| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 1352, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2603550295857989e-05, | |
| "loss": 0.3604, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8746428489685059, | |
| "eval_loss": 0.3286072015762329, | |
| "eval_runtime": 25.372, | |
| "eval_samples_per_second": 551.79, | |
| "eval_steps_per_second": 8.632, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.207100591715976e-06, | |
| "loss": 0.2243, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8787857294082642, | |
| "eval_loss": 0.32322901487350464, | |
| "eval_runtime": 18.0969, | |
| "eval_samples_per_second": 773.613, | |
| "eval_steps_per_second": 12.102, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1352, | |
| "total_flos": 5691618349547520.0, | |
| "train_loss": 0.268575707836264, | |
| "train_runtime": 1091.6903, | |
| "train_samples_per_second": 633.526, | |
| "train_steps_per_second": 1.238 | |
| } | |
| ], | |
| "max_steps": 1352, | |
| "num_train_epochs": 2, | |
| "total_flos": 5691618349547520.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |