| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.7711654268508679, | |
| "eval_steps": 500, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017711654268508677, | |
| "grad_norm": 840.4213256835938, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 284.3595, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.035423308537017355, | |
| "grad_norm": 1422.4412841796875, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 132.2688, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.053134962805526036, | |
| "grad_norm": 521.0823974609375, | |
| "learning_rate": 3e-06, | |
| "loss": 143.3815, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07084661707403471, | |
| "grad_norm": 1643.88134765625, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 138.0503, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.08855827134254339, | |
| "grad_norm": 624.1228637695312, | |
| "learning_rate": 5e-06, | |
| "loss": 91.7494, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.10626992561105207, | |
| "grad_norm": 327.2431640625, | |
| "learning_rate": 6e-06, | |
| "loss": 78.6758, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12398157987956075, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7e-06, | |
| "loss": 86.4779, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14169323414806942, | |
| "grad_norm": 659.7210083007812, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 103.229, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.1594048884165781, | |
| "grad_norm": 1251.806396484375, | |
| "learning_rate": 9e-06, | |
| "loss": 65.7367, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.17711654268508678, | |
| "grad_norm": 148.77456665039062, | |
| "learning_rate": 1e-05, | |
| "loss": 69.1247, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19482819695359546, | |
| "grad_norm": 401.77276611328125, | |
| "learning_rate": 9.99695413509548e-06, | |
| "loss": 187.7779, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.21253985122210414, | |
| "grad_norm": 2775.739501953125, | |
| "learning_rate": 9.987820251299121e-06, | |
| "loss": 86.0222, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.23025150549061282, | |
| "grad_norm": 606.4810791015625, | |
| "learning_rate": 9.972609476841368e-06, | |
| "loss": 109.5479, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.2479631597591215, | |
| "grad_norm": 111.0719985961914, | |
| "learning_rate": 9.951340343707852e-06, | |
| "loss": 76.4427, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.26567481402763016, | |
| "grad_norm": 652.419921875, | |
| "learning_rate": 9.924038765061042e-06, | |
| "loss": 99.8802, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.28338646829613884, | |
| "grad_norm": 538.8716430664062, | |
| "learning_rate": 9.890738003669029e-06, | |
| "loss": 85.5841, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.3010981225646475, | |
| "grad_norm": 577.9231567382812, | |
| "learning_rate": 9.851478631379982e-06, | |
| "loss": 62.3816, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.3188097768331562, | |
| "grad_norm": 1849.4739990234375, | |
| "learning_rate": 9.806308479691595e-06, | |
| "loss": 72.0953, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.3365214311016649, | |
| "grad_norm": 1891.8170166015625, | |
| "learning_rate": 9.755282581475769e-06, | |
| "loss": 94.3675, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.35423308537017356, | |
| "grad_norm": 1724.4739990234375, | |
| "learning_rate": 9.698463103929542e-06, | |
| "loss": 86.6209, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.37194473963868224, | |
| "grad_norm": 1946.1993408203125, | |
| "learning_rate": 9.635919272833938e-06, | |
| "loss": 95.5281, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.3896563939071909, | |
| "grad_norm": 352.8471374511719, | |
| "learning_rate": 9.567727288213005e-06, | |
| "loss": 65.7924, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.4073680481756996, | |
| "grad_norm": 1524.947509765625, | |
| "learning_rate": 9.493970231495836e-06, | |
| "loss": 88.3867, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.4250797024442083, | |
| "grad_norm": 583.3900756835938, | |
| "learning_rate": 9.414737964294636e-06, | |
| "loss": 90.2626, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.44279135671271697, | |
| "grad_norm": 1394.978759765625, | |
| "learning_rate": 9.330127018922195e-06, | |
| "loss": 65.8816, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.46050301098122565, | |
