| { |
| "best_global_step": 7000, |
| "best_metric": 0.6139324903488159, |
| "best_model_checkpoint": "/scratch/ameyk/lma/model-finetuned/checkpoints/checkpoint-7000", |
| "epoch": 4.087049808429119, |
| "eval_steps": 500, |
| "global_step": 10000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.04086845466155811, |
| "grad_norm": 60.702266693115234, |
| "learning_rate": 2.9700000000000004e-06, |
| "loss": 17.819, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.08173690932311622, |
| "grad_norm": 47.32134246826172, |
| "learning_rate": 5.9700000000000004e-06, |
| "loss": 7.7058, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.12260536398467432, |
| "grad_norm": 31.054290771484375, |
| "learning_rate": 8.97e-06, |
| "loss": 5.103, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.16347381864623245, |
| "grad_norm": 25.974103927612305, |
| "learning_rate": 1.197e-05, |
| "loss": 4.0185, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.20434227330779056, |
| "grad_norm": 19.508167266845703, |
| "learning_rate": 1.497e-05, |
| "loss": 3.3012, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.20434227330779056, |
| "eval_accuracy": 0.0, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 2000, |
| 0 |
| ] |
| ], |
| "eval_f1_macro": 0.0, |
| "eval_humor_english_accuracy": 0.0, |
| "eval_humor_english_avg_loss": 0.7475204364605462, |
| "eval_humor_english_f1": 0.0, |
| "eval_humor_english_precision": 0.0, |
| "eval_humor_english_recall": 0.0, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.0, |
| "eval_humor_hindi_avg_loss": 0.591034443622271, |
| "eval_humor_hindi_f1": 0.0, |
| "eval_humor_hindi_precision": 0.0, |
| "eval_humor_hindi_recall": 0.0, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0, |
| "eval_humor_overall_avg_loss": 0.6701178589073499, |
| "eval_humor_overall_f1": 0.0, |
| "eval_humor_overall_precision": 0.0, |
| "eval_humor_overall_recall": 0.0, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 1.5324292182922363, |
| "eval_perplexity": 4.629409284567538, |
| "eval_precision_macro": 0.0, |
| "eval_recall_macro": 0.0, |
| "eval_runtime": 1346.3326, |
| "eval_samples_per_second": 1.486, |
| "eval_steps_per_second": 0.743, |
| "eval_temporal_english_avg_loss": 0.9373093255760758, |
| "eval_temporal_hindi_avg_loss": 1.4527543705181314, |
| "eval_temporal_overall_avg_loss": 1.1912949920150167, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.24521072796934865, |
| "grad_norm": 14.092637062072754, |
| "learning_rate": 1.797e-05, |
| "loss": 2.8053, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.28607918263090676, |
| "grad_norm": 9.543878555297852, |
| "learning_rate": 2.097e-05, |
| "loss": 2.393, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.3269476372924649, |
| "grad_norm": 7.029173851013184, |
| "learning_rate": 2.3970000000000003e-05, |
| "loss": 2.1078, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.367816091954023, |
| "grad_norm": 6.117612361907959, |
| "learning_rate": 2.697e-05, |
| "loss": 1.924, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.4086845466155811, |
| "grad_norm": 6.306148052215576, |
| "learning_rate": 2.997e-05, |
| "loss": 1.7717, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.4086845466155811, |
| "eval_accuracy": 0.0, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 2000, |
| 0 |
| ] |
| ], |
| "eval_f1_macro": 0.0, |
| "eval_humor_english_accuracy": 0.0, |
| "eval_humor_english_avg_loss": 0.5384818354636839, |
| "eval_humor_english_f1": 0.0, |
| "eval_humor_english_precision": 0.0, |
| "eval_humor_english_recall": 0.0, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.0, |
| "eval_humor_hindi_avg_loss": 0.3663946280421942, |
| "eval_humor_hindi_f1": 0.0, |
| "eval_humor_hindi_precision": 0.0, |
| "eval_humor_hindi_recall": 0.0, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0, |
| "eval_humor_overall_avg_loss": 0.4519576721899176, |
| "eval_humor_overall_f1": 0.0, |
| "eval_humor_overall_precision": 0.0, |
| "eval_humor_overall_recall": 0.0, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.9522949457168579, |
| "eval_perplexity": 2.5916505672889882, |
| "eval_precision_macro": 0.0, |
| "eval_recall_macro": 0.0, |
| "eval_runtime": 1361.0739, |
| "eval_samples_per_second": 1.469, |
| "eval_steps_per_second": 0.735, |
| "eval_temporal_english_avg_loss": 0.5830070118860753, |
| "eval_temporal_hindi_avg_loss": 0.9119034483581334, |
| "eval_temporal_overall_avg_loss": 0.7474362385274357, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.4495530012771392, |
| "grad_norm": 5.973597526550293, |
| "learning_rate": 2.9991044225324593e-05, |
| "loss": 1.6392, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.4904214559386973, |
| "grad_norm": 5.6661834716796875, |
| "learning_rate": 2.996382508630622e-05, |
| "loss": 1.5923, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.5312899106002554, |
| "grad_norm": 5.6116790771484375, |
| "learning_rate": 2.9918374831420734e-05, |
| "loss": 1.5159, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.5721583652618135, |
| "grad_norm": 6.842437744140625, |
| "learning_rate": 2.9854748834802644e-05, |
| "loss": 1.5072, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.6130268199233716, |
| "grad_norm": 4.068236827850342, |
| "learning_rate": 2.977302461492799e-05, |
| "loss": 1.362, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.6130268199233716, |
| "eval_accuracy": 0.0, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 2000, |
| 0 |
| ] |
| ], |
