diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,47686 @@ +{ + "best_metric": 0.776286930718617, + "best_model_checkpoint": "videomae-base-finetuned-ucf101-epoch20/checkpoint-79120", + "epoch": 19.05, + "global_step": 79120, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 6.319514661274014e-08, + "loss": 4.8137, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 1.263902932254803e-07, + "loss": 4.5871, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 1.8958543983822043e-07, + "loss": 4.6141, + "step": 30 + }, + { + "epoch": 0.0, + "learning_rate": 2.527805864509606e-07, + "loss": 4.7492, + "step": 40 + }, + { + "epoch": 0.0, + "learning_rate": 3.159757330637007e-07, + "loss": 4.523, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 3.7917087967644087e-07, + "loss": 4.6734, + "step": 60 + }, + { + "epoch": 0.0, + "learning_rate": 4.42366026289181e-07, + "loss": 4.5762, + "step": 70 + }, + { + "epoch": 0.0, + "learning_rate": 5.055611729019212e-07, + "loss": 4.5658, + "step": 80 + }, + { + "epoch": 0.0, + "learning_rate": 5.687563195146614e-07, + "loss": 4.5506, + "step": 90 + }, + { + "epoch": 0.0, + "learning_rate": 6.319514661274014e-07, + "loss": 4.7235, + "step": 100 + }, + { + "epoch": 0.0, + "learning_rate": 6.951466127401416e-07, + "loss": 4.8277, + "step": 110 + }, + { + "epoch": 0.0, + "learning_rate": 7.583417593528817e-07, + "loss": 4.7116, + "step": 120 + }, + { + "epoch": 0.0, + "learning_rate": 8.215369059656218e-07, + "loss": 4.6663, + "step": 130 + }, + { + "epoch": 0.0, + "learning_rate": 8.84732052578362e-07, + "loss": 4.6786, + "step": 140 + }, + { + "epoch": 0.0, + "learning_rate": 9.479271991911022e-07, + "loss": 4.7358, + "step": 150 + }, + { + "epoch": 0.0, + "learning_rate": 1.0111223458038423e-06, + "loss": 4.6732, + "step": 160 + }, + { + "epoch": 0.0, + "learning_rate": 1.0743174924165826e-06, + "loss": 4.5606, + "step": 170 + }, + { + "epoch": 0.0, + "learning_rate": 1.1375126390293227e-06, + "loss": 4.6964, + "step": 180 + }, + { + "epoch": 0.0, + "learning_rate": 1.2007077856420628e-06, + "loss": 4.651, + "step": 190 + }, + { + "epoch": 0.0, + "learning_rate": 1.2639029322548029e-06, + "loss": 4.6526, + "step": 200 + }, + { + "epoch": 0.0, + "learning_rate": 1.3270980788675432e-06, + "loss": 4.6003, + "step": 210 + }, + { + "epoch": 0.0, + "learning_rate": 1.3902932254802833e-06, + "loss": 4.6634, + "step": 220 + }, + { + "epoch": 0.0, + "learning_rate": 1.4534883720930234e-06, + "loss": 4.7354, + "step": 230 + }, + { + "epoch": 0.0, + "learning_rate": 1.5166835187057635e-06, + "loss": 4.6788, + "step": 240 + }, + { + "epoch": 0.0, + "learning_rate": 1.5798786653185038e-06, + "loss": 4.678, + "step": 250 + }, + { + "epoch": 0.0, + "learning_rate": 1.6430738119312437e-06, + "loss": 4.5274, + "step": 260 + }, + { + "epoch": 0.0, + "learning_rate": 1.706268958543984e-06, + "loss": 4.7338, + "step": 270 + }, + { + "epoch": 0.0, + "learning_rate": 1.769464105156724e-06, + "loss": 4.7868, + "step": 280 + }, + { + "epoch": 0.0, + "learning_rate": 1.8326592517694641e-06, + "loss": 4.6697, + "step": 290 + }, + { + "epoch": 0.0, + "learning_rate": 1.8958543983822044e-06, + "loss": 4.5723, + "step": 300 + }, + { + "epoch": 0.0, + "learning_rate": 1.9590495449949447e-06, + "loss": 4.7154, + "step": 310 + }, + { + "epoch": 0.0, + "learning_rate": 2.0222446916076846e-06, + "loss": 4.7551, + "step": 320 + }, + { + "epoch": 0.0, + "learning_rate": 2.085439838220425e-06, + "loss": 4.5266, + "step": 330 + }, + { + "epoch": 0.0, + "learning_rate": 2.1486349848331652e-06, + "loss": 4.7777, + "step": 340 + }, + { + "epoch": 0.0, + "learning_rate": 2.211830131445905e-06, + "loss": 4.6002, + "step": 350 + }, + { + "epoch": 0.0, + "learning_rate": 2.2750252780586454e-06, + "loss": 4.6696, + "step": 360 + }, + { + "epoch": 0.0, + "learning_rate": 2.3382204246713853e-06, + "loss": 4.727, + "step": 370 + }, + { + "epoch": 0.0, + "learning_rate": 2.4014155712841256e-06, + "loss": 4.6541, + "step": 380 + }, + { + "epoch": 0.0, + "learning_rate": 2.464610717896866e-06, + "loss": 4.6874, + "step": 390 + }, + { + "epoch": 0.01, + "learning_rate": 2.5278058645096058e-06, + "loss": 4.5731, + "step": 400 + }, + { + "epoch": 0.01, + "learning_rate": 2.591001011122346e-06, + "loss": 4.7717, + "step": 410 + }, + { + "epoch": 0.01, + "learning_rate": 2.6541961577350864e-06, + "loss": 4.5935, + "step": 420 + }, + { + "epoch": 0.01, + "learning_rate": 2.7173913043478263e-06, + "loss": 4.6032, + "step": 430 + }, + { + "epoch": 0.01, + "learning_rate": 2.7805864509605666e-06, + "loss": 4.7271, + "step": 440 + }, + { + "epoch": 0.01, + "learning_rate": 2.8437815975733065e-06, + "loss": 4.6907, + "step": 450 + }, + { + "epoch": 0.01, + "learning_rate": 2.9069767441860468e-06, + "loss": 4.6249, + "step": 460 + }, + { + "epoch": 0.01, + "learning_rate": 2.970171890798787e-06, + "loss": 4.5562, + "step": 470 + }, + { + "epoch": 0.01, + "learning_rate": 3.033367037411527e-06, + "loss": 4.5678, + "step": 480 + }, + { + "epoch": 0.01, + "learning_rate": 3.0965621840242672e-06, + "loss": 4.7829, + "step": 490 + }, + { + "epoch": 0.01, + "learning_rate": 3.1597573306370075e-06, + "loss": 4.8224, + "step": 500 + }, + { + "epoch": 0.01, + "learning_rate": 3.2229524772497474e-06, + "loss": 4.7253, + "step": 510 + }, + { + "epoch": 0.01, + "learning_rate": 3.2861476238624873e-06, + "loss": 4.6637, + "step": 520 + }, + { + "epoch": 0.01, + "learning_rate": 3.3493427704752276e-06, + "loss": 4.6635, + "step": 530 + }, + { + "epoch": 0.01, + "learning_rate": 3.412537917087968e-06, + "loss": 4.729, + "step": 540 + }, + { + "epoch": 0.01, + "learning_rate": 3.475733063700708e-06, + "loss": 4.6745, + "step": 550 + }, + { + "epoch": 0.01, + "learning_rate": 3.538928210313448e-06, + "loss": 4.4553, + "step": 560 + }, + { + "epoch": 0.01, + "learning_rate": 3.6021233569261884e-06, + "loss": 4.6859, + "step": 570 + }, + { + "epoch": 0.01, + "learning_rate": 3.6653185035389283e-06, + "loss": 4.7117, + "step": 580 + }, + { + "epoch": 0.01, + "learning_rate": 3.7285136501516686e-06, + "loss": 4.6074, + "step": 590 + }, + { + "epoch": 0.01, + "learning_rate": 3.791708796764409e-06, + "loss": 4.6138, + "step": 600 + }, + { + "epoch": 0.01, + "learning_rate": 3.854903943377149e-06, + "loss": 4.5707, + "step": 610 + }, + { + "epoch": 0.01, + "learning_rate": 3.9180990899898895e-06, + "loss": 4.6139, + "step": 620 + }, + { + "epoch": 0.01, + "learning_rate": 3.981294236602629e-06, + "loss": 4.6674, + "step": 630 + }, + { + "epoch": 0.01, + "learning_rate": 4.044489383215369e-06, + "loss": 4.5383, + "step": 640 + }, + { + "epoch": 0.01, + "learning_rate": 4.10768452982811e-06, + "loss": 4.6174, + "step": 650 + }, + { + "epoch": 0.01, + "learning_rate": 4.17087967644085e-06, + "loss": 4.6462, + "step": 660 + }, + { + "epoch": 0.01, + "learning_rate": 4.23407482305359e-06, + "loss": 4.5949, + "step": 670 + }, + { + "epoch": 0.01, + "learning_rate": 4.2972699696663305e-06, + "loss": 4.4447, + "step": 680 + }, + { + "epoch": 0.01, + "learning_rate": 4.36046511627907e-06, + "loss": 4.6029, + "step": 690 + }, + { + "epoch": 0.01, + "learning_rate": 4.42366026289181e-06, + "loss": 4.6738, + "step": 700 + }, + { + "epoch": 0.01, + "learning_rate": 4.48685540950455e-06, + "loss": 4.645, + "step": 710 + }, + { + "epoch": 0.01, + "learning_rate": 4.550050556117291e-06, + "loss": 4.6181, + "step": 720 + }, + { + "epoch": 0.01, + "learning_rate": 4.613245702730031e-06, + "loss": 4.6318, + "step": 730 + }, + { + "epoch": 0.01, + "learning_rate": 4.676440849342771e-06, + "loss": 4.6973, + "step": 740 + }, + { + "epoch": 0.01, + "learning_rate": 4.739635995955511e-06, + "loss": 4.5497, + "step": 750 + }, + { + "epoch": 0.01, + "learning_rate": 4.802831142568251e-06, + "loss": 4.6404, + "step": 760 + }, + { + "epoch": 0.01, + "learning_rate": 4.866026289180991e-06, + "loss": 4.5752, + "step": 770 + }, + { + "epoch": 0.01, + "learning_rate": 4.929221435793732e-06, + "loss": 4.6471, + "step": 780 + }, + { + "epoch": 0.01, + "learning_rate": 4.992416582406472e-06, + "loss": 4.6019, + "step": 790 + }, + { + "epoch": 0.01, + "learning_rate": 5.0556117290192116e-06, + "loss": 4.58, + "step": 800 + }, + { + "epoch": 0.01, + "learning_rate": 5.118806875631952e-06, + "loss": 4.5285, + "step": 810 + }, + { + "epoch": 0.01, + "learning_rate": 5.182002022244692e-06, + "loss": 4.6475, + "step": 820 + }, + { + "epoch": 0.01, + "learning_rate": 5.245197168857432e-06, + "loss": 4.5705, + "step": 830 + }, + { + "epoch": 0.01, + "learning_rate": 5.308392315470173e-06, + "loss": 4.5052, + "step": 840 + }, + { + "epoch": 0.01, + "learning_rate": 5.371587462082913e-06, + "loss": 4.7354, + "step": 850 + }, + { + "epoch": 0.01, + "learning_rate": 5.4347826086956525e-06, + "loss": 4.5762, + "step": 860 + }, + { + "epoch": 0.01, + "learning_rate": 5.497977755308393e-06, + "loss": 4.6558, + "step": 870 + }, + { + "epoch": 0.01, + "learning_rate": 5.561172901921133e-06, + "loss": 4.6685, + "step": 880 + }, + { + "epoch": 0.01, + "learning_rate": 5.624368048533873e-06, + "loss": 4.6789, + "step": 890 + }, + { + "epoch": 0.01, + "learning_rate": 5.687563195146613e-06, + "loss": 4.5414, + "step": 900 + }, + { + "epoch": 0.01, + "learning_rate": 5.750758341759354e-06, + "loss": 4.6925, + "step": 910 + }, + { + "epoch": 0.01, + "learning_rate": 5.8139534883720935e-06, + "loss": 4.5305, + "step": 920 + }, + { + "epoch": 0.01, + "learning_rate": 5.877148634984833e-06, + "loss": 4.6553, + "step": 930 + }, + { + "epoch": 0.01, + "learning_rate": 5.940343781597574e-06, + "loss": 4.5568, + "step": 940 + }, + { + "epoch": 0.01, + "learning_rate": 6.003538928210314e-06, + "loss": 4.5699, + "step": 950 + }, + { + "epoch": 0.01, + "learning_rate": 6.066734074823054e-06, + "loss": 4.5284, + "step": 960 + }, + { + "epoch": 0.01, + "learning_rate": 6.129929221435795e-06, + "loss": 4.5164, + "step": 970 + }, + { + "epoch": 0.01, + "learning_rate": 6.1931243680485345e-06, + "loss": 4.5356, + "step": 980 + }, + { + "epoch": 0.01, + "learning_rate": 6.256319514661274e-06, + "loss": 4.542, + "step": 990 + }, + { + "epoch": 0.01, + "learning_rate": 6.319514661274015e-06, + "loss": 4.4818, + "step": 1000 + }, + { + "epoch": 0.01, + "learning_rate": 6.382709807886754e-06, + "loss": 4.522, + "step": 1010 + }, + { + "epoch": 0.01, + "learning_rate": 6.445904954499495e-06, + "loss": 4.5998, + "step": 1020 + }, + { + "epoch": 0.01, + "learning_rate": 6.509100101112236e-06, + "loss": 4.4969, + "step": 1030 + }, + { + "epoch": 0.01, + "learning_rate": 6.572295247724975e-06, + "loss": 4.5777, + "step": 1040 + }, + { + "epoch": 0.01, + "learning_rate": 6.635490394337715e-06, + "loss": 4.5045, + "step": 1050 + }, + { + "epoch": 0.01, + "learning_rate": 6.698685540950455e-06, + "loss": 4.5134, + "step": 1060 + }, + { + "epoch": 0.01, + "learning_rate": 6.761880687563195e-06, + "loss": 4.4869, + "step": 1070 + }, + { + "epoch": 0.01, + "learning_rate": 6.825075834175936e-06, + "loss": 4.4444, + "step": 1080 + }, + { + "epoch": 0.01, + "learning_rate": 6.888270980788676e-06, + "loss": 4.6518, + "step": 1090 + }, + { + "epoch": 0.01, + "learning_rate": 6.951466127401416e-06, + "loss": 4.4599, + "step": 1100 + }, + { + "epoch": 0.01, + "learning_rate": 7.014661274014156e-06, + "loss": 4.627, + "step": 1110 + }, + { + "epoch": 0.01, + "learning_rate": 7.077856420626896e-06, + "loss": 4.3437, + "step": 1120 + }, + { + "epoch": 0.01, + "learning_rate": 7.141051567239636e-06, + "loss": 4.497, + "step": 1130 + }, + { + "epoch": 0.01, + "learning_rate": 7.204246713852377e-06, + "loss": 4.5048, + "step": 1140 + }, + { + "epoch": 0.01, + "learning_rate": 7.267441860465117e-06, + "loss": 4.5502, + "step": 1150 + }, + { + "epoch": 0.01, + "learning_rate": 7.3306370070778566e-06, + "loss": 4.5298, + "step": 1160 + }, + { + "epoch": 0.01, + "learning_rate": 7.393832153690597e-06, + "loss": 4.4805, + "step": 1170 + }, + { + "epoch": 0.01, + "learning_rate": 7.457027300303337e-06, + "loss": 4.5296, + "step": 1180 + }, + { + "epoch": 0.02, + "learning_rate": 7.520222446916077e-06, + "loss": 4.4115, + "step": 1190 + }, + { + "epoch": 0.02, + "learning_rate": 7.583417593528818e-06, + "loss": 4.4393, + "step": 1200 + }, + { + "epoch": 0.02, + "learning_rate": 7.646612740141559e-06, + "loss": 4.5296, + "step": 1210 + }, + { + "epoch": 0.02, + "learning_rate": 7.709807886754298e-06, + "loss": 4.5208, + "step": 1220 + }, + { + "epoch": 0.02, + "learning_rate": 7.773003033367038e-06, + "loss": 4.4125, + "step": 1230 + }, + { + "epoch": 0.02, + "learning_rate": 7.836198179979779e-06, + "loss": 4.497, + "step": 1240 + }, + { + "epoch": 0.02, + "learning_rate": 7.899393326592518e-06, + "loss": 4.4995, + "step": 1250 + }, + { + "epoch": 0.02, + "learning_rate": 7.962588473205259e-06, + "loss": 4.448, + "step": 1260 + }, + { + "epoch": 0.02, + "learning_rate": 8.025783619818e-06, + "loss": 4.3839, + "step": 1270 + }, + { + "epoch": 0.02, + "learning_rate": 8.088978766430739e-06, + "loss": 4.4644, + "step": 1280 + }, + { + "epoch": 0.02, + "learning_rate": 8.15217391304348e-06, + "loss": 4.6661, + "step": 1290 + }, + { + "epoch": 0.02, + "learning_rate": 8.21536905965622e-06, + "loss": 4.3639, + "step": 1300 + }, + { + "epoch": 0.02, + "learning_rate": 8.278564206268959e-06, + "loss": 4.313, + "step": 1310 + }, + { + "epoch": 0.02, + "learning_rate": 8.3417593528817e-06, + "loss": 4.2005, + "step": 1320 + }, + { + "epoch": 0.02, + "learning_rate": 8.40495449949444e-06, + "loss": 4.3866, + "step": 1330 + }, + { + "epoch": 0.02, + "learning_rate": 8.46814964610718e-06, + "loss": 4.472, + "step": 1340 + }, + { + "epoch": 0.02, + "learning_rate": 8.53134479271992e-06, + "loss": 4.5965, + "step": 1350 + }, + { + "epoch": 0.02, + "learning_rate": 8.594539939332661e-06, + "loss": 4.2823, + "step": 1360 + }, + { + "epoch": 0.02, + "learning_rate": 8.6577350859454e-06, + "loss": 4.3901, + "step": 1370 + }, + { + "epoch": 0.02, + "learning_rate": 8.72093023255814e-06, + "loss": 4.2836, + "step": 1380 + }, + { + "epoch": 0.02, + "learning_rate": 8.78412537917088e-06, + "loss": 4.4204, + "step": 1390 + }, + { + "epoch": 0.02, + "learning_rate": 8.84732052578362e-06, + "loss": 4.3263, + "step": 1400 + }, + { + "epoch": 0.02, + "learning_rate": 8.910515672396361e-06, + "loss": 4.5639, + "step": 1410 + }, + { + "epoch": 0.02, + "learning_rate": 8.9737108190091e-06, + "loss": 4.3572, + "step": 1420 + }, + { + "epoch": 0.02, + "learning_rate": 9.036905965621841e-06, + "loss": 4.4036, + "step": 1430 + }, + { + "epoch": 0.02, + "learning_rate": 9.100101112234582e-06, + "loss": 4.3568, + "step": 1440 + }, + { + "epoch": 0.02, + "learning_rate": 9.16329625884732e-06, + "loss": 4.2013, + "step": 1450 + }, + { + "epoch": 0.02, + "learning_rate": 9.226491405460061e-06, + "loss": 4.3776, + "step": 1460 + }, + { + "epoch": 0.02, + "learning_rate": 9.289686552072802e-06, + "loss": 4.3542, + "step": 1470 + }, + { + "epoch": 0.02, + "learning_rate": 9.352881698685541e-06, + "loss": 4.3279, + "step": 1480 + }, + { + "epoch": 0.02, + "learning_rate": 9.416076845298282e-06, + "loss": 4.2633, + "step": 1490 + }, + { + "epoch": 0.02, + "learning_rate": 9.479271991911023e-06, + "loss": 4.3342, + "step": 1500 + }, + { + "epoch": 0.02, + "learning_rate": 9.542467138523762e-06, + "loss": 4.3596, + "step": 1510 + }, + { + "epoch": 0.02, + "learning_rate": 9.605662285136502e-06, + "loss": 4.2599, + "step": 1520 + }, + { + "epoch": 0.02, + "learning_rate": 9.668857431749243e-06, + "loss": 4.2129, + "step": 1530 + }, + { + "epoch": 0.02, + "learning_rate": 9.732052578361982e-06, + "loss": 4.3694, + "step": 1540 + }, + { + "epoch": 0.02, + "learning_rate": 9.795247724974723e-06, + "loss": 4.3714, + "step": 1550 + }, + { + "epoch": 0.02, + "learning_rate": 9.858442871587464e-06, + "loss": 4.2316, + "step": 1560 + }, + { + "epoch": 0.02, + "learning_rate": 9.921638018200203e-06, + "loss": 4.2767, + "step": 1570 + }, + { + "epoch": 0.02, + "learning_rate": 9.984833164812943e-06, + "loss": 4.2708, + "step": 1580 + }, + { + "epoch": 0.02, + "learning_rate": 1.0048028311425684e-05, + "loss": 4.4327, + "step": 1590 + }, + { + "epoch": 0.02, + "learning_rate": 1.0111223458038423e-05, + "loss": 4.0266, + "step": 1600 + }, + { + "epoch": 0.02, + "learning_rate": 1.0174418604651164e-05, + "loss": 4.2784, + "step": 1610 + }, + { + "epoch": 0.02, + "learning_rate": 1.0237613751263905e-05, + "loss": 4.4782, + "step": 1620 + }, + { + "epoch": 0.02, + "learning_rate": 1.0300808897876644e-05, + "loss": 4.0133, + "step": 1630 + }, + { + "epoch": 0.02, + "learning_rate": 1.0364004044489384e-05, + "loss": 4.1972, + "step": 1640 + }, + { + "epoch": 0.02, + "learning_rate": 1.0427199191102125e-05, + "loss": 4.3027, + "step": 1650 + }, + { + "epoch": 0.02, + "learning_rate": 1.0490394337714864e-05, + "loss": 4.3939, + "step": 1660 + }, + { + "epoch": 0.02, + "learning_rate": 1.0553589484327605e-05, + "loss": 4.0384, + "step": 1670 + }, + { + "epoch": 0.02, + "learning_rate": 1.0616784630940346e-05, + "loss": 4.205, + "step": 1680 + }, + { + "epoch": 0.02, + "learning_rate": 1.0679979777553085e-05, + "loss": 4.1956, + "step": 1690 + }, + { + "epoch": 0.02, + "learning_rate": 1.0743174924165825e-05, + "loss": 4.1742, + "step": 1700 + }, + { + "epoch": 0.02, + "learning_rate": 1.0806370070778566e-05, + "loss": 4.1863, + "step": 1710 + }, + { + "epoch": 0.02, + "learning_rate": 1.0869565217391305e-05, + "loss": 3.9666, + "step": 1720 + }, + { + "epoch": 0.02, + "learning_rate": 1.0932760364004046e-05, + "loss": 4.1567, + "step": 1730 + }, + { + "epoch": 0.02, + "learning_rate": 1.0995955510616787e-05, + "loss": 4.2431, + "step": 1740 + }, + { + "epoch": 0.02, + "learning_rate": 1.1059150657229526e-05, + "loss": 4.0643, + "step": 1750 + }, + { + "epoch": 0.02, + "learning_rate": 1.1122345803842266e-05, + "loss": 4.2388, + "step": 1760 + }, + { + "epoch": 0.02, + "learning_rate": 1.1185540950455007e-05, + "loss": 4.3002, + "step": 1770 + }, + { + "epoch": 0.02, + "learning_rate": 1.1248736097067746e-05, + "loss": 4.0982, + "step": 1780 + }, + { + "epoch": 0.02, + "learning_rate": 1.1311931243680487e-05, + "loss": 4.1627, + "step": 1790 + }, + { + "epoch": 0.02, + "learning_rate": 1.1375126390293226e-05, + "loss": 4.1522, + "step": 1800 + }, + { + "epoch": 0.02, + "learning_rate": 1.1438321536905967e-05, + "loss": 4.2056, + "step": 1810 + }, + { + "epoch": 0.02, + "learning_rate": 1.1501516683518707e-05, + "loss": 3.9712, + "step": 1820 + }, + { + "epoch": 0.02, + "learning_rate": 1.1564711830131446e-05, + "loss": 3.8599, + "step": 1830 + }, + { + "epoch": 0.02, + "learning_rate": 1.1627906976744187e-05, + "loss": 4.0825, + "step": 1840 + }, + { + "epoch": 0.02, + "learning_rate": 1.1691102123356928e-05, + "loss": 4.3179, + "step": 1850 + }, + { + "epoch": 0.02, + "learning_rate": 1.1754297269969667e-05, + "loss": 4.1472, + "step": 1860 + }, + { + "epoch": 0.02, + "learning_rate": 1.1817492416582408e-05, + "loss": 3.805, + "step": 1870 + }, + { + "epoch": 0.02, + "learning_rate": 1.1880687563195148e-05, + "loss": 4.0287, + "step": 1880 + }, + { + "epoch": 0.02, + "learning_rate": 1.1943882709807887e-05, + "loss": 4.075, + "step": 1890 + }, + { + "epoch": 0.02, + "learning_rate": 1.2007077856420628e-05, + "loss": 4.0923, + "step": 1900 + }, + { + "epoch": 0.02, + "learning_rate": 1.2070273003033369e-05, + "loss": 4.053, + "step": 1910 + }, + { + "epoch": 0.02, + "learning_rate": 1.2133468149646108e-05, + "loss": 3.9915, + "step": 1920 + }, + { + "epoch": 0.02, + "learning_rate": 1.2196663296258848e-05, + "loss": 4.0775, + "step": 1930 + }, + { + "epoch": 0.02, + "learning_rate": 1.225985844287159e-05, + "loss": 3.8254, + "step": 1940 + }, + { + "epoch": 0.02, + "learning_rate": 1.2323053589484328e-05, + "loss": 3.9222, + "step": 1950 + }, + { + "epoch": 0.02, + "learning_rate": 1.2386248736097069e-05, + "loss": 4.0325, + "step": 1960 + }, + { + "epoch": 0.02, + "learning_rate": 1.244944388270981e-05, + "loss": 3.8016, + "step": 1970 + }, + { + "epoch": 0.03, + "learning_rate": 1.2512639029322549e-05, + "loss": 3.9808, + "step": 1980 + }, + { + "epoch": 0.03, + "learning_rate": 1.257583417593529e-05, + "loss": 3.7906, + "step": 1990 + }, + { + "epoch": 0.03, + "learning_rate": 1.263902932254803e-05, + "loss": 4.0861, + "step": 2000 + }, + { + "epoch": 0.03, + "learning_rate": 1.270222446916077e-05, + "loss": 3.8172, + "step": 2010 + }, + { + "epoch": 0.03, + "learning_rate": 1.2765419615773508e-05, + "loss": 4.0993, + "step": 2020 + }, + { + "epoch": 0.03, + "learning_rate": 1.2828614762386249e-05, + "loss": 3.9353, + "step": 2030 + }, + { + "epoch": 0.03, + "learning_rate": 1.289180990899899e-05, + "loss": 3.8593, + "step": 2040 + }, + { + "epoch": 0.03, + "learning_rate": 1.295500505561173e-05, + "loss": 4.1449, + "step": 2050 + }, + { + "epoch": 0.03, + "learning_rate": 1.3018200202224471e-05, + "loss": 4.0715, + "step": 2060 + }, + { + "epoch": 0.03, + "learning_rate": 1.308139534883721e-05, + "loss": 4.1413, + "step": 2070 + }, + { + "epoch": 0.03, + "learning_rate": 1.314459049544995e-05, + "loss": 3.9075, + "step": 2080 + }, + { + "epoch": 0.03, + "learning_rate": 1.320778564206269e-05, + "loss": 3.7543, + "step": 2090 + }, + { + "epoch": 0.03, + "learning_rate": 1.327098078867543e-05, + "loss": 3.9419, + "step": 2100 + }, + { + "epoch": 0.03, + "learning_rate": 1.3334175935288171e-05, + "loss": 3.8122, + "step": 2110 + }, + { + "epoch": 0.03, + "learning_rate": 1.339737108190091e-05, + "loss": 4.0084, + "step": 2120 + }, + { + "epoch": 0.03, + "learning_rate": 1.3460566228513651e-05, + "loss": 3.6395, + "step": 2130 + }, + { + "epoch": 0.03, + "learning_rate": 1.352376137512639e-05, + "loss": 3.7254, + "step": 2140 + }, + { + "epoch": 0.03, + "learning_rate": 1.3586956521739131e-05, + "loss": 3.9213, + "step": 2150 + }, + { + "epoch": 0.03, + "learning_rate": 1.3650151668351872e-05, + "loss": 3.8688, + "step": 2160 + }, + { + "epoch": 0.03, + "learning_rate": 1.3713346814964612e-05, + "loss": 3.7157, + "step": 2170 + }, + { + "epoch": 0.03, + "learning_rate": 1.3776541961577351e-05, + "loss": 3.9638, + "step": 2180 + }, + { + "epoch": 0.03, + "learning_rate": 1.3839737108190092e-05, + "loss": 3.7266, + "step": 2190 + }, + { + "epoch": 0.03, + "learning_rate": 1.3902932254802831e-05, + "loss": 3.5752, + "step": 2200 + }, + { + "epoch": 0.03, + "learning_rate": 1.3966127401415572e-05, + "loss": 3.7281, + "step": 2210 + }, + { + "epoch": 0.03, + "learning_rate": 1.4029322548028313e-05, + "loss": 3.5565, + "step": 2220 + }, + { + "epoch": 0.03, + "learning_rate": 1.4092517694641053e-05, + "loss": 3.8869, + "step": 2230 + }, + { + "epoch": 0.03, + "learning_rate": 1.4155712841253792e-05, + "loss": 3.796, + "step": 2240 + }, + { + "epoch": 0.03, + "learning_rate": 1.4218907987866533e-05, + "loss": 3.8631, + "step": 2250 + }, + { + "epoch": 0.03, + "learning_rate": 1.4282103134479272e-05, + "loss": 3.5033, + "step": 2260 + }, + { + "epoch": 0.03, + "learning_rate": 1.4345298281092013e-05, + "loss": 3.5252, + "step": 2270 + }, + { + "epoch": 0.03, + "learning_rate": 1.4408493427704754e-05, + "loss": 3.8213, + "step": 2280 + }, + { + "epoch": 0.03, + "learning_rate": 1.4471688574317493e-05, + "loss": 3.6509, + "step": 2290 + }, + { + "epoch": 0.03, + "learning_rate": 1.4534883720930233e-05, + "loss": 3.4573, + "step": 2300 + }, + { + "epoch": 0.03, + "learning_rate": 1.4598078867542974e-05, + "loss": 3.5626, + "step": 2310 + }, + { + "epoch": 0.03, + "learning_rate": 1.4661274014155713e-05, + "loss": 3.7017, + "step": 2320 + }, + { + "epoch": 0.03, + "learning_rate": 1.4724469160768454e-05, + "loss": 3.8538, + "step": 2330 + }, + { + "epoch": 0.03, + "learning_rate": 1.4787664307381195e-05, + "loss": 3.8036, + "step": 2340 + }, + { + "epoch": 0.03, + "learning_rate": 1.4850859453993934e-05, + "loss": 3.6317, + "step": 2350 + }, + { + "epoch": 0.03, + "learning_rate": 1.4914054600606674e-05, + "loss": 3.5713, + "step": 2360 + }, + { + "epoch": 0.03, + "learning_rate": 1.4977249747219415e-05, + "loss": 3.3497, + "step": 2370 + }, + { + "epoch": 0.03, + "learning_rate": 1.5040444893832154e-05, + "loss": 3.7972, + "step": 2380 + }, + { + "epoch": 0.03, + "learning_rate": 1.5103640040444895e-05, + "loss": 3.269, + "step": 2390 + }, + { + "epoch": 0.03, + "learning_rate": 1.5166835187057636e-05, + "loss": 3.3064, + "step": 2400 + }, + { + "epoch": 0.03, + "learning_rate": 1.5230030333670375e-05, + "loss": 3.4208, + "step": 2410 + }, + { + "epoch": 0.03, + "learning_rate": 1.5293225480283117e-05, + "loss": 3.5935, + "step": 2420 + }, + { + "epoch": 0.03, + "learning_rate": 1.5356420626895853e-05, + "loss": 3.5516, + "step": 2430 + }, + { + "epoch": 0.03, + "learning_rate": 1.5419615773508595e-05, + "loss": 3.664, + "step": 2440 + }, + { + "epoch": 0.03, + "learning_rate": 1.5482810920121334e-05, + "loss": 3.686, + "step": 2450 + }, + { + "epoch": 0.03, + "learning_rate": 1.5546006066734077e-05, + "loss": 3.7139, + "step": 2460 + }, + { + "epoch": 0.03, + "learning_rate": 1.5609201213346816e-05, + "loss": 3.6894, + "step": 2470 + }, + { + "epoch": 0.03, + "learning_rate": 1.5672396359959558e-05, + "loss": 3.452, + "step": 2480 + }, + { + "epoch": 0.03, + "learning_rate": 1.5735591506572294e-05, + "loss": 3.1836, + "step": 2490 + }, + { + "epoch": 0.03, + "learning_rate": 1.5798786653185036e-05, + "loss": 3.6827, + "step": 2500 + }, + { + "epoch": 0.03, + "learning_rate": 1.5861981799797775e-05, + "loss": 3.2805, + "step": 2510 + }, + { + "epoch": 0.03, + "learning_rate": 1.5925176946410518e-05, + "loss": 3.4733, + "step": 2520 + }, + { + "epoch": 0.03, + "learning_rate": 1.5988372093023257e-05, + "loss": 3.5676, + "step": 2530 + }, + { + "epoch": 0.03, + "learning_rate": 1.6051567239636e-05, + "loss": 3.7344, + "step": 2540 + }, + { + "epoch": 0.03, + "learning_rate": 1.6114762386248735e-05, + "loss": 3.2077, + "step": 2550 + }, + { + "epoch": 0.03, + "learning_rate": 1.6177957532861477e-05, + "loss": 3.569, + "step": 2560 + }, + { + "epoch": 0.03, + "learning_rate": 1.6241152679474216e-05, + "loss": 3.5911, + "step": 2570 + }, + { + "epoch": 0.03, + "learning_rate": 1.630434782608696e-05, + "loss": 3.1481, + "step": 2580 + }, + { + "epoch": 0.03, + "learning_rate": 1.6367542972699698e-05, + "loss": 3.8807, + "step": 2590 + }, + { + "epoch": 0.03, + "learning_rate": 1.643073811931244e-05, + "loss": 3.7296, + "step": 2600 + }, + { + "epoch": 0.03, + "learning_rate": 1.6493933265925176e-05, + "loss": 3.6879, + "step": 2610 + }, + { + "epoch": 0.03, + "learning_rate": 1.6557128412537918e-05, + "loss": 3.2696, + "step": 2620 + }, + { + "epoch": 0.03, + "learning_rate": 1.6620323559150657e-05, + "loss": 3.5097, + "step": 2630 + }, + { + "epoch": 0.03, + "learning_rate": 1.66835187057634e-05, + "loss": 3.4804, + "step": 2640 + }, + { + "epoch": 0.03, + "learning_rate": 1.674671385237614e-05, + "loss": 3.6257, + "step": 2650 + }, + { + "epoch": 0.03, + "learning_rate": 1.680990899898888e-05, + "loss": 3.6508, + "step": 2660 + }, + { + "epoch": 0.03, + "learning_rate": 1.6873104145601617e-05, + "loss": 3.6276, + "step": 2670 + }, + { + "epoch": 0.03, + "learning_rate": 1.693629929221436e-05, + "loss": 3.5054, + "step": 2680 + }, + { + "epoch": 0.03, + "learning_rate": 1.6999494438827098e-05, + "loss": 3.4006, + "step": 2690 + }, + { + "epoch": 0.03, + "learning_rate": 1.706268958543984e-05, + "loss": 3.1905, + "step": 2700 + }, + { + "epoch": 0.03, + "learning_rate": 1.712588473205258e-05, + "loss": 3.2191, + "step": 2710 + }, + { + "epoch": 0.03, + "learning_rate": 1.7189079878665322e-05, + "loss": 3.3957, + "step": 2720 + }, + { + "epoch": 0.03, + "learning_rate": 1.7252275025278058e-05, + "loss": 3.2479, + "step": 2730 + }, + { + "epoch": 0.03, + "learning_rate": 1.73154701718908e-05, + "loss": 3.2937, + "step": 2740 + }, + { + "epoch": 0.03, + "learning_rate": 1.737866531850354e-05, + "loss": 3.024, + "step": 2750 + }, + { + "epoch": 0.03, + "learning_rate": 1.744186046511628e-05, + "loss": 3.1543, + "step": 2760 + }, + { + "epoch": 0.04, + "learning_rate": 1.750505561172902e-05, + "loss": 3.0059, + "step": 2770 + }, + { + "epoch": 0.04, + "learning_rate": 1.756825075834176e-05, + "loss": 3.2246, + "step": 2780 + }, + { + "epoch": 0.04, + "learning_rate": 1.76314459049545e-05, + "loss": 3.2096, + "step": 2790 + }, + { + "epoch": 0.04, + "learning_rate": 1.769464105156724e-05, + "loss": 3.3073, + "step": 2800 + }, + { + "epoch": 0.04, + "learning_rate": 1.775783619817998e-05, + "loss": 2.7326, + "step": 2810 + }, + { + "epoch": 0.04, + "learning_rate": 1.7821031344792722e-05, + "loss": 3.0153, + "step": 2820 + }, + { + "epoch": 0.04, + "learning_rate": 1.788422649140546e-05, + "loss": 3.2977, + "step": 2830 + }, + { + "epoch": 0.04, + "learning_rate": 1.79474216380182e-05, + "loss": 3.0521, + "step": 2840 + }, + { + "epoch": 0.04, + "learning_rate": 1.801061678463094e-05, + "loss": 3.2618, + "step": 2850 + }, + { + "epoch": 0.04, + "learning_rate": 1.8073811931243682e-05, + "loss": 3.4206, + "step": 2860 + }, + { + "epoch": 0.04, + "learning_rate": 1.813700707785642e-05, + "loss": 3.3683, + "step": 2870 + }, + { + "epoch": 0.04, + "learning_rate": 1.8200202224469163e-05, + "loss": 3.1959, + "step": 2880 + }, + { + "epoch": 0.04, + "learning_rate": 1.8263397371081902e-05, + "loss": 3.2729, + "step": 2890 + }, + { + "epoch": 0.04, + "learning_rate": 1.832659251769464e-05, + "loss": 3.1314, + "step": 2900 + }, + { + "epoch": 0.04, + "learning_rate": 1.838978766430738e-05, + "loss": 3.2666, + "step": 2910 + }, + { + "epoch": 0.04, + "learning_rate": 1.8452982810920123e-05, + "loss": 3.3068, + "step": 2920 + }, + { + "epoch": 0.04, + "learning_rate": 1.8516177957532862e-05, + "loss": 2.9326, + "step": 2930 + }, + { + "epoch": 0.04, + "learning_rate": 1.8579373104145604e-05, + "loss": 3.2599, + "step": 2940 + }, + { + "epoch": 0.04, + "learning_rate": 1.8642568250758343e-05, + "loss": 3.01, + "step": 2950 + }, + { + "epoch": 0.04, + "learning_rate": 1.8705763397371082e-05, + "loss": 3.129, + "step": 2960 + }, + { + "epoch": 0.04, + "learning_rate": 1.876895854398382e-05, + "loss": 3.0911, + "step": 2970 + }, + { + "epoch": 0.04, + "learning_rate": 1.8832153690596564e-05, + "loss": 3.2024, + "step": 2980 + }, + { + "epoch": 0.04, + "learning_rate": 1.8895348837209303e-05, + "loss": 2.8037, + "step": 2990 + }, + { + "epoch": 0.04, + "learning_rate": 1.8958543983822045e-05, + "loss": 3.3293, + "step": 3000 + }, + { + "epoch": 0.04, + "learning_rate": 1.9021739130434784e-05, + "loss": 3.0673, + "step": 3010 + }, + { + "epoch": 0.04, + "learning_rate": 1.9084934277047523e-05, + "loss": 2.8306, + "step": 3020 + }, + { + "epoch": 0.04, + "learning_rate": 1.9148129423660262e-05, + "loss": 2.833, + "step": 3030 + }, + { + "epoch": 0.04, + "learning_rate": 1.9211324570273005e-05, + "loss": 2.9429, + "step": 3040 + }, + { + "epoch": 0.04, + "learning_rate": 1.9274519716885744e-05, + "loss": 3.1177, + "step": 3050 + }, + { + "epoch": 0.04, + "learning_rate": 1.9337714863498486e-05, + "loss": 3.2254, + "step": 3060 + }, + { + "epoch": 0.04, + "learning_rate": 1.9400910010111225e-05, + "loss": 2.9748, + "step": 3070 + }, + { + "epoch": 0.04, + "learning_rate": 1.9464105156723964e-05, + "loss": 2.8905, + "step": 3080 + }, + { + "epoch": 0.04, + "learning_rate": 1.9527300303336703e-05, + "loss": 2.6819, + "step": 3090 + }, + { + "epoch": 0.04, + "learning_rate": 1.9590495449949446e-05, + "loss": 2.9618, + "step": 3100 + }, + { + "epoch": 0.04, + "learning_rate": 1.9653690596562185e-05, + "loss": 3.0473, + "step": 3110 + }, + { + "epoch": 0.04, + "learning_rate": 1.9716885743174927e-05, + "loss": 2.9152, + "step": 3120 + }, + { + "epoch": 0.04, + "learning_rate": 1.9780080889787666e-05, + "loss": 3.0444, + "step": 3130 + }, + { + "epoch": 0.04, + "learning_rate": 1.9843276036400405e-05, + "loss": 3.0475, + "step": 3140 + }, + { + "epoch": 0.04, + "learning_rate": 1.9906471183013144e-05, + "loss": 2.971, + "step": 3150 + }, + { + "epoch": 0.04, + "learning_rate": 1.9969666329625887e-05, + "loss": 2.4389, + "step": 3160 + }, + { + "epoch": 0.04, + "learning_rate": 2.0032861476238626e-05, + "loss": 2.6158, + "step": 3170 + }, + { + "epoch": 0.04, + "learning_rate": 2.0096056622851368e-05, + "loss": 3.111, + "step": 3180 + }, + { + "epoch": 0.04, + "learning_rate": 2.0159251769464104e-05, + "loss": 2.6425, + "step": 3190 + }, + { + "epoch": 0.04, + "learning_rate": 2.0222446916076846e-05, + "loss": 2.6529, + "step": 3200 + }, + { + "epoch": 0.04, + "learning_rate": 2.0285642062689585e-05, + "loss": 2.7563, + "step": 3210 + }, + { + "epoch": 0.04, + "learning_rate": 2.0348837209302328e-05, + "loss": 3.1355, + "step": 3220 + }, + { + "epoch": 0.04, + "learning_rate": 2.0412032355915067e-05, + "loss": 2.7316, + "step": 3230 + }, + { + "epoch": 0.04, + "learning_rate": 2.047522750252781e-05, + "loss": 3.0771, + "step": 3240 + }, + { + "epoch": 0.04, + "learning_rate": 2.0538422649140545e-05, + "loss": 2.7386, + "step": 3250 + }, + { + "epoch": 0.04, + "learning_rate": 2.0601617795753287e-05, + "loss": 3.0685, + "step": 3260 + }, + { + "epoch": 0.04, + "learning_rate": 2.0664812942366026e-05, + "loss": 3.0143, + "step": 3270 + }, + { + "epoch": 0.04, + "learning_rate": 2.072800808897877e-05, + "loss": 2.6404, + "step": 3280 + }, + { + "epoch": 0.04, + "learning_rate": 2.0791203235591508e-05, + "loss": 2.9624, + "step": 3290 + }, + { + "epoch": 0.04, + "learning_rate": 2.085439838220425e-05, + "loss": 2.8812, + "step": 3300 + }, + { + "epoch": 0.04, + "learning_rate": 2.0917593528816986e-05, + "loss": 2.5523, + "step": 3310 + }, + { + "epoch": 0.04, + "learning_rate": 2.0980788675429728e-05, + "loss": 2.535, + "step": 3320 + }, + { + "epoch": 0.04, + "learning_rate": 2.1043983822042467e-05, + "loss": 3.0393, + "step": 3330 + }, + { + "epoch": 0.04, + "learning_rate": 2.110717896865521e-05, + "loss": 2.5007, + "step": 3340 + }, + { + "epoch": 0.04, + "learning_rate": 2.117037411526795e-05, + "loss": 2.8221, + "step": 3350 + }, + { + "epoch": 0.04, + "learning_rate": 2.123356926188069e-05, + "loss": 2.8118, + "step": 3360 + }, + { + "epoch": 0.04, + "learning_rate": 2.1296764408493427e-05, + "loss": 2.6009, + "step": 3370 + }, + { + "epoch": 0.04, + "learning_rate": 2.135995955510617e-05, + "loss": 2.7411, + "step": 3380 + }, + { + "epoch": 0.04, + "learning_rate": 2.1423154701718908e-05, + "loss": 2.6422, + "step": 3390 + }, + { + "epoch": 0.04, + "learning_rate": 2.148634984833165e-05, + "loss": 2.9583, + "step": 3400 + }, + { + "epoch": 0.04, + "learning_rate": 2.154954499494439e-05, + "loss": 2.6321, + "step": 3410 + }, + { + "epoch": 0.04, + "learning_rate": 2.1612740141557132e-05, + "loss": 2.7626, + "step": 3420 + }, + { + "epoch": 0.04, + "learning_rate": 2.1675935288169868e-05, + "loss": 2.3105, + "step": 3430 + }, + { + "epoch": 0.04, + "learning_rate": 2.173913043478261e-05, + "loss": 2.5993, + "step": 3440 + }, + { + "epoch": 0.04, + "learning_rate": 2.180232558139535e-05, + "loss": 2.5323, + "step": 3450 + }, + { + "epoch": 0.04, + "learning_rate": 2.186552072800809e-05, + "loss": 2.2182, + "step": 3460 + }, + { + "epoch": 0.04, + "learning_rate": 2.192871587462083e-05, + "loss": 2.746, + "step": 3470 + }, + { + "epoch": 0.04, + "learning_rate": 2.1991911021233573e-05, + "loss": 2.7207, + "step": 3480 + }, + { + "epoch": 0.04, + "learning_rate": 2.205510616784631e-05, + "loss": 2.4982, + "step": 3490 + }, + { + "epoch": 0.04, + "learning_rate": 2.211830131445905e-05, + "loss": 2.5163, + "step": 3500 + }, + { + "epoch": 0.04, + "learning_rate": 2.218149646107179e-05, + "loss": 3.2339, + "step": 3510 + }, + { + "epoch": 0.04, + "learning_rate": 2.2244691607684533e-05, + "loss": 2.4154, + "step": 3520 + }, + { + "epoch": 0.04, + "learning_rate": 2.230788675429727e-05, + "loss": 2.9864, + "step": 3530 + }, + { + "epoch": 0.04, + "learning_rate": 2.2371081900910014e-05, + "loss": 2.2323, + "step": 3540 + }, + { + "epoch": 0.04, + "learning_rate": 2.243427704752275e-05, + "loss": 2.5058, + "step": 3550 + }, + { + "epoch": 0.04, + "learning_rate": 2.2497472194135492e-05, + "loss": 2.695, + "step": 3560 + }, + { + "epoch": 0.05, + "learning_rate": 2.256066734074823e-05, + "loss": 2.5402, + "step": 3570 + }, + { + "epoch": 0.05, + "learning_rate": 2.2623862487360974e-05, + "loss": 2.5119, + "step": 3580 + }, + { + "epoch": 0.05, + "learning_rate": 2.2687057633973713e-05, + "loss": 2.4513, + "step": 3590 + }, + { + "epoch": 0.05, + "learning_rate": 2.275025278058645e-05, + "loss": 2.2057, + "step": 3600 + }, + { + "epoch": 0.05, + "learning_rate": 2.281344792719919e-05, + "loss": 2.3299, + "step": 3610 + }, + { + "epoch": 0.05, + "learning_rate": 2.2876643073811933e-05, + "loss": 2.1939, + "step": 3620 + }, + { + "epoch": 0.05, + "learning_rate": 2.2939838220424672e-05, + "loss": 3.0215, + "step": 3630 + }, + { + "epoch": 0.05, + "learning_rate": 2.3003033367037415e-05, + "loss": 2.6346, + "step": 3640 + }, + { + "epoch": 0.05, + "learning_rate": 2.3066228513650154e-05, + "loss": 2.7174, + "step": 3650 + }, + { + "epoch": 0.05, + "learning_rate": 2.3129423660262893e-05, + "loss": 2.6709, + "step": 3660 + }, + { + "epoch": 0.05, + "learning_rate": 2.319261880687563e-05, + "loss": 2.5589, + "step": 3670 + }, + { + "epoch": 0.05, + "learning_rate": 2.3255813953488374e-05, + "loss": 2.2971, + "step": 3680 + }, + { + "epoch": 0.05, + "learning_rate": 2.3319009100101113e-05, + "loss": 2.5036, + "step": 3690 + }, + { + "epoch": 0.05, + "learning_rate": 2.3382204246713856e-05, + "loss": 2.2793, + "step": 3700 + }, + { + "epoch": 0.05, + "learning_rate": 2.3445399393326595e-05, + "loss": 2.769, + "step": 3710 + }, + { + "epoch": 0.05, + "learning_rate": 2.3508594539939334e-05, + "loss": 2.1892, + "step": 3720 + }, + { + "epoch": 0.05, + "learning_rate": 2.3571789686552073e-05, + "loss": 2.7442, + "step": 3730 + }, + { + "epoch": 0.05, + "learning_rate": 2.3634984833164815e-05, + "loss": 2.4694, + "step": 3740 + }, + { + "epoch": 0.05, + "learning_rate": 2.3698179979777554e-05, + "loss": 2.5707, + "step": 3750 + }, + { + "epoch": 0.05, + "learning_rate": 2.3761375126390297e-05, + "loss": 3.2404, + "step": 3760 + }, + { + "epoch": 0.05, + "learning_rate": 2.3824570273003036e-05, + "loss": 2.3573, + "step": 3770 + }, + { + "epoch": 0.05, + "learning_rate": 2.3887765419615775e-05, + "loss": 2.6777, + "step": 3780 + }, + { + "epoch": 0.05, + "learning_rate": 2.3950960566228514e-05, + "loss": 2.5962, + "step": 3790 + }, + { + "epoch": 0.05, + "learning_rate": 2.4014155712841256e-05, + "loss": 2.3535, + "step": 3800 + }, + { + "epoch": 0.05, + "learning_rate": 2.4077350859453995e-05, + "loss": 2.2846, + "step": 3810 + }, + { + "epoch": 0.05, + "learning_rate": 2.4140546006066737e-05, + "loss": 2.5489, + "step": 3820 + }, + { + "epoch": 0.05, + "learning_rate": 2.4203741152679477e-05, + "loss": 2.4501, + "step": 3830 + }, + { + "epoch": 0.05, + "learning_rate": 2.4266936299292216e-05, + "loss": 1.9336, + "step": 3840 + }, + { + "epoch": 0.05, + "learning_rate": 2.4330131445904955e-05, + "loss": 2.7265, + "step": 3850 + }, + { + "epoch": 0.05, + "learning_rate": 2.4393326592517697e-05, + "loss": 1.8973, + "step": 3860 + }, + { + "epoch": 0.05, + "learning_rate": 2.4456521739130436e-05, + "loss": 2.0162, + "step": 3870 + }, + { + "epoch": 0.05, + "learning_rate": 2.451971688574318e-05, + "loss": 2.1607, + "step": 3880 + }, + { + "epoch": 0.05, + "learning_rate": 2.4582912032355917e-05, + "loss": 2.3048, + "step": 3890 + }, + { + "epoch": 0.05, + "learning_rate": 2.4646107178968657e-05, + "loss": 2.3436, + "step": 3900 + }, + { + "epoch": 0.05, + "learning_rate": 2.4709302325581396e-05, + "loss": 2.1047, + "step": 3910 + }, + { + "epoch": 0.05, + "learning_rate": 2.4772497472194138e-05, + "loss": 2.3984, + "step": 3920 + }, + { + "epoch": 0.05, + "learning_rate": 2.4835692618806877e-05, + "loss": 2.445, + "step": 3930 + }, + { + "epoch": 0.05, + "learning_rate": 2.489888776541962e-05, + "loss": 2.461, + "step": 3940 + }, + { + "epoch": 0.05, + "learning_rate": 2.496208291203236e-05, + "loss": 1.9278, + "step": 3950 + }, + { + "epoch": 0.05, + "eval_accuracy": 0.3483421494000774, + "eval_loss": 2.7015833854675293, + "eval_runtime": 1214.1682, + "eval_samples_per_second": 6.384, + "eval_steps_per_second": 3.192, + "step": 3956 + }, + { + "epoch": 1.0, + "learning_rate": 2.5025278058645097e-05, + "loss": 2.4069, + "step": 3960 + }, + { + "epoch": 1.0, + "learning_rate": 2.508847320525784e-05, + "loss": 2.4144, + "step": 3970 + }, + { + "epoch": 1.0, + "learning_rate": 2.515166835187058e-05, + "loss": 1.9401, + "step": 3980 + }, + { + "epoch": 1.0, + "learning_rate": 2.5214863498483315e-05, + "loss": 2.1604, + "step": 3990 + }, + { + "epoch": 1.0, + "learning_rate": 2.527805864509606e-05, + "loss": 2.0431, + "step": 4000 + }, + { + "epoch": 1.0, + "learning_rate": 2.5341253791708796e-05, + "loss": 2.382, + "step": 4010 + }, + { + "epoch": 1.0, + "learning_rate": 2.540444893832154e-05, + "loss": 2.3504, + "step": 4020 + }, + { + "epoch": 1.0, + "learning_rate": 2.5467644084934277e-05, + "loss": 1.9599, + "step": 4030 + }, + { + "epoch": 1.0, + "learning_rate": 2.5530839231547017e-05, + "loss": 2.5499, + "step": 4040 + }, + { + "epoch": 1.0, + "learning_rate": 2.559403437815976e-05, + "loss": 2.4014, + "step": 4050 + }, + { + "epoch": 1.0, + "learning_rate": 2.5657229524772498e-05, + "loss": 2.0668, + "step": 4060 + }, + { + "epoch": 1.0, + "learning_rate": 2.572042467138524e-05, + "loss": 2.0949, + "step": 4070 + }, + { + "epoch": 1.0, + "learning_rate": 2.578361981799798e-05, + "loss": 2.0247, + "step": 4080 + }, + { + "epoch": 1.0, + "learning_rate": 2.5846814964610722e-05, + "loss": 1.8933, + "step": 4090 + }, + { + "epoch": 1.0, + "learning_rate": 2.591001011122346e-05, + "loss": 2.2305, + "step": 4100 + }, + { + "epoch": 1.0, + "learning_rate": 2.5973205257836197e-05, + "loss": 2.4658, + "step": 4110 + }, + { + "epoch": 1.0, + "learning_rate": 2.6036400404448942e-05, + "loss": 2.3094, + "step": 4120 + }, + { + "epoch": 1.0, + "learning_rate": 2.6099595551061678e-05, + "loss": 1.8873, + "step": 4130 + }, + { + "epoch": 1.0, + "learning_rate": 2.616279069767442e-05, + "loss": 1.8086, + "step": 4140 + }, + { + "epoch": 1.0, + "learning_rate": 2.622598584428716e-05, + "loss": 1.8955, + "step": 4150 + }, + { + "epoch": 1.0, + "learning_rate": 2.62891809908999e-05, + "loss": 1.9182, + "step": 4160 + }, + { + "epoch": 1.0, + "learning_rate": 2.635237613751264e-05, + "loss": 2.1472, + "step": 4170 + }, + { + "epoch": 1.0, + "learning_rate": 2.641557128412538e-05, + "loss": 1.7235, + "step": 4180 + }, + { + "epoch": 1.0, + "learning_rate": 2.6478766430738122e-05, + "loss": 2.2514, + "step": 4190 + }, + { + "epoch": 1.0, + "learning_rate": 2.654196157735086e-05, + "loss": 2.023, + "step": 4200 + }, + { + "epoch": 1.0, + "learning_rate": 2.6605156723963604e-05, + "loss": 1.9005, + "step": 4210 + }, + { + "epoch": 1.0, + "learning_rate": 2.6668351870576343e-05, + "loss": 2.4229, + "step": 4220 + }, + { + "epoch": 1.0, + "learning_rate": 2.673154701718908e-05, + "loss": 2.3923, + "step": 4230 + }, + { + "epoch": 1.0, + "learning_rate": 2.679474216380182e-05, + "loss": 1.8501, + "step": 4240 + }, + { + "epoch": 1.0, + "learning_rate": 2.685793731041456e-05, + "loss": 1.923, + "step": 4250 + }, + { + "epoch": 1.0, + "learning_rate": 2.6921132457027302e-05, + "loss": 1.9329, + "step": 4260 + }, + { + "epoch": 1.0, + "learning_rate": 2.698432760364004e-05, + "loss": 2.1817, + "step": 4270 + }, + { + "epoch": 1.0, + "learning_rate": 2.704752275025278e-05, + "loss": 1.7928, + "step": 4280 + }, + { + "epoch": 1.0, + "learning_rate": 2.7110717896865523e-05, + "loss": 1.9392, + "step": 4290 + }, + { + "epoch": 1.0, + "learning_rate": 2.7173913043478262e-05, + "loss": 1.7639, + "step": 4300 + }, + { + "epoch": 1.0, + "learning_rate": 2.7237108190091004e-05, + "loss": 1.701, + "step": 4310 + }, + { + "epoch": 1.0, + "learning_rate": 2.7300303336703743e-05, + "loss": 1.987, + "step": 4320 + }, + { + "epoch": 1.0, + "learning_rate": 2.7363498483316486e-05, + "loss": 1.662, + "step": 4330 + }, + { + "epoch": 1.0, + "learning_rate": 2.7426693629929225e-05, + "loss": 1.4077, + "step": 4340 + }, + { + "epoch": 1.0, + "learning_rate": 2.748988877654196e-05, + "loss": 1.4389, + "step": 4350 + }, + { + "epoch": 1.01, + "learning_rate": 2.7553083923154703e-05, + "loss": 1.8565, + "step": 4360 + }, + { + "epoch": 1.01, + "learning_rate": 2.7616279069767442e-05, + "loss": 1.8253, + "step": 4370 + }, + { + "epoch": 1.01, + "learning_rate": 2.7679474216380184e-05, + "loss": 1.9225, + "step": 4380 + }, + { + "epoch": 1.01, + "learning_rate": 2.7742669362992923e-05, + "loss": 1.9639, + "step": 4390 + }, + { + "epoch": 1.01, + "learning_rate": 2.7805864509605662e-05, + "loss": 2.14, + "step": 4400 + }, + { + "epoch": 1.01, + "learning_rate": 2.7869059656218405e-05, + "loss": 1.9855, + "step": 4410 + }, + { + "epoch": 1.01, + "learning_rate": 2.7932254802831144e-05, + "loss": 1.7251, + "step": 4420 + }, + { + "epoch": 1.01, + "learning_rate": 2.7995449949443886e-05, + "loss": 2.0943, + "step": 4430 + }, + { + "epoch": 1.01, + "learning_rate": 2.8058645096056625e-05, + "loss": 1.8432, + "step": 4440 + }, + { + "epoch": 1.01, + "learning_rate": 2.812184024266936e-05, + "loss": 2.358, + "step": 4450 + }, + { + "epoch": 1.01, + "learning_rate": 2.8185035389282107e-05, + "loss": 1.7749, + "step": 4460 + }, + { + "epoch": 1.01, + "learning_rate": 2.8248230535894842e-05, + "loss": 1.4156, + "step": 4470 + }, + { + "epoch": 1.01, + "learning_rate": 2.8311425682507585e-05, + "loss": 2.4294, + "step": 4480 + }, + { + "epoch": 1.01, + "learning_rate": 2.8374620829120324e-05, + "loss": 2.4287, + "step": 4490 + }, + { + "epoch": 1.01, + "learning_rate": 2.8437815975733066e-05, + "loss": 1.9334, + "step": 4500 + }, + { + "epoch": 1.01, + "learning_rate": 2.8501011122345805e-05, + "loss": 2.2001, + "step": 4510 + }, + { + "epoch": 1.01, + "learning_rate": 2.8564206268958544e-05, + "loss": 1.9227, + "step": 4520 + }, + { + "epoch": 1.01, + "learning_rate": 2.8627401415571287e-05, + "loss": 2.2894, + "step": 4530 + }, + { + "epoch": 1.01, + "learning_rate": 2.8690596562184026e-05, + "loss": 2.0885, + "step": 4540 + }, + { + "epoch": 1.01, + "learning_rate": 2.8753791708796768e-05, + "loss": 1.5449, + "step": 4550 + }, + { + "epoch": 1.01, + "learning_rate": 2.8816986855409507e-05, + "loss": 1.6547, + "step": 4560 + }, + { + "epoch": 1.01, + "learning_rate": 2.8880182002022243e-05, + "loss": 1.8572, + "step": 4570 + }, + { + "epoch": 1.01, + "learning_rate": 2.8943377148634985e-05, + "loss": 1.8954, + "step": 4580 + }, + { + "epoch": 1.01, + "learning_rate": 2.9006572295247724e-05, + "loss": 2.2409, + "step": 4590 + }, + { + "epoch": 1.01, + "learning_rate": 2.9069767441860467e-05, + "loss": 2.1448, + "step": 4600 + }, + { + "epoch": 1.01, + "learning_rate": 2.9132962588473206e-05, + "loss": 1.8475, + "step": 4610 + }, + { + "epoch": 1.01, + "learning_rate": 2.9196157735085948e-05, + "loss": 1.7558, + "step": 4620 + }, + { + "epoch": 1.01, + "learning_rate": 2.9259352881698687e-05, + "loss": 1.9998, + "step": 4630 + }, + { + "epoch": 1.01, + "learning_rate": 2.9322548028311426e-05, + "loss": 1.954, + "step": 4640 + }, + { + "epoch": 1.01, + "learning_rate": 2.938574317492417e-05, + "loss": 1.8021, + "step": 4650 + }, + { + "epoch": 1.01, + "learning_rate": 2.9448938321536908e-05, + "loss": 1.5341, + "step": 4660 + }, + { + "epoch": 1.01, + "learning_rate": 2.951213346814965e-05, + "loss": 1.8172, + "step": 4670 + }, + { + "epoch": 1.01, + "learning_rate": 2.957532861476239e-05, + "loss": 1.428, + "step": 4680 + }, + { + "epoch": 1.01, + "learning_rate": 2.9638523761375125e-05, + "loss": 1.5525, + "step": 4690 + }, + { + "epoch": 1.01, + "learning_rate": 2.9701718907987867e-05, + "loss": 1.4755, + "step": 4700 + }, + { + "epoch": 1.01, + "learning_rate": 2.9764914054600606e-05, + "loss": 1.698, + "step": 4710 + }, + { + "epoch": 1.01, + "learning_rate": 2.982810920121335e-05, + "loss": 1.5498, + "step": 4720 + }, + { + "epoch": 1.01, + "learning_rate": 2.9891304347826088e-05, + "loss": 1.9093, + "step": 4730 + }, + { + "epoch": 1.01, + "learning_rate": 2.995449949443883e-05, + "loss": 2.2121, + "step": 4740 + }, + { + "epoch": 1.01, + "learning_rate": 3.001769464105157e-05, + "loss": 1.7604, + "step": 4750 + }, + { + "epoch": 1.01, + "learning_rate": 3.0080889787664308e-05, + "loss": 2.0317, + "step": 4760 + }, + { + "epoch": 1.01, + "learning_rate": 3.014408493427705e-05, + "loss": 1.6414, + "step": 4770 + }, + { + "epoch": 1.01, + "learning_rate": 3.020728008088979e-05, + "loss": 2.0402, + "step": 4780 + }, + { + "epoch": 1.01, + "learning_rate": 3.0270475227502532e-05, + "loss": 1.7019, + "step": 4790 + }, + { + "epoch": 1.01, + "learning_rate": 3.033367037411527e-05, + "loss": 2.2473, + "step": 4800 + }, + { + "epoch": 1.01, + "learning_rate": 3.0396865520728007e-05, + "loss": 1.8549, + "step": 4810 + }, + { + "epoch": 1.01, + "learning_rate": 3.046006066734075e-05, + "loss": 1.5342, + "step": 4820 + }, + { + "epoch": 1.01, + "learning_rate": 3.052325581395349e-05, + "loss": 2.0673, + "step": 4830 + }, + { + "epoch": 1.01, + "learning_rate": 3.0586450960566234e-05, + "loss": 1.8738, + "step": 4840 + }, + { + "epoch": 1.01, + "learning_rate": 3.064964610717897e-05, + "loss": 1.6196, + "step": 4850 + }, + { + "epoch": 1.01, + "learning_rate": 3.0712841253791705e-05, + "loss": 1.4519, + "step": 4860 + }, + { + "epoch": 1.01, + "learning_rate": 3.077603640040445e-05, + "loss": 2.085, + "step": 4870 + }, + { + "epoch": 1.01, + "learning_rate": 3.083923154701719e-05, + "loss": 1.796, + "step": 4880 + }, + { + "epoch": 1.01, + "learning_rate": 3.090242669362993e-05, + "loss": 1.7531, + "step": 4890 + }, + { + "epoch": 1.01, + "learning_rate": 3.096562184024267e-05, + "loss": 1.6934, + "step": 4900 + }, + { + "epoch": 1.01, + "learning_rate": 3.1028816986855414e-05, + "loss": 1.9619, + "step": 4910 + }, + { + "epoch": 1.01, + "learning_rate": 3.109201213346815e-05, + "loss": 2.2563, + "step": 4920 + }, + { + "epoch": 1.01, + "learning_rate": 3.115520728008089e-05, + "loss": 1.8669, + "step": 4930 + }, + { + "epoch": 1.01, + "learning_rate": 3.121840242669363e-05, + "loss": 1.4466, + "step": 4940 + }, + { + "epoch": 1.01, + "learning_rate": 3.128159757330637e-05, + "loss": 1.7559, + "step": 4950 + }, + { + "epoch": 1.01, + "learning_rate": 3.1344792719919116e-05, + "loss": 1.2325, + "step": 4960 + }, + { + "epoch": 1.01, + "learning_rate": 3.1407987866531855e-05, + "loss": 2.0537, + "step": 4970 + }, + { + "epoch": 1.01, + "learning_rate": 3.147118301314459e-05, + "loss": 1.2909, + "step": 4980 + }, + { + "epoch": 1.01, + "learning_rate": 3.153437815975733e-05, + "loss": 1.5425, + "step": 4990 + }, + { + "epoch": 1.01, + "learning_rate": 3.159757330637007e-05, + "loss": 2.1147, + "step": 5000 + }, + { + "epoch": 1.01, + "learning_rate": 3.166076845298281e-05, + "loss": 2.2409, + "step": 5010 + }, + { + "epoch": 1.01, + "learning_rate": 3.172396359959555e-05, + "loss": 2.1161, + "step": 5020 + }, + { + "epoch": 1.01, + "learning_rate": 3.1787158746208296e-05, + "loss": 1.4589, + "step": 5030 + }, + { + "epoch": 1.01, + "learning_rate": 3.1850353892821035e-05, + "loss": 1.3382, + "step": 5040 + }, + { + "epoch": 1.01, + "learning_rate": 3.1913549039433774e-05, + "loss": 1.5513, + "step": 5050 + }, + { + "epoch": 1.01, + "learning_rate": 3.197674418604651e-05, + "loss": 1.504, + "step": 5060 + }, + { + "epoch": 1.01, + "learning_rate": 3.203993933265925e-05, + "loss": 1.8869, + "step": 5070 + }, + { + "epoch": 1.01, + "learning_rate": 3.2103134479272e-05, + "loss": 1.7294, + "step": 5080 + }, + { + "epoch": 1.01, + "learning_rate": 3.216632962588473e-05, + "loss": 1.5804, + "step": 5090 + }, + { + "epoch": 1.01, + "learning_rate": 3.222952477249747e-05, + "loss": 1.4326, + "step": 5100 + }, + { + "epoch": 1.01, + "learning_rate": 3.2292719919110215e-05, + "loss": 1.0694, + "step": 5110 + }, + { + "epoch": 1.01, + "learning_rate": 3.2355915065722954e-05, + "loss": 1.5897, + "step": 5120 + }, + { + "epoch": 1.01, + "learning_rate": 3.241911021233569e-05, + "loss": 1.4416, + "step": 5130 + }, + { + "epoch": 1.01, + "learning_rate": 3.248230535894843e-05, + "loss": 1.6944, + "step": 5140 + }, + { + "epoch": 1.02, + "learning_rate": 3.254550050556117e-05, + "loss": 1.5548, + "step": 5150 + }, + { + "epoch": 1.02, + "learning_rate": 3.260869565217392e-05, + "loss": 1.6551, + "step": 5160 + }, + { + "epoch": 1.02, + "learning_rate": 3.2671890798786656e-05, + "loss": 1.8062, + "step": 5170 + }, + { + "epoch": 1.02, + "learning_rate": 3.2735085945399395e-05, + "loss": 1.6986, + "step": 5180 + }, + { + "epoch": 1.02, + "learning_rate": 3.2798281092012134e-05, + "loss": 1.8364, + "step": 5190 + }, + { + "epoch": 1.02, + "learning_rate": 3.286147623862488e-05, + "loss": 1.5671, + "step": 5200 + }, + { + "epoch": 1.02, + "learning_rate": 3.292467138523761e-05, + "loss": 1.1343, + "step": 5210 + }, + { + "epoch": 1.02, + "learning_rate": 3.298786653185035e-05, + "loss": 1.5405, + "step": 5220 + }, + { + "epoch": 1.02, + "learning_rate": 3.30510616784631e-05, + "loss": 1.4792, + "step": 5230 + }, + { + "epoch": 1.02, + "learning_rate": 3.3114256825075836e-05, + "loss": 1.8239, + "step": 5240 + }, + { + "epoch": 1.02, + "learning_rate": 3.3177451971688575e-05, + "loss": 1.5594, + "step": 5250 + }, + { + "epoch": 1.02, + "learning_rate": 3.3240647118301314e-05, + "loss": 1.5367, + "step": 5260 + }, + { + "epoch": 1.02, + "learning_rate": 3.330384226491405e-05, + "loss": 1.996, + "step": 5270 + }, + { + "epoch": 1.02, + "learning_rate": 3.33670374115268e-05, + "loss": 1.6591, + "step": 5280 + }, + { + "epoch": 1.02, + "learning_rate": 3.343023255813954e-05, + "loss": 1.3933, + "step": 5290 + }, + { + "epoch": 1.02, + "learning_rate": 3.349342770475228e-05, + "loss": 1.9823, + "step": 5300 + }, + { + "epoch": 1.02, + "learning_rate": 3.3556622851365016e-05, + "loss": 1.974, + "step": 5310 + }, + { + "epoch": 1.02, + "learning_rate": 3.361981799797776e-05, + "loss": 1.2526, + "step": 5320 + }, + { + "epoch": 1.02, + "learning_rate": 3.3683013144590494e-05, + "loss": 1.6057, + "step": 5330 + }, + { + "epoch": 1.02, + "learning_rate": 3.374620829120323e-05, + "loss": 1.2758, + "step": 5340 + }, + { + "epoch": 1.02, + "learning_rate": 3.380940343781598e-05, + "loss": 2.1118, + "step": 5350 + }, + { + "epoch": 1.02, + "learning_rate": 3.387259858442872e-05, + "loss": 1.7469, + "step": 5360 + }, + { + "epoch": 1.02, + "learning_rate": 3.393579373104146e-05, + "loss": 1.7399, + "step": 5370 + }, + { + "epoch": 1.02, + "learning_rate": 3.3998988877654196e-05, + "loss": 1.2015, + "step": 5380 + }, + { + "epoch": 1.02, + "learning_rate": 3.4062184024266935e-05, + "loss": 1.5381, + "step": 5390 + }, + { + "epoch": 1.02, + "learning_rate": 3.412537917087968e-05, + "loss": 1.3604, + "step": 5400 + }, + { + "epoch": 1.02, + "learning_rate": 3.418857431749242e-05, + "loss": 1.8152, + "step": 5410 + }, + { + "epoch": 1.02, + "learning_rate": 3.425176946410516e-05, + "loss": 1.5965, + "step": 5420 + }, + { + "epoch": 1.02, + "learning_rate": 3.43149646107179e-05, + "loss": 1.668, + "step": 5430 + }, + { + "epoch": 1.02, + "learning_rate": 3.4378159757330644e-05, + "loss": 1.4882, + "step": 5440 + }, + { + "epoch": 1.02, + "learning_rate": 3.4441354903943376e-05, + "loss": 1.6251, + "step": 5450 + }, + { + "epoch": 1.02, + "learning_rate": 3.4504550050556115e-05, + "loss": 1.6572, + "step": 5460 + }, + { + "epoch": 1.02, + "learning_rate": 3.456774519716886e-05, + "loss": 1.6473, + "step": 5470 + }, + { + "epoch": 1.02, + "learning_rate": 3.46309403437816e-05, + "loss": 1.7867, + "step": 5480 + }, + { + "epoch": 1.02, + "learning_rate": 3.469413549039434e-05, + "loss": 1.7049, + "step": 5490 + }, + { + "epoch": 1.02, + "learning_rate": 3.475733063700708e-05, + "loss": 1.7788, + "step": 5500 + }, + { + "epoch": 1.02, + "learning_rate": 3.482052578361982e-05, + "loss": 1.5405, + "step": 5510 + }, + { + "epoch": 1.02, + "learning_rate": 3.488372093023256e-05, + "loss": 1.7232, + "step": 5520 + }, + { + "epoch": 1.02, + "learning_rate": 3.49469160768453e-05, + "loss": 1.3913, + "step": 5530 + }, + { + "epoch": 1.02, + "learning_rate": 3.501011122345804e-05, + "loss": 1.2379, + "step": 5540 + }, + { + "epoch": 1.02, + "learning_rate": 3.507330637007078e-05, + "loss": 1.7427, + "step": 5550 + }, + { + "epoch": 1.02, + "learning_rate": 3.513650151668352e-05, + "loss": 1.5744, + "step": 5560 + }, + { + "epoch": 1.02, + "learning_rate": 3.519969666329626e-05, + "loss": 1.9389, + "step": 5570 + }, + { + "epoch": 1.02, + "learning_rate": 3.5262891809909e-05, + "loss": 1.436, + "step": 5580 + }, + { + "epoch": 1.02, + "learning_rate": 3.532608695652174e-05, + "loss": 1.6208, + "step": 5590 + }, + { + "epoch": 1.02, + "learning_rate": 3.538928210313448e-05, + "loss": 1.782, + "step": 5600 + }, + { + "epoch": 1.02, + "learning_rate": 3.545247724974722e-05, + "loss": 1.6182, + "step": 5610 + }, + { + "epoch": 1.02, + "learning_rate": 3.551567239635996e-05, + "loss": 1.1525, + "step": 5620 + }, + { + "epoch": 1.02, + "learning_rate": 3.55788675429727e-05, + "loss": 1.2885, + "step": 5630 + }, + { + "epoch": 1.02, + "learning_rate": 3.5642062689585445e-05, + "loss": 1.7939, + "step": 5640 + }, + { + "epoch": 1.02, + "learning_rate": 3.5705257836198184e-05, + "loss": 1.6161, + "step": 5650 + }, + { + "epoch": 1.02, + "learning_rate": 3.576845298281092e-05, + "loss": 1.6074, + "step": 5660 + }, + { + "epoch": 1.02, + "learning_rate": 3.583164812942366e-05, + "loss": 1.6745, + "step": 5670 + }, + { + "epoch": 1.02, + "learning_rate": 3.58948432760364e-05, + "loss": 1.6982, + "step": 5680 + }, + { + "epoch": 1.02, + "learning_rate": 3.595803842264914e-05, + "loss": 1.7199, + "step": 5690 + }, + { + "epoch": 1.02, + "learning_rate": 3.602123356926188e-05, + "loss": 1.3847, + "step": 5700 + }, + { + "epoch": 1.02, + "learning_rate": 3.6084428715874625e-05, + "loss": 1.8448, + "step": 5710 + }, + { + "epoch": 1.02, + "learning_rate": 3.6147623862487364e-05, + "loss": 1.142, + "step": 5720 + }, + { + "epoch": 1.02, + "learning_rate": 3.62108190091001e-05, + "loss": 1.5672, + "step": 5730 + }, + { + "epoch": 1.02, + "learning_rate": 3.627401415571284e-05, + "loss": 1.4054, + "step": 5740 + }, + { + "epoch": 1.02, + "learning_rate": 3.633720930232558e-05, + "loss": 1.9846, + "step": 5750 + }, + { + "epoch": 1.02, + "learning_rate": 3.640040444893833e-05, + "loss": 1.4499, + "step": 5760 + }, + { + "epoch": 1.02, + "learning_rate": 3.646359959555106e-05, + "loss": 1.5732, + "step": 5770 + }, + { + "epoch": 1.02, + "learning_rate": 3.6526794742163805e-05, + "loss": 1.616, + "step": 5780 + }, + { + "epoch": 1.02, + "learning_rate": 3.6589989888776544e-05, + "loss": 1.4784, + "step": 5790 + }, + { + "epoch": 1.02, + "learning_rate": 3.665318503538928e-05, + "loss": 1.0925, + "step": 5800 + }, + { + "epoch": 1.02, + "learning_rate": 3.671638018200202e-05, + "loss": 1.5826, + "step": 5810 + }, + { + "epoch": 1.02, + "learning_rate": 3.677957532861476e-05, + "loss": 2.0011, + "step": 5820 + }, + { + "epoch": 1.02, + "learning_rate": 3.684277047522751e-05, + "loss": 1.4164, + "step": 5830 + }, + { + "epoch": 1.02, + "learning_rate": 3.6905965621840246e-05, + "loss": 1.4582, + "step": 5840 + }, + { + "epoch": 1.02, + "learning_rate": 3.6969160768452985e-05, + "loss": 1.7216, + "step": 5850 + }, + { + "epoch": 1.02, + "learning_rate": 3.7032355915065724e-05, + "loss": 1.4333, + "step": 5860 + }, + { + "epoch": 1.02, + "learning_rate": 3.709555106167846e-05, + "loss": 1.7032, + "step": 5870 + }, + { + "epoch": 1.02, + "learning_rate": 3.715874620829121e-05, + "loss": 1.4221, + "step": 5880 + }, + { + "epoch": 1.02, + "learning_rate": 3.722194135490394e-05, + "loss": 1.3474, + "step": 5890 + }, + { + "epoch": 1.02, + "learning_rate": 3.728513650151669e-05, + "loss": 1.1665, + "step": 5900 + }, + { + "epoch": 1.02, + "learning_rate": 3.7348331648129426e-05, + "loss": 1.6232, + "step": 5910 + }, + { + "epoch": 1.02, + "learning_rate": 3.7411526794742165e-05, + "loss": 0.97, + "step": 5920 + }, + { + "epoch": 1.02, + "learning_rate": 3.7474721941354904e-05, + "loss": 1.6169, + "step": 5930 + }, + { + "epoch": 1.03, + "learning_rate": 3.753791708796764e-05, + "loss": 1.3697, + "step": 5940 + }, + { + "epoch": 1.03, + "learning_rate": 3.760111223458039e-05, + "loss": 0.8898, + "step": 5950 + }, + { + "epoch": 1.03, + "learning_rate": 3.766430738119313e-05, + "loss": 1.3221, + "step": 5960 + }, + { + "epoch": 1.03, + "learning_rate": 3.772750252780587e-05, + "loss": 1.3691, + "step": 5970 + }, + { + "epoch": 1.03, + "learning_rate": 3.7790697674418606e-05, + "loss": 1.9215, + "step": 5980 + }, + { + "epoch": 1.03, + "learning_rate": 3.7853892821031345e-05, + "loss": 1.7605, + "step": 5990 + }, + { + "epoch": 1.03, + "learning_rate": 3.791708796764409e-05, + "loss": 0.9549, + "step": 6000 + }, + { + "epoch": 1.03, + "learning_rate": 3.798028311425682e-05, + "loss": 1.0527, + "step": 6010 + }, + { + "epoch": 1.03, + "learning_rate": 3.804347826086957e-05, + "loss": 1.2393, + "step": 6020 + }, + { + "epoch": 1.03, + "learning_rate": 3.810667340748231e-05, + "loss": 1.2701, + "step": 6030 + }, + { + "epoch": 1.03, + "learning_rate": 3.816986855409505e-05, + "loss": 1.2478, + "step": 6040 + }, + { + "epoch": 1.03, + "learning_rate": 3.8233063700707786e-05, + "loss": 1.6303, + "step": 6050 + }, + { + "epoch": 1.03, + "learning_rate": 3.8296258847320525e-05, + "loss": 1.3766, + "step": 6060 + }, + { + "epoch": 1.03, + "learning_rate": 3.835945399393327e-05, + "loss": 1.5025, + "step": 6070 + }, + { + "epoch": 1.03, + "learning_rate": 3.842264914054601e-05, + "loss": 1.2978, + "step": 6080 + }, + { + "epoch": 1.03, + "learning_rate": 3.848584428715875e-05, + "loss": 1.358, + "step": 6090 + }, + { + "epoch": 1.03, + "learning_rate": 3.854903943377149e-05, + "loss": 1.1737, + "step": 6100 + }, + { + "epoch": 1.03, + "learning_rate": 3.861223458038423e-05, + "loss": 1.3736, + "step": 6110 + }, + { + "epoch": 1.03, + "learning_rate": 3.867542972699697e-05, + "loss": 1.3223, + "step": 6120 + }, + { + "epoch": 1.03, + "learning_rate": 3.8738624873609705e-05, + "loss": 1.0108, + "step": 6130 + }, + { + "epoch": 1.03, + "learning_rate": 3.880182002022245e-05, + "loss": 1.2669, + "step": 6140 + }, + { + "epoch": 1.03, + "learning_rate": 3.886501516683519e-05, + "loss": 1.0154, + "step": 6150 + }, + { + "epoch": 1.03, + "learning_rate": 3.892821031344793e-05, + "loss": 1.7241, + "step": 6160 + }, + { + "epoch": 1.03, + "learning_rate": 3.899140546006067e-05, + "loss": 1.4356, + "step": 6170 + }, + { + "epoch": 1.03, + "learning_rate": 3.905460060667341e-05, + "loss": 1.3747, + "step": 6180 + }, + { + "epoch": 1.03, + "learning_rate": 3.911779575328615e-05, + "loss": 1.0838, + "step": 6190 + }, + { + "epoch": 1.03, + "learning_rate": 3.918099089989889e-05, + "loss": 1.2585, + "step": 6200 + }, + { + "epoch": 1.03, + "learning_rate": 3.924418604651163e-05, + "loss": 1.1956, + "step": 6210 + }, + { + "epoch": 1.03, + "learning_rate": 3.930738119312437e-05, + "loss": 1.5237, + "step": 6220 + }, + { + "epoch": 1.03, + "learning_rate": 3.937057633973711e-05, + "loss": 1.7448, + "step": 6230 + }, + { + "epoch": 1.03, + "learning_rate": 3.9433771486349854e-05, + "loss": 1.2874, + "step": 6240 + }, + { + "epoch": 1.03, + "learning_rate": 3.949696663296259e-05, + "loss": 1.486, + "step": 6250 + }, + { + "epoch": 1.03, + "learning_rate": 3.956016177957533e-05, + "loss": 1.3999, + "step": 6260 + }, + { + "epoch": 1.03, + "learning_rate": 3.962335692618807e-05, + "loss": 0.8174, + "step": 6270 + }, + { + "epoch": 1.03, + "learning_rate": 3.968655207280081e-05, + "loss": 1.7789, + "step": 6280 + }, + { + "epoch": 1.03, + "learning_rate": 3.974974721941355e-05, + "loss": 1.3221, + "step": 6290 + }, + { + "epoch": 1.03, + "learning_rate": 3.981294236602629e-05, + "loss": 0.9945, + "step": 6300 + }, + { + "epoch": 1.03, + "learning_rate": 3.9876137512639034e-05, + "loss": 1.0097, + "step": 6310 + }, + { + "epoch": 1.03, + "learning_rate": 3.9939332659251774e-05, + "loss": 1.5705, + "step": 6320 + }, + { + "epoch": 1.03, + "learning_rate": 4.000252780586451e-05, + "loss": 1.6138, + "step": 6330 + }, + { + "epoch": 1.03, + "learning_rate": 4.006572295247725e-05, + "loss": 0.9936, + "step": 6340 + }, + { + "epoch": 1.03, + "learning_rate": 4.012891809908999e-05, + "loss": 1.7602, + "step": 6350 + }, + { + "epoch": 1.03, + "learning_rate": 4.0192113245702736e-05, + "loss": 1.2059, + "step": 6360 + }, + { + "epoch": 1.03, + "learning_rate": 4.025530839231547e-05, + "loss": 1.3485, + "step": 6370 + }, + { + "epoch": 1.03, + "learning_rate": 4.031850353892821e-05, + "loss": 1.3269, + "step": 6380 + }, + { + "epoch": 1.03, + "learning_rate": 4.0381698685540954e-05, + "loss": 1.1835, + "step": 6390 + }, + { + "epoch": 1.03, + "learning_rate": 4.044489383215369e-05, + "loss": 1.5449, + "step": 6400 + }, + { + "epoch": 1.03, + "learning_rate": 4.050808897876643e-05, + "loss": 1.0622, + "step": 6410 + }, + { + "epoch": 1.03, + "learning_rate": 4.057128412537917e-05, + "loss": 1.1236, + "step": 6420 + }, + { + "epoch": 1.03, + "learning_rate": 4.0634479271991916e-05, + "loss": 1.1591, + "step": 6430 + }, + { + "epoch": 1.03, + "learning_rate": 4.0697674418604655e-05, + "loss": 1.0707, + "step": 6440 + }, + { + "epoch": 1.03, + "learning_rate": 4.076086956521739e-05, + "loss": 1.0028, + "step": 6450 + }, + { + "epoch": 1.03, + "learning_rate": 4.0824064711830134e-05, + "loss": 1.484, + "step": 6460 + }, + { + "epoch": 1.03, + "learning_rate": 4.088725985844287e-05, + "loss": 1.2171, + "step": 6470 + }, + { + "epoch": 1.03, + "learning_rate": 4.095045500505562e-05, + "loss": 1.4152, + "step": 6480 + }, + { + "epoch": 1.03, + "learning_rate": 4.101365015166835e-05, + "loss": 1.2687, + "step": 6490 + }, + { + "epoch": 1.03, + "learning_rate": 4.107684529828109e-05, + "loss": 0.9454, + "step": 6500 + }, + { + "epoch": 1.03, + "learning_rate": 4.1140040444893835e-05, + "loss": 1.6375, + "step": 6510 + }, + { + "epoch": 1.03, + "learning_rate": 4.1203235591506575e-05, + "loss": 1.1483, + "step": 6520 + }, + { + "epoch": 1.03, + "learning_rate": 4.1266430738119314e-05, + "loss": 1.035, + "step": 6530 + }, + { + "epoch": 1.03, + "learning_rate": 4.132962588473205e-05, + "loss": 1.6631, + "step": 6540 + }, + { + "epoch": 1.03, + "learning_rate": 4.13928210313448e-05, + "loss": 0.7222, + "step": 6550 + }, + { + "epoch": 1.03, + "learning_rate": 4.145601617795754e-05, + "loss": 1.1543, + "step": 6560 + }, + { + "epoch": 1.03, + "learning_rate": 4.151921132457027e-05, + "loss": 1.198, + "step": 6570 + }, + { + "epoch": 1.03, + "learning_rate": 4.1582406471183015e-05, + "loss": 1.1302, + "step": 6580 + }, + { + "epoch": 1.03, + "learning_rate": 4.1645601617795755e-05, + "loss": 1.1208, + "step": 6590 + }, + { + "epoch": 1.03, + "learning_rate": 4.17087967644085e-05, + "loss": 0.8966, + "step": 6600 + }, + { + "epoch": 1.03, + "learning_rate": 4.177199191102123e-05, + "loss": 1.1334, + "step": 6610 + }, + { + "epoch": 1.03, + "learning_rate": 4.183518705763397e-05, + "loss": 1.2916, + "step": 6620 + }, + { + "epoch": 1.03, + "learning_rate": 4.189838220424672e-05, + "loss": 1.5907, + "step": 6630 + }, + { + "epoch": 1.03, + "learning_rate": 4.1961577350859456e-05, + "loss": 1.4373, + "step": 6640 + }, + { + "epoch": 1.03, + "learning_rate": 4.2024772497472195e-05, + "loss": 1.0371, + "step": 6650 + }, + { + "epoch": 1.03, + "learning_rate": 4.2087967644084935e-05, + "loss": 1.5451, + "step": 6660 + }, + { + "epoch": 1.03, + "learning_rate": 4.215116279069768e-05, + "loss": 1.6836, + "step": 6670 + }, + { + "epoch": 1.03, + "learning_rate": 4.221435793731042e-05, + "loss": 1.5367, + "step": 6680 + }, + { + "epoch": 1.03, + "learning_rate": 4.227755308392315e-05, + "loss": 1.3506, + "step": 6690 + }, + { + "epoch": 1.03, + "learning_rate": 4.23407482305359e-05, + "loss": 1.2998, + "step": 6700 + }, + { + "epoch": 1.03, + "learning_rate": 4.2403943377148636e-05, + "loss": 1.02, + "step": 6710 + }, + { + "epoch": 1.03, + "learning_rate": 4.246713852376138e-05, + "loss": 1.8236, + "step": 6720 + }, + { + "epoch": 1.04, + "learning_rate": 4.2530333670374115e-05, + "loss": 1.2729, + "step": 6730 + }, + { + "epoch": 1.04, + "learning_rate": 4.2593528816986854e-05, + "loss": 0.8606, + "step": 6740 + }, + { + "epoch": 1.04, + "learning_rate": 4.26567239635996e-05, + "loss": 1.4699, + "step": 6750 + }, + { + "epoch": 1.04, + "learning_rate": 4.271991911021234e-05, + "loss": 1.1563, + "step": 6760 + }, + { + "epoch": 1.04, + "learning_rate": 4.278311425682508e-05, + "loss": 1.263, + "step": 6770 + }, + { + "epoch": 1.04, + "learning_rate": 4.2846309403437816e-05, + "loss": 1.5499, + "step": 6780 + }, + { + "epoch": 1.04, + "learning_rate": 4.2909504550050556e-05, + "loss": 1.0288, + "step": 6790 + }, + { + "epoch": 1.04, + "learning_rate": 4.29726996966633e-05, + "loss": 1.1736, + "step": 6800 + }, + { + "epoch": 1.04, + "learning_rate": 4.3035894843276034e-05, + "loss": 1.001, + "step": 6810 + }, + { + "epoch": 1.04, + "learning_rate": 4.309908998988878e-05, + "loss": 1.1109, + "step": 6820 + }, + { + "epoch": 1.04, + "learning_rate": 4.316228513650152e-05, + "loss": 0.6067, + "step": 6830 + }, + { + "epoch": 1.04, + "learning_rate": 4.3225480283114264e-05, + "loss": 1.0773, + "step": 6840 + }, + { + "epoch": 1.04, + "learning_rate": 4.3288675429726996e-05, + "loss": 1.2838, + "step": 6850 + }, + { + "epoch": 1.04, + "learning_rate": 4.3351870576339736e-05, + "loss": 1.4128, + "step": 6860 + }, + { + "epoch": 1.04, + "learning_rate": 4.341506572295248e-05, + "loss": 1.3133, + "step": 6870 + }, + { + "epoch": 1.04, + "learning_rate": 4.347826086956522e-05, + "loss": 1.1862, + "step": 6880 + }, + { + "epoch": 1.04, + "learning_rate": 4.354145601617796e-05, + "loss": 0.9493, + "step": 6890 + }, + { + "epoch": 1.04, + "learning_rate": 4.36046511627907e-05, + "loss": 1.2753, + "step": 6900 + }, + { + "epoch": 1.04, + "learning_rate": 4.366784630940344e-05, + "loss": 1.2221, + "step": 6910 + }, + { + "epoch": 1.04, + "learning_rate": 4.373104145601618e-05, + "loss": 0.7976, + "step": 6920 + }, + { + "epoch": 1.04, + "learning_rate": 4.3794236602628916e-05, + "loss": 1.3584, + "step": 6930 + }, + { + "epoch": 1.04, + "learning_rate": 4.385743174924166e-05, + "loss": 1.8375, + "step": 6940 + }, + { + "epoch": 1.04, + "learning_rate": 4.39206268958544e-05, + "loss": 0.6735, + "step": 6950 + }, + { + "epoch": 1.04, + "learning_rate": 4.3983822042467146e-05, + "loss": 1.1708, + "step": 6960 + }, + { + "epoch": 1.04, + "learning_rate": 4.404701718907988e-05, + "loss": 0.9881, + "step": 6970 + }, + { + "epoch": 1.04, + "learning_rate": 4.411021233569262e-05, + "loss": 1.3723, + "step": 6980 + }, + { + "epoch": 1.04, + "learning_rate": 4.417340748230536e-05, + "loss": 1.0193, + "step": 6990 + }, + { + "epoch": 1.04, + "learning_rate": 4.42366026289181e-05, + "loss": 0.7854, + "step": 7000 + }, + { + "epoch": 1.04, + "learning_rate": 4.429979777553084e-05, + "loss": 0.4923, + "step": 7010 + }, + { + "epoch": 1.04, + "learning_rate": 4.436299292214358e-05, + "loss": 1.1542, + "step": 7020 + }, + { + "epoch": 1.04, + "learning_rate": 4.442618806875632e-05, + "loss": 1.1944, + "step": 7030 + }, + { + "epoch": 1.04, + "learning_rate": 4.4489383215369065e-05, + "loss": 1.7082, + "step": 7040 + }, + { + "epoch": 1.04, + "learning_rate": 4.45525783619818e-05, + "loss": 1.5915, + "step": 7050 + }, + { + "epoch": 1.04, + "learning_rate": 4.461577350859454e-05, + "loss": 1.5452, + "step": 7060 + }, + { + "epoch": 1.04, + "learning_rate": 4.467896865520728e-05, + "loss": 0.7324, + "step": 7070 + }, + { + "epoch": 1.04, + "learning_rate": 4.474216380182003e-05, + "loss": 0.8718, + "step": 7080 + }, + { + "epoch": 1.04, + "learning_rate": 4.480535894843276e-05, + "loss": 1.1223, + "step": 7090 + }, + { + "epoch": 1.04, + "learning_rate": 4.48685540950455e-05, + "loss": 1.0553, + "step": 7100 + }, + { + "epoch": 1.04, + "learning_rate": 4.4931749241658245e-05, + "loss": 1.0315, + "step": 7110 + }, + { + "epoch": 1.04, + "learning_rate": 4.4994944388270984e-05, + "loss": 0.9717, + "step": 7120 + }, + { + "epoch": 1.04, + "learning_rate": 4.505813953488372e-05, + "loss": 0.7506, + "step": 7130 + }, + { + "epoch": 1.04, + "learning_rate": 4.512133468149646e-05, + "loss": 1.2199, + "step": 7140 + }, + { + "epoch": 1.04, + "learning_rate": 4.51845298281092e-05, + "loss": 1.0544, + "step": 7150 + }, + { + "epoch": 1.04, + "learning_rate": 4.524772497472195e-05, + "loss": 1.4411, + "step": 7160 + }, + { + "epoch": 1.04, + "learning_rate": 4.531092012133468e-05, + "loss": 0.8859, + "step": 7170 + }, + { + "epoch": 1.04, + "learning_rate": 4.5374115267947425e-05, + "loss": 1.4895, + "step": 7180 + }, + { + "epoch": 1.04, + "learning_rate": 4.5437310414560164e-05, + "loss": 1.4667, + "step": 7190 + }, + { + "epoch": 1.04, + "learning_rate": 4.55005055611729e-05, + "loss": 1.5267, + "step": 7200 + }, + { + "epoch": 1.04, + "learning_rate": 4.556370070778564e-05, + "loss": 1.1389, + "step": 7210 + }, + { + "epoch": 1.04, + "learning_rate": 4.562689585439838e-05, + "loss": 1.193, + "step": 7220 + }, + { + "epoch": 1.04, + "learning_rate": 4.569009100101113e-05, + "loss": 1.2311, + "step": 7230 + }, + { + "epoch": 1.04, + "learning_rate": 4.5753286147623866e-05, + "loss": 1.1444, + "step": 7240 + }, + { + "epoch": 1.04, + "learning_rate": 4.5816481294236605e-05, + "loss": 0.9953, + "step": 7250 + }, + { + "epoch": 1.04, + "learning_rate": 4.5879676440849344e-05, + "loss": 0.7042, + "step": 7260 + }, + { + "epoch": 1.04, + "learning_rate": 4.594287158746208e-05, + "loss": 0.7985, + "step": 7270 + }, + { + "epoch": 1.04, + "learning_rate": 4.600606673407483e-05, + "loss": 1.3149, + "step": 7280 + }, + { + "epoch": 1.04, + "learning_rate": 4.606926188068756e-05, + "loss": 0.9187, + "step": 7290 + }, + { + "epoch": 1.04, + "learning_rate": 4.613245702730031e-05, + "loss": 0.415, + "step": 7300 + }, + { + "epoch": 1.04, + "learning_rate": 4.6195652173913046e-05, + "loss": 0.702, + "step": 7310 + }, + { + "epoch": 1.04, + "learning_rate": 4.6258847320525785e-05, + "loss": 0.7427, + "step": 7320 + }, + { + "epoch": 1.04, + "learning_rate": 4.6322042467138524e-05, + "loss": 1.2195, + "step": 7330 + }, + { + "epoch": 1.04, + "learning_rate": 4.638523761375126e-05, + "loss": 1.1088, + "step": 7340 + }, + { + "epoch": 1.04, + "learning_rate": 4.644843276036401e-05, + "loss": 1.1246, + "step": 7350 + }, + { + "epoch": 1.04, + "learning_rate": 4.651162790697675e-05, + "loss": 0.8654, + "step": 7360 + }, + { + "epoch": 1.04, + "learning_rate": 4.657482305358949e-05, + "loss": 0.8603, + "step": 7370 + }, + { + "epoch": 1.04, + "learning_rate": 4.6638018200202226e-05, + "loss": 1.1118, + "step": 7380 + }, + { + "epoch": 1.04, + "learning_rate": 4.6701213346814965e-05, + "loss": 1.2194, + "step": 7390 + }, + { + "epoch": 1.04, + "learning_rate": 4.676440849342771e-05, + "loss": 0.6745, + "step": 7400 + }, + { + "epoch": 1.04, + "learning_rate": 4.682760364004044e-05, + "loss": 0.7672, + "step": 7410 + }, + { + "epoch": 1.04, + "learning_rate": 4.689079878665319e-05, + "loss": 0.6604, + "step": 7420 + }, + { + "epoch": 1.04, + "learning_rate": 4.695399393326593e-05, + "loss": 1.4772, + "step": 7430 + }, + { + "epoch": 1.04, + "learning_rate": 4.701718907987867e-05, + "loss": 0.8097, + "step": 7440 + }, + { + "epoch": 1.04, + "learning_rate": 4.7080384226491406e-05, + "loss": 1.1015, + "step": 7450 + }, + { + "epoch": 1.04, + "learning_rate": 4.7143579373104145e-05, + "loss": 1.6311, + "step": 7460 + }, + { + "epoch": 1.04, + "learning_rate": 4.720677451971689e-05, + "loss": 1.3734, + "step": 7470 + }, + { + "epoch": 1.04, + "learning_rate": 4.726996966632963e-05, + "loss": 1.1042, + "step": 7480 + }, + { + "epoch": 1.04, + "learning_rate": 4.733316481294237e-05, + "loss": 1.1821, + "step": 7490 + }, + { + "epoch": 1.04, + "learning_rate": 4.739635995955511e-05, + "loss": 1.1527, + "step": 7500 + }, + { + "epoch": 1.04, + "learning_rate": 4.745955510616785e-05, + "loss": 0.6089, + "step": 7510 + }, + { + "epoch": 1.05, + "learning_rate": 4.752275025278059e-05, + "loss": 1.0304, + "step": 7520 + }, + { + "epoch": 1.05, + "learning_rate": 4.7585945399393325e-05, + "loss": 1.2711, + "step": 7530 + }, + { + "epoch": 1.05, + "learning_rate": 4.764914054600607e-05, + "loss": 1.0455, + "step": 7540 + }, + { + "epoch": 1.05, + "learning_rate": 4.771233569261881e-05, + "loss": 0.7922, + "step": 7550 + }, + { + "epoch": 1.05, + "learning_rate": 4.777553083923155e-05, + "loss": 0.8806, + "step": 7560 + }, + { + "epoch": 1.05, + "learning_rate": 4.783872598584429e-05, + "loss": 0.8546, + "step": 7570 + }, + { + "epoch": 1.05, + "learning_rate": 4.790192113245703e-05, + "loss": 1.3455, + "step": 7580 + }, + { + "epoch": 1.05, + "learning_rate": 4.796511627906977e-05, + "loss": 2.1216, + "step": 7590 + }, + { + "epoch": 1.05, + "learning_rate": 4.802831142568251e-05, + "loss": 1.603, + "step": 7600 + }, + { + "epoch": 1.05, + "learning_rate": 4.8091506572295244e-05, + "loss": 1.3879, + "step": 7610 + }, + { + "epoch": 1.05, + "learning_rate": 4.815470171890799e-05, + "loss": 1.8445, + "step": 7620 + }, + { + "epoch": 1.05, + "learning_rate": 4.821789686552073e-05, + "loss": 1.374, + "step": 7630 + }, + { + "epoch": 1.05, + "learning_rate": 4.8281092012133475e-05, + "loss": 0.8166, + "step": 7640 + }, + { + "epoch": 1.05, + "learning_rate": 4.834428715874621e-05, + "loss": 1.1314, + "step": 7650 + }, + { + "epoch": 1.05, + "learning_rate": 4.840748230535895e-05, + "loss": 1.2528, + "step": 7660 + }, + { + "epoch": 1.05, + "learning_rate": 4.847067745197169e-05, + "loss": 1.2661, + "step": 7670 + }, + { + "epoch": 1.05, + "learning_rate": 4.853387259858443e-05, + "loss": 0.8197, + "step": 7680 + }, + { + "epoch": 1.05, + "learning_rate": 4.859706774519717e-05, + "loss": 1.3669, + "step": 7690 + }, + { + "epoch": 1.05, + "learning_rate": 4.866026289180991e-05, + "loss": 1.0629, + "step": 7700 + }, + { + "epoch": 1.05, + "learning_rate": 4.8723458038422655e-05, + "loss": 0.4791, + "step": 7710 + }, + { + "epoch": 1.05, + "learning_rate": 4.8786653185035394e-05, + "loss": 0.9313, + "step": 7720 + }, + { + "epoch": 1.05, + "learning_rate": 4.8849848331648126e-05, + "loss": 1.0688, + "step": 7730 + }, + { + "epoch": 1.05, + "learning_rate": 4.891304347826087e-05, + "loss": 1.4551, + "step": 7740 + }, + { + "epoch": 1.05, + "learning_rate": 4.897623862487361e-05, + "loss": 0.7147, + "step": 7750 + }, + { + "epoch": 1.05, + "learning_rate": 4.903943377148636e-05, + "loss": 0.5926, + "step": 7760 + }, + { + "epoch": 1.05, + "learning_rate": 4.910262891809909e-05, + "loss": 1.485, + "step": 7770 + }, + { + "epoch": 1.05, + "learning_rate": 4.9165824064711835e-05, + "loss": 1.1985, + "step": 7780 + }, + { + "epoch": 1.05, + "learning_rate": 4.9229019211324574e-05, + "loss": 0.9516, + "step": 7790 + }, + { + "epoch": 1.05, + "learning_rate": 4.929221435793731e-05, + "loss": 0.7388, + "step": 7800 + }, + { + "epoch": 1.05, + "learning_rate": 4.935540950455005e-05, + "loss": 1.7702, + "step": 7810 + }, + { + "epoch": 1.05, + "learning_rate": 4.941860465116279e-05, + "loss": 1.2636, + "step": 7820 + }, + { + "epoch": 1.05, + "learning_rate": 4.948179979777554e-05, + "loss": 0.6471, + "step": 7830 + }, + { + "epoch": 1.05, + "learning_rate": 4.9544994944388276e-05, + "loss": 1.0884, + "step": 7840 + }, + { + "epoch": 1.05, + "learning_rate": 4.960819009100101e-05, + "loss": 0.5902, + "step": 7850 + }, + { + "epoch": 1.05, + "learning_rate": 4.9671385237613754e-05, + "loss": 0.8991, + "step": 7860 + }, + { + "epoch": 1.05, + "learning_rate": 4.973458038422649e-05, + "loss": 1.4636, + "step": 7870 + }, + { + "epoch": 1.05, + "learning_rate": 4.979777553083924e-05, + "loss": 0.9824, + "step": 7880 + }, + { + "epoch": 1.05, + "learning_rate": 4.986097067745197e-05, + "loss": 1.6099, + "step": 7890 + }, + { + "epoch": 1.05, + "learning_rate": 4.992416582406472e-05, + "loss": 0.5584, + "step": 7900 + }, + { + "epoch": 1.05, + "learning_rate": 4.9987360970677456e-05, + "loss": 0.6674, + "step": 7910 + }, + { + "epoch": 1.05, + "eval_accuracy": 0.591536575925687, + "eval_loss": 1.5344293117523193, + "eval_runtime": 1299.3107, + "eval_samples_per_second": 5.965, + "eval_steps_per_second": 2.983, + "step": 7912 + }, + { + "epoch": 2.0, + "learning_rate": 4.999438265363443e-05, + "loss": 0.7109, + "step": 7920 + }, + { + "epoch": 2.0, + "learning_rate": 4.9987360970677456e-05, + "loss": 1.1526, + "step": 7930 + }, + { + "epoch": 2.0, + "learning_rate": 4.9980339287720483e-05, + "loss": 0.7717, + "step": 7940 + }, + { + "epoch": 2.0, + "learning_rate": 4.997331760476352e-05, + "loss": 0.9997, + "step": 7950 + }, + { + "epoch": 2.0, + "learning_rate": 4.996629592180654e-05, + "loss": 1.0304, + "step": 7960 + }, + { + "epoch": 2.0, + "learning_rate": 4.9959274238849566e-05, + "loss": 1.9314, + "step": 7970 + }, + { + "epoch": 2.0, + "learning_rate": 4.99522525558926e-05, + "loss": 0.5047, + "step": 7980 + }, + { + "epoch": 2.0, + "learning_rate": 4.994523087293563e-05, + "loss": 0.9022, + "step": 7990 + }, + { + "epoch": 2.0, + "learning_rate": 4.9938209189978655e-05, + "loss": 0.7496, + "step": 8000 + }, + { + "epoch": 2.0, + "learning_rate": 4.993118750702169e-05, + "loss": 0.3453, + "step": 8010 + }, + { + "epoch": 2.0, + "learning_rate": 4.992416582406472e-05, + "loss": 1.0139, + "step": 8020 + }, + { + "epoch": 2.0, + "learning_rate": 4.9917144141107744e-05, + "loss": 1.3372, + "step": 8030 + }, + { + "epoch": 2.0, + "learning_rate": 4.991012245815077e-05, + "loss": 0.9973, + "step": 8040 + }, + { + "epoch": 2.0, + "learning_rate": 4.99031007751938e-05, + "loss": 1.4845, + "step": 8050 + }, + { + "epoch": 2.0, + "learning_rate": 4.9896079092236834e-05, + "loss": 1.0658, + "step": 8060 + }, + { + "epoch": 2.0, + "learning_rate": 4.988905740927986e-05, + "loss": 0.6433, + "step": 8070 + }, + { + "epoch": 2.0, + "learning_rate": 4.988203572632289e-05, + "loss": 1.1081, + "step": 8080 + }, + { + "epoch": 2.0, + "learning_rate": 4.9875014043365916e-05, + "loss": 0.4482, + "step": 8090 + }, + { + "epoch": 2.0, + "learning_rate": 4.9867992360408944e-05, + "loss": 0.6706, + "step": 8100 + }, + { + "epoch": 2.0, + "learning_rate": 4.986097067745197e-05, + "loss": 0.9956, + "step": 8110 + }, + { + "epoch": 2.0, + "learning_rate": 4.9853948994495005e-05, + "loss": 0.9077, + "step": 8120 + }, + { + "epoch": 2.0, + "learning_rate": 4.984692731153803e-05, + "loss": 0.9749, + "step": 8130 + }, + { + "epoch": 2.0, + "learning_rate": 4.983990562858106e-05, + "loss": 0.7219, + "step": 8140 + }, + { + "epoch": 2.0, + "learning_rate": 4.983288394562409e-05, + "loss": 0.9671, + "step": 8150 + }, + { + "epoch": 2.0, + "learning_rate": 4.9825862262667115e-05, + "loss": 0.6152, + "step": 8160 + }, + { + "epoch": 2.0, + "learning_rate": 4.981884057971015e-05, + "loss": 0.882, + "step": 8170 + }, + { + "epoch": 2.0, + "learning_rate": 4.981181889675318e-05, + "loss": 1.2113, + "step": 8180 + }, + { + "epoch": 2.0, + "learning_rate": 4.9804797213796205e-05, + "loss": 0.7253, + "step": 8190 + }, + { + "epoch": 2.0, + "learning_rate": 4.979777553083924e-05, + "loss": 0.4313, + "step": 8200 + }, + { + "epoch": 2.0, + "learning_rate": 4.9790753847882266e-05, + "loss": 1.5566, + "step": 8210 + }, + { + "epoch": 2.0, + "learning_rate": 4.978373216492529e-05, + "loss": 1.1919, + "step": 8220 + }, + { + "epoch": 2.0, + "learning_rate": 4.977671048196832e-05, + "loss": 1.1472, + "step": 8230 + }, + { + "epoch": 2.0, + "learning_rate": 4.976968879901135e-05, + "loss": 0.4307, + "step": 8240 + }, + { + "epoch": 2.0, + "learning_rate": 4.9762667116054376e-05, + "loss": 1.434, + "step": 8250 + }, + { + "epoch": 2.0, + "learning_rate": 4.975564543309741e-05, + "loss": 0.9392, + "step": 8260 + }, + { + "epoch": 2.0, + "learning_rate": 4.974862375014044e-05, + "loss": 1.4062, + "step": 8270 + }, + { + "epoch": 2.0, + "learning_rate": 4.9741602067183466e-05, + "loss": 0.6974, + "step": 8280 + }, + { + "epoch": 2.0, + "learning_rate": 4.973458038422649e-05, + "loss": 0.6546, + "step": 8290 + }, + { + "epoch": 2.0, + "learning_rate": 4.972755870126952e-05, + "loss": 0.9109, + "step": 8300 + }, + { + "epoch": 2.01, + "learning_rate": 4.9720537018312555e-05, + "loss": 0.6335, + "step": 8310 + }, + { + "epoch": 2.01, + "learning_rate": 4.971351533535558e-05, + "loss": 0.3017, + "step": 8320 + }, + { + "epoch": 2.01, + "learning_rate": 4.970649365239861e-05, + "loss": 1.6681, + "step": 8330 + }, + { + "epoch": 2.01, + "learning_rate": 4.969947196944164e-05, + "loss": 0.6734, + "step": 8340 + }, + { + "epoch": 2.01, + "learning_rate": 4.9692450286484665e-05, + "loss": 1.15, + "step": 8350 + }, + { + "epoch": 2.01, + "learning_rate": 4.968542860352769e-05, + "loss": 0.6318, + "step": 8360 + }, + { + "epoch": 2.01, + "learning_rate": 4.9678406920570727e-05, + "loss": 0.9682, + "step": 8370 + }, + { + "epoch": 2.01, + "learning_rate": 4.9671385237613754e-05, + "loss": 0.8751, + "step": 8380 + }, + { + "epoch": 2.01, + "learning_rate": 4.966436355465678e-05, + "loss": 1.1719, + "step": 8390 + }, + { + "epoch": 2.01, + "learning_rate": 4.9657341871699816e-05, + "loss": 0.732, + "step": 8400 + }, + { + "epoch": 2.01, + "learning_rate": 4.9650320188742836e-05, + "loss": 1.2234, + "step": 8410 + }, + { + "epoch": 2.01, + "learning_rate": 4.964329850578587e-05, + "loss": 0.2968, + "step": 8420 + }, + { + "epoch": 2.01, + "learning_rate": 4.96362768228289e-05, + "loss": 0.6511, + "step": 8430 + }, + { + "epoch": 2.01, + "learning_rate": 4.9629255139871926e-05, + "loss": 1.051, + "step": 8440 + }, + { + "epoch": 2.01, + "learning_rate": 4.962223345691496e-05, + "loss": 1.0669, + "step": 8450 + }, + { + "epoch": 2.01, + "learning_rate": 4.961521177395799e-05, + "loss": 1.0184, + "step": 8460 + }, + { + "epoch": 2.01, + "learning_rate": 4.960819009100101e-05, + "loss": 1.078, + "step": 8470 + }, + { + "epoch": 2.01, + "learning_rate": 4.960116840804404e-05, + "loss": 1.0336, + "step": 8480 + }, + { + "epoch": 2.01, + "learning_rate": 4.959414672508707e-05, + "loss": 0.8304, + "step": 8490 + }, + { + "epoch": 2.01, + "learning_rate": 4.95871250421301e-05, + "loss": 1.0518, + "step": 8500 + }, + { + "epoch": 2.01, + "learning_rate": 4.958010335917313e-05, + "loss": 0.81, + "step": 8510 + }, + { + "epoch": 2.01, + "learning_rate": 4.957308167621616e-05, + "loss": 0.4732, + "step": 8520 + }, + { + "epoch": 2.01, + "learning_rate": 4.956605999325919e-05, + "loss": 1.3087, + "step": 8530 + }, + { + "epoch": 2.01, + "learning_rate": 4.9559038310302214e-05, + "loss": 1.0968, + "step": 8540 + }, + { + "epoch": 2.01, + "learning_rate": 4.955201662734524e-05, + "loss": 0.838, + "step": 8550 + }, + { + "epoch": 2.01, + "learning_rate": 4.9544994944388276e-05, + "loss": 1.7122, + "step": 8560 + }, + { + "epoch": 2.01, + "learning_rate": 4.9537973261431303e-05, + "loss": 0.8689, + "step": 8570 + }, + { + "epoch": 2.01, + "learning_rate": 4.953095157847433e-05, + "loss": 0.5449, + "step": 8580 + }, + { + "epoch": 2.01, + "learning_rate": 4.9523929895517365e-05, + "loss": 0.8814, + "step": 8590 + }, + { + "epoch": 2.01, + "learning_rate": 4.9516908212560386e-05, + "loss": 0.8935, + "step": 8600 + }, + { + "epoch": 2.01, + "learning_rate": 4.9509886529603413e-05, + "loss": 0.4516, + "step": 8610 + }, + { + "epoch": 2.01, + "learning_rate": 4.950286484664645e-05, + "loss": 0.7804, + "step": 8620 + }, + { + "epoch": 2.01, + "learning_rate": 4.9495843163689475e-05, + "loss": 0.8973, + "step": 8630 + }, + { + "epoch": 2.01, + "learning_rate": 4.94888214807325e-05, + "loss": 0.5898, + "step": 8640 + }, + { + "epoch": 2.01, + "learning_rate": 4.948179979777554e-05, + "loss": 0.7066, + "step": 8650 + }, + { + "epoch": 2.01, + "learning_rate": 4.947477811481856e-05, + "loss": 0.7828, + "step": 8660 + }, + { + "epoch": 2.01, + "learning_rate": 4.946775643186159e-05, + "loss": 0.8352, + "step": 8670 + }, + { + "epoch": 2.01, + "learning_rate": 4.946073474890462e-05, + "loss": 0.6758, + "step": 8680 + }, + { + "epoch": 2.01, + "learning_rate": 4.945371306594765e-05, + "loss": 0.9748, + "step": 8690 + }, + { + "epoch": 2.01, + "learning_rate": 4.944669138299068e-05, + "loss": 0.5942, + "step": 8700 + }, + { + "epoch": 2.01, + "learning_rate": 4.943966970003371e-05, + "loss": 0.9475, + "step": 8710 + }, + { + "epoch": 2.01, + "learning_rate": 4.9432648017076736e-05, + "loss": 1.0586, + "step": 8720 + }, + { + "epoch": 2.01, + "learning_rate": 4.9425626334119764e-05, + "loss": 0.8553, + "step": 8730 + }, + { + "epoch": 2.01, + "learning_rate": 4.941860465116279e-05, + "loss": 0.8951, + "step": 8740 + }, + { + "epoch": 2.01, + "learning_rate": 4.941158296820582e-05, + "loss": 0.817, + "step": 8750 + }, + { + "epoch": 2.01, + "learning_rate": 4.940456128524885e-05, + "loss": 0.885, + "step": 8760 + }, + { + "epoch": 2.01, + "learning_rate": 4.939753960229188e-05, + "loss": 0.6457, + "step": 8770 + }, + { + "epoch": 2.01, + "learning_rate": 4.939051791933491e-05, + "loss": 0.9861, + "step": 8780 + }, + { + "epoch": 2.01, + "learning_rate": 4.9383496236377935e-05, + "loss": 0.94, + "step": 8790 + }, + { + "epoch": 2.01, + "learning_rate": 4.937647455342096e-05, + "loss": 0.5432, + "step": 8800 + }, + { + "epoch": 2.01, + "learning_rate": 4.9369452870464e-05, + "loss": 0.6227, + "step": 8810 + }, + { + "epoch": 2.01, + "learning_rate": 4.9362431187507025e-05, + "loss": 0.7659, + "step": 8820 + }, + { + "epoch": 2.01, + "learning_rate": 4.935540950455005e-05, + "loss": 0.3485, + "step": 8830 + }, + { + "epoch": 2.01, + "learning_rate": 4.9348387821593086e-05, + "loss": 1.6669, + "step": 8840 + }, + { + "epoch": 2.01, + "learning_rate": 4.934136613863611e-05, + "loss": 1.2286, + "step": 8850 + }, + { + "epoch": 2.01, + "learning_rate": 4.9334344455679135e-05, + "loss": 1.1936, + "step": 8860 + }, + { + "epoch": 2.01, + "learning_rate": 4.932732277272217e-05, + "loss": 1.4821, + "step": 8870 + }, + { + "epoch": 2.01, + "learning_rate": 4.9320301089765196e-05, + "loss": 1.0694, + "step": 8880 + }, + { + "epoch": 2.01, + "learning_rate": 4.9313279406808224e-05, + "loss": 0.6418, + "step": 8890 + }, + { + "epoch": 2.01, + "learning_rate": 4.930625772385126e-05, + "loss": 0.5328, + "step": 8900 + }, + { + "epoch": 2.01, + "learning_rate": 4.9299236040894286e-05, + "loss": 0.5129, + "step": 8910 + }, + { + "epoch": 2.01, + "learning_rate": 4.929221435793731e-05, + "loss": 1.6964, + "step": 8920 + }, + { + "epoch": 2.01, + "learning_rate": 4.928519267498034e-05, + "loss": 0.8817, + "step": 8930 + }, + { + "epoch": 2.01, + "learning_rate": 4.927817099202337e-05, + "loss": 0.7372, + "step": 8940 + }, + { + "epoch": 2.01, + "learning_rate": 4.92711493090664e-05, + "loss": 1.6331, + "step": 8950 + }, + { + "epoch": 2.01, + "learning_rate": 4.926412762610943e-05, + "loss": 0.7444, + "step": 8960 + }, + { + "epoch": 2.01, + "learning_rate": 4.925710594315246e-05, + "loss": 0.4741, + "step": 8970 + }, + { + "epoch": 2.01, + "learning_rate": 4.9250084260195485e-05, + "loss": 1.1827, + "step": 8980 + }, + { + "epoch": 2.01, + "learning_rate": 4.924306257723851e-05, + "loss": 1.1358, + "step": 8990 + }, + { + "epoch": 2.01, + "learning_rate": 4.9236040894281547e-05, + "loss": 0.954, + "step": 9000 + }, + { + "epoch": 2.01, + "learning_rate": 4.9229019211324574e-05, + "loss": 1.1683, + "step": 9010 + }, + { + "epoch": 2.01, + "learning_rate": 4.92219975283676e-05, + "loss": 0.9467, + "step": 9020 + }, + { + "epoch": 2.01, + "learning_rate": 4.9214975845410636e-05, + "loss": 0.7619, + "step": 9030 + }, + { + "epoch": 2.01, + "learning_rate": 4.9207954162453656e-05, + "loss": 0.5651, + "step": 9040 + }, + { + "epoch": 2.01, + "learning_rate": 4.9200932479496684e-05, + "loss": 0.5229, + "step": 9050 + }, + { + "epoch": 2.01, + "learning_rate": 4.919391079653972e-05, + "loss": 1.1342, + "step": 9060 + }, + { + "epoch": 2.01, + "learning_rate": 4.9186889113582746e-05, + "loss": 0.5606, + "step": 9070 + }, + { + "epoch": 2.01, + "learning_rate": 4.917986743062577e-05, + "loss": 0.7783, + "step": 9080 + }, + { + "epoch": 2.01, + "learning_rate": 4.917284574766881e-05, + "loss": 0.6299, + "step": 9090 + }, + { + "epoch": 2.02, + "learning_rate": 4.9165824064711835e-05, + "loss": 0.4648, + "step": 9100 + }, + { + "epoch": 2.02, + "learning_rate": 4.915880238175486e-05, + "loss": 0.5477, + "step": 9110 + }, + { + "epoch": 2.02, + "learning_rate": 4.915178069879789e-05, + "loss": 0.3168, + "step": 9120 + }, + { + "epoch": 2.02, + "learning_rate": 4.914475901584092e-05, + "loss": 0.3818, + "step": 9130 + }, + { + "epoch": 2.02, + "learning_rate": 4.913773733288395e-05, + "loss": 0.5832, + "step": 9140 + }, + { + "epoch": 2.02, + "learning_rate": 4.913071564992698e-05, + "loss": 0.719, + "step": 9150 + }, + { + "epoch": 2.02, + "learning_rate": 4.912369396697001e-05, + "loss": 0.5045, + "step": 9160 + }, + { + "epoch": 2.02, + "learning_rate": 4.9116672284013034e-05, + "loss": 0.5967, + "step": 9170 + }, + { + "epoch": 2.02, + "learning_rate": 4.910965060105606e-05, + "loss": 1.2943, + "step": 9180 + }, + { + "epoch": 2.02, + "learning_rate": 4.910262891809909e-05, + "loss": 1.0839, + "step": 9190 + }, + { + "epoch": 2.02, + "learning_rate": 4.9095607235142123e-05, + "loss": 0.7102, + "step": 9200 + }, + { + "epoch": 2.02, + "learning_rate": 4.908858555218515e-05, + "loss": 0.5315, + "step": 9210 + }, + { + "epoch": 2.02, + "learning_rate": 4.908156386922818e-05, + "loss": 0.8828, + "step": 9220 + }, + { + "epoch": 2.02, + "learning_rate": 4.9074542186271206e-05, + "loss": 1.0149, + "step": 9230 + }, + { + "epoch": 2.02, + "learning_rate": 4.906752050331423e-05, + "loss": 1.3922, + "step": 9240 + }, + { + "epoch": 2.02, + "learning_rate": 4.906049882035727e-05, + "loss": 0.4923, + "step": 9250 + }, + { + "epoch": 2.02, + "learning_rate": 4.9053477137400295e-05, + "loss": 0.7293, + "step": 9260 + }, + { + "epoch": 2.02, + "learning_rate": 4.904645545444332e-05, + "loss": 0.4422, + "step": 9270 + }, + { + "epoch": 2.02, + "learning_rate": 4.903943377148636e-05, + "loss": 0.4348, + "step": 9280 + }, + { + "epoch": 2.02, + "learning_rate": 4.9032412088529384e-05, + "loss": 0.6886, + "step": 9290 + }, + { + "epoch": 2.02, + "learning_rate": 4.9025390405572405e-05, + "loss": 0.5319, + "step": 9300 + }, + { + "epoch": 2.02, + "learning_rate": 4.901836872261544e-05, + "loss": 0.7447, + "step": 9310 + }, + { + "epoch": 2.02, + "learning_rate": 4.901134703965847e-05, + "loss": 0.853, + "step": 9320 + }, + { + "epoch": 2.02, + "learning_rate": 4.9004325356701494e-05, + "loss": 0.4824, + "step": 9330 + }, + { + "epoch": 2.02, + "learning_rate": 4.899730367374453e-05, + "loss": 1.1617, + "step": 9340 + }, + { + "epoch": 2.02, + "learning_rate": 4.8990281990787556e-05, + "loss": 0.5413, + "step": 9350 + }, + { + "epoch": 2.02, + "learning_rate": 4.8983260307830584e-05, + "loss": 0.8986, + "step": 9360 + }, + { + "epoch": 2.02, + "learning_rate": 4.897623862487361e-05, + "loss": 0.6986, + "step": 9370 + }, + { + "epoch": 2.02, + "learning_rate": 4.896921694191664e-05, + "loss": 0.7204, + "step": 9380 + }, + { + "epoch": 2.02, + "learning_rate": 4.896219525895967e-05, + "loss": 0.3319, + "step": 9390 + }, + { + "epoch": 2.02, + "learning_rate": 4.89551735760027e-05, + "loss": 1.1386, + "step": 9400 + }, + { + "epoch": 2.02, + "learning_rate": 4.894815189304573e-05, + "loss": 0.703, + "step": 9410 + }, + { + "epoch": 2.02, + "learning_rate": 4.8941130210088755e-05, + "loss": 0.1462, + "step": 9420 + }, + { + "epoch": 2.02, + "learning_rate": 4.893410852713178e-05, + "loss": 0.8621, + "step": 9430 + }, + { + "epoch": 2.02, + "learning_rate": 4.892708684417481e-05, + "loss": 1.049, + "step": 9440 + }, + { + "epoch": 2.02, + "learning_rate": 4.8920065161217845e-05, + "loss": 0.7409, + "step": 9450 + }, + { + "epoch": 2.02, + "learning_rate": 4.891304347826087e-05, + "loss": 0.8497, + "step": 9460 + }, + { + "epoch": 2.02, + "learning_rate": 4.89060217953039e-05, + "loss": 1.2697, + "step": 9470 + }, + { + "epoch": 2.02, + "learning_rate": 4.8899000112346934e-05, + "loss": 0.3274, + "step": 9480 + }, + { + "epoch": 2.02, + "learning_rate": 4.8891978429389955e-05, + "loss": 0.6796, + "step": 9490 + }, + { + "epoch": 2.02, + "learning_rate": 4.888495674643299e-05, + "loss": 0.5291, + "step": 9500 + }, + { + "epoch": 2.02, + "learning_rate": 4.8877935063476016e-05, + "loss": 1.1741, + "step": 9510 + }, + { + "epoch": 2.02, + "learning_rate": 4.8870913380519044e-05, + "loss": 0.5173, + "step": 9520 + }, + { + "epoch": 2.02, + "learning_rate": 4.886389169756208e-05, + "loss": 0.5347, + "step": 9530 + }, + { + "epoch": 2.02, + "learning_rate": 4.8856870014605106e-05, + "loss": 1.1272, + "step": 9540 + }, + { + "epoch": 2.02, + "learning_rate": 4.8849848331648126e-05, + "loss": 1.1414, + "step": 9550 + }, + { + "epoch": 2.02, + "learning_rate": 4.884282664869116e-05, + "loss": 0.5831, + "step": 9560 + }, + { + "epoch": 2.02, + "learning_rate": 4.883580496573419e-05, + "loss": 1.0383, + "step": 9570 + }, + { + "epoch": 2.02, + "learning_rate": 4.8828783282777215e-05, + "loss": 0.3539, + "step": 9580 + }, + { + "epoch": 2.02, + "learning_rate": 4.882176159982025e-05, + "loss": 0.4879, + "step": 9590 + }, + { + "epoch": 2.02, + "learning_rate": 4.881473991686328e-05, + "loss": 0.8955, + "step": 9600 + }, + { + "epoch": 2.02, + "learning_rate": 4.8807718233906305e-05, + "loss": 1.3461, + "step": 9610 + }, + { + "epoch": 2.02, + "learning_rate": 4.880069655094933e-05, + "loss": 1.1566, + "step": 9620 + }, + { + "epoch": 2.02, + "learning_rate": 4.879367486799236e-05, + "loss": 0.9787, + "step": 9630 + }, + { + "epoch": 2.02, + "learning_rate": 4.8786653185035394e-05, + "loss": 0.5939, + "step": 9640 + }, + { + "epoch": 2.02, + "learning_rate": 4.877963150207842e-05, + "loss": 0.2955, + "step": 9650 + }, + { + "epoch": 2.02, + "learning_rate": 4.877260981912145e-05, + "loss": 1.0346, + "step": 9660 + }, + { + "epoch": 2.02, + "learning_rate": 4.876558813616448e-05, + "loss": 0.5063, + "step": 9670 + }, + { + "epoch": 2.02, + "learning_rate": 4.8758566453207504e-05, + "loss": 0.3263, + "step": 9680 + }, + { + "epoch": 2.02, + "learning_rate": 4.875154477025053e-05, + "loss": 0.1908, + "step": 9690 + }, + { + "epoch": 2.02, + "learning_rate": 4.8744523087293566e-05, + "loss": 0.5915, + "step": 9700 + }, + { + "epoch": 2.02, + "learning_rate": 4.873750140433659e-05, + "loss": 1.3992, + "step": 9710 + }, + { + "epoch": 2.02, + "learning_rate": 4.873047972137962e-05, + "loss": 0.6923, + "step": 9720 + }, + { + "epoch": 2.02, + "learning_rate": 4.8723458038422655e-05, + "loss": 0.9391, + "step": 9730 + }, + { + "epoch": 2.02, + "learning_rate": 4.8716436355465676e-05, + "loss": 1.1005, + "step": 9740 + }, + { + "epoch": 2.02, + "learning_rate": 4.870941467250871e-05, + "loss": 0.9501, + "step": 9750 + }, + { + "epoch": 2.02, + "learning_rate": 4.870239298955174e-05, + "loss": 1.1727, + "step": 9760 + }, + { + "epoch": 2.02, + "learning_rate": 4.8695371306594765e-05, + "loss": 0.7199, + "step": 9770 + }, + { + "epoch": 2.02, + "learning_rate": 4.86883496236378e-05, + "loss": 0.3405, + "step": 9780 + }, + { + "epoch": 2.02, + "learning_rate": 4.868132794068083e-05, + "loss": 0.4157, + "step": 9790 + }, + { + "epoch": 2.02, + "learning_rate": 4.8674306257723854e-05, + "loss": 0.6786, + "step": 9800 + }, + { + "epoch": 2.02, + "learning_rate": 4.866728457476688e-05, + "loss": 0.9901, + "step": 9810 + }, + { + "epoch": 2.02, + "learning_rate": 4.866026289180991e-05, + "loss": 0.4955, + "step": 9820 + }, + { + "epoch": 2.02, + "learning_rate": 4.865324120885294e-05, + "loss": 0.9081, + "step": 9830 + }, + { + "epoch": 2.02, + "learning_rate": 4.864621952589597e-05, + "loss": 1.004, + "step": 9840 + }, + { + "epoch": 2.02, + "learning_rate": 4.8639197842939e-05, + "loss": 1.4742, + "step": 9850 + }, + { + "epoch": 2.02, + "learning_rate": 4.8632176159982026e-05, + "loss": 0.4744, + "step": 9860 + }, + { + "epoch": 2.02, + "learning_rate": 4.862515447702505e-05, + "loss": 0.5454, + "step": 9870 + }, + { + "epoch": 2.02, + "learning_rate": 4.861813279406808e-05, + "loss": 0.8284, + "step": 9880 + }, + { + "epoch": 2.02, + "learning_rate": 4.8611111111111115e-05, + "loss": 0.5385, + "step": 9890 + }, + { + "epoch": 2.03, + "learning_rate": 4.860408942815414e-05, + "loss": 1.3958, + "step": 9900 + }, + { + "epoch": 2.03, + "learning_rate": 4.859706774519717e-05, + "loss": 1.5061, + "step": 9910 + }, + { + "epoch": 2.03, + "learning_rate": 4.8590046062240204e-05, + "loss": 0.6482, + "step": 9920 + }, + { + "epoch": 2.03, + "learning_rate": 4.8583024379283225e-05, + "loss": 0.3625, + "step": 9930 + }, + { + "epoch": 2.03, + "learning_rate": 4.857600269632626e-05, + "loss": 0.238, + "step": 9940 + }, + { + "epoch": 2.03, + "learning_rate": 4.856898101336929e-05, + "loss": 0.8451, + "step": 9950 + }, + { + "epoch": 2.03, + "learning_rate": 4.8561959330412314e-05, + "loss": 0.6215, + "step": 9960 + }, + { + "epoch": 2.03, + "learning_rate": 