Upload trainer_state.json
Browse files- trainer_state.json +1270 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1270 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 4600,
|
| 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 |
+
"eval_accuracy": 0.531720741599073,
|
| 13 |
+
"eval_f1": 0.022944550669216066,
|
| 14 |
+
"eval_loss": 1.042980432510376,
|
| 15 |
+
"eval_precision": 0.013824884792626729,
|
| 16 |
+
"eval_recall": 0.06741573033707865,
|
| 17 |
+
"eval_runtime": 2.6201,
|
| 18 |
+
"eval_samples_per_second": 52.67,
|
| 19 |
+
"eval_steps_per_second": 4.58,
|
| 20 |
+
"step": 46
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.0,
|
| 24 |
+
"eval_accuracy": 0.5824159907300116,
|
| 25 |
+
"eval_f1": 0.05216284987277353,
|
| 26 |
+
"eval_loss": 0.9423891305923462,
|
| 27 |
+
"eval_precision": 0.03637976929902396,
|
| 28 |
+
"eval_recall": 0.09213483146067415,
|
| 29 |
+
"eval_runtime": 2.6433,
|
| 30 |
+
"eval_samples_per_second": 52.207,
|
| 31 |
+
"eval_steps_per_second": 4.54,
|
| 32 |
+
"step": 92
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.0,
|
| 36 |
+
"eval_accuracy": 0.6006662804171495,
|
| 37 |
+
"eval_f1": 0.06494287432351173,
|
| 38 |
+
"eval_loss": 0.8916851282119751,
|
| 39 |
+
"eval_precision": 0.04433497536945813,
|
| 40 |
+
"eval_recall": 0.12134831460674157,
|
| 41 |
+
"eval_runtime": 2.6927,
|
| 42 |
+
"eval_samples_per_second": 51.249,
|
| 43 |
+
"eval_steps_per_second": 4.456,
|
| 44 |
+
"step": 138
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.6420191193511008,
|
| 49 |
+
"eval_f1": 0.10659560293137908,
|
| 50 |
+
"eval_loss": 0.826358437538147,
|
| 51 |
+
"eval_precision": 0.07575757575757576,
|
| 52 |
+
"eval_recall": 0.1797752808988764,
|
| 53 |
+
"eval_runtime": 2.6923,
|
| 54 |
+
"eval_samples_per_second": 51.257,
|
| 55 |
+
"eval_steps_per_second": 4.457,
|
| 56 |
+
"step": 184
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 5.0,
|
| 60 |
+
"eval_accuracy": 0.6395567786790266,
|
| 61 |
+
"eval_f1": 0.17535903250188964,
|
| 62 |
+
"eval_loss": 0.9059988260269165,
|
| 63 |
+
"eval_precision": 0.13211845102505695,
|
| 64 |
+
"eval_recall": 0.2606741573033708,
|
| 65 |
+
"eval_runtime": 2.718,
|
| 66 |
+
"eval_samples_per_second": 50.774,
|
| 67 |
+
"eval_steps_per_second": 4.415,
|
| 68 |
+
"step": 230
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 6.0,
|
| 72 |
+
"eval_accuracy": 0.6408603707995365,
|
| 73 |
+
"eval_f1": 0.16282420749279541,
|
| 74 |
+
"eval_loss": 0.9093501567840576,
|
| 75 |
+
"eval_precision": 0.11983032873807,
|
| 76 |
+
"eval_recall": 0.2539325842696629,
|
| 77 |
+
"eval_runtime": 2.7417,
|
| 78 |
+
"eval_samples_per_second": 50.334,
|
| 79 |
+
"eval_steps_per_second": 4.377,
|
| 80 |
+
"step": 276
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 7.0,
|
| 84 |
+
"eval_accuracy": 0.6427433371958285,
|
| 85 |
+
"eval_f1": 0.17168481742472771,
|
| 86 |
+
"eval_loss": 1.0060389041900635,
|
| 87 |
+
"eval_precision": 0.12007168458781362,
|
| 88 |
+
"eval_recall": 0.30112359550561796,
|
| 89 |
+
"eval_runtime": 2.7409,
|
| 90 |
+
"eval_samples_per_second": 50.348,
|
| 91 |
+
"eval_steps_per_second": 4.378,
|
| 92 |
+
"step": 322
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.6558516801853997,
|
| 97 |
+
"eval_f1": 0.17363344051446947,
|
| 98 |
+
"eval_loss": 1.1526994705200195,
|
| 99 |
+
"eval_precision": 0.12162162162162163,
|
| 100 |
+
"eval_recall": 0.30337078651685395,
|
| 101 |
+
"eval_runtime": 2.7362,
|
| 102 |
+
"eval_samples_per_second": 50.435,
|
| 103 |
+
"eval_steps_per_second": 4.386,
|
| 104 |
+
"step": 368
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 9.0,
|
| 108 |
+
"eval_accuracy": 0.6541859791425261,
|
| 109 |
+
"eval_f1": 0.16133004926108374,
|
| 110 |
+
"eval_loss": 1.0466786623001099,
|
| 111 |
+
"eval_precision": 0.1111111111111111,
|
| 112 |
+
"eval_recall": 0.2943820224719101,
|
| 113 |
+
"eval_runtime": 2.7242,
|
| 114 |
+
"eval_samples_per_second": 50.657,
|
| 115 |
+
"eval_steps_per_second": 4.405,
|
| 116 |
+
"step": 414
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 10.0,
|
| 120 |
+
"eval_accuracy": 0.6379634994206257,
|
| 121 |
+
"eval_f1": 0.19889502762430938,
|
| 122 |
+
"eval_loss": 1.185328722000122,
|
| 123 |
+
"eval_precision": 0.14356929212362912,
|
| 124 |
+
"eval_recall": 0.3235955056179775,
|
| 125 |
+
"eval_runtime": 2.7468,
|
| 126 |
+
"eval_samples_per_second": 50.24,
|
| 127 |
+
"eval_steps_per_second": 4.369,
|
| 128 |
+
"step": 460
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 10.87,
|
| 132 |
+
"learning_rate": 1.782608695652174e-05,
|
| 133 |
+
"loss": 0.5814,
|
| 134 |
+
"step": 500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 11.0,
|
| 138 |
+
"eval_accuracy": 0.6622972190034763,
|
| 139 |
+
"eval_f1": 0.24944974321349964,
|
| 140 |
+
"eval_loss": 1.2020127773284912,
|
| 141 |
+
"eval_precision": 0.18518518518518517,
|
| 142 |
+
"eval_recall": 0.38202247191011235,
|
| 143 |
+
"eval_runtime": 2.7642,
|
| 144 |
+
"eval_samples_per_second": 49.924,
|
| 145 |
+
"eval_steps_per_second": 4.341,
|
| 146 |
+
"step": 506
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 12.0,
|
| 150 |
+
"eval_accuracy": 0.6589658169177288,
|
| 151 |
+
"eval_f1": 0.2267481330617787,
|
| 152 |
+
"eval_loss": 1.3076611757278442,
|
| 153 |
+
"eval_precision": 0.16245136186770429,
|
| 154 |
+
"eval_recall": 0.3752808988764045,
|
| 155 |
+
"eval_runtime": 2.7844,
|
| 156 |
+
"eval_samples_per_second": 49.562,
|
| 157 |
+
"eval_steps_per_second": 4.31,
|
| 158 |
+
"step": 552
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 13.0,
|
| 162 |
+
"eval_accuracy": 0.6307213209733488,
|
| 163 |
+
"eval_f1": 0.2636015325670498,
|
| 164 |
+
"eval_loss": 1.294709324836731,
|
| 165 |
+
"eval_precision": 0.2,
|
| 166 |
+
"eval_recall": 0.3865168539325843,