| "grad_norm": 460.3956604003906, | |
| "learning_rate": 9.24024048078213e-06, | |
| "loss": 98.1954, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.4782146652497343, | |
| "grad_norm": 1408.5103759765625, | |
| "learning_rate": 9.145187862775208e-06, | |
| "loss": 66.155, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.495926319518243, | |
| "grad_norm": 2149.823486328125, | |
| "learning_rate": 9.045084971874738e-06, | |
| "loss": 59.4641, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.5136379737867517, | |
| "grad_norm": 1315.8822021484375, | |
| "learning_rate": 8.94005376803361e-06, | |
| "loss": 54.4708, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.5313496280552603, | |
| "grad_norm": 5779.1162109375, | |
| "learning_rate": 8.83022221559489e-06, | |
| "loss": 65.8693, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.549061282323769, | |
| "grad_norm": 1683.06689453125, | |
| "learning_rate": 8.715724127386971e-06, | |
| "loss": 59.8766, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.5667729365922777, | |
| "grad_norm": 249.68614196777344, | |
| "learning_rate": 8.596699001693257e-06, | |
| "loss": 54.9801, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.5844845908607864, | |
| "grad_norm": 807.7084350585938, | |
| "learning_rate": 8.473291852294986e-06, | |
| "loss": 88.5092, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.602196245129295, | |
| "grad_norm": 1282.7015380859375, | |
| "learning_rate": 8.345653031794292e-06, | |
| "loss": 52.5884, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.6199078993978038, | |
| "grad_norm": 719.1715087890625, | |
| "learning_rate": 8.213938048432697e-06, | |
| "loss": 68.8219, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.6376195536663124, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.078307376628292e-06, | |
| "loss": 50.8381, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.6553312079348211, | |
| "grad_norm": 335.9521484375, | |
| "learning_rate": 7.938926261462366e-06, | |
| "loss": 64.6055, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.6730428622033298, | |
| "grad_norm": 548.884765625, | |
| "learning_rate": 7.795964517353734e-06, | |
| "loss": 55.1281, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.6907545164718385, | |
| "grad_norm": 3670.740478515625, | |
| "learning_rate": 7.649596321166024e-06, | |
| "loss": 54.0147, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.7084661707403471, | |
| "grad_norm": 444.0411376953125, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 64.277, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.7261778250088559, | |
| "grad_norm": 691.2824096679688, | |
| "learning_rate": 7.347357813929455e-06, | |
| "loss": 75.0035, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.7438894792773645, | |
| "grad_norm": 1208.0020751953125, | |
| "learning_rate": 7.191855733945388e-06, | |
| "loss": 50.5697, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.7616011335458732, | |
| "grad_norm": 254.70382690429688, | |
| "learning_rate": 7.033683215379002e-06, | |
| "loss": 75.5335, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.7793127878143818, | |
| "grad_norm": 292.61505126953125, | |
| "learning_rate": 6.873032967079562e-06, | |
| "loss": 48.5479, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.7970244420828906, | |
| "grad_norm": 166.0381622314453, | |
| "learning_rate": 6.710100716628345e-06, | |
| "loss": 64.7102, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.8147360963513992, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.545084971874738e-06, | |
| "loss": 79.722, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.832447750619908, | |
| "grad_norm": 155.06393432617188, | |
| "learning_rate": 6.378186779084996e-06, | |
| "loss": 82.8574, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.8501594048884166, | |
| "grad_norm": 617.4712524414062, | |
| "learning_rate": 6.209609477998339e-06, | |
| "loss": 89.0717, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.8678710591569253, | |
| "grad_norm": 772.412109375, | |
| "learning_rate": 6.039558454088796e-06, | |
| "loss": 82.9971, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.8855827134254339, | |
| "grad_norm": 862.8140258789062, | |
| "learning_rate": 5.8682408883346535e-06, | |
| "loss": 63.886, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.9032943676939427, | |
| "grad_norm": 382.09161376953125, | |
| "learning_rate": 5.695865504800328e-06, | |
| "loss": 53.4964, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.9210060219624513, | |
| "grad_norm": 460.5621643066406, | |
| "learning_rate": 5.522642316338268e-06, | |
| "loss": 46.1012, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.9387176762309599, | |
| "grad_norm": 1686.7965087890625, | |
| "learning_rate": 5.348782368720627e-06, | |
| "loss": 61.8082, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.9564293304994687, | |
| "grad_norm": 15522.65234375, | |
| "learning_rate": 5.174497483512506e-06, | |
| "loss": 75.6515, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.9741409847679773, | |
| "grad_norm": 8405.8447265625, | |
| "learning_rate": 5e-06, | |
| "loss": 52.1935, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.991852639036486, | |
| "grad_norm": 1334.54541015625, | |
| "learning_rate": 4.825502516487497e-06, | |
| "loss": 40.6468, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 176.01559448242188, | |
| "eval_runtime": 57.5774, | |
| "eval_samples_per_second": 43.594, | |
| "eval_steps_per_second": 10.907, | |
| "step": 5646 | |
| }, | |
| { | |
| "epoch": 1.0095642933049946, | |
| "grad_norm": 765.7606811523438, | |
| "learning_rate": 4.651217631279374e-06, | |
| "loss": 63.3825, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.0272759475735034, | |
| "grad_norm": 527.5181274414062, | |
| "learning_rate": 4.477357683661734e-06, | |
| "loss": 43.7208, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.0449876018420121, | |
| "grad_norm": 211.73487854003906, | |
| "learning_rate": 4.304134495199675e-06, | |
| "loss": 40.4991, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.0626992561105206, | |
| "grad_norm": 400.1595764160156, | |
| "learning_rate": 4.131759111665349e-06, | |
| "loss": 43.3378, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.0804109103790294, | |
| "grad_norm": 389.70013427734375, | |
| "learning_rate": 3.960441545911205e-06, | |
| "loss": 41.4699, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.098122564647538, | |
| "grad_norm": 326.5900573730469, | |
| "learning_rate": 3.790390522001662e-06, | |
| "loss": 47.9553, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.1158342189160468, | |
| "grad_norm": 7330.3310546875, | |
| "learning_rate": 3.6218132209150047e-06, | |
| "loss": 50.2975, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.1335458731845554, | |
| "grad_norm": 216.81912231445312, | |
| "learning_rate": 3.4549150281252635e-06, | |
| "loss": 45.4736, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.151257527453064, | |
| "grad_norm": 660.7682495117188, | |
| "learning_rate": 3.289899283371657e-06, | |
| "loss": 40.2617, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.1689691817215728, | |
| "grad_norm": 389.1190490722656, | |
| "learning_rate": 3.12696703292044e-06, | |
| "loss": 35.4133, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.1866808359900816, | |
| "grad_norm": 877.7334594726562, | |
| "learning_rate": 2.966316784621e-06, | |
| "loss": 71.6576, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.20439249025859, | |
| "grad_norm": 378.35980224609375, | |
| "learning_rate": 2.8081442660546126e-06, | |
| "loss": 34.5238, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.2221041445270988, | |
| "grad_norm": 1453.03515625, | |
| "learning_rate": 2.6526421860705474e-06, | |
| "loss": 62.9686, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.2398157987956075, | |
| "grad_norm": 132.0270538330078, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 52.0945, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.257527453064116, | |
| "grad_norm": 440.1296691894531, | |
| "learning_rate": 2.3504036788339763e-06, | |
| "loss": 39.2655, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.2752391073326248, | |
| "grad_norm": 770.677734375, | |
| "learning_rate": 2.204035482646267e-06, | |
| "loss": 38.8261, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.2929507616011335, | |
| "grad_norm": 114.07510375976562, | |
| "learning_rate": 2.061073738537635e-06, | |
| "loss": 47.3239, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.3106624158696423, | |
| "grad_norm": 10688.4658203125, | |
| "learning_rate": 1.9216926233717087e-06, | |
| "loss": 62.3463, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.328374070138151, | |
| "grad_norm": 1633.7628173828125, | |
| "learning_rate": 1.7860619515673034e-06, | |
