| { | |
| "best_metric": 0.6516264428121721, | |
| "best_model_checkpoint": "/media/cse/HDD/Shawon/shawon/Timesformer_BdSLW60/checkpoint-1754", | |
| "epoch": 2.1990867579908677, | |
| "eval_steps": 500, | |
| "global_step": 2190, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0045662100456621, | |
| "grad_norm": 13.168639183044434, | |
| "learning_rate": 2.2831050228310503e-06, | |
| "loss": 4.2832, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0091324200913242, | |
| "grad_norm": 12.616332054138184, | |
| "learning_rate": 4.566210045662101e-06, | |
| "loss": 4.2594, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0136986301369863, | |
| "grad_norm": 12.632943153381348, | |
| "learning_rate": 6.849315068493151e-06, | |
| "loss": 4.1827, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0182648401826484, | |
| "grad_norm": 12.6991548538208, | |
| "learning_rate": 9.132420091324201e-06, | |
| "loss": 4.0751, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0228310502283105, | |
| "grad_norm": 13.425009727478027, | |
| "learning_rate": 1.1415525114155251e-05, | |
| "loss": 4.0899, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0273972602739726, | |
| "grad_norm": 13.551825523376465, | |
| "learning_rate": 1.3698630136986302e-05, | |
| "loss": 3.9588, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0319634703196347, | |
| "grad_norm": 12.522140502929688, | |
| "learning_rate": 1.5981735159817352e-05, | |
| "loss": 3.9251, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0365296803652968, | |
| "grad_norm": 12.543455123901367, | |
| "learning_rate": 1.8264840182648402e-05, | |
| "loss": 3.8922, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0410958904109589, | |
| "grad_norm": 13.488080024719238, | |
| "learning_rate": 2.0547945205479453e-05, | |
| "loss": 4.0068, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.045662100456621, | |
| "grad_norm": 12.935967445373535, | |
| "learning_rate": 2.2831050228310503e-05, | |
| "loss": 3.6687, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0502283105022831, | |
| "grad_norm": 12.15580940246582, | |
| "learning_rate": 2.5114155251141553e-05, | |
| "loss": 3.5509, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0547945205479452, | |
| "grad_norm": 13.309331893920898, | |
| "learning_rate": 2.7397260273972603e-05, | |
| "loss": 3.4915, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0593607305936073, | |
| "grad_norm": 12.072916030883789, | |
| "learning_rate": 2.9680365296803654e-05, | |
| "loss": 3.3421, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0639269406392694, | |
| "grad_norm": 13.800947189331055, | |
| "learning_rate": 3.1963470319634704e-05, | |
| "loss": 3.3284, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0684931506849315, | |
| "grad_norm": 14.214157104492188, | |
| "learning_rate": 3.424657534246575e-05, | |
| "loss": 3.1462, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0730593607305936, | |
| "grad_norm": 12.837231636047363, | |
| "learning_rate": 3.6529680365296805e-05, | |
| "loss": 2.9169, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0776255707762557, | |
| "grad_norm": 14.62553882598877, | |
| "learning_rate": 3.881278538812785e-05, | |
| "loss": 2.9004, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0821917808219178, | |
| "grad_norm": 14.03178596496582, | |
| "learning_rate": 4.1095890410958905e-05, | |
| "loss": 2.9808, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0867579908675799, | |
| "grad_norm": 12.56568717956543, | |
| "learning_rate": 4.337899543378995e-05, | |
| "loss": 2.6627, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.091324200913242, | |
| "grad_norm": 14.119081497192383, | |
| "learning_rate": 4.5662100456621006e-05, | |
| "loss": 2.4799, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0958904109589041, | |
| "grad_norm": 12.76337718963623, | |
| "learning_rate": 4.794520547945205e-05, | |
| "loss": 2.318, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1004566210045662, | |
| "grad_norm": 12.352985382080078, | |
| "learning_rate": 4.997463216641299e-05, | |
| "loss": 2.0831, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.1050228310502283, | |
| "grad_norm": 11.853866577148438, | |
| "learning_rate": 4.9720953830542875e-05, | |
| "loss": 2.4107, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.1095890410958904, | |
| "grad_norm": 13.12740421295166, | |
| "learning_rate": 4.9467275494672755e-05, | |
| "loss": 2.0582, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1141552511415525, | |
| "grad_norm": 11.828140258789062, | |
| "learning_rate": 4.9213597158802636e-05, | |
| "loss": 1.8289, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1187214611872146, | |
| "grad_norm": 12.555204391479492, | |
| "learning_rate": 4.895991882293252e-05, | |
| "loss": 1.5664, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1232876712328767, | |
| "grad_norm": 10.697933197021484, | |
| "learning_rate": 4.8706240487062404e-05, | |
| "loss": 1.5136, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1278538812785388, | |
