| { | |
| "best_metric": 0.3499479191731958, | |
| "best_model_checkpoint": "/content/drive/MyDrive/model_outputs/trial_2/checkpoint-1752", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 1752, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017123287671232876, | |
| "grad_norm": 11.892024993896484, | |
| "learning_rate": 7.592988034631688e-07, | |
| "loss": 2.215, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03424657534246575, | |
| "grad_norm": 11.759550094604492, | |
| "learning_rate": 1.5185976069263376e-06, | |
| "loss": 2.2173, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05136986301369863, | |
| "grad_norm": 11.668200492858887, | |
| "learning_rate": 2.2778964103895065e-06, | |
| "loss": 2.2151, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0684931506849315, | |
| "grad_norm": 10.476677894592285, | |
| "learning_rate": 3.0371952138526753e-06, | |
| "loss": 2.1641, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08561643835616438, | |
| "grad_norm": 9.34625244140625, | |
| "learning_rate": 3.7964940173158444e-06, | |
| "loss": 2.1653, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10273972602739725, | |
| "grad_norm": 9.768696784973145, | |
| "learning_rate": 4.555792820779013e-06, | |
| "loss": 2.1354, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11986301369863013, | |
| "grad_norm": 8.377198219299316, | |
| "learning_rate": 5.315091624242182e-06, | |
| "loss": 2.1244, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.136986301369863, | |
| "grad_norm": 11.591525077819824, | |
| "learning_rate": 6.0743904277053505e-06, | |
| "loss": 2.0764, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1541095890410959, | |
| "grad_norm": 9.914732933044434, | |
| "learning_rate": 6.83368923116852e-06, | |
| "loss": 2.1816, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17123287671232876, | |
| "grad_norm": 11.065092086791992, | |
| "learning_rate": 7.592988034631689e-06, | |
| "loss": 2.0771, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18835616438356165, | |
| "grad_norm": 11.352612495422363, | |
| "learning_rate": 8.352286838094857e-06, | |
| "loss": 2.0857, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2054794520547945, | |
| "grad_norm": 11.720881462097168, | |
| "learning_rate": 9.111585641558026e-06, | |
| "loss": 2.0766, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2226027397260274, | |
| "grad_norm": 10.508756637573242, | |
| "learning_rate": 9.870884445021195e-06, | |
| "loss": 2.1546, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.23972602739726026, | |
| "grad_norm": 11.728009223937988, | |
| "learning_rate": 1.0630183248484364e-05, | |
| "loss": 2.0462, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2568493150684932, | |
| "grad_norm": 9.81351089477539, | |
| "learning_rate": 1.1389482051947532e-05, | |
| "loss": 2.0667, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.273972602739726, | |
| "grad_norm": 8.76012134552002, | |
| "learning_rate": 1.2148780855410701e-05, | |
| "loss": 2.0924, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2910958904109589, | |
| "grad_norm": 9.990751266479492, | |
| "learning_rate": 1.290807965887387e-05, | |
| "loss": 2.0672, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3082191780821918, | |
| "grad_norm": 10.576468467712402, | |
| "learning_rate": 1.366737846233704e-05, | |
| "loss": 2.0109, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3253424657534247, | |
| "grad_norm": 9.048011779785156, | |
| "learning_rate": 1.4426677265800208e-05, | |
| "loss": 2.0392, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3424657534246575, | |
| "grad_norm": 11.787674903869629, | |
| "learning_rate": 1.5185976069263378e-05, | |
| "loss": 2.034, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3595890410958904, | |
| "grad_norm": 10.221675872802734, | |
| "learning_rate": 1.5945274872726547e-05, | |
| "loss": 1.9762, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3767123287671233, | |
| "grad_norm": 9.597102165222168, | |
| "learning_rate": 1.6704573676189714e-05, | |
| "loss": 2.0765, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3938356164383562, | |
| "grad_norm": 11.017401695251465, | |
| "learning_rate": 1.7463872479652885e-05, | |
| "loss": 2.0997, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.410958904109589, | |
| "grad_norm": 10.641432762145996, | |
| "learning_rate": 1.8223171283116052e-05, | |
| "loss": 2.0985, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4280821917808219, | |
| "grad_norm": 8.237184524536133, | |
| "learning_rate": 1.8982470086579223e-05, | |
| "loss": 2.056, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4452054794520548, | |
| "grad_norm": 8.095678329467773, | |
| "learning_rate": 1.974176889004239e-05, | |
| "loss": 2.0807, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4623287671232877, | |
| "grad_norm": 9.02453899383545, | |
| "learning_rate": 2.050106769350556e-05, | |
| "loss": 2.102, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4794520547945205, | |
| "grad_norm": 9.920637130737305, | |
| "learning_rate": 2.126036649696873e-05, | |
| "loss": 1.9904, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4965753424657534, | |
| "grad_norm": 10.004680633544922, | |
| "learning_rate": 2.2019665300431896e-05, | |
| "loss": 2.0086, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5136986301369864, | |
| "grad_norm": 8.83816909790039, | |
| "learning_rate": 2.2140715952718072e-05, | |
| "loss": 1.9748, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5308219178082192, | |
| "grad_norm": 8.490972518920898, | |
| "learning_rate": 2.1988231683071254e-05, | |
| "loss": 2.0026, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.547945205479452, | |
| "grad_norm": 9.475632667541504, | |
| "learning_rate": 2.1835747413424435e-05, | |
| "loss": 2.0512, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.565068493150685, | |
| "grad_norm": 9.864670753479004, | |
| "learning_rate": 2.1683263143777617e-05, | |
| "loss": 2.0304, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5821917808219178, | |
| "grad_norm": 10.569999694824219, | |
| "learning_rate": 2.15307788741308e-05, | |
| "loss": 1.9682, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5993150684931506, | |
| "grad_norm": 9.928328514099121, | |
| "learning_rate": 2.137829460448398e-05, | |
| "loss": 2.0027, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6164383561643836, | |
| "grad_norm": 8.040825843811035, | |
| "learning_rate": 2.1225810334837162e-05, | |
| "loss": 1.9177, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6335616438356164, | |
| "grad_norm": 10.714200019836426, | |
| "learning_rate": 2.1073326065190344e-05, | |
| "loss": 2.0449, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6506849315068494, | |
| "grad_norm": 8.81334114074707, | |
| "learning_rate": 2.0920841795543526e-05, | |
| "loss": 2.0512, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6678082191780822, | |
| "grad_norm": 10.270297050476074, | |
| "learning_rate": 2.0768357525896707e-05, | |
| "loss": 1.9713, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.684931506849315, | |
| "grad_norm": 10.366915702819824, | |
| "learning_rate": 2.061587325624989e-05, | |
| "loss": 1.9791, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.702054794520548, | |
| "grad_norm": 10.001801490783691, | |
| "learning_rate": 2.0463388986603067e-05, | |
| "loss": 1.9183, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7191780821917808, | |
| "grad_norm": 11.41006088256836, | |
| "learning_rate": 2.031090471695625e-05, | |
| "loss": 1.9395, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7363013698630136, | |
| "grad_norm": 9.574875831604004, | |
| "learning_rate": 2.015842044730943e-05, | |
| "loss": 1.9356, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7534246575342466, | |
| "grad_norm": 11.301505088806152, | |
| "learning_rate": 2.0005936177662612e-05, | |
| "loss": 2.059, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7705479452054794, | |
| "grad_norm": 11.491235733032227, | |
| "learning_rate": 1.9853451908015794e-05, | |
| "loss": 2.0022, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7876712328767124, | |
| "grad_norm": 9.770740509033203, | |
| "learning_rate": 1.9700967638368976e-05, | |
| "loss": 1.9812, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8047945205479452, | |
| "grad_norm": 9.801970481872559, | |
| "learning_rate": 1.9548483368722154e-05, | |
| "loss": 2.0287, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.821917808219178, | |
| "grad_norm": 10.885910034179688, | |
| "learning_rate": 1.9395999099075336e-05, | |
| "loss": 2.0574, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.839041095890411, | |
| "grad_norm": 10.215595245361328, | |
| "learning_rate": 1.9243514829428517e-05, | |
| "loss": 2.0379, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8561643835616438, | |
| "grad_norm": 10.391200065612793, | |
| "learning_rate": 1.9091030559781702e-05, | |
| "loss": 1.9701, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8732876712328768, | |
