{ "best_metric": 1.0080007314682007, "best_model_checkpoint": "finetune_bert-base-uncased/checkpoint-5476", "epoch": 2.0, "eval_steps": 500, "global_step": 5476, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.18, "learning_rate": 1.8787436084733382e-05, "loss": 2.324, "step": 500 }, { "epoch": 0.37, "learning_rate": 1.7570002434867303e-05, "loss": 1.3702, "step": 1000 }, { "epoch": 0.55, "learning_rate": 1.635500365230095e-05, "loss": 1.2429, "step": 1500 }, { "epoch": 0.73, "learning_rate": 1.513757000243487e-05, "loss": 1.1599, "step": 2000 }, { "epoch": 0.91, "learning_rate": 1.3920136352568787e-05, "loss": 1.1298, "step": 2500 }, { "epoch": 1.0, "eval_loss": 1.0416048765182495, "eval_runtime": 80.4387, "eval_samples_per_second": 131.404, "eval_steps_per_second": 1.032, "step": 2738 }, { "epoch": 1.1, "learning_rate": 1.2702702702702702e-05, "loss": 0.9945, "step": 3000 }, { "epoch": 1.28, "learning_rate": 1.1485269052836621e-05, "loss": 0.888, "step": 3500 }, { "epoch": 1.46, "learning_rate": 1.0267835402970538e-05, "loss": 0.8653, "step": 4000 }, { "epoch": 1.64, "learning_rate": 9.052836620404189e-06, "loss": 0.8768, "step": 4500 }, { "epoch": 1.83, "learning_rate": 7.835402970538106e-06, "loss": 0.8736, "step": 5000 }, { "epoch": 2.0, "eval_loss": 1.0080007314682007, "eval_runtime": 79.7471, "eval_samples_per_second": 132.544, "eval_steps_per_second": 1.041, "step": 5476 } ], "logging_steps": 500, "max_steps": 8214, "num_train_epochs": 3, "save_steps": 500, "total_flos": 3.4334001889975296e+16, "trial_name": null, "trial_params": null }