{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 2.9865067466266866, | |
"eval_steps": 500, | |
"global_step": 498, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.8, | |
"learning_rate": 8.191964285714287e-05, | |
"loss": 0.7441, | |
"step": 133 | |
}, | |
{ | |
"epoch": 1.6, | |
"learning_rate": 5.223214285714286e-05, | |
"loss": 0.3712, | |
"step": 266 | |
}, | |
{ | |
"epoch": 2.39, | |
"learning_rate": 2.2544642857142857e-05, | |
"loss": 0.3396, | |
"step": 399 | |
} | |
], | |
"logging_steps": 133, | |
"max_steps": 498, | |
"num_input_tokens_seen": 0, | |
"num_train_epochs": 3, | |
"save_steps": 500, | |
"total_flos": 8.732487119441756e+17, | |
"train_batch_size": 10, | |
"trial_name": null, | |
"trial_params": null | |
} | |