{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 0, "global_step": 3290, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00030395136778115504, "grad_norm": 0.392578125, "learning_rate": 9.99696048632219e-06, "loss": 1.7732, "step": 1 }, { "epoch": 0.0006079027355623101, "grad_norm": 0.41015625, "learning_rate": 9.993920972644378e-06, "loss": 1.7862, "step": 2 }, { "epoch": 0.000911854103343465, "grad_norm": 0.3828125, "learning_rate": 9.990881458966565e-06, "loss": 1.8057, "step": 3 }, { "epoch": 0.0012158054711246201, "grad_norm": 0.3828125, "learning_rate": 9.987841945288754e-06, "loss": 1.8368, "step": 4 }, { "epoch": 0.001519756838905775, "grad_norm": 0.365234375, "learning_rate": 9.984802431610943e-06, "loss": 1.7164, "step": 5 }, { "epoch": 0.00182370820668693, "grad_norm": 0.380859375, "learning_rate": 9.981762917933131e-06, "loss": 1.7691, "step": 6 }, { "epoch": 0.002127659574468085, "grad_norm": 0.333984375, "learning_rate": 9.97872340425532e-06, "loss": 1.7161, "step": 7 }, { "epoch": 0.0024316109422492403, "grad_norm": 0.302734375, "learning_rate": 9.975683890577509e-06, "loss": 1.7941, "step": 8 }, { "epoch": 0.002735562310030395, "grad_norm": 0.298828125, "learning_rate": 9.972644376899697e-06, "loss": 1.7409, "step": 9 }, { "epoch": 0.00303951367781155, "grad_norm": 0.29296875, "learning_rate": 9.969604863221886e-06, "loss": 1.761, "step": 10 }, { "epoch": 0.0033434650455927053, "grad_norm": 0.28515625, "learning_rate": 9.966565349544075e-06, "loss": 1.8032, "step": 11 }, { "epoch": 0.00364741641337386, "grad_norm": 0.2470703125, "learning_rate": 9.963525835866262e-06, "loss": 1.6851, "step": 12 }, { "epoch": 0.003951367781155016, "grad_norm": 0.234375, "learning_rate": 9.96048632218845e-06, "loss": 1.6766, "step": 13 }, { "epoch": 0.00425531914893617, "grad_norm": 0.2421875, "learning_rate": 9.957446808510639e-06, "loss": 1.6984, "step": 14 }, { "epoch": 0.004559270516717325, "grad_norm": 0.248046875, "learning_rate": 9.954407294832828e-06, "loss": 1.6502, "step": 15 }, { "epoch": 0.004863221884498481, "grad_norm": 0.2353515625, "learning_rate": 9.951367781155016e-06, "loss": 1.6413, "step": 16 }, { "epoch": 0.005167173252279635, "grad_norm": 0.2275390625, "learning_rate": 9.948328267477203e-06, "loss": 1.6161, "step": 17 }, { "epoch": 0.00547112462006079, "grad_norm": 0.228515625, "learning_rate": 9.945288753799392e-06, "loss": 1.5899, "step": 18 }, { "epoch": 0.005775075987841946, "grad_norm": 0.212890625, "learning_rate": 9.94224924012158e-06, "loss": 1.6689, "step": 19 }, { "epoch": 0.0060790273556231, "grad_norm": 0.2099609375, "learning_rate": 9.939209726443771e-06, "loss": 1.6254, "step": 20 }, { "epoch": 0.006382978723404255, "grad_norm": 0.212890625, "learning_rate": 9.936170212765958e-06, "loss": 1.6068, "step": 21 }, { "epoch": 0.006686930091185411, "grad_norm": 0.19921875, "learning_rate": 9.933130699088147e-06, "loss": 1.6764, "step": 22 }, { "epoch": 0.006990881458966565, "grad_norm": 0.19140625, "learning_rate": 9.930091185410335e-06, "loss": 1.6416, "step": 23 }, { "epoch": 0.00729483282674772, "grad_norm": 0.1982421875, "learning_rate": 9.927051671732524e-06, "loss": 1.6309, "step": 24 }, { "epoch": 0.007598784194528876, "grad_norm": 0.1923828125, "learning_rate": 9.924012158054713e-06, "loss": 1.5178, "step": 25 }, { "epoch": 0.007902735562310031, "grad_norm": 0.18359375, "learning_rate": 9.9209726443769e-06, "loss": 1.5448, "step": 26 }, { "epoch": 0.008206686930091186, "grad_norm": 0.17578125, "learning_rate": 9.917933130699089e-06, "loss": 1.6313, "step": 27 }, { "epoch": 0.00851063829787234, "grad_norm": 0.1806640625, "learning_rate": 9.914893617021277e-06, "loss": 1.5107, "step": 28 }, { "epoch": 0.008814589665653495, "grad_norm": 0.17578125, "learning_rate": 9.911854103343466e-06, "loss": 1.5642, "step": 29 }, { "epoch": 0.00911854103343465, "grad_norm": 0.1904296875, "learning_rate": 9.908814589665655e-06, "loss": 1.559, "step": 30 }, { "epoch": 0.009422492401215805, "grad_norm": 0.1630859375, "learning_rate": 9.905775075987842e-06, "loss": 1.5433, "step": 31 }, { "epoch": 0.009726443768996961, "grad_norm": 0.1650390625, "learning_rate": 9.90273556231003e-06, "loss": 1.5423, "step": 32 }, { "epoch": 0.010030395136778116, "grad_norm": 0.169921875, "learning_rate": 9.899696048632219e-06, "loss": 1.6182, "step": 33 }, { "epoch": 0.01033434650455927, "grad_norm": 0.1552734375, "learning_rate": 9.89665653495441e-06, "loss": 1.6109, "step": 34 }, { "epoch": 0.010638297872340425, "grad_norm": 0.162109375, "learning_rate": 9.893617021276596e-06, "loss": 1.5072, "step": 35 }, { "epoch": 0.01094224924012158, "grad_norm": 0.158203125, "learning_rate": 9.890577507598785e-06, "loss": 1.4997, "step": 36 }, { "epoch": 0.011246200607902736, "grad_norm": 0.1591796875, "learning_rate": 9.887537993920974e-06, "loss": 1.4627, "step": 37 }, { "epoch": 0.011550151975683891, "grad_norm": 0.154296875, "learning_rate": 9.884498480243162e-06, "loss": 1.5406, "step": 38 }, { "epoch": 0.011854103343465046, "grad_norm": 0.1572265625, "learning_rate": 9.881458966565351e-06, "loss": 1.56, "step": 39 }, { "epoch": 0.0121580547112462, "grad_norm": 0.1484375, "learning_rate": 9.878419452887538e-06, "loss": 1.4746, "step": 40 }, { "epoch": 0.012462006079027355, "grad_norm": 0.1572265625, "learning_rate": 9.875379939209727e-06, "loss": 1.5389, "step": 41 }, { "epoch": 0.01276595744680851, "grad_norm": 0.1708984375, "learning_rate": 9.872340425531915e-06, "loss": 1.5061, "step": 42 }, { "epoch": 0.013069908814589666, "grad_norm": 0.1552734375, "learning_rate": 9.869300911854104e-06, "loss": 1.4602, "step": 43 }, { "epoch": 0.013373860182370821, "grad_norm": 0.150390625, "learning_rate": 9.866261398176293e-06, "loss": 1.5035, "step": 44 }, { "epoch": 0.013677811550151976, "grad_norm": 0.1513671875, "learning_rate": 9.86322188449848e-06, "loss": 1.5337, "step": 45 }, { "epoch": 0.01398176291793313, "grad_norm": 0.1552734375, "learning_rate": 9.860182370820668e-06, "loss": 1.5305, "step": 46 }, { "epoch": 0.014285714285714285, "grad_norm": 0.1416015625, "learning_rate": 9.857142857142859e-06, "loss": 1.4701, "step": 47 }, { "epoch": 0.01458966565349544, "grad_norm": 0.142578125, "learning_rate": 9.854103343465048e-06, "loss": 1.485, "step": 48 }, { "epoch": 0.014893617021276596, "grad_norm": 0.134765625, "learning_rate": 9.851063829787235e-06, "loss": 1.4775, "step": 49 }, { "epoch": 0.015197568389057751, "grad_norm": 0.1494140625, "learning_rate": 9.848024316109423e-06, "loss": 1.3473, "step": 50 }, { "epoch": 0.015501519756838906, "grad_norm": 0.140625, "learning_rate": 9.844984802431612e-06, "loss": 1.4807, "step": 51 }, { "epoch": 0.015805471124620062, "grad_norm": 0.1416015625, "learning_rate": 9.8419452887538e-06, "loss": 1.4614, "step": 52 }, { "epoch": 0.016109422492401215, "grad_norm": 0.13671875, "learning_rate": 9.83890577507599e-06, "loss": 1.4871, "step": 53 }, { "epoch": 0.01641337386018237, "grad_norm": 0.130859375, "learning_rate": 9.835866261398176e-06, "loss": 1.5185, "step": 54 }, { "epoch": 0.016717325227963525, "grad_norm": 0.134765625, "learning_rate": 9.832826747720365e-06, "loss": 1.4098, "step": 55 }, { "epoch": 0.01702127659574468, "grad_norm": 0.1357421875, "learning_rate": 9.829787234042554e-06, "loss": 1.491, "step": 56 }, { "epoch": 0.017325227963525838, "grad_norm": 0.1396484375, "learning_rate": 9.826747720364742e-06, "loss": 1.3946, "step": 57 }, { "epoch": 0.01762917933130699, "grad_norm": 0.140625, "learning_rate": 9.823708206686931e-06, "loss": 1.4475, "step": 58 }, { "epoch": 0.017933130699088147, "grad_norm": 0.1328125, "learning_rate": 9.82066869300912e-06, "loss": 1.4315, "step": 59 }, { "epoch": 0.0182370820668693, "grad_norm": 0.1611328125, "learning_rate": 9.817629179331308e-06, "loss": 1.4593, "step": 60 }, { "epoch": 0.018541033434650456, "grad_norm": 0.142578125, "learning_rate": 9.814589665653497e-06, "loss": 1.3794, "step": 61 }, { "epoch": 0.01884498480243161, "grad_norm": 0.1416015625, "learning_rate": 9.811550151975686e-06, "loss": 1.4753, "step": 62 }, { "epoch": 0.019148936170212766, "grad_norm": 0.130859375, "learning_rate": 9.808510638297873e-06, "loss": 1.3962, "step": 63 }, { "epoch": 0.019452887537993922, "grad_norm": 0.13671875, "learning_rate": 9.805471124620061e-06, "loss": 1.4661, "step": 64 }, { "epoch": 0.019756838905775075, "grad_norm": 0.1416015625, "learning_rate": 9.80243161094225e-06, "loss": 1.4993, "step": 65 }, { "epoch": 0.02006079027355623, "grad_norm": 0.134765625, "learning_rate": 9.799392097264439e-06, "loss": 1.4878, "step": 66 }, { "epoch": 0.020364741641337385, "grad_norm": 0.1337890625, "learning_rate": 9.796352583586627e-06, "loss": 1.4965, "step": 67 }, { "epoch": 0.02066869300911854, "grad_norm": 0.1328125, "learning_rate": 9.793313069908814e-06, "loss": 1.4214, "step": 68 }, { "epoch": 0.020972644376899698, "grad_norm": 0.138671875, "learning_rate": 9.790273556231003e-06, "loss": 1.4793, "step": 69 }, { "epoch": 0.02127659574468085, "grad_norm": 0.1318359375, "learning_rate": 9.787234042553192e-06, "loss": 1.4492, "step": 70 }, { "epoch": 0.021580547112462007, "grad_norm": 0.14453125, "learning_rate": 9.78419452887538e-06, "loss": 1.4017, "step": 71 }, { "epoch": 0.02188449848024316, "grad_norm": 0.1455078125, "learning_rate": 9.78115501519757e-06, "loss": 1.386, "step": 72 }, { "epoch": 0.022188449848024316, "grad_norm": 0.2578125, "learning_rate": 9.778115501519758e-06, "loss": 1.3891, "step": 73 }, { "epoch": 0.022492401215805473, "grad_norm": 0.140625, "learning_rate": 9.775075987841947e-06, "loss": 1.4128, "step": 74 }, { "epoch": 0.022796352583586626, "grad_norm": 0.1376953125, "learning_rate": 9.772036474164135e-06, "loss": 1.395, "step": 75 }, { "epoch": 0.023100303951367782, "grad_norm": 0.1318359375, "learning_rate": 9.768996960486324e-06, "loss": 1.4429, "step": 76 }, { "epoch": 0.023404255319148935, "grad_norm": 0.138671875, "learning_rate": 9.765957446808511e-06, "loss": 1.4807, "step": 77 }, { "epoch": 0.02370820668693009, "grad_norm": 0.1357421875, "learning_rate": 9.7629179331307e-06, "loss": 1.3831, "step": 78 }, { "epoch": 0.024012158054711245, "grad_norm": 0.169921875, "learning_rate": 9.759878419452888e-06, "loss": 1.4234, "step": 79 }, { "epoch": 0.0243161094224924, "grad_norm": 0.1357421875, "learning_rate": 9.756838905775077e-06, "loss": 1.3879, "step": 80 }, { "epoch": 0.024620060790273558, "grad_norm": 0.12890625, "learning_rate": 9.753799392097266e-06, "loss": 1.369, "step": 81 }, { "epoch": 0.02492401215805471, "grad_norm": 0.154296875, "learning_rate": 9.750759878419453e-06, "loss": 1.4923, "step": 82 }, { "epoch": 0.025227963525835867, "grad_norm": 0.138671875, "learning_rate": 9.747720364741641e-06, "loss": 1.3184, "step": 83 }, { "epoch": 0.02553191489361702, "grad_norm": 0.1513671875, "learning_rate": 9.74468085106383e-06, "loss": 1.3347, "step": 84 }, { "epoch": 0.025835866261398176, "grad_norm": 0.1416015625, "learning_rate": 9.741641337386019e-06, "loss": 1.3469, "step": 85 }, { "epoch": 0.026139817629179333, "grad_norm": 0.1787109375, "learning_rate": 9.738601823708207e-06, "loss": 1.5045, "step": 86 }, { "epoch": 0.026443768996960486, "grad_norm": 0.162109375, "learning_rate": 9.735562310030396e-06, "loss": 1.3845, "step": 87 }, { "epoch": 0.026747720364741642, "grad_norm": 0.15625, "learning_rate": 9.732522796352585e-06, "loss": 1.3668, "step": 88 }, { "epoch": 0.027051671732522795, "grad_norm": 0.14453125, "learning_rate": 9.729483282674773e-06, "loss": 1.4213, "step": 89 }, { "epoch": 0.02735562310030395, "grad_norm": 0.140625, "learning_rate": 9.726443768996962e-06, "loss": 1.4503, "step": 90 }, { "epoch": 0.027659574468085105, "grad_norm": 0.1572265625, "learning_rate": 9.723404255319149e-06, "loss": 1.2939, "step": 91 }, { "epoch": 0.02796352583586626, "grad_norm": 0.1435546875, "learning_rate": 9.720364741641338e-06, "loss": 1.3585, "step": 92 }, { "epoch": 0.028267477203647418, "grad_norm": 0.140625, "learning_rate": 9.717325227963526e-06, "loss": 1.3273, "step": 93 }, { "epoch": 0.02857142857142857, "grad_norm": 0.1435546875, "learning_rate": 9.714285714285715e-06, "loss": 1.2947, "step": 94 }, { "epoch": 0.028875379939209727, "grad_norm": 0.142578125, "learning_rate": 9.711246200607904e-06, "loss": 1.4172, "step": 95 }, { "epoch": 0.02917933130699088, "grad_norm": 0.1328125, "learning_rate": 9.708206686930093e-06, "loss": 1.3992, "step": 96 }, { "epoch": 0.029483282674772036, "grad_norm": 0.142578125, "learning_rate": 9.70516717325228e-06, "loss": 1.4069, "step": 97 }, { "epoch": 0.029787234042553193, "grad_norm": 0.158203125, "learning_rate": 9.702127659574468e-06, "loss": 1.3173, "step": 98 }, { "epoch": 0.030091185410334346, "grad_norm": 0.169921875, "learning_rate": 9.699088145896657e-06, "loss": 1.3319, "step": 99 }, { "epoch": 0.030395136778115502, "grad_norm": 0.15234375, "learning_rate": 9.696048632218846e-06, "loss": 1.417, "step": 100 }, { "epoch": 0.030699088145896655, "grad_norm": 0.1396484375, "learning_rate": 9.693009118541034e-06, "loss": 1.3696, "step": 101 }, { "epoch": 0.03100303951367781, "grad_norm": 0.13671875, "learning_rate": 9.689969604863223e-06, "loss": 1.3691, "step": 102 }, { "epoch": 0.03130699088145897, "grad_norm": 0.1513671875, "learning_rate": 9.686930091185412e-06, "loss": 1.3581, "step": 103 }, { "epoch": 0.031610942249240125, "grad_norm": 0.1416015625, "learning_rate": 9.6838905775076e-06, "loss": 1.3476, "step": 104 }, { "epoch": 0.031914893617021274, "grad_norm": 0.15625, "learning_rate": 9.680851063829787e-06, "loss": 1.4596, "step": 105 }, { "epoch": 0.03221884498480243, "grad_norm": 0.1474609375, "learning_rate": 9.677811550151976e-06, "loss": 1.3698, "step": 106 }, { "epoch": 0.03252279635258359, "grad_norm": 0.1337890625, "learning_rate": 9.674772036474165e-06, "loss": 1.3268, "step": 107 }, { "epoch": 0.03282674772036474, "grad_norm": 0.140625, "learning_rate": 9.671732522796353e-06, "loss": 1.3606, "step": 108 }, { "epoch": 0.0331306990881459, "grad_norm": 0.1748046875, "learning_rate": 9.668693009118542e-06, "loss": 1.3282, "step": 109 }, { "epoch": 0.03343465045592705, "grad_norm": 0.1416015625, "learning_rate": 9.66565349544073e-06, "loss": 1.3786, "step": 110 }, { "epoch": 0.033738601823708206, "grad_norm": 0.1416015625, "learning_rate": 9.662613981762918e-06, "loss": 1.3186, "step": 111 }, { "epoch": 0.03404255319148936, "grad_norm": 0.1455078125, "learning_rate": 9.659574468085106e-06, "loss": 1.3575, "step": 112 }, { "epoch": 0.03434650455927052, "grad_norm": 0.1396484375, "learning_rate": 9.656534954407297e-06, "loss": 1.4137, "step": 113 }, { "epoch": 0.034650455927051675, "grad_norm": 0.14453125, "learning_rate": 9.653495440729484e-06, "loss": 1.3537, "step": 114 }, { "epoch": 0.034954407294832825, "grad_norm": 0.1533203125, "learning_rate": 9.650455927051672e-06, "loss": 1.3927, "step": 115 }, { "epoch": 0.03525835866261398, "grad_norm": 0.1484375, "learning_rate": 9.647416413373861e-06, "loss": 1.3414, "step": 116 }, { "epoch": 0.03556231003039514, "grad_norm": 0.1435546875, "learning_rate": 9.64437689969605e-06, "loss": 1.3332, "step": 117 }, { "epoch": 0.035866261398176294, "grad_norm": 0.1962890625, "learning_rate": 9.641337386018239e-06, "loss": 1.3668, "step": 118 }, { "epoch": 0.036170212765957444, "grad_norm": 0.1396484375, "learning_rate": 9.638297872340426e-06, "loss": 1.3197, "step": 119 }, { "epoch": 0.0364741641337386, "grad_norm": 0.17578125, "learning_rate": 9.635258358662614e-06, "loss": 1.3218, "step": 120 }, { "epoch": 0.036778115501519756, "grad_norm": 0.1669921875, "learning_rate": 9.632218844984803e-06, "loss": 1.3143, "step": 121 }, { "epoch": 0.03708206686930091, "grad_norm": 0.15234375, "learning_rate": 9.629179331306992e-06, "loss": 1.2176, "step": 122 }, { "epoch": 0.03738601823708207, "grad_norm": 0.1552734375, "learning_rate": 9.62613981762918e-06, "loss": 1.3651, "step": 123 }, { "epoch": 0.03768996960486322, "grad_norm": 0.150390625, "learning_rate": 9.623100303951369e-06, "loss": 1.3621, "step": 124 }, { "epoch": 0.037993920972644375, "grad_norm": 0.1630859375, "learning_rate": 9.620060790273556e-06, "loss": 1.3325, "step": 125 }, { "epoch": 0.03829787234042553, "grad_norm": 0.1552734375, "learning_rate": 9.617021276595745e-06, "loss": 1.2468, "step": 126 }, { "epoch": 0.03860182370820669, "grad_norm": 0.1572265625, "learning_rate": 9.613981762917935e-06, "loss": 1.331, "step": 127 }, { "epoch": 0.038905775075987845, "grad_norm": 0.1884765625, "learning_rate": 9.610942249240122e-06, "loss": 1.2179, "step": 128 }, { "epoch": 0.039209726443768994, "grad_norm": 0.171875, "learning_rate": 9.60790273556231e-06, "loss": 1.2703, "step": 129 }, { "epoch": 0.03951367781155015, "grad_norm": 0.1494140625, "learning_rate": 9.6048632218845e-06, "loss": 1.3175, "step": 130 }, { "epoch": 0.03981762917933131, "grad_norm": 0.1689453125, "learning_rate": 9.601823708206688e-06, "loss": 1.2629, "step": 131 }, { "epoch": 0.04012158054711246, "grad_norm": 0.1806640625, "learning_rate": 9.598784194528877e-06, "loss": 1.2874, "step": 132 }, { "epoch": 0.04042553191489362, "grad_norm": 0.1611328125, "learning_rate": 9.595744680851065e-06, "loss": 1.2644, "step": 133 }, { "epoch": 0.04072948328267477, "grad_norm": 0.1611328125, "learning_rate": 9.592705167173252e-06, "loss": 1.399, "step": 134 }, { "epoch": 0.041033434650455926, "grad_norm": 0.17578125, "learning_rate": 9.589665653495441e-06, "loss": 1.3488, "step": 135 }, { "epoch": 0.04133738601823708, "grad_norm": 0.171875, "learning_rate": 9.58662613981763e-06, "loss": 1.3002, "step": 136 }, { "epoch": 0.04164133738601824, "grad_norm": 0.16015625, "learning_rate": 9.583586626139818e-06, "loss": 1.2279, "step": 137 }, { "epoch": 0.041945288753799395, "grad_norm": 0.158203125, "learning_rate": 9.580547112462007e-06, "loss": 1.2976, "step": 138 }, { "epoch": 0.042249240121580545, "grad_norm": 0.1845703125, "learning_rate": 9.577507598784194e-06, "loss": 1.2442, "step": 139 }, { "epoch": 0.0425531914893617, "grad_norm": 0.1572265625, "learning_rate": 9.574468085106385e-06, "loss": 1.2743, "step": 140 }, { "epoch": 0.04285714285714286, "grad_norm": 0.1611328125, "learning_rate": 9.571428571428573e-06, "loss": 1.3167, "step": 141 }, { "epoch": 0.043161094224924014, "grad_norm": 0.177734375, "learning_rate": 9.56838905775076e-06, "loss": 1.3418, "step": 142 }, { "epoch": 0.04346504559270517, "grad_norm": 0.1748046875, "learning_rate": 9.565349544072949e-06, "loss": 1.2118, "step": 143 }, { "epoch": 0.04376899696048632, "grad_norm": 0.16015625, "learning_rate": 9.562310030395138e-06, "loss": 1.2595, "step": 144 }, { "epoch": 0.044072948328267476, "grad_norm": 0.2294921875, "learning_rate": 9.559270516717326e-06, "loss": 1.259, "step": 145 }, { "epoch": 0.04437689969604863, "grad_norm": 0.1552734375, "learning_rate": 9.556231003039515e-06, "loss": 1.2891, "step": 146 }, { "epoch": 0.04468085106382979, "grad_norm": 0.1689453125, "learning_rate": 9.553191489361704e-06, "loss": 1.1982, "step": 147 }, { "epoch": 0.044984802431610946, "grad_norm": 0.1640625, "learning_rate": 9.55015197568389e-06, "loss": 1.2906, "step": 148 }, { "epoch": 0.045288753799392095, "grad_norm": 0.193359375, "learning_rate": 9.54711246200608e-06, "loss": 1.4037, "step": 149 }, { "epoch": 0.04559270516717325, "grad_norm": 0.1591796875, "learning_rate": 9.544072948328268e-06, "loss": 1.2413, "step": 150 }, { "epoch": 0.04589665653495441, "grad_norm": 0.158203125, "learning_rate": 9.541033434650457e-06, "loss": 1.2876, "step": 151 }, { "epoch": 0.046200607902735565, "grad_norm": 0.171875, "learning_rate": 9.537993920972645e-06, "loss": 1.2657, "step": 152 }, { "epoch": 0.046504559270516714, "grad_norm": 0.1591796875, "learning_rate": 9.534954407294834e-06, "loss": 1.2802, "step": 153 }, { "epoch": 0.04680851063829787, "grad_norm": 0.162109375, "learning_rate": 9.531914893617023e-06, "loss": 1.3168, "step": 154 }, { "epoch": 0.04711246200607903, "grad_norm": 0.33203125, "learning_rate": 9.528875379939211e-06, "loss": 1.2118, "step": 155 }, { "epoch": 0.04741641337386018, "grad_norm": 0.251953125, "learning_rate": 9.525835866261398e-06, "loss": 1.2677, "step": 156 }, { "epoch": 0.04772036474164134, "grad_norm": 0.185546875, "learning_rate": 9.522796352583587e-06, "loss": 1.2617, "step": 157 }, { "epoch": 0.04802431610942249, "grad_norm": 0.1591796875, "learning_rate": 9.519756838905776e-06, "loss": 1.2789, "step": 158 }, { "epoch": 0.048328267477203646, "grad_norm": 0.1611328125, "learning_rate": 9.516717325227964e-06, "loss": 1.2337, "step": 159 }, { "epoch": 0.0486322188449848, "grad_norm": 0.1826171875, "learning_rate": 9.513677811550153e-06, "loss": 1.2375, "step": 160 }, { "epoch": 0.04893617021276596, "grad_norm": 0.1796875, "learning_rate": 9.510638297872342e-06, "loss": 1.2442, "step": 161 }, { "epoch": 0.049240121580547115, "grad_norm": 0.1748046875, "learning_rate": 9.507598784194529e-06, "loss": 1.256, "step": 162 }, { "epoch": 0.049544072948328265, "grad_norm": 0.1943359375, "learning_rate": 9.504559270516717e-06, "loss": 1.2923, "step": 163 }, { "epoch": 0.04984802431610942, "grad_norm": 0.1630859375, "learning_rate": 9.501519756838906e-06, "loss": 1.299, "step": 164 }, { "epoch": 0.05015197568389058, "grad_norm": 0.16796875, "learning_rate": 9.498480243161095e-06, "loss": 1.2387, "step": 165 }, { "epoch": 0.050455927051671734, "grad_norm": 0.1689453125, "learning_rate": 9.495440729483284e-06, "loss": 1.2756, "step": 166 }, { "epoch": 0.05075987841945289, "grad_norm": 0.171875, "learning_rate": 9.492401215805472e-06, "loss": 1.2914, "step": 167 }, { "epoch": 0.05106382978723404, "grad_norm": 0.1806640625, "learning_rate": 9.489361702127661e-06, "loss": 1.2922, "step": 168 }, { "epoch": 0.051367781155015196, "grad_norm": 0.173828125, "learning_rate": 9.48632218844985e-06, "loss": 1.311, "step": 169 }, { "epoch": 0.05167173252279635, "grad_norm": 0.193359375, "learning_rate": 9.483282674772038e-06, "loss": 1.1823, "step": 170 }, { "epoch": 0.05197568389057751, "grad_norm": 0.1748046875, "learning_rate": 9.480243161094225e-06, "loss": 1.2408, "step": 171 }, { "epoch": 0.052279635258358666, "grad_norm": 0.1806640625, "learning_rate": 9.477203647416414e-06, "loss": 1.2402, "step": 172 }, { "epoch": 0.052583586626139815, "grad_norm": 0.2236328125, "learning_rate": 9.474164133738603e-06, "loss": 1.3589, "step": 173 }, { "epoch": 0.05288753799392097, "grad_norm": 0.166015625, "learning_rate": 9.471124620060791e-06, "loss": 1.1678, "step": 174 }, { "epoch": 0.05319148936170213, "grad_norm": 0.18359375, "learning_rate": 9.46808510638298e-06, "loss": 1.2512, "step": 175 }, { "epoch": 0.053495440729483285, "grad_norm": 0.1806640625, "learning_rate": 9.465045592705167e-06, "loss": 1.1461, "step": 176 }, { "epoch": 0.05379939209726444, "grad_norm": 0.169921875, "learning_rate": 9.462006079027356e-06, "loss": 1.2284, "step": 177 }, { "epoch": 0.05410334346504559, "grad_norm": 0.173828125, "learning_rate": 9.458966565349544e-06, "loss": 1.2766, "step": 178 }, { "epoch": 0.05440729483282675, "grad_norm": 0.1826171875, "learning_rate": 9.455927051671733e-06, "loss": 1.3468, "step": 179 }, { "epoch": 0.0547112462006079, "grad_norm": 0.169921875, "learning_rate": 9.452887537993922e-06, "loss": 1.1995, "step": 180 }, { "epoch": 0.05501519756838906, "grad_norm": 0.185546875, "learning_rate": 9.44984802431611e-06, "loss": 1.2376, "step": 181 }, { "epoch": 0.05531914893617021, "grad_norm": 0.189453125, "learning_rate": 9.446808510638299e-06, "loss": 1.263, "step": 182 }, { "epoch": 0.055623100303951366, "grad_norm": 0.1884765625, "learning_rate": 9.443768996960488e-06, "loss": 1.1698, "step": 183 }, { "epoch": 0.05592705167173252, "grad_norm": 0.1884765625, "learning_rate": 9.440729483282676e-06, "loss": 1.2098, "step": 184 }, { "epoch": 0.05623100303951368, "grad_norm": 0.19140625, "learning_rate": 9.437689969604863e-06, "loss": 1.2565, "step": 185 }, { "epoch": 0.056534954407294835, "grad_norm": 0.203125, "learning_rate": 9.434650455927052e-06, "loss": 1.1391, "step": 186 }, { "epoch": 0.056838905775075985, "grad_norm": 0.181640625, "learning_rate": 9.431610942249241e-06, "loss": 1.2338, "step": 187 }, { "epoch": 0.05714285714285714, "grad_norm": 0.177734375, "learning_rate": 9.42857142857143e-06, "loss": 1.1856, "step": 188 }, { "epoch": 0.0574468085106383, "grad_norm": 0.1884765625, "learning_rate": 9.425531914893618e-06, "loss": 1.2162, "step": 189 }, { "epoch": 0.057750759878419454, "grad_norm": 0.1923828125, "learning_rate": 9.422492401215805e-06, "loss": 1.1983, "step": 190 }, { "epoch": 0.05805471124620061, "grad_norm": 0.1904296875, "learning_rate": 9.419452887537994e-06, "loss": 1.2694, "step": 191 }, { "epoch": 0.05835866261398176, "grad_norm": 0.1748046875, "learning_rate": 9.416413373860183e-06, "loss": 1.1764, "step": 192 }, { "epoch": 0.058662613981762916, "grad_norm": 0.171875, "learning_rate": 9.413373860182371e-06, "loss": 1.2191, "step": 193 }, { "epoch": 0.05896656534954407, "grad_norm": 0.181640625, "learning_rate": 9.41033434650456e-06, "loss": 1.2479, "step": 194 }, { "epoch": 0.05927051671732523, "grad_norm": 0.19921875, "learning_rate": 9.407294832826749e-06, "loss": 1.2135, "step": 195 }, { "epoch": 0.059574468085106386, "grad_norm": 0.1865234375, "learning_rate": 9.404255319148937e-06, "loss": 1.248, "step": 196 }, { "epoch": 0.059878419452887535, "grad_norm": 0.2119140625, "learning_rate": 9.401215805471126e-06, "loss": 1.1747, "step": 197 }, { "epoch": 0.06018237082066869, "grad_norm": 0.193359375, "learning_rate": 9.398176291793315e-06, "loss": 1.2408, "step": 198 }, { "epoch": 0.06048632218844985, "grad_norm": 0.177734375, "learning_rate": 9.395136778115502e-06, "loss": 1.1828, "step": 199 }, { "epoch": 0.060790273556231005, "grad_norm": 0.228515625, "learning_rate": 9.39209726443769e-06, "loss": 1.257, "step": 200 }, { "epoch": 0.06109422492401216, "grad_norm": 0.1875, "learning_rate": 9.389057750759879e-06, "loss": 1.2769, "step": 201 }, { "epoch": 0.06139817629179331, "grad_norm": 0.1953125, "learning_rate": 9.386018237082068e-06, "loss": 1.1767, "step": 202 }, { "epoch": 0.06170212765957447, "grad_norm": 0.1806640625, "learning_rate": 9.382978723404256e-06, "loss": 1.2967, "step": 203 }, { "epoch": 0.06200607902735562, "grad_norm": 0.2041015625, "learning_rate": 9.379939209726443e-06, "loss": 1.2399, "step": 204 }, { "epoch": 0.06231003039513678, "grad_norm": 0.1796875, "learning_rate": 9.376899696048632e-06, "loss": 1.2917, "step": 205 }, { "epoch": 0.06261398176291794, "grad_norm": 0.18359375, "learning_rate": 9.373860182370822e-06, "loss": 1.1698, "step": 206 }, { "epoch": 0.06291793313069909, "grad_norm": 0.1865234375, "learning_rate": 9.370820668693011e-06, "loss": 1.1435, "step": 207 }, { "epoch": 0.06322188449848025, "grad_norm": 0.1845703125, "learning_rate": 9.367781155015198e-06, "loss": 1.1893, "step": 208 }, { "epoch": 0.0635258358662614, "grad_norm": 0.1806640625, "learning_rate": 9.364741641337387e-06, "loss": 1.2196, "step": 209 }, { "epoch": 0.06382978723404255, "grad_norm": 0.19140625, "learning_rate": 9.361702127659576e-06, "loss": 1.2243, "step": 210 }, { "epoch": 0.06413373860182371, "grad_norm": 0.18359375, "learning_rate": 9.358662613981764e-06, "loss": 1.2542, "step": 211 }, { "epoch": 0.06443768996960486, "grad_norm": 0.1845703125, "learning_rate": 9.355623100303953e-06, "loss": 1.3045, "step": 212 }, { "epoch": 0.06474164133738602, "grad_norm": 0.1806640625, "learning_rate": 9.35258358662614e-06, "loss": 1.2459, "step": 213 }, { "epoch": 0.06504559270516717, "grad_norm": 0.208984375, "learning_rate": 9.349544072948329e-06, "loss": 1.1791, "step": 214 }, { "epoch": 0.06534954407294832, "grad_norm": 0.18359375, "learning_rate": 9.346504559270517e-06, "loss": 1.2311, "step": 215 }, { "epoch": 0.06565349544072949, "grad_norm": 0.2001953125, "learning_rate": 9.343465045592706e-06, "loss": 1.1532, "step": 216 }, { "epoch": 0.06595744680851064, "grad_norm": 0.2099609375, "learning_rate": 9.340425531914895e-06, "loss": 1.2047, "step": 217 }, { "epoch": 0.0662613981762918, "grad_norm": 0.1923828125, "learning_rate": 9.337386018237082e-06, "loss": 1.1562, "step": 218 }, { "epoch": 0.06656534954407295, "grad_norm": 0.1884765625, "learning_rate": 9.334346504559272e-06, "loss": 1.1996, "step": 219 }, { "epoch": 0.0668693009118541, "grad_norm": 0.1904296875, "learning_rate": 9.33130699088146e-06, "loss": 1.1816, "step": 220 }, { "epoch": 0.06717325227963526, "grad_norm": 0.197265625, "learning_rate": 9.32826747720365e-06, "loss": 1.2008, "step": 221 }, { "epoch": 0.06747720364741641, "grad_norm": 0.197265625, "learning_rate": 9.325227963525836e-06, "loss": 1.2054, "step": 222 }, { "epoch": 0.06778115501519758, "grad_norm": 0.1796875, "learning_rate": 9.322188449848025e-06, "loss": 1.1573, "step": 223 }, { "epoch": 0.06808510638297872, "grad_norm": 0.2265625, "learning_rate": 9.319148936170214e-06, "loss": 1.1668, "step": 224 }, { "epoch": 0.06838905775075987, "grad_norm": 0.1982421875, "learning_rate": 9.316109422492402e-06, "loss": 1.166, "step": 225 }, { "epoch": 0.06869300911854104, "grad_norm": 0.1982421875, "learning_rate": 9.313069908814591e-06, "loss": 1.2649, "step": 226 }, { "epoch": 0.06899696048632219, "grad_norm": 0.185546875, "learning_rate": 9.310030395136778e-06, "loss": 1.2789, "step": 227 }, { "epoch": 0.06930091185410335, "grad_norm": 0.1884765625, "learning_rate": 9.306990881458967e-06, "loss": 1.2067, "step": 228 }, { "epoch": 0.0696048632218845, "grad_norm": 0.2158203125, "learning_rate": 9.303951367781155e-06, "loss": 1.1201, "step": 229 }, { "epoch": 0.06990881458966565, "grad_norm": 0.1865234375, "learning_rate": 9.300911854103344e-06, "loss": 1.2371, "step": 230 }, { "epoch": 0.07021276595744681, "grad_norm": 0.193359375, "learning_rate": 9.297872340425533e-06, "loss": 1.2044, "step": 231 }, { "epoch": 0.07051671732522796, "grad_norm": 0.203125, "learning_rate": 9.29483282674772e-06, "loss": 1.2521, "step": 232 }, { "epoch": 0.07082066869300911, "grad_norm": 0.19921875, "learning_rate": 9.29179331306991e-06, "loss": 1.2284, "step": 233 }, { "epoch": 0.07112462006079028, "grad_norm": 0.1884765625, "learning_rate": 9.288753799392099e-06, "loss": 1.175, "step": 234 }, { "epoch": 0.07142857142857142, "grad_norm": 0.1865234375, "learning_rate": 9.285714285714288e-06, "loss": 1.2787, "step": 235 }, { "epoch": 0.07173252279635259, "grad_norm": 0.19140625, "learning_rate": 9.282674772036475e-06, "loss": 1.2148, "step": 236 }, { "epoch": 0.07203647416413374, "grad_norm": 0.18359375, "learning_rate": 9.279635258358663e-06, "loss": 1.1812, "step": 237 }, { "epoch": 0.07234042553191489, "grad_norm": 0.1943359375, "learning_rate": 9.276595744680852e-06, "loss": 1.2547, "step": 238 }, { "epoch": 0.07264437689969605, "grad_norm": 0.1962890625, "learning_rate": 9.27355623100304e-06, "loss": 1.1768, "step": 239 }, { "epoch": 0.0729483282674772, "grad_norm": 0.1962890625, "learning_rate": 9.27051671732523e-06, "loss": 1.196, "step": 240 }, { "epoch": 0.07325227963525836, "grad_norm": 0.1923828125, "learning_rate": 9.267477203647416e-06, "loss": 1.2243, "step": 241 }, { "epoch": 0.07355623100303951, "grad_norm": 0.1943359375, "learning_rate": 9.264437689969605e-06, "loss": 1.2857, "step": 242 }, { "epoch": 0.07386018237082066, "grad_norm": 0.1845703125, "learning_rate": 9.261398176291794e-06, "loss": 1.2122, "step": 243 }, { "epoch": 0.07416413373860183, "grad_norm": 0.189453125, "learning_rate": 9.258358662613982e-06, "loss": 1.2325, "step": 244 }, { "epoch": 0.07446808510638298, "grad_norm": 0.1923828125, "learning_rate": 9.255319148936171e-06, "loss": 1.118, "step": 245 }, { "epoch": 0.07477203647416414, "grad_norm": 0.2109375, "learning_rate": 9.25227963525836e-06, "loss": 1.1408, "step": 246 }, { "epoch": 0.07507598784194529, "grad_norm": 0.193359375, "learning_rate": 9.249240121580548e-06, "loss": 1.2214, "step": 247 }, { "epoch": 0.07537993920972644, "grad_norm": 0.1962890625, "learning_rate": 9.246200607902737e-06, "loss": 1.1958, "step": 248 }, { "epoch": 0.0756838905775076, "grad_norm": 0.185546875, "learning_rate": 9.243161094224926e-06, "loss": 1.2385, "step": 249 }, { "epoch": 0.07598784194528875, "grad_norm": 0.2041015625, "learning_rate": 9.240121580547113e-06, "loss": 1.1787, "step": 250 }, { "epoch": 0.07629179331306991, "grad_norm": 0.2099609375, "learning_rate": 9.237082066869301e-06, "loss": 1.2026, "step": 251 }, { "epoch": 0.07659574468085106, "grad_norm": 0.203125, "learning_rate": 9.23404255319149e-06, "loss": 1.1827, "step": 252 }, { "epoch": 0.07689969604863221, "grad_norm": 0.189453125, "learning_rate": 9.231003039513679e-06, "loss": 1.1968, "step": 253 }, { "epoch": 0.07720364741641338, "grad_norm": 0.205078125, "learning_rate": 9.227963525835867e-06, "loss": 1.2158, "step": 254 }, { "epoch": 0.07750759878419453, "grad_norm": 0.19140625, "learning_rate": 9.224924012158054e-06, "loss": 1.1788, "step": 255 }, { "epoch": 0.07781155015197569, "grad_norm": 0.1962890625, "learning_rate": 9.221884498480243e-06, "loss": 1.2151, "step": 256 }, { "epoch": 0.07811550151975684, "grad_norm": 0.201171875, "learning_rate": 9.218844984802432e-06, "loss": 1.1682, "step": 257 }, { "epoch": 0.07841945288753799, "grad_norm": 0.212890625, "learning_rate": 9.21580547112462e-06, "loss": 1.1243, "step": 258 }, { "epoch": 0.07872340425531915, "grad_norm": 0.201171875, "learning_rate": 9.21276595744681e-06, "loss": 1.1776, "step": 259 }, { "epoch": 0.0790273556231003, "grad_norm": 0.1865234375, "learning_rate": 9.209726443768998e-06, "loss": 1.1906, "step": 260 }, { "epoch": 0.07933130699088146, "grad_norm": 0.1962890625, "learning_rate": 9.206686930091187e-06, "loss": 1.2751, "step": 261 }, { "epoch": 0.07963525835866261, "grad_norm": 0.2119140625, "learning_rate": 9.203647416413375e-06, "loss": 1.1639, "step": 262 }, { "epoch": 0.07993920972644376, "grad_norm": 0.19140625, "learning_rate": 9.200607902735564e-06, "loss": 1.2376, "step": 263 }, { "epoch": 0.08024316109422493, "grad_norm": 0.181640625, "learning_rate": 9.197568389057751e-06, "loss": 1.2166, "step": 264 }, { "epoch": 0.08054711246200608, "grad_norm": 0.2001953125, "learning_rate": 9.19452887537994e-06, "loss": 1.2285, "step": 265 }, { "epoch": 0.08085106382978724, "grad_norm": 0.1962890625, "learning_rate": 9.191489361702128e-06, "loss": 1.2225, "step": 266 }, { "epoch": 0.08115501519756839, "grad_norm": 0.1962890625, "learning_rate": 9.188449848024317e-06, "loss": 1.1923, "step": 267 }, { "epoch": 0.08145896656534954, "grad_norm": 0.19921875, "learning_rate": 9.185410334346506e-06, "loss": 1.1716, "step": 268 }, { "epoch": 0.0817629179331307, "grad_norm": 0.197265625, "learning_rate": 9.182370820668693e-06, "loss": 1.1449, "step": 269 }, { "epoch": 0.08206686930091185, "grad_norm": 0.203125, "learning_rate": 9.179331306990881e-06, "loss": 1.1421, "step": 270 }, { "epoch": 0.08237082066869301, "grad_norm": 0.203125, "learning_rate": 9.17629179331307e-06, "loss": 1.1732, "step": 271 }, { "epoch": 0.08267477203647416, "grad_norm": 0.2158203125, "learning_rate": 9.17325227963526e-06, "loss": 1.1603, "step": 272 }, { "epoch": 0.08297872340425531, "grad_norm": 0.19921875, "learning_rate": 9.170212765957447e-06, "loss": 1.2195, "step": 273 }, { "epoch": 0.08328267477203648, "grad_norm": 0.193359375, "learning_rate": 9.167173252279636e-06, "loss": 1.1775, "step": 274 }, { "epoch": 0.08358662613981763, "grad_norm": 0.19921875, "learning_rate": 9.164133738601825e-06, "loss": 1.1843, "step": 275 }, { "epoch": 0.08389057750759879, "grad_norm": 0.2060546875, "learning_rate": 9.161094224924013e-06, "loss": 1.1358, "step": 276 }, { "epoch": 0.08419452887537994, "grad_norm": 0.1982421875, "learning_rate": 9.158054711246202e-06, "loss": 1.2488, "step": 277 }, { "epoch": 0.08449848024316109, "grad_norm": 0.2021484375, "learning_rate": 9.15501519756839e-06, "loss": 1.1549, "step": 278 }, { "epoch": 0.08480243161094225, "grad_norm": 0.2021484375, "learning_rate": 9.151975683890578e-06, "loss": 1.1845, "step": 279 }, { "epoch": 0.0851063829787234, "grad_norm": 0.193359375, "learning_rate": 9.148936170212767e-06, "loss": 1.1184, "step": 280 }, { "epoch": 0.08541033434650457, "grad_norm": 0.2021484375, "learning_rate": 9.145896656534955e-06, "loss": 1.2206, "step": 281 }, { "epoch": 0.08571428571428572, "grad_norm": 0.212890625, "learning_rate": 9.142857142857144e-06, "loss": 1.1201, "step": 282 }, { "epoch": 0.08601823708206686, "grad_norm": 0.3203125, "learning_rate": 9.139817629179331e-06, "loss": 1.1344, "step": 283 }, { "epoch": 0.08632218844984803, "grad_norm": 0.216796875, "learning_rate": 9.13677811550152e-06, "loss": 1.1416, "step": 284 }, { "epoch": 0.08662613981762918, "grad_norm": 0.1953125, "learning_rate": 9.133738601823708e-06, "loss": 1.207, "step": 285 }, { "epoch": 0.08693009118541034, "grad_norm": 0.1962890625, "learning_rate": 9.130699088145899e-06, "loss": 1.1189, "step": 286 }, { "epoch": 0.08723404255319149, "grad_norm": 0.2001953125, "learning_rate": 9.127659574468086e-06, "loss": 1.1902, "step": 287 }, { "epoch": 0.08753799392097264, "grad_norm": 0.1923828125, "learning_rate": 9.124620060790274e-06, "loss": 1.1688, "step": 288 }, { "epoch": 0.0878419452887538, "grad_norm": 0.205078125, "learning_rate": 9.121580547112463e-06, "loss": 1.0341, "step": 289 }, { "epoch": 0.08814589665653495, "grad_norm": 0.19921875, "learning_rate": 9.118541033434652e-06, "loss": 1.2262, "step": 290 }, { "epoch": 0.08844984802431612, "grad_norm": 0.1943359375, "learning_rate": 9.11550151975684e-06, "loss": 1.1057, "step": 291 }, { "epoch": 0.08875379939209727, "grad_norm": 0.2001953125, "learning_rate": 9.112462006079027e-06, "loss": 1.1571, "step": 292 }, { "epoch": 0.08905775075987842, "grad_norm": 0.2421875, "learning_rate": 9.109422492401216e-06, "loss": 1.1191, "step": 293 }, { "epoch": 0.08936170212765958, "grad_norm": 0.2177734375, "learning_rate": 9.106382978723405e-06, "loss": 1.1437, "step": 294 }, { "epoch": 0.08966565349544073, "grad_norm": 0.197265625, "learning_rate": 9.103343465045593e-06, "loss": 1.2028, "step": 295 }, { "epoch": 0.08996960486322189, "grad_norm": 0.2060546875, "learning_rate": 9.100303951367782e-06, "loss": 1.1126, "step": 296 }, { "epoch": 0.09027355623100304, "grad_norm": 0.197265625, "learning_rate": 9.097264437689969e-06, "loss": 1.1968, "step": 297 }, { "epoch": 0.09057750759878419, "grad_norm": 0.205078125, "learning_rate": 9.094224924012158e-06, "loss": 1.146, "step": 298 }, { "epoch": 0.09088145896656535, "grad_norm": 0.2001953125, "learning_rate": 9.091185410334348e-06, "loss": 1.1342, "step": 299 }, { "epoch": 0.0911854103343465, "grad_norm": 0.2021484375, "learning_rate": 9.088145896656537e-06, "loss": 1.1912, "step": 300 }, { "epoch": 0.09148936170212765, "grad_norm": 0.20703125, "learning_rate": 9.085106382978724e-06, "loss": 1.1076, "step": 301 }, { "epoch": 0.09179331306990882, "grad_norm": 0.2119140625, "learning_rate": 9.082066869300913e-06, "loss": 1.1889, "step": 302 }, { "epoch": 0.09209726443768997, "grad_norm": 0.1943359375, "learning_rate": 9.079027355623101e-06, "loss": 1.1275, "step": 303 }, { "epoch": 0.09240121580547113, "grad_norm": 0.19921875, "learning_rate": 9.07598784194529e-06, "loss": 1.0812, "step": 304 }, { "epoch": 0.09270516717325228, "grad_norm": 0.2001953125, "learning_rate": 9.072948328267479e-06, "loss": 1.1682, "step": 305 }, { "epoch": 0.09300911854103343, "grad_norm": 0.2109375, "learning_rate": 9.069908814589666e-06, "loss": 1.2064, "step": 306 }, { "epoch": 0.09331306990881459, "grad_norm": 0.203125, "learning_rate": 9.066869300911854e-06, "loss": 1.1564, "step": 307 }, { "epoch": 0.09361702127659574, "grad_norm": 0.2119140625, "learning_rate": 9.063829787234043e-06, "loss": 1.2003, "step": 308 }, { "epoch": 0.0939209726443769, "grad_norm": 0.212890625, "learning_rate": 9.060790273556232e-06, "loss": 1.2474, "step": 309 }, { "epoch": 0.09422492401215805, "grad_norm": 0.23828125, "learning_rate": 9.05775075987842e-06, "loss": 1.1259, "step": 310 }, { "epoch": 0.0945288753799392, "grad_norm": 0.2080078125, "learning_rate": 9.054711246200609e-06, "loss": 1.2094, "step": 311 }, { "epoch": 0.09483282674772037, "grad_norm": 0.2099609375, "learning_rate": 9.051671732522798e-06, "loss": 1.1734, "step": 312 }, { "epoch": 0.09513677811550152, "grad_norm": 0.208984375, "learning_rate": 9.048632218844986e-06, "loss": 1.1641, "step": 313 }, { "epoch": 0.09544072948328268, "grad_norm": 0.201171875, "learning_rate": 9.045592705167175e-06, "loss": 1.2167, "step": 314 }, { "epoch": 0.09574468085106383, "grad_norm": 0.203125, "learning_rate": 9.042553191489362e-06, "loss": 1.2242, "step": 315 }, { "epoch": 0.09604863221884498, "grad_norm": 0.2177734375, "learning_rate": 9.03951367781155e-06, "loss": 1.1531, "step": 316 }, { "epoch": 0.09635258358662614, "grad_norm": 0.2158203125, "learning_rate": 9.03647416413374e-06, "loss": 1.2071, "step": 317 }, { "epoch": 0.09665653495440729, "grad_norm": 0.205078125, "learning_rate": 9.033434650455928e-06, "loss": 1.1607, "step": 318 }, { "epoch": 0.09696048632218845, "grad_norm": 0.205078125, "learning_rate": 9.030395136778117e-06, "loss": 1.2422, "step": 319 }, { "epoch": 0.0972644376899696, "grad_norm": 0.2080078125, "learning_rate": 9.027355623100304e-06, "loss": 1.2199, "step": 320 }, { "epoch": 0.09756838905775075, "grad_norm": 0.205078125, "learning_rate": 9.024316109422492e-06, "loss": 1.1579, "step": 321 }, { "epoch": 0.09787234042553192, "grad_norm": 0.2041015625, "learning_rate": 9.021276595744681e-06, "loss": 1.2171, "step": 322 }, { "epoch": 0.09817629179331307, "grad_norm": 0.2041015625, "learning_rate": 9.01823708206687e-06, "loss": 1.1942, "step": 323 }, { "epoch": 0.09848024316109423, "grad_norm": 0.2158203125, "learning_rate": 9.015197568389058e-06, "loss": 1.2111, "step": 324 }, { "epoch": 0.09878419452887538, "grad_norm": 0.220703125, "learning_rate": 9.012158054711247e-06, "loss": 1.142, "step": 325 }, { "epoch": 0.09908814589665653, "grad_norm": 0.220703125, "learning_rate": 9.009118541033436e-06, "loss": 1.0939, "step": 326 }, { "epoch": 0.09939209726443769, "grad_norm": 0.2158203125, "learning_rate": 9.006079027355625e-06, "loss": 1.1872, "step": 327 }, { "epoch": 0.09969604863221884, "grad_norm": 0.2099609375, "learning_rate": 9.003039513677813e-06, "loss": 1.1099, "step": 328 }, { "epoch": 0.1, "grad_norm": 0.21484375, "learning_rate": 9e-06, "loss": 1.1395, "step": 329 }, { "epoch": 0.10030395136778116, "grad_norm": 0.2197265625, "learning_rate": 8.996960486322189e-06, "loss": 1.1309, "step": 330 }, { "epoch": 0.1006079027355623, "grad_norm": 0.2353515625, "learning_rate": 8.993920972644378e-06, "loss": 1.1566, "step": 331 }, { "epoch": 0.10091185410334347, "grad_norm": 0.2138671875, "learning_rate": 8.990881458966566e-06, "loss": 1.174, "step": 332 }, { "epoch": 0.10121580547112462, "grad_norm": 0.2275390625, "learning_rate": 8.987841945288755e-06, "loss": 1.1825, "step": 333 }, { "epoch": 0.10151975683890578, "grad_norm": 0.2021484375, "learning_rate": 8.984802431610942e-06, "loss": 1.236, "step": 334 }, { "epoch": 0.10182370820668693, "grad_norm": 0.2255859375, "learning_rate": 8.98176291793313e-06, "loss": 1.1955, "step": 335 }, { "epoch": 0.10212765957446808, "grad_norm": 0.232421875, "learning_rate": 8.97872340425532e-06, "loss": 1.089, "step": 336 }, { "epoch": 0.10243161094224924, "grad_norm": 0.23046875, "learning_rate": 8.975683890577508e-06, "loss": 1.1838, "step": 337 }, { "epoch": 0.10273556231003039, "grad_norm": 0.1982421875, "learning_rate": 8.972644376899697e-06, "loss": 1.1734, "step": 338 }, { "epoch": 0.10303951367781156, "grad_norm": 0.20703125, "learning_rate": 8.969604863221885e-06, "loss": 1.1612, "step": 339 }, { "epoch": 0.1033434650455927, "grad_norm": 0.2119140625, "learning_rate": 8.966565349544074e-06, "loss": 1.0977, "step": 340 }, { "epoch": 0.10364741641337386, "grad_norm": 0.2080078125, "learning_rate": 8.963525835866263e-06, "loss": 1.2368, "step": 341 }, { "epoch": 0.10395136778115502, "grad_norm": 0.2412109375, "learning_rate": 8.960486322188451e-06, "loss": 1.0306, "step": 342 }, { "epoch": 0.10425531914893617, "grad_norm": 0.20703125, "learning_rate": 8.957446808510638e-06, "loss": 1.1257, "step": 343 }, { "epoch": 0.10455927051671733, "grad_norm": 0.21484375, "learning_rate": 8.954407294832827e-06, "loss": 1.2072, "step": 344 }, { "epoch": 0.10486322188449848, "grad_norm": 0.2314453125, "learning_rate": 8.951367781155016e-06, "loss": 1.1901, "step": 345 }, { "epoch": 0.10516717325227963, "grad_norm": 0.2060546875, "learning_rate": 8.948328267477204e-06, "loss": 1.1904, "step": 346 }, { "epoch": 0.1054711246200608, "grad_norm": 0.22265625, "learning_rate": 8.945288753799393e-06, "loss": 1.1658, "step": 347 }, { "epoch": 0.10577507598784194, "grad_norm": 0.2265625, "learning_rate": 8.942249240121582e-06, "loss": 1.0453, "step": 348 }, { "epoch": 0.1060790273556231, "grad_norm": 0.2060546875, "learning_rate": 8.939209726443769e-06, "loss": 1.111, "step": 349 }, { "epoch": 0.10638297872340426, "grad_norm": 0.23828125, "learning_rate": 8.936170212765958e-06, "loss": 1.1802, "step": 350 }, { "epoch": 0.1066869300911854, "grad_norm": 0.259765625, "learning_rate": 8.933130699088146e-06, "loss": 1.1455, "step": 351 }, { "epoch": 0.10699088145896657, "grad_norm": 0.2099609375, "learning_rate": 8.930091185410335e-06, "loss": 1.1794, "step": 352 }, { "epoch": 0.10729483282674772, "grad_norm": 0.21875, "learning_rate": 8.927051671732524e-06, "loss": 1.0908, "step": 353 }, { "epoch": 0.10759878419452888, "grad_norm": 0.208984375, "learning_rate": 8.924012158054712e-06, "loss": 1.1417, "step": 354 }, { "epoch": 0.10790273556231003, "grad_norm": 0.2177734375, "learning_rate": 8.920972644376901e-06, "loss": 1.1748, "step": 355 }, { "epoch": 0.10820668693009118, "grad_norm": 0.205078125, "learning_rate": 8.91793313069909e-06, "loss": 1.1147, "step": 356 }, { "epoch": 0.10851063829787234, "grad_norm": 0.2236328125, "learning_rate": 8.914893617021277e-06, "loss": 1.1051, "step": 357 }, { "epoch": 0.1088145896656535, "grad_norm": 0.2734375, "learning_rate": 8.911854103343465e-06, "loss": 1.1196, "step": 358 }, { "epoch": 0.10911854103343466, "grad_norm": 0.21484375, "learning_rate": 8.908814589665654e-06, "loss": 1.0944, "step": 359 }, { "epoch": 0.1094224924012158, "grad_norm": 0.2158203125, "learning_rate": 8.905775075987843e-06, "loss": 1.1264, "step": 360 }, { "epoch": 0.10972644376899696, "grad_norm": 0.2578125, "learning_rate": 8.902735562310031e-06, "loss": 1.2017, "step": 361 }, { "epoch": 0.11003039513677812, "grad_norm": 0.205078125, "learning_rate": 8.89969604863222e-06, "loss": 1.1849, "step": 362 }, { "epoch": 0.11033434650455927, "grad_norm": 0.21875, "learning_rate": 8.896656534954407e-06, "loss": 1.1455, "step": 363 }, { "epoch": 0.11063829787234042, "grad_norm": 0.255859375, "learning_rate": 8.893617021276596e-06, "loss": 1.1294, "step": 364 }, { "epoch": 0.11094224924012158, "grad_norm": 0.318359375, "learning_rate": 8.890577507598786e-06, "loss": 1.1652, "step": 365 }, { "epoch": 0.11124620060790273, "grad_norm": 0.203125, "learning_rate": 8.887537993920973e-06, "loss": 1.1389, "step": 366 }, { "epoch": 0.1115501519756839, "grad_norm": 0.2158203125, "learning_rate": 8.884498480243162e-06, "loss": 1.1288, "step": 367 }, { "epoch": 0.11185410334346504, "grad_norm": 0.203125, "learning_rate": 8.88145896656535e-06, "loss": 1.066, "step": 368 }, { "epoch": 0.1121580547112462, "grad_norm": 0.208984375, "learning_rate": 8.87841945288754e-06, "loss": 1.1934, "step": 369 }, { "epoch": 0.11246200607902736, "grad_norm": 0.208984375, "learning_rate": 8.875379939209728e-06, "loss": 1.2113, "step": 370 }, { "epoch": 0.1127659574468085, "grad_norm": 0.2158203125, "learning_rate": 8.872340425531915e-06, "loss": 1.0893, "step": 371 }, { "epoch": 0.11306990881458967, "grad_norm": 0.1982421875, "learning_rate": 8.869300911854104e-06, "loss": 1.2049, "step": 372 }, { "epoch": 0.11337386018237082, "grad_norm": 0.2119140625, "learning_rate": 8.866261398176292e-06, "loss": 1.2143, "step": 373 }, { "epoch": 0.11367781155015197, "grad_norm": 0.208984375, "learning_rate": 8.863221884498481e-06, "loss": 1.2179, "step": 374 }, { "epoch": 0.11398176291793313, "grad_norm": 0.2138671875, "learning_rate": 8.86018237082067e-06, "loss": 1.1569, "step": 375 }, { "epoch": 0.11428571428571428, "grad_norm": 0.23046875, "learning_rate": 8.857142857142858e-06, "loss": 1.123, "step": 376 }, { "epoch": 0.11458966565349545, "grad_norm": 0.2275390625, "learning_rate": 8.854103343465045e-06, "loss": 1.0708, "step": 377 }, { "epoch": 0.1148936170212766, "grad_norm": 0.2080078125, "learning_rate": 8.851063829787234e-06, "loss": 1.1655, "step": 378 }, { "epoch": 0.11519756838905774, "grad_norm": 0.20703125, "learning_rate": 8.848024316109424e-06, "loss": 1.1771, "step": 379 }, { "epoch": 0.11550151975683891, "grad_norm": 0.21484375, "learning_rate": 8.844984802431611e-06, "loss": 1.1258, "step": 380 }, { "epoch": 0.11580547112462006, "grad_norm": 0.208984375, "learning_rate": 8.8419452887538e-06, "loss": 1.1832, "step": 381 }, { "epoch": 0.11610942249240122, "grad_norm": 0.232421875, "learning_rate": 8.838905775075989e-06, "loss": 1.1084, "step": 382 }, { "epoch": 0.11641337386018237, "grad_norm": 0.2138671875, "learning_rate": 8.835866261398177e-06, "loss": 1.1964, "step": 383 }, { "epoch": 0.11671732522796352, "grad_norm": 0.2197265625, "learning_rate": 8.832826747720366e-06, "loss": 1.1227, "step": 384 }, { "epoch": 0.11702127659574468, "grad_norm": 0.234375, "learning_rate": 8.829787234042555e-06, "loss": 1.1582, "step": 385 }, { "epoch": 0.11732522796352583, "grad_norm": 0.2216796875, "learning_rate": 8.826747720364742e-06, "loss": 1.1712, "step": 386 }, { "epoch": 0.117629179331307, "grad_norm": 0.22265625, "learning_rate": 8.82370820668693e-06, "loss": 1.1976, "step": 387 }, { "epoch": 0.11793313069908815, "grad_norm": 0.20703125, "learning_rate": 8.820668693009119e-06, "loss": 1.1536, "step": 388 }, { "epoch": 0.1182370820668693, "grad_norm": 0.216796875, "learning_rate": 8.817629179331308e-06, "loss": 1.1487, "step": 389 }, { "epoch": 0.11854103343465046, "grad_norm": 0.2109375, "learning_rate": 8.814589665653496e-06, "loss": 1.1096, "step": 390 }, { "epoch": 0.11884498480243161, "grad_norm": 0.23828125, "learning_rate": 8.811550151975683e-06, "loss": 1.1272, "step": 391 }, { "epoch": 0.11914893617021277, "grad_norm": 0.2314453125, "learning_rate": 8.808510638297874e-06, "loss": 1.1776, "step": 392 }, { "epoch": 0.11945288753799392, "grad_norm": 0.21484375, "learning_rate": 8.805471124620063e-06, "loss": 1.1928, "step": 393 }, { "epoch": 0.11975683890577507, "grad_norm": 0.2138671875, "learning_rate": 8.80243161094225e-06, "loss": 1.1618, "step": 394 }, { "epoch": 0.12006079027355623, "grad_norm": 0.2080078125, "learning_rate": 8.799392097264438e-06, "loss": 1.0857, "step": 395 }, { "epoch": 0.12036474164133738, "grad_norm": 0.23046875, "learning_rate": 8.796352583586627e-06, "loss": 1.1456, "step": 396 }, { "epoch": 0.12066869300911855, "grad_norm": 0.2255859375, "learning_rate": 8.793313069908816e-06, "loss": 1.1855, "step": 397 }, { "epoch": 0.1209726443768997, "grad_norm": 0.2197265625, "learning_rate": 8.790273556231004e-06, "loss": 1.1794, "step": 398 }, { "epoch": 0.12127659574468085, "grad_norm": 0.271484375, "learning_rate": 8.787234042553193e-06, "loss": 1.094, "step": 399 }, { "epoch": 0.12158054711246201, "grad_norm": 0.20703125, "learning_rate": 8.78419452887538e-06, "loss": 1.174, "step": 400 }, { "epoch": 0.12188449848024316, "grad_norm": 0.2158203125, "learning_rate": 8.781155015197569e-06, "loss": 1.1149, "step": 401 }, { "epoch": 0.12218844984802432, "grad_norm": 0.2177734375, "learning_rate": 8.778115501519757e-06, "loss": 1.1103, "step": 402 }, { "epoch": 0.12249240121580547, "grad_norm": 0.294921875, "learning_rate": 8.775075987841946e-06, "loss": 1.1737, "step": 403 }, { "epoch": 0.12279635258358662, "grad_norm": 0.2158203125, "learning_rate": 8.772036474164135e-06, "loss": 1.1507, "step": 404 }, { "epoch": 0.12310030395136778, "grad_norm": 0.2158203125, "learning_rate": 8.768996960486323e-06, "loss": 1.1643, "step": 405 }, { "epoch": 0.12340425531914893, "grad_norm": 0.2451171875, "learning_rate": 8.765957446808512e-06, "loss": 1.099, "step": 406 }, { "epoch": 0.1237082066869301, "grad_norm": 0.21484375, "learning_rate": 8.7629179331307e-06, "loss": 1.1365, "step": 407 }, { "epoch": 0.12401215805471125, "grad_norm": 0.2158203125, "learning_rate": 8.759878419452888e-06, "loss": 1.2345, "step": 408 }, { "epoch": 0.1243161094224924, "grad_norm": 0.21875, "learning_rate": 8.756838905775076e-06, "loss": 1.1507, "step": 409 }, { "epoch": 0.12462006079027356, "grad_norm": 