| "grad_norm": 10.24344253540039, | |
| "learning_rate": 1.893854629013488e-05, | |
| "loss": 1.9256, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.8904109589041096, | |
| "grad_norm": 11.228612899780273, | |
| "learning_rate": 1.8786062020488062e-05, | |
| "loss": 2.029, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9075342465753424, | |
| "grad_norm": 13.008833885192871, | |
| "learning_rate": 1.8633577750841244e-05, | |
| "loss": 1.9837, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9246575342465754, | |
| "grad_norm": 10.637398719787598, | |
| "learning_rate": 1.8481093481194426e-05, | |
| "loss": 1.8445, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9417808219178082, | |
| "grad_norm": 13.296565055847168, | |
| "learning_rate": 1.8328609211547607e-05, | |
| "loss": 1.9631, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.958904109589041, | |
| "grad_norm": 11.811957359313965, | |
| "learning_rate": 1.817612494190079e-05, | |
| "loss": 1.961, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.976027397260274, | |
| "grad_norm": 11.553853988647461, | |
| "learning_rate": 1.8023640672253967e-05, | |
| "loss": 1.9842, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.9931506849315068, | |
| "grad_norm": 12.016478538513184, | |
| "learning_rate": 1.787115640260715e-05, | |
| "loss": 1.9183, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.273, | |
| "ar": { | |
| "f1-score": 0.22297297297297297, | |
| "precision": 0.36666666666666664, | |
| "recall": 0.16019417475728157, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.06876790830945559, | |
| "precision": 0.2033898305084746, | |
| "recall": 0.041379310344827586, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.3118081180811808, | |
| "precision": 0.21311475409836064, | |
| "recall": 0.5807560137457045, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.33212996389891697, | |
| "precision": 0.33454545454545453, | |
| "recall": 0.32974910394265233, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.2506678702739225, | |
| "precision": 0.29977219111270315, | |
| "recall": 0.2607198556236934, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.26535626535626533, | |
| "precision": 0.46551724137931033, | |
| "recall": 0.18556701030927836, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.21484375, | |
| "precision": 0.248868778280543, | |
| "recall": 0.18900343642611683, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.574585635359116, | |
| "precision": 0.65, | |
| "recall": 0.5148514851485149, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.26554621848739496, | |
| "precision": 0.21584699453551912, | |
| "recall": 0.34497816593886466, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.2539284929025767, | |
| "precision": 0.30641812645562344, | |
| "recall": 0.273, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.2506678702739225, | |
| "eval_loss": 1.8817169666290283, | |
| "eval_runtime": 2.8277, | |
| "eval_samples_per_second": 707.278, | |
| "eval_steps_per_second": 88.41, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.0102739726027397, | |
| "grad_norm": 11.690625190734863, | |
| "learning_rate": 1.771867213296033e-05, | |
| "loss": 1.8131, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0273972602739727, | |
| "grad_norm": 12.48835563659668, | |
| "learning_rate": 1.7566187863313512e-05, | |
| "loss": 1.7617, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.0445205479452055, | |
| "grad_norm": 15.49386978149414, | |
| "learning_rate": 1.7413703593666694e-05, | |
| "loss": 1.7815, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.0616438356164384, | |
| "grad_norm": 17.888931274414062, | |
| "learning_rate": 1.7261219324019876e-05, | |
| "loss": 1.7337, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.0787671232876712, | |
| "grad_norm": 15.020529747009277, | |
| "learning_rate": 1.7108735054373058e-05, | |
| "loss": 1.6467, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.095890410958904, | |
| "grad_norm": 13.33208179473877, | |
| "learning_rate": 1.695625078472624e-05, | |
| "loss": 1.8363, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.1130136986301369, | |
| "grad_norm": 18.24432373046875, | |
| "learning_rate": 1.680376651507942e-05, | |
| "loss": 1.9009, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.13013698630137, | |
| "grad_norm": 13.63119125366211, | |
| "learning_rate": 1.6651282245432603e-05, | |
| "loss": 1.7742, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.1472602739726028, | |
