{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 1.5723270440251573, | |
"eval_steps": 500, | |
"global_step": 500, | |
"is_hyper_param_search": true, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 1.0, | |
"eval_accuracy": 0.5861290322580646, | |
"eval_loss": 0.1950901746749878, | |
"eval_runtime": 1.47, | |
"eval_samples_per_second": 2108.896, | |
"eval_steps_per_second": 44.219, | |
"step": 318 | |
}, | |
{ | |
"epoch": 1.57, | |
"learning_rate": 1.685534591194969e-05, | |
"loss": 0.315, | |
"step": 500 | |
} | |
], | |
"logging_steps": 500, | |
"max_steps": 3180, | |
"num_input_tokens_seen": 0, | |
"num_train_epochs": 10, | |
"save_steps": 500, | |
"total_flos": 271860022520268.0, | |
"train_batch_size": 48, | |
"trial_name": null, | |
"trial_params": { | |
"alpha": 0.17426119812070984, | |
"num_train_epochs": 10, | |
"temperature": 13 | |
} | |
} | |