Upload trainer_state.json
Browse files- trainer_state.json +46 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9143889652364229,
|
| 3 |
+
"best_model_checkpoint": "./resultsv3p2/output/checkpoint-196352",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"global_step": 196352,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 3.1928048978707266e-06,
|
| 13 |
+
"loss": 0.4695,
|
| 14 |
+
"step": 98176
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_accuracy": 0.9100117066218761,
|
| 19 |
+
"eval_loss": 0.3915119469165802,
|
| 20 |
+
"eval_runtime": 352.6229,
|
| 21 |
+
"eval_samples_per_second": 55.717,
|
| 22 |
+
"eval_steps_per_second": 13.93,
|
| 23 |
+
"step": 98176
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 2.0,
|
| 27 |
+
"learning_rate": 2.4055913745462424e-09,
|
| 28 |
+
"loss": 0.2675,
|
| 29 |
+
"step": 196352
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.0,
|
| 33 |
+
"eval_accuracy": 0.9143889652364229,
|
| 34 |
+
"eval_loss": 0.4945842921733856,
|
| 35 |
+
"eval_runtime": 227.1799,
|
| 36 |
+
"eval_samples_per_second": 86.482,
|
| 37 |
+
"eval_steps_per_second": 21.622,
|
| 38 |
+
"step": 196352
|
| 39 |
+
}
|
| 40 |
+
],
|
| 41 |
+
"max_steps": 196352,
|
| 42 |
+
"num_train_epochs": 2,
|
| 43 |
+
"total_flos": 8.31341361422996e+16,
|
| 44 |
+
"trial_name": null,
|
| 45 |
+
"trial_params": null
|
| 46 |
+
}
|