| "grad_norm": 11.606474876403809, | |
| "learning_rate": 4.8452562151192285e-05, | |
| "loss": 1.6229, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1324200913242009, | |
| "grad_norm": 9.681914329528809, | |
| "learning_rate": 4.819888381532217e-05, | |
| "loss": 1.7203, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.136986301369863, | |
| "grad_norm": 10.477850914001465, | |
| "learning_rate": 4.794520547945205e-05, | |
| "loss": 1.5071, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1415525114155251, | |
| "grad_norm": 11.697348594665527, | |
| "learning_rate": 4.769152714358194e-05, | |
| "loss": 1.6212, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1461187214611872, | |
| "grad_norm": 11.230805397033691, | |
| "learning_rate": 4.743784880771182e-05, | |
| "loss": 1.3522, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1506849315068493, | |
| "grad_norm": 8.138943672180176, | |
| "learning_rate": 4.71841704718417e-05, | |
| "loss": 1.2338, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1552511415525114, | |
| "grad_norm": 10.334748268127441, | |
| "learning_rate": 4.693049213597159e-05, | |
| "loss": 1.3393, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.1598173515981735, | |
| "grad_norm": 9.896265029907227, | |
| "learning_rate": 4.667681380010147e-05, | |
| "loss": 1.261, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1643835616438356, | |
| "grad_norm": 10.67037582397461, | |
| "learning_rate": 4.642313546423136e-05, | |
| "loss": 1.2365, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1689497716894977, | |
| "grad_norm": 7.820501327514648, | |
| "learning_rate": 4.616945712836124e-05, | |
| "loss": 1.0388, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.1735159817351598, | |
| "grad_norm": 8.09377670288086, | |
| "learning_rate": 4.591577879249112e-05, | |
| "loss": 1.0919, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.1780821917808219, | |
| "grad_norm": 9.110677719116211, | |
| "learning_rate": 4.5662100456621006e-05, | |
| "loss": 1.0683, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.182648401826484, | |
| "grad_norm": 11.42436695098877, | |
| "learning_rate": 4.5408422120750886e-05, | |
| "loss": 1.0649, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1872146118721461, | |
| "grad_norm": 9.24187183380127, | |
| "learning_rate": 4.5154743784880774e-05, | |
| "loss": 0.8407, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1917808219178082, | |
| "grad_norm": 8.836689949035645, | |
| "learning_rate": 4.4901065449010655e-05, | |
| "loss": 0.8219, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.1963470319634703, | |
| "grad_norm": 13.152447700500488, | |
| "learning_rate": 4.4647387113140535e-05, | |
| "loss": 1.064, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.2009132420091324, | |
| "grad_norm": 10.30240249633789, | |
| "learning_rate": 4.439370877727042e-05, | |
| "loss": 0.7285, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.2054794520547945, | |
| "grad_norm": 6.908230781555176, | |
| "learning_rate": 4.41400304414003e-05, | |
| "loss": 0.9826, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2100456621004566, | |
| "grad_norm": 6.0176920890808105, | |
| "learning_rate": 4.3886352105530184e-05, | |
| "loss": 0.7598, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.2146118721461187, | |
| "grad_norm": 6.647716999053955, | |
| "learning_rate": 4.363267376966007e-05, | |
| "loss": 0.8257, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.2191780821917808, | |
| "grad_norm": 6.749544620513916, | |
| "learning_rate": 4.337899543378995e-05, | |
| "loss": 0.8272, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2237442922374429, | |
| "grad_norm": 7.512877464294434, | |
| "learning_rate": 4.312531709791984e-05, | |
| "loss": 0.8007, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.228310502283105, | |
| "grad_norm": 6.311679363250732, | |
| "learning_rate": 4.287163876204972e-05, | |
| "loss": 0.7219, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2328767123287671, | |
| "grad_norm": 6.0948591232299805, | |
| "learning_rate": 4.26179604261796e-05, | |
| "loss": 0.5973, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.2374429223744292, | |
| "grad_norm": 9.038314819335938, | |
| "learning_rate": 4.236428209030949e-05, | |
| "loss": 0.784, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.2420091324200913, | |
| "grad_norm": 5.299352645874023, | |
| "learning_rate": 4.211060375443937e-05, | |
| "loss": 0.5438, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2465753424657534, | |
| "grad_norm": 7.578807353973389, | |
| "learning_rate": 4.1856925418569256e-05, | |
| "loss": 0.7382, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.2511415525114155, | |
| "grad_norm": 9.358942985534668, | |
| "learning_rate": 4.160324708269914e-05, | |
| "loss": 0.7584, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2557077625570776, | |
| "grad_norm": 9.18459701538086, | |
| "learning_rate": 4.134956874682902e-05, | |
| "loss": 0.8444, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.2602739726027397, | |
| "grad_norm": 5.982091426849365, | |
| "learning_rate": 4.1095890410958905e-05, | |
| "loss": 0.4002, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.2648401826484018, | |
| "grad_norm": 4.503027439117432, | |
| "learning_rate": 4.0842212075088786e-05, | |
| "loss": 0.5723, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.2694063926940639, | |
| "grad_norm": 7.2720794677734375, | |
| "learning_rate": 4.058853373921867e-05, | |
| "loss": 0.5846, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.273972602739726, | |
| "grad_norm": 6.161769390106201, | |
| "learning_rate": 4.0334855403348554e-05, | |
| "loss": 0.5559, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2785388127853881, | |
| "grad_norm": 8.040207862854004, | |
| "learning_rate": 4.0081177067478435e-05, | |
| "loss": 0.5126, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.2831050228310502, | |
| "grad_norm": 5.092132568359375, | |
| "learning_rate": 3.982749873160832e-05, | |
| "loss": 0.5201, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2876712328767123, | |
| "grad_norm": 9.829987525939941, | |
| "learning_rate": 3.95738203957382e-05, | |
| "loss": 0.4563, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.2922374429223744, | |
| "grad_norm": 7.451713562011719, | |
| "learning_rate": 3.932014205986809e-05, | |
| "loss": 0.7734, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2968036529680365, | |
| "grad_norm": 9.06767463684082, | |
| "learning_rate": 3.906646372399797e-05, | |
| "loss": 0.4621, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3013698630136986, | |
| "grad_norm": 8.293368339538574, | |
| "learning_rate": 3.881278538812785e-05, | |
| "loss": 0.5231, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.3059360730593607, | |
| "grad_norm": 9.901214599609375, | |
| "learning_rate": 3.855910705225774e-05, | |
| "loss": 0.6035, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.3105022831050228, | |
| "grad_norm": 10.380690574645996, | |
| "learning_rate": 3.830542871638762e-05, | |
| "loss": 0.3312, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3150684931506849, | |
| "grad_norm": 12.21066665649414, | |
| "learning_rate": 3.80517503805175e-05, | |
| "loss": 0.5626, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.319634703196347, | |
| "grad_norm": 8.75660514831543, | |
| "learning_rate": 3.779807204464739e-05, | |
| "loss": 0.5877, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3242009132420091, | |
| "grad_norm": 5.634203910827637, | |
| "learning_rate": 3.754439370877727e-05, | |
| "loss": 0.7606, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.3287671232876712, | |
| "grad_norm": 4.741566181182861, | |
| "learning_rate": 3.7290715372907156e-05, | |
| "loss": 0.5575, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 5.109163284301758, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.4544, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.3378995433789954, | |
| "grad_norm": 10.15185832977295, | |
| "learning_rate": 3.678335870116692e-05, | |
| "loss": 0.408, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3424657534246575, | |
| "grad_norm": 8.459287643432617, | |
| "learning_rate": 3.6529680365296805e-05, | |
| "loss": 0.518, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.3470319634703196, | |
| "grad_norm": 4.05381441116333, | |
| "learning_rate": 3.6276002029426685e-05, | |
| "loss": 0.5425, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3515981735159817, | |
| "grad_norm": 7.6227335929870605, | |
| "learning_rate": 3.602232369355657e-05, | |
| "loss": 0.4797, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.3561643835616438, | |
| "grad_norm": 4.937866687774658, | |
| "learning_rate": 3.5768645357686453e-05, | |
| "loss": 0.447, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.3607305936073059, | |
| "grad_norm": 6.663622856140137, | |
| "learning_rate": 3.5514967021816334e-05, | |
| "loss": 0.6484, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.365296803652968, | |
| "grad_norm": 2.132852792739868, | |
| "learning_rate": 3.526128868594622e-05, | |
| "loss": 0.5992, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3698630136986301, | |
| "grad_norm": 3.264267921447754, | |
| "learning_rate": 3.50076103500761e-05, | |
| "loss": 0.3265, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.3744292237442922, | |
| "grad_norm": 1.495285153388977, | |
| "learning_rate": 3.475393201420599e-05, | |
| "loss": 0.2921, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3789954337899543, | |
| "grad_norm": 1.2416194677352905, | |
| "learning_rate": 3.450025367833587e-05, | |
| "loss": 0.3851, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.3835616438356164, | |
| "grad_norm": 6.346595287322998, | |
| "learning_rate": 3.424657534246575e-05, | |
| "loss": 0.3337, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3881278538812785, | |
| "grad_norm": 3.725011110305786, | |
| "learning_rate": 3.399289700659564e-05, | |
| "loss": 0.3301, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3926940639269406, | |
| "grad_norm": 5.670325756072998, | |
| "learning_rate": 3.373921867072552e-05, | |
| "loss": 0.499, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.3972602739726027, | |
| "grad_norm": 7.0639424324035645, | |
| "learning_rate": 3.34855403348554e-05, | |
| "loss": 0.3419, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.4004566210045662, | |
| "eval_accuracy": 0.5540398740818469, | |
| "eval_loss": 1.5632456541061401, | |
| "eval_runtime": 168.0602, | |
| "eval_samples_per_second": 5.671, | |
| "eval_steps_per_second": 2.838, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.0013698630136987, | |
| "grad_norm": 2.454108953475952, | |
| "learning_rate": 3.323186199898529e-05, | |
| "loss": 0.334, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.0059360730593607, | |
| "grad_norm": 12.557369232177734, | |
| "learning_rate": 3.297818366311517e-05, | |
| "loss": 0.2568, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.0105022831050228, | |
| "grad_norm": 4.680070400238037, | |
| "learning_rate": 3.2724505327245055e-05, | |
| "loss": 0.2696, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.015068493150685, | |
| "grad_norm": 2.2027194499969482, | |
| "learning_rate": 3.2470826991374936e-05, | |
| "loss": 0.4776, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.019634703196347, | |
| "grad_norm": 8.404707908630371, | |
| "learning_rate": 3.221714865550482e-05, | |
| "loss": 0.2588, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.0242009132420091, | |
| "grad_norm": 1.3184863328933716, | |
| "learning_rate": 3.1963470319634704e-05, | |
| "loss": 0.2601, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0287671232876712, | |
| "grad_norm": 3.3989500999450684, | |
| "learning_rate": 3.1709791983764585e-05, | |
| "loss": 0.3734, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.0333333333333334, | |
| "grad_norm": 6.5863142013549805, | |
| "learning_rate": 3.145611364789447e-05, | |
| "loss": 0.4799, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0378995433789955, | |
| "grad_norm": 5.605319023132324, | |
| "learning_rate": 3.120243531202435e-05, | |
| "loss": 0.2629, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0424657534246575, | |
| "grad_norm": 5.677365779876709, | |
| "learning_rate": 3.0948756976154234e-05, | |
| "loss": 0.4172, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.0470319634703196, | |
| "grad_norm": 6.943356037139893, | |
| "learning_rate": 3.069507864028412e-05, | |
| "loss": 0.4699, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.0515981735159818, | |
| "grad_norm": 1.7512027025222778, | |
| "learning_rate": 3.0441400304414e-05, | |
| "loss": 0.1905, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.0561643835616439, | |
| "grad_norm": 3.9878628253936768, | |
| "learning_rate": 3.0187721968543886e-05, | |
| "loss": 0.4574, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.060730593607306, | |
| "grad_norm": 1.668309211730957, | |
| "learning_rate": 2.993404363267377e-05, | |
| "loss": 0.3441, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.065296803652968, | |
| "grad_norm": 14.312126159667969, | |
| "learning_rate": 2.9680365296803654e-05, | |
| "loss": 0.4114, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.0698630136986302, | |
| "grad_norm": 2.793670654296875, | |
| "learning_rate": 2.9426686960933534e-05, | |
| "loss": 0.292, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.0744292237442923, | |
| "grad_norm": 0.9054046273231506, | |
| "learning_rate": 2.917300862506342e-05, | |
| "loss": 0.2573, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.0789954337899543, | |
| "grad_norm": 3.951117515563965, | |
| "learning_rate": 2.8919330289193303e-05, | |
| "loss": 0.4457, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.0835616438356164, | |
| "grad_norm": 10.71568775177002, | |
| "learning_rate": 2.8665651953323187e-05, | |
| "loss": 0.2332, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.0881278538812786, | |
| "grad_norm": 5.735054969787598, | |
| "learning_rate": 2.841197361745307e-05, | |
| "loss": 0.2876, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.0926940639269407, | |
| "grad_norm": 5.521303176879883, | |
| "learning_rate": 2.815829528158295e-05, | |
| "loss": 0.3894, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.0972602739726027, | |
| "grad_norm": 10.683832168579102, | |
| "learning_rate": 2.7904616945712835e-05, | |
| "loss": 0.3966, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.1018264840182648, | |
| "grad_norm": 1.0494991540908813, | |
| "learning_rate": 2.765093860984272e-05, | |
| "loss": 0.3233, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.106392694063927, | |
| "grad_norm": 4.736073017120361, | |
| "learning_rate": 2.7397260273972603e-05, | |
| "loss": 0.3742, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.110958904109589, | |
| "grad_norm": 8.027792930603027, | |
| "learning_rate": 2.7143581938102484e-05, | |
| "loss": 0.3813, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.1155251141552511, | |
| "grad_norm": 2.795949935913086, | |
| "learning_rate": 2.6889903602232368e-05, | |