0.22265625, "learning_rate": 8.753799392097265e-06, "loss": 1.0768, "step": 410 }, { "epoch": 0.12492401215805471, "grad_norm": 0.2392578125, "learning_rate": 8.750759878419454e-06, "loss": 1.2029, "step": 411 }, { "epoch": 0.12522796352583587, "grad_norm": 0.232421875, "learning_rate": 8.747720364741642e-06, "loss": 1.0744, "step": 412 }, { "epoch": 0.125531914893617, "grad_norm": 0.2177734375, "learning_rate": 8.744680851063831e-06, "loss": 1.0858, "step": 413 }, { "epoch": 0.12583586626139817, "grad_norm": 0.2314453125, "learning_rate": 8.741641337386018e-06, "loss": 1.0881, "step": 414 }, { "epoch": 0.12613981762917933, "grad_norm": 0.2158203125, "learning_rate": 8.738601823708207e-06, "loss": 1.1327, "step": 415 }, { "epoch": 0.1264437689969605, "grad_norm": 0.23828125, "learning_rate": 8.735562310030395e-06, "loss": 1.1347, "step": 416 }, { "epoch": 0.12674772036474163, "grad_norm": 0.279296875, "learning_rate": 8.732522796352584e-06, "loss": 1.1122, "step": 417 }, { "epoch": 0.1270516717325228, "grad_norm": 0.248046875, "learning_rate": 8.729483282674773e-06, "loss": 1.0635, "step": 418 }, { "epoch": 0.12735562310030396, "grad_norm": 0.21875, "learning_rate": 8.726443768996962e-06, "loss": 1.0837, "step": 419 }, { "epoch": 0.1276595744680851, "grad_norm": 0.2373046875, "learning_rate": 8.72340425531915e-06, "loss": 1.2012, "step": 420 }, { "epoch": 0.12796352583586626, "grad_norm": 0.220703125, "learning_rate": 8.720364741641339e-06, "loss": 1.0837, "step": 421 }, { "epoch": 0.12826747720364742, "grad_norm": 0.2353515625, "learning_rate": 8.717325227963528e-06, "loss": 1.1291, "step": 422 }, { "epoch": 0.12857142857142856, "grad_norm": 0.22265625, "learning_rate": 8.714285714285715e-06, "loss": 1.1114, "step": 423 }, { "epoch": 0.12887537993920972, "grad_norm": 0.224609375, "learning_rate": 8.711246200607903e-06, "loss": 1.1013, "step": 424 }, { "epoch": 0.12917933130699089, "grad_norm": 0.244140625, "learning_rate": 8.708206686930092e-06, "loss": 1.1135, "step": 425 }, { "epoch": 0.12948328267477205, "grad_norm": 0.2451171875, "learning_rate": 8.70516717325228e-06, "loss": 1.1421, "step": 426 }, { "epoch": 0.12978723404255318, "grad_norm": 0.2451171875, "learning_rate": 8.70212765957447e-06, "loss": 1.1213, "step": 427 }, { "epoch": 0.13009118541033435, "grad_norm": 0.2197265625, "learning_rate": 8.699088145896656e-06, "loss": 1.2355, "step": 428 }, { "epoch": 0.1303951367781155, "grad_norm": 0.275390625, "learning_rate": 8.696048632218845e-06, "loss": 1.1754, "step": 429 }, { "epoch": 0.13069908814589665, "grad_norm": 0.2314453125, "learning_rate": 8.693009118541034e-06, "loss": 1.1204, "step": 430 }, { "epoch": 0.1310030395136778, "grad_norm": 0.2216796875, "learning_rate": 8.689969604863222e-06, "loss": 1.1428, "step": 431 }, { "epoch": 0.13130699088145897, "grad_norm": 0.2109375, "learning_rate": 8.686930091185411e-06, "loss": 1.1631, "step": 432 }, { "epoch": 0.1316109422492401, "grad_norm": 0.21875, "learning_rate": 8.6838905775076e-06, "loss": 1.2235, "step": 433 }, { "epoch": 0.13191489361702127, "grad_norm": 0.2294921875, "learning_rate": 8.680851063829788e-06, "loss": 1.1666, "step": 434 }, { "epoch": 0.13221884498480244, "grad_norm": 0.2373046875, "learning_rate": 8.677811550151977e-06, "loss": 1.1546, "step": 435 }, { "epoch": 0.1325227963525836, "grad_norm": 0.240234375, "learning_rate": 8.674772036474166e-06, "loss": 1.0957, "step": 436 }, { "epoch": 0.13282674772036474, "grad_norm": 0.216796875, "learning_rate": 8.671732522796353e-06, "loss": 1.1955, "step": 437 }, { "epoch": 0.1331306990881459, "grad_norm": 0.28515625, "learning_rate": 8.668693009118541e-06, "loss": 1.2011, "step": 438 }, { "epoch": 0.13343465045592706, "grad_norm": 0.2138671875, "learning_rate": 8.66565349544073e-06, "loss": 1.0997, "step": 439 }, { "epoch": 0.1337386018237082, "grad_norm": 0.228515625, "learning_rate": 8.662613981762919e-06, "loss": 1.0701, "step": 440 }, { "epoch": 0.13404255319148936, "grad_norm": 0.25390625, "learning_rate": 8.659574468085108e-06, "loss": 1.1451, "step": 441 }, { "epoch": 0.13434650455927052, "grad_norm": 0.2216796875, "learning_rate": 8.656534954407295e-06, "loss": 1.1463, "step": 442 }, { "epoch": 0.13465045592705166, "grad_norm": 0.2294921875, "learning_rate": 8.653495440729483e-06, "loss": 1.1914, "step": 443 }, { "epoch": 0.13495440729483282, "grad_norm": 0.2197265625, "learning_rate": 8.650455927051672e-06, "loss": 1.0847, "step": 444 }, { "epoch": 0.135258358662614, "grad_norm": 0.2265625, "learning_rate": 8.64741641337386e-06, "loss": 1.1234, "step": 445 }, { "epoch": 0.13556231003039515, "grad_norm": 0.2265625, "learning_rate": 8.64437689969605e-06, "loss": 1.064, "step": 446 }, { "epoch": 0.13586626139817629, "grad_norm": 0.236328125, "learning_rate": 8.641337386018238e-06, "loss": 1.1551, "step": 447 }, { "epoch": 0.13617021276595745, "grad_norm": 0.2333984375, "learning_rate": 8.638297872340427e-06, "loss": 0.9941, "step": 448 }, { "epoch": 0.1364741641337386, "grad_norm": 0.22265625, "learning_rate": 8.635258358662615e-06, "loss": 1.1141, "step": 449 }, { "epoch": 0.13677811550151975, "grad_norm": 0.25, "learning_rate": 8.632218844984804e-06, "loss": 1.0894, "step": 450 }, { "epoch": 0.1370820668693009, "grad_norm": 0.216796875, "learning_rate": 8.629179331306991e-06, "loss": 1.1026, "step": 451 }, { "epoch": 0.13738601823708207, "grad_norm": 0.22265625, "learning_rate": 8.62613981762918e-06, "loss": 1.0941, "step": 452 }, { "epoch": 0.1376899696048632, "grad_norm": 0.2216796875, "learning_rate": 8.623100303951368e-06, "loss": 1.1359, "step": 453 }, { "epoch": 0.13799392097264437, "grad_norm": 0.2275390625, "learning_rate": 8.620060790273557e-06, "loss": 1.0801, "step": 454 }, { "epoch": 0.13829787234042554, "grad_norm": 0.228515625, "learning_rate": 8.617021276595746e-06, "loss": 1.1122, "step": 455 }, { "epoch": 0.1386018237082067, "grad_norm": 0.2216796875, "learning_rate": 8.613981762917933e-06, "loss": 1.1588, "step": 456 }, { "epoch": 0.13890577507598784, "grad_norm": 0.2265625, "learning_rate": 8.610942249240121e-06, "loss": 1.1887, "step": 457 }, { "epoch": 0.139209726443769, "grad_norm": 0.22265625, "learning_rate": 8.607902735562312e-06, "loss": 1.1681, "step": 458 }, { "epoch": 0.13951367781155016, "grad_norm": 0.224609375, "learning_rate": 8.6048632218845e-06, "loss": 1.1512, "step": 459 }, { "epoch": 0.1398176291793313, "grad_norm": 0.224609375, "learning_rate": 8.601823708206687e-06, "loss": 1.089, "step": 460 }, { "epoch": 0.14012158054711246, "grad_norm": 0.2353515625, "learning_rate": 8.598784194528876e-06, "loss": 1.2509, "step": 461 }, { "epoch": 0.14042553191489363, "grad_norm": 0.2109375, "learning_rate": 8.595744680851065e-06, "loss": 1.1401, "step": 462 }, { "epoch": 0.14072948328267476, "grad_norm": 0.224609375, "learning_rate": 8.592705167173254e-06, "loss": 1.0963, "step": 463 }, { "epoch": 0.14103343465045592, "grad_norm": 0.216796875, "learning_rate": 8.589665653495442e-06, "loss": 1.0796, "step": 464 }, { "epoch": 0.1413373860182371, "grad_norm": 0.2255859375, "learning_rate": 8.58662613981763e-06, "loss": 1.094, "step": 465 }, { "epoch": 0.14164133738601822, "grad_norm": 0.2177734375, "learning_rate": 8.583586626139818e-06, "loss": 1.1365, "step": 466 }, { "epoch": 0.1419452887537994, "grad_norm": 0.2294921875, "learning_rate": 8.580547112462007e-06, "loss": 1.209, "step": 467 }, { "epoch": 0.14224924012158055, "grad_norm": 0.21875, "learning_rate": 8.577507598784195e-06, "loss": 1.1549, "step": 468 }, { "epoch": 0.1425531914893617, "grad_norm": 0.2255859375, "learning_rate": 8.574468085106384e-06, "loss": 1.0753, "step": 469 }, { "epoch": 0.14285714285714285, "grad_norm": 0.2294921875, "learning_rate": 8.571428571428571e-06, "loss": 1.0792, "step": 470 }, { "epoch": 0.143161094224924, "grad_norm": 0.25390625, "learning_rate": 8.568389057750761e-06, "loss": 1.1214, "step": 471 }, { "epoch": 0.14346504559270518, "grad_norm": 0.224609375, "learning_rate": 8.56534954407295e-06, "loss": 1.1122, "step": 472 }, { "epoch": 0.1437689969604863, "grad_norm": 0.265625, "learning_rate": 8.562310030395139e-06, "loss": 1.1221, "step": 473 }, { "epoch": 0.14407294832826747, "grad_norm": 0.234375, "learning_rate": 8.559270516717326e-06, "loss": 1.0543, "step": 474 }, { "epoch": 0.14437689969604864, "grad_norm": 0.23046875, "learning_rate": 8.556231003039514e-06, "loss": 1.1121, "step": 475 }, { "epoch": 0.14468085106382977, "grad_norm": 0.2294921875, "learning_rate": 8.553191489361703e-06, "loss": 1.1265, "step": 476 }, { "epoch": 0.14498480243161094, "grad_norm": 0.21875, "learning_rate": 8.550151975683892e-06, "loss": 1.2198, "step": 477 }, { "epoch": 0.1452887537993921, "grad_norm": 0.2158203125, "learning_rate": 8.54711246200608e-06, "loss": 1.2375, "step": 478 }, { "epoch": 0.14559270516717326, "grad_norm": 0.2197265625, "learning_rate": 8.544072948328267e-06, "loss": 1.1216, "step": 479 }, { "epoch": 0.1458966565349544, "grad_norm": 0.2197265625, "learning_rate": 8.541033434650456e-06, "loss": 1.1571, "step": 480 }, { "epoch": 0.14620060790273556, "grad_norm": 0.216796875, "learning_rate": 8.537993920972645e-06, "loss": 1.1555, "step": 481 }, { "epoch": 0.14650455927051673, "grad_norm": 0.234375, "learning_rate": 8.534954407294833e-06, "loss": 1.1489, "step": 482 }, { "epoch": 0.14680851063829786, "grad_norm": 0.2275390625, "learning_rate": 8.531914893617022e-06, "loss": 1.1868, "step": 483 }, { "epoch": 0.14711246200607903, "grad_norm": 0.248046875, "learning_rate": 8.528875379939209e-06, "loss": 1.0698, "step": 484 }, { "epoch": 0.1474164133738602, "grad_norm": 0.224609375, "learning_rate": 8.5258358662614e-06, "loss": 1.1138, "step": 485 }, { "epoch": 0.14772036474164132, "grad_norm": 0.22265625, "learning_rate": 8.522796352583588e-06, "loss": 1.0779, "step": 486 }, { "epoch": 0.1480243161094225, "grad_norm": 0.2275390625, "learning_rate": 8.519756838905777e-06, "loss": 1.1607, "step": 487 }, { "epoch": 0.14832826747720365, "grad_norm": 0.224609375, "learning_rate": 8.516717325227964e-06, "loss": 1.0437, "step": 488 }, { "epoch": 0.14863221884498481, "grad_norm": 0.224609375, "learning_rate": 8.513677811550153e-06, "loss": 1.1635, "step": 489 }, { "epoch": 0.14893617021276595, "grad_norm": 0.2490234375, "learning_rate": 8.510638297872341e-06, "loss": 1.0869, "step": 490 }, { "epoch": 0.1492401215805471, "grad_norm": 0.28125, "learning_rate": 8.50759878419453e-06, "loss": 0.9742, "step": 491 }, { "epoch": 0.14954407294832828, "grad_norm": 0.2392578125, "learning_rate": 8.504559270516719e-06, "loss": 1.079, "step": 492 }, { "epoch": 0.1498480243161094, "grad_norm": 0.2197265625, "learning_rate": 8.501519756838906e-06, "loss": 1.1653, "step": 493 }, { "epoch": 0.15015197568389058, "grad_norm": 0.234375, "learning_rate": 8.498480243161094e-06, "loss": 1.1138, "step": 494 }, { "epoch": 0.15045592705167174, "grad_norm": 0.212890625, "learning_rate": 8.495440729483283e-06, "loss": 1.1824, "step": 495 }, { "epoch": 0.15075987841945288, "grad_norm": 0.220703125, "learning_rate": 8.492401215805472e-06, "loss": 1.1925, "step": 496 }, { "epoch": 0.15106382978723404, "grad_norm": 0.2421875, "learning_rate": 8.48936170212766e-06, "loss": 1.1281, "step": 497 }, { "epoch": 0.1513677811550152, "grad_norm": 0.23046875, "learning_rate": 8.486322188449849e-06, "loss": 1.1013, "step": 498 }, { "epoch": 0.15167173252279637, "grad_norm": 0.2255859375, "learning_rate": 8.483282674772038e-06, "loss": 1.1907, "step": 499 }, { "epoch": 0.1519756838905775, "grad_norm": 0.2373046875, "learning_rate": 8.480243161094226e-06, "loss": 1.0905, "step": 500 }, { "epoch": 0.15227963525835866, "grad_norm": 0.232421875, "learning_rate": 8.477203647416415e-06, "loss": 1.1299, "step": 501 }, { "epoch": 0.15258358662613983, "grad_norm": 0.259765625, "learning_rate": 8.474164133738602e-06, "loss": 1.1294, "step": 502 }, { "epoch": 0.15288753799392096, "grad_norm": 0.23046875, "learning_rate": 8.47112462006079e-06, "loss": 1.1225, "step": 503 }, { "epoch": 0.15319148936170213, "grad_norm": 0.2353515625, "learning_rate": 8.46808510638298e-06, "loss": 1.1531, "step": 504 }, { "epoch": 0.1534954407294833, "grad_norm": 0.240234375, "learning_rate": 8.465045592705168e-06, "loss": 1.1074, "step": 505 }, { "epoch": 0.15379939209726443, "grad_norm": 0.2197265625, "learning_rate": 8.462006079027357e-06, "loss": 1.0975, "step": 506 }, { "epoch": 0.1541033434650456, "grad_norm": 0.234375, "learning_rate": 8.458966565349544e-06, "loss": 1.1505, "step": 507 }, { "epoch": 0.15440729483282675, "grad_norm": 0.2412109375, "learning_rate": 8.455927051671732e-06, "loss": 1.1202, "step": 508 }, { "epoch": 0.15471124620060792, "grad_norm": 0.224609375, "learning_rate": 8.452887537993921e-06, "loss": 1.0899, "step": 509 }, { "epoch": 0.15501519756838905, "grad_norm": 0.26953125, "learning_rate": 8.44984802431611e-06, "loss": 1.0767, "step": 510 }, { "epoch": 0.15531914893617021, "grad_norm": 0.2265625, "learning_rate": 8.446808510638299e-06, "loss": 1.1123, "step": 511 }, { "epoch": 0.15562310030395138, "grad_norm": 0.2294921875, "learning_rate": 8.443768996960487e-06, "loss": 1.1409, "step": 512 }, { "epoch": 0.15592705167173251, "grad_norm": 0.23828125, "learning_rate": 8.440729483282676e-06, "loss": 1.147, "step": 513 }, { "epoch": 0.15623100303951368, "grad_norm": 0.240234375, "learning_rate": 8.437689969604865e-06, "loss": 1.1614, "step": 514 }, { "epoch": 0.15653495440729484, "grad_norm": 0.2216796875, "learning_rate": 8.434650455927053e-06, "loss": 1.0973, "step": 515 }, { "epoch": 0.15683890577507598, "grad_norm": 0.22265625, "learning_rate": 8.43161094224924e-06, "loss": 1.1644, "step": 516 }, { "epoch": 0.15714285714285714, "grad_norm": 0.236328125, "learning_rate": 8.428571428571429e-06, "loss": 1.0586, "step": 517 }, { "epoch": 0.1574468085106383, "grad_norm": 0.2275390625, "learning_rate": 8.425531914893618e-06, "loss": 1.0923, "step": 518 }, { "epoch": 0.15775075987841947, "grad_norm": 0.22265625, "learning_rate": 8.422492401215806e-06, "loss": 1.1247, "step": 519 }, { "epoch": 0.1580547112462006, "grad_norm": 0.236328125, "learning_rate": 8.419452887537995e-06, "loss": 1.2089, "step": 520 }, { "epoch": 0.15835866261398177, "grad_norm": 0.234375, "learning_rate": 8.416413373860182e-06, "loss": 1.0919, "step": 521 }, { "epoch": 0.15866261398176293, "grad_norm": 0.2294921875, "learning_rate": 8.41337386018237e-06, "loss": 1.1132, "step": 522 }, { "epoch": 0.15896656534954406, "grad_norm": 0.2392578125, "learning_rate": 8.41033434650456e-06, "loss": 1.0616, "step": 523 }, { "epoch": 0.15927051671732523, "grad_norm": 0.2275390625, "learning_rate": 8.407294832826748e-06, "loss": 1.2061, "step": 524 }, { "epoch": 0.1595744680851064, "grad_norm": 0.24609375, "learning_rate": 8.404255319148937e-06, "loss": 1.2191, "step": 525 }, { "epoch": 0.15987841945288753, "grad_norm": 0.271484375, "learning_rate": 8.401215805471125e-06, "loss": 1.0368, "step": 526 }, { "epoch": 0.1601823708206687, "grad_norm": 0.251953125, "learning_rate": 8.398176291793314e-06, "loss": 1.129, "step": 527 }, { "epoch": 0.16048632218844985, "grad_norm": 0.234375, "learning_rate": 8.395136778115503e-06, "loss": 1.0845, "step": 528 }, { "epoch": 0.160790273556231, "grad_norm": 0.2294921875, "learning_rate": 8.392097264437691e-06, "loss": 1.1008, "step": 529 }, { "epoch": 0.16109422492401215, "grad_norm": 0.2255859375, "learning_rate": 8.389057750759878e-06, "loss": 1.1728, "step": 530 }, { "epoch": 0.16139817629179332, "grad_norm": 0.248046875, "learning_rate": 8.386018237082067e-06, "loss": 1.1444, "step": 531 }, { "epoch": 0.16170212765957448, "grad_norm": 0.2373046875, "learning_rate": 8.382978723404256e-06, "loss": 1.1492, "step": 532 }, { "epoch": 0.16200607902735562, "grad_norm": 0.2392578125, "learning_rate": 8.379939209726445e-06, "loss": 1.0466, "step": 533 }, { "epoch": 0.16231003039513678, "grad_norm": 0.2353515625, "learning_rate": 8.376899696048633e-06, "loss": 1.0237, "step": 534 }, { "epoch": 0.16261398176291794, "grad_norm": 0.236328125, "learning_rate": 8.37386018237082e-06, "loss": 1.0965, "step": 535 }, { "epoch": 0.16291793313069908, "grad_norm": 0.2412109375, "learning_rate": 8.370820668693009e-06, "loss": 1.0814, "step": 536 }, { "epoch": 0.16322188449848024, "grad_norm": 0.2412109375, "learning_rate": 8.367781155015198e-06, "loss": 1.1304, "step": 537 }, { "epoch": 0.1635258358662614, "grad_norm": 0.23046875, "learning_rate": 8.364741641337388e-06, "loss": 1.1457, "step": 538 }, { "epoch": 0.16382978723404254, "grad_norm": 0.2236328125, "learning_rate": 8.361702127659575e-06, "loss": 1.0666, "step": 539 }, { "epoch": 0.1641337386018237, "grad_norm": 0.2353515625, "learning_rate": 8.358662613981764e-06, "loss": 1.1254, "step": 540 }, { "epoch": 0.16443768996960487, "grad_norm": 0.240234375, "learning_rate": 8.355623100303952e-06, "loss": 1.0838, "step": 541 }, { "epoch": 0.16474164133738603, "grad_norm": 0.240234375, "learning_rate": 8.352583586626141e-06, "loss": 1.0705, "step": 542 }, { "epoch": 0.16504559270516717, "grad_norm": 0.2294921875, "learning_rate": 8.34954407294833e-06, "loss": 1.1626, "step": 543 }, { "epoch": 0.16534954407294833, "grad_norm": 0.28125, "learning_rate": 8.346504559270517e-06, "loss": 1.118, "step": 544 }, { "epoch": 0.1656534954407295, "grad_norm": 0.2353515625, "learning_rate": 8.343465045592705e-06, "loss": 1.1583, "step": 545 }, { "epoch": 0.16595744680851063, "grad_norm": 0.2333984375, "learning_rate": 8.340425531914894e-06, "loss": 1.1528, "step": 546 }, { "epoch": 0.1662613981762918, "grad_norm": 0.232421875, "learning_rate": 8.337386018237083e-06, "loss": 1.1674, "step": 547 }, { "epoch": 0.16656534954407295, "grad_norm": 0.2431640625, "learning_rate": 8.334346504559271e-06, "loss": 1.0644, "step": 548 }, { "epoch": 0.1668693009118541, "grad_norm": 0.236328125, "learning_rate": 8.331306990881458e-06, "loss": 1.0609, "step": 549 }, { "epoch": 0.16717325227963525, "grad_norm": 0.23046875, "learning_rate": 8.328267477203647e-06, "loss": 1.0636, "step": 550 }, { "epoch": 0.16747720364741642, "grad_norm": 0.2333984375, "learning_rate": 8.325227963525837e-06, "loss": 1.1098, "step": 551 }, { "epoch": 0.16778115501519758, "grad_norm": 0.2490234375, "learning_rate": 8.322188449848026e-06, "loss": 1.0945, "step": 552 }, { "epoch": 0.16808510638297872, "grad_norm": 0.228515625, "learning_rate": 8.319148936170213e-06, "loss": 1.0324, "step": 553 }, { "epoch": 0.16838905775075988, "grad_norm": 0.2265625, "learning_rate": 8.316109422492402e-06, "loss": 1.114, "step": 554 }, { "epoch": 0.16869300911854104, "grad_norm": 0.2353515625, "learning_rate": 8.31306990881459e-06, "loss": 1.105, "step": 555 }, { "epoch": 0.16899696048632218, "grad_norm": 0.2294921875, "learning_rate": 8.31003039513678e-06, "loss": 1.0641, "step": 556 }, { "epoch": 0.16930091185410334, "grad_norm": 0.248046875, "learning_rate": 8.306990881458968e-06, "loss": 1.1434, "step": 557 }, { "epoch": 0.1696048632218845, "grad_norm": 0.228515625, "learning_rate": 8.303951367781155e-06, "loss": 1.0776, "step": 558 }, { "epoch": 0.16990881458966564, "grad_norm": 0.2373046875, "learning_rate": 8.300911854103344e-06, "loss": 1.1188, "step": 559 }, { "epoch": 0.1702127659574468, "grad_norm": 0.2275390625, "learning_rate": 8.297872340425532e-06, "loss": 1.1748, "step": 560 }, { "epoch": 0.17051671732522797, "grad_norm": 0.236328125, "learning_rate": 8.294832826747721e-06, "loss": 1.1072, "step": 561 }, { "epoch": 0.17082066869300913, "grad_norm": 0.255859375, "learning_rate": 8.29179331306991e-06, "loss": 1.1091, "step": 562 }, { "epoch": 0.17112462006079027, "grad_norm": 0.2392578125, "learning_rate": 8.288753799392098e-06, "loss": 1.1029, "step": 563 }, { "epoch": 0.17142857142857143, "grad_norm": 0.2412109375, "learning_rate": 8.285714285714287e-06, "loss": 1.0719, "step": 564 }, { "epoch": 0.1717325227963526, "grad_norm": 0.2333984375, "learning_rate": 8.282674772036476e-06, "loss": 1.0859, "step": 565 }, { "epoch": 0.17203647416413373, "grad_norm": 0.2412109375, "learning_rate": 8.279635258358664e-06, "loss": 1.1299, "step": 566 }, { "epoch": 0.1723404255319149, "grad_norm": 0.2314453125, "learning_rate": 8.276595744680851e-06, "loss": 1.0626, "step": 567 }, { "epoch": 0.17264437689969606, "grad_norm": 0.23828125, "learning_rate": 8.27355623100304e-06, "loss": 1.1613, "step": 568 }, { "epoch": 0.1729483282674772, "grad_norm": 0.2353515625, "learning_rate": 8.270516717325229e-06, "loss": 1.0832, "step": 569 }, { "epoch": 0.17325227963525835, "grad_norm": 0.2294921875, "learning_rate": 8.267477203647417e-06, "loss": 1.0862, "step": 570 }, { "epoch": 0.17355623100303952, "grad_norm": 0.236328125, "learning_rate": 8.264437689969606e-06, "loss": 1.1025, "step": 571 }, { "epoch": 0.17386018237082068, "grad_norm": 0.2197265625, "learning_rate": 8.261398176291793e-06, "loss": 1.1498, "step": 572 }, { "epoch": 0.17416413373860182, "grad_norm": 0.2392578125, "learning_rate": 8.258358662613982e-06, "loss": 1.137, "step": 573 }, { "epoch": 0.17446808510638298, "grad_norm": 0.2314453125, "learning_rate": 8.25531914893617e-06, "loss": 1.14, "step": 574 }, { "epoch": 0.17477203647416414, "grad_norm": 0.2392578125, "learning_rate": 8.252279635258359e-06, "loss": 1.0733, "step": 575 }, { "epoch": 0.17507598784194528, "grad_norm": 0.24609375, "learning_rate": 8.249240121580548e-06, "loss": 1.1063, "step": 576 }, { "epoch": 0.17537993920972644, "grad_norm": 0.2216796875, "learning_rate": 8.246200607902736e-06, "loss": 1.0803, "step": 577 }, { "epoch": 0.1756838905775076, "grad_norm": 0.232421875, "learning_rate": 8.243161094224925e-06, "loss": 1.0704, "step": 578 }, { "epoch": 0.17598784194528874, "grad_norm": 0.2373046875, "learning_rate": 8.240121580547114e-06, "loss": 1.1177, "step": 579 }, { "epoch": 0.1762917933130699, "grad_norm": 0.2412109375, "learning_rate": 8.237082066869303e-06, "loss": 1.0695, "step": 580 }, { "epoch": 0.17659574468085107, "grad_norm": 0.2294921875, "learning_rate": 8.23404255319149e-06, "loss": 1.1325, "step": 581 }, { "epoch": 0.17689969604863223, "grad_norm": 0.2265625, "learning_rate": 8.231003039513678e-06, "loss": 1.1012, "step": 582 }, { "epoch": 0.17720364741641337, "grad_norm": 0.263671875, "learning_rate": 8.227963525835867e-06, "loss": 1.1499, "step": 583 }, { "epoch": 0.17750759878419453, "grad_norm": 0.2333984375, "learning_rate": 8.224924012158056e-06, "loss": 1.0809, "step": 584 }, { "epoch": 0.1778115501519757, "grad_norm": 0.39453125, "learning_rate": 8.221884498480244e-06, "loss": 1.1217, "step": 585 }, { "epoch": 0.17811550151975683, "grad_norm": 0.27734375, "learning_rate": 8.218844984802431e-06, "loss": 1.1296, "step": 586 }, { "epoch": 0.178419452887538, "grad_norm": 0.25390625, "learning_rate": 8.21580547112462e-06, "loss": 1.17, "step": 587 }, { "epoch": 0.17872340425531916, "grad_norm": 0.248046875, "learning_rate": 8.212765957446809e-06, "loss": 1.1288, "step": 588 }, { "epoch": 0.1790273556231003, "grad_norm": 0.25, "learning_rate": 8.209726443768997e-06, "loss": 1.1402, "step": 589 }, { "epoch": 0.17933130699088146, "grad_norm": 0.2451171875, "learning_rate": 8.206686930091186e-06, "loss": 1.1238, "step": 590 }, { "epoch": 0.17963525835866262, "grad_norm": 0.240234375, "learning_rate": 8.203647416413375e-06, "loss": 1.0936, "step": 591 }, { "epoch": 0.17993920972644378, "grad_norm": 0.240234375, "learning_rate": 8.200607902735563e-06, "loss": 1.0782, "step": 592 }, { "epoch": 0.18024316109422492, "grad_norm": 0.248046875, "learning_rate": 8.197568389057752e-06, "loss": 1.029, "step": 593 }, { "epoch": 0.18054711246200608, "grad_norm": 0.2373046875, "learning_rate": 8.19452887537994e-06, "loss": 1.1144, "step": 594 }, { "epoch": 0.18085106382978725, "grad_norm": 0.25, "learning_rate": 8.191489361702128e-06, "loss": 1.1556, "step": 595 }, { "epoch": 0.18115501519756838, "grad_norm": 0.240234375, "learning_rate": 8.188449848024316e-06, "loss": 1.0929, "step": 596 }, { "epoch": 0.18145896656534954, "grad_norm": 0.2294921875, "learning_rate": 8.185410334346505e-06, "loss": 1.1586, "step": 597 }, { "epoch": 0.1817629179331307, "grad_norm": 0.2392578125, "learning_rate": 8.182370820668694e-06, "loss": 1.1251, "step": 598 }, { "epoch": 0.18206686930091184, "grad_norm": 0.240234375, "learning_rate": 8.179331306990882e-06, "loss": 1.1043, "step": 599 }, { "epoch": 0.182370820668693, "grad_norm": 0.2412109375, "learning_rate": 8.176291793313071e-06, "loss": 1.1088, "step": 600 }, { "epoch": 0.18267477203647417, "grad_norm": 0.2470703125, "learning_rate": 8.173252279635258e-06, "loss": 1.0762, "step": 601 }, { "epoch": 0.1829787234042553, "grad_norm": 0.251953125, "learning_rate": 8.170212765957447e-06, "loss": 1.1449, "step": 602 }, { "epoch": 0.18328267477203647, "grad_norm": 0.2412109375, "learning_rate": 8.167173252279636e-06, "loss": 1.1208, "step": 603 }, { "epoch": 0.18358662613981763, "grad_norm": 0.2421875, "learning_rate": 8.164133738601824e-06, "loss": 1.117, "step": 604 }, { "epoch": 0.1838905775075988, "grad_norm": 0.24609375, "learning_rate": 8.161094224924013e-06, "loss": 1.1448, "step": 605 }, { "epoch": 0.18419452887537993, "grad_norm": 0.2490234375, "learning_rate": 8.158054711246202e-06, "loss": 1.1729, "step": 606 }, { "epoch": 0.1844984802431611, "grad_norm": 0.234375, "learning_rate": 8.15501519756839e-06, "loss": 1.0911, "step": 607 }, { "epoch": 0.18480243161094226, "grad_norm": 0.259765625, "learning_rate": 8.151975683890579e-06, "loss": 1.1078, "step": 608 }, { "epoch": 0.1851063829787234, "grad_norm": 0.24609375, "learning_rate": 8.148936170212766e-06, "loss": 1.1521, "step": 609 }, { "epoch": 0.18541033434650456, "grad_norm": 0.2490234375, "learning_rate": 8.145896656534955e-06, "loss": 1.1362, "step": 610 }, { "epoch": 0.18571428571428572, "grad_norm": 0.236328125, "learning_rate": 8.142857142857143e-06, "loss": 1.045, "step": 611 }, { "epoch": 0.18601823708206686, "grad_norm": 0.228515625, "learning_rate": 8.139817629179332e-06, "loss": 1.1327, "step": 612 }, { "epoch": 0.18632218844984802, "grad_norm": 0.2421875, "learning_rate": 8.13677811550152e-06, "loss": 1.0929, "step": 613 }, { "epoch": 0.18662613981762918, "grad_norm": 0.24609375, "learning_rate": 8.13373860182371e-06, "loss": 1.2257, "step": 614 }, { "epoch": 0.18693009118541035, "grad_norm": 0.26171875, "learning_rate": 8.130699088145896e-06, "loss": 1.2015, "step": 615 }, { "epoch": 0.18723404255319148, "grad_norm": 0.251953125, "learning_rate": 8.127659574468085e-06, "loss": 1.0675, "step": 616 }, { "epoch": 0.18753799392097265, "grad_norm": 0.2294921875, "learning_rate": 8.124620060790275e-06, "loss": 1.1313, "step": 617 }, { "epoch": 0.1878419452887538, "grad_norm": 0.251953125, "learning_rate": 8.121580547112462e-06, "loss": 1.0771, "step": 618 }, { "epoch": 0.18814589665653494, "grad_norm": 0.27734375, "learning_rate": 8.118541033434651e-06, "loss": 1.0439, "step": 619 }, { "epoch": 0.1884498480243161, "grad_norm": 0.26953125, "learning_rate": 8.11550151975684e-06, "loss": 1.1547, "step": 620 }, { "epoch": 0.18875379939209727, "grad_norm": 0.263671875, "learning_rate": 8.112462006079028e-06, "loss": 1.1048, "step": 621 }, { "epoch": 0.1890577507598784, "grad_norm": 0.36328125, "learning_rate": 8.109422492401217e-06, "loss": 1.1075, "step": 622 }, { "epoch": 0.18936170212765957, "grad_norm": 0.2373046875, "learning_rate": 8.106382978723404e-06, "loss": 1.2019, "step": 623 }, { "epoch": 0.18966565349544073, "grad_norm": 0.2412109375, "learning_rate": 8.103343465045593e-06, "loss": 1.043, "step": 624 }, { "epoch": 0.1899696048632219, "grad_norm": 0.2333984375, "learning_rate": 8.100303951367782e-06, "loss": 1.0861, "step": 625 }, { "epoch": 0.19027355623100303, "grad_norm": 0.244140625, "learning_rate": 8.09726443768997e-06, "loss": 1.1367, "step": 626 }, { "epoch": 0.1905775075987842, "grad_norm": 0.26953125, "learning_rate": 8.094224924012159e-06, "loss": 1.0616, "step": 627 }, { "epoch": 0.19088145896656536, "grad_norm": 0.240234375, "learning_rate": 8.091185410334348e-06, "loss": 1.0589, "step": 628 }, { "epoch": 0.1911854103343465, "grad_norm": 0.23828125, "learning_rate": 8.088145896656535e-06, "loss": 1.054, "step": 629 }, { "epoch": 0.19148936170212766, "grad_norm": 0.2333984375, "learning_rate": 8.085106382978723e-06, "loss": 1.1097, "step": 630 }, { "epoch": 0.19179331306990882, "grad_norm": 0.244140625, "learning_rate": 8.082066869300914e-06, "loss": 1.0795, "step": 631 }, { "epoch": 0.19209726443768996, "grad_norm": 0.2431640625, "learning_rate": 8.0790273556231e-06, "loss": 1.0585, "step": 632 }, { "epoch": 0.19240121580547112, "grad_norm": 0.2333984375, "learning_rate": 8.07598784194529e-06, "loss": 1.1216, "step": 633 }, { "epoch": 0.19270516717325228, "grad_norm": 0.2353515625, "learning_rate": 8.072948328267478e-06, "loss": 1.0791, "step": 634 }, { "epoch": 0.19300911854103345, "grad_norm": 0.251953125, "learning_rate": 8.069908814589667e-06, "loss": 1.07, "step": 635 }, { "epoch": 0.19331306990881458, "grad_norm": 0.240234375, "learning_rate": 8.066869300911855e-06, "loss": 1.1122, "step": 636 }, { "epoch": 0.19361702127659575, "grad_norm": 0.23046875, "learning_rate": 8.063829787234044e-06, "loss": 1.0548, "step": 637 }, { "epoch": 0.1939209726443769, "grad_norm": 0.2431640625, "learning_rate": 8.060790273556231e-06, "loss": 1.1107, "step": 638 }, { "epoch": 0.19422492401215805, "grad_norm": 0.248046875, "learning_rate": 8.05775075987842e-06, "loss": 1.1051, "step": 639 }, { "epoch": 0.1945288753799392, "grad_norm": 0.25390625, "learning_rate": 8.054711246200608e-06, "loss": 1.057, "step": 640 }, { "epoch": 0.19483282674772037, "grad_norm": 0.2412109375, "learning_rate": 8.051671732522797e-06, "loss": 1.0985, "step": 641 }, { "epoch": 0.1951367781155015, "grad_norm": 0.2392578125, "learning_rate": 8.048632218844986e-06, "loss": 1.0485, "step": 642 }, { "epoch": 0.19544072948328267, "grad_norm": 0.2373046875, "learning_rate": 8.045592705167173e-06, "loss": 1.1506, "step": 643 }, { "epoch": 0.19574468085106383, "grad_norm": 0.248046875, "learning_rate": 8.042553191489363e-06, "loss": 1.089, "step": 644 }, { "epoch": 0.196048632218845, "grad_norm": 0.2431640625, "learning_rate": 8.039513677811552e-06, "loss": 1.0916, "step": 645 }, { "epoch": 0.19635258358662613, "grad_norm": 0.259765625, "learning_rate": 8.036474164133739e-06, "loss": 1.0776, "step": 646 }, { "epoch": 0.1966565349544073, "grad_norm": 0.244140625, "learning_rate": 8.033434650455927e-06, "loss": 1.1585, "step": 647 }, { "epoch": 0.19696048632218846, "grad_norm": 0.251953125, "learning_rate": 8.030395136778116e-06, "loss": 1.1211, "step": 648 }, { "epoch": 0.1972644376899696, "grad_norm": 0.2431640625, "learning_rate": 8.027355623100305e-06, "loss": 1.1301, "step": 649 }, { "epoch": 0.19756838905775076, "grad_norm": 0.240234375, "learning_rate": 8.024316109422494e-06, "loss": 1.1583, "step": 650 }, { "epoch": 0.19787234042553192, "grad_norm": 0.25390625, "learning_rate": 8.021276595744682e-06, "loss": 1.0942, "step": 651 }, { "epoch": 0.19817629179331306, "grad_norm": 0.2333984375, "learning_rate": 8.01823708206687e-06, "loss": 1.1099, "step": 652 }, { "epoch": 0.19848024316109422, "grad_norm": 0.2333984375, "learning_rate": 8.015197568389058e-06, "loss": 1.1736, "step": 653 }, { "epoch": 0.19878419452887539, "grad_norm": 0.2333984375, "learning_rate": 8.012158054711247e-06, "loss": 1.1261, "step": 654 }, { "epoch": 0.19908814589665655, "grad_norm": 0.25, "learning_rate": 8.009118541033435e-06, "loss": 1.0726, "step": 655 }, { "epoch": 0.19939209726443768, "grad_norm": 0.251953125, "learning_rate": 8.006079027355624e-06, "loss": 1.1492, "step": 656 }, { "epoch": 0.19969604863221885, "grad_norm": 0.234375, "learning_rate": 8.003039513677813e-06, "loss": 1.0985, "step": 657 }, { "epoch": 0.2, "grad_norm": 0.25, "learning_rate": 8.000000000000001e-06, "loss": 1.1085, "step": 658 }, { "epoch": 0.20030395136778115, "grad_norm": 0.2470703125, "learning_rate": 7.99696048632219e-06, "loss": 1.0549, "step": 659 }, { "epoch": 0.2006079027355623, "grad_norm": 0.28125, "learning_rate": 7.993920972644377e-06, "loss": 1.1213, "step": 660 }, { "epoch": 0.20091185410334347, "grad_norm": 0.236328125, "learning_rate": 7.990881458966566e-06, "loss": 1.1406, "step": 661 }, { "epoch": 0.2012158054711246, "grad_norm": 0.240234375, "learning_rate": 7.987841945288754e-06, "loss": 1.0516, "step": 662 }, { "epoch": 0.20151975683890577, "grad_norm": 0.2412109375, "learning_rate": 7.984802431610943e-06, "loss": 1.0675, "step": 663 }, { "epoch": 0.20182370820668694, "grad_norm": 0.251953125, "learning_rate": 7.981762917933132e-06, "loss": 1.1468, "step": 664 }, { "epoch": 0.20212765957446807, "grad_norm": 0.2392578125, "learning_rate": 7.97872340425532e-06, "loss": 1.0709, "step": 665 }, { "epoch": 0.20243161094224923, "grad_norm": 0.2431640625, "learning_rate": 7.975683890577507e-06, "loss": 1.0698, "step": 666 }, { "epoch": 0.2027355623100304, "grad_norm": 0.24609375, "learning_rate": 7.972644376899696e-06, "loss": 1.078, "step": 667 }, { "epoch": 0.20303951367781156, "grad_norm": 0.26171875, "learning_rate": 7.969604863221885e-06, "loss": 1.0952, "step": 668 }, { "epoch": 0.2033434650455927, "grad_norm": 0.232421875, "learning_rate": 7.966565349544073e-06, "loss": 1.1106, "step": 669 }, { "epoch": 0.20364741641337386, "grad_norm": 0.23828125, "learning_rate": 7.963525835866262e-06, "loss": 1.1522, "step": 670 }, { "epoch": 0.20395136778115502, "grad_norm": 0.236328125, "learning_rate": 7.960486322188451e-06, "loss": 1.103, "step": 671 }, { "epoch": 0.20425531914893616, "grad_norm": 0.2353515625, "learning_rate": 7.95744680851064e-06, "loss": 1.0589, "step": 672 }, { "epoch": 0.20455927051671732, "grad_norm": 0.24609375, "learning_rate": 7.954407294832828e-06, "loss": 1.1333, "step": 673 }, { "epoch": 0.2048632218844985, "grad_norm": 0.232421875, "learning_rate": 7.951367781155017e-06, "loss": 1.1048, "step": 674 }, { "epoch": 0.20516717325227962, "grad_norm": 0.25390625, "learning_rate": 7.948328267477204e-06, "loss": 1.208, "step": 675 }, { "epoch": 0.20547112462006079, "grad_norm": 0.251953125, "learning_rate": 7.945288753799393e-06, "loss": 1.0512, "step": 676 }, { "epoch": 0.20577507598784195, "grad_norm": 0.24609375, "learning_rate": 7.942249240121581e-06, "loss": 1.0919, "step": 677 }, { "epoch": 0.2060790273556231, "grad_norm": 0.2421875, "learning_rate": 7.93920972644377e-06, "loss": 1.1232, "step": 678 }, { "epoch": 0.20638297872340425, "grad_norm": 0.240234375, "learning_rate": 7.936170212765959e-06, "loss": 1.0246, "step": 679 }, { "epoch": 0.2066869300911854, "grad_norm": 0.2578125, "learning_rate": 7.933130699088146e-06, "loss": 1.0348, "step": 680 }, { "epoch": 0.20699088145896657, "grad_norm": 0.2421875, "learning_rate": 7.930091185410334e-06, "loss": 1.1106, "step": 681 }, { "epoch": 0.2072948328267477, "grad_norm": 0.259765625, "learning_rate": 7.927051671732523e-06, "loss": 1.0486, "step": 682 }, { "epoch": 0.20759878419452887, "grad_norm": 0.2470703125, "learning_rate": 7.924012158054712e-06, "loss": 1.0725, "step": 683 }, { "epoch": 0.20790273556231004, "grad_norm": 0.2451171875, "learning_rate": 7.9209726443769e-06, "loss": 1.0823, "step": 684 }, { "epoch": 0.20820668693009117, "grad_norm": 0.2451171875, "learning_rate": 7.917933130699089e-06, "loss": 1.1734, "step": 685 }, { "epoch": 0.20851063829787234, "grad_norm": 0.4921875, "learning_rate": 7.914893617021278e-06, "loss": 1.052, "step": 686 }, { "epoch": 0.2088145896656535, "grad_norm": 0.24609375, "learning_rate": 7.911854103343466e-06, "loss": 1.0755, "step": 687 }, { "epoch": 0.20911854103343466, "grad_norm": 0.259765625, "learning_rate": 7.908814589665655e-06, "loss": 1.2172, "step": 688 }, { "epoch": 0.2094224924012158, "grad_norm": 0.240234375, "learning_rate": 7.905775075987842e-06, "loss": 1.1453, "step": 689 }, { "epoch": 0.20972644376899696, "grad_norm": 0.2392578125, "learning_rate": 7.90273556231003e-06, "loss": 1.0912, "step": 690 }, { "epoch": 0.21003039513677813, "grad_norm": 0.2431640625, "learning_rate": 7.89969604863222e-06, "loss": 1.1248, "step": 691 }, { "epoch": 0.21033434650455926, "grad_norm": 0.271484375, "learning_rate": 7.896656534954408e-06, "loss": 1.0488, "step": 692 }, { "epoch": 0.21063829787234042, "grad_norm": 0.26171875, "learning_rate": 7.893617021276597e-06, "loss": 1.0384, "step": 693 }, { "epoch": 0.2109422492401216, "grad_norm": 0.2431640625, "learning_rate": 7.890577507598784e-06, "loss": 1.0377, "step": 694 }, { "epoch": 0.21124620060790272, "grad_norm": 0.251953125, "learning_rate": 7.887537993920973e-06, "loss": 1.0925, "step": 695 }, { "epoch": 0.2115501519756839, "grad_norm": 0.2353515625, "learning_rate": 7.884498480243161e-06, "loss": 1.0745, "step": 696 }, { "epoch": 0.21185410334346505, "grad_norm": 0.259765625, "learning_rate": 7.88145896656535e-06, "loss": 1.0927, "step": 697 }, { "epoch": 0.2121580547112462, "grad_norm": 0.234375, "learning_rate": 7.878419452887539e-06, "loss": 1.1056, "step": 698 }, { "epoch": 0.21246200607902735, "grad_norm": 0.2470703125, "learning_rate": 7.875379939209727e-06, "loss": 1.0429, "step": 699 }, { "epoch": 0.2127659574468085, "grad_norm": 0.2412109375, "learning_rate": 7.872340425531916e-06, "loss": 1.116, "step": 700 }, { "epoch": 0.21306990881458968, "grad_norm": 0.2373046875, "learning_rate": 7.869300911854105e-06, "loss": 1.0975, "step": 701 }, { "epoch": 0.2133738601823708, "grad_norm": 0.244140625, "learning_rate": 7.866261398176293e-06, "loss": 1.1162, "step": 702 }, { "epoch": 0.21367781155015197, "grad_norm": 0.248046875, "learning_rate": 7.86322188449848e-06, "loss": 1.1307, "step": 703 }, { "epoch": 0.21398176291793314, "grad_norm": 0.35546875, "learning_rate": 7.860182370820669e-06, "loss": 1.0833, "step": 704 }, { "epoch": 0.21428571428571427, "grad_norm": 0.25, "learning_rate": 7.857142857142858e-06, "loss": 1.174, "step": 705 }, { "epoch": 0.21458966565349544, "grad_norm": 0.27734375, "learning_rate": 7.854103343465046e-06, "loss": 1.0522, "step": 706 }, { "epoch": 0.2148936170212766, "grad_norm": 0.267578125, "learning_rate": 7.851063829787235e-06, "loss": 1.0789, "step": 707 }, { "epoch": 0.21519756838905776, "grad_norm": 0.275390625, "learning_rate": 7.848024316109422e-06, "loss": 1.0145, "step": 708 }, { "epoch": 0.2155015197568389, "grad_norm": 0.25390625, "learning_rate": 7.84498480243161e-06, "loss": 1.083, "step": 709 }, { "epoch": 0.21580547112462006, "grad_norm": 0.24609375, "learning_rate": 7.841945288753801e-06, "loss": 1.1292, "step": 710 }, { "epoch": 0.21610942249240123, "grad_norm": 0.259765625, "learning_rate": 7.83890577507599e-06, "loss": 1.0969, "step": 711 }, { "epoch": 0.21641337386018236, "grad_norm": 0.25390625, "learning_rate": 7.835866261398177e-06, "loss": 1.0533, "step": 712 }, { "epoch": 0.21671732522796353, "grad_norm": 0.25, "learning_rate": 7.832826747720365e-06, "loss": 1.1207, "step": 713 }, { "epoch": 0.2170212765957447, "grad_norm": 0.2392578125, "learning_rate": 7.829787234042554e-06, "loss": 1.0737, "step": 714 }, { "epoch": 0.21732522796352582, "grad_norm": 0.25390625, "learning_rate": 7.826747720364743e-06, "loss": 1.0773, "step": 715 }, { "epoch": 0.217629179331307, "grad_norm": 0.2451171875, "learning_rate": 7.823708206686932e-06, "loss": 1.0748, "step": 716 }, { "epoch": 0.21793313069908815, "grad_norm": 0.2421875, "learning_rate": 7.820668693009118e-06, "loss": 1.062, "step": 717 }, { "epoch": 0.21823708206686931, "grad_norm": 0.255859375, "learning_rate": 7.817629179331307e-06, "loss": 1.0952, "step": 718 }, { "epoch": 0.21854103343465045, "grad_norm": 0.310546875, "learning_rate": 7.814589665653496e-06, "loss": 1.0857, "step": 719 }, { "epoch": 0.2188449848024316, "grad_norm": 0.251953125, "learning_rate": 7.811550151975685e-06, "loss": 1.1096, "step": 720 }, { "epoch": 0.21914893617021278, "grad_norm": 0.296875, "learning_rate": 7.808510638297873e-06, "loss": 1.1054, "step": 721 }, { "epoch": 0.2194528875379939, "grad_norm": 0.2578125, "learning_rate": 7.80547112462006e-06, "loss": 1.1565, "step": 722 }, { "epoch": 0.21975683890577508, "grad_norm": 0.23828125, "learning_rate": 7.802431610942249e-06, "loss": 1.091, "step": 723 }, { "epoch": 0.22006079027355624, "grad_norm": 0.25390625, "learning_rate": 7.79939209726444e-06, "loss": 1.0689, "step": 724 }, { "epoch": 0.22036474164133737, "grad_norm": 0.2431640625, "learning_rate": 7.796352583586628e-06, "loss": 1.0659, "step": 725 }, { "epoch": 0.22066869300911854, "grad_norm": 0.2421875, "learning_rate": 7.793313069908815e-06, "loss": 1.1103, "step": 726 }, { "epoch": 0.2209726443768997, "grad_norm": 0.2392578125, "learning_rate": 7.790273556231004e-06, "loss": 1.1242, "step": 727 }, { "epoch": 0.22127659574468084, "grad_norm": 0.2734375, "learning_rate": 7.787234042553192e-06, "loss": 1.1023, "step": 728 }, { "epoch": 0.221580547112462, "grad_norm": 0.2353515625, "learning_rate": 7.784194528875381e-06, "loss": 1.1285, "step": 729 }, { "epoch": 0.22188449848024316, "grad_norm": 0.25, "learning_rate": 7.78115501519757e-06, "loss": 1.0761, "step": 730 }, { "epoch": 0.22218844984802433, "grad_norm": 0.2373046875, "learning_rate": 7.778115501519757e-06, "loss": 1.0552, "step": 731 }, { "epoch": 0.22249240121580546, "grad_norm": 0.248046875, "learning_rate": 7.775075987841945e-06, "loss": 1.0818, "step": 732 }, { "epoch": 0.22279635258358663, "grad_norm": 0.255859375, "learning_rate": 7.772036474164134e-06, "loss": 1.161, "step": 733 }, { "epoch": 0.2231003039513678, "grad_norm": 0.240234375, "learning_rate": 7.768996960486323e-06, "loss": 1.0569, "step": 734 }, { "epoch": 0.22340425531914893, "grad_norm": 0.24609375, "learning_rate": 7.765957446808511e-06, "loss": 1.0756, "step": 735 }, { "epoch": 0.2237082066869301, "grad_norm": 0.244140625, "learning_rate": 7.762917933130698e-06, "loss": 1.1014, "step": 736 }, { "epoch": 0.22401215805471125, "grad_norm": 0.2578125, "learning_rate": 7.759878419452889e-06, "loss": 1.1226, "step": 737 }, { "epoch": 0.2243161094224924, "grad_norm": 0.263671875, "learning_rate": 7.756838905775077e-06, "loss": 1.0781, "step": 738 }, { "epoch": 0.22462006079027355, "grad_norm": 0.265625, "learning_rate": 7.753799392097266e-06, "loss": 1.0564, "step": 739 }, { "epoch": 0.22492401215805471, "grad_norm": 0.244140625, "learning_rate": 7.750759878419453e-06, "loss": 1.1283, "step": 740 }, { "epoch": 0.22522796352583588, "grad_norm": 0.248046875, "learning_rate": 7.747720364741642e-06, "loss": 1.0197, "step": 741 }, { "epoch": 0.225531914893617, "grad_norm": 0.2392578125, "learning_rate": 7.74468085106383e-06, "loss": 1.1204, "step": 742 }, { "epoch": 0.22583586626139818, "grad_norm": 0.2451171875, "learning_rate": 7.74164133738602e-06, "loss": 1.1315, "step": 743 }, { "epoch": 0.22613981762917934, "grad_norm": 0.255859375, "learning_rate": 7.738601823708208e-06, "loss": 1.1439, "step": 744 }, { "epoch": 0.22644376899696048, "grad_norm": 0.24609375, "learning_rate": 7.735562310030395e-06, "loss": 1.0534, "step": 745 }, { "epoch": 0.22674772036474164, "grad_norm": 0.240234375, "learning_rate": 7.732522796352584e-06, "loss": 1.0914, "step": 746 }, { "epoch": 0.2270516717325228, "grad_norm": 0.240234375, "learning_rate": 7.729483282674772e-06, "loss": 1.0622, "step": 747 }, { "epoch": 0.22735562310030394, "grad_norm": 0.240234375, "learning_rate": 7.726443768996961e-06, "loss": 1.0676, "step": 748 }, { "epoch": 0.2276595744680851, "grad_norm": 0.255859375, "learning_rate": 7.72340425531915e-06, "loss": 1.0928, "step": 749 }, { "epoch": 0.22796352583586627, "grad_norm": 0.25, "learning_rate": 7.720364741641338e-06, "loss": 1.0991, "step": 750 }, { "epoch": 0.22826747720364743, "grad_norm": 0.2490234375, "learning_rate": 7.717325227963527e-06, "loss": 1.1539, "step": 751 }, { "epoch": 0.22857142857142856, "grad_norm": 0.31640625, "learning_rate": 7.714285714285716e-06, "loss": 1.0592, "step": 752 }, { "epoch": 0.22887537993920973, "grad_norm": 0.26953125, "learning_rate": 7.711246200607904e-06, "loss": 1.1252, "step": 753 }, { "epoch": 0.2291793313069909, "grad_norm": 0.3046875, "learning_rate": 7.708206686930091e-06, "loss": 1.0652, "step": 754 }, { "epoch": 0.22948328267477203, "grad_norm": 0.244140625, "learning_rate": 7.70516717325228e-06, "loss": 0.9838, "step": 755 }, { "epoch": 0.2297872340425532, "grad_norm": 0.26953125, "learning_rate": 7.702127659574469e-06, "loss": 1.11, "step": 756 }, { "epoch": 0.23009118541033435, "grad_norm": 0.279296875, "learning_rate": 7.699088145896657e-06, "loss": 1.071, "step": 757 }, { "epoch": 0.2303951367781155, "grad_norm": 0.255859375, "learning_rate": 7.696048632218846e-06, "loss": 1.0347, "step": 758 }, { "epoch": 0.23069908814589665, "grad_norm": 0.24609375, "learning_rate": 7.693009118541033e-06, "loss": 1.174, "step": 759 }, { "epoch": 0.23100303951367782, "grad_norm": 0.2412109375, "learning_rate": 7.689969604863222e-06, "loss": 1.0932, "step": 760 }, { "epoch": 0.23130699088145898, "grad_norm": 0.2734375, "learning_rate": 7.68693009118541e-06, "loss": 1.1144, "step": 761 }, { "epoch": 0.23161094224924011, "grad_norm": 0.263671875, "learning_rate": 7.683890577507599e-06, "loss": 1.0191, "step": 762 }, { "epoch": 0.23191489361702128, "grad_norm": 0.25390625, "learning_rate": 7.680851063829788e-06, "loss": 1.0689, "step": 763 }, { "epoch": 0.23221884498480244, "grad_norm": 0.263671875, "learning_rate": 7.677811550151977e-06, "loss": 1.0356, "step": 764 }, { "epoch": 0.23252279635258358, "grad_norm": 0.2578125, "learning_rate": 7.674772036474165e-06, "loss": 1.0037, "step": 765 }, { "epoch": 0.23282674772036474, "grad_norm": 0.26171875, "learning_rate": 7.671732522796354e-06, "loss": 1.0622, "step": 766 }, { "epoch": 0.2331306990881459, "grad_norm": 0.26171875, "learning_rate": 7.668693009118543e-06, "loss": 1.0905, "step": 767 }, { "epoch": 0.23343465045592704, "grad_norm": 0.2373046875, "learning_rate": 7.66565349544073e-06, "loss": 1.1765, "step": 768 }, { "epoch": 0.2337386018237082, "grad_norm": 0.259765625, "learning_rate": 7.662613981762918e-06, "loss": 1.0881, "step": 769 }, { "epoch": 0.23404255319148937, "grad_norm": 0.2412109375, "learning_rate": 7.659574468085107e-06, "loss": 1.0542, "step": 770 }, { "epoch": 0.23434650455927053, "grad_norm": 0.267578125, "learning_rate": 7.656534954407296e-06, "loss": 1.0524, "step": 771 }, { "epoch": 0.23465045592705167, "grad_norm": 0.2451171875, "learning_rate": 7.653495440729484e-06, "loss": 1.0641, "step": 772 }, { "epoch": 0.23495440729483283, "grad_norm": 0.25390625, "learning_rate": 7.650455927051671e-06, "loss": 1.1025, "step": 773 }, { "epoch": 0.235258358662614, "grad_norm": 0.244140625, "learning_rate": 7.64741641337386e-06, "loss": 1.0962, "step": 774 }, { "epoch": 0.23556231003039513, "grad_norm": 0.255859375, "learning_rate": 7.644376899696049e-06, "loss": 1.0636, "step": 775 }, { "epoch": 0.2358662613981763, "grad_norm": 0.25, "learning_rate": 7.641337386018237e-06, "loss": 1.0937, "step": 776 }, { "epoch": 0.23617021276595745, "grad_norm": 0.265625, "learning_rate": 7.638297872340426e-06, "loss": 1.1317, "step": 777 }, { "epoch": 0.2364741641337386, "grad_norm": 0.2578125, "learning_rate": 7.635258358662615e-06, "loss": 1.1726, "step": 778 }, { "epoch": 0.23677811550151975, "grad_norm": 0.26171875, "learning_rate": 7.632218844984803e-06, "loss": 1.0404, "step": 779 }, { "epoch": 0.23708206686930092, "grad_norm": 0.244140625, "learning_rate": 7.629179331306992e-06, "loss": 1.1031, "step": 780 }, { "epoch": 0.23738601823708208, "grad_norm": 0.244140625, "learning_rate": 7.62613981762918e-06, "loss": 1.1055, "step": 781 }, { "epoch": 0.23768996960486322, "grad_norm": 0.24609375, "learning_rate": 7.623100303951369e-06, "loss": 1.0813, "step": 782 }, { "epoch": 0.23799392097264438, "grad_norm": 0.2734375, "learning_rate": 7.6200607902735565e-06, "loss": 1.0024, "step": 783 }, { "epoch": 0.23829787234042554, "grad_norm": 0.248046875, "learning_rate": 7.617021276595745e-06, "loss": 1.083, "step": 784 }, { "epoch": 0.23860182370820668, "grad_norm": 0.259765625, "learning_rate": 7.613981762917934e-06, "loss": 1.0999, "step": 785 }, { "epoch": 0.23890577507598784, "grad_norm": 0.2412109375, "learning_rate": 7.610942249240122e-06, "loss": 1.0681, "step": 786 }, { "epoch": 0.239209726443769, "grad_norm": 0.26171875, "learning_rate": 7.60790273556231e-06, "loss": 1.0576, "step": 787 }, { "epoch": 0.23951367781155014, "grad_norm": 0.2421875, "learning_rate": 7.604863221884499e-06, "loss": 1.0553, "step": 788 }, { "epoch": 0.2398176291793313, "grad_norm": 0.24609375, "learning_rate": 7.601823708206687e-06, "loss": 1.0639, "step": 789 }, { "epoch": 0.24012158054711247, "grad_norm": 0.25, "learning_rate": 7.598784194528876e-06, "loss": 1.1072, "step": 790 }, { "epoch": 0.2404255319148936, "grad_norm": 0.2421875, "learning_rate": 7.595744680851065e-06, "loss": 1.0664, "step": 791 }, { "epoch": 0.24072948328267477, "grad_norm": 0.244140625, "learning_rate": 7.592705167173253e-06, "loss": 1.0913, "step": 792 }, { "epoch": 0.24103343465045593, "grad_norm": 0.2451171875, "learning_rate": 7.589665653495442e-06, "loss": 1.144, "step": 793 }, { "epoch": 0.2413373860182371, "grad_norm": 0.3046875, "learning_rate": 7.58662613981763e-06, "loss": 1.0801, "step": 794 }, { "epoch": 0.24164133738601823, "grad_norm": 0.2578125, "learning_rate": 7.583586626139818e-06, "loss": 1.0419, "step": 795 }, { "epoch": 0.2419452887537994, "grad_norm": 0.255859375, "learning_rate": 7.580547112462007e-06, "loss": 1.0554, "step": 796 }, { "epoch": 0.24224924012158056, "grad_norm": 0.234375, "learning_rate": 7.577507598784195e-06, "loss": 1.0771, "step": 797 }, { "epoch": 0.2425531914893617, "grad_norm": 0.2392578125, "learning_rate": 7.574468085106383e-06, "loss": 1.1114, "step": 798 }, { "epoch": 0.24285714285714285, "grad_norm": 0.255859375, "learning_rate": 7.571428571428572e-06, "loss": 1.13, "step": 799 }, { "epoch": 0.24316109422492402, "grad_norm": 0.28515625, "learning_rate": 7.56838905775076e-06, "loss": 1.1208, "step": 800 }, { "epoch": 0.24346504559270515, "grad_norm": 0.2470703125, "learning_rate": 7.5653495440729485e-06, "loss": 1.079, "step": 801 }, { "epoch": 0.24376899696048632, "grad_norm": 0.255859375, "learning_rate": 7.562310030395137e-06, "loss": 1.1281, "step": 802 }, { "epoch": 0.24407294832826748, "grad_norm": 0.2578125, "learning_rate": 7.559270516717327e-06, "loss": 1.0784, "step": 803 }, { "epoch": 0.24437689969604864, "grad_norm": 0.259765625, "learning_rate": 7.556231003039515e-06, "loss": 1.127, "step": 804 }, { "epoch": 0.24468085106382978, "grad_norm": 0.267578125, "learning_rate": 7.553191489361703e-06, "loss": 1.0495, "step": 805 }, { "epoch": 0.24498480243161094, "grad_norm": 0.24609375, "learning_rate": 7.550151975683891e-06, "loss": 1.0746, "step": 806 }, { "epoch": 0.2452887537993921, "grad_norm": 0.248046875, "learning_rate": 7.54711246200608e-06, "loss": 1.1173, "step": 807 }, { "epoch": 0.24559270516717324, "grad_norm": 0.2470703125, "learning_rate": 7.5440729483282685e-06, "loss": 1.0287, "step": 808 }, { "epoch": 0.2458966565349544, "grad_norm": 0.248046875, "learning_rate": 7.541033434650456e-06, "loss": 1.1524, "step": 809 }, { "epoch": 0.24620060790273557, "grad_norm": 0.2490234375, "learning_rate": 7.537993920972645e-06, "loss": 1.077, "step": 810 }, { "epoch": 0.2465045592705167, "grad_norm": 0.326171875, "learning_rate": 7.534954407294834e-06, "loss": 1.1136, "step": 811 }, { "epoch": 0.24680851063829787, "grad_norm": 0.251953125, "learning_rate": 7.5319148936170215e-06, "loss": 1.0987, "step": 812 }, { "epoch": 0.24711246200607903, "grad_norm": 0.2431640625, "learning_rate": 7.52887537993921e-06, "loss": 1.1047, "step": 813 }, { "epoch": 0.2474164133738602, "grad_norm": 0.259765625, "learning_rate": 7.525835866261398e-06, "loss": 1.1077, "step": 814 }, { "epoch": 0.24772036474164133, "grad_norm": 0.25, "learning_rate": 7.522796352583587e-06, "loss": 1.1244, "step": 815 }, { "epoch": 0.2480243161094225, "grad_norm": 0.2734375, "learning_rate": 7.519756838905776e-06, "loss": 1.0404, "step": 816 }, { "epoch": 0.24832826747720366, "grad_norm": 0.2431640625, "learning_rate": 7.516717325227965e-06, "loss": 1.1003, "step": 817 }, { "epoch": 0.2486322188449848, "grad_norm": 0.248046875, "learning_rate": 7.513677811550153e-06, "loss": 1.1279, "step": 818 }, { "epoch": 0.24893617021276596, "grad_norm": 0.251953125, "learning_rate": 7.5106382978723415e-06, "loss": 1.146, "step": 819 }, { "epoch": 0.24924012158054712, "grad_norm": 0.23828125, "learning_rate": 7.507598784194529e-06, "loss": 1.1373, "step": 820 }, { "epoch": 0.24954407294832825, "grad_norm": 0.2431640625, "learning_rate": 7.504559270516718e-06, "loss": 1.049, "step": 821 }, { "epoch": 0.24984802431610942, "grad_norm": 