|
{ |
|
"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": 4225.109375, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 308.9427, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.035423308537017355, |
|
"grad_norm": 27728.119140625, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 355.2513, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.053134962805526036, |
|
"grad_norm": 830.27294921875, |
|
"learning_rate": 3e-06, |
|
"loss": 161.2664, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.07084661707403471, |
|
"grad_norm": 127.9210205078125, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 110.5367, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.08855827134254339, |
|
"grad_norm": 306.7760314941406, |
|
"learning_rate": 5e-06, |
|
"loss": 164.8705, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.10626992561105207, |
|
"grad_norm": 297.30364990234375, |
|
"learning_rate": 6e-06, |
|
"loss": 297.7952, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.12398157987956075, |
|
"grad_norm": 1049.6143798828125, |
|
"learning_rate": 7e-06, |
|
"loss": 98.5203, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.14169323414806942, |
|
"grad_norm": 376.938232421875, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 168.0427, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.1594048884165781, |
|
"grad_norm": 503.4874267578125, |
|
"learning_rate": 9e-06, |
|
"loss": 185.119, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.17711654268508678, |
|
"grad_norm": 799.0496826171875, |
|
"learning_rate": 1e-05, |
|
"loss": 70.9623, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.19482819695359546, |
|
"grad_norm": 293.7897033691406, |
|
"learning_rate": 9.99695413509548e-06, |
|
"loss": 149.6974, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.21253985122210414, |
|
"grad_norm": 940.9512939453125, |
|
"learning_rate": 9.987820251299121e-06, |
|
"loss": 86.5666, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.23025150549061282, |
|
"grad_norm": 900.4419555664062, |
|
"learning_rate": 9.972609476841368e-06, |
|
"loss": 97.8882, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.2479631597591215, |
|
"grad_norm": 484.7297058105469, |
|
"learning_rate": 9.951340343707852e-06, |
|
"loss": 139.5794, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.26567481402763016, |
|
"grad_norm": 466.8398742675781, |
|
"learning_rate": 9.924038765061042e-06, |
|
"loss": 135.3206, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.28338646829613884, |
|
"grad_norm": 860.2133178710938, |
|
"learning_rate": 9.890738003669029e-06, |
|
"loss": 89.5288, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.3010981225646475, |
|
"grad_norm": 1099.1881103515625, |
|
"learning_rate": 9.851478631379982e-06, |
|
"loss": 91.6384, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.3188097768331562, |
|
"grad_norm": 250.72775268554688, |
|
"learning_rate": 9.806308479691595e-06, |
|
"loss": 154.9306, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.3365214311016649, |
|
"grad_norm": 12806.482421875, |
|
"learning_rate": 9.755282581475769e-06, |
|
"loss": 191.4907, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.35423308537017356, |
|
"grad_norm": 248.52146911621094, |
|
"learning_rate": 9.698463103929542e-06, |
|
"loss": 99.3807, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.37194473963868224, |
|
"grad_norm": 591.319091796875, |
|
"learning_rate": 9.635919272833938e-06, |
|
"loss": 107.2973, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.3896563939071909, |
|
"grad_norm": 1257.0123291015625, |
|
"learning_rate": 9.567727288213005e-06, |
|
"loss": 121.5465, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.4073680481756996, |
|
"grad_norm": 501.8513488769531, |
|
"learning_rate": 9.493970231495836e-06, |
|
"loss": 130.1157, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.4250797024442083, |
|
"grad_norm": 4555.56787109375, |
|
"learning_rate": 9.414737964294636e-06, |
|
"loss": 61.3746, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.44279135671271697, |
|
"grad_norm": 1030.1964111328125, |
|
"learning_rate": 9.330127018922195e-06, |
|
"loss": 123.0796, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.46050301098122565, |
|
"grad_norm": 788.5494384765625, |
|
"learning_rate": 9.24024048078213e-06, |
|
"loss": 92.3144, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.4782146652497343, |
|
"grad_norm": 486.0594177246094, |
|
"learning_rate": 9.145187862775208e-06, |
|
"loss": 137.4813, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.495926319518243, |
|
"grad_norm": 648.8170776367188, |
|
"learning_rate": 9.045084971874738e-06, |
|
"loss": 106.6566, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.5136379737867517, |
|
"grad_norm": 1567.68505859375, |
|
"learning_rate": 8.94005376803361e-06, |
|
"loss": 90.1854, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.5313496280552603, |
|
"grad_norm": 631.2234497070312, |
|
"learning_rate": 8.83022221559489e-06, |
|
"loss": 143.8329, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.549061282323769, |
|
"grad_norm": 327.2228088378906, |
|
"learning_rate": 8.715724127386971e-06, |
|
"loss": 126.8167, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.5667729365922777, |
|
"grad_norm": 524.2578125, |
|
"learning_rate": 8.596699001693257e-06, |
|
"loss": 135.6904, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.5844845908607864, |
|
"grad_norm": 587.7288208007812, |
|
"learning_rate": 8.473291852294986e-06, |
|
"loss": 144.2552, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.602196245129295, |
|
"grad_norm": 292.5152893066406, |
|
"learning_rate": 8.345653031794292e-06, |
|
"loss": 107.0582, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.6199078993978038, |
|
"grad_norm": 2563.22216796875, |
|
"learning_rate": 8.213938048432697e-06, |
|
"loss": 107.5712, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.6376195536663124, |
|
"grad_norm": 2963.86279296875, |
|
"learning_rate": 8.078307376628292e-06, |
|
"loss": 75.9044, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.6553312079348211, |
|
"grad_norm": 216.5026397705078, |
|
"learning_rate": 7.938926261462366e-06, |
|
"loss": 79.0156, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.6730428622033298, |
|
"grad_norm": 674.3363037109375, |
|
"learning_rate": 7.795964517353734e-06, |
|
"loss": 67.8441, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.6907545164718385, |
|
"grad_norm": 284.2720947265625, |
|
"learning_rate": 7.649596321166024e-06, |
|
"loss": 77.3812, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.7084661707403471, |
|
"grad_norm": 4744.10791015625, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 102.2295, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.7261778250088559, |
|
"grad_norm": 3887.924560546875, |
|
"learning_rate": 7.347357813929455e-06, |
|
"loss": 119.0388, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.7438894792773645, |
|
"grad_norm": 6186.1201171875, |
|
"learning_rate": 7.191855733945388e-06, |
|
"loss": 96.2807, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.7616011335458732, |
|
"grad_norm": 229.56248474121094, |
|
"learning_rate": 7.033683215379002e-06, |
|
"loss": 101.2205, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.7793127878143818, |
|
"grad_norm": 306.1286926269531, |
|
"learning_rate": 6.873032967079562e-06, |
|
"loss": 77.5637, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.7970244420828906, |
|
"grad_norm": 549.1785888671875, |
|
"learning_rate": 6.710100716628345e-06, |
|
"loss": 72.9524, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.8147360963513992, |
|
"grad_norm": 560.0545654296875, |
|
"learning_rate": 6.545084971874738e-06, |
|
"loss": 96.7195, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 0.832447750619908, |
|
"grad_norm": 492.44091796875, |
|
"learning_rate": 6.378186779084996e-06, |
|
"loss": 100.4574, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 0.8501594048884166, |
|
"grad_norm": 471.7704162597656, |
|
"learning_rate": 6.209609477998339e-06, |
|
"loss": 97.3029, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 0.8678710591569253, |
|
"grad_norm": 266.0693359375, |
|
"learning_rate": 6.039558454088796e-06, |
|
"loss": 84.8428, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 0.8855827134254339, |
|
"grad_norm": 863.2924194335938, |
|
"learning_rate": 5.8682408883346535e-06, |
|
"loss": 67.9786, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.9032943676939427, |
|
"grad_norm": 1584.0009765625, |
|
"learning_rate": 5.695865504800328e-06, |
|
"loss": 100.1995, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 0.9210060219624513, |
|
"grad_norm": 2196.005126953125, |
|
"learning_rate": 5.522642316338268e-06, |
|
"loss": 90.8572, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 0.9387176762309599, |
|
"grad_norm": 598.8994140625, |
|
"learning_rate": 5.348782368720627e-06, |
|
"loss": 66.768, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 0.9564293304994687, |
|
"grad_norm": 696.6085815429688, |
|
"learning_rate": 5.174497483512506e-06, |
|
"loss": 78.7763, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 0.9741409847679773, |
|
"grad_norm": 273.0209655761719, |
|
"learning_rate": 5e-06, |
|
"loss": 235.4915, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.991852639036486, |
|
"grad_norm": 1731.2950439453125, |
|
"learning_rate": 4.825502516487497e-06, |
|
"loss": 82.1385, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 226.7201690673828, |
|
"eval_runtime": 21.734, |
|
"eval_samples_per_second": 115.487, |
|
"eval_steps_per_second": 28.895, |
|
"step": 5646 |
|
}, |
|
{ |
|
"epoch": 1.0095642933049946, |
|
"grad_norm": 538.6300659179688, |
|
"learning_rate": 4.651217631279374e-06, |
|
"loss": 56.6562, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 1.0272759475735034, |
|
"grad_norm": 1225.543212890625, |
|
"learning_rate": 4.477357683661734e-06, |
|
"loss": 82.8729, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 1.0449876018420121, |
|
"grad_norm": 652.039794921875, |
|
"learning_rate": 4.304134495199675e-06, |
|
"loss": 72.214, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 1.0626992561105206, |
|
"grad_norm": 248.68992614746094, |
|
"learning_rate": 4.131759111665349e-06, |
|
"loss": 142.3135, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 1.0804109103790294, |
|
"grad_norm": 457.009033203125, |
|
"learning_rate": 3.960441545911205e-06, |
|
"loss": 66.6412, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 1.098122564647538, |
|
"grad_norm": 817.4805297851562, |
|
"learning_rate": 3.790390522001662e-06, |
|
"loss": 97.1646, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 1.1158342189160468, |
|
"grad_norm": 1199.6221923828125, |
|
"learning_rate": 3.6218132209150047e-06, |
|
"loss": 88.5047, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 1.1335458731845554, |
|
"grad_norm": 553.9771728515625, |
|
"learning_rate": 3.4549150281252635e-06, |
|
"loss": 101.1097, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 1.151257527453064, |
|
"grad_norm": 814.3031616210938, |
|
"learning_rate": 3.289899283371657e-06, |
|
"loss": 82.8931, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 1.1689691817215728, |
|
"grad_norm": 274.961181640625, |
|
"learning_rate": 3.12696703292044e-06, |
|
"loss": 92.692, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 1.1866808359900816, |
|
"grad_norm": 1075.9053955078125, |
|
"learning_rate": 2.966316784621e-06, |
|
"loss": 102.8753, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 1.20439249025859, |
|
"grad_norm": 117.70056915283203, |
|
"learning_rate": 2.8081442660546126e-06, |
|
"loss": 87.495, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 1.2221041445270988, |
|
"grad_norm": 457.9502258300781, |
|
"learning_rate": 2.6526421860705474e-06, |
|
"loss": 57.6712, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 1.2398157987956075, |
|
"grad_norm": 1044.0828857421875, |
|
"learning_rate": 2.5000000000000015e-06, |
|
"loss": 100.2862, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 1.257527453064116, |
|
"grad_norm": 454.90802001953125, |
|
"learning_rate": 2.3504036788339763e-06, |
|
"loss": 153.9108, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 1.2752391073326248, |
|
"grad_norm": 275.4113464355469, |
|
"learning_rate": 2.204035482646267e-06, |
|
"loss": 133.3204, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 1.2929507616011335, |
|
"grad_norm": 323.5784606933594, |
|
"learning_rate": 2.061073738537635e-06, |
|
"loss": 59.3089, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 1.3106624158696423, |
|
"grad_norm": 2257.790771484375, |
|
"learning_rate": 1.9216926233717087e-06, |
|
"loss": 152.4049, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 1.328374070138151, |
|
"grad_norm": 503.2570495605469, |
|
"learning_rate": 1.7860619515673034e-06, |
|
"loss": 96.8615, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 1.3460857244066595, |
|
"grad_norm": 3797.780517578125, |
|
"learning_rate": 1.6543469682057105e-06, |
|
"loss": 61.9906, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 1.3637973786751683, |
|
"grad_norm": 1134.3297119140625, |
|
"learning_rate": 1.5267081477050132e-06, |
|
"loss": 72.4744, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 1.381509032943677, |
|
"grad_norm": 187.08917236328125, |
|
"learning_rate": 1.4033009983067454e-06, |
|
"loss": 99.3734, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 1.3992206872121855, |
|
"grad_norm": 617.7765502929688, |
|
"learning_rate": 1.2842758726130283e-06, |
|
"loss": 143.0522, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 1.4169323414806942, |
|
"grad_norm": 1724.1121826171875, |
|
"learning_rate": 1.1697777844051105e-06, |
|
"loss": 88.8854, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 1.434643995749203, |
|
"grad_norm": 1899.291259765625, |
|
"learning_rate": 1.0599462319663906e-06, |
|
"loss": 60.2541, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 1.4523556500177117, |
|
"grad_norm": 775.2619018554688, |
|
"learning_rate": 9.549150281252633e-07, |
|
"loss": 60.3232, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 1.4700673042862205, |
|
"grad_norm": 522.8294067382812, |
|
"learning_rate": 8.54812137224792e-07, |
|
"loss": 68.5027, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 1.487778958554729, |
|
"grad_norm": 804.1600341796875, |
|
"learning_rate": 7.597595192178702e-07, |
|
"loss": 133.9367, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 1.5054906128232377, |
|
"grad_norm": 582.5191040039062, |
|
"learning_rate": 6.698729810778065e-07, |
|
"loss": 55.4999, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 1.5232022670917464, |
|
"grad_norm": 406.82830810546875, |
|
"learning_rate": 5.852620357053651e-07, |
|
"loss": 55.5675, |
|
"step": 8600 |
|
}, |
|
{ |
|
"epoch": 1.540913921360255, |
|
"grad_norm": 523.9061889648438, |
|
"learning_rate": 5.06029768504166e-07, |
|
"loss": 74.5364, |
|
"step": 8700 |
|
}, |
|
{ |
|
"epoch": 1.5586255756287637, |
|
"grad_norm": 397.27508544921875, |
|
"learning_rate": 4.322727117869951e-07, |
|
"loss": 95.4718, |
|
"step": 8800 |
|
}, |
|
{ |
|
"epoch": 1.5763372298972724, |
|
"grad_norm": 827.7362670898438, |
|
"learning_rate": 3.6408072716606346e-07, |
|
"loss": 61.7358, |
|
"step": 8900 |
|
}, |
|
{ |
|
"epoch": 1.594048884165781, |
|
"grad_norm": 8013.48486328125, |
|
"learning_rate": 3.015368960704584e-07, |
|
"loss": 67.585, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 1.61176053843429, |
|
"grad_norm": 4467.837890625, |
|
"learning_rate": 2.447174185242324e-07, |
|
"loss": 108.6042, |
|
"step": 9100 |
|
}, |
|
{ |
|
"epoch": 1.6294721927027984, |
|
"grad_norm": 1124.7557373046875, |
|
"learning_rate": 1.9369152030840553e-07, |
|
"loss": 88.8729, |
|
"step": 9200 |
|
}, |
|
{ |
|
"epoch": 1.6471838469713072, |
|
"grad_norm": 461.0145263671875, |
|
"learning_rate": 1.4852136862001766e-07, |
|
"loss": 106.0446, |
|
"step": 9300 |
|
}, |
|
{ |
|
"epoch": 1.664895501239816, |
|
"grad_norm": 1302.7359619140625, |
|
"learning_rate": 1.0926199633097156e-07, |
|
"loss": 52.0522, |
|
"step": 9400 |
|
}, |
|
{ |
|
"epoch": 1.6826071555083244, |
|
"grad_norm": 570.8071899414062, |
|
"learning_rate": 7.59612349389599e-08, |
|
"loss": 66.1858, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 1.7003188097768331, |
|
"grad_norm": 791.3758544921875, |
|
"learning_rate": 4.865965629214819e-08, |
|
"loss": 91.2349, |
|
"step": 9600 |
|
}, |
|
{ |
|
"epoch": 1.7180304640453419, |
|
"grad_norm": 1101.4031982421875, |
|
"learning_rate": 2.7390523158633552e-08, |
|
"loss": 102.8052, |
|
"step": 9700 |
|
}, |
|
{ |
|
"epoch": 1.7357421183138504, |
|
"grad_norm": 1016.6182250976562, |
|
"learning_rate": 1.2179748700879013e-08, |
|
"loss": 132.7038, |
|
"step": 9800 |
|
}, |
|
{ |
|
"epoch": 1.7534537725823593, |
|
"grad_norm": 265.1811218261719, |
|
"learning_rate": 3.0458649045211897e-09, |
|
"loss": 116.7223, |
|
"step": 9900 |
|
}, |
|
{ |
|
"epoch": 1.7711654268508679, |
|
"grad_norm": 1466.9412841796875, |
|
"learning_rate": 0.0, |
|
"loss": 103.9828, |
|
"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 |
|
} |
|
|