|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.1851851851851851, |
|
"eval_steps": 100, |
|
"global_step": 800, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 0.0, |
|
"loss": 8.5238, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"eval_runtime": 788.2037, |
|
"eval_samples_per_second": 1.903, |
|
"eval_steps_per_second": 0.239, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 0.0, |
|
"loss": 8.4542, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"eval_runtime": 788.181, |
|
"eval_samples_per_second": 1.903, |
|
"eval_steps_per_second": 0.239, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 0.0, |
|
"loss": 8.4751, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"eval_runtime": 787.7273, |
|
"eval_samples_per_second": 1.904, |
|
"eval_steps_per_second": 0.239, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 0.0, |
|
"loss": 8.6043, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"eval_runtime": 787.7907, |
|
"eval_samples_per_second": 1.904, |
|
"eval_steps_per_second": 0.239, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 0.0, |
|
"loss": 8.2845, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"eval_runtime": 788.246, |
|
"eval_samples_per_second": 1.903, |
|
"eval_steps_per_second": 0.239, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 0.0, |
|
"loss": 8.7866, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"eval_runtime": 787.5816, |
|
"eval_samples_per_second": 1.905, |
|
"eval_steps_per_second": 0.239, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 0.0, |
|
"loss": 8.4979, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"eval_runtime": 788.4314, |
|
"eval_samples_per_second": 1.903, |
|
"eval_steps_per_second": 0.238, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 0.0, |
|
"loss": 8.4584, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"eval_runtime": 787.875, |
|
"eval_samples_per_second": 1.904, |
|
"eval_steps_per_second": 0.239, |
|
"step": 800 |
|
} |
|
], |
|
"logging_steps": 100, |
|
"max_steps": 2025, |
|
"num_train_epochs": 3, |
|
"save_steps": 100, |
|
"total_flos": 1.66592319062016e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|