| "eval_f1_macro": 0.0, |
| "eval_humor_english_accuracy": 0.0, |
| "eval_humor_english_avg_loss": 0.43862412279962604, |
| "eval_humor_english_f1": 0.0, |
| "eval_humor_english_precision": 0.0, |
| "eval_humor_english_recall": 0.0, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.0, |
| "eval_humor_hindi_avg_loss": 0.2803115557623819, |
| "eval_humor_hindi_f1": 0.0, |
| "eval_humor_hindi_precision": 0.0, |
| "eval_humor_hindi_recall": 0.0, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0, |
| "eval_humor_overall_avg_loss": 0.3596194935054261, |
| "eval_humor_overall_f1": 0.0, |
| "eval_humor_overall_precision": 0.0, |
| "eval_humor_overall_recall": 0.0, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.8224886059761047, |
| "eval_perplexity": 2.2761572637241803, |
| "eval_precision_macro": 0.0, |
| "eval_recall_macro": 0.0, |
| "eval_runtime": 1364.4034, |
| "eval_samples_per_second": 1.466, |
| "eval_steps_per_second": 0.733, |
| "eval_temporal_english_avg_loss": 0.431482280489928, |
| "eval_temporal_hindi_avg_loss": 0.6994093691276505, |
| "eval_temporal_overall_avg_loss": 0.5653328100846917, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.6538952745849298, |
| "grad_norm": 6.672731876373291, |
| "learning_rate": 2.967330174017004e-05, |
| "loss": 1.3573, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.6947637292464879, |
| "grad_norm": 4.476126670837402, |
| "learning_rate": 2.9555701707490556e-05, |
| "loss": 1.3249, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.735632183908046, |
| "grad_norm": 4.317950248718262, |
| "learning_rate": 2.942036779441446e-05, |
| "loss": 1.2761, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.776500638569604, |
| "grad_norm": 4.1325578689575195, |
| "learning_rate": 2.9267464884468245e-05, |
| "loss": 1.2085, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.8173690932311622, |
| "grad_norm": 5.426914215087891, |
| "learning_rate": 2.9097179266294794e-05, |
| "loss": 1.2643, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.8173690932311622, |
| "eval_accuracy": 0.0015, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1997, |
| 3 |
| ] |
| ], |
| "eval_f1_macro": 0.0014977533699450823, |
| "eval_humor_english_accuracy": 0.0030211480362537764, |
| "eval_humor_english_avg_loss": 0.38365533949005975, |
| "eval_humor_english_f1": 0.0030120481927710845, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.0015105740181268882, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.0, |
| "eval_humor_hindi_avg_loss": 0.23210459279744125, |
| "eval_humor_hindi_f1": 0.0, |
| "eval_humor_hindi_precision": 0.0, |
| "eval_humor_hindi_recall": 0.0, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0015, |
| "eval_humor_overall_avg_loss": 0.3081119789219082, |
| "eval_humor_overall_f1": 0.0014977533699450823, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.00075, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.7675837278366089, |
| "eval_perplexity": 2.1545540160759593, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.00075, |
| "eval_runtime": 1360.1194, |
| "eval_samples_per_second": 1.47, |
| "eval_steps_per_second": 0.735, |
| "eval_temporal_english_avg_loss": 0.3506820346533618, |
| "eval_temporal_hindi_avg_loss": 0.5820792646224661, |
| "eval_temporal_overall_avg_loss": 0.4664542024764292, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.8582375478927203, |
| "grad_norm": 5.104623794555664, |
| "learning_rate": 2.8909718406689366e-05, |
| "loss": 1.2051, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.8991060025542784, |
| "grad_norm": 4.696566104888916, |
| "learning_rate": 2.870531069783325e-05, |
| "loss": 1.1831, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.9399744572158365, |
| "grad_norm": 4.637947082519531, |
| "learning_rate": 2.8484205179033096e-05, |
| "loss": 1.1282, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.9808429118773946, |
| "grad_norm": 5.258996486663818, |
| "learning_rate": 2.824667123330487e-05, |
| "loss": 1.1429, |
| "step": 2400 |
| }, |
| { |
| "epoch": 1.021660280970626, |
| "grad_norm": 4.4945597648620605, |
| "learning_rate": 2.7992998259172142e-05, |
| "loss": 1.0593, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.021660280970626, |
| "eval_accuracy": 0.0745, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1851, |
| 149 |
| ] |
| ], |
| "eval_f1_macro": 0.0693345742205677, |
| "eval_humor_english_accuracy": 0.07653575025176233, |
| "eval_humor_english_avg_loss": 0.34921365948140487, |
| "eval_humor_english_f1": 0.07109448082319925, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.038267875125881166, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.07249255213505462, |
| "eval_humor_hindi_avg_loss": 0.20103979772989347, |
| "eval_humor_hindi_f1": 0.06759259259259259, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.03624627606752731, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0745, |
| "eval_humor_overall_avg_loss": 0.27532809556280224, |
| "eval_humor_overall_f1": 0.0693345742205677, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.03725, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.7367945313453674, |
| "eval_perplexity": 2.0892278709676293, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.03725, |
| "eval_runtime": 1359.6405, |
| "eval_samples_per_second": 1.471, |
| "eval_steps_per_second": 0.735, |
| "eval_temporal_english_avg_loss": 0.2974580233862613, |
| "eval_temporal_hindi_avg_loss": 0.5058095154879095, |
