611816,
      "learning_rate": 8.675295706195845e-06,
      "loss": 0.6565,
      "step": 8694
    },
    {
      "epoch": 0.4785623864824701,
      "grad_norm": 0.7236714959144592,
      "learning_rate": 8.675001800721114e-06,
      "loss": 0.6849,
      "step": 8695
    },
    {
      "epoch": 0.4786174252848258,
      "grad_norm": 0.7567160129547119,
      "learning_rate": 8.674707867626056e-06,
      "loss": 0.8289,
      "step": 8696
    },
    {
      "epoch": 0.47867246408718145,
      "grad_norm": 0.7004136443138123,
      "learning_rate": 8.674413906912876e-06,
      "loss": 0.7466,
      "step": 8697
    },
    {
      "epoch": 0.47872750288953714,
      "grad_norm": 0.713835597038269,
      "learning_rate": 8.674119918583783e-06,
      "loss": 0.7875,
      "step": 8698
    },
    {
      "epoch": 0.47878254169189277,
      "grad_norm": 0.8476874232292175,
      "learning_rate": 8.67382590264099e-06,
      "loss": 0.8028,
      "step": 8699
    },
    {
      "epoch": 0.47883758049424846,
      "grad_norm": 0.720273494720459,
      "learning_rate": 8.673531859086706e-06,
      "loss": 0.7829,
      "step": 8700
    },
    {
      "epoch": 0.4788926192966041,
      "grad_norm": 0.8042417168617249,
      "learning_rate": 8.673237787923137e-06,
      "loss": 0.7914,
      "step": 8701
    },
    {
      "epoch": 0.4789476580989598,
      "grad_norm": 0.7779260277748108,
      "learning_rate": 8.672943689152498e-06,
      "loss": 0.6921,
      "step": 8702
    },
    {
      "epoch": 0.4790026969013154,
      "grad_norm": 0.7957637906074524,
      "learning_rate": 8.672649562776997e-06,
      "loss": 0.8761,
      "step": 8703
    },
    {
      "epoch": 0.4790577357036711,
      "grad_norm": 0.7467649579048157,
      "learning_rate": 8.672355408798845e-06,
      "loss": 0.7984,
      "step": 8704
    },
    {
      "epoch": 0.47911277450602674,
      "grad_norm": 0.6746538877487183,
      "learning_rate": 8.672061227220252e-06,
      "loss": 0.7392,
      "step": 8705
    },
    {
      "epoch": 0.47916781330838243,
      "grad_norm": 0.7331795692443848,
      "learning_rate": 8.671767018043432e-06,
      "loss": 0.7171,
      "step": 8706
    },
    {
      "epoch": 0.47922285211073806,
      "grad_norm": 0.7879608273506165,
      "learning_rate": 8.671472781270592e-06,
      "loss": 0.8497,
      "step": 8707
    },
    {
      "epoch": 0.47927789091309375,
      "grad_norm": 0.8659428358078003,
      "learning_rate": 8.671178516903946e-06,
      "loss": 0.8102,
      "step": 8708
    },
    {
      "epoch": 0.4793329297154494,
      "grad_norm": 0.6489408612251282,
      "learning_rate": 8.670884224945704e-06,
      "loss": 0.6752,
      "step": 8709
    },
    {
      "epoch": 0.4793879685178051,
      "grad_norm": 0.8182825446128845,
      "learning_rate": 8.670589905398079e-06,
      "loss": 0.7972,
      "step": 8710
    },
    {
      "epoch": 0.4794430073201607,
      "grad_norm": 0.7759343981742859,
      "learning_rate": 8.670295558263285e-06,
      "loss": 0.7856,
      "step": 8711
    },
    {
      "epoch": 0.4794980461225164,
      "grad_norm": 0.7421835064888,
      "learning_rate": 8.670001183543528e-06,
      "loss": 0.8165,
      "step": 8712
    },
    {
      "epoch": 0.47955308492487203,
      "grad_norm": 0.6498512625694275,
      "learning_rate": 8.669706781241028e-06,
      "loss": 0.7212,
      "step": 8713
    },
    {
      "epoch": 0.4796081237272277,
      "grad_norm": 0.8493219614028931,
      "learning_rate": 8.669412351357993e-06,
      "loss": 0.8036,
      "step": 8714
    },
    {
      "epoch": 0.47966316252958335,
      "grad_norm": 0.6834331750869751,
      "learning_rate": 8.669117893896637e-06,
      "loss": 0.8127,
      "step": 8715
    },
    {
      "epoch": 0.47971820133193904,
      "grad_norm": 0.7793670296669006,
      "learning_rate": 8.668823408859172e-06,
      "loss": 0.7276,
      "step": 8716
    },
    {
      "epoch": 0.4797732401342947,
      "grad_norm": 0.7108075022697449,
      "learning_rate": 8.668528896247815e-06,
      "loss": 0.8328,
      "step": 8717
    },
    {
      "epoch": 0.47982827893665037,
      "grad_norm": 0.6662433743476868,
      "learning_rate": 8.668234356064774e-06,
      "loss": 0.6751,
      "step": 8718
    },
    {
      "epoch": 0.479883317739006,
      "grad_norm": 0.6595591902732849,
      "learning_rate": 8.667939788312267e-06,
      "loss": 0.707,
      "step": 8719
    },
    {
      "epoch": 0.47993835654136163,
      "grad_norm": 0.7435836791992188,
      "learning_rate": 8.667645192992506e-06,
      "loss": 0.7885,
      "step": 8720
    },
    {
      "epoch": 0.4799933953437173,
      "grad_norm": 0.6999356746673584,
      "learning_rate": 8.667350570107706e-06,
      "loss": 0.7538,
      "step": 8721
    },
    {
      "epoch": 0.48004843414607296,
      "grad_norm": 0.7111191749572754,
      "learning_rate": 8.66705591966008e-06,
      "loss": 0.6814,
      "step": 8722
    },
    {
      "epoch": 0.48010347294842864,
      "grad_norm": 0.6752734780311584,
      "learning_rate": 8.666761241651844e-06,
      "loss": 0.7221,
      "step": 8723
    },
    {
      "epoch": 0.4801585117507843,
      "grad_norm": 0.7432951331138611,
      "learning_rate": 8.666466536085212e-06,
      "loss": 0.7689,
      "step": 8724
    },
    {
      "epoch": 0.48021355055313997,
      "grad_norm": 0.7384392023086548,
      "learning_rate": 8.666171802962398e-06,
      "loss": 0.7862,
      "step": 8725
    },
    {
      "epoch": 0.4802685893554956,
      "grad_norm": 0.6878762245178223,
      "learning_rate": 8.66587704228562e-06,
      "loss": 0.7246,
      "step": 8726
    },
    {
      "epoch": 0.4803236281578513,
      "grad_norm": 0.6640586853027344,
      "learning_rate": 8.66558225405709e-06,
      "loss": 0.7181,
      "step": 8727
    },
    {
      "epoch": 0.4803786669602069,
      "grad_norm": 0.6808595061302185,
      "learning_rate": 8.665287438279024e-06,
      "loss": 0.7866,
      "step": 8728
    },
    {
      "epoch": 0.4804337057625626,
      "grad_norm": 0.5966268181800842,
      "learning_rate": 8.66499259495364e-06,
      "loss": 0.6755,
      "step": 8729
    },
    {
      "epoch": 0.48048874456491825,
      "grad_norm": 0.742016077041626,
      "learning_rate": 8.664697724083152e-06,
      "loss": 0.8682,
      "step": 8730
    },
    {
      "epoch": 0.48054378336727394,
      "grad_norm": 0.6621154546737671,
      "learning_rate": 8.66440282566978e-06,
      "loss": 0.7525,
      "step": 8731
    },
    {
      "epoch": 0.48059882216962957,
      "grad_norm": 0.7347434759140015,
      "learning_rate": 8.664107899715733e-06,
      "loss": 0.7919,
      "step": 8732
    },
    {
      "epoch": 0.48065386097198526,
      "grad_norm": 0.7564681172370911,
      "learning_rate": 8.663812946223234e-06,
      "loss": 0.9172,
      "step": 8733
    },
    {
      "epoch": 0.4807088997743409,
      "grad_norm": 0.7193084359169006,
      "learning_rate": 8.663517965194497e-06,
      "loss": 0.7931,
      "step": 8734
    },
    {
      "epoch": 0.4807639385766966,
      "grad_norm": 0.6882064938545227,
      "learning_rate": 8.66322295663174e-06,
      "loss": 0.7678,
      "step": 8735
    },
    {
      "epoch": 0.4808189773790522,
      "grad_norm": 0.7954713106155396,
      "learning_rate": 8.662927920537179e-06,
      "loss": 0.6357,
      "step": 8736
    },
    {
      "epoch": 0.4808740161814079,
      "grad_norm": 0.7123041749000549,
      "learning_rate": 8.662632856913034e-06,
      "loss": 0.7234,
      "step": 8737
    },
    {
      "epoch": 0.48092905498376354,
      "grad_norm": 0.745145320892334,
      "learning_rate": 8.66233776576152e-06,
      "loss": 0.7516,
      "step": 8738
    },
    {
      "epoch": 0.4809840937861192,
      "grad_norm": 0.6904219388961792,
      "learning_rate": 8.662042647084856e-06,
      "loss": 0.7995,
      "step": 8739
    },
    {
      "epoch": 0.48103913258847486,
      "grad_norm": 0.71831214427948,
      "learning_rate": 8.661747500885258e-06,
      "loss": 0.7965,
      "step": 8740
    },
    {
      "epoch": 0.48109417139083055,
      "grad_norm": 0.8514378666877747,
      "learning_rate": 8.661452327164948e-06,
      "loss": 0.8023,
      "step": 8741
    },
    {
      "epoch": 0.4811492101931862,
      "grad_norm": 0.7411143779754639,
      "learning_rate": 8.66115712592614e-06,
      "loss": 0.797,
      "step": 8742
    },
    {
      "epoch": 0.4812042489955419,
      "grad_norm": 0.737178385257721,
      "learning_rate": 8.660861897171057e-06,
      "loss": 0.7286,
      "step": 8743
    },
    {
      "epoch": 0.4812592877978975,
      "grad_norm": 0.6823513507843018,
      "learning_rate": 8.660566640901918e-06,
      "loss": 0.7482,
      "step": 8744
    },
    {
      "epoch": 0.4813143266002532,
      "grad_norm": 0.7205879092216492,
      "learning_rate": 8.660271357120937e-06,
      "loss": 0.8294,
      "step": 8745
    },
    {
      "epoch": 0.48136936540260883,
      "grad_norm": 0.6887338757514954,
      "learning_rate": 8.659976045830337e-06,
      "loss": 0.7711,
      "step": 8746
    },
    {
      "epoch": 0.4814244042049645,
      "grad_norm": 0.7498533129692078,
      "learning_rate": 8.659680707032336e-06,
      "loss": 0.7296,
      "step": 8747
    },
    {
      "epoch": 0.48147944300732015,
      "grad_norm": 0.8041636943817139,
      "learning_rate": 8.659385340729155e-06,
      "loss": 0.9213,
      "step": 8748
    },
    {
      "epoch": 0.48153448180967584,
      "grad_norm": 0.8623721599578857,
      "learning_rate": 8.659089946923014e-06,
      "loss": 0.8024,
      "step": 8749
    },
    {
      "epoch": 0.4815895206120315,
      "grad_norm": 0.7212050557136536,
      "learning_rate": 8.658794525616132e-06,
      "loss": 0.732,
      "step": 8750
    },
    {
      "epoch": 0.48164455941438716,
      "grad_norm": 0.7141492366790771,
      "learning_rate": 8.658499076810729e-06,
      "loss": 0.8062,
      "step": 8751
    },
    {
      "epoch": 0.4816995982167428,
      "grad_norm": 0.7191516160964966,
      "learning_rate": 8.658203600509027e-06,
      "loss": 0.805,
      "step": 8752
    },
    {
      "epoch": 0.4817546370190985,
      "grad_norm": 0.71059650182724,
      "learning_rate": 8.657908096713245e-06,
      "loss": 0.6755,
      "step": 8753
    },
    {
      "epoch": 0.4818096758214541,
      "grad_norm": 0.6715459823608398,
      "learning_rate": 8.657612565425607e-06,
      "loss": 0.8093,
      "step": 8754
    },
    {
      "epoch": 0.4818647146238098,
      "grad_norm": 0.7438814640045166,
      "learning_rate": 8.65731700664833e-06,
      "loss": 0.8059,
      "step": 8755
    },
    {
      "epoch": 0.48191975342616544,
      "grad_norm": 0.7295387387275696,
      "learning_rate": 8.657021420383637e-06,
      "loss": 0.8437,
      "step": 8756
    },
    {
      "epoch": 0.48197479222852113,
      "grad_norm": 0.7053797245025635,
      "learning_rate": 8.656725806633753e-06,
      "loss": 0.8424,
      "step": 8757
    },
    {
      "epoch": 0.48202983103087677,
      "grad_norm": 0.6902007460594177,
      "learning_rate": 8.656430165400894e-06,
      "loss": 0.6967,
      "step": 8758
    },
    {
      "epoch": 0.48208486983323245,
      "grad_norm": 0.66749507188797,
      "learning_rate": 8.656134496687286e-06,
      "loss": 0.7858,
      "step": 8759
    },
    {
      "epoch": 0.4821399086355881,
      "grad_norm": 0.6755428314208984,
      "learning_rate": 8.65583880049515e-06,
      "loss": 0.6669,
      "step": 8760
    },
    {
      "epoch": 0.4821949474379438,
      "grad_norm": 0.921096920967102,
      "learning_rate": 8.655543076826706e-06,
      "loss": 0.8545,
      "step": 8761
    },
    {
      "epoch": 0.4822499862402994,
      "grad_norm": 0.7931553721427917,
      "learning_rate": 8.65524732568418e-06,
      "loss": 0.8708,
      "step": 8762
    },
    {
      "epoch": 0.48230502504265504,
      "grad_norm": 0.7891780734062195,
      "learning_rate": 8.654951547069794e-06,
      "loss": 0.687,
      "step": 8763
    },
    {
      "epoch": 0.48236006384501073,
      "grad_norm": 0.747662365436554,
      "learning_rate": 8.65465574098577e-06,
      "loss": 0.8153,
      "step": 8764
    },
    {
      "epoch": 0.48241510264736637,
      "grad_norm": 0.7758497595787048,
      "learning_rate": 8.65435990743433e-06,
      "loss": 0.8018,
      "step": 8765
    },
    {
      "epoch": 0.48247014144972206,
      "grad_norm": 0.6997805237770081,
      "learning_rate": 8.654064046417703e-06,
      "loss": 0.7845,
      "step": 8766
    },
    {
      "epoch": 0.4825251802520777,
      "grad_norm": 0.7188366651535034,
      "learning_rate": 8.653768157938106e-06,
      "loss": 0.7528,
      "step": 8767
    },
    {
      "epoch": 0.4825802190544334,
      "grad_norm": 0.6848055124282837,
      "learning_rate": 8.653472241997767e-06,
      "loss": 0.7658,
      "step": 8768
    },
    {
      "epoch": 0.482635257856789,
      "grad_norm": 1.0603824853897095,
      "learning_rate": 8.653176298598907e-06,
      "loss": 0.7692,
      "step": 8769
    },
    {
      "epoch": 0.4826902966591447,
      "grad_norm": 0.8191514611244202,
      "learning_rate": 8.652880327743753e-06,
      "loss": 0.7706,
      "step": 8770
    },
    {
      "epoch": 0.48274533546150034,
      "grad_norm": 0.6318503618240356,
      "learning_rate": 8.652584329434527e-06,
      "loss": 0.6635,
      "step": 8771
    },
    {
      "epoch": 0.482800374263856,
      "grad_norm": 0.6860769391059875,
      "learning_rate": 8.652288303673457e-06,
      "loss": 0.739,
      "step": 8772
    },
    {
      "epoch": 0.48285541306621166,
      "grad_norm": 0.7414761185646057,
      "learning_rate": 8.651992250462765e-06,
      "loss": 0.7949,
      "step": 8773
    },
    {
      "epoch": 0.48291045186856735,
      "grad_norm": 0.7255183458328247,
      "learning_rate": 8.651696169804676e-06,
      "loss": 0.8569,
      "step": 8774
    },
    {
      "epoch": 0.482965490670923,
      "grad_norm": 0.7034135460853577,
      "learning_rate": 8.651400061701417e-06,
      "loss": 0.7562,
      "step": 8775
    },
    {
      "epoch": 0.48302052947327867,
      "grad_norm": 0.7041038274765015,
      "learning_rate": 8.651103926155212e-06,
      "loss": 0.7194,
      "step": 8776
    },
    {
      "epoch": 0.4830755682756343,
      "grad_norm": 1.0965619087219238,
      "learning_rate": 8.650807763168287e-06,
      "loss": 0.9033,
      "step": 8777
    },
    {
      "epoch": 0.48313060707799,
      "grad_norm": 0.7400044798851013,
      "learning_rate": 8.650511572742869e-06,
      "loss": 0.7626,
      "step": 8778
    },
    {
      "epoch": 0.4831856458803456,
      "grad_norm": 0.6957885026931763,
      "learning_rate": 8.650215354881182e-06,
      "loss": 0.7283,
      "step": 8779
    },
    {
      "epoch": 0.4832406846827013,
      "grad_norm": 0.7992473840713501,
      "learning_rate": 8.649919109585454e-06,
      "loss": 0.8376,
      "step": 8780
    },
    {
      "epoch": 0.48329572348505695,
      "grad_norm": 0.8556981086730957,
      "learning_rate": 8.649622836857911e-06,
      "loss": 0.7737,
      "step": 8781
    },
    {
      "epoch": 0.48335076228741264,
      "grad_norm": 0.8476192355155945,
      "learning_rate": 8.64932653670078e-06,
      "loss": 0.8926,
      "step": 8782
    },
    {
      "epoch": 0.48340580108976827,
      "grad_norm": 0.6461093425750732,
      "learning_rate": 8.649030209116289e-06,
      "loss": 0.7452,
      "step": 8783
    },
    {
      "epoch": 0.48346083989212396,
      "grad_norm": 0.6997528076171875,
      "learning_rate": 8.648733854106661e-06,
      "loss": 0.7962,
      "step": 8784
    },
    {
      "epoch": 0.4835158786944796,
      "grad_norm": 0.7606356739997864,
      "learning_rate": 8.648437471674128e-06,
      "loss": 0.6517,
      "step": 8785
    },
    {
      "epoch": 0.4835709174968353,
      "grad_norm": 0.8118630051612854,
      "learning_rate": 8.648141061820913e-06,
      "loss": 0.7539,
      "step": 8786
    },
    {
      "epoch": 0.4836259562991909,
      "grad_norm": 0.8778805136680603,
      "learning_rate": 8.64784462454925e-06,
      "loss": 0.763,
      "step": 8787
    },
    {
      "epoch": 0.4836809951015466,
      "grad_norm": 0.7741022706031799,
      "learning_rate": 8.647548159861361e-06,
      "loss": 0.7749,
      "step": 8788
    },
    {
      "epoch": 0.48373603390390224,
      "grad_norm": 0.76578688621521,
      "learning_rate": 8.647251667759478e-06,
      "loss": 0.6968,
      "step": 8789
    },
    {
      "epoch": 0.48379107270625793,
      "grad_norm": 0.8477250933647156,
      "learning_rate": 8.646955148245827e-06,
      "loss": 0.8364,
      "step": 8790
    },
    {
      "epoch": 0.48384611150861356,
      "grad_norm": 0.9105041027069092,
      "learning_rate": 8.646658601322635e-06,
      "loss": 0.823,
      "step": 8791
    },
    {
      "epoch": 0.48390115031096925,
      "grad_norm": 0.7642726898193359,
      "learning_rate": 8.646362026992135e-06,
      "loss": 0.721,
      "step": 8792
    },
    {
      "epoch": 0.4839561891133249,
      "grad_norm": 0.7567259669303894,
      "learning_rate": 8.646065425256555e-06,
      "loss": 0.7876,
      "step": 8793
    },
    {
      "epoch": 0.4840112279156806,
      "grad_norm": 0.7691231966018677,
      "learning_rate": 8.64576879611812e-06,
      "loss": 0.8308,
      "step": 8794
    },
    {
      "epoch": 0.4840662667180362,
      "grad_norm": 1.0769426822662354,
      "learning_rate": 8.645472139579067e-06,
      "loss": 0.892,
      "step": 8795
    },
    {
      "epoch": 0.4841213055203919,
      "grad_norm": 0.6987955570220947,
      "learning_rate": 8.64517545564162e-06,
      "loss": 0.8254,
      "step": 8796
    },
    {
      "epoch": 0.48417634432274753,
      "grad_norm": 0.7736005783081055,
      "learning_rate": 8.644878744308007e-06,
      "loss": 0.7666,
      "step": 8797
    },
    {
      "epoch": 0.4842313831251032,
      "grad_norm": 0.6233380436897278,
      "learning_rate": 8.644582005580464e-06,
      "loss": 0.6443,
      "step": 8798
    },
    {
      "epoch": 0.48428642192745885,
      "grad_norm": 0.7343530654907227,
      "learning_rate": 8.644285239461217e-06,
      "loss": 0.724,
      "step": 8799
    },
    {
      "epoch": 0.48434146072981454,
      "grad_norm": 0.725321352481842,
      "learning_rate": 8.643988445952499e-06,
      "loss": 0.7249,
      "step": 8800
    },
    {
      "epoch": 0.4843964995321702,
      "grad_norm": 0.7256256341934204,
      "learning_rate": 8.643691625056539e-06,
      "loss": 0.8656,
      "step": 8801
    },
    {
      "epoch": 0.48445153833452587,
      "grad_norm": 0.8559528589248657,
      "learning_rate": 8.643394776775567e-06,
      "loss": 0.9186,
      "step": 8802
    },
    {
      "epoch": 0.4845065771368815,
      "grad_norm": 0.6735692024230957,
      "learning_rate": 8.643097901111815e-06,
      "loss": 0.7007,
      "step": 8803
    },
    {
      "epoch": 0.4845616159392372,
      "grad_norm": 0.8373280167579651,
      "learning_rate": 8.642800998067515e-06,
      "loss": 0.8774,
      "step": 8804
    },
    {
      "epoch": 0.4846166547415928,
      "grad_norm": 0.731311023235321,
      "learning_rate": 8.642504067644898e-06,
      "loss": 0.7102,
      "step": 8805
    },
    {
      "epoch": 0.48467169354394846,
      "grad_norm": 0.7259742617607117,
      "learning_rate": 8.642207109846195e-06,
      "loss": 0.7174,
      "step": 8806
    },
    {
      "epoch": 0.48472673234630415,
      "grad_norm": 0.6454386115074158,
      "learning_rate": 8.641910124673638e-06,
      "loss": 0.7656,
      "step": 8807
    },
    {
      "epoch": 0.4847817711486598,
      "grad_norm": 0.7701624631881714,
      "learning_rate": 8.641613112129462e-06,
      "loss": 0.7926,
      "step": 8808
    },
    {
      "epoch": 0.48483680995101547,
      "grad_norm": 0.6812854409217834,
      "learning_rate": 8.641316072215893e-06,
      "loss": 0.7072,
      "step": 8809
    },
    {
      "epoch": 0.4848918487533711,
      "grad_norm": 0.8180119395256042,
      "learning_rate": 8.641019004935169e-06,
      "loss": 0.8621,
      "step": 8810
    },
    {
      "epoch": 0.4849468875557268,
      "grad_norm": 0.6346331834793091,
      "learning_rate": 8.64072191028952e-06,
      "loss": 0.6907,
      "step": 8811
    },
    {
      "epoch": 0.4850019263580824,
      "grad_norm": 0.6819741129875183,
      "learning_rate": 8.64042478828118e-06,
      "loss": 0.77,
      "step": 8812
    },
    {
      "epoch": 0.4850569651604381,
      "grad_norm": 0.9074214100837708,
      "learning_rate": 8.640127638912383e-06,
      "loss": 0.7799,
      "step": 8813
    },
    {
      "epoch": 0.48511200396279375,
      "grad_norm": 0.8065158724784851,
      "learning_rate": 8.63983046218536e-06,
      "loss": 0.8033,
      "step": 8814
    },
    {
      "epoch": 0.48516704276514944,
      "grad_norm": 0.6241241097450256,
      "learning_rate": 8.639533258102345e-06,
      "loss": 0.6936,
      "step": 8815
    },
    {
      "epoch": 0.48522208156750507,
      "grad_norm": 0.6928265690803528,
      "learning_rate": 8.639236026665573e-06,
      "loss": 0.7526,
      "step": 8816
    },
    {
      "epoch": 0.48527712036986076,
      "grad_norm": 0.8171425461769104,
      "learning_rate": 8.638938767877276e-06,
      "loss": 0.8227,
      "step": 8817
    },
    {
      "epoch": 0.4853321591722164,
      "grad_norm": 0.7007083296775818,
      "learning_rate": 8.638641481739692e-06,
      "loss": 0.7439,
      "step": 8818
    },
    {
      "epoch": 0.4853871979745721,
      "grad_norm": 0.8905115127563477,
      "learning_rate": 8.63834416825505e-06,
      "loss": 0.6873,
      "step": 8819
    },
    {
      "epoch": 0.4854422367769277,
      "grad_norm": 0.702198326587677,
      "learning_rate": 8.638046827425588e-06,
      "loss": 0.7999,
      "step": 8820
    },
    {
      "epoch": 0.4854972755792834,
      "grad_norm": 0.7280104160308838,
      "learning_rate": 8.63774945925354e-06,
      "loss": 0.8562,
      "step": 8821
    },
    {
      "epoch": 0.48555231438163904,
      "grad_norm": 0.9803630113601685,
      "learning_rate": 8.63745206374114e-06,
      "loss": 0.8347,
      "step": 8822
    },
    {
      "epoch": 0.4856073531839947,
      "grad_norm": 0.6781168580055237,
      "learning_rate": 8.637154640890625e-06,
      "loss": 0.8124,
      "step": 8823
    },
    {
      "epoch": 0.48566239198635036,
      "grad_norm": 0.7219669222831726,
      "learning_rate": 8.63685719070423e-06,
      "loss": 0.8053,
      "step": 8824
    },
    {
      "epoch": 0.48571743078870605,
      "grad_norm": 0.7077241539955139,
      "learning_rate": 8.636559713184187e-06,
      "loss": 0.7534,
      "step": 8825
    },
    {
      "epoch": 0.4857724695910617,
      "grad_norm": 0.70063316822052,
      "learning_rate": 8.636262208332737e-06,
      "loss": 0.7509,
      "step": 8826
    },
    {
      "epoch": 0.4858275083934174,
      "grad_norm": 0.7292184233665466,
      "learning_rate": 8.635964676152114e-06,
      "loss": 0.7485,
      "step": 8827
    },
    {
      "epoch": 0.485882547195773,
      "grad_norm": 0.7970258593559265,
      "learning_rate": 8.635667116644552e-06,
      "loss": 0.8874,
      "step": 8828
    },
    {
      "epoch": 0.4859375859981287,
      "grad_norm": 0.7090024352073669,
      "learning_rate": 8.63536952981229e-06,
      "loss": 0.7665,
      "step": 8829
    },
    {
      "epoch": 0.48599262480048433,
      "grad_norm": 0.761409342288971,
      "learning_rate": 8.635071915657565e-06,
      "loss": 0.7977,
      "step": 8830
    },
    {
      "epoch": 0.48604766360284,
      "grad_norm": 0.724896252155304,
      "learning_rate": 8.634774274182611e-06,
      "loss": 0.8591,
      "step": 8831
    },
    {
      "epoch": 0.48610270240519565,
      "grad_norm": 0.737424910068512,
      "learning_rate": 8.634476605389666e-06,
      "loss": 0.8256,
      "step": 8832
    },
    {
      "epoch": 0.48615774120755134,
      "grad_norm": 0.8261227607727051,
      "learning_rate": 8.63417890928097e-06,
      "loss": 0.8089,
      "step": 8833
    },
    {
      "epoch": 0.486212780009907,
      "grad_norm": 0.6744595766067505,
      "learning_rate": 8.633881185858756e-06,
      "loss": 0.7821,
      "step": 8834
    },
    {
      "epoch": 0.48626781881226266,
      "grad_norm": 0.6717672944068909,
      "learning_rate": 8.633583435125263e-06,
      "loss": 0.7823,
      "step": 8835
    },
    {
      "epoch": 0.4863228576146183,
      "grad_norm": 0.753616213798523,
      "learning_rate": 8.633285657082732e-06,
      "loss": 0.8044,
      "step": 8836
    },
    {
      "epoch": 0.486377896416974,
      "grad_norm": 0.6910914182662964,
      "learning_rate": 8.632987851733397e-06,
      "loss": 0.8244,
      "step": 8837
    },
    {
      "epoch": 0.4864329352193296,
      "grad_norm": 0.9127064347267151,
      "learning_rate": 8.632690019079499e-06,
      "loss": 0.7918,
      "step": 8838
    },
    {
      "epoch": 0.4864879740216853,
      "grad_norm": 0.715918779373169,
      "learning_rate": 8.632392159123274e-06,
      "loss": 0.744,
      "step": 8839
    },
    {
      "epoch": 0.48654301282404094,
      "grad_norm": 0.8206684589385986,
      "learning_rate": 8.632094271866963e-06,
      "loss": 0.7852,
      "step": 8840
    },
    {
      "epoch": 0.48659805162639663,
      "grad_norm": 0.6502171158790588,
      "learning_rate": 8.631796357312802e-06,
      "loss": 0.7653,
      "step": 8841
    },
    {
      "epoch": 0.48665309042875227,
      "grad_norm": 0.6987786889076233,
      "learning_rate": 8.631498415463033e-06,
      "loss": 0.7669,
      "step": 8842
    },
    {
      "epoch": 0.48670812923110796,
      "grad_norm": 0.7902390360832214,
      "learning_rate": 8.631200446319894e-06,
      "loss": 0.8438,
      "step": 8843
    },
    {
      "epoch": 0.4867631680334636,
      "grad_norm": 0.7464659810066223,
      "learning_rate": 8.630902449885625e-06,
      "loss": 0.8276,
      "step": 8844
    },
    {
      "epoch": 0.4868182068358193,
      "grad_norm": 0.7375630736351013,
      "learning_rate": 8.630604426162465e-06,
      "loss": 0.7921,
      "step": 8845
    },
    {
      "epoch": 0.4868732456381749,
      "grad_norm": 0.7206295728683472,
      "learning_rate": 8.630306375152653e-06,
      "loss": 0.8424,
      "step": 8846
    },
    {
      "epoch": 0.4869282844405306,
      "grad_norm": 0.7384368181228638,
      "learning_rate": 8.63000829685843e-06,
      "loss": 0.8702,
      "step": 8847
    },
    {
      "epoch": 0.48698332324288623,
      "grad_norm": 0.7839015126228333,
      "learning_rate": 8.629710191282037e-06,
      "loss": 0.7064,
      "step": 8848
    },
    {
      "epoch": 0.48703836204524187,
      "grad_norm": 0.6909724473953247,
      "learning_rate": 8.629412058425712e-06,
      "loss": 0.6924,
      "step": 8849
    },
    {
      "epoch": 0.48709340084759756,
      "grad_norm": 0.6553036570549011,
      "learning_rate": 8.6291138982917e-06,
      "loss": 0.6526,
      "step": 8850
    },
    {
      "epoch": 0.4871484396499532,
      "grad_norm": 0.7202072143554688,
      "learning_rate": 8.628815710882239e-06,
      "loss": 0.7272,
      "step": 8851
    },
    {
      "epoch": 0.4872034784523089,
      "grad_norm": 0.6898619532585144,
      "learning_rate": 8.62851749619957e-06,
      "loss": 0.7687,
      "step": 8852
    },
    {
      "epoch": 0.4872585172546645,
      "grad_norm": 0.7888908386230469,
      "learning_rate": 8.628219254245935e-06,
      "loss": 0.7654,
      "step": 8853
    },
    {
      "epoch": 0.4873135560570202,
      "grad_norm": 0.7312424778938293,
      "learning_rate": 8.627920985023575e-06,
      "loss": 0.8053,
      "step": 8854
    },
    {
      "epoch": 0.48736859485937584,
      "grad_norm": 0.6588439345359802,
      "learning_rate": 8.627622688534731e-06,
      "loss": 0.7229,
      "step": 8855
    },
    {
      "epoch": 0.4874236336617315,
      "grad_norm": 0.8292293548583984,
      "learning_rate": 8.627324364781647e-06,
      "loss": 0.8482,
      "step": 8856
    },
    {
      "epoch": 0.48747867246408716,
      "grad_norm": 0.7573973536491394,
      "learning_rate": 8.627026013766564e-06,
      "loss": 0.7282,
      "step": 8857
    },
    {
      "epoch": 0.48753371126644285,
      "grad_norm": 1.2215768098831177,
      "learning_rate": 8.626727635491726e-06,
      "loss": 0.7771,
      "step": 8858
    },
    {
      "epoch": 0.4875887500687985,
      "grad_norm": 0.7324759364128113,
      "learning_rate": 8.626429229959369e-06,
      "loss": 0.781,
      "step": 8859
    },
    {
      "epoch": 0.48764378887115417,
      "grad_norm": 0.6995676159858704,
      "learning_rate": 8.626130797171745e-06,
      "loss": 0.6907,
      "step": 8860
    },
    {
      "epoch": 0.4876988276735098,
      "grad_norm": 0.7400509119033813,
      "learning_rate": 8.625832337131092e-06,
      "loss": 0.6572,
      "step": 8861
    },
    {
      "epoch": 0.4877538664758655,
      "grad_norm": 0.6634842753410339,
      "learning_rate": 8.625533849839653e-06,
      "loss": 0.7229,
      "step": 8862
    },
    {
      "epoch": 0.4878089052782211,
      "grad_norm": 0.7357299327850342,
      "learning_rate": 8.625235335299673e-06,
      "loss": 0.6418,
      "step": 8863
    },
    {
      "epoch": 0.4878639440805768,
      "grad_norm": 0.6473466157913208,
      "learning_rate": 8.624936793513394e-06,
      "loss": 0.6796,
      "step": 8864
    },
    {
      "epoch": 0.48791898288293245,
      "grad_norm": 0.9110734462738037,
      "learning_rate": 8.62463822448306e-06,
      "loss": 0.8143,
      "step": 8865
    },
    {
      "epoch": 0.48797402168528814,
      "grad_norm": 0.7932308316230774,
      "learning_rate": 8.624339628210916e-06,
      "loss": 0.9103,
      "step": 8866
    },
    {
      "epoch": 0.4880290604876438,
      "grad_norm": 0.6677752137184143,
      "learning_rate": 8.624041004699205e-06,
      "loss": 0.8073,
      "step": 8867
    },
    {
      "epoch": 0.48808409928999946,
      "grad_norm": 0.7379121780395508,
      "learning_rate": 8.623742353950171e-06,
      "loss": 0.8643,
      "step": 8868
    },
    {
      "epoch": 0.4881391380923551,
      "grad_norm": 0.7479479312896729,
      "learning_rate": 8.623443675966062e-06,
      "loss": 0.6117,
      "step": 8869
    },
    {
      "epoch": 0.4881941768947108,
      "grad_norm": 0.7822794914245605,
      "learning_rate": 8.623144970749118e-06,
      "loss": 0.8629,
      "step": 8870
    },
    {
      "epoch": 0.4882492156970664,
      "grad_norm": 0.7040950655937195,
      "learning_rate": 8.622846238301587e-06,
      "loss": 0.7519,
      "step": 8871
    },
    {
      "epoch": 0.4883042544994221,
      "grad_norm": 0.747368574142456,
      "learning_rate": 8.622547478625714e-06,
      "loss": 0.7459,
      "step": 8872
    },
    {
      "epoch": 0.48835929330177774,
      "grad_norm": 0.6755948066711426,
      "learning_rate": 8.622248691723742e-06,
      "loss": 0.7515,
      "step": 8873
    },
    {
      "epoch": 0.48841433210413343,
      "grad_norm": 0.7265586256980896,
      "learning_rate": 8.62194987759792e-06,
      "loss": 0.7691,
      "step": 8874
    },
    {
      "epoch": 0.48846937090648906,
      "grad_norm": 0.6696380972862244,
      "learning_rate": 8.621651036250493e-06,
      "loss": 0.778,
      "step": 8875
    },
    {
      "epoch": 0.48852440970884475,
      "grad_norm": 0.7666454911231995,
      "learning_rate": 8.621352167683705e-06,
      "loss": 0.7396,
      "step": 8876
    },
    {
      "epoch": 0.4885794485112004,
      "grad_norm": 0.7079235315322876,
      "learning_rate": 8.621053271899803e-06,
      "loss": 0.7917,
      "step": 8877
    },
    {
      "epoch": 0.4886344873135561,
      "grad_norm": 0.6888919472694397,
      "learning_rate": 8.620754348901034e-06,
      "loss": 0.605,
      "step": 8878
    },
    {
      "epoch": 0.4886895261159117,
      "grad_norm": 0.7177572250366211,
      "learning_rate": 8.620455398689645e-06,
      "loss": 0.7534,
      "step": 8879
    },
    {
      "epoch": 0.4887445649182674,
      "grad_norm": 0.7268772721290588,
      "learning_rate": 8.620156421267883e-06,
      "loss": 0.7748,
      "step": 8880
    },
    {
      "epoch": 0.48879960372062303,
      "grad_norm": 0.8015080690383911,
      "learning_rate": 8.619857416637993e-06,
      "loss": 0.6716,
      "step": 8881
    },
    {
      "epoch": 0.4888546425229787,
      "grad_norm": 0.7464118599891663,
      "learning_rate": 8.619558384802226e-06,
      "loss": 0.796,
      "step": 8882
    },
    {
      "epoch": 0.48890968132533436,
      "grad_norm": 0.6829718351364136,
      "learning_rate": 8.619259325762826e-06,
      "loss": 0.788,
      "step": 8883
    },
    {
      "epoch": 0.48896472012769004,
      "grad_norm": 0.6553084850311279,
      "learning_rate": 8.618960239522041e-06,
      "loss": 0.7215,
      "step": 8884
    },
    {
      "epoch": 0.4890197589300457,
      "grad_norm": 0.8056252598762512,
      "learning_rate": 8.618661126082119e-06,
      "loss": 0.8588,
      "step": 8885
    },
    {
      "epoch": 0.48907479773240137,
      "grad_norm": 0.8145674467086792,
      "learning_rate": 8.618361985445309e-06,
      "loss": 0.8095,
      "step": 8886
    },
    {
      "epoch": 0.489129836534757,
      "grad_norm": 0.740031898021698,
      "learning_rate": 8.61806281761386e-06,
      "loss": 0.7029,
      "step": 8887
    },
    {
      "epoch": 0.4891848753371127,
      "grad_norm": 0.7442640662193298,
      "learning_rate": 8.617763622590019e-06,
      "loss": 0.782,
      "step": 8888
    },
    {
      "epoch": 0.4892399141394683,
      "grad_norm": 0.6992725133895874,
      "learning_rate": 8.617464400376035e-06,
      "loss": 0.7877,
      "step": 8889
    },
    {
      "epoch": 0.489294952941824,
      "grad_norm": 1.19756281375885,
      "learning_rate": 8.617165150974157e-06,
      "loss": 0.6985,
      "step": 8890
    },
    {
      "epoch": 0.48934999174417965,
      "grad_norm": 0.6418262720108032,
      "learning_rate": 8.616865874386633e-06,
      "loss": 0.7385,
      "step": 8891
    },
    {
      "epoch": 0.4894050305465353,
      "grad_norm": 0.787406325340271,
      "learning_rate": 8.616566570615714e-06,
      "loss": 0.8686,
      "step": 8892
    },
    {
      "epoch": 0.48946006934889097,
      "grad_norm": 0.6990430951118469,
      "learning_rate": 8.616267239663648e-06,
      "loss": 0.7683,
      "step": 8893
    },
    {
      "epoch": 0.4895151081512466,
      "grad_norm": 0.7180235981941223,
      "learning_rate": 8.615967881532687e-06,
      "loss": 0.8337,
      "step": 8894
    },
    {
      "epoch": 0.4895701469536023,
      "grad_norm": 0.7647475600242615,
      "learning_rate": 8.615668496225077e-06,
      "loss": 0.8668,
      "step": 8895
    },
    {
      "epoch": 0.4896251857559579,
      "grad_norm": 0.843063473701477,
      "learning_rate": 8.615369083743072e-06,
      "loss": 0.7968,
      "step": 8896
    },
    {
      "epoch": 0.4896802245583136,
      "grad_norm": 0.9526075124740601,
      "learning_rate": 8.61506964408892e-06,
      "loss": 0.8766,
      "step": 8897
    },
    {
      "epoch": 0.48973526336066925,
      "grad_norm": 0.7850056290626526,
      "learning_rate": 8.614770177264874e-06,
      "loss": 0.8033,
      "step": 8898
    },
    {
      "epoch": 0.48979030216302494,
      "grad_norm": 0.8658629655838013,
      "learning_rate": 8.614470683273182e-06,
      "loss": 0.8206,
      "step": 8899
    },
    {
      "epoch": 0.48984534096538057,
      "grad_norm": 0.8060176968574524,
      "learning_rate": 8.614171162116096e-06,
      "loss": 0.7602,
      "step": 8900
    },
    {
      "epoch": 0.48990037976773626,
      "grad_norm": 0.7398280501365662,
      "learning_rate": 8.613871613795865e-06,
      "loss": 0.8067,
      "step": 8901
    },
    {
      "epoch": 0.4899554185700919,
      "grad_norm": 0.7341256141662598,
      "learning_rate": 8.613572038314744e-06,
      "loss": 0.7305,
      "step": 8902
    },
    {
      "epoch": 0.4900104573724476,
      "grad_norm": 0.7832887172698975,
      "learning_rate": 8.613272435674984e-06,
      "loss": 0.7012,
      "step": 8903
    },
    {
      "epoch": 0.4900654961748032,
      "grad_norm": 0.6536995768547058,
      "learning_rate": 8.612972805878834e-06,
      "loss": 0.745,
      "step": 8904
    },
    {
      "epoch": 0.4901205349771589,
      "grad_norm": 0.7511856555938721,
      "learning_rate": 8.612673148928547e-06,
      "loss": 0.7741,
      "step": 8905
    },
    {
      "epoch": 0.49017557377951454,
      "grad_norm": 0.6117261648178101,
      "learning_rate": 8.612373464826377e-06,
      "loss": 0.5813,
      "step": 8906
    },
    {
      "epoch": 0.49023061258187023,
      "grad_norm": 0.7832254767417908,
      "learning_rate": 8.612073753574574e-06,
      "loss": 0.7426,
      "step": 8907
    },
    {
      "epoch": 0.49028565138422586,
      "grad_norm": 0.7516622543334961,
      "learning_rate": 8.611774015175393e-06,
      "loss": 0.8205,
      "step": 8908
    },
    {
      "epoch": 0.49034069018658155,
      "grad_norm": 0.7776936888694763,
      "learning_rate": 8.611474249631085e-06,
      "loss": 0.8457,
      "step": 8909
    },
    {
      "epoch": 0.4903957289889372,
      "grad_norm": 0.9364853501319885,
      "learning_rate": 8.6111744569439e-06,
      "loss": 0.9114,
      "step": 8910
    },
    {
      "epoch": 0.4904507677912929,
      "grad_norm": 0.7584181427955627,
      "learning_rate": 8.610874637116099e-06,
      "loss": 0.6852,
      "step": 8911
    },
    {
      "epoch": 0.4905058065936485,
      "grad_norm": 0.7326254844665527,
      "learning_rate": 8.610574790149929e-06,
      "loss": 0.7843,
      "step": 8912
    },
    {
      "epoch": 0.4905608453960042,
      "grad_norm": 0.918258547782898,
      "learning_rate": 8.610274916047645e-06,
      "loss": 0.766,
      "step": 8913
    },
    {
      "epoch": 0.49061588419835983,
      "grad_norm": 1.0083420276641846,
      "learning_rate": 8.609975014811502e-06,
      "loss": 0.7436,
      "step": 8914
    },
    {
      "epoch": 0.4906709230007155,
      "grad_norm": 0.712664783000946,
      "learning_rate": 8.609675086443752e-06,
      "loss": 0.7891,
      "step": 8915
    },
    {
      "epoch": 0.49072596180307115,
      "grad_norm": 0.7635206580162048,
      "learning_rate": 8.609375130946651e-06,
      "loss": 0.7842,
      "step": 8916
    },
    {
      "epoch": 0.49078100060542684,
      "grad_norm": 0.7567723989486694,
      "learning_rate": 8.609075148322452e-06,
      "loss": 0.8435,
      "step": 8917
    },
    {
      "epoch": 0.4908360394077825,
      "grad_norm": 0.8918718099594116,
      "learning_rate": 8.60877513857341e-06,
      "loss": 0.8015,
      "step": 8918
    },
    {
      "epoch": 0.49089107821013817,
      "grad_norm": 0.8701914548873901,
      "learning_rate": 8.608475101701781e-06,
      "loss": 0.7806,
      "step": 8919
    },
    {
      "epoch": 0.4909461170124938,
      "grad_norm": 0.7528215646743774,
      "learning_rate": 8.608175037709819e-06,
      "loss": 0.7958,
      "step": 8920
    },
    {
      "epoch": 0.4910011558148495,
      "grad_norm": 0.7277387380599976,
      "learning_rate": 8.60787494659978e-06,
      "loss": 0.7878,
      "step": 8921
    },
    {
      "epoch": 0.4910561946172051,
      "grad_norm": 0.6739892959594727,
      "learning_rate": 8.607574828373917e-06,
      "loss": 0.7212,
      "step": 8922
    },
    {
      "epoch": 0.4911112334195608,
      "grad_norm": 0.712480366230011,
      "learning_rate": 8.607274683034487e-06,
      "loss": 0.7966,
      "step": 8923
    },
    {
      "epoch": 0.49116627222191644,
      "grad_norm": 0.7192126512527466,
      "learning_rate": 8.606974510583747e-06,
      "loss": 0.7032,
      "step": 8924
    },
    {
      "epoch": 0.49122131102427213,
      "grad_norm": 0.7502614855766296,
      "learning_rate": 8.606674311023953e-06,
      "loss": 0.7465,
      "step": 8925
    },
    {
      "epoch": 0.49127634982662777,
      "grad_norm": 0.8475236892700195,
      "learning_rate": 8.606374084357361e-06,
      "loss": 0.8083,
      "step": 8926
    },
    {
      "epoch": 0.49133138862898346,
      "grad_norm": 0.6972761750221252,
      "learning_rate": 8.606073830586224e-06,
      "loss": 0.7206,
      "step": 8927
    },
    {
      "epoch": 0.4913864274313391,
      "grad_norm": 0.6209561824798584,
      "learning_rate": 8.605773549712803e-06,
      "loss": 0.6664,
      "step": 8928
    },
    {
      "epoch": 0.4914414662336948,
      "grad_norm": 0.7905771732330322,
      "learning_rate": 8.605473241739353e-06,
      "loss": 0.7243,
      "step": 8929
    },
    {
      "epoch": 0.4914965050360504,
      "grad_norm": 0.762959897518158,
      "learning_rate": 8.605172906668131e-06,
      "loss": 0.7747,
      "step": 8930
    },
    {
      "epoch": 0.4915515438384061,
      "grad_norm": 0.7297530174255371,
      "learning_rate": 8.604872544501394e-06,
      "loss": 0.7441,
      "step": 8931
    },
    {
      "epoch": 0.49160658264076174,
      "grad_norm": 0.6732318997383118,
      "learning_rate": 8.6045721552414e-06,
      "loss": 0.7621,
      "step": 8932
    },
    {
      "epoch": 0.4916616214431174,
      "grad_norm": 0.7010045647621155,
      "learning_rate": 8.604271738890407e-06,
      "loss": 0.7971,
      "step": 8933
    },
    {
      "epoch": 0.49171666024547306,
      "grad_norm": 0.6996648907661438,
      "learning_rate": 8.603971295450672e-06,
      "loss": 0.8119,
      "step": 8934
    },
    {
      "epoch": 0.4917716990478287,
      "grad_norm": 0.7679941058158875,
      "learning_rate": 8.603670824924456e-06,
      "loss": 0.8035,
      "step": 8935
    },
    {
      "epoch": 0.4918267378501844,
      "grad_norm": 0.8009630441665649,
      "learning_rate": 8.603370327314011e-06,
      "loss": 0.7817,
      "step": 8936
    },
    {
      "epoch": 0.49188177665254,
      "grad_norm": 0.7167709469795227,
      "learning_rate": 8.603069802621601e-06,
      "loss": 0.7621,
      "step": 8937
    },
    {
      "epoch": 0.4919368154548957,
      "grad_norm": 0.7447960376739502,
      "learning_rate": 8.602769250849483e-06,
      "loss": 0.7664,
      "step": 8938
    },
    {
      "epoch": 0.49199185425725134,
      "grad_norm": 0.653131365776062,
      "learning_rate": 8.602468671999915e-06,
      "loss": 0.6927,
      "step": 8939
    },
    {
      "epoch": 0.492046893059607,
      "grad_norm": 0.6758691072463989,
      "learning_rate": 8.602168066075158e-06,
      "loss": 0.7519,
      "step": 8940
    },
    {
      "epoch": 0.49210193186196266,
      "grad_norm": 0.9186220765113831,
      "learning_rate": 8.60186743307747e-06,
      "loss": 0.7265,
      "step": 8941
    },
    {
      "epoch": 0.49215697066431835,
      "grad_norm": 0.6781855225563049,
      "learning_rate": 8.60156677300911e-06,
      "loss": 0.6719,
      "step": 8942
    },
    {
      "epoch": 0.492212009466674,
      "grad_norm": 0.7262865304946899,
      "learning_rate": 8.601266085872336e-06,
      "loss": 0.6449,
      "step": 8943
    },
    {
      "epoch": 0.4922670482690297,
      "grad_norm": 0.6877585053443909,
      "learning_rate": 8.600965371669411e-06,
      "loss": 0.6999,
      "step": 8944
    },
    {
      "epoch": 0.4923220870713853,
      "grad_norm": 1.1133443117141724,
      "learning_rate": 8.600664630402596e-06,
      "loss": 0.7842,
      "step": 8945
    },
    {
      "epoch": 0.492377125873741,
      "grad_norm": 0.643478274345398,
      "learning_rate": 8.600363862074149e-06,
      "loss": 0.7009,
      "step": 8946
    },
    {
      "epoch": 0.49243216467609663,
      "grad_norm": 0.7692574262619019,
      "learning_rate": 8.600063066686331e-06,
      "loss": 0.7777,
      "step": 8947
    },
    {
      "epoch": 0.4924872034784523,
      "grad_norm": 0.884963870048523,
      "learning_rate": 8.599762244241403e-06,
      "loss": 0.7789,
      "step": 8948
    },
    {
      "epoch": 0.49254224228080795,
      "grad_norm": 0.6918813586235046,
      "learning_rate": 8.599461394741624e-06,
      "loss": 0.7769,
      "step": 8949
    },
    {
      "epoch": 0.49259728108316364,
      "grad_norm": 0.7432044148445129,
      "learning_rate": 8.599160518189258e-06,
      "loss": 0.7972,
      "step": 8950
    },
    {
      "epoch": 0.4926523198855193,
      "grad_norm": 0.7530491948127747,
      "learning_rate": 8.598859614586564e-06,
      "loss": 0.8812,
      "step": 8951
    },
    {
      "epoch": 0.49270735868787496,
      "grad_norm": 0.8738592267036438,
      "learning_rate": 8.598558683935806e-06,
      "loss": 0.6967,
      "step": 8952
    },
    {
      "epoch": 0.4927623974902306,
      "grad_norm": 1.032084584236145,
      "learning_rate": 8.598257726239242e-06,
      "loss": 0.8513,
      "step": 8953
    },
    {
      "epoch": 0.4928174362925863,
      "grad_norm": 0.8717961311340332,
      "learning_rate": 8.597956741499136e-06,
      "loss": 0.7703,
      "step": 8954
    },
    {
      "epoch": 0.4928724750949419,
      "grad_norm": 0.6788356900215149,
      "learning_rate": 8.597655729717753e-06,
      "loss": 0.7649,
      "step": 8955
    },
    {
      "epoch": 0.4929275138972976,
      "grad_norm": 1.0595613718032837,
      "learning_rate": 8.59735469089735e-06,
      "loss": 0.6967,
      "step": 8956
    },
    {
      "epoch": 0.49298255269965324,
      "grad_norm": 0.7583820819854736,
      "learning_rate": 8.597053625040193e-06,
      "loss": 0.8384,
      "step": 8957
    },
    {
      "epoch": 0.49303759150200893,
      "grad_norm": 0.7232168912887573,
      "learning_rate": 8.596752532148545e-06,
      "loss": 0.7643,
      "step": 8958
    },
    {
      "epoch": 0.49309263030436457,
      "grad_norm": 0.727190375328064,
      "learning_rate": 8.596451412224666e-06,
      "loss": 0.845,
      "step": 8959
    },
    {
      "epoch": 0.49314766910672025,
      "grad_norm": 0.6844252347946167,
      "learning_rate": 8.596150265270821e-06,
      "loss": 0.7099,
      "step": 8960
    },
    {
      "epoch": 0.4932027079090759,
      "grad_norm": 0.7379910945892334,
      "learning_rate": 8.595849091289275e-06,
      "loss": 0.8168,
      "step": 8961
    },
    {
      "epoch": 0.4932577467114316,
      "grad_norm": 0.77718186378479,
      "learning_rate": 8.595547890282288e-06,
      "loss": 0.8457,
      "step": 8962
    },
    {
      "epoch": 0.4933127855137872,
      "grad_norm": 0.686126172542572,
      "learning_rate": 8.595246662252127e-06,
      "loss": 0.7918,
      "step": 8963
    },
    {
      "epoch": 0.4933678243161429,
      "grad_norm": 0.7406145930290222,
      "learning_rate": 8.594945407201051e-06,
      "loss": 0.6866,
      "step": 8964
    },
    {
      "epoch": 0.49342286311849853,
      "grad_norm": 0.9543277025222778,
      "learning_rate": 8.594644125131331e-06,
      "loss": 0.8444,
      "step": 8965
    },
    {
      "epoch": 0.4934779019208542,
      "grad_norm": 0.8659517765045166,
      "learning_rate": 8.594342816045228e-06,
      "loss": 0.7661,
      "step": 8966
    },
    {
      "epoch": 0.49353294072320986,
      "grad_norm": 0.7289552092552185,
      "learning_rate": 8.594041479945005e-06,
      "loss": 0.7734,
      "step": 8967
    },
    {
      "epoch": 0.49358797952556555,
      "grad_norm": 0.7232840657234192,
      "learning_rate": 8.59374011683293e-06,
      "loss": 0.8557,
      "step": 8968
    },
    {
      "epoch": 0.4936430183279212,
      "grad_norm": 0.738684356212616,
      "learning_rate": 8.593438726711265e-06,
      "loss": 0.7779,
      "step": 8969
    },
    {
      "epoch": 0.49369805713027687,
      "grad_norm": 0.7486668229103088,
      "learning_rate": 8.593137309582276e-06,
      "loss": 0.7326,
      "step": 8970
    },
    {
      "epoch": 0.4937530959326325,
      "grad_norm": 0.6564297080039978,
      "learning_rate": 8.59283586544823e-06,
      "loss": 0.6927,
      "step": 8971
    },
    {
      "epoch": 0.4938081347349882,
      "grad_norm": 0.722540557384491,
      "learning_rate": 8.592534394311392e-06,
      "loss": 0.7254,
      "step": 8972
    },
    {
      "epoch": 0.4938631735373438,
      "grad_norm": 0.7466141581535339,
      "learning_rate": 8.592232896174026e-06,
      "loss": 0.8551,
      "step": 8973
    },
    {
      "epoch": 0.4939182123396995,
      "grad_norm": 0.7819109559059143,
      "learning_rate": 8.591931371038398e-06,
      "loss": 0.7271,
      "step": 8974
    },
    {
      "epoch": 0.49397325114205515,
      "grad_norm": 0.7847672700881958,
      "learning_rate": 8.591629818906776e-06,
      "loss": 0.8404,
      "step": 8975
    },
    {
      "epoch": 0.49402828994441084,
      "grad_norm": 0.8167426586151123,
      "learning_rate": 8.591328239781428e-06,
      "loss": 0.7375,
      "step": 8976
    },
    {
      "epoch": 0.49408332874676647,
      "grad_norm": 0.7894755005836487,
      "learning_rate": 8.591026633664615e-06,
      "loss": 0.7872,
      "step": 8977
    },
    {
      "epoch": 0.4941383675491221,
      "grad_norm": 0.726204514503479,
      "learning_rate": 8.590725000558609e-06,
      "loss": 0.7289,
      "step": 8978
    },
    {
      "epoch": 0.4941934063514778,
      "grad_norm": 0.7116577625274658,
      "learning_rate": 8.590423340465675e-06,
      "loss": 0.7379,
      "step": 8979
    },
    {
      "epoch": 0.4942484451538334,
      "grad_norm": 0.7302193641662598,
      "learning_rate": 8.59012165338808e-06,
      "loss": 0.7951,
      "step": 8980
    },
    {
      "epoch": 0.4943034839561891,
      "grad_norm": 0.680555522441864,
      "learning_rate": 8.58981993932809e-06,
      "loss": 0.7609,
      "step": 8981
    },
    {
      "epoch": 0.49435852275854475,
      "grad_norm": 0.874546229839325,
      "learning_rate": 8.589518198287976e-06,
      "loss": 0.8025,
      "step": 8982
    },
    {
      "epoch": 0.49441356156090044,
      "grad_norm": 0.7164583206176758,
      "learning_rate": 8.589216430270004e-06,
      "loss": 0.7466,
      "step": 8983
    },
    {
      "epoch": 0.49446860036325607,
      "grad_norm": 0.9155141115188599,
      "learning_rate": 8.588914635276442e-06,
      "loss": 0.7896,
      "step": 8984
    },
    {
      "epoch": 0.49452363916561176,
      "grad_norm": 0.6777059435844421,
      "learning_rate": 8.588612813309558e-06,
      "loss": 0.7468,
      "step": 8985
    },
    {
      "epoch": 0.4945786779679674,
      "grad_norm": 0.7100371718406677,
      "learning_rate": 8.58831096437162e-06,
      "loss": 0.7216,
      "step": 8986
    },
    {
      "epoch": 0.4946337167703231,
      "grad_norm": 0.6842584609985352,
      "learning_rate": 8.5880090884649e-06,
      "loss": 0.7103,
      "step": 8987
    },
    {
      "epoch": 0.4946887555726787,
      "grad_norm": 0.6347573399543762,
      "learning_rate": 8.587707185591661e-06,
      "loss": 0.7103,
      "step": 8988
    },
    {
      "epoch": 0.4947437943750344,
      "grad_norm": 0.7175829410552979,
      "learning_rate": 8.587405255754177e-06,
      "loss": 0.8375,
      "step": 8989
    },
    {
      "epoch": 0.49479883317739004,
      "grad_norm": 0.8402735590934753,
      "learning_rate": 8.587103298954715e-06,
      "loss": 0.6841,
      "step": 8990
    },
    {
      "epoch": 0.49485387197974573,
      "grad_norm": 0.6988743543624878,
      "learning_rate": 8.586801315195545e-06,
      "loss": 0.7637,
      "step": 8991
    },
    {
      "epoch": 0.49490891078210136,
      "grad_norm": 0.6672561168670654,
      "learning_rate": 8.586499304478934e-06,
      "loss": 0.7103,
      "step": 8992
    },
    {
      "epoch": 0.49496394958445705,
      "grad_norm": 0.6821330189704895,
      "learning_rate": 8.586197266807158e-06,
      "loss": 0.6881,
      "step": 8993
    },
    {
      "epoch": 0.4950189883868127,
      "grad_norm": 0.7886170744895935,
      "learning_rate": 8.585895202182482e-06,
      "loss": 0.7892,
      "step": 8994
    },
    {
      "epoch": 0.4950740271891684,
      "grad_norm": 0.7348074913024902,
      "learning_rate": 8.585593110607177e-06,
      "loss": 0.7835,
      "step": 8995
    },
    {
      "epoch": 0.495129065991524,
      "grad_norm": 0.9375506639480591,
      "learning_rate": 8.585290992083514e-06,
      "loss": 0.8017,
      "step": 8996
    },
    {
      "epoch": 0.4951841047938797,
      "grad_norm": 0.7442331910133362,
      "learning_rate": 8.584988846613765e-06,
      "loss": 0.72,
      "step": 8997
    },
    {
      "epoch": 0.49523914359623533,
      "grad_norm": 0.7347918748855591,
      "learning_rate": 8.584686674200197e-06,
      "loss": 0.8229,
      "step": 8998
    },
    {
      "epoch": 0.495294182398591,
      "grad_norm": 0.7168740630149841,
      "learning_rate": 8.584384474845084e-06,
      "loss": 0.7288,
      "step": 8999
    },
    {
      "epoch": 0.49534922120094665,
      "grad_norm": 0.7834853529930115,
      "learning_rate": 8.584082248550697e-06,
      "loss": 0.8521,
      "step": 9000
    },
    {
      "epoch": 0.49540426000330234,
      "grad_norm": 0.6499035358428955,
      "learning_rate": 8.58377999531931e-06,
      "loss": 0.6887,
      "step": 9001
    },
    {
      "epoch": 0.495459298805658,
      "grad_norm": 0.8000181913375854,
      "learning_rate": 8.583477715153189e-06,
      "loss": 0.8688,
      "step": 9002
    },
    {
      "epoch": 0.49551433760801367,
      "grad_norm": 0.7539342045783997,
      "learning_rate": 8.58317540805461e-06,
      "loss": 0.6151,
      "step": 9003
    },
    {
      "epoch": 0.4955693764103693,
      "grad_norm": 0.7677812576293945,
      "learning_rate": 8.582873074025841e-06,
      "loss": 0.8168,
      "step": 9004
    },
    {
      "epoch": 0.495624415212725,
      "grad_norm": 0.7679157853126526,
      "learning_rate": 8.58257071306916e-06,
      "loss": 0.7719,
      "step": 9005
    },
    {
      "epoch": 0.4956794540150806,
      "grad_norm": 0.9745703935623169,
      "learning_rate": 8.582268325186836e-06,
      "loss": 0.8272,
      "step": 9006
    },
    {
      "epoch": 0.4957344928174363,
      "grad_norm": 0.66932612657547,
      "learning_rate": 8.581965910381143e-06,
      "loss": 0.7256,
      "step": 9007
    },
    {
      "epoch": 0.49578953161979195,
      "grad_norm": 0.7630981206893921,
      "learning_rate": 8.581663468654351e-06,
      "loss": 0.7594,
      "step": 9008
    },
    {
      "epoch": 0.49584457042214763,
      "grad_norm": 0.7420778870582581,
      "learning_rate": 8.581361000008737e-06,
      "loss": 0.7834,
      "step": 9009
    },
    {
      "epoch": 0.49589960922450327,
      "grad_norm": 0.6775205731391907,
      "learning_rate": 8.58105850444657e-06,
      "loss": 0.7609,
      "step": 9010
    },
    {
      "epoch": 0.49595464802685896,
      "grad_norm": 0.6588264107704163,
      "learning_rate": 8.580755981970128e-06,
      "loss": 0.805,
      "step": 9011
    },
    {
      "epoch": 0.4960096868292146,
      "grad_norm": 0.7325689196586609,
      "learning_rate": 8.580453432581681e-06,
      "loss": 0.8817,
      "step": 9012
    },
    {
      "epoch": 0.4960647256315703,
      "grad_norm": 0.7319273948669434,
      "learning_rate": 8.580150856283505e-06,
      "loss": 0.8001,
      "step": 9013
    },
    {
      "epoch": 0.4961197644339259,
      "grad_norm": 0.7841789126396179,
      "learning_rate": 8.579848253077875e-06,
      "loss": 0.8415,
      "step": 9014
    },
    {
      "epoch": 0.4961748032362816,
      "grad_norm": 0.7593979239463806,
      "learning_rate": 8.579545622967062e-06,
      "loss": 0.8238,
      "step": 9015
    },
    {
      "epoch": 0.49622984203863724,
      "grad_norm": 0.6938808560371399,
      "learning_rate": 8.579242965953343e-06,
      "loss": 0.7325,
      "step": 9016
    },
    {
      "epoch": 0.4962848808409929,
      "grad_norm": 0.7907594442367554,
      "learning_rate": 8.578940282038993e-06,
      "loss": 0.6947,
      "step": 9017
    },
    {
      "epoch": 0.49633991964334856,
      "grad_norm": 0.708703875541687,
      "learning_rate": 8.578637571226283e-06,
      "loss": 0.6712,
      "step": 9018
    },
    {
      "epoch": 0.49639495844570425,
      "grad_norm": 0.6820377707481384,
      "learning_rate": 8.578334833517492e-06,
      "loss": 0.7269,
      "step": 9019
    },
    {
      "epoch": 0.4964499972480599,
      "grad_norm": 0.6858653426170349,