0.255859375, "learning_rate": 7.501519756838907e-06, "loss": 1.1164, "step": 822 }, { "epoch": 0.2501519756838906, "grad_norm": 0.2451171875, "learning_rate": 7.4984802431610945e-06, "loss": 1.0161, "step": 823 }, { "epoch": 0.25045592705167175, "grad_norm": 0.25390625, "learning_rate": 7.495440729483283e-06, "loss": 1.0972, "step": 824 }, { "epoch": 0.2507598784194529, "grad_norm": 0.259765625, "learning_rate": 7.492401215805472e-06, "loss": 1.1938, "step": 825 }, { "epoch": 0.251063829787234, "grad_norm": 0.267578125, "learning_rate": 7.48936170212766e-06, "loss": 1.18, "step": 826 }, { "epoch": 0.2513677811550152, "grad_norm": 0.263671875, "learning_rate": 7.4863221884498484e-06, "loss": 1.0438, "step": 827 }, { "epoch": 0.25167173252279634, "grad_norm": 0.251953125, "learning_rate": 7.483282674772036e-06, "loss": 0.9638, "step": 828 }, { "epoch": 0.2519756838905775, "grad_norm": 0.251953125, "learning_rate": 7.480243161094225e-06, "loss": 1.0625, "step": 829 }, { "epoch": 0.25227963525835867, "grad_norm": 0.31640625, "learning_rate": 7.4772036474164145e-06, "loss": 1.0148, "step": 830 }, { "epoch": 0.25258358662613983, "grad_norm": 0.251953125, "learning_rate": 7.474164133738603e-06, "loss": 1.0959, "step": 831 }, { "epoch": 0.252887537993921, "grad_norm": 0.2734375, "learning_rate": 7.471124620060791e-06, "loss": 0.9821, "step": 832 }, { "epoch": 0.2531914893617021, "grad_norm": 0.255859375, "learning_rate": 7.46808510638298e-06, "loss": 1.1165, "step": 833 }, { "epoch": 0.25349544072948327, "grad_norm": 0.255859375, "learning_rate": 7.4650455927051675e-06, "loss": 1.083, "step": 834 }, { "epoch": 0.25379939209726443, "grad_norm": 0.2470703125, "learning_rate": 7.462006079027356e-06, "loss": 1.1096, "step": 835 }, { "epoch": 0.2541033434650456, "grad_norm": 0.2431640625, "learning_rate": 7.458966565349545e-06, "loss": 1.0479, "step": 836 }, { "epoch": 0.25440729483282676, "grad_norm": 0.255859375, "learning_rate": 7.455927051671733e-06, "loss": 1.0684, "step": 837 }, { "epoch": 0.2547112462006079, "grad_norm": 0.263671875, "learning_rate": 7.452887537993921e-06, "loss": 1.099, "step": 838 }, { "epoch": 0.2550151975683891, "grad_norm": 0.267578125, "learning_rate": 7.44984802431611e-06, "loss": 1.1013, "step": 839 }, { "epoch": 0.2553191489361702, "grad_norm": 0.375, "learning_rate": 7.446808510638298e-06, "loss": 1.084, "step": 840 }, { "epoch": 0.25562310030395136, "grad_norm": 0.26171875, "learning_rate": 7.443768996960487e-06, "loss": 1.0806, "step": 841 }, { "epoch": 0.2559270516717325, "grad_norm": 0.2578125, "learning_rate": 7.4407294832826745e-06, "loss": 1.1255, "step": 842 }, { "epoch": 0.2562310030395137, "grad_norm": 0.357421875, "learning_rate": 7.437689969604864e-06, "loss": 1.062, "step": 843 }, { "epoch": 0.25653495440729485, "grad_norm": 0.255859375, "learning_rate": 7.434650455927053e-06, "loss": 1.0091, "step": 844 }, { "epoch": 0.256838905775076, "grad_norm": 0.2578125, "learning_rate": 7.431610942249241e-06, "loss": 1.0728, "step": 845 }, { "epoch": 0.2571428571428571, "grad_norm": 0.248046875, "learning_rate": 7.428571428571429e-06, "loss": 1.0571, "step": 846 }, { "epoch": 0.2574468085106383, "grad_norm": 0.306640625, "learning_rate": 7.425531914893618e-06, "loss": 1.0659, "step": 847 }, { "epoch": 0.25775075987841944, "grad_norm": 0.26953125, "learning_rate": 7.4224924012158066e-06, "loss": 1.0636, "step": 848 }, { "epoch": 0.2580547112462006, "grad_norm": 0.2734375, "learning_rate": 7.419452887537994e-06, "loss": 1.0423, "step": 849 }, { "epoch": 0.25835866261398177, "grad_norm": 0.248046875, "learning_rate": 7.416413373860183e-06, "loss": 1.1502, "step": 850 }, { "epoch": 0.25866261398176293, "grad_norm": 0.248046875, "learning_rate": 7.413373860182371e-06, "loss": 1.1628, "step": 851 }, { "epoch": 0.2589665653495441, "grad_norm": 0.248046875, "learning_rate": 7.41033434650456e-06, "loss": 0.9826, "step": 852 }, { "epoch": 0.2592705167173252, "grad_norm": 0.27734375, "learning_rate": 7.407294832826748e-06, "loss": 0.9675, "step": 853 }, { "epoch": 0.25957446808510637, "grad_norm": 0.2578125, "learning_rate": 7.404255319148936e-06, "loss": 1.0753, "step": 854 }, { "epoch": 0.25987841945288753, "grad_norm": 0.248046875, "learning_rate": 7.401215805471125e-06, "loss": 1.1001, "step": 855 }, { "epoch": 0.2601823708206687, "grad_norm": 0.2470703125, "learning_rate": 7.398176291793314e-06, "loss": 1.1137, "step": 856 }, { "epoch": 0.26048632218844986, "grad_norm": 0.263671875, "learning_rate": 7.395136778115502e-06, "loss": 1.1134, "step": 857 }, { "epoch": 0.260790273556231, "grad_norm": 0.255859375, "learning_rate": 7.392097264437691e-06, "loss": 1.1059, "step": 858 }, { "epoch": 0.26109422492401213, "grad_norm": 0.26953125, "learning_rate": 7.3890577507598796e-06, "loss": 1.0569, "step": 859 }, { "epoch": 0.2613981762917933, "grad_norm": 0.25390625, "learning_rate": 7.386018237082067e-06, "loss": 1.0729, "step": 860 }, { "epoch": 0.26170212765957446, "grad_norm": 0.25390625, "learning_rate": 7.382978723404256e-06, "loss": 1.071, "step": 861 }, { "epoch": 0.2620060790273556, "grad_norm": 0.251953125, "learning_rate": 7.379939209726445e-06, "loss": 1.0465, "step": 862 }, { "epoch": 0.2623100303951368, "grad_norm": 0.26171875, "learning_rate": 7.376899696048633e-06, "loss": 1.1018, "step": 863 }, { "epoch": 0.26261398176291795, "grad_norm": 0.25, "learning_rate": 7.373860182370821e-06, "loss": 1.091, "step": 864 }, { "epoch": 0.2629179331306991, "grad_norm": 0.251953125, "learning_rate": 7.370820668693009e-06, "loss": 1.1419, "step": 865 }, { "epoch": 0.2632218844984802, "grad_norm": 0.244140625, "learning_rate": 7.367781155015198e-06, "loss": 1.0874, "step": 866 }, { "epoch": 0.2635258358662614, "grad_norm": 0.2412109375, "learning_rate": 7.3647416413373865e-06, "loss": 1.1287, "step": 867 }, { "epoch": 0.26382978723404255, "grad_norm": 0.248046875, "learning_rate": 7.361702127659574e-06, "loss": 1.0643, "step": 868 }, { "epoch": 0.2641337386018237, "grad_norm": 0.2470703125, "learning_rate": 7.358662613981764e-06, "loss": 1.0841, "step": 869 }, { "epoch": 0.26443768996960487, "grad_norm": 0.27734375, "learning_rate": 7.3556231003039526e-06, "loss": 1.0063, "step": 870 }, { "epoch": 0.26474164133738604, "grad_norm": 0.255859375, "learning_rate": 7.35258358662614e-06, "loss": 1.125, "step": 871 }, { "epoch": 0.2650455927051672, "grad_norm": 0.244140625, "learning_rate": 7.349544072948329e-06, "loss": 1.062, "step": 872 }, { "epoch": 0.2653495440729483, "grad_norm": 0.29296875, "learning_rate": 7.346504559270518e-06, "loss": 1.0278, "step": 873 }, { "epoch": 0.26565349544072947, "grad_norm": 0.26171875, "learning_rate": 7.343465045592706e-06, "loss": 1.134, "step": 874 }, { "epoch": 0.26595744680851063, "grad_norm": 0.25, "learning_rate": 7.340425531914894e-06, "loss": 1.0941, "step": 875 }, { "epoch": 0.2662613981762918, "grad_norm": 0.255859375, "learning_rate": 7.337386018237083e-06, "loss": 1.0829, "step": 876 }, { "epoch": 0.26656534954407296, "grad_norm": 0.24609375, "learning_rate": 7.334346504559271e-06, "loss": 1.0723, "step": 877 }, { "epoch": 0.2668693009118541, "grad_norm": 0.2490234375, "learning_rate": 7.3313069908814595e-06, "loss": 1.0592, "step": 878 }, { "epoch": 0.26717325227963523, "grad_norm": 0.255859375, "learning_rate": 7.328267477203647e-06, "loss": 1.0839, "step": 879 }, { "epoch": 0.2674772036474164, "grad_norm": 0.265625, "learning_rate": 7.325227963525836e-06, "loss": 1.0745, "step": 880 }, { "epoch": 0.26778115501519756, "grad_norm": 0.439453125, "learning_rate": 7.322188449848025e-06, "loss": 1.0194, "step": 881 }, { "epoch": 0.2680851063829787, "grad_norm": 0.2578125, "learning_rate": 7.3191489361702125e-06, "loss": 1.0223, "step": 882 }, { "epoch": 0.2683890577507599, "grad_norm": 0.265625, "learning_rate": 7.316109422492402e-06, "loss": 1.1026, "step": 883 }, { "epoch": 0.26869300911854105, "grad_norm": 0.2578125, "learning_rate": 7.313069908814591e-06, "loss": 1.1272, "step": 884 }, { "epoch": 0.2689969604863222, "grad_norm": 0.265625, "learning_rate": 7.3100303951367794e-06, "loss": 1.1224, "step": 885 }, { "epoch": 0.2693009118541033, "grad_norm": 0.255859375, "learning_rate": 7.306990881458967e-06, "loss": 1.0879, "step": 886 }, { "epoch": 0.2696048632218845, "grad_norm": 0.2431640625, "learning_rate": 7.303951367781156e-06, "loss": 1.1601, "step": 887 }, { "epoch": 0.26990881458966565, "grad_norm": 0.251953125, "learning_rate": 7.300911854103344e-06, "loss": 1.0319, "step": 888 }, { "epoch": 0.2702127659574468, "grad_norm": 0.24609375, "learning_rate": 7.2978723404255325e-06, "loss": 1.1199, "step": 889 }, { "epoch": 0.270516717325228, "grad_norm": 0.2578125, "learning_rate": 7.294832826747721e-06, "loss": 1.0771, "step": 890 }, { "epoch": 0.27082066869300914, "grad_norm": 0.41015625, "learning_rate": 7.291793313069909e-06, "loss": 1.0288, "step": 891 }, { "epoch": 0.2711246200607903, "grad_norm": 0.2490234375, "learning_rate": 7.288753799392098e-06, "loss": 1.1024, "step": 892 }, { "epoch": 0.2714285714285714, "grad_norm": 0.26171875, "learning_rate": 7.285714285714286e-06, "loss": 1.0911, "step": 893 }, { "epoch": 0.27173252279635257, "grad_norm": 0.26171875, "learning_rate": 7.282674772036474e-06, "loss": 1.0637, "step": 894 }, { "epoch": 0.27203647416413373, "grad_norm": 0.25390625, "learning_rate": 7.279635258358663e-06, "loss": 1.1337, "step": 895 }, { "epoch": 0.2723404255319149, "grad_norm": 0.2578125, "learning_rate": 7.2765957446808524e-06, "loss": 1.0743, "step": 896 }, { "epoch": 0.27264437689969606, "grad_norm": 0.5078125, "learning_rate": 7.27355623100304e-06, "loss": 1.0642, "step": 897 }, { "epoch": 0.2729483282674772, "grad_norm": 0.322265625, "learning_rate": 7.270516717325229e-06, "loss": 1.1632, "step": 898 }, { "epoch": 0.27325227963525833, "grad_norm": 0.283203125, "learning_rate": 7.267477203647418e-06, "loss": 1.002, "step": 899 }, { "epoch": 0.2735562310030395, "grad_norm": 0.265625, "learning_rate": 7.2644376899696055e-06, "loss": 1.1679, "step": 900 }, { "epoch": 0.27386018237082066, "grad_norm": 0.26171875, "learning_rate": 7.261398176291794e-06, "loss": 1.0293, "step": 901 }, { "epoch": 0.2741641337386018, "grad_norm": 0.2578125, "learning_rate": 7.258358662613982e-06, "loss": 1.0753, "step": 902 }, { "epoch": 0.274468085106383, "grad_norm": 0.265625, "learning_rate": 7.255319148936171e-06, "loss": 1.0356, "step": 903 }, { "epoch": 0.27477203647416415, "grad_norm": 0.259765625, "learning_rate": 7.252279635258359e-06, "loss": 1.0512, "step": 904 }, { "epoch": 0.2750759878419453, "grad_norm": 0.255859375, "learning_rate": 7.249240121580547e-06, "loss": 1.0158, "step": 905 }, { "epoch": 0.2753799392097264, "grad_norm": 0.255859375, "learning_rate": 7.246200607902736e-06, "loss": 1.0467, "step": 906 }, { "epoch": 0.2756838905775076, "grad_norm": 0.2490234375, "learning_rate": 7.243161094224925e-06, "loss": 1.0785, "step": 907 }, { "epoch": 0.27598784194528875, "grad_norm": 0.26171875, "learning_rate": 7.240121580547112e-06, "loss": 1.0815, "step": 908 }, { "epoch": 0.2762917933130699, "grad_norm": 0.267578125, "learning_rate": 7.237082066869302e-06, "loss": 1.0676, "step": 909 }, { "epoch": 0.2765957446808511, "grad_norm": 0.265625, "learning_rate": 7.234042553191491e-06, "loss": 1.0478, "step": 910 }, { "epoch": 0.27689969604863224, "grad_norm": 0.265625, "learning_rate": 7.2310030395136785e-06, "loss": 1.1263, "step": 911 }, { "epoch": 0.2772036474164134, "grad_norm": 0.2734375, "learning_rate": 7.227963525835867e-06, "loss": 1.2066, "step": 912 }, { "epoch": 0.2775075987841945, "grad_norm": 0.26171875, "learning_rate": 7.224924012158056e-06, "loss": 1.0854, "step": 913 }, { "epoch": 0.2778115501519757, "grad_norm": 0.251953125, "learning_rate": 7.221884498480244e-06, "loss": 0.9851, "step": 914 }, { "epoch": 0.27811550151975684, "grad_norm": 0.2578125, "learning_rate": 7.218844984802432e-06, "loss": 1.0555, "step": 915 }, { "epoch": 0.278419452887538, "grad_norm": 0.26171875, "learning_rate": 7.21580547112462e-06, "loss": 1.0573, "step": 916 }, { "epoch": 0.27872340425531916, "grad_norm": 0.2890625, "learning_rate": 7.212765957446809e-06, "loss": 1.1133, "step": 917 }, { "epoch": 0.2790273556231003, "grad_norm": 0.259765625, "learning_rate": 7.209726443768998e-06, "loss": 1.0961, "step": 918 }, { "epoch": 0.27933130699088143, "grad_norm": 0.25390625, "learning_rate": 7.206686930091185e-06, "loss": 1.0775, "step": 919 }, { "epoch": 0.2796352583586626, "grad_norm": 0.244140625, "learning_rate": 7.203647416413374e-06, "loss": 1.1109, "step": 920 }, { "epoch": 0.27993920972644376, "grad_norm": 0.2578125, "learning_rate": 7.200607902735563e-06, "loss": 1.1333, "step": 921 }, { "epoch": 0.2802431610942249, "grad_norm": 0.400390625, "learning_rate": 7.197568389057752e-06, "loss": 1.0539, "step": 922 }, { "epoch": 0.2805471124620061, "grad_norm": 0.2578125, "learning_rate": 7.19452887537994e-06, "loss": 1.0221, "step": 923 }, { "epoch": 0.28085106382978725, "grad_norm": 0.255859375, "learning_rate": 7.191489361702129e-06, "loss": 1.0713, "step": 924 }, { "epoch": 0.2811550151975684, "grad_norm": 0.28515625, "learning_rate": 7.188449848024317e-06, "loss": 1.0551, "step": 925 }, { "epoch": 0.2814589665653495, "grad_norm": 0.279296875, "learning_rate": 7.185410334346505e-06, "loss": 1.0985, "step": 926 }, { "epoch": 0.2817629179331307, "grad_norm": 0.333984375, "learning_rate": 7.182370820668694e-06, "loss": 1.0879, "step": 927 }, { "epoch": 0.28206686930091185, "grad_norm": 0.259765625, "learning_rate": 7.179331306990882e-06, "loss": 1.0171, "step": 928 }, { "epoch": 0.282370820668693, "grad_norm": 0.24609375, "learning_rate": 7.1762917933130706e-06, "loss": 1.0963, "step": 929 }, { "epoch": 0.2826747720364742, "grad_norm": 0.25, "learning_rate": 7.173252279635258e-06, "loss": 1.0366, "step": 930 }, { "epoch": 0.28297872340425534, "grad_norm": 0.25390625, "learning_rate": 7.170212765957447e-06, "loss": 1.1719, "step": 931 }, { "epoch": 0.28328267477203645, "grad_norm": 0.2578125, "learning_rate": 7.167173252279636e-06, "loss": 1.1352, "step": 932 }, { "epoch": 0.2835866261398176, "grad_norm": 0.267578125, "learning_rate": 7.164133738601824e-06, "loss": 1.0695, "step": 933 }, { "epoch": 0.2838905775075988, "grad_norm": 0.2470703125, "learning_rate": 7.161094224924012e-06, "loss": 1.0693, "step": 934 }, { "epoch": 0.28419452887537994, "grad_norm": 0.255859375, "learning_rate": 7.158054711246201e-06, "loss": 1.0725, "step": 935 }, { "epoch": 0.2844984802431611, "grad_norm": 0.265625, "learning_rate": 7.1550151975683905e-06, "loss": 1.1367, "step": 936 }, { "epoch": 0.28480243161094226, "grad_norm": 0.265625, "learning_rate": 7.151975683890578e-06, "loss": 1.0615, "step": 937 }, { "epoch": 0.2851063829787234, "grad_norm": 0.255859375, "learning_rate": 7.148936170212767e-06, "loss": 1.0785, "step": 938 }, { "epoch": 0.28541033434650454, "grad_norm": 0.275390625, "learning_rate": 7.145896656534955e-06, "loss": 1.0268, "step": 939 }, { "epoch": 0.2857142857142857, "grad_norm": 0.271484375, "learning_rate": 7.1428571428571436e-06, "loss": 1.0574, "step": 940 }, { "epoch": 0.28601823708206686, "grad_norm": 0.251953125, "learning_rate": 7.139817629179332e-06, "loss": 1.0767, "step": 941 }, { "epoch": 0.286322188449848, "grad_norm": 0.2578125, "learning_rate": 7.13677811550152e-06, "loss": 0.9763, "step": 942 }, { "epoch": 0.2866261398176292, "grad_norm": 0.26953125, "learning_rate": 7.133738601823709e-06, "loss": 1.113, "step": 943 }, { "epoch": 0.28693009118541035, "grad_norm": 0.24609375, "learning_rate": 7.1306990881458975e-06, "loss": 1.0558, "step": 944 }, { "epoch": 0.2872340425531915, "grad_norm": 0.26171875, "learning_rate": 7.127659574468085e-06, "loss": 1.1149, "step": 945 }, { "epoch": 0.2875379939209726, "grad_norm": 0.26171875, "learning_rate": 7.124620060790274e-06, "loss": 1.0489, "step": 946 }, { "epoch": 0.2878419452887538, "grad_norm": 0.251953125, "learning_rate": 7.121580547112462e-06, "loss": 1.0302, "step": 947 }, { "epoch": 0.28814589665653495, "grad_norm": 0.265625, "learning_rate": 7.1185410334346505e-06, "loss": 1.0035, "step": 948 }, { "epoch": 0.2884498480243161, "grad_norm": 0.294921875, "learning_rate": 7.11550151975684e-06, "loss": 1.0012, "step": 949 }, { "epoch": 0.2887537993920973, "grad_norm": 0.283203125, "learning_rate": 7.112462006079029e-06, "loss": 1.1044, "step": 950 }, { "epoch": 0.28905775075987844, "grad_norm": 0.263671875, "learning_rate": 7.1094224924012166e-06, "loss": 1.0772, "step": 951 }, { "epoch": 0.28936170212765955, "grad_norm": 0.25390625, "learning_rate": 7.106382978723405e-06, "loss": 1.0433, "step": 952 }, { "epoch": 0.2896656534954407, "grad_norm": 0.251953125, "learning_rate": 7.103343465045593e-06, "loss": 1.1211, "step": 953 }, { "epoch": 0.2899696048632219, "grad_norm": 0.255859375, "learning_rate": 7.100303951367782e-06, "loss": 1.0531, "step": 954 }, { "epoch": 0.29027355623100304, "grad_norm": 0.2578125, "learning_rate": 7.0972644376899704e-06, "loss": 1.0975, "step": 955 }, { "epoch": 0.2905775075987842, "grad_norm": 0.26953125, "learning_rate": 7.094224924012158e-06, "loss": 1.0573, "step": 956 }, { "epoch": 0.29088145896656536, "grad_norm": 0.26171875, "learning_rate": 7.091185410334347e-06, "loss": 1.0418, "step": 957 }, { "epoch": 0.29118541033434653, "grad_norm": 0.30859375, "learning_rate": 7.088145896656536e-06, "loss": 1.0671, "step": 958 }, { "epoch": 0.29148936170212764, "grad_norm": 0.265625, "learning_rate": 7.0851063829787235e-06, "loss": 1.0594, "step": 959 }, { "epoch": 0.2917933130699088, "grad_norm": 0.2451171875, "learning_rate": 7.082066869300912e-06, "loss": 1.1177, "step": 960 }, { "epoch": 0.29209726443768996, "grad_norm": 0.2734375, "learning_rate": 7.0790273556231e-06, "loss": 1.0974, "step": 961 }, { "epoch": 0.2924012158054711, "grad_norm": 0.337890625, "learning_rate": 7.0759878419452895e-06, "loss": 1.012, "step": 962 }, { "epoch": 0.2927051671732523, "grad_norm": 0.263671875, "learning_rate": 7.072948328267478e-06, "loss": 1.1246, "step": 963 }, { "epoch": 0.29300911854103345, "grad_norm": 0.298828125, "learning_rate": 7.069908814589667e-06, "loss": 1.012, "step": 964 }, { "epoch": 0.2933130699088146, "grad_norm": 0.25390625, "learning_rate": 7.066869300911855e-06, "loss": 1.1149, "step": 965 }, { "epoch": 0.2936170212765957, "grad_norm": 0.265625, "learning_rate": 7.0638297872340434e-06, "loss": 1.0149, "step": 966 }, { "epoch": 0.2939209726443769, "grad_norm": 0.251953125, "learning_rate": 7.060790273556231e-06, "loss": 1.0711, "step": 967 }, { "epoch": 0.29422492401215805, "grad_norm": 0.259765625, "learning_rate": 7.05775075987842e-06, "loss": 1.0263, "step": 968 }, { "epoch": 0.2945288753799392, "grad_norm": 0.2734375, "learning_rate": 7.054711246200609e-06, "loss": 1.0123, "step": 969 }, { "epoch": 0.2948328267477204, "grad_norm": 0.255859375, "learning_rate": 7.0516717325227965e-06, "loss": 1.0957, "step": 970 }, { "epoch": 0.29513677811550154, "grad_norm": 0.267578125, "learning_rate": 7.048632218844985e-06, "loss": 1.1011, "step": 971 }, { "epoch": 0.29544072948328265, "grad_norm": 0.27734375, "learning_rate": 7.045592705167174e-06, "loss": 1.0547, "step": 972 }, { "epoch": 0.2957446808510638, "grad_norm": 0.265625, "learning_rate": 7.042553191489362e-06, "loss": 1.0826, "step": 973 }, { "epoch": 0.296048632218845, "grad_norm": 0.251953125, "learning_rate": 7.03951367781155e-06, "loss": 1.0405, "step": 974 }, { "epoch": 0.29635258358662614, "grad_norm": 0.2490234375, "learning_rate": 7.036474164133738e-06, "loss": 1.0685, "step": 975 }, { "epoch": 0.2966565349544073, "grad_norm": 0.26953125, "learning_rate": 7.033434650455928e-06, "loss": 1.1103, "step": 976 }, { "epoch": 0.29696048632218847, "grad_norm": 0.251953125, "learning_rate": 7.0303951367781164e-06, "loss": 1.0833, "step": 977 }, { "epoch": 0.29726443768996963, "grad_norm": 0.26953125, "learning_rate": 7.027355623100305e-06, "loss": 1.0755, "step": 978 }, { "epoch": 0.29756838905775074, "grad_norm": 0.26953125, "learning_rate": 7.024316109422493e-06, "loss": 1.0986, "step": 979 }, { "epoch": 0.2978723404255319, "grad_norm": 0.2578125, "learning_rate": 7.021276595744682e-06, "loss": 1.0793, "step": 980 }, { "epoch": 0.29817629179331306, "grad_norm": 0.26953125, "learning_rate": 7.01823708206687e-06, "loss": 1.0548, "step": 981 }, { "epoch": 0.2984802431610942, "grad_norm": 0.31640625, "learning_rate": 7.015197568389058e-06, "loss": 1.0812, "step": 982 }, { "epoch": 0.2987841945288754, "grad_norm": 0.271484375, "learning_rate": 7.012158054711247e-06, "loss": 1.0519, "step": 983 }, { "epoch": 0.29908814589665655, "grad_norm": 0.259765625, "learning_rate": 7.009118541033435e-06, "loss": 1.1047, "step": 984 }, { "epoch": 0.2993920972644377, "grad_norm": 0.25390625, "learning_rate": 7.006079027355623e-06, "loss": 1.0921, "step": 985 }, { "epoch": 0.2996960486322188, "grad_norm": 0.26171875, "learning_rate": 7.003039513677812e-06, "loss": 1.139, "step": 986 }, { "epoch": 0.3, "grad_norm": 0.25390625, "learning_rate": 7e-06, "loss": 1.0366, "step": 987 }, { "epoch": 0.30030395136778115, "grad_norm": 0.267578125, "learning_rate": 6.996960486322189e-06, "loss": 1.0499, "step": 988 }, { "epoch": 0.3006079027355623, "grad_norm": 0.2578125, "learning_rate": 6.993920972644378e-06, "loss": 1.1328, "step": 989 }, { "epoch": 0.3009118541033435, "grad_norm": 0.279296875, "learning_rate": 6.990881458966566e-06, "loss": 1.0266, "step": 990 }, { "epoch": 0.30121580547112464, "grad_norm": 0.251953125, "learning_rate": 6.987841945288755e-06, "loss": 1.1265, "step": 991 }, { "epoch": 0.30151975683890575, "grad_norm": 0.26171875, "learning_rate": 6.984802431610943e-06, "loss": 1.0217, "step": 992 }, { "epoch": 0.3018237082066869, "grad_norm": 0.267578125, "learning_rate": 6.981762917933131e-06, "loss": 1.0872, "step": 993 }, { "epoch": 0.3021276595744681, "grad_norm": 0.2890625, "learning_rate": 6.97872340425532e-06, "loss": 1.0288, "step": 994 }, { "epoch": 0.30243161094224924, "grad_norm": 0.29296875, "learning_rate": 6.9756838905775085e-06, "loss": 1.0147, "step": 995 }, { "epoch": 0.3027355623100304, "grad_norm": 0.263671875, "learning_rate": 6.972644376899696e-06, "loss": 1.0955, "step": 996 }, { "epoch": 0.30303951367781157, "grad_norm": 0.251953125, "learning_rate": 6.969604863221885e-06, "loss": 1.1061, "step": 997 }, { "epoch": 0.30334346504559273, "grad_norm": 0.26171875, "learning_rate": 6.966565349544073e-06, "loss": 1.0774, "step": 998 }, { "epoch": 0.30364741641337384, "grad_norm": 0.263671875, "learning_rate": 6.9635258358662616e-06, "loss": 1.0695, "step": 999 }, { "epoch": 0.303951367781155, "grad_norm": 0.267578125, "learning_rate": 6.96048632218845e-06, "loss": 1.0666, "step": 1000 }, { "epoch": 0.30425531914893617, "grad_norm": 0.267578125, "learning_rate": 6.957446808510638e-06, "loss": 1.1522, "step": 1001 }, { "epoch": 0.30455927051671733, "grad_norm": 0.259765625, "learning_rate": 6.954407294832828e-06, "loss": 1.1058, "step": 1002 }, { "epoch": 0.3048632218844985, "grad_norm": 0.294921875, "learning_rate": 6.951367781155016e-06, "loss": 1.0718, "step": 1003 }, { "epoch": 0.30516717325227966, "grad_norm": 0.251953125, "learning_rate": 6.948328267477204e-06, "loss": 1.0986, "step": 1004 }, { "epoch": 0.30547112462006076, "grad_norm": 0.2578125, "learning_rate": 6.945288753799393e-06, "loss": 1.0829, "step": 1005 }, { "epoch": 0.3057750759878419, "grad_norm": 0.25390625, "learning_rate": 6.9422492401215815e-06, "loss": 1.1523, "step": 1006 }, { "epoch": 0.3060790273556231, "grad_norm": 0.2578125, "learning_rate": 6.939209726443769e-06, "loss": 1.1753, "step": 1007 }, { "epoch": 0.30638297872340425, "grad_norm": 0.251953125, "learning_rate": 6.936170212765958e-06, "loss": 1.0904, "step": 1008 }, { "epoch": 0.3066869300911854, "grad_norm": 0.310546875, "learning_rate": 6.933130699088147e-06, "loss": 1.0118, "step": 1009 }, { "epoch": 0.3069908814589666, "grad_norm": 0.265625, "learning_rate": 6.9300911854103346e-06, "loss": 1.1277, "step": 1010 }, { "epoch": 0.30729483282674774, "grad_norm": 0.26953125, "learning_rate": 6.927051671732523e-06, "loss": 1.0454, "step": 1011 }, { "epoch": 0.30759878419452885, "grad_norm": 0.2578125, "learning_rate": 6.924012158054711e-06, "loss": 1.0354, "step": 1012 }, { "epoch": 0.30790273556231, "grad_norm": 0.26953125, "learning_rate": 6.9209726443769e-06, "loss": 1.0711, "step": 1013 }, { "epoch": 0.3082066869300912, "grad_norm": 0.28515625, "learning_rate": 6.9179331306990885e-06, "loss": 1.0951, "step": 1014 }, { "epoch": 0.30851063829787234, "grad_norm": 0.259765625, "learning_rate": 6.914893617021278e-06, "loss": 1.1314, "step": 1015 }, { "epoch": 0.3088145896656535, "grad_norm": 0.26171875, "learning_rate": 6.911854103343466e-06, "loss": 1.1691, "step": 1016 }, { "epoch": 0.30911854103343467, "grad_norm": 0.2470703125, "learning_rate": 6.9088145896656545e-06, "loss": 1.0812, "step": 1017 }, { "epoch": 0.30942249240121583, "grad_norm": 0.259765625, "learning_rate": 6.905775075987843e-06, "loss": 1.0538, "step": 1018 }, { "epoch": 0.30972644376899694, "grad_norm": 0.251953125, "learning_rate": 6.902735562310031e-06, "loss": 1.0825, "step": 1019 }, { "epoch": 0.3100303951367781, "grad_norm": 0.296875, "learning_rate": 6.89969604863222e-06, "loss": 1.0874, "step": 1020 }, { "epoch": 0.31033434650455927, "grad_norm": 0.26953125, "learning_rate": 6.8966565349544076e-06, "loss": 1.0433, "step": 1021 }, { "epoch": 0.31063829787234043, "grad_norm": 0.271484375, "learning_rate": 6.893617021276596e-06, "loss": 1.126, "step": 1022 }, { "epoch": 0.3109422492401216, "grad_norm": 0.2578125, "learning_rate": 6.890577507598785e-06, "loss": 1.0707, "step": 1023 }, { "epoch": 0.31124620060790276, "grad_norm": 0.259765625, "learning_rate": 6.887537993920973e-06, "loss": 1.0529, "step": 1024 }, { "epoch": 0.31155015197568386, "grad_norm": 0.263671875, "learning_rate": 6.8844984802431615e-06, "loss": 1.1355, "step": 1025 }, { "epoch": 0.31185410334346503, "grad_norm": 0.26171875, "learning_rate": 6.88145896656535e-06, "loss": 1.0441, "step": 1026 }, { "epoch": 0.3121580547112462, "grad_norm": 0.263671875, "learning_rate": 6.878419452887538e-06, "loss": 1.0633, "step": 1027 }, { "epoch": 0.31246200607902735, "grad_norm": 0.3125, "learning_rate": 6.875379939209727e-06, "loss": 1.0968, "step": 1028 }, { "epoch": 0.3127659574468085, "grad_norm": 0.28125, "learning_rate": 6.872340425531916e-06, "loss": 1.0812, "step": 1029 }, { "epoch": 0.3130699088145897, "grad_norm": 0.251953125, "learning_rate": 6.869300911854104e-06, "loss": 1.0936, "step": 1030 }, { "epoch": 0.31337386018237084, "grad_norm": 0.25390625, "learning_rate": 6.866261398176293e-06, "loss": 1.0972, "step": 1031 }, { "epoch": 0.31367781155015195, "grad_norm": 0.271484375, "learning_rate": 6.863221884498481e-06, "loss": 0.9841, "step": 1032 }, { "epoch": 0.3139817629179331, "grad_norm": 0.265625, "learning_rate": 6.860182370820669e-06, "loss": 1.1033, "step": 1033 }, { "epoch": 0.3142857142857143, "grad_norm": 0.255859375, "learning_rate": 6.857142857142858e-06, "loss": 1.0342, "step": 1034 }, { "epoch": 0.31458966565349544, "grad_norm": 0.255859375, "learning_rate": 6.854103343465046e-06, "loss": 1.0498, "step": 1035 }, { "epoch": 0.3148936170212766, "grad_norm": 0.251953125, "learning_rate": 6.8510638297872344e-06, "loss": 1.1319, "step": 1036 }, { "epoch": 0.31519756838905777, "grad_norm": 0.27734375, "learning_rate": 6.848024316109423e-06, "loss": 1.0118, "step": 1037 }, { "epoch": 0.31550151975683893, "grad_norm": 0.26171875, "learning_rate": 6.844984802431611e-06, "loss": 1.0485, "step": 1038 }, { "epoch": 0.31580547112462004, "grad_norm": 0.259765625, "learning_rate": 6.8419452887538e-06, "loss": 1.1479, "step": 1039 }, { "epoch": 0.3161094224924012, "grad_norm": 0.26953125, "learning_rate": 6.838905775075988e-06, "loss": 1.0511, "step": 1040 }, { "epoch": 0.31641337386018237, "grad_norm": 0.2451171875, "learning_rate": 6.835866261398176e-06, "loss": 1.0844, "step": 1041 }, { "epoch": 0.31671732522796353, "grad_norm": 0.25390625, "learning_rate": 6.832826747720366e-06, "loss": 1.1124, "step": 1042 }, { "epoch": 0.3170212765957447, "grad_norm": 0.341796875, "learning_rate": 6.829787234042554e-06, "loss": 1.1521, "step": 1043 }, { "epoch": 0.31732522796352586, "grad_norm": 0.263671875, "learning_rate": 6.826747720364742e-06, "loss": 1.1236, "step": 1044 }, { "epoch": 0.31762917933130697, "grad_norm": 0.251953125, "learning_rate": 6.823708206686931e-06, "loss": 1.0813, "step": 1045 }, { "epoch": 0.31793313069908813, "grad_norm": 0.263671875, "learning_rate": 6.82066869300912e-06, "loss": 1.0811, "step": 1046 }, { "epoch": 0.3182370820668693, "grad_norm": 0.27734375, "learning_rate": 6.8176291793313074e-06, "loss": 1.1279, "step": 1047 }, { "epoch": 0.31854103343465046, "grad_norm": 0.271484375, "learning_rate": 6.814589665653496e-06, "loss": 1.1205, "step": 1048 }, { "epoch": 0.3188449848024316, "grad_norm": 0.26953125, "learning_rate": 6.811550151975684e-06, "loss": 1.0649, "step": 1049 }, { "epoch": 0.3191489361702128, "grad_norm": 0.44140625, "learning_rate": 6.808510638297873e-06, "loss": 1.0555, "step": 1050 }, { "epoch": 0.31945288753799395, "grad_norm": 0.26953125, "learning_rate": 6.805471124620061e-06, "loss": 1.0752, "step": 1051 }, { "epoch": 0.31975683890577505, "grad_norm": 0.267578125, "learning_rate": 6.802431610942249e-06, "loss": 1.0303, "step": 1052 }, { "epoch": 0.3200607902735562, "grad_norm": 0.279296875, "learning_rate": 6.799392097264438e-06, "loss": 1.0697, "step": 1053 }, { "epoch": 0.3203647416413374, "grad_norm": 0.2578125, "learning_rate": 6.7963525835866265e-06, "loss": 1.0974, "step": 1054 }, { "epoch": 0.32066869300911854, "grad_norm": 0.25390625, "learning_rate": 6.793313069908816e-06, "loss": 1.1062, "step": 1055 }, { "epoch": 0.3209726443768997, "grad_norm": 0.259765625, "learning_rate": 6.790273556231004e-06, "loss": 1.1083, "step": 1056 }, { "epoch": 0.32127659574468087, "grad_norm": 0.2890625, "learning_rate": 6.787234042553193e-06, "loss": 1.0049, "step": 1057 }, { "epoch": 0.321580547112462, "grad_norm": 0.2578125, "learning_rate": 6.7841945288753804e-06, "loss": 1.0904, "step": 1058 }, { "epoch": 0.32188449848024314, "grad_norm": 0.25390625, "learning_rate": 6.781155015197569e-06, "loss": 1.0862, "step": 1059 }, { "epoch": 0.3221884498480243, "grad_norm": 0.255859375, "learning_rate": 6.778115501519758e-06, "loss": 1.1415, "step": 1060 }, { "epoch": 0.32249240121580547, "grad_norm": 0.275390625, "learning_rate": 6.775075987841946e-06, "loss": 1.1401, "step": 1061 }, { "epoch": 0.32279635258358663, "grad_norm": 0.26171875, "learning_rate": 6.772036474164134e-06, "loss": 1.0428, "step": 1062 }, { "epoch": 0.3231003039513678, "grad_norm": 0.265625, "learning_rate": 6.768996960486323e-06, "loss": 1.0853, "step": 1063 }, { "epoch": 0.32340425531914896, "grad_norm": 0.25, "learning_rate": 6.765957446808511e-06, "loss": 1.1508, "step": 1064 }, { "epoch": 0.32370820668693007, "grad_norm": 0.259765625, "learning_rate": 6.7629179331306995e-06, "loss": 1.1456, "step": 1065 }, { "epoch": 0.32401215805471123, "grad_norm": 0.2734375, "learning_rate": 6.759878419452887e-06, "loss": 1.0368, "step": 1066 }, { "epoch": 0.3243161094224924, "grad_norm": 0.287109375, "learning_rate": 6.756838905775076e-06, "loss": 1.0568, "step": 1067 }, { "epoch": 0.32462006079027356, "grad_norm": 0.255859375, "learning_rate": 6.753799392097266e-06, "loss": 1.0586, "step": 1068 }, { "epoch": 0.3249240121580547, "grad_norm": 0.25390625, "learning_rate": 6.750759878419454e-06, "loss": 1.08, "step": 1069 }, { "epoch": 0.3252279635258359, "grad_norm": 0.265625, "learning_rate": 6.747720364741642e-06, "loss": 1.0515, "step": 1070 }, { "epoch": 0.32553191489361705, "grad_norm": 0.267578125, "learning_rate": 6.744680851063831e-06, "loss": 1.0393, "step": 1071 }, { "epoch": 0.32583586626139815, "grad_norm": 0.275390625, "learning_rate": 6.741641337386019e-06, "loss": 1.004, "step": 1072 }, { "epoch": 0.3261398176291793, "grad_norm": 0.2890625, "learning_rate": 6.738601823708207e-06, "loss": 1.0964, "step": 1073 }, { "epoch": 0.3264437689969605, "grad_norm": 0.294921875, "learning_rate": 6.735562310030396e-06, "loss": 1.0805, "step": 1074 }, { "epoch": 0.32674772036474165, "grad_norm": 0.267578125, "learning_rate": 6.732522796352584e-06, "loss": 0.9962, "step": 1075 }, { "epoch": 0.3270516717325228, "grad_norm": 0.287109375, "learning_rate": 6.7294832826747725e-06, "loss": 1.1249, "step": 1076 }, { "epoch": 0.32735562310030397, "grad_norm": 0.2490234375, "learning_rate": 6.726443768996961e-06, "loss": 1.0901, "step": 1077 }, { "epoch": 0.3276595744680851, "grad_norm": 0.26171875, "learning_rate": 6.723404255319149e-06, "loss": 1.1016, "step": 1078 }, { "epoch": 0.32796352583586624, "grad_norm": 0.265625, "learning_rate": 6.720364741641338e-06, "loss": 1.0666, "step": 1079 }, { "epoch": 0.3282674772036474, "grad_norm": 0.3359375, "learning_rate": 6.7173252279635256e-06, "loss": 1.0732, "step": 1080 }, { "epoch": 0.32857142857142857, "grad_norm": 0.3203125, "learning_rate": 6.714285714285714e-06, "loss": 1.1487, "step": 1081 }, { "epoch": 0.32887537993920973, "grad_norm": 0.275390625, "learning_rate": 6.711246200607904e-06, "loss": 1.0865, "step": 1082 }, { "epoch": 0.3291793313069909, "grad_norm": 0.27734375, "learning_rate": 6.7082066869300925e-06, "loss": 1.0561, "step": 1083 }, { "epoch": 0.32948328267477206, "grad_norm": 0.3203125, "learning_rate": 6.70516717325228e-06, "loss": 1.0529, "step": 1084 }, { "epoch": 0.32978723404255317, "grad_norm": 0.2578125, "learning_rate": 6.702127659574469e-06, "loss": 1.0799, "step": 1085 }, { "epoch": 0.33009118541033433, "grad_norm": 0.255859375, "learning_rate": 6.699088145896657e-06, "loss": 1.0186, "step": 1086 }, { "epoch": 0.3303951367781155, "grad_norm": 0.2470703125, "learning_rate": 6.6960486322188455e-06, "loss": 1.1215, "step": 1087 }, { "epoch": 0.33069908814589666, "grad_norm": 0.263671875, "learning_rate": 6.693009118541034e-06, "loss": 1.1034, "step": 1088 }, { "epoch": 0.3310030395136778, "grad_norm": 0.267578125, "learning_rate": 6.689969604863222e-06, "loss": 1.0116, "step": 1089 }, { "epoch": 0.331306990881459, "grad_norm": 0.251953125, "learning_rate": 6.686930091185411e-06, "loss": 1.1176, "step": 1090 }, { "epoch": 0.33161094224924015, "grad_norm": 0.2578125, "learning_rate": 6.683890577507599e-06, "loss": 1.1002, "step": 1091 }, { "epoch": 0.33191489361702126, "grad_norm": 0.265625, "learning_rate": 6.680851063829787e-06, "loss": 1.0532, "step": 1092 }, { "epoch": 0.3322188449848024, "grad_norm": 0.2734375, "learning_rate": 6.677811550151976e-06, "loss": 1.0624, "step": 1093 }, { "epoch": 0.3325227963525836, "grad_norm": 0.275390625, "learning_rate": 6.674772036474164e-06, "loss": 1.0274, "step": 1094 }, { "epoch": 0.33282674772036475, "grad_norm": 0.2578125, "learning_rate": 6.671732522796353e-06, "loss": 1.0888, "step": 1095 }, { "epoch": 0.3331306990881459, "grad_norm": 0.255859375, "learning_rate": 6.668693009118542e-06, "loss": 1.0001, "step": 1096 }, { "epoch": 0.3334346504559271, "grad_norm": 0.2578125, "learning_rate": 6.665653495440731e-06, "loss": 1.0735, "step": 1097 }, { "epoch": 0.3337386018237082, "grad_norm": 0.2734375, "learning_rate": 6.6626139817629185e-06, "loss": 1.1272, "step": 1098 }, { "epoch": 0.33404255319148934, "grad_norm": 0.353515625, "learning_rate": 6.659574468085107e-06, "loss": 1.0553, "step": 1099 }, { "epoch": 0.3343465045592705, "grad_norm": 0.271484375, "learning_rate": 6.656534954407296e-06, "loss": 1.0355, "step": 1100 }, { "epoch": 0.33465045592705167, "grad_norm": 0.26953125, "learning_rate": 6.653495440729484e-06, "loss": 1.0921, "step": 1101 }, { "epoch": 0.33495440729483283, "grad_norm": 0.267578125, "learning_rate": 6.650455927051672e-06, "loss": 1.0777, "step": 1102 }, { "epoch": 0.335258358662614, "grad_norm": 0.26171875, "learning_rate": 6.64741641337386e-06, "loss": 1.1069, "step": 1103 }, { "epoch": 0.33556231003039516, "grad_norm": 0.296875, "learning_rate": 6.644376899696049e-06, "loss": 1.0424, "step": 1104 }, { "epoch": 0.33586626139817627, "grad_norm": 0.263671875, "learning_rate": 6.641337386018238e-06, "loss": 1.12, "step": 1105 }, { "epoch": 0.33617021276595743, "grad_norm": 0.28515625, "learning_rate": 6.6382978723404254e-06, "loss": 1.1068, "step": 1106 }, { "epoch": 0.3364741641337386, "grad_norm": 0.2734375, "learning_rate": 6.635258358662614e-06, "loss": 1.0468, "step": 1107 }, { "epoch": 0.33677811550151976, "grad_norm": 0.263671875, "learning_rate": 6.632218844984804e-06, "loss": 1.0393, "step": 1108 }, { "epoch": 0.3370820668693009, "grad_norm": 0.279296875, "learning_rate": 6.6291793313069915e-06, "loss": 1.0261, "step": 1109 }, { "epoch": 0.3373860182370821, "grad_norm": 0.2578125, "learning_rate": 6.62613981762918e-06, "loss": 1.066, "step": 1110 }, { "epoch": 0.33768996960486325, "grad_norm": 0.271484375, "learning_rate": 6.623100303951369e-06, "loss": 1.0939, "step": 1111 }, { "epoch": 0.33799392097264436, "grad_norm": 0.267578125, "learning_rate": 6.620060790273557e-06, "loss": 1.046, "step": 1112 }, { "epoch": 0.3382978723404255, "grad_norm": 0.267578125, "learning_rate": 6.617021276595745e-06, "loss": 1.0513, "step": 1113 }, { "epoch": 0.3386018237082067, "grad_norm": 0.263671875, "learning_rate": 6.613981762917934e-06, "loss": 1.0338, "step": 1114 }, { "epoch": 0.33890577507598785, "grad_norm": 0.263671875, "learning_rate": 6.610942249240122e-06, "loss": 1.0851, "step": 1115 }, { "epoch": 0.339209726443769, "grad_norm": 0.267578125, "learning_rate": 6.607902735562311e-06, "loss": 1.0103, "step": 1116 }, { "epoch": 0.3395136778115502, "grad_norm": 0.283203125, "learning_rate": 6.6048632218844984e-06, "loss": 1.1253, "step": 1117 }, { "epoch": 0.3398176291793313, "grad_norm": 0.267578125, "learning_rate": 6.601823708206687e-06, "loss": 0.9874, "step": 1118 }, { "epoch": 0.34012158054711245, "grad_norm": 0.26171875, "learning_rate": 6.598784194528876e-06, "loss": 1.1077, "step": 1119 }, { "epoch": 0.3404255319148936, "grad_norm": 0.267578125, "learning_rate": 6.595744680851064e-06, "loss": 1.049, "step": 1120 }, { "epoch": 0.34072948328267477, "grad_norm": 0.26171875, "learning_rate": 6.592705167173253e-06, "loss": 1.0558, "step": 1121 }, { "epoch": 0.34103343465045594, "grad_norm": 0.251953125, "learning_rate": 6.589665653495442e-06, "loss": 1.1204, "step": 1122 }, { "epoch": 0.3413373860182371, "grad_norm": 0.267578125, "learning_rate": 6.58662613981763e-06, "loss": 1.0835, "step": 1123 }, { "epoch": 0.34164133738601826, "grad_norm": 0.26953125, "learning_rate": 6.583586626139818e-06, "loss": 1.0028, "step": 1124 }, { "epoch": 0.34194528875379937, "grad_norm": 0.26953125, "learning_rate": 6.580547112462007e-06, "loss": 1.065, "step": 1125 }, { "epoch": 0.34224924012158053, "grad_norm": 0.2890625, "learning_rate": 6.577507598784195e-06, "loss": 1.0587, "step": 1126 }, { "epoch": 0.3425531914893617, "grad_norm": 0.267578125, "learning_rate": 6.574468085106384e-06, "loss": 1.0946, "step": 1127 }, { "epoch": 0.34285714285714286, "grad_norm": 0.265625, "learning_rate": 6.571428571428572e-06, "loss": 1.0981, "step": 1128 }, { "epoch": 0.343161094224924, "grad_norm": 0.267578125, "learning_rate": 6.56838905775076e-06, "loss": 1.0609, "step": 1129 }, { "epoch": 0.3434650455927052, "grad_norm": 0.29296875, "learning_rate": 6.565349544072949e-06, "loss": 1.0967, "step": 1130 }, { "epoch": 0.3437689969604863, "grad_norm": 0.26953125, "learning_rate": 6.562310030395137e-06, "loss": 1.0345, "step": 1131 }, { "epoch": 0.34407294832826746, "grad_norm": 0.267578125, "learning_rate": 6.559270516717325e-06, "loss": 1.0859, "step": 1132 }, { "epoch": 0.3443768996960486, "grad_norm": 0.2578125, "learning_rate": 6.556231003039514e-06, "loss": 1.0643, "step": 1133 }, { "epoch": 0.3446808510638298, "grad_norm": 0.2578125, "learning_rate": 6.553191489361702e-06, "loss": 1.0437, "step": 1134 }, { "epoch": 0.34498480243161095, "grad_norm": 0.27734375, "learning_rate": 6.550151975683891e-06, "loss": 1.0239, "step": 1135 }, { "epoch": 0.3452887537993921, "grad_norm": 0.291015625, "learning_rate": 6.54711246200608e-06, "loss": 1.1095, "step": 1136 }, { "epoch": 0.3455927051671733, "grad_norm": 0.28125, "learning_rate": 6.544072948328269e-06, "loss": 1.0234, "step": 1137 }, { "epoch": 0.3458966565349544, "grad_norm": 0.275390625, "learning_rate": 6.541033434650457e-06, "loss": 0.9641, "step": 1138 }, { "epoch": 0.34620060790273555, "grad_norm": 0.27734375, "learning_rate": 6.537993920972645e-06, "loss": 1.102, "step": 1139 }, { "epoch": 0.3465045592705167, "grad_norm": 0.287109375, "learning_rate": 6.534954407294833e-06, "loss": 1.0548, "step": 1140 }, { "epoch": 0.3468085106382979, "grad_norm": 0.271484375, "learning_rate": 6.531914893617022e-06, "loss": 1.097, "step": 1141 }, { "epoch": 0.34711246200607904, "grad_norm": 0.2578125, "learning_rate": 6.5288753799392105e-06, "loss": 1.1099, "step": 1142 }, { "epoch": 0.3474164133738602, "grad_norm": 0.2578125, "learning_rate": 6.525835866261398e-06, "loss": 1.0999, "step": 1143 }, { "epoch": 0.34772036474164136, "grad_norm": 0.2734375, "learning_rate": 6.522796352583587e-06, "loss": 1.1546, "step": 1144 }, { "epoch": 0.34802431610942247, "grad_norm": 0.27734375, "learning_rate": 6.519756838905775e-06, "loss": 1.0336, "step": 1145 }, { "epoch": 0.34832826747720363, "grad_norm": 0.279296875, "learning_rate": 6.5167173252279635e-06, "loss": 0.9827, "step": 1146 }, { "epoch": 0.3486322188449848, "grad_norm": 0.294921875, "learning_rate": 6.513677811550152e-06, "loss": 1.0358, "step": 1147 }, { "epoch": 0.34893617021276596, "grad_norm": 0.271484375, "learning_rate": 6.510638297872342e-06, "loss": 1.1467, "step": 1148 }, { "epoch": 0.3492401215805471, "grad_norm": 0.263671875, "learning_rate": 6.50759878419453e-06, "loss": 1.1478, "step": 1149 }, { "epoch": 0.3495440729483283, "grad_norm": 0.2734375, "learning_rate": 6.504559270516718e-06, "loss": 1.0851, "step": 1150 }, { "epoch": 0.3498480243161094, "grad_norm": 0.271484375, "learning_rate": 6.501519756838907e-06, "loss": 1.0895, "step": 1151 }, { "epoch": 0.35015197568389056, "grad_norm": 0.34375, "learning_rate": 6.498480243161095e-06, "loss": 1.0444, "step": 1152 }, { "epoch": 0.3504559270516717, "grad_norm": 0.31640625, "learning_rate": 6.4954407294832835e-06, "loss": 1.0649, "step": 1153 }, { "epoch": 0.3507598784194529, "grad_norm": 0.255859375, "learning_rate": 6.492401215805471e-06, "loss": 1.1457, "step": 1154 }, { "epoch": 0.35106382978723405, "grad_norm": 0.35546875, "learning_rate": 6.48936170212766e-06, "loss": 1.049, "step": 1155 }, { "epoch": 0.3513677811550152, "grad_norm": 0.259765625, "learning_rate": 6.486322188449849e-06, "loss": 1.0763, "step": 1156 }, { "epoch": 0.3516717325227964, "grad_norm": 0.267578125, "learning_rate": 6.4832826747720365e-06, "loss": 1.1199, "step": 1157 }, { "epoch": 0.3519756838905775, "grad_norm": 0.25390625, "learning_rate": 6.480243161094225e-06, "loss": 1.0489, "step": 1158 }, { "epoch": 0.35227963525835865, "grad_norm": 0.275390625, "learning_rate": 6.477203647416414e-06, "loss": 1.0595, "step": 1159 }, { "epoch": 0.3525835866261398, "grad_norm": 0.259765625, "learning_rate": 6.474164133738602e-06, "loss": 1.0265, "step": 1160 }, { "epoch": 0.352887537993921, "grad_norm": 0.275390625, "learning_rate": 6.471124620060791e-06, "loss": 1.0871, "step": 1161 }, { "epoch": 0.35319148936170214, "grad_norm": 0.287109375, "learning_rate": 6.46808510638298e-06, "loss": 1.0346, "step": 1162 }, { "epoch": 0.3534954407294833, "grad_norm": 0.271484375, "learning_rate": 6.465045592705168e-06, "loss": 1.0664, "step": 1163 }, { "epoch": 0.35379939209726446, "grad_norm": 0.265625, "learning_rate": 6.4620060790273565e-06, "loss": 1.099, "step": 1164 }, { "epoch": 0.3541033434650456, "grad_norm": 0.26171875, "learning_rate": 6.458966565349545e-06, "loss": 1.0471, "step": 1165 }, { "epoch": 0.35440729483282674, "grad_norm": 0.267578125, "learning_rate": 6.455927051671733e-06, "loss": 1.0344, "step": 1166 }, { "epoch": 0.3547112462006079, "grad_norm": 0.30859375, "learning_rate": 6.452887537993922e-06, "loss": 1.1169, "step": 1167 }, { "epoch": 0.35501519756838906, "grad_norm": 0.263671875, "learning_rate": 6.4498480243161095e-06, "loss": 1.136, "step": 1168 }, { "epoch": 0.3553191489361702, "grad_norm": 0.2470703125, "learning_rate": 6.446808510638298e-06, "loss": 1.0689, "step": 1169 }, { "epoch": 0.3556231003039514, "grad_norm": 0.267578125, "learning_rate": 6.443768996960487e-06, "loss": 1.1263, "step": 1170 }, { "epoch": 0.3559270516717325, "grad_norm": 0.2734375, "learning_rate": 6.440729483282675e-06, "loss": 0.9996, "step": 1171 }, { "epoch": 0.35623100303951366, "grad_norm": 0.26171875, "learning_rate": 6.437689969604863e-06, "loss": 1.1465, "step": 1172 }, { "epoch": 0.3565349544072948, "grad_norm": 0.265625, "learning_rate": 6.434650455927052e-06, "loss": 1.0086, "step": 1173 }, { "epoch": 0.356838905775076, "grad_norm": 0.2734375, "learning_rate": 6.43161094224924e-06, "loss": 1.1143, "step": 1174 }, { "epoch": 0.35714285714285715, "grad_norm": 0.267578125, "learning_rate": 6.4285714285714295e-06, "loss": 1.0913, "step": 1175 }, { "epoch": 0.3574468085106383, "grad_norm": 0.478515625, "learning_rate": 6.425531914893618e-06, "loss": 0.991, "step": 1176 }, { "epoch": 0.3577507598784195, "grad_norm": 0.34375, "learning_rate": 6.422492401215806e-06, "loss": 1.0531, "step": 1177 }, { "epoch": 0.3580547112462006, "grad_norm": 0.27734375, "learning_rate": 6.419452887537995e-06, "loss": 1.0561, "step": 1178 }, { "epoch": 0.35835866261398175, "grad_norm": 0.279296875, "learning_rate": 6.416413373860183e-06, "loss": 1.0965, "step": 1179 }, { "epoch": 0.3586626139817629, "grad_norm": 0.26953125, "learning_rate": 6.413373860182371e-06, "loss": 1.1441, "step": 1180 }, { "epoch": 0.3589665653495441, "grad_norm": 0.279296875, "learning_rate": 6.41033434650456e-06, "loss": 1.0896, "step": 1181 }, { "epoch": 0.35927051671732524, "grad_norm": 0.2734375, "learning_rate": 6.407294832826748e-06, "loss": 1.0842, "step": 1182 }, { "epoch": 0.3595744680851064, "grad_norm": 0.263671875, "learning_rate": 6.404255319148936e-06, "loss": 1.1053, "step": 1183 }, { "epoch": 0.35987841945288757, "grad_norm": 0.259765625, "learning_rate": 6.401215805471125e-06, "loss": 1.1439, "step": 1184 }, { "epoch": 0.3601823708206687, "grad_norm": 0.263671875, "learning_rate": 6.398176291793313e-06, "loss": 1.0342, "step": 1185 }, { "epoch": 0.36048632218844984, "grad_norm": 0.267578125, "learning_rate": 6.395136778115502e-06, "loss": 1.1416, "step": 1186 }, { "epoch": 0.360790273556231, "grad_norm": 0.267578125, "learning_rate": 6.39209726443769e-06, "loss": 1.0515, "step": 1187 }, { "epoch": 0.36109422492401216, "grad_norm": 0.263671875, "learning_rate": 6.38905775075988e-06, "loss": 1.1041, "step": 1188 }, { "epoch": 0.3613981762917933, "grad_norm": 0.265625, "learning_rate": 6.386018237082068e-06, "loss": 1.0465, "step": 1189 }, { "epoch": 0.3617021276595745, "grad_norm": 0.265625, "learning_rate": 6.382978723404256e-06, "loss": 1.0206, "step": 1190 }, { "epoch": 0.3620060790273556, "grad_norm": 0.267578125, "learning_rate": 6.379939209726444e-06, "loss": 1.078, "step": 1191 }, { "epoch": 0.36231003039513676, "grad_norm": 0.263671875, "learning_rate": 6.376899696048633e-06, "loss": 1.1082, "step": 1192 }, { "epoch": 0.3626139817629179, "grad_norm": 0.267578125, "learning_rate": 6.3738601823708216e-06, "loss": 1.0414, "step": 1193 }, { "epoch": 0.3629179331306991, "grad_norm": 0.267578125, "learning_rate": 6.370820668693009e-06, "loss": 1.0062, "step": 1194 }, { "epoch": 0.36322188449848025, "grad_norm": 0.263671875, "learning_rate": 6.367781155015198e-06, "loss": 1.0155, "step": 1195 }, { "epoch": 0.3635258358662614, "grad_norm": 0.265625, "learning_rate": 6.364741641337387e-06, "loss": 1.1378, "step": 1196 }, { "epoch": 0.3638297872340426, "grad_norm": 0.263671875, "learning_rate": 6.361702127659575e-06, "loss": 1.0098, "step": 1197 }, { "epoch": 0.3641337386018237, "grad_norm": 0.267578125, "learning_rate": 6.358662613981763e-06, "loss": 1.0264, "step": 1198 }, { "epoch": 0.36443768996960485, "grad_norm": 0.271484375, "learning_rate": 6.355623100303951e-06, "loss": 1.0137, "step": 1199 }, { "epoch": 0.364741641337386, "grad_norm": 0.287109375, "learning_rate": 6.35258358662614e-06, "loss": 1.0932, "step": 1200 }, { "epoch": 0.3650455927051672, "grad_norm": 0.265625, "learning_rate": 6.349544072948329e-06, "loss": 1.1042, "step": 1201 }, { "epoch": 0.36534954407294834, "grad_norm": 0.275390625, "learning_rate": 6.346504559270518e-06, "loss": 1.1179, "step": 1202 }, { "epoch": 0.3656534954407295, "grad_norm": 0.2578125, "learning_rate": 6.343465045592706e-06, "loss": 1.0747, "step": 1203 }, { "epoch": 0.3659574468085106, "grad_norm": 0.27734375, "learning_rate": 6.3404255319148945e-06, "loss": 1.201, "step": 1204 }, { "epoch": 0.3662613981762918, "grad_norm": 0.28125, "learning_rate": 6.337386018237082e-06, "loss": 1.1096, "step": 1205 }, { "epoch": 0.36656534954407294, "grad_norm": 0.265625, "learning_rate": 6.334346504559271e-06, "loss": 1.0432, "step": 1206 }, { "epoch": 0.3668693009118541, "grad_norm": 0.26171875, "learning_rate": 6.33130699088146e-06, "loss": 1.1247, "step": 1207 }, { "epoch": 0.36717325227963526, "grad_norm": 0.30859375, "learning_rate": 6.328267477203648e-06, "loss": 1.0282, "step": 1208 }, { "epoch": 0.36747720364741643, "grad_norm": 0.27734375, "learning_rate": 6.325227963525836e-06, "loss": 1.0882, "step": 1209 }, { "epoch": 0.3677811550151976, "grad_norm": 0.26953125, "learning_rate": 6.322188449848025e-06, "loss": 1.064, "step": 1210 }, { "epoch": 0.3680851063829787, "grad_norm": 0.9609375, "learning_rate": 6.319148936170213e-06, "loss": 1.0655, "step": 1211 }, { "epoch": 0.36838905775075986, "grad_norm": 0.275390625, "learning_rate": 6.3161094224924015e-06, "loss": 0.9565, "step": 1212 }, { "epoch": 0.368693009118541, "grad_norm": 0.259765625, "learning_rate": 6.313069908814589e-06, "loss": 1.097, "step": 1213 }, { "epoch": 0.3689969604863222, "grad_norm": 0.26953125, "learning_rate": 6.310030395136779e-06, "loss": 1.0343, "step": 1214 }, { "epoch": 0.36930091185410335, "grad_norm": 0.271484375, "learning_rate": 6.3069908814589675e-06, "loss": 1.0816, "step": 1215 }, { "epoch": 0.3696048632218845, "grad_norm": 0.26953125, "learning_rate": 6.303951367781156e-06, "loss": 1.098, "step": 1216 }, { "epoch": 0.3699088145896657, "grad_norm": 0.26953125, "learning_rate": 6.300911854103344e-06, "loss": 1.0363, "step": 1217 }, { "epoch": 0.3702127659574468, "grad_norm": 0.26953125, "learning_rate": 6.297872340425533e-06, "loss": 1.0986, "step": 1218 }, { "epoch": 0.37051671732522795, "grad_norm": 0.26171875, "learning_rate": 6.294832826747721e-06, "loss": 1.0775, "step": 1219 }, { "epoch": 0.3708206686930091, "grad_norm": 0.26953125, "learning_rate": 6.291793313069909e-06, "loss": 0.9819, "step": 1220 }, { "epoch": 0.3711246200607903, "grad_norm": 0.337890625, "learning_rate": 6.288753799392098e-06, "loss": 1.106, "step": 1221 }, { "epoch": 0.37142857142857144, "grad_norm": 0.271484375, "learning_rate": 6.285714285714286e-06, "loss": 1.1498, "step": 1222 }, { "epoch": 0.3717325227963526, "grad_norm": 0.30078125, "learning_rate": 6.2826747720364745e-06, "loss": 0.9928, "step": 1223 }, { "epoch": 0.3720364741641337, "grad_norm": 0.302734375, "learning_rate": 6.279635258358663e-06, "loss": 1.053, "step": 1224 }, { "epoch": 0.3723404255319149, "grad_norm": 0.267578125, "learning_rate": 6.276595744680851e-06, "loss": 1.0824, "step": 1225 }, { "epoch": 0.37264437689969604, "grad_norm": 0.27734375, "learning_rate": 6.27355623100304e-06, "loss": 1.0764, "step": 1226 }, { "epoch": 0.3729483282674772, "grad_norm": 0.265625, "learning_rate": 6.2705167173252275e-06, "loss": 1.0432, "step": 1227 }, { "epoch": 0.37325227963525837, "grad_norm": 0.279296875, "learning_rate": 6.267477203647417e-06, "loss": 1.0309, "step": 1228 }, { "epoch": 0.37355623100303953, "grad_norm": 0.271484375, "learning_rate": 6.264437689969606e-06, "loss": 1.0839, "step": 1229 }, { "epoch": 0.3738601823708207, "grad_norm": 0.259765625, "learning_rate": 6.2613981762917944e-06, "loss": 1.0786, "step": 1230 }, { "epoch": 0.3741641337386018, "grad_norm": 0.26953125, "learning_rate": 6.258358662613982e-06, "loss": 0.9982, "step": 1231 }, { "epoch": 0.37446808510638296, "grad_norm": 0.26953125, "learning_rate": 6.255319148936171e-06, "loss": 1.0382, "step": 1232 }, { "epoch": 0.3747720364741641, "grad_norm": 0.271484375, "learning_rate": 6.25227963525836e-06, "loss": 1.026, "step": 1233 }, { "epoch": 0.3750759878419453, "grad_norm": 0.259765625, "learning_rate": 6.2492401215805475e-06, "loss": 0.9705, "step": 1234 }, { "epoch": 0.37537993920972645, "grad_norm": 0.26953125, "learning_rate": 6.246200607902736e-06, "loss": 1.0621, "step": 1235 }, { "epoch": 0.3756838905775076, "grad_norm": 0.2734375, "learning_rate": 6.243161094224924e-06, "loss": 1.0241, "step": 1236 }, { "epoch": 0.3759878419452888, "grad_norm": 0.3125, "learning_rate": 6.240121580547113e-06, "loss": 1.2197, "step": 1237 }, { "epoch": 0.3762917933130699, "grad_norm": 0.2578125, "learning_rate": 6.237082066869301e-06, "loss": 1.0456, "step": 1238 }, { "epoch": 0.37659574468085105, "grad_norm": 0.275390625, "learning_rate": 6.234042553191489e-06, "loss": 1.0036, "step": 1239 }, { "epoch": 0.3768996960486322, "grad_norm": 0.291015625, "learning_rate": 6.231003039513678e-06, "loss": 0.9695, "step": 1240 }, { "epoch": 0.3772036474164134, "grad_norm": 0.26953125, "learning_rate": 6.227963525835867e-06, "loss": 1.0704, "step": 1241 }, { "epoch": 0.37750759878419454, "grad_norm": 0.259765625, "learning_rate": 6.224924012158055e-06, "loss": 1.1212, "step": 1242 }, { "epoch": 0.3778115501519757, "grad_norm": 0.275390625, "learning_rate": 6.221884498480244e-06, "loss": 0.9709, "step": 1243 }, { "epoch": 0.3781155015197568, "grad_norm": 0.28125, "learning_rate": 6.218844984802433e-06, "loss": 0.9902, "step": 1244 }, { "epoch": 0.378419452887538, "grad_norm": 0.27734375, "learning_rate": 6.2158054711246205e-06, "loss": 1.0306, "step": 1245 }, { "epoch": 0.37872340425531914, "grad_norm": 0.265625, "learning_rate": 6.212765957446809e-06, "loss": 1.0334, "step": 1246 }, { "epoch": 0.3790273556231003, "grad_norm": 0.287109375, "learning_rate": 6.209726443768998e-06, "loss": 1.1382, "step": 1247 }, { "epoch": 0.37933130699088147, "grad_norm": 0.283203125, "learning_rate": 6.206686930091186e-06, "loss": 1.0754, "step": 1248 }, { "epoch": 0.37963525835866263, "grad_norm": 0.2734375, "learning_rate": 6.203647416413374e-06, "loss": 1.1659, "step": 1249 }, { "epoch": 0.3799392097264438, "grad_norm": 0.26953125, "learning_rate": 6.200607902735562e-06, "loss": 1.158, "step": 1250 }, { "epoch": 0.3802431610942249, "grad_norm": 0.28125, "learning_rate": 6.197568389057751e-06, "loss": 1.0454, "step": 1251 }, { "epoch": 0.38054711246200607, "grad_norm": 0.271484375, "learning_rate": 6.1945288753799396e-06, "loss": 1.0085, "step": 1252 }, { "epoch": 0.38085106382978723, "grad_norm": 0.294921875, "learning_rate": 6.191489361702127e-06, "loss": 1.0302, "step": 1253 }, { "epoch": 0.3811550151975684, "grad_norm": 0.265625, "learning_rate": 6.188449848024317e-06, "loss": 1.0666, "step": 1254 }, { "epoch": 0.38145896656534956, "grad_norm": 0.33984375, "learning_rate": 6.185410334346506e-06, "loss": 1.0977, "step": 1255 }, { "epoch": 0.3817629179331307, "grad_norm": 0.2734375, "learning_rate": 6.1823708206686935e-06, "loss": 1.0405, "step": 1256 }, { "epoch": 0.3820668693009118, "grad_norm": 0.26171875, "learning_rate": 6.179331306990882e-06, "loss": 1.0597, "step": 1257 }, { "epoch": 0.382370820668693, "grad_norm": 0.296875, "learning_rate": 6.176291793313071e-06, "loss": 1.1542, "step": 1258 }, { "epoch": 0.38267477203647415, "grad_norm": 0.275390625, "learning_rate": 6.173252279635259e-06, "loss": 1.0877, "step": 1259 }, { "epoch": 0.3829787234042553, "grad_norm": 0.26953125, "learning_rate": 6.170212765957447e-06, "loss": 1.1103, "step": 1260 }, { "epoch": 0.3832826747720365, "grad_norm": 0.263671875, "learning_rate": 6.167173252279636e-06, "loss": 0.9736, "step": 1261 }, { "epoch": 0.38358662613981764, "grad_norm": 0.27734375, "learning_rate": 6.164133738601824e-06, "loss": 1.0721, "step": 1262 }, { "epoch": 0.3838905775075988, "grad_norm": 0.2734375, "learning_rate": 6.1610942249240126e-06, "loss": 1.1162, "step": 1263 }, { "epoch": 0.3841945288753799, "grad_norm": 0.26953125, "learning_rate": 6.1580547112462e-06, "loss": 1.1358, "step": 1264 }, { "epoch": 0.3844984802431611, "grad_norm": 0.275390625, "learning_rate": 6.155015197568389e-06, "loss": 1.0083, "step": 1265 }, { "epoch": 0.38480243161094224, "grad_norm": 0.259765625, "learning_rate": 6.151975683890578e-06, "loss": 1.1044, "step": 1266 }, { "epoch": 0.3851063829787234, "grad_norm": 0.26171875, "learning_rate": 6.148936170212767e-06, "loss": 1.0804, "step": 1267 }, { "epoch": 0.38541033434650457, "grad_norm": 0.25, "learning_rate": 6.145896656534955e-06, "loss": 1.0896, "step": 1268 }, { "epoch": 0.38571428571428573, "grad_norm": 0.27734375, "learning_rate": 6.142857142857144e-06, "loss": 1.1699, "step": 1269 }, { "epoch": 0.3860182370820669, "grad_norm": 0.28515625, "learning_rate": 6.1398176291793325e-06, "loss": 1.0562, "step": 1270 }, { "epoch": 0.386322188449848, "grad_norm": 0.267578125, "learning_rate": 6.13677811550152e-06, "loss": 1.1001, "step": 1271 }, { "epoch": 0.38662613981762917, "grad_norm": 0.263671875, "learning_rate": 6.133738601823709e-06, "loss": 1.065, "step": 1272 }, { "epoch": 0.38693009118541033, "grad_norm": 0.26171875, "learning_rate": 6.130699088145897e-06, "loss": 1.1292, "step": 1273 }, { "epoch": 0.3872340425531915, "grad_norm": 0.26953125, "learning_rate": 6.1276595744680855e-06, "loss": 1.1764, "step": 1274 }, { "epoch": 0.38753799392097266, "grad_norm": 0.29296875, "learning_rate": 6.124620060790274e-06, "loss": 1.0132, "step": 1275 }, { "epoch": 0.3878419452887538, "grad_norm": 0.267578125, "learning_rate": 6.121580547112462e-06, "loss": 1.0469, "step": 1276 }, { "epoch": 0.38814589665653493, "grad_norm": 0.2734375, "learning_rate": 6.118541033434651e-06, "loss": 1.0432, "step": 1277 }, { "epoch": 0.3884498480243161, "grad_norm": 0.271484375, "learning_rate": 6.1155015197568394e-06, "loss": 1.0456, "step": 1278 }, { "epoch": 0.38875379939209725, "grad_norm": 0.2890625, "learning_rate": 6.112462006079027e-06, "loss": 1.0476, "step": 1279 }, { "epoch": 0.3890577507598784, "grad_norm": 0.267578125, "learning_rate": 6.109422492401216e-06, "loss": 1.0092, "step": 1280 }, { "epoch": 0.3893617021276596, "grad_norm": 0.267578125, "learning_rate": 6.1063829787234055e-06, "loss": 1.0997, "step": 1281 }, { "epoch": 0.38966565349544074, "grad_norm": 0.28125, "learning_rate": 6.103343465045593e-06, "loss": 1.0295, "step": 1282 }, { "epoch": 0.3899696048632219, "grad_norm": 0.2734375, "learning_rate": 6.100303951367782e-06, "loss": 1.0867, "step": 1283 }, { "epoch": 0.390273556231003, "grad_norm": 0.2734375, "learning_rate": 6.097264437689971e-06, "loss": 1.0469, "step": 1284 }, { "epoch": 0.3905775075987842, "grad_norm": 0.28125, "learning_rate": 6.0942249240121585e-06, "loss": 1.0828, "step": 1285 }, { "epoch": 0.39088145896656534, "grad_norm": 0.291015625, "learning_rate": 6.091185410334347e-06, "loss": 1.1224, "step": 1286 }, { "epoch": 0.3911854103343465, "grad_norm": 0.298828125, "learning_rate": 6.088145896656535e-06, "loss": 1.1323, "step": 1287 }, { "epoch": 0.39148936170212767, "grad_norm": 0.28125, "learning_rate": 6.085106382978724e-06, "loss": 1.0229, "step": 1288 }, { "epoch": 0.39179331306990883, "grad_norm": 0.28125, "learning_rate": 6.0820668693009124e-06, "loss": 0.982, "step": 1289 }, { "epoch": 0.39209726443769, "grad_norm": 0.26953125, "learning_rate": 6.0790273556231e-06, "loss": 1.0438, "step": 1290 }, { "epoch": 0.3924012158054711, "grad_norm": 0.373046875, "learning_rate": 6.075987841945289e-06, "loss": 1.0593, "step": 1291 }, { "epoch": 0.39270516717325227, "grad_norm": 0.265625, "learning_rate": 6.072948328267478e-06, "loss": 1.0002, "step": 1292 }, { "epoch": 0.39300911854103343, "grad_norm": 0.322265625, "learning_rate": 6.0699088145896655e-06, "loss": 1.083, "step": 1293 }, { "epoch": 0.3933130699088146, "grad_norm": 0.275390625, "learning_rate": 6.066869300911855e-06, "loss": 1.0569, "step": 1294 }, { "epoch": 0.39361702127659576, "grad_norm": 0.31640625, "learning_rate": 6.063829787234044e-06, "loss": 1.0991, "step": 1295 }, { "epoch": 0.3939209726443769, "grad_norm": 0.306640625, "learning_rate": 6.0607902735562315e-06, "loss": 1.137, "step": 1296 }, { "epoch": 0.39422492401215803, "grad_norm": 0.271484375, "learning_rate": 6.05775075987842e-06, "loss": 1.1106, "step": 1297 }, { "epoch": 0.3945288753799392, "grad_norm": 0.259765625, "learning_rate": 6.054711246200609e-06, "loss": 1.086, "step": 1298 }, { "epoch": 0.39483282674772036, "grad_norm": 0.265625, "learning_rate": 6.051671732522797e-06, "loss": 1.0804, "step": 1299 }, { "epoch": 0.3951367781155015, "grad_norm": 0.26953125, "learning_rate": 6.0486322188449854e-06, "loss": 1.0765, "step": 1300 }, { "epoch": 0.3954407294832827, "grad_norm": 0.2890625, "learning_rate": 6.045592705167173e-06, "loss": 1.0372, "step": 1301 }, { "epoch": 0.39574468085106385, "grad_norm": 0.279296875, "learning_rate": 6.042553191489362e-06, "loss": 1.0319, "step": 1302 }, { "epoch": 0.396048632218845, "grad_norm": 0.2734375, "learning_rate": 6.039513677811551e-06, "loss": 1.0398, "step": 1303 }, { "epoch": 0.3963525835866261, "grad_norm": 0.27734375, "learning_rate": 6.0364741641337385e-06, "loss": 1.0792, "step": 1304 }, { "epoch": 0.3966565349544073, "grad_norm": 0.26953125, "learning_rate": 6.033434650455927e-06, "loss": 1.0382, "step": 1305 }, { "epoch": 0.39696048632218844, "grad_norm": 0.2734375, "learning_rate": 6.030395136778116e-06, "loss": 1.0818, "step": 1306 }, { "epoch": 0.3972644376899696, "grad_norm": 0.275390625, "learning_rate": 6.027355623100305e-06, "loss": 1.1133, "step": 1307 }, { "epoch": 0.39756838905775077, "grad_norm": 0.3359375, "learning_rate": 6.024316109422493e-06, "loss": 1.0309, "step": 1308 }, { "epoch": 0.39787234042553193, "grad_norm": 0.27734375, "learning_rate": 6.021276595744682e-06, "loss": 1.0626, "step": 1309 }, { "epoch": 0.3981762917933131, "grad_norm": 0.29296875, "learning_rate": 6.01823708206687e-06, "loss": 1.0937, "step": 1310 }, { "epoch": 0.3984802431610942, "grad_norm": 0.2734375, "learning_rate": 6.015197568389058e-06, "loss": 1.0569, "step": 1311 }, { "epoch": 0.39878419452887537, "grad_norm": 0.265625, "learning_rate": 6.012158054711247e-06, "loss": 1.0905, "step": 1312 }, { "epoch": 0.39908814589665653, "grad_norm": 0.283203125, "learning_rate": 6.009118541033435e-06, "loss": 1.1056, "step": 1313 }, { "epoch": 0.3993920972644377, "grad_norm": 0.26171875, "learning_rate": 6.006079027355624e-06, "loss": 1.0705, "step": 1314 }, { "epoch": 0.39969604863221886, "grad_norm": 0.271484375, "learning_rate": 6.003039513677812e-06, "loss": 1.0041, "step": 1315 }, { "epoch": 0.4, "grad_norm": 0.27734375, "learning_rate": 6e-06, "loss": 1.0449, "step": 1316 }, { "epoch": 0.40030395136778113, "grad_norm": 0.271484375, "learning_rate": 5.996960486322189e-06, "loss": 1.0381, "step": 1317 }, { "epoch": 0.4006079027355623, "grad_norm": 0.259765625, "learning_rate": 5.993920972644377e-06, "loss": 1.0405, "step": 1318 }, { "epoch": 0.40091185410334346, "grad_norm": 0.298828125, "learning_rate": 5.990881458966565e-06, "loss": 1.0603, "step": 1319 }, { "epoch": 0.4012158054711246, "grad_norm": 0.279296875, "learning_rate": 5.987841945288755e-06, "loss": 1.0325, "step": 1320 }, { "epoch": 0.4015197568389058, "grad_norm": 0.267578125, "learning_rate": 5.984802431610944e-06, "loss": 1.0358, "step": 1321 }, { "epoch": 0.40182370820668695, "grad_norm": 0.271484375, "learning_rate": 5.981762917933131e-06, "loss": 1.034, "step": 1322 }, { "epoch": 0.4021276595744681, "grad_norm": 0.27734375, "learning_rate": 5.97872340425532e-06, "loss": 1.034, "step": 1323 }, { "epoch": 0.4024316109422492, "grad_norm": 0.2734375, "learning_rate": 5.975683890577508e-06, "loss": 1.0393, "step": 1324 }, { "epoch": 0.4027355623100304, "grad_norm": 0.271484375, "learning_rate": 5.972644376899697e-06, "loss": 0.9965, "step": 1325 }, { "epoch": 0.40303951367781155, "grad_norm": 0.271484375, "learning_rate": 5.969604863221885e-06, "loss": 1.1066, "step": 1326 }, { "epoch": 0.4033434650455927, "grad_norm": 0.267578125, "learning_rate": 5.966565349544073e-06, "loss": 0.9945, "step": 1327 }, { "epoch": 0.40364741641337387, "grad_norm": 0.265625, "learning_rate": 5.963525835866262e-06, "loss": 1.0729, "step": 1328 }, { "epoch": 0.40395136778115504, "grad_norm": 0.263671875, "learning_rate": 5.9604863221884505e-06, "loss": 1.1076, "step": 1329 }, { "epoch": 0.40425531914893614, "grad_norm": 0.2734375, "learning_rate": 5.957446808510638e-06, "loss": 1.0016, "step": 1330 }, { "epoch": 0.4045592705167173, "grad_norm": 0.26953125, "learning_rate": 5.954407294832827e-06, "loss": 1.1149, "step": 1331 }, { "epoch": 0.40486322188449847, "grad_norm": 0.275390625, "learning_rate": 5.951367781155015e-06, "loss": 1.0636, "step": 1332 }, { "epoch": 0.40516717325227963, "grad_norm": 0.267578125, "learning_rate": 5.9483282674772036e-06, "loss": 1.0011, "step": 1333 }, { "epoch": 0.4054711246200608, "grad_norm": 0.2734375, "learning_rate": 5.945288753799393e-06, "loss": 1.0109, "step": 1334 }, { "epoch": 0.40577507598784196, "grad_norm": 0.28515625, "learning_rate": 5.942249240121582e-06, "loss": 0.9464, "step": 1335 }, { "epoch": 0.4060790273556231, "grad_norm": 0.26953125, "learning_rate": 5.93920972644377e-06, "loss": 1.0821, "step": 1336 }, { "epoch": 0.40638297872340423, "grad_norm": 0.271484375, "learning_rate": 5.936170212765958e-06, "loss": 1.0035, "step": 1337 }, { "epoch": 0.4066869300911854, "grad_norm": 0.283203125, "learning_rate": 5.933130699088146e-06, "loss": 1.0636, "step": 1338 }, { "epoch": 0.40699088145896656, "grad_norm": 0.283203125, "learning_rate": 5.930091185410335e-06, "loss": 1.036, "step": 1339 }, { "epoch": 0.4072948328267477, "grad_norm": 0.2734375, "learning_rate": 5.9270516717325235e-06, "loss": 1.0507, "step": 1340 }, { "epoch": 0.4075987841945289, "grad_norm": 0.283203125, "learning_rate": 5.924012158054711e-06, "loss": 1.1475, "step": 1341 }, { "epoch": 0.40790273556231005, "grad_norm": 0.322265625, "learning_rate": 5.9209726443769e-06, "loss": 0.9847, "step": 1342 }, { "epoch": 0.4082066869300912, "grad_norm": 0.265625, "learning_rate": 5.917933130699089e-06, "loss": 1.0683, "step": 1343 }, { "epoch": 0.4085106382978723, "grad_norm": 0.318359375, "learning_rate": 5.9148936170212766e-06, "loss": 0.9945, "step": 1344 }, { "epoch": 0.4088145896656535, "grad_norm": 0.28125, "learning_rate": 5.911854103343465e-06, "loss": 1.1052, "step": 1345 }, { "epoch": 0.40911854103343465, "grad_norm": 0.2734375, "learning_rate": 5.908814589665653e-06, "loss": 1.05, "step": 1346 }, { "epoch": 0.4094224924012158, "grad_norm": 0.275390625, "learning_rate": 5.905775075987843e-06, "loss": 1.0716, "step": 1347 }, { "epoch": 0.409726443768997, "grad_norm": 0.267578125, "learning_rate": 5.902735562310031e-06, "loss": 1.0524, "step": 1348 }, { "epoch": 0.41003039513677814, "grad_norm": 0.26953125, "learning_rate": 5.89969604863222e-06, "loss": 1.005, "step": 1349 }, { "epoch": 0.41033434650455924, "grad_norm": 0.26953125, "learning_rate": 5.896656534954408e-06, "loss": 1.0683, "step": 1350 }, { "epoch": 0.4106382978723404, "grad_norm": 0.2734375, "learning_rate": 5.8936170212765965e-06, "loss": 1.0904, "step": 1351 }, { "epoch": 0.41094224924012157, "grad_norm": 0.259765625, "learning_rate": 5.890577507598785e-06, "loss": 1.0708, "step": 1352 }, { "epoch": 0.41124620060790273, "grad_norm": 0.263671875, "learning_rate": 5.887537993920973e-06, "loss": 1.0861, "step": 1353 }, { "epoch": 0.4115501519756839, "grad_norm": 0.251953125, "learning_rate": 5.884498480243162e-06, "loss": 1.0555, "step": 1354 }, { "epoch": 0.41185410334346506, "grad_norm": 0.26953125, "learning_rate": 5.8814589665653495e-06, "loss": 1.0913, "step": 1355 }, { "epoch": 0.4121580547112462, "grad_norm": 0.2734375, "learning_rate": 5.878419452887538e-06, "loss": 1.0258, "step": 1356 }, { "epoch": 0.41246200607902733, "grad_norm": 0.2734375, "learning_rate": 5.875379939209727e-06, "loss": 0.9939, "step": 1357 }, { "epoch": 0.4127659574468085, "grad_norm": 0.2734375, "learning_rate": 5.872340425531915e-06, "loss": 1.1127, "step": 1358 }, { "epoch": 0.41306990881458966, "grad_norm": 0.26953125, "learning_rate": 5.8693009118541034e-06, "loss": 1.0387, "step": 1359 }, { "epoch": 0.4133738601823708, "grad_norm": 0.271484375, "learning_rate": 5.866261398176293e-06, "loss": 1.0812, "step": 1360 }, { "epoch": 0.413677811550152, "grad_norm": 0.263671875, "learning_rate": 5.863221884498481e-06, "loss": 1.0269, "step": 1361 }, { "epoch": 0.41398176291793315, "grad_norm": 0.26171875, "learning_rate": 5.8601823708206695e-06, "loss": 1.0944, "step": 1362 }, { "epoch": 0.4142857142857143, "grad_norm": 0.29296875, "learning_rate": 5.857142857142858e-06, "loss": 1.0927, "step": 1363 }, { "epoch": 0.4145896656534954, "grad_norm": 0.275390625, "learning_rate": 5.854103343465046e-06, "loss": 1.0798, "step": 1364 }, { "epoch": 0.4148936170212766, "grad_norm": 0.2890625, "learning_rate": 5.851063829787235e-06, "loss": 1.0962, "step": 1365 }, { "epoch": 0.41519756838905775, "grad_norm": 0.275390625, "learning_rate": 5.848024316109423e-06, "loss": 1.0358, "step": 1366 }, { "epoch": 0.4155015197568389, "grad_norm": 0.2734375, "learning_rate": 5.844984802431611e-06, "loss": 0.9598, "step": 1367 }, { "epoch": 0.4158054711246201, "grad_norm": 0.2734375, "learning_rate": 5.8419452887538e-06, "loss": 1.0577, "step": 1368 }, { "epoch": 0.41610942249240124, "grad_norm": 0.27734375, "learning_rate": 5.838905775075988e-06, "loss": 1.0045, "step": 1369 }, { "epoch": 0.41641337386018235, "grad_norm": 0.28515625, "learning_rate": 5.8358662613981764e-06, "loss": 1.0271, "step": 1370 }, { "epoch": 0.4167173252279635, "grad_norm": 0.26953125, "learning_rate": 5.832826747720365e-06, "loss": 1.0607, "step": 1371 }, { "epoch": 0.41702127659574467, "grad_norm": 0.3828125, "learning_rate": 5.829787234042553e-06, "loss": 0.9929, "step": 1372 }, { "epoch": 0.41732522796352584, "grad_norm": 0.2890625, "learning_rate": 5.826747720364742e-06, "loss": 1.1061, "step": 1373 }, { "epoch": 0.417629179331307, "grad_norm": 0.279296875, "learning_rate": 5.823708206686931e-06, "loss": 1.1267, "step": 1374 }, { "epoch": 0.41793313069908816, "grad_norm": 0.263671875, "learning_rate": 5.820668693009119e-06, "loss": 0.9701, "step": 1375 }, { "epoch": 0.4182370820668693, "grad_norm": 0.287109375, "learning_rate": 5.817629179331308e-06, "loss": 0.9893, "step": 1376 }, { "epoch": 0.41854103343465043, "grad_norm": 0.279296875, "learning_rate": 5.814589665653496e-06, "loss": 1.0481, "step": 1377 }, { "epoch": 0.4188449848024316, "grad_norm": 0.26953125, "learning_rate": 5.811550151975684e-06, "loss": 1.1125, "step": 1378 }, { "epoch": 0.41914893617021276, "grad_norm": 0.2734375, "learning_rate": 5.808510638297873e-06, "loss": 1.0137, "step": 1379 }, { "epoch": 0.4194528875379939, "grad_norm": 0.296875, "learning_rate": 5.805471124620062e-06, "loss": 1.0047, "step": 1380 }, { "epoch": 0.4197568389057751, "grad_norm": 0.26953125, "learning_rate": 5.8024316109422494e-06, "loss": 1.1128, "step": 1381 }, { "epoch": 0.42006079027355625, "grad_norm": 0.296875, "learning_rate": 5.799392097264438e-06, "loss": 1.0876, "step": 1382 }, { "epoch": 0.42036474164133736, "grad_norm": 0.279296875, "learning_rate": 5.796352583586626e-06, "loss": 1.0618, "step": 1383 }, { "epoch": 0.4206686930091185, "grad_norm": 0.31640625, "learning_rate": 5.793313069908815e-06, "loss": 1.0708, "step": 1384 }, { "epoch": 0.4209726443768997, "grad_norm": 0.267578125, "learning_rate": 5.790273556231003e-06, "loss": 1.0447, "step": 1385 }, { "epoch": 0.42127659574468085, "grad_norm": 0.275390625, "learning_rate": 5.787234042553191e-06, "loss": 1.0951, "step": 1386 }, { "epoch": 0.421580547112462, "grad_norm": 0.27734375, "learning_rate": 5.784194528875381e-06, "loss": 1.1157, "step": 1387 }, { "epoch": 0.4218844984802432, "grad_norm": 0.275390625, "learning_rate": 5.781155015197569e-06, "loss": 1.0442, "step": 1388 }, { "epoch": 0.42218844984802434, "grad_norm": 0.265625, "learning_rate": 5.778115501519758e-06, "loss": 1.0537, "step": 1389 }, { "epoch": 0.42249240121580545, "grad_norm": 0.271484375, "learning_rate": 5.775075987841946e-06, "loss": 1.0484, "step": 1390 }, { "epoch": 0.4227963525835866, "grad_norm": 0.283203125, "learning_rate": 5.772036474164135e-06, "loss": 1.0832, "step": 1391 }, { "epoch": 0.4231003039513678, "grad_norm": 0.2734375, "learning_rate": 5.768996960486322e-06, "loss": 1.021, "step": 1392 }, { "epoch": 0.42340425531914894, "grad_norm": 0.26953125, "learning_rate": 5.765957446808511e-06, "loss": 0.993, "step": 1393 }, { "epoch": 0.4237082066869301, "grad_norm": 0.2734375, "learning_rate": 5.7629179331307e-06, "loss": 1.1607, "step": 1394 }, { "epoch": 0.42401215805471126, "grad_norm": 0.26953125, "learning_rate": 5.759878419452888e-06, "loss": 1.064, "step": 1395 }, { "epoch": 0.4243161094224924, "grad_norm": 0.259765625, "learning_rate": 5.756838905775076e-06, "loss": 1.0893, "step": 1396 }, { "epoch": 0.42462006079027353, "grad_norm": 0.263671875, "learning_rate": 5.753799392097264e-06, "loss": 1.1539, "step": 1397 }, { "epoch": 0.4249240121580547, "grad_norm": 0.271484375, "learning_rate": 5.750759878419453e-06, "loss": 1.0041, "step": 1398 }, { "epoch": 0.42522796352583586, "grad_norm": 0.279296875, "learning_rate": 5.7477203647416415e-06, "loss": 1.1022, "step": 1399 }, { "epoch": 0.425531914893617, "grad_norm": 0.26953125, "learning_rate": 5.744680851063831e-06, "loss": 1.0391, "step": 1400 }, { "epoch": 0.4258358662613982, "grad_norm": 0.28515625, "learning_rate": 5.741641337386019e-06, "loss": 0.9389, "step": 1401 }, { "epoch": 0.42613981762917935, "grad_norm": 0.2734375, "learning_rate": 5.7386018237082076e-06, "loss": 1.0753, "step": 1402 }, { "epoch": 0.42644376899696046, "grad_norm": 0.275390625, "learning_rate": 5.735562310030396e-06, "loss": 1.0515, "step": 1403 }, { "epoch": 0.4267477203647416, "grad_norm": 0.2734375, "learning_rate": 5.732522796352584e-06, "loss": 1.0921, "step": 1404 }, { "epoch": 0.4270516717325228, "grad_norm": 0.2734375, "learning_rate": 5.729483282674773e-06, "loss": 1.0685, "step": 1405 }, { "epoch": 0.42735562310030395, "grad_norm": 0.259765625, "learning_rate": 5.726443768996961e-06, "loss": 1.0772, "step": 1406 }, { "epoch": 0.4276595744680851, "grad_norm": 0.283203125, "learning_rate": 5.723404255319149e-06, "loss": 1.1009, "step": 1407 }, { "epoch": 0.4279635258358663, "grad_norm": 0.2734375, "learning_rate": 5.720364741641338e-06, "loss": 1.0766, "step": 1408 }, { "epoch": 0.42826747720364744, "grad_norm": 0.263671875, "learning_rate": 5.717325227963526e-06, "loss": 1.0612, "step": 1409 }, { "epoch": 0.42857142857142855, "grad_norm": 0.283203125, "learning_rate": 5.7142857142857145e-06, "loss": 1.0765, "step": 1410 }, { "epoch": 0.4288753799392097, "grad_norm": 0.263671875, "learning_rate": 5.711246200607903e-06, "loss": 1.0666, "step": 1411 }, { "epoch": 0.4291793313069909, "grad_norm": 0.26953125, "learning_rate": 5.708206686930091e-06, "loss": 1.0683, "step": 1412 }, { "epoch": 0.42948328267477204, "grad_norm": 0.70703125, "learning_rate": 5.7051671732522806e-06, "loss": 0.9997, "step": 1413 }, { "epoch": 0.4297872340425532, "grad_norm": 0.310546875, "learning_rate": 5.702127659574469e-06, "loss": 1.1176, "step": 1414 }, { "epoch": 0.43009118541033436, "grad_norm": 0.27734375, "learning_rate": 5.699088145896657e-06, "loss": 0.9799, "step": 1415 }, { "epoch": 0.43039513677811553, "grad_norm": 0.275390625, "learning_rate": 5.696048632218846e-06, "loss": 1.0016, "step": 1416 }, { "epoch": 0.43069908814589664, "grad_norm": 0.275390625, "learning_rate": 5.6930091185410345e-06, "loss": 1.0598, "step": 1417 }, { "epoch": 0.4310030395136778, "grad_norm": 0.28125, "learning_rate": 5.689969604863222e-06, "loss": 1.0307, "step": 1418 }, { "epoch": 0.43130699088145896, "grad_norm": 0.287109375, "learning_rate": 5.686930091185411e-06, "loss": 1.0749, "step": 1419 }, { "epoch": 0.4316109422492401, "grad_norm": 0.271484375, "learning_rate": 5.683890577507599e-06, "loss": 1.0041, "step": 1420 }, { "epoch": 0.4319148936170213, "grad_norm": 0.3046875, "learning_rate": 5.6808510638297875e-06, "loss": 1.07, "step": 1421 }, { "epoch": 0.43221884498480245, "grad_norm": 0.267578125, "learning_rate": 5.677811550151976e-06, "loss": 1.0026, "step": 1422 }, { "epoch": 0.43252279635258356, "grad_norm": 0.263671875, "learning_rate": 5.674772036474164e-06, "loss": 1.1045, "step": 1423 }, { "epoch": 0.4328267477203647, "grad_norm": 0.27734375, "learning_rate": 5.671732522796353e-06, "loss": 1.0917, "step": 1424 }, { "epoch": 0.4331306990881459, "grad_norm": 0.267578125, "learning_rate": 5.668693009118541e-06, "loss": 1.0392, "step": 1425 }, { "epoch": 0.43343465045592705, "grad_norm": 0.28515625, "learning_rate": 5.665653495440729e-06, "loss": 1.027, "step": 1426 }, { "epoch": 0.4337386018237082, "grad_norm": 0.298828125, "learning_rate": 5.662613981762919e-06, "loss": 1.0487, "step": 1427 }, { "epoch": 0.4340425531914894, "grad_norm": 0.2734375, "learning_rate": 5.6595744680851075e-06, "loss": 1.0196, "step": 1428 }, { "epoch": 0.43434650455927054, "grad_norm": 0.275390625, "learning_rate": 5.656534954407295e-06, "loss": 1.0299, "step": 1429 }, { "epoch": 0.43465045592705165, "grad_norm": 0.27734375, "learning_rate": 5.653495440729484e-06, "loss": 0.9696, "step": 1430 }, { "epoch": 0.4349544072948328, "grad_norm": 0.314453125, "learning_rate": 5.650455927051673e-06, "loss": 0.9869, "step": 1431 }, { "epoch": 0.435258358662614, "grad_norm": 0.265625, "learning_rate": 5.6474164133738605e-06, "loss": 1.0785, "step": 1432 }, { "epoch": 0.43556231003039514, "grad_norm": 0.26953125, "learning_rate": 5.644376899696049e-06, "loss": 1.0669, "step": 1433 }, { "epoch": 0.4358662613981763, "grad_norm": 0.28515625, "learning_rate": 5.641337386018237e-06, "loss": 1.0073, "step": 1434 }, { "epoch": 0.43617021276595747, "grad_norm": 0.283203125, "learning_rate": 5.638297872340426e-06, "loss": 1.073, "step": 1435 }, { "epoch": 0.43647416413373863, "grad_norm": 0.26953125, "learning_rate": 5.635258358662614e-06, "loss": 1.0961, "step": 1436 }, { "epoch": 0.43677811550151974, "grad_norm": 0.28515625, "learning_rate": 5.632218844984802e-06, "loss": 1.11, "step": 1437 }, { "epoch": 0.4370820668693009, "grad_norm": 0.28515625, "learning_rate": 5.629179331306991e-06, "loss": 1.044, "step": 1438 }, { "epoch": 0.43738601823708206, "grad_norm": 0.275390625, "learning_rate": 5.62613981762918e-06, "loss": 1.0956, "step": 1439 }, { "epoch": 0.4376899696048632, "grad_norm": 0.265625, "learning_rate": 5.623100303951369e-06, "loss": 1.0897, "step": 1440 }, { "epoch": 0.4379939209726444, "grad_norm": 0.26953125, "learning_rate": 5.620060790273557e-06, "loss": 1.0816, "step": 1441 }, { "epoch": 0.43829787234042555, "grad_norm": 0.26953125, "learning_rate": 5.617021276595746e-06, "loss": 1.0297, "step": 1442 }, { "epoch": 0.43860182370820666, "grad_norm": 0.2734375, "learning_rate": 5.6139817629179335e-06, "loss": 1.1134, "step": 1443 }, { "epoch": 0.4389057750759878, "grad_norm": 0.275390625, "learning_rate": 5.610942249240122e-06, "loss": 0.9807, "step": 1444 }, { "epoch": 0.439209726443769, "grad_norm": 0.27734375, "learning_rate": 5.607902735562311e-06, "loss": 1.052, "step": 1445 }, { "epoch": 0.43951367781155015, "grad_norm": 0.26171875, "learning_rate": 5.604863221884499e-06, "loss": 1.0064, "step": 1446 }, { "epoch": 0.4398176291793313, "grad_norm": 0.462890625, "learning_rate": 5.601823708206687e-06, "loss": 1.085, "step": 1447 }, { "epoch": 0.4401215805471125, "grad_norm": 0.26953125, "learning_rate": 5.598784194528876e-06, "loss": 1.022, "step": 1448 }, { "epoch": 0.44042553191489364, "grad_norm": 0.275390625, "learning_rate": 5.595744680851064e-06, "loss": 1.0853, "step": 1449 }, { "epoch": 0.44072948328267475, "grad_norm": 0.279296875, "learning_rate": 5.592705167173253e-06, "loss": 1.0287, "step": 1450 }, { "epoch": 0.4410334346504559, "grad_norm": 0.376953125, "learning_rate": 5.5896656534954404e-06, "loss": 1.074, "step": 1451 }, { "epoch": 0.4413373860182371, "grad_norm": 0.294921875, "learning_rate": 5.586626139817629e-06, "loss": 1.1117, "step": 1452 }, { "epoch": 0.44164133738601824, "grad_norm": 0.291015625, "learning_rate": 5.583586626139819e-06, "loss": 1.0224, "step": 1453 }, { "epoch": 0.4419452887537994, "grad_norm": 0.271484375, "learning_rate": 5.580547112462007e-06, "loss": 1.0818, "step": 1454 }, { "epoch": 0.44224924012158057, "grad_norm": 0.287109375, "learning_rate": 5.577507598784195e-06, "loss": 1.0175, "step": 1455 }, { "epoch": 0.4425531914893617, "grad_norm": 0.3046875, "learning_rate": 5.574468085106384e-06, "loss": 1.0644, "step": 1456 }, { "epoch": 0.44285714285714284, "grad_norm": 0.27734375, "learning_rate": 5.571428571428572e-06, "loss": 1.1023, "step": 1457 }, { "epoch": 0.443161094224924, "grad_norm": 0.279296875, "learning_rate": 5.56838905775076e-06, "loss": 1.0117, "step": 1458 }, { "epoch": 0.44346504559270516, "grad_norm": 0.2734375, "learning_rate": 5.565349544072949e-06, "loss": 1.042, "step": 1459 }, { "epoch": 0.44376899696048633, "grad_norm": 0.26953125, "learning_rate": 5.562310030395137e-06, "loss": 1.0468, "step": 1460 }, { "epoch": 0.4440729483282675, "grad_norm": 0.267578125, "learning_rate": 5.559270516717326e-06, "loss": 1.0889, "step": 1461 }, { "epoch": 0.44437689969604866, "grad_norm": 0.275390625, "learning_rate": 5.556231003039514e-06, "loss": 1.1092, "step": 1462 }, { "epoch": 0.44468085106382976, "grad_norm": 0.27734375, "learning_rate": 5.553191489361702e-06, "loss": 1.0678, "step": 1463 }, { "epoch": 0.4449848024316109, "grad_norm": 0.28515625, "learning_rate": 5.550151975683891e-06, "loss": 0.9944, "step": 1464 }, { "epoch": 0.4452887537993921, "grad_norm": 0.29296875, "learning_rate": 5.547112462006079e-06, "loss": 1.0671, "step": 1465 }, { "epoch": 0.44559270516717325, "grad_norm": 1.3984375, "learning_rate": 5.544072948328268e-06, "loss": 1.101, "step": 1466 }, { "epoch": 0.4458966565349544, "grad_norm": 0.275390625, "learning_rate": 5.541033434650457e-06, "loss": 1.0147, "step": 1467 }, { "epoch": 0.4462006079027356, "grad_norm": 0.271484375, "learning_rate": 5.5379939209726455e-06, "loss": 1.1809, "step": 1468 }, { "epoch": 0.44650455927051674, "grad_norm": 0.2734375, "learning_rate": 5.534954407294833e-06, "loss": 1.1128, "step": 1469 }, { "epoch": 0.44680851063829785, "grad_norm": 0.27734375, "learning_rate": 5.531914893617022e-06, "loss": 1.0619, "step": 1470 }, { "epoch": 0.447112462006079, "grad_norm": 0.291015625, "learning_rate": 5.52887537993921e-06, "loss": 1.0417, "step": 1471 }, { "epoch": 0.4474164133738602, "grad_norm": 0.279296875, "learning_rate": 5.5258358662613986e-06, "loss": 1.0563, "step": 1472 }, { "epoch": 0.44772036474164134, "grad_norm": 0.31640625, "learning_rate": 5.522796352583587e-06, "loss": 1.0313, "step": 1473 }, { "epoch": 0.4480243161094225, "grad_norm": 0.275390625, "learning_rate": 5.519756838905775e-06, "loss": 1.0908, "step": 1474 }, { "epoch": 0.44832826747720367, "grad_norm": 0.28515625, "learning_rate": 5.516717325227964e-06, "loss": 0.9894, "step": 1475 }, { "epoch": 0.4486322188449848, "grad_norm": 0.275390625, "learning_rate": 5.5136778115501525e-06, "loss": 0.9792, "step": 1476 }, { "epoch": 0.44893617021276594, "grad_norm": 0.271484375, "learning_rate": 5.51063829787234e-06, "loss": 1.052, "step": 1477 }, { "epoch": 0.4492401215805471, "grad_norm": 0.279296875, "learning_rate": 5.507598784194529e-06, "loss": 1.0498, "step": 1478 }, { "epoch": 0.44954407294832827, "grad_norm": 0.283203125, "learning_rate": 5.504559270516717e-06, "loss": 1.0726, "step": 1479 }, { "epoch": 0.44984802431610943, "grad_norm": 0.26953125, "learning_rate": 5.501519756838906e-06, "loss": 1.037, "step": 1480 }, { "epoch": 0.4501519756838906, "grad_norm": 0.275390625, "learning_rate": 5.498480243161095e-06, "loss": 1.0348, "step": 1481 }, { "epoch": 0.45045592705167176, "grad_norm": 0.26953125, "learning_rate": 5.495440729483284e-06, "loss": 1.1525, "step": 1482 }, { "epoch": 0.45075987841945286, "grad_norm": 0.271484375, "learning_rate": 5.4924012158054716e-06, "loss": 0.9548, "step": 1483 }, { "epoch": 0.451063829787234, "grad_norm": 0.275390625, "learning_rate": 5.48936170212766e-06, "loss": 1.0385, "step": 1484 }, { "epoch": 0.4513677811550152, "grad_norm": 0.271484375, "learning_rate": 5.486322188449849e-06, "loss": 0.9956, "step": 1485 }, { "epoch": 0.45167173252279635, "grad_norm": 0.291015625, "learning_rate": 5.483282674772037e-06, "loss": 1.0347, "step": 1486 }, { "epoch": 0.4519756838905775, "grad_norm": 0.28515625, "learning_rate": 5.4802431610942255e-06, "loss": 0.992, "step": 1487 }, { "epoch": 0.4522796352583587, "grad_norm": 0.28125, "learning_rate": 5.477203647416413e-06, "loss": 0.9476, "step": 1488 }, { "epoch": 0.45258358662613984, "grad_norm": 0.296875, "learning_rate": 5.474164133738602e-06, "loss": 1.08, "step": 1489 }, { "epoch": 0.45288753799392095, "grad_norm": 0.26953125, "learning_rate": 5.471124620060791e-06, "loss": 1.0224, "step": 1490 }, { "epoch": 0.4531914893617021, "grad_norm": 0.271484375, "learning_rate": 5.4680851063829785e-06, "loss": 0.9936, "step": 1491 }, { "epoch": 0.4534954407294833, "grad_norm": 0.2734375, "learning_rate": 5.465045592705167e-06, "loss": 1.0146, "step": 1492 }, { "epoch": 0.45379939209726444, "grad_norm": 0.26953125, "learning_rate": 5.462006079027357e-06, "loss": 1.1068, "step": 1493 }, { "epoch": 0.4541033434650456, "grad_norm": 0.28515625, "learning_rate": 5.4589665653495446e-06, "loss": 1.1023, "step": 1494 }, { "epoch": 0.45440729483282677, "grad_norm": 0.27734375, "learning_rate": 5.455927051671733e-06, "loss": 1.0706, "step": 1495 }, { "epoch": 0.4547112462006079, "grad_norm": 0.287109375, "learning_rate": 5.452887537993922e-06, "loss": 1.1302, "step": 1496 }, { "epoch": 0.45501519756838904, "grad_norm": 0.265625, "learning_rate": 5.44984802431611e-06, "loss": 1.0029, "step": 1497 }, { "epoch": 0.4553191489361702, "grad_norm": 0.275390625, "learning_rate": 5.4468085106382985e-06, "loss": 1.0566, "step": 1498 }, { "epoch": 0.45562310030395137, "grad_norm": 0.271484375, "learning_rate": 5.443768996960487e-06, "loss": 1.0263, "step": 1499 }, { "epoch": 0.45592705167173253, "grad_norm": 0.28125, "learning_rate": 5.440729483282675e-06, "loss": 1.0633, "step": 1500 }, { "epoch": 0.4562310030395137, "grad_norm": 0.28125, "learning_rate": 5.437689969604864e-06, "loss": 1.067, "step": 1501 }, { "epoch": 0.45653495440729486, "grad_norm": 0.287109375, "learning_rate": 5.4346504559270515e-06, "loss": 1.0297, "step": 1502 }, { "epoch": 0.45683890577507597, "grad_norm": 0.28515625, "learning_rate": 5.43161094224924e-06, "loss": 1.0001, "step": 1503 }, { "epoch": 0.45714285714285713, "grad_norm": 0.3046875, "learning_rate": 5.428571428571429e-06, "loss": 1.0939, "step": 1504 }, { "epoch": 0.4574468085106383, "grad_norm": 0.40625, "learning_rate": 5.425531914893617e-06, "loss": 1.048, "step": 1505 }, { "epoch": 0.45775075987841946, "grad_norm": 0.275390625, "learning_rate": 5.422492401215806e-06, "loss": 1.0283, "step": 1506 }, { "epoch": 0.4580547112462006, "grad_norm": 0.291015625, "learning_rate": 5.419452887537995e-06, "loss": 1.0926, "step": 1507 }, { "epoch": 0.4583586626139818, "grad_norm": 0.265625, "learning_rate": 5.416413373860183e-06, "loss": 1.0787, "step": 1508 }, { "epoch": 0.45866261398176295, "grad_norm": 0.27734375, "learning_rate": 5.4133738601823714e-06, "loss": 1.0099, "step": 1509 }, { "epoch": 0.45896656534954405, "grad_norm": 0.2890625, "learning_rate": 5.41033434650456e-06, "loss": 1.072, "step": 1510 }, { "epoch": 0.4592705167173252, "grad_norm": 0.26953125, "learning_rate": 5.407294832826748e-06, "loss": 1.0628, "step": 1511 }, { "epoch": 0.4595744680851064, "grad_norm": 0.2734375, "learning_rate": 5.404255319148937e-06, "loss": 1.0982, "step": 1512 }, { "epoch": 0.45987841945288754, "grad_norm": 0.27734375, "learning_rate": 5.401215805471125e-06, "loss": 1.0023, "step": 1513 }, { "epoch": 0.4601823708206687, "grad_norm": 0.298828125, "learning_rate": 5.398176291793313e-06, "loss": 1.0443, "step": 1514 }, { "epoch": 0.46048632218844987, "grad_norm": 0.28515625, "learning_rate": 5.395136778115502e-06, "loss": 1.0802, "step": 1515 }, { "epoch": 0.460790273556231, "grad_norm": 0.287109375, "learning_rate": 5.39209726443769e-06, "loss": 1.0102, "step": 1516 }, { "epoch": 0.46109422492401214, "grad_norm": 0.271484375, "learning_rate": 5.389057750759878e-06, "loss": 1.0885, "step": 1517 }, { "epoch": 0.4613981762917933, "grad_norm": 0.28515625, "learning_rate": 5.386018237082067e-06, "loss": 1.0393, "step": 1518 }, { "epoch": 0.46170212765957447, "grad_norm": 0.28515625, "learning_rate": 5.382978723404257e-06, "loss": 1.0843, "step": 1519 }, { "epoch": 0.46200607902735563, "grad_norm": 0.275390625, "learning_rate": 5.3799392097264444e-06, "loss": 0.998, "step": 1520 }, { "epoch": 0.4623100303951368, "grad_norm": 0.291015625, "learning_rate": 5.376899696048633e-06, "loss": 1.11, "step": 1521 }, { "epoch": 0.46261398176291796, "grad_norm": 0.283203125, "learning_rate": 5.373860182370822e-06, "loss": 1.0336, "step": 1522 }, { "epoch": 0.46291793313069907, "grad_norm": 0.27734375, "learning_rate": 5.37082066869301e-06, "loss": 1.0772, "step": 1523 }, { "epoch": 0.46322188449848023, "grad_norm": 0.279296875, "learning_rate": 5.367781155015198e-06, "loss": 1.0616, "step": 1524 }, { "epoch": 0.4635258358662614, "grad_norm": 0.28125, "learning_rate": 5.364741641337386e-06, "loss": 1.0499, "step": 1525 }, { "epoch": 0.46382978723404256, "grad_norm": 0.30859375, "learning_rate": 5.361702127659575e-06, "loss": 1.0442, "step": 1526 }, { "epoch": 0.4641337386018237, "grad_norm": 0.28515625, "learning_rate": 5.3586626139817635e-06, "loss": 0.9965, "step": 1527 }, { "epoch": 0.4644376899696049, "grad_norm": 0.275390625, "learning_rate": 5.355623100303951e-06, "loss": 1.0133, "step": 1528 }, { "epoch": 0.464741641337386, "grad_norm": 0.291015625, "learning_rate": 5.35258358662614e-06, "loss": 1.0123, "step": 1529 }, { "epoch": 0.46504559270516715, "grad_norm": 0.27734375, "learning_rate": 5.349544072948329e-06, "loss": 1.0059, "step": 1530 }, { "epoch": 0.4653495440729483, "grad_norm": 0.28125, "learning_rate": 5.346504559270517e-06, "loss": 1.0036, "step": 1531 }, { "epoch": 0.4656534954407295, "grad_norm": 0.279296875, "learning_rate": 5.343465045592705e-06, "loss": 1.0183, "step": 1532 }, { "epoch": 0.46595744680851064, "grad_norm": 0.283203125, "learning_rate": 5.340425531914895e-06, "loss": 1.0715, "step": 1533 }, { "epoch": 0.4662613981762918, "grad_norm": 0.271484375, "learning_rate": 5.337386018237083e-06, "loss": 1.035, "step": 1534 }, { "epoch": 0.46656534954407297, "grad_norm": 0.28125, "learning_rate": 5.334346504559271e-06, "loss": 1.0792, "step": 1535 }, { "epoch": 0.4668693009118541, "grad_norm": 0.27734375, "learning_rate": 5.33130699088146e-06, "loss": 1.0887, "step": 1536 }, { "epoch": 0.46717325227963524, "grad_norm": 0.28125, "learning_rate": 5.328267477203648e-06, "loss": 0.9987, "step": 1537 }, { "epoch": 0.4674772036474164, "grad_norm": 0.271484375, "learning_rate": 5.3252279635258365e-06, "loss": 1.1046, "step": 1538 }, { "epoch": 0.46778115501519757, "grad_norm": 0.275390625, "learning_rate": 5.322188449848024e-06, "loss": 1.0273, "step": 1539 }, { "epoch": 0.46808510638297873, "grad_norm": 0.26171875, "learning_rate": 5.319148936170213e-06, "loss": 1.0641, "step": 1540 }, { "epoch": 0.4683890577507599, "grad_norm": 0.28515625, "learning_rate": 5.316109422492402e-06, "loss": 1.0688, "step": 1541 }, { "epoch": 0.46869300911854106, "grad_norm": 0.279296875, "learning_rate": 5.31306990881459e-06, "loss": 1.0128, "step": 1542 }, { "epoch": 0.46899696048632217, "grad_norm": 0.318359375, "learning_rate": 5.310030395136778e-06, "loss": 1.0699, "step": 1543 }, { "epoch": 0.46930091185410333, "grad_norm": 0.302734375, "learning_rate": 5.306990881458967e-06, "loss": 0.9921, "step": 1544 }, { "epoch": 0.4696048632218845, "grad_norm": 0.287109375, "learning_rate": 5.303951367781155e-06, "loss": 1.1284, "step": 1545 }, { "epoch": 0.46990881458966566, "grad_norm": 0.29296875, "learning_rate": 5.300911854103344e-06, "loss": 1.0378, "step": 1546 }, { "epoch": 0.4702127659574468, "grad_norm": 0.2734375, "learning_rate": 5.297872340425533e-06, "loss": 1.0725, "step": 1547 }, { "epoch": 0.470516717325228, "grad_norm": 0.26953125, "learning_rate": 5.294832826747721e-06, "loss": 1.0531, "step": 1548 }, { "epoch": 0.4708206686930091, "grad_norm": 0.287109375, "learning_rate": 5.2917933130699095e-06, "loss": 0.9839, "step": 1549 }, { "epoch": 0.47112462006079026, "grad_norm": 0.27734375, "learning_rate": 5.288753799392098e-06, "loss": 1.0816, "step": 1550 }, { "epoch": 0.4714285714285714, "grad_norm": 0.26953125, "learning_rate": 5.285714285714286e-06, "loss": 1.0505, "step": 1551 }, { "epoch": 0.4717325227963526, "grad_norm": 0.271484375, "learning_rate": 5.282674772036475e-06, "loss": 1.0444, "step": 1552 }, { "epoch": 0.47203647416413375, "grad_norm": 0.265625, "learning_rate": 5.2796352583586626e-06, "loss": 1.0664, "step": 1553 }, { "epoch": 0.4723404255319149, "grad_norm": 0.291015625, "learning_rate": 5.276595744680851e-06, "loss": 1.0953, "step": 1554 }, { "epoch": 0.4726443768996961, "grad_norm": 0.8671875, "learning_rate": 5.27355623100304e-06, "loss": 0.9956, "step": 1555 }, { "epoch": 0.4729483282674772, "grad_norm": 0.275390625, "learning_rate": 5.270516717325228e-06, "loss": 1.0463, "step": 1556 }, { "epoch": 0.47325227963525834, "grad_norm": 0.26953125, "learning_rate": 5.2674772036474165e-06, "loss": 1.0897, "step": 1557 }, { "epoch": 0.4735562310030395, "grad_norm": 0.26953125, "learning_rate": 5.264437689969605e-06, "loss": 1.0494, "step": 1558 }, { "epoch": 0.47386018237082067, "grad_norm": 0.271484375, "learning_rate": 5.261398176291795e-06, "loss": 1.0205, "step": 1559 }, { "epoch": 0.47416413373860183, "grad_norm": 0.2734375, "learning_rate": 5.2583586626139825e-06, "loss": 1.0682, "step": 1560 }, { "epoch": 0.474468085106383, "grad_norm": 0.265625, "learning_rate": 5.255319148936171e-06, "loss": 1.1026, "step": 1561 }, { "epoch": 0.47477203647416416, "grad_norm": 0.28125, "learning_rate": 5.252279635258359e-06, "loss": 1.0822, "step": 1562 }, { "epoch": 0.47507598784194527, "grad_norm": 0.28125, "learning_rate": 5.249240121580548e-06, "loss": 1.1008, "step": 1563 }, { "epoch": 0.47537993920972643, "grad_norm": 0.2734375, "learning_rate": 5.246200607902736e-06, "loss": 1.0137, "step": 1564 }, { "epoch": 0.4756838905775076, "grad_norm": 0.271484375, "learning_rate": 5.243161094224924e-06, "loss": 1.0793, "step": 1565 }, { "epoch": 0.47598784194528876, "grad_norm": 0.26953125, "learning_rate": 5.240121580547113e-06, "loss": 1.0412, "step": 1566 }, { "epoch": 0.4762917933130699, "grad_norm": 0.271484375, "learning_rate": 5.237082066869302e-06, "loss": 1.059, "step": 1567 }, { "epoch": 0.4765957446808511, "grad_norm": 0.26953125, "learning_rate": 5.2340425531914895e-06, "loss": 1.1102, "step": 1568 }, { "epoch": 0.4768996960486322, "grad_norm": 0.28515625, "learning_rate": 5.231003039513678e-06, "loss": 1.0715, "step": 1569 }, { "epoch": 0.47720364741641336, "grad_norm": 0.27734375, "learning_rate": 5.227963525835866e-06, "loss": 1.0791, "step": 1570 }, { "epoch": 0.4775075987841945, "grad_norm": 0.283203125, "learning_rate": 5.224924012158055e-06, "loss": 1.0033, "step": 1571 }, { "epoch": 0.4778115501519757, "grad_norm": 0.26953125, "learning_rate": 5.221884498480243e-06, "loss": 1.0687, "step": 1572 }, { "epoch": 0.47811550151975685, "grad_norm": 0.27734375, "learning_rate": 5.218844984802433e-06, "loss": 1.0566, "step": 1573 }, { "epoch": 0.478419452887538, "grad_norm": 0.294921875, "learning_rate": 5.215805471124621e-06, "loss": 1.0048, "step": 1574 }, { "epoch": 0.4787234042553192, "grad_norm": 0.28125, "learning_rate": 5.212765957446809e-06, "loss": 1.0085, "step": 1575 }, { "epoch": 0.4790273556231003, "grad_norm": 0.271484375, "learning_rate": 5.209726443768997e-06, "loss": 1.053, "step": 1576 }, { "epoch": 0.47933130699088144, "grad_norm": 0.26953125, "learning_rate": 5.206686930091186e-06, "loss": 1.0287, "step": 1577 }, { "epoch": 0.4796352583586626, "grad_norm": 0.28515625, "learning_rate": 5.203647416413375e-06, "loss": 1.0538, "step": 1578 }, { "epoch": 0.47993920972644377, "grad_norm": 0.26953125, "learning_rate": 5.2006079027355625e-06, "loss": 1.0447, "step": 1579 }, { "epoch": 0.48024316109422494, "grad_norm": 0.2734375, "learning_rate": 5.197568389057751e-06, "loss": 1.1108, "step": 1580 }, { "epoch": 0.4805471124620061, "grad_norm": 0.2890625, "learning_rate": 5.19452887537994e-06, "loss": 1.1023, "step": 1581 }, { "epoch": 0.4808510638297872, "grad_norm": 0.265625, "learning_rate": 5.191489361702128e-06, "loss": 1.0609, "step": 1582 }, { "epoch": 0.48115501519756837, "grad_norm": 0.29296875, "learning_rate": 5.188449848024316e-06, "loss": 1.0808, "step": 1583 }, { "epoch": 0.48145896656534953, "grad_norm": 0.267578125, "learning_rate": 5.185410334346504e-06, "loss": 1.0796, "step": 1584 }, { "epoch": 0.4817629179331307, "grad_norm": 0.287109375, "learning_rate": 5.182370820668693e-06, "loss": 1.0514, "step": 1585 }, { "epoch": 0.48206686930091186, "grad_norm": 0.28125, "learning_rate": 5.179331306990882e-06, "loss": 1.0905, "step": 1586 }, { "epoch": 0.482370820668693, "grad_norm": 0.279296875, "learning_rate": 5.176291793313071e-06, "loss": 1.0696, "step": 1587 }, { "epoch": 0.4826747720364742, "grad_norm": 0.28125, "learning_rate": 5.173252279635259e-06, "loss": 1.0036, "step": 1588 }, { "epoch": 0.4829787234042553, "grad_norm": 0.283203125, "learning_rate": 5.170212765957448e-06, "loss": 1.0433, "step": 1589 }, { "epoch": 0.48328267477203646, "grad_norm": 0.265625, "learning_rate": 5.1671732522796354e-06, "loss": 1.1414, "step": 1590 }, { "epoch": 0.4835866261398176, "grad_norm": 0.314453125, "learning_rate": 5.164133738601824e-06, "loss": 1.0547, "step": 1591 }, { "epoch": 0.4838905775075988, "grad_norm": 0.259765625, "learning_rate": 5.161094224924013e-06, "loss": 1.0739, "step": 1592 }, { "epoch": 0.48419452887537995, "grad_norm": 0.28125, "learning_rate": 5.158054711246201e-06, "loss": 1.1008, "step": 1593 }, { "epoch": 0.4844984802431611, "grad_norm": 0.263671875, "learning_rate": 5.155015197568389e-06, "loss": 1.1236, "step": 1594 }, { "epoch": 0.4848024316109423, "grad_norm": 0.2734375, "learning_rate": 5.151975683890578e-06, "loss": 0.9998, "step": 1595 }, { "epoch": 0.4851063829787234, "grad_norm": 0.271484375, "learning_rate": 5.148936170212766e-06, "loss": 1.0978, "step": 1596 }, { "epoch": 0.48541033434650455, "grad_norm": 0.28125, "learning_rate": 5.1458966565349545e-06, "loss": 1.0264, "step": 1597 }, { "epoch": 0.4857142857142857, "grad_norm": 0.271484375, "learning_rate": 5.142857142857142e-06, "loss": 0.9808, "step": 1598 }, { "epoch": 0.4860182370820669, "grad_norm": 0.271484375, "learning_rate": 5.139817629179332e-06, "loss": 1.025, "step": 1599 }, { "epoch": 0.48632218844984804, "grad_norm": 0.27734375, "learning_rate": 5.136778115501521e-06, "loss": 1.0047, "step": 1600 }, { "epoch": 0.4866261398176292, "grad_norm": 0.265625, "learning_rate": 5.133738601823709e-06, "loss": 1.0712, "step": 1601 }, { "epoch": 0.4869300911854103, "grad_norm": 0.29296875, "learning_rate": 5.130699088145897e-06, "loss": 1.0328, "step": 1602 }, { "epoch": 0.48723404255319147, "grad_norm": 0.26953125, "learning_rate": 5.127659574468086e-06, "loss": 1.0023, "step": 1603 }, { "epoch": 0.48753799392097263, "grad_norm": 0.27734375, "learning_rate": 5.1246200607902745e-06, "loss": 1.0765, "step": 1604 }, { "epoch": 0.4878419452887538, "grad_norm": 0.26953125, "learning_rate": 5.121580547112462e-06, "loss": 1.0376, "step": 1605 }, { "epoch": 0.48814589665653496, "grad_norm": 0.275390625, "learning_rate": 5.118541033434651e-06, "loss": 1.0239, "step": 1606 }, { "epoch": 0.4884498480243161, "grad_norm": 0.28125, "learning_rate": 5.115501519756839e-06, "loss": 1.0785, "step": 1607 }, { "epoch": 0.4887537993920973, "grad_norm": 0.271484375, "learning_rate": 5.1124620060790275e-06, "loss": 0.9819, "step": 1608 }, { "epoch": 0.4890577507598784, "grad_norm": 0.28515625, "learning_rate": 5.109422492401216e-06, "loss": 1.0149, "step": 1609 }, { "epoch": 0.48936170212765956, "grad_norm": 0.275390625, "learning_rate": 5.106382978723404e-06, "loss": 1.0401, "step": 1610 }, { "epoch": 0.4896656534954407, "grad_norm": 0.291015625, "learning_rate": 5.103343465045593e-06, "loss": 1.0882, "step": 1611 }, { "epoch": 0.4899696048632219, "grad_norm": 0.27734375, "learning_rate": 5.100303951367782e-06, "loss": 1.0704, "step": 1612 }, { "epoch": 0.49027355623100305, "grad_norm": 0.310546875, "learning_rate": 5.09726443768997e-06, "loss": 1.0143, "step": 1613 }, { "epoch": 0.4905775075987842, "grad_norm": 0.265625, "learning_rate": 5.094224924012159e-06, "loss": 1.108, "step": 1614 }, { "epoch": 0.4908814589665654, "grad_norm": 0.267578125, "learning_rate": 5.0911854103343475e-06, "loss": 1.1045, "step": 1615 }, { "epoch": 0.4911854103343465, "grad_norm": 0.271484375, "learning_rate": 5.088145896656535e-06, "loss": 1.0096, "step": 1616 }, { "epoch": 0.49148936170212765, "grad_norm": 0.26953125, "learning_rate": 5.085106382978724e-06, "loss": 1.1126, "step": 1617 }, { "epoch": 0.4917933130699088, "grad_norm": 0.267578125, "learning_rate": 5.082066869300913e-06, "loss": 1.0453, "step": 1618 }, { "epoch": 0.49209726443769, "grad_norm": 0.279296875, "learning_rate": 5.0790273556231005e-06, "loss": 0.9829, "step": 1619 }, { "epoch": 0.49240121580547114, "grad_norm": 0.27734375, "learning_rate": 5.075987841945289e-06, "loss": 1.0834, "step": 1620 }, { "epoch": 0.4927051671732523, "grad_norm": 0.28125, "learning_rate": 5.072948328267477e-06, "loss": 1.0562, "step": 1621 }, { "epoch": 0.4930091185410334, "grad_norm": 0.28515625, "learning_rate": 5.069908814589666e-06, "loss": 1.0375, "step": 1622 }, { "epoch": 0.49331306990881457, "grad_norm": 0.267578125, "learning_rate": 5.066869300911854e-06, "loss": 1.0839, "step": 1623 }, { "epoch": 0.49361702127659574, "grad_norm": 0.283203125, "learning_rate": 5.063829787234042e-06, "loss": 1.0817, "step": 1624 }, { "epoch": 0.4939209726443769, "grad_norm": 0.28515625, "learning_rate": 5.060790273556231e-06, "loss": 1.0556, "step": 1625 }, { "epoch": 0.49422492401215806, "grad_norm": 0.287109375, "learning_rate": 5.0577507598784205e-06, "loss": 1.0107, "step": 1626 }, { "epoch": 0.4945288753799392, "grad_norm": 0.279296875, "learning_rate": 5.054711246200608e-06, "loss": 1.0643, "step": 1627 }, { "epoch": 0.4948328267477204, "grad_norm": 0.28515625, "learning_rate": 5.051671732522797e-06, "loss": 1.0312, "step": 1628 }, { "epoch": 0.4951367781155015, "grad_norm": 0.3125, "learning_rate": 5.048632218844986e-06, "loss": 1.135, "step": 1629 }, { "epoch": 0.49544072948328266, "grad_norm": 0.275390625, "learning_rate": 5.0455927051671735e-06, "loss": 1.0496, "step": 1630 }, { "epoch": 0.4957446808510638, "grad_norm": 0.27734375, "learning_rate": 5.042553191489362e-06, "loss": 1.0419, "step": 1631 }, { "epoch": 0.496048632218845, "grad_norm": 0.263671875, "learning_rate": 5.039513677811551e-06, "loss": 1.0408, "step": 1632 }, { "epoch": 0.49635258358662615, "grad_norm": 0.28515625, "learning_rate": 5.036474164133739e-06, "loss": 1.0425, "step": 1633 }, { "epoch": 0.4966565349544073, "grad_norm": 0.2890625, "learning_rate": 5.033434650455927e-06, "loss": 1.0773, "step": 1634 }, { "epoch": 0.4969604863221885, "grad_norm": 0.275390625, "learning_rate": 5.030395136778115e-06, "loss": 1.0741, "step": 1635 }, { "epoch": 0.4972644376899696, "grad_norm": 0.28125, "learning_rate": 5.027355623100304e-06, "loss": 1.1116, "step": 1636 }, { "epoch": 0.49756838905775075, "grad_norm": 0.294921875, "learning_rate": 5.024316109422493e-06, "loss": 0.9836, "step": 1637 }, { "epoch": 0.4978723404255319, "grad_norm": 0.294921875, "learning_rate": 5.0212765957446805e-06, "loss": 1.0041, "step": 1638 }, { "epoch": 0.4981762917933131, "grad_norm": 0.28125, "learning_rate": 5.01823708206687e-06, "loss": 1.1083, "step": 1639 }, { "epoch": 0.49848024316109424, "grad_norm": 0.26953125, "learning_rate": 5.015197568389059e-06, "loss": 1.0743, "step": 1640 }, { "epoch": 0.4987841945288754, "grad_norm": 0.271484375, "learning_rate": 5.012158054711247e-06, "loss": 1.0481, "step": 1641 }, { "epoch": 0.4990881458966565, "grad_norm": 0.267578125, "learning_rate": 5.009118541033435e-06, "loss": 1.029, "step": 1642 }, { "epoch": 0.4993920972644377, "grad_norm": 0.267578125, "learning_rate": 5.006079027355624e-06, "loss": 1.0465, "step": 1643 }, { "epoch": 0.49969604863221884, "grad_norm": 0.3046875, "learning_rate": 5.003039513677812e-06, "loss": 0.9985, "step": 1644 }, { "epoch": 0.5, "grad_norm": 0.27734375, "learning_rate": 5e-06, "loss": 