4.855493764745535e-05, + "loss": 0.2356, + "step": 9970 + }, + { + "epoch": 2.03, + "learning_rate": 4.8547915964498376e-05, + "loss": 0.5704, + "step": 9980 + }, + { + "epoch": 2.03, + "learning_rate": 4.8540894281541404e-05, + "loss": 0.6429, + "step": 9990 + }, + { + "epoch": 2.03, + "learning_rate": 4.853387259858443e-05, + "loss": 0.8264, + "step": 10000 + }, + { + "epoch": 2.03, + "learning_rate": 4.852685091562746e-05, + "loss": 0.4024, + "step": 10010 + }, + { + "epoch": 2.03, + "learning_rate": 4.8519829232670486e-05, + "loss": 1.4077, + "step": 10020 + }, + { + "epoch": 2.03, + "learning_rate": 4.851280754971352e-05, + "loss": 0.7027, + "step": 10030 + }, + { + "epoch": 2.03, + "learning_rate": 4.850578586675655e-05, + "loss": 0.5796, + "step": 10040 + }, + { + "epoch": 2.03, + "learning_rate": 4.8498764183799575e-05, + "loss": 0.7401, + "step": 10050 + }, + { + "epoch": 2.03, + "learning_rate": 4.84917425008426e-05, + "loss": 0.2925, + "step": 10060 + }, + { + "epoch": 2.03, + "learning_rate": 4.848472081788563e-05, + "loss": 0.2851, + "step": 10070 + }, + { + "epoch": 2.03, + "learning_rate": 4.8477699134928665e-05, + "loss": 0.4985, + "step": 10080 + }, + { + "epoch": 2.03, + "learning_rate": 4.847067745197169e-05, + "loss": 1.6844, + "step": 10090 + }, + { + "epoch": 2.03, + "learning_rate": 4.846365576901472e-05, + "loss": 0.8282, + "step": 10100 + }, + { + "epoch": 2.03, + "learning_rate": 4.8456634086057754e-05, + "loss": 1.1244, + "step": 10110 + }, + { + "epoch": 2.03, + "learning_rate": 4.8449612403100775e-05, + "loss": 0.5211, + "step": 10120 + }, + { + "epoch": 2.03, + "learning_rate": 4.84425907201438e-05, + "loss": 1.2646, + "step": 10130 + }, + { + "epoch": 2.03, + "learning_rate": 4.8435569037186836e-05, + "loss": 0.5883, + "step": 10140 + }, + { + "epoch": 2.03, + "learning_rate": 4.8428547354229864e-05, + "loss": 0.6104, + "step": 10150 + }, + { + "epoch": 2.03, + "learning_rate": 4.842152567127289e-05, + "loss": 0.9094, + "step": 10160 + }, + { + "epoch": 2.03, + "learning_rate": 4.8414503988315926e-05, + "loss": 1.0124, + "step": 10170 + }, + { + "epoch": 2.03, + "learning_rate": 4.840748230535895e-05, + "loss": 1.2088, + "step": 10180 + }, + { + "epoch": 2.03, + "learning_rate": 4.840046062240198e-05, + "loss": 1.0647, + "step": 10190 + }, + { + "epoch": 2.03, + "learning_rate": 4.839343893944501e-05, + "loss": 0.9796, + "step": 10200 + }, + { + "epoch": 2.03, + "learning_rate": 4.8386417256488035e-05, + "loss": 0.3048, + "step": 10210 + }, + { + "epoch": 2.03, + "learning_rate": 4.837939557353107e-05, + "loss": 0.6877, + "step": 10220 + }, + { + "epoch": 2.03, + "learning_rate": 4.83723738905741e-05, + "loss": 0.6995, + "step": 10230 + }, + { + "epoch": 2.03, + "learning_rate": 4.8365352207617125e-05, + "loss": 0.424, + "step": 10240 + }, + { + "epoch": 2.03, + "learning_rate": 4.835833052466015e-05, + "loss": 1.282, + "step": 10250 + }, + { + "epoch": 2.03, + "learning_rate": 4.835130884170318e-05, + "loss": 0.6136, + "step": 10260 + }, + { + "epoch": 2.03, + "learning_rate": 4.834428715874621e-05, + "loss": 0.2591, + "step": 10270 + }, + { + "epoch": 2.03, + "learning_rate": 4.833726547578924e-05, + "loss": 0.5458, + "step": 10280 + }, + { + "epoch": 2.03, + "learning_rate": 4.833024379283227e-05, + "loss": 1.1679, + "step": 10290 + }, + { + "epoch": 2.03, + "learning_rate": 4.8323222109875296e-05, + "loss": 0.627, + "step": 10300 + }, + { + "epoch": 2.03, + "learning_rate": 4.831620042691833e-05, + "loss": 0.4581, + "step": 10310 + }, + { + "epoch": 2.03, + "learning_rate": 4.830917874396135e-05, + "loss": 0.7572, + "step": 10320 + }, + { + "epoch": 2.03, + "learning_rate": 4.8302157061004386e-05, + "loss": 0.8336, + "step": 10330 + }, + { + "epoch": 2.03, + "learning_rate": 4.829513537804741e-05, + "loss": 0.6813, + "step": 10340 + }, + { + "epoch": 2.03, + "learning_rate": 4.828811369509044e-05, + "loss": 0.8724, + "step": 10350 + }, + { + "epoch": 2.03, + "learning_rate": 4.8281092012133475e-05, + "loss": 0.537, + "step": 10360 + }, + { + "epoch": 2.03, + "learning_rate": 4.82740703291765e-05, + "loss": 0.457, + "step": 10370 + }, + { + "epoch": 2.03, + "learning_rate": 4.826704864621952e-05, + "loss": 0.9986, + "step": 10380 + }, + { + "epoch": 2.03, + "learning_rate": 4.826002696326256e-05, + "loss": 0.8907, + "step": 10390 + }, + { + "epoch": 2.03, + "learning_rate": 4.8253005280305585e-05, + "loss": 0.0468, + "step": 10400 + }, + { + "epoch": 2.03, + "learning_rate": 4.824598359734861e-05, + "loss": 0.944, + "step": 10410 + }, + { + "epoch": 2.03, + "learning_rate": 4.823896191439165e-05, + "loss": 0.8912, + "step": 10420 + }, + { + "epoch": 2.03, + "learning_rate": 4.8231940231434674e-05, + "loss": 0.5158, + "step": 10430 + }, + { + "epoch": 2.03, + "learning_rate": 4.82249185484777e-05, + "loss": 0.9149, + "step": 10440 + }, + { + "epoch": 2.03, + "learning_rate": 4.821789686552073e-05, + "loss": 1.1499, + "step": 10450 + }, + { + "epoch": 2.03, + "learning_rate": 4.821087518256376e-05, + "loss": 0.6026, + "step": 10460 + }, + { + "epoch": 2.03, + "learning_rate": 4.820385349960679e-05, + "loss": 0.4563, + "step": 10470 + }, + { + "epoch": 2.03, + "learning_rate": 4.819683181664982e-05, + "loss": 0.4433, + "step": 10480 + }, + { + "epoch": 2.03, + "learning_rate": 4.8189810133692846e-05, + "loss": 0.1351, + "step": 10490 + }, + { + "epoch": 2.03, + "learning_rate": 4.818278845073588e-05, + "loss": 0.4414, + "step": 10500 + }, + { + "epoch": 2.03, + "learning_rate": 4.81757667677789e-05, + "loss": 0.2493, + "step": 10510 + }, + { + "epoch": 2.03, + "learning_rate": 4.816874508482193e-05, + "loss": 0.5099, + "step": 10520 + }, + { + "epoch": 2.03, + "learning_rate": 4.816172340186496e-05, + "loss": 0.8358, + "step": 10530 + }, + { + "epoch": 2.03, + "learning_rate": 4.815470171890799e-05, + "loss": 0.7346, + "step": 10540 + }, + { + "epoch": 2.03, + "learning_rate": 4.814768003595102e-05, + "loss": 0.6918, + "step": 10550 + }, + { + "epoch": 2.03, + "learning_rate": 4.814065835299405e-05, + "loss": 0.6532, + "step": 10560 + }, + { + "epoch": 2.03, + "learning_rate": 4.813363667003707e-05, + "loss": 0.5273, + "step": 10570 + }, + { + "epoch": 2.03, + "learning_rate": 4.812661498708011e-05, + "loss": 0.7836, + "step": 10580 + }, + { + "epoch": 2.03, + "learning_rate": 4.8119593304123134e-05, + "loss": 0.886, + "step": 10590 + }, + { + "epoch": 2.03, + "learning_rate": 4.811257162116616e-05, + "loss": 0.972, + "step": 10600 + }, + { + "epoch": 2.03, + "learning_rate": 4.8105549938209196e-05, + "loss": 0.6477, + "step": 10610 + }, + { + "epoch": 2.03, + "learning_rate": 4.8098528255252224e-05, + "loss": 0.9054, + "step": 10620 + }, + { + "epoch": 2.03, + "learning_rate": 4.8091506572295244e-05, + "loss": 1.0275, + "step": 10630 + }, + { + "epoch": 2.03, + "learning_rate": 4.808448488933828e-05, + "loss": 0.6503, + "step": 10640 + }, + { + "epoch": 2.03, + "learning_rate": 4.8077463206381306e-05, + "loss": 1.1547, + "step": 10650 + }, + { + "epoch": 2.03, + "learning_rate": 4.8070441523424334e-05, + "loss": 0.6116, + "step": 10660 + }, + { + "epoch": 2.03, + "learning_rate": 4.806341984046737e-05, + "loss": 0.5774, + "step": 10670 + }, + { + "epoch": 2.03, + "learning_rate": 4.8056398157510395e-05, + "loss": 0.0856, + "step": 10680 + }, + { + "epoch": 2.04, + "learning_rate": 4.804937647455342e-05, + "loss": 0.3214, + "step": 10690 + }, + { + "epoch": 2.04, + "learning_rate": 4.804235479159645e-05, + "loss": 0.3087, + "step": 10700 + }, + { + "epoch": 2.04, + "learning_rate": 4.803533310863948e-05, + "loss": 1.0098, + "step": 10710 + }, + { + "epoch": 2.04, + "learning_rate": 4.802831142568251e-05, + "loss": 0.3977, + "step": 10720 + }, + { + "epoch": 2.04, + "learning_rate": 4.802128974272554e-05, + "loss": 0.3339, + "step": 10730 + }, + { + "epoch": 2.04, + "learning_rate": 4.801426805976857e-05, + "loss": 0.358, + "step": 10740 + }, + { + "epoch": 2.04, + "learning_rate": 4.80072463768116e-05, + "loss": 0.7154, + "step": 10750 + }, + { + "epoch": 2.04, + "learning_rate": 4.800022469385462e-05, + "loss": 0.3323, + "step": 10760 + }, + { + "epoch": 2.04, + "learning_rate": 4.799320301089765e-05, + "loss": 0.6743, + "step": 10770 + }, + { + "epoch": 2.04, + "learning_rate": 4.7986181327940684e-05, + "loss": 1.0641, + "step": 10780 + }, + { + "epoch": 2.04, + "learning_rate": 4.797915964498371e-05, + "loss": 0.2426, + "step": 10790 + }, + { + "epoch": 2.04, + "learning_rate": 4.797213796202674e-05, + "loss": 0.6633, + "step": 10800 + }, + { + "epoch": 2.04, + "learning_rate": 4.796511627906977e-05, + "loss": 0.6288, + "step": 10810 + }, + { + "epoch": 2.04, + "learning_rate": 4.7958094596112794e-05, + "loss": 0.3103, + "step": 10820 + }, + { + "epoch": 2.04, + "learning_rate": 4.795107291315583e-05, + "loss": 0.8735, + "step": 10830 + }, + { + "epoch": 2.04, + "learning_rate": 4.7944051230198855e-05, + "loss": 0.4551, + "step": 10840 + }, + { + "epoch": 2.04, + "learning_rate": 4.793702954724188e-05, + "loss": 0.8446, + "step": 10850 + }, + { + "epoch": 2.04, + "learning_rate": 4.793000786428492e-05, + "loss": 0.8748, + "step": 10860 + }, + { + "epoch": 2.04, + "learning_rate": 4.7922986181327945e-05, + "loss": 1.2554, + "step": 10870 + }, + { + "epoch": 2.04, + "learning_rate": 4.791596449837097e-05, + "loss": 1.1323, + "step": 10880 + }, + { + "epoch": 2.04, + "learning_rate": 4.7908942815414e-05, + "loss": 0.3389, + "step": 10890 + }, + { + "epoch": 2.04, + "learning_rate": 4.790192113245703e-05, + "loss": 0.2464, + "step": 10900 + }, + { + "epoch": 2.04, + "learning_rate": 4.7894899449500055e-05, + "loss": 0.909, + "step": 10910 + }, + { + "epoch": 2.04, + "learning_rate": 4.788787776654309e-05, + "loss": 0.8414, + "step": 10920 + }, + { + "epoch": 2.04, + "learning_rate": 4.7880856083586116e-05, + "loss": 0.1223, + "step": 10930 + }, + { + "epoch": 2.04, + "learning_rate": 4.7873834400629144e-05, + "loss": 0.3594, + "step": 10940 + }, + { + "epoch": 2.04, + "learning_rate": 4.786681271767217e-05, + "loss": 0.4888, + "step": 10950 + }, + { + "epoch": 2.04, + "learning_rate": 4.78597910347152e-05, + "loss": 0.5202, + "step": 10960 + }, + { + "epoch": 2.04, + "learning_rate": 4.785276935175823e-05, + "loss": 0.9503, + "step": 10970 + }, + { + "epoch": 2.04, + "learning_rate": 4.784574766880126e-05, + "loss": 0.2493, + "step": 10980 + }, + { + "epoch": 2.04, + "learning_rate": 4.783872598584429e-05, + "loss": 1.4139, + "step": 10990 + }, + { + "epoch": 2.04, + "learning_rate": 4.783170430288732e-05, + "loss": 0.2602, + "step": 11000 + }, + { + "epoch": 2.04, + "learning_rate": 4.782468261993035e-05, + "loss": 1.1004, + "step": 11010 + }, + { + "epoch": 2.04, + "learning_rate": 4.781766093697338e-05, + "loss": 0.2803, + "step": 11020 + }, + { + "epoch": 2.04, + "learning_rate": 4.7810639254016405e-05, + "loss": 0.5344, + "step": 11030 + }, + { + "epoch": 2.04, + "learning_rate": 4.780361757105943e-05, + "loss": 1.1285, + "step": 11040 + }, + { + "epoch": 2.04, + "learning_rate": 4.779659588810247e-05, + "loss": 0.4942, + "step": 11050 + }, + { + "epoch": 2.04, + "learning_rate": 4.7789574205145494e-05, + "loss": 0.7212, + "step": 11060 + }, + { + "epoch": 2.04, + "learning_rate": 4.778255252218852e-05, + "loss": 0.6673, + "step": 11070 + }, + { + "epoch": 2.04, + "learning_rate": 4.777553083923155e-05, + "loss": 0.697, + "step": 11080 + }, + { + "epoch": 2.04, + "learning_rate": 4.776850915627458e-05, + "loss": 1.8318, + "step": 11090 + }, + { + "epoch": 2.04, + "learning_rate": 4.7761487473317604e-05, + "loss": 0.7076, + "step": 11100 + }, + { + "epoch": 2.04, + "learning_rate": 4.775446579036064e-05, + "loss": 0.8304, + "step": 11110 + }, + { + "epoch": 2.04, + "learning_rate": 4.7747444107403666e-05, + "loss": 0.181, + "step": 11120 + }, + { + "epoch": 2.04, + "learning_rate": 4.774042242444669e-05, + "loss": 0.2028, + "step": 11130 + }, + { + "epoch": 2.04, + "learning_rate": 4.773340074148972e-05, + "loss": 1.4598, + "step": 11140 + }, + { + "epoch": 2.04, + "learning_rate": 4.772637905853275e-05, + "loss": 0.8776, + "step": 11150 + }, + { + "epoch": 2.04, + "learning_rate": 4.771935737557578e-05, + "loss": 1.0532, + "step": 11160 + }, + { + "epoch": 2.04, + "learning_rate": 4.771233569261881e-05, + "loss": 0.6016, + "step": 11170 + }, + { + "epoch": 2.04, + "learning_rate": 4.770531400966184e-05, + "loss": 0.3232, + "step": 11180 + }, + { + "epoch": 2.04, + "learning_rate": 4.769829232670487e-05, + "loss": 0.2391, + "step": 11190 + }, + { + "epoch": 2.04, + "learning_rate": 4.76912706437479e-05, + "loss": 0.7103, + "step": 11200 + }, + { + "epoch": 2.04, + "learning_rate": 4.768424896079092e-05, + "loss": 0.3399, + "step": 11210 + }, + { + "epoch": 2.04, + "learning_rate": 4.7677227277833954e-05, + "loss": 0.7041, + "step": 11220 + }, + { + "epoch": 2.04, + "learning_rate": 4.767020559487698e-05, + "loss": 0.783, + "step": 11230 + }, + { + "epoch": 2.04, + "learning_rate": 4.766318391192001e-05, + "loss": 0.9811, + "step": 11240 + }, + { + "epoch": 2.04, + "learning_rate": 4.7656162228963044e-05, + "loss": 1.1581, + "step": 11250 + }, + { + "epoch": 2.04, + "learning_rate": 4.764914054600607e-05, + "loss": 1.0759, + "step": 11260 + }, + { + "epoch": 2.04, + "learning_rate": 4.76421188630491e-05, + "loss": 1.0752, + "step": 11270 + }, + { + "epoch": 2.04, + "learning_rate": 4.7635097180092126e-05, + "loss": 0.3745, + "step": 11280 + }, + { + "epoch": 2.04, + "learning_rate": 4.7628075497135154e-05, + "loss": 0.128, + "step": 11290 + }, + { + "epoch": 2.04, + "learning_rate": 4.762105381417819e-05, + "loss": 0.7648, + "step": 11300 + }, + { + "epoch": 2.04, + "learning_rate": 4.7614032131221215e-05, + "loss": 0.4624, + "step": 11310 + }, + { + "epoch": 2.04, + "learning_rate": 4.760701044826424e-05, + "loss": 0.5345, + "step": 11320 + }, + { + "epoch": 2.04, + "learning_rate": 4.759998876530727e-05, + "loss": 0.6443, + "step": 11330 + }, + { + "epoch": 2.04, + "learning_rate": 4.75929670823503e-05, + "loss": 0.5734, + "step": 11340 + }, + { + "epoch": 2.04, + "learning_rate": 4.7585945399393325e-05, + "loss": 0.8904, + "step": 11350 + }, + { + "epoch": 2.04, + "learning_rate": 4.757892371643636e-05, + "loss": 0.5029, + "step": 11360 + }, + { + "epoch": 2.04, + "learning_rate": 4.757190203347939e-05, + "loss": 0.5533, + "step": 11370 + }, + { + "epoch": 2.04, + "learning_rate": 4.7564880350522415e-05, + "loss": 0.6399, + "step": 11380 + }, + { + "epoch": 2.04, + "learning_rate": 4.755785866756545e-05, + "loss": 0.7892, + "step": 11390 + }, + { + "epoch": 2.04, + "learning_rate": 4.755083698460847e-05, + "loss": 0.3524, + "step": 11400 + }, + { + "epoch": 2.04, + "learning_rate": 4.7543815301651504e-05, + "loss": 0.6781, + "step": 11410 + }, + { + "epoch": 2.04, + "learning_rate": 4.753679361869453e-05, + "loss": 0.6549, + "step": 11420 + }, + { + "epoch": 2.04, + "learning_rate": 4.752977193573756e-05, + "loss": 0.6583, + "step": 11430 + }, + { + "epoch": 2.04, + "learning_rate": 4.752275025278059e-05, + "loss": 0.7213, + "step": 11440 + }, + { + "epoch": 2.04, + "learning_rate": 4.751572856982362e-05, + "loss": 0.2021, + "step": 11450 + }, + { + "epoch": 2.04, + "learning_rate": 4.750870688686664e-05, + "loss": 0.3574, + "step": 11460 + }, + { + "epoch": 2.04, + "learning_rate": 4.7501685203909675e-05, + "loss": 0.8602, + "step": 11470 + }, + { + "epoch": 2.05, + "learning_rate": 4.74946635209527e-05, + "loss": 0.4062, + "step": 11480 + }, + { + "epoch": 2.05, + "learning_rate": 4.748764183799573e-05, + "loss": 0.8534, + "step": 11490 + }, + { + "epoch": 2.05, + "learning_rate": 4.7480620155038765e-05, + "loss": 0.3926, + "step": 11500 + }, + { + "epoch": 2.05, + "learning_rate": 4.747359847208179e-05, + "loss": 0.4779, + "step": 11510 + }, + { + "epoch": 2.05, + "learning_rate": 4.746657678912482e-05, + "loss": 0.6154, + "step": 11520 + }, + { + "epoch": 2.05, + "learning_rate": 4.745955510616785e-05, + "loss": 0.3432, + "step": 11530 + }, + { + "epoch": 2.05, + "learning_rate": 4.7452533423210875e-05, + "loss": 0.3499, + "step": 11540 + }, + { + "epoch": 2.05, + "learning_rate": 4.744551174025391e-05, + "loss": 1.0412, + "step": 11550 + }, + { + "epoch": 2.05, + "learning_rate": 4.7438490057296936e-05, + "loss": 0.5076, + "step": 11560 + }, + { + "epoch": 2.05, + "learning_rate": 4.7431468374339964e-05, + "loss": 0.6185, + "step": 11570 + }, + { + "epoch": 2.05, + "learning_rate": 4.7424446691383e-05, + "loss": 0.3891, + "step": 11580 + }, + { + "epoch": 2.05, + "learning_rate": 4.741742500842602e-05, + "loss": 0.4084, + "step": 11590 + }, + { + "epoch": 2.05, + "learning_rate": 4.7410403325469046e-05, + "loss": 0.507, + "step": 11600 + }, + { + "epoch": 2.05, + "learning_rate": 4.740338164251208e-05, + "loss": 0.6221, + "step": 11610 + }, + { + "epoch": 2.05, + "learning_rate": 4.739635995955511e-05, + "loss": 0.9327, + "step": 11620 + }, + { + "epoch": 2.05, + "learning_rate": 4.7389338276598136e-05, + "loss": 0.4568, + "step": 11630 + }, + { + "epoch": 2.05, + "learning_rate": 4.738231659364117e-05, + "loss": 0.4686, + "step": 11640 + }, + { + "epoch": 2.05, + "learning_rate": 4.737529491068419e-05, + "loss": 0.169, + "step": 11650 + }, + { + "epoch": 2.05, + "learning_rate": 4.7368273227727225e-05, + "loss": 0.6117, + "step": 11660 + }, + { + "epoch": 2.05, + "learning_rate": 4.736125154477025e-05, + "loss": 0.6964, + "step": 11670 + }, + { + "epoch": 2.05, + "learning_rate": 4.735422986181328e-05, + "loss": 0.5856, + "step": 11680 + }, + { + "epoch": 2.05, + "learning_rate": 4.7347208178856314e-05, + "loss": 0.5145, + "step": 11690 + }, + { + "epoch": 2.05, + "learning_rate": 4.734018649589934e-05, + "loss": 0.6437, + "step": 11700 + }, + { + "epoch": 2.05, + "learning_rate": 4.733316481294237e-05, + "loss": 1.0695, + "step": 11710 + }, + { + "epoch": 2.05, + "learning_rate": 4.73261431299854e-05, + "loss": 1.695, + "step": 11720 + }, + { + "epoch": 2.05, + "learning_rate": 4.7319121447028424e-05, + "loss": 0.7638, + "step": 11730 + }, + { + "epoch": 2.05, + "learning_rate": 4.731209976407145e-05, + "loss": 0.4896, + "step": 11740 + }, + { + "epoch": 2.05, + "learning_rate": 4.7305078081114486e-05, + "loss": 0.4114, + "step": 11750 + }, + { + "epoch": 2.05, + "learning_rate": 4.729805639815751e-05, + "loss": 0.2666, + "step": 11760 + }, + { + "epoch": 2.05, + "learning_rate": 4.729103471520054e-05, + "loss": 0.2912, + "step": 11770 + }, + { + "epoch": 2.05, + "learning_rate": 4.728401303224357e-05, + "loss": 1.3753, + "step": 11780 + }, + { + "epoch": 2.05, + "learning_rate": 4.7276991349286596e-05, + "loss": 0.0397, + "step": 11790 + }, + { + "epoch": 2.05, + "learning_rate": 4.726996966632963e-05, + "loss": 0.6896, + "step": 11800 + }, + { + "epoch": 2.05, + "learning_rate": 4.726294798337266e-05, + "loss": 0.4477, + "step": 11810 + }, + { + "epoch": 2.05, + "learning_rate": 4.7255926300415685e-05, + "loss": 0.743, + "step": 11820 + }, + { + "epoch": 2.05, + "learning_rate": 4.724890461745872e-05, + "loss": 0.8092, + "step": 11830 + }, + { + "epoch": 2.05, + "learning_rate": 4.724188293450174e-05, + "loss": 0.3391, + "step": 11840 + }, + { + "epoch": 2.05, + "learning_rate": 4.723486125154477e-05, + "loss": 0.8784, + "step": 11850 + }, + { + "epoch": 2.05, + "learning_rate": 4.72278395685878e-05, + "loss": 0.7501, + "step": 11860 + }, + { + "epoch": 2.05, + "eval_accuracy": 0.6476583666623661, + "eval_loss": 1.5601611137390137, + "eval_runtime": 1227.0832, + "eval_samples_per_second": 6.317, + "eval_steps_per_second": 3.159, + "step": 11868 + }, + { + "epoch": 3.0, + "learning_rate": 4.722081788563083e-05, + "loss": 0.3143, + "step": 11870 + }, + { + "epoch": 3.0, + "learning_rate": 4.721379620267386e-05, + "loss": 1.0078, + "step": 11880 + }, + { + "epoch": 3.0, + "learning_rate": 4.720677451971689e-05, + "loss": 0.8437, + "step": 11890 + }, + { + "epoch": 3.0, + "learning_rate": 4.719975283675992e-05, + "loss": 0.5359, + "step": 11900 + }, + { + "epoch": 3.0, + "learning_rate": 4.7192731153802946e-05, + "loss": 0.4032, + "step": 11910 + }, + { + "epoch": 3.0, + "learning_rate": 4.7185709470845974e-05, + "loss": 0.3134, + "step": 11920 + }, + { + "epoch": 3.0, + "learning_rate": 4.7178687787889e-05, + "loss": 0.2581, + "step": 11930 + }, + { + "epoch": 3.0, + "learning_rate": 4.7171666104932035e-05, + "loss": 0.1437, + "step": 11940 + }, + { + "epoch": 3.0, + "learning_rate": 4.716464442197506e-05, + "loss": 0.2373, + "step": 11950 + }, + { + "epoch": 3.0, + "learning_rate": 4.715762273901809e-05, + "loss": 0.0609, + "step": 11960 + }, + { + "epoch": 3.0, + "learning_rate": 4.715060105606112e-05, + "loss": 0.3543, + "step": 11970 + }, + { + "epoch": 3.0, + "learning_rate": 4.7143579373104145e-05, + "loss": 0.5592, + "step": 11980 + }, + { + "epoch": 3.0, + "learning_rate": 4.713655769014718e-05, + "loss": 0.7185, + "step": 11990 + }, + { + "epoch": 3.0, + "learning_rate": 4.712953600719021e-05, + "loss": 0.4144, + "step": 12000 + }, + { + "epoch": 3.0, + "learning_rate": 4.7122514324233235e-05, + "loss": 0.3487, + "step": 12010 + }, + { + "epoch": 3.0, + "learning_rate": 4.711549264127627e-05, + "loss": 0.5302, + "step": 12020 + }, + { + "epoch": 3.0, + "learning_rate": 4.710847095831929e-05, + "loss": 0.8708, + "step": 12030 + }, + { + "epoch": 3.0, + "learning_rate": 4.710144927536232e-05, + "loss": 0.2747, + "step": 12040 + }, + { + "epoch": 3.0, + "learning_rate": 4.709442759240535e-05, + "loss": 0.7926, + "step": 12050 + }, + { + "epoch": 3.0, + "learning_rate": 4.708740590944838e-05, + "loss": 0.6743, + "step": 12060 + }, + { + "epoch": 3.0, + "learning_rate": 4.7080384226491406e-05, + "loss": 0.2431, + "step": 12070 + }, + { + "epoch": 3.0, + "learning_rate": 4.707336254353444e-05, + "loss": 0.3252, + "step": 12080 + }, + { + "epoch": 3.0, + "learning_rate": 4.706634086057747e-05, + "loss": 0.1324, + "step": 12090 + }, + { + "epoch": 3.0, + "learning_rate": 4.7059319177620495e-05, + "loss": 0.238, + "step": 12100 + }, + { + "epoch": 3.0, + "learning_rate": 4.705229749466352e-05, + "loss": 0.1514, + "step": 12110 + }, + { + "epoch": 3.0, + "learning_rate": 4.704527581170655e-05, + "loss": 1.0484, + "step": 12120 + }, + { + "epoch": 3.0, + "learning_rate": 4.7038254128749585e-05, + "loss": 0.4404, + "step": 12130 + }, + { + "epoch": 3.0, + "learning_rate": 4.703123244579261e-05, + "loss": 0.4533, + "step": 12140 + }, + { + "epoch": 3.0, + "learning_rate": 4.702421076283564e-05, + "loss": 0.9046, + "step": 12150 + }, + { + "epoch": 3.0, + "learning_rate": 4.701718907987867e-05, + "loss": 0.3901, + "step": 12160 + }, + { + "epoch": 3.0, + "learning_rate": 4.7010167396921695e-05, + "loss": 0.2893, + "step": 12170 + }, + { + "epoch": 3.0, + "learning_rate": 4.700314571396472e-05, + "loss": 0.125, + "step": 12180 + }, + { + "epoch": 3.0, + "learning_rate": 4.6996124031007756e-05, + "loss": 0.0608, + "step": 12190 + }, + { + "epoch": 3.0, + "learning_rate": 4.6989102348050784e-05, + "loss": 0.4398, + "step": 12200 + }, + { + "epoch": 3.0, + "learning_rate": 4.698208066509381e-05, + "loss": 0.2148, + "step": 12210 + }, + { + "epoch": 3.0, + "learning_rate": 4.697505898213684e-05, + "loss": 0.6907, + "step": 12220 + }, + { + "epoch": 3.0, + "learning_rate": 4.6968037299179866e-05, + "loss": 0.2142, + "step": 12230 + }, + { + "epoch": 3.0, + "learning_rate": 4.69610156162229e-05, + "loss": 0.2534, + "step": 12240 + }, + { + "epoch": 3.0, + "learning_rate": 4.695399393326593e-05, + "loss": 0.9024, + "step": 12250 + }, + { + "epoch": 3.0, + "learning_rate": 4.6946972250308956e-05, + "loss": 0.082, + "step": 12260 + }, + { + "epoch": 3.01, + "learning_rate": 4.693995056735199e-05, + "loss": 0.1234, + "step": 12270 + }, + { + "epoch": 3.01, + "learning_rate": 4.693292888439502e-05, + "loss": 0.5319, + "step": 12280 + }, + { + "epoch": 3.01, + "learning_rate": 4.692590720143804e-05, + "loss": 0.7733, + "step": 12290 + }, + { + "epoch": 3.01, + "learning_rate": 4.691888551848107e-05, + "loss": 0.4241, + "step": 12300 + }, + { + "epoch": 3.01, + "learning_rate": 4.69118638355241e-05, + "loss": 1.1588, + "step": 12310 + }, + { + "epoch": 3.01, + "learning_rate": 4.690484215256713e-05, + "loss": 0.3644, + "step": 12320 + }, + { + "epoch": 3.01, + "learning_rate": 4.689782046961016e-05, + "loss": 0.3535, + "step": 12330 + }, + { + "epoch": 3.01, + "learning_rate": 4.689079878665319e-05, + "loss": 0.369, + "step": 12340 + }, + { + "epoch": 3.01, + "learning_rate": 4.688377710369622e-05, + "loss": 0.6029, + "step": 12350 + }, + { + "epoch": 3.01, + "learning_rate": 4.6876755420739244e-05, + "loss": 0.3571, + "step": 12360 + }, + { + "epoch": 3.01, + "learning_rate": 4.686973373778227e-05, + "loss": 0.7849, + "step": 12370 + }, + { + "epoch": 3.01, + "learning_rate": 4.6862712054825306e-05, + "loss": 0.2195, + "step": 12380 + }, + { + "epoch": 3.01, + "learning_rate": 4.685569037186833e-05, + "loss": 0.2342, + "step": 12390 + }, + { + "epoch": 3.01, + "learning_rate": 4.684866868891136e-05, + "loss": 0.5316, + "step": 12400 + }, + { + "epoch": 3.01, + "learning_rate": 4.6841647005954395e-05, + "loss": 0.3965, + "step": 12410 + }, + { + "epoch": 3.01, + "learning_rate": 4.6834625322997416e-05, + "loss": 0.0272, + "step": 12420 + }, + { + "epoch": 3.01, + "learning_rate": 4.682760364004044e-05, + "loss": 0.6089, + "step": 12430 + }, + { + "epoch": 3.01, + "learning_rate": 4.682058195708348e-05, + "loss": 0.265, + "step": 12440 + }, + { + "epoch": 3.01, + "learning_rate": 4.6813560274126505e-05, + "loss": 0.0436, + "step": 12450 + }, + { + "epoch": 3.01, + "learning_rate": 4.680653859116953e-05, + "loss": 0.1923, + "step": 12460 + }, + { + "epoch": 3.01, + "learning_rate": 4.679951690821257e-05, + "loss": 0.679, + "step": 12470 + }, + { + "epoch": 3.01, + "learning_rate": 4.679249522525559e-05, + "loss": 0.4698, + "step": 12480 + }, + { + "epoch": 3.01, + "learning_rate": 4.678547354229862e-05, + "loss": 2.1826, + "step": 12490 + }, + { + "epoch": 3.01, + "learning_rate": 4.677845185934165e-05, + "loss": 0.4964, + "step": 12500 + }, + { + "epoch": 3.01, + "learning_rate": 4.677143017638468e-05, + "loss": 0.6847, + "step": 12510 + }, + { + "epoch": 3.01, + "learning_rate": 4.676440849342771e-05, + "loss": 1.0087, + "step": 12520 + }, + { + "epoch": 3.01, + "learning_rate": 4.675738681047074e-05, + "loss": 0.3967, + "step": 12530 + }, + { + "epoch": 3.01, + "learning_rate": 4.675036512751376e-05, + "loss": 0.3194, + "step": 12540 + }, + { + "epoch": 3.01, + "learning_rate": 4.6743343444556794e-05, + "loss": 0.0721, + "step": 12550 + }, + { + "epoch": 3.01, + "learning_rate": 4.673632176159982e-05, + "loss": 0.7591, + "step": 12560 + }, + { + "epoch": 3.01, + "learning_rate": 4.672930007864285e-05, + "loss": 0.8378, + "step": 12570 + }, + { + "epoch": 3.01, + "learning_rate": 4.672227839568588e-05, + "loss": 0.1648, + "step": 12580 + }, + { + "epoch": 3.01, + "learning_rate": 4.671525671272891e-05, + "loss": 0.0791, + "step": 12590 + }, + { + "epoch": 3.01, + "learning_rate": 4.670823502977194e-05, + "loss": 0.2022, + "step": 12600 + }, + { + "epoch": 3.01, + "learning_rate": 4.6701213346814965e-05, + "loss": 0.4152, + "step": 12610 + }, + { + "epoch": 3.01, + "learning_rate": 4.669419166385799e-05, + "loss": 0.8237, + "step": 12620 + }, + { + "epoch": 3.01, + "learning_rate": 4.668716998090103e-05, + "loss": 0.2269, + "step": 12630 + }, + { + "epoch": 3.01, + "learning_rate": 4.6680148297944054e-05, + "loss": 0.4906, + "step": 12640 + }, + { + "epoch": 3.01, + "learning_rate": 4.667312661498708e-05, + "loss": 0.0533, + "step": 12650 + }, + { + "epoch": 3.01, + "learning_rate": 4.6666104932030116e-05, + "loss": 0.6052, + "step": 12660 + }, + { + "epoch": 3.01, + "learning_rate": 4.665908324907314e-05, + "loss": 0.0801, + "step": 12670 + }, + { + "epoch": 3.01, + "learning_rate": 4.6652061566116164e-05, + "loss": 0.3692, + "step": 12680 + }, + { + "epoch": 3.01, + "learning_rate": 4.66450398831592e-05, + "loss": 0.585, + "step": 12690 + }, + { + "epoch": 3.01, + "learning_rate": 4.6638018200202226e-05, + "loss": 0.4378, + "step": 12700 + }, + { + "epoch": 3.01, + "learning_rate": 4.6630996517245254e-05, + "loss": 0.3132, + "step": 12710 + }, + { + "epoch": 3.01, + "learning_rate": 4.662397483428829e-05, + "loss": 0.4663, + "step": 12720 + }, + { + "epoch": 3.01, + "learning_rate": 4.661695315133131e-05, + "loss": 0.5782, + "step": 12730 + }, + { + "epoch": 3.01, + "learning_rate": 4.660993146837434e-05, + "loss": 0.4679, + "step": 12740 + }, + { + "epoch": 3.01, + "learning_rate": 4.660290978541737e-05, + "loss": 0.3372, + "step": 12750 + }, + { + "epoch": 3.01, + "learning_rate": 4.65958881024604e-05, + "loss": 0.5742, + "step": 12760 + }, + { + "epoch": 3.01, + "learning_rate": 4.658886641950343e-05, + "loss": 0.123, + "step": 12770 + }, + { + "epoch": 3.01, + "learning_rate": 4.658184473654646e-05, + "loss": 0.6549, + "step": 12780 + }, + { + "epoch": 3.01, + "learning_rate": 4.657482305358949e-05, + "loss": 0.28, + "step": 12790 + }, + { + "epoch": 3.01, + "learning_rate": 4.6567801370632515e-05, + "loss": 0.0228, + "step": 12800 + }, + { + "epoch": 3.01, + "learning_rate": 4.656077968767554e-05, + "loss": 0.6003, + "step": 12810 + }, + { + "epoch": 3.01, + "learning_rate": 4.655375800471857e-05, + "loss": 0.5381, + "step": 12820 + }, + { + "epoch": 3.01, + "learning_rate": 4.6546736321761604e-05, + "loss": 0.9116, + "step": 12830 + }, + { + "epoch": 3.01, + "learning_rate": 4.653971463880463e-05, + "loss": 0.4522, + "step": 12840 + }, + { + "epoch": 3.01, + "learning_rate": 4.653269295584766e-05, + "loss": 0.3821, + "step": 12850 + }, + { + "epoch": 3.01, + "learning_rate": 4.6525671272890686e-05, + "loss": 0.4233, + "step": 12860 + }, + { + "epoch": 3.01, + "learning_rate": 4.6518649589933714e-05, + "loss": 0.6587, + "step": 12870 + }, + { + "epoch": 3.01, + "learning_rate": 4.651162790697675e-05, + "loss": 0.7896, + "step": 12880 + }, + { + "epoch": 3.01, + "learning_rate": 4.6504606224019776e-05, + "loss": 0.239, + "step": 12890 + }, + { + "epoch": 3.01, + "learning_rate": 4.64975845410628e-05, + "loss": 0.3021, + "step": 12900 + }, + { + "epoch": 3.01, + "learning_rate": 4.649056285810584e-05, + "loss": 0.344, + "step": 12910 + }, + { + "epoch": 3.01, + "learning_rate": 4.648354117514886e-05, + "loss": 0.074, + "step": 12920 + }, + { + "epoch": 3.01, + "learning_rate": 4.6476519492191886e-05, + "loss": 0.5993, + "step": 12930 + }, + { + "epoch": 3.01, + "learning_rate": 4.646949780923492e-05, + "loss": 0.5399, + "step": 12940 + }, + { + "epoch": 3.01, + "learning_rate": 4.646247612627795e-05, + "loss": 0.6622, + "step": 12950 + }, + { + "epoch": 3.01, + "learning_rate": 4.6455454443320975e-05, + "loss": 0.5662, + "step": 12960 + }, + { + "epoch": 3.01, + "learning_rate": 4.644843276036401e-05, + "loss": 0.2485, + "step": 12970 + }, + { + "epoch": 3.01, + "learning_rate": 4.6441411077407037e-05, + "loss": 0.861, + "step": 12980 + }, + { + "epoch": 3.01, + "learning_rate": 4.6434389394450064e-05, + "loss": 0.0614, + "step": 12990 + }, + { + "epoch": 3.01, + "learning_rate": 4.642736771149309e-05, + "loss": 0.0925, + "step": 13000 + }, + { + "epoch": 3.01, + "learning_rate": 4.642034602853612e-05, + "loss": 0.0323, + "step": 13010 + }, + { + "epoch": 3.01, + "learning_rate": 4.641332434557915e-05, + "loss": 0.5408, + "step": 13020 + }, + { + "epoch": 3.01, + "learning_rate": 4.640630266262218e-05, + "loss": 0.0098, + "step": 13030 + }, + { + "epoch": 3.01, + "learning_rate": 4.639928097966521e-05, + "loss": 0.6243, + "step": 13040 + }, + { + "epoch": 3.01, + "learning_rate": 4.6392259296708236e-05, + "loss": 0.3056, + "step": 13050 + }, + { + "epoch": 3.02, + "learning_rate": 4.638523761375126e-05, + "loss": 0.6436, + "step": 13060 + }, + { + "epoch": 3.02, + "learning_rate": 4.63782159307943e-05, + "loss": 0.2953, + "step": 13070 + }, + { + "epoch": 3.02, + "learning_rate": 4.6371194247837325e-05, + "loss": 0.7828, + "step": 13080 + }, + { + "epoch": 3.02, + "learning_rate": 4.636417256488035e-05, + "loss": 0.1911, + "step": 13090 + }, + { + "epoch": 3.02, + "learning_rate": 4.635715088192339e-05, + "loss": 0.673, + "step": 13100 + }, + { + "epoch": 3.02, + "learning_rate": 4.6350129198966414e-05, + "loss": 0.2671, + "step": 13110 + }, + { + "epoch": 3.02, + "learning_rate": 4.6343107516009435e-05, + "loss": 0.6332, + "step": 13120 + }, + { + "epoch": 3.02, + "learning_rate": 4.633608583305247e-05, + "loss": 0.1835, + "step": 13130 + }, + { + "epoch": 3.02, + "learning_rate": 4.63290641500955e-05, + "loss": 0.6446, + "step": 13140 + }, + { + "epoch": 3.02, + "learning_rate": 4.6322042467138524e-05, + "loss": 0.6259, + "step": 13150 + }, + { + "epoch": 3.02, + "learning_rate": 4.631502078418156e-05, + "loss": 0.4761, + "step": 13160 + }, + { + "epoch": 3.02, + "learning_rate": 4.6307999101224586e-05, + "loss": 0.3725, + "step": 13170 + }, + { + "epoch": 3.02, + "learning_rate": 4.6300977418267614e-05, + "loss": 0.0589, + "step": 13180 + }, + { + "epoch": 3.02, + "learning_rate": 4.629395573531064e-05, + "loss": 0.4839, + "step": 13190 + }, + { + "epoch": 3.02, + "learning_rate": 4.628693405235367e-05, + "loss": 0.1157, + "step": 13200 + }, + { + "epoch": 3.02, + "learning_rate": 4.62799123693967e-05, + "loss": 0.4466, + "step": 13210 + }, + { + "epoch": 3.02, + "learning_rate": 4.627289068643973e-05, + "loss": 0.5788, + "step": 13220 + }, + { + "epoch": 3.02, + "learning_rate": 4.626586900348276e-05, + "loss": 0.2678, + "step": 13230 + }, + { + "epoch": 3.02, + "learning_rate": 4.6258847320525785e-05, + "loss": 0.0354, + "step": 13240 + }, + { + "epoch": 3.02, + "learning_rate": 4.625182563756881e-05, + "loss": 0.2195, + "step": 13250 + }, + { + "epoch": 3.02, + "learning_rate": 4.624480395461184e-05, + "loss": 1.0414, + "step": 13260 + }, + { + "epoch": 3.02, + "learning_rate": 4.6237782271654874e-05, + "loss": 0.9989, + "step": 13270 + }, + { + "epoch": 3.02, + "learning_rate": 4.62307605886979e-05, + "loss": 1.0054, + "step": 13280 + }, + { + "epoch": 3.02, + "learning_rate": 4.622373890574093e-05, + "loss": 0.7873, + "step": 13290 + }, + { + "epoch": 3.02, + "learning_rate": 4.6216717222783964e-05, + "loss": 0.6889, + "step": 13300 + }, + { + "epoch": 3.02, + "learning_rate": 4.6209695539826984e-05, + "loss": 0.1809, + "step": 13310 + }, + { + "epoch": 3.02, + "learning_rate": 4.620267385687002e-05, + "loss": 0.367, + "step": 13320 + }, + { + "epoch": 3.02, + "learning_rate": 4.6195652173913046e-05, + "loss": 0.9204, + "step": 13330 + }, + { + "epoch": 3.02, + "learning_rate": 4.6188630490956074e-05, + "loss": 0.5189, + "step": 13340 + }, + { + "epoch": 3.02, + "learning_rate": 4.618160880799911e-05, + "loss": 0.4226, + "step": 13350 + }, + { + "epoch": 3.02, + "learning_rate": 4.6174587125042135e-05, + "loss": 0.1222, + "step": 13360 + }, + { + "epoch": 3.02, + "learning_rate": 4.6167565442085156e-05, + "loss": 0.423, + "step": 13370 + }, + { + "epoch": 3.02, + "learning_rate": 4.616054375912819e-05, + "loss": 0.0163, + "step": 13380 + }, + { + "epoch": 3.02, + "learning_rate": 4.615352207617122e-05, + "loss": 0.1339, + "step": 13390 + }, + { + "epoch": 3.02, + "learning_rate": 4.6146500393214245e-05, + "loss": 0.5178, + "step": 13400 + }, + { + "epoch": 3.02, + "learning_rate": 4.613947871025728e-05, + "loss": 0.42, + "step": 13410 + }, + { + "epoch": 3.02, + "learning_rate": 4.613245702730031e-05, + "loss": 0.0923, + "step": 13420 + }, + { + "epoch": 3.02, + "learning_rate": 4.6125435344343335e-05, + "loss": 0.832, + "step": 13430 + }, + { + "epoch": 3.02, + "learning_rate": 4.611841366138636e-05, + "loss": 0.3838, + "step": 13440 + }, + { + "epoch": 3.02, + "learning_rate": 4.611139197842939e-05, + "loss": 0.0617, + "step": 13450 + }, + { + "epoch": 3.02, + "learning_rate": 4.6104370295472424e-05, + "loss": 0.5253, + "step": 13460 + }, + { + "epoch": 3.02, + "learning_rate": 4.609734861251545e-05, + "loss": 0.1055, + "step": 13470 + }, + { + "epoch": 3.02, + "learning_rate": 4.609032692955848e-05, + "loss": 0.6798, + "step": 13480 + }, + { + "epoch": 3.02, + "learning_rate": 4.608330524660151e-05, + "loss": 0.1073, + "step": 13490 + }, + { + "epoch": 3.02, + "learning_rate": 4.6076283563644534e-05, + "loss": 0.3697, + "step": 13500 + }, + { + "epoch": 3.02, + "learning_rate": 4.606926188068756e-05, + "loss": 0.5177, + "step": 13510 + }, + { + "epoch": 3.02, + "learning_rate": 4.6062240197730596e-05, + "loss": 1.3425, + "step": 13520 + }, + { + "epoch": 3.02, + "learning_rate": 4.605521851477362e-05, + "loss": 0.433, + "step": 13530 + }, + { + "epoch": 3.02, + "learning_rate": 4.604819683181665e-05, + "loss": 0.0726, + "step": 13540 + }, + { + "epoch": 3.02, + "learning_rate": 4.6041175148859685e-05, + "loss": 0.5098, + "step": 13550 + }, + { + "epoch": 3.02, + "learning_rate": 4.6034153465902706e-05, + "loss": 0.5076, + "step": 13560 + }, + { + "epoch": 3.02, + "learning_rate": 4.602713178294574e-05, + "loss": 0.2923, + "step": 13570 + }, + { + "epoch": 3.02, + "learning_rate": 4.602011009998877e-05, + "loss": 0.0816, + "step": 13580 + }, + { + "epoch": 3.02, + "learning_rate": 4.6013088417031795e-05, + "loss": 0.2785, + "step": 13590 + }, + { + "epoch": 3.02, + "learning_rate": 4.600606673407483e-05, + "loss": 0.0071, + "step": 13600 + }, + { + "epoch": 3.02, + "learning_rate": 4.5999045051117857e-05, + "loss": 0.2931, + "step": 13610 + }, + { + "epoch": 3.02, + "learning_rate": 4.599202336816088e-05, + "loss": 0.1735, + "step": 13620 + }, + { + "epoch": 3.02, + "learning_rate": 4.598500168520391e-05, + "loss": 0.78, + "step": 13630 + }, + { + "epoch": 3.02, + "learning_rate": 4.597798000224694e-05, + "loss": 0.3484, + "step": 13640 + }, + { + "epoch": 3.02, + "learning_rate": 4.5970958319289967e-05, + "loss": 1.0231, + "step": 13650 + }, + { + "epoch": 3.02, + "learning_rate": 4.5963936636333e-05, + "loss": 0.5152, + "step": 13660 + }, + { + "epoch": 3.02, + "learning_rate": 4.595691495337603e-05, + "loss": 0.359, + "step": 13670 + }, + { + "epoch": 3.02, + "learning_rate": 4.5949893270419056e-05, + "loss": 0.3902, + "step": 13680 + }, + { + "epoch": 3.02, + "learning_rate": 4.594287158746208e-05, + "loss": 0.5662, + "step": 13690 + }, + { + "epoch": 3.02, + "learning_rate": 4.593584990450511e-05, + "loss": 0.7707, + "step": 13700 + }, + { + "epoch": 3.02, + "learning_rate": 4.5928828221548145e-05, + "loss": 0.4148, + "step": 13710 + }, + { + "epoch": 3.02, + "learning_rate": 4.592180653859117e-05, + "loss": 0.3084, + "step": 13720 + }, + { + "epoch": 3.02, + "learning_rate": 4.59147848556342e-05, + "loss": 1.2329, + "step": 13730 + }, + { + "epoch": 3.02, + "learning_rate": 4.5907763172677234e-05, + "loss": 0.5036, + "step": 13740 + }, + { + "epoch": 3.02, + "learning_rate": 4.5900741489720255e-05, + "loss": 0.3222, + "step": 13750 + }, + { + "epoch": 3.02, + "learning_rate": 4.589371980676328e-05, + "loss": 0.0068, + "step": 13760 + }, + { + "epoch": 3.02, + "learning_rate": 4.588669812380632e-05, + "loss": 1.1158, + "step": 13770 + }, + { + "epoch": 3.02, + "learning_rate": 4.5879676440849344e-05, + "loss": 0.6201, + "step": 13780 + }, + { + "epoch": 3.02, + "learning_rate": 4.587265475789237e-05, + "loss": 0.0957, + "step": 13790 + }, + { + "epoch": 3.02, + "learning_rate": 4.5865633074935406e-05, + "loss": 0.6328, + "step": 13800 + }, + { + "epoch": 3.02, + "learning_rate": 4.585861139197843e-05, + "loss": 0.5255, + "step": 13810 + }, + { + "epoch": 3.02, + "learning_rate": 4.585158970902146e-05, + "loss": 0.2835, + "step": 13820 + }, + { + "epoch": 3.02, + "learning_rate": 4.584456802606449e-05, + "loss": 0.552, + "step": 13830 + }, + { + "epoch": 3.02, + "learning_rate": 4.5837546343107516e-05, + "loss": 0.3165, + "step": 13840 + }, + { + "epoch": 3.03, + "learning_rate": 4.583052466015055e-05, + "loss": 0.76, + "step": 13850 + }, + { + "epoch": 3.03, + "learning_rate": 4.582350297719358e-05, + "loss": 0.2596, + "step": 13860 + }, + { + "epoch": 3.03, + "learning_rate": 4.5816481294236605e-05, + "loss": 1.9244, + "step": 13870 + }, + { + "epoch": 3.03, + "learning_rate": 4.580945961127963e-05, + "loss": 0.2606, + "step": 13880 + }, + { + "epoch": 3.03, + "learning_rate": 4.580243792832266e-05, + "loss": 0.3859, + "step": 13890 + }, + { + "epoch": 3.03, + "learning_rate": 4.579541624536569e-05, + "loss": 0.2225, + "step": 13900 + }, + { + "epoch": 3.03, + "learning_rate": 4.578839456240872e-05, + "loss": 0.058, + "step": 13910 + }, + { + "epoch": 3.03, + "learning_rate": 4.578137287945175e-05, + "loss": 0.8261, + "step": 13920 + }, + { + "epoch": 3.03, + "learning_rate": 4.577435119649478e-05, + "loss": 0.5438, + "step": 13930 + }, + { + "epoch": 3.03, + "learning_rate": 4.5767329513537804e-05, + "loss": 0.8559, + "step": 13940 + }, + { + "epoch": 3.03, + "learning_rate": 4.576030783058083e-05, + "loss": 0.4489, + "step": 13950 + }, + { + "epoch": 3.03, + "learning_rate": 4.5753286147623866e-05, + "loss": 0.4355, + "step": 13960 + }, + { + "epoch": 3.03, + "learning_rate": 4.5746264464666894e-05, + "loss": 0.4661, + "step": 13970 + }, + { + "epoch": 3.03, + "learning_rate": 4.573924278170992e-05, + "loss": 0.84, + "step": 13980 + }, + { + "epoch": 3.03, + "learning_rate": 4.5732221098752955e-05, + "loss": 0.3722, + "step": 13990 + }, + { + "epoch": 3.03, + "learning_rate": 4.572519941579598e-05, + "loss": 0.7899, + "step": 14000 + }, + { + "epoch": 3.03, + "learning_rate": 4.571817773283901e-05, + "loss": 0.2706, + "step": 14010 + }, + { + "epoch": 3.03, + "learning_rate": 4.571115604988204e-05, + "loss": 1.6878, + "step": 14020 + }, + { + "epoch": 3.03, + "learning_rate": 4.5704134366925065e-05, + "loss": 1.2727, + "step": 14030 + }, + { + "epoch": 3.03, + "learning_rate": 4.56971126839681e-05, + "loss": 0.275, + "step": 14040 + }, + { + "epoch": 3.03, + "learning_rate": 4.569009100101113e-05, + "loss": 0.068, + "step": 14050 + }, + { + "epoch": 3.03, + "learning_rate": 4.5683069318054155e-05, + "loss": 0.0644, + "step": 14060 + }, + { + "epoch": 3.03, + "learning_rate": 4.567604763509718e-05, + "loss": 0.3697, + "step": 14070 + }, + { + "epoch": 3.03, + "learning_rate": 4.566902595214021e-05, + "loss": 0.6663, + "step": 14080 + }, + { + "epoch": 3.03, + "learning_rate": 4.566200426918324e-05, + "loss": 0.4164, + "step": 14090 + }, + { + "epoch": 3.03, + "learning_rate": 4.565498258622627e-05, + "loss": 0.1007, + "step": 14100 + }, + { + "epoch": 3.03, + "learning_rate": 4.56479609032693e-05, + "loss": 0.6764, + "step": 14110 + }, + { + "epoch": 3.03, + "learning_rate": 4.5640939220312326e-05, + "loss": 0.4913, + "step": 14120 + }, + { + "epoch": 3.03, + "learning_rate": 4.5633917537355354e-05, + "loss": 0.1382, + "step": 14130 + }, + { + "epoch": 3.03, + "learning_rate": 4.562689585439838e-05, + "loss": 0.9021, + "step": 14140 + }, + { + "epoch": 3.03, + "learning_rate": 4.5619874171441416e-05, + "loss": 0.2341, + "step": 14150 + }, + { + "epoch": 3.03, + "learning_rate": 4.561285248848444e-05, + "loss": 0.0247, + "step": 14160 + }, + { + "epoch": 3.03, + "learning_rate": 4.560583080552747e-05, + "loss": 0.7011, + "step": 14170 + }, + { + "epoch": 3.03, + "learning_rate": 4.5598809122570505e-05, + "loss": 0.6415, + "step": 14180 + }, + { + "epoch": 3.03, + "learning_rate": 4.559178743961353e-05, + "loss": 0.2448, + "step": 14190 + }, + { + "epoch": 3.03, + "learning_rate": 4.558476575665655e-05, + "loss": 0.3765, + "step": 14200 + }, + { + "epoch": 3.03, + "learning_rate": 4.557774407369959e-05, + "loss": 0.72, + "step": 14210 + }, + { + "epoch": 3.03, + "learning_rate": 4.5570722390742615e-05, + "loss": 0.2395, + "step": 14220 + }, + { + "epoch": 3.03, + "learning_rate": 4.556370070778564e-05, + "loss": 0.0853, + "step": 14230 + }, + { + "epoch": 3.03, + "learning_rate": 4.5556679024828677e-05, + "loss": 0.6115, + "step": 14240 + }, + { + "epoch": 3.03, + "learning_rate": 4.5549657341871704e-05, + "loss": 0.3561, + "step": 14250 + }, + { + "epoch": 3.03, + "learning_rate": 4.554263565891473e-05, + "loss": 0.5638, + "step": 14260 + }, + { + "epoch": 3.03, + "learning_rate": 4.553561397595776e-05, + "loss": 0.1966, + "step": 14270 + }, + { + "epoch": 3.03, + "learning_rate": 4.5528592293000787e-05, + "loss": 0.6441, + "step": 14280 + }, + { + "epoch": 3.03, + "learning_rate": 4.552157061004382e-05, + "loss": 0.3167, + "step": 14290 + }, + { + "epoch": 3.03, + "learning_rate": 4.551454892708685e-05, + "loss": 0.5829, + "step": 14300 + }, + { + "epoch": 3.03, + "learning_rate": 4.5507527244129876e-05, + "loss": 0.152, + "step": 14310 + }, + { + "epoch": 3.03, + "learning_rate": 4.55005055611729e-05, + "loss": 0.2009, + "step": 14320 + }, + { + "epoch": 3.03, + "learning_rate": 4.549348387821593e-05, + "loss": 0.2085, + "step": 14330 + }, + { + "epoch": 3.03, + "learning_rate": 4.548646219525896e-05, + "loss": 0.2535, + "step": 14340 + }, + { + "epoch": 3.03, + "learning_rate": 4.547944051230199e-05, + "loss": 0.5962, + "step": 14350 + }, + { + "epoch": 3.03, + "learning_rate": 4.547241882934502e-05, + "loss": 0.6582, + "step": 14360 + }, + { + "epoch": 3.03, + "learning_rate": 4.546539714638805e-05, + "loss": 0.2353, + "step": 14370 + }, + { + "epoch": 3.03, + "learning_rate": 4.545837546343108e-05, + "loss": 0.0929, + "step": 14380 + }, + { + "epoch": 3.03, + "learning_rate": 4.54513537804741e-05, + "loss": 0.8531, + "step": 14390 + }, + { + "epoch": 3.03, + "learning_rate": 4.544433209751714e-05, + "loss": 0.4861, + "step": 14400 + }, + { + "epoch": 3.03, + "learning_rate": 4.5437310414560164e-05, + "loss": 0.1536, + "step": 14410 + }, + { + "epoch": 3.03, + "learning_rate": 4.543028873160319e-05, + "loss": 1.1297, + "step": 14420 + }, + { + "epoch": 3.03, + "learning_rate": 4.5423267048646226e-05, + "loss": 0.6034, + "step": 14430 + }, + { + "epoch": 3.03, + "learning_rate": 4.5416245365689254e-05, + "loss": 0.8316, + "step": 14440 + }, + { + "epoch": 3.03, + "learning_rate": 4.5409223682732274e-05, + "loss": 0.2494, + "step": 14450 + }, + { + "epoch": 3.03, + "learning_rate": 4.540220199977531e-05, + "loss": 0.1555, + "step": 14460 + }, + { + "epoch": 3.03, + "learning_rate": 4.5395180316818336e-05, + "loss": 0.1696, + "step": 14470 + }, + { + "epoch": 3.03, + "learning_rate": 4.5388158633861363e-05, + "loss": 0.534, + "step": 14480 + }, + { + "epoch": 3.03, + "learning_rate": 4.53811369509044e-05, + "loss": 0.4202, + "step": 14490 + }, + { + "epoch": 3.03, + "learning_rate": 4.5374115267947425e-05, + "loss": 0.8026, + "step": 14500 + }, + { + "epoch": 3.03, + "learning_rate": 4.536709358499045e-05, + "loss": 0.0605, + "step": 14510 + }, + { + "epoch": 3.03, + "learning_rate": 4.536007190203348e-05, + "loss": 0.1074, + "step": 14520 + }, + { + "epoch": 3.03, + "learning_rate": 4.535305021907651e-05, + "loss": 0.5159, + "step": 14530 + }, + { + "epoch": 3.03, + "learning_rate": 4.534602853611954e-05, + "loss": 0.0425, + "step": 14540 + }, + { + "epoch": 3.03, + "learning_rate": 4.533900685316257e-05, + "loss": 0.7374, + "step": 14550 + }, + { + "epoch": 3.03, + "learning_rate": 4.53319851702056e-05, + "loss": 0.012, + "step": 14560 + }, + { + "epoch": 3.03, + "learning_rate": 4.532496348724863e-05, + "loss": 0.2777, + "step": 14570 + }, + { + "epoch": 3.03, + "learning_rate": 4.531794180429165e-05, + "loss": 0.2228, + "step": 14580 + }, + { + "epoch": 3.03, + "learning_rate": 4.531092012133468e-05, + "loss": 0.534, + "step": 14590 + }, + { + "epoch": 3.03, + "learning_rate": 4.5303898438377714e-05, + "loss": 0.1507, + "step": 14600 + }, + { + "epoch": 3.03, + "learning_rate": 4.529687675542074e-05, + "loss": 0.5023, + "step": 14610 + }, + { + "epoch": 3.03, + "learning_rate": 4.528985507246377e-05, + "loss": 0.4888, + "step": 14620 + }, + { + "epoch": 3.03, + "learning_rate": 4.52828333895068e-05, + "loss": 0.5932, + "step": 14630 + }, + { + "epoch": 3.04, + "learning_rate": 4.5275811706549824e-05, + "loss": 1.2883, + "step": 14640 + }, + { + "epoch": 3.04, + "learning_rate": 4.526879002359286e-05, + "loss": 0.098, + "step": 14650 + }, + { + "epoch": 3.04, + "learning_rate": 4.5261768340635885e-05, + "loss": 0.5878, + "step": 14660 + }, + { + "epoch": 3.04, + "learning_rate": 4.525474665767891e-05, + "loss": 0.3416, + "step": 14670 + }, + { + "epoch": 3.04, + "learning_rate": 4.524772497472195e-05, + "loss": 0.4698, + "step": 14680 + }, + { + "epoch": 3.04, + "learning_rate": 4.5240703291764975e-05, + "loss": 0.339, + "step": 14690 + }, + { + "epoch": 3.04, + "learning_rate": 4.5233681608808e-05, + "loss": 0.183, + "step": 14700 + }, + { + "epoch": 3.04, + "learning_rate": 4.522665992585103e-05, + "loss": 0.6586, + "step": 14710 + }, + { + "epoch": 3.04, + "learning_rate": 4.521963824289406e-05, + "loss": 0.2144, + "step": 14720 + }, + { + "epoch": 3.04, + "learning_rate": 4.5212616559937085e-05, + "loss": 0.4956, + "step": 14730 + }, + { + "epoch": 3.04, + "learning_rate": 4.520559487698012e-05, + "loss": 0.3143, + "step": 14740 + }, + { + "epoch": 3.04, + "learning_rate": 4.5198573194023146e-05, + "loss": 0.7629, + "step": 14750 + }, + { + "epoch": 3.04, + "learning_rate": 4.5191551511066174e-05, + "loss": 0.0481, + "step": 14760 + }, + { + "epoch": 3.04, + "learning_rate": 4.51845298281092e-05, + "loss": 0.2197, + "step": 14770 + }, + { + "epoch": 3.04, + "learning_rate": 4.517750814515223e-05, + "loss": 0.1949, + "step": 14780 + }, + { + "epoch": 3.04, + "learning_rate": 4.517048646219526e-05, + "loss": 1.6349, + "step": 14790 + }, + { + "epoch": 3.04, + "learning_rate": 4.516346477923829e-05, + "loss": 0.7234, + "step": 14800 + }, + { + "epoch": 3.04, + "learning_rate": 4.515644309628132e-05, + "loss": 0.2992, + "step": 14810 + }, + { + "epoch": 3.04, + "learning_rate": 4.514942141332435e-05, + "loss": 0.3752, + "step": 14820 + }, + { + "epoch": 3.04, + "learning_rate": 4.514239973036737e-05, + "loss": 0.5452, + "step": 14830 + }, + { + "epoch": 3.04, + "learning_rate": 4.51353780474104e-05, + "loss": 0.382, + "step": 14840 + }, + { + "epoch": 3.04, + "learning_rate": 4.5128356364453435e-05, + "loss": 1.1462, + "step": 14850 + }, + { + "epoch": 3.04, + "learning_rate": 4.512133468149646e-05, + "loss": 0.2113, + "step": 14860 + }, + { + "epoch": 3.04, + "learning_rate": 4.511431299853949e-05, + "loss": 0.9684, + "step": 14870 + }, + { + "epoch": 3.04, + "learning_rate": 4.5107291315582524e-05, + "loss": 0.6265, + "step": 14880 + }, + { + "epoch": 3.04, + "learning_rate": 4.510026963262555e-05, + "loss": 0.3632, + "step": 14890 + }, + { + "epoch": 3.04, + "learning_rate": 4.509324794966858e-05, + "loss": 0.5235, + "step": 14900 + }, + { + "epoch": 3.04, + "learning_rate": 4.5086226266711607e-05, + "loss": 0.3524, + "step": 14910 + }, + { + "epoch": 3.04, + "learning_rate": 4.5079204583754634e-05, + "loss": 0.0847, + "step": 14920 + }, + { + "epoch": 3.04, + "learning_rate": 4.507218290079767e-05, + "loss": 0.5692, + "step": 14930 + }, + { + "epoch": 3.04, + "learning_rate": 4.5065161217840696e-05, + "loss": 0.5134, + "step": 14940 + }, + { + "epoch": 3.04, + "learning_rate": 4.505813953488372e-05, + "loss": 0.3147, + "step": 14950 + }, + { + "epoch": 3.04, + "learning_rate": 4.505111785192675e-05, + "loss": 0.2393, + "step": 14960 + }, + { + "epoch": 3.04, + "learning_rate": 4.504409616896978e-05, + "loss": 0.1223, + "step": 14970 + }, + { + "epoch": 3.04, + "learning_rate": 4.5037074486012806e-05, + "loss": 0.2165, + "step": 14980 + }, + { + "epoch": 3.04, + "learning_rate": 4.503005280305584e-05, + "loss": 0.3428, + "step": 14990 + }, + { + "epoch": 3.04, + "learning_rate": 4.502303112009887e-05, + "loss": 0.2675, + "step": 15000 + }, + { + "epoch": 3.04, + "learning_rate": 4.5016009437141895e-05, + "loss": 0.0015, + "step": 15010 + }, + { + "epoch": 3.04, + "learning_rate": 4.500898775418492e-05, + "loss": 0.4128, + "step": 15020 + }, + { + "epoch": 3.04, + "learning_rate": 4.500196607122795e-05, + "loss": 0.3243, + "step": 15030 + }, + { + "epoch": 3.04, + "learning_rate": 4.4994944388270984e-05, + "loss": 0.9503, + "step": 15040 + }, + { + "epoch": 3.04, + "learning_rate": 4.498792270531401e-05, + "loss": 1.1233, + "step": 15050 + }, + { + "epoch": 3.04, + "learning_rate": 4.498090102235704e-05, + "loss": 0.4463, + "step": 15060 + }, + { + "epoch": 3.04, + "learning_rate": 4.4973879339400074e-05, + "loss": 0.1861, + "step": 15070 + }, + { + "epoch": 3.04, + "learning_rate": 4.49668576564431e-05, + "loss": 0.3081, + "step": 15080 + }, + { + "epoch": 3.04, + "learning_rate": 4.495983597348613e-05, + "loss": 0.4784, + "step": 15090 + }, + { + "epoch": 3.04, + "learning_rate": 4.4952814290529156e-05, + "loss": 0.7833, + "step": 15100 + }, + { + "epoch": 3.04, + "learning_rate": 4.4945792607572183e-05, + "loss": 0.4654, + "step": 15110 + }, + { + "epoch": 3.04, + "learning_rate": 4.493877092461522e-05, + "loss": 0.8979, + "step": 15120 + }, + { + "epoch": 3.04, + "learning_rate": 4.4931749241658245e-05, + "loss": 0.2478, + "step": 15130 + }, + { + "epoch": 3.04, + "learning_rate": 4.492472755870127e-05, + "loss": 0.0789, + "step": 15140 + }, + { + "epoch": 3.04, + "learning_rate": 4.49177058757443e-05, + "loss": 0.3908, + "step": 15150 + }, + { + "epoch": 3.04, + "learning_rate": 4.491068419278733e-05, + "loss": 0.2983, + "step": 15160 + }, + { + "epoch": 3.04, + "learning_rate": 4.4903662509830355e-05, + "loss": 0.2472, + "step": 15170 + }, + { + "epoch": 3.04, + "learning_rate": 4.489664082687339e-05, + "loss": 0.1016, + "step": 15180 + }, + { + "epoch": 3.04, + "learning_rate": 4.488961914391642e-05, + "loss": 0.5434, + "step": 15190 + }, + { + "epoch": 3.04, + "learning_rate": 4.4882597460959444e-05, + "loss": 0.1571, + "step": 15200 + }, + { + "epoch": 3.04, + "learning_rate": 4.487557577800247e-05, + "loss": 0.1041, + "step": 15210 + }, + { + "epoch": 3.04, + "learning_rate": 4.48685540950455e-05, + "loss": 0.1455, + "step": 15220 + }, + { + "epoch": 3.04, + "learning_rate": 4.4861532412088534e-05, + "loss": 0.4242, + "step": 15230 + }, + { + "epoch": 3.04, + "learning_rate": 4.485451072913156e-05, + "loss": 0.0219, + "step": 15240 + }, + { + "epoch": 3.04, + "learning_rate": 4.484748904617459e-05, + "loss": 0.8647, + "step": 15250 + }, + { + "epoch": 3.04, + "learning_rate": 4.484046736321762e-05, + "loss": 0.4104, + "step": 15260 + }, + { + "epoch": 3.04, + "learning_rate": 4.483344568026065e-05, + "loss": 0.877, + "step": 15270 + }, + { + "epoch": 3.04, + "learning_rate": 4.482642399730367e-05, + "loss": 0.5787, + "step": 15280 + }, + { + "epoch": 3.04, + "learning_rate": 4.4819402314346705e-05, + "loss": 0.3878, + "step": 15290 + }, + { + "epoch": 3.04, + "learning_rate": 4.481238063138973e-05, + "loss": 0.632, + "step": 15300 + }, + { + "epoch": 3.04, + "learning_rate": 4.480535894843276e-05, + "loss": 0.4179, + "step": 15310 + }, + { + "epoch": 3.04, + "learning_rate": 4.4798337265475795e-05, + "loss": 0.1773, + "step": 15320 + }, + { + "epoch": 3.04, + "learning_rate": 4.479131558251882e-05, + "loss": 0.484, + "step": 15330 + }, + { + "epoch": 3.04, + "learning_rate": 4.478429389956185e-05, + "loss": 0.7073, + "step": 15340 + }, + { + "epoch": 3.04, + "learning_rate": 4.477727221660488e-05, + "loss": 0.7751, + "step": 15350 + }, + { + "epoch": 3.04, + "learning_rate": 4.4770250533647905e-05, + "loss": 0.3569, + "step": 15360 + }, + { + "epoch": 3.04, + "learning_rate": 4.476322885069094e-05, + "loss": 0.3943, + "step": 15370 + }, + { + "epoch": 3.04, + "learning_rate": 4.4756207167733966e-05, + "loss": 0.2973, + "step": 15380 + }, + { + "epoch": 3.04, + "learning_rate": 4.4749185484776994e-05, + "loss": 0.1976, + "step": 15390 + }, + { + "epoch": 3.04, + "learning_rate": 4.474216380182003e-05, + "loss": 0.2228, + "step": 15400 + }, + { + "epoch": 3.04, + "learning_rate": 4.473514211886305e-05, + "loss": 0.9329, + "step": 15410 + }, + { + "epoch": 3.04, + "learning_rate": 4.4728120435906076e-05, + "loss": 0.9275, + "step": 15420 + }, + { + "epoch": 3.05, + "learning_rate": 4.472109875294911e-05, + "loss": 0.3051, + "step": 15430 + }, + { + "epoch": 3.05, + "learning_rate": 4.471407706999214e-05, + "loss": 0.1625, + "step": 15440 + }, + { + "epoch": 3.05, + "learning_rate": 4.4707055387035166e-05, + "loss": 0.5141, + "step": 15450 + }, + { + "epoch": 3.05, + "learning_rate": 4.47000337040782e-05, + "loss": 0.1721, + "step": 15460 + }, + { + "epoch": 3.05, + "learning_rate": 4.469301202112122e-05, + "loss": 0.0739, + "step": 15470 + }, + { + "epoch": 3.05, + "learning_rate": 4.4685990338164255e-05, + "loss": 0.3032, + "step": 15480 + }, + { + "epoch": 3.05, + "learning_rate": 4.467896865520728e-05, + "loss": 0.132, + "step": 15490 + }, + { + "epoch": 3.05, + "learning_rate": 4.467194697225031e-05, + "loss": 0.766, + "step": 15500 + }, + { + "epoch": 3.05, + "learning_rate": 4.4664925289293344e-05, + "loss": 0.4584, + "step": 15510 + }, + { + "epoch": 3.05, + "learning_rate": 4.465790360633637e-05, + "loss": 0.8463, + "step": 15520 + }, + { + "epoch": 3.05, + "learning_rate": 4.465088192337939e-05, + "loss": 0.1061, + "step": 15530 + }, + { + "epoch": 3.05, + "learning_rate": 4.4643860240422427e-05, + "loss": 0.4329, + "step": 15540 + }, + { + "epoch": 3.05, + "learning_rate": 4.4636838557465454e-05, + "loss": 0.3351, + "step": 15550 + }, + { + "epoch": 3.05, + "learning_rate": 4.462981687450848e-05, + "loss": 0.3213, + "step": 15560 + }, + { + "epoch": 3.05, + "learning_rate": 4.4622795191551516e-05, + "loss": 0.1966, + "step": 15570 + }, + { + "epoch": 3.05, + "learning_rate": 4.461577350859454e-05, + "loss": 0.3244, + "step": 15580 + }, + { + "epoch": 3.05, + "learning_rate": 4.460875182563757e-05, + "loss": 0.4696, + "step": 15590 + }, + { + "epoch": 3.05, + "learning_rate": 4.46017301426806e-05, + "loss": 0.6769, + "step": 15600 + }, + { + "epoch": 3.05, + "learning_rate": 4.4594708459723626e-05, + "loss": 0.6404, + "step": 15610 + }, + { + "epoch": 3.05, + "learning_rate": 4.458768677676666e-05, + "loss": 0.4358, + "step": 15620 + }, + { + "epoch": 3.05, + "learning_rate": 4.458066509380969e-05, + "loss": 0.4559, + "step": 15630 + }, + { + "epoch": 3.05, + "learning_rate": 4.4573643410852715e-05, + "loss": 0.2625, + "step": 15640 + }, + { + "epoch": 3.05, + "learning_rate": 4.456662172789575e-05, + "loss": 0.3927, + "step": 15650 + }, + { + "epoch": 3.05, + "learning_rate": 4.455960004493877e-05, + "loss": 0.3318, + "step": 15660 + }, + { + "epoch": 3.05, + "learning_rate": 4.45525783619818e-05, + "loss": 0.4804, + "step": 15670 + }, + { + "epoch": 3.05, + "learning_rate": 4.454555667902483e-05, + "loss": 0.9837, + "step": 15680 + }, + { + "epoch": 3.05, + "learning_rate": 4.453853499606786e-05, + "loss": 0.5476, + "step": 15690 + }, + { + "epoch": 3.05, + "learning_rate": 4.453151331311089e-05, + "loss": 0.0231, + "step": 15700 + }, + { + "epoch": 3.05, + "learning_rate": 4.452449163015392e-05, + "loss": 0.1448, + "step": 15710 + }, + { + "epoch": 3.05, + "learning_rate": 4.451746994719694e-05, + "loss": 0.6308, + "step": 15720 + }, + { + "epoch": 3.05, + "learning_rate": 4.4510448264239976e-05, + "loss": 0.8901, + "step": 15730 + }, + { + "epoch": 3.05, + "learning_rate": 4.4503426581283003e-05, + "loss": 0.0232, + "step": 15740 + }, + { + "epoch": 3.05, + "learning_rate": 4.449640489832603e-05, + "loss": 0.2968, + "step": 15750 + }, + { + "epoch": 3.05, + "learning_rate": 4.4489383215369065e-05, + "loss": 0.2622, + "step": 15760 + }, + { + "epoch": 3.05, + "learning_rate": 4.448236153241209e-05, + "loss": 0.5842, + "step": 15770 + }, + { + "epoch": 3.05, + "learning_rate": 4.447533984945512e-05, + "loss": 0.2136, + "step": 15780 + }, + { + "epoch": 3.05, + "learning_rate": 4.446831816649815e-05, + "loss": 0.0722, + "step": 15790 + }, + { + "epoch": 3.05, + "learning_rate": 4.4461296483541175e-05, + "loss": 0.5193, + "step": 15800 + }, + { + "epoch": 3.05, + "learning_rate": 4.44542748005842e-05, + "loss": 0.853, + "step": 15810 + }, + { + "epoch": 3.05, + "learning_rate": 4.444725311762724e-05, + "loss": 0.3266, + "step": 15820 + }, + { + "epoch": 3.05, + "eval_accuracy": 0.680686363049929, + "eval_loss": 1.6152681112289429, + "eval_runtime": 1180.847, + "eval_samples_per_second": 6.564, + "eval_steps_per_second": 3.282, + "step": 15824 + }, + { + "epoch": 4.0, + "learning_rate": 4.4440231434670264e-05, + "loss": 0.026, + "step": 15830 + }, + { + "epoch": 4.0, + "learning_rate": 4.443320975171329e-05, + "loss": 0.5773, + "step": 15840 + }, + { + "epoch": 4.0, + "learning_rate": 4.442618806875632e-05, + "loss": 0.9706, + "step": 15850 + }, + { + "epoch": 4.0, + "learning_rate": 4.441916638579935e-05, + "loss": 0.4446, + "step": 15860 + }, + { + "epoch": 4.0, + "learning_rate": 4.441214470284238e-05, + "loss": 1.1639, + "step": 15870 + }, + { + "epoch": 4.0, + "learning_rate": 4.440512301988541e-05, + "loss": 0.9097, + "step": 15880 + }, + { + "epoch": 4.0, + "learning_rate": 4.4398101336928436e-05, + "loss": 0.4402, + "step": 15890 + }, + { + "epoch": 4.0, + "learning_rate": 4.439107965397147e-05, + "loss": 0.3964, + "step": 15900 + }, + { + "epoch": 4.0, + "learning_rate": 4.438405797101449e-05, + "loss": 0.1012, + "step": 15910 + }, + { + "epoch": 4.0, + "learning_rate": 4.437703628805752e-05, + "loss": 0.3148, + "step": 15920 + }, + { + "epoch": 4.0, + "learning_rate": 4.437001460510055e-05, + "loss": 1.0531, + "step": 15930 + }, + { + "epoch": 4.0, + "learning_rate": 4.436299292214358e-05, + "loss": 0.4194, + "step": 15940 + }, + { + "epoch": 4.0, + "learning_rate": 4.435597123918661e-05, + "loss": 0.1872, + "step": 15950 + }, + { + "epoch": 4.0, + "learning_rate": 4.434894955622964e-05, + "loss": 0.0137, + "step": 15960 + }, + { + "epoch": 4.0, + "learning_rate": 4.434192787327267e-05, + "loss": 0.0814, + "step": 15970 + }, + { + "epoch": 4.0, + "learning_rate": 4.43349061903157e-05, + "loss": 0.0053, + "step": 15980 + }, + { + "epoch": 4.0, + "learning_rate": 4.4327884507358725e-05, + "loss": 0.4069, + "step": 15990 + }, + { + "epoch": 4.0, + "learning_rate": 4.432086282440175e-05, + "loss": 0.0038, + "step": 16000 + }, + { + "epoch": 4.0, + "learning_rate": 4.4313841141444786e-05, + "loss": 0.4972, + "step": 16010 + }, + { + "epoch": 4.0, + "learning_rate": 4.4306819458487814e-05, + "loss": 0.259, + "step": 16020 + }, + { + "epoch": 4.0, + "learning_rate": 4.429979777553084e-05, + "loss": 0.3287, + "step": 16030 + }, + { + "epoch": 4.0, + "learning_rate": 4.429277609257387e-05, + "loss": 0.4776, + "step": 16040 + }, + { + "epoch": 4.0, + "learning_rate": 4.4285754409616896e-05, + "loss": 0.1387, + "step": 16050 + }, + { + "epoch": 4.0, + "learning_rate": 4.427873272665993e-05, + "loss": 0.1164, + "step": 16060 + }, + { + "epoch": 4.0, + "learning_rate": 4.427171104370296e-05, + "loss": 0.9065, + "step": 16070 + }, + { + "epoch": 4.0, + "learning_rate": 4.4264689360745986e-05, + "loss": 0.6364, + "step": 16080 + }, + { + "epoch": 4.0, + "learning_rate": 4.425766767778902e-05, + "loss": 0.3867, + "step": 16090 + }, + { + "epoch": 4.0, + "learning_rate": 4.425064599483205e-05, + "loss": 0.3426, + "step": 16100 + }, + { + "epoch": 4.0, + "learning_rate": 4.424362431187507e-05, + "loss": 0.8831, + "step": 16110 + }, + { + "epoch": 4.0, + "learning_rate": 4.42366026289181e-05, + "loss": 0.1117, + "step": 16120 + }, + { + "epoch": 4.0, + "learning_rate": 4.422958094596113e-05, + "loss": 0.4362, + "step": 16130 + }, + { + "epoch": 4.0, + "learning_rate": 4.422255926300416e-05, + "loss": 0.57, + "step": 16140 + }, + { + "epoch": 4.0, + "learning_rate": 4.421553758004719e-05, + "loss": 0.2413, + "step": 16150 + }, + { + "epoch": 4.0, + "learning_rate": 4.420851589709022e-05, + "loss": 0.0854, + "step": 16160 + }, + { + "epoch": 4.0, + "learning_rate": 4.4201494214133247e-05, + "loss": 0.4474, + "step": 16170 + }, + { + "epoch": 4.0, + "learning_rate": 4.4194472531176274e-05, + "loss": 0.0717, + "step": 16180 + }, + { + "epoch": 4.0, + "learning_rate": 4.41874508482193e-05, + "loss": 0.3291, + "step": 16190 + }, + { + "epoch": 4.0, + "learning_rate": 4.4180429165262336e-05, + "loss": 0.6586, + "step": 16200 + }, + { + "epoch": 4.0, + "learning_rate": 4.417340748230536e-05, + "loss": 0.66, + "step": 16210 + }, + { + "epoch": 4.01, + "learning_rate": 4.416638579934839e-05, + "loss": 0.3102, + "step": 16220 + }, + { + "epoch": 4.01, + "learning_rate": 4.415936411639142e-05, + "loss": 0.0044, + "step": 16230 + }, + { + "epoch": 4.01, + "learning_rate": 4.4152342433434446e-05, + "loss": 0.4972, + "step": 16240 + }, + { + "epoch": 4.01, + "learning_rate": 4.414532075047747e-05, + "loss": 0.2298, + "step": 16250 + }, + { + "epoch": 4.01, + "learning_rate": 4.413829906752051e-05, + "loss": 0.1591, + "step": 16260 + }, + { + "epoch": 4.01, + "learning_rate": 4.4131277384563535e-05, + "loss": 0.5762, + "step": 16270 + }, + { + "epoch": 4.01, + "learning_rate": 4.412425570160656e-05, + "loss": 0.1731, + "step": 16280 + }, + { + "epoch": 4.01, + "learning_rate": 4.41172340186496e-05, + "loss": 1.1924, + "step": 16290 + }, + { + "epoch": 4.01, + "learning_rate": 4.411021233569262e-05, + "loss": 0.5324, + "step": 16300 + }, + { + "epoch": 4.01, + "learning_rate": 4.410319065273565e-05, + "loss": 0.2821, + "step": 16310 + }, + { + "epoch": 4.01, + "learning_rate": 4.409616896977868e-05, + "loss": 0.3033, + "step": 16320 + }, + { + "epoch": 4.01, + "learning_rate": 4.408914728682171e-05, + "loss": 0.2199, + "step": 16330 + }, + { + "epoch": 4.01, + "learning_rate": 4.408212560386474e-05, + "loss": 0.6283, + "step": 16340 + }, + { + "epoch": 4.01, + "learning_rate": 4.407510392090777e-05, + "loss": 0.122, + "step": 16350 + }, + { + "epoch": 4.01, + "learning_rate": 4.406808223795079e-05, + "loss": 0.1444, + "step": 16360 + }, + { + "epoch": 4.01, + "learning_rate": 4.4061060554993823e-05, + "loss": 0.0787, + "step": 16370 + }, + { + "epoch": 4.01, + "learning_rate": 4.405403887203685e-05, + "loss": 0.4394, + "step": 16380 + }, + { + "epoch": 4.01, + "learning_rate": 4.404701718907988e-05, + "loss": 0.0103, + "step": 16390 + }, + { + "epoch": 4.01, + "learning_rate": 4.403999550612291e-05, + "loss": 0.9547, + "step": 16400 + }, + { + "epoch": 4.01, + "learning_rate": 4.403297382316594e-05, + "loss": 0.6987, + "step": 16410 + }, + { + "epoch": 4.01, + "learning_rate": 4.402595214020897e-05, + "loss": 0.1724, + "step": 16420 + }, + { + "epoch": 4.01, + "learning_rate": 4.4018930457251995e-05, + "loss": 0.009, + "step": 16430 + }, + { + "epoch": 4.01, + "learning_rate": 4.401190877429502e-05, + "loss": 0.2335, + "step": 16440 + }, + { + "epoch": 4.01, + "learning_rate": 4.400488709133806e-05, + "loss": 0.0942, + "step": 16450 + }, + { + "epoch": 4.01, + "learning_rate": 4.3997865408381084e-05, + "loss": 0.4987, + "step": 16460 + }, + { + "epoch": 4.01, + "learning_rate": 4.399084372542411e-05, + "loss": 0.0781, + "step": 16470 + }, + { + "epoch": 4.01, + "learning_rate": 4.3983822042467146e-05, + "loss": 0.0119, + "step": 16480 + }, + { + "epoch": 4.01, + "learning_rate": 4.397680035951017e-05, + "loss": 0.0608, + "step": 16490 + }, + { + "epoch": 4.01, + "learning_rate": 4.3969778676553194e-05, + "loss": 0.0058, + "step": 16500 + }, + { + "epoch": 4.01, + "learning_rate": 4.396275699359623e-05, + "loss": 0.4475, + "step": 16510 + }, + { + "epoch": 4.01, + "learning_rate": 4.3955735310639256e-05, + "loss": 0.5772, + "step": 16520 + }, + { + "epoch": 4.01, + "learning_rate": 4.3948713627682284e-05, + "loss": 0.0475, + "step": 16530 + }, + { + "epoch": 4.01, + "learning_rate": 4.394169194472532e-05, + "loss": 0.2842, + "step": 16540 + }, + { + "epoch": 4.01, + "learning_rate": 4.393467026176834e-05, + "loss": 0.0681, + "step": 16550 + }, + { + "epoch": 4.01, + "learning_rate": 4.392764857881137e-05, + "loss": 0.3023, + "step": 16560 + }, + { + "epoch": 4.01, + "learning_rate": 4.39206268958544e-05, + "loss": 0.3014, + "step": 16570 + }, + { + "epoch": 4.01, + "learning_rate": 4.391360521289743e-05, + "loss": 0.8247, + "step": 16580 + }, + { + "epoch": 4.01, + "learning_rate": 4.390658352994046e-05, + "loss": 0.631, + "step": 16590 + }, + { + "epoch": 4.01, + "learning_rate": 4.389956184698349e-05, + "loss": 0.2519, + "step": 16600 + }, + { + "epoch": 4.01, + "learning_rate": 4.389254016402651e-05, + "loss": 0.0578, + "step": 16610 + }, + { + "epoch": 4.01, + "learning_rate": 4.3885518481069545e-05, + "loss": 0.3181, + "step": 16620 + }, + { + "epoch": 4.01, + "learning_rate": 4.387849679811257e-05, + "loss": 0.3388, + "step": 16630 + }, + { + "epoch": 4.01, + "learning_rate": 4.38714751151556e-05, + "loss": 0.082, + "step": 16640 + }, + { + "epoch": 4.01, + "learning_rate": 4.3864453432198634e-05, + "loss": 0.0565, + "step": 16650 + }, + { + "epoch": 4.01, + "learning_rate": 4.385743174924166e-05, + "loss": 0.9469, + "step": 16660 + }, + { + "epoch": 4.01, + "learning_rate": 4.385041006628469e-05, + "loss": 0.2321, + "step": 16670 + }, + { + "epoch": 4.01, + "learning_rate": 4.3843388383327716e-05, + "loss": 0.5888, + "step": 16680 + }, + { + "epoch": 4.01, + "learning_rate": 4.3836366700370744e-05, + "loss": 0.1409, + "step": 16690 + }, + { + "epoch": 4.01, + "learning_rate": 4.382934501741378e-05, + "loss": 0.0401, + "step": 16700 + }, + { + "epoch": 4.01, + "learning_rate": 4.3822323334456806e-05, + "loss": 0.8206, + "step": 16710 + }, + { + "epoch": 4.01, + "learning_rate": 4.381530165149983e-05, + "loss": 0.2427, + "step": 16720 + }, + { + "epoch": 4.01, + "learning_rate": 4.380827996854287e-05, + "loss": 0.3657, + "step": 16730 + }, + { + "epoch": 4.01, + "learning_rate": 4.380125828558589e-05, + "loss": 0.1549, + "step": 16740 + }, + { + "epoch": 4.01, + "learning_rate": 4.3794236602628916e-05, + "loss": 0.1409, + "step": 16750 + }, + { + "epoch": 4.01, + "learning_rate": 4.378721491967195e-05, + "loss": 0.0737, + "step": 16760 + }, + { + "epoch": 4.01, + "learning_rate": 4.378019323671498e-05, + "loss": 0.5768, + "step": 16770 + }, + { + "epoch": 4.01, + "learning_rate": 4.3773171553758005e-05, + "loss": 0.2412, + "step": 16780 + }, + { + "epoch": 4.01, + "learning_rate": 4.376614987080104e-05, + "loss": 0.797, + "step": 16790 + }, + { + "epoch": 4.01, + "learning_rate": 4.3759128187844067e-05, + "loss": 0.116, + "step": 16800 + }, + { + "epoch": 4.01, + "learning_rate": 4.3752106504887094e-05, + "loss": 0.5539, + "step": 16810 + }, + { + "epoch": 4.01, + "learning_rate": 4.374508482193012e-05, + "loss": 0.0365, + "step": 16820 + }, + { + "epoch": 4.01, + "learning_rate": 4.373806313897315e-05, + "loss": 0.6585, + "step": 16830 + }, + { + "epoch": 4.01, + "learning_rate": 4.373104145601618e-05, + "loss": 0.4766, + "step": 16840 + }, + { + "epoch": 4.01, + "learning_rate": 4.372401977305921e-05, + "loss": 0.1718, + "step": 16850 + }, + { + "epoch": 4.01, + "learning_rate": 4.371699809010224e-05, + "loss": 0.0732, + "step": 16860 + }, + { + "epoch": 4.01, + "learning_rate": 4.3709976407145266e-05, + "loss": 0.3982, + "step": 16870 + }, + { + "epoch": 4.01, + "learning_rate": 4.370295472418829e-05, + "loss": 0.006, + "step": 16880 + }, + { + "epoch": 4.01, + "learning_rate": 4.369593304123132e-05, + "loss": 0.3897, + "step": 16890 + }, + { + "epoch": 4.01, + "learning_rate": 4.3688911358274355e-05, + "loss": 0.0007, + "step": 16900 + }, + { + "epoch": 4.01, + "learning_rate": 4.368188967531738e-05, + "loss": 0.2844, + "step": 16910 + }, + { + "epoch": 4.01, + "learning_rate": 4.367486799236041e-05, + "loss": 0.3209, + "step": 16920 + }, + { + "epoch": 4.01, + "learning_rate": 4.366784630940344e-05, + "loss": 0.3425, + "step": 16930 + }, + { + "epoch": 4.01, + "learning_rate": 4.3660824626446465e-05, + "loss": 0.117, + "step": 16940 + }, + { + "epoch": 4.01, + "learning_rate": 4.36538029434895e-05, + "loss": 0.2642, + "step": 16950 + }, + { + "epoch": 4.01, + "learning_rate": 4.364678126053253e-05, + "loss": 0.257, + "step": 16960 + }, + { + "epoch": 4.01, + "learning_rate": 4.3639759577575554e-05, + "loss": 0.9548, + "step": 16970 + }, + { + "epoch": 4.01, + "learning_rate": 4.363273789461859e-05, + "loss": 0.1607, + "step": 16980 + }, + { + "epoch": 4.01, + "learning_rate": 4.3625716211661616e-05, + "loss": 0.883, + "step": 16990 + }, + { + "epoch": 4.01, + "learning_rate": 4.361869452870464e-05, + "loss": 0.2493, + "step": 17000 + }, + { + "epoch": 4.01, + "learning_rate": 4.361167284574767e-05, + "loss": 0.3497, + "step": 17010 + }, + { + "epoch": 4.02, + "learning_rate": 4.36046511627907e-05, + "loss": 0.5835, + "step": 17020 + }, + { + "epoch": 4.02, + "learning_rate": 4.3597629479833726e-05, + "loss": 0.1059, + "step": 17030 + }, + { + "epoch": 4.02, + "learning_rate": 4.359060779687676e-05, + "loss": 0.3145, + "step": 17040 + }, + { + "epoch": 4.02, + "learning_rate": 4.358358611391979e-05, + "loss": 0.5939, + "step": 17050 + }, + { + "epoch": 4.02, + "learning_rate": 4.3576564430962815e-05, + "loss": 0.3623, + "step": 17060 + }, + { + "epoch": 4.02, + "learning_rate": 4.356954274800584e-05, + "loss": 0.2493, + "step": 17070 + }, + { + "epoch": 4.02, + "learning_rate": 4.356252106504887e-05, + "loss": 0.424, + "step": 17080 + }, + { + "epoch": 4.02, + "learning_rate": 4.3555499382091904e-05, + "loss": 0.4244, + "step": 17090 + }, + { + "epoch": 4.02, + "learning_rate": 4.354847769913493e-05, + "loss": 0.6282, + "step": 17100 + }, + { + "epoch": 4.02, + "learning_rate": 4.354145601617796e-05, + "loss": 0.4529, + "step": 17110 + }, + { + "epoch": 4.02, + "learning_rate": 4.353443433322099e-05, + "loss": 0.235, + "step": 17120 + }, + { + "epoch": 4.02, + "learning_rate": 4.3527412650264014e-05, + "loss": 0.3818, + "step": 17130 + }, + { + "epoch": 4.02, + "learning_rate": 4.352039096730705e-05, + "loss": 0.0334, + "step": 17140 + }, + { + "epoch": 4.02, + "learning_rate": 4.3513369284350076e-05, + "loss": 0.7513, + "step": 17150 + }, + { + "epoch": 4.02, + "learning_rate": 4.3506347601393104e-05, + "loss": 0.1333, + "step": 17160 + }, + { + "epoch": 4.02, + "learning_rate": 4.349932591843614e-05, + "loss": 0.6374, + "step": 17170 + }, + { + "epoch": 4.02, + "learning_rate": 4.3492304235479165e-05, + "loss": 0.7738, + "step": 17180 + }, + { + "epoch": 4.02, + "learning_rate": 4.3485282552522186e-05, + "loss": 0.8072, + "step": 17190 + }, + { + "epoch": 4.02, + "learning_rate": 4.347826086956522e-05, + "loss": 0.8342, + "step": 17200 + }, + { + "epoch": 4.02, + "learning_rate": 4.347123918660825e-05, + "loss": 0.3204, + "step": 17210 + }, + { + "epoch": 4.02, + "learning_rate": 4.3464217503651275e-05, + "loss": 0.3193, + "step": 17220 + }, + { + "epoch": 4.02, + "learning_rate": 4.345719582069431e-05, + "loss": 0.2529, + "step": 17230 + }, + { + "epoch": 4.02, + "learning_rate": 4.345017413773734e-05, + "loss": 0.4349, + "step": 17240 + }, + { + "epoch": 4.02, + "learning_rate": 4.3443152454780365e-05, + "loss": 0.0524, + "step": 17250 + }, + { + "epoch": 4.02, + "learning_rate": 4.343613077182339e-05, + "loss": 0.0435, + "step": 17260 + }, + { + "epoch": 4.02, + "learning_rate": 4.342910908886642e-05, + "loss": 0.4047, + "step": 17270 + }, + { + "epoch": 4.02, + "learning_rate": 4.3422087405909454e-05, + "loss": 0.3021, + "step": 17280 + }, + { + "epoch": 4.02, + "learning_rate": 4.341506572295248e-05, + "loss": 0.3547, + "step": 17290 + }, + { + "epoch": 4.02, + "learning_rate": 4.340804403999551e-05, + "loss": 0.4034, + "step": 17300 + }, + { + "epoch": 4.02, + "learning_rate": 4.3401022357038536e-05, + "loss": 0.999, + "step": 17310 + }, + { + "epoch": 4.02, + "learning_rate": 4.3394000674081564e-05, + "loss": 0.1375, + "step": 17320 + }, + { + "epoch": 4.02, + "learning_rate": 4.338697899112459e-05, + "loss": 0.2811, + "step": 17330 + }, + { + "epoch": 4.02, + "learning_rate": 4.3379957308167626e-05, + "loss": 0.2804, + "step": 17340 + }, + { + "epoch": 4.02, + "learning_rate": 4.337293562521065e-05, + "loss": 0.7592, + "step": 17350 + }, + { + "epoch": 4.02, + "learning_rate": 4.336591394225368e-05, + "loss": 0.3536, + "step": 17360 + }, + { + "epoch": 4.02, + "learning_rate": 4.3358892259296715e-05, + "loss": 0.9396, + "step": 17370 + }, + { + "epoch": 4.02, + "learning_rate": 4.3351870576339736e-05, + "loss": 0.5309, + "step": 17380 + }, + { + "epoch": 4.02, + "learning_rate": 4.334484889338277e-05, + "loss": 0.1876, + "step": 17390 + }, + { + "epoch": 4.02, + "learning_rate": 4.33378272104258e-05, + "loss": 0.7003, + "step": 17400 + }, + { + "epoch": 4.02, + "learning_rate": 4.3330805527468825e-05, + "loss": 0.0037, + "step": 17410 + }, + { + "epoch": 4.02, + "learning_rate": 4.332378384451186e-05, + "loss": 0.0425, + "step": 17420 + }, + { + "epoch": 4.02, + "learning_rate": 4.3316762161554887e-05, + "loss": 0.4801, + "step": 17430 + }, + { + "epoch": 4.02, + "learning_rate": 4.330974047859791e-05, + "loss": 0.0381, + "step": 17440 + }, + { + "epoch": 4.02, + "learning_rate": 4.330271879564094e-05, + "loss": 0.1457, + "step": 17450 + }, + { + "epoch": 4.02, + "learning_rate": 4.329569711268397e-05, + "loss": 0.2762, + "step": 17460 + }, + { + "epoch": 4.02, + "learning_rate": 4.3288675429726996e-05, + "loss": 0.0102, + "step": 17470 + }, + { + "epoch": 4.02, + "learning_rate": 4.328165374677003e-05, + "loss": 0.6173, + "step": 17480 + }, + { + "epoch": 4.02, + "learning_rate": 4.327463206381306e-05, + "loss": 0.3212, + "step": 17490 + }, + { + "epoch": 4.02, + "learning_rate": 4.3267610380856086e-05, + "loss": 0.6573, + "step": 17500 + }, + { + "epoch": 4.02, + "learning_rate": 4.326058869789911e-05, + "loss": 0.027, + "step": 17510 + }, + { + "epoch": 4.02, + "learning_rate": 4.325356701494214e-05, + "loss": 0.5279, + "step": 17520 + }, + { + "epoch": 4.02, + "learning_rate": 4.3246545331985175e-05, + "loss": 0.025, + "step": 17530 + }, + { + "epoch": 4.02, + "learning_rate": 4.32395236490282e-05, + "loss": 0.2999, + "step": 17540 + }, + { + "epoch": 4.02, + "learning_rate": 4.323250196607123e-05, + "loss": 0.3207, + "step": 17550 + }, + { + "epoch": 4.02, + "learning_rate": 4.3225480283114264e-05, + "loss": 0.9542, + "step": 17560 + }, + { + "epoch": 4.02, + "learning_rate": 4.3218458600157285e-05, + "loss": 0.7599, + "step": 17570 + }, + { + "epoch": 4.02, + "learning_rate": 4.321143691720031e-05, + "loss": 0.1383, + "step": 17580 + }, + { + "epoch": 4.02, + "learning_rate": 4.320441523424335e-05, + "loss": 0.3806, + "step": 17590 + }, + { + "epoch": 4.02, + "learning_rate": 4.3197393551286374e-05, + "loss": 0.6663, + "step": 17600 + }, + { + "epoch": 4.02, + "learning_rate": 4.31903718683294e-05, + "loss": 0.2113, + "step": 17610 + }, + { + "epoch": 4.02, + "learning_rate": 4.3183350185372436e-05, + "loss": 0.8357, + "step": 17620 + }, + { + "epoch": 4.02, + "learning_rate": 4.317632850241546e-05, + "loss": 0.8118, + "step": 17630 + }, + { + "epoch": 4.02, + "learning_rate": 4.316930681945849e-05, + "loss": 0.0766, + "step": 17640 + }, + { + "epoch": 4.02, + "learning_rate": 4.316228513650152e-05, + "loss": 0.2071, + "step": 17650 + }, + { + "epoch": 4.02, + "learning_rate": 4.3155263453544546e-05, + "loss": 0.282, + "step": 17660 + }, + { + "epoch": 4.02, + "learning_rate": 4.314824177058758e-05, + "loss": 0.4383, + "step": 17670 + }, + { + "epoch": 4.02, + "learning_rate": 4.314122008763061e-05, + "loss": 0.1589, + "step": 17680 + }, + { + "epoch": 4.02, + "learning_rate": 4.3134198404673635e-05, + "loss": 0.1519, + "step": 17690 + }, + { + "epoch": 4.02, + "learning_rate": 4.312717672171666e-05, + "loss": 0.0091, + "step": 17700 + }, + { + "epoch": 4.02, + "learning_rate": 4.312015503875969e-05, + "loss": 0.1173, + "step": 17710 + }, + { + "epoch": 4.02, + "learning_rate": 4.311313335580272e-05, + "loss": 0.415, + "step": 17720 + }, + { + "epoch": 4.02, + "learning_rate": 4.310611167284575e-05, + "loss": 0.0312, + "step": 17730 + }, + { + "epoch": 4.02, + "learning_rate": 4.309908998988878e-05, + "loss": 0.1583, + "step": 17740 + }, + { + "epoch": 4.02, + "learning_rate": 4.309206830693181e-05, + "loss": 0.1716, + "step": 17750 + }, + { + "epoch": 4.02, + "learning_rate": 4.3085046623974834e-05, + "loss": 0.4887, + "step": 17760 + }, + { + "epoch": 4.02, + "learning_rate": 4.307802494101786e-05, + "loss": 0.2562, + "step": 17770 + }, + { + "epoch": 4.02, + "learning_rate": 4.3071003258060896e-05, + "loss": 0.769, + "step": 17780 + }, + { + "epoch": 4.02, + "learning_rate": 4.3063981575103924e-05, + "loss": 0.6407, + "step": 17790 + }, + { + "epoch": 4.02, + "learning_rate": 4.305695989214695e-05, + "loss": 0.1667, + "step": 17800 + }, + { + "epoch": 4.03, + "learning_rate": 4.3049938209189985e-05, + "loss": 0.3828, + "step": 17810 + }, + { + "epoch": 4.03, + "learning_rate": 4.3042916526233006e-05, + "loss": 0.0025, + "step": 17820 + }, + { + "epoch": 4.03, + "learning_rate": 4.3035894843276034e-05, + "loss": 0.0688, + "step": 17830 + }, + { + "epoch": 4.03, + "learning_rate": 4.302887316031907e-05, + "loss": 0.022, + "step": 17840 + }, + { + "epoch": 4.03, + "learning_rate": 4.3021851477362095e-05, + "loss": 0.543, + "step": 17850 + }, + { + "epoch": 4.03, + "learning_rate": 4.301482979440512e-05, + "loss": 0.0012, + "step": 17860 + }, + { + "epoch": 4.03, + "learning_rate": 4.300780811144816e-05, + "loss": 0.7413, + "step": 17870 + }, + { + "epoch": 4.03, + "learning_rate": 4.3000786428491185e-05, + "loss": 0.1632, + "step": 17880 + }, + { + "epoch": 4.03, + "learning_rate": 4.299376474553421e-05, + "loss": 0.2114, + "step": 17890 + }, + { + "epoch": 4.03, + "learning_rate": 4.298674306257724e-05, + "loss": 0.5477, + "step": 17900 + }, + { + "epoch": 4.03, + "learning_rate": 4.297972137962027e-05, + "loss": 0.0258, + "step": 17910 + }, + { + "epoch": 4.03, + "learning_rate": 4.29726996966633e-05, + "loss": 0.322, + "step": 17920 + }, + { + "epoch": 4.03, + "learning_rate": 4.296567801370633e-05, + "loss": 0.103, + "step": 17930 + }, + { + "epoch": 4.03, + "learning_rate": 4.2958656330749356e-05, + "loss": 0.0069, + "step": 17940 + }, + { + "epoch": 4.03, + "learning_rate": 4.2951634647792384e-05, + "loss": 0.3623, + "step": 17950 + }, + { + "epoch": 4.03, + "learning_rate": 4.294461296483541e-05, + "loss": 0.7134, + "step": 17960 + }, + { + "epoch": 4.03, + "learning_rate": 4.293759128187844e-05, + "loss": 0.5983, + "step": 17970 + }, + { + "epoch": 4.03, + "learning_rate": 4.293056959892147e-05, + "loss": 0.0095, + "step": 17980 + }, + { + "epoch": 4.03, + "learning_rate": 4.29235479159645e-05, + "loss": 0.1268, + "step": 17990 + }, + { + "epoch": 4.03, + "learning_rate": 4.291652623300753e-05, + "loss": 0.8201, + "step": 18000 + }, + { + "epoch": 4.03, + "learning_rate": 4.2909504550050556e-05, + "loss": 0.0608, + "step": 18010 + }, + { + "epoch": 4.03, + "learning_rate": 4.290248286709358e-05, + "loss": 0.1335, + "step": 18020 + }, + { + "epoch": 4.03, + "learning_rate": 4.289546118413662e-05, + "loss": 0.3689, + "step": 18030 + }, + { + "epoch": 4.03, + "learning_rate": 4.2888439501179645e-05, + "loss": 0.3436, + "step": 18040 + }, + { + "epoch": 4.03, + "learning_rate": 4.288141781822267e-05, + "loss": 0.2885, + "step": 18050 + }, + { + "epoch": 4.03, + "learning_rate": 4.2874396135265707e-05, + "loss": 0.0173, + "step": 18060 + }, + { + "epoch": 4.03, + "learning_rate": 4.2867374452308734e-05, + "loss": 0.3822, + "step": 18070 + }, + { + "epoch": 4.03, + "learning_rate": 4.286035276935176e-05, + "loss": 0.5167, + "step": 18080 + }, + { + "epoch": 4.03, + "learning_rate": 4.285333108639479e-05, + "loss": 0.61, + "step": 18090 + }, + { + "epoch": 4.03, + "learning_rate": 4.2846309403437816e-05, + "loss": 0.0222, + "step": 18100 + }, + { + "epoch": 4.03, + "learning_rate": 4.283928772048085e-05, + "loss": 0.2067, + "step": 18110 + }, + { + "epoch": 4.03, + "learning_rate": 4.283226603752388e-05, + "loss": 0.1775, + "step": 18120 + }, + { + "epoch": 4.03, + "learning_rate": 4.2825244354566906e-05, + "loss": 0.0515, + "step": 18130 + }, + { + "epoch": 4.03, + "learning_rate": 4.281822267160993e-05, + "loss": 0.3773, + "step": 18140 + }, + { + "epoch": 4.03, + "learning_rate": 4.281120098865296e-05, + "loss": 0.0461, + "step": 18150 + }, + { + "epoch": 4.03, + "learning_rate": 4.280417930569599e-05, + "loss": 0.072, + "step": 18160 + }, + { + "epoch": 4.03, + "learning_rate": 4.279715762273902e-05, + "loss": 0.002, + "step": 18170 + }, + { + "epoch": 4.03, + "learning_rate": 4.279013593978205e-05, + "loss": 0.1392, + "step": 18180 + }, + { + "epoch": 4.03, + "learning_rate": 4.278311425682508e-05, + "loss": 0.016, + "step": 18190 + }, + { + "epoch": 4.03, + "learning_rate": 4.277609257386811e-05, + "loss": 0.2001, + "step": 18200 + }, + { + "epoch": 4.03, + "learning_rate": 4.276907089091113e-05, + "loss": 0.0228, + "step": 18210 + }, + { + "epoch": 4.03, + "learning_rate": 4.276204920795417e-05, + "loss": 0.8933, + "step": 18220 + }, + { + "epoch": 4.03, + "learning_rate": 4.2755027524997194e-05, + "loss": 0.1167, + "step": 18230 + }, + { + "epoch": 4.03, + "learning_rate": 4.274800584204022e-05, + "loss": 0.493, + "step": 18240 + }, + { + "epoch": 4.03, + "learning_rate": 4.2740984159083256e-05, + "loss": 0.2059, + "step": 18250 + }, + { + "epoch": 4.03, + "learning_rate": 4.2733962476126283e-05, + "loss": 0.3072, + "step": 18260 + }, + { + "epoch": 4.03, + "learning_rate": 4.2726940793169304e-05, + "loss": 0.5644, + "step": 18270 + }, + { + "epoch": 4.03, + "learning_rate": 4.271991911021234e-05, + "loss": 0.278, + "step": 18280 + }, + { + "epoch": 4.03, + "learning_rate": 4.2712897427255366e-05, + "loss": 0.3131, + "step": 18290 + }, + { + "epoch": 4.03, + "learning_rate": 4.2705875744298393e-05, + "loss": 0.4197, + "step": 18300 + }, + { + "epoch": 4.03, + "learning_rate": 4.269885406134143e-05, + "loss": 0.2874, + "step": 18310 + }, + { + "epoch": 4.03, + "learning_rate": 4.2691832378384455e-05, + "loss": 0.3993, + "step": 18320 + }, + { + "epoch": 4.03, + "learning_rate": 4.268481069542748e-05, + "loss": 0.2259, + "step": 18330 + }, + { + "epoch": 4.03, + "learning_rate": 4.267778901247051e-05, + "loss": 1.1462, + "step": 18340 + }, + { + "epoch": 4.03, + "learning_rate": 4.267076732951354e-05, + "loss": 0.8444, + "step": 18350 + }, + { + "epoch": 4.03, + "learning_rate": 4.266374564655657e-05, + "loss": 0.5872, + "step": 18360 + }, + { + "epoch": 4.03, + "learning_rate": 4.26567239635996e-05, + "loss": 0.5974, + "step": 18370 + }, + { + "epoch": 4.03, + "learning_rate": 4.264970228064263e-05, + "loss": 0.6346, + "step": 18380 + }, + { + "epoch": 4.03, + "learning_rate": 4.264268059768566e-05, + "loss": 0.2639, + "step": 18390 + }, + { + "epoch": 4.03, + "learning_rate": 4.263565891472868e-05, + "loss": 0.5373, + "step": 18400 + }, + { + "epoch": 4.03, + "learning_rate": 4.262863723177171e-05, + "loss": 0.4375, + "step": 18410 + }, + { + "epoch": 4.03, + "learning_rate": 4.2621615548814744e-05, + "loss": 1.1992, + "step": 18420 + }, + { + "epoch": 4.03, + "learning_rate": 4.261459386585777e-05, + "loss": 0.1925, + "step": 18430 + }, + { + "epoch": 4.03, + "learning_rate": 4.26075721829008e-05, + "loss": 0.0224, + "step": 18440 + }, + { + "epoch": 4.03, + "learning_rate": 4.260055049994383e-05, + "loss": 0.567, + "step": 18450 + }, + { + "epoch": 4.03, + "learning_rate": 4.2593528816986854e-05, + "loss": 0.4246, + "step": 18460 + }, + { + "epoch": 4.03, + "learning_rate": 4.258650713402989e-05, + "loss": 0.2773, + "step": 18470 + }, + { + "epoch": 4.03, + "learning_rate": 4.2579485451072915e-05, + "loss": 0.1581, + "step": 18480 + }, + { + "epoch": 4.03, + "learning_rate": 4.257246376811594e-05, + "loss": 0.303, + "step": 18490 + }, + { + "epoch": 4.03, + "learning_rate": 4.256544208515898e-05, + "loss": 0.3366, + "step": 18500 + }, + { + "epoch": 4.03, + "learning_rate": 4.2558420402202005e-05, + "loss": 0.051, + "step": 18510 + }, + { + "epoch": 4.03, + "learning_rate": 4.2551398719245025e-05, + "loss": 0.0816, + "step": 18520 + }, + { + "epoch": 4.03, + "learning_rate": 4.254437703628806e-05, + "loss": 0.3003, + "step": 18530 + }, + { + "epoch": 4.03, + "learning_rate": 4.253735535333109e-05, + "loss": 0.3555, + "step": 18540 + }, + { + "epoch": 4.03, + "learning_rate": 4.2530333670374115e-05, + "loss": 0.0214, + "step": 18550 + }, + { + "epoch": 4.03, + "learning_rate": 4.252331198741715e-05, + "loss": 0.0161, + "step": 18560 + }, + { + "epoch": 4.03, + "learning_rate": 4.2516290304460176e-05, + "loss": 0.4587, + "step": 18570 + }, + { + "epoch": 4.03, + "learning_rate": 4.2509268621503204e-05, + "loss": 0.2694, + "step": 18580 + }, + { + "epoch": 4.03, + "learning_rate": 4.250224693854623e-05, + "loss": 0.8612, + "step": 18590 + }, + { + "epoch": 4.04, + "learning_rate": 4.249522525558926e-05, + "loss": 0.3936, + "step": 18600 + }, + { + "epoch": 4.04, + "learning_rate": 4.248820357263229e-05, + "loss": 0.314, + "step": 18610 + }, + { + "epoch": 4.04, + "learning_rate": 4.248118188967532e-05, + "loss": 0.7967, + "step": 18620 + }, + { + "epoch": 4.04, + "learning_rate": 4.247416020671835e-05, + "loss": 0.0603, + "step": 18630 + }, + { + "epoch": 4.04, + "learning_rate": 4.246713852376138e-05, + "loss": 0.0054, + "step": 18640 + }, + { + "epoch": 4.04, + "learning_rate": 4.24601168408044e-05, + "loss": 0.5488, + "step": 18650 + }, + { + "epoch": 4.04, + "learning_rate": 4.245309515784743e-05, + "loss": 0.4361, + "step": 18660 + }, + { + "epoch": 4.04, + "learning_rate": 4.2446073474890465e-05, + "loss": 0.6551, + "step": 18670 + }, + { + "epoch": 4.04, + "learning_rate": 4.243905179193349e-05, + "loss": 0.186, + "step": 18680 + }, + { + "epoch": 4.04, + "learning_rate": 4.243203010897652e-05, + "loss": 0.8202, + "step": 18690 + }, + { + "epoch": 4.04, + "learning_rate": 4.2425008426019554e-05, + "loss": 0.5632, + "step": 18700 + }, + { + "epoch": 4.04, + "learning_rate": 4.2417986743062575e-05, + "loss": 0.3043, + "step": 18710 + }, + { + "epoch": 4.04, + "learning_rate": 4.241096506010561e-05, + "loss": 1.282, + "step": 18720 + }, + { + "epoch": 4.04, + "learning_rate": 4.2403943377148636e-05, + "loss": 0.8839, + "step": 18730 + }, + { + "epoch": 4.04, + "learning_rate": 4.2396921694191664e-05, + "loss": 0.2358, + "step": 18740 + }, + { + "epoch": 4.04, + "learning_rate": 4.23899000112347e-05, + "loss": 0.2491, + "step": 18750 + }, + { + "epoch": 4.04, + "learning_rate": 4.2382878328277726e-05, + "loss": 0.0338, + "step": 18760 + }, + { + "epoch": 4.04, + "learning_rate": 4.237585664532075e-05, + "loss": 0.126, + "step": 18770 + }, + { + "epoch": 4.04, + "learning_rate": 4.236883496236378e-05, + "loss": 0.0888, + "step": 18780 + }, + { + "epoch": 4.04, + "learning_rate": 4.236181327940681e-05, + "loss": 1.0392, + "step": 18790 + }, + { + "epoch": 4.04, + "learning_rate": 4.2354791596449836e-05, + "loss": 0.3894, + "step": 18800 + }, + { + "epoch": 4.04, + "learning_rate": 4.234776991349287e-05, + "loss": 0.1548, + "step": 18810 + }, + { + "epoch": 4.04, + "learning_rate": 4.23407482305359e-05, + "loss": 0.3386, + "step": 18820 + }, + { + "epoch": 4.04, + "learning_rate": 4.2333726547578925e-05, + "loss": 0.574, + "step": 18830 + }, + { + "epoch": 4.04, + "learning_rate": 4.232670486462195e-05, + "loss": 0.6928, + "step": 18840 + }, + { + "epoch": 4.04, + "learning_rate": 4.231968318166498e-05, + "loss": 0.2358, + "step": 18850 + }, + { + "epoch": 4.04, + "learning_rate": 4.2312661498708014e-05, + "loss": 0.4685, + "step": 18860 + }, + { + "epoch": 4.04, + "learning_rate": 4.230563981575104e-05, + "loss": 0.0699, + "step": 18870 + }, + { + "epoch": 4.04, + "learning_rate": 4.229861813279407e-05, + "loss": 0.2197, + "step": 18880 + }, + { + "epoch": 4.04, + "learning_rate": 4.2291596449837103e-05, + "loss": 0.3226, + "step": 18890 + }, + { + "epoch": 4.04, + "learning_rate": 4.2284574766880124e-05, + "loss": 0.0874, + "step": 18900 + }, + { + "epoch": 4.04, + "learning_rate": 4.227755308392315e-05, + "loss": 0.2611, + "step": 18910 + }, + { + "epoch": 4.04, + "learning_rate": 4.2270531400966186e-05, + "loss": 1.1402, + "step": 18920 + }, + { + "epoch": 4.04, + "learning_rate": 4.226350971800921e-05, + "loss": 0.2299, + "step": 18930 + }, + { + "epoch": 4.04, + "learning_rate": 4.225648803505224e-05, + "loss": 0.2082, + "step": 18940 + }, + { + "epoch": 4.04, + "learning_rate": 4.2249466352095275e-05, + "loss": 0.6044, + "step": 18950 + }, + { + "epoch": 4.04, + "learning_rate": 4.22424446691383e-05, + "loss": 0.3496, + "step": 18960 + }, + { + "epoch": 4.04, + "learning_rate": 4.223542298618133e-05, + "loss": 1.352, + "step": 18970 + }, + { + "epoch": 4.04, + "learning_rate": 4.222840130322436e-05, + "loss": 0.2067, + "step": 18980 + }, + { + "epoch": 4.04, + "learning_rate": 4.2221379620267385e-05, + "loss": 0.0111, + "step": 18990 + }, + { + "epoch": 4.04, + "learning_rate": 4.221435793731042e-05, + "loss": 0.5011, + "step": 19000 + }, + { + "epoch": 4.04, + "learning_rate": 4.220733625435345e-05, + "loss": 1.4403, + "step": 19010 + }, + { + "epoch": 4.04, + "learning_rate": 4.2200314571396474e-05, + "loss": 0.5286, + "step": 19020 + }, + { + "epoch": 4.04, + "learning_rate": 4.21932928884395e-05, + "loss": 1.2216, + "step": 19030 + }, + { + "epoch": 4.04, + "learning_rate": 4.218627120548253e-05, + "loss": 0.0122, + "step": 19040 + }, + { + "epoch": 4.04, + "learning_rate": 4.217924952252556e-05, + "loss": 0.3497, + "step": 19050 + }, + { + "epoch": 4.04, + "learning_rate": 4.217222783956859e-05, + "loss": 0.3681, + "step": 19060 + }, + { + "epoch": 4.04, + "learning_rate": 4.216520615661162e-05, + "loss": 0.0189, + "step": 19070 + }, + { + "epoch": 4.04, + "learning_rate": 4.2158184473654646e-05, + "loss": 0.0676, + "step": 19080 + }, + { + "epoch": 4.04, + "learning_rate": 4.215116279069768e-05, + "loss": 0.0372, + "step": 19090 + }, + { + "epoch": 4.04, + "learning_rate": 4.21441411077407e-05, + "loss": 0.043, + "step": 19100 + }, + { + "epoch": 4.04, + "learning_rate": 4.2137119424783735e-05, + "loss": 0.055, + "step": 19110 + }, + { + "epoch": 4.04, + "learning_rate": 4.213009774182676e-05, + "loss": 0.6016, + "step": 19120 + }, + { + "epoch": 4.04, + "learning_rate": 4.212307605886979e-05, + "loss": 0.1234, + "step": 19130 + }, + { + "epoch": 4.04, + "learning_rate": 4.2116054375912825e-05, + "loss": 0.2356, + "step": 19140 + }, + { + "epoch": 4.04, + "learning_rate": 4.210903269295585e-05, + "loss": 0.0178, + "step": 19150 + }, + { + "epoch": 4.04, + "learning_rate": 4.210201100999888e-05, + "loss": 0.5614, + "step": 19160 + }, + { + "epoch": 4.04, + "learning_rate": 4.209498932704191e-05, + "loss": 0.2364, + "step": 19170 + }, + { + "epoch": 4.04, + "learning_rate": 4.2087967644084935e-05, + "loss": 0.6089, + "step": 19180 + }, + { + "epoch": 4.04, + "learning_rate": 4.208094596112797e-05, + "loss": 0.5251, + "step": 19190 + }, + { + "epoch": 4.04, + "learning_rate": 4.2073924278170996e-05, + "loss": 0.4731, + "step": 19200 + }, + { + "epoch": 4.04, + "learning_rate": 4.2066902595214024e-05, + "loss": 0.3806, + "step": 19210 + }, + { + "epoch": 4.04, + "learning_rate": 4.205988091225705e-05, + "loss": 0.0127, + "step": 19220 + }, + { + "epoch": 4.04, + "learning_rate": 4.205285922930008e-05, + "loss": 1.0005, + "step": 19230 + }, + { + "epoch": 4.04, + "learning_rate": 4.2045837546343106e-05, + "loss": 1.1088, + "step": 19240 + }, + { + "epoch": 4.04, + "learning_rate": 4.203881586338614e-05, + "loss": 0.4342, + "step": 19250 + }, + { + "epoch": 4.04, + "learning_rate": 4.203179418042917e-05, + "loss": 0.0791, + "step": 19260 + }, + { + "epoch": 4.04, + "learning_rate": 4.2024772497472195e-05, + "loss": 0.6531, + "step": 19270 + }, + { + "epoch": 4.04, + "learning_rate": 4.201775081451523e-05, + "loss": 0.2368, + "step": 19280 + }, + { + "epoch": 4.04, + "learning_rate": 4.201072913155825e-05, + "loss": 0.4924, + "step": 19290 + }, + { + "epoch": 4.04, + "learning_rate": 4.2003707448601285e-05, + "loss": 0.7764, + "step": 19300 + }, + { + "epoch": 4.04, + "learning_rate": 4.199668576564431e-05, + "loss": 0.485, + "step": 19310 + }, + { + "epoch": 4.04, + "learning_rate": 4.198966408268734e-05, + "loss": 0.0576, + "step": 19320 + }, + { + "epoch": 4.04, + "learning_rate": 4.1982642399730374e-05, + "loss": 0.3486, + "step": 19330 + }, + { + "epoch": 4.04, + "learning_rate": 4.19756207167734e-05, + "loss": 0.0141, + "step": 19340 + }, + { + "epoch": 4.04, + "learning_rate": 4.196859903381642e-05, + "loss": 0.5185, + "step": 19350 + }, + { + "epoch": 4.04, + "learning_rate": 4.1961577350859456e-05, + "loss": 0.3166, + "step": 19360 + }, + { + "epoch": 4.04, + "learning_rate": 4.1954555667902484e-05, + "loss": 0.7583, + "step": 19370 + }, + { + "epoch": 4.04, + "learning_rate": 4.194753398494551e-05, + "loss": 0.0401, + "step": 19380 + }, + { + "epoch": 4.05, + "learning_rate": 4.1940512301988546e-05, + "loss": 0.3695, + "step": 19390 + }, + { + "epoch": 4.05, + "learning_rate": 4.193349061903157e-05, + "loss": 0.3012, + "step": 19400 + }, + { + "epoch": 4.05, + "learning_rate": 4.19264689360746e-05, + "loss": 0.4053, + "step": 19410 + }, + { + "epoch": 4.05, + "learning_rate": 4.191944725311763e-05, + "loss": 1.2622, + "step": 19420 + }, + { + "epoch": 4.05, + "learning_rate": 4.1912425570160656e-05, + "loss": 0.4409, + "step": 19430 + }, + { + "epoch": 4.05, + "learning_rate": 4.190540388720369e-05, + "loss": 0.6796, + "step": 19440 + }, + { + "epoch": 4.05, + "learning_rate": 4.189838220424672e-05, + "loss": 0.0345, + "step": 19450 + }, + { + "epoch": 4.05, + "learning_rate": 4.1891360521289745e-05, + "loss": 0.5538, + "step": 19460 + }, + { + "epoch": 4.05, + "learning_rate": 4.188433883833278e-05, + "loss": 0.2858, + "step": 19470 + }, + { + "epoch": 4.05, + "learning_rate": 4.18773171553758e-05, + "loss": 1.2019, + "step": 19480 + }, + { + "epoch": 4.05, + "learning_rate": 4.187029547241883e-05, + "loss": 0.0073, + "step": 19490 + }, + { + "epoch": 4.05, + "learning_rate": 4.186327378946186e-05, + "loss": 1.0368, + "step": 19500 + }, + { + "epoch": 4.05, + "learning_rate": 4.185625210650489e-05, + "loss": 0.1548, + "step": 19510 + }, + { + "epoch": 4.05, + "learning_rate": 4.184923042354792e-05, + "loss": 0.105, + "step": 19520 + }, + { + "epoch": 4.05, + "learning_rate": 4.184220874059095e-05, + "loss": 0.4202, + "step": 19530 + }, + { + "epoch": 4.05, + "learning_rate": 4.183518705763397e-05, + "loss": 0.4431, + "step": 19540 + }, + { + "epoch": 4.05, + "learning_rate": 4.1828165374677006e-05, + "loss": 0.0315, + "step": 19550 + }, + { + "epoch": 4.05, + "learning_rate": 4.182114369172003e-05, + "loss": 0.2672, + "step": 19560 + }, + { + "epoch": 4.05, + "learning_rate": 4.181412200876306e-05, + "loss": 0.0162, + "step": 19570 + }, + { + "epoch": 4.05, + "learning_rate": 4.1807100325806095e-05, + "loss": 0.2237, + "step": 19580 + }, + { + "epoch": 4.05, + "learning_rate": 4.180007864284912e-05, + "loss": 0.2805, + "step": 19590 + }, + { + "epoch": 4.05, + "learning_rate": 4.179305695989214e-05, + "loss": 0.1041, + "step": 19600 + }, + { + "epoch": 4.05, + "learning_rate": 4.178603527693518e-05, + "loss": 0.2633, + "step": 19610 + }, + { + "epoch": 4.05, + "learning_rate": 4.1779013593978205e-05, + "loss": 0.2737, + "step": 19620 + }, + { + "epoch": 4.05, + "learning_rate": 4.177199191102123e-05, + "loss": 1.1189, + "step": 19630 + }, + { + "epoch": 4.05, + "learning_rate": 4.176497022806427e-05, + "loss": 0.1607, + "step": 19640 + }, + { + "epoch": 4.05, + "learning_rate": 4.1757948545107294e-05, + "loss": 0.5202, + "step": 19650 + }, + { + "epoch": 4.05, + "learning_rate": 4.175092686215032e-05, + "loss": 0.1825, + "step": 19660 + }, + { + "epoch": 4.05, + "learning_rate": 4.174390517919335e-05, + "loss": 0.6251, + "step": 19670 + }, + { + "epoch": 4.05, + "learning_rate": 4.173688349623638e-05, + "loss": 0.5171, + "step": 19680 + }, + { + "epoch": 4.05, + "learning_rate": 4.172986181327941e-05, + "loss": 0.3907, + "step": 19690 + }, + { + "epoch": 4.05, + "learning_rate": 4.172284013032244e-05, + "loss": 0.6252, + "step": 19700 + }, + { + "epoch": 4.05, + "learning_rate": 4.1715818447365466e-05, + "loss": 0.0939, + "step": 19710 + }, + { + "epoch": 4.05, + "learning_rate": 4.17087967644085e-05, + "loss": 0.5122, + "step": 19720 + }, + { + "epoch": 4.05, + "learning_rate": 4.170177508145152e-05, + "loss": 0.8829, + "step": 19730 + }, + { + "epoch": 4.05, + "learning_rate": 4.169475339849455e-05, + "loss": 0.7042, + "step": 19740 + }, + { + "epoch": 4.05, + "learning_rate": 4.168773171553758e-05, + "loss": 0.2323, + "step": 19750 + }, + { + "epoch": 4.05, + "learning_rate": 4.168071003258061e-05, + "loss": 0.6806, + "step": 19760 + }, + { + "epoch": 4.05, + "learning_rate": 4.167368834962364e-05, + "loss": 0.2257, + "step": 19770 + }, + { + "epoch": 4.05, + "learning_rate": 4.166666666666667e-05, + "loss": 0.6674, + "step": 19780 + }, + { + "epoch": 4.05, + "eval_accuracy": 0.7147464843246033, + "eval_loss": 1.500740647315979, + "eval_runtime": 1228.7695, + "eval_samples_per_second": 6.308, + "eval_steps_per_second": 3.154, + "step": 19780 + }, + { + "epoch": 5.0, + "learning_rate": 4.16596449837097e-05, + "loss": 0.0071, + "step": 19790 + }, + { + "epoch": 5.0, + "learning_rate": 4.165262330075273e-05, + "loss": 0.3421, + "step": 19800 + }, + { + "epoch": 5.0, + "learning_rate": 4.1645601617795755e-05, + "loss": 0.03, + "step": 19810 + }, + { + "epoch": 5.0, + "learning_rate": 4.163857993483878e-05, + "loss": 0.2115, + "step": 19820 + }, + { + "epoch": 5.0, + "learning_rate": 4.1631558251881816e-05, + "loss": 0.2927, + "step": 19830 + }, + { + "epoch": 5.0, + "learning_rate": 