{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 1.0, | |
"eval_steps": 248, | |
"global_step": 248, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.2, | |
"learning_rate": 3e-05, | |
"loss": 2.8428, | |
"step": 50 | |
}, | |
{ | |
"epoch": 0.4, | |
"learning_rate": 3e-05, | |
"loss": 2.1589, | |
"step": 100 | |
}, | |
{ | |
"epoch": 0.6, | |
"learning_rate": 3e-05, | |
"loss": 2.0527, | |
"step": 150 | |
}, | |
{ | |
"epoch": 0.81, | |
"learning_rate": 3e-05, | |
"loss": 1.9823, | |
"step": 200 | |
}, | |
{ | |
"epoch": 1.0, | |
"eval_accuracy": 0.6398457059679767, | |
"eval_loss": 1.6149228811264038, | |
"eval_runtime": 17.4504, | |
"eval_samples_per_second": 28.653, | |
"eval_steps_per_second": 1.834, | |
"step": 248 | |
}, | |
{ | |
"epoch": 1.0, | |
"eval_exact_match": 9.0, | |
"eval_f1": 13.45666666666666, | |
"eval_qa_bleu": 7.063432441196736, | |
"eval_qa_exact_match": 0.09, | |
"eval_recite_bleu": 15.344658855358656, | |
"eval_recite_exact_match": 0.0, | |
"step": 248 | |
} | |
], | |
"logging_steps": 50, | |
"max_steps": 4960, | |
"num_train_epochs": 20, | |
"save_steps": 500, | |
"total_flos": 1.15393402827264e+16, | |
"trial_name": null, | |
"trial_params": null | |
} | |