      "learning_rate": 8.578032068914896e-06,
      "loss": 0.7325,
      "step": 9020
    },
    {
      "epoch": 0.4965050360504155,
      "grad_norm": 0.8758736848831177,
      "learning_rate": 8.577729277420768e-06,
      "loss": 0.6652,
      "step": 9021
    },
    {
      "epoch": 0.4965600748527712,
      "grad_norm": 0.731316328048706,
      "learning_rate": 8.577426459037383e-06,
      "loss": 0.7835,
      "step": 9022
    },
    {
      "epoch": 0.49661511365512684,
      "grad_norm": 0.813778817653656,
      "learning_rate": 8.57712361376702e-06,
      "loss": 0.8025,
      "step": 9023
    },
    {
      "epoch": 0.4966701524574825,
      "grad_norm": 0.7167351841926575,
      "learning_rate": 8.576820741611952e-06,
      "loss": 0.7483,
      "step": 9024
    },
    {
      "epoch": 0.49672519125983816,
      "grad_norm": 0.7243192791938782,
      "learning_rate": 8.576517842574457e-06,
      "loss": 0.8411,
      "step": 9025
    },
    {
      "epoch": 0.49678023006219385,
      "grad_norm": 0.5869036316871643,
      "learning_rate": 8.576214916656814e-06,
      "loss": 0.6661,
      "step": 9026
    },
    {
      "epoch": 0.4968352688645495,
      "grad_norm": 0.7502203583717346,
      "learning_rate": 8.575911963861293e-06,
      "loss": 0.8838,
      "step": 9027
    },
    {
      "epoch": 0.4968903076669052,
      "grad_norm": 0.687562108039856,
      "learning_rate": 8.575608984190177e-06,
      "loss": 0.7446,
      "step": 9028
    },
    {
      "epoch": 0.4969453464692608,
      "grad_norm": 0.7735342383384705,
      "learning_rate": 8.57530597764574e-06,
      "loss": 0.8464,
      "step": 9029
    },
    {
      "epoch": 0.4970003852716165,
      "grad_norm": 0.7828487753868103,
      "learning_rate": 8.575002944230261e-06,
      "loss": 0.7504,
      "step": 9030
    },
    {
      "epoch": 0.49705542407397213,
      "grad_norm": 0.6359286904335022,
      "learning_rate": 8.574699883946018e-06,
      "loss": 0.6805,
      "step": 9031
    },
    {
      "epoch": 0.4971104628763278,
      "grad_norm": 0.7462830543518066,
      "learning_rate": 8.574396796795285e-06,
      "loss": 0.8317,
      "step": 9032
    },
    {
      "epoch": 0.49716550167868345,
      "grad_norm": 0.705115795135498,
      "learning_rate": 8.574093682780344e-06,
      "loss": 0.7401,
      "step": 9033
    },
    {
      "epoch": 0.49722054048103914,
      "grad_norm": 0.6466538310050964,
      "learning_rate": 8.573790541903472e-06,
      "loss": 0.7761,
      "step": 9034
    },
    {
      "epoch": 0.4972755792833948,
      "grad_norm": 0.7479867339134216,
      "learning_rate": 8.573487374166946e-06,
      "loss": 0.8394,
      "step": 9035
    },
    {
      "epoch": 0.49733061808575046,
      "grad_norm": 0.7378019094467163,
      "learning_rate": 8.573184179573046e-06,
      "loss": 0.8215,
      "step": 9036
    },
    {
      "epoch": 0.4973856568881061,
      "grad_norm": 0.6526094675064087,
      "learning_rate": 8.57288095812405e-06,
      "loss": 0.8055,
      "step": 9037
    },
    {
      "epoch": 0.4974406956904618,
      "grad_norm": 0.679595947265625,
      "learning_rate": 8.572577709822238e-06,
      "loss": 0.8241,
      "step": 9038
    },
    {
      "epoch": 0.4974957344928174,
      "grad_norm": 0.753466010093689,
      "learning_rate": 8.572274434669886e-06,
      "loss": 0.896,
      "step": 9039
    },
    {
      "epoch": 0.4975507732951731,
      "grad_norm": 0.7068368792533875,
      "learning_rate": 8.571971132669277e-06,
      "loss": 0.778,
      "step": 9040
    },
    {
      "epoch": 0.49760581209752874,
      "grad_norm": 0.7397973537445068,
      "learning_rate": 8.571667803822689e-06,
      "loss": 0.782,
      "step": 9041
    },
    {
      "epoch": 0.49766085089988443,
      "grad_norm": 0.7837033271789551,
      "learning_rate": 8.571364448132402e-06,
      "loss": 0.7509,
      "step": 9042
    },
    {
      "epoch": 0.49771588970224007,
      "grad_norm": 0.6808765530586243,
      "learning_rate": 8.571061065600696e-06,
      "loss": 0.672,
      "step": 9043
    },
    {
      "epoch": 0.49777092850459576,
      "grad_norm": 0.6574100255966187,
      "learning_rate": 8.570757656229852e-06,
      "loss": 0.751,
      "step": 9044
    },
    {
      "epoch": 0.4978259673069514,
      "grad_norm": 0.7357671856880188,
      "learning_rate": 8.570454220022146e-06,
      "loss": 0.7977,
      "step": 9045
    },
    {
      "epoch": 0.4978810061093071,
      "grad_norm": 0.7937216758728027,
      "learning_rate": 8.570150756979865e-06,
      "loss": 0.8151,
      "step": 9046
    },
    {
      "epoch": 0.4979360449116627,
      "grad_norm": 0.7050907611846924,
      "learning_rate": 8.569847267105285e-06,
      "loss": 0.7667,
      "step": 9047
    },
    {
      "epoch": 0.4979910837140184,
      "grad_norm": 0.7105300426483154,
      "learning_rate": 8.569543750400688e-06,
      "loss": 0.7031,
      "step": 9048
    },
    {
      "epoch": 0.49804612251637403,
      "grad_norm": 0.7174646854400635,
      "learning_rate": 8.569240206868358e-06,
      "loss": 0.7692,
      "step": 9049
    },
    {
      "epoch": 0.4981011613187297,
      "grad_norm": 0.7525906562805176,
      "learning_rate": 8.568936636510573e-06,
      "loss": 0.7584,
      "step": 9050
    },
    {
      "epoch": 0.49815620012108536,
      "grad_norm": 1.5518100261688232,
      "learning_rate": 8.568633039329615e-06,
      "loss": 0.7932,
      "step": 9051
    },
    {
      "epoch": 0.49821123892344105,
      "grad_norm": 0.7037720084190369,
      "learning_rate": 8.568329415327766e-06,
      "loss": 0.8345,
      "step": 9052
    },
    {
      "epoch": 0.4982662777257967,
      "grad_norm": 0.6422694325447083,
      "learning_rate": 8.568025764507308e-06,
      "loss": 0.7396,
      "step": 9053
    },
    {
      "epoch": 0.49832131652815237,
      "grad_norm": 0.777306854724884,
      "learning_rate": 8.567722086870525e-06,
      "loss": 0.8605,
      "step": 9054
    },
    {
      "epoch": 0.498376355330508,
      "grad_norm": 0.6619865298271179,
      "learning_rate": 8.567418382419697e-06,
      "loss": 0.7395,
      "step": 9055
    },
    {
      "epoch": 0.4984313941328637,
      "grad_norm": 0.7214456796646118,
      "learning_rate": 8.567114651157106e-06,
      "loss": 0.7932,
      "step": 9056
    },
    {
      "epoch": 0.4984864329352193,
      "grad_norm": 0.75806725025177,
      "learning_rate": 8.566810893085037e-06,
      "loss": 0.7998,
      "step": 9057
    },
    {
      "epoch": 0.498541471737575,
      "grad_norm": 0.8089895844459534,
      "learning_rate": 8.566507108205773e-06,
      "loss": 0.7849,
      "step": 9058
    },
    {
      "epoch": 0.49859651053993065,
      "grad_norm": 0.817814290523529,
      "learning_rate": 8.566203296521597e-06,
      "loss": 0.7261,
      "step": 9059
    },
    {
      "epoch": 0.49865154934228634,
      "grad_norm": 0.7417539954185486,
      "learning_rate": 8.56589945803479e-06,
      "loss": 0.7087,
      "step": 9060
    },
    {
      "epoch": 0.49870658814464197,
      "grad_norm": 0.7518000602722168,
      "learning_rate": 8.565595592747639e-06,
      "loss": 0.7245,
      "step": 9061
    },
    {
      "epoch": 0.49876162694699766,
      "grad_norm": 0.9537304043769836,
      "learning_rate": 8.565291700662423e-06,
      "loss": 0.901,
      "step": 9062
    },
    {
      "epoch": 0.4988166657493533,
      "grad_norm": 0.784545361995697,
      "learning_rate": 8.56498778178143e-06,
      "loss": 0.7813,
      "step": 9063
    },
    {
      "epoch": 0.4988717045517089,
      "grad_norm": 0.9218429923057556,
      "learning_rate": 8.564683836106945e-06,
      "loss": 0.8452,
      "step": 9064
    },
    {
      "epoch": 0.4989267433540646,
      "grad_norm": 0.6902065277099609,
      "learning_rate": 8.56437986364125e-06,
      "loss": 0.7527,
      "step": 9065
    },
    {
      "epoch": 0.49898178215642025,
      "grad_norm": 0.7388677000999451,
      "learning_rate": 8.56407586438663e-06,
      "loss": 0.82,
      "step": 9066
    },
    {
      "epoch": 0.49903682095877594,
      "grad_norm": 0.6959313154220581,
      "learning_rate": 8.563771838345369e-06,
      "loss": 0.7274,
      "step": 9067
    },
    {
      "epoch": 0.4990918597611316,
      "grad_norm": 0.6582610607147217,
      "learning_rate": 8.563467785519753e-06,
      "loss": 0.6518,
      "step": 9068
    },
    {
      "epoch": 0.49914689856348726,
      "grad_norm": 0.6525924801826477,
      "learning_rate": 8.563163705912066e-06,
      "loss": 0.7006,
      "step": 9069
    },
    {
      "epoch": 0.4992019373658429,
      "grad_norm": 0.8092843890190125,
      "learning_rate": 8.562859599524596e-06,
      "loss": 0.6915,
      "step": 9070
    },
    {
      "epoch": 0.4992569761681986,
      "grad_norm": 0.6540575623512268,
      "learning_rate": 8.562555466359626e-06,
      "loss": 0.6729,
      "step": 9071
    },
    {
      "epoch": 0.4993120149705542,
      "grad_norm": 0.8220445513725281,
      "learning_rate": 8.562251306419443e-06,
      "loss": 0.8172,
      "step": 9072
    },
    {
      "epoch": 0.4993670537729099,
      "grad_norm": 0.7461502552032471,
      "learning_rate": 8.561947119706334e-06,
      "loss": 0.6902,
      "step": 9073
    },
    {
      "epoch": 0.49942209257526554,
      "grad_norm": 0.8166316151618958,
      "learning_rate": 8.56164290622258e-06,
      "loss": 0.8238,
      "step": 9074
    },
    {
      "epoch": 0.49947713137762123,
      "grad_norm": 0.8453896641731262,
      "learning_rate": 8.561338665970476e-06,
      "loss": 0.7697,
      "step": 9075
    },
    {
      "epoch": 0.49953217017997686,
      "grad_norm": 0.7606340050697327,
      "learning_rate": 8.5610343989523e-06,
      "loss": 0.6951,
      "step": 9076
    },
    {
      "epoch": 0.49958720898233255,
      "grad_norm": 0.7408013343811035,
      "learning_rate": 8.560730105170345e-06,
      "loss": 0.8298,
      "step": 9077
    },
    {
      "epoch": 0.4996422477846882,
      "grad_norm": 0.7625541090965271,
      "learning_rate": 8.560425784626896e-06,
      "loss": 0.6738,
      "step": 9078
    },
    {
      "epoch": 0.4996972865870439,
      "grad_norm": 0.6940996646881104,
      "learning_rate": 8.560121437324238e-06,
      "loss": 0.78,
      "step": 9079
    },
    {
      "epoch": 0.4997523253893995,
      "grad_norm": 0.8087461590766907,
      "learning_rate": 8.559817063264661e-06,
      "loss": 0.7831,
      "step": 9080
    },
    {
      "epoch": 0.4998073641917552,
      "grad_norm": 0.7418510317802429,
      "learning_rate": 8.559512662450452e-06,
      "loss": 0.801,
      "step": 9081
    },
    {
      "epoch": 0.49986240299411083,
      "grad_norm": 0.6793946027755737,
      "learning_rate": 8.5592082348839e-06,
      "loss": 0.7329,
      "step": 9082
    },
    {
      "epoch": 0.4999174417964665,
      "grad_norm": 0.8197429180145264,
      "learning_rate": 8.55890378056729e-06,
      "loss": 0.804,
      "step": 9083
    },
    {
      "epoch": 0.49997248059882216,
      "grad_norm": 0.7526460886001587,
      "learning_rate": 8.558599299502912e-06,
      "loss": 0.8378,
      "step": 9084
    },
    {
      "epoch": 0.5000275194011778,
      "grad_norm": 0.8169133067131042,
      "learning_rate": 8.558294791693055e-06,
      "loss": 0.828,
      "step": 9085
    },
    {
      "epoch": 0.5000825582035335,
      "grad_norm": 0.8386932015419006,
      "learning_rate": 8.557990257140007e-06,
      "loss": 0.7961,
      "step": 9086
    },
    {
      "epoch": 0.5001375970058891,
      "grad_norm": 0.7183443903923035,
      "learning_rate": 8.557685695846057e-06,
      "loss": 0.6964,
      "step": 9087
    },
    {
      "epoch": 0.5001926358082448,
      "grad_norm": 0.77079176902771,
      "learning_rate": 8.557381107813491e-06,
      "loss": 0.8222,
      "step": 9088
    },
    {
      "epoch": 0.5002476746106005,
      "grad_norm": 0.6519342660903931,
      "learning_rate": 8.557076493044603e-06,
      "loss": 0.772,
      "step": 9089
    },
    {
      "epoch": 0.5003027134129562,
      "grad_norm": 0.7039975523948669,
      "learning_rate": 8.556771851541678e-06,
      "loss": 0.7491,
      "step": 9090
    },
    {
      "epoch": 0.5003577522153118,
      "grad_norm": 0.6459039449691772,
      "learning_rate": 8.556467183307012e-06,
      "loss": 0.7104,
      "step": 9091
    },
    {
      "epoch": 0.5004127910176674,
      "grad_norm": 0.7359183430671692,
      "learning_rate": 8.556162488342887e-06,
      "loss": 0.829,
      "step": 9092
    },
    {
      "epoch": 0.5004678298200231,
      "grad_norm": 0.7029602527618408,
      "learning_rate": 8.555857766651599e-06,
      "loss": 0.8163,
      "step": 9093
    },
    {
      "epoch": 0.5005228686223788,
      "grad_norm": 0.6687049865722656,
      "learning_rate": 8.555553018235435e-06,
      "loss": 0.7589,
      "step": 9094
    },
    {
      "epoch": 0.5005779074247344,
      "grad_norm": 0.7277147173881531,
      "learning_rate": 8.555248243096686e-06,
      "loss": 0.8334,
      "step": 9095
    },
    {
      "epoch": 0.5006329462270901,
      "grad_norm": 0.6512065529823303,
      "learning_rate": 8.554943441237642e-06,
      "loss": 0.7174,
      "step": 9096
    },
    {
      "epoch": 0.5006879850294458,
      "grad_norm": 0.725351095199585,
      "learning_rate": 8.554638612660594e-06,
      "loss": 0.6514,
      "step": 9097
    },
    {
      "epoch": 0.5007430238318015,
      "grad_norm": 0.7983208894729614,
      "learning_rate": 8.554333757367836e-06,
      "loss": 0.8385,
      "step": 9098
    },
    {
      "epoch": 0.500798062634157,
      "grad_norm": 0.6631388068199158,
      "learning_rate": 8.554028875361657e-06,
      "loss": 0.7103,
      "step": 9099
    },
    {
      "epoch": 0.5008531014365127,
      "grad_norm": 0.730421245098114,
      "learning_rate": 8.553723966644347e-06,
      "loss": 0.8005,
      "step": 9100
    },
    {
      "epoch": 0.5009081402388684,
      "grad_norm": 0.7385838627815247,
      "learning_rate": 8.5534190312182e-06,
      "loss": 0.7586,
      "step": 9101
    },
    {
      "epoch": 0.5009631790412241,
      "grad_norm": 0.712458610534668,
      "learning_rate": 8.553114069085506e-06,
      "loss": 0.7587,
      "step": 9102
    },
    {
      "epoch": 0.5010182178435797,
      "grad_norm": 0.7393542528152466,
      "learning_rate": 8.552809080248559e-06,
      "loss": 0.746,
      "step": 9103
    },
    {
      "epoch": 0.5010732566459354,
      "grad_norm": 0.6596370935440063,
      "learning_rate": 8.552504064709649e-06,
      "loss": 0.6968,
      "step": 9104
    },
    {
      "epoch": 0.5011282954482911,
      "grad_norm": 0.7340545654296875,
      "learning_rate": 8.552199022471069e-06,
      "loss": 0.8326,
      "step": 9105
    },
    {
      "epoch": 0.5011833342506467,
      "grad_norm": 0.6586140990257263,
      "learning_rate": 8.55189395353511e-06,
      "loss": 0.7144,
      "step": 9106
    },
    {
      "epoch": 0.5012383730530023,
      "grad_norm": 0.6875959038734436,
      "learning_rate": 8.551588857904071e-06,
      "loss": 0.721,
      "step": 9107
    },
    {
      "epoch": 0.501293411855358,
      "grad_norm": 0.6754499077796936,
      "learning_rate": 8.551283735580238e-06,
      "loss": 0.6771,
      "step": 9108
    },
    {
      "epoch": 0.5013484506577137,
      "grad_norm": 0.8027325868606567,
      "learning_rate": 8.55097858656591e-06,
      "loss": 0.8196,
      "step": 9109
    },
    {
      "epoch": 0.5014034894600693,
      "grad_norm": 0.6992260217666626,
      "learning_rate": 8.550673410863376e-06,
      "loss": 0.7923,
      "step": 9110
    },
    {
      "epoch": 0.501458528262425,
      "grad_norm": 0.741205632686615,
      "learning_rate": 8.550368208474928e-06,
      "loss": 0.7036,
      "step": 9111
    },
    {
      "epoch": 0.5015135670647807,
      "grad_norm": 0.6485981345176697,
      "learning_rate": 8.550062979402866e-06,
      "loss": 0.6351,
      "step": 9112
    },
    {
      "epoch": 0.5015686058671364,
      "grad_norm": 0.6984226703643799,
      "learning_rate": 8.549757723649481e-06,
      "loss": 0.7714,
      "step": 9113
    },
    {
      "epoch": 0.5016236446694919,
      "grad_norm": 0.7773998975753784,
      "learning_rate": 8.549452441217067e-06,
      "loss": 0.8901,
      "step": 9114
    },
    {
      "epoch": 0.5016786834718476,
      "grad_norm": 0.6912227272987366,
      "learning_rate": 8.549147132107918e-06,
      "loss": 0.7702,
      "step": 9115
    },
    {
      "epoch": 0.5017337222742033,
      "grad_norm": 0.6742583513259888,
      "learning_rate": 8.54884179632433e-06,
      "loss": 0.7789,
      "step": 9116
    },
    {
      "epoch": 0.501788761076559,
      "grad_norm": 0.7896195650100708,
      "learning_rate": 8.548536433868595e-06,
      "loss": 0.7358,
      "step": 9117
    },
    {
      "epoch": 0.5018437998789146,
      "grad_norm": 0.7112523913383484,
      "learning_rate": 8.548231044743011e-06,
      "loss": 0.7286,
      "step": 9118
    },
    {
      "epoch": 0.5018988386812703,
      "grad_norm": 0.9162774085998535,
      "learning_rate": 8.547925628949873e-06,
      "loss": 0.935,
      "step": 9119
    },
    {
      "epoch": 0.501953877483626,
      "grad_norm": 0.6319599747657776,
      "learning_rate": 8.547620186491477e-06,
      "loss": 0.625,
      "step": 9120
    },
    {
      "epoch": 0.5020089162859817,
      "grad_norm": 0.7074719667434692,
      "learning_rate": 8.547314717370115e-06,
      "loss": 0.6614,
      "step": 9121
    },
    {
      "epoch": 0.5020639550883372,
      "grad_norm": 0.7417262196540833,
      "learning_rate": 8.547009221588086e-06,
      "loss": 0.8476,
      "step": 9122
    },
    {
      "epoch": 0.5021189938906929,
      "grad_norm": 0.7057339549064636,
      "learning_rate": 8.546703699147685e-06,
      "loss": 0.805,
      "step": 9123
    },
    {
      "epoch": 0.5021740326930486,
      "grad_norm": 0.7420887351036072,
      "learning_rate": 8.546398150051207e-06,
      "loss": 0.7331,
      "step": 9124
    },
    {
      "epoch": 0.5022290714954043,
      "grad_norm": 0.9526195526123047,
      "learning_rate": 8.546092574300953e-06,
      "loss": 0.7803,
      "step": 9125
    },
    {
      "epoch": 0.5022841102977599,
      "grad_norm": 0.748130202293396,
      "learning_rate": 8.545786971899214e-06,
      "loss": 0.7998,
      "step": 9126
    },
    {
      "epoch": 0.5023391491001156,
      "grad_norm": 0.7266026139259338,
      "learning_rate": 8.545481342848289e-06,
      "loss": 0.8377,
      "step": 9127
    },
    {
      "epoch": 0.5023941879024713,
      "grad_norm": 0.6762456893920898,
      "learning_rate": 8.545175687150478e-06,
      "loss": 0.7312,
      "step": 9128
    },
    {
      "epoch": 0.502449226704827,
      "grad_norm": 0.7011429667472839,
      "learning_rate": 8.544870004808072e-06,
      "loss": 0.7666,
      "step": 9129
    },
    {
      "epoch": 0.5025042655071825,
      "grad_norm": 0.6652229428291321,
      "learning_rate": 8.544564295823375e-06,
      "loss": 0.6904,
      "step": 9130
    },
    {
      "epoch": 0.5025593043095382,
      "grad_norm": 0.8333765268325806,
      "learning_rate": 8.54425856019868e-06,
      "loss": 0.7318,
      "step": 9131
    },
    {
      "epoch": 0.5026143431118939,
      "grad_norm": 0.6827245950698853,
      "learning_rate": 8.543952797936285e-06,
      "loss": 0.7692,
      "step": 9132
    },
    {
      "epoch": 0.5026693819142496,
      "grad_norm": 0.8744323253631592,
      "learning_rate": 8.543647009038491e-06,
      "loss": 0.7316,
      "step": 9133
    },
    {
      "epoch": 0.5027244207166052,
      "grad_norm": 0.7024276852607727,
      "learning_rate": 8.543341193507594e-06,
      "loss": 0.7008,
      "step": 9134
    },
    {
      "epoch": 0.5027794595189609,
      "grad_norm": 0.8786055445671082,
      "learning_rate": 8.543035351345895e-06,
      "loss": 0.7054,
      "step": 9135
    },
    {
      "epoch": 0.5028344983213165,
      "grad_norm": 0.727924108505249,
      "learning_rate": 8.54272948255569e-06,
      "loss": 0.8049,
      "step": 9136
    },
    {
      "epoch": 0.5028895371236722,
      "grad_norm": 0.8366256356239319,
      "learning_rate": 8.542423587139277e-06,
      "loss": 0.7926,
      "step": 9137
    },
    {
      "epoch": 0.5029445759260278,
      "grad_norm": 0.7657913565635681,
      "learning_rate": 8.542117665098958e-06,
      "loss": 0.8152,
      "step": 9138
    },
    {
      "epoch": 0.5029996147283835,
      "grad_norm": 0.7543498277664185,
      "learning_rate": 8.54181171643703e-06,
      "loss": 0.7566,
      "step": 9139
    },
    {
      "epoch": 0.5030546535307392,
      "grad_norm": 0.7771349549293518,
      "learning_rate": 8.541505741155794e-06,
      "loss": 0.7907,
      "step": 9140
    },
    {
      "epoch": 0.5031096923330949,
      "grad_norm": 0.6661877632141113,
      "learning_rate": 8.541199739257548e-06,
      "loss": 0.7481,
      "step": 9141
    },
    {
      "epoch": 0.5031647311354505,
      "grad_norm": 0.7700417637825012,
      "learning_rate": 8.540893710744593e-06,
      "loss": 0.7544,
      "step": 9142
    },
    {
      "epoch": 0.5032197699378061,
      "grad_norm": 0.6476640105247498,
      "learning_rate": 8.54058765561923e-06,
      "loss": 0.7221,
      "step": 9143
    },
    {
      "epoch": 0.5032748087401618,
      "grad_norm": 0.7098944187164307,
      "learning_rate": 8.540281573883755e-06,
      "loss": 0.8083,
      "step": 9144
    },
    {
      "epoch": 0.5033298475425175,
      "grad_norm": 0.9733545184135437,
      "learning_rate": 8.539975465540473e-06,
      "loss": 0.7381,
      "step": 9145
    },
    {
      "epoch": 0.5033848863448731,
      "grad_norm": 0.641211986541748,
      "learning_rate": 8.539669330591685e-06,
      "loss": 0.7511,
      "step": 9146
    },
    {
      "epoch": 0.5034399251472288,
      "grad_norm": 0.626027524471283,
      "learning_rate": 8.539363169039687e-06,
      "loss": 0.7321,
      "step": 9147
    },
    {
      "epoch": 0.5034949639495845,
      "grad_norm": 0.7627241611480713,
      "learning_rate": 8.539056980886785e-06,
      "loss": 0.7269,
      "step": 9148
    },
    {
      "epoch": 0.5035500027519401,
      "grad_norm": 0.6711145639419556,
      "learning_rate": 8.538750766135275e-06,
      "loss": 0.8179,
      "step": 9149
    },
    {
      "epoch": 0.5036050415542958,
      "grad_norm": 0.6981950998306274,
      "learning_rate": 8.538444524787463e-06,
      "loss": 0.8095,
      "step": 9150
    },
    {
      "epoch": 0.5036600803566514,
      "grad_norm": 0.8869871497154236,
      "learning_rate": 8.53813825684565e-06,
      "loss": 0.8549,
      "step": 9151
    },
    {
      "epoch": 0.5037151191590071,
      "grad_norm": 0.6461544036865234,
      "learning_rate": 8.537831962312137e-06,
      "loss": 0.7388,
      "step": 9152
    },
    {
      "epoch": 0.5037701579613627,
      "grad_norm": 0.8279222249984741,
      "learning_rate": 8.537525641189224e-06,
      "loss": 0.8609,
      "step": 9153
    },
    {
      "epoch": 0.5038251967637184,
      "grad_norm": 0.7117578387260437,
      "learning_rate": 8.537219293479217e-06,
      "loss": 0.802,
      "step": 9154
    },
    {
      "epoch": 0.5038802355660741,
      "grad_norm": 0.6831860542297363,
      "learning_rate": 8.536912919184416e-06,
      "loss": 0.7821,
      "step": 9155
    },
    {
      "epoch": 0.5039352743684298,
      "grad_norm": 1.1528539657592773,
      "learning_rate": 8.536606518307125e-06,
      "loss": 0.8578,
      "step": 9156
    },
    {
      "epoch": 0.5039903131707854,
      "grad_norm": 0.6545060873031616,
      "learning_rate": 8.536300090849645e-06,
      "loss": 0.7744,
      "step": 9157
    },
    {
      "epoch": 0.504045351973141,
      "grad_norm": 0.7176601886749268,
      "learning_rate": 8.535993636814281e-06,
      "loss": 0.8104,
      "step": 9158
    },
    {
      "epoch": 0.5041003907754967,
      "grad_norm": 0.8458410501480103,
      "learning_rate": 8.535687156203334e-06,
      "loss": 0.8653,
      "step": 9159
    },
    {
      "epoch": 0.5041554295778524,
      "grad_norm": 0.7500274777412415,
      "learning_rate": 8.53538064901911e-06,
      "loss": 0.8043,
      "step": 9160
    },
    {
      "epoch": 0.504210468380208,
      "grad_norm": 0.6982965469360352,
      "learning_rate": 8.535074115263911e-06,
      "loss": 0.7564,
      "step": 9161
    },
    {
      "epoch": 0.5042655071825637,
      "grad_norm": 0.8344218134880066,
      "learning_rate": 8.534767554940042e-06,
      "loss": 0.7575,
      "step": 9162
    },
    {
      "epoch": 0.5043205459849194,
      "grad_norm": 0.7527137398719788,
      "learning_rate": 8.534460968049806e-06,
      "loss": 0.7757,
      "step": 9163
    },
    {
      "epoch": 0.5043755847872751,
      "grad_norm": 0.7136969566345215,
      "learning_rate": 8.534154354595508e-06,
      "loss": 0.826,
      "step": 9164
    },
    {
      "epoch": 0.5044306235896306,
      "grad_norm": 0.8102819919586182,
      "learning_rate": 8.533847714579449e-06,
      "loss": 0.7247,
      "step": 9165
    },
    {
      "epoch": 0.5044856623919863,
      "grad_norm": 0.7568309903144836,
      "learning_rate": 8.53354104800394e-06,
      "loss": 0.8509,
      "step": 9166
    },
    {
      "epoch": 0.504540701194342,
      "grad_norm": 0.7719592452049255,
      "learning_rate": 8.53323435487128e-06,
      "loss": 0.8039,
      "step": 9167
    },
    {
      "epoch": 0.5045957399966977,
      "grad_norm": 0.7514411807060242,
      "learning_rate": 8.532927635183778e-06,
      "loss": 0.8759,
      "step": 9168
    },
    {
      "epoch": 0.5046507787990533,
      "grad_norm": 0.9781903028488159,
      "learning_rate": 8.532620888943736e-06,
      "loss": 0.8022,
      "step": 9169
    },
    {
      "epoch": 0.504705817601409,
      "grad_norm": 0.7713304758071899,
      "learning_rate": 8.532314116153462e-06,
      "loss": 0.8372,
      "step": 9170
    },
    {
      "epoch": 0.5047608564037647,
      "grad_norm": 0.7519709467887878,
      "learning_rate": 8.53200731681526e-06,
      "loss": 0.7374,
      "step": 9171
    },
    {
      "epoch": 0.5048158952061204,
      "grad_norm": 0.6923980712890625,
      "learning_rate": 8.531700490931438e-06,
      "loss": 0.7511,
      "step": 9172
    },
    {
      "epoch": 0.5048709340084759,
      "grad_norm": 0.682357907295227,
      "learning_rate": 8.5313936385043e-06,
      "loss": 0.7647,
      "step": 9173
    },
    {
      "epoch": 0.5049259728108316,
      "grad_norm": 0.8255659341812134,
      "learning_rate": 8.531086759536152e-06,
      "loss": 0.7533,
      "step": 9174
    },
    {
      "epoch": 0.5049810116131873,
      "grad_norm": 0.6774975061416626,
      "learning_rate": 8.530779854029301e-06,
      "loss": 0.7019,
      "step": 9175
    },
    {
      "epoch": 0.505036050415543,
      "grad_norm": 0.7973241209983826,
      "learning_rate": 8.530472921986053e-06,
      "loss": 0.7824,
      "step": 9176
    },
    {
      "epoch": 0.5050910892178986,
      "grad_norm": 0.8216109275817871,
      "learning_rate": 8.530165963408716e-06,
      "loss": 0.8063,
      "step": 9177
    },
    {
      "epoch": 0.5051461280202543,
      "grad_norm": 0.7277935743331909,
      "learning_rate": 8.5298589782996e-06,
      "loss": 0.7631,
      "step": 9178
    },
    {
      "epoch": 0.50520116682261,
      "grad_norm": 0.6647855043411255,
      "learning_rate": 8.529551966661004e-06,
      "loss": 0.7462,
      "step": 9179
    },
    {
      "epoch": 0.5052562056249656,
      "grad_norm": 0.766272783279419,
      "learning_rate": 8.529244928495241e-06,
      "loss": 0.8075,
      "step": 9180
    },
    {
      "epoch": 0.5053112444273212,
      "grad_norm": 0.7276293635368347,
      "learning_rate": 8.52893786380462e-06,
      "loss": 0.7908,
      "step": 9181
    },
    {
      "epoch": 0.5053662832296769,
      "grad_norm": 0.7864169478416443,
      "learning_rate": 8.528630772591447e-06,
      "loss": 0.8082,
      "step": 9182
    },
    {
      "epoch": 0.5054213220320326,
      "grad_norm": 0.9106804132461548,
      "learning_rate": 8.528323654858028e-06,
      "loss": 0.8989,
      "step": 9183
    },
    {
      "epoch": 0.5054763608343883,
      "grad_norm": 0.7288523316383362,
      "learning_rate": 8.52801651060667e-06,
      "loss": 0.7972,
      "step": 9184
    },
    {
      "epoch": 0.5055313996367439,
      "grad_norm": 0.7149643301963806,
      "learning_rate": 8.527709339839689e-06,
      "loss": 0.8191,
      "step": 9185
    },
    {
      "epoch": 0.5055864384390996,
      "grad_norm": 0.6661714911460876,
      "learning_rate": 8.527402142559388e-06,
      "loss": 0.6596,
      "step": 9186
    },
    {
      "epoch": 0.5056414772414553,
      "grad_norm": 0.7071447372436523,
      "learning_rate": 8.527094918768076e-06,
      "loss": 0.7633,
      "step": 9187
    },
    {
      "epoch": 0.5056965160438109,
      "grad_norm": 0.7314093112945557,
      "learning_rate": 8.526787668468064e-06,
      "loss": 0.7815,
      "step": 9188
    },
    {
      "epoch": 0.5057515548461665,
      "grad_norm": 0.8200539946556091,
      "learning_rate": 8.526480391661657e-06,
      "loss": 0.8376,
      "step": 9189
    },
    {
      "epoch": 0.5058065936485222,
      "grad_norm": 0.7422435283660889,
      "learning_rate": 8.52617308835117e-06,
      "loss": 0.8783,
      "step": 9190
    },
    {
      "epoch": 0.5058616324508779,
      "grad_norm": 0.7845084071159363,
      "learning_rate": 8.525865758538909e-06,
      "loss": 0.8005,
      "step": 9191
    },
    {
      "epoch": 0.5059166712532335,
      "grad_norm": 0.6854296922683716,
      "learning_rate": 8.525558402227185e-06,
      "loss": 0.8118,
      "step": 9192
    },
    {
      "epoch": 0.5059717100555892,
      "grad_norm": 0.6805297136306763,
      "learning_rate": 8.525251019418309e-06,
      "loss": 0.6765,
      "step": 9193
    },
    {
      "epoch": 0.5060267488579449,
      "grad_norm": 0.7194867134094238,
      "learning_rate": 8.524943610114587e-06,
      "loss": 0.6752,
      "step": 9194
    },
    {
      "epoch": 0.5060817876603005,
      "grad_norm": 0.6935137510299683,
      "learning_rate": 8.524636174318335e-06,
      "loss": 0.7122,
      "step": 9195
    },
    {
      "epoch": 0.5061368264626561,
      "grad_norm": 0.8652825951576233,
      "learning_rate": 8.52432871203186e-06,
      "loss": 0.7725,
      "step": 9196
    },
    {
      "epoch": 0.5061918652650118,
      "grad_norm": 0.9104461669921875,
      "learning_rate": 8.524021223257472e-06,
      "loss": 0.8589,
      "step": 9197
    },
    {
      "epoch": 0.5062469040673675,
      "grad_norm": 0.7680580019950867,
      "learning_rate": 8.523713707997486e-06,
      "loss": 0.842,
      "step": 9198
    },
    {
      "epoch": 0.5063019428697232,
      "grad_norm": 0.7324872612953186,
      "learning_rate": 8.52340616625421e-06,
      "loss": 0.802,
      "step": 9199
    },
    {
      "epoch": 0.5063569816720788,
      "grad_norm": 0.8812359571456909,
      "learning_rate": 8.523098598029958e-06,
      "loss": 0.8286,
      "step": 9200
    },
    {
      "epoch": 0.5064120204744345,
      "grad_norm": 0.6992496848106384,
      "learning_rate": 8.522791003327038e-06,
      "loss": 0.811,
      "step": 9201
    },
    {
      "epoch": 0.5064670592767901,
      "grad_norm": 0.8191942572593689,
      "learning_rate": 8.522483382147766e-06,
      "loss": 0.7192,
      "step": 9202
    },
    {
      "epoch": 0.5065220980791458,
      "grad_norm": 0.9354501366615295,
      "learning_rate": 8.522175734494452e-06,
      "loss": 0.7424,
      "step": 9203
    },
    {
      "epoch": 0.5065771368815014,
      "grad_norm": 0.6481999754905701,
      "learning_rate": 8.521868060369405e-06,
      "loss": 0.6385,
      "step": 9204
    },
    {
      "epoch": 0.5066321756838571,
      "grad_norm": 0.7158499360084534,
      "learning_rate": 8.521560359774943e-06,
      "loss": 0.6116,
      "step": 9205
    },
    {
      "epoch": 0.5066872144862128,
      "grad_norm": 0.8738408088684082,
      "learning_rate": 8.521252632713376e-06,
      "loss": 0.894,
      "step": 9206
    },
    {
      "epoch": 0.5067422532885685,
      "grad_norm": 0.7037062644958496,
      "learning_rate": 8.520944879187015e-06,
      "loss": 0.6958,
      "step": 9207
    },
    {
      "epoch": 0.5067972920909241,
      "grad_norm": 0.7205594778060913,
      "learning_rate": 8.520637099198175e-06,
      "loss": 0.7188,
      "step": 9208
    },
    {
      "epoch": 0.5068523308932797,
      "grad_norm": 0.6761966347694397,
      "learning_rate": 8.520329292749169e-06,
      "loss": 0.7669,
      "step": 9209
    },
    {
      "epoch": 0.5069073696956354,
      "grad_norm": 0.682556688785553,
      "learning_rate": 8.520021459842312e-06,
      "loss": 0.7745,
      "step": 9210
    },
    {
      "epoch": 0.5069624084979911,
      "grad_norm": 0.6687794923782349,
      "learning_rate": 8.519713600479913e-06,
      "loss": 0.7814,
      "step": 9211
    },
    {
      "epoch": 0.5070174473003467,
      "grad_norm": 0.6391967535018921,
      "learning_rate": 8.51940571466429e-06,
      "loss": 0.7331,
      "step": 9212
    },
    {
      "epoch": 0.5070724861027024,
      "grad_norm": 0.8420151472091675,
      "learning_rate": 8.519097802397758e-06,
      "loss": 0.8257,
      "step": 9213
    },
    {
      "epoch": 0.5071275249050581,
      "grad_norm": 0.692787230014801,
      "learning_rate": 8.518789863682625e-06,
      "loss": 0.7179,
      "step": 9214
    },
    {
      "epoch": 0.5071825637074138,
      "grad_norm": 0.6874318718910217,
      "learning_rate": 8.518481898521213e-06,
      "loss": 0.6847,
      "step": 9215
    },
    {
      "epoch": 0.5072376025097693,
      "grad_norm": 0.8107750415802002,
      "learning_rate": 8.518173906915832e-06,
      "loss": 0.8459,
      "step": 9216
    },
    {
      "epoch": 0.507292641312125,
      "grad_norm": 0.7952812910079956,
      "learning_rate": 8.517865888868797e-06,
      "loss": 0.8503,
      "step": 9217
    },
    {
      "epoch": 0.5073476801144807,
      "grad_norm": 0.6926921606063843,
      "learning_rate": 8.517557844382424e-06,
      "loss": 0.6713,
      "step": 9218
    },
    {
      "epoch": 0.5074027189168364,
      "grad_norm": 0.8203585147857666,
      "learning_rate": 8.517249773459026e-06,
      "loss": 0.8483,
      "step": 9219
    },
    {
      "epoch": 0.507457757719192,
      "grad_norm": 0.6788125038146973,
      "learning_rate": 8.516941676100923e-06,
      "loss": 0.7521,
      "step": 9220
    },
    {
      "epoch": 0.5075127965215477,
      "grad_norm": 0.6439838409423828,
      "learning_rate": 8.516633552310426e-06,
      "loss": 0.7359,
      "step": 9221
    },
    {
      "epoch": 0.5075678353239034,
      "grad_norm": 0.6872217655181885,
      "learning_rate": 8.516325402089854e-06,
      "loss": 0.73,
      "step": 9222
    },
    {
      "epoch": 0.5076228741262591,
      "grad_norm": 0.6695985794067383,
      "learning_rate": 8.51601722544152e-06,
      "loss": 0.7519,
      "step": 9223
    },
    {
      "epoch": 0.5076779129286146,
      "grad_norm": 0.7779402136802673,
      "learning_rate": 8.515709022367741e-06,
      "loss": 0.7325,
      "step": 9224
    },
    {
      "epoch": 0.5077329517309703,
      "grad_norm": 0.9289746284484863,
      "learning_rate": 8.515400792870836e-06,
      "loss": 0.7839,
      "step": 9225
    },
    {
      "epoch": 0.507787990533326,
      "grad_norm": 0.6949248313903809,
      "learning_rate": 8.51509253695312e-06,
      "loss": 0.7363,
      "step": 9226
    },
    {
      "epoch": 0.5078430293356817,
      "grad_norm": 0.6463130116462708,
      "learning_rate": 8.514784254616908e-06,
      "loss": 0.7607,
      "step": 9227
    },
    {
      "epoch": 0.5078980681380373,
      "grad_norm": 0.7332046031951904,
      "learning_rate": 8.514475945864519e-06,
      "loss": 0.6833,
      "step": 9228
    },
    {
      "epoch": 0.507953106940393,
      "grad_norm": 0.8674100637435913,
      "learning_rate": 8.51416761069827e-06,
      "loss": 0.669,
      "step": 9229
    },
    {
      "epoch": 0.5080081457427487,
      "grad_norm": 0.8073185682296753,
      "learning_rate": 8.513859249120477e-06,
      "loss": 0.7215,
      "step": 9230
    },
    {
      "epoch": 0.5080631845451044,
      "grad_norm": 0.674117386341095,
      "learning_rate": 8.51355086113346e-06,
      "loss": 0.7813,
      "step": 9231
    },
    {
      "epoch": 0.5081182233474599,
      "grad_norm": 0.8564596176147461,
      "learning_rate": 8.513242446739534e-06,
      "loss": 0.7393,
      "step": 9232
    },
    {
      "epoch": 0.5081732621498156,
      "grad_norm": 0.684637188911438,
      "learning_rate": 8.512934005941015e-06,
      "loss": 0.781,
      "step": 9233
    },
    {
      "epoch": 0.5082283009521713,
      "grad_norm": 0.816123902797699,
      "learning_rate": 8.51262553874023e-06,
      "loss": 0.8597,
      "step": 9234
    },
    {
      "epoch": 0.5082833397545269,
      "grad_norm": 0.6582320332527161,
      "learning_rate": 8.512317045139488e-06,
      "loss": 0.6654,
      "step": 9235
    },
    {
      "epoch": 0.5083383785568826,
      "grad_norm": 1.0153518915176392,
      "learning_rate": 8.512008525141113e-06,
      "loss": 0.7946,
      "step": 9236
    },
    {
      "epoch": 0.5083934173592383,
      "grad_norm": 0.7455416917800903,
      "learning_rate": 8.511699978747422e-06,
      "loss": 0.8365,
      "step": 9237
    },
    {
      "epoch": 0.508448456161594,
      "grad_norm": 0.6498221755027771,
      "learning_rate": 8.511391405960733e-06,
      "loss": 0.7252,
      "step": 9238
    },
    {
      "epoch": 0.5085034949639495,
      "grad_norm": 0.6856792569160461,
      "learning_rate": 8.511082806783368e-06,
      "loss": 0.7282,
      "step": 9239
    },
    {
      "epoch": 0.5085585337663052,
      "grad_norm": 0.6930065751075745,
      "learning_rate": 8.510774181217643e-06,
      "loss": 0.7404,
      "step": 9240
    },
    {
      "epoch": 0.5086135725686609,
      "grad_norm": 0.6953150033950806,
      "learning_rate": 8.51046552926588e-06,
      "loss": 0.7684,
      "step": 9241
    },
    {
      "epoch": 0.5086686113710166,
      "grad_norm": 0.7307711839675903,
      "learning_rate": 8.510156850930395e-06,
      "loss": 0.7557,
      "step": 9242
    },
    {
      "epoch": 0.5087236501733722,
      "grad_norm": 0.7296478152275085,
      "learning_rate": 8.509848146213513e-06,
      "loss": 0.7469,
      "step": 9243
    },
    {
      "epoch": 0.5087786889757279,
      "grad_norm": 0.7035672664642334,
      "learning_rate": 8.509539415117553e-06,
      "loss": 0.7151,
      "step": 9244
    },
    {
      "epoch": 0.5088337277780836,
      "grad_norm": 0.7818698883056641,
      "learning_rate": 8.509230657644832e-06,
      "loss": 0.7134,
      "step": 9245
    },
    {
      "epoch": 0.5088887665804392,
      "grad_norm": 0.7503119111061096,
      "learning_rate": 8.508921873797674e-06,
      "loss": 0.7028,
      "step": 9246
    },
    {
      "epoch": 0.5089438053827948,
      "grad_norm": 0.7733498215675354,
      "learning_rate": 8.508613063578397e-06,
      "loss": 0.8159,
      "step": 9247
    },
    {
      "epoch": 0.5089988441851505,
      "grad_norm": 0.9236353635787964,
      "learning_rate": 8.508304226989326e-06,
      "loss": 0.8013,
      "step": 9248
    },
    {
      "epoch": 0.5090538829875062,
      "grad_norm": 0.6567198634147644,
      "learning_rate": 8.507995364032777e-06,
      "loss": 0.8285,
      "step": 9249
    },
    {
      "epoch": 0.5091089217898619,
      "grad_norm": 0.6555445790290833,
      "learning_rate": 8.507686474711074e-06,
      "loss": 0.6917,
      "step": 9250
    },
    {
      "epoch": 0.5091639605922175,
      "grad_norm": 0.8505375385284424,
      "learning_rate": 8.507377559026539e-06,
      "loss": 0.824,
      "step": 9251
    },
    {
      "epoch": 0.5092189993945732,
      "grad_norm": 0.703413188457489,
      "learning_rate": 8.507068616981493e-06,
      "loss": 0.7162,
      "step": 9252
    },
    {
      "epoch": 0.5092740381969288,
      "grad_norm": 0.7257823944091797,
      "learning_rate": 8.50675964857826e-06,
      "loss": 0.8031,
      "step": 9253
    },
    {
      "epoch": 0.5093290769992845,
      "grad_norm": 0.6861198544502258,
      "learning_rate": 8.506450653819159e-06,
      "loss": 0.7724,
      "step": 9254
    },
    {
      "epoch": 0.5093841158016401,
      "grad_norm": 0.7733107209205627,
      "learning_rate": 8.506141632706512e-06,
      "loss": 0.7834,
      "step": 9255
    },
    {
      "epoch": 0.5094391546039958,
      "grad_norm": 0.7472217082977295,
      "learning_rate": 8.505832585242644e-06,
      "loss": 0.7594,
      "step": 9256
    },
    {
      "epoch": 0.5094941934063515,
      "grad_norm": 0.6273325085639954,
      "learning_rate": 8.505523511429876e-06,
      "loss": 0.6798,
      "step": 9257
    },
    {
      "epoch": 0.5095492322087072,
      "grad_norm": 0.7366517186164856,
      "learning_rate": 8.505214411270533e-06,
      "loss": 0.7916,
      "step": 9258
    },
    {
      "epoch": 0.5096042710110628,
      "grad_norm": 0.6654453873634338,
      "learning_rate": 8.504905284766936e-06,
      "loss": 0.7228,
      "step": 9259
    },
    {
      "epoch": 0.5096593098134184,
      "grad_norm": 0.7926275134086609,
      "learning_rate": 8.50459613192141e-06,
      "loss": 0.8303,
      "step": 9260
    },
    {
      "epoch": 0.5097143486157741,
      "grad_norm": 0.7256377935409546,
      "learning_rate": 8.504286952736277e-06,
      "loss": 0.7977,
      "step": 9261
    },
    {
      "epoch": 0.5097693874181298,
      "grad_norm": 0.7333946824073792,
      "learning_rate": 8.50397774721386e-06,
      "loss": 0.7978,
      "step": 9262
    },
    {
      "epoch": 0.5098244262204854,
      "grad_norm": 0.6102882623672485,
      "learning_rate": 8.503668515356485e-06,
      "loss": 0.6386,
      "step": 9263
    },
    {
      "epoch": 0.5098794650228411,
      "grad_norm": 0.7939823865890503,
      "learning_rate": 8.503359257166477e-06,
      "loss": 0.7328,
      "step": 9264
    },
    {
      "epoch": 0.5099345038251968,
      "grad_norm": 0.7245013117790222,
      "learning_rate": 8.503049972646157e-06,
      "loss": 0.795,
      "step": 9265
    },
    {
      "epoch": 0.5099895426275525,
      "grad_norm": 0.6722108125686646,
      "learning_rate": 8.502740661797852e-06,
      "loss": 0.7062,
      "step": 9266
    },
    {
      "epoch": 0.510044581429908,
      "grad_norm": 0.6759012341499329,
      "learning_rate": 8.502431324623884e-06,
      "loss": 0.7427,
      "step": 9267
    },
    {
      "epoch": 0.5100996202322637,
      "grad_norm": 0.6448835730552673,
      "learning_rate": 8.502121961126581e-06,
      "loss": 0.7381,
      "step": 9268
    },
    {
      "epoch": 0.5101546590346194,
      "grad_norm": 0.6437426209449768,
      "learning_rate": 8.501812571308266e-06,
      "loss": 0.6733,
      "step": 9269
    },
    {
      "epoch": 0.5102096978369751,
      "grad_norm": 0.6879013776779175,
      "learning_rate": 8.501503155171267e-06,
      "loss": 0.7227,
      "step": 9270
    },
    {
      "epoch": 0.5102647366393307,
      "grad_norm": 0.6628512740135193,
      "learning_rate": 8.501193712717906e-06,
      "loss": 0.7151,
      "step": 9271
    },
    {
      "epoch": 0.5103197754416864,
      "grad_norm": 0.7653747797012329,
      "learning_rate": 8.500884243950511e-06,
      "loss": 0.8189,
      "step": 9272
    },
    {
      "epoch": 0.5103748142440421,
      "grad_norm": 0.7180060148239136,
      "learning_rate": 8.500574748871407e-06,
      "loss": 0.7633,
      "step": 9273
    },
    {
      "epoch": 0.5104298530463978,
      "grad_norm": 0.7045086622238159,
      "learning_rate": 8.50026522748292e-06,
      "loss": 0.746,
      "step": 9274
    },
    {
      "epoch": 0.5104848918487533,
      "grad_norm": 0.6224614381790161,
      "learning_rate": 8.499955679787376e-06,
      "loss": 0.7436,
      "step": 9275
    },
    {
      "epoch": 0.510539930651109,
      "grad_norm": 0.6716495156288147,
      "learning_rate": 8.499646105787103e-06,
      "loss": 0.8006,
      "step": 9276
    },
    {
      "epoch": 0.5105949694534647,
      "grad_norm": 0.83705735206604,
      "learning_rate": 8.499336505484426e-06,
      "loss": 0.886,
      "step": 9277
    },
    {
      "epoch": 0.5106500082558203,
      "grad_norm": 0.7942199110984802,
      "learning_rate": 8.499026878881673e-06,
      "loss": 0.7709,
      "step": 9278
    },
    {
      "epoch": 0.510705047058176,
      "grad_norm": 0.7500330209732056,
      "learning_rate": 8.49871722598117e-06,
      "loss": 0.7737,
      "step": 9279
    },
    {
      "epoch": 0.5107600858605317,
      "grad_norm": 0.7283433675765991,
      "learning_rate": 8.498407546785245e-06,
      "loss": 0.8345,
      "step": 9280
    },
    {
      "epoch": 0.5108151246628874,
      "grad_norm": 0.6970989108085632,
      "learning_rate": 8.498097841296224e-06,
      "loss": 0.7451,
      "step": 9281
    },
    {
      "epoch": 0.5108701634652429,
      "grad_norm": 0.8338573575019836,
      "learning_rate": 8.497788109516438e-06,
      "loss": 0.8198,
      "step": 9282
    },
    {
      "epoch": 0.5109252022675986,
      "grad_norm": 0.6544861197471619,
      "learning_rate": 8.497478351448213e-06,
      "loss": 0.7549,
      "step": 9283
    },
    {
      "epoch": 0.5109802410699543,
      "grad_norm": 0.6627360582351685,
      "learning_rate": 8.497168567093876e-06,
      "loss": 0.7136,
      "step": 9284
    },
    {
      "epoch": 0.51103527987231,
      "grad_norm": 0.7176669239997864,
      "learning_rate": 8.496858756455755e-06,
      "loss": 0.766,
      "step": 9285
    },
    {
      "epoch": 0.5110903186746656,
      "grad_norm": 0.8260897397994995,
      "learning_rate": 8.496548919536183e-06,
      "loss": 0.8167,
      "step": 9286
    },
    {
      "epoch": 0.5111453574770213,
      "grad_norm": 0.7077773809432983,
      "learning_rate": 8.496239056337483e-06,
      "loss": 0.776,
      "step": 9287
    },
    {
      "epoch": 0.511200396279377,
      "grad_norm": 0.7609447836875916,
      "learning_rate": 8.495929166861988e-06,
      "loss": 0.7339,
      "step": 9288
    },
    {
      "epoch": 0.5112554350817327,
      "grad_norm": 0.6896487474441528,
      "learning_rate": 8.495619251112022e-06,
      "loss": 0.7639,
      "step": 9289
    },
    {
      "epoch": 0.5113104738840882,
      "grad_norm": 0.6946871280670166,
      "learning_rate": 8.495309309089918e-06,
      "loss": 0.8242,
      "step": 9290
    },
    {
      "epoch": 0.5113655126864439,
      "grad_norm": 0.79847252368927,
      "learning_rate": 8.494999340798007e-06,
      "loss": 0.8226,
      "step": 9291
    },
    {
      "epoch": 0.5114205514887996,
      "grad_norm": 0.7845447063446045,
      "learning_rate": 8.494689346238615e-06,
      "loss": 0.8593,
      "step": 9292
    },
    {
      "epoch": 0.5114755902911553,
      "grad_norm": 1.1577119827270508,
      "learning_rate": 8.494379325414074e-06,
      "loss": 0.746,
      "step": 9293
    },
    {
      "epoch": 0.5115306290935109,
      "grad_norm": 0.6720938682556152,
      "learning_rate": 8.494069278326713e-06,
      "loss": 0.6768,
      "step": 9294
    },
    {
      "epoch": 0.5115856678958666,
      "grad_norm": 0.7389395833015442,
      "learning_rate": 8.493759204978862e-06,
      "loss": 0.8126,
      "step": 9295
    },
    {
      "epoch": 0.5116407066982223,
      "grad_norm": 0.7629536986351013,
      "learning_rate": 8.493449105372853e-06,
      "loss": 0.7107,
      "step": 9296
    },
    {
      "epoch": 0.511695745500578,
      "grad_norm": 0.7339474558830261,
      "learning_rate": 8.493138979511015e-06,
      "loss": 0.8144,
      "step": 9297
    },
    {
      "epoch": 0.5117507843029335,
      "grad_norm": 0.7222825288772583,
      "learning_rate": 8.49282882739568e-06,
      "loss": 0.7512,
      "step": 9298
    },
    {
      "epoch": 0.5118058231052892,
      "grad_norm": 0.676659107208252,
      "learning_rate": 8.49251864902918e-06,
      "loss": 0.6515,
      "step": 9299
    },
    {
      "epoch": 0.5118608619076449,
      "grad_norm": 0.6336323618888855,
      "learning_rate": 8.492208444413844e-06,
      "loss": 0.719,
      "step": 9300
    },
    {
      "epoch": 0.5119159007100006,
      "grad_norm": 0.701543927192688,
      "learning_rate": 8.491898213552e-06,
      "loss": 0.728,
      "step": 9301
    },
    {
      "epoch": 0.5119709395123562,
      "grad_norm": 0.6809069514274597,
      "learning_rate": 8.491587956445988e-06,
      "loss": 0.8844,
      "step": 9302
    },
    {
      "epoch": 0.5120259783147119,
      "grad_norm": 0.8046489357948303,
      "learning_rate": 8.491277673098135e-06,
      "loss": 0.817,
      "step": 9303
    },
    {
      "epoch": 0.5120810171170675,
      "grad_norm": 0.8630616068840027,
      "learning_rate": 8.490967363510774e-06,
      "loss": 0.7745,
      "step": 9304
    },
    {
      "epoch": 0.5121360559194232,
      "grad_norm": 0.7457678914070129,
      "learning_rate": 8.490657027686235e-06,
      "loss": 0.7956,
      "step": 9305
    },
    {
      "epoch": 0.5121910947217788,
      "grad_norm": 0.6383466124534607,
      "learning_rate": 8.490346665626854e-06,
      "loss": 0.8046,
      "step": 9306
    },
    {
      "epoch": 0.5122461335241345,
      "grad_norm": 0.7658202052116394,
      "learning_rate": 8.49003627733496e-06,
      "loss": 0.7905,
      "step": 9307
    },
    {
      "epoch": 0.5123011723264902,
      "grad_norm": 0.6793283224105835,
      "learning_rate": 8.48972586281289e-06,
      "loss": 0.6646,
      "step": 9308
    },
    {
      "epoch": 0.5123562111288459,
      "grad_norm": 0.7345246076583862,
      "learning_rate": 8.489415422062972e-06,
      "loss": 0.788,
      "step": 9309
    },
    {
      "epoch": 0.5124112499312015,
      "grad_norm": 0.6665463447570801,
      "learning_rate": 8.489104955087542e-06,
      "loss": 0.706,
      "step": 9310
    },
    {
      "epoch": 0.5124662887335572,
      "grad_norm": 0.7895458936691284,
      "learning_rate": 8.488794461888934e-06,
      "loss": 0.7464,
      "step": 9311
    },
    {
      "epoch": 0.5125213275359128,
      "grad_norm": 0.7375221252441406,
      "learning_rate": 8.488483942469481e-06,
      "loss": 0.8029,
      "step": 9312
    },
    {
      "epoch": 0.5125763663382685,
      "grad_norm": 0.792348325252533,
      "learning_rate": 8.488173396831514e-06,
      "loss": 0.7324,
      "step": 9313
    },
    {
      "epoch": 0.5126314051406241,
      "grad_norm": 0.6500192880630493,
      "learning_rate": 8.487862824977373e-06,
      "loss": 0.7331,
      "step": 9314
    },
    {
      "epoch": 0.5126864439429798,
      "grad_norm": 0.6607314348220825,
      "learning_rate": 8.487552226909386e-06,
      "loss": 0.7782,
      "step": 9315
    },
    {
      "epoch": 0.5127414827453355,
      "grad_norm": 0.8261791467666626,
      "learning_rate": 8.487241602629892e-06,
      "loss": 0.8036,
      "step": 9316
    },
    {
      "epoch": 0.5127965215476912,
      "grad_norm": 0.8301663994789124,
      "learning_rate": 8.486930952141222e-06,
      "loss": 0.7928,
      "step": 9317
    },
    {
      "epoch": 0.5128515603500468,
      "grad_norm": 0.6957940459251404,
      "learning_rate": 8.486620275445713e-06,
      "loss": 0.7359,
      "step": 9318
    },
    {
      "epoch": 0.5129065991524024,
      "grad_norm": 0.7562606334686279,
      "learning_rate": 8.4863095725457e-06,
      "loss": 0.7546,
      "step": 9319
    },
    {
      "epoch": 0.5129616379547581,
      "grad_norm": 0.795886218547821,
      "learning_rate": 8.485998843443517e-06,
      "loss": 0.7558,
      "step": 9320
    },
    {
      "epoch": 0.5130166767571137,
      "grad_norm": 0.6558147072792053,
      "learning_rate": 8.4856880881415e-06,
      "loss": 0.6832,
      "step": 9321
    },
    {
      "epoch": 0.5130717155594694,
      "grad_norm": 0.7300151586532593,
      "learning_rate": 8.485377306641984e-06,
      "loss": 0.8018,
      "step": 9322
    },
    {
      "epoch": 0.5131267543618251,
      "grad_norm": 0.7114105224609375,
      "learning_rate": 8.485066498947305e-06,
      "loss": 0.7374,
      "step": 9323
    },
    {
      "epoch": 0.5131817931641808,
      "grad_norm": 0.7061085104942322,
      "learning_rate": 8.484755665059798e-06,
      "loss": 0.7905,
      "step": 9324
    },
    {
      "epoch": 0.5132368319665364,
      "grad_norm": 0.8481647968292236,
      "learning_rate": 8.484444804981802e-06,
      "loss": 0.8518,
      "step": 9325
    },
    {
      "epoch": 0.513291870768892,
      "grad_norm": 0.7583557367324829,
      "learning_rate": 8.48413391871565e-06,
      "loss": 0.8328,
      "step": 9326
    },
    {
      "epoch": 0.5133469095712477,
      "grad_norm": 0.7381925582885742,
      "learning_rate": 8.483823006263683e-06,
      "loss": 0.76,
      "step": 9327
    },
    {
      "epoch": 0.5134019483736034,
      "grad_norm": 0.8037852644920349,
      "learning_rate": 8.483512067628232e-06,
      "loss": 0.711,
      "step": 9328
    },
    {
      "epoch": 0.513456987175959,
      "grad_norm": 0.6682618260383606,
      "learning_rate": 8.483201102811637e-06,
      "loss": 0.7479,
      "step": 9329
    },
    {
      "epoch": 0.5135120259783147,
      "grad_norm": 0.662234365940094,
      "learning_rate": 8.482890111816237e-06,
      "loss": 0.7701,
      "step": 9330
    },
    {
      "epoch": 0.5135670647806704,
      "grad_norm": 0.7081482410430908,
      "learning_rate": 8.482579094644365e-06,
      "loss": 0.8255,
      "step": 9331
    },
    {
      "epoch": 0.5136221035830261,
      "grad_norm": 0.9659954905509949,
      "learning_rate": 8.482268051298364e-06,
      "loss": 0.8742,
      "step": 9332
    },
    {
      "epoch": 0.5136771423853816,
      "grad_norm": 0.7837772369384766,
      "learning_rate": 8.481956981780564e-06,
      "loss": 0.7692,
      "step": 9333
    },
    {
      "epoch": 0.5137321811877373,
      "grad_norm": 0.681918203830719,
      "learning_rate": 8.481645886093311e-06,
      "loss": 0.6952,
      "step": 9334
    },
    {
      "epoch": 0.513787219990093,
      "grad_norm": 0.7253187894821167,
      "learning_rate": 8.481334764238937e-06,
      "loss": 0.7074,
      "step": 9335
    },
    {
      "epoch": 0.5138422587924487,
      "grad_norm": 0.8845877051353455,
      "learning_rate": 8.481023616219783e-06,
      "loss": 0.675,
      "step": 9336
    },
    {
      "epoch": 0.5138972975948043,
      "grad_norm": 0.6569344401359558,
      "learning_rate": 8.480712442038188e-06,
      "loss": 0.7181,
      "step": 9337
    },
    {
      "epoch": 0.51395233639716,
      "grad_norm": 0.7372813820838928,
      "learning_rate": 8.480401241696491e-06,
      "loss": 0.8137,
      "step": 9338
    },
    {
      "epoch": 0.5140073751995157,
      "grad_norm": 0.843099057674408,
      "learning_rate": 8.48009001519703e-06,
      "loss": 0.7648,
      "step": 9339
    },
    {
      "epoch": 0.5140624140018714,
      "grad_norm": 0.7762032747268677,
      "learning_rate": 8.479778762542142e-06,
      "loss": 0.7805,
      "step": 9340
    },
    {
      "epoch": 0.5141174528042269,
      "grad_norm": 0.739086925983429,
      "learning_rate": 8.479467483734169e-06,
      "loss": 0.7125,
      "step": 9341
    },
    {
      "epoch": 0.5141724916065826,
      "grad_norm": 0.7351683974266052,
      "learning_rate": 8.479156178775451e-06,
      "loss": 0.7855,
      "step": 9342
    },
    {
      "epoch": 0.5142275304089383,
      "grad_norm": 0.7601314187049866,
      "learning_rate": 8.478844847668325e-06,
      "loss": 0.8349,
      "step": 9343
    },
    {
      "epoch": 0.514282569211294,
      "grad_norm": 0.6841638684272766,
      "learning_rate": 8.478533490415133e-06,
      "loss": 0.7986,
      "step": 9344
    },
    {
      "epoch": 0.5143376080136496,
      "grad_norm": 0.6734872460365295,
      "learning_rate": 8.478222107018213e-06,
      "loss": 0.6941,
      "step": 9345
    },
    {
      "epoch": 0.5143926468160053,
      "grad_norm": 0.801930844783783,
      "learning_rate": 8.47791069747991e-06,
      "loss": 0.8537,
      "step": 9346
    },
    {
      "epoch": 0.514447685618361,
      "grad_norm": 0.6960629224777222,