| "grad_norm": 15.010072708129883, | |
| "learning_rate": 1.6498797975785784e-05, | |
| "loss": 1.6258, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.1643835616438356, | |
| "grad_norm": 11.895198822021484, | |
| "learning_rate": 1.6346313706138963e-05, | |
| "loss": 1.8729, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.1815068493150684, | |
| "grad_norm": 14.77038288116455, | |
| "learning_rate": 1.6193829436492144e-05, | |
| "loss": 1.7128, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.1986301369863013, | |
| "grad_norm": 11.579257011413574, | |
| "learning_rate": 1.6041345166845326e-05, | |
| "loss": 1.8186, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.2157534246575343, | |
| "grad_norm": 13.447906494140625, | |
| "learning_rate": 1.5888860897198508e-05, | |
| "loss": 1.7262, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.2328767123287672, | |
| "grad_norm": 16.56707763671875, | |
| "learning_rate": 1.573637662755169e-05, | |
| "loss": 1.8165, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 12.793381690979004, | |
| "learning_rate": 1.558389235790487e-05, | |
| "loss": 1.7352, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.2671232876712328, | |
| "grad_norm": 14.821900367736816, | |
| "learning_rate": 1.543140808825805e-05, | |
| "loss": 1.8093, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.2842465753424657, | |
| "grad_norm": 13.674068450927734, | |
| "learning_rate": 1.527892381861123e-05, | |
| "loss": 1.7997, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3013698630136985, | |
| "grad_norm": 21.423152923583984, | |
| "learning_rate": 1.5126439548964413e-05, | |
| "loss": 1.8501, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3184931506849316, | |
| "grad_norm": 13.842451095581055, | |
| "learning_rate": 1.4973955279317594e-05, | |
| "loss": 1.6396, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.3356164383561644, | |
| "grad_norm": 12.872172355651855, | |
| "learning_rate": 1.4821471009670778e-05, | |
| "loss": 1.7377, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.3527397260273972, | |
| "grad_norm": 18.74566650390625, | |
| "learning_rate": 1.466898674002396e-05, | |
| "loss": 1.7576, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.36986301369863, | |
| "grad_norm": 15.825615882873535, | |
| "learning_rate": 1.451650247037714e-05, | |
| "loss": 1.738, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.3869863013698631, | |
| "grad_norm": 13.375591278076172, | |
| "learning_rate": 1.4364018200730321e-05, | |
| "loss": 1.6459, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.404109589041096, | |
| "grad_norm": 17.51374053955078, | |
| "learning_rate": 1.4211533931083503e-05, | |
| "loss": 1.7292, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4212328767123288, | |
| "grad_norm": 15.895259857177734, | |
| "learning_rate": 1.4059049661436683e-05, | |
| "loss": 1.8173, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.4383561643835616, | |
| "grad_norm": 24.653060913085938, | |
| "learning_rate": 1.3906565391789864e-05, | |
| "loss": 1.7303, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.4554794520547945, | |
| "grad_norm": 19.205734252929688, | |
| "learning_rate": 1.3754081122143046e-05, | |
| "loss": 1.5581, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.4726027397260273, | |
| "grad_norm": 13.14504337310791, | |
| "learning_rate": 1.3601596852496226e-05, | |
| "loss": 1.7028, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.4897260273972603, | |
| "grad_norm": 19.409351348876953, | |
| "learning_rate": 1.3449112582849408e-05, | |
| "loss": 1.7864, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5068493150684932, | |
| "grad_norm": 20.153881072998047, | |
| "learning_rate": 1.329662831320259e-05, | |
| "loss": 1.5898, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.523972602739726, | |
| "grad_norm": 13.584912300109863, | |
| "learning_rate": 1.314414404355577e-05, | |
| "loss": 1.6459, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.541095890410959, | |
| "grad_norm": 20.62996482849121, | |
| "learning_rate": 1.2991659773908951e-05, | |
| "loss": 1.8207, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.558219178082192, | |
| "grad_norm": 16.82609748840332, | |
| "learning_rate": 1.2839175504262133e-05, | |
| "loss": 1.6256, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.5753424657534247, | |
| "grad_norm": 19.674978256225586, | |
| "learning_rate": 1.2686691234615316e-05, | |
| "loss": 1.6743, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.5924657534246576, | |