| "loss": 55.4716, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.3460857244066595, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6543469682057105e-06, | |
| "loss": 59.58, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.3637973786751683, | |
| "grad_norm": 159.12704467773438, | |
| "learning_rate": 1.5267081477050132e-06, | |
| "loss": 41.1704, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.381509032943677, | |
| "grad_norm": 801.6414184570312, | |
| "learning_rate": 1.4033009983067454e-06, | |
| "loss": 51.0953, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.3992206872121855, | |
| "grad_norm": 857.81201171875, | |
| "learning_rate": 1.2842758726130283e-06, | |
| "loss": 40.7129, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.4169323414806942, | |
| "grad_norm": 575.1051635742188, | |
| "learning_rate": 1.1697777844051105e-06, | |
| "loss": 46.1708, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.434643995749203, | |
| "grad_norm": 2294.723388671875, | |
| "learning_rate": 1.0599462319663906e-06, | |
| "loss": 60.1358, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.4523556500177117, | |
| "grad_norm": 932.974853515625, | |
| "learning_rate": 9.549150281252633e-07, | |
| "loss": 43.8091, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.4700673042862205, | |
| "grad_norm": 874.0687255859375, | |
| "learning_rate": 8.54812137224792e-07, | |
| "loss": 37.7215, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.487778958554729, | |
| "grad_norm": 262.20599365234375, | |
| "learning_rate": 7.597595192178702e-07, | |
| "loss": 85.4072, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.5054906128232377, | |
| "grad_norm": 1489.546875, | |
| "learning_rate": 6.698729810778065e-07, | |
| "loss": 32.7225, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.5232022670917464, | |
| "grad_norm": 160.3507537841797, | |
| "learning_rate": 5.852620357053651e-07, | |
| "loss": 56.6163, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.540913921360255, | |
| "grad_norm": 224.78094482421875, | |
| "learning_rate": 5.06029768504166e-07, | |
| "loss": 40.1567, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.5586255756287637, | |
| "grad_norm": 2062.70263671875, | |
| "learning_rate": 4.322727117869951e-07, | |
| "loss": 44.4478, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.5763372298972724, | |
| "grad_norm": 2095.88525390625, | |
| "learning_rate": 3.6408072716606346e-07, | |
| "loss": 59.679, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.594048884165781, | |
| "grad_norm": 3169.360107421875, | |
| "learning_rate": 3.015368960704584e-07, | |
| "loss": 46.9302, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.61176053843429, | |
| "grad_norm": 172.900634765625, | |
| "learning_rate": 2.447174185242324e-07, | |
| "loss": 37.7532, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.6294721927027984, | |
| "grad_norm": 281.58856201171875, | |
| "learning_rate": 1.9369152030840553e-07, | |
| "loss": 32.5123, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.6471838469713072, | |
| "grad_norm": 367.6300048828125, | |
| "learning_rate": 1.4852136862001766e-07, | |
| "loss": 40.0032, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.664895501239816, | |
| "grad_norm": 373.369873046875, | |
| "learning_rate": 1.0926199633097156e-07, | |
| "loss": 37.6046, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 1.6826071555083244, | |
| "grad_norm": 758.5933227539062, | |
| "learning_rate": 7.59612349389599e-08, | |
| "loss": 31.245, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.7003188097768331, | |
| "grad_norm": 557.6469116210938, | |
| "learning_rate": 4.865965629214819e-08, | |
| "loss": 68.3517, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.7180304640453419, | |
| "grad_norm": 297.67327880859375, | |
| "learning_rate": 2.7390523158633552e-08, | |
| "loss": 45.4609, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 1.7357421183138504, | |
| "grad_norm": 98.50409698486328, | |
| "learning_rate": 1.2179748700879013e-08, | |
| "loss": 35.9993, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 1.7534537725823593, | |
| "grad_norm": 911.8038940429688, | |
| "learning_rate": 3.0458649045211897e-09, | |
| "loss": 41.5838, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.7711654268508679, | |
| "grad_norm": 755.3094482421875, | |
| "learning_rate": 0.0, | |
| "loss": 57.0455, | |
| "step": 10000 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 5000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |