End of training
Browse files- all_results.json +25 -0
- eval_results.json +12 -0
- predict_results.json +10 -0
- predictions.txt +0 -0
- train_results.json +8 -0
- trainer_state.json +2161 -0
all_results.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"eval_accuracy": 0.7937549199685122,
|
| 4 |
+
"eval_f1": 0.6471614548382768,
|
| 5 |
+
"eval_loss": 2.2270121574401855,
|
| 6 |
+
"eval_precision": 0.6231584388730939,
|
| 7 |
+
"eval_recall": 0.6730876605248465,
|
| 8 |
+
"eval_runtime": 2.7334,
|
| 9 |
+
"eval_samples": 1112,
|
| 10 |
+
"eval_samples_per_second": 406.825,
|
| 11 |
+
"eval_steps_per_second": 25.609,
|
| 12 |
+
"predict_accuracy": 0.7971507299056106,
|
| 13 |
+
"predict_f1": 0.6426393875753675,
|
| 14 |
+
"predict_loss": 2.1785261631011963,
|
| 15 |
+
"predict_precision": 0.6145374449339207,
|
| 16 |
+
"predict_recall": 0.6734346159307113,
|
| 17 |
+
"predict_runtime": 5.5541,
|
| 18 |
+
"predict_samples_per_second": 400.602,
|
| 19 |
+
"predict_steps_per_second": 25.206,
|
| 20 |
+
"train_loss": 0.039664112105659635,
|
| 21 |
+
"train_runtime": 7090.7344,
|
| 22 |
+
"train_samples": 7785,
|
| 23 |
+
"train_samples_per_second": 131.749,
|
| 24 |
+
"train_steps_per_second": 8.242
|
| 25 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"eval_accuracy": 0.7937549199685122,
|
| 4 |
+
"eval_f1": 0.6471614548382768,
|
| 5 |
+
"eval_loss": 2.2270121574401855,
|
| 6 |
+
"eval_precision": 0.6231584388730939,
|
| 7 |
+
"eval_recall": 0.6730876605248465,
|
| 8 |
+
"eval_runtime": 2.7334,
|
| 9 |
+
"eval_samples": 1112,
|
| 10 |
+
"eval_samples_per_second": 406.825,
|
| 11 |
+
"eval_steps_per_second": 25.609
|
| 12 |
+
}
|
predict_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"predict_accuracy": 0.7971507299056106,
|
| 3 |
+
"predict_f1": 0.6426393875753675,
|
| 4 |
+
"predict_loss": 2.1785261631011963,
|
| 5 |
+
"predict_precision": 0.6145374449339207,
|
| 6 |
+
"predict_recall": 0.6734346159307113,
|
| 7 |
+
"predict_runtime": 5.5541,
|
| 8 |
+
"predict_samples_per_second": 400.602,
|
| 9 |
+
"predict_steps_per_second": 25.206
|
| 10 |
+
}
|
predictions.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"train_loss": 0.039664112105659635,
|
| 4 |
+
"train_runtime": 7090.7344,
|
| 5 |
+
"train_samples": 7785,
|
| 6 |
+
"train_samples_per_second": 131.749,
|
| 7 |
+
"train_steps_per_second": 8.242
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2161 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6471614548382768,
|
| 3 |
+
"best_model_checkpoint": "/tmp/test-ner1_/checkpoint-50648",
|
| 4 |
+
"epoch": 120.0,
|
| 5 |
+
"global_step": 58440,
|
| 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": 0.7624102478471411,
|
| 13 |
+
"eval_f1": 0.47653262255186,
|
| 14 |
+
"eval_loss": 0.7562578320503235,
|
| 15 |
+
"eval_precision": 0.40737071527640184,
|
| 16 |
+
"eval_recall": 0.5739810161920714,
|
| 17 |
+
"eval_runtime": 2.7545,
|
| 18 |
+
"eval_samples_per_second": 403.708,
|
| 19 |
+
"eval_steps_per_second": 25.413,
|
| 20 |
+
"step": 487
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.03,
|
| 24 |
+
"learning_rate": 2.9743326488706367e-05,
|
| 25 |
+
"loss": 1.1928,
|
| 26 |
+
"step": 500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"eval_accuracy": 0.7788459244769924,
|
| 31 |
+
"eval_f1": 0.5465686274509804,
|
| 32 |
+
"eval_loss": 0.6944684386253357,
|
| 33 |
+
"eval_precision": 0.4871122761031018,
|
| 34 |
+
"eval_recall": 0.6225572305974316,
|
| 35 |
+
"eval_runtime": 2.779,
|
| 36 |
+
"eval_samples_per_second": 400.142,
|
| 37 |
+
"eval_steps_per_second": 25.189,
|
| 38 |
+
"step": 974
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.05,
|
| 42 |
+
"learning_rate": 2.9486652977412733e-05,
|
| 43 |
+
"loss": 0.6432,
|
| 44 |
+
"step": 1000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 3.0,
|
| 48 |
+
"eval_accuracy": 0.7798001001884497,
|
| 49 |
+
"eval_f1": 0.5616201531242281,
|
| 50 |
+
"eval_loss": 0.7379429936408997,
|
| 51 |
+
"eval_precision": 0.5035429583702391,
|
| 52 |
+
"eval_recall": 0.6348408710217756,
|
| 53 |
+
"eval_runtime": 2.7688,
|
| 54 |
+
"eval_samples_per_second": 401.624,
|
| 55 |
+
"eval_steps_per_second": 25.282,
|
| 56 |
+
"step": 1461
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 3.08,
|
| 60 |
+
"learning_rate": 2.9229979466119096e-05,
|
| 61 |
+
"loss": 0.4783,
|
| 62 |
+
"step": 1500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 4.0,
|
| 66 |
+
"eval_accuracy": 0.7848572314591732,
|
| 67 |
+
"eval_f1": 0.56262278978389,
|
| 68 |
+
"eval_loss": 0.7334258556365967,
|
| 69 |
+
"eval_precision": 0.502192021043402,
|
| 70 |
+
"eval_recall": 0.6395868230039085,
|
| 71 |
+
"eval_runtime": 2.7553,
|
| 72 |
+
"eval_samples_per_second": 403.589,
|
| 73 |
+
"eval_steps_per_second": 25.406,
|
| 74 |
+
"step": 1948
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 4.11,
|
| 78 |
+
"learning_rate": 2.8973305954825462e-05,
|
| 79 |
+
"loss": 0.3609,
|
| 80 |
+
"step": 2000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 5.0,
|
| 84 |
+
"eval_accuracy": 0.7818754323608692,
|
| 85 |
+
"eval_f1": 0.5866432777847388,
|
| 86 |
+
"eval_loss": 0.8052919507026672,
|
| 87 |
+
"eval_precision": 0.5321664014548761,
|
| 88 |
+
"eval_recall": 0.6535455053042992,
|
| 89 |
+
"eval_runtime": 2.7356,
|
| 90 |
+
"eval_samples_per_second": 406.499,
|
| 91 |
+
"eval_steps_per_second": 25.589,
|
| 92 |
+
"step": 2435
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 5.13,
|
| 96 |
+
"learning_rate": 2.8716632443531828e-05,
|
| 97 |
+
"loss": 0.2735,
|
| 98 |
+
"step": 2500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 6.0,
|
| 102 |
+
"eval_accuracy": 0.7850242122086782,
|
| 103 |
+
"eval_f1": 0.5788942186316586,
|
| 104 |
+
"eval_loss": 0.8289278745651245,
|
| 105 |
+
"eval_precision": 0.5283410138248847,
|
| 106 |
+
"eval_recall": 0.6401451702959241,
|
| 107 |
+
"eval_runtime": 2.794,
|
| 108 |
+
"eval_samples_per_second": 397.99,
|
| 109 |
+
"eval_steps_per_second": 25.053,
|
| 110 |
+
"step": 2922
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 6.16,
|
| 114 |
+
"learning_rate": 2.8459958932238194e-05,
|
| 115 |
+
"loss": 0.2243,
|
| 116 |
+
"step": 3000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 7.0,
|
| 120 |
+
"eval_accuracy": 0.7811836549700627,
|
| 121 |
+
"eval_f1": 0.5892489711934157,
|
| 122 |
+
"eval_loss": 0.9322538375854492,
|
| 123 |
+
"eval_precision": 0.5462565569861707,
|
| 124 |
+
"eval_recall": 0.6395868230039085,
|
| 125 |
+
"eval_runtime": 2.7662,
|
| 126 |
+
"eval_samples_per_second": 401.992,
|
| 127 |
+
"eval_steps_per_second": 25.305,
|
| 128 |
+
"step": 3409
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 7.19,
|
| 132 |
+
"learning_rate": 2.8203285420944557e-05,
|
| 133 |
+
"loss": 0.1753,
|
| 134 |
+
"step": 3500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 8.0,
|
| 138 |
+
"eval_accuracy": 0.7798001001884497,
|
| 139 |
+
"eval_f1": 0.5883828879688872,
|
| 140 |
+
"eval_loss": 0.9913250803947449,
|
| 141 |
+
"eval_precision": 0.5342902711323764,
|
| 142 |
+
"eval_recall": 0.6546621998883305,
|
| 143 |
+
"eval_runtime": 2.787,
|
| 144 |
+
"eval_samples_per_second": 399.0,
|
| 145 |
+
"eval_steps_per_second": 25.117,
|
| 146 |
+
"step": 3896
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 8.21,
|
| 150 |
+
"learning_rate": 2.7946611909650923e-05,
|
| 151 |
+
"loss": 0.147,
|
| 152 |
+
"step": 4000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"eval_accuracy": 0.7737649388134825,
|
| 157 |
+
"eval_f1": 0.5875266658300916,
|
| 158 |
+
"eval_loss": 1.070313572883606,
|
| 159 |
+
"eval_precision": 0.5336220651926146,
|
| 160 |
+
"eval_recall": 0.6535455053042992,
|
| 161 |
+
"eval_runtime": 2.7595,
|
| 162 |
+
"eval_samples_per_second": 402.973,
|
| 163 |
+
"eval_steps_per_second": 25.367,
|
| 164 |
+
"step": 4383
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 9.24,
|
| 168 |
+
"learning_rate": 2.7689938398357292e-05,
|
| 169 |
+
"loss": 0.1204,
|
| 170 |
+
"step": 4500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 10.0,
|
| 174 |
+
"eval_accuracy": 0.785286610529329,
|
| 175 |
+
"eval_f1": 0.5879922530664945,
|
| 176 |
+
"eval_loss": 1.0797334909439087,
|
| 177 |
+
"eval_precision": 0.5469613259668509,
|
| 178 |
+
"eval_recall": 0.635678391959799,
|
| 179 |
+
"eval_runtime": 2.8881,
|
| 180 |
+
"eval_samples_per_second": 385.035,
|
| 181 |
+
"eval_steps_per_second": 24.238,
|
| 182 |
+
"step": 4870
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 10.27,
|
| 186 |
+
"learning_rate": 2.743326488706366e-05,
|
| 187 |
+
"loss": 0.1046,
|
| 188 |
+
"step": 5000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 11.0,
|
| 192 |
+
"eval_accuracy": 0.7803726056153241,
|
| 193 |
+
"eval_f1": 0.5878064110622251,
|
| 194 |
+
"eval_loss": 1.128252625465393,
|
| 195 |
+
"eval_precision": 0.5346444088726275,
|
| 196 |
+
"eval_recall": 0.6527079843662759,
|
| 197 |
+
"eval_runtime": 2.7415,
|
| 198 |
+
"eval_samples_per_second": 405.623,
|
| 199 |
+
"eval_steps_per_second": 25.534,
|
| 200 |
+
"step": 5357
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 11.29,
|
| 204 |
+
"learning_rate": 2.717659137577002e-05,
|
| 205 |
+
"loss": 0.0839,
|
| 206 |
+
"step": 5500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 12.0,
|
| 210 |
+
"eval_accuracy": 0.7841415996755803,
|
| 211 |
+
"eval_f1": 0.5948770230661399,
|
| 212 |
+
"eval_loss": 1.1799981594085693,
|
| 213 |
+
"eval_precision": 0.5472450175849941,
|
| 214 |
+
"eval_recall": 0.6515912897822446,
|
| 215 |
+
"eval_runtime": 2.7908,
|
| 216 |
+
"eval_samples_per_second": 398.457,
|
| 217 |
+
"eval_steps_per_second": 25.083,
|
| 218 |
+
"step": 5844
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 12.32,
|
| 222 |
+
"learning_rate": 2.6919917864476387e-05,
|
| 223 |
+
"loss": 0.0702,
|
| 224 |
+
"step": 6000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 13.0,
|
| 228 |
+
"eval_accuracy": 0.7898427995515375,
|
| 229 |
+
"eval_f1": 0.6095139873662497,
|
| 230 |
+
"eval_loss": 1.2153304815292358,
|
| 231 |
+
"eval_precision": 0.5662275449101797,
|
| 232 |
+
"eval_recall": 0.6599664991624791,
|
| 233 |
+
"eval_runtime": 2.747,
|
| 234 |
+
"eval_samples_per_second": 404.804,
|
| 235 |
+
"eval_steps_per_second": 25.482,
|
| 236 |
+
"step": 6331
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 13.35,
|
| 240 |
+
"learning_rate": 2.6663244353182754e-05,
|
| 241 |
+
"loss": 0.0631,
|
| 242 |
+
"step": 6500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 14.0,
|
| 246 |
+
"eval_accuracy": 0.7823763746093844,
|
| 247 |
+
"eval_f1": 0.5905834186284545,
|
| 248 |
+
"eval_loss": 1.2911711931228638,
|
| 249 |
+
"eval_precision": 0.5451110061407652,
|
| 250 |
+
"eval_recall": 0.6443327749860414,
|
| 251 |
+
"eval_runtime": 2.7671,
|
| 252 |
+
"eval_samples_per_second": 401.867,
|
| 253 |
+
"eval_steps_per_second": 25.297,
|
| 254 |
+
"step": 6818
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 14.37,
|
| 258 |
+
"learning_rate": 2.640657084188912e-05,
|
| 259 |
+
"loss": 0.0508,
|
| 260 |
+
"step": 7000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 15.0,
|
| 264 |
+
"eval_accuracy": 0.7855251544571933,
|
| 265 |
+
"eval_f1": 0.6055880441845354,
|
| 266 |
+
"eval_loss": 1.3288202285766602,
|
| 267 |
+
"eval_precision": 0.5664964745927547,
|
| 268 |
+
"eval_recall": 0.6504745951982133,
|
| 269 |
+
"eval_runtime": 2.7475,
|
| 270 |
+
"eval_samples_per_second": 404.729,
|
| 271 |
+
"eval_steps_per_second": 25.478,
|
| 272 |
+
"step": 7305
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 15.4,
|
| 276 |
+
"learning_rate": 2.6149897330595482e-05,
|
| 277 |
+
"loss": 0.0463,
|
| 278 |
+
"step": 7500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 16.0,
|
| 282 |
+
"eval_accuracy": 0.7857398439922711,
|
| 283 |
+
"eval_f1": 0.6076379806434737,
|
| 284 |
+
"eval_loss": 1.4110201597213745,
|
| 285 |
+
"eval_precision": 0.5716043307086615,
|
| 286 |
+
"eval_recall": 0.6485203796761586,
|
| 287 |
+
"eval_runtime": 2.7575,
|
| 288 |
+
"eval_samples_per_second": 403.266,
|
| 289 |
+
"eval_steps_per_second": 25.385,
|
| 290 |
+
"step": 7792
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 16.43,
|
| 294 |
+
"learning_rate": 2.589322381930185e-05,
|
| 295 |
+
"loss": 0.0387,
|
| 296 |
+
"step": 8000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 17.0,
|
| 300 |
+
"eval_accuracy": 0.7860499510984947,
|
| 301 |
+
"eval_f1": 0.606709677419355,
|
| 302 |
+
"eval_loss": 1.4022138118743896,
|
| 303 |
+
"eval_precision": 0.5640595009596929,
|
| 304 |
+
"eval_recall": 0.6563372417643775,
|
| 305 |
+
"eval_runtime": 2.7523,
|
| 306 |
+
"eval_samples_per_second": 404.029,
|
| 307 |
+
"eval_steps_per_second": 25.433,
|
| 308 |
+
"step": 8279
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 17.45,
|
| 312 |
+
"learning_rate": 2.5636550308008215e-05,
|
| 313 |
+
"loss": 0.0352,
|
| 314 |
+
"step": 8500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 18.0,
|
| 318 |
+
"eval_accuracy": 0.7891510221607309,
|
| 319 |
+
"eval_f1": 0.6005115089514067,
|
| 320 |
+
"eval_loss": 1.4306492805480957,
|
| 321 |
+
"eval_precision": 0.5540349221330817,
|
| 322 |
+
"eval_recall": 0.655499720826354,
|
| 323 |
+
"eval_runtime": 2.7318,
|
| 324 |
+
"eval_samples_per_second": 407.059,
|
| 325 |
+
"eval_steps_per_second": 25.624,
|
| 326 |
+
"step": 8766
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 18.48,
|
| 330 |
+
"learning_rate": 2.537987679671458e-05,
|
| 331 |
+
"loss": 0.031,
|
| 332 |
+
"step": 9000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 19.0,
|
| 336 |
+
"eval_accuracy": 0.7930869969704921,
|
| 337 |
+
"eval_f1": 0.6042940793754067,
|
| 338 |
+
"eval_loss": 1.4502052068710327,
|
| 339 |
+
"eval_precision": 0.5659273702169144,
|
| 340 |
+
"eval_recall": 0.6482412060301508,
|
| 341 |
+
"eval_runtime": 2.7492,
|
| 342 |
+
"eval_samples_per_second": 404.487,
|
| 343 |
+
"eval_steps_per_second": 25.462,
|
| 344 |
+
"step": 9253
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 19.51,
|
| 348 |
+
"learning_rate": 2.5123203285420943e-05,
|
| 349 |
+
"loss": 0.0286,
|
| 350 |
+
"step": 9500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 20.0,
|
| 354 |
+
"eval_accuracy": 0.7866940197037284,
|
| 355 |
+
"eval_f1": 0.5970073548059852,
|
| 356 |
+
"eval_loss": 1.5110787153244019,
|
| 357 |
+
"eval_precision": 0.5469330855018587,
|
| 358 |
+
"eval_recall": 0.6571747627024009,
|
| 359 |
+
"eval_runtime": 2.7416,
|
| 360 |
+
"eval_samples_per_second": 405.609,
|
| 361 |
+
"eval_steps_per_second": 25.533,
|
| 362 |
+
"step": 9740
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 20.53,
|
| 366 |
+
"learning_rate": 2.486652977412731e-05,
|
| 367 |
+
"loss": 0.0262,
|
| 368 |
+
"step": 10000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 21.0,
|
| 372 |
+
"eval_accuracy": 0.7825672097516758,
|
| 373 |
+
"eval_f1": 0.6085357846355877,
|
| 374 |
+
"eval_loss": 1.6086113452911377,
|
| 375 |
+
"eval_precision": 0.5745102901066204,
|
| 376 |
+
"eval_recall": 0.6468453378001117,
|
| 377 |
+
"eval_runtime": 2.7751,
|
| 378 |
+
"eval_samples_per_second": 400.703,
|
| 379 |
+
"eval_steps_per_second": 25.224,
|
| 380 |
+
"step": 10227
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 21.56,
|
| 384 |
+
"learning_rate": 2.4609856262833676e-05,
|
| 385 |
+
"loss": 0.0212,
|
| 386 |
+
"step": 10500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 22.0,
|
| 390 |
+
"eval_accuracy": 0.7844755611745903,
|
| 391 |
+
"eval_f1": 0.6177887745800236,
|
| 392 |
+
"eval_loss": 1.6133506298065186,
|
| 393 |
+
"eval_precision": 0.5789602147913108,
|
| 394 |
+
"eval_recall": 0.6621998883305416,
|
| 395 |
+
"eval_runtime": 2.7516,
|
| 396 |
+
"eval_samples_per_second": 404.131,
|
| 397 |
+
"eval_steps_per_second": 25.44,
|
| 398 |
+
"step": 10714
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 22.59,
|
| 402 |
+
"learning_rate": 2.4353182751540042e-05,
|
| 403 |
+
"loss": 0.0188,
|
| 404 |
+
"step": 11000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 23.0,
|
| 408 |
+
"eval_accuracy": 0.7830681520001909,
|
| 409 |
+
"eval_f1": 0.6066104990278678,
|
| 410 |
+
"eval_loss": 1.6032028198242188,
|
| 411 |
+
"eval_precision": 0.5661746915073796,
|
| 412 |
+
"eval_recall": 0.6532663316582915,
|
| 413 |
+
"eval_runtime": 2.7575,
|
| 414 |
+
"eval_samples_per_second": 403.261,
|
| 415 |
+
"eval_steps_per_second": 25.385,
|
| 416 |
+
"step": 11201
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 23.61,
|
| 420 |
+
"learning_rate": 2.4096509240246405e-05,
|
| 421 |
+
"loss": 0.0179,
|
| 422 |
+
"step": 11500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 24.0,
|
| 426 |
+
"eval_accuracy": 0.7821139762887336,
|
| 427 |
+
"eval_f1": 0.6070782743477138,
|
| 428 |
+
"eval_loss": 1.6523734331130981,
|
| 429 |
+
"eval_precision": 0.5649038461538461,
|
| 430 |
+
"eval_recall": 0.6560580681183696,
|
| 431 |
+
"eval_runtime": 2.7371,
|
| 432 |
+
"eval_samples_per_second": 406.274,
|
| 433 |
+
"eval_steps_per_second": 25.575,
|
| 434 |
+
"step": 11688
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 24.64,
|
| 438 |
+
"learning_rate": 2.383983572895277e-05,
|
| 439 |
+
"loss": 0.0156,
|
| 440 |
+
"step": 12000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 25.0,
|
| 444 |
+
"eval_accuracy": 0.783187423964123,
|
| 445 |
+
"eval_f1": 0.6089974293059126,
|
| 446 |
+
"eval_loss": 1.6467918157577515,
|
| 447 |
+
"eval_precision": 0.5643163411148165,
|
| 448 |
+
"eval_recall": 0.6613623673925182,
|
| 449 |
+
"eval_runtime": 2.8919,
|
| 450 |
+
"eval_samples_per_second": 384.523,
|
| 451 |
+
"eval_steps_per_second": 24.206,
|
| 452 |
+
"step": 12175
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 25.67,
|
| 456 |
+
"learning_rate": 2.3583162217659137e-05,
|
| 457 |
+
"loss": 0.0165,
|
| 458 |
+
"step": 12500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 26.0,
|
| 462 |
+
"eval_accuracy": 0.7857398439922711,
|
| 463 |
+
"eval_f1": 0.6114599686028258,
|
| 464 |
+
"eval_loss": 1.6683323383331299,
|
| 465 |
+
"eval_precision": 0.5753323485967504,
|
| 466 |
+
"eval_recall": 0.652428810720268,
|
| 467 |
+
"eval_runtime": 2.7393,
|
| 468 |
+
"eval_samples_per_second": 405.942,
|
| 469 |
+
"eval_steps_per_second": 25.554,
|
| 470 |
+
"step": 12662
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 26.69,
|
| 474 |
+
"learning_rate": 2.3326488706365506e-05,
|
| 475 |
+
"loss": 0.0141,
|
| 476 |
+
"step": 13000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 27.0,
|
| 480 |
+
"eval_accuracy": 0.7868371460604471,
|
| 481 |
+
"eval_f1": 0.6095836606441478,
|
| 482 |
+
"eval_loss": 1.688219428062439,
|
| 483 |
+
"eval_precision": 0.5739644970414202,
|
| 484 |
+
"eval_recall": 0.6499162479061976,
|
| 485 |
+
"eval_runtime": 2.7856,
|
| 486 |
+
"eval_samples_per_second": 399.201,
|
| 487 |
+
"eval_steps_per_second": 25.13,
|
| 488 |
+
"step": 13149
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 27.72,
|
| 492 |
+
"learning_rate": 2.306981519507187e-05,
|
| 493 |
+
"loss": 0.013,
|
| 494 |
+
"step": 13500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 28.0,
|
| 498 |
+
"eval_accuracy": 0.7891748765535174,
|
| 499 |
+
"eval_f1": 0.6083289404322615,
|
| 500 |
+
"eval_loss": 1.7041257619857788,
|
| 501 |
+
"eval_precision": 0.5761357963055417,
|
| 502 |
+
"eval_recall": 0.6443327749860414,
|
| 503 |
+
"eval_runtime": 2.7487,
|
| 504 |
+
"eval_samples_per_second": 404.554,
|
| 505 |
+
"eval_steps_per_second": 25.467,
|
| 506 |
+
"step": 13636
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 28.75,
|
| 510 |
+
"learning_rate": 2.2813141683778235e-05,
|
| 511 |
+
"loss": 0.0125,
|
| 512 |
+
"step": 14000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 29.0,
|
| 516 |
+
"eval_accuracy": 0.7868371460604471,
|
| 517 |
+
"eval_f1": 0.615344262295082,
|
| 518 |
+
"eval_loss": 1.7665446996688843,
|
| 519 |
+
"eval_precision": 0.5802621815483552,
|
| 520 |
+
"eval_recall": 0.6549413735343383,
|
| 521 |
+
"eval_runtime": 2.7723,
|
| 522 |
+
"eval_samples_per_second": 401.108,
|
| 523 |
+
"eval_steps_per_second": 25.25,
|
| 524 |
+
"step": 14123
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 29.77,
|
| 528 |
+
"learning_rate": 2.25564681724846e-05,
|
| 529 |
+
"loss": 0.0113,
|
| 530 |
+
"step": 14500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 30.0,
|
| 534 |
+
"eval_accuracy": 0.7934925216478614,
|
| 535 |
+
"eval_f1": 0.6155059132720105,
|
| 536 |
+
"eval_loss": 1.723854899406433,
|
| 537 |
+
"eval_precision": 0.5814299900695135,
|
| 538 |
+
"eval_recall": 0.653824678950307,
|
| 539 |
+
"eval_runtime": 2.7827,
|
| 540 |
+
"eval_samples_per_second": 399.605,
|
| 541 |
+
"eval_steps_per_second": 25.155,
|
| 542 |
+
"step": 14610
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 30.8,
|
| 546 |
+
"learning_rate": 2.2299794661190967e-05,
|
| 547 |
+
"loss": 0.0115,
|
| 548 |
+
"step": 15000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 31.0,
|
| 552 |
+
"eval_accuracy": 0.7847618138880275,
|
| 553 |
+
"eval_f1": 0.6101120667187907,
|
| 554 |
+
"eval_loss": 1.8082807064056396,
|
| 555 |
+
"eval_precision": 0.5720918866080157,
|
| 556 |
+
"eval_recall": 0.6535455053042992,
|
| 557 |
+
"eval_runtime": 2.7565,
|
| 558 |
+
"eval_samples_per_second": 403.407,
|
| 559 |
+
"eval_steps_per_second": 25.394,
|
| 560 |
+
"step": 15097
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 31.83,
|
| 564 |
+
"learning_rate": 2.204312114989733e-05,
|
| 565 |
+
"loss": 0.0121,
|
| 566 |
+
"step": 15500
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 32.0,
|
| 570 |
+
"eval_accuracy": 0.7925144915436177,
|
| 571 |
+
"eval_f1": 0.6105967078189302,
|
| 572 |
+
"eval_loss": 1.7592214345932007,
|
| 573 |
+
"eval_precision": 0.5660467334287077,
|
| 574 |
+
"eval_recall": 0.6627582356225572,
|
| 575 |
+
"eval_runtime": 2.7597,
|
| 576 |
+
"eval_samples_per_second": 402.936,
|
| 577 |
+
"eval_steps_per_second": 25.365,
|
| 578 |
+
"step": 15584
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 32.85,
|
| 582 |
+
"learning_rate": 2.1786447638603696e-05,
|
| 583 |
+
"loss": 0.0105,
|
| 584 |
+
"step": 16000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 33.0,
|
| 588 |
+
"eval_accuracy": 0.7882207008420601,
|
| 589 |
+
"eval_f1": 0.6161497186232169,
|
| 590 |
+
"eval_loss": 1.780286431312561,
|
| 591 |
+
"eval_precision": 0.5799457994579946,
|
| 592 |
+
"eval_recall": 0.6571747627024009,
|
| 593 |
+
"eval_runtime": 2.7565,
|
| 594 |
+
"eval_samples_per_second": 403.416,
|
| 595 |
+
"eval_steps_per_second": 25.395,
|
| 596 |
+
"step": 16071
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 33.88,
|
| 600 |
+
"learning_rate": 2.1529774127310062e-05,
|
| 601 |
+
"loss": 0.0089,
|
| 602 |
+
"step": 16500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 34.0,
|
| 606 |
+
"eval_accuracy": 0.7870995443810977,
|
| 607 |
+
"eval_f1": 0.6128184922511163,
|
| 608 |
+
"eval_loss": 1.8191593885421753,
|
| 609 |
+
"eval_precision": 0.5786210317460317,
|
| 610 |
+
"eval_recall": 0.6513121161362367,
|
| 611 |
+
"eval_runtime": 2.745,
|
| 612 |
+
"eval_samples_per_second": 405.095,
|
| 613 |
+
"eval_steps_per_second": 25.501,
|
| 614 |
+
"step": 16558
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 34.91,
|
| 618 |
+
"learning_rate": 2.127310061601643e-05,
|
| 619 |
+
"loss": 0.0107,
|
| 620 |
+
"step": 17000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 35.0,
|
| 624 |
+
"eval_accuracy": 0.7860260967057083,
|
| 625 |
+
"eval_f1": 0.6097277770610243,
|
| 626 |
+
"eval_loss": 1.8328877687454224,
|
| 627 |
+
"eval_precision": 0.5668025905492924,
|
| 628 |
+
"eval_recall": 0.6596873255164712,
|
| 629 |
+
"eval_runtime": 2.7678,
|
| 630 |
+
"eval_samples_per_second": 401.766,
|
| 631 |
+
"eval_steps_per_second": 25.291,
|
| 632 |
+
"step": 17045
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 35.93,
|
| 636 |
+
"learning_rate": 2.1016427104722795e-05,
|
| 637 |
+
"loss": 0.011,
|
| 638 |
+
"step": 17500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 36.0,
|
| 642 |
+
"eval_accuracy": 0.7834021134992009,
|
| 643 |
+
"eval_f1": 0.6102003642987249,
|
| 644 |
+
"eval_loss": 1.8010284900665283,
|
| 645 |
+
"eval_precision": 0.5713937621832359,
|
| 646 |
+
"eval_recall": 0.6546621998883305,
|
| 647 |
+
"eval_runtime": 2.7737,
|
| 648 |
+
"eval_samples_per_second": 400.906,
|
| 649 |
+
"eval_steps_per_second": 25.237,
|
| 650 |
+
"step": 17532
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 36.96,
|
| 654 |
+
"learning_rate": 2.0759753593429157e-05,
|
| 655 |
+
"loss": 0.0087,
|
| 656 |
+
"step": 18000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 37.0,
|
| 660 |
+
"eval_accuracy": 0.789818945158751,
|
| 661 |
+
"eval_f1": 0.6208449212024898,
|
| 662 |
+
"eval_loss": 1.8313595056533813,
|
| 663 |
+
"eval_precision": 0.5905769715293525,
|
| 664 |
+
"eval_recall": 0.6543830262423227,
|
| 665 |
+
"eval_runtime": 2.7538,
|
| 666 |
+
"eval_samples_per_second": 403.804,
|
| 667 |
+
"eval_steps_per_second": 25.419,
|
| 668 |
+
"step": 18019
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 37.99,
|
| 672 |
+
"learning_rate": 2.0503080082135523e-05,
|
| 673 |
+
"loss": 0.0075,
|
| 674 |
+
"step": 18500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 38.0,
|
| 678 |
+
"eval_accuracy": 0.7912740631187233,
|
| 679 |
+
"eval_f1": 0.6227038456455664,
|
| 680 |
+
"eval_loss": 1.8428292274475098,
|
| 681 |
+
"eval_precision": 0.5912170639899623,
|
| 682 |
+
"eval_recall": 0.6577331099944165,
|
| 683 |
+
"eval_runtime": 2.7724,
|
| 684 |
+
"eval_samples_per_second": 401.093,
|
| 685 |
+
"eval_steps_per_second": 25.249,
|
| 686 |
+
"step": 18506
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 39.0,
|
| 690 |
+
"eval_accuracy": 0.7892702941246631,
|
| 691 |
+
"eval_f1": 0.6217024041585445,
|
| 692 |
+
"eval_loss": 1.8756595849990845,
|
| 693 |
+
"eval_precision": 0.5815706297106735,
|
| 694 |
+
"eval_recall": 0.6677833612506979,
|
| 695 |
+
"eval_runtime": 2.7463,
|
| 696 |
+
"eval_samples_per_second": 404.906,
|
| 697 |
+
"eval_steps_per_second": 25.489,
|
| 698 |
+
"step": 18993
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 39.01,
|
| 702 |
+
"learning_rate": 2.024640657084189e-05,
|
| 703 |
+
"loss": 0.0079,
|
| 704 |
+
"step": 19000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 40.0,
|
| 708 |
+
"eval_accuracy": 0.7897235275876052,
|
| 709 |
+
"eval_f1": 0.6222632550778159,
|
| 710 |
+
"eval_loss": 1.8513692617416382,
|
| 711 |
+
"eval_precision": 0.58975,
|
| 712 |
+
"eval_recall": 0.65857063093244,
|
| 713 |
+
"eval_runtime": 2.8669,
|
| 714 |
+
"eval_samples_per_second": 387.875,
|
| 715 |
+
"eval_steps_per_second": 24.417,
|
| 716 |
+
"step": 19480
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 40.04,
|
| 720 |
+
"learning_rate": 1.9989733059548256e-05,
|
| 721 |
+
"loss": 0.0086,
|
| 722 |
+
"step": 19500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 41.0,
|
| 726 |
+
"eval_accuracy": 0.7896758188020324,
|
| 727 |
+
"eval_f1": 0.6242471851269966,
|
| 728 |
+
"eval_loss": 1.8782603740692139,
|
| 729 |
+
"eval_precision": 0.5877712031558185,
|
| 730 |
+
"eval_recall": 0.6655499720826354,
|
| 731 |
+
"eval_runtime": 2.7468,
|
| 732 |
+
"eval_samples_per_second": 404.838,
|
| 733 |
+
"eval_steps_per_second": 25.484,
|
| 734 |
+
"step": 19967
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 41.07,
|
| 738 |
+
"learning_rate": 1.973305954825462e-05,
|
| 739 |
+
"loss": 0.0075,
|
| 740 |
+
"step": 20000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 42.0,
|
| 744 |
+
"eval_accuracy": 0.7950669115717659,
|
| 745 |
+
"eval_f1": 0.6231997905210788,
|
| 746 |
+
"eval_loss": 1.8176987171173096,
|
| 747 |
+
"eval_precision": 0.5867850098619329,
|
| 748 |
+
"eval_recall": 0.6644332774986041,
|
| 749 |
+
"eval_runtime": 2.7491,
|
| 750 |
+
"eval_samples_per_second": 404.496,
|
| 751 |
+
"eval_steps_per_second": 25.463,
|
| 752 |
+
"step": 20454
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 42.09,
|
| 756 |
+
"learning_rate": 1.9476386036960984e-05,
|
| 757 |
+
"loss": 0.0071,
|
| 758 |
+
"step": 20500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 43.0,
|
| 762 |
+
"eval_accuracy": 0.794017318289163,
|
| 763 |
+
"eval_f1": 0.6329214826624153,
|
| 764 |
+
"eval_loss": 1.88504159450531,
|
| 765 |
+
"eval_precision": 0.6038022813688213,
|
| 766 |
+
"eval_recall": 0.6649916247906198,
|
| 767 |
+
"eval_runtime": 2.7727,
|
| 768 |
+
"eval_samples_per_second": 401.056,
|
| 769 |
+
"eval_steps_per_second": 25.246,
|
| 770 |
+
"step": 20941
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 43.12,
|
| 774 |
+
"learning_rate": 1.921971252566735e-05,
|
| 775 |
+
"loss": 0.0068,
|
| 776 |
+
"step": 21000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 44.0,
|
| 780 |
+
"eval_accuracy": 0.7917511509744519,
|
| 781 |
+
"eval_f1": 0.6311334479566195,
|
| 782 |
+
"eval_loss": 1.9209691286087036,
|
| 783 |
+
"eval_precision": 0.5996481528022116,
|
| 784 |
+
"eval_recall": 0.666108319374651,
|
| 785 |
+
"eval_runtime": 2.8074,
|
| 786 |
+
"eval_samples_per_second": 396.092,
|
| 787 |
+
"eval_steps_per_second": 24.934,
|
| 788 |
+
"step": 21428
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 44.15,
|
| 792 |
+
"learning_rate": 1.896303901437372e-05,
|
| 793 |
+
"loss": 0.006,
|
| 794 |
+
"step": 21500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 45.0,
|
| 798 |
+
"eval_accuracy": 0.7912502087259369,
|
| 799 |
+
"eval_f1": 0.6239327466176278,
|
| 800 |
+
"eval_loss": 1.9289366006851196,
|
| 801 |
+
"eval_precision": 0.5891838253535103,
|
| 802 |
+
"eval_recall": 0.663037409268565,
|
| 803 |
+
"eval_runtime": 2.7508,
|
| 804 |
+
"eval_samples_per_second": 404.251,
|
| 805 |
+
"eval_steps_per_second": 25.447,
|
| 806 |
+
"step": 21915
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 45.17,
|
| 810 |
+
"learning_rate": 1.8706365503080083e-05,
|
| 811 |
+
"loss": 0.0077,
|
| 812 |
+
"step": 22000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 46.0,
|
| 816 |
+
"eval_accuracy": 0.793802628754085,
|
| 817 |
+
"eval_f1": 0.6217957144735113,
|
| 818 |
+
"eval_loss": 1.901074767112732,
|
| 819 |
+
"eval_precision": 0.5875776397515527,
|
| 820 |
+
"eval_recall": 0.6602456728084869,
|
| 821 |
+
"eval_runtime": 2.7683,
|
| 822 |
+
"eval_samples_per_second": 401.688,
|
| 823 |
+
"eval_steps_per_second": 25.286,
|
| 824 |
+
"step": 22402
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 46.2,
|
| 828 |
+
"learning_rate": 1.844969199178645e-05,
|
| 829 |
+
"loss": 0.0047,
|
| 830 |
+
"step": 22500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 47.0,
|
| 834 |
+
"eval_accuracy": 0.7934448128622886,
|
| 835 |
+
"eval_f1": 0.6218905472636815,
|
| 836 |
+
"eval_loss": 1.909180998802185,
|
| 837 |
+
"eval_precision": 0.585552268244576,
|
| 838 |
+
"eval_recall": 0.663037409268565,
|
| 839 |
+
"eval_runtime": 2.7768,
|
| 840 |
+
"eval_samples_per_second": 400.465,
|
| 841 |
+
"eval_steps_per_second": 25.209,
|
| 842 |
+
"step": 22889
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 47.23,
|
| 846 |
+
"learning_rate": 1.8193018480492815e-05,
|
| 847 |
+
"loss": 0.0073,
|
| 848 |
+
"step": 23000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 48.0,
|
| 852 |
+
"eval_accuracy": 0.7884592447699244,
|
| 853 |
+
"eval_f1": 0.6239832065074784,
|
| 854 |
+
"eval_loss": 1.9654145240783691,
|
| 855 |
+
"eval_precision": 0.5886138613861386,
|
| 856 |
+
"eval_recall": 0.6638749302065885,
|
| 857 |
+
"eval_runtime": 2.7606,
|
| 858 |
+
"eval_samples_per_second": 402.811,
|
| 859 |
+
"eval_steps_per_second": 25.357,
|
| 860 |
+
"step": 23376
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 48.25,
|
| 864 |
+
"learning_rate": 1.793634496919918e-05,
|
| 865 |
+
"loss": 0.0058,
|
| 866 |
+
"step": 23500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 49.0,
|
| 870 |
+
"eval_accuracy": 0.7884115359843515,
|
| 871 |
+
"eval_f1": 0.6195935383011986,
|
| 872 |
+
"eval_loss": 1.948256254196167,
|
| 873 |
+
"eval_precision": 0.5808500244259892,
|
| 874 |
+
"eval_recall": 0.6638749302065885,
|
| 875 |
+
"eval_runtime": 2.7711,
|
| 876 |
+
"eval_samples_per_second": 401.285,
|
| 877 |
+
"eval_steps_per_second": 25.261,
|
| 878 |
+
"step": 23863
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 49.28,
|
| 882 |
+
"learning_rate": 1.7679671457905544e-05,
|
| 883 |
+
"loss": 0.0081,
|
| 884 |
+
"step": 24000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 50.0,
|
| 888 |
+
"eval_accuracy": 0.7899382171226832,
|
| 889 |
+
"eval_f1": 0.626782145236509,
|
| 890 |
+
"eval_loss": 1.943403959274292,
|
| 891 |
+
"eval_precision": 0.599541167473872,
|
| 892 |
+
"eval_recall": 0.6566164154103853,
|
| 893 |
+
"eval_runtime": 2.7621,
|
| 894 |
+
"eval_samples_per_second": 402.586,
|
| 895 |
+
"eval_steps_per_second": 25.343,
|
| 896 |
+
"step": 24350
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 50.31,
|
| 900 |
+
"learning_rate": 1.742299794661191e-05,
|
| 901 |
+
"loss": 0.0063,
|
| 902 |
+
"step": 24500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 51.0,
|
| 906 |
+
"eval_accuracy": 0.7902960330144796,
|
| 907 |
+
"eval_f1": 0.626861737182022,
|
| 908 |
+
"eval_loss": 1.9489706754684448,
|
| 909 |
+
"eval_precision": 0.5937578027465668,
|
| 910 |
+
"eval_recall": 0.6638749302065885,
|
| 911 |
+
"eval_runtime": 2.7691,
|
| 912 |
+
"eval_samples_per_second": 401.574,
|
| 913 |
+
"eval_steps_per_second": 25.279,
|
| 914 |
+
"step": 24837
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 51.33,
|
| 918 |
+
"learning_rate": 1.7166324435318276e-05,
|
| 919 |
+
"loss": 0.0052,
|
| 920 |
+
"step": 25000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 52.0,
|
| 924 |
+
"eval_accuracy": 0.78788673934305,
|
| 925 |
+
"eval_f1": 0.6303209346045388,
|
| 926 |
+
"eval_loss": 1.9653586149215698,
|
| 927 |
+
"eval_precision": 0.6072445019404916,
|
| 928 |
+
"eval_recall": 0.6552205471803462,
|
| 929 |
+
"eval_runtime": 2.7625,
|
| 930 |
+
"eval_samples_per_second": 402.535,
|
| 931 |
+
"eval_steps_per_second": 25.339,
|
| 932 |
+
"step": 25324
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 52.36,
|
| 936 |
+
"learning_rate": 1.6909650924024642e-05,
|
| 937 |
+
"loss": 0.007,
|
| 938 |
+
"step": 25500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 53.0,
|
| 942 |
+
"eval_accuracy": 0.7880298656997686,
|
| 943 |
+
"eval_f1": 0.6263430163151613,
|
| 944 |
+
"eval_loss": 1.9698514938354492,
|
| 945 |
+
"eval_precision": 0.5966641394996209,
|
| 946 |
+
"eval_recall": 0.6591289782244556,
|
| 947 |
+
"eval_runtime": 2.7557,
|
| 948 |
+
"eval_samples_per_second": 403.531,
|
| 949 |
+
"eval_steps_per_second": 25.402,
|
| 950 |
+
"step": 25811
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 53.39,
|
| 954 |
+
"learning_rate": 1.6652977412731005e-05,
|
| 955 |
+
"loss": 0.0047,
|
| 956 |
+
"step": 26000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 54.0,
|
| 960 |
+
"eval_accuracy": 0.7909401016197133,
|
| 961 |
+
"eval_f1": 0.6273834745762713,
|
| 962 |
+
"eval_loss": 1.971285343170166,
|
| 963 |
+
"eval_precision": 0.5967254408060454,
|
| 964 |
+
"eval_recall": 0.6613623673925182,
|
| 965 |
+
"eval_runtime": 2.7453,
|
| 966 |
+
"eval_samples_per_second": 405.054,
|
| 967 |
+
"eval_steps_per_second": 25.498,
|
| 968 |
+
"step": 26298
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 54.41,
|
| 972 |
+
"learning_rate": 1.639630390143737e-05,
|
| 973 |
+
"loss": 0.0041,
|
| 974 |
+
"step": 26500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 55.0,
|
| 978 |
+
"eval_accuracy": 0.7894611292669546,
|
| 979 |
+
"eval_f1": 0.6249177739771082,
|
| 980 |
+
"eval_loss": 1.9533612728118896,
|
| 981 |
+
"eval_precision": 0.5909430206519034,
|
| 982 |
+
"eval_recall": 0.663037409268565,
|
| 983 |
+
"eval_runtime": 2.8696,
|
| 984 |
+
"eval_samples_per_second": 387.509,
|
| 985 |
+
"eval_steps_per_second": 24.394,
|
| 986 |
+
"step": 26785
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 55.44,
|
| 990 |
+
"learning_rate": 1.6139630390143737e-05,
|
| 991 |
+
"loss": 0.0042,
|
| 992 |
+
"step": 27000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 56.0,
|
| 996 |
+
"eval_accuracy": 0.7940650270747358,
|
| 997 |
+
"eval_f1": 0.631480989098644,
|
| 998 |
+
"eval_loss": 1.9982216358184814,
|
| 999 |
+
"eval_precision": 0.6027918781725888,
|
| 1000 |
+
"eval_recall": 0.663037409268565,
|
| 1001 |
+
"eval_runtime": 2.7416,
|
| 1002 |
+
"eval_samples_per_second": 405.607,
|
| 1003 |
+
"eval_steps_per_second": 25.533,
|
| 1004 |
+
"step": 27272
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 56.47,
|
| 1008 |
+
"learning_rate": 1.5882956878850103e-05,
|
| 1009 |
+
"loss": 0.0045,
|
| 1010 |
+
"step": 27500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 57.0,
|
| 1014 |
+
"eval_accuracy": 0.792085112473462,
|
| 1015 |
+
"eval_f1": 0.6291772916387062,
|
| 1016 |
+
"eval_loss": 1.996762990951538,
|
| 1017 |
+
"eval_precision": 0.6058413026621866,
|
| 1018 |
+
"eval_recall": 0.6543830262423227,
|
| 1019 |
+
"eval_runtime": 2.7399,
|
| 1020 |
+
"eval_samples_per_second": 405.854,
|
| 1021 |
+
"eval_steps_per_second": 25.548,
|
| 1022 |
+
"step": 27759
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 57.49,
|
| 1026 |
+
"learning_rate": 1.5626283367556466e-05,
|
| 1027 |
+
"loss": 0.0045,
|
| 1028 |
+
"step": 28000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 58.0,
|
| 1032 |
+
"eval_accuracy": 0.7904630137639846,
|
| 1033 |
+
"eval_f1": 0.6297929191716767,
|
| 1034 |
+
"eval_loss": 1.985123634338379,
|
| 1035 |
+
"eval_precision": 0.603894440174225,
|
| 1036 |
+
"eval_recall": 0.6580122836404243,
|
| 1037 |
+
"eval_runtime": 2.7713,
|
| 1038 |
+
"eval_samples_per_second": 401.262,
|
| 1039 |
+
"eval_steps_per_second": 25.259,
|
| 1040 |
+
"step": 28246
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 58.52,
|
| 1044 |
+
"learning_rate": 1.5369609856262832e-05,
|
| 1045 |
+
"loss": 0.0039,
|
| 1046 |
+
"step": 28500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 59.0,
|
| 1050 |
+
"eval_accuracy": 0.789103313375158,
|
| 1051 |
+
"eval_f1": 0.6346205059920107,
|
| 1052 |
+
"eval_loss": 2.043133497238159,
|
| 1053 |
+
"eval_precision": 0.6066700610997964,
|
| 1054 |
+
"eval_recall": 0.6652707984366276,
|
| 1055 |
+
"eval_runtime": 2.7416,
|
| 1056 |
+
"eval_samples_per_second": 405.599,
|
| 1057 |
+
"eval_steps_per_second": 25.532,
|
| 1058 |
+
"step": 28733
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 59.55,
|
| 1062 |
+
"learning_rate": 1.5112936344969198e-05,
|
| 1063 |
+
"loss": 0.0048,
|
| 1064 |
+
"step": 29000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 60.0,
|
| 1068 |
+
"eval_accuracy": 0.7877913217719043,
|
| 1069 |
+
"eval_f1": 0.6211777273334224,
|
| 1070 |
+
"eval_loss": 2.0035526752471924,
|
| 1071 |
+
"eval_precision": 0.5953416943946762,
|
| 1072 |
+
"eval_recall": 0.649357900614182,
|
| 1073 |
+
"eval_runtime": 2.7429,
|
| 1074 |
+
"eval_samples_per_second": 405.406,
|
| 1075 |
+
"eval_steps_per_second": 25.52,
|
| 1076 |
+
"step": 29220
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 60.57,
|
| 1080 |
+
"learning_rate": 1.4856262833675564e-05,
|
| 1081 |
+
"loss": 0.004,
|
| 1082 |
+
"step": 29500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 61.0,
|
| 1086 |
+
"eval_accuracy": 0.791369480689869,
|
| 1087 |
+
"eval_f1": 0.6329315141078288,
|
| 1088 |
+
"eval_loss": 1.9970779418945312,
|
| 1089 |
+
"eval_precision": 0.6022183009831107,
|
| 1090 |
+
"eval_recall": 0.6669458403126745,
|
| 1091 |
+
"eval_runtime": 2.7486,
|
| 1092 |
+
"eval_samples_per_second": 404.568,
|
| 1093 |
+
"eval_steps_per_second": 25.467,
|
| 1094 |
+
"step": 29707
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 61.6,
|
| 1098 |
+
"learning_rate": 1.459958932238193e-05,
|
| 1099 |
+
"loss": 0.0032,
|
| 1100 |
+
"step": 30000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 62.0,
|
| 1104 |
+
"eval_accuracy": 0.7912024999403641,
|
| 1105 |
+
"eval_f1": 0.6301771207883874,
|
| 1106 |
+
"eval_loss": 2.007345199584961,
|
| 1107 |
+
"eval_precision": 0.6024955436720143,
|
| 1108 |
+
"eval_recall": 0.6605248464544947,
|
| 1109 |
+
"eval_runtime": 2.7495,
|
| 1110 |
+
"eval_samples_per_second": 404.438,
|
| 1111 |
+
"eval_steps_per_second": 25.459,
|
| 1112 |
+
"step": 30194
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 62.63,
|
| 1116 |
+
"learning_rate": 1.4342915811088295e-05,
|
| 1117 |
+
"loss": 0.0033,
|
| 1118 |
+
"step": 30500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 63.0,
|
| 1122 |
+
"eval_accuracy": 0.7917511509744519,
|
| 1123 |
+
"eval_f1": 0.626853813559322,
|
| 1124 |
+
"eval_loss": 2.0134215354919434,
|
| 1125 |
+
"eval_precision": 0.5962216624685138,
|
| 1126 |
+
"eval_recall": 0.6608040201005025,
|
| 1127 |
+
"eval_runtime": 2.7805,
|
| 1128 |
+
"eval_samples_per_second": 399.935,
|
| 1129 |
+
"eval_steps_per_second": 25.176,
|
| 1130 |
+
"step": 30681
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 63.66,
|
| 1134 |
+
"learning_rate": 1.4086242299794661e-05,
|
| 1135 |
+
"loss": 0.0035,
|
| 1136 |
+
"step": 31000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 64.0,
|
| 1140 |
+
"eval_accuracy": 0.7937072111829393,
|
| 1141 |
+
"eval_f1": 0.628412403922608,
|
| 1142 |
+
"eval_loss": 2.0015182495117188,
|
| 1143 |
+
"eval_precision": 0.5981331987891019,
|
| 1144 |
+
"eval_recall": 0.6619207146845337,
|
| 1145 |
+
"eval_runtime": 2.7746,
|
| 1146 |
+
"eval_samples_per_second": 400.774,
|
| 1147 |
+
"eval_steps_per_second": 25.229,
|
| 1148 |
+
"step": 31168
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 64.68,
|
| 1152 |
+
"learning_rate": 1.3829568788501027e-05,
|
| 1153 |
+
"loss": 0.0032,
|
| 1154 |
+
"step": 31500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 65.0,
|
| 1158 |
+
"eval_accuracy": 0.7939934638963765,
|
| 1159 |
+
"eval_f1": 0.6255252100840337,
|
| 1160 |
+
"eval_loss": 1.9973812103271484,
|
| 1161 |
+
"eval_precision": 0.5904809122459098,
|
| 1162 |
+
"eval_recall": 0.6649916247906198,
|
| 1163 |
+
"eval_runtime": 2.7988,
|
| 1164 |
+
"eval_samples_per_second": 397.311,
|
| 1165 |
+
"eval_steps_per_second": 25.011,
|
| 1166 |
+
"step": 31655
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 65.71,
|
| 1170 |
+
"learning_rate": 1.3572895277207393e-05,
|
| 1171 |
+
"loss": 0.0036,
|
| 1172 |
+
"step": 32000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 66.0,
|
| 1176 |
+
"eval_accuracy": 0.7891987309463038,
|
| 1177 |
+
"eval_f1": 0.6282034432908398,
|
| 1178 |
+
"eval_loss": 2.052255392074585,
|
| 1179 |
+
"eval_precision": 0.5934939160665508,
|
| 1180 |
+
"eval_recall": 0.6672250139586823,
|
| 1181 |
+
"eval_runtime": 2.7831,
|
| 1182 |
+
"eval_samples_per_second": 399.549,
|
| 1183 |
+
"eval_steps_per_second": 25.151,
|
| 1184 |
+
"step": 32142
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 66.74,
|
| 1188 |
+
"learning_rate": 1.331622176591376e-05,
|
| 1189 |
+
"loss": 0.0027,
|
| 1190 |
+
"step": 32500
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 67.0,
|
| 1194 |
+
"eval_accuracy": 0.7900813434794017,
|
| 1195 |
+
"eval_f1": 0.6333861595351294,
|
| 1196 |
+
"eval_loss": 2.068305730819702,
|
| 1197 |
+
"eval_precision": 0.6010025062656642,
|
| 1198 |
+
"eval_recall": 0.6694584031267449,
|
| 1199 |
+
"eval_runtime": 2.801,
|
| 1200 |
+
"eval_samples_per_second": 397.005,
|
| 1201 |
+
"eval_steps_per_second": 24.991,
|
| 1202 |
+
"step": 32629
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 67.76,
|
| 1206 |
+
"learning_rate": 1.3059548254620124e-05,
|
| 1207 |
+
"loss": 0.0039,
|
| 1208 |
+
"step": 33000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 68.0,
|
| 1212 |
+
"eval_accuracy": 0.7876481954151857,
|
| 1213 |
+
"eval_f1": 0.6244558765334389,
|
| 1214 |
+
"eval_loss": 2.1081135272979736,
|
| 1215 |
+
"eval_precision": 0.5918979744936234,
|
| 1216 |
+
"eval_recall": 0.6608040201005025,
|
| 1217 |
+
"eval_runtime": 2.7627,
|
| 1218 |
+
"eval_samples_per_second": 402.499,
|
| 1219 |
+
"eval_steps_per_second": 25.337,
|
| 1220 |
+
"step": 33116
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 68.79,
|
| 1224 |
+
"learning_rate": 1.280287474332649e-05,
|
| 1225 |
+
"loss": 0.0027,
|
| 1226 |
+
"step": 33500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 69.0,
|
| 1230 |
+
"eval_accuracy": 0.7922759476157535,
|
| 1231 |
+
"eval_f1": 0.6296051762841676,
|
| 1232 |
+
"eval_loss": 2.0554866790771484,
|
| 1233 |
+
"eval_precision": 0.5973440240541218,
|
| 1234 |
+
"eval_recall": 0.6655499720826354,
|
| 1235 |
+
"eval_runtime": 2.8053,
|
| 1236 |
+
"eval_samples_per_second": 396.39,
|
| 1237 |
+
"eval_steps_per_second": 24.953,
|
| 1238 |
+
"step": 33603
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 69.82,
|
| 1242 |
+
"learning_rate": 1.2546201232032854e-05,
|
| 1243 |
+
"loss": 0.003,
|
| 1244 |
+
"step": 34000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 70.0,
|
| 1248 |
+
"eval_accuracy": 0.7879821569141957,
|
| 1249 |
+
"eval_f1": 0.6243246804585585,
|
| 1250 |
+
"eval_loss": 2.100715398788452,
|
| 1251 |
+
"eval_precision": 0.59121537309708,
|
| 1252 |
+
"eval_recall": 0.6613623673925182,
|
| 1253 |
+
"eval_runtime": 2.8873,
|
| 1254 |
+
"eval_samples_per_second": 385.136,
|
| 1255 |
+
"eval_steps_per_second": 24.244,
|
| 1256 |
+
"step": 34090
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 70.84,
|
| 1260 |
+
"learning_rate": 1.228952772073922e-05,
|
| 1261 |
+
"loss": 0.0023,
|
| 1262 |
+
"step": 34500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 71.0,
|
| 1266 |
+
"eval_accuracy": 0.7936595023973665,
|
| 1267 |
+
"eval_f1": 0.6381622626477227,
|
| 1268 |
+
"eval_loss": 2.091642379760742,
|
| 1269 |
+
"eval_precision": 0.6085084831602937,
|
| 1270 |
+
"eval_recall": 0.6708542713567839,
|
| 1271 |
+
"eval_runtime": 2.7949,
|
| 1272 |
+
"eval_samples_per_second": 397.869,
|
| 1273 |
+
"eval_steps_per_second": 25.046,
|
| 1274 |
+
"step": 34577
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 71.87,
|
| 1278 |
+
"learning_rate": 1.2032854209445585e-05,
|
| 1279 |
+
"loss": 0.0016,
|
| 1280 |
+
"step": 35000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 72.0,
|
| 1284 |
+
"eval_accuracy": 0.7908208296557812,
|
| 1285 |
+
"eval_f1": 0.6252314202591908,
|
| 1286 |
+
"eval_loss": 2.1563851833343506,
|
| 1287 |
+
"eval_precision": 0.5939698492462312,
|
| 1288 |
+
"eval_recall": 0.6599664991624791,
|
| 1289 |
+
"eval_runtime": 2.7897,
|
| 1290 |
+
"eval_samples_per_second": 398.603,
|
| 1291 |
+
"eval_steps_per_second": 25.092,
|
| 1292 |
+
"step": 35064
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 72.9,
|
| 1296 |
+
"learning_rate": 1.1776180698151951e-05,
|
| 1297 |
+
"loss": 0.0028,
|
| 1298 |
+
"step": 35500
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 73.0,
|
| 1302 |
+
"eval_accuracy": 0.7862646406335727,
|
| 1303 |
+
"eval_f1": 0.6271611455721261,
|
| 1304 |
+
"eval_loss": 2.161975383758545,
|
| 1305 |
+
"eval_precision": 0.5947434292866083,
|
| 1306 |
+
"eval_recall": 0.6633165829145728,
|
| 1307 |
+
"eval_runtime": 2.7929,
|
| 1308 |
+
"eval_samples_per_second": 398.158,
|
| 1309 |
+
"eval_steps_per_second": 25.064,
|
| 1310 |
+
"step": 35551
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 73.92,
|
| 1314 |
+
"learning_rate": 1.1519507186858315e-05,
|
| 1315 |
+
"loss": 0.0028,
|
| 1316 |
+
"step": 36000
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 74.0,
|
| 1320 |
+
"eval_accuracy": 0.7891510221607309,
|
| 1321 |
+
"eval_f1": 0.6318289786223278,
|
| 1322 |
+
"eval_loss": 2.1389882564544678,
|
| 1323 |
+
"eval_precision": 0.5990990990990991,
|
| 1324 |
+
"eval_recall": 0.6683417085427136,
|
| 1325 |
+
"eval_runtime": 2.7738,
|
| 1326 |
+
"eval_samples_per_second": 400.894,
|
| 1327 |
+
"eval_steps_per_second": 25.236,
|
| 1328 |
+
"step": 36038
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 74.95,
|
| 1332 |
+
"learning_rate": 1.1262833675564683e-05,
|
| 1333 |
+
"loss": 0.0025,
|
| 1334 |
+
"step": 36500
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 75.0,
|
| 1338 |
+