      "learning_rate": 8.477599261802558e-06,
      "loss": 0.6629,
      "step": 9347
    },
    {
      "epoch": 0.5145027244207167,
      "grad_norm": 0.7791358232498169,
      "learning_rate": 8.477287799988502e-06,
      "loss": 0.8777,
      "step": 9348
    },
    {
      "epoch": 0.5145577632230722,
      "grad_norm": 0.7022722959518433,
      "learning_rate": 8.476976312040082e-06,
      "loss": 0.7116,
      "step": 9349
    },
    {
      "epoch": 0.5146128020254279,
      "grad_norm": 0.7791306376457214,
      "learning_rate": 8.476664797959639e-06,
      "loss": 0.7262,
      "step": 9350
    },
    {
      "epoch": 0.5146678408277836,
      "grad_norm": 0.7391177415847778,
      "learning_rate": 8.476353257749514e-06,
      "loss": 0.7308,
      "step": 9351
    },
    {
      "epoch": 0.5147228796301393,
      "grad_norm": 0.6989552974700928,
      "learning_rate": 8.476041691412046e-06,
      "loss": 0.7754,
      "step": 9352
    },
    {
      "epoch": 0.5147779184324949,
      "grad_norm": 0.7639930844306946,
      "learning_rate": 8.475730098949582e-06,
      "loss": 0.8385,
      "step": 9353
    },
    {
      "epoch": 0.5148329572348506,
      "grad_norm": 0.7687931060791016,
      "learning_rate": 8.47541848036446e-06,
      "loss": 0.8118,
      "step": 9354
    },
    {
      "epoch": 0.5148879960372063,
      "grad_norm": 0.8831589221954346,
      "learning_rate": 8.475106835659024e-06,
      "loss": 0.7705,
      "step": 9355
    },
    {
      "epoch": 0.5149430348395619,
      "grad_norm": 0.7585502862930298,
      "learning_rate": 8.474795164835614e-06,
      "loss": 0.8167,
      "step": 9356
    },
    {
      "epoch": 0.5149980736419175,
      "grad_norm": 0.7078690528869629,
      "learning_rate": 8.474483467896572e-06,
      "loss": 0.7412,
      "step": 9357
    },
    {
      "epoch": 0.5150531124442732,
      "grad_norm": 0.8950889706611633,
      "learning_rate": 8.474171744844246e-06,
      "loss": 0.8132,
      "step": 9358
    },
    {
      "epoch": 0.5151081512466289,
      "grad_norm": 0.7196077704429626,
      "learning_rate": 8.473859995680973e-06,
      "loss": 0.8041,
      "step": 9359
    },
    {
      "epoch": 0.5151631900489846,
      "grad_norm": 0.7705141305923462,
      "learning_rate": 8.473548220409099e-06,
      "loss": 0.8437,
      "step": 9360
    },
    {
      "epoch": 0.5152182288513402,
      "grad_norm": 0.6507467031478882,
      "learning_rate": 8.473236419030966e-06,
      "loss": 0.7713,
      "step": 9361
    },
    {
      "epoch": 0.5152732676536959,
      "grad_norm": 0.7120817303657532,
      "learning_rate": 8.472924591548917e-06,
      "loss": 0.7688,
      "step": 9362
    },
    {
      "epoch": 0.5153283064560515,
      "grad_norm": 0.7830487489700317,
      "learning_rate": 8.472612737965297e-06,
      "loss": 0.8875,
      "step": 9363
    },
    {
      "epoch": 0.5153833452584071,
      "grad_norm": 0.8790529370307922,
      "learning_rate": 8.47230085828245e-06,
      "loss": 0.7648,
      "step": 9364
    },
    {
      "epoch": 0.5154383840607628,
      "grad_norm": 0.8956806659698486,
      "learning_rate": 8.471988952502718e-06,
      "loss": 0.7891,
      "step": 9365
    },
    {
      "epoch": 0.5154934228631185,
      "grad_norm": 0.7370011210441589,
      "learning_rate": 8.471677020628448e-06,
      "loss": 0.7609,
      "step": 9366
    },
    {
      "epoch": 0.5155484616654742,
      "grad_norm": 0.6794238090515137,
      "learning_rate": 8.471365062661982e-06,
      "loss": 0.6679,
      "step": 9367
    },
    {
      "epoch": 0.5156035004678298,
      "grad_norm": 0.7330273985862732,
      "learning_rate": 8.471053078605664e-06,
      "loss": 0.7276,
      "step": 9368
    },
    {
      "epoch": 0.5156585392701855,
      "grad_norm": 0.7796601057052612,
      "learning_rate": 8.470741068461843e-06,
      "loss": 0.7897,
      "step": 9369
    },
    {
      "epoch": 0.5157135780725411,
      "grad_norm": 0.6834099888801575,
      "learning_rate": 8.470429032232858e-06,
      "loss": 0.7924,
      "step": 9370
    },
    {
      "epoch": 0.5157686168748968,
      "grad_norm": 0.6991616487503052,
      "learning_rate": 8.47011696992106e-06,
      "loss": 0.7901,
      "step": 9371
    },
    {
      "epoch": 0.5158236556772524,
      "grad_norm": 0.7321401834487915,
      "learning_rate": 8.469804881528792e-06,
      "loss": 0.6718,
      "step": 9372
    },
    {
      "epoch": 0.5158786944796081,
      "grad_norm": 0.7091043591499329,
      "learning_rate": 8.469492767058398e-06,
      "loss": 0.8204,
      "step": 9373
    },
    {
      "epoch": 0.5159337332819638,
      "grad_norm": 0.8777012825012207,
      "learning_rate": 8.469180626512223e-06,
      "loss": 0.8045,
      "step": 9374
    },
    {
      "epoch": 0.5159887720843195,
      "grad_norm": 0.6652738451957703,
      "learning_rate": 8.468868459892619e-06,
      "loss": 0.7248,
      "step": 9375
    },
    {
      "epoch": 0.5160438108866751,
      "grad_norm": 0.7209659218788147,
      "learning_rate": 8.468556267201925e-06,
      "loss": 0.7508,
      "step": 9376
    },
    {
      "epoch": 0.5160988496890307,
      "grad_norm": 0.7685441970825195,
      "learning_rate": 8.468244048442494e-06,
      "loss": 0.7501,
      "step": 9377
    },
    {
      "epoch": 0.5161538884913864,
      "grad_norm": 0.6773725152015686,
      "learning_rate": 8.467931803616665e-06,
      "loss": 0.8036,
      "step": 9378
    },
    {
      "epoch": 0.5162089272937421,
      "grad_norm": 0.7167890071868896,
      "learning_rate": 8.467619532726792e-06,
      "loss": 0.7229,
      "step": 9379
    },
    {
      "epoch": 0.5162639660960977,
      "grad_norm": 0.7066929340362549,
      "learning_rate": 8.467307235775218e-06,
      "loss": 0.7433,
      "step": 9380
    },
    {
      "epoch": 0.5163190048984534,
      "grad_norm": 0.7261828780174255,
      "learning_rate": 8.46699491276429e-06,
      "loss": 0.7873,
      "step": 9381
    },
    {
      "epoch": 0.5163740437008091,
      "grad_norm": 0.7442463636398315,
      "learning_rate": 8.466682563696356e-06,
      "loss": 0.7953,
      "step": 9382
    },
    {
      "epoch": 0.5164290825031648,
      "grad_norm": 0.5668768286705017,
      "learning_rate": 8.466370188573765e-06,
      "loss": 0.5602,
      "step": 9383
    },
    {
      "epoch": 0.5164841213055203,
      "grad_norm": 0.7364997267723083,
      "learning_rate": 8.466057787398864e-06,
      "loss": 0.8274,
      "step": 9384
    },
    {
      "epoch": 0.516539160107876,
      "grad_norm": 0.7793132066726685,
      "learning_rate": 8.465745360174e-06,
      "loss": 0.7832,
      "step": 9385
    },
    {
      "epoch": 0.5165941989102317,
      "grad_norm": 0.6818128824234009,
      "learning_rate": 8.46543290690152e-06,
      "loss": 0.8314,
      "step": 9386
    },
    {
      "epoch": 0.5166492377125874,
      "grad_norm": 0.7392195463180542,
      "learning_rate": 8.465120427583778e-06,
      "loss": 0.8124,
      "step": 9387
    },
    {
      "epoch": 0.516704276514943,
      "grad_norm": 0.8582521677017212,
      "learning_rate": 8.464807922223115e-06,
      "loss": 0.7417,
      "step": 9388
    },
    {
      "epoch": 0.5167593153172987,
      "grad_norm": 0.7322097420692444,
      "learning_rate": 8.464495390821882e-06,
      "loss": 0.7408,
      "step": 9389
    },
    {
      "epoch": 0.5168143541196544,
      "grad_norm": 0.8177433013916016,
      "learning_rate": 8.464182833382432e-06,
      "loss": 0.87,
      "step": 9390
    },
    {
      "epoch": 0.5168693929220101,
      "grad_norm": 0.7088115215301514,
      "learning_rate": 8.46387024990711e-06,
      "loss": 0.7748,
      "step": 9391
    },
    {
      "epoch": 0.5169244317243656,
      "grad_norm": 0.6648650169372559,
      "learning_rate": 8.463557640398268e-06,
      "loss": 0.6302,
      "step": 9392
    },
    {
      "epoch": 0.5169794705267213,
      "grad_norm": 0.6688859462738037,
      "learning_rate": 8.463245004858251e-06,
      "loss": 0.7252,
      "step": 9393
    },
    {
      "epoch": 0.517034509329077,
      "grad_norm": 0.7231030464172363,
      "learning_rate": 8.462932343289412e-06,
      "loss": 0.8497,
      "step": 9394
    },
    {
      "epoch": 0.5170895481314327,
      "grad_norm": 0.7142065763473511,
      "learning_rate": 8.462619655694103e-06,
      "loss": 0.7041,
      "step": 9395
    },
    {
      "epoch": 0.5171445869337883,
      "grad_norm": 0.7197136878967285,
      "learning_rate": 8.462306942074669e-06,
      "loss": 0.7022,
      "step": 9396
    },
    {
      "epoch": 0.517199625736144,
      "grad_norm": 0.7620192766189575,
      "learning_rate": 8.461994202433463e-06,
      "loss": 0.8243,
      "step": 9397
    },
    {
      "epoch": 0.5172546645384997,
      "grad_norm": 0.7697533965110779,
      "learning_rate": 8.461681436772836e-06,
      "loss": 0.7861,
      "step": 9398
    },
    {
      "epoch": 0.5173097033408554,
      "grad_norm": 0.7224711179733276,
      "learning_rate": 8.461368645095138e-06,
      "loss": 0.7588,
      "step": 9399
    },
    {
      "epoch": 0.5173647421432109,
      "grad_norm": 0.9285979270935059,
      "learning_rate": 8.46105582740272e-06,
      "loss": 0.8113,
      "step": 9400
    },
    {
      "epoch": 0.5174197809455666,
      "grad_norm": 0.7297842502593994,
      "learning_rate": 8.460742983697934e-06,
      "loss": 0.7115,
      "step": 9401
    },
    {
      "epoch": 0.5174748197479223,
      "grad_norm": 0.6712872982025146,
      "learning_rate": 8.460430113983126e-06,
      "loss": 0.751,
      "step": 9402
    },
    {
      "epoch": 0.517529858550278,
      "grad_norm": 0.7807186245918274,
      "learning_rate": 8.460117218260657e-06,
      "loss": 0.8375,
      "step": 9403
    },
    {
      "epoch": 0.5175848973526336,
      "grad_norm": 0.621530294418335,
      "learning_rate": 8.45980429653287e-06,
      "loss": 0.638,
      "step": 9404
    },
    {
      "epoch": 0.5176399361549893,
      "grad_norm": 0.7086256146430969,
      "learning_rate": 8.45949134880212e-06,
      "loss": 0.8304,
      "step": 9405
    },
    {
      "epoch": 0.517694974957345,
      "grad_norm": 0.62705397605896,
      "learning_rate": 8.45917837507076e-06,
      "loss": 0.7008,
      "step": 9406
    },
    {
      "epoch": 0.5177500137597005,
      "grad_norm": 0.9109121561050415,
      "learning_rate": 8.458865375341142e-06,
      "loss": 0.7529,
      "step": 9407
    },
    {
      "epoch": 0.5178050525620562,
      "grad_norm": 0.6909900903701782,
      "learning_rate": 8.458552349615615e-06,
      "loss": 0.8453,
      "step": 9408
    },
    {
      "epoch": 0.5178600913644119,
      "grad_norm": 0.7548434138298035,
      "learning_rate": 8.458239297896536e-06,
      "loss": 0.7516,
      "step": 9409
    },
    {
      "epoch": 0.5179151301667676,
      "grad_norm": 0.7595730423927307,
      "learning_rate": 8.457926220186257e-06,
      "loss": 0.7599,
      "step": 9410
    },
    {
      "epoch": 0.5179701689691232,
      "grad_norm": 0.7449337840080261,
      "learning_rate": 8.45761311648713e-06,
      "loss": 0.8236,
      "step": 9411
    },
    {
      "epoch": 0.5180252077714789,
      "grad_norm": 0.7529160976409912,
      "learning_rate": 8.457299986801507e-06,
      "loss": 0.8655,
      "step": 9412
    },
    {
      "epoch": 0.5180802465738346,
      "grad_norm": 0.6777701377868652,
      "learning_rate": 8.456986831131742e-06,
      "loss": 0.7737,
      "step": 9413
    },
    {
      "epoch": 0.5181352853761902,
      "grad_norm": 0.9363510012626648,
      "learning_rate": 8.456673649480191e-06,
      "loss": 0.8227,
      "step": 9414
    },
    {
      "epoch": 0.5181903241785458,
      "grad_norm": 0.798001229763031,
      "learning_rate": 8.456360441849206e-06,
      "loss": 0.8881,
      "step": 9415
    },
    {
      "epoch": 0.5182453629809015,
      "grad_norm": 0.7212072610855103,
      "learning_rate": 8.456047208241141e-06,
      "loss": 0.8165,
      "step": 9416
    },
    {
      "epoch": 0.5183004017832572,
      "grad_norm": 0.6918027997016907,
      "learning_rate": 8.45573394865835e-06,
      "loss": 0.8048,
      "step": 9417
    },
    {
      "epoch": 0.5183554405856129,
      "grad_norm": 0.6474916338920593,
      "learning_rate": 8.455420663103187e-06,
      "loss": 0.6502,
      "step": 9418
    },
    {
      "epoch": 0.5184104793879685,
      "grad_norm": 0.6592364311218262,
      "learning_rate": 8.455107351578008e-06,
      "loss": 0.7509,
      "step": 9419
    },
    {
      "epoch": 0.5184655181903242,
      "grad_norm": 0.7658745646476746,
      "learning_rate": 8.454794014085168e-06,
      "loss": 0.8444,
      "step": 9420
    },
    {
      "epoch": 0.5185205569926798,
      "grad_norm": 0.6814215183258057,
      "learning_rate": 8.45448065062702e-06,
      "loss": 0.7367,
      "step": 9421
    },
    {
      "epoch": 0.5185755957950355,
      "grad_norm": 0.644740104675293,
      "learning_rate": 8.45416726120592e-06,
      "loss": 0.7456,
      "step": 9422
    },
    {
      "epoch": 0.5186306345973911,
      "grad_norm": 0.8578751087188721,
      "learning_rate": 8.453853845824225e-06,
      "loss": 0.8481,
      "step": 9423
    },
    {
      "epoch": 0.5186856733997468,
      "grad_norm": 0.6630389094352722,
      "learning_rate": 8.453540404484288e-06,
      "loss": 0.7487,
      "step": 9424
    },
    {
      "epoch": 0.5187407122021025,
      "grad_norm": 0.7756431698799133,
      "learning_rate": 8.453226937188466e-06,
      "loss": 0.798,
      "step": 9425
    },
    {
      "epoch": 0.5187957510044582,
      "grad_norm": 0.7856318354606628,
      "learning_rate": 8.452913443939113e-06,
      "loss": 0.785,
      "step": 9426
    },
    {
      "epoch": 0.5188507898068138,
      "grad_norm": 0.7563977837562561,
      "learning_rate": 8.45259992473859e-06,
      "loss": 0.8182,
      "step": 9427
    },
    {
      "epoch": 0.5189058286091695,
      "grad_norm": 0.6945043802261353,
      "learning_rate": 8.452286379589247e-06,
      "loss": 0.7262,
      "step": 9428
    },
    {
      "epoch": 0.5189608674115251,
      "grad_norm": 0.6607717275619507,
      "learning_rate": 8.451972808493444e-06,
      "loss": 0.7257,
      "step": 9429
    },
    {
      "epoch": 0.5190159062138808,
      "grad_norm": 0.6682843565940857,
      "learning_rate": 8.451659211453539e-06,
      "loss": 0.6775,
      "step": 9430
    },
    {
      "epoch": 0.5190709450162364,
      "grad_norm": 0.7175559401512146,
      "learning_rate": 8.451345588471886e-06,
      "loss": 0.7154,
      "step": 9431
    },
    {
      "epoch": 0.5191259838185921,
      "grad_norm": 0.7499119639396667,
      "learning_rate": 8.451031939550845e-06,
      "loss": 0.7537,
      "step": 9432
    },
    {
      "epoch": 0.5191810226209478,
      "grad_norm": 0.65048748254776,
      "learning_rate": 8.450718264692771e-06,
      "loss": 0.7253,
      "step": 9433
    },
    {
      "epoch": 0.5192360614233035,
      "grad_norm": 0.7067640423774719,
      "learning_rate": 8.450404563900022e-06,
      "loss": 0.7245,
      "step": 9434
    },
    {
      "epoch": 0.519291100225659,
      "grad_norm": 0.7079932689666748,
      "learning_rate": 8.450090837174956e-06,
      "loss": 0.7776,
      "step": 9435
    },
    {
      "epoch": 0.5193461390280147,
      "grad_norm": 0.8260107636451721,
      "learning_rate": 8.44977708451993e-06,
      "loss": 0.8529,
      "step": 9436
    },
    {
      "epoch": 0.5194011778303704,
      "grad_norm": 0.6412167549133301,
      "learning_rate": 8.449463305937304e-06,
      "loss": 0.7371,
      "step": 9437
    },
    {
      "epoch": 0.5194562166327261,
      "grad_norm": 0.7067576050758362,
      "learning_rate": 8.449149501429435e-06,
      "loss": 0.7161,
      "step": 9438
    },
    {
      "epoch": 0.5195112554350817,
      "grad_norm": 0.6966904997825623,
      "learning_rate": 8.448835670998681e-06,
      "loss": 0.7285,
      "step": 9439
    },
    {
      "epoch": 0.5195662942374374,
      "grad_norm": 0.8066132664680481,
      "learning_rate": 8.448521814647401e-06,
      "loss": 0.8265,
      "step": 9440
    },
    {
      "epoch": 0.5196213330397931,
      "grad_norm": 0.7597149610519409,
      "learning_rate": 8.448207932377957e-06,
      "loss": 0.7721,
      "step": 9441
    },
    {
      "epoch": 0.5196763718421488,
      "grad_norm": 0.6965302228927612,
      "learning_rate": 8.447894024192702e-06,
      "loss": 0.749,
      "step": 9442
    },
    {
      "epoch": 0.5197314106445043,
      "grad_norm": 0.7032600045204163,
      "learning_rate": 8.447580090094e-06,
      "loss": 0.7923,
      "step": 9443
    },
    {
      "epoch": 0.51978644944686,
      "grad_norm": 0.7255309820175171,
      "learning_rate": 8.447266130084208e-06,
      "loss": 0.6739,
      "step": 9444
    },
    {
      "epoch": 0.5198414882492157,
      "grad_norm": 0.6602993011474609,
      "learning_rate": 8.446952144165686e-06,
      "loss": 0.7886,
      "step": 9445
    },
    {
      "epoch": 0.5198965270515714,
      "grad_norm": 0.7017884850502014,
      "learning_rate": 8.446638132340796e-06,
      "loss": 0.7554,
      "step": 9446
    },
    {
      "epoch": 0.519951565853927,
      "grad_norm": 0.7234843969345093,
      "learning_rate": 8.446324094611894e-06,
      "loss": 0.8294,
      "step": 9447
    },
    {
      "epoch": 0.5200066046562827,
      "grad_norm": 0.6859332919120789,
      "learning_rate": 8.446010030981347e-06,
      "loss": 0.7563,
      "step": 9448
    },
    {
      "epoch": 0.5200616434586384,
      "grad_norm": 0.7759458422660828,
      "learning_rate": 8.445695941451507e-06,
      "loss": 0.7577,
      "step": 9449
    },
    {
      "epoch": 0.520116682260994,
      "grad_norm": 0.7852263450622559,
      "learning_rate": 8.44538182602474e-06,
      "loss": 0.7446,
      "step": 9450
    },
    {
      "epoch": 0.5201717210633496,
      "grad_norm": 0.8143053650856018,
      "learning_rate": 8.445067684703406e-06,
      "loss": 0.7995,
      "step": 9451
    },
    {
      "epoch": 0.5202267598657053,
      "grad_norm": 0.692738950252533,
      "learning_rate": 8.444753517489865e-06,
      "loss": 0.7185,
      "step": 9452
    },
    {
      "epoch": 0.520281798668061,
      "grad_norm": 0.6615390181541443,
      "learning_rate": 8.444439324386478e-06,
      "loss": 0.7128,
      "step": 9453
    },
    {
      "epoch": 0.5203368374704166,
      "grad_norm": 0.7360419034957886,
      "learning_rate": 8.444125105395608e-06,
      "loss": 0.6565,
      "step": 9454
    },
    {
      "epoch": 0.5203918762727723,
      "grad_norm": 0.7280182838439941,
      "learning_rate": 8.443810860519615e-06,
      "loss": 0.7295,
      "step": 9455
    },
    {
      "epoch": 0.520446915075128,
      "grad_norm": 0.787367582321167,
      "learning_rate": 8.44349658976086e-06,
      "loss": 0.7342,
      "step": 9456
    },
    {
      "epoch": 0.5205019538774837,
      "grad_norm": 0.7496024966239929,
      "learning_rate": 8.44318229312171e-06,
      "loss": 0.7499,
      "step": 9457
    },
    {
      "epoch": 0.5205569926798392,
      "grad_norm": 0.9167383909225464,
      "learning_rate": 8.44286797060452e-06,
      "loss": 0.7797,
      "step": 9458
    },
    {
      "epoch": 0.5206120314821949,
      "grad_norm": 0.7032341957092285,
      "learning_rate": 8.442553622211659e-06,
      "loss": 0.7627,
      "step": 9459
    },
    {
      "epoch": 0.5206670702845506,
      "grad_norm": 1.2905993461608887,
      "learning_rate": 8.442239247945485e-06,
      "loss": 0.7841,
      "step": 9460
    },
    {
      "epoch": 0.5207221090869063,
      "grad_norm": 0.6909230351448059,
      "learning_rate": 8.441924847808362e-06,
      "loss": 0.7234,
      "step": 9461
    },
    {
      "epoch": 0.5207771478892619,
      "grad_norm": 0.6632175445556641,
      "learning_rate": 8.441610421802653e-06,
      "loss": 0.6733,
      "step": 9462
    },
    {
      "epoch": 0.5208321866916176,
      "grad_norm": 0.7838154435157776,
      "learning_rate": 8.441295969930722e-06,
      "loss": 0.7583,
      "step": 9463
    },
    {
      "epoch": 0.5208872254939733,
      "grad_norm": 0.6380481123924255,
      "learning_rate": 8.440981492194932e-06,
      "loss": 0.7109,
      "step": 9464
    },
    {
      "epoch": 0.520942264296329,
      "grad_norm": 0.6859052181243896,
      "learning_rate": 8.440666988597646e-06,
      "loss": 0.7387,
      "step": 9465
    },
    {
      "epoch": 0.5209973030986845,
      "grad_norm": 0.7411379814147949,
      "learning_rate": 8.440352459141226e-06,
      "loss": 0.7852,
      "step": 9466
    },
    {
      "epoch": 0.5210523419010402,
      "grad_norm": 0.6925216913223267,
      "learning_rate": 8.44003790382804e-06,
      "loss": 0.8228,
      "step": 9467
    },
    {
      "epoch": 0.5211073807033959,
      "grad_norm": 0.7136396169662476,
      "learning_rate": 8.43972332266045e-06,
      "loss": 0.8168,
      "step": 9468
    },
    {
      "epoch": 0.5211624195057516,
      "grad_norm": 0.719639003276825,
      "learning_rate": 8.43940871564082e-06,
      "loss": 0.6728,
      "step": 9469
    },
    {
      "epoch": 0.5212174583081072,
      "grad_norm": 0.647861897945404,
      "learning_rate": 8.439094082771513e-06,
      "loss": 0.6986,
      "step": 9470
    },
    {
      "epoch": 0.5212724971104629,
      "grad_norm": 0.6644579172134399,
      "learning_rate": 8.438779424054897e-06,
      "loss": 0.6263,
      "step": 9471
    },
    {
      "epoch": 0.5213275359128186,
      "grad_norm": 0.7157352566719055,
      "learning_rate": 8.438464739493335e-06,
      "loss": 0.827,
      "step": 9472
    },
    {
      "epoch": 0.5213825747151742,
      "grad_norm": 0.793765127658844,
      "learning_rate": 8.438150029089193e-06,
      "loss": 0.741,
      "step": 9473
    },
    {
      "epoch": 0.5214376135175298,
      "grad_norm": 0.7078518867492676,
      "learning_rate": 8.437835292844836e-06,
      "loss": 0.7618,
      "step": 9474
    },
    {
      "epoch": 0.5214926523198855,
      "grad_norm": 0.7492140531539917,
      "learning_rate": 8.437520530762628e-06,
      "loss": 0.7894,
      "step": 9475
    },
    {
      "epoch": 0.5215476911222412,
      "grad_norm": 0.6534473299980164,
      "learning_rate": 8.437205742844937e-06,
      "loss": 0.7567,
      "step": 9476
    },
    {
      "epoch": 0.5216027299245969,
      "grad_norm": 0.8745388984680176,
      "learning_rate": 8.436890929094126e-06,
      "loss": 0.8758,
      "step": 9477
    },
    {
      "epoch": 0.5216577687269525,
      "grad_norm": 0.6804752349853516,
      "learning_rate": 8.436576089512564e-06,
      "loss": 0.7841,
      "step": 9478
    },
    {
      "epoch": 0.5217128075293082,
      "grad_norm": 0.712065577507019,
      "learning_rate": 8.436261224102615e-06,
      "loss": 0.8079,
      "step": 9479
    },
    {
      "epoch": 0.5217678463316638,
      "grad_norm": 0.8733783960342407,
      "learning_rate": 8.435946332866648e-06,
      "loss": 0.8295,
      "step": 9480
    },
    {
      "epoch": 0.5218228851340195,
      "grad_norm": 0.6871289610862732,
      "learning_rate": 8.435631415807028e-06,
      "loss": 0.7087,
      "step": 9481
    },
    {
      "epoch": 0.5218779239363751,
      "grad_norm": 0.8363185524940491,
      "learning_rate": 8.43531647292612e-06,
      "loss": 0.7329,
      "step": 9482
    },
    {
      "epoch": 0.5219329627387308,
      "grad_norm": 0.6845195293426514,
      "learning_rate": 8.435001504226295e-06,
      "loss": 0.7651,
      "step": 9483
    },
    {
      "epoch": 0.5219880015410865,
      "grad_norm": 0.7527645826339722,
      "learning_rate": 8.434686509709917e-06,
      "loss": 0.6856,
      "step": 9484
    },
    {
      "epoch": 0.5220430403434422,
      "grad_norm": 0.6945710778236389,
      "learning_rate": 8.434371489379356e-06,
      "loss": 0.6875,
      "step": 9485
    },
    {
      "epoch": 0.5220980791457978,
      "grad_norm": 0.7668873071670532,
      "learning_rate": 8.434056443236977e-06,
      "loss": 0.7662,
      "step": 9486
    },
    {
      "epoch": 0.5221531179481534,
      "grad_norm": 0.9873473048210144,
      "learning_rate": 8.433741371285148e-06,
      "loss": 0.7662,
      "step": 9487
    },
    {
      "epoch": 0.5222081567505091,
      "grad_norm": 0.8635447025299072,
      "learning_rate": 8.43342627352624e-06,
      "loss": 0.645,
      "step": 9488
    },
    {
      "epoch": 0.5222631955528648,
      "grad_norm": 0.7836978435516357,
      "learning_rate": 8.43311114996262e-06,
      "loss": 0.7647,
      "step": 9489
    },
    {
      "epoch": 0.5223182343552204,
      "grad_norm": 0.8370835185050964,
      "learning_rate": 8.432796000596652e-06,
      "loss": 0.8402,
      "step": 9490
    },
    {
      "epoch": 0.5223732731575761,
      "grad_norm": 0.9627843499183655,
      "learning_rate": 8.432480825430712e-06,
      "loss": 0.6985,
      "step": 9491
    },
    {
      "epoch": 0.5224283119599318,
      "grad_norm": 0.6774263978004456,
      "learning_rate": 8.432165624467163e-06,
      "loss": 0.7051,
      "step": 9492
    },
    {
      "epoch": 0.5224833507622874,
      "grad_norm": 0.6590597033500671,
      "learning_rate": 8.431850397708375e-06,
      "loss": 0.7147,
      "step": 9493
    },
    {
      "epoch": 0.522538389564643,
      "grad_norm": 0.8153522610664368,
      "learning_rate": 8.43153514515672e-06,
      "loss": 0.6759,
      "step": 9494
    },
    {
      "epoch": 0.5225934283669987,
      "grad_norm": 0.7457708716392517,
      "learning_rate": 8.431219866814563e-06,
      "loss": 0.7168,
      "step": 9495
    },
    {
      "epoch": 0.5226484671693544,
      "grad_norm": 0.6994161009788513,
      "learning_rate": 8.430904562684278e-06,
      "loss": 0.8393,
      "step": 9496
    },
    {
      "epoch": 0.52270350597171,
      "grad_norm": 0.780337393283844,
      "learning_rate": 8.430589232768232e-06,
      "loss": 0.6528,
      "step": 9497
    },
    {
      "epoch": 0.5227585447740657,
      "grad_norm": 0.6833232641220093,
      "learning_rate": 8.430273877068796e-06,
      "loss": 0.7545,
      "step": 9498
    },
    {
      "epoch": 0.5228135835764214,
      "grad_norm": 0.7330057621002197,
      "learning_rate": 8.42995849558834e-06,
      "loss": 0.7932,
      "step": 9499
    },
    {
      "epoch": 0.5228686223787771,
      "grad_norm": 0.8131541609764099,
      "learning_rate": 8.429643088329233e-06,
      "loss": 0.7546,
      "step": 9500
    },
    {
      "epoch": 0.5229236611811326,
      "grad_norm": 0.7353833317756653,
      "learning_rate": 8.42932765529385e-06,
      "loss": 0.7508,
      "step": 9501
    },
    {
      "epoch": 0.5229786999834883,
      "grad_norm": 0.7166246771812439,
      "learning_rate": 8.429012196484554e-06,
      "loss": 0.728,
      "step": 9502
    },
    {
      "epoch": 0.523033738785844,
      "grad_norm": 0.732064962387085,
      "learning_rate": 8.428696711903721e-06,
      "loss": 0.8306,
      "step": 9503
    },
    {
      "epoch": 0.5230887775881997,
      "grad_norm": 0.6858934164047241,
      "learning_rate": 8.428381201553721e-06,
      "loss": 0.7801,
      "step": 9504
    },
    {
      "epoch": 0.5231438163905553,
      "grad_norm": 0.7046478986740112,
      "learning_rate": 8.428065665436928e-06,
      "loss": 0.7365,
      "step": 9505
    },
    {
      "epoch": 0.523198855192911,
      "grad_norm": 0.6669325828552246,
      "learning_rate": 8.42775010355571e-06,
      "loss": 0.7764,
      "step": 9506
    },
    {
      "epoch": 0.5232538939952667,
      "grad_norm": 0.655619740486145,
      "learning_rate": 8.427434515912438e-06,
      "loss": 0.7919,
      "step": 9507
    },
    {
      "epoch": 0.5233089327976224,
      "grad_norm": 0.6236690878868103,
      "learning_rate": 8.427118902509487e-06,
      "loss": 0.6653,
      "step": 9508
    },
    {
      "epoch": 0.5233639715999779,
      "grad_norm": 0.8233165740966797,
      "learning_rate": 8.426803263349228e-06,
      "loss": 0.8012,
      "step": 9509
    },
    {
      "epoch": 0.5234190104023336,
      "grad_norm": 0.6626759171485901,
      "learning_rate": 8.426487598434035e-06,
      "loss": 0.7728,
