{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 1.995249406175772, | |
"eval_steps": 500, | |
"global_step": 630, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.8, | |
"learning_rate": 0.00013333333333333334, | |
"loss": 1.5622, | |
"step": 252 | |
}, | |
{ | |
"epoch": 1.6, | |
"learning_rate": 4.4444444444444447e-05, | |
"loss": 1.2174, | |
"step": 504 | |
} | |
], | |
"logging_steps": 252, | |
"max_steps": 630, | |
"num_input_tokens_seen": 0, | |
"num_train_epochs": 2, | |
"save_steps": 500, | |
"total_flos": 514265587200000.0, | |
"train_batch_size": 4, | |
"trial_name": null, | |
"trial_params": null | |
} | |