| "eval_temporal_overall_avg_loss": 0.40173021953064236, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.062528735632184, |
| "grad_norm": 4.575131416320801, |
| "learning_rate": 2.7723495318078564e-05, |
| "loss": 0.9702, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.103397190293742, |
| "grad_norm": 4.862764835357666, |
| "learning_rate": 2.743849075784412e-05, |
| "loss": 0.9822, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.1442656449553001, |
| "grad_norm": 4.071940898895264, |
| "learning_rate": 2.713833181262386e-05, |
| "loss": 0.9633, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.1851340996168582, |
| "grad_norm": 4.650261878967285, |
| "learning_rate": 2.6823384179856602e-05, |
| "loss": 1.0072, |
| "step": 2900 |
| }, |
| { |
| "epoch": 1.2260025542784163, |
| "grad_norm": 4.2909321784973145, |
| "learning_rate": 2.64940315747189e-05, |
| "loss": 0.9524, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.2260025542784163, |
| "eval_accuracy": 0.111, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1778, |
| 222 |
| ] |
| ], |
| "eval_f1_macro": 0.0999099909990999, |
| "eval_humor_english_accuracy": 0.06847935548841894, |
| "eval_humor_english_avg_loss": 0.3228903253947321, |
| "eval_humor_english_f1": 0.0640904806786051, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.03423967774420947, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.1529294935451837, |
| "eval_humor_hindi_avg_loss": 0.1783881702906315, |
| "eval_humor_hindi_f1": 0.1326442721791559, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.07646474677259185, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.111, |
| "eval_humor_overall_avg_loss": 0.2508712931905895, |
| "eval_humor_overall_f1": 0.0999099909990999, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.0555, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.71168053150177, |
| "eval_perplexity": 2.037412286720291, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.0555, |
| "eval_runtime": 1359.5963, |
| "eval_samples_per_second": 1.471, |
| "eval_steps_per_second": 0.736, |
| "eval_temporal_english_avg_loss": 0.25916867434431917, |
| "eval_temporal_hindi_avg_loss": 0.449419908061452, |
| "eval_temporal_overall_avg_loss": 0.35426122447886077, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.2668710089399744, |
| "grad_norm": 4.282441139221191, |
| "learning_rate": 2.6150675262627243e-05, |
| "loss": 0.9654, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1.3077394636015325, |
| "grad_norm": 4.853513240814209, |
| "learning_rate": 2.5793733570357923e-05, |
| "loss": 0.9484, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1.3486079182630908, |
| "grad_norm": 3.8319053649902344, |
| "learning_rate": 2.5423641376380324e-05, |
| "loss": 0.9162, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1.3894763729246487, |
| "grad_norm": 4.360899448394775, |
| "learning_rate": 2.5040849581024466e-05, |
| "loss": 0.9263, |
| "step": 3400 |
| }, |
| { |
| "epoch": 1.430344827586207, |
| "grad_norm": 3.914219379425049, |
| "learning_rate": 2.464582455712844e-05, |
| "loss": 0.8967, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.430344827586207, |
| "eval_accuracy": 0.0075, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1985, |
| 15 |
| ] |
| ], |
| "eval_f1_macro": 0.007444168734491315, |
| "eval_humor_english_accuracy": 0.004028197381671702, |
| "eval_humor_english_avg_loss": 0.3037518717352607, |
| "eval_humor_english_f1": 0.004012036108324975, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.002014098690835851, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.010923535253227408, |
| "eval_humor_hindi_avg_loss": 0.16131232707652512, |
| "eval_humor_hindi_f1": 0.010805500982318271, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.005461767626613704, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0075, |
| "eval_humor_overall_avg_loss": 0.2326810370445648, |
| "eval_humor_overall_f1": 0.007444168734491315, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.00375, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6824359893798828, |
| "eval_perplexity": 1.9786919166070414, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.00375, |
| "eval_runtime": 1319.0244, |
| "eval_samples_per_second": 1.516, |
| "eval_steps_per_second": 0.758, |
| "eval_temporal_english_avg_loss": 0.23054835682895794, |
| "eval_temporal_hindi_avg_loss": 0.4091248675787164, |
| "eval_temporal_overall_avg_loss": 0.31947562442693717, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.471213282247765, |
| "grad_norm": 4.441721439361572, |
| "learning_rate": 2.423904758183493e-05, |
| "loss": 0.9344, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1.5120817369093231, |
| "grad_norm": 4.262256145477295, |
| "learning_rate": 2.3821014250229128e-05, |
| "loss": 0.8854, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1.5529501915708812, |
| "grad_norm": 4.789700984954834, |
| "learning_rate": 2.3392233871532504e-05, |
| "loss": 0.8741, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1.5938186462324393, |
| "grad_norm": 4.243940830230713, |
| "learning_rate": 2.2953228848587946e-05, |
| "loss": 0.8728, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1.6346871008939976, |
| "grad_norm": 3.6989076137542725, |
| "learning_rate": 2.2504534041392377e-05, |
| "loss": 0.8662, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.6346871008939976, |
| "eval_accuracy": 0.0345, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1931, |
| 69 |
| ] |
| ], |
| "eval_f1_macro": 0.0333494441759304, |
