{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 4.25, | |
"eval_steps": 500, | |
"global_step": 5, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.25, | |
"learning_rate": 0.0001, | |
"loss": 2.2049, | |
"step": 1 | |
}, | |
{ | |
"epoch": 1.25, | |
"learning_rate": 0.0002, | |
"loss": 2.2049, | |
"step": 2 | |
}, | |
{ | |
"epoch": 2.25, | |
"learning_rate": 0.00018888888888888888, | |
"loss": 2.1056, | |
"step": 3 | |
}, | |
{ | |
"epoch": 3.25, | |
"learning_rate": 0.00017777777777777779, | |
"loss": 1.9492, | |
"step": 4 | |
}, | |
{ | |
"epoch": 4.25, | |
"learning_rate": 0.0001666666666666667, | |
"loss": 1.8287, | |
"step": 5 | |
} | |
], | |
"logging_steps": 1, | |
"max_steps": 20, | |
"num_train_epochs": 5, | |
"save_steps": 500, | |
"total_flos": 103763977175040.0, | |
"trial_name": null, | |
"trial_params": null | |
} | |