| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 7721, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "step": 1103, | |
| "train_eval_accuracy": 0.7887755102040817, | |
| "train_eval_f1": 0.7499765233661925, | |
| "train_eval_loss": 0.6592453122138977, | |
| "train_eval_precision": 0.7927385476759929, | |
| "train_eval_recall": 0.7853664471213178, | |
| "train_loss": 0.6592453718185425, | |
| "train_runtime": 296.3413, | |
| "train_samples_per_second": 29.763, | |
| "train_steps_per_second": 3.722 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7685185185185185, | |
| "eval_f1": 0.7339765316819828, | |
| "eval_loss": 0.7040889859199524, | |
| "eval_precision": 0.7863260957104617, | |
| "eval_recall": 0.7767501150267258, | |
| "eval_runtime": 126.8841, | |
| "eval_samples_per_second": 29.791, | |
| "eval_steps_per_second": 3.728, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 2206, | |
| "train_eval_accuracy": 0.8997732426303855, | |
| "train_eval_f1": 0.8934703863536795, | |
| "train_eval_loss": 0.3266890048980713, | |
| "train_eval_precision": 0.9101662801242012, | |
| "train_eval_recall": 0.8995948840878806, | |
| "train_loss": 0.3266890048980713, | |
| "train_runtime": 296.0976, | |
| "train_samples_per_second": 29.787, | |
| "train_steps_per_second": 3.725 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8748677248677249, | |
| "eval_f1": 0.8663052621269062, | |
| "eval_loss": 0.41356033086776733, | |
| "eval_precision": 0.8855161789508988, | |
| "eval_recall": 0.8760987779104309, | |
| "eval_runtime": 127.0363, | |
| "eval_samples_per_second": 29.755, | |
| "eval_steps_per_second": 3.723, | |
| "step": 2206 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 3309, | |
| "train_eval_accuracy": 0.9326530612244898, | |
| "train_eval_f1": 0.9319607388705672, | |
| "train_eval_loss": 0.25093671679496765, | |
| "train_eval_precision": 0.934719996865382, | |
| "train_eval_recall": 0.9321842163347528, | |
| "train_loss": 0.25093671679496765, | |
| "train_runtime": 296.2824, | |
| "train_samples_per_second": 29.769, | |
| "train_steps_per_second": 3.723 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9042328042328043, | |
| "eval_f1": 0.9039529271468891, | |
| "eval_loss": 0.39229774475097656, | |
| "eval_precision": 0.9080753320256745, | |
| "eval_recall": 0.9062393448267169, | |
| "eval_runtime": 126.7759, | |
| "eval_samples_per_second": 29.816, | |
| "eval_steps_per_second": 3.731, | |
| "step": 3309 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 4412, | |
| "train_eval_accuracy": 0.9439909297052154, | |
| "train_eval_f1": 0.9436709080792405, | |
| "train_eval_loss": 0.2164522260427475, | |
| "train_eval_precision": 0.9456422075056565, | |
| "train_eval_recall": 0.9435377727151975, | |
| "train_loss": 0.2164521962404251, | |
| "train_runtime": 296.0068, | |
| "train_samples_per_second": 29.797, | |
| "train_steps_per_second": 3.726 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9050264550264551, | |
| "eval_f1": 0.9051255458970469, | |
| "eval_loss": 0.42613422870635986, | |
| "eval_precision": 0.9081998981335793, | |
| "eval_recall": 0.9074460110563182, | |
| "eval_runtime": 126.5384, | |
| "eval_samples_per_second": 29.872, | |
| "eval_steps_per_second": 3.738, | |
| "step": 4412 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 5515, | |
| "train_eval_accuracy": 0.9621315192743765, | |
| "train_eval_f1": 0.9621058175036092, | |
| "train_eval_loss": 0.1465868204832077, | |
| "train_eval_precision": 0.963177617453184, | |
| "train_eval_recall": 0.9619998226212546, | |
| "train_loss": 0.1465868204832077, | |
| "train_runtime": 295.9817, | |
| "train_samples_per_second": 29.799, | |
| "train_steps_per_second": 3.727 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.903968253968254, | |
