|
{ |
|
"best_metric": 1.3715944290161133, |
|
"best_model_checkpoint": "flan-t5-base-samsum/checkpoint-3684", |
|
"epoch": 2.0, |
|
"global_step": 3684, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 4.728555917480999e-05, |
|
"loss": 1.4811, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.457111834961998e-05, |
|
"loss": 1.4545, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 4.185667752442997e-05, |
|
"loss": 1.4379, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_gen_len": 17.11965811965812, |
|
"eval_loss": 1.3805454969406128, |
|
"eval_rouge1": 47.1438, |
|
"eval_rouge2": 23.6153, |
|
"eval_rougeL": 39.6699, |
|
"eval_rougeLsum": 43.5505, |
|
"eval_runtime": 126.1042, |
|
"eval_samples_per_second": 6.495, |
|
"eval_steps_per_second": 0.817, |
|
"step": 1842 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 3.914223669923996e-05, |
|
"loss": 1.4094, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 3.642779587404995e-05, |
|
"loss": 1.3318, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 3.3713355048859935e-05, |
|
"loss": 1.3396, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 3.099891422366993e-05, |
|
"loss": 1.3559, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_gen_len": 17.39072039072039, |
|
"eval_loss": 1.3715944290161133, |
|
"eval_rouge1": 47.2663, |
|
"eval_rouge2": 23.5327, |
|
"eval_rougeL": 39.6491, |
|
"eval_rougeLsum": 43.3169, |
|
"eval_runtime": 125.9833, |
|
"eval_samples_per_second": 6.501, |
|
"eval_steps_per_second": 0.818, |
|
"step": 3684 |
|
} |
|
], |
|
"max_steps": 9210, |
|
"num_train_epochs": 5, |
|
"total_flos": 2.017569063252787e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|