{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 0.96, | |
"eval_steps": 500, | |
"global_step": 12, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.8, | |
"learning_rate": 0.0005, | |
"loss": 0.7096, | |
"step": 10 | |
}, | |
{ | |
"epoch": 0.96, | |
"step": 12, | |
"total_flos": 1241604054319104.0, | |
"train_loss": 0.6211734215418497, | |
"train_runtime": 23.6128, | |
"train_samples_per_second": 8.47, | |
"train_steps_per_second": 0.508 | |
} | |
], | |
"logging_steps": 10, | |
"max_steps": 12, | |
"num_input_tokens_seen": 0, | |
"num_train_epochs": 1, | |
"save_steps": 100, | |
"total_flos": 1241604054319104.0, | |
"train_batch_size": 4, | |
"trial_name": null, | |
"trial_params": null | |
} | |