| "eval_humor_english_accuracy": 0.025176233635448138, |
| "eval_humor_english_avg_loss": 0.2894040331679491, |
| "eval_humor_english_f1": 0.02455795677799607, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.012588116817724069, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.04369414101290963, |
| "eval_humor_hindi_avg_loss": 0.1483633538700107, |
| "eval_humor_hindi_f1": 0.04186489058039962, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.021847070506454815, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0345, |
| "eval_humor_overall_avg_loss": 0.21891726321483457, |
| "eval_humor_overall_f1": 0.0333494441759304, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.01725, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6712037324905396, |
| "eval_perplexity": 1.9565911477290814, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.01725, |
| "eval_runtime": 1317.9709, |
| "eval_samples_per_second": 1.517, |
| "eval_steps_per_second": 0.759, |
| "eval_temporal_english_avg_loss": 0.20886689280834833, |
| "eval_temporal_hindi_avg_loss": 0.37523903380129076, |
| "eval_temporal_overall_avg_loss": 0.29192309029086067, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.6755555555555555, |
| "grad_norm": 4.29471492767334, |
| "learning_rate": 2.204669611545222e-05, |
| "loss": 0.8442, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1.7164240102171138, |
| "grad_norm": 3.6822047233581543, |
| "learning_rate": 2.1580272875755707e-05, |
| "loss": 0.8287, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1.7572924648786716, |
| "grad_norm": 3.862697124481201, |
| "learning_rate": 2.1105832587173418e-05, |
| "loss": 0.9005, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1.79816091954023, |
| "grad_norm": 4.998415946960449, |
| "learning_rate": 2.0623953282115073e-05, |
| "loss": 0.8541, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1.839029374201788, |
| "grad_norm": 4.094598770141602, |
| "learning_rate": 2.013522205628606e-05, |
| "loss": 0.8411, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.839029374201788, |
| "eval_accuracy": 0.0705, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1859, |
| 141 |
| ] |
| ], |
| "eval_f1_macro": 0.0658570761326483, |
| "eval_humor_english_accuracy": 0.07452165156092648, |
| "eval_humor_english_avg_loss": 0.27731017850001527, |
| "eval_humor_english_f1": 0.06935332708528585, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.03726082578046324, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.06653426017874876, |
| "eval_humor_hindi_avg_loss": 0.1377803079530714, |
| "eval_humor_hindi_f1": 0.06238361266294227, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.03326713008937438, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0705, |
| "eval_humor_overall_avg_loss": 0.20768647205886417, |
| "eval_humor_overall_f1": 0.0658570761326483, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.03525, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6517492532730103, |
| "eval_perplexity": 1.9188945335622285, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.03525, |
| "eval_runtime": 1361.0394, |
| "eval_samples_per_second": 1.469, |
| "eval_steps_per_second": 0.735, |
| "eval_temporal_english_avg_loss": 0.19179030200470823, |
| "eval_temporal_hindi_avg_loss": 0.348848874167915, |
| "eval_temporal_overall_avg_loss": 0.2703094946072132, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.879897828863346, |
| "grad_norm": 4.929368495941162, |
| "learning_rate": 1.964023435340178e-05, |
| "loss": 0.8146, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1.9207662835249042, |
| "grad_norm": 3.4482617378234863, |
| "learning_rate": 1.913959323973119e-05, |
| "loss": 0.8337, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1.9616347381864623, |
| "grad_norm": 4.268481254577637, |
| "learning_rate": 1.863390866935344e-05, |
| "loss": 0.8111, |
| "step": 4800 |
| }, |
| { |
| "epoch": 2.0024521072796935, |
| "grad_norm": 3.2703258991241455, |
| "learning_rate": 1.8123796741022803e-05, |
| "loss": 0.8167, |
| "step": 4900 |
| }, |
| { |
| "epoch": 2.043320561941252, |
| "grad_norm": 3.365675926208496, |
| "learning_rate": 1.7609878947547232e-05, |
| "loss": 0.7072, |
| "step": 5000 |
| }, |
| { |
| "epoch": 2.043320561941252, |
| "eval_accuracy": 0.031, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1938, |
| 62 |
| ] |
| ], |
| "eval_f1_macro": 0.030067895247332686, |
| "eval_humor_english_accuracy": 0.02920443101711984, |
| "eval_humor_english_avg_loss": 0.26709887150925354, |
| "eval_humor_english_f1": 0.02837573385518591, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.01460221550855992, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.03277060575968222, |
| "eval_humor_hindi_avg_loss": 0.1286927389570682, |
| "eval_humor_hindi_f1": 0.03173076923076923, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.01638530287984111, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.031, |
| "eval_humor_overall_avg_loss": 0.19798794310908208, |
| "eval_humor_overall_f1": 0.030067895247332686, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.0155, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6472519040107727, |
| "eval_perplexity": 1.9102839848633415, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.0155, |
| "eval_runtime": 1320.0922, |
| "eval_samples_per_second": 1.515, |
| "eval_steps_per_second": 0.758, |
| "eval_temporal_english_avg_loss": 0.17779056256882214, |
| "eval_temporal_hindi_avg_loss": 0.3266544083817946, |
