Commit
•
5abbe53
1
Parent(s):
339a1fb
End of training
Browse files- README.md +2 -2
- all_results.json +8 -0
- test_results.json +8 -0
- trainer_state.json +3112 -0
README.md
CHANGED
|
@@ -17,8 +17,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 17 |
|
| 18 |
This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingface.co/MCG-NJU/videomae-base) on an unknown dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss:
|
| 21 |
-
- Accuracy: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
|
|
| 17 |
|
| 18 |
This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingface.co/MCG-NJU/videomae-base) on an unknown dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 1.6457
|
| 21 |
+
- Accuracy: 0.5962
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.17,
|
| 3 |
+
"eval_accuracy": 0.5961538461538461,
|
| 4 |
+
"eval_loss": 1.6456845998764038,
|
| 5 |
+
"eval_runtime": 40.3879,
|
| 6 |
+
"eval_samples_per_second": 2.575,
|
| 7 |
+
"eval_steps_per_second": 0.223
|
| 8 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.17,
|
| 3 |
+
"eval_accuracy": 0.5961538461538461,
|
| 4 |
+
"eval_loss": 1.6456845998764038,
|
| 5 |
+
"eval_runtime": 40.3879,
|
| 6 |
+
"eval_samples_per_second": 2.575,
|
| 7 |
+
"eval_steps_per_second": 0.223
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8643669149353195,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-cctv-brawl_extended/checkpoint-4190",
|
| 4 |
+
"epoch": 5.165671047391477,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 5022,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 1.988071570576541e-07,
|
| 14 |
+
"loss": 0.7219,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.0,
|
| 19 |
+
"learning_rate": 3.976143141153082e-07,
|
| 20 |
+
"loss": 0.7194,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 5.964214711729623e-07,
|
| 26 |
+
"loss": 0.7111,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 7.952286282306164e-07,
|
| 32 |
+
"loss": 0.7259,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.01,
|
| 37 |
+
"learning_rate": 9.940357852882705e-07,
|
| 38 |
+
"loss": 0.7108,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.01,
|
| 43 |
+
"learning_rate": 1.1928429423459246e-06,
|
| 44 |
+
"loss": 0.7128,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.01,
|
| 49 |
+
"learning_rate": 1.3916500994035787e-06,
|
| 50 |
+
"loss": 0.7096,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"learning_rate": 1.5904572564612327e-06,
|
| 56 |
+
"loss": 0.6754,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.02,
|
| 61 |
+
"learning_rate": 1.7892644135188868e-06,
|
| 62 |
+
"loss": 0.6289,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.02,
|
| 67 |
+
"learning_rate": 1.988071570576541e-06,
|
| 68 |
+
"loss": 0.6735,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.02,
|
| 73 |
+
"learning_rate": 2.186878727634195e-06,
|
| 74 |
+
"loss": 0.6414,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.02,
|
| 79 |
+
"learning_rate": 2.3856858846918492e-06,
|
| 80 |
+
"loss": 0.6591,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.03,
|
| 85 |
+
"learning_rate": 2.5844930417495035e-06,
|
| 86 |
+
"loss": 0.6247,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.03,
|
| 91 |
+
"learning_rate": 2.7833001988071574e-06,
|
| 92 |
+
"loss": 0.5978,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.03,
|
| 97 |
+
"learning_rate": 2.9821073558648112e-06,
|
| 98 |
+
"loss": 0.6092,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.03,
|
| 103 |
+
"learning_rate": 3.1809145129224655e-06,
|
| 104 |
+
"loss": 0.5812,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.03,
|
| 109 |
+
"learning_rate": 3.3797216699801194e-06,
|
| 110 |
+
"loss": 0.5962,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.04,
|
| 115 |
+
"learning_rate": 3.5785288270377736e-06,
|
| 116 |
+
"loss": 0.5872,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.04,
|
| 121 |
+
"learning_rate": 3.7773359840954275e-06,
|
| 122 |
+
"loss": 0.5884,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.04,
|
| 127 |
+
"learning_rate": 3.976143141153082e-06,
|
| 128 |
+
"loss": 0.6023,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.04,
|
| 133 |
+
"learning_rate": 4.174950298210736e-06,
|
| 134 |
+
"loss": 0.5678,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.04,
|
| 139 |
+
"learning_rate": 4.37375745526839e-06,
|
| 140 |
+
"loss": 0.5665,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.05,
|
| 145 |
+
"learning_rate": 4.572564612326044e-06,
|
| 146 |
+
"loss": 0.5642,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.05,
|
| 151 |
+
"learning_rate": 4.7713717693836985e-06,
|
| 152 |
+
"loss": 0.5253,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.05,
|
| 157 |
+
"learning_rate": 4.970178926441353e-06,
|
| 158 |
+
"loss": 0.4764,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.05,
|
| 163 |
+
"learning_rate": 5.168986083499007e-06,
|
| 164 |
+
"loss": 0.5652,
|
| 165 |
+
"step": 260
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.05,
|
| 169 |
+
"learning_rate": 5.3677932405566604e-06,
|
| 170 |
+
"loss": 0.5636,
|
| 171 |
+
"step": 270
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.06,
|
| 175 |
+
"learning_rate": 5.566600397614315e-06,
|
| 176 |
+
"loss": 0.5912,
|
| 177 |
+
"step": 280
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.06,
|
| 181 |
+
"learning_rate": 5.765407554671969e-06,
|
| 182 |
+
"loss": 0.5503,
|
| 183 |
+
"step": 290
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.06,
|
| 187 |
+
"learning_rate": 5.9642147117296224e-06,
|
| 188 |
+
"loss": 0.5331,
|
| 189 |
+
"step": 300
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.06,
|
| 193 |
+
"learning_rate": 6.163021868787277e-06,
|
| 194 |
+
"loss": 0.5082,
|
| 195 |
+
"step": 310
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.06,
|
| 199 |
+
"learning_rate": 6.361829025844931e-06,
|
| 200 |
+
"loss": 0.4525,
|
| 201 |
+
"step": 320
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.07,
|
| 205 |
+
"learning_rate": 6.560636182902585e-06,
|
| 206 |
+
"loss": 0.4899,
|
| 207 |
+
"step": 330
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.07,
|
| 211 |
+
"learning_rate": 6.759443339960239e-06,
|
| 212 |
+
"loss": 0.4781,
|
| 213 |
+
"step": 340
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.07,
|
| 217 |
+
"learning_rate": 6.958250497017893e-06,
|
| 218 |
+
"loss": 0.4776,
|
| 219 |
+
"step": 350
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.07,
|
| 223 |
+
"learning_rate": 7.157057654075547e-06,
|
| 224 |
+
"loss": 0.4492,
|
| 225 |
+
"step": 360
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.07,
|
| 229 |
+
"learning_rate": 7.355864811133201e-06,
|
| 230 |
+
"loss": 0.4572,
|
| 231 |
+
"step": 370
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.08,
|
| 235 |
+
"learning_rate": 7.554671968190855e-06,
|
| 236 |
+
"loss": 0.3901,
|
| 237 |
+
"step": 380
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.08,
|
| 241 |
+
"learning_rate": 7.75347912524851e-06,
|
| 242 |
+
"loss": 0.4159,
|
| 243 |
+
"step": 390
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.08,
|
| 247 |
+
"learning_rate": 7.952286282306164e-06,
|
| 248 |
+
"loss": 0.3229,
|
| 249 |
+
"step": 400
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.08,
|
| 253 |
+
"learning_rate": 8.151093439363818e-06,
|
| 254 |
+
"loss": 0.4575,
|
| 255 |
+
"step": 410
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.08,
|
| 259 |
+
"learning_rate": 8.349900596421471e-06,
|
| 260 |
+
"loss": 0.3253,
|
| 261 |
+
"step": 420
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.09,
|
| 265 |
+
"learning_rate": 8.548707753479126e-06,
|
| 266 |
+
"loss": 0.4011,
|
| 267 |
+
"step": 430
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.09,
|
| 271 |
+
"learning_rate": 8.74751491053678e-06,
|
| 272 |
+
"loss": 0.3972,
|
| 273 |
+
"step": 440
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.09,
|
| 277 |
+
"learning_rate": 8.946322067594433e-06,
|
| 278 |
+
"loss": 0.4415,
|
| 279 |
+
"step": 450
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.09,
|
| 283 |
+
"learning_rate": 9.145129224652088e-06,
|
| 284 |
+
"loss": 0.3809,
|
| 285 |
+
"step": 460
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.09,
|
| 289 |
+
"learning_rate": 9.343936381709743e-06,
|
| 290 |
+
"loss": 0.4083,
|
| 291 |
+
"step": 470
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.1,
|
| 295 |
+
"learning_rate": 9.542743538767397e-06,
|
| 296 |
+
"loss": 0.3875,
|
| 297 |
+
"step": 480
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.1,
|
| 301 |
+
"learning_rate": 9.74155069582505e-06,
|
| 302 |
+
"loss": 0.4106,
|
| 303 |
+
"step": 490
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.1,
|
| 307 |
+
"learning_rate": 9.940357852882705e-06,
|
| 308 |
+
"loss": 0.4624,
|
| 309 |
+
"step": 500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.1,
|
| 313 |
+
"learning_rate": 9.984509847311354e-06,
|
| 314 |
+
"loss": 0.4058,
|
| 315 |
+
"step": 510
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.1,
|
| 319 |
+
"learning_rate": 9.962381057756142e-06,
|
| 320 |
+
"loss": 0.4908,
|
| 321 |
+
"step": 520
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.11,
|
| 325 |
+
"learning_rate": 9.94025226820093e-06,
|
| 326 |
+
"loss": 0.3859,
|
| 327 |
+
"step": 530
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.11,
|
| 331 |
+
"learning_rate": 9.918123478645718e-06,
|
| 332 |
+
"loss": 0.4327,
|
| 333 |
+
"step": 540
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.11,
|
| 337 |
+
"learning_rate": 9.895994689090508e-06,
|
| 338 |
+
"loss": 0.4058,
|
| 339 |
+
"step": 550
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.11,
|
| 343 |
+
"learning_rate": 9.873865899535297e-06,
|
| 344 |
+
"loss": 0.4407,
|
| 345 |
+
"step": 560
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.11,
|
| 349 |
+
"learning_rate": 9.851737109980085e-06,
|
| 350 |
+
"loss": 0.34,
|
| 351 |
+
"step": 570
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.12,
|
| 355 |
+
"learning_rate": 9.829608320424873e-06,
|
| 356 |
+
"loss": 0.3745,
|
| 357 |
+
"step": 580
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.12,
|
| 361 |
+
"learning_rate": 9.807479530869663e-06,
|
| 362 |
+
"loss": 0.3693,
|
| 363 |
+
"step": 590
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.12,
|
| 367 |
+
"learning_rate": 9.785350741314451e-06,
|
| 368 |
+
"loss": 0.3525,
|
| 369 |
+
"step": 600
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.12,
|
| 373 |
+
"learning_rate": 9.76322195175924e-06,
|
| 374 |
+
"loss": 0.3235,
|
| 375 |
+
"step": 610
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.12,
|
| 379 |
+
"learning_rate": 9.741093162204028e-06,
|
| 380 |
+
"loss": 0.2694,
|
| 381 |
+
"step": 620
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.13,
|
| 385 |
+
"learning_rate": 9.718964372648817e-06,
|
| 386 |
+
"loss": 0.3649,
|
| 387 |
+
"step": 630
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.13,
|
| 391 |
+
"learning_rate": 9.696835583093606e-06,
|
| 392 |
+
"loss": 0.3023,
|
| 393 |
+
"step": 640
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.13,
|
| 397 |
+
"learning_rate": 9.674706793538394e-06,
|
| 398 |
+
"loss": 0.3799,
|
| 399 |
+
"step": 650
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.13,
|
| 403 |
+
"learning_rate": 9.652578003983182e-06,
|
| 404 |
+
"loss": 0.4931,
|
| 405 |
+
"step": 660
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.13,
|
| 409 |
+
"learning_rate": 9.630449214427972e-06,
|
| 410 |
+
"loss": 0.3699,
|
| 411 |
+
"step": 670
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.14,
|
| 415 |
+
"learning_rate": 9.60832042487276e-06,
|
| 416 |
+
"loss": 0.3177,
|
| 417 |
+
"step": 680
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.14,
|
| 421 |
+
"learning_rate": 9.586191635317548e-06,
|
| 422 |
+
"loss": 0.3574,
|
| 423 |
+
"step": 690
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.14,
|
| 427 |
+
"learning_rate": 9.564062845762337e-06,
|
| 428 |
+
"loss": 0.3777,
|
| 429 |
+
"step": 700
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.14,
|
| 433 |
+
"learning_rate": 9.541934056207127e-06,
|
| 434 |
+
"loss": 0.2595,
|
| 435 |
+
"step": 710
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.14,
|
| 439 |
+
"learning_rate": 9.519805266651915e-06,
|
| 440 |
+
"loss": 0.2837,
|
| 441 |
+
"step": 720
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.15,
|
| 445 |
+
"learning_rate": 9.497676477096703e-06,
|
| 446 |
+
"loss": 0.2935,
|
| 447 |
+
"step": 730
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.15,
|
| 451 |
+
"learning_rate": 9.475547687541491e-06,
|
| 452 |
+
"loss": 0.2305,
|
| 453 |
+
"step": 740
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.15,
|
| 457 |
+
"learning_rate": 9.453418897986281e-06,
|
| 458 |
+
"loss": 0.3272,
|
| 459 |
+
"step": 750
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.15,
|
| 463 |
+
"learning_rate": 9.43129010843107e-06,
|
| 464 |
+
"loss": 0.2583,
|
| 465 |
+
"step": 760
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.15,
|
| 469 |
+
"learning_rate": 9.409161318875858e-06,
|
| 470 |
+
"loss": 0.3476,
|
| 471 |
+
"step": 770
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.16,
|
| 475 |
+
"learning_rate": 9.387032529320646e-06,
|
| 476 |
+
"loss": 0.2818,
|
| 477 |
+
"step": 780
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.16,
|
| 481 |
+
"learning_rate": 9.364903739765436e-06,
|
| 482 |
+
"loss": 0.3278,
|
| 483 |
+
"step": 790
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.16,
|
| 487 |
+
"learning_rate": 9.342774950210224e-06,
|
| 488 |
+
"loss": 0.306,
|
| 489 |
+
"step": 800
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.16,
|
| 493 |
+
"learning_rate": 9.320646160655012e-06,
|
| 494 |
+
"loss": 0.406,
|
| 495 |
+
"step": 810
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.16,
|
| 499 |
+
"learning_rate": 9.2985173710998e-06,
|
| 500 |
+
"loss": 0.2104,
|
| 501 |
+
"step": 820
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.17,
|
| 505 |
+
"learning_rate": 9.27638858154459e-06,
|
| 506 |
+
"loss": 0.232,
|
| 507 |
+
"step": 830
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.17,
|
| 511 |
+
"eval_accuracy": 0.7812622500980008,
|
| 512 |
+
"eval_loss": 0.5134419202804565,
|
| 513 |
+
"eval_runtime": 577.7526,
|
| 514 |
+
"eval_samples_per_second": 4.415,
|
| 515 |
+
"eval_steps_per_second": 0.369,
|
| 516 |
+
"step": 838
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 1.0,
|
| 520 |
+
"learning_rate": 9.254259791989379e-06,
|
| 521 |
+
"loss": 0.3927,
|
| 522 |
+
"step": 840
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.0,
|
| 526 |
+
"learning_rate": 9.232131002434168e-06,
|
| 527 |
+
"loss": 0.3709,
|
| 528 |
+
"step": 850
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.0,
|
| 532 |
+
"learning_rate": 9.210002212878957e-06,
|
| 533 |
+
"loss": 0.3054,
|
| 534 |
+
"step": 860
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 1.01,
|
| 538 |
+
"learning_rate": 9.187873423323745e-06,
|
| 539 |
+
"loss": 0.2988,
|
| 540 |
+
"step": 870
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.01,
|
| 544 |
+
"learning_rate": 9.165744633768535e-06,
|
| 545 |
+
"loss": 0.2673,
|
| 546 |
+
"step": 880
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.01,
|
| 550 |
+
"learning_rate": 9.143615844213323e-06,
|
| 551 |
+
"loss": 0.3064,
|
| 552 |
+
"step": 890
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.01,
|
| 556 |
+
"learning_rate": 9.121487054658111e-06,
|
| 557 |
+
"loss": 0.3245,
|
| 558 |
+
"step": 900
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 1.01,
|
| 562 |
+
"learning_rate": 9.0993582651029e-06,
|
| 563 |
+
"loss": 0.2463,
|
| 564 |
+
"step": 910
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.02,
|
| 568 |
+
"learning_rate": 9.07722947554769e-06,
|
| 569 |
+
"loss": 0.2583,
|
| 570 |
+
"step": 920
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.02,
|
| 574 |
+
"learning_rate": 9.055100685992478e-06,
|
| 575 |
+
"loss": 0.302,
|
| 576 |
+
"step": 930
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.02,
|
| 580 |
+
"learning_rate": 9.032971896437266e-06,
|
| 581 |
+
"loss": 0.2733,
|
| 582 |
+
"step": 940
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.02,
|
| 586 |
+
"learning_rate": 9.010843106882054e-06,
|
| 587 |
+
"loss": 0.2683,
|
| 588 |
+
"step": 950
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 1.02,
|
| 592 |
+
"learning_rate": 8.988714317326844e-06,
|
| 593 |
+
"loss": 0.2354,
|
| 594 |
+
"step": 960
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.03,
|
| 598 |
+
"learning_rate": 8.966585527771632e-06,
|
| 599 |
+
"loss": 0.1559,
|
| 600 |
+
"step": 970
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.03,
|
| 604 |
+
"learning_rate": 8.94445673821642e-06,
|
| 605 |
+
"loss": 0.4634,
|
| 606 |
+
"step": 980
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.03,
|
| 610 |
+
"learning_rate": 8.922327948661209e-06,
|
| 611 |
+
"loss": 0.2703,
|
| 612 |
+
"step": 990
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.03,
|
| 616 |
+
"learning_rate": 8.900199159105998e-06,
|
| 617 |
+
"loss": 0.3276,
|
| 618 |
+
"step": 1000
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.03,
|
| 622 |
+
"learning_rate": 8.878070369550787e-06,
|
| 623 |
+
"loss": 0.3241,
|
| 624 |
+
"step": 1010
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 1.04,
|
| 628 |
+
"learning_rate": 8.855941579995575e-06,
|
| 629 |
+
"loss": 0.339,
|
| 630 |
+
"step": 1020
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.04,
|
| 634 |
+
"learning_rate": 8.833812790440363e-06,
|
| 635 |
+
"loss": 0.2769,
|
| 636 |
+
"step": 1030
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.04,
|
| 640 |
+
"learning_rate": 8.811684000885153e-06,
|
| 641 |
+
"loss": 0.3192,
|
| 642 |
+
"step": 1040
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.04,
|
| 646 |
+
"learning_rate": 8.789555211329941e-06,
|
| 647 |
+
"loss": 0.2547,
|
| 648 |
+
"step": 1050
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.04,
|
| 652 |
+
"learning_rate": 8.76742642177473e-06,
|
| 653 |
+
"loss": 0.3132,
|
| 654 |
+
"step": 1060
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.05,
|
| 658 |
+
"learning_rate": 8.745297632219518e-06,
|
| 659 |
+
"loss": 0.2438,
|
| 660 |
+
"step": 1070
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 1.05,
|
| 664 |
+
"learning_rate": 8.723168842664308e-06,
|
| 665 |
+
"loss": 0.2702,
|
| 666 |
+
"step": 1080
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 1.05,
|
| 670 |
+
"learning_rate": 8.701040053109096e-06,
|
| 671 |
+
"loss": 0.2292,
|
| 672 |
+
"step": 1090
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.05,
|
| 676 |
+
"learning_rate": 8.678911263553884e-06,
|
| 677 |
+
"loss": 0.1375,
|
| 678 |
+
"step": 1100
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.05,
|
| 682 |
+
"learning_rate": 8.656782473998672e-06,
|
| 683 |
+
"loss": 0.2329,
|
| 684 |
+
"step": 1110
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.06,
|
| 688 |
+
"learning_rate": 8.634653684443462e-06,
|
| 689 |
+
"loss": 0.2198,
|
| 690 |
+
"step": 1120
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.06,
|
| 694 |
+
"learning_rate": 8.61252489488825e-06,
|
| 695 |
+
"loss": 0.2714,
|
| 696 |
+
"step": 1130
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.06,
|
| 700 |
+
"learning_rate": 8.590396105333039e-06,
|
| 701 |
+
"loss": 0.2835,
|
| 702 |
+
"step": 1140
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.06,
|
| 706 |
+
"learning_rate": 8.568267315777827e-06,
|
| 707 |
+
"loss": 0.2106,
|
| 708 |
+
"step": 1150
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.06,
|
| 712 |
+
"learning_rate": 8.546138526222617e-06,
|
| 713 |
+
"loss": 0.225,
|
| 714 |
+
"step": 1160
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.07,
|
| 718 |
+
"learning_rate": 8.524009736667405e-06,
|
| 719 |
+
"loss": 0.3149,
|
| 720 |
+
"step": 1170
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.07,
|
| 724 |
+
"learning_rate": 8.501880947112193e-06,
|
| 725 |
+
"loss": 0.2412,
|
| 726 |
+
"step": 1180
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 1.07,
|
| 730 |
+
"learning_rate": 8.479752157556981e-06,
|
| 731 |
+
"loss": 0.2139,
|
| 732 |
+
"step": 1190
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.07,
|
| 736 |
+
"learning_rate": 8.457623368001771e-06,
|
| 737 |
+
"loss": 0.2128,
|
| 738 |
+
"step": 1200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.07,
|
| 742 |
+
"learning_rate": 8.43549457844656e-06,
|
| 743 |
+
"loss": 0.1696,
|
| 744 |
+
"step": 1210
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.08,
|
| 748 |
+
"learning_rate": 8.413365788891348e-06,
|
| 749 |
+
"loss": 0.196,
|
| 750 |
+
"step": 1220
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.08,
|
| 754 |
+
"learning_rate": 8.391236999336136e-06,
|
| 755 |
+
"loss": 0.3033,
|
| 756 |
+
"step": 1230
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.08,
|
| 760 |
+
"learning_rate": 8.369108209780926e-06,
|
| 761 |
+
"loss": 0.2646,
|
| 762 |
+
"step": 1240
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.08,
|
| 766 |
+
"learning_rate": 8.346979420225714e-06,
|
| 767 |
+
"loss": 0.344,
|
| 768 |
+
"step": 1250
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 1.08,
|
| 772 |
+
"learning_rate": 8.324850630670504e-06,
|
| 773 |
+
"loss": 0.1595,
|
| 774 |
+
"step": 1260
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.09,
|
| 778 |
+
"learning_rate": 8.302721841115292e-06,
|
| 779 |
+
"loss": 0.3235,
|
| 780 |
+
"step": 1270
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.09,
|
| 784 |
+
"learning_rate": 8.28059305156008e-06,
|
| 785 |
+
"loss": 0.1695,
|
| 786 |
+
"step": 1280
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.09,
|
| 790 |
+
"learning_rate": 8.258464262004869e-06,
|
| 791 |
+
"loss": 0.2382,
|
| 792 |
+
"step": 1290
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.09,
|
| 796 |
+
"learning_rate": 8.236335472449659e-06,
|
| 797 |
+
"loss": 0.3283,
|
| 798 |
+
"step": 1300
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 1.09,
|
| 802 |
+
"learning_rate": 8.214206682894447e-06,
|
| 803 |
+
"loss": 0.3028,
|
| 804 |
+
"step": 1310
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.1,
|
| 808 |
+
"learning_rate": 8.192077893339235e-06,
|
| 809 |
+
"loss": 0.231,
|
| 810 |
+
"step": 1320
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 1.1,
|
| 814 |
+
"learning_rate": 8.169949103784023e-06,
|
| 815 |
+
"loss": 0.319,
|
| 816 |
+
"step": 1330
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.1,
|
| 820 |
+
"learning_rate": 8.147820314228813e-06,
|
| 821 |
+
"loss": 0.2392,
|
| 822 |
+
"step": 1340
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.1,
|
| 826 |
+
"learning_rate": 8.125691524673601e-06,
|
| 827 |
+
"loss": 0.2225,
|
| 828 |
+
"step": 1350
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.1,
|
| 832 |
+
"learning_rate": 8.10356273511839e-06,
|
| 833 |
+
"loss": 0.449,
|
| 834 |
+
"step": 1360
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.11,
|
| 838 |
+
"learning_rate": 8.081433945563178e-06,
|
| 839 |
+
"loss": 0.3471,
|
| 840 |
+
"step": 1370
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 1.11,
|
| 844 |
+
"learning_rate": 8.059305156007968e-06,
|
| 845 |
+
"loss": 0.3111,
|
| 846 |
+
"step": 1380
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.11,
|
| 850 |
+
"learning_rate": 8.037176366452756e-06,
|
| 851 |
+
"loss": 0.2577,
|
| 852 |
+
"step": 1390
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.11,
|
| 856 |
+
"learning_rate": 8.015047576897544e-06,
|
| 857 |
+
"loss": 0.205,
|
| 858 |
+
"step": 1400
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.11,
|
| 862 |
+
"learning_rate": 7.992918787342332e-06,
|
| 863 |
+
"loss": 0.2821,
|
| 864 |
+
"step": 1410
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.12,
|
| 868 |
+
"learning_rate": 7.970789997787122e-06,
|
| 869 |
+
"loss": 0.2689,
|
| 870 |
+
"step": 1420
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.12,
|
| 874 |
+
"learning_rate": 7.94866120823191e-06,
|
| 875 |
+
"loss": 0.2167,
|
| 876 |
+
"step": 1430
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.12,
|
| 880 |
+
"learning_rate": 7.926532418676699e-06,
|
| 881 |
+
"loss": 0.2621,
|
| 882 |
+
"step": 1440
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.12,
|
| 886 |
+
"learning_rate": 7.904403629121487e-06,
|
| 887 |
+
"loss": 0.3638,
|
| 888 |
+
"step": 1450
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.12,
|
| 892 |
+
"learning_rate": 7.882274839566277e-06,
|
| 893 |
+
"loss": 0.2401,
|
| 894 |
+
"step": 1460
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 1.13,
|
| 898 |
+
"learning_rate": 7.860146050011065e-06,
|
| 899 |
+
"loss": 0.2378,
|
| 900 |
+
"step": 1470
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 1.13,
|
| 904 |
+
"learning_rate": 7.838017260455853e-06,
|
| 905 |
+
"loss": 0.2154,
|
| 906 |
+
"step": 1480
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.13,
|
| 910 |
+
"learning_rate": 7.815888470900642e-06,
|
| 911 |
+
"loss": 0.2149,
|
| 912 |
+
"step": 1490
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.13,
|
| 916 |
+
"learning_rate": 7.793759681345431e-06,
|
| 917 |
+
"loss": 0.2284,
|
| 918 |
+
"step": 1500
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.13,
|
| 922 |
+
"learning_rate": 7.77163089179022e-06,
|
| 923 |
+
"loss": 0.1492,
|
| 924 |
+
"step": 1510
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.14,
|
| 928 |
+
"learning_rate": 7.749502102235008e-06,
|
| 929 |
+
"loss": 0.3516,
|
| 930 |
+
"step": 1520
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.14,
|
| 934 |
+
"learning_rate": 7.727373312679796e-06,
|
| 935 |
+
"loss": 0.2962,
|
| 936 |
+
"step": 1530
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 1.14,
|
| 940 |
+
"learning_rate": 7.705244523124586e-06,
|
| 941 |
+
"loss": 0.1978,
|
| 942 |
+
"step": 1540
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 1.14,
|
| 946 |
+
"learning_rate": 7.683115733569374e-06,
|
| 947 |
+
"loss": 0.2855,
|
| 948 |
+
"step": 1550
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.14,
|
| 952 |
+
"learning_rate": 7.660986944014162e-06,
|
| 953 |
+
"loss": 0.2107,
|
| 954 |
+
"step": 1560
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.15,
|
| 958 |
+
"learning_rate": 7.63885815445895e-06,
|
| 959 |
+
"loss": 0.1989,
|
| 960 |
+
"step": 1570
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.15,
|
| 964 |
+
"learning_rate": 7.61672936490374e-06,
|
| 965 |
+
"loss": 0.1848,
|
| 966 |
+
"step": 1580
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 1.15,
|
| 970 |
+
"learning_rate": 7.594600575348529e-06,
|
| 971 |
+
"loss": 0.2676,
|
| 972 |
+
"step": 1590
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 1.15,
|
| 976 |
+
"learning_rate": 7.572471785793317e-06,
|
| 977 |
+
"loss": 0.2119,
|
| 978 |
+
"step": 1600
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.15,
|
| 982 |
+
"learning_rate": 7.550342996238106e-06,
|
| 983 |
+
"loss": 0.3076,
|
| 984 |
+
"step": 1610
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.16,
|
| 988 |
+
"learning_rate": 7.528214206682896e-06,
|
| 989 |
+
"loss": 0.2303,
|
| 990 |
+
"step": 1620
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.16,
|
| 994 |
+
"learning_rate": 7.506085417127684e-06,
|
| 995 |
+
"loss": 0.1646,
|
| 996 |
+
"step": 1630
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.16,
|
| 1000 |
+
"learning_rate": 7.4839566275724725e-06,
|
| 1001 |
+
"loss": 0.2384,
|
| 1002 |
+
"step": 1640
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.16,
|
| 1006 |
+
"learning_rate": 7.4618278380172615e-06,
|
| 1007 |
+
"loss": 0.1908,
|
| 1008 |
+
"step": 1650
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.16,
|
| 1012 |
+
"learning_rate": 7.43969904846205e-06,
|
| 1013 |
+
"loss": 0.3155,
|
| 1014 |
+
"step": 1660
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.17,
|
| 1018 |
+
"learning_rate": 7.417570258906839e-06,
|
| 1019 |
+
"loss": 0.1099,
|
| 1020 |
+
"step": 1670
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.17,
|
| 1024 |
+
"eval_accuracy": 0.782438259506076,
|
| 1025 |
+
"eval_loss": 0.7191832661628723,
|
| 1026 |
+
"eval_runtime": 556.1282,
|
| 1027 |
+
"eval_samples_per_second": 4.587,
|
| 1028 |
+
"eval_steps_per_second": 0.383,
|
| 1029 |
+
"step": 1676
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 2.0,
|
| 1033 |
+
"learning_rate": 7.395441469351627e-06,
|
| 1034 |
+
"loss": 0.1875,
|
| 1035 |
+
"step": 1680
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.0,
|
| 1039 |
+
"learning_rate": 7.373312679796416e-06,
|
| 1040 |
+
"loss": 0.2107,
|
| 1041 |
+
"step": 1690
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.0,
|
| 1045 |
+
"learning_rate": 7.351183890241204e-06,
|
| 1046 |
+
"loss": 0.374,
|
| 1047 |
+
"step": 1700
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 2.01,
|
| 1051 |
+
"learning_rate": 7.329055100685993e-06,
|
| 1052 |
+
"loss": 0.2293,
|
| 1053 |
+
"step": 1710
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.01,
|
| 1057 |
+
"learning_rate": 7.306926311130782e-06,
|
| 1058 |
+
"loss": 0.2004,
|
| 1059 |
+
"step": 1720
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 2.01,
|
| 1063 |
+
"learning_rate": 7.284797521575571e-06,
|
| 1064 |
+
"loss": 0.2731,
|
| 1065 |
+
"step": 1730
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 2.01,
|
| 1069 |
+
"learning_rate": 7.262668732020359e-06,
|
| 1070 |
+
"loss": 0.2156,
|
| 1071 |
+
"step": 1740
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.01,
|
| 1075 |
+
"learning_rate": 7.240539942465148e-06,
|
| 1076 |
+
"loss": 0.3421,
|
| 1077 |
+
"step": 1750
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 2.02,
|
| 1081 |
+
"learning_rate": 7.218411152909936e-06,
|
| 1082 |
+
"loss": 0.1165,
|
| 1083 |
+
"step": 1760
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.02,
|
| 1087 |
+
"learning_rate": 7.196282363354725e-06,
|
| 1088 |
+
"loss": 0.1781,
|
| 1089 |
+
"step": 1770
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.02,
|
| 1093 |
+
"learning_rate": 7.1741535737995134e-06,
|
| 1094 |
+
"loss": 0.1665,
|
| 1095 |
+
"step": 1780
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.02,
|
| 1099 |
+
"learning_rate": 7.1520247842443025e-06,
|
| 1100 |
+
"loss": 0.1456,
|
| 1101 |
+
"step": 1790
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 2.02,
|
| 1105 |
+
"learning_rate": 7.129895994689091e-06,
|
| 1106 |
+
"loss": 0.2544,
|
| 1107 |
+
"step": 1800
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.03,
|
| 1111 |
+
"learning_rate": 7.10776720513388e-06,
|
| 1112 |
+
"loss": 0.2747,
|
| 1113 |
+
"step": 1810
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 2.03,
|
| 1117 |
+
"learning_rate": 7.085638415578668e-06,
|
| 1118 |
+
"loss": 0.2413,
|
| 1119 |
+
"step": 1820
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.03,
|
| 1123 |
+
"learning_rate": 7.063509626023457e-06,
|
| 1124 |
+
"loss": 0.2187,
|
| 1125 |
+
"step": 1830
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.03,
|
| 1129 |
+
"learning_rate": 7.041380836468245e-06,
|
| 1130 |
+
"loss": 0.2101,
|
| 1131 |
+
"step": 1840
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 2.03,
|
| 1135 |
+
"learning_rate": 7.019252046913034e-06,
|
| 1136 |
+
"loss": 0.2719,
|
| 1137 |
+
"step": 1850
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.04,
|
| 1141 |
+
"learning_rate": 6.997123257357823e-06,
|
| 1142 |
+
"loss": 0.2055,
|
| 1143 |
+
"step": 1860
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.04,
|
| 1147 |
+
"learning_rate": 6.974994467802612e-06,
|
| 1148 |
+
"loss": 0.1508,
|
| 1149 |
+
"step": 1870
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.04,
|
| 1153 |
+
"learning_rate": 6.9528656782474e-06,
|
| 1154 |
+
"loss": 0.2017,
|
| 1155 |
+
"step": 1880
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 2.04,
|
| 1159 |
+
"learning_rate": 6.930736888692189e-06,
|
| 1160 |
+
"loss": 0.1292,
|
| 1161 |
+
"step": 1890
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.04,
|
| 1165 |
+
"learning_rate": 6.908608099136977e-06,
|
| 1166 |
+
"loss": 0.1796,
|
| 1167 |
+
"step": 1900
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 2.05,
|
| 1171 |
+
"learning_rate": 6.886479309581766e-06,
|
| 1172 |
+
"loss": 0.2117,
|
| 1173 |
+
"step": 1910
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 2.05,
|
| 1177 |
+
"learning_rate": 6.8643505200265544e-06,
|
| 1178 |
+
"loss": 0.2878,
|
| 1179 |
+
"step": 1920
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.05,
|
| 1183 |
+
"learning_rate": 6.8422217304713435e-06,
|
| 1184 |
+
"loss": 0.1418,
|
| 1185 |
+
"step": 1930
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 2.05,
|
| 1189 |
+
"learning_rate": 6.820092940916132e-06,
|
| 1190 |
+
"loss": 0.0803,
|
| 1191 |
+
"step": 1940
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 2.05,
|
| 1195 |
+
"learning_rate": 6.797964151360921e-06,
|
| 1196 |
+
"loss": 0.3428,
|
| 1197 |
+
"step": 1950
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 2.06,
|
| 1201 |
+
"learning_rate": 6.775835361805709e-06,
|
| 1202 |
+
"loss": 0.2008,
|
| 1203 |
+
"step": 1960
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 2.06,
|
| 1207 |
+
"learning_rate": 6.753706572250498e-06,
|
| 1208 |
+
"loss": 0.2444,
|
| 1209 |
+
"step": 1970
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 2.06,
|
| 1213 |
+
"learning_rate": 6.731577782695286e-06,
|
| 1214 |
+
"loss": 0.2775,
|
| 1215 |
+
"step": 1980
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 2.06,
|
| 1219 |
+
"learning_rate": 6.709448993140075e-06,
|
| 1220 |
+
"loss": 0.1756,
|
| 1221 |
+
"step": 1990
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.06,
|
| 1225 |
+
"learning_rate": 6.687320203584865e-06,
|
| 1226 |
+
"loss": 0.2606,
|
| 1227 |
+
"step": 2000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 2.07,
|
| 1231 |
+
"learning_rate": 6.6651914140296535e-06,
|
| 1232 |
+
"loss": 0.2213,
|
| 1233 |
+
"step": 2010
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 2.07,
|
| 1237 |
+
"learning_rate": 6.6430626244744426e-06,
|
| 1238 |
+
"loss": 0.1869,
|
| 1239 |
+
"step": 2020
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 2.07,
|
| 1243 |
+
"learning_rate": 6.620933834919231e-06,
|
| 1244 |
+
"loss": 0.2149,
|
| 1245 |
+
"step": 2030
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 2.07,
|
| 1249 |
+
"learning_rate": 6.59880504536402e-06,
|
| 1250 |
+
"loss": 0.2346,
|
| 1251 |
+
"step": 2040
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 2.07,
|
| 1255 |
+
"learning_rate": 6.576676255808808e-06,
|
| 1256 |
+
"loss": 0.1693,
|
| 1257 |
+
"step": 2050
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 2.08,
|
| 1261 |
+
"learning_rate": 6.554547466253597e-06,
|
| 1262 |
+
"loss": 0.2225,
|
| 1263 |
+
"step": 2060
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.08,
|
| 1267 |
+
"learning_rate": 6.532418676698385e-06,
|
| 1268 |
+
"loss": 0.2497,
|
| 1269 |
+
"step": 2070
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 2.08,
|
| 1273 |
+
"learning_rate": 6.510289887143174e-06,
|
| 1274 |
+
"loss": 0.2825,
|
| 1275 |
+
"step": 2080
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 2.08,
|
| 1279 |
+
"learning_rate": 6.488161097587963e-06,
|
| 1280 |
+
"loss": 0.2909,
|
| 1281 |
+
"step": 2090
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 2.08,
|
| 1285 |
+
"learning_rate": 6.466032308032752e-06,
|
| 1286 |
+
"loss": 0.3198,
|
| 1287 |
+
"step": 2100
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 2.09,
|
| 1291 |
+
"learning_rate": 6.44390351847754e-06,
|
| 1292 |
+
"loss": 0.1954,
|
| 1293 |
+
"step": 2110
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 2.09,
|
| 1297 |
+
"learning_rate": 6.421774728922329e-06,
|
| 1298 |
+
"loss": 0.2166,
|
| 1299 |
+
"step": 2120
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 2.09,
|
| 1303 |
+
"learning_rate": 6.399645939367117e-06,
|
| 1304 |
+
"loss": 0.3095,
|
| 1305 |
+
"step": 2130
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.09,
|
| 1309 |
+
"learning_rate": 6.377517149811906e-06,
|
| 1310 |
+
"loss": 0.0666,
|
| 1311 |
+
"step": 2140
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 2.09,
|
| 1315 |
+
"learning_rate": 6.3553883602566945e-06,
|
| 1316 |
+
"loss": 0.2469,
|
| 1317 |
+
"step": 2150
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 2.1,
|
| 1321 |
+
"learning_rate": 6.3332595707014835e-06,
|
| 1322 |
+
"loss": 0.1981,
|
| 1323 |
+
"step": 2160
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 2.1,
|
| 1327 |
+
"learning_rate": 6.311130781146272e-06,
|
| 1328 |
+
"loss": 0.0946,
|
| 1329 |
+
"step": 2170
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 2.1,
|
| 1333 |
+
"learning_rate": 6.289001991591061e-06,
|
| 1334 |
+
"loss": 0.209,
|
| 1335 |
+
"step": 2180
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 2.1,
|
| 1339 |
+
"learning_rate": 6.266873202035849e-06,
|
| 1340 |
+
"loss": 0.1831,
|
| 1341 |
+
"step": 2190
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 2.1,
|
| 1345 |
+
"learning_rate": 6.244744412480638e-06,
|
| 1346 |
+
"loss": 0.226,
|
| 1347 |
+
"step": 2200
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.11,
|
| 1351 |
+
"learning_rate": 6.222615622925426e-06,
|
| 1352 |
+
"loss": 0.1695,
|
| 1353 |
+
"step": 2210
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 2.11,
|
| 1357 |
+
"learning_rate": 6.200486833370215e-06,
|
| 1358 |
+
"loss": 0.2599,
|
| 1359 |
+
"step": 2220
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 2.11,
|
| 1363 |
+
"learning_rate": 6.178358043815004e-06,
|
| 1364 |
+
"loss": 0.0863,
|
| 1365 |
+
"step": 2230
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 2.11,
|
| 1369 |
+
"learning_rate": 6.156229254259793e-06,
|
| 1370 |
+
"loss": 0.2787,
|
| 1371 |
+
"step": 2240
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 2.11,
|
| 1375 |
+
"learning_rate": 6.134100464704581e-06,
|
| 1376 |
+
"loss": 0.3413,
|
| 1377 |
+
"step": 2250
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 2.12,
|
| 1381 |
+
"learning_rate": 6.11197167514937e-06,
|
| 1382 |
+
"loss": 0.2457,
|
| 1383 |
+
"step": 2260
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 2.12,
|
| 1387 |
+
"learning_rate": 6.089842885594158e-06,
|
| 1388 |
+
"loss": 0.1474,
|
| 1389 |
+
"step": 2270
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.12,
|
| 1393 |
+
"learning_rate": 6.067714096038947e-06,
|
| 1394 |
+
"loss": 0.1682,
|
| 1395 |
+
"step": 2280
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 2.12,
|
| 1399 |
+
"learning_rate": 6.0455853064837355e-06,
|
| 1400 |
+
"loss": 0.236,
|
| 1401 |
+
"step": 2290
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 2.12,
|
| 1405 |
+
"learning_rate": 6.0234565169285245e-06,
|
| 1406 |
+
"loss": 0.1295,
|
| 1407 |
+
"step": 2300
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 2.13,
|
| 1411 |
+
"learning_rate": 6.001327727373313e-06,
|
| 1412 |
+
"loss": 0.1731,
|
| 1413 |
+
"step": 2310
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 2.13,
|
| 1417 |
+
"learning_rate": 5.979198937818102e-06,
|
| 1418 |
+
"loss": 0.1829,
|
| 1419 |
+
"step": 2320
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 2.13,
|
| 1423 |
+
"learning_rate": 5.95707014826289e-06,
|
| 1424 |
+
"loss": 0.2921,
|
| 1425 |
+
"step": 2330
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 2.13,
|
| 1429 |
+
"learning_rate": 5.934941358707679e-06,
|
| 1430 |
+
"loss": 0.2537,
|
| 1431 |
+
"step": 2340
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.13,
|
| 1435 |
+
"learning_rate": 5.912812569152467e-06,
|
| 1436 |
+
"loss": 0.2838,
|
| 1437 |
+
"step": 2350
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 2.14,
|
| 1441 |
+
"learning_rate": 5.890683779597256e-06,
|
| 1442 |
+
"loss": 0.2342,
|
| 1443 |
+
"step": 2360
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 2.14,
|
| 1447 |
+
"learning_rate": 5.868554990042045e-06,
|
| 1448 |
+
"loss": 0.1875,
|
| 1449 |
+
"step": 2370
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 2.14,
|
| 1453 |
+
"learning_rate": 5.8464262004868345e-06,
|
| 1454 |
+
"loss": 0.2775,
|
| 1455 |
+
"step": 2380
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 2.14,
|
| 1459 |
+
"learning_rate": 5.824297410931623e-06,
|
| 1460 |
+
"loss": 0.2561,
|
| 1461 |
+
"step": 2390
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 2.14,
|
| 1465 |
+
"learning_rate": 5.802168621376412e-06,
|
| 1466 |
+
"loss": 0.292,
|
| 1467 |
+
"step": 2400
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 2.15,
|
| 1471 |
+
"learning_rate": 5.7800398318212e-06,
|
| 1472 |
+
"loss": 0.2448,
|
| 1473 |
+
"step": 2410
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 2.15,
|
| 1477 |
+
"learning_rate": 5.757911042265989e-06,
|
| 1478 |
+
"loss": 0.2695,
|
| 1479 |
+
"step": 2420
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 2.15,
|
| 1483 |
+
"learning_rate": 5.735782252710777e-06,
|
| 1484 |
+
"loss": 0.1965,
|
| 1485 |
+
"step": 2430
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 2.15,
|
| 1489 |
+
"learning_rate": 5.713653463155566e-06,
|
| 1490 |
+
"loss": 0.203,
|
| 1491 |
+
"step": 2440
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 2.15,
|
| 1495 |
+
"learning_rate": 5.691524673600355e-06,
|
| 1496 |
+
"loss": 0.239,
|
| 1497 |
+
"step": 2450
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 2.16,
|
| 1501 |
+
"learning_rate": 5.669395884045144e-06,
|
| 1502 |
+
"loss": 0.2028,
|
| 1503 |
+
"step": 2460
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 2.16,
|
| 1507 |
+
"learning_rate": 5.647267094489932e-06,
|
| 1508 |
+
"loss": 0.1763,
|
| 1509 |
+
"step": 2470
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 2.16,
|
| 1513 |
+
"learning_rate": 5.625138304934721e-06,
|
| 1514 |
+
"loss": 0.2413,
|
| 1515 |
+
"step": 2480
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 2.16,
|
| 1519 |
+
"learning_rate": 5.603009515379509e-06,
|
| 1520 |
+
"loss": 0.2726,
|
| 1521 |
+
"step": 2490
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 2.16,
|
| 1525 |
+
"learning_rate": 5.580880725824298e-06,
|
| 1526 |
+
"loss": 0.2998,
|
| 1527 |
+
"step": 2500
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 2.17,
|
| 1531 |
+
"learning_rate": 5.5587519362690865e-06,
|
| 1532 |
+
"loss": 0.2178,
|
| 1533 |
+
"step": 2510
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 2.17,
|
| 1537 |
+
"eval_accuracy": 0.8298706389651117,
|
| 1538 |
+
"eval_loss": 0.4418592154979706,
|
| 1539 |
+
"eval_runtime": 455.5394,
|
| 1540 |
+
"eval_samples_per_second": 5.6,
|
| 1541 |
+
"eval_steps_per_second": 0.468,
|
| 1542 |
+
"step": 2514
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 3.0,
|
| 1546 |
+
"learning_rate": 5.5366231467138755e-06,
|
| 1547 |
+
"loss": 0.1391,
|
| 1548 |
+
"step": 2520
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 3.0,
|
| 1552 |
+
"learning_rate": 5.514494357158664e-06,
|
| 1553 |
+
"loss": 0.1202,
|
| 1554 |
+
"step": 2530
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 3.01,
|
| 1558 |
+
"learning_rate": 5.492365567603453e-06,
|
| 1559 |
+
"loss": 0.1799,
|
| 1560 |
+
"step": 2540
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 3.01,
|
| 1564 |
+
"learning_rate": 5.470236778048241e-06,
|
| 1565 |
+
"loss": 0.1843,
|
| 1566 |
+
"step": 2550
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 3.01,
|
| 1570 |
+
"learning_rate": 5.44810798849303e-06,
