End of training
Browse files- README.md +2 -2
- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +2998 -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 [microsoft/swinv2-large-patch4-window12to16-192to256-22kto1k-ft](https://huggingface.co/microsoft/swinv2-large-patch4-window12to16-192to256-22kto1k-ft) on the None dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss: 0.
|
| 21 |
-
- Accuracy: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
|
|
| 17 |
|
| 18 |
This model is a fine-tuned version of [microsoft/swinv2-large-patch4-window12to16-192to256-22kto1k-ft](https://huggingface.co/microsoft/swinv2-large-patch4-window12to16-192to256-22kto1k-ft) on the None dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0161
|
| 21 |
+
- Accuracy: 0.9972
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"eval_accuracy": 0.9971724787935909,
|
| 4 |
+
"eval_loss": 0.01605476438999176,
|
| 5 |
+
"eval_runtime": 89.3043,
|
| 6 |
+
"eval_samples_per_second": 23.761,
|
| 7 |
+
"eval_steps_per_second": 2.979,
|
| 8 |
+
"total_flos": 1.7585593436847735e+19,
|
| 9 |
+
"train_loss": 0.12566840387472886,
|
| 10 |
+
"train_runtime": 12106.0173,
|
| 11 |
+
"train_samples_per_second": 6.308,
|
| 12 |
+
"train_steps_per_second": 0.197
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"eval_accuracy": 0.9971724787935909,
|
| 4 |
+
"eval_loss": 0.01605476438999176,
|
| 5 |
+
"eval_runtime": 89.3043,
|
| 6 |
+
"eval_samples_per_second": 23.761,
|
| 7 |
+
"eval_steps_per_second": 2.979
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"total_flos": 1.7585593436847735e+19,
|
| 4 |
+
"train_loss": 0.12566840387472886,
|
| 5 |
+
"train_runtime": 12106.0173,
|
| 6 |
+
"train_samples_per_second": 6.308,
|
| 7 |
+
"train_steps_per_second": 0.197
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2998 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9971724787935909,
|
| 3 |
+
"best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-BreastCancer-BreakHis-AH-Shuff-3/checkpoint-2189",
|
| 4 |
+
"epoch": 12.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2388,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03,
|
| 13 |
+
"learning_rate": 2.0938023450586265e-07,
|
| 14 |
+
"loss": 0.7386,
|
| 15 |
+
"step": 5
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.05,
|
| 19 |
+
"learning_rate": 4.187604690117253e-07,
|
| 20 |
+
"loss": 0.7211,
|
| 21 |
+
"step": 10
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.08,
|
| 25 |
+
"learning_rate": 6.28140703517588e-07,
|
| 26 |
+
"loss": 0.7177,
|
| 27 |
+
"step": 15
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.1,
|
| 31 |
+
"learning_rate": 8.375209380234506e-07,
|
| 32 |
+
"loss": 0.7531,
|
| 33 |
+
"step": 20
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.13,
|
| 37 |
+
"learning_rate": 1.0469011725293133e-06,
|
| 38 |
+
"loss": 0.6788,
|
| 39 |
+
"step": 25
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.15,
|
| 43 |
+
"learning_rate": 1.256281407035176e-06,
|
| 44 |
+
"loss": 0.7,
|
| 45 |
+
"step": 30
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.18,
|
| 49 |
+
"learning_rate": 1.4656616415410386e-06,
|
| 50 |
+
"loss": 0.6658,
|
| 51 |
+
"step": 35
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.2,
|
| 55 |
+
"learning_rate": 1.6750418760469012e-06,
|
| 56 |
+
"loss": 0.6642,
|
| 57 |
+
"step": 40
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.23,
|
| 61 |
+
"learning_rate": 1.8844221105527638e-06,
|
| 62 |
+
"loss": 0.641,
|
| 63 |
+
"step": 45
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.25,
|
| 67 |
+
"learning_rate": 2.0938023450586266e-06,
|
| 68 |
+
"loss": 0.6131,
|
| 69 |
+
"step": 50
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.28,
|
| 73 |
+
"learning_rate": 2.3031825795644893e-06,
|
| 74 |
+
"loss": 0.5781,
|
| 75 |
+
"step": 55
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.3,
|
| 79 |
+
"learning_rate": 2.512562814070352e-06,
|
| 80 |
+
"loss": 0.5835,
|
| 81 |
+
"step": 60
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.33,
|
| 85 |
+
"learning_rate": 2.7219430485762145e-06,
|
| 86 |
+
"loss": 0.5773,
|
| 87 |
+
"step": 65
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.35,
|
| 91 |
+
"learning_rate": 2.931323283082077e-06,
|
| 92 |
+
"loss": 0.55,
|
| 93 |
+
"step": 70
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.38,
|
| 97 |
+
"learning_rate": 3.1407035175879398e-06,
|
| 98 |
+
"loss": 0.4751,
|
| 99 |
+
"step": 75
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.4,
|
| 103 |
+
"learning_rate": 3.3500837520938024e-06,
|
| 104 |
+
"loss": 0.4837,
|
| 105 |
+
"step": 80
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.43,
|
| 109 |
+
"learning_rate": 3.559463986599665e-06,
|
| 110 |
+
"loss": 0.4874,
|
| 111 |
+
"step": 85
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.45,
|
| 115 |
+
"learning_rate": 3.7688442211055276e-06,
|
| 116 |
+
"loss": 0.4528,
|
| 117 |
+
"step": 90
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.48,
|
| 121 |
+
"learning_rate": 3.978224455611391e-06,
|
| 122 |
+
"loss": 0.4497,
|
| 123 |
+
"step": 95
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.5,
|
| 127 |
+
"learning_rate": 4.187604690117253e-06,
|
| 128 |
+
"loss": 0.407,
|
| 129 |
+
"step": 100
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.53,
|
| 133 |
+
"learning_rate": 4.396984924623116e-06,
|
| 134 |
+
"loss": 0.3557,
|
| 135 |
+
"step": 105
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.55,
|
| 139 |
+
"learning_rate": 4.6063651591289785e-06,
|
| 140 |
+
"loss": 0.3435,
|
| 141 |
+
"step": 110
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.58,
|
| 145 |
+
"learning_rate": 4.815745393634841e-06,
|
| 146 |
+
"loss": 0.334,
|
| 147 |
+
"step": 115
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.6,
|
| 151 |
+
"learning_rate": 5.025125628140704e-06,
|
| 152 |
+
"loss": 0.2579,
|
| 153 |
+
"step": 120
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.63,
|
| 157 |
+
"learning_rate": 5.234505862646566e-06,
|
| 158 |
+
"loss": 0.298,
|
| 159 |
+
"step": 125
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.65,
|
| 163 |
+
"learning_rate": 5.443886097152429e-06,
|
| 164 |
+
"loss": 0.2904,
|
| 165 |
+
"step": 130
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.68,
|
| 169 |
+
"learning_rate": 5.653266331658292e-06,
|
| 170 |
+
"loss": 0.2653,
|
| 171 |
+
"step": 135
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.7,
|
| 175 |
+
"learning_rate": 5.862646566164154e-06,
|
| 176 |
+
"loss": 0.2814,
|
| 177 |
+
"step": 140
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.73,
|
| 181 |
+
"learning_rate": 6.072026800670017e-06,
|
| 182 |
+
"loss": 0.2375,
|
| 183 |
+
"step": 145
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.75,
|
| 187 |
+
"learning_rate": 6.2814070351758795e-06,
|
| 188 |
+
"loss": 0.2557,
|
| 189 |
+
"step": 150
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.78,
|
| 193 |
+
"learning_rate": 6.490787269681742e-06,
|
| 194 |
+
"loss": 0.301,
|
| 195 |
+
"step": 155
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.8,
|
| 199 |
+
"learning_rate": 6.700167504187605e-06,
|
| 200 |
+
"loss": 0.2171,
|
| 201 |
+
"step": 160
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.83,
|
| 205 |
+
"learning_rate": 6.909547738693467e-06,
|
| 206 |
+
"loss": 0.1899,
|
| 207 |
+
"step": 165
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.85,
|
| 211 |
+
"learning_rate": 7.11892797319933e-06,
|
| 212 |
+
"loss": 0.2101,
|
| 213 |
+
"step": 170
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.88,
|
| 217 |
+
"learning_rate": 7.328308207705193e-06,
|
| 218 |
+
"loss": 0.2239,
|
| 219 |
+
"step": 175
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.9,
|
| 223 |
+
"learning_rate": 7.537688442211055e-06,
|
| 224 |
+
"loss": 0.2252,
|
| 225 |
+
"step": 180
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.93,
|
| 229 |
+
"learning_rate": 7.747068676716917e-06,
|
| 230 |
+
"loss": 0.2012,
|
| 231 |
+
"step": 185
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.95,
|
| 235 |
+
"learning_rate": 7.956448911222781e-06,
|
| 236 |
+
"loss": 0.1873,
|
| 237 |
+
"step": 190
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.98,
|
| 241 |
+
"learning_rate": 8.165829145728644e-06,
|
| 242 |
+
"loss": 0.2598,
|
| 243 |
+
"step": 195
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 1.0,
|
| 247 |
+
"eval_accuracy": 0.941564561734213,
|
| 248 |
+
"eval_loss": 0.1445263922214508,
|
| 249 |
+
"eval_runtime": 535.3563,
|
| 250 |
+
"eval_samples_per_second": 3.964,
|
| 251 |
+
"eval_steps_per_second": 0.497,
|
| 252 |
+
"step": 199
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 1.01,
|
| 256 |
+
"learning_rate": 8.375209380234507e-06,
|
| 257 |
+
"loss": 0.2636,
|
| 258 |
+
"step": 200
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.03,
|
| 262 |
+
"learning_rate": 8.58458961474037e-06,
|
| 263 |
+
"loss": 0.193,
|
| 264 |
+
"step": 205
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.06,
|
| 268 |
+
"learning_rate": 8.793969849246232e-06,
|
| 269 |
+
"loss": 0.2517,
|
| 270 |
+
"step": 210
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.08,
|
| 274 |
+
"learning_rate": 9.003350083752094e-06,
|
| 275 |
+
"loss": 0.1317,
|
| 276 |
+
"step": 215
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.11,
|
| 280 |
+
"learning_rate": 9.212730318257957e-06,
|
| 281 |
+
"loss": 0.2586,
|
| 282 |
+
"step": 220
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.13,
|
| 286 |
+
"learning_rate": 9.42211055276382e-06,
|
| 287 |
+
"loss": 0.199,
|
| 288 |
+
"step": 225
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.16,
|
| 292 |
+
"learning_rate": 9.631490787269682e-06,
|
| 293 |
+
"loss": 0.1048,
|
| 294 |
+
"step": 230
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.18,
|
| 298 |
+
"learning_rate": 9.840871021775545e-06,
|
| 299 |
+
"loss": 0.1902,
|
| 300 |
+
"step": 235
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.21,
|
| 304 |
+
"learning_rate": 1.0050251256281408e-05,
|
| 305 |
+
"loss": 0.2831,
|
| 306 |
+
"step": 240
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.23,
|
| 310 |
+
"learning_rate": 1.025963149078727e-05,
|
| 311 |
+
"loss": 0.279,
|
| 312 |
+
"step": 245
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.26,
|
| 316 |
+
"learning_rate": 1.0469011725293133e-05,
|
| 317 |
+
"loss": 0.3919,
|
| 318 |
+
"step": 250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.28,
|
| 322 |
+
"learning_rate": 1.0678391959798995e-05,
|
| 323 |
+
"loss": 0.2474,
|
| 324 |
+
"step": 255
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.31,
|
| 328 |
+
"learning_rate": 1.0887772194304858e-05,
|
| 329 |
+
"loss": 0.1412,
|
| 330 |
+
"step": 260
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.33,
|
| 334 |
+
"learning_rate": 1.109715242881072e-05,
|
| 335 |
+
"loss": 0.2199,
|
| 336 |
+
"step": 265
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.36,
|
| 340 |
+
"learning_rate": 1.1306532663316583e-05,
|
| 341 |
+
"loss": 0.1251,
|
| 342 |
+
"step": 270
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.38,
|
| 346 |
+
"learning_rate": 1.1515912897822446e-05,
|
| 347 |
+
"loss": 0.2052,
|
| 348 |
+
"step": 275
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.41,
|
| 352 |
+
"learning_rate": 1.1725293132328309e-05,
|
| 353 |
+
"loss": 0.1302,
|
| 354 |
+
"step": 280
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.43,
|
| 358 |
+
"learning_rate": 1.1934673366834171e-05,
|
| 359 |
+
"loss": 0.1979,
|
| 360 |
+
"step": 285
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.46,
|
| 364 |
+
"learning_rate": 1.2144053601340034e-05,
|
| 365 |
+
"loss": 0.1516,
|
| 366 |
+
"step": 290
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.48,
|
| 370 |
+
"learning_rate": 1.2353433835845896e-05,
|
| 371 |
+
"loss": 0.2435,
|
| 372 |
+
"step": 295
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.51,
|
| 376 |
+
"learning_rate": 1.2562814070351759e-05,
|
| 377 |
+
"loss": 0.1363,
|
| 378 |
+
"step": 300
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.53,
|
| 382 |
+
"learning_rate": 1.2772194304857622e-05,
|
| 383 |
+
"loss": 0.1761,
|
| 384 |
+
"step": 305
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.56,
|
| 388 |
+
"learning_rate": 1.2981574539363484e-05,
|
| 389 |
+
"loss": 0.1312,
|
| 390 |
+
"step": 310
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.58,
|
| 394 |
+
"learning_rate": 1.3190954773869347e-05,
|
| 395 |
+
"loss": 0.1731,
|
| 396 |
+
"step": 315
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.61,
|
| 400 |
+
"learning_rate": 1.340033500837521e-05,
|
| 401 |
+
"loss": 0.1808,
|
| 402 |
+
"step": 320
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.63,
|
| 406 |
+
"learning_rate": 1.3609715242881072e-05,
|
| 407 |
+
"loss": 0.2518,
|
| 408 |
+
"step": 325
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.66,
|
| 412 |
+
"learning_rate": 1.3819095477386935e-05,
|
| 413 |
+
"loss": 0.1414,
|
| 414 |
+
"step": 330
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.68,
|
| 418 |
+
"learning_rate": 1.4028475711892797e-05,
|
| 419 |
+
"loss": 0.1738,
|
| 420 |
+
"step": 335
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.71,
|
| 424 |
+
"learning_rate": 1.423785594639866e-05,
|
| 425 |
+
"loss": 0.2038,
|
| 426 |
+
"step": 340
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.73,
|
| 430 |
+
"learning_rate": 1.4447236180904523e-05,
|
| 431 |
+
"loss": 0.1958,
|
| 432 |
+
"step": 345
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.76,
|
| 436 |
+
"learning_rate": 1.4656616415410385e-05,
|
| 437 |
+
"loss": 0.1442,
|
| 438 |
+
"step": 350
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.78,
|
| 442 |
+
"learning_rate": 1.4865996649916248e-05,
|
| 443 |
+
"loss": 0.1938,
|
| 444 |
+
"step": 355
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.81,
|
| 448 |
+
"learning_rate": 1.507537688442211e-05,
|
| 449 |
+
"loss": 0.1803,
|
| 450 |
+
"step": 360
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.83,
|
| 454 |
+
"learning_rate": 1.528475711892797e-05,
|
| 455 |
+
"loss": 0.1918,
|
| 456 |
+
"step": 365
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.86,
|
| 460 |
+
"learning_rate": 1.5494137353433834e-05,
|
| 461 |
+
"loss": 0.2579,
|
| 462 |
+
"step": 370
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.88,
|
| 466 |
+
"learning_rate": 1.57035175879397e-05,
|
| 467 |
+
"loss": 0.2417,
|
| 468 |
+
"step": 375
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.91,
|
| 472 |
+
"learning_rate": 1.5912897822445563e-05,
|
| 473 |
+
"loss": 0.116,
|
| 474 |
+
"step": 380
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 1.93,
|
| 478 |
+
"learning_rate": 1.6122278056951425e-05,
|
| 479 |
+
"loss": 0.3688,
|
| 480 |
+
"step": 385
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.96,
|
| 484 |
+
"learning_rate": 1.6331658291457288e-05,
|
| 485 |
+
"loss": 0.1942,
|
| 486 |
+
"step": 390
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.98,
|
| 490 |
+
"learning_rate": 1.654103852596315e-05,
|
| 491 |
+
"loss": 0.194,
|
| 492 |
+
"step": 395
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 2.0,
|
| 496 |
+
"eval_accuracy": 0.9524033930254477,
|
| 497 |
+
"eval_loss": 0.11212717741727829,
|
| 498 |
+
"eval_runtime": 91.4664,
|
| 499 |
+
"eval_samples_per_second": 23.2,
|
| 500 |
+
"eval_steps_per_second": 2.908,
|
| 501 |
+
"step": 398
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.01,
|
| 505 |
+
"learning_rate": 1.6750418760469013e-05,
|
| 506 |
+
"loss": 0.1765,
|
| 507 |
+
"step": 400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.04,
|
| 511 |
+
"learning_rate": 1.6959798994974876e-05,
|
| 512 |
+
"loss": 0.1709,
|
| 513 |
+
"step": 405
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.06,
|
| 517 |
+
"learning_rate": 1.716917922948074e-05,
|
| 518 |
+
"loss": 0.1229,
|
| 519 |
+
"step": 410
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.09,
|
| 523 |
+
"learning_rate": 1.73785594639866e-05,
|
| 524 |
+
"loss": 0.0914,
|
| 525 |
+
"step": 415
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.11,
|
| 529 |
+
"learning_rate": 1.7587939698492464e-05,
|
| 530 |
+
"loss": 0.1755,
|
| 531 |
+
"step": 420
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.14,
|
| 535 |
+
"learning_rate": 1.7797319932998326e-05,
|
| 536 |
+
"loss": 0.2204,
|
| 537 |
+
"step": 425
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 2.16,
|
| 541 |
+
"learning_rate": 1.800670016750419e-05,
|
| 542 |
+
"loss": 0.1749,
|
| 543 |
+
"step": 430
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 2.19,
|
| 547 |
+
"learning_rate": 1.821608040201005e-05,
|
| 548 |
+
"loss": 0.0697,
|
| 549 |
+
"step": 435
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.21,
|
| 553 |
+
"learning_rate": 1.8425460636515914e-05,
|
| 554 |
+
"loss": 0.1144,
|
| 555 |
+
"step": 440
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.24,
|
| 559 |
+
"learning_rate": 1.8634840871021777e-05,
|
| 560 |
+
"loss": 0.1236,
|
| 561 |
+
"step": 445
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 2.26,
|
| 565 |
+
"learning_rate": 1.884422110552764e-05,
|
| 566 |
+
"loss": 0.2386,
|
| 567 |
+
"step": 450
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 2.29,
|
| 571 |
+
"learning_rate": 1.9053601340033502e-05,
|
| 572 |
+
"loss": 0.156,
|
| 573 |
+
"step": 455
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.31,
|
| 577 |
+
"learning_rate": 1.9262981574539365e-05,
|
| 578 |
+
"loss": 0.2102,
|
| 579 |
+
"step": 460
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.34,
|
| 583 |
+
"learning_rate": 1.9472361809045227e-05,
|
| 584 |
+
"loss": 0.1128,
|
| 585 |
+
"step": 465
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.36,
|
| 589 |
+
"learning_rate": 1.968174204355109e-05,
|
| 590 |
+
"loss": 0.112,
|
| 591 |
+
"step": 470
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.39,
|
| 595 |
+
"learning_rate": 1.9891122278056953e-05,
|
| 596 |
+
"loss": 0.1304,
|
| 597 |
+
"step": 475
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 2.41,
|
| 601 |
+
"learning_rate": 2.0100502512562815e-05,
|
| 602 |
+
"loss": 0.1724,
|
| 603 |
+
"step": 480
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 2.44,
|
| 607 |
+
"learning_rate": 2.0309882747068678e-05,
|
| 608 |
+
"loss": 0.3854,
|
| 609 |
+
"step": 485
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.46,
|
| 613 |
+
"learning_rate": 2.051926298157454e-05,
|
| 614 |
+
"loss": 0.2293,
|
| 615 |
+
"step": 490
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 2.49,
|
| 619 |
+
"learning_rate": 2.0728643216080403e-05,
|
| 620 |
+
"loss": 0.1072,
|
| 621 |
+
"step": 495
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.51,
|
| 625 |
+
"learning_rate": 2.0938023450586266e-05,
|
| 626 |
+
"loss": 0.1487,
|
| 627 |
+
"step": 500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.54,
|
| 631 |
+
"learning_rate": 2.1147403685092128e-05,
|
| 632 |
+
"loss": 0.1017,
|
| 633 |
+
"step": 505
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.56,
|
| 637 |
+
"learning_rate": 2.135678391959799e-05,
|
| 638 |
+
"loss": 0.1227,
|
| 639 |
+
"step": 510
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.59,
|
| 643 |
+
"learning_rate": 2.1566164154103854e-05,
|
| 644 |
+
"loss": 0.1068,
|
| 645 |
+
"step": 515
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.61,
|
| 649 |
+
"learning_rate": 2.1775544388609716e-05,
|
| 650 |
+
"loss": 0.0725,
|
| 651 |
+
"step": 520
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.64,
|
| 655 |
+
"learning_rate": 2.198492462311558e-05,
|
| 656 |
+
"loss": 0.1454,
|
| 657 |
+
"step": 525
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 2.66,
|
| 661 |
+
"learning_rate": 2.219430485762144e-05,
|
| 662 |
+
"loss": 0.1174,
|
| 663 |
+
"step": 530
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.69,
|
| 667 |
+
"learning_rate": 2.2403685092127304e-05,
|
| 668 |
+
"loss": 0.1054,
|
| 669 |
+
"step": 535
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.71,
|
| 673 |
+
"learning_rate": 2.2613065326633167e-05,
|
| 674 |
+
"loss": 0.1123,
|
| 675 |
+
"step": 540
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.74,
|
| 679 |
+
"learning_rate": 2.282244556113903e-05,
|
| 680 |
+
"loss": 0.1119,
|
| 681 |
+
"step": 545
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.76,
|
| 685 |
+
"learning_rate": 2.3031825795644892e-05,
|
| 686 |
+
"loss": 0.0954,
|
| 687 |
+
"step": 550
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 2.79,
|
| 691 |
+
"learning_rate": 2.3241206030150754e-05,
|
| 692 |
+
"loss": 0.1004,
|
| 693 |
+
"step": 555
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.81,
|
| 697 |
+
"learning_rate": 2.3450586264656617e-05,
|
| 698 |
+
"loss": 0.1512,
|
| 699 |
+
"step": 560
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.84,
|
| 703 |
+
"learning_rate": 2.365996649916248e-05,
|
| 704 |
+
"loss": 0.1917,
|
| 705 |
+
"step": 565
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.86,
|
| 709 |
+
"learning_rate": 2.3869346733668342e-05,
|
| 710 |
+
"loss": 0.0982,
|
| 711 |
+
"step": 570
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.89,
|
| 715 |
+
"learning_rate": 2.4078726968174205e-05,
|
| 716 |
+
"loss": 0.0821,
|
| 717 |
+
"step": 575
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.91,
|
| 721 |
+
"learning_rate": 2.4288107202680068e-05,
