|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 10.0, |
|
"global_step": 3980, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_bleu": 0.014, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.8135625720024109, |
|
"eval_runtime": 38.7924, |
|
"eval_samples_per_second": 20.494, |
|
"eval_steps_per_second": 1.289, |
|
"step": 398 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.3718592964824124e-05, |
|
"loss": 1.246, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_bleu": 0.1109, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6999887228012085, |
|
"eval_runtime": 38.7142, |
|
"eval_samples_per_second": 20.535, |
|
"eval_steps_per_second": 1.292, |
|
"step": 796 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 3.7437185929648245e-05, |
|
"loss": 0.8026, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_bleu": 0.1504, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6451420783996582, |
|
"eval_runtime": 38.8554, |
|
"eval_samples_per_second": 20.46, |
|
"eval_steps_per_second": 1.287, |
|
"step": 1194 |
|
}, |
|
{ |
|
"epoch": 3.77, |
|
"learning_rate": 3.1155778894472366e-05, |
|
"loss": 0.7162, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_bleu": 0.1224, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6195945143699646, |
|
"eval_runtime": 38.7838, |
|
"eval_samples_per_second": 20.498, |
|
"eval_steps_per_second": 1.289, |
|
"step": 1592 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_bleu": 0.1486, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.596439778804779, |
|
"eval_runtime": 38.8878, |
|
"eval_samples_per_second": 20.443, |
|
"eval_steps_per_second": 1.286, |
|
"step": 1990 |
|
}, |
|
{ |
|
"epoch": 5.03, |
|
"learning_rate": 2.4874371859296484e-05, |
|
"loss": 0.6704, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_bleu": 0.2053, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.5814200639724731, |
|
"eval_runtime": 38.8468, |
|
"eval_samples_per_second": 20.465, |
|
"eval_steps_per_second": 1.287, |
|
"step": 2388 |
|
}, |
|
{ |
|
"epoch": 6.28, |
|
"learning_rate": 1.8592964824120602e-05, |
|
"loss": 0.6399, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"eval_bleu": 0.2048, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.5735769271850586, |
|
"eval_runtime": 38.9201, |
|
"eval_samples_per_second": 20.426, |
|
"eval_steps_per_second": 1.285, |
|
"step": 2786 |
|
}, |
|
{ |
|
"epoch": 7.54, |
|
"learning_rate": 1.2311557788944725e-05, |
|
"loss": 0.6201, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_bleu": 0.1875, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.568035900592804, |
|
"eval_runtime": 39.4161, |
|
"eval_samples_per_second": 20.169, |
|
"eval_steps_per_second": 1.269, |
|
"step": 3184 |
|
}, |
|
{ |
|
"epoch": 8.79, |
|
"learning_rate": 6.030150753768844e-06, |
|
"loss": 0.6074, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"eval_bleu": 0.2047, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.5626117587089539, |
|
"eval_runtime": 38.8108, |
|
"eval_samples_per_second": 20.484, |
|
"eval_steps_per_second": 1.288, |
|
"step": 3582 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_bleu": 0.2074, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.5602744221687317, |
|
"eval_runtime": 39.1814, |
|
"eval_samples_per_second": 20.29, |
|
"eval_steps_per_second": 1.276, |
|
"step": 3980 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"step": 3980, |
|
"total_flos": 2.26168179720192e+16, |
|
"train_loss": 0.7384387069012053, |
|
"train_runtime": 2101.0033, |
|
"train_samples_per_second": 30.271, |
|
"train_steps_per_second": 1.894 |
|
} |
|
], |
|
"max_steps": 3980, |
|
"num_train_epochs": 10, |
|
"total_flos": 2.26168179720192e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|