| "loss": 0.2194, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.1200913242009132, | |
| "grad_norm": 16.23459815979004, | |
| "learning_rate": 2.6636225266362252e-05, | |
| "loss": 0.4115, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.1246575342465754, | |
| "grad_norm": 8.320725440979004, | |
| "learning_rate": 2.6382546930492136e-05, | |
| "loss": 0.2755, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.1292237442922375, | |
| "grad_norm": 2.192859649658203, | |
| "learning_rate": 2.612886859462202e-05, | |
| "loss": 0.2373, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.1337899543378995, | |
| "grad_norm": 6.194462299346924, | |
| "learning_rate": 2.58751902587519e-05, | |
| "loss": 0.2048, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.1383561643835616, | |
| "grad_norm": 6.522688388824463, | |
| "learning_rate": 2.5621511922881785e-05, | |
| "loss": 0.2106, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.1429223744292236, | |
| "grad_norm": 7.976909160614014, | |
| "learning_rate": 2.536783358701167e-05, | |
| "loss": 0.1977, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.1474885844748859, | |
| "grad_norm": 10.373821258544922, | |
| "learning_rate": 2.5114155251141553e-05, | |
| "loss": 0.4609, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.152054794520548, | |
| "grad_norm": 8.701196670532227, | |
| "learning_rate": 2.4860476915271437e-05, | |
| "loss": 0.3093, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.15662100456621, | |
| "grad_norm": 0.8994976878166199, | |
| "learning_rate": 2.4606798579401318e-05, | |
| "loss": 0.1242, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.1611872146118722, | |
| "grad_norm": 4.23283576965332, | |
| "learning_rate": 2.4353120243531202e-05, | |
| "loss": 0.1627, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.1657534246575343, | |
| "grad_norm": 4.330513000488281, | |
| "learning_rate": 2.4099441907661086e-05, | |
| "loss": 0.3118, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.1703196347031963, | |
| "grad_norm": 2.304389715194702, | |
| "learning_rate": 2.384576357179097e-05, | |
| "loss": 0.218, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1748858447488584, | |
| "grad_norm": 6.340790271759033, | |
| "learning_rate": 2.359208523592085e-05, | |
| "loss": 0.2833, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.1794520547945206, | |
| "grad_norm": 2.6649558544158936, | |
| "learning_rate": 2.3338406900050735e-05, | |
| "loss": 0.2335, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.1840182648401827, | |
| "grad_norm": 0.8412348628044128, | |
| "learning_rate": 2.308472856418062e-05, | |
| "loss": 0.1889, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.1885844748858447, | |
| "grad_norm": 5.932245254516602, | |
| "learning_rate": 2.2831050228310503e-05, | |
| "loss": 0.2006, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.1931506849315068, | |
| "grad_norm": 9.092105865478516, | |
| "learning_rate": 2.2577371892440387e-05, | |
| "loss": 0.3411, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.197716894977169, | |
| "grad_norm": 6.397707939147949, | |
| "learning_rate": 2.2323693556570268e-05, | |
| "loss": 0.1739, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.202283105022831, | |
| "grad_norm": 3.637505292892456, | |
| "learning_rate": 2.207001522070015e-05, | |
| "loss": 0.2351, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.2068493150684931, | |
| "grad_norm": 3.8445942401885986, | |
| "learning_rate": 2.1816336884830036e-05, | |
| "loss": 0.3909, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.2114155251141552, | |
| "grad_norm": 1.331699252128601, | |
| "learning_rate": 2.156265854895992e-05, | |
| "loss": 0.154, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.2159817351598174, | |
| "grad_norm": 9.704405784606934, | |
| "learning_rate": 2.13089802130898e-05, | |
| "loss": 0.1631, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.2205479452054795, | |
| "grad_norm": 2.271298885345459, | |
| "learning_rate": 2.1055301877219685e-05, | |
| "loss": 0.1551, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.2251141552511415, | |
| "grad_norm": 10.894644737243652, | |
| "learning_rate": 2.080162354134957e-05, | |
| "loss": 0.3072, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.2296803652968036, | |
| "grad_norm": 10.65208625793457, | |
| "learning_rate": 2.0547945205479453e-05, | |
| "loss": 0.2142, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.2342465753424658, | |
| "grad_norm": 15.530343055725098, | |
| "learning_rate": 2.0294266869609337e-05, | |
| "loss": 0.2107, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.238812785388128, | |
| "grad_norm": 7.166269779205322, | |
| "learning_rate": 2.0040588533739217e-05, | |
| "loss": 0.2599, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.24337899543379, | |
| "grad_norm": 1.9615888595581055, | |
| "learning_rate": 1.97869101978691e-05, | |
| "loss": 0.1511, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.247945205479452, | |
| "grad_norm": 6.840981960296631, | |