|
| 722 |
+
"loss": 0.119,
|
| 723 |
+
"step": 580
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.94,
|
| 727 |
+
"learning_rate": 2.449748743718593e-05,
|
| 728 |
+
"loss": 0.142,
|
| 729 |
+
"step": 585
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.96,
|
| 733 |
+
"learning_rate": 2.4706867671691793e-05,
|
| 734 |
+
"loss": 0.0796,
|
| 735 |
+
"step": 590
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.99,
|
| 739 |
+
"learning_rate": 2.4916247906197655e-05,
|
| 740 |
+
"loss": 0.0967,
|
| 741 |
+
"step": 595
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 3.0,
|
| 745 |
+
"eval_accuracy": 0.9825636192271442,
|
| 746 |
+
"eval_loss": 0.050433963537216187,
|
| 747 |
+
"eval_runtime": 91.4591,
|
| 748 |
+
"eval_samples_per_second": 23.202,
|
| 749 |
+
"eval_steps_per_second": 2.908,
|
| 750 |
+
"step": 597
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.02,
|
| 754 |
+
"learning_rate": 2.5125628140703518e-05,
|
| 755 |
+
"loss": 0.1725,
|
| 756 |
+
"step": 600
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.04,
|
| 760 |
+
"learning_rate": 2.533500837520938e-05,
|
| 761 |
+
"loss": 0.0554,
|
| 762 |
+
"step": 605
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.07,
|
| 766 |
+
"learning_rate": 2.5544388609715243e-05,
|
| 767 |
+
"loss": 0.4003,
|
| 768 |
+
"step": 610
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.09,
|
| 772 |
+
"learning_rate": 2.5753768844221106e-05,
|
| 773 |
+
"loss": 0.2184,
|
| 774 |
+
"step": 615
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.12,
|
| 778 |
+
"learning_rate": 2.596314907872697e-05,
|
| 779 |
+
"loss": 0.1262,
|
| 780 |
+
"step": 620
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.14,
|
| 784 |
+
"learning_rate": 2.617252931323283e-05,
|
| 785 |
+
"loss": 0.0931,
|
| 786 |
+
"step": 625
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.17,
|
| 790 |
+
"learning_rate": 2.6381909547738694e-05,
|
| 791 |
+
"loss": 0.0755,
|
| 792 |
+
"step": 630
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.19,
|
| 796 |
+
"learning_rate": 2.6591289782244556e-05,
|
| 797 |
+
"loss": 0.0888,
|
| 798 |
+
"step": 635
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 3.22,
|
| 802 |
+
"learning_rate": 2.680067001675042e-05,
|
| 803 |
+
"loss": 0.1824,
|
| 804 |
+
"step": 640
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.24,
|
| 808 |
+
"learning_rate": 2.7010050251256282e-05,
|
| 809 |
+
"loss": 0.1061,
|
| 810 |
+
"step": 645
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 3.27,
|
| 814 |
+
"learning_rate": 2.7219430485762144e-05,
|
| 815 |
+
"loss": 0.1057,
|
| 816 |
+
"step": 650
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 3.29,
|
| 820 |
+
"learning_rate": 2.7428810720268007e-05,
|
| 821 |
+
"loss": 0.2637,
|
| 822 |
+
"step": 655
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.32,
|
| 826 |
+
"learning_rate": 2.763819095477387e-05,
|
| 827 |
+
"loss": 0.1097,
|
| 828 |
+
"step": 660
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 3.34,
|
| 832 |
+
"learning_rate": 2.7847571189279732e-05,
|
| 833 |
+
"loss": 0.1488,
|
| 834 |
+
"step": 665
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.37,
|
| 838 |
+
"learning_rate": 2.8056951423785595e-05,
|
| 839 |
+
"loss": 0.0566,
|
| 840 |
+
"step": 670
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.39,
|
| 844 |
+
"learning_rate": 2.8266331658291457e-05,
|
| 845 |
+
"loss": 0.1574,
|
| 846 |
+
"step": 675
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.42,
|
| 850 |
+
"learning_rate": 2.847571189279732e-05,
|
| 851 |
+
"loss": 0.1386,
|
| 852 |
+
"step": 680
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 3.44,
|
| 856 |
+
"learning_rate": 2.8685092127303183e-05,
|
| 857 |
+
"loss": 0.1321,
|
| 858 |
+
"step": 685
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.47,
|
| 862 |
+
"learning_rate": 2.8894472361809045e-05,
|
| 863 |
+
"loss": 0.2852,
|
| 864 |
+
"step": 690
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.49,
|
| 868 |
+
"learning_rate": 2.9103852596314908e-05,
|
| 869 |
+
"loss": 0.1582,
|
| 870 |
+
"step": 695
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 3.52,
|
| 874 |
+
"learning_rate": 2.931323283082077e-05,
|
| 875 |
+
"loss": 0.2385,
|
| 876 |
+
"step": 700
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.54,
|
| 880 |
+
"learning_rate": 2.9522613065326633e-05,
|
| 881 |
+
"loss": 0.2513,
|
| 882 |
+
"step": 705
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 3.57,
|
| 886 |
+
"learning_rate": 2.9731993299832496e-05,
|
| 887 |
+
"loss": 0.0796,
|
| 888 |
+
"step": 710
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.59,
|
| 892 |
+
"learning_rate": 2.994137353433836e-05,
|
| 893 |
+
"loss": 0.0922,
|
| 894 |
+
"step": 715
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.62,
|
| 898 |
+
"learning_rate": 3.015075376884422e-05,
|
| 899 |
+
"loss": 0.1955,
|
| 900 |
+
"step": 720
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 3.64,
|
| 904 |
+
"learning_rate": 3.0360134003350084e-05,
|
| 905 |
+
"loss": 0.1581,
|
| 906 |
+
"step": 725
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 3.67,
|
| 910 |
+
"learning_rate": 3.056951423785594e-05,
|
| 911 |
+
"loss": 0.0307,
|
| 912 |
+
"step": 730
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 3.69,
|
| 916 |
+
"learning_rate": 3.077889447236181e-05,
|
| 917 |
+
"loss": 0.1641,
|
| 918 |
+
"step": 735
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 3.72,
|
| 922 |
+
"learning_rate": 3.098827470686767e-05,
|
| 923 |
+
"loss": 0.3599,
|
| 924 |
+
"step": 740
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 3.74,
|
| 928 |
+
"learning_rate": 3.1197654941373534e-05,
|
| 929 |
+
"loss": 0.112,
|
| 930 |
+
"step": 745
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 3.77,
|
| 934 |
+
"learning_rate": 3.14070351758794e-05,
|
| 935 |
+
"loss": 0.1384,
|
| 936 |
+
"step": 750
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 3.79,
|
| 940 |
+
"learning_rate": 3.1616415410385266e-05,
|
| 941 |
+
"loss": 0.2174,
|
| 942 |
+
"step": 755
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 3.82,
|
| 946 |
+
"learning_rate": 3.1825795644891125e-05,
|
| 947 |
+
"loss": 0.1579,
|
| 948 |
+
"step": 760
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 3.84,
|
| 952 |
+
"learning_rate": 3.203517587939699e-05,
|
| 953 |
+
"loss": 0.1173,
|
| 954 |
+
"step": 765
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 3.87,
|
| 958 |
+
"learning_rate": 3.224455611390285e-05,
|
| 959 |
+
"loss": 0.0456,
|
| 960 |
+
"step": 770
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 3.89,
|
| 964 |
+
"learning_rate": 3.245393634840872e-05,
|
| 965 |
+
"loss": 0.1281,
|
| 966 |
+
"step": 775
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 3.92,
|
| 970 |
+
"learning_rate": 3.2663316582914576e-05,
|
| 971 |
+
"loss": 0.1152,
|
| 972 |
+
"step": 780
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 3.94,
|
| 976 |
+
"learning_rate": 3.287269681742044e-05,
|
| 977 |
+
"loss": 0.1576,
|
| 978 |
+
"step": 785
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.97,
|
| 982 |
+
"learning_rate": 3.30820770519263e-05,
|
| 983 |
+
"loss": 0.1334,
|
| 984 |
+
"step": 790
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 3.99,
|
| 988 |
+
"learning_rate": 3.329145728643217e-05,
|
| 989 |
+
"loss": 0.0809,
|
| 990 |
+
"step": 795
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 4.0,
|
| 994 |
+
"eval_accuracy": 0.9448633364750235,
|
| 995 |
+
"eval_loss": 0.1603740155696869,
|
| 996 |
+
"eval_runtime": 91.2767,
|
| 997 |
+
"eval_samples_per_second": 23.248,
|
| 998 |
+
"eval_steps_per_second": 2.914,
|
| 999 |
+
"step": 796
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.02,
|
| 1003 |
+
"learning_rate": 3.3500837520938026e-05,
|
| 1004 |
+
"loss": 0.1905,
|
| 1005 |
+
"step": 800
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 4.05,
|
| 1009 |
+
"learning_rate": 3.371021775544389e-05,
|
| 1010 |
+
"loss": 0.1124,
|
| 1011 |
+
"step": 805
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 4.07,
|
| 1015 |
+
"learning_rate": 3.391959798994975e-05,
|
| 1016 |
+
"loss": 0.1095,
|
| 1017 |
+
"step": 810
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 4.1,
|
| 1021 |
+
"learning_rate": 3.412897822445562e-05,
|
| 1022 |
+
"loss": 0.1477,
|
| 1023 |
+
"step": 815
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.12,
|
| 1027 |
+
"learning_rate": 3.433835845896148e-05,
|
| 1028 |
+
"loss": 0.0943,
|
| 1029 |
+
"step": 820
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.15,
|
| 1033 |
+
"learning_rate": 3.454773869346734e-05,
|
| 1034 |
+
"loss": 0.1347,
|
| 1035 |
+
"step": 825
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 4.17,
|
| 1039 |
+
"learning_rate": 3.47571189279732e-05,
|
| 1040 |
+
"loss": 0.0956,
|
| 1041 |
+
"step": 830
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 4.2,
|
| 1045 |
+
"learning_rate": 3.496649916247907e-05,
|
| 1046 |
+
"loss": 0.2531,
|
| 1047 |
+
"step": 835
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 4.22,
|
| 1051 |
+
"learning_rate": 3.517587939698493e-05,
|
| 1052 |
+
"loss": 0.2285,
|
| 1053 |
+
"step": 840
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.25,
|
| 1057 |
+
"learning_rate": 3.538525963149079e-05,
|
| 1058 |
+
"loss": 0.1616,
|
| 1059 |
+
"step": 845
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.27,
|
| 1063 |
+
"learning_rate": 3.559463986599665e-05,
|
| 1064 |
+
"loss": 0.0924,
|
| 1065 |
+
"step": 850
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 4.3,
|
| 1069 |
+
"learning_rate": 3.580402010050252e-05,
|
| 1070 |
+
"loss": 0.167,
|
| 1071 |
+
"step": 855
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 4.32,
|
| 1075 |
+
"learning_rate": 3.601340033500838e-05,
|
| 1076 |
+
"loss": 0.0795,
|
| 1077 |
+
"step": 860
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 4.35,
|
| 1081 |
+
"learning_rate": 3.6222780569514244e-05,
|
| 1082 |
+
"loss": 0.059,
|
| 1083 |
+
"step": 865
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 4.37,
|
| 1087 |
+
"learning_rate": 3.64321608040201e-05,
|
| 1088 |
+
"loss": 0.069,
|
| 1089 |
+
"step": 870
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.4,
|
| 1093 |
+
"learning_rate": 3.664154103852597e-05,
|
| 1094 |
+
"loss": 0.083,
|
| 1095 |
+
"step": 875
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.42,
|
| 1099 |
+
"learning_rate": 3.685092127303183e-05,
|
| 1100 |
+
"loss": 0.1194,
|
| 1101 |
+
"step": 880
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 4.45,
|
| 1105 |
+
"learning_rate": 3.7060301507537694e-05,
|
| 1106 |
+
"loss": 0.1297,
|
| 1107 |
+
"step": 885
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 4.47,