| "eval_f1": 0.9038650845517326, | |
| "eval_loss": 0.4323442280292511, | |
| "eval_precision": 0.9057544963051973, | |
| "eval_recall": 0.9056543685348158, | |
| "eval_runtime": 126.9132, | |
| "eval_samples_per_second": 29.784, | |
| "eval_steps_per_second": 3.727, | |
| "step": 5515 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 6618, | |
| "train_eval_accuracy": 0.9688208616780045, | |
| "train_eval_f1": 0.9686094121707401, | |
| "train_eval_loss": 0.12130556255578995, | |
| "train_eval_precision": 0.9692092193201128, | |
| "train_eval_recall": 0.968624092631756, | |
| "train_loss": 0.12130556255578995, | |
| "train_runtime": 295.4832, | |
| "train_samples_per_second": 29.849, | |
| "train_steps_per_second": 3.733 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9074074074074074, | |
| "eval_f1": 0.9074718452978288, | |
| "eval_loss": 0.4809330999851227, | |
| "eval_precision": 0.9099023812111475, | |
| "eval_recall": 0.9094968555241478, | |
| "eval_runtime": 126.5478, | |
| "eval_samples_per_second": 29.87, | |
| "eval_steps_per_second": 3.738, | |
| "step": 6618 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 7721, | |
| "train_eval_accuracy": 0.9763038548752835, | |
| "train_eval_f1": 0.9762453458735343, | |
| "train_eval_loss": 0.09657437354326248, | |
| "train_eval_precision": 0.9765774895508033, | |
| "train_eval_recall": 0.9763401145817671, | |
| "train_loss": 0.09657437354326248, | |
| "train_runtime": 296.9603, | |
| "train_samples_per_second": 29.701, | |
| "train_steps_per_second": 3.714 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9082010582010582, | |
| "eval_f1": 0.9079087150920282, | |
| "eval_loss": 0.4926854968070984, | |
| "eval_precision": 0.9100340807600719, | |
| "eval_recall": 0.9087758232905191, | |
| "eval_runtime": 127.7522, | |
| "eval_samples_per_second": 29.589, | |
| "eval_steps_per_second": 3.702, | |
| "step": 7721 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 7721, | |
| "total_flos": 1.62503106619392e+16, | |
| "train_loss": 0.5412681450144088, | |
| "train_runtime": 9360.1511, | |
| "train_samples_per_second": 14.134, | |
| "train_steps_per_second": 1.768 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9082010582010582, | |
| "eval_f1": 0.9079087150920282, | |
| "eval_loss": 0.4926854968070984, | |
| "eval_precision": 0.9100340807600719, | |
| "eval_recall": 0.9087758232905191, | |
| "eval_runtime": 127.3673, | |
| "eval_samples_per_second": 29.678, | |
| "eval_steps_per_second": 3.714, | |
| "step": 7721 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 7721, | |
| "train_en_eval_accuracy": 0.9763038548752835, | |
| "train_en_eval_f1": 0.9762453458735343, | |
| "train_en_eval_loss": 0.09657437354326248, | |
| "train_en_eval_precision": 0.9765774895508033, | |
| "train_en_eval_recall": 0.9763401145817671, | |
| "train_en_loss": 0.09657437354326248, | |
| "train_en_runtime": 296.2956, | |
| "train_en_samples_per_second": 29.768, | |
| "train_en_steps_per_second": 3.723 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 7721, | |
| "test_en_eval_accuracy": 0.9082010582010582, | |
| "test_en_eval_f1": 0.9079087150920282, | |
| "test_en_eval_loss": 0.4926854372024536, | |
| "test_en_eval_precision": 0.9100340807600719, | |
| "test_en_eval_recall": 0.9087758232905191, | |
| "test_en_loss": 0.4926854968070984, | |
| "test_en_runtime": 126.7992, | |
| "test_en_samples_per_second": 29.811, | |
| "test_en_steps_per_second": 3.73 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 16545, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "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": 1.62503106619392e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |