| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 410, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.24390243902439024, |
| "grad_norm": 23.37420654296875, |
| "learning_rate": 1.5e-06, |
| "loss": 3.2045, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.4878048780487805, |
| "grad_norm": 11.02387809753418, |
| "learning_rate": 3.5e-06, |
| "loss": 2.0765, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.7317073170731707, |
| "grad_norm": 4.315557479858398, |
| "learning_rate": 5.500000000000001e-06, |
| "loss": 1.1171, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.975609756097561, |
| "grad_norm": 3.7738795280456543, |
| "learning_rate": 7.500000000000001e-06, |
| "loss": 0.6946, |
| "step": 80 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_avg_cer": 0.38252571063317314, |
| "eval_avg_wer": 0.8156832493942604, |
| "eval_gilaki_cer": 0.37511361177234365, |
| "eval_gilaki_wer": 0.9556110684089162, |
| "eval_hawrami_cer": 0.11549464078794902, |
| "eval_hawrami_wer": 0.5245801526717557, |
| "eval_laki_kurdish_cer": 0.2733040507564666, |
| "eval_laki_kurdish_wer": 0.7938650306748466, |
| "eval_loss": 0.766200602054596, |
| "eval_mazanderani_cer": 0.32235210768685796, |
| "eval_mazanderani_wer": 0.8120224146714213, |
| "eval_runtime": 281.2263, |
| "eval_samples_per_second": 11.585, |
| "eval_southern_kurdish_cer": 0.2192075497883561, |
| "eval_southern_kurdish_wer": 0.6800766283524904, |
| "eval_steps_per_second": 0.046, |
| "eval_talysh_cer": 0.3333333333333333, |
| "eval_talysh_wer": 0.9166666666666666, |
| "eval_zazaki_cer": 1.0388746803069053, |
| "eval_zazaki_wer": 1.0269607843137254, |
| "step": 82 |
| }, |
| { |
| "epoch": 1.2195121951219512, |
| "grad_norm": 2.4077467918395996, |
| "learning_rate": 9.5e-06, |
| "loss": 0.5333, |
| "step": 100 |
| }, |
| { |
| "epoch": 1.4634146341463414, |
| "grad_norm": 2.5292022228240967, |
| "learning_rate": 9.516129032258065e-06, |
| "loss": 0.4533, |
| "step": 120 |
| }, |
| { |
| "epoch": 1.7073170731707317, |
| "grad_norm": 2.2705225944519043, |
| "learning_rate": 8.870967741935484e-06, |
| "loss": 0.4231, |
| "step": 140 |
| }, |
| { |
| "epoch": 1.951219512195122, |
| "grad_norm": 2.4976840019226074, |
| "learning_rate": 8.225806451612904e-06, |
| "loss": 0.3864, |
| "step": 160 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_avg_cer": 0.24537730055466347, |
| "eval_avg_wer": 0.686525530003659, |
| "eval_gilaki_cer": 0.34542306859987015, |
| "eval_gilaki_wer": 0.9579169869331283, |
| "eval_hawrami_cer": 0.07734646581691773, |
| "eval_hawrami_wer": 0.380559796437659, |
| "eval_laki_kurdish_cer": 0.15470961444607126, |
| "eval_laki_kurdish_wer": 0.5644171779141104, |
| "eval_loss": 0.5171846747398376, |
| "eval_mazanderani_cer": 0.18455543747786043, |
| "eval_mazanderani_wer": 0.609271523178808, |
| "eval_runtime": 283.0635, |
| "eval_samples_per_second": 11.51, |
| "eval_southern_kurdish_cer": 0.1742765942682673, |
| "eval_southern_kurdish_wer": 0.5484151863462209, |
| "eval_steps_per_second": 0.046, |
| "eval_talysh_cer": 0.5, |
| "eval_talysh_wer": 1.0, |
| "eval_zazaki_cer": 0.2813299232736573, |
| "eval_zazaki_wer": 0.7450980392156863, |
| "step": 164 |
| }, |
| { |
| "epoch": 2.1951219512195124, |
| "grad_norm": 2.0281219482421875, |
| "learning_rate": 7.580645161290323e-06, |
| "loss": 0.3523, |
| "step": 180 |
| }, |
| { |
| "epoch": 2.4390243902439024, |
| "grad_norm": 2.4135963916778564, |
| "learning_rate": 6.935483870967743e-06, |
| "loss": 0.3342, |
| "step": 200 |
| }, |
| { |
| "epoch": 2.682926829268293, |
| "grad_norm": 2.4130258560180664, |
| "learning_rate": 6.290322580645162e-06, |
| "loss": 0.3199, |
| "step": 220 |
| }, |
| { |
| "epoch": 2.926829268292683, |
| "grad_norm": 1.6657346487045288, |
| "learning_rate": 5.645161290322582e-06, |
| "loss": 0.3234, |
| "step": 240 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_avg_cer": 0.2326738623090893, |
| "eval_avg_wer": 0.6506993324980173, |
| "eval_gilaki_cer": 0.35135252109932913, |
| "eval_gilaki_wer": 0.9632974634896234, |
| "eval_hawrami_cer": 0.07220451911935111, |
| "eval_hawrami_wer": 0.3532824427480916, |
| "eval_laki_kurdish_cer": 0.13624532292175046, |
| "eval_laki_kurdish_wer": 0.5231083844580777, |
| "eval_loss": 0.4686383008956909, |
| "eval_mazanderani_cer": 0.16519069547762427, |
| "eval_mazanderani_wer": 0.5476311767702496, |
| "eval_runtime": 284.8477, |
| "eval_samples_per_second": 11.438, |
| "eval_southern_kurdish_cer": 0.16740684199569772, |
| "eval_southern_kurdish_wer": 0.4935562521769418, |
| "eval_steps_per_second": 0.046, |
| "eval_talysh_cer": 0.5, |
| "eval_talysh_wer": 1.0, |
| "eval_zazaki_cer": 0.23631713554987213, |
| "eval_zazaki_wer": 0.6740196078431373, |
| "step": 246 |
| }, |
| { |
| "epoch": 3.1707317073170733, |
| "grad_norm": 1.7249436378479004, |
| "learning_rate": 5e-06, |
| "loss": 0.299, |
| "step": 260 |
| }, |
| { |
| "epoch": 3.4146341463414633, |
| "grad_norm": 1.6258125305175781, |
| "learning_rate": 4.35483870967742e-06, |
| "loss": 0.2806, |
| "step": 280 |
| }, |
| { |
| "epoch": 3.658536585365854, |
| "grad_norm": 1.9203271865844727, |
| "learning_rate": 3.7096774193548392e-06, |
| "loss": 0.2753, |
| "step": 300 |
| }, |
| { |
| "epoch": 3.902439024390244, |
| "grad_norm": 1.7945189476013184, |
| "learning_rate": 3.0645161290322584e-06, |
| "loss": 0.2786, |
| "step": 320 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_avg_cer": 0.2249207301185327, |
| "eval_avg_wer": 0.632319681761422, |
| "eval_gilaki_cer": 0.3179398398615018, |
| "eval_gilaki_wer": 0.9083397386625672, |
| "eval_hawrami_cer": 0.07321842410196987, |
| "eval_hawrami_wer": 0.34849872773536894, |
| "eval_laki_kurdish_cer": 0.1333984057263706, |
| "eval_laki_kurdish_wer": 0.5047034764826176, |
| "eval_loss": 0.4527292251586914, |
| "eval_mazanderani_cer": 0.1596410438068249, |
| "eval_mazanderani_wer": 0.5277636271013755, |
| "eval_runtime": 287.001, |
| "eval_samples_per_second": 11.352, |
| "eval_southern_kurdish_cer": 0.1769481645964888, |
| "eval_southern_kurdish_wer": 0.4947753396029258, |
| "eval_steps_per_second": 0.045, |
| "eval_talysh_cer": 0.5, |
| "eval_talysh_wer": 1.0, |
| "eval_zazaki_cer": 0.2132992327365729, |
| "eval_zazaki_wer": 0.6421568627450981, |
| "step": 328 |
| }, |
| { |
| "epoch": 4.146341463414634, |
| "grad_norm": 1.4399691820144653, |
| "learning_rate": 2.4193548387096776e-06, |
| "loss": 0.2609, |
| "step": 340 |
| }, |
| { |
| "epoch": 4.390243902439025, |
| "grad_norm": 1.4866726398468018, |
| "learning_rate": 1.774193548387097e-06, |
| "loss": 0.2484, |
| "step": 360 |
| }, |
| { |
| "epoch": 4.634146341463414, |
| "grad_norm": 1.4835667610168457, |
| "learning_rate": 1.1290322580645162e-06, |
| "loss": 0.2567, |
| "step": 380 |
| }, |
| { |
| "epoch": 4.878048780487805, |
| "grad_norm": 1.4296433925628662, |
| "learning_rate": 4.838709677419355e-07, |
| "loss": 0.2586, |
| "step": 400 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_avg_cer": 0.2212028617810692, |
| "eval_avg_wer": 0.6295649617906507, |
| "eval_gilaki_cer": 0.3175070331097165, |
| "eval_gilaki_wer": 0.9175634127594159, |
| "eval_hawrami_cer": 0.07082850521436848, |
| "eval_hawrami_wer": 0.345852417302799, |
| "eval_laki_kurdish_cer": 0.12469497315763788, |
| "eval_laki_kurdish_wer": 0.49038854805725973, |
| "eval_loss": 0.45085304975509644, |
| "eval_mazanderani_cer": 0.16153028692879914, |
| "eval_mazanderani_wer": 0.5277636271013755, |
| "eval_runtime": 287.0262, |
| "eval_samples_per_second": 11.351, |
| "eval_southern_kurdish_cer": 0.17181319825133579, |
| "eval_southern_kurdish_wer": 0.4930337861372344, |
| "eval_steps_per_second": 0.045, |
| "eval_talysh_cer": 0.5, |
| "eval_talysh_wer": 1.0, |
| "eval_zazaki_cer": 0.2020460358056266, |
| "eval_zazaki_wer": 0.6323529411764706, |
| "step": 410 |
| }, |
| { |
| "epoch": 5.0, |
| "step": 410, |
| "total_flos": 3.028992382327587e+19, |
| "train_loss": 0.6099171120946, |
| "train_runtime": 8367.268, |
| "train_samples_per_second": 12.473, |
| "train_steps_per_second": 0.049 |
| } |
| ], |
| "logging_steps": 20, |
| "max_steps": 410, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "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": 3.028992382327587e+19, |
| "train_batch_size": 128, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