| "learning_rate": 1.9533231861998985e-05, | |
| "loss": 0.3887, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.252511415525114, | |
| "grad_norm": 5.138172626495361, | |
| "learning_rate": 1.927955352612887e-05, | |
| "loss": 0.2284, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.2570776255707763, | |
| "grad_norm": 5.770719051361084, | |
| "learning_rate": 1.902587519025875e-05, | |
| "loss": 0.3119, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.2616438356164383, | |
| "grad_norm": 1.5914487838745117, | |
| "learning_rate": 1.8772196854388634e-05, | |
| "loss": 0.0873, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.2662100456621004, | |
| "grad_norm": 2.5335633754730225, | |
| "learning_rate": 1.8518518518518518e-05, | |
| "loss": 0.2629, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.2707762557077626, | |
| "grad_norm": 10.850088119506836, | |
| "learning_rate": 1.8264840182648402e-05, | |
| "loss": 0.3098, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.2753424657534247, | |
| "grad_norm": 2.9598915576934814, | |
| "learning_rate": 1.8011161846778286e-05, | |
| "loss": 0.0882, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.2799086757990867, | |
| "grad_norm": 2.021665096282959, | |
| "learning_rate": 1.7757483510908167e-05, | |
| "loss": 0.2079, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.2844748858447488, | |
| "grad_norm": 6.0909833908081055, | |
| "learning_rate": 1.750380517503805e-05, | |
| "loss": 0.1287, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.2890410958904108, | |
| "grad_norm": 7.84592866897583, | |
| "learning_rate": 1.7250126839167935e-05, | |
| "loss": 0.1406, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.293607305936073, | |
| "grad_norm": 3.269815444946289, | |
| "learning_rate": 1.699644850329782e-05, | |
| "loss": 0.2997, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.2981735159817351, | |
| "grad_norm": 0.7504135966300964, | |
| "learning_rate": 1.67427701674277e-05, | |
| "loss": 0.1724, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.3027397260273972, | |
| "grad_norm": 2.8693363666534424, | |
| "learning_rate": 1.6489091831557584e-05, | |
| "loss": 0.1543, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.3073059360730594, | |
| "grad_norm": 6.736435890197754, | |
| "learning_rate": 1.6235413495687468e-05, | |
| "loss": 0.1396, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.3118721461187215, | |
| "grad_norm": 0.3472974896430969, | |
| "learning_rate": 1.5981735159817352e-05, | |
| "loss": 0.2981, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.3164383561643835, | |
| "grad_norm": 4.580450057983398, | |
| "learning_rate": 1.5728056823947236e-05, | |
| "loss": 0.2362, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.3210045662100456, | |
| "grad_norm": 1.7592848539352417, | |
| "learning_rate": 1.5474378488077117e-05, | |
| "loss": 0.2429, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.3255707762557076, | |
| "grad_norm": 1.0239814519882202, | |
| "learning_rate": 1.5220700152207e-05, | |
| "loss": 0.1309, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.33013698630137, | |
| "grad_norm": 1.68032705783844, | |
| "learning_rate": 1.4967021816336885e-05, | |
| "loss": 0.2172, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.334703196347032, | |
| "grad_norm": 5.960744380950928, | |
| "learning_rate": 1.4713343480466767e-05, | |
| "loss": 0.2709, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.339269406392694, | |
| "grad_norm": 9.478919982910156, | |
| "learning_rate": 1.4459665144596651e-05, | |
| "loss": 0.2803, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.3438356164383563, | |
| "grad_norm": 0.2718620002269745, | |
| "learning_rate": 1.4205986808726535e-05, | |
| "loss": 0.1481, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.3484018264840183, | |
| "grad_norm": 0.5140727162361145, | |
| "learning_rate": 1.3952308472856418e-05, | |
| "loss": 0.1621, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.3529680365296803, | |
| "grad_norm": 8.044154167175293, | |
| "learning_rate": 1.3698630136986302e-05, | |
| "loss": 0.2557, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.3575342465753424, | |
| "grad_norm": 0.42045390605926514, | |
| "learning_rate": 1.3444951801116184e-05, | |
| "loss": 0.1946, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.3621004566210044, | |
| "grad_norm": 1.3292018175125122, | |
| "learning_rate": 1.3191273465246068e-05, | |
| "loss": 0.1813, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.3666666666666667, | |
| "grad_norm": 0.8583564758300781, | |
| "learning_rate": 1.293759512937595e-05, | |
| "loss": 0.1753, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.3712328767123287, | |
| "grad_norm": 5.7116546630859375, | |
| "learning_rate": 1.2683916793505835e-05, | |
| "loss": 0.2727, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.3757990867579908, | |
| "grad_norm": 0.348532497882843, | |
| "learning_rate": 1.2430238457635719e-05, | |
| "loss": 0.1961, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.380365296803653, | |