4.1624536568924844e-05, + "loss": 0.4403, + "step": 19840 + }, + { + "epoch": 5.0, + "learning_rate": 4.161751488596787e-05, + "loss": 0.0801, + "step": 19850 + }, + { + "epoch": 5.0, + "learning_rate": 4.16104932030109e-05, + "loss": 0.0873, + "step": 19860 + }, + { + "epoch": 5.0, + "learning_rate": 4.1603471520053926e-05, + "loss": 0.0575, + "step": 19870 + }, + { + "epoch": 5.0, + "learning_rate": 4.1596449837096954e-05, + "loss": 0.0134, + "step": 19880 + }, + { + "epoch": 5.0, + "learning_rate": 4.158942815413999e-05, + "loss": 0.0172, + "step": 19890 + }, + { + "epoch": 5.0, + "learning_rate": 4.1582406471183015e-05, + "loss": 0.001, + "step": 19900 + }, + { + "epoch": 5.0, + "learning_rate": 4.157538478822604e-05, + "loss": 0.3875, + "step": 19910 + }, + { + "epoch": 5.0, + "learning_rate": 4.156836310526907e-05, + "loss": 0.5486, + "step": 19920 + }, + { + "epoch": 5.0, + "learning_rate": 4.15613414223121e-05, + "loss": 0.2297, + "step": 19930 + }, + { + "epoch": 5.0, + "learning_rate": 4.155431973935513e-05, + "loss": 0.0091, + "step": 19940 + }, + { + "epoch": 5.0, + "learning_rate": 4.154729805639816e-05, + "loss": 0.3023, + "step": 19950 + }, + { + "epoch": 5.0, + "learning_rate": 4.154027637344119e-05, + "loss": 0.2442, + "step": 19960 + }, + { + "epoch": 5.0, + "learning_rate": 4.153325469048422e-05, + "loss": 0.0385, + "step": 19970 + }, + { + "epoch": 5.0, + "learning_rate": 4.152623300752725e-05, + "loss": 0.275, + "step": 19980 + }, + { + "epoch": 5.0, + "learning_rate": 4.151921132457027e-05, + "loss": 0.0589, + "step": 19990 + }, + { + "epoch": 5.0, + "learning_rate": 4.1512189641613304e-05, + "loss": 0.083, + "step": 20000 + }, + { + "epoch": 5.0, + "learning_rate": 4.150516795865633e-05, + "loss": 0.0755, + "step": 20010 + }, + { + "epoch": 5.0, + "learning_rate": 4.149814627569936e-05, + "loss": 0.0992, + "step": 20020 + }, + { + "epoch": 5.0, + "learning_rate": 4.149112459274239e-05, + "loss": 0.0105, + "step": 20030 + }, + { + "epoch": 5.0, + "learning_rate": 4.148410290978542e-05, + "loss": 0.0749, + "step": 20040 + }, + { + "epoch": 5.0, + "learning_rate": 4.147708122682845e-05, + "loss": 0.1422, + "step": 20050 + }, + { + "epoch": 5.0, + "learning_rate": 4.1470059543871476e-05, + "loss": 0.0863, + "step": 20060 + }, + { + "epoch": 5.0, + "learning_rate": 4.14630378609145e-05, + "loss": 0.1184, + "step": 20070 + }, + { + "epoch": 5.0, + "learning_rate": 4.145601617795754e-05, + "loss": 0.6078, + "step": 20080 + }, + { + "epoch": 5.0, + "learning_rate": 4.1448994495000565e-05, + "loss": 0.209, + "step": 20090 + }, + { + "epoch": 5.0, + "learning_rate": 4.144197281204359e-05, + "loss": 0.9421, + "step": 20100 + }, + { + "epoch": 5.0, + "learning_rate": 4.143495112908662e-05, + "loss": 0.2165, + "step": 20110 + }, + { + "epoch": 5.0, + "learning_rate": 4.142792944612965e-05, + "loss": 0.2071, + "step": 20120 + }, + { + "epoch": 5.0, + "learning_rate": 4.142090776317268e-05, + "loss": 0.1787, + "step": 20130 + }, + { + "epoch": 5.0, + "learning_rate": 4.141388608021571e-05, + "loss": 0.0444, + "step": 20140 + }, + { + "epoch": 5.0, + "learning_rate": 4.140686439725874e-05, + "loss": 0.001, + "step": 20150 + }, + { + "epoch": 5.0, + "learning_rate": 4.139984271430177e-05, + "loss": 0.0347, + "step": 20160 + }, + { + "epoch": 5.0, + "learning_rate": 4.13928210313448e-05, + "loss": 0.4662, + "step": 20170 + }, + { + "epoch": 5.01, + "learning_rate": 4.138579934838782e-05, + "loss": 0.0319, + "step": 20180 + }, + { + "epoch": 5.01, + "learning_rate": 4.137877766543085e-05, + "loss": 0.0003, + "step": 20190 + }, + { + "epoch": 5.01, + "learning_rate": 4.137175598247388e-05, + "loss": 0.5792, + "step": 20200 + }, + { + "epoch": 5.01, + "learning_rate": 4.136473429951691e-05, + "loss": 0.0542, + "step": 20210 + }, + { + "epoch": 5.01, + "learning_rate": 4.135771261655994e-05, + "loss": 0.2656, + "step": 20220 + }, + { + "epoch": 5.01, + "learning_rate": 4.135069093360297e-05, + "loss": 0.2506, + "step": 20230 + }, + { + "epoch": 5.01, + "learning_rate": 4.1343669250646e-05, + "loss": 0.0088, + "step": 20240 + }, + { + "epoch": 5.01, + "learning_rate": 4.1336647567689025e-05, + "loss": 0.2305, + "step": 20250 + }, + { + "epoch": 5.01, + "learning_rate": 4.132962588473205e-05, + "loss": 0.3892, + "step": 20260 + }, + { + "epoch": 5.01, + "learning_rate": 4.132260420177509e-05, + "loss": 0.4709, + "step": 20270 + }, + { + "epoch": 5.01, + "learning_rate": 4.1315582518818114e-05, + "loss": 0.4627, + "step": 20280 + }, + { + "epoch": 5.01, + "learning_rate": 4.130856083586114e-05, + "loss": 0.2746, + "step": 20290 + }, + { + "epoch": 5.01, + "learning_rate": 4.130153915290417e-05, + "loss": 0.0454, + "step": 20300 + }, + { + "epoch": 5.01, + "learning_rate": 4.12945174699472e-05, + "loss": 0.0057, + "step": 20310 + }, + { + "epoch": 5.01, + "learning_rate": 4.1287495786990224e-05, + "loss": 0.3903, + "step": 20320 + }, + { + "epoch": 5.01, + "learning_rate": 4.128047410403326e-05, + "loss": 0.2519, + "step": 20330 + }, + { + "epoch": 5.01, + "learning_rate": 4.1273452421076286e-05, + "loss": 0.1728, + "step": 20340 + }, + { + "epoch": 5.01, + "learning_rate": 4.1266430738119314e-05, + "loss": 0.017, + "step": 20350 + }, + { + "epoch": 5.01, + "learning_rate": 4.125940905516235e-05, + "loss": 0.3841, + "step": 20360 + }, + { + "epoch": 5.01, + "learning_rate": 4.125238737220537e-05, + "loss": 0.3162, + "step": 20370 + }, + { + "epoch": 5.01, + "learning_rate": 4.12453656892484e-05, + "loss": 0.0437, + "step": 20380 + }, + { + "epoch": 5.01, + "learning_rate": 4.123834400629143e-05, + "loss": 0.2675, + "step": 20390 + }, + { + "epoch": 5.01, + "learning_rate": 4.123132232333446e-05, + "loss": 0.1909, + "step": 20400 + }, + { + "epoch": 5.01, + "learning_rate": 4.122430064037749e-05, + "loss": 0.0472, + "step": 20410 + }, + { + "epoch": 5.01, + "learning_rate": 4.121727895742052e-05, + "loss": 0.1, + "step": 20420 + }, + { + "epoch": 5.01, + "learning_rate": 4.121025727446354e-05, + "loss": 0.2975, + "step": 20430 + }, + { + "epoch": 5.01, + "learning_rate": 4.1203235591506575e-05, + "loss": 0.3657, + "step": 20440 + }, + { + "epoch": 5.01, + "learning_rate": 4.11962139085496e-05, + "loss": 0.1538, + "step": 20450 + }, + { + "epoch": 5.01, + "learning_rate": 4.118919222559263e-05, + "loss": 0.0315, + "step": 20460 + }, + { + "epoch": 5.01, + "learning_rate": 4.1182170542635664e-05, + "loss": 0.3873, + "step": 20470 + }, + { + "epoch": 5.01, + "learning_rate": 4.117514885967869e-05, + "loss": 0.0769, + "step": 20480 + }, + { + "epoch": 5.01, + "learning_rate": 4.116812717672172e-05, + "loss": 0.0617, + "step": 20490 + }, + { + "epoch": 5.01, + "learning_rate": 4.1161105493764746e-05, + "loss": 0.2738, + "step": 20500 + }, + { + "epoch": 5.01, + "learning_rate": 4.1154083810807774e-05, + "loss": 0.0774, + "step": 20510 + }, + { + "epoch": 5.01, + "learning_rate": 4.114706212785081e-05, + "loss": 0.2776, + "step": 20520 + }, + { + "epoch": 5.01, + "learning_rate": 4.1140040444893835e-05, + "loss": 0.0241, + "step": 20530 + }, + { + "epoch": 5.01, + "learning_rate": 4.113301876193686e-05, + "loss": 0.3237, + "step": 20540 + }, + { + "epoch": 5.01, + "learning_rate": 4.11259970789799e-05, + "loss": 0.0003, + "step": 20550 + }, + { + "epoch": 5.01, + "learning_rate": 4.111897539602292e-05, + "loss": 0.0022, + "step": 20560 + }, + { + "epoch": 5.01, + "learning_rate": 4.1111953713065945e-05, + "loss": 0.1374, + "step": 20570 + }, + { + "epoch": 5.01, + "learning_rate": 4.110493203010898e-05, + "loss": 0.4644, + "step": 20580 + }, + { + "epoch": 5.01, + "learning_rate": 4.109791034715201e-05, + "loss": 0.0003, + "step": 20590 + }, + { + "epoch": 5.01, + "learning_rate": 4.1090888664195035e-05, + "loss": 0.0091, + "step": 20600 + }, + { + "epoch": 5.01, + "learning_rate": 4.108386698123807e-05, + "loss": 0.2079, + "step": 20610 + }, + { + "epoch": 5.01, + "learning_rate": 4.107684529828109e-05, + "loss": 0.0767, + "step": 20620 + }, + { + "epoch": 5.01, + "learning_rate": 4.1069823615324124e-05, + "loss": 0.0122, + "step": 20630 + }, + { + "epoch": 5.01, + "learning_rate": 4.106280193236715e-05, + "loss": 0.4263, + "step": 20640 + }, + { + "epoch": 5.01, + "learning_rate": 4.105578024941018e-05, + "loss": 0.0022, + "step": 20650 + }, + { + "epoch": 5.01, + "learning_rate": 4.104875856645321e-05, + "loss": 0.5443, + "step": 20660 + }, + { + "epoch": 5.01, + "learning_rate": 4.104173688349624e-05, + "loss": 0.353, + "step": 20670 + }, + { + "epoch": 5.01, + "learning_rate": 4.103471520053927e-05, + "loss": 0.1342, + "step": 20680 + }, + { + "epoch": 5.01, + "learning_rate": 4.1027693517582296e-05, + "loss": 0.0044, + "step": 20690 + }, + { + "epoch": 5.01, + "learning_rate": 4.102067183462532e-05, + "loss": 0.3486, + "step": 20700 + }, + { + "epoch": 5.01, + "learning_rate": 4.101365015166835e-05, + "loss": 0.2352, + "step": 20710 + }, + { + "epoch": 5.01, + "learning_rate": 4.1006628468711385e-05, + "loss": 0.1087, + "step": 20720 + }, + { + "epoch": 5.01, + "learning_rate": 4.099960678575441e-05, + "loss": 0.1015, + "step": 20730 + }, + { + "epoch": 5.01, + "learning_rate": 4.099258510279744e-05, + "loss": 0.4749, + "step": 20740 + }, + { + "epoch": 5.01, + "learning_rate": 4.098556341984047e-05, + "loss": 0.4295, + "step": 20750 + }, + { + "epoch": 5.01, + "learning_rate": 4.0978541736883495e-05, + "loss": 0.1008, + "step": 20760 + }, + { + "epoch": 5.01, + "learning_rate": 4.097152005392653e-05, + "loss": 0.1517, + "step": 20770 + }, + { + "epoch": 5.01, + "learning_rate": 4.096449837096956e-05, + "loss": 0.4754, + "step": 20780 + }, + { + "epoch": 5.01, + "learning_rate": 4.0957476688012584e-05, + "loss": 0.0042, + "step": 20790 + }, + { + "epoch": 5.01, + "learning_rate": 4.095045500505562e-05, + "loss": 0.0639, + "step": 20800 + }, + { + "epoch": 5.01, + "learning_rate": 4.094343332209864e-05, + "loss": 0.4922, + "step": 20810 + }, + { + "epoch": 5.01, + "learning_rate": 4.0936411639141667e-05, + "loss": 0.3571, + "step": 20820 + }, + { + "epoch": 5.01, + "learning_rate": 4.09293899561847e-05, + "loss": 0.669, + "step": 20830 + }, + { + "epoch": 5.01, + "learning_rate": 4.092236827322773e-05, + "loss": 0.3443, + "step": 20840 + }, + { + "epoch": 5.01, + "learning_rate": 4.0915346590270756e-05, + "loss": 0.0156, + "step": 20850 + }, + { + "epoch": 5.01, + "learning_rate": 4.090832490731379e-05, + "loss": 0.4376, + "step": 20860 + }, + { + "epoch": 5.01, + "learning_rate": 4.090130322435682e-05, + "loss": 0.423, + "step": 20870 + }, + { + "epoch": 5.01, + "learning_rate": 4.0894281541399845e-05, + "loss": 1.0439, + "step": 20880 + }, + { + "epoch": 5.01, + "learning_rate": 4.088725985844287e-05, + "loss": 1.3657, + "step": 20890 + }, + { + "epoch": 5.01, + "learning_rate": 4.08802381754859e-05, + "loss": 0.2527, + "step": 20900 + }, + { + "epoch": 5.01, + "learning_rate": 4.0873216492528934e-05, + "loss": 0.2635, + "step": 20910 + }, + { + "epoch": 5.01, + "learning_rate": 4.086619480957196e-05, + "loss": 0.9315, + "step": 20920 + }, + { + "epoch": 5.01, + "learning_rate": 4.085917312661499e-05, + "loss": 0.2953, + "step": 20930 + }, + { + "epoch": 5.01, + "learning_rate": 4.085215144365802e-05, + "loss": 0.0406, + "step": 20940 + }, + { + "epoch": 5.01, + "learning_rate": 4.0845129760701044e-05, + "loss": 0.0524, + "step": 20950 + }, + { + "epoch": 5.01, + "learning_rate": 4.083810807774407e-05, + "loss": 0.0587, + "step": 20960 + }, + { + "epoch": 5.02, + "learning_rate": 4.0831086394787106e-05, + "loss": 0.2526, + "step": 20970 + }, + { + "epoch": 5.02, + "learning_rate": 4.0824064711830134e-05, + "loss": 0.0514, + "step": 20980 + }, + { + "epoch": 5.02, + "learning_rate": 4.081704302887316e-05, + "loss": 0.6689, + "step": 20990 + }, + { + "epoch": 5.02, + "learning_rate": 4.081002134591619e-05, + "loss": 0.5746, + "step": 21000 + }, + { + "epoch": 5.02, + "learning_rate": 4.0802999662959216e-05, + "loss": 0.335, + "step": 21010 + }, + { + "epoch": 5.02, + "learning_rate": 4.079597798000225e-05, + "loss": 0.0373, + "step": 21020 + }, + { + "epoch": 5.02, + "learning_rate": 4.078895629704528e-05, + "loss": 0.0012, + "step": 21030 + }, + { + "epoch": 5.02, + "learning_rate": 4.0781934614088305e-05, + "loss": 0.1193, + "step": 21040 + }, + { + "epoch": 5.02, + "learning_rate": 4.077491293113134e-05, + "loss": 0.2885, + "step": 21050 + }, + { + "epoch": 5.02, + "learning_rate": 4.076789124817437e-05, + "loss": 1.0699, + "step": 21060 + }, + { + "epoch": 5.02, + "learning_rate": 4.076086956521739e-05, + "loss": 0.4431, + "step": 21070 + }, + { + "epoch": 5.02, + "learning_rate": 4.075384788226042e-05, + "loss": 0.0019, + "step": 21080 + }, + { + "epoch": 5.02, + "learning_rate": 4.074682619930345e-05, + "loss": 0.1069, + "step": 21090 + }, + { + "epoch": 5.02, + "learning_rate": 4.073980451634648e-05, + "loss": 0.271, + "step": 21100 + }, + { + "epoch": 5.02, + "learning_rate": 4.073278283338951e-05, + "loss": 0.3448, + "step": 21110 + }, + { + "epoch": 5.02, + "learning_rate": 4.072576115043254e-05, + "loss": 0.0006, + "step": 21120 + }, + { + "epoch": 5.02, + "learning_rate": 4.0718739467475566e-05, + "loss": 0.2013, + "step": 21130 + }, + { + "epoch": 5.02, + "learning_rate": 4.0711717784518594e-05, + "loss": 0.6106, + "step": 21140 + }, + { + "epoch": 5.02, + "learning_rate": 4.070469610156162e-05, + "loss": 0.243, + "step": 21150 + }, + { + "epoch": 5.02, + "learning_rate": 4.0697674418604655e-05, + "loss": 0.0477, + "step": 21160 + }, + { + "epoch": 5.02, + "learning_rate": 4.069065273564768e-05, + "loss": 1.2592, + "step": 21170 + }, + { + "epoch": 5.02, + "learning_rate": 4.068363105269071e-05, + "loss": 0.181, + "step": 21180 + }, + { + "epoch": 5.02, + "learning_rate": 4.0676609369733745e-05, + "loss": 0.0054, + "step": 21190 + }, + { + "epoch": 5.02, + "learning_rate": 4.0669587686776765e-05, + "loss": 0.1149, + "step": 21200 + }, + { + "epoch": 5.02, + "learning_rate": 4.06625660038198e-05, + "loss": 0.4368, + "step": 21210 + }, + { + "epoch": 5.02, + "learning_rate": 4.065554432086283e-05, + "loss": 0.047, + "step": 21220 + }, + { + "epoch": 5.02, + "learning_rate": 4.0648522637905855e-05, + "loss": 0.0462, + "step": 21230 + }, + { + "epoch": 5.02, + "learning_rate": 4.064150095494889e-05, + "loss": 0.2406, + "step": 21240 + }, + { + "epoch": 5.02, + "learning_rate": 4.0634479271991916e-05, + "loss": 0.1731, + "step": 21250 + }, + { + "epoch": 5.02, + "learning_rate": 4.062745758903494e-05, + "loss": 0.063, + "step": 21260 + }, + { + "epoch": 5.02, + "learning_rate": 4.062043590607797e-05, + "loss": 0.4046, + "step": 21270 + }, + { + "epoch": 5.02, + "learning_rate": 4.0613414223121e-05, + "loss": 0.5177, + "step": 21280 + }, + { + "epoch": 5.02, + "learning_rate": 4.0606392540164026e-05, + "loss": 0.1799, + "step": 21290 + }, + { + "epoch": 5.02, + "learning_rate": 4.059937085720706e-05, + "loss": 1.5399, + "step": 21300 + }, + { + "epoch": 5.02, + "learning_rate": 4.059234917425009e-05, + "loss": 0.0009, + "step": 21310 + }, + { + "epoch": 5.02, + "learning_rate": 4.0585327491293116e-05, + "loss": 0.6043, + "step": 21320 + }, + { + "epoch": 5.02, + "learning_rate": 4.057830580833614e-05, + "loss": 0.607, + "step": 21330 + }, + { + "epoch": 5.02, + "learning_rate": 4.057128412537917e-05, + "loss": 0.2742, + "step": 21340 + }, + { + "epoch": 5.02, + "learning_rate": 4.0564262442422205e-05, + "loss": 0.017, + "step": 21350 + }, + { + "epoch": 5.02, + "learning_rate": 4.055724075946523e-05, + "loss": 0.4012, + "step": 21360 + }, + { + "epoch": 5.02, + "learning_rate": 4.055021907650826e-05, + "loss": 0.4771, + "step": 21370 + }, + { + "epoch": 5.02, + "learning_rate": 4.0543197393551294e-05, + "loss": 0.1536, + "step": 21380 + }, + { + "epoch": 5.02, + "learning_rate": 4.0536175710594315e-05, + "loss": 0.0012, + "step": 21390 + }, + { + "epoch": 5.02, + "learning_rate": 4.052915402763734e-05, + "loss": 0.4098, + "step": 21400 + }, + { + "epoch": 5.02, + "learning_rate": 4.052213234468038e-05, + "loss": 0.0021, + "step": 21410 + }, + { + "epoch": 5.02, + "learning_rate": 4.0515110661723404e-05, + "loss": 0.0691, + "step": 21420 + }, + { + "epoch": 5.02, + "learning_rate": 4.050808897876643e-05, + "loss": 0.0151, + "step": 21430 + }, + { + "epoch": 5.02, + "learning_rate": 4.0501067295809466e-05, + "loss": 0.1553, + "step": 21440 + }, + { + "epoch": 5.02, + "learning_rate": 4.0494045612852487e-05, + "loss": 0.0981, + "step": 21450 + }, + { + "epoch": 5.02, + "learning_rate": 4.048702392989552e-05, + "loss": 0.5916, + "step": 21460 + }, + { + "epoch": 5.02, + "learning_rate": 4.048000224693855e-05, + "loss": 0.003, + "step": 21470 + }, + { + "epoch": 5.02, + "learning_rate": 4.0472980563981576e-05, + "loss": 0.043, + "step": 21480 + }, + { + "epoch": 5.02, + "learning_rate": 4.046595888102461e-05, + "loss": 0.0369, + "step": 21490 + }, + { + "epoch": 5.02, + "learning_rate": 4.045893719806764e-05, + "loss": 0.0218, + "step": 21500 + }, + { + "epoch": 5.02, + "learning_rate": 4.045191551511066e-05, + "loss": 0.7087, + "step": 21510 + }, + { + "epoch": 5.02, + "learning_rate": 4.044489383215369e-05, + "loss": 0.1049, + "step": 21520 + }, + { + "epoch": 5.02, + "learning_rate": 4.043787214919672e-05, + "loss": 0.1221, + "step": 21530 + }, + { + "epoch": 5.02, + "learning_rate": 4.043085046623975e-05, + "loss": 0.1104, + "step": 21540 + }, + { + "epoch": 5.02, + "learning_rate": 4.042382878328278e-05, + "loss": 0.2662, + "step": 21550 + }, + { + "epoch": 5.02, + "learning_rate": 4.041680710032581e-05, + "loss": 0.0057, + "step": 21560 + }, + { + "epoch": 5.02, + "learning_rate": 4.040978541736884e-05, + "loss": 0.0005, + "step": 21570 + }, + { + "epoch": 5.02, + "learning_rate": 4.0402763734411864e-05, + "loss": 0.3357, + "step": 21580 + }, + { + "epoch": 5.02, + "learning_rate": 4.039574205145489e-05, + "loss": 0.6727, + "step": 21590 + }, + { + "epoch": 5.02, + "learning_rate": 4.0388720368497926e-05, + "loss": 0.6819, + "step": 21600 + }, + { + "epoch": 5.02, + "learning_rate": 4.0381698685540954e-05, + "loss": 0.1953, + "step": 21610 + }, + { + "epoch": 5.02, + "learning_rate": 4.037467700258398e-05, + "loss": 0.0731, + "step": 21620 + }, + { + "epoch": 5.02, + "learning_rate": 4.0367655319627015e-05, + "loss": 0.3223, + "step": 21630 + }, + { + "epoch": 5.02, + "learning_rate": 4.0360633636670036e-05, + "loss": 0.4572, + "step": 21640 + }, + { + "epoch": 5.02, + "learning_rate": 4.0353611953713063e-05, + "loss": 0.4104, + "step": 21650 + }, + { + "epoch": 5.02, + "learning_rate": 4.03465902707561e-05, + "loss": 0.0509, + "step": 21660 + }, + { + "epoch": 5.02, + "learning_rate": 4.0339568587799125e-05, + "loss": 0.3232, + "step": 21670 + }, + { + "epoch": 5.02, + "learning_rate": 4.033254690484215e-05, + "loss": 0.0158, + "step": 21680 + }, + { + "epoch": 5.02, + "learning_rate": 4.032552522188519e-05, + "loss": 0.8785, + "step": 21690 + }, + { + "epoch": 5.02, + "learning_rate": 4.031850353892821e-05, + "loss": 0.5797, + "step": 21700 + }, + { + "epoch": 5.02, + "learning_rate": 4.031148185597124e-05, + "loss": 0.0141, + "step": 21710 + }, + { + "epoch": 5.02, + "learning_rate": 4.030446017301427e-05, + "loss": 0.5504, + "step": 21720 + }, + { + "epoch": 5.02, + "learning_rate": 4.02974384900573e-05, + "loss": 0.429, + "step": 21730 + }, + { + "epoch": 5.02, + "learning_rate": 4.029041680710033e-05, + "loss": 0.5101, + "step": 21740 + }, + { + "epoch": 5.02, + "learning_rate": 4.028339512414336e-05, + "loss": 0.8366, + "step": 21750 + }, + { + "epoch": 5.03, + "learning_rate": 4.0276373441186386e-05, + "loss": 0.0043, + "step": 21760 + }, + { + "epoch": 5.03, + "learning_rate": 4.0269351758229414e-05, + "loss": 0.4016, + "step": 21770 + }, + { + "epoch": 5.03, + "learning_rate": 4.026233007527244e-05, + "loss": 0.0164, + "step": 21780 + }, + { + "epoch": 5.03, + "learning_rate": 4.025530839231547e-05, + "loss": 0.294, + "step": 21790 + }, + { + "epoch": 5.03, + "learning_rate": 4.02482867093585e-05, + "loss": 0.1081, + "step": 21800 + }, + { + "epoch": 5.03, + "learning_rate": 4.024126502640153e-05, + "loss": 0.2577, + "step": 21810 + }, + { + "epoch": 5.03, + "learning_rate": 4.023424334344456e-05, + "loss": 0.1366, + "step": 21820 + }, + { + "epoch": 5.03, + "learning_rate": 4.0227221660487585e-05, + "loss": 0.0002, + "step": 21830 + }, + { + "epoch": 5.03, + "learning_rate": 4.022019997753061e-05, + "loss": 0.0094, + "step": 21840 + }, + { + "epoch": 5.03, + "learning_rate": 4.021317829457365e-05, + "loss": 0.9207, + "step": 21850 + }, + { + "epoch": 5.03, + "learning_rate": 4.0206156611616675e-05, + "loss": 0.8645, + "step": 21860 + }, + { + "epoch": 5.03, + "learning_rate": 4.01991349286597e-05, + "loss": 0.4848, + "step": 21870 + }, + { + "epoch": 5.03, + "learning_rate": 4.0192113245702736e-05, + "loss": 0.3265, + "step": 21880 + }, + { + "epoch": 5.03, + "learning_rate": 4.0185091562745764e-05, + "loss": 0.5164, + "step": 21890 + }, + { + "epoch": 5.03, + "learning_rate": 4.0178069879788785e-05, + "loss": 0.3176, + "step": 21900 + }, + { + "epoch": 5.03, + "learning_rate": 4.017104819683182e-05, + "loss": 0.8989, + "step": 21910 + }, + { + "epoch": 5.03, + "learning_rate": 4.0164026513874846e-05, + "loss": 0.2551, + "step": 21920 + }, + { + "epoch": 5.03, + "learning_rate": 4.0157004830917874e-05, + "loss": 0.9977, + "step": 21930 + }, + { + "epoch": 5.03, + "learning_rate": 4.014998314796091e-05, + "loss": 0.1575, + "step": 21940 + }, + { + "epoch": 5.03, + "learning_rate": 4.0142961465003936e-05, + "loss": 0.8459, + "step": 21950 + }, + { + "epoch": 5.03, + "learning_rate": 4.013593978204696e-05, + "loss": 0.1274, + "step": 21960 + }, + { + "epoch": 5.03, + "learning_rate": 4.012891809908999e-05, + "loss": 0.6365, + "step": 21970 + }, + { + "epoch": 5.03, + "learning_rate": 4.012189641613302e-05, + "loss": 0.2802, + "step": 21980 + }, + { + "epoch": 5.03, + "learning_rate": 4.011487473317605e-05, + "loss": 0.5251, + "step": 21990 + }, + { + "epoch": 5.03, + "learning_rate": 4.010785305021908e-05, + "loss": 0.0827, + "step": 22000 + }, + { + "epoch": 5.03, + "learning_rate": 4.010083136726211e-05, + "loss": 0.0149, + "step": 22010 + }, + { + "epoch": 5.03, + "learning_rate": 4.0093809684305135e-05, + "loss": 0.2067, + "step": 22020 + }, + { + "epoch": 5.03, + "learning_rate": 4.008678800134816e-05, + "loss": 0.5906, + "step": 22030 + }, + { + "epoch": 5.03, + "learning_rate": 4.007976631839119e-05, + "loss": 0.6435, + "step": 22040 + }, + { + "epoch": 5.03, + "learning_rate": 4.0072744635434224e-05, + "loss": 0.8878, + "step": 22050 + }, + { + "epoch": 5.03, + "learning_rate": 4.006572295247725e-05, + "loss": 0.7509, + "step": 22060 + }, + { + "epoch": 5.03, + "learning_rate": 4.005870126952028e-05, + "loss": 0.6226, + "step": 22070 + }, + { + "epoch": 5.03, + "learning_rate": 4.005167958656331e-05, + "loss": 0.3121, + "step": 22080 + }, + { + "epoch": 5.03, + "learning_rate": 4.0044657903606334e-05, + "loss": 0.077, + "step": 22090 + }, + { + "epoch": 5.03, + "learning_rate": 4.003763622064937e-05, + "loss": 0.1639, + "step": 22100 + }, + { + "epoch": 5.03, + "learning_rate": 4.0030614537692396e-05, + "loss": 0.729, + "step": 22110 + }, + { + "epoch": 5.03, + "learning_rate": 4.002359285473542e-05, + "loss": 0.1176, + "step": 22120 + }, + { + "epoch": 5.03, + "learning_rate": 4.001657117177846e-05, + "loss": 0.6888, + "step": 22130 + }, + { + "epoch": 5.03, + "learning_rate": 4.0009549488821485e-05, + "loss": 0.0209, + "step": 22140 + }, + { + "epoch": 5.03, + "learning_rate": 4.000252780586451e-05, + "loss": 0.7488, + "step": 22150 + }, + { + "epoch": 5.03, + "learning_rate": 3.999550612290754e-05, + "loss": 0.2688, + "step": 22160 + }, + { + "epoch": 5.03, + "learning_rate": 3.998848443995057e-05, + "loss": 0.5351, + "step": 22170 + }, + { + "epoch": 5.03, + "learning_rate": 3.99814627569936e-05, + "loss": 0.6054, + "step": 22180 + }, + { + "epoch": 5.03, + "learning_rate": 3.997444107403663e-05, + "loss": 0.0066, + "step": 22190 + }, + { + "epoch": 5.03, + "learning_rate": 3.996741939107966e-05, + "loss": 0.8341, + "step": 22200 + }, + { + "epoch": 5.03, + "learning_rate": 3.9960397708122684e-05, + "loss": 0.2035, + "step": 22210 + }, + { + "epoch": 5.03, + "learning_rate": 3.995337602516571e-05, + "loss": 0.5068, + "step": 22220 + }, + { + "epoch": 5.03, + "learning_rate": 3.994635434220874e-05, + "loss": 0.5424, + "step": 22230 + }, + { + "epoch": 5.03, + "learning_rate": 3.9939332659251774e-05, + "loss": 0.1384, + "step": 22240 + }, + { + "epoch": 5.03, + "learning_rate": 3.99323109762948e-05, + "loss": 0.0446, + "step": 22250 + }, + { + "epoch": 5.03, + "learning_rate": 3.992528929333783e-05, + "loss": 0.0167, + "step": 22260 + }, + { + "epoch": 5.03, + "learning_rate": 3.991826761038086e-05, + "loss": 0.4953, + "step": 22270 + }, + { + "epoch": 5.03, + "learning_rate": 3.9911245927423883e-05, + "loss": 0.4652, + "step": 22280 + }, + { + "epoch": 5.03, + "learning_rate": 3.990422424446692e-05, + "loss": 0.0105, + "step": 22290 + }, + { + "epoch": 5.03, + "learning_rate": 3.9897202561509945e-05, + "loss": 0.1332, + "step": 22300 + }, + { + "epoch": 5.03, + "learning_rate": 3.989018087855297e-05, + "loss": 0.2316, + "step": 22310 + }, + { + "epoch": 5.03, + "learning_rate": 3.988315919559601e-05, + "loss": 0.0865, + "step": 22320 + }, + { + "epoch": 5.03, + "learning_rate": 3.9876137512639034e-05, + "loss": 0.0068, + "step": 22330 + }, + { + "epoch": 5.03, + "learning_rate": 3.9869115829682055e-05, + "loss": 0.2753, + "step": 22340 + }, + { + "epoch": 5.03, + "learning_rate": 3.986209414672509e-05, + "loss": 0.3534, + "step": 22350 + }, + { + "epoch": 5.03, + "learning_rate": 3.985507246376812e-05, + "loss": 0.002, + "step": 22360 + }, + { + "epoch": 5.03, + "learning_rate": 3.9848050780811144e-05, + "loss": 0.2003, + "step": 22370 + }, + { + "epoch": 5.03, + "learning_rate": 3.984102909785418e-05, + "loss": 0.4544, + "step": 22380 + }, + { + "epoch": 5.03, + "learning_rate": 3.9834007414897206e-05, + "loss": 0.3123, + "step": 22390 + }, + { + "epoch": 5.03, + "learning_rate": 3.9826985731940234e-05, + "loss": 0.2909, + "step": 22400 + }, + { + "epoch": 5.03, + "learning_rate": 3.981996404898326e-05, + "loss": 0.003, + "step": 22410 + }, + { + "epoch": 5.03, + "learning_rate": 3.981294236602629e-05, + "loss": 0.0545, + "step": 22420 + }, + { + "epoch": 5.03, + "learning_rate": 3.980592068306932e-05, + "loss": 0.2582, + "step": 22430 + }, + { + "epoch": 5.03, + "learning_rate": 3.979889900011235e-05, + "loss": 0.3318, + "step": 22440 + }, + { + "epoch": 5.03, + "learning_rate": 3.979187731715538e-05, + "loss": 0.3301, + "step": 22450 + }, + { + "epoch": 5.03, + "learning_rate": 3.978485563419841e-05, + "loss": 0.0074, + "step": 22460 + }, + { + "epoch": 5.03, + "learning_rate": 3.977783395124143e-05, + "loss": 0.2764, + "step": 22470 + }, + { + "epoch": 5.03, + "learning_rate": 3.977081226828446e-05, + "loss": 0.3036, + "step": 22480 + }, + { + "epoch": 5.03, + "learning_rate": 3.9763790585327495e-05, + "loss": 0.7488, + "step": 22490 + }, + { + "epoch": 5.03, + "learning_rate": 3.975676890237052e-05, + "loss": 0.5424, + "step": 22500 + }, + { + "epoch": 5.03, + "learning_rate": 3.974974721941355e-05, + "loss": 0.5065, + "step": 22510 + }, + { + "epoch": 5.03, + "learning_rate": 3.9742725536456584e-05, + "loss": 0.5173, + "step": 22520 + }, + { + "epoch": 5.03, + "learning_rate": 3.9735703853499605e-05, + "loss": 0.845, + "step": 22530 + }, + { + "epoch": 5.03, + "learning_rate": 3.972868217054264e-05, + "loss": 0.0889, + "step": 22540 + }, + { + "epoch": 5.04, + "learning_rate": 3.9721660487585666e-05, + "loss": 0.5275, + "step": 22550 + }, + { + "epoch": 5.04, + "learning_rate": 3.9714638804628694e-05, + "loss": 0.4327, + "step": 22560 + }, + { + "epoch": 5.04, + "learning_rate": 3.970761712167173e-05, + "loss": 0.1533, + "step": 22570 + }, + { + "epoch": 5.04, + "learning_rate": 3.9700595438714756e-05, + "loss": 0.1701, + "step": 22580 + }, + { + "epoch": 5.04, + "learning_rate": 3.969357375575778e-05, + "loss": 0.0562, + "step": 22590 + }, + { + "epoch": 5.04, + "learning_rate": 3.968655207280081e-05, + "loss": 0.3313, + "step": 22600 + }, + { + "epoch": 5.04, + "learning_rate": 3.967953038984384e-05, + "loss": 0.0009, + "step": 22610 + }, + { + "epoch": 5.04, + "learning_rate": 3.9672508706886866e-05, + "loss": 0.6619, + "step": 22620 + }, + { + "epoch": 5.04, + "learning_rate": 3.96654870239299e-05, + "loss": 0.357, + "step": 22630 + }, + { + "epoch": 5.04, + "learning_rate": 3.965846534097293e-05, + "loss": 0.7874, + "step": 22640 + }, + { + "epoch": 5.04, + "learning_rate": 3.9651443658015955e-05, + "loss": 0.0085, + "step": 22650 + }, + { + "epoch": 5.04, + "learning_rate": 3.964442197505898e-05, + "loss": 0.3643, + "step": 22660 + }, + { + "epoch": 5.04, + "learning_rate": 3.963740029210201e-05, + "loss": 0.6606, + "step": 22670 + }, + { + "epoch": 5.04, + "learning_rate": 3.9630378609145044e-05, + "loss": 0.128, + "step": 22680 + }, + { + "epoch": 5.04, + "learning_rate": 3.962335692618807e-05, + "loss": 0.357, + "step": 22690 + }, + { + "epoch": 5.04, + "learning_rate": 3.96163352432311e-05, + "loss": 0.582, + "step": 22700 + }, + { + "epoch": 5.04, + "learning_rate": 3.960931356027413e-05, + "loss": 0.6935, + "step": 22710 + }, + { + "epoch": 5.04, + "learning_rate": 3.9602291877317154e-05, + "loss": 0.2393, + "step": 22720 + }, + { + "epoch": 5.04, + "learning_rate": 3.959527019436018e-05, + "loss": 0.3193, + "step": 22730 + }, + { + "epoch": 5.04, + "learning_rate": 3.9588248511403216e-05, + "loss": 0.4765, + "step": 22740 + }, + { + "epoch": 5.04, + "learning_rate": 3.958122682844624e-05, + "loss": 1.0248, + "step": 22750 + }, + { + "epoch": 5.04, + "learning_rate": 3.957420514548927e-05, + "loss": 0.1894, + "step": 22760 + }, + { + "epoch": 5.04, + "learning_rate": 3.9567183462532305e-05, + "loss": 0.1251, + "step": 22770 + }, + { + "epoch": 5.04, + "learning_rate": 3.956016177957533e-05, + "loss": 0.0772, + "step": 22780 + }, + { + "epoch": 5.04, + "learning_rate": 3.955314009661836e-05, + "loss": 0.2637, + "step": 22790 + }, + { + "epoch": 5.04, + "learning_rate": 3.954611841366139e-05, + "loss": 0.4514, + "step": 22800 + }, + { + "epoch": 5.04, + "learning_rate": 3.9539096730704415e-05, + "loss": 0.1116, + "step": 22810 + }, + { + "epoch": 5.04, + "learning_rate": 3.953207504774745e-05, + "loss": 0.2629, + "step": 22820 + }, + { + "epoch": 5.04, + "learning_rate": 3.952505336479048e-05, + "loss": 0.164, + "step": 22830 + }, + { + "epoch": 5.04, + "learning_rate": 3.9518031681833504e-05, + "loss": 0.1629, + "step": 22840 + }, + { + "epoch": 5.04, + "learning_rate": 3.951100999887653e-05, + "loss": 0.1923, + "step": 22850 + }, + { + "epoch": 5.04, + "learning_rate": 3.950398831591956e-05, + "loss": 0.0282, + "step": 22860 + }, + { + "epoch": 5.04, + "learning_rate": 3.949696663296259e-05, + "loss": 0.7025, + "step": 22870 + }, + { + "epoch": 5.04, + "learning_rate": 3.948994495000562e-05, + "loss": 0.0356, + "step": 22880 + }, + { + "epoch": 5.04, + "learning_rate": 3.948292326704865e-05, + "loss": 0.2152, + "step": 22890 + }, + { + "epoch": 5.04, + "learning_rate": 3.9475901584091676e-05, + "loss": 0.3462, + "step": 22900 + }, + { + "epoch": 5.04, + "learning_rate": 3.9468879901134703e-05, + "loss": 0.0067, + "step": 22910 + }, + { + "epoch": 5.04, + "learning_rate": 3.946185821817773e-05, + "loss": 0.3377, + "step": 22920 + }, + { + "epoch": 5.04, + "learning_rate": 3.9454836535220765e-05, + "loss": 0.5784, + "step": 22930 + }, + { + "epoch": 5.04, + "learning_rate": 3.944781485226379e-05, + "loss": 0.3562, + "step": 22940 + }, + { + "epoch": 5.04, + "learning_rate": 3.944079316930682e-05, + "loss": 0.0044, + "step": 22950 + }, + { + "epoch": 5.04, + "learning_rate": 3.9433771486349854e-05, + "loss": 0.2235, + "step": 22960 + }, + { + "epoch": 5.04, + "learning_rate": 3.942674980339288e-05, + "loss": 0.4581, + "step": 22970 + }, + { + "epoch": 5.04, + "learning_rate": 3.94197281204359e-05, + "loss": 0.354, + "step": 22980 + }, + { + "epoch": 5.04, + "learning_rate": 3.941270643747894e-05, + "loss": 0.0714, + "step": 22990 + }, + { + "epoch": 5.04, + "learning_rate": 3.9405684754521964e-05, + "loss": 0.6007, + "step": 23000 + }, + { + "epoch": 5.04, + "learning_rate": 3.939866307156499e-05, + "loss": 0.2147, + "step": 23010 + }, + { + "epoch": 5.04, + "learning_rate": 3.9391641388608026e-05, + "loss": 0.3189, + "step": 23020 + }, + { + "epoch": 5.04, + "learning_rate": 3.9384619705651054e-05, + "loss": 0.002, + "step": 23030 + }, + { + "epoch": 5.04, + "learning_rate": 3.937759802269408e-05, + "loss": 0.3123, + "step": 23040 + }, + { + "epoch": 5.04, + "learning_rate": 3.937057633973711e-05, + "loss": 0.0655, + "step": 23050 + }, + { + "epoch": 5.04, + "learning_rate": 3.9363554656780136e-05, + "loss": 0.1952, + "step": 23060 + }, + { + "epoch": 5.04, + "learning_rate": 3.935653297382317e-05, + "loss": 0.1146, + "step": 23070 + }, + { + "epoch": 5.04, + "learning_rate": 3.93495112908662e-05, + "loss": 0.3841, + "step": 23080 + }, + { + "epoch": 5.04, + "learning_rate": 3.9342489607909225e-05, + "loss": 0.0568, + "step": 23090 + }, + { + "epoch": 5.04, + "learning_rate": 3.933546792495225e-05, + "loss": 0.0676, + "step": 23100 + }, + { + "epoch": 5.04, + "learning_rate": 3.932844624199528e-05, + "loss": 1.2651, + "step": 23110 + }, + { + "epoch": 5.04, + "learning_rate": 3.932142455903831e-05, + "loss": 1.0039, + "step": 23120 + }, + { + "epoch": 5.04, + "learning_rate": 3.931440287608134e-05, + "loss": 0.0042, + "step": 23130 + }, + { + "epoch": 5.04, + "learning_rate": 3.930738119312437e-05, + "loss": 0.2997, + "step": 23140 + }, + { + "epoch": 5.04, + "learning_rate": 3.93003595101674e-05, + "loss": 0.6777, + "step": 23150 + }, + { + "epoch": 5.04, + "learning_rate": 3.929333782721043e-05, + "loss": 0.4172, + "step": 23160 + }, + { + "epoch": 5.04, + "learning_rate": 3.928631614425345e-05, + "loss": 0.0899, + "step": 23170 + }, + { + "epoch": 5.04, + "learning_rate": 3.9279294461296486e-05, + "loss": 0.0241, + "step": 23180 + }, + { + "epoch": 5.04, + "learning_rate": 3.9272272778339514e-05, + "loss": 0.4873, + "step": 23190 + }, + { + "epoch": 5.04, + "learning_rate": 3.926525109538254e-05, + "loss": 0.367, + "step": 23200 + }, + { + "epoch": 5.04, + "learning_rate": 3.9258229412425576e-05, + "loss": 0.0226, + "step": 23210 + }, + { + "epoch": 5.04, + "learning_rate": 3.92512077294686e-05, + "loss": 0.2475, + "step": 23220 + }, + { + "epoch": 5.04, + "learning_rate": 3.924418604651163e-05, + "loss": 0.76, + "step": 23230 + }, + { + "epoch": 5.04, + "learning_rate": 3.923716436355466e-05, + "loss": 0.116, + "step": 23240 + }, + { + "epoch": 5.04, + "learning_rate": 3.9230142680597686e-05, + "loss": 0.4541, + "step": 23250 + }, + { + "epoch": 5.04, + "learning_rate": 3.922312099764072e-05, + "loss": 0.0112, + "step": 23260 + }, + { + "epoch": 5.04, + "learning_rate": 3.921609931468375e-05, + "loss": 0.1989, + "step": 23270 + }, + { + "epoch": 5.04, + "learning_rate": 3.9209077631726775e-05, + "loss": 0.6206, + "step": 23280 + }, + { + "epoch": 5.04, + "learning_rate": 3.92020559487698e-05, + "loss": 0.2571, + "step": 23290 + }, + { + "epoch": 5.04, + "learning_rate": 3.919503426581283e-05, + "loss": 0.5866, + "step": 23300 + }, + { + "epoch": 5.04, + "learning_rate": 3.918801258285586e-05, + "loss": 0.8862, + "step": 23310 + }, + { + "epoch": 5.04, + "learning_rate": 3.918099089989889e-05, + "loss": 0.3894, + "step": 23320 + }, + { + "epoch": 5.04, + "learning_rate": 3.917396921694192e-05, + "loss": 0.7141, + "step": 23330 + }, + { + "epoch": 5.04, + "learning_rate": 3.9166947533984947e-05, + "loss": 0.474, + "step": 23340 + }, + { + "epoch": 5.05, + "learning_rate": 3.915992585102798e-05, + "loss": 0.0369, + "step": 23350 + }, + { + "epoch": 5.05, + "learning_rate": 3.9152904168071e-05, + "loss": 0.0003, + "step": 23360 + }, + { + "epoch": 5.05, + "learning_rate": 3.9145882485114036e-05, + "loss": 0.0006, + "step": 23370 + }, + { + "epoch": 5.05, + "learning_rate": 3.913886080215706e-05, + "loss": 0.0197, + "step": 23380 + }, + { + "epoch": 5.05, + "learning_rate": 3.913183911920009e-05, + "loss": 0.0059, + "step": 23390 + }, + { + "epoch": 5.05, + "learning_rate": 3.9124817436243125e-05, + "loss": 0.4995, + "step": 23400 + }, + { + "epoch": 5.05, + "learning_rate": 3.911779575328615e-05, + "loss": 0.6545, + "step": 23410 + }, + { + "epoch": 5.05, + "learning_rate": 3.911077407032917e-05, + "loss": 0.1229, + "step": 23420 + }, + { + "epoch": 5.05, + "learning_rate": 3.910375238737221e-05, + "loss": 0.4182, + "step": 23430 + }, + { + "epoch": 5.05, + "learning_rate": 3.9096730704415235e-05, + "loss": 0.1852, + "step": 23440 + }, + { + "epoch": 5.05, + "learning_rate": 3.908970902145826e-05, + "loss": 0.3708, + "step": 23450 + }, + { + "epoch": 5.05, + "learning_rate": 3.90826873385013e-05, + "loss": 0.0594, + "step": 23460 + }, + { + "epoch": 5.05, + "learning_rate": 3.9075665655544324e-05, + "loss": 0.0483, + "step": 23470 + }, + { + "epoch": 5.05, + "learning_rate": 3.906864397258735e-05, + "loss": 0.2925, + "step": 23480 + }, + { + "epoch": 5.05, + "learning_rate": 3.906162228963038e-05, + "loss": 0.4294, + "step": 23490 + }, + { + "epoch": 5.05, + "learning_rate": 3.905460060667341e-05, + "loss": 0.0251, + "step": 23500 + }, + { + "epoch": 5.05, + "learning_rate": 3.904757892371644e-05, + "loss": 0.6397, + "step": 23510 + }, + { + "epoch": 5.05, + "learning_rate": 3.904055724075947e-05, + "loss": 0.1216, + "step": 23520 + }, + { + "epoch": 5.05, + "learning_rate": 3.9033535557802496e-05, + "loss": 0.0683, + "step": 23530 + }, + { + "epoch": 5.05, + "learning_rate": 3.902651387484553e-05, + "loss": 0.01, + "step": 23540 + }, + { + "epoch": 5.05, + "learning_rate": 3.901949219188855e-05, + "loss": 0.0192, + "step": 23550 + }, + { + "epoch": 5.05, + "learning_rate": 3.901247050893158e-05, + "loss": 0.0068, + "step": 23560 + }, + { + "epoch": 5.05, + "learning_rate": 3.900544882597461e-05, + "loss": 0.0448, + "step": 23570 + }, + { + "epoch": 5.05, + "learning_rate": 3.899842714301764e-05, + "loss": 0.4838, + "step": 23580 + }, + { + "epoch": 5.05, + "learning_rate": 3.899140546006067e-05, + "loss": 0.7024, + "step": 23590 + }, + { + "epoch": 5.05, + "learning_rate": 3.89843837771037e-05, + "loss": 0.0024, + "step": 23600 + }, + { + "epoch": 5.05, + "learning_rate": 3.897736209414672e-05, + "loss": 0.6181, + "step": 23610 + }, + { + "epoch": 5.05, + "learning_rate": 3.897034041118976e-05, + "loss": 0.5901, + "step": 23620 + }, + { + "epoch": 5.05, + "learning_rate": 3.8963318728232784e-05, + "loss": 0.0581, + "step": 23630 + }, + { + "epoch": 5.05, + "learning_rate": 3.895629704527581e-05, + "loss": 0.0018, + "step": 23640 + }, + { + "epoch": 5.05, + "learning_rate": 3.8949275362318846e-05, + "loss": 0.2831, + "step": 23650 + }, + { + "epoch": 5.05, + "learning_rate": 3.8942253679361874e-05, + "loss": 0.4832, + "step": 23660 + }, + { + "epoch": 5.05, + "learning_rate": 3.89352319964049e-05, + "loss": 0.9927, + "step": 23670 + }, + { + "epoch": 5.05, + "learning_rate": 3.892821031344793e-05, + "loss": 0.1838, + "step": 23680 + }, + { + "epoch": 5.05, + "learning_rate": 3.8921188630490956e-05, + "loss": 0.4305, + "step": 23690 + }, + { + "epoch": 5.05, + "learning_rate": 3.8914166947533984e-05, + "loss": 0.4126, + "step": 23700 + }, + { + "epoch": 5.05, + "learning_rate": 3.890714526457702e-05, + "loss": 0.1518, + "step": 23710 + }, + { + "epoch": 5.05, + "learning_rate": 3.8900123581620045e-05, + "loss": 0.597, + "step": 23720 + }, + { + "epoch": 5.05, + "learning_rate": 3.889310189866307e-05, + "loss": 0.0025, + "step": 23730 + }, + { + "epoch": 5.05, + "eval_accuracy": 0.7064894852277126, + "eval_loss": 1.573199987411499, + "eval_runtime": 1222.7474, + "eval_samples_per_second": 6.339, + "eval_steps_per_second": 3.17, + "step": 23736 + }, + { + "epoch": 6.0, + "learning_rate": 3.88860802157061e-05, + "loss": 0.0147, + "step": 23740 + }, + { + "epoch": 6.0, + "learning_rate": 3.887905853274913e-05, + "loss": 0.2025, + "step": 23750 + }, + { + "epoch": 6.0, + "learning_rate": 3.887203684979216e-05, + "loss": 0.8512, + "step": 23760 + }, + { + "epoch": 6.0, + "learning_rate": 3.886501516683519e-05, + "loss": 0.1094, + "step": 23770 + }, + { + "epoch": 6.0, + "learning_rate": 3.885799348387822e-05, + "loss": 0.8718, + "step": 23780 + }, + { + "epoch": 6.0, + "learning_rate": 3.885097180092125e-05, + "loss": 0.4455, + "step": 23790 + }, + { + "epoch": 6.0, + "learning_rate": 3.884395011796427e-05, + "loss": 0.1114, + "step": 23800 + }, + { + "epoch": 6.0, + "learning_rate": 3.88369284350073e-05, + "loss": 0.0002, + "step": 23810 + }, + { + "epoch": 6.0, + "learning_rate": 3.8829906752050334e-05, + "loss": 0.1276, + "step": 23820 + }, + { + "epoch": 6.0, + "learning_rate": 3.882288506909336e-05, + "loss": 0.011, + "step": 23830 + }, + { + "epoch": 6.0, + "learning_rate": 3.881586338613639e-05, + "loss": 0.2087, + "step": 23840 + }, + { + "epoch": 6.0, + "learning_rate": 3.880884170317942e-05, + "loss": 0.1767, + "step": 23850 + }, + { + "epoch": 6.0, + "learning_rate": 3.880182002022245e-05, + "loss": 0.034, + "step": 23860 + }, + { + "epoch": 6.0, + "learning_rate": 3.879479833726548e-05, + "loss": 0.3853, + "step": 23870 + }, + { + "epoch": 6.0, + "learning_rate": 3.8787776654308506e-05, + "loss": 0.0549, + "step": 23880 + }, + { + "epoch": 6.0, + "learning_rate": 3.878075497135153e-05, + "loss": 0.2103, + "step": 23890 + }, + { + "epoch": 6.0, + "learning_rate": 3.877373328839457e-05, + "loss": 0.0619, + "step": 23900 + }, + { + "epoch": 6.0, + "learning_rate": 3.8766711605437595e-05, + "loss": 0.7083, + "step": 23910 + }, + { + "epoch": 6.0, + "learning_rate": 3.875968992248062e-05, + "loss": 0.0011, + "step": 23920 + }, + { + "epoch": 6.0, + "learning_rate": 3.875266823952365e-05, + "loss": 0.0061, + "step": 23930 + }, + { + "epoch": 6.0, + "learning_rate": 3.874564655656668e-05, + "loss": 0.0063, + "step": 23940 + }, + { + "epoch": 6.0, + "learning_rate": 3.8738624873609705e-05, + "loss": 0.4345, + "step": 23950 + }, + { + "epoch": 6.0, + "learning_rate": 3.873160319065274e-05, + "loss": 0.004, + "step": 23960 + }, + { + "epoch": 6.0, + "learning_rate": 3.8724581507695767e-05, + "loss": 0.0004, + "step": 23970 + }, + { + "epoch": 6.0, + "learning_rate": 3.8717559824738794e-05, + "loss": 0.0133, + "step": 23980 + }, + { + "epoch": 6.0, + "learning_rate": 3.871053814178182e-05, + "loss": 0.0007, + "step": 23990 + }, + { + "epoch": 6.0, + "learning_rate": 3.870351645882485e-05, + "loss": 0.2953, + "step": 24000 + }, + { + "epoch": 6.0, + "learning_rate": 3.869649477586788e-05, + "loss": 0.1378, + "step": 24010 + }, + { + "epoch": 6.0, + "learning_rate": 3.868947309291091e-05, + "loss": 0.0034, + "step": 24020 + }, + { + "epoch": 6.0, + "learning_rate": 3.868245140995394e-05, + "loss": 0.0018, + "step": 24030 + }, + { + "epoch": 6.0, + "learning_rate": 3.867542972699697e-05, + "loss": 0.0244, + "step": 24040 + }, + { + "epoch": 6.0, + "learning_rate": 3.866840804404e-05, + "loss": 0.3075, + "step": 24050 + }, + { + "epoch": 6.0, + "learning_rate": 3.866138636108302e-05, + "loss": 0.0041, + "step": 24060 + }, + { + "epoch": 6.0, + "learning_rate": 3.8654364678126055e-05, + "loss": 0.3307, + "step": 24070 + }, + { + "epoch": 6.0, + "learning_rate": 3.864734299516908e-05, + "loss": 0.0004, + "step": 24080 + }, + { + "epoch": 6.0, + "learning_rate": 3.864032131221211e-05, + "loss": 0.3166, + "step": 24090 + }, + { + "epoch": 6.0, + "learning_rate": 3.8633299629255144e-05, + "loss": 0.8682, + "step": 24100 + }, + { + "epoch": 6.0, + "learning_rate": 3.862627794629817e-05, + "loss": 0.003, + "step": 24110 + }, + { + "epoch": 6.0, + "learning_rate": 3.86192562633412e-05, + "loss": 0.02, + "step": 24120 + }, + { + "epoch": 6.0, + "learning_rate": 3.861223458038423e-05, + "loss": 0.4391, + "step": 24130 + }, + { + "epoch": 6.01, + "learning_rate": 3.8605212897427254e-05, + "loss": 0.0906, + "step": 24140 + }, + { + "epoch": 6.01, + "learning_rate": 3.859819121447029e-05, + "loss": 0.2667, + "step": 24150 + }, + { + "epoch": 6.01, + "learning_rate": 3.8591169531513316e-05, + "loss": 0.0532, + "step": 24160 + }, + { + "epoch": 6.01, + "learning_rate": 3.8584147848556343e-05, + "loss": 0.9993, + "step": 24170 + }, + { + "epoch": 6.01, + "learning_rate": 3.857712616559938e-05, + "loss": 0.0008, + "step": 24180 + }, + { + "epoch": 6.01, + "learning_rate": 3.85701044826424e-05, + "loss": 0.0463, + "step": 24190 + }, + { + "epoch": 6.01, + "learning_rate": 3.856308279968543e-05, + "loss": 0.6413, + "step": 24200 + }, + { + "epoch": 6.01, + "learning_rate": 3.855606111672846e-05, + "loss": 0.6506, + "step": 24210 + }, + { + "epoch": 6.01, + "learning_rate": 3.854903943377149e-05, + "loss": 0.3106, + "step": 24220 + }, + { + "epoch": 6.01, + "learning_rate": 3.854201775081452e-05, + "loss": 0.4717, + "step": 24230 + }, + { + "epoch": 6.01, + "learning_rate": 3.853499606785755e-05, + "loss": 0.662, + "step": 24240 + }, + { + "epoch": 6.01, + "learning_rate": 3.852797438490057e-05, + "loss": 0.1547, + "step": 24250 + }, + { + "epoch": 6.01, + "learning_rate": 3.8520952701943604e-05, + "loss": 0.15, + "step": 24260 + }, + { + "epoch": 6.01, + "learning_rate": 3.851393101898663e-05, + "loss": 0.0969, + "step": 24270 + }, + { + "epoch": 6.01, + "learning_rate": 3.850690933602966e-05, + "loss": 0.3112, + "step": 24280 + }, + { + "epoch": 6.01, + "learning_rate": 3.8499887653072694e-05, + "loss": 0.6392, + "step": 24290 + }, + { + "epoch": 6.01, + "learning_rate": 3.849286597011572e-05, + "loss": 0.1817, + "step": 24300 + }, + { + "epoch": 6.01, + "learning_rate": 3.848584428715875e-05, + "loss": 0.4131, + "step": 24310 + }, + { + "epoch": 6.01, + "learning_rate": 3.8478822604201776e-05, + "loss": 0.1765, + "step": 24320 + }, + { + "epoch": 6.01, + "learning_rate": 3.8471800921244804e-05, + "loss": 0.2671, + "step": 24330 + }, + { + "epoch": 6.01, + "learning_rate": 3.846477923828784e-05, + "loss": 0.0687, + "step": 24340 + }, + { + "epoch": 6.01, + "learning_rate": 3.8457757555330865e-05, + "loss": 0.3037, + "step": 24350 + }, + { + "epoch": 6.01, + "learning_rate": 3.845073587237389e-05, + "loss": 0.5952, + "step": 24360 + }, + { + "epoch": 6.01, + "learning_rate": 3.844371418941693e-05, + "loss": 0.3804, + "step": 24370 + }, + { + "epoch": 6.01, + "learning_rate": 3.843669250645995e-05, + "loss": 0.0009, + "step": 24380 + }, + { + "epoch": 6.01, + "learning_rate": 3.8429670823502975e-05, + "loss": 0.1322, + "step": 24390 + }, + { + "epoch": 6.01, + "learning_rate": 3.842264914054601e-05, + "loss": 0.0094, + "step": 24400 + }, + { + "epoch": 6.01, + "learning_rate": 3.841562745758904e-05, + "loss": 0.0009, + "step": 24410 + }, + { + "epoch": 6.01, + "learning_rate": 3.8408605774632065e-05, + "loss": 0.3703, + "step": 24420 + }, + { + "epoch": 6.01, + "learning_rate": 3.84015840916751e-05, + "loss": 0.5393, + "step": 24430 + }, + { + "epoch": 6.01, + "learning_rate": 3.839456240871812e-05, + "loss": 0.6295, + "step": 24440 + }, + { + "epoch": 6.01, + "learning_rate": 3.8387540725761154e-05, + "loss": 0.6145, + "step": 24450 + }, + { + "epoch": 6.01, + "learning_rate": 3.838051904280418e-05, + "loss": 0.0032, + "step": 24460 + }, + { + "epoch": 6.01, + "learning_rate": 3.837349735984721e-05, + "loss": 0.1562, + "step": 24470 + }, + { + "epoch": 6.01, + "learning_rate": 3.836647567689024e-05, + "loss": 0.5566, + "step": 24480 + }, + { + "epoch": 6.01, + "learning_rate": 3.835945399393327e-05, + "loss": 0.0026, + "step": 24490 + }, + { + "epoch": 6.01, + "learning_rate": 3.835243231097629e-05, + "loss": 0.2902, + "step": 24500 + }, + { + "epoch": 6.01, + "learning_rate": 3.8345410628019326e-05, + "loss": 0.448, + "step": 24510 + }, + { + "epoch": 6.01, + "learning_rate": 3.833838894506235e-05, + "loss": 0.1989, + "step": 24520 + }, + { + "epoch": 6.01, + "learning_rate": 3.833136726210538e-05, + "loss": 0.0017, + "step": 24530 + }, + { + "epoch": 6.01, + "learning_rate": 3.8324345579148415e-05, + "loss": 0.275, + "step": 24540 + }, + { + "epoch": 6.01, + "learning_rate": 3.831732389619144e-05, + "loss": 0.0016, + "step": 24550 + }, + { + "epoch": 6.01, + "learning_rate": 3.831030221323447e-05, + "loss": 0.3677, + "step": 24560 + }, + { + "epoch": 6.01, + "learning_rate": 3.83032805302775e-05, + "loss": 0.0057, + "step": 24570 + }, + { + "epoch": 6.01, + "learning_rate": 3.8296258847320525e-05, + "loss": 0.0479, + "step": 24580 + }, + { + "epoch": 6.01, + "learning_rate": 3.828923716436356e-05, + "loss": 0.7014, + "step": 24590 + }, + { + "epoch": 6.01, + "learning_rate": 3.8282215481406587e-05, + "loss": 0.0064, + "step": 24600 + }, + { + "epoch": 6.01, + "learning_rate": 3.8275193798449614e-05, + "loss": 0.5305, + "step": 24610 + }, + { + "epoch": 6.01, + "learning_rate": 3.826817211549265e-05, + "loss": 0.0166, + "step": 24620 + }, + { + "epoch": 6.01, + "learning_rate": 3.826115043253567e-05, + "loss": 0.0083, + "step": 24630 + }, + { + "epoch": 6.01, + "learning_rate": 3.8254128749578697e-05, + "loss": 0.0036, + "step": 24640 + }, + { + "epoch": 6.01, + "learning_rate": 3.824710706662173e-05, + "loss": 0.002, + "step": 24650 + }, + { + "epoch": 6.01, + "learning_rate": 3.824008538366476e-05, + "loss": 0.0032, + "step": 24660 + }, + { + "epoch": 6.01, + "learning_rate": 3.8233063700707786e-05, + "loss": 0.4356, + "step": 24670 + }, + { + "epoch": 6.01, + "learning_rate": 3.822604201775082e-05, + "loss": 0.4848, + "step": 24680 + }, + { + "epoch": 6.01, + "learning_rate": 3.821902033479384e-05, + "loss": 0.014, + "step": 24690 + }, + { + "epoch": 6.01, + "learning_rate": 3.8211998651836875e-05, + "loss": 0.003, + "step": 24700 + }, + { + "epoch": 6.01, + "learning_rate": 3.82049769688799e-05, + "loss": 0.8338, + "step": 24710 + }, + { + "epoch": 6.01, + "learning_rate": 3.819795528592293e-05, + "loss": 0.3657, + "step": 24720 + }, + { + "epoch": 6.01, + "learning_rate": 3.8190933602965964e-05, + "loss": 0.2609, + "step": 24730 + }, + { + "epoch": 6.01, + "learning_rate": 3.818391192000899e-05, + "loss": 0.1945, + "step": 24740 + }, + { + "epoch": 6.01, + "learning_rate": 3.817689023705202e-05, + "loss": 0.1867, + "step": 24750 + }, + { + "epoch": 6.01, + "learning_rate": 3.816986855409505e-05, + "loss": 0.2197, + "step": 24760 + }, + { + "epoch": 6.01, + "learning_rate": 3.8162846871138074e-05, + "loss": 0.017, + "step": 24770 + }, + { + "epoch": 6.01, + "learning_rate": 3.81558251881811e-05, + "loss": 0.0445, + "step": 24780 + }, + { + "epoch": 6.01, + "learning_rate": 3.8148803505224136e-05, + "loss": 0.0087, + "step": 24790 + }, + { + "epoch": 6.01, + "learning_rate": 3.8141781822267163e-05, + "loss": 0.2765, + "step": 24800 + }, + { + "epoch": 6.01, + "learning_rate": 3.813476013931019e-05, + "loss": 0.3395, + "step": 24810 + }, + { + "epoch": 6.01, + "learning_rate": 3.812773845635322e-05, + "loss": 0.3436, + "step": 24820 + }, + { + "epoch": 6.01, + "learning_rate": 3.8120716773396246e-05, + "loss": 0.1448, + "step": 24830 + }, + { + "epoch": 6.01, + "learning_rate": 3.811369509043928e-05, + "loss": 0.0023, + "step": 24840 + }, + { + "epoch": 6.01, + "learning_rate": 3.810667340748231e-05, + "loss": 0.0142, + "step": 24850 + }, + { + "epoch": 6.01, + "learning_rate": 3.8099651724525335e-05, + "loss": 0.2859, + "step": 24860 + }, + { + "epoch": 6.01, + "learning_rate": 3.809263004156837e-05, + "loss": 0.9813, + "step": 24870 + }, + { + "epoch": 6.01, + "learning_rate": 3.80856083586114e-05, + "loss": 0.0237, + "step": 24880 + }, + { + "epoch": 6.01, + "learning_rate": 3.807858667565442e-05, + "loss": 0.1827, + "step": 24890 + }, + { + "epoch": 6.01, + "learning_rate": 3.807156499269745e-05, + "loss": 0.4976, + "step": 24900 + }, + { + "epoch": 6.01, + "learning_rate": 3.806454330974048e-05, + "loss": 0.1168, + "step": 24910 + }, + { + "epoch": 6.01, + "learning_rate": 3.805752162678351e-05, + "loss": 0.1007, + "step": 24920 + }, + { + "epoch": 6.02, + "learning_rate": 3.805049994382654e-05, + "loss": 0.0406, + "step": 24930 + }, + { + "epoch": 6.02, + "learning_rate": 3.804347826086957e-05, + "loss": 0.9261, + "step": 24940 + }, + { + "epoch": 6.02, + "learning_rate": 3.8036456577912596e-05, + "loss": 0.1076, + "step": 24950 + }, + { + "epoch": 6.02, + "learning_rate": 3.8029434894955624e-05, + "loss": 0.5967, + "step": 24960 + }, + { + "epoch": 6.02, + "learning_rate": 3.802241321199865e-05, + "loss": 0.0064, + "step": 24970 + }, + { + "epoch": 6.02, + "learning_rate": 3.8015391529041685e-05, + "loss": 0.0082, + "step": 24980 + }, + { + "epoch": 6.02, + "learning_rate": 3.800836984608471e-05, + "loss": 0.4952, + "step": 24990 + }, + { + "epoch": 6.02, + "learning_rate": 3.800134816312774e-05, + "loss": 0.0075, + "step": 25000 + }, + { + "epoch": 6.02, + "learning_rate": 3.799432648017077e-05, + "loss": 0.4248, + "step": 25010 + }, + { + "epoch": 6.02, + "learning_rate": 3.7987304797213795e-05, + "loss": 0.0157, + "step": 25020 + }, + { + "epoch": 6.02, + "learning_rate": 3.798028311425682e-05, + "loss": 0.0927, + "step": 25030 + }, + { + "epoch": 6.02, + "learning_rate": 3.797326143129986e-05, + "loss": 0.1675, + "step": 25040 + }, + { + "epoch": 6.02, + "learning_rate": 3.7966239748342885e-05, + "loss": 0.2663, + "step": 25050 + }, + { + "epoch": 6.02, + "learning_rate": 3.795921806538591e-05, + "loss": 0.008, + "step": 25060 + }, + { + "epoch": 6.02, + "learning_rate": 3.7952196382428946e-05, + "loss": 0.1999, + "step": 25070 + }, + { + "epoch": 6.02, + "learning_rate": 3.794517469947197e-05, + "loss": 0.0987, + "step": 25080 + }, + { + "epoch": 6.02, + "learning_rate": 3.7938153016515e-05, + "loss": 0.0028, + "step": 25090 + }, + { + "epoch": 6.02, + "learning_rate": 3.793113133355803e-05, + "loss": 0.7511, + "step": 25100 + }, + { + "epoch": 6.02, + "learning_rate": 3.7924109650601056e-05, + "loss": 0.2222, + "step": 25110 + }, + { + "epoch": 6.02, + "learning_rate": 3.791708796764409e-05, + "loss": 0.4223, + "step": 25120 + }, + { + "epoch": 6.02, + "learning_rate": 3.791006628468712e-05, + "loss": 0.6036, + "step": 25130 + }, + { + "epoch": 6.02, + "learning_rate": 3.790304460173014e-05, + "loss": 0.0135, + "step": 25140 + }, + { + "epoch": 6.02, + "learning_rate": 3.789602291877317e-05, + "loss": 0.0518, + "step": 25150 + }, + { + "epoch": 6.02, + "learning_rate": 3.78890012358162e-05, + "loss": 0.5503, + "step": 25160 + }, + { + "epoch": 6.02, + "learning_rate": 3.788197955285923e-05, + "loss": 0.2223, + "step": 25170 + }, + { + "epoch": 6.02, + "learning_rate": 3.787495786990226e-05, + "loss": 0.2021, + "step": 25180 + }, + { + "epoch": 6.02, + "learning_rate": 3.786793618694529e-05, + "loss": 0.0026, + "step": 25190 + }, + { + "epoch": 6.02, + "learning_rate": 3.786091450398832e-05, + "loss": 0.4532, + "step": 25200 + }, + { + "epoch": 6.02, + "learning_rate": 3.7853892821031345e-05, + "loss": 0.007, + "step": 25210 + }, + { + "epoch": 6.02, + "learning_rate": 3.784687113807437e-05, + "loss": 0.2655, + "step": 25220 + }, + { + "epoch": 6.02, + "learning_rate": 3.7839849455117407e-05, + "loss": 0.0007, + "step": 25230 + }, + { + "epoch": 6.02, + "learning_rate": 3.7832827772160434e-05, + "loss": 0.4196, + "step": 25240 + }, + { + "epoch": 6.02, + "learning_rate": 3.782580608920346e-05, + "loss": 0.5951, + "step": 25250 + }, + { + "epoch": 6.02, + "learning_rate": 3.7818784406246496e-05, + "loss": 0.0036, + "step": 25260 + }, + { + "epoch": 6.02, + "learning_rate": 3.7811762723289517e-05, + "loss": 0.0743, + "step": 25270 + }, + { + "epoch": 6.02, + "learning_rate": 3.780474104033255e-05, + "loss": 0.0314, + "step": 25280 + }, + { + "epoch": 6.02, + "learning_rate": 3.779771935737558e-05, + "loss": 0.1623, + "step": 25290 + }, + { + "epoch": 6.02, + "learning_rate": 3.7790697674418606e-05, + "loss": 0.5138, + "step": 25300 + }, + { + "epoch": 6.02, + "learning_rate": 3.778367599146164e-05, + "loss": 0.058, + "step": 25310 + }, + { + "epoch": 6.02, + "learning_rate": 3.777665430850467e-05, + "loss": 0.347, + "step": 25320 + }, + { + "epoch": 6.02, + "learning_rate": 3.776963262554769e-05, + "loss": 0.0236, + "step": 25330 + }, + { + "epoch": 6.02, + "learning_rate": 3.776261094259072e-05, + "loss": 0.0058, + "step": 25340 + }, + { + "epoch": 6.02, + "learning_rate": 3.775558925963375e-05, + "loss": 0.2422, + "step": 25350 + }, + { + "epoch": 6.02, + "learning_rate": 3.774856757667678e-05, + "loss": 0.0111, + "step": 25360 + }, + { + "epoch": 6.02, + "learning_rate": 3.774154589371981e-05, + "loss": 0.7988, + "step": 25370 + }, + { + "epoch": 6.02, + "learning_rate": 3.773452421076284e-05, + "loss": 0.0002, + "step": 25380 + }, + { + "epoch": 6.02, + "learning_rate": 3.772750252780587e-05, + "loss": 0.5258, + "step": 25390 + }, + { + "epoch": 6.02, + "learning_rate": 3.7720480844848894e-05, + "loss": 0.4104, + "step": 25400 + }, + { + "epoch": 6.02, + "learning_rate": 3.771345916189192e-05, + "loss": 0.303, + "step": 25410 + }, + { + "epoch": 6.02, + "learning_rate": 3.7706437478934956e-05, + "loss": 0.0061, + "step": 25420 + }, + { + "epoch": 6.02, + "learning_rate": 3.7699415795977983e-05, + "loss": 0.001, + "step": 25430 + }, + { + "epoch": 6.02, + "learning_rate": 3.769239411302101e-05, + "loss": 0.003, + "step": 25440 + }, + { + "epoch": 6.02, + "learning_rate": 3.7685372430064045e-05, + "loss": 0.2896, + "step": 25450 + }, + { + "epoch": 6.02, + "learning_rate": 3.7678350747107066e-05, + "loss": 0.5275, + "step": 25460 + }, + { + "epoch": 6.02, + "learning_rate": 3.7671329064150093e-05, + "loss": 0.6238, + "step": 25470 + }, + { + "epoch": 6.02, + "learning_rate": 3.766430738119313e-05, + "loss": 0.0679, + "step": 25480 + }, + { + "epoch": 6.02, + "learning_rate": 3.7657285698236155e-05, + "loss": 0.4135, + "step": 25490 + }, + { + "epoch": 6.02, + "learning_rate": 3.765026401527918e-05, + "loss": 0.0017, + "step": 25500 + }, + { + "epoch": 6.02, + "learning_rate": 3.764324233232222e-05, + "loss": 0.0515, + "step": 25510 + }, + { + "epoch": 6.02, + "learning_rate": 3.763622064936524e-05, + "loss": 0.4121, + "step": 25520 + }, + { + "epoch": 6.02, + "learning_rate": 3.762919896640827e-05, + "loss": 0.0124, + "step": 25530 + }, + { + "epoch": 6.02, + "learning_rate": 3.76221772834513e-05, + "loss": 0.0022, + "step": 25540 + }, + { + "epoch": 6.02, + "learning_rate": 3.761515560049433e-05, + "loss": 0.0003, + "step": 25550 + }, + { + "epoch": 6.02, + "learning_rate": 3.760813391753736e-05, + "loss": 0.0048, + "step": 25560 + }, + { + "epoch": 6.02, + "learning_rate": 3.760111223458039e-05, + "loss": 0.2637, + "step": 25570 + }, + { + "epoch": 6.02, + "learning_rate": 3.7594090551623416e-05, + "loss": 0.0003, + "step": 25580 + }, + { + "epoch": 6.02, + "learning_rate": 3.7587068868666444e-05, + "loss": 0.0125, + "step": 25590 + }, + { + "epoch": 6.02, + "learning_rate": 3.758004718570947e-05, + "loss": 0.0177, + "step": 25600 + }, + { + "epoch": 6.02, + "learning_rate": 3.75730255027525e-05, + "loss": 0.5735, + "step": 25610 + }, + { + "epoch": 6.02, + "learning_rate": 3.756600381979553e-05, + "loss": 0.0645, + "step": 25620 + }, + { + "epoch": 6.02, + "learning_rate": 3.755898213683856e-05, + "loss": 0.0306, + "step": 25630 + }, + { + "epoch": 6.02, + "learning_rate": 3.755196045388159e-05, + "loss": 0.0027, + "step": 25640 + }, + { + "epoch": 6.02, + "learning_rate": 3.7544938770924615e-05, + "loss": 0.5213, + "step": 25650 + }, + { + "epoch": 6.02, + "learning_rate": 3.753791708796764e-05, + "loss": 0.7319, + "step": 25660 + }, + { + "epoch": 6.02, + "learning_rate": 3.753089540501068e-05, + "loss": 0.1354, + "step": 25670 + }, + { + "epoch": 6.02, + "learning_rate": 3.7523873722053705e-05, + "loss": 0.1037, + "step": 25680 + }, + { + "epoch": 6.02, + "learning_rate": 3.751685203909673e-05, + "loss": 0.0696, + "step": 25690 + }, + { + "epoch": 6.02, + "learning_rate": 3.7509830356139766e-05, + "loss": 0.4739, + "step": 25700 + }, + { + "epoch": 6.02, + "learning_rate": 3.750280867318279e-05, + "loss": 0.7503, + "step": 25710 + }, + { + "epoch": 6.03, + "learning_rate": 3.7495786990225815e-05, + "loss": 0.417, + "step": 25720 + }, + { + "epoch": 6.03, + "learning_rate": 3.748876530726885e-05, + "loss": 0.4642, + "step": 25730 + }, + { + "epoch": 6.03, + "learning_rate": 3.7481743624311876e-05, + "loss": 0.8133, + "step": 25740 + }, + { + "epoch": 6.03, + "learning_rate": 3.7474721941354904e-05, + "loss": 0.8836, + "step": 25750 + }, + { + "epoch": 6.03, + "learning_rate": 3.746770025839794e-05, + "loss": 0.0914, + "step": 25760 + }, + { + "epoch": 6.03, + "learning_rate": 3.7460678575440966e-05, + "loss": 0.0004, + "step": 25770 + }, + { + "epoch": 6.03, + "learning_rate": 3.745365689248399e-05, + "loss": 0.0446, + "step": 25780 + }, + { + "epoch": 6.03, + "learning_rate": 3.744663520952702e-05, + "loss": 0.0741, + "step": 25790 + }, + { + "epoch": 6.03, + "learning_rate": 3.743961352657005e-05, + "loss": 0.0296, + "step": 25800 + }, + { + "epoch": 6.03, + "learning_rate": 3.743259184361308e-05, + "loss": 0.0113, + "step": 25810 + }, + { + "epoch": 6.03, + "learning_rate": 3.742557016065611e-05, + "loss": 0.5611, + "step": 25820 + }, + { + "epoch": 6.03, + "learning_rate": 3.741854847769914e-05, + "loss": 0.4513, + "step": 25830 + }, + { + "epoch": 6.03, + "learning_rate": 3.7411526794742165e-05, + "loss": 0.6155, + "step": 25840 + }, + { + "epoch": 6.03, + "learning_rate": 3.740450511178519e-05, + "loss": 0.769, + "step": 25850 + }, + { + "epoch": 6.03, + "learning_rate": 3.739748342882822e-05, + "loss": 0.1689, + "step": 25860 + }, + { + "epoch": 6.03, + "learning_rate": 3.7390461745871254e-05, + "loss": 0.0003, + "step": 25870 + }, + { + "epoch": 6.03, + "learning_rate": 3.738344006291428e-05, + "loss": 0.8232, + "step": 25880 + }, + { + "epoch": 6.03, + "learning_rate": 3.737641837995731e-05, + "loss": 0.2764, + "step": 25890 + }, + { + "epoch": 6.03, + "learning_rate": 3.7369396697000336e-05, + "loss": 0.0061, + "step": 25900 + }, + { + "epoch": 6.03, + "learning_rate": 3.7362375014043364e-05, + "loss": 0.0909, + "step": 25910 + }, + { + "epoch": 6.03, + "learning_rate": 3.73553533310864e-05, + "loss": 0.1014, + "step": 25920 + }, + { + "epoch": 6.03, + "learning_rate": 3.7348331648129426e-05, + "loss": 0.2242, + "step": 25930 + }, + { + "epoch": 6.03, + "learning_rate": 3.734130996517245e-05, + "loss": 0.3879, + "step": 25940 + }, + { + "epoch": 6.03, + "learning_rate": 3.733428828221549e-05, + "loss": 0.0435, + "step": 25950 + }, + { + "epoch": 6.03, + "learning_rate": 3.7327266599258515e-05, + "loss": 0.3426, + "step": 25960 + }, + { + "epoch": 6.03, + "learning_rate": 3.7320244916301536e-05, + "loss": 0.0162, + "step": 25970 + }, + { + "epoch": 6.03, + "learning_rate": 3.731322323334457e-05, + "loss": 0.0391, + "step": 25980 + }, + { + "epoch": 6.03, + "learning_rate": 3.73062015503876e-05, + "loss": 0.0699, + "step": 25990 + }, + { + "epoch": 6.03, + "learning_rate": 3.7299179867430625e-05, + "loss": 0.2113, + "step": 26000 + }, + { + "epoch": 6.03, + "learning_rate": 3.729215818447366e-05, + "loss": 0.7397, + "step": 26010 + }, + { + "epoch": 6.03, + "learning_rate": 3.728513650151669e-05, + "loss": 0.0492, + "step": 26020 + }, + { + "epoch": 6.03, + "learning_rate": 3.7278114818559714e-05, + "loss": 0.602, + "step": 26030 + }, + { + "epoch": 6.03, + "learning_rate": 3.727109313560274e-05, + "loss": 0.0208, + "step": 26040 + }, + { + "epoch": 6.03, + "learning_rate": 3.726407145264577e-05, + "loss": 0.2615, + "step": 26050 + }, + { + "epoch": 6.03, + "learning_rate": 3.7257049769688803e-05, + "loss": 0.0017, + "step": 26060 + }, + { + "epoch": 6.03, + "learning_rate": 3.725002808673183e-05, + "loss": 0.0001, + "step": 26070 + }, + { + "epoch": 6.03, + "learning_rate": 3.724300640377486e-05, + "loss": 0.0105, + "step": 26080 + }, + { + "epoch": 6.03, + "learning_rate": 3.7235984720817886e-05, + "loss": 0.3485, + "step": 26090 + }, + { + "epoch": 6.03, + "learning_rate": 3.7228963037860913e-05, + "loss": 0.5605, + "step": 26100 + }, + { + "epoch": 6.03, + "learning_rate": 3.722194135490394e-05, + "loss": 0.0348, + "step": 26110 + }, + { + "epoch": 6.03, + "learning_rate": 3.7214919671946975e-05, + "loss": 0.4793, + "step": 26120 + }, + { + "epoch": 6.03, + "learning_rate": 3.720789798899e-05, + "loss": 0.0048, + "step": 26130 + }, + { + "epoch": 6.03, + "learning_rate": 3.720087630603303e-05, + "loss": 0.0101, + "step": 26140 + }, + { + "epoch": 6.03, + "learning_rate": 3.7193854623076064e-05, + "loss": 0.0826, + "step": 26150 + }, + { + "epoch": 6.03, + "learning_rate": 3.7186832940119085e-05, + "loss": 0.0055, + "step": 26160 + }, + { + "epoch": 6.03, + "learning_rate": 3.717981125716212e-05, + "loss": 0.0047, + "step": 26170 + }, + { + "epoch": 6.03, + "learning_rate": 3.717278957420515e-05, + "loss": 0.0309, + "step": 26180 + }, + { + "epoch": 6.03, + "learning_rate": 3.7165767891248174e-05, + "loss": 0.3236, + "step": 26190 + }, + { + "epoch": 6.03, + "learning_rate": 3.715874620829121e-05, + "loss": 0.0048, + "step": 26200 + }, + { + "epoch": 6.03, + "learning_rate": 3.7151724525334236e-05, + "loss": 0.6334, + "step": 26210 + }, + { + "epoch": 6.03, + "learning_rate": 3.7144702842377264e-05, + "loss": 0.0005, + "step": 26220 + }, + { + "epoch": 6.03, + "learning_rate": 3.713768115942029e-05, + "loss": 0.0081, + "step": 26230 + }, + { + "epoch": 6.03, + "learning_rate": 3.713065947646332e-05, + "loss": 0.4601, + "step": 26240 + }, + { + "epoch": 6.03, + "learning_rate": 3.712363779350635e-05, + "loss": 0.1685, + "step": 26250 + }, + { + "epoch": 6.03, + "learning_rate": 3.711661611054938e-05, + "loss": 0.0046, + "step": 26260 + }, + { + "epoch": 6.03, + "learning_rate": 3.710959442759241e-05, + "loss": 0.1018, + "step": 26270 + }, + { + "epoch": 6.03, + "learning_rate": 3.710257274463544e-05, + "loss": 0.4142, + "step": 26280 + }, + { + "epoch": 6.03, + "learning_rate": 3.709555106167846e-05, + "loss": 0.4348, + "step": 26290 + }, + { + "epoch": 6.03, + "learning_rate": 3.708852937872149e-05, + "loss": 0.9754, + "step": 26300 + }, + { + "epoch": 6.03, + "learning_rate": 3.7081507695764525e-05, + "loss": 0.2764, + "step": 26310 + }, + { + "epoch": 6.03, + "learning_rate": 3.707448601280755e-05, + "loss": 0.4001, + "step": 26320 + }, + { + "epoch": 6.03, + "learning_rate": 3.706746432985058e-05, + "loss": 0.001, + "step": 26330 + }, + { + "epoch": 6.03, + "learning_rate": 3.7060442646893614e-05, + "loss": 0.6406, + "step": 26340 + }, + { + "epoch": 6.03, + "learning_rate": 3.7053420963936635e-05, + "loss": 0.0598, + "step": 26350 + }, + { + "epoch": 6.03, + "learning_rate": 3.704639928097967e-05, + "loss": 0.1162, + "step": 26360 + }, + { + "epoch": 6.03, + "learning_rate": 3.7039377598022696e-05, + "loss": 0.0559, + "step": 26370 + }, + { + "epoch": 6.03, + "learning_rate": 3.7032355915065724e-05, + "loss": 0.041, + "step": 26380 + }, + { + "epoch": 6.03, + "learning_rate": 3.702533423210876e-05, + "loss": 0.0002, + "step": 26390 + }, + { + "epoch": 6.03, + "learning_rate": 3.7018312549151786e-05, + "loss": 0.619, + "step": 26400 + }, + { + "epoch": 6.03, + "learning_rate": 3.7011290866194806e-05, + "loss": 0.6035, + "step": 26410 + }, + { + "epoch": 6.03, + "learning_rate": 3.700426918323784e-05, + "loss": 0.2843, + "step": 26420 + }, + { + "epoch": 6.03, + "learning_rate": 3.699724750028087e-05, + "loss": 0.2225, + "step": 26430 + }, + { + "epoch": 6.03, + "learning_rate": 3.6990225817323896e-05, + "loss": 0.1261, + "step": 26440 + }, + { + "epoch": 6.03, + "learning_rate": 3.698320413436693e-05, + "loss": 0.2374, + "step": 26450 + }, + { + "epoch": 6.03, + "learning_rate": 3.697618245140996e-05, + "loss": 0.7946, + "step": 26460 + }, + { + "epoch": 6.03, + "learning_rate": 3.6969160768452985e-05, + "loss": 0.0056, + "step": 26470 + }, + { + "epoch": 6.03, + "learning_rate": 3.696213908549601e-05, + "loss": 0.6467, + "step": 26480 + }, + { + "epoch": 6.03, + "learning_rate": 3.695511740253904e-05, + "loss": 0.0011, + "step": 26490 + }, + { + "epoch": 6.03, + "learning_rate": 3.6948095719582074e-05, + "loss": 0.0001, + "step": 26500 + }, + { + "epoch": 6.04, + "learning_rate": 3.69410740366251e-05, + "loss": 0.0249, + "step": 26510 + }, + { + "epoch": 6.04, + "learning_rate": 3.693405235366813e-05, + "loss": 0.2286, + "step": 26520 + }, + { + "epoch": 6.04, + "learning_rate": 3.692703067071116e-05, + "loss": 0.2254, + "step": 26530 + }, + { + "epoch": 6.04, + "learning_rate": 3.6920008987754184e-05, + "loss": 0.0413, + "step": 26540 + }, + { + "epoch": 6.04, + "learning_rate": 3.691298730479721e-05, + "loss": 0.2523, + "step": 26550 + }, + { + "epoch": 6.04, + "learning_rate": 3.6905965621840246e-05, + "loss": 0.3633, + "step": 26560 + }, + { + "epoch": 6.04, + "learning_rate": 3.689894393888327e-05, + "loss": 0.0664, + "step": 26570 + }, + { + "epoch": 6.04, + "learning_rate": 3.68919222559263e-05, + "loss": 1.0771, + "step": 26580 + }, + { + "epoch": 6.04, + "learning_rate": 3.6884900572969335e-05, + "loss": 0.3229, + "step": 26590 + }, + { + "epoch": 6.04, + "learning_rate": 3.6877878890012356e-05, + "loss": 0.6796, + "step": 26600 + }, + { + "epoch": 6.04, + "learning_rate": 3.687085720705539e-05, + "loss": 0.0293, + "step": 26610 + }, + { + "epoch": 6.04, + "learning_rate": 3.686383552409842e-05, + "loss": 0.119, + "step": 26620 + }, + { + "epoch": 6.04, + "learning_rate": 3.6856813841141445e-05, + "loss": 0.0073, + "step": 26630 + }, + { + "epoch": 6.04, + "learning_rate": 3.684979215818448e-05, + "loss": 0.3176, + "step": 26640 + }, + { + "epoch": 6.04, + "learning_rate": 3.684277047522751e-05, + "loss": 0.0003, + "step": 26650 + }, + { + "epoch": 6.04, + "learning_rate": 3.6835748792270534e-05, + "loss": 0.1201, + "step": 26660 + }, + { + "epoch": 6.04, + "learning_rate": 3.682872710931356e-05, + "loss": 0.2782, + "step": 26670 + }, + { + "epoch": 6.04, + "learning_rate": 3.682170542635659e-05, + "loss": 0.0851, + "step": 26680 + }, + { + "epoch": 6.04, + "learning_rate": 3.681468374339962e-05, + "loss": 0.003, + "step": 26690 + }, + { + "epoch": 6.04, + "learning_rate": 3.680766206044265e-05, + "loss": 0.111, + "step": 26700 + }, + { + "epoch": 6.04, + "learning_rate": 3.680064037748568e-05, + "loss": 0.001, + "step": 26710 + }, + { + "epoch": 6.04, + "learning_rate": 3.6793618694528706e-05, + "loss": 0.1466, + "step": 26720 + }, + { + "epoch": 6.04, + "learning_rate": 3.6786597011571733e-05, + "loss": 0.004, + "step": 26730 + }, + { + "epoch": 6.04, + "learning_rate": 3.677957532861476e-05, + "loss": 0.2867, + "step": 26740 + }, + { + "epoch": 6.04, + "learning_rate": 3.6772553645657795e-05, + "loss": 0.0594, + "step": 26750 + }, + { + "epoch": 6.04, + "learning_rate": 3.676553196270082e-05, + "loss": 0.2029, + "step": 26760 + }, + { + "epoch": 6.04, + "learning_rate": 3.675851027974385e-05, + "loss": 0.0365, + "step": 26770 + }, + { + "epoch": 6.04, + "learning_rate": 3.6751488596786884e-05, + "loss": 0.8334, + "step": 26780 + }, + { + "epoch": 6.04, + "learning_rate": 3.6744466913829905e-05, + "loss": 0.0057, + "step": 26790 + }, + { + "epoch": 6.04, + "learning_rate": 3.673744523087293e-05, + "loss": 0.5338, + "step": 26800 + }, + { + "epoch": 6.04, + "learning_rate": 3.673042354791597e-05, + "loss": 0.0152, + "step": 26810 + }, + { + "epoch": 6.04, + "learning_rate": 3.6723401864958994e-05, + "loss": 0.4646, + "step": 26820 + }, + { + "epoch": 6.04, + "learning_rate": 3.671638018200202e-05, + "loss": 0.245, + "step": 26830 + }, + { + "epoch": 6.04, + "learning_rate": 3.6709358499045056e-05, + "loss": 0.0079, + "step": 26840 + }, + { + "epoch": 6.04, + "learning_rate": 3.6702336816088084e-05, + "loss": 0.8329, + "step": 26850 + }, + { + "epoch": 6.04, + "learning_rate": 3.669531513313111e-05, + "loss": 0.0549, + "step": 26860 + }, + { + "epoch": 6.04, + "learning_rate": 3.668829345017414e-05, + "loss": 0.0295, + "step": 26870 + }, + { + "epoch": 6.04, + "learning_rate": 3.6681271767217166e-05, + "loss": 0.1756, + "step": 26880 + }, + { + "epoch": 6.04, + "learning_rate": 3.66742500842602e-05, + "loss": 0.3095, + "step": 26890 + }, + { + "epoch": 6.04, + "learning_rate": 3.666722840130323e-05, + "loss": 0.0078, + "step": 26900 + }, + { + "epoch": 6.04, + "learning_rate": 3.6660206718346255e-05, + "loss": 0.1504, + "step": 26910 + }, + { + "epoch": 6.04, + "learning_rate": 3.665318503538928e-05, + "loss": 0.5866, + "step": 26920 + }, + { + "epoch": 6.04, + "learning_rate": 3.664616335243231e-05, + "loss": 0.4116, + "step": 26930 + }, + { + "epoch": 6.04, + "learning_rate": 3.663914166947534e-05, + "loss": 0.004, + "step": 26940 + }, + { + "epoch": 6.04, + "learning_rate": 3.663211998651837e-05, + "loss": 0.1242, + "step": 26950 + }, + { + "epoch": 6.04, + "learning_rate": 3.66250983035614e-05, + "loss": 0.4462, + "step": 26960 + }, + { + "epoch": 6.04, + "learning_rate": 3.661807662060443e-05, + "loss": 0.524, + "step": 26970 + }, + { + "epoch": 6.04, + "learning_rate": 3.661105493764746e-05, + "loss": 0.0011, + "step": 26980 + }, + { + "epoch": 6.04, + "learning_rate": 3.660403325469048e-05, + "loss": 0.0019, + "step": 26990 + }, + { + "epoch": 6.04, + "learning_rate": 3.6597011571733516e-05, + "loss": 0.3936, + "step": 27000 + }, + { + "epoch": 6.04, + "learning_rate": 3.6589989888776544e-05, + "loss": 0.0889, + "step": 27010 + }, + { + "epoch": 6.04, + "learning_rate": 3.658296820581957e-05, + "loss": 0.1896, + "step": 27020 + }, + { + "epoch": 6.04, + "learning_rate": 3.6575946522862606e-05, + "loss": 0.7003, + "step": 27030 + }, + { + "epoch": 6.04, + "learning_rate": 3.656892483990563e-05, + "loss": 0.0064, + "step": 27040 + }, + { + "epoch": 6.04, + "learning_rate": 3.6561903156948654e-05, + "loss": 0.093, + "step": 27050 + }, + { + "epoch": 6.04, + "learning_rate": 3.655488147399169e-05, + "loss": 0.0022, + "step": 27060 + }, + { + "epoch": 6.04, + "learning_rate": 3.6547859791034716e-05, + "loss": 0.4901, + "step": 27070 + }, + { + "epoch": 6.04, + "learning_rate": 3.654083810807774e-05, + "loss": 1.2747, + "step": 27080 + }, + { + "epoch": 6.04, + "learning_rate": 3.653381642512078e-05, + "loss": 0.0199, + "step": 27090 + }, + { + "epoch": 6.04, + "learning_rate": 3.6526794742163805e-05, + "loss": 0.0371, + "step": 27100 + }, + { + "epoch": 6.04, + "learning_rate": 3.651977305920683e-05, + "loss": 0.797, + "step": 27110 + }, + { + "epoch": 6.04, + "learning_rate": 3.651275137624986e-05, + "loss": 0.7422, + "step": 27120 + }, + { + "epoch": 6.04, + "learning_rate": 3.650572969329289e-05, + "loss": 0.0101, + "step": 27130 + }, + { + "epoch": 6.04, + "learning_rate": 3.649870801033592e-05, + "loss": 0.0001, + "step": 27140 + }, + { + "epoch": 6.04, + "learning_rate": 3.649168632737895e-05, + "loss": 0.0021, + "step": 27150 + }, + { + "epoch": 6.04, + "learning_rate": 3.6484664644421976e-05, + "loss": 0.7335, + "step": 27160 + }, + { + "epoch": 6.04, + "learning_rate": 3.647764296146501e-05, + "loss": 0.1274, + "step": 27170 + }, + { + "epoch": 6.04, + "learning_rate": 3.647062127850803e-05, + "loss": 0.0006, + "step": 27180 + }, + { + "epoch": 6.04, + "learning_rate": 3.646359959555106e-05, + "loss": 0.0915, + "step": 27190 + }, + { + "epoch": 6.04, + "learning_rate": 3.645657791259409e-05, + "loss": 0.0009, + "step": 27200 + }, + { + "epoch": 6.04, + "learning_rate": 3.644955622963712e-05, + "loss": 0.3436, + "step": 27210 + }, + { + "epoch": 6.04, + "learning_rate": 3.644253454668015e-05, + "loss": 0.5576, + "step": 27220 + }, + { + "epoch": 6.04, + "learning_rate": 3.643551286372318e-05, + "loss": 0.0003, + "step": 27230 + }, + { + "epoch": 6.04, + "learning_rate": 3.64284911807662e-05, + "loss": 0.0034, + "step": 27240 + }, + { + "epoch": 6.04, + "learning_rate": 3.642146949780924e-05, + "loss": 0.473, + "step": 27250 + }, + { + "epoch": 6.04, + "learning_rate": 3.6414447814852265e-05, + "loss": 0.0146, + "step": 27260 + }, + { + "epoch": 6.04, + "learning_rate": 3.640742613189529e-05, + "loss": 0.0194, + "step": 27270 + }, + { + "epoch": 6.04, + "learning_rate": 3.640040444893833e-05, + "loss": 0.0033, + "step": 27280 + }, + { + "epoch": 6.04, + "learning_rate": 3.6393382765981354e-05, + "loss": 0.0789, + "step": 27290 + }, + { + "epoch": 6.05, + "learning_rate": 3.638636108302438e-05, + "loss": 0.7161, + "step": 27300 + }, + { + "epoch": 6.05, + "learning_rate": 3.637933940006741e-05, + "loss": 1.1196, + "step": 27310 + }, + { + "epoch": 6.05, + "learning_rate": 3.637231771711044e-05, + "loss": 0.0014, + "step": 27320 + }, + { + "epoch": 6.05, + "learning_rate": 3.636529603415347e-05, + "loss": 0.0204, + "step": 27330 + }, + { + "epoch": 6.05, + "learning_rate": 3.63582743511965e-05, + "loss": 0.0689, + "step": 27340 + }, + { + "epoch": 6.05, + "learning_rate": 3.6351252668239526e-05, + "loss": 0.1117, + "step": 27350 + }, + { + "epoch": 6.05, + "learning_rate": 3.634423098528256e-05, + "loss": 0.1221, + "step": 27360 + }, + { + "epoch": 6.05, + "learning_rate": 3.633720930232558e-05, + "loss": 0.0344, + "step": 27370 + }, + { + "epoch": 6.05, + "learning_rate": 3.633018761936861e-05, + "loss": 0.5914, + "step": 27380 + }, + { + "epoch": 6.05, + "learning_rate": 3.632316593641164e-05, + "loss": 0.0157, + "step": 27390 + }, + { + "epoch": 6.05, + "learning_rate": 3.631614425345467e-05, + "loss": 0.4351, + "step": 27400 + }, + { + "epoch": 6.05, + "learning_rate": 3.63091225704977e-05, + "loss": 0.0035, + "step": 27410 + }, + { + "epoch": 6.05, + "learning_rate": 3.630210088754073e-05, + "loss": 0.061, + "step": 27420 + }, + { + "epoch": 6.05, + "learning_rate": 3.629507920458375e-05, + "loss": 0.6405, + "step": 27430 + }, + { + "epoch": 6.05, + "learning_rate": 3.628805752162679e-05, + "loss": 0.0035, + "step": 27440 + }, + { + "epoch": 6.05, + "learning_rate": 3.6281035838669814e-05, + "loss": 0.2484, + "step": 27450 + }, + { + "epoch": 6.05, + "learning_rate": 3.627401415571284e-05, + "loss": 0.0345, + "step": 27460 + }, + { + "epoch": 6.05, + "learning_rate": 3.6266992472755876e-05, + "loss": 0.5987, + "step": 27470 + }, + { + "epoch": 6.05, + "learning_rate": 3.6259970789798904e-05, + "loss": 0.1555, + "step": 27480 + }, + { + "epoch": 6.05, + "learning_rate": 3.6252949106841924e-05, + "loss": 0.9912, + "step": 27490 + }, + { + "epoch": 6.05, + "learning_rate": 3.624592742388496e-05, + "loss": 0.0824, + "step": 27500 + }, + { + "epoch": 6.05, + "learning_rate": 3.6238905740927986e-05, + "loss": 0.01, + "step": 27510 + }, + { + "epoch": 6.05, + "learning_rate": 3.6231884057971014e-05, + "loss": 0.1639, + "step": 27520 + }, + { + "epoch": 6.05, + "learning_rate": 3.622486237501405e-05, + "loss": 0.0021, + "step": 27530 + }, + { + "epoch": 6.05, + "learning_rate": 3.6217840692057075e-05, + "loss": 0.0013, + "step": 27540 + }, + { + "epoch": 6.05, + "learning_rate": 3.62108190091001e-05, + "loss": 0.0108, + "step": 27550 + }, + { + "epoch": 6.05, + "learning_rate": 3.620379732614313e-05, + "loss": 0.0004, + "step": 27560 + }, + { + "epoch": 6.05, + "learning_rate": 3.619677564318616e-05, + "loss": 0.2031, + "step": 27570 + }, + { + "epoch": 6.05, + "learning_rate": 3.618975396022919e-05, + "loss": 0.1731, + "step": 27580 + }, + { + "epoch": 6.05, + "learning_rate": 3.618273227727222e-05, + "loss": 0.5768, + "step": 27590 + }, + { + "epoch": 6.05, + "learning_rate": 3.617571059431525e-05, + "loss": 0.4574, + "step": 27600 + }, + { + "epoch": 6.05, + "learning_rate": 3.616868891135828e-05, + "loss": 0.3211, + "step": 27610 + }, + { + "epoch": 6.05, + "learning_rate": 3.61616672284013e-05, + "loss": 0.0006, + "step": 27620 + }, + { + "epoch": 6.05, + "learning_rate": 3.615464554544433e-05, + "loss": 0.0007, + "step": 27630 + }, + { + "epoch": 6.05, + "learning_rate": 3.6147623862487364e-05, + "loss": 0.1417, + "step": 27640 + }, + { + "epoch": 6.05, + "learning_rate": 3.614060217953039e-05, + "loss": 0.0664, + "step": 27650 + }, + { + "epoch": 6.05, + "learning_rate": 3.613358049657342e-05, + "loss": 0.0414, + "step": 27660 + }, + { + "epoch": 6.05, + "learning_rate": 3.612655881361645e-05, + "loss": 0.7866, + "step": 27670 + }, + { + "epoch": 6.05, + "learning_rate": 3.6119537130659474e-05, + "loss": 0.7867, + "step": 27680 + }, + { + "epoch": 6.05, + "learning_rate": 3.611251544770251e-05, + "loss": 0.4971, + "step": 27690 + }, + { + "epoch": 6.05, + "eval_accuracy": 0.7175848277641594, + "eval_loss": 1.7131096124649048, + "eval_runtime": 1394.7548, + "eval_samples_per_second": 5.557, + "eval_steps_per_second": 2.779, + "step": 27692 + }, + { + "epoch": 7.0, + "learning_rate": 3.6105493764745536e-05, + "loss": 0.272, + "step": 27700 + }, + { + "epoch": 7.0, + "learning_rate": 3.609847208178856e-05, + "loss": 0.0072, + "step": 27710 + }, + { + "epoch": 7.0, + "learning_rate": 3.60914503988316e-05, + "loss": 0.2494, + "step": 27720 + }, + { + "epoch": 7.0, + "learning_rate": 3.6084428715874625e-05, + "loss": 0.3872, + "step": 27730 + }, + { + "epoch": 7.0, + "learning_rate": 3.607740703291765e-05, + "loss": 0.4893, + "step": 27740 + }, + { + "epoch": 7.0, + "learning_rate": 3.607038534996068e-05, + "loss": 0.2327, + "step": 27750 + }, + { + "epoch": 7.0, + "learning_rate": 3.606336366700371e-05, + "loss": 0.0145, + "step": 27760 + }, + { + "epoch": 7.0, + "learning_rate": 3.6056341984046735e-05, + "loss": 0.3899, + "step": 27770 + }, + { + "epoch": 7.0, + "learning_rate": 3.604932030108977e-05, + "loss": 0.2911, + "step": 27780 + }, + { + "epoch": 7.0, + "learning_rate": 3.6042298618132796e-05, + "loss": 0.0005, + "step": 27790 + }, + { + "epoch": 7.0, + "learning_rate": 3.6035276935175824e-05, + "loss": 0.0309, + "step": 27800 + }, + { + "epoch": 7.0, + "learning_rate": 3.602825525221885e-05, + "loss": 0.0006, + "step": 27810 + }, + { + "epoch": 7.0, + "learning_rate": 3.602123356926188e-05, + "loss": 0.463, + "step": 27820 + }, + { + "epoch": 7.0, + "learning_rate": 3.601421188630491e-05, + "loss": 0.0001, + "step": 27830 + }, + { + "epoch": 7.0, + "learning_rate": 3.600719020334794e-05, + "loss": 0.0106, + "step": 27840 + }, + { + "epoch": 7.0, + "learning_rate": 3.600016852039097e-05, + "loss": 0.0941, + "step": 27850 + }, + { + "epoch": 7.0, + "learning_rate": 3.5993146837434e-05, + "loss": 0.2478, + "step": 27860 + }, + { + "epoch": 7.0, + "learning_rate": 3.598612515447703e-05, + "loss": 0.3401, + "step": 27870 + }, + { + "epoch": 7.0, + "learning_rate": 3.597910347152005e-05, + "loss": 0.0784, + "step": 27880 + }, + { + "epoch": 7.0, + "learning_rate": 3.5972081788563085e-05, + "loss": 0.014, + "step": 27890 + }, + { + "epoch": 7.0, + "learning_rate": 3.596506010560611e-05, + "loss": 0.422, + "step": 27900 + }, + { + "epoch": 7.0, + "learning_rate": 3.595803842264914e-05, + "loss": 0.0008, + "step": 27910 + }, + { + "epoch": 7.0, + "learning_rate": 3.5951016739692174e-05, + "loss": 0.0017, + "step": 27920 + }, + { + "epoch": 7.0, + "learning_rate": 3.59439950567352e-05, + "loss": 0.1601, + "step": 27930 + }, + { + "epoch": 7.0, + "learning_rate": 3.593697337377823e-05, + "loss": 0.347, + "step": 27940 + }, + { + "epoch": 7.0, + "learning_rate": 3.592995169082126e-05, + "loss": 0.2947, + "step": 27950 + }, + { + "epoch": 7.0, + "learning_rate": 3.5922930007864284e-05, + "loss": 0.0042, + "step": 27960 + }, + { + "epoch": 7.0, + "learning_rate": 3.591590832490732e-05, + "loss": 0.0002, + "step": 27970 + }, + { + "epoch": 7.0, + "learning_rate": 3.5908886641950346e-05, + "loss": 0.6514, + "step": 27980 + }, + { + "epoch": 7.0, + "learning_rate": 3.5901864958993373e-05, + "loss": 0.4686, + "step": 27990 + }, + { + "epoch": 7.0, + "learning_rate": 3.58948432760364e-05, + "loss": 0.0002, + "step": 28000 + }, + { + "epoch": 7.0, + "learning_rate": 3.588782159307943e-05, + "loss": 0.5549, + "step": 28010 + }, + { + "epoch": 7.0, + "learning_rate": 3.5880799910122456e-05, + "loss": 0.0004, + "step": 28020 + }, + { + "epoch": 7.0, + "learning_rate": 3.587377822716549e-05, + "loss": 0.4606, + "step": 28030 + }, + { + "epoch": 7.0, + "learning_rate": 3.586675654420852e-05, + "loss": 0.0909, + "step": 28040 + }, + { + "epoch": 7.0, + "learning_rate": 3.5859734861251545e-05, + "loss": 0.0003, + "step": 28050 + }, + { + "epoch": 7.0, + "learning_rate": 3.585271317829458e-05, + "loss": 0.0632, + "step": 28060 + }, + { + "epoch": 7.0, + "learning_rate": 3.58456914953376e-05, + "loss": 0.123, + "step": 28070 + }, + { + "epoch": 7.0, + "learning_rate": 3.5838669812380634e-05, + "loss": 0.3751, + "step": 28080 + }, + { + "epoch": 7.01, + "learning_rate": 3.583164812942366e-05, + "loss": 0.0006, + "step": 28090 + }, + { + "epoch": 7.01, + "learning_rate": 3.582462644646669e-05, + "loss": 0.0791, + "step": 28100 + }, + { + "epoch": 7.01, + "learning_rate": 3.5817604763509724e-05, + "loss": 0.0018, + "step": 28110 + }, + { + "epoch": 7.01, + "learning_rate": 3.581058308055275e-05, + "loss": 0.5267, + "step": 28120 + }, + { + "epoch": 7.01, + "learning_rate": 3.580356139759577e-05, + "loss": 0.3585, + "step": 28130 + }, + { + "epoch": 7.01, + "learning_rate": 3.5796539714638806e-05, + "loss": 0.0035, + "step": 28140 + }, + { + "epoch": 7.01, + "learning_rate": 3.5789518031681834e-05, + "loss": 0.1072, + "step": 28150 + }, + { + "epoch": 7.01, + "learning_rate": 3.578249634872486e-05, + "loss": 0.0013, + "step": 28160 + }, + { + "epoch": 7.01, + "learning_rate": 3.5775474665767895e-05, + "loss": 0.083, + "step": 28170 + }, + { + "epoch": 7.01, + "learning_rate": 3.576845298281092e-05, + "loss": 0.2895, + "step": 28180 + }, + { + "epoch": 7.01, + "learning_rate": 3.576143129985395e-05, + "loss": 0.0109, + "step": 28190 + }, + { + "epoch": 7.01, + "learning_rate": 3.575440961689698e-05, + "loss": 0.0041, + "step": 28200 + }, + { + "epoch": 7.01, + "learning_rate": 3.5747387933940005e-05, + "loss": 0.1891, + "step": 28210 + }, + { + "epoch": 7.01, + "learning_rate": 3.574036625098304e-05, + "loss": 0.003, + "step": 28220 + }, + { + "epoch": 7.01, + "learning_rate": 3.573334456802607e-05, + "loss": 0.0087, + "step": 28230 + }, + { + "epoch": 7.01, + "learning_rate": 3.5726322885069095e-05, + "loss": 0.0054, + "step": 28240 + }, + { + "epoch": 7.01, + "learning_rate": 3.571930120211213e-05, + "loss": 0.1366, + "step": 28250 + }, + { + "epoch": 7.01, + "learning_rate": 3.571227951915515e-05, + "loss": 0.5405, + "step": 28260 + }, + { + "epoch": 7.01, + "learning_rate": 3.5705257836198184e-05, + "loss": 0.0003, + "step": 28270 + }, + { + "epoch": 7.01, + "learning_rate": 3.569823615324121e-05, + "loss": 0.736, + "step": 28280 + }, + { + "epoch": 7.01, + "learning_rate": 3.569121447028424e-05, + "loss": 0.0106, + "step": 28290 + }, + { + "epoch": 7.01, + "learning_rate": 3.568419278732727e-05, + "loss": 0.008, + "step": 28300 + }, + { + "epoch": 7.01, + "learning_rate": 3.56771711043703e-05, + "loss": 0.0224, + "step": 28310 + }, + { + "epoch": 7.01, + "learning_rate": 3.567014942141332e-05, + "loss": 0.4835, + "step": 28320 + }, + { + "epoch": 7.01, + "learning_rate": 3.5663127738456356e-05, + "loss": 0.2162, + "step": 28330 + }, + { + "epoch": 7.01, + "learning_rate": 3.565610605549938e-05, + "loss": 0.3026, + "step": 28340 + }, + { + "epoch": 7.01, + "learning_rate": 3.564908437254241e-05, + "loss": 0.0768, + "step": 28350 + }, + { + "epoch": 7.01, + "learning_rate": 3.5642062689585445e-05, + "loss": 0.0926, + "step": 28360 + }, + { + "epoch": 7.01, + "learning_rate": 3.563504100662847e-05, + "loss": 0.0573, + "step": 28370 + }, + { + "epoch": 7.01, + "learning_rate": 3.56280193236715e-05, + "loss": 0.1824, + "step": 28380 + }, + { + "epoch": 7.01, + "learning_rate": 3.562099764071453e-05, + "loss": 0.0278, + "step": 28390 + }, + { + "epoch": 7.01, + "learning_rate": 3.5613975957757555e-05, + "loss": 0.0138, + "step": 28400 + }, + { + "epoch": 7.01, + "learning_rate": 3.560695427480059e-05, + "loss": 0.22, + "step": 28410 + }, + { + "epoch": 7.01, + "learning_rate": 3.5599932591843616e-05, + "loss": 0.5692, + "step": 28420 + }, + { + "epoch": 7.01, + "learning_rate": 3.5592910908886644e-05, + "loss": 0.0006, + "step": 28430 + }, + { + "epoch": 7.01, + "learning_rate": 3.558588922592968e-05, + "loss": 0.016, + "step": 28440 + }, + { + "epoch": 7.01, + "learning_rate": 3.55788675429727e-05, + "loss": 0.0001, + "step": 28450 + }, + { + "epoch": 7.01, + "learning_rate": 3.5571845860015726e-05, + "loss": 0.0342, + "step": 28460 + }, + { + "epoch": 7.01, + "learning_rate": 3.556482417705876e-05, + "loss": 0.0937, + "step": 28470 + }, + { + "epoch": 7.01, + "learning_rate": 3.555780249410179e-05, + "loss": 0.1361, + "step": 28480 + }, + { + "epoch": 7.01, + "learning_rate": 3.5550780811144816e-05, + "loss": 0.0781, + "step": 28490 + }, + { + "epoch": 7.01, + "learning_rate": 3.554375912818785e-05, + "loss": 0.0881, + "step": 28500 + }, + { + "epoch": 7.01, + "learning_rate": 3.553673744523087e-05, + "loss": 0.2175, + "step": 28510 + }, + { + "epoch": 7.01, + "learning_rate": 3.5529715762273905e-05, + "loss": 0.0002, + "step": 28520 + }, + { + "epoch": 7.01, + "learning_rate": 3.552269407931693e-05, + "loss": 0.6173, + "step": 28530 + }, + { + "epoch": 7.01, + "learning_rate": 3.551567239635996e-05, + "loss": 0.5094, + "step": 28540 + }, + { + "epoch": 7.01, + "learning_rate": 3.5508650713402994e-05, + "loss": 0.0174, + "step": 28550 + }, + { + "epoch": 7.01, + "learning_rate": 3.550162903044602e-05, + "loss": 0.1899, + "step": 28560 + }, + { + "epoch": 7.01, + "learning_rate": 3.549460734748905e-05, + "loss": 0.023, + "step": 28570 + }, + { + "epoch": 7.01, + "learning_rate": 3.548758566453208e-05, + "loss": 0.0005, + "step": 28580 + }, + { + "epoch": 7.01, + "learning_rate": 3.5480563981575104e-05, + "loss": 0.0006, + "step": 28590 + }, + { + "epoch": 7.01, + "learning_rate": 3.547354229861813e-05, + "loss": 0.0024, + "step": 28600 + }, + { + "epoch": 7.01, + "learning_rate": 3.5466520615661166e-05, + "loss": 0.0074, + "step": 28610 + }, + { + "epoch": 7.01, + "learning_rate": 3.545949893270419e-05, + "loss": 0.0004, + "step": 28620 + }, + { + "epoch": 7.01, + "learning_rate": 3.545247724974722e-05, + "loss": 0.0002, + "step": 28630 + }, + { + "epoch": 7.01, + "learning_rate": 3.544545556679025e-05, + "loss": 0.0002, + "step": 28640 + }, + { + "epoch": 7.01, + "learning_rate": 3.5438433883833276e-05, + "loss": 0.7221, + "step": 28650 + }, + { + "epoch": 7.01, + "learning_rate": 3.543141220087631e-05, + "loss": 0.4285, + "step": 28660 + }, + { + "epoch": 7.01, + "learning_rate": 3.542439051791934e-05, + "loss": 0.0015, + "step": 28670 + }, + { + "epoch": 7.01, + "learning_rate": 3.5417368834962365e-05, + "loss": 0.1156, + "step": 28680 + }, + { + "epoch": 7.01, + "learning_rate": 3.54103471520054e-05, + "loss": 0.5857, + "step": 28690 + }, + { + "epoch": 7.01, + "learning_rate": 3.540332546904842e-05, + "loss": 0.0472, + "step": 28700 + }, + { + "epoch": 7.01, + "learning_rate": 3.539630378609145e-05, + "loss": 0.0123, + "step": 28710 + }, + { + "epoch": 7.01, + "learning_rate": 3.538928210313448e-05, + "loss": 0.3325, + "step": 28720 + }, + { + "epoch": 7.01, + "learning_rate": 3.538226042017751e-05, + "loss": 0.0007, + "step": 28730 + }, + { + "epoch": 7.01, + "learning_rate": 3.537523873722054e-05, + "loss": 0.4211, + "step": 28740 + }, + { + "epoch": 7.01, + "learning_rate": 3.536821705426357e-05, + "loss": 0.0632, + "step": 28750 + }, + { + "epoch": 7.01, + "learning_rate": 3.53611953713066e-05, + "loss": 0.0235, + "step": 28760 + }, + { + "epoch": 7.01, + "learning_rate": 3.5354173688349626e-05, + "loss": 0.0002, + "step": 28770 + }, + { + "epoch": 7.01, + "learning_rate": 3.5347152005392654e-05, + "loss": 0.3494, + "step": 28780 + }, + { + "epoch": 7.01, + "learning_rate": 3.534013032243568e-05, + "loss": 0.0149, + "step": 28790 + }, + { + "epoch": 7.01, + "learning_rate": 3.5333108639478715e-05, + "loss": 0.0004, + "step": 28800 + }, + { + "epoch": 7.01, + "learning_rate": 3.532608695652174e-05, + "loss": 0.0022, + "step": 28810 + }, + { + "epoch": 7.01, + "learning_rate": 3.531906527356477e-05, + "loss": 0.0023, + "step": 28820 + }, + { + "epoch": 7.01, + "learning_rate": 3.53120435906078e-05, + "loss": 0.0015, + "step": 28830 + }, + { + "epoch": 7.01, + "learning_rate": 3.5305021907650825e-05, + "loss": 0.077, + "step": 28840 + }, + { + "epoch": 7.01, + "learning_rate": 3.529800022469385e-05, + "loss": 0.0223, + "step": 28850 + }, + { + "epoch": 7.01, + "learning_rate": 3.529097854173689e-05, + "loss": 0.2964, + "step": 28860 + }, + { + "epoch": 7.01, + "learning_rate": 3.5283956858779915e-05, + "loss": 0.2392, + "step": 28870 + }, + { + "epoch": 7.02, + "learning_rate": 3.527693517582294e-05, + "loss": 0.0018, + "step": 28880 + }, + { + "epoch": 7.02, + "learning_rate": 3.526991349286597e-05, + "loss": 0.2846, + "step": 28890 + }, + { + "epoch": 7.02, + "learning_rate": 3.5262891809909e-05, + "loss": 0.6291, + "step": 28900 + }, + { + "epoch": 7.02, + "learning_rate": 3.525587012695203e-05, + "loss": 1.0295, + "step": 28910 + }, + { + "epoch": 7.02, + "learning_rate": 3.524884844399506e-05, + "loss": 1.4269, + "step": 28920 + }, + { + "epoch": 7.02, + "learning_rate": 3.5241826761038086e-05, + "loss": 0.0138, + "step": 28930 + }, + { + "epoch": 7.02, + "learning_rate": 3.523480507808112e-05, + "loss": 0.0782, + "step": 28940 + }, + { + "epoch": 7.02, + "learning_rate": 3.522778339512415e-05, + "loss": 0.311, + "step": 28950 + }, + { + "epoch": 7.02, + "learning_rate": 3.522076171216717e-05, + "loss": 0.0975, + "step": 28960 + }, + { + "epoch": 7.02, + "learning_rate": 3.52137400292102e-05, + "loss": 0.0021, + "step": 28970 + }, + { + "epoch": 7.02, + "learning_rate": 3.520671834625323e-05, + "loss": 0.0018, + "step": 28980 + }, + { + "epoch": 7.02, + "learning_rate": 3.519969666329626e-05, + "loss": 0.554, + "step": 28990 + }, + { + "epoch": 7.02, + "learning_rate": 3.519267498033929e-05, + "loss": 0.026, + "step": 29000 + }, + { + "epoch": 7.02, + "learning_rate": 3.518565329738232e-05, + "loss": 0.0937, + "step": 29010 + }, + { + "epoch": 7.02, + "learning_rate": 3.517863161442535e-05, + "loss": 0.0122, + "step": 29020 + }, + { + "epoch": 7.02, + "learning_rate": 3.5171609931468375e-05, + "loss": 0.3284, + "step": 29030 + }, + { + "epoch": 7.02, + "learning_rate": 3.51645882485114e-05, + "loss": 0.0158, + "step": 29040 + }, + { + "epoch": 7.02, + "learning_rate": 3.5157566565554436e-05, + "loss": 0.0046, + "step": 29050 + }, + { + "epoch": 7.02, + "learning_rate": 3.5150544882597464e-05, + "loss": 0.0012, + "step": 29060 + }, + { + "epoch": 7.02, + "learning_rate": 3.514352319964049e-05, + "loss": 0.0105, + "step": 29070 + }, + { + "epoch": 7.02, + "learning_rate": 3.513650151668352e-05, + "loss": 0.0004, + "step": 29080 + }, + { + "epoch": 7.02, + "learning_rate": 3.5129479833726546e-05, + "loss": 0.0032, + "step": 29090 + }, + { + "epoch": 7.02, + "learning_rate": 3.5122458150769574e-05, + "loss": 0.0007, + "step": 29100 + }, + { + "epoch": 7.02, + "learning_rate": 3.511543646781261e-05, + "loss": 0.0002, + "step": 29110 + }, + { + "epoch": 7.02, + "learning_rate": 3.5108414784855636e-05, + "loss": 0.027, + "step": 29120 + }, + { + "epoch": 7.02, + "learning_rate": 3.510139310189866e-05, + "loss": 0.0002, + "step": 29130 + }, + { + "epoch": 7.02, + "learning_rate": 3.50943714189417e-05, + "loss": 0.0293, + "step": 29140 + }, + { + "epoch": 7.02, + "learning_rate": 3.508734973598472e-05, + "loss": 0.0005, + "step": 29150 + }, + { + "epoch": 7.02, + "learning_rate": 3.508032805302775e-05, + "loss": 0.0215, + "step": 29160 + }, + { + "epoch": 7.02, + "learning_rate": 3.507330637007078e-05, + "loss": 0.4081, + "step": 29170 + }, + { + "epoch": 7.02, + "learning_rate": 3.506628468711381e-05, + "loss": 0.0332, + "step": 29180 + }, + { + "epoch": 7.02, + "learning_rate": 3.505926300415684e-05, + "loss": 0.1187, + "step": 29190 + }, + { + "epoch": 7.02, + "learning_rate": 3.505224132119987e-05, + "loss": 0.1211, + "step": 29200 + }, + { + "epoch": 7.02, + "learning_rate": 3.504521963824289e-05, + "loss": 0.3056, + "step": 29210 + }, + { + "epoch": 7.02, + "learning_rate": 3.5038197955285924e-05, + "loss": 0.193, + "step": 29220 + }, + { + "epoch": 7.02, + "learning_rate": 3.503117627232895e-05, + "loss": 0.3352, + "step": 29230 + }, + { + "epoch": 7.02, + "learning_rate": 3.502415458937198e-05, + "loss": 0.0603, + "step": 29240 + }, + { + "epoch": 7.02, + "learning_rate": 3.501713290641501e-05, + "loss": 0.1328, + "step": 29250 + }, + { + "epoch": 7.02, + "learning_rate": 3.501011122345804e-05, + "loss": 0.0417, + "step": 29260 + }, + { + "epoch": 7.02, + "learning_rate": 3.500308954050107e-05, + "loss": 0.0061, + "step": 29270 + }, + { + "epoch": 7.02, + "learning_rate": 3.4996067857544096e-05, + "loss": 0.0004, + "step": 29280 + }, + { + "epoch": 7.02, + "learning_rate": 3.498904617458712e-05, + "loss": 0.7128, + "step": 29290 + }, + { + "epoch": 7.02, + "learning_rate": 3.498202449163016e-05, + "loss": 0.0015, + "step": 29300 + }, + { + "epoch": 7.02, + "learning_rate": 3.4975002808673185e-05, + "loss": 0.012, + "step": 29310 + }, + { + "epoch": 7.02, + "learning_rate": 3.496798112571621e-05, + "loss": 0.1025, + "step": 29320 + }, + { + "epoch": 7.02, + "learning_rate": 3.496095944275925e-05, + "loss": 0.063, + "step": 29330 + }, + { + "epoch": 7.02, + "learning_rate": 3.495393775980227e-05, + "loss": 0.0023, + "step": 29340 + }, + { + "epoch": 7.02, + "learning_rate": 3.49469160768453e-05, + "loss": 0.3747, + "step": 29350 + }, + { + "epoch": 7.02, + "learning_rate": 3.493989439388833e-05, + "loss": 0.023, + "step": 29360 + }, + { + "epoch": 7.02, + "learning_rate": 3.493287271093136e-05, + "loss": 0.0006, + "step": 29370 + }, + { + "epoch": 7.02, + "learning_rate": 3.492585102797439e-05, + "loss": 0.4015, + "step": 29380 + }, + { + "epoch": 7.02, + "learning_rate": 3.491882934501742e-05, + "loss": 0.0005, + "step": 29390 + }, + { + "epoch": 7.02, + "learning_rate": 3.491180766206044e-05, + "loss": 0.0595, + "step": 29400 + }, + { + "epoch": 7.02, + "learning_rate": 3.4904785979103474e-05, + "loss": 0.0001, + "step": 29410 + }, + { + "epoch": 7.02, + "learning_rate": 3.48977642961465e-05, + "loss": 0.0028, + "step": 29420 + }, + { + "epoch": 7.02, + "learning_rate": 3.489074261318953e-05, + "loss": 0.0222, + "step": 29430 + }, + { + "epoch": 7.02, + "learning_rate": 3.488372093023256e-05, + "loss": 0.1849, + "step": 29440 + }, + { + "epoch": 7.02, + "learning_rate": 3.487669924727559e-05, + "loss": 0.7227, + "step": 29450 + }, + { + "epoch": 7.02, + "learning_rate": 3.486967756431862e-05, + "loss": 0.0002, + "step": 29460 + }, + { + "epoch": 7.02, + "learning_rate": 3.4862655881361645e-05, + "loss": 0.0009, + "step": 29470 + }, + { + "epoch": 7.02, + "learning_rate": 3.485563419840467e-05, + "loss": 0.3463, + "step": 29480 + }, + { + "epoch": 7.02, + "learning_rate": 3.484861251544771e-05, + "loss": 0.141, + "step": 29490 + }, + { + "epoch": 7.02, + "learning_rate": 3.4841590832490735e-05, + "loss": 0.1455, + "step": 29500 + }, + { + "epoch": 7.02, + "learning_rate": 3.483456914953376e-05, + "loss": 0.0004, + "step": 29510 + }, + { + "epoch": 7.02, + "learning_rate": 3.4827547466576796e-05, + "loss": 0.4195, + "step": 29520 + }, + { + "epoch": 7.02, + "learning_rate": 3.482052578361982e-05, + "loss": 0.0117, + "step": 29530 + }, + { + "epoch": 7.02, + "learning_rate": 3.4813504100662844e-05, + "loss": 0.0181, + "step": 29540 + }, + { + "epoch": 7.02, + "learning_rate": 3.480648241770588e-05, + "loss": 0.5655, + "step": 29550 + }, + { + "epoch": 7.02, + "learning_rate": 3.4799460734748906e-05, + "loss": 0.0662, + "step": 29560 + }, + { + "epoch": 7.02, + "learning_rate": 3.4792439051791934e-05, + "loss": 0.0878, + "step": 29570 + }, + { + "epoch": 7.02, + "learning_rate": 3.478541736883497e-05, + "loss": 0.2163, + "step": 29580 + }, + { + "epoch": 7.02, + "learning_rate": 3.477839568587799e-05, + "loss": 0.0687, + "step": 29590 + }, + { + "epoch": 7.02, + "learning_rate": 3.477137400292102e-05, + "loss": 0.0031, + "step": 29600 + }, + { + "epoch": 7.02, + "learning_rate": 3.476435231996405e-05, + "loss": 0.0136, + "step": 29610 + }, + { + "epoch": 7.02, + "learning_rate": 3.475733063700708e-05, + "loss": 0.0088, + "step": 29620 + }, + { + "epoch": 7.02, + "learning_rate": 3.475030895405011e-05, + "loss": 0.4367, + "step": 29630 + }, + { + "epoch": 7.02, + "learning_rate": 3.474328727109314e-05, + "loss": 0.4239, + "step": 29640 + }, + { + "epoch": 7.02, + "learning_rate": 3.473626558813617e-05, + "loss": 0.0006, + "step": 29650 + }, + { + "epoch": 7.02, + "learning_rate": 3.4729243905179195e-05, + "loss": 0.4047, + "step": 29660 + }, + { + "epoch": 7.03, + "learning_rate": 3.472222222222222e-05, + "loss": 0.0002, + "step": 29670 + }, + { + "epoch": 7.03, + "learning_rate": 3.471520053926525e-05, + "loss": 0.0021, + "step": 29680 + }, + { + "epoch": 7.03, + "learning_rate": 3.4708178856308284e-05, + "loss": 0.5187, + "step": 29690 + }, + { + "epoch": 7.03, + "learning_rate": 3.470115717335131e-05, + "loss": 0.0043, + "step": 29700 + }, + { + "epoch": 7.03, + "learning_rate": 3.469413549039434e-05, + "loss": 0.3976, + "step": 29710 + }, + { + "epoch": 7.03, + "learning_rate": 3.4687113807437366e-05, + "loss": 0.0715, + "step": 29720 + }, + { + "epoch": 7.03, + "learning_rate": 3.4680092124480394e-05, + "loss": 0.1434, + "step": 