|
{ |
|
"best_metric": 7.917535781860352, |
|
"best_model_checkpoint": "checkpoints-mistral-300M-FA2/checkpoint-100", |
|
"epoch": 0.18117480537862704, |
|
"eval_steps": 100, |
|
"global_step": 100, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 5.999999999999999e-06, |
|
"loss": 10.9312, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.1999999999999999e-05, |
|
"loss": 10.3611, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.7999999999999997e-05, |
|
"loss": 9.6675, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 2.3999999999999997e-05, |
|
"loss": 9.2233, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.9999999999999997e-05, |
|
"loss": 8.9619, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 3.5999999999999994e-05, |
|
"loss": 8.7956, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 4.2e-05, |
|
"loss": 8.5973, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 4.7999999999999994e-05, |
|
"loss": 8.37, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 5.399999999999999e-05, |
|
"loss": 8.1088, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 5.9999999999999995e-05, |
|
"loss": 7.9131, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"eval_loss": 7.917535781860352, |
|
"eval_runtime": 1155.2404, |
|
"eval_samples_per_second": 7.467, |
|
"eval_steps_per_second": 7.467, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"step": 100, |
|
"total_flos": 4.78084620877824e+16, |
|
"train_loss": 9.09298583984375, |
|
"train_runtime": 7786.6651, |
|
"train_samples_per_second": 3.288, |
|
"train_steps_per_second": 0.013 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 100, |
|
"num_train_epochs": 1, |
|
"save_steps": 100, |
|
"total_flos": 4.78084620877824e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|