0.9995, "step": 1645 }, { "epoch": 0.5003039513677812, "grad_norm": 0.27734375, "learning_rate": 4.996960486322189e-06, "loss": 1.0261, "step": 1646 }, { "epoch": 0.5006079027355623, "grad_norm": 0.2890625, "learning_rate": 4.993920972644377e-06, "loss": 1.0594, "step": 1647 }, { "epoch": 0.5009118541033435, "grad_norm": 0.287109375, "learning_rate": 4.990881458966566e-06, "loss": 1.0339, "step": 1648 }, { "epoch": 0.5012158054711247, "grad_norm": 0.28125, "learning_rate": 4.987841945288754e-06, "loss": 1.0442, "step": 1649 }, { "epoch": 0.5015197568389058, "grad_norm": 0.27734375, "learning_rate": 4.984802431610943e-06, "loss": 1.0511, "step": 1650 }, { "epoch": 0.501823708206687, "grad_norm": 0.279296875, "learning_rate": 4.981762917933131e-06, "loss": 1.0138, "step": 1651 }, { "epoch": 0.502127659574468, "grad_norm": 0.275390625, "learning_rate": 4.9787234042553195e-06, "loss": 1.0874, "step": 1652 }, { "epoch": 0.5024316109422492, "grad_norm": 0.30859375, "learning_rate": 4.975683890577508e-06, "loss": 1.0435, "step": 1653 }, { "epoch": 0.5027355623100304, "grad_norm": 0.27734375, "learning_rate": 4.972644376899696e-06, "loss": 1.0122, "step": 1654 }, { "epoch": 0.5030395136778115, "grad_norm": 0.27734375, "learning_rate": 4.9696048632218856e-06, "loss": 1.1249, "step": 1655 }, { "epoch": 0.5033434650455927, "grad_norm": 0.30078125, "learning_rate": 4.966565349544073e-06, "loss": 1.005, "step": 1656 }, { "epoch": 0.5036474164133738, "grad_norm": 0.29296875, "learning_rate": 4.963525835866262e-06, "loss": 0.9957, "step": 1657 }, { "epoch": 0.503951367781155, "grad_norm": 0.28125, "learning_rate": 4.96048632218845e-06, "loss": 1.072, "step": 1658 }, { "epoch": 0.5042553191489362, "grad_norm": 0.291015625, "learning_rate": 4.957446808510639e-06, "loss": 1.0084, "step": 1659 }, { "epoch": 0.5045592705167173, "grad_norm": 0.326171875, "learning_rate": 4.954407294832827e-06, "loss": 0.9701, "step": 1660 }, { "epoch": 0.5048632218844985, "grad_norm": 0.283203125, "learning_rate": 4.951367781155015e-06, "loss": 1.089, "step": 1661 }, { "epoch": 0.5051671732522797, "grad_norm": 0.29296875, "learning_rate": 4.948328267477205e-06, "loss": 1.0388, "step": 1662 }, { "epoch": 0.5054711246200608, "grad_norm": 0.28515625, "learning_rate": 4.9452887537993925e-06, "loss": 1.0064, "step": 1663 }, { "epoch": 0.505775075987842, "grad_norm": 0.27734375, "learning_rate": 4.942249240121581e-06, "loss": 1.0434, "step": 1664 }, { "epoch": 0.506079027355623, "grad_norm": 0.26953125, "learning_rate": 4.939209726443769e-06, "loss": 1.0899, "step": 1665 }, { "epoch": 0.5063829787234042, "grad_norm": 0.287109375, "learning_rate": 4.936170212765958e-06, "loss": 1.0349, "step": 1666 }, { "epoch": 0.5066869300911854, "grad_norm": 0.28515625, "learning_rate": 4.933130699088146e-06, "loss": 1.048, "step": 1667 }, { "epoch": 0.5069908814589665, "grad_norm": 0.375, "learning_rate": 4.930091185410334e-06, "loss": 1.0487, "step": 1668 }, { "epoch": 0.5072948328267477, "grad_norm": 0.28125, "learning_rate": 4.927051671732524e-06, "loss": 1.0397, "step": 1669 }, { "epoch": 0.5075987841945289, "grad_norm": 0.279296875, "learning_rate": 4.924012158054712e-06, "loss": 1.0363, "step": 1670 }, { "epoch": 0.50790273556231, "grad_norm": 0.275390625, "learning_rate": 4.9209726443769e-06, "loss": 1.0401, "step": 1671 }, { "epoch": 0.5082066869300912, "grad_norm": 0.267578125, "learning_rate": 4.917933130699088e-06, "loss": 1.0296, "step": 1672 }, { "epoch": 0.5085106382978724, "grad_norm": 0.2890625, "learning_rate": 4.914893617021277e-06, "loss": 1.1571, "step": 1673 }, { "epoch": 0.5088145896656535, "grad_norm": 0.294921875, "learning_rate": 4.9118541033434655e-06, "loss": 1.0444, "step": 1674 }, { "epoch": 0.5091185410334347, "grad_norm": 0.28125, "learning_rate": 4.908814589665654e-06, "loss": 1.1075, "step": 1675 }, { "epoch": 0.5094224924012158, "grad_norm": 0.287109375, "learning_rate": 4.905775075987843e-06, "loss": 1.0288, "step": 1676 }, { "epoch": 0.509726443768997, "grad_norm": 0.28515625, "learning_rate": 4.902735562310031e-06, "loss": 1.0903, "step": 1677 }, { "epoch": 0.5100303951367782, "grad_norm": 0.2734375, "learning_rate": 4.899696048632219e-06, "loss": 1.0271, "step": 1678 }, { "epoch": 0.5103343465045592, "grad_norm": 0.287109375, "learning_rate": 4.896656534954407e-06, "loss": 1.0992, "step": 1679 }, { "epoch": 0.5106382978723404, "grad_norm": 0.287109375, "learning_rate": 4.893617021276596e-06, "loss": 1.1159, "step": 1680 }, { "epoch": 0.5109422492401215, "grad_norm": 0.294921875, "learning_rate": 4.890577507598785e-06, "loss": 0.9833, "step": 1681 }, { "epoch": 0.5112462006079027, "grad_norm": 0.2734375, "learning_rate": 4.887537993920973e-06, "loss": 1.031, "step": 1682 }, { "epoch": 0.5115501519756839, "grad_norm": 0.28515625, "learning_rate": 4.884498480243162e-06, "loss": 1.0647, "step": 1683 }, { "epoch": 0.511854103343465, "grad_norm": 0.2734375, "learning_rate": 4.88145896656535e-06, "loss": 1.0064, "step": 1684 }, { "epoch": 0.5121580547112462, "grad_norm": 0.28515625, "learning_rate": 4.8784194528875385e-06, "loss": 1.0548, "step": 1685 }, { "epoch": 0.5124620060790274, "grad_norm": 0.28125, "learning_rate": 4.875379939209726e-06, "loss": 1.0457, "step": 1686 }, { "epoch": 0.5127659574468085, "grad_norm": 0.294921875, "learning_rate": 4.872340425531915e-06, "loss": 1.132, "step": 1687 }, { "epoch": 0.5130699088145897, "grad_norm": 0.271484375, "learning_rate": 4.869300911854104e-06, "loss": 1.1326, "step": 1688 }, { "epoch": 0.5133738601823709, "grad_norm": 0.314453125, "learning_rate": 4.866261398176292e-06, "loss": 1.0933, "step": 1689 }, { "epoch": 0.513677811550152, "grad_norm": 0.287109375, "learning_rate": 4.863221884498481e-06, "loss": 1.0983, "step": 1690 }, { "epoch": 0.5139817629179332, "grad_norm": 0.2890625, "learning_rate": 4.860182370820669e-06, "loss": 1.0225, "step": 1691 }, { "epoch": 0.5142857142857142, "grad_norm": 0.287109375, "learning_rate": 4.857142857142858e-06, "loss": 1.1257, "step": 1692 }, { "epoch": 0.5145896656534954, "grad_norm": 0.294921875, "learning_rate": 4.854103343465046e-06, "loss": 1.0108, "step": 1693 }, { "epoch": 0.5148936170212766, "grad_norm": 0.294921875, "learning_rate": 4.851063829787234e-06, "loss": 1.0466, "step": 1694 }, { "epoch": 0.5151975683890577, "grad_norm": 0.28515625, "learning_rate": 4.848024316109423e-06, "loss": 1.0794, "step": 1695 }, { "epoch": 0.5155015197568389, "grad_norm": 0.283203125, "learning_rate": 4.8449848024316115e-06, "loss": 1.1131, "step": 1696 }, { "epoch": 0.51580547112462, "grad_norm": 0.29296875, "learning_rate": 4.8419452887538e-06, "loss": 1.0188, "step": 1697 }, { "epoch": 0.5161094224924012, "grad_norm": 0.279296875, "learning_rate": 4.838905775075988e-06, "loss": 1.0674, "step": 1698 }, { "epoch": 0.5164133738601824, "grad_norm": 0.279296875, "learning_rate": 4.835866261398177e-06, "loss": 1.0484, "step": 1699 }, { "epoch": 0.5167173252279635, "grad_norm": 0.26953125, "learning_rate": 4.832826747720365e-06, "loss": 1.0381, "step": 1700 }, { "epoch": 0.5170212765957447, "grad_norm": 0.283203125, "learning_rate": 4.829787234042553e-06, "loss": 1.0132, "step": 1701 }, { "epoch": 0.5173252279635259, "grad_norm": 0.26953125, "learning_rate": 4.826747720364742e-06, "loss": 1.0291, "step": 1702 }, { "epoch": 0.517629179331307, "grad_norm": 0.279296875, "learning_rate": 4.823708206686931e-06, "loss": 1.0868, "step": 1703 }, { "epoch": 0.5179331306990882, "grad_norm": 0.28125, "learning_rate": 4.820668693009119e-06, "loss": 1.0977, "step": 1704 }, { "epoch": 0.5182370820668692, "grad_norm": 0.306640625, "learning_rate": 4.817629179331307e-06, "loss": 1.0505, "step": 1705 }, { "epoch": 0.5185410334346504, "grad_norm": 0.28125, "learning_rate": 4.814589665653496e-06, "loss": 1.0902, "step": 1706 }, { "epoch": 0.5188449848024316, "grad_norm": 0.30078125, "learning_rate": 4.8115501519756845e-06, "loss": 1.0255, "step": 1707 }, { "epoch": 0.5191489361702127, "grad_norm": 0.28125, "learning_rate": 4.808510638297872e-06, "loss": 1.0967, "step": 1708 }, { "epoch": 0.5194528875379939, "grad_norm": 0.27734375, "learning_rate": 4.805471124620061e-06, "loss": 1.0091, "step": 1709 }, { "epoch": 0.5197568389057751, "grad_norm": 0.287109375, "learning_rate": 4.80243161094225e-06, "loss": 1.1093, "step": 1710 }, { "epoch": 0.5200607902735562, "grad_norm": 0.287109375, "learning_rate": 4.799392097264438e-06, "loss": 1.0525, "step": 1711 }, { "epoch": 0.5203647416413374, "grad_norm": 0.322265625, "learning_rate": 4.796352583586626e-06, "loss": 0.99, "step": 1712 }, { "epoch": 0.5206686930091186, "grad_norm": 0.287109375, "learning_rate": 4.793313069908815e-06, "loss": 1.0186, "step": 1713 }, { "epoch": 0.5209726443768997, "grad_norm": 0.2734375, "learning_rate": 4.7902735562310036e-06, "loss": 1.0822, "step": 1714 }, { "epoch": 0.5212765957446809, "grad_norm": 0.271484375, "learning_rate": 4.787234042553192e-06, "loss": 1.07, "step": 1715 }, { "epoch": 0.521580547112462, "grad_norm": 0.28515625, "learning_rate": 4.78419452887538e-06, "loss": 0.9785, "step": 1716 }, { "epoch": 0.5218844984802432, "grad_norm": 0.2734375, "learning_rate": 4.781155015197569e-06, "loss": 1.1287, "step": 1717 }, { "epoch": 0.5221884498480243, "grad_norm": 0.28125, "learning_rate": 4.7781155015197575e-06, "loss": 1.0808, "step": 1718 }, { "epoch": 0.5224924012158054, "grad_norm": 0.296875, "learning_rate": 4.775075987841945e-06, "loss": 1.038, "step": 1719 }, { "epoch": 0.5227963525835866, "grad_norm": 0.279296875, "learning_rate": 4.772036474164134e-06, "loss": 1.0585, "step": 1720 }, { "epoch": 0.5231003039513678, "grad_norm": 0.28125, "learning_rate": 4.768996960486323e-06, "loss": 1.1211, "step": 1721 }, { "epoch": 0.5234042553191489, "grad_norm": 0.28125, "learning_rate": 4.765957446808511e-06, "loss": 1.075, "step": 1722 }, { "epoch": 0.5237082066869301, "grad_norm": 0.28125, "learning_rate": 4.762917933130699e-06, "loss": 1.0072, "step": 1723 }, { "epoch": 0.5240121580547112, "grad_norm": 0.275390625, "learning_rate": 4.759878419452888e-06, "loss": 1.0055, "step": 1724 }, { "epoch": 0.5243161094224924, "grad_norm": 0.283203125, "learning_rate": 4.7568389057750766e-06, "loss": 1.0976, "step": 1725 }, { "epoch": 0.5246200607902736, "grad_norm": 0.279296875, "learning_rate": 4.753799392097264e-06, "loss": 1.0571, "step": 1726 }, { "epoch": 0.5249240121580547, "grad_norm": 0.275390625, "learning_rate": 4.750759878419453e-06, "loss": 1.0491, "step": 1727 }, { "epoch": 0.5252279635258359, "grad_norm": 0.27734375, "learning_rate": 4.747720364741642e-06, "loss": 1.087, "step": 1728 }, { "epoch": 0.5255319148936171, "grad_norm": 0.2890625, "learning_rate": 4.7446808510638305e-06, "loss": 1.0076, "step": 1729 }, { "epoch": 0.5258358662613982, "grad_norm": 0.283203125, "learning_rate": 4.741641337386019e-06, "loss": 1.0207, "step": 1730 }, { "epoch": 0.5261398176291794, "grad_norm": 0.298828125, "learning_rate": 4.738601823708207e-06, "loss": 1.0441, "step": 1731 }, { "epoch": 0.5264437689969604, "grad_norm": 0.298828125, "learning_rate": 4.735562310030396e-06, "loss": 1.0511, "step": 1732 }, { "epoch": 0.5267477203647416, "grad_norm": 0.27734375, "learning_rate": 4.7325227963525835e-06, "loss": 1.0718, "step": 1733 }, { "epoch": 0.5270516717325228, "grad_norm": 0.26953125, "learning_rate": 4.729483282674772e-06, "loss": 1.0951, "step": 1734 }, { "epoch": 0.5273556231003039, "grad_norm": 0.279296875, "learning_rate": 4.726443768996961e-06, "loss": 1.0528, "step": 1735 }, { "epoch": 0.5276595744680851, "grad_norm": 0.26953125, "learning_rate": 4.7234042553191496e-06, "loss": 1.0751, "step": 1736 }, { "epoch": 0.5279635258358663, "grad_norm": 0.283203125, "learning_rate": 4.720364741641338e-06, "loss": 1.0648, "step": 1737 }, { "epoch": 0.5282674772036474, "grad_norm": 0.2734375, "learning_rate": 4.717325227963526e-06, "loss": 1.0396, "step": 1738 }, { "epoch": 0.5285714285714286, "grad_norm": 0.2890625, "learning_rate": 4.714285714285715e-06, "loss": 1.0195, "step": 1739 }, { "epoch": 0.5288753799392097, "grad_norm": 0.287109375, "learning_rate": 4.711246200607903e-06, "loss": 1.024, "step": 1740 }, { "epoch": 0.5291793313069909, "grad_norm": 0.291015625, "learning_rate": 4.708206686930091e-06, "loss": 1.0489, "step": 1741 }, { "epoch": 0.5294832826747721, "grad_norm": 0.275390625, "learning_rate": 4.70516717325228e-06, "loss": 1.0948, "step": 1742 }, { "epoch": 0.5297872340425532, "grad_norm": 0.29296875, "learning_rate": 4.702127659574469e-06, "loss": 1.0468, "step": 1743 }, { "epoch": 0.5300911854103344, "grad_norm": 0.275390625, "learning_rate": 4.699088145896657e-06, "loss": 1.0486, "step": 1744 }, { "epoch": 0.5303951367781155, "grad_norm": 0.287109375, "learning_rate": 4.696048632218845e-06, "loss": 0.9789, "step": 1745 }, { "epoch": 0.5306990881458966, "grad_norm": 0.294921875, "learning_rate": 4.693009118541034e-06, "loss": 1.1035, "step": 1746 }, { "epoch": 0.5310030395136778, "grad_norm": 0.466796875, "learning_rate": 4.689969604863222e-06, "loss": 1.0337, "step": 1747 }, { "epoch": 0.5313069908814589, "grad_norm": 0.2734375, "learning_rate": 4.686930091185411e-06, "loss": 1.0359, "step": 1748 }, { "epoch": 0.5316109422492401, "grad_norm": 0.279296875, "learning_rate": 4.683890577507599e-06, "loss": 1.0358, "step": 1749 }, { "epoch": 0.5319148936170213, "grad_norm": 0.287109375, "learning_rate": 4.680851063829788e-06, "loss": 1.0667, "step": 1750 }, { "epoch": 0.5322188449848024, "grad_norm": 0.283203125, "learning_rate": 4.6778115501519764e-06, "loss": 1.0445, "step": 1751 }, { "epoch": 0.5325227963525836, "grad_norm": 0.271484375, "learning_rate": 4.674772036474164e-06, "loss": 1.0409, "step": 1752 }, { "epoch": 0.5328267477203648, "grad_norm": 0.30859375, "learning_rate": 4.671732522796353e-06, "loss": 1.0569, "step": 1753 }, { "epoch": 0.5331306990881459, "grad_norm": 0.29296875, "learning_rate": 4.668693009118541e-06, "loss": 1.0117, "step": 1754 }, { "epoch": 0.5334346504559271, "grad_norm": 0.298828125, "learning_rate": 4.66565349544073e-06, "loss": 1.0619, "step": 1755 }, { "epoch": 0.5337386018237082, "grad_norm": 0.275390625, "learning_rate": 4.662613981762918e-06, "loss": 1.0709, "step": 1756 }, { "epoch": 0.5340425531914894, "grad_norm": 0.302734375, "learning_rate": 4.659574468085107e-06, "loss": 1.0419, "step": 1757 }, { "epoch": 0.5343465045592705, "grad_norm": 0.29296875, "learning_rate": 4.6565349544072955e-06, "loss": 0.9905, "step": 1758 }, { "epoch": 0.5346504559270516, "grad_norm": 0.29296875, "learning_rate": 4.653495440729483e-06, "loss": 0.9734, "step": 1759 }, { "epoch": 0.5349544072948328, "grad_norm": 0.287109375, "learning_rate": 4.650455927051672e-06, "loss": 1.016, "step": 1760 }, { "epoch": 0.535258358662614, "grad_norm": 0.27734375, "learning_rate": 4.64741641337386e-06, "loss": 1.0831, "step": 1761 }, { "epoch": 0.5355623100303951, "grad_norm": 0.2734375, "learning_rate": 4.6443768996960494e-06, "loss": 1.0053, "step": 1762 }, { "epoch": 0.5358662613981763, "grad_norm": 0.27734375, "learning_rate": 4.641337386018237e-06, "loss": 1.0238, "step": 1763 }, { "epoch": 0.5361702127659574, "grad_norm": 0.27734375, "learning_rate": 4.638297872340426e-06, "loss": 1.079, "step": 1764 }, { "epoch": 0.5364741641337386, "grad_norm": 0.275390625, "learning_rate": 4.635258358662615e-06, "loss": 1.1094, "step": 1765 }, { "epoch": 0.5367781155015198, "grad_norm": 0.29296875, "learning_rate": 4.6322188449848025e-06, "loss": 1.0741, "step": 1766 }, { "epoch": 0.5370820668693009, "grad_norm": 0.33984375, "learning_rate": 4.629179331306991e-06, "loss": 1.103, "step": 1767 }, { "epoch": 0.5373860182370821, "grad_norm": 0.2734375, "learning_rate": 4.62613981762918e-06, "loss": 1.0885, "step": 1768 }, { "epoch": 0.5376899696048633, "grad_norm": 0.2890625, "learning_rate": 4.6231003039513685e-06, "loss": 1.0377, "step": 1769 }, { "epoch": 0.5379939209726444, "grad_norm": 0.279296875, "learning_rate": 4.620060790273556e-06, "loss": 1.1023, "step": 1770 }, { "epoch": 0.5382978723404256, "grad_norm": 0.283203125, "learning_rate": 4.617021276595745e-06, "loss": 1.0926, "step": 1771 }, { "epoch": 0.5386018237082066, "grad_norm": 0.28515625, "learning_rate": 4.613981762917934e-06, "loss": 1.084, "step": 1772 }, { "epoch": 0.5389057750759878, "grad_norm": 0.275390625, "learning_rate": 4.610942249240122e-06, "loss": 1.0932, "step": 1773 }, { "epoch": 0.539209726443769, "grad_norm": 0.30078125, "learning_rate": 4.60790273556231e-06, "loss": 1.0178, "step": 1774 }, { "epoch": 0.5395136778115501, "grad_norm": 0.28125, "learning_rate": 4.604863221884499e-06, "loss": 1.0112, "step": 1775 }, { "epoch": 0.5398176291793313, "grad_norm": 0.349609375, "learning_rate": 4.601823708206688e-06, "loss": 0.9964, "step": 1776 }, { "epoch": 0.5401215805471125, "grad_norm": 0.2734375, "learning_rate": 4.5987841945288755e-06, "loss": 1.1135, "step": 1777 }, { "epoch": 0.5404255319148936, "grad_norm": 0.26171875, "learning_rate": 4.595744680851064e-06, "loss": 1.0739, "step": 1778 }, { "epoch": 0.5407294832826748, "grad_norm": 0.279296875, "learning_rate": 4.592705167173253e-06, "loss": 1.049, "step": 1779 }, { "epoch": 0.541033434650456, "grad_norm": 0.28515625, "learning_rate": 4.589665653495441e-06, "loss": 1.0659, "step": 1780 }, { "epoch": 0.5413373860182371, "grad_norm": 0.287109375, "learning_rate": 4.58662613981763e-06, "loss": 1.0262, "step": 1781 }, { "epoch": 0.5416413373860183, "grad_norm": 0.291015625, "learning_rate": 4.583586626139818e-06, "loss": 1.0189, "step": 1782 }, { "epoch": 0.5419452887537994, "grad_norm": 0.2734375, "learning_rate": 4.580547112462007e-06, "loss": 1.0224, "step": 1783 }, { "epoch": 0.5422492401215806, "grad_norm": 0.28515625, "learning_rate": 4.577507598784195e-06, "loss": 1.0743, "step": 1784 }, { "epoch": 0.5425531914893617, "grad_norm": 0.28515625, "learning_rate": 4.574468085106383e-06, "loss": 1.0227, "step": 1785 }, { "epoch": 0.5428571428571428, "grad_norm": 0.296875, "learning_rate": 4.571428571428572e-06, "loss": 1.0903, "step": 1786 }, { "epoch": 0.543161094224924, "grad_norm": 0.279296875, "learning_rate": 4.56838905775076e-06, "loss": 1.0312, "step": 1787 }, { "epoch": 0.5434650455927051, "grad_norm": 0.32421875, "learning_rate": 4.565349544072949e-06, "loss": 1.0746, "step": 1788 }, { "epoch": 0.5437689969604863, "grad_norm": 0.28515625, "learning_rate": 4.562310030395137e-06, "loss": 1.0614, "step": 1789 }, { "epoch": 0.5440729483282675, "grad_norm": 0.28515625, "learning_rate": 4.559270516717326e-06, "loss": 1.0015, "step": 1790 }, { "epoch": 0.5443768996960486, "grad_norm": 0.291015625, "learning_rate": 4.556231003039514e-06, "loss": 1.0147, "step": 1791 }, { "epoch": 0.5446808510638298, "grad_norm": 0.373046875, "learning_rate": 4.553191489361702e-06, "loss": 1.0485, "step": 1792 }, { "epoch": 0.544984802431611, "grad_norm": 0.3359375, "learning_rate": 4.550151975683891e-06, "loss": 1.0786, "step": 1793 }, { "epoch": 0.5452887537993921, "grad_norm": 0.279296875, "learning_rate": 4.547112462006079e-06, "loss": 0.9846, "step": 1794 }, { "epoch": 0.5455927051671733, "grad_norm": 0.283203125, "learning_rate": 4.544072948328268e-06, "loss": 1.0492, "step": 1795 }, { "epoch": 0.5458966565349544, "grad_norm": 0.267578125, "learning_rate": 4.541033434650456e-06, "loss": 1.1091, "step": 1796 }, { "epoch": 0.5462006079027356, "grad_norm": 0.31640625, "learning_rate": 4.537993920972645e-06, "loss": 1.0217, "step": 1797 }, { "epoch": 0.5465045592705167, "grad_norm": 0.27734375, "learning_rate": 4.534954407294833e-06, "loss": 1.1259, "step": 1798 }, { "epoch": 0.5468085106382978, "grad_norm": 0.298828125, "learning_rate": 4.5319148936170215e-06, "loss": 1.1508, "step": 1799 }, { "epoch": 0.547112462006079, "grad_norm": 0.283203125, "learning_rate": 4.52887537993921e-06, "loss": 1.1161, "step": 1800 }, { "epoch": 0.5474164133738602, "grad_norm": 0.283203125, "learning_rate": 4.525835866261399e-06, "loss": 1.0113, "step": 1801 }, { "epoch": 0.5477203647416413, "grad_norm": 0.279296875, "learning_rate": 4.5227963525835875e-06, "loss": 1.0588, "step": 1802 }, { "epoch": 0.5480243161094225, "grad_norm": 0.28125, "learning_rate": 4.519756838905775e-06, "loss": 1.0892, "step": 1803 }, { "epoch": 0.5483282674772036, "grad_norm": 0.287109375, "learning_rate": 4.516717325227964e-06, "loss": 1.0442, "step": 1804 }, { "epoch": 0.5486322188449848, "grad_norm": 0.28515625, "learning_rate": 4.513677811550152e-06, "loss": 1.0686, "step": 1805 }, { "epoch": 0.548936170212766, "grad_norm": 0.283203125, "learning_rate": 4.5106382978723406e-06, "loss": 1.0953, "step": 1806 }, { "epoch": 0.5492401215805471, "grad_norm": 0.27734375, "learning_rate": 4.507598784194529e-06, "loss": 1.0509, "step": 1807 }, { "epoch": 0.5495440729483283, "grad_norm": 0.29296875, "learning_rate": 4.504559270516718e-06, "loss": 1.1422, "step": 1808 }, { "epoch": 0.5498480243161095, "grad_norm": 0.279296875, "learning_rate": 4.501519756838907e-06, "loss": 1.0342, "step": 1809 }, { "epoch": 0.5501519756838906, "grad_norm": 0.38671875, "learning_rate": 4.4984802431610945e-06, "loss": 1.0531, "step": 1810 }, { "epoch": 0.5504559270516717, "grad_norm": 0.28125, "learning_rate": 4.495440729483283e-06, "loss": 1.038, "step": 1811 }, { "epoch": 0.5507598784194528, "grad_norm": 0.28125, "learning_rate": 4.492401215805471e-06, "loss": 1.0902, "step": 1812 }, { "epoch": 0.551063829787234, "grad_norm": 0.28515625, "learning_rate": 4.48936170212766e-06, "loss": 1.0893, "step": 1813 }, { "epoch": 0.5513677811550152, "grad_norm": 0.3046875, "learning_rate": 4.486322188449848e-06, "loss": 1.1149, "step": 1814 }, { "epoch": 0.5516717325227963, "grad_norm": 0.3125, "learning_rate": 4.483282674772037e-06, "loss": 1.0568, "step": 1815 }, { "epoch": 0.5519756838905775, "grad_norm": 0.28125, "learning_rate": 4.480243161094226e-06, "loss": 1.0409, "step": 1816 }, { "epoch": 0.5522796352583587, "grad_norm": 0.28125, "learning_rate": 4.4772036474164136e-06, "loss": 1.0133, "step": 1817 }, { "epoch": 0.5525835866261398, "grad_norm": 0.28515625, "learning_rate": 4.474164133738602e-06, "loss": 1.0594, "step": 1818 }, { "epoch": 0.552887537993921, "grad_norm": 0.275390625, "learning_rate": 4.471124620060791e-06, "loss": 1.1537, "step": 1819 }, { "epoch": 0.5531914893617021, "grad_norm": 0.267578125, "learning_rate": 4.468085106382979e-06, "loss": 1.0734, "step": 1820 }, { "epoch": 0.5534954407294833, "grad_norm": 0.296875, "learning_rate": 4.4650455927051674e-06, "loss": 1.0428, "step": 1821 }, { "epoch": 0.5537993920972645, "grad_norm": 0.27734375, "learning_rate": 4.462006079027356e-06, "loss": 0.9828, "step": 1822 }, { "epoch": 0.5541033434650456, "grad_norm": 0.279296875, "learning_rate": 4.458966565349545e-06, "loss": 1.0961, "step": 1823 }, { "epoch": 0.5544072948328268, "grad_norm": 0.279296875, "learning_rate": 4.455927051671733e-06, "loss": 1.0874, "step": 1824 }, { "epoch": 0.5547112462006079, "grad_norm": 0.283203125, "learning_rate": 4.452887537993921e-06, "loss": 1.0266, "step": 1825 }, { "epoch": 0.555015197568389, "grad_norm": 0.279296875, "learning_rate": 4.44984802431611e-06, "loss": 1.0661, "step": 1826 }, { "epoch": 0.5553191489361702, "grad_norm": 0.271484375, "learning_rate": 4.446808510638298e-06, "loss": 1.0737, "step": 1827 }, { "epoch": 0.5556231003039513, "grad_norm": 0.271484375, "learning_rate": 4.4437689969604865e-06, "loss": 1.0452, "step": 1828 }, { "epoch": 0.5559270516717325, "grad_norm": 0.275390625, "learning_rate": 4.440729483282675e-06, "loss": 1.0685, "step": 1829 }, { "epoch": 0.5562310030395137, "grad_norm": 0.48046875, "learning_rate": 4.437689969604864e-06, "loss": 1.0723, "step": 1830 }, { "epoch": 0.5565349544072948, "grad_norm": 0.26953125, "learning_rate": 4.434650455927052e-06, "loss": 1.1614, "step": 1831 }, { "epoch": 0.556838905775076, "grad_norm": 0.279296875, "learning_rate": 4.4316109422492404e-06, "loss": 0.9881, "step": 1832 }, { "epoch": 0.5571428571428572, "grad_norm": 0.27734375, "learning_rate": 4.428571428571429e-06, "loss": 0.9831, "step": 1833 }, { "epoch": 0.5574468085106383, "grad_norm": 0.298828125, "learning_rate": 4.425531914893617e-06, "loss": 0.9781, "step": 1834 }, { "epoch": 0.5577507598784195, "grad_norm": 0.283203125, "learning_rate": 4.422492401215806e-06, "loss": 1.0462, "step": 1835 }, { "epoch": 0.5580547112462007, "grad_norm": 0.275390625, "learning_rate": 4.419452887537994e-06, "loss": 1.0629, "step": 1836 }, { "epoch": 0.5583586626139818, "grad_norm": 0.322265625, "learning_rate": 4.416413373860183e-06, "loss": 1.0986, "step": 1837 }, { "epoch": 0.5586626139817629, "grad_norm": 0.279296875, "learning_rate": 4.413373860182371e-06, "loss": 1.0883, "step": 1838 }, { "epoch": 0.558966565349544, "grad_norm": 0.271484375, "learning_rate": 4.4103343465045595e-06, "loss": 1.0921, "step": 1839 }, { "epoch": 0.5592705167173252, "grad_norm": 0.283203125, "learning_rate": 4.407294832826748e-06, "loss": 1.1066, "step": 1840 }, { "epoch": 0.5595744680851064, "grad_norm": 0.2734375, "learning_rate": 4.404255319148937e-06, "loss": 1.0679, "step": 1841 }, { "epoch": 0.5598784194528875, "grad_norm": 0.275390625, "learning_rate": 4.401215805471125e-06, "loss": 1.1165, "step": 1842 }, { "epoch": 0.5601823708206687, "grad_norm": 0.28515625, "learning_rate": 4.3981762917933134e-06, "loss": 0.9928, "step": 1843 }, { "epoch": 0.5604863221884498, "grad_norm": 0.2734375, "learning_rate": 4.395136778115502e-06, "loss": 1.0653, "step": 1844 }, { "epoch": 0.560790273556231, "grad_norm": 0.28515625, "learning_rate": 4.39209726443769e-06, "loss": 1.0692, "step": 1845 }, { "epoch": 0.5610942249240122, "grad_norm": 0.279296875, "learning_rate": 4.389057750759879e-06, "loss": 1.0898, "step": 1846 }, { "epoch": 0.5613981762917933, "grad_norm": 0.279296875, "learning_rate": 4.386018237082067e-06, "loss": 1.0453, "step": 1847 }, { "epoch": 0.5617021276595745, "grad_norm": 0.302734375, "learning_rate": 4.382978723404256e-06, "loss": 1.0179, "step": 1848 }, { "epoch": 0.5620060790273557, "grad_norm": 0.287109375, "learning_rate": 4.379939209726444e-06, "loss": 0.9817, "step": 1849 }, { "epoch": 0.5623100303951368, "grad_norm": 0.28125, "learning_rate": 4.3768996960486325e-06, "loss": 1.0707, "step": 1850 }, { "epoch": 0.5626139817629179, "grad_norm": 0.283203125, "learning_rate": 4.373860182370821e-06, "loss": 1.0623, "step": 1851 }, { "epoch": 0.562917933130699, "grad_norm": 0.265625, "learning_rate": 4.370820668693009e-06, "loss": 1.0632, "step": 1852 }, { "epoch": 0.5632218844984802, "grad_norm": 0.2890625, "learning_rate": 4.367781155015198e-06, "loss": 0.9972, "step": 1853 }, { "epoch": 0.5635258358662614, "grad_norm": 0.294921875, "learning_rate": 4.3647416413373864e-06, "loss": 0.994, "step": 1854 }, { "epoch": 0.5638297872340425, "grad_norm": 0.296875, "learning_rate": 4.361702127659575e-06, "loss": 1.0918, "step": 1855 }, { "epoch": 0.5641337386018237, "grad_norm": 0.28125, "learning_rate": 4.358662613981764e-06, "loss": 1.0998, "step": 1856 }, { "epoch": 0.5644376899696049, "grad_norm": 0.283203125, "learning_rate": 4.355623100303952e-06, "loss": 1.094, "step": 1857 }, { "epoch": 0.564741641337386, "grad_norm": 0.28125, "learning_rate": 4.35258358662614e-06, "loss": 1.0425, "step": 1858 }, { "epoch": 0.5650455927051672, "grad_norm": 0.265625, "learning_rate": 4.349544072948328e-06, "loss": 1.0612, "step": 1859 }, { "epoch": 0.5653495440729484, "grad_norm": 0.2734375, "learning_rate": 4.346504559270517e-06, "loss": 1.0573, "step": 1860 }, { "epoch": 0.5656534954407295, "grad_norm": 0.283203125, "learning_rate": 4.3434650455927055e-06, "loss": 0.978, "step": 1861 }, { "epoch": 0.5659574468085107, "grad_norm": 0.291015625, "learning_rate": 4.340425531914894e-06, "loss": 1.0607, "step": 1862 }, { "epoch": 0.5662613981762918, "grad_norm": 0.28515625, "learning_rate": 4.337386018237083e-06, "loss": 1.0684, "step": 1863 }, { "epoch": 0.5665653495440729, "grad_norm": 0.287109375, "learning_rate": 4.334346504559271e-06, "loss": 1.0737, "step": 1864 }, { "epoch": 0.5668693009118541, "grad_norm": 0.28125, "learning_rate": 4.331306990881459e-06, "loss": 1.0452, "step": 1865 }, { "epoch": 0.5671732522796352, "grad_norm": 0.2734375, "learning_rate": 4.328267477203647e-06, "loss": 1.1132, "step": 1866 }, { "epoch": 0.5674772036474164, "grad_norm": 0.28125, "learning_rate": 4.325227963525836e-06, "loss": 0.9893, "step": 1867 }, { "epoch": 0.5677811550151975, "grad_norm": 0.27734375, "learning_rate": 4.322188449848025e-06, "loss": 1.0643, "step": 1868 }, { "epoch": 0.5680851063829787, "grad_norm": 0.298828125, "learning_rate": 4.319148936170213e-06, "loss": 0.9954, "step": 1869 }, { "epoch": 0.5683890577507599, "grad_norm": 0.27734375, "learning_rate": 4.316109422492402e-06, "loss": 1.0094, "step": 1870 }, { "epoch": 0.568693009118541, "grad_norm": 0.27734375, "learning_rate": 4.31306990881459e-06, "loss": 1.0954, "step": 1871 }, { "epoch": 0.5689969604863222, "grad_norm": 0.2890625, "learning_rate": 4.3100303951367785e-06, "loss": 1.0211, "step": 1872 }, { "epoch": 0.5693009118541034, "grad_norm": 0.29296875, "learning_rate": 4.306990881458966e-06, "loss": 1.1193, "step": 1873 }, { "epoch": 0.5696048632218845, "grad_norm": 0.283203125, "learning_rate": 4.303951367781156e-06, "loss": 0.9544, "step": 1874 }, { "epoch": 0.5699088145896657, "grad_norm": 0.28515625, "learning_rate": 4.300911854103344e-06, "loss": 1.092, "step": 1875 }, { "epoch": 0.5702127659574469, "grad_norm": 0.283203125, "learning_rate": 4.297872340425532e-06, "loss": 1.0657, "step": 1876 }, { "epoch": 0.570516717325228, "grad_norm": 0.2734375, "learning_rate": 4.294832826747721e-06, "loss": 1.0171, "step": 1877 }, { "epoch": 0.5708206686930091, "grad_norm": 0.275390625, "learning_rate": 4.291793313069909e-06, "loss": 1.0682, "step": 1878 }, { "epoch": 0.5711246200607902, "grad_norm": 0.28125, "learning_rate": 4.288753799392098e-06, "loss": 1.0147, "step": 1879 }, { "epoch": 0.5714285714285714, "grad_norm": 0.302734375, "learning_rate": 4.2857142857142855e-06, "loss": 1.1167, "step": 1880 }, { "epoch": 0.5717325227963526, "grad_norm": 0.2890625, "learning_rate": 4.282674772036475e-06, "loss": 1.056, "step": 1881 }, { "epoch": 0.5720364741641337, "grad_norm": 0.29296875, "learning_rate": 4.279635258358663e-06, "loss": 1.0192, "step": 1882 }, { "epoch": 0.5723404255319149, "grad_norm": 0.298828125, "learning_rate": 4.2765957446808515e-06, "loss": 1.0415, "step": 1883 }, { "epoch": 0.572644376899696, "grad_norm": 0.306640625, "learning_rate": 4.27355623100304e-06, "loss": 1.0014, "step": 1884 }, { "epoch": 0.5729483282674772, "grad_norm": 0.291015625, "learning_rate": 4.270516717325228e-06, "loss": 1.0442, "step": 1885 }, { "epoch": 0.5732522796352584, "grad_norm": 0.30078125, "learning_rate": 4.267477203647417e-06, "loss": 1.0345, "step": 1886 }, { "epoch": 0.5735562310030395, "grad_norm": 0.27734375, "learning_rate": 4.2644376899696046e-06, "loss": 1.045, "step": 1887 }, { "epoch": 0.5738601823708207, "grad_norm": 0.27734375, "learning_rate": 4.261398176291794e-06, "loss": 1.0342, "step": 1888 }, { "epoch": 0.5741641337386019, "grad_norm": 0.294921875, "learning_rate": 4.258358662613982e-06, "loss": 1.0338, "step": 1889 }, { "epoch": 0.574468085106383, "grad_norm": 0.29296875, "learning_rate": 4.255319148936171e-06, "loss": 1.0903, "step": 1890 }, { "epoch": 0.5747720364741641, "grad_norm": 0.279296875, "learning_rate": 4.252279635258359e-06, "loss": 0.9938, "step": 1891 }, { "epoch": 0.5750759878419452, "grad_norm": 0.31640625, "learning_rate": 4.249240121580547e-06, "loss": 1.0728, "step": 1892 }, { "epoch": 0.5753799392097264, "grad_norm": 0.27734375, "learning_rate": 4.246200607902736e-06, "loss": 1.0649, "step": 1893 }, { "epoch": 0.5756838905775076, "grad_norm": 0.2734375, "learning_rate": 4.2431610942249245e-06, "loss": 1.12, "step": 1894 }, { "epoch": 0.5759878419452887, "grad_norm": 0.2890625, "learning_rate": 4.240121580547113e-06, "loss": 1.0354, "step": 1895 }, { "epoch": 0.5762917933130699, "grad_norm": 0.27734375, "learning_rate": 4.237082066869301e-06, "loss": 1.07, "step": 1896 }, { "epoch": 0.5765957446808511, "grad_norm": 0.279296875, "learning_rate": 4.23404255319149e-06, "loss": 1.0404, "step": 1897 }, { "epoch": 0.5768996960486322, "grad_norm": 0.279296875, "learning_rate": 4.231003039513678e-06, "loss": 1.122, "step": 1898 }, { "epoch": 0.5772036474164134, "grad_norm": 0.279296875, "learning_rate": 4.227963525835866e-06, "loss": 1.0058, "step": 1899 }, { "epoch": 0.5775075987841946, "grad_norm": 0.279296875, "learning_rate": 4.224924012158055e-06, "loss": 1.1424, "step": 1900 }, { "epoch": 0.5778115501519757, "grad_norm": 0.3125, "learning_rate": 4.221884498480244e-06, "loss": 1.0488, "step": 1901 }, { "epoch": 0.5781155015197569, "grad_norm": 0.283203125, "learning_rate": 4.218844984802432e-06, "loss": 1.0675, "step": 1902 }, { "epoch": 0.578419452887538, "grad_norm": 0.291015625, "learning_rate": 4.21580547112462e-06, "loss": 1.0461, "step": 1903 }, { "epoch": 0.5787234042553191, "grad_norm": 0.357421875, "learning_rate": 4.212765957446809e-06, "loss": 1.03, "step": 1904 }, { "epoch": 0.5790273556231003, "grad_norm": 0.279296875, "learning_rate": 4.2097264437689975e-06, "loss": 1.0684, "step": 1905 }, { "epoch": 0.5793313069908814, "grad_norm": 0.296875, "learning_rate": 4.206686930091185e-06, "loss": 1.0806, "step": 1906 }, { "epoch": 0.5796352583586626, "grad_norm": 0.283203125, "learning_rate": 4.203647416413374e-06, "loss": 1.0263, "step": 1907 }, { "epoch": 0.5799392097264437, "grad_norm": 0.291015625, "learning_rate": 4.200607902735563e-06, "loss": 0.9585, "step": 1908 }, { "epoch": 0.5802431610942249, "grad_norm": 0.27734375, "learning_rate": 4.197568389057751e-06, "loss": 1.0953, "step": 1909 }, { "epoch": 0.5805471124620061, "grad_norm": 0.298828125, "learning_rate": 4.194528875379939e-06, "loss": 1.0857, "step": 1910 }, { "epoch": 0.5808510638297872, "grad_norm": 0.296875, "learning_rate": 4.191489361702128e-06, "loss": 1.0945, "step": 1911 }, { "epoch": 0.5811550151975684, "grad_norm": 0.28125, "learning_rate": 4.188449848024317e-06, "loss": 1.0341, "step": 1912 }, { "epoch": 0.5814589665653496, "grad_norm": 0.35546875, "learning_rate": 4.1854103343465044e-06, "loss": 1.0107, "step": 1913 }, { "epoch": 0.5817629179331307, "grad_norm": 0.28125, "learning_rate": 4.182370820668694e-06, "loss": 1.0345, "step": 1914 }, { "epoch": 0.5820668693009119, "grad_norm": 0.28125, "learning_rate": 4.179331306990882e-06, "loss": 1.0988, "step": 1915 }, { "epoch": 0.5823708206686931, "grad_norm": 0.275390625, "learning_rate": 4.1762917933130705e-06, "loss": 1.0648, "step": 1916 }, { "epoch": 0.5826747720364741, "grad_norm": 0.28125, "learning_rate": 4.173252279635258e-06, "loss": 1.0756, "step": 1917 }, { "epoch": 0.5829787234042553, "grad_norm": 0.2890625, "learning_rate": 4.170212765957447e-06, "loss": 1.1036, "step": 1918 }, { "epoch": 0.5832826747720364, "grad_norm": 0.2734375, "learning_rate": 4.167173252279636e-06, "loss": 1.1345, "step": 1919 }, { "epoch": 0.5835866261398176, "grad_norm": 0.451171875, "learning_rate": 4.1641337386018235e-06, "loss": 0.9759, "step": 1920 }, { "epoch": 0.5838905775075988, "grad_norm": 0.2890625, "learning_rate": 4.161094224924013e-06, "loss": 1.0322, "step": 1921 }, { "epoch": 0.5841945288753799, "grad_norm": 0.302734375, "learning_rate": 4.158054711246201e-06, "loss": 1.0387, "step": 1922 }, { "epoch": 0.5844984802431611, "grad_norm": 0.275390625, "learning_rate": 4.15501519756839e-06, "loss": 1.075, "step": 1923 }, { "epoch": 0.5848024316109423, "grad_norm": 0.27734375, "learning_rate": 4.1519756838905774e-06, "loss": 1.1, "step": 1924 }, { "epoch": 0.5851063829787234, "grad_norm": 0.2734375, "learning_rate": 4.148936170212766e-06, "loss": 0.9945, "step": 1925 }, { "epoch": 0.5854103343465046, "grad_norm": 0.28515625, "learning_rate": 4.145896656534955e-06, "loss": 1.0596, "step": 1926 }, { "epoch": 0.5857142857142857, "grad_norm": 0.28515625, "learning_rate": 4.1428571428571435e-06, "loss": 1.0837, "step": 1927 }, { "epoch": 0.5860182370820669, "grad_norm": 0.27734375, "learning_rate": 4.139817629179332e-06, "loss": 1.0206, "step": 1928 }, { "epoch": 0.5863221884498481, "grad_norm": 0.287109375, "learning_rate": 4.13677811550152e-06, "loss": 1.0052, "step": 1929 }, { "epoch": 0.5866261398176292, "grad_norm": 0.3046875, "learning_rate": 4.133738601823709e-06, "loss": 1.03, "step": 1930 }, { "epoch": 0.5869300911854103, "grad_norm": 0.283203125, "learning_rate": 4.1306990881458965e-06, "loss": 1.0828, "step": 1931 }, { "epoch": 0.5872340425531914, "grad_norm": 0.291015625, "learning_rate": 4.127659574468085e-06, "loss": 1.0397, "step": 1932 }, { "epoch": 0.5875379939209726, "grad_norm": 0.291015625, "learning_rate": 4.124620060790274e-06, "loss": 1.0293, "step": 1933 }, { "epoch": 0.5878419452887538, "grad_norm": 0.279296875, "learning_rate": 4.121580547112463e-06, "loss": 1.0688, "step": 1934 }, { "epoch": 0.5881458966565349, "grad_norm": 0.29296875, "learning_rate": 4.118541033434651e-06, "loss": 1.1206, "step": 1935 }, { "epoch": 0.5884498480243161, "grad_norm": 0.3125, "learning_rate": 4.115501519756839e-06, "loss": 1.0545, "step": 1936 }, { "epoch": 0.5887537993920973, "grad_norm": 0.279296875, "learning_rate": 4.112462006079028e-06, "loss": 1.1074, "step": 1937 }, { "epoch": 0.5890577507598784, "grad_norm": 0.275390625, "learning_rate": 4.109422492401216e-06, "loss": 1.055, "step": 1938 }, { "epoch": 0.5893617021276596, "grad_norm": 0.306640625, "learning_rate": 4.106382978723404e-06, "loss": 1.135, "step": 1939 }, { "epoch": 0.5896656534954408, "grad_norm": 0.279296875, "learning_rate": 4.103343465045593e-06, "loss": 1.0193, "step": 1940 }, { "epoch": 0.5899696048632219, "grad_norm": 0.3046875, "learning_rate": 4.100303951367782e-06, "loss": 0.9693, "step": 1941 }, { "epoch": 0.5902735562310031, "grad_norm": 0.30078125, "learning_rate": 4.09726443768997e-06, "loss": 1.0128, "step": 1942 }, { "epoch": 0.5905775075987842, "grad_norm": 0.296875, "learning_rate": 4.094224924012158e-06, "loss": 1.071, "step": 1943 }, { "epoch": 0.5908814589665653, "grad_norm": 0.2734375, "learning_rate": 4.091185410334347e-06, "loss": 1.0381, "step": 1944 }, { "epoch": 0.5911854103343465, "grad_norm": 0.275390625, "learning_rate": 4.088145896656536e-06, "loss": 1.1016, "step": 1945 }, { "epoch": 0.5914893617021276, "grad_norm": 0.291015625, "learning_rate": 4.085106382978723e-06, "loss": 1.0438, "step": 1946 }, { "epoch": 0.5917933130699088, "grad_norm": 0.380859375, "learning_rate": 4.082066869300912e-06, "loss": 1.0118, "step": 1947 }, { "epoch": 0.59209726443769, "grad_norm": 0.287109375, "learning_rate": 4.079027355623101e-06, "loss": 1.1267, "step": 1948 }, { "epoch": 0.5924012158054711, "grad_norm": 0.283203125, "learning_rate": 4.0759878419452895e-06, "loss": 1.0803, "step": 1949 }, { "epoch": 0.5927051671732523, "grad_norm": 0.283203125, "learning_rate": 4.072948328267477e-06, "loss": 1.0978, "step": 1950 }, { "epoch": 0.5930091185410334, "grad_norm": 0.287109375, "learning_rate": 4.069908814589666e-06, "loss": 0.9734, "step": 1951 }, { "epoch": 0.5933130699088146, "grad_norm": 0.27734375, "learning_rate": 4.066869300911855e-06, "loss": 1.0336, "step": 1952 }, { "epoch": 0.5936170212765958, "grad_norm": 0.283203125, "learning_rate": 4.0638297872340425e-06, "loss": 1.0372, "step": 1953 }, { "epoch": 0.5939209726443769, "grad_norm": 0.275390625, "learning_rate": 4.060790273556231e-06, "loss": 1.1091, "step": 1954 }, { "epoch": 0.5942249240121581, "grad_norm": 0.294921875, "learning_rate": 4.05775075987842e-06, "loss": 1.0809, "step": 1955 }, { "epoch": 0.5945288753799393, "grad_norm": 0.31640625, "learning_rate": 4.0547112462006086e-06, "loss": 1.0209, "step": 1956 }, { "epoch": 0.5948328267477203, "grad_norm": 0.279296875, "learning_rate": 4.051671732522796e-06, "loss": 1.1016, "step": 1957 }, { "epoch": 0.5951367781155015, "grad_norm": 0.279296875, "learning_rate": 4.048632218844985e-06, "loss": 1.1114, "step": 1958 }, { "epoch": 0.5954407294832826, "grad_norm": 0.283203125, "learning_rate": 4.045592705167174e-06, "loss": 1.0558, "step": 1959 }, { "epoch": 0.5957446808510638, "grad_norm": 0.306640625, "learning_rate": 4.042553191489362e-06, "loss": 1.0598, "step": 1960 }, { "epoch": 0.596048632218845, "grad_norm": 0.291015625, "learning_rate": 4.03951367781155e-06, "loss": 1.0733, "step": 1961 }, { "epoch": 0.5963525835866261, "grad_norm": 0.27734375, "learning_rate": 4.036474164133739e-06, "loss": 1.0287, "step": 1962 }, { "epoch": 0.5966565349544073, "grad_norm": 0.314453125, "learning_rate": 4.033434650455928e-06, "loss": 1.0635, "step": 1963 }, { "epoch": 0.5969604863221885, "grad_norm": 0.291015625, "learning_rate": 4.0303951367781155e-06, "loss": 1.0587, "step": 1964 }, { "epoch": 0.5972644376899696, "grad_norm": 0.28125, "learning_rate": 4.027355623100304e-06, "loss": 1.0384, "step": 1965 }, { "epoch": 0.5975683890577508, "grad_norm": 0.30078125, "learning_rate": 4.024316109422493e-06, "loss": 1.0405, "step": 1966 }, { "epoch": 0.597872340425532, "grad_norm": 0.2890625, "learning_rate": 4.0212765957446816e-06, "loss": 1.1449, "step": 1967 }, { "epoch": 0.5981762917933131, "grad_norm": 0.28125, "learning_rate": 4.018237082066869e-06, "loss": 1.0833, "step": 1968 }, { "epoch": 0.5984802431610943, "grad_norm": 0.2734375, "learning_rate": 4.015197568389058e-06, "loss": 1.0634, "step": 1969 }, { "epoch": 0.5987841945288754, "grad_norm": 0.291015625, "learning_rate": 4.012158054711247e-06, "loss": 1.0671, "step": 1970 }, { "epoch": 0.5990881458966565, "grad_norm": 0.28125, "learning_rate": 4.009118541033435e-06, "loss": 1.1162, "step": 1971 }, { "epoch": 0.5993920972644377, "grad_norm": 0.27734375, "learning_rate": 4.006079027355623e-06, "loss": 1.0295, "step": 1972 }, { "epoch": 0.5996960486322188, "grad_norm": 0.2734375, "learning_rate": 4.003039513677812e-06, "loss": 1.0047, "step": 1973 }, { "epoch": 0.6, "grad_norm": 0.287109375, "learning_rate": 4.000000000000001e-06, "loss": 1.0454, "step": 1974 }, { "epoch": 0.6003039513677811, "grad_norm": 0.283203125, "learning_rate": 3.9969604863221885e-06, "loss": 1.0317, "step": 1975 }, { "epoch": 0.6006079027355623, "grad_norm": 0.29296875, "learning_rate": 3.993920972644377e-06, "loss": 1.1341, "step": 1976 }, { "epoch": 0.6009118541033435, "grad_norm": 0.26953125, "learning_rate": 3.990881458966566e-06, "loss": 1.1335, "step": 1977 }, { "epoch": 0.6012158054711246, "grad_norm": 0.27734375, "learning_rate": 3.987841945288754e-06, "loss": 1.0608, "step": 1978 }, { "epoch": 0.6015197568389058, "grad_norm": 0.30078125, "learning_rate": 3.984802431610942e-06, "loss": 0.9985, "step": 1979 }, { "epoch": 0.601823708206687, "grad_norm": 0.326171875, "learning_rate": 3.981762917933131e-06, "loss": 0.9978, "step": 1980 }, { "epoch": 0.6021276595744681, "grad_norm": 0.322265625, "learning_rate": 3.97872340425532e-06, "loss": 1.0142, "step": 1981 }, { "epoch": 0.6024316109422493, "grad_norm": 0.279296875, "learning_rate": 3.9756838905775085e-06, "loss": 1.043, "step": 1982 }, { "epoch": 0.6027355623100304, "grad_norm": 0.29296875, "learning_rate": 3.972644376899696e-06, "loss": 0.9673, "step": 1983 }, { "epoch": 0.6030395136778115, "grad_norm": 0.283203125, "learning_rate": 3.969604863221885e-06, "loss": 1.0303, "step": 1984 }, { "epoch": 0.6033434650455927, "grad_norm": 0.283203125, "learning_rate": 3.966565349544073e-06, "loss": 1.0882, "step": 1985 }, { "epoch": 0.6036474164133738, "grad_norm": 0.2734375, "learning_rate": 3.9635258358662615e-06, "loss": 1.0696, "step": 1986 }, { "epoch": 0.603951367781155, "grad_norm": 0.283203125, "learning_rate": 3.96048632218845e-06, "loss": 1.1231, "step": 1987 }, { "epoch": 0.6042553191489362, "grad_norm": 0.29296875, "learning_rate": 3.957446808510639e-06, "loss": 1.0159, "step": 1988 }, { "epoch": 0.6045592705167173, "grad_norm": 0.279296875, "learning_rate": 3.9544072948328276e-06, "loss": 1.0514, "step": 1989 }, { "epoch": 0.6048632218844985, "grad_norm": 0.28125, "learning_rate": 3.951367781155015e-06, "loss": 1.0685, "step": 1990 }, { "epoch": 0.6051671732522796, "grad_norm": 0.283203125, "learning_rate": 3.948328267477204e-06, "loss": 1.0454, "step": 1991 }, { "epoch": 0.6054711246200608, "grad_norm": 0.359375, "learning_rate": 3.945288753799392e-06, "loss": 1.1023, "step": 1992 }, { "epoch": 0.605775075987842, "grad_norm": 0.28515625, "learning_rate": 3.942249240121581e-06, "loss": 1.0812, "step": 1993 }, { "epoch": 0.6060790273556231, "grad_norm": 0.298828125, "learning_rate": 3.939209726443769e-06, "loss": 1.0286, "step": 1994 }, { "epoch": 0.6063829787234043, "grad_norm": 0.28515625, "learning_rate": 3.936170212765958e-06, "loss": 1.1316, "step": 1995 }, { "epoch": 0.6066869300911855, "grad_norm": 0.2890625, "learning_rate": 3.933130699088147e-06, "loss": 0.9885, "step": 1996 }, { "epoch": 0.6069908814589665, "grad_norm": 0.275390625, "learning_rate": 3.9300911854103345e-06, "loss": 1.0254, "step": 1997 }, { "epoch": 0.6072948328267477, "grad_norm": 0.29296875, "learning_rate": 3.927051671732523e-06, "loss": 1.1529, "step": 1998 }, { "epoch": 0.6075987841945288, "grad_norm": 0.287109375, "learning_rate": 3.924012158054711e-06, "loss": 1.0401, "step": 1999 }, { "epoch": 0.60790273556231, "grad_norm": 0.279296875, "learning_rate": 3.9209726443769005e-06, "loss": 1.0157, "step": 2000 }, { "epoch": 0.6082066869300912, "grad_norm": 0.28515625, "learning_rate": 3.917933130699088e-06, "loss": 1.0252, "step": 2001 }, { "epoch": 0.6085106382978723, "grad_norm": 0.291015625, "learning_rate": 3.914893617021277e-06, "loss": 1.0344, "step": 2002 }, { "epoch": 0.6088145896656535, "grad_norm": 0.279296875, "learning_rate": 3.911854103343466e-06, "loss": 1.0922, "step": 2003 }, { "epoch": 0.6091185410334347, "grad_norm": 0.376953125, "learning_rate": 3.908814589665654e-06, "loss": 1.0083, "step": 2004 }, { "epoch": 0.6094224924012158, "grad_norm": 0.326171875, "learning_rate": 3.905775075987842e-06, "loss": 0.9871, "step": 2005 }, { "epoch": 0.609726443768997, "grad_norm": 0.267578125, "learning_rate": 3.90273556231003e-06, "loss": 1.0706, "step": 2006 }, { "epoch": 0.6100303951367781, "grad_norm": 0.302734375, "learning_rate": 3.89969604863222e-06, "loss": 1.0479, "step": 2007 }, { "epoch": 0.6103343465045593, "grad_norm": 0.287109375, "learning_rate": 3.8966565349544075e-06, "loss": 1.037, "step": 2008 }, { "epoch": 0.6106382978723405, "grad_norm": 0.294921875, "learning_rate": 3.893617021276596e-06, "loss": 1.0172, "step": 2009 }, { "epoch": 0.6109422492401215, "grad_norm": 0.318359375, "learning_rate": 3.890577507598785e-06, "loss": 1.0753, "step": 2010 }, { "epoch": 0.6112462006079027, "grad_norm": 0.28125, "learning_rate": 3.887537993920973e-06, "loss": 1.0755, "step": 2011 }, { "epoch": 0.6115501519756839, "grad_norm": 0.2890625, "learning_rate": 3.884498480243161e-06, "loss": 1.0449, "step": 2012 }, { "epoch": 0.611854103343465, "grad_norm": 0.29296875, "learning_rate": 3.881458966565349e-06, "loss": 1.0056, "step": 2013 }, { "epoch": 0.6121580547112462, "grad_norm": 0.28515625, "learning_rate": 3.878419452887539e-06, "loss": 1.0289, "step": 2014 }, { "epoch": 0.6124620060790273, "grad_norm": 0.298828125, "learning_rate": 3.875379939209727e-06, "loss": 1.1198, "step": 2015 }, { "epoch": 0.6127659574468085, "grad_norm": 0.275390625, "learning_rate": 3.872340425531915e-06, "loss": 1.0661, "step": 2016 }, { "epoch": 0.6130699088145897, "grad_norm": 0.28515625, "learning_rate": 3.869300911854104e-06, "loss": 1.055, "step": 2017 }, { "epoch": 0.6133738601823708, "grad_norm": 0.41015625, "learning_rate": 3.866261398176292e-06, "loss": 1.0762, "step": 2018 }, { "epoch": 0.613677811550152, "grad_norm": 0.279296875, "learning_rate": 3.8632218844984805e-06, "loss": 1.0461, "step": 2019 }, { "epoch": 0.6139817629179332, "grad_norm": 0.275390625, "learning_rate": 3.860182370820669e-06, "loss": 1.1066, "step": 2020 }, { "epoch": 0.6142857142857143, "grad_norm": 0.283203125, "learning_rate": 3.857142857142858e-06, "loss": 0.9762, "step": 2021 }, { "epoch": 0.6145896656534955, "grad_norm": 0.283203125, "learning_rate": 3.854103343465046e-06, "loss": 1.0252, "step": 2022 }, { "epoch": 0.6148936170212767, "grad_norm": 0.271484375, "learning_rate": 3.851063829787234e-06, "loss": 1.1155, "step": 2023 }, { "epoch": 0.6151975683890577, "grad_norm": 0.2890625, "learning_rate": 3.848024316109423e-06, "loss": 1.0179, "step": 2024 }, { "epoch": 0.6155015197568389, "grad_norm": 0.28125, "learning_rate": 3.844984802431611e-06, "loss": 1.0802, "step": 2025 }, { "epoch": 0.61580547112462, "grad_norm": 0.279296875, "learning_rate": 3.8419452887537996e-06, "loss": 1.069, "step": 2026 }, { "epoch": 0.6161094224924012, "grad_norm": 0.28515625, "learning_rate": 3.838905775075988e-06, "loss": 1.0236, "step": 2027 }, { "epoch": 0.6164133738601824, "grad_norm": 0.3046875, "learning_rate": 3.835866261398177e-06, "loss": 1.1477, "step": 2028 }, { "epoch": 0.6167173252279635, "grad_norm": 0.28125, "learning_rate": 3.832826747720365e-06, "loss": 1.0435, "step": 2029 }, { "epoch": 0.6170212765957447, "grad_norm": 0.283203125, "learning_rate": 3.8297872340425535e-06, "loss": 1.0814, "step": 2030 }, { "epoch": 0.6173252279635258, "grad_norm": 0.279296875, "learning_rate": 3.826747720364742e-06, "loss": 1.1096, "step": 2031 }, { "epoch": 0.617629179331307, "grad_norm": 0.275390625, "learning_rate": 3.82370820668693e-06, "loss": 1.0501, "step": 2032 }, { "epoch": 0.6179331306990882, "grad_norm": 0.28125, "learning_rate": 3.820668693009119e-06, "loss": 1.0282, "step": 2033 }, { "epoch": 0.6182370820668693, "grad_norm": 0.298828125, "learning_rate": 3.817629179331307e-06, "loss": 0.989, "step": 2034 }, { "epoch": 0.6185410334346505, "grad_norm": 0.28125, "learning_rate": 3.814589665653496e-06, "loss": 1.0378, "step": 2035 }, { "epoch": 0.6188449848024317, "grad_norm": 0.28125, "learning_rate": 3.8115501519756843e-06, "loss": 1.0334, "step": 2036 }, { "epoch": 0.6191489361702127, "grad_norm": 0.291015625, "learning_rate": 3.8085106382978726e-06, "loss": 1.0344, "step": 2037 }, { "epoch": 0.6194528875379939, "grad_norm": 0.294921875, "learning_rate": 3.805471124620061e-06, "loss": 1.0888, "step": 2038 }, { "epoch": 0.619756838905775, "grad_norm": 0.28125, "learning_rate": 3.8024316109422495e-06, "loss": 1.0446, "step": 2039 }, { "epoch": 0.6200607902735562, "grad_norm": 0.27734375, "learning_rate": 3.799392097264438e-06, "loss": 1.0846, "step": 2040 }, { "epoch": 0.6203647416413374, "grad_norm": 0.279296875, "learning_rate": 3.7963525835866265e-06, "loss": 1.036, "step": 2041 }, { "epoch": 0.6206686930091185, "grad_norm": 0.29296875, "learning_rate": 3.793313069908815e-06, "loss": 1.0368, "step": 2042 }, { "epoch": 0.6209726443768997, "grad_norm": 0.30859375, "learning_rate": 3.7902735562310034e-06, "loss": 1.1107, "step": 2043 }, { "epoch": 0.6212765957446809, "grad_norm": 0.30078125, "learning_rate": 3.7872340425531917e-06, "loss": 1.0646, "step": 2044 }, { "epoch": 0.621580547112462, "grad_norm": 0.283203125, "learning_rate": 3.78419452887538e-06, "loss": 1.0728, "step": 2045 }, { "epoch": 0.6218844984802432, "grad_norm": 0.30078125, "learning_rate": 3.7811550151975686e-06, "loss": 1.1844, "step": 2046 }, { "epoch": 0.6221884498480243, "grad_norm": 0.29296875, "learning_rate": 3.7781155015197573e-06, "loss": 1.087, "step": 2047 }, { "epoch": 0.6224924012158055, "grad_norm": 0.27734375, "learning_rate": 3.7750759878419456e-06, "loss": 1.0594, "step": 2048 }, { "epoch": 0.6227963525835867, "grad_norm": 0.271484375, "learning_rate": 3.7720364741641342e-06, "loss": 1.0777, "step": 2049 }, { "epoch": 0.6231003039513677, "grad_norm": 0.2734375, "learning_rate": 3.7689969604863225e-06, "loss": 1.0383, "step": 2050 }, { "epoch": 0.6234042553191489, "grad_norm": 0.291015625, "learning_rate": 3.7659574468085108e-06, "loss": 1.0119, "step": 2051 }, { "epoch": 0.6237082066869301, "grad_norm": 0.291015625, "learning_rate": 3.762917933130699e-06, "loss": 0.9689, "step": 2052 }, { "epoch": 0.6240121580547112, "grad_norm": 0.279296875, "learning_rate": 3.759878419452888e-06, "loss": 1.077, "step": 2053 }, { "epoch": 0.6243161094224924, "grad_norm": 0.287109375, "learning_rate": 3.7568389057750764e-06, "loss": 1.0297, "step": 2054 }, { "epoch": 0.6246200607902735, "grad_norm": 0.275390625, "learning_rate": 3.7537993920972647e-06, "loss": 0.9794, "step": 2055 }, { "epoch": 0.6249240121580547, "grad_norm": 0.296875, "learning_rate": 3.7507598784194533e-06, "loss": 1.0536, "step": 2056 }, { "epoch": 0.6252279635258359, "grad_norm": 0.28125, "learning_rate": 3.7477203647416416e-06, "loss": 1.0334, "step": 2057 }, { "epoch": 0.625531914893617, "grad_norm": 0.27734375, "learning_rate": 