| "grad_norm": 9.204560279846191, | |
| "learning_rate": 1.2176560121765601e-05, | |
| "loss": 0.257, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.384931506849315, | |
| "grad_norm": 5.398329257965088, | |
| "learning_rate": 1.1922881785895485e-05, | |
| "loss": 0.1348, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.3894977168949771, | |
| "grad_norm": 0.43983083963394165, | |
| "learning_rate": 1.1669203450025367e-05, | |
| "loss": 0.2507, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.3940639269406392, | |
| "grad_norm": 0.9157253503799438, | |
| "learning_rate": 1.1415525114155251e-05, | |
| "loss": 0.193, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.3986301369863012, | |
| "grad_norm": 0.855165421962738, | |
| "learning_rate": 1.1161846778285134e-05, | |
| "loss": 0.1291, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.4004566210045661, | |
| "eval_accuracy": 0.6516264428121721, | |
| "eval_loss": 1.277839183807373, | |
| "eval_runtime": 178.2589, | |
| "eval_samples_per_second": 5.346, | |
| "eval_steps_per_second": 2.676, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 2.0027397260273974, | |
| "grad_norm": 8.00808334350586, | |
| "learning_rate": 1.0908168442415018e-05, | |
| "loss": 0.1597, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.007305936073059, | |
| "grad_norm": 0.1681767702102661, | |
| "learning_rate": 1.06544901065449e-05, | |
| "loss": 0.0838, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.0118721461187214, | |
| "grad_norm": 12.16120719909668, | |
| "learning_rate": 1.0400811770674784e-05, | |
| "loss": 0.1621, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.0164383561643837, | |
| "grad_norm": 1.8593238592147827, | |
| "learning_rate": 1.0147133434804668e-05, | |
| "loss": 0.0698, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.0210045662100455, | |
| "grad_norm": 1.4579780101776123, | |
| "learning_rate": 9.89345509893455e-06, | |
| "loss": 0.2562, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.025570776255708, | |
| "grad_norm": 2.344836711883545, | |
| "learning_rate": 9.639776763064435e-06, | |
| "loss": 0.257, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.03013698630137, | |
| "grad_norm": 8.14854621887207, | |
| "learning_rate": 9.386098427194317e-06, | |
| "loss": 0.2149, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.034703196347032, | |
| "grad_norm": 11.254741668701172, | |
| "learning_rate": 9.132420091324201e-06, | |
| "loss": 0.2175, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.039269406392694, | |
| "grad_norm": 8.984929084777832, | |
| "learning_rate": 8.878741755454084e-06, | |
| "loss": 0.1927, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.043835616438356, | |
| "grad_norm": 13.033331871032715, | |
| "learning_rate": 8.625063419583968e-06, | |
| "loss": 0.2202, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.0484018264840183, | |
| "grad_norm": 1.0010125637054443, | |
| "learning_rate": 8.37138508371385e-06, | |
| "loss": 0.1063, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.0529680365296805, | |
| "grad_norm": 0.2957708239555359, | |
| "learning_rate": 8.117706747843734e-06, | |
| "loss": 0.1509, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.0575342465753423, | |
| "grad_norm": 0.20592088997364044, | |
| "learning_rate": 7.864028411973618e-06, | |
| "loss": 0.1519, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.0621004566210046, | |
| "grad_norm": 2.5575947761535645, | |
| "learning_rate": 7.6103500761035e-06, | |
| "loss": 0.1487, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.066666666666667, | |
| "grad_norm": 3.2097675800323486, | |
| "learning_rate": 7.356671740233384e-06, | |
| "loss": 0.1506, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.0712328767123287, | |
| "grad_norm": 7.255710124969482, | |
| "learning_rate": 7.102993404363268e-06, | |
| "loss": 0.2091, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.075799086757991, | |
| "grad_norm": 0.6733071804046631, | |
| "learning_rate": 6.849315068493151e-06, | |
| "loss": 0.1179, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.080365296803653, | |
| "grad_norm": 5.307835102081299, | |
| "learning_rate": 6.595636732623034e-06, | |
| "loss": 0.2213, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.084931506849315, | |
| "grad_norm": 2.307908535003662, | |
| "learning_rate": 6.341958396752917e-06, | |
| "loss": 0.0695, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.0894977168949773, | |
| "grad_norm": 9.304193496704102, | |
| "learning_rate": 6.0882800608828005e-06, | |
| "loss": 0.13, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.094063926940639, | |
| "grad_norm": 0.5198363661766052, | |
| "learning_rate": 5.834601725012684e-06, | |
| "loss": 0.1376, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.0986301369863014, | |
| "grad_norm": 4.95096492767334, | |
| "learning_rate": 5.580923389142567e-06, | |
| "loss": 0.1322, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.1031963470319637, | |
| "grad_norm": 4.969466686248779, | |