      "step": 9510
    },
    {
      "epoch": 0.5234740492046893,
      "grad_norm": 0.9209974408149719,
      "learning_rate": 8.426171907766275e-06,
      "loss": 0.769,
      "step": 9511
    },
    {
      "epoch": 0.523529088007045,
      "grad_norm": 0.6297587156295776,
      "learning_rate": 8.425856191348325e-06,
      "loss": 0.7333,
      "step": 9512
    },
    {
      "epoch": 0.5235841268094006,
      "grad_norm": 0.6995256543159485,
      "learning_rate": 8.425540449182558e-06,
      "loss": 0.7486,
      "step": 9513
    },
    {
      "epoch": 0.5236391656117563,
      "grad_norm": 0.8076607584953308,
      "learning_rate": 8.425224681271345e-06,
      "loss": 0.8533,
      "step": 9514
    },
    {
      "epoch": 0.523694204414112,
      "grad_norm": 1.2198601961135864,
      "learning_rate": 8.42490888761706e-06,
      "loss": 0.7291,
      "step": 9515
    },
    {
      "epoch": 0.5237492432164677,
      "grad_norm": 0.7047159671783447,
      "learning_rate": 8.424593068222076e-06,
      "loss": 0.713,
      "step": 9516
    },
    {
      "epoch": 0.5238042820188232,
      "grad_norm": 0.7652333378791809,
      "learning_rate": 8.424277223088768e-06,
      "loss": 0.8149,
      "step": 9517
    },
    {
      "epoch": 0.5238593208211789,
      "grad_norm": 1.1311010122299194,
      "learning_rate": 8.42396135221951e-06,
      "loss": 0.8195,
      "step": 9518
    },
    {
      "epoch": 0.5239143596235346,
      "grad_norm": 0.7855533957481384,
      "learning_rate": 8.423645455616674e-06,
      "loss": 0.7901,
      "step": 9519
    },
    {
      "epoch": 0.5239693984258903,
      "grad_norm": 0.7028971314430237,
      "learning_rate": 8.423329533282635e-06,
      "loss": 0.8006,
      "step": 9520
    },
    {
      "epoch": 0.5240244372282459,
      "grad_norm": 0.703809916973114,
      "learning_rate": 8.423013585219769e-06,
      "loss": 0.7581,
      "step": 9521
    },
    {
      "epoch": 0.5240794760306016,
      "grad_norm": 0.94233238697052,
      "learning_rate": 8.422697611430448e-06,
      "loss": 0.7689,
      "step": 9522
    },
    {
      "epoch": 0.5241345148329573,
      "grad_norm": 0.8164071440696716,
      "learning_rate": 8.422381611917047e-06,
      "loss": 0.8761,
      "step": 9523
    },
    {
      "epoch": 0.5241895536353129,
      "grad_norm": 0.6242091059684753,
      "learning_rate": 8.422065586681944e-06,
      "loss": 0.6975,
      "step": 9524
    },
    {
      "epoch": 0.5242445924376685,
      "grad_norm": 0.6607261300086975,
      "learning_rate": 8.42174953572751e-06,
      "loss": 0.6847,
      "step": 9525
    },
    {
      "epoch": 0.5242996312400242,
      "grad_norm": 0.7174261212348938,
      "learning_rate": 8.421433459056123e-06,
      "loss": 0.7905,
      "step": 9526
    },
    {
      "epoch": 0.5243546700423799,
      "grad_norm": 0.7414089441299438,
      "learning_rate": 8.42111735667016e-06,
      "loss": 0.7788,
      "step": 9527
    },
    {
      "epoch": 0.5244097088447356,
      "grad_norm": 0.7347442507743835,
      "learning_rate": 8.420801228571992e-06,
      "loss": 0.7691,
      "step": 9528
    },
    {
      "epoch": 0.5244647476470912,
      "grad_norm": 0.6947832107543945,
      "learning_rate": 8.420485074763999e-06,
      "loss": 0.6702,
      "step": 9529
    },
    {
      "epoch": 0.5245197864494469,
      "grad_norm": 0.6865423321723938,
      "learning_rate": 8.420168895248557e-06,
      "loss": 0.7577,
      "step": 9530
    },
    {
      "epoch": 0.5245748252518025,
      "grad_norm": 0.7023190855979919,
      "learning_rate": 8.419852690028039e-06,
      "loss": 0.7711,
      "step": 9531
    },
    {
      "epoch": 0.5246298640541582,
      "grad_norm": 0.8312145471572876,
      "learning_rate": 8.419536459104824e-06,
      "loss": 0.7999,
      "step": 9532
    },
    {
      "epoch": 0.5246849028565138,
      "grad_norm": 0.6700688600540161,
      "learning_rate": 8.419220202481288e-06,
      "loss": 0.7163,
      "step": 9533
    },
    {
      "epoch": 0.5247399416588695,
      "grad_norm": 0.767062246799469,
      "learning_rate": 8.418903920159809e-06,
      "loss": 0.7451,
      "step": 9534
    },
    {
      "epoch": 0.5247949804612252,
      "grad_norm": 0.6814010143280029,
      "learning_rate": 8.418587612142763e-06,
      "loss": 0.771,
      "step": 9535
    },
    {
      "epoch": 0.5248500192635808,
      "grad_norm": 0.6728426218032837,
      "learning_rate": 8.418271278432528e-06,
      "loss": 0.8336,
      "step": 9536
    },
    {
      "epoch": 0.5249050580659365,
      "grad_norm": 0.7112382650375366,
      "learning_rate": 8.417954919031482e-06,
      "loss": 0.7392,
      "step": 9537
    },
    {
      "epoch": 0.5249600968682921,
      "grad_norm": 0.7371365427970886,
      "learning_rate": 8.417638533942e-06,
      "loss": 0.8233,
      "step": 9538
    },
    {
      "epoch": 0.5250151356706478,
      "grad_norm": 0.6593502163887024,
      "learning_rate": 8.41732212316646e-06,
      "loss": 0.7455,
      "step": 9539
    },
    {
      "epoch": 0.5250701744730034,
      "grad_norm": 0.685553252696991,
      "learning_rate": 8.417005686707245e-06,
      "loss": 0.7783,
      "step": 9540
    },
    {
      "epoch": 0.5251252132753591,
      "grad_norm": 0.7003353238105774,
      "learning_rate": 8.41668922456673e-06,
      "loss": 0.7733,
      "step": 9541
    },
    {
      "epoch": 0.5251802520777148,
      "grad_norm": 0.7602891325950623,
      "learning_rate": 8.416372736747292e-06,
      "loss": 0.7236,
      "step": 9542
    },
    {
      "epoch": 0.5252352908800705,
      "grad_norm": 0.647531270980835,
      "learning_rate": 8.41605622325131e-06,
      "loss": 0.7388,
      "step": 9543
    },
    {
      "epoch": 0.5252903296824261,
      "grad_norm": 0.7309756875038147,
      "learning_rate": 8.415739684081165e-06,
      "loss": 0.7178,
      "step": 9544
    },
    {
      "epoch": 0.5253453684847817,
      "grad_norm": 0.6991532444953918,
      "learning_rate": 8.415423119239236e-06,
      "loss": 0.8078,
      "step": 9545
    },
    {
      "epoch": 0.5254004072871374,
      "grad_norm": 0.7392330765724182,
      "learning_rate": 8.4151065287279e-06,
      "loss": 0.8452,
      "step": 9546
    },
    {
      "epoch": 0.5254554460894931,
      "grad_norm": 0.7617329955101013,
      "learning_rate": 8.414789912549537e-06,
      "loss": 0.7885,
      "step": 9547
    },
    {
      "epoch": 0.5255104848918487,
      "grad_norm": 1.160125732421875,
      "learning_rate": 8.414473270706527e-06,
      "loss": 0.9628,
      "step": 9548
    },
    {
      "epoch": 0.5255655236942044,
      "grad_norm": 0.7578685879707336,
      "learning_rate": 8.414156603201252e-06,
      "loss": 0.7745,
      "step": 9549
    },
    {
      "epoch": 0.5256205624965601,
      "grad_norm": 0.6963017582893372,
      "learning_rate": 8.413839910036089e-06,
      "loss": 0.7693,
      "step": 9550
    },
    {
      "epoch": 0.5256756012989158,
      "grad_norm": 0.6631398797035217,
      "learning_rate": 8.413523191213415e-06,
      "loss": 0.6606,
      "step": 9551
    },
    {
      "epoch": 0.5257306401012714,
      "grad_norm": 0.707343339920044,
      "learning_rate": 8.41320644673562e-06,
      "loss": 0.7161,
      "step": 9552
    },
    {
      "epoch": 0.525785678903627,
      "grad_norm": 0.833448588848114,
      "learning_rate": 8.412889676605075e-06,
      "loss": 0.7509,
      "step": 9553
    },
    {
      "epoch": 0.5258407177059827,
      "grad_norm": 0.6214264631271362,
      "learning_rate": 8.412572880824168e-06,
      "loss": 0.7436,
      "step": 9554
    },
    {
      "epoch": 0.5258957565083384,
      "grad_norm": 0.6479233503341675,
      "learning_rate": 8.412256059395274e-06,
      "loss": 0.7359,
      "step": 9555
    },
    {
      "epoch": 0.525950795310694,
      "grad_norm": 0.7596501111984253,
      "learning_rate": 8.411939212320778e-06,
      "loss": 0.7422,
      "step": 9556
    },
    {
      "epoch": 0.5260058341130497,
      "grad_norm": 0.8040934205055237,
      "learning_rate": 8.41162233960306e-06,
      "loss": 0.7721,
      "step": 9557
    },
    {
      "epoch": 0.5260608729154054,
      "grad_norm": 0.7190027832984924,
      "learning_rate": 8.411305441244505e-06,
      "loss": 0.8794,
      "step": 9558
    },
    {
      "epoch": 0.5261159117177611,
      "grad_norm": 0.8002649545669556,
      "learning_rate": 8.410988517247486e-06,
      "loss": 0.7958,
      "step": 9559
    },
    {
      "epoch": 0.5261709505201166,
      "grad_norm": 0.7151750326156616,
      "learning_rate": 8.410671567614394e-06,
      "loss": 0.7597,
      "step": 9560
    },
    {
      "epoch": 0.5262259893224723,
      "grad_norm": 0.9718102812767029,
      "learning_rate": 8.410354592347607e-06,
      "loss": 0.8272,
      "step": 9561
    },
    {
      "epoch": 0.526281028124828,
      "grad_norm": 0.701932966709137,
      "learning_rate": 8.410037591449506e-06,
      "loss": 0.808,
      "step": 9562
    },
    {
      "epoch": 0.5263360669271837,
      "grad_norm": 0.8247585296630859,
      "learning_rate": 8.409720564922476e-06,
      "loss": 0.7598,
      "step": 9563
    },
    {
      "epoch": 0.5263911057295393,
      "grad_norm": 0.7305104732513428,
      "learning_rate": 8.409403512768899e-06,
      "loss": 0.8161,
      "step": 9564
    },
    {
      "epoch": 0.526446144531895,
      "grad_norm": 0.8726410865783691,
      "learning_rate": 8.409086434991158e-06,
      "loss": 0.8598,
      "step": 9565
    },
    {
      "epoch": 0.5265011833342507,
      "grad_norm": 0.7329155802726746,
      "learning_rate": 8.408769331591637e-06,
      "loss": 0.7355,
      "step": 9566
    },
    {
      "epoch": 0.5265562221366064,
      "grad_norm": 0.8227902054786682,
      "learning_rate": 8.408452202572716e-06,
      "loss": 0.7888,
      "step": 9567
    },
    {
      "epoch": 0.5266112609389619,
      "grad_norm": 0.7190666794776917,
      "learning_rate": 8.408135047936783e-06,
      "loss": 0.669,
      "step": 9568
    },
    {
      "epoch": 0.5266662997413176,
      "grad_norm": 0.6529938578605652,
      "learning_rate": 8.407817867686217e-06,
      "loss": 0.7345,
      "step": 9569
    },
    {
      "epoch": 0.5267213385436733,
      "grad_norm": 0.6985379457473755,
      "learning_rate": 8.407500661823407e-06,
      "loss": 0.852,
      "step": 9570
    },
    {
      "epoch": 0.526776377346029,
      "grad_norm": 0.7480047345161438,
      "learning_rate": 8.407183430350732e-06,
      "loss": 0.7422,
      "step": 9571
    },
    {
      "epoch": 0.5268314161483846,
      "grad_norm": 0.7599420547485352,
      "learning_rate": 8.406866173270579e-06,
      "loss": 0.7499,
      "step": 9572
    },
    {
      "epoch": 0.5268864549507403,
      "grad_norm": 0.813448965549469,
      "learning_rate": 8.406548890585331e-06,
      "loss": 0.7979,
      "step": 9573
    },
    {
      "epoch": 0.526941493753096,
      "grad_norm": 0.6029278039932251,
      "learning_rate": 8.406231582297374e-06,
      "loss": 0.7289,
      "step": 9574
    },
    {
      "epoch": 0.5269965325554516,
      "grad_norm": 0.656829297542572,
      "learning_rate": 8.40591424840909e-06,
      "loss": 0.6778,
      "step": 9575
    },
    {
      "epoch": 0.5270515713578072,
      "grad_norm": 0.7147198915481567,
      "learning_rate": 8.405596888922869e-06,
      "loss": 0.7212,
      "step": 9576
    },
    {
      "epoch": 0.5271066101601629,
      "grad_norm": 0.7722035050392151,
      "learning_rate": 8.405279503841094e-06,
      "loss": 0.8008,
      "step": 9577
    },
    {
      "epoch": 0.5271616489625186,
      "grad_norm": 0.6828493475914001,
      "learning_rate": 8.40496209316615e-06,
      "loss": 0.787,
      "step": 9578
    },
    {
      "epoch": 0.5272166877648742,
      "grad_norm": 0.6965187788009644,
      "learning_rate": 8.40464465690042e-06,
      "loss": 0.6803,
      "step": 9579
    },
    {
      "epoch": 0.5272717265672299,
      "grad_norm": 0.7300547957420349,
      "learning_rate": 8.404327195046293e-06,
      "loss": 0.8165,
      "step": 9580
    },
    {
      "epoch": 0.5273267653695856,
      "grad_norm": 0.7367526292800903,
      "learning_rate": 8.404009707606153e-06,
      "loss": 0.7709,
      "step": 9581
    },
    {
      "epoch": 0.5273818041719412,
      "grad_norm": 0.6694689989089966,
      "learning_rate": 8.40369219458239e-06,
      "loss": 0.7971,
      "step": 9582
    },
    {
      "epoch": 0.5274368429742968,
      "grad_norm": 0.6723141074180603,
      "learning_rate": 8.403374655977384e-06,
      "loss": 0.695,
      "step": 9583
    },
    {
      "epoch": 0.5274918817766525,
      "grad_norm": 0.7737089395523071,
      "learning_rate": 8.403057091793528e-06,
      "loss": 0.7765,
      "step": 9584
    },
    {
      "epoch": 0.5275469205790082,
      "grad_norm": 0.8378487825393677,
      "learning_rate": 8.402739502033204e-06,
      "loss": 0.7984,
      "step": 9585
    },
    {
      "epoch": 0.5276019593813639,
      "grad_norm": 0.7496509552001953,
      "learning_rate": 8.402421886698802e-06,
      "loss": 0.7846,
      "step": 9586
    },
    {
      "epoch": 0.5276569981837195,
      "grad_norm": 0.7020435929298401,
      "learning_rate": 8.402104245792706e-06,
      "loss": 0.8102,
      "step": 9587
    },
    {
      "epoch": 0.5277120369860752,
      "grad_norm": 0.8877277374267578,
      "learning_rate": 8.401786579317308e-06,
      "loss": 0.6995,
      "step": 9588
    },
    {
      "epoch": 0.5277670757884309,
      "grad_norm": 0.6975196599960327,
      "learning_rate": 8.401468887274991e-06,
      "loss": 0.7475,
      "step": 9589
    },
    {
      "epoch": 0.5278221145907865,
      "grad_norm": 0.8267357349395752,
      "learning_rate": 8.401151169668144e-06,
      "loss": 0.7091,
      "step": 9590
    },
    {
      "epoch": 0.5278771533931421,
      "grad_norm": 0.6778179407119751,
      "learning_rate": 8.400833426499156e-06,
      "loss": 0.8198,
      "step": 9591
    },
    {
      "epoch": 0.5279321921954978,
      "grad_norm": 0.7343330979347229,
      "learning_rate": 8.400515657770414e-06,
      "loss": 0.7565,
      "step": 9592
    },
    {
      "epoch": 0.5279872309978535,
      "grad_norm": 0.7745271325111389,
      "learning_rate": 8.400197863484307e-06,
      "loss": 0.7991,
      "step": 9593
    },
    {
      "epoch": 0.5280422698002092,
      "grad_norm": 0.7652345895767212,
      "learning_rate": 8.399880043643224e-06,
      "loss": 0.7752,
      "step": 9594
    },
    {
      "epoch": 0.5280973086025648,
      "grad_norm": 0.9764432311058044,
      "learning_rate": 8.399562198249551e-06,
      "loss": 0.784,
      "step": 9595
    },
    {
      "epoch": 0.5281523474049205,
      "grad_norm": 0.6763052940368652,
      "learning_rate": 8.399244327305678e-06,
      "loss": 0.7695,
      "step": 9596
    },
    {
      "epoch": 0.5282073862072761,
      "grad_norm": 0.7788934111595154,
      "learning_rate": 8.398926430813996e-06,
      "loss": 0.8152,
      "step": 9597
    },
    {
      "epoch": 0.5282624250096318,
      "grad_norm": 0.8088317513465881,
      "learning_rate": 8.398608508776894e-06,
      "loss": 0.7751,
      "step": 9598
    },
    {
      "epoch": 0.5283174638119874,
      "grad_norm": 0.6735319495201111,
      "learning_rate": 8.398290561196756e-06,
      "loss": 0.7305,
      "step": 9599
    },
    {
      "epoch": 0.5283725026143431,
      "grad_norm": 0.7279297113418579,
      "learning_rate": 8.39797258807598e-06,
      "loss": 0.7381,
      "step": 9600
    },
    {
      "epoch": 0.5284275414166988,
      "grad_norm": 0.74604332447052,
      "learning_rate": 8.39765458941695e-06,
      "loss": 0.8138,
      "step": 9601
    },
    {
      "epoch": 0.5284825802190545,
      "grad_norm": 0.7735850214958191,
      "learning_rate": 8.397336565222057e-06,
      "loss": 0.7364,
      "step": 9602
    },
    {
      "epoch": 0.52853761902141,
      "grad_norm": 0.7890003323554993,
      "learning_rate": 8.397018515493693e-06,
      "loss": 0.8301,
      "step": 9603
    },
    {
      "epoch": 0.5285926578237657,
      "grad_norm": 0.739054262638092,
      "learning_rate": 8.396700440234245e-06,
      "loss": 0.7503,
      "step": 9604
    },
    {
      "epoch": 0.5286476966261214,
      "grad_norm": 0.7611023783683777,
      "learning_rate": 8.396382339446108e-06,
      "loss": 0.7225,
      "step": 9605
    },
    {
      "epoch": 0.5287027354284771,
      "grad_norm": 0.770602285861969,
      "learning_rate": 8.39606421313167e-06,
      "loss": 0.71,
      "step": 9606
    },
    {
      "epoch": 0.5287577742308327,
      "grad_norm": 0.7495261430740356,
      "learning_rate": 8.395746061293322e-06,
      "loss": 0.7729,
      "step": 9607
    },
    {
      "epoch": 0.5288128130331884,
      "grad_norm": 0.7159668207168579,
      "learning_rate": 8.395427883933456e-06,
      "loss": 0.8457,
      "step": 9608
    },
    {
      "epoch": 0.5288678518355441,
      "grad_norm": 0.7663426399230957,
      "learning_rate": 8.395109681054463e-06,
      "loss": 0.784,
      "step": 9609
    },
    {
      "epoch": 0.5289228906378998,
      "grad_norm": 0.7271933555603027,
      "learning_rate": 8.394791452658732e-06,
      "loss": 0.7981,
      "step": 9610
    },
    {
      "epoch": 0.5289779294402553,
      "grad_norm": 0.7782096266746521,
      "learning_rate": 8.394473198748661e-06,
      "loss": 0.7953,
      "step": 9611
    },
    {
      "epoch": 0.529032968242611,
      "grad_norm": 0.8318955302238464,
      "learning_rate": 8.394154919326636e-06,
      "loss": 0.6875,
      "step": 9612
    },
    {
      "epoch": 0.5290880070449667,
      "grad_norm": 0.7402167916297913,
      "learning_rate": 8.393836614395051e-06,
      "loss": 0.7805,
      "step": 9613
    },
    {
      "epoch": 0.5291430458473224,
      "grad_norm": 0.6314370632171631,
      "learning_rate": 8.393518283956299e-06,
      "loss": 0.6841,
      "step": 9614
    },
    {
      "epoch": 0.529198084649678,
      "grad_norm": 0.8387365937232971,
      "learning_rate": 8.393199928012772e-06,
      "loss": 0.8503,
      "step": 9615
    },
    {
      "epoch": 0.5292531234520337,
      "grad_norm": 0.7066243886947632,
      "learning_rate": 8.392881546566863e-06,
      "loss": 0.8494,
      "step": 9616
    },
    {
      "epoch": 0.5293081622543894,
      "grad_norm": 0.7034226059913635,
      "learning_rate": 8.392563139620964e-06,
      "loss": 0.7335,
      "step": 9617
    },
    {
      "epoch": 0.5293632010567451,
      "grad_norm": 0.6969622373580933,
      "learning_rate": 8.392244707177468e-06,
      "loss": 0.7203,
      "step": 9618
    },
    {
      "epoch": 0.5294182398591006,
      "grad_norm": 0.7694050073623657,
      "learning_rate": 8.391926249238768e-06,
      "loss": 0.7864,
      "step": 9619
    },
    {
      "epoch": 0.5294732786614563,
      "grad_norm": 0.7284281253814697,
      "learning_rate": 8.391607765807262e-06,
      "loss": 0.6704,
      "step": 9620
    },
    {
      "epoch": 0.529528317463812,
      "grad_norm": 1.0466688871383667,
      "learning_rate": 8.391289256885337e-06,
      "loss": 0.7807,
      "step": 9621
    },
    {
      "epoch": 0.5295833562661676,
      "grad_norm": 0.7118388414382935,
      "learning_rate": 8.39097072247539e-06,
      "loss": 0.738,
      "step": 9622
    },
    {
      "epoch": 0.5296383950685233,
      "grad_norm": 0.794377863407135,
      "learning_rate": 8.390652162579815e-06,
      "loss": 0.6831,
      "step": 9623
    },
    {
      "epoch": 0.529693433870879,
      "grad_norm": 0.6042492389678955,
      "learning_rate": 8.390333577201007e-06,
      "loss": 0.6773,
      "step": 9624
    },
    {
      "epoch": 0.5297484726732347,
      "grad_norm": 0.6452521681785583,
      "learning_rate": 8.390014966341357e-06,
      "loss": 0.7168,
      "step": 9625
    },
    {
      "epoch": 0.5298035114755902,
      "grad_norm": 0.7113651633262634,
      "learning_rate": 8.389696330003265e-06,
      "loss": 0.709,
      "step": 9626
    },
    {
      "epoch": 0.5298585502779459,
      "grad_norm": 0.6469250917434692,
      "learning_rate": 8.38937766818912e-06,
      "loss": 0.6804,
      "step": 9627
    },
    {
      "epoch": 0.5299135890803016,
      "grad_norm": 0.7529417872428894,
      "learning_rate": 8.389058980901322e-06,
      "loss": 0.8537,
      "step": 9628
    },
    {
      "epoch": 0.5299686278826573,
      "grad_norm": 0.7681186199188232,
      "learning_rate": 8.388740268142262e-06,
      "loss": 0.7383,
      "step": 9629
    },
    {
      "epoch": 0.5300236666850129,
      "grad_norm": 0.6585648655891418,
      "learning_rate": 8.388421529914337e-06,
      "loss": 0.7535,
      "step": 9630
    },
    {
      "epoch": 0.5300787054873686,
      "grad_norm": 0.7432085871696472,
      "learning_rate": 8.388102766219943e-06,
      "loss": 0.7391,
      "step": 9631
    },
    {
      "epoch": 0.5301337442897243,
      "grad_norm": 0.6672815084457397,
      "learning_rate": 8.387783977061476e-06,
      "loss": 0.8056,
      "step": 9632
    },
    {
      "epoch": 0.53018878309208,
      "grad_norm": 0.7566675543785095,
      "learning_rate": 8.387465162441332e-06,
      "loss": 0.7858,
      "step": 9633
    },
    {
      "epoch": 0.5302438218944355,
      "grad_norm": 0.6522077322006226,
      "learning_rate": 8.387146322361907e-06,
      "loss": 0.759,
      "step": 9634
    },
    {
      "epoch": 0.5302988606967912,
      "grad_norm": 0.7246397137641907,
      "learning_rate": 8.386827456825597e-06,
      "loss": 0.8158,
      "step": 9635
    },
    {
      "epoch": 0.5303538994991469,
      "grad_norm": 0.7577807307243347,
      "learning_rate": 8.386508565834797e-06,
      "loss": 0.7495,
      "step": 9636
    },
    {
      "epoch": 0.5304089383015026,
      "grad_norm": 0.7080703973770142,
      "learning_rate": 8.386189649391906e-06,
      "loss": 0.8086,
      "step": 9637
    },
    {
      "epoch": 0.5304639771038582,
      "grad_norm": 0.7505277395248413,
      "learning_rate": 8.385870707499321e-06,
      "loss": 0.7206,
      "step": 9638
    },
    {
      "epoch": 0.5305190159062139,
      "grad_norm": 0.7044165134429932,
      "learning_rate": 8.385551740159437e-06,
      "loss": 0.7838,
      "step": 9639
    },
    {
      "epoch": 0.5305740547085696,
      "grad_norm": 0.7921645641326904,
      "learning_rate": 8.385232747374652e-06,
      "loss": 0.7604,
      "step": 9640
    },
    {
      "epoch": 0.5306290935109252,
      "grad_norm": 0.9930111169815063,
      "learning_rate": 8.384913729147364e-06,
      "loss": 0.7839,
      "step": 9641
    },
    {
      "epoch": 0.5306841323132808,
      "grad_norm": 0.7333244681358337,
      "learning_rate": 8.38459468547997e-06,
      "loss": 0.7941,
      "step": 9642
    },
    {
      "epoch": 0.5307391711156365,
      "grad_norm": 0.7857590913772583,
      "learning_rate": 8.384275616374868e-06,
      "loss": 0.8535,
      "step": 9643
    },
    {
      "epoch": 0.5307942099179922,
      "grad_norm": 0.8568746447563171,
      "learning_rate": 8.383956521834459e-06,
      "loss": 0.6586,
      "step": 9644
    },
    {
      "epoch": 0.5308492487203479,
      "grad_norm": 0.7061276435852051,
      "learning_rate": 8.383637401861136e-06,
      "loss": 0.7288,
      "step": 9645
    },
    {
      "epoch": 0.5309042875227035,
      "grad_norm": 0.7348940968513489,
      "learning_rate": 8.383318256457303e-06,
      "loss": 0.8099,
      "step": 9646
    },
    {
      "epoch": 0.5309593263250592,
      "grad_norm": 0.6526725888252258,
      "learning_rate": 8.382999085625353e-06,
      "loss": 0.6702,
      "step": 9647
    },
    {
      "epoch": 0.5310143651274148,
      "grad_norm": 0.8122747540473938,
      "learning_rate": 8.382679889367687e-06,
      "loss": 0.67,
      "step": 9648
    },
    {
      "epoch": 0.5310694039297705,
      "grad_norm": 0.9145376682281494,
      "learning_rate": 8.382360667686706e-06,
      "loss": 0.7719,
      "step": 9649
    },
    {
      "epoch": 0.5311244427321261,
      "grad_norm": 0.6659818887710571,
      "learning_rate": 8.382041420584807e-06,
      "loss": 0.806,
      "step": 9650
    },
    {
      "epoch": 0.5311794815344818,
      "grad_norm": 0.7088539004325867,
      "learning_rate": 8.381722148064391e-06,
      "loss": 0.7046,
      "step": 9651
    },
    {
      "epoch": 0.5312345203368375,
      "grad_norm": 0.8610590696334839,
      "learning_rate": 8.381402850127854e-06,
      "loss": 0.6998,
      "step": 9652
    },
    {
      "epoch": 0.5312895591391932,
      "grad_norm": 0.775830864906311,
      "learning_rate": 8.3810835267776e-06,
      "loss": 0.8874,
      "step": 9653
    },
    {
      "epoch": 0.5313445979415488,
      "grad_norm": 0.6871606707572937,
      "learning_rate": 8.380764178016028e-06,
      "loss": 0.7903,
      "step": 9654
    },
    {
      "epoch": 0.5313996367439044,
      "grad_norm": 0.7005272507667542,
      "learning_rate": 8.380444803845537e-06,
      "loss": 0.6685,
      "step": 9655
    },
    {
      "epoch": 0.5314546755462601,
      "grad_norm": 0.8922042846679688,
      "learning_rate": 8.380125404268527e-06,
      "loss": 0.7797,
      "step": 9656
    },
    {
      "epoch": 0.5315097143486158,
      "grad_norm": 0.7242267727851868,
      "learning_rate": 8.3798059792874e-06,
      "loss": 0.863,
      "step": 9657
    },
    {
      "epoch": 0.5315647531509714,
      "grad_norm": 0.6625328660011292,
      "learning_rate": 8.379486528904555e-06,
      "loss": 0.7,
      "step": 9658
    },
    {
      "epoch": 0.5316197919533271,
      "grad_norm": 0.9882226586341858,
      "learning_rate": 8.379167053122394e-06,
      "loss": 0.7534,
      "step": 9659
    },
    {
      "epoch": 0.5316748307556828,
      "grad_norm": 0.6894702911376953,
      "learning_rate": 8.378847551943318e-06,
      "loss": 0.7503,
      "step": 9660
    },
    {
      "epoch": 0.5317298695580385,
      "grad_norm": 0.6820259690284729,
      "learning_rate": 8.37852802536973e-06,
      "loss": 0.7713,
      "step": 9661
    },
    {
      "epoch": 0.531784908360394,
      "grad_norm": 0.667918860912323,
      "learning_rate": 8.378208473404028e-06,
      "loss": 0.7524,
      "step": 9662
    },
    {
      "epoch": 0.5318399471627497,
      "grad_norm": 0.7789241075515747,
      "learning_rate": 8.377888896048617e-06,
      "loss": 0.6906,
      "step": 9663
    },
    {
      "epoch": 0.5318949859651054,
      "grad_norm": 0.7264542579650879,
      "learning_rate": 8.377569293305894e-06,
      "loss": 0.7836,
      "step": 9664
    },
    {
      "epoch": 0.531950024767461,
      "grad_norm": 0.6979835629463196,
      "learning_rate": 8.377249665178267e-06,
      "loss": 0.7739,
      "step": 9665
    },
    {
      "epoch": 0.5320050635698167,
      "grad_norm": 0.8008072376251221,
      "learning_rate": 8.376930011668136e-06,
      "loss": 0.7853,
      "step": 9666
    },
    {
      "epoch": 0.5320601023721724,
      "grad_norm": 0.7185621857643127,
      "learning_rate": 8.376610332777901e-06,
      "loss": 0.7311,
      "step": 9667
    },
    {
      "epoch": 0.5321151411745281,
      "grad_norm": 0.7644047141075134,
      "learning_rate": 8.376290628509969e-06,
      "loss": 0.6919,
      "step": 9668
    },
    {
      "epoch": 0.5321701799768837,
      "grad_norm": 0.7387600541114807,
      "learning_rate": 8.37597089886674e-06,
      "loss": 0.7285,
      "step": 9669
    },
    {
      "epoch": 0.5322252187792393,
      "grad_norm": 0.7344895005226135,
      "learning_rate": 8.375651143850614e-06,
      "loss": 0.7514,
      "step": 9670
    },
    {
      "epoch": 0.532280257581595,
      "grad_norm": 0.6930707097053528,
      "learning_rate": 8.375331363464002e-06,
      "loss": 0.8318,
      "step": 9671
    },
    {
      "epoch": 0.5323352963839507,
      "grad_norm": 0.678162693977356,
      "learning_rate": 8.3750115577093e-06,
      "loss": 0.7123,
      "step": 9672
    },
    {
      "epoch": 0.5323903351863063,
      "grad_norm": 0.7780481576919556,
      "learning_rate": 8.374691726588914e-06,
      "loss": 0.7672,
      "step": 9673
    },
    {
      "epoch": 0.532445373988662,
      "grad_norm": 0.6664674282073975,
      "learning_rate": 8.374371870105252e-06,
      "loss": 0.6994,
      "step": 9674
    },
    {
      "epoch": 0.5325004127910177,
      "grad_norm": 0.6952562928199768,
      "learning_rate": 8.374051988260712e-06,
      "loss": 0.8638,
      "step": 9675
    },
    {
      "epoch": 0.5325554515933734,
      "grad_norm": 0.764005184173584,
      "learning_rate": 8.373732081057699e-06,
      "loss": 0.756,
      "step": 9676
    },
    {
      "epoch": 0.5326104903957289,
      "grad_norm": 0.9434393048286438,
      "learning_rate": 8.373412148498621e-06,
      "loss": 0.8668,
      "step": 9677
    },
    {
      "epoch": 0.5326655291980846,
      "grad_norm": 0.752609133720398,
      "learning_rate": 8.373092190585878e-06,
      "loss": 0.8078,
      "step": 9678
    },
    {
      "epoch": 0.5327205680004403,
      "grad_norm": 0.671940803527832,
      "learning_rate": 8.37277220732188e-06,
      "loss": 0.7726,
      "step": 9679
    },
    {
      "epoch": 0.532775606802796,
      "grad_norm": 0.7824863791465759,
      "learning_rate": 8.372452198709027e-06,
      "loss": 0.8246,
      "step": 9680
    },
    {
      "epoch": 0.5328306456051516,
      "grad_norm": 0.7300587892532349,
      "learning_rate": 8.372132164749726e-06,
      "loss": 0.7953,
      "step": 9681
    },
    {
      "epoch": 0.5328856844075073,
      "grad_norm": 0.7146018743515015,
      "learning_rate": 8.371812105446384e-06,
      "loss": 0.7409,
      "step": 9682
    },
    {
      "epoch": 0.532940723209863,
      "grad_norm": 0.73857581615448,
      "learning_rate": 8.371492020801404e-06,
      "loss": 0.8067,
      "step": 9683
    },
    {
      "epoch": 0.5329957620122187,
      "grad_norm": 0.6760877966880798,
      "learning_rate": 8.37117191081719e-06,
      "loss": 0.7363,
      "step": 9684
    },
    {
      "epoch": 0.5330508008145742,
      "grad_norm": 0.766482412815094,
      "learning_rate": 8.370851775496154e-06,
      "loss": 0.7358,
      "step": 9685
    },
    {
      "epoch": 0.5331058396169299,
      "grad_norm": 0.7230576276779175,
      "learning_rate": 8.370531614840697e-06,
      "loss": 0.8154,
      "step": 9686
    },
    {
      "epoch": 0.5331608784192856,
      "grad_norm": 0.7357933521270752,
      "learning_rate": 8.370211428853225e-06,
      "loss": 0.7187,
      "step": 9687
    },
    {
      "epoch": 0.5332159172216413,
      "grad_norm": 0.8208534121513367,
      "learning_rate": 8.369891217536148e-06,
      "loss": 0.8037,
      "step": 9688
    },
    {
      "epoch": 0.5332709560239969,
      "grad_norm": 0.6771863698959351,
      "learning_rate": 8.36957098089187e-06,
      "loss": 0.733,
      "step": 9689
    },
    {
      "epoch": 0.5333259948263526,
      "grad_norm": 0.6382480263710022,
      "learning_rate": 8.369250718922798e-06,
      "loss": 0.7391,
      "step": 9690
    },
    {
      "epoch": 0.5333810336287083,
      "grad_norm": 0.6638994812965393,
      "learning_rate": 8.368930431631342e-06,
      "loss": 0.7176,
      "step": 9691
    },
    {
      "epoch": 0.533436072431064,
      "grad_norm": 0.7599604725837708,
      "learning_rate": 8.368610119019903e-06,
      "loss": 0.8814,
      "step": 9692
    },
    {
      "epoch": 0.5334911112334195,
      "grad_norm": 0.6896547079086304,
      "learning_rate": 8.368289781090894e-06,
      "loss": 0.7618,
      "step": 9693
    },
    {
      "epoch": 0.5335461500357752,
      "grad_norm": 0.7081224918365479,
      "learning_rate": 8.36796941784672e-06,
      "loss": 0.656,
      "step": 9694
    },
    {
      "epoch": 0.5336011888381309,
      "grad_norm": 0.8819646835327148,
      "learning_rate": 8.367649029289791e-06,
      "loss": 0.8946,
      "step": 9695
    },
    {
      "epoch": 0.5336562276404866,
      "grad_norm": 0.6597925424575806,
      "learning_rate": 8.367328615422512e-06,
      "loss": 0.6891,
      "step": 9696
    },
    {
      "epoch": 0.5337112664428422,
      "grad_norm": 0.6855770945549011,
      "learning_rate": 8.367008176247294e-06,
      "loss": 0.7158,
      "step": 9697
    },
    {
      "epoch": 0.5337663052451979,
      "grad_norm": 0.6874905228614807,
      "learning_rate": 8.366687711766541e-06,
      "loss": 0.7445,
      "step": 9698
    },
    {
      "epoch": 0.5338213440475535,
      "grad_norm": 0.6990895867347717,
      "learning_rate": 8.366367221982666e-06,
      "loss": 0.6189,
      "step": 9699
    },
    {
      "epoch": 0.5338763828499092,
      "grad_norm": 0.7235365509986877,
      "learning_rate": 8.366046706898075e-06,
      "loss": 0.6406,
      "step": 9700
    },
    {
      "epoch": 0.5339314216522648,
      "grad_norm": 0.7563154697418213,
      "learning_rate": 8.36572616651518e-06,
      "loss": 0.7798,
      "step": 9701
    },
    {
      "epoch": 0.5339864604546205,
      "grad_norm": 0.6845980286598206,
      "learning_rate": 8.365405600836387e-06,
      "loss": 0.7665,
      "step": 9702
    },
    {
      "epoch": 0.5340414992569762,
      "grad_norm": 0.6374378204345703,
      "learning_rate": 8.365085009864106e-06,
      "loss": 0.6935,
      "step": 9703
    },
    {
      "epoch": 0.5340965380593319,
      "grad_norm": 0.726672887802124,
      "learning_rate": 8.364764393600747e-06,
      "loss": 0.7821,
      "step": 9704
    },
    {
      "epoch": 0.5341515768616875,
      "grad_norm": 0.6784456372261047,
      "learning_rate": 8.364443752048719e-06,
      "loss": 0.7722,
      "step": 9705
    },
    {
      "epoch": 0.5342066156640431,
      "grad_norm": 0.6344080567359924,
      "learning_rate": 8.364123085210433e-06,
      "loss": 0.7256,
      "step": 9706
    },
    {
      "epoch": 0.5342616544663988,
      "grad_norm": 0.7913152575492859,
      "learning_rate": 8.363802393088299e-06,
      "loss": 0.7892,
      "step": 9707
    },
    {
      "epoch": 0.5343166932687544,
      "grad_norm": 0.6792107820510864,
      "learning_rate": 8.363481675684726e-06,
      "loss": 0.7374,
      "step": 9708
    },
    {
      "epoch": 0.5343717320711101,
      "grad_norm": 1.0153685808181763,
      "learning_rate": 8.363160933002126e-06,
      "loss": 0.7396,
      "step": 9709
    },
    {
      "epoch": 0.5344267708734658,
      "grad_norm": 0.7655258774757385,
      "learning_rate": 8.362840165042906e-06,
      "loss": 0.7746,
      "step": 9710
    },
    {
      "epoch": 0.5344818096758215,
      "grad_norm": 0.7830179929733276,
      "learning_rate": 8.362519371809483e-06,
      "loss": 0.7082,
      "step": 9711
    },
    {
      "epoch": 0.5345368484781771,
      "grad_norm": 0.7410556674003601,
      "learning_rate": 8.362198553304261e-06,
      "loss": 0.7055,
      "step": 9712
    },
    {
      "epoch": 0.5345918872805328,
      "grad_norm": 0.6542297005653381,
      "learning_rate": 8.361877709529658e-06,
      "loss": 0.7153,
      "step": 9713
    },
    {
      "epoch": 0.5346469260828884,
      "grad_norm": 0.6752653121948242,
      "learning_rate": 8.36155684048808e-06,
      "loss": 0.6901,
      "step": 9714
    },
    {
      "epoch": 0.5347019648852441,
      "grad_norm": 0.7158684134483337,
      "learning_rate": 8.361235946181943e-06,
      "loss": 0.7775,
      "step": 9715
    },
    {
      "epoch": 0.5347570036875997,
      "grad_norm": 0.6174392700195312,
      "learning_rate": 8.360915026613652e-06,
      "loss": 0.6501,
      "step": 9716
    },
    {
      "epoch": 0.5348120424899554,
      "grad_norm": 0.7110500931739807,
      "learning_rate": 8.360594081785627e-06,
      "loss": 0.742,
      "step": 9717
    },
    {
      "epoch": 0.5348670812923111,
      "grad_norm": 0.8456488251686096,
      "learning_rate": 8.360273111700276e-06,
      "loss": 0.8237,
      "step": 9718
    },
    {
      "epoch": 0.5349221200946668,
      "grad_norm": 0.6660711169242859,
      "learning_rate": 8.359952116360011e-06,
      "loss": 0.7856,
      "step": 9719
    },
    {
      "epoch": 0.5349771588970224,
      "grad_norm": 0.7661204934120178,
      "learning_rate": 8.359631095767244e-06,
      "loss": 0.8336,
      "step": 9720
    },
    {
      "epoch": 0.535032197699378,
      "grad_norm": 0.7747855186462402,
      "learning_rate": 8.359310049924392e-06,
      "loss": 0.7302,
      "step": 9721
    },
    {
      "epoch": 0.5350872365017337,
      "grad_norm": 0.8156001567840576,
      "learning_rate": 8.358988978833864e-06,
      "loss": 0.7878,
      "step": 9722
    },
    {
      "epoch": 0.5351422753040894,
      "grad_norm": 0.7371010780334473,
      "learning_rate": 8.358667882498073e-06,
      "loss": 0.803,
      "step": 9723
    },
    {
      "epoch": 0.535197314106445,
      "grad_norm": 0.7141744494438171,
      "learning_rate": 8.358346760919431e-06,
      "loss": 0.687,
      "step": 9724
    },
    {
      "epoch": 0.5352523529088007,
      "grad_norm": 0.6395956873893738,
      "learning_rate": 8.358025614100358e-06,
      "loss": 0.7052,
      "step": 9725
    },
    {
      "epoch": 0.5353073917111564,
      "grad_norm": 0.7135289311408997,
      "learning_rate": 8.35770444204326e-06,
      "loss": 0.7882,
      "step": 9726
    },
    {
      "epoch": 0.5353624305135121,
      "grad_norm": 0.702408492565155,
      "learning_rate": 8.357383244750557e-06,
      "loss": 0.6965,
      "step": 9727
    },
    {
      "epoch": 0.5354174693158676,
      "grad_norm": 0.731193482875824,
      "learning_rate": 8.357062022224658e-06,
      "loss": 0.7525,
      "step": 9728
    },
    {
      "epoch": 0.5354725081182233,
      "grad_norm": 0.8115057945251465,
      "learning_rate": 8.356740774467982e-06,
      "loss": 0.7466,
      "step": 9729
    },
    {
      "epoch": 0.535527546920579,
      "grad_norm": 0.8644380569458008,
      "learning_rate": 8.356419501482938e-06,
      "loss": 0.7989,
      "step": 9730
    },
    {
      "epoch": 0.5355825857229347,
      "grad_norm": 1.414620041847229,
      "learning_rate": 8.356098203271945e-06,
      "loss": 0.7782,
      "step": 9731
    },
    {
      "epoch": 0.5356376245252903,
      "grad_norm": 0.7355421185493469,
      "learning_rate": 8.355776879837417e-06,
      "loss": 0.7163,
      "step": 9732
    },
    {
      "epoch": 0.535692663327646,
      "grad_norm": 0.6556879281997681,
      "learning_rate": 8.355455531181766e-06,
      "loss": 0.7543,
      "step": 9733
    },
    {
      "epoch": 0.5357477021300017,
      "grad_norm": 0.6632516980171204,
      "learning_rate": 8.355134157307412e-06,
      "loss": 0.7382,
      "step": 9734
    },
    {
      "epoch": 0.5358027409323574,
      "grad_norm": 0.7096145153045654,
      "learning_rate": 8.354812758216767e-06,
      "loss": 0.7797,
      "step": 9735
    },
    {
      "epoch": 0.5358577797347129,
      "grad_norm": 0.6404649019241333,
      "learning_rate": 8.354491333912244e-06,
      "loss": 0.6637,
      "step": 9736
    },
    {
      "epoch": 0.5359128185370686,
      "grad_norm": 0.6987022757530212,
      "learning_rate": 8.354169884396266e-06,
      "loss": 0.7682,
      "step": 9737
    },
    {
      "epoch": 0.5359678573394243,
      "grad_norm": 0.6593581438064575,
      "learning_rate": 8.353848409671245e-06,
      "loss": 0.6747,
      "step": 9738
    },
    {
      "epoch": 0.53602289614178,
      "grad_norm": 0.6999880075454712,
      "learning_rate": 8.353526909739596e-06,
      "loss": 0.6659,
      "step": 9739
    },
    {
      "epoch": 0.5360779349441356,
      "grad_norm": 0.6448989510536194,
      "learning_rate": 8.353205384603735e-06,
      "loss": 0.7297,
      "step": 9740
    },
    {
      "epoch": 0.5361329737464913,
      "grad_norm": 0.6666765213012695,
      "learning_rate": 8.352883834266082e-06,
      "loss": 0.6459,
      "step": 9741
    },
    {
      "epoch": 0.536188012548847,
      "grad_norm": 0.8020225763320923,
      "learning_rate": 8.352562258729051e-06,
      "loss": 0.8122,
      "step": 9742
    },
    {
      "epoch": 0.5362430513512026,
      "grad_norm": 0.6883382201194763,
      "learning_rate": 8.35224065799506e-06,
      "loss": 0.7084,
      "step": 9743
    },
    {
      "epoch": 0.5362980901535582,
      "grad_norm": 0.7366660237312317,
      "learning_rate": 8.351919032066525e-06,
      "loss": 0.848,
      "step": 9744
    },
    {
      "epoch": 0.5363531289559139,
      "grad_norm": 0.7408311367034912,
      "learning_rate": 8.351597380945863e-06,
      "loss": 0.798,
      "step": 9745
    },
    {
      "epoch": 0.5364081677582696,
      "grad_norm": 0.6841676235198975,
      "learning_rate": 8.351275704635495e-06,
      "loss": 0.7372,
      "step": 9746
    },
    {
      "epoch": 0.5364632065606253,
      "grad_norm": 0.6903505325317383,
      "learning_rate": 8.350954003137833e-06,
      "loss": 0.7371,
      "step": 9747
    },
    {
      "epoch": 0.5365182453629809,
      "grad_norm": 0.6444700956344604,
      "learning_rate": 8.350632276455298e-06,
      "loss": 0.6685,
      "step": 9748
    },
    {
      "epoch": 0.5365732841653366,
      "grad_norm": 0.6821029186248779,
      "learning_rate": 8.350310524590307e-06,
      "loss": 0.8796,
      "step": 9749
    },
    {
      "epoch": 0.5366283229676923,
      "grad_norm": 0.6733999848365784,
      "learning_rate": 8.349988747545282e-06,
      "loss": 0.6833,
      "step": 9750
    },
    {
      "epoch": 0.5366833617700478,
      "grad_norm": 0.8097321391105652,
      "learning_rate": 8.349666945322636e-06,
      "loss": 0.834,
      "step": 9751
    },
    {
      "epoch": 0.5367384005724035,
      "grad_norm": 0.7692395448684692,
      "learning_rate": 8.34934511792479e-06,
      "loss": 0.7866,
      "step": 9752
    },
    {
      "epoch": 0.5367934393747592,
      "grad_norm": 0.7551112174987793,
      "learning_rate": 8.349023265354164e-06,
      "loss": 0.8378,
      "step": 9753
    },
    {
      "epoch": 0.5368484781771149,
      "grad_norm": 0.5796393156051636,
      "learning_rate": 8.348701387613176e-06,
      "loss": 0.5995,
      "step": 9754
    },
    {
      "epoch": 0.5369035169794705,
      "grad_norm": 0.6839799284934998,
      "learning_rate": 8.348379484704244e-06,
      "loss": 0.8262,
      "step": 9755
    },
    {
      "epoch": 0.5369585557818262,
      "grad_norm": 0.7710869908332825,
      "learning_rate": 8.348057556629786e-06,
      "loss": 0.7796,
      "step": 9756
    },
    {
      "epoch": 0.5370135945841819,
      "grad_norm": 0.733096718788147,
      "learning_rate": 8.347735603392225e-06,
      "loss": 0.8233,
      "step": 9757
    },
    {
      "epoch": 0.5370686333865375,
      "grad_norm": 0.6438466906547546,
      "learning_rate": 8.347413624993982e-06,
      "loss": 0.7582,
      "step": 9758
    },
    {
      "epoch": 0.5371236721888931,
      "grad_norm": 0.6877560615539551,
      "learning_rate": 8.34709162143747e-06,
      "loss": 0.7428,
      "step": 9759
    },
    {
      "epoch": 0.5371787109912488,
      "grad_norm": 1.060831069946289,
      "learning_rate": 8.346769592725115e-06,
      "loss": 0.8636,
      "step": 9760
    },
    {
      "epoch": 0.5372337497936045,
      "grad_norm": 0.6828434467315674,
      "learning_rate": 8.346447538859334e-06,
      "loss": 0.7801,
      "step": 9761
    },
    {
      "epoch": 0.5372887885959602,
      "grad_norm": 0.6784753203392029,
      "learning_rate": 8.346125459842552e-06,
      "loss": 0.7356,
      "step": 9762
    },
    {
      "epoch": 0.5373438273983158,
      "grad_norm": 0.6493560075759888,
      "learning_rate": 8.345803355677185e-06,
      "loss": 0.749,
      "step": 9763
    },
    {
      "epoch": 0.5373988662006715,
      "grad_norm": 0.7109258770942688,
      "learning_rate": 8.345481226365657e-06,
      "loss": 0.7599,
      "step": 9764
    },
    {
      "epoch": 0.5374539050030271,
      "grad_norm": 0.8526985049247742,
      "learning_rate": 8.345159071910387e-06,
      "loss": 0.6605,
      "step": 9765
    },
    {
      "epoch": 0.5375089438053828,
      "grad_norm": 0.9194039702415466,
      "learning_rate": 8.344836892313797e-06,
      "loss": 0.794,
      "step": 9766
    },
    {
      "epoch": 0.5375639826077384,
      "grad_norm": 0.7258954048156738,
      "learning_rate": 8.344514687578307e-06,
      "loss": 0.871,
      "step": 9767
    },
    {
      "epoch": 0.5376190214100941,
      "grad_norm": 0.7099377512931824,
      "learning_rate": 8.34419245770634e-06,
      "loss": 0.8098,
      "step": 9768
    },
    {
      "epoch": 0.5376740602124498,
      "grad_norm": 0.7883020639419556,
      "learning_rate": 8.34387020270032e-06,
      "loss": 0.8383,
      "step": 9769
    },
    {
      "epoch": 0.5377290990148055,
      "grad_norm": 0.7009730339050293,
      "learning_rate": 8.343547922562664e-06,
      "loss": 0.7794,
      "step": 9770
    },
    {
      "epoch": 0.5377841378171611,
      "grad_norm": 0.6569581031799316,
      "learning_rate": 8.343225617295798e-06,
      "loss": 0.7574,
      "step": 9771
    },
    {
      "epoch": 0.5378391766195167,
      "grad_norm": 0.6159278154373169,
      "learning_rate": 8.342903286902142e-06,
      "loss": 0.7136,
      "step": 9772
    },
    {
      "epoch": 0.5378942154218724,
      "grad_norm": 0.6594879627227783,
      "learning_rate": 8.342580931384121e-06,
      "loss": 0.6906,
      "step": 9773
    },
    {
      "epoch": 0.5379492542242281,
      "grad_norm": 0.7002933025360107,
      "learning_rate": 8.342258550744156e-06,
      "loss": 0.7272,
      "step": 9774
    },
    {
      "epoch": 0.5380042930265837,
      "grad_norm": 0.8243216276168823,
      "learning_rate": 8.341936144984672e-06,
      "loss": 0.8105,
      "step": 9775
    },
    {
      "epoch": 0.5380593318289394,
      "grad_norm": 0.8358921408653259,
      "learning_rate": 8.34161371410809e-06,
      "loss": 0.7118,
      "step": 9776
    },
    {
      "epoch": 0.5381143706312951,
      "grad_norm": 0.6339066028594971,
      "learning_rate": 8.34129125811683e-06,
      "loss": 0.7035,
      "step": 9777
    },
    {
      "epoch": 0.5381694094336508,
      "grad_norm": 0.7407625317573547,
      "learning_rate": 8.340968777013324e-06,
      "loss": 0.7447,
      "step": 9778
    },
    {
      "epoch": 0.5382244482360063,
      "grad_norm": 0.6876600384712219,
      "learning_rate": 8.340646270799991e-06,
      "loss": 0.7298,
      "step": 9779
    },
    {
      "epoch": 0.538279487038362,
      "grad_norm": 0.7021264433860779,
      "learning_rate": 8.340323739479251e-06,
      "loss": 0.7869,
      "step": 9780
    },
    {
      "epoch": 0.5383345258407177,
      "grad_norm": 0.7341023087501526,
      "learning_rate": 8.340001183053535e-06,
      "loss": 0.7447,
      "step": 9781
    },
    {
      "epoch": 0.5383895646430734,
      "grad_norm": 0.6829406023025513,
      "learning_rate": 8.339678601525263e-06,
      "loss": 0.7438,
      "step": 9782
    },
    {
      "epoch": 0.538444603445429,
      "grad_norm": 0.7671583294868469,
      "learning_rate": 8.33935599489686e-06,
      "loss": 0.8678,
      "step": 9783
    },
    {
      "epoch": 0.5384996422477847,
      "grad_norm": 0.701797366142273,
      "learning_rate": 8.339033363170753e-06,
      "loss": 0.8431,
      "step": 9784
    },
    {
      "epoch": 0.5385546810501404,
      "grad_norm": 0.748235285282135,
      "learning_rate": 8.338710706349363e-06,
      "loss": 0.7905,
      "step": 9785
    },
    {
      "epoch": 0.5386097198524961,
      "grad_norm": 0.8202430605888367,
      "learning_rate": 8.338388024435119e-06,
      "loss": 0.7734,
      "step": 9786
    },
    {
      "epoch": 0.5386647586548516,
      "grad_norm": 0.8218014240264893,
      "learning_rate": 8.338065317430442e-06,
      "loss": 0.846,
      "step": 9787
    },
    {
      "epoch": 0.5387197974572073,
      "grad_norm": 0.6773214936256409,
      "learning_rate": 8.337742585337762e-06,
      "loss": 0.7692,
      "step": 9788
    },
    {
      "epoch": 0.538774836259563,
      "grad_norm": 0.7011464834213257,
      "learning_rate": 8.337419828159501e-06,
      "loss": 0.7534,
      "step": 9789
    },
    {
      "epoch": 0.5388298750619187,
      "grad_norm": 0.8299004435539246,
      "learning_rate": 8.337097045898087e-06,
      "loss": 0.7997,
      "step": 9790
    },
    {
      "epoch": 0.5388849138642743,
      "grad_norm": 0.8600753545761108,
      "learning_rate": 8.336774238555942e-06,
      "loss": 0.8307,
      "step": 9791
    },
    {
      "epoch": 0.53893995266663,
      "grad_norm": 0.676490843296051,
      "learning_rate": 8.336451406135498e-06,
      "loss": 0.7748,
      "step": 9792
    },
    {
      "epoch": 0.5389949914689857,
      "grad_norm": 0.7094627618789673,
      "learning_rate": 8.336128548639177e-06,
      "loss": 0.7524,
      "step": 9793
    },
    {
      "epoch": 0.5390500302713412,
      "grad_norm": 0.6804066896438599,
      "learning_rate": 8.335805666069407e-06,
      "loss": 0.8299,
      "step": 9794
    },
    {
      "epoch": 0.5391050690736969,
      "grad_norm": 0.6992025971412659,
      "learning_rate": 8.335482758428614e-06,
      "loss": 0.7548,
      "step": 9795
    },
    {
      "epoch": 0.5391601078760526,
      "grad_norm": 0.6649640798568726,
      "learning_rate": 8.335159825719227e-06,
      "loss": 0.6595,
      "step": 9796
    },
    {
      "epoch": 0.5392151466784083,
      "grad_norm": 0.7292002439498901,
      "learning_rate": 8.33483686794367e-06,
      "loss": 0.7944,
      "step": 9797
    },
    {
      "epoch": 0.5392701854807639,
      "grad_norm": 0.9124587178230286,
      "learning_rate": 8.334513885104375e-06,
      "loss": 0.8586,
      "step": 9798
    },
    {
      "epoch": 0.5393252242831196,
      "grad_norm": 0.7091020941734314,
      "learning_rate": 8.334190877203761e-06,
      "loss": 0.7019,
      "step": 9799
    },
    {
      "epoch": 0.5393802630854753,
      "grad_norm": 0.7470952272415161,
      "learning_rate": 8.333867844244265e-06,
      "loss": 0.7866,
      "step": 9800
    },
    {
      "epoch": 0.539435301887831,
      "grad_norm": 0.7368966341018677,
      "learning_rate": 8.333544786228309e-06,
      "loss": 0.8135,
      "step": 9801
    },
    {
      "epoch": 0.5394903406901865,
      "grad_norm": 0.668305516242981,
      "learning_rate": 8.333221703158322e-06,
      "loss": 0.7549,
      "step": 9802
    },
    {
      "epoch": 0.5395453794925422,
      "grad_norm": 0.6788874268531799,
      "learning_rate": 8.332898595036735e-06,
      "loss": 0.8077,
      "step": 9803
    },
    {
      "epoch": 0.5396004182948979,
      "grad_norm": 0.654863715171814,
      "learning_rate": 8.332575461865972e-06,
      "loss": 0.7695,
      "step": 9804
    },
    {
      "epoch": 0.5396554570972536,
      "grad_norm": 0.7460314631462097,
      "learning_rate": 8.332252303648464e-06,
      "loss": 0.7711,
      "step": 9805
    },
    {
      "epoch": 0.5397104958996092,
      "grad_norm": 0.7923582792282104,
      "learning_rate": 8.331929120386643e-06,
      "loss": 0.7348,
      "step": 9806
    },
    {
      "epoch": 0.5397655347019649,
      "grad_norm": 0.6570843458175659,
      "learning_rate": 8.331605912082932e-06,
      "loss": 0.7029,
      "step": 9807
    },
    {
      "epoch": 0.5398205735043206,
      "grad_norm": 0.7728865742683411,
      "learning_rate": 8.331282678739762e-06,
      "loss": 0.8249,
      "step": 9808
    },
    {
      "epoch": 0.5398756123066762,
      "grad_norm": 0.7121468186378479,
      "learning_rate": 8.330959420359565e-06,
      "loss": 0.8698,
      "step": 9809
    },
    {
      "epoch": 0.5399306511090318,
      "grad_norm": 0.7779444456100464,
      "learning_rate": 8.330636136944768e-06,
      "loss": 0.7448,
      "step": 9810
    },
    {
      "epoch": 0.5399856899113875,
      "grad_norm": 0.7770833373069763,
      "learning_rate": 8.330312828497801e-06,
      "loss": 0.8489,
      "step": 9811
    },
    {
      "epoch": 0.5400407287137432,
      "grad_norm": 0.6705769896507263,
      "learning_rate": 8.329989495021096e-06,
      "loss": 0.7349,
      "step": 9812
    },
    {
      "epoch": 0.5400957675160989,
      "grad_norm": 0.6775381565093994,
      "learning_rate": 8.329666136517079e-06,
      "loss": 0.8093,
      "step": 9813
    },
    {
      "epoch": 0.5401508063184545,
      "grad_norm": 0.6621832251548767,
      "learning_rate": 8.329342752988183e-06,
      "loss": 0.7877,
      "step": 9814
    },
    {
      "epoch": 0.5402058451208102,
      "grad_norm": 0.704339861869812,
      "learning_rate": 8.329019344436839e-06,
      "loss": 0.7708,
      "step": 9815
    },
    {
      "epoch": 0.5402608839231658,
      "grad_norm": 0.789944052696228,
      "learning_rate": 8.328695910865476e-06,
      "loss": 0.7563,
      "step": 9816
    },
    {
      "epoch": 0.5403159227255215,
      "grad_norm": 0.6997420191764832,
      "learning_rate": 8.328372452276525e-06,
      "loss": 0.7023,
      "step": 9817
    },
    {
      "epoch": 0.5403709615278771,
      "grad_norm": 0.6453180313110352,
      "learning_rate": 8.328048968672418e-06,
      "loss": 0.7193,
      "step": 9818
    },
    {
      "epoch": 0.5404260003302328,
      "grad_norm": 0.7059640884399414,
      "learning_rate": 8.327725460055586e-06,
      "loss": 0.7875,
      "step": 9819
    },
    {
      "epoch": 0.5404810391325885,
      "grad_norm": 0.7725005745887756,
      "learning_rate": 8.327401926428461e-06,
      "loss": 0.7503,
      "step": 9820
    },
    {
      "epoch": 0.5405360779349442,
      "grad_norm": 0.7710940837860107,
      "learning_rate": 8.327078367793473e-06,
      "loss": 0.8314,
      "step": 9821
    },
    {
      "epoch": 0.5405911167372998,
      "grad_norm": 0.9090666770935059,
      "learning_rate": 8.326754784153055e-06,
      "loss": 0.8021,
      "step": 9822
    },
    {
      "epoch": 0.5406461555396554,
      "grad_norm": 0.7135322690010071,
      "learning_rate": 8.326431175509638e-06,
      "loss": 0.8084,
      "step": 9823
    },
    {
      "epoch": 0.5407011943420111,
      "grad_norm": 0.9126102328300476,
      "learning_rate": 8.326107541865656e-06,
      "loss": 0.75,
      "step": 9824
    },
    {
      "epoch": 0.5407562331443668,
      "grad_norm": 0.7263361215591431,
      "learning_rate": 8.325783883223539e-06,
      "loss": 0.6808,
      "step": 9825
    },
    {
      "epoch": 0.5408112719467224,
      "grad_norm": 0.7234700918197632,
      "learning_rate": 8.32546019958572e-06,
      "loss": 0.7582,
      "step": 9826
    },
    {
      "epoch": 0.5408663107490781,
      "grad_norm": 0.7043294310569763,
      "learning_rate": 8.325136490954633e-06,
      "loss": 0.8421,
      "step": 9827
    },
    {
      "epoch": 0.5409213495514338,
      "grad_norm": 0.7947664856910706,
      "learning_rate": 8.32481275733271e-06,
      "loss": 0.8672,
      "step": 9828
    },
    {
      "epoch": 0.5409763883537895,
      "grad_norm": 0.704590916633606,
      "learning_rate": 8.324488998722384e-06,
      "loss": 0.7356,
      "step": 9829
    },
    {
      "epoch": 0.541031427156145,
      "grad_norm": 0.7630662322044373,
      "learning_rate": 8.32416521512609e-06,
      "loss": 0.7082,
      "step": 9830
    },
    {
      "epoch": 0.5410864659585007,
      "grad_norm": 0.728721022605896,
      "learning_rate": 8.323841406546259e-06,
      "loss": 0.7987,
      "step": 9831
    },
    {
      "epoch": 0.5411415047608564,
      "grad_norm": 0.7164294719696045,
      "learning_rate": 8.323517572985326e-06,
      "loss": 0.721,
      "step": 9832
    },
    {
      "epoch": 0.5411965435632121,
      "grad_norm": 0.7555723190307617,
      "learning_rate": 8.323193714445722e-06,
      "loss": 0.814,
      "step": 9833
    },
    {
      "epoch": 0.5412515823655677,
      "grad_norm": 0.827485978603363,
      "learning_rate": 8.322869830929887e-06,
      "loss": 0.8817,
      "step": 9834
    },
    {
      "epoch": 0.5413066211679234,
      "grad_norm": 0.718950092792511,
      "learning_rate": 8.322545922440252e-06,
      "loss": 0.8648,
      "step": 9835
    },
    {
      "epoch": 0.5413616599702791,
      "grad_norm": 0.7361611723899841,
      "learning_rate": 8.32222198897925e-06,
      "loss": 0.7392,
      "step": 9836
    },
    {
      "epoch": 0.5414166987726347,
      "grad_norm": 0.6712168455123901,
      "learning_rate": 8.321898030549316e-06,
      "loss": 0.7505,
      "step": 9837
    },
    {
      "epoch": 0.5414717375749903,
      "grad_norm": 0.7475710511207581,
      "learning_rate": 8.321574047152887e-06,
      "loss": 0.7969,
      "step": 9838
    },
    {
      "epoch": 0.541526776377346,
      "grad_norm": 0.9751361608505249,
      "learning_rate": 8.321250038792397e-06,
      "loss": 0.8534,
      "step": 9839
    },
    {
      "epoch": 0.5415818151797017,
      "grad_norm": 0.6858723163604736,
      "learning_rate": 8.32092600547028e-06,
      "loss": 0.8277,
      "step": 9840
    },
    {
      "epoch": 0.5416368539820573,
      "grad_norm": 0.8899725675582886,
      "learning_rate": 8.320601947188971e-06,
      "loss": 0.8599,
      "step": 9841
    },
    {
      "epoch": 0.541691892784413,
      "grad_norm": 0.7140665650367737,
      "learning_rate": 8.320277863950907e-06,
      "loss": 0.7429,
      "step": 9842
    },
    {
      "epoch": 0.5417469315867687,
      "grad_norm": 0.7467615604400635,
      "learning_rate": 8.319953755758525e-06,
      "loss": 0.7826,
      "step": 9843
    },
    {
      "epoch": 0.5418019703891244,
      "grad_norm": 0.6578202843666077,
      "learning_rate": 8.319629622614258e-06,
      "loss": 0.6833,
      "step": 9844
    },
    {
      "epoch": 0.5418570091914799,
      "grad_norm": 0.9430698156356812,
      "learning_rate": 8.319305464520543e-06,
      "loss": 0.8243,
      "step": 9845
    },
    {
      "epoch": 0.5419120479938356,
      "grad_norm": 0.8632097840309143,
      "learning_rate": 8.318981281479817e-06,
      "loss": 0.7975,
      "step": 9846
    },
    {
      "epoch": 0.5419670867961913,
      "grad_norm": 0.7241839170455933,
      "learning_rate": 8.318657073494517e-06,
      "loss": 0.7226,
      "step": 9847
    },
    {
      "epoch": 0.542022125598547,
      "grad_norm": 0.6927164196968079,
      "learning_rate": 8.318332840567078e-06,
      "loss": 0.7125,
      "step": 9848
    },
    {
      "epoch": 0.5420771644009026,
      "grad_norm": 0.6414939761161804,
      "learning_rate": 8.318008582699937e-06,
      "loss": 0.7366,
      "step": 9849
    },
    {
      "epoch": 0.5421322032032583,
      "grad_norm": 0.7584436535835266,
      "learning_rate": 8.317684299895533e-06,
      "loss": 0.8601,
      "step": 9850
    },
    {
      "epoch": 0.542187242005614,
      "grad_norm": 0.6045856475830078,
      "learning_rate": 8.317359992156302e-06,
      "loss": 0.6697,
      "step": 9851
    },
    {
      "epoch": 0.5422422808079697,
      "grad_norm": 0.715048611164093,
      "learning_rate": 8.31703565948468e-06,
      "loss": 0.7535,
      "step": 9852
    },
    {
      "epoch": 0.5422973196103252,
      "grad_norm": 0.6925113201141357,
      "learning_rate": 8.316711301883106e-06,
      "loss": 0.8122,
      "step": 9853
    },
    {
      "epoch": 0.5423523584126809,
      "grad_norm": 0.6787780523300171,
      "learning_rate": 8.316386919354018e-06,
      "loss": 0.7428,
      "step": 9854
    },
    {
      "epoch": 0.5424073972150366,
      "grad_norm": 0.6831366419792175,
      "learning_rate": 8.316062511899855e-06,
      "loss": 0.767,
      "step": 9855
    },
    {
      "epoch": 0.5424624360173923,
      "grad_norm": 0.6865691542625427,
      "learning_rate": 8.315738079523053e-06,
      "loss": 0.6549,
      "step": 9856
    },
    {
      "epoch": 0.5425174748197479,
      "grad_norm": 0.7149406671524048,
      "learning_rate": 8.31541362222605e-06,
      "loss": 0.8127,
      "step": 9857
    },
    {
      "epoch": 0.5425725136221036,
      "grad_norm": 0.6826779842376709,
      "learning_rate": 8.315089140011286e-06,
      "loss": 0.706,
      "step": 9858
    },
    {
      "epoch": 0.5426275524244593,
      "grad_norm": 0.688204288482666,
      "learning_rate": 8.3147646328812e-06,
      "loss": 0.8675,
      "step": 9859
    },
    {
      "epoch": 0.542682591226815,
      "grad_norm": 0.6659492254257202,
      "learning_rate": 8.31444010083823e-06,
      "loss": 0.7851,
      "step": 9860
    },
    {
      "epoch": 0.5427376300291705,
      "grad_norm": 0.8049291372299194,
      "learning_rate": 8.314115543884816e-06,
      "loss": 0.7442,
      "step": 9861
    },
    {
      "epoch": 0.5427926688315262,
      "grad_norm": 0.7505989670753479,
      "learning_rate": 8.313790962023397e-06,
      "loss": 0.8391,
      "step": 9862
    },
    {
      "epoch": 0.5428477076338819,
      "grad_norm": 0.6810199618339539,
      "learning_rate": 8.31346635525641e-06,
      "loss": 0.8131,
      "step": 9863
    },
    {
      "epoch": 0.5429027464362376,
      "grad_norm": 0.6724215745925903,
      "learning_rate": 8.313141723586298e-06,
      "loss": 0.75,
      "step": 9864
    },
    {
      "epoch": 0.5429577852385932,
      "grad_norm": 0.7804376482963562,
      "learning_rate": 8.3128170670155e-06,
      "loss": 0.704,
      "step": 9865
    },
    {
      "epoch": 0.5430128240409489,
      "grad_norm": 0.9494230151176453,
      "learning_rate": 8.312492385546455e-06,
      "loss": 0.8578,
      "step": 9866
    },
    {
      "epoch": 0.5430678628433045,
      "grad_norm": 0.6780333518981934,
      "learning_rate": 8.312167679181606e-06,
      "loss": 0.701,
      "step": 9867
    },
    {
      "epoch": 0.5431229016456602,
      "grad_norm": 0.7407701015472412,
      "learning_rate": 8.31184294792339e-06,
      "loss": 0.8505,
      "step": 9868
    },
    {
      "epoch": 0.5431779404480158,
      "grad_norm": 0.680903434753418,
      "learning_rate": 8.311518191774249e-06,
      "loss": 0.7645,
      "step": 9869
    },
    {
      "epoch": 0.5432329792503715,
      "grad_norm": 0.6695752143859863,
      "learning_rate": 8.311193410736622e-06,
      "loss": 0.816,
      "step": 9870
    },
    {
      "epoch": 0.5432880180527272,
      "grad_norm": 0.6725142598152161,
      "learning_rate": 8.310868604812954e-06,
      "loss": 0.7044,
      "step": 9871
    },
    {
      "epoch": 0.5433430568550829,
      "grad_norm": 0.922627866268158,
      "learning_rate": 8.310543774005684e-06,
      "loss": 0.7589,
      "step": 9872
    },
    {
      "epoch": 0.5433980956574385,
      "grad_norm": 1.0136839151382446,
      "learning_rate": 8.310218918317251e-06,
      "loss": 0.7573,
      "step": 9873
    },
    {
      "epoch": 0.5434531344597942,
      "grad_norm": 0.9053532481193542,
      "learning_rate": 8.309894037750099e-06,
      "loss": 0.8269,
      "step": 9874
    },
    {
      "epoch": 0.5435081732621498,
      "grad_norm": 0.6800149083137512,
      "learning_rate": 8.309569132306671e-06,
      "loss": 0.716,
      "step": 9875
    },
    {
      "epoch": 0.5435632120645055,
      "grad_norm": 0.7157679796218872,
      "learning_rate": 8.309244201989408e-06,
      "loss": 0.7433,
      "step": 9876
    },
    {
      "epoch": 0.5436182508668611,
      "grad_norm": 0.9316089749336243,
      "learning_rate": 8.308919246800748e-06,
      "loss": 0.7499,
      "step": 9877
    },
    {
      "epoch": 0.5436732896692168,
      "grad_norm": 0.6682490110397339,
      "learning_rate": 8.308594266743139e-06,
      "loss": 0.7286,
      "step": 9878
    },
    {
      "epoch": 0.5437283284715725,
      "grad_norm": 0.7241143584251404,
      "learning_rate": 8.308269261819022e-06,
      "loss": 0.7934,
      "step": 9879
    },
    {
      "epoch": 0.5437833672739281,
      "grad_norm": 0.7402396202087402,
      "learning_rate": 8.307944232030838e-06,
      "loss": 0.7361,
      "step": 9880
    },
    {
      "epoch": 0.5438384060762838,
      "grad_norm": 0.6839993596076965,
      "learning_rate": 8.307619177381029e-06,
      "loss": 0.749,
      "step": 9881
    },
    {
      "epoch": 0.5438934448786394,
      "grad_norm": 0.6536363363265991,
      "learning_rate": 8.307294097872041e-06,
      "loss": 0.706,
      "step": 9882
    },
    {
      "epoch": 0.5439484836809951,
      "grad_norm": 0.602644681930542,
      "learning_rate": 8.306968993506317e-06,
      "loss": 0.6857,
      "step": 9883
    },
    {
      "epoch": 0.5440035224833507,
      "grad_norm": 0.6567881107330322,
      "learning_rate": 8.306643864286297e-06,
      "loss": 0.6989,
      "step": 9884
    },
    {
      "epoch": 0.5440585612857064,
      "grad_norm": 1.0013506412506104,
      "learning_rate": 8.306318710214427e-06,
      "loss": 0.7251,
      "step": 9885
    },
    {
      "epoch": 0.5441136000880621,
      "grad_norm": 0.7016813158988953,
      "learning_rate": 8.305993531293153e-06,
      "loss": 0.7535,
      "step": 9886
    },
    {
      "epoch": 0.5441686388904178,
      "grad_norm": 0.7345741391181946,
      "learning_rate": 8.305668327524915e-06,
      "loss": 0.887,
      "step": 9887
    },
    {
      "epoch": 0.5442236776927734,
      "grad_norm": 1.0925754308700562,
      "learning_rate": 8.305343098912158e-06,
      "loss": 0.7779,
      "step": 9888
    },
    {
      "epoch": 0.544278716495129,
      "grad_norm": 0.79815274477005,
      "learning_rate": 8.305017845457328e-06,
      "loss": 0.7736,
      "step": 9889
    },
    {
      "epoch": 0.5443337552974847,
      "grad_norm": 0.6324154138565063,
      "learning_rate": 8.304692567162868e-06,
      "loss": 0.6823,
      "step": 9890
    },
    {
      "epoch": 0.5443887940998404,
      "grad_norm": 0.6990262866020203,
      "learning_rate": 8.304367264031223e-06,
      "loss": 0.7804,
      "step": 9891
    },
    {
      "epoch": 0.544443832902196,
      "grad_norm": 1.4203195571899414,
      "learning_rate": 8.304041936064839e-06,
      "loss": 0.8702,
      "step": 9892
    },
    {
      "epoch": 0.5444988717045517,
      "grad_norm": 0.6986544132232666,
      "learning_rate": 8.303716583266161e-06,
      "loss": 0.7666,
      "step": 9893
    },
    {
      "epoch": 0.5445539105069074,
      "grad_norm": 0.7037138938903809,
      "learning_rate": 8.303391205637632e-06,
      "loss": 0.7995,
      "step": 9894
    },
    {
      "epoch": 0.5446089493092631,
      "grad_norm": 0.7101728320121765,
      "learning_rate": 8.3030658031817e-06,
      "loss": 0.8185,
      "step": 9895
    },
    {
      "epoch": 0.5446639881116186,
      "grad_norm": 0.6571425795555115,
      "learning_rate": 8.302740375900808e-06,
      "loss": 0.6152,
      "step": 9896
    },
    {
      "epoch": 0.5447190269139743,
      "grad_norm": 0.7560263276100159,
      "learning_rate": 8.302414923797406e-06,
      "loss": 0.9037,
      "step": 9897
    },
    {
      "epoch": 0.54477406571633,
      "grad_norm": 0.8692007064819336,
      "learning_rate": 8.302089446873935e-06,
      "loss": 0.7689,
      "step": 9898
    },
    {
      "epoch": 0.5448291045186857,
      "grad_norm": 0.7533506751060486,
      "learning_rate": 8.301763945132845e-06,
      "loss": 0.7671,
      "step": 9899
    },
    {
      "epoch": 0.5448841433210413,
      "grad_norm": 0.6992233991622925,
      "learning_rate": 8.301438418576581e-06,
      "loss": 0.723,
      "step": 9900
    },
    {
      "epoch": 0.544939182123397,
      "grad_norm": 0.7966120839118958,
      "learning_rate": 8.301112867207589e-06,
      "loss": 0.7968,
      "step": 9901
    },
    {
      "epoch": 0.5449942209257527,
      "grad_norm": 0.800558865070343,
      "learning_rate": 8.300787291028316e-06,
      "loss": 0.8583,
      "step": 9902
    },
    {
      "epoch": 0.5450492597281084,
      "grad_norm": 0.7019909024238586,
      "learning_rate": 8.30046169004121e-06,
      "loss": 0.7045,
      "step": 9903
    },
    {
      "epoch": 0.5451042985304639,
      "grad_norm": 0.7778449654579163,
      "learning_rate": 8.300136064248717e-06,
      "loss": 0.7964,
      "step": 9904
    },
    {
      "epoch": 0.5451593373328196,
      "grad_norm": 0.6894309520721436,
      "learning_rate": 8.299810413653284e-06,
      "loss": 0.7382,
      "step": 9905
    },
    {
      "epoch": 0.5452143761351753,
      "grad_norm": 0.6942182183265686,
      "learning_rate": 8.299484738257361e-06,
      "loss": 0.73,
      "step": 9906
    },
    {
      "epoch": 0.545269414937531,
      "grad_norm": 0.6607787609100342,
      "learning_rate": 8.299159038063394e-06,
      "loss": 0.6987,
      "step": 9907
    },
    {
      "epoch": 0.5453244537398866,
      "grad_norm": 0.7447709441184998,
      "learning_rate": 8.29883331307383e-06,
      "loss": 0.7787,
      "step": 9908
    },
    {
      "epoch": 0.5453794925422423,
      "grad_norm": 0.6315301656723022,
      "learning_rate": 8.298507563291116e-06,
      "loss": 0.7047,
      "step": 9909
    },
    {
      "epoch": 0.545434531344598,
      "grad_norm": 0.8095656633377075,
      "learning_rate": 8.298181788717705e-06,
      "loss": 0.691,
      "step": 9910
    },
    {
      "epoch": 0.5454895701469537,
      "grad_norm": 0.6419453024864197,
      "learning_rate": 8.29785598935604e-06,
      "loss": 0.7333,
      "step": 9911
    },
    {
      "epoch": 0.5455446089493092,
      "grad_norm": 0.7209222316741943,
      "learning_rate": 8.297530165208574e-06,
      "loss": 0.8174,
      "step": 9912
    },
    {
      "epoch": 0.5455996477516649,
      "grad_norm": 0.6778598427772522,
      "learning_rate": 8.297204316277754e-06,
      "loss": 0.7696,
      "step": 9913
    },
    {
      "epoch": 0.5456546865540206,
      "grad_norm": 0.6573307514190674,
      "learning_rate": 8.296878442566028e-06,
      "loss": 0.7843,
      "step": 9914
    },
    {
      "epoch": 0.5457097253563763,
      "grad_norm": 0.6987473964691162,
      "learning_rate": 8.296552544075847e-06,
      "loss": 0.809,
      "step": 9915
    },
    {
      "epoch": 0.5457647641587319,
      "grad_norm": 0.7149204015731812,
      "learning_rate": 8.29622662080966e-06,
      "loss": 0.848,
      "step": 9916
    },
    {
      "epoch": 0.5458198029610876,
      "grad_norm": 0.6252632141113281,
      "learning_rate": 8.295900672769913e-06,
      "loss": 0.7029,
      "step": 9917
    },
    {
      "epoch": 0.5458748417634433,
      "grad_norm": 0.713376522064209,
      "learning_rate": 8.295574699959062e-06,
      "loss": 0.726,
      "step": 9918
    },
    {
      "epoch": 0.5459298805657989,
      "grad_norm": 0.6864717602729797,
      "learning_rate": 8.295248702379552e-06,
      "loss": 0.7428,
      "step": 9919
    },
    {
      "epoch": 0.5459849193681545,
      "grad_norm": 0.8085678219795227,
      "learning_rate": 8.294922680033837e-06,
      "loss": 0.8697,
      "step": 9920
    },
    {
      "epoch": 0.5460399581705102,
      "grad_norm": 0.7366700768470764,
      "learning_rate": 8.294596632924363e-06,
      "loss": 0.7714,
      "step": 9921
    },
    {
      "epoch": 0.5460949969728659,
      "grad_norm": 0.670632541179657,
      "learning_rate": 8.294270561053583e-06,
      "loss": 0.7032,
      "step": 9922
    },
    {
      "epoch": 0.5461500357752215,
      "grad_norm": 0.7867220640182495,
      "learning_rate": 8.293944464423946e-06,
      "loss": 0.8903,
      "step": 9923
    },
    {
      "epoch": 0.5462050745775772,
      "grad_norm": 0.8441565632820129,
      "learning_rate": 8.293618343037907e-06,
      "loss": 0.8694,
      "step": 9924
    },
    {
      "epoch": 0.5462601133799329,
      "grad_norm": 0.7048027515411377,
      "learning_rate": 8.293292196897913e-06,
      "loss": 0.8226,
      "step": 9925
    },
    {
      "epoch": 0.5463151521822885,
      "grad_norm": 0.6344078779220581,
      "learning_rate": 8.292966026006416e-06,
      "loss": 0.7615,
      "step": 9926
    },
    {
      "epoch": 0.5463701909846441,
      "grad_norm": 0.6744484901428223,
      "learning_rate": 8.292639830365867e-06,
      "loss": 0.6944,
      "step": 9927
    },
    {
      "epoch": 0.5464252297869998,
      "grad_norm": 0.8113303780555725,
      "learning_rate": 8.292313609978721e-06,
      "loss": 0.7558,
      "step": 9928
    },
    {
      "epoch": 0.5464802685893555,
      "grad_norm": 0.640190839767456,
      "learning_rate": 8.291987364847425e-06,
      "loss": 0.7167,
      "step": 9929
    },
    {
      "epoch": 0.5465353073917112,
      "grad_norm": 0.7714816331863403,
      "learning_rate": 8.291661094974434e-06,
      "loss": 0.8662,
      "step": 9930
    },
    {
      "epoch": 0.5465903461940668,
      "grad_norm": 0.6785402894020081,
      "learning_rate": 8.291334800362199e-06,
      "loss": 0.6835,
      "step": 9931
    },
    {
      "epoch": 0.5466453849964225,
      "grad_norm": 0.704868495464325,
      "learning_rate": 8.291008481013173e-06,
      "loss": 0.7343,
      "step": 9932
    },
    {
      "epoch": 0.5467004237987781,
      "grad_norm": 0.7587466239929199,
      "learning_rate": 8.290682136929809e-06,
      "loss": 0.7856,
      "step": 9933
    },
    {
      "epoch": 0.5467554626011338,
      "grad_norm": 0.7460505962371826,
      "learning_rate": 8.290355768114557e-06,
      "loss": 0.7463,
      "step": 9934
    },
    {
      "epoch": 0.5468105014034894,
      "grad_norm": 0.7185021042823792,
      "learning_rate": 8.290029374569873e-06,
      "loss": 0.8106,
      "step": 9935
    },
    {
      "epoch": 0.5468655402058451,
      "grad_norm": 0.7023874521255493,
      "learning_rate": 8.289702956298209e-06,
      "loss": 0.6863,
      "step": 9936
    },
    {
      "epoch": 0.5469205790082008,
      "grad_norm": 0.8688495755195618,
      "learning_rate": 8.289376513302017e-06,
      "loss": 0.8898,
      "step": 9937
    },
    {
      "epoch": 0.5469756178105565,
      "grad_norm": 0.6405122876167297,
      "learning_rate": 8.289050045583752e-06,
      "loss": 0.6804,
      "step": 9938
    },
    {
      "epoch": 0.5470306566129121,
      "grad_norm": 0.8364881277084351,
      "learning_rate": 8.288723553145868e-06,
      "loss": 0.8356,
      "step": 9939
    },
    {
      "epoch": 0.5470856954152677,
      "grad_norm": 0.6621617078781128,
      "learning_rate": 8.288397035990818e-06,
      "loss": 0.7508,
      "step": 9940
    },
    {
      "epoch": 0.5471407342176234,
      "grad_norm": 0.6822347640991211,
      "learning_rate": 8.288070494121056e-06,
      "loss": 0.7722,
      "step": 9941
    },
    {
      "epoch": 0.5471957730199791,
      "grad_norm": 0.6727223992347717,
      "learning_rate": 8.287743927539036e-06,
      "loss": 0.743,
      "step": 9942
    },
    {
      "epoch": 0.5472508118223347,
      "grad_norm": 0.7852441668510437,
      "learning_rate": 8.287417336247214e-06,
      "loss": 0.8321,
      "step": 9943
    },
    {
      "epoch": 0.5473058506246904,
      "grad_norm": 0.6982126235961914,
      "learning_rate": 8.287090720248041e-06,
      "loss": 0.6669,
      "step": 9944
    },
    {
      "epoch": 0.5473608894270461,
      "grad_norm": 0.7820166945457458,
      "learning_rate": 8.286764079543976e-06,
      "loss": 0.7592,
      "step": 9945
    },
    {
      "epoch": 0.5474159282294018,
      "grad_norm": 0.6868422627449036,
      "learning_rate": 8.28643741413747e-06,
      "loss": 0.8308,
      "step": 9946
    },
    {
      "epoch": 0.5474709670317573,
      "grad_norm": 0.8227942585945129,
      "learning_rate": 8.286110724030982e-06,
      "loss": 0.7982,
      "step": 9947
    },
    {
      "epoch": 0.547526005834113,
      "grad_norm": 0.6838171482086182,
      "learning_rate": 8.285784009226964e-06,
      "loss": 0.7907,
      "step": 9948
    },
    {
      "epoch": 0.5475810446364687,
      "grad_norm": 0.7200812697410583,
      "learning_rate": 8.285457269727875e-06,
      "loss": 0.88,
      "step": 9949
    },
    {
      "epoch": 0.5476360834388244,
      "grad_norm": 0.7469412684440613,
      "learning_rate": 8.285130505536168e-06,
      "loss": 0.8167,
      "step": 9950
    },
    {
      "epoch": 0.54769112224118,
      "grad_norm": 0.6660227179527283,
      "learning_rate": 8.284803716654298e-06,
      "loss": 0.7685,
      "step": 9951
    },
    {
      "epoch": 0.5477461610435357,
      "grad_norm": 0.7116572260856628,
      "learning_rate": 8.284476903084723e-06,
      "loss": 0.7415,
      "step": 9952
    },
    {
      "epoch": 0.5478011998458914,
      "grad_norm": 0.6540791988372803,
      "learning_rate": 8.284150064829899e-06,
      "loss": 0.6571,
      "step": 9953
    },
    {
      "epoch": 0.5478562386482471,
      "grad_norm": 0.7527759075164795,
      "learning_rate": 8.283823201892283e-06,
      "loss": 0.8678,
      "step": 9954
    },
    {
      "epoch": 0.5479112774506026,
      "grad_norm": 0.7795953750610352,
      "learning_rate": 8.283496314274331e-06,
      "loss": 0.8086,
      "step": 9955
    },
    {
      "epoch": 0.5479663162529583,
      "grad_norm": 0.862503170967102,
      "learning_rate": 8.283169401978498e-06,
      "loss": 0.7442,
      "step": 9956
    },
    {
      "epoch": 0.548021355055314,
      "grad_norm": 0.6552054286003113,
      "learning_rate": 8.282842465007244e-06,
      "loss": 0.6664,
      "step": 9957
    },
    {
      "epoch": 0.5480763938576697,
      "grad_norm": 0.7242427468299866,
      "learning_rate": 8.282515503363024e-06,
      "loss": 0.8199,
      "step": 9958
    },
    {
      "epoch": 0.5481314326600253,
      "grad_norm": 0.7529763579368591,
      "learning_rate": 8.282188517048295e-06,
      "loss": 0.761,
      "step": 9959
    },
    {
      "epoch": 0.548186471462381,
      "grad_norm": 0.7909425497055054,
      "learning_rate": 8.281861506065519e-06,
      "loss": 0.7389,
      "step": 9960
    },
    {
      "epoch": 0.5482415102647367,
      "grad_norm": 0.6594850420951843,
      "learning_rate": 8.281534470417147e-06,
      "loss": 0.7473,
      "step": 9961
    },
    {
      "epoch": 0.5482965490670924,
      "grad_norm": 0.6900844573974609,
      "learning_rate": 8.281207410105642e-06,
      "loss": 0.7551,
      "step": 9962
    },
    {
      "epoch": 0.5483515878694479,
      "grad_norm": 0.6922640204429626,
      "learning_rate": 8.28088032513346e-06,
      "loss": 0.7654,
      "step": 9963
    },
    {
      "epoch": 0.5484066266718036,
      "grad_norm": 0.7758432626724243,
      "learning_rate": 8.28055321550306e-06,
      "loss": 0.8033,
      "step": 9964
    },
    {
      "epoch": 0.5484616654741593,
      "grad_norm": 0.7074280977249146,
      "learning_rate": 8.2802260812169e-06,
      "loss": 0.7302,
      "step": 9965
    },
    {
      "epoch": 0.5485167042765149,
      "grad_norm": 0.7724928259849548,
      "learning_rate": 8.27989892227744e-06,
      "loss": 0.7621,
      "step": 9966
    },
    {
      "epoch": 0.5485717430788706,
      "grad_norm": 0.7364168167114258,
      "learning_rate": 8.279571738687137e-06,
      "loss": 0.7587,
      "step": 9967
    },
    {
      "epoch": 0.5486267818812263,
      "grad_norm": 0.7298350930213928,
      "learning_rate": 8.27924453044845e-06,
      "loss": 0.7371,
      "step": 9968
    },
    {
      "epoch": 0.548681820683582,
      "grad_norm": 0.8056737780570984,
      "learning_rate": 8.27891729756384e-06,
      "loss": 0.9871,
      "step": 9969
    },
    {
      "epoch": 0.5487368594859375,
      "grad_norm": 0.7499688267707825,
      "learning_rate": 8.278590040035763e-06,
      "loss": 0.8574,
      "step": 9970
    },
    {
      "epoch": 0.5487918982882932,
      "grad_norm": 0.7398175001144409,
      "learning_rate": 8.278262757866683e-06,
      "loss": 0.744,
      "step": 9971
    },
    {
      "epoch": 0.5488469370906489,
      "grad_norm": 0.7099171876907349,
      "learning_rate": 8.277935451059058e-06,
      "loss": 0.7108,
      "step": 9972
    },
    {
      "epoch": 0.5489019758930046,
      "grad_norm": 0.6720188856124878,
      "learning_rate": 8.277608119615345e-06,
      "loss": 0.8565,
      "step": 9973
    },
    {
      "epoch": 0.5489570146953602,
      "grad_norm": 0.7870737910270691,
      "learning_rate": 8.27728076353801e-06,
      "loss": 0.7429,
      "step": 9974
    },
    {
      "epoch": 0.5490120534977159,
      "grad_norm": 0.7358133792877197,
      "learning_rate": 8.276953382829507e-06,
      "loss": 0.7549,
      "step": 9975
    },
    {
      "epoch": 0.5490670923000716,
      "grad_norm": 0.8968467116355896,
      "learning_rate": 8.276625977492303e-06,
      "loss": 0.6983,
      "step": 9976
    },
    {
      "epoch": 0.5491221311024272,
      "grad_norm": 0.7346875071525574,
      "learning_rate": 8.276298547528852e-06,
      "loss": 0.8541,
      "step": 9977
    },
    {
      "epoch": 0.5491771699047828,
      "grad_norm": 0.7297229170799255,
      "learning_rate": 8.27597109294162e-06,
      "loss": 0.8378,
      "step": 9978
    },
    {
      "epoch": 0.5492322087071385,
      "grad_norm": 0.6907635927200317,
      "learning_rate": 8.275643613733064e-06,
      "loss": 0.7058,
      "step": 9979
    },
    {
      "epoch": 0.5492872475094942,
      "grad_norm": 0.7612239718437195,
      "learning_rate": 8.27531610990565e-06,
      "loss": 0.6827,
      "step": 9980
    },
    {
      "epoch": 0.5493422863118499,
      "grad_norm": 1.3160386085510254,
      "learning_rate": 8.274988581461837e-06,
      "loss": 0.7357,
      "step": 9981
    },
    {
      "epoch": 0.5493973251142055,
      "grad_norm": 0.6370541453361511,
      "learning_rate": 8.274661028404083e-06,
      "loss": 0.7323,
      "step": 9982
    },
    {
      "epoch": 0.5494523639165612,
      "grad_norm": 0.7051724195480347,
      "learning_rate": 8.274333450734856e-06,
      "loss": 0.7714,
      "step": 9983
    },
    {
      "epoch": 0.5495074027189168,
      "grad_norm": 0.7452969551086426,
      "learning_rate": 8.274005848456614e-06,
      "loss": 0.7516,
      "step": 9984
    },
    {
      "epoch": 0.5495624415212725,
      "grad_norm": 0.7132626175880432,
      "learning_rate": 8.273678221571823e-06,
      "loss": 0.6417,
      "step": 9985
    },
    {
      "epoch": 0.5496174803236281,
      "grad_norm": 0.7873446345329285,
      "learning_rate": 8.273350570082941e-06,
      "loss": 0.8457,
      "step": 9986
    },
    {
      "epoch": 0.5496725191259838,
      "grad_norm": 0.691470205783844,
      "learning_rate": 8.273022893992432e-06,
      "loss": 0.7871,
      "step": 9987
    },
    {
      "epoch": 0.5497275579283395,
      "grad_norm": 0.6671431064605713,
      "learning_rate": 8.27269519330276e-06,
      "loss": 0.6919,
      "step": 9988
    },
    {
      "epoch": 0.5497825967306952,
      "grad_norm": 0.8026914596557617,
      "learning_rate": 8.272367468016387e-06,
      "loss": 0.6885,
      "step": 9989
    },
    {
      "epoch": 0.5498376355330508,
      "grad_norm": 0.9003152251243591,
      "learning_rate": 8.272039718135774e-06,
      "loss": 0.7671,
      "step": 9990
    },
    {
      "epoch": 0.5498926743354065,
      "grad_norm": 0.6515254378318787,
      "learning_rate": 8.271711943663388e-06,
      "loss": 0.7589,
      "step": 9991
    },
    {
      "epoch": 0.5499477131377621,
      "grad_norm": 0.6495782136917114,
      "learning_rate": 8.27138414460169e-06,
      "loss": 0.7277,
      "step": 9992
    },
    {
      "epoch": 0.5500027519401178,
      "grad_norm": 0.7564565539360046,
      "learning_rate": 8.271056320953146e-06,
      "loss": 0.6977,
      "step": 9993
    },
    {
      "epoch": 0.5500577907424734,
      "grad_norm": 0.8551548719406128,
      "learning_rate": 8.270728472720218e-06,
      "loss": 0.684,
      "step": 9994
    },
    {
      "epoch": 0.5501128295448291,
      "grad_norm": 0.6614843010902405,
      "learning_rate": 8.270400599905369e-06,
      "loss": 0.6559,
      "step": 9995
    },
    {
      "epoch": 0.5501678683471848,
      "grad_norm": 0.6920068264007568,
      "learning_rate": 8.270072702511065e-06,
      "loss": 0.7497,
      "step": 9996
    },
    {
      "epoch": 0.5502229071495405,
      "grad_norm": 0.7426198124885559,
      "learning_rate": 8.26974478053977e-06,
      "loss": 0.7434,
      "step": 9997
    },
    {
      "epoch": 0.550277945951896,
      "grad_norm": 1.2630934715270996,
      "learning_rate": 8.269416833993949e-06,
      "loss": 0.7306,
      "step": 9998
    },
    {
      "epoch": 0.5503329847542517,
      "grad_norm": 0.7069457769393921,
      "learning_rate": 8.269088862876066e-06,
      "loss": 0.6735,
      "step": 9999
    }
  ],
  "logging_steps": 1,
  "max_steps": 36338,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 909,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.9507661855341937e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}