{ "best_metric": 0.8578811369509044, "best_model_checkpoint": "distilhubert-finetuned-mixed-data/checkpoint-242", "epoch": 12.989690721649485, "eval_steps": 500, "global_step": 315, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.9896907216494846, "eval_accuracy": 0.6356589147286822, "eval_f1": 0.5920563016978556, "eval_loss": 0.980873703956604, "eval_precision": 0.5920482291587493, "eval_recall": 0.6356589147286822, "eval_runtime": 1.6668, "eval_samples_per_second": 232.175, "eval_steps_per_second": 29.397, "step": 24 }, { "epoch": 1.9793814432989691, "eval_accuracy": 0.7157622739018088, "eval_f1": 0.6905410405322238, "eval_loss": 0.7444477081298828, "eval_precision": 0.6992377248989063, "eval_recall": 0.7157622739018088, "eval_runtime": 1.6941, "eval_samples_per_second": 228.443, "eval_steps_per_second": 28.924, "step": 48 }, { "epoch": 2.9690721649484537, "eval_accuracy": 0.7493540051679587, "eval_f1": 0.744898505571463, "eval_loss": 0.6171658039093018, "eval_precision": 0.7437592422989429, "eval_recall": 0.7493540051679587, "eval_runtime": 1.6943, "eval_samples_per_second": 228.408, "eval_steps_per_second": 28.92, "step": 72 }, { "epoch": 4.0, "eval_accuracy": 0.7984496124031008, "eval_f1": 0.7873621619744228, "eval_loss": 0.5430988073348999, "eval_precision": 0.79180344284319, "eval_recall": 0.7984496124031008, "eval_runtime": 1.7027, "eval_samples_per_second": 227.289, "eval_steps_per_second": 28.778, "step": 97 }, { "epoch": 4.989690721649485, "eval_accuracy": 0.8010335917312662, "eval_f1": 0.7974946178390901, "eval_loss": 0.5268548130989075, "eval_precision": 0.8005965453214461, "eval_recall": 0.8010335917312662, "eval_runtime": 1.6829, "eval_samples_per_second": 229.957, "eval_steps_per_second": 29.116, "step": 121 }, { "epoch": 5.979381443298969, "eval_accuracy": 0.7493540051679587, "eval_f1": 0.7551329793037762, "eval_loss": 0.5810549259185791, "eval_precision": 0.7802262423287315, "eval_recall": 0.7493540051679587, "eval_runtime": 1.6999, "eval_samples_per_second": 227.658, "eval_steps_per_second": 28.825, "step": 145 }, { "epoch": 6.969072164948454, "eval_accuracy": 0.8397932816537468, "eval_f1": 0.8354840008265724, "eval_loss": 0.44080850481987, "eval_precision": 0.8365717854569443, "eval_recall": 0.8397932816537468, "eval_runtime": 1.6785, "eval_samples_per_second": 230.56, "eval_steps_per_second": 29.192, "step": 169 }, { "epoch": 8.0, "eval_accuracy": 0.8191214470284238, "eval_f1": 0.8167837576120855, "eval_loss": 0.46871018409729004, "eval_precision": 0.8187643627479353, "eval_recall": 0.8191214470284238, "eval_runtime": 1.6719, "eval_samples_per_second": 231.477, "eval_steps_per_second": 29.308, "step": 194 }, { "epoch": 8.989690721649485, "eval_accuracy": 0.8475452196382429, "eval_f1": 0.8473710740005564, "eval_loss": 0.4363822937011719, "eval_precision": 0.8483238707679635, "eval_recall": 0.8475452196382429, "eval_runtime": 1.6838, "eval_samples_per_second": 229.835, "eval_steps_per_second": 29.101, "step": 218 }, { "epoch": 9.97938144329897, "eval_accuracy": 0.8578811369509044, "eval_f1": 0.8567532661685897, "eval_loss": 0.42906680703163147, "eval_precision": 0.8560504853170988, "eval_recall": 0.8578811369509044, "eval_runtime": 1.6875, "eval_samples_per_second": 229.333, "eval_steps_per_second": 29.037, "step": 242 }, { "epoch": 10.969072164948454, "eval_accuracy": 0.8501291989664083, "eval_f1": 0.8527674409653727, "eval_loss": 0.46987947821617126, "eval_precision": 0.858217159873796, "eval_recall": 0.8501291989664083, "eval_runtime": 1.6795, "eval_samples_per_second": 230.425, "eval_steps_per_second": 29.175, "step": 266 }, { "epoch": 12.0, "eval_accuracy": 0.8449612403100775, "eval_f1": 0.8479775479649662, "eval_loss": 0.48616188764572144, "eval_precision": 0.8536273463382107, "eval_recall": 0.8449612403100775, "eval_runtime": 1.6948, "eval_samples_per_second": 228.339, "eval_steps_per_second": 28.911, "step": 291 }, { "epoch": 12.989690721649485, "eval_accuracy": 0.8475452196382429, "eval_f1": 0.8483528512434715, "eval_loss": 0.4765341877937317, "eval_precision": 0.8497431189750078, "eval_recall": 0.8475452196382429, "eval_runtime": 1.688, "eval_samples_per_second": 229.264, "eval_steps_per_second": 29.028, "step": 315 } ], "logging_steps": 500, "max_steps": 360, "num_input_tokens_seen": 0, "num_train_epochs": 15, "save_steps": 500, "stateful_callbacks": { "EarlyStoppingCallback": { "args": { "early_stopping_patience": 3, "early_stopping_threshold": 0.0 }, "attributes": { "early_stopping_patience_counter": 0 } }, "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 4.56778404576e+16, "train_batch_size": 8, "trial_name": null, "trial_params": null }