|
{ |
|
"best_metric": 19.1114, |
|
"best_model_checkpoint": "/local1/hfs/gs_stuff/ft-wmt14/0-0002/checkpoint-10000", |
|
"epoch": 0.2777777777777778, |
|
"eval_steps": 10000, |
|
"global_step": 10000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.2777777777777778, |
|
"grad_norm": 1.3005393743515015, |
|
"learning_rate": 0.0, |
|
"loss": 1.9517, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.2777777777777778, |
|
"eval_bleu": 19.1114, |
|
"eval_gen_len": 28.331, |
|
"eval_loss": 1.9202933311462402, |
|
"eval_runtime": 249.0743, |
|
"eval_samples_per_second": 12.045, |
|
"eval_steps_per_second": 1.506, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.2777777777777778, |
|
"step": 10000, |
|
"total_flos": 3807008386449408.0, |
|
"train_loss": 1.95171328125, |
|
"train_runtime": 1612.6909, |
|
"train_samples_per_second": 99.213, |
|
"train_steps_per_second": 6.201 |
|
} |
|
], |
|
"logging_steps": 10000, |
|
"max_steps": 10000, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 10000, |
|
"total_flos": 3807008386449408.0, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|