|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"eval_steps": 1452, |
|
"global_step": 4356, |
|
"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": 2.2721, |
|
"step": 291 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 3e-05, |
|
"loss": 2.0539, |
|
"step": 582 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 3e-05, |
|
"loss": 2.0079, |
|
"step": 873 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3e-05, |
|
"loss": 1.9686, |
|
"step": 1164 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.668149394347241, |
|
"eval_loss": 1.5662202835083008, |
|
"eval_runtime": 137.7572, |
|
"eval_samples_per_second": 36.296, |
|
"eval_steps_per_second": 2.272, |
|
"step": 1452 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_exact_match": 9.04, |
|
"eval_f1": 13.087046780017364, |
|
"eval_qa_bleu": 5.297766450359973, |
|
"eval_qa_exact_match": 0.085, |
|
"eval_recite_bleu": 14.716037077738477, |
|
"eval_recite_exact_match": 0.0, |
|
"step": 1452 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3e-05, |
|
"loss": 1.9076, |
|
"step": 1455 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 3e-05, |
|
"loss": 1.5211, |
|
"step": 1746 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 3e-05, |
|
"loss": 1.5039, |
|
"step": 2037 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 3e-05, |
|
"loss": 1.4834, |
|
"step": 2328 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 3e-05, |
|
"loss": 1.4604, |
|
"step": 2619 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.6914514131897712, |
|
"eval_loss": 1.206884741783142, |
|
"eval_runtime": 139.7305, |
|
"eval_samples_per_second": 35.783, |
|
"eval_steps_per_second": 2.24, |
|
"step": 2904 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_exact_match": 11.64, |
|
"eval_f1": 16.917593370997537, |
|
"eval_qa_bleu": 7.627957176996819, |
|
"eval_qa_exact_match": 0.1106, |
|
"eval_recite_bleu": 18.356743056908567, |
|
"eval_recite_exact_match": 0.0, |
|
"step": 2904 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 3e-05, |
|
"loss": 1.4368, |
|
"step": 2910 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 3e-05, |
|
"loss": 1.0705, |
|
"step": 3201 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 3e-05, |
|
"loss": 1.0685, |
|
"step": 3492 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 3e-05, |
|
"loss": 1.0554, |
|
"step": 3783 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 3e-05, |
|
"loss": 1.0522, |
|
"step": 4074 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.7148896366083446, |
|
"eval_loss": 0.9045500159263611, |
|
"eval_runtime": 141.6455, |
|
"eval_samples_per_second": 35.299, |
|
"eval_steps_per_second": 2.21, |
|
"step": 4356 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_exact_match": 12.84, |
|
"eval_f1": 18.555356874635468, |
|
"eval_qa_bleu": 8.395879921702353, |
|
"eval_qa_exact_match": 0.1188, |
|
"eval_recite_bleu": 20.824221134285214, |
|
"eval_recite_exact_match": 0.0, |
|
"step": 4356 |
|
} |
|
], |
|
"logging_steps": 291, |
|
"max_steps": 29040, |
|
"num_train_epochs": 20, |
|
"save_steps": 500, |
|
"total_flos": 1.99049246312448e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|