| "learning_rate": 5.32724505327245e-06, | |
| "loss": 0.1839, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.1077625570776255, | |
| "grad_norm": 0.14146043360233307, | |
| "learning_rate": 5.073566717402334e-06, | |
| "loss": 0.2181, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.1123287671232878, | |
| "grad_norm": 0.31362447142601013, | |
| "learning_rate": 4.819888381532217e-06, | |
| "loss": 0.2004, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.1168949771689496, | |
| "grad_norm": 9.072457313537598, | |
| "learning_rate": 4.566210045662101e-06, | |
| "loss": 0.158, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.121461187214612, | |
| "grad_norm": 6.152866363525391, | |
| "learning_rate": 4.312531709791984e-06, | |
| "loss": 0.1529, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.126027397260274, | |
| "grad_norm": 2.435375452041626, | |
| "learning_rate": 4.058853373921867e-06, | |
| "loss": 0.172, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.130593607305936, | |
| "grad_norm": 7.807186603546143, | |
| "learning_rate": 3.80517503805175e-06, | |
| "loss": 0.1801, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.135159817351598, | |
| "grad_norm": 2.6992197036743164, | |
| "learning_rate": 3.551496702181634e-06, | |
| "loss": 0.1777, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.1397260273972605, | |
| "grad_norm": 9.397564888000488, | |
| "learning_rate": 3.297818366311517e-06, | |
| "loss": 0.0939, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.1442922374429223, | |
| "grad_norm": 3.4112205505371094, | |
| "learning_rate": 3.0441400304414002e-06, | |
| "loss": 0.1201, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.1488584474885846, | |
| "grad_norm": 0.36758846044540405, | |
| "learning_rate": 2.7904616945712835e-06, | |
| "loss": 0.167, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.1534246575342464, | |
| "grad_norm": 0.20398499071598053, | |
| "learning_rate": 2.536783358701167e-06, | |
| "loss": 0.0453, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.1579908675799087, | |
| "grad_norm": 0.14668098092079163, | |
| "learning_rate": 2.2831050228310503e-06, | |
| "loss": 0.0674, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.162557077625571, | |
| "grad_norm": 3.2575571537017822, | |
| "learning_rate": 2.0294266869609335e-06, | |
| "loss": 0.0917, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.1671232876712327, | |
| "grad_norm": 0.5656366348266602, | |
| "learning_rate": 1.775748351090817e-06, | |
| "loss": 0.1194, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.171689497716895, | |
| "grad_norm": 7.743866443634033, | |
| "learning_rate": 1.5220700152207001e-06, | |
| "loss": 0.1199, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.1762557077625573, | |
| "grad_norm": 1.3446861505508423, | |
| "learning_rate": 1.2683916793505835e-06, | |
| "loss": 0.0521, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.180821917808219, | |
| "grad_norm": 0.33492910861968994, | |
| "learning_rate": 1.0147133434804667e-06, | |
| "loss": 0.0444, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.1853881278538814, | |
| "grad_norm": 0.12891508638858795, | |
| "learning_rate": 7.610350076103501e-07, | |
| "loss": 0.1172, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.189954337899543, | |
| "grad_norm": 5.329843521118164, | |
| "learning_rate": 5.073566717402334e-07, | |
| "loss": 0.1813, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.1945205479452055, | |
| "grad_norm": 2.9594337940216064, | |
| "learning_rate": 2.536783358701167e-07, | |
| "loss": 0.2105, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.1990867579908677, | |
| "grad_norm": 1.7173917293548584, | |
| "learning_rate": 0.0, | |
| "loss": 0.183, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.1990867579908677, | |
| "eval_accuracy": 0.6295907660020986, | |
| "eval_loss": 1.3431484699249268, | |
| "eval_runtime": 185.0677, | |
| "eval_samples_per_second": 5.149, | |
| "eval_steps_per_second": 2.577, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.1990867579908677, | |
| "step": 2190, | |
| "total_flos": 1.5353993990024331e+19, | |
| "train_loss": 0.7283184543456117, | |
| "train_runtime": 5503.9878, | |
| "train_samples_per_second": 3.183, | |
| "train_steps_per_second": 0.398 | |
| }, | |
| { | |
| "epoch": 2.1990867579908677, | |
| "eval_accuracy": 0.6516264428121721, | |
| "eval_loss": 1.277839183807373, | |
| "eval_runtime": 188.0153, | |
| "eval_samples_per_second": 5.069, | |
| "eval_steps_per_second": 2.537, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.1990867579908677, | |
| "eval_accuracy": 0.7749803304484658, | |
| "eval_loss": 0.9055941104888916, | |
| "eval_runtime": 256.8953, | |
| "eval_samples_per_second": 4.948, | |
| "eval_steps_per_second": 2.476, | |
| "step": 2190 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2190, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 2, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.5353993990024331e+19, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |