|
{ |
|
"best_metric": 1.292348861694336, |
|
"best_model_checkpoint": "../llama2-9439-21sept/checkpoint-400", |
|
"epoch": 0.6946454413892909, |
|
"eval_steps": 200, |
|
"global_step": 400, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 5.9999999999999995e-05, |
|
"loss": 3.4609, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 0.00011999999999999999, |
|
"loss": 2.8346, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.00017999999999999998, |
|
"loss": 1.7597, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 0.00023999999999999998, |
|
"loss": 1.542, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 0.0003, |
|
"loss": 1.4567, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 0.0002963076923076923, |
|
"loss": 1.4081, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 0.0002926153846153846, |
|
"loss": 1.3805, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 0.0002889230769230769, |
|
"loss": 1.3605, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 0.00028523076923076923, |
|
"loss": 1.3329, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 0.0002815384615384615, |
|
"loss": 1.3367, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"eval_loss": 1.3450770378112793, |
|
"eval_runtime": 6.306, |
|
"eval_samples_per_second": 47.574, |
|
"eval_steps_per_second": 1.269, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 0.0002778461538461538, |
|
"loss": 1.3152, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 0.0002741538461538461, |
|
"loss": 1.312, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 0.00027046153846153843, |
|
"loss": 1.2883, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 0.00026676923076923074, |
|
"loss": 1.2843, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 0.00026307692307692306, |
|
"loss": 1.2609, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 0.00025938461538461537, |
|
"loss": 1.2707, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 0.0002556923076923077, |
|
"loss": 1.2624, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 0.00025199999999999995, |
|
"loss": 1.273, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 0.0002483076923076923, |
|
"loss": 1.251, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 0.0002446153846153846, |
|
"loss": 1.2648, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"eval_loss": 1.292348861694336, |
|
"eval_runtime": 6.304, |
|
"eval_samples_per_second": 47.589, |
|
"eval_steps_per_second": 1.269, |
|
"step": 400 |
|
} |
|
], |
|
"logging_steps": 20, |
|
"max_steps": 1725, |
|
"num_train_epochs": 3, |
|
"save_steps": 200, |
|
"total_flos": 9.568312680146534e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|