|
| 1571 |
+
"loss": 0.1504,
|
| 1572 |
+
"step": 2560
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 3.01,
|
| 1576 |
+
"learning_rate": 5.425979198937818e-06,
|
| 1577 |
+
"loss": 0.1064,
|
| 1578 |
+
"step": 2570
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.01,
|
| 1582 |
+
"learning_rate": 5.403850409382607e-06,
|
| 1583 |
+
"loss": 0.397,
|
| 1584 |
+
"step": 2580
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 3.02,
|
| 1588 |
+
"learning_rate": 5.381721619827396e-06,
|
| 1589 |
+
"loss": 0.1298,
|
| 1590 |
+
"step": 2590
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 3.02,
|
| 1594 |
+
"learning_rate": 5.359592830272185e-06,
|
| 1595 |
+
"loss": 0.1984,
|
| 1596 |
+
"step": 2600
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 3.02,
|
| 1600 |
+
"learning_rate": 5.337464040716973e-06,
|
| 1601 |
+
"loss": 0.3349,
|
| 1602 |
+
"step": 2610
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 3.02,
|
| 1606 |
+
"learning_rate": 5.315335251161762e-06,
|
| 1607 |
+
"loss": 0.1957,
|
| 1608 |
+
"step": 2620
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 3.02,
|
| 1612 |
+
"learning_rate": 5.29320646160655e-06,
|
| 1613 |
+
"loss": 0.2681,
|
| 1614 |
+
"step": 2630
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 3.03,
|
| 1618 |
+
"learning_rate": 5.271077672051339e-06,
|
| 1619 |
+
"loss": 0.0977,
|
| 1620 |
+
"step": 2640
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.03,
|
| 1624 |
+
"learning_rate": 5.2489488824961274e-06,
|
| 1625 |
+
"loss": 0.3088,
|
| 1626 |
+
"step": 2650
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 3.03,
|
| 1630 |
+
"learning_rate": 5.2268200929409165e-06,
|
| 1631 |
+
"loss": 0.2288,
|
| 1632 |
+
"step": 2660
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 3.03,
|
| 1636 |
+
"learning_rate": 5.204691303385705e-06,
|
| 1637 |
+
"loss": 0.2878,
|
| 1638 |
+
"step": 2670
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 3.03,
|
| 1642 |
+
"learning_rate": 5.182562513830494e-06,
|
| 1643 |
+
"loss": 0.2257,
|
| 1644 |
+
"step": 2680
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 3.04,
|
| 1648 |
+
"learning_rate": 5.160433724275282e-06,
|
| 1649 |
+
"loss": 0.15,
|
| 1650 |
+
"step": 2690
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 3.04,
|
| 1654 |
+
"learning_rate": 5.138304934720071e-06,
|
| 1655 |
+
"loss": 0.2424,
|
| 1656 |
+
"step": 2700
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 3.04,
|
| 1660 |
+
"learning_rate": 5.116176145164859e-06,
|
| 1661 |
+
"loss": 0.2213,
|
| 1662 |
+
"step": 2710
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 3.04,
|
| 1666 |
+
"learning_rate": 5.094047355609648e-06,
|
| 1667 |
+
"loss": 0.1488,
|
| 1668 |
+
"step": 2720
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 3.04,
|
| 1672 |
+
"learning_rate": 5.071918566054437e-06,
|
| 1673 |
+
"loss": 0.1004,
|
| 1674 |
+
"step": 2730
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 3.05,
|
| 1678 |
+
"learning_rate": 5.049789776499226e-06,
|
| 1679 |
+
"loss": 0.2068,
|
| 1680 |
+
"step": 2740
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 3.05,
|
| 1684 |
+
"learning_rate": 5.027660986944014e-06,
|
| 1685 |
+
"loss": 0.3556,
|
| 1686 |
+
"step": 2750
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 3.05,
|
| 1690 |
+
"learning_rate": 5.005532197388803e-06,
|
| 1691 |
+
"loss": 0.2188,
|
| 1692 |
+
"step": 2760
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 3.05,
|
| 1696 |
+
"learning_rate": 4.983403407833592e-06,
|
| 1697 |
+
"loss": 0.2531,
|
| 1698 |
+
"step": 2770
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 3.05,
|
| 1702 |
+
"learning_rate": 4.96127461827838e-06,
|
| 1703 |
+
"loss": 0.2659,
|
| 1704 |
+
"step": 2780
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 3.05,
|
| 1708 |
+
"learning_rate": 4.939145828723169e-06,
|
| 1709 |
+
"loss": 0.2333,
|
| 1710 |
+
"step": 2790
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 3.06,
|
| 1714 |
+
"learning_rate": 4.9170170391679575e-06,
|
| 1715 |
+
"loss": 0.2603,
|
| 1716 |
+
"step": 2800
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 3.06,
|
| 1720 |
+
"learning_rate": 4.8948882496127466e-06,
|
| 1721 |
+
"loss": 0.1662,
|
| 1722 |
+
"step": 2810
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 3.06,
|
| 1726 |
+
"learning_rate": 4.872759460057535e-06,
|
| 1727 |
+
"loss": 0.1891,
|
| 1728 |
+
"step": 2820
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 3.06,
|
| 1732 |
+
"learning_rate": 4.850630670502324e-06,
|
| 1733 |
+
"loss": 0.1719,
|
| 1734 |
+
"step": 2830
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 3.06,
|
| 1738 |
+
"learning_rate": 4.828501880947112e-06,
|
| 1739 |
+
"loss": 0.1205,
|
| 1740 |
+
"step": 2840
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 3.07,
|
| 1744 |
+
"learning_rate": 4.806373091391901e-06,
|
| 1745 |
+
"loss": 0.1061,
|
| 1746 |
+
"step": 2850
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 3.07,
|
| 1750 |
+
"learning_rate": 4.78424430183669e-06,
|
| 1751 |
+
"loss": 0.2271,
|
| 1752 |
+
"step": 2860
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 3.07,
|
| 1756 |
+
"learning_rate": 4.762115512281478e-06,
|
| 1757 |
+
"loss": 0.2392,
|
| 1758 |
+
"step": 2870
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 3.07,
|
| 1762 |
+
"learning_rate": 4.7399867227262675e-06,
|
| 1763 |
+
"loss": 0.2029,
|
| 1764 |
+
"step": 2880
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 3.07,
|
| 1768 |
+
"learning_rate": 4.717857933171056e-06,
|
| 1769 |
+
"loss": 0.2938,
|
| 1770 |
+
"step": 2890
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 3.08,
|
| 1774 |
+
"learning_rate": 4.695729143615845e-06,
|
| 1775 |
+
"loss": 0.2804,
|
| 1776 |
+
"step": 2900
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 3.08,
|
| 1780 |
+
"learning_rate": 4.673600354060633e-06,
|
| 1781 |
+
"loss": 0.2101,
|
| 1782 |
+
"step": 2910
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 3.08,
|
| 1786 |
+
"learning_rate": 4.651471564505422e-06,
|
| 1787 |
+
"loss": 0.2806,
|
| 1788 |
+
"step": 2920
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 3.08,
|
| 1792 |
+
"learning_rate": 4.62934277495021e-06,
|
| 1793 |
+
"loss": 0.1367,
|
| 1794 |
+
"step": 2930
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 3.08,
|
| 1798 |
+
"learning_rate": 4.607213985394999e-06,
|
| 1799 |
+
"loss": 0.2148,
|
| 1800 |
+
"step": 2940
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 3.09,
|
| 1804 |
+
"learning_rate": 4.5850851958397876e-06,
|
| 1805 |
+
"loss": 0.0975,
|
| 1806 |
+
"step": 2950
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 3.09,
|
| 1810 |
+
"learning_rate": 4.562956406284577e-06,
|
| 1811 |
+
"loss": 0.1914,
|
| 1812 |
+
"step": 2960
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 3.09,
|
| 1816 |
+
"learning_rate": 4.540827616729365e-06,
|
| 1817 |
+
"loss": 0.1859,
|
| 1818 |
+
"step": 2970
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 3.09,
|
| 1822 |
+
"learning_rate": 4.518698827174154e-06,
|
| 1823 |
+
"loss": 0.1547,
|
| 1824 |
+
"step": 2980
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 3.09,
|
| 1828 |
+
"learning_rate": 4.496570037618942e-06,
|
| 1829 |
+
"loss": 0.1743,
|
| 1830 |
+
"step": 2990
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 3.1,
|
| 1834 |
+
"learning_rate": 4.474441248063731e-06,
|
| 1835 |
+
"loss": 0.2247,
|
| 1836 |
+
"step": 3000
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 3.1,
|
| 1840 |
+
"learning_rate": 4.452312458508519e-06,
|
| 1841 |
+
"loss": 0.2033,
|
| 1842 |
+
"step": 3010
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 3.1,
|
| 1846 |
+
"learning_rate": 4.4301836689533085e-06,
|
| 1847 |
+
"loss": 0.1757,
|
| 1848 |
+
"step": 3020
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 3.1,
|
| 1852 |
+
"learning_rate": 4.408054879398097e-06,
|
| 1853 |
+
"loss": 0.212,
|
| 1854 |
+
"step": 3030
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 3.1,
|
| 1858 |
+
"learning_rate": 4.385926089842886e-06,
|
| 1859 |
+
"loss": 0.2183,
|
| 1860 |
+
"step": 3040
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 3.11,
|
| 1864 |
+
"learning_rate": 4.363797300287675e-06,
|
| 1865 |
+
"loss": 0.2498,
|
| 1866 |
+
"step": 3050
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 3.11,
|
| 1870 |
+
"learning_rate": 4.341668510732464e-06,
|
| 1871 |
+
"loss": 0.0668,
|
| 1872 |
+
"step": 3060
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 3.11,
|
| 1876 |
+
"learning_rate": 4.319539721177252e-06,
|
| 1877 |
+
"loss": 0.1867,
|
| 1878 |
+
"step": 3070
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 3.11,
|
| 1882 |
+
"learning_rate": 4.297410931622041e-06,
|
| 1883 |
+
"loss": 0.1403,
|
| 1884 |
+
"step": 3080
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 3.11,
|
| 1888 |
+
"learning_rate": 4.275282142066829e-06,
|
| 1889 |
+
"loss": 0.2343,
|
| 1890 |
+
"step": 3090
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 3.12,
|
| 1894 |
+
"learning_rate": 4.2531533525116185e-06,
|
| 1895 |
+
"loss": 0.1536,
|
| 1896 |
+
"step": 3100
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 3.12,
|
| 1900 |
+
"learning_rate": 4.231024562956407e-06,
|
| 1901 |
+
"loss": 0.1656,
|
| 1902 |
+
"step": 3110
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 3.12,
|
| 1906 |
+
"learning_rate": 4.208895773401196e-06,
|
| 1907 |
+
"loss": 0.2044,
|
| 1908 |
+
"step": 3120
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 3.12,
|
| 1912 |
+
"learning_rate": 4.186766983845984e-06,
|
| 1913 |
+
"loss": 0.1283,
|
| 1914 |
+
"step": 3130
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 3.12,
|
| 1918 |
+
"learning_rate": 4.164638194290773e-06,
|
| 1919 |
+
"loss": 0.0793,
|
| 1920 |
+
"step": 3140
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 3.13,
|
| 1924 |
+
"learning_rate": 4.142509404735561e-06,
|
| 1925 |
+
"loss": 0.2157,
|
| 1926 |
+
"step": 3150
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 3.13,
|
| 1930 |
+
"learning_rate": 4.12038061518035e-06,
|
| 1931 |
+
"loss": 0.2458,
|
| 1932 |
+
"step": 3160
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 3.13,
|
| 1936 |
+
"learning_rate": 4.0982518256251385e-06,
|
| 1937 |
+
"loss": 0.1386,
|
| 1938 |
+
"step": 3170
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 3.13,
|
| 1942 |
+
"learning_rate": 4.076123036069928e-06,
|
| 1943 |
+
"loss": 0.1956,
|
| 1944 |
+
"step": 3180
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 3.13,
|
| 1948 |
+
"learning_rate": 4.053994246514716e-06,
|
| 1949 |
+
"loss": 0.1353,
|
| 1950 |
+
"step": 3190
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 3.14,
|
| 1954 |
+
"learning_rate": 4.031865456959505e-06,
|
| 1955 |
+
"loss": 0.1785,
|
| 1956 |
+
"step": 3200
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 3.14,
|
| 1960 |
+
"learning_rate": 4.009736667404293e-06,
