{ "best_global_step": 800, "best_metric": 0.6842105263157895, "best_model_checkpoint": "/home/pricie/trusca/ner/models_ocr/detect_perplex_param/model_lambda0.2_th0.9/checkpoint-800", "epoch": 160.0, "eval_steps": 200, "global_step": 800, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 10.0, "grad_norm": 0.18833546340465546, "learning_rate": 4.902e-05, "loss": 0.1822, "step": 50 }, { "epoch": 20.0, "grad_norm": 0.13318823277950287, "learning_rate": 4.8020000000000004e-05, "loss": 0.0069, "step": 100 }, { "epoch": 30.0, "grad_norm": 0.11782731115818024, "learning_rate": 4.702e-05, "loss": 0.0022, "step": 150 }, { "epoch": 40.0, "grad_norm": 0.03338160738348961, "learning_rate": 4.602e-05, "loss": 0.001, "step": 200 }, { "epoch": 40.0, "eval_LOC": { "f1": 0.0, "number": 1, "precision": 0.0, "recall": 0.0 }, "eval_PERS": { "f1": 0.6796116504854368, "number": 56, "precision": 0.7446808510638298, "recall": 0.625 }, "eval_loss": 0.0714135617017746, "eval_overall_accuracy": 0.9904264577893821, "eval_overall_f1": 0.6603773584905661, "eval_overall_precision": 0.7142857142857143, "eval_overall_recall": 0.6140350877192983, "eval_runtime": 0.1069, "eval_samples_per_second": 140.301, "eval_steps_per_second": 9.353, "step": 200 }, { "epoch": 50.0, "grad_norm": 0.004488145466893911, "learning_rate": 4.502e-05, "loss": 0.0006, "step": 250 }, { "epoch": 60.0, "grad_norm": 0.0032296597491949797, "learning_rate": 4.402e-05, "loss": 0.0005, "step": 300 }, { "epoch": 70.0, "grad_norm": 0.06219864264130592, "learning_rate": 4.3020000000000005e-05, "loss": 0.0005, "step": 350 }, { "epoch": 80.0, "grad_norm": 0.037636227905750275, "learning_rate": 4.202e-05, "loss": 0.0004, "step": 400 }, { "epoch": 80.0, "eval_LOC": { "f1": 0.0, "number": 1, "precision": 0.0, "recall": 0.0 }, "eval_PERS": { "f1": 0.7307692307692307, "number": 56, "precision": 0.7916666666666666, "recall": 0.6785714285714286 }, "eval_loss": 0.07451974600553513, "eval_overall_accuracy": 0.9915868871482448, "eval_overall_f1": 0.7238095238095237, "eval_overall_precision": 0.7916666666666666, "eval_overall_recall": 0.6666666666666666, "eval_runtime": 0.0932, "eval_samples_per_second": 160.973, "eval_steps_per_second": 10.732, "step": 400 }, { "epoch": 90.0, "grad_norm": 0.01358821615576744, "learning_rate": 4.1020000000000004e-05, "loss": 0.0005, "step": 450 }, { "epoch": 100.0, "grad_norm": 0.0060653723776340485, "learning_rate": 4.002e-05, "loss": 0.0004, "step": 500 }, { "epoch": 110.0, "grad_norm": 0.015429244376718998, "learning_rate": 3.902e-05, "loss": 0.0003, "step": 550 }, { "epoch": 120.0, "grad_norm": 0.0023402925580739975, "learning_rate": 3.802e-05, "loss": 0.0003, "step": 600 }, { "epoch": 120.0, "eval_LOC": { "f1": 0.0, "number": 1, "precision": 0.0, "recall": 0.0 }, "eval_PERS": { "f1": 0.6923076923076924, "number": 56, "precision": 0.75, "recall": 0.6428571428571429 }, "eval_loss": 0.07376705855131149, "eval_overall_accuracy": 0.9907165651290978, "eval_overall_f1": 0.6857142857142857, "eval_overall_precision": 0.75, "eval_overall_recall": 0.631578947368421, "eval_runtime": 0.0895, "eval_samples_per_second": 167.643, "eval_steps_per_second": 11.176, "step": 600 }, { "epoch": 130.0, "grad_norm": 0.09169554710388184, "learning_rate": 3.702e-05, "loss": 0.0003, "step": 650 }, { "epoch": 140.0, "grad_norm": 0.011466496624052525, "learning_rate": 3.6020000000000004e-05, "loss": 0.0004, "step": 700 }, { "epoch": 150.0, "grad_norm": 0.0007753117242828012, "learning_rate": 3.502e-05, "loss": 0.0003, "step": 750 }, { "epoch": 160.0, "grad_norm": 0.0963754802942276, "learning_rate": 3.402e-05, "loss": 0.0003, "step": 800 }, { "epoch": 160.0, "eval_LOC": { "f1": 0.0, "number": 1, "precision": 0.0, "recall": 0.0 }, "eval_PERS": { "f1": 0.7090909090909091, "number": 56, "precision": 0.7222222222222222, "recall": 0.6964285714285714 }, "eval_loss": 0.07110230624675751, "eval_overall_accuracy": 0.9912967798085291, "eval_overall_f1": 0.7027027027027027, "eval_overall_precision": 0.7222222222222222, "eval_overall_recall": 0.6842105263157895, "eval_runtime": 0.0914, "eval_samples_per_second": 164.058, "eval_steps_per_second": 10.937, "step": 800 } ], "logging_steps": 50, "max_steps": 2500, "num_input_tokens_seen": 0, "num_train_epochs": 500, "save_steps": 200, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 1.1915455279104e+16, "train_batch_size": 64, "trial_name": null, "trial_params": null }