3.74468085106383e-06, "loss": 0.9318, "step": 2058 }, { "epoch": 0.6258358662613982, "grad_norm": 0.28515625, "learning_rate": 3.741641337386018e-06, "loss": 1.0182, "step": 2059 }, { "epoch": 0.6261398176291794, "grad_norm": 0.283203125, "learning_rate": 3.7386018237082072e-06, "loss": 1.0776, "step": 2060 }, { "epoch": 0.6264437689969605, "grad_norm": 0.29296875, "learning_rate": 3.7355623100303955e-06, "loss": 1.1111, "step": 2061 }, { "epoch": 0.6267477203647417, "grad_norm": 0.271484375, "learning_rate": 3.7325227963525838e-06, "loss": 1.0449, "step": 2062 }, { "epoch": 0.6270516717325227, "grad_norm": 0.2890625, "learning_rate": 3.7294832826747724e-06, "loss": 1.0146, "step": 2063 }, { "epoch": 0.6273556231003039, "grad_norm": 0.28515625, "learning_rate": 3.7264437689969607e-06, "loss": 1.0194, "step": 2064 }, { "epoch": 0.6276595744680851, "grad_norm": 0.294921875, "learning_rate": 3.723404255319149e-06, "loss": 1.1033, "step": 2065 }, { "epoch": 0.6279635258358662, "grad_norm": 0.28125, "learning_rate": 3.7203647416413372e-06, "loss": 1.1007, "step": 2066 }, { "epoch": 0.6282674772036474, "grad_norm": 0.279296875, "learning_rate": 3.7173252279635263e-06, "loss": 0.9903, "step": 2067 }, { "epoch": 0.6285714285714286, "grad_norm": 0.27734375, "learning_rate": 3.7142857142857146e-06, "loss": 1.0656, "step": 2068 }, { "epoch": 0.6288753799392097, "grad_norm": 0.275390625, "learning_rate": 3.7112462006079033e-06, "loss": 1.0707, "step": 2069 }, { "epoch": 0.6291793313069909, "grad_norm": 0.279296875, "learning_rate": 3.7082066869300915e-06, "loss": 1.1111, "step": 2070 }, { "epoch": 0.629483282674772, "grad_norm": 0.34375, "learning_rate": 3.70516717325228e-06, "loss": 1.0432, "step": 2071 }, { "epoch": 0.6297872340425532, "grad_norm": 0.28515625, "learning_rate": 3.702127659574468e-06, "loss": 1.046, "step": 2072 }, { "epoch": 0.6300911854103344, "grad_norm": 0.28125, "learning_rate": 3.699088145896657e-06, "loss": 1.0726, "step": 2073 }, { "epoch": 0.6303951367781155, "grad_norm": 0.26953125, "learning_rate": 3.6960486322188454e-06, "loss": 1.0868, "step": 2074 }, { "epoch": 0.6306990881458967, "grad_norm": 0.28125, "learning_rate": 3.6930091185410337e-06, "loss": 0.976, "step": 2075 }, { "epoch": 0.6310030395136779, "grad_norm": 0.294921875, "learning_rate": 3.6899696048632224e-06, "loss": 1.0057, "step": 2076 }, { "epoch": 0.6313069908814589, "grad_norm": 0.28515625, "learning_rate": 3.6869300911854106e-06, "loss": 0.9865, "step": 2077 }, { "epoch": 0.6316109422492401, "grad_norm": 0.279296875, "learning_rate": 3.683890577507599e-06, "loss": 1.0748, "step": 2078 }, { "epoch": 0.6319148936170212, "grad_norm": 0.27734375, "learning_rate": 3.680851063829787e-06, "loss": 1.0494, "step": 2079 }, { "epoch": 0.6322188449848024, "grad_norm": 0.318359375, "learning_rate": 3.6778115501519763e-06, "loss": 1.1538, "step": 2080 }, { "epoch": 0.6325227963525836, "grad_norm": 0.291015625, "learning_rate": 3.6747720364741645e-06, "loss": 1.0515, "step": 2081 }, { "epoch": 0.6328267477203647, "grad_norm": 0.291015625, "learning_rate": 3.671732522796353e-06, "loss": 1.0494, "step": 2082 }, { "epoch": 0.6331306990881459, "grad_norm": 0.3125, "learning_rate": 3.6686930091185415e-06, "loss": 1.0188, "step": 2083 }, { "epoch": 0.6334346504559271, "grad_norm": 0.28125, "learning_rate": 3.6656534954407297e-06, "loss": 1.0744, "step": 2084 }, { "epoch": 0.6337386018237082, "grad_norm": 0.287109375, "learning_rate": 3.662613981762918e-06, "loss": 1.076, "step": 2085 }, { "epoch": 0.6340425531914894, "grad_norm": 0.279296875, "learning_rate": 3.6595744680851063e-06, "loss": 1.0639, "step": 2086 }, { "epoch": 0.6343465045592706, "grad_norm": 0.283203125, "learning_rate": 3.6565349544072954e-06, "loss": 1.0516, "step": 2087 }, { "epoch": 0.6346504559270517, "grad_norm": 0.345703125, "learning_rate": 3.6534954407294836e-06, "loss": 0.9772, "step": 2088 }, { "epoch": 0.6349544072948329, "grad_norm": 0.28125, "learning_rate": 3.650455927051672e-06, "loss": 1.0543, "step": 2089 }, { "epoch": 0.6352583586626139, "grad_norm": 0.283203125, "learning_rate": 3.6474164133738606e-06, "loss": 1.0356, "step": 2090 }, { "epoch": 0.6355623100303951, "grad_norm": 0.29296875, "learning_rate": 3.644376899696049e-06, "loss": 1.0611, "step": 2091 }, { "epoch": 0.6358662613981763, "grad_norm": 0.287109375, "learning_rate": 3.641337386018237e-06, "loss": 1.0627, "step": 2092 }, { "epoch": 0.6361702127659574, "grad_norm": 0.287109375, "learning_rate": 3.6382978723404262e-06, "loss": 0.983, "step": 2093 }, { "epoch": 0.6364741641337386, "grad_norm": 0.2890625, "learning_rate": 3.6352583586626145e-06, "loss": 1.0536, "step": 2094 }, { "epoch": 0.6367781155015197, "grad_norm": 0.30078125, "learning_rate": 3.6322188449848027e-06, "loss": 1.0127, "step": 2095 }, { "epoch": 0.6370820668693009, "grad_norm": 0.26953125, "learning_rate": 3.629179331306991e-06, "loss": 1.0622, "step": 2096 }, { "epoch": 0.6373860182370821, "grad_norm": 0.283203125, "learning_rate": 3.6261398176291797e-06, "loss": 1.1342, "step": 2097 }, { "epoch": 0.6376899696048632, "grad_norm": 0.27734375, "learning_rate": 3.623100303951368e-06, "loss": 1.0411, "step": 2098 }, { "epoch": 0.6379939209726444, "grad_norm": 0.279296875, "learning_rate": 3.620060790273556e-06, "loss": 1.0693, "step": 2099 }, { "epoch": 0.6382978723404256, "grad_norm": 0.279296875, "learning_rate": 3.6170212765957453e-06, "loss": 1.1231, "step": 2100 }, { "epoch": 0.6386018237082067, "grad_norm": 0.27734375, "learning_rate": 3.6139817629179336e-06, "loss": 0.9702, "step": 2101 }, { "epoch": 0.6389057750759879, "grad_norm": 0.3359375, "learning_rate": 3.610942249240122e-06, "loss": 1.0171, "step": 2102 }, { "epoch": 0.6392097264437689, "grad_norm": 0.318359375, "learning_rate": 3.60790273556231e-06, "loss": 0.9676, "step": 2103 }, { "epoch": 0.6395136778115501, "grad_norm": 0.283203125, "learning_rate": 3.604863221884499e-06, "loss": 1.0334, "step": 2104 }, { "epoch": 0.6398176291793313, "grad_norm": 0.279296875, "learning_rate": 3.601823708206687e-06, "loss": 1.0737, "step": 2105 }, { "epoch": 0.6401215805471124, "grad_norm": 0.2734375, "learning_rate": 3.598784194528876e-06, "loss": 1.0678, "step": 2106 }, { "epoch": 0.6404255319148936, "grad_norm": 0.283203125, "learning_rate": 3.5957446808510644e-06, "loss": 0.9606, "step": 2107 }, { "epoch": 0.6407294832826748, "grad_norm": 0.28515625, "learning_rate": 3.5927051671732527e-06, "loss": 1.0088, "step": 2108 }, { "epoch": 0.6410334346504559, "grad_norm": 0.29296875, "learning_rate": 3.589665653495441e-06, "loss": 1.0779, "step": 2109 }, { "epoch": 0.6413373860182371, "grad_norm": 0.291015625, "learning_rate": 3.586626139817629e-06, "loss": 0.9837, "step": 2110 }, { "epoch": 0.6416413373860183, "grad_norm": 0.306640625, "learning_rate": 3.583586626139818e-06, "loss": 1.0282, "step": 2111 }, { "epoch": 0.6419452887537994, "grad_norm": 0.275390625, "learning_rate": 3.580547112462006e-06, "loss": 1.0593, "step": 2112 }, { "epoch": 0.6422492401215806, "grad_norm": 0.291015625, "learning_rate": 3.5775075987841953e-06, "loss": 1.0836, "step": 2113 }, { "epoch": 0.6425531914893617, "grad_norm": 0.27734375, "learning_rate": 3.5744680851063835e-06, "loss": 1.0045, "step": 2114 }, { "epoch": 0.6428571428571429, "grad_norm": 0.283203125, "learning_rate": 3.5714285714285718e-06, "loss": 1.0402, "step": 2115 }, { "epoch": 0.643161094224924, "grad_norm": 0.28125, "learning_rate": 3.56838905775076e-06, "loss": 1.0591, "step": 2116 }, { "epoch": 0.6434650455927051, "grad_norm": 0.287109375, "learning_rate": 3.5653495440729487e-06, "loss": 1.1262, "step": 2117 }, { "epoch": 0.6437689969604863, "grad_norm": 0.283203125, "learning_rate": 3.562310030395137e-06, "loss": 1.1158, "step": 2118 }, { "epoch": 0.6440729483282674, "grad_norm": 0.2890625, "learning_rate": 3.5592705167173253e-06, "loss": 1.019, "step": 2119 }, { "epoch": 0.6443768996960486, "grad_norm": 0.287109375, "learning_rate": 3.5562310030395144e-06, "loss": 1.0986, "step": 2120 }, { "epoch": 0.6446808510638298, "grad_norm": 0.275390625, "learning_rate": 3.5531914893617026e-06, "loss": 1.1826, "step": 2121 }, { "epoch": 0.6449848024316109, "grad_norm": 0.287109375, "learning_rate": 3.550151975683891e-06, "loss": 1.0526, "step": 2122 }, { "epoch": 0.6452887537993921, "grad_norm": 0.287109375, "learning_rate": 3.547112462006079e-06, "loss": 1.071, "step": 2123 }, { "epoch": 0.6455927051671733, "grad_norm": 0.27734375, "learning_rate": 3.544072948328268e-06, "loss": 1.0684, "step": 2124 }, { "epoch": 0.6458966565349544, "grad_norm": 0.296875, "learning_rate": 3.541033434650456e-06, "loss": 1.0146, "step": 2125 }, { "epoch": 0.6462006079027356, "grad_norm": 0.296875, "learning_rate": 3.5379939209726448e-06, "loss": 0.9867, "step": 2126 }, { "epoch": 0.6465045592705168, "grad_norm": 0.28515625, "learning_rate": 3.5349544072948335e-06, "loss": 1.0548, "step": 2127 }, { "epoch": 0.6468085106382979, "grad_norm": 0.287109375, "learning_rate": 3.5319148936170217e-06, "loss": 1.0287, "step": 2128 }, { "epoch": 0.6471124620060791, "grad_norm": 0.291015625, "learning_rate": 3.52887537993921e-06, "loss": 1.0368, "step": 2129 }, { "epoch": 0.6474164133738601, "grad_norm": 0.28125, "learning_rate": 3.5258358662613982e-06, "loss": 1.1127, "step": 2130 }, { "epoch": 0.6477203647416413, "grad_norm": 0.296875, "learning_rate": 3.522796352583587e-06, "loss": 1.0888, "step": 2131 }, { "epoch": 0.6480243161094225, "grad_norm": 0.27734375, "learning_rate": 3.519756838905775e-06, "loss": 1.0314, "step": 2132 }, { "epoch": 0.6483282674772036, "grad_norm": 0.287109375, "learning_rate": 3.516717325227964e-06, "loss": 1.095, "step": 2133 }, { "epoch": 0.6486322188449848, "grad_norm": 0.296875, "learning_rate": 3.5136778115501526e-06, "loss": 1.0232, "step": 2134 }, { "epoch": 0.648936170212766, "grad_norm": 0.28125, "learning_rate": 3.510638297872341e-06, "loss": 1.0142, "step": 2135 }, { "epoch": 0.6492401215805471, "grad_norm": 0.291015625, "learning_rate": 3.507598784194529e-06, "loss": 1.0658, "step": 2136 }, { "epoch": 0.6495440729483283, "grad_norm": 0.279296875, "learning_rate": 3.5045592705167173e-06, "loss": 1.0307, "step": 2137 }, { "epoch": 0.6498480243161094, "grad_norm": 0.283203125, "learning_rate": 3.501519756838906e-06, "loss": 1.1055, "step": 2138 }, { "epoch": 0.6501519756838906, "grad_norm": 0.296875, "learning_rate": 3.4984802431610943e-06, "loss": 0.9458, "step": 2139 }, { "epoch": 0.6504559270516718, "grad_norm": 0.404296875, "learning_rate": 3.495440729483283e-06, "loss": 1.0733, "step": 2140 }, { "epoch": 0.6507598784194529, "grad_norm": 0.283203125, "learning_rate": 3.4924012158054717e-06, "loss": 1.067, "step": 2141 }, { "epoch": 0.6510638297872341, "grad_norm": 0.27734375, "learning_rate": 3.48936170212766e-06, "loss": 1.0596, "step": 2142 }, { "epoch": 0.6513677811550151, "grad_norm": 0.279296875, "learning_rate": 3.486322188449848e-06, "loss": 1.0118, "step": 2143 }, { "epoch": 0.6516717325227963, "grad_norm": 0.28125, "learning_rate": 3.4832826747720364e-06, "loss": 1.0408, "step": 2144 }, { "epoch": 0.6519756838905775, "grad_norm": 0.2890625, "learning_rate": 3.480243161094225e-06, "loss": 1.0801, "step": 2145 }, { "epoch": 0.6522796352583586, "grad_norm": 0.291015625, "learning_rate": 3.477203647416414e-06, "loss": 1.0196, "step": 2146 }, { "epoch": 0.6525835866261398, "grad_norm": 0.2734375, "learning_rate": 3.474164133738602e-06, "loss": 1.02, "step": 2147 }, { "epoch": 0.652887537993921, "grad_norm": 0.3125, "learning_rate": 3.4711246200607908e-06, "loss": 1.0352, "step": 2148 }, { "epoch": 0.6531914893617021, "grad_norm": 0.2890625, "learning_rate": 3.468085106382979e-06, "loss": 1.0463, "step": 2149 }, { "epoch": 0.6534954407294833, "grad_norm": 0.294921875, "learning_rate": 3.4650455927051673e-06, "loss": 1.0546, "step": 2150 }, { "epoch": 0.6537993920972645, "grad_norm": 0.298828125, "learning_rate": 3.4620060790273555e-06, "loss": 1.0497, "step": 2151 }, { "epoch": 0.6541033434650456, "grad_norm": 0.28515625, "learning_rate": 3.4589665653495442e-06, "loss": 1.0892, "step": 2152 }, { "epoch": 0.6544072948328268, "grad_norm": 0.2734375, "learning_rate": 3.455927051671733e-06, "loss": 0.9629, "step": 2153 }, { "epoch": 0.6547112462006079, "grad_norm": 0.27734375, "learning_rate": 3.4528875379939216e-06, "loss": 1.0469, "step": 2154 }, { "epoch": 0.6550151975683891, "grad_norm": 0.306640625, "learning_rate": 3.44984802431611e-06, "loss": 1.0427, "step": 2155 }, { "epoch": 0.6553191489361702, "grad_norm": 0.28515625, "learning_rate": 3.446808510638298e-06, "loss": 1.0272, "step": 2156 }, { "epoch": 0.6556231003039513, "grad_norm": 0.28515625, "learning_rate": 3.4437689969604864e-06, "loss": 1.1235, "step": 2157 }, { "epoch": 0.6559270516717325, "grad_norm": 0.29296875, "learning_rate": 3.440729483282675e-06, "loss": 1.0123, "step": 2158 }, { "epoch": 0.6562310030395136, "grad_norm": 0.28515625, "learning_rate": 3.4376899696048633e-06, "loss": 1.0749, "step": 2159 }, { "epoch": 0.6565349544072948, "grad_norm": 0.283203125, "learning_rate": 3.434650455927052e-06, "loss": 1.0726, "step": 2160 }, { "epoch": 0.656838905775076, "grad_norm": 0.291015625, "learning_rate": 3.4316109422492407e-06, "loss": 1.0122, "step": 2161 }, { "epoch": 0.6571428571428571, "grad_norm": 0.283203125, "learning_rate": 3.428571428571429e-06, "loss": 1.105, "step": 2162 }, { "epoch": 0.6574468085106383, "grad_norm": 0.287109375, "learning_rate": 3.4255319148936172e-06, "loss": 1.0368, "step": 2163 }, { "epoch": 0.6577507598784195, "grad_norm": 0.291015625, "learning_rate": 3.4224924012158055e-06, "loss": 1.009, "step": 2164 }, { "epoch": 0.6580547112462006, "grad_norm": 0.291015625, "learning_rate": 3.419452887537994e-06, "loss": 1.0545, "step": 2165 }, { "epoch": 0.6583586626139818, "grad_norm": 0.28125, "learning_rate": 3.416413373860183e-06, "loss": 1.0838, "step": 2166 }, { "epoch": 0.658662613981763, "grad_norm": 0.287109375, "learning_rate": 3.413373860182371e-06, "loss": 1.0873, "step": 2167 }, { "epoch": 0.6589665653495441, "grad_norm": 0.2890625, "learning_rate": 3.41033434650456e-06, "loss": 1.0665, "step": 2168 }, { "epoch": 0.6592705167173253, "grad_norm": 0.2890625, "learning_rate": 3.407294832826748e-06, "loss": 1.0076, "step": 2169 }, { "epoch": 0.6595744680851063, "grad_norm": 0.28515625, "learning_rate": 3.4042553191489363e-06, "loss": 1.1086, "step": 2170 }, { "epoch": 0.6598784194528875, "grad_norm": 0.27734375, "learning_rate": 3.4012158054711246e-06, "loss": 1.0961, "step": 2171 }, { "epoch": 0.6601823708206687, "grad_norm": 0.283203125, "learning_rate": 3.3981762917933133e-06, "loss": 1.0035, "step": 2172 }, { "epoch": 0.6604863221884498, "grad_norm": 0.28125, "learning_rate": 3.395136778115502e-06, "loss": 1.065, "step": 2173 }, { "epoch": 0.660790273556231, "grad_norm": 0.287109375, "learning_rate": 3.3920972644376902e-06, "loss": 1.0857, "step": 2174 }, { "epoch": 0.6610942249240122, "grad_norm": 0.296875, "learning_rate": 3.389057750759879e-06, "loss": 1.0588, "step": 2175 }, { "epoch": 0.6613981762917933, "grad_norm": 0.2890625, "learning_rate": 3.386018237082067e-06, "loss": 1.0651, "step": 2176 }, { "epoch": 0.6617021276595745, "grad_norm": 0.2890625, "learning_rate": 3.3829787234042554e-06, "loss": 1.0796, "step": 2177 }, { "epoch": 0.6620060790273556, "grad_norm": 0.28515625, "learning_rate": 3.3799392097264437e-06, "loss": 1.0466, "step": 2178 }, { "epoch": 0.6623100303951368, "grad_norm": 0.328125, "learning_rate": 3.376899696048633e-06, "loss": 1.0091, "step": 2179 }, { "epoch": 0.662613981762918, "grad_norm": 0.27734375, "learning_rate": 3.373860182370821e-06, "loss": 1.0871, "step": 2180 }, { "epoch": 0.6629179331306991, "grad_norm": 0.283203125, "learning_rate": 3.3708206686930093e-06, "loss": 0.987, "step": 2181 }, { "epoch": 0.6632218844984803, "grad_norm": 0.279296875, "learning_rate": 3.367781155015198e-06, "loss": 1.1005, "step": 2182 }, { "epoch": 0.6635258358662613, "grad_norm": 0.263671875, "learning_rate": 3.3647416413373863e-06, "loss": 1.0665, "step": 2183 }, { "epoch": 0.6638297872340425, "grad_norm": 0.2734375, "learning_rate": 3.3617021276595745e-06, "loss": 1.0072, "step": 2184 }, { "epoch": 0.6641337386018237, "grad_norm": 0.298828125, "learning_rate": 3.3586626139817628e-06, "loss": 1.0915, "step": 2185 }, { "epoch": 0.6644376899696048, "grad_norm": 0.2890625, "learning_rate": 3.355623100303952e-06, "loss": 0.9977, "step": 2186 }, { "epoch": 0.664741641337386, "grad_norm": 0.294921875, "learning_rate": 3.35258358662614e-06, "loss": 1.0839, "step": 2187 }, { "epoch": 0.6650455927051672, "grad_norm": 0.31640625, "learning_rate": 3.3495440729483284e-06, "loss": 1.0582, "step": 2188 }, { "epoch": 0.6653495440729483, "grad_norm": 0.291015625, "learning_rate": 3.346504559270517e-06, "loss": 1.0469, "step": 2189 }, { "epoch": 0.6656534954407295, "grad_norm": 0.298828125, "learning_rate": 3.3434650455927054e-06, "loss": 0.9913, "step": 2190 }, { "epoch": 0.6659574468085107, "grad_norm": 0.27734375, "learning_rate": 3.3404255319148936e-06, "loss": 1.1254, "step": 2191 }, { "epoch": 0.6662613981762918, "grad_norm": 0.287109375, "learning_rate": 3.337386018237082e-06, "loss": 1.0577, "step": 2192 }, { "epoch": 0.666565349544073, "grad_norm": 0.28125, "learning_rate": 3.334346504559271e-06, "loss": 1.0217, "step": 2193 }, { "epoch": 0.6668693009118541, "grad_norm": 0.279296875, "learning_rate": 3.3313069908814593e-06, "loss": 1.0378, "step": 2194 }, { "epoch": 0.6671732522796353, "grad_norm": 0.287109375, "learning_rate": 3.328267477203648e-06, "loss": 1.0611, "step": 2195 }, { "epoch": 0.6674772036474164, "grad_norm": 0.279296875, "learning_rate": 3.325227963525836e-06, "loss": 1.0782, "step": 2196 }, { "epoch": 0.6677811550151975, "grad_norm": 0.287109375, "learning_rate": 3.3221884498480245e-06, "loss": 1.0374, "step": 2197 }, { "epoch": 0.6680851063829787, "grad_norm": 0.28515625, "learning_rate": 3.3191489361702127e-06, "loss": 1.0718, "step": 2198 }, { "epoch": 0.6683890577507599, "grad_norm": 0.279296875, "learning_rate": 3.316109422492402e-06, "loss": 1.0228, "step": 2199 }, { "epoch": 0.668693009118541, "grad_norm": 0.291015625, "learning_rate": 3.31306990881459e-06, "loss": 1.0642, "step": 2200 }, { "epoch": 0.6689969604863222, "grad_norm": 0.283203125, "learning_rate": 3.3100303951367784e-06, "loss": 1.0634, "step": 2201 }, { "epoch": 0.6693009118541033, "grad_norm": 0.287109375, "learning_rate": 3.306990881458967e-06, "loss": 1.0634, "step": 2202 }, { "epoch": 0.6696048632218845, "grad_norm": 0.28125, "learning_rate": 3.3039513677811553e-06, "loss": 1.1221, "step": 2203 }, { "epoch": 0.6699088145896657, "grad_norm": 0.283203125, "learning_rate": 3.3009118541033436e-06, "loss": 1.054, "step": 2204 }, { "epoch": 0.6702127659574468, "grad_norm": 0.2890625, "learning_rate": 3.297872340425532e-06, "loss": 1.1403, "step": 2205 }, { "epoch": 0.670516717325228, "grad_norm": 0.28515625, "learning_rate": 3.294832826747721e-06, "loss": 0.961, "step": 2206 }, { "epoch": 0.6708206686930092, "grad_norm": 0.28125, "learning_rate": 3.291793313069909e-06, "loss": 1.0171, "step": 2207 }, { "epoch": 0.6711246200607903, "grad_norm": 0.314453125, "learning_rate": 3.2887537993920975e-06, "loss": 0.982, "step": 2208 }, { "epoch": 0.6714285714285714, "grad_norm": 0.287109375, "learning_rate": 3.285714285714286e-06, "loss": 1.0695, "step": 2209 }, { "epoch": 0.6717325227963525, "grad_norm": 0.42578125, "learning_rate": 3.2826747720364744e-06, "loss": 1.0269, "step": 2210 }, { "epoch": 0.6720364741641337, "grad_norm": 0.287109375, "learning_rate": 3.2796352583586627e-06, "loss": 1.0767, "step": 2211 }, { "epoch": 0.6723404255319149, "grad_norm": 0.283203125, "learning_rate": 3.276595744680851e-06, "loss": 1.086, "step": 2212 }, { "epoch": 0.672644376899696, "grad_norm": 0.2890625, "learning_rate": 3.27355623100304e-06, "loss": 1.083, "step": 2213 }, { "epoch": 0.6729483282674772, "grad_norm": 0.27734375, "learning_rate": 3.2705167173252283e-06, "loss": 1.0259, "step": 2214 }, { "epoch": 0.6732522796352584, "grad_norm": 0.296875, "learning_rate": 3.2674772036474166e-06, "loss": 1.0818, "step": 2215 }, { "epoch": 0.6735562310030395, "grad_norm": 0.306640625, "learning_rate": 3.2644376899696052e-06, "loss": 1.0638, "step": 2216 }, { "epoch": 0.6738601823708207, "grad_norm": 0.291015625, "learning_rate": 3.2613981762917935e-06, "loss": 0.992, "step": 2217 }, { "epoch": 0.6741641337386018, "grad_norm": 0.29296875, "learning_rate": 3.2583586626139818e-06, "loss": 0.9888, "step": 2218 }, { "epoch": 0.674468085106383, "grad_norm": 0.283203125, "learning_rate": 3.255319148936171e-06, "loss": 1.0014, "step": 2219 }, { "epoch": 0.6747720364741642, "grad_norm": 0.27734375, "learning_rate": 3.252279635258359e-06, "loss": 1.1262, "step": 2220 }, { "epoch": 0.6750759878419453, "grad_norm": 0.2734375, "learning_rate": 3.2492401215805474e-06, "loss": 1.072, "step": 2221 }, { "epoch": 0.6753799392097265, "grad_norm": 0.28515625, "learning_rate": 3.2462006079027357e-06, "loss": 1.0814, "step": 2222 }, { "epoch": 0.6756838905775076, "grad_norm": 0.28515625, "learning_rate": 3.2431610942249243e-06, "loss": 1.0917, "step": 2223 }, { "epoch": 0.6759878419452887, "grad_norm": 0.28125, "learning_rate": 3.2401215805471126e-06, "loss": 0.9747, "step": 2224 }, { "epoch": 0.6762917933130699, "grad_norm": 0.283203125, "learning_rate": 3.237082066869301e-06, "loss": 1.1261, "step": 2225 }, { "epoch": 0.676595744680851, "grad_norm": 0.3125, "learning_rate": 3.23404255319149e-06, "loss": 0.9932, "step": 2226 }, { "epoch": 0.6768996960486322, "grad_norm": 0.287109375, "learning_rate": 3.2310030395136782e-06, "loss": 1.0234, "step": 2227 }, { "epoch": 0.6772036474164134, "grad_norm": 0.3203125, "learning_rate": 3.2279635258358665e-06, "loss": 1.0915, "step": 2228 }, { "epoch": 0.6775075987841945, "grad_norm": 0.294921875, "learning_rate": 3.2249240121580548e-06, "loss": 1.0888, "step": 2229 }, { "epoch": 0.6778115501519757, "grad_norm": 0.275390625, "learning_rate": 3.2218844984802434e-06, "loss": 0.9826, "step": 2230 }, { "epoch": 0.6781155015197569, "grad_norm": 0.29296875, "learning_rate": 3.2188449848024317e-06, "loss": 0.9942, "step": 2231 }, { "epoch": 0.678419452887538, "grad_norm": 0.298828125, "learning_rate": 3.21580547112462e-06, "loss": 1.1313, "step": 2232 }, { "epoch": 0.6787234042553192, "grad_norm": 0.27734375, "learning_rate": 3.212765957446809e-06, "loss": 1.0099, "step": 2233 }, { "epoch": 0.6790273556231003, "grad_norm": 0.294921875, "learning_rate": 3.2097264437689973e-06, "loss": 1.0356, "step": 2234 }, { "epoch": 0.6793313069908815, "grad_norm": 0.279296875, "learning_rate": 3.2066869300911856e-06, "loss": 1.0371, "step": 2235 }, { "epoch": 0.6796352583586626, "grad_norm": 0.28125, "learning_rate": 3.203647416413374e-06, "loss": 1.0376, "step": 2236 }, { "epoch": 0.6799392097264437, "grad_norm": 0.310546875, "learning_rate": 3.2006079027355625e-06, "loss": 1.1436, "step": 2237 }, { "epoch": 0.6802431610942249, "grad_norm": 0.294921875, "learning_rate": 3.197568389057751e-06, "loss": 1.0751, "step": 2238 }, { "epoch": 0.680547112462006, "grad_norm": 0.296875, "learning_rate": 3.19452887537994e-06, "loss": 0.9595, "step": 2239 }, { "epoch": 0.6808510638297872, "grad_norm": 0.28125, "learning_rate": 3.191489361702128e-06, "loss": 1.0574, "step": 2240 }, { "epoch": 0.6811550151975684, "grad_norm": 0.28515625, "learning_rate": 3.1884498480243164e-06, "loss": 0.9832, "step": 2241 }, { "epoch": 0.6814589665653495, "grad_norm": 0.294921875, "learning_rate": 3.1854103343465047e-06, "loss": 1.0503, "step": 2242 }, { "epoch": 0.6817629179331307, "grad_norm": 0.2890625, "learning_rate": 3.1823708206686934e-06, "loss": 1.0505, "step": 2243 }, { "epoch": 0.6820668693009119, "grad_norm": 0.29296875, "learning_rate": 3.1793313069908816e-06, "loss": 1.1019, "step": 2244 }, { "epoch": 0.682370820668693, "grad_norm": 0.29296875, "learning_rate": 3.17629179331307e-06, "loss": 1.0249, "step": 2245 }, { "epoch": 0.6826747720364742, "grad_norm": 0.2890625, "learning_rate": 3.173252279635259e-06, "loss": 1.0501, "step": 2246 }, { "epoch": 0.6829787234042554, "grad_norm": 0.28125, "learning_rate": 3.1702127659574473e-06, "loss": 1.0228, "step": 2247 }, { "epoch": 0.6832826747720365, "grad_norm": 0.287109375, "learning_rate": 3.1671732522796355e-06, "loss": 1.0403, "step": 2248 }, { "epoch": 0.6835866261398176, "grad_norm": 0.2890625, "learning_rate": 3.164133738601824e-06, "loss": 1.1047, "step": 2249 }, { "epoch": 0.6838905775075987, "grad_norm": 0.291015625, "learning_rate": 3.1610942249240125e-06, "loss": 1.042, "step": 2250 }, { "epoch": 0.6841945288753799, "grad_norm": 0.279296875, "learning_rate": 3.1580547112462007e-06, "loss": 1.0565, "step": 2251 }, { "epoch": 0.6844984802431611, "grad_norm": 0.294921875, "learning_rate": 3.1550151975683894e-06, "loss": 0.9813, "step": 2252 }, { "epoch": 0.6848024316109422, "grad_norm": 0.29296875, "learning_rate": 3.151975683890578e-06, "loss": 1.12, "step": 2253 }, { "epoch": 0.6851063829787234, "grad_norm": 0.298828125, "learning_rate": 3.1489361702127664e-06, "loss": 1.2079, "step": 2254 }, { "epoch": 0.6854103343465046, "grad_norm": 0.298828125, "learning_rate": 3.1458966565349546e-06, "loss": 0.9644, "step": 2255 }, { "epoch": 0.6857142857142857, "grad_norm": 0.287109375, "learning_rate": 3.142857142857143e-06, "loss": 1.0912, "step": 2256 }, { "epoch": 0.6860182370820669, "grad_norm": 0.34765625, "learning_rate": 3.1398176291793316e-06, "loss": 1.0789, "step": 2257 }, { "epoch": 0.686322188449848, "grad_norm": 0.2890625, "learning_rate": 3.13677811550152e-06, "loss": 1.0342, "step": 2258 }, { "epoch": 0.6866261398176292, "grad_norm": 0.279296875, "learning_rate": 3.1337386018237085e-06, "loss": 1.0156, "step": 2259 }, { "epoch": 0.6869300911854104, "grad_norm": 0.328125, "learning_rate": 3.1306990881458972e-06, "loss": 1.067, "step": 2260 }, { "epoch": 0.6872340425531915, "grad_norm": 0.27734375, "learning_rate": 3.1276595744680855e-06, "loss": 1.017, "step": 2261 }, { "epoch": 0.6875379939209726, "grad_norm": 0.2890625, "learning_rate": 3.1246200607902737e-06, "loss": 1.1223, "step": 2262 }, { "epoch": 0.6878419452887538, "grad_norm": 0.30078125, "learning_rate": 3.121580547112462e-06, "loss": 1.0246, "step": 2263 }, { "epoch": 0.6881458966565349, "grad_norm": 0.2890625, "learning_rate": 3.1185410334346507e-06, "loss": 1.0895, "step": 2264 }, { "epoch": 0.6884498480243161, "grad_norm": 0.30859375, "learning_rate": 3.115501519756839e-06, "loss": 1.0652, "step": 2265 }, { "epoch": 0.6887537993920972, "grad_norm": 0.296875, "learning_rate": 3.1124620060790276e-06, "loss": 1.1024, "step": 2266 }, { "epoch": 0.6890577507598784, "grad_norm": 0.28125, "learning_rate": 3.1094224924012163e-06, "loss": 1.1097, "step": 2267 }, { "epoch": 0.6893617021276596, "grad_norm": 0.291015625, "learning_rate": 3.1063829787234046e-06, "loss": 1.1078, "step": 2268 }, { "epoch": 0.6896656534954407, "grad_norm": 0.28515625, "learning_rate": 3.103343465045593e-06, "loss": 1.0758, "step": 2269 }, { "epoch": 0.6899696048632219, "grad_norm": 0.4140625, "learning_rate": 3.100303951367781e-06, "loss": 1.0726, "step": 2270 }, { "epoch": 0.6902735562310031, "grad_norm": 0.296875, "learning_rate": 3.0972644376899698e-06, "loss": 1.0927, "step": 2271 }, { "epoch": 0.6905775075987842, "grad_norm": 0.29296875, "learning_rate": 3.0942249240121585e-06, "loss": 1.0305, "step": 2272 }, { "epoch": 0.6908814589665654, "grad_norm": 0.291015625, "learning_rate": 3.0911854103343467e-06, "loss": 1.0332, "step": 2273 }, { "epoch": 0.6911854103343466, "grad_norm": 0.28515625, "learning_rate": 3.0881458966565354e-06, "loss": 1.0603, "step": 2274 }, { "epoch": 0.6914893617021277, "grad_norm": 0.283203125, "learning_rate": 3.0851063829787237e-06, "loss": 1.0755, "step": 2275 }, { "epoch": 0.6917933130699088, "grad_norm": 0.283203125, "learning_rate": 3.082066869300912e-06, "loss": 1.0606, "step": 2276 }, { "epoch": 0.6920972644376899, "grad_norm": 0.27734375, "learning_rate": 3.0790273556231e-06, "loss": 1.0464, "step": 2277 }, { "epoch": 0.6924012158054711, "grad_norm": 0.310546875, "learning_rate": 3.075987841945289e-06, "loss": 1.0533, "step": 2278 }, { "epoch": 0.6927051671732523, "grad_norm": 0.31640625, "learning_rate": 3.0729483282674776e-06, "loss": 1.0907, "step": 2279 }, { "epoch": 0.6930091185410334, "grad_norm": 0.30859375, "learning_rate": 3.0699088145896663e-06, "loss": 0.9784, "step": 2280 }, { "epoch": 0.6933130699088146, "grad_norm": 0.298828125, "learning_rate": 3.0668693009118545e-06, "loss": 1.0023, "step": 2281 }, { "epoch": 0.6936170212765957, "grad_norm": 0.28515625, "learning_rate": 3.0638297872340428e-06, "loss": 1.024, "step": 2282 }, { "epoch": 0.6939209726443769, "grad_norm": 0.28515625, "learning_rate": 3.060790273556231e-06, "loss": 1.1065, "step": 2283 }, { "epoch": 0.6942249240121581, "grad_norm": 0.2890625, "learning_rate": 3.0577507598784197e-06, "loss": 1.0438, "step": 2284 }, { "epoch": 0.6945288753799392, "grad_norm": 0.302734375, "learning_rate": 3.054711246200608e-06, "loss": 1.0333, "step": 2285 }, { "epoch": 0.6948328267477204, "grad_norm": 0.296875, "learning_rate": 3.0516717325227967e-06, "loss": 0.9629, "step": 2286 }, { "epoch": 0.6951367781155016, "grad_norm": 0.28125, "learning_rate": 3.0486322188449854e-06, "loss": 1.0013, "step": 2287 }, { "epoch": 0.6954407294832827, "grad_norm": 0.298828125, "learning_rate": 3.0455927051671736e-06, "loss": 1.0691, "step": 2288 }, { "epoch": 0.6957446808510638, "grad_norm": 0.287109375, "learning_rate": 3.042553191489362e-06, "loss": 1.0517, "step": 2289 }, { "epoch": 0.6960486322188449, "grad_norm": 0.2890625, "learning_rate": 3.03951367781155e-06, "loss": 1.0528, "step": 2290 }, { "epoch": 0.6963525835866261, "grad_norm": 0.287109375, "learning_rate": 3.036474164133739e-06, "loss": 1.024, "step": 2291 }, { "epoch": 0.6966565349544073, "grad_norm": 0.291015625, "learning_rate": 3.0334346504559275e-06, "loss": 1.0211, "step": 2292 }, { "epoch": 0.6969604863221884, "grad_norm": 0.2734375, "learning_rate": 3.0303951367781158e-06, "loss": 1.063, "step": 2293 }, { "epoch": 0.6972644376899696, "grad_norm": 0.283203125, "learning_rate": 3.0273556231003045e-06, "loss": 1.0073, "step": 2294 }, { "epoch": 0.6975683890577508, "grad_norm": 0.296875, "learning_rate": 3.0243161094224927e-06, "loss": 1.0188, "step": 2295 }, { "epoch": 0.6978723404255319, "grad_norm": 0.294921875, "learning_rate": 3.021276595744681e-06, "loss": 1.0655, "step": 2296 }, { "epoch": 0.6981762917933131, "grad_norm": 0.287109375, "learning_rate": 3.0182370820668692e-06, "loss": 1.112, "step": 2297 }, { "epoch": 0.6984802431610942, "grad_norm": 0.28125, "learning_rate": 3.015197568389058e-06, "loss": 1.0289, "step": 2298 }, { "epoch": 0.6987841945288754, "grad_norm": 0.2734375, "learning_rate": 3.0121580547112466e-06, "loss": 1.073, "step": 2299 }, { "epoch": 0.6990881458966566, "grad_norm": 0.287109375, "learning_rate": 3.009118541033435e-06, "loss": 1.0773, "step": 2300 }, { "epoch": 0.6993920972644377, "grad_norm": 0.28515625, "learning_rate": 3.0060790273556236e-06, "loss": 1.0404, "step": 2301 }, { "epoch": 0.6996960486322188, "grad_norm": 0.291015625, "learning_rate": 3.003039513677812e-06, "loss": 1.0131, "step": 2302 }, { "epoch": 0.7, "grad_norm": 0.2734375, "learning_rate": 3e-06, "loss": 1.0349, "step": 2303 }, { "epoch": 0.7003039513677811, "grad_norm": 0.28125, "learning_rate": 2.9969604863221883e-06, "loss": 1.0469, "step": 2304 }, { "epoch": 0.7006079027355623, "grad_norm": 0.287109375, "learning_rate": 2.9939209726443774e-06, "loss": 0.9936, "step": 2305 }, { "epoch": 0.7009118541033434, "grad_norm": 0.294921875, "learning_rate": 2.9908814589665657e-06, "loss": 1.0707, "step": 2306 }, { "epoch": 0.7012158054711246, "grad_norm": 0.283203125, "learning_rate": 2.987841945288754e-06, "loss": 0.9326, "step": 2307 }, { "epoch": 0.7015197568389058, "grad_norm": 0.30859375, "learning_rate": 2.9848024316109427e-06, "loss": 1.0338, "step": 2308 }, { "epoch": 0.7018237082066869, "grad_norm": 0.283203125, "learning_rate": 2.981762917933131e-06, "loss": 0.9966, "step": 2309 }, { "epoch": 0.7021276595744681, "grad_norm": 0.30078125, "learning_rate": 2.978723404255319e-06, "loss": 1.0576, "step": 2310 }, { "epoch": 0.7024316109422493, "grad_norm": 0.27734375, "learning_rate": 2.9756838905775074e-06, "loss": 1.0621, "step": 2311 }, { "epoch": 0.7027355623100304, "grad_norm": 0.283203125, "learning_rate": 2.9726443768996965e-06, "loss": 1.0613, "step": 2312 }, { "epoch": 0.7030395136778116, "grad_norm": 0.28125, "learning_rate": 2.969604863221885e-06, "loss": 1.0713, "step": 2313 }, { "epoch": 0.7033434650455928, "grad_norm": 0.279296875, "learning_rate": 2.966565349544073e-06, "loss": 1.0541, "step": 2314 }, { "epoch": 0.7036474164133738, "grad_norm": 0.28515625, "learning_rate": 2.9635258358662618e-06, "loss": 1.1047, "step": 2315 }, { "epoch": 0.703951367781155, "grad_norm": 0.29296875, "learning_rate": 2.96048632218845e-06, "loss": 1.0335, "step": 2316 }, { "epoch": 0.7042553191489361, "grad_norm": 0.384765625, "learning_rate": 2.9574468085106383e-06, "loss": 1.0992, "step": 2317 }, { "epoch": 0.7045592705167173, "grad_norm": 0.28125, "learning_rate": 2.9544072948328265e-06, "loss": 1.0209, "step": 2318 }, { "epoch": 0.7048632218844985, "grad_norm": 0.29296875, "learning_rate": 2.9513677811550156e-06, "loss": 1.1687, "step": 2319 }, { "epoch": 0.7051671732522796, "grad_norm": 0.279296875, "learning_rate": 2.948328267477204e-06, "loss": 1.1649, "step": 2320 }, { "epoch": 0.7054711246200608, "grad_norm": 0.31640625, "learning_rate": 2.9452887537993926e-06, "loss": 1.0086, "step": 2321 }, { "epoch": 0.705775075987842, "grad_norm": 0.294921875, "learning_rate": 2.942249240121581e-06, "loss": 1.0323, "step": 2322 }, { "epoch": 0.7060790273556231, "grad_norm": 0.28125, "learning_rate": 2.939209726443769e-06, "loss": 1.0119, "step": 2323 }, { "epoch": 0.7063829787234043, "grad_norm": 0.27734375, "learning_rate": 2.9361702127659574e-06, "loss": 1.0324, "step": 2324 }, { "epoch": 0.7066869300911854, "grad_norm": 0.30078125, "learning_rate": 2.9331306990881465e-06, "loss": 1.0387, "step": 2325 }, { "epoch": 0.7069908814589666, "grad_norm": 0.302734375, "learning_rate": 2.9300911854103347e-06, "loss": 0.9853, "step": 2326 }, { "epoch": 0.7072948328267478, "grad_norm": 0.302734375, "learning_rate": 2.927051671732523e-06, "loss": 1.0517, "step": 2327 }, { "epoch": 0.7075987841945289, "grad_norm": 0.296875, "learning_rate": 2.9240121580547117e-06, "loss": 1.0439, "step": 2328 }, { "epoch": 0.70790273556231, "grad_norm": 0.298828125, "learning_rate": 2.9209726443769e-06, "loss": 1.0654, "step": 2329 }, { "epoch": 0.7082066869300911, "grad_norm": 0.2734375, "learning_rate": 2.9179331306990882e-06, "loss": 1.1552, "step": 2330 }, { "epoch": 0.7085106382978723, "grad_norm": 0.279296875, "learning_rate": 2.9148936170212765e-06, "loss": 1.04, "step": 2331 }, { "epoch": 0.7088145896656535, "grad_norm": 0.29296875, "learning_rate": 2.9118541033434656e-06, "loss": 1.0429, "step": 2332 }, { "epoch": 0.7091185410334346, "grad_norm": 0.275390625, "learning_rate": 2.908814589665654e-06, "loss": 1.0511, "step": 2333 }, { "epoch": 0.7094224924012158, "grad_norm": 0.2890625, "learning_rate": 2.905775075987842e-06, "loss": 1.05, "step": 2334 }, { "epoch": 0.709726443768997, "grad_norm": 0.30859375, "learning_rate": 2.902735562310031e-06, "loss": 1.0897, "step": 2335 }, { "epoch": 0.7100303951367781, "grad_norm": 0.279296875, "learning_rate": 2.899696048632219e-06, "loss": 1.0488, "step": 2336 }, { "epoch": 0.7103343465045593, "grad_norm": 0.28125, "learning_rate": 2.8966565349544073e-06, "loss": 1.085, "step": 2337 }, { "epoch": 0.7106382978723405, "grad_norm": 0.287109375, "learning_rate": 2.8936170212765956e-06, "loss": 1.0354, "step": 2338 }, { "epoch": 0.7109422492401216, "grad_norm": 0.287109375, "learning_rate": 2.8905775075987847e-06, "loss": 1.0846, "step": 2339 }, { "epoch": 0.7112462006079028, "grad_norm": 0.283203125, "learning_rate": 2.887537993920973e-06, "loss": 1.0841, "step": 2340 }, { "epoch": 0.7115501519756839, "grad_norm": 0.2890625, "learning_rate": 2.884498480243161e-06, "loss": 0.9815, "step": 2341 }, { "epoch": 0.711854103343465, "grad_norm": 0.279296875, "learning_rate": 2.88145896656535e-06, "loss": 1.0832, "step": 2342 }, { "epoch": 0.7121580547112462, "grad_norm": 0.275390625, "learning_rate": 2.878419452887538e-06, "loss": 1.058, "step": 2343 }, { "epoch": 0.7124620060790273, "grad_norm": 0.28515625, "learning_rate": 2.8753799392097264e-06, "loss": 1.0285, "step": 2344 }, { "epoch": 0.7127659574468085, "grad_norm": 0.453125, "learning_rate": 2.8723404255319155e-06, "loss": 0.9927, "step": 2345 }, { "epoch": 0.7130699088145896, "grad_norm": 0.28125, "learning_rate": 2.8693009118541038e-06, "loss": 1.023, "step": 2346 }, { "epoch": 0.7133738601823708, "grad_norm": 0.28515625, "learning_rate": 2.866261398176292e-06, "loss": 1.0224, "step": 2347 }, { "epoch": 0.713677811550152, "grad_norm": 0.30859375, "learning_rate": 2.8632218844984803e-06, "loss": 1.0131, "step": 2348 }, { "epoch": 0.7139817629179331, "grad_norm": 0.28125, "learning_rate": 2.860182370820669e-06, "loss": 0.9541, "step": 2349 }, { "epoch": 0.7142857142857143, "grad_norm": 0.27734375, "learning_rate": 2.8571428571428573e-06, "loss": 1.0562, "step": 2350 }, { "epoch": 0.7145896656534955, "grad_norm": 0.29296875, "learning_rate": 2.8541033434650455e-06, "loss": 1.0948, "step": 2351 }, { "epoch": 0.7148936170212766, "grad_norm": 0.283203125, "learning_rate": 2.8510638297872346e-06, "loss": 1.0583, "step": 2352 }, { "epoch": 0.7151975683890578, "grad_norm": 0.2890625, "learning_rate": 2.848024316109423e-06, "loss": 0.965, "step": 2353 }, { "epoch": 0.715501519756839, "grad_norm": 0.28125, "learning_rate": 2.844984802431611e-06, "loss": 1.1166, "step": 2354 }, { "epoch": 0.71580547112462, "grad_norm": 0.287109375, "learning_rate": 2.8419452887537994e-06, "loss": 1.0097, "step": 2355 }, { "epoch": 0.7161094224924012, "grad_norm": 0.287109375, "learning_rate": 2.838905775075988e-06, "loss": 0.9844, "step": 2356 }, { "epoch": 0.7164133738601823, "grad_norm": 0.2890625, "learning_rate": 2.8358662613981764e-06, "loss": 1.0941, "step": 2357 }, { "epoch": 0.7167173252279635, "grad_norm": 0.296875, "learning_rate": 2.8328267477203646e-06, "loss": 1.0436, "step": 2358 }, { "epoch": 0.7170212765957447, "grad_norm": 0.27734375, "learning_rate": 2.8297872340425537e-06, "loss": 0.9918, "step": 2359 }, { "epoch": 0.7173252279635258, "grad_norm": 0.2890625, "learning_rate": 2.826747720364742e-06, "loss": 1.0583, "step": 2360 }, { "epoch": 0.717629179331307, "grad_norm": 0.283203125, "learning_rate": 2.8237082066869302e-06, "loss": 0.9704, "step": 2361 }, { "epoch": 0.7179331306990882, "grad_norm": 0.28125, "learning_rate": 2.8206686930091185e-06, "loss": 1.0414, "step": 2362 }, { "epoch": 0.7182370820668693, "grad_norm": 0.296875, "learning_rate": 2.817629179331307e-06, "loss": 1.135, "step": 2363 }, { "epoch": 0.7185410334346505, "grad_norm": 0.310546875, "learning_rate": 2.8145896656534955e-06, "loss": 1.022, "step": 2364 }, { "epoch": 0.7188449848024316, "grad_norm": 0.29296875, "learning_rate": 2.8115501519756846e-06, "loss": 1.0778, "step": 2365 }, { "epoch": 0.7191489361702128, "grad_norm": 0.28515625, "learning_rate": 2.808510638297873e-06, "loss": 0.9995, "step": 2366 }, { "epoch": 0.719452887537994, "grad_norm": 0.298828125, "learning_rate": 2.805471124620061e-06, "loss": 1.0118, "step": 2367 }, { "epoch": 0.7197568389057751, "grad_norm": 0.287109375, "learning_rate": 2.8024316109422493e-06, "loss": 1.0186, "step": 2368 }, { "epoch": 0.7200607902735562, "grad_norm": 0.294921875, "learning_rate": 2.799392097264438e-06, "loss": 0.9893, "step": 2369 }, { "epoch": 0.7203647416413373, "grad_norm": 0.29296875, "learning_rate": 2.7963525835866263e-06, "loss": 1.0655, "step": 2370 }, { "epoch": 0.7206686930091185, "grad_norm": 0.275390625, "learning_rate": 2.7933130699088146e-06, "loss": 1.0348, "step": 2371 }, { "epoch": 0.7209726443768997, "grad_norm": 0.29296875, "learning_rate": 2.7902735562310037e-06, "loss": 1.0931, "step": 2372 }, { "epoch": 0.7212765957446808, "grad_norm": 0.283203125, "learning_rate": 2.787234042553192e-06, "loss": 1.0095, "step": 2373 }, { "epoch": 0.721580547112462, "grad_norm": 0.28125, "learning_rate": 2.78419452887538e-06, "loss": 1.0613, "step": 2374 }, { "epoch": 0.7218844984802432, "grad_norm": 0.32421875, "learning_rate": 2.7811550151975684e-06, "loss": 1.0104, "step": 2375 }, { "epoch": 0.7221884498480243, "grad_norm": 0.291015625, "learning_rate": 2.778115501519757e-06, "loss": 1.0519, "step": 2376 }, { "epoch": 0.7224924012158055, "grad_norm": 0.29296875, "learning_rate": 2.7750759878419454e-06, "loss": 1.0025, "step": 2377 }, { "epoch": 0.7227963525835867, "grad_norm": 0.283203125, "learning_rate": 2.772036474164134e-06, "loss": 1.0121, "step": 2378 }, { "epoch": 0.7231003039513678, "grad_norm": 0.30078125, "learning_rate": 2.7689969604863228e-06, "loss": 1.0324, "step": 2379 }, { "epoch": 0.723404255319149, "grad_norm": 0.287109375, "learning_rate": 2.765957446808511e-06, "loss": 1.0728, "step": 2380 }, { "epoch": 0.7237082066869301, "grad_norm": 0.3046875, "learning_rate": 2.7629179331306993e-06, "loss": 1.0532, "step": 2381 }, { "epoch": 0.7240121580547112, "grad_norm": 0.306640625, "learning_rate": 2.7598784194528875e-06, "loss": 1.1327, "step": 2382 }, { "epoch": 0.7243161094224924, "grad_norm": 0.306640625, "learning_rate": 2.7568389057750762e-06, "loss": 1.1111, "step": 2383 }, { "epoch": 0.7246200607902735, "grad_norm": 0.291015625, "learning_rate": 2.7537993920972645e-06, "loss": 1.0797, "step": 2384 }, { "epoch": 0.7249240121580547, "grad_norm": 0.28515625, "learning_rate": 2.750759878419453e-06, "loss": 1.0794, "step": 2385 }, { "epoch": 0.7252279635258359, "grad_norm": 0.2890625, "learning_rate": 2.747720364741642e-06, "loss": 1.1235, "step": 2386 }, { "epoch": 0.725531914893617, "grad_norm": 0.30859375, "learning_rate": 2.74468085106383e-06, "loss": 1.071, "step": 2387 }, { "epoch": 0.7258358662613982, "grad_norm": 0.28515625, "learning_rate": 2.7416413373860184e-06, "loss": 1.0642, "step": 2388 }, { "epoch": 0.7261398176291793, "grad_norm": 0.283203125, "learning_rate": 2.7386018237082066e-06, "loss": 1.0675, "step": 2389 }, { "epoch": 0.7264437689969605, "grad_norm": 0.294921875, "learning_rate": 2.7355623100303953e-06, "loss": 1.1602, "step": 2390 }, { "epoch": 0.7267477203647417, "grad_norm": 0.275390625, "learning_rate": 2.7325227963525836e-06, "loss": 1.0946, "step": 2391 }, { "epoch": 0.7270516717325228, "grad_norm": 0.28515625, "learning_rate": 2.7294832826747723e-06, "loss": 1.0575, "step": 2392 }, { "epoch": 0.727355623100304, "grad_norm": 0.279296875, "learning_rate": 2.726443768996961e-06, "loss": 1.0965, "step": 2393 }, { "epoch": 0.7276595744680852, "grad_norm": 0.291015625, "learning_rate": 2.7234042553191492e-06, "loss": 0.9855, "step": 2394 }, { "epoch": 0.7279635258358662, "grad_norm": 0.28515625, "learning_rate": 2.7203647416413375e-06, "loss": 1.0422, "step": 2395 }, { "epoch": 0.7282674772036474, "grad_norm": 0.283203125, "learning_rate": 2.7173252279635258e-06, "loss": 1.0389, "step": 2396 }, { "epoch": 0.7285714285714285, "grad_norm": 0.28125, "learning_rate": 2.7142857142857144e-06, "loss": 1.0259, "step": 2397 }, { "epoch": 0.7288753799392097, "grad_norm": 0.2890625, "learning_rate": 2.711246200607903e-06, "loss": 1.1004, "step": 2398 }, { "epoch": 0.7291793313069909, "grad_norm": 0.291015625, "learning_rate": 2.7082066869300914e-06, "loss": 1.0928, "step": 2399 }, { "epoch": 0.729483282674772, "grad_norm": 0.287109375, "learning_rate": 2.70516717325228e-06, "loss": 1.1738, "step": 2400 }, { "epoch": 0.7297872340425532, "grad_norm": 0.279296875, "learning_rate": 2.7021276595744683e-06, "loss": 1.084, "step": 2401 }, { "epoch": 0.7300911854103344, "grad_norm": 0.275390625, "learning_rate": 2.6990881458966566e-06, "loss": 1.0699, "step": 2402 }, { "epoch": 0.7303951367781155, "grad_norm": 0.29296875, "learning_rate": 2.696048632218845e-06, "loss": 1.0661, "step": 2403 }, { "epoch": 0.7306990881458967, "grad_norm": 0.3125, "learning_rate": 2.6930091185410335e-06, "loss": 0.9703, "step": 2404 }, { "epoch": 0.7310030395136778, "grad_norm": 0.27734375, "learning_rate": 2.6899696048632222e-06, "loss": 1.0545, "step": 2405 }, { "epoch": 0.731306990881459, "grad_norm": 0.37890625, "learning_rate": 2.686930091185411e-06, "loss": 1.0943, "step": 2406 }, { "epoch": 0.7316109422492402, "grad_norm": 0.283203125, "learning_rate": 2.683890577507599e-06, "loss": 1.0074, "step": 2407 }, { "epoch": 0.7319148936170212, "grad_norm": 0.29296875, "learning_rate": 2.6808510638297874e-06, "loss": 1.1211, "step": 2408 }, { "epoch": 0.7322188449848024, "grad_norm": 0.287109375, "learning_rate": 2.6778115501519757e-06, "loss": 1.0556, "step": 2409 }, { "epoch": 0.7325227963525835, "grad_norm": 0.296875, "learning_rate": 2.6747720364741644e-06, "loss": 1.0649, "step": 2410 }, { "epoch": 0.7328267477203647, "grad_norm": 0.306640625, "learning_rate": 2.6717325227963526e-06, "loss": 1.0652, "step": 2411 }, { "epoch": 0.7331306990881459, "grad_norm": 0.291015625, "learning_rate": 2.6686930091185413e-06, "loss": 1.0272, "step": 2412 }, { "epoch": 0.733434650455927, "grad_norm": 0.71875, "learning_rate": 2.66565349544073e-06, "loss": 0.9574, "step": 2413 }, { "epoch": 0.7337386018237082, "grad_norm": 0.35546875, "learning_rate": 2.6626139817629183e-06, "loss": 1.0446, "step": 2414 }, { "epoch": 0.7340425531914894, "grad_norm": 0.2890625, "learning_rate": 2.6595744680851065e-06, "loss": 0.9842, "step": 2415 }, { "epoch": 0.7343465045592705, "grad_norm": 0.34375, "learning_rate": 2.656534954407295e-06, "loss": 1.0967, "step": 2416 }, { "epoch": 0.7346504559270517, "grad_norm": 0.28515625, "learning_rate": 2.6534954407294835e-06, "loss": 1.0723, "step": 2417 }, { "epoch": 0.7349544072948329, "grad_norm": 0.298828125, "learning_rate": 2.650455927051672e-06, "loss": 1.0515, "step": 2418 }, { "epoch": 0.735258358662614, "grad_norm": 0.30078125, "learning_rate": 2.6474164133738604e-06, "loss": 1.004, "step": 2419 }, { "epoch": 0.7355623100303952, "grad_norm": 0.30078125, "learning_rate": 2.644376899696049e-06, "loss": 1.0314, "step": 2420 }, { "epoch": 0.7358662613981763, "grad_norm": 0.29296875, "learning_rate": 2.6413373860182374e-06, "loss": 1.0409, "step": 2421 }, { "epoch": 0.7361702127659574, "grad_norm": 0.2734375, "learning_rate": 2.6382978723404256e-06, "loss": 1.0363, "step": 2422 }, { "epoch": 0.7364741641337386, "grad_norm": 0.287109375, "learning_rate": 2.635258358662614e-06, "loss": 1.0098, "step": 2423 }, { "epoch": 0.7367781155015197, "grad_norm": 0.279296875, "learning_rate": 2.6322188449848026e-06, "loss": 0.9894, "step": 2424 }, { "epoch": 0.7370820668693009, "grad_norm": 0.27734375, "learning_rate": 2.6291793313069913e-06, "loss": 1.0446, "step": 2425 }, { "epoch": 0.737386018237082, "grad_norm": 0.271484375, "learning_rate": 2.6261398176291795e-06, "loss": 1.0742, "step": 2426 }, { "epoch": 0.7376899696048632, "grad_norm": 0.271484375, "learning_rate": 2.623100303951368e-06, "loss": 1.0265, "step": 2427 }, { "epoch": 0.7379939209726444, "grad_norm": 0.275390625, "learning_rate": 2.6200607902735565e-06, "loss": 1.0521, "step": 2428 }, { "epoch": 0.7382978723404255, "grad_norm": 0.287109375, "learning_rate": 2.6170212765957447e-06, "loss": 0.9676, "step": 2429 }, { "epoch": 0.7386018237082067, "grad_norm": 0.314453125, "learning_rate": 2.613981762917933e-06, "loss": 1.0116, "step": 2430 }, { "epoch": 0.7389057750759879, "grad_norm": 0.298828125, "learning_rate": 2.6109422492401217e-06, "loss": 1.0236, "step": 2431 }, { "epoch": 0.739209726443769, "grad_norm": 0.30078125, "learning_rate": 2.6079027355623104e-06, "loss": 1.0892, "step": 2432 }, { "epoch": 0.7395136778115502, "grad_norm": 0.279296875, "learning_rate": 2.6048632218844986e-06, "loss": 1.0451, "step": 2433 }, { "epoch": 0.7398176291793314, "grad_norm": 0.296875, "learning_rate": 2.6018237082066873e-06, "loss": 0.971, "step": 2434 }, { "epoch": 0.7401215805471124, "grad_norm": 0.28515625, "learning_rate": 2.5987841945288756e-06, "loss": 1.0719, "step": 2435 }, { "epoch": 0.7404255319148936, "grad_norm": 0.314453125, "learning_rate": 2.595744680851064e-06, "loss": 1.0975, "step": 2436 }, { "epoch": 0.7407294832826747, "grad_norm": 0.283203125, "learning_rate": 2.592705167173252e-06, "loss": 1.0396, "step": 2437 }, { "epoch": 0.7410334346504559, "grad_norm": 0.287109375, "learning_rate": 2.589665653495441e-06, "loss": 1.0501, "step": 2438 }, { "epoch": 0.7413373860182371, "grad_norm": 0.279296875, "learning_rate": 2.5866261398176295e-06, "loss": 1.0731, "step": 2439 }, { "epoch": 0.7416413373860182, "grad_norm": 0.2890625, "learning_rate": 2.5835866261398177e-06, "loss": 1.0946, "step": 2440 }, { "epoch": 0.7419452887537994, "grad_norm": 0.28515625, "learning_rate": 2.5805471124620064e-06, "loss": 1.1333, "step": 2441 }, { "epoch": 0.7422492401215806, "grad_norm": 0.296875, "learning_rate": 2.5775075987841947e-06, "loss": 1.0367, "step": 2442 }, { "epoch": 0.7425531914893617, "grad_norm": 0.302734375, "learning_rate": 2.574468085106383e-06, "loss": 1.0262, "step": 2443 }, { "epoch": 0.7428571428571429, "grad_norm": 0.287109375, "learning_rate": 2.571428571428571e-06, "loss": 1.069, "step": 2444 }, { "epoch": 0.743161094224924, "grad_norm": 0.287109375, "learning_rate": 2.5683890577507603e-06, "loss": 1.1207, "step": 2445 }, { "epoch": 0.7434650455927052, "grad_norm": 0.29296875, "learning_rate": 2.5653495440729486e-06, "loss": 1.0693, "step": 2446 }, { "epoch": 0.7437689969604864, "grad_norm": 0.29296875, "learning_rate": 2.5623100303951372e-06, "loss": 1.0446, "step": 2447 }, { "epoch": 0.7440729483282674, "grad_norm": 0.283203125, "learning_rate": 2.5592705167173255e-06, "loss": 1.0542, "step": 2448 }, { "epoch": 0.7443768996960486, "grad_norm": 0.28515625, "learning_rate": 2.5562310030395138e-06, "loss": 1.0893, "step": 2449 }, { "epoch": 0.7446808510638298, "grad_norm": 0.283203125, "learning_rate": 2.553191489361702e-06, "loss": 1.0565, "step": 2450 }, { "epoch": 0.7449848024316109, "grad_norm": 0.30078125, "learning_rate": 2.550151975683891e-06, "loss": 1.0183, "step": 2451 }, { "epoch": 0.7452887537993921, "grad_norm": 0.275390625, "learning_rate": 2.5471124620060794e-06, "loss": 1.1015, "step": 2452 }, { "epoch": 0.7455927051671732, "grad_norm": 0.28515625, "learning_rate": 2.5440729483282677e-06, "loss": 1.0545, "step": 2453 }, { "epoch": 0.7458966565349544, "grad_norm": 0.29296875, "learning_rate": 2.5410334346504563e-06, "loss": 1.052, "step": 2454 }, { "epoch": 0.7462006079027356, "grad_norm": 0.28515625, "learning_rate": 2.5379939209726446e-06, "loss": 1.069, "step": 2455 }, { "epoch": 0.7465045592705167, "grad_norm": 0.30078125, "learning_rate": 2.534954407294833e-06, "loss": 1.035, "step": 2456 }, { "epoch": 0.7468085106382979, "grad_norm": 0.291015625, "learning_rate": 2.531914893617021e-06, "loss": 0.9992, "step": 2457 }, { "epoch": 0.7471124620060791, "grad_norm": 0.27734375, "learning_rate": 2.5288753799392102e-06, "loss": 1.0342, "step": 2458 }, { "epoch": 0.7474164133738602, "grad_norm": 0.279296875, "learning_rate": 2.5258358662613985e-06, "loss": 0.9736, "step": 2459 }, { "epoch": 0.7477203647416414, "grad_norm": 0.287109375, "learning_rate": 2.5227963525835868e-06, "loss": 1.0541, "step": 2460 }, { "epoch": 0.7480243161094224, "grad_norm": 0.28515625, "learning_rate": 2.5197568389057754e-06, "loss": 1.0676, "step": 2461 }, { "epoch": 0.7483282674772036, "grad_norm": 0.81640625, "learning_rate": 2.5167173252279637e-06, "loss": 1.0174, "step": 2462 }, { "epoch": 0.7486322188449848, "grad_norm": 0.29296875, "learning_rate": 2.513677811550152e-06, "loss": 1.031, "step": 2463 }, { "epoch": 0.7489361702127659, "grad_norm": 0.283203125, "learning_rate": 2.5106382978723402e-06, "loss": 0.9655, "step": 2464 }, { "epoch": 0.7492401215805471, "grad_norm": 0.29296875, "learning_rate": 2.5075987841945293e-06, "loss": 1.0604, "step": 2465 }, { "epoch": 0.7495440729483283, "grad_norm": 0.29296875, "learning_rate": 2.5045592705167176e-06, "loss": 1.0977, "step": 2466 }, { "epoch": 0.7498480243161094, "grad_norm": 0.279296875, "learning_rate": 2.501519756838906e-06, "loss": 1.042, "step": 2467 }, { "epoch": 0.7501519756838906, "grad_norm": 0.287109375, "learning_rate": 2.4984802431610945e-06, "loss": 1.034, "step": 2468 }, { "epoch": 0.7504559270516717, "grad_norm": 0.283203125, "learning_rate": 2.495440729483283e-06, "loss": 1.1116, "step": 