|
| 1111 |
+
"learning_rate": 3.7269681742043554e-05,
|
| 1112 |
+
"loss": 0.2179,
|
| 1113 |
+
"step": 890
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 4.5,
|
| 1117 |
+
"learning_rate": 3.747906197654942e-05,
|
| 1118 |
+
"loss": 0.2596,
|
| 1119 |
+
"step": 895
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.52,
|
| 1123 |
+
"learning_rate": 3.768844221105528e-05,
|
| 1124 |
+
"loss": 0.1906,
|
| 1125 |
+
"step": 900
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 4.55,
|
| 1129 |
+
"learning_rate": 3.7897822445561145e-05,
|
| 1130 |
+
"loss": 0.0549,
|
| 1131 |
+
"step": 905
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.57,
|
| 1135 |
+
"learning_rate": 3.8107202680067004e-05,
|
| 1136 |
+
"loss": 0.1066,
|
| 1137 |
+
"step": 910
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 4.6,
|
| 1141 |
+
"learning_rate": 3.831658291457287e-05,
|
| 1142 |
+
"loss": 0.1355,
|
| 1143 |
+
"step": 915
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 4.62,
|
| 1147 |
+
"learning_rate": 3.852596314907873e-05,
|
| 1148 |
+
"loss": 0.1011,
|
| 1149 |
+
"step": 920
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 4.65,
|
| 1153 |
+
"learning_rate": 3.8735343383584595e-05,
|
| 1154 |
+
"loss": 0.1629,
|
| 1155 |
+
"step": 925
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 4.67,
|
| 1159 |
+
"learning_rate": 3.8944723618090455e-05,
|
| 1160 |
+
"loss": 0.1952,
|
| 1161 |
+
"step": 930
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 4.7,
|
| 1165 |
+
"learning_rate": 3.915410385259632e-05,
|
| 1166 |
+
"loss": 0.1166,
|
| 1167 |
+
"step": 935
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 4.72,
|
| 1171 |
+
"learning_rate": 3.936348408710218e-05,
|
| 1172 |
+
"loss": 0.1182,
|
| 1173 |
+
"step": 940
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 4.75,
|
| 1177 |
+
"learning_rate": 3.9572864321608046e-05,
|
| 1178 |
+
"loss": 0.1669,
|
| 1179 |
+
"step": 945
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 4.77,
|
| 1183 |
+
"learning_rate": 3.9782244556113905e-05,
|
| 1184 |
+
"loss": 0.1174,
|
| 1185 |
+
"step": 950
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 4.8,
|
| 1189 |
+
"learning_rate": 3.999162479061977e-05,
|
| 1190 |
+
"loss": 0.0891,
|
| 1191 |
+
"step": 955
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 4.82,
|
| 1195 |
+
"learning_rate": 4.020100502512563e-05,
|
| 1196 |
+
"loss": 0.2477,
|
| 1197 |
+
"step": 960
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 4.85,
|
| 1201 |
+
"learning_rate": 4.0410385259631496e-05,
|
| 1202 |
+
"loss": 0.2259,
|
| 1203 |
+
"step": 965
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 4.87,
|
| 1207 |
+
"learning_rate": 4.0619765494137356e-05,
|
| 1208 |
+
"loss": 0.0424,
|
| 1209 |
+
"step": 970
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 4.9,
|
| 1213 |
+
"learning_rate": 4.082914572864322e-05,
|
| 1214 |
+
"loss": 0.0697,
|
| 1215 |
+
"step": 975
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 4.92,
|
| 1219 |
+
"learning_rate": 4.103852596314908e-05,
|
| 1220 |
+
"loss": 0.1165,
|
| 1221 |
+
"step": 980
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 4.95,
|
| 1225 |
+
"learning_rate": 4.124790619765495e-05,
|
| 1226 |
+
"loss": 0.1085,
|
| 1227 |
+
"step": 985
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 4.97,
|
| 1231 |
+
"learning_rate": 4.1457286432160806e-05,
|
| 1232 |
+
"loss": 0.0367,
|
| 1233 |
+
"step": 990
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 5.0,
|
| 1237 |
+
"learning_rate": 4.166666666666667e-05,
|
| 1238 |
+
"loss": 0.1531,
|
| 1239 |
+
"step": 995
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.0,
|
| 1243 |
+
"eval_accuracy": 0.9806786050895382,
|
| 1244 |
+
"eval_loss": 0.06729210168123245,
|
| 1245 |
+
"eval_runtime": 91.2343,
|
| 1246 |
+
"eval_samples_per_second": 23.259,
|
| 1247 |
+
"eval_steps_per_second": 2.916,
|
| 1248 |
+
"step": 995
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 5.03,
|
| 1252 |
+
"learning_rate": 4.187604690117253e-05,
|
| 1253 |
+
"loss": 0.0984,
|
| 1254 |
+
"step": 1000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 5.05,
|
| 1258 |
+
"learning_rate": 4.20854271356784e-05,
|
| 1259 |
+
"loss": 0.1212,
|
| 1260 |
+
"step": 1005
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 5.08,
|
| 1264 |
+
"learning_rate": 4.2294807370184257e-05,
|
| 1265 |
+
"loss": 0.1852,
|
| 1266 |
+
"step": 1010
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 5.1,
|
| 1270 |
+
"learning_rate": 4.250418760469012e-05,
|
| 1271 |
+
"loss": 0.1014,
|
| 1272 |
+
"step": 1015
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 5.13,
|
| 1276 |
+
"learning_rate": 4.271356783919598e-05,
|
| 1277 |
+
"loss": 0.1069,
|
| 1278 |
+
"step": 1020
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 5.15,
|
| 1282 |
+
"learning_rate": 4.292294807370185e-05,
|
| 1283 |
+
"loss": 0.05,
|
| 1284 |
+
"step": 1025
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.18,
|
| 1288 |
+
"learning_rate": 4.313232830820771e-05,
|
| 1289 |
+
"loss": 0.1147,
|
| 1290 |
+
"step": 1030
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 5.2,
|
| 1294 |
+
"learning_rate": 4.334170854271357e-05,
|
| 1295 |
+
"loss": 0.1277,
|
| 1296 |
+
"step": 1035
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 5.23,
|
| 1300 |
+
"learning_rate": 4.355108877721943e-05,
|
| 1301 |
+
"loss": 0.1375,
|
| 1302 |
+
"step": 1040
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 5.25,
|
| 1306 |
+
"learning_rate": 4.37604690117253e-05,
|
| 1307 |
+
"loss": 0.0598,
|
| 1308 |
+
"step": 1045
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 5.28,
|
| 1312 |
+
"learning_rate": 4.396984924623116e-05,
|
| 1313 |
+
"loss": 0.0845,
|
| 1314 |
+
"step": 1050
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.3,
|
| 1318 |
+
"learning_rate": 4.4179229480737024e-05,
|
| 1319 |
+
"loss": 0.0603,
|
| 1320 |
+
"step": 1055
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 5.33,
|
| 1324 |
+
"learning_rate": 4.438860971524288e-05,
|
| 1325 |
+
"loss": 0.1557,
|
| 1326 |
+
"step": 1060
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 5.35,
|
| 1330 |
+
"learning_rate": 4.459798994974875e-05,
|
| 1331 |
+
"loss": 0.0632,
|
| 1332 |
+
"step": 1065
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 5.38,
|
| 1336 |
+
"learning_rate": 4.480737018425461e-05,
|
| 1337 |
+
"loss": 0.1375,
|
| 1338 |
+
"step": 1070
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 5.4,
|
| 1342 |
+
"learning_rate": 4.5016750418760474e-05,
|
| 1343 |
+
"loss": 0.1563,
|
| 1344 |
+
"step": 1075
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.43,
|
| 1348 |
+
"learning_rate": 4.522613065326633e-05,
|
| 1349 |
+
"loss": 0.0756,
|
| 1350 |
+
"step": 1080
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 5.45,
|
| 1354 |
+
"learning_rate": 4.54355108877722e-05,
|
| 1355 |
+
"loss": 0.1598,
|
| 1356 |
+
"step": 1085
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 5.48,
|
| 1360 |
+
"learning_rate": 4.564489112227806e-05,
|
| 1361 |
+
"loss": 0.1176,
|
| 1362 |
+
"step": 1090
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 5.5,
|
| 1366 |
+
"learning_rate": 4.5854271356783925e-05,
|
| 1367 |
+
"loss": 0.0973,
|
| 1368 |
+
"step": 1095
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 5.53,
|
| 1372 |
+
"learning_rate": 4.6063651591289784e-05,
|
| 1373 |
+
"loss": 0.1257,
|
| 1374 |
+
"step": 1100
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 5.55,
|
| 1378 |
+
"learning_rate": 4.627303182579565e-05,
|
| 1379 |
+
"loss": 0.0995,
|
| 1380 |
+
"step": 1105
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 5.58,
|
| 1384 |
+
"learning_rate": 4.648241206030151e-05,
|
| 1385 |
+
"loss": 0.0587,
|
| 1386 |
+
"step": 1110
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.6,
|
| 1390 |
+
"learning_rate": 4.6691792294807375e-05,
|
| 1391 |
+
"loss": 0.053,
|
| 1392 |
+
"step": 1115
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 5.63,
|
| 1396 |
+
"learning_rate": 4.6901172529313234e-05,
|
| 1397 |
+
"loss": 0.0598,
|
| 1398 |
+
"step": 1120
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 5.65,
|
| 1402 |
+
"learning_rate": 4.71105527638191e-05,
|
| 1403 |
+
"loss": 0.0378,
|
| 1404 |
+
"step": 1125
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 5.68,
|
| 1408 |
+
"learning_rate": 4.731993299832496e-05,
|
| 1409 |
+
"loss": 0.1141,
|
| 1410 |
+
"step": 1130
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 5.7,
|
| 1414 |
+
"learning_rate": 4.7529313232830825e-05,
|
| 1415 |
+
"loss": 0.1103,
|
| 1416 |
+
"step": 1135
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 5.73,
|
| 1420 |
+
"learning_rate": 4.7738693467336685e-05,
|
| 1421 |
+
"loss": 0.0986,
|
| 1422 |
+
"step": 1140
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 5.75,
|
| 1426 |
+
"learning_rate": 4.794807370184255e-05,
|
| 1427 |
+
"loss": 0.0762,
|
| 1428 |
+
"step": 1145
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.78,
|
| 1432 |
+
"learning_rate": 4.815745393634841e-05,
|
| 1433 |
+
"loss": 0.0736,
|
| 1434 |
+
"step": 1150
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 5.8,
|
| 1438 |
+
"learning_rate": 4.8366834170854276e-05,
|
| 1439 |
+
"loss": 0.1732,
|
| 1440 |
+
"step": 1155
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 5.83,
|
| 1444 |
+
"learning_rate": 4.8576214405360135e-05,
|
| 1445 |
+
"loss": 0.1787,
|
| 1446 |
+
"step": 1160
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 5.85,
|
| 1450 |
+
"learning_rate": 4.8785594639866e-05,
|
| 1451 |
+
"loss": 0.2364,
|
| 1452 |
+
"step": 1165
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 5.88,
|
| 1456 |
+
"learning_rate": 4.899497487437186e-05,
|
| 1457 |
+
"loss": 0.0701,
|
| 1458 |
+
"step": 1170
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 5.9,
|
| 1462 |
+
"learning_rate": 4.9204355108877726e-05,
|
| 1463 |
+
"loss": 0.065,
|
| 1464 |
+
"step": 1175
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 5.93,
|
| 1468 |
+
"learning_rate": 4.9413735343383586e-05,
|
| 1469 |
+
"loss": 0.2027,
|
| 1470 |
+
"step": 1180
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 5.95,
|
| 1474 |
+
"learning_rate": 4.962311557788945e-05,
|
| 1475 |
+
"loss": 0.1818,
|
| 1476 |
+
"step": 1185
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 5.98,
|
| 1480 |
+
"learning_rate": 4.983249581239531e-05,
|
| 1481 |
+
"loss": 0.0941,
|
| 1482 |
+
"step": 1190
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 6.0,
|
| 1486 |
+
"eval_accuracy": 0.9679547596606974,
|
| 1487 |
+
"eval_loss": 0.08656678348779678,
|
| 1488 |
+
"eval_runtime": 91.4801,
|
| 1489 |
+
"eval_samples_per_second": 23.196,
|
| 1490 |
+
"eval_steps_per_second": 2.908,
|
| 1491 |
+
"step": 1194
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 6.01,
|
| 1495 |
+
"learning_rate": 4.995812395309883e-05,
|
| 1496 |
+
"loss": 0.0423,
|
| 1497 |
+
"step": 1195
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 6.03,
|
| 1501 |
+
"learning_rate": 4.974874371859297e-05,
|
| 1502 |
+
"loss": 0.1607,
|
| 1503 |
+
"step": 1200
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 6.06,
|
| 1507 |
+
"learning_rate": 4.95393634840871e-05,
|
| 1508 |
+
"loss": 0.2889,
|
| 1509 |
+
"step": 1205
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 6.08,
|
| 1513 |
+
"learning_rate": 4.932998324958124e-05,
|
| 1514 |
+
"loss": 0.1447,
|
| 1515 |
+
"step": 1210
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 6.11,
|
| 1519 |
+
"learning_rate": 4.912060301507538e-05,
|
| 1520 |
+
"loss": 0.0918,
|
| 1521 |
+
"step": 1215
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 6.13,
|
| 1525 |
+
"learning_rate": 4.891122278056952e-05,
|
| 1526 |
+
"loss": 0.1959,
|
| 1527 |
+
"step": 1220
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 6.16,
|
| 1531 |
+
"learning_rate": 4.870184254606365e-05,
|
| 1532 |
+
"loss": 0.0893,
|
| 1533 |
+
"step": 1225
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 6.18,
|
| 1537 |
+
"learning_rate": 4.849246231155779e-05,
|
| 1538 |
+
"loss": 0.0551,
|
| 1539 |
+
"step": 1230
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.21,
|
| 1543 |
+
"learning_rate": 4.828308207705193e-05,
|
| 1544 |
+
"loss": 0.0628,
|
| 1545 |
+
"step": 1235
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 6.23,
|
| 1549 |
+
"learning_rate": 4.807370184254607e-05,
|
| 1550 |
+
"loss": 0.0512,
|
| 1551 |
+
"step": 1240
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 6.26,
|
| 1555 |
+
"learning_rate": 4.78643216080402e-05,
|
| 1556 |
+
"loss": 0.2542,
|
| 1557 |
+
"step": 1245
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 6.28,
|
| 1561 |
+
"learning_rate": 4.765494137353434e-05,
|
| 1562 |
+
"loss": 0.0462,
|
| 1563 |
+
"step": 1250
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 6.31,
|
| 1567 |
+
"learning_rate": 4.7445561139028476e-05,
|
| 1568 |
+
"loss": 0.1519,
|
| 1569 |
+
"step": 1255
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.33,
|
| 1573 |
+
"learning_rate": 4.723618090452262e-05,
|
| 1574 |
+
"loss": 0.0917,
|
| 1575 |
+
"step": 1260
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 6.36,
|
| 1579 |
+
"learning_rate": 4.702680067001675e-05,
|
| 1580 |
+
"loss": 0.2339,
|
| 1581 |
+
"step": 1265
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 6.38,
|
| 1585 |
+
"learning_rate": 4.681742043551089e-05,
|
| 1586 |
+
"loss": 0.1873,
|
| 1587 |
+
"step": 1270
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 6.41,
|
| 1591 |
+
"learning_rate": 4.6608040201005026e-05,
|
| 1592 |
+
"loss": 0.0984,
|
| 1593 |
+
"step": 1275
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 6.43,
|
| 1597 |
+
"learning_rate": 4.639865996649917e-05,
|
| 1598 |
+
"loss": 0.1295,
|
| 1599 |
+
"step": 1280
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.46,
|
| 1603 |
+
"learning_rate": 4.61892797319933e-05,
|
| 1604 |
+
"loss": 0.0943,
|
| 1605 |
+
"step": 1285
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 6.48,
|
| 1609 |
+
"learning_rate": 4.597989949748744e-05,
|
| 1610 |
+
"loss": 0.1148,
|
| 1611 |
+
"step": 1290
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 6.51,
|
| 1615 |
+
"learning_rate": 4.5770519262981575e-05,
|
| 1616 |
+
"loss": 0.0837,
|
| 1617 |
+
"step": 1295
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 6.53,
|
| 1621 |
+
"learning_rate": 4.5561139028475716e-05,
|
| 1622 |
+
"loss": 0.1483,
|
| 1623 |
+
"step": 1300
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 6.56,
|
| 1627 |
+
"learning_rate": 4.535175879396985e-05,
|
| 1628 |
+
"loss": 0.0798,
|
| 1629 |
+
"step": 1305
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 6.58,
|
| 1633 |
+
"learning_rate": 4.514237855946399e-05,
|
| 1634 |
+
"loss": 0.0404,
|
| 1635 |
+
"step": 1310
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 6.61,
|
| 1639 |
+
"learning_rate": 4.4932998324958125e-05,
|
| 1640 |
+
"loss": 0.0347,
|
| 1641 |
+
"step": 1315
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.63,
|
| 1645 |
+
"learning_rate": 4.4723618090452266e-05,
|
| 1646 |
+
"loss": 0.0504,
|
| 1647 |
+
"step": 1320
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 6.66,
|
| 1651 |
+
"learning_rate": 4.45142378559464e-05,
|
| 1652 |
+
"loss": 0.1347,
|
| 1653 |
+
"step": 1325
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 6.68,
|
| 1657 |
+
"learning_rate": 4.430485762144054e-05,
|
| 1658 |
+
"loss": 0.0369,
|
| 1659 |
+
"step": 1330
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 6.71,
|
| 1663 |
+
"learning_rate": 4.4095477386934674e-05,
|
| 1664 |
+
"loss": 0.1117,
|
| 1665 |
+
"step": 1335
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 6.73,
|
| 1669 |
+
"learning_rate": 4.3886097152428815e-05,
|
| 1670 |
+
"loss": 0.175,
|
| 1671 |
+
"step": 1340
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 6.76,
|
| 1675 |
+
"learning_rate": 4.367671691792295e-05,
|
| 1676 |
+
"loss": 0.0575,
|
| 1677 |
+
"step": 1345
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 6.78,
|
| 1681 |
+
"learning_rate": 4.346733668341709e-05,
|
| 1682 |
+
"loss": 0.0205,
|
| 1683 |
+
"step": 1350
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 6.81,
|
| 1687 |
+
"learning_rate": 4.3257956448911224e-05,
|
| 1688 |
+
"loss": 0.0523,
|
| 1689 |
+
"step": 1355
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 6.83,
|
| 1693 |
+
"learning_rate": 4.3048576214405365e-05,
|
| 1694 |
+
"loss": 0.0719,
|
| 1695 |
+
"step": 1360
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 6.86,
|
| 1699 |
+
"learning_rate": 4.28391959798995e-05,
|
| 1700 |
+
"loss": 0.0567,
|
| 1701 |
+
"step": 1365
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 6.88,
|
| 1705 |
+
"learning_rate": 4.262981574539364e-05,
|
| 1706 |
+
"loss": 0.0985,
|
| 1707 |
+
"step": 1370
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 6.91,
|
| 1711 |
+
"learning_rate": 4.2420435510887773e-05,
|
| 1712 |
+
"loss": 0.2237,
|
| 1713 |
+
"step": 1375
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 6.93,
|
| 1717 |
+
"learning_rate": 4.2211055276381914e-05,
|
| 1718 |
+
"loss": 0.0886,
|
| 1719 |
+
"step": 1380
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 6.96,
|
| 1723 |
+
"learning_rate": 4.200167504187605e-05,
|
| 1724 |
+
"loss": 0.0679,
|
| 1725 |
+
"step": 1385
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.98,
|
| 1729 |
+
"learning_rate": 4.179229480737019e-05,
|
| 1730 |
+
"loss": 0.1157,
|
| 1731 |
+
"step": 1390
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 7.0,
|
| 1735 |
+
"eval_accuracy": 0.9844486333647502,
|
| 1736 |
+
"eval_loss": 0.05253003165125847,
|
| 1737 |
+
"eval_runtime": 91.534,
|
| 1738 |
+
"eval_samples_per_second": 23.183,
|
| 1739 |
+
"eval_steps_per_second": 2.906,
|
| 1740 |
+
"step": 1393
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 7.01,
|
| 1744 |
+
"learning_rate": 4.158291457286432e-05,
|
| 1745 |
+
"loss": 0.141,
|
| 1746 |
+
"step": 1395
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 7.04,
|
| 1750 |
+
"learning_rate": 4.1373534338358464e-05,
|
| 1751 |
+
"loss": 0.0745,
|
| 1752 |
+
"step": 1400
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 7.06,
|
| 1756 |
+
"learning_rate": 4.11641541038526e-05,
|
| 1757 |
+
"loss": 0.0342,
|
| 1758 |
+
"step": 1405
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 7.09,
|
| 1762 |
+
"learning_rate": 4.095477386934674e-05,
|
| 1763 |
+
"loss": 0.056,
|
| 1764 |
+
"step": 1410
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 7.11,
|
| 1768 |
+
"learning_rate": 4.074539363484087e-05,
|
| 1769 |
+
"loss": 0.16,
|
| 1770 |
+
"step": 1415
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 7.14,
|
| 1774 |
+
"learning_rate": 4.053601340033501e-05,
|
| 1775 |
+
"loss": 0.109,
|
| 1776 |
+
"step": 1420
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 7.16,
|
| 1780 |
+
"learning_rate": 4.032663316582915e-05,
|
| 1781 |
+
"loss": 0.152,
|
| 1782 |
+
"step": 1425
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 7.19,
|
| 1786 |
+
"learning_rate": 4.011725293132329e-05,
|
| 1787 |
+
"loss": 0.0268,
|
| 1788 |
+
"step": 1430
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 7.21,
|
| 1792 |
+
"learning_rate": 3.990787269681742e-05,
|
| 1793 |
+
"loss": 0.0469,
|
| 1794 |
+
"step": 1435
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 7.24,
|
| 1798 |
+
"learning_rate": 3.969849246231156e-05,
|
| 1799 |
+
"loss": 0.0486,
|
| 1800 |
+
"step": 1440
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 7.26,
|
| 1804 |
+
"learning_rate": 3.94891122278057e-05,
|
| 1805 |
+
"loss": 0.0799,
|
| 1806 |
+
"step": 1445
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 7.29,
|
| 1810 |
+
"learning_rate": 3.927973199329984e-05,
|
| 1811 |
+
"loss": 0.039,
|
| 1812 |
+
"step": 1450
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 7.31,
|
| 1816 |
+
"learning_rate": 3.907035175879397e-05,
|
| 1817 |
+
"loss": 0.1076,
|
| 1818 |
+
"step": 1455
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 7.34,
|
| 1822 |
+
"learning_rate": 3.886097152428811e-05,
|
| 1823 |
+
"loss": 0.0738,
|
| 1824 |
+
"step": 1460
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 7.36,
|
| 1828 |
+
"learning_rate": 3.8651591289782246e-05,
|
| 1829 |
+
"loss": 0.1493,
|
| 1830 |
+
"step": 1465
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 7.39,
|
| 1834 |
+
"learning_rate": 3.844221105527639e-05,
|
| 1835 |
+
"loss": 0.1027,
|
| 1836 |
+
"step": 1470
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 7.41,
|
| 1840 |
+
"learning_rate": 3.823283082077052e-05,
|
| 1841 |
+
"loss": 0.0343,
|
| 1842 |
+
"step": 1475
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 7.44,
|
| 1846 |
+
"learning_rate": 3.802345058626466e-05,
|
| 1847 |
+
"loss": 0.1157,
|
| 1848 |
+
"step": 1480
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 7.46,
|
| 1852 |
+
"learning_rate": 3.7814070351758796e-05,
|
| 1853 |
+
"loss": 0.1568,
|
| 1854 |
+
"step": 1485
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 7.49,
|
| 1858 |
+
"learning_rate": 3.7604690117252936e-05,
|
| 1859 |
+
"loss": 0.0784,
|
| 1860 |
+
"step": 1490
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 7.51,
|
| 1864 |
+
"learning_rate": 3.739530988274707e-05,
|
| 1865 |
+
"loss": 0.0659,
|
| 1866 |
+
"step": 1495
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 7.54,
|
| 1870 |
+