| "eval_temporal_overall_avg_loss": 0.2523379425825881, |
| "step": 5000 |
| }, |
| { |
| "epoch": 2.0841890166028096, |
| "grad_norm": 4.201941013336182, |
| "learning_rate": 1.709278141859511e-05, |
| "loss": 0.7132, |
| "step": 5100 |
| }, |
| { |
| "epoch": 2.125057471264368, |
| "grad_norm": 4.3990888595581055, |
| "learning_rate": 1.6573134157852686e-05, |
| "loss": 0.6732, |
| "step": 5200 |
| }, |
| { |
| "epoch": 2.165925925925926, |
| "grad_norm": 4.231266021728516, |
| "learning_rate": 1.6051570275461666e-05, |
| "loss": 0.6784, |
| "step": 5300 |
| }, |
| { |
| "epoch": 2.206794380587484, |
| "grad_norm": 3.714615821838379, |
| "learning_rate": 1.552872521667198e-05, |
| "loss": 0.6785, |
| "step": 5400 |
| }, |
| { |
| "epoch": 2.247662835249042, |
| "grad_norm": 3.3602356910705566, |
| "learning_rate": 1.5005235987649652e-05, |
| "loss": 0.6877, |
| "step": 5500 |
| }, |
| { |
| "epoch": 2.247662835249042, |
| "eval_accuracy": 0.005, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1990, |
| 10 |
| ] |
| ], |
| "eval_f1_macro": 0.004975124378109453, |
| "eval_humor_english_accuracy": 0.004028197381671702, |
| "eval_humor_english_avg_loss": 0.2567401882446778, |
| "eval_humor_english_f1": 0.004012036108324975, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.002014098690835851, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.005958291956305859, |
| "eval_humor_hindi_avg_loss": 0.12068944840764959, |
| "eval_humor_hindi_f1": 0.005923000987166831, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.0029791459781529296, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.005, |
| "eval_humor_overall_avg_loss": 0.18884940875979583, |
| "eval_humor_overall_f1": 0.004975124378109453, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.0025, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6365848183631897, |
| "eval_perplexity": 1.8900151414329684, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.0025, |
| "eval_runtime": 1323.2498, |
| "eval_samples_per_second": 1.511, |
| "eval_steps_per_second": 0.756, |
| "eval_temporal_english_avg_loss": 0.16544690548644253, |
| "eval_temporal_hindi_avg_loss": 0.3068489971076771, |
| "eval_temporal_overall_avg_loss": 0.23624011447018664, |
| "step": 5500 |
| }, |
| { |
| "epoch": 2.2885312899106003, |
| "grad_norm": 4.01371431350708, |
| "learning_rate": 1.4481740379382916e-05, |
| "loss": 0.688, |
| "step": 5600 |
| }, |
| { |
| "epoch": 2.329399744572158, |
| "grad_norm": 4.780395984649658, |
| "learning_rate": 1.3958876190632131e-05, |
| "loss": 0.6912, |
| "step": 5700 |
| }, |
| { |
| "epoch": 2.3702681992337165, |
| "grad_norm": 4.804982662200928, |
| "learning_rate": 1.3437280450870225e-05, |
| "loss": 0.6524, |
| "step": 5800 |
| }, |
| { |
| "epoch": 2.4111366538952748, |
| "grad_norm": 4.125489711761475, |
| "learning_rate": 1.2917588644160447e-05, |
| "loss": 0.659, |
| "step": 5900 |
| }, |
| { |
| "epoch": 2.4520051085568326, |
| "grad_norm": 3.899745225906372, |
| "learning_rate": 1.2400433934916879e-05, |
| "loss": 0.6537, |
| "step": 6000 |
| }, |
| { |
| "epoch": 2.4520051085568326, |
| "eval_accuracy": 0.044, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1912, |
| 88 |
| ] |
| ], |
| "eval_f1_macro": 0.0421455938697318, |
| "eval_humor_english_accuracy": 0.05236656596173213, |
| "eval_humor_english_avg_loss": 0.24773680959614575, |
| "eval_humor_english_f1": 0.049760765550239235, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.026183282980866064, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.03574975173783516, |
| "eval_humor_hindi_avg_loss": 0.1137336753887836, |
| "eval_humor_hindi_f1": 0.03451581975071908, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.01787487586891758, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.044, |
| "eval_humor_overall_avg_loss": 0.18087796053705443, |
| "eval_humor_overall_f1": 0.0421455938697318, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.022, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6272982358932495, |
| "eval_perplexity": 1.8725445511265306, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.022, |
| "eval_runtime": 1361.0844, |
| "eval_samples_per_second": 1.469, |
| "eval_steps_per_second": 0.735, |
| "eval_temporal_english_avg_loss": 0.15524752458450397, |
| "eval_temporal_hindi_avg_loss": 0.29028965216244074, |
| "eval_temporal_overall_avg_loss": 0.22285711327104782, |
| "step": 6000 |
| }, |
| { |
| "epoch": 2.492873563218391, |
| "grad_norm": 3.0799901485443115, |
| "learning_rate": 1.1886446396491155e-05, |
| "loss": 0.6469, |
| "step": 6100 |
| }, |
| { |
| "epoch": 2.533742017879949, |
| "grad_norm": 4.32665491104126, |
| "learning_rate": 1.1376252243525146e-05, |
| "loss": 0.6748, |
| "step": 6200 |
| }, |
| { |
| "epoch": 2.574610472541507, |
| "grad_norm": 4.455280780792236, |
| "learning_rate": 1.0870473069004852e-05, |
| "loss": 0.654, |
| "step": 6300 |
| }, |
| { |
| "epoch": 2.615478927203065, |
| "grad_norm": 3.6130049228668213, |
| "learning_rate": 1.0369725086945106e-05, |
| "loss": 0.6495, |
| "step": 6400 |
| }, |
| { |
| "epoch": 2.6563473818646233, |
| "grad_norm": 4.499478816986084, |
| "learning_rate": 9.874618381627751e-06, |
| "loss": 0.6443, |
| "step": 6500 |
| }, |
| { |
| "epoch": 2.6563473818646233, |
| "eval_accuracy": 0.117, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1766, |
| 234 |
| ] |
| ], |
| "eval_f1_macro": 0.10474485228290063, |
| "eval_humor_english_accuracy": 0.0986908358509567, |
| "eval_humor_english_avg_loss": 0.24009977730751, |
| "eval_humor_english_f1": 0.08982584784601283, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.04934541792547835, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.13505461767626614, |
| "eval_humor_hindi_avg_loss": 0.10769947945037281, |
| "eval_humor_hindi_f1": 0.1189851268591426, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.06752730883813307, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.117, |
| "eval_humor_overall_avg_loss": 0.17395776182075373, |
| "eval_humor_overall_f1": 0.10474485228290063, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.0585, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6204598546028137, |
| "eval_perplexity": 1.8597831080769023, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.0585, |
| "eval_runtime": 1361.8863, |
| "eval_samples_per_second": 1.469, |
| "eval_steps_per_second": 0.734, |
| "eval_temporal_english_avg_loss": 0.1464234281813041, |
| "eval_temporal_hindi_avg_loss": 0.27630730096865136, |
| "eval_temporal_overall_avg_loss": 0.21141392534496303, |
| "step": 6500 |
| }, |
| { |
| "epoch": 2.6972158365261816, |
| "grad_norm": 3.4488909244537354, |
| "learning_rate": 9.385756164307868e-06, |
| "loss": 0.6831, |
| "step": 6600 |
| }, |
| { |
| "epoch": 2.7380842911877394, |
| "grad_norm": 4.000279903411865, |
| "learning_rate": 8.903734038293804e-06, |
| "loss": 0.6437, |
| "step": 6700 |
| }, |
| { |
| "epoch": 2.7789527458492973, |
| "grad_norm": 4.4706220626831055, |
| "learning_rate": 8.429139273296304e-06, |
| "loss": 0.6284, |
| "step": 6800 |
| }, |
| { |
| "epoch": 2.8198212005108556, |
| "grad_norm": 3.7017948627471924, |
| "learning_rate": 7.96255008993078e-06, |
| "loss": 0.6357, |
| "step": 6900 |
| }, |
| { |
| "epoch": 2.860689655172414, |
| "grad_norm": 4.603327751159668, |
| "learning_rate": 7.504534955244629e-06, |
| "loss": 0.6207, |
| "step": 7000 |
| }, |
| { |
| "epoch": 2.860689655172414, |
| "eval_accuracy": 0.0635, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1873, |
| 127 |
| ] |
| ], |
| "eval_f1_macro": 0.059708509637987774, |
| "eval_humor_english_accuracy": 0.05639476334340383, |
| "eval_humor_english_avg_loss": 0.2334255897185822, |
| "eval_humor_english_f1": 0.05338417540514776, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.028197381671701913, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.070506454816286, |
| "eval_humor_hindi_avg_loss": 0.10243915956694087, |
| "eval_humor_hindi_f1": 0.06586270871985157, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.035253227408143, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0635, |
| "eval_humor_overall_avg_loss": 0.1679845057619467, |
| "eval_humor_overall_f1": 0.059708509637987774, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.03175, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6139324903488159, |
| "eval_perplexity": 1.847683260757382, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.03175, |
| "eval_runtime": 1315.241, |
| "eval_samples_per_second": 1.521, |
| "eval_steps_per_second": 0.76, |
| "eval_temporal_english_avg_loss": 0.138804246816629, |
| "eval_temporal_hindi_avg_loss": 0.2640184406992659, |
| "eval_temporal_overall_avg_loss": 0.20143722001935127, |
| "step": 7000 |
| }, |
| { |
| "epoch": 2.9015581098339718, |
| "grad_norm": 3.4093639850616455, |
| "learning_rate": 7.055651890127625e-06, |
| "loss": 0.631, |
| "step": 7100 |
| }, |
| { |
| "epoch": 2.94242656449553, |
| "grad_norm": 3.2841861248016357, |
| "learning_rate": 6.616447789449488e-06, |
| "loss": 0.6271, |
| "step": 7200 |
| }, |
| { |
| "epoch": 2.9832950191570884, |
| "grad_norm": 2.856577157974243, |
| "learning_rate": 6.187457755752684e-06, |
| "loss": 0.6457, |
| "step": 7300 |
| }, |
| { |
| "epoch": 3.024112388250319, |
| "grad_norm": 4.275922775268555, |
| "learning_rate": 5.7692044473124276e-06, |
| "loss": 0.5875, |
| "step": 7400 |
| }, |
| { |
| "epoch": 3.0649808429118774, |
| "grad_norm": 4.011717319488525, |
| "learning_rate": 5.362197441358068e-06, |
| "loss": 0.5295, |
| "step": 7500 |
| }, |
| { |
| "epoch": 3.0649808429118774, |
| "eval_accuracy": 0.1215, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1757, |
| 243 |
| ] |
| ], |
| "eval_f1_macro": 0.10833704859563086, |
| "eval_humor_english_accuracy": 0.12185297079556898, |
| "eval_humor_english_avg_loss": 0.22710613908185748, |
| "eval_humor_english_f1": 0.10861759425493717, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.06092648539778449, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.1211519364448858, |
| "eval_humor_hindi_avg_loss": 0.09781679747076281, |
| "eval_humor_hindi_f1": 0.10806023029229407, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.0605759682224429, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.1215, |
| "eval_humor_overall_avg_loss": 0.16250835471025796, |
| "eval_humor_overall_f1": 0.10833704859563086, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.06075, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6209594011306763, |
| "eval_perplexity": 1.8607123988418854, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.06075, |
| "eval_runtime": 1319.9889, |
| "eval_samples_per_second": 1.515, |
| "eval_steps_per_second": 0.758, |
| "eval_temporal_english_avg_loss": 0.13217057422883527, |
| "eval_temporal_hindi_avg_loss": 0.25253982628289234, |
| "eval_temporal_overall_avg_loss": 0.19244430216349742, |
| "step": 7500 |
| }, |
| { |
| "epoch": 3.1058492975734353, |
| "grad_norm": 3.4078967571258545, |
