Upload trainer_state.json
Browse files- trainer_state.json +67 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 3909,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.38,
|
| 12 |
+
"learning_rate": 4.360450243028908e-05,
|
| 13 |
+
"loss": 2.6273,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.77,
|
| 18 |
+
"learning_rate": 3.7209004860578156e-05,
|
| 19 |
+
"loss": 2.3755,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.15,
|
| 24 |
+
"learning_rate": 3.0813507290867236e-05,
|
| 25 |
+
"loss": 2.1927,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.53,
|
| 30 |
+
"learning_rate": 2.4418009721156306e-05,
|
| 31 |
+
"loss": 1.9905,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.92,
|
| 36 |
+
"learning_rate": 1.8022512151445383e-05,
|
| 37 |
+
"loss": 1.9672,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.3,
|
| 42 |
+
"learning_rate": 1.162701458173446e-05,
|
| 43 |
+
"loss": 1.841,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.69,
|
| 48 |
+
"learning_rate": 5.231517012023536e-06,
|
| 49 |
+
"loss": 1.8011,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"step": 3909,
|
| 55 |
+
"total_flos": 2119732555677696.0,
|
| 56 |
+
"train_loss": 2.077282390072389,
|
| 57 |
+
"train_runtime": 3363.6598,
|
| 58 |
+
"train_samples_per_second": 4.648,
|
| 59 |
+
"train_steps_per_second": 1.162
|
| 60 |
+
}
|
| 61 |
+
],
|
| 62 |
+
"max_steps": 3909,
|
| 63 |
+
"num_train_epochs": 3,
|
| 64 |
+
"total_flos": 2119732555677696.0,
|
| 65 |
+
"trial_name": null,
|
| 66 |
+
"trial_params": null
|
| 67 |
+
}
|