| "grad_norm": 23.367334365844727, | |
| "learning_rate": 1.2534206964968498e-05, | |
| "loss": 1.602, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.6095890410958904, | |
| "grad_norm": 19.46713638305664, | |
| "learning_rate": 1.2381722695321678e-05, | |
| "loss": 1.6916, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.6267123287671232, | |
| "grad_norm": 24.35736083984375, | |
| "learning_rate": 1.222923842567486e-05, | |
| "loss": 1.6058, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.643835616438356, | |
| "grad_norm": 24.921846389770508, | |
| "learning_rate": 1.2076754156028041e-05, | |
| "loss": 1.7183, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.660958904109589, | |
| "grad_norm": 15.05435848236084, | |
| "learning_rate": 1.1924269886381221e-05, | |
| "loss": 1.6577, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.678082191780822, | |
| "grad_norm": 15.788996696472168, | |
| "learning_rate": 1.1771785616734403e-05, | |
| "loss": 1.6803, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.6952054794520548, | |
| "grad_norm": 18.309133529663086, | |
| "learning_rate": 1.1619301347087585e-05, | |
| "loss": 1.5949, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.7123287671232876, | |
| "grad_norm": 20.53139877319336, | |
| "learning_rate": 1.1466817077440765e-05, | |
| "loss": 1.6097, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.7294520547945207, | |
| "grad_norm": 23.069534301757812, | |
| "learning_rate": 1.1314332807793946e-05, | |
| "loss": 1.6068, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.7465753424657535, | |
| "grad_norm": 22.394153594970703, | |
| "learning_rate": 1.1161848538147128e-05, | |
| "loss": 1.6641, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.7636986301369864, | |
| "grad_norm": 20.755950927734375, | |
| "learning_rate": 1.100936426850031e-05, | |
| "loss": 1.4966, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.7808219178082192, | |
| "grad_norm": 16.427976608276367, | |
| "learning_rate": 1.0856879998853491e-05, | |
| "loss": 1.7212, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.797945205479452, | |
| "grad_norm": 33.8179817199707, | |
| "learning_rate": 1.0704395729206671e-05, | |
| "loss": 1.7811, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.8150684931506849, | |
| "grad_norm": 20.979110717773438, | |
| "learning_rate": 1.0551911459559853e-05, | |
| "loss": 1.583, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.8321917808219177, | |
| "grad_norm": 17.085895538330078, | |
| "learning_rate": 1.0399427189913035e-05, | |
| "loss": 1.7014, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.8493150684931505, | |
| "grad_norm": 18.183605194091797, | |
| "learning_rate": 1.0246942920266215e-05, | |
| "loss": 1.5758, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.8664383561643836, | |
| "grad_norm": 16.887876510620117, | |
| "learning_rate": 1.0094458650619396e-05, | |
| "loss": 1.4666, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.8835616438356164, | |
| "grad_norm": 18.789949417114258, | |
| "learning_rate": 9.94197438097258e-06, | |
| "loss": 1.7645, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.9006849315068495, | |
| "grad_norm": 18.971622467041016, | |
| "learning_rate": 9.78949011132576e-06, | |
| "loss": 1.6875, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.9178082191780823, | |
| "grad_norm": 24.858366012573242, | |
| "learning_rate": 9.637005841678942e-06, | |
| "loss": 1.6846, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.9349315068493151, | |
| "grad_norm": 13.931944847106934, | |
| "learning_rate": 9.484521572032123e-06, | |
| "loss": 1.6025, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.952054794520548, | |
| "grad_norm": 17.908315658569336, | |
| "learning_rate": 9.332037302385303e-06, | |
| "loss": 1.4988, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.9691780821917808, | |
| "grad_norm": 16.93858528137207, | |
| "learning_rate": 9.179553032738485e-06, | |
| "loss": 1.7157, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.9863013698630136, | |
| "grad_norm": 19.648006439208984, | |
| "learning_rate": 9.027068763091667e-06, | |
| "loss": 1.5337, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.3535, | |
| "ar": { | |
| "f1-score": 0.31309904153354634, | |
| "precision": 0.45794392523364486, | |
| "recall": 0.23786407766990292, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.23689320388349513, | |
| "precision": 0.27111111111111114, | |
