Upload trainer_state.json
Browse files- trainer_state.json +49 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"global_step": 2060,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 4.85,
|
| 12 |
+
"learning_rate": 1.5145631067961166e-05,
|
| 13 |
+
"loss": 0.4008,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 9.71,
|
| 18 |
+
"learning_rate": 1.029126213592233e-05,
|
| 19 |
+
"loss": 0.1154,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 14.56,
|
| 24 |
+
"learning_rate": 5.436893203883496e-06,
|
| 25 |
+
"loss": 0.0677,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 19.42,
|
| 30 |
+
"learning_rate": 5.825242718446603e-07,
|
| 31 |
+
"loss": 0.0481,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 20.0,
|
| 36 |
+
"step": 2060,
|
| 37 |
+
"total_flos": 2.284156053808128e+16,
|
| 38 |
+
"train_loss": 0.15463279879208908,
|
| 39 |
+
"train_runtime": 2030.3525,
|
| 40 |
+
"train_samples_per_second": 64.383,
|
| 41 |
+
"train_steps_per_second": 1.015
|
| 42 |
+
}
|
| 43 |
+
],
|
| 44 |
+
"max_steps": 2060,
|
| 45 |
+
"num_train_epochs": 20,
|
| 46 |
+
"total_flos": 2.284156053808128e+16,
|
| 47 |
+
"trial_name": null,
|
| 48 |
+
"trial_params": null
|
| 49 |
+
}
|