"eval_accuracy": 0.7925144915436177,
|
| 1339 |
+
"eval_f1": 0.6336555011253806,
|
| 1340 |
+
"eval_loss": 2.120399236679077,
|
| 1341 |
+
"eval_precision": 0.6026189876605389,
|
| 1342 |
+
"eval_recall": 0.6680625348967058,
|
| 1343 |
+
"eval_runtime": 2.7787,
|
| 1344 |
+
"eval_samples_per_second": 400.19,
|
| 1345 |
+
"eval_steps_per_second": 25.192,
|
| 1346 |
+
"step": 36525
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 75.98,
|
| 1350 |
+
"learning_rate": 1.1006160164271048e-05,
|
| 1351 |
+
"loss": 0.0026,
|
| 1352 |
+
"step": 37000
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 76.0,
|
| 1356 |
+
"eval_accuracy": 0.7884115359843515,
|
| 1357 |
+
"eval_f1": 0.6298019407151402,
|
| 1358 |
+
"eval_loss": 2.1699509620666504,
|
| 1359 |
+
"eval_precision": 0.6011164679015478,
|
| 1360 |
+
"eval_recall": 0.6613623673925182,
|
| 1361 |
+
"eval_runtime": 2.772,
|
| 1362 |
+
"eval_samples_per_second": 401.15,
|
| 1363 |
+
"eval_steps_per_second": 25.252,
|
| 1364 |
+
"step": 37012
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 77.0,
|
| 1368 |
+
"eval_accuracy": 0.7923713651868992,
|
| 1369 |
+
"eval_f1": 0.6300172208239503,
|
| 1370 |
+
"eval_loss": 2.1477670669555664,
|
| 1371 |
+
"eval_precision": 0.5994454247542224,
|
| 1372 |
+
"eval_recall": 0.6638749302065885,
|
| 1373 |
+
"eval_runtime": 2.7809,
|
| 1374 |
+
"eval_samples_per_second": 399.866,
|
| 1375 |
+
"eval_steps_per_second": 25.171,
|
| 1376 |
+
"step": 37499
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 77.0,
|
| 1380 |
+
"learning_rate": 1.0749486652977414e-05,
|
| 1381 |
+
"loss": 0.0022,
|
| 1382 |
+
"step": 37500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 78.0,
|
| 1386 |
+
"eval_accuracy": 0.7879344481286229,
|
| 1387 |
+
"eval_f1": 0.6282473954899117,
|
| 1388 |
+
"eval_loss": 2.15474009513855,
|
| 1389 |
+
"eval_precision": 0.5953511622094476,
|
| 1390 |
+
"eval_recall": 0.6649916247906198,
|
| 1391 |
+
"eval_runtime": 2.7719,
|
| 1392 |
+
"eval_samples_per_second": 401.163,
|
| 1393 |
+
"eval_steps_per_second": 25.253,
|
| 1394 |
+
"step": 37986
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 78.03,
|
| 1398 |
+
"learning_rate": 1.0492813141683778e-05,
|
| 1399 |
+
"loss": 0.0026,
|
| 1400 |
+
"step": 38000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 79.0,
|
| 1404 |
+
"eval_accuracy": 0.78788673934305,
|
| 1405 |
+
"eval_f1": 0.6241042345276874,
|
| 1406 |
+
"eval_loss": 2.1489484310150146,
|
| 1407 |
+
"eval_precision": 0.5851453701441486,
|
| 1408 |
+
"eval_recall": 0.6686208821887214,
|
| 1409 |
+
"eval_runtime": 2.7999,
|
| 1410 |
+
"eval_samples_per_second": 397.156,
|
| 1411 |
+
"eval_steps_per_second": 25.001,
|
| 1412 |
+
"step": 38473
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 79.06,
|
| 1416 |
+
"learning_rate": 1.0236139630390144e-05,
|
| 1417 |
+
"loss": 0.0017,
|
| 1418 |
+
"step": 38500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 80.0,
|
| 1422 |
+
"eval_accuracy": 0.7869564180243792,
|
| 1423 |
+
"eval_f1": 0.6278917788524376,
|
| 1424 |
+
"eval_loss": 2.1788620948791504,
|
| 1425 |
+
"eval_precision": 0.5903170312116,
|
| 1426 |
+
"eval_recall": 0.6705750977107761,
|
| 1427 |
+
"eval_runtime": 2.7799,
|
| 1428 |
+
"eval_samples_per_second": 400.01,
|
| 1429 |
+
"eval_steps_per_second": 25.18,
|
| 1430 |
+
"step": 38960
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 80.08,
|
| 1434 |
+
"learning_rate": 9.979466119096509e-06,
|
| 1435 |
+
"loss": 0.0016,
|
| 1436 |
+
"step": 39000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 81.0,
|
| 1440 |
+
"eval_accuracy": 0.7877436129863314,
|
| 1441 |
+
"eval_f1": 0.6317747077577046,
|
| 1442 |
+
"eval_loss": 2.188213586807251,
|
| 1443 |
+
"eval_precision": 0.6026355803345159,
|
| 1444 |
+
"eval_recall": 0.6638749302065885,
|
| 1445 |
+
"eval_runtime": 2.8389,
|
| 1446 |
+
"eval_samples_per_second": 391.698,
|
| 1447 |
+
"eval_steps_per_second": 24.657,
|
| 1448 |
+
"step": 39447
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 81.11,
|
| 1452 |
+
"learning_rate": 9.722792607802875e-06,
|
| 1453 |
+
"loss": 0.0014,
|
| 1454 |
+
"step": 39500
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 82.0,
|
| 1458 |
+
"eval_accuracy": 0.7879821569141957,
|
| 1459 |
+
"eval_f1": 0.6343844834410872,
|
| 1460 |
+
"eval_loss": 2.1824533939361572,
|
| 1461 |
+
"eval_precision": 0.6014510883162372,
|
| 1462 |
+
"eval_recall": 0.6711334450027917,
|
| 1463 |
+
"eval_runtime": 2.7717,
|
| 1464 |
+
"eval_samples_per_second": 401.192,
|
| 1465 |
+
"eval_steps_per_second": 25.255,
|
| 1466 |
+
"step": 39934
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 82.14,
|
| 1470 |
+
"learning_rate": 9.46611909650924e-06,
|
| 1471 |
+
"loss": 0.0019,
|
| 1472 |
+
"step": 40000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 83.0,
|
| 1476 |
+
"eval_accuracy": 0.7902721786216932,
|
| 1477 |
+
"eval_f1": 0.6320529801324504,
|
| 1478 |
+
"eval_loss": 2.1753077507019043,
|
| 1479 |
+
"eval_precision": 0.6013104838709677,
|
| 1480 |
+
"eval_recall": 0.666108319374651,
|
| 1481 |
+
"eval_runtime": 2.7935,
|
| 1482 |
+
"eval_samples_per_second": 398.061,
|
| 1483 |
+
"eval_steps_per_second": 25.058,
|
| 1484 |
+
"step": 40421
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 83.16,
|
| 1488 |
+
"learning_rate": 9.209445585215607e-06,
|
| 1489 |
+
"loss": 0.0014,
|
| 1490 |
+
"step": 40500
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 84.0,
|
| 1494 |
+
"eval_accuracy": 0.7911070823692183,
|
| 1495 |
+
"eval_f1": 0.6313839640116432,
|
| 1496 |
+
"eval_loss": 2.1886837482452393,
|
| 1497 |
+
"eval_precision": 0.6001006036217303,
|
| 1498 |
+
"eval_recall": 0.666108319374651,
|
| 1499 |
+
"eval_runtime": 2.7541,
|
| 1500 |
+
"eval_samples_per_second": 403.769,
|
| 1501 |
+
"eval_steps_per_second": 25.417,
|
| 1502 |
+
"step": 40908
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 84.19,
|
| 1506 |
+
"learning_rate": 8.952772073921972e-06,
|
| 1507 |
+
"loss": 0.0011,
|
| 1508 |
+
"step": 41000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 85.0,
|
| 1512 |
+
"eval_accuracy": 0.7912979175115098,
|
| 1513 |
+
"eval_f1": 0.6346000531490833,
|
| 1514 |
+
"eval_loss": 2.1973717212677,
|
| 1515 |
+
"eval_precision": 0.6054766734279919,
|
| 1516 |
+
"eval_recall": 0.6666666666666666,
|
| 1517 |
+
"eval_runtime": 2.8017,
|
| 1518 |
+
"eval_samples_per_second": 396.905,
|
| 1519 |
+
"eval_steps_per_second": 24.985,
|
| 1520 |
+
"step": 41395
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 85.22,
|
| 1524 |
+
"learning_rate": 8.696098562628338e-06,
|
| 1525 |
+
"loss": 0.0019,
|
| 1526 |
+
"step": 41500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 86.0,
|
| 1530 |
+
"eval_accuracy": 0.7913217719042962,
|
| 1531 |
+
"eval_f1": 0.6334745762711864,
|
| 1532 |
+
"eval_loss": 2.1918320655822754,
|
| 1533 |
+
"eval_precision": 0.6025188916876574,
|
| 1534 |
+
"eval_recall": 0.6677833612506979,
|
| 1535 |
+
"eval_runtime": 2.8875,
|
| 1536 |
+
"eval_samples_per_second": 385.106,
|
| 1537 |
+
"eval_steps_per_second": 24.242,
|
| 1538 |
+
"step": 41882
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 86.24,
|
| 1542 |
+
"learning_rate": 8.439425051334702e-06,
|
| 1543 |
+
"loss": 0.0014,
|
| 1544 |
+
"step": 42000
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 87.0,
|
| 1548 |
+
"eval_accuracy": 0.7900813434794017,
|
| 1549 |
+
"eval_f1": 0.6352624495289367,
|
| 1550 |
+
"eval_loss": 2.196187734603882,
|
| 1551 |
+
"eval_precision": 0.6133056133056133,
|
| 1552 |
+
"eval_recall": 0.6588498045784478,
|
| 1553 |
+
"eval_runtime": 2.7929,
|
| 1554 |
+
"eval_samples_per_second": 398.148,
|
| 1555 |
+
"eval_steps_per_second": 25.063,
|
| 1556 |
+
"step": 42369
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 87.27,
|
| 1560 |
+
"learning_rate": 8.182751540041068e-06,
|
| 1561 |
+
"loss": 0.0019,
|
| 1562 |
+
"step": 42500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 88.0,
|
| 1566 |
+
"eval_accuracy": 0.7902006154433339,
|
| 1567 |
+
"eval_f1": 0.6272126816380448,
|
| 1568 |
+
"eval_loss": 2.197391986846924,
|
| 1569 |
+
"eval_precision": 0.5952858575727181,
|
| 1570 |
+
"eval_recall": 0.6627582356225572,
|
| 1571 |
+
"eval_runtime": 2.7744,
|
| 1572 |
+
"eval_samples_per_second": 400.806,
|
| 1573 |
+
"eval_steps_per_second": 25.231,
|
| 1574 |
+
"step": 42856
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 88.3,
|
| 1578 |
+
"learning_rate": 7.926078028747433e-06,
|
| 1579 |
+
"loss": 0.0009,
|
| 1580 |
+
"step": 43000
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 89.0,
|
| 1584 |
+
"eval_accuracy": 0.7918227141528112,
|
| 1585 |
+
"eval_f1": 0.6316624785081338,
|
| 1586 |
+
"eval_loss": 2.181831121444702,
|
| 1587 |
+
"eval_precision": 0.600150791656195,
|
| 1588 |
+
"eval_recall": 0.6666666666666666,
|
| 1589 |
+
"eval_runtime": 2.7914,
|
| 1590 |
+
"eval_samples_per_second": 398.372,
|
| 1591 |
+
"eval_steps_per_second": 25.077,
|
| 1592 |
+
"step": 43343
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 89.32,
|
| 1596 |
+
"learning_rate": 7.669404517453799e-06,
|
| 1597 |
+
"loss": 0.0016,
|
| 1598 |
+
"step": 43500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 90.0,
|
| 1602 |
+
"eval_accuracy": 0.7944705517521051,
|
| 1603 |
+
"eval_f1": 0.6410461702695489,
|
| 1604 |
+
"eval_loss": 2.205850124359131,
|
| 1605 |
+
"eval_precision": 0.6140081799591002,
|
| 1606 |
+
"eval_recall": 0.6705750977107761,
|
| 1607 |
+
"eval_runtime": 2.7843,
|
| 1608 |
+
"eval_samples_per_second": 399.377,
|
| 1609 |
+
"eval_steps_per_second": 25.141,
|
| 1610 |
+
"step": 43830
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 90.35,
|
| 1614 |
+
"learning_rate": 7.412731006160165e-06,
|
| 1615 |
+
"loss": 0.0013,
|
| 1616 |
+
"step": 44000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 91.0,
|
| 1620 |
+
"eval_accuracy": 0.7921805300446078,
|
| 1621 |
+
"eval_f1": 0.6387156693644687,
|
| 1622 |
+
"eval_loss": 2.2012627124786377,
|
| 1623 |
+
"eval_precision": 0.6085967130214918,
|
| 1624 |
+
"eval_recall": 0.6719709659408152,
|
| 1625 |
+
"eval_runtime": 2.7984,
|
| 1626 |
+
"eval_samples_per_second": 397.373,
|
| 1627 |
+
"eval_steps_per_second": 25.014,
|
| 1628 |
+
"step": 44317
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 91.38,
|
| 1632 |
+
"learning_rate": 7.15605749486653e-06,
|
| 1633 |
+
"loss": 0.001,
|
| 1634 |
+
"step": 44500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 92.0,
|
| 1638 |
+
"eval_accuracy": 0.7944944061448915,
|
| 1639 |
+
"eval_f1": 0.6372340425531915,
|
| 1640 |
+
"eval_loss": 2.172316551208496,
|
| 1641 |
+
"eval_precision": 0.6084306754697816,
|
| 1642 |
+
"eval_recall": 0.6689000558347292,
|
| 1643 |
+
"eval_runtime": 2.7639,
|
| 1644 |
+
"eval_samples_per_second": 402.326,
|
| 1645 |
+
"eval_steps_per_second": 25.326,
|
| 1646 |
+
"step": 44804
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 92.4,
|
| 1650 |
+
"learning_rate": 6.899383983572895e-06,
|
| 1651 |
+
"loss": 0.0012,
|
| 1652 |
+
"step": 45000
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 93.0,
|
| 1656 |
+
"eval_accuracy": 0.7965935927100976,
|
| 1657 |
+
"eval_f1": 0.6390847412531595,
|
| 1658 |
+
"eval_loss": 2.1967203617095947,
|
| 1659 |
+
"eval_precision": 0.6104193138500635,
|
| 1660 |
+
"eval_recall": 0.6705750977107761,
|
| 1661 |
+
"eval_runtime": 2.8557,
|
| 1662 |
+
"eval_samples_per_second": 389.403,
|
| 1663 |
+
"eval_steps_per_second": 24.513,
|
| 1664 |
+
"step": 45291
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 93.43,
|
| 1668 |
+
"learning_rate": 6.642710472279261e-06,
|
| 1669 |
+
"loss": 0.0023,
|
| 1670 |
+
"step": 45500
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 94.0,
|
| 1674 |
+
"eval_accuracy": 0.7938741919324444,
|
| 1675 |
+
"eval_f1": 0.6414337301056573,
|
| 1676 |
+
"eval_loss": 2.202401876449585,
|
| 1677 |
+
"eval_precision": 0.6156611039794608,
|
| 1678 |
+
"eval_recall": 0.6694584031267449,
|
| 1679 |
+
"eval_runtime": 2.795,
|
| 1680 |
+
"eval_samples_per_second": 397.859,
|
| 1681 |
+
"eval_steps_per_second": 25.045,
|
| 1682 |
+
"step": 45778
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 94.46,
|
| 1686 |
+
"learning_rate": 6.386036960985627e-06,
|
| 1687 |
+
"loss": 0.0012,
|
| 1688 |
+
"step": 46000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 95.0,
|
| 1692 |
+
"eval_accuracy": 0.7929200162209871,
|
| 1693 |
+
"eval_f1": 0.6406042936655181,
|
| 1694 |
+
"eval_loss": 2.225001096725464,
|
| 1695 |
+
"eval_precision": 0.6097376387487387,
|
| 1696 |
+
"eval_recall": 0.6747627024008933,
|
| 1697 |
+
"eval_runtime": 2.7893,
|
| 1698 |
+
"eval_samples_per_second": 398.662,
|
| 1699 |
+
"eval_steps_per_second": 25.096,
|
| 1700 |
+
"step": 46265
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 95.48,
|
| 1704 |
+
"learning_rate": 6.129363449691992e-06,
|
| 1705 |
+
"loss": 0.0015,
|
| 1706 |
+
"step": 46500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 96.0,
|
| 1710 |
+
"eval_accuracy": 0.7913933350826555,
|
| 1711 |
+
"eval_f1": 0.6457831325301205,
|
| 1712 |
+
"eval_loss": 2.193800210952759,
|
| 1713 |
+
"eval_precision": 0.6203703703703703,
|
| 1714 |
+
"eval_recall": 0.6733668341708543,
|
| 1715 |
+
"eval_runtime": 2.8031,
|
| 1716 |
+
"eval_samples_per_second": 396.697,
|
| 1717 |
+
"eval_steps_per_second": 24.972,
|
| 1718 |
+
"step": 46752
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 96.51,
|
| 1722 |
+
"learning_rate": 5.872689938398357e-06,
|
| 1723 |
+
"loss": 0.0012,
|
| 1724 |
+
"step": 47000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 97.0,
|
| 1728 |
+
"eval_accuracy": 0.7897473819803917,
|
| 1729 |
+
"eval_f1": 0.638197537332984,
|
| 1730 |
+
"eval_loss": 2.1853880882263184,
|
| 1731 |
+
"eval_precision": 0.6011846001974334,
|
| 1732 |
+
"eval_recall": 0.6800670016750419,
|
| 1733 |
+
"eval_runtime": 2.7798,
|
| 1734 |
+
"eval_samples_per_second": 400.033,
|
| 1735 |
+
"eval_steps_per_second": 25.182,
|
| 1736 |
+
"step": 47239
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 97.54,
|
| 1740 |
+
"learning_rate": 5.6160164271047226e-06,
|
| 1741 |
+
"loss": 0.0008,
|
| 1742 |
+
"step": 47500
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 98.0,
|
| 1746 |
+
"eval_accuracy": 0.7930392881849192,
|
| 1747 |
+
"eval_f1": 0.6455238859895625,
|
| 1748 |
+
"eval_loss": 2.200537919998169,
|
| 1749 |
+
"eval_precision": 0.6198920585967618,
|
| 1750 |
+
"eval_recall": 0.6733668341708543,
|
| 1751 |
+
"eval_runtime": 2.7585,
|
| 1752 |
+
"eval_samples_per_second": 403.112,
|
| 1753 |
+
"eval_steps_per_second": 25.376,
|
| 1754 |
+
"step": 47726
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 98.56,
|
| 1758 |
+
"learning_rate": 5.359342915811089e-06,
|
| 1759 |
+
"loss": 0.0008,
|
| 1760 |
+
"step": 48000
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 99.0,
|
| 1764 |
+
"eval_accuracy": 0.7896042556236731,
|
| 1765 |
+
"eval_f1": 0.6393529567753912,
|
| 1766 |
+
"eval_loss": 2.1999213695526123,
|
| 1767 |
+
"eval_precision": 0.6088383838383838,
|
| 1768 |
+
"eval_recall": 0.6730876605248465,
|
| 1769 |
+
"eval_runtime": 2.8138,
|
| 1770 |
+
"eval_samples_per_second": 395.198,
|
| 1771 |
+
"eval_steps_per_second": 24.878,
|
| 1772 |
+
"step": 48213
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 99.59,
|
| 1776 |
+
"learning_rate": 5.102669404517454e-06,
|
| 1777 |
+
"loss": 0.0011,
|
| 1778 |
+
"step": 48500
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 100.0,
|
| 1782 |
+
"eval_accuracy": 0.7930631425777057,
|
| 1783 |
+
"eval_f1": 0.6375963839404414,
|
| 1784 |
+
"eval_loss": 2.2227866649627686,
|
| 1785 |
+
"eval_precision": 0.6086294416243655,
|
| 1786 |
+
"eval_recall": 0.6694584031267449,
|
| 1787 |
+
"eval_runtime": 2.7694,
|
| 1788 |
+
"eval_samples_per_second": 401.531,
|
| 1789 |
+
"eval_steps_per_second": 25.276,
|
| 1790 |
+
"step": 48700
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 100.62,
|
| 1794 |
+
"learning_rate": 4.845995893223819e-06,
|
| 1795 |
+
"loss": 0.0006,
|
| 1796 |
+
"step": 49000
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 101.0,
|
| 1800 |
+
"eval_accuracy": 0.7925144915436177,
|
| 1801 |
+
"eval_f1": 0.6429421881201217,
|
| 1802 |
+
"eval_loss": 2.2299931049346924,
|
| 1803 |
+
"eval_precision": 0.6110133266281117,
|
| 1804 |
+
"eval_recall": 0.678391959798995,
|
| 1805 |
+
"eval_runtime": 2.9094,
|
| 1806 |
+
"eval_samples_per_second": 382.209,
|
| 1807 |
+
"eval_steps_per_second": 24.06,
|
| 1808 |
+
"step": 49187
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 101.64,
|
| 1812 |
+
"learning_rate": 4.5893223819301845e-06,
|
| 1813 |
+
"loss": 0.001,
|
| 1814 |
+
"step": 49500
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 102.0,
|
| 1818 |
+
"eval_accuracy": 0.7917034421888791,
|
| 1819 |
+
"eval_f1": 0.6384889710738344,
|
| 1820 |
+
"eval_loss": 2.2193572521209717,
|
| 1821 |
+
"eval_precision": 0.6059162697417899,
|
| 1822 |
+
"eval_recall": 0.6747627024008933,
|
| 1823 |
+
"eval_runtime": 2.7802,
|
| 1824 |
+
"eval_samples_per_second": 399.973,
|
| 1825 |
+
"eval_steps_per_second": 25.178,
|
| 1826 |
+
"step": 49674
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 102.67,
|
| 1830 |
+
"learning_rate": 4.332648870636551e-06,
|
| 1831 |
+
"loss": 0.0007,
|
| 1832 |
+
"step": 50000
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 103.0,
|
| 1836 |
+
"eval_accuracy": 0.794685241287183,
|
| 1837 |
+
"eval_f1": 0.6422018348623854,
|
| 1838 |
+
"eval_loss": 2.2047975063323975,
|
| 1839 |
+
"eval_precision": 0.6130997715156131,
|
| 1840 |
+
"eval_recall": 0.6742043551088778,
|
| 1841 |
+
"eval_runtime": 2.8095,
|
| 1842 |
+
"eval_samples_per_second": 395.8,
|
| 1843 |
+
"eval_steps_per_second": 24.915,
|
| 1844 |
+
"step": 50161
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 103.7,
|
| 1848 |
+
"learning_rate": 4.075975359342916e-06,
|
| 1849 |
+
"loss": 0.0003,
|
| 1850 |
+
"step": 50500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 104.0,
|
| 1854 |
+
"eval_accuracy": 0.7937549199685122,
|
| 1855 |
+
"eval_f1": 0.6471614548382768,
|
| 1856 |
+
"eval_loss": 2.2270121574401855,
|
| 1857 |
+
"eval_precision": 0.6231584388730939,
|
| 1858 |
+
"eval_recall": 0.6730876605248465,
|
| 1859 |
+
"eval_runtime": 2.7614,
|
| 1860 |
+
"eval_samples_per_second": 402.698,
|
| 1861 |
+
"eval_steps_per_second": 25.35,
|
| 1862 |
+
"step": 50648
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 104.72,
|
| 1866 |
+
"learning_rate": 3.819301848049281e-06,
|
| 1867 |
+
"loss": 0.0008,
|
| 1868 |
+
"step": 51000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 105.0,
|
| 1872 |
+
"eval_accuracy": 0.7951861835356981,
|
| 1873 |
+
"eval_f1": 0.6451182048884735,
|
| 1874 |
+
"eval_loss": 2.228421211242676,
|
| 1875 |
+
"eval_precision": 0.618437900128041,
|
| 1876 |
+
"eval_recall": 0.6742043551088778,
|
| 1877 |
+
"eval_runtime": 2.7968,
|
| 1878 |
+
"eval_samples_per_second": 397.592,
|
| 1879 |
+
"eval_steps_per_second": 25.028,
|
| 1880 |
+
"step": 51135
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 105.75,
|
| 1884 |
+
"learning_rate": 3.5626283367556473e-06,
|
| 1885 |
+
"loss": 0.0005,
|
| 1886 |
+
"step": 51500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 106.0,
|
| 1890 |
+
"eval_accuracy": 0.7920612580806755,
|
| 1891 |
+
"eval_f1": 0.6393963463065925,
|
| 1892 |
+
"eval_loss": 2.2278361320495605,
|
| 1893 |
+
"eval_precision": 0.6080060422960725,
|
| 1894 |
+
"eval_recall": 0.6742043551088778,
|
| 1895 |
+
"eval_runtime": 2.7765,
|
| 1896 |
+
"eval_samples_per_second": 400.508,
|
| 1897 |
+
"eval_steps_per_second": 25.212,
|
| 1898 |
+
"step": 51622
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 106.78,
|
| 1902 |
+
"learning_rate": 3.3059548254620125e-06,
|
| 1903 |
+
"loss": 0.0004,
|
| 1904 |
+
"step": 52000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 107.0,
|
| 1908 |
+
"eval_accuracy": 0.7925622003291907,
|
| 1909 |
+
"eval_f1": 0.6443971253659835,
|
| 1910 |
+
"eval_loss": 2.257086992263794,
|
| 1911 |
+
"eval_precision": 0.6157171922685656,
|
| 1912 |
+
"eval_recall": 0.6758793969849246,
|
| 1913 |
+
"eval_runtime": 2.7854,
|
| 1914 |
+
"eval_samples_per_second": 399.221,
|
| 1915 |
+
"eval_steps_per_second": 25.131,
|
| 1916 |
+
"step": 52109
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 107.8,
|
| 1920 |
+
"learning_rate": 3.049281314168378e-06,
|
| 1921 |
+
"loss": 0.0006,
|
| 1922 |
+
"step": 52500
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 108.0,
|
| 1926 |
+
"eval_accuracy": 0.7924906371508313,
|
| 1927 |
+
"eval_f1": 0.6378807947019867,
|
| 1928 |
+
"eval_loss": 2.256169319152832,
|
| 1929 |
+
"eval_precision": 0.6068548387096774,
|
| 1930 |
+
"eval_recall": 0.672250139586823,
|
| 1931 |
+
"eval_runtime": 2.7803,
|
| 1932 |
+
"eval_samples_per_second": 399.96,
|
| 1933 |
+
"eval_steps_per_second": 25.177,
|
| 1934 |
+
"step": 52596
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 108.83,
|
| 1938 |
+
"learning_rate": 2.7926078028747435e-06,
|
| 1939 |
+
"loss": 0.0005,
|
| 1940 |
+
"step": 53000
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 109.0,
|
| 1944 |
+
"eval_accuracy": 0.7949714940006202,
|
| 1945 |
+
"eval_f1": 0.6433155080213904,
|
| 1946 |
+
"eval_loss": 2.225450277328491,
|
| 1947 |
+
"eval_precision": 0.6172396100564392,
|
| 1948 |
+
"eval_recall": 0.6716917922948074,
|
| 1949 |
+
"eval_runtime": 2.7966,
|
| 1950 |
+
"eval_samples_per_second": 397.621,
|
| 1951 |
+
"eval_steps_per_second": 25.03,
|
| 1952 |
+
"step": 53083
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 109.86,
|
| 1956 |
+
"learning_rate": 2.5359342915811088e-06,
|
| 1957 |
+
"loss": 0.0006,
|
| 1958 |
+
"step": 53500
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 110.0,
|
| 1962 |
+
"eval_accuracy": 0.7930869969704921,
|
| 1963 |
+
"eval_f1": 0.6414944356120826,
|
| 1964 |
+
"eval_loss": 2.242906332015991,
|
| 1965 |
+
"eval_precision": 0.6104387291981845,
|
| 1966 |
+
"eval_recall": 0.6758793969849246,
|
| 1967 |
+
"eval_runtime": 2.7792,
|
| 1968 |
+
"eval_samples_per_second": 400.111,
|
| 1969 |
+
"eval_steps_per_second": 25.187,
|
| 1970 |
+
"step": 53570
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 110.88,
|
| 1974 |
+
"learning_rate": 2.2792607802874745e-06,
|
| 1975 |
+
"loss": 0.0004,
|
| 1976 |
+
"step": 54000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 111.0,
|
| 1980 |
+
"eval_accuracy": 0.7927291810786956,
|
| 1981 |
+
"eval_f1": 0.6417751793781558,
|
| 1982 |
+
"eval_loss": 2.241628885269165,
|
| 1983 |
+
"eval_precision": 0.6123225152129818,
|
| 1984 |
+
"eval_recall": 0.6742043551088778,
|
| 1985 |
+
"eval_runtime": 2.7871,
|
| 1986 |
+
"eval_samples_per_second": 398.979,
|
| 1987 |
+
"eval_steps_per_second": 25.116,
|
| 1988 |
+
"step": 54057
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 111.91,
|
| 1992 |
+
"learning_rate": 2.0225872689938397e-06,
|
| 1993 |
+
"loss": 0.0004,
|
| 1994 |
+
"step": 54500
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 112.0,
|
| 1998 |
+
"eval_accuracy": 0.7938980463252308,
|
| 1999 |
+
"eval_f1": 0.6393595730486992,
|
| 2000 |
+
"eval_loss": 2.262876033782959,
|
| 2001 |
+
"eval_precision": 0.612317914643496,
|
| 2002 |
+
"eval_recall": 0.6689000558347292,
|
| 2003 |
+
"eval_runtime": 2.7846,
|
| 2004 |
+
"eval_samples_per_second": 399.344,
|
| 2005 |
+
"eval_steps_per_second": 25.139,
|
| 2006 |
+
"step": 54544
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 112.94,
|
| 2010 |
+
"learning_rate": 1.7659137577002054e-06,
|
| 2011 |
+
"loss": 0.0004,
|
| 2012 |
+
"step": 55000
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 113.0,
|
| 2016 |
+
"eval_accuracy": 0.7931824145416378,
|
| 2017 |
+
"eval_f1": 0.6427336790320436,
|
| 2018 |
+
"eval_loss": 2.2645206451416016,
|
| 2019 |
+
"eval_precision": 0.6136075145976136,
|
| 2020 |
+
"eval_recall": 0.6747627024008933,
|
| 2021 |
+
"eval_runtime": 2.7845,
|
| 2022 |
+
"eval_samples_per_second": 399.355,
|
| 2023 |
+
"eval_steps_per_second": 25.139,
|
| 2024 |
+
"step": 55031
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 113.96,
|
| 2028 |
+
"learning_rate": 1.509240246406571e-06,
|
| 2029 |
+
"loss": 0.0003,
|
| 2030 |
+
"step": 55500
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 114.0,
|
| 2034 |
+
"eval_accuracy": 0.7943751341809594,
|
| 2035 |
+
"eval_f1": 0.6461373677868524,
|
| 2036 |
+
"eval_loss": 2.2760562896728516,
|
| 2037 |
+
"eval_precision": 0.6207872395163365,
|
| 2038 |
+
"eval_recall": 0.6736460078168621,
|
| 2039 |
+
"eval_runtime": 2.7804,
|
| 2040 |
+
"eval_samples_per_second": 399.941,
|
| 2041 |
+
"eval_steps_per_second": 25.176,
|
| 2042 |
+
"step": 55518
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 114.99,
|
| 2046 |
+
"learning_rate": 1.2525667351129362e-06,
|
| 2047 |
+
"loss": 0.0004,
|
| 2048 |
+
"step": 56000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 115.0,
|
| 2052 |
+
"eval_accuracy": 0.7936595023973665,
|
| 2053 |
+
"eval_f1": 0.6438374417055296,
|
| 2054 |
+
"eval_loss": 2.2683961391448975,
|
| 2055 |
+
"eval_precision": 0.6158552128473107,
|
| 2056 |
+
"eval_recall": 0.6744835287548855,
|
| 2057 |
+
"eval_runtime": 2.8007,
|
| 2058 |
+
"eval_samples_per_second": 397.04,
|
| 2059 |
+
"eval_steps_per_second": 24.994,
|
| 2060 |
+
"step": 56005
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 116.0,
|
| 2064 |
+
"eval_accuracy": 0.7926337635075499,
|
| 2065 |
+
"eval_f1": 0.6435524736631552,
|
| 2066 |
+
"eval_loss": 2.274099349975586,
|
| 2067 |
+
"eval_precision": 0.6160326780699515,
|
| 2068 |
+
"eval_recall": 0.6736460078168621,
|
| 2069 |
+
"eval_runtime": 2.8997,
|
| 2070 |
+
"eval_samples_per_second": 383.493,
|
| 2071 |
+
"eval_steps_per_second": 24.141,
|
| 2072 |
+
"step": 56492
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 116.02,
|
| 2076 |
+
"learning_rate": 9.958932238193019e-07,
|
| 2077 |
+
"loss": 0.0003,
|
| 2078 |
+
"step": 56500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 117.0,
|
| 2082 |
+
"eval_accuracy": 0.7939457551108037,
|
| 2083 |
+
"eval_f1": 0.6435524736631552,
|
| 2084 |
+
"eval_loss": 2.257563591003418,
|
| 2085 |
+
"eval_precision": 0.6160326780699515,
|
| 2086 |
+
"eval_recall": 0.6736460078168621,
|
| 2087 |
+
"eval_runtime": 2.7639,
|
| 2088 |
+
"eval_samples_per_second": 402.33,
|
| 2089 |
+
"eval_steps_per_second": 25.327,
|
| 2090 |
+
"step": 56979
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 117.04,
|
| 2094 |
+
"learning_rate": 7.392197125256674e-07,
|
| 2095 |
+
"loss": 0.001,
|
| 2096 |
+
"step": 57000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 118.0,
|
| 2100 |
+
"eval_accuracy": 0.7942558622170273,
|
| 2101 |
+
"eval_f1": 0.643104827954121,
|
| 2102 |
+
"eval_loss": 2.2542884349823,
|
| 2103 |
+
"eval_precision": 0.615679264555669,
|
| 2104 |
+
"eval_recall": 0.6730876605248465,
|
| 2105 |
+
"eval_runtime": 2.7671,
|
| 2106 |
+
"eval_samples_per_second": 401.861,
|
| 2107 |
+
"eval_steps_per_second": 25.297,
|
| 2108 |
+
"step": 57466
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 118.07,
|
| 2112 |
+
"learning_rate": 4.82546201232033e-07,
|
| 2113 |
+
"loss": 0.0003,
|
| 2114 |
+
"step": 57500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 119.0,
|
| 2118 |
+
"eval_accuracy": 0.794685241287183,
|
| 2119 |
+
"eval_f1": 0.6438191758901186,
|
| 2120 |
+
"eval_loss": 2.2540647983551025,
|
| 2121 |
+
"eval_precision": 0.6162879754914475,
|
| 2122 |
+
"eval_recall": 0.67392518146287,
|
| 2123 |
+
"eval_runtime": 2.7786,
|
| 2124 |
+
"eval_samples_per_second": 400.208,
|
| 2125 |
+
"eval_steps_per_second": 25.193,
|
| 2126 |
+
"step": 57953
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 119.1,
|
| 2130 |
+
"learning_rate": 2.2587268993839835e-07,
|
| 2131 |
+
"loss": 0.0005,
|
| 2132 |
+
"step": 58000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 120.0,
|
| 2136 |
+
"eval_accuracy": 0.7944228429665323,
|
| 2137 |
+
"eval_f1": 0.6433529097704217,
|
| 2138 |
+
"eval_loss": 2.2551677227020264,
|
| 2139 |
+
"eval_precision": 0.6163682864450127,
|
| 2140 |
+
"eval_recall": 0.6728084868788387,
|
| 2141 |
+
"eval_runtime": 2.7762,
|
| 2142 |
+
"eval_samples_per_second": 400.553,
|
| 2143 |
+
"eval_steps_per_second": 25.215,
|
| 2144 |
+
"step": 58440
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 120.0,
|
| 2148 |
+
"step": 58440,
|
| 2149 |
+
"total_flos": 1.220726808511488e+17,
|
| 2150 |
+
"train_loss": 0.039664112105659635,
|
| 2151 |
+
"train_runtime": 7090.7344,
|
| 2152 |
+
"train_samples_per_second": 131.749,
|
| 2153 |
+
"train_steps_per_second": 8.242
|
| 2154 |
+
}
|
| 2155 |
+
],
|
| 2156 |
+
"max_steps": 58440,
|
| 2157 |
+
"num_train_epochs": 120,
|
| 2158 |
+
"total_flos": 1.220726808511488e+17,
|
| 2159 |
+
"trial_name": null,
|
| 2160 |
+
"trial_params": null
|
| 2161 |
+
}
|