| "learning_rate": 4.966932613231739e-06, |
| "loss": 0.5529, |
| "step": 7600 |
| }, |
| { |
| "epoch": 3.1467177522349936, |
| "grad_norm": 3.757827043533325, |
| "learning_rate": 4.583891532240636e-06, |
| "loss": 0.5268, |
| "step": 7700 |
| }, |
| { |
| "epoch": 3.187586206896552, |
| "grad_norm": 3.6647462844848633, |
| "learning_rate": 4.213540874938976e-06, |
| "loss": 0.5249, |
| "step": 7800 |
| }, |
| { |
| "epoch": 3.22845466155811, |
| "grad_norm": 3.4517507553100586, |
| "learning_rate": 3.856331856554558e-06, |
| "loss": 0.5423, |
| "step": 7900 |
| }, |
| { |
| "epoch": 3.269323116219668, |
| "grad_norm": 3.9157795906066895, |
| "learning_rate": 3.5126996812524377e-06, |
| "loss": 0.5311, |
| "step": 8000 |
| }, |
| { |
| "epoch": 3.269323116219668, |
| "eval_accuracy": 0.1045, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1791, |
| 209 |
| ] |
| ], |
| "eval_f1_macro": 0.0946129470348574, |
| "eval_humor_english_accuracy": 0.10372608257804633, |
| "eval_humor_english_avg_loss": 0.2202950545789041, |
| "eval_humor_english_f1": 0.09397810218978102, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.051863041289023165, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.10526315789473684, |
| "eval_humor_hindi_avg_loss": 0.09350429856929683, |
| "eval_humor_hindi_f1": 0.09523809523809523, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.05263157894736842, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.1045, |
| "eval_humor_overall_avg_loss": 0.15700095058811175, |
| "eval_humor_overall_f1": 0.0946129470348574, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.05225, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.621131181716919, |
| "eval_perplexity": 1.861031944826067, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.05225, |
| "eval_runtime": 1318.5114, |
| "eval_samples_per_second": 1.517, |
| "eval_steps_per_second": 0.758, |
| "eval_temporal_english_avg_loss": 0.1261224628154746, |
| "eval_temporal_hindi_avg_loss": 0.2422587038177751, |
| "eval_temporal_overall_avg_loss": 0.18421073652827494, |
| "step": 8000 |
| }, |
| { |
| "epoch": 3.310191570881226, |
| "grad_norm": 3.881890058517456, |
| "learning_rate": 3.183063011905744e-06, |
| "loss": 0.5225, |
| "step": 8100 |
| }, |
| { |
| "epoch": 3.3510600255427843, |
| "grad_norm": 4.492738723754883, |
| "learning_rate": 2.8678234600194414e-06, |
| "loss": 0.5358, |
| "step": 8200 |
| }, |
| { |
| "epoch": 3.391928480204342, |
| "grad_norm": 3.7062699794769287, |
| "learning_rate": 2.5673650964285718e-06, |
| "loss": 0.531, |
| "step": 8300 |
| }, |
| { |
| "epoch": 3.4327969348659004, |
| "grad_norm": 4.436982154846191, |
| "learning_rate": 2.2820539833671085e-06, |
| "loss": 0.5394, |
| "step": 8400 |
| }, |
| { |
| "epoch": 3.4736653895274587, |
| "grad_norm": 3.07208251953125, |
| "learning_rate": 2.0122377284775136e-06, |
| "loss": 0.5136, |
| "step": 8500 |
| }, |
| { |
| "epoch": 3.4736653895274587, |
| "eval_accuracy": 0.097, |
| "eval_confusion_matrix": [ |
| [ |
| 0, |
| 0 |
| ], |
| [ |
| 1806, |
| 194 |
| ] |
| ], |
| "eval_f1_macro": 0.08842297174111212, |
| "eval_humor_english_accuracy": 0.09667673716012085, |
| "eval_humor_english_avg_loss": 0.21453529215166584, |
| "eval_humor_english_f1": 0.0881542699724518, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.04833836858006042, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.09731876861966236, |
| "eval_humor_hindi_avg_loss": 0.08955117555728288, |
| "eval_humor_hindi_f1": 0.08868778280542987, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.04865938430983118, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.097, |
| "eval_humor_overall_avg_loss": 0.15204923117292898, |
| "eval_humor_overall_f1": 0.08842297174111212, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.0485, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.618234395980835, |
| "eval_perplexity": 1.8556488288019473, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.0485, |
| "eval_runtime": 1365.0675, |
| "eval_samples_per_second": 1.465, |
| "eval_steps_per_second": 0.733, |
| "eval_temporal_english_avg_loss": 0.12083532439477838, |
| "eval_temporal_hindi_avg_loss": 0.23295580905530663, |
| "eval_temporal_overall_avg_loss": 0.17692150853084088, |
| "step": 8500 |
| }, |
| { |
| "epoch": 3.5145338441890166, |
| "grad_norm": 4.052462577819824, |
| "learning_rate": 2.1103441209615935e-05, |
| "loss": 0.5403, |
| "step": 8600 |
| }, |
| { |
| "epoch": 3.5554022988505745, |
| "grad_norm": 4.443219184875488, |
| "learning_rate": 2.0863376714130244e-05, |
| "loss": 0.5796, |
| "step": 8700 |
| }, |
| { |
| "epoch": 3.5962707535121328, |
| "grad_norm": 4.265786170959473, |
| "learning_rate": 2.0621526177297613e-05, |
| "loss": 0.5538, |
| "step": 8800 |
| }, |
| { |
| "epoch": 3.637139208173691, |
| "grad_norm": 3.9799318313598633, |
| "learning_rate": 2.037796326913444e-05, |
| "loss": 0.5683, |
| "step": 8900 |
| }, |
| { |
| "epoch": 3.678007662835249, |
| "grad_norm": 5.084273815155029, |
| "learning_rate": 2.0132762181261985e-05, |
| "loss": 0.5716, |
| "step": 9000 |
| }, |
| { |
| "epoch": 3.678007662835249, |
| "eval_accuracy": 0.0995, |
| "eval_f1_macro": 0.0904956798544793, |
| "eval_humor_english_accuracy": 0.05740181268882175, |
| "eval_humor_english_avg_loss": 0.117691616244732, |
| "eval_humor_english_f1": 0.054285714285714284, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.028700906344410877, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.141012909632572, |
| "eval_humor_hindi_avg_loss": 0.031129881199932288, |
| "eval_humor_hindi_f1": 0.12358572671888599, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.070506454816286, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0995, |
| "eval_humor_overall_avg_loss": 0.07511869748413463, |
| "eval_humor_overall_f1": 0.0904956798544793, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.04975, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6332020163536072, |
| "eval_perplexity": 1.8836323104356882, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.04975, |
| "eval_runtime": 1319.1571, |
| "eval_samples_per_second": 1.516, |
| "eval_steps_per_second": 0.758, |
| "eval_temporal_english_avg_loss": 0.03940154156217748, |
| "eval_temporal_hindi_avg_loss": 0.09286631036147072, |
| "eval_temporal_overall_avg_loss": 0.0664199314402633, |
| "step": 9000 |
| }, |
| { |
| "epoch": 3.7188761174968072, |
| "grad_norm": 4.418786525726318, |
| "learning_rate": 1.9885997604306933e-05, |
| "loss": 0.5554, |
| "step": 9100 |
| }, |
| { |
| "epoch": 3.759744572158365, |
| "grad_norm": 3.633528470993042, |
| "learning_rate": 1.9637744705149887e-05, |
| "loss": 0.5581, |
| "step": 9200 |
| }, |
| { |
| "epoch": 3.8006130268199234, |
| "grad_norm": 4.484210014343262, |
| "learning_rate": 1.938807910402881e-05, |
| "loss": 0.5744, |
| "step": 9300 |
| }, |
| { |
| "epoch": 3.8414814814814813, |
| "grad_norm": 3.976949453353882, |
| "learning_rate": 1.913707685150437e-05, |
| "loss": 0.5831, |
| "step": 9400 |
| }, |
| { |
| "epoch": 3.8823499361430396, |
| "grad_norm": 4.015331745147705, |
| "learning_rate": 1.88848144052942e-05, |
| "loss": 0.5658, |
| "step": 9500 |
| }, |
| { |
| "epoch": 3.8823499361430396, |
| "eval_accuracy": 0.0505, |
| "eval_f1_macro": 0.04807234650166587, |
| "eval_humor_english_accuracy": 0.04934541792547835, |
| "eval_humor_english_avg_loss": 0.11883179473594759, |
| "eval_humor_english_f1": 0.04702495201535509, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.024672708962739175, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.05163853028798411, |
| "eval_humor_hindi_avg_loss": 0.030766832375286615, |
| "eval_humor_hindi_f1": 0.049102927289896126, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.025819265143992055, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.0505, |
| "eval_humor_overall_avg_loss": 0.07535547554798036, |
| "eval_humor_overall_f1": 0.04807234650166587, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.02525, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6302640438079834, |
| "eval_perplexity": 1.8781064395523288, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.02525, |
| "eval_runtime": 1320.57, |
| "eval_samples_per_second": 1.514, |
| "eval_steps_per_second": 0.757, |
| "eval_temporal_english_avg_loss": 0.03935090346366541, |
| "eval_temporal_hindi_avg_loss": 0.0936906353848563, |
| "eval_temporal_overall_avg_loss": 0.06659941324144447, |
| "step": 9500 |
| }, |
| { |
| "epoch": 3.923218390804598, |
| "grad_norm": 3.6775312423706055, |
| "learning_rate": 1.8631368606983134e-05, |
| "loss": 0.5588, |
| "step": 9600 |
| }, |
| { |
| "epoch": 3.9640868454661558, |
| "grad_norm": 3.3025190830230713, |
| "learning_rate": 1.8376816658616527e-05, |
| "loss": 0.5638, |
| "step": 9700 |
| }, |
| { |
| "epoch": 4.005312899106003, |
| "grad_norm": 5.129138469696045, |
| "learning_rate": 1.8121236099183743e-05, |
| "loss": 0.5504, |
| "step": 9800 |
| }, |
| { |
| "epoch": 4.046181353767561, |
| "grad_norm": 4.347407817840576, |
| "learning_rate": 1.7864704780999054e-05, |
| "loss": 0.5361, |
| "step": 9900 |
| }, |
| { |
| "epoch": 4.087049808429119, |
| "grad_norm": 5.224924564361572, |
| "learning_rate": 1.7607300845987055e-05, |
| "loss": 0.5143, |
| "step": 10000 |
| }, |
| { |
| "epoch": 4.087049808429119, |
| "eval_accuracy": 0.132, |
| "eval_f1_macro": 0.1166077738515901, |
| "eval_humor_english_accuracy": 0.1379657603222558, |
| "eval_humor_english_avg_loss": 0.117569191649303, |
| "eval_humor_english_f1": 0.12123893805309735, |
| "eval_humor_english_precision": 0.5, |
| "eval_humor_english_recall": 0.0689828801611279, |
| "eval_humor_english_samples": 993, |
| "eval_humor_hindi_accuracy": 0.12611717974180736, |
| "eval_humor_hindi_avg_loss": 0.030592078067631787, |
| "eval_humor_hindi_f1": 0.11199294532627865, |
| "eval_humor_hindi_precision": 0.5, |
| "eval_humor_hindi_recall": 0.06305858987090368, |
| "eval_humor_hindi_samples": 1007, |
| "eval_humor_overall_accuracy": 0.132, |
| "eval_humor_overall_avg_loss": 0.07420280479598583, |
| "eval_humor_overall_f1": 0.1166077738515901, |
| "eval_humor_overall_precision": 0.5, |
| "eval_humor_overall_recall": 0.066, |
| "eval_humor_overall_samples": 2000, |
| "eval_loss": 0.6409682035446167, |
| "eval_perplexity": 1.8983179897511162, |
| "eval_precision_macro": 0.5, |
| "eval_recall_macro": 0.066, |
| "eval_runtime": 1308.5558, |
| "eval_samples_per_second": 1.528, |
| "eval_steps_per_second": 0.764, |
| "eval_temporal_english_avg_loss": 0.039093032986785355, |
| "eval_temporal_hindi_avg_loss": 0.09288990841599172, |
| "eval_temporal_overall_avg_loss": 0.0659479490747604, |
| "step": 10000 |
| } |
| ], |
| "logging_steps": 100, |
| "max_steps": 20000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 9, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 3, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 3 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 3.2097341741727744e+16, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