|
| 1961 |
+
"loss": 0.2949,
|
| 1962 |
+
"step": 3210
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 3.14,
|
| 1966 |
+
"learning_rate": 3.987607877849082e-06,
|
| 1967 |
+
"loss": 0.2012,
|
| 1968 |
+
"step": 3220
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 3.14,
|
| 1972 |
+
"learning_rate": 3.96547908829387e-06,
|
| 1973 |
+
"loss": 0.1827,
|
| 1974 |
+
"step": 3230
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 3.14,
|
| 1978 |
+
"learning_rate": 3.9433502987386595e-06,
|
| 1979 |
+
"loss": 0.2234,
|
| 1980 |
+
"step": 3240
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 3.15,
|
| 1984 |
+
"learning_rate": 3.9212215091834485e-06,
|
| 1985 |
+
"loss": 0.1809,
|
| 1986 |
+
"step": 3250
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 3.15,
|
| 1990 |
+
"learning_rate": 3.899092719628237e-06,
|
| 1991 |
+
"loss": 0.1309,
|
| 1992 |
+
"step": 3260
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 3.15,
|
| 1996 |
+
"learning_rate": 3.876963930073026e-06,
|
| 1997 |
+
"loss": 0.1616,
|
| 1998 |
+
"step": 3270
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 3.15,
|
| 2002 |
+
"learning_rate": 3.854835140517814e-06,
|
| 2003 |
+
"loss": 0.2984,
|
| 2004 |
+
"step": 3280
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 3.15,
|
| 2008 |
+
"learning_rate": 3.832706350962603e-06,
|
| 2009 |
+
"loss": 0.0939,
|
| 2010 |
+
"step": 3290
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 3.16,
|
| 2014 |
+
"learning_rate": 3.8105775614073913e-06,
|
| 2015 |
+
"loss": 0.1858,
|
| 2016 |
+
"step": 3300
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 3.16,
|
| 2020 |
+
"learning_rate": 3.78844877185218e-06,
|
| 2021 |
+
"loss": 0.2291,
|
| 2022 |
+
"step": 3310
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 3.16,
|
| 2026 |
+
"learning_rate": 3.7663199822969686e-06,
|
| 2027 |
+
"loss": 0.2996,
|
| 2028 |
+
"step": 3320
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 3.16,
|
| 2032 |
+
"learning_rate": 3.7441911927417572e-06,
|
| 2033 |
+
"loss": 0.1286,
|
| 2034 |
+
"step": 3330
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 3.16,
|
| 2038 |
+
"learning_rate": 3.722062403186546e-06,
|
| 2039 |
+
"loss": 0.0735,
|
| 2040 |
+
"step": 3340
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 3.17,
|
| 2044 |
+
"learning_rate": 3.6999336136313345e-06,
|
| 2045 |
+
"loss": 0.0697,
|
| 2046 |
+
"step": 3350
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 3.17,
|
| 2050 |
+
"eval_accuracy": 0.8447667581340651,
|
| 2051 |
+
"eval_loss": 0.6156530976295471,
|
| 2052 |
+
"eval_runtime": 478.3766,
|
| 2053 |
+
"eval_samples_per_second": 5.333,
|
| 2054 |
+
"eval_steps_per_second": 0.445,
|
| 2055 |
+
"step": 3352
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 4.0,
|
| 2059 |
+
"learning_rate": 3.677804824076123e-06,
|
| 2060 |
+
"loss": 0.0433,
|
| 2061 |
+
"step": 3360
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 4.0,
|
| 2065 |
+
"learning_rate": 3.655676034520912e-06,
|
| 2066 |
+
"loss": 0.3206,
|
| 2067 |
+
"step": 3370
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 4.01,
|
| 2071 |
+
"learning_rate": 3.6335472449657004e-06,
|
| 2072 |
+
"loss": 0.228,
|
| 2073 |
+
"step": 3380
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 4.01,
|
| 2077 |
+
"learning_rate": 3.611418455410489e-06,
|
| 2078 |
+
"loss": 0.2148,
|
| 2079 |
+
"step": 3390
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 4.01,
|
| 2083 |
+
"learning_rate": 3.5892896658552777e-06,
|
| 2084 |
+
"loss": 0.1528,
|
| 2085 |
+
"step": 3400
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 4.01,
|
| 2089 |
+
"learning_rate": 3.5671608763000664e-06,
|
| 2090 |
+
"loss": 0.2124,
|
| 2091 |
+
"step": 3410
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 4.01,
|
| 2095 |
+
"learning_rate": 3.545032086744855e-06,
|
| 2096 |
+
"loss": 0.0536,
|
| 2097 |
+
"step": 3420
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 4.02,
|
| 2101 |
+
"learning_rate": 3.5229032971896437e-06,
|
| 2102 |
+
"loss": 0.2198,
|
| 2103 |
+
"step": 3430
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 4.02,
|
| 2107 |
+
"learning_rate": 3.500774507634433e-06,
|
| 2108 |
+
"loss": 0.1849,
|
| 2109 |
+
"step": 3440
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 4.02,
|
| 2113 |
+
"learning_rate": 3.478645718079222e-06,
|
| 2114 |
+
"loss": 0.2319,
|
| 2115 |
+
"step": 3450
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 4.02,
|
| 2119 |
+
"learning_rate": 3.45651692852401e-06,
|
| 2120 |
+
"loss": 0.0702,
|
| 2121 |
+
"step": 3460
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 4.02,
|
| 2125 |
+
"learning_rate": 3.4343881389687987e-06,
|
| 2126 |
+
"loss": 0.1208,
|
| 2127 |
+
"step": 3470
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 4.03,
|
| 2131 |
+
"learning_rate": 3.4122593494135873e-06,
|
| 2132 |
+
"loss": 0.108,
|
| 2133 |
+
"step": 3480
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 4.03,
|
| 2137 |
+
"learning_rate": 3.390130559858376e-06,
|
| 2138 |
+
"loss": 0.2169,
|
| 2139 |
+
"step": 3490
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 4.03,
|
| 2143 |
+
"learning_rate": 3.3680017703031646e-06,
|
| 2144 |
+
"loss": 0.1446,
|
| 2145 |
+
"step": 3500
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 4.03,
|
| 2149 |
+
"learning_rate": 3.3458729807479532e-06,
|
| 2150 |
+
"loss": 0.1869,
|
| 2151 |
+
"step": 3510
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 4.03,
|
| 2155 |
+
"learning_rate": 3.323744191192742e-06,
|
| 2156 |
+
"loss": 0.3277,
|
| 2157 |
+
"step": 3520
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 4.04,
|
| 2161 |
+
"learning_rate": 3.3016154016375305e-06,
|
| 2162 |
+
"loss": 0.2244,
|
| 2163 |
+
"step": 3530
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 4.04,
|
| 2167 |
+
"learning_rate": 3.279486612082319e-06,
|
| 2168 |
+
"loss": 0.1608,
|
| 2169 |
+
"step": 3540
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 4.04,
|
| 2173 |
+
"learning_rate": 3.257357822527108e-06,
|
| 2174 |
+
"loss": 0.1947,
|
| 2175 |
+
"step": 3550
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 4.04,
|
| 2179 |
+
"learning_rate": 3.2352290329718964e-06,
|
| 2180 |
+
"loss": 0.2295,
|
| 2181 |
+
"step": 3560
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 4.04,
|
| 2185 |
+
"learning_rate": 3.213100243416685e-06,
|
| 2186 |
+
"loss": 0.1933,
|
| 2187 |
+
"step": 3570
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 4.05,
|
| 2191 |
+
"learning_rate": 3.1909714538614737e-06,
|
| 2192 |
+
"loss": 0.1222,
|
| 2193 |
+
"step": 3580
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 4.05,
|
| 2197 |
+
"learning_rate": 3.1688426643062624e-06,
|
| 2198 |
+
"loss": 0.1036,
|
| 2199 |
+
"step": 3590
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 4.05,
|
| 2203 |
+
"learning_rate": 3.146713874751051e-06,
|
| 2204 |
+
"loss": 0.1496,
|
| 2205 |
+
"step": 3600
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 4.05,
|
| 2209 |
+
"learning_rate": 3.1245850851958396e-06,
|
| 2210 |
+
"loss": 0.0575,
|
| 2211 |
+
"step": 3610
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 4.05,
|
| 2215 |
+
"learning_rate": 3.1024562956406283e-06,
|
| 2216 |
+
"loss": 0.0818,
|
| 2217 |
+
"step": 3620
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 4.06,
|
| 2221 |
+
"learning_rate": 3.0803275060854178e-06,
|
| 2222 |
+
"loss": 0.216,
|
| 2223 |
+
"step": 3630
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 4.06,
|
| 2227 |
+
"learning_rate": 3.0581987165302064e-06,
|
| 2228 |
+
"loss": 0.097,
|
| 2229 |
+
"step": 3640
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 4.06,
|
| 2233 |
+
"learning_rate": 3.036069926974995e-06,
|
| 2234 |
+
"loss": 0.0846,
|
| 2235 |
+
"step": 3650
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 4.06,
|
| 2239 |
+
"learning_rate": 3.0139411374197837e-06,
|
| 2240 |
+
"loss": 0.178,
|
| 2241 |
+
"step": 3660
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 4.06,
|
| 2245 |
+
"learning_rate": 2.9918123478645723e-06,
|
| 2246 |
+
"loss": 0.3417,
|
| 2247 |
+
"step": 3670
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 4.07,
|
| 2251 |
+
"learning_rate": 2.969683558309361e-06,
|
| 2252 |
+
"loss": 0.1691,
|
| 2253 |
+
"step": 3680
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 4.07,
|
| 2257 |
+
"learning_rate": 2.9475547687541496e-06,
|
| 2258 |
+
"loss": 0.2325,
|
| 2259 |
+
"step": 3690
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 4.07,
|
| 2263 |
+
"learning_rate": 2.9254259791989383e-06,
|
| 2264 |
+
"loss": 0.0148,
|
| 2265 |
+
"step": 3700
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 4.07,
|
| 2269 |
+
"learning_rate": 2.903297189643727e-06,
|
| 2270 |
+
"loss": 0.3002,
|
| 2271 |
+
"step": 3710
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 4.07,
|
| 2275 |
+
"learning_rate": 2.8811684000885156e-06,
|
| 2276 |
+
"loss": 0.1167,
|
| 2277 |
+
"step": 3720
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 4.08,
|
| 2281 |
+
"learning_rate": 2.859039610533304e-06,
|
| 2282 |
+
"loss": 0.3281,
|
| 2283 |
+
"step": 3730
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 4.08,
|
| 2287 |
+
"learning_rate": 2.836910820978093e-06,
|
| 2288 |
+
"loss": 0.1938,
|
| 2289 |
+
"step": 3740
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 4.08,
|
| 2293 |
+
"learning_rate": 2.8147820314228815e-06,
|
| 2294 |
+
"loss": 0.1634,
|
| 2295 |
+
"step": 3750
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 4.08,
|
| 2299 |
+
"learning_rate": 2.79265324186767e-06,
|
| 2300 |
+
"loss": 0.1627,
|
| 2301 |
+
"step": 3760
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 4.08,
|
| 2305 |
+
"learning_rate": 2.7705244523124588e-06,
|
| 2306 |
+
"loss": 0.1519,
|
| 2307 |
+
"step": 3770
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 4.09,
|
| 2311 |
+
"learning_rate": 2.7483956627572474e-06,
|
| 2312 |
+
"loss": 0.0827,
|
| 2313 |
+
"step": 3780
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 4.09,
|
| 2317 |
+
"learning_rate": 2.726266873202036e-06,
|
| 2318 |
+
"loss": 0.181,
|
| 2319 |
+
"step": 3790
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 4.09,
|
| 2323 |
+
"learning_rate": 2.7041380836468247e-06,
|
| 2324 |
+
"loss": 0.1907,
|
| 2325 |
+
"step": 3800
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 4.09,
|
| 2329 |
+
"learning_rate": 2.6820092940916133e-06,
|
| 2330 |
+
"loss": 0.1519,
|
| 2331 |
+
"step": 3810
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 4.09,
|
| 2335 |
+
"learning_rate": 2.6598805045364024e-06,
|
| 2336 |
+
"loss": 0.2911,
|
| 2337 |
+
"step": 3820
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 4.1,
|
| 2341 |
+
"learning_rate": 2.637751714981191e-06,
|
| 2342 |
+
"loss": 0.2609,
|
| 2343 |
+
"step": 3830
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 4.1,
|
| 2347 |
+
"learning_rate": 2.6156229254259797e-06,
|
| 2348 |
+
"loss": 0.1138,
|
| 2349 |
+
"step": 3840
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 4.1,
|
| 2353 |
+
"learning_rate": 2.5934941358707683e-06,
|
| 2354 |
+
"loss": 0.1796,
|
| 2355 |
+
"step": 3850
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 4.1,
|
| 2359 |
+
"learning_rate": 2.571365346315557e-06,
|
| 2360 |
+
"loss": 0.1406,
|
| 2361 |
+
"step": 3860
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 4.1,
|
| 2365 |
+
"learning_rate": 2.5492365567603456e-06,
|
| 2366 |
+
"loss": 0.3324,
|
| 2367 |
+
"step": 3870
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 4.11,
|
| 2371 |
+
"learning_rate": 2.5271077672051343e-06,
|
| 2372 |
+
"loss": 0.1073,
|
| 2373 |
+
"step": 3880
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 4.11,
|
| 2377 |
+
"learning_rate": 2.504978977649923e-06,
|
| 2378 |
+
"loss": 0.1545,
|
| 2379 |
+
"step": 3890
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 4.11,
|
| 2383 |
+
"learning_rate": 2.4828501880947115e-06,
|
| 2384 |
+
"loss": 0.0673,
|
| 2385 |
+
"step": 3900
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 4.11,
|
| 2389 |
+
"learning_rate": 2.4607213985395e-06,
|
| 2390 |
+
"loss": 0.1819,
|
| 2391 |
+
"step": 3910
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 4.11,
|
| 2395 |
+
"learning_rate": 2.438592608984289e-06,
|
| 2396 |
+
"loss": 0.0896,
|
| 2397 |
+
"step": 3920
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 4.12,
|
| 2401 |
+
"learning_rate": 2.4164638194290775e-06,
|
| 2402 |
+
"loss": 0.1293,
|
| 2403 |
+
"step": 3930
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 4.12,
|
| 2407 |
+
"learning_rate": 2.394335029873866e-06,
|
| 2408 |
+
"loss": 0.2196,
|
| 2409 |
+
"step": 3940
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 4.12,
|
| 2413 |
+
"learning_rate": 2.3722062403186548e-06,
|
| 2414 |
+
"loss": 0.2235,
|
| 2415 |
+
"step": 3950
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 4.12,
|
| 2419 |
+
"learning_rate": 2.3500774507634434e-06,
|
| 2420 |
+
"loss": 0.1067,
|
| 2421 |
+
"step": 3960
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 4.12,
|
| 2425 |
+
"learning_rate": 2.327948661208232e-06,
|
| 2426 |
+
"loss": 0.2652,
|
| 2427 |
+
"step": 3970
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 4.13,
|
| 2431 |
+
"learning_rate": 2.3058198716530207e-06,
|
| 2432 |
+
"loss": 0.2255,
|
| 2433 |
+
"step": 3980
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 4.13,
|
| 2437 |
+
"learning_rate": 2.2836910820978093e-06,
|
| 2438 |
+
"loss": 0.2405,
|
| 2439 |
+
"step": 3990
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 4.13,
|
| 2443 |
+
"learning_rate": 2.261562292542598e-06,
|
| 2444 |
+
"loss": 0.137,
|
| 2445 |
+
"step": 4000
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 4.13,
|
| 2449 |
+
"learning_rate": 2.2394335029873866e-06,
|
| 2450 |
+
"loss": 0.133,
|
| 2451 |
+
"step": 4010
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 4.13,
|
| 2455 |
+
"learning_rate": 2.2173047134321752e-06,
|
| 2456 |
+
"loss": 0.1682,
|
| 2457 |
+
"step": 4020
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 4.14,
|
| 2461 |
+
"learning_rate": 2.195175923876964e-06,
|
| 2462 |
+
"loss": 0.2465,
|
| 2463 |
+
"step": 4030
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 4.14,
|
| 2467 |
+
"learning_rate": 2.1730471343217525e-06,
|
| 2468 |
+
"loss": 0.1982,
|
| 2469 |
+
"step": 4040
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 4.14,
|
| 2473 |
+
"learning_rate": 2.150918344766541e-06,
|
| 2474 |
+
"loss": 0.1348,
|
| 2475 |
+
"step": 4050
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 4.14,
|
| 2479 |
+
"learning_rate": 2.1287895552113302e-06,
|
| 2480 |
+
"loss": 0.1349,
|
| 2481 |
+
"step": 4060
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 4.14,
|
| 2485 |
+
"learning_rate": 2.106660765656119e-06,
|
| 2486 |
+
"loss": 0.1451,
|
| 2487 |
+
"step": 4070
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 4.14,
|
| 2491 |
+
"learning_rate": 2.0845319761009075e-06,
|
| 2492 |
+
"loss": 0.2781,
|
| 2493 |
+
"step": 4080
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 4.15,
|
| 2497 |
+
"learning_rate": 2.062403186545696e-06,
|
| 2498 |
+
"loss": 0.1213,
|
| 2499 |
+
"step": 4090
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 4.15,
|
| 2503 |
+
"learning_rate": 2.040274396990485e-06,
|
| 2504 |
+
"loss": 0.1548,
|
| 2505 |
+
"step": 4100
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 4.15,
|
| 2509 |
+
"learning_rate": 2.0181456074352735e-06,
|
| 2510 |
+
"loss": 0.1743,
|
| 2511 |
+
"step": 4110
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 4.15,
|
| 2515 |
+
"learning_rate": 1.996016817880062e-06,
|
| 2516 |
+
"loss": 0.163,
|
| 2517 |
+
"step": 4120
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 4.15,
|
| 2521 |
+
"learning_rate": 1.9738880283248507e-06,
|
| 2522 |
+
"loss": 0.1655,
|
| 2523 |
+
"step": 4130
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 4.16,
|
| 2527 |
+
"learning_rate": 1.9517592387696394e-06,
|
| 2528 |
+
"loss": 0.1672,
|
| 2529 |
+
"step": 4140
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 4.16,
|
| 2533 |
+
"learning_rate": 1.929630449214428e-06,
|
| 2534 |
+
"loss": 0.0966,
|
| 2535 |
+
"step": 4150
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 4.16,
|
| 2539 |
+
"learning_rate": 1.907501659659217e-06,
|
| 2540 |
+
"loss": 0.1906,
|
| 2541 |
+
"step": 4160
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 4.16,
|
| 2545 |
+
"learning_rate": 1.8853728701040055e-06,
|
| 2546 |
+
"loss": 0.2156,
|
| 2547 |
+
"step": 4170
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 4.16,
|
| 2551 |
+
"learning_rate": 1.8632440805487942e-06,
|
| 2552 |
+
"loss": 0.1546,
|
| 2553 |
+
"step": 4180
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 4.17,
|
| 2557 |
+
"learning_rate": 1.8411152909935828e-06,
|
| 2558 |
+
"loss": 0.2116,
|
| 2559 |
+
"step": 4190
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 4.17,
|
| 2563 |
+
"eval_accuracy": 0.8643669149353195,
|
| 2564 |
+
"eval_loss": 0.4997994899749756,
|
| 2565 |
+
"eval_runtime": 440.6246,
|
| 2566 |
+
"eval_samples_per_second": 5.79,
|
| 2567 |
+
"eval_steps_per_second": 0.483,
|
| 2568 |
+
"step": 4190
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 5.0,
|
| 2572 |
+
"learning_rate": 1.8189865014383714e-06,
|
| 2573 |
+
"loss": 0.1667,
|
| 2574 |
+
"step": 4200
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 5.0,
|
| 2578 |
+
"learning_rate": 1.79685771188316e-06,
|
| 2579 |
+
"loss": 0.396,
|
| 2580 |
+
"step": 4210
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 5.01,
|
| 2584 |
+
"learning_rate": 1.7747289223279487e-06,
|
| 2585 |
+
"loss": 0.12,
|
| 2586 |
+
"step": 4220
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 5.01,
|
| 2590 |
+
"learning_rate": 1.7526001327727374e-06,
|
| 2591 |
+
"loss": 0.1906,
|
| 2592 |
+
"step": 4230
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 5.01,
|
| 2596 |
+
"learning_rate": 1.730471343217526e-06,
|
| 2597 |
+
"loss": 0.3117,
|
| 2598 |
+
"step": 4240
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 5.01,
|
| 2602 |
+
"learning_rate": 1.7083425536623149e-06,
|
| 2603 |
+
"loss": 0.1781,
|
| 2604 |
+
"step": 4250
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 5.01,
|
| 2608 |
+
"learning_rate": 1.6862137641071035e-06,
|
| 2609 |
+
"loss": 0.0814,
|
| 2610 |
+
"step": 4260
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 5.02,
|
| 2614 |
+
"learning_rate": 1.6640849745518922e-06,
|
| 2615 |
+
"loss": 0.1559,
|
| 2616 |
+
"step": 4270
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 5.02,
|
| 2620 |
+
"learning_rate": 1.6419561849966808e-06,
|
| 2621 |
+
"loss": 0.2145,
|
| 2622 |
+
"step": 4280
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 5.02,
|
| 2626 |
+
"learning_rate": 1.6198273954414694e-06,
|
| 2627 |
+
"loss": 0.1389,
|
| 2628 |
+
"step": 4290
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 5.02,
|
| 2632 |
+
"learning_rate": 1.597698605886258e-06,
|
| 2633 |
+
"loss": 0.0715,
|
| 2634 |
+
"step": 4300
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 5.02,
|
| 2638 |
+
"learning_rate": 1.5755698163310467e-06,
|
| 2639 |
+
"loss": 0.2595,
|
| 2640 |
+
"step": 4310
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 5.03,
|
| 2644 |
+
"learning_rate": 1.5534410267758354e-06,
|
| 2645 |
+
"loss": 0.119,
|
| 2646 |
+
"step": 4320
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 5.03,
|
| 2650 |
+
"learning_rate": 1.531312237220624e-06,
|
| 2651 |
+
"loss": 0.0869,
|
| 2652 |
+
"step": 4330
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 5.03,
|
| 2656 |
+
"learning_rate": 1.5091834476654126e-06,
|
| 2657 |
+
"loss": 0.1115,
|
| 2658 |
+
"step": 4340
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 5.03,
|
| 2662 |
+
"learning_rate": 1.4870546581102017e-06,
|
| 2663 |
+
"loss": 0.249,
|
| 2664 |
+
"step": 4350
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 5.03,
|
| 2668 |
+
"learning_rate": 1.4649258685549901e-06,
|
| 2669 |
+
"loss": 0.1659,
|
| 2670 |
+
"step": 4360
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 5.04,
|
| 2674 |
+
"learning_rate": 1.4427970789997788e-06,
|
| 2675 |
+
"loss": 0.1521,
|
| 2676 |
+
"step": 4370
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 5.04,
|
| 2680 |
+
"learning_rate": 1.4206682894445674e-06,
|
| 2681 |
+
"loss": 0.1376,
|
| 2682 |
+
"step": 4380
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 5.04,
|
| 2686 |
+
"learning_rate": 1.398539499889356e-06,
|
| 2687 |
+
"loss": 0.1015,
|
| 2688 |
+
"step": 4390
|
| 2689 |
+
},
|
| 2690 |
+
{
|
| 2691 |
+
"epoch": 5.04,
|
| 2692 |
+
"learning_rate": 1.3764107103341447e-06,
|
| 2693 |
+
"loss": 0.2383,
|
| 2694 |
+
"step": 4400
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 5.04,
|
| 2698 |
+
"learning_rate": 1.3542819207789334e-06,
|
| 2699 |
+
"loss": 0.1839,
|
| 2700 |
+
"step": 4410
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 5.05,
|
| 2704 |
+
"learning_rate": 1.332153131223722e-06,
|
| 2705 |
+
"loss": 0.0749,
|
| 2706 |
+
"step": 4420
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 5.05,
|
| 2710 |
+
"learning_rate": 1.3100243416685106e-06,
|
| 2711 |
+
"loss": 0.2221,
|
| 2712 |
+
"step": 4430
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 5.05,
|
| 2716 |
+
"learning_rate": 1.2878955521132997e-06,
|
| 2717 |
+
"loss": 0.1028,
|
| 2718 |
+
"step": 4440
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 5.05,
|
| 2722 |
+
"learning_rate": 1.2657667625580883e-06,
|
| 2723 |
+
"loss": 0.1556,
|
| 2724 |
+
"step": 4450
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 5.05,
|
| 2728 |
+
"learning_rate": 1.2436379730028768e-06,
|
| 2729 |
+
"loss": 0.3584,
|
| 2730 |
+
"step": 4460
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 5.06,
|
| 2734 |
+
"learning_rate": 1.2215091834476656e-06,
|
| 2735 |
+
"loss": 0.1174,
|
| 2736 |
+
"step": 4470
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 5.06,
|
| 2740 |
+
"learning_rate": 1.1993803938924543e-06,
|
| 2741 |
+
"loss": 0.2224,
|
| 2742 |
+
"step": 4480
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 5.06,
|
| 2746 |
+
"learning_rate": 1.177251604337243e-06,
|
| 2747 |
+
"loss": 0.0997,
|
| 2748 |
+
"step": 4490
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 5.06,
|
| 2752 |
+
"learning_rate": 1.1551228147820316e-06,
|
| 2753 |
+
"loss": 0.1691,
|
| 2754 |
+
"step": 4500
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 5.06,
|
| 2758 |
+
"learning_rate": 1.1329940252268202e-06,
|
| 2759 |
+
"loss": 0.1221,
|
| 2760 |
+
"step": 4510
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 5.07,
|
| 2764 |
+
"learning_rate": 1.1108652356716088e-06,
|
| 2765 |
+
"loss": 0.1768,
|
| 2766 |
+
"step": 4520
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 5.07,
|
| 2770 |
+
"learning_rate": 1.0887364461163975e-06,
|
| 2771 |
+
"loss": 0.1979,
|
| 2772 |
+
"step": 4530
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 5.07,
|
| 2776 |
+
"learning_rate": 1.0666076565611861e-06,
|
| 2777 |
+
"loss": 0.2115,
|
| 2778 |
+
"step": 4540
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 5.07,
|
| 2782 |
+
"learning_rate": 1.0444788670059748e-06,
|
| 2783 |
+
"loss": 0.1275,
|
| 2784 |
+
"step": 4550
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 5.07,
|
| 2788 |
+
"learning_rate": 1.0223500774507636e-06,
|
| 2789 |
+
"loss": 0.1195,
|
| 2790 |
+
"step": 4560
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 5.08,
|
| 2794 |
+
"learning_rate": 1.0002212878955523e-06,
|
| 2795 |
+
"loss": 0.0997,
|
| 2796 |
+
"step": 4570
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 5.08,
|
| 2800 |
+
"learning_rate": 9.78092498340341e-07,
|
| 2801 |
+
"loss": 0.1918,
|
| 2802 |
+
"step": 4580
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 5.08,
|
| 2806 |
+
"learning_rate": 9.559637087851296e-07,
|
| 2807 |
+
"loss": 0.1582,
|
| 2808 |
+
"step": 4590
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 5.08,
|
| 2812 |
+
"learning_rate": 9.338349192299182e-07,
|
| 2813 |
+
"loss": 0.2074,
|
| 2814 |
+
"step": 4600
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 5.08,
|
| 2818 |
+
"learning_rate": 9.117061296747069e-07,
|
| 2819 |
+
"loss": 0.3362,
|
| 2820 |
+
"step": 4610
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 5.09,
|
| 2824 |
+
"learning_rate": 8.895773401194956e-07,
|
| 2825 |
+
"loss": 0.2022,
|
| 2826 |
+
"step": 4620
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 5.09,
|
| 2830 |
+
"learning_rate": 8.674485505642842e-07,
|
| 2831 |
+
"loss": 0.1409,
|
| 2832 |
+
"step": 4630
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 5.09,
|
| 2836 |
+
"learning_rate": 8.453197610090729e-07,
|
| 2837 |
+
"loss": 0.1028,
|
| 2838 |
+
"step": 4640
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 5.09,
|
| 2842 |
+
"learning_rate": 8.231909714538615e-07,
|
| 2843 |
+
"loss": 0.0855,
|
| 2844 |
+
"step": 4650
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 5.09,
|
| 2848 |
+
"learning_rate": 8.010621818986503e-07,
|
| 2849 |
+
"loss": 0.1453,
|
| 2850 |
+
"step": 4660
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 5.1,
|
| 2854 |
+
"learning_rate": 7.789333923434389e-07,
|
| 2855 |
+
"loss": 0.1247,
|
| 2856 |
+
"step": 4670
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 5.1,
|
| 2860 |
+
"learning_rate": 7.568046027882275e-07,
|
| 2861 |
+
"loss": 0.0749,
|
| 2862 |
+
"step": 4680
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 5.1,
|
| 2866 |
+
"learning_rate": 7.346758132330162e-07,
|
| 2867 |
+
"loss": 0.1832,
|
| 2868 |
+
"step": 4690
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 5.1,
|
| 2872 |
+
"learning_rate": 7.125470236778048e-07,
|
| 2873 |
+
"loss": 0.0402,
|
| 2874 |
+
"step": 4700
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 5.1,
|
| 2878 |
+
"learning_rate": 6.904182341225936e-07,
|
| 2879 |
+
"loss": 0.1689,
|
| 2880 |
+
"step": 4710
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 5.11,
|
| 2884 |
+
"learning_rate": 6.682894445673822e-07,
|
| 2885 |
+
"loss": 0.1668,
|
| 2886 |
+
"step": 4720
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 5.11,
|
| 2890 |
+
"learning_rate": 6.461606550121709e-07,
|
| 2891 |
+
"loss": 0.0635,
|
| 2892 |
+
"step": 4730
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 5.11,
|
| 2896 |
+
"learning_rate": 6.240318654569596e-07,
|
| 2897 |
+
"loss": 0.3186,
|
| 2898 |
+
"step": 4740
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 5.11,
|
| 2902 |
+
"learning_rate": 6.019030759017483e-07,
|
| 2903 |
+
"loss": 0.0793,
|
| 2904 |
+
"step": 4750
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 5.11,
|
| 2908 |
+
"learning_rate": 5.797742863465369e-07,
|
| 2909 |
+
"loss": 0.1578,
|
| 2910 |
+
"step": 4760
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 5.12,
|
| 2914 |
+
"learning_rate": 5.576454967913255e-07,
|
| 2915 |
+
"loss": 0.0247,
|
| 2916 |
+
"step": 4770
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 5.12,
|
| 2920 |
+
"learning_rate": 5.355167072361142e-07,
|
| 2921 |
+
"loss": 0.1533,
|
| 2922 |
+
"step": 4780
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 5.12,
|
| 2926 |
+
"learning_rate": 5.133879176809029e-07,
|
| 2927 |
+
"loss": 0.089,
|
| 2928 |
+
"step": 4790
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 5.12,
|
| 2932 |
+
"learning_rate": 4.912591281256916e-07,
|
| 2933 |
+
"loss": 0.1192,
|
| 2934 |
+
"step": 4800
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 5.12,
|
| 2938 |
+
"learning_rate": 4.6913033857048026e-07,
|
| 2939 |
+
"loss": 0.1591,
|
| 2940 |
+
"step": 4810
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 5.13,
|
| 2944 |
+
"learning_rate": 4.470015490152689e-07,
|
| 2945 |
+
"loss": 0.169,
|
| 2946 |
+
"step": 4820
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 5.13,
|
| 2950 |
+
"learning_rate": 4.2487275946005755e-07,
|
| 2951 |
+
"loss": 0.0843,
|
| 2952 |
+
"step": 4830
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 5.13,
|
| 2956 |
+
"learning_rate": 4.0274396990484624e-07,
|
| 2957 |
+
"loss": 0.1278,
|
| 2958 |
+
"step": 4840
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 5.13,
|
| 2962 |
+
"learning_rate": 3.806151803496349e-07,
|
| 2963 |
+
"loss": 0.1355,
|
| 2964 |
+
"step": 4850
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 5.13,
|
| 2968 |
+
"learning_rate": 3.584863907944236e-07,
|
| 2969 |
+
"loss": 0.1263,
|
| 2970 |
+
"step": 4860
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 5.14,
|
| 2974 |
+
"learning_rate": 3.363576012392122e-07,
|
| 2975 |
+
"loss": 0.1327,
|
| 2976 |
+
"step": 4870
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 5.14,
|
| 2980 |
+
"learning_rate": 3.1422881168400087e-07,
|
| 2981 |
+
"loss": 0.1344,
|
| 2982 |
+
"step": 4880
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 5.14,
|
| 2986 |
+
"learning_rate": 2.9210002212878956e-07,
|
| 2987 |
+
"loss": 0.0959,
|
| 2988 |
+
"step": 4890
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 5.14,
|
| 2992 |
+
"learning_rate": 2.6997123257357826e-07,
|
| 2993 |
+
"loss": 0.1882,
|
| 2994 |
+
"step": 4900
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 5.14,
|
| 2998 |
+
"learning_rate": 2.478424430183669e-07,
|
| 2999 |
+
"loss": 0.2254,
|
| 3000 |
+
"step": 4910
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 5.15,
|
| 3004 |
+
"learning_rate": 2.2571365346315557e-07,
|
| 3005 |
+
"loss": 0.0817,
|
| 3006 |
+
"step": 4920
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 5.15,
|
| 3010 |
+
"learning_rate": 2.0358486390794424e-07,
|
| 3011 |
+
"loss": 0.1513,
|
| 3012 |
+
"step": 4930
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 5.15,
|
| 3016 |
+
"learning_rate": 1.8145607435273293e-07,
|
| 3017 |
+
"loss": 0.0656,
|
| 3018 |
+
"step": 4940
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 5.15,
|
| 3022 |
+
"learning_rate": 1.593272847975216e-07,
|
| 3023 |
+
"loss": 0.1388,
|
| 3024 |
+
"step": 4950
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 5.15,
|
| 3028 |
+
"learning_rate": 1.3719849524231024e-07,
|
| 3029 |
+
"loss": 0.2887,
|
| 3030 |
+
"step": 4960
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 5.16,
|
| 3034 |
+
"learning_rate": 1.1506970568709893e-07,
|
| 3035 |
+
"loss": 0.2351,
|
| 3036 |
+
"step": 4970
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 5.16,
|
| 3040 |
+
"learning_rate": 9.294091613188761e-08,
|
| 3041 |
+
"loss": 0.1093,
|
| 3042 |
+
"step": 4980
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 5.16,
|
| 3046 |
+
"learning_rate": 7.081212657667626e-08,
|
| 3047 |
+
"loss": 0.231,
|
| 3048 |
+
"step": 4990
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 5.16,
|
| 3052 |
+
"learning_rate": 4.868333702146493e-08,
|
| 3053 |
+
"loss": 0.0945,
|
| 3054 |
+
"step": 5000
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 5.16,
|
| 3058 |
+
"learning_rate": 2.6554547466253597e-08,
|
| 3059 |
+
"loss": 0.119,
|
| 3060 |
+
"step": 5010
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 5.17,
|
| 3064 |
+
"learning_rate": 4.4257579110422664e-09,
|
| 3065 |
+
"loss": 0.1655,
|
| 3066 |
+
"step": 5020
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 5.17,
|
| 3070 |
+
"eval_accuracy": 0.8541748333986672,
|
| 3071 |
+
"eval_loss": 0.5614621043205261,
|
| 3072 |
+
"eval_runtime": 458.2183,
|
| 3073 |
+
"eval_samples_per_second": 5.567,
|
| 3074 |
+
"eval_steps_per_second": 0.465,
|
| 3075 |
+
"step": 5022
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 5.17,
|
| 3079 |
+
"step": 5022,
|
| 3080 |
+
"total_flos": 7.508660609697782e+19,
|
| 3081 |
+
"train_loss": 0.24504626986812852,
|
| 3082 |
+
"train_runtime": 16648.2238,
|
| 3083 |
+
"train_samples_per_second": 3.62,
|
| 3084 |
+
"train_steps_per_second": 0.302
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 5.17,
|
| 3088 |
+
"eval_accuracy": 0.8643669149353195,
|
| 3089 |
+
"eval_loss": 0.4997994899749756,
|
| 3090 |
+
"eval_runtime": 558.362,
|
| 3091 |
+
"eval_samples_per_second": 4.569,
|
| 3092 |
+
"eval_steps_per_second": 0.381,
|
| 3093 |
+
"step": 5022
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 5.17,
|
| 3097 |
+
"eval_accuracy": 0.5961538461538461,
|
| 3098 |
+
"eval_loss": 1.6456845998764038,
|
| 3099 |
+
"eval_runtime": 40.3879,
|
| 3100 |
+
"eval_samples_per_second": 2.575,
|
| 3101 |
+
"eval_steps_per_second": 0.223,
|
| 3102 |
+
"step": 5022
|
| 3103 |
+
}
|
| 3104 |
+
],
|
| 3105 |
+
"logging_steps": 10,
|
| 3106 |
+
"max_steps": 5022,
|
| 3107 |
+
"num_train_epochs": 9223372036854775807,
|
| 3108 |
+
"save_steps": 500,
|
| 3109 |
+
"total_flos": 7.508660609697782e+19,
|
| 3110 |
+
"trial_name": null,
|
| 3111 |
+
"trial_params": null
|
| 3112 |
+
}
|