| "recall": 0.2103448275862069, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.36268068331143233, | |
| "precision": 0.2936170212765957, | |
| "recall": 0.4742268041237113, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.38976377952755903, | |
| "precision": 0.43231441048034935, | |
| "recall": 0.3548387096774194, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.3353596555098448, | |
| "precision": 0.3519727641328481, | |
| "recall": 0.3389548790003148, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.36254980079681276, | |
| "precision": 0.4312796208530806, | |
| "recall": 0.3127147766323024, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.31550068587105623, | |
| "precision": 0.2625570776255708, | |
| "recall": 0.3951890034364261, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6543778801843319, | |
| "precision": 0.6120689655172413, | |
| "recall": 0.7029702970297029, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.3833718244803695, | |
| "precision": 0.4068627450980392, | |
| "recall": 0.3624454148471616, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.34933930516945594, | |
| "precision": 0.3679569789689199, | |
| "recall": 0.3535, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.3353596555098448, | |
| "eval_loss": 1.793221116065979, | |
| "eval_runtime": 2.7925, | |
| "eval_samples_per_second": 716.208, | |
| "eval_steps_per_second": 89.526, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.0034246575342465, | |
| "grad_norm": 20.4486083984375, | |
| "learning_rate": 8.874584493444848e-06, | |
| "loss": 1.5645, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.0205479452054793, | |
| "grad_norm": 17.64771270751953, | |
| "learning_rate": 8.72210022379803e-06, | |
| "loss": 1.3974, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.037671232876712, | |
| "grad_norm": 13.944244384765625, | |
| "learning_rate": 8.56961595415121e-06, | |
| "loss": 1.2199, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.0547945205479454, | |
| "grad_norm": 16.506717681884766, | |
| "learning_rate": 8.417131684504392e-06, | |
| "loss": 1.2971, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.0719178082191783, | |
| "grad_norm": 22.669170379638672, | |
| "learning_rate": 8.264647414857573e-06, | |
| "loss": 1.2465, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.089041095890411, | |
| "grad_norm": 20.12689971923828, | |
| "learning_rate": 8.112163145210753e-06, | |
| "loss": 1.1901, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.106164383561644, | |
| "grad_norm": 16.720157623291016, | |
| "learning_rate": 7.959678875563935e-06, | |
| "loss": 1.1476, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.1232876712328768, | |
| "grad_norm": 13.676715850830078, | |
| "learning_rate": 7.807194605917117e-06, | |
| "loss": 1.2803, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.1404109589041096, | |
| "grad_norm": 16.257658004760742, | |
| "learning_rate": 7.654710336270298e-06, | |
| "loss": 1.148, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.1575342465753424, | |
| "grad_norm": 20.69976234436035, | |
| "learning_rate": 7.50222606662348e-06, | |
| "loss": 1.3063, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.1746575342465753, | |
| "grad_norm": 21.628828048706055, | |
| "learning_rate": 7.349741796976661e-06, | |
| "loss": 1.3074, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.191780821917808, | |
| "grad_norm": 18.611053466796875, | |
| "learning_rate": 7.197257527329842e-06, | |
| "loss": 1.2918, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.208904109589041, | |
| "grad_norm": 19.288227081298828, | |
| "learning_rate": 7.044773257683023e-06, | |
| "loss": 1.1346, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.2260273972602738, | |
| "grad_norm": 27.32809066772461, | |
| "learning_rate": 6.892288988036204e-06, | |
| "loss": 1.4171, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.243150684931507, | |
| "grad_norm": 26.971960067749023, | |
| "learning_rate": 6.739804718389387e-06, | |
| "loss": 1.0316, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.26027397260274, | |
| "grad_norm": 30.222131729125977, | |
| "learning_rate": 6.587320448742568e-06, | |
| "loss": 1.1692, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.2773972602739727, | |
| "grad_norm": 20.51207160949707, | |
| "learning_rate": 6.434836179095749e-06, | |
| "loss": 1.0899, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.2945205479452055, | |
| "grad_norm": 19.466644287109375, | |
| "learning_rate": 6.28235190944893e-06, | |
| "loss": 1.2698, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.3116438356164384, | |
| "grad_norm": 21.64017677307129, | |
| "learning_rate": 6.129867639802111e-06, | |
| "loss": 1.244, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.328767123287671, | |
| "grad_norm": 20.915922164916992, | |
| "learning_rate": 5.977383370155293e-06, | |
| "loss": 1.0774, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.345890410958904, | |
| "grad_norm": 24.89851188659668, | |
| "learning_rate": 5.8248991005084735e-06, | |
| "loss": 1.2141, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.363013698630137, | |
| "grad_norm": 24.925865173339844, | |
| "learning_rate": 5.672414830861656e-06, | |
| "loss": 1.2304, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.3801369863013697, | |
| "grad_norm": 27.173961639404297, | |
| "learning_rate": 5.519930561214836e-06, | |
| "loss": 1.1165, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.3972602739726026, | |
| "grad_norm": 19.15679931640625, | |
| "learning_rate": 5.367446291568018e-06, | |
| "loss": 1.3572, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.4143835616438354, | |
| "grad_norm": 20.516983032226562, | |
| "learning_rate": 5.214962021921199e-06, | |
| "loss": 1.2764, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.4315068493150687, | |
| "grad_norm": 19.259366989135742, | |
| "learning_rate": 5.06247775227438e-06, | |
| "loss": 1.2762, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.4486301369863015, | |
| "grad_norm": 25.636960983276367, | |
| "learning_rate": 4.909993482627562e-06, | |
| "loss": 1.1584, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.4657534246575343, | |
| "grad_norm": 17.860868453979492, | |
| "learning_rate": 4.757509212980744e-06, | |
| "loss": 0.9371, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.482876712328767, | |
| "grad_norm": 31.28371238708496, | |
| "learning_rate": 4.6050249433339244e-06, | |
| "loss": 1.1262, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 19.444276809692383, | |
| "learning_rate": 4.452540673687105e-06, | |
| "loss": 1.2705, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.517123287671233, | |
| "grad_norm": 16.713764190673828, | |
| "learning_rate": 4.300056404040287e-06, | |
| "loss": 1.1658, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.5342465753424657, | |
| "grad_norm": 24.59725570678711, | |
| "learning_rate": 4.147572134393469e-06, | |
| "loss": 1.303, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.5513698630136985, | |
| "grad_norm": 28.553564071655273, | |
| "learning_rate": 3.9950878647466495e-06, | |
| "loss": 1.2434, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.5684931506849313, | |
| "grad_norm": 25.321197509765625, | |
| "learning_rate": 3.84260359509983e-06, | |
| "loss": 1.2506, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.5856164383561646, | |
| "grad_norm": 29.10701560974121, | |
| "learning_rate": 3.6901193254530124e-06, | |
| "loss": 1.3008, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.602739726027397, | |
| "grad_norm": 23.615144729614258, | |
| "learning_rate": 3.5376350558061937e-06, | |
| "loss": 1.0486, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.6198630136986303, | |
| "grad_norm": 28.57093620300293, | |
| "learning_rate": 3.3851507861593745e-06, | |
| "loss": 1.2169, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.636986301369863, | |
| "grad_norm": 18.41623306274414, | |
| "learning_rate": 3.2326665165125566e-06, | |
| "loss": 1.2028, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.654109589041096, | |
| "grad_norm": 23.891878128051758, | |
| "learning_rate": 3.0801822468657375e-06, | |
| "loss": 1.2732, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.671232876712329, | |
| "grad_norm": 23.33771514892578, | |
| "learning_rate": 2.9276979772189187e-06, | |
| "loss": 1.2175, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.6883561643835616, | |
| "grad_norm": 17.881248474121094, | |
| "learning_rate": 2.7752137075721e-06, | |
| "loss": 1.2555, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.7054794520547945, | |
| "grad_norm": 19.803930282592773, | |
| "learning_rate": 2.6227294379252813e-06, | |
| "loss": 1.2051, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.7226027397260273, | |
| "grad_norm": 22.89762306213379, | |
| "learning_rate": 2.470245168278463e-06, | |
| "loss": 1.1713, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.73972602739726, | |
| "grad_norm": 16.94407844543457, | |
| "learning_rate": 2.3177608986316442e-06, | |