2469 }, { "epoch": 0.7507598784194529, "grad_norm": 0.28515625, "learning_rate": 2.4924012158054715e-06, "loss": 1.0916, "step": 2470 }, { "epoch": 0.7510638297872341, "grad_norm": 0.28515625, "learning_rate": 2.4893617021276598e-06, "loss": 1.0169, "step": 2471 }, { "epoch": 0.7513677811550152, "grad_norm": 0.275390625, "learning_rate": 2.486322188449848e-06, "loss": 1.1135, "step": 2472 }, { "epoch": 0.7516717325227964, "grad_norm": 0.29296875, "learning_rate": 2.4832826747720367e-06, "loss": 1.0388, "step": 2473 }, { "epoch": 0.7519756838905776, "grad_norm": 0.287109375, "learning_rate": 2.480243161094225e-06, "loss": 0.9879, "step": 2474 }, { "epoch": 0.7522796352583586, "grad_norm": 0.2890625, "learning_rate": 2.4772036474164136e-06, "loss": 1.0471, "step": 2475 }, { "epoch": 0.7525835866261398, "grad_norm": 0.345703125, "learning_rate": 2.4741641337386023e-06, "loss": 1.045, "step": 2476 }, { "epoch": 0.7528875379939209, "grad_norm": 0.271484375, "learning_rate": 2.4711246200607906e-06, "loss": 1.0906, "step": 2477 }, { "epoch": 0.7531914893617021, "grad_norm": 0.27734375, "learning_rate": 2.468085106382979e-06, "loss": 1.039, "step": 2478 }, { "epoch": 0.7534954407294833, "grad_norm": 0.283203125, "learning_rate": 2.465045592705167e-06, "loss": 1.0764, "step": 2479 }, { "epoch": 0.7537993920972644, "grad_norm": 0.291015625, "learning_rate": 2.462006079027356e-06, "loss": 0.9949, "step": 2480 }, { "epoch": 0.7541033434650456, "grad_norm": 0.283203125, "learning_rate": 2.458966565349544e-06, "loss": 1.1691, "step": 2481 }, { "epoch": 0.7544072948328268, "grad_norm": 0.30078125, "learning_rate": 2.4559270516717327e-06, "loss": 1.0529, "step": 2482 }, { "epoch": 0.7547112462006079, "grad_norm": 0.26953125, "learning_rate": 2.4528875379939214e-06, "loss": 1.0898, "step": 2483 }, { "epoch": 0.7550151975683891, "grad_norm": 0.287109375, "learning_rate": 2.4498480243161097e-06, "loss": 1.0557, "step": 2484 }, { "epoch": 0.7553191489361702, "grad_norm": 0.306640625, "learning_rate": 2.446808510638298e-06, "loss": 1.0348, "step": 2485 }, { "epoch": 0.7556231003039514, "grad_norm": 0.29296875, "learning_rate": 2.4437689969604866e-06, "loss": 1.0607, "step": 2486 }, { "epoch": 0.7559270516717326, "grad_norm": 0.30078125, "learning_rate": 2.440729483282675e-06, "loss": 1.0417, "step": 2487 }, { "epoch": 0.7562310030395136, "grad_norm": 0.28125, "learning_rate": 2.437689969604863e-06, "loss": 1.0261, "step": 2488 }, { "epoch": 0.7565349544072948, "grad_norm": 0.341796875, "learning_rate": 2.434650455927052e-06, "loss": 1.1157, "step": 2489 }, { "epoch": 0.756838905775076, "grad_norm": 0.294921875, "learning_rate": 2.4316109422492405e-06, "loss": 1.0602, "step": 2490 }, { "epoch": 0.7571428571428571, "grad_norm": 0.283203125, "learning_rate": 2.428571428571429e-06, "loss": 1.0736, "step": 2491 }, { "epoch": 0.7574468085106383, "grad_norm": 0.283203125, "learning_rate": 2.425531914893617e-06, "loss": 1.1038, "step": 2492 }, { "epoch": 0.7577507598784194, "grad_norm": 0.31640625, "learning_rate": 2.4224924012158057e-06, "loss": 1.0596, "step": 2493 }, { "epoch": 0.7580547112462006, "grad_norm": 0.2734375, "learning_rate": 2.419452887537994e-06, "loss": 1.0056, "step": 2494 }, { "epoch": 0.7583586626139818, "grad_norm": 0.291015625, "learning_rate": 2.4164133738601827e-06, "loss": 1.0468, "step": 2495 }, { "epoch": 0.7586626139817629, "grad_norm": 0.28515625, "learning_rate": 2.413373860182371e-06, "loss": 1.0307, "step": 2496 }, { "epoch": 0.7589665653495441, "grad_norm": 0.28515625, "learning_rate": 2.4103343465045596e-06, "loss": 1.0674, "step": 2497 }, { "epoch": 0.7592705167173253, "grad_norm": 0.279296875, "learning_rate": 2.407294832826748e-06, "loss": 1.0608, "step": 2498 }, { "epoch": 0.7595744680851064, "grad_norm": 0.287109375, "learning_rate": 2.404255319148936e-06, "loss": 0.9742, "step": 2499 }, { "epoch": 0.7598784194528876, "grad_norm": 0.2890625, "learning_rate": 2.401215805471125e-06, "loss": 0.9737, "step": 2500 }, { "epoch": 0.7601823708206686, "grad_norm": 0.28125, "learning_rate": 2.398176291793313e-06, "loss": 1.0569, "step": 2501 }, { "epoch": 0.7604863221884498, "grad_norm": 0.28515625, "learning_rate": 2.3951367781155018e-06, "loss": 1.007, "step": 2502 }, { "epoch": 0.760790273556231, "grad_norm": 0.29296875, "learning_rate": 2.39209726443769e-06, "loss": 1.0492, "step": 2503 }, { "epoch": 0.7610942249240121, "grad_norm": 0.27734375, "learning_rate": 2.3890577507598787e-06, "loss": 1.0788, "step": 2504 }, { "epoch": 0.7613981762917933, "grad_norm": 0.29296875, "learning_rate": 2.386018237082067e-06, "loss": 1.0103, "step": 2505 }, { "epoch": 0.7617021276595745, "grad_norm": 0.28125, "learning_rate": 2.3829787234042557e-06, "loss": 1.0275, "step": 2506 }, { "epoch": 0.7620060790273556, "grad_norm": 0.2890625, "learning_rate": 2.379939209726444e-06, "loss": 1.0053, "step": 2507 }, { "epoch": 0.7623100303951368, "grad_norm": 0.275390625, "learning_rate": 2.376899696048632e-06, "loss": 1.0136, "step": 2508 }, { "epoch": 0.762613981762918, "grad_norm": 0.287109375, "learning_rate": 2.373860182370821e-06, "loss": 1.0302, "step": 2509 }, { "epoch": 0.7629179331306991, "grad_norm": 0.283203125, "learning_rate": 2.3708206686930096e-06, "loss": 1.0473, "step": 2510 }, { "epoch": 0.7632218844984803, "grad_norm": 0.279296875, "learning_rate": 2.367781155015198e-06, "loss": 1.025, "step": 2511 }, { "epoch": 0.7635258358662614, "grad_norm": 0.27734375, "learning_rate": 2.364741641337386e-06, "loss": 1.0442, "step": 2512 }, { "epoch": 0.7638297872340426, "grad_norm": 0.298828125, "learning_rate": 2.3617021276595748e-06, "loss": 1.1018, "step": 2513 }, { "epoch": 0.7641337386018237, "grad_norm": 0.28515625, "learning_rate": 2.358662613981763e-06, "loss": 1.0675, "step": 2514 }, { "epoch": 0.7644376899696048, "grad_norm": 0.287109375, "learning_rate": 2.3556231003039513e-06, "loss": 1.0346, "step": 2515 }, { "epoch": 0.764741641337386, "grad_norm": 0.296875, "learning_rate": 2.35258358662614e-06, "loss": 1.0361, "step": 2516 }, { "epoch": 0.7650455927051671, "grad_norm": 0.279296875, "learning_rate": 2.3495440729483287e-06, "loss": 1.0308, "step": 2517 }, { "epoch": 0.7653495440729483, "grad_norm": 0.28515625, "learning_rate": 2.346504559270517e-06, "loss": 1.0177, "step": 2518 }, { "epoch": 0.7656534954407295, "grad_norm": 0.28515625, "learning_rate": 2.3434650455927056e-06, "loss": 1.0816, "step": 2519 }, { "epoch": 0.7659574468085106, "grad_norm": 0.2890625, "learning_rate": 2.340425531914894e-06, "loss": 1.0358, "step": 2520 }, { "epoch": 0.7662613981762918, "grad_norm": 0.3359375, "learning_rate": 2.337386018237082e-06, "loss": 1.1002, "step": 2521 }, { "epoch": 0.766565349544073, "grad_norm": 0.2890625, "learning_rate": 2.3343465045592704e-06, "loss": 1.113, "step": 2522 }, { "epoch": 0.7668693009118541, "grad_norm": 0.279296875, "learning_rate": 2.331306990881459e-06, "loss": 1.076, "step": 2523 }, { "epoch": 0.7671732522796353, "grad_norm": 0.2890625, "learning_rate": 2.3282674772036478e-06, "loss": 1.0391, "step": 2524 }, { "epoch": 0.7674772036474165, "grad_norm": 0.283203125, "learning_rate": 2.325227963525836e-06, "loss": 1.0429, "step": 2525 }, { "epoch": 0.7677811550151976, "grad_norm": 0.279296875, "learning_rate": 2.3221884498480247e-06, "loss": 1.0599, "step": 2526 }, { "epoch": 0.7680851063829788, "grad_norm": 0.291015625, "learning_rate": 2.319148936170213e-06, "loss": 1.0719, "step": 2527 }, { "epoch": 0.7683890577507598, "grad_norm": 0.27734375, "learning_rate": 2.3161094224924012e-06, "loss": 1.0176, "step": 2528 }, { "epoch": 0.768693009118541, "grad_norm": 0.326171875, "learning_rate": 2.31306990881459e-06, "loss": 1.1069, "step": 2529 }, { "epoch": 0.7689969604863222, "grad_norm": 0.2890625, "learning_rate": 2.310030395136778e-06, "loss": 1.0274, "step": 2530 }, { "epoch": 0.7693009118541033, "grad_norm": 0.27734375, "learning_rate": 2.306990881458967e-06, "loss": 1.0786, "step": 2531 }, { "epoch": 0.7696048632218845, "grad_norm": 0.294921875, "learning_rate": 2.303951367781155e-06, "loss": 1.0732, "step": 2532 }, { "epoch": 0.7699088145896656, "grad_norm": 0.275390625, "learning_rate": 2.300911854103344e-06, "loss": 1.0962, "step": 2533 }, { "epoch": 0.7702127659574468, "grad_norm": 0.326171875, "learning_rate": 2.297872340425532e-06, "loss": 1.024, "step": 2534 }, { "epoch": 0.770516717325228, "grad_norm": 0.283203125, "learning_rate": 2.2948328267477203e-06, "loss": 1.0522, "step": 2535 }, { "epoch": 0.7708206686930091, "grad_norm": 0.287109375, "learning_rate": 2.291793313069909e-06, "loss": 1.1418, "step": 2536 }, { "epoch": 0.7711246200607903, "grad_norm": 0.287109375, "learning_rate": 2.2887537993920973e-06, "loss": 0.9621, "step": 2537 }, { "epoch": 0.7714285714285715, "grad_norm": 0.28125, "learning_rate": 2.285714285714286e-06, "loss": 1.0961, "step": 2538 }, { "epoch": 0.7717325227963526, "grad_norm": 0.28515625, "learning_rate": 2.2826747720364747e-06, "loss": 1.0857, "step": 2539 }, { "epoch": 0.7720364741641338, "grad_norm": 0.333984375, "learning_rate": 2.279635258358663e-06, "loss": 1.0232, "step": 2540 }, { "epoch": 0.7723404255319148, "grad_norm": 0.298828125, "learning_rate": 2.276595744680851e-06, "loss": 1.0981, "step": 2541 }, { "epoch": 0.772644376899696, "grad_norm": 0.28125, "learning_rate": 2.2735562310030394e-06, "loss": 1.0752, "step": 2542 }, { "epoch": 0.7729483282674772, "grad_norm": 0.279296875, "learning_rate": 2.270516717325228e-06, "loss": 1.0665, "step": 2543 }, { "epoch": 0.7732522796352583, "grad_norm": 0.296875, "learning_rate": 2.2674772036474164e-06, "loss": 0.976, "step": 2544 }, { "epoch": 0.7735562310030395, "grad_norm": 0.283203125, "learning_rate": 2.264437689969605e-06, "loss": 1.1011, "step": 2545 }, { "epoch": 0.7738601823708207, "grad_norm": 0.2890625, "learning_rate": 2.2613981762917938e-06, "loss": 1.0673, "step": 2546 }, { "epoch": 0.7741641337386018, "grad_norm": 0.298828125, "learning_rate": 2.258358662613982e-06, "loss": 0.9776, "step": 2547 }, { "epoch": 0.774468085106383, "grad_norm": 0.310546875, "learning_rate": 2.2553191489361703e-06, "loss": 1.0863, "step": 2548 }, { "epoch": 0.7747720364741641, "grad_norm": 0.27734375, "learning_rate": 2.252279635258359e-06, "loss": 1.0441, "step": 2549 }, { "epoch": 0.7750759878419453, "grad_norm": 0.28515625, "learning_rate": 2.2492401215805472e-06, "loss": 1.0138, "step": 2550 }, { "epoch": 0.7753799392097265, "grad_norm": 0.28515625, "learning_rate": 2.2462006079027355e-06, "loss": 1.006, "step": 2551 }, { "epoch": 0.7756838905775076, "grad_norm": 0.28515625, "learning_rate": 2.243161094224924e-06, "loss": 1.0027, "step": 2552 }, { "epoch": 0.7759878419452888, "grad_norm": 0.287109375, "learning_rate": 2.240121580547113e-06, "loss": 1.0528, "step": 2553 }, { "epoch": 0.7762917933130699, "grad_norm": 0.29296875, "learning_rate": 2.237082066869301e-06, "loss": 1.0609, "step": 2554 }, { "epoch": 0.776595744680851, "grad_norm": 0.291015625, "learning_rate": 2.2340425531914894e-06, "loss": 1.0233, "step": 2555 }, { "epoch": 0.7768996960486322, "grad_norm": 0.296875, "learning_rate": 2.231003039513678e-06, "loss": 1.091, "step": 2556 }, { "epoch": 0.7772036474164133, "grad_norm": 0.30859375, "learning_rate": 2.2279635258358663e-06, "loss": 1.077, "step": 2557 }, { "epoch": 0.7775075987841945, "grad_norm": 0.287109375, "learning_rate": 2.224924012158055e-06, "loss": 1.1207, "step": 2558 }, { "epoch": 0.7778115501519757, "grad_norm": 0.279296875, "learning_rate": 2.2218844984802433e-06, "loss": 1.1064, "step": 2559 }, { "epoch": 0.7781155015197568, "grad_norm": 0.294921875, "learning_rate": 2.218844984802432e-06, "loss": 1.0571, "step": 2560 }, { "epoch": 0.778419452887538, "grad_norm": 0.27734375, "learning_rate": 2.2158054711246202e-06, "loss": 1.0327, "step": 2561 }, { "epoch": 0.7787234042553192, "grad_norm": 0.287109375, "learning_rate": 2.2127659574468085e-06, "loss": 1.0108, "step": 2562 }, { "epoch": 0.7790273556231003, "grad_norm": 0.3046875, "learning_rate": 2.209726443768997e-06, "loss": 1.0496, "step": 2563 }, { "epoch": 0.7793313069908815, "grad_norm": 0.283203125, "learning_rate": 2.2066869300911854e-06, "loss": 1.0436, "step": 2564 }, { "epoch": 0.7796352583586627, "grad_norm": 0.28125, "learning_rate": 2.203647416413374e-06, "loss": 1.0087, "step": 2565 }, { "epoch": 0.7799392097264438, "grad_norm": 0.28515625, "learning_rate": 2.2006079027355624e-06, "loss": 1.0988, "step": 2566 }, { "epoch": 0.7802431610942249, "grad_norm": 0.283203125, "learning_rate": 2.197568389057751e-06, "loss": 1.0161, "step": 2567 }, { "epoch": 0.780547112462006, "grad_norm": 0.28125, "learning_rate": 2.1945288753799393e-06, "loss": 1.1044, "step": 2568 }, { "epoch": 0.7808510638297872, "grad_norm": 0.28515625, "learning_rate": 2.191489361702128e-06, "loss": 1.0562, "step": 2569 }, { "epoch": 0.7811550151975684, "grad_norm": 0.373046875, "learning_rate": 2.1884498480243163e-06, "loss": 1.0325, "step": 2570 }, { "epoch": 0.7814589665653495, "grad_norm": 0.2890625, "learning_rate": 2.1854103343465045e-06, "loss": 0.9705, "step": 2571 }, { "epoch": 0.7817629179331307, "grad_norm": 0.294921875, "learning_rate": 2.1823708206686932e-06, "loss": 1.0594, "step": 2572 }, { "epoch": 0.7820668693009118, "grad_norm": 0.28515625, "learning_rate": 2.179331306990882e-06, "loss": 1.0424, "step": 2573 }, { "epoch": 0.782370820668693, "grad_norm": 0.296875, "learning_rate": 2.17629179331307e-06, "loss": 1.0354, "step": 2574 }, { "epoch": 0.7826747720364742, "grad_norm": 0.28515625, "learning_rate": 2.1732522796352584e-06, "loss": 1.0959, "step": 2575 }, { "epoch": 0.7829787234042553, "grad_norm": 0.287109375, "learning_rate": 2.170212765957447e-06, "loss": 1.0887, "step": 2576 }, { "epoch": 0.7832826747720365, "grad_norm": 0.279296875, "learning_rate": 2.1671732522796354e-06, "loss": 1.123, "step": 2577 }, { "epoch": 0.7835866261398177, "grad_norm": 0.30078125, "learning_rate": 2.1641337386018236e-06, "loss": 0.9757, "step": 2578 }, { "epoch": 0.7838905775075988, "grad_norm": 0.279296875, "learning_rate": 2.1610942249240123e-06, "loss": 0.9984, "step": 2579 }, { "epoch": 0.78419452887538, "grad_norm": 0.28515625, "learning_rate": 2.158054711246201e-06, "loss": 1.0128, "step": 2580 }, { "epoch": 0.784498480243161, "grad_norm": 0.294921875, "learning_rate": 2.1550151975683893e-06, "loss": 1.0102, "step": 2581 }, { "epoch": 0.7848024316109422, "grad_norm": 0.28515625, "learning_rate": 2.151975683890578e-06, "loss": 0.9486, "step": 2582 }, { "epoch": 0.7851063829787234, "grad_norm": 0.27734375, "learning_rate": 2.148936170212766e-06, "loss": 1.1278, "step": 2583 }, { "epoch": 0.7854103343465045, "grad_norm": 0.279296875, "learning_rate": 2.1458966565349545e-06, "loss": 1.1001, "step": 2584 }, { "epoch": 0.7857142857142857, "grad_norm": 0.287109375, "learning_rate": 2.1428571428571427e-06, "loss": 1.0455, "step": 2585 }, { "epoch": 0.7860182370820669, "grad_norm": 0.279296875, "learning_rate": 2.1398176291793314e-06, "loss": 1.1274, "step": 2586 }, { "epoch": 0.786322188449848, "grad_norm": 0.30078125, "learning_rate": 2.13677811550152e-06, "loss": 0.9584, "step": 2587 }, { "epoch": 0.7866261398176292, "grad_norm": 0.294921875, "learning_rate": 2.1337386018237084e-06, "loss": 1.0537, "step": 2588 }, { "epoch": 0.7869300911854104, "grad_norm": 0.291015625, "learning_rate": 2.130699088145897e-06, "loss": 1.0509, "step": 2589 }, { "epoch": 0.7872340425531915, "grad_norm": 0.294921875, "learning_rate": 2.1276595744680853e-06, "loss": 1.0801, "step": 2590 }, { "epoch": 0.7875379939209727, "grad_norm": 0.283203125, "learning_rate": 2.1246200607902736e-06, "loss": 1.0051, "step": 2591 }, { "epoch": 0.7878419452887538, "grad_norm": 0.296875, "learning_rate": 2.1215805471124623e-06, "loss": 1.0677, "step": 2592 }, { "epoch": 0.788145896656535, "grad_norm": 0.27734375, "learning_rate": 2.1185410334346505e-06, "loss": 1.0735, "step": 2593 }, { "epoch": 0.7884498480243161, "grad_norm": 0.29296875, "learning_rate": 2.115501519756839e-06, "loss": 1.024, "step": 2594 }, { "epoch": 0.7887537993920972, "grad_norm": 0.302734375, "learning_rate": 2.1124620060790275e-06, "loss": 0.9411, "step": 2595 }, { "epoch": 0.7890577507598784, "grad_norm": 0.291015625, "learning_rate": 2.109422492401216e-06, "loss": 0.9807, "step": 2596 }, { "epoch": 0.7893617021276595, "grad_norm": 0.28515625, "learning_rate": 2.1063829787234044e-06, "loss": 1.0334, "step": 2597 }, { "epoch": 0.7896656534954407, "grad_norm": 0.27734375, "learning_rate": 2.1033434650455927e-06, "loss": 1.073, "step": 2598 }, { "epoch": 0.7899696048632219, "grad_norm": 0.291015625, "learning_rate": 2.1003039513677814e-06, "loss": 1.037, "step": 2599 }, { "epoch": 0.790273556231003, "grad_norm": 0.287109375, "learning_rate": 2.0972644376899696e-06, "loss": 1.1117, "step": 2600 }, { "epoch": 0.7905775075987842, "grad_norm": 0.2890625, "learning_rate": 2.0942249240121583e-06, "loss": 0.9867, "step": 2601 }, { "epoch": 0.7908814589665654, "grad_norm": 0.294921875, "learning_rate": 2.091185410334347e-06, "loss": 1.0379, "step": 2602 }, { "epoch": 0.7911854103343465, "grad_norm": 0.287109375, "learning_rate": 2.0881458966565352e-06, "loss": 1.0318, "step": 2603 }, { "epoch": 0.7914893617021277, "grad_norm": 0.28125, "learning_rate": 2.0851063829787235e-06, "loss": 0.9932, "step": 2604 }, { "epoch": 0.7917933130699089, "grad_norm": 0.291015625, "learning_rate": 2.0820668693009118e-06, "loss": 1.0565, "step": 2605 }, { "epoch": 0.79209726443769, "grad_norm": 0.291015625, "learning_rate": 2.0790273556231005e-06, "loss": 1.0242, "step": 2606 }, { "epoch": 0.7924012158054711, "grad_norm": 0.28515625, "learning_rate": 2.0759878419452887e-06, "loss": 1.0918, "step": 2607 }, { "epoch": 0.7927051671732522, "grad_norm": 0.291015625, "learning_rate": 2.0729483282674774e-06, "loss": 1.0714, "step": 2608 }, { "epoch": 0.7930091185410334, "grad_norm": 0.28515625, "learning_rate": 2.069908814589666e-06, "loss": 1.0639, "step": 2609 }, { "epoch": 0.7933130699088146, "grad_norm": 0.279296875, "learning_rate": 2.0668693009118543e-06, "loss": 1.1242, "step": 2610 }, { "epoch": 0.7936170212765957, "grad_norm": 0.302734375, "learning_rate": 2.0638297872340426e-06, "loss": 1.1591, "step": 2611 }, { "epoch": 0.7939209726443769, "grad_norm": 0.27734375, "learning_rate": 2.0607902735562313e-06, "loss": 1.1019, "step": 2612 }, { "epoch": 0.794224924012158, "grad_norm": 0.294921875, "learning_rate": 2.0577507598784196e-06, "loss": 1.0764, "step": 2613 }, { "epoch": 0.7945288753799392, "grad_norm": 0.330078125, "learning_rate": 2.054711246200608e-06, "loss": 1.0403, "step": 2614 }, { "epoch": 0.7948328267477204, "grad_norm": 0.296875, "learning_rate": 2.0516717325227965e-06, "loss": 1.006, "step": 2615 }, { "epoch": 0.7951367781155015, "grad_norm": 0.30078125, "learning_rate": 2.048632218844985e-06, "loss": 1.036, "step": 2616 }, { "epoch": 0.7954407294832827, "grad_norm": 0.279296875, "learning_rate": 2.0455927051671734e-06, "loss": 1.0601, "step": 2617 }, { "epoch": 0.7957446808510639, "grad_norm": 0.2890625, "learning_rate": 2.0425531914893617e-06, "loss": 1.0081, "step": 2618 }, { "epoch": 0.796048632218845, "grad_norm": 0.29296875, "learning_rate": 2.0395136778115504e-06, "loss": 0.9724, "step": 2619 }, { "epoch": 0.7963525835866262, "grad_norm": 0.28515625, "learning_rate": 2.0364741641337387e-06, "loss": 1.0087, "step": 2620 }, { "epoch": 0.7966565349544072, "grad_norm": 0.291015625, "learning_rate": 2.0334346504559273e-06, "loss": 1.1071, "step": 2621 }, { "epoch": 0.7969604863221884, "grad_norm": 0.29296875, "learning_rate": 2.0303951367781156e-06, "loss": 1.0887, "step": 2622 }, { "epoch": 0.7972644376899696, "grad_norm": 0.287109375, "learning_rate": 2.0273556231003043e-06, "loss": 1.0586, "step": 2623 }, { "epoch": 0.7975683890577507, "grad_norm": 0.28515625, "learning_rate": 2.0243161094224925e-06, "loss": 1.0375, "step": 2624 }, { "epoch": 0.7978723404255319, "grad_norm": 0.283203125, "learning_rate": 2.021276595744681e-06, "loss": 0.9968, "step": 2625 }, { "epoch": 0.7981762917933131, "grad_norm": 0.291015625, "learning_rate": 2.0182370820668695e-06, "loss": 1.0127, "step": 2626 }, { "epoch": 0.7984802431610942, "grad_norm": 0.296875, "learning_rate": 2.0151975683890578e-06, "loss": 0.9585, "step": 2627 }, { "epoch": 0.7987841945288754, "grad_norm": 0.3046875, "learning_rate": 2.0121580547112464e-06, "loss": 1.0426, "step": 2628 }, { "epoch": 0.7990881458966566, "grad_norm": 0.287109375, "learning_rate": 2.0091185410334347e-06, "loss": 1.06, "step": 2629 }, { "epoch": 0.7993920972644377, "grad_norm": 0.333984375, "learning_rate": 2.0060790273556234e-06, "loss": 1.0988, "step": 2630 }, { "epoch": 0.7996960486322189, "grad_norm": 0.28515625, "learning_rate": 2.0030395136778116e-06, "loss": 1.0125, "step": 2631 }, { "epoch": 0.8, "grad_norm": 0.287109375, "learning_rate": 2.0000000000000003e-06, "loss": 1.0538, "step": 2632 }, { "epoch": 0.8003039513677812, "grad_norm": 0.296875, "learning_rate": 1.9969604863221886e-06, "loss": 1.0884, "step": 2633 }, { "epoch": 0.8006079027355623, "grad_norm": 0.2890625, "learning_rate": 1.993920972644377e-06, "loss": 1.0028, "step": 2634 }, { "epoch": 0.8009118541033434, "grad_norm": 0.328125, "learning_rate": 1.9908814589665655e-06, "loss": 0.9978, "step": 2635 }, { "epoch": 0.8012158054711246, "grad_norm": 0.279296875, "learning_rate": 1.9878419452887542e-06, "loss": 1.1268, "step": 2636 }, { "epoch": 0.8015197568389058, "grad_norm": 0.28515625, "learning_rate": 1.9848024316109425e-06, "loss": 1.0584, "step": 2637 }, { "epoch": 0.8018237082066869, "grad_norm": 0.287109375, "learning_rate": 1.9817629179331307e-06, "loss": 1.0578, "step": 2638 }, { "epoch": 0.8021276595744681, "grad_norm": 0.283203125, "learning_rate": 1.9787234042553194e-06, "loss": 1.0073, "step": 2639 }, { "epoch": 0.8024316109422492, "grad_norm": 0.28125, "learning_rate": 1.9756838905775077e-06, "loss": 1.0528, "step": 2640 }, { "epoch": 0.8027355623100304, "grad_norm": 0.287109375, "learning_rate": 1.972644376899696e-06, "loss": 1.0224, "step": 2641 }, { "epoch": 0.8030395136778116, "grad_norm": 0.287109375, "learning_rate": 1.9696048632218846e-06, "loss": 1.0867, "step": 2642 }, { "epoch": 0.8033434650455927, "grad_norm": 0.3046875, "learning_rate": 1.9665653495440733e-06, "loss": 1.0699, "step": 2643 }, { "epoch": 0.8036474164133739, "grad_norm": 0.275390625, "learning_rate": 1.9635258358662616e-06, "loss": 1.0341, "step": 2644 }, { "epoch": 0.8039513677811551, "grad_norm": 0.287109375, "learning_rate": 1.9604863221884503e-06, "loss": 1.0722, "step": 2645 }, { "epoch": 0.8042553191489362, "grad_norm": 0.28515625, "learning_rate": 1.9574468085106385e-06, "loss": 1.0397, "step": 2646 }, { "epoch": 0.8045592705167173, "grad_norm": 0.279296875, "learning_rate": 1.954407294832827e-06, "loss": 1.0391, "step": 2647 }, { "epoch": 0.8048632218844984, "grad_norm": 0.318359375, "learning_rate": 1.951367781155015e-06, "loss": 1.1227, "step": 2648 }, { "epoch": 0.8051671732522796, "grad_norm": 0.28515625, "learning_rate": 1.9483282674772037e-06, "loss": 1.0347, "step": 2649 }, { "epoch": 0.8054711246200608, "grad_norm": 0.30078125, "learning_rate": 1.9452887537993924e-06, "loss": 1.1227, "step": 2650 }, { "epoch": 0.8057750759878419, "grad_norm": 0.27734375, "learning_rate": 1.9422492401215807e-06, "loss": 0.9895, "step": 2651 }, { "epoch": 0.8060790273556231, "grad_norm": 0.27734375, "learning_rate": 1.9392097264437694e-06, "loss": 1.1211, "step": 2652 }, { "epoch": 0.8063829787234043, "grad_norm": 0.27734375, "learning_rate": 1.9361702127659576e-06, "loss": 0.975, "step": 2653 }, { "epoch": 0.8066869300911854, "grad_norm": 0.28515625, "learning_rate": 1.933130699088146e-06, "loss": 1.0452, "step": 2654 }, { "epoch": 0.8069908814589666, "grad_norm": 0.294921875, "learning_rate": 1.9300911854103346e-06, "loss": 1.0235, "step": 2655 }, { "epoch": 0.8072948328267477, "grad_norm": 0.287109375, "learning_rate": 1.927051671732523e-06, "loss": 0.9938, "step": 2656 }, { "epoch": 0.8075987841945289, "grad_norm": 0.28125, "learning_rate": 1.9240121580547115e-06, "loss": 1.0753, "step": 2657 }, { "epoch": 0.8079027355623101, "grad_norm": 0.28515625, "learning_rate": 1.9209726443768998e-06, "loss": 1.0153, "step": 2658 }, { "epoch": 0.8082066869300912, "grad_norm": 0.291015625, "learning_rate": 1.9179331306990885e-06, "loss": 1.0427, "step": 2659 }, { "epoch": 0.8085106382978723, "grad_norm": 0.375, "learning_rate": 1.9148936170212767e-06, "loss": 1.0495, "step": 2660 }, { "epoch": 0.8088145896656534, "grad_norm": 0.294921875, "learning_rate": 1.911854103343465e-06, "loss": 1.068, "step": 2661 }, { "epoch": 0.8091185410334346, "grad_norm": 0.294921875, "learning_rate": 1.9088145896656537e-06, "loss": 0.9932, "step": 2662 }, { "epoch": 0.8094224924012158, "grad_norm": 0.294921875, "learning_rate": 1.9057750759878422e-06, "loss": 1.0173, "step": 2663 }, { "epoch": 0.8097264437689969, "grad_norm": 0.275390625, "learning_rate": 1.9027355623100304e-06, "loss": 1.0907, "step": 2664 }, { "epoch": 0.8100303951367781, "grad_norm": 0.294921875, "learning_rate": 1.899696048632219e-06, "loss": 1.0388, "step": 2665 }, { "epoch": 0.8103343465045593, "grad_norm": 0.29296875, "learning_rate": 1.8966565349544076e-06, "loss": 1.0212, "step": 2666 }, { "epoch": 0.8106382978723404, "grad_norm": 0.27734375, "learning_rate": 1.8936170212765958e-06, "loss": 1.1039, "step": 2667 }, { "epoch": 0.8109422492401216, "grad_norm": 0.28515625, "learning_rate": 1.8905775075987843e-06, "loss": 1.0712, "step": 2668 }, { "epoch": 0.8112462006079028, "grad_norm": 0.275390625, "learning_rate": 1.8875379939209728e-06, "loss": 1.0631, "step": 2669 }, { "epoch": 0.8115501519756839, "grad_norm": 0.287109375, "learning_rate": 1.8844984802431613e-06, "loss": 1.1011, "step": 2670 }, { "epoch": 0.8118541033434651, "grad_norm": 0.2890625, "learning_rate": 1.8814589665653495e-06, "loss": 1.0609, "step": 2671 }, { "epoch": 0.8121580547112462, "grad_norm": 0.29296875, "learning_rate": 1.8784194528875382e-06, "loss": 1.1205, "step": 2672 }, { "epoch": 0.8124620060790274, "grad_norm": 0.294921875, "learning_rate": 1.8753799392097267e-06, "loss": 1.1097, "step": 2673 }, { "epoch": 0.8127659574468085, "grad_norm": 0.294921875, "learning_rate": 1.872340425531915e-06, "loss": 0.9742, "step": 2674 }, { "epoch": 0.8130699088145896, "grad_norm": 0.2890625, "learning_rate": 1.8693009118541036e-06, "loss": 1.0444, "step": 2675 }, { "epoch": 0.8133738601823708, "grad_norm": 0.29296875, "learning_rate": 1.8662613981762919e-06, "loss": 1.0834, "step": 2676 }, { "epoch": 0.813677811550152, "grad_norm": 0.29296875, "learning_rate": 1.8632218844984804e-06, "loss": 1.0745, "step": 2677 }, { "epoch": 0.8139817629179331, "grad_norm": 0.546875, "learning_rate": 1.8601823708206686e-06, "loss": 0.9781, "step": 2678 }, { "epoch": 0.8142857142857143, "grad_norm": 0.28125, "learning_rate": 1.8571428571428573e-06, "loss": 1.0415, "step": 2679 }, { "epoch": 0.8145896656534954, "grad_norm": 0.28515625, "learning_rate": 1.8541033434650458e-06, "loss": 1.0328, "step": 2680 }, { "epoch": 0.8148936170212766, "grad_norm": 0.34765625, "learning_rate": 1.851063829787234e-06, "loss": 1.0473, "step": 2681 }, { "epoch": 0.8151975683890578, "grad_norm": 0.291015625, "learning_rate": 1.8480243161094227e-06, "loss": 1.0289, "step": 2682 }, { "epoch": 0.8155015197568389, "grad_norm": 0.294921875, "learning_rate": 1.8449848024316112e-06, "loss": 1.0943, "step": 2683 }, { "epoch": 0.8158054711246201, "grad_norm": 0.291015625, "learning_rate": 1.8419452887537995e-06, "loss": 1.0764, "step": 2684 }, { "epoch": 0.8161094224924013, "grad_norm": 0.291015625, "learning_rate": 1.8389057750759881e-06, "loss": 1.0654, "step": 2685 }, { "epoch": 0.8164133738601824, "grad_norm": 0.2890625, "learning_rate": 1.8358662613981764e-06, "loss": 1.0317, "step": 2686 }, { "epoch": 0.8167173252279635, "grad_norm": 0.28125, "learning_rate": 1.8328267477203649e-06, "loss": 1.0093, "step": 2687 }, { "epoch": 0.8170212765957446, "grad_norm": 0.283203125, "learning_rate": 1.8297872340425531e-06, "loss": 1.0503, "step": 2688 }, { "epoch": 0.8173252279635258, "grad_norm": 0.298828125, "learning_rate": 1.8267477203647418e-06, "loss": 1.0049, "step": 2689 }, { "epoch": 0.817629179331307, "grad_norm": 0.28515625, "learning_rate": 1.8237082066869303e-06, "loss": 1.2082, "step": 2690 }, { "epoch": 0.8179331306990881, "grad_norm": 0.28125, "learning_rate": 1.8206686930091186e-06, "loss": 1.0855, "step": 2691 }, { "epoch": 0.8182370820668693, "grad_norm": 0.283203125, "learning_rate": 1.8176291793313072e-06, "loss": 1.0897, "step": 2692 }, { "epoch": 0.8185410334346505, "grad_norm": 0.2890625, "learning_rate": 1.8145896656534955e-06, "loss": 1.0765, "step": 2693 }, { "epoch": 0.8188449848024316, "grad_norm": 0.28515625, "learning_rate": 1.811550151975684e-06, "loss": 1.0163, "step": 2694 }, { "epoch": 0.8191489361702128, "grad_norm": 0.291015625, "learning_rate": 1.8085106382978727e-06, "loss": 1.0478, "step": 2695 }, { "epoch": 0.819452887537994, "grad_norm": 0.283203125, "learning_rate": 1.805471124620061e-06, "loss": 1.0504, "step": 2696 }, { "epoch": 0.8197568389057751, "grad_norm": 0.291015625, "learning_rate": 1.8024316109422494e-06, "loss": 1.009, "step": 2697 }, { "epoch": 0.8200607902735563, "grad_norm": 0.279296875, "learning_rate": 1.799392097264438e-06, "loss": 1.0416, "step": 2698 }, { "epoch": 0.8203647416413374, "grad_norm": 0.451171875, "learning_rate": 1.7963525835866263e-06, "loss": 0.9992, "step": 2699 }, { "epoch": 0.8206686930091185, "grad_norm": 0.32421875, "learning_rate": 1.7933130699088146e-06, "loss": 1.0981, "step": 2700 }, { "epoch": 0.8209726443768997, "grad_norm": 0.296875, "learning_rate": 1.790273556231003e-06, "loss": 0.9941, "step": 2701 }, { "epoch": 0.8212765957446808, "grad_norm": 0.298828125, "learning_rate": 1.7872340425531918e-06, "loss": 1.0596, "step": 2702 }, { "epoch": 0.821580547112462, "grad_norm": 0.287109375, "learning_rate": 1.78419452887538e-06, "loss": 1.0395, "step": 2703 }, { "epoch": 0.8218844984802431, "grad_norm": 0.28125, "learning_rate": 1.7811550151975685e-06, "loss": 1.037, "step": 2704 }, { "epoch": 0.8221884498480243, "grad_norm": 0.287109375, "learning_rate": 1.7781155015197572e-06, "loss": 1.0289, "step": 2705 }, { "epoch": 0.8224924012158055, "grad_norm": 0.279296875, "learning_rate": 1.7750759878419454e-06, "loss": 1.0458, "step": 2706 }, { "epoch": 0.8227963525835866, "grad_norm": 0.298828125, "learning_rate": 1.772036474164134e-06, "loss": 1.0522, "step": 2707 }, { "epoch": 0.8231003039513678, "grad_norm": 0.32421875, "learning_rate": 1.7689969604863224e-06, "loss": 1.0863, "step": 2708 }, { "epoch": 0.823404255319149, "grad_norm": 0.279296875, "learning_rate": 1.7659574468085109e-06, "loss": 1.0578, "step": 2709 }, { "epoch": 0.8237082066869301, "grad_norm": 0.302734375, "learning_rate": 1.7629179331306991e-06, "loss": 0.9945, "step": 2710 }, { "epoch": 0.8240121580547113, "grad_norm": 0.3046875, "learning_rate": 1.7598784194528876e-06, "loss": 1.0561, "step": 2711 }, { "epoch": 0.8243161094224924, "grad_norm": 0.3125, "learning_rate": 1.7568389057750763e-06, "loss": 1.1019, "step": 2712 }, { "epoch": 0.8246200607902735, "grad_norm": 0.28125, "learning_rate": 1.7537993920972645e-06, "loss": 0.973, "step": 2713 }, { "epoch": 0.8249240121580547, "grad_norm": 0.291015625, "learning_rate": 1.750759878419453e-06, "loss": 0.9344, "step": 2714 }, { "epoch": 0.8252279635258358, "grad_norm": 0.298828125, "learning_rate": 1.7477203647416415e-06, "loss": 1.0239, "step": 2715 }, { "epoch": 0.825531914893617, "grad_norm": 0.279296875, "learning_rate": 1.74468085106383e-06, "loss": 1.0655, "step": 2716 }, { "epoch": 0.8258358662613982, "grad_norm": 0.29296875, "learning_rate": 1.7416413373860182e-06, "loss": 1.1666, "step": 2717 }, { "epoch": 0.8261398176291793, "grad_norm": 0.287109375, "learning_rate": 1.738601823708207e-06, "loss": 1.1058, "step": 2718 }, { "epoch": 0.8264437689969605, "grad_norm": 0.2890625, "learning_rate": 1.7355623100303954e-06, "loss": 1.0068, "step": 2719 }, { "epoch": 0.8267477203647416, "grad_norm": 0.298828125, "learning_rate": 1.7325227963525836e-06, "loss": 1.0231, "step": 2720 }, { "epoch": 0.8270516717325228, "grad_norm": 0.29296875, "learning_rate": 1.7294832826747721e-06, "loss": 1.0956, "step": 2721 }, { "epoch": 0.827355623100304, "grad_norm": 0.306640625, "learning_rate": 1.7264437689969608e-06, "loss": 1.0131, "step": 2722 }, { "epoch": 0.8276595744680851, "grad_norm": 0.306640625, "learning_rate": 1.723404255319149e-06, "loss": 1.0079, "step": 2723 }, { "epoch": 0.8279635258358663, "grad_norm": 0.294921875, "learning_rate": 1.7203647416413375e-06, "loss": 0.9916, "step": 2724 }, { "epoch": 0.8282674772036475, "grad_norm": 0.279296875, "learning_rate": 1.717325227963526e-06, "loss": 1.111, "step": 2725 }, { "epoch": 0.8285714285714286, "grad_norm": 0.3046875, "learning_rate": 1.7142857142857145e-06, "loss": 1.1101, "step": 2726 }, { "epoch": 0.8288753799392097, "grad_norm": 0.283203125, "learning_rate": 1.7112462006079027e-06, "loss": 1.0516, "step": 2727 }, { "epoch": 0.8291793313069908, "grad_norm": 0.294921875, "learning_rate": 1.7082066869300914e-06, "loss": 1.0543, "step": 2728 }, { "epoch": 0.829483282674772, "grad_norm": 0.30859375, "learning_rate": 1.70516717325228e-06, "loss": 1.0397, "step": 2729 }, { "epoch": 0.8297872340425532, "grad_norm": 0.3125, "learning_rate": 1.7021276595744682e-06, "loss": 1.1156, "step": 2730 }, { "epoch": 0.8300911854103343, "grad_norm": 0.2890625, "learning_rate": 1.6990881458966566e-06, "loss": 1.0476, "step": 2731 }, { "epoch": 0.8303951367781155, "grad_norm": 0.2890625, "learning_rate": 1.6960486322188451e-06, "loss": 1.0143, "step": 2732 }, { "epoch": 0.8306990881458967, "grad_norm": 0.279296875, "learning_rate": 1.6930091185410336e-06, "loss": 0.9428, "step": 2733 }, { "epoch": 0.8310030395136778, "grad_norm": 0.28515625, "learning_rate": 1.6899696048632218e-06, "loss": 1.0363, "step": 2734 }, { "epoch": 0.831306990881459, "grad_norm": 0.2890625, "learning_rate": 1.6869300911854105e-06, "loss": 1.0877, "step": 2735 }, { "epoch": 0.8316109422492401, "grad_norm": 0.28125, "learning_rate": 1.683890577507599e-06, "loss": 1.0557, "step": 2736 }, { "epoch": 0.8319148936170213, "grad_norm": 0.365234375, "learning_rate": 1.6808510638297873e-06, "loss": 1.0364, "step": 2737 }, { "epoch": 0.8322188449848025, "grad_norm": 0.3046875, "learning_rate": 1.677811550151976e-06, "loss": 1.049, "step": 2738 }, { "epoch": 0.8325227963525836, "grad_norm": 0.283203125, "learning_rate": 1.6747720364741642e-06, "loss": 1.0329, "step": 2739 }, { "epoch": 0.8328267477203647, "grad_norm": 0.291015625, "learning_rate": 1.6717325227963527e-06, "loss": 1.1064, "step": 2740 }, { "epoch": 0.8331306990881459, "grad_norm": 0.279296875, "learning_rate": 1.668693009118541e-06, "loss": 1.0454, "step": 2741 }, { "epoch": 0.833434650455927, "grad_norm": 0.28515625, "learning_rate": 1.6656534954407296e-06, "loss": 0.9906, "step": 2742 }, { "epoch": 0.8337386018237082, "grad_norm": 0.26953125, "learning_rate": 1.662613981762918e-06, "loss": 1.0928, "step": 2743 }, { "epoch": 0.8340425531914893, "grad_norm": 0.28515625, "learning_rate": 1.6595744680851064e-06, "loss": 1.0905, "step": 2744 }, { "epoch": 0.8343465045592705, "grad_norm": 0.287109375, "learning_rate": 1.656534954407295e-06, "loss": 1.0381, "step": 2745 }, { "epoch": 0.8346504559270517, "grad_norm": 0.29296875, "learning_rate": 1.6534954407294835e-06, "loss": 0.9922, "step": 2746 }, { "epoch": 0.8349544072948328, "grad_norm": 0.28125, "learning_rate": 1.6504559270516718e-06, "loss": 1.0368, "step": 2747 }, { "epoch": 0.835258358662614, "grad_norm": 0.296875, "learning_rate": 1.6474164133738605e-06, "loss": 1.0294, "step": 2748 }, { "epoch": 0.8355623100303952, "grad_norm": 0.279296875, "learning_rate": 1.6443768996960487e-06, "loss": 1.0419, "step": 2749 }, { "epoch": 0.8358662613981763, "grad_norm": 0.28125, "learning_rate": 1.6413373860182372e-06, "loss": 1.1328, "step": 2750 }, { "epoch": 0.8361702127659575, "grad_norm": 0.287109375, "learning_rate": 1.6382978723404255e-06, "loss": 1.0285, "step": 2751 }, { "epoch": 0.8364741641337387, "grad_norm": 0.27734375, "learning_rate": 1.6352583586626141e-06, "loss": 1.067, "step": 2752 }, { "epoch": 0.8367781155015197, "grad_norm": 0.294921875, "learning_rate": 1.6322188449848026e-06, "loss": 1.0877, "step": 2753 }, { "epoch": 0.8370820668693009, "grad_norm": 0.2890625, "learning_rate": 1.6291793313069909e-06, "loss": 0.9899, "step": 2754 }, { "epoch": 0.837386018237082, "grad_norm": 0.296875, "learning_rate": 1.6261398176291796e-06, "loss": 1.0946, "step": 2755 }, { "epoch": 0.8376899696048632, "grad_norm": 0.28515625, "learning_rate": 1.6231003039513678e-06, "loss": 0.9967, "step": 2756 }, { "epoch": 0.8379939209726444, "grad_norm": 0.294921875, "learning_rate": 1.6200607902735563e-06, "loss": 1.0541, "step": 2757 }, { "epoch": 0.8382978723404255, "grad_norm": 0.2890625, "learning_rate": 1.617021276595745e-06, "loss": 1.0353, "step": 2758 }, { "epoch": 0.8386018237082067, "grad_norm": 0.326171875, "learning_rate": 1.6139817629179332e-06, "loss": 0.9789, "step": 2759 }, { "epoch": 0.8389057750759878, "grad_norm": 0.294921875, "learning_rate": 1.6109422492401217e-06, "loss": 0.9588, "step": 2760 }, { "epoch": 0.839209726443769, "grad_norm": 0.29296875, "learning_rate": 1.60790273556231e-06, "loss": 1.1202, "step": 2761 }, { "epoch": 0.8395136778115502, "grad_norm": 0.28515625, "learning_rate": 1.6048632218844987e-06, "loss": 1.0673, "step": 2762 }, { "epoch": 0.8398176291793313, "grad_norm": 0.279296875, "learning_rate": 1.601823708206687e-06, "loss": 1.0571, "step": 2763 }, { "epoch": 0.8401215805471125, "grad_norm": 0.330078125, "learning_rate": 1.5987841945288754e-06, "loss": 1.0181, "step": 2764 }, { "epoch": 0.8404255319148937, "grad_norm": 0.30078125, "learning_rate": 1.595744680851064e-06, "loss": 1.0213, "step": 2765 }, { "epoch": 0.8407294832826747, "grad_norm": 0.294921875, "learning_rate": 1.5927051671732523e-06, "loss": 1.0323, "step": 2766 }, { "epoch": 0.8410334346504559, "grad_norm": 0.361328125, "learning_rate": 1.5896656534954408e-06, "loss": 1.1495, "step": 2767 }, { "epoch": 0.841337386018237, "grad_norm": 0.28125, "learning_rate": 1.5866261398176295e-06, "loss": 1.1034, "step": 2768 }, { "epoch": 0.8416413373860182, "grad_norm": 0.29296875, "learning_rate": 1.5835866261398178e-06, "loss": 1.0909, "step": 2769 }, { "epoch": 0.8419452887537994, "grad_norm": 0.279296875, "learning_rate": 1.5805471124620062e-06, "loss": 1.0444, "step": 2770 }, { "epoch": 0.8422492401215805, "grad_norm": 0.2890625, "learning_rate": 1.5775075987841947e-06, "loss": 1.1065, "step": 2771 }, { "epoch": 0.8425531914893617, "grad_norm": 0.283203125, "learning_rate": 1.5744680851063832e-06, "loss": 1.0625, "step": 2772 }, { "epoch": 0.8428571428571429, "grad_norm": 0.2890625, "learning_rate": 1.5714285714285714e-06, "loss": 1.096, "step": 2773 }, { "epoch": 0.843161094224924, "grad_norm": 0.310546875, "learning_rate": 1.56838905775076e-06, "loss": 1.0467, "step": 2774 }, { "epoch": 0.8434650455927052, "grad_norm": 0.291015625, "learning_rate": 1.5653495440729486e-06, "loss": 1.0021, "step": 2775 }, { "epoch": 0.8437689969604864, "grad_norm": 0.34765625, "learning_rate": 1.5623100303951369e-06, "loss": 1.0795, "step": 2776 }, { "epoch": 0.8440729483282675, "grad_norm": 0.291015625, "learning_rate": 1.5592705167173253e-06, "loss": 1.0418, "step": 2777 }, { "epoch": 0.8443768996960487, "grad_norm": 0.28515625, "learning_rate": 1.5562310030395138e-06, "loss": 1.0276, "step": 2778 }, { "epoch": 0.8446808510638298, "grad_norm": 0.275390625, "learning_rate": 1.5531914893617023e-06, "loss": 1.0967, "step": 2779 }, { "epoch": 0.8449848024316109, "grad_norm": 0.28125, "learning_rate": 1.5501519756838905e-06, "loss": 1.0018, "step": 2780 }, { "epoch": 0.8452887537993921, "grad_norm": 0.279296875, "learning_rate": 1.5471124620060792e-06, "loss": 1.02, "step": 2781 }, { "epoch": 0.8455927051671732, "grad_norm": 0.28125, "learning_rate": 1.5440729483282677e-06, "loss": 1.0578, "step": 2782 }, { "epoch": 0.8458966565349544, "grad_norm": 0.28125, "learning_rate": 1.541033434650456e-06, "loss": 0.9228, "step": 2783 }, { "epoch": 0.8462006079027355, "grad_norm": 0.27734375, "learning_rate": 1.5379939209726444e-06, "loss": 1.1032, "step": 2784 }, { "epoch": 0.8465045592705167, "grad_norm": 0.29296875, "learning_rate": 1.5349544072948331e-06, "loss": 1.0417, "step": 2785 }, { "epoch": 0.8468085106382979, "grad_norm": 0.302734375, "learning_rate": 1.5319148936170214e-06, "loss": 1.0623, "step": 2786 }, { "epoch": 0.847112462006079, "grad_norm": 0.310546875, "learning_rate": 1.5288753799392099e-06, "loss": 1.019, "step": 2787 }, { "epoch": 0.8474164133738602, "grad_norm": 0.291015625, "learning_rate": 1.5258358662613983e-06, "loss": 1.0648, "step": 2788 }, { "epoch": 0.8477203647416414, "grad_norm": 0.298828125, "learning_rate": 1.5227963525835868e-06, "loss": 1.0502, "step": 2789 }, { "epoch": 0.8480243161094225, "grad_norm": 0.294921875, "learning_rate": 1.519756838905775e-06, "loss": 1.0473, "step": 2790 }, { "epoch": 0.8483282674772037, "grad_norm": 0.330078125, "learning_rate": 1.5167173252279638e-06, "loss": 1.0625, "step": 2791 }, { "epoch": 0.8486322188449849, "grad_norm": 0.28515625, "learning_rate": 1.5136778115501522e-06, "loss": 1.0639, "step": 2792 }, { "epoch": 0.8489361702127659, "grad_norm": 0.287109375, "learning_rate": 1.5106382978723405e-06, "loss": 1.12, "step": 2793 }, { "epoch": 0.8492401215805471, "grad_norm": 0.296875, "learning_rate": 1.507598784194529e-06, "loss": 1.1004, "step": 2794 }, { "epoch": 0.8495440729483282, "grad_norm": 0.28515625, "learning_rate": 1.5045592705167174e-06, "loss": 1.0367, "step": 2795 }, { "epoch": 0.8498480243161094, "grad_norm": 0.51953125, "learning_rate": 1.501519756838906e-06, "loss": 1.0284, "step": 2796 }, { "epoch": 0.8501519756838906, "grad_norm": 0.29296875, "learning_rate": 1.4984802431610942e-06, "loss": 0.9621, "step": 2797 }, { "epoch": 0.8504559270516717, "grad_norm": 0.287109375, "learning_rate": 1.4954407294832829e-06, "loss": 1.0189, "step": 2798 }, { "epoch": 0.8507598784194529, "grad_norm": 0.294921875, "learning_rate": 1.4924012158054713e-06, "loss": 1.0889, "step": 2799 }, { "epoch": 0.851063829787234, "grad_norm": 0.294921875, "learning_rate": 1.4893617021276596e-06, "loss": 1.0307, "step": 2800 }, { "epoch": 0.8513677811550152, "grad_norm": 0.3046875, "learning_rate": 1.4863221884498483e-06, "loss": 1.0598, "step": 2801 }, { "epoch": 0.8516717325227964, "grad_norm": 0.298828125, "learning_rate": 1.4832826747720365e-06, "loss": 1.0635, "step": 2802 }, { "epoch": 0.8519756838905775, "grad_norm": 0.283203125, "learning_rate": 1.480243161094225e-06, "loss": 1.0639, "step": 2803 }, { "epoch": 0.8522796352583587, "grad_norm": 0.283203125, "learning_rate": 1.4772036474164133e-06, "loss": 0.9633, "step": 2804 }, { "epoch": 0.8525835866261399, "grad_norm": 0.396484375, "learning_rate": 1.474164133738602e-06, "loss": 1.0092, "step": 2805 }, { "epoch": 0.8528875379939209, "grad_norm": 0.3125, "learning_rate": 1.4711246200607904e-06, "loss": 1.0336, "step": 2806 }, { "epoch": 0.8531914893617021, "grad_norm": 0.28125, "learning_rate": 1.4680851063829787e-06, "loss": 0.9777, "step": 2807 }, { "epoch": 0.8534954407294832, "grad_norm": 0.279296875, "learning_rate": 1.4650455927051674e-06, "loss": 1.076, "step": 2808 }, { "epoch": 0.8537993920972644, "grad_norm": 0.287109375, "learning_rate": 1.4620060790273558e-06, "loss": 1.0668, "step": 2809 }, { "epoch": 0.8541033434650456, "grad_norm": 0.29296875, "learning_rate": 1.4589665653495441e-06, "loss": 1.002, "step": 2810 }, { "epoch": 0.8544072948328267, "grad_norm": 0.287109375, "learning_rate": 1.4559270516717328e-06, "loss": 1.0474, "step": 2811 }, { "epoch": 0.8547112462006079, "grad_norm": 0.33203125, "learning_rate": 1.452887537993921e-06, "loss": 1.0206, "step": 2812 }, { "epoch": 0.8550151975683891, "grad_norm": 0.291015625, "learning_rate": 1.4498480243161095e-06, "loss": 1.0446, "step": 2813 }, { "epoch": 0.8553191489361702, "grad_norm": 0.2890625, "learning_rate": 1.4468085106382978e-06, "loss": 1.1023, "step": 2814 }, { "epoch": 0.8556231003039514, "grad_norm": 0.33203125, "learning_rate": 1.4437689969604865e-06, "loss": 1.0728, "step": 2815 }, { "epoch": 0.8559270516717326, "grad_norm": 0.296875, "learning_rate": 1.440729483282675e-06, "loss": 1.0646, "step": 2816 }, { "epoch": 0.8562310030395137, "grad_norm": 0.28125, "learning_rate": 1.4376899696048632e-06, "loss": 1.045, "step": 2817 }, { "epoch": 0.8565349544072949, "grad_norm": 0.302734375, "learning_rate": 1.4346504559270519e-06, "loss": 1.0105, "step": 2818 }, { "epoch": 0.856838905775076, "grad_norm": 0.291015625, "learning_rate": 1.4316109422492402e-06, "loss": 1.0139, "step": 2819 }, { "epoch": 0.8571428571428571, "grad_norm": 0.283203125, "learning_rate": 1.4285714285714286e-06, "loss": 1.1326, "step": 2820 }, { "epoch": 0.8574468085106383, "grad_norm": 0.2890625, "learning_rate": 1.4255319148936173e-06, "loss": 1.0275, "step": 2821 }, { "epoch": 0.8577507598784194, "grad_norm": 0.28515625, "learning_rate": 1.4224924012158056e-06, "loss": 0.9966, "step": 2822 }, { "epoch": 0.8580547112462006, "grad_norm": 0.34765625, "learning_rate": 1.419452887537994e-06, "loss": 1.017, "step": 2823 }, { "epoch": 0.8583586626139817, "grad_norm": 0.28515625, "learning_rate": 1.4164133738601823e-06, "loss": 1.0319, "step": 2824 }, { "epoch": 0.8586626139817629, "grad_norm": 0.283203125, "learning_rate": 1.413373860182371e-06, "loss": 0.9815, "step": 2825 }, { "epoch": 0.8589665653495441, "grad_norm": 0.28515625, "learning_rate": 1.4103343465045593e-06, "loss": 1.1023, "step": 2826 }, { "epoch": 0.8592705167173252, "grad_norm": 0.291015625, "learning_rate": 1.4072948328267477e-06, "loss": 1.0737, "step": 2827 }, { "epoch": 0.8595744680851064, "grad_norm": 0.298828125, "learning_rate": 1.4042553191489364e-06, "loss": 1.1596, "step": 2828 }, { "epoch": 0.8598784194528876, "grad_norm": 0.310546875, "learning_rate": 1.4012158054711247e-06, "loss": 1.0096, "step": 2829 }, { "epoch": 0.8601823708206687, "grad_norm": 0.298828125, "learning_rate": 1.3981762917933131e-06, "loss": 1.0472, "step": 2830 }, { "epoch": 0.8604863221884499, "grad_norm": 0.287109375, "learning_rate": 1.3951367781155018e-06, "loss": 1.0287, "step": 2831 }, { "epoch": 0.8607902735562311, "grad_norm": 0.333984375, "learning_rate": 1.39209726443769e-06, "loss": 1.0227, "step": 2832 }, { "epoch": 0.8610942249240121, "grad_norm": 0.3046875, "learning_rate": 1.3890577507598786e-06, "loss": 1.0689, "step": 2833 }, { "epoch": 0.8613981762917933, "grad_norm": 0.28125, "learning_rate": 1.386018237082067e-06, "loss": 1.0716, "step": 2834 }, { "epoch": 0.8617021276595744, "grad_norm": 0.28125, "learning_rate": 1.3829787234042555e-06, "loss": 1.1071, "step": 2835 }, { "epoch": 0.8620060790273556, "grad_norm": 0.2890625, "learning_rate": 1.3799392097264438e-06, "loss": 1.0745, "step": 2836 }, { "epoch": 0.8623100303951368, "grad_norm": 0.330078125, "learning_rate": 1.3768996960486322e-06, "loss": 1.0358, "step": 2837 }, { "epoch": 0.8626139817629179, "grad_norm": 0.2890625, "learning_rate": 1.373860182370821e-06, "loss": 1.0749, "step": 2838 }, { "epoch": 0.8629179331306991, "grad_norm": 0.30859375, "learning_rate": 1.3708206686930092e-06, "loss": 1.1142, "step": 2839 }, { "epoch": 0.8632218844984803, "grad_norm": 0.2890625, "learning_rate": 1.3677811550151977e-06, "loss": 1.1008, "step": 2840 }, { "epoch": 0.8635258358662614, "grad_norm": 0.283203125, "learning_rate": 1.3647416413373861e-06, "loss": 1.1111, "step": 2841 }, { "epoch": 0.8638297872340426, "grad_norm": 0.2890625, "learning_rate": 1.3617021276595746e-06, "loss": 1.0569, "step": 2842 }, { "epoch": 0.8641337386018237, "grad_norm": 0.302734375, "learning_rate": 1.3586626139817629e-06, "loss": 1.0465, "step": 2843 }, { "epoch": 0.8644376899696049, "grad_norm": 0.291015625, "learning_rate": 1.3556231003039516e-06, "loss": 1.0801, "step": 2844 }, { "epoch": 0.8647416413373861, "grad_norm": 0.28515625, "learning_rate": 1.35258358662614e-06, "loss": 1.068, "step": 2845 }, { "epoch": 0.8650455927051671, "grad_norm": 0.2890625, "learning_rate": 1.3495440729483283e-06, "loss": 0.9933, "step": 2846 }, { "epoch": 0.8653495440729483, "grad_norm": 0.283203125, "learning_rate": 1.3465045592705168e-06, "loss": 1.0781, "step": 2847 }, { "epoch": 0.8656534954407294, "grad_norm": 0.287109375, "learning_rate": 1.3434650455927055e-06, "loss": 1.0273, "step": 2848 }, { "epoch": 0.8659574468085106, "grad_norm": 0.291015625, "learning_rate": 1.3404255319148937e-06, "loss": 1.0365, "step": 2849 }, { "epoch": 0.8662613981762918, "grad_norm": 0.28515625, "learning_rate": 1.3373860182370822e-06, "loss": 1.0362, "step": 2850 }, { "epoch": 0.8665653495440729, "grad_norm": 0.2890625, "learning_rate": 1.3343465045592707e-06, "loss": 1.1233, "step": 2851 }, { "epoch": 0.8668693009118541, "grad_norm": 0.287109375, "learning_rate": 1.3313069908814591e-06, "loss": 1.0145, "step": 2852 }, { "epoch": 0.8671732522796353, "grad_norm": 0.275390625, "learning_rate": 1.3282674772036474e-06, "loss": 1.1037, "step": 2853 }, { "epoch": 0.8674772036474164, "grad_norm": 0.3203125, "learning_rate": 1.325227963525836e-06, "loss": 0.9856, "step": 2854 }, { "epoch": 0.8677811550151976, "grad_norm": 0.310546875, "learning_rate": 1.3221884498480246e-06, "loss": 1.0245, "step": 2855 }, { "epoch": 0.8680851063829788, "grad_norm": 0.28125, "learning_rate": 1.3191489361702128e-06, "loss": 1.0973, "step": 2856 }, { "epoch": 0.8683890577507599, "grad_norm": 0.27734375, "learning_rate": 1.3161094224924013e-06, "loss": 1.0634, "step": 2857 }, { "epoch": 0.8686930091185411, "grad_norm": 0.29296875, "learning_rate": 1.3130699088145898e-06, "loss": 0.9708, "step": 2858 }, { "epoch": 0.8689969604863221, "grad_norm": 0.302734375, "learning_rate": 1.3100303951367782e-06, "loss": 1.1074, "step": 2859 }, { "epoch": 0.8693009118541033, "grad_norm": 0.2890625, "learning_rate": 1.3069908814589665e-06, "loss": 1.0357, "step": 2860 }, { "epoch": 0.8696048632218845, "grad_norm": 0.29296875, "learning_rate": 1.3039513677811552e-06, "loss": 1.0333, "step": 2861 }, { "epoch": 0.8699088145896656, "grad_norm": 0.28125, "learning_rate": 1.3009118541033437e-06, "loss": 1.1555, "step": 2862 }, { "epoch": 0.8702127659574468, "grad_norm": 0.2890625, "learning_rate": 1.297872340425532e-06, "loss": 1.0112, "step": 2863 }, { "epoch": 0.870516717325228, "grad_norm": 0.28125, "learning_rate": 1.2948328267477206e-06, "loss": 1.0822, "step": 2864 }, { "epoch": 0.8708206686930091, "grad_norm": 0.287109375, "learning_rate": 1.2917933130699089e-06, "loss": 1.0395, "step": 2865 }, { "epoch": 0.8711246200607903, "grad_norm": 0.291015625, "learning_rate": 1.2887537993920973e-06, "loss": 1.1131, "step": 2866 }, { "epoch": 0.8714285714285714, "grad_norm": 0.30078125, "learning_rate": 1.2857142857142856e-06, "loss": 0.999, "step": 2867 }, { "epoch": 0.8717325227963526, "grad_norm": 0.3203125, "learning_rate": 1.2826747720364743e-06, "loss": 0.9677, "step": 2868 }, { "epoch": 0.8720364741641338, "grad_norm": 0.283203125, "learning_rate": 1.2796352583586628e-06, "loss": 1.066, "step": 2869 }, { "epoch": 0.8723404255319149, "grad_norm": 0.28515625, "learning_rate": 1.276595744680851e-06, "loss": 1.1098, "step": 2870 }, { "epoch": 0.8726443768996961, "grad_norm": 0.29296875, "learning_rate": 1.2735562310030397e-06, "loss": 1.0883, "step": 2871 }, { "epoch": 0.8729483282674773, "grad_norm": 0.294921875, "learning_rate": 1.2705167173252282e-06, "loss": 1.0817, "step": 2872 }, { "epoch": 0.8732522796352583, "grad_norm": 0.298828125, "learning_rate": 1.2674772036474164e-06, "loss": 1.0875, "step": 2873 }, { "epoch": 0.8735562310030395, "grad_norm": 0.298828125, "learning_rate": 1.2644376899696051e-06, "loss": 1.0599, "step": 2874 }, { "epoch": 0.8738601823708206, "grad_norm": 0.287109375, "learning_rate": 1.2613981762917934e-06, "loss": 1.0253, "step": 2875 }, { "epoch": 0.8741641337386018, "grad_norm": 0.27734375, "learning_rate": 1.2583586626139819e-06, "loss": 1.0387, "step": 2876 }, { "epoch": 0.874468085106383, "grad_norm": 0.3125, "learning_rate": 1.2553191489361701e-06, "loss": 1.0589, "step": 2877 }, { "epoch": 0.8747720364741641, "grad_norm": 0.466796875, "learning_rate": 1.2522796352583588e-06, "loss": 1.0608, "step": 2878 }, { "epoch": 0.8750759878419453, "grad_norm": 0.28515625, "learning_rate": 1.2492401215805473e-06, "loss": 1.0363, "step": 2879 }, { "epoch": 0.8753799392097265, "grad_norm": 