29730 + }, + { + "epoch": 7.03, + "learning_rate": 3.467307044152343e-05, + "loss": 0.3178, + "step": 29740 + }, + { + "epoch": 7.03, + "learning_rate": 3.4666048758566456e-05, + "loss": 0.0021, + "step": 29750 + }, + { + "epoch": 7.03, + "learning_rate": 3.465902707560948e-05, + "loss": 0.5062, + "step": 29760 + }, + { + "epoch": 7.03, + "learning_rate": 3.465200539265252e-05, + "loss": 0.6172, + "step": 29770 + }, + { + "epoch": 7.03, + "learning_rate": 3.464498370969554e-05, + "loss": 0.3, + "step": 29780 + }, + { + "epoch": 7.03, + "learning_rate": 3.4637962026738566e-05, + "loss": 0.2517, + "step": 29790 + }, + { + "epoch": 7.03, + "learning_rate": 3.46309403437816e-05, + "loss": 0.0003, + "step": 29800 + }, + { + "epoch": 7.03, + "learning_rate": 3.462391866082463e-05, + "loss": 0.7938, + "step": 29810 + }, + { + "epoch": 7.03, + "learning_rate": 3.4616896977867655e-05, + "loss": 0.1922, + "step": 29820 + }, + { + "epoch": 7.03, + "learning_rate": 3.460987529491069e-05, + "loss": 0.0187, + "step": 29830 + }, + { + "epoch": 7.03, + "learning_rate": 3.460285361195372e-05, + "loss": 0.4691, + "step": 29840 + }, + { + "epoch": 7.03, + "learning_rate": 3.4595831928996744e-05, + "loss": 0.4495, + "step": 29850 + }, + { + "epoch": 7.03, + "learning_rate": 3.458881024603977e-05, + "loss": 0.3445, + "step": 29860 + }, + { + "epoch": 7.03, + "learning_rate": 3.45817885630828e-05, + "loss": 0.0018, + "step": 29870 + }, + { + "epoch": 7.03, + "learning_rate": 3.457476688012583e-05, + "loss": 0.0496, + "step": 29880 + }, + { + "epoch": 7.03, + "learning_rate": 3.456774519716886e-05, + "loss": 0.2771, + "step": 29890 + }, + { + "epoch": 7.03, + "learning_rate": 3.456072351421189e-05, + "loss": 0.2365, + "step": 29900 + }, + { + "epoch": 7.03, + "learning_rate": 3.4553701831254916e-05, + "loss": 0.0022, + "step": 29910 + }, + { + "epoch": 7.03, + "learning_rate": 3.454668014829794e-05, + "loss": 0.0035, + "step": 29920 + }, + { + "epoch": 7.03, + "learning_rate": 3.453965846534097e-05, + "loss": 0.0024, + "step": 29930 + }, + { + "epoch": 7.03, + "learning_rate": 3.4532636782384005e-05, + "loss": 0.7458, + "step": 29940 + }, + { + "epoch": 7.03, + "learning_rate": 3.452561509942703e-05, + "loss": 0.4568, + "step": 29950 + }, + { + "epoch": 7.03, + "learning_rate": 3.451859341647006e-05, + "loss": 0.012, + "step": 29960 + }, + { + "epoch": 7.03, + "learning_rate": 3.4511571733513094e-05, + "loss": 0.006, + "step": 29970 + }, + { + "epoch": 7.03, + "learning_rate": 3.4504550050556115e-05, + "loss": 0.3732, + "step": 29980 + }, + { + "epoch": 7.03, + "learning_rate": 3.449752836759915e-05, + "loss": 0.2219, + "step": 29990 + }, + { + "epoch": 7.03, + "learning_rate": 3.449050668464218e-05, + "loss": 0.0031, + "step": 30000 + }, + { + "epoch": 7.03, + "learning_rate": 3.4483485001685204e-05, + "loss": 0.2734, + "step": 30010 + }, + { + "epoch": 7.03, + "learning_rate": 3.447646331872824e-05, + "loss": 0.0202, + "step": 30020 + }, + { + "epoch": 7.03, + "learning_rate": 3.4469441635771266e-05, + "loss": 0.2631, + "step": 30030 + }, + { + "epoch": 7.03, + "learning_rate": 3.446241995281429e-05, + "loss": 0.307, + "step": 30040 + }, + { + "epoch": 7.03, + "learning_rate": 3.445539826985732e-05, + "loss": 0.0131, + "step": 30050 + }, + { + "epoch": 7.03, + "learning_rate": 3.444837658690035e-05, + "loss": 0.1145, + "step": 30060 + }, + { + "epoch": 7.03, + "learning_rate": 3.4441354903943376e-05, + "loss": 0.0217, + "step": 30070 + }, + { + "epoch": 7.03, + "learning_rate": 3.443433322098641e-05, + "loss": 0.2386, + "step": 30080 + }, + { + "epoch": 7.03, + "learning_rate": 3.442731153802944e-05, + "loss": 0.0589, + "step": 30090 + }, + { + "epoch": 7.03, + "learning_rate": 3.4420289855072465e-05, + "loss": 0.9327, + "step": 30100 + }, + { + "epoch": 7.03, + "learning_rate": 3.441326817211549e-05, + "loss": 0.0082, + "step": 30110 + }, + { + "epoch": 7.03, + "learning_rate": 3.440624648915852e-05, + "loss": 0.0025, + "step": 30120 + }, + { + "epoch": 7.03, + "learning_rate": 3.4399224806201555e-05, + "loss": 0.0521, + "step": 30130 + }, + { + "epoch": 7.03, + "learning_rate": 3.439220312324458e-05, + "loss": 0.0685, + "step": 30140 + }, + { + "epoch": 7.03, + "learning_rate": 3.438518144028761e-05, + "loss": 0.0012, + "step": 30150 + }, + { + "epoch": 7.03, + "learning_rate": 3.4378159757330644e-05, + "loss": 0.1586, + "step": 30160 + }, + { + "epoch": 7.03, + "learning_rate": 3.4371138074373664e-05, + "loss": 0.0436, + "step": 30170 + }, + { + "epoch": 7.03, + "learning_rate": 3.436411639141669e-05, + "loss": 0.0643, + "step": 30180 + }, + { + "epoch": 7.03, + "learning_rate": 3.4357094708459726e-05, + "loss": 0.4885, + "step": 30190 + }, + { + "epoch": 7.03, + "learning_rate": 3.4350073025502754e-05, + "loss": 0.3713, + "step": 30200 + }, + { + "epoch": 7.03, + "learning_rate": 3.434305134254578e-05, + "loss": 0.3066, + "step": 30210 + }, + { + "epoch": 7.03, + "learning_rate": 3.4336029659588815e-05, + "loss": 0.1509, + "step": 30220 + }, + { + "epoch": 7.03, + "learning_rate": 3.4329007976631836e-05, + "loss": 0.3691, + "step": 30230 + }, + { + "epoch": 7.03, + "learning_rate": 3.432198629367487e-05, + "loss": 0.2895, + "step": 30240 + }, + { + "epoch": 7.03, + "learning_rate": 3.43149646107179e-05, + "loss": 0.2452, + "step": 30250 + }, + { + "epoch": 7.03, + "learning_rate": 3.4307942927760925e-05, + "loss": 0.2877, + "step": 30260 + }, + { + "epoch": 7.03, + "learning_rate": 3.430092124480396e-05, + "loss": 0.0377, + "step": 30270 + }, + { + "epoch": 7.03, + "learning_rate": 3.429389956184699e-05, + "loss": 0.0099, + "step": 30280 + }, + { + "epoch": 7.03, + "learning_rate": 3.4286877878890015e-05, + "loss": 0.1655, + "step": 30290 + }, + { + "epoch": 7.03, + "learning_rate": 3.427985619593304e-05, + "loss": 0.2542, + "step": 30300 + }, + { + "epoch": 7.03, + "learning_rate": 3.427283451297607e-05, + "loss": 0.0011, + "step": 30310 + }, + { + "epoch": 7.03, + "learning_rate": 3.4265812830019104e-05, + "loss": 0.0703, + "step": 30320 + }, + { + "epoch": 7.03, + "learning_rate": 3.425879114706213e-05, + "loss": 0.121, + "step": 30330 + }, + { + "epoch": 7.03, + "learning_rate": 3.425176946410516e-05, + "loss": 0.1228, + "step": 30340 + }, + { + "epoch": 7.03, + "learning_rate": 3.424474778114819e-05, + "loss": 0.5062, + "step": 30350 + }, + { + "epoch": 7.03, + "learning_rate": 3.4237726098191214e-05, + "loss": 0.0698, + "step": 30360 + }, + { + "epoch": 7.03, + "learning_rate": 3.423070441523424e-05, + "loss": 0.4472, + "step": 30370 + }, + { + "epoch": 7.03, + "learning_rate": 3.4223682732277276e-05, + "loss": 0.3525, + "step": 30380 + }, + { + "epoch": 7.03, + "learning_rate": 3.42166610493203e-05, + "loss": 0.2165, + "step": 30390 + }, + { + "epoch": 7.03, + "learning_rate": 3.420963936636333e-05, + "loss": 0.04, + "step": 30400 + }, + { + "epoch": 7.03, + "learning_rate": 3.4202617683406365e-05, + "loss": 0.0033, + "step": 30410 + }, + { + "epoch": 7.03, + "learning_rate": 3.4195596000449386e-05, + "loss": 0.2513, + "step": 30420 + }, + { + "epoch": 7.03, + "learning_rate": 3.418857431749242e-05, + "loss": 0.0381, + "step": 30430 + }, + { + "epoch": 7.03, + "learning_rate": 3.418155263453545e-05, + "loss": 0.1535, + "step": 30440 + }, + { + "epoch": 7.03, + "learning_rate": 3.4174530951578475e-05, + "loss": 0.069, + "step": 30450 + }, + { + "epoch": 7.03, + "learning_rate": 3.416750926862151e-05, + "loss": 0.0778, + "step": 30460 + }, + { + "epoch": 7.04, + "learning_rate": 3.416048758566454e-05, + "loss": 0.3365, + "step": 30470 + }, + { + "epoch": 7.04, + "learning_rate": 3.415346590270756e-05, + "loss": 0.2696, + "step": 30480 + }, + { + "epoch": 7.04, + "learning_rate": 3.414644421975059e-05, + "loss": 0.1984, + "step": 30490 + }, + { + "epoch": 7.04, + "learning_rate": 3.413942253679362e-05, + "loss": 0.3765, + "step": 30500 + }, + { + "epoch": 7.04, + "learning_rate": 3.4132400853836647e-05, + "loss": 0.1453, + "step": 30510 + }, + { + "epoch": 7.04, + "learning_rate": 3.412537917087968e-05, + "loss": 0.6647, + "step": 30520 + }, + { + "epoch": 7.04, + "learning_rate": 3.411835748792271e-05, + "loss": 0.0981, + "step": 30530 + }, + { + "epoch": 7.04, + "learning_rate": 3.4111335804965736e-05, + "loss": 0.6452, + "step": 30540 + }, + { + "epoch": 7.04, + "learning_rate": 3.410431412200876e-05, + "loss": 0.0009, + "step": 30550 + }, + { + "epoch": 7.04, + "learning_rate": 3.409729243905179e-05, + "loss": 0.1792, + "step": 30560 + }, + { + "epoch": 7.04, + "learning_rate": 3.4090270756094825e-05, + "loss": 0.0072, + "step": 30570 + }, + { + "epoch": 7.04, + "learning_rate": 3.408324907313785e-05, + "loss": 0.0565, + "step": 30580 + }, + { + "epoch": 7.04, + "learning_rate": 3.407622739018088e-05, + "loss": 0.0015, + "step": 30590 + }, + { + "epoch": 7.04, + "learning_rate": 3.4069205707223914e-05, + "loss": 0.0006, + "step": 30600 + }, + { + "epoch": 7.04, + "learning_rate": 3.4062184024266935e-05, + "loss": 0.0159, + "step": 30610 + }, + { + "epoch": 7.04, + "learning_rate": 3.405516234130996e-05, + "loss": 0.0004, + "step": 30620 + }, + { + "epoch": 7.04, + "learning_rate": 3.4048140658353e-05, + "loss": 0.8674, + "step": 30630 + }, + { + "epoch": 7.04, + "learning_rate": 3.4041118975396024e-05, + "loss": 0.2494, + "step": 30640 + }, + { + "epoch": 7.04, + "learning_rate": 3.403409729243905e-05, + "loss": 0.0008, + "step": 30650 + }, + { + "epoch": 7.04, + "learning_rate": 3.4027075609482086e-05, + "loss": 0.0003, + "step": 30660 + }, + { + "epoch": 7.04, + "learning_rate": 3.4020053926525114e-05, + "loss": 0.001, + "step": 30670 + }, + { + "epoch": 7.04, + "learning_rate": 3.401303224356814e-05, + "loss": 0.0023, + "step": 30680 + }, + { + "epoch": 7.04, + "learning_rate": 3.400601056061117e-05, + "loss": 0.0009, + "step": 30690 + }, + { + "epoch": 7.04, + "learning_rate": 3.3998988877654196e-05, + "loss": 0.3457, + "step": 30700 + }, + { + "epoch": 7.04, + "learning_rate": 3.399196719469723e-05, + "loss": 0.1078, + "step": 30710 + }, + { + "epoch": 7.04, + "learning_rate": 3.398494551174026e-05, + "loss": 0.7915, + "step": 30720 + }, + { + "epoch": 7.04, + "learning_rate": 3.3977923828783285e-05, + "loss": 0.1264, + "step": 30730 + }, + { + "epoch": 7.04, + "learning_rate": 3.397090214582631e-05, + "loss": 0.041, + "step": 30740 + }, + { + "epoch": 7.04, + "learning_rate": 3.396388046286934e-05, + "loss": 0.0005, + "step": 30750 + }, + { + "epoch": 7.04, + "learning_rate": 3.395685877991237e-05, + "loss": 0.0026, + "step": 30760 + }, + { + "epoch": 7.04, + "learning_rate": 3.39498370969554e-05, + "loss": 0.886, + "step": 30770 + }, + { + "epoch": 7.04, + "learning_rate": 3.394281541399843e-05, + "loss": 0.0001, + "step": 30780 + }, + { + "epoch": 7.04, + "learning_rate": 3.393579373104146e-05, + "loss": 0.034, + "step": 30790 + }, + { + "epoch": 7.04, + "learning_rate": 3.3928772048084484e-05, + "loss": 0.0589, + "step": 30800 + }, + { + "epoch": 7.04, + "learning_rate": 3.392175036512751e-05, + "loss": 0.0538, + "step": 30810 + }, + { + "epoch": 7.04, + "learning_rate": 3.3914728682170546e-05, + "loss": 0.3075, + "step": 30820 + }, + { + "epoch": 7.04, + "learning_rate": 3.3907706999213574e-05, + "loss": 0.4502, + "step": 30830 + }, + { + "epoch": 7.04, + "learning_rate": 3.39006853162566e-05, + "loss": 0.0057, + "step": 30840 + }, + { + "epoch": 7.04, + "learning_rate": 3.3893663633299635e-05, + "loss": 0.0005, + "step": 30850 + }, + { + "epoch": 7.04, + "learning_rate": 3.388664195034266e-05, + "loss": 0.1559, + "step": 30860 + }, + { + "epoch": 7.04, + "learning_rate": 3.3879620267385684e-05, + "loss": 0.2074, + "step": 30870 + }, + { + "epoch": 7.04, + "learning_rate": 3.387259858442872e-05, + "loss": 0.4827, + "step": 30880 + }, + { + "epoch": 7.04, + "learning_rate": 3.3865576901471745e-05, + "loss": 0.0012, + "step": 30890 + }, + { + "epoch": 7.04, + "learning_rate": 3.385855521851477e-05, + "loss": 0.0011, + "step": 30900 + }, + { + "epoch": 7.04, + "learning_rate": 3.385153353555781e-05, + "loss": 0.0005, + "step": 30910 + }, + { + "epoch": 7.04, + "learning_rate": 3.3844511852600835e-05, + "loss": 0.3459, + "step": 30920 + }, + { + "epoch": 7.04, + "learning_rate": 3.383749016964386e-05, + "loss": 0.027, + "step": 30930 + }, + { + "epoch": 7.04, + "learning_rate": 3.383046848668689e-05, + "loss": 0.0053, + "step": 30940 + }, + { + "epoch": 7.04, + "learning_rate": 3.382344680372992e-05, + "loss": 0.0538, + "step": 30950 + }, + { + "epoch": 7.04, + "learning_rate": 3.381642512077295e-05, + "loss": 0.0011, + "step": 30960 + }, + { + "epoch": 7.04, + "learning_rate": 3.380940343781598e-05, + "loss": 0.2906, + "step": 30970 + }, + { + "epoch": 7.04, + "learning_rate": 3.3802381754859006e-05, + "loss": 0.0006, + "step": 30980 + }, + { + "epoch": 7.04, + "learning_rate": 3.3795360071902034e-05, + "loss": 0.0017, + "step": 30990 + }, + { + "epoch": 7.04, + "learning_rate": 3.378833838894506e-05, + "loss": 0.3713, + "step": 31000 + }, + { + "epoch": 7.04, + "learning_rate": 3.378131670598809e-05, + "loss": 0.6157, + "step": 31010 + }, + { + "epoch": 7.04, + "learning_rate": 3.377429502303112e-05, + "loss": 0.0001, + "step": 31020 + }, + { + "epoch": 7.04, + "learning_rate": 3.376727334007415e-05, + "loss": 0.0007, + "step": 31030 + }, + { + "epoch": 7.04, + "learning_rate": 3.376025165711718e-05, + "loss": 0.0005, + "step": 31040 + }, + { + "epoch": 7.04, + "learning_rate": 3.375322997416021e-05, + "loss": 0.0061, + "step": 31050 + }, + { + "epoch": 7.04, + "learning_rate": 3.374620829120323e-05, + "loss": 0.4049, + "step": 31060 + }, + { + "epoch": 7.04, + "learning_rate": 3.373918660824627e-05, + "loss": 0.2725, + "step": 31070 + }, + { + "epoch": 7.04, + "learning_rate": 3.3732164925289295e-05, + "loss": 0.0035, + "step": 31080 + }, + { + "epoch": 7.04, + "learning_rate": 3.372514324233232e-05, + "loss": 0.4818, + "step": 31090 + }, + { + "epoch": 7.04, + "learning_rate": 3.371812155937536e-05, + "loss": 0.0001, + "step": 31100 + }, + { + "epoch": 7.04, + "learning_rate": 3.3711099876418384e-05, + "loss": 0.08, + "step": 31110 + }, + { + "epoch": 7.04, + "learning_rate": 3.3704078193461405e-05, + "loss": 0.0627, + "step": 31120 + }, + { + "epoch": 7.04, + "learning_rate": 3.369705651050444e-05, + "loss": 0.903, + "step": 31130 + }, + { + "epoch": 7.04, + "learning_rate": 3.3690034827547467e-05, + "loss": 0.0669, + "step": 31140 + }, + { + "epoch": 7.04, + "learning_rate": 3.3683013144590494e-05, + "loss": 0.0195, + "step": 31150 + }, + { + "epoch": 7.04, + "learning_rate": 3.367599146163353e-05, + "loss": 0.0003, + "step": 31160 + }, + { + "epoch": 7.04, + "learning_rate": 3.3668969778676556e-05, + "loss": 0.0059, + "step": 31170 + }, + { + "epoch": 7.04, + "learning_rate": 3.366194809571958e-05, + "loss": 0.0441, + "step": 31180 + }, + { + "epoch": 7.04, + "learning_rate": 3.365492641276261e-05, + "loss": 0.1429, + "step": 31190 + }, + { + "epoch": 7.04, + "learning_rate": 3.364790472980564e-05, + "loss": 0.6086, + "step": 31200 + }, + { + "epoch": 7.04, + "learning_rate": 3.364088304684867e-05, + "loss": 0.4145, + "step": 31210 + }, + { + "epoch": 7.04, + "learning_rate": 3.36338613638917e-05, + "loss": 0.2412, + "step": 31220 + }, + { + "epoch": 7.04, + "learning_rate": 3.362683968093473e-05, + "loss": 1.3075, + "step": 31230 + }, + { + "epoch": 7.04, + "learning_rate": 3.361981799797776e-05, + "loss": 0.3457, + "step": 31240 + }, + { + "epoch": 7.04, + "learning_rate": 3.361279631502078e-05, + "loss": 0.341, + "step": 31250 + }, + { + "epoch": 7.05, + "learning_rate": 3.360577463206381e-05, + "loss": 0.0099, + "step": 31260 + }, + { + "epoch": 7.05, + "learning_rate": 3.3598752949106844e-05, + "loss": 0.0019, + "step": 31270 + }, + { + "epoch": 7.05, + "learning_rate": 3.359173126614987e-05, + "loss": 0.0078, + "step": 31280 + }, + { + "epoch": 7.05, + "learning_rate": 3.35847095831929e-05, + "loss": 0.0506, + "step": 31290 + }, + { + "epoch": 7.05, + "learning_rate": 3.3577687900235934e-05, + "loss": 0.0056, + "step": 31300 + }, + { + "epoch": 7.05, + "learning_rate": 3.3570666217278954e-05, + "loss": 0.1082, + "step": 31310 + }, + { + "epoch": 7.05, + "learning_rate": 3.356364453432199e-05, + "loss": 0.0034, + "step": 31320 + }, + { + "epoch": 7.05, + "learning_rate": 3.3556622851365016e-05, + "loss": 0.0777, + "step": 31330 + }, + { + "epoch": 7.05, + "learning_rate": 3.3549601168408043e-05, + "loss": 0.0542, + "step": 31340 + }, + { + "epoch": 7.05, + "learning_rate": 3.354257948545108e-05, + "loss": 0.26, + "step": 31350 + }, + { + "epoch": 7.05, + "learning_rate": 3.3535557802494105e-05, + "loss": 0.0912, + "step": 31360 + }, + { + "epoch": 7.05, + "learning_rate": 3.352853611953713e-05, + "loss": 0.6288, + "step": 31370 + }, + { + "epoch": 7.05, + "learning_rate": 3.352151443658016e-05, + "loss": 0.0038, + "step": 31380 + }, + { + "epoch": 7.05, + "learning_rate": 3.351449275362319e-05, + "loss": 0.1041, + "step": 31390 + }, + { + "epoch": 7.05, + "learning_rate": 3.350747107066622e-05, + "loss": 0.0003, + "step": 31400 + }, + { + "epoch": 7.05, + "learning_rate": 3.350044938770925e-05, + "loss": 0.0782, + "step": 31410 + }, + { + "epoch": 7.05, + "learning_rate": 3.349342770475228e-05, + "loss": 0.2855, + "step": 31420 + }, + { + "epoch": 7.05, + "learning_rate": 3.348640602179531e-05, + "loss": 0.0012, + "step": 31430 + }, + { + "epoch": 7.05, + "learning_rate": 3.347938433883833e-05, + "loss": 0.0054, + "step": 31440 + }, + { + "epoch": 7.05, + "learning_rate": 3.347236265588136e-05, + "loss": 0.1221, + "step": 31450 + }, + { + "epoch": 7.05, + "learning_rate": 3.3465340972924394e-05, + "loss": 1.0928, + "step": 31460 + }, + { + "epoch": 7.05, + "learning_rate": 3.345831928996742e-05, + "loss": 0.1192, + "step": 31470 + }, + { + "epoch": 7.05, + "learning_rate": 3.345129760701045e-05, + "loss": 0.0688, + "step": 31480 + }, + { + "epoch": 7.05, + "learning_rate": 3.344427592405348e-05, + "loss": 0.0117, + "step": 31490 + }, + { + "epoch": 7.05, + "learning_rate": 3.3437254241096504e-05, + "loss": 0.0154, + "step": 31500 + }, + { + "epoch": 7.05, + "learning_rate": 3.343023255813954e-05, + "loss": 0.0159, + "step": 31510 + }, + { + "epoch": 7.05, + "learning_rate": 3.3423210875182565e-05, + "loss": 0.0077, + "step": 31520 + }, + { + "epoch": 7.05, + "learning_rate": 3.341618919222559e-05, + "loss": 0.0002, + "step": 31530 + }, + { + "epoch": 7.05, + "learning_rate": 3.340916750926863e-05, + "loss": 0.0025, + "step": 31540 + }, + { + "epoch": 7.05, + "learning_rate": 3.3402145826311655e-05, + "loss": 0.0001, + "step": 31550 + }, + { + "epoch": 7.05, + "learning_rate": 3.339512414335468e-05, + "loss": 0.0069, + "step": 31560 + }, + { + "epoch": 7.05, + "learning_rate": 3.338810246039771e-05, + "loss": 0.3249, + "step": 31570 + }, + { + "epoch": 7.05, + "learning_rate": 3.338108077744074e-05, + "loss": 0.0563, + "step": 31580 + }, + { + "epoch": 7.05, + "learning_rate": 3.3374059094483765e-05, + "loss": 0.0033, + "step": 31590 + }, + { + "epoch": 7.05, + "learning_rate": 3.33670374115268e-05, + "loss": 0.2022, + "step": 31600 + }, + { + "epoch": 7.05, + "learning_rate": 3.3360015728569826e-05, + "loss": 0.2046, + "step": 31610 + }, + { + "epoch": 7.05, + "learning_rate": 3.3352994045612854e-05, + "loss": 0.0058, + "step": 31620 + }, + { + "epoch": 7.05, + "learning_rate": 3.334597236265588e-05, + "loss": 0.1427, + "step": 31630 + }, + { + "epoch": 7.05, + "learning_rate": 3.333895067969891e-05, + "loss": 0.0001, + "step": 31640 + }, + { + "epoch": 7.05, + "eval_accuracy": 0.7481615275448329, + "eval_loss": 1.5946505069732666, + "eval_runtime": 1276.1937, + "eval_samples_per_second": 6.074, + "eval_steps_per_second": 3.037, + "step": 31648 + }, + { + "epoch": 8.0, + "learning_rate": 3.333192899674194e-05, + "loss": 0.5077, + "step": 31650 + }, + { + "epoch": 8.0, + "learning_rate": 3.332490731378497e-05, + "loss": 0.0003, + "step": 31660 + }, + { + "epoch": 8.0, + "learning_rate": 3.3317885630828e-05, + "loss": 0.0407, + "step": 31670 + }, + { + "epoch": 8.0, + "learning_rate": 3.331086394787103e-05, + "loss": 0.3532, + "step": 31680 + }, + { + "epoch": 8.0, + "learning_rate": 3.330384226491405e-05, + "loss": 0.1548, + "step": 31690 + }, + { + "epoch": 8.0, + "learning_rate": 3.329682058195708e-05, + "loss": 0.2169, + "step": 31700 + }, + { + "epoch": 8.0, + "learning_rate": 3.3289798899000115e-05, + "loss": 0.2426, + "step": 31710 + }, + { + "epoch": 8.0, + "learning_rate": 3.328277721604314e-05, + "loss": 0.3527, + "step": 31720 + }, + { + "epoch": 8.0, + "learning_rate": 3.327575553308617e-05, + "loss": 0.0006, + "step": 31730 + }, + { + "epoch": 8.0, + "learning_rate": 3.3268733850129204e-05, + "loss": 0.0008, + "step": 31740 + }, + { + "epoch": 8.0, + "learning_rate": 3.326171216717223e-05, + "loss": 1.0696, + "step": 31750 + }, + { + "epoch": 8.0, + "learning_rate": 3.325469048421526e-05, + "loss": 0.0213, + "step": 31760 + }, + { + "epoch": 8.0, + "learning_rate": 3.3247668801258287e-05, + "loss": 0.0971, + "step": 31770 + }, + { + "epoch": 8.0, + "learning_rate": 3.3240647118301314e-05, + "loss": 0.0022, + "step": 31780 + }, + { + "epoch": 8.0, + "learning_rate": 3.323362543534435e-05, + "loss": 0.0013, + "step": 31790 + }, + { + "epoch": 8.0, + "learning_rate": 3.3226603752387376e-05, + "loss": 0.1716, + "step": 31800 + }, + { + "epoch": 8.0, + "learning_rate": 3.32195820694304e-05, + "loss": 0.167, + "step": 31810 + }, + { + "epoch": 8.0, + "learning_rate": 3.321256038647343e-05, + "loss": 0.4907, + "step": 31820 + }, + { + "epoch": 8.0, + "learning_rate": 3.320553870351646e-05, + "loss": 0.0035, + "step": 31830 + }, + { + "epoch": 8.0, + "learning_rate": 3.3198517020559486e-05, + "loss": 0.0085, + "step": 31840 + }, + { + "epoch": 8.0, + "learning_rate": 3.319149533760252e-05, + "loss": 0.1655, + "step": 31850 + }, + { + "epoch": 8.0, + "learning_rate": 3.318447365464555e-05, + "loss": 0.0009, + "step": 31860 + }, + { + "epoch": 8.0, + "learning_rate": 3.3177451971688575e-05, + "loss": 0.1614, + "step": 31870 + }, + { + "epoch": 8.0, + "learning_rate": 3.31704302887316e-05, + "loss": 0.0012, + "step": 31880 + }, + { + "epoch": 8.0, + "learning_rate": 3.316340860577463e-05, + "loss": 0.1424, + "step": 31890 + }, + { + "epoch": 8.0, + "learning_rate": 3.3156386922817664e-05, + "loss": 0.0001, + "step": 31900 + }, + { + "epoch": 8.0, + "learning_rate": 3.314936523986069e-05, + "loss": 0.0004, + "step": 31910 + }, + { + "epoch": 8.0, + "learning_rate": 3.314234355690372e-05, + "loss": 0.0713, + "step": 31920 + }, + { + "epoch": 8.0, + "learning_rate": 3.3135321873946754e-05, + "loss": 1.1175, + "step": 31930 + }, + { + "epoch": 8.0, + "learning_rate": 3.312830019098978e-05, + "loss": 0.2867, + "step": 31940 + }, + { + "epoch": 8.0, + "learning_rate": 3.31212785080328e-05, + "loss": 0.3437, + "step": 31950 + }, + { + "epoch": 8.0, + "learning_rate": 3.3114256825075836e-05, + "loss": 0.0109, + "step": 31960 + }, + { + "epoch": 8.0, + "learning_rate": 3.3107235142118863e-05, + "loss": 0.0041, + "step": 31970 + }, + { + "epoch": 8.0, + "learning_rate": 3.310021345916189e-05, + "loss": 0.4439, + "step": 31980 + }, + { + "epoch": 8.0, + "learning_rate": 3.3093191776204925e-05, + "loss": 0.5916, + "step": 31990 + }, + { + "epoch": 8.0, + "learning_rate": 3.308617009324795e-05, + "loss": 0.0353, + "step": 32000 + }, + { + "epoch": 8.0, + "learning_rate": 3.307914841029098e-05, + "loss": 0.0104, + "step": 32010 + }, + { + "epoch": 8.0, + "learning_rate": 3.307212672733401e-05, + "loss": 0.0096, + "step": 32020 + }, + { + "epoch": 8.0, + "learning_rate": 3.3065105044377035e-05, + "loss": 0.0361, + "step": 32030 + }, + { + "epoch": 8.0, + "learning_rate": 3.305808336142007e-05, + "loss": 0.0051, + "step": 32040 + }, + { + "epoch": 8.01, + "learning_rate": 3.30510616784631e-05, + "loss": 0.4574, + "step": 32050 + }, + { + "epoch": 8.01, + "learning_rate": 3.3044039995506124e-05, + "loss": 0.001, + "step": 32060 + }, + { + "epoch": 8.01, + "learning_rate": 3.303701831254916e-05, + "loss": 0.0013, + "step": 32070 + }, + { + "epoch": 8.01, + "learning_rate": 3.302999662959218e-05, + "loss": 0.0252, + "step": 32080 + }, + { + "epoch": 8.01, + "learning_rate": 3.302297494663521e-05, + "loss": 0.0001, + "step": 32090 + }, + { + "epoch": 8.01, + "learning_rate": 3.301595326367824e-05, + "loss": 0.2373, + "step": 32100 + }, + { + "epoch": 8.01, + "learning_rate": 3.300893158072127e-05, + "loss": 0.0004, + "step": 32110 + }, + { + "epoch": 8.01, + "learning_rate": 3.3001909897764296e-05, + "loss": 0.0003, + "step": 32120 + }, + { + "epoch": 8.01, + "learning_rate": 3.299488821480733e-05, + "loss": 0.1445, + "step": 32130 + }, + { + "epoch": 8.01, + "learning_rate": 3.298786653185035e-05, + "loss": 0.0233, + "step": 32140 + }, + { + "epoch": 8.01, + "learning_rate": 3.2980844848893385e-05, + "loss": 0.2702, + "step": 32150 + }, + { + "epoch": 8.01, + "learning_rate": 3.297382316593641e-05, + "loss": 0.3014, + "step": 32160 + }, + { + "epoch": 8.01, + "learning_rate": 3.296680148297944e-05, + "loss": 0.5491, + "step": 32170 + }, + { + "epoch": 8.01, + "learning_rate": 3.2959779800022475e-05, + "loss": 0.0228, + "step": 32180 + }, + { + "epoch": 8.01, + "learning_rate": 3.29527581170655e-05, + "loss": 0.5336, + "step": 32190 + }, + { + "epoch": 8.01, + "learning_rate": 3.294573643410852e-05, + "loss": 0.177, + "step": 32200 + }, + { + "epoch": 8.01, + "learning_rate": 3.293871475115156e-05, + "loss": 0.008, + "step": 32210 + }, + { + "epoch": 8.01, + "learning_rate": 3.2931693068194585e-05, + "loss": 0.0047, + "step": 32220 + }, + { + "epoch": 8.01, + "learning_rate": 3.292467138523761e-05, + "loss": 0.1882, + "step": 32230 + }, + { + "epoch": 8.01, + "learning_rate": 3.2917649702280646e-05, + "loss": 0.6219, + "step": 32240 + }, + { + "epoch": 8.01, + "learning_rate": 3.2910628019323674e-05, + "loss": 0.0625, + "step": 32250 + }, + { + "epoch": 8.01, + "learning_rate": 3.29036063363667e-05, + "loss": 0.0276, + "step": 32260 + }, + { + "epoch": 8.01, + "learning_rate": 3.289658465340973e-05, + "loss": 0.0005, + "step": 32270 + }, + { + "epoch": 8.01, + "learning_rate": 3.2889562970452756e-05, + "loss": 0.016, + "step": 32280 + }, + { + "epoch": 8.01, + "learning_rate": 3.288254128749579e-05, + "loss": 0.3161, + "step": 32290 + }, + { + "epoch": 8.01, + "learning_rate": 3.287551960453882e-05, + "loss": 0.0585, + "step": 32300 + }, + { + "epoch": 8.01, + "learning_rate": 3.2868497921581846e-05, + "loss": 0.0295, + "step": 32310 + }, + { + "epoch": 8.01, + "learning_rate": 3.286147623862488e-05, + "loss": 0.0315, + "step": 32320 + }, + { + "epoch": 8.01, + "learning_rate": 3.28544545556679e-05, + "loss": 0.0044, + "step": 32330 + }, + { + "epoch": 8.01, + "learning_rate": 3.2847432872710935e-05, + "loss": 0.0001, + "step": 32340 + }, + { + "epoch": 8.01, + "learning_rate": 3.284041118975396e-05, + "loss": 0.0086, + "step": 32350 + }, + { + "epoch": 8.01, + "learning_rate": 3.283338950679699e-05, + "loss": 0.0023, + "step": 32360 + }, + { + "epoch": 8.01, + "learning_rate": 3.2826367823840024e-05, + "loss": 0.0037, + "step": 32370 + }, + { + "epoch": 8.01, + "learning_rate": 3.281934614088305e-05, + "loss": 0.3903, + "step": 32380 + }, + { + "epoch": 8.01, + "learning_rate": 3.281232445792607e-05, + "loss": 0.1154, + "step": 32390 + }, + { + "epoch": 8.01, + "learning_rate": 3.2805302774969107e-05, + "loss": 0.0048, + "step": 32400 + }, + { + "epoch": 8.01, + "learning_rate": 3.2798281092012134e-05, + "loss": 0.0002, + "step": 32410 + }, + { + "epoch": 8.01, + "learning_rate": 3.279125940905516e-05, + "loss": 0.0003, + "step": 32420 + }, + { + "epoch": 8.01, + "learning_rate": 3.2784237726098196e-05, + "loss": 0.4289, + "step": 32430 + }, + { + "epoch": 8.01, + "learning_rate": 3.277721604314122e-05, + "loss": 0.0001, + "step": 32440 + }, + { + "epoch": 8.01, + "learning_rate": 3.277019436018425e-05, + "loss": 0.2983, + "step": 32450 + }, + { + "epoch": 8.01, + "learning_rate": 3.276317267722728e-05, + "loss": 0.5355, + "step": 32460 + }, + { + "epoch": 8.01, + "learning_rate": 3.2756150994270306e-05, + "loss": 0.3568, + "step": 32470 + }, + { + "epoch": 8.01, + "learning_rate": 3.274912931131334e-05, + "loss": 0.0047, + "step": 32480 + }, + { + "epoch": 8.01, + "learning_rate": 3.274210762835637e-05, + "loss": 0.0986, + "step": 32490 + }, + { + "epoch": 8.01, + "learning_rate": 3.2735085945399395e-05, + "loss": 0.0007, + "step": 32500 + }, + { + "epoch": 8.01, + "learning_rate": 3.272806426244243e-05, + "loss": 0.008, + "step": 32510 + }, + { + "epoch": 8.01, + "learning_rate": 3.272104257948545e-05, + "loss": 0.1386, + "step": 32520 + }, + { + "epoch": 8.01, + "learning_rate": 3.271402089652848e-05, + "loss": 0.0001, + "step": 32530 + }, + { + "epoch": 8.01, + "learning_rate": 3.270699921357151e-05, + "loss": 0.0058, + "step": 32540 + }, + { + "epoch": 8.01, + "learning_rate": 3.269997753061454e-05, + "loss": 0.2799, + "step": 32550 + }, + { + "epoch": 8.01, + "learning_rate": 3.269295584765757e-05, + "loss": 0.6197, + "step": 32560 + }, + { + "epoch": 8.01, + "learning_rate": 3.26859341647006e-05, + "loss": 0.0309, + "step": 32570 + }, + { + "epoch": 8.01, + "learning_rate": 3.267891248174362e-05, + "loss": 0.0002, + "step": 32580 + }, + { + "epoch": 8.01, + "learning_rate": 3.2671890798786656e-05, + "loss": 0.142, + "step": 32590 + }, + { + "epoch": 8.01, + "learning_rate": 3.2664869115829683e-05, + "loss": 0.0046, + "step": 32600 + }, + { + "epoch": 8.01, + "learning_rate": 3.265784743287271e-05, + "loss": 0.4663, + "step": 32610 + }, + { + "epoch": 8.01, + "learning_rate": 3.2650825749915745e-05, + "loss": 0.0811, + "step": 32620 + }, + { + "epoch": 8.01, + "learning_rate": 3.264380406695877e-05, + "loss": 0.0014, + "step": 32630 + }, + { + "epoch": 8.01, + "learning_rate": 3.26367823840018e-05, + "loss": 0.1216, + "step": 32640 + }, + { + "epoch": 8.01, + "learning_rate": 3.262976070104483e-05, + "loss": 0.9207, + "step": 32650 + }, + { + "epoch": 8.01, + "learning_rate": 3.2622739018087855e-05, + "loss": 0.0074, + "step": 32660 + }, + { + "epoch": 8.01, + "learning_rate": 3.261571733513088e-05, + "loss": 0.5065, + "step": 32670 + }, + { + "epoch": 8.01, + "learning_rate": 3.260869565217392e-05, + "loss": 0.009, + "step": 32680 + }, + { + "epoch": 8.01, + "learning_rate": 3.2601673969216944e-05, + "loss": 0.2742, + "step": 32690 + }, + { + "epoch": 8.01, + "learning_rate": 3.259465228625997e-05, + "loss": 0.0043, + "step": 32700 + }, + { + "epoch": 8.01, + "learning_rate": 3.2587630603303e-05, + "loss": 0.2973, + "step": 32710 + }, + { + "epoch": 8.01, + "learning_rate": 3.258060892034603e-05, + "loss": 0.0843, + "step": 32720 + }, + { + "epoch": 8.01, + "learning_rate": 3.257358723738906e-05, + "loss": 0.1155, + "step": 32730 + }, + { + "epoch": 8.01, + "learning_rate": 3.256656555443209e-05, + "loss": 0.0505, + "step": 32740 + }, + { + "epoch": 8.01, + "learning_rate": 3.2559543871475116e-05, + "loss": 0.1602, + "step": 32750 + }, + { + "epoch": 8.01, + "learning_rate": 3.255252218851815e-05, + "loss": 0.0004, + "step": 32760 + }, + { + "epoch": 8.01, + "learning_rate": 3.254550050556117e-05, + "loss": 0.0005, + "step": 32770 + }, + { + "epoch": 8.01, + "learning_rate": 3.25384788226042e-05, + "loss": 0.0172, + "step": 32780 + }, + { + "epoch": 8.01, + "learning_rate": 3.253145713964723e-05, + "loss": 0.0172, + "step": 32790 + }, + { + "epoch": 8.01, + "learning_rate": 3.252443545669026e-05, + "loss": 0.3917, + "step": 32800 + }, + { + "epoch": 8.01, + "learning_rate": 3.251741377373329e-05, + "loss": 0.0099, + "step": 32810 + }, + { + "epoch": 8.01, + "learning_rate": 3.251039209077632e-05, + "loss": 0.0009, + "step": 32820 + }, + { + "epoch": 8.01, + "learning_rate": 3.250337040781935e-05, + "loss": 0.4246, + "step": 32830 + }, + { + "epoch": 8.02, + "learning_rate": 3.249634872486238e-05, + "loss": 0.0062, + "step": 32840 + }, + { + "epoch": 8.02, + "learning_rate": 3.2489327041905405e-05, + "loss": 0.0744, + "step": 32850 + }, + { + "epoch": 8.02, + "learning_rate": 3.248230535894843e-05, + "loss": 0.2148, + "step": 32860 + }, + { + "epoch": 8.02, + "learning_rate": 3.2475283675991466e-05, + "loss": 0.0085, + "step": 32870 + }, + { + "epoch": 8.02, + "learning_rate": 3.2468261993034494e-05, + "loss": 0.0008, + "step": 32880 + }, + { + "epoch": 8.02, + "learning_rate": 3.246124031007752e-05, + "loss": 0.0003, + "step": 32890 + }, + { + "epoch": 8.02, + "learning_rate": 3.245421862712055e-05, + "loss": 0.7163, + "step": 32900 + }, + { + "epoch": 8.02, + "learning_rate": 3.2447196944163576e-05, + "loss": 0.2547, + "step": 32910 + }, + { + "epoch": 8.02, + "learning_rate": 3.2440175261206604e-05, + "loss": 0.0059, + "step": 32920 + }, + { + "epoch": 8.02, + "learning_rate": 3.243315357824964e-05, + "loss": 0.184, + "step": 32930 + }, + { + "epoch": 8.02, + "learning_rate": 3.2426131895292666e-05, + "loss": 0.4133, + "step": 32940 + }, + { + "epoch": 8.02, + "learning_rate": 3.241911021233569e-05, + "loss": 0.0319, + "step": 32950 + }, + { + "epoch": 8.02, + "learning_rate": 3.241208852937873e-05, + "loss": 0.4454, + "step": 32960 + }, + { + "epoch": 8.02, + "learning_rate": 3.240506684642175e-05, + "loss": 0.1975, + "step": 32970 + }, + { + "epoch": 8.02, + "learning_rate": 3.239804516346478e-05, + "loss": 0.003, + "step": 32980 + }, + { + "epoch": 8.02, + "learning_rate": 3.239102348050781e-05, + "loss": 0.4202, + "step": 32990 + }, + { + "epoch": 8.02, + "learning_rate": 3.238400179755084e-05, + "loss": 0.002, + "step": 33000 + }, + { + "epoch": 8.02, + "learning_rate": 3.237698011459387e-05, + "loss": 0.4502, + "step": 33010 + }, + { + "epoch": 8.02, + "learning_rate": 3.23699584316369e-05, + "loss": 0.0004, + "step": 33020 + }, + { + "epoch": 8.02, + "learning_rate": 3.236293674867992e-05, + "loss": 0.37, + "step": 33030 + }, + { + "epoch": 8.02, + "learning_rate": 3.2355915065722954e-05, + "loss": 0.0, + "step": 33040 + }, + { + "epoch": 8.02, + "learning_rate": 3.234889338276598e-05, + "loss": 0.2142, + "step": 33050 + }, + { + "epoch": 8.02, + "learning_rate": 3.234187169980901e-05, + "loss": 0.0223, + "step": 33060 + }, + { + "epoch": 8.02, + "learning_rate": 3.233485001685204e-05, + "loss": 0.0001, + "step": 33070 + }, + { + "epoch": 8.02, + "learning_rate": 3.232782833389507e-05, + "loss": 0.0406, + "step": 33080 + }, + { + "epoch": 8.02, + "learning_rate": 3.23208066509381e-05, + "loss": 0.0261, + "step": 33090 + }, + { + "epoch": 8.02, + "learning_rate": 3.2313784967981126e-05, + "loss": 0.0021, + "step": 33100 + }, + { + "epoch": 8.02, + "learning_rate": 3.230676328502415e-05, + "loss": 0.1187, + "step": 33110 + }, + { + "epoch": 8.02, + "learning_rate": 3.229974160206719e-05, + "loss": 0.315, + "step": 33120 + }, + { + "epoch": 8.02, + "learning_rate": 3.2292719919110215e-05, + "loss": 0.0007, + "step": 33130 + }, + { + "epoch": 8.02, + "learning_rate": 3.228569823615324e-05, + "loss": 0.4922, + "step": 33140 + }, + { + "epoch": 8.02, + "learning_rate": 3.227867655319628e-05, + "loss": 0.0003, + "step": 33150 + }, + { + "epoch": 8.02, + "learning_rate": 3.22716548702393e-05, + "loss": 0.0397, + "step": 33160 + }, + { + "epoch": 8.02, + "learning_rate": 3.2264633187282325e-05, + "loss": 0.0723, + "step": 33170 + }, + { + "epoch": 8.02, + "learning_rate": 3.225761150432536e-05, + "loss": 0.1389, + "step": 33180 + }, + { + "epoch": 8.02, + "learning_rate": 3.225058982136839e-05, + "loss": 0.0001, + "step": 33190 + }, + { + "epoch": 8.02, + "learning_rate": 3.2243568138411414e-05, + "loss": 0.0067, + "step": 33200 + }, + { + "epoch": 8.02, + "learning_rate": 3.223654645545445e-05, + "loss": 0.0129, + "step": 33210 + }, + { + "epoch": 8.02, + "learning_rate": 3.222952477249747e-05, + "loss": 0.2119, + "step": 33220 + }, + { + "epoch": 8.02, + "learning_rate": 3.2222503089540503e-05, + "loss": 0.0354, + "step": 33230 + }, + { + "epoch": 8.02, + "learning_rate": 3.221548140658353e-05, + "loss": 0.33, + "step": 33240 + }, + { + "epoch": 8.02, + "learning_rate": 3.220845972362656e-05, + "loss": 0.2476, + "step": 33250 + }, + { + "epoch": 8.02, + "learning_rate": 3.220143804066959e-05, + "loss": 0.0081, + "step": 33260 + }, + { + "epoch": 8.02, + "learning_rate": 3.219441635771262e-05, + "loss": 0.0292, + "step": 33270 + }, + { + "epoch": 8.02, + "learning_rate": 3.218739467475564e-05, + "loss": 0.0345, + "step": 33280 + }, + { + "epoch": 8.02, + "learning_rate": 3.2180372991798675e-05, + "loss": 0.0834, + "step": 33290 + }, + { + "epoch": 8.02, + "learning_rate": 3.21733513088417e-05, + "loss": 0.002, + "step": 33300 + }, + { + "epoch": 8.02, + "learning_rate": 3.216632962588473e-05, + "loss": 0.2443, + "step": 33310 + }, + { + "epoch": 8.02, + "learning_rate": 3.2159307942927764e-05, + "loss": 0.0003, + "step": 33320 + }, + { + "epoch": 8.02, + "learning_rate": 3.215228625997079e-05, + "loss": 0.0032, + "step": 33330 + }, + { + "epoch": 8.02, + "learning_rate": 3.214526457701382e-05, + "loss": 0.2329, + "step": 33340 + }, + { + "epoch": 8.02, + "learning_rate": 3.213824289405685e-05, + "loss": 0.1568, + "step": 33350 + }, + { + "epoch": 8.02, + "learning_rate": 3.2131221211099874e-05, + "loss": 0.0048, + "step": 33360 + }, + { + "epoch": 8.02, + "learning_rate": 3.212419952814291e-05, + "loss": 0.2933, + "step": 33370 + }, + { + "epoch": 8.02, + "learning_rate": 3.2117177845185936e-05, + "loss": 0.0412, + "step": 33380 + }, + { + "epoch": 8.02, + "learning_rate": 3.2110156162228964e-05, + "loss": 0.0005, + "step": 33390 + }, + { + "epoch": 8.02, + "learning_rate": 3.2103134479272e-05, + "loss": 0.5077, + "step": 33400 + }, + { + "epoch": 8.02, + "learning_rate": 3.209611279631502e-05, + "loss": 0.1549, + "step": 33410 + }, + { + "epoch": 8.02, + "learning_rate": 3.208909111335805e-05, + "loss": 0.002, + "step": 33420 + }, + { + "epoch": 8.02, + "learning_rate": 3.208206943040108e-05, + "loss": 1.7583, + "step": 33430 + }, + { + "epoch": 8.02, + "learning_rate": 3.207504774744411e-05, + "loss": 0.1903, + "step": 33440 + }, + { + "epoch": 8.02, + "learning_rate": 3.206802606448714e-05, + "loss": 0.001, + "step": 33450 + }, + { + "epoch": 8.02, + "learning_rate": 3.206100438153017e-05, + "loss": 0.0136, + "step": 33460 + }, + { + "epoch": 8.02, + "learning_rate": 3.205398269857319e-05, + "loss": 0.0001, + "step": 33470 + }, + { + "epoch": 8.02, + "learning_rate": 3.2046961015616225e-05, + "loss": 1.6762, + "step": 33480 + }, + { + "epoch": 8.02, + "learning_rate": 3.203993933265925e-05, + "loss": 0.003, + "step": 33490 + }, + { + "epoch": 8.02, + "learning_rate": 3.203291764970228e-05, + "loss": 0.0116, + "step": 33500 + }, + { + "epoch": 8.02, + "learning_rate": 3.2025895966745314e-05, + "loss": 0.0003, + "step": 33510 + }, + { + "epoch": 8.02, + "learning_rate": 3.201887428378834e-05, + "loss": 0.2125, + "step": 33520 + }, + { + "epoch": 8.02, + "learning_rate": 3.201185260083137e-05, + "loss": 0.0029, + "step": 33530 + }, + { + "epoch": 8.02, + "learning_rate": 3.2004830917874396e-05, + "loss": 0.1889, + "step": 33540 + }, + { + "epoch": 8.02, + "learning_rate": 3.1997809234917424e-05, + "loss": 0.0002, + "step": 33550 + }, + { + "epoch": 8.02, + "learning_rate": 3.199078755196046e-05, + "loss": 0.7915, + "step": 33560 + }, + { + "epoch": 8.02, + "learning_rate": 3.1983765869003486e-05, + "loss": 0.0717, + "step": 33570 + }, + { + "epoch": 8.02, + "learning_rate": 3.197674418604651e-05, + "loss": 0.5634, + "step": 33580 + }, + { + "epoch": 8.02, + "learning_rate": 3.196972250308955e-05, + "loss": 0.0001, + "step": 33590 + }, + { + "epoch": 8.02, + "learning_rate": 3.196270082013257e-05, + "loss": 0.1022, + "step": 33600 + }, + { + "epoch": 8.02, + "learning_rate": 3.1955679137175596e-05, + "loss": 0.1102, + "step": 33610 + }, + { + "epoch": 8.02, + "learning_rate": 3.194865745421863e-05, + "loss": 0.1985, + "step": 33620 + }, + { + "epoch": 8.03, + "learning_rate": 3.194163577126166e-05, + "loss": 0.0001, + "step": 33630 + }, + { + "epoch": 8.03, + "learning_rate": 3.1934614088304685e-05, + "loss": 0.1628, + "step": 33640 + }, + { + "epoch": 8.03, + "learning_rate": 3.192759240534772e-05, + "loss": 0.6929, + "step": 33650 + }, + { + "epoch": 8.03, + "learning_rate": 3.1920570722390747e-05, + "loss": 0.0148, + "step": 33660 + }, + { + "epoch": 8.03, + "learning_rate": 3.1913549039433774e-05, + "loss": 0.7625, + "step": 33670 + }, + { + "epoch": 8.03, + "learning_rate": 3.19065273564768e-05, + "loss": 0.7876, + "step": 33680 + }, + { + "epoch": 8.03, + "learning_rate": 3.189950567351983e-05, + "loss": 0.0839, + "step": 33690 + }, + { + "epoch": 8.03, + "learning_rate": 3.189248399056286e-05, + "loss": 0.1266, + "step": 33700 + }, + { + "epoch": 8.03, + "learning_rate": 3.188546230760589e-05, + "loss": 0.8669, + "step": 33710 + }, + { + "epoch": 8.03, + "learning_rate": 3.187844062464892e-05, + "loss": 0.0002, + "step": 33720 + }, + { + "epoch": 8.03, + "learning_rate": 3.1871418941691946e-05, + "loss": 0.0309, + "step": 33730 + }, + { + "epoch": 8.03, + "learning_rate": 3.186439725873497e-05, + "loss": 0.2499, + "step": 33740 + }, + { + "epoch": 8.03, + "learning_rate": 3.1857375575778e-05, + "loss": 0.0693, + "step": 33750 + }, + { + "epoch": 8.03, + "learning_rate": 3.1850353892821035e-05, + "loss": 0.0428, + "step": 33760 + }, + { + "epoch": 8.03, + "learning_rate": 3.184333220986406e-05, + "loss": 0.0012, + "step": 33770 + }, + { + "epoch": 8.03, + "learning_rate": 3.183631052690709e-05, + "loss": 0.5643, + "step": 33780 + }, + { + "epoch": 8.03, + "learning_rate": 3.182928884395012e-05, + "loss": 0.0003, + "step": 33790 + }, + { + "epoch": 8.03, + "learning_rate": 3.1822267160993145e-05, + "loss": 0.0003, + "step": 33800 + }, + { + "epoch": 8.03, + "learning_rate": 3.181524547803618e-05, + "loss": 0.026, + "step": 33810 + }, + { + "epoch": 8.03, + "learning_rate": 3.180822379507921e-05, + "loss": 0.0037, + "step": 33820 + }, + { + "epoch": 8.03, + "learning_rate": 3.1801202112122234e-05, + "loss": 0.0001, + "step": 33830 + }, + { + "epoch": 8.03, + "learning_rate": 3.179418042916527e-05, + "loss": 0.0006, + "step": 33840 + }, + { + "epoch": 8.03, + "learning_rate": 3.1787158746208296e-05, + "loss": 0.286, + "step": 33850 + }, + { + "epoch": 8.03, + "learning_rate": 3.178013706325132e-05, + "loss": 0.0001, + "step": 33860 + }, + { + "epoch": 8.03, + "learning_rate": 3.177311538029435e-05, + "loss": 0.0004, + "step": 33870 + }, + { + "epoch": 8.03, + "learning_rate": 3.176609369733738e-05, + "loss": 0.0008, + "step": 33880 + }, + { + "epoch": 8.03, + "learning_rate": 3.1759072014380406e-05, + "loss": 0.0003, + "step": 33890 + }, + { + "epoch": 8.03, + "learning_rate": 3.175205033142344e-05, + "loss": 0.107, + "step": 33900 + }, + { + "epoch": 8.03, + "learning_rate": 3.174502864846647e-05, + "loss": 0.0018, + "step": 33910 + }, + { + "epoch": 8.03, + "learning_rate": 3.1738006965509495e-05, + "loss": 0.1424, + "step": 33920 + }, + { + "epoch": 8.03, + "learning_rate": 3.173098528255252e-05, + "loss": 0.3862, + "step": 33930 + }, + { + "epoch": 8.03, + "learning_rate": 3.172396359959555e-05, + "loss": 0.2243, + "step": 33940 + }, + { + "epoch": 8.03, + "learning_rate": 3.1716941916638584e-05, + "loss": 0.0001, + "step": 33950 + }, + { + "epoch": 8.03, + "learning_rate": 3.170992023368161e-05, + "loss": 0.4445, + "step": 33960 + }, + { + "epoch": 8.03, + "learning_rate": 3.170289855072464e-05, + "loss": 0.0016, + "step": 33970 + }, + { + "epoch": 8.03, + "learning_rate": 3.169587686776767e-05, + "loss": 0.3909, + "step": 33980 + }, + { + "epoch": 8.03, + "learning_rate": 3.1688855184810694e-05, + "loss": 0.1209, + "step": 33990 + }, + { + "epoch": 8.03, + "learning_rate": 3.168183350185372e-05, + "loss": 0.0027, + "step": 34000 + }, + { + "epoch": 8.03, + "learning_rate": 3.1674811818896756e-05, + "loss": 0.0916, + "step": 34010 + }, + { + "epoch": 8.03, + "learning_rate": 3.1667790135939784e-05, + "loss": 0.0056, + "step": 34020 + }, + { + "epoch": 8.03, + "learning_rate": 3.166076845298281e-05, + "loss": 0.0311, + "step": 34030 + }, + { + "epoch": 8.03, + "learning_rate": 3.1653746770025845e-05, + "loss": 0.0036, + "step": 34040 + }, + { + "epoch": 8.03, + "learning_rate": 3.1646725087068866e-05, + "loss": 0.0181, + "step": 34050 + }, + { + "epoch": 8.03, + "learning_rate": 3.16397034041119e-05, + "loss": 0.6601, + "step": 34060 + }, + { + "epoch": 8.03, + "learning_rate": 3.163268172115493e-05, + "loss": 0.5842, + "step": 34070 + }, + { + "epoch": 8.03, + "learning_rate": 3.1625660038197955e-05, + "loss": 0.0002, + "step": 34080 + }, + { + "epoch": 8.03, + "learning_rate": 3.161863835524099e-05, + "loss": 0.0009, + "step": 34090 + }, + { + "epoch": 8.03, + "learning_rate": 3.161161667228402e-05, + "loss": 0.1228, + "step": 34100 + }, + { + "epoch": 8.03, + "learning_rate": 3.160459498932704e-05, + "loss": 0.1574, + "step": 34110 + }, + { + "epoch": 8.03, + "learning_rate": 3.159757330637007e-05, + "loss": 0.3391, + "step": 34120 + }, + { + "epoch": 8.03, + "learning_rate": 3.15905516234131e-05, + "loss": 0.0007, + "step": 34130 + }, + { + "epoch": 8.03, + "learning_rate": 3.158352994045613e-05, + "loss": 0.129, + "step": 34140 + }, + { + "epoch": 8.03, + "learning_rate": 3.157650825749916e-05, + "loss": 0.0007, + "step": 34150 + }, + { + "epoch": 8.03, + "learning_rate": 3.156948657454219e-05, + "loss": 0.0716, + "step": 34160 + }, + { + "epoch": 8.03, + "learning_rate": 3.1562464891585216e-05, + "loss": 0.0002, + "step": 34170 + }, + { + "epoch": 8.03, + "learning_rate": 3.1555443208628244e-05, + "loss": 1.298, + "step": 34180 + }, + { + "epoch": 8.03, + "learning_rate": 3.154842152567127e-05, + "loss": 0.0621, + "step": 34190 + }, + { + "epoch": 8.03, + "learning_rate": 3.1541399842714306e-05, + "loss": 0.2694, + "step": 34200 + }, + { + "epoch": 8.03, + "learning_rate": 3.153437815975733e-05, + "loss": 0.0001, + "step": 34210 + }, + { + "epoch": 8.03, + "learning_rate": 3.152735647680036e-05, + "loss": 0.403, + "step": 34220 + }, + { + "epoch": 8.03, + "learning_rate": 3.1520334793843395e-05, + "loss": 0.0001, + "step": 34230 + }, + { + "epoch": 8.03, + "learning_rate": 3.1513313110886416e-05, + "loss": 0.0075, + "step": 34240 + }, + { + "epoch": 8.03, + "learning_rate": 3.150629142792944e-05, + "loss": 0.2443, + "step": 34250 + }, + { + "epoch": 8.03, + "learning_rate": 3.149926974497248e-05, + "loss": 0.4734, + "step": 34260 + }, + { + "epoch": 8.03, + "learning_rate": 3.1492248062015505e-05, + "loss": 0.1344, + "step": 34270 + }, + { + "epoch": 8.03, + "learning_rate": 3.148522637905853e-05, + "loss": 0.513, + "step": 34280 + }, + { + "epoch": 8.03, + "learning_rate": 3.1478204696101567e-05, + "loss": 0.0824, + "step": 34290 + }, + { + "epoch": 8.03, + "learning_rate": 3.147118301314459e-05, + "loss": 1.2133, + "step": 34300 + }, + { + "epoch": 8.03, + "learning_rate": 3.146416133018762e-05, + "loss": 0.0051, + "step": 34310 + }, + { + "epoch": 8.03, + "learning_rate": 3.145713964723065e-05, + "loss": 0.425, + "step": 34320 + }, + { + "epoch": 8.03, + "learning_rate": 3.1450117964273677e-05, + "loss": 0.0138, + "step": 34330 + }, + { + "epoch": 8.03, + "learning_rate": 3.144309628131671e-05, + "loss": 0.2434, + "step": 34340 + }, + { + "epoch": 8.03, + "learning_rate": 3.143607459835974e-05, + "loss": 0.0164, + "step": 34350 + }, + { + "epoch": 8.03, + "learning_rate": 3.1429052915402766e-05, + "loss": 0.0016, + "step": 34360 + }, + { + "epoch": 8.03, + "learning_rate": 3.142203123244579e-05, + "loss": 0.0623, + "step": 34370 + }, + { + "epoch": 8.03, + "learning_rate": 3.141500954948882e-05, + "loss": 0.0312, + "step": 34380 + }, + { + "epoch": 8.03, + "learning_rate": 3.1407987866531855e-05, + "loss": 0.0124, + "step": 34390 + }, + { + "epoch": 8.03, + "learning_rate": 3.140096618357488e-05, + "loss": 0.0008, + "step": 34400 + }, + { + "epoch": 8.03, + "learning_rate": 3.139394450061791e-05, + "loss": 0.2786, + "step": 34410 + }, + { + "epoch": 8.04, + "learning_rate": 3.1386922817660944e-05, + "loss": 0.0022, + "step": 34420 + }, + { + "epoch": 8.04, + "learning_rate": 3.1379901134703965e-05, + "loss": 0.1052, + "step": 34430 + }, + { + "epoch": 8.04, + "learning_rate": 3.137287945174699e-05, + "loss": 0.001, + "step": 34440 + }, + { + "epoch": 8.04, + "learning_rate": 3.136585776879003e-05, + "loss": 0.1113, + "step": 34450 + }, + { + "epoch": 8.04, + "learning_rate": 3.1358836085833054e-05, + "loss": 0.0002, + "step": 34460 + }, + { + "epoch": 8.04, + "learning_rate": 3.135181440287608e-05, + "loss": 0.2457, + "step": 34470 + }, + { + "epoch": 8.04, + "learning_rate": 3.1344792719919116e-05, + "loss": 0.4748, + "step": 34480 + }, + { + "epoch": 8.04, + "learning_rate": 3.133777103696214e-05, + "loss": 0.2272, + "step": 34490 + }, + { + "epoch": 8.04, + "learning_rate": 3.133074935400517e-05, + "loss": 0.232, + "step": 34500 + }, + { + "epoch": 8.04, + "learning_rate": 3.13237276710482e-05, + "loss": 0.7007, + "step": 34510 + }, + { + "epoch": 8.04, + "learning_rate": 3.1316705988091226e-05, + "loss": 0.0002, + "step": 34520 + }, + { + "epoch": 8.04, + "learning_rate": 3.130968430513426e-05, + "loss": 0.326, + "step": 34530 + }, + { + "epoch": 8.04, + "learning_rate": 3.130266262217729e-05, + "loss": 0.0255, + "step": 34540 + }, + { + "epoch": 8.04, + "learning_rate": 3.1295640939220315e-05, + "loss": 0.3082, + "step": 34550 + }, + { + "epoch": 8.04, + "learning_rate": 3.128861925626334e-05, + "loss": 0.0538, + "step": 34560 + }, + { + "epoch": 8.04, + "learning_rate": 3.128159757330637e-05, + "loss": 0.49, + "step": 34570 + }, + { + "epoch": 8.04, + "learning_rate": 3.12745758903494e-05, + "loss": 0.3683, + "step": 34580 + }, + { + "epoch": 8.04, + "learning_rate": 3.126755420739243e-05, + "loss": 0.678, + "step": 34590 + }, + { + "epoch": 8.04, + "learning_rate": 3.126053252443546e-05, + "loss": 0.3811, + "step": 34600 + }, + { + "epoch": 8.04, + "learning_rate": 3.125351084147849e-05, + "loss": 0.0168, + "step": 34610 + }, + { + "epoch": 8.04, + "learning_rate": 3.1246489158521514e-05, + "loss": 0.0147, + "step": 34620 + }, + { + "epoch": 8.04, + "learning_rate": 3.123946747556454e-05, + "loss": 0.2432, + "step": 34630 + }, + { + "epoch": 8.04, + "learning_rate": 3.1232445792607576e-05, + "loss": 0.3247, + "step": 34640 + }, + { + "epoch": 8.04, + "learning_rate": 3.1225424109650604e-05, + "loss": 0.1676, + "step": 34650 + }, + { + "epoch": 8.04, + "learning_rate": 3.121840242669363e-05, + "loss": 0.1202, + "step": 34660 + }, + { + "epoch": 8.04, + "learning_rate": 3.1211380743736665e-05, + "loss": 0.298, + "step": 34670 + }, + { + "epoch": 8.04, + "learning_rate": 3.1204359060779686e-05, + "loss": 0.0019, + "step": 34680 + }, + { + "epoch": 8.04, + "learning_rate": 3.1197337377822714e-05, + "loss": 0.1133, + "step": 34690 + }, + { + "epoch": 8.04, + "learning_rate": 3.119031569486575e-05, + "loss": 0.0003, + "step": 34700 + }, + { + "epoch": 8.04, + "learning_rate": 3.1183294011908775e-05, + "loss": 0.1052, + "step": 34710 + }, + { + "epoch": 8.04, + "learning_rate": 3.11762723289518e-05, + "loss": 0.004, + "step": 34720 + }, + { + "epoch": 8.04, + "learning_rate": 3.116925064599484e-05, + "loss": 0.3272, + "step": 34730 + }, + { + "epoch": 8.04, + "learning_rate": 3.1162228963037865e-05, + "loss": 0.2624, + "step": 34740 + }, + { + "epoch": 8.04, + "learning_rate": 3.115520728008089e-05, + "loss": 0.8137, + "step": 34750 + }, + { + "epoch": 8.04, + "learning_rate": 3.114818559712392e-05, + "loss": 0.0288, + "step": 34760 + }, + { + "epoch": 8.04, + "learning_rate": 3.114116391416695e-05, + "loss": 0.0069, + "step": 34770 + }, + { + "epoch": 8.04, + "learning_rate": 3.113414223120998e-05, + "loss": 0.0087, + "step": 34780 + }, + { + "epoch": 8.04, + "learning_rate": 3.112712054825301e-05, + "loss": 0.7169, + "step": 34790 + }, + { + "epoch": 8.04, + "learning_rate": 3.1120098865296036e-05, + "loss": 0.2302, + "step": 34800 + }, + { + "epoch": 8.04, + "learning_rate": 3.1113077182339064e-05, + "loss": 0.0018, + "step": 34810 + }, + { + "epoch": 8.04, + "learning_rate": 3.110605549938209e-05, + "loss": 0.013, + "step": 34820 + }, + { + "epoch": 8.04, + "learning_rate": 3.109903381642512e-05, + "loss": 0.0072, + "step": 34830 + }, + { + "epoch": 8.04, + "learning_rate": 3.109201213346815e-05, + "loss": 0.2886, + "step": 34840 + }, + { + "epoch": 8.04, + "learning_rate": 3.108499045051118e-05, + "loss": 0.512, + "step": 34850 + }, + { + "epoch": 8.04, + "learning_rate": 3.107796876755421e-05, + "loss": 0.0002, + "step": 34860 + }, + { + "epoch": 8.04, + "learning_rate": 3.1070947084597236e-05, + "loss": 0.0008, + "step": 34870 + }, + { + "epoch": 8.04, + "learning_rate": 3.106392540164026e-05, + "loss": 0.1088, + "step": 34880 + }, + { + "epoch": 8.04, + "learning_rate": 3.10569037186833e-05, + "loss": 0.001, + "step": 34890 + }, + { + "epoch": 8.04, + "learning_rate": 3.1049882035726325e-05, + "loss": 0.0007, + "step": 34900 + }, + { + "epoch": 8.04, + "learning_rate": 3.104286035276935e-05, + "loss": 0.0077, + "step": 34910 + }, + { + "epoch": 8.04, + "learning_rate": 3.1035838669812387e-05, + "loss": 0.1058, + "step": 34920 + }, + { + "epoch": 8.04, + "learning_rate": 3.1028816986855414e-05, + "loss": 0.4805, + "step": 34930 + }, + { + "epoch": 8.04, + "learning_rate": 3.1021795303898435e-05, + "loss": 0.1165, + "step": 34940 + }, + { + "epoch": 8.04, + "learning_rate": 3.101477362094147e-05, + "loss": 0.1794, + "step": 34950 + }, + { + "epoch": 8.04, + "learning_rate": 3.1007751937984497e-05, + "loss": 0.1606, + "step": 34960 + }, + { + "epoch": 8.04, + "learning_rate": 3.1000730255027524e-05, + "loss": 0.0014, + "step": 34970 + }, + { + "epoch": 8.04, + "learning_rate": 3.099370857207056e-05, + "loss": 0.1262, + "step": 34980 + }, + { + "epoch": 8.04, + "learning_rate": 3.0986686889113586e-05, + "loss": 0.325, + "step": 34990 + }, + { + "epoch": 8.04, + "learning_rate": 3.097966520615661e-05, + "loss": 0.0412, + "step": 35000 + }, + { + "epoch": 8.04, + "learning_rate": 3.097264352319964e-05, + "loss": 0.1357, + "step": 35010 + }, + { + "epoch": 8.04, + "learning_rate": 3.096562184024267e-05, + "loss": 0.0017, + "step": 35020 + }, + { + "epoch": 8.04, + "learning_rate": 3.09586001572857e-05, + "loss": 0.002, + "step": 35030 + }, + { + "epoch": 8.04, + "learning_rate": 3.095157847432873e-05, + "loss": 0.3587, + "step": 35040 + }, + { + "epoch": 8.04, + "learning_rate": 3.094455679137176e-05, + "loss": 0.3321, + "step": 35050 + }, + { + "epoch": 8.04, + "learning_rate": 3.093753510841479e-05, + "loss": 0.0037, + "step": 35060 + }, + { + "epoch": 8.04, + "learning_rate": 3.093051342545781e-05, + "loss": 0.0002, + "step": 35070 + }, + { + "epoch": 8.04, + "learning_rate": 3.092349174250084e-05, + "loss": 0.0004, + "step": 35080 + }, + { + "epoch": 8.04, + "learning_rate": 3.0916470059543874e-05, + "loss": 0.1129, + "step": 35090 + }, + { + "epoch": 8.04, + "learning_rate": 3.09094483765869e-05, + "loss": 0.1991, + "step": 35100 + }, + { + "epoch": 8.04, + "learning_rate": 3.090242669362993e-05, + "loss": 0.0019, + "step": 35110 + }, + { + "epoch": 8.04, + "learning_rate": 3.0895405010672963e-05, + "loss": 0.0094, + "step": 35120 + }, + { + "epoch": 8.04, + "learning_rate": 3.0888383327715984e-05, + "loss": 0.0, + "step": 35130 + }, + { + "epoch": 8.04, + "learning_rate": 3.088136164475902e-05, + "loss": 0.1415, + "step": 35140 + }, + { + "epoch": 8.04, + "learning_rate": 3.0874339961802046e-05, + "loss": 0.0057, + "step": 35150 + }, + { + "epoch": 8.04, + "learning_rate": 3.0867318278845073e-05, + "loss": 0.0003, + "step": 35160 + }, + { + "epoch": 8.04, + "learning_rate": 3.086029659588811e-05, + "loss": 0.4072, + "step": 35170 + }, + { + "epoch": 8.04, + "learning_rate": 3.0853274912931135e-05, + "loss": 0.0027, + "step": 35180 + }, + { + "epoch": 8.04, + "learning_rate": 3.0846253229974156e-05, + "loss": 0.0031, + "step": 35190 + }, + { + "epoch": 8.04, + "learning_rate": 3.083923154701719e-05, + "loss": 0.0001, + "step": 35200 + }, + { + "epoch": 8.05, + "learning_rate": 3.083220986406022e-05, + "loss": 0.0417, + "step": 35210 + }, + { + "epoch": 8.05, + "learning_rate": 3.0825188181103245e-05, + "loss": 0.0009, + "step": 35220 + }, + { + "epoch": 8.05, + "learning_rate": 3.081816649814628e-05, + "loss": 0.0007, + "step": 35230 + }, + { + "epoch": 8.05, + "learning_rate": 3.081114481518931e-05, + "loss": 0.0276, + "step": 35240 + }, + { + "epoch": 8.05, + "learning_rate": 3.0804123132232334e-05, + "loss": 0.9242, + "step": 35250 + }, + { + "epoch": 8.05, + "learning_rate": 3.079710144927536e-05, + "loss": 0.0013, + "step": 35260 + }, + { + "epoch": 8.05, + "learning_rate": 3.079007976631839e-05, + "loss": 0.4291, + "step": 35270 + }, + { + "epoch": 8.05, + "learning_rate": 3.0783058083361424e-05, + "loss": 0.1552, + "step": 35280 + }, + { + "epoch": 8.05, + "learning_rate": 3.077603640040445e-05, + "loss": 0.0978, + "step": 35290 + }, + { + "epoch": 8.05, + "learning_rate": 3.076901471744748e-05, + "loss": 0.0013, + "step": 35300 + }, + { + "epoch": 8.05, + "learning_rate": 3.076199303449051e-05, + "loss": 0.0001, + "step": 35310 + }, + { + "epoch": 8.05, + "learning_rate": 3.0754971351533534e-05, + "loss": 0.5878, + "step": 35320 + }, + { + "epoch": 8.05, + "learning_rate": 3.074794966857656e-05, + "loss": 0.0001, + "step": 35330 + }, + { + "epoch": 8.05, + "learning_rate": 3.0740927985619595e-05, + "loss": 1.034, + "step": 35340 + }, + { + "epoch": 8.05, + "learning_rate": 3.073390630266262e-05, + "loss": 0.0025, + "step": 35350 + }, + { + "epoch": 8.05, + "learning_rate": 3.072688461970565e-05, + "loss": 0.0001, + "step": 35360 + }, + { + "epoch": 8.05, + "learning_rate": 3.0719862936748685e-05, + "loss": 0.3064, + "step": 35370 + }, + { + "epoch": 8.05, + "learning_rate": 3.0712841253791705e-05, + "loss": 0.1403, + "step": 35380 + }, + { + "epoch": 8.05, + "learning_rate": 3.070581957083474e-05, + "loss": 0.002, + "step": 35390 + }, + { + "epoch": 8.05, + "learning_rate": 3.069879788787777e-05, + "loss": 0.3044, + "step": 35400 + }, + { + "epoch": 8.05, + "learning_rate": 3.0691776204920795e-05, + "loss": 0.2775, + "step": 35410 + }, + { + "epoch": 8.05, + "learning_rate": 3.068475452196383e-05, + "loss": 0.5785, + "step": 35420 + }, + { + "epoch": 8.05, + "learning_rate": 3.0677732839006856e-05, + "loss": 0.0006, + "step": 35430 + }, + { + "epoch": 8.05, + "learning_rate": 3.0670711156049884e-05, + "loss": 0.4995, + "step": 35440 + }, + { + "epoch": 8.05, + "learning_rate": 3.066368947309291e-05, + "loss": 0.2444, + "step": 35450 + }, + { + "epoch": 8.05, + "learning_rate": 3.065666779013594e-05, + "loss": 0.0011, + "step": 35460 + }, + { + "epoch": 8.05, + "learning_rate": 3.064964610717897e-05, + "loss": 0.7211, + "step": 35470 + }, + { + "epoch": 8.05, + "learning_rate": 3.0642624424222e-05, + "loss": 0.108, + "step": 35480 + }, + { + "epoch": 8.05, + "learning_rate": 3.063560274126503e-05, + "loss": 0.0007, + "step": 35490 + }, + { + "epoch": 8.05, + "learning_rate": 3.062858105830806e-05, + "loss": 0.2328, + "step": 35500 + }, + { + "epoch": 8.05, + "learning_rate": 3.062155937535108e-05, + "loss": 0.0003, + "step": 35510 + }, + { + "epoch": 8.05, + "learning_rate": 3.061453769239411e-05, + "loss": 0.0773, + "step": 35520 + }, + { + "epoch": 8.05, + "learning_rate": 3.0607516009437145e-05, + "loss": 0.204, + "step": 35530 + }, + { + "epoch": 8.05, + "learning_rate": 3.060049432648017e-05, + "loss": 0.0001, + "step": 35540 + }, + { + "epoch": 8.05, + "learning_rate": 3.05934726435232e-05, + "loss": 0.0019, + "step": 35550 + }, + { + "epoch": 8.05, + "learning_rate": 3.0586450960566234e-05, + "loss": 0.2156, + "step": 35560 + }, + { + "epoch": 8.05, + "learning_rate": 3.0579429277609255e-05, + "loss": 0.0063, + "step": 35570 + }, + { + "epoch": 8.05, + "learning_rate": 3.057240759465229e-05, + "loss": 0.0539, + "step": 35580 + }, + { + "epoch": 8.05, + "learning_rate": 3.0565385911695317e-05, + "loss": 0.4159, + "step": 35590 + }, + { + "epoch": 8.05, + "learning_rate": 3.0558364228738344e-05, + "loss": 0.0003, + "step": 35600 + }, + { + "epoch": 8.05, + "eval_accuracy": 0.6957811895239324, + "eval_loss": 2.055114507675171, + "eval_runtime": 1295.8513, + "eval_samples_per_second": 5.981, + "eval_steps_per_second": 2.991, + "step": 35604 + }, + { + "epoch": 9.0, + "learning_rate": 3.055134254578138e-05, + "loss": 0.0009, + "step": 35610 + }, + { + "epoch": 9.0, + "learning_rate": 3.0544320862824406e-05, + "loss": 0.0002, + "step": 35620 + }, + { + "epoch": 9.0, + "learning_rate": 3.053729917986743e-05, + "loss": 0.0003, + "step": 35630 + }, + { + "epoch": 9.0, + "learning_rate": 3.053027749691046e-05, + "loss": 0.0164, + "step": 35640 + }, + { + "epoch": 9.0, + "learning_rate": 3.052325581395349e-05, + "loss": 0.2862, + "step": 35650 + }, + { + "epoch": 9.0, + "learning_rate": 3.051623413099652e-05, + "loss": 0.0064, + "step": 35660 + }, + { + "epoch": 9.0, + "learning_rate": 3.0509212448039547e-05, + "loss": 0.0054, + "step": 35670 + }, + { + "epoch": 9.0, + "learning_rate": 3.0502190765082577e-05, + "loss": 0.0001, + "step": 35680 + }, + { + "epoch": 9.0, + "learning_rate": 3.049516908212561e-05, + "loss": 0.2647, + "step": 35690 + }, + { + "epoch": 9.0, + "learning_rate": 3.0488147399168632e-05, + "loss": 0.0165, + "step": 35700 + }, + { + "epoch": 9.0, + "learning_rate": 3.048112571621166e-05, + "loss": 0.0017, + "step": 35710 + }, + { + "epoch": 9.0, + "learning_rate": 3.047410403325469e-05, + "loss": 0.0006, + "step": 35720 + }, + { + "epoch": 9.0, + "learning_rate": 3.0467082350297722e-05, + "loss": 0.0004, + "step": 35730 + }, + { + "epoch": 9.0, + "learning_rate": 3.046006066734075e-05, + "loss": 0.0001, + "step": 35740 + }, + { + "epoch": 9.0, + "learning_rate": 3.045303898438378e-05, + "loss": 0.001, + "step": 35750 + }, + { + "epoch": 9.0, + "learning_rate": 3.044601730142681e-05, + "loss": 0.0003, + "step": 35760 + }, + { + "epoch": 9.0, + "learning_rate": 3.0438995618469835e-05, + "loss": 0.2428, + "step": 35770 + }, + { + "epoch": 9.0, + "learning_rate": 3.0431973935512863e-05, + "loss": 0.2229, + "step": 35780 + }, + { + "epoch": 9.0, + "learning_rate": 3.0424952252555893e-05, + "loss": 0.0001, + "step": 35790 + }, + { + "epoch": 9.0, + "learning_rate": 3.0417930569598924e-05, + "loss": 0.0078, + "step": 35800 + }, + { + "epoch": 9.0, + "learning_rate": 3.0410908886641952e-05, + "loss": 0.0027, + "step": 35810 + }, + { + "epoch": 9.0, + "learning_rate": 3.0403887203684983e-05, + "loss": 0.0, + "step": 35820 + }, + { + "epoch": 9.0, + "learning_rate": 3.0396865520728007e-05, + "loss": 0.0905, + "step": 35830 + }, + { + "epoch": 9.0, + "learning_rate": 3.0389843837771038e-05, + "loss": 0.0004, + "step": 35840 + }, + { + "epoch": 9.0, + "learning_rate": 3.0382822154814065e-05, + "loss": 0.0, + "step": 35850 + }, + { + "epoch": 9.0, + "learning_rate": 3.0375800471857096e-05, + "loss": 0.004, + "step": 35860 + }, + { + "epoch": 9.0, + "learning_rate": 3.0368778788900127e-05, + "loss": 0.0006, + "step": 35870 + }, + { + "epoch": 9.0, + "learning_rate": 3.0361757105943154e-05, + "loss": 0.0001, + "step": 35880 + }, + { + "epoch": 9.0, + "learning_rate": 3.0354735422986182e-05, + "loss": 0.4913, + "step": 35890 + }, + { + "epoch": 9.0, + "learning_rate": 3.034771374002921e-05, + "loss": 0.02, + "step": 35900 + }, + { + "epoch": 9.0, + "learning_rate": 3.034069205707224e-05, + "loss": 0.0418, + "step": 35910 + }, + { + "epoch": 9.0, + "learning_rate": 3.033367037411527e-05, + "loss": 0.1947, + "step": 35920 + }, + { + "epoch": 9.0, + "learning_rate": 3.03266486911583e-05, + "loss": 0.0623, + "step": 35930 + }, + { + "epoch": 9.0, + "learning_rate": 3.031962700820133e-05, + "loss": 0.0012, + "step": 35940 + }, + { + "epoch": 9.0, + "learning_rate": 3.031260532524436e-05, + "loss": 0.0019, + "step": 35950 + }, + { + "epoch": 9.0, + "learning_rate": 3.0305583642287384e-05, + "loss": 0.0, + "step": 35960 + }, + { + "epoch": 9.0, + "learning_rate": 3.0298561959330412e-05, + "loss": 0.0069, + "step": 35970 + }, + { + "epoch": 9.0, + "learning_rate": 3.0291540276373443e-05, + "loss": 0.0029, + "step": 35980 + }, + { + "epoch": 9.0, + "learning_rate": 3.0284518593416474e-05, + "loss": 0.0003, + "step": 35990 + }, + { + "epoch": 9.01, + "learning_rate": 3.02774969104595e-05, + "loss": 0.0639, + "step": 36000 + }, + { + "epoch": 9.01, + "learning_rate": 3.0270475227502532e-05, + "loss": 0.2623, + "step": 36010 + }, + { + "epoch": 9.01, + "learning_rate": 3.0263453544545556e-05, + "loss": 0.2823, + "step": 36020 + }, + { + "epoch": 9.01, + "learning_rate": 3.0256431861588587e-05, + "loss": 0.0092, + "step": 36030 + }, + { + "epoch": 9.01, + "learning_rate": 3.0249410178631615e-05, + "loss": 0.1899, + "step": 36040 + }, + { + "epoch": 9.01, + "learning_rate": 3.0242388495674645e-05, + "loss": 0.0029, + "step": 36050 + }, + { + "epoch": 9.01, + "learning_rate": 3.0235366812717676e-05, + "loss": 0.0002, + "step": 36060 + }, + { + "epoch": 9.01, + "learning_rate": 3.0228345129760704e-05, + "loss": 0.0013, + "step": 36070 + }, + { + "epoch": 9.01, + "learning_rate": 3.0221323446803728e-05, + "loss": 0.0001, + "step": 36080 + }, + { + "epoch": 9.01, + "learning_rate": 3.021430176384676e-05, + "loss": 0.0022, + "step": 36090 + }, + { + "epoch": 9.01, + "learning_rate": 3.020728008088979e-05, + "loss": 0.1276, + "step": 36100 + }, + { + "epoch": 9.01, + "learning_rate": 3.0200258397932817e-05, + "loss": 0.0024, + "step": 36110 + }, + { + "epoch": 9.01, + "learning_rate": 3.0193236714975848e-05, + "loss": 0.0001, + "step": 36120 + }, + { + "epoch": 9.01, + "learning_rate": 3.018621503201888e-05, + "loss": 0.0192, + "step": 36130 + }, + { + "epoch": 9.01, + "learning_rate": 3.0179193349061906e-05, + "loss": 0.0001, + "step": 36140 + }, + { + "epoch": 9.01, + "learning_rate": 3.017217166610493e-05, + "loss": 0.0101, + "step": 36150 + }, + { + "epoch": 9.01, + "learning_rate": 3.016514998314796e-05, + "loss": 0.519, + "step": 36160 + }, + { + "epoch": 9.01, + "learning_rate": 3.0158128300190992e-05, + "loss": 0.0002, + "step": 36170 + }, + { + "epoch": 9.01, + "learning_rate": 3.015110661723402e-05, + "loss": 0.0002, + "step": 36180 + }, + { + "epoch": 9.01, + "learning_rate": 3.014408493427705e-05, + "loss": 0.0001, + "step": 36190 + }, + { + "epoch": 9.01, + "learning_rate": 3.013706325132008e-05, + "loss": 0.0008, + "step": 36200 + }, + { + "epoch": 9.01, + "learning_rate": 3.0130041568363106e-05, + "loss": 0.0057, + "step": 36210 + }, + { + "epoch": 9.01, + "learning_rate": 3.0123019885406133e-05, + "loss": 0.0999, + "step": 36220 + }, + { + "epoch": 9.01, + "learning_rate": 3.0115998202449164e-05, + "loss": 0.1977, + "step": 36230 + }, + { + "epoch": 9.01, + "learning_rate": 3.0108976519492195e-05, + "loss": 0.0519, + "step": 36240 + }, + { + "epoch": 9.01, + "learning_rate": 3.0101954836535222e-05, + "loss": 0.011, + "step": 36250 + }, + { + "epoch": 9.01, + "learning_rate": 3.0094933153578253e-05, + "loss": 0.1238, + "step": 36260 + }, + { + "epoch": 9.01, + "learning_rate": 3.0087911470621277e-05, + "loss": 0.0903, + "step": 36270 + }, + { + "epoch": 9.01, + "learning_rate": 3.0080889787664308e-05, + "loss": 0.0946, + "step": 36280 + }, + { + "epoch": 9.01, + "learning_rate": 3.0073868104707336e-05, + "loss": 0.2004, + "step": 36290 + }, + { + "epoch": 9.01, + "learning_rate": 3.0066846421750367e-05, + "loss": 0.4675, + "step": 36300 + }, + { + "epoch": 9.01, + "learning_rate": 3.0059824738793397e-05, + "loss": 0.0096, + "step": 36310 + }, + { + "epoch": 9.01, + "learning_rate": 3.0052803055836425e-05, + "loss": 0.0289, + "step": 36320 + }, + { + "epoch": 9.01, + "learning_rate": 3.0045781372879456e-05, + "loss": 0.5527, + "step": 36330 + }, + { + "epoch": 9.01, + "learning_rate": 3.003875968992248e-05, + "loss": 0.2882, + "step": 36340 + }, + { + "epoch": 9.01, + "learning_rate": 3.003173800696551e-05, + "loss": 0.2136, + "step": 36350 + }, + { + "epoch": 9.01, + "learning_rate": 3.0024716324008538e-05, + "loss": 0.0356, + "step": 36360 + }, + { + "epoch": 9.01, + "learning_rate": 3.001769464105157e-05, + "loss": 0.0005, + "step": 36370 + }, + { + "epoch": 9.01, + "learning_rate": 3.00106729580946e-05, + "loss": 0.3856, + "step": 36380 + }, + { + "epoch": 9.01, + "learning_rate": 3.0003651275137628e-05, + "loss": 0.1088, + "step": 36390 + }, + { + "epoch": 9.01, + "learning_rate": 2.999662959218065e-05, + "loss": 0.0095, + "step": 36400 + }, + { + "epoch": 9.01, + "learning_rate": 2.9989607909223683e-05, + "loss": 0.0016, + "step": 36410 + }, + { + "epoch": 9.01, + "learning_rate": 2.9982586226266713e-05, + "loss": 1.1894, + "step": 36420 + }, + { + "epoch": 9.01, + "learning_rate": 2.997556454330974e-05, + "loss": 0.433, + "step": 36430 + }, + { + "epoch": 9.01, + "learning_rate": 2.9968542860352772e-05, + "loss": 0.1048, + "step": 36440 + }, + { + "epoch": 9.01, + "learning_rate": 2.9961521177395803e-05, + "loss": 0.0003, + "step": 36450 + }, + { + "epoch": 9.01, + "learning_rate": 2.995449949443883e-05, + "loss": 0.1512, + "step": 36460 + }, + { + "epoch": 9.01, + "learning_rate": 2.9947477811481854e-05, + "loss": 0.2023, + "step": 36470 + }, + { + "epoch": 9.01, + "learning_rate": 2.9940456128524885e-05, + "loss": 0.0001, + "step": 36480 + }, + { + "epoch": 9.01, + "learning_rate": 2.9933434445567916e-05, + "loss": 0.2142, + "step": 36490 + }, + { + "epoch": 9.01, + "learning_rate": 2.9926412762610943e-05, + "loss": 0.0022, + "step": 36500 + }, + { + "epoch": 9.01, + "learning_rate": 2.9919391079653974e-05, + "loss": 0.2948, + "step": 36510 + }, + { + "epoch": 9.01, + "learning_rate": 2.9912369396697005e-05, + "loss": 0.0258, + "step": 36520 + }, + { + "epoch": 9.01, + "learning_rate": 2.990534771374003e-05, + "loss": 0.0047, + "step": 36530 + }, + { + "epoch": 9.01, + "learning_rate": 2.9898326030783057e-05, + "loss": 0.0865, + "step": 36540 + }, + { + "epoch": 9.01, + "learning_rate": 2.9891304347826088e-05, + "loss": 0.0649, + "step": 36550 + }, + { + "epoch": 9.01, + "learning_rate": 2.988428266486912e-05, + "loss": 0.0001, + "step": 36560 + }, + { + "epoch": 9.01, + "learning_rate": 2.9877260981912146e-05, + "loss": 0.6622, + "step": 36570 + }, + { + "epoch": 9.01, + "learning_rate": 2.9870239298955177e-05, + "loss": 0.0056, + "step": 36580 + }, + { + "epoch": 9.01, + "learning_rate": 2.98632176159982e-05, + "loss": 0.1631, + "step": 36590 + }, + { + "epoch": 9.01, + "learning_rate": 2.9856195933041232e-05, + "loss": 0.4805, + "step": 36600 + }, + { + "epoch": 9.01, + "learning_rate": 2.984917425008426e-05, + "loss": 0.4517, + "step": 36610 + }, + { + "epoch": 9.01, + "learning_rate": 2.984215256712729e-05, + "loss": 0.0005, + "step": 36620 + }, + { + "epoch": 9.01, + "learning_rate": 2.983513088417032e-05, + "loss": 0.0005, + "step": 36630 + }, + { + "epoch": 9.01, + "learning_rate": 2.982810920121335e-05, + "loss": 0.0002, + "step": 36640 + }, + { + "epoch": 9.01, + "learning_rate": 2.982108751825638e-05, + "loss": 0.0687, + "step": 36650 + }, + { + "epoch": 9.01, + "learning_rate": 2.9814065835299404e-05, + "loss": 0.4051, + "step": 36660 + }, + { + "epoch": 9.01, + "learning_rate": 2.9807044152342435e-05, + "loss": 0.2592, + "step": 36670 + }, + { + "epoch": 9.01, + "learning_rate": 2.9800022469385462e-05, + "loss": 0.0013, + "step": 36680 + }, + { + "epoch": 9.01, + "learning_rate": 2.9793000786428493e-05, + "loss": 0.0244, + "step": 36690 + }, + { + "epoch": 9.01, + "learning_rate": 2.9785979103471524e-05, + "loss": 0.1839, + "step": 36700 + }, + { + "epoch": 9.01, + "learning_rate": 2.977895742051455e-05, + "loss": 0.3085, + "step": 36710 + }, + { + "epoch": 9.01, + "learning_rate": 2.9771935737557575e-05, + "loss": 0.0032, + "step": 36720 + }, + { + "epoch": 9.01, + "learning_rate": 2.9764914054600606e-05, + "loss": 0.0021, + "step": 36730 + }, + { + "epoch": 9.01, + "learning_rate": 2.9757892371643637e-05, + "loss": 0.0031, + "step": 36740 + }, + { + "epoch": 9.01, + "learning_rate": 2.9750870688686665e-05, + "loss": 0.0001, + "step": 36750 + }, + { + "epoch": 9.01, + "learning_rate": 2.9743849005729696e-05, + "loss": 0.2438, + "step": 36760 + }, + { + "epoch": 9.01, + "learning_rate": 2.9736827322772726e-05, + "loss": 0.0006, + "step": 36770 + }, + { + "epoch": 9.01, + "learning_rate": 2.972980563981575e-05, + "loss": 0.0045, + "step": 36780 + }, + { + "epoch": 9.01, + "learning_rate": 2.9722783956858778e-05, + "loss": 0.0909, + "step": 36790 + }, + { + "epoch": 9.02, + "learning_rate": 2.971576227390181e-05, + "loss": 0.0008, + "step": 36800 + }, + { + "epoch": 9.02, + "learning_rate": 2.970874059094484e-05, + "loss": 0.001, + "step": 36810 + }, + { + "epoch": 9.02, + "learning_rate": 2.9701718907987867e-05, + "loss": 0.0041, + "step": 36820 + }, + { + "epoch": 9.02, + "learning_rate": 2.9694697225030898e-05, + "loss": 0.0663, + "step": 36830 + }, + { + "epoch": 9.02, + "learning_rate": 2.968767554207393e-05, + "loss": 0.0064, + "step": 36840 + }, + { + "epoch": 9.02, + "learning_rate": 2.9680653859116953e-05, + "loss": 0.0, + "step": 36850 + }, + { + "epoch": 9.02, + "learning_rate": 2.967363217615998e-05, + "loss": 0.309, + "step": 36860 + }, + { + "epoch": 9.02, + "learning_rate": 2.966661049320301e-05, + "loss": 0.0003, + "step": 36870 + }, + { + "epoch": 9.02, + "learning_rate": 2.9659588810246042e-05, + "loss": 0.0745, + "step": 36880 + }, + { + "epoch": 9.02, + "learning_rate": 2.965256712728907e-05, + "loss": 0.3719, + "step": 36890 + }, + { + "epoch": 9.02, + "learning_rate": 2.96455454443321e-05, + "loss": 0.0003, + "step": 36900 + }, + { + "epoch": 9.02, + "learning_rate": 2.9638523761375125e-05, + "loss": 0.1768, + "step": 36910 + }, + { + "epoch": 9.02, + "learning_rate": 2.9631502078418156e-05, + "loss": 0.0191, + "step": 36920 + }, + { + "epoch": 9.02, + "learning_rate": 2.9624480395461187e-05, + "loss": 1.124, + "step": 36930 + }, + { + "epoch": 9.02, + "learning_rate": 2.9617458712504214e-05, + "loss": 0.1613, + "step": 36940 + }, + { + "epoch": 9.02, + "learning_rate": 2.9610437029547245e-05, + "loss": 0.0007, + "step": 36950 + }, + { + "epoch": 9.02, + "learning_rate": 2.9603415346590276e-05, + "loss": 0.0398, + "step": 36960 + }, + { + "epoch": 9.02, + "learning_rate": 2.95963936636333e-05, + "loss": 0.0, + "step": 36970 + }, + { + "epoch": 9.02, + "learning_rate": 2.9589371980676327e-05, + "loss": 0.0055, + "step": 36980 + }, + { + "epoch": 9.02, + "learning_rate": 2.9582350297719358e-05, + "loss": 0.3511, + "step": 36990 + }, + { + "epoch": 9.02, + "learning_rate": 2.957532861476239e-05, + "loss": 0.0086, + "step": 37000 + }, + { + "epoch": 9.02, + "learning_rate": 2.9568306931805417e-05, + "loss": 0.0047, + "step": 37010 + }, + { + "epoch": 9.02, + "learning_rate": 2.9561285248848448e-05, + "loss": 0.0227, + "step": 37020 + }, + { + "epoch": 9.02, + "learning_rate": 2.955426356589148e-05, + "loss": 0.0347, + "step": 37030 + }, + { + "epoch": 9.02, + "learning_rate": 2.9547241882934503e-05, + "loss": 0.0045, + "step": 37040 + }, + { + "epoch": 9.02, + "learning_rate": 2.954022019997753e-05, + "loss": 0.3219, + "step": 37050 + }, + { + "epoch": 9.02, + "learning_rate": 2.953319851702056e-05, + "loss": 0.0002, + "step": 37060 + }, + { + "epoch": 9.02, + "learning_rate": 2.9526176834063592e-05, + "loss": 0.01, + "step": 37070 + }, + { + "epoch": 9.02, + "learning_rate": 2.951915515110662e-05, + "loss": 0.0772, + "step": 37080 + }, + { + "epoch": 9.02, + "learning_rate": 2.951213346814965e-05, + "loss": 0.0002, + "step": 37090 + }, + { + "epoch": 9.02, + "learning_rate": 2.9505111785192674e-05, + "loss": 0.0002, + "step": 37100 + }, + { + "epoch": 9.02, + "learning_rate": 2.9498090102235705e-05, + "loss": 0.0634, + "step": 37110 + }, + { + "epoch": 9.02, + "learning_rate": 2.9491068419278733e-05, + "loss": 0.0008, + "step": 37120 + }, + { + "epoch": 9.02, + "learning_rate": 2.9484046736321763e-05, + "loss": 0.3721, + "step": 37130 + }, + { + "epoch": 9.02, + "learning_rate": 2.9477025053364794e-05, + "loss": 0.0003, + "step": 37140 + }, + { + "epoch": 9.02, + "learning_rate": 2.9470003370407822e-05, + "loss": 0.0002, + "step": 37150 + }, + { + "epoch": 9.02, + "learning_rate": 2.9462981687450846e-05, + "loss": 0.0002, + "step": 37160 + }, + { + "epoch": 9.02, + "learning_rate": 2.9455960004493877e-05, + "loss": 0.2725, + "step": 37170 + }, + { + "epoch": 9.02, + "learning_rate": 2.9448938321536908e-05, + "loss": 0.1791, + "step": 37180 + }, + { + "epoch": 9.02, + "learning_rate": 2.9441916638579935e-05, + "loss": 0.0038, + "step": 37190 + }, + { + "epoch": 9.02, + "learning_rate": 2.9434894955622966e-05, + "loss": 0.2067, + "step": 37200 + }, + { + "epoch": 9.02, + "learning_rate": 2.9427873272665997e-05, + "loss": 0.0005, + "step": 37210 + }, + { + "epoch": 9.02, + "learning_rate": 2.9420851589709024e-05, + "loss": 0.0015, + "step": 37220 + }, + { + "epoch": 9.02, + "learning_rate": 2.941382990675205e-05, + "loss": 0.1993, + "step": 37230 + }, + { + "epoch": 9.02, + "learning_rate": 2.940680822379508e-05, + "loss": 0.0657, + "step": 37240 + }, + { + "epoch": 9.02, + "learning_rate": 2.939978654083811e-05, + "loss": 0.0639, + "step": 37250 + }, + { + "epoch": 9.02, + "learning_rate": 2.9392764857881138e-05, + "loss": 0.0001, + "step": 37260 + }, + { + "epoch": 9.02, + "learning_rate": 2.938574317492417e-05, + "loss": 0.0098, + "step": 37270 + }, + { + "epoch": 9.02, + "learning_rate": 2.93787214919672e-05, + "loss": 0.0178, + "step": 37280 + }, + { + "epoch": 9.02, + "learning_rate": 2.9371699809010224e-05, + "loss": 0.2002, + "step": 37290 + }, + { + "epoch": 9.02, + "learning_rate": 2.936467812605325e-05, + "loss": 0.0006, + "step": 37300 + }, + { + "epoch": 9.02, + "learning_rate": 2.9357656443096282e-05, + "loss": 0.0006, + "step": 37310 + }, + { + "epoch": 9.02, + "learning_rate": 2.9350634760139313e-05, + "loss": 0.0018, + "step": 37320 + }, + { + "epoch": 9.02, + "learning_rate": 2.934361307718234e-05, + "loss": 0.3532, + "step": 37330 + }, + { + "epoch": 9.02, + "learning_rate": 2.933659139422537e-05, + "loss": 0.0001, + "step": 37340 + }, + { + "epoch": 9.02, + "learning_rate": 2.9329569711268402e-05, + "loss": 0.0, + "step": 37350 + }, + { + "epoch": 9.02, + "learning_rate": 2.9322548028311426e-05, + "loss": 0.7083, + "step": 37360 + }, + { + "epoch": 9.02, + "learning_rate": 2.9315526345354454e-05, + "loss": 0.3812, + "step": 37370 + }, + { + "epoch": 9.02, + "learning_rate": 2.9308504662397485e-05, + "loss": 0.0029, + "step": 37380 + }, + { + "epoch": 9.02, + "learning_rate": 2.9301482979440516e-05, + "loss": 0.6736, + "step": 37390 + }, + { + "epoch": 9.02, + "learning_rate": 2.9294461296483543e-05, + "loss": 0.0, + "step": 37400 + }, + { + "epoch": 9.02, + "learning_rate": 2.9287439613526574e-05, + "loss": 0.0001, + "step": 37410 + }, + { + "epoch": 9.02, + "learning_rate": 2.9280417930569598e-05, + "loss": 0.0164, + "step": 37420 + }, + { + "epoch": 9.02, + "learning_rate": 2.927339624761263e-05, + "loss": 0.0118, + "step": 37430 + }, + { + "epoch": 9.02, + "learning_rate": 2.9266374564655656e-05, + "loss": 0.0001, + "step": 37440 + }, + { + "epoch": 9.02, + "learning_rate": 2.9259352881698687e-05, + "loss": 0.0005, + "step": 37450 + }, + { + "epoch": 9.02, + "learning_rate": 2.9252331198741718e-05, + "loss": 0.0, + "step": 37460 + }, + { + "epoch": 9.02, + "learning_rate": 2.9245309515784746e-05, + "loss": 0.0194, + "step": 37470 + }, + { + "epoch": 9.02, + "learning_rate": 2.923828783282777e-05, + "loss": 0.0029, + "step": 37480 + }, + { + "epoch": 9.02, + "learning_rate": 2.92312661498708e-05, + "loss": 0.0021, + "step": 37490 + }, + { + "epoch": 9.02, + "learning_rate": 2.922424446691383e-05, + "loss": 0.0155, + "step": 37500 + }, + { + "epoch": 9.02, + "learning_rate": 2.921722278395686e-05, + "loss": 0.0478, + "step": 37510 + }, + { + "epoch": 9.02, + "learning_rate": 2.921020110099989e-05, + "loss": 0.0009, + "step": 37520 + }, + { + "epoch": 9.02, + "learning_rate": 2.920317941804292e-05, + "loss": 0.0003, + "step": 37530 + }, + { + "epoch": 9.02, + "learning_rate": 2.9196157735085948e-05, + "loss": 0.596, + "step": 37540 + }, + { + "epoch": 9.02, + "learning_rate": 2.9189136052128972e-05, + "loss": 0.0001, + "step": 37550 + }, + { + "epoch": 9.02, + "learning_rate": 2.9182114369172003e-05, + "loss": 0.4764, + "step": 37560 + }, + { + "epoch": 9.02, + "learning_rate": 2.9175092686215034e-05, + "loss": 0.0047, + "step": 37570 + }, + { + "epoch": 9.02, + "learning_rate": 2.916807100325806e-05, + "loss": 0.0063, + "step": 37580 + }, + { + "epoch": 9.03, + "learning_rate": 2.9161049320301092e-05, + "loss": 0.1078, + "step": 37590 + }, + { + "epoch": 9.03, + "learning_rate": 2.9154027637344123e-05, + "loss": 0.0043, + "step": 37600 + }, + { + "epoch": 9.03, + "learning_rate": 2.9147005954387147e-05, + "loss": 0.2379, + "step": 37610 + }, + { + "epoch": 9.03, + "learning_rate": 2.9139984271430175e-05, + "loss": 0.001, + "step": 37620 + }, + { + "epoch": 9.03, + "learning_rate": 2.9132962588473206e-05, + "loss": 0.02, + "step": 37630 + }, + { + "epoch": 9.03, + "learning_rate": 2.9125940905516237e-05, + "loss": 0.0566, + "step": 37640 + }, + { + "epoch": 9.03, + "learning_rate": 2.9118919222559264e-05, + "loss": 0.239, + "step": 37650 + }, + { + "epoch": 9.03, + "learning_rate": 2.9111897539602295e-05, + "loss": 0.2261, + "step": 37660 + }, + { + "epoch": 9.03, + "learning_rate": 2.910487585664532e-05, + "loss": 0.0001, + "step": 37670 + }, + { + "epoch": 9.03, + "learning_rate": 2.909785417368835e-05, + "loss": 0.0679, + "step": 37680 + }, + { + "epoch": 9.03, + "learning_rate": 2.9090832490731377e-05, + "loss": 0.0019, + "step": 37690 + }, + { + "epoch": 9.03, + "learning_rate": 2.908381080777441e-05, + "loss": 0.0001, + "step": 37700 + }, + { + "epoch": 9.03, + "learning_rate": 2.907678912481744e-05, + "loss": 0.0167, + "step": 37710 + }, + { + "epoch": 9.03, + "learning_rate": 2.9069767441860467e-05, + "loss": 0.8766, + "step": 37720 + }, + { + "epoch": 9.03, + "learning_rate": 2.9062745758903498e-05, + "loss": 0.0753, + "step": 37730 + }, + { + "epoch": 9.03, + "learning_rate": 2.9055724075946522e-05, + "loss": 0.1337, + "step": 37740 + }, + { + "epoch": 9.03, + "learning_rate": 2.9048702392989553e-05, + "loss": 0.0019, + "step": 37750 + }, + { + "epoch": 9.03, + "learning_rate": 2.904168071003258e-05, + "loss": 0.0002, + "step": 37760 + }, + { + "epoch": 9.03, + "learning_rate": 2.903465902707561e-05, + "loss": 0.4535, + "step": 37770 + }, + { + "epoch": 9.03, + "learning_rate": 2.9027637344118642e-05, + "loss": 0.0992, + "step": 37780 + }, + { + "epoch": 9.03, + "learning_rate": 2.902061566116167e-05, + "loss": 0.0001, + "step": 37790 + }, + { + "epoch": 9.03, + "learning_rate": 2.9013593978204693e-05, + "loss": 0.1081, + "step": 37800 + }, + { + "epoch": 9.03, + "learning_rate": 2.9006572295247724e-05, + "loss": 0.104, + "step": 37810 + }, + { + "epoch": 9.03, + "learning_rate": 2.8999550612290755e-05, + "loss": 0.4337, + "step": 37820 + }, + { + "epoch": 9.03, + "learning_rate": 2.8992528929333783e-05, + "loss": 0.001, + "step": 37830 + }, + { + "epoch": 9.03, + "learning_rate": 2.8985507246376814e-05, + "loss": 0.0001, + "step": 37840 + }, + { + "epoch": 9.03, + "learning_rate": 2.8978485563419844e-05, + "loss": 0.0003, + "step": 37850 + }, + { + "epoch": 9.03, + "learning_rate": 2.897146388046287e-05, + "loss": 0.018, + "step": 37860 + }, + { + "epoch": 9.03, + "learning_rate": 2.8964442197505896e-05, + "loss": 0.4052, + "step": 37870 + }, + { + "epoch": 9.03, + "learning_rate": 2.8957420514548927e-05, + "loss": 0.0087, + "step": 37880 + }, + { + "epoch": 9.03, + "learning_rate": 2.8950398831591958e-05, + "loss": 0.0001, + "step": 37890 + }, + { + "epoch": 9.03, + "learning_rate": 2.8943377148634985e-05, + "loss": 0.0, + "step": 37900 + }, + { + "epoch": 9.03, + "learning_rate": 2.8936355465678016e-05, + "loss": 0.0254, + "step": 37910 + }, + { + "epoch": 9.03, + "learning_rate": 2.8929333782721047e-05, + "loss": 0.045, + "step": 37920 + }, + { + "epoch": 9.03, + "learning_rate": 2.892231209976407e-05, + "loss": 0.0001, + "step": 37930 + }, + { + "epoch": 9.03, + "learning_rate": 2.8915290416807102e-05, + "loss": 0.0, + "step": 37940 + }, + { + "epoch": 9.03, + "learning_rate": 2.890826873385013e-05, + "loss": 0.0242, + "step": 37950 + }, + { + "epoch": 9.03, + "learning_rate": 2.890124705089316e-05, + "loss": 0.1953, + "step": 37960 + }, + { + "epoch": 9.03, + "learning_rate": 2.889422536793619e-05, + "loss": 0.4858, + "step": 37970 + }, + { + "epoch": 9.03, + "learning_rate": 2.888720368497922e-05, + "loss": 0.1003, + "step": 37980 + }, + { + "epoch": 9.03, + "learning_rate": 2.8880182002022243e-05, + "loss": 0.1829, + "step": 37990 + }, + { + "epoch": 9.03, + "learning_rate": 2.8873160319065274e-05, + "loss": 0.0002, + "step": 38000 + }, + { + "epoch": 9.03, + "learning_rate": 2.8866138636108305e-05, + "loss": 0.0003, + "step": 38010 + }, + { + "epoch": 9.03, + "learning_rate": 2.8859116953151332e-05, + "loss": 0.4911, + "step": 38020 + }, + { + "epoch": 9.03, + "learning_rate": 2.8852095270194363e-05, + "loss": 0.0001, + "step": 38030 + }, + { + "epoch": 9.03, + "learning_rate": 2.8845073587237394e-05, + "loss": 0.4309, + "step": 38040 + }, + { + "epoch": 9.03, + "learning_rate": 2.883805190428042e-05, + "loss": 0.091, + "step": 38050 + }, + { + "epoch": 9.03, + "learning_rate": 2.8831030221323445e-05, + "loss": 0.0003, + "step": 38060 + }, + { + "epoch": 9.03, + "learning_rate": 2.8824008538366476e-05, + "loss": 0.0033, + "step": 38070 + }, + { + "epoch": 9.03, + "learning_rate": 2.8816986855409507e-05, + "loss": 0.0, + "step": 38080 + }, + { + "epoch": 9.03, + "learning_rate": 2.8809965172452535e-05, + "loss": 0.0003, + "step": 38090 + }, + { + "epoch": 9.03, + "learning_rate": 2.8802943489495566e-05, + "loss": 0.1397, + "step": 38100 + }, + { + "epoch": 9.03, + "learning_rate": 2.8795921806538596e-05, + "loss": 0.0, + "step": 38110 + }, + { + "epoch": 9.03, + "learning_rate": 2.878890012358162e-05, + "loss": 0.1276, + "step": 38120 + }, + { + "epoch": 9.03, + "learning_rate": 2.8781878440624648e-05, + "loss": 0.4788, + "step": 38130 + }, + { + "epoch": 9.03, + "learning_rate": 2.877485675766768e-05, + "loss": 0.4064, + "step": 38140 + }, + { + "epoch": 9.03, + "learning_rate": 2.876783507471071e-05, + "loss": 0.4755, + "step": 38150 + }, + { + "epoch": 9.03, + "learning_rate": 2.8760813391753737e-05, + "loss": 0.0195, + "step": 38160 + }, + { + "epoch": 9.03, + "learning_rate": 2.8753791708796768e-05, + "loss": 0.1915, + "step": 38170 + }, + { + "epoch": 9.03, + "learning_rate": 2.8746770025839792e-05, + "loss": 0.3409, + "step": 38180 + }, + { + "epoch": 9.03, + "learning_rate": 2.8739748342882823e-05, + "loss": 0.1778, + "step": 38190 + }, + { + "epoch": 9.03, + "learning_rate": 2.873272665992585e-05, + "loss": 0.8771, + "step": 38200 + }, + { + "epoch": 9.03, + "learning_rate": 2.872570497696888e-05, + "loss": 0.0034, + "step": 38210 + }, + { + "epoch": 9.03, + "learning_rate": 2.8718683294011912e-05, + "loss": 0.0393, + "step": 38220 + }, + { + "epoch": 9.03, + "learning_rate": 2.871166161105494e-05, + "loss": 0.0004, + "step": 38230 + }, + { + "epoch": 9.03, + "learning_rate": 2.870463992809797e-05, + "loss": 0.228, + "step": 38240 + }, + { + "epoch": 9.03, + "learning_rate": 2.8697618245140995e-05, + "loss": 0.0242, + "step": 38250 + }, + { + "epoch": 9.03, + "learning_rate": 2.8690596562184026e-05, + "loss": 0.3688, + "step": 38260 + }, + { + "epoch": 9.03, + "learning_rate": 2.8683574879227053e-05, + "loss": 0.0073, + "step": 38270 + }, + { + "epoch": 9.03, + "learning_rate": 2.8676553196270084e-05, + "loss": 0.0117, + "step": 38280 + }, + { + "epoch": 9.03, + "learning_rate": 2.8669531513313115e-05, + "loss": 0.1525, + "step": 38290 + }, + { + "epoch": 9.03, + "learning_rate": 2.8662509830356143e-05, + "loss": 0.0057, + "step": 38300 + }, + { + "epoch": 9.03, + "learning_rate": 2.8655488147399167e-05, + "loss": 0.5723, + "step": 38310 + }, + { + "epoch": 9.03, + "learning_rate": 2.8648466464442197e-05, + "loss": 0.0065, + "step": 38320 + }, + { + "epoch": 9.03, + "learning_rate": 2.864144478148523e-05, + "loss": 0.1541, + "step": 38330 + }, + { + "epoch": 9.03, + "learning_rate": 2.8634423098528256e-05, + "loss": 0.0117, + "step": 38340 + }, + { + "epoch": 9.03, + "learning_rate": 2.8627401415571287e-05, + "loss": 0.0137, + "step": 38350 + }, + { + "epoch": 9.03, + "learning_rate": 2.8620379732614318e-05, + "loss": 0.1072, + "step": 38360 + }, + { + "epoch": 9.03, + "learning_rate": 2.8613358049657342e-05, + "loss": 0.2208, + "step": 38370 + }, + { + "epoch": 9.04, + "learning_rate": 2.860633636670037e-05, + "loss": 0.1395, + "step": 38380 + }, + { + "epoch": 9.04, + "learning_rate": 2.85993146837434e-05, + "loss": 0.0702, + "step": 38390 + }, + { + "epoch": 9.04, + "learning_rate": 2.859229300078643e-05, + "loss": 0.108, + "step": 38400 + }, + { + "epoch": 9.04, + "learning_rate": 2.858527131782946e-05, + "loss": 0.0001, + "step": 38410 + }, + { + "epoch": 9.04, + "learning_rate": 2.857824963487249e-05, + "loss": 0.0072, + "step": 38420 + }, + { + "epoch": 9.04, + "learning_rate": 2.857122795191552e-05, + "loss": 0.4558, + "step": 38430 + }, + { + "epoch": 9.04, + "learning_rate": 2.8564206268958544e-05, + "loss": 0.242, + "step": 38440 + }, + { + "epoch": 9.04, + "learning_rate": 2.8557184586001572e-05, + "loss": 0.6701, + "step": 38450 + }, + { + "epoch": 9.04, + "learning_rate": 2.8550162903044603e-05, + "loss": 0.0491, + "step": 38460 + }, + { + "epoch": 9.04, + "learning_rate": 2.8543141220087634e-05, + "loss": 0.4135, + "step": 38470 + }, + { + "epoch": 9.04, + "learning_rate": 2.853611953713066e-05, + "loss": 0.0012, + "step": 38480 + }, + { + "epoch": 9.04, + "learning_rate": 2.8529097854173692e-05, + "loss": 0.6546, + "step": 38490 + }, + { + "epoch": 9.04, + "learning_rate": 2.8522076171216716e-05, + "loss": 0.0315, + "step": 38500 + }, + { + "epoch": 9.04, + "learning_rate": 2.8515054488259747e-05, + "loss": 0.1845, + "step": 38510 + }, + { + "epoch": 9.04, + "learning_rate": 2.8508032805302774e-05, + "loss": 0.0004, + "step": 38520 + }, + { + "epoch": 9.04, + "learning_rate": 2.8501011122345805e-05, + "loss": 0.0001, + "step": 38530 + }, + { + "epoch": 9.04, + "learning_rate": 2.8493989439388836e-05, + "loss": 0.0137, + "step": 38540 + }, + { + "epoch": 9.04, + "learning_rate": 2.8486967756431864e-05, + "loss": 0.3089, + "step": 38550 + }, + { + "epoch": 9.04, + "learning_rate": 2.8479946073474888e-05, + "loss": 0.0001, + "step": 38560 + }, + { + "epoch": 9.04, + "learning_rate": 2.847292439051792e-05, + "loss": 0.0002, + "step": 38570 + }, + { + "epoch": 9.04, + "learning_rate": 2.846590270756095e-05, + "loss": 0.0, + "step": 38580 + }, + { + "epoch": 9.04, + "learning_rate": 2.8458881024603977e-05, + "loss": 0.0005, + "step": 38590 + }, + { + "epoch": 9.04, + "learning_rate": 2.8451859341647008e-05, + "loss": 0.0726, + "step": 38600 + }, + { + "epoch": 9.04, + "learning_rate": 2.844483765869004e-05, + "loss": 0.1651, + "step": 38610 + }, + { + "epoch": 9.04, + "learning_rate": 2.8437815975733066e-05, + "loss": 0.0004, + "step": 38620 + }, + { + "epoch": 9.04, + "learning_rate": 2.843079429277609e-05, + "loss": 0.3549, + "step": 38630 + }, + { + "epoch": 9.04, + "learning_rate": 2.842377260981912e-05, + "loss": 0.0723, + "step": 38640 + }, + { + "epoch": 9.04, + "learning_rate": 2.8416750926862152e-05, + "loss": 0.0294, + "step": 38650 + }, + { + "epoch": 9.04, + "learning_rate": 2.840972924390518e-05, + "loss": 0.0066, + "step": 38660 + }, + { + "epoch": 9.04, + "learning_rate": 2.840270756094821e-05, + "loss": 0.1154, + "step": 38670 + }, + { + "epoch": 9.04, + "learning_rate": 2.839568587799124e-05, + "loss": 0.5324, + "step": 38680 + }, + { + "epoch": 9.04, + "learning_rate": 2.8388664195034265e-05, + "loss": 0.5059, + "step": 38690 + }, + { + "epoch": 9.04, + "learning_rate": 2.8381642512077293e-05, + "loss": 0.3097, + "step": 38700 + }, + { + "epoch": 9.04, + "learning_rate": 2.8374620829120324e-05, + "loss": 0.0002, + "step": 38710 + }, + { + "epoch": 9.04, + "learning_rate": 2.8367599146163355e-05, + "loss": 0.0001, + "step": 38720 + }, + { + "epoch": 9.04, + "learning_rate": 2.8360577463206382e-05, + "loss": 0.202, + "step": 38730 + }, + { + "epoch": 9.04, + "learning_rate": 2.8353555780249413e-05, + "loss": 0.0167, + "step": 38740 + }, + { + "epoch": 9.04, + "learning_rate": 2.8346534097292444e-05, + "loss": 0.4624, + "step": 38750 + }, + { + "epoch": 9.04, + "learning_rate": 2.8339512414335468e-05, + "loss": 0.078, + "step": 38760 + }, + { + "epoch": 9.04, + "learning_rate": 2.8332490731378496e-05, + "loss": 0.0063, + "step": 38770 + }, + { + "epoch": 9.04, + "learning_rate": 2.8325469048421526e-05, + "loss": 0.0056, + "step": 38780 + }, + { + "epoch": 9.04, + "learning_rate": 2.8318447365464557e-05, + "loss": 0.1693, + "step": 38790 + }, + { + "epoch": 9.04, + "learning_rate": 2.8311425682507585e-05, + "loss": 0.0, + "step": 38800 + }, + { + "epoch": 9.04, + "learning_rate": 2.8304403999550616e-05, + "loss": 0.0003, + "step": 38810 + }, + { + "epoch": 9.04, + "learning_rate": 2.829738231659364e-05, + "loss": 0.0028, + "step": 38820 + }, + { + "epoch": 9.04, + "learning_rate": 2.829036063363667e-05, + "loss": 0.8018, + "step": 38830 + }, + { + "epoch": 9.04, + "learning_rate": 2.8283338950679698e-05, + "loss": 0.4797, + "step": 38840 + }, + { + "epoch": 9.04, + "learning_rate": 2.827631726772273e-05, + "loss": 0.0004, + "step": 38850 + }, + { + "epoch": 9.04, + "learning_rate": 2.826929558476576e-05, + "loss": 0.1022, + "step": 38860 + }, + { + "epoch": 9.04, + "learning_rate": 2.8262273901808787e-05, + "loss": 0.0005, + "step": 38870 + }, + { + "epoch": 9.04, + "learning_rate": 2.825525221885181e-05, + "loss": 0.0007, + "step": 38880 + }, + { + "epoch": 9.04, + "learning_rate": 2.8248230535894842e-05, + "loss": 0.2844, + "step": 38890 + }, + { + "epoch": 9.04, + "learning_rate": 2.8241208852937873e-05, + "loss": 0.0306, + "step": 38900 + }, + { + "epoch": 9.04, + "learning_rate": 2.82341871699809e-05, + "loss": 0.3681, + "step": 38910 + }, + { + "epoch": 9.04, + "learning_rate": 2.822716548702393e-05, + "loss": 0.0136, + "step": 38920 + }, + { + "epoch": 9.04, + "learning_rate": 2.8220143804066963e-05, + "loss": 0.2214, + "step": 38930 + }, + { + "epoch": 9.04, + "learning_rate": 2.821312212110999e-05, + "loss": 0.0416, + "step": 38940 + }, + { + "epoch": 9.04, + "learning_rate": 2.8206100438153017e-05, + "loss": 0.4924, + "step": 38950 + }, + { + "epoch": 9.04, + "learning_rate": 2.8199078755196045e-05, + "loss": 0.1647, + "step": 38960 + }, + { + "epoch": 9.04, + "learning_rate": 2.8192057072239076e-05, + "loss": 0.0019, + "step": 38970 + }, + { + "epoch": 9.04, + "learning_rate": 2.8185035389282107e-05, + "loss": 0.0085, + "step": 38980 + }, + { + "epoch": 9.04, + "learning_rate": 2.8178013706325134e-05, + "loss": 0.6963, + "step": 38990 + }, + { + "epoch": 9.04, + "learning_rate": 2.8170992023368165e-05, + "loss": 0.0062, + "step": 39000 + }, + { + "epoch": 9.04, + "learning_rate": 2.816397034041119e-05, + "loss": 0.1529, + "step": 39010 + }, + { + "epoch": 9.04, + "learning_rate": 2.815694865745422e-05, + "loss": 0.0005, + "step": 39020 + }, + { + "epoch": 9.04, + "learning_rate": 2.8149926974497248e-05, + "loss": 0.0008, + "step": 39030 + }, + { + "epoch": 9.04, + "learning_rate": 2.814290529154028e-05, + "loss": 0.006, + "step": 39040 + }, + { + "epoch": 9.04, + "learning_rate": 2.813588360858331e-05, + "loss": 0.4566, + "step": 39050 + }, + { + "epoch": 9.04, + "learning_rate": 2.8128861925626337e-05, + "loss": 0.452, + "step": 39060 + }, + { + "epoch": 9.04, + "learning_rate": 2.812184024266936e-05, + "loss": 0.1115, + "step": 39070 + }, + { + "epoch": 9.04, + "learning_rate": 2.8114818559712392e-05, + "loss": 0.094, + "step": 39080 + }, + { + "epoch": 9.04, + "learning_rate": 2.8107796876755423e-05, + "loss": 0.0176, + "step": 39090 + }, + { + "epoch": 9.04, + "learning_rate": 2.810077519379845e-05, + "loss": 0.0661, + "step": 39100 + }, + { + "epoch": 9.04, + "learning_rate": 2.809375351084148e-05, + "loss": 0.0001, + "step": 39110 + }, + { + "epoch": 9.04, + "learning_rate": 2.8086731827884512e-05, + "loss": 0.0113, + "step": 39120 + }, + { + "epoch": 9.04, + "learning_rate": 2.807971014492754e-05, + "loss": 0.1888, + "step": 39130 + }, + { + "epoch": 9.04, + "learning_rate": 2.8072688461970564e-05, + "loss": 0.1134, + "step": 39140 + }, + { + "epoch": 9.04, + "learning_rate": 2.8065666779013594e-05, + "loss": 0.0015, + "step": 39150 + }, + { + "epoch": 9.04, + "learning_rate": 2.8058645096056625e-05, + "loss": 0.0503, + "step": 39160 + }, + { + "epoch": 9.05, + "learning_rate": 2.8051623413099653e-05, + "loss": 0.0729, + "step": 39170 + }, + { + "epoch": 9.05, + "learning_rate": 2.8044601730142684e-05, + "loss": 0.0033, + "step": 39180 + }, + { + "epoch": 9.05, + "learning_rate": 2.8037580047185715e-05, + "loss": 0.0535, + "step": 39190 + }, + { + "epoch": 9.05, + "learning_rate": 2.803055836422874e-05, + "loss": 0.7629, + "step": 39200 + }, + { + "epoch": 9.05, + "learning_rate": 2.8023536681271766e-05, + "loss": 0.0009, + "step": 39210 + }, + { + "epoch": 9.05, + "learning_rate": 2.8016514998314797e-05, + "loss": 0.0541, + "step": 39220 + }, + { + "epoch": 9.05, + "learning_rate": 2.8009493315357828e-05, + "loss": 0.1706, + "step": 39230 + }, + { + "epoch": 9.05, + "learning_rate": 2.8002471632400855e-05, + "loss": 0.0003, + "step": 39240 + }, + { + "epoch": 9.05, + "learning_rate": 2.7995449949443886e-05, + "loss": 0.0016, + "step": 39250 + }, + { + "epoch": 9.05, + "learning_rate": 2.798842826648691e-05, + "loss": 0.0012, + "step": 39260 + }, + { + "epoch": 9.05, + "learning_rate": 2.798140658352994e-05, + "loss": 0.1053, + "step": 39270 + }, + { + "epoch": 9.05, + "learning_rate": 2.797438490057297e-05, + "loss": 0.4971, + "step": 39280 + }, + { + "epoch": 9.05, + "learning_rate": 2.7967363217616e-05, + "loss": 0.1326, + "step": 39290 + }, + { + "epoch": 9.05, + "learning_rate": 2.796034153465903e-05, + "loss": 0.2364, + "step": 39300 + }, + { + "epoch": 9.05, + "learning_rate": 2.7953319851702058e-05, + "loss": 0.0059, + "step": 39310 + }, + { + "epoch": 9.05, + "learning_rate": 2.794629816874509e-05, + "loss": 0.1938, + "step": 39320 + }, + { + "epoch": 9.05, + "learning_rate": 2.7939276485788113e-05, + "loss": 0.2057, + "step": 39330 + }, + { + "epoch": 9.05, + "learning_rate": 2.7932254802831144e-05, + "loss": 0.5138, + "step": 39340 + }, + { + "epoch": 9.05, + "learning_rate": 2.792523311987417e-05, + "loss": 0.0001, + "step": 39350 + }, + { + "epoch": 9.05, + "learning_rate": 2.7918211436917202e-05, + "loss": 0.014, + "step": 39360 + }, + { + "epoch": 9.05, + "learning_rate": 2.7911189753960233e-05, + "loss": 0.005, + "step": 39370 + }, + { + "epoch": 9.05, + "learning_rate": 2.790416807100326e-05, + "loss": 0.001, + "step": 39380 + }, + { + "epoch": 9.05, + "learning_rate": 2.7897146388046285e-05, + "loss": 0.0029, + "step": 39390 + }, + { + "epoch": 9.05, + "learning_rate": 2.7890124705089316e-05, + "loss": 0.606, + "step": 39400 + }, + { + "epoch": 9.05, + "learning_rate": 2.7883103022132346e-05, + "loss": 0.0053, + "step": 39410 + }, + { + "epoch": 9.05, + "learning_rate": 2.7876081339175374e-05, + "loss": 0.1722, + "step": 39420 + }, + { + "epoch": 9.05, + "learning_rate": 2.7869059656218405e-05, + "loss": 0.0589, + "step": 39430 + }, + { + "epoch": 9.05, + "learning_rate": 2.7862037973261436e-05, + "loss": 0.3836, + "step": 39440 + }, + { + "epoch": 9.05, + "learning_rate": 2.7855016290304463e-05, + "loss": 0.0001, + "step": 39450 + }, + { + "epoch": 9.05, + "learning_rate": 2.7847994607347487e-05, + "loss": 0.1416, + "step": 39460 + }, + { + "epoch": 9.05, + "learning_rate": 2.7840972924390518e-05, + "loss": 0.331, + "step": 39470 + }, + { + "epoch": 9.05, + "learning_rate": 2.783395124143355e-05, + "loss": 0.0143, + "step": 39480 + }, + { + "epoch": 9.05, + "learning_rate": 2.7826929558476577e-05, + "loss": 0.0002, + "step": 39490 + }, + { + "epoch": 9.05, + "learning_rate": 2.7819907875519607e-05, + "loss": 0.0, + "step": 39500 + }, + { + "epoch": 9.05, + "learning_rate": 2.7812886192562638e-05, + "loss": 0.0002, + "step": 39510 + }, + { + "epoch": 9.05, + "learning_rate": 2.7805864509605662e-05, + "loss": 0.0001, + "step": 39520 + }, + { + "epoch": 9.05, + "learning_rate": 2.779884282664869e-05, + "loss": 0.0002, + "step": 39530 + }, + { + "epoch": 9.05, + "learning_rate": 2.779182114369172e-05, + "loss": 0.0002, + "step": 39540 + }, + { + "epoch": 9.05, + "learning_rate": 2.778479946073475e-05, + "loss": 0.0018, + "step": 39550 + }, + { + "epoch": 9.05, + "learning_rate": 2.777777777777778e-05, + "loss": 0.2192, + "step": 39560 + }, + { + "epoch": 9.05, + "eval_accuracy": 0.7356470132886079, + "eval_loss": 1.7949692010879517, + "eval_runtime": 1269.9259, + "eval_samples_per_second": 6.104, + "eval_steps_per_second": 3.052, + "step": 39560 + }, + { + "epoch": 10.0, + "learning_rate": 2.777075609482081e-05, + "loss": 0.1373, + "step": 39570 + }, + { + "epoch": 10.0, + "learning_rate": 2.7763734411863834e-05, + "loss": 0.4282, + "step": 39580 + }, + { + "epoch": 10.0, + "learning_rate": 2.7756712728906865e-05, + "loss": 0.1095, + "step": 39590 + }, + { + "epoch": 10.0, + "learning_rate": 2.7749691045949892e-05, + "loss": 0.0, + "step": 39600 + }, + { + "epoch": 10.0, + "learning_rate": 2.7742669362992923e-05, + "loss": 0.0066, + "step": 39610 + }, + { + "epoch": 10.0, + "learning_rate": 2.7735647680035954e-05, + "loss": 0.0033, + "step": 39620 + }, + { + "epoch": 10.0, + "learning_rate": 2.7728625997078982e-05, + "loss": 0.236, + "step": 39630 + }, + { + "epoch": 10.0, + "learning_rate": 2.7721604314122013e-05, + "loss": 0.0, + "step": 39640 + }, + { + "epoch": 10.0, + "learning_rate": 2.7714582631165037e-05, + "loss": 0.0001, + "step": 39650 + }, + { + "epoch": 10.0, + "learning_rate": 2.7707560948208068e-05, + "loss": 0.0024, + "step": 39660 + }, + { + "epoch": 10.0, + "learning_rate": 2.7700539265251095e-05, + "loss": 0.0839, + "step": 39670 + }, + { + "epoch": 10.0, + "learning_rate": 2.7693517582294126e-05, + "loss": 0.1489, + "step": 39680 + }, + { + "epoch": 10.0, + "learning_rate": 2.7686495899337157e-05, + "loss": 0.0, + "step": 39690 + }, + { + "epoch": 10.0, + "learning_rate": 2.7679474216380184e-05, + "loss": 0.0287, + "step": 39700 + }, + { + "epoch": 10.0, + "learning_rate": 2.767245253342321e-05, + "loss": 0.0002, + "step": 39710 + }, + { + "epoch": 10.0, + "learning_rate": 2.766543085046624e-05, + "loss": 0.0001, + "step": 39720 + }, + { + "epoch": 10.0, + "learning_rate": 2.765840916750927e-05, + "loss": 0.0529, + "step": 39730 + }, + { + "epoch": 10.0, + "learning_rate": 2.7651387484552298e-05, + "loss": 0.0117, + "step": 39740 + }, + { + "epoch": 10.0, + "learning_rate": 2.764436580159533e-05, + "loss": 0.0301, + "step": 39750 + }, + { + "epoch": 10.0, + "learning_rate": 2.763734411863836e-05, + "loss": 0.0004, + "step": 39760 + }, + { + "epoch": 10.0, + "learning_rate": 2.7630322435681384e-05, + "loss": 0.2174, + "step": 39770 + }, + { + "epoch": 10.0, + "learning_rate": 2.762330075272441e-05, + "loss": 0.0096, + "step": 39780 + }, + { + "epoch": 10.0, + "learning_rate": 2.7616279069767442e-05, + "loss": 0.0123, + "step": 39790 + }, + { + "epoch": 10.0, + "learning_rate": 2.7609257386810473e-05, + "loss": 0.1014, + "step": 39800 + }, + { + "epoch": 10.0, + "learning_rate": 2.76022357038535e-05, + "loss": 0.0001, + "step": 39810 + }, + { + "epoch": 10.0, + "learning_rate": 2.759521402089653e-05, + "loss": 0.0027, + "step": 39820 + }, + { + "epoch": 10.0, + "learning_rate": 2.7588192337939562e-05, + "loss": 0.008, + "step": 39830 + }, + { + "epoch": 10.0, + "learning_rate": 2.7581170654982586e-05, + "loss": 0.0001, + "step": 39840 + }, + { + "epoch": 10.0, + "learning_rate": 2.7574148972025614e-05, + "loss": 0.0639, + "step": 39850 + }, + { + "epoch": 10.0, + "learning_rate": 2.7567127289068644e-05, + "loss": 0.1609, + "step": 39860 + }, + { + "epoch": 10.0, + "learning_rate": 2.7560105606111675e-05, + "loss": 0.0001, + "step": 39870 + }, + { + "epoch": 10.0, + "learning_rate": 2.7553083923154703e-05, + "loss": 0.0001, + "step": 39880 + }, + { + "epoch": 10.0, + "learning_rate": 2.7546062240197734e-05, + "loss": 0.0, + "step": 39890 + }, + { + "epoch": 10.0, + "learning_rate": 2.7539040557240758e-05, + "loss": 0.1924, + "step": 39900 + }, + { + "epoch": 10.0, + "learning_rate": 2.753201887428379e-05, + "loss": 0.099, + "step": 39910 + }, + { + "epoch": 10.0, + "learning_rate": 2.7524997191326816e-05, + "loss": 0.0227, + "step": 39920 + }, + { + "epoch": 10.0, + "learning_rate": 2.7517975508369847e-05, + "loss": 0.181, + "step": 39930 + }, + { + "epoch": 10.0, + "learning_rate": 2.7510953825412878e-05, + "loss": 0.0116, + "step": 39940 + }, + { + "epoch": 10.0, + "learning_rate": 2.7503932142455905e-05, + "loss": 0.0, + "step": 39950 + }, + { + "epoch": 10.01, + "learning_rate": 2.7496910459498933e-05, + "loss": 0.0207, + "step": 39960 + }, + { + "epoch": 10.01, + "learning_rate": 2.748988877654196e-05, + "loss": 0.0604, + "step": 39970 + }, + { + "epoch": 10.01, + "learning_rate": 2.748286709358499e-05, + "loss": 0.0005, + "step": 39980 + }, + { + "epoch": 10.01, + "learning_rate": 2.7475845410628022e-05, + "loss": 0.0002, + "step": 39990 + }, + { + "epoch": 10.01, + "learning_rate": 2.746882372767105e-05, + "loss": 0.2885, + "step": 40000 + }, + { + "epoch": 10.01, + "learning_rate": 2.746180204471408e-05, + "loss": 0.0, + "step": 40010 + }, + { + "epoch": 10.01, + "learning_rate": 2.745478036175711e-05, + "loss": 0.1813, + "step": 40020 + }, + { + "epoch": 10.01, + "learning_rate": 2.7447758678800136e-05, + "loss": 0.0001, + "step": 40030 + }, + { + "epoch": 10.01, + "learning_rate": 2.7440736995843163e-05, + "loss": 0.1275, + "step": 40040 + }, + { + "epoch": 10.01, + "learning_rate": 2.7433715312886194e-05, + "loss": 0.4736, + "step": 40050 + }, + { + "epoch": 10.01, + "learning_rate": 2.7426693629929225e-05, + "loss": 0.1471, + "step": 40060 + }, + { + "epoch": 10.01, + "learning_rate": 2.7419671946972252e-05, + "loss": 0.0009, + "step": 40070 + }, + { + "epoch": 10.01, + "learning_rate": 2.7412650264015283e-05, + "loss": 0.0014, + "step": 40080 + }, + { + "epoch": 10.01, + "learning_rate": 2.7405628581058307e-05, + "loss": 0.0001, + "step": 40090 + }, + { + "epoch": 10.01, + "learning_rate": 2.7398606898101338e-05, + "loss": 0.0003, + "step": 40100 + }, + { + "epoch": 10.01, + "learning_rate": 2.7391585215144366e-05, + "loss": 0.0001, + "step": 40110 + }, + { + "epoch": 10.01, + "learning_rate": 2.7384563532187397e-05, + "loss": 0.152, + "step": 40120 + }, + { + "epoch": 10.01, + "learning_rate": 2.7377541849230427e-05, + "loss": 0.0001, + "step": 40130 + }, + { + "epoch": 10.01, + "learning_rate": 2.7370520166273455e-05, + "loss": 0.0002, + "step": 40140 + }, + { + "epoch": 10.01, + "learning_rate": 2.7363498483316486e-05, + "loss": 0.0599, + "step": 40150 + }, + { + "epoch": 10.01, + "learning_rate": 2.735647680035951e-05, + "loss": 0.1004, + "step": 40160 + }, + { + "epoch": 10.01, + "learning_rate": 2.734945511740254e-05, + "loss": 0.0146, + "step": 40170 + }, + { + "epoch": 10.01, + "learning_rate": 2.7342433434445568e-05, + "loss": 0.0002, + "step": 40180 + }, + { + "epoch": 10.01, + "learning_rate": 2.73354117514886e-05, + "loss": 0.0642, + "step": 40190 + }, + { + "epoch": 10.01, + "learning_rate": 2.732839006853163e-05, + "loss": 0.014, + "step": 40200 + }, + { + "epoch": 10.01, + "learning_rate": 2.7321368385574657e-05, + "loss": 0.0003, + "step": 40210 + }, + { + "epoch": 10.01, + "learning_rate": 2.731434670261768e-05, + "loss": 0.3189, + "step": 40220 + }, + { + "epoch": 10.01, + "learning_rate": 2.7307325019660712e-05, + "loss": 0.0008, + "step": 40230 + }, + { + "epoch": 10.01, + "learning_rate": 2.7300303336703743e-05, + "loss": 0.1914, + "step": 40240 + }, + { + "epoch": 10.01, + "learning_rate": 2.729328165374677e-05, + "loss": 0.0001, + "step": 40250 + }, + { + "epoch": 10.01, + "learning_rate": 2.7286259970789802e-05, + "loss": 0.0, + "step": 40260 + }, + { + "epoch": 10.01, + "learning_rate": 2.7279238287832833e-05, + "loss": 0.0, + "step": 40270 + }, + { + "epoch": 10.01, + "learning_rate": 2.7272216604875857e-05, + "loss": 0.003, + "step": 40280 + }, + { + "epoch": 10.01, + "learning_rate": 2.7265194921918884e-05, + "loss": 0.1549, + "step": 40290 + }, + { + "epoch": 10.01, + "learning_rate": 2.7258173238961915e-05, + "loss": 0.0113, + "step": 40300 + }, + { + "epoch": 10.01, + "learning_rate": 2.7251151556004946e-05, + "loss": 0.0056, + "step": 40310 + }, + { + "epoch": 10.01, + "learning_rate": 2.7244129873047973e-05, + "loss": 0.0728, + "step": 40320 + }, + { + "epoch": 10.01, + "learning_rate": 2.7237108190091004e-05, + "loss": 0.0156, + "step": 40330 + }, + { + "epoch": 10.01, + "learning_rate": 2.7230086507134035e-05, + "loss": 0.0162, + "step": 40340 + }, + { + "epoch": 10.01, + "learning_rate": 2.722306482417706e-05, + "loss": 0.3812, + "step": 40350 + }, + { + "epoch": 10.01, + "learning_rate": 2.7216043141220087e-05, + "loss": 0.0006, + "step": 40360 + }, + { + "epoch": 10.01, + "learning_rate": 2.7209021458263118e-05, + "loss": 0.3794, + "step": 40370 + }, + { + "epoch": 10.01, + "learning_rate": 2.720199977530615e-05, + "loss": 0.001, + "step": 40380 + }, + { + "epoch": 10.01, + "learning_rate": 2.7194978092349176e-05, + "loss": 0.2317, + "step": 40390 + }, + { + "epoch": 10.01, + "learning_rate": 2.7187956409392207e-05, + "loss": 0.0, + "step": 40400 + }, + { + "epoch": 10.01, + "learning_rate": 2.718093472643523e-05, + "loss": 0.8344, + "step": 40410 + }, + { + "epoch": 10.01, + "learning_rate": 2.7173913043478262e-05, + "loss": 0.2935, + "step": 40420 + }, + { + "epoch": 10.01, + "learning_rate": 2.716689136052129e-05, + "loss": 0.4339, + "step": 40430 + }, + { + "epoch": 10.01, + "learning_rate": 2.715986967756432e-05, + "loss": 0.0059, + "step": 40440 + }, + { + "epoch": 10.01, + "learning_rate": 2.715284799460735e-05, + "loss": 0.0001, + "step": 40450 + }, + { + "epoch": 10.01, + "learning_rate": 2.714582631165038e-05, + "loss": 0.0095, + "step": 40460 + }, + { + "epoch": 10.01, + "learning_rate": 2.7138804628693403e-05, + "loss": 0.4542, + "step": 40470 + }, + { + "epoch": 10.01, + "learning_rate": 2.7131782945736434e-05, + "loss": 0.0974, + "step": 40480 + }, + { + "epoch": 10.01, + "learning_rate": 2.7124761262779464e-05, + "loss": 0.0012, + "step": 40490 + }, + { + "epoch": 10.01, + "learning_rate": 2.7117739579822492e-05, + "loss": 0.3787, + "step": 40500 + }, + { + "epoch": 10.01, + "learning_rate": 2.7110717896865523e-05, + "loss": 0.0002, + "step": 40510 + }, + { + "epoch": 10.01, + "learning_rate": 2.7103696213908554e-05, + "loss": 0.0032, + "step": 40520 + }, + { + "epoch": 10.01, + "learning_rate": 2.709667453095158e-05, + "loss": 0.2823, + "step": 40530 + }, + { + "epoch": 10.01, + "learning_rate": 2.7089652847994605e-05, + "loss": 0.0, + "step": 40540 + }, + { + "epoch": 10.01, + "learning_rate": 2.7082631165037636e-05, + "loss": 0.0487, + "step": 40550 + }, + { + "epoch": 10.01, + "learning_rate": 2.7075609482080667e-05, + "loss": 0.0001, + "step": 40560 + }, + { + "epoch": 10.01, + "learning_rate": 2.7068587799123695e-05, + "loss": 0.0, + "step": 40570 + }, + { + "epoch": 10.01, + "learning_rate": 2.7061566116166725e-05, + "loss": 0.0261, + "step": 40580 + }, + { + "epoch": 10.01, + "learning_rate": 2.7054544433209756e-05, + "loss": 0.0094, + "step": 40590 + }, + { + "epoch": 10.01, + "learning_rate": 2.704752275025278e-05, + "loss": 0.0006, + "step": 40600 + }, + { + "epoch": 10.01, + "learning_rate": 2.7040501067295808e-05, + "loss": 0.14, + "step": 40610 + }, + { + "epoch": 10.01, + "learning_rate": 2.703347938433884e-05, + "loss": 0.0002, + "step": 40620 + }, + { + "epoch": 10.01, + "learning_rate": 2.702645770138187e-05, + "loss": 0.0603, + "step": 40630 + }, + { + "epoch": 10.01, + "learning_rate": 2.7019436018424897e-05, + "loss": 0.112, + "step": 40640 + }, + { + "epoch": 10.01, + "learning_rate": 2.7012414335467928e-05, + "loss": 0.4157, + "step": 40650 + }, + { + "epoch": 10.01, + "learning_rate": 2.7005392652510952e-05, + "loss": 0.0006, + "step": 40660 + }, + { + "epoch": 10.01, + "learning_rate": 2.6998370969553983e-05, + "loss": 0.0419, + "step": 40670 + }, + { + "epoch": 10.01, + "learning_rate": 2.699134928659701e-05, + "loss": 0.0001, + "step": 40680 + }, + { + "epoch": 10.01, + "learning_rate": 2.698432760364004e-05, + "loss": 0.0639, + "step": 40690 + }, + { + "epoch": 10.01, + "learning_rate": 2.6977305920683072e-05, + "loss": 0.0002, + "step": 40700 + }, + { + "epoch": 10.01, + "learning_rate": 2.69702842377261e-05, + "loss": 0.5044, + "step": 40710 + }, + { + "epoch": 10.01, + "learning_rate": 2.696326255476913e-05, + "loss": 0.1466, + "step": 40720 + }, + { + "epoch": 10.01, + "learning_rate": 2.6956240871812155e-05, + "loss": 0.0108, + "step": 40730 + }, + { + "epoch": 10.01, + "learning_rate": 2.6949219188855186e-05, + "loss": 0.0, + "step": 40740 + }, + { + "epoch": 10.02, + "learning_rate": 2.6942197505898213e-05, + "loss": 0.0287, + "step": 40750 + }, + { + "epoch": 10.02, + "learning_rate": 2.6935175822941244e-05, + "loss": 0.7738, + "step": 40760 + }, + { + "epoch": 10.02, + "learning_rate": 2.6928154139984275e-05, + "loss": 0.5562, + "step": 40770 + }, + { + "epoch": 10.02, + "learning_rate": 2.6921132457027302e-05, + "loss": 0.0331, + "step": 40780 + }, + { + "epoch": 10.02, + "learning_rate": 2.6914110774070326e-05, + "loss": 0.0002, + "step": 40790 + }, + { + "epoch": 10.02, + "learning_rate": 2.6907089091113357e-05, + "loss": 0.473, + "step": 40800 + }, + { + "epoch": 10.02, + "learning_rate": 2.6900067408156388e-05, + "loss": 0.1137, + "step": 40810 + }, + { + "epoch": 10.02, + "learning_rate": 2.6893045725199416e-05, + "loss": 0.1053, + "step": 40820 + }, + { + "epoch": 10.02, + "learning_rate": 2.6886024042242447e-05, + "loss": 0.0, + "step": 40830 + }, + { + "epoch": 10.02, + "learning_rate": 2.6879002359285477e-05, + "loss": 0.1829, + "step": 40840 + }, + { + "epoch": 10.02, + "learning_rate": 2.6871980676328505e-05, + "loss": 0.0104, + "step": 40850 + }, + { + "epoch": 10.02, + "learning_rate": 2.686495899337153e-05, + "loss": 0.2175, + "step": 40860 + }, + { + "epoch": 10.02, + "learning_rate": 2.685793731041456e-05, + "loss": 0.7501, + "step": 40870 + }, + { + "epoch": 10.02, + "learning_rate": 2.685091562745759e-05, + "loss": 0.0003, + "step": 40880 + }, + { + "epoch": 10.02, + "learning_rate": 2.6843893944500618e-05, + "loss": 0.0161, + "step": 40890 + }, + { + "epoch": 10.02, + "learning_rate": 2.683687226154365e-05, + "loss": 0.0001, + "step": 40900 + }, + { + "epoch": 10.02, + "learning_rate": 2.682985057858668e-05, + "loss": 0.0001, + "step": 40910 + }, + { + "epoch": 10.02, + "learning_rate": 2.6822828895629704e-05, + "loss": 0.3804, + "step": 40920 + }, + { + "epoch": 10.02, + "learning_rate": 2.681580721267273e-05, + "loss": 0.3669, + "step": 40930 + }, + { + "epoch": 10.02, + "learning_rate": 2.6808785529715763e-05, + "loss": 0.0001, + "step": 40940 + }, + { + "epoch": 10.02, + "learning_rate": 2.6801763846758793e-05, + "loss": 0.0004, + "step": 40950 + }, + { + "epoch": 10.02, + "learning_rate": 2.679474216380182e-05, + "loss": 0.0613, + "step": 40960 + }, + { + "epoch": 10.02, + "learning_rate": 2.6787720480844852e-05, + "loss": 0.1143, + "step": 40970 + }, + { + "epoch": 10.02, + "learning_rate": 2.6780698797887876e-05, + "loss": 0.0304, + "step": 40980 + }, + { + "epoch": 10.02, + "learning_rate": 2.6773677114930907e-05, + "loss": 0.1842, + "step": 40990 + }, + { + "epoch": 10.02, + "learning_rate": 2.6766655431973938e-05, + "loss": 0.0001, + "step": 41000 + }, + { + "epoch": 10.02, + "learning_rate": 2.6759633749016965e-05, + "loss": 0.0438, + "step": 41010 + }, + { + "epoch": 10.02, + "learning_rate": 2.6752612066059996e-05, + "loss": 0.0, + "step": 41020 + }, + { + "epoch": 10.02, + "learning_rate": 2.6745590383103027e-05, + "loss": 0.0001, + "step": 41030 + }, + { + "epoch": 10.02, + "learning_rate": 2.6738568700146054e-05, + "loss": 0.0, + "step": 41040 + }, + { + "epoch": 10.02, + "learning_rate": 2.673154701718908e-05, + "loss": 0.0115, + "step": 41050 + }, + { + "epoch": 10.02, + "learning_rate": 2.672452533423211e-05, + "loss": 0.0286, + "step": 41060 + }, + { + "epoch": 10.02, + "learning_rate": 2.671750365127514e-05, + "loss": 0.0008, + "step": 41070 + }, + { + "epoch": 10.02, + "learning_rate": 2.6710481968318168e-05, + "loss": 0.1852, + "step": 41080 + }, + { + "epoch": 10.02, + "learning_rate": 2.67034602853612e-05, + "loss": 0.0, + "step": 41090 + }, + { + "epoch": 10.02, + "learning_rate": 2.669643860240423e-05, + "loss": 0.0033, + "step": 41100 + }, + { + "epoch": 10.02, + "learning_rate": 2.6689416919447254e-05, + "loss": 0.9245, + "step": 41110 + }, + { + "epoch": 10.02, + "learning_rate": 2.668239523649028e-05, + "loss": 0.046, + "step": 41120 + }, + { + "epoch": 10.02, + "learning_rate": 2.6675373553533312e-05, + "loss": 0.004, + "step": 41130 + }, + { + "epoch": 10.02, + "learning_rate": 2.6668351870576343e-05, + "loss": 0.0025, + "step": 41140 + }, + { + "epoch": 10.02, + "learning_rate": 2.666133018761937e-05, + "loss": 0.0, + "step": 41150 + }, + { + "epoch": 10.02, + "learning_rate": 2.66543085046624e-05, + "loss": 0.0018, + "step": 41160 + }, + { + "epoch": 10.02, + "learning_rate": 2.6647286821705425e-05, + "loss": 0.0022, + "step": 41170 + }, + { + "epoch": 10.02, + "learning_rate": 2.6640265138748456e-05, + "loss": 0.0011, + "step": 41180 + }, + { + "epoch": 10.02, + "learning_rate": 2.6633243455791484e-05, + "loss": 0.0013, + "step": 41190 + }, + { + "epoch": 10.02, + "learning_rate": 2.6626221772834515e-05, + "loss": 0.1989, + "step": 41200 + }, + { + "epoch": 10.02, + "learning_rate": 2.6619200089877545e-05, + "loss": 0.0001, + "step": 41210 + }, + { + "epoch": 10.02, + "learning_rate": 2.6612178406920573e-05, + "loss": 0.0002, + "step": 41220 + }, + { + "epoch": 10.02, + "learning_rate": 2.6605156723963604e-05, + "loss": 0.0068, + "step": 41230 + }, + { + "epoch": 10.02, + "learning_rate": 2.6598135041006628e-05, + "loss": 0.1272, + "step": 41240 + }, + { + "epoch": 10.02, + "learning_rate": 2.659111335804966e-05, + "loss": 0.3693, + "step": 41250 + }, + { + "epoch": 10.02, + "learning_rate": 2.6584091675092686e-05, + "loss": 0.0142, + "step": 41260 + }, + { + "epoch": 10.02, + "learning_rate": 2.6577069992135717e-05, + "loss": 0.0212, + "step": 41270 + }, + { + "epoch": 10.02, + "learning_rate": 2.6570048309178748e-05, + "loss": 0.3833, + "step": 41280 + }, + { + "epoch": 10.02, + "learning_rate": 2.6563026626221776e-05, + "loss": 0.0, + "step": 41290 + }, + { + "epoch": 10.02, + "learning_rate": 2.65560049432648e-05, + "loss": 0.0377, + "step": 41300 + }, + { + "epoch": 10.02, + "learning_rate": 2.654898326030783e-05, + "loss": 0.0, + "step": 41310 + }, + { + "epoch": 10.02, + "learning_rate": 2.654196157735086e-05, + "loss": 0.0381, + "step": 41320 + }, + { + "epoch": 10.02, + "learning_rate": 2.653493989439389e-05, + "loss": 0.0002, + "step": 41330 + }, + { + "epoch": 10.02, + "learning_rate": 2.652791821143692e-05, + "loss": 0.9546, + "step": 41340 + }, + { + "epoch": 10.02, + "learning_rate": 2.652089652847995e-05, + "loss": 0.0001, + "step": 41350 + }, + { + "epoch": 10.02, + "learning_rate": 2.6513874845522975e-05, + "loss": 0.1766, + "step": 41360 + }, + { + "epoch": 10.02, + "learning_rate": 2.6506853162566002e-05, + "loss": 0.397, + "step": 41370 + }, + { + "epoch": 10.02, + "learning_rate": 2.6499831479609033e-05, + "loss": 0.011, + "step": 41380 + }, + { + "epoch": 10.02, + "learning_rate": 2.6492809796652064e-05, + "loss": 0.8399, + "step": 41390 + }, + { + "epoch": 10.02, + "learning_rate": 2.648578811369509e-05, + "loss": 0.413, + "step": 41400 + }, + { + "epoch": 10.02, + "learning_rate": 2.6478766430738122e-05, + "loss": 0.0017, + "step": 41410 + }, + { + "epoch": 10.02, + "learning_rate": 2.6471744747781153e-05, + "loss": 0.0018, + "step": 41420 + }, + { + "epoch": 10.02, + "learning_rate": 2.6464723064824177e-05, + "loss": 0.2884, + "step": 41430 + }, + { + "epoch": 10.02, + "learning_rate": 2.6457701381867205e-05, + "loss": 0.3766, + "step": 41440 + }, + { + "epoch": 10.02, + "learning_rate": 2.6450679698910236e-05, + "loss": 0.0256, + "step": 41450 + }, + { + "epoch": 10.02, + "learning_rate": 2.6443658015953267e-05, + "loss": 0.0003, + "step": 41460 + }, + { + "epoch": 10.02, + "learning_rate": 2.6436636332996294e-05, + "loss": 0.0002, + "step": 41470 + }, + { + "epoch": 10.02, + "learning_rate": 2.6429614650039325e-05, + "loss": 0.1291, + "step": 41480 + }, + { + "epoch": 10.02, + "learning_rate": 2.642259296708235e-05, + "loss": 0.0, + "step": 41490 + }, + { + "epoch": 10.02, + "learning_rate": 2.641557128412538e-05, + "loss": 0.002, + "step": 41500 + }, + { + "epoch": 10.02, + "learning_rate": 2.6408549601168407e-05, + "loss": 0.0729, + "step": 41510 + }, + { + "epoch": 10.02, + "learning_rate": 2.6401527918211438e-05, + "loss": 0.0065, + "step": 41520 + }, + { + "epoch": 10.02, + "learning_rate": 2.639450623525447e-05, + "loss": 0.003, + "step": 41530 + }, + { + "epoch": 10.03, + "learning_rate": 2.6387484552297497e-05, + "loss": 0.0002, + "step": 41540 + }, + { + "epoch": 10.03, + "learning_rate": 2.6380462869340528e-05, + "loss": 0.0038, + "step": 41550 + }, + { + "epoch": 10.03, + "learning_rate": 2.637344118638355e-05, + "loss": 0.3201, + "step": 41560 + }, + { + "epoch": 10.03, + "learning_rate": 2.6366419503426583e-05, + "loss": 0.0001, + "step": 41570 + }, + { + "epoch": 10.03, + "learning_rate": 2.635939782046961e-05, + "loss": 0.0003, + "step": 41580 + }, + { + "epoch": 10.03, + "learning_rate": 2.635237613751264e-05, + "loss": 0.0004, + "step": 41590 + }, + { + "epoch": 10.03, + "learning_rate": 2.6345354454555672e-05, + "loss": 0.1528, + "step": 41600 + }, + { + "epoch": 10.03, + "learning_rate": 2.63383327715987e-05, + "loss": 0.5348, + "step": 41610 + }, + { + "epoch": 10.03, + "learning_rate": 2.6331311088641723e-05, + "loss": 0.6717, + "step": 41620 + }, + { + "epoch": 10.03, + "learning_rate": 2.6324289405684754e-05, + "loss": 0.0001, + "step": 41630 + }, + { + "epoch": 10.03, + "learning_rate": 2.6317267722727785e-05, + "loss": 0.0027, + "step": 41640 + }, + { + "epoch": 10.03, + "learning_rate": 2.6310246039770813e-05, + "loss": 0.1722, + "step": 41650 + }, + { + "epoch": 10.03, + "learning_rate": 2.6303224356813843e-05, + "loss": 0.0018, + "step": 41660 + }, + { + "epoch": 10.03, + "learning_rate": 2.6296202673856874e-05, + "loss": 0.0001, + "step": 41670 + }, + { + "epoch": 10.03, + "learning_rate": 2.62891809908999e-05, + "loss": 0.5277, + "step": 41680 + }, + { + "epoch": 10.03, + "learning_rate": 2.6282159307942926e-05, + "loss": 0.0024, + "step": 41690 + }, + { + "epoch": 10.03, + "learning_rate": 2.6275137624985957e-05, + "loss": 0.1112, + "step": 41700 + }, + { + "epoch": 10.03, + "learning_rate": 2.6268115942028988e-05, + "loss": 0.1065, + "step": 41710 + }, + { + "epoch": 10.03, + "learning_rate": 2.6261094259072015e-05, + "loss": 0.005, + "step": 41720 + }, + { + "epoch": 10.03, + "learning_rate": 2.6254072576115046e-05, + "loss": 0.0001, + "step": 41730 + }, + { + "epoch": 10.03, + "learning_rate": 2.6247050893158077e-05, + "loss": 0.0004, + "step": 41740 + }, + { + "epoch": 10.03, + "learning_rate": 2.62400292102011e-05, + "loss": 0.0001, + "step": 41750 + }, + { + "epoch": 10.03, + "learning_rate": 2.623300752724413e-05, + "loss": 0.0, + "step": 41760 + }, + { + "epoch": 10.03, + "learning_rate": 2.622598584428716e-05, + "loss": 0.0, + "step": 41770 + }, + { + "epoch": 10.03, + "learning_rate": 2.621896416133019e-05, + "loss": 0.0, + "step": 41780 + }, + { + "epoch": 10.03, + "learning_rate": 2.6211942478373218e-05, + "loss": 0.0004, + "step": 41790 + }, + { + "epoch": 10.03, + "learning_rate": 2.620492079541625e-05, + "loss": 0.9106, + "step": 41800 + }, + { + "epoch": 10.03, + "learning_rate": 2.6197899112459273e-05, + "loss": 0.1026, + "step": 41810 + }, + { + "epoch": 10.03, + "learning_rate": 2.6190877429502304e-05, + "loss": 0.1634, + "step": 41820 + }, + { + "epoch": 10.03, + "learning_rate": 2.618385574654533e-05, + "loss": 0.0144, + "step": 41830 + }, + { + "epoch": 10.03, + "learning_rate": 2.6176834063588362e-05, + "loss": 0.008, + "step": 41840 + }, + { + "epoch": 10.03, + "learning_rate": 2.6169812380631393e-05, + "loss": 0.0, + "step": 41850 + }, + { + "epoch": 10.03, + "learning_rate": 2.616279069767442e-05, + "loss": 0.0001, + "step": 41860 + }, + { + "epoch": 10.03, + "learning_rate": 2.6155769014717445e-05, + "loss": 0.0011, + "step": 41870 + }, + { + "epoch": 10.03, + "learning_rate": 2.6148747331760475e-05, + "loss": 1.1024, + "step": 41880 + }, + { + "epoch": 10.03, + "learning_rate": 2.6141725648803506e-05, + "loss": 0.5252, + "step": 41890 + }, + { + "epoch": 10.03, + "learning_rate": 2.6134703965846534e-05, + "loss": 0.0087, + "step": 41900 + }, + { + "epoch": 10.03, + "learning_rate": 2.6127682282889565e-05, + "loss": 0.0004, + "step": 41910 + }, + { + "epoch": 10.03, + "learning_rate": 2.6120660599932596e-05, + "loss": 0.0138, + "step": 41920 + }, + { + "epoch": 10.03, + "learning_rate": 2.6113638916975623e-05, + "loss": 0.161, + "step": 41930 + }, + { + "epoch": 10.03, + "learning_rate": 2.6106617234018647e-05, + "loss": 0.0054, + "step": 41940 + }, + { + "epoch": 10.03, + "learning_rate": 2.6099595551061678e-05, + "loss": 0.234, + "step": 41950 + }, + { + "epoch": 10.03, + "learning_rate": 2.609257386810471e-05, + "loss": 0.0002, + "step": 41960 + }, + { + "epoch": 10.03, + "learning_rate": 2.6085552185147736e-05, + "loss": 0.0001, + "step": 41970 + }, + { + "epoch": 10.03, + "learning_rate": 2.6078530502190767e-05, + "loss": 0.0001, + "step": 41980 + }, + { + "epoch": 10.03, + "learning_rate": 2.6071508819233798e-05, + "loss": 0.3436, + "step": 41990 + }, + { + "epoch": 10.03, + "learning_rate": 2.6064487136276822e-05, + "loss": 0.0003, + "step": 42000 + }, + { + "epoch": 10.03, + "learning_rate": 2.6057465453319853e-05, + "loss": 0.0823, + "step": 42010 + }, + { + "epoch": 10.03, + "learning_rate": 2.605044377036288e-05, + "loss": 0.5461, + "step": 42020 + }, + { + "epoch": 10.03, + "learning_rate": 2.604342208740591e-05, + "loss": 0.1521, + "step": 42030 + }, + { + "epoch": 10.03, + "learning_rate": 2.6036400404448942e-05, + "loss": 0.0143, + "step": 42040 + }, + { + "epoch": 10.03, + "learning_rate": 2.602937872149197e-05, + "loss": 0.3366, + "step": 42050 + }, + { + "epoch": 10.03, + "learning_rate": 2.6022357038534994e-05, + "loss": 0.0017, + "step": 42060 + }, + { + "epoch": 10.03, + "learning_rate": 2.6015335355578025e-05, + "loss": 0.0222, + "step": 42070 + }, + { + "epoch": 10.03, + "learning_rate": 2.6008313672621056e-05, + "loss": 0.0001, + "step": 42080 + }, + { + "epoch": 10.03, + "learning_rate": 2.6001291989664083e-05, + "loss": 0.0124, + "step": 42090 + }, + { + "epoch": 10.03, + "learning_rate": 2.5994270306707114e-05, + "loss": 0.0426, + "step": 42100 + }, + { + "epoch": 10.03, + "learning_rate": 2.5987248623750145e-05, + "loss": 0.0015, + "step": 42110 + }, + { + "epoch": 10.03, + "learning_rate": 2.5980226940793172e-05, + "loss": 0.0, + "step": 42120 + }, + { + "epoch": 10.03, + "learning_rate": 2.5973205257836197e-05, + "loss": 0.0102, + "step": 42130 + }, + { + "epoch": 10.03, + "learning_rate": 2.5966183574879227e-05, + "loss": 0.0001, + "step": 42140 + }, + { + "epoch": 10.03, + "learning_rate": 2.5959161891922258e-05, + "loss": 0.0, + "step": 42150 + }, + { + "epoch": 10.03, + "learning_rate": 2.5952140208965286e-05, + "loss": 0.0008, + "step": 42160 + }, + { + "epoch": 10.03, + "learning_rate": 2.5945118526008317e-05, + "loss": 0.879, + "step": 42170 + }, + { + "epoch": 10.03, + "learning_rate": 2.5938096843051348e-05, + "loss": 0.0, + "step": 42180 + }, + { + "epoch": 10.03, + "learning_rate": 2.593107516009437e-05, + "loss": 0.0011, + "step": 42190 + }, + { + "epoch": 10.03, + "learning_rate": 2.59240534771374e-05, + "loss": 0.067, + "step": 42200 + }, + { + "epoch": 10.03, + "learning_rate": 2.591703179418043e-05, + "loss": 0.1197, + "step": 42210 + }, + { + "epoch": 10.03, + "learning_rate": 2.591001011122346e-05, + "loss": 0.047, + "step": 42220 + }, + { + "epoch": 10.03, + "learning_rate": 2.590298842826649e-05, + "loss": 0.0001, + "step": 42230 + }, + { + "epoch": 10.03, + "learning_rate": 2.589596674530952e-05, + "loss": 0.0022, + "step": 42240 + }, + { + "epoch": 10.03, + "learning_rate": 2.5888945062352543e-05, + "loss": 0.0207, + "step": 42250 + }, + { + "epoch": 10.03, + "learning_rate": 2.5881923379395574e-05, + "loss": 0.0187, + "step": 42260 + }, + { + "epoch": 10.03, + "learning_rate": 2.5874901696438602e-05, + "loss": 0.0001, + "step": 42270 + }, + { + "epoch": 10.03, + "learning_rate": 2.5867880013481633e-05, + "loss": 0.0649, + "step": 42280 + }, + { + "epoch": 10.03, + "learning_rate": 2.5860858330524663e-05, + "loss": 0.3642, + "step": 42290 + }, + { + "epoch": 10.03, + "learning_rate": 2.585383664756769e-05, + "loss": 0.0777, + "step": 42300 + }, + { + "epoch": 10.03, + "learning_rate": 2.5846814964610722e-05, + "loss": 0.0679, + "step": 42310 + }, + { + "epoch": 10.03, + "learning_rate": 2.5839793281653746e-05, + "loss": 0.0003, + "step": 42320 + }, + { + "epoch": 10.04, + "learning_rate": 2.5832771598696777e-05, + "loss": 0.0003, + "step": 42330 + }, + { + "epoch": 10.04, + "learning_rate": 2.5825749915739804e-05, + "loss": 0.0012, + "step": 42340 + }, + { + "epoch": 10.04, + "learning_rate": 2.5818728232782835e-05, + "loss": 0.1295, + "step": 42350 + }, + { + "epoch": 10.04, + "learning_rate": 2.5811706549825866e-05, + "loss": 0.0005, + "step": 42360 + }, + { + "epoch": 10.04, + "learning_rate": 2.5804684866868894e-05, + "loss": 0.0724, + "step": 42370 + }, + { + "epoch": 10.04, + "learning_rate": 2.5797663183911918e-05, + "loss": 0.005, + "step": 42380 + }, + { + "epoch": 10.04, + "learning_rate": 2.579064150095495e-05, + "loss": 0.0078, + "step": 42390 + }, + { + "epoch": 10.04, + "learning_rate": 2.578361981799798e-05, + "loss": 0.0004, + "step": 42400 + }, + { + "epoch": 10.04, + "learning_rate": 2.5776598135041007e-05, + "loss": 0.4343, + "step": 42410 + }, + { + "epoch": 10.04, + "learning_rate": 2.5769576452084038e-05, + "loss": 0.0175, + "step": 42420 + }, + { + "epoch": 10.04, + "learning_rate": 2.576255476912707e-05, + "loss": 0.0029, + "step": 42430 + }, + { + "epoch": 10.04, + "learning_rate": 2.5755533086170096e-05, + "loss": 0.0, + "step": 42440 + }, + { + "epoch": 10.04, + "learning_rate": 2.574851140321312e-05, + "loss": 0.0004, + "step": 42450 + }, + { + "epoch": 10.04, + "learning_rate": 2.574148972025615e-05, + "loss": 0.0, + "step": 42460 + }, + { + "epoch": 10.04, + "learning_rate": 2.5734468037299182e-05, + "loss": 0.0001, + "step": 42470 + }, + { + "epoch": 10.04, + "learning_rate": 2.572744635434221e-05, + "loss": 0.0087, + "step": 42480 + }, + { + "epoch": 10.04, + "learning_rate": 2.572042467138524e-05, + "loss": 0.0153, + "step": 42490 + }, + { + "epoch": 10.04, + "learning_rate": 2.571340298842827e-05, + "loss": 0.2635, + "step": 42500 + }, + { + "epoch": 10.04, + "learning_rate": 2.5706381305471295e-05, + "loss": 0.4586, + "step": 42510 + }, + { + "epoch": 10.04, + "learning_rate": 2.5699359622514323e-05, + "loss": 0.0712, + "step": 42520 + }, + { + "epoch": 10.04, + "learning_rate": 2.5692337939557354e-05, + "loss": 0.3717, + "step": 42530 + }, + { + "epoch": 10.04, + "learning_rate": 2.5685316256600385e-05, + "loss": 0.0707, + "step": 42540 + }, + { + "epoch": 10.04, + "learning_rate": 2.5678294573643412e-05, + "loss": 0.0531, + "step": 42550 + }, + { + "epoch": 10.04, + "learning_rate": 2.5671272890686443e-05, + "loss": 0.0002, + "step": 42560 + }, + { + "epoch": 10.04, + "learning_rate": 2.5664251207729467e-05, + "loss": 0.7325, + "step": 42570 + }, + { + "epoch": 10.04, + "learning_rate": 2.5657229524772498e-05, + "loss": 0.0446, + "step": 42580 + }, + { + "epoch": 10.04, + "learning_rate": 2.5650207841815525e-05, + "loss": 0.0057, + "step": 42590 + }, + { + "epoch": 10.04, + "learning_rate": 2.5643186158858556e-05, + "loss": 0.0003, + "step": 42600 + }, + { + "epoch": 10.04, + "learning_rate": 2.5636164475901587e-05, + "loss": 0.2722, + "step": 42610 + }, + { + "epoch": 10.04, + "learning_rate": 2.5629142792944615e-05, + "loss": 0.001, + "step": 42620 + }, + { + "epoch": 10.04, + "learning_rate": 2.5622121109987646e-05, + "loss": 0.0, + "step": 42630 + }, + { + "epoch": 10.04, + "learning_rate": 2.561509942703067e-05, + "loss": 0.0001, + "step": 42640 + }, + { + "epoch": 10.04, + "learning_rate": 2.56080777440737e-05, + "loss": 0.0005, + "step": 42650 + }, + { + "epoch": 10.04, + "learning_rate": 2.5601056061116728e-05, + "loss": 0.0006, + "step": 42660 + }, + { + "epoch": 10.04, + "learning_rate": 2.559403437815976e-05, + "loss": 0.0004, + "step": 42670 + }, + { + "epoch": 10.04, + "learning_rate": 2.558701269520279e-05, + "loss": 0.0112, + "step": 42680 + }, + { + "epoch": 10.04, + "learning_rate": 2.5579991012245817e-05, + "loss": 0.3039, + "step": 42690 + }, + { + "epoch": 10.04, + "learning_rate": 2.557296932928884e-05, + "loss": 0.0, + "step": 42700 + }, + { + "epoch": 10.04, + "learning_rate": 2.5565947646331872e-05, + "loss": 0.0, + "step": 42710 + }, + { + "epoch": 10.04, + "learning_rate": 2.5558925963374903e-05, + "loss": 0.117, + "step": 42720 + }, + { + "epoch": 10.04, + "learning_rate": 2.555190428041793e-05, + "loss": 0.0002, + "step": 42730 + }, + { + "epoch": 10.04, + "learning_rate": 2.554488259746096e-05, + "loss": 0.0005, + "step": 42740 + }, + { + "epoch": 10.04, + "learning_rate": 2.5537860914503992e-05, + "loss": 0.0, + "step": 42750 + }, + { + "epoch": 10.04, + "learning_rate": 2.5530839231547017e-05, + "loss": 0.0003, + "step": 42760 + }, + { + "epoch": 10.04, + "learning_rate": 2.5523817548590044e-05, + "loss": 0.3189, + "step": 42770 + }, + { + "epoch": 10.04, + "learning_rate": 2.5516795865633075e-05, + "loss": 0.0006, + "step": 42780 + }, + { + "epoch": 10.04, + "learning_rate": 2.5509774182676106e-05, + "loss": 0.0, + "step": 42790 + }, + { + "epoch": 10.04, + "learning_rate": 2.5502752499719133e-05, + "loss": 0.0001, + "step": 42800 + }, + { + "epoch": 10.04, + "learning_rate": 2.5495730816762164e-05, + "loss": 0.0003, + "step": 42810 + }, + { + "epoch": 10.04, + "learning_rate": 2.5488709133805195e-05, + "loss": 0.1221, + "step": 42820 + }, + { + "epoch": 10.04, + "learning_rate": 2.548168745084822e-05, + "loss": 0.0011, + "step": 42830 + }, + { + "epoch": 10.04, + "learning_rate": 2.5474665767891247e-05, + "loss": 0.4467, + "step": 42840 + }, + { + "epoch": 10.04, + "learning_rate": 2.5467644084934277e-05, + "loss": 0.347, + "step": 42850 + }, + { + "epoch": 10.04, + "learning_rate": 2.546062240197731e-05, + "loss": 0.2797, + "step": 42860 + }, + { + "epoch": 10.04, + "learning_rate": 2.5453600719020336e-05, + "loss": 0.0, + "step": 42870 + }, + { + "epoch": 10.04, + "learning_rate": 2.5446579036063367e-05, + "loss": 0.1823, + "step": 42880 + }, + { + "epoch": 10.04, + "learning_rate": 2.543955735310639e-05, + "loss": 0.011, + "step": 42890 + }, + { + "epoch": 10.04, + "learning_rate": 2.5432535670149422e-05, + "loss": 0.0026, + "step": 42900 + }, + { + "epoch": 10.04, + "learning_rate": 2.542551398719245e-05, + "loss": 0.0004, + "step": 42910 + }, + { + "epoch": 10.04, + "learning_rate": 2.541849230423548e-05, + "loss": 0.0402, + "step": 42920 + }, + { + "epoch": 10.04, + "learning_rate": 2.541147062127851e-05, + "loss": 0.001, + "step": 42930 + }, + { + "epoch": 10.04, + "learning_rate": 2.540444893832154e-05, + "loss": 0.156, + "step": 42940 + }, + { + "epoch": 10.04, + "learning_rate": 2.5397427255364563e-05, + "loss": 0.0, + "step": 42950 + }, + { + "epoch": 10.04, + "learning_rate": 2.5390405572407593e-05, + "loss": 0.0267, + "step": 42960 + }, + { + "epoch": 10.04, + "learning_rate": 2.5383383889450624e-05, + "loss": 0.0016, + "step": 42970 + }, + { + "epoch": 10.04, + "learning_rate": 2.5376362206493652e-05, + "loss": 0.0072, + "step": 42980 + }, + { + "epoch": 10.04, + "learning_rate": 2.5369340523536683e-05, + "loss": 0.0045, + "step": 42990 + }, + { + "epoch": 10.04, + "learning_rate": 2.5362318840579714e-05, + "loss": 0.0001, + "step": 43000 + }, + { + "epoch": 10.04, + "learning_rate": 2.535529715762274e-05, + "loss": 0.0, + "step": 43010 + }, + { + "epoch": 10.04, + "learning_rate": 2.534827547466577e-05, + "loss": 0.3149, + "step": 43020 + }, + { + "epoch": 10.04, + "learning_rate": 2.5341253791708796e-05, + "loss": 0.4275, + "step": 43030 + }, + { + "epoch": 10.04, + "learning_rate": 2.5334232108751827e-05, + "loss": 0.0001, + "step": 43040 + }, + { + "epoch": 10.04, + "learning_rate": 2.5327210425794858e-05, + "loss": 0.0015, + "step": 43050 + }, + { + "epoch": 10.04, + "learning_rate": 2.5320188742837885e-05, + "loss": 0.184, + "step": 43060 + }, + { + "epoch": 10.04, + "learning_rate": 2.5313167059880916e-05, + "loss": 0.2994, + "step": 43070 + }, + { + "epoch": 10.04, + "learning_rate": 2.530614537692394e-05, + "loss": 0.0254, + "step": 43080 + }, + { + "epoch": 10.04, + "learning_rate": 2.529912369396697e-05, + "loss": 0.0002, + "step": 43090 + }, + { + "epoch": 10.04, + "learning_rate": 2.529210201101e-05, + "loss": 0.2108, + "step": 43100 + }, + { + "epoch": 10.04, + "learning_rate": 2.528508032805303e-05, + "loss": 0.2323, + "step": 43110 + }, + { + "epoch": 10.04, + "learning_rate": 2.527805864509606e-05, + "loss": 0.0, + "step": 43120 + }, + { + "epoch": 10.05, + "learning_rate": 2.5271036962139088e-05, + "loss": 0.4947, + "step": 43130 + }, + { + "epoch": 10.05, + "learning_rate": 2.526401527918212e-05, + "loss": 0.2767, + "step": 43140 + }, + { + "epoch": 10.05, + "learning_rate": 2.5256993596225143e-05, + "loss": 0.0011, + "step": 43150 + }, + { + "epoch": 10.05, + "learning_rate": 2.5249971913268174e-05, + "loss": 0.2476, + "step": 43160 + }, + { + "epoch": 10.05, + "learning_rate": 2.52429502303112e-05, + "loss": 0.0003, + "step": 43170 + }, + { + "epoch": 10.05, + "learning_rate": 2.5235928547354232e-05, + "loss": 0.0637, + "step": 43180 + }, + { + "epoch": 10.05, + "learning_rate": 2.5228906864397263e-05, + "loss": 0.1102, + "step": 43190 + }, + { + "epoch": 10.05, + "learning_rate": 2.522188518144029e-05, + "loss": 0.0009, + "step": 43200 + }, + { + "epoch": 10.05, + "learning_rate": 2.5214863498483315e-05, + "loss": 0.1155, + "step": 43210 + }, + { + "epoch": 10.05, + "learning_rate": 2.5207841815526345e-05, + "loss": 0.0, + "step": 43220 + }, + { + "epoch": 10.05, + "learning_rate": 2.5200820132569376e-05, + "loss": 0.0, + "step": 43230 + }, + { + "epoch": 10.05, + "learning_rate": 2.5193798449612404e-05, + "loss": 0.0003, + "step": 43240 + }, + { + "epoch": 10.05, + "learning_rate": 2.5186776766655435e-05, + "loss": 0.005, + "step": 43250 + }, + { + "epoch": 10.05, + "learning_rate": 2.5179755083698466e-05, + "loss": 0.0004, + "step": 43260 + }, + { + "epoch": 10.05, + "learning_rate": 2.517273340074149e-05, + "loss": 0.0004, + "step": 43270 + }, + { + "epoch": 10.05, + "learning_rate": 2.5165711717784517e-05, + "loss": 0.0002, + "step": 43280 + }, + { + "epoch": 10.05, + "learning_rate": 2.5158690034827548e-05, + "loss": 0.3934, + "step": 43290 + }, + { + "epoch": 10.05, + "learning_rate": 2.515166835187058e-05, + "loss": 0.0, + "step": 43300 + }, + { + "epoch": 10.05, + "learning_rate": 2.5144646668913606e-05, + "loss": 0.0299, + "step": 43310 + }, + { + "epoch": 10.05, + "learning_rate": 2.5137624985956637e-05, + "loss": 0.2304, + "step": 43320 + }, + { + "epoch": 10.05, + "learning_rate": 2.5130603302999668e-05, + "loss": 0.0308, + "step": 43330 + }, + { + "epoch": 10.05, + "learning_rate": 2.5123581620042692e-05, + "loss": 0.004, + "step": 43340 + }, + { + "epoch": 10.05, + "learning_rate": 2.511655993708572e-05, + "loss": 0.0001, + "step": 43350 + }, + { + "epoch": 10.05, + "learning_rate": 2.510953825412875e-05, + "loss": 0.5118, + "step": 43360 + }, + { + "epoch": 10.05, + "learning_rate": 2.510251657117178e-05, + "loss": 0.0001, + "step": 43370 + }, + { + "epoch": 10.05, + "learning_rate": 2.509549488821481e-05, + "loss": 0.3302, + "step": 43380 + }, + { + "epoch": 10.05, + "learning_rate": 2.508847320525784e-05, + "loss": 0.0567, + "step": 43390 + }, + { + "epoch": 10.05, + "learning_rate": 2.5081451522300864e-05, + "loss": 0.0006, + "step": 43400 + }, + { + "epoch": 10.05, + "learning_rate": 2.5074429839343895e-05, + "loss": 0.0486, + "step": 43410 + }, + { + "epoch": 10.05, + "learning_rate": 2.5067408156386922e-05, + "loss": 0.0801, + "step": 43420 + }, + { + "epoch": 10.05, + "learning_rate": 2.5060386473429953e-05, + "loss": 0.0002, + "step": 43430 + }, + { + "epoch": 10.05, + "learning_rate": 2.5053364790472984e-05, + "loss": 0.0024, + "step": 43440 + }, + { + "epoch": 10.05, + "learning_rate": 2.504634310751601e-05, + "loss": 0.1085, + "step": 43450 + }, + { + "epoch": 10.05, + "learning_rate": 2.5039321424559036e-05, + "loss": 0.0001, + "step": 43460 + }, + { + "epoch": 10.05, + "learning_rate": 2.5032299741602067e-05, + "loss": 0.0001, + "step": 43470 + }, + { + "epoch": 10.05, + "learning_rate": 2.5025278058645097e-05, + "loss": 0.0022, + "step": 43480 + }, + { + "epoch": 10.05, + "learning_rate": 2.5018256375688125e-05, + "loss": 0.0001, + "step": 43490 + }, + { + "epoch": 10.05, + "learning_rate": 2.5011234692731156e-05, + "loss": 0.0098, + "step": 43500 + }, + { + "epoch": 10.05, + "learning_rate": 2.5004213009774187e-05, + "loss": 0.1505, + "step": 43510 + }, + { + "epoch": 10.05, + "eval_accuracy": 0.7160366404334925, + "eval_loss": 1.948011875152588, + "eval_runtime": 1265.7946, + "eval_samples_per_second": 6.123, + "eval_steps_per_second": 3.062, + "step": 43516 + }, + { + "epoch": 11.0, + "learning_rate": 2.4997191326817214e-05, + "loss": 0.0583, + "step": 43520 + }, + { + "epoch": 11.0, + "learning_rate": 2.4990169643860242e-05, + "loss": 0.0, + "step": 43530 + }, + { + "epoch": 11.0, + "learning_rate": 2.498314796090327e-05, + "loss": 0.4449, + "step": 43540 + }, + { + "epoch": 11.0, + "learning_rate": 2.49761262779463e-05, + "loss": 0.001, + "step": 43550 + }, + { + "epoch": 11.0, + "learning_rate": 2.4969104594989328e-05, + "loss": 0.0151, + "step": 43560 + }, + { + "epoch": 11.0, + "learning_rate": 2.496208291203236e-05, + "loss": 0.0037, + "step": 43570 + }, + { + "epoch": 11.0, + "learning_rate": 2.4955061229075386e-05, + "loss": 0.0041, + "step": 43580 + }, + { + "epoch": 11.0, + "learning_rate": 2.4948039546118417e-05, + "loss": 0.0, + "step": 43590 + }, + { + "epoch": 11.0, + "learning_rate": 2.4941017863161444e-05, + "loss": 0.3306, + "step": 43600 + }, + { + "epoch": 11.0, + "learning_rate": 2.4933996180204472e-05, + "loss": 0.0001, + "step": 43610 + }, + { + "epoch": 11.0, + "learning_rate": 2.4926974497247503e-05, + "loss": 0.0006, + "step": 43620 + }, + { + "epoch": 11.0, + "learning_rate": 2.491995281429053e-05, + "loss": 0.0005, + "step": 43630 + }, + { + "epoch": 11.0, + "learning_rate": 2.4912931131333558e-05, + "loss": 0.0014, + "step": 43640 + }, + { + "epoch": 11.0, + "learning_rate": 2.490590944837659e-05, + "loss": 0.0025, + "step": 43650 + }, + { + "epoch": 11.0, + "learning_rate": 2.489888776541962e-05, + "loss": 0.0, + "step": 43660 + }, + { + "epoch": 11.0, + "learning_rate": 2.4891866082462644e-05, + "loss": 0.0853, + "step": 43670 + }, + { + "epoch": 11.0, + "learning_rate": 2.4884844399505674e-05, + "loss": 0.0001, + "step": 43680 + }, + { + "epoch": 11.0, + "learning_rate": 2.4877822716548705e-05, + "loss": 0.2855, + "step": 43690 + }, + { + "epoch": 11.0, + "learning_rate": 2.4870801033591733e-05, + "loss": 0.0795, + "step": 43700 + }, + { + "epoch": 11.0, + "learning_rate": 2.486377935063476e-05, + "loss": 0.2941, + "step": 43710 + }, + { + "epoch": 11.0, + "learning_rate": 2.485675766767779e-05, + "loss": 0.2563, + "step": 43720 + }, + { + "epoch": 11.0, + "learning_rate": 2.484973598472082e-05, + "loss": 0.0033, + "step": 43730 + }, + { + "epoch": 11.0, + "learning_rate": 2.4842714301763846e-05, + "loss": 0.0199, + "step": 43740 + }, + { + "epoch": 11.0, + "learning_rate": 2.4835692618806877e-05, + "loss": 0.0001, + "step": 43750 + }, + { + "epoch": 11.0, + "learning_rate": 2.4828670935849908e-05, + "loss": 0.0002, + "step": 43760 + }, + { + "epoch": 11.0, + "learning_rate": 2.4821649252892935e-05, + "loss": 0.0004, + "step": 43770 + }, + { + "epoch": 11.0, + "learning_rate": 2.4814627569935963e-05, + "loss": 0.4987, + "step": 43780 + }, + { + "epoch": 11.0, + "learning_rate": 2.4807605886978994e-05, + "loss": 0.0001, + "step": 43790 + }, + { + "epoch": 11.0, + "learning_rate": 2.480058420402202e-05, + "loss": 0.0029, + "step": 43800 + }, + { + "epoch": 11.0, + "learning_rate": 2.479356252106505e-05, + "loss": 0.0031, + "step": 43810 + }, + { + "epoch": 11.0, + "learning_rate": 2.478654083810808e-05, + "loss": 0.0034, + "step": 43820 + }, + { + "epoch": 11.0, + "learning_rate": 2.4779519155151107e-05, + "loss": 0.038, + "step": 43830 + }, + { + "epoch": 11.0, + "learning_rate": 2.4772497472194138e-05, + "loss": 0.4361, + "step": 43840 + }, + { + "epoch": 11.0, + "learning_rate": 2.4765475789237165e-05, + "loss": 0.0923, + "step": 43850 + }, + { + "epoch": 11.0, + "learning_rate": 2.4758454106280193e-05, + "loss": 0.0361, + "step": 43860 + }, + { + "epoch": 11.0, + "learning_rate": 2.4751432423323224e-05, + "loss": 0.0002, + "step": 43870 + }, + { + "epoch": 11.0, + "learning_rate": 2.474441074036625e-05, + "loss": 0.0001, + "step": 43880 + }, + { + "epoch": 11.0, + "learning_rate": 2.473738905740928e-05, + "loss": 0.0034, + "step": 43890 + }, + { + "epoch": 11.0, + "learning_rate": 2.473036737445231e-05, + "loss": 0.0, + "step": 43900 + }, + { + "epoch": 11.0, + "learning_rate": 2.472334569149534e-05, + "loss": 0.0576, + "step": 43910 + }, + { + "epoch": 11.01, + "learning_rate": 2.4716324008538368e-05, + "loss": 0.0002, + "step": 43920 + }, + { + "epoch": 11.01, + "learning_rate": 2.4709302325581396e-05, + "loss": 0.0007, + "step": 43930 + }, + { + "epoch": 11.01, + "learning_rate": 2.4702280642624426e-05, + "loss": 0.0, + "step": 43940 + }, + { + "epoch": 11.01, + "learning_rate": 2.4695258959667454e-05, + "loss": 0.0001, + "step": 43950 + }, + { + "epoch": 11.01, + "learning_rate": 2.468823727671048e-05, + "loss": 0.6739, + "step": 43960 + }, + { + "epoch": 11.01, + "learning_rate": 2.4681215593753512e-05, + "loss": 0.0865, + "step": 43970 + }, + { + "epoch": 11.01, + "learning_rate": 2.4674193910796543e-05, + "loss": 0.0031, + "step": 43980 + }, + { + "epoch": 11.01, + "learning_rate": 2.4667172227839567e-05, + "loss": 0.0003, + "step": 43990 + }, + { + "epoch": 11.01, + "learning_rate": 2.4660150544882598e-05, + "loss": 0.0028, + "step": 44000 + }, + { + "epoch": 11.01, + "learning_rate": 2.465312886192563e-05, + "loss": 0.0, + "step": 44010 + }, + { + "epoch": 11.01, + "learning_rate": 2.4646107178968657e-05, + "loss": 0.7629, + "step": 44020 + }, + { + "epoch": 11.01, + "learning_rate": 2.4639085496011684e-05, + "loss": 0.1707, + "step": 44030 + }, + { + "epoch": 11.01, + "learning_rate": 2.4632063813054715e-05, + "loss": 0.0837, + "step": 44040 + }, + { + "epoch": 11.01, + "learning_rate": 2.4625042130097742e-05, + "loss": 0.0081, + "step": 44050 + }, + { + "epoch": 11.01, + "learning_rate": 2.4618020447140773e-05, + "loss": 0.0002, + "step": 44060 + }, + { + "epoch": 11.01, + "learning_rate": 2.46109987641838e-05, + "loss": 0.1646, + "step": 44070 + }, + { + "epoch": 11.01, + "learning_rate": 2.4603977081226828e-05, + "loss": 0.0001, + "step": 44080 + }, + { + "epoch": 11.01, + "learning_rate": 2.459695539826986e-05, + "loss": 0.0015, + "step": 44090 + }, + { + "epoch": 11.01, + "learning_rate": 2.4589933715312887e-05, + "loss": 0.0, + "step": 44100 + }, + { + "epoch": 11.01, + "learning_rate": 2.4582912032355917e-05, + "loss": 0.0001, + "step": 44110 + }, + { + "epoch": 11.01, + "learning_rate": 2.4575890349398945e-05, + "loss": 0.0185, + "step": 44120 + }, + { + "epoch": 11.01, + "learning_rate": 2.4568868666441976e-05, + "loss": 0.4935, + "step": 44130 + }, + { + "epoch": 11.01, + "learning_rate": 2.4561846983485003e-05, + "loss": 0.0055, + "step": 44140 + }, + { + "epoch": 11.01, + "learning_rate": 2.455482530052803e-05, + "loss": 0.0377, + "step": 44150 + }, + { + "epoch": 11.01, + "learning_rate": 2.4547803617571062e-05, + "loss": 0.0968, + "step": 44160 + }, + { + "epoch": 11.01, + "learning_rate": 2.454078193461409e-05, + "loss": 0.0981, + "step": 44170 + }, + { + "epoch": 11.01, + "learning_rate": 2.4533760251657117e-05, + "loss": 0.0039, + "step": 44180 + }, + { + "epoch": 11.01, + "learning_rate": 2.4526738568700148e-05, + "loss": 0.0001, + "step": 44190 + }, + { + "epoch": 11.01, + "learning_rate": 2.451971688574318e-05, + "loss": 0.0001, + "step": 44200 + }, + { + "epoch": 11.01, + "learning_rate": 2.4512695202786203e-05, + "loss": 0.0345, + "step": 44210 + }, + { + "epoch": 11.01, + "learning_rate": 2.4505673519829233e-05, + "loss": 0.0, + "step": 44220 + }, + { + "epoch": 11.01, + "learning_rate": 2.4498651836872264e-05, + "loss": 0.0002, + "step": 44230 + }, + { + "epoch": 11.01, + "learning_rate": 2.4491630153915292e-05, + "loss": 0.0042, + "step": 44240 + }, + { + "epoch": 11.01, + "learning_rate": 2.448460847095832e-05, + "loss": 0.0007, + "step": 44250 + }, + { + "epoch": 11.01, + "learning_rate": 2.447758678800135e-05, + "loss": 0.0006, + "step": 44260 + }, + { + "epoch": 11.01, + "learning_rate": 2.4470565105044378e-05, + "loss": 0.0002, + "step": 44270 + }, + { + "epoch": 11.01, + "learning_rate": 2.4463543422087405e-05, + "loss": 0.0066, + "step": 44280 + }, + { + "epoch": 11.01, + "learning_rate": 2.4456521739130436e-05, + "loss": 0.0055, + "step": 44290 + }, + { + "epoch": 11.01, + "learning_rate": 2.4449500056173467e-05, + "loss": 0.4205, + "step": 44300 + }, + { + "epoch": 11.01, + "learning_rate": 2.4442478373216494e-05, + "loss": 0.0348, + "step": 44310 + }, + { + "epoch": 11.01, + "learning_rate": 2.4435456690259522e-05, + "loss": 0.0373, + "step": 44320 + }, + { + "epoch": 11.01, + "learning_rate": 2.4428435007302553e-05, + "loss": 0.0, + "step": 44330 + }, + { + "epoch": 11.01, + "learning_rate": 2.442141332434558e-05, + "loss": 0.0001, + "step": 44340 + }, + { + "epoch": 11.01, + "learning_rate": 2.4414391641388608e-05, + "loss": 0.1353, + "step": 44350 + }, + { + "epoch": 11.01, + "learning_rate": 2.440736995843164e-05, + "loss": 0.0116, + "step": 44360 + }, + { + "epoch": 11.01, + "learning_rate": 2.4400348275474666e-05, + "loss": 0.0005, + "step": 44370 + }, + { + "epoch": 11.01, + "learning_rate": 2.4393326592517697e-05, + "loss": 0.1134, + "step": 44380 + }, + { + "epoch": 11.01, + "learning_rate": 2.4386304909560724e-05, + "loss": 0.328, + "step": 44390 + }, + { + "epoch": 11.01, + "learning_rate": 2.4379283226603752e-05, + "loss": 0.0081, + "step": 44400 + }, + { + "epoch": 11.01, + "learning_rate": 2.4372261543646783e-05, + "loss": 0.0008, + "step": 44410 + }, + { + "epoch": 11.01, + "learning_rate": 2.436523986068981e-05, + "loss": 0.0772, + "step": 44420 + }, + { + "epoch": 11.01, + "learning_rate": 2.4358218177732838e-05, + "loss": 0.009, + "step": 44430 + }, + { + "epoch": 11.01, + "learning_rate": 2.435119649477587e-05, + "loss": 0.0004, + "step": 44440 + }, + { + "epoch": 11.01, + "learning_rate": 2.43441748118189e-05, + "loss": 0.0003, + "step": 44450 + }, + { + "epoch": 11.01, + "learning_rate": 2.4337153128861927e-05, + "loss": 0.0035, + "step": 44460 + }, + { + "epoch": 11.01, + "learning_rate": 2.4330131445904955e-05, + "loss": 0.461, + "step": 44470 + }, + { + "epoch": 11.01, + "learning_rate": 2.4323109762947985e-05, + "loss": 0.0, + "step": 44480 + }, + { + "epoch": 11.01, + "learning_rate": 2.4316088079991013e-05, + "loss": 0.0003, + "step": 44490 + }, + { + "epoch": 11.01, + "learning_rate": 2.430906639703404e-05, + "loss": 0.0002, + "step": 44500 + }, + { + "epoch": 11.01, + "learning_rate": 2.430204471407707e-05, + "loss": 0.0, + "step": 44510 + }, + { + "epoch": 11.01, + "learning_rate": 2.4295023031120102e-05, + "loss": 0.0562, + "step": 44520 + }, + { + "epoch": 11.01, + "learning_rate": 2.428800134816313e-05, + "loss": 0.0007, + "step": 44530 + }, + { + "epoch": 11.01, + "learning_rate": 2.4280979665206157e-05, + "loss": 0.2862, + "step": 44540 + }, + { + "epoch": 11.01, + "learning_rate": 2.4273957982249188e-05, + "loss": 0.0513, + "step": 44550 + }, + { + "epoch": 11.01, + "learning_rate": 2.4266936299292216e-05, + "loss": 0.0, + "step": 44560 + }, + { + "epoch": 11.01, + "learning_rate": 2.4259914616335243e-05, + "loss": 0.1455, + "step": 44570 + }, + { + "epoch": 11.01, + "learning_rate": 2.4252892933378274e-05, + "loss": 0.0012, + "step": 44580 + }, + { + "epoch": 11.01, + "learning_rate": 2.42458712504213e-05, + "loss": 0.354, + "step": 44590 + }, + { + "epoch": 11.01, + "learning_rate": 2.4238849567464332e-05, + "loss": 0.0003, + "step": 44600 + }, + { + "epoch": 11.01, + "learning_rate": 2.423182788450736e-05, + "loss": 0.1954, + "step": 44610 + }, + { + "epoch": 11.01, + "learning_rate": 2.4224806201550387e-05, + "loss": 0.1426, + "step": 44620 + }, + { + "epoch": 11.01, + "learning_rate": 2.4217784518593418e-05, + "loss": 0.0069, + "step": 44630 + }, + { + "epoch": 11.01, + "learning_rate": 2.4210762835636446e-05, + "loss": 0.2261, + "step": 44640 + }, + { + "epoch": 11.01, + "learning_rate": 2.4203741152679477e-05, + "loss": 0.3166, + "step": 44650 + }, + { + "epoch": 11.01, + "learning_rate": 2.4196719469722504e-05, + "loss": 0.0027, + "step": 44660 + }, + { + "epoch": 11.01, + "learning_rate": 2.4189697786765535e-05, + "loss": 0.0086, + "step": 44670 + }, + { + "epoch": 11.01, + "learning_rate": 2.4182676103808562e-05, + "loss": 0.0001, + "step": 44680 + }, + { + "epoch": 11.01, + "learning_rate": 2.417565442085159e-05, + "loss": 0.0, + "step": 44690 + }, + { + "epoch": 11.01, + "learning_rate": 2.416863273789462e-05, + "loss": 0.0082, + "step": 44700 + }, + { + "epoch": 11.02, + "learning_rate": 2.4161611054937648e-05, + "loss": 0.1848, + "step": 44710 + }, + { + "epoch": 11.02, + "learning_rate": 2.4154589371980676e-05, + "loss": 0.2063, + "step": 44720 + }, + { + "epoch": 11.02, + "learning_rate": 2.4147567689023707e-05, + "loss": 0.3568, + "step": 44730 + }, + { + "epoch": 11.02, + "learning_rate": 2.4140546006066737e-05, + "loss": 0.0005, + "step": 44740 + }, + { + "epoch": 11.02, + "learning_rate": 2.413352432310976e-05, + "loss": 0.276, + "step": 44750 + }, + { + "epoch": 11.02, + "learning_rate": 2.4126502640152792e-05, + "loss": 0.2834, + "step": 44760 + }, + { + "epoch": 11.02, + "learning_rate": 2.4119480957195823e-05, + "loss": 0.0387, + "step": 44770 + }, + { + "epoch": 11.02, + "learning_rate": 2.411245927423885e-05, + "loss": 0.0003, + "step": 44780 + }, + { + "epoch": 11.02, + "learning_rate": 2.410543759128188e-05, + "loss": 0.2552, + "step": 44790 + }, + { + "epoch": 11.02, + "learning_rate": 2.409841590832491e-05, + "loss": 0.0075, + "step": 44800 + }, + { + "epoch": 11.02, + "learning_rate": 2.409139422536794e-05, + "loss": 0.0001, + "step": 44810 + }, + { + "epoch": 11.02, + "learning_rate": 2.4084372542410964e-05, + "loss": 0.0583, + "step": 44820 + }, + { + "epoch": 11.02, + "learning_rate": 2.4077350859453995e-05, + "loss": 0.0, + "step": 44830 + }, + { + "epoch": 11.02, + "learning_rate": 2.4070329176497026e-05, + "loss": 0.619, + "step": 44840 + }, + { + "epoch": 11.02, + "learning_rate": 2.4063307493540053e-05, + "loss": 0.0001, + "step": 44850 + }, + { + "epoch": 11.02, + "learning_rate": 2.405628581058308e-05, + "loss": 0.0001, + "step": 44860 + }, + { + "epoch": 11.02, + "learning_rate": 2.4049264127626112e-05, + "loss": 0.2141, + "step": 44870 + }, + { + "epoch": 11.02, + "learning_rate": 2.404224244466914e-05, + "loss": 0.0, + "step": 44880 + }, + { + "epoch": 11.02, + "learning_rate": 2.4035220761712167e-05, + "loss": 0.0003, + "step": 44890 + }, + { + "epoch": 11.02, + "learning_rate": 2.4028199078755198e-05, + "loss": 0.0, + "step": 44900 + }, + { + "epoch": 11.02, + "learning_rate": 2.4021177395798225e-05, + "loss": 0.2695, + "step": 44910 + }, + { + "epoch": 11.02, + "learning_rate": 2.4014155712841256e-05, + "loss": 0.0002, + "step": 44920 + }, + { + "epoch": 11.02, + "learning_rate": 2.4007134029884284e-05, + "loss": 0.0059, + "step": 44930 + }, + { + "epoch": 11.02, + "learning_rate": 2.400011234692731e-05, + "loss": 0.4904, + "step": 44940 + }, + { + "epoch": 11.02, + "learning_rate": 2.3993090663970342e-05, + "loss": 0.0, + "step": 44950 + }, + { + "epoch": 11.02, + "learning_rate": 2.398606898101337e-05, + "loss": 0.0001, + "step": 44960 + }, + { + "epoch": 11.02, + "learning_rate": 2.3979047298056397e-05, + "loss": 0.1884, + "step": 44970 + }, + { + "epoch": 11.02, + "learning_rate": 2.3972025615099428e-05, + "loss": 0.0241, + "step": 44980 + }, + { + "epoch": 11.02, + "learning_rate": 2.396500393214246e-05, + "loss": 0.4218, + "step": 44990 + }, + { + "epoch": 11.02, + "learning_rate": 2.3957982249185486e-05, + "loss": 0.0254, + "step": 45000 + }, + { + "epoch": 11.02, + "learning_rate": 2.3950960566228514e-05, + "loss": 0.4745, + "step": 45010 + }, + { + "epoch": 11.02, + "learning_rate": 2.3943938883271544e-05, + "loss": 0.4155, + "step": 45020 + }, + { + "epoch": 11.02, + "learning_rate": 2.3936917200314572e-05, + "loss": 0.0096, + "step": 45030 + }, + { + "epoch": 11.02, + "learning_rate": 2.39298955173576e-05, + "loss": 0.0002, + "step": 45040 + }, + { + "epoch": 11.02, + "learning_rate": 2.392287383440063e-05, + "loss": 0.0002, + "step": 45050 + }, + { + "epoch": 11.02, + "learning_rate": 2.391585215144366e-05, + "loss": 0.0003, + "step": 45060 + }, + { + "epoch": 11.02, + "learning_rate": 2.390883046848669e-05, + "loss": 0.0007, + "step": 45070 + }, + { + "epoch": 11.02, + "learning_rate": 2.3901808785529716e-05, + "loss": 0.0187, + "step": 45080 + }, + { + "epoch": 11.02, + "learning_rate": 2.3894787102572747e-05, + "loss": 0.3417, + "step": 45090 + }, + { + "epoch": 11.02, + "learning_rate": 2.3887765419615775e-05, + "loss": 0.3386, + "step": 45100 + }, + { + "epoch": 11.02, + "learning_rate": 2.3880743736658802e-05, + "loss": 0.0009, + "step": 45110 + }, + { + "epoch": 11.02, + "learning_rate": 2.3873722053701833e-05, + "loss": 0.0022, + "step": 45120 + }, + { + "epoch": 11.02, + "learning_rate": 2.386670037074486e-05, + "loss": 0.0136, + "step": 45130 + }, + { + "epoch": 11.02, + "learning_rate": 2.385967868778789e-05, + "loss": 0.0007, + "step": 45140 + }, + { + "epoch": 11.02, + "learning_rate": 2.385265700483092e-05, + "loss": 0.0005, + "step": 45150 + }, + { + "epoch": 11.02, + "learning_rate": 2.384563532187395e-05, + "loss": 0.0038, + "step": 45160 + }, + { + "epoch": 11.02, + "learning_rate": 2.3838613638916977e-05, + "loss": 0.4275, + "step": 45170 + }, + { + "epoch": 11.02, + "learning_rate": 2.3831591955960005e-05, + "loss": 0.0009, + "step": 45180 + }, + { + "epoch": 11.02, + "learning_rate": 2.3824570273003036e-05, + "loss": 0.0053, + "step": 45190 + }, + { + "epoch": 11.02, + "learning_rate": 2.3817548590046063e-05, + "loss": 0.1584, + "step": 45200 + }, + { + "epoch": 11.02, + "learning_rate": 2.3810526907089094e-05, + "loss": 0.1051, + "step": 45210 + }, + { + "epoch": 11.02, + "learning_rate": 2.380350522413212e-05, + "loss": 0.0714, + "step": 45220 + }, + { + "epoch": 11.02, + "learning_rate": 2.379648354117515e-05, + "loss": 0.1053, + "step": 45230 + }, + { + "epoch": 11.02, + "learning_rate": 2.378946185821818e-05, + "loss": 0.2469, + "step": 45240 + }, + { + "epoch": 11.02, + "learning_rate": 2.3782440175261207e-05, + "loss": 0.0, + "step": 45250 + }, + { + "epoch": 11.02, + "learning_rate": 2.3775418492304235e-05, + "loss": 0.0, + "step": 45260 + }, + { + "epoch": 11.02, + "learning_rate": 2.3768396809347266e-05, + "loss": 0.2903, + "step": 45270 + }, + { + "epoch": 11.02, + "learning_rate": 2.3761375126390297e-05, + "loss": 0.0005, + "step": 45280 + }, + { + "epoch": 11.02, + "learning_rate": 2.375435344343332e-05, + "loss": 0.1252, + "step": 45290 + }, + { + "epoch": 11.02, + "learning_rate": 2.374733176047635e-05, + "loss": 0.1793, + "step": 45300 + }, + { + "epoch": 11.02, + "learning_rate": 2.3740310077519382e-05, + "loss": 0.002, + "step": 45310 + }, + { + "epoch": 11.02, + "learning_rate": 2.373328839456241e-05, + "loss": 0.0004, + "step": 45320 + }, + { + "epoch": 11.02, + "learning_rate": 2.3726266711605437e-05, + "loss": 0.0, + "step": 45330 + }, + { + "epoch": 11.02, + "learning_rate": 2.3719245028648468e-05, + "loss": 0.0001, + "step": 45340 + }, + { + "epoch": 11.02, + "learning_rate": 2.37122233456915e-05, + "loss": 0.0221, + "step": 45350 + }, + { + "epoch": 11.02, + "learning_rate": 2.3705201662734523e-05, + "loss": 0.0, + "step": 45360 + }, + { + "epoch": 11.02, + "learning_rate": 2.3698179979777554e-05, + "loss": 0.0001, + "step": 45370 + }, + { + "epoch": 11.02, + "learning_rate": 2.3691158296820585e-05, + "loss": 0.0386, + "step": 45380 + }, + { + "epoch": 11.02, + "learning_rate": 2.3684136613863612e-05, + "loss": 0.0694, + "step": 45390 + }, + { + "epoch": 11.02, + "learning_rate": 2.367711493090664e-05, + "loss": 0.2517, + "step": 45400 + }, + { + "epoch": 11.02, + "learning_rate": 2.367009324794967e-05, + "loss": 0.0029, + "step": 45410 + }, + { + "epoch": 11.02, + "learning_rate": 2.36630715649927e-05, + "loss": 0.4179, + "step": 45420 + }, + { + "epoch": 11.02, + "learning_rate": 2.3656049882035726e-05, + "loss": 0.0009, + "step": 45430 + }, + { + "epoch": 11.02, + "learning_rate": 2.3649028199078757e-05, + "loss": 0.0004, + "step": 45440 + }, + { + "epoch": 11.02, + "learning_rate": 2.3642006516121784e-05, + "loss": 0.0003, + "step": 45450 + }, + { + "epoch": 11.02, + "learning_rate": 2.3634984833164815e-05, + "loss": 0.0002, + "step": 45460 + }, + { + "epoch": 11.02, + "learning_rate": 2.3627963150207843e-05, + "loss": 0.0052, + "step": 45470 + }, + { + "epoch": 11.02, + "learning_rate": 2.362094146725087e-05, + "loss": 0.0001, + "step": 45480 + }, + { + "epoch": 11.02, + "learning_rate": 2.36139197842939e-05, + "loss": 0.413, + "step": 45490 + }, + { + "epoch": 11.03, + "learning_rate": 2.360689810133693e-05, + "loss": 0.3637, + "step": 45500 + }, + { + "epoch": 11.03, + "learning_rate": 2.359987641837996e-05, + "loss": 0.3883, + "step": 45510 + }, + { + "epoch": 11.03, + "learning_rate": 2.3592854735422987e-05, + "loss": 0.0019, + "step": 45520 + }, + { + "epoch": 11.03, + "learning_rate": 2.3585833052466018e-05, + "loss": 0.3931, + "step": 45530 + }, + { + "epoch": 11.03, + "learning_rate": 2.3578811369509045e-05, + "loss": 0.2699, + "step": 45540 + }, + { + "epoch": 11.03, + "learning_rate": 2.3571789686552073e-05, + "loss": 0.0036, + "step": 45550 + }, + { + "epoch": 11.03, + "learning_rate": 2.3564768003595104e-05, + "loss": 0.1574, + "step": 45560 + }, + { + "epoch": 11.03, + "learning_rate": 2.3557746320638134e-05, + "loss": 0.0322, + "step": 45570 + }, + { + "epoch": 11.03, + "learning_rate": 2.355072463768116e-05, + "loss": 0.4254, + "step": 45580 + }, + { + "epoch": 11.03, + "learning_rate": 2.354370295472419e-05, + "loss": 0.0003, + "step": 45590 + }, + { + "epoch": 11.03, + "learning_rate": 2.353668127176722e-05, + "loss": 0.0653, + "step": 45600 + }, + { + "epoch": 11.03, + "learning_rate": 2.3529659588810248e-05, + "loss": 0.0002, + "step": 45610 + }, + { + "epoch": 11.03, + "learning_rate": 2.3522637905853275e-05, + "loss": 0.0, + "step": 45620 + }, + { + "epoch": 11.03, + "learning_rate": 2.3515616222896306e-05, + "loss": 0.0034, + "step": 45630 + }, + { + "epoch": 11.03, + "learning_rate": 2.3508594539939334e-05, + "loss": 0.332, + "step": 45640 + }, + { + "epoch": 11.03, + "learning_rate": 2.350157285698236e-05, + "loss": 0.5208, + "step": 45650 + }, + { + "epoch": 11.03, + "learning_rate": 2.3494551174025392e-05, + "loss": 0.004, + "step": 45660 + }, + { + "epoch": 11.03, + "learning_rate": 2.348752949106842e-05, + "loss": 0.0001, + "step": 45670 + }, + { + "epoch": 11.03, + "learning_rate": 2.348050780811145e-05, + "loss": 0.1595, + "step": 45680 + }, + { + "epoch": 11.03, + "learning_rate": 2.3473486125154478e-05, + "loss": 0.4135, + "step": 45690 + }, + { + "epoch": 11.03, + "learning_rate": 2.346646444219751e-05, + "loss": 0.0207, + "step": 45700 + }, + { + "epoch": 11.03, + "learning_rate": 2.3459442759240536e-05, + "loss": 0.0001, + "step": 45710 + }, + { + "epoch": 11.03, + "learning_rate": 2.3452421076283564e-05, + "loss": 0.0, + "step": 45720 + }, + { + "epoch": 11.03, + "learning_rate": 2.3445399393326595e-05, + "loss": 0.0001, + "step": 45730 + }, + { + "epoch": 11.03, + "learning_rate": 2.3438377710369622e-05, + "loss": 0.2679, + "step": 45740 + }, + { + "epoch": 11.03, + "learning_rate": 2.3431356027412653e-05, + "loss": 0.0006, + "step": 45750 + }, + { + "epoch": 11.03, + "learning_rate": 2.342433434445568e-05, + "loss": 0.0003, + "step": 45760 + }, + { + "epoch": 11.03, + "learning_rate": 2.3417312661498708e-05, + "loss": 0.0001, + "step": 45770 + }, + { + "epoch": 11.03, + "learning_rate": 2.341029097854174e-05, + "loss": 0.0005, + "step": 45780 + }, + { + "epoch": 11.03, + "learning_rate": 2.3403269295584766e-05, + "loss": 0.0026, + "step": 45790 + }, + { + "epoch": 11.03, + "learning_rate": 2.3396247612627794e-05, + "loss": 0.0073, + "step": 45800 + }, + { + "epoch": 11.03, + "learning_rate": 2.3389225929670825e-05, + "loss": 0.0058, + "step": 45810 + }, + { + "epoch": 11.03, + "learning_rate": 2.3382204246713856e-05, + "loss": 0.4276, + "step": 45820 + }, + { + "epoch": 11.03, + "learning_rate": 2.337518256375688e-05, + "loss": 0.0005, + "step": 45830 + }, + { + "epoch": 11.03, + "learning_rate": 2.336816088079991e-05, + "loss": 0.005, + "step": 45840 + }, + { + "epoch": 11.03, + "learning_rate": 2.336113919784294e-05, + "loss": 0.0, + "step": 45850 + }, + { + "epoch": 11.03, + "learning_rate": 2.335411751488597e-05, + "loss": 0.0003, + "step": 45860 + }, + { + "epoch": 11.03, + "learning_rate": 2.3347095831928996e-05, + "loss": 0.0191, + "step": 45870 + }, + { + "epoch": 11.03, + "learning_rate": 2.3340074148972027e-05, + "loss": 0.0087, + "step": 45880 + }, + { + "epoch": 11.03, + "learning_rate": 2.3333052466015058e-05, + "loss": 0.0033, + "step": 45890 + }, + { + "epoch": 11.03, + "learning_rate": 2.3326030783058082e-05, + "loss": 0.0767, + "step": 45900 + }, + { + "epoch": 11.03, + "learning_rate": 2.3319009100101113e-05, + "loss": 0.0003, + "step": 45910 + }, + { + "epoch": 11.03, + "learning_rate": 2.3311987417144144e-05, + "loss": 0.3487, + "step": 45920 + }, + { + "epoch": 11.03, + "learning_rate": 2.330496573418717e-05, + "loss": 0.0011, + "step": 45930 + }, + { + "epoch": 11.03, + "learning_rate": 2.32979440512302e-05, + "loss": 0.4928, + "step": 45940 + }, + { + "epoch": 11.03, + "learning_rate": 2.329092236827323e-05, + "loss": 0.0241, + "step": 45950 + }, + { + "epoch": 11.03, + "learning_rate": 2.3283900685316257e-05, + "loss": 0.0004, + "step": 45960 + }, + { + "epoch": 11.03, + "learning_rate": 2.3276879002359285e-05, + "loss": 0.0002, + "step": 45970 + }, + { + "epoch": 11.03, + "learning_rate": 2.3269857319402316e-05, + "loss": 0.0002, + "step": 45980 + }, + { + "epoch": 11.03, + "learning_rate": 2.3262835636445343e-05, + "loss": 0.0029, + "step": 45990 + }, + { + "epoch": 11.03, + "learning_rate": 2.3255813953488374e-05, + "loss": 0.0004, + "step": 46000 + }, + { + "epoch": 11.03, + "learning_rate": 2.32487922705314e-05, + "loss": 0.3707, + "step": 46010 + }, + { + "epoch": 11.03, + "learning_rate": 2.324177058757443e-05, + "loss": 0.1171, + "step": 46020 + }, + { + "epoch": 11.03, + "learning_rate": 2.323474890461746e-05, + "loss": 0.0, + "step": 46030 + }, + { + "epoch": 11.03, + "learning_rate": 2.3227727221660487e-05, + "loss": 0.0, + "step": 46040 + }, + { + "epoch": 11.03, + "learning_rate": 2.3220705538703518e-05, + "loss": 0.004, + "step": 46050 + }, + { + "epoch": 11.03, + "learning_rate": 2.3213683855746546e-05, + "loss": 0.0025, + "step": 46060 + }, + { + "epoch": 11.03, + "learning_rate": 2.3206662172789577e-05, + "loss": 0.1226, + "step": 46070 + }, + { + "epoch": 11.03, + "learning_rate": 2.3199640489832604e-05, + "loss": 0.0193, + "step": 46080 + }, + { + "epoch": 11.03, + "learning_rate": 2.319261880687563e-05, + "loss": 0.0295, + "step": 46090 + }, + { + "epoch": 11.03, + "learning_rate": 2.3185597123918663e-05, + "loss": 0.0001, + "step": 46100 + }, + { + "epoch": 11.03, + "learning_rate": 2.3178575440961693e-05, + "loss": 0.0768, + "step": 46110 + }, + { + "epoch": 11.03, + "learning_rate": 2.3171553758004718e-05, + "loss": 0.0001, + "step": 46120 + }, + { + "epoch": 11.03, + "learning_rate": 2.316453207504775e-05, + "loss": 0.0, + "step": 46130 + }, + { + "epoch": 11.03, + "learning_rate": 2.315751039209078e-05, + "loss": 0.0002, + "step": 46140 + }, + { + "epoch": 11.03, + "learning_rate": 2.3150488709133807e-05, + "loss": 0.0, + "step": 46150 + }, + { + "epoch": 11.03, + "learning_rate": 2.3143467026176834e-05, + "loss": 0.1951, + "step": 46160 + }, + { + "epoch": 11.03, + "learning_rate": 2.3136445343219865e-05, + "loss": 0.0001, + "step": 46170 + }, + { + "epoch": 11.03, + "learning_rate": 2.3129423660262893e-05, + "loss": 0.3537, + "step": 46180 + }, + { + "epoch": 11.03, + "learning_rate": 2.312240197730592e-05, + "loss": 0.0228, + "step": 46190 + }, + { + "epoch": 11.03, + "learning_rate": 2.311538029434895e-05, + "loss": 0.3772, + "step": 46200 + }, + { + "epoch": 11.03, + "learning_rate": 2.3108358611391982e-05, + "loss": 0.0049, + "step": 46210 + }, + { + "epoch": 11.03, + "learning_rate": 2.310133692843501e-05, + "loss": 0.0007, + "step": 46220 + }, + { + "epoch": 11.03, + "learning_rate": 2.3094315245478037e-05, + "loss": 0.0002, + "step": 46230 + }, + { + "epoch": 11.03, + "learning_rate": 2.3087293562521068e-05, + "loss": 0.0018, + "step": 46240 + }, + { + "epoch": 11.03, + "learning_rate": 2.3080271879564095e-05, + "loss": 0.1855, + "step": 46250 + }, + { + "epoch": 11.03, + "learning_rate": 2.3073250196607123e-05, + "loss": 0.2041, + "step": 46260 + }, + { + "epoch": 11.03, + "learning_rate": 2.3066228513650154e-05, + "loss": 0.0589, + "step": 46270 + }, + { + "epoch": 11.03, + "learning_rate": 2.305920683069318e-05, + "loss": 0.039, + "step": 46280 + }, + { + "epoch": 11.04, + "learning_rate": 2.3052185147736212e-05, + "loss": 0.543, + "step": 46290 + }, + { + "epoch": 11.04, + "learning_rate": 2.304516346477924e-05, + "loss": 0.0001, + "step": 46300 + }, + { + "epoch": 11.04, + "learning_rate": 2.3038141781822267e-05, + "loss": 0.3284, + "step": 46310 + }, + { + "epoch": 11.04, + "learning_rate": 2.3031120098865298e-05, + "loss": 0.0009, + "step": 46320 + }, + { + "epoch": 11.04, + "learning_rate": 2.3024098415908325e-05, + "loss": 0.0001, + "step": 46330 + }, + { + "epoch": 11.04, + "learning_rate": 2.3017076732951353e-05, + "loss": 0.5456, + "step": 46340 + }, + { + "epoch": 11.04, + "learning_rate": 2.3010055049994384e-05, + "loss": 0.5065, + "step": 46350 + }, + { + "epoch": 11.04, + "learning_rate": 2.3003033367037415e-05, + "loss": 0.0218, + "step": 46360 + }, + { + "epoch": 11.04, + "learning_rate": 2.299601168408044e-05, + "loss": 0.1919, + "step": 46370 + }, + { + "epoch": 11.04, + "learning_rate": 2.298899000112347e-05, + "loss": 0.0359, + "step": 46380 + }, + { + "epoch": 11.04, + "learning_rate": 2.29819683181665e-05, + "loss": 0.2901, + "step": 46390 + }, + { + "epoch": 11.04, + "learning_rate": 2.2974946635209528e-05, + "loss": 0.0006, + "step": 46400 + }, + { + "epoch": 11.04, + "learning_rate": 2.2967924952252555e-05, + "loss": 0.0002, + "step": 46410 + }, + { + "epoch": 11.04, + "learning_rate": 2.2960903269295586e-05, + "loss": 0.0839, + "step": 46420 + }, + { + "epoch": 11.04, + "learning_rate": 2.2953881586338617e-05, + "loss": 0.0679, + "step": 46430 + }, + { + "epoch": 11.04, + "learning_rate": 2.294685990338164e-05, + "loss": 0.0, + "step": 46440 + }, + { + "epoch": 11.04, + "learning_rate": 2.2939838220424672e-05, + "loss": 0.0486, + "step": 46450 + }, + { + "epoch": 11.04, + "learning_rate": 2.2932816537467703e-05, + "loss": 0.134, + "step": 46460 + }, + { + "epoch": 11.04, + "learning_rate": 2.292579485451073e-05, + "loss": 0.0, + "step": 46470 + }, + { + "epoch": 11.04, + "learning_rate": 2.2918773171553758e-05, + "loss": 0.0007, + "step": 46480 + }, + { + "epoch": 11.04, + "learning_rate": 2.291175148859679e-05, + "loss": 0.0006, + "step": 46490 + }, + { + "epoch": 11.04, + "learning_rate": 2.2904729805639816e-05, + "loss": 0.2159, + "step": 46500 + }, + { + "epoch": 11.04, + "learning_rate": 2.2897708122682844e-05, + "loss": 0.0, + "step": 46510 + }, + { + "epoch": 11.04, + "learning_rate": 2.2890686439725875e-05, + "loss": 0.0356, + "step": 46520 + }, + { + "epoch": 11.04, + "learning_rate": 2.2883664756768902e-05, + "loss": 0.0, + "step": 46530 + }, + { + "epoch": 11.04, + "learning_rate": 2.2876643073811933e-05, + "loss": 0.0001, + "step": 46540 + }, + { + "epoch": 11.04, + "learning_rate": 2.286962139085496e-05, + "loss": 0.8437, + "step": 46550 + }, + { + "epoch": 11.04, + "learning_rate": 2.286259970789799e-05, + "loss": 0.0, + "step": 46560 + }, + { + "epoch": 11.04, + "learning_rate": 2.285557802494102e-05, + "loss": 0.0253, + "step": 46570 + }, + { + "epoch": 11.04, + "learning_rate": 2.284855634198405e-05, + "loss": 0.0205, + "step": 46580 + }, + { + "epoch": 11.04, + "learning_rate": 2.2841534659027077e-05, + "loss": 0.0, + "step": 46590 + }, + { + "epoch": 11.04, + "learning_rate": 2.2834512976070105e-05, + "loss": 0.3656, + "step": 46600 + }, + { + "epoch": 11.04, + "learning_rate": 2.2827491293113136e-05, + "loss": 0.0756, + "step": 46610 + }, + { + "epoch": 11.04, + "learning_rate": 2.2820469610156163e-05, + "loss": 0.0004, + "step": 46620 + }, + { + "epoch": 11.04, + "learning_rate": 2.281344792719919e-05, + "loss": 0.0, + "step": 46630 + }, + { + "epoch": 11.04, + "learning_rate": 2.280642624424222e-05, + "loss": 0.1053, + "step": 46640 + }, + { + "epoch": 11.04, + "learning_rate": 2.2799404561285252e-05, + "loss": 0.0003, + "step": 46650 + }, + { + "epoch": 11.04, + "learning_rate": 2.2792382878328277e-05, + "loss": 0.0001, + "step": 46660 + }, + { + "epoch": 11.04, + "learning_rate": 2.2785361195371307e-05, + "loss": 0.0184, + "step": 46670 + }, + { + "epoch": 11.04, + "learning_rate": 2.2778339512414338e-05, + "loss": 0.0003, + "step": 46680 + }, + { + "epoch": 11.04, + "learning_rate": 2.2771317829457366e-05, + "loss": 0.0, + "step": 46690 + }, + { + "epoch": 11.04, + "learning_rate": 2.2764296146500393e-05, + "loss": 0.0001, + "step": 46700 + }, + { + "epoch": 11.04, + "learning_rate": 2.2757274463543424e-05, + "loss": 0.6094, + "step": 46710 + }, + { + "epoch": 11.04, + "learning_rate": 2.275025278058645e-05, + "loss": 0.0004, + "step": 46720 + }, + { + "epoch": 11.04, + "learning_rate": 2.274323109762948e-05, + "loss": 0.0005, + "step": 46730 + }, + { + "epoch": 11.04, + "learning_rate": 2.273620941467251e-05, + "loss": 0.5278, + "step": 46740 + }, + { + "epoch": 11.04, + "learning_rate": 2.272918773171554e-05, + "loss": 0.039, + "step": 46750 + }, + { + "epoch": 11.04, + "learning_rate": 2.272216604875857e-05, + "loss": 0.0325, + "step": 46760 + }, + { + "epoch": 11.04, + "learning_rate": 2.2715144365801596e-05, + "loss": 0.1069, + "step": 46770 + }, + { + "epoch": 11.04, + "learning_rate": 2.2708122682844627e-05, + "loss": 0.0002, + "step": 46780 + }, + { + "epoch": 11.04, + "learning_rate": 2.2701100999887654e-05, + "loss": 0.2367, + "step": 46790 + }, + { + "epoch": 11.04, + "learning_rate": 2.2694079316930682e-05, + "loss": 0.3016, + "step": 46800 + }, + { + "epoch": 11.04, + "learning_rate": 2.2687057633973713e-05, + "loss": 0.0004, + "step": 46810 + }, + { + "epoch": 11.04, + "learning_rate": 2.268003595101674e-05, + "loss": 0.0002, + "step": 46820 + }, + { + "epoch": 11.04, + "learning_rate": 2.267301426805977e-05, + "loss": 0.1439, + "step": 46830 + }, + { + "epoch": 11.04, + "learning_rate": 2.26659925851028e-05, + "loss": 0.7615, + "step": 46840 + }, + { + "epoch": 11.04, + "learning_rate": 2.2658970902145826e-05, + "loss": 0.0875, + "step": 46850 + }, + { + "epoch": 11.04, + "learning_rate": 2.2651949219188857e-05, + "loss": 0.0018, + "step": 46860 + }, + { + "epoch": 11.04, + "learning_rate": 2.2644927536231884e-05, + "loss": 0.006, + "step": 46870 + }, + { + "epoch": 11.04, + "learning_rate": 2.2637905853274912e-05, + "loss": 0.1166, + "step": 46880 + }, + { + "epoch": 11.04, + "learning_rate": 2.2630884170317943e-05, + "loss": 0.0553, + "step": 46890 + }, + { + "epoch": 11.04, + "learning_rate": 2.2623862487360974e-05, + "loss": 0.0013, + "step": 46900 + }, + { + "epoch": 11.04, + "learning_rate": 2.2616840804404e-05, + "loss": 0.0036, + "step": 46910 + }, + { + "epoch": 11.04, + "learning_rate": 2.260981912144703e-05, + "loss": 0.001, + "step": 46920 + }, + { + "epoch": 11.04, + "learning_rate": 2.260279743849006e-05, + "loss": 0.0192, + "step": 46930 + }, + { + "epoch": 11.04, + "learning_rate": 2.2595775755533087e-05, + "loss": 0.0126, + "step": 46940 + }, + { + "epoch": 11.04, + "learning_rate": 2.2588754072576114e-05, + "loss": 0.3203, + "step": 46950 + }, + { + "epoch": 11.04, + "learning_rate": 2.2581732389619145e-05, + "loss": 0.0029, + "step": 46960 + }, + { + "epoch": 11.04, + "learning_rate": 2.2574710706662176e-05, + "loss": 0.0004, + "step": 46970 + }, + { + "epoch": 11.04, + "learning_rate": 2.25676890237052e-05, + "loss": 0.0739, + "step": 46980 + }, + { + "epoch": 11.04, + "learning_rate": 2.256066734074823e-05, + "loss": 0.0, + "step": 46990 + }, + { + "epoch": 11.04, + "learning_rate": 2.2553645657791262e-05, + "loss": 0.1236, + "step": 47000 + }, + { + "epoch": 11.04, + "learning_rate": 2.254662397483429e-05, + "loss": 0.0003, + "step": 47010 + }, + { + "epoch": 11.04, + "learning_rate": 2.2539602291877317e-05, + "loss": 0.0, + "step": 47020 + }, + { + "epoch": 11.04, + "learning_rate": 2.2532580608920348e-05, + "loss": 0.0136, + "step": 47030 + }, + { + "epoch": 11.04, + "learning_rate": 2.2525558925963375e-05, + "loss": 0.0068, + "step": 47040 + }, + { + "epoch": 11.04, + "learning_rate": 2.2518537243006403e-05, + "loss": 0.582, + "step": 47050 + }, + { + "epoch": 11.04, + "learning_rate": 2.2511515560049434e-05, + "loss": 0.0, + "step": 47060 + }, + { + "epoch": 11.04, + "learning_rate": 2.250449387709246e-05, + "loss": 0.0002, + "step": 47070 + }, + { + "epoch": 11.05, + "learning_rate": 2.2497472194135492e-05, + "loss": 0.0742, + "step": 47080 + }, + { + "epoch": 11.05, + "learning_rate": 2.249045051117852e-05, + "loss": 0.0002, + "step": 47090 + }, + { + "epoch": 11.05, + "learning_rate": 2.248342882822155e-05, + "loss": 0.009, + "step": 47100 + }, + { + "epoch": 11.05, + "learning_rate": 2.2476407145264578e-05, + "loss": 0.0, + "step": 47110 + }, + { + "epoch": 11.05, + "learning_rate": 2.246938546230761e-05, + "loss": 0.0, + "step": 47120 + }, + { + "epoch": 11.05, + "learning_rate": 2.2462363779350636e-05, + "loss": 0.0, + "step": 47130 + }, + { + "epoch": 11.05, + "learning_rate": 2.2455342096393664e-05, + "loss": 0.0, + "step": 47140 + }, + { + "epoch": 11.05, + "learning_rate": 2.2448320413436695e-05, + "loss": 0.0, + "step": 47150 + }, + { + "epoch": 11.05, + "learning_rate": 2.2441298730479722e-05, + "loss": 0.0, + "step": 47160 + }, + { + "epoch": 11.05, + "learning_rate": 2.243427704752275e-05, + "loss": 0.0541, + "step": 47170 + }, + { + "epoch": 11.05, + "learning_rate": 2.242725536456578e-05, + "loss": 0.0, + "step": 47180 + }, + { + "epoch": 11.05, + "learning_rate": 2.242023368160881e-05, + "loss": 0.0028, + "step": 47190 + }, + { + "epoch": 11.05, + "learning_rate": 2.2413211998651836e-05, + "loss": 0.0053, + "step": 47200 + }, + { + "epoch": 11.05, + "learning_rate": 2.2406190315694866e-05, + "loss": 0.0001, + "step": 47210 + }, + { + "epoch": 11.05, + "learning_rate": 2.2399168632737897e-05, + "loss": 0.0016, + "step": 47220 + }, + { + "epoch": 11.05, + "learning_rate": 2.2392146949780925e-05, + "loss": 0.0108, + "step": 47230 + }, + { + "epoch": 11.05, + "learning_rate": 2.2385125266823952e-05, + "loss": 0.0, + "step": 47240 + }, + { + "epoch": 11.05, + "learning_rate": 2.2378103583866983e-05, + "loss": 0.0012, + "step": 47250 + }, + { + "epoch": 11.05, + "learning_rate": 2.2371081900910014e-05, + "loss": 0.0004, + "step": 47260 + }, + { + "epoch": 11.05, + "learning_rate": 2.2364060217953038e-05, + "loss": 0.0, + "step": 47270 + }, + { + "epoch": 11.05, + "learning_rate": 2.235703853499607e-05, + "loss": 0.0, + "step": 47280 + }, + { + "epoch": 11.05, + "learning_rate": 2.23500168520391e-05, + "loss": 0.0, + "step": 47290 + }, + { + "epoch": 11.05, + "learning_rate": 2.2342995169082127e-05, + "loss": 0.0, + "step": 47300 + }, + { + "epoch": 11.05, + "learning_rate": 2.2335973486125155e-05, + "loss": 0.0, + "step": 47310 + }, + { + "epoch": 11.05, + "learning_rate": 2.2328951803168186e-05, + "loss": 0.0574, + "step": 47320 + }, + { + "epoch": 11.05, + "learning_rate": 2.2321930120211213e-05, + "loss": 0.0987, + "step": 47330 + }, + { + "epoch": 11.05, + "learning_rate": 2.231490843725424e-05, + "loss": 0.0266, + "step": 47340 + }, + { + "epoch": 11.05, + "learning_rate": 2.230788675429727e-05, + "loss": 0.0117, + "step": 47350 + }, + { + "epoch": 11.05, + "learning_rate": 2.23008650713403e-05, + "loss": 0.0, + "step": 47360 + }, + { + "epoch": 11.05, + "learning_rate": 2.229384338838333e-05, + "loss": 0.0005, + "step": 47370 + }, + { + "epoch": 11.05, + "learning_rate": 2.2286821705426357e-05, + "loss": 0.0, + "step": 47380 + }, + { + "epoch": 11.05, + "learning_rate": 2.2279800022469385e-05, + "loss": 0.0007, + "step": 47390 + }, + { + "epoch": 11.05, + "learning_rate": 2.2272778339512416e-05, + "loss": 0.0446, + "step": 47400 + }, + { + "epoch": 11.05, + "learning_rate": 2.2265756656555443e-05, + "loss": 0.0001, + "step": 47410 + }, + { + "epoch": 11.05, + "learning_rate": 2.225873497359847e-05, + "loss": 0.0081, + "step": 47420 + }, + { + "epoch": 11.05, + "learning_rate": 2.2251713290641502e-05, + "loss": 0.3327, + "step": 47430 + }, + { + "epoch": 11.05, + "learning_rate": 2.2244691607684533e-05, + "loss": 0.2027, + "step": 47440 + }, + { + "epoch": 11.05, + "learning_rate": 2.223766992472756e-05, + "loss": 0.0004, + "step": 47450 + }, + { + "epoch": 11.05, + "learning_rate": 2.2230648241770588e-05, + "loss": 0.0001, + "step": 47460 + }, + { + "epoch": 11.05, + "learning_rate": 2.222362655881362e-05, + "loss": 0.0, + "step": 47470 + }, + { + "epoch": 11.05, + "eval_accuracy": 0.7573216359179461, + "eval_loss": 1.5405784845352173, + "eval_runtime": 1314.8335, + "eval_samples_per_second": 5.895, + "eval_steps_per_second": 2.948, + "step": 47472 + }, + { + "epoch": 12.0, + "learning_rate": 2.2216604875856646e-05, + "loss": 0.0, + "step": 47480 + }, + { + "epoch": 12.0, + "learning_rate": 2.2209583192899673e-05, + "loss": 0.0, + "step": 47490 + }, + { + "epoch": 12.0, + "learning_rate": 2.2202561509942704e-05, + "loss": 0.0114, + "step": 47500 + }, + { + "epoch": 12.0, + "learning_rate": 2.2195539826985735e-05, + "loss": 0.0786, + "step": 47510 + }, + { + "epoch": 12.0, + "learning_rate": 2.218851814402876e-05, + "loss": 0.1488, + "step": 47520 + }, + { + "epoch": 12.0, + "learning_rate": 2.218149646107179e-05, + "loss": 0.0008, + "step": 47530 + }, + { + "epoch": 12.0, + "learning_rate": 2.217447477811482e-05, + "loss": 0.0, + "step": 47540 + }, + { + "epoch": 12.0, + "learning_rate": 2.216745309515785e-05, + "loss": 0.1265, + "step": 47550 + }, + { + "epoch": 12.0, + "learning_rate": 2.2160431412200876e-05, + "loss": 0.0001, + "step": 47560 + }, + { + "epoch": 12.0, + "learning_rate": 2.2153409729243907e-05, + "loss": 0.0002, + "step": 47570 + }, + { + "epoch": 12.0, + "learning_rate": 2.2146388046286934e-05, + "loss": 0.0, + "step": 47580 + }, + { + "epoch": 12.0, + "learning_rate": 2.2139366363329965e-05, + "loss": 0.0141, + "step": 47590 + }, + { + "epoch": 12.0, + "learning_rate": 2.2132344680372993e-05, + "loss": 0.0, + "step": 47600 + }, + { + "epoch": 12.0, + "learning_rate": 2.2125322997416024e-05, + "loss": 0.0, + "step": 47610 + }, + { + "epoch": 12.0, + "learning_rate": 2.211830131445905e-05, + "loss": 0.0, + "step": 47620 + }, + { + "epoch": 12.0, + "learning_rate": 2.211127963150208e-05, + "loss": 0.0001, + "step": 47630 + }, + { + "epoch": 12.0, + "learning_rate": 2.210425794854511e-05, + "loss": 0.0371, + "step": 47640 + }, + { + "epoch": 12.0, + "learning_rate": 2.2097236265588137e-05, + "loss": 0.0001, + "step": 47650 + }, + { + "epoch": 12.0, + "learning_rate": 2.2090214582631168e-05, + "loss": 0.0015, + "step": 47660 + }, + { + "epoch": 12.0, + "learning_rate": 2.2083192899674195e-05, + "loss": 0.0471, + "step": 47670 + }, + { + "epoch": 12.0, + "learning_rate": 2.2076171216717223e-05, + "loss": 0.0, + "step": 47680 + }, + { + "epoch": 12.0, + "learning_rate": 2.2069149533760254e-05, + "loss": 0.0, + "step": 47690 + }, + { + "epoch": 12.0, + "learning_rate": 2.206212785080328e-05, + "loss": 0.0341, + "step": 47700 + }, + { + "epoch": 12.0, + "learning_rate": 2.205510616784631e-05, + "loss": 0.0, + "step": 47710 + }, + { + "epoch": 12.0, + "learning_rate": 2.204808448488934e-05, + "loss": 0.0, + "step": 47720 + }, + { + "epoch": 12.0, + "learning_rate": 2.204106280193237e-05, + "loss": 0.0048, + "step": 47730 + }, + { + "epoch": 12.0, + "learning_rate": 2.2034041118975395e-05, + "loss": 0.0002, + "step": 47740 + }, + { + "epoch": 12.0, + "learning_rate": 2.2027019436018425e-05, + "loss": 0.0, + "step": 47750 + }, + { + "epoch": 12.0, + "learning_rate": 2.2019997753061456e-05, + "loss": 0.0001, + "step": 47760 + }, + { + "epoch": 12.0, + "learning_rate": 2.2012976070104484e-05, + "loss": 0.0, + "step": 47770 + }, + { + "epoch": 12.0, + "learning_rate": 2.200595438714751e-05, + "loss": 0.0001, + "step": 47780 + }, + { + "epoch": 12.0, + "learning_rate": 2.1998932704190542e-05, + "loss": 0.1879, + "step": 47790 + }, + { + "epoch": 12.0, + "learning_rate": 2.1991911021233573e-05, + "loss": 0.0, + "step": 47800 + }, + { + "epoch": 12.0, + "learning_rate": 2.1984889338276597e-05, + "loss": 0.0082, + "step": 47810 + }, + { + "epoch": 12.0, + "learning_rate": 2.1977867655319628e-05, + "loss": 0.0, + "step": 47820 + }, + { + "epoch": 12.0, + "learning_rate": 2.197084597236266e-05, + "loss": 0.0011, + "step": 47830 + }, + { + "epoch": 12.0, + "learning_rate": 2.1963824289405686e-05, + "loss": 0.0531, + "step": 47840 + }, + { + "epoch": 12.0, + "learning_rate": 2.1956802606448714e-05, + "loss": 0.0049, + "step": 47850 + }, + { + "epoch": 12.0, + "learning_rate": 2.1949780923491745e-05, + "loss": 0.0001, + "step": 47860 + }, + { + "epoch": 12.01, + "learning_rate": 2.1942759240534772e-05, + "loss": 0.0002, + "step": 47870 + }, + { + "epoch": 12.01, + "learning_rate": 2.19357375575778e-05, + "loss": 0.0004, + "step": 47880 + }, + { + "epoch": 12.01, + "learning_rate": 2.192871587462083e-05, + "loss": 0.0, + "step": 47890 + }, + { + "epoch": 12.01, + "learning_rate": 2.1921694191663858e-05, + "loss": 0.0, + "step": 47900 + }, + { + "epoch": 12.01, + "learning_rate": 2.191467250870689e-05, + "loss": 0.0, + "step": 47910 + }, + { + "epoch": 12.01, + "learning_rate": 2.1907650825749917e-05, + "loss": 0.0, + "step": 47920 + }, + { + "epoch": 12.01, + "learning_rate": 2.1900629142792944e-05, + "loss": 0.0001, + "step": 47930 + }, + { + "epoch": 12.01, + "learning_rate": 2.1893607459835975e-05, + "loss": 0.0167, + "step": 47940 + }, + { + "epoch": 12.01, + "learning_rate": 2.1886585776879002e-05, + "loss": 0.0, + "step": 47950 + }, + { + "epoch": 12.01, + "learning_rate": 2.1879564093922033e-05, + "loss": 0.019, + "step": 47960 + }, + { + "epoch": 12.01, + "learning_rate": 2.187254241096506e-05, + "loss": 0.003, + "step": 47970 + }, + { + "epoch": 12.01, + "learning_rate": 2.186552072800809e-05, + "loss": 0.0, + "step": 47980 + }, + { + "epoch": 12.01, + "learning_rate": 2.185849904505112e-05, + "loss": 0.0001, + "step": 47990 + }, + { + "epoch": 12.01, + "learning_rate": 2.1851477362094147e-05, + "loss": 0.0, + "step": 48000 + }, + { + "epoch": 12.01, + "learning_rate": 2.1844455679137177e-05, + "loss": 0.0001, + "step": 48010 + }, + { + "epoch": 12.01, + "learning_rate": 2.1837433996180205e-05, + "loss": 0.0016, + "step": 48020 + }, + { + "epoch": 12.01, + "learning_rate": 2.1830412313223232e-05, + "loss": 0.7588, + "step": 48030 + }, + { + "epoch": 12.01, + "learning_rate": 2.1823390630266263e-05, + "loss": 0.0004, + "step": 48040 + }, + { + "epoch": 12.01, + "learning_rate": 2.1816368947309294e-05, + "loss": 0.0001, + "step": 48050 + }, + { + "epoch": 12.01, + "learning_rate": 2.180934726435232e-05, + "loss": 0.0073, + "step": 48060 + }, + { + "epoch": 12.01, + "learning_rate": 2.180232558139535e-05, + "loss": 0.0, + "step": 48070 + }, + { + "epoch": 12.01, + "learning_rate": 2.179530389843838e-05, + "loss": 0.0, + "step": 48080 + }, + { + "epoch": 12.01, + "learning_rate": 2.1788282215481408e-05, + "loss": 0.0031, + "step": 48090 + }, + { + "epoch": 12.01, + "learning_rate": 2.1781260532524435e-05, + "loss": 0.0101, + "step": 48100 + }, + { + "epoch": 12.01, + "learning_rate": 2.1774238849567466e-05, + "loss": 0.1959, + "step": 48110 + }, + { + "epoch": 12.01, + "learning_rate": 2.1767217166610493e-05, + "loss": 0.0002, + "step": 48120 + }, + { + "epoch": 12.01, + "learning_rate": 2.1760195483653524e-05, + "loss": 0.0001, + "step": 48130 + }, + { + "epoch": 12.01, + "learning_rate": 2.1753173800696552e-05, + "loss": 0.2306, + "step": 48140 + }, + { + "epoch": 12.01, + "learning_rate": 2.1746152117739583e-05, + "loss": 0.0, + "step": 48150 + }, + { + "epoch": 12.01, + "learning_rate": 2.173913043478261e-05, + "loss": 0.0114, + "step": 48160 + }, + { + "epoch": 12.01, + "learning_rate": 2.1732108751825638e-05, + "loss": 0.0199, + "step": 48170 + }, + { + "epoch": 12.01, + "learning_rate": 2.172508706886867e-05, + "loss": 0.0013, + "step": 48180 + }, + { + "epoch": 12.01, + "learning_rate": 2.1718065385911696e-05, + "loss": 0.0, + "step": 48190 + }, + { + "epoch": 12.01, + "learning_rate": 2.1711043702954727e-05, + "loss": 0.3639, + "step": 48200 + }, + { + "epoch": 12.01, + "learning_rate": 2.1704022019997754e-05, + "loss": 0.0001, + "step": 48210 + }, + { + "epoch": 12.01, + "learning_rate": 2.1697000337040782e-05, + "loss": 0.0002, + "step": 48220 + }, + { + "epoch": 12.01, + "learning_rate": 2.1689978654083813e-05, + "loss": 0.1115, + "step": 48230 + }, + { + "epoch": 12.01, + "learning_rate": 2.168295697112684e-05, + "loss": 0.0006, + "step": 48240 + }, + { + "epoch": 12.01, + "learning_rate": 2.1675935288169868e-05, + "loss": 0.0, + "step": 48250 + }, + { + "epoch": 12.01, + "learning_rate": 2.16689136052129e-05, + "loss": 0.0, + "step": 48260 + }, + { + "epoch": 12.01, + "learning_rate": 2.166189192225593e-05, + "loss": 0.3734, + "step": 48270 + }, + { + "epoch": 12.01, + "learning_rate": 2.1654870239298954e-05, + "loss": 0.0019, + "step": 48280 + }, + { + "epoch": 12.01, + "learning_rate": 2.1647848556341984e-05, + "loss": 0.0, + "step": 48290 + }, + { + "epoch": 12.01, + "learning_rate": 2.1640826873385015e-05, + "loss": 0.001, + "step": 48300 + }, + { + "epoch": 12.01, + "learning_rate": 2.1633805190428043e-05, + "loss": 0.438, + "step": 48310 + }, + { + "epoch": 12.01, + "learning_rate": 2.162678350747107e-05, + "loss": 0.0002, + "step": 48320 + }, + { + "epoch": 12.01, + "learning_rate": 2.16197618245141e-05, + "loss": 0.067, + "step": 48330 + }, + { + "epoch": 12.01, + "learning_rate": 2.1612740141557132e-05, + "loss": 1.0177, + "step": 48340 + }, + { + "epoch": 12.01, + "learning_rate": 2.1605718458600156e-05, + "loss": 0.0388, + "step": 48350 + }, + { + "epoch": 12.01, + "learning_rate": 2.1598696775643187e-05, + "loss": 0.1265, + "step": 48360 + }, + { + "epoch": 12.01, + "learning_rate": 2.1591675092686218e-05, + "loss": 0.0001, + "step": 48370 + }, + { + "epoch": 12.01, + "learning_rate": 2.1584653409729245e-05, + "loss": 0.0, + "step": 48380 + }, + { + "epoch": 12.01, + "learning_rate": 2.1577631726772273e-05, + "loss": 0.0008, + "step": 48390 + }, + { + "epoch": 12.01, + "learning_rate": 2.1570610043815304e-05, + "loss": 0.007, + "step": 48400 + }, + { + "epoch": 12.01, + "learning_rate": 2.156358836085833e-05, + "loss": 0.0001, + "step": 48410 + }, + { + "epoch": 12.01, + "learning_rate": 2.155656667790136e-05, + "loss": 0.0118, + "step": 48420 + }, + { + "epoch": 12.01, + "learning_rate": 2.154954499494439e-05, + "loss": 0.0002, + "step": 48430 + }, + { + "epoch": 12.01, + "learning_rate": 2.1542523311987417e-05, + "loss": 0.48, + "step": 48440 + }, + { + "epoch": 12.01, + "learning_rate": 2.1535501629030448e-05, + "loss": 0.6302, + "step": 48450 + }, + { + "epoch": 12.01, + "learning_rate": 2.1528479946073476e-05, + "loss": 0.0001, + "step": 48460 + }, + { + "epoch": 12.01, + "learning_rate": 2.1521458263116503e-05, + "loss": 0.1606, + "step": 48470 + }, + { + "epoch": 12.01, + "learning_rate": 2.1514436580159534e-05, + "loss": 0.002, + "step": 48480 + }, + { + "epoch": 12.01, + "learning_rate": 2.150741489720256e-05, + "loss": 0.0001, + "step": 48490 + }, + { + "epoch": 12.01, + "learning_rate": 2.1500393214245592e-05, + "loss": 0.0, + "step": 48500 + }, + { + "epoch": 12.01, + "learning_rate": 2.149337153128862e-05, + "loss": 0.0178, + "step": 48510 + }, + { + "epoch": 12.01, + "learning_rate": 2.148634984833165e-05, + "loss": 0.0037, + "step": 48520 + }, + { + "epoch": 12.01, + "learning_rate": 2.1479328165374678e-05, + "loss": 0.0, + "step": 48530 + }, + { + "epoch": 12.01, + "learning_rate": 2.1472306482417706e-05, + "loss": 0.0, + "step": 48540 + }, + { + "epoch": 12.01, + "learning_rate": 2.1465284799460737e-05, + "loss": 0.0007, + "step": 48550 + }, + { + "epoch": 12.01, + "learning_rate": 2.1458263116503764e-05, + "loss": 0.2653, + "step": 48560 + }, + { + "epoch": 12.01, + "learning_rate": 2.145124143354679e-05, + "loss": 0.012, + "step": 48570 + }, + { + "epoch": 12.01, + "learning_rate": 2.1444219750589822e-05, + "loss": 0.0008, + "step": 48580 + }, + { + "epoch": 12.01, + "learning_rate": 2.1437198067632853e-05, + "loss": 0.0001, + "step": 48590 + }, + { + "epoch": 12.01, + "learning_rate": 2.143017638467588e-05, + "loss": 0.0, + "step": 48600 + }, + { + "epoch": 12.01, + "learning_rate": 2.1423154701718908e-05, + "loss": 0.0004, + "step": 48610 + }, + { + "epoch": 12.01, + "learning_rate": 2.141613301876194e-05, + "loss": 0.012, + "step": 48620 + }, + { + "epoch": 12.01, + "learning_rate": 2.1409111335804967e-05, + "loss": 0.0003, + "step": 48630 + }, + { + "epoch": 12.01, + "learning_rate": 2.1402089652847994e-05, + "loss": 0.2952, + "step": 48640 + }, + { + "epoch": 12.01, + "learning_rate": 2.1395067969891025e-05, + "loss": 0.4434, + "step": 48650 + }, + { + "epoch": 12.02, + "learning_rate": 2.1388046286934056e-05, + "loss": 0.0001, + "step": 48660 + }, + { + "epoch": 12.02, + "learning_rate": 2.1381024603977083e-05, + "loss": 0.7248, + "step": 48670 + }, + { + "epoch": 12.02, + "learning_rate": 2.137400292102011e-05, + "loss": 0.3707, + "step": 48680 + }, + { + "epoch": 12.02, + "learning_rate": 2.1366981238063142e-05, + "loss": 0.0018, + "step": 48690 + }, + { + "epoch": 12.02, + "learning_rate": 2.135995955510617e-05, + "loss": 0.0001, + "step": 48700 + }, + { + "epoch": 12.02, + "learning_rate": 2.1352937872149197e-05, + "loss": 0.0002, + "step": 48710 + }, + { + "epoch": 12.02, + "learning_rate": 2.1345916189192228e-05, + "loss": 0.3446, + "step": 48720 + }, + { + "epoch": 12.02, + "learning_rate": 2.1338894506235255e-05, + "loss": 0.0001, + "step": 48730 + }, + { + "epoch": 12.02, + "learning_rate": 2.1331872823278286e-05, + "loss": 0.0009, + "step": 48740 + }, + { + "epoch": 12.02, + "learning_rate": 2.1324851140321313e-05, + "loss": 0.0069, + "step": 48750 + }, + { + "epoch": 12.02, + "learning_rate": 2.131782945736434e-05, + "loss": 0.0221, + "step": 48760 + }, + { + "epoch": 12.02, + "learning_rate": 2.1310807774407372e-05, + "loss": 0.0017, + "step": 48770 + }, + { + "epoch": 12.02, + "learning_rate": 2.13037860914504e-05, + "loss": 0.1316, + "step": 48780 + }, + { + "epoch": 12.02, + "learning_rate": 2.1296764408493427e-05, + "loss": 0.0, + "step": 48790 + }, + { + "epoch": 12.02, + "learning_rate": 2.1289742725536458e-05, + "loss": 0.0007, + "step": 48800 + }, + { + "epoch": 12.02, + "learning_rate": 2.128272104257949e-05, + "loss": 0.4672, + "step": 48810 + }, + { + "epoch": 12.02, + "learning_rate": 2.1275699359622513e-05, + "loss": 0.0001, + "step": 48820 + }, + { + "epoch": 12.02, + "learning_rate": 2.1268677676665544e-05, + "loss": 0.0, + "step": 48830 + }, + { + "epoch": 12.02, + "learning_rate": 2.1261655993708574e-05, + "loss": 0.0002, + "step": 48840 + }, + { + "epoch": 12.02, + "learning_rate": 2.1254634310751602e-05, + "loss": 0.0, + "step": 48850 + }, + { + "epoch": 12.02, + "learning_rate": 2.124761262779463e-05, + "loss": 0.0006, + "step": 48860 + }, + { + "epoch": 12.02, + "learning_rate": 2.124059094483766e-05, + "loss": 0.0, + "step": 48870 + }, + { + "epoch": 12.02, + "learning_rate": 2.123356926188069e-05, + "loss": 0.0, + "step": 48880 + }, + { + "epoch": 12.02, + "learning_rate": 2.1226547578923715e-05, + "loss": 0.0061, + "step": 48890 + }, + { + "epoch": 12.02, + "learning_rate": 2.1219525895966746e-05, + "loss": 0.0001, + "step": 48900 + }, + { + "epoch": 12.02, + "learning_rate": 2.1212504213009777e-05, + "loss": 0.0033, + "step": 48910 + }, + { + "epoch": 12.02, + "learning_rate": 2.1205482530052804e-05, + "loss": 0.4841, + "step": 48920 + }, + { + "epoch": 12.02, + "learning_rate": 2.1198460847095832e-05, + "loss": 0.0, + "step": 48930 + }, + { + "epoch": 12.02, + "learning_rate": 2.1191439164138863e-05, + "loss": 0.0, + "step": 48940 + }, + { + "epoch": 12.02, + "learning_rate": 2.118441748118189e-05, + "loss": 0.3721, + "step": 48950 + }, + { + "epoch": 12.02, + "learning_rate": 2.1177395798224918e-05, + "loss": 0.0002, + "step": 48960 + }, + { + "epoch": 12.02, + "learning_rate": 2.117037411526795e-05, + "loss": 0.0, + "step": 48970 + }, + { + "epoch": 12.02, + "learning_rate": 2.1163352432310976e-05, + "loss": 0.1689, + "step": 48980 + }, + { + "epoch": 12.02, + "learning_rate": 2.1156330749354007e-05, + "loss": 0.0886, + "step": 48990 + }, + { + "epoch": 12.02, + "learning_rate": 2.1149309066397035e-05, + "loss": 0.0009, + "step": 49000 + }, + { + "epoch": 12.02, + "learning_rate": 2.1142287383440062e-05, + "loss": 0.0001, + "step": 49010 + }, + { + "epoch": 12.02, + "learning_rate": 2.1135265700483093e-05, + "loss": 0.0024, + "step": 49020 + }, + { + "epoch": 12.02, + "learning_rate": 2.112824401752612e-05, + "loss": 0.0, + "step": 49030 + }, + { + "epoch": 12.02, + "learning_rate": 2.112122233456915e-05, + "loss": 0.0001, + "step": 49040 + }, + { + "epoch": 12.02, + "learning_rate": 2.111420065161218e-05, + "loss": 0.0192, + "step": 49050 + }, + { + "epoch": 12.02, + "learning_rate": 2.110717896865521e-05, + "loss": 0.2795, + "step": 49060 + }, + { + "epoch": 12.02, + "learning_rate": 2.1100157285698237e-05, + "loss": 0.1856, + "step": 49070 + }, + { + "epoch": 12.02, + "learning_rate": 2.1093135602741265e-05, + "loss": 0.0019, + "step": 49080 + }, + { + "epoch": 12.02, + "learning_rate": 2.1086113919784296e-05, + "loss": 0.0, + "step": 49090 + }, + { + "epoch": 12.02, + "learning_rate": 2.1079092236827323e-05, + "loss": 0.0001, + "step": 49100 + }, + { + "epoch": 12.02, + "learning_rate": 2.107207055387035e-05, + "loss": 0.047, + "step": 49110 + }, + { + "epoch": 12.02, + "learning_rate": 2.106504887091338e-05, + "loss": 0.2265, + "step": 49120 + }, + { + "epoch": 12.02, + "learning_rate": 2.1058027187956412e-05, + "loss": 0.0385, + "step": 49130 + }, + { + "epoch": 12.02, + "learning_rate": 2.105100550499944e-05, + "loss": 0.0, + "step": 49140 + }, + { + "epoch": 12.02, + "learning_rate": 2.1043983822042467e-05, + "loss": 0.0, + "step": 49150 + }, + { + "epoch": 12.02, + "learning_rate": 2.1036962139085498e-05, + "loss": 0.0064, + "step": 49160 + }, + { + "epoch": 12.02, + "learning_rate": 2.1029940456128526e-05, + "loss": 0.0, + "step": 49170 + }, + { + "epoch": 12.02, + "learning_rate": 2.1022918773171553e-05, + "loss": 0.0001, + "step": 49180 + }, + { + "epoch": 12.02, + "learning_rate": 2.1015897090214584e-05, + "loss": 0.0012, + "step": 49190 + }, + { + "epoch": 12.02, + "learning_rate": 2.1008875407257615e-05, + "loss": 0.0001, + "step": 49200 + }, + { + "epoch": 12.02, + "learning_rate": 2.1001853724300642e-05, + "loss": 0.0, + "step": 49210 + }, + { + "epoch": 12.02, + "learning_rate": 2.099483204134367e-05, + "loss": 0.0, + "step": 49220 + }, + { + "epoch": 12.02, + "learning_rate": 2.09878103583867e-05, + "loss": 0.0, + "step": 49230 + }, + { + "epoch": 12.02, + "learning_rate": 2.0980788675429728e-05, + "loss": 0.0, + "step": 49240 + }, + { + "epoch": 12.02, + "learning_rate": 2.0973766992472756e-05, + "loss": 0.4453, + "step": 49250 + }, + { + "epoch": 12.02, + "learning_rate": 2.0966745309515787e-05, + "loss": 0.0255, + "step": 49260 + }, + { + "epoch": 12.02, + "learning_rate": 2.0959723626558814e-05, + "loss": 0.0245, + "step": 49270 + }, + { + "epoch": 12.02, + "learning_rate": 2.0952701943601845e-05, + "loss": 0.0, + "step": 49280 + }, + { + "epoch": 12.02, + "learning_rate": 2.0945680260644872e-05, + "loss": 0.0001, + "step": 49290 + }, + { + "epoch": 12.02, + "learning_rate": 2.09386585776879e-05, + "loss": 0.0034, + "step": 49300 + }, + { + "epoch": 12.02, + "learning_rate": 2.093163689473093e-05, + "loss": 0.0023, + "step": 49310 + }, + { + "epoch": 12.02, + "learning_rate": 2.092461521177396e-05, + "loss": 0.0, + "step": 49320 + }, + { + "epoch": 12.02, + "learning_rate": 2.0917593528816986e-05, + "loss": 0.0, + "step": 49330 + }, + { + "epoch": 12.02, + "learning_rate": 2.0910571845860017e-05, + "loss": 0.0021, + "step": 49340 + }, + { + "epoch": 12.02, + "learning_rate": 2.0903550162903048e-05, + "loss": 0.0006, + "step": 49350 + }, + { + "epoch": 12.02, + "learning_rate": 2.089652847994607e-05, + "loss": 0.0022, + "step": 49360 + }, + { + "epoch": 12.02, + "learning_rate": 2.0889506796989103e-05, + "loss": 0.0007, + "step": 49370 + }, + { + "epoch": 12.02, + "learning_rate": 2.0882485114032133e-05, + "loss": 0.1424, + "step": 49380 + }, + { + "epoch": 12.02, + "learning_rate": 2.087546343107516e-05, + "loss": 