|
| 167 |
+
"eval_runtime": 2.7404,
|
| 168 |
+
"eval_samples_per_second": 50.358,
|
| 169 |
+
"eval_steps_per_second": 4.379,
|
| 170 |
+
"step": 598
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 14.0,
|
| 174 |
+
"eval_accuracy": 0.6591830822711472,
|
| 175 |
+
"eval_f1": 0.3390663390663391,
|
| 176 |
+
"eval_loss": 1.4205528497695923,
|
| 177 |
+
"eval_precision": 0.2667525773195876,
|
| 178 |
+
"eval_recall": 0.4651685393258427,
|
| 179 |
+
"eval_runtime": 2.7797,
|
| 180 |
+
"eval_samples_per_second": 49.646,
|
| 181 |
+
"eval_steps_per_second": 4.317,
|
| 182 |
+
"step": 644
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 15.0,
|
| 186 |
+
"eval_accuracy": 0.6625869061413673,
|
| 187 |
+
"eval_f1": 0.33782712133227594,
|
| 188 |
+
"eval_loss": 1.475420355796814,
|
| 189 |
+
"eval_precision": 0.2610294117647059,
|
| 190 |
+
"eval_recall": 0.4786516853932584,
|
| 191 |
+
"eval_runtime": 2.7429,
|
| 192 |
+
"eval_samples_per_second": 50.312,
|
| 193 |
+
"eval_steps_per_second": 4.375,
|
| 194 |
+
"step": 690
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 16.0,
|
| 198 |
+
"eval_accuracy": 0.6528099652375434,
|
| 199 |
+
"eval_f1": 0.3605728727885425,
|
| 200 |
+
"eval_loss": 1.6823292970657349,
|
| 201 |
+
"eval_precision": 0.2884097035040431,
|
| 202 |
+
"eval_recall": 0.48089887640449436,
|
| 203 |
+
"eval_runtime": 2.7637,
|
| 204 |
+
"eval_samples_per_second": 49.933,
|
| 205 |
+
"eval_steps_per_second": 4.342,
|
| 206 |
+
"step": 736
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 17.0,
|
| 210 |
+
"eval_accuracy": 0.662369640787949,
|
| 211 |
+
"eval_f1": 0.3175832687838885,
|
| 212 |
+
"eval_loss": 1.6413644552230835,
|
| 213 |
+
"eval_precision": 0.24231678486997635,
|
| 214 |
+
"eval_recall": 0.4606741573033708,
|
| 215 |
+
"eval_runtime": 2.7668,
|
| 216 |
+
"eval_samples_per_second": 49.877,
|
| 217 |
+
"eval_steps_per_second": 4.337,
|
| 218 |
+
"step": 782
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 18.0,
|
| 222 |
+
"eval_accuracy": 0.6634559675550405,
|
| 223 |
+
"eval_f1": 0.2549647661755285,
|
| 224 |
+
"eval_loss": 1.5539870262145996,
|
| 225 |
+
"eval_precision": 0.17831541218637992,
|
| 226 |
+
"eval_recall": 0.44719101123595506,
|
| 227 |
+
"eval_runtime": 2.7564,
|
| 228 |
+
"eval_samples_per_second": 50.065,
|
| 229 |
+
"eval_steps_per_second": 4.353,
|
| 230 |
+
"step": 828
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 19.0,
|
| 234 |
+
"eval_accuracy": 0.65454808806489,
|
| 235 |
+
"eval_f1": 0.32420429311621024,
|
| 236 |
+
"eval_loss": 1.7641867399215698,
|
| 237 |
+
"eval_precision": 0.24172185430463577,
|
| 238 |
+
"eval_recall": 0.49213483146067416,
|
| 239 |
+
"eval_runtime": 2.7607,
|
| 240 |
+
"eval_samples_per_second": 49.987,
|
| 241 |
+
"eval_steps_per_second": 4.347,
|
| 242 |
+
"step": 874
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 20.0,
|
| 246 |
+
"eval_accuracy": 0.6743192352259559,
|
| 247 |
+
"eval_f1": 0.29172510518934086,
|
| 248 |
+
"eval_loss": 1.6112134456634521,
|
| 249 |
+
"eval_precision": 0.21202854230377166,
|
| 250 |
+
"eval_recall": 0.46741573033707867,
|
| 251 |
+
"eval_runtime": 2.7419,
|
| 252 |
+
"eval_samples_per_second": 50.331,
|
| 253 |
+
"eval_steps_per_second": 4.377,
|
| 254 |
+
"step": 920
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 21.0,
|
| 258 |
+
"eval_accuracy": 0.6320973348783314,
|
| 259 |
+
"eval_f1": 0.2980225988700565,
|
| 260 |
+
"eval_loss": 1.9230778217315674,
|
| 261 |
+
"eval_precision": 0.21730175077239958,
|
| 262 |
+
"eval_recall": 0.47415730337078654,
|
| 263 |
+
"eval_runtime": 2.7602,
|
| 264 |
+
"eval_samples_per_second": 49.997,
|
| 265 |
+
"eval_steps_per_second": 4.348,
|
| 266 |
+
"step": 966
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 21.74,
|
| 270 |
+
"learning_rate": 1.565217391304348e-05,
|
| 271 |
+
"loss": 0.1098,
|
| 272 |
+
"step": 1000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 22.0,
|
| 276 |
+
"eval_accuracy": 0.6402085747392816,
|
| 277 |
+
"eval_f1": 0.3536776212832551,
|
| 278 |
+
"eval_loss": 2.1778900623321533,
|
| 279 |
+
"eval_precision": 0.27130852340936373,
|
| 280 |
+
"eval_recall": 0.5078651685393258,
|
| 281 |
+
"eval_runtime": 2.7589,
|
| 282 |
+
"eval_samples_per_second": 50.021,
|
| 283 |
+
"eval_steps_per_second": 4.35,
|
| 284 |
+
"step": 1012
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 23.0,
|
| 288 |
+
"eval_accuracy": 0.6542584009269988,
|
| 289 |
+
"eval_f1": 0.31053351573187415,
|
| 290 |
+
"eval_loss": 1.8866859674453735,
|
| 291 |
+
"eval_precision": 0.2232055063913471,
|
| 292 |
+
"eval_recall": 0.5101123595505618,
|
| 293 |
+
"eval_runtime": 2.7457,
|
| 294 |
+
"eval_samples_per_second": 50.26,
|
| 295 |
+
"eval_steps_per_second": 4.37,
|
| 296 |
+
"step": 1058
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 24.0,
|
| 300 |
+
"eval_accuracy": 0.6286935110081112,
|
| 301 |
+
"eval_f1": 0.29891304347826086,
|
| 302 |
+
"eval_loss": 2.057832956314087,
|
| 303 |
+
"eval_precision": 0.21421616358325218,
|
| 304 |
+
"eval_recall": 0.4943820224719101,
|
| 305 |
+
"eval_runtime": 2.7704,
|
| 306 |
+
"eval_samples_per_second": 49.813,
|
| 307 |
+
"eval_steps_per_second": 4.332,
|
| 308 |
+
"step": 1104
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 25.0,
|
| 312 |
+
"eval_accuracy": 0.6264484356894554,
|
| 313 |
+
"eval_f1": 0.2706571242680546,
|
| 314 |
+
"eval_loss": 2.031670093536377,
|
| 315 |
+
"eval_precision": 0.19047619047619047,
|
| 316 |
+
"eval_recall": 0.46741573033707867,
|
| 317 |
+
"eval_runtime": 2.7593,
|
| 318 |
+
"eval_samples_per_second": 50.013,
|
| 319 |
+
"eval_steps_per_second": 4.349,
|
| 320 |
+
"step": 1150
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 26.0,
|
| 324 |
+
"eval_accuracy": 0.6575898030127463,
|
| 325 |
+
"eval_f1": 0.34562211981566826,
|
| 326 |
+
"eval_loss": 1.8547980785369873,
|
| 327 |
+
"eval_precision": 0.26254375729288215,
|
| 328 |
+
"eval_recall": 0.5056179775280899,
|
| 329 |
+
"eval_runtime": 2.745,
|
| 330 |
+
"eval_samples_per_second": 50.273,
|
| 331 |
+
"eval_steps_per_second": 4.372,
|
| 332 |
+
"step": 1196
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 27.0,
|
| 336 |
+
"eval_accuracy": 0.6420915411355735,
|
| 337 |
+
"eval_f1": 0.3376230128690386,
|
| 338 |
+
"eval_loss": 2.0632588863372803,
|
| 339 |
+
"eval_precision": 0.2545662100456621,
|
| 340 |
+
"eval_recall": 0.501123595505618,
|
| 341 |
+
"eval_runtime": 2.7583,
|
| 342 |
+
"eval_samples_per_second": 50.03,
|
| 343 |
+
"eval_steps_per_second": 4.35,
|
| 344 |
+
"step": 1242
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 28.0,
|
| 348 |
+
"eval_accuracy": 0.666787369640788,
|
| 349 |
+
"eval_f1": 0.36335160532498045,
|
| 350 |
+
"eval_loss": 1.943156123161316,
|
| 351 |
+
"eval_precision": 0.27884615384615385,
|
| 352 |
+
"eval_recall": 0.5213483146067416,
|
| 353 |
+
"eval_runtime": 2.7624,
|
| 354 |
+
"eval_samples_per_second": 49.957,
|
| 355 |
+
"eval_steps_per_second": 4.344,
|
| 356 |
+
"step": 1288
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 29.0,
|
| 360 |
+
"eval_accuracy": 0.6518684820393974,
|
| 361 |
+
"eval_f1": 0.32880629020729096,
|
| 362 |
+
"eval_loss": 1.9604640007019043,
|
| 363 |
+
"eval_precision": 0.24109014675052412,
|
| 364 |
+
"eval_recall": 0.5168539325842697,
|
| 365 |
+
"eval_runtime": 2.7551,
|
| 366 |
+
"eval_samples_per_second": 50.089,
|
| 367 |
+
"eval_steps_per_second": 4.356,
|
| 368 |
+
"step": 1334
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 30.0,
|
| 372 |
+
"eval_accuracy": 0.6587485515643106,
|
| 373 |
+
"eval_f1": 0.3465045592705167,
|
| 374 |
+
"eval_loss": 2.0531818866729736,
|
| 375 |
+
"eval_precision": 0.2617680826636051,
|
| 376 |
+
"eval_recall": 0.5123595505617977,
|
| 377 |
+
"eval_runtime": 2.7577,
|
| 378 |
+
"eval_samples_per_second": 50.042,
|
| 379 |
+
"eval_steps_per_second": 4.352,
|
| 380 |
+
"step": 1380
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 31.0,
|
| 384 |
+
"eval_accuracy": 0.6528099652375434,
|
| 385 |
+
"eval_f1": 0.30739045127534337,
|
| 386 |
+
"eval_loss": 1.9861180782318115,
|
| 387 |
+
"eval_precision": 0.21678966789667897,
|
| 388 |
+
"eval_recall": 0.5280898876404494,
|
| 389 |
+
"eval_runtime": 2.7639,
|
| 390 |
+
"eval_samples_per_second": 49.929,
|
| 391 |
+
"eval_steps_per_second": 4.342,
|
| 392 |
+
"step": 1426
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 32.0,
|
| 396 |
+
"eval_accuracy": 0.6265932792584009,
|
| 397 |
+
"eval_f1": 0.25979112271540467,
|
| 398 |
+
"eval_loss": 2.3644657135009766,
|
| 399 |
+
"eval_precision": 0.18307267709291627,
|
| 400 |
+
"eval_recall": 0.44719101123595506,
|
| 401 |
+
"eval_runtime": 2.7617,
|
| 402 |
+
"eval_samples_per_second": 49.969,
|
| 403 |
+
"eval_steps_per_second": 4.345,
|
| 404 |
+
"step": 1472
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 32.61,
|
| 408 |
+
"learning_rate": 1.3478260869565218e-05,
|
| 409 |
+
"loss": 0.0461,
|
| 410 |
+
"step": 1500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 33.0,
|
| 414 |
+
"eval_accuracy": 0.6571552723059096,
|
| 415 |
+
"eval_f1": 0.2733908427339084,
|
| 416 |
+
"eval_loss": 2.0718517303466797,
|
| 417 |
+
"eval_precision": 0.19397363465160075,
|
| 418 |
+
"eval_recall": 0.46292134831460674,
|
| 419 |
+
"eval_runtime": 2.7478,
|
| 420 |
+
"eval_samples_per_second": 50.223,
|
| 421 |
+
"eval_steps_per_second": 4.367,
|
| 422 |
+
"step": 1518
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 34.0,
|
| 426 |
+
"eval_accuracy": 0.6454953650057937,
|
| 427 |
+
"eval_f1": 0.2763246143527833,
|
| 428 |
+
"eval_loss": 2.050283193588257,
|
| 429 |
+
"eval_precision": 0.19694072657743786,
|
| 430 |
+
"eval_recall": 0.46292134831460674,
|
| 431 |
+
"eval_runtime": 2.7452,
|
| 432 |
+
"eval_samples_per_second": 50.27,
|
| 433 |
+
"eval_steps_per_second": 4.371,
|
| 434 |
+
"step": 1564
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 35.0,
|
| 438 |
+
"eval_accuracy": 0.6310110081112399,
|
| 439 |
+
"eval_f1": 0.26170622193713916,
|
| 440 |
+
"eval_loss": 2.261981964111328,
|
| 441 |
+
"eval_precision": 0.18312387791741472,
|
| 442 |
+
"eval_recall": 0.4584269662921348,
|
| 443 |
+
"eval_runtime": 2.7577,
|
| 444 |
+
"eval_samples_per_second": 50.041,
|
| 445 |
+
"eval_steps_per_second": 4.351,
|
| 446 |
+
"step": 1610
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 36.0,
|
| 450 |
+
"eval_accuracy": 0.6399188876013905,
|
| 451 |
+
"eval_f1": 0.3131313131313131,
|
| 452 |
+
"eval_loss": 2.217377185821533,
|
| 453 |
+
"eval_precision": 0.230605738575983,
|
| 454 |
+
"eval_recall": 0.48764044943820223,
|
| 455 |
+
"eval_runtime": 2.7643,
|
| 456 |
+
"eval_samples_per_second": 49.922,
|
| 457 |
+
"eval_steps_per_second": 4.341,
|
| 458 |
+
"step": 1656
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 37.0,
|
| 462 |
+
"eval_accuracy": 0.6441917728852838,
|
| 463 |
+
"eval_f1": 0.314410480349345,
|
| 464 |
+
"eval_loss": 2.3862390518188477,
|
| 465 |
+
"eval_precision": 0.232508073196986,
|
| 466 |
+
"eval_recall": 0.4853932584269663,
|
| 467 |
+
"eval_runtime": 2.7654,
|
| 468 |
+
"eval_samples_per_second": 49.903,
|
| 469 |
+
"eval_steps_per_second": 4.339,
|
| 470 |
+
"step": 1702
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 38.0,
|
| 474 |
+
"eval_accuracy": 0.6550550405561993,
|
| 475 |
+
"eval_f1": 0.3128571428571429,
|
| 476 |
+
"eval_loss": 2.0702590942382812,
|
| 477 |
+
"eval_precision": 0.2293193717277487,
|
| 478 |
+
"eval_recall": 0.49213483146067416,
|
| 479 |
+
"eval_runtime": 2.7536,
|
| 480 |
+
"eval_samples_per_second": 50.116,
|
| 481 |
+
"eval_steps_per_second": 4.358,
|
| 482 |
+