"learning_rate": 3.7185929648241204e-05,
|
| 1871 |
+
"loss": 0.1141,
|
| 1872 |
+
"step": 1500
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 7.56,
|
| 1876 |
+
"learning_rate": 3.6976549413735345e-05,
|
| 1877 |
+
"loss": 0.1225,
|
| 1878 |
+
"step": 1505
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 7.59,
|
| 1882 |
+
"learning_rate": 3.676716917922948e-05,
|
| 1883 |
+
"loss": 0.112,
|
| 1884 |
+
"step": 1510
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 7.61,
|
| 1888 |
+
"learning_rate": 3.655778894472362e-05,
|
| 1889 |
+
"loss": 0.1435,
|
| 1890 |
+
"step": 1515
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 7.64,
|
| 1894 |
+
"learning_rate": 3.6348408710217754e-05,
|
| 1895 |
+
"loss": 0.0802,
|
| 1896 |
+
"step": 1520
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.66,
|
| 1900 |
+
"learning_rate": 3.6139028475711895e-05,
|
| 1901 |
+
"loss": 0.0237,
|
| 1902 |
+
"step": 1525
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 7.69,
|
| 1906 |
+
"learning_rate": 3.592964824120603e-05,
|
| 1907 |
+
"loss": 0.1263,
|
| 1908 |
+
"step": 1530
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 7.71,
|
| 1912 |
+
"learning_rate": 3.572026800670017e-05,
|
| 1913 |
+
"loss": 0.0624,
|
| 1914 |
+
"step": 1535
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 7.74,
|
| 1918 |
+
"learning_rate": 3.5510887772194303e-05,
|
| 1919 |
+
"loss": 0.0963,
|
| 1920 |
+
"step": 1540
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 7.76,
|
| 1924 |
+
"learning_rate": 3.5301507537688444e-05,
|
| 1925 |
+
"loss": 0.0227,
|
| 1926 |
+
"step": 1545
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 7.79,
|
| 1930 |
+
"learning_rate": 3.509212730318258e-05,
|
| 1931 |
+
"loss": 0.0827,
|
| 1932 |
+
"step": 1550
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 7.81,
|
| 1936 |
+
"learning_rate": 3.488274706867672e-05,
|
| 1937 |
+
"loss": 0.1123,
|
| 1938 |
+
"step": 1555
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 7.84,
|
| 1942 |
+
"learning_rate": 3.467336683417085e-05,
|
| 1943 |
+
"loss": 0.0659,
|
| 1944 |
+
"step": 1560
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 7.86,
|
| 1948 |
+
"learning_rate": 3.4463986599664994e-05,
|
| 1949 |
+
"loss": 0.0376,
|
| 1950 |
+
"step": 1565
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 7.89,
|
| 1954 |
+
"learning_rate": 3.425460636515913e-05,
|
| 1955 |
+
"loss": 0.0089,
|
| 1956 |
+
"step": 1570
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 7.91,
|
| 1960 |
+
"learning_rate": 3.404522613065327e-05,
|
| 1961 |
+
"loss": 0.0912,
|
| 1962 |
+
"step": 1575
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 7.94,
|
| 1966 |
+
"learning_rate": 3.38358458961474e-05,
|
| 1967 |
+
"loss": 0.073,
|
| 1968 |
+
"step": 1580
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 7.96,
|
| 1972 |
+
"learning_rate": 3.362646566164154e-05,
|
| 1973 |
+
"loss": 0.0478,
|
| 1974 |
+
"step": 1585
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 7.99,
|
| 1978 |
+
"learning_rate": 3.341708542713568e-05,
|
| 1979 |
+
"loss": 0.0684,
|
| 1980 |
+
"step": 1590
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 8.0,
|
| 1984 |
+
"eval_accuracy": 0.9759660697455231,
|
| 1985 |
+
"eval_loss": 0.1003836989402771,
|
| 1986 |
+
"eval_runtime": 91.5881,
|
| 1987 |
+
"eval_samples_per_second": 23.169,
|
| 1988 |
+
"eval_steps_per_second": 2.904,
|
| 1989 |
+
"step": 1592
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 8.02,
|
| 1993 |
+
"learning_rate": 3.320770519262982e-05,
|
| 1994 |
+
"loss": 0.0423,
|
| 1995 |
+
"step": 1595
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 8.04,
|
| 1999 |
+
"learning_rate": 3.299832495812395e-05,
|
| 2000 |
+
"loss": 0.0516,
|
| 2001 |
+
"step": 1600
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 8.07,
|
| 2005 |
+
"learning_rate": 3.278894472361809e-05,
|
| 2006 |
+
"loss": 0.0599,
|
| 2007 |
+
"step": 1605
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 8.09,
|
| 2011 |
+
"learning_rate": 3.257956448911223e-05,
|
| 2012 |
+
"loss": 0.1022,
|
| 2013 |
+
"step": 1610
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 8.12,
|
| 2017 |
+
"learning_rate": 3.237018425460637e-05,
|
| 2018 |
+
"loss": 0.0228,
|
| 2019 |
+
"step": 1615
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 8.14,
|
| 2023 |
+
"learning_rate": 3.21608040201005e-05,
|
| 2024 |
+
"loss": 0.0516,
|
| 2025 |
+
"step": 1620
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 8.17,
|
| 2029 |
+
"learning_rate": 3.195142378559464e-05,
|
| 2030 |
+
"loss": 0.0463,
|
| 2031 |
+
"step": 1625
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 8.19,
|
| 2035 |
+
"learning_rate": 3.1742043551088776e-05,
|
| 2036 |
+
"loss": 0.0959,
|
| 2037 |
+
"step": 1630
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 8.22,
|
| 2041 |
+
"learning_rate": 3.153266331658292e-05,
|
| 2042 |
+
"loss": 0.0109,
|
| 2043 |
+
"step": 1635
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 8.24,
|
| 2047 |
+
"learning_rate": 3.132328308207705e-05,
|
| 2048 |
+
"loss": 0.059,
|
| 2049 |
+
"step": 1640
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 8.27,
|
| 2053 |
+
"learning_rate": 3.111390284757119e-05,
|
| 2054 |
+
"loss": 0.0971,
|
| 2055 |
+
"step": 1645
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 8.29,
|
| 2059 |
+
"learning_rate": 3.0904522613065326e-05,
|
| 2060 |
+
"loss": 0.0721,
|
| 2061 |
+
"step": 1650
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 8.32,
|
| 2065 |
+
"learning_rate": 3.0695142378559467e-05,
|
| 2066 |
+
"loss": 0.1205,
|
| 2067 |
+
"step": 1655
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 8.34,
|
| 2071 |
+
"learning_rate": 3.0485762144053604e-05,
|
| 2072 |
+
"loss": 0.0985,
|
| 2073 |
+
"step": 1660
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 8.37,
|
| 2077 |
+
"learning_rate": 3.027638190954774e-05,
|
| 2078 |
+
"loss": 0.0457,
|
| 2079 |
+
"step": 1665
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 8.39,
|
| 2083 |
+
"learning_rate": 3.006700167504188e-05,
|
| 2084 |
+
"loss": 0.1018,
|
| 2085 |
+
"step": 1670
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 8.42,
|
| 2089 |
+
"learning_rate": 2.9857621440536016e-05,
|
| 2090 |
+
"loss": 0.1062,
|
| 2091 |
+
"step": 1675
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 8.44,
|
| 2095 |
+
"learning_rate": 2.9648241206030153e-05,
|
| 2096 |
+
"loss": 0.0627,
|
| 2097 |
+
"step": 1680
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 8.47,
|
| 2101 |
+
"learning_rate": 2.943886097152429e-05,
|
| 2102 |
+
"loss": 0.0603,
|
| 2103 |
+
"step": 1685
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 8.49,
|
| 2107 |
+
"learning_rate": 2.9229480737018428e-05,
|
| 2108 |
+
"loss": 0.0543,
|
| 2109 |
+
"step": 1690
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 8.52,
|
| 2113 |
+
"learning_rate": 2.9020100502512566e-05,
|
| 2114 |
+
"loss": 0.0975,
|
| 2115 |
+
"step": 1695
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 8.54,
|
| 2119 |
+
"learning_rate": 2.8810720268006703e-05,
|
| 2120 |
+
"loss": 0.0124,
|
| 2121 |
+
"step": 1700
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 8.57,
|
| 2125 |
+
"learning_rate": 2.860134003350084e-05,
|
| 2126 |
+
"loss": 0.0914,
|
| 2127 |
+
"step": 1705
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 8.59,
|
| 2131 |
+
"learning_rate": 2.8391959798994978e-05,
|
| 2132 |
+
"loss": 0.0461,
|
| 2133 |
+
"step": 1710
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 8.62,
|
| 2137 |
+
"learning_rate": 2.8182579564489115e-05,
|
| 2138 |
+
"loss": 0.0174,
|
| 2139 |
+
"step": 1715
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 8.64,
|
| 2143 |
+
"learning_rate": 2.7973199329983252e-05,
|
| 2144 |
+
"loss": 0.0043,
|
| 2145 |
+
"step": 1720
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 8.67,
|
| 2149 |
+
"learning_rate": 2.776381909547739e-05,
|
| 2150 |
+
"loss": 0.048,
|
| 2151 |
+
"step": 1725
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.69,
|
| 2155 |
+
"learning_rate": 2.7554438860971527e-05,
|
| 2156 |
+
"loss": 0.0277,
|
| 2157 |
+
"step": 1730
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 8.72,
|
| 2161 |
+
"learning_rate": 2.7345058626465665e-05,
|
| 2162 |
+
"loss": 0.125,
|
| 2163 |
+
"step": 1735
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 8.74,
|
| 2167 |
+
"learning_rate": 2.7135678391959802e-05,
|
| 2168 |
+
"loss": 0.0723,
|
| 2169 |
+
"step": 1740
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 8.77,
|
| 2173 |
+
"learning_rate": 2.692629815745394e-05,
|
| 2174 |
+
"loss": 0.1748,
|
| 2175 |
+
"step": 1745
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 8.79,
|
| 2179 |
+
"learning_rate": 2.6716917922948077e-05,
|
| 2180 |
+
"loss": 0.0646,
|
| 2181 |
+
"step": 1750
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 8.82,
|
| 2185 |
+
"learning_rate": 2.6507537688442214e-05,
|
| 2186 |
+
"loss": 0.0415,
|
| 2187 |
+
"step": 1755
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 8.84,
|
| 2191 |
+
"learning_rate": 2.629815745393635e-05,
|
| 2192 |
+
"loss": 0.0726,
|
| 2193 |
+
"step": 1760
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 8.87,
|
| 2197 |
+
"learning_rate": 2.608877721943049e-05,
|
| 2198 |
+
"loss": 0.0047,
|
| 2199 |
+
"step": 1765
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 8.89,
|
| 2203 |
+
"learning_rate": 2.5879396984924626e-05,
|
| 2204 |
+
"loss": 0.0561,
|
| 2205 |
+
"step": 1770
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 8.92,
|
| 2209 |
+
"learning_rate": 2.5670016750418764e-05,
|
| 2210 |
+
"loss": 0.0529,
|
| 2211 |
+
"step": 1775
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 8.94,
|
| 2215 |
+
"learning_rate": 2.54606365159129e-05,
|
| 2216 |
+
"loss": 0.0009,
|
| 2217 |
+
"step": 1780
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 8.97,
|
| 2221 |
+
"learning_rate": 2.525125628140704e-05,
|
| 2222 |
+
"loss": 0.0378,
|
| 2223 |
+
"step": 1785
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 8.99,
|
| 2227 |
+
"learning_rate": 2.5041876046901176e-05,
|
| 2228 |
+
"loss": 0.0141,
|
| 2229 |
+
"step": 1790
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 9.0,
|
| 2233 |
+
"eval_accuracy": 0.9872761545711592,
|
| 2234 |
+
"eval_loss": 0.05207793414592743,
|
| 2235 |
+
"eval_runtime": 91.6702,
|
| 2236 |
+
"eval_samples_per_second": 23.148,
|
| 2237 |
+
"eval_steps_per_second": 2.902,
|
| 2238 |
+
"step": 1791
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 9.02,
|
| 2242 |
+
"learning_rate": 2.4832495812395313e-05,
|
| 2243 |
+
"loss": 0.0666,
|
| 2244 |
+
"step": 1795
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 9.05,
|
| 2248 |
+
"learning_rate": 2.462311557788945e-05,
|
| 2249 |
+
"loss": 0.0038,
|
| 2250 |
+
"step": 1800
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 9.07,
|
| 2254 |
+
"learning_rate": 2.4413735343383588e-05,
|
| 2255 |
+
"loss": 0.1081,
|
| 2256 |
+
"step": 1805
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 9.1,
|
| 2260 |
+
"learning_rate": 2.4204355108877725e-05,
|
| 2261 |
+
"loss": 0.0273,
|
| 2262 |
+
"step": 1810
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 9.12,
|
| 2266 |
+
"learning_rate": 2.3994974874371863e-05,
|
| 2267 |
+
"loss": 0.0519,
|
| 2268 |
+
"step": 1815
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 9.15,
|
| 2272 |
+
"learning_rate": 2.3785594639866e-05,
|
| 2273 |
+
"loss": 0.0233,
|
| 2274 |
+
"step": 1820
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 9.17,
|
| 2278 |
+
"learning_rate": 2.3576214405360137e-05,
|
| 2279 |
+
"loss": 0.0128,
|
| 2280 |
+
"step": 1825
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 9.2,
|
| 2284 |
+
"learning_rate": 2.3366834170854275e-05,
|
| 2285 |
+
"loss": 0.0657,
|
| 2286 |
+
"step": 1830
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 9.22,
|
| 2290 |
+
"learning_rate": 2.3157453936348412e-05,
|
| 2291 |
+
"loss": 0.016,
|
| 2292 |
+
"step": 1835
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 9.25,
|
| 2296 |
+
"learning_rate": 2.294807370184255e-05,
|
| 2297 |
+
"loss": 0.0528,
|
| 2298 |
+
"step": 1840
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 9.27,
|
| 2302 |
+
"learning_rate": 2.2738693467336687e-05,
|
| 2303 |
+
"loss": 0.0561,
|
| 2304 |
+
"step": 1845
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 9.3,
|
| 2308 |
+
"learning_rate": 2.2529313232830824e-05,
|
| 2309 |
+
"loss": 0.0826,
|
| 2310 |
+
"step": 1850
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 9.32,
|
| 2314 |
+
"learning_rate": 2.231993299832496e-05,
|
| 2315 |
+
"loss": 0.0392,
|
| 2316 |
+
"step": 1855
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 9.35,
|
| 2320 |
+
"learning_rate": 2.21105527638191e-05,
|
| 2321 |
+
"loss": 0.0815,
|
| 2322 |
+
"step": 1860
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 9.37,
|
| 2326 |
+
"learning_rate": 2.1901172529313236e-05,
|
| 2327 |
+
"loss": 0.0392,
|
| 2328 |
+
"step": 1865
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 9.4,
|
| 2332 |
+
"learning_rate": 2.169179229480737e-05,
|
| 2333 |
+
"loss": 0.0209,
|
| 2334 |
+
"step": 1870
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 9.42,
|
| 2338 |
+
"learning_rate": 2.1482412060301508e-05,
|
| 2339 |
+
"loss": 0.1453,
|
| 2340 |
+
"step": 1875
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 9.45,
|
| 2344 |
+
"learning_rate": 2.1273031825795645e-05,
|
| 2345 |
+
"loss": 0.0127,
|
| 2346 |
+
"step": 1880
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 9.47,
|
| 2350 |
+
"learning_rate": 2.1063651591289783e-05,
|
| 2351 |
+
"loss": 0.0012,
|
| 2352 |
+
"step": 1885
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 9.5,
|
| 2356 |
+
"learning_rate": 2.085427135678392e-05,
|
| 2357 |
+
"loss": 0.0525,
|
| 2358 |
+
"step": 1890
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 9.52,
|
| 2362 |
+
"learning_rate": 2.0644891122278057e-05,
|
| 2363 |
+
"loss": 0.0213,
|
| 2364 |
+
"step": 1895
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 9.55,
|
| 2368 |
+
"learning_rate": 2.0435510887772195e-05,
|
| 2369 |
+
"loss": 0.0595,
|
| 2370 |
+
"step": 1900
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 9.57,
|
| 2374 |
+
"learning_rate": 2.0226130653266332e-05,
|
| 2375 |
+
"loss": 0.0434,
|
| 2376 |
+
"step": 1905
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 9.6,
|
| 2380 |
+
"learning_rate": 2.001675041876047e-05,
|
| 2381 |
+
"loss": 0.0081,
|
| 2382 |
+
"step": 1910
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 9.62,
|
| 2386 |
+
"learning_rate": 1.9807370184254607e-05,
|
| 2387 |
+
"loss": 0.0385,
|
| 2388 |
+
"step": 1915
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 9.65,
|
| 2392 |
+
"learning_rate": 1.9597989949748744e-05,
|
| 2393 |
+
"loss": 0.054,
|
| 2394 |
+
"step": 1920
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 9.67,
|
| 2398 |
+
"learning_rate": 1.938860971524288e-05,
|
| 2399 |
+
"loss": 0.0406,
|
| 2400 |
+
"step": 1925
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 9.7,
|
| 2404 |
+
"learning_rate": 1.917922948073702e-05,
|
| 2405 |
+
"loss": 0.0373,
|
| 2406 |
+
"step": 1930
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.72,
|
| 2410 |
+
"learning_rate": 1.8969849246231156e-05,
|
| 2411 |
+
"loss": 0.0314,
|
| 2412 |
+
"step": 1935
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 9.75,
|
| 2416 |
+
"learning_rate": 1.8760469011725294e-05,
|
| 2417 |
+
"loss": 0.0017,
|
| 2418 |
+
"step": 1940
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 9.77,
|
| 2422 |
+
"learning_rate": 1.855108877721943e-05,
|
| 2423 |
+
"loss": 0.0315,
|
| 2424 |
+
"step": 1945
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 9.8,
|
| 2428 |
+
"learning_rate": 1.834170854271357e-05,
|
| 2429 |
+
"loss": 0.002,
|
| 2430 |
+
"step": 1950
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 9.82,
|
| 2434 |
+
"learning_rate": 1.8132328308207706e-05,
|
| 2435 |
+
"loss": 0.076,
|
| 2436 |
+
"step": 1955
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 9.85,
|
| 2440 |
+
"learning_rate": 1.7922948073701843e-05,
|
| 2441 |
+
"loss": 0.049,
|
| 2442 |
+
"step": 1960
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 9.87,
|
| 2446 |
+
"learning_rate": 1.771356783919598e-05,
|
| 2447 |
+
"loss": 0.0174,
|
| 2448 |
+
"step": 1965
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 9.9,
|
| 2452 |
+
"learning_rate": 1.7504187604690118e-05,
|
| 2453 |
+
"loss": 0.0064,
|
| 2454 |
+
"step": 1970
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 9.92,
|
| 2458 |
+
"learning_rate": 1.7294807370184255e-05,
|
| 2459 |
+
"loss": 0.0382,
|
| 2460 |
+
"step": 1975
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 9.95,
|
| 2464 |
+
"learning_rate": 1.7085427135678393e-05,
|
| 2465 |
+
"loss": 0.0646,
|
| 2466 |
+
"step": 1980
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 9.97,
|
| 2470 |
+
"learning_rate": 1.687604690117253e-05,
|
| 2471 |
+
"loss": 0.003,
|
| 2472 |
+
"step": 1985
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 10.0,
|
| 2476 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 2477 |
+
"loss": 0.038,
|
| 2478 |
+
"step": 1990
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 10.0,
|
| 2482 |
+
"eval_accuracy": 0.9934024505183789,
|
| 2483 |
+
"eval_loss": 0.02926916442811489,
|
| 2484 |
+
"eval_runtime": 91.6568,
|
| 2485 |
+
"eval_samples_per_second": 23.152,
|
| 2486 |
+
"eval_steps_per_second": 2.902,
|
| 2487 |
+
"step": 1990
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 10.03,
|
| 2491 |
+
"learning_rate": 1.6457286432160805e-05,
|
| 2492 |
+
"loss": 0.0329,
|
| 2493 |
+
"step": 1995
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 10.05,
|
| 2497 |
+
"learning_rate": 1.6247906197654942e-05,
|
| 2498 |
+
"loss": 0.0641,
|
| 2499 |
+
"step": 2000
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 10.08,
|
| 2503 |
+
"learning_rate": 1.603852596314908e-05,
|
| 2504 |
+
"loss": 0.0806,
|
| 2505 |
+
"step": 2005
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 10.1,
|
| 2509 |
+
"learning_rate": 1.5829145728643217e-05,
|
| 2510 |
+
"loss": 0.0285,
|
| 2511 |
+
"step": 2010
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 10.13,
|
| 2515 |
+
"learning_rate": 1.5619765494137354e-05,
|
| 2516 |
+
"loss": 0.0191,
|
| 2517 |
+
"step": 2015
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 10.15,
|
| 2521 |
+
"learning_rate": 1.5410385259631492e-05,
|
| 2522 |
+
"loss": 0.0372,
|
| 2523 |
+
"step": 2020
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 10.18,
|
| 2527 |
+
"learning_rate": 1.5201005025125627e-05,
|
| 2528 |
+
"loss": 0.0013,
|
| 2529 |
+
"step": 2025
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 10.2,
|
| 2533 |
+
"learning_rate": 1.4991624790619765e-05,
|
| 2534 |
+
"loss": 0.0009,
|
| 2535 |
+
"step": 2030
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 10.23,
|
| 2539 |
+
"learning_rate": 1.4782244556113902e-05,
|
| 2540 |
+
"loss": 0.0111,
|
| 2541 |
+
"step": 2035
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 10.25,
|
| 2545 |
+
"learning_rate": 1.457286432160804e-05,
|
| 2546 |
+
"loss": 0.0003,
|
| 2547 |
+
"step": 2040
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 10.28,
|
| 2551 |
+
"learning_rate": 1.4363484087102177e-05,
|
| 2552 |
+
"loss": 0.0002,
|
| 2553 |
+
"step": 2045
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 10.3,
|
| 2557 |
+
"learning_rate": 1.4154103852596314e-05,
|
| 2558 |
+
"loss": 0.005,
|
| 2559 |
+
"step": 2050
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 10.33,
|
| 2563 |
+
"learning_rate": 1.3944723618090452e-05,
|
| 2564 |
+
"loss": 0.0094,
|
| 2565 |
+
"step": 2055
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 10.35,
|
| 2569 |
+
"learning_rate": 1.3735343383584589e-05,
|
| 2570 |
+
"loss": 0.0001,
|
| 2571 |
+
"step": 2060
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 10.38,
|
| 2575 |
+
"learning_rate": 1.3525963149078726e-05,
|
| 2576 |
+
"loss": 0.0084,
|
| 2577 |
+
"step": 2065
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 10.4,
|
| 2581 |
+
"learning_rate": 1.3316582914572864e-05,
|
| 2582 |
+
"loss": 0.0192,
|
| 2583 |
+
"step": 2070
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 10.43,
|
| 2587 |
+
"learning_rate": 1.3107202680067001e-05,
|
| 2588 |
+
"loss": 0.0324,
|
| 2589 |
+
"step": 2075
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 10.45,
|
| 2593 |
+
"learning_rate": 1.2897822445561139e-05,
|
| 2594 |
+
"loss": 0.0463,
|
| 2595 |
+
"step": 2080
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 10.48,
|
| 2599 |
+
"learning_rate": 1.2688442211055276e-05,
|
| 2600 |
+
"loss": 0.0644,
|
| 2601 |
+
"step": 2085
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 10.5,
|
| 2605 |
+
"learning_rate": 1.2479061976549413e-05,
|
| 2606 |
+
"loss": 0.0702,
|
| 2607 |
+
"step": 2090
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 10.53,
|
| 2611 |
+
"learning_rate": 1.226968174204355e-05,
|
| 2612 |
+
"loss": 0.0377,
|
| 2613 |
+
"step": 2095
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 10.55,
|
| 2617 |
+
"learning_rate": 1.2060301507537688e-05,
|
| 2618 |
+
"loss": 0.0688,
|
| 2619 |
+
"step": 2100
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 10.58,
|
| 2623 |
+
"learning_rate": 1.1850921273031825e-05,
|
| 2624 |
+
"loss": 0.0039,
|
| 2625 |
+
"step": 2105
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 10.6,
|
| 2629 |
+
"learning_rate": 1.1641541038525963e-05,
|
| 2630 |
+
"loss": 0.0028,
|
| 2631 |
+
"step": 2110
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 10.63,
|
| 2635 |
+
"learning_rate": 1.14321608040201e-05,
|
| 2636 |
+
"loss": 0.0518,
|
| 2637 |
+
"step": 2115
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 10.65,
|
| 2641 |
+
"learning_rate": 1.1222780569514238e-05,
|
| 2642 |
+
"loss": 0.0279,
|
| 2643 |
+
"step": 2120
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 10.68,
|
| 2647 |
+
"learning_rate": 1.1013400335008375e-05,
|
| 2648 |
+
"loss": 0.0001,
|
| 2649 |
+
"step": 2125
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 10.7,
|
| 2653 |
+
"learning_rate": 1.0804020100502512e-05,
|
| 2654 |
+
"loss": 0.0108,
|
| 2655 |
+
"step": 2130
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 10.73,
|
| 2659 |
+
"learning_rate": 1.059463986599665e-05,
|
| 2660 |
+
"loss": 0.0513,
|
| 2661 |
+
"step": 2135
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 10.75,
|
| 2665 |
+
"learning_rate": 1.0385259631490787e-05,
|
| 2666 |
+
"loss": 0.0208,
|
| 2667 |
+
"step": 2140
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 10.78,
|
| 2671 |
+
"learning_rate": 1.0175879396984924e-05,
|
| 2672 |
+
"loss": 0.0141,
|
| 2673 |
+
"step": 2145
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 10.8,
|
| 2677 |
+
"learning_rate": 9.966499162479062e-06,
|
| 2678 |
+
"loss": 0.0063,
|
| 2679 |
+
"step": 2150
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 10.83,
|
| 2683 |
+
"learning_rate": 9.7571189279732e-06,
|
| 2684 |
+
"loss": 0.0723,
|
| 2685 |
+
"step": 2155
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 10.85,
|
| 2689 |
+
"learning_rate": 9.547738693467337e-06,
|
| 2690 |
+
"loss": 0.0002,
|
| 2691 |
+
"step": 2160
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 10.88,
|
| 2695 |
+
"learning_rate": 9.338358458961474e-06,
|
| 2696 |
+
"loss": 0.0477,
|
| 2697 |
+
"step": 2165
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 10.9,
|
| 2701 |
+
"learning_rate": 9.128978224455611e-06,
|
| 2702 |
+
"loss": 0.0868,
|
| 2703 |
+
"step": 2170
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 10.93,
|
| 2707 |
+
"learning_rate": 8.919597989949749e-06,
|
| 2708 |
+
"loss": 0.0424,
|
| 2709 |
+
"step": 2175
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 10.95,
|
| 2713 |
+
"learning_rate": 8.710217755443886e-06,
|
| 2714 |
+
"loss": 0.0728,
|
| 2715 |
+
"step": 2180
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 10.98,
|
| 2719 |
+
"learning_rate": 8.500837520938023e-06,
|
| 2720 |
+
"loss": 0.0044,
|
| 2721 |
+
"step": 2185
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 11.0,
|
| 2725 |
+
"eval_accuracy": 0.9971724787935909,
|
| 2726 |
+
"eval_loss": 0.01605476438999176,
|
| 2727 |
+
"eval_runtime": 91.6819,
|
| 2728 |
+
"eval_samples_per_second": 23.145,
|
| 2729 |
+
"eval_steps_per_second": 2.901,
|
| 2730 |
+
"step": 2189
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 11.01,
|
| 2734 |
+
"learning_rate": 8.291457286432161e-06,
|
| 2735 |
+
"loss": 0.0324,
|
| 2736 |
+
"step": 2190
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 11.03,
|
| 2740 |
+
"learning_rate": 8.082077051926298e-06,
|
| 2741 |
+
"loss": 0.0002,
|
| 2742 |
+
"step": 2195
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 11.06,
|
| 2746 |
+
"learning_rate": 7.872696817420436e-06,
|
| 2747 |
+
"loss": 0.0001,
|
| 2748 |
+
"step": 2200
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 11.08,
|
| 2752 |
+
"learning_rate": 7.663316582914573e-06,
|
| 2753 |
+
"loss": 0.0036,
|
| 2754 |
+
"step": 2205
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 11.11,
|
| 2758 |
+
"learning_rate": 7.45393634840871e-06,
|
| 2759 |
+
"loss": 0.151,
|
| 2760 |
+
"step": 2210
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 11.13,
|
| 2764 |
+
"learning_rate": 7.244556113902848e-06,
|
| 2765 |
+
"loss": 0.004,
|
| 2766 |
+
"step": 2215
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 11.16,
|
| 2770 |
+
"learning_rate": 7.035175879396985e-06,
|
| 2771 |
+
"loss": 0.0008,
|
| 2772 |
+
"step": 2220
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 11.18,
|
| 2776 |
+
"learning_rate": 6.8257956448911225e-06,
|
| 2777 |
+
"loss": 0.0643,
|
| 2778 |
+
"step": 2225
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 11.21,
|
| 2782 |
+
"learning_rate": 6.61641541038526e-06,
|
| 2783 |
+
"loss": 0.0754,
|
| 2784 |
+
"step": 2230
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 11.23,
|
| 2788 |
+
"learning_rate": 6.407035175879397e-06,
|
| 2789 |
+
"loss": 0.0048,
|
| 2790 |
+
"step": 2235
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 11.26,
|
| 2794 |
+
"learning_rate": 6.197654941373535e-06,
|
| 2795 |
+
"loss": 0.0028,
|
| 2796 |
+
"step": 2240
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 11.28,
|
| 2800 |
+
"learning_rate": 5.988274706867672e-06,
|
| 2801 |
+
"loss": 0.0378,
|
| 2802 |
+
"step": 2245
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 11.31,
|
| 2806 |
+
"learning_rate": 5.778894472361809e-06,
|
| 2807 |
+
"loss": 0.0561,
|
| 2808 |
+
"step": 2250
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 11.33,
|
| 2812 |
+
"learning_rate": 5.569514237855947e-06,
|
| 2813 |
+
"loss": 0.0005,
|
| 2814 |
+
"step": 2255
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 11.36,
|
| 2818 |
+
"learning_rate": 5.360134003350084e-06,
|
| 2819 |
+
"loss": 0.0361,
|
| 2820 |
+
"step": 2260
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 11.38,
|
| 2824 |
+
"learning_rate": 5.1507537688442215e-06,
|
| 2825 |
+
"loss": 0.0348,
|
| 2826 |
+
"step": 2265
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 11.41,
|
| 2830 |
+
"learning_rate": 4.941373534338359e-06,
|
| 2831 |
+
"loss": 0.0509,
|
| 2832 |
+
"step": 2270
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 11.43,
|
| 2836 |
+
"learning_rate": 4.731993299832496e-06,
|
| 2837 |
+
"loss": 0.0062,
|
| 2838 |
+
"step": 2275
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 11.46,
|
| 2842 |
+
"learning_rate": 4.522613065326634e-06,
|
| 2843 |
+
"loss": 0.0228,
|
| 2844 |
+
"step": 2280
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 11.48,
|
| 2848 |
+
"learning_rate": 4.313232830820771e-06,
|
| 2849 |
+
"loss": 0.0244,
|
| 2850 |
+
"step": 2285
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 11.51,
|
| 2854 |
+
"learning_rate": 4.103852596314908e-06,
|
| 2855 |
+
"loss": 0.0003,
|
| 2856 |
+
"step": 2290
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 11.53,
|
| 2860 |
+
"learning_rate": 3.894472361809046e-06,
|
| 2861 |
+
"loss": 0.0384,
|
| 2862 |
+
"step": 2295
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 11.56,
|
| 2866 |
+
"learning_rate": 3.685092127303183e-06,
|
| 2867 |
+
"loss": 0.0251,
|
| 2868 |
+
"step": 2300
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 11.58,
|
| 2872 |
+
"learning_rate": 3.4757118927973206e-06,
|
| 2873 |
+
"loss": 0.0548,
|
| 2874 |
+
"step": 2305
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 11.61,
|
| 2878 |
+
"learning_rate": 3.2663316582914575e-06,
|
| 2879 |
+
"loss": 0.0022,
|
| 2880 |
+
"step": 2310
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 11.63,
|
| 2884 |
+
"learning_rate": 3.056951423785595e-06,
|
| 2885 |
+
"loss": 0.0004,
|
| 2886 |
+
"step": 2315
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 11.66,
|
| 2890 |
+
"learning_rate": 2.8475711892797323e-06,
|
| 2891 |
+
"loss": 0.0002,
|
| 2892 |
+
"step": 2320
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 11.68,
|
| 2896 |
+
"learning_rate": 2.6381909547738696e-06,
|
| 2897 |
+
"loss": 0.0001,
|
| 2898 |
+
"step": 2325
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 11.71,
|
| 2902 |
+
"learning_rate": 2.428810720268007e-06,
|
| 2903 |
+
"loss": 0.0004,
|
| 2904 |
+
"step": 2330
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 11.73,
|
| 2908 |
+
"learning_rate": 2.219430485762144e-06,
|
| 2909 |
+
"loss": 0.0211,
|
| 2910 |
+
"step": 2335
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 11.76,
|
| 2914 |
+
"learning_rate": 2.0100502512562813e-06,
|
| 2915 |
+
"loss": 0.0036,
|
| 2916 |
+
"step": 2340
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 11.78,
|
| 2920 |
+
"learning_rate": 1.800670016750419e-06,
|
| 2921 |
+
"loss": 0.0122,
|
| 2922 |
+
"step": 2345
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 11.81,
|
| 2926 |
+
"learning_rate": 1.5912897822445563e-06,
|
| 2927 |
+
"loss": 0.0265,
|
| 2928 |
+
"step": 2350
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 11.83,
|
| 2932 |
+
"learning_rate": 1.3819095477386937e-06,
|
| 2933 |
+
"loss": 0.01,
|
| 2934 |
+
"step": 2355
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 11.86,
|
| 2938 |
+
"learning_rate": 1.1725293132328309e-06,
|
| 2939 |
+
"loss": 0.0021,
|
| 2940 |
+
"step": 2360
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 11.88,
|
| 2944 |
+
"learning_rate": 9.631490787269682e-07,
|
| 2945 |
+
"loss": 0.0,
|
| 2946 |
+
"step": 2365
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 11.91,
|
| 2950 |
+
"learning_rate": 7.537688442211055e-07,
|
| 2951 |
+
"loss": 0.0001,
|
| 2952 |
+
"step": 2370
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 11.93,
|
| 2956 |
+
"learning_rate": 5.443886097152429e-07,
|
| 2957 |
+
"loss": 0.0212,
|
| 2958 |
+
"step": 2375
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 11.96,
|
| 2962 |
+
"learning_rate": 3.3500837520938026e-07,
|
| 2963 |
+
"loss": 0.0109,
|
| 2964 |
+
"step": 2380
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 11.98,
|
| 2968 |
+
"learning_rate": 1.2562814070351758e-07,
|
| 2969 |
+
"loss": 0.0006,
|
| 2970 |
+
"step": 2385
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 12.0,
|
| 2974 |
+
"eval_accuracy": 0.9967012252591895,
|
| 2975 |
+
"eval_loss": 0.019685085862874985,
|
| 2976 |
+
"eval_runtime": 91.8593,
|
| 2977 |
+
"eval_samples_per_second": 23.101,
|
| 2978 |
+
"eval_steps_per_second": 2.896,
|
| 2979 |
+
"step": 2388
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 12.0,
|
| 2983 |
+
"step": 2388,
|
| 2984 |
+
"total_flos": 1.7585593436847735e+19,
|
| 2985 |
+
"train_loss": 0.12566840387472886,
|
| 2986 |
+
"train_runtime": 12106.0173,
|
| 2987 |
+
"train_samples_per_second": 6.308,
|
| 2988 |
+
"train_steps_per_second": 0.197
|
| 2989 |
+
}
|
| 2990 |
+
],
|
| 2991 |
+
"logging_steps": 5,
|
| 2992 |
+
"max_steps": 2388,
|
| 2993 |
+
"num_train_epochs": 12,
|
| 2994 |
+
"save_steps": 500,
|
| 2995 |
+
"total_flos": 1.7585593436847735e+19,
|
| 2996 |
+
"trial_name": null,
|
| 2997 |
+
"trial_params": null
|
| 2998 |
+
}
|