|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 10.0, |
|
"global_step": 3970, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_bleu": 0.0481, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 1.043367624282837, |
|
"eval_runtime": 38.6251, |
|
"eval_samples_per_second": 20.531, |
|
"eval_steps_per_second": 1.294, |
|
"step": 397 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 4.370277078085643e-05, |
|
"loss": 1.5127, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_bleu": 0.0772, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.9015085101127625, |
|
"eval_runtime": 38.4043, |
|
"eval_samples_per_second": 20.649, |
|
"eval_steps_per_second": 1.302, |
|
"step": 794 |
|
}, |
|
{ |
|
"epoch": 2.52, |
|
"learning_rate": 3.7405541561712845e-05, |
|
"loss": 1.0539, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_bleu": 0.0991, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.8207009434700012, |
|
"eval_runtime": 38.7237, |
|
"eval_samples_per_second": 20.478, |
|
"eval_steps_per_second": 1.291, |
|
"step": 1191 |
|
}, |
|
{ |
|
"epoch": 3.78, |
|
"learning_rate": 3.1108312342569276e-05, |
|
"loss": 0.932, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_bleu": 0.0925, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.7715731859207153, |
|
"eval_runtime": 38.873, |
|
"eval_samples_per_second": 20.4, |
|
"eval_steps_per_second": 1.286, |
|
"step": 1588 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_bleu": 0.1292, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.7394161820411682, |
|
"eval_runtime": 38.8257, |
|
"eval_samples_per_second": 20.425, |
|
"eval_steps_per_second": 1.288, |
|
"step": 1985 |
|
}, |
|
{ |
|
"epoch": 5.04, |
|
"learning_rate": 2.4811083123425694e-05, |
|
"loss": 0.8578, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_bleu": 0.1101, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.7167691588401794, |
|
"eval_runtime": 38.8738, |
|
"eval_samples_per_second": 20.399, |
|
"eval_steps_per_second": 1.286, |
|
"step": 2382 |
|
}, |
|
{ |
|
"epoch": 6.3, |
|
"learning_rate": 1.8513853904282116e-05, |
|
"loss": 0.8102, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"eval_bleu": 0.1465, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6981162428855896, |
|
"eval_runtime": 38.9078, |
|
"eval_samples_per_second": 20.382, |
|
"eval_steps_per_second": 1.285, |
|
"step": 2779 |
|
}, |
|
{ |
|
"epoch": 7.56, |
|
"learning_rate": 1.2216624685138539e-05, |
|
"loss": 0.7837, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_bleu": 0.1574, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6897701621055603, |
|
"eval_runtime": 38.6725, |
|
"eval_samples_per_second": 20.506, |
|
"eval_steps_per_second": 1.293, |
|
"step": 3176 |
|
}, |
|
{ |
|
"epoch": 8.82, |
|
"learning_rate": 5.919395465994963e-06, |
|
"loss": 0.7597, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"eval_bleu": 0.1562, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6845841407775879, |
|
"eval_runtime": 38.8027, |
|
"eval_samples_per_second": 20.437, |
|
"eval_steps_per_second": 1.289, |
|
"step": 3573 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_bleu": 0.1592, |
|
"eval_gen_len": 19.0, |
|
"eval_loss": 0.6838726997375488, |
|
"eval_runtime": 38.9407, |
|
"eval_samples_per_second": 20.364, |
|
"eval_steps_per_second": 1.284, |
|
"step": 3970 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"step": 3970, |
|
"total_flos": 2.25634764201984e+16, |
|
"train_loss": 0.9340425890098591, |
|
"train_runtime": 2097.0146, |
|
"train_samples_per_second": 30.257, |
|
"train_steps_per_second": 1.893 |
|
} |
|
], |
|
"max_steps": 3970, |
|
"num_train_epochs": 10, |
|
"total_flos": 2.25634764201984e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|