0.0, + "step": 49390 + }, + { + "epoch": 12.02, + "learning_rate": 2.086844174811819e-05, + "loss": 0.0816, + "step": 49400 + }, + { + "epoch": 12.02, + "learning_rate": 2.086142006516122e-05, + "loss": 0.0058, + "step": 49410 + }, + { + "epoch": 12.02, + "learning_rate": 2.085439838220425e-05, + "loss": 0.0002, + "step": 49420 + }, + { + "epoch": 12.02, + "learning_rate": 2.0847376699247274e-05, + "loss": 0.0002, + "step": 49430 + }, + { + "epoch": 12.02, + "learning_rate": 2.0840355016290305e-05, + "loss": 0.0003, + "step": 49440 + }, + { + "epoch": 12.03, + "learning_rate": 2.0833333333333336e-05, + "loss": 0.0002, + "step": 49450 + }, + { + "epoch": 12.03, + "learning_rate": 2.0826311650376364e-05, + "loss": 0.0003, + "step": 49460 + }, + { + "epoch": 12.03, + "learning_rate": 2.081928996741939e-05, + "loss": 0.0033, + "step": 49470 + }, + { + "epoch": 12.03, + "learning_rate": 2.0812268284462422e-05, + "loss": 0.0001, + "step": 49480 + }, + { + "epoch": 12.03, + "learning_rate": 2.080524660150545e-05, + "loss": 0.0002, + "step": 49490 + }, + { + "epoch": 12.03, + "learning_rate": 2.0798224918548477e-05, + "loss": 0.0001, + "step": 49500 + }, + { + "epoch": 12.03, + "learning_rate": 2.0791203235591508e-05, + "loss": 0.0, + "step": 49510 + }, + { + "epoch": 12.03, + "learning_rate": 2.0784181552634535e-05, + "loss": 0.0, + "step": 49520 + }, + { + "epoch": 12.03, + "learning_rate": 2.0777159869677566e-05, + "loss": 0.0908, + "step": 49530 + }, + { + "epoch": 12.03, + "learning_rate": 2.0770138186720594e-05, + "loss": 0.0004, + "step": 49540 + }, + { + "epoch": 12.03, + "learning_rate": 2.0763116503763624e-05, + "loss": 0.0, + "step": 49550 + }, + { + "epoch": 12.03, + "learning_rate": 2.0756094820806652e-05, + "loss": 0.0026, + "step": 49560 + }, + { + "epoch": 12.03, + "learning_rate": 2.074907313784968e-05, + "loss": 0.0004, + "step": 49570 + }, + { + "epoch": 12.03, + "learning_rate": 2.074205145489271e-05, + "loss": 0.0, + "step": 49580 + }, + { + "epoch": 12.03, + "learning_rate": 2.0735029771935738e-05, + "loss": 0.3153, + "step": 49590 + }, + { + "epoch": 12.03, + "learning_rate": 2.072800808897877e-05, + "loss": 0.164, + "step": 49600 + }, + { + "epoch": 12.03, + "learning_rate": 2.0720986406021796e-05, + "loss": 0.0408, + "step": 49610 + }, + { + "epoch": 12.03, + "learning_rate": 2.0713964723064824e-05, + "loss": 0.031, + "step": 49620 + }, + { + "epoch": 12.03, + "learning_rate": 2.0706943040107855e-05, + "loss": 0.0612, + "step": 49630 + }, + { + "epoch": 12.03, + "learning_rate": 2.0699921357150885e-05, + "loss": 0.1663, + "step": 49640 + }, + { + "epoch": 12.03, + "learning_rate": 2.069289967419391e-05, + "loss": 0.0, + "step": 49650 + }, + { + "epoch": 12.03, + "learning_rate": 2.068587799123694e-05, + "loss": 0.7401, + "step": 49660 + }, + { + "epoch": 12.03, + "learning_rate": 2.067885630827997e-05, + "loss": 0.0227, + "step": 49670 + }, + { + "epoch": 12.03, + "learning_rate": 2.0671834625323e-05, + "loss": 0.0, + "step": 49680 + }, + { + "epoch": 12.03, + "learning_rate": 2.0664812942366026e-05, + "loss": 0.1347, + "step": 49690 + }, + { + "epoch": 12.03, + "learning_rate": 2.0657791259409057e-05, + "loss": 0.3155, + "step": 49700 + }, + { + "epoch": 12.03, + "learning_rate": 2.0650769576452085e-05, + "loss": 0.5072, + "step": 49710 + }, + { + "epoch": 12.03, + "learning_rate": 2.0643747893495112e-05, + "loss": 0.0, + "step": 49720 + }, + { + "epoch": 12.03, + "learning_rate": 2.0636726210538143e-05, + "loss": 0.0001, + "step": 49730 + }, + { + "epoch": 12.03, + "learning_rate": 2.0629704527581174e-05, + "loss": 0.0805, + "step": 49740 + }, + { + "epoch": 12.03, + "learning_rate": 2.06226828446242e-05, + "loss": 0.0, + "step": 49750 + }, + { + "epoch": 12.03, + "learning_rate": 2.061566116166723e-05, + "loss": 0.0022, + "step": 49760 + }, + { + "epoch": 12.03, + "learning_rate": 2.060863947871026e-05, + "loss": 0.0, + "step": 49770 + }, + { + "epoch": 12.03, + "learning_rate": 2.0601617795753287e-05, + "loss": 0.4356, + "step": 49780 + }, + { + "epoch": 12.03, + "learning_rate": 2.0594596112796315e-05, + "loss": 0.0001, + "step": 49790 + }, + { + "epoch": 12.03, + "learning_rate": 2.0587574429839346e-05, + "loss": 0.0022, + "step": 49800 + }, + { + "epoch": 12.03, + "learning_rate": 2.0580552746882373e-05, + "loss": 0.0, + "step": 49810 + }, + { + "epoch": 12.03, + "learning_rate": 2.0573531063925404e-05, + "loss": 0.0, + "step": 49820 + }, + { + "epoch": 12.03, + "learning_rate": 2.056650938096843e-05, + "loss": 0.1846, + "step": 49830 + }, + { + "epoch": 12.03, + "learning_rate": 2.055948769801146e-05, + "loss": 0.2317, + "step": 49840 + }, + { + "epoch": 12.03, + "learning_rate": 2.055246601505449e-05, + "loss": 0.0472, + "step": 49850 + }, + { + "epoch": 12.03, + "learning_rate": 2.0545444332097517e-05, + "loss": 0.0021, + "step": 49860 + }, + { + "epoch": 12.03, + "learning_rate": 2.0538422649140545e-05, + "loss": 0.0, + "step": 49870 + }, + { + "epoch": 12.03, + "learning_rate": 2.0531400966183576e-05, + "loss": 0.3346, + "step": 49880 + }, + { + "epoch": 12.03, + "learning_rate": 2.0524379283226607e-05, + "loss": 0.002, + "step": 49890 + }, + { + "epoch": 12.03, + "learning_rate": 2.0517357600269634e-05, + "loss": 0.0003, + "step": 49900 + }, + { + "epoch": 12.03, + "learning_rate": 2.051033591731266e-05, + "loss": 0.0123, + "step": 49910 + }, + { + "epoch": 12.03, + "learning_rate": 2.0503314234355692e-05, + "loss": 0.0001, + "step": 49920 + }, + { + "epoch": 12.03, + "learning_rate": 2.049629255139872e-05, + "loss": 0.0131, + "step": 49930 + }, + { + "epoch": 12.03, + "learning_rate": 2.0489270868441747e-05, + "loss": 0.0, + "step": 49940 + }, + { + "epoch": 12.03, + "learning_rate": 2.048224918548478e-05, + "loss": 0.0001, + "step": 49950 + }, + { + "epoch": 12.03, + "learning_rate": 2.047522750252781e-05, + "loss": 0.4968, + "step": 49960 + }, + { + "epoch": 12.03, + "learning_rate": 2.0468205819570833e-05, + "loss": 0.017, + "step": 49970 + }, + { + "epoch": 12.03, + "learning_rate": 2.0461184136613864e-05, + "loss": 0.0, + "step": 49980 + }, + { + "epoch": 12.03, + "learning_rate": 2.0454162453656895e-05, + "loss": 0.19, + "step": 49990 + }, + { + "epoch": 12.03, + "learning_rate": 2.0447140770699923e-05, + "loss": 0.0043, + "step": 50000 + }, + { + "epoch": 12.03, + "learning_rate": 2.044011908774295e-05, + "loss": 0.0149, + "step": 50010 + }, + { + "epoch": 12.03, + "learning_rate": 2.043309740478598e-05, + "loss": 0.0009, + "step": 50020 + }, + { + "epoch": 12.03, + "learning_rate": 2.042607572182901e-05, + "loss": 0.1492, + "step": 50030 + }, + { + "epoch": 12.03, + "learning_rate": 2.0419054038872036e-05, + "loss": 0.4887, + "step": 50040 + }, + { + "epoch": 12.03, + "learning_rate": 2.0412032355915067e-05, + "loss": 0.0121, + "step": 50050 + }, + { + "epoch": 12.03, + "learning_rate": 2.0405010672958094e-05, + "loss": 0.0855, + "step": 50060 + }, + { + "epoch": 12.03, + "learning_rate": 2.0397988990001125e-05, + "loss": 0.1531, + "step": 50070 + }, + { + "epoch": 12.03, + "learning_rate": 2.0390967307044153e-05, + "loss": 0.0001, + "step": 50080 + }, + { + "epoch": 12.03, + "learning_rate": 2.0383945624087184e-05, + "loss": 0.0035, + "step": 50090 + }, + { + "epoch": 12.03, + "learning_rate": 2.037692394113021e-05, + "loss": 0.3237, + "step": 50100 + }, + { + "epoch": 12.03, + "learning_rate": 2.036990225817324e-05, + "loss": 0.0, + "step": 50110 + }, + { + "epoch": 12.03, + "learning_rate": 2.036288057521627e-05, + "loss": 0.4153, + "step": 50120 + }, + { + "epoch": 12.03, + "learning_rate": 2.0355858892259297e-05, + "loss": 0.0, + "step": 50130 + }, + { + "epoch": 12.03, + "learning_rate": 2.0348837209302328e-05, + "loss": 0.0003, + "step": 50140 + }, + { + "epoch": 12.03, + "learning_rate": 2.0341815526345355e-05, + "loss": 0.0, + "step": 50150 + }, + { + "epoch": 12.03, + "learning_rate": 2.0334793843388383e-05, + "loss": 0.1283, + "step": 50160 + }, + { + "epoch": 12.03, + "learning_rate": 2.0327772160431414e-05, + "loss": 0.0, + "step": 50170 + }, + { + "epoch": 12.03, + "learning_rate": 2.0320750477474444e-05, + "loss": 0.0002, + "step": 50180 + }, + { + "epoch": 12.03, + "learning_rate": 2.031372879451747e-05, + "loss": 0.0, + "step": 50190 + }, + { + "epoch": 12.03, + "learning_rate": 2.03067071115605e-05, + "loss": 0.084, + "step": 50200 + }, + { + "epoch": 12.03, + "learning_rate": 2.029968542860353e-05, + "loss": 0.2528, + "step": 50210 + }, + { + "epoch": 12.03, + "learning_rate": 2.0292663745646558e-05, + "loss": 0.9576, + "step": 50220 + }, + { + "epoch": 12.03, + "learning_rate": 2.0285642062689585e-05, + "loss": 0.3412, + "step": 50230 + }, + { + "epoch": 12.03, + "learning_rate": 2.0278620379732616e-05, + "loss": 0.0051, + "step": 50240 + }, + { + "epoch": 12.04, + "learning_rate": 2.0271598696775647e-05, + "loss": 0.0006, + "step": 50250 + }, + { + "epoch": 12.04, + "learning_rate": 2.026457701381867e-05, + "loss": 0.0255, + "step": 50260 + }, + { + "epoch": 12.04, + "learning_rate": 2.0257555330861702e-05, + "loss": 0.0, + "step": 50270 + }, + { + "epoch": 12.04, + "learning_rate": 2.0250533647904733e-05, + "loss": 0.0081, + "step": 50280 + }, + { + "epoch": 12.04, + "learning_rate": 2.024351196494776e-05, + "loss": 0.0146, + "step": 50290 + }, + { + "epoch": 12.04, + "learning_rate": 2.0236490281990788e-05, + "loss": 0.0, + "step": 50300 + }, + { + "epoch": 12.04, + "learning_rate": 2.022946859903382e-05, + "loss": 0.1895, + "step": 50310 + }, + { + "epoch": 12.04, + "learning_rate": 2.0222446916076846e-05, + "loss": 0.0255, + "step": 50320 + }, + { + "epoch": 12.04, + "learning_rate": 2.0215425233119874e-05, + "loss": 0.0067, + "step": 50330 + }, + { + "epoch": 12.04, + "learning_rate": 2.0208403550162905e-05, + "loss": 0.0, + "step": 50340 + }, + { + "epoch": 12.04, + "learning_rate": 2.0201381867205932e-05, + "loss": 0.0002, + "step": 50350 + }, + { + "epoch": 12.04, + "learning_rate": 2.0194360184248963e-05, + "loss": 0.0916, + "step": 50360 + }, + { + "epoch": 12.04, + "learning_rate": 2.018733850129199e-05, + "loss": 0.0, + "step": 50370 + }, + { + "epoch": 12.04, + "learning_rate": 2.0180316818335018e-05, + "loss": 0.0001, + "step": 50380 + }, + { + "epoch": 12.04, + "learning_rate": 2.017329513537805e-05, + "loss": 0.0, + "step": 50390 + }, + { + "epoch": 12.04, + "learning_rate": 2.0166273452421076e-05, + "loss": 0.0001, + "step": 50400 + }, + { + "epoch": 12.04, + "learning_rate": 2.0159251769464104e-05, + "loss": 0.0, + "step": 50410 + }, + { + "epoch": 12.04, + "learning_rate": 2.0152230086507135e-05, + "loss": 0.0019, + "step": 50420 + }, + { + "epoch": 12.04, + "learning_rate": 2.0145208403550166e-05, + "loss": 0.0001, + "step": 50430 + }, + { + "epoch": 12.04, + "learning_rate": 2.0138186720593193e-05, + "loss": 0.0, + "step": 50440 + }, + { + "epoch": 12.04, + "learning_rate": 2.013116503763622e-05, + "loss": 0.0001, + "step": 50450 + }, + { + "epoch": 12.04, + "learning_rate": 2.012414335467925e-05, + "loss": 0.0506, + "step": 50460 + }, + { + "epoch": 12.04, + "learning_rate": 2.011712167172228e-05, + "loss": 0.0, + "step": 50470 + }, + { + "epoch": 12.04, + "learning_rate": 2.0110099988765306e-05, + "loss": 0.0, + "step": 50480 + }, + { + "epoch": 12.04, + "learning_rate": 2.0103078305808337e-05, + "loss": 0.0562, + "step": 50490 + }, + { + "epoch": 12.04, + "learning_rate": 2.0096056622851368e-05, + "loss": 0.0, + "step": 50500 + }, + { + "epoch": 12.04, + "learning_rate": 2.0089034939894392e-05, + "loss": 0.027, + "step": 50510 + }, + { + "epoch": 12.04, + "learning_rate": 2.0082013256937423e-05, + "loss": 0.0079, + "step": 50520 + }, + { + "epoch": 12.04, + "learning_rate": 2.0074991573980454e-05, + "loss": 0.0, + "step": 50530 + }, + { + "epoch": 12.04, + "learning_rate": 2.006796989102348e-05, + "loss": 0.0, + "step": 50540 + }, + { + "epoch": 12.04, + "learning_rate": 2.006094820806651e-05, + "loss": 0.0001, + "step": 50550 + }, + { + "epoch": 12.04, + "learning_rate": 2.005392652510954e-05, + "loss": 0.0164, + "step": 50560 + }, + { + "epoch": 12.04, + "learning_rate": 2.0046904842152567e-05, + "loss": 0.0001, + "step": 50570 + }, + { + "epoch": 12.04, + "learning_rate": 2.0039883159195595e-05, + "loss": 0.0001, + "step": 50580 + }, + { + "epoch": 12.04, + "learning_rate": 2.0032861476238626e-05, + "loss": 0.5307, + "step": 50590 + }, + { + "epoch": 12.04, + "learning_rate": 2.0025839793281657e-05, + "loss": 0.006, + "step": 50600 + }, + { + "epoch": 12.04, + "learning_rate": 2.0018818110324684e-05, + "loss": 0.0001, + "step": 50610 + }, + { + "epoch": 12.04, + "learning_rate": 2.001179642736771e-05, + "loss": 0.0114, + "step": 50620 + }, + { + "epoch": 12.04, + "learning_rate": 2.0004774744410743e-05, + "loss": 0.0003, + "step": 50630 + }, + { + "epoch": 12.04, + "learning_rate": 1.999775306145377e-05, + "loss": 0.0074, + "step": 50640 + }, + { + "epoch": 12.04, + "learning_rate": 1.99907313784968e-05, + "loss": 0.01, + "step": 50650 + }, + { + "epoch": 12.04, + "learning_rate": 1.998370969553983e-05, + "loss": 0.0001, + "step": 50660 + }, + { + "epoch": 12.04, + "learning_rate": 1.9976688012582856e-05, + "loss": 0.1403, + "step": 50670 + }, + { + "epoch": 12.04, + "learning_rate": 1.9969666329625887e-05, + "loss": 0.0001, + "step": 50680 + }, + { + "epoch": 12.04, + "learning_rate": 1.9962644646668914e-05, + "loss": 0.008, + "step": 50690 + }, + { + "epoch": 12.04, + "learning_rate": 1.9955622963711942e-05, + "loss": 0.0, + "step": 50700 + }, + { + "epoch": 12.04, + "learning_rate": 1.9948601280754973e-05, + "loss": 0.4151, + "step": 50710 + }, + { + "epoch": 12.04, + "learning_rate": 1.9941579597798004e-05, + "loss": 0.0012, + "step": 50720 + }, + { + "epoch": 12.04, + "learning_rate": 1.9934557914841028e-05, + "loss": 0.0, + "step": 50730 + }, + { + "epoch": 12.04, + "learning_rate": 1.992753623188406e-05, + "loss": 0.0029, + "step": 50740 + }, + { + "epoch": 12.04, + "learning_rate": 1.992051454892709e-05, + "loss": 0.0034, + "step": 50750 + }, + { + "epoch": 12.04, + "learning_rate": 1.9913492865970117e-05, + "loss": 0.0, + "step": 50760 + }, + { + "epoch": 12.04, + "learning_rate": 1.9906471183013144e-05, + "loss": 0.0568, + "step": 50770 + }, + { + "epoch": 12.04, + "learning_rate": 1.9899449500056175e-05, + "loss": 0.0001, + "step": 50780 + }, + { + "epoch": 12.04, + "learning_rate": 1.9892427817099206e-05, + "loss": 0.0031, + "step": 50790 + }, + { + "epoch": 12.04, + "learning_rate": 1.988540613414223e-05, + "loss": 0.0108, + "step": 50800 + }, + { + "epoch": 12.04, + "learning_rate": 1.987838445118526e-05, + "loss": 0.0004, + "step": 50810 + }, + { + "epoch": 12.04, + "learning_rate": 1.9871362768228292e-05, + "loss": 0.005, + "step": 50820 + }, + { + "epoch": 12.04, + "learning_rate": 1.986434108527132e-05, + "loss": 0.0115, + "step": 50830 + }, + { + "epoch": 12.04, + "learning_rate": 1.9857319402314347e-05, + "loss": 0.0, + "step": 50840 + }, + { + "epoch": 12.04, + "learning_rate": 1.9850297719357378e-05, + "loss": 0.0, + "step": 50850 + }, + { + "epoch": 12.04, + "learning_rate": 1.9843276036400405e-05, + "loss": 0.0007, + "step": 50860 + }, + { + "epoch": 12.04, + "learning_rate": 1.9836254353443433e-05, + "loss": 0.0, + "step": 50870 + }, + { + "epoch": 12.04, + "learning_rate": 1.9829232670486464e-05, + "loss": 0.0, + "step": 50880 + }, + { + "epoch": 12.04, + "learning_rate": 1.982221098752949e-05, + "loss": 0.0, + "step": 50890 + }, + { + "epoch": 12.04, + "learning_rate": 1.9815189304572522e-05, + "loss": 0.0009, + "step": 50900 + }, + { + "epoch": 12.04, + "learning_rate": 1.980816762161555e-05, + "loss": 0.0004, + "step": 50910 + }, + { + "epoch": 12.04, + "learning_rate": 1.9801145938658577e-05, + "loss": 0.0642, + "step": 50920 + }, + { + "epoch": 12.04, + "learning_rate": 1.9794124255701608e-05, + "loss": 0.0009, + "step": 50930 + }, + { + "epoch": 12.04, + "learning_rate": 1.9787102572744635e-05, + "loss": 0.0305, + "step": 50940 + }, + { + "epoch": 12.04, + "learning_rate": 1.9780080889787666e-05, + "loss": 0.0047, + "step": 50950 + }, + { + "epoch": 12.04, + "learning_rate": 1.9773059206830694e-05, + "loss": 0.392, + "step": 50960 + }, + { + "epoch": 12.04, + "learning_rate": 1.9766037523873725e-05, + "loss": 0.0, + "step": 50970 + }, + { + "epoch": 12.04, + "learning_rate": 1.9759015840916752e-05, + "loss": 0.0, + "step": 50980 + }, + { + "epoch": 12.04, + "learning_rate": 1.975199415795978e-05, + "loss": 0.4877, + "step": 50990 + }, + { + "epoch": 12.04, + "learning_rate": 1.974497247500281e-05, + "loss": 0.2772, + "step": 51000 + }, + { + "epoch": 12.04, + "learning_rate": 1.9737950792045838e-05, + "loss": 0.0, + "step": 51010 + }, + { + "epoch": 12.04, + "learning_rate": 1.9730929109088865e-05, + "loss": 0.0, + "step": 51020 + }, + { + "epoch": 12.04, + "learning_rate": 1.9723907426131896e-05, + "loss": 0.0, + "step": 51030 + }, + { + "epoch": 12.05, + "learning_rate": 1.9716885743174927e-05, + "loss": 0.0001, + "step": 51040 + }, + { + "epoch": 12.05, + "learning_rate": 1.970986406021795e-05, + "loss": 0.0, + "step": 51050 + }, + { + "epoch": 12.05, + "learning_rate": 1.9702842377260982e-05, + "loss": 0.2685, + "step": 51060 + }, + { + "epoch": 12.05, + "learning_rate": 1.9695820694304013e-05, + "loss": 0.0, + "step": 51070 + }, + { + "epoch": 12.05, + "learning_rate": 1.968879901134704e-05, + "loss": 0.0709, + "step": 51080 + }, + { + "epoch": 12.05, + "learning_rate": 1.9681777328390068e-05, + "loss": 0.2393, + "step": 51090 + }, + { + "epoch": 12.05, + "learning_rate": 1.96747556454331e-05, + "loss": 0.0, + "step": 51100 + }, + { + "epoch": 12.05, + "learning_rate": 1.9667733962476126e-05, + "loss": 0.0002, + "step": 51110 + }, + { + "epoch": 12.05, + "learning_rate": 1.9660712279519154e-05, + "loss": 0.0004, + "step": 51120 + }, + { + "epoch": 12.05, + "learning_rate": 1.9653690596562185e-05, + "loss": 0.0863, + "step": 51130 + }, + { + "epoch": 12.05, + "learning_rate": 1.9646668913605216e-05, + "loss": 0.0011, + "step": 51140 + }, + { + "epoch": 12.05, + "learning_rate": 1.9639647230648243e-05, + "loss": 0.0005, + "step": 51150 + }, + { + "epoch": 12.05, + "learning_rate": 1.963262554769127e-05, + "loss": 0.0019, + "step": 51160 + }, + { + "epoch": 12.05, + "learning_rate": 1.96256038647343e-05, + "loss": 0.0001, + "step": 51170 + }, + { + "epoch": 12.05, + "learning_rate": 1.961858218177733e-05, + "loss": 0.767, + "step": 51180 + }, + { + "epoch": 12.05, + "learning_rate": 1.961156049882036e-05, + "loss": 0.0, + "step": 51190 + }, + { + "epoch": 12.05, + "learning_rate": 1.9604538815863387e-05, + "loss": 0.0, + "step": 51200 + }, + { + "epoch": 12.05, + "learning_rate": 1.9597517132906415e-05, + "loss": 0.0025, + "step": 51210 + }, + { + "epoch": 12.05, + "learning_rate": 1.9590495449949446e-05, + "loss": 0.0191, + "step": 51220 + }, + { + "epoch": 12.05, + "learning_rate": 1.9583473766992473e-05, + "loss": 0.0001, + "step": 51230 + }, + { + "epoch": 12.05, + "learning_rate": 1.95764520840355e-05, + "loss": 0.0001, + "step": 51240 + }, + { + "epoch": 12.05, + "learning_rate": 1.956943040107853e-05, + "loss": 0.2818, + "step": 51250 + }, + { + "epoch": 12.05, + "learning_rate": 1.9562408718121563e-05, + "loss": 0.3832, + "step": 51260 + }, + { + "epoch": 12.05, + "learning_rate": 1.9555387035164587e-05, + "loss": 0.023, + "step": 51270 + }, + { + "epoch": 12.05, + "learning_rate": 1.9548365352207618e-05, + "loss": 0.0655, + "step": 51280 + }, + { + "epoch": 12.05, + "learning_rate": 1.954134366925065e-05, + "loss": 0.0684, + "step": 51290 + }, + { + "epoch": 12.05, + "learning_rate": 1.9534321986293676e-05, + "loss": 0.2302, + "step": 51300 + }, + { + "epoch": 12.05, + "learning_rate": 1.9527300303336703e-05, + "loss": 0.4798, + "step": 51310 + }, + { + "epoch": 12.05, + "learning_rate": 1.9520278620379734e-05, + "loss": 0.0005, + "step": 51320 + }, + { + "epoch": 12.05, + "learning_rate": 1.9513256937422765e-05, + "loss": 0.0, + "step": 51330 + }, + { + "epoch": 12.05, + "learning_rate": 1.950623525446579e-05, + "loss": 0.0, + "step": 51340 + }, + { + "epoch": 12.05, + "learning_rate": 1.949921357150882e-05, + "loss": 0.0005, + "step": 51350 + }, + { + "epoch": 12.05, + "learning_rate": 1.949219188855185e-05, + "loss": 0.818, + "step": 51360 + }, + { + "epoch": 12.05, + "learning_rate": 1.948517020559488e-05, + "loss": 0.0019, + "step": 51370 + }, + { + "epoch": 12.05, + "learning_rate": 1.9478148522637906e-05, + "loss": 0.0, + "step": 51380 + }, + { + "epoch": 12.05, + "learning_rate": 1.9471126839680937e-05, + "loss": 0.0, + "step": 51390 + }, + { + "epoch": 12.05, + "learning_rate": 1.9464105156723964e-05, + "loss": 0.0007, + "step": 51400 + }, + { + "epoch": 12.05, + "learning_rate": 1.9457083473766992e-05, + "loss": 0.4402, + "step": 51410 + }, + { + "epoch": 12.05, + "learning_rate": 1.9450061790810023e-05, + "loss": 0.0016, + "step": 51420 + }, + { + "epoch": 12.05, + "eval_accuracy": 0.7206812024254935, + "eval_loss": 1.9810506105422974, + "eval_runtime": 1234.8657, + "eval_samples_per_second": 6.277, + "eval_steps_per_second": 3.139, + "step": 51428 + }, + { + "epoch": 13.0, + "learning_rate": 1.944304010785305e-05, + "loss": 0.6745, + "step": 51430 + }, + { + "epoch": 13.0, + "learning_rate": 1.943601842489608e-05, + "loss": 0.0002, + "step": 51440 + }, + { + "epoch": 13.0, + "learning_rate": 1.942899674193911e-05, + "loss": 0.0157, + "step": 51450 + }, + { + "epoch": 13.0, + "learning_rate": 1.9421975058982136e-05, + "loss": 0.0102, + "step": 51460 + }, + { + "epoch": 13.0, + "learning_rate": 1.9414953376025167e-05, + "loss": 0.0, + "step": 51470 + }, + { + "epoch": 13.0, + "learning_rate": 1.9407931693068194e-05, + "loss": 0.0, + "step": 51480 + }, + { + "epoch": 13.0, + "learning_rate": 1.9400910010111225e-05, + "loss": 0.0887, + "step": 51490 + }, + { + "epoch": 13.0, + "learning_rate": 1.9393888327154253e-05, + "loss": 0.0005, + "step": 51500 + }, + { + "epoch": 13.0, + "learning_rate": 1.9386866644197284e-05, + "loss": 0.026, + "step": 51510 + }, + { + "epoch": 13.0, + "learning_rate": 1.937984496124031e-05, + "loss": 0.0004, + "step": 51520 + }, + { + "epoch": 13.0, + "learning_rate": 1.937282327828334e-05, + "loss": 0.0108, + "step": 51530 + }, + { + "epoch": 13.0, + "learning_rate": 1.936580159532637e-05, + "loss": 0.0, + "step": 51540 + }, + { + "epoch": 13.0, + "learning_rate": 1.9358779912369397e-05, + "loss": 0.0, + "step": 51550 + }, + { + "epoch": 13.0, + "learning_rate": 1.9351758229412425e-05, + "loss": 0.0002, + "step": 51560 + }, + { + "epoch": 13.0, + "learning_rate": 1.9344736546455455e-05, + "loss": 0.0, + "step": 51570 + }, + { + "epoch": 13.0, + "learning_rate": 1.9337714863498486e-05, + "loss": 0.0001, + "step": 51580 + }, + { + "epoch": 13.0, + "learning_rate": 1.933069318054151e-05, + "loss": 0.0006, + "step": 51590 + }, + { + "epoch": 13.0, + "learning_rate": 1.932367149758454e-05, + "loss": 0.0, + "step": 51600 + }, + { + "epoch": 13.0, + "learning_rate": 1.9316649814627572e-05, + "loss": 0.0, + "step": 51610 + }, + { + "epoch": 13.0, + "learning_rate": 1.93096281316706e-05, + "loss": 0.0001, + "step": 51620 + }, + { + "epoch": 13.0, + "learning_rate": 1.9302606448713627e-05, + "loss": 0.0, + "step": 51630 + }, + { + "epoch": 13.0, + "learning_rate": 1.9295584765756658e-05, + "loss": 0.0, + "step": 51640 + }, + { + "epoch": 13.0, + "learning_rate": 1.928856308279969e-05, + "loss": 0.7972, + "step": 51650 + }, + { + "epoch": 13.0, + "learning_rate": 1.9281541399842716e-05, + "loss": 0.2388, + "step": 51660 + }, + { + "epoch": 13.0, + "learning_rate": 1.9274519716885744e-05, + "loss": 0.0, + "step": 51670 + }, + { + "epoch": 13.0, + "learning_rate": 1.9267498033928775e-05, + "loss": 0.0001, + "step": 51680 + }, + { + "epoch": 13.0, + "learning_rate": 1.9260476350971802e-05, + "loss": 0.0001, + "step": 51690 + }, + { + "epoch": 13.0, + "learning_rate": 1.925345466801483e-05, + "loss": 0.0, + "step": 51700 + }, + { + "epoch": 13.0, + "learning_rate": 1.924643298505786e-05, + "loss": 0.0104, + "step": 51710 + }, + { + "epoch": 13.0, + "learning_rate": 1.9239411302100888e-05, + "loss": 0.0021, + "step": 51720 + }, + { + "epoch": 13.0, + "learning_rate": 1.923238961914392e-05, + "loss": 0.0001, + "step": 51730 + }, + { + "epoch": 13.0, + "learning_rate": 1.9225367936186946e-05, + "loss": 0.0057, + "step": 51740 + }, + { + "epoch": 13.0, + "learning_rate": 1.9218346253229974e-05, + "loss": 0.0, + "step": 51750 + }, + { + "epoch": 13.0, + "learning_rate": 1.9211324570273005e-05, + "loss": 0.0922, + "step": 51760 + }, + { + "epoch": 13.0, + "learning_rate": 1.9204302887316032e-05, + "loss": 0.0025, + "step": 51770 + }, + { + "epoch": 13.0, + "learning_rate": 1.919728120435906e-05, + "loss": 0.0027, + "step": 51780 + }, + { + "epoch": 13.0, + "learning_rate": 1.919025952140209e-05, + "loss": 0.1292, + "step": 51790 + }, + { + "epoch": 13.0, + "learning_rate": 1.918323783844512e-05, + "loss": 0.0, + "step": 51800 + }, + { + "epoch": 13.0, + "learning_rate": 1.9176216155488146e-05, + "loss": 0.0715, + "step": 51810 + }, + { + "epoch": 13.0, + "learning_rate": 1.9169194472531177e-05, + "loss": 0.0, + "step": 51820 + }, + { + "epoch": 13.01, + "learning_rate": 1.9162172789574207e-05, + "loss": 0.0014, + "step": 51830 + }, + { + "epoch": 13.01, + "learning_rate": 1.9155151106617235e-05, + "loss": 0.0, + "step": 51840 + }, + { + "epoch": 13.01, + "learning_rate": 1.9148129423660262e-05, + "loss": 0.0, + "step": 51850 + }, + { + "epoch": 13.01, + "learning_rate": 1.9141107740703293e-05, + "loss": 0.0043, + "step": 51860 + }, + { + "epoch": 13.01, + "learning_rate": 1.9134086057746324e-05, + "loss": 0.0005, + "step": 51870 + }, + { + "epoch": 13.01, + "learning_rate": 1.9127064374789348e-05, + "loss": 0.1109, + "step": 51880 + }, + { + "epoch": 13.01, + "learning_rate": 1.912004269183238e-05, + "loss": 0.0, + "step": 51890 + }, + { + "epoch": 13.01, + "learning_rate": 1.911302100887541e-05, + "loss": 0.0, + "step": 51900 + }, + { + "epoch": 13.01, + "learning_rate": 1.9105999325918438e-05, + "loss": 0.0035, + "step": 51910 + }, + { + "epoch": 13.01, + "learning_rate": 1.9098977642961465e-05, + "loss": 0.0, + "step": 51920 + }, + { + "epoch": 13.01, + "learning_rate": 1.9091955960004496e-05, + "loss": 0.0, + "step": 51930 + }, + { + "epoch": 13.01, + "learning_rate": 1.9084934277047523e-05, + "loss": 0.004, + "step": 51940 + }, + { + "epoch": 13.01, + "learning_rate": 1.907791259409055e-05, + "loss": 0.0, + "step": 51950 + }, + { + "epoch": 13.01, + "learning_rate": 1.9070890911133582e-05, + "loss": 0.1042, + "step": 51960 + }, + { + "epoch": 13.01, + "learning_rate": 1.906386922817661e-05, + "loss": 0.0016, + "step": 51970 + }, + { + "epoch": 13.01, + "learning_rate": 1.905684754521964e-05, + "loss": 0.0001, + "step": 51980 + }, + { + "epoch": 13.01, + "learning_rate": 1.9049825862262668e-05, + "loss": 0.0, + "step": 51990 + }, + { + "epoch": 13.01, + "learning_rate": 1.90428041793057e-05, + "loss": 0.1256, + "step": 52000 + }, + { + "epoch": 13.01, + "learning_rate": 1.9035782496348726e-05, + "loss": 0.0, + "step": 52010 + }, + { + "epoch": 13.01, + "learning_rate": 1.9028760813391753e-05, + "loss": 0.0, + "step": 52020 + }, + { + "epoch": 13.01, + "learning_rate": 1.9021739130434784e-05, + "loss": 0.0004, + "step": 52030 + }, + { + "epoch": 13.01, + "learning_rate": 1.9014717447477812e-05, + "loss": 0.0, + "step": 52040 + }, + { + "epoch": 13.01, + "learning_rate": 1.9007695764520843e-05, + "loss": 0.0001, + "step": 52050 + }, + { + "epoch": 13.01, + "learning_rate": 1.900067408156387e-05, + "loss": 0.0001, + "step": 52060 + }, + { + "epoch": 13.01, + "learning_rate": 1.8993652398606898e-05, + "loss": 0.0, + "step": 52070 + }, + { + "epoch": 13.01, + "learning_rate": 1.898663071564993e-05, + "loss": 0.0001, + "step": 52080 + }, + { + "epoch": 13.01, + "learning_rate": 1.8979609032692956e-05, + "loss": 0.0, + "step": 52090 + }, + { + "epoch": 13.01, + "learning_rate": 1.8972587349735984e-05, + "loss": 0.3969, + "step": 52100 + }, + { + "epoch": 13.01, + "learning_rate": 1.8965565666779014e-05, + "loss": 0.0002, + "step": 52110 + }, + { + "epoch": 13.01, + "learning_rate": 1.8958543983822045e-05, + "loss": 0.0, + "step": 52120 + }, + { + "epoch": 13.01, + "learning_rate": 1.895152230086507e-05, + "loss": 0.0, + "step": 52130 + }, + { + "epoch": 13.01, + "learning_rate": 1.89445006179081e-05, + "loss": 0.0, + "step": 52140 + }, + { + "epoch": 13.01, + "learning_rate": 1.893747893495113e-05, + "loss": 0.0, + "step": 52150 + }, + { + "epoch": 13.01, + "learning_rate": 1.893045725199416e-05, + "loss": 0.0037, + "step": 52160 + }, + { + "epoch": 13.01, + "learning_rate": 1.8923435569037186e-05, + "loss": 0.1173, + "step": 52170 + }, + { + "epoch": 13.01, + "learning_rate": 1.8916413886080217e-05, + "loss": 0.0001, + "step": 52180 + }, + { + "epoch": 13.01, + "learning_rate": 1.8909392203123248e-05, + "loss": 0.0, + "step": 52190 + }, + { + "epoch": 13.01, + "learning_rate": 1.8902370520166275e-05, + "loss": 0.0003, + "step": 52200 + }, + { + "epoch": 13.01, + "learning_rate": 1.8895348837209303e-05, + "loss": 0.0, + "step": 52210 + }, + { + "epoch": 13.01, + "learning_rate": 1.8888327154252334e-05, + "loss": 0.2224, + "step": 52220 + }, + { + "epoch": 13.01, + "learning_rate": 1.888130547129536e-05, + "loss": 0.0, + "step": 52230 + }, + { + "epoch": 13.01, + "learning_rate": 1.887428378833839e-05, + "loss": 0.2798, + "step": 52240 + }, + { + "epoch": 13.01, + "learning_rate": 1.886726210538142e-05, + "loss": 0.5383, + "step": 52250 + }, + { + "epoch": 13.01, + "learning_rate": 1.8860240422424447e-05, + "loss": 0.0001, + "step": 52260 + }, + { + "epoch": 13.01, + "learning_rate": 1.8853218739467478e-05, + "loss": 0.0153, + "step": 52270 + }, + { + "epoch": 13.01, + "learning_rate": 1.8846197056510505e-05, + "loss": 0.0001, + "step": 52280 + }, + { + "epoch": 13.01, + "learning_rate": 1.8839175373553533e-05, + "loss": 0.002, + "step": 52290 + }, + { + "epoch": 13.01, + "learning_rate": 1.8832153690596564e-05, + "loss": 0.0038, + "step": 52300 + }, + { + "epoch": 13.01, + "learning_rate": 1.882513200763959e-05, + "loss": 0.2871, + "step": 52310 + }, + { + "epoch": 13.01, + "learning_rate": 1.881811032468262e-05, + "loss": 0.0234, + "step": 52320 + }, + { + "epoch": 13.01, + "learning_rate": 1.881108864172565e-05, + "loss": 0.47, + "step": 52330 + }, + { + "epoch": 13.01, + "learning_rate": 1.880406695876868e-05, + "loss": 0.0001, + "step": 52340 + }, + { + "epoch": 13.01, + "learning_rate": 1.8797045275811708e-05, + "loss": 0.0, + "step": 52350 + }, + { + "epoch": 13.01, + "learning_rate": 1.8790023592854736e-05, + "loss": 0.0, + "step": 52360 + }, + { + "epoch": 13.01, + "learning_rate": 1.8783001909897766e-05, + "loss": 0.0054, + "step": 52370 + }, + { + "epoch": 13.01, + "learning_rate": 1.8775980226940794e-05, + "loss": 0.0438, + "step": 52380 + }, + { + "epoch": 13.01, + "learning_rate": 1.876895854398382e-05, + "loss": 0.0, + "step": 52390 + }, + { + "epoch": 13.01, + "learning_rate": 1.8761936861026852e-05, + "loss": 0.0, + "step": 52400 + }, + { + "epoch": 13.01, + "learning_rate": 1.8754915178069883e-05, + "loss": 0.0001, + "step": 52410 + }, + { + "epoch": 13.01, + "learning_rate": 1.8747893495112907e-05, + "loss": 0.0, + "step": 52420 + }, + { + "epoch": 13.01, + "learning_rate": 1.8740871812155938e-05, + "loss": 0.0079, + "step": 52430 + }, + { + "epoch": 13.01, + "learning_rate": 1.873385012919897e-05, + "loss": 0.0, + "step": 52440 + }, + { + "epoch": 13.01, + "learning_rate": 1.8726828446241997e-05, + "loss": 0.0009, + "step": 52450 + }, + { + "epoch": 13.01, + "learning_rate": 1.8719806763285024e-05, + "loss": 0.0002, + "step": 52460 + }, + { + "epoch": 13.01, + "learning_rate": 1.8712785080328055e-05, + "loss": 0.0, + "step": 52470 + }, + { + "epoch": 13.01, + "learning_rate": 1.8705763397371082e-05, + "loss": 0.0, + "step": 52480 + }, + { + "epoch": 13.01, + "learning_rate": 1.869874171441411e-05, + "loss": 0.0, + "step": 52490 + }, + { + "epoch": 13.01, + "learning_rate": 1.869172003145714e-05, + "loss": 0.0, + "step": 52500 + }, + { + "epoch": 13.01, + "learning_rate": 1.8684698348500168e-05, + "loss": 0.0, + "step": 52510 + }, + { + "epoch": 13.01, + "learning_rate": 1.86776766655432e-05, + "loss": 0.0003, + "step": 52520 + }, + { + "epoch": 13.01, + "learning_rate": 1.8670654982586227e-05, + "loss": 0.0001, + "step": 52530 + }, + { + "epoch": 13.01, + "learning_rate": 1.8663633299629257e-05, + "loss": 0.0001, + "step": 52540 + }, + { + "epoch": 13.01, + "learning_rate": 1.8656611616672285e-05, + "loss": 0.2261, + "step": 52550 + }, + { + "epoch": 13.01, + "learning_rate": 1.8649589933715312e-05, + "loss": 0.0, + "step": 52560 + }, + { + "epoch": 13.01, + "learning_rate": 1.8642568250758343e-05, + "loss": 0.0339, + "step": 52570 + }, + { + "epoch": 13.01, + "learning_rate": 1.863554656780137e-05, + "loss": 0.0715, + "step": 52580 + }, + { + "epoch": 13.01, + "learning_rate": 1.8628524884844402e-05, + "loss": 0.0, + "step": 52590 + }, + { + "epoch": 13.01, + "learning_rate": 1.862150320188743e-05, + "loss": 0.0037, + "step": 52600 + }, + { + "epoch": 13.01, + "learning_rate": 1.8614481518930457e-05, + "loss": 0.0003, + "step": 52610 + }, + { + "epoch": 13.02, + "learning_rate": 1.8607459835973488e-05, + "loss": 0.1538, + "step": 52620 + }, + { + "epoch": 13.02, + "learning_rate": 1.8600438153016515e-05, + "loss": 0.0, + "step": 52630 + }, + { + "epoch": 13.02, + "learning_rate": 1.8593416470059543e-05, + "loss": 0.0, + "step": 52640 + }, + { + "epoch": 13.02, + "learning_rate": 1.8586394787102573e-05, + "loss": 0.3644, + "step": 52650 + }, + { + "epoch": 13.02, + "learning_rate": 1.8579373104145604e-05, + "loss": 0.0001, + "step": 52660 + }, + { + "epoch": 13.02, + "learning_rate": 1.8572351421188632e-05, + "loss": 0.0187, + "step": 52670 + }, + { + "epoch": 13.02, + "learning_rate": 1.856532973823166e-05, + "loss": 0.0006, + "step": 52680 + }, + { + "epoch": 13.02, + "learning_rate": 1.855830805527469e-05, + "loss": 0.0, + "step": 52690 + }, + { + "epoch": 13.02, + "learning_rate": 1.855128637231772e-05, + "loss": 0.0155, + "step": 52700 + }, + { + "epoch": 13.02, + "learning_rate": 1.8544264689360745e-05, + "loss": 0.0251, + "step": 52710 + }, + { + "epoch": 13.02, + "learning_rate": 1.8537243006403776e-05, + "loss": 0.0001, + "step": 52720 + }, + { + "epoch": 13.02, + "learning_rate": 1.8530221323446807e-05, + "loss": 0.1756, + "step": 52730 + }, + { + "epoch": 13.02, + "learning_rate": 1.8523199640489834e-05, + "loss": 0.1006, + "step": 52740 + }, + { + "epoch": 13.02, + "learning_rate": 1.8516177957532862e-05, + "loss": 0.0001, + "step": 52750 + }, + { + "epoch": 13.02, + "learning_rate": 1.8509156274575893e-05, + "loss": 0.0149, + "step": 52760 + }, + { + "epoch": 13.02, + "learning_rate": 1.850213459161892e-05, + "loss": 0.0003, + "step": 52770 + }, + { + "epoch": 13.02, + "learning_rate": 1.8495112908661948e-05, + "loss": 0.0002, + "step": 52780 + }, + { + "epoch": 13.02, + "learning_rate": 1.848809122570498e-05, + "loss": 0.1618, + "step": 52790 + }, + { + "epoch": 13.02, + "learning_rate": 1.8481069542748006e-05, + "loss": 0.1566, + "step": 52800 + }, + { + "epoch": 13.02, + "learning_rate": 1.8474047859791037e-05, + "loss": 0.0, + "step": 52810 + }, + { + "epoch": 13.02, + "learning_rate": 1.8467026176834065e-05, + "loss": 0.0, + "step": 52820 + }, + { + "epoch": 13.02, + "learning_rate": 1.8460004493877092e-05, + "loss": 0.0003, + "step": 52830 + }, + { + "epoch": 13.02, + "learning_rate": 1.8452982810920123e-05, + "loss": 0.0, + "step": 52840 + }, + { + "epoch": 13.02, + "learning_rate": 1.844596112796315e-05, + "loss": 0.0012, + "step": 52850 + }, + { + "epoch": 13.02, + "learning_rate": 1.8438939445006178e-05, + "loss": 0.0001, + "step": 52860 + }, + { + "epoch": 13.02, + "learning_rate": 1.843191776204921e-05, + "loss": 0.0195, + "step": 52870 + }, + { + "epoch": 13.02, + "learning_rate": 1.842489607909224e-05, + "loss": 0.0596, + "step": 52880 + }, + { + "epoch": 13.02, + "learning_rate": 1.8417874396135267e-05, + "loss": 0.0, + "step": 52890 + }, + { + "epoch": 13.02, + "learning_rate": 1.8410852713178295e-05, + "loss": 0.0, + "step": 52900 + }, + { + "epoch": 13.02, + "learning_rate": 1.8403831030221325e-05, + "loss": 0.1333, + "step": 52910 + }, + { + "epoch": 13.02, + "learning_rate": 1.8396809347264353e-05, + "loss": 0.5977, + "step": 52920 + }, + { + "epoch": 13.02, + "learning_rate": 1.838978766430738e-05, + "loss": 0.0002, + "step": 52930 + }, + { + "epoch": 13.02, + "learning_rate": 1.838276598135041e-05, + "loss": 0.0, + "step": 52940 + }, + { + "epoch": 13.02, + "learning_rate": 1.8375744298393442e-05, + "loss": 0.0, + "step": 52950 + }, + { + "epoch": 13.02, + "learning_rate": 1.8368722615436466e-05, + "loss": 0.0001, + "step": 52960 + }, + { + "epoch": 13.02, + "learning_rate": 1.8361700932479497e-05, + "loss": 0.0016, + "step": 52970 + }, + { + "epoch": 13.02, + "learning_rate": 1.8354679249522528e-05, + "loss": 0.0002, + "step": 52980 + }, + { + "epoch": 13.02, + "learning_rate": 1.8347657566565556e-05, + "loss": 0.0004, + "step": 52990 + }, + { + "epoch": 13.02, + "learning_rate": 1.8340635883608583e-05, + "loss": 0.0387, + "step": 53000 + }, + { + "epoch": 13.02, + "learning_rate": 1.8333614200651614e-05, + "loss": 0.0001, + "step": 53010 + }, + { + "epoch": 13.02, + "learning_rate": 1.832659251769464e-05, + "loss": 0.0, + "step": 53020 + }, + { + "epoch": 13.02, + "learning_rate": 1.831957083473767e-05, + "loss": 0.003, + "step": 53030 + }, + { + "epoch": 13.02, + "learning_rate": 1.83125491517807e-05, + "loss": 0.0007, + "step": 53040 + }, + { + "epoch": 13.02, + "learning_rate": 1.830552746882373e-05, + "loss": 0.0, + "step": 53050 + }, + { + "epoch": 13.02, + "learning_rate": 1.8298505785866758e-05, + "loss": 0.0, + "step": 53060 + }, + { + "epoch": 13.02, + "learning_rate": 1.8291484102909786e-05, + "loss": 0.0003, + "step": 53070 + }, + { + "epoch": 13.02, + "learning_rate": 1.8284462419952817e-05, + "loss": 0.0007, + "step": 53080 + }, + { + "epoch": 13.02, + "learning_rate": 1.8277440736995844e-05, + "loss": 0.0, + "step": 53090 + }, + { + "epoch": 13.02, + "learning_rate": 1.827041905403887e-05, + "loss": 0.0005, + "step": 53100 + }, + { + "epoch": 13.02, + "learning_rate": 1.8263397371081902e-05, + "loss": 0.0003, + "step": 53110 + }, + { + "epoch": 13.02, + "learning_rate": 1.825637568812493e-05, + "loss": 0.1097, + "step": 53120 + }, + { + "epoch": 13.02, + "learning_rate": 1.824935400516796e-05, + "loss": 0.0, + "step": 53130 + }, + { + "epoch": 13.02, + "learning_rate": 1.8242332322210988e-05, + "loss": 0.4737, + "step": 53140 + }, + { + "epoch": 13.02, + "learning_rate": 1.8235310639254016e-05, + "loss": 0.2225, + "step": 53150 + }, + { + "epoch": 13.02, + "learning_rate": 1.8228288956297047e-05, + "loss": 0.0, + "step": 53160 + }, + { + "epoch": 13.02, + "learning_rate": 1.8221267273340074e-05, + "loss": 0.0, + "step": 53170 + }, + { + "epoch": 13.02, + "learning_rate": 1.82142455903831e-05, + "loss": 0.0001, + "step": 53180 + }, + { + "epoch": 13.02, + "learning_rate": 1.8207223907426132e-05, + "loss": 0.0, + "step": 53190 + }, + { + "epoch": 13.02, + "learning_rate": 1.8200202224469163e-05, + "loss": 0.0091, + "step": 53200 + }, + { + "epoch": 13.02, + "learning_rate": 1.819318054151219e-05, + "loss": 0.0241, + "step": 53210 + }, + { + "epoch": 13.02, + "learning_rate": 1.818615885855522e-05, + "loss": 0.0001, + "step": 53220 + }, + { + "epoch": 13.02, + "learning_rate": 1.817913717559825e-05, + "loss": 0.3982, + "step": 53230 + }, + { + "epoch": 13.02, + "learning_rate": 1.817211549264128e-05, + "loss": 0.0, + "step": 53240 + }, + { + "epoch": 13.02, + "learning_rate": 1.8165093809684304e-05, + "loss": 0.0001, + "step": 53250 + }, + { + "epoch": 13.02, + "learning_rate": 1.8158072126727335e-05, + "loss": 0.0, + "step": 53260 + }, + { + "epoch": 13.02, + "learning_rate": 1.8151050443770366e-05, + "loss": 0.0008, + "step": 53270 + }, + { + "epoch": 13.02, + "learning_rate": 1.8144028760813393e-05, + "loss": 0.0001, + "step": 53280 + }, + { + "epoch": 13.02, + "learning_rate": 1.813700707785642e-05, + "loss": 0.0001, + "step": 53290 + }, + { + "epoch": 13.02, + "learning_rate": 1.8129985394899452e-05, + "loss": 0.0051, + "step": 53300 + }, + { + "epoch": 13.02, + "learning_rate": 1.812296371194248e-05, + "loss": 0.3457, + "step": 53310 + }, + { + "epoch": 13.02, + "learning_rate": 1.8115942028985507e-05, + "loss": 0.0, + "step": 53320 + }, + { + "epoch": 13.02, + "learning_rate": 1.8108920346028538e-05, + "loss": 0.0005, + "step": 53330 + }, + { + "epoch": 13.02, + "learning_rate": 1.8101898663071565e-05, + "loss": 0.0, + "step": 53340 + }, + { + "epoch": 13.02, + "learning_rate": 1.8094876980114596e-05, + "loss": 0.2339, + "step": 53350 + }, + { + "epoch": 13.02, + "learning_rate": 1.8087855297157624e-05, + "loss": 0.0, + "step": 53360 + }, + { + "epoch": 13.02, + "learning_rate": 1.808083361420065e-05, + "loss": 0.0, + "step": 53370 + }, + { + "epoch": 13.02, + "learning_rate": 1.8073811931243682e-05, + "loss": 0.1258, + "step": 53380 + }, + { + "epoch": 13.02, + "learning_rate": 1.806679024828671e-05, + "loss": 0.0027, + "step": 53390 + }, + { + "epoch": 13.02, + "learning_rate": 1.8059768565329737e-05, + "loss": 0.0093, + "step": 53400 + }, + { + "epoch": 13.03, + "learning_rate": 1.8052746882372768e-05, + "loss": 0.0001, + "step": 53410 + }, + { + "epoch": 13.03, + "learning_rate": 1.80457251994158e-05, + "loss": 0.0, + "step": 53420 + }, + { + "epoch": 13.03, + "learning_rate": 1.8038703516458826e-05, + "loss": 0.0228, + "step": 53430 + }, + { + "epoch": 13.03, + "learning_rate": 1.8031681833501854e-05, + "loss": 0.4496, + "step": 53440 + }, + { + "epoch": 13.03, + "learning_rate": 1.8024660150544884e-05, + "loss": 0.0038, + "step": 53450 + }, + { + "epoch": 13.03, + "learning_rate": 1.8017638467587912e-05, + "loss": 0.0929, + "step": 53460 + }, + { + "epoch": 13.03, + "learning_rate": 1.801061678463094e-05, + "loss": 0.0077, + "step": 53470 + }, + { + "epoch": 13.03, + "learning_rate": 1.800359510167397e-05, + "loss": 0.0, + "step": 53480 + }, + { + "epoch": 13.03, + "learning_rate": 1.7996573418717e-05, + "loss": 0.0, + "step": 53490 + }, + { + "epoch": 13.03, + "learning_rate": 1.7989551735760025e-05, + "loss": 0.0018, + "step": 53500 + }, + { + "epoch": 13.03, + "learning_rate": 1.7982530052803056e-05, + "loss": 0.0, + "step": 53510 + }, + { + "epoch": 13.03, + "learning_rate": 1.7975508369846087e-05, + "loss": 0.0, + "step": 53520 + }, + { + "epoch": 13.03, + "learning_rate": 1.7968486686889115e-05, + "loss": 0.0004, + "step": 53530 + }, + { + "epoch": 13.03, + "learning_rate": 1.7961465003932142e-05, + "loss": 0.0, + "step": 53540 + }, + { + "epoch": 13.03, + "learning_rate": 1.7954443320975173e-05, + "loss": 0.0592, + "step": 53550 + }, + { + "epoch": 13.03, + "learning_rate": 1.79474216380182e-05, + "loss": 0.0, + "step": 53560 + }, + { + "epoch": 13.03, + "learning_rate": 1.7940399955061228e-05, + "loss": 0.0, + "step": 53570 + }, + { + "epoch": 13.03, + "learning_rate": 1.793337827210426e-05, + "loss": 0.0, + "step": 53580 + }, + { + "epoch": 13.03, + "learning_rate": 1.792635658914729e-05, + "loss": 0.0001, + "step": 53590 + }, + { + "epoch": 13.03, + "learning_rate": 1.7919334906190317e-05, + "loss": 0.0, + "step": 53600 + }, + { + "epoch": 13.03, + "learning_rate": 1.7912313223233345e-05, + "loss": 0.0, + "step": 53610 + }, + { + "epoch": 13.03, + "learning_rate": 1.7905291540276376e-05, + "loss": 0.1128, + "step": 53620 + }, + { + "epoch": 13.03, + "learning_rate": 1.7898269857319403e-05, + "loss": 0.2468, + "step": 53630 + }, + { + "epoch": 13.03, + "learning_rate": 1.789124817436243e-05, + "loss": 0.0001, + "step": 53640 + }, + { + "epoch": 13.03, + "learning_rate": 1.788422649140546e-05, + "loss": 0.0001, + "step": 53650 + }, + { + "epoch": 13.03, + "learning_rate": 1.787720480844849e-05, + "loss": 0.0001, + "step": 53660 + }, + { + "epoch": 13.03, + "learning_rate": 1.787018312549152e-05, + "loss": 0.0001, + "step": 53670 + }, + { + "epoch": 13.03, + "learning_rate": 1.7863161442534547e-05, + "loss": 0.0, + "step": 53680 + }, + { + "epoch": 13.03, + "learning_rate": 1.7856139759577575e-05, + "loss": 0.0, + "step": 53690 + }, + { + "epoch": 13.03, + "learning_rate": 1.7849118076620606e-05, + "loss": 0.0006, + "step": 53700 + }, + { + "epoch": 13.03, + "learning_rate": 1.7842096393663637e-05, + "loss": 0.0288, + "step": 53710 + }, + { + "epoch": 13.03, + "learning_rate": 1.783507471070666e-05, + "loss": 0.0, + "step": 53720 + }, + { + "epoch": 13.03, + "learning_rate": 1.782805302774969e-05, + "loss": 0.0014, + "step": 53730 + }, + { + "epoch": 13.03, + "learning_rate": 1.7821031344792722e-05, + "loss": 0.012, + "step": 53740 + }, + { + "epoch": 13.03, + "learning_rate": 1.781400966183575e-05, + "loss": 0.1161, + "step": 53750 + }, + { + "epoch": 13.03, + "learning_rate": 1.7806987978878777e-05, + "loss": 0.0001, + "step": 53760 + }, + { + "epoch": 13.03, + "learning_rate": 1.7799966295921808e-05, + "loss": 0.0768, + "step": 53770 + }, + { + "epoch": 13.03, + "learning_rate": 1.779294461296484e-05, + "loss": 0.0, + "step": 53780 + }, + { + "epoch": 13.03, + "learning_rate": 1.7785922930007863e-05, + "loss": 0.0, + "step": 53790 + }, + { + "epoch": 13.03, + "learning_rate": 1.7778901247050894e-05, + "loss": 0.0, + "step": 53800 + }, + { + "epoch": 13.03, + "learning_rate": 1.7771879564093925e-05, + "loss": 0.0, + "step": 53810 + }, + { + "epoch": 13.03, + "learning_rate": 1.7764857881136952e-05, + "loss": 0.0141, + "step": 53820 + }, + { + "epoch": 13.03, + "learning_rate": 1.775783619817998e-05, + "loss": 0.1587, + "step": 53830 + }, + { + "epoch": 13.03, + "learning_rate": 1.775081451522301e-05, + "loss": 0.0015, + "step": 53840 + }, + { + "epoch": 13.03, + "learning_rate": 1.774379283226604e-05, + "loss": 0.0001, + "step": 53850 + }, + { + "epoch": 13.03, + "learning_rate": 1.7736771149309066e-05, + "loss": 0.0, + "step": 53860 + }, + { + "epoch": 13.03, + "learning_rate": 1.7729749466352097e-05, + "loss": 0.0, + "step": 53870 + }, + { + "epoch": 13.03, + "learning_rate": 1.7722727783395124e-05, + "loss": 0.0, + "step": 53880 + }, + { + "epoch": 13.03, + "learning_rate": 1.7715706100438155e-05, + "loss": 0.0, + "step": 53890 + }, + { + "epoch": 13.03, + "learning_rate": 1.7708684417481183e-05, + "loss": 0.015, + "step": 53900 + }, + { + "epoch": 13.03, + "learning_rate": 1.770166273452421e-05, + "loss": 0.2302, + "step": 53910 + }, + { + "epoch": 13.03, + "learning_rate": 1.769464105156724e-05, + "loss": 0.0, + "step": 53920 + }, + { + "epoch": 13.03, + "learning_rate": 1.768761936861027e-05, + "loss": 0.0, + "step": 53930 + }, + { + "epoch": 13.03, + "learning_rate": 1.76805976856533e-05, + "loss": 0.0, + "step": 53940 + }, + { + "epoch": 13.03, + "learning_rate": 1.7673576002696327e-05, + "loss": 0.0, + "step": 53950 + }, + { + "epoch": 13.03, + "learning_rate": 1.7666554319739358e-05, + "loss": 0.0014, + "step": 53960 + }, + { + "epoch": 13.03, + "learning_rate": 1.7659532636782385e-05, + "loss": 0.0003, + "step": 53970 + }, + { + "epoch": 13.03, + "learning_rate": 1.7652510953825413e-05, + "loss": 0.0, + "step": 53980 + }, + { + "epoch": 13.03, + "learning_rate": 1.7645489270868444e-05, + "loss": 0.0001, + "step": 53990 + }, + { + "epoch": 13.03, + "learning_rate": 1.763846758791147e-05, + "loss": 0.0004, + "step": 54000 + }, + { + "epoch": 13.03, + "learning_rate": 1.76314459049545e-05, + "loss": 0.0126, + "step": 54010 + }, + { + "epoch": 13.03, + "learning_rate": 1.762442422199753e-05, + "loss": 0.0003, + "step": 54020 + }, + { + "epoch": 13.03, + "learning_rate": 1.761740253904056e-05, + "loss": 0.0, + "step": 54030 + }, + { + "epoch": 13.03, + "learning_rate": 1.7610380856083584e-05, + "loss": 0.0019, + "step": 54040 + }, + { + "epoch": 13.03, + "learning_rate": 1.7603359173126615e-05, + "loss": 0.0001, + "step": 54050 + }, + { + "epoch": 13.03, + "learning_rate": 1.7596337490169646e-05, + "loss": 0.0008, + "step": 54060 + }, + { + "epoch": 13.03, + "learning_rate": 1.7589315807212674e-05, + "loss": 0.2618, + "step": 54070 + }, + { + "epoch": 13.03, + "learning_rate": 1.75822941242557e-05, + "loss": 0.1005, + "step": 54080 + }, + { + "epoch": 13.03, + "learning_rate": 1.7575272441298732e-05, + "loss": 0.0001, + "step": 54090 + }, + { + "epoch": 13.03, + "learning_rate": 1.756825075834176e-05, + "loss": 0.001, + "step": 54100 + }, + { + "epoch": 13.03, + "learning_rate": 1.7561229075384787e-05, + "loss": 0.0001, + "step": 54110 + }, + { + "epoch": 13.03, + "learning_rate": 1.7554207392427818e-05, + "loss": 0.017, + "step": 54120 + }, + { + "epoch": 13.03, + "learning_rate": 1.754718570947085e-05, + "loss": 0.0045, + "step": 54130 + }, + { + "epoch": 13.03, + "learning_rate": 1.7540164026513876e-05, + "loss": 0.0478, + "step": 54140 + }, + { + "epoch": 13.03, + "learning_rate": 1.7533142343556904e-05, + "loss": 0.0001, + "step": 54150 + }, + { + "epoch": 13.03, + "learning_rate": 1.7526120660599935e-05, + "loss": 0.02, + "step": 54160 + }, + { + "epoch": 13.03, + "learning_rate": 1.7519098977642962e-05, + "loss": 0.0169, + "step": 54170 + }, + { + "epoch": 13.03, + "learning_rate": 1.751207729468599e-05, + "loss": 0.0159, + "step": 54180 + }, + { + "epoch": 13.03, + "learning_rate": 1.750505561172902e-05, + "loss": 0.0002, + "step": 54190 + }, + { + "epoch": 13.04, + "learning_rate": 1.7498033928772048e-05, + "loss": 0.0001, + "step": 54200 + }, + { + "epoch": 13.04, + "learning_rate": 1.749101224581508e-05, + "loss": 0.0064, + "step": 54210 + }, + { + "epoch": 13.04, + "learning_rate": 1.7483990562858106e-05, + "loss": 0.2201, + "step": 54220 + }, + { + "epoch": 13.04, + "learning_rate": 1.7476968879901134e-05, + "loss": 0.3602, + "step": 54230 + }, + { + "epoch": 13.04, + "learning_rate": 1.7469947196944165e-05, + "loss": 0.0, + "step": 54240 + }, + { + "epoch": 13.04, + "learning_rate": 1.7462925513987196e-05, + "loss": 0.0, + "step": 54250 + }, + { + "epoch": 13.04, + "learning_rate": 1.745590383103022e-05, + "loss": 0.0, + "step": 54260 + }, + { + "epoch": 13.04, + "learning_rate": 1.744888214807325e-05, + "loss": 0.0, + "step": 54270 + }, + { + "epoch": 13.04, + "learning_rate": 1.744186046511628e-05, + "loss": 0.1672, + "step": 54280 + }, + { + "epoch": 13.04, + "learning_rate": 1.743483878215931e-05, + "loss": 0.0001, + "step": 54290 + }, + { + "epoch": 13.04, + "learning_rate": 1.7427817099202336e-05, + "loss": 0.001, + "step": 54300 + }, + { + "epoch": 13.04, + "learning_rate": 1.7420795416245367e-05, + "loss": 0.0004, + "step": 54310 + }, + { + "epoch": 13.04, + "learning_rate": 1.7413773733288398e-05, + "loss": 0.01, + "step": 54320 + }, + { + "epoch": 13.04, + "learning_rate": 1.7406752050331422e-05, + "loss": 0.0009, + "step": 54330 + }, + { + "epoch": 13.04, + "learning_rate": 1.7399730367374453e-05, + "loss": 0.0, + "step": 54340 + }, + { + "epoch": 13.04, + "learning_rate": 1.7392708684417484e-05, + "loss": 0.0004, + "step": 54350 + }, + { + "epoch": 13.04, + "learning_rate": 1.738568700146051e-05, + "loss": 0.0, + "step": 54360 + }, + { + "epoch": 13.04, + "learning_rate": 1.737866531850354e-05, + "loss": 0.0, + "step": 54370 + }, + { + "epoch": 13.04, + "learning_rate": 1.737164363554657e-05, + "loss": 0.0001, + "step": 54380 + }, + { + "epoch": 13.04, + "learning_rate": 1.7364621952589597e-05, + "loss": 0.0, + "step": 54390 + }, + { + "epoch": 13.04, + "learning_rate": 1.7357600269632625e-05, + "loss": 0.0, + "step": 54400 + }, + { + "epoch": 13.04, + "learning_rate": 1.7350578586675656e-05, + "loss": 0.0, + "step": 54410 + }, + { + "epoch": 13.04, + "learning_rate": 1.7343556903718683e-05, + "loss": 0.0008, + "step": 54420 + }, + { + "epoch": 13.04, + "learning_rate": 1.7336535220761714e-05, + "loss": 0.0001, + "step": 54430 + }, + { + "epoch": 13.04, + "learning_rate": 1.732951353780474e-05, + "loss": 0.0003, + "step": 54440 + }, + { + "epoch": 13.04, + "learning_rate": 1.732249185484777e-05, + "loss": 0.0003, + "step": 54450 + }, + { + "epoch": 13.04, + "learning_rate": 1.73154701718908e-05, + "loss": 0.0001, + "step": 54460 + }, + { + "epoch": 13.04, + "learning_rate": 1.7308448488933827e-05, + "loss": 0.0338, + "step": 54470 + }, + { + "epoch": 13.04, + "learning_rate": 1.730142680597686e-05, + "loss": 0.0001, + "step": 54480 + }, + { + "epoch": 13.04, + "learning_rate": 1.7294405123019886e-05, + "loss": 0.0, + "step": 54490 + }, + { + "epoch": 13.04, + "learning_rate": 1.7287383440062917e-05, + "loss": 0.0016, + "step": 54500 + }, + { + "epoch": 13.04, + "learning_rate": 1.7280361757105944e-05, + "loss": 0.0003, + "step": 54510 + }, + { + "epoch": 13.04, + "learning_rate": 1.727334007414897e-05, + "loss": 0.3979, + "step": 54520 + }, + { + "epoch": 13.04, + "learning_rate": 1.7266318391192003e-05, + "loss": 0.0, + "step": 54530 + }, + { + "epoch": 13.04, + "learning_rate": 1.725929670823503e-05, + "loss": 0.0, + "step": 54540 + }, + { + "epoch": 13.04, + "learning_rate": 1.7252275025278058e-05, + "loss": 0.3041, + "step": 54550 + }, + { + "epoch": 13.04, + "learning_rate": 1.724525334232109e-05, + "loss": 0.0004, + "step": 54560 + }, + { + "epoch": 13.04, + "learning_rate": 1.723823165936412e-05, + "loss": 0.0868, + "step": 54570 + }, + { + "epoch": 13.04, + "learning_rate": 1.7231209976407143e-05, + "loss": 0.1211, + "step": 54580 + }, + { + "epoch": 13.04, + "learning_rate": 1.7224188293450174e-05, + "loss": 0.0, + "step": 54590 + }, + { + "epoch": 13.04, + "learning_rate": 1.7217166610493205e-05, + "loss": 0.334, + "step": 54600 + }, + { + "epoch": 13.04, + "learning_rate": 1.7210144927536233e-05, + "loss": 0.0001, + "step": 54610 + }, + { + "epoch": 13.04, + "learning_rate": 1.720312324457926e-05, + "loss": 0.1139, + "step": 54620 + }, + { + "epoch": 13.04, + "learning_rate": 1.719610156162229e-05, + "loss": 0.0106, + "step": 54630 + }, + { + "epoch": 13.04, + "learning_rate": 1.7189079878665322e-05, + "loss": 0.0, + "step": 54640 + }, + { + "epoch": 13.04, + "learning_rate": 1.7182058195708346e-05, + "loss": 0.0, + "step": 54650 + }, + { + "epoch": 13.04, + "learning_rate": 1.7175036512751377e-05, + "loss": 0.0, + "step": 54660 + }, + { + "epoch": 13.04, + "learning_rate": 1.7168014829794408e-05, + "loss": 0.0, + "step": 54670 + }, + { + "epoch": 13.04, + "learning_rate": 1.7160993146837435e-05, + "loss": 0.0, + "step": 54680 + }, + { + "epoch": 13.04, + "learning_rate": 1.7153971463880463e-05, + "loss": 0.0, + "step": 54690 + }, + { + "epoch": 13.04, + "learning_rate": 1.7146949780923494e-05, + "loss": 0.0687, + "step": 54700 + }, + { + "epoch": 13.04, + "learning_rate": 1.713992809796652e-05, + "loss": 0.0014, + "step": 54710 + }, + { + "epoch": 13.04, + "learning_rate": 1.7132906415009552e-05, + "loss": 0.0, + "step": 54720 + }, + { + "epoch": 13.04, + "learning_rate": 1.712588473205258e-05, + "loss": 0.0001, + "step": 54730 + }, + { + "epoch": 13.04, + "learning_rate": 1.7118863049095607e-05, + "loss": 0.0001, + "step": 54740 + }, + { + "epoch": 13.04, + "learning_rate": 1.7111841366138638e-05, + "loss": 0.0806, + "step": 54750 + }, + { + "epoch": 13.04, + "learning_rate": 1.7104819683181665e-05, + "loss": 0.0001, + "step": 54760 + }, + { + "epoch": 13.04, + "learning_rate": 1.7097798000224693e-05, + "loss": 0.0282, + "step": 54770 + }, + { + "epoch": 13.04, + "learning_rate": 1.7090776317267724e-05, + "loss": 0.6291, + "step": 54780 + }, + { + "epoch": 13.04, + "learning_rate": 1.7083754634310755e-05, + "loss": 0.0004, + "step": 54790 + }, + { + "epoch": 13.04, + "learning_rate": 1.707673295135378e-05, + "loss": 0.0005, + "step": 54800 + }, + { + "epoch": 13.04, + "learning_rate": 1.706971126839681e-05, + "loss": 0.5125, + "step": 54810 + }, + { + "epoch": 13.04, + "learning_rate": 1.706268958543984e-05, + "loss": 0.0, + "step": 54820 + }, + { + "epoch": 13.04, + "learning_rate": 1.7055667902482868e-05, + "loss": 0.4034, + "step": 54830 + }, + { + "epoch": 13.04, + "learning_rate": 1.7048646219525895e-05, + "loss": 0.0, + "step": 54840 + }, + { + "epoch": 13.04, + "learning_rate": 1.7041624536568926e-05, + "loss": 0.0487, + "step": 54850 + }, + { + "epoch": 13.04, + "learning_rate": 1.7034602853611957e-05, + "loss": 0.0, + "step": 54860 + }, + { + "epoch": 13.04, + "learning_rate": 1.702758117065498e-05, + "loss": 0.1891, + "step": 54870 + }, + { + "epoch": 13.04, + "learning_rate": 1.7020559487698012e-05, + "loss": 0.0035, + "step": 54880 + }, + { + "epoch": 13.04, + "learning_rate": 1.7013537804741043e-05, + "loss": 0.0001, + "step": 54890 + }, + { + "epoch": 13.04, + "learning_rate": 1.700651612178407e-05, + "loss": 0.2243, + "step": 54900 + }, + { + "epoch": 13.04, + "learning_rate": 1.6999494438827098e-05, + "loss": 0.1194, + "step": 54910 + }, + { + "epoch": 13.04, + "learning_rate": 1.699247275587013e-05, + "loss": 0.4969, + "step": 54920 + }, + { + "epoch": 13.04, + "learning_rate": 1.6985451072913156e-05, + "loss": 0.0003, + "step": 54930 + }, + { + "epoch": 13.04, + "learning_rate": 1.6978429389956184e-05, + "loss": 0.0041, + "step": 54940 + }, + { + "epoch": 13.04, + "learning_rate": 1.6971407706999215e-05, + "loss": 0.0119, + "step": 54950 + }, + { + "epoch": 13.04, + "learning_rate": 1.6964386024042242e-05, + "loss": 0.14, + "step": 54960 + }, + { + "epoch": 13.04, + "learning_rate": 1.6957364341085273e-05, + "loss": 0.0001, + "step": 54970 + }, + { + "epoch": 13.04, + "learning_rate": 1.69503426581283e-05, + "loss": 0.0, + "step": 54980 + }, + { + "epoch": 13.05, + "learning_rate": 1.694332097517133e-05, + "loss": 0.0, + "step": 54990 + }, + { + "epoch": 13.05, + "learning_rate": 1.693629929221436e-05, + "loss": 0.0, + "step": 55000 + }, + { + "epoch": 13.05, + "learning_rate": 1.6929277609257386e-05, + "loss": 0.0, + "step": 55010 + }, + { + "epoch": 13.05, + "learning_rate": 1.6922255926300417e-05, + "loss": 0.0, + "step": 55020 + }, + { + "epoch": 13.05, + "learning_rate": 1.6915234243343445e-05, + "loss": 0.0, + "step": 55030 + }, + { + "epoch": 13.05, + "learning_rate": 1.6908212560386476e-05, + "loss": 0.3837, + "step": 55040 + }, + { + "epoch": 13.05, + "learning_rate": 1.6901190877429503e-05, + "loss": 0.0007, + "step": 55050 + }, + { + "epoch": 13.05, + "learning_rate": 1.689416919447253e-05, + "loss": 0.0074, + "step": 55060 + }, + { + "epoch": 13.05, + "learning_rate": 1.688714751151556e-05, + "loss": 0.001, + "step": 55070 + }, + { + "epoch": 13.05, + "learning_rate": 1.688012582855859e-05, + "loss": 0.0237, + "step": 55080 + }, + { + "epoch": 13.05, + "learning_rate": 1.6873104145601617e-05, + "loss": 0.0356, + "step": 55090 + }, + { + "epoch": 13.05, + "learning_rate": 1.6866082462644647e-05, + "loss": 0.002, + "step": 55100 + }, + { + "epoch": 13.05, + "learning_rate": 1.685906077968768e-05, + "loss": 0.0, + "step": 55110 + }, + { + "epoch": 13.05, + "learning_rate": 1.6852039096730702e-05, + "loss": 0.0, + "step": 55120 + }, + { + "epoch": 13.05, + "learning_rate": 1.6845017413773733e-05, + "loss": 0.003, + "step": 55130 + }, + { + "epoch": 13.05, + "learning_rate": 1.6837995730816764e-05, + "loss": 0.0546, + "step": 55140 + }, + { + "epoch": 13.05, + "learning_rate": 1.683097404785979e-05, + "loss": 0.002, + "step": 55150 + }, + { + "epoch": 13.05, + "learning_rate": 1.682395236490282e-05, + "loss": 0.3438, + "step": 55160 + }, + { + "epoch": 13.05, + "learning_rate": 1.681693068194585e-05, + "loss": 0.0, + "step": 55170 + }, + { + "epoch": 13.05, + "learning_rate": 1.680990899898888e-05, + "loss": 0.0, + "step": 55180 + }, + { + "epoch": 13.05, + "learning_rate": 1.6802887316031905e-05, + "loss": 0.0001, + "step": 55190 + }, + { + "epoch": 13.05, + "learning_rate": 1.6795865633074936e-05, + "loss": 0.2728, + "step": 55200 + }, + { + "epoch": 13.05, + "learning_rate": 1.6788843950117967e-05, + "loss": 0.5667, + "step": 55210 + }, + { + "epoch": 13.05, + "learning_rate": 1.6781822267160994e-05, + "loss": 0.0, + "step": 55220 + }, + { + "epoch": 13.05, + "learning_rate": 1.6774800584204022e-05, + "loss": 0.0526, + "step": 55230 + }, + { + "epoch": 13.05, + "learning_rate": 1.6767778901247053e-05, + "loss": 0.106, + "step": 55240 + }, + { + "epoch": 13.05, + "learning_rate": 1.676075721829008e-05, + "loss": 0.0, + "step": 55250 + }, + { + "epoch": 13.05, + "learning_rate": 1.675373553533311e-05, + "loss": 0.0013, + "step": 55260 + }, + { + "epoch": 13.05, + "learning_rate": 1.674671385237614e-05, + "loss": 0.2648, + "step": 55270 + }, + { + "epoch": 13.05, + "learning_rate": 1.6739692169419166e-05, + "loss": 0.0159, + "step": 55280 + }, + { + "epoch": 13.05, + "learning_rate": 1.6732670486462197e-05, + "loss": 0.0068, + "step": 55290 + }, + { + "epoch": 13.05, + "learning_rate": 1.6725648803505224e-05, + "loss": 0.0009, + "step": 55300 + }, + { + "epoch": 13.05, + "learning_rate": 1.6718627120548252e-05, + "loss": 0.0023, + "step": 55310 + }, + { + "epoch": 13.05, + "learning_rate": 1.6711605437591283e-05, + "loss": 0.1607, + "step": 55320 + }, + { + "epoch": 13.05, + "learning_rate": 1.6704583754634314e-05, + "loss": 0.004, + "step": 55330 + }, + { + "epoch": 13.05, + "learning_rate": 1.669756207167734e-05, + "loss": 0.0001, + "step": 55340 + }, + { + "epoch": 13.05, + "learning_rate": 1.669054038872037e-05, + "loss": 0.0, + "step": 55350 + }, + { + "epoch": 13.05, + "learning_rate": 1.66835187057634e-05, + "loss": 0.0001, + "step": 55360 + }, + { + "epoch": 13.05, + "learning_rate": 1.6676497022806427e-05, + "loss": 0.0088, + "step": 55370 + }, + { + "epoch": 13.05, + "learning_rate": 1.6669475339849454e-05, + "loss": 0.0005, + "step": 55380 + }, + { + "epoch": 13.05, + "eval_accuracy": 0.7384853567281641, + "eval_loss": 1.819637417793274, + "eval_runtime": 1242.5471, + "eval_samples_per_second": 6.238, + "eval_steps_per_second": 3.119, + "step": 55384 + }, + { + "epoch": 14.0, + "learning_rate": 1.6662453656892485e-05, + "loss": 0.0244, + "step": 55390 + }, + { + "epoch": 14.0, + "learning_rate": 1.6655431973935516e-05, + "loss": 0.0, + "step": 55400 + }, + { + "epoch": 14.0, + "learning_rate": 1.664841029097854e-05, + "loss": 0.0, + "step": 55410 + }, + { + "epoch": 14.0, + "learning_rate": 1.664138860802157e-05, + "loss": 0.0001, + "step": 55420 + }, + { + "epoch": 14.0, + "learning_rate": 1.6634366925064602e-05, + "loss": 0.0, + "step": 55430 + }, + { + "epoch": 14.0, + "learning_rate": 1.662734524210763e-05, + "loss": 0.0002, + "step": 55440 + }, + { + "epoch": 14.0, + "learning_rate": 1.6620323559150657e-05, + "loss": 0.0, + "step": 55450 + }, + { + "epoch": 14.0, + "learning_rate": 1.6613301876193688e-05, + "loss": 0.0, + "step": 55460 + }, + { + "epoch": 14.0, + "learning_rate": 1.6606280193236715e-05, + "loss": 0.0, + "step": 55470 + }, + { + "epoch": 14.0, + "learning_rate": 1.6599258510279743e-05, + "loss": 0.0, + "step": 55480 + }, + { + "epoch": 14.0, + "learning_rate": 1.6592236827322774e-05, + "loss": 0.0, + "step": 55490 + }, + { + "epoch": 14.0, + "learning_rate": 1.65852151443658e-05, + "loss": 0.0, + "step": 55500 + }, + { + "epoch": 14.0, + "learning_rate": 1.6578193461408832e-05, + "loss": 0.0, + "step": 55510 + }, + { + "epoch": 14.0, + "learning_rate": 1.657117177845186e-05, + "loss": 0.0058, + "step": 55520 + }, + { + "epoch": 14.0, + "learning_rate": 1.656415009549489e-05, + "loss": 0.0, + "step": 55530 + }, + { + "epoch": 14.0, + "learning_rate": 1.6557128412537918e-05, + "loss": 0.0, + "step": 55540 + }, + { + "epoch": 14.0, + "learning_rate": 1.6550106729580945e-05, + "loss": 0.0524, + "step": 55550 + }, + { + "epoch": 14.0, + "learning_rate": 1.6543085046623976e-05, + "loss": 0.0, + "step": 55560 + }, + { + "epoch": 14.0, + "learning_rate": 1.6536063363667004e-05, + "loss": 0.0289, + "step": 55570 + }, + { + "epoch": 14.0, + "learning_rate": 1.6529041680710035e-05, + "loss": 0.0, + "step": 55580 + }, + { + "epoch": 14.0, + "learning_rate": 1.6522019997753062e-05, + "loss": 0.0, + "step": 55590 + }, + { + "epoch": 14.0, + "learning_rate": 1.651499831479609e-05, + "loss": 0.0009, + "step": 55600 + }, + { + "epoch": 14.0, + "learning_rate": 1.650797663183912e-05, + "loss": 0.0, + "step": 55610 + }, + { + "epoch": 14.0, + "learning_rate": 1.6500954948882148e-05, + "loss": 0.0007, + "step": 55620 + }, + { + "epoch": 14.0, + "learning_rate": 1.6493933265925176e-05, + "loss": 0.0, + "step": 55630 + }, + { + "epoch": 14.0, + "learning_rate": 1.6486911582968206e-05, + "loss": 0.0084, + "step": 55640 + }, + { + "epoch": 14.0, + "learning_rate": 1.6479889900011237e-05, + "loss": 0.0001, + "step": 55650 + }, + { + "epoch": 14.0, + "learning_rate": 1.647286821705426e-05, + "loss": 0.0, + "step": 55660 + }, + { + "epoch": 14.0, + "learning_rate": 1.6465846534097292e-05, + "loss": 0.0005, + "step": 55670 + }, + { + "epoch": 14.0, + "learning_rate": 1.6458824851140323e-05, + "loss": 0.0, + "step": 55680 + }, + { + "epoch": 14.0, + "learning_rate": 1.645180316818335e-05, + "loss": 0.0678, + "step": 55690 + }, + { + "epoch": 14.0, + "learning_rate": 1.6444781485226378e-05, + "loss": 0.0001, + "step": 55700 + }, + { + "epoch": 14.0, + "learning_rate": 1.643775980226941e-05, + "loss": 0.0, + "step": 55710 + }, + { + "epoch": 14.0, + "learning_rate": 1.643073811931244e-05, + "loss": 0.0, + "step": 55720 + }, + { + "epoch": 14.0, + "learning_rate": 1.6423716436355467e-05, + "loss": 0.0, + "step": 55730 + }, + { + "epoch": 14.0, + "learning_rate": 1.6416694753398495e-05, + "loss": 0.2497, + "step": 55740 + }, + { + "epoch": 14.0, + "learning_rate": 1.6409673070441526e-05, + "loss": 0.0, + "step": 55750 + }, + { + "epoch": 14.0, + "learning_rate": 1.6402651387484553e-05, + "loss": 0.011, + "step": 55760 + }, + { + "epoch": 14.0, + "learning_rate": 1.639562970452758e-05, + "loss": 0.0001, + "step": 55770 + }, + { + "epoch": 14.01, + "learning_rate": 1.638860802157061e-05, + "loss": 0.0, + "step": 55780 + }, + { + "epoch": 14.01, + "learning_rate": 1.638158633861364e-05, + "loss": 0.0, + "step": 55790 + }, + { + "epoch": 14.01, + "learning_rate": 1.637456465565667e-05, + "loss": 0.0001, + "step": 55800 + }, + { + "epoch": 14.01, + "learning_rate": 1.6367542972699698e-05, + "loss": 0.0, + "step": 55810 + }, + { + "epoch": 14.01, + "learning_rate": 1.6360521289742725e-05, + "loss": 0.0001, + "step": 55820 + }, + { + "epoch": 14.01, + "learning_rate": 1.6353499606785756e-05, + "loss": 0.0, + "step": 55830 + }, + { + "epoch": 14.01, + "learning_rate": 1.6346477923828783e-05, + "loss": 0.0001, + "step": 55840 + }, + { + "epoch": 14.01, + "learning_rate": 1.633945624087181e-05, + "loss": 0.0383, + "step": 55850 + }, + { + "epoch": 14.01, + "learning_rate": 1.6332434557914842e-05, + "loss": 0.0, + "step": 55860 + }, + { + "epoch": 14.01, + "learning_rate": 1.6325412874957873e-05, + "loss": 0.0, + "step": 55870 + }, + { + "epoch": 14.01, + "learning_rate": 1.63183911920009e-05, + "loss": 0.0, + "step": 55880 + }, + { + "epoch": 14.01, + "learning_rate": 1.6311369509043928e-05, + "loss": 0.0, + "step": 55890 + }, + { + "epoch": 14.01, + "learning_rate": 1.630434782608696e-05, + "loss": 0.0, + "step": 55900 + }, + { + "epoch": 14.01, + "learning_rate": 1.6297326143129986e-05, + "loss": 0.0, + "step": 55910 + }, + { + "epoch": 14.01, + "learning_rate": 1.6290304460173013e-05, + "loss": 0.0001, + "step": 55920 + }, + { + "epoch": 14.01, + "learning_rate": 1.6283282777216044e-05, + "loss": 0.0, + "step": 55930 + }, + { + "epoch": 14.01, + "learning_rate": 1.6276261094259075e-05, + "loss": 0.0001, + "step": 55940 + }, + { + "epoch": 14.01, + "learning_rate": 1.62692394113021e-05, + "loss": 0.0, + "step": 55950 + }, + { + "epoch": 14.01, + "learning_rate": 1.626221772834513e-05, + "loss": 0.0, + "step": 55960 + }, + { + "epoch": 14.01, + "learning_rate": 1.625519604538816e-05, + "loss": 0.0, + "step": 55970 + }, + { + "epoch": 14.01, + "learning_rate": 1.624817436243119e-05, + "loss": 0.0003, + "step": 55980 + }, + { + "epoch": 14.01, + "learning_rate": 1.6241152679474216e-05, + "loss": 0.0, + "step": 55990 + }, + { + "epoch": 14.01, + "learning_rate": 1.6234130996517247e-05, + "loss": 0.1192, + "step": 56000 + }, + { + "epoch": 14.01, + "learning_rate": 1.6227109313560274e-05, + "loss": 0.0022, + "step": 56010 + }, + { + "epoch": 14.01, + "learning_rate": 1.6220087630603302e-05, + "loss": 0.0, + "step": 56020 + }, + { + "epoch": 14.01, + "learning_rate": 1.6213065947646333e-05, + "loss": 0.0, + "step": 56030 + }, + { + "epoch": 14.01, + "learning_rate": 1.6206044264689364e-05, + "loss": 0.3319, + "step": 56040 + }, + { + "epoch": 14.01, + "learning_rate": 1.619902258173239e-05, + "loss": 0.0066, + "step": 56050 + }, + { + "epoch": 14.01, + "learning_rate": 1.619200089877542e-05, + "loss": 0.0, + "step": 56060 + }, + { + "epoch": 14.01, + "learning_rate": 1.618497921581845e-05, + "loss": 0.0001, + "step": 56070 + }, + { + "epoch": 14.01, + "learning_rate": 1.6177957532861477e-05, + "loss": 0.2876, + "step": 56080 + }, + { + "epoch": 14.01, + "learning_rate": 1.6170935849904505e-05, + "loss": 0.0001, + "step": 56090 + }, + { + "epoch": 14.01, + "learning_rate": 1.6163914166947535e-05, + "loss": 0.0, + "step": 56100 + }, + { + "epoch": 14.01, + "learning_rate": 1.6156892483990563e-05, + "loss": 0.0002, + "step": 56110 + }, + { + "epoch": 14.01, + "learning_rate": 1.6149870801033594e-05, + "loss": 0.0728, + "step": 56120 + }, + { + "epoch": 14.01, + "learning_rate": 1.614284911807662e-05, + "loss": 0.0002, + "step": 56130 + }, + { + "epoch": 14.01, + "learning_rate": 1.613582743511965e-05, + "loss": 0.0407, + "step": 56140 + }, + { + "epoch": 14.01, + "learning_rate": 1.612880575216268e-05, + "loss": 0.0, + "step": 56150 + }, + { + "epoch": 14.01, + "learning_rate": 1.6121784069205707e-05, + "loss": 0.0, + "step": 56160 + }, + { + "epoch": 14.01, + "learning_rate": 1.6114762386248735e-05, + "loss": 0.0011, + "step": 56170 + }, + { + "epoch": 14.01, + "learning_rate": 1.6107740703291765e-05, + "loss": 0.0003, + "step": 56180 + }, + { + "epoch": 14.01, + "learning_rate": 1.6100719020334796e-05, + "loss": 0.0, + "step": 56190 + }, + { + "epoch": 14.01, + "learning_rate": 1.609369733737782e-05, + "loss": 0.0006, + "step": 56200 + }, + { + "epoch": 14.01, + "learning_rate": 1.608667565442085e-05, + "loss": 0.0, + "step": 56210 + }, + { + "epoch": 14.01, + "learning_rate": 1.6079653971463882e-05, + "loss": 0.2914, + "step": 56220 + }, + { + "epoch": 14.01, + "learning_rate": 1.607263228850691e-05, + "loss": 0.003, + "step": 56230 + }, + { + "epoch": 14.01, + "learning_rate": 1.6065610605549937e-05, + "loss": 0.2634, + "step": 56240 + }, + { + "epoch": 14.01, + "learning_rate": 1.6058588922592968e-05, + "loss": 0.0001, + "step": 56250 + }, + { + "epoch": 14.01, + "learning_rate": 1.6051567239636e-05, + "loss": 0.0016, + "step": 56260 + }, + { + "epoch": 14.01, + "learning_rate": 1.6044545556679026e-05, + "loss": 0.1719, + "step": 56270 + }, + { + "epoch": 14.01, + "learning_rate": 1.6037523873722054e-05, + "loss": 0.0001, + "step": 56280 + }, + { + "epoch": 14.01, + "learning_rate": 1.6030502190765085e-05, + "loss": 0.0028, + "step": 56290 + }, + { + "epoch": 14.01, + "learning_rate": 1.6023480507808112e-05, + "loss": 0.0, + "step": 56300 + }, + { + "epoch": 14.01, + "learning_rate": 1.601645882485114e-05, + "loss": 0.0001, + "step": 56310 + }, + { + "epoch": 14.01, + "learning_rate": 1.600943714189417e-05, + "loss": 0.1435, + "step": 56320 + }, + { + "epoch": 14.01, + "learning_rate": 1.6002415458937198e-05, + "loss": 0.0001, + "step": 56330 + }, + { + "epoch": 14.01, + "learning_rate": 1.599539377598023e-05, + "loss": 0.0, + "step": 56340 + }, + { + "epoch": 14.01, + "learning_rate": 1.5988372093023257e-05, + "loss": 0.0, + "step": 56350 + }, + { + "epoch": 14.01, + "learning_rate": 1.5981350410066284e-05, + "loss": 0.0039, + "step": 56360 + }, + { + "epoch": 14.01, + "learning_rate": 1.5974328727109315e-05, + "loss": 0.0, + "step": 56370 + }, + { + "epoch": 14.01, + "learning_rate": 1.5967307044152342e-05, + "loss": 0.0, + "step": 56380 + }, + { + "epoch": 14.01, + "learning_rate": 1.5960285361195373e-05, + "loss": 0.0, + "step": 56390 + }, + { + "epoch": 14.01, + "learning_rate": 1.59532636782384e-05, + "loss": 0.0, + "step": 56400 + }, + { + "epoch": 14.01, + "learning_rate": 1.594624199528143e-05, + "loss": 0.008, + "step": 56410 + }, + { + "epoch": 14.01, + "learning_rate": 1.593922031232446e-05, + "loss": 0.0193, + "step": 56420 + }, + { + "epoch": 14.01, + "learning_rate": 1.5932198629367487e-05, + "loss": 0.5883, + "step": 56430 + }, + { + "epoch": 14.01, + "learning_rate": 1.5925176946410518e-05, + "loss": 0.0002, + "step": 56440 + }, + { + "epoch": 14.01, + "learning_rate": 1.5918155263453545e-05, + "loss": 0.0001, + "step": 56450 + }, + { + "epoch": 14.01, + "learning_rate": 1.5911133580496572e-05, + "loss": 0.0028, + "step": 56460 + }, + { + "epoch": 14.01, + "learning_rate": 1.5904111897539603e-05, + "loss": 0.0001, + "step": 56470 + }, + { + "epoch": 14.01, + "learning_rate": 1.5897090214582634e-05, + "loss": 0.0, + "step": 56480 + }, + { + "epoch": 14.01, + "learning_rate": 1.589006853162566e-05, + "loss": 0.0001, + "step": 56490 + }, + { + "epoch": 14.01, + "learning_rate": 1.588304684866869e-05, + "loss": 0.0003, + "step": 56500 + }, + { + "epoch": 14.01, + "learning_rate": 1.587602516571172e-05, + "loss": 0.0001, + "step": 56510 + }, + { + "epoch": 14.01, + "learning_rate": 1.5869003482754748e-05, + "loss": 0.0, + "step": 56520 + }, + { + "epoch": 14.01, + "learning_rate": 1.5861981799797775e-05, + "loss": 0.0, + "step": 56530 + }, + { + "epoch": 14.01, + "learning_rate": 1.5854960116840806e-05, + "loss": 0.0, + "step": 56540 + }, + { + "epoch": 14.01, + "learning_rate": 1.5847938433883833e-05, + "loss": 0.018, + "step": 56550 + }, + { + "epoch": 14.01, + "learning_rate": 1.584091675092686e-05, + "loss": 0.0001, + "step": 56560 + }, + { + "epoch": 14.01, + "learning_rate": 1.5833895067969892e-05, + "loss": 0.0854, + "step": 56570 + }, + { + "epoch": 14.02, + "learning_rate": 1.5826873385012923e-05, + "loss": 0.0645, + "step": 56580 + }, + { + "epoch": 14.02, + "learning_rate": 1.581985170205595e-05, + "loss": 0.0002, + "step": 56590 + }, + { + "epoch": 14.02, + "learning_rate": 1.5812830019098978e-05, + "loss": 0.001, + "step": 56600 + }, + { + "epoch": 14.02, + "learning_rate": 1.580580833614201e-05, + "loss": 0.0, + "step": 56610 + }, + { + "epoch": 14.02, + "learning_rate": 1.5798786653185036e-05, + "loss": 0.0046, + "step": 56620 + }, + { + "epoch": 14.02, + "learning_rate": 1.5791764970228064e-05, + "loss": 0.0051, + "step": 56630 + }, + { + "epoch": 14.02, + "learning_rate": 1.5784743287271094e-05, + "loss": 0.0, + "step": 56640 + }, + { + "epoch": 14.02, + "learning_rate": 1.5777721604314122e-05, + "loss": 0.0012, + "step": 56650 + }, + { + "epoch": 14.02, + "learning_rate": 1.5770699921357153e-05, + "loss": 0.0021, + "step": 56660 + }, + { + "epoch": 14.02, + "learning_rate": 1.576367823840018e-05, + "loss": 0.0001, + "step": 56670 + }, + { + "epoch": 14.02, + "learning_rate": 1.5756656555443208e-05, + "loss": 0.0, + "step": 56680 + }, + { + "epoch": 14.02, + "learning_rate": 1.574963487248624e-05, + "loss": 0.1017, + "step": 56690 + }, + { + "epoch": 14.02, + "learning_rate": 1.5742613189529266e-05, + "loss": 0.0, + "step": 56700 + }, + { + "epoch": 14.02, + "learning_rate": 1.5735591506572294e-05, + "loss": 0.0092, + "step": 56710 + }, + { + "epoch": 14.02, + "learning_rate": 1.5728569823615325e-05, + "loss": 0.0, + "step": 56720 + }, + { + "epoch": 14.02, + "learning_rate": 1.5721548140658355e-05, + "loss": 0.0001, + "step": 56730 + }, + { + "epoch": 14.02, + "learning_rate": 1.5714526457701383e-05, + "loss": 0.0, + "step": 56740 + }, + { + "epoch": 14.02, + "learning_rate": 1.570750477474441e-05, + "loss": 0.0007, + "step": 56750 + }, + { + "epoch": 14.02, + "learning_rate": 1.570048309178744e-05, + "loss": 0.0001, + "step": 56760 + }, + { + "epoch": 14.02, + "learning_rate": 1.5693461408830472e-05, + "loss": 0.0, + "step": 56770 + }, + { + "epoch": 14.02, + "learning_rate": 1.5686439725873496e-05, + "loss": 0.0015, + "step": 56780 + }, + { + "epoch": 14.02, + "learning_rate": 1.5679418042916527e-05, + "loss": 0.0001, + "step": 56790 + }, + { + "epoch": 14.02, + "learning_rate": 1.5672396359959558e-05, + "loss": 0.0, + "step": 56800 + }, + { + "epoch": 14.02, + "learning_rate": 1.5665374677002585e-05, + "loss": 0.0, + "step": 56810 + }, + { + "epoch": 14.02, + "learning_rate": 1.5658352994045613e-05, + "loss": 0.0, + "step": 56820 + }, + { + "epoch": 14.02, + "learning_rate": 1.5651331311088644e-05, + "loss": 0.0, + "step": 56830 + }, + { + "epoch": 14.02, + "learning_rate": 1.564430962813167e-05, + "loss": 0.0108, + "step": 56840 + }, + { + "epoch": 14.02, + "learning_rate": 1.56372879451747e-05, + "loss": 0.0, + "step": 56850 + }, + { + "epoch": 14.02, + "learning_rate": 1.563026626221773e-05, + "loss": 0.0028, + "step": 56860 + }, + { + "epoch": 14.02, + "learning_rate": 1.5623244579260757e-05, + "loss": 0.226, + "step": 56870 + }, + { + "epoch": 14.02, + "learning_rate": 1.5616222896303788e-05, + "loss": 0.0001, + "step": 56880 + }, + { + "epoch": 14.02, + "learning_rate": 1.5609201213346816e-05, + "loss": 0.0, + "step": 56890 + }, + { + "epoch": 14.02, + "learning_rate": 1.5602179530389843e-05, + "loss": 0.0006, + "step": 56900 + }, + { + "epoch": 14.02, + "learning_rate": 1.5595157847432874e-05, + "loss": 0.0946, + "step": 56910 + }, + { + "epoch": 14.02, + "learning_rate": 1.55881361644759e-05, + "loss": 0.0, + "step": 56920 + }, + { + "epoch": 14.02, + "learning_rate": 1.5581114481518932e-05, + "loss": 0.0, + "step": 56930 + }, + { + "epoch": 14.02, + "learning_rate": 1.557409279856196e-05, + "loss": 0.0001, + "step": 56940 + }, + { + "epoch": 14.02, + "learning_rate": 1.556707111560499e-05, + "loss": 0.0, + "step": 56950 + }, + { + "epoch": 14.02, + "learning_rate": 1.5560049432648018e-05, + "loss": 0.0, + "step": 56960 + }, + { + "epoch": 14.02, + "learning_rate": 1.5553027749691046e-05, + "loss": 0.1718, + "step": 56970 + }, + { + "epoch": 14.02, + "learning_rate": 1.5546006066734077e-05, + "loss": 0.0004, + "step": 56980 + }, + { + "epoch": 14.02, + "learning_rate": 1.5538984383777104e-05, + "loss": 0.0, + "step": 56990 + }, + { + "epoch": 14.02, + "learning_rate": 1.553196270082013e-05, + "loss": 0.0156, + "step": 57000 + }, + { + "epoch": 14.02, + "learning_rate": 1.5524941017863162e-05, + "loss": 0.0, + "step": 57010 + }, + { + "epoch": 14.02, + "learning_rate": 1.5517919334906193e-05, + "loss": 0.0, + "step": 57020 + }, + { + "epoch": 14.02, + "learning_rate": 1.5510897651949217e-05, + "loss": 0.2133, + "step": 57030 + }, + { + "epoch": 14.02, + "learning_rate": 1.5503875968992248e-05, + "loss": 0.0003, + "step": 57040 + }, + { + "epoch": 14.02, + "learning_rate": 1.549685428603528e-05, + "loss": 0.0, + "step": 57050 + }, + { + "epoch": 14.02, + "learning_rate": 1.5489832603078307e-05, + "loss": 0.2614, + "step": 57060 + }, + { + "epoch": 14.02, + "learning_rate": 1.5482810920121334e-05, + "loss": 0.0, + "step": 57070 + }, + { + "epoch": 14.02, + "learning_rate": 1.5475789237164365e-05, + "loss": 0.0001, + "step": 57080 + }, + { + "epoch": 14.02, + "learning_rate": 1.5468767554207396e-05, + "loss": 0.0, + "step": 57090 + }, + { + "epoch": 14.02, + "learning_rate": 1.546174587125042e-05, + "loss": 0.0, + "step": 57100 + }, + { + "epoch": 14.02, + "learning_rate": 1.545472418829345e-05, + "loss": 0.0446, + "step": 57110 + }, + { + "epoch": 14.02, + "learning_rate": 1.5447702505336482e-05, + "loss": 0.0006, + "step": 57120 + }, + { + "epoch": 14.02, + "learning_rate": 1.544068082237951e-05, + "loss": 0.0071, + "step": 57130 + }, + { + "epoch": 14.02, + "learning_rate": 1.5433659139422537e-05, + "loss": 0.0, + "step": 57140 + }, + { + "epoch": 14.02, + "learning_rate": 1.5426637456465568e-05, + "loss": 0.0, + "step": 57150 + }, + { + "epoch": 14.02, + "learning_rate": 1.5419615773508595e-05, + "loss": 0.0002, + "step": 57160 + }, + { + "epoch": 14.02, + "learning_rate": 1.5412594090551623e-05, + "loss": 0.0014, + "step": 57170 + }, + { + "epoch": 14.02, + "learning_rate": 1.5405572407594653e-05, + "loss": 0.0121, + "step": 57180 + }, + { + "epoch": 14.02, + "learning_rate": 1.539855072463768e-05, + "loss": 0.0264, + "step": 57190 + }, + { + "epoch": 14.02, + "learning_rate": 1.5391529041680712e-05, + "loss": 0.0002, + "step": 57200 + }, + { + "epoch": 14.02, + "learning_rate": 1.538450735872374e-05, + "loss": 0.0001, + "step": 57210 + }, + { + "epoch": 14.02, + "learning_rate": 1.5377485675766767e-05, + "loss": 0.0, + "step": 57220 + }, + { + "epoch": 14.02, + "learning_rate": 1.5370463992809798e-05, + "loss": 0.0125, + "step": 57230 + }, + { + "epoch": 14.02, + "learning_rate": 1.5363442309852825e-05, + "loss": 0.0449, + "step": 57240 + }, + { + "epoch": 14.02, + "learning_rate": 1.5356420626895853e-05, + "loss": 0.0716, + "step": 57250 + }, + { + "epoch": 14.02, + "learning_rate": 1.5349398943938884e-05, + "loss": 0.0, + "step": 57260 + }, + { + "epoch": 14.02, + "learning_rate": 1.5342377260981914e-05, + "loss": 0.0017, + "step": 57270 + }, + { + "epoch": 14.02, + "learning_rate": 1.5335355578024942e-05, + "loss": 0.0008, + "step": 57280 + }, + { + "epoch": 14.02, + "learning_rate": 1.532833389506797e-05, + "loss": 0.0, + "step": 57290 + }, + { + "epoch": 14.02, + "learning_rate": 1.5321312212111e-05, + "loss": 0.2691, + "step": 57300 + }, + { + "epoch": 14.02, + "learning_rate": 1.531429052915403e-05, + "loss": 0.0, + "step": 57310 + }, + { + "epoch": 14.02, + "learning_rate": 1.5307268846197055e-05, + "loss": 0.0001, + "step": 57320 + }, + { + "epoch": 14.02, + "learning_rate": 1.5300247163240086e-05, + "loss": 0.003, + "step": 57330 + }, + { + "epoch": 14.02, + "learning_rate": 1.5293225480283117e-05, + "loss": 0.0, + "step": 57340 + }, + { + "epoch": 14.02, + "learning_rate": 1.5286203797326145e-05, + "loss": 0.0, + "step": 57350 + }, + { + "epoch": 14.02, + "learning_rate": 1.5279182114369172e-05, + "loss": 0.0, + "step": 57360 + }, + { + "epoch": 14.03, + "learning_rate": 1.5272160431412203e-05, + "loss": 0.0002, + "step": 57370 + }, + { + "epoch": 14.03, + "learning_rate": 1.526513874845523e-05, + "loss": 0.0002, + "step": 57380 + }, + { + "epoch": 14.03, + "learning_rate": 1.525811706549826e-05, + "loss": 0.0001, + "step": 57390 + }, + { + "epoch": 14.03, + "learning_rate": 1.5251095382541289e-05, + "loss": 0.0006, + "step": 57400 + }, + { + "epoch": 14.03, + "learning_rate": 1.5244073699584316e-05, + "loss": 0.0, + "step": 57410 + }, + { + "epoch": 14.03, + "learning_rate": 1.5237052016627345e-05, + "loss": 0.0, + "step": 57420 + }, + { + "epoch": 14.03, + "learning_rate": 1.5230030333670375e-05, + "loss": 0.0007, + "step": 57430 + }, + { + "epoch": 14.03, + "learning_rate": 1.5223008650713405e-05, + "loss": 0.0002, + "step": 57440 + }, + { + "epoch": 14.03, + "learning_rate": 1.5215986967756431e-05, + "loss": 0.152, + "step": 57450 + }, + { + "epoch": 14.03, + "learning_rate": 1.5208965284799462e-05, + "loss": 0.1168, + "step": 57460 + }, + { + "epoch": 14.03, + "learning_rate": 1.5201943601842491e-05, + "loss": 0.088, + "step": 57470 + }, + { + "epoch": 14.03, + "learning_rate": 1.5194921918885519e-05, + "loss": 0.0, + "step": 57480 + }, + { + "epoch": 14.03, + "learning_rate": 1.5187900235928548e-05, + "loss": 0.0, + "step": 57490 + }, + { + "epoch": 14.03, + "learning_rate": 1.5180878552971577e-05, + "loss": 0.0001, + "step": 57500 + }, + { + "epoch": 14.03, + "learning_rate": 1.5173856870014605e-05, + "loss": 0.0, + "step": 57510 + }, + { + "epoch": 14.03, + "learning_rate": 1.5166835187057636e-05, + "loss": 0.0, + "step": 57520 + }, + { + "epoch": 14.03, + "learning_rate": 1.5159813504100665e-05, + "loss": 0.0, + "step": 57530 + }, + { + "epoch": 14.03, + "learning_rate": 1.5152791821143692e-05, + "loss": 0.0, + "step": 57540 + }, + { + "epoch": 14.03, + "learning_rate": 1.5145770138186721e-05, + "loss": 0.003, + "step": 57550 + }, + { + "epoch": 14.03, + "learning_rate": 1.513874845522975e-05, + "loss": 0.0, + "step": 57560 + }, + { + "epoch": 14.03, + "learning_rate": 1.5131726772272778e-05, + "loss": 0.0004, + "step": 57570 + }, + { + "epoch": 14.03, + "learning_rate": 1.5124705089315807e-05, + "loss": 0.0001, + "step": 57580 + }, + { + "epoch": 14.03, + "learning_rate": 1.5117683406358838e-05, + "loss": 0.0, + "step": 57590 + }, + { + "epoch": 14.03, + "learning_rate": 1.5110661723401864e-05, + "loss": 0.0, + "step": 57600 + }, + { + "epoch": 14.03, + "learning_rate": 1.5103640040444895e-05, + "loss": 0.0, + "step": 57610 + }, + { + "epoch": 14.03, + "learning_rate": 1.5096618357487924e-05, + "loss": 0.0, + "step": 57620 + }, + { + "epoch": 14.03, + "learning_rate": 1.5089596674530953e-05, + "loss": 0.0, + "step": 57630 + }, + { + "epoch": 14.03, + "learning_rate": 1.508257499157398e-05, + "loss": 0.0, + "step": 57640 + }, + { + "epoch": 14.03, + "learning_rate": 1.507555330861701e-05, + "loss": 0.0, + "step": 57650 + }, + { + "epoch": 14.03, + "learning_rate": 1.506853162566004e-05, + "loss": 0.0, + "step": 57660 + }, + { + "epoch": 14.03, + "learning_rate": 1.5061509942703067e-05, + "loss": 0.0, + "step": 57670 + }, + { + "epoch": 14.03, + "learning_rate": 1.5054488259746097e-05, + "loss": 0.0, + "step": 57680 + }, + { + "epoch": 14.03, + "learning_rate": 1.5047466576789127e-05, + "loss": 0.0, + "step": 57690 + }, + { + "epoch": 14.03, + "learning_rate": 1.5040444893832154e-05, + "loss": 0.0, + "step": 57700 + }, + { + "epoch": 14.03, + "learning_rate": 1.5033423210875183e-05, + "loss": 0.001, + "step": 57710 + }, + { + "epoch": 14.03, + "learning_rate": 1.5026401527918212e-05, + "loss": 0.0, + "step": 57720 + }, + { + "epoch": 14.03, + "learning_rate": 1.501937984496124e-05, + "loss": 0.0, + "step": 57730 + }, + { + "epoch": 14.03, + "learning_rate": 1.5012358162004269e-05, + "loss": 0.0, + "step": 57740 + }, + { + "epoch": 14.03, + "learning_rate": 1.50053364790473e-05, + "loss": 0.0, + "step": 57750 + }, + { + "epoch": 14.03, + "learning_rate": 1.4998314796090326e-05, + "loss": 0.0001, + "step": 57760 + }, + { + "epoch": 14.03, + "learning_rate": 1.4991293113133357e-05, + "loss": 0.0, + "step": 57770 + }, + { + "epoch": 14.03, + "learning_rate": 1.4984271430176386e-05, + "loss": 0.0001, + "step": 57780 + }, + { + "epoch": 14.03, + "learning_rate": 1.4977249747219415e-05, + "loss": 0.0, + "step": 57790 + }, + { + "epoch": 14.03, + "learning_rate": 1.4970228064262443e-05, + "loss": 0.0, + "step": 57800 + }, + { + "epoch": 14.03, + "learning_rate": 1.4963206381305472e-05, + "loss": 0.0, + "step": 57810 + }, + { + "epoch": 14.03, + "learning_rate": 1.4956184698348503e-05, + "loss": 0.0, + "step": 57820 + }, + { + "epoch": 14.03, + "learning_rate": 1.4949163015391528e-05, + "loss": 0.0001, + "step": 57830 + }, + { + "epoch": 14.03, + "learning_rate": 1.494214133243456e-05, + "loss": 0.0, + "step": 57840 + }, + { + "epoch": 14.03, + "learning_rate": 1.4935119649477588e-05, + "loss": 0.0003, + "step": 57850 + }, + { + "epoch": 14.03, + "learning_rate": 1.4928097966520616e-05, + "loss": 0.0001, + "step": 57860 + }, + { + "epoch": 14.03, + "learning_rate": 1.4921076283563645e-05, + "loss": 0.0, + "step": 57870 + }, + { + "epoch": 14.03, + "learning_rate": 1.4914054600606674e-05, + "loss": 0.0001, + "step": 57880 + }, + { + "epoch": 14.03, + "learning_rate": 1.4907032917649702e-05, + "loss": 0.0, + "step": 57890 + }, + { + "epoch": 14.03, + "learning_rate": 1.4900011234692731e-05, + "loss": 0.0, + "step": 57900 + }, + { + "epoch": 14.03, + "learning_rate": 1.4892989551735762e-05, + "loss": 0.1787, + "step": 57910 + }, + { + "epoch": 14.03, + "learning_rate": 1.4885967868778788e-05, + "loss": 0.0, + "step": 57920 + }, + { + "epoch": 14.03, + "learning_rate": 1.4878946185821819e-05, + "loss": 0.0, + "step": 57930 + }, + { + "epoch": 14.03, + "learning_rate": 1.4871924502864848e-05, + "loss": 0.0, + "step": 57940 + }, + { + "epoch": 14.03, + "learning_rate": 1.4864902819907875e-05, + "loss": 0.0402, + "step": 57950 + }, + { + "epoch": 14.03, + "learning_rate": 1.4857881136950904e-05, + "loss": 0.0001, + "step": 57960 + }, + { + "epoch": 14.03, + "learning_rate": 1.4850859453993934e-05, + "loss": 0.3572, + "step": 57970 + }, + { + "epoch": 14.03, + "learning_rate": 1.4843837771036964e-05, + "loss": 0.1007, + "step": 57980 + }, + { + "epoch": 14.03, + "learning_rate": 1.483681608807999e-05, + "loss": 0.0, + "step": 57990 + }, + { + "epoch": 14.03, + "learning_rate": 1.4829794405123021e-05, + "loss": 0.0, + "step": 58000 + }, + { + "epoch": 14.03, + "learning_rate": 1.482277272216605e-05, + "loss": 0.0, + "step": 58010 + }, + { + "epoch": 14.03, + "learning_rate": 1.4815751039209078e-05, + "loss": 0.0, + "step": 58020 + }, + { + "epoch": 14.03, + "learning_rate": 1.4808729356252107e-05, + "loss": 0.0, + "step": 58030 + }, + { + "epoch": 14.03, + "learning_rate": 1.4801707673295138e-05, + "loss": 0.0075, + "step": 58040 + }, + { + "epoch": 14.03, + "learning_rate": 1.4794685990338164e-05, + "loss": 0.1999, + "step": 58050 + }, + { + "epoch": 14.03, + "learning_rate": 1.4787664307381195e-05, + "loss": 0.3427, + "step": 58060 + }, + { + "epoch": 14.03, + "learning_rate": 1.4780642624424224e-05, + "loss": 0.0, + "step": 58070 + }, + { + "epoch": 14.03, + "learning_rate": 1.4773620941467251e-05, + "loss": 0.5486, + "step": 58080 + }, + { + "epoch": 14.03, + "learning_rate": 1.476659925851028e-05, + "loss": 0.0, + "step": 58090 + }, + { + "epoch": 14.03, + "learning_rate": 1.475957757555331e-05, + "loss": 0.0003, + "step": 58100 + }, + { + "epoch": 14.03, + "learning_rate": 1.4752555892596337e-05, + "loss": 0.0, + "step": 58110 + }, + { + "epoch": 14.03, + "learning_rate": 1.4745534209639366e-05, + "loss": 0.0019, + "step": 58120 + }, + { + "epoch": 14.03, + "learning_rate": 1.4738512526682397e-05, + "loss": 0.0342, + "step": 58130 + }, + { + "epoch": 14.03, + "learning_rate": 1.4731490843725423e-05, + "loss": 0.0, + "step": 58140 + }, + { + "epoch": 14.03, + "learning_rate": 1.4724469160768454e-05, + "loss": 0.0, + "step": 58150 + }, + { + "epoch": 14.04, + "learning_rate": 1.4717447477811483e-05, + "loss": 0.0001, + "step": 58160 + }, + { + "epoch": 14.04, + "learning_rate": 1.4710425794854512e-05, + "loss": 0.0, + "step": 58170 + }, + { + "epoch": 14.04, + "learning_rate": 1.470340411189754e-05, + "loss": 0.0001, + "step": 58180 + }, + { + "epoch": 14.04, + "learning_rate": 1.4696382428940569e-05, + "loss": 0.0003, + "step": 58190 + }, + { + "epoch": 14.04, + "learning_rate": 1.46893607459836e-05, + "loss": 0.0107, + "step": 58200 + }, + { + "epoch": 14.04, + "learning_rate": 1.4682339063026626e-05, + "loss": 0.2274, + "step": 58210 + }, + { + "epoch": 14.04, + "learning_rate": 1.4675317380069656e-05, + "loss": 0.0, + "step": 58220 + }, + { + "epoch": 14.04, + "learning_rate": 1.4668295697112686e-05, + "loss": 0.0, + "step": 58230 + }, + { + "epoch": 14.04, + "learning_rate": 1.4661274014155713e-05, + "loss": 0.0001, + "step": 58240 + }, + { + "epoch": 14.04, + "learning_rate": 1.4654252331198742e-05, + "loss": 0.0, + "step": 58250 + }, + { + "epoch": 14.04, + "learning_rate": 1.4647230648241772e-05, + "loss": 0.065, + "step": 58260 + }, + { + "epoch": 14.04, + "learning_rate": 1.4640208965284799e-05, + "loss": 0.0001, + "step": 58270 + }, + { + "epoch": 14.04, + "learning_rate": 1.4633187282327828e-05, + "loss": 0.0193, + "step": 58280 + }, + { + "epoch": 14.04, + "learning_rate": 1.4626165599370859e-05, + "loss": 0.0073, + "step": 58290 + }, + { + "epoch": 14.04, + "learning_rate": 1.4619143916413885e-05, + "loss": 0.0, + "step": 58300 + }, + { + "epoch": 14.04, + "learning_rate": 1.4612122233456916e-05, + "loss": 0.0, + "step": 58310 + }, + { + "epoch": 14.04, + "learning_rate": 1.4605100550499945e-05, + "loss": 0.0, + "step": 58320 + }, + { + "epoch": 14.04, + "learning_rate": 1.4598078867542974e-05, + "loss": 0.0, + "step": 58330 + }, + { + "epoch": 14.04, + "learning_rate": 1.4591057184586002e-05, + "loss": 0.0, + "step": 58340 + }, + { + "epoch": 14.04, + "learning_rate": 1.458403550162903e-05, + "loss": 0.0, + "step": 58350 + }, + { + "epoch": 14.04, + "learning_rate": 1.4577013818672062e-05, + "loss": 0.0004, + "step": 58360 + }, + { + "epoch": 14.04, + "learning_rate": 1.4569992135715087e-05, + "loss": 0.0, + "step": 58370 + }, + { + "epoch": 14.04, + "learning_rate": 1.4562970452758118e-05, + "loss": 0.3394, + "step": 58380 + }, + { + "epoch": 14.04, + "learning_rate": 1.4555948769801148e-05, + "loss": 0.0, + "step": 58390 + }, + { + "epoch": 14.04, + "learning_rate": 1.4548927086844175e-05, + "loss": 0.0001, + "step": 58400 + }, + { + "epoch": 14.04, + "learning_rate": 1.4541905403887204e-05, + "loss": 0.013, + "step": 58410 + }, + { + "epoch": 14.04, + "learning_rate": 1.4534883720930233e-05, + "loss": 0.0002, + "step": 58420 + }, + { + "epoch": 14.04, + "learning_rate": 1.4527862037973261e-05, + "loss": 0.0531, + "step": 58430 + }, + { + "epoch": 14.04, + "learning_rate": 1.452084035501629e-05, + "loss": 0.0, + "step": 58440 + }, + { + "epoch": 14.04, + "learning_rate": 1.4513818672059321e-05, + "loss": 0.0, + "step": 58450 + }, + { + "epoch": 14.04, + "learning_rate": 1.4506796989102347e-05, + "loss": 0.0, + "step": 58460 + }, + { + "epoch": 14.04, + "learning_rate": 1.4499775306145378e-05, + "loss": 0.0001, + "step": 58470 + }, + { + "epoch": 14.04, + "learning_rate": 1.4492753623188407e-05, + "loss": 0.0204, + "step": 58480 + }, + { + "epoch": 14.04, + "learning_rate": 1.4485731940231434e-05, + "loss": 0.0, + "step": 58490 + }, + { + "epoch": 14.04, + "learning_rate": 1.4478710257274463e-05, + "loss": 0.0, + "step": 58500 + }, + { + "epoch": 14.04, + "learning_rate": 1.4471688574317493e-05, + "loss": 0.0, + "step": 58510 + }, + { + "epoch": 14.04, + "learning_rate": 1.4464666891360524e-05, + "loss": 0.0, + "step": 58520 + }, + { + "epoch": 14.04, + "learning_rate": 1.4457645208403551e-05, + "loss": 0.0002, + "step": 58530 + }, + { + "epoch": 14.04, + "learning_rate": 1.445062352544658e-05, + "loss": 0.0003, + "step": 58540 + }, + { + "epoch": 14.04, + "learning_rate": 1.444360184248961e-05, + "loss": 0.0, + "step": 58550 + }, + { + "epoch": 14.04, + "learning_rate": 1.4436580159532637e-05, + "loss": 0.0, + "step": 58560 + }, + { + "epoch": 14.04, + "learning_rate": 1.4429558476575666e-05, + "loss": 0.0009, + "step": 58570 + }, + { + "epoch": 14.04, + "learning_rate": 1.4422536793618697e-05, + "loss": 0.0, + "step": 58580 + }, + { + "epoch": 14.04, + "learning_rate": 1.4415515110661723e-05, + "loss": 0.0, + "step": 58590 + }, + { + "epoch": 14.04, + "learning_rate": 1.4408493427704754e-05, + "loss": 0.0148, + "step": 58600 + }, + { + "epoch": 14.04, + "learning_rate": 1.4401471744747783e-05, + "loss": 0.0, + "step": 58610 + }, + { + "epoch": 14.04, + "learning_rate": 1.439445006179081e-05, + "loss": 0.0, + "step": 58620 + }, + { + "epoch": 14.04, + "learning_rate": 1.438742837883384e-05, + "loss": 0.0, + "step": 58630 + }, + { + "epoch": 14.04, + "learning_rate": 1.4380406695876869e-05, + "loss": 0.0, + "step": 58640 + }, + { + "epoch": 14.04, + "learning_rate": 1.4373385012919896e-05, + "loss": 0.0, + "step": 58650 + }, + { + "epoch": 14.04, + "learning_rate": 1.4366363329962925e-05, + "loss": 0.0022, + "step": 58660 + }, + { + "epoch": 14.04, + "learning_rate": 1.4359341647005956e-05, + "loss": 0.1539, + "step": 58670 + }, + { + "epoch": 14.04, + "learning_rate": 1.4352319964048985e-05, + "loss": 0.0, + "step": 58680 + }, + { + "epoch": 14.04, + "learning_rate": 1.4345298281092013e-05, + "loss": 0.0, + "step": 58690 + }, + { + "epoch": 14.04, + "learning_rate": 1.4338276598135042e-05, + "loss": 0.0, + "step": 58700 + }, + { + "epoch": 14.04, + "learning_rate": 1.4331254915178071e-05, + "loss": 0.0, + "step": 58710 + }, + { + "epoch": 14.04, + "learning_rate": 1.4324233232221099e-05, + "loss": 0.0, + "step": 58720 + }, + { + "epoch": 14.04, + "learning_rate": 1.4317211549264128e-05, + "loss": 0.0093, + "step": 58730 + }, + { + "epoch": 14.04, + "learning_rate": 1.4310189866307159e-05, + "loss": 0.0019, + "step": 58740 + }, + { + "epoch": 14.04, + "learning_rate": 1.4303168183350185e-05, + "loss": 0.5166, + "step": 58750 + }, + { + "epoch": 14.04, + "learning_rate": 1.4296146500393215e-05, + "loss": 0.0003, + "step": 58760 + }, + { + "epoch": 14.04, + "learning_rate": 1.4289124817436245e-05, + "loss": 0.0, + "step": 58770 + }, + { + "epoch": 14.04, + "learning_rate": 1.4282103134479272e-05, + "loss": 0.0001, + "step": 58780 + }, + { + "epoch": 14.04, + "learning_rate": 1.4275081451522301e-05, + "loss": 0.0036, + "step": 58790 + }, + { + "epoch": 14.04, + "learning_rate": 1.426805976856533e-05, + "loss": 0.0, + "step": 58800 + }, + { + "epoch": 14.04, + "learning_rate": 1.4261038085608358e-05, + "loss": 0.0004, + "step": 58810 + }, + { + "epoch": 14.04, + "learning_rate": 1.4254016402651387e-05, + "loss": 0.0003, + "step": 58820 + }, + { + "epoch": 14.04, + "learning_rate": 1.4246994719694418e-05, + "loss": 0.7018, + "step": 58830 + }, + { + "epoch": 14.04, + "learning_rate": 1.4239973036737444e-05, + "loss": 0.0001, + "step": 58840 + }, + { + "epoch": 14.04, + "learning_rate": 1.4232951353780475e-05, + "loss": 0.0003, + "step": 58850 + }, + { + "epoch": 14.04, + "learning_rate": 1.4225929670823504e-05, + "loss": 0.0012, + "step": 58860 + }, + { + "epoch": 14.04, + "learning_rate": 1.4218907987866533e-05, + "loss": 0.0, + "step": 58870 + }, + { + "epoch": 14.04, + "learning_rate": 1.421188630490956e-05, + "loss": 0.001, + "step": 58880 + }, + { + "epoch": 14.04, + "learning_rate": 1.420486462195259e-05, + "loss": 0.0, + "step": 58890 + }, + { + "epoch": 14.04, + "learning_rate": 1.419784293899562e-05, + "loss": 0.0, + "step": 58900 + }, + { + "epoch": 14.04, + "learning_rate": 1.4190821256038646e-05, + "loss": 0.0201, + "step": 58910 + }, + { + "epoch": 14.04, + "learning_rate": 1.4183799573081677e-05, + "loss": 0.0, + "step": 58920 + }, + { + "epoch": 14.04, + "learning_rate": 1.4176777890124707e-05, + "loss": 0.0006, + "step": 58930 + }, + { + "epoch": 14.04, + "learning_rate": 1.4169756207167734e-05, + "loss": 0.0, + "step": 58940 + }, + { + "epoch": 14.05, + "learning_rate": 1.4162734524210763e-05, + "loss": 0.0001, + "step": 58950 + }, + { + "epoch": 14.05, + "learning_rate": 1.4155712841253792e-05, + "loss": 0.0, + "step": 58960 + }, + { + "epoch": 14.05, + "learning_rate": 1.414869115829682e-05, + "loss": 0.3201, + "step": 58970 + }, + { + "epoch": 14.05, + "learning_rate": 1.4141669475339849e-05, + "loss": 0.0018, + "step": 58980 + }, + { + "epoch": 14.05, + "learning_rate": 1.413464779238288e-05, + "loss": 0.0, + "step": 58990 + }, + { + "epoch": 14.05, + "learning_rate": 1.4127626109425906e-05, + "loss": 0.0154, + "step": 59000 + }, + { + "epoch": 14.05, + "learning_rate": 1.4120604426468937e-05, + "loss": 0.1947, + "step": 59010 + }, + { + "epoch": 14.05, + "learning_rate": 1.4113582743511966e-05, + "loss": 0.0, + "step": 59020 + }, + { + "epoch": 14.05, + "learning_rate": 1.4106561060554995e-05, + "loss": 0.0, + "step": 59030 + }, + { + "epoch": 14.05, + "learning_rate": 1.4099539377598022e-05, + "loss": 0.0, + "step": 59040 + }, + { + "epoch": 14.05, + "learning_rate": 1.4092517694641053e-05, + "loss": 0.0, + "step": 59050 + }, + { + "epoch": 14.05, + "learning_rate": 1.4085496011684083e-05, + "loss": 0.0, + "step": 59060 + }, + { + "epoch": 14.05, + "learning_rate": 1.407847432872711e-05, + "loss": 0.0, + "step": 59070 + }, + { + "epoch": 14.05, + "learning_rate": 1.407145264577014e-05, + "loss": 0.0002, + "step": 59080 + }, + { + "epoch": 14.05, + "learning_rate": 1.4064430962813168e-05, + "loss": 0.0, + "step": 59090 + }, + { + "epoch": 14.05, + "learning_rate": 1.4057409279856196e-05, + "loss": 0.0001, + "step": 59100 + }, + { + "epoch": 14.05, + "learning_rate": 1.4050387596899225e-05, + "loss": 0.0, + "step": 59110 + }, + { + "epoch": 14.05, + "learning_rate": 1.4043365913942256e-05, + "loss": 0.0133, + "step": 59120 + }, + { + "epoch": 14.05, + "learning_rate": 1.4036344230985282e-05, + "loss": 0.0, + "step": 59130 + }, + { + "epoch": 14.05, + "learning_rate": 1.4029322548028313e-05, + "loss": 0.0531, + "step": 59140 + }, + { + "epoch": 14.05, + "learning_rate": 1.4022300865071342e-05, + "loss": 0.0, + "step": 59150 + }, + { + "epoch": 14.05, + "learning_rate": 1.401527918211437e-05, + "loss": 0.0, + "step": 59160 + }, + { + "epoch": 14.05, + "learning_rate": 1.4008257499157398e-05, + "loss": 0.0, + "step": 59170 + }, + { + "epoch": 14.05, + "learning_rate": 1.4001235816200428e-05, + "loss": 0.082, + "step": 59180 + }, + { + "epoch": 14.05, + "learning_rate": 1.3994214133243455e-05, + "loss": 0.0, + "step": 59190 + }, + { + "epoch": 14.05, + "learning_rate": 1.3987192450286484e-05, + "loss": 0.0001, + "step": 59200 + }, + { + "epoch": 14.05, + "learning_rate": 1.3980170767329515e-05, + "loss": 0.0002, + "step": 59210 + }, + { + "epoch": 14.05, + "learning_rate": 1.3973149084372544e-05, + "loss": 0.2338, + "step": 59220 + }, + { + "epoch": 14.05, + "learning_rate": 1.3966127401415572e-05, + "loss": 0.0001, + "step": 59230 + }, + { + "epoch": 14.05, + "learning_rate": 1.3959105718458601e-05, + "loss": 0.0, + "step": 59240 + }, + { + "epoch": 14.05, + "learning_rate": 1.395208403550163e-05, + "loss": 