{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 1.0, | |
"eval_steps": 2179, | |
"global_step": 2179, | |
"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": 2.2464, | |
"step": 436 | |
}, | |
{ | |
"epoch": 0.4, | |
"learning_rate": 3e-05, | |
"loss": 2.0544, | |
"step": 872 | |
}, | |
{ | |
"epoch": 0.6, | |
"learning_rate": 3e-05, | |
"loss": 1.9968, | |
"step": 1308 | |
}, | |
{ | |
"epoch": 0.8, | |
"learning_rate": 3e-05, | |
"loss": 1.9537, | |
"step": 1744 | |
}, | |
{ | |
"epoch": 1.0, | |
"eval_accuracy": 0.6693020521036767, | |
"eval_loss": 1.5623117685317993, | |
"eval_runtime": 220.2436, | |
"eval_samples_per_second": 33.622, | |
"eval_steps_per_second": 2.102, | |
"step": 2179 | |
}, | |
{ | |
"epoch": 1.0, | |
"eval_exact_match": 10.24983119513842, | |
"eval_f1": 15.257097040945366, | |
"eval_qa_bleu": 5.8382901471491575, | |
"eval_qa_exact_match": 0.09736664415935178, | |
"eval_recite_bleu": 17.0930417063702, | |
"eval_recite_exact_match": 0.0, | |
"step": 2179 | |
} | |
], | |
"logging_steps": 436, | |
"max_steps": 43580, | |
"num_train_epochs": 20, | |
"save_steps": 500, | |
"total_flos": 1.00774010674176e+17, | |
"trial_name": null, | |
"trial_params": null | |
} | |