"step": 1748
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 39.0,
|
| 486 |
+
"eval_accuracy": 0.6577346465816918,
|
| 487 |
+
"eval_f1": 0.343939393939394,
|
| 488 |
+
"eval_loss": 2.151489019393921,
|
| 489 |
+
"eval_precision": 0.25942857142857145,
|
| 490 |
+
"eval_recall": 0.5101123595505618,
|
| 491 |
+
"eval_runtime": 2.7686,
|
| 492 |
+
"eval_samples_per_second": 49.844,
|
| 493 |
+
"eval_steps_per_second": 4.334,
|
| 494 |
+
"step": 1794
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 40.0,
|
| 498 |
+
"eval_accuracy": 0.647595596755504,
|
| 499 |
+
"eval_f1": 0.3431151241534989,
|
| 500 |
+
"eval_loss": 2.1895194053649902,
|
| 501 |
+
"eval_precision": 0.2579185520361991,
|
| 502 |
+
"eval_recall": 0.5123595505617977,
|
| 503 |
+
"eval_runtime": 2.7697,
|
| 504 |
+
"eval_samples_per_second": 49.825,
|
| 505 |
+
"eval_steps_per_second": 4.333,
|
| 506 |
+
"step": 1840
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 41.0,
|
| 510 |
+
"eval_accuracy": 0.6629490150637312,
|
| 511 |
+
"eval_f1": 0.3240810202550638,
|
| 512 |
+
"eval_loss": 2.175402879714966,
|
| 513 |
+
"eval_precision": 0.24324324324324326,
|
| 514 |
+
"eval_recall": 0.4853932584269663,
|
| 515 |
+
"eval_runtime": 2.7386,
|
| 516 |
+
"eval_samples_per_second": 50.39,
|
| 517 |
+
"eval_steps_per_second": 4.382,
|
| 518 |
+
"step": 1886
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 42.0,
|
| 522 |
+
"eval_accuracy": 0.650130359212051,
|
| 523 |
+
"eval_f1": 0.31381055676066527,
|
| 524 |
+
"eval_loss": 2.3337295055389404,
|
| 525 |
+
"eval_precision": 0.23134328358208955,
|
| 526 |
+
"eval_recall": 0.48764044943820223,
|
| 527 |
+
"eval_runtime": 2.775,
|
| 528 |
+
"eval_samples_per_second": 49.729,
|
| 529 |
+
"eval_steps_per_second": 4.324,
|
| 530 |
+
"step": 1932
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 43.0,
|
| 534 |
+
"eval_accuracy": 0.6625869061413673,
|
| 535 |
+
"eval_f1": 0.312180889861415,
|
| 536 |
+
"eval_loss": 2.1595537662506104,
|
| 537 |
+
"eval_precision": 0.23110151187904968,
|
| 538 |
+
"eval_recall": 0.48089887640449436,
|
| 539 |
+
"eval_runtime": 2.7514,
|
| 540 |
+
"eval_samples_per_second": 50.156,
|
| 541 |
+
"eval_steps_per_second": 4.361,
|
| 542 |
+
"step": 1978
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 43.48,
|
| 546 |
+
"learning_rate": 1.1304347826086957e-05,
|
| 547 |
+
"loss": 0.0323,
|
| 548 |
+
"step": 2000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 44.0,
|
| 552 |
+
"eval_accuracy": 0.6530272305909618,
|
| 553 |
+
"eval_f1": 0.3159379407616361,
|
| 554 |
+
"eval_loss": 2.2453536987304688,
|
| 555 |
+
"eval_precision": 0.2302158273381295,
|
| 556 |
+
"eval_recall": 0.503370786516854,
|
| 557 |
+
"eval_runtime": 2.7499,
|
| 558 |
+
"eval_samples_per_second": 50.184,
|
| 559 |
+
"eval_steps_per_second": 4.364,
|
| 560 |
+
"step": 2024
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 45.0,
|
| 564 |
+
"eval_accuracy": 0.6525926998841252,
|
| 565 |
+
"eval_f1": 0.2999299229152067,
|
| 566 |
+
"eval_loss": 2.1538333892822266,
|
| 567 |
+
"eval_precision": 0.21792260692464357,
|
| 568 |
+
"eval_recall": 0.48089887640449436,
|
| 569 |
+
"eval_runtime": 2.772,
|
| 570 |
+
"eval_samples_per_second": 49.784,
|
| 571 |
+
"eval_steps_per_second": 4.329,
|
| 572 |
+
"step": 2070
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 46.0,
|
| 576 |
+
"eval_accuracy": 0.6556344148319815,
|
| 577 |
+
"eval_f1": 0.2761290322580645,
|
| 578 |
+
"eval_loss": 2.1632461547851562,
|
| 579 |
+
"eval_precision": 0.19366515837104073,
|
| 580 |
+
"eval_recall": 0.48089887640449436,
|
| 581 |
+
"eval_runtime": 2.7668,
|
| 582 |
+
"eval_samples_per_second": 49.877,
|
| 583 |
+
"eval_steps_per_second": 4.337,
|
| 584 |
+
"step": 2116
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 47.0,
|
| 588 |
+
"eval_accuracy": 0.6575898030127463,
|
| 589 |
+
"eval_f1": 0.2993762993762994,
|
| 590 |
+
"eval_loss": 2.178040027618408,
|
| 591 |
+
"eval_precision": 0.21643286573146292,
|
| 592 |
+
"eval_recall": 0.4853932584269663,
|
| 593 |
+
"eval_runtime": 2.7774,
|
| 594 |
+
"eval_samples_per_second": 49.686,
|
| 595 |
+
"eval_steps_per_second": 4.321,
|
| 596 |
+
"step": 2162
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 48.0,
|
| 600 |
+
"eval_accuracy": 0.6502027809965237,
|
| 601 |
+
"eval_f1": 0.29047949965253644,
|
| 602 |
+
"eval_loss": 2.3481132984161377,
|
| 603 |
+
"eval_precision": 0.21026156941649898,
|
| 604 |
+
"eval_recall": 0.4696629213483146,
|
| 605 |
+
"eval_runtime": 2.758,
|
| 606 |
+
"eval_samples_per_second": 50.036,
|
| 607 |
+
"eval_steps_per_second": 4.351,
|
| 608 |
+
"step": 2208
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 49.0,
|
| 612 |
+
"eval_accuracy": 0.6594003476245655,
|
| 613 |
+
"eval_f1": 0.2608695652173913,
|
| 614 |
+
"eval_loss": 2.1968982219696045,
|
| 615 |
+
"eval_precision": 0.18452935694315004,
|
| 616 |
+
"eval_recall": 0.4449438202247191,
|
| 617 |
+
"eval_runtime": 2.7601,
|
| 618 |
+
"eval_samples_per_second": 49.997,
|
| 619 |
+
"eval_steps_per_second": 4.348,
|
| 620 |
+
"step": 2254
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 50.0,
|
| 624 |
+
"eval_accuracy": 0.6613557358053302,
|
| 625 |
+
"eval_f1": 0.27965540092776675,
|
| 626 |
+
"eval_loss": 2.1524665355682373,
|
| 627 |
+
"eval_precision": 0.19830827067669174,
|
| 628 |
+
"eval_recall": 0.47415730337078654,
|
| 629 |
+
"eval_runtime": 2.7622,
|
| 630 |
+
"eval_samples_per_second": 49.96,
|
| 631 |
+
"eval_steps_per_second": 4.344,
|
| 632 |
+
"step": 2300
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 51.0,
|
| 636 |
+
"eval_accuracy": 0.6413673232908459,
|
| 637 |
+
"eval_f1": 0.26915647134578236,
|
| 638 |
+
"eval_loss": 2.244340658187866,
|
| 639 |
+
"eval_precision": 0.18862815884476533,
|
| 640 |
+
"eval_recall": 0.4696629213483146,