0.0059, + "step": 59250 + }, + { + "epoch": 14.05, + "learning_rate": 1.3945062352544658e-05, + "loss": 0.0001, + "step": 59260 + }, + { + "epoch": 14.05, + "learning_rate": 1.3938040669587687e-05, + "loss": 0.0033, + "step": 59270 + }, + { + "epoch": 14.05, + "learning_rate": 1.3931018986630718e-05, + "loss": 0.0, + "step": 59280 + }, + { + "epoch": 14.05, + "learning_rate": 1.3923997303673744e-05, + "loss": 0.0002, + "step": 59290 + }, + { + "epoch": 14.05, + "learning_rate": 1.3916975620716775e-05, + "loss": 0.0, + "step": 59300 + }, + { + "epoch": 14.05, + "learning_rate": 1.3909953937759804e-05, + "loss": 0.0611, + "step": 59310 + }, + { + "epoch": 14.05, + "learning_rate": 1.3902932254802831e-05, + "loss": 0.0, + "step": 59320 + }, + { + "epoch": 14.05, + "learning_rate": 1.389591057184586e-05, + "loss": 0.0001, + "step": 59330 + }, + { + "epoch": 14.05, + "learning_rate": 1.388888888888889e-05, + "loss": 0.0, + "step": 59340 + }, + { + "epoch": 14.05, + "eval_accuracy": 0.7520319958715005, + "eval_loss": 1.8392677307128906, + "eval_runtime": 1232.308, + "eval_samples_per_second": 6.29, + "eval_steps_per_second": 3.145, + "step": 59340 + }, + { + "epoch": 15.0, + "learning_rate": 1.3881867205931917e-05, + "loss": 0.0, + "step": 59350 + }, + { + "epoch": 15.0, + "learning_rate": 1.3874845522974946e-05, + "loss": 0.0, + "step": 59360 + }, + { + "epoch": 15.0, + "learning_rate": 1.3867823840017977e-05, + "loss": 0.0, + "step": 59370 + }, + { + "epoch": 15.0, + "learning_rate": 1.3860802157061006e-05, + "loss": 0.0001, + "step": 59380 + }, + { + "epoch": 15.0, + "learning_rate": 1.3853780474104034e-05, + "loss": 0.0, + "step": 59390 + }, + { + "epoch": 15.0, + "learning_rate": 1.3846758791147063e-05, + "loss": 0.1887, + "step": 59400 + }, + { + "epoch": 15.0, + "learning_rate": 1.3839737108190092e-05, + "loss": 0.0, + "step": 59410 + }, + { + "epoch": 15.0, + "learning_rate": 1.383271542523312e-05, + "loss": 0.0, + "step": 59420 + }, + { + "epoch": 15.0, + "learning_rate": 1.3825693742276149e-05, + "loss": 0.0004, + "step": 59430 + }, + { + "epoch": 15.0, + "learning_rate": 1.381867205931918e-05, + "loss": 0.0, + "step": 59440 + }, + { + "epoch": 15.0, + "learning_rate": 1.3811650376362206e-05, + "loss": 0.2473, + "step": 59450 + }, + { + "epoch": 15.0, + "learning_rate": 1.3804628693405236e-05, + "loss": 0.0, + "step": 59460 + }, + { + "epoch": 15.0, + "learning_rate": 1.3797607010448266e-05, + "loss": 0.0001, + "step": 59470 + }, + { + "epoch": 15.0, + "learning_rate": 1.3790585327491293e-05, + "loss": 0.1758, + "step": 59480 + }, + { + "epoch": 15.0, + "learning_rate": 1.3783563644534322e-05, + "loss": 0.0001, + "step": 59490 + }, + { + "epoch": 15.0, + "learning_rate": 1.3776541961577351e-05, + "loss": 0.0002, + "step": 59500 + }, + { + "epoch": 15.0, + "learning_rate": 1.3769520278620379e-05, + "loss": 0.0, + "step": 59510 + }, + { + "epoch": 15.0, + "learning_rate": 1.3762498595663408e-05, + "loss": 0.0006, + "step": 59520 + }, + { + "epoch": 15.0, + "learning_rate": 1.3755476912706439e-05, + "loss": 0.0, + "step": 59530 + }, + { + "epoch": 15.0, + "learning_rate": 1.3748455229749466e-05, + "loss": 0.1103, + "step": 59540 + }, + { + "epoch": 15.0, + "learning_rate": 1.3741433546792496e-05, + "loss": 0.0, + "step": 59550 + }, + { + "epoch": 15.0, + "learning_rate": 1.3734411863835525e-05, + "loss": 0.0025, + "step": 59560 + }, + { + "epoch": 15.0, + "learning_rate": 1.3727390180878556e-05, + "loss": 0.2099, + "step": 59570 + }, + { + "epoch": 15.0, + "learning_rate": 1.3720368497921582e-05, + "loss": 0.0, + "step": 59580 + }, + { + "epoch": 15.0, + "learning_rate": 1.3713346814964612e-05, + "loss": 0.1191, + "step": 59590 + }, + { + "epoch": 15.0, + "learning_rate": 1.3706325132007642e-05, + "loss": 0.0011, + "step": 59600 + }, + { + "epoch": 15.0, + "learning_rate": 1.3699303449050669e-05, + "loss": 0.0, + "step": 59610 + }, + { + "epoch": 15.0, + "learning_rate": 1.3692281766093698e-05, + "loss": 0.0003, + "step": 59620 + }, + { + "epoch": 15.0, + "learning_rate": 1.3685260083136727e-05, + "loss": 0.0, + "step": 59630 + }, + { + "epoch": 15.0, + "learning_rate": 1.3678238400179755e-05, + "loss": 0.0, + "step": 59640 + }, + { + "epoch": 15.0, + "learning_rate": 1.3671216717222784e-05, + "loss": 0.0003, + "step": 59650 + }, + { + "epoch": 15.0, + "learning_rate": 1.3664195034265815e-05, + "loss": 0.0, + "step": 59660 + }, + { + "epoch": 15.0, + "learning_rate": 1.365717335130884e-05, + "loss": 0.0367, + "step": 59670 + }, + { + "epoch": 15.0, + "learning_rate": 1.3650151668351872e-05, + "loss": 0.0371, + "step": 59680 + }, + { + "epoch": 15.0, + "learning_rate": 1.3643129985394901e-05, + "loss": 0.0006, + "step": 59690 + }, + { + "epoch": 15.0, + "learning_rate": 1.3636108302437928e-05, + "loss": 0.0, + "step": 59700 + }, + { + "epoch": 15.0, + "learning_rate": 1.3629086619480958e-05, + "loss": 0.0001, + "step": 59710 + }, + { + "epoch": 15.0, + "learning_rate": 1.3622064936523987e-05, + "loss": 0.0268, + "step": 59720 + }, + { + "epoch": 15.0, + "learning_rate": 1.3615043253567018e-05, + "loss": 0.0, + "step": 59730 + }, + { + "epoch": 15.01, + "learning_rate": 1.3608021570610043e-05, + "loss": 0.0, + "step": 59740 + }, + { + "epoch": 15.01, + "learning_rate": 1.3600999887653074e-05, + "loss": 0.0, + "step": 59750 + }, + { + "epoch": 15.01, + "learning_rate": 1.3593978204696103e-05, + "loss": 0.0081, + "step": 59760 + }, + { + "epoch": 15.01, + "learning_rate": 1.3586956521739131e-05, + "loss": 0.1137, + "step": 59770 + }, + { + "epoch": 15.01, + "learning_rate": 1.357993483878216e-05, + "loss": 0.0, + "step": 59780 + }, + { + "epoch": 15.01, + "learning_rate": 1.357291315582519e-05, + "loss": 0.1853, + "step": 59790 + }, + { + "epoch": 15.01, + "learning_rate": 1.3565891472868217e-05, + "loss": 0.0003, + "step": 59800 + }, + { + "epoch": 15.01, + "learning_rate": 1.3558869789911246e-05, + "loss": 0.0, + "step": 59810 + }, + { + "epoch": 15.01, + "learning_rate": 1.3551848106954277e-05, + "loss": 0.0009, + "step": 59820 + }, + { + "epoch": 15.01, + "learning_rate": 1.3544826423997303e-05, + "loss": 0.0, + "step": 59830 + }, + { + "epoch": 15.01, + "learning_rate": 1.3537804741040334e-05, + "loss": 0.0001, + "step": 59840 + }, + { + "epoch": 15.01, + "learning_rate": 1.3530783058083363e-05, + "loss": 0.0, + "step": 59850 + }, + { + "epoch": 15.01, + "learning_rate": 1.352376137512639e-05, + "loss": 0.0001, + "step": 59860 + }, + { + "epoch": 15.01, + "learning_rate": 1.351673969216942e-05, + "loss": 0.0001, + "step": 59870 + }, + { + "epoch": 15.01, + "learning_rate": 1.3509718009212449e-05, + "loss": 0.0058, + "step": 59880 + }, + { + "epoch": 15.01, + "learning_rate": 1.3502696326255476e-05, + "loss": 0.0302, + "step": 59890 + }, + { + "epoch": 15.01, + "learning_rate": 1.3495674643298505e-05, + "loss": 0.3061, + "step": 59900 + }, + { + "epoch": 15.01, + "learning_rate": 1.3488652960341536e-05, + "loss": 0.0, + "step": 59910 + }, + { + "epoch": 15.01, + "learning_rate": 1.3481631277384565e-05, + "loss": 0.1453, + "step": 59920 + }, + { + "epoch": 15.01, + "learning_rate": 1.3474609594427593e-05, + "loss": 0.5439, + "step": 59930 + }, + { + "epoch": 15.01, + "learning_rate": 1.3467587911470622e-05, + "loss": 0.0213, + "step": 59940 + }, + { + "epoch": 15.01, + "learning_rate": 1.3460566228513651e-05, + "loss": 0.0, + "step": 59950 + }, + { + "epoch": 15.01, + "learning_rate": 1.3453544545556679e-05, + "loss": 0.0, + "step": 59960 + }, + { + "epoch": 15.01, + "learning_rate": 1.3446522862599708e-05, + "loss": 0.0, + "step": 59970 + }, + { + "epoch": 15.01, + "learning_rate": 1.3439501179642739e-05, + "loss": 0.0, + "step": 59980 + }, + { + "epoch": 15.01, + "learning_rate": 1.3432479496685765e-05, + "loss": 0.0005, + "step": 59990 + }, + { + "epoch": 15.01, + "learning_rate": 1.3425457813728795e-05, + "loss": 0.0016, + "step": 60000 + }, + { + "epoch": 15.01, + "learning_rate": 1.3418436130771825e-05, + "loss": 0.0, + "step": 60010 + }, + { + "epoch": 15.01, + "learning_rate": 1.3411414447814852e-05, + "loss": 0.0, + "step": 60020 + }, + { + "epoch": 15.01, + "learning_rate": 1.3404392764857881e-05, + "loss": 0.0003, + "step": 60030 + }, + { + "epoch": 15.01, + "learning_rate": 1.339737108190091e-05, + "loss": 0.0, + "step": 60040 + }, + { + "epoch": 15.01, + "learning_rate": 1.3390349398943938e-05, + "loss": 0.2857, + "step": 60050 + }, + { + "epoch": 15.01, + "learning_rate": 1.3383327715986969e-05, + "loss": 0.0, + "step": 60060 + }, + { + "epoch": 15.01, + "learning_rate": 1.3376306033029998e-05, + "loss": 0.0001, + "step": 60070 + }, + { + "epoch": 15.01, + "learning_rate": 1.3369284350073027e-05, + "loss": 0.0005, + "step": 60080 + }, + { + "epoch": 15.01, + "learning_rate": 1.3362262667116055e-05, + "loss": 0.0, + "step": 60090 + }, + { + "epoch": 15.01, + "learning_rate": 1.3355240984159084e-05, + "loss": 0.0, + "step": 60100 + }, + { + "epoch": 15.01, + "learning_rate": 1.3348219301202115e-05, + "loss": 0.3067, + "step": 60110 + }, + { + "epoch": 15.01, + "learning_rate": 1.334119761824514e-05, + "loss": 0.0, + "step": 60120 + }, + { + "epoch": 15.01, + "learning_rate": 1.3334175935288171e-05, + "loss": 0.0, + "step": 60130 + }, + { + "epoch": 15.01, + "learning_rate": 1.33271542523312e-05, + "loss": 0.0, + "step": 60140 + }, + { + "epoch": 15.01, + "learning_rate": 1.3320132569374228e-05, + "loss": 0.0, + "step": 60150 + }, + { + "epoch": 15.01, + "learning_rate": 1.3313110886417257e-05, + "loss": 0.0001, + "step": 60160 + }, + { + "epoch": 15.01, + "learning_rate": 1.3306089203460286e-05, + "loss": 0.0087, + "step": 60170 + }, + { + "epoch": 15.01, + "learning_rate": 1.3299067520503314e-05, + "loss": 0.0, + "step": 60180 + }, + { + "epoch": 15.01, + "learning_rate": 1.3292045837546343e-05, + "loss": 0.0, + "step": 60190 + }, + { + "epoch": 15.01, + "learning_rate": 1.3285024154589374e-05, + "loss": 0.0, + "step": 60200 + }, + { + "epoch": 15.01, + "learning_rate": 1.32780024716324e-05, + "loss": 0.2908, + "step": 60210 + }, + { + "epoch": 15.01, + "learning_rate": 1.327098078867543e-05, + "loss": 0.0001, + "step": 60220 + }, + { + "epoch": 15.01, + "learning_rate": 1.326395910571846e-05, + "loss": 0.0001, + "step": 60230 + }, + { + "epoch": 15.01, + "learning_rate": 1.3256937422761487e-05, + "loss": 0.0, + "step": 60240 + }, + { + "epoch": 15.01, + "learning_rate": 1.3249915739804517e-05, + "loss": 0.0001, + "step": 60250 + }, + { + "epoch": 15.01, + "learning_rate": 1.3242894056847546e-05, + "loss": 0.0, + "step": 60260 + }, + { + "epoch": 15.01, + "learning_rate": 1.3235872373890577e-05, + "loss": 0.0, + "step": 60270 + }, + { + "epoch": 15.01, + "learning_rate": 1.3228850690933602e-05, + "loss": 0.0323, + "step": 60280 + }, + { + "epoch": 15.01, + "learning_rate": 1.3221829007976633e-05, + "loss": 0.0, + "step": 60290 + }, + { + "epoch": 15.01, + "learning_rate": 1.3214807325019662e-05, + "loss": 0.0, + "step": 60300 + }, + { + "epoch": 15.01, + "learning_rate": 1.320778564206269e-05, + "loss": 0.0, + "step": 60310 + }, + { + "epoch": 15.01, + "learning_rate": 1.3200763959105719e-05, + "loss": 0.0001, + "step": 60320 + }, + { + "epoch": 15.01, + "learning_rate": 1.3193742276148748e-05, + "loss": 0.0, + "step": 60330 + }, + { + "epoch": 15.01, + "learning_rate": 1.3186720593191776e-05, + "loss": 0.0, + "step": 60340 + }, + { + "epoch": 15.01, + "learning_rate": 1.3179698910234805e-05, + "loss": 0.1836, + "step": 60350 + }, + { + "epoch": 15.01, + "learning_rate": 1.3172677227277836e-05, + "loss": 0.0004, + "step": 60360 + }, + { + "epoch": 15.01, + "learning_rate": 1.3165655544320862e-05, + "loss": 0.0, + "step": 60370 + }, + { + "epoch": 15.01, + "learning_rate": 1.3158633861363893e-05, + "loss": 0.0, + "step": 60380 + }, + { + "epoch": 15.01, + "learning_rate": 1.3151612178406922e-05, + "loss": 0.0042, + "step": 60390 + }, + { + "epoch": 15.01, + "learning_rate": 1.314459049544995e-05, + "loss": 0.0, + "step": 60400 + }, + { + "epoch": 15.01, + "learning_rate": 1.3137568812492978e-05, + "loss": 0.0006, + "step": 60410 + }, + { + "epoch": 15.01, + "learning_rate": 1.3130547129536008e-05, + "loss": 0.0168, + "step": 60420 + }, + { + "epoch": 15.01, + "learning_rate": 1.3123525446579038e-05, + "loss": 0.1204, + "step": 60430 + }, + { + "epoch": 15.01, + "learning_rate": 1.3116503763622064e-05, + "loss": 0.0171, + "step": 60440 + }, + { + "epoch": 15.01, + "learning_rate": 1.3109482080665095e-05, + "loss": 0.0, + "step": 60450 + }, + { + "epoch": 15.01, + "learning_rate": 1.3102460397708124e-05, + "loss": 0.0, + "step": 60460 + }, + { + "epoch": 15.01, + "learning_rate": 1.3095438714751152e-05, + "loss": 0.0, + "step": 60470 + }, + { + "epoch": 15.01, + "learning_rate": 1.3088417031794181e-05, + "loss": 0.0, + "step": 60480 + }, + { + "epoch": 15.01, + "learning_rate": 1.308139534883721e-05, + "loss": 0.0, + "step": 60490 + }, + { + "epoch": 15.01, + "learning_rate": 1.3074373665880238e-05, + "loss": 0.0002, + "step": 60500 + }, + { + "epoch": 15.01, + "learning_rate": 1.3067351982923267e-05, + "loss": 0.0, + "step": 60510 + }, + { + "epoch": 15.01, + "learning_rate": 1.3060330299966298e-05, + "loss": 0.0004, + "step": 60520 + }, + { + "epoch": 15.02, + "learning_rate": 1.3053308617009324e-05, + "loss": 0.5345, + "step": 60530 + }, + { + "epoch": 15.02, + "learning_rate": 1.3046286934052354e-05, + "loss": 0.0485, + "step": 60540 + }, + { + "epoch": 15.02, + "learning_rate": 1.3039265251095384e-05, + "loss": 0.0, + "step": 60550 + }, + { + "epoch": 15.02, + "learning_rate": 1.3032243568138411e-05, + "loss": 0.0066, + "step": 60560 + }, + { + "epoch": 15.02, + "learning_rate": 1.302522188518144e-05, + "loss": 0.0004, + "step": 60570 + }, + { + "epoch": 15.02, + "learning_rate": 1.3018200202224471e-05, + "loss": 0.0091, + "step": 60580 + }, + { + "epoch": 15.02, + "learning_rate": 1.3011178519267497e-05, + "loss": 0.0014, + "step": 60590 + }, + { + "epoch": 15.02, + "learning_rate": 1.3004156836310528e-05, + "loss": 0.1114, + "step": 60600 + }, + { + "epoch": 15.02, + "learning_rate": 1.2997135153353557e-05, + "loss": 0.008, + "step": 60610 + }, + { + "epoch": 15.02, + "learning_rate": 1.2990113470396586e-05, + "loss": 0.0, + "step": 60620 + }, + { + "epoch": 15.02, + "learning_rate": 1.2983091787439614e-05, + "loss": 0.0, + "step": 60630 + }, + { + "epoch": 15.02, + "learning_rate": 1.2976070104482643e-05, + "loss": 0.4507, + "step": 60640 + }, + { + "epoch": 15.02, + "learning_rate": 1.2969048421525674e-05, + "loss": 0.0032, + "step": 60650 + }, + { + "epoch": 15.02, + "learning_rate": 1.29620267385687e-05, + "loss": 0.3757, + "step": 60660 + }, + { + "epoch": 15.02, + "learning_rate": 1.295500505561173e-05, + "loss": 0.0, + "step": 60670 + }, + { + "epoch": 15.02, + "learning_rate": 1.294798337265476e-05, + "loss": 0.0003, + "step": 60680 + }, + { + "epoch": 15.02, + "learning_rate": 1.2940961689697787e-05, + "loss": 0.0338, + "step": 60690 + }, + { + "epoch": 15.02, + "learning_rate": 1.2933940006740816e-05, + "loss": 0.0012, + "step": 60700 + }, + { + "epoch": 15.02, + "learning_rate": 1.2926918323783845e-05, + "loss": 0.0143, + "step": 60710 + }, + { + "epoch": 15.02, + "learning_rate": 1.2919896640826873e-05, + "loss": 0.2541, + "step": 60720 + }, + { + "epoch": 15.02, + "learning_rate": 1.2912874957869902e-05, + "loss": 0.0, + "step": 60730 + }, + { + "epoch": 15.02, + "learning_rate": 1.2905853274912933e-05, + "loss": 0.0, + "step": 60740 + }, + { + "epoch": 15.02, + "learning_rate": 1.2898831591955959e-05, + "loss": 0.0, + "step": 60750 + }, + { + "epoch": 15.02, + "learning_rate": 1.289180990899899e-05, + "loss": 0.0, + "step": 60760 + }, + { + "epoch": 15.02, + "learning_rate": 1.2884788226042019e-05, + "loss": 0.0, + "step": 60770 + }, + { + "epoch": 15.02, + "learning_rate": 1.2877766543085048e-05, + "loss": 0.0, + "step": 60780 + }, + { + "epoch": 15.02, + "learning_rate": 1.2870744860128076e-05, + "loss": 0.1443, + "step": 60790 + }, + { + "epoch": 15.02, + "learning_rate": 1.2863723177171105e-05, + "loss": 0.0, + "step": 60800 + }, + { + "epoch": 15.02, + "learning_rate": 1.2856701494214136e-05, + "loss": 0.0007, + "step": 60810 + }, + { + "epoch": 15.02, + "learning_rate": 1.2849679811257161e-05, + "loss": 0.0, + "step": 60820 + }, + { + "epoch": 15.02, + "learning_rate": 1.2842658128300192e-05, + "loss": 0.3606, + "step": 60830 + }, + { + "epoch": 15.02, + "learning_rate": 1.2835636445343222e-05, + "loss": 0.0001, + "step": 60840 + }, + { + "epoch": 15.02, + "learning_rate": 1.2828614762386249e-05, + "loss": 0.0115, + "step": 60850 + }, + { + "epoch": 15.02, + "learning_rate": 1.2821593079429278e-05, + "loss": 0.0, + "step": 60860 + }, + { + "epoch": 15.02, + "learning_rate": 1.2814571396472307e-05, + "loss": 0.0007, + "step": 60870 + }, + { + "epoch": 15.02, + "learning_rate": 1.2807549713515335e-05, + "loss": 0.0, + "step": 60880 + }, + { + "epoch": 15.02, + "learning_rate": 1.2800528030558364e-05, + "loss": 0.0001, + "step": 60890 + }, + { + "epoch": 15.02, + "learning_rate": 1.2793506347601395e-05, + "loss": 0.0002, + "step": 60900 + }, + { + "epoch": 15.02, + "learning_rate": 1.278648466464442e-05, + "loss": 0.0046, + "step": 60910 + }, + { + "epoch": 15.02, + "learning_rate": 1.2779462981687452e-05, + "loss": 0.0, + "step": 60920 + }, + { + "epoch": 15.02, + "learning_rate": 1.277244129873048e-05, + "loss": 0.0001, + "step": 60930 + }, + { + "epoch": 15.02, + "learning_rate": 1.2765419615773508e-05, + "loss": 0.0, + "step": 60940 + }, + { + "epoch": 15.02, + "learning_rate": 1.2758397932816537e-05, + "loss": 0.0009, + "step": 60950 + }, + { + "epoch": 15.02, + "learning_rate": 1.2751376249859567e-05, + "loss": 0.0001, + "step": 60960 + }, + { + "epoch": 15.02, + "learning_rate": 1.2744354566902598e-05, + "loss": 0.0, + "step": 60970 + }, + { + "epoch": 15.02, + "learning_rate": 1.2737332883945623e-05, + "loss": 0.0543, + "step": 60980 + }, + { + "epoch": 15.02, + "learning_rate": 1.2730311200988654e-05, + "loss": 0.3729, + "step": 60990 + }, + { + "epoch": 15.02, + "learning_rate": 1.2723289518031683e-05, + "loss": 0.0, + "step": 61000 + }, + { + "epoch": 15.02, + "learning_rate": 1.2716267835074711e-05, + "loss": 0.368, + "step": 61010 + }, + { + "epoch": 15.02, + "learning_rate": 1.270924615211774e-05, + "loss": 0.0013, + "step": 61020 + }, + { + "epoch": 15.02, + "learning_rate": 1.270222446916077e-05, + "loss": 0.0002, + "step": 61030 + }, + { + "epoch": 15.02, + "learning_rate": 1.2695202786203797e-05, + "loss": 0.0, + "step": 61040 + }, + { + "epoch": 15.02, + "learning_rate": 1.2688181103246826e-05, + "loss": 0.2367, + "step": 61050 + }, + { + "epoch": 15.02, + "learning_rate": 1.2681159420289857e-05, + "loss": 0.0, + "step": 61060 + }, + { + "epoch": 15.02, + "learning_rate": 1.2674137737332884e-05, + "loss": 0.0, + "step": 61070 + }, + { + "epoch": 15.02, + "learning_rate": 1.2667116054375913e-05, + "loss": 0.01, + "step": 61080 + }, + { + "epoch": 15.02, + "learning_rate": 1.2660094371418943e-05, + "loss": 0.0127, + "step": 61090 + }, + { + "epoch": 15.02, + "learning_rate": 1.265307268846197e-05, + "loss": 0.322, + "step": 61100 + }, + { + "epoch": 15.02, + "learning_rate": 1.2646051005505e-05, + "loss": 0.0, + "step": 61110 + }, + { + "epoch": 15.02, + "learning_rate": 1.263902932254803e-05, + "loss": 0.0, + "step": 61120 + }, + { + "epoch": 15.02, + "learning_rate": 1.263200763959106e-05, + "loss": 0.0002, + "step": 61130 + }, + { + "epoch": 15.02, + "learning_rate": 1.2624985956634087e-05, + "loss": 0.0, + "step": 61140 + }, + { + "epoch": 15.02, + "learning_rate": 1.2617964273677116e-05, + "loss": 0.0, + "step": 61150 + }, + { + "epoch": 15.02, + "learning_rate": 1.2610942590720145e-05, + "loss": 0.0, + "step": 61160 + }, + { + "epoch": 15.02, + "learning_rate": 1.2603920907763173e-05, + "loss": 0.0, + "step": 61170 + }, + { + "epoch": 15.02, + "learning_rate": 1.2596899224806202e-05, + "loss": 0.0004, + "step": 61180 + }, + { + "epoch": 15.02, + "learning_rate": 1.2589877541849233e-05, + "loss": 0.0, + "step": 61190 + }, + { + "epoch": 15.02, + "learning_rate": 1.2582855858892259e-05, + "loss": 0.0003, + "step": 61200 + }, + { + "epoch": 15.02, + "learning_rate": 1.257583417593529e-05, + "loss": 0.0001, + "step": 61210 + }, + { + "epoch": 15.02, + "learning_rate": 1.2568812492978319e-05, + "loss": 0.0, + "step": 61220 + }, + { + "epoch": 15.02, + "learning_rate": 1.2561790810021346e-05, + "loss": 0.0004, + "step": 61230 + }, + { + "epoch": 15.02, + "learning_rate": 1.2554769127064375e-05, + "loss": 0.0001, + "step": 61240 + }, + { + "epoch": 15.02, + "learning_rate": 1.2547747444107405e-05, + "loss": 0.0002, + "step": 61250 + }, + { + "epoch": 15.02, + "learning_rate": 1.2540725761150432e-05, + "loss": 0.0, + "step": 61260 + }, + { + "epoch": 15.02, + "learning_rate": 1.2533704078193461e-05, + "loss": 0.0002, + "step": 61270 + }, + { + "epoch": 15.02, + "learning_rate": 1.2526682395236492e-05, + "loss": 0.0224, + "step": 61280 + }, + { + "epoch": 15.02, + "learning_rate": 1.2519660712279518e-05, + "loss": 0.0, + "step": 61290 + }, + { + "epoch": 15.02, + "learning_rate": 1.2512639029322549e-05, + "loss": 0.0229, + "step": 61300 + }, + { + "epoch": 15.02, + "learning_rate": 1.2505617346365578e-05, + "loss": 0.0006, + "step": 61310 + }, + { + "epoch": 15.03, + "learning_rate": 1.2498595663408607e-05, + "loss": 0.0008, + "step": 61320 + }, + { + "epoch": 15.03, + "learning_rate": 1.2491573980451635e-05, + "loss": 0.0, + "step": 61330 + }, + { + "epoch": 15.03, + "learning_rate": 1.2484552297494664e-05, + "loss": 0.0, + "step": 61340 + }, + { + "epoch": 15.03, + "learning_rate": 1.2477530614537693e-05, + "loss": 0.0004, + "step": 61350 + }, + { + "epoch": 15.03, + "learning_rate": 1.2470508931580722e-05, + "loss": 0.0001, + "step": 61360 + }, + { + "epoch": 15.03, + "learning_rate": 1.2463487248623751e-05, + "loss": 0.0001, + "step": 61370 + }, + { + "epoch": 15.03, + "learning_rate": 1.2456465565666779e-05, + "loss": 0.0, + "step": 61380 + }, + { + "epoch": 15.03, + "learning_rate": 1.244944388270981e-05, + "loss": 0.0, + "step": 61390 + }, + { + "epoch": 15.03, + "learning_rate": 1.2442422199752837e-05, + "loss": 0.0, + "step": 61400 + }, + { + "epoch": 15.03, + "learning_rate": 1.2435400516795866e-05, + "loss": 0.0, + "step": 61410 + }, + { + "epoch": 15.03, + "learning_rate": 1.2428378833838896e-05, + "loss": 0.0, + "step": 61420 + }, + { + "epoch": 15.03, + "learning_rate": 1.2421357150881923e-05, + "loss": 0.0, + "step": 61430 + }, + { + "epoch": 15.03, + "learning_rate": 1.2414335467924954e-05, + "loss": 0.0, + "step": 61440 + }, + { + "epoch": 15.03, + "learning_rate": 1.2407313784967981e-05, + "loss": 0.0, + "step": 61450 + }, + { + "epoch": 15.03, + "learning_rate": 1.240029210201101e-05, + "loss": 0.0, + "step": 61460 + }, + { + "epoch": 15.03, + "learning_rate": 1.239327041905404e-05, + "loss": 0.0806, + "step": 61470 + }, + { + "epoch": 15.03, + "learning_rate": 1.2386248736097069e-05, + "loss": 0.0468, + "step": 61480 + }, + { + "epoch": 15.03, + "learning_rate": 1.2379227053140096e-05, + "loss": 0.0, + "step": 61490 + }, + { + "epoch": 15.03, + "learning_rate": 1.2372205370183126e-05, + "loss": 0.0035, + "step": 61500 + }, + { + "epoch": 15.03, + "learning_rate": 1.2365183687226155e-05, + "loss": 0.0006, + "step": 61510 + }, + { + "epoch": 15.03, + "learning_rate": 1.2358162004269184e-05, + "loss": 0.0, + "step": 61520 + }, + { + "epoch": 15.03, + "learning_rate": 1.2351140321312213e-05, + "loss": 0.0353, + "step": 61530 + }, + { + "epoch": 15.03, + "learning_rate": 1.234411863835524e-05, + "loss": 0.0, + "step": 61540 + }, + { + "epoch": 15.03, + "learning_rate": 1.2337096955398272e-05, + "loss": 0.0, + "step": 61550 + }, + { + "epoch": 15.03, + "learning_rate": 1.2330075272441299e-05, + "loss": 0.0003, + "step": 61560 + }, + { + "epoch": 15.03, + "learning_rate": 1.2323053589484328e-05, + "loss": 0.0017, + "step": 61570 + }, + { + "epoch": 15.03, + "learning_rate": 1.2316031906527357e-05, + "loss": 0.0003, + "step": 61580 + }, + { + "epoch": 15.03, + "learning_rate": 1.2309010223570387e-05, + "loss": 0.0004, + "step": 61590 + }, + { + "epoch": 15.03, + "learning_rate": 1.2301988540613414e-05, + "loss": 0.0001, + "step": 61600 + }, + { + "epoch": 15.03, + "learning_rate": 1.2294966857656443e-05, + "loss": 0.0001, + "step": 61610 + }, + { + "epoch": 15.03, + "learning_rate": 1.2287945174699472e-05, + "loss": 0.0001, + "step": 61620 + }, + { + "epoch": 15.03, + "learning_rate": 1.2280923491742502e-05, + "loss": 0.0, + "step": 61630 + }, + { + "epoch": 15.03, + "learning_rate": 1.2273901808785531e-05, + "loss": 0.0, + "step": 61640 + }, + { + "epoch": 15.03, + "learning_rate": 1.2266880125828558e-05, + "loss": 0.0, + "step": 61650 + }, + { + "epoch": 15.03, + "learning_rate": 1.225985844287159e-05, + "loss": 0.0, + "step": 61660 + }, + { + "epoch": 15.03, + "learning_rate": 1.2252836759914617e-05, + "loss": 0.0002, + "step": 61670 + }, + { + "epoch": 15.03, + "learning_rate": 1.2245815076957646e-05, + "loss": 0.0052, + "step": 61680 + }, + { + "epoch": 15.03, + "learning_rate": 1.2238793394000675e-05, + "loss": 0.0001, + "step": 61690 + }, + { + "epoch": 15.03, + "learning_rate": 1.2231771711043703e-05, + "loss": 0.0, + "step": 61700 + }, + { + "epoch": 15.03, + "learning_rate": 1.2224750028086733e-05, + "loss": 0.0003, + "step": 61710 + }, + { + "epoch": 15.03, + "learning_rate": 1.2217728345129761e-05, + "loss": 0.0, + "step": 61720 + }, + { + "epoch": 15.03, + "learning_rate": 1.221070666217279e-05, + "loss": 0.1739, + "step": 61730 + }, + { + "epoch": 15.03, + "learning_rate": 1.220368497921582e-05, + "loss": 0.0, + "step": 61740 + }, + { + "epoch": 15.03, + "learning_rate": 1.2196663296258848e-05, + "loss": 0.124, + "step": 61750 + }, + { + "epoch": 15.03, + "learning_rate": 1.2189641613301876e-05, + "loss": 0.0002, + "step": 61760 + }, + { + "epoch": 15.03, + "learning_rate": 1.2182619930344905e-05, + "loss": 0.0, + "step": 61770 + }, + { + "epoch": 15.03, + "learning_rate": 1.2175598247387934e-05, + "loss": 0.0, + "step": 61780 + }, + { + "epoch": 15.03, + "learning_rate": 1.2168576564430964e-05, + "loss": 0.0006, + "step": 61790 + }, + { + "epoch": 15.03, + "learning_rate": 1.2161554881473993e-05, + "loss": 0.0, + "step": 61800 + }, + { + "epoch": 15.03, + "learning_rate": 1.215453319851702e-05, + "loss": 0.0, + "step": 61810 + }, + { + "epoch": 15.03, + "learning_rate": 1.2147511515560051e-05, + "loss": 0.0, + "step": 61820 + }, + { + "epoch": 15.03, + "learning_rate": 1.2140489832603079e-05, + "loss": 0.0042, + "step": 61830 + }, + { + "epoch": 15.03, + "learning_rate": 1.2133468149646108e-05, + "loss": 0.0, + "step": 61840 + }, + { + "epoch": 15.03, + "learning_rate": 1.2126446466689137e-05, + "loss": 0.0, + "step": 61850 + }, + { + "epoch": 15.03, + "learning_rate": 1.2119424783732166e-05, + "loss": 0.0018, + "step": 61860 + }, + { + "epoch": 15.03, + "learning_rate": 1.2112403100775194e-05, + "loss": 0.1049, + "step": 61870 + }, + { + "epoch": 15.03, + "learning_rate": 1.2105381417818223e-05, + "loss": 0.0, + "step": 61880 + }, + { + "epoch": 15.03, + "learning_rate": 1.2098359734861252e-05, + "loss": 0.0, + "step": 61890 + }, + { + "epoch": 15.03, + "learning_rate": 1.2091338051904281e-05, + "loss": 0.0, + "step": 61900 + }, + { + "epoch": 15.03, + "learning_rate": 1.208431636894731e-05, + "loss": 0.0, + "step": 61910 + }, + { + "epoch": 15.03, + "learning_rate": 1.2077294685990338e-05, + "loss": 0.0, + "step": 61920 + }, + { + "epoch": 15.03, + "learning_rate": 1.2070273003033369e-05, + "loss": 0.0001, + "step": 61930 + }, + { + "epoch": 15.03, + "learning_rate": 1.2063251320076396e-05, + "loss": 0.0001, + "step": 61940 + }, + { + "epoch": 15.03, + "learning_rate": 1.2056229637119425e-05, + "loss": 0.0021, + "step": 61950 + }, + { + "epoch": 15.03, + "learning_rate": 1.2049207954162455e-05, + "loss": 0.0, + "step": 61960 + }, + { + "epoch": 15.03, + "learning_rate": 1.2042186271205482e-05, + "loss": 0.0, + "step": 61970 + }, + { + "epoch": 15.03, + "learning_rate": 1.2035164588248513e-05, + "loss": 0.0006, + "step": 61980 + }, + { + "epoch": 15.03, + "learning_rate": 1.202814290529154e-05, + "loss": 0.0, + "step": 61990 + }, + { + "epoch": 15.03, + "learning_rate": 1.202112122233457e-05, + "loss": 0.0, + "step": 62000 + }, + { + "epoch": 15.03, + "learning_rate": 1.2014099539377599e-05, + "loss": 0.0, + "step": 62010 + }, + { + "epoch": 15.03, + "learning_rate": 1.2007077856420628e-05, + "loss": 0.0, + "step": 62020 + }, + { + "epoch": 15.03, + "learning_rate": 1.2000056173463656e-05, + "loss": 0.0, + "step": 62030 + }, + { + "epoch": 15.03, + "learning_rate": 1.1993034490506685e-05, + "loss": 0.0, + "step": 62040 + }, + { + "epoch": 15.03, + "learning_rate": 1.1986012807549714e-05, + "loss": 0.0, + "step": 62050 + }, + { + "epoch": 15.03, + "learning_rate": 1.1978991124592743e-05, + "loss": 0.0, + "step": 62060 + }, + { + "epoch": 15.03, + "learning_rate": 1.1971969441635772e-05, + "loss": 0.0, + "step": 62070 + }, + { + "epoch": 15.03, + "learning_rate": 1.19649477586788e-05, + "loss": 0.0, + "step": 62080 + }, + { + "epoch": 15.03, + "learning_rate": 1.195792607572183e-05, + "loss": 0.0426, + "step": 62090 + }, + { + "epoch": 15.03, + "learning_rate": 1.1950904392764858e-05, + "loss": 0.0, + "step": 62100 + }, + { + "epoch": 15.04, + "learning_rate": 1.1943882709807887e-05, + "loss": 0.0001, + "step": 62110 + }, + { + "epoch": 15.04, + "learning_rate": 1.1936861026850916e-05, + "loss": 0.0, + "step": 62120 + }, + { + "epoch": 15.04, + "learning_rate": 1.1929839343893946e-05, + "loss": 0.0, + "step": 62130 + }, + { + "epoch": 15.04, + "learning_rate": 1.1922817660936975e-05, + "loss": 0.0014, + "step": 62140 + }, + { + "epoch": 15.04, + "learning_rate": 1.1915795977980002e-05, + "loss": 0.0065, + "step": 62150 + }, + { + "epoch": 15.04, + "learning_rate": 1.1908774295023032e-05, + "loss": 0.0, + "step": 62160 + }, + { + "epoch": 15.04, + "learning_rate": 1.190175261206606e-05, + "loss": 0.0, + "step": 62170 + }, + { + "epoch": 15.04, + "learning_rate": 1.189473092910909e-05, + "loss": 0.0001, + "step": 62180 + }, + { + "epoch": 15.04, + "learning_rate": 1.1887709246152117e-05, + "loss": 0.0, + "step": 62190 + }, + { + "epoch": 15.04, + "learning_rate": 1.1880687563195148e-05, + "loss": 0.0, + "step": 62200 + }, + { + "epoch": 15.04, + "learning_rate": 1.1873665880238176e-05, + "loss": 0.0, + "step": 62210 + }, + { + "epoch": 15.04, + "learning_rate": 1.1866644197281205e-05, + "loss": 0.0, + "step": 62220 + }, + { + "epoch": 15.04, + "learning_rate": 1.1859622514324234e-05, + "loss": 0.0, + "step": 62230 + }, + { + "epoch": 15.04, + "learning_rate": 1.1852600831367262e-05, + "loss": 0.1077, + "step": 62240 + }, + { + "epoch": 15.04, + "learning_rate": 1.1845579148410292e-05, + "loss": 0.0157, + "step": 62250 + }, + { + "epoch": 15.04, + "learning_rate": 1.183855746545332e-05, + "loss": 0.0247, + "step": 62260 + }, + { + "epoch": 15.04, + "learning_rate": 1.183153578249635e-05, + "loss": 0.0, + "step": 62270 + }, + { + "epoch": 15.04, + "learning_rate": 1.1824514099539378e-05, + "loss": 0.0, + "step": 62280 + }, + { + "epoch": 15.04, + "learning_rate": 1.1817492416582408e-05, + "loss": 0.0, + "step": 62290 + }, + { + "epoch": 15.04, + "learning_rate": 1.1810470733625435e-05, + "loss": 0.0, + "step": 62300 + }, + { + "epoch": 15.04, + "learning_rate": 1.1803449050668464e-05, + "loss": 0.0461, + "step": 62310 + }, + { + "epoch": 15.04, + "learning_rate": 1.1796427367711493e-05, + "loss": 0.0, + "step": 62320 + }, + { + "epoch": 15.04, + "learning_rate": 1.1789405684754523e-05, + "loss": 0.0, + "step": 62330 + }, + { + "epoch": 15.04, + "learning_rate": 1.1782384001797552e-05, + "loss": 0.19, + "step": 62340 + }, + { + "epoch": 15.04, + "learning_rate": 1.177536231884058e-05, + "loss": 0.0002, + "step": 62350 + }, + { + "epoch": 15.04, + "learning_rate": 1.176834063588361e-05, + "loss": 0.0, + "step": 62360 + }, + { + "epoch": 15.04, + "learning_rate": 1.1761318952926638e-05, + "loss": 0.0, + "step": 62370 + }, + { + "epoch": 15.04, + "learning_rate": 1.1754297269969667e-05, + "loss": 0.0, + "step": 62380 + }, + { + "epoch": 15.04, + "learning_rate": 1.1747275587012696e-05, + "loss": 0.0, + "step": 62390 + }, + { + "epoch": 15.04, + "learning_rate": 1.1740253904055725e-05, + "loss": 0.0, + "step": 62400 + }, + { + "epoch": 15.04, + "learning_rate": 1.1733232221098754e-05, + "loss": 0.0, + "step": 62410 + }, + { + "epoch": 15.04, + "learning_rate": 1.1726210538141782e-05, + "loss": 0.0036, + "step": 62420 + }, + { + "epoch": 15.04, + "learning_rate": 1.1719188855184811e-05, + "loss": 0.0, + "step": 62430 + }, + { + "epoch": 15.04, + "learning_rate": 1.171216717222784e-05, + "loss": 0.0, + "step": 62440 + }, + { + "epoch": 15.04, + "learning_rate": 1.170514548927087e-05, + "loss": 0.0, + "step": 62450 + }, + { + "epoch": 15.04, + "learning_rate": 1.1698123806313897e-05, + "loss": 0.2186, + "step": 62460 + }, + { + "epoch": 15.04, + "learning_rate": 1.1691102123356928e-05, + "loss": 0.0004, + "step": 62470 + }, + { + "epoch": 15.04, + "learning_rate": 1.1684080440399955e-05, + "loss": 0.0, + "step": 62480 + }, + { + "epoch": 15.04, + "learning_rate": 1.1677058757442984e-05, + "loss": 0.0, + "step": 62490 + }, + { + "epoch": 15.04, + "learning_rate": 1.1670037074486014e-05, + "loss": 0.0, + "step": 62500 + }, + { + "epoch": 15.04, + "learning_rate": 1.1663015391529041e-05, + "loss": 0.0002, + "step": 62510 + }, + { + "epoch": 15.04, + "learning_rate": 1.1655993708572072e-05, + "loss": 0.0007, + "step": 62520 + }, + { + "epoch": 15.04, + "learning_rate": 1.16489720256151e-05, + "loss": 0.0, + "step": 62530 + }, + { + "epoch": 15.04, + "learning_rate": 1.1641950342658129e-05, + "loss": 0.0027, + "step": 62540 + }, + { + "epoch": 15.04, + "learning_rate": 1.1634928659701158e-05, + "loss": 0.0, + "step": 62550 + }, + { + "epoch": 15.04, + "learning_rate": 1.1627906976744187e-05, + "loss": 0.0, + "step": 62560 + }, + { + "epoch": 15.04, + "learning_rate": 1.1620885293787215e-05, + "loss": 0.0, + "step": 62570 + }, + { + "epoch": 15.04, + "learning_rate": 1.1613863610830244e-05, + "loss": 0.0002, + "step": 62580 + }, + { + "epoch": 15.04, + "learning_rate": 1.1606841927873273e-05, + "loss": 0.0, + "step": 62590 + }, + { + "epoch": 15.04, + "learning_rate": 1.1599820244916302e-05, + "loss": 0.0, + "step": 62600 + }, + { + "epoch": 15.04, + "learning_rate": 1.1592798561959331e-05, + "loss": 0.413, + "step": 62610 + }, + { + "epoch": 15.04, + "learning_rate": 1.1585776879002359e-05, + "loss": 0.0324, + "step": 62620 + }, + { + "epoch": 15.04, + "learning_rate": 1.157875519604539e-05, + "loss": 0.0116, + "step": 62630 + }, + { + "epoch": 15.04, + "learning_rate": 1.1571733513088417e-05, + "loss": 0.0, + "step": 62640 + }, + { + "epoch": 15.04, + "learning_rate": 1.1564711830131446e-05, + "loss": 0.0002, + "step": 62650 + }, + { + "epoch": 15.04, + "learning_rate": 1.1557690147174475e-05, + "loss": 0.0, + "step": 62660 + }, + { + "epoch": 15.04, + "learning_rate": 1.1550668464217505e-05, + "loss": 0.0003, + "step": 62670 + }, + { + "epoch": 15.04, + "learning_rate": 1.1543646781260534e-05, + "loss": 0.0001, + "step": 62680 + }, + { + "epoch": 15.04, + "learning_rate": 1.1536625098303561e-05, + "loss": 0.0, + "step": 62690 + }, + { + "epoch": 15.04, + "learning_rate": 1.152960341534659e-05, + "loss": 0.0, + "step": 62700 + }, + { + "epoch": 15.04, + "learning_rate": 1.152258173238962e-05, + "loss": 0.0003, + "step": 62710 + }, + { + "epoch": 15.04, + "learning_rate": 1.1515560049432649e-05, + "loss": 0.0042, + "step": 62720 + }, + { + "epoch": 15.04, + "learning_rate": 1.1508538366475676e-05, + "loss": 0.0, + "step": 62730 + }, + { + "epoch": 15.04, + "learning_rate": 1.1501516683518707e-05, + "loss": 0.076, + "step": 62740 + }, + { + "epoch": 15.04, + "learning_rate": 1.1494495000561735e-05, + "loss": 0.0, + "step": 62750 + }, + { + "epoch": 15.04, + "learning_rate": 1.1487473317604764e-05, + "loss": 0.0, + "step": 62760 + }, + { + "epoch": 15.04, + "learning_rate": 1.1480451634647793e-05, + "loss": 0.0022, + "step": 62770 + }, + { + "epoch": 15.04, + "learning_rate": 1.147342995169082e-05, + "loss": 0.0, + "step": 62780 + }, + { + "epoch": 15.04, + "learning_rate": 1.1466408268733852e-05, + "loss": 0.0, + "step": 62790 + }, + { + "epoch": 15.04, + "learning_rate": 1.1459386585776879e-05, + "loss": 0.0, + "step": 62800 + }, + { + "epoch": 15.04, + "learning_rate": 1.1452364902819908e-05, + "loss": 0.0, + "step": 62810 + }, + { + "epoch": 15.04, + "learning_rate": 1.1445343219862937e-05, + "loss": 0.0, + "step": 62820 + }, + { + "epoch": 15.04, + "learning_rate": 1.1438321536905967e-05, + "loss": 0.112, + "step": 62830 + }, + { + "epoch": 15.04, + "learning_rate": 1.1431299853948996e-05, + "loss": 0.0, + "step": 62840 + }, + { + "epoch": 15.04, + "learning_rate": 1.1424278170992025e-05, + "loss": 0.0008, + "step": 62850 + }, + { + "epoch": 15.04, + "learning_rate": 1.1417256488035052e-05, + "loss": 0.0008, + "step": 62860 + }, + { + "epoch": 15.04, + "learning_rate": 1.1410234805078082e-05, + "loss": 0.0001, + "step": 62870 + }, + { + "epoch": 15.04, + "learning_rate": 1.140321312212111e-05, + "loss": 0.0, + "step": 62880 + }, + { + "epoch": 15.04, + "learning_rate": 1.1396191439164138e-05, + "loss": 0.0, + "step": 62890 + }, + { + "epoch": 15.04, + "learning_rate": 1.1389169756207169e-05, + "loss": 0.0001, + "step": 62900 + }, + { + "epoch": 15.05, + "learning_rate": 1.1382148073250197e-05, + "loss": 0.0, + "step": 62910 + }, + { + "epoch": 15.05, + "learning_rate": 1.1375126390293226e-05, + "loss": 0.0082, + "step": 62920 + }, + { + "epoch": 15.05, + "learning_rate": 1.1368104707336255e-05, + "loss": 0.0, + "step": 62930 + }, + { + "epoch": 15.05, + "learning_rate": 1.1361083024379284e-05, + "loss": 0.0003, + "step": 62940 + }, + { + "epoch": 15.05, + "learning_rate": 1.1354061341422313e-05, + "loss": 0.0, + "step": 62950 + }, + { + "epoch": 15.05, + "learning_rate": 1.1347039658465341e-05, + "loss": 0.0, + "step": 62960 + }, + { + "epoch": 15.05, + "learning_rate": 1.134001797550837e-05, + "loss": 0.0, + "step": 62970 + }, + { + "epoch": 15.05, + "learning_rate": 1.13329962925514e-05, + "loss": 0.0, + "step": 62980 + }, + { + "epoch": 15.05, + "learning_rate": 1.1325974609594428e-05, + "loss": 0.0001, + "step": 62990 + }, + { + "epoch": 15.05, + "learning_rate": 1.1318952926637456e-05, + "loss": 0.0, + "step": 63000 + }, + { + "epoch": 15.05, + "learning_rate": 1.1311931243680487e-05, + "loss": 0.0, + "step": 63010 + }, + { + "epoch": 15.05, + "learning_rate": 1.1304909560723514e-05, + "loss": 0.0, + "step": 63020 + }, + { + "epoch": 15.05, + "learning_rate": 1.1297887877766543e-05, + "loss": 0.0001, + "step": 63030 + }, + { + "epoch": 15.05, + "learning_rate": 1.1290866194809573e-05, + "loss": 0.0, + "step": 63040 + }, + { + "epoch": 15.05, + "learning_rate": 1.12838445118526e-05, + "loss": 0.0, + "step": 63050 + }, + { + "epoch": 15.05, + "learning_rate": 1.1276822828895631e-05, + "loss": 0.0, + "step": 63060 + }, + { + "epoch": 15.05, + "learning_rate": 1.1269801145938659e-05, + "loss": 0.2037, + "step": 63070 + }, + { + "epoch": 15.05, + "learning_rate": 1.1262779462981688e-05, + "loss": 0.0, + "step": 63080 + }, + { + "epoch": 15.05, + "learning_rate": 1.1255757780024717e-05, + "loss": 0.0, + "step": 63090 + }, + { + "epoch": 15.05, + "learning_rate": 1.1248736097067746e-05, + "loss": 0.0, + "step": 63100 + }, + { + "epoch": 15.05, + "learning_rate": 1.1241714414110775e-05, + "loss": 0.0, + "step": 63110 + }, + { + "epoch": 15.05, + "learning_rate": 1.1234692731153804e-05, + "loss": 0.009, + "step": 63120 + }, + { + "epoch": 15.05, + "learning_rate": 1.1227671048196832e-05, + "loss": 0.0004, + "step": 63130 + }, + { + "epoch": 15.05, + "learning_rate": 1.1220649365239861e-05, + "loss": 0.0, + "step": 63140 + }, + { + "epoch": 15.05, + "learning_rate": 1.121362768228289e-05, + "loss": 0.0, + "step": 63150 + }, + { + "epoch": 15.05, + "learning_rate": 1.1206605999325918e-05, + "loss": 0.0002, + "step": 63160 + }, + { + "epoch": 15.05, + "learning_rate": 1.1199584316368949e-05, + "loss": 0.0001, + "step": 63170 + }, + { + "epoch": 15.05, + "learning_rate": 1.1192562633411976e-05, + "loss": 0.2093, + "step": 63180 + }, + { + "epoch": 15.05, + "learning_rate": 1.1185540950455007e-05, + "loss": 0.0, + "step": 63190 + }, + { + "epoch": 15.05, + "learning_rate": 1.1178519267498035e-05, + "loss": 0.1589, + "step": 63200 + }, + { + "epoch": 15.05, + "learning_rate": 1.1171497584541064e-05, + "loss": 0.127, + "step": 63210 + }, + { + "epoch": 15.05, + "learning_rate": 1.1164475901584093e-05, + "loss": 0.0, + "step": 63220 + }, + { + "epoch": 15.05, + "learning_rate": 1.115745421862712e-05, + "loss": 0.0, + "step": 63230 + }, + { + "epoch": 15.05, + "learning_rate": 1.115043253567015e-05, + "loss": 0.0, + "step": 63240 + }, + { + "epoch": 15.05, + "learning_rate": 1.1143410852713179e-05, + "loss": 0.0, + "step": 63250 + }, + { + "epoch": 15.05, + "learning_rate": 1.1136389169756208e-05, + "loss": 0.0, + "step": 63260 + }, + { + "epoch": 15.05, + "learning_rate": 1.1129367486799235e-05, + "loss": 0.1117, + "step": 63270 + }, + { + "epoch": 15.05, + "learning_rate": 1.1122345803842266e-05, + "loss": 0.0002, + "step": 63280 + }, + { + "epoch": 15.05, + "learning_rate": 1.1115324120885294e-05, + "loss": 0.0, + "step": 63290 + }, + { + "epoch": 15.05, + "eval_accuracy": 0.766739775512837, + "eval_loss": 1.673811435699463, + "eval_runtime": 1264.3549, + "eval_samples_per_second": 6.13, + "eval_steps_per_second": 3.066, + "step": 63296 + }, + { + "epoch": 16.0, + "learning_rate": 1.1108302437928323e-05, + "loss": 0.0001, + "step": 63300 + }, + { + "epoch": 16.0, + "learning_rate": 1.1101280754971352e-05, + "loss": 0.0, + "step": 63310 + }, + { + "epoch": 16.0, + "learning_rate": 1.109425907201438e-05, + "loss": 0.0, + "step": 63320 + }, + { + "epoch": 16.0, + "learning_rate": 1.108723738905741e-05, + "loss": 0.0, + "step": 63330 + }, + { + "epoch": 16.0, + "learning_rate": 1.1080215706100438e-05, + "loss": 0.0, + "step": 63340 + }, + { + "epoch": 16.0, + "learning_rate": 1.1073194023143467e-05, + "loss": 0.0001, + "step": 63350 + }, + { + "epoch": 16.0, + "learning_rate": 1.1066172340186496e-05, + "loss": 0.0, + "step": 63360 + }, + { + "epoch": 16.0, + "learning_rate": 1.1059150657229526e-05, + "loss": 0.0, + "step": 63370 + }, + { + "epoch": 16.0, + "learning_rate": 1.1052128974272555e-05, + "loss": 0.0004, + "step": 63380 + }, + { + "epoch": 16.0, + "learning_rate": 1.1045107291315584e-05, + "loss": 0.1237, + "step": 63390 + }, + { + "epoch": 16.0, + "learning_rate": 1.1038085608358611e-05, + "loss": 0.0, + "step": 63400 + }, + { + "epoch": 16.0, + "learning_rate": 1.103106392540164e-05, + "loss": 0.0001, + "step": 63410 + }, + { + "epoch": 16.0, + "learning_rate": 1.102404224244467e-05, + "loss": 0.0, + "step": 63420 + }, + { + "epoch": 16.0, + "learning_rate": 1.1017020559487697e-05, + "loss": 0.0536, + "step": 63430 + }, + { + "epoch": 16.0, + "learning_rate": 1.1009998876530728e-05, + "loss": 0.0001, + "step": 63440 + }, + { + "epoch": 16.0, + "learning_rate": 1.1002977193573756e-05, + "loss": 0.0, + "step": 63450 + }, + { + "epoch": 16.0, + "learning_rate": 1.0995955510616787e-05, + "loss": 0.0, + "step": 63460 + }, + { + "epoch": 16.0, + "learning_rate": 1.0988933827659814e-05, + "loss": 0.0, + "step": 63470 + }, + { + "epoch": 16.0, + "learning_rate": 1.0981912144702843e-05, + "loss": 0.4758, + "step": 63480 + }, + { + "epoch": 16.0, + "learning_rate": 1.0974890461745872e-05, + "loss": 0.0, + "step": 63490 + }, + { + "epoch": 16.0, + "learning_rate": 1.09678687787889e-05, + "loss": 0.0001, + "step": 63500 + }, + { + "epoch": 16.0, + "learning_rate": 1.0960847095831929e-05, + "loss": 0.0, + "step": 63510 + }, + { + "epoch": 16.0, + "learning_rate": 1.0953825412874958e-05, + "loss": 0.0, + "step": 63520 + }, + { + "epoch": 16.0, + "learning_rate": 1.0946803729917987e-05, + "loss": 0.0063, + "step": 63530 + }, + { + "epoch": 16.0, + "learning_rate": 1.0939782046961017e-05, + "loss": 0.0, + "step": 63540 + }, + { + "epoch": 16.0, + "learning_rate": 1.0932760364004046e-05, + "loss": 0.0, + "step": 63550 + }, + { + "epoch": 16.0, + "learning_rate": 1.0925738681047073e-05, + "loss": 0.0039, + "step": 63560 + }, + { + "epoch": 16.0, + "learning_rate": 1.0918716998090102e-05, + "loss": 0.0, + "step": 63570 + }, + { + "epoch": 16.0, + "learning_rate": 1.0911695315133132e-05, + "loss": 0.0, + "step": 63580 + }, + { + "epoch": 16.0, + "learning_rate": 1.090467363217616e-05, + "loss": 0.0, + "step": 63590 + }, + { + "epoch": 16.0, + "learning_rate": 1.089765194921919e-05, + "loss": 0.0002, + "step": 63600 + }, + { + "epoch": 16.0, + "learning_rate": 1.0890630266262218e-05, + "loss": 0.0001, + "step": 63610 + }, + { + "epoch": 16.0, + "learning_rate": 1.0883608583305247e-05, + "loss": 0.0, + "step": 63620 + }, + { + "epoch": 16.0, + "learning_rate": 1.0876586900348276e-05, + "loss": 0.0016, + "step": 63630 + }, + { + "epoch": 16.0, + "learning_rate": 1.0869565217391305e-05, + "loss": 0.001, + "step": 63640 + }, + { + "epoch": 16.0, + "learning_rate": 1.0862543534434334e-05, + "loss": 0.0001, + "step": 63650 + }, + { + "epoch": 16.0, + "learning_rate": 1.0855521851477363e-05, + "loss": 0.0053, + "step": 63660 + }, + { + "epoch": 16.0, + "learning_rate": 1.0848500168520391e-05, + "loss": 0.0002, + "step": 63670 + }, + { + "epoch": 16.0, + "learning_rate": 1.084147848556342e-05, + "loss": 0.0, + "step": 63680 + }, + { + "epoch": 16.0, + "learning_rate": 1.083445680260645e-05, + "loss": 0.0122, + "step": 63690 + }, + { + "epoch": 16.01, + "learning_rate": 1.0827435119649477e-05, + "loss": 0.0, + "step": 63700 + }, + { + "epoch": 16.01, + "learning_rate": 1.0820413436692508e-05, + "loss": 0.0, + "step": 63710 + }, + { + "epoch": 16.01, + "learning_rate": 1.0813391753735535e-05, + "loss": 0.1678, + "step": 63720 + }, + { + "epoch": 16.01, + "learning_rate": 1.0806370070778566e-05, + "loss": 0.0001, + "step": 63730 + }, + { + "epoch": 16.01, + "learning_rate": 1.0799348387821594e-05, + "loss": 0.0, + "step": 63740 + }, + { + "epoch": 16.01, + "learning_rate": 1.0792326704864623e-05, + "loss": 0.0001, + "step": 63750 + }, + { + "epoch": 16.01, + "learning_rate": 1.0785305021907652e-05, + "loss": 0.0, + "step": 63760 + }, + { + "epoch": 16.01, + "learning_rate": 1.077828333895068e-05, + "loss": 0.0, + "step": 63770 + }, + { + "epoch": 16.01, + "learning_rate": 1.0771261655993709e-05, + "loss": 0.0, + "step": 63780 + }, + { + "epoch": 16.01, + "learning_rate": 1.0764239973036738e-05, + "loss": 0.0, + "step": 63790 + }, + { + "epoch": 16.01, + "learning_rate": 1.0757218290079767e-05, + "loss": 0.0, + "step": 63800 + }, + { + "epoch": 16.01, + "learning_rate": 1.0750196607122796e-05, + "loss": 0.0, + "step": 63810 + }, + { + "epoch": 16.01, + "learning_rate": 1.0743174924165825e-05, + "loss": 0.0004, + "step": 63820 + }, + { + "epoch": 16.01, + "learning_rate": 1.0736153241208853e-05, + "loss": 0.0, + "step": 63830 + }, + { + "epoch": 16.01, + "learning_rate": 1.0729131558251882e-05, + "loss": 0.0, + "step": 63840 + }, + { + "epoch": 16.01, + "learning_rate": 1.0722109875294911e-05, + "loss": 0.0, + "step": 63850 + }, + { + "epoch": 16.01, + "learning_rate": 1.071508819233794e-05, + "loss": 0.0, + "step": 63860 + }, + { + "epoch": 16.01, + "learning_rate": 1.070806650938097e-05, + "loss": 0.0001, + "step": 63870 + }, + { + "epoch": 16.01, + "learning_rate": 1.0701044826423997e-05, + "loss": 0.0, + "step": 63880 + }, + { + "epoch": 16.01, + "learning_rate": 1.0694023143467028e-05, + "loss": 0.0, + "step": 63890 + }, + { + "epoch": 16.01, + "learning_rate": 1.0687001460510055e-05, + "loss": 0.0, + "step": 63900 + }, + { + "epoch": 16.01, + "learning_rate": 1.0679979777553085e-05, + "loss": 0.0, + "step": 63910 + }, + { + "epoch": 16.01, + "learning_rate": 1.0672958094596114e-05, + "loss": 0.0079, + "step": 63920 + }, + { + "epoch": 16.01, + "learning_rate": 1.0665936411639143e-05, + "loss": 0.0004, + "step": 63930 + }, + { + "epoch": 16.01, + "learning_rate": 1.065891472868217e-05, + "loss": 0.0001, + "step": 63940 + }, + { + "epoch": 16.01, + "learning_rate": 1.06518930457252e-05, + "loss": 0.0, + "step": 63950 + }, + { + "epoch": 16.01, + "learning_rate": 1.0644871362768229e-05, + "loss": 0.1996, + "step": 63960 + }, + { + "epoch": 16.01, + "learning_rate": 1.0637849679811256e-05, + "loss": 0.0, + "step": 63970 + }, + { + "epoch": 16.01, + "learning_rate": 1.0630827996854287e-05, + "loss": 0.0, + "step": 63980 + }, + { + "epoch": 16.01, + "learning_rate": 1.0623806313897315e-05, + "loss": 0.002, + "step": 63990 + }, + { + "epoch": 16.01, + "learning_rate": 1.0616784630940346e-05, + "loss": 0.0, + "step": 64000 + }, + { + "epoch": 16.01, + "learning_rate": 1.0609762947983373e-05, + "loss": 0.0832, + "step": 64010 + }, + { + "epoch": 16.01, + "learning_rate": 1.0602741265026402e-05, + "loss": 0.0, + "step": 64020 + }, + { + "epoch": 16.01, + "learning_rate": 1.0595719582069431e-05, + "loss": 0.0, + "step": 64030 + }, + { + "epoch": 16.01, + "learning_rate": 1.0588697899112459e-05, + "loss": 0.0, + "step": 64040 + }, + { + "epoch": 16.01, + "learning_rate": 1.0581676216155488e-05, + "loss": 0.0257, + "step": 64050 + }, + { + "epoch": 16.01, + "learning_rate": 1.0574654533198517e-05, + "loss": 0.0074, + "step": 64060 + }, + { + "epoch": 16.01, + "learning_rate": 1.0567632850241546e-05, + "loss": 0.0, + "step": 64070 + }, + { + "epoch": 16.01, + "learning_rate": 1.0560611167284576e-05, + "loss": 0.0, + "step": 64080 + }, + { + "epoch": 16.01, + "learning_rate": 1.0553589484327605e-05, + "loss": 0.0, + "step": 64090 + }, + { + "epoch": 16.01, + "learning_rate": 1.0546567801370632e-05, + "loss": 0.0003, + "step": 64100 + }, + { + "epoch": 16.01, + "learning_rate": 1.0539546118413662e-05, + "loss": 0.0089, + "step": 64110 + }, + { + "epoch": 16.01, + "learning_rate": 1.053252443545669e-05, + "loss": 0.0002, + "step": 64120 + }, + { + "epoch": 16.01, + "learning_rate": 1.052550275249972e-05, + "loss": 0.0, + "step": 64130 + }, + { + "epoch": 16.01, + "learning_rate": 1.0518481069542749e-05, + "loss": 0.0, + "step": 64140 + }, + { + "epoch": 16.01, + "learning_rate": 1.0511459386585777e-05, + "loss": 0.0, + "step": 64150 + }, + { + "epoch": 16.01, + "learning_rate": 1.0504437703628807e-05, + "loss": 0.0, + "step": 64160 + }, + { + "epoch": 16.01, + "learning_rate": 1.0497416020671835e-05, + "loss": 0.0002, + "step": 64170 + }, + { + "epoch": 16.01, + "learning_rate": 1.0490394337714864e-05, + "loss": 0.0012, + "step": 64180 + }, + { + "epoch": 16.01, + "learning_rate": 1.0483372654757893e-05, + "loss": 0.0, + "step": 64190 + }, + { + "epoch": 16.01, + "learning_rate": 1.0476350971800922e-05, + "loss": 0.0, + "step": 64200 + }, + { + "epoch": 16.01, + "learning_rate": 1.046932928884395e-05, + "loss": 0.0, + "step": 64210 + }, + { + "epoch": 16.01, + "learning_rate": 1.046230760588698e-05, + "loss": 0.0, + "step": 64220 + }, + { + "epoch": 16.01, + "learning_rate": 1.0455285922930008e-05, + "loss": 0.0, + "step": 64230 + }, + { + "epoch": 16.01, + "learning_rate": 1.0448264239973036e-05, + "loss": 0.0, + "step": 64240 + }, + { + "epoch": 16.01, + "learning_rate": 1.0441242557016067e-05, + "loss": 0.0013, + "step": 64250 + }, + { + "epoch": 16.01, + "learning_rate": 1.0434220874059094e-05, + "loss": 0.0001, + "step": 64260 + }, + { + "epoch": 16.01, + "learning_rate": 1.0427199191102125e-05, + "loss": 0.0, + "step": 64270 + }, + { + "epoch": 16.01, + "learning_rate": 1.0420177508145153e-05, + "loss": 0.0, + "step": 64280 + }, + { + "epoch": 16.01, + "learning_rate": 1.0413155825188182e-05, + "loss": 0.0, + "step": 64290 + }, + { + "epoch": 16.01, + "learning_rate": 1.0406134142231211e-05, + "loss": 0.0, + "step": 64300 + }, + { + "epoch": 16.01, + "learning_rate": 1.0399112459274238e-05, + "loss": 0.0, + "step": 64310 + }, + { + "epoch": 16.01, + "learning_rate": 1.0392090776317268e-05, + "loss": 0.0, + "step": 64320 + }, + { + "epoch": 16.01, + "learning_rate": 1.0385069093360297e-05, + "loss": 0.0, + "step": 64330 + }, + { + "epoch": 16.01, + "learning_rate": 1.0378047410403326e-05, + "loss": 0.0009, + "step": 64340 + }, + { + "epoch": 16.01, + "learning_rate": 1.0371025727446355e-05, + "loss": 0.0, + "step": 64350 + }, + { + "epoch": 16.01, + "learning_rate": 1.0364004044489384e-05, + "loss": 0.0004, + "step": 64360 + }, + { + "epoch": 16.01, + "learning_rate": 1.0356982361532412e-05, + "loss": 0.2821, + "step": 64370 + }, + { + "epoch": 16.01, + "learning_rate": 1.0349960678575443e-05, + "loss": 0.0, + "step": 64380 + }, + { + "epoch": 16.01, + "learning_rate": 1.034293899561847e-05, + "loss": 0.0, + "step": 64390 + }, + { + "epoch": 16.01, + "learning_rate": 1.03359173126615e-05, + "loss": 0.2507, + "step": 64400 + }, + { + "epoch": 16.01, + "learning_rate": 1.0328895629704529e-05, + "loss": 0.0, + "step": 64410 + }, + { + "epoch": 16.01, + "learning_rate": 1.0321873946747556e-05, + "loss": 0.0, + "step": 64420 + }, + { + "epoch": 16.01, + "learning_rate": 1.0314852263790587e-05, + "loss": 0.0001, + "step": 64430 + }, + { + "epoch": 16.01, + "learning_rate": 1.0307830580833614e-05, + "loss": 0.0, + "step": 64440 + }, + { + "epoch": 16.01, + "learning_rate": 1.0300808897876644e-05, + "loss": 0.0246, + "step": 64450 + }, + { + "epoch": 16.01, + "learning_rate": 1.0293787214919673e-05, + "loss": 0.0, + "step": 64460 + }, + { + "epoch": 16.01, + "learning_rate": 1.0286765531962702e-05, + "loss": 0.0282, + "step": 64470 + }, + { + "epoch": 16.01, + "learning_rate": 1.027974384900573e-05, + "loss": 0.0002, + "step": 64480 + }, + { + "epoch": 16.02, + "learning_rate": 1.0272722166048759e-05, + "loss": 0.0132, + "step": 64490 + }, + { + "epoch": 16.02, + "learning_rate": 1.0265700483091788e-05, + "loss": 0.0, + "step": 64500 + }, + { + "epoch": 16.02, + "learning_rate": 1.0258678800134817e-05, + "loss": 0.0053, + "step": 64510 + }, + { + "epoch": 16.02, + "learning_rate": 1.0251657117177846e-05, + "loss": 0.0, + "step": 64520 + }, + { + "epoch": 16.02, + "learning_rate": 1.0244635434220874e-05, + "loss": 0.0, + "step": 64530 + }, + { + "epoch": 16.02, + "learning_rate": 1.0237613751263905e-05, + "loss": 0.0, + "step": 64540 + }, + { + "epoch": 16.02, + "learning_rate": 1.0230592068306932e-05, + "loss": 0.0, + "step": 64550 + }, + { + "epoch": 16.02, + "learning_rate": 1.0223570385349961e-05, + "loss": 0.0011, + "step": 64560 + }, + { + "epoch": 16.02, + "learning_rate": 1.021654870239299e-05, + "loss": 0.0, + "step": 64570 + }, + { + "epoch": 16.02, + "learning_rate": 1.0209527019436018e-05, + "loss": 0.0, + "step": 64580 + }, + { + "epoch": 16.02, + "learning_rate": 1.0202505336479047e-05, + "loss": 0.0003, + "step": 64590 + }, + { + "epoch": 16.02, + "learning_rate": 1.0195483653522076e-05, + "loss": 0.0, + "step": 64600 + }, + { + "epoch": 16.02, + "learning_rate": 1.0188461970565106e-05, + "loss": 0.0, + "step": 64610 + }, + { + "epoch": 16.02, + "learning_rate": 1.0181440287608135e-05, + "loss": 0.0002, + "step": 64620 + }, + { + "epoch": 16.02, + "learning_rate": 1.0174418604651164e-05, + "loss": 0.0003, + "step": 64630 + }, + { + "epoch": 16.02, + "learning_rate": 1.0167396921694191e-05, + "loss": 0.0, + "step": 64640 + }, + { + "epoch": 16.02, + "learning_rate": 1.0160375238737222e-05, + "loss": 0.0, + "step": 64650 + }, + { + "epoch": 16.02, + "learning_rate": 1.015335355578025e-05, + "loss": 0.0, + "step": 64660 + }, + { + "epoch": 16.02, + "learning_rate": 1.0146331872823279e-05, + "loss": 0.0, + "step": 64670 + }, + { + "epoch": 16.02, + "learning_rate": 1.0139310189866308e-05, + "loss": 0.0002, + "step": 64680 + }, + { + "epoch": 16.02, + "learning_rate": 1.0132288506909336e-05, + "loss": 0.0001, + "step": 64690 + }, + { + "epoch": 16.02, + "learning_rate": 1.0125266823952366e-05, + "loss": 0.0, + "step": 64700 + }, + { + "epoch": 16.02, + "learning_rate": 1.0118245140995394e-05, + "loss": 0.0005, + "step": 64710 + }, + { + "epoch": 16.02, + "learning_rate": 1.0111223458038423e-05, + "loss": 0.0, + "step": 64720 + }, + { + "epoch": 16.02, + "learning_rate": 1.0104201775081452e-05, + "loss": 0.0, + "step": 64730 + }, + { + "epoch": 16.02, + "learning_rate": 1.0097180092124482e-05, + "loss": 0.0, + "step": 64740 + }, + { + "epoch": 16.02, + "learning_rate": 1.0090158409167509e-05, + "loss": 0.0002, + "step": 64750 + }, + { + "epoch": 16.02, + "learning_rate": 1.0083136726210538e-05, + "loss": 0.0, + "step": 64760 + }, + { + "epoch": 16.02, + "learning_rate": 1.0076115043253567e-05, + "loss": 0.0, + "step": 64770 + }, + { + "epoch": 16.02, + "learning_rate": 1.0069093360296597e-05, + "loss": 0.0005, + "step": 64780 + }, + { + "epoch": 16.02, + "learning_rate": 1.0062071677339626e-05, + "loss": 0.0002, + "step": 64790 + }, + { + "epoch": 16.02, + "learning_rate": 1.0055049994382653e-05, + "loss": 0.0, + "step": 64800 + }, + { + "epoch": 16.02, + "learning_rate": 1.0048028311425684e-05, + "loss": 0.0023, + "step": 64810 + }, + { + "epoch": 16.02, + "learning_rate": 1.0041006628468712e-05, + "loss": 0.0029, + "step": 64820 + }, + { + "epoch": 16.02, + "learning_rate": 1.003398494551174e-05, + "loss": 0.0001, + "step": 64830 + }, + { + "epoch": 16.02, + "learning_rate": 1.002696326255477e-05, + "loss": 0.0, + "step": 64840 + }, + { + "epoch": 16.02, + "learning_rate": 1.0019941579597797e-05, + "loss": 0.0, + "step": 64850 + }, + { + "epoch": 16.02, + "learning_rate": 1.0012919896640828e-05, + "loss": 0.0, + "step": 64860 + }, + { + "epoch": 16.02, + "learning_rate": 1.0005898213683856e-05, + "loss": 0.0017, + "step": 64870 + }, + { + "epoch": 16.02, + "learning_rate": 9.998876530726885e-06, + "loss": 0.0, + "step": 64880 + }, + { + "epoch": 16.02, + "learning_rate": 9.991854847769914e-06, + "loss": 0.0, + "step": 64890 + }, + { + "epoch": 16.02, + "learning_rate": 9.984833164812943e-06, + "loss": 0.0001, + "step": 64900 + }, + { + "epoch": 16.02, + "learning_rate": 9.977811481855971e-06, + "loss": 0.0002, + "step": 64910 + }, + { + "epoch": 16.02, + "learning_rate": 9.970789798899002e-06, + "loss": 0.0, + "step": 64920 + }, + { + "epoch": 16.02, + "learning_rate": 9.96376811594203e-06, + "loss": 0.0, + "step": 64930 + }, + { + "epoch": 16.02, + "learning_rate": 9.956746432985058e-06, + "loss": 0.0, + "step": 64940 + }, + { + "epoch": 16.02, + "learning_rate": 9.949724750028088e-06, + "loss": 0.0, + "step": 64950 + }, + { + "epoch": 16.02, + "learning_rate": 9.942703067071115e-06, + "loss": 0.0, + "step": 64960 + }, + { + "epoch": 16.02, + "learning_rate": 9.935681384114146e-06, + "loss": 0.0, + "step": 64970 + }, + { + "epoch": 16.02, + "learning_rate": 9.928659701157173e-06, + "loss": 0.0002, + "step": 64980 + }, + { + "epoch": 16.02, + "learning_rate": 9.921638018200203e-06, + "loss": 0.0, + "step": 64990 + }, + { + "epoch": 16.02, + "learning_rate": 9.914616335243232e-06, + "loss": 0.0, + "step": 65000 + }, + { + "epoch": 16.02, + "learning_rate": 9.907594652286261e-06, + "loss": 0.0, + "step": 65010 + }, + { + "epoch": 16.02, + "learning_rate": 9.900572969329289e-06, + "loss": 0.0, + "step": 65020 + }, + { + "epoch": 16.02, + "learning_rate": 9.893551286372318e-06, + "loss": 0.0001, + "step": 65030 + }, + { + "epoch": 16.02, + "learning_rate": 9.886529603415347e-06, + "loss": 0.0, + "step": 65040 + }, + { + "epoch": 16.02, + "learning_rate": 9.879507920458376e-06, + "loss": 0.0, + "step": 65050 + }, + { + "epoch": 16.02, + "learning_rate": 9.872486237501405e-06, + "loss": 0.0, + "step": 65060 + }, + { + "epoch": 16.02, + "learning_rate": 9.865464554544433e-06, + "loss": 0.0, + "step": 65070 + }, + { + "epoch": 16.02, + "learning_rate": 9.858442871587464e-06, + "loss": 0.0, + "step": 65080 + }, + { + "epoch": 16.02, + "learning_rate": 9.851421188630491e-06, + "loss": 0.0001, + "step": 65090 + }, + { + "epoch": 16.02, + "learning_rate": 9.84439950567352e-06, + "loss": 0.0, + "step": 65100 + }, + { + "epoch": 16.02, + "learning_rate": 9.83737782271655e-06, + "loss": 0.0, + "step": 65110 + }, + { + "epoch": 16.02, + "learning_rate": 9.830356139759577e-06, + "loss": 0.0, + "step": 65120 + }, + { + "epoch": 16.02, + "learning_rate": 9.823334456802608e-06, + "loss": 0.0, + "step": 65130 + }, + { + "epoch": 16.02, + "learning_rate": 9.816312773845635e-06, + "loss": 0.0, + "step": 65140 + }, + { + "epoch": 16.02, + "learning_rate": 9.809291090888665e-06, + "loss": 0.0001, + "step": 65150 + }, + { + "epoch": 16.02, + "learning_rate": 9.802269407931694e-06, + "loss": 0.0, + "step": 65160 + }, + { + "epoch": 16.02, + "learning_rate": 9.795247724974723e-06, + "loss": 0.0, + "step": 65170 + }, + { + "epoch": 16.02, + "learning_rate": 9.78822604201775e-06, + "loss": 0.0, + "step": 65180 + }, + { + "epoch": 16.02, + "learning_rate": 9.781204359060781e-06, + "loss": 0.0, + "step": 65190 + }, + { + "epoch": 16.02, + "learning_rate": 9.774182676103809e-06, + "loss": 0.0, + "step": 65200 + }, + { + "epoch": 16.02, + "learning_rate": 9.767160993146838e-06, + "loss": 0.0, + "step": 65210 + }, + { + "epoch": 16.02, + "learning_rate": 9.760139310189867e-06, + "loss": 0.0, + "step": 65220 + }, + { + "epoch": 16.02, + "learning_rate": 9.753117627232895e-06, + "loss": 0.4694, + "step": 65230 + }, + { + "epoch": 16.02, + "learning_rate": 9.746095944275925e-06, + "loss": 0.0, + "step": 65240 + }, + { + "epoch": 16.02, + "learning_rate": 9.739074261318953e-06, + "loss": 0.0, + "step": 65250 + }, + { + "epoch": 16.02, + "learning_rate": 9.732052578361982e-06, + "loss": 0.0, + "step": 65260 + }, + { + "epoch": 16.02, + "learning_rate": 9.725030895405011e-06, + "loss": 0.0095, + "step": 65270 + }, + { + "epoch": 16.03, + "learning_rate": 9.71800921244804e-06, + "loss": 0.0297, + "step": 65280 + }, + { + "epoch": 16.03, + "learning_rate": 9.710987529491068e-06, + "loss": 0.0001, + "step": 65290 + }, + { + "epoch": 16.03, + "learning_rate": 9.703965846534097e-06, + "loss": 0.0022, + "step": 65300 + }, + { + "epoch": 16.03, + "learning_rate": 9.696944163577126e-06, + "loss": 0.0, + "step": 65310 + }, + { + "epoch": 16.03, + "learning_rate": 9.689922480620156e-06, + "loss": 0.0, + "step": 65320 + }, + { + "epoch": 16.03, + "learning_rate": 9.682900797663185e-06, + "loss": 0.0002, + "step": 65330 + }, + { + "epoch": 16.03, + "learning_rate": 9.675879114706212e-06, + "loss": 0.0, + "step": 65340 + }, + { + "epoch": 16.03, + "learning_rate": 9.668857431749243e-06, + "loss": 0.5037, + "step": 65350 + }, + { + "epoch": 16.03, + "learning_rate": 9.66183574879227e-06, + "loss": 0.0, + "step": 65360 + }, + { + "epoch": 16.03, + "learning_rate": 9.6548140658353e-06, + "loss": 0.0, + "step": 65370 + }, + { + "epoch": 16.03, + "learning_rate": 9.647792382878329e-06, + "loss": 0.0, + "step": 65380 + }, + { + "epoch": 16.03, + "learning_rate": 9.640770699921358e-06, + "loss": 0.0, + "step": 65390 + }, + { + "epoch": 16.03, + "learning_rate": 9.633749016964387e-06, + "loss": 0.0, + "step": 65400 + }, + { + "epoch": 16.03, + "learning_rate": 9.626727334007415e-06, + "loss": 0.0001, + "step": 65410 + }, + { + "epoch": 16.03, + "learning_rate": 9.619705651050444e-06, + "loss": 0.2667, + "step": 65420 + }, + { + "epoch": 16.03, + "learning_rate": 9.612683968093473e-06, + "loss": 0.0, + "step": 65430 + }, + { + "epoch": 16.03, + "learning_rate": 9.605662285136502e-06, + "loss": 0.0, + "step": 65440 + }, + { + "epoch": 16.03, + "learning_rate": 9.59864060217953e-06, + "loss": 0.0, + "step": 65450 + }, + { + "epoch": 16.03, + "learning_rate": 9.59161891922256e-06, + "loss": 0.0, + "step": 65460 + }, + { + "epoch": 16.03, + "learning_rate": 9.584597236265588e-06, + "loss": 0.0298, + "step": 65470 + }, + { + "epoch": 16.03, + "learning_rate": 9.577575553308617e-06, + "loss": 0.0, + "step": 65480 + }, + { + "epoch": 16.03, + "learning_rate": 9.570553870351647e-06, + "loss": 0.0, + "step": 65490 + }, + { + "epoch": 16.03, + "learning_rate": 9.563532187394674e-06, + "loss": 0.0, + "step": 65500 + }, + { + "epoch": 16.03, + "learning_rate": 9.556510504437705e-06, + "loss": 0.0, + "step": 65510 + }, + { + "epoch": 16.03, + "learning_rate": 9.549488821480732e-06, + "loss": 0.0, + "step": 65520 + }, + { + "epoch": 16.03, + "learning_rate": 9.542467138523762e-06, + "loss": 0.0, + "step": 65530 + }, + { + "epoch": 16.03, + "learning_rate": 9.535445455566791e-06, + "loss": 0.014, + "step": 65540 + }, + { + "epoch": 16.03, + "learning_rate": 9.52842377260982e-06, + "loss": 0.0, + "step": 65550 + }, + { + "epoch": 16.03, + "learning_rate": 9.52140208965285e-06, + "loss": 0.0, + "step": 65560 + }, + { + "epoch": 16.03, + "learning_rate": 9.514380406695877e-06, + "loss": 0.0, + "step": 65570 + }, + { + "epoch": 16.03, + "learning_rate": 9.507358723738906e-06, + "loss": 0.0001, + "step": 65580 + }, + { + "epoch": 16.03, + "learning_rate": 9.500337040781935e-06, + "loss": 0.0, + "step": 65590 + }, + { + "epoch": 16.03, + "learning_rate": 9.493315357824964e-06, + "loss": 0.0, + "step": 65600 + }, + { + "epoch": 16.03, + "learning_rate": 9.486293674867992e-06, + "loss": 0.0, + "step": 65610 + }, + { + "epoch": 16.03, + "learning_rate": 9.479271991911023e-06, + "loss": 0.0, + "step": 65620 + }, + { + "epoch": 16.03, + "learning_rate": 9.47225030895405e-06, + "loss": 0.0978, + "step": 65630 + }, + { + "epoch": 16.03, + "learning_rate": 9.46522862599708e-06, + "loss": 0.0, + "step": 65640 + }, + { + "epoch": 16.03, + "learning_rate": 9.458206943040109e-06, + "loss": 0.0, + "step": 65650 + }, + { + "epoch": 16.03, + "learning_rate": 9.451185260083138e-06, + "loss": 0.0, + "step": 65660 + }, + { + "epoch": 16.03, + "learning_rate": 9.444163577126167e-06, + "loss": 0.0, + "step": 65670 + }, + { + "epoch": 16.03, + "learning_rate": 9.437141894169194e-06, + "loss": 0.0, + "step": 65680 + }, + { + "epoch": 16.03, + "learning_rate": 9.430120211212224e-06, + "loss": 0.0, + "step": 65690 + }, + { + "epoch": 16.03, + "learning_rate": 9.423098528255253e-06, + "loss": 0.0, + "step": 65700 + }, + { + "epoch": 16.03, + "learning_rate": 9.416076845298282e-06, + "loss": 0.0004, + "step": 65710 + }, + { + "epoch": 16.03, + "learning_rate": 9.40905516234131e-06, + "loss": 0.0, + "step": 65720 + }, + { + "epoch": 16.03, + "learning_rate": 9.40203347938434e-06, + "loss": 0.0001, + "step": 65730 + }, + { + "epoch": 16.03, + "learning_rate": 9.395011796427368e-06, + "loss": 0.0, + "step": 65740 + }, + { + "epoch": 16.03, + "learning_rate": 9.387990113470397e-06, + "loss": 0.0, + "step": 65750 + }, + { + "epoch": 16.03, + "learning_rate": 9.380968430513426e-06, + "loss": 0.0002, + "step": 65760 + }, + { + "epoch": 16.03, + "learning_rate": 9.373946747556454e-06, + "loss": 0.0, + "step": 65770 + }, + { + "epoch": 16.03, + "learning_rate": 9.366925064599485e-06, + "loss": 0.0, + "step": 65780 + }, + { + "epoch": 16.03, + "learning_rate": 9.359903381642512e-06, + "loss": 0.0024, + "step": 65790 + }, + { + "epoch": 16.03, + "learning_rate": 9.352881698685541e-06, + "loss": 0.0, + "step": 65800 + }, + { + "epoch": 16.03, + "learning_rate": 9.34586001572857e-06, + "loss": 0.0016, + "step": 65810 + }, + { + "epoch": 16.03, + "learning_rate": 9.3388383327716e-06, + "loss": 0.0008, + "step": 65820 + }, + { + "epoch": 16.03, + "learning_rate": 9.331816649814629e-06, + "loss": 0.0, + "step": 65830 + }, + { + "epoch": 16.03, + "learning_rate": 9.324794966857656e-06, + "loss": 0.037, + "step": 65840 + }, + { + "epoch": 16.03, + "learning_rate": 9.317773283900685e-06, + "loss": 0.0, + "step": 65850 + }, + { + "epoch": 16.03, + "learning_rate": 9.310751600943715e-06, + "loss": 0.0, + "step": 65860 + }, + { + "epoch": 16.03, + "learning_rate": 9.303729917986744e-06, + "loss": 0.0003, + "step": 65870 + }, + { + "epoch": 16.03, + "learning_rate": 9.296708235029771e-06, + "loss": 0.0, + "step": 65880 + }, + { + "epoch": 16.03, + "learning_rate": 9.289686552072802e-06, + "loss": 0.0, + "step": 65890 + }, + { + "epoch": 16.03, + "learning_rate": 9.28266486911583e-06, + "loss": 0.0, + "step": 65900 + }, + { + "epoch": 16.03, + "learning_rate": 9.27564318615886e-06, + "loss": 0.0002, + "step": 65910 + }, + { + "epoch": 16.03, + "learning_rate": 9.268621503201888e-06, + "loss": 0.0, + "step": 65920 + }, + { + "epoch": 16.03, + "learning_rate": 9.261599820244917e-06, + "loss": 0.0003, + "step": 65930 + }, + { + "epoch": 16.03, + "learning_rate": 9.254578137287946e-06, + "loss": 0.0002, + "step": 65940 + }, + { + "epoch": 16.03, + "learning_rate": 9.247556454330974e-06, + "loss": 0.0, + "step": 65950 + }, + { + "epoch": 16.03, + "learning_rate": 9.240534771374003e-06, + "loss": 0.0, + "step": 65960 + }, + { + "epoch": 16.03, + "learning_rate": 9.233513088417032e-06, + "loss": 0.0012, + "step": 65970 + }, + { + "epoch": 16.03, + "learning_rate": 9.226491405460061e-06, + "loss": 0.0007, + "step": 65980 + }, + { + "epoch": 16.03, + "learning_rate": 9.219469722503089e-06, + "loss": 0.0232, + "step": 65990 + }, + { + "epoch": 16.03, + "learning_rate": 9.21244803954612e-06, + "loss": 0.0003, + "step": 66000 + }, + { + "epoch": 16.03, + "learning_rate": 9.205426356589147e-06, + "loss": 0.0, + "step": 66010 + }, + { + "epoch": 16.03, + "learning_rate": 9.198404673632176e-06, + "loss": 0.0001, + "step": 66020 + }, + { + "epoch": 16.03, + "learning_rate": 9.191382990675206e-06, + "loss": 0.0002, + "step": 66030 + }, + { + "epoch": 16.03, + "learning_rate": 9.184361307718233e-06, + "loss": 0.0, + "step": 66040 + }, + { + "epoch": 16.03, + "learning_rate": 9.177339624761264e-06, + "loss": 0.0, + "step": 66050 + }, + { + "epoch": 16.03, + "learning_rate": 9.170317941804292e-06, + "loss": 0.0003, + "step": 66060 + }, + { + "epoch": 16.04, + "learning_rate": 9.16329625884732e-06, + "loss": 0.0, + "step": 66070 + }, + { + "epoch": 16.04, + "learning_rate": 9.15627457589035e-06, + "loss": 0.0, + "step": 66080 + }, + { + "epoch": 16.04, + "learning_rate": 9.149252892933379e-06, + "loss": 0.0001, + "step": 66090 + }, + { + "epoch": 16.04, + "learning_rate": 9.142231209976408e-06, + "loss": 0.0, + "step": 66100 + }, + { + "epoch": 16.04, + "learning_rate": 9.135209527019436e-06, + "loss": 0.0001, + "step": 66110 + }, + { + "epoch": 16.04, + "learning_rate": 9.128187844062465e-06, + "loss": 0.0, + "step": 66120 + }, + { + "epoch": 16.04, + "learning_rate": 9.121166161105494e-06, + "loss": 0.014, + "step": 66130 + }, + { + "epoch": 16.04, + "learning_rate": 9.114144478148523e-06, + "loss": 0.0022, + "step": 66140 + }, + { + "epoch": 16.04, + "learning_rate": 9.10712279519155e-06, + "loss": 0.0, + "step": 66150 + }, + { + "epoch": 16.04, + "learning_rate": 9.100101112234582e-06, + "loss": 0.0, + "step": 66160 + }, + { + "epoch": 16.04, + "learning_rate": 9.09307942927761e-06, + "loss": 0.0, + "step": 66170 + }, + { + "epoch": 16.04, + "learning_rate": 9.08605774632064e-06, + "loss": 0.3525, + "step": 66180 + }, + { + "epoch": 16.04, + "learning_rate": 9.079036063363668e-06, + "loss": 0.0002, + "step": 66190 + }, + { + "epoch": 16.04, + "learning_rate": 9.072014380406697e-06, + "loss": 0.0019, + "step": 66200 + }, + { + "epoch": 16.04, + "learning_rate": 9.064992697449726e-06, + "loss": 0.0, + "step": 66210 + }, + { + "epoch": 16.04, + "learning_rate": 9.057971014492753e-06, + "loss": 0.0015, + "step": 66220 + }, + { + "epoch": 16.04, + "learning_rate": 9.050949331535783e-06, + "loss": 0.0, + "step": 66230 + }, + { + "epoch": 16.04, + "learning_rate": 9.043927648578812e-06, + "loss": 0.0, + "step": 66240 + }, + { + "epoch": 16.04, + "learning_rate": 9.036905965621841e-06, + "loss": 0.0, + "step": 66250 + }, + { + "epoch": 16.04, + "learning_rate": 9.029884282664868e-06, + "loss": 0.1715, + "step": 66260 + }, + { + "epoch": 16.04, + "learning_rate": 9.0228625997079e-06, + "loss": 0.0, + "step": 66270 + }, + { + "epoch": 16.04, + "learning_rate": 9.015840916750927e-06, + "loss": 0.0, + "step": 66280 + }, + { + "epoch": 16.04, + "learning_rate": 9.008819233793956e-06, + "loss": 0.0, + "step": 66290 + }, + { + "epoch": 16.04, + "learning_rate": 9.001797550836985e-06, + "loss": 0.0, + "step": 66300 + }, + { + "epoch": 16.04, + "learning_rate": 8.994775867880013e-06, + "loss": 0.0, + "step": 66310 + }, + { + "epoch": 16.04, + "learning_rate": 8.987754184923044e-06, + "loss": 0.0022, + "step": 66320 + }, + { + "epoch": 16.04, + "learning_rate": 8.980732501966071e-06, + "loss": 0.0003, + "step": 66330 + }, + { + "epoch": 16.04, + "learning_rate": 8.9737108190091e-06, + "loss": 0.0619, + "step": 66340 + }, + { + "epoch": 16.04, + "learning_rate": 8.96668913605213e-06, + "loss": 0.7071, + "step": 66350 + }, + { + "epoch": 16.04, + "learning_rate": 8.959667453095159e-06, + "loss": 0.0, + "step": 66360 + }, + { + "epoch": 16.04, + "learning_rate": 8.952645770138188e-06, + "loss": 0.0, + "step": 66370 + }, + { + "epoch": 16.04, + "learning_rate": 8.945624087181215e-06, + "loss": 0.0, + "step": 66380 + }, + { + "epoch": 16.04, + "learning_rate": 8.938602404224244e-06, + "loss": 0.0025, + "step": 66390 + }, + { + "epoch": 16.04, + "learning_rate": 8.931580721267274e-06, + "loss": 0.0006, + "step": 66400 + }, + { + "epoch": 16.04, + "learning_rate": 8.924559038310303e-06, + "loss": 0.0, + "step": 66410 + }, + { + "epoch": 16.04, + "learning_rate": 8.91753735535333e-06, + "loss": 0.6254, + "step": 66420 + }, + { + "epoch": 16.04, + "learning_rate": 8.910515672396361e-06, + "loss": 0.0, + "step": 66430 + }, + { + "epoch": 16.04, + "learning_rate": 8.903493989439389e-06, + "loss": 0.0002, + "step": 66440 + }, + { + "epoch": 16.04, + "learning_rate": 8.89647230648242e-06, + "loss": 0.0, + "step": 66450 + }, + { + "epoch": 16.04, + "learning_rate": 8.889450623525447e-06, + "loss": 0.0, + "step": 66460 + }, + { + "epoch": 16.04, + "learning_rate": 8.882428940568476e-06, + "loss": 0.0032, + "step": 66470 + }, + { + "epoch": 16.04, + "learning_rate": 8.875407257611505e-06, + "loss": 0.0, + "step": 66480 + }, + { + "epoch": 16.04, + "learning_rate": 8.868385574654533e-06, + "loss": 0.0, + "step": 66490 + }, + { + "epoch": 16.04, + "learning_rate": 8.861363891697562e-06, + "loss": 0.0, + "step": 66500 + }, + { + "epoch": 16.04, + "learning_rate": 8.854342208740591e-06, + "loss": 0.0, + "step": 66510 + }, + { + "epoch": 16.04, + "learning_rate": 8.84732052578362e-06, + "loss": 0.0, + "step": 66520 + }, + { + "epoch": 16.04, + "learning_rate": 8.84029884282665e-06, + "loss": 0.0, + "step": 66530 + }, + { + "epoch": 16.04, + "learning_rate": 8.833277159869679e-06, + "loss": 0.0384, + "step": 66540 + }, + { + "epoch": 16.04, + "learning_rate": 8.826255476912706e-06, + "loss": 0.0, + "step": 66550 + }, + { + "epoch": 16.04, + "learning_rate": 8.819233793955736e-06, + "loss": 0.025, + "step": 66560 + }, + { + "epoch": 16.04, + "learning_rate": 8.812212110998765e-06, + "loss": 0.0, + "step": 66570 + }, + { + "epoch": 16.04, + "learning_rate": 8.805190428041792e-06, + "loss": 0.0, + "step": 66580 + }, + { + "epoch": 16.04, + "learning_rate": 8.798168745084823e-06, + "loss": 0.0, + "step": 66590 + }, + { + "epoch": 16.04, + "learning_rate": 8.79114706212785e-06, + "loss": 0.0, + "step": 66600 + }, + { + "epoch": 16.04, + "learning_rate": 8.78412537917088e-06, + "loss": 0.0, + "step": 66610 + }, + { + "epoch": 16.04, + "learning_rate": 8.777103696213909e-06, + "loss": 0.0, + "step": 66620 + }, + { + "epoch": 16.04, + "learning_rate": 8.770082013256938e-06, + "loss": 0.0, + "step": 66630 + }, + { + "epoch": 16.04, + "learning_rate": 8.763060330299967e-06, + "loss": 0.0, + "step": 66640 + }, + { + "epoch": 16.04, + "learning_rate": 8.756038647342995e-06, + "loss": 0.0, + "step": 66650 + }, + { + "epoch": 16.04, + "learning_rate": 8.749016964386024e-06, + "loss": 0.0, + "step": 66660 + }, + { + "epoch": 16.04, + "learning_rate": 8.741995281429053e-06, + "loss": 0.0581, + "step": 66670 + }, + { + "epoch": 16.04, + "learning_rate": 8.734973598472082e-06, + "loss": 0.0, + "step": 66680 + }, + { + "epoch": 16.04, + "learning_rate": 8.72795191551511e-06, + "loss": 0.0, + "step": 66690 + }, + { + "epoch": 16.04, + "learning_rate": 8.72093023255814e-06, + "loss": 0.0, + "step": 66700 + }, + { + "epoch": 16.04, + "learning_rate": 8.713908549601168e-06, + "loss": 0.0001, + "step": 66710 + }, + { + "epoch": 16.04, + "learning_rate": 8.706886866644199e-06, + "loss": 0.0, + "step": 66720 + }, + { + "epoch": 16.04, + "learning_rate": 8.699865183687227e-06, + "loss": 0.0, + "step": 66730 + }, + { + "epoch": 16.04, + "learning_rate": 8.692843500730256e-06, + "loss": 0.0, + "step": 66740 + }, + { + "epoch": 16.04, + "learning_rate": 8.685821817773285e-06, + "loss": 0.0, + "step": 66750 + }, + { + "epoch": 16.04, + "learning_rate": 8.678800134816312e-06, + "loss": 0.0, + "step": 66760 + }, + { + "epoch": 16.04, + "learning_rate": 8.671778451859342e-06, + "loss": 0.0, + "step": 66770 + }, + { + "epoch": 16.04, + "learning_rate": 8.66475676890237e-06, + "loss": 0.0, + "step": 66780 + }, + { + "epoch": 16.04, + "learning_rate": 8.6577350859454e-06, + "loss": 0.0, + "step": 66790 + }, + { + "epoch": 16.04, + "learning_rate": 8.65071340298843e-06, + "loss": 0.0, + "step": 66800 + }, + { + "epoch": 16.04, + "learning_rate": 8.643691720031458e-06, + "loss": 0.0001, + "step": 66810 + }, + { + "epoch": 16.04, + "learning_rate": 8.636670037074486e-06, + "loss": 0.0, + "step": 66820 + }, + { + "epoch": 16.04, + "learning_rate": 8.629648354117515e-06, + "loss": 0.0002, + "step": 66830 + }, + { + "epoch": 16.04, + "learning_rate": 8.622626671160544e-06, + "loss": 0.0, + "step": 66840 + }, + { + "epoch": 16.04, + "learning_rate": 8.615604988203572e-06, + "loss": 0.0001, + "step": 66850 + }, + { + "epoch": 16.05, + "learning_rate": 8.608583305246603e-06, + "loss": 0.0, + "step": 66860 + }, + { + "epoch": 16.05, + "learning_rate": 8.60156162228963e-06, + "loss": 0.0, + "step": 66870 + }, + { + "epoch": 16.05, + "learning_rate": 8.594539939332661e-06, + "loss": 0.0, + "step": 66880 + }, + { + "epoch": 16.05, + "learning_rate": 8.587518256375688e-06, + "loss": 0.0016, + "step": 66890 + }, + { + "epoch": 16.05, + "learning_rate": 8.580496573418718e-06, + "loss": 0.0, + "step": 66900 + }, + { + "epoch": 16.05, + "learning_rate": 8.573474890461747e-06, + "loss": 0.0049, + "step": 66910 + }, + { + "epoch": 16.05, + "learning_rate": 8.566453207504776e-06, + "loss": 0.0235, + "step": 66920 + }, + { + "epoch": 16.05, + "learning_rate": 8.559431524547803e-06, + "loss": 0.2714, + "step": 66930 + }, + { + "epoch": 16.05, + "learning_rate": 8.552409841590833e-06, + "loss": 0.0, + "step": 66940 + }, + { + "epoch": 16.05, + "learning_rate": 8.545388158633862e-06, + "loss": 0.0001, + "step": 66950 + }, + { + "epoch": 16.05, + "learning_rate": 8.53836647567689e-06, + "loss": 0.0, + "step": 66960 + }, + { + "epoch": 16.05, + "learning_rate": 8.53134479271992e-06, + "loss": 0.0, + "step": 66970 + }, + { + "epoch": 16.05, + "learning_rate": 8.524323109762948e-06, + "loss": 0.0, + "step": 66980 + }, + { + "epoch": 16.05, + "learning_rate": 8.517301426805979e-06, + "loss": 0.0, + "step": 66990 + }, + { + "epoch": 16.05, + "learning_rate": 8.510279743849006e-06, + "loss": 0.0001, + "step": 67000 + }, + { + "epoch": 16.05, + "learning_rate": 8.503258060892035e-06, + "loss": 0.0, + "step": 67010 + }, + { + "epoch": 16.05, + "learning_rate": 8.496236377935064e-06, + "loss": 0.0, + "step": 67020 + }, + { + "epoch": 16.05, + "learning_rate": 8.489214694978092e-06, + "loss": 0.3789, + "step": 67030 + }, + { + "epoch": 16.05, + "learning_rate": 8.482193012021121e-06, + "loss": 0.0162, + "step": 67040 + }, + { + "epoch": 16.05, + "learning_rate": 8.47517132906415e-06, + "loss": 0.0, + "step": 67050 + }, + { + "epoch": 16.05, + "learning_rate": 8.46814964610718e-06, + "loss": 0.0, + "step": 67060 + }, + { + "epoch": 16.05, + "learning_rate": 8.461127963150209e-06, + "loss": 0.0001, + "step": 67070 + }, + { + "epoch": 16.05, + "learning_rate": 8.454106280193238e-06, + "loss": 0.0, + "step": 67080 + }, + { + "epoch": 16.05, + "learning_rate": 8.447084597236265e-06, + "loss": 0.0632, + "step": 67090 + }, + { + "epoch": 16.05, + "learning_rate": 8.440062914279295e-06, + "loss": 0.0, + "step": 67100 + }, + { + "epoch": 16.05, + "learning_rate": 8.433041231322324e-06, + "loss": 0.2193, + "step": 67110 + }, + { + "epoch": 16.05, + "learning_rate": 8.426019548365351e-06, + "loss": 0.0, + "step": 67120 + }, + { + "epoch": 16.05, + "learning_rate": 8.418997865408382e-06, + "loss": 0.0538, + "step": 67130 + }, + { + "epoch": 16.05, + "learning_rate": 8.41197618245141e-06, + "loss": 0.0, + "step": 67140 + }, + { + "epoch": 16.05, + "learning_rate": 8.40495449949444e-06, + "loss": 0.0001, + "step": 67150 + }, + { + "epoch": 16.05, + "learning_rate": 8.397932816537468e-06, + "loss": 0.4657, + "step": 67160 + }, + { + "epoch": 16.05, + "learning_rate": 8.390911133580497e-06, + "loss": 0.0, + "step": 67170 + }, + { + "epoch": 16.05, + "learning_rate": 8.383889450623526e-06, + "loss": 0.3601, + "step": 67180 + }, + { + "epoch": 16.05, + "learning_rate": 8.376867767666555e-06, + "loss": 0.0, + "step": 67190 + }, + { + "epoch": 16.05, + "learning_rate": 8.369846084709583e-06, + "loss": 0.1349, + "step": 67200 + }, + { + "epoch": 16.05, + "learning_rate": 8.362824401752612e-06, + "loss": 0.0, + "step": 67210 + }, + { + "epoch": 16.05, + "learning_rate": 8.355802718795641e-06, + "loss": 0.0, + "step": 67220 + }, + { + "epoch": 16.05, + "learning_rate": 8.34878103583867e-06, + "loss": 0.0001, + "step": 67230 + }, + { + "epoch": 16.05, + "learning_rate": 8.3417593528817e-06, + "loss": 0.0035, + "step": 67240 + }, + { + "epoch": 16.05, + "learning_rate": 8.334737669924727e-06, + "loss": 0.0, + "step": 67250 + }, + { + "epoch": 16.05, + "eval_accuracy": 0.7593858856921688, + "eval_loss": 1.7354708909988403, + "eval_runtime": 1271.3316, + "eval_samples_per_second": 6.097, + "eval_steps_per_second": 3.049, + "step": 67252 + }, + { + "epoch": 17.0, + "learning_rate": 8.327715986967758e-06, + "loss": 0.0, + "step": 67260 + }, + { + "epoch": 17.0, + "learning_rate": 8.320694304010786e-06, + "loss": 0.0, + "step": 67270 + }, + { + "epoch": 17.0, + "learning_rate": 8.313672621053815e-06, + "loss": 0.0002, + "step": 67280 + }, + { + "epoch": 17.0, + "learning_rate": 8.306650938096844e-06, + "loss": 0.0001, + "step": 67290 + }, + { + "epoch": 17.0, + "learning_rate": 8.299629255139871e-06, + "loss": 0.0, + "step": 67300 + }, + { + "epoch": 17.0, + "learning_rate": 8.2926075721829e-06, + "loss": 0.0, + "step": 67310 + }, + { + "epoch": 17.0, + "learning_rate": 8.28558588922593e-06, + "loss": 0.0, + "step": 67320 + }, + { + "epoch": 17.0, + "learning_rate": 8.278564206268959e-06, + "loss": 0.0, + "step": 67330 + }, + { + "epoch": 17.0, + "learning_rate": 8.271542523311988e-06, + "loss": 0.0, + "step": 67340 + }, + { + "epoch": 17.0, + "learning_rate": 8.264520840355017e-06, + "loss": 0.3809, + "step": 67350 + }, + { + "epoch": 17.0, + "learning_rate": 8.257499157398045e-06, + "loss": 0.0, + "step": 67360 + }, + { + "epoch": 17.0, + "learning_rate": 8.250477474441074e-06, + "loss": 0.0, + "step": 67370 + }, + { + "epoch": 17.0, + "learning_rate": 8.243455791484103e-06, + "loss": 0.0, + "step": 67380 + }, + { + "epoch": 17.0, + "learning_rate": 8.23643410852713e-06, + "loss": 0.0, + "step": 67390 + }, + { + "epoch": 17.0, + "learning_rate": 8.229412425570162e-06, + "loss": 0.0, + "step": 67400 + }, + { + "epoch": 17.0, + "learning_rate": 8.222390742613189e-06, + "loss": 0.079, + "step": 67410 + }, + { + "epoch": 17.0, + "learning_rate": 8.21536905965622e-06, + "loss": 0.1336, + "step": 67420 + }, + { + "epoch": 17.0, + "learning_rate": 8.208347376699247e-06, + "loss": 0.0, + "step": 67430 + }, + { + "epoch": 17.0, + "learning_rate": 8.201325693742277e-06, + "loss": 0.0, + "step": 67440 + }, + { + "epoch": 17.0, + "learning_rate": 8.194304010785306e-06, + "loss": 0.0, + "step": 67450 + }, + { + "epoch": 17.0, + "learning_rate": 8.187282327828335e-06, + "loss": 0.0, + "step": 67460 + }, + { + "epoch": 17.0, + "learning_rate": 8.180260644871363e-06, + "loss": 0.0, + "step": 67470 + }, + { + "epoch": 17.0, + "learning_rate": 8.173238961914392e-06, + "loss": 0.0, + "step": 67480 + }, + { + "epoch": 17.0, + "learning_rate": 8.166217278957421e-06, + "loss": 0.0, + "step": 67490 + }, + { + "epoch": 17.0, + "learning_rate": 8.15919559600045e-06, + "loss": 0.0, + "step": 67500 + }, + { + "epoch": 17.0, + "learning_rate": 8.15217391304348e-06, + "loss": 0.0, + "step": 67510 + }, + { + "epoch": 17.0, + "learning_rate": 8.145152230086507e-06, + "loss": 0.0, + "step": 67520 + }, + { + "epoch": 17.0, + "learning_rate": 8.138130547129538e-06, + "loss": 0.0, + "step": 67530 + }, + { + "epoch": 17.0, + "learning_rate": 8.131108864172565e-06, + "loss": 0.0, + "step": 67540 + }, + { + "epoch": 17.0, + "learning_rate": 8.124087181215594e-06, + "loss": 0.0068, + "step": 67550 + }, + { + "epoch": 17.0, + "learning_rate": 8.117065498258623e-06, + "loss": 0.0265, + "step": 67560 + }, + { + "epoch": 17.0, + "learning_rate": 8.110043815301651e-06, + "loss": 0.0, + "step": 67570 + }, + { + "epoch": 17.0, + "learning_rate": 8.103022132344682e-06, + "loss": 0.0, + "step": 67580 + }, + { + "epoch": 17.0, + "learning_rate": 8.09600044938771e-06, + "loss": 0.0001, + "step": 67590 + }, + { + "epoch": 17.0, + "learning_rate": 8.088978766430739e-06, + "loss": 0.0, + "step": 67600 + }, + { + "epoch": 17.0, + "learning_rate": 8.081957083473768e-06, + "loss": 0.0, + "step": 67610 + }, + { + "epoch": 17.0, + "learning_rate": 8.074935400516797e-06, + "loss": 0.0, + "step": 67620 + }, + { + "epoch": 17.0, + "learning_rate": 8.067913717559824e-06, + "loss": 0.0, + "step": 67630 + }, + { + "epoch": 17.0, + "learning_rate": 8.060892034602854e-06, + "loss": 0.0, + "step": 67640 + }, + { + "epoch": 17.01, + "learning_rate": 8.053870351645883e-06, + "loss": 0.0, + "step": 67650 + }, + { + "epoch": 17.01, + "learning_rate": 8.04684866868891e-06, + "loss": 0.0, + "step": 67660 + }, + { + "epoch": 17.01, + "learning_rate": 8.039826985731941e-06, + "loss": 0.0, + "step": 67670 + }, + { + "epoch": 17.01, + "learning_rate": 8.032805302774969e-06, + "loss": 0.0002, + "step": 67680 + }, + { + "epoch": 17.01, + "learning_rate": 8.025783619818e-06, + "loss": 0.0, + "step": 67690 + }, + { + "epoch": 17.01, + "learning_rate": 8.018761936861027e-06, + "loss": 0.0, + "step": 67700 + }, + { + "epoch": 17.01, + "learning_rate": 8.011740253904056e-06, + "loss": 0.0, + "step": 67710 + }, + { + "epoch": 17.01, + "learning_rate": 8.004718570947085e-06, + "loss": 0.3215, + "step": 67720 + }, + { + "epoch": 17.01, + "learning_rate": 7.997696887990115e-06, + "loss": 0.0, + "step": 67730 + }, + { + "epoch": 17.01, + "learning_rate": 7.990675205033142e-06, + "loss": 0.0, + "step": 67740 + }, + { + "epoch": 17.01, + "learning_rate": 7.983653522076171e-06, + "loss": 0.0, + "step": 67750 + }, + { + "epoch": 17.01, + "learning_rate": 7.9766318391192e-06, + "loss": 0.0, + "step": 67760 + }, + { + "epoch": 17.01, + "learning_rate": 7.96961015616223e-06, + "loss": 0.0, + "step": 67770 + }, + { + "epoch": 17.01, + "learning_rate": 7.962588473205259e-06, + "loss": 0.0, + "step": 67780 + }, + { + "epoch": 17.01, + "learning_rate": 7.955566790248286e-06, + "loss": 0.0, + "step": 67790 + }, + { + "epoch": 17.01, + "learning_rate": 7.948545107291317e-06, + "loss": 0.0, + "step": 67800 + }, + { + "epoch": 17.01, + "learning_rate": 7.941523424334345e-06, + "loss": 0.1635, + "step": 67810 + }, + { + "epoch": 17.01, + "learning_rate": 7.934501741377374e-06, + "loss": 0.0, + "step": 67820 + }, + { + "epoch": 17.01, + "learning_rate": 7.927480058420403e-06, + "loss": 0.0, + "step": 67830 + }, + { + "epoch": 17.01, + "learning_rate": 7.92045837546343e-06, + "loss": 0.0, + "step": 67840 + }, + { + "epoch": 17.01, + "learning_rate": 7.913436692506461e-06, + "loss": 0.0108, + "step": 67850 + }, + { + "epoch": 17.01, + "learning_rate": 7.906415009549489e-06, + "loss": 0.0, + "step": 67860 + }, + { + "epoch": 17.01, + "learning_rate": 7.899393326592518e-06, + "loss": 0.0, + "step": 67870 + }, + { + "epoch": 17.01, + "learning_rate": 7.892371643635547e-06, + "loss": 0.0, + "step": 67880 + }, + { + "epoch": 17.01, + "learning_rate": 7.885349960678576e-06, + "loss": 0.0, + "step": 67890 + }, + { + "epoch": 17.01, + "learning_rate": 7.878328277721604e-06, + "loss": 0.0, + "step": 67900 + }, + { + "epoch": 17.01, + "learning_rate": 7.871306594764633e-06, + "loss": 0.0, + "step": 67910 + }, + { + "epoch": 17.01, + "learning_rate": 7.864284911807662e-06, + "loss": 0.0009, + "step": 67920 + }, + { + "epoch": 17.01, + "learning_rate": 7.857263228850691e-06, + "loss": 0.0, + "step": 67930 + }, + { + "epoch": 17.01, + "learning_rate": 7.85024154589372e-06, + "loss": 0.0, + "step": 67940 + }, + { + "epoch": 17.01, + "learning_rate": 7.843219862936748e-06, + "loss": 0.0, + "step": 67950 + }, + { + "epoch": 17.01, + "learning_rate": 7.836198179979779e-06, + "loss": 0.0717, + "step": 67960 + }, + { + "epoch": 17.01, + "learning_rate": 7.829176497022806e-06, + "loss": 0.0, + "step": 67970 + }, + { + "epoch": 17.01, + "learning_rate": 7.822154814065836e-06, + "loss": 0.1188, + "step": 67980 + }, + { + "epoch": 17.01, + "learning_rate": 7.815133131108865e-06, + "loss": 0.0, + "step": 67990 + }, + { + "epoch": 17.01, + "learning_rate": 7.808111448151894e-06, + "loss": 0.0614, + "step": 68000 + }, + { + "epoch": 17.01, + "learning_rate": 7.801089765194922e-06, + "loss": 0.0, + "step": 68010 + }, + { + "epoch": 17.01, + "learning_rate": 7.79406808223795e-06, + "loss": 0.0, + "step": 68020 + }, + { + "epoch": 17.01, + "learning_rate": 7.78704639928098e-06, + "loss": 0.0001, + "step": 68030 + }, + { + "epoch": 17.01, + "learning_rate": 7.780024716324009e-06, + "loss": 0.0, + "step": 68040 + }, + { + "epoch": 17.01, + "learning_rate": 7.773003033367038e-06, + "loss": 0.0, + "step": 68050 + }, + { + "epoch": 17.01, + "learning_rate": 7.765981350410066e-06, + "loss": 0.0007, + "step": 68060 + }, + { + "epoch": 17.01, + "learning_rate": 7.758959667453097e-06, + "loss": 0.0, + "step": 68070 + }, + { + "epoch": 17.01, + "learning_rate": 7.751937984496124e-06, + "loss": 0.0001, + "step": 68080 + }, + { + "epoch": 17.01, + "learning_rate": 7.744916301539153e-06, + "loss": 0.0, + "step": 68090 + }, + { + "epoch": 17.01, + "learning_rate": 7.737894618582182e-06, + "loss": 0.0, + "step": 68100 + }, + { + "epoch": 17.01, + "learning_rate": 7.73087293562521e-06, + "loss": 0.0, + "step": 68110 + }, + { + "epoch": 17.01, + "learning_rate": 7.723851252668241e-06, + "loss": 0.0001, + "step": 68120 + }, + { + "epoch": 17.01, + "learning_rate": 7.716829569711268e-06, + "loss": 0.0, + "step": 68130 + }, + { + "epoch": 17.01, + "learning_rate": 7.709807886754298e-06, + "loss": 0.0, + "step": 68140 + }, + { + "epoch": 17.01, + "learning_rate": 7.702786203797327e-06, + "loss": 0.0, + "step": 68150 + }, + { + "epoch": 17.01, + "learning_rate": 7.695764520840356e-06, + "loss": 0.0, + "step": 68160 + }, + { + "epoch": 17.01, + "learning_rate": 7.688742837883383e-06, + "loss": 0.0, + "step": 68170 + }, + { + "epoch": 17.01, + "learning_rate": 7.681721154926413e-06, + "loss": 0.0, + "step": 68180 + }, + { + "epoch": 17.01, + "learning_rate": 7.674699471969442e-06, + "loss": 0.0, + "step": 68190 + }, + { + "epoch": 17.01, + "learning_rate": 7.667677789012471e-06, + "loss": 0.0, + "step": 68200 + }, + { + "epoch": 17.01, + "learning_rate": 7.6606561060555e-06, + "loss": 0.0, + "step": 68210 + }, + { + "epoch": 17.01, + "learning_rate": 7.653634423098528e-06, + "loss": 0.0, + "step": 68220 + }, + { + "epoch": 17.01, + "learning_rate": 7.646612740141559e-06, + "loss": 0.0, + "step": 68230 + }, + { + "epoch": 17.01, + "learning_rate": 7.639591057184586e-06, + "loss": 0.0, + "step": 68240 + }, + { + "epoch": 17.01, + "learning_rate": 7.632569374227615e-06, + "loss": 0.0911, + "step": 68250 + }, + { + "epoch": 17.01, + "learning_rate": 7.625547691270644e-06, + "loss": 0.0, + "step": 68260 + }, + { + "epoch": 17.01, + "learning_rate": 7.618526008313673e-06, + "loss": 0.0, + "step": 68270 + }, + { + "epoch": 17.01, + "learning_rate": 7.611504325356703e-06, + "loss": 0.5443, + "step": 68280 + }, + { + "epoch": 17.01, + "learning_rate": 7.604482642399731e-06, + "loss": 0.0, + "step": 68290 + }, + { + "epoch": 17.01, + "learning_rate": 7.597460959442759e-06, + "loss": 0.0, + "step": 68300 + }, + { + "epoch": 17.01, + "learning_rate": 7.590439276485789e-06, + "loss": 0.0, + "step": 68310 + }, + { + "epoch": 17.01, + "learning_rate": 7.583417593528818e-06, + "loss": 0.0, + "step": 68320 + }, + { + "epoch": 17.01, + "learning_rate": 7.576395910571846e-06, + "loss": 0.0, + "step": 68330 + }, + { + "epoch": 17.01, + "learning_rate": 7.569374227614875e-06, + "loss": 0.0, + "step": 68340 + }, + { + "epoch": 17.01, + "learning_rate": 7.562352544657904e-06, + "loss": 0.0005, + "step": 68350 + }, + { + "epoch": 17.01, + "learning_rate": 7.555330861700932e-06, + "loss": 0.0001, + "step": 68360 + }, + { + "epoch": 17.01, + "learning_rate": 7.548309178743962e-06, + "loss": 0.0, + "step": 68370 + }, + { + "epoch": 17.01, + "learning_rate": 7.54128749578699e-06, + "loss": 0.0, + "step": 68380 + }, + { + "epoch": 17.01, + "learning_rate": 7.53426581283002e-06, + "loss": 0.0, + "step": 68390 + }, + { + "epoch": 17.01, + "learning_rate": 7.527244129873049e-06, + "loss": 0.0009, + "step": 68400 + }, + { + "epoch": 17.01, + "learning_rate": 7.520222446916077e-06, + "loss": 0.0001, + "step": 68410 + }, + { + "epoch": 17.01, + "learning_rate": 7.513200763959106e-06, + "loss": 0.0, + "step": 68420 + }, + { + "epoch": 17.01, + "learning_rate": 7.5061790810021346e-06, + "loss": 0.0, + "step": 68430 + }, + { + "epoch": 17.02, + "learning_rate": 7.499157398045163e-06, + "loss": 0.0, + "step": 68440 + }, + { + "epoch": 17.02, + "learning_rate": 7.492135715088193e-06, + "loss": 0.0, + "step": 68450 + }, + { + "epoch": 17.02, + "learning_rate": 7.485114032131221e-06, + "loss": 0.0, + "step": 68460 + }, + { + "epoch": 17.02, + "learning_rate": 7.478092349174251e-06, + "loss": 0.0602, + "step": 68470 + }, + { + "epoch": 17.02, + "learning_rate": 7.47107066621728e-06, + "loss": 0.0, + "step": 68480 + }, + { + "epoch": 17.02, + "learning_rate": 7.464048983260308e-06, + "loss": 0.0, + "step": 68490 + }, + { + "epoch": 17.02, + "learning_rate": 7.457027300303337e-06, + "loss": 0.0, + "step": 68500 + }, + { + "epoch": 17.02, + "learning_rate": 7.4500056173463655e-06, + "loss": 0.0, + "step": 68510 + }, + { + "epoch": 17.02, + "learning_rate": 7.442983934389394e-06, + "loss": 0.0, + "step": 68520 + }, + { + "epoch": 17.02, + "learning_rate": 7.435962251432424e-06, + "loss": 0.0, + "step": 68530 + }, + { + "epoch": 17.02, + "learning_rate": 7.428940568475452e-06, + "loss": 0.0, + "step": 68540 + }, + { + "epoch": 17.02, + "learning_rate": 7.421918885518482e-06, + "loss": 0.0, + "step": 68550 + }, + { + "epoch": 17.02, + "learning_rate": 7.414897202561511e-06, + "loss": 0.0, + "step": 68560 + }, + { + "epoch": 17.02, + "learning_rate": 7.407875519604539e-06, + "loss": 0.0, + "step": 68570 + }, + { + "epoch": 17.02, + "learning_rate": 7.400853836647569e-06, + "loss": 0.0, + "step": 68580 + }, + { + "epoch": 17.02, + "learning_rate": 7.393832153690597e-06, + "loss": 0.0005, + "step": 68590 + }, + { + "epoch": 17.02, + "learning_rate": 7.386810470733626e-06, + "loss": 0.0, + "step": 68600 + }, + { + "epoch": 17.02, + "learning_rate": 7.379788787776655e-06, + "loss": 0.0, + "step": 68610 + }, + { + "epoch": 17.02, + "learning_rate": 7.372767104819683e-06, + "loss": 0.0, + "step": 68620 + }, + { + "epoch": 17.02, + "learning_rate": 7.3657454218627115e-06, + "loss": 0.0022, + "step": 68630 + }, + { + "epoch": 17.02, + "learning_rate": 7.3587237389057415e-06, + "loss": 0.0, + "step": 68640 + }, + { + "epoch": 17.02, + "learning_rate": 7.35170205594877e-06, + "loss": 0.0, + "step": 68650 + }, + { + "epoch": 17.02, + "learning_rate": 7.3446803729918e-06, + "loss": 0.0, + "step": 68660 + }, + { + "epoch": 17.02, + "learning_rate": 7.337658690034828e-06, + "loss": 0.0, + "step": 68670 + }, + { + "epoch": 17.02, + "learning_rate": 7.3306370070778566e-06, + "loss": 0.0, + "step": 68680 + }, + { + "epoch": 17.02, + "learning_rate": 7.323615324120886e-06, + "loss": 0.0003, + "step": 68690 + }, + { + "epoch": 17.02, + "learning_rate": 7.316593641163914e-06, + "loss": 0.0, + "step": 68700 + }, + { + "epoch": 17.02, + "learning_rate": 7.3095719582069424e-06, + "loss": 0.0, + "step": 68710 + }, + { + "epoch": 17.02, + "learning_rate": 7.3025502752499725e-06, + "loss": 0.0, + "step": 68720 + }, + { + "epoch": 17.02, + "learning_rate": 7.295528592293001e-06, + "loss": 0.0, + "step": 68730 + }, + { + "epoch": 17.02, + "learning_rate": 7.288506909336031e-06, + "loss": 0.0, + "step": 68740 + }, + { + "epoch": 17.02, + "learning_rate": 7.281485226379059e-06, + "loss": 0.0, + "step": 68750 + }, + { + "epoch": 17.02, + "learning_rate": 7.2744635434220875e-06, + "loss": 0.0, + "step": 68760 + }, + { + "epoch": 17.02, + "learning_rate": 7.267441860465117e-06, + "loss": 0.0, + "step": 68770 + }, + { + "epoch": 17.02, + "learning_rate": 7.260420177508145e-06, + "loss": 0.0, + "step": 68780 + }, + { + "epoch": 17.02, + "learning_rate": 7.253398494551173e-06, + "loss": 0.0, + "step": 68790 + }, + { + "epoch": 17.02, + "learning_rate": 7.246376811594203e-06, + "loss": 0.0, + "step": 68800 + }, + { + "epoch": 17.02, + "learning_rate": 7.239355128637232e-06, + "loss": 0.3315, + "step": 68810 + }, + { + "epoch": 17.02, + "learning_rate": 7.232333445680262e-06, + "loss": 0.0024, + "step": 68820 + }, + { + "epoch": 17.02, + "learning_rate": 7.22531176272329e-06, + "loss": 0.0, + "step": 68830 + }, + { + "epoch": 17.02, + "learning_rate": 7.2182900797663184e-06, + "loss": 0.0, + "step": 68840 + }, + { + "epoch": 17.02, + "learning_rate": 7.2112683968093485e-06, + "loss": 0.0, + "step": 68850 + }, + { + "epoch": 17.02, + "learning_rate": 7.204246713852377e-06, + "loss": 0.0, + "step": 68860 + }, + { + "epoch": 17.02, + "learning_rate": 7.197225030895405e-06, + "loss": 0.0, + "step": 68870 + }, + { + "epoch": 17.02, + "learning_rate": 7.190203347938434e-06, + "loss": 0.0001, + "step": 68880 + }, + { + "epoch": 17.02, + "learning_rate": 7.183181664981463e-06, + "loss": 0.0, + "step": 68890 + }, + { + "epoch": 17.02, + "learning_rate": 7.176159982024493e-06, + "loss": 0.0, + "step": 68900 + }, + { + "epoch": 17.02, + "learning_rate": 7.169138299067521e-06, + "loss": 0.0, + "step": 68910 + }, + { + "epoch": 17.02, + "learning_rate": 7.162116616110549e-06, + "loss": 0.0, + "step": 68920 + }, + { + "epoch": 17.02, + "learning_rate": 7.155094933153579e-06, + "loss": 0.0, + "step": 68930 + }, + { + "epoch": 17.02, + "learning_rate": 7.148073250196608e-06, + "loss": 0.0, + "step": 68940 + }, + { + "epoch": 17.02, + "learning_rate": 7.141051567239636e-06, + "loss": 0.0, + "step": 68950 + }, + { + "epoch": 17.02, + "learning_rate": 7.134029884282665e-06, + "loss": 0.0, + "step": 68960 + }, + { + "epoch": 17.02, + "learning_rate": 7.127008201325694e-06, + "loss": 0.0001, + "step": 68970 + }, + { + "epoch": 17.02, + "learning_rate": 7.119986518368722e-06, + "loss": 0.0, + "step": 68980 + }, + { + "epoch": 17.02, + "learning_rate": 7.112964835411752e-06, + "loss": 0.0, + "step": 68990 + }, + { + "epoch": 17.02, + "learning_rate": 7.10594315245478e-06, + "loss": 0.0, + "step": 69000 + }, + { + "epoch": 17.02, + "learning_rate": 7.09892146949781e-06, + "loss": 0.0108, + "step": 69010 + }, + { + "epoch": 17.02, + "learning_rate": 7.091899786540839e-06, + "loss": 0.0001, + "step": 69020 + }, + { + "epoch": 17.02, + "learning_rate": 7.084878103583867e-06, + "loss": 0.0139, + "step": 69030 + }, + { + "epoch": 17.02, + "learning_rate": 7.077856420626896e-06, + "loss": 0.0, + "step": 69040 + }, + { + "epoch": 17.02, + "learning_rate": 7.0708347376699245e-06, + "loss": 0.0, + "step": 69050 + }, + { + "epoch": 17.02, + "learning_rate": 7.063813054712953e-06, + "loss": 0.0, + "step": 69060 + }, + { + "epoch": 17.02, + "learning_rate": 7.056791371755983e-06, + "loss": 0.0002, + "step": 69070 + }, + { + "epoch": 17.02, + "learning_rate": 7.049769688799011e-06, + "loss": 0.0006, + "step": 69080 + }, + { + "epoch": 17.02, + "learning_rate": 7.042748005842041e-06, + "loss": 0.0002, + "step": 69090 + }, + { + "epoch": 17.02, + "learning_rate": 7.03572632288507e-06, + "loss": 0.0, + "step": 69100 + }, + { + "epoch": 17.02, + "learning_rate": 7.028704639928098e-06, + "loss": 0.0, + "step": 69110 + }, + { + "epoch": 17.02, + "learning_rate": 7.021682956971128e-06, + "loss": 0.0, + "step": 69120 + }, + { + "epoch": 17.02, + "learning_rate": 7.014661274014156e-06, + "loss": 0.0, + "step": 69130 + }, + { + "epoch": 17.02, + "learning_rate": 7.007639591057185e-06, + "loss": 0.0, + "step": 69140 + }, + { + "epoch": 17.02, + "learning_rate": 7.000617908100214e-06, + "loss": 0.0, + "step": 69150 + }, + { + "epoch": 17.02, + "learning_rate": 6.993596225143242e-06, + "loss": 0.0, + "step": 69160 + }, + { + "epoch": 17.02, + "learning_rate": 6.986574542186272e-06, + "loss": 0.0, + "step": 69170 + }, + { + "epoch": 17.02, + "learning_rate": 6.9795528592293005e-06, + "loss": 0.0, + "step": 69180 + }, + { + "epoch": 17.02, + "learning_rate": 6.972531176272329e-06, + "loss": 0.0, + "step": 69190 + }, + { + "epoch": 17.02, + "learning_rate": 6.965509493315359e-06, + "loss": 0.0, + "step": 69200 + }, + { + "epoch": 17.02, + "learning_rate": 6.958487810358387e-06, + "loss": 0.0, + "step": 69210 + }, + { + "epoch": 17.02, + "learning_rate": 6.951466127401416e-06, + "loss": 0.0, + "step": 69220 + }, + { + "epoch": 17.02, + "learning_rate": 6.944444444444445e-06, + "loss": 0.0002, + "step": 69230 + }, + { + "epoch": 17.03, + "learning_rate": 6.937422761487473e-06, + "loss": 0.0, + "step": 69240 + }, + { + "epoch": 17.03, + "learning_rate": 6.930401078530503e-06, + "loss": 0.0, + "step": 69250 + }, + { + "epoch": 17.03, + "learning_rate": 6.9233793955735315e-06, + "loss": 0.0, + "step": 69260 + }, + { + "epoch": 17.03, + "learning_rate": 6.91635771261656e-06, + "loss": 0.0, + "step": 69270 + }, + { + "epoch": 17.03, + "learning_rate": 6.90933602965959e-06, + "loss": 0.0, + "step": 69280 + }, + { + "epoch": 17.03, + "learning_rate": 6.902314346702618e-06, + "loss": 0.0, + "step": 69290 + }, + { + "epoch": 17.03, + "learning_rate": 6.8952926637456465e-06, + "loss": 0.0001, + "step": 69300 + }, + { + "epoch": 17.03, + "learning_rate": 6.888270980788676e-06, + "loss": 0.0, + "step": 69310 + }, + { + "epoch": 17.03, + "learning_rate": 6.881249297831704e-06, + "loss": 0.0, + "step": 69320 + }, + { + "epoch": 17.03, + "learning_rate": 6.874227614874733e-06, + "loss": 0.0, + "step": 69330 + }, + { + "epoch": 17.03, + "learning_rate": 6.867205931917762e-06, + "loss": 0.0, + "step": 69340 + }, + { + "epoch": 17.03, + "learning_rate": 6.860184248960791e-06, + "loss": 0.5607, + "step": 69350 + }, + { + "epoch": 17.03, + "learning_rate": 6.853162566003821e-06, + "loss": 0.1465, + "step": 69360 + }, + { + "epoch": 17.03, + "learning_rate": 6.846140883046849e-06, + "loss": 0.0, + "step": 69370 + }, + { + "epoch": 17.03, + "learning_rate": 6.8391192000898775e-06, + "loss": 0.3887, + "step": 69380 + }, + { + "epoch": 17.03, + "learning_rate": 6.8320975171329075e-06, + "loss": 0.1948, + "step": 69390 + }, + { + "epoch": 17.03, + "learning_rate": 6.825075834175936e-06, + "loss": 0.0, + "step": 69400 + }, + { + "epoch": 17.03, + "learning_rate": 6.818054151218964e-06, + "loss": 0.0, + "step": 69410 + }, + { + "epoch": 17.03, + "learning_rate": 6.811032468261993e-06, + "loss": 0.0, + "step": 69420 + }, + { + "epoch": 17.03, + "learning_rate": 6.804010785305022e-06, + "loss": 0.0, + "step": 69430 + }, + { + "epoch": 17.03, + "learning_rate": 6.796989102348052e-06, + "loss": 0.0, + "step": 69440 + }, + { + "epoch": 17.03, + "learning_rate": 6.78996741939108e-06, + "loss": 0.0, + "step": 69450 + }, + { + "epoch": 17.03, + "learning_rate": 6.782945736434108e-06, + "loss": 0.0, + "step": 69460 + }, + { + "epoch": 17.03, + "learning_rate": 6.7759240534771384e-06, + "loss": 0.0, + "step": 69470 + }, + { + "epoch": 17.03, + "learning_rate": 6.768902370520167e-06, + "loss": 0.0007, + "step": 69480 + }, + { + "epoch": 17.03, + "learning_rate": 6.761880687563195e-06, + "loss": 0.0, + "step": 69490 + }, + { + "epoch": 17.03, + "learning_rate": 6.754859004606224e-06, + "loss": 0.0, + "step": 69500 + }, + { + "epoch": 17.03, + "learning_rate": 6.747837321649253e-06, + "loss": 0.2926, + "step": 69510 + }, + { + "epoch": 17.03, + "learning_rate": 6.740815638692283e-06, + "loss": 0.0042, + "step": 69520 + }, + { + "epoch": 17.03, + "learning_rate": 6.733793955735311e-06, + "loss": 0.0, + "step": 69530 + }, + { + "epoch": 17.03, + "learning_rate": 6.726772272778339e-06, + "loss": 0.0, + "step": 69540 + }, + { + "epoch": 17.03, + "learning_rate": 6.719750589821369e-06, + "loss": 0.0, + "step": 69550 + }, + { + "epoch": 17.03, + "learning_rate": 6.712728906864398e-06, + "loss": 0.0, + "step": 69560 + }, + { + "epoch": 17.03, + "learning_rate": 6.705707223907426e-06, + "loss": 0.0, + "step": 69570 + }, + { + "epoch": 17.03, + "learning_rate": 6.698685540950455e-06, + "loss": 0.0, + "step": 69580 + }, + { + "epoch": 17.03, + "learning_rate": 6.691663857993484e-06, + "loss": 0.0013, + "step": 69590 + }, + { + "epoch": 17.03, + "learning_rate": 6.684642175036514e-06, + "loss": 0.0001, + "step": 69600 + }, + { + "epoch": 17.03, + "learning_rate": 6.677620492079542e-06, + "loss": 0.0, + "step": 69610 + }, + { + "epoch": 17.03, + "learning_rate": 6.67059880912257e-06, + "loss": 0.0, + "step": 69620 + }, + { + "epoch": 17.03, + "learning_rate": 6.6635771261656e-06, + "loss": 0.0, + "step": 69630 + }, + { + "epoch": 17.03, + "learning_rate": 6.656555443208629e-06, + "loss": 0.0, + "step": 69640 + }, + { + "epoch": 17.03, + "learning_rate": 6.649533760251657e-06, + "loss": 0.0, + "step": 69650 + }, + { + "epoch": 17.03, + "learning_rate": 6.642512077294687e-06, + "loss": 0.0035, + "step": 69660 + }, + { + "epoch": 17.03, + "learning_rate": 6.635490394337715e-06, + "loss": 0.0, + "step": 69670 + }, + { + "epoch": 17.03, + "learning_rate": 6.628468711380744e-06, + "loss": 0.0, + "step": 69680 + }, + { + "epoch": 17.03, + "learning_rate": 6.621447028423773e-06, + "loss": 0.4267, + "step": 69690 + }, + { + "epoch": 17.03, + "learning_rate": 6.614425345466801e-06, + "loss": 0.0005, + "step": 69700 + }, + { + "epoch": 17.03, + "learning_rate": 6.607403662509831e-06, + "loss": 0.0001, + "step": 69710 + }, + { + "epoch": 17.03, + "learning_rate": 6.6003819795528596e-06, + "loss": 0.0, + "step": 69720 + }, + { + "epoch": 17.03, + "learning_rate": 6.593360296595888e-06, + "loss": 0.0, + "step": 69730 + }, + { + "epoch": 17.03, + "learning_rate": 6.586338613638918e-06, + "loss": 0.0, + "step": 69740 + }, + { + "epoch": 17.03, + "learning_rate": 6.579316930681946e-06, + "loss": 0.0, + "step": 69750 + }, + { + "epoch": 17.03, + "learning_rate": 6.572295247724975e-06, + "loss": 0.0115, + "step": 69760 + }, + { + "epoch": 17.03, + "learning_rate": 6.565273564768004e-06, + "loss": 0.0001, + "step": 69770 + }, + { + "epoch": 17.03, + "learning_rate": 6.558251881811032e-06, + "loss": 0.0, + "step": 69780 + }, + { + "epoch": 17.03, + "learning_rate": 6.551230198854062e-06, + "loss": 0.0, + "step": 69790 + }, + { + "epoch": 17.03, + "learning_rate": 6.5442085158970905e-06, + "loss": 0.0, + "step": 69800 + }, + { + "epoch": 17.03, + "learning_rate": 6.537186832940119e-06, + "loss": 0.0, + "step": 69810 + }, + { + "epoch": 17.03, + "learning_rate": 6.530165149983149e-06, + "loss": 0.0, + "step": 69820 + }, + { + "epoch": 17.03, + "learning_rate": 6.523143467026177e-06, + "loss": 0.0, + "step": 69830 + }, + { + "epoch": 17.03, + "learning_rate": 6.5161217840692056e-06, + "loss": 0.0, + "step": 69840 + }, + { + "epoch": 17.03, + "learning_rate": 6.509100101112236e-06, + "loss": 0.0, + "step": 69850 + }, + { + "epoch": 17.03, + "learning_rate": 6.502078418155264e-06, + "loss": 0.0, + "step": 69860 + }, + { + "epoch": 17.03, + "learning_rate": 6.495056735198293e-06, + "loss": 0.0, + "step": 69870 + }, + { + "epoch": 17.03, + "learning_rate": 6.4880350522413214e-06, + "loss": 0.0, + "step": 69880 + }, + { + "epoch": 17.03, + "learning_rate": 6.48101336928435e-06, + "loss": 0.3839, + "step": 69890 + }, + { + "epoch": 17.03, + "learning_rate": 6.47399168632738e-06, + "loss": 0.0, + "step": 69900 + }, + { + "epoch": 17.03, + "learning_rate": 6.466970003370408e-06, + "loss": 0.6541, + "step": 69910 + }, + { + "epoch": 17.03, + "learning_rate": 6.4599483204134365e-06, + "loss": 0.0, + "step": 69920 + }, + { + "epoch": 17.03, + "learning_rate": 6.4529266374564665e-06, + "loss": 0.0, + "step": 69930 + }, + { + "epoch": 17.03, + "learning_rate": 6.445904954499495e-06, + "loss": 0.0, + "step": 69940 + }, + { + "epoch": 17.03, + "learning_rate": 6.438883271542524e-06, + "loss": 0.0, + "step": 69950 + }, + { + "epoch": 17.03, + "learning_rate": 6.431861588585552e-06, + "loss": 0.0, + "step": 69960 + }, + { + "epoch": 17.03, + "learning_rate": 6.424839905628581e-06, + "loss": 0.0, + "step": 69970 + }, + { + "epoch": 17.03, + "learning_rate": 6.417818222671611e-06, + "loss": 0.0, + "step": 69980 + }, + { + "epoch": 17.03, + "learning_rate": 6.410796539714639e-06, + "loss": 0.0, + "step": 69990 + }, + { + "epoch": 17.03, + "learning_rate": 6.4037748567576674e-06, + "loss": 0.0, + "step": 70000 + }, + { + "epoch": 17.03, + "learning_rate": 6.3967531738006975e-06, + "loss": 0.0, + "step": 70010 + }, + { + "epoch": 17.03, + "learning_rate": 6.389731490843726e-06, + "loss": 0.0, + "step": 70020 + }, + { + "epoch": 17.04, + "learning_rate": 6.382709807886754e-06, + "loss": 0.0, + "step": 70030 + }, + { + "epoch": 17.04, + "learning_rate": 6.375688124929783e-06, + "loss": 0.0001, + "step": 70040 + }, + { + "epoch": 17.04, + "learning_rate": 6.368666441972812e-06, + "loss": 0.0045, + "step": 70050 + }, + { + "epoch": 17.04, + "learning_rate": 6.361644759015842e-06, + "loss": 0.0, + "step": 70060 + }, + { + "epoch": 17.04, + "learning_rate": 6.35462307605887e-06, + "loss": 0.0, + "step": 70070 + }, + { + "epoch": 17.04, + "learning_rate": 6.347601393101898e-06, + "loss": 0.0, + "step": 70080 + }, + { + "epoch": 17.04, + "learning_rate": 6.340579710144928e-06, + "loss": 0.0, + "step": 70090 + }, + { + "epoch": 17.04, + "learning_rate": 6.333558027187957e-06, + "loss": 0.0, + "step": 70100 + }, + { + "epoch": 17.04, + "learning_rate": 6.326536344230985e-06, + "loss": 0.0, + "step": 70110 + }, + { + "epoch": 17.04, + "learning_rate": 6.319514661274015e-06, + "loss": 0.0, + "step": 70120 + }, + { + "epoch": 17.04, + "learning_rate": 6.3124929783170434e-06, + "loss": 0.0, + "step": 70130 + }, + { + "epoch": 17.04, + "learning_rate": 6.305471295360073e-06, + "loss": 0.0, + "step": 70140 + }, + { + "epoch": 17.04, + "learning_rate": 6.298449612403101e-06, + "loss": 0.0064, + "step": 70150 + }, + { + "epoch": 17.04, + "learning_rate": 6.291427929446129e-06, + "loss": 0.0, + "step": 70160 + }, + { + "epoch": 17.04, + "learning_rate": 6.284406246489159e-06, + "loss": 0.0, + "step": 70170 + }, + { + "epoch": 17.04, + "learning_rate": 6.277384563532188e-06, + "loss": 0.0, + "step": 70180 + }, + { + "epoch": 17.04, + "learning_rate": 6.270362880575216e-06, + "loss": 0.0, + "step": 70190 + }, + { + "epoch": 17.04, + "learning_rate": 6.263341197618246e-06, + "loss": 0.0, + "step": 70200 + }, + { + "epoch": 17.04, + "learning_rate": 6.256319514661274e-06, + "loss": 0.0, + "step": 70210 + }, + { + "epoch": 17.04, + "learning_rate": 6.2492978317043036e-06, + "loss": 0.0, + "step": 70220 + }, + { + "epoch": 17.04, + "learning_rate": 6.242276148747332e-06, + "loss": 0.0, + "step": 70230 + }, + { + "epoch": 17.04, + "learning_rate": 6.235254465790361e-06, + "loss": 0.161, + "step": 70240 + }, + { + "epoch": 17.04, + "learning_rate": 6.228232782833389e-06, + "loss": 0.0, + "step": 70250 + }, + { + "epoch": 17.04, + "learning_rate": 6.221211099876419e-06, + "loss": 0.0, + "step": 70260 + }, + { + "epoch": 17.04, + "learning_rate": 6.214189416919448e-06, + "loss": 0.0, + "step": 70270 + }, + { + "epoch": 17.04, + "learning_rate": 6.207167733962477e-06, + "loss": 0.0, + "step": 70280 + }, + { + "epoch": 17.04, + "learning_rate": 6.200146051005505e-06, + "loss": 0.0001, + "step": 70290 + }, + { + "epoch": 17.04, + "learning_rate": 6.1931243680485345e-06, + "loss": 0.0, + "step": 70300 + }, + { + "epoch": 17.04, + "learning_rate": 6.186102685091563e-06, + "loss": 0.0, + "step": 70310 + }, + { + "epoch": 17.04, + "learning_rate": 6.179081002134592e-06, + "loss": 0.0, + "step": 70320 + }, + { + "epoch": 17.04, + "learning_rate": 6.17205931917762e-06, + "loss": 0.0, + "step": 70330 + }, + { + "epoch": 17.04, + "learning_rate": 6.1650376362206495e-06, + "loss": 0.0001, + "step": 70340 + }, + { + "epoch": 17.04, + "learning_rate": 6.158015953263679e-06, + "loss": 0.0, + "step": 70350 + }, + { + "epoch": 17.04, + "learning_rate": 6.150994270306707e-06, + "loss": 0.0002, + "step": 70360 + }, + { + "epoch": 17.04, + "learning_rate": 6.143972587349736e-06, + "loss": 0.5287, + "step": 70370 + }, + { + "epoch": 17.04, + "learning_rate": 6.1369509043927654e-06, + "loss": 0.0, + "step": 70380 + }, + { + "epoch": 17.04, + "learning_rate": 6.129929221435795e-06, + "loss": 0.0, + "step": 70390 + }, + { + "epoch": 17.04, + "learning_rate": 6.122907538478823e-06, + "loss": 0.0, + "step": 70400 + }, + { + "epoch": 17.04, + "learning_rate": 6.115885855521851e-06, + "loss": 0.0, + "step": 70410 + }, + { + "epoch": 17.04, + "learning_rate": 6.1088641725648805e-06, + "loss": 0.0025, + "step": 70420 + }, + { + "epoch": 17.04, + "learning_rate": 6.10184248960791e-06, + "loss": 0.0, + "step": 70430 + }, + { + "epoch": 17.04, + "learning_rate": 6.094820806650938e-06, + "loss": 0.067, + "step": 70440 + }, + { + "epoch": 17.04, + "learning_rate": 6.087799123693967e-06, + "loss": 0.0, + "step": 70450 + }, + { + "epoch": 17.04, + "learning_rate": 6.080777440736996e-06, + "loss": 0.0, + "step": 70460 + }, + { + "epoch": 17.04, + "learning_rate": 6.0737557577800255e-06, + "loss": 0.0, + "step": 70470 + }, + { + "epoch": 17.04, + "learning_rate": 6.066734074823054e-06, + "loss": 0.0, + "step": 70480 + }, + { + "epoch": 17.04, + "learning_rate": 6.059712391866083e-06, + "loss": 0.1261, + "step": 70490 + }, + { + "epoch": 17.04, + "learning_rate": 6.052690708909111e-06, + "loss": 0.0, + "step": 70500 + }, + { + "epoch": 17.04, + "learning_rate": 6.045669025952141e-06, + "loss": 0.0, + "step": 70510 + }, + { + "epoch": 17.04, + "learning_rate": 6.038647342995169e-06, + "loss": 0.0002, + "step": 70520 + }, + { + "epoch": 17.04, + "learning_rate": 6.031625660038198e-06, + "loss": 0.4609, + "step": 70530 + }, + { + "epoch": 17.04, + "learning_rate": 6.024603977081227e-06, + "loss": 0.0, + "step": 70540 + }, + { + "epoch": 17.04, + "learning_rate": 6.0175822941242565e-06, + "loss": 0.0, + "step": 70550 + }, + { + "epoch": 17.04, + "learning_rate": 6.010560611167285e-06, + "loss": 0.0012, + "step": 70560 + }, + { + "epoch": 17.04, + "learning_rate": 6.003538928210314e-06, + "loss": 0.0, + "step": 70570 + }, + { + "epoch": 17.04, + "learning_rate": 5.996517245253342e-06, + "loss": 0.0401, + "step": 70580 + }, + { + "epoch": 17.04, + "learning_rate": 5.9894955622963715e-06, + "loss": 0.0, + "step": 70590 + }, + { + "epoch": 17.04, + "learning_rate": 5.9824738793394e-06, + "loss": 0.0021, + "step": 70600 + }, + { + "epoch": 17.04, + "learning_rate": 5.975452196382429e-06, + "loss": 0.001, + "step": 70610 + }, + { + "epoch": 17.04, + "learning_rate": 5.968430513425458e-06, + "loss": 0.0001, + "step": 70620 + }, + { + "epoch": 17.04, + "learning_rate": 5.961408830468487e-06, + "loss": 0.0004, + "step": 70630 + }, + { + "epoch": 17.04, + "learning_rate": 5.954387147511516e-06, + "loss": 0.0, + "step": 70640 + }, + { + "epoch": 17.04, + "learning_rate": 5.947365464554545e-06, + "loss": 0.0883, + "step": 70650 + }, + { + "epoch": 17.04, + "learning_rate": 5.940343781597574e-06, + "loss": 0.0, + "step": 70660 + }, + { + "epoch": 17.04, + "learning_rate": 5.9333220986406025e-06, + "loss": 0.0, + "step": 70670 + }, + { + "epoch": 17.04, + "learning_rate": 5.926300415683631e-06, + "loss": 0.0, + "step": 70680 + }, + { + "epoch": 17.04, + "learning_rate": 5.91927873272666e-06, + "loss": 0.0, + "step": 70690 + }, + { + "epoch": 17.04, + "learning_rate": 5.912257049769689e-06, + "loss": 0.0, + "step": 70700 + }, + { + "epoch": 17.04, + "learning_rate": 5.9052353668127175e-06, + "loss": 0.0002, + "step": 70710 + }, + { + "epoch": 17.04, + "learning_rate": 5.898213683855747e-06, + "loss": 0.0, + "step": 70720 + }, + { + "epoch": 17.04, + "learning_rate": 5.891192000898776e-06, + "loss": 0.0, + "step": 70730 + }, + { + "epoch": 17.04, + "learning_rate": 5.884170317941805e-06, + "loss": 0.0367, + "step": 70740 + }, + { + "epoch": 17.04, + "learning_rate": 5.877148634984833e-06, + "loss": 0.0, + "step": 70750 + }, + { + "epoch": 17.04, + "learning_rate": 5.870126952027863e-06, + "loss": 0.0, + "step": 70760 + }, + { + "epoch": 17.04, + "learning_rate": 5.863105269070891e-06, + "loss": 0.0, + "step": 70770 + }, + { + "epoch": 17.04, + "learning_rate": 5.85608358611392e-06, + "loss": 0.0, + "step": 70780 + }, + { + "epoch": 17.04, + "learning_rate": 5.8490619031569484e-06, + "loss": 0.0, + "step": 70790 + }, + { + "epoch": 17.04, + "learning_rate": 5.842040220199978e-06, + "loss": 0.0179, + "step": 70800 + }, + { + "epoch": 17.04, + "learning_rate": 5.835018537243007e-06, + "loss": 0.0003, + "step": 70810 + }, + { + "epoch": 17.05, + "learning_rate": 5.827996854286036e-06, + "loss": 0.0, + "step": 70820 + }, + { + "epoch": 17.05, + "learning_rate": 5.820975171329064e-06, + "loss": 0.0, + "step": 70830 + }, + { + "epoch": 17.05, + "learning_rate": 5.8139534883720935e-06, + "loss": 0.0, + "step": 70840 + }, + { + "epoch": 17.05, + "learning_rate": 5.806931805415122e-06, + "loss": 0.0, + "step": 70850 + }, + { + "epoch": 17.05, + "learning_rate": 5.799910122458151e-06, + "loss": 0.0, + "step": 70860 + }, + { + "epoch": 17.05, + "learning_rate": 5.792888439501179e-06, + "loss": 0.0, + "step": 70870 + }, + { + "epoch": 17.05, + "learning_rate": 5.7858667565442086e-06, + "loss": 0.0, + "step": 70880 + }, + { + "epoch": 17.05, + "learning_rate": 5.778845073587238e-06, + "loss": 0.0, + "step": 70890 + }, + { + "epoch": 17.05, + "learning_rate": 5.771823390630267e-06, + "loss": 0.0, + "step": 70900 + }, + { + "epoch": 17.05, + "learning_rate": 5.764801707673295e-06, + "loss": 0.0, + "step": 70910 + }, + { + "epoch": 17.05, + "learning_rate": 5.7577800247163245e-06, + "loss": 0.0, + "step": 70920 + }, + { + "epoch": 17.05, + "learning_rate": 5.750758341759354e-06, + "loss": 0.0, + "step": 70930 + }, + { + "epoch": 17.05, + "learning_rate": 5.743736658802382e-06, + "loss": 0.0, + "step": 70940 + }, + { + "epoch": 17.05, + "learning_rate": 5.73671497584541e-06, + "loss": 0.0, + "step": 70950 + }, + { + "epoch": 17.05, + "learning_rate": 5.7296932928884395e-06, + "loss": 0.0, + "step": 70960 + }, + { + "epoch": 17.05, + "learning_rate": 5.722671609931469e-06, + "loss": 0.0046, + "step": 70970 + }, + { + "epoch": 17.05, + "learning_rate": 5.715649926974498e-06, + "loss": 0.0, + "step": 70980 + }, + { + "epoch": 17.05, + "learning_rate": 5.708628244017526e-06, + "loss": 0.0, + "step": 70990 + }, + { + "epoch": 17.05, + "learning_rate": 5.701606561060555e-06, + "loss": 0.0002, + "step": 71000 + }, + { + "epoch": 17.05, + "learning_rate": 5.6945848781035846e-06, + "loss": 0.0002, + "step": 71010 + }, + { + "epoch": 17.05, + "learning_rate": 5.687563195146613e-06, + "loss": 0.0, + "step": 71020 + }, + { + "epoch": 17.05, + "learning_rate": 5.680541512189642e-06, + "loss": 0.0, + "step": 71030 + }, + { + "epoch": 17.05, + "learning_rate": 5.6735198292326704e-06, + "loss": 0.0018, + "step": 71040 + }, + { + "epoch": 17.05, + "learning_rate": 5.6664981462757e-06, + "loss": 0.0, + "step": 71050 + }, + { + "epoch": 17.05, + "learning_rate": 5.659476463318728e-06, + "loss": 0.0, + "step": 71060 + }, + { + "epoch": 17.05, + "learning_rate": 5.652454780361757e-06, + "loss": 0.0032, + "step": 71070 + }, + { + "epoch": 17.05, + "learning_rate": 5.645433097404786e-06, + "loss": 0.0039, + "step": 71080 + }, + { + "epoch": 17.05, + "learning_rate": 5.6384114144478155e-06, + "loss": 0.0, + "step": 71090 + }, + { + "epoch": 17.05, + "learning_rate": 5.631389731490844e-06, + "loss": 0.0, + "step": 71100 + }, + { + "epoch": 17.05, + "learning_rate": 5.624368048533873e-06, + "loss": 0.0, + "step": 71110 + }, + { + "epoch": 17.05, + "learning_rate": 5.617346365576902e-06, + "loss": 0.0, + "step": 71120 + }, + { + "epoch": 17.05, + "learning_rate": 5.6103246826199306e-06, + "loss": 0.0, + "step": 71130 + }, + { + "epoch": 17.05, + "learning_rate": 5.603302999662959e-06, + "loss": 0.0003, + "step": 71140 + }, + { + "epoch": 17.05, + "learning_rate": 5.596281316705988e-06, + "loss": 0.0011, + "step": 71150 + }, + { + "epoch": 17.05, + "learning_rate": 5.589259633749017e-06, + "loss": 0.0, + "step": 71160 + }, + { + "epoch": 17.05, + "learning_rate": 5.5822379507920464e-06, + "loss": 0.0, + "step": 71170 + }, + { + "epoch": 17.05, + "learning_rate": 5.575216267835075e-06, + "loss": 0.0, + "step": 71180 + }, + { + "epoch": 17.05, + "learning_rate": 5.568194584878104e-06, + "loss": 0.0, + "step": 71190 + }, + { + "epoch": 17.05, + "learning_rate": 5.561172901921133e-06, + "loss": 0.0416, + "step": 71200 + }, + { + "epoch": 17.05, + "eval_accuracy": 0.7637724164623919, + "eval_loss": 1.7013074159622192, + "eval_runtime": 1249.6172, + "eval_samples_per_second": 6.203, + "eval_steps_per_second": 3.102, + "step": 71208 + }, + { + "epoch": 18.0, + "learning_rate": 5.5541512189641615e-06, + "loss": 0.0005, + "step": 71210 + }, + { + "epoch": 18.0, + "learning_rate": 5.54712953600719e-06, + "loss": 0.0, + "step": 71220 + }, + { + "epoch": 18.0, + "learning_rate": 5.540107853050219e-06, + "loss": 0.0003, + "step": 71230 + }, + { + "epoch": 18.0, + "learning_rate": 5.533086170093248e-06, + "loss": 0.0, + "step": 71240 + }, + { + "epoch": 18.0, + "learning_rate": 5.526064487136277e-06, + "loss": 0.0026, + "step": 71250 + }, + { + "epoch": 18.0, + "learning_rate": 5.519042804179306e-06, + "loss": 0.0107, + "step": 71260 + }, + { + "epoch": 18.0, + "learning_rate": 5.512021121222335e-06, + "loss": 0.0, + "step": 71270 + }, + { + "epoch": 18.0, + "learning_rate": 5.504999438265364e-06, + "loss": 0.0, + "step": 71280 + }, + { + "epoch": 18.0, + "learning_rate": 5.497977755308393e-06, + "loss": 0.0, + "step": 71290 + }, + { + "epoch": 18.0, + "learning_rate": 5.490956072351422e-06, + "loss": 0.0, + "step": 71300 + }, + { + "epoch": 18.0, + "learning_rate": 5.48393438939445e-06, + "loss": 0.0, + "step": 71310 + }, + { + "epoch": 18.0, + "learning_rate": 5.476912706437479e-06, + "loss": 0.0, + "step": 71320 + }, + { + "epoch": 18.0, + "learning_rate": 5.469891023480508e-06, + "loss": 0.0, + "step": 71330 + }, + { + "epoch": 18.0, + "learning_rate": 5.462869340523537e-06, + "loss": 0.0, + "step": 71340 + }, + { + "epoch": 18.0, + "learning_rate": 5.455847657566566e-06, + "loss": 0.0001, + "step": 71350 + }, + { + "epoch": 18.0, + "learning_rate": 5.448825974609595e-06, + "loss": 0.0, + "step": 71360 + }, + { + "epoch": 18.0, + "learning_rate": 5.441804291652623e-06, + "loss": 0.0, + "step": 71370 + }, + { + "epoch": 18.0, + "learning_rate": 5.4347826086956525e-06, + "loss": 0.0, + "step": 71380 + }, + { + "epoch": 18.0, + "learning_rate": 5.427760925738682e-06, + "loss": 0.0005, + "step": 71390 + }, + { + "epoch": 18.0, + "learning_rate": 5.42073924278171e-06, + "loss": 0.0, + "step": 71400 + }, + { + "epoch": 18.0, + "learning_rate": 5.413717559824738e-06, + "loss": 0.0, + "step": 71410 + }, + { + "epoch": 18.0, + "learning_rate": 5.406695876867768e-06, + "loss": 0.0, + "step": 71420 + }, + { + "epoch": 18.0, + "learning_rate": 5.399674193910797e-06, + "loss": 0.0004, + "step": 71430 + }, + { + "epoch": 18.0, + "learning_rate": 5.392652510953826e-06, + "loss": 0.0, + "step": 71440 + }, + { + "epoch": 18.0, + "learning_rate": 5.385630827996854e-06, + "loss": 0.0, + "step": 71450 + }, + { + "epoch": 18.0, + "learning_rate": 5.3786091450398835e-06, + "loss": 0.0, + "step": 71460 + }, + { + "epoch": 18.0, + "learning_rate": 5.371587462082913e-06, + "loss": 0.0, + "step": 71470 + }, + { + "epoch": 18.0, + "learning_rate": 5.364565779125941e-06, + "loss": 0.001, + "step": 71480 + }, + { + "epoch": 18.0, + "learning_rate": 5.35754409616897e-06, + "loss": 0.0, + "step": 71490 + }, + { + "epoch": 18.0, + "learning_rate": 5.3505224132119985e-06, + "loss": 0.0, + "step": 71500 + }, + { + "epoch": 18.0, + "learning_rate": 5.343500730255028e-06, + "loss": 0.0, + "step": 71510 + }, + { + "epoch": 18.0, + "learning_rate": 5.336479047298057e-06, + "loss": 0.0001, + "step": 71520 + }, + { + "epoch": 18.0, + "learning_rate": 5.329457364341085e-06, + "loss": 0.0, + "step": 71530 + }, + { + "epoch": 18.0, + "learning_rate": 5.322435681384114e-06, + "loss": 0.0, + "step": 71540 + }, + { + "epoch": 18.0, + "learning_rate": 5.315413998427144e-06, + "loss": 0.0001, + "step": 71550 + }, + { + "epoch": 18.0, + "learning_rate": 5.308392315470173e-06, + "loss": 0.0, + "step": 71560 + }, + { + "epoch": 18.0, + "learning_rate": 5.301370632513201e-06, + "loss": 0.0001, + "step": 71570 + }, + { + "epoch": 18.0, + "learning_rate": 5.2943489495562295e-06, + "loss": 0.0, + "step": 71580 + }, + { + "epoch": 18.0, + "learning_rate": 5.287327266599259e-06, + "loss": 0.0, + "step": 71590 + }, + { + "epoch": 18.0, + "learning_rate": 5.280305583642288e-06, + "loss": 0.0, + "step": 71600 + }, + { + "epoch": 18.01, + "learning_rate": 5.273283900685316e-06, + "loss": 0.0, + "step": 71610 + }, + { + "epoch": 18.01, + "learning_rate": 5.266262217728345e-06, + "loss": 0.0, + "step": 71620 + }, + { + "epoch": 18.01, + "learning_rate": 5.2592405347713745e-06, + "loss": 0.0, + "step": 71630 + }, + { + "epoch": 18.01, + "learning_rate": 5.252218851814404e-06, + "loss": 0.0, + "step": 71640 + }, + { + "epoch": 18.01, + "learning_rate": 5.245197168857432e-06, + "loss": 0.0, + "step": 71650 + }, + { + "epoch": 18.01, + "learning_rate": 5.238175485900461e-06, + "loss": 0.0, + "step": 71660 + }, + { + "epoch": 18.01, + "learning_rate": 5.23115380294349e-06, + "loss": 0.0, + "step": 71670 + }, + { + "epoch": 18.01, + "learning_rate": 5.224132119986518e-06, + "loss": 0.0, + "step": 71680 + }, + { + "epoch": 18.01, + "learning_rate": 5.217110437029547e-06, + "loss": 0.0, + "step": 71690 + }, + { + "epoch": 18.01, + "learning_rate": 5.210088754072576e-06, + "loss": 0.0, + "step": 71700 + }, + { + "epoch": 18.01, + "learning_rate": 5.2030670711156055e-06, + "loss": 0.0, + "step": 71710 + }, + { + "epoch": 18.01, + "learning_rate": 5.196045388158634e-06, + "loss": 0.1824, + "step": 71720 + }, + { + "epoch": 18.01, + "learning_rate": 5.189023705201663e-06, + "loss": 0.0, + "step": 71730 + }, + { + "epoch": 18.01, + "learning_rate": 5.182002022244692e-06, + "loss": 0.0, + "step": 71740 + }, + { + "epoch": 18.01, + "learning_rate": 5.174980339287721e-06, + "loss": 0.4306, + "step": 71750 + }, + { + "epoch": 18.01, + "learning_rate": 5.16795865633075e-06, + "loss": 0.0001, + "step": 71760 + }, + { + "epoch": 18.01, + "learning_rate": 5.160936973373778e-06, + "loss": 0.0, + "step": 71770 + }, + { + "epoch": 18.01, + "learning_rate": 5.153915290416807e-06, + "loss": 0.0, + "step": 71780 + }, + { + "epoch": 18.01, + "learning_rate": 5.146893607459836e-06, + "loss": 0.0, + "step": 71790 + }, + { + "epoch": 18.01, + "learning_rate": 5.139871924502865e-06, + "loss": 0.0, + "step": 71800 + }, + { + "epoch": 18.01, + "learning_rate": 5.132850241545894e-06, + "loss": 0.0, + "step": 71810 + }, + { + "epoch": 18.01, + "learning_rate": 5.125828558588923e-06, + "loss": 0.0, + "step": 71820 + }, + { + "epoch": 18.01, + "learning_rate": 5.118806875631952e-06, + "loss": 0.0, + "step": 71830 + }, + { + "epoch": 18.01, + "learning_rate": 5.111785192674981e-06, + "loss": 0.0084, + "step": 71840 + }, + { + "epoch": 18.01, + "learning_rate": 5.104763509718009e-06, + "loss": 0.0, + "step": 71850 + }, + { + "epoch": 18.01, + "learning_rate": 5.097741826761038e-06, + "loss": 0.0, + "step": 71860 + }, + { + "epoch": 18.01, + "learning_rate": 5.090720143804067e-06, + "loss": 0.0, + "step": 71870 + }, + { + "epoch": 18.01, + "learning_rate": 5.083698460847096e-06, + "loss": 0.0, + "step": 71880 + }, + { + "epoch": 18.01, + "learning_rate": 5.076676777890125e-06, + "loss": 0.0, + "step": 71890 + }, + { + "epoch": 18.01, + "learning_rate": 5.069655094933154e-06, + "loss": 0.0, + "step": 71900 + }, + { + "epoch": 18.01, + "learning_rate": 5.062633411976183e-06, + "loss": 0.0, + "step": 71910 + }, + { + "epoch": 18.01, + "learning_rate": 5.0556117290192116e-06, + "loss": 0.0, + "step": 71920 + }, + { + "epoch": 18.01, + "learning_rate": 5.048590046062241e-06, + "loss": 0.0, + "step": 71930 + }, + { + "epoch": 18.01, + "learning_rate": 5.041568363105269e-06, + "loss": 0.0042, + "step": 71940 + }, + { + "epoch": 18.01, + "learning_rate": 5.034546680148298e-06, + "loss": 0.0178, + "step": 71950 + }, + { + "epoch": 18.01, + "learning_rate": 5.027524997191327e-06, + "loss": 0.0, + "step": 71960 + }, + { + "epoch": 18.01, + "learning_rate": 5.020503314234356e-06, + "loss": 0.0, + "step": 71970 + }, + { + "epoch": 18.01, + "learning_rate": 5.013481631277385e-06, + "loss": 0.0003, + "step": 71980 + }, + { + "epoch": 18.01, + "learning_rate": 5.006459948320414e-06, + "loss": 0.0, + "step": 71990 + }, + { + "epoch": 18.01, + "learning_rate": 4.9994382653634425e-06, + "loss": 0.0, + "step": 72000 + }, + { + "epoch": 18.01, + "learning_rate": 4.992416582406472e-06, + "loss": 0.0, + "step": 72010 + }, + { + "epoch": 18.01, + "learning_rate": 4.985394899449501e-06, + "loss": 0.0056, + "step": 72020 + }, + { + "epoch": 18.01, + "learning_rate": 4.978373216492529e-06, + "loss": 0.0, + "step": 72030 + }, + { + "epoch": 18.01, + "learning_rate": 4.9713515335355576e-06, + "loss": 0.0, + "step": 72040 + }, + { + "epoch": 18.01, + "learning_rate": 4.964329850578587e-06, + "loss": 0.0, + "step": 72050 + }, + { + "epoch": 18.01, + "learning_rate": 4.957308167621616e-06, + "loss": 0.0, + "step": 72060 + }, + { + "epoch": 18.01, + "learning_rate": 4.950286484664644e-06, + "loss": 0.0, + "step": 72070 + }, + { + "epoch": 18.01, + "learning_rate": 4.9432648017076734e-06, + "loss": 0.0, + "step": 72080 + }, + { + "epoch": 18.01, + "learning_rate": 4.936243118750703e-06, + "loss": 0.0, + "step": 72090 + }, + { + "epoch": 18.01, + "learning_rate": 4.929221435793732e-06, + "loss": 0.0, + "step": 72100 + }, + { + "epoch": 18.01, + "learning_rate": 4.92219975283676e-06, + "loss": 0.0, + "step": 72110 + }, + { + "epoch": 18.01, + "learning_rate": 4.9151780698797885e-06, + "loss": 0.0, + "step": 72120 + }, + { + "epoch": 18.01, + "learning_rate": 4.908156386922818e-06, + "loss": 0.0, + "step": 72130 + }, + { + "epoch": 18.01, + "learning_rate": 4.901134703965847e-06, + "loss": 0.0004, + "step": 72140 + }, + { + "epoch": 18.01, + "learning_rate": 4.894113021008875e-06, + "loss": 0.0, + "step": 72150 + }, + { + "epoch": 18.01, + "learning_rate": 4.887091338051904e-06, + "loss": 0.0, + "step": 72160 + }, + { + "epoch": 18.01, + "learning_rate": 4.8800696550949336e-06, + "loss": 0.0, + "step": 72170 + }, + { + "epoch": 18.01, + "learning_rate": 4.873047972137963e-06, + "loss": 0.0, + "step": 72180 + }, + { + "epoch": 18.01, + "learning_rate": 4.866026289180991e-06, + "loss": 0.0, + "step": 72190 + }, + { + "epoch": 18.01, + "learning_rate": 4.85900460622402e-06, + "loss": 0.002, + "step": 72200 + }, + { + "epoch": 18.01, + "learning_rate": 4.851982923267049e-06, + "loss": 0.0, + "step": 72210 + }, + { + "epoch": 18.01, + "learning_rate": 4.844961240310078e-06, + "loss": 0.0, + "step": 72220 + }, + { + "epoch": 18.01, + "learning_rate": 4.837939557353106e-06, + "loss": 0.1272, + "step": 72230 + }, + { + "epoch": 18.01, + "learning_rate": 4.830917874396135e-06, + "loss": 0.0, + "step": 72240 + }, + { + "epoch": 18.01, + "learning_rate": 4.8238961914391645e-06, + "loss": 0.0, + "step": 72250 + }, + { + "epoch": 18.01, + "learning_rate": 4.816874508482194e-06, + "loss": 0.0, + "step": 72260 + }, + { + "epoch": 18.01, + "learning_rate": 4.809852825525222e-06, + "loss": 0.0, + "step": 72270 + }, + { + "epoch": 18.01, + "learning_rate": 4.802831142568251e-06, + "loss": 0.0, + "step": 72280 + }, + { + "epoch": 18.01, + "learning_rate": 4.79580945961128e-06, + "loss": 0.0, + "step": 72290 + }, + { + "epoch": 18.01, + "learning_rate": 4.788787776654309e-06, + "loss": 0.0, + "step": 72300 + }, + { + "epoch": 18.01, + "learning_rate": 4.781766093697337e-06, + "loss": 0.0, + "step": 72310 + }, + { + "epoch": 18.01, + "learning_rate": 4.774744410740366e-06, + "loss": 0.0, + "step": 72320 + }, + { + "epoch": 18.01, + "learning_rate": 4.7677227277833954e-06, + "loss": 0.0, + "step": 72330 + }, + { + "epoch": 18.01, + "learning_rate": 4.760701044826425e-06, + "loss": 0.0, + "step": 72340 + }, + { + "epoch": 18.01, + "learning_rate": 4.753679361869453e-06, + "loss": 0.356, + "step": 72350 + }, + { + "epoch": 18.01, + "learning_rate": 4.746657678912482e-06, + "loss": 0.0, + "step": 72360 + }, + { + "epoch": 18.01, + "learning_rate": 4.739635995955511e-06, + "loss": 0.0, + "step": 72370 + }, + { + "epoch": 18.01, + "learning_rate": 4.73261431299854e-06, + "loss": 0.0, + "step": 72380 + }, + { + "epoch": 18.01, + "learning_rate": 4.725592630041569e-06, + "loss": 0.001, + "step": 72390 + }, + { + "epoch": 18.02, + "learning_rate": 4.718570947084597e-06, + "loss": 0.0, + "step": 72400 + }, + { + "epoch": 18.02, + "learning_rate": 4.711549264127626e-06, + "loss": 0.0, + "step": 72410 + }, + { + "epoch": 18.02, + "learning_rate": 4.704527581170655e-06, + "loss": 0.0, + "step": 72420 + }, + { + "epoch": 18.02, + "learning_rate": 4.697505898213684e-06, + "loss": 0.0, + "step": 72430 + }, + { + "epoch": 18.02, + "learning_rate": 4.690484215256713e-06, + "loss": 0.0, + "step": 72440 + }, + { + "epoch": 18.02, + "learning_rate": 4.683462532299742e-06, + "loss": 0.0007, + "step": 72450 + }, + { + "epoch": 18.02, + "learning_rate": 4.676440849342771e-06, + "loss": 0.0, + "step": 72460 + }, + { + "epoch": 18.02, + "learning_rate": 4.6694191663858e-06, + "loss": 0.0, + "step": 72470 + }, + { + "epoch": 18.02, + "learning_rate": 4.662397483428828e-06, + "loss": 0.0, + "step": 72480 + }, + { + "epoch": 18.02, + "learning_rate": 4.655375800471857e-06, + "loss": 0.0003, + "step": 72490 + }, + { + "epoch": 18.02, + "learning_rate": 4.648354117514886e-06, + "loss": 0.0, + "step": 72500 + }, + { + "epoch": 18.02, + "learning_rate": 4.641332434557915e-06, + "loss": 0.0, + "step": 72510 + }, + { + "epoch": 18.02, + "learning_rate": 4.634310751600944e-06, + "loss": 0.0002, + "step": 72520 + }, + { + "epoch": 18.02, + "learning_rate": 4.627289068643973e-06, + "loss": 0.0, + "step": 72530 + }, + { + "epoch": 18.02, + "learning_rate": 4.6202673856870015e-06, + "loss": 0.0005, + "step": 72540 + }, + { + "epoch": 18.02, + "learning_rate": 4.613245702730031e-06, + "loss": 0.0, + "step": 72550 + }, + { + "epoch": 18.02, + "learning_rate": 4.60622401977306e-06, + "loss": 0.0, + "step": 72560 + }, + { + "epoch": 18.02, + "learning_rate": 4.599202336816088e-06, + "loss": 0.0, + "step": 72570 + }, + { + "epoch": 18.02, + "learning_rate": 4.592180653859117e-06, + "loss": 0.0, + "step": 72580 + }, + { + "epoch": 18.02, + "learning_rate": 4.585158970902146e-06, + "loss": 0.0, + "step": 72590 + }, + { + "epoch": 18.02, + "learning_rate": 4.578137287945175e-06, + "loss": 0.0, + "step": 72600 + }, + { + "epoch": 18.02, + "learning_rate": 4.571115604988204e-06, + "loss": 0.0, + "step": 72610 + }, + { + "epoch": 18.02, + "learning_rate": 4.5640939220312325e-06, + "loss": 0.0, + "step": 72620 + }, + { + "epoch": 18.02, + "learning_rate": 4.557072239074262e-06, + "loss": 0.0, + "step": 72630 + }, + { + "epoch": 18.02, + "learning_rate": 4.550050556117291e-06, + "loss": 0.0504, + "step": 72640 + }, + { + "epoch": 18.02, + "learning_rate": 4.54302887316032e-06, + "loss": 0.0004, + "step": 72650 + }, + { + "epoch": 18.02, + "learning_rate": 4.536007190203348e-06, + "loss": 0.0, + "step": 72660 + }, + { + "epoch": 18.02, + "learning_rate": 4.528985507246377e-06, + "loss": 0.0, + "step": 72670 + }, + { + "epoch": 18.02, + "learning_rate": 4.521963824289406e-06, + "loss": 0.0, + "step": 72680 + }, + { + "epoch": 18.02, + "learning_rate": 4.514942141332434e-06, + "loss": 0.0, + "step": 72690 + }, + { + "epoch": 18.02, + "learning_rate": 4.507920458375463e-06, + "loss": 0.0, + "step": 72700 + }, + { + "epoch": 18.02, + "learning_rate": 4.500898775418493e-06, + "loss": 0.0, + "step": 72710 + }, + { + "epoch": 18.02, + "learning_rate": 4.493877092461522e-06, + "loss": 0.2625, + "step": 72720 + }, + { + "epoch": 18.02, + "learning_rate": 4.48685540950455e-06, + "loss": 0.0, + "step": 72730 + }, + { + "epoch": 18.02, + "learning_rate": 4.479833726547579e-06, + "loss": 0.0, + "step": 72740 + }, + { + "epoch": 18.02, + "learning_rate": 4.472812043590608e-06, + "loss": 0.0, + "step": 72750 + }, + { + "epoch": 18.02, + "learning_rate": 4.465790360633637e-06, + "loss": 0.0001, + "step": 72760 + }, + { + "epoch": 18.02, + "learning_rate": 4.458768677676665e-06, + "loss": 0.0, + "step": 72770 + }, + { + "epoch": 18.02, + "learning_rate": 4.451746994719694e-06, + "loss": 0.0, + "step": 72780 + }, + { + "epoch": 18.02, + "learning_rate": 4.4447253117627235e-06, + "loss": 0.0, + "step": 72790 + }, + { + "epoch": 18.02, + "learning_rate": 4.437703628805753e-06, + "loss": 0.0002, + "step": 72800 + }, + { + "epoch": 18.02, + "learning_rate": 4.430681945848781e-06, + "loss": 0.0, + "step": 72810 + }, + { + "epoch": 18.02, + "learning_rate": 4.42366026289181e-06, + "loss": 0.0, + "step": 72820 + }, + { + "epoch": 18.02, + "learning_rate": 4.416638579934839e-06, + "loss": 0.0012, + "step": 72830 + }, + { + "epoch": 18.02, + "learning_rate": 4.409616896977868e-06, + "loss": 0.0, + "step": 72840 + }, + { + "epoch": 18.02, + "learning_rate": 4.402595214020896e-06, + "loss": 0.0, + "step": 72850 + }, + { + "epoch": 18.02, + "learning_rate": 4.395573531063925e-06, + "loss": 0.0001, + "step": 72860 + }, + { + "epoch": 18.02, + "learning_rate": 4.3885518481069545e-06, + "loss": 0.1738, + "step": 72870 + }, + { + "epoch": 18.02, + "learning_rate": 4.381530165149984e-06, + "loss": 0.0042, + "step": 72880 + }, + { + "epoch": 18.02, + "learning_rate": 4.374508482193012e-06, + "loss": 0.0, + "step": 72890 + }, + { + "epoch": 18.02, + "learning_rate": 4.367486799236041e-06, + "loss": 0.0, + "step": 72900 + }, + { + "epoch": 18.02, + "learning_rate": 4.36046511627907e-06, + "loss": 0.0, + "step": 72910 + }, + { + "epoch": 18.02, + "learning_rate": 4.3534434333220995e-06, + "loss": 0.0, + "step": 72920 + }, + { + "epoch": 18.02, + "learning_rate": 4.346421750365128e-06, + "loss": 0.0, + "step": 72930 + }, + { + "epoch": 18.02, + "learning_rate": 4.339400067408156e-06, + "loss": 0.0, + "step": 72940 + }, + { + "epoch": 18.02, + "learning_rate": 4.332378384451185e-06, + "loss": 0.0, + "step": 72950 + }, + { + "epoch": 18.02, + "learning_rate": 4.325356701494215e-06, + "loss": 0.0, + "step": 72960 + }, + { + "epoch": 18.02, + "learning_rate": 4.318335018537243e-06, + "loss": 0.0002, + "step": 72970 + }, + { + "epoch": 18.02, + "learning_rate": 4.311313335580272e-06, + "loss": 0.0, + "step": 72980 + }, + { + "epoch": 18.02, + "learning_rate": 4.304291652623301e-06, + "loss": 0.0, + "step": 72990 + }, + { + "epoch": 18.02, + "learning_rate": 4.2972699696663305e-06, + "loss": 0.0, + "step": 73000 + }, + { + "epoch": 18.02, + "learning_rate": 4.290248286709359e-06, + "loss": 0.0, + "step": 73010 + }, + { + "epoch": 18.02, + "learning_rate": 4.283226603752388e-06, + "loss": 0.0, + "step": 73020 + }, + { + "epoch": 18.02, + "learning_rate": 4.276204920795416e-06, + "loss": 0.0, + "step": 73030 + }, + { + "epoch": 18.02, + "learning_rate": 4.269183237838445e-06, + "loss": 0.0029, + "step": 73040 + }, + { + "epoch": 18.02, + "learning_rate": 4.262161554881474e-06, + "loss": 0.0003, + "step": 73050 + }, + { + "epoch": 18.02, + "learning_rate": 4.255139871924503e-06, + "loss": 0.0, + "step": 73060 + }, + { + "epoch": 18.02, + "learning_rate": 4.248118188967532e-06, + "loss": 0.0, + "step": 73070 + }, + { + "epoch": 18.02, + "learning_rate": 4.2410965060105606e-06, + "loss": 0.0082, + "step": 73080 + }, + { + "epoch": 18.02, + "learning_rate": 4.23407482305359e-06, + "loss": 0.0, + "step": 73090 + }, + { + "epoch": 18.02, + "learning_rate": 4.227053140096619e-06, + "loss": 0.0, + "step": 73100 + }, + { + "epoch": 18.02, + "learning_rate": 4.220031457139647e-06, + "loss": 0.0, + "step": 73110 + }, + { + "epoch": 18.02, + "learning_rate": 4.213009774182676e-06, + "loss": 0.0, + "step": 73120 + }, + { + "epoch": 18.02, + "learning_rate": 4.205988091225705e-06, + "loss": 0.0, + "step": 73130 + }, + { + "epoch": 18.02, + "learning_rate": 4.198966408268734e-06, + "loss": 0.0005, + "step": 73140 + }, + { + "epoch": 18.02, + "learning_rate": 4.191944725311763e-06, + "loss": 0.0, + "step": 73150 + }, + { + "epoch": 18.02, + "learning_rate": 4.1849230423547915e-06, + "loss": 0.0, + "step": 73160 + }, + { + "epoch": 18.02, + "learning_rate": 4.177901359397821e-06, + "loss": 0.0, + "step": 73170 + }, + { + "epoch": 18.02, + "learning_rate": 4.17087967644085e-06, + "loss": 0.0, + "step": 73180 + }, + { + "epoch": 18.03, + "learning_rate": 4.163857993483879e-06, + "loss": 0.0, + "step": 73190 + }, + { + "epoch": 18.03, + "learning_rate": 4.156836310526907e-06, + "loss": 0.0, + "step": 73200 + }, + { + "epoch": 18.03, + "learning_rate": 4.149814627569936e-06, + "loss": 0.0, + "step": 73210 + }, + { + "epoch": 18.03, + "learning_rate": 4.142792944612965e-06, + "loss": 0.0, + "step": 73220 + }, + { + "epoch": 18.03, + "learning_rate": 4.135771261655994e-06, + "loss": 0.0, + "step": 73230 + }, + { + "epoch": 18.03, + "learning_rate": 4.1287495786990224e-06, + "loss": 0.0, + "step": 73240 + }, + { + "epoch": 18.03, + "learning_rate": 4.121727895742052e-06, + "loss": 0.0, + "step": 73250 + }, + { + "epoch": 18.03, + "learning_rate": 4.114706212785081e-06, + "loss": 0.0008, + "step": 73260 + }, + { + "epoch": 18.03, + "learning_rate": 4.10768452982811e-06, + "loss": 0.0, + "step": 73270 + }, + { + "epoch": 18.03, + "learning_rate": 4.100662846871138e-06, + "loss": 0.0, + "step": 73280 + }, + { + "epoch": 18.03, + "learning_rate": 4.0936411639141675e-06, + "loss": 0.0, + "step": 73290 + }, + { + "epoch": 18.03, + "learning_rate": 4.086619480957196e-06, + "loss": 0.0, + "step": 73300 + }, + { + "epoch": 18.03, + "learning_rate": 4.079597798000225e-06, + "loss": 0.0, + "step": 73310 + }, + { + "epoch": 18.03, + "learning_rate": 4.072576115043253e-06, + "loss": 0.0, + "step": 73320 + }, + { + "epoch": 18.03, + "learning_rate": 4.0655544320862826e-06, + "loss": 0.0, + "step": 73330 + }, + { + "epoch": 18.03, + "learning_rate": 4.058532749129312e-06, + "loss": 0.0, + "step": 73340 + }, + { + "epoch": 18.03, + "learning_rate": 4.051511066172341e-06, + "loss": 0.0, + "step": 73350 + }, + { + "epoch": 18.03, + "learning_rate": 4.044489383215369e-06, + "loss": 0.0, + "step": 73360 + }, + { + "epoch": 18.03, + "learning_rate": 4.0374677002583984e-06, + "loss": 0.0, + "step": 73370 + }, + { + "epoch": 18.03, + "learning_rate": 4.030446017301427e-06, + "loss": 0.0, + "step": 73380 + }, + { + "epoch": 18.03, + "learning_rate": 4.023424334344455e-06, + "loss": 0.0, + "step": 73390 + }, + { + "epoch": 18.03, + "learning_rate": 4.016402651387484e-06, + "loss": 0.0, + "step": 73400 + }, + { + "epoch": 18.03, + "learning_rate": 4.0093809684305135e-06, + "loss": 0.0, + "step": 73410 + }, + { + "epoch": 18.03, + "learning_rate": 4.002359285473543e-06, + "loss": 0.0, + "step": 73420 + }, + { + "epoch": 18.03, + "learning_rate": 3.995337602516571e-06, + "loss": 0.0, + "step": 73430 + }, + { + "epoch": 18.03, + "learning_rate": 3.9883159195596e-06, + "loss": 0.0, + "step": 73440 + }, + { + "epoch": 18.03, + "learning_rate": 3.981294236602629e-06, + "loss": 0.0, + "step": 73450 + }, + { + "epoch": 18.03, + "learning_rate": 3.9742725536456586e-06, + "loss": 0.0011, + "step": 73460 + }, + { + "epoch": 18.03, + "learning_rate": 3.967250870688687e-06, + "loss": 0.0, + "step": 73470 + }, + { + "epoch": 18.03, + "learning_rate": 3.960229187731715e-06, + "loss": 0.0, + "step": 73480 + }, + { + "epoch": 18.03, + "learning_rate": 3.953207504774744e-06, + "loss": 0.0, + "step": 73490 + }, + { + "epoch": 18.03, + "learning_rate": 3.946185821817774e-06, + "loss": 0.0, + "step": 73500 + }, + { + "epoch": 18.03, + "learning_rate": 3.939164138860802e-06, + "loss": 0.0, + "step": 73510 + }, + { + "epoch": 18.03, + "learning_rate": 3.932142455903831e-06, + "loss": 0.0117, + "step": 73520 + }, + { + "epoch": 18.03, + "learning_rate": 3.92512077294686e-06, + "loss": 0.0, + "step": 73530 + }, + { + "epoch": 18.03, + "learning_rate": 3.9180990899898895e-06, + "loss": 0.0, + "step": 73540 + }, + { + "epoch": 18.03, + "learning_rate": 3.911077407032918e-06, + "loss": 0.0, + "step": 73550 + }, + { + "epoch": 18.03, + "learning_rate": 3.904055724075947e-06, + "loss": 0.0, + "step": 73560 + }, + { + "epoch": 18.03, + "learning_rate": 3.897034041118975e-06, + "loss": 0.0, + "step": 73570 + }, + { + "epoch": 18.03, + "learning_rate": 3.8900123581620045e-06, + "loss": 0.0, + "step": 73580 + }, + { + "epoch": 18.03, + "learning_rate": 3.882990675205033e-06, + "loss": 0.0, + "step": 73590 + }, + { + "epoch": 18.03, + "learning_rate": 3.875968992248062e-06, + "loss": 0.0, + "step": 73600 + }, + { + "epoch": 18.03, + "learning_rate": 3.868947309291091e-06, + "loss": 0.0, + "step": 73610 + }, + { + "epoch": 18.03, + "learning_rate": 3.8619256263341204e-06, + "loss": 0.0, + "step": 73620 + }, + { + "epoch": 18.03, + "learning_rate": 3.854903943377149e-06, + "loss": 0.0, + "step": 73630 + }, + { + "epoch": 18.03, + "learning_rate": 3.847882260420178e-06, + "loss": 0.0, + "step": 73640 + }, + { + "epoch": 18.03, + "learning_rate": 3.840860577463206e-06, + "loss": 0.0, + "step": 73650 + }, + { + "epoch": 18.03, + "learning_rate": 3.8338388945062355e-06, + "loss": 0.0, + "step": 73660 + }, + { + "epoch": 18.03, + "learning_rate": 3.826817211549264e-06, + "loss": 0.0, + "step": 73670 + }, + { + "epoch": 18.03, + "learning_rate": 3.819795528592293e-06, + "loss": 0.0, + "step": 73680 + }, + { + "epoch": 18.03, + "learning_rate": 3.812773845635322e-06, + "loss": 0.0, + "step": 73690 + }, + { + "epoch": 18.03, + "learning_rate": 3.8057521626783514e-06, + "loss": 0.0, + "step": 73700 + }, + { + "epoch": 18.03, + "learning_rate": 3.7987304797213797e-06, + "loss": 0.0, + "step": 73710 + }, + { + "epoch": 18.03, + "learning_rate": 3.791708796764409e-06, + "loss": 0.0, + "step": 73720 + }, + { + "epoch": 18.03, + "learning_rate": 3.7846871138074377e-06, + "loss": 0.0, + "step": 73730 + }, + { + "epoch": 18.03, + "learning_rate": 3.777665430850466e-06, + "loss": 0.0, + "step": 73740 + }, + { + "epoch": 18.03, + "learning_rate": 3.770643747893495e-06, + "loss": 0.3526, + "step": 73750 + }, + { + "epoch": 18.03, + "learning_rate": 3.7636220649365244e-06, + "loss": 0.0, + "step": 73760 + }, + { + "epoch": 18.03, + "learning_rate": 3.756600381979553e-06, + "loss": 0.1663, + "step": 73770 + }, + { + "epoch": 18.03, + "learning_rate": 3.7495786990225815e-06, + "loss": 0.0, + "step": 73780 + }, + { + "epoch": 18.03, + "learning_rate": 3.7425570160656106e-06, + "loss": 0.0291, + "step": 73790 + }, + { + "epoch": 18.03, + "learning_rate": 3.73553533310864e-06, + "loss": 0.0, + "step": 73800 + }, + { + "epoch": 18.03, + "learning_rate": 3.7285136501516686e-06, + "loss": 0.0014, + "step": 73810 + }, + { + "epoch": 18.03, + "learning_rate": 3.721491967194697e-06, + "loss": 0.0, + "step": 73820 + }, + { + "epoch": 18.03, + "learning_rate": 3.714470284237726e-06, + "loss": 0.0, + "step": 73830 + }, + { + "epoch": 18.03, + "learning_rate": 3.7074486012807553e-06, + "loss": 0.0, + "step": 73840 + }, + { + "epoch": 18.03, + "learning_rate": 3.7004269183237845e-06, + "loss": 0.0, + "step": 73850 + }, + { + "epoch": 18.03, + "learning_rate": 3.693405235366813e-06, + "loss": 0.0, + "step": 73860 + }, + { + "epoch": 18.03, + "learning_rate": 3.6863835524098416e-06, + "loss": 0.0, + "step": 73870 + }, + { + "epoch": 18.03, + "learning_rate": 3.6793618694528708e-06, + "loss": 0.0, + "step": 73880 + }, + { + "epoch": 18.03, + "learning_rate": 3.6723401864959e-06, + "loss": 0.0, + "step": 73890 + }, + { + "epoch": 18.03, + "learning_rate": 3.6653185035389283e-06, + "loss": 0.0, + "step": 73900 + }, + { + "epoch": 18.03, + "learning_rate": 3.658296820581957e-06, + "loss": 0.0, + "step": 73910 + }, + { + "epoch": 18.03, + "learning_rate": 3.6512751376249862e-06, + "loss": 0.0001, + "step": 73920 + }, + { + "epoch": 18.03, + "learning_rate": 3.6442534546680154e-06, + "loss": 0.0, + "step": 73930 + }, + { + "epoch": 18.03, + "learning_rate": 3.6372317717110438e-06, + "loss": 0.0, + "step": 73940 + }, + { + "epoch": 18.03, + "learning_rate": 3.6302100887540725e-06, + "loss": 0.0, + "step": 73950 + }, + { + "epoch": 18.03, + "learning_rate": 3.6231884057971017e-06, + "loss": 0.0, + "step": 73960 + }, + { + "epoch": 18.03, + "learning_rate": 3.616166722840131e-06, + "loss": 0.0006, + "step": 73970 + }, + { + "epoch": 18.04, + "learning_rate": 3.6091450398831592e-06, + "loss": 0.0, + "step": 73980 + }, + { + "epoch": 18.04, + "learning_rate": 3.6021233569261884e-06, + "loss": 0.0, + "step": 73990 + }, + { + "epoch": 18.04, + "learning_rate": 3.595101673969217e-06, + "loss": 0.0, + "step": 74000 + }, + { + "epoch": 18.04, + "learning_rate": 3.5880799910122463e-06, + "loss": 0.0, + "step": 74010 + }, + { + "epoch": 18.04, + "learning_rate": 3.5810583080552747e-06, + "loss": 0.0, + "step": 74020 + }, + { + "epoch": 18.04, + "learning_rate": 3.574036625098304e-06, + "loss": 0.0, + "step": 74030 + }, + { + "epoch": 18.04, + "learning_rate": 3.5670149421413326e-06, + "loss": 0.0, + "step": 74040 + }, + { + "epoch": 18.04, + "learning_rate": 3.559993259184361e-06, + "loss": 0.1418, + "step": 74050 + }, + { + "epoch": 18.04, + "learning_rate": 3.55297157622739e-06, + "loss": 0.0, + "step": 74060 + }, + { + "epoch": 18.04, + "learning_rate": 3.5459498932704193e-06, + "loss": 0.0, + "step": 74070 + }, + { + "epoch": 18.04, + "learning_rate": 3.538928210313448e-06, + "loss": 0.0, + "step": 74080 + }, + { + "epoch": 18.04, + "learning_rate": 3.5319065273564764e-06, + "loss": 0.0, + "step": 74090 + }, + { + "epoch": 18.04, + "learning_rate": 3.5248848443995056e-06, + "loss": 0.0, + "step": 74100 + }, + { + "epoch": 18.04, + "learning_rate": 3.517863161442535e-06, + "loss": 0.0, + "step": 74110 + }, + { + "epoch": 18.04, + "learning_rate": 3.510841478485564e-06, + "loss": 0.0, + "step": 74120 + }, + { + "epoch": 18.04, + "learning_rate": 3.5038197955285923e-06, + "loss": 0.0004, + "step": 74130 + }, + { + "epoch": 18.04, + "learning_rate": 3.496798112571621e-06, + "loss": 0.0, + "step": 74140 + }, + { + "epoch": 18.04, + "learning_rate": 3.4897764296146503e-06, + "loss": 0.0001, + "step": 74150 + }, + { + "epoch": 18.04, + "learning_rate": 3.4827547466576795e-06, + "loss": 0.0001, + "step": 74160 + }, + { + "epoch": 18.04, + "learning_rate": 3.475733063700708e-06, + "loss": 0.0, + "step": 74170 + }, + { + "epoch": 18.04, + "learning_rate": 3.4687113807437366e-06, + "loss": 0.0, + "step": 74180 + }, + { + "epoch": 18.04, + "learning_rate": 3.4616896977867657e-06, + "loss": 0.0, + "step": 74190 + }, + { + "epoch": 18.04, + "learning_rate": 3.454668014829795e-06, + "loss": 0.0, + "step": 74200 + }, + { + "epoch": 18.04, + "learning_rate": 3.4476463318728233e-06, + "loss": 0.076, + "step": 74210 + }, + { + "epoch": 18.04, + "learning_rate": 3.440624648915852e-06, + "loss": 0.2755, + "step": 74220 + }, + { + "epoch": 18.04, + "learning_rate": 3.433602965958881e-06, + "loss": 0.0, + "step": 74230 + }, + { + "epoch": 18.04, + "learning_rate": 3.4265812830019104e-06, + "loss": 0.0006, + "step": 74240 + }, + { + "epoch": 18.04, + "learning_rate": 3.4195596000449387e-06, + "loss": 0.0, + "step": 74250 + }, + { + "epoch": 18.04, + "learning_rate": 3.412537917087968e-06, + "loss": 0.0, + "step": 74260 + }, + { + "epoch": 18.04, + "learning_rate": 3.4055162341309967e-06, + "loss": 0.0, + "step": 74270 + }, + { + "epoch": 18.04, + "learning_rate": 3.398494551174026e-06, + "loss": 0.0, + "step": 74280 + }, + { + "epoch": 18.04, + "learning_rate": 3.391472868217054e-06, + "loss": 0.0, + "step": 74290 + }, + { + "epoch": 18.04, + "learning_rate": 3.3844511852600834e-06, + "loss": 0.0, + "step": 74300 + }, + { + "epoch": 18.04, + "learning_rate": 3.377429502303112e-06, + "loss": 0.0, + "step": 74310 + }, + { + "epoch": 18.04, + "learning_rate": 3.3704078193461413e-06, + "loss": 0.0, + "step": 74320 + }, + { + "epoch": 18.04, + "learning_rate": 3.3633861363891697e-06, + "loss": 0.0104, + "step": 74330 + }, + { + "epoch": 18.04, + "learning_rate": 3.356364453432199e-06, + "loss": 0.0, + "step": 74340 + }, + { + "epoch": 18.04, + "learning_rate": 3.3493427704752276e-06, + "loss": 0.0, + "step": 74350 + }, + { + "epoch": 18.04, + "learning_rate": 3.342321087518257e-06, + "loss": 0.0, + "step": 74360 + }, + { + "epoch": 18.04, + "learning_rate": 3.335299404561285e-06, + "loss": 0.0, + "step": 74370 + }, + { + "epoch": 18.04, + "learning_rate": 3.3282777216043143e-06, + "loss": 0.0, + "step": 74380 + }, + { + "epoch": 18.04, + "learning_rate": 3.3212560386473435e-06, + "loss": 0.0, + "step": 74390 + }, + { + "epoch": 18.04, + "learning_rate": 3.314234355690372e-06, + "loss": 0.1066, + "step": 74400 + }, + { + "epoch": 18.04, + "learning_rate": 3.3072126727334006e-06, + "loss": 0.0, + "step": 74410 + }, + { + "epoch": 18.04, + "learning_rate": 3.3001909897764298e-06, + "loss": 0.0, + "step": 74420 + }, + { + "epoch": 18.04, + "learning_rate": 3.293169306819459e-06, + "loss": 0.0, + "step": 74430 + }, + { + "epoch": 18.04, + "learning_rate": 3.2861476238624873e-06, + "loss": 0.0, + "step": 74440 + }, + { + "epoch": 18.04, + "learning_rate": 3.279125940905516e-06, + "loss": 0.0, + "step": 74450 + }, + { + "epoch": 18.04, + "learning_rate": 3.2721042579485453e-06, + "loss": 0.0, + "step": 74460 + }, + { + "epoch": 18.04, + "learning_rate": 3.2650825749915744e-06, + "loss": 0.0, + "step": 74470 + }, + { + "epoch": 18.04, + "learning_rate": 3.2580608920346028e-06, + "loss": 0.0, + "step": 74480 + }, + { + "epoch": 18.04, + "learning_rate": 3.251039209077632e-06, + "loss": 0.0, + "step": 74490 + }, + { + "epoch": 18.04, + "learning_rate": 3.2440175261206607e-06, + "loss": 0.0, + "step": 74500 + }, + { + "epoch": 18.04, + "learning_rate": 3.23699584316369e-06, + "loss": 0.0, + "step": 74510 + }, + { + "epoch": 18.04, + "learning_rate": 3.2299741602067182e-06, + "loss": 0.0, + "step": 74520 + }, + { + "epoch": 18.04, + "learning_rate": 3.2229524772497474e-06, + "loss": 0.0, + "step": 74530 + }, + { + "epoch": 18.04, + "learning_rate": 3.215930794292776e-06, + "loss": 0.0, + "step": 74540 + }, + { + "epoch": 18.04, + "learning_rate": 3.2089091113358054e-06, + "loss": 0.0357, + "step": 74550 + }, + { + "epoch": 18.04, + "learning_rate": 3.2018874283788337e-06, + "loss": 0.0, + "step": 74560 + }, + { + "epoch": 18.04, + "learning_rate": 3.194865745421863e-06, + "loss": 0.0, + "step": 74570 + }, + { + "epoch": 18.04, + "learning_rate": 3.1878440624648917e-06, + "loss": 0.0001, + "step": 74580 + }, + { + "epoch": 18.04, + "learning_rate": 3.180822379507921e-06, + "loss": 0.0, + "step": 74590 + }, + { + "epoch": 18.04, + "learning_rate": 3.173800696550949e-06, + "loss": 0.0, + "step": 74600 + }, + { + "epoch": 18.04, + "learning_rate": 3.1667790135939784e-06, + "loss": 0.0088, + "step": 74610 + }, + { + "epoch": 18.04, + "learning_rate": 3.1597573306370075e-06, + "loss": 0.041, + "step": 74620 + }, + { + "epoch": 18.04, + "learning_rate": 3.1527356476800363e-06, + "loss": 0.0, + "step": 74630 + }, + { + "epoch": 18.04, + "learning_rate": 3.1457139647230646e-06, + "loss": 0.0, + "step": 74640 + }, + { + "epoch": 18.04, + "learning_rate": 3.138692281766094e-06, + "loss": 0.0, + "step": 74650 + }, + { + "epoch": 18.04, + "learning_rate": 3.131670598809123e-06, + "loss": 0.0, + "step": 74660 + }, + { + "epoch": 18.04, + "learning_rate": 3.1246489158521518e-06, + "loss": 0.0, + "step": 74670 + }, + { + "epoch": 18.04, + "learning_rate": 3.1176272328951805e-06, + "loss": 0.0001, + "step": 74680 + }, + { + "epoch": 18.04, + "learning_rate": 3.1106055499382093e-06, + "loss": 0.0, + "step": 74690 + }, + { + "epoch": 18.04, + "learning_rate": 3.1035838669812385e-06, + "loss": 0.0, + "step": 74700 + }, + { + "epoch": 18.04, + "learning_rate": 3.0965621840242672e-06, + "loss": 0.0, + "step": 74710 + }, + { + "epoch": 18.04, + "learning_rate": 3.089540501067296e-06, + "loss": 0.0, + "step": 74720 + }, + { + "epoch": 18.04, + "learning_rate": 3.0825188181103248e-06, + "loss": 0.0, + "step": 74730 + }, + { + "epoch": 18.04, + "learning_rate": 3.0754971351533535e-06, + "loss": 0.0, + "step": 74740 + }, + { + "epoch": 18.04, + "learning_rate": 3.0684754521963827e-06, + "loss": 0.0, + "step": 74750 + }, + { + "epoch": 18.04, + "learning_rate": 3.0614537692394115e-06, + "loss": 0.0, + "step": 74760 + }, + { + "epoch": 18.05, + "learning_rate": 3.0544320862824402e-06, + "loss": 0.0, + "step": 74770 + }, + { + "epoch": 18.05, + "learning_rate": 3.047410403325469e-06, + "loss": 0.0, + "step": 74780 + }, + { + "epoch": 18.05, + "learning_rate": 3.040388720368498e-06, + "loss": 0.0, + "step": 74790 + }, + { + "epoch": 18.05, + "learning_rate": 3.033367037411527e-06, + "loss": 0.0, + "step": 74800 + }, + { + "epoch": 18.05, + "learning_rate": 3.0263453544545557e-06, + "loss": 0.008, + "step": 74810 + }, + { + "epoch": 18.05, + "learning_rate": 3.0193236714975845e-06, + "loss": 0.0004, + "step": 74820 + }, + { + "epoch": 18.05, + "learning_rate": 3.0123019885406136e-06, + "loss": 0.0, + "step": 74830 + }, + { + "epoch": 18.05, + "learning_rate": 3.0052803055836424e-06, + "loss": 0.0, + "step": 74840 + }, + { + "epoch": 18.05, + "learning_rate": 2.998258622626671e-06, + "loss": 0.0008, + "step": 74850 + }, + { + "epoch": 18.05, + "learning_rate": 2.9912369396697e-06, + "loss": 0.0, + "step": 74860 + }, + { + "epoch": 18.05, + "learning_rate": 2.984215256712729e-06, + "loss": 0.0, + "step": 74870 + }, + { + "epoch": 18.05, + "learning_rate": 2.977193573755758e-06, + "loss": 0.0004, + "step": 74880 + }, + { + "epoch": 18.05, + "learning_rate": 2.970171890798787e-06, + "loss": 0.0, + "step": 74890 + }, + { + "epoch": 18.05, + "learning_rate": 2.9631502078418154e-06, + "loss": 0.0, + "step": 74900 + }, + { + "epoch": 18.05, + "learning_rate": 2.9561285248848446e-06, + "loss": 0.0377, + "step": 74910 + }, + { + "epoch": 18.05, + "learning_rate": 2.9491068419278733e-06, + "loss": 0.0, + "step": 74920 + }, + { + "epoch": 18.05, + "learning_rate": 2.9420851589709025e-06, + "loss": 0.0, + "step": 74930 + }, + { + "epoch": 18.05, + "learning_rate": 2.9350634760139313e-06, + "loss": 0.0, + "step": 74940 + }, + { + "epoch": 18.05, + "learning_rate": 2.92804179305696e-06, + "loss": 0.0, + "step": 74950 + }, + { + "epoch": 18.05, + "learning_rate": 2.921020110099989e-06, + "loss": 0.0, + "step": 74960 + }, + { + "epoch": 18.05, + "learning_rate": 2.913998427143018e-06, + "loss": 0.0, + "step": 74970 + }, + { + "epoch": 18.05, + "learning_rate": 2.9069767441860468e-06, + "loss": 0.229, + "step": 74980 + }, + { + "epoch": 18.05, + "learning_rate": 2.8999550612290755e-06, + "loss": 0.0, + "step": 74990 + }, + { + "epoch": 18.05, + "learning_rate": 2.8929333782721043e-06, + "loss": 0.0002, + "step": 75000 + }, + { + "epoch": 18.05, + "learning_rate": 2.8859116953151335e-06, + "loss": 0.0, + "step": 75010 + }, + { + "epoch": 18.05, + "learning_rate": 2.8788900123581622e-06, + "loss": 0.0, + "step": 75020 + }, + { + "epoch": 18.05, + "learning_rate": 2.871868329401191e-06, + "loss": 0.0, + "step": 75030 + }, + { + "epoch": 18.05, + "learning_rate": 2.8648466464442197e-06, + "loss": 0.0, + "step": 75040 + }, + { + "epoch": 18.05, + "learning_rate": 2.857824963487249e-06, + "loss": 0.0, + "step": 75050 + }, + { + "epoch": 18.05, + "learning_rate": 2.8508032805302777e-06, + "loss": 0.0, + "step": 75060 + }, + { + "epoch": 18.05, + "learning_rate": 2.8437815975733065e-06, + "loss": 0.0, + "step": 75070 + }, + { + "epoch": 18.05, + "learning_rate": 2.8367599146163352e-06, + "loss": 0.0, + "step": 75080 + }, + { + "epoch": 18.05, + "learning_rate": 2.829738231659364e-06, + "loss": 0.0, + "step": 75090 + }, + { + "epoch": 18.05, + "learning_rate": 2.822716548702393e-06, + "loss": 0.0073, + "step": 75100 + }, + { + "epoch": 18.05, + "learning_rate": 2.815694865745422e-06, + "loss": 0.0001, + "step": 75110 + }, + { + "epoch": 18.05, + "learning_rate": 2.808673182788451e-06, + "loss": 0.0, + "step": 75120 + }, + { + "epoch": 18.05, + "learning_rate": 2.8016514998314794e-06, + "loss": 0.0, + "step": 75130 + }, + { + "epoch": 18.05, + "learning_rate": 2.7946298168745086e-06, + "loss": 0.0, + "step": 75140 + }, + { + "epoch": 18.05, + "learning_rate": 2.7876081339175374e-06, + "loss": 0.0, + "step": 75150 + }, + { + "epoch": 18.05, + "learning_rate": 2.7805864509605666e-06, + "loss": 0.0, + "step": 75160 + }, + { + "epoch": 18.05, + "eval_accuracy": 0.7708682750612824, + "eval_loss": 1.6373403072357178, + "eval_runtime": 1290.1574, + "eval_samples_per_second": 6.008, + "eval_steps_per_second": 3.004, + "step": 75164 + }, + { + "epoch": 19.0, + "learning_rate": 2.773564768003595e-06, + "loss": 0.0001, + "step": 75170 + }, + { + "epoch": 19.0, + "learning_rate": 2.766543085046624e-06, + "loss": 0.0, + "step": 75180 + }, + { + "epoch": 19.0, + "learning_rate": 2.759521402089653e-06, + "loss": 0.0, + "step": 75190 + }, + { + "epoch": 19.0, + "learning_rate": 2.752499719132682e-06, + "loss": 0.0001, + "step": 75200 + }, + { + "epoch": 19.0, + "learning_rate": 2.745478036175711e-06, + "loss": 0.0, + "step": 75210 + }, + { + "epoch": 19.0, + "learning_rate": 2.7384563532187396e-06, + "loss": 0.0, + "step": 75220 + }, + { + "epoch": 19.0, + "learning_rate": 2.7314346702617683e-06, + "loss": 0.0008, + "step": 75230 + }, + { + "epoch": 19.0, + "learning_rate": 2.7244129873047975e-06, + "loss": 0.0, + "step": 75240 + }, + { + "epoch": 19.0, + "learning_rate": 2.7173913043478263e-06, + "loss": 0.0, + "step": 75250 + }, + { + "epoch": 19.0, + "learning_rate": 2.710369621390855e-06, + "loss": 0.0, + "step": 75260 + }, + { + "epoch": 19.0, + "learning_rate": 2.703347938433884e-06, + "loss": 0.0, + "step": 75270 + }, + { + "epoch": 19.0, + "learning_rate": 2.696326255476913e-06, + "loss": 0.0189, + "step": 75280 + }, + { + "epoch": 19.0, + "learning_rate": 2.6893045725199417e-06, + "loss": 0.0, + "step": 75290 + }, + { + "epoch": 19.0, + "learning_rate": 2.6822828895629705e-06, + "loss": 0.0, + "step": 75300 + }, + { + "epoch": 19.0, + "learning_rate": 2.6752612066059993e-06, + "loss": 0.0, + "step": 75310 + }, + { + "epoch": 19.0, + "learning_rate": 2.6682395236490284e-06, + "loss": 0.0, + "step": 75320 + }, + { + "epoch": 19.0, + "learning_rate": 2.661217840692057e-06, + "loss": 0.0, + "step": 75330 + }, + { + "epoch": 19.0, + "learning_rate": 2.6541961577350864e-06, + "loss": 0.022, + "step": 75340 + }, + { + "epoch": 19.0, + "learning_rate": 2.6471744747781147e-06, + "loss": 0.0, + "step": 75350 + }, + { + "epoch": 19.0, + "learning_rate": 2.640152791821144e-06, + "loss": 0.0, + "step": 75360 + }, + { + "epoch": 19.0, + "learning_rate": 2.6331311088641727e-06, + "loss": 0.0, + "step": 75370 + }, + { + "epoch": 19.0, + "learning_rate": 2.626109425907202e-06, + "loss": 0.0, + "step": 75380 + }, + { + "epoch": 19.0, + "learning_rate": 2.6190877429502306e-06, + "loss": 0.0, + "step": 75390 + }, + { + "epoch": 19.0, + "learning_rate": 2.612066059993259e-06, + "loss": 0.0, + "step": 75400 + }, + { + "epoch": 19.0, + "learning_rate": 2.605044377036288e-06, + "loss": 0.0, + "step": 75410 + }, + { + "epoch": 19.0, + "learning_rate": 2.598022694079317e-06, + "loss": 0.0, + "step": 75420 + }, + { + "epoch": 19.0, + "learning_rate": 2.591001011122346e-06, + "loss": 0.0, + "step": 75430 + }, + { + "epoch": 19.0, + "learning_rate": 2.583979328165375e-06, + "loss": 0.0, + "step": 75440 + }, + { + "epoch": 19.0, + "learning_rate": 2.5769576452084036e-06, + "loss": 0.0, + "step": 75450 + }, + { + "epoch": 19.0, + "learning_rate": 2.5699359622514324e-06, + "loss": 0.0, + "step": 75460 + }, + { + "epoch": 19.0, + "learning_rate": 2.5629142792944616e-06, + "loss": 0.0, + "step": 75470 + }, + { + "epoch": 19.0, + "learning_rate": 2.5558925963374903e-06, + "loss": 0.0, + "step": 75480 + }, + { + "epoch": 19.0, + "learning_rate": 2.548870913380519e-06, + "loss": 0.0, + "step": 75490 + }, + { + "epoch": 19.0, + "learning_rate": 2.541849230423548e-06, + "loss": 0.0, + "step": 75500 + }, + { + "epoch": 19.0, + "learning_rate": 2.534827547466577e-06, + "loss": 0.0, + "step": 75510 + }, + { + "epoch": 19.0, + "learning_rate": 2.5278058645096058e-06, + "loss": 0.0675, + "step": 75520 + }, + { + "epoch": 19.0, + "learning_rate": 2.5207841815526345e-06, + "loss": 0.0, + "step": 75530 + }, + { + "epoch": 19.0, + "learning_rate": 2.5137624985956633e-06, + "loss": 0.0, + "step": 75540 + }, + { + "epoch": 19.0, + "learning_rate": 2.5067408156386925e-06, + "loss": 0.0, + "step": 75550 + }, + { + "epoch": 19.01, + "learning_rate": 2.4997191326817213e-06, + "loss": 0.0019, + "step": 75560 + }, + { + "epoch": 19.01, + "learning_rate": 2.4926974497247504e-06, + "loss": 0.0, + "step": 75570 + }, + { + "epoch": 19.01, + "learning_rate": 2.4856757667677788e-06, + "loss": 0.0, + "step": 75580 + }, + { + "epoch": 19.01, + "learning_rate": 2.478654083810808e-06, + "loss": 0.0, + "step": 75590 + }, + { + "epoch": 19.01, + "learning_rate": 2.4716324008538367e-06, + "loss": 0.2787, + "step": 75600 + }, + { + "epoch": 19.01, + "learning_rate": 2.464610717896866e-06, + "loss": 0.0, + "step": 75610 + }, + { + "epoch": 19.01, + "learning_rate": 2.4575890349398942e-06, + "loss": 0.0, + "step": 75620 + }, + { + "epoch": 19.01, + "learning_rate": 2.4505673519829234e-06, + "loss": 0.0001, + "step": 75630 + }, + { + "epoch": 19.01, + "learning_rate": 2.443545669025952e-06, + "loss": 0.0, + "step": 75640 + }, + { + "epoch": 19.01, + "learning_rate": 2.4365239860689814e-06, + "loss": 0.0, + "step": 75650 + }, + { + "epoch": 19.01, + "learning_rate": 2.42950230311201e-06, + "loss": 0.0, + "step": 75660 + }, + { + "epoch": 19.01, + "learning_rate": 2.422480620155039e-06, + "loss": 0.0769, + "step": 75670 + }, + { + "epoch": 19.01, + "learning_rate": 2.4154589371980677e-06, + "loss": 0.0, + "step": 75680 + }, + { + "epoch": 19.01, + "learning_rate": 2.408437254241097e-06, + "loss": 0.0, + "step": 75690 + }, + { + "epoch": 19.01, + "learning_rate": 2.4014155712841256e-06, + "loss": 0.0, + "step": 75700 + }, + { + "epoch": 19.01, + "learning_rate": 2.3943938883271544e-06, + "loss": 0.0, + "step": 75710 + }, + { + "epoch": 19.01, + "learning_rate": 2.387372205370183e-06, + "loss": 0.0, + "step": 75720 + }, + { + "epoch": 19.01, + "learning_rate": 2.3803505224132123e-06, + "loss": 0.0002, + "step": 75730 + }, + { + "epoch": 19.01, + "learning_rate": 2.373328839456241e-06, + "loss": 0.0, + "step": 75740 + }, + { + "epoch": 19.01, + "learning_rate": 2.36630715649927e-06, + "loss": 0.0, + "step": 75750 + }, + { + "epoch": 19.01, + "learning_rate": 2.3592854735422986e-06, + "loss": 0.2863, + "step": 75760 + }, + { + "epoch": 19.01, + "learning_rate": 2.3522637905853274e-06, + "loss": 0.0, + "step": 75770 + }, + { + "epoch": 19.01, + "learning_rate": 2.3452421076283565e-06, + "loss": 0.0, + "step": 75780 + }, + { + "epoch": 19.01, + "learning_rate": 2.3382204246713853e-06, + "loss": 0.0, + "step": 75790 + }, + { + "epoch": 19.01, + "learning_rate": 2.331198741714414e-06, + "loss": 0.0, + "step": 75800 + }, + { + "epoch": 19.01, + "learning_rate": 2.324177058757443e-06, + "loss": 0.0, + "step": 75810 + }, + { + "epoch": 19.01, + "learning_rate": 2.317155375800472e-06, + "loss": 0.0, + "step": 75820 + }, + { + "epoch": 19.01, + "learning_rate": 2.3101336928435008e-06, + "loss": 0.0, + "step": 75830 + }, + { + "epoch": 19.01, + "learning_rate": 2.30311200988653e-06, + "loss": 0.0, + "step": 75840 + }, + { + "epoch": 19.01, + "learning_rate": 2.2960903269295583e-06, + "loss": 0.0, + "step": 75850 + }, + { + "epoch": 19.01, + "learning_rate": 2.2890686439725875e-06, + "loss": 0.0, + "step": 75860 + }, + { + "epoch": 19.01, + "learning_rate": 2.2820469610156162e-06, + "loss": 0.0, + "step": 75870 + }, + { + "epoch": 19.01, + "learning_rate": 2.2750252780586454e-06, + "loss": 0.0, + "step": 75880 + }, + { + "epoch": 19.01, + "learning_rate": 2.268003595101674e-06, + "loss": 0.0, + "step": 75890 + }, + { + "epoch": 19.01, + "learning_rate": 2.260981912144703e-06, + "loss": 0.0, + "step": 75900 + }, + { + "epoch": 19.01, + "learning_rate": 2.2539602291877317e-06, + "loss": 0.1, + "step": 75910 + }, + { + "epoch": 19.01, + "learning_rate": 2.246938546230761e-06, + "loss": 0.0, + "step": 75920 + }, + { + "epoch": 19.01, + "learning_rate": 2.2399168632737896e-06, + "loss": 0.0, + "step": 75930 + }, + { + "epoch": 19.01, + "learning_rate": 2.2328951803168184e-06, + "loss": 0.0, + "step": 75940 + }, + { + "epoch": 19.01, + "learning_rate": 2.225873497359847e-06, + "loss": 0.0, + "step": 75950 + }, + { + "epoch": 19.01, + "learning_rate": 2.2188518144028764e-06, + "loss": 0.0, + "step": 75960 + }, + { + "epoch": 19.01, + "learning_rate": 2.211830131445905e-06, + "loss": 0.0001, + "step": 75970 + }, + { + "epoch": 19.01, + "learning_rate": 2.204808448488934e-06, + "loss": 0.0, + "step": 75980 + }, + { + "epoch": 19.01, + "learning_rate": 2.1977867655319626e-06, + "loss": 0.0, + "step": 75990 + }, + { + "epoch": 19.01, + "learning_rate": 2.190765082574992e-06, + "loss": 0.0, + "step": 76000 + }, + { + "epoch": 19.01, + "learning_rate": 2.1837433996180206e-06, + "loss": 0.0, + "step": 76010 + }, + { + "epoch": 19.01, + "learning_rate": 2.1767217166610498e-06, + "loss": 0.0, + "step": 76020 + }, + { + "epoch": 19.01, + "learning_rate": 2.169700033704078e-06, + "loss": 0.0, + "step": 76030 + }, + { + "epoch": 19.01, + "learning_rate": 2.1626783507471073e-06, + "loss": 0.0002, + "step": 76040 + }, + { + "epoch": 19.01, + "learning_rate": 2.155656667790136e-06, + "loss": 0.0, + "step": 76050 + }, + { + "epoch": 19.01, + "learning_rate": 2.1486349848331652e-06, + "loss": 0.0, + "step": 76060 + }, + { + "epoch": 19.01, + "learning_rate": 2.141613301876194e-06, + "loss": 0.0, + "step": 76070 + }, + { + "epoch": 19.01, + "learning_rate": 2.1345916189192223e-06, + "loss": 0.0, + "step": 76080 + }, + { + "epoch": 19.01, + "learning_rate": 2.1275699359622515e-06, + "loss": 0.0, + "step": 76090 + }, + { + "epoch": 19.01, + "learning_rate": 2.1205482530052803e-06, + "loss": 0.0, + "step": 76100 + }, + { + "epoch": 19.01, + "learning_rate": 2.1135265700483095e-06, + "loss": 0.0, + "step": 76110 + }, + { + "epoch": 19.01, + "learning_rate": 2.106504887091338e-06, + "loss": 0.0, + "step": 76120 + }, + { + "epoch": 19.01, + "learning_rate": 2.099483204134367e-06, + "loss": 0.0, + "step": 76130 + }, + { + "epoch": 19.01, + "learning_rate": 2.0924615211773957e-06, + "loss": 0.0, + "step": 76140 + }, + { + "epoch": 19.01, + "learning_rate": 2.085439838220425e-06, + "loss": 0.0, + "step": 76150 + }, + { + "epoch": 19.01, + "learning_rate": 2.0784181552634537e-06, + "loss": 0.0003, + "step": 76160 + }, + { + "epoch": 19.01, + "learning_rate": 2.0713964723064825e-06, + "loss": 0.0, + "step": 76170 + }, + { + "epoch": 19.01, + "learning_rate": 2.0643747893495112e-06, + "loss": 0.0, + "step": 76180 + }, + { + "epoch": 19.01, + "learning_rate": 2.0573531063925404e-06, + "loss": 0.0, + "step": 76190 + }, + { + "epoch": 19.01, + "learning_rate": 2.050331423435569e-06, + "loss": 0.0, + "step": 76200 + }, + { + "epoch": 19.01, + "learning_rate": 2.043309740478598e-06, + "loss": 0.0, + "step": 76210 + }, + { + "epoch": 19.01, + "learning_rate": 2.0362880575216267e-06, + "loss": 0.0, + "step": 76220 + }, + { + "epoch": 19.01, + "learning_rate": 2.029266374564656e-06, + "loss": 0.0, + "step": 76230 + }, + { + "epoch": 19.01, + "learning_rate": 2.0222446916076846e-06, + "loss": 0.0, + "step": 76240 + }, + { + "epoch": 19.01, + "learning_rate": 2.0152230086507134e-06, + "loss": 0.0, + "step": 76250 + }, + { + "epoch": 19.01, + "learning_rate": 2.008201325693742e-06, + "loss": 0.0017, + "step": 76260 + }, + { + "epoch": 19.01, + "learning_rate": 2.0011796427367713e-06, + "loss": 0.0, + "step": 76270 + }, + { + "epoch": 19.01, + "learning_rate": 1.9941579597798e-06, + "loss": 0.0, + "step": 76280 + }, + { + "epoch": 19.01, + "learning_rate": 1.9871362768228293e-06, + "loss": 0.0, + "step": 76290 + }, + { + "epoch": 19.01, + "learning_rate": 1.9801145938658576e-06, + "loss": 0.0, + "step": 76300 + }, + { + "epoch": 19.01, + "learning_rate": 1.973092910908887e-06, + "loss": 0.0, + "step": 76310 + }, + { + "epoch": 19.01, + "learning_rate": 1.9660712279519156e-06, + "loss": 0.0, + "step": 76320 + }, + { + "epoch": 19.01, + "learning_rate": 1.9590495449949447e-06, + "loss": 0.0, + "step": 76330 + }, + { + "epoch": 19.01, + "learning_rate": 1.9520278620379735e-06, + "loss": 0.0, + "step": 76340 + }, + { + "epoch": 19.01, + "learning_rate": 1.9450061790810023e-06, + "loss": 0.0, + "step": 76350 + }, + { + "epoch": 19.02, + "learning_rate": 1.937984496124031e-06, + "loss": 0.0, + "step": 76360 + }, + { + "epoch": 19.02, + "learning_rate": 1.9309628131670602e-06, + "loss": 0.0, + "step": 76370 + }, + { + "epoch": 19.02, + "learning_rate": 1.923941130210089e-06, + "loss": 0.0, + "step": 76380 + }, + { + "epoch": 19.02, + "learning_rate": 1.9169194472531177e-06, + "loss": 0.0001, + "step": 76390 + }, + { + "epoch": 19.02, + "learning_rate": 1.9098977642961465e-06, + "loss": 0.0, + "step": 76400 + }, + { + "epoch": 19.02, + "learning_rate": 1.9028760813391757e-06, + "loss": 0.0, + "step": 76410 + }, + { + "epoch": 19.02, + "learning_rate": 1.8958543983822044e-06, + "loss": 0.0, + "step": 76420 + }, + { + "epoch": 19.02, + "learning_rate": 1.888832715425233e-06, + "loss": 0.0, + "step": 76430 + }, + { + "epoch": 19.02, + "learning_rate": 1.8818110324682622e-06, + "loss": 0.0, + "step": 76440 + }, + { + "epoch": 19.02, + "learning_rate": 1.8747893495112907e-06, + "loss": 0.0, + "step": 76450 + }, + { + "epoch": 19.02, + "learning_rate": 1.86776766655432e-06, + "loss": 0.0, + "step": 76460 + }, + { + "epoch": 19.02, + "learning_rate": 1.8607459835973485e-06, + "loss": 0.0, + "step": 76470 + }, + { + "epoch": 19.02, + "learning_rate": 1.8537243006403776e-06, + "loss": 0.0, + "step": 76480 + }, + { + "epoch": 19.02, + "learning_rate": 1.8467026176834064e-06, + "loss": 0.0, + "step": 76490 + }, + { + "epoch": 19.02, + "learning_rate": 1.8396809347264354e-06, + "loss": 0.0, + "step": 76500 + }, + { + "epoch": 19.02, + "learning_rate": 1.8326592517694641e-06, + "loss": 0.0, + "step": 76510 + }, + { + "epoch": 19.02, + "learning_rate": 1.8256375688124931e-06, + "loss": 0.0, + "step": 76520 + }, + { + "epoch": 19.02, + "learning_rate": 1.8186158858555219e-06, + "loss": 0.0, + "step": 76530 + }, + { + "epoch": 19.02, + "learning_rate": 1.8115942028985508e-06, + "loss": 0.0, + "step": 76540 + }, + { + "epoch": 19.02, + "learning_rate": 1.8045725199415796e-06, + "loss": 0.0, + "step": 76550 + }, + { + "epoch": 19.02, + "learning_rate": 1.7975508369846086e-06, + "loss": 0.0, + "step": 76560 + }, + { + "epoch": 19.02, + "learning_rate": 1.7905291540276373e-06, + "loss": 0.0, + "step": 76570 + }, + { + "epoch": 19.02, + "learning_rate": 1.7835074710706663e-06, + "loss": 0.0, + "step": 76580 + }, + { + "epoch": 19.02, + "learning_rate": 1.776485788113695e-06, + "loss": 0.0, + "step": 76590 + }, + { + "epoch": 19.02, + "learning_rate": 1.769464105156724e-06, + "loss": 0.0, + "step": 76600 + }, + { + "epoch": 19.02, + "learning_rate": 1.7624424221997528e-06, + "loss": 0.0, + "step": 76610 + }, + { + "epoch": 19.02, + "learning_rate": 1.755420739242782e-06, + "loss": 0.0, + "step": 76620 + }, + { + "epoch": 19.02, + "learning_rate": 1.7483990562858105e-06, + "loss": 0.0, + "step": 76630 + }, + { + "epoch": 19.02, + "learning_rate": 1.7413773733288397e-06, + "loss": 0.0, + "step": 76640 + }, + { + "epoch": 19.02, + "learning_rate": 1.7343556903718683e-06, + "loss": 0.0, + "step": 76650 + }, + { + "epoch": 19.02, + "learning_rate": 1.7273340074148975e-06, + "loss": 0.0, + "step": 76660 + }, + { + "epoch": 19.02, + "learning_rate": 1.720312324457926e-06, + "loss": 0.0, + "step": 76670 + }, + { + "epoch": 19.02, + "learning_rate": 1.7132906415009552e-06, + "loss": 0.0, + "step": 76680 + }, + { + "epoch": 19.02, + "learning_rate": 1.706268958543984e-06, + "loss": 0.0, + "step": 76690 + }, + { + "epoch": 19.02, + "learning_rate": 1.699247275587013e-06, + "loss": 0.0, + "step": 76700 + }, + { + "epoch": 19.02, + "learning_rate": 1.6922255926300417e-06, + "loss": 0.0, + "step": 76710 + }, + { + "epoch": 19.02, + "learning_rate": 1.6852039096730707e-06, + "loss": 0.0001, + "step": 76720 + }, + { + "epoch": 19.02, + "learning_rate": 1.6781822267160994e-06, + "loss": 0.0, + "step": 76730 + }, + { + "epoch": 19.02, + "learning_rate": 1.6711605437591284e-06, + "loss": 0.0, + "step": 76740 + }, + { + "epoch": 19.02, + "learning_rate": 1.6641388608021572e-06, + "loss": 0.0565, + "step": 76750 + }, + { + "epoch": 19.02, + "learning_rate": 1.657117177845186e-06, + "loss": 0.0, + "step": 76760 + }, + { + "epoch": 19.02, + "learning_rate": 1.6500954948882149e-06, + "loss": 0.0018, + "step": 76770 + }, + { + "epoch": 19.02, + "learning_rate": 1.6430738119312437e-06, + "loss": 0.0, + "step": 76780 + }, + { + "epoch": 19.02, + "learning_rate": 1.6360521289742726e-06, + "loss": 0.0, + "step": 76790 + }, + { + "epoch": 19.02, + "learning_rate": 1.6290304460173014e-06, + "loss": 0.0, + "step": 76800 + }, + { + "epoch": 19.02, + "learning_rate": 1.6220087630603304e-06, + "loss": 0.0, + "step": 76810 + }, + { + "epoch": 19.02, + "learning_rate": 1.6149870801033591e-06, + "loss": 0.0, + "step": 76820 + }, + { + "epoch": 19.02, + "learning_rate": 1.607965397146388e-06, + "loss": 0.0, + "step": 76830 + }, + { + "epoch": 19.02, + "learning_rate": 1.6009437141894169e-06, + "loss": 0.0, + "step": 76840 + }, + { + "epoch": 19.02, + "learning_rate": 1.5939220312324458e-06, + "loss": 0.0, + "step": 76850 + }, + { + "epoch": 19.02, + "learning_rate": 1.5869003482754746e-06, + "loss": 0.0, + "step": 76860 + }, + { + "epoch": 19.02, + "learning_rate": 1.5798786653185038e-06, + "loss": 0.0, + "step": 76870 + }, + { + "epoch": 19.02, + "learning_rate": 1.5728569823615323e-06, + "loss": 0.0, + "step": 76880 + }, + { + "epoch": 19.02, + "learning_rate": 1.5658352994045615e-06, + "loss": 0.0, + "step": 76890 + }, + { + "epoch": 19.02, + "learning_rate": 1.5588136164475903e-06, + "loss": 0.0, + "step": 76900 + }, + { + "epoch": 19.02, + "learning_rate": 1.5517919334906192e-06, + "loss": 0.0, + "step": 76910 + }, + { + "epoch": 19.02, + "learning_rate": 1.544770250533648e-06, + "loss": 0.0, + "step": 76920 + }, + { + "epoch": 19.02, + "learning_rate": 1.5377485675766768e-06, + "loss": 0.0, + "step": 76930 + }, + { + "epoch": 19.02, + "learning_rate": 1.5307268846197057e-06, + "loss": 0.1419, + "step": 76940 + }, + { + "epoch": 19.02, + "learning_rate": 1.5237052016627345e-06, + "loss": 0.0, + "step": 76950 + }, + { + "epoch": 19.02, + "learning_rate": 1.5166835187057635e-06, + "loss": 0.0, + "step": 76960 + }, + { + "epoch": 19.02, + "learning_rate": 1.5096618357487922e-06, + "loss": 0.0, + "step": 76970 + }, + { + "epoch": 19.02, + "learning_rate": 1.5026401527918212e-06, + "loss": 0.0001, + "step": 76980 + }, + { + "epoch": 19.02, + "learning_rate": 1.49561846983485e-06, + "loss": 0.0, + "step": 76990 + }, + { + "epoch": 19.02, + "learning_rate": 1.488596786877879e-06, + "loss": 0.0, + "step": 77000 + }, + { + "epoch": 19.02, + "learning_rate": 1.4815751039209077e-06, + "loss": 0.0, + "step": 77010 + }, + { + "epoch": 19.02, + "learning_rate": 1.4745534209639367e-06, + "loss": 0.0, + "step": 77020 + }, + { + "epoch": 19.02, + "learning_rate": 1.4675317380069656e-06, + "loss": 0.0, + "step": 77030 + }, + { + "epoch": 19.02, + "learning_rate": 1.4605100550499944e-06, + "loss": 0.0, + "step": 77040 + }, + { + "epoch": 19.02, + "learning_rate": 1.4534883720930234e-06, + "loss": 0.0, + "step": 77050 + }, + { + "epoch": 19.02, + "learning_rate": 1.4464666891360521e-06, + "loss": 0.0, + "step": 77060 + }, + { + "epoch": 19.02, + "learning_rate": 1.4394450061790811e-06, + "loss": 0.0, + "step": 77070 + }, + { + "epoch": 19.02, + "learning_rate": 1.4324233232221099e-06, + "loss": 0.0, + "step": 77080 + }, + { + "epoch": 19.02, + "learning_rate": 1.4254016402651388e-06, + "loss": 0.0, + "step": 77090 + }, + { + "epoch": 19.02, + "learning_rate": 1.4183799573081676e-06, + "loss": 0.0, + "step": 77100 + }, + { + "epoch": 19.02, + "learning_rate": 1.4113582743511966e-06, + "loss": 0.0, + "step": 77110 + }, + { + "epoch": 19.02, + "learning_rate": 1.4043365913942256e-06, + "loss": 0.0, + "step": 77120 + }, + { + "epoch": 19.02, + "learning_rate": 1.3973149084372543e-06, + "loss": 0.0, + "step": 77130 + }, + { + "epoch": 19.02, + "learning_rate": 1.3902932254802833e-06, + "loss": 0.0021, + "step": 77140 + }, + { + "epoch": 19.03, + "learning_rate": 1.383271542523312e-06, + "loss": 0.0, + "step": 77150 + }, + { + "epoch": 19.03, + "learning_rate": 1.376249859566341e-06, + "loss": 0.0, + "step": 77160 + }, + { + "epoch": 19.03, + "learning_rate": 1.3692281766093698e-06, + "loss": 0.0, + "step": 77170 + }, + { + "epoch": 19.03, + "learning_rate": 1.3622064936523988e-06, + "loss": 0.0, + "step": 77180 + }, + { + "epoch": 19.03, + "learning_rate": 1.3551848106954275e-06, + "loss": 0.0, + "step": 77190 + }, + { + "epoch": 19.03, + "learning_rate": 1.3481631277384565e-06, + "loss": 0.0, + "step": 77200 + }, + { + "epoch": 19.03, + "learning_rate": 1.3411414447814853e-06, + "loss": 0.0, + "step": 77210 + }, + { + "epoch": 19.03, + "learning_rate": 1.3341197618245142e-06, + "loss": 0.0, + "step": 77220 + }, + { + "epoch": 19.03, + "learning_rate": 1.3270980788675432e-06, + "loss": 0.0, + "step": 77230 + }, + { + "epoch": 19.03, + "learning_rate": 1.320076395910572e-06, + "loss": 0.0, + "step": 77240 + }, + { + "epoch": 19.03, + "learning_rate": 1.313054712953601e-06, + "loss": 0.0, + "step": 77250 + }, + { + "epoch": 19.03, + "learning_rate": 1.3060330299966295e-06, + "loss": 0.0, + "step": 77260 + }, + { + "epoch": 19.03, + "learning_rate": 1.2990113470396585e-06, + "loss": 0.0227, + "step": 77270 + }, + { + "epoch": 19.03, + "learning_rate": 1.2919896640826874e-06, + "loss": 0.0, + "step": 77280 + }, + { + "epoch": 19.03, + "learning_rate": 1.2849679811257162e-06, + "loss": 0.0, + "step": 77290 + }, + { + "epoch": 19.03, + "learning_rate": 1.2779462981687452e-06, + "loss": 0.0, + "step": 77300 + }, + { + "epoch": 19.03, + "learning_rate": 1.270924615211774e-06, + "loss": 0.0578, + "step": 77310 + }, + { + "epoch": 19.03, + "learning_rate": 1.2639029322548029e-06, + "loss": 0.0006, + "step": 77320 + }, + { + "epoch": 19.03, + "learning_rate": 1.2568812492978317e-06, + "loss": 0.2058, + "step": 77330 + }, + { + "epoch": 19.03, + "learning_rate": 1.2498595663408606e-06, + "loss": 0.0, + "step": 77340 + }, + { + "epoch": 19.03, + "learning_rate": 1.2428378833838894e-06, + "loss": 0.0, + "step": 77350 + }, + { + "epoch": 19.03, + "learning_rate": 1.2358162004269184e-06, + "loss": 0.0002, + "step": 77360 + }, + { + "epoch": 19.03, + "learning_rate": 1.2287945174699471e-06, + "loss": 0.0, + "step": 77370 + }, + { + "epoch": 19.03, + "learning_rate": 1.221772834512976e-06, + "loss": 0.0, + "step": 77380 + }, + { + "epoch": 19.03, + "learning_rate": 1.214751151556005e-06, + "loss": 0.0, + "step": 77390 + }, + { + "epoch": 19.03, + "learning_rate": 1.2077294685990338e-06, + "loss": 0.0, + "step": 77400 + }, + { + "epoch": 19.03, + "learning_rate": 1.2007077856420628e-06, + "loss": 0.0001, + "step": 77410 + }, + { + "epoch": 19.03, + "learning_rate": 1.1936861026850916e-06, + "loss": 0.0001, + "step": 77420 + }, + { + "epoch": 19.03, + "learning_rate": 1.1866644197281205e-06, + "loss": 0.0, + "step": 77430 + }, + { + "epoch": 19.03, + "learning_rate": 1.1796427367711493e-06, + "loss": 0.0, + "step": 77440 + }, + { + "epoch": 19.03, + "learning_rate": 1.1726210538141783e-06, + "loss": 0.0, + "step": 77450 + }, + { + "epoch": 19.03, + "learning_rate": 1.165599370857207e-06, + "loss": 0.0, + "step": 77460 + }, + { + "epoch": 19.03, + "learning_rate": 1.158577687900236e-06, + "loss": 0.0, + "step": 77470 + }, + { + "epoch": 19.03, + "learning_rate": 1.151556004943265e-06, + "loss": 0.0, + "step": 77480 + }, + { + "epoch": 19.03, + "learning_rate": 1.1445343219862937e-06, + "loss": 0.0, + "step": 77490 + }, + { + "epoch": 19.03, + "learning_rate": 1.1375126390293227e-06, + "loss": 0.0, + "step": 77500 + }, + { + "epoch": 19.03, + "learning_rate": 1.1304909560723515e-06, + "loss": 0.0, + "step": 77510 + }, + { + "epoch": 19.03, + "learning_rate": 1.1234692731153804e-06, + "loss": 0.0, + "step": 77520 + }, + { + "epoch": 19.03, + "learning_rate": 1.1164475901584092e-06, + "loss": 0.3182, + "step": 77530 + }, + { + "epoch": 19.03, + "learning_rate": 1.1094259072014382e-06, + "loss": 0.0019, + "step": 77540 + }, + { + "epoch": 19.03, + "learning_rate": 1.102404224244467e-06, + "loss": 0.0, + "step": 77550 + }, + { + "epoch": 19.03, + "learning_rate": 1.095382541287496e-06, + "loss": 0.0, + "step": 77560 + }, + { + "epoch": 19.03, + "learning_rate": 1.0883608583305249e-06, + "loss": 0.0, + "step": 77570 + }, + { + "epoch": 19.03, + "learning_rate": 1.0813391753735536e-06, + "loss": 0.0, + "step": 77580 + }, + { + "epoch": 19.03, + "learning_rate": 1.0743174924165826e-06, + "loss": 0.0, + "step": 77590 + }, + { + "epoch": 19.03, + "learning_rate": 1.0672958094596112e-06, + "loss": 0.0, + "step": 77600 + }, + { + "epoch": 19.03, + "learning_rate": 1.0602741265026401e-06, + "loss": 0.0, + "step": 77610 + }, + { + "epoch": 19.03, + "learning_rate": 1.053252443545669e-06, + "loss": 0.0, + "step": 77620 + }, + { + "epoch": 19.03, + "learning_rate": 1.0462307605886979e-06, + "loss": 0.0, + "step": 77630 + }, + { + "epoch": 19.03, + "learning_rate": 1.0392090776317268e-06, + "loss": 0.0005, + "step": 77640 + }, + { + "epoch": 19.03, + "learning_rate": 1.0321873946747556e-06, + "loss": 0.0, + "step": 77650 + }, + { + "epoch": 19.03, + "learning_rate": 1.0251657117177846e-06, + "loss": 0.0, + "step": 77660 + }, + { + "epoch": 19.03, + "learning_rate": 1.0181440287608133e-06, + "loss": 0.0, + "step": 77670 + }, + { + "epoch": 19.03, + "learning_rate": 1.0111223458038423e-06, + "loss": 0.0, + "step": 77680 + }, + { + "epoch": 19.03, + "learning_rate": 1.004100662846871e-06, + "loss": 0.0, + "step": 77690 + }, + { + "epoch": 19.03, + "learning_rate": 9.970789798899e-07, + "loss": 0.0, + "step": 77700 + }, + { + "epoch": 19.03, + "learning_rate": 9.900572969329288e-07, + "loss": 0.001, + "step": 77710 + }, + { + "epoch": 19.03, + "learning_rate": 9.830356139759578e-07, + "loss": 0.0001, + "step": 77720 + }, + { + "epoch": 19.03, + "learning_rate": 9.760139310189868e-07, + "loss": 0.0, + "step": 77730 + }, + { + "epoch": 19.03, + "learning_rate": 9.689922480620155e-07, + "loss": 0.0002, + "step": 77740 + }, + { + "epoch": 19.03, + "learning_rate": 9.619705651050445e-07, + "loss": 0.021, + "step": 77750 + }, + { + "epoch": 19.03, + "learning_rate": 9.549488821480732e-07, + "loss": 0.4044, + "step": 77760 + }, + { + "epoch": 19.03, + "learning_rate": 9.479271991911022e-07, + "loss": 0.0, + "step": 77770 + }, + { + "epoch": 19.03, + "learning_rate": 9.409055162341311e-07, + "loss": 0.0, + "step": 77780 + }, + { + "epoch": 19.03, + "learning_rate": 9.3388383327716e-07, + "loss": 0.0, + "step": 77790 + }, + { + "epoch": 19.03, + "learning_rate": 9.268621503201888e-07, + "loss": 0.0, + "step": 77800 + }, + { + "epoch": 19.03, + "learning_rate": 9.198404673632177e-07, + "loss": 0.0, + "step": 77810 + }, + { + "epoch": 19.03, + "learning_rate": 9.128187844062466e-07, + "loss": 0.0, + "step": 77820 + }, + { + "epoch": 19.03, + "learning_rate": 9.057971014492754e-07, + "loss": 0.0, + "step": 77830 + }, + { + "epoch": 19.03, + "learning_rate": 8.987754184923043e-07, + "loss": 0.0, + "step": 77840 + }, + { + "epoch": 19.03, + "learning_rate": 8.917537355353332e-07, + "loss": 0.0, + "step": 77850 + }, + { + "epoch": 19.03, + "learning_rate": 8.84732052578362e-07, + "loss": 0.0001, + "step": 77860 + }, + { + "epoch": 19.03, + "learning_rate": 8.77710369621391e-07, + "loss": 0.0, + "step": 77870 + }, + { + "epoch": 19.03, + "learning_rate": 8.706886866644199e-07, + "loss": 0.0, + "step": 77880 + }, + { + "epoch": 19.03, + "learning_rate": 8.636670037074487e-07, + "loss": 0.0, + "step": 77890 + }, + { + "epoch": 19.03, + "learning_rate": 8.566453207504776e-07, + "loss": 0.0, + "step": 77900 + }, + { + "epoch": 19.03, + "learning_rate": 8.496236377935065e-07, + "loss": 0.0, + "step": 77910 + }, + { + "epoch": 19.03, + "learning_rate": 8.426019548365353e-07, + "loss": 0.0, + "step": 77920 + }, + { + "epoch": 19.03, + "learning_rate": 8.355802718795642e-07, + "loss": 0.0, + "step": 77930 + }, + { + "epoch": 19.04, + "learning_rate": 8.28558588922593e-07, + "loss": 0.0007, + "step": 77940 + }, + { + "epoch": 19.04, + "learning_rate": 8.215369059656218e-07, + "loss": 0.0, + "step": 77950 + }, + { + "epoch": 19.04, + "learning_rate": 8.145152230086507e-07, + "loss": 0.0, + "step": 77960 + }, + { + "epoch": 19.04, + "learning_rate": 8.074935400516796e-07, + "loss": 0.0, + "step": 77970 + }, + { + "epoch": 19.04, + "learning_rate": 8.004718570947084e-07, + "loss": 0.2458, + "step": 77980 + }, + { + "epoch": 19.04, + "learning_rate": 7.934501741377373e-07, + "loss": 0.0002, + "step": 77990 + }, + { + "epoch": 19.04, + "learning_rate": 7.864284911807662e-07, + "loss": 0.0, + "step": 78000 + }, + { + "epoch": 19.04, + "learning_rate": 7.794068082237951e-07, + "loss": 0.0, + "step": 78010 + }, + { + "epoch": 19.04, + "learning_rate": 7.72385125266824e-07, + "loss": 0.0, + "step": 78020 + }, + { + "epoch": 19.04, + "learning_rate": 7.653634423098529e-07, + "loss": 0.0, + "step": 78030 + }, + { + "epoch": 19.04, + "learning_rate": 7.583417593528817e-07, + "loss": 0.0, + "step": 78040 + }, + { + "epoch": 19.04, + "learning_rate": 7.513200763959106e-07, + "loss": 0.0, + "step": 78050 + }, + { + "epoch": 19.04, + "learning_rate": 7.442983934389395e-07, + "loss": 0.0, + "step": 78060 + }, + { + "epoch": 19.04, + "learning_rate": 7.372767104819683e-07, + "loss": 0.0, + "step": 78070 + }, + { + "epoch": 19.04, + "learning_rate": 7.302550275249972e-07, + "loss": 0.0, + "step": 78080 + }, + { + "epoch": 19.04, + "learning_rate": 7.232333445680261e-07, + "loss": 0.0, + "step": 78090 + }, + { + "epoch": 19.04, + "learning_rate": 7.162116616110549e-07, + "loss": 0.0, + "step": 78100 + }, + { + "epoch": 19.04, + "learning_rate": 7.091899786540838e-07, + "loss": 0.0, + "step": 78110 + }, + { + "epoch": 19.04, + "learning_rate": 7.021682956971128e-07, + "loss": 0.0, + "step": 78120 + }, + { + "epoch": 19.04, + "learning_rate": 6.951466127401416e-07, + "loss": 0.0, + "step": 78130 + }, + { + "epoch": 19.04, + "learning_rate": 6.881249297831705e-07, + "loss": 0.0, + "step": 78140 + }, + { + "epoch": 19.04, + "learning_rate": 6.811032468261994e-07, + "loss": 0.0007, + "step": 78150 + }, + { + "epoch": 19.04, + "learning_rate": 6.740815638692282e-07, + "loss": 0.0, + "step": 78160 + }, + { + "epoch": 19.04, + "learning_rate": 6.670598809122571e-07, + "loss": 0.0001, + "step": 78170 + }, + { + "epoch": 19.04, + "learning_rate": 6.60038197955286e-07, + "loss": 0.0001, + "step": 78180 + }, + { + "epoch": 19.04, + "learning_rate": 6.530165149983147e-07, + "loss": 0.0, + "step": 78190 + }, + { + "epoch": 19.04, + "learning_rate": 6.459948320413437e-07, + "loss": 0.0, + "step": 78200 + }, + { + "epoch": 19.04, + "learning_rate": 6.389731490843726e-07, + "loss": 0.0, + "step": 78210 + }, + { + "epoch": 19.04, + "learning_rate": 6.319514661274014e-07, + "loss": 0.0, + "step": 78220 + }, + { + "epoch": 19.04, + "learning_rate": 6.249297831704303e-07, + "loss": 0.0002, + "step": 78230 + }, + { + "epoch": 19.04, + "learning_rate": 6.179081002134592e-07, + "loss": 0.0, + "step": 78240 + }, + { + "epoch": 19.04, + "learning_rate": 6.10886417256488e-07, + "loss": 0.0, + "step": 78250 + }, + { + "epoch": 19.04, + "learning_rate": 6.038647342995169e-07, + "loss": 0.0, + "step": 78260 + }, + { + "epoch": 19.04, + "learning_rate": 5.968430513425458e-07, + "loss": 0.0, + "step": 78270 + }, + { + "epoch": 19.04, + "learning_rate": 5.898213683855746e-07, + "loss": 0.0, + "step": 78280 + }, + { + "epoch": 19.04, + "learning_rate": 5.827996854286035e-07, + "loss": 0.0, + "step": 78290 + }, + { + "epoch": 19.04, + "learning_rate": 5.757780024716325e-07, + "loss": 0.0, + "step": 78300 + }, + { + "epoch": 19.04, + "learning_rate": 5.687563195146614e-07, + "loss": 0.0, + "step": 78310 + }, + { + "epoch": 19.04, + "learning_rate": 5.617346365576902e-07, + "loss": 0.0, + "step": 78320 + }, + { + "epoch": 19.04, + "learning_rate": 5.547129536007191e-07, + "loss": 0.0, + "step": 78330 + }, + { + "epoch": 19.04, + "learning_rate": 5.47691270643748e-07, + "loss": 0.0, + "step": 78340 + }, + { + "epoch": 19.04, + "learning_rate": 5.406695876867768e-07, + "loss": 0.6078, + "step": 78350 + }, + { + "epoch": 19.04, + "learning_rate": 5.336479047298056e-07, + "loss": 0.0, + "step": 78360 + }, + { + "epoch": 19.04, + "learning_rate": 5.266262217728345e-07, + "loss": 0.0, + "step": 78370 + }, + { + "epoch": 19.04, + "learning_rate": 5.196045388158634e-07, + "loss": 0.0, + "step": 78380 + }, + { + "epoch": 19.04, + "learning_rate": 5.125828558588923e-07, + "loss": 0.0, + "step": 78390 + }, + { + "epoch": 19.04, + "learning_rate": 5.055611729019212e-07, + "loss": 0.0, + "step": 78400 + }, + { + "epoch": 19.04, + "learning_rate": 4.9853948994495e-07, + "loss": 0.0, + "step": 78410 + }, + { + "epoch": 19.04, + "learning_rate": 4.915178069879789e-07, + "loss": 0.0, + "step": 78420 + }, + { + "epoch": 19.04, + "learning_rate": 4.844961240310078e-07, + "loss": 0.042, + "step": 78430 + }, + { + "epoch": 19.04, + "learning_rate": 4.774744410740366e-07, + "loss": 0.0, + "step": 78440 + }, + { + "epoch": 19.04, + "learning_rate": 4.7045275811706554e-07, + "loss": 0.0, + "step": 78450 + }, + { + "epoch": 19.04, + "learning_rate": 4.634310751600944e-07, + "loss": 0.0, + "step": 78460 + }, + { + "epoch": 19.04, + "learning_rate": 4.564093922031233e-07, + "loss": 0.0, + "step": 78470 + }, + { + "epoch": 19.04, + "learning_rate": 4.4938770924615215e-07, + "loss": 0.0, + "step": 78480 + }, + { + "epoch": 19.04, + "learning_rate": 4.42366026289181e-07, + "loss": 0.0, + "step": 78490 + }, + { + "epoch": 19.04, + "learning_rate": 4.3534434333220993e-07, + "loss": 0.0, + "step": 78500 + }, + { + "epoch": 19.04, + "learning_rate": 4.283226603752388e-07, + "loss": 0.0, + "step": 78510 + }, + { + "epoch": 19.04, + "learning_rate": 4.2130097741826767e-07, + "loss": 0.0, + "step": 78520 + }, + { + "epoch": 19.04, + "learning_rate": 4.142792944612965e-07, + "loss": 0.0, + "step": 78530 + }, + { + "epoch": 19.04, + "learning_rate": 4.0725761150432535e-07, + "loss": 0.0, + "step": 78540 + }, + { + "epoch": 19.04, + "learning_rate": 4.002359285473542e-07, + "loss": 0.0, + "step": 78550 + }, + { + "epoch": 19.04, + "learning_rate": 3.932142455903831e-07, + "loss": 0.0, + "step": 78560 + }, + { + "epoch": 19.04, + "learning_rate": 3.86192562633412e-07, + "loss": 0.0, + "step": 78570 + }, + { + "epoch": 19.04, + "learning_rate": 3.7917087967644087e-07, + "loss": 0.0, + "step": 78580 + }, + { + "epoch": 19.04, + "learning_rate": 3.7214919671946973e-07, + "loss": 0.0, + "step": 78590 + }, + { + "epoch": 19.04, + "learning_rate": 3.651275137624986e-07, + "loss": 0.0, + "step": 78600 + }, + { + "epoch": 19.04, + "learning_rate": 3.5810583080552747e-07, + "loss": 0.0, + "step": 78610 + }, + { + "epoch": 19.04, + "learning_rate": 3.510841478485564e-07, + "loss": 0.0, + "step": 78620 + }, + { + "epoch": 19.04, + "learning_rate": 3.4406246489158526e-07, + "loss": 0.0, + "step": 78630 + }, + { + "epoch": 19.04, + "learning_rate": 3.370407819346141e-07, + "loss": 0.0, + "step": 78640 + }, + { + "epoch": 19.04, + "learning_rate": 3.30019098977643e-07, + "loss": 0.0, + "step": 78650 + }, + { + "epoch": 19.04, + "learning_rate": 3.2299741602067186e-07, + "loss": 0.0, + "step": 78660 + }, + { + "epoch": 19.04, + "learning_rate": 3.159757330637007e-07, + "loss": 0.0, + "step": 78670 + }, + { + "epoch": 19.04, + "learning_rate": 3.089540501067296e-07, + "loss": 0.0, + "step": 78680 + }, + { + "epoch": 19.04, + "learning_rate": 3.0193236714975846e-07, + "loss": 0.0, + "step": 78690 + }, + { + "epoch": 19.04, + "learning_rate": 2.949106841927873e-07, + "loss": 0.0, + "step": 78700 + }, + { + "epoch": 19.04, + "learning_rate": 2.8788900123581624e-07, + "loss": 0.0, + "step": 78710 + }, + { + "epoch": 19.04, + "learning_rate": 2.808673182788451e-07, + "loss": 0.0, + "step": 78720 + }, + { + "epoch": 19.05, + "learning_rate": 2.73845635321874e-07, + "loss": 0.0, + "step": 78730 + }, + { + "epoch": 19.05, + "learning_rate": 2.668239523649028e-07, + "loss": 0.0, + "step": 78740 + }, + { + "epoch": 19.05, + "learning_rate": 2.598022694079317e-07, + "loss": 0.046, + "step": 78750 + }, + { + "epoch": 19.05, + "learning_rate": 2.527805864509606e-07, + "loss": 0.0, + "step": 78760 + }, + { + "epoch": 19.05, + "learning_rate": 2.4575890349398945e-07, + "loss": 0.0, + "step": 78770 + }, + { + "epoch": 19.05, + "learning_rate": 2.387372205370183e-07, + "loss": 0.0, + "step": 78780 + }, + { + "epoch": 19.05, + "learning_rate": 2.317155375800472e-07, + "loss": 0.0, + "step": 78790 + }, + { + "epoch": 19.05, + "learning_rate": 2.2469385462307607e-07, + "loss": 0.0, + "step": 78800 + }, + { + "epoch": 19.05, + "learning_rate": 2.1767217166610497e-07, + "loss": 0.0, + "step": 78810 + }, + { + "epoch": 19.05, + "learning_rate": 2.1065048870913383e-07, + "loss": 0.0, + "step": 78820 + }, + { + "epoch": 19.05, + "learning_rate": 2.0362880575216267e-07, + "loss": 0.0001, + "step": 78830 + }, + { + "epoch": 19.05, + "learning_rate": 1.9660712279519154e-07, + "loss": 0.0, + "step": 78840 + }, + { + "epoch": 19.05, + "learning_rate": 1.8958543983822043e-07, + "loss": 0.0, + "step": 78850 + }, + { + "epoch": 19.05, + "learning_rate": 1.825637568812493e-07, + "loss": 0.0, + "step": 78860 + }, + { + "epoch": 19.05, + "learning_rate": 1.755420739242782e-07, + "loss": 0.0, + "step": 78870 + }, + { + "epoch": 19.05, + "learning_rate": 1.6852039096730706e-07, + "loss": 0.0004, + "step": 78880 + }, + { + "epoch": 19.05, + "learning_rate": 1.6149870801033593e-07, + "loss": 0.0, + "step": 78890 + }, + { + "epoch": 19.05, + "learning_rate": 1.544770250533648e-07, + "loss": 0.0, + "step": 78900 + }, + { + "epoch": 19.05, + "learning_rate": 1.4745534209639366e-07, + "loss": 0.0, + "step": 78910 + }, + { + "epoch": 19.05, + "learning_rate": 1.4043365913942256e-07, + "loss": 0.0, + "step": 78920 + }, + { + "epoch": 19.05, + "learning_rate": 1.334119761824514e-07, + "loss": 0.0, + "step": 78930 + }, + { + "epoch": 19.05, + "learning_rate": 1.263902932254803e-07, + "loss": 0.0, + "step": 78940 + }, + { + "epoch": 19.05, + "learning_rate": 1.1936861026850916e-07, + "loss": 0.0, + "step": 78950 + }, + { + "epoch": 19.05, + "learning_rate": 1.1234692731153804e-07, + "loss": 0.0, + "step": 78960 + }, + { + "epoch": 19.05, + "learning_rate": 1.0532524435456692e-07, + "loss": 0.0, + "step": 78970 + }, + { + "epoch": 19.05, + "learning_rate": 9.830356139759577e-08, + "loss": 0.0, + "step": 78980 + }, + { + "epoch": 19.05, + "learning_rate": 9.128187844062465e-08, + "loss": 0.0, + "step": 78990 + }, + { + "epoch": 19.05, + "learning_rate": 8.426019548365353e-08, + "loss": 0.0001, + "step": 79000 + }, + { + "epoch": 19.05, + "learning_rate": 7.72385125266824e-08, + "loss": 0.0, + "step": 79010 + }, + { + "epoch": 19.05, + "learning_rate": 7.021682956971128e-08, + "loss": 0.0, + "step": 79020 + }, + { + "epoch": 19.05, + "learning_rate": 6.319514661274014e-08, + "loss": 0.0, + "step": 79030 + }, + { + "epoch": 19.05, + "learning_rate": 5.617346365576902e-08, + "loss": 0.0, + "step": 79040 + }, + { + "epoch": 19.05, + "learning_rate": 4.9151780698797885e-08, + "loss": 0.0, + "step": 79050 + }, + { + "epoch": 19.05, + "learning_rate": 4.2130097741826765e-08, + "loss": 0.0, + "step": 79060 + }, + { + "epoch": 19.05, + "learning_rate": 3.510841478485564e-08, + "loss": 0.0, + "step": 79070 + }, + { + "epoch": 19.05, + "learning_rate": 2.808673182788451e-08, + "loss": 0.0005, + "step": 79080 + }, + { + "epoch": 19.05, + "learning_rate": 2.1065048870913383e-08, + "loss": 0.0, + "step": 79090 + }, + { + "epoch": 19.05, + "learning_rate": 1.4043365913942255e-08, + "loss": 0.0, + "step": 79100 + }, + { + "epoch": 19.05, + "learning_rate": 7.021682956971127e-09, + "loss": 0.0, + "step": 79110 + }, + { + "epoch": 19.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 79120 + }, + { + "epoch": 19.05, + "eval_accuracy": 0.776286930718617, + "eval_loss": 1.5900887250900269, + "eval_runtime": 1282.4743, + "eval_samples_per_second": 6.044, + "eval_steps_per_second": 3.022, + "step": 79120 + }, + { + "epoch": 19.05, + "step": 79120, + "total_flos": 1.9735134763473175e+20, + "train_loss": 0.41154708696418907, + "train_runtime": 77030.0979, + "train_samples_per_second": 2.054, + "train_steps_per_second": 1.027 + }, + { + "epoch": 19.05, + "eval_accuracy": 0.770395509724579, + "eval_loss": 1.8347108364105225, + "eval_runtime": 1273.3257, + "eval_samples_per_second": 6.017, + "eval_steps_per_second": 3.009, + "step": 79120 + } + ], + "max_steps": 79120, + "num_train_epochs": 9223372036854775807, + "total_flos": 1.9735134763473175e+20, + "trial_name": null, + "trial_params": null +}