0.310546875, "learning_rate": 1.2462006079027357e-06, "loss": 1.0684, "step": 2880 }, { "epoch": 0.8756838905775076, "grad_norm": 0.279296875, "learning_rate": 1.243161094224924e-06, "loss": 0.9959, "step": 2881 }, { "epoch": 0.8759878419452888, "grad_norm": 0.279296875, "learning_rate": 1.2401215805471125e-06, "loss": 1.0723, "step": 2882 }, { "epoch": 0.8762917933130699, "grad_norm": 0.2890625, "learning_rate": 1.2370820668693012e-06, "loss": 1.0382, "step": 2883 }, { "epoch": 0.8765957446808511, "grad_norm": 0.33203125, "learning_rate": 1.2340425531914894e-06, "loss": 1.0172, "step": 2884 }, { "epoch": 0.8768996960486323, "grad_norm": 0.287109375, "learning_rate": 1.231003039513678e-06, "loss": 1.0446, "step": 2885 }, { "epoch": 0.8772036474164133, "grad_norm": 0.29296875, "learning_rate": 1.2279635258358664e-06, "loss": 1.0374, "step": 2886 }, { "epoch": 0.8775075987841945, "grad_norm": 0.29296875, "learning_rate": 1.2249240121580548e-06, "loss": 1.0521, "step": 2887 }, { "epoch": 0.8778115501519757, "grad_norm": 0.28515625, "learning_rate": 1.2218844984802433e-06, "loss": 1.0136, "step": 2888 }, { "epoch": 0.8781155015197568, "grad_norm": 0.28125, "learning_rate": 1.2188449848024316e-06, "loss": 1.1052, "step": 2889 }, { "epoch": 0.878419452887538, "grad_norm": 0.283203125, "learning_rate": 1.2158054711246203e-06, "loss": 1.0882, "step": 2890 }, { "epoch": 0.8787234042553191, "grad_norm": 0.294921875, "learning_rate": 1.2127659574468085e-06, "loss": 1.0312, "step": 2891 }, { "epoch": 0.8790273556231003, "grad_norm": 0.2890625, "learning_rate": 1.209726443768997e-06, "loss": 1.0301, "step": 2892 }, { "epoch": 0.8793313069908815, "grad_norm": 0.28125, "learning_rate": 1.2066869300911855e-06, "loss": 1.1105, "step": 2893 }, { "epoch": 0.8796352583586626, "grad_norm": 0.287109375, "learning_rate": 1.203647416413374e-06, "loss": 1.0049, "step": 2894 }, { "epoch": 0.8799392097264438, "grad_norm": 0.275390625, "learning_rate": 1.2006079027355624e-06, "loss": 1.1387, "step": 2895 }, { "epoch": 0.880243161094225, "grad_norm": 0.3359375, "learning_rate": 1.1975683890577509e-06, "loss": 1.0336, "step": 2896 }, { "epoch": 0.8805471124620061, "grad_norm": 0.283203125, "learning_rate": 1.1945288753799394e-06, "loss": 1.0646, "step": 2897 }, { "epoch": 0.8808510638297873, "grad_norm": 0.357421875, "learning_rate": 1.1914893617021278e-06, "loss": 1.0448, "step": 2898 }, { "epoch": 0.8811550151975683, "grad_norm": 0.28125, "learning_rate": 1.188449848024316e-06, "loss": 1.0488, "step": 2899 }, { "epoch": 0.8814589665653495, "grad_norm": 0.279296875, "learning_rate": 1.1854103343465048e-06, "loss": 1.0941, "step": 2900 }, { "epoch": 0.8817629179331307, "grad_norm": 0.275390625, "learning_rate": 1.182370820668693e-06, "loss": 1.0799, "step": 2901 }, { "epoch": 0.8820668693009118, "grad_norm": 0.28125, "learning_rate": 1.1793313069908815e-06, "loss": 1.1336, "step": 2902 }, { "epoch": 0.882370820668693, "grad_norm": 0.283203125, "learning_rate": 1.17629179331307e-06, "loss": 0.9381, "step": 2903 }, { "epoch": 0.8826747720364742, "grad_norm": 0.2890625, "learning_rate": 1.1732522796352585e-06, "loss": 1.0037, "step": 2904 }, { "epoch": 0.8829787234042553, "grad_norm": 0.291015625, "learning_rate": 1.170212765957447e-06, "loss": 1.003, "step": 2905 }, { "epoch": 0.8832826747720365, "grad_norm": 0.28515625, "learning_rate": 1.1671732522796352e-06, "loss": 1.0014, "step": 2906 }, { "epoch": 0.8835866261398176, "grad_norm": 0.3125, "learning_rate": 1.1641337386018239e-06, "loss": 1.0237, "step": 2907 }, { "epoch": 0.8838905775075988, "grad_norm": 0.2890625, "learning_rate": 1.1610942249240124e-06, "loss": 1.0607, "step": 2908 }, { "epoch": 0.88419452887538, "grad_norm": 0.291015625, "learning_rate": 1.1580547112462006e-06, "loss": 1.0098, "step": 2909 }, { "epoch": 0.8844984802431611, "grad_norm": 0.28515625, "learning_rate": 1.155015197568389e-06, "loss": 1.107, "step": 2910 }, { "epoch": 0.8848024316109423, "grad_norm": 0.279296875, "learning_rate": 1.1519756838905776e-06, "loss": 1.0223, "step": 2911 }, { "epoch": 0.8851063829787233, "grad_norm": 0.287109375, "learning_rate": 1.148936170212766e-06, "loss": 1.0744, "step": 2912 }, { "epoch": 0.8854103343465045, "grad_norm": 0.306640625, "learning_rate": 1.1458966565349545e-06, "loss": 0.983, "step": 2913 }, { "epoch": 0.8857142857142857, "grad_norm": 0.279296875, "learning_rate": 1.142857142857143e-06, "loss": 1.0194, "step": 2914 }, { "epoch": 0.8860182370820668, "grad_norm": 0.28125, "learning_rate": 1.1398176291793315e-06, "loss": 1.0608, "step": 2915 }, { "epoch": 0.886322188449848, "grad_norm": 0.28515625, "learning_rate": 1.1367781155015197e-06, "loss": 1.0837, "step": 2916 }, { "epoch": 0.8866261398176292, "grad_norm": 0.28125, "learning_rate": 1.1337386018237082e-06, "loss": 1.1445, "step": 2917 }, { "epoch": 0.8869300911854103, "grad_norm": 0.28125, "learning_rate": 1.1306990881458969e-06, "loss": 1.0681, "step": 2918 }, { "epoch": 0.8872340425531915, "grad_norm": 0.30078125, "learning_rate": 1.1276595744680851e-06, "loss": 1.0192, "step": 2919 }, { "epoch": 0.8875379939209727, "grad_norm": 0.2890625, "learning_rate": 1.1246200607902736e-06, "loss": 1.0093, "step": 2920 }, { "epoch": 0.8878419452887538, "grad_norm": 0.279296875, "learning_rate": 1.121580547112462e-06, "loss": 1.0859, "step": 2921 }, { "epoch": 0.888145896656535, "grad_norm": 0.283203125, "learning_rate": 1.1185410334346506e-06, "loss": 1.048, "step": 2922 }, { "epoch": 0.8884498480243161, "grad_norm": 0.28125, "learning_rate": 1.115501519756839e-06, "loss": 1.029, "step": 2923 }, { "epoch": 0.8887537993920973, "grad_norm": 0.31640625, "learning_rate": 1.1124620060790275e-06, "loss": 1.0835, "step": 2924 }, { "epoch": 0.8890577507598785, "grad_norm": 0.306640625, "learning_rate": 1.109422492401216e-06, "loss": 0.9932, "step": 2925 }, { "epoch": 0.8893617021276595, "grad_norm": 0.287109375, "learning_rate": 1.1063829787234042e-06, "loss": 1.0818, "step": 2926 }, { "epoch": 0.8896656534954407, "grad_norm": 0.28125, "learning_rate": 1.1033434650455927e-06, "loss": 1.0838, "step": 2927 }, { "epoch": 0.8899696048632219, "grad_norm": 0.2890625, "learning_rate": 1.1003039513677812e-06, "loss": 1.1134, "step": 2928 }, { "epoch": 0.890273556231003, "grad_norm": 0.2890625, "learning_rate": 1.0972644376899697e-06, "loss": 1.0361, "step": 2929 }, { "epoch": 0.8905775075987842, "grad_norm": 0.296875, "learning_rate": 1.0942249240121581e-06, "loss": 1.0895, "step": 2930 }, { "epoch": 0.8908814589665653, "grad_norm": 0.291015625, "learning_rate": 1.0911854103343466e-06, "loss": 0.9386, "step": 2931 }, { "epoch": 0.8911854103343465, "grad_norm": 0.29296875, "learning_rate": 1.088145896656535e-06, "loss": 1.0773, "step": 2932 }, { "epoch": 0.8914893617021277, "grad_norm": 0.296875, "learning_rate": 1.0851063829787236e-06, "loss": 1.0264, "step": 2933 }, { "epoch": 0.8917933130699088, "grad_norm": 0.27734375, "learning_rate": 1.0820668693009118e-06, "loss": 0.9727, "step": 2934 }, { "epoch": 0.89209726443769, "grad_norm": 0.283203125, "learning_rate": 1.0790273556231005e-06, "loss": 1.0762, "step": 2935 }, { "epoch": 0.8924012158054712, "grad_norm": 0.28125, "learning_rate": 1.075987841945289e-06, "loss": 1.0226, "step": 2936 }, { "epoch": 0.8927051671732523, "grad_norm": 0.291015625, "learning_rate": 1.0729483282674772e-06, "loss": 1.0219, "step": 2937 }, { "epoch": 0.8930091185410335, "grad_norm": 0.28125, "learning_rate": 1.0699088145896657e-06, "loss": 1.0791, "step": 2938 }, { "epoch": 0.8933130699088145, "grad_norm": 0.291015625, "learning_rate": 1.0668693009118542e-06, "loss": 1.0023, "step": 2939 }, { "epoch": 0.8936170212765957, "grad_norm": 0.296875, "learning_rate": 1.0638297872340427e-06, "loss": 1.0125, "step": 2940 }, { "epoch": 0.8939209726443769, "grad_norm": 0.27734375, "learning_rate": 1.0607902735562311e-06, "loss": 1.0472, "step": 2941 }, { "epoch": 0.894224924012158, "grad_norm": 0.296875, "learning_rate": 1.0577507598784196e-06, "loss": 0.9857, "step": 2942 }, { "epoch": 0.8945288753799392, "grad_norm": 0.3125, "learning_rate": 1.054711246200608e-06, "loss": 1.0487, "step": 2943 }, { "epoch": 0.8948328267477204, "grad_norm": 0.294921875, "learning_rate": 1.0516717325227963e-06, "loss": 1.0403, "step": 2944 }, { "epoch": 0.8951367781155015, "grad_norm": 0.291015625, "learning_rate": 1.0486322188449848e-06, "loss": 1.0502, "step": 2945 }, { "epoch": 0.8954407294832827, "grad_norm": 0.28515625, "learning_rate": 1.0455927051671735e-06, "loss": 1.0228, "step": 2946 }, { "epoch": 0.8957446808510638, "grad_norm": 0.298828125, "learning_rate": 1.0425531914893618e-06, "loss": 0.9912, "step": 2947 }, { "epoch": 0.896048632218845, "grad_norm": 0.28125, "learning_rate": 1.0395136778115502e-06, "loss": 1.0524, "step": 2948 }, { "epoch": 0.8963525835866262, "grad_norm": 0.287109375, "learning_rate": 1.0364741641337387e-06, "loss": 1.1155, "step": 2949 }, { "epoch": 0.8966565349544073, "grad_norm": 0.408203125, "learning_rate": 1.0334346504559272e-06, "loss": 1.0177, "step": 2950 }, { "epoch": 0.8969604863221885, "grad_norm": 0.294921875, "learning_rate": 1.0303951367781156e-06, "loss": 1.027, "step": 2951 }, { "epoch": 0.8972644376899696, "grad_norm": 0.28125, "learning_rate": 1.027355623100304e-06, "loss": 1.0299, "step": 2952 }, { "epoch": 0.8975683890577507, "grad_norm": 0.291015625, "learning_rate": 1.0243161094224926e-06, "loss": 1.0534, "step": 2953 }, { "epoch": 0.8978723404255319, "grad_norm": 0.291015625, "learning_rate": 1.0212765957446809e-06, "loss": 1.025, "step": 2954 }, { "epoch": 0.898176291793313, "grad_norm": 0.376953125, "learning_rate": 1.0182370820668693e-06, "loss": 1.0416, "step": 2955 }, { "epoch": 0.8984802431610942, "grad_norm": 0.275390625, "learning_rate": 1.0151975683890578e-06, "loss": 1.1466, "step": 2956 }, { "epoch": 0.8987841945288754, "grad_norm": 0.275390625, "learning_rate": 1.0121580547112463e-06, "loss": 1.1228, "step": 2957 }, { "epoch": 0.8990881458966565, "grad_norm": 0.29296875, "learning_rate": 1.0091185410334347e-06, "loss": 1.0843, "step": 2958 }, { "epoch": 0.8993920972644377, "grad_norm": 0.296875, "learning_rate": 1.0060790273556232e-06, "loss": 1.0633, "step": 2959 }, { "epoch": 0.8996960486322189, "grad_norm": 0.291015625, "learning_rate": 1.0030395136778117e-06, "loss": 0.9965, "step": 2960 }, { "epoch": 0.9, "grad_norm": 0.28125, "learning_rate": 1.0000000000000002e-06, "loss": 1.0726, "step": 2961 }, { "epoch": 0.9003039513677812, "grad_norm": 0.283203125, "learning_rate": 9.969604863221884e-07, "loss": 0.9838, "step": 2962 }, { "epoch": 0.9006079027355623, "grad_norm": 0.283203125, "learning_rate": 9.939209726443771e-07, "loss": 1.0547, "step": 2963 }, { "epoch": 0.9009118541033435, "grad_norm": 0.287109375, "learning_rate": 9.908814589665654e-07, "loss": 1.0454, "step": 2964 }, { "epoch": 0.9012158054711246, "grad_norm": 0.306640625, "learning_rate": 9.878419452887538e-07, "loss": 1.0496, "step": 2965 }, { "epoch": 0.9015197568389057, "grad_norm": 0.28125, "learning_rate": 9.848024316109423e-07, "loss": 1.1259, "step": 2966 }, { "epoch": 0.9018237082066869, "grad_norm": 0.28125, "learning_rate": 9.817629179331308e-07, "loss": 1.0413, "step": 2967 }, { "epoch": 0.902127659574468, "grad_norm": 0.28515625, "learning_rate": 9.787234042553193e-07, "loss": 1.0351, "step": 2968 }, { "epoch": 0.9024316109422492, "grad_norm": 0.294921875, "learning_rate": 9.756838905775075e-07, "loss": 0.9411, "step": 2969 }, { "epoch": 0.9027355623100304, "grad_norm": 0.28125, "learning_rate": 9.726443768996962e-07, "loss": 1.0764, "step": 2970 }, { "epoch": 0.9030395136778115, "grad_norm": 0.294921875, "learning_rate": 9.696048632218847e-07, "loss": 1.1026, "step": 2971 }, { "epoch": 0.9033434650455927, "grad_norm": 0.275390625, "learning_rate": 9.66565349544073e-07, "loss": 1.0527, "step": 2972 }, { "epoch": 0.9036474164133739, "grad_norm": 0.30859375, "learning_rate": 9.635258358662614e-07, "loss": 1.036, "step": 2973 }, { "epoch": 0.903951367781155, "grad_norm": 0.2890625, "learning_rate": 9.604863221884499e-07, "loss": 1.0455, "step": 2974 }, { "epoch": 0.9042553191489362, "grad_norm": 0.291015625, "learning_rate": 9.574468085106384e-07, "loss": 1.0773, "step": 2975 }, { "epoch": 0.9045592705167174, "grad_norm": 0.2890625, "learning_rate": 9.544072948328268e-07, "loss": 1.0245, "step": 2976 }, { "epoch": 0.9048632218844985, "grad_norm": 0.30859375, "learning_rate": 9.513677811550152e-07, "loss": 1.0351, "step": 2977 }, { "epoch": 0.9051671732522797, "grad_norm": 0.28125, "learning_rate": 9.483282674772038e-07, "loss": 1.0236, "step": 2978 }, { "epoch": 0.9054711246200607, "grad_norm": 0.29296875, "learning_rate": 9.452887537993922e-07, "loss": 1.0476, "step": 2979 }, { "epoch": 0.9057750759878419, "grad_norm": 0.302734375, "learning_rate": 9.422492401215806e-07, "loss": 0.9928, "step": 2980 }, { "epoch": 0.9060790273556231, "grad_norm": 0.296875, "learning_rate": 9.392097264437691e-07, "loss": 1.1343, "step": 2981 }, { "epoch": 0.9063829787234042, "grad_norm": 0.29296875, "learning_rate": 9.361702127659575e-07, "loss": 1.0162, "step": 2982 }, { "epoch": 0.9066869300911854, "grad_norm": 0.3125, "learning_rate": 9.331306990881459e-07, "loss": 1.186, "step": 2983 }, { "epoch": 0.9069908814589666, "grad_norm": 0.298828125, "learning_rate": 9.300911854103343e-07, "loss": 1.0204, "step": 2984 }, { "epoch": 0.9072948328267477, "grad_norm": 0.3046875, "learning_rate": 9.270516717325229e-07, "loss": 1.0734, "step": 2985 }, { "epoch": 0.9075987841945289, "grad_norm": 0.294921875, "learning_rate": 9.240121580547114e-07, "loss": 1.0055, "step": 2986 }, { "epoch": 0.90790273556231, "grad_norm": 0.294921875, "learning_rate": 9.209726443768997e-07, "loss": 1.108, "step": 2987 }, { "epoch": 0.9082066869300912, "grad_norm": 0.2890625, "learning_rate": 9.179331306990882e-07, "loss": 1.0924, "step": 2988 }, { "epoch": 0.9085106382978724, "grad_norm": 0.375, "learning_rate": 9.148936170212766e-07, "loss": 1.0191, "step": 2989 }, { "epoch": 0.9088145896656535, "grad_norm": 0.28515625, "learning_rate": 9.118541033434651e-07, "loss": 0.9698, "step": 2990 }, { "epoch": 0.9091185410334347, "grad_norm": 0.2890625, "learning_rate": 9.088145896656536e-07, "loss": 1.1029, "step": 2991 }, { "epoch": 0.9094224924012158, "grad_norm": 0.2890625, "learning_rate": 9.05775075987842e-07, "loss": 1.0611, "step": 2992 }, { "epoch": 0.9097264437689969, "grad_norm": 0.30078125, "learning_rate": 9.027355623100305e-07, "loss": 1.0641, "step": 2993 }, { "epoch": 0.9100303951367781, "grad_norm": 0.28125, "learning_rate": 8.99696048632219e-07, "loss": 1.0829, "step": 2994 }, { "epoch": 0.9103343465045592, "grad_norm": 0.291015625, "learning_rate": 8.966565349544073e-07, "loss": 1.0494, "step": 2995 }, { "epoch": 0.9106382978723404, "grad_norm": 0.294921875, "learning_rate": 8.936170212765959e-07, "loss": 0.9858, "step": 2996 }, { "epoch": 0.9109422492401216, "grad_norm": 0.298828125, "learning_rate": 8.905775075987842e-07, "loss": 0.9833, "step": 2997 }, { "epoch": 0.9112462006079027, "grad_norm": 0.287109375, "learning_rate": 8.875379939209727e-07, "loss": 1.0494, "step": 2998 }, { "epoch": 0.9115501519756839, "grad_norm": 0.29296875, "learning_rate": 8.844984802431612e-07, "loss": 1.0482, "step": 2999 }, { "epoch": 0.9118541033434651, "grad_norm": 0.291015625, "learning_rate": 8.814589665653496e-07, "loss": 1.0795, "step": 3000 }, { "epoch": 0.9121580547112462, "grad_norm": 0.330078125, "learning_rate": 8.784194528875381e-07, "loss": 1.0388, "step": 3001 }, { "epoch": 0.9124620060790274, "grad_norm": 0.283203125, "learning_rate": 8.753799392097265e-07, "loss": 1.0491, "step": 3002 }, { "epoch": 0.9127659574468086, "grad_norm": 0.296875, "learning_rate": 8.72340425531915e-07, "loss": 1.0129, "step": 3003 }, { "epoch": 0.9130699088145897, "grad_norm": 0.28515625, "learning_rate": 8.693009118541035e-07, "loss": 1.0896, "step": 3004 }, { "epoch": 0.9133738601823708, "grad_norm": 0.28515625, "learning_rate": 8.662613981762918e-07, "loss": 1.0606, "step": 3005 }, { "epoch": 0.9136778115501519, "grad_norm": 0.302734375, "learning_rate": 8.632218844984804e-07, "loss": 1.0941, "step": 3006 }, { "epoch": 0.9139817629179331, "grad_norm": 0.29296875, "learning_rate": 8.601823708206688e-07, "loss": 1.0137, "step": 3007 }, { "epoch": 0.9142857142857143, "grad_norm": 0.2890625, "learning_rate": 8.571428571428572e-07, "loss": 1.0986, "step": 3008 }, { "epoch": 0.9145896656534954, "grad_norm": 0.28515625, "learning_rate": 8.541033434650457e-07, "loss": 1.03, "step": 3009 }, { "epoch": 0.9148936170212766, "grad_norm": 0.287109375, "learning_rate": 8.510638297872341e-07, "loss": 1.0074, "step": 3010 }, { "epoch": 0.9151975683890577, "grad_norm": 0.296875, "learning_rate": 8.480243161094226e-07, "loss": 1.1641, "step": 3011 }, { "epoch": 0.9155015197568389, "grad_norm": 0.291015625, "learning_rate": 8.449848024316109e-07, "loss": 1.0381, "step": 3012 }, { "epoch": 0.9158054711246201, "grad_norm": 0.283203125, "learning_rate": 8.419452887537995e-07, "loss": 1.0536, "step": 3013 }, { "epoch": 0.9161094224924012, "grad_norm": 0.275390625, "learning_rate": 8.38905775075988e-07, "loss": 1.0301, "step": 3014 }, { "epoch": 0.9164133738601824, "grad_norm": 0.2890625, "learning_rate": 8.358662613981763e-07, "loss": 1.0624, "step": 3015 }, { "epoch": 0.9167173252279636, "grad_norm": 0.275390625, "learning_rate": 8.328267477203648e-07, "loss": 1.0061, "step": 3016 }, { "epoch": 0.9170212765957447, "grad_norm": 0.28515625, "learning_rate": 8.297872340425532e-07, "loss": 1.0392, "step": 3017 }, { "epoch": 0.9173252279635259, "grad_norm": 0.298828125, "learning_rate": 8.267477203647418e-07, "loss": 1.0319, "step": 3018 }, { "epoch": 0.9176291793313069, "grad_norm": 0.2890625, "learning_rate": 8.237082066869302e-07, "loss": 1.0531, "step": 3019 }, { "epoch": 0.9179331306990881, "grad_norm": 0.328125, "learning_rate": 8.206686930091186e-07, "loss": 1.1054, "step": 3020 }, { "epoch": 0.9182370820668693, "grad_norm": 0.28125, "learning_rate": 8.176291793313071e-07, "loss": 1.1344, "step": 3021 }, { "epoch": 0.9185410334346504, "grad_norm": 0.283203125, "learning_rate": 8.145896656534954e-07, "loss": 1.0442, "step": 3022 }, { "epoch": 0.9188449848024316, "grad_norm": 0.287109375, "learning_rate": 8.115501519756839e-07, "loss": 1.0281, "step": 3023 }, { "epoch": 0.9191489361702128, "grad_norm": 0.287109375, "learning_rate": 8.085106382978725e-07, "loss": 1.0392, "step": 3024 }, { "epoch": 0.9194528875379939, "grad_norm": 0.271484375, "learning_rate": 8.054711246200609e-07, "loss": 1.0375, "step": 3025 }, { "epoch": 0.9197568389057751, "grad_norm": 0.291015625, "learning_rate": 8.024316109422493e-07, "loss": 1.0246, "step": 3026 }, { "epoch": 0.9200607902735563, "grad_norm": 0.28515625, "learning_rate": 7.993920972644377e-07, "loss": 1.0765, "step": 3027 }, { "epoch": 0.9203647416413374, "grad_norm": 0.29296875, "learning_rate": 7.963525835866262e-07, "loss": 0.9797, "step": 3028 }, { "epoch": 0.9206686930091186, "grad_norm": 0.306640625, "learning_rate": 7.933130699088148e-07, "loss": 1.0459, "step": 3029 }, { "epoch": 0.9209726443768997, "grad_norm": 0.283203125, "learning_rate": 7.902735562310031e-07, "loss": 1.116, "step": 3030 }, { "epoch": 0.9212765957446809, "grad_norm": 0.27734375, "learning_rate": 7.872340425531916e-07, "loss": 1.0505, "step": 3031 }, { "epoch": 0.921580547112462, "grad_norm": 0.287109375, "learning_rate": 7.8419452887538e-07, "loss": 1.0239, "step": 3032 }, { "epoch": 0.9218844984802431, "grad_norm": 0.318359375, "learning_rate": 7.811550151975684e-07, "loss": 1.0295, "step": 3033 }, { "epoch": 0.9221884498480243, "grad_norm": 0.302734375, "learning_rate": 7.781155015197569e-07, "loss": 1.0315, "step": 3034 }, { "epoch": 0.9224924012158054, "grad_norm": 0.27734375, "learning_rate": 7.750759878419453e-07, "loss": 0.9975, "step": 3035 }, { "epoch": 0.9227963525835866, "grad_norm": 0.296875, "learning_rate": 7.720364741641339e-07, "loss": 1.029, "step": 3036 }, { "epoch": 0.9231003039513678, "grad_norm": 0.287109375, "learning_rate": 7.689969604863222e-07, "loss": 1.08, "step": 3037 }, { "epoch": 0.9234042553191489, "grad_norm": 0.28125, "learning_rate": 7.659574468085107e-07, "loss": 1.0803, "step": 3038 }, { "epoch": 0.9237082066869301, "grad_norm": 0.28515625, "learning_rate": 7.629179331306992e-07, "loss": 1.0785, "step": 3039 }, { "epoch": 0.9240121580547113, "grad_norm": 0.28515625, "learning_rate": 7.598784194528875e-07, "loss": 1.0227, "step": 3040 }, { "epoch": 0.9243161094224924, "grad_norm": 0.296875, "learning_rate": 7.568389057750761e-07, "loss": 1.0905, "step": 3041 }, { "epoch": 0.9246200607902736, "grad_norm": 0.291015625, "learning_rate": 7.537993920972645e-07, "loss": 1.0625, "step": 3042 }, { "epoch": 0.9249240121580548, "grad_norm": 0.2890625, "learning_rate": 7.50759878419453e-07, "loss": 0.9838, "step": 3043 }, { "epoch": 0.9252279635258359, "grad_norm": 0.28125, "learning_rate": 7.477203647416414e-07, "loss": 1.0319, "step": 3044 }, { "epoch": 0.925531914893617, "grad_norm": 0.283203125, "learning_rate": 7.446808510638298e-07, "loss": 1.0451, "step": 3045 }, { "epoch": 0.9258358662613981, "grad_norm": 0.30078125, "learning_rate": 7.416413373860183e-07, "loss": 1.0662, "step": 3046 }, { "epoch": 0.9261398176291793, "grad_norm": 0.330078125, "learning_rate": 7.386018237082066e-07, "loss": 1.1192, "step": 3047 }, { "epoch": 0.9264437689969605, "grad_norm": 0.298828125, "learning_rate": 7.355623100303952e-07, "loss": 1.1, "step": 3048 }, { "epoch": 0.9267477203647416, "grad_norm": 0.2890625, "learning_rate": 7.325227963525837e-07, "loss": 1.0418, "step": 3049 }, { "epoch": 0.9270516717325228, "grad_norm": 0.2890625, "learning_rate": 7.294832826747721e-07, "loss": 1.0543, "step": 3050 }, { "epoch": 0.927355623100304, "grad_norm": 0.291015625, "learning_rate": 7.264437689969605e-07, "loss": 1.0397, "step": 3051 }, { "epoch": 0.9276595744680851, "grad_norm": 0.287109375, "learning_rate": 7.234042553191489e-07, "loss": 1.0599, "step": 3052 }, { "epoch": 0.9279635258358663, "grad_norm": 0.28125, "learning_rate": 7.203647416413375e-07, "loss": 0.9753, "step": 3053 }, { "epoch": 0.9282674772036474, "grad_norm": 0.291015625, "learning_rate": 7.173252279635259e-07, "loss": 1.0431, "step": 3054 }, { "epoch": 0.9285714285714286, "grad_norm": 0.298828125, "learning_rate": 7.142857142857143e-07, "loss": 1.0371, "step": 3055 }, { "epoch": 0.9288753799392098, "grad_norm": 0.275390625, "learning_rate": 7.112462006079028e-07, "loss": 1.0622, "step": 3056 }, { "epoch": 0.9291793313069909, "grad_norm": 0.279296875, "learning_rate": 7.082066869300912e-07, "loss": 1.1224, "step": 3057 }, { "epoch": 0.929483282674772, "grad_norm": 0.2890625, "learning_rate": 7.051671732522796e-07, "loss": 0.9584, "step": 3058 }, { "epoch": 0.9297872340425531, "grad_norm": 0.283203125, "learning_rate": 7.021276595744682e-07, "loss": 1.0357, "step": 3059 }, { "epoch": 0.9300911854103343, "grad_norm": 0.291015625, "learning_rate": 6.990881458966566e-07, "loss": 1.0266, "step": 3060 }, { "epoch": 0.9303951367781155, "grad_norm": 0.29296875, "learning_rate": 6.96048632218845e-07, "loss": 1.0438, "step": 3061 }, { "epoch": 0.9306990881458966, "grad_norm": 0.296875, "learning_rate": 6.930091185410335e-07, "loss": 1.0945, "step": 3062 }, { "epoch": 0.9310030395136778, "grad_norm": 0.287109375, "learning_rate": 6.899696048632219e-07, "loss": 1.0846, "step": 3063 }, { "epoch": 0.931306990881459, "grad_norm": 0.28125, "learning_rate": 6.869300911854105e-07, "loss": 1.0404, "step": 3064 }, { "epoch": 0.9316109422492401, "grad_norm": 0.279296875, "learning_rate": 6.838905775075988e-07, "loss": 1.0381, "step": 3065 }, { "epoch": 0.9319148936170213, "grad_norm": 0.279296875, "learning_rate": 6.808510638297873e-07, "loss": 1.0773, "step": 3066 }, { "epoch": 0.9322188449848025, "grad_norm": 0.28515625, "learning_rate": 6.778115501519758e-07, "loss": 1.1105, "step": 3067 }, { "epoch": 0.9325227963525836, "grad_norm": 0.310546875, "learning_rate": 6.747720364741641e-07, "loss": 1.083, "step": 3068 }, { "epoch": 0.9328267477203648, "grad_norm": 0.296875, "learning_rate": 6.717325227963527e-07, "loss": 1.0491, "step": 3069 }, { "epoch": 0.9331306990881459, "grad_norm": 0.287109375, "learning_rate": 6.686930091185411e-07, "loss": 1.0812, "step": 3070 }, { "epoch": 0.9334346504559271, "grad_norm": 0.28125, "learning_rate": 6.656534954407296e-07, "loss": 1.0623, "step": 3071 }, { "epoch": 0.9337386018237082, "grad_norm": 0.310546875, "learning_rate": 6.62613981762918e-07, "loss": 0.9714, "step": 3072 }, { "epoch": 0.9340425531914893, "grad_norm": 0.29296875, "learning_rate": 6.595744680851064e-07, "loss": 1.0694, "step": 3073 }, { "epoch": 0.9343465045592705, "grad_norm": 0.314453125, "learning_rate": 6.565349544072949e-07, "loss": 0.9961, "step": 3074 }, { "epoch": 0.9346504559270516, "grad_norm": 0.35546875, "learning_rate": 6.534954407294832e-07, "loss": 1.0303, "step": 3075 }, { "epoch": 0.9349544072948328, "grad_norm": 0.28125, "learning_rate": 6.504559270516718e-07, "loss": 1.1177, "step": 3076 }, { "epoch": 0.935258358662614, "grad_norm": 0.29296875, "learning_rate": 6.474164133738603e-07, "loss": 1.0478, "step": 3077 }, { "epoch": 0.9355623100303951, "grad_norm": 0.287109375, "learning_rate": 6.443768996960487e-07, "loss": 1.0203, "step": 3078 }, { "epoch": 0.9358662613981763, "grad_norm": 0.287109375, "learning_rate": 6.413373860182371e-07, "loss": 1.0482, "step": 3079 }, { "epoch": 0.9361702127659575, "grad_norm": 0.291015625, "learning_rate": 6.382978723404255e-07, "loss": 1.0022, "step": 3080 }, { "epoch": 0.9364741641337386, "grad_norm": 0.29296875, "learning_rate": 6.352583586626141e-07, "loss": 1.012, "step": 3081 }, { "epoch": 0.9367781155015198, "grad_norm": 0.28125, "learning_rate": 6.322188449848026e-07, "loss": 1.0487, "step": 3082 }, { "epoch": 0.937082066869301, "grad_norm": 0.3203125, "learning_rate": 6.291793313069909e-07, "loss": 1.0215, "step": 3083 }, { "epoch": 0.9373860182370821, "grad_norm": 0.29296875, "learning_rate": 6.261398176291794e-07, "loss": 1.0275, "step": 3084 }, { "epoch": 0.9376899696048632, "grad_norm": 0.27734375, "learning_rate": 6.231003039513679e-07, "loss": 1.0689, "step": 3085 }, { "epoch": 0.9379939209726443, "grad_norm": 0.2890625, "learning_rate": 6.200607902735562e-07, "loss": 1.057, "step": 3086 }, { "epoch": 0.9382978723404255, "grad_norm": 0.291015625, "learning_rate": 6.170212765957447e-07, "loss": 1.087, "step": 3087 }, { "epoch": 0.9386018237082067, "grad_norm": 0.28125, "learning_rate": 6.139817629179332e-07, "loss": 0.9969, "step": 3088 }, { "epoch": 0.9389057750759878, "grad_norm": 0.482421875, "learning_rate": 6.109422492401217e-07, "loss": 1.0687, "step": 3089 }, { "epoch": 0.939209726443769, "grad_norm": 0.302734375, "learning_rate": 6.079027355623101e-07, "loss": 1.017, "step": 3090 }, { "epoch": 0.9395136778115502, "grad_norm": 0.28515625, "learning_rate": 6.048632218844985e-07, "loss": 1.0422, "step": 3091 }, { "epoch": 0.9398176291793313, "grad_norm": 0.2890625, "learning_rate": 6.01823708206687e-07, "loss": 1.0543, "step": 3092 }, { "epoch": 0.9401215805471125, "grad_norm": 0.306640625, "learning_rate": 5.987841945288754e-07, "loss": 1.0128, "step": 3093 }, { "epoch": 0.9404255319148936, "grad_norm": 0.36328125, "learning_rate": 5.957446808510639e-07, "loss": 1.0438, "step": 3094 }, { "epoch": 0.9407294832826748, "grad_norm": 0.287109375, "learning_rate": 5.927051671732524e-07, "loss": 1.0451, "step": 3095 }, { "epoch": 0.941033434650456, "grad_norm": 0.287109375, "learning_rate": 5.896656534954408e-07, "loss": 1.0384, "step": 3096 }, { "epoch": 0.9413373860182371, "grad_norm": 0.28125, "learning_rate": 5.866261398176292e-07, "loss": 1.0382, "step": 3097 }, { "epoch": 0.9416413373860182, "grad_norm": 0.287109375, "learning_rate": 5.835866261398176e-07, "loss": 1.0024, "step": 3098 }, { "epoch": 0.9419452887537993, "grad_norm": 0.318359375, "learning_rate": 5.805471124620062e-07, "loss": 1.0838, "step": 3099 }, { "epoch": 0.9422492401215805, "grad_norm": 0.298828125, "learning_rate": 5.775075987841945e-07, "loss": 1.1166, "step": 3100 }, { "epoch": 0.9425531914893617, "grad_norm": 0.298828125, "learning_rate": 5.74468085106383e-07, "loss": 1.0185, "step": 3101 }, { "epoch": 0.9428571428571428, "grad_norm": 0.283203125, "learning_rate": 5.714285714285715e-07, "loss": 1.05, "step": 3102 }, { "epoch": 0.943161094224924, "grad_norm": 0.294921875, "learning_rate": 5.683890577507599e-07, "loss": 1.0449, "step": 3103 }, { "epoch": 0.9434650455927052, "grad_norm": 0.296875, "learning_rate": 5.653495440729484e-07, "loss": 1.0566, "step": 3104 }, { "epoch": 0.9437689969604863, "grad_norm": 0.29296875, "learning_rate": 5.623100303951368e-07, "loss": 1.0407, "step": 3105 }, { "epoch": 0.9440729483282675, "grad_norm": 0.330078125, "learning_rate": 5.592705167173253e-07, "loss": 1.0216, "step": 3106 }, { "epoch": 0.9443768996960487, "grad_norm": 0.3046875, "learning_rate": 5.562310030395138e-07, "loss": 1.1462, "step": 3107 }, { "epoch": 0.9446808510638298, "grad_norm": 0.29296875, "learning_rate": 5.531914893617021e-07, "loss": 1.0224, "step": 3108 }, { "epoch": 0.944984802431611, "grad_norm": 0.287109375, "learning_rate": 5.501519756838906e-07, "loss": 1.0304, "step": 3109 }, { "epoch": 0.9452887537993921, "grad_norm": 0.28515625, "learning_rate": 5.471124620060791e-07, "loss": 1.0788, "step": 3110 }, { "epoch": 0.9455927051671732, "grad_norm": 0.287109375, "learning_rate": 5.440729483282675e-07, "loss": 1.0161, "step": 3111 }, { "epoch": 0.9458966565349544, "grad_norm": 0.302734375, "learning_rate": 5.410334346504559e-07, "loss": 0.9867, "step": 3112 }, { "epoch": 0.9462006079027355, "grad_norm": 0.2890625, "learning_rate": 5.379939209726445e-07, "loss": 1.0493, "step": 3113 }, { "epoch": 0.9465045592705167, "grad_norm": 0.28515625, "learning_rate": 5.349544072948329e-07, "loss": 1.0653, "step": 3114 }, { "epoch": 0.9468085106382979, "grad_norm": 0.29296875, "learning_rate": 5.319148936170213e-07, "loss": 1.0622, "step": 3115 }, { "epoch": 0.947112462006079, "grad_norm": 0.275390625, "learning_rate": 5.288753799392098e-07, "loss": 1.0622, "step": 3116 }, { "epoch": 0.9474164133738602, "grad_norm": 0.287109375, "learning_rate": 5.258358662613982e-07, "loss": 1.0958, "step": 3117 }, { "epoch": 0.9477203647416413, "grad_norm": 0.294921875, "learning_rate": 5.227963525835867e-07, "loss": 1.0847, "step": 3118 }, { "epoch": 0.9480243161094225, "grad_norm": 0.30078125, "learning_rate": 5.197568389057751e-07, "loss": 1.0486, "step": 3119 }, { "epoch": 0.9483282674772037, "grad_norm": 0.291015625, "learning_rate": 5.167173252279636e-07, "loss": 1.049, "step": 3120 }, { "epoch": 0.9486322188449848, "grad_norm": 0.29296875, "learning_rate": 5.13677811550152e-07, "loss": 0.9873, "step": 3121 }, { "epoch": 0.948936170212766, "grad_norm": 0.28125, "learning_rate": 5.106382978723404e-07, "loss": 1.108, "step": 3122 }, { "epoch": 0.9492401215805472, "grad_norm": 0.287109375, "learning_rate": 5.075987841945289e-07, "loss": 1.0427, "step": 3123 }, { "epoch": 0.9495440729483283, "grad_norm": 0.291015625, "learning_rate": 5.045592705167174e-07, "loss": 1.0846, "step": 3124 }, { "epoch": 0.9498480243161094, "grad_norm": 0.2890625, "learning_rate": 5.015197568389058e-07, "loss": 1.0528, "step": 3125 }, { "epoch": 0.9501519756838905, "grad_norm": 0.28515625, "learning_rate": 4.984802431610942e-07, "loss": 1.0491, "step": 3126 }, { "epoch": 0.9504559270516717, "grad_norm": 0.28125, "learning_rate": 4.954407294832827e-07, "loss": 1.0714, "step": 3127 }, { "epoch": 0.9507598784194529, "grad_norm": 0.279296875, "learning_rate": 4.924012158054712e-07, "loss": 1.0843, "step": 3128 }, { "epoch": 0.951063829787234, "grad_norm": 0.2890625, "learning_rate": 4.893617021276596e-07, "loss": 1.0437, "step": 3129 }, { "epoch": 0.9513677811550152, "grad_norm": 0.298828125, "learning_rate": 4.863221884498481e-07, "loss": 0.9894, "step": 3130 }, { "epoch": 0.9516717325227964, "grad_norm": 0.287109375, "learning_rate": 4.832826747720365e-07, "loss": 1.0949, "step": 3131 }, { "epoch": 0.9519756838905775, "grad_norm": 0.296875, "learning_rate": 4.802431610942249e-07, "loss": 0.9864, "step": 3132 }, { "epoch": 0.9522796352583587, "grad_norm": 0.291015625, "learning_rate": 4.772036474164134e-07, "loss": 1.0781, "step": 3133 }, { "epoch": 0.9525835866261398, "grad_norm": 0.296875, "learning_rate": 4.741641337386019e-07, "loss": 0.9742, "step": 3134 }, { "epoch": 0.952887537993921, "grad_norm": 0.294921875, "learning_rate": 4.711246200607903e-07, "loss": 0.9778, "step": 3135 }, { "epoch": 0.9531914893617022, "grad_norm": 0.275390625, "learning_rate": 4.6808510638297873e-07, "loss": 1.1103, "step": 3136 }, { "epoch": 0.9534954407294833, "grad_norm": 0.28125, "learning_rate": 4.6504559270516715e-07, "loss": 0.9754, "step": 3137 }, { "epoch": 0.9537993920972644, "grad_norm": 0.29296875, "learning_rate": 4.620060790273557e-07, "loss": 1.0237, "step": 3138 }, { "epoch": 0.9541033434650456, "grad_norm": 0.2890625, "learning_rate": 4.589665653495441e-07, "loss": 1.0791, "step": 3139 }, { "epoch": 0.9544072948328267, "grad_norm": 0.283203125, "learning_rate": 4.5592705167173257e-07, "loss": 1.0184, "step": 3140 }, { "epoch": 0.9547112462006079, "grad_norm": 0.275390625, "learning_rate": 4.52887537993921e-07, "loss": 1.1191, "step": 3141 }, { "epoch": 0.955015197568389, "grad_norm": 0.283203125, "learning_rate": 4.498480243161095e-07, "loss": 1.1473, "step": 3142 }, { "epoch": 0.9553191489361702, "grad_norm": 0.296875, "learning_rate": 4.4680851063829794e-07, "loss": 1.098, "step": 3143 }, { "epoch": 0.9556231003039514, "grad_norm": 0.28125, "learning_rate": 4.4376899696048636e-07, "loss": 1.0411, "step": 3144 }, { "epoch": 0.9559270516717325, "grad_norm": 0.41796875, "learning_rate": 4.407294832826748e-07, "loss": 1.0456, "step": 3145 }, { "epoch": 0.9562310030395137, "grad_norm": 0.291015625, "learning_rate": 4.3768996960486325e-07, "loss": 1.0074, "step": 3146 }, { "epoch": 0.9565349544072949, "grad_norm": 0.275390625, "learning_rate": 4.3465045592705173e-07, "loss": 1.0682, "step": 3147 }, { "epoch": 0.956838905775076, "grad_norm": 0.287109375, "learning_rate": 4.316109422492402e-07, "loss": 1.0438, "step": 3148 }, { "epoch": 0.9571428571428572, "grad_norm": 0.3125, "learning_rate": 4.285714285714286e-07, "loss": 1.083, "step": 3149 }, { "epoch": 0.9574468085106383, "grad_norm": 0.400390625, "learning_rate": 4.2553191489361704e-07, "loss": 1.0725, "step": 3150 }, { "epoch": 0.9577507598784194, "grad_norm": 0.296875, "learning_rate": 4.2249240121580546e-07, "loss": 1.064, "step": 3151 }, { "epoch": 0.9580547112462006, "grad_norm": 0.2890625, "learning_rate": 4.19452887537994e-07, "loss": 0.9835, "step": 3152 }, { "epoch": 0.9583586626139817, "grad_norm": 0.29296875, "learning_rate": 4.164133738601824e-07, "loss": 1.0848, "step": 3153 }, { "epoch": 0.9586626139817629, "grad_norm": 0.283203125, "learning_rate": 4.133738601823709e-07, "loss": 1.1024, "step": 3154 }, { "epoch": 0.958966565349544, "grad_norm": 0.291015625, "learning_rate": 4.103343465045593e-07, "loss": 1.0123, "step": 3155 }, { "epoch": 0.9592705167173252, "grad_norm": 0.28515625, "learning_rate": 4.072948328267477e-07, "loss": 1.0234, "step": 3156 }, { "epoch": 0.9595744680851064, "grad_norm": 0.287109375, "learning_rate": 4.0425531914893625e-07, "loss": 1.0337, "step": 3157 }, { "epoch": 0.9598784194528875, "grad_norm": 0.294921875, "learning_rate": 4.0121580547112467e-07, "loss": 1.0852, "step": 3158 }, { "epoch": 0.9601823708206687, "grad_norm": 0.2890625, "learning_rate": 3.981762917933131e-07, "loss": 1.0391, "step": 3159 }, { "epoch": 0.9604863221884499, "grad_norm": 0.3046875, "learning_rate": 3.9513677811550156e-07, "loss": 1.0446, "step": 3160 }, { "epoch": 0.960790273556231, "grad_norm": 0.302734375, "learning_rate": 3.9209726443769e-07, "loss": 1.0584, "step": 3161 }, { "epoch": 0.9610942249240122, "grad_norm": 0.28515625, "learning_rate": 3.8905775075987845e-07, "loss": 1.1011, "step": 3162 }, { "epoch": 0.9613981762917934, "grad_norm": 0.283203125, "learning_rate": 3.8601823708206693e-07, "loss": 1.066, "step": 3163 }, { "epoch": 0.9617021276595744, "grad_norm": 0.287109375, "learning_rate": 3.8297872340425535e-07, "loss": 1.0831, "step": 3164 }, { "epoch": 0.9620060790273556, "grad_norm": 0.287109375, "learning_rate": 3.7993920972644377e-07, "loss": 0.9536, "step": 3165 }, { "epoch": 0.9623100303951367, "grad_norm": 0.3046875, "learning_rate": 3.7689969604863224e-07, "loss": 1.0375, "step": 3166 }, { "epoch": 0.9626139817629179, "grad_norm": 0.28515625, "learning_rate": 3.738601823708207e-07, "loss": 0.9866, "step": 3167 }, { "epoch": 0.9629179331306991, "grad_norm": 0.279296875, "learning_rate": 3.7082066869300913e-07, "loss": 1.0702, "step": 3168 }, { "epoch": 0.9632218844984802, "grad_norm": 0.287109375, "learning_rate": 3.677811550151976e-07, "loss": 1.038, "step": 3169 }, { "epoch": 0.9635258358662614, "grad_norm": 0.287109375, "learning_rate": 3.6474164133738603e-07, "loss": 1.0533, "step": 3170 }, { "epoch": 0.9638297872340426, "grad_norm": 0.302734375, "learning_rate": 3.6170212765957445e-07, "loss": 1.1236, "step": 3171 }, { "epoch": 0.9641337386018237, "grad_norm": 0.29296875, "learning_rate": 3.5866261398176297e-07, "loss": 1.0655, "step": 3172 }, { "epoch": 0.9644376899696049, "grad_norm": 0.291015625, "learning_rate": 3.556231003039514e-07, "loss": 1.0273, "step": 3173 }, { "epoch": 0.964741641337386, "grad_norm": 0.306640625, "learning_rate": 3.525835866261398e-07, "loss": 1.143, "step": 3174 }, { "epoch": 0.9650455927051672, "grad_norm": 0.287109375, "learning_rate": 3.495440729483283e-07, "loss": 1.1362, "step": 3175 }, { "epoch": 0.9653495440729484, "grad_norm": 0.291015625, "learning_rate": 3.4650455927051676e-07, "loss": 1.0669, "step": 3176 }, { "epoch": 0.9656534954407295, "grad_norm": 0.51953125, "learning_rate": 3.4346504559270523e-07, "loss": 1.0352, "step": 3177 }, { "epoch": 0.9659574468085106, "grad_norm": 0.287109375, "learning_rate": 3.4042553191489365e-07, "loss": 1.122, "step": 3178 }, { "epoch": 0.9662613981762918, "grad_norm": 0.2734375, "learning_rate": 3.373860182370821e-07, "loss": 1.0805, "step": 3179 }, { "epoch": 0.9665653495440729, "grad_norm": 0.287109375, "learning_rate": 3.3434650455927055e-07, "loss": 1.0717, "step": 3180 }, { "epoch": 0.9668693009118541, "grad_norm": 0.37109375, "learning_rate": 3.31306990881459e-07, "loss": 1.1668, "step": 3181 }, { "epoch": 0.9671732522796352, "grad_norm": 0.28515625, "learning_rate": 3.2826747720364744e-07, "loss": 1.0894, "step": 3182 }, { "epoch": 0.9674772036474164, "grad_norm": 0.29296875, "learning_rate": 3.252279635258359e-07, "loss": 1.0095, "step": 3183 }, { "epoch": 0.9677811550151976, "grad_norm": 0.279296875, "learning_rate": 3.2218844984802433e-07, "loss": 1.0623, "step": 3184 }, { "epoch": 0.9680851063829787, "grad_norm": 0.287109375, "learning_rate": 3.1914893617021275e-07, "loss": 1.0998, "step": 3185 }, { "epoch": 0.9683890577507599, "grad_norm": 0.447265625, "learning_rate": 3.161094224924013e-07, "loss": 1.0446, "step": 3186 }, { "epoch": 0.9686930091185411, "grad_norm": 0.3046875, "learning_rate": 3.130699088145897e-07, "loss": 0.9542, "step": 3187 }, { "epoch": 0.9689969604863222, "grad_norm": 0.28515625, "learning_rate": 3.100303951367781e-07, "loss": 1.0438, "step": 3188 }, { "epoch": 0.9693009118541034, "grad_norm": 0.283203125, "learning_rate": 3.069908814589666e-07, "loss": 1.0119, "step": 3189 }, { "epoch": 0.9696048632218845, "grad_norm": 0.291015625, "learning_rate": 3.0395136778115507e-07, "loss": 1.0522, "step": 3190 }, { "epoch": 0.9699088145896656, "grad_norm": 0.275390625, "learning_rate": 3.009118541033435e-07, "loss": 1.0957, "step": 3191 }, { "epoch": 0.9702127659574468, "grad_norm": 0.29296875, "learning_rate": 2.9787234042553196e-07, "loss": 1.0426, "step": 3192 }, { "epoch": 0.9705167173252279, "grad_norm": 0.3046875, "learning_rate": 2.948328267477204e-07, "loss": 1.0537, "step": 3193 }, { "epoch": 0.9708206686930091, "grad_norm": 0.283203125, "learning_rate": 2.917933130699088e-07, "loss": 1.0371, "step": 3194 }, { "epoch": 0.9711246200607903, "grad_norm": 0.2890625, "learning_rate": 2.887537993920973e-07, "loss": 0.9955, "step": 3195 }, { "epoch": 0.9714285714285714, "grad_norm": 0.2890625, "learning_rate": 2.8571428571428575e-07, "loss": 1.0544, "step": 3196 }, { "epoch": 0.9717325227963526, "grad_norm": 0.28125, "learning_rate": 2.826747720364742e-07, "loss": 1.0125, "step": 3197 }, { "epoch": 0.9720364741641337, "grad_norm": 0.3203125, "learning_rate": 2.7963525835866264e-07, "loss": 1.022, "step": 3198 }, { "epoch": 0.9723404255319149, "grad_norm": 0.302734375, "learning_rate": 2.7659574468085106e-07, "loss": 1.0049, "step": 3199 }, { "epoch": 0.9726443768996961, "grad_norm": 0.283203125, "learning_rate": 2.7355623100303953e-07, "loss": 1.0677, "step": 3200 }, { "epoch": 0.9729483282674772, "grad_norm": 0.29296875, "learning_rate": 2.7051671732522795e-07, "loss": 1.0209, "step": 3201 }, { "epoch": 0.9732522796352584, "grad_norm": 0.275390625, "learning_rate": 2.6747720364741643e-07, "loss": 1.0594, "step": 3202 }, { "epoch": 0.9735562310030396, "grad_norm": 0.275390625, "learning_rate": 2.644376899696049e-07, "loss": 1.0641, "step": 3203 }, { "epoch": 0.9738601823708206, "grad_norm": 0.27734375, "learning_rate": 2.6139817629179337e-07, "loss": 1.0889, "step": 3204 }, { "epoch": 0.9741641337386018, "grad_norm": 0.28515625, "learning_rate": 2.583586626139818e-07, "loss": 1.0665, "step": 3205 }, { "epoch": 0.9744680851063829, "grad_norm": 0.2890625, "learning_rate": 2.553191489361702e-07, "loss": 1.1531, "step": 3206 }, { "epoch": 0.9747720364741641, "grad_norm": 0.294921875, "learning_rate": 2.522796352583587e-07, "loss": 0.9933, "step": 3207 }, { "epoch": 0.9750759878419453, "grad_norm": 0.28515625, "learning_rate": 2.492401215805471e-07, "loss": 1.0307, "step": 3208 }, { "epoch": 0.9753799392097264, "grad_norm": 0.41796875, "learning_rate": 2.462006079027356e-07, "loss": 1.0111, "step": 3209 }, { "epoch": 0.9756838905775076, "grad_norm": 0.29296875, "learning_rate": 2.4316109422492405e-07, "loss": 1.0465, "step": 3210 }, { "epoch": 0.9759878419452888, "grad_norm": 0.291015625, "learning_rate": 2.4012158054711247e-07, "loss": 1.078, "step": 3211 }, { "epoch": 0.9762917933130699, "grad_norm": 0.29296875, "learning_rate": 2.3708206686930095e-07, "loss": 1.0363, "step": 3212 }, { "epoch": 0.9765957446808511, "grad_norm": 0.2890625, "learning_rate": 2.3404255319148937e-07, "loss": 0.9807, "step": 3213 }, { "epoch": 0.9768996960486322, "grad_norm": 0.29296875, "learning_rate": 2.3100303951367784e-07, "loss": 1.0085, "step": 3214 }, { "epoch": 0.9772036474164134, "grad_norm": 0.287109375, "learning_rate": 2.2796352583586629e-07, "loss": 1.0147, "step": 3215 }, { "epoch": 0.9775075987841946, "grad_norm": 0.291015625, "learning_rate": 2.2492401215805476e-07, "loss": 1.0642, "step": 3216 }, { "epoch": 0.9778115501519757, "grad_norm": 0.283203125, "learning_rate": 2.2188449848024318e-07, "loss": 1.098, "step": 3217 }, { "epoch": 0.9781155015197568, "grad_norm": 1.0078125, "learning_rate": 2.1884498480243163e-07, "loss": 1.0147, "step": 3218 }, { "epoch": 0.978419452887538, "grad_norm": 0.28515625, "learning_rate": 2.158054711246201e-07, "loss": 1.0459, "step": 3219 }, { "epoch": 0.9787234042553191, "grad_norm": 0.28125, "learning_rate": 2.1276595744680852e-07, "loss": 1.0184, "step": 3220 }, { "epoch": 0.9790273556231003, "grad_norm": 0.29296875, "learning_rate": 2.09726443768997e-07, "loss": 1.0734, "step": 3221 }, { "epoch": 0.9793313069908814, "grad_norm": 0.275390625, "learning_rate": 2.0668693009118544e-07, "loss": 1.0874, "step": 3222 }, { "epoch": 0.9796352583586626, "grad_norm": 0.29296875, "learning_rate": 2.0364741641337386e-07, "loss": 1.0664, "step": 3223 }, { "epoch": 0.9799392097264438, "grad_norm": 0.2890625, "learning_rate": 2.0060790273556233e-07, "loss": 1.0472, "step": 3224 }, { "epoch": 0.9802431610942249, "grad_norm": 0.2890625, "learning_rate": 1.9756838905775078e-07, "loss": 1.0976, "step": 3225 }, { "epoch": 0.9805471124620061, "grad_norm": 0.287109375, "learning_rate": 1.9452887537993923e-07, "loss": 1.0951, "step": 3226 }, { "epoch": 0.9808510638297873, "grad_norm": 0.28515625, "learning_rate": 1.9148936170212767e-07, "loss": 1.1261, "step": 3227 }, { "epoch": 0.9811550151975684, "grad_norm": 0.27734375, "learning_rate": 1.8844984802431612e-07, "loss": 1.0182, "step": 3228 }, { "epoch": 0.9814589665653496, "grad_norm": 0.283203125, "learning_rate": 1.8541033434650457e-07, "loss": 1.1241, "step": 3229 }, { "epoch": 0.9817629179331308, "grad_norm": 0.3046875, "learning_rate": 1.8237082066869301e-07, "loss": 1.0144, "step": 3230 }, { "epoch": 0.9820668693009118, "grad_norm": 0.296875, "learning_rate": 1.7933130699088149e-07, "loss": 0.9249, "step": 3231 }, { "epoch": 0.982370820668693, "grad_norm": 0.28125, "learning_rate": 1.762917933130699e-07, "loss": 1.0591, "step": 3232 }, { "epoch": 0.9826747720364741, "grad_norm": 0.29296875, "learning_rate": 1.7325227963525838e-07, "loss": 1.0967, "step": 3233 }, { "epoch": 0.9829787234042553, "grad_norm": 0.28515625, "learning_rate": 1.7021276595744683e-07, "loss": 1.0476, "step": 3234 }, { "epoch": 0.9832826747720365, "grad_norm": 0.287109375, "learning_rate": 1.6717325227963527e-07, "loss": 1.0668, "step": 3235 }, { "epoch": 0.9835866261398176, "grad_norm": 0.369140625, "learning_rate": 1.6413373860182372e-07, "loss": 0.9934, "step": 3236 }, { "epoch": 0.9838905775075988, "grad_norm": 0.310546875, "learning_rate": 1.6109422492401217e-07, "loss": 1.037, "step": 3237 }, { "epoch": 0.98419452887538, "grad_norm": 0.283203125, "learning_rate": 1.5805471124620064e-07, "loss": 1.0974, "step": 3238 }, { "epoch": 0.9844984802431611, "grad_norm": 0.2890625, "learning_rate": 1.5501519756838906e-07, "loss": 1.1549, "step": 3239 }, { "epoch": 0.9848024316109423, "grad_norm": 0.283203125, "learning_rate": 1.5197568389057753e-07, "loss": 1.0355, "step": 3240 }, { "epoch": 0.9851063829787234, "grad_norm": 0.298828125, "learning_rate": 1.4893617021276598e-07, "loss": 0.98, "step": 3241 }, { "epoch": 0.9854103343465046, "grad_norm": 0.28125, "learning_rate": 1.458966565349544e-07, "loss": 1.0866, "step": 3242 }, { "epoch": 0.9857142857142858, "grad_norm": 0.2890625, "learning_rate": 1.4285714285714287e-07, "loss": 1.0303, "step": 3243 }, { "epoch": 0.9860182370820668, "grad_norm": 0.3125, "learning_rate": 1.3981762917933132e-07, "loss": 0.9391, "step": 3244 }, { "epoch": 0.986322188449848, "grad_norm": 0.294921875, "learning_rate": 1.3677811550151977e-07, "loss": 1.0125, "step": 3245 }, { "epoch": 0.9866261398176291, "grad_norm": 0.291015625, "learning_rate": 1.3373860182370821e-07, "loss": 1.0552, "step": 3246 }, { "epoch": 0.9869300911854103, "grad_norm": 0.30078125, "learning_rate": 1.3069908814589669e-07, "loss": 1.0891, "step": 3247 }, { "epoch": 0.9872340425531915, "grad_norm": 0.283203125, "learning_rate": 1.276595744680851e-07, "loss": 1.0419, "step": 3248 }, { "epoch": 0.9875379939209726, "grad_norm": 0.29296875, "learning_rate": 1.2462006079027355e-07, "loss": 1.115, "step": 3249 }, { "epoch": 0.9878419452887538, "grad_norm": 0.283203125, "learning_rate": 1.2158054711246203e-07, "loss": 1.0669, "step": 3250 }, { "epoch": 0.988145896656535, "grad_norm": 0.306640625, "learning_rate": 1.1854103343465047e-07, "loss": 0.9959, "step": 3251 }, { "epoch": 0.9884498480243161, "grad_norm": 0.287109375, "learning_rate": 1.1550151975683892e-07, "loss": 1.1169, "step": 3252 }, { "epoch": 0.9887537993920973, "grad_norm": 0.29296875, "learning_rate": 1.1246200607902738e-07, "loss": 1.1066, "step": 3253 }, { "epoch": 0.9890577507598785, "grad_norm": 0.291015625, "learning_rate": 1.0942249240121581e-07, "loss": 1.0476, "step": 3254 }, { "epoch": 0.9893617021276596, "grad_norm": 0.27734375, "learning_rate": 1.0638297872340426e-07, "loss": 1.074, "step": 3255 }, { "epoch": 0.9896656534954408, "grad_norm": 0.291015625, "learning_rate": 1.0334346504559272e-07, "loss": 1.0734, "step": 3256 }, { "epoch": 0.9899696048632218, "grad_norm": 0.314453125, "learning_rate": 1.0030395136778117e-07, "loss": 1.0335, "step": 3257 }, { "epoch": 0.990273556231003, "grad_norm": 0.30859375, "learning_rate": 9.726443768996961e-08, "loss": 1.0401, "step": 3258 }, { "epoch": 0.9905775075987842, "grad_norm": 0.30859375, "learning_rate": 9.422492401215806e-08, "loss": 1.1035, "step": 3259 }, { "epoch": 0.9908814589665653, "grad_norm": 0.3125, "learning_rate": 9.118541033434651e-08, "loss": 1.0946, "step": 3260 }, { "epoch": 0.9911854103343465, "grad_norm": 0.3046875, "learning_rate": 8.814589665653495e-08, "loss": 1.0724, "step": 3261 }, { "epoch": 0.9914893617021276, "grad_norm": 0.294921875, "learning_rate": 8.510638297872341e-08, "loss": 1.0735, "step": 3262 }, { "epoch": 0.9917933130699088, "grad_norm": 0.287109375, "learning_rate": 8.206686930091186e-08, "loss": 1.0443, "step": 3263 }, { "epoch": 0.99209726443769, "grad_norm": 0.31640625, "learning_rate": 7.902735562310032e-08, "loss": 1.0461, "step": 3264 }, { "epoch": 0.9924012158054711, "grad_norm": 0.279296875, "learning_rate": 7.598784194528877e-08, "loss": 1.0428, "step": 3265 }, { "epoch": 0.9927051671732523, "grad_norm": 0.302734375, "learning_rate": 7.29483282674772e-08, "loss": 1.0149, "step": 3266 }, { "epoch": 0.9930091185410335, "grad_norm": 0.2890625, "learning_rate": 6.990881458966566e-08, "loss": 1.0313, "step": 3267 }, { "epoch": 0.9933130699088146, "grad_norm": 0.2734375, "learning_rate": 6.686930091185411e-08, "loss": 1.0764, "step": 3268 }, { "epoch": 0.9936170212765958, "grad_norm": 0.294921875, "learning_rate": 6.382978723404255e-08, "loss": 1.1292, "step": 3269 }, { "epoch": 0.993920972644377, "grad_norm": 0.287109375, "learning_rate": 6.079027355623101e-08, "loss": 1.05, "step": 3270 }, { "epoch": 0.994224924012158, "grad_norm": 0.27734375, "learning_rate": 5.775075987841946e-08, "loss": 1.1569, "step": 3271 }, { "epoch": 0.9945288753799392, "grad_norm": 0.283203125, "learning_rate": 5.471124620060791e-08, "loss": 0.9703, "step": 3272 }, { "epoch": 0.9948328267477203, "grad_norm": 0.27734375, "learning_rate": 5.167173252279636e-08, "loss": 1.0575, "step": 3273 }, { "epoch": 0.9951367781155015, "grad_norm": 0.291015625, "learning_rate": 4.8632218844984807e-08, "loss": 1.1022, "step": 3274 }, { "epoch": 0.9954407294832827, "grad_norm": 0.3203125, "learning_rate": 4.5592705167173253e-08, "loss": 1.0922, "step": 3275 }, { "epoch": 0.9957446808510638, "grad_norm": 0.28515625, "learning_rate": 4.2553191489361707e-08, "loss": 1.0618, "step": 3276 }, { "epoch": 0.996048632218845, "grad_norm": 0.28515625, "learning_rate": 3.951367781155016e-08, "loss": 1.0288, "step": 3277 }, { "epoch": 0.9963525835866262, "grad_norm": 0.330078125, "learning_rate": 3.64741641337386e-08, "loss": 1.0146, "step": 3278 }, { "epoch": 0.9966565349544073, "grad_norm": 0.291015625, "learning_rate": 3.3434650455927053e-08, "loss": 1.006, "step": 3279 }, { "epoch": 0.9969604863221885, "grad_norm": 0.28125, "learning_rate": 3.0395136778115507e-08, "loss": 1.0084, "step": 3280 }, { "epoch": 0.9972644376899696, "grad_norm": 0.291015625, "learning_rate": 2.7355623100303953e-08, "loss": 1.0903, "step": 3281 }, { "epoch": 0.9975683890577508, "grad_norm": 0.28125, "learning_rate": 2.4316109422492403e-08, "loss": 1.0909, "step": 3282 }, { "epoch": 0.997872340425532, "grad_norm": 0.2890625, "learning_rate": 2.1276595744680853e-08, "loss": 1.1004, "step": 3283 }, { "epoch": 0.998176291793313, "grad_norm": 0.283203125, "learning_rate": 1.82370820668693e-08, "loss": 1.1513, "step": 3284 }, { "epoch": 0.9984802431610942, "grad_norm": 0.29296875, "learning_rate": 1.5197568389057753e-08, "loss": 1.0121, "step": 3285 }, { "epoch": 0.9987841945288753, "grad_norm": 0.28515625, "learning_rate": 1.2158054711246202e-08, "loss": 1.0755, "step": 3286 }, { "epoch": 0.9990881458966565, "grad_norm": 0.287109375, "learning_rate": 9.11854103343465e-09, "loss": 1.1004, "step": 3287 }, { "epoch": 0.9993920972644377, "grad_norm": 0.287109375, "learning_rate": 6.079027355623101e-09, "loss": 1.0958, "step": 3288 }, { "epoch": 0.9996960486322188, "grad_norm": 0.326171875, "learning_rate": 3.0395136778115504e-09, "loss": 1.1129, "step": 3289 }, { "epoch": 1.0, "grad_norm": 0.283203125, "learning_rate": 0.0, "loss": 1.029, "step": 3290 } ], "logging_steps": 1.0, "max_steps": 3290, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 0, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.079138300593596e+19, "train_batch_size": 1, "trial_name": null, "trial_params": null }