|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 6.25, |
|
"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": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.0541, |
|
"eval_samples_per_second": 2940.887, |
|
"eval_steps_per_second": 16.127, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.0855, |
|
"eval_samples_per_second": 2855.842, |
|
"eval_steps_per_second": 15.661, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.1066, |
|
"eval_samples_per_second": 2801.344, |
|
"eval_steps_per_second": 15.362, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_accuracy": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.0705, |
|
"eval_samples_per_second": 2895.912, |
|
"eval_steps_per_second": 15.881, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_accuracy": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.0601, |
|
"eval_samples_per_second": 2924.316, |
|
"eval_steps_per_second": 16.037, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_accuracy": 0.0064516129032258064, |
|
"eval_loss": NaN, |
|
"eval_runtime": 1.0689, |
|
"eval_samples_per_second": 2900.115, |
|
"eval_steps_per_second": 15.904, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 6.25, |
|
"learning_rate": 4.3750000000000005e-06, |
|
"loss": 0.0, |
|
"step": 500 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 640, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 8, |
|
"save_steps": 500, |
|
"total_flos": 591309336775836.0, |
|
"train_batch_size": 192, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.2985555267876274, |
|
"num_train_epochs": 8, |
|
"temperature": 13 |
|
} |
|
} |
|
|