|
| 641 |
+
"eval_runtime": 2.7474,
|
| 642 |
+
"eval_samples_per_second": 50.229,
|
| 643 |
+
"eval_steps_per_second": 4.368,
|
| 644 |
+
"step": 2346
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 52.0,
|
| 648 |
+
"eval_accuracy": 0.6659907300115875,
|
| 649 |
+
"eval_f1": 0.29979181124219295,
|
| 650 |
+
"eval_loss": 2.1262786388397217,
|
| 651 |
+
"eval_precision": 0.21686746987951808,
|
| 652 |
+
"eval_recall": 0.4853932584269663,
|
| 653 |
+
"eval_runtime": 2.7827,
|
| 654 |
+
"eval_samples_per_second": 49.592,
|
| 655 |
+
"eval_steps_per_second": 4.312,
|
| 656 |
+
"step": 2392
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 53.0,
|
| 660 |
+
"eval_accuracy": 0.6543308227114716,
|
| 661 |
+
"eval_f1": 0.27220447284345045,
|
| 662 |
+
"eval_loss": 2.178636074066162,
|
| 663 |
+
"eval_precision": 0.19017857142857142,
|
| 664 |
+
"eval_recall": 0.4786516853932584,
|
| 665 |
+
"eval_runtime": 2.7405,
|
| 666 |
+
"eval_samples_per_second": 50.356,
|
| 667 |
+
"eval_steps_per_second": 4.379,
|
| 668 |
+
"step": 2438
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 54.0,
|
| 672 |
+
"eval_accuracy": 0.6599797219003476,
|
| 673 |
+
"eval_f1": 0.3191800878477306,
|
| 674 |
+
"eval_loss": 2.2686634063720703,
|
| 675 |
+
"eval_precision": 0.23669923995656894,
|
| 676 |
+
"eval_recall": 0.4898876404494382,
|
| 677 |
+
"eval_runtime": 2.7587,
|
| 678 |
+
"eval_samples_per_second": 50.024,
|
| 679 |
+
"eval_steps_per_second": 4.35,
|
| 680 |
+
"step": 2484
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 54.35,
|
| 684 |
+
"learning_rate": 9.130434782608697e-06,
|
| 685 |
+
"loss": 0.0289,
|
| 686 |
+
"step": 2500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 55.0,
|
| 690 |
+
"eval_accuracy": 0.6491164542294322,
|
| 691 |
+
"eval_f1": 0.3539412673879443,
|
| 692 |
+
"eval_loss": 2.4480252265930176,
|
| 693 |
+
"eval_precision": 0.2697290930506478,
|
| 694 |
+
"eval_recall": 0.5146067415730337,
|
| 695 |
+
"eval_runtime": 2.7586,
|
| 696 |
+
"eval_samples_per_second": 50.026,
|
| 697 |
+
"eval_steps_per_second": 4.35,
|
| 698 |
+
"step": 2530
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 56.0,
|
| 702 |
+
"eval_accuracy": 0.658603707995365,
|
| 703 |
+
"eval_f1": 0.32152974504249293,
|
| 704 |
+
"eval_loss": 2.241455078125,
|
| 705 |
+
"eval_precision": 0.23474663908996898,
|
| 706 |
+
"eval_recall": 0.5101123595505618,
|
| 707 |
+
"eval_runtime": 2.7669,
|
| 708 |
+
"eval_samples_per_second": 49.875,
|
| 709 |
+
"eval_steps_per_second": 4.337,
|
| 710 |
+
"step": 2576
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 57.0,
|
| 714 |
+
"eval_accuracy": 0.6625144843568945,
|
| 715 |
+
"eval_f1": 0.33824670287044223,
|
| 716 |
+
"eval_loss": 2.2077813148498535,
|
| 717 |
+
"eval_precision": 0.25829383886255924,
|
| 718 |
+
"eval_recall": 0.4898876404494382,
|
| 719 |
+
"eval_runtime": 2.7555,
|
| 720 |
+
"eval_samples_per_second": 50.082,
|
| 721 |
+
"eval_steps_per_second": 4.355,
|
| 722 |
+
"step": 2622
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 58.0,
|
| 726 |
+
"eval_accuracy": 0.6491164542294322,
|
| 727 |
+
"eval_f1": 0.30736240171551105,
|
| 728 |
+
"eval_loss": 2.3162100315093994,
|
| 729 |
+
"eval_precision": 0.22536687631027252,
|
| 730 |
+
"eval_recall": 0.48314606741573035,
|
| 731 |
+
"eval_runtime": 2.7573,
|
| 732 |
+
"eval_samples_per_second": 50.049,
|
| 733 |
+
"eval_steps_per_second": 4.352,
|
| 734 |
+
"step": 2668
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 59.0,
|
| 738 |
+
"eval_accuracy": 0.654837775202781,
|
| 739 |
+
"eval_f1": 0.35776201733648544,
|
| 740 |
+
"eval_loss": 2.3420791625976562,
|
| 741 |
+
"eval_precision": 0.27548543689320387,
|
| 742 |
+
"eval_recall": 0.5101123595505618,
|
| 743 |
+
"eval_runtime": 2.7605,
|
| 744 |
+
"eval_samples_per_second": 49.991,
|
| 745 |
+
"eval_steps_per_second": 4.347,
|
| 746 |
+
"step": 2714
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 60.0,
|
| 750 |
+
"eval_accuracy": 0.6544032444959443,
|
| 751 |
+
"eval_f1": 0.3475609756097561,
|
| 752 |
+
"eval_loss": 2.3971171379089355,
|
| 753 |
+
"eval_precision": 0.2629757785467128,
|
| 754 |
+
"eval_recall": 0.5123595505617977,
|
| 755 |
+
"eval_runtime": 2.7631,
|
| 756 |
+
"eval_samples_per_second": 49.943,
|
| 757 |
+
"eval_steps_per_second": 4.343,
|
| 758 |
+
"step": 2760
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 61.0,
|
| 762 |
+
"eval_accuracy": 0.6589658169177288,
|
| 763 |
+
"eval_f1": 0.356979405034325,
|
| 764 |
+
"eval_loss": 2.2836458683013916,
|
| 765 |
+
"eval_precision": 0.2702078521939954,
|
| 766 |
+
"eval_recall": 0.5258426966292135,
|
| 767 |
+
"eval_runtime": 2.7541,
|
| 768 |
+
"eval_samples_per_second": 50.107,
|
| 769 |
+
"eval_steps_per_second": 4.357,
|
| 770 |
+
"step": 2806
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 62.0,
|
| 774 |
+
"eval_accuracy": 0.6542584009269988,
|
| 775 |
+
"eval_f1": 0.35784685367702807,
|
| 776 |
+
"eval_loss": 2.3359830379486084,
|
| 777 |
+
"eval_precision": 0.2700228832951945,
|
| 778 |
+
"eval_recall": 0.5303370786516854,
|
| 779 |
+
"eval_runtime": 2.7479,
|
| 780 |
+
"eval_samples_per_second": 50.22,
|
| 781 |
+
"eval_steps_per_second": 4.367,
|
| 782 |
+
"step": 2852
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 63.0,
|
| 786 |
+
"eval_accuracy": 0.6491164542294322,
|
| 787 |
+
"eval_f1": 0.36519036519036524,
|
| 788 |
+
"eval_loss": 2.386730194091797,
|
| 789 |
+
"eval_precision": 0.27909738717339666,
|
| 790 |
+
"eval_recall": 0.5280898876404494,
|
| 791 |
+
"eval_runtime": 2.7514,
|
| 792 |
+
"eval_samples_per_second": 50.155,
|
| 793 |
+
"eval_steps_per_second": 4.361,
|
| 794 |
+
"step": 2898
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 64.0,
|
| 798 |
+
"eval_accuracy": 0.6441917728852838,
|
| 799 |
+
"eval_f1": 0.35186595582635183,
|
| 800 |
+
"eval_loss": 2.383052349090576,
|
| 801 |
+
"eval_precision": 0.2661290322580645,
|
| 802 |
+
"eval_recall": 0.5191011235955056,
|
| 803 |
+
"eval_runtime": 2.7602,
|
| 804 |
+
"eval_samples_per_second": 49.996,
|
| 805 |
+
"eval_steps_per_second": 4.347,
|
| 806 |
+
"step": 2944
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 65.0,
|
| 810 |
+
"eval_accuracy": 0.6536790266512167,
|
| 811 |
+
"eval_f1": 0.3841166936790924,
|
| 812 |
+
"eval_loss": 2.5153658390045166,
|
| 813 |
+
"eval_precision": 0.30038022813688214,
|
| 814 |
+
"eval_recall": 0.5325842696629214,
|
| 815 |
+
"eval_runtime": 2.7644,
|
| 816 |
+
"eval_samples_per_second": 49.92,
|
| 817 |
+
"eval_steps_per_second": 4.341,
|
| 818 |
+
"step": 2990
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 65.22,
|
| 822 |
+
"learning_rate": 6.956521739130435e-06,
|
| 823 |
+
"loss": 0.0147,
|
| 824 |
+
"step": 3000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 66.0,
|
| 828 |
+
"eval_accuracy": 0.664904403244496,
|
| 829 |
+
"eval_f1": 0.37846655791190864,
|
| 830 |
+
"eval_loss": 2.3971505165100098,
|
| 831 |
+
"eval_precision": 0.2970550576184379,
|
| 832 |
+
"eval_recall": 0.5213483146067416,
|
| 833 |
+
"eval_runtime": 2.7665,
|
| 834 |
+
"eval_samples_per_second": 49.882,
|
| 835 |
+
"eval_steps_per_second": 4.338,
|
| 836 |
+
"step": 3036
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 67.0,
|
| 840 |
+
"eval_accuracy": 0.6633835457705678,
|
| 841 |
+
"eval_f1": 0.36391912908242613,
|
| 842 |
+
"eval_loss": 2.3396737575531006,
|
| 843 |
+
"eval_precision": 0.27824019024970276,
|
| 844 |
+
"eval_recall": 0.5258426966292135,
|
| 845 |
+
"eval_runtime": 2.7947,
|
| 846 |
+
"eval_samples_per_second": 49.379,
|
| 847 |
+
"eval_steps_per_second": 4.294,
|
| 848 |
+
"step": 3082
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 68.0,
|
| 852 |
+
"eval_accuracy": 0.6564310544611819,
|
| 853 |
+
"eval_f1": 0.35294117647058826,
|
| 854 |
+
"eval_loss": 2.459906816482544,
|
| 855 |
+
"eval_precision": 0.269185360094451,
|
| 856 |
+
"eval_recall": 0.5123595505617977,
|
| 857 |
+
"eval_runtime": 2.748,
|
| 858 |
+
"eval_samples_per_second": 50.218,
|
| 859 |
+
"eval_steps_per_second": 4.367,
|
| 860 |
+
"step": 3128
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 69.0,
|
| 864 |
+
"eval_accuracy": 0.6593279258400927,
|
| 865 |
+
"eval_f1": 0.36205533596837947,
|
| 866 |
+
"eval_loss": 2.3763880729675293,
|
| 867 |
+
"eval_precision": 0.27926829268292686,
|
| 868 |
+
"eval_recall": 0.5146067415730337,
|
| 869 |
+
"eval_runtime": 2.7664,
|
| 870 |
+
"eval_samples_per_second": 49.884,
|
| 871 |
+
"eval_steps_per_second": 4.338,
|
| 872 |
+
"step": 3174
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 70.0,
|
| 876 |
+
"eval_accuracy": 0.654837775202781,
|
| 877 |
+
"eval_f1": 0.3348148148148148,
|
| 878 |
+
"eval_loss": 2.368340492248535,
|
| 879 |
+
"eval_precision": 0.24972375690607734,
|
| 880 |
+
"eval_recall": 0.5078651685393258,
|
| 881 |
+
"eval_runtime": 2.7667,
|
| 882 |
+
"eval_samples_per_second": 49.878,
|
| 883 |
+
"eval_steps_per_second": 4.337,
|
| 884 |
+
"step": 3220
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 71.0,
|
| 888 |
+
"eval_accuracy": 0.6617902665121669,
|
| 889 |
+
"eval_f1": 0.32901751713632904,
|
| 890 |
+
"eval_loss": 2.3155479431152344,
|
| 891 |
+
"eval_precision": 0.2488479262672811,
|
| 892 |
+
"eval_recall": 0.4853932584269663,
|
| 893 |
+
"eval_runtime": 2.7611,
|
| 894 |
+
"eval_samples_per_second": 49.981,
|
| 895 |
+
"eval_steps_per_second": 4.346,
|
| 896 |
+
"step": 3266
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 72.0,
|
| 900 |
+
"eval_accuracy": 0.6662804171494786,
|
| 901 |
+
"eval_f1": 0.3543307086614173,
|
| 902 |
+
"eval_loss": 2.3285837173461914,
|
| 903 |
+
"eval_precision": 0.2727272727272727,
|
| 904 |
+
"eval_recall": 0.5056179775280899,
|
| 905 |
+
"eval_runtime": 2.7664,
|
| 906 |
+
"eval_samples_per_second": 49.885,
|
| 907 |
+
"eval_steps_per_second": 4.338,
|
| 908 |
+
"step": 3312
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 73.0,
|
| 912 |
+
"eval_accuracy": 0.6607763615295481,
|
| 913 |
+
"eval_f1": 0.33206397562833206,
|
| 914 |
+
"eval_loss": 2.3957254886627197,
|
| 915 |
+
"eval_precision": 0.2511520737327189,
|
| 916 |
+
"eval_recall": 0.4898876404494382,
|
| 917 |
+
"eval_runtime": 2.7653,
|
| 918 |
+
"eval_samples_per_second": 49.905,
|
| 919 |
+
"eval_steps_per_second": 4.34,
|
| 920 |
+
"step": 3358
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 74.0,
|
| 924 |
+
"eval_accuracy": 0.6594727694090382,
|
| 925 |
+
"eval_f1": 0.3111432706222865,
|
| 926 |
+
"eval_loss": 2.3470306396484375,
|
| 927 |
+
"eval_precision": 0.2294557097118463,
|
| 928 |
+
"eval_recall": 0.48314606741573035,
|
| 929 |
+
"eval_runtime": 2.7615,
|
| 930 |
+
"eval_samples_per_second": 49.973,
|
| 931 |
+
"eval_steps_per_second": 4.345,
|
| 932 |
+
"step": 3404
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 75.0,
|
| 936 |
+
"eval_accuracy": 0.6536066048667439,
|
| 937 |
+
"eval_f1": 0.32770022042615726,
|
| 938 |
+
"eval_loss": 2.451366424560547,
|
| 939 |
+
"eval_precision": 0.24344978165938866,
|
| 940 |
+
"eval_recall": 0.501123595505618,
|
| 941 |
+
"eval_runtime": 2.7612,
|
| 942 |
+
"eval_samples_per_second": 49.977,
|
| 943 |
+
"eval_steps_per_second": 4.346,
|
| 944 |
+
"step": 3450
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 76.0,
|
| 948 |
+
"eval_accuracy": 0.659545191193511,
|
| 949 |
+
"eval_f1": 0.3306152705707932,
|
| 950 |
+
"eval_loss": 2.346513271331787,
|
| 951 |
+
"eval_precision": 0.24668141592920353,
|
| 952 |
+
"eval_recall": 0.501123595505618,
|
| 953 |
+
"eval_runtime": 2.7429,
|
| 954 |
+
"eval_samples_per_second": 50.311,
|
| 955 |
+
"eval_steps_per_second": 4.375,
|
| 956 |
+
"step": 3496
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 76.09,
|
| 960 |
+
"learning_rate": 4.782608695652174e-06,
|
| 961 |
+
"loss": 0.0156,
|
| 962 |
+
"step": 3500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 77.0,
|
| 966 |
+
"eval_accuracy": 0.6538238702201622,
|
| 967 |
+
"eval_f1": 0.3167746580273578,
|
| 968 |
+
"eval_loss": 2.42856502532959,
|
| 969 |
+
"eval_precision": 0.2330508474576271,
|
| 970 |
+
"eval_recall": 0.4943820224719101,
|
| 971 |
+
"eval_runtime": 2.7674,
|
| 972 |
+
"eval_samples_per_second": 49.867,
|
| 973 |
+
"eval_steps_per_second": 4.336,
|
| 974 |
+
"step": 3542
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 78.0,
|
| 978 |
+
"eval_accuracy": 0.6603418308227115,
|
| 979 |
+
"eval_f1": 0.336559940431869,
|
| 980 |
+
"eval_loss": 2.3454573154449463,
|
| 981 |
+
"eval_precision": 0.2516703786191537,
|
| 982 |
+
"eval_recall": 0.5078651685393258,
|
| 983 |
+
"eval_runtime": 2.7701,
|
| 984 |
+
"eval_samples_per_second": 49.818,
|
| 985 |
+
"eval_steps_per_second": 4.332,
|
| 986 |
+
"step": 3588
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 79.0,
|
| 990 |
+
"eval_accuracy": 0.655489571263036,
|
| 991 |
+
"eval_f1": 0.3257520176082172,
|
| 992 |
+
"eval_loss": 2.332895040512085,
|
| 993 |
+
"eval_precision": 0.24183006535947713,
|
| 994 |
+
"eval_recall": 0.49887640449438203,
|
| 995 |
+
"eval_runtime": 2.7724,
|
| 996 |
+
"eval_samples_per_second": 49.776,
|
| 997 |
+
"eval_steps_per_second": 4.328,
|
| 998 |
+
"step": 3634
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 80.0,
|
| 1002 |
+
"eval_accuracy": 0.6509269988412515,
|
| 1003 |
+
"eval_f1": 0.3096960926193922,
|
| 1004 |
+
"eval_loss": 2.4176864624023438,
|
| 1005 |
+
"eval_precision": 0.22838847385272146,
|
| 1006 |
+
"eval_recall": 0.48089887640449436,
|
| 1007 |
+
"eval_runtime": 2.7816,
|
| 1008 |
+
"eval_samples_per_second": 49.612,
|
| 1009 |
+
"eval_steps_per_second": 4.314,
|
| 1010 |
+
"step": 3680
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 81.0,
|
| 1014 |
+
"eval_accuracy": 0.663311123986095,
|
| 1015 |
+
"eval_f1": 0.31703703703703706,
|
| 1016 |
+
"eval_loss": 2.322171926498413,
|
| 1017 |
+
"eval_precision": 0.23646408839779007,
|
| 1018 |
+
"eval_recall": 0.48089887640449436,
|
| 1019 |
+
"eval_runtime": 2.7495,
|
| 1020 |
+
"eval_samples_per_second": 50.191,
|
| 1021 |
+
"eval_steps_per_second": 4.364,
|
| 1022 |
+
"step": 3726
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 82.0,
|
| 1026 |
+
"eval_accuracy": 0.6552723059096176,
|
| 1027 |
+
"eval_f1": 0.3291512915129151,
|
| 1028 |
+
"eval_loss": 2.358896493911743,
|
| 1029 |
+
"eval_precision": 0.24505494505494504,
|
| 1030 |
+
"eval_recall": 0.501123595505618,
|
| 1031 |
+
"eval_runtime": 2.7748,
|
| 1032 |
+
"eval_samples_per_second": 49.733,
|
| 1033 |
+
"eval_steps_per_second": 4.325,
|
| 1034 |
+
"step": 3772
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 83.0,
|
| 1038 |
+
"eval_accuracy": 0.6573001158748552,
|
| 1039 |
+
"eval_f1": 0.3404907975460122,
|
| 1040 |
+
"eval_loss": 2.3836281299591064,
|
| 1041 |
+
"eval_precision": 0.25844004656577413,
|
| 1042 |
+
"eval_recall": 0.49887640449438203,
|
| 1043 |
+
"eval_runtime": 2.7586,
|
| 1044 |
+
"eval_samples_per_second": 50.025,
|
| 1045 |
+
"eval_steps_per_second": 4.35,
|
| 1046 |
+
"step": 3818
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 84.0,
|
| 1050 |
+
"eval_accuracy": 0.6519409038238703,
|
| 1051 |
+
"eval_f1": 0.35154394299287417,
|
| 1052 |
+
"eval_loss": 2.417782783508301,
|
| 1053 |
+
"eval_precision": 0.2713936430317848,
|
| 1054 |
+
"eval_recall": 0.49887640449438203,
|
| 1055 |
+
"eval_runtime": 2.751,
|
| 1056 |
+
"eval_samples_per_second": 50.164,
|
| 1057 |
+
"eval_steps_per_second": 4.362,
|
| 1058 |
+
"step": 3864
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 85.0,
|
| 1062 |
+
"eval_accuracy": 0.6506373117033604,
|
| 1063 |
+
"eval_f1": 0.35266084193804603,
|
| 1064 |
+
"eval_loss": 2.494265079498291,
|
| 1065 |
+
"eval_precision": 0.2727272727272727,
|
| 1066 |
+
"eval_recall": 0.49887640449438203,
|
| 1067 |
+
"eval_runtime": 2.791,
|
| 1068 |
+
"eval_samples_per_second": 49.444,
|
| 1069 |
+
"eval_steps_per_second": 4.299,
|
| 1070 |
+
"step": 3910
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 86.0,
|
| 1074 |
+
"eval_accuracy": 0.6568655851680185,
|
| 1075 |
+
"eval_f1": 0.34029389017788086,
|
| 1076 |
+
"eval_loss": 2.3840575218200684,
|
| 1077 |
+
"eval_precision": 0.25943396226415094,
|
| 1078 |
+
"eval_recall": 0.4943820224719101,
|
| 1079 |
+
"eval_runtime": 2.7591,
|
| 1080 |
+
"eval_samples_per_second": 50.016,
|
| 1081 |
+
"eval_steps_per_second": 4.349,
|
| 1082 |
+
"step": 3956
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 86.96,
|
| 1086 |
+
"learning_rate": 2.6086956521739132e-06,
|
| 1087 |
+
"loss": 0.0172,
|
| 1088 |
+
"step": 4000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 87.0,
|
| 1092 |
+
"eval_accuracy": 0.6472334878331402,
|
| 1093 |
+
"eval_f1": 0.3413498836307215,
|
| 1094 |
+
"eval_loss": 2.444718599319458,
|
| 1095 |
+
"eval_precision": 0.26066350710900477,
|
| 1096 |
+
"eval_recall": 0.4943820224719101,
|
| 1097 |
+
"eval_runtime": 2.7661,
|
| 1098 |
+
"eval_samples_per_second": 49.89,
|
| 1099 |
+
"eval_steps_per_second": 4.338,
|
| 1100 |
+
"step": 4002
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 88.0,
|
| 1104 |
+
"eval_accuracy": 0.6479577056778679,
|
| 1105 |
+
"eval_f1": 0.36450079239302696,
|
| 1106 |
+
"eval_loss": 2.4771909713745117,
|
| 1107 |
+
"eval_precision": 0.28151774785801714,
|
| 1108 |
+
"eval_recall": 0.5168539325842697,
|
| 1109 |
+
"eval_runtime": 2.7594,
|
| 1110 |
+
"eval_samples_per_second": 50.01,
|
| 1111 |
+
"eval_steps_per_second": 4.349,
|
| 1112 |
+
"step": 4048
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 89.0,
|
| 1116 |
+
"eval_accuracy": 0.6564310544611819,
|
| 1117 |
+
"eval_f1": 0.36450839328537166,
|
| 1118 |
+
"eval_loss": 2.3940951824188232,
|
| 1119 |
+
"eval_precision": 0.28287841191067,
|
| 1120 |
+
"eval_recall": 0.5123595505617977,
|
| 1121 |
+
"eval_runtime": 2.766,
|
| 1122 |
+
"eval_samples_per_second": 49.892,
|
| 1123 |
+
"eval_steps_per_second": 4.338,
|
| 1124 |
+
"step": 4094
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 90.0,
|
| 1128 |
+
"eval_accuracy": 0.664904403244496,
|
| 1129 |
+
"eval_f1": 0.36421725239616615,
|
| 1130 |
+
"eval_loss": 2.3723766803741455,
|
| 1131 |
+
"eval_precision": 0.2825278810408922,
|
| 1132 |
+
"eval_recall": 0.5123595505617977,
|
| 1133 |
+
"eval_runtime": 2.7724,
|
| 1134 |
+
"eval_samples_per_second": 49.776,
|
| 1135 |
+
"eval_steps_per_second": 4.328,
|
| 1136 |
+
"step": 4140
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 91.0,
|
| 1140 |
+
"eval_accuracy": 0.6589658169177288,
|
| 1141 |
+
"eval_f1": 0.35759493670886083,
|
| 1142 |
+
"eval_loss": 2.412285566329956,
|
| 1143 |
+
"eval_precision": 0.27594627594627597,
|
| 1144 |
+
"eval_recall": 0.5078651685393258,
|
| 1145 |
+
"eval_runtime": 2.7583,
|
| 1146 |
+
"eval_samples_per_second": 50.03,
|
| 1147 |
+
"eval_steps_per_second": 4.35,
|
| 1148 |
+
"step": 4186
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 92.0,
|
| 1152 |
+
"eval_accuracy": 0.6568655851680185,
|
| 1153 |
+
"eval_f1": 0.3537735849056604,
|
| 1154 |
+
"eval_loss": 2.4169280529022217,
|
| 1155 |
+
"eval_precision": 0.2720677146311971,
|
| 1156 |
+
"eval_recall": 0.5056179775280899,
|
| 1157 |
+
"eval_runtime": 2.7606,
|
| 1158 |
+
"eval_samples_per_second": 49.989,
|
| 1159 |
+
"eval_steps_per_second": 4.347,
|
| 1160 |
+
"step": 4232
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 93.0,
|
| 1164 |
+
"eval_accuracy": 0.6580243337195828,
|
| 1165 |
+
"eval_f1": 0.34843749999999996,
|
| 1166 |
+
"eval_loss": 2.4057414531707764,
|
| 1167 |
+
"eval_precision": 0.26706586826347306,
|
| 1168 |
+
"eval_recall": 0.501123595505618,
|
| 1169 |
+
"eval_runtime": 2.7672,
|
| 1170 |
+
"eval_samples_per_second": 49.87,
|
| 1171 |
+
"eval_steps_per_second": 4.337,
|
| 1172 |
+
"step": 4278
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 94.0,
|
| 1176 |
+
"eval_accuracy": 0.6558516801853997,
|
| 1177 |
+
"eval_f1": 0.34493426140757927,
|
| 1178 |
+
"eval_loss": 2.417436361312866,
|
| 1179 |
+
"eval_precision": 0.26297169811320753,
|
| 1180 |
+
"eval_recall": 0.501123595505618,
|
| 1181 |
+
"eval_runtime": 2.7713,
|
| 1182 |
+
"eval_samples_per_second": 49.796,
|
| 1183 |
+
"eval_steps_per_second": 4.33,
|
| 1184 |
+
"step": 4324
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 95.0,
|
| 1188 |
+
"eval_accuracy": 0.6587485515643106,
|
| 1189 |
+
"eval_f1": 0.3555555555555556,
|
| 1190 |
+
"eval_loss": 2.4058446884155273,
|
| 1191 |
+
"eval_precision": 0.2748466257668712,
|
| 1192 |
+
"eval_recall": 0.503370786516854,
|
| 1193 |
+
"eval_runtime": 2.7567,
|
| 1194 |
+
"eval_samples_per_second": 50.06,
|
| 1195 |
+
"eval_steps_per_second": 4.353,
|
| 1196 |
+
"step": 4370
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 96.0,
|
| 1200 |
+
"eval_accuracy": 0.6549101969872537,
|
| 1201 |
+
"eval_f1": 0.3468118195956454,
|
| 1202 |
+
"eval_loss": 2.423997163772583,
|
| 1203 |
+
"eval_precision": 0.26516052318668254,
|
| 1204 |
+
"eval_recall": 0.501123595505618,
|
| 1205 |
+
"eval_runtime": 2.7522,
|
| 1206 |
+
"eval_samples_per_second": 50.142,
|
| 1207 |
+
"eval_steps_per_second": 4.36,
|
| 1208 |
+
"step": 4416
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 97.0,
|
| 1212 |
+
"eval_accuracy": 0.6596900347624566,
|
| 1213 |
+
"eval_f1": 0.3534956794972506,
|
| 1214 |
+
"eval_loss": 2.406991720199585,
|
| 1215 |
+
"eval_precision": 0.2717391304347826,
|
| 1216 |
+
"eval_recall": 0.5056179775280899,
|
| 1217 |
+
"eval_runtime": 2.765,
|
| 1218 |
+
"eval_samples_per_second": 49.91,
|
| 1219 |
+
"eval_steps_per_second": 4.34,
|
| 1220 |
+
"step": 4462
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 97.83,
|
| 1224 |
+
"learning_rate": 4.347826086956522e-07,
|
| 1225 |
+
"loss": 0.0106,
|
| 1226 |
+
"step": 4500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 98.0,
|
| 1230 |
+
"eval_accuracy": 0.6607763615295481,
|
| 1231 |
+
"eval_f1": 0.34850863422292,
|
| 1232 |
+
"eval_loss": 2.402587413787842,
|
| 1233 |
+
"eval_precision": 0.2677925211097708,
|
| 1234 |
+
"eval_recall": 0.49887640449438203,
|
| 1235 |
+
"eval_runtime": 2.7737,
|
| 1236 |
+
"eval_samples_per_second": 49.753,
|
| 1237 |
+
"eval_steps_per_second": 4.326,
|
| 1238 |
+
"step": 4508
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 99.0,
|
| 1242 |
+
"eval_accuracy": 0.6622247972190035,
|
| 1243 |
+
"eval_f1": 0.35062893081761004,
|
| 1244 |
+
"eval_loss": 2.4004111289978027,
|
| 1245 |
+
"eval_precision": 0.26964933494558646,
|
| 1246 |
+
"eval_recall": 0.501123595505618,
|
| 1247 |
+
"eval_runtime": 2.7443,
|
| 1248 |
+
"eval_samples_per_second": 50.286,
|
| 1249 |
+
"eval_steps_per_second": 4.373,
|
| 1250 |
+
"step": 4554
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 100.0,
|
| 1254 |
+
"eval_accuracy": 0.6621523754345308,
|
| 1255 |
+
"eval_f1": 0.35035349567949725,
|
| 1256 |
+
"eval_loss": 2.4005491733551025,
|
| 1257 |
+
"eval_precision": 0.2693236714975845,
|
| 1258 |
+
"eval_recall": 0.501123595505618,
|
| 1259 |
+
"eval_runtime": 2.749,
|
| 1260 |
+
"eval_samples_per_second": 50.2,
|
| 1261 |
+
"eval_steps_per_second": 4.365,
|
| 1262 |
+
"step": 4600
|
| 1263 |
+
}
|
| 1264 |
+
],
|
| 1265 |
+
"max_steps": 4600,
|
| 1266 |
+
"num_train_epochs": 100,
|
| 1267 |
+
"total_flos": 8533183236037200.0,
|
| 1268 |
+
"trial_name": null,
|
| 1269 |
+
"trial_params": null
|
| 1270 |
+
}
|