| "loss": 1.1175, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.756849315068493, | |
| "grad_norm": 23.772077560424805, | |
| "learning_rate": 2.1652766289848255e-06, | |
| "loss": 1.1266, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.7739726027397262, | |
| "grad_norm": 38.49664306640625, | |
| "learning_rate": 2.0127923593380067e-06, | |
| "loss": 1.3238, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.791095890410959, | |
| "grad_norm": 27.785099029541016, | |
| "learning_rate": 1.860308089691188e-06, | |
| "loss": 1.1356, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.808219178082192, | |
| "grad_norm": 34.19268798828125, | |
| "learning_rate": 1.7078238200443693e-06, | |
| "loss": 1.3087, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.8253424657534247, | |
| "grad_norm": 24.389307022094727, | |
| "learning_rate": 1.5553395503975505e-06, | |
| "loss": 1.3934, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.8424657534246576, | |
| "grad_norm": 41.80609130859375, | |
| "learning_rate": 1.402855280750732e-06, | |
| "loss": 1.2443, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.8595890410958904, | |
| "grad_norm": 25.80476188659668, | |
| "learning_rate": 1.2503710111039133e-06, | |
| "loss": 1.3135, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.8767123287671232, | |
| "grad_norm": 22.425390243530273, | |
| "learning_rate": 1.0978867414570945e-06, | |
| "loss": 1.0686, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.893835616438356, | |
| "grad_norm": 35.875511169433594, | |
| "learning_rate": 9.454024718102758e-07, | |
| "loss": 1.2933, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.910958904109589, | |
| "grad_norm": 43.623722076416016, | |
| "learning_rate": 7.929182021634573e-07, | |
| "loss": 1.2489, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.928082191780822, | |
| "grad_norm": 29.01580047607422, | |
| "learning_rate": 6.404339325166384e-07, | |
| "loss": 1.2077, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.9452054794520546, | |
| "grad_norm": 19.448177337646484, | |
| "learning_rate": 4.879496628698198e-07, | |
| "loss": 1.0898, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.962328767123288, | |
| "grad_norm": 29.721946716308594, | |
| "learning_rate": 3.354653932230011e-07, | |
| "loss": 1.1101, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.9794520547945207, | |
| "grad_norm": 23.694095611572266, | |
| "learning_rate": 1.8298112357618242e-07, | |
| "loss": 1.0485, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.9965753424657535, | |
| "grad_norm": 27.28858757019043, | |
| "learning_rate": 3.049685392936374e-08, | |
| "loss": 1.1386, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.368, | |
| "ar": { | |
| "f1-score": 0.34759358288770054, | |
| "precision": 0.3869047619047619, | |
| "recall": 0.3155339805825243, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.23636363636363636, | |
| "precision": 0.25, | |
| "recall": 0.22413793103448276, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.38351822503961963, | |
| "precision": 0.3558823529411765, | |
| "recall": 0.41580756013745707, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.39919354838709675, | |
| "precision": 0.45622119815668205, | |
| "recall": 0.3548387096774194, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.3499479191731958, | |
| "precision": 0.3551406608650635, | |
| "recall": 0.3502956304842808, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.38202247191011235, | |
| "precision": 0.35843373493975905, | |
| "recall": 0.40893470790378006, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.3446244477172312, | |
| "precision": 0.3015463917525773, | |
| "recall": 0.4020618556701031, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6666666666666666, | |
| "precision": 0.6601941747572816, | |
| "recall": 0.6732673267326733, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.38954869358669836, | |
| "precision": 0.4270833333333333, | |
| "recall": 0.35807860262008734, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.3655613869415472, | |
| "precision": 0.36979288589763315, | |
| "recall": 0.368, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.3499479191731958, | |
| "eval_loss": 1.8162137269973755, | |
| "eval_runtime": 2.8908, | |
| "eval_samples_per_second": 691.846, | |
| "eval_steps_per_second": 86.481, | |
| "step": 1752 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1752, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 230203751995008.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |