End of training
Browse files- README.md +5 -5
- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Apr03_09-11-38_66437d7433cf/events.out.tfevents.1712139455.66437d7433cf.8793.4 +3 -0
- train_results.json +8 -0
- trainer_state.json +3701 -0
README.md
CHANGED
|
@@ -20,11 +20,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 20 |
|
| 21 |
This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
|
| 22 |
It achieves the following results on the evaluation set:
|
| 23 |
-
- Loss: 0.
|
| 24 |
-
- Accuracy: 0.
|
| 25 |
-
- Precision: 0.
|
| 26 |
-
- Recall: 0.
|
| 27 |
-
- F1: 0.
|
| 28 |
|
| 29 |
## Model description
|
| 30 |
|
|
|
|
| 20 |
|
| 21 |
This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
|
| 22 |
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.1502
|
| 24 |
+
- Accuracy: 0.9755
|
| 25 |
+
- Precision: 0.9755
|
| 26 |
+
- Recall: 0.9755
|
| 27 |
+
- F1: 0.9755
|
| 28 |
|
| 29 |
## Model description
|
| 30 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9754901960784313,
|
| 4 |
+
"eval_f1": 0.9754901960784313,
|
| 5 |
+
"eval_loss": 0.15021921694278717,
|
| 6 |
+
"eval_precision": 0.9754901960784313,
|
| 7 |
+
"eval_recall": 0.9754901960784313,
|
| 8 |
+
"eval_runtime": 16.9389,
|
| 9 |
+
"eval_samples_per_second": 60.217,
|
| 10 |
+
"eval_steps_per_second": 7.557,
|
| 11 |
+
"total_flos": 5.562769847811564e+18,
|
| 12 |
+
"train_loss": 0.025327244219600944,
|
| 13 |
+
"train_runtime": 3846.586,
|
| 14 |
+
"train_samples_per_second": 18.637,
|
| 15 |
+
"train_steps_per_second": 1.167
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9754901960784313,
|
| 4 |
+
"eval_f1": 0.9754901960784313,
|
| 5 |
+
"eval_loss": 0.15021921694278717,
|
| 6 |
+
"eval_precision": 0.9754901960784313,
|
| 7 |
+
"eval_recall": 0.9754901960784313,
|
| 8 |
+
"eval_runtime": 16.9389,
|
| 9 |
+
"eval_samples_per_second": 60.217,
|
| 10 |
+
"eval_steps_per_second": 7.557
|
| 11 |
+
}
|
runs/Apr03_09-11-38_66437d7433cf/events.out.tfevents.1712139455.66437d7433cf.8793.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34d20b3fe1ef836a4b70ba79f59bb4c91b9dc9808bce8f6718bb4c7949c0c87c
|
| 3 |
+
size 560
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 5.562769847811564e+18,
|
| 4 |
+
"train_loss": 0.025327244219600944,
|
| 5 |
+
"train_runtime": 3846.586,
|
| 6 |
+
"train_samples_per_second": 18.637,
|
| 7 |
+
"train_steps_per_second": 1.167
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3701 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.15021921694278717,
|
| 3 |
+
"best_model_checkpoint": "ViT_Flower102_2/checkpoint-1600",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 4490,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02,
|
| 13 |
+
"grad_norm": 0.0026773272547870874,
|
| 14 |
+
"learning_rate": 0.00019955456570155904,
|
| 15 |
+
"loss": 0.0009,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04,
|
| 20 |
+
"grad_norm": 0.005304301157593727,
|
| 21 |
+
"learning_rate": 0.00019910913140311804,
|
| 22 |
+
"loss": 0.0009,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07,
|
| 27 |
+
"grad_norm": 20.306926727294922,
|
| 28 |
+
"learning_rate": 0.00019866369710467706,
|
| 29 |
+
"loss": 0.0194,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.09,
|
| 34 |
+
"grad_norm": 2.321718692779541,
|
| 35 |
+
"learning_rate": 0.0001982182628062361,
|
| 36 |
+
"loss": 0.1544,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.11,
|
| 41 |
+
"grad_norm": 0.021045241504907608,
|
| 42 |
+
"learning_rate": 0.00019777282850779511,
|
| 43 |
+
"loss": 0.0701,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.13,
|
| 48 |
+
"grad_norm": 0.02391223795711994,
|
| 49 |
+
"learning_rate": 0.00019732739420935414,
|
| 50 |
+
"loss": 0.1303,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.16,
|
| 55 |
+
"grad_norm": 0.20434625446796417,
|
| 56 |
+
"learning_rate": 0.00019688195991091317,
|
| 57 |
+
"loss": 0.1413,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.18,
|
| 62 |
+
"grad_norm": 0.2549870014190674,
|
| 63 |
+
"learning_rate": 0.00019643652561247217,
|
| 64 |
+
"loss": 0.1047,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.2,
|
| 69 |
+
"grad_norm": 6.997387886047363,
|
| 70 |
+
"learning_rate": 0.0001959910913140312,
|
| 71 |
+
"loss": 0.1234,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.22,
|
| 76 |
+
"grad_norm": 5.206876277923584,
|
| 77 |
+
"learning_rate": 0.0001955456570155902,
|
| 78 |
+
"loss": 0.053,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.22,
|
| 83 |
+
"eval_accuracy": 0.9235294117647059,
|
| 84 |
+
"eval_f1": 0.9235294117647059,
|
| 85 |
+
"eval_loss": 0.3198450803756714,
|
| 86 |
+
"eval_precision": 0.9235294117647059,
|
| 87 |
+
"eval_recall": 0.9235294117647059,
|
| 88 |
+
"eval_runtime": 17.2015,
|
| 89 |
+
"eval_samples_per_second": 59.297,
|
| 90 |
+
"eval_steps_per_second": 7.441,
|
| 91 |
+
"step": 100
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 0.24,
|
| 95 |
+
"grad_norm": 6.388282299041748,
|
| 96 |
+
"learning_rate": 0.00019510022271714922,
|
| 97 |
+
"loss": 0.1377,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.27,
|
| 102 |
+
"grad_norm": 5.129618167877197,
|
| 103 |
+
"learning_rate": 0.00019465478841870825,
|
| 104 |
+
"loss": 0.0561,
|
| 105 |
+
"step": 120
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.29,
|
| 109 |
+
"grad_norm": 8.198286056518555,
|
| 110 |
+
"learning_rate": 0.00019420935412026727,
|
| 111 |
+
"loss": 0.0417,
|
| 112 |
+
"step": 130
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.31,
|
| 116 |
+
"grad_norm": 7.592113494873047,
|
| 117 |
+
"learning_rate": 0.0001937639198218263,
|
| 118 |
+
"loss": 0.0388,
|
| 119 |
+
"step": 140
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.33,
|
| 123 |
+
"grad_norm": 2.7412798404693604,
|
| 124 |
+
"learning_rate": 0.00019331848552338533,
|
| 125 |
+
"loss": 0.0921,
|
| 126 |
+
"step": 150
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.36,
|
| 130 |
+
"grad_norm": 0.12602120637893677,
|
| 131 |
+
"learning_rate": 0.00019287305122494432,
|
| 132 |
+
"loss": 0.1905,
|
| 133 |
+
"step": 160
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 0.38,
|
| 137 |
+
"grad_norm": 5.073046684265137,
|
| 138 |
+
"learning_rate": 0.00019242761692650335,
|
| 139 |
+
"loss": 0.1331,
|
| 140 |
+
"step": 170
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.4,
|
| 144 |
+
"grad_norm": 0.5400639176368713,
|
| 145 |
+
"learning_rate": 0.00019198218262806238,
|
| 146 |
+
"loss": 0.0315,
|
| 147 |
+
"step": 180
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.42,
|
| 151 |
+
"grad_norm": 11.878341674804688,
|
| 152 |
+
"learning_rate": 0.00019153674832962138,
|
| 153 |
+
"loss": 0.2794,
|
| 154 |
+
"step": 190
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.45,
|
| 158 |
+
"grad_norm": 0.48547011613845825,
|
| 159 |
+
"learning_rate": 0.0001910913140311804,
|
| 160 |
+
"loss": 0.1225,
|
| 161 |
+
"step": 200
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.45,
|
| 165 |
+
"eval_accuracy": 0.9166666666666666,
|
| 166 |
+
"eval_f1": 0.9166666666666666,
|
| 167 |
+
"eval_loss": 0.40865278244018555,
|
| 168 |
+
"eval_precision": 0.9166666666666666,
|
| 169 |
+
"eval_recall": 0.9166666666666666,
|
| 170 |
+
"eval_runtime": 17.0881,
|
| 171 |
+
"eval_samples_per_second": 59.691,
|
| 172 |
+
"eval_steps_per_second": 7.491,
|
| 173 |
+
"step": 200
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.47,
|
| 177 |
+
"grad_norm": 0.015034608542919159,
|
| 178 |
+
"learning_rate": 0.00019064587973273943,
|
| 179 |
+
"loss": 0.0953,
|
| 180 |
+
"step": 210
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.49,
|
| 184 |
+
"grad_norm": 1.665817141532898,
|
| 185 |
+
"learning_rate": 0.00019020044543429846,
|
| 186 |
+
"loss": 0.081,
|
| 187 |
+
"step": 220
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.51,
|
| 191 |
+
"grad_norm": 1.737998127937317,
|
| 192 |
+
"learning_rate": 0.00018975501113585748,
|
| 193 |
+
"loss": 0.037,
|
| 194 |
+
"step": 230
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.53,
|
| 198 |
+
"grad_norm": 0.08088938146829605,
|
| 199 |
+
"learning_rate": 0.00018930957683741648,
|
| 200 |
+
"loss": 0.2113,
|
| 201 |
+
"step": 240
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.56,
|
| 205 |
+
"grad_norm": 6.148426055908203,
|
| 206 |
+
"learning_rate": 0.0001888641425389755,
|
| 207 |
+
"loss": 0.0674,
|
| 208 |
+
"step": 250
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.58,
|
| 212 |
+
"grad_norm": 0.07585727423429489,
|
| 213 |
+
"learning_rate": 0.00018841870824053454,
|
| 214 |
+
"loss": 0.1698,
|
| 215 |
+
"step": 260
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.6,
|
| 219 |
+
"grad_norm": 4.297791481018066,
|
| 220 |
+
"learning_rate": 0.00018797327394209353,
|
| 221 |
+
"loss": 0.1087,
|
| 222 |
+
"step": 270
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.62,
|
| 226 |
+
"grad_norm": 0.039582639932632446,
|
| 227 |
+
"learning_rate": 0.00018752783964365256,
|
| 228 |
+
"loss": 0.082,
|
| 229 |
+
"step": 280
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.65,
|
| 233 |
+
"grad_norm": 4.647770881652832,
|
| 234 |
+
"learning_rate": 0.0001870824053452116,
|
| 235 |
+
"loss": 0.0256,
|
| 236 |
+
"step": 290
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.67,
|
| 240 |
+
"grad_norm": 0.05055054649710655,
|
| 241 |
+
"learning_rate": 0.00018663697104677061,
|
| 242 |
+
"loss": 0.1985,
|
| 243 |
+
"step": 300
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.67,
|
| 247 |
+
"eval_accuracy": 0.9568627450980393,
|
| 248 |
+
"eval_f1": 0.9568627450980393,
|
| 249 |
+
"eval_loss": 0.20681221783161163,
|
| 250 |
+
"eval_precision": 0.9568627450980393,
|
| 251 |
+
"eval_recall": 0.9568627450980393,
|
| 252 |
+
"eval_runtime": 17.1131,
|
| 253 |
+
"eval_samples_per_second": 59.603,
|
| 254 |
+
"eval_steps_per_second": 7.48,
|
| 255 |
+
"step": 300
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.69,
|
| 259 |
+
"grad_norm": 12.765353202819824,
|
| 260 |
+
"learning_rate": 0.00018619153674832964,
|
| 261 |
+
"loss": 0.0698,
|
| 262 |
+
"step": 310
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.71,
|
| 266 |
+
"grad_norm": 9.829974174499512,
|
| 267 |
+
"learning_rate": 0.00018574610244988867,
|
| 268 |
+
"loss": 0.076,
|
| 269 |
+
"step": 320
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.73,
|
| 273 |
+
"grad_norm": 0.024047628045082092,
|
| 274 |
+
"learning_rate": 0.00018530066815144767,
|
| 275 |
+
"loss": 0.0335,
|
| 276 |
+
"step": 330
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.76,
|
| 280 |
+
"grad_norm": 6.1861066818237305,
|
| 281 |
+
"learning_rate": 0.0001848552338530067,
|
| 282 |
+
"loss": 0.1085,
|
| 283 |
+
"step": 340
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.78,
|
| 287 |
+
"grad_norm": 1.6630871295928955,
|
| 288 |
+
"learning_rate": 0.00018440979955456572,
|
| 289 |
+
"loss": 0.1537,
|
| 290 |
+
"step": 350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.8,
|
| 294 |
+
"grad_norm": 0.012870335020124912,
|
| 295 |
+
"learning_rate": 0.00018396436525612472,
|
| 296 |
+
"loss": 0.0729,
|
| 297 |
+
"step": 360
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.82,
|
| 301 |
+
"grad_norm": 0.9247069954872131,
|
| 302 |
+
"learning_rate": 0.00018351893095768375,
|
| 303 |
+
"loss": 0.0293,
|
| 304 |
+
"step": 370
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.85,
|
| 308 |
+
"grad_norm": 0.0049523478373885155,
|
| 309 |
+
"learning_rate": 0.00018307349665924277,
|
| 310 |
+
"loss": 0.043,
|
| 311 |
+
"step": 380
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.87,
|
| 315 |
+
"grad_norm": 6.6150336265563965,
|
| 316 |
+
"learning_rate": 0.0001826280623608018,
|
| 317 |
+
"loss": 0.0184,
|
| 318 |
+
"step": 390
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.89,
|
| 322 |
+
"grad_norm": 0.16694368422031403,
|
| 323 |
+
"learning_rate": 0.00018218262806236082,
|
| 324 |
+
"loss": 0.0804,
|
| 325 |
+
"step": 400
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.89,
|
| 329 |
+
"eval_accuracy": 0.9333333333333333,
|
| 330 |
+
"eval_f1": 0.9333333333333333,
|
| 331 |
+
"eval_loss": 0.3181270360946655,
|
| 332 |
+
"eval_precision": 0.9333333333333333,
|
| 333 |
+
"eval_recall": 0.9333333333333333,
|
| 334 |
+
"eval_runtime": 16.8894,
|
| 335 |
+
"eval_samples_per_second": 60.393,
|
| 336 |
+
"eval_steps_per_second": 7.579,
|
| 337 |
+
"step": 400
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.91,
|
| 341 |
+
"grad_norm": 2.3894243240356445,
|
| 342 |
+
"learning_rate": 0.00018173719376391982,
|
| 343 |
+
"loss": 0.0611,
|
| 344 |
+
"step": 410
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.94,
|
| 348 |
+
"grad_norm": 0.014911720529198647,
|
| 349 |
+
"learning_rate": 0.00018129175946547885,
|
| 350 |
+
"loss": 0.0172,
|
| 351 |
+
"step": 420
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.96,
|
| 355 |
+
"grad_norm": 0.01411391980946064,
|
| 356 |
+
"learning_rate": 0.00018084632516703788,
|
| 357 |
+
"loss": 0.2164,
|
| 358 |
+
"step": 430
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.98,
|
| 362 |
+
"grad_norm": 13.27253532409668,
|
| 363 |
+
"learning_rate": 0.00018040089086859688,
|
| 364 |
+
"loss": 0.1459,
|
| 365 |
+
"step": 440
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.0,
|
| 369 |
+
"grad_norm": 0.028676768764853477,
|
| 370 |
+
"learning_rate": 0.0001799554565701559,
|
| 371 |
+
"loss": 0.0702,
|
| 372 |
+
"step": 450
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.02,
|
| 376 |
+
"grad_norm": 4.484796047210693,
|
| 377 |
+
"learning_rate": 0.00017951002227171493,
|
| 378 |
+
"loss": 0.1319,
|
| 379 |
+
"step": 460
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 1.05,
|
| 383 |
+
"grad_norm": 5.17644739151001,
|
| 384 |
+
"learning_rate": 0.00017906458797327396,
|
| 385 |
+
"loss": 0.1276,
|
| 386 |
+
"step": 470
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.07,
|
| 390 |
+
"grad_norm": 4.405980110168457,
|
| 391 |
+
"learning_rate": 0.00017861915367483298,
|
| 392 |
+
"loss": 0.1718,
|
| 393 |
+
"step": 480
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.09,
|
| 397 |
+
"grad_norm": 0.1352098286151886,
|
| 398 |
+
"learning_rate": 0.000178173719376392,
|
| 399 |
+
"loss": 0.1796,
|
| 400 |
+
"step": 490
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 1.11,
|
| 404 |
+
"grad_norm": 8.556909561157227,
|
| 405 |
+
"learning_rate": 0.000177728285077951,
|
| 406 |
+
"loss": 0.1672,
|
| 407 |
+
"step": 500
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.11,
|
| 411 |
+
"eval_accuracy": 0.9274509803921569,
|
| 412 |
+
"eval_f1": 0.9274509803921569,
|
| 413 |
+
"eval_loss": 0.35819730162620544,
|
| 414 |
+
"eval_precision": 0.9274509803921569,
|
| 415 |
+
"eval_recall": 0.9274509803921569,
|
| 416 |
+
"eval_runtime": 16.8086,
|
| 417 |
+
"eval_samples_per_second": 60.683,
|
| 418 |
+
"eval_steps_per_second": 7.615,
|
| 419 |
+
"step": 500
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.14,
|
| 423 |
+
"grad_norm": 0.4172447919845581,
|
| 424 |
+
"learning_rate": 0.00017728285077951003,
|
| 425 |
+
"loss": 0.0686,
|
| 426 |
+
"step": 510
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.16,
|
| 430 |
+
"grad_norm": 3.2105464935302734,
|
| 431 |
+
"learning_rate": 0.00017683741648106903,
|
| 432 |
+
"loss": 0.0347,
|
| 433 |
+
"step": 520
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 1.18,
|
| 437 |
+
"grad_norm": 0.23049691319465637,
|
| 438 |
+
"learning_rate": 0.00017639198218262806,
|
| 439 |
+
"loss": 0.1714,
|
| 440 |
+
"step": 530
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.2,
|
| 444 |
+
"grad_norm": 0.008479436859488487,
|
| 445 |
+
"learning_rate": 0.0001759465478841871,
|
| 446 |
+
"loss": 0.0519,
|
| 447 |
+
"step": 540
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.22,
|
| 451 |
+
"grad_norm": 0.11945914477109909,
|
| 452 |
+
"learning_rate": 0.00017550111358574611,
|
| 453 |
+
"loss": 0.0858,
|
| 454 |
+
"step": 550
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.25,
|
| 458 |
+
"grad_norm": 1.7262502908706665,
|
| 459 |
+
"learning_rate": 0.00017505567928730514,
|
| 460 |
+
"loss": 0.0753,
|
| 461 |
+
"step": 560
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.27,
|
| 465 |
+
"grad_norm": 0.02038051374256611,
|
| 466 |
+
"learning_rate": 0.00017461024498886417,
|
| 467 |
+
"loss": 0.0571,
|
| 468 |
+
"step": 570
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.29,
|
| 472 |
+
"grad_norm": 9.348058700561523,
|
| 473 |
+
"learning_rate": 0.00017416481069042317,
|
| 474 |
+
"loss": 0.128,
|
| 475 |
+
"step": 580
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.31,
|
| 479 |
+
"grad_norm": 0.039238784462213516,
|
| 480 |
+
"learning_rate": 0.0001737193763919822,
|
| 481 |
+
"loss": 0.0551,
|
| 482 |
+
"step": 590
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.34,
|
| 486 |
+
"grad_norm": 5.897979736328125,
|
| 487 |
+
"learning_rate": 0.00017327394209354122,
|
| 488 |
+
"loss": 0.1287,
|
| 489 |
+
"step": 600
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.34,
|
| 493 |
+
"eval_accuracy": 0.9450980392156862,
|
| 494 |
+
"eval_f1": 0.9450980392156862,
|
| 495 |
+
"eval_loss": 0.27001550793647766,
|
| 496 |
+
"eval_precision": 0.9450980392156862,
|
| 497 |
+
"eval_recall": 0.9450980392156862,
|
| 498 |
+
"eval_runtime": 17.21,
|
| 499 |
+
"eval_samples_per_second": 59.268,
|
| 500 |
+
"eval_steps_per_second": 7.438,
|
| 501 |
+
"step": 600
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.36,
|
| 505 |
+
"grad_norm": 5.809938907623291,
|
| 506 |
+
"learning_rate": 0.00017282850779510022,
|
| 507 |
+
"loss": 0.0894,
|
| 508 |
+
"step": 610
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.38,
|
| 512 |
+
"grad_norm": 0.013608383946120739,
|
| 513 |
+
"learning_rate": 0.00017238307349665924,
|
| 514 |
+
"loss": 0.0312,
|
| 515 |
+
"step": 620
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.4,
|
| 519 |
+
"grad_norm": 0.14918509125709534,
|
| 520 |
+
"learning_rate": 0.00017193763919821827,
|
| 521 |
+
"loss": 0.0813,
|
| 522 |
+
"step": 630
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.43,
|
| 526 |
+
"grad_norm": 0.09317727386951447,
|
| 527 |
+
"learning_rate": 0.0001714922048997773,
|
| 528 |
+
"loss": 0.1021,
|
| 529 |
+
"step": 640
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.45,
|
| 533 |
+
"grad_norm": 0.12424630671739578,
|
| 534 |
+
"learning_rate": 0.00017104677060133632,
|
| 535 |
+
"loss": 0.0382,
|
| 536 |
+
"step": 650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.47,
|
| 540 |
+
"grad_norm": 0.004964092746376991,
|
| 541 |
+
"learning_rate": 0.00017060133630289532,
|
| 542 |
+
"loss": 0.0729,
|
| 543 |
+
"step": 660
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.49,
|
| 547 |
+
"grad_norm": 3.553861379623413,
|
| 548 |
+
"learning_rate": 0.00017015590200445435,
|
| 549 |
+
"loss": 0.0475,
|
| 550 |
+
"step": 670
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.51,
|
| 554 |
+
"grad_norm": 0.0814567431807518,
|
| 555 |
+
"learning_rate": 0.00016971046770601338,
|
| 556 |
+
"loss": 0.0424,
|
| 557 |
+
"step": 680
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.54,
|
| 561 |
+
"grad_norm": 0.5184776186943054,
|
| 562 |
+
"learning_rate": 0.00016926503340757238,
|
| 563 |
+
"loss": 0.0182,
|
| 564 |
+
"step": 690
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.56,
|
| 568 |
+
"grad_norm": 0.0049703894183039665,
|
| 569 |
+
"learning_rate": 0.0001688195991091314,
|
| 570 |
+
"loss": 0.0147,
|
| 571 |
+
"step": 700
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.56,
|
| 575 |
+
"eval_accuracy": 0.9205882352941176,
|
| 576 |
+
"eval_f1": 0.9205882352941176,
|
| 577 |
+
"eval_loss": 0.369125634431839,
|
| 578 |
+
"eval_precision": 0.9205882352941176,
|
| 579 |
+
"eval_recall": 0.9205882352941176,
|
| 580 |
+
"eval_runtime": 16.9323,
|
| 581 |
+
"eval_samples_per_second": 60.24,
|
| 582 |
+
"eval_steps_per_second": 7.56,
|
| 583 |
+
"step": 700
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.58,
|
| 587 |
+
"grad_norm": 0.021510396152734756,
|
| 588 |
+
"learning_rate": 0.00016837416481069043,
|
| 589 |
+
"loss": 0.0821,
|
| 590 |
+
"step": 710
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.6,
|
| 594 |
+
"grad_norm": 0.06858960539102554,
|
| 595 |
+
"learning_rate": 0.00016792873051224946,
|
| 596 |
+
"loss": 0.0053,
|
| 597 |
+
"step": 720
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.63,
|
| 601 |
+
"grad_norm": 0.00354547961615026,
|
| 602 |
+
"learning_rate": 0.00016748329621380848,
|
| 603 |
+
"loss": 0.027,
|
| 604 |
+
"step": 730
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.65,
|
| 608 |
+
"grad_norm": 0.0027330678422003984,
|
| 609 |
+
"learning_rate": 0.0001670378619153675,
|
| 610 |
+
"loss": 0.0038,
|
| 611 |
+
"step": 740
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.67,
|
| 615 |
+
"grad_norm": 0.0024623360950499773,
|
| 616 |
+
"learning_rate": 0.0001665924276169265,
|
| 617 |
+
"loss": 0.0435,
|
| 618 |
+
"step": 750
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.69,
|
| 622 |
+
"grad_norm": 0.006295201368629932,
|
| 623 |
+
"learning_rate": 0.00016614699331848553,
|
| 624 |
+
"loss": 0.063,
|
| 625 |
+
"step": 760
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.71,
|
| 629 |
+
"grad_norm": 0.8971105813980103,
|
| 630 |
+
"learning_rate": 0.00016570155902004456,
|
| 631 |
+
"loss": 0.0625,
|
| 632 |
+
"step": 770
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 1.74,
|
| 636 |
+
"grad_norm": 0.004554128274321556,
|
| 637 |
+
"learning_rate": 0.00016525612472160356,
|
| 638 |
+
"loss": 0.004,
|
| 639 |
+
"step": 780
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.76,
|
| 643 |
+
"grad_norm": 1.1439096927642822,
|
| 644 |
+
"learning_rate": 0.0001648106904231626,
|
| 645 |
+
"loss": 0.0959,
|
| 646 |
+
"step": 790
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.78,
|
| 650 |
+
"grad_norm": 0.12275100499391556,
|
| 651 |
+
"learning_rate": 0.0001643652561247216,
|
| 652 |
+
"loss": 0.0416,
|
| 653 |
+
"step": 800
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 1.78,
|
| 657 |
+
"eval_accuracy": 0.9470588235294117,
|
| 658 |
+
"eval_f1": 0.9470588235294117,
|
| 659 |
+
"eval_loss": 0.25350436568260193,
|
| 660 |
+
"eval_precision": 0.9470588235294117,
|
| 661 |
+
"eval_recall": 0.9470588235294117,
|
| 662 |
+
"eval_runtime": 16.7158,
|
| 663 |
+
"eval_samples_per_second": 61.02,
|
| 664 |
+
"eval_steps_per_second": 7.657,
|
| 665 |
+
"step": 800
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.8,
|
| 669 |
+
"grad_norm": 0.0064841569401323795,
|
| 670 |
+
"learning_rate": 0.00016391982182628064,
|
| 671 |
+
"loss": 0.0479,
|
| 672 |
+
"step": 810
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.83,
|
| 676 |
+
"grad_norm": 0.006001554429531097,
|
| 677 |
+
"learning_rate": 0.00016347438752783967,
|
| 678 |
+
"loss": 0.0624,
|
| 679 |
+
"step": 820
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.85,
|
| 683 |
+
"grad_norm": 0.1525709182024002,
|
| 684 |
+
"learning_rate": 0.00016302895322939867,
|
| 685 |
+
"loss": 0.0855,
|
| 686 |
+
"step": 830
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.87,
|
| 690 |
+
"grad_norm": 0.017199277877807617,
|
| 691 |
+
"learning_rate": 0.0001625835189309577,
|
| 692 |
+
"loss": 0.0199,
|
| 693 |
+
"step": 840
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.89,
|
| 697 |
+
"grad_norm": 0.40461423993110657,
|
| 698 |
+
"learning_rate": 0.00016213808463251672,
|
| 699 |
+
"loss": 0.0613,
|
| 700 |
+
"step": 850
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.92,
|
| 704 |
+
"grad_norm": 0.006222237832844257,
|
| 705 |
+
"learning_rate": 0.00016169265033407572,
|
| 706 |
+
"loss": 0.0168,
|
| 707 |
+
"step": 860
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.94,
|
| 711 |
+
"grad_norm": 0.0056493207812309265,
|
| 712 |
+
"learning_rate": 0.00016124721603563474,
|
| 713 |
+
"loss": 0.0247,
|
| 714 |
+
"step": 870
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.96,
|
| 718 |
+
"grad_norm": 0.004911198280751705,
|
| 719 |
+
"learning_rate": 0.00016080178173719377,
|
| 720 |
+
"loss": 0.074,
|
| 721 |
+
"step": 880
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.98,
|
| 725 |
+
"grad_norm": 0.0023081935942173004,
|
| 726 |
+
"learning_rate": 0.0001603563474387528,
|
| 727 |
+
"loss": 0.1029,
|
| 728 |
+
"step": 890
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.0,
|
| 732 |
+
"grad_norm": 0.010029925964772701,
|
| 733 |
+
"learning_rate": 0.00015991091314031182,
|
| 734 |
+
"loss": 0.0211,
|
| 735 |
+
"step": 900
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.0,
|
| 739 |
+
"eval_accuracy": 0.9470588235294117,
|
| 740 |
+
"eval_f1": 0.9470588235294117,
|
| 741 |
+
"eval_loss": 0.25747954845428467,
|
| 742 |
+
"eval_precision": 0.9470588235294117,
|
| 743 |
+
"eval_recall": 0.9470588235294117,
|
| 744 |
+
"eval_runtime": 16.8113,
|
| 745 |
+
"eval_samples_per_second": 60.674,
|
| 746 |
+
"eval_steps_per_second": 7.614,
|
| 747 |
+
"step": 900
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 2.03,
|
| 751 |
+
"grad_norm": 0.003727799979969859,
|
| 752 |
+
"learning_rate": 0.00015946547884187085,
|
| 753 |
+
"loss": 0.0126,
|
| 754 |
+
"step": 910
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 2.05,
|
| 758 |
+
"grad_norm": 0.7938788533210754,
|
| 759 |
+
"learning_rate": 0.00015902004454342985,
|
| 760 |
+
"loss": 0.0702,
|
| 761 |
+
"step": 920
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 2.07,
|
| 765 |
+
"grad_norm": 0.48882561922073364,
|
| 766 |
+
"learning_rate": 0.00015857461024498888,
|
| 767 |
+
"loss": 0.0513,
|
| 768 |
+
"step": 930
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 2.09,
|
| 772 |
+
"grad_norm": 10.071187973022461,
|
| 773 |
+
"learning_rate": 0.0001581291759465479,
|
| 774 |
+
"loss": 0.109,
|
| 775 |
+
"step": 940
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 2.12,
|
| 779 |
+
"grad_norm": 0.009675947949290276,
|
| 780 |
+
"learning_rate": 0.0001576837416481069,
|
| 781 |
+
"loss": 0.0034,
|
| 782 |
+
"step": 950
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.14,
|
| 786 |
+
"grad_norm": 0.004574100486934185,
|
| 787 |
+
"learning_rate": 0.00015723830734966593,
|
| 788 |
+
"loss": 0.0018,
|
| 789 |
+
"step": 960
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.16,
|
| 793 |
+
"grad_norm": 0.0033382533583790064,
|
| 794 |
+
"learning_rate": 0.00015679287305122495,
|
| 795 |
+
"loss": 0.0719,
|
| 796 |
+
"step": 970
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.18,
|
| 800 |
+
"grad_norm": 4.464893341064453,
|
| 801 |
+
"learning_rate": 0.00015634743875278398,
|
| 802 |
+
"loss": 0.0099,
|
| 803 |
+
"step": 980
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 2.2,
|
| 807 |
+
"grad_norm": 0.004051242955029011,
|
| 808 |
+
"learning_rate": 0.000155902004454343,
|
| 809 |
+
"loss": 0.0846,
|
| 810 |
+
"step": 990
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 2.23,
|
| 814 |
+
"grad_norm": 0.047729793936014175,
|
| 815 |
+
"learning_rate": 0.000155456570155902,
|
| 816 |
+
"loss": 0.088,
|
| 817 |
+
"step": 1000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 2.23,
|
| 821 |
+
"eval_accuracy": 0.9529411764705882,
|
| 822 |
+
"eval_f1": 0.9529411764705882,
|
| 823 |
+
"eval_loss": 0.19075074791908264,
|
| 824 |
+
"eval_precision": 0.9529411764705882,
|
| 825 |
+
"eval_recall": 0.9529411764705882,
|
| 826 |
+
"eval_runtime": 16.8471,
|
| 827 |
+
"eval_samples_per_second": 60.544,
|
| 828 |
+
"eval_steps_per_second": 7.598,
|
| 829 |
+
"step": 1000
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 2.25,
|
| 833 |
+
"grad_norm": 0.008470825850963593,
|
| 834 |
+
"learning_rate": 0.00015501113585746103,
|
| 835 |
+
"loss": 0.0022,
|
| 836 |
+
"step": 1010
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.27,
|
| 840 |
+
"grad_norm": 2.529259204864502,
|
| 841 |
+
"learning_rate": 0.00015456570155902006,
|
| 842 |
+
"loss": 0.0917,
|
| 843 |
+
"step": 1020
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.29,
|
| 847 |
+
"grad_norm": 0.3583894670009613,
|
| 848 |
+
"learning_rate": 0.00015412026726057906,
|
| 849 |
+
"loss": 0.0793,
|
| 850 |
+
"step": 1030
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 2.32,
|
| 854 |
+
"grad_norm": 0.039421286433935165,
|
| 855 |
+
"learning_rate": 0.00015367483296213809,
|
| 856 |
+
"loss": 0.0742,
|
| 857 |
+
"step": 1040
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.34,
|
| 861 |
+
"grad_norm": 0.2551974654197693,
|
| 862 |
+
"learning_rate": 0.0001532293986636971,
|
| 863 |
+
"loss": 0.0029,
|
| 864 |
+
"step": 1050
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.36,
|
| 868 |
+
"grad_norm": 0.3370533883571625,
|
| 869 |
+
"learning_rate": 0.00015278396436525614,
|
| 870 |
+
"loss": 0.0563,
|
| 871 |
+
"step": 1060
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.38,
|
| 875 |
+
"grad_norm": 1.642697811126709,
|
| 876 |
+
"learning_rate": 0.00015233853006681517,
|
| 877 |
+
"loss": 0.0362,
|
| 878 |
+
"step": 1070
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.41,
|
| 882 |
+
"grad_norm": 12.125362396240234,
|
| 883 |
+
"learning_rate": 0.0001518930957683742,
|
| 884 |
+
"loss": 0.0949,
|
| 885 |
+
"step": 1080
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.43,
|
| 889 |
+
"grad_norm": 0.0022143302485346794,
|
| 890 |
+
"learning_rate": 0.0001514476614699332,
|
| 891 |
+
"loss": 0.0219,
|
| 892 |
+
"step": 1090
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 2.45,
|
| 896 |
+
"grad_norm": 0.0026613217778503895,
|
| 897 |
+
"learning_rate": 0.00015100222717149222,
|
| 898 |
+
"loss": 0.1849,
|
| 899 |
+
"step": 1100
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 2.45,
|
| 903 |
+
"eval_accuracy": 0.9529411764705882,
|
| 904 |
+
"eval_f1": 0.9529411764705882,
|
| 905 |
+
"eval_loss": 0.2200697511434555,
|
| 906 |
+
"eval_precision": 0.9529411764705882,
|
| 907 |
+
"eval_recall": 0.9529411764705882,
|
| 908 |
+
"eval_runtime": 16.9063,
|
| 909 |
+
"eval_samples_per_second": 60.332,
|
| 910 |
+
"eval_steps_per_second": 7.571,
|
| 911 |
+
"step": 1100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.47,
|
| 915 |
+
"grad_norm": 0.5232903957366943,
|
| 916 |
+
"learning_rate": 0.00015055679287305122,
|
| 917 |
+
"loss": 0.0013,
|
| 918 |
+
"step": 1110
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.49,
|
| 922 |
+
"grad_norm": 1.5729717016220093,
|
| 923 |
+
"learning_rate": 0.00015011135857461024,
|
| 924 |
+
"loss": 0.0433,
|
| 925 |
+
"step": 1120
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.52,
|
| 929 |
+
"grad_norm": 0.005640827585011721,
|
| 930 |
+
"learning_rate": 0.00014966592427616927,
|
| 931 |
+
"loss": 0.0007,
|
| 932 |
+
"step": 1130
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.54,
|
| 936 |
+
"grad_norm": 7.901826858520508,
|
| 937 |
+
"learning_rate": 0.0001492204899777283,
|
| 938 |
+
"loss": 0.1131,
|
| 939 |
+
"step": 1140
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.56,
|
| 943 |
+
"grad_norm": 0.4438769221305847,
|
| 944 |
+
"learning_rate": 0.00014877505567928732,
|
| 945 |
+
"loss": 0.0557,
|
| 946 |
+
"step": 1150
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.58,
|
| 950 |
+
"grad_norm": 0.0032986474689096212,
|
| 951 |
+
"learning_rate": 0.00014832962138084635,
|
| 952 |
+
"loss": 0.001,
|
| 953 |
+
"step": 1160
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.61,
|
| 957 |
+
"grad_norm": 0.09250050783157349,
|
| 958 |
+
"learning_rate": 0.00014788418708240535,
|
| 959 |
+
"loss": 0.02,
|
| 960 |
+
"step": 1170
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.63,
|
| 964 |
+
"grad_norm": 0.005981163587421179,
|
| 965 |
+
"learning_rate": 0.00014743875278396438,
|
| 966 |
+
"loss": 0.0147,
|
| 967 |
+
"step": 1180
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.65,
|
| 971 |
+
"grad_norm": 0.02270863950252533,
|
| 972 |
+
"learning_rate": 0.0001469933184855234,
|
| 973 |
+
"loss": 0.0246,
|
| 974 |
+
"step": 1190
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.67,
|
| 978 |
+
"grad_norm": 0.014892498031258583,
|
| 979 |
+
"learning_rate": 0.0001465478841870824,
|
| 980 |
+
"loss": 0.0009,
|
| 981 |
+
"step": 1200
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.67,
|
| 985 |
+
"eval_accuracy": 0.9549019607843138,
|
| 986 |
+
"eval_f1": 0.9549019607843138,
|
| 987 |
+
"eval_loss": 0.22289611399173737,
|
| 988 |
+
"eval_precision": 0.9549019607843138,
|
| 989 |
+
"eval_recall": 0.9549019607843138,
|
| 990 |
+
"eval_runtime": 16.9051,
|
| 991 |
+
"eval_samples_per_second": 60.337,
|
| 992 |
+
"eval_steps_per_second": 7.572,
|
| 993 |
+
"step": 1200
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.69,
|
| 997 |
+
"grad_norm": 10.428900718688965,
|
| 998 |
+
"learning_rate": 0.00014610244988864143,
|
| 999 |
+
"loss": 0.0115,
|
| 1000 |
+
"step": 1210
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.72,
|
| 1004 |
+
"grad_norm": 0.0370117723941803,
|
| 1005 |
+
"learning_rate": 0.00014565701559020045,
|
| 1006 |
+
"loss": 0.0525,
|
| 1007 |
+
"step": 1220
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.74,
|
| 1011 |
+
"grad_norm": 0.0029497628565877676,
|
| 1012 |
+
"learning_rate": 0.00014521158129175948,
|
| 1013 |
+
"loss": 0.0009,
|
| 1014 |
+
"step": 1230
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.76,
|
| 1018 |
+
"grad_norm": 0.8202322125434875,
|
| 1019 |
+
"learning_rate": 0.0001447661469933185,
|
| 1020 |
+
"loss": 0.051,
|
| 1021 |
+
"step": 1240
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.78,
|
| 1025 |
+
"grad_norm": 0.012352533638477325,
|
| 1026 |
+
"learning_rate": 0.0001443207126948775,
|
| 1027 |
+
"loss": 0.0034,
|
| 1028 |
+
"step": 1250
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.81,
|
| 1032 |
+
"grad_norm": 0.5958288908004761,
|
| 1033 |
+
"learning_rate": 0.00014387527839643653,
|
| 1034 |
+
"loss": 0.0016,
|
| 1035 |
+
"step": 1260
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.83,
|
| 1039 |
+
"grad_norm": 0.01864822395145893,
|
| 1040 |
+
"learning_rate": 0.00014342984409799556,
|
| 1041 |
+
"loss": 0.0113,
|
| 1042 |
+
"step": 1270
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 2.85,
|
| 1046 |
+
"grad_norm": 0.053810037672519684,
|
| 1047 |
+
"learning_rate": 0.00014298440979955456,
|
| 1048 |
+
"loss": 0.0044,
|
| 1049 |
+
"step": 1280
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.87,
|
| 1053 |
+
"grad_norm": 0.07594209164381027,
|
| 1054 |
+
"learning_rate": 0.00014253897550111359,
|
| 1055 |
+
"loss": 0.0022,
|
| 1056 |
+
"step": 1290
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.9,
|
| 1060 |
+
"grad_norm": 0.003192998468875885,
|
| 1061 |
+
"learning_rate": 0.0001420935412026726,
|
| 1062 |
+
"loss": 0.0599,
|
| 1063 |
+
"step": 1300
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 2.9,
|
| 1067 |
+
"eval_accuracy": 0.9607843137254902,
|
| 1068 |
+
"eval_f1": 0.9607843137254902,
|
| 1069 |
+
"eval_loss": 0.1780730038881302,
|
| 1070 |
+
"eval_precision": 0.9607843137254902,
|
| 1071 |
+
"eval_recall": 0.9607843137254902,
|
| 1072 |
+
"eval_runtime": 16.8581,
|
| 1073 |
+
"eval_samples_per_second": 60.505,
|
| 1074 |
+
"eval_steps_per_second": 7.593,
|
| 1075 |
+
"step": 1300
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 2.92,
|
| 1079 |
+
"grad_norm": 0.0019150603329762816,
|
| 1080 |
+
"learning_rate": 0.00014164810690423164,
|
| 1081 |
+
"loss": 0.0136,
|
| 1082 |
+
"step": 1310
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 2.94,
|
| 1086 |
+
"grad_norm": 0.0017021102830767632,
|
| 1087 |
+
"learning_rate": 0.00014120267260579067,
|
| 1088 |
+
"loss": 0.0004,
|
| 1089 |
+
"step": 1320
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.96,
|
| 1093 |
+
"grad_norm": 0.002855106256902218,
|
| 1094 |
+
"learning_rate": 0.0001407572383073497,
|
| 1095 |
+
"loss": 0.0028,
|
| 1096 |
+
"step": 1330
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 2.98,
|
| 1100 |
+
"grad_norm": 0.0011817626655101776,
|
| 1101 |
+
"learning_rate": 0.0001403118040089087,
|
| 1102 |
+
"loss": 0.0146,
|
| 1103 |
+
"step": 1340
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 3.01,
|
| 1107 |
+
"grad_norm": 11.652885437011719,
|
| 1108 |
+
"learning_rate": 0.00013986636971046772,
|
| 1109 |
+
"loss": 0.0155,
|
| 1110 |
+
"step": 1350
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 3.03,
|
| 1114 |
+
"grad_norm": 0.002146989107131958,
|
| 1115 |
+
"learning_rate": 0.00013942093541202674,
|
| 1116 |
+
"loss": 0.007,
|
| 1117 |
+
"step": 1360
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 3.05,
|
| 1121 |
+
"grad_norm": 0.0012873058440163732,
|
| 1122 |
+
"learning_rate": 0.00013897550111358574,
|
| 1123 |
+
"loss": 0.0005,
|
| 1124 |
+
"step": 1370
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 3.07,
|
| 1128 |
+
"grad_norm": 0.001976664876565337,
|
| 1129 |
+
"learning_rate": 0.00013853006681514477,
|
| 1130 |
+
"loss": 0.0067,
|
| 1131 |
+
"step": 1380
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 3.1,
|
| 1135 |
+
"grad_norm": 0.004611461888998747,
|
| 1136 |
+
"learning_rate": 0.00013808463251670377,
|
| 1137 |
+
"loss": 0.068,
|
| 1138 |
+
"step": 1390
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 3.12,
|
| 1142 |
+
"grad_norm": 0.05266120657324791,
|
| 1143 |
+
"learning_rate": 0.00013763919821826282,
|
| 1144 |
+
"loss": 0.0004,
|
| 1145 |
+
"step": 1400
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 3.12,
|
| 1149 |
+
"eval_accuracy": 0.9666666666666667,
|
| 1150 |
+
"eval_f1": 0.9666666666666667,
|
| 1151 |
+
"eval_loss": 0.1750936210155487,
|
| 1152 |
+
"eval_precision": 0.9666666666666667,
|
| 1153 |
+
"eval_recall": 0.9666666666666667,
|
| 1154 |
+
"eval_runtime": 16.8283,
|
| 1155 |
+
"eval_samples_per_second": 60.612,
|
| 1156 |
+
"eval_steps_per_second": 7.606,
|
| 1157 |
+
"step": 1400
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 3.14,
|
| 1161 |
+
"grad_norm": 0.005359290167689323,
|
| 1162 |
+
"learning_rate": 0.00013719376391982185,
|
| 1163 |
+
"loss": 0.038,
|
| 1164 |
+
"step": 1410
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 3.16,
|
| 1168 |
+
"grad_norm": 0.0019557911437004805,
|
| 1169 |
+
"learning_rate": 0.00013674832962138085,
|
| 1170 |
+
"loss": 0.0006,
|
| 1171 |
+
"step": 1420
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 3.18,
|
| 1175 |
+
"grad_norm": 0.002684570848941803,
|
| 1176 |
+
"learning_rate": 0.00013630289532293988,
|
| 1177 |
+
"loss": 0.0179,
|
| 1178 |
+
"step": 1430
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 3.21,
|
| 1182 |
+
"grad_norm": 2.459765911102295,
|
| 1183 |
+
"learning_rate": 0.0001358574610244989,
|
| 1184 |
+
"loss": 0.0164,
|
| 1185 |
+
"step": 1440
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 3.23,
|
| 1189 |
+
"grad_norm": 0.010600595735013485,
|
| 1190 |
+
"learning_rate": 0.0001354120267260579,
|
| 1191 |
+
"loss": 0.0008,
|
| 1192 |
+
"step": 1450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 3.25,
|
| 1196 |
+
"grad_norm": 0.12665680050849915,
|
| 1197 |
+
"learning_rate": 0.00013496659242761693,
|
| 1198 |
+
"loss": 0.1161,
|
| 1199 |
+
"step": 1460
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 3.27,
|
| 1203 |
+
"grad_norm": 0.0013257049722597003,
|
| 1204 |
+
"learning_rate": 0.00013452115812917595,
|
| 1205 |
+
"loss": 0.0718,
|
| 1206 |
+
"step": 1470
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 3.3,
|
| 1210 |
+
"grad_norm": 2.888887882232666,
|
| 1211 |
+
"learning_rate": 0.00013407572383073498,
|
| 1212 |
+
"loss": 0.0028,
|
| 1213 |
+
"step": 1480
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 3.32,
|
| 1217 |
+
"grad_norm": 0.0022232867777347565,
|
| 1218 |
+
"learning_rate": 0.000133630289532294,
|
| 1219 |
+
"loss": 0.0009,
|
| 1220 |
+
"step": 1490
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 3.34,
|
| 1224 |
+
"grad_norm": 0.0026971769984811544,
|
| 1225 |
+
"learning_rate": 0.00013318485523385303,
|
| 1226 |
+
"loss": 0.0004,
|
| 1227 |
+
"step": 1500
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 3.34,
|
| 1231 |
+
"eval_accuracy": 0.9686274509803922,
|
| 1232 |
+
"eval_f1": 0.9686274509803922,
|
| 1233 |
+
"eval_loss": 0.1684277504682541,
|
| 1234 |
+
"eval_precision": 0.9686274509803922,
|
| 1235 |
+
"eval_recall": 0.9686274509803922,
|
| 1236 |
+
"eval_runtime": 16.8277,
|
| 1237 |
+
"eval_samples_per_second": 60.614,
|
| 1238 |
+
"eval_steps_per_second": 7.606,
|
| 1239 |
+
"step": 1500
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 3.36,
|
| 1243 |
+
"grad_norm": 0.06871479004621506,
|
| 1244 |
+
"learning_rate": 0.00013273942093541203,
|
| 1245 |
+
"loss": 0.0004,
|
| 1246 |
+
"step": 1510
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 3.39,
|
| 1250 |
+
"grad_norm": 0.010680126026272774,
|
| 1251 |
+
"learning_rate": 0.00013229398663697106,
|
| 1252 |
+
"loss": 0.0623,
|
| 1253 |
+
"step": 1520
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 3.41,
|
| 1257 |
+
"grad_norm": 0.0024642229545861483,
|
| 1258 |
+
"learning_rate": 0.00013184855233853006,
|
| 1259 |
+
"loss": 0.0005,
|
| 1260 |
+
"step": 1530
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 3.43,
|
| 1264 |
+
"grad_norm": 0.0014489213936030865,
|
| 1265 |
+
"learning_rate": 0.00013140311804008909,
|
| 1266 |
+
"loss": 0.0004,
|
| 1267 |
+
"step": 1540
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 3.45,
|
| 1271 |
+
"grad_norm": 0.004346741829067469,
|
| 1272 |
+
"learning_rate": 0.0001309576837416481,
|
| 1273 |
+
"loss": 0.0025,
|
| 1274 |
+
"step": 1550
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 3.47,
|
| 1278 |
+
"grad_norm": 0.006028163246810436,
|
| 1279 |
+
"learning_rate": 0.0001305122494432071,
|
| 1280 |
+
"loss": 0.0546,
|
| 1281 |
+
"step": 1560
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 3.5,
|
| 1285 |
+
"grad_norm": 0.0034053712151944637,
|
| 1286 |
+
"learning_rate": 0.00013006681514476616,
|
| 1287 |
+
"loss": 0.028,
|
| 1288 |
+
"step": 1570
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 3.52,
|
| 1292 |
+
"grad_norm": 0.004035326186567545,
|
| 1293 |
+
"learning_rate": 0.0001296213808463252,
|
| 1294 |
+
"loss": 0.0042,
|
| 1295 |
+
"step": 1580
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 3.54,
|
| 1299 |
+
"grad_norm": 0.0025597705971449614,
|
| 1300 |
+
"learning_rate": 0.0001291759465478842,
|
| 1301 |
+
"loss": 0.0201,
|
| 1302 |
+
"step": 1590
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 3.56,
|
| 1306 |
+
"grad_norm": 0.0010411799885332584,
|
| 1307 |
+
"learning_rate": 0.00012873051224944322,
|
| 1308 |
+
"loss": 0.0352,
|
| 1309 |
+
"step": 1600
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 3.56,
|
| 1313 |
+
"eval_accuracy": 0.9754901960784313,
|
| 1314 |
+
"eval_f1": 0.9754901960784313,
|
| 1315 |
+
"eval_loss": 0.15021921694278717,
|
| 1316 |
+
"eval_precision": 0.9754901960784313,
|
| 1317 |
+
"eval_recall": 0.9754901960784313,
|
| 1318 |
+
"eval_runtime": 16.7579,
|
| 1319 |
+
"eval_samples_per_second": 60.867,
|
| 1320 |
+
"eval_steps_per_second": 7.638,
|
| 1321 |
+
"step": 1600
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 3.59,
|
| 1325 |
+
"grad_norm": 0.04698515310883522,
|
| 1326 |
+
"learning_rate": 0.00012828507795100224,
|
| 1327 |
+
"loss": 0.0036,
|
| 1328 |
+
"step": 1610
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 3.61,
|
| 1332 |
+
"grad_norm": 0.013074472546577454,
|
| 1333 |
+
"learning_rate": 0.00012783964365256124,
|
| 1334 |
+
"loss": 0.0626,
|
| 1335 |
+
"step": 1620
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 3.63,
|
| 1339 |
+
"grad_norm": 0.002302026841789484,
|
| 1340 |
+
"learning_rate": 0.00012739420935412027,
|
| 1341 |
+
"loss": 0.0004,
|
| 1342 |
+
"step": 1630
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 3.65,
|
| 1346 |
+
"grad_norm": 0.012451753951609135,
|
| 1347 |
+
"learning_rate": 0.0001269487750556793,
|
| 1348 |
+
"loss": 0.0418,
|
| 1349 |
+
"step": 1640
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 3.67,
|
| 1353 |
+
"grad_norm": 0.010159431956708431,
|
| 1354 |
+
"learning_rate": 0.00012650334075723832,
|
| 1355 |
+
"loss": 0.0008,
|
| 1356 |
+
"step": 1650
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 3.7,
|
| 1360 |
+
"grad_norm": 0.20833130180835724,
|
| 1361 |
+
"learning_rate": 0.00012605790645879735,
|
| 1362 |
+
"loss": 0.0006,
|
| 1363 |
+
"step": 1660
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 3.72,
|
| 1367 |
+
"grad_norm": 0.0008345023961737752,
|
| 1368 |
+
"learning_rate": 0.00012561247216035635,
|
| 1369 |
+
"loss": 0.0013,
|
| 1370 |
+
"step": 1670
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 3.74,
|
| 1374 |
+
"grad_norm": 0.0027376762591302395,
|
| 1375 |
+
"learning_rate": 0.00012516703786191537,
|
| 1376 |
+
"loss": 0.0003,
|
| 1377 |
+
"step": 1680
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 3.76,
|
| 1381 |
+
"grad_norm": 0.0012037215055897832,
|
| 1382 |
+
"learning_rate": 0.0001247216035634744,
|
| 1383 |
+
"loss": 0.0003,
|
| 1384 |
+
"step": 1690
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 3.79,
|
| 1388 |
+
"grad_norm": 0.02287732996046543,
|
| 1389 |
+
"learning_rate": 0.0001242761692650334,
|
| 1390 |
+
"loss": 0.0003,
|
| 1391 |
+
"step": 1700
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 3.79,
|
| 1395 |
+
"eval_accuracy": 0.9745098039215686,
|
| 1396 |
+
"eval_f1": 0.9745098039215686,
|
| 1397 |
+
"eval_loss": 0.15970657765865326,
|
| 1398 |
+
"eval_precision": 0.9745098039215686,
|
| 1399 |
+
"eval_recall": 0.9745098039215686,
|
| 1400 |
+
"eval_runtime": 16.8382,
|
| 1401 |
+
"eval_samples_per_second": 60.576,
|
| 1402 |
+
"eval_steps_per_second": 7.602,
|
| 1403 |
+
"step": 1700
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.81,
|
| 1407 |
+
"grad_norm": 0.0013968138955533504,
|
| 1408 |
+
"learning_rate": 0.00012383073496659243,
|
| 1409 |
+
"loss": 0.0005,
|
| 1410 |
+
"step": 1710
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.83,
|
| 1414 |
+
"grad_norm": 0.003733535995706916,
|
| 1415 |
+
"learning_rate": 0.00012338530066815145,
|
| 1416 |
+
"loss": 0.0003,
|
| 1417 |
+
"step": 1720
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.85,
|
| 1421 |
+
"grad_norm": 0.976151168346405,
|
| 1422 |
+
"learning_rate": 0.00012293986636971045,
|
| 1423 |
+
"loss": 0.001,
|
| 1424 |
+
"step": 1730
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.88,
|
| 1428 |
+
"grad_norm": 0.003409826662391424,
|
| 1429 |
+
"learning_rate": 0.0001224944320712695,
|
| 1430 |
+
"loss": 0.0003,
|
| 1431 |
+
"step": 1740
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.9,
|
| 1435 |
+
"grad_norm": 0.004897921811789274,
|
| 1436 |
+
"learning_rate": 0.00012204899777282852,
|
| 1437 |
+
"loss": 0.0013,
|
| 1438 |
+
"step": 1750
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.92,
|
| 1442 |
+
"grad_norm": 0.0010848743841052055,
|
| 1443 |
+
"learning_rate": 0.00012160356347438753,
|
| 1444 |
+
"loss": 0.0003,
|
| 1445 |
+
"step": 1760
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.94,
|
| 1449 |
+
"grad_norm": 0.0013221738627180457,
|
| 1450 |
+
"learning_rate": 0.00012115812917594656,
|
| 1451 |
+
"loss": 0.0002,
|
| 1452 |
+
"step": 1770
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 3.96,
|
| 1456 |
+
"grad_norm": 0.09072667360305786,
|
| 1457 |
+
"learning_rate": 0.00012071269487750559,
|
| 1458 |
+
"loss": 0.0029,
|
| 1459 |
+
"step": 1780
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.99,
|
| 1463 |
+
"grad_norm": 0.00137105374597013,
|
| 1464 |
+
"learning_rate": 0.00012026726057906458,
|
| 1465 |
+
"loss": 0.0007,
|
| 1466 |
+
"step": 1790
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 4.01,
|
| 1470 |
+
"grad_norm": 0.004980257712304592,
|
| 1471 |
+
"learning_rate": 0.00011982182628062361,
|
| 1472 |
+
"loss": 0.0003,
|
| 1473 |
+
"step": 1800
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 4.01,
|
| 1477 |
+
"eval_accuracy": 0.9558823529411765,
|
| 1478 |
+
"eval_f1": 0.9558823529411765,
|
| 1479 |
+
"eval_loss": 0.25734347105026245,
|
| 1480 |
+
"eval_precision": 0.9558823529411765,
|
| 1481 |
+
"eval_recall": 0.9558823529411765,
|
| 1482 |
+
"eval_runtime": 16.754,
|
| 1483 |
+
"eval_samples_per_second": 60.881,
|
| 1484 |
+
"eval_steps_per_second": 7.64,
|
| 1485 |
+
"step": 1800
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 4.03,
|
| 1489 |
+
"grad_norm": 0.0013974602334201336,
|
| 1490 |
+
"learning_rate": 0.00011937639198218265,
|
| 1491 |
+
"loss": 0.0015,
|
| 1492 |
+
"step": 1810
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 4.05,
|
| 1496 |
+
"grad_norm": 0.0027338312938809395,
|
| 1497 |
+
"learning_rate": 0.00011893095768374165,
|
| 1498 |
+
"loss": 0.0028,
|
| 1499 |
+
"step": 1820
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 4.08,
|
| 1503 |
+
"grad_norm": 0.0018153024138882756,
|
| 1504 |
+
"learning_rate": 0.00011848552338530068,
|
| 1505 |
+
"loss": 0.0003,
|
| 1506 |
+
"step": 1830
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 4.1,
|
| 1510 |
+
"grad_norm": 0.0008053297642618418,
|
| 1511 |
+
"learning_rate": 0.00011804008908685969,
|
| 1512 |
+
"loss": 0.0145,
|
| 1513 |
+
"step": 1840
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 4.12,
|
| 1517 |
+
"grad_norm": 0.0016097394982352853,
|
| 1518 |
+
"learning_rate": 0.00011759465478841872,
|
| 1519 |
+
"loss": 0.0003,
|
| 1520 |
+
"step": 1850
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 4.14,
|
| 1524 |
+
"grad_norm": 0.0011555146193131804,
|
| 1525 |
+
"learning_rate": 0.00011714922048997774,
|
| 1526 |
+
"loss": 0.0007,
|
| 1527 |
+
"step": 1860
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 4.16,
|
| 1531 |
+
"grad_norm": 0.0013847779482603073,
|
| 1532 |
+
"learning_rate": 0.00011670378619153674,
|
| 1533 |
+
"loss": 0.0005,
|
| 1534 |
+
"step": 1870
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 4.19,
|
| 1538 |
+
"grad_norm": 0.026971347630023956,
|
| 1539 |
+
"learning_rate": 0.00011625835189309577,
|
| 1540 |
+
"loss": 0.04,
|
| 1541 |
+
"step": 1880
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 4.21,
|
| 1545 |
+
"grad_norm": 0.004224107600748539,
|
| 1546 |
+
"learning_rate": 0.0001158129175946548,
|
| 1547 |
+
"loss": 0.0004,
|
| 1548 |
+
"step": 1890
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 4.23,
|
| 1552 |
+
"grad_norm": 0.008265385404229164,
|
| 1553 |
+
"learning_rate": 0.00011536748329621381,
|
| 1554 |
+
"loss": 0.0005,
|
| 1555 |
+
"step": 1900
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 4.23,
|
| 1559 |
+
"eval_accuracy": 0.9666666666666667,
|
| 1560 |
+
"eval_f1": 0.9666666666666667,
|
| 1561 |
+
"eval_loss": 0.19066497683525085,
|
| 1562 |
+
"eval_precision": 0.9666666666666667,
|
| 1563 |
+
"eval_recall": 0.9666666666666667,
|
| 1564 |
+
"eval_runtime": 16.876,
|
| 1565 |
+
"eval_samples_per_second": 60.441,
|
| 1566 |
+
"eval_steps_per_second": 7.585,
|
| 1567 |
+
"step": 1900
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 4.25,
|
| 1571 |
+
"grad_norm": 0.0009556623990647495,
|
| 1572 |
+
"learning_rate": 0.00011492204899777283,
|
| 1573 |
+
"loss": 0.0103,
|
| 1574 |
+
"step": 1910
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 4.28,
|
| 1578 |
+
"grad_norm": 0.0008651684038341045,
|
| 1579 |
+
"learning_rate": 0.00011447661469933186,
|
| 1580 |
+
"loss": 0.0003,
|
| 1581 |
+
"step": 1920
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 4.3,
|
| 1585 |
+
"grad_norm": 0.0021915507968515158,
|
| 1586 |
+
"learning_rate": 0.00011403118040089087,
|
| 1587 |
+
"loss": 0.0017,
|
| 1588 |
+
"step": 1930
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 4.32,
|
| 1592 |
+
"grad_norm": 0.993601381778717,
|
| 1593 |
+
"learning_rate": 0.0001135857461024499,
|
| 1594 |
+
"loss": 0.0016,
|
| 1595 |
+
"step": 1940
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 4.34,
|
| 1599 |
+
"grad_norm": 0.012279433198273182,
|
| 1600 |
+
"learning_rate": 0.00011314031180400893,
|
| 1601 |
+
"loss": 0.0004,
|
| 1602 |
+
"step": 1950
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 4.37,
|
| 1606 |
+
"grad_norm": 0.0192144475877285,
|
| 1607 |
+
"learning_rate": 0.00011269487750556793,
|
| 1608 |
+
"loss": 0.0211,
|
| 1609 |
+
"step": 1960
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 4.39,
|
| 1613 |
+
"grad_norm": 0.002850558841601014,
|
| 1614 |
+
"learning_rate": 0.00011224944320712695,
|
| 1615 |
+
"loss": 0.0011,
|
| 1616 |
+
"step": 1970
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 4.41,
|
| 1620 |
+
"grad_norm": 0.016757028177380562,
|
| 1621 |
+
"learning_rate": 0.00011180400890868597,
|
| 1622 |
+
"loss": 0.0004,
|
| 1623 |
+
"step": 1980
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 4.43,
|
| 1627 |
+
"grad_norm": 0.009729539044201374,
|
| 1628 |
+
"learning_rate": 0.00011135857461024499,
|
| 1629 |
+
"loss": 0.0005,
|
| 1630 |
+
"step": 1990
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 4.45,
|
| 1634 |
+
"grad_norm": 0.0012516066199168563,
|
| 1635 |
+
"learning_rate": 0.00011091314031180402,
|
| 1636 |
+
"loss": 0.0741,
|
| 1637 |
+
"step": 2000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 4.45,
|
| 1641 |
+
"eval_accuracy": 0.9637254901960784,
|
| 1642 |
+
"eval_f1": 0.9637254901960784,
|
| 1643 |
+
"eval_loss": 0.20377103984355927,
|
| 1644 |
+
"eval_precision": 0.9637254901960784,
|
| 1645 |
+
"eval_recall": 0.9637254901960784,
|
| 1646 |
+
"eval_runtime": 16.9265,
|
| 1647 |
+
"eval_samples_per_second": 60.26,
|
| 1648 |
+
"eval_steps_per_second": 7.562,
|
| 1649 |
+
"step": 2000
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 4.48,
|
| 1653 |
+
"grad_norm": 0.03771669417619705,
|
| 1654 |
+
"learning_rate": 0.00011046770601336303,
|
| 1655 |
+
"loss": 0.0005,
|
| 1656 |
+
"step": 2010
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 4.5,
|
| 1660 |
+
"grad_norm": 0.0029582425486296415,
|
| 1661 |
+
"learning_rate": 0.00011002227171492206,
|
| 1662 |
+
"loss": 0.044,
|
| 1663 |
+
"step": 2020
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 4.52,
|
| 1667 |
+
"grad_norm": 0.0017603106098249555,
|
| 1668 |
+
"learning_rate": 0.00010957683741648108,
|
| 1669 |
+
"loss": 0.0002,
|
| 1670 |
+
"step": 2030
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 4.54,
|
| 1674 |
+
"grad_norm": 0.006093455944210291,
|
| 1675 |
+
"learning_rate": 0.00010913140311804008,
|
| 1676 |
+
"loss": 0.0004,
|
| 1677 |
+
"step": 2040
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 4.57,
|
| 1681 |
+
"grad_norm": 0.024271611124277115,
|
| 1682 |
+
"learning_rate": 0.00010868596881959911,
|
| 1683 |
+
"loss": 0.0004,
|
| 1684 |
+
"step": 2050
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 4.59,
|
| 1688 |
+
"grad_norm": 0.0059431749396026134,
|
| 1689 |
+
"learning_rate": 0.00010824053452115814,
|
| 1690 |
+
"loss": 0.0171,
|
| 1691 |
+
"step": 2060
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 4.61,
|
| 1695 |
+
"grad_norm": 0.001350950333289802,
|
| 1696 |
+
"learning_rate": 0.00010779510022271715,
|
| 1697 |
+
"loss": 0.0169,
|
| 1698 |
+
"step": 2070
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 4.63,
|
| 1702 |
+
"grad_norm": 0.11208397895097733,
|
| 1703 |
+
"learning_rate": 0.00010734966592427618,
|
| 1704 |
+
"loss": 0.0009,
|
| 1705 |
+
"step": 2080
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 4.65,
|
| 1709 |
+
"grad_norm": 0.000902643718291074,
|
| 1710 |
+
"learning_rate": 0.0001069042316258352,
|
| 1711 |
+
"loss": 0.0002,
|
| 1712 |
+
"step": 2090
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 4.68,
|
| 1716 |
+
"grad_norm": 0.0008040807442739606,
|
| 1717 |
+
"learning_rate": 0.00010645879732739422,
|
| 1718 |
+
"loss": 0.0025,
|
| 1719 |
+
"step": 2100
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 4.68,
|
| 1723 |
+
"eval_accuracy": 0.9647058823529412,
|
| 1724 |
+
"eval_f1": 0.9647058823529412,
|
| 1725 |
+
"eval_loss": 0.1929028332233429,
|
| 1726 |
+
"eval_precision": 0.9647058823529412,
|
| 1727 |
+
"eval_recall": 0.9647058823529412,
|
| 1728 |
+
"eval_runtime": 17.2457,
|
| 1729 |
+
"eval_samples_per_second": 59.145,
|
| 1730 |
+
"eval_steps_per_second": 7.422,
|
| 1731 |
+
"step": 2100
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 4.7,
|
| 1735 |
+
"grad_norm": 0.001127121620811522,
|
| 1736 |
+
"learning_rate": 0.00010601336302895324,
|
| 1737 |
+
"loss": 0.0025,
|
| 1738 |
+
"step": 2110
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 4.72,
|
| 1742 |
+
"grad_norm": 0.0010411780094727874,
|
| 1743 |
+
"learning_rate": 0.00010556792873051224,
|
| 1744 |
+
"loss": 0.0002,
|
| 1745 |
+
"step": 2120
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 4.74,
|
| 1749 |
+
"grad_norm": 0.0012262547388672829,
|
| 1750 |
+
"learning_rate": 0.00010512249443207127,
|
| 1751 |
+
"loss": 0.0031,
|
| 1752 |
+
"step": 2130
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 4.77,
|
| 1756 |
+
"grad_norm": 0.06668848544359207,
|
| 1757 |
+
"learning_rate": 0.0001046770601336303,
|
| 1758 |
+
"loss": 0.0011,
|
| 1759 |
+
"step": 2140
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 4.79,
|
| 1763 |
+
"grad_norm": 11.083710670471191,
|
| 1764 |
+
"learning_rate": 0.00010423162583518931,
|
| 1765 |
+
"loss": 0.1101,
|
| 1766 |
+
"step": 2150
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 4.81,
|
| 1770 |
+
"grad_norm": 0.0041348133236169815,
|
| 1771 |
+
"learning_rate": 0.00010378619153674833,
|
| 1772 |
+
"loss": 0.0312,
|
| 1773 |
+
"step": 2160
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 4.83,
|
| 1777 |
+
"grad_norm": 0.0014749247347936034,
|
| 1778 |
+
"learning_rate": 0.00010334075723830736,
|
| 1779 |
+
"loss": 0.002,
|
| 1780 |
+
"step": 2170
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 4.86,
|
| 1784 |
+
"grad_norm": 0.00816721748560667,
|
| 1785 |
+
"learning_rate": 0.00010289532293986637,
|
| 1786 |
+
"loss": 0.0374,
|
| 1787 |
+
"step": 2180
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 4.88,
|
| 1791 |
+
"grad_norm": 0.0029881075024604797,
|
| 1792 |
+
"learning_rate": 0.0001024498886414254,
|
| 1793 |
+
"loss": 0.0245,
|
| 1794 |
+
"step": 2190
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 4.9,
|
| 1798 |
+
"grad_norm": 0.01441921480000019,
|
| 1799 |
+
"learning_rate": 0.00010200445434298443,
|
| 1800 |
+
"loss": 0.0293,
|
| 1801 |
+
"step": 2200
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 4.9,
|
| 1805 |
+
"eval_accuracy": 0.9607843137254902,
|
| 1806 |
+
"eval_f1": 0.9607843137254902,
|
| 1807 |
+
"eval_loss": 0.17395375669002533,
|
| 1808 |
+
"eval_precision": 0.9607843137254902,
|
| 1809 |
+
"eval_recall": 0.9607843137254902,
|
| 1810 |
+
"eval_runtime": 17.1376,
|
| 1811 |
+
"eval_samples_per_second": 59.518,
|
| 1812 |
+
"eval_steps_per_second": 7.469,
|
| 1813 |
+
"step": 2200
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 4.92,
|
| 1817 |
+
"grad_norm": 0.005346087273210287,
|
| 1818 |
+
"learning_rate": 0.00010155902004454343,
|
| 1819 |
+
"loss": 0.054,
|
| 1820 |
+
"step": 2210
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 4.94,
|
| 1824 |
+
"grad_norm": 0.0087255435064435,
|
| 1825 |
+
"learning_rate": 0.00010111358574610245,
|
| 1826 |
+
"loss": 0.0008,
|
| 1827 |
+
"step": 2220
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 4.97,
|
| 1831 |
+
"grad_norm": 0.0033257934264838696,
|
| 1832 |
+
"learning_rate": 0.00010066815144766148,
|
| 1833 |
+
"loss": 0.0034,
|
| 1834 |
+
"step": 2230
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 4.99,
|
| 1838 |
+
"grad_norm": 0.0017741642659530044,
|
| 1839 |
+
"learning_rate": 0.00010022271714922049,
|
| 1840 |
+
"loss": 0.0008,
|
| 1841 |
+
"step": 2240
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 5.01,
|
| 1845 |
+
"grad_norm": 0.03513794392347336,
|
| 1846 |
+
"learning_rate": 9.977728285077952e-05,
|
| 1847 |
+
"loss": 0.0064,
|
| 1848 |
+
"step": 2250
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 5.03,
|
| 1852 |
+
"grad_norm": 0.0020874643232673407,
|
| 1853 |
+
"learning_rate": 9.933184855233853e-05,
|
| 1854 |
+
"loss": 0.0003,
|
| 1855 |
+
"step": 2260
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 5.06,
|
| 1859 |
+
"grad_norm": 0.0035891502629965544,
|
| 1860 |
+
"learning_rate": 9.888641425389756e-05,
|
| 1861 |
+
"loss": 0.0489,
|
| 1862 |
+
"step": 2270
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 5.08,
|
| 1866 |
+
"grad_norm": 0.001030069775879383,
|
| 1867 |
+
"learning_rate": 9.844097995545658e-05,
|
| 1868 |
+
"loss": 0.0011,
|
| 1869 |
+
"step": 2280
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 5.1,
|
| 1873 |
+
"grad_norm": 0.053950581699609756,
|
| 1874 |
+
"learning_rate": 9.79955456570156e-05,
|
| 1875 |
+
"loss": 0.0016,
|
| 1876 |
+
"step": 2290
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 5.12,
|
| 1880 |
+
"grad_norm": 0.0023863562382757664,
|
| 1881 |
+
"learning_rate": 9.755011135857461e-05,
|
| 1882 |
+
"loss": 0.0003,
|
| 1883 |
+
"step": 2300
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 5.12,
|
| 1887 |
+
"eval_accuracy": 0.9568627450980393,
|
| 1888 |
+
"eval_f1": 0.9568627450980393,
|
| 1889 |
+
"eval_loss": 0.25984036922454834,
|
| 1890 |
+
"eval_precision": 0.9568627450980393,
|
| 1891 |
+
"eval_recall": 0.9568627450980393,
|
| 1892 |
+
"eval_runtime": 17.3202,
|
| 1893 |
+
"eval_samples_per_second": 58.891,
|
| 1894 |
+
"eval_steps_per_second": 7.39,
|
| 1895 |
+
"step": 2300
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 5.14,
|
| 1899 |
+
"grad_norm": 0.054378170520067215,
|
| 1900 |
+
"learning_rate": 9.710467706013364e-05,
|
| 1901 |
+
"loss": 0.0005,
|
| 1902 |
+
"step": 2310
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 5.17,
|
| 1906 |
+
"grad_norm": 0.003204792272299528,
|
| 1907 |
+
"learning_rate": 9.665924276169266e-05,
|
| 1908 |
+
"loss": 0.0485,
|
| 1909 |
+
"step": 2320
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 5.19,
|
| 1913 |
+
"grad_norm": 0.001054179621860385,
|
| 1914 |
+
"learning_rate": 9.621380846325168e-05,
|
| 1915 |
+
"loss": 0.0456,
|
| 1916 |
+
"step": 2330
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 5.21,
|
| 1920 |
+
"grad_norm": 0.0008275459986180067,
|
| 1921 |
+
"learning_rate": 9.576837416481069e-05,
|
| 1922 |
+
"loss": 0.0003,
|
| 1923 |
+
"step": 2340
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 5.23,
|
| 1927 |
+
"grad_norm": 0.06051745265722275,
|
| 1928 |
+
"learning_rate": 9.532293986636972e-05,
|
| 1929 |
+
"loss": 0.0005,
|
| 1930 |
+
"step": 2350
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 5.26,
|
| 1934 |
+
"grad_norm": 0.0037743300199508667,
|
| 1935 |
+
"learning_rate": 9.487750556792874e-05,
|
| 1936 |
+
"loss": 0.0034,
|
| 1937 |
+
"step": 2360
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 5.28,
|
| 1941 |
+
"grad_norm": 0.00043078724411316216,
|
| 1942 |
+
"learning_rate": 9.443207126948775e-05,
|
| 1943 |
+
"loss": 0.0004,
|
| 1944 |
+
"step": 2370
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 5.3,
|
| 1948 |
+
"grad_norm": 0.001160395797342062,
|
| 1949 |
+
"learning_rate": 9.398663697104677e-05,
|
| 1950 |
+
"loss": 0.0009,
|
| 1951 |
+
"step": 2380
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 5.32,
|
| 1955 |
+
"grad_norm": 0.0006127849337644875,
|
| 1956 |
+
"learning_rate": 9.35412026726058e-05,
|
| 1957 |
+
"loss": 0.0001,
|
| 1958 |
+
"step": 2390
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 5.35,
|
| 1962 |
+
"grad_norm": 0.0023824446834623814,
|
| 1963 |
+
"learning_rate": 9.309576837416482e-05,
|
| 1964 |
+
"loss": 0.0037,
|
| 1965 |
+
"step": 2400
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 5.35,
|
| 1969 |
+
"eval_accuracy": 0.961764705882353,
|
| 1970 |
+
"eval_f1": 0.961764705882353,
|
| 1971 |
+
"eval_loss": 0.17718201875686646,
|
| 1972 |
+
"eval_precision": 0.961764705882353,
|
| 1973 |
+
"eval_recall": 0.961764705882353,
|
| 1974 |
+
"eval_runtime": 17.3617,
|
| 1975 |
+
"eval_samples_per_second": 58.75,
|
| 1976 |
+
"eval_steps_per_second": 7.373,
|
| 1977 |
+
"step": 2400
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 5.37,
|
| 1981 |
+
"grad_norm": 0.0015472627710551023,
|
| 1982 |
+
"learning_rate": 9.265033407572383e-05,
|
| 1983 |
+
"loss": 0.0132,
|
| 1984 |
+
"step": 2410
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 5.39,
|
| 1988 |
+
"grad_norm": 0.002203689655289054,
|
| 1989 |
+
"learning_rate": 9.220489977728286e-05,
|
| 1990 |
+
"loss": 0.0003,
|
| 1991 |
+
"step": 2420
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 5.41,
|
| 1995 |
+
"grad_norm": 0.003569718450307846,
|
| 1996 |
+
"learning_rate": 9.175946547884187e-05,
|
| 1997 |
+
"loss": 0.0006,
|
| 1998 |
+
"step": 2430
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 5.43,
|
| 2002 |
+
"grad_norm": 0.0020932252518832684,
|
| 2003 |
+
"learning_rate": 9.13140311804009e-05,
|
| 2004 |
+
"loss": 0.0051,
|
| 2005 |
+
"step": 2440
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 5.46,
|
| 2009 |
+
"grad_norm": 0.0030166106298565865,
|
| 2010 |
+
"learning_rate": 9.086859688195991e-05,
|
| 2011 |
+
"loss": 0.0008,
|
| 2012 |
+
"step": 2450
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 5.48,
|
| 2016 |
+
"grad_norm": 0.001667293719947338,
|
| 2017 |
+
"learning_rate": 9.042316258351894e-05,
|
| 2018 |
+
"loss": 0.0004,
|
| 2019 |
+
"step": 2460
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 5.5,
|
| 2023 |
+
"grad_norm": 0.0005573901580646634,
|
| 2024 |
+
"learning_rate": 8.997772828507795e-05,
|
| 2025 |
+
"loss": 0.0097,
|
| 2026 |
+
"step": 2470
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 5.52,
|
| 2030 |
+
"grad_norm": 0.0013365427730605006,
|
| 2031 |
+
"learning_rate": 8.953229398663698e-05,
|
| 2032 |
+
"loss": 0.0002,
|
| 2033 |
+
"step": 2480
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 5.55,
|
| 2037 |
+
"grad_norm": 0.0031586128752678633,
|
| 2038 |
+
"learning_rate": 8.9086859688196e-05,
|
| 2039 |
+
"loss": 0.0022,
|
| 2040 |
+
"step": 2490
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 5.57,
|
| 2044 |
+
"grad_norm": 0.21919859945774078,
|
| 2045 |
+
"learning_rate": 8.864142538975502e-05,
|
| 2046 |
+
"loss": 0.0213,
|
| 2047 |
+
"step": 2500
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 5.57,
|
| 2051 |
+
"eval_accuracy": 0.9519607843137254,
|
| 2052 |
+
"eval_f1": 0.9519607843137254,
|
| 2053 |
+
"eval_loss": 0.2910812199115753,
|
| 2054 |
+
"eval_precision": 0.9519607843137254,
|
| 2055 |
+
"eval_recall": 0.9519607843137254,
|
| 2056 |
+
"eval_runtime": 17.3785,
|
| 2057 |
+
"eval_samples_per_second": 58.693,
|
| 2058 |
+
"eval_steps_per_second": 7.365,
|
| 2059 |
+
"step": 2500
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 5.59,
|
| 2063 |
+
"grad_norm": 0.002107376931235194,
|
| 2064 |
+
"learning_rate": 8.819599109131403e-05,
|
| 2065 |
+
"loss": 0.001,
|
| 2066 |
+
"step": 2510
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 5.61,
|
| 2070 |
+
"grad_norm": 11.032261848449707,
|
| 2071 |
+
"learning_rate": 8.775055679287306e-05,
|
| 2072 |
+
"loss": 0.0331,
|
| 2073 |
+
"step": 2520
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 5.63,
|
| 2077 |
+
"grad_norm": 0.0010789623484015465,
|
| 2078 |
+
"learning_rate": 8.730512249443208e-05,
|
| 2079 |
+
"loss": 0.0002,
|
| 2080 |
+
"step": 2530
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 5.66,
|
| 2084 |
+
"grad_norm": 0.0013054576702415943,
|
| 2085 |
+
"learning_rate": 8.68596881959911e-05,
|
| 2086 |
+
"loss": 0.0019,
|
| 2087 |
+
"step": 2540
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 5.68,
|
| 2091 |
+
"grad_norm": 0.001707877148874104,
|
| 2092 |
+
"learning_rate": 8.641425389755011e-05,
|
| 2093 |
+
"loss": 0.0004,
|
| 2094 |
+
"step": 2550
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 5.7,
|
| 2098 |
+
"grad_norm": 0.0006996811716817319,
|
| 2099 |
+
"learning_rate": 8.596881959910914e-05,
|
| 2100 |
+
"loss": 0.0172,
|
| 2101 |
+
"step": 2560
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 5.72,
|
| 2105 |
+
"grad_norm": 0.00302655971609056,
|
| 2106 |
+
"learning_rate": 8.552338530066816e-05,
|
| 2107 |
+
"loss": 0.0461,
|
| 2108 |
+
"step": 2570
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 5.75,
|
| 2112 |
+
"grad_norm": 0.0005682900664396584,
|
| 2113 |
+
"learning_rate": 8.507795100222718e-05,
|
| 2114 |
+
"loss": 0.0002,
|
| 2115 |
+
"step": 2580
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 5.77,
|
| 2119 |
+
"grad_norm": 0.0009555260185152292,
|
| 2120 |
+
"learning_rate": 8.463251670378619e-05,
|
| 2121 |
+
"loss": 0.0002,
|
| 2122 |
+
"step": 2590
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 5.79,
|
| 2126 |
+
"grad_norm": 0.020968729630112648,
|
| 2127 |
+
"learning_rate": 8.418708240534521e-05,
|
| 2128 |
+
"loss": 0.027,
|
| 2129 |
+
"step": 2600
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 5.79,
|
| 2133 |
+
"eval_accuracy": 0.9519607843137254,
|
| 2134 |
+
"eval_f1": 0.9519607843137254,
|
| 2135 |
+
"eval_loss": 0.25403299927711487,
|
| 2136 |
+
"eval_precision": 0.9519607843137254,
|
| 2137 |
+
"eval_recall": 0.9519607843137254,
|
| 2138 |
+
"eval_runtime": 17.2499,
|
| 2139 |
+
"eval_samples_per_second": 59.131,
|
| 2140 |
+
"eval_steps_per_second": 7.42,
|
| 2141 |
+
"step": 2600
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 5.81,
|
| 2145 |
+
"grad_norm": 0.10158411413431168,
|
| 2146 |
+
"learning_rate": 8.374164810690424e-05,
|
| 2147 |
+
"loss": 0.0003,
|
| 2148 |
+
"step": 2610
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 5.84,
|
| 2152 |
+
"grad_norm": 0.016370078548789024,
|
| 2153 |
+
"learning_rate": 8.329621380846325e-05,
|
| 2154 |
+
"loss": 0.0981,
|
| 2155 |
+
"step": 2620
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 5.86,
|
| 2159 |
+
"grad_norm": 0.0012518821749836206,
|
| 2160 |
+
"learning_rate": 8.285077951002228e-05,
|
| 2161 |
+
"loss": 0.0004,
|
| 2162 |
+
"step": 2630
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 5.88,
|
| 2166 |
+
"grad_norm": 0.0029411010909825563,
|
| 2167 |
+
"learning_rate": 8.24053452115813e-05,
|
| 2168 |
+
"loss": 0.0007,
|
| 2169 |
+
"step": 2640
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 5.9,
|
| 2173 |
+
"grad_norm": 0.000937216158490628,
|
| 2174 |
+
"learning_rate": 8.195991091314032e-05,
|
| 2175 |
+
"loss": 0.0177,
|
| 2176 |
+
"step": 2650
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 5.92,
|
| 2180 |
+
"grad_norm": 0.0034318570978939533,
|
| 2181 |
+
"learning_rate": 8.151447661469933e-05,
|
| 2182 |
+
"loss": 0.0312,
|
| 2183 |
+
"step": 2660
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 5.95,
|
| 2187 |
+
"grad_norm": 0.0007262133876793087,
|
| 2188 |
+
"learning_rate": 8.106904231625836e-05,
|
| 2189 |
+
"loss": 0.0002,
|
| 2190 |
+
"step": 2670
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 5.97,
|
| 2194 |
+
"grad_norm": 0.0007804339984431863,
|
| 2195 |
+
"learning_rate": 8.062360801781737e-05,
|
| 2196 |
+
"loss": 0.0002,
|
| 2197 |
+
"step": 2680
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 5.99,
|
| 2201 |
+
"grad_norm": 0.001342720352113247,
|
| 2202 |
+
"learning_rate": 8.01781737193764e-05,
|
| 2203 |
+
"loss": 0.0002,
|
| 2204 |
+
"step": 2690
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 6.01,
|
| 2208 |
+
"grad_norm": 0.001933095627464354,
|
| 2209 |
+
"learning_rate": 7.973273942093543e-05,
|
| 2210 |
+
"loss": 0.0155,
|
| 2211 |
+
"step": 2700
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 6.01,
|
| 2215 |
+
"eval_accuracy": 0.9549019607843138,
|
| 2216 |
+
"eval_f1": 0.9549019607843138,
|
| 2217 |
+
"eval_loss": 0.22524712979793549,
|
| 2218 |
+
"eval_precision": 0.9549019607843138,
|
| 2219 |
+
"eval_recall": 0.9549019607843138,
|
| 2220 |
+
"eval_runtime": 17.4105,
|
| 2221 |
+
"eval_samples_per_second": 58.585,
|
| 2222 |
+
"eval_steps_per_second": 7.352,
|
| 2223 |
+
"step": 2700
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 6.04,
|
| 2227 |
+
"grad_norm": 0.0011171975638717413,
|
| 2228 |
+
"learning_rate": 7.928730512249444e-05,
|
| 2229 |
+
"loss": 0.0002,
|
| 2230 |
+
"step": 2710
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 6.06,
|
| 2234 |
+
"grad_norm": 0.001169139752164483,
|
| 2235 |
+
"learning_rate": 7.884187082405345e-05,
|
| 2236 |
+
"loss": 0.0002,
|
| 2237 |
+
"step": 2720
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 6.08,
|
| 2241 |
+
"grad_norm": 0.002055455232039094,
|
| 2242 |
+
"learning_rate": 7.839643652561248e-05,
|
| 2243 |
+
"loss": 0.0082,
|
| 2244 |
+
"step": 2730
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 6.1,
|
| 2248 |
+
"grad_norm": 0.05008271709084511,
|
| 2249 |
+
"learning_rate": 7.79510022271715e-05,
|
| 2250 |
+
"loss": 0.0321,
|
| 2251 |
+
"step": 2740
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 6.12,
|
| 2255 |
+
"grad_norm": 0.000607622554525733,
|
| 2256 |
+
"learning_rate": 7.750556792873052e-05,
|
| 2257 |
+
"loss": 0.0397,
|
| 2258 |
+
"step": 2750
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 6.15,
|
| 2262 |
+
"grad_norm": 0.000795868574641645,
|
| 2263 |
+
"learning_rate": 7.706013363028953e-05,
|
| 2264 |
+
"loss": 0.0064,
|
| 2265 |
+
"step": 2760
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 6.17,
|
| 2269 |
+
"grad_norm": 0.010306187905371189,
|
| 2270 |
+
"learning_rate": 7.661469933184856e-05,
|
| 2271 |
+
"loss": 0.004,
|
| 2272 |
+
"step": 2770
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 6.19,
|
| 2276 |
+
"grad_norm": 0.001497789635322988,
|
| 2277 |
+
"learning_rate": 7.616926503340758e-05,
|
| 2278 |
+
"loss": 0.005,
|
| 2279 |
+
"step": 2780
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 6.21,
|
| 2283 |
+
"grad_norm": 0.12043255567550659,
|
| 2284 |
+
"learning_rate": 7.57238307349666e-05,
|
| 2285 |
+
"loss": 0.003,
|
| 2286 |
+
"step": 2790
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 6.24,
|
| 2290 |
+
"grad_norm": 0.0034782905131578445,
|
| 2291 |
+
"learning_rate": 7.527839643652561e-05,
|
| 2292 |
+
"loss": 0.0002,
|
| 2293 |
+
"step": 2800
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 6.24,
|
| 2297 |
+
"eval_accuracy": 0.9431372549019608,
|
| 2298 |
+
"eval_f1": 0.9431372549019608,
|
| 2299 |
+
"eval_loss": 0.3040062189102173,
|
| 2300 |
+
"eval_precision": 0.9431372549019608,
|
| 2301 |
+
"eval_recall": 0.9431372549019608,
|
| 2302 |
+
"eval_runtime": 17.1818,
|
| 2303 |
+
"eval_samples_per_second": 59.365,
|
| 2304 |
+
"eval_steps_per_second": 7.45,
|
| 2305 |
+
"step": 2800
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 6.26,
|
| 2309 |
+
"grad_norm": 5.399389266967773,
|
| 2310 |
+
"learning_rate": 7.483296213808464e-05,
|
| 2311 |
+
"loss": 0.0092,
|
| 2312 |
+
"step": 2810
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 6.28,
|
| 2316 |
+
"grad_norm": 0.0007157580694183707,
|
| 2317 |
+
"learning_rate": 7.438752783964366e-05,
|
| 2318 |
+
"loss": 0.0134,
|
| 2319 |
+
"step": 2820
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 6.3,
|
| 2323 |
+
"grad_norm": 0.005470567848533392,
|
| 2324 |
+
"learning_rate": 7.394209354120267e-05,
|
| 2325 |
+
"loss": 0.147,
|
| 2326 |
+
"step": 2830
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 6.33,
|
| 2330 |
+
"grad_norm": 0.01675906591117382,
|
| 2331 |
+
"learning_rate": 7.34966592427617e-05,
|
| 2332 |
+
"loss": 0.0813,
|
| 2333 |
+
"step": 2840
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 6.35,
|
| 2337 |
+
"grad_norm": 5.728336334228516,
|
| 2338 |
+
"learning_rate": 7.305122494432071e-05,
|
| 2339 |
+
"loss": 0.0275,
|
| 2340 |
+
"step": 2850
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 6.37,
|
| 2344 |
+
"grad_norm": 0.003522884799167514,
|
| 2345 |
+
"learning_rate": 7.260579064587974e-05,
|
| 2346 |
+
"loss": 0.0109,
|
| 2347 |
+
"step": 2860
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 6.39,
|
| 2351 |
+
"grad_norm": 0.026970118284225464,
|
| 2352 |
+
"learning_rate": 7.216035634743875e-05,
|
| 2353 |
+
"loss": 0.0004,
|
| 2354 |
+
"step": 2870
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 6.41,
|
| 2358 |
+
"grad_norm": 0.12984509766101837,
|
| 2359 |
+
"learning_rate": 7.171492204899778e-05,
|
| 2360 |
+
"loss": 0.0007,
|
| 2361 |
+
"step": 2880
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 6.44,
|
| 2365 |
+
"grad_norm": 0.009168056771159172,
|
| 2366 |
+
"learning_rate": 7.126948775055679e-05,
|
| 2367 |
+
"loss": 0.0006,
|
| 2368 |
+
"step": 2890
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 6.46,
|
| 2372 |
+
"grad_norm": 0.0009597400785423815,
|
| 2373 |
+
"learning_rate": 7.082405345211582e-05,
|
| 2374 |
+
"loss": 0.011,
|
| 2375 |
+
"step": 2900
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 6.46,
|
| 2379 |
+
"eval_accuracy": 0.9598039215686275,
|
| 2380 |
+
"eval_f1": 0.9598039215686275,
|
| 2381 |
+
"eval_loss": 0.1923176795244217,
|
| 2382 |
+
"eval_precision": 0.9598039215686275,
|
| 2383 |
+
"eval_recall": 0.9598039215686275,
|
| 2384 |
+
"eval_runtime": 17.225,
|
| 2385 |
+
"eval_samples_per_second": 59.216,
|
| 2386 |
+
"eval_steps_per_second": 7.431,
|
| 2387 |
+
"step": 2900
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 6.48,
|
| 2391 |
+
"grad_norm": 0.0009739417000673711,
|
| 2392 |
+
"learning_rate": 7.037861915367485e-05,
|
| 2393 |
+
"loss": 0.0014,
|
| 2394 |
+
"step": 2910
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 6.5,
|
| 2398 |
+
"grad_norm": 0.0022935476154088974,
|
| 2399 |
+
"learning_rate": 6.993318485523386e-05,
|
| 2400 |
+
"loss": 0.0133,
|
| 2401 |
+
"step": 2920
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 6.53,
|
| 2405 |
+
"grad_norm": 0.0005638069123961031,
|
| 2406 |
+
"learning_rate": 6.948775055679287e-05,
|
| 2407 |
+
"loss": 0.0009,
|
| 2408 |
+
"step": 2930
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 6.55,
|
| 2412 |
+
"grad_norm": 0.014625852927565575,
|
| 2413 |
+
"learning_rate": 6.904231625835188e-05,
|
| 2414 |
+
"loss": 0.0006,
|
| 2415 |
+
"step": 2940
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 6.57,
|
| 2419 |
+
"grad_norm": 0.001474756863899529,
|
| 2420 |
+
"learning_rate": 6.859688195991092e-05,
|
| 2421 |
+
"loss": 0.0128,
|
| 2422 |
+
"step": 2950
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 6.59,
|
| 2426 |
+
"grad_norm": 0.0029620621353387833,
|
| 2427 |
+
"learning_rate": 6.815144766146994e-05,
|
| 2428 |
+
"loss": 0.0002,
|
| 2429 |
+
"step": 2960
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 6.61,
|
| 2433 |
+
"grad_norm": 0.0016939816996455193,
|
| 2434 |
+
"learning_rate": 6.770601336302895e-05,
|
| 2435 |
+
"loss": 0.0003,
|
| 2436 |
+
"step": 2970
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 6.64,
|
| 2440 |
+
"grad_norm": 0.0009252108866348863,
|
| 2441 |
+
"learning_rate": 6.726057906458798e-05,
|
| 2442 |
+
"loss": 0.0011,
|
| 2443 |
+
"step": 2980
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 6.66,
|
| 2447 |
+
"grad_norm": 0.1285027265548706,
|
| 2448 |
+
"learning_rate": 6.6815144766147e-05,
|
| 2449 |
+
"loss": 0.0071,
|
| 2450 |
+
"step": 2990
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 6.68,
|
| 2454 |
+
"grad_norm": 0.00125114805996418,
|
| 2455 |
+
"learning_rate": 6.636971046770602e-05,
|
| 2456 |
+
"loss": 0.0006,
|
| 2457 |
+
"step": 3000
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 6.68,
|
| 2461 |
+
"eval_accuracy": 0.9637254901960784,
|
| 2462 |
+
"eval_f1": 0.9637254901960784,
|
| 2463 |
+
"eval_loss": 0.20890936255455017,
|
| 2464 |
+
"eval_precision": 0.9637254901960784,
|
| 2465 |
+
"eval_recall": 0.9637254901960784,
|
| 2466 |
+
"eval_runtime": 17.1679,
|
| 2467 |
+
"eval_samples_per_second": 59.413,
|
| 2468 |
+
"eval_steps_per_second": 7.456,
|
| 2469 |
+
"step": 3000
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 6.7,
|
| 2473 |
+
"grad_norm": 0.0017677777213975787,
|
| 2474 |
+
"learning_rate": 6.592427616926503e-05,
|
| 2475 |
+
"loss": 0.03,
|
| 2476 |
+
"step": 3010
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 6.73,
|
| 2480 |
+
"grad_norm": 0.0006067939684726298,
|
| 2481 |
+
"learning_rate": 6.547884187082406e-05,
|
| 2482 |
+
"loss": 0.0003,
|
| 2483 |
+
"step": 3020
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 6.75,
|
| 2487 |
+
"grad_norm": 0.0010102881351485848,
|
| 2488 |
+
"learning_rate": 6.503340757238308e-05,
|
| 2489 |
+
"loss": 0.0002,
|
| 2490 |
+
"step": 3030
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 6.77,
|
| 2494 |
+
"grad_norm": 0.0009144017240032554,
|
| 2495 |
+
"learning_rate": 6.45879732739421e-05,
|
| 2496 |
+
"loss": 0.0282,
|
| 2497 |
+
"step": 3040
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 6.79,
|
| 2501 |
+
"grad_norm": 0.0007274287054315209,
|
| 2502 |
+
"learning_rate": 6.414253897550112e-05,
|
| 2503 |
+
"loss": 0.0002,
|
| 2504 |
+
"step": 3050
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 6.82,
|
| 2508 |
+
"grad_norm": 0.0034935837611556053,
|
| 2509 |
+
"learning_rate": 6.369710467706013e-05,
|
| 2510 |
+
"loss": 0.0046,
|
| 2511 |
+
"step": 3060
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 6.84,
|
| 2515 |
+
"grad_norm": 0.004235483705997467,
|
| 2516 |
+
"learning_rate": 6.325167037861916e-05,
|
| 2517 |
+
"loss": 0.0002,
|
| 2518 |
+
"step": 3070
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 6.86,
|
| 2522 |
+
"grad_norm": 0.0013753636740148067,
|
| 2523 |
+
"learning_rate": 6.280623608017817e-05,
|
| 2524 |
+
"loss": 0.0101,
|
| 2525 |
+
"step": 3080
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 6.88,
|
| 2529 |
+
"grad_norm": 0.0008035491337068379,
|
| 2530 |
+
"learning_rate": 6.23608017817372e-05,
|
| 2531 |
+
"loss": 0.0002,
|
| 2532 |
+
"step": 3090
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 6.9,
|
| 2536 |
+
"grad_norm": 0.00173095241189003,
|
| 2537 |
+
"learning_rate": 6.191536748329621e-05,
|
| 2538 |
+
"loss": 0.0002,
|
| 2539 |
+
"step": 3100
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 6.9,
|
| 2543 |
+
"eval_accuracy": 0.957843137254902,
|
| 2544 |
+
"eval_f1": 0.957843137254902,
|
| 2545 |
+
"eval_loss": 0.22062458097934723,
|
| 2546 |
+
"eval_precision": 0.957843137254902,
|
| 2547 |
+
"eval_recall": 0.957843137254902,
|
| 2548 |
+
"eval_runtime": 17.3063,
|
| 2549 |
+
"eval_samples_per_second": 58.938,
|
| 2550 |
+
"eval_steps_per_second": 7.396,
|
| 2551 |
+
"step": 3100
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 6.93,
|
| 2555 |
+
"grad_norm": 0.001496517681516707,
|
| 2556 |
+
"learning_rate": 6.146993318485523e-05,
|
| 2557 |
+
"loss": 0.0002,
|
| 2558 |
+
"step": 3110
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 6.95,
|
| 2562 |
+
"grad_norm": 0.0005152082885615528,
|
| 2563 |
+
"learning_rate": 6.102449888641426e-05,
|
| 2564 |
+
"loss": 0.0002,
|
| 2565 |
+
"step": 3120
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 6.97,
|
| 2569 |
+
"grad_norm": 0.0005918457172811031,
|
| 2570 |
+
"learning_rate": 6.057906458797328e-05,
|
| 2571 |
+
"loss": 0.0192,
|
| 2572 |
+
"step": 3130
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 6.99,
|
| 2576 |
+
"grad_norm": 0.000515251827891916,
|
| 2577 |
+
"learning_rate": 6.013363028953229e-05,
|
| 2578 |
+
"loss": 0.0003,
|
| 2579 |
+
"step": 3140
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 7.02,
|
| 2583 |
+
"grad_norm": 0.00043858023127540946,
|
| 2584 |
+
"learning_rate": 5.9688195991091325e-05,
|
| 2585 |
+
"loss": 0.0105,
|
| 2586 |
+
"step": 3150
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 7.04,
|
| 2590 |
+
"grad_norm": 0.0018106413772329688,
|
| 2591 |
+
"learning_rate": 5.924276169265034e-05,
|
| 2592 |
+
"loss": 0.0059,
|
| 2593 |
+
"step": 3160
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 7.06,
|
| 2597 |
+
"grad_norm": 0.000563229201361537,
|
| 2598 |
+
"learning_rate": 5.879732739420936e-05,
|
| 2599 |
+
"loss": 0.0003,
|
| 2600 |
+
"step": 3170
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 7.08,
|
| 2604 |
+
"grad_norm": 0.001515958341769874,
|
| 2605 |
+
"learning_rate": 5.835189309576837e-05,
|
| 2606 |
+
"loss": 0.0058,
|
| 2607 |
+
"step": 3180
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 7.1,
|
| 2611 |
+
"grad_norm": 0.0005047390004619956,
|
| 2612 |
+
"learning_rate": 5.79064587973274e-05,
|
| 2613 |
+
"loss": 0.0002,
|
| 2614 |
+
"step": 3190
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 7.13,
|
| 2618 |
+
"grad_norm": 0.03174121677875519,
|
| 2619 |
+
"learning_rate": 5.746102449888642e-05,
|
| 2620 |
+
"loss": 0.0006,
|
| 2621 |
+
"step": 3200
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 7.13,
|
| 2625 |
+
"eval_accuracy": 0.9627450980392157,
|
| 2626 |
+
"eval_f1": 0.9627450980392157,
|
| 2627 |
+
"eval_loss": 0.22668223083019257,
|
| 2628 |
+
"eval_precision": 0.9627450980392157,
|
| 2629 |
+
"eval_recall": 0.9627450980392157,
|
| 2630 |
+
"eval_runtime": 16.905,
|
| 2631 |
+
"eval_samples_per_second": 60.337,
|
| 2632 |
+
"eval_steps_per_second": 7.572,
|
| 2633 |
+
"step": 3200
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 7.15,
|
| 2637 |
+
"grad_norm": 0.0019246222218498588,
|
| 2638 |
+
"learning_rate": 5.701559020044544e-05,
|
| 2639 |
+
"loss": 0.0002,
|
| 2640 |
+
"step": 3210
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 7.17,
|
| 2644 |
+
"grad_norm": 8.673022270202637,
|
| 2645 |
+
"learning_rate": 5.6570155902004463e-05,
|
| 2646 |
+
"loss": 0.0058,
|
| 2647 |
+
"step": 3220
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 7.19,
|
| 2651 |
+
"grad_norm": 0.0006804656004533172,
|
| 2652 |
+
"learning_rate": 5.6124721603563476e-05,
|
| 2653 |
+
"loss": 0.0002,
|
| 2654 |
+
"step": 3230
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 7.22,
|
| 2658 |
+
"grad_norm": 0.0013651803601533175,
|
| 2659 |
+
"learning_rate": 5.5679287305122496e-05,
|
| 2660 |
+
"loss": 0.0007,
|
| 2661 |
+
"step": 3240
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 7.24,
|
| 2665 |
+
"grad_norm": 0.0014620161382481456,
|
| 2666 |
+
"learning_rate": 5.5233853006681516e-05,
|
| 2667 |
+
"loss": 0.0002,
|
| 2668 |
+
"step": 3250
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 7.26,
|
| 2672 |
+
"grad_norm": 0.0009020831785164773,
|
| 2673 |
+
"learning_rate": 5.478841870824054e-05,
|
| 2674 |
+
"loss": 0.0003,
|
| 2675 |
+
"step": 3260
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 7.28,
|
| 2679 |
+
"grad_norm": 0.0010328377829864621,
|
| 2680 |
+
"learning_rate": 5.4342984409799555e-05,
|
| 2681 |
+
"loss": 0.0016,
|
| 2682 |
+
"step": 3270
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 7.31,
|
| 2686 |
+
"grad_norm": 0.002698230091482401,
|
| 2687 |
+
"learning_rate": 5.3897550111358575e-05,
|
| 2688 |
+
"loss": 0.0002,
|
| 2689 |
+
"step": 3280
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 7.33,
|
| 2693 |
+
"grad_norm": 0.0025662758853286505,
|
| 2694 |
+
"learning_rate": 5.34521158129176e-05,
|
| 2695 |
+
"loss": 0.0002,
|
| 2696 |
+
"step": 3290
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 7.35,
|
| 2700 |
+
"grad_norm": 0.0005825618281960487,
|
| 2701 |
+
"learning_rate": 5.300668151447662e-05,
|
| 2702 |
+
"loss": 0.0001,
|
| 2703 |
+
"step": 3300
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 7.35,
|
| 2707 |
+
"eval_accuracy": 0.9637254901960784,
|
| 2708 |
+
"eval_f1": 0.9637254901960784,
|
| 2709 |
+
"eval_loss": 0.17346832156181335,
|
| 2710 |
+
"eval_precision": 0.9637254901960784,
|
| 2711 |
+
"eval_recall": 0.9637254901960784,
|
| 2712 |
+
"eval_runtime": 17.0715,
|
| 2713 |
+
"eval_samples_per_second": 59.749,
|
| 2714 |
+
"eval_steps_per_second": 7.498,
|
| 2715 |
+
"step": 3300
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 7.37,
|
| 2719 |
+
"grad_norm": 0.001066404627636075,
|
| 2720 |
+
"learning_rate": 5.2561247216035634e-05,
|
| 2721 |
+
"loss": 0.0001,
|
| 2722 |
+
"step": 3310
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 7.39,
|
| 2726 |
+
"grad_norm": 0.0007172970799729228,
|
| 2727 |
+
"learning_rate": 5.2115812917594654e-05,
|
| 2728 |
+
"loss": 0.0002,
|
| 2729 |
+
"step": 3320
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 7.42,
|
| 2733 |
+
"grad_norm": 0.000634915370028466,
|
| 2734 |
+
"learning_rate": 5.167037861915368e-05,
|
| 2735 |
+
"loss": 0.0001,
|
| 2736 |
+
"step": 3330
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 7.44,
|
| 2740 |
+
"grad_norm": 0.004406619351357222,
|
| 2741 |
+
"learning_rate": 5.12249443207127e-05,
|
| 2742 |
+
"loss": 0.0001,
|
| 2743 |
+
"step": 3340
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 7.46,
|
| 2747 |
+
"grad_norm": 0.015614562667906284,
|
| 2748 |
+
"learning_rate": 5.077951002227171e-05,
|
| 2749 |
+
"loss": 0.0002,
|
| 2750 |
+
"step": 3350
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 7.48,
|
| 2754 |
+
"grad_norm": 0.0017906671855598688,
|
| 2755 |
+
"learning_rate": 5.033407572383074e-05,
|
| 2756 |
+
"loss": 0.0003,
|
| 2757 |
+
"step": 3360
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 7.51,
|
| 2761 |
+
"grad_norm": 0.002051855204626918,
|
| 2762 |
+
"learning_rate": 4.988864142538976e-05,
|
| 2763 |
+
"loss": 0.0003,
|
| 2764 |
+
"step": 3370
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 7.53,
|
| 2768 |
+
"grad_norm": 0.0007296734838746488,
|
| 2769 |
+
"learning_rate": 4.944320712694878e-05,
|
| 2770 |
+
"loss": 0.0001,
|
| 2771 |
+
"step": 3380
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 7.55,
|
| 2775 |
+
"grad_norm": 0.0005030659376643598,
|
| 2776 |
+
"learning_rate": 4.89977728285078e-05,
|
| 2777 |
+
"loss": 0.001,
|
| 2778 |
+
"step": 3390
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 7.57,
|
| 2782 |
+
"grad_norm": 0.00040412909584119916,
|
| 2783 |
+
"learning_rate": 4.855233853006682e-05,
|
| 2784 |
+
"loss": 0.0001,
|
| 2785 |
+
"step": 3400
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 7.57,
|
| 2789 |
+
"eval_accuracy": 0.9686274509803922,
|
| 2790 |
+
"eval_f1": 0.9686274509803922,
|
| 2791 |
+
"eval_loss": 0.16111387312412262,
|
| 2792 |
+
"eval_precision": 0.9686274509803922,
|
| 2793 |
+
"eval_recall": 0.9686274509803922,
|
| 2794 |
+
"eval_runtime": 16.979,
|
| 2795 |
+
"eval_samples_per_second": 60.074,
|
| 2796 |
+
"eval_steps_per_second": 7.539,
|
| 2797 |
+
"step": 3400
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 7.59,
|
| 2801 |
+
"grad_norm": 0.0005902125267311931,
|
| 2802 |
+
"learning_rate": 4.810690423162584e-05,
|
| 2803 |
+
"loss": 0.0002,
|
| 2804 |
+
"step": 3410
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 7.62,
|
| 2808 |
+
"grad_norm": 0.0005297433235682547,
|
| 2809 |
+
"learning_rate": 4.766146993318486e-05,
|
| 2810 |
+
"loss": 0.0002,
|
| 2811 |
+
"step": 3420
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 7.64,
|
| 2815 |
+
"grad_norm": 0.0005097580142319202,
|
| 2816 |
+
"learning_rate": 4.721603563474388e-05,
|
| 2817 |
+
"loss": 0.0001,
|
| 2818 |
+
"step": 3430
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 7.66,
|
| 2822 |
+
"grad_norm": 0.0006824088632129133,
|
| 2823 |
+
"learning_rate": 4.67706013363029e-05,
|
| 2824 |
+
"loss": 0.0002,
|
| 2825 |
+
"step": 3440
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 7.68,
|
| 2829 |
+
"grad_norm": 0.0007693713996559381,
|
| 2830 |
+
"learning_rate": 4.632516703786192e-05,
|
| 2831 |
+
"loss": 0.0002,
|
| 2832 |
+
"step": 3450
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 7.71,
|
| 2836 |
+
"grad_norm": 0.0010686744935810566,
|
| 2837 |
+
"learning_rate": 4.5879732739420936e-05,
|
| 2838 |
+
"loss": 0.0001,
|
| 2839 |
+
"step": 3460
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 7.73,
|
| 2843 |
+
"grad_norm": 0.0008867672295309603,
|
| 2844 |
+
"learning_rate": 4.5434298440979956e-05,
|
| 2845 |
+
"loss": 0.0001,
|
| 2846 |
+
"step": 3470
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 7.75,
|
| 2850 |
+
"grad_norm": 0.029097959399223328,
|
| 2851 |
+
"learning_rate": 4.4988864142538976e-05,
|
| 2852 |
+
"loss": 0.0002,
|
| 2853 |
+
"step": 3480
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 7.77,
|
| 2857 |
+
"grad_norm": 1.0295618772506714,
|
| 2858 |
+
"learning_rate": 4.4543429844098e-05,
|
| 2859 |
+
"loss": 0.0232,
|
| 2860 |
+
"step": 3490
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 7.8,
|
| 2864 |
+
"grad_norm": 0.01833498664200306,
|
| 2865 |
+
"learning_rate": 4.4097995545657015e-05,
|
| 2866 |
+
"loss": 0.0003,
|
| 2867 |
+
"step": 3500
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 7.8,
|
| 2871 |
+
"eval_accuracy": 0.9676470588235294,
|
| 2872 |
+
"eval_f1": 0.9676470588235294,
|
| 2873 |
+
"eval_loss": 0.15838229656219482,
|
| 2874 |
+
"eval_precision": 0.9676470588235294,
|
| 2875 |
+
"eval_recall": 0.9676470588235294,
|
| 2876 |
+
"eval_runtime": 17.1614,
|
| 2877 |
+
"eval_samples_per_second": 59.436,
|
| 2878 |
+
"eval_steps_per_second": 7.459,
|
| 2879 |
+
"step": 3500
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 7.82,
|
| 2883 |
+
"grad_norm": 0.00047049217391759157,
|
| 2884 |
+
"learning_rate": 4.365256124721604e-05,
|
| 2885 |
+
"loss": 0.0002,
|
| 2886 |
+
"step": 3510
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 7.84,
|
| 2890 |
+
"grad_norm": 0.0005326452082954347,
|
| 2891 |
+
"learning_rate": 4.3207126948775055e-05,
|
| 2892 |
+
"loss": 0.0001,
|
| 2893 |
+
"step": 3520
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 7.86,
|
| 2897 |
+
"grad_norm": 0.004704196471720934,
|
| 2898 |
+
"learning_rate": 4.276169265033408e-05,
|
| 2899 |
+
"loss": 0.0001,
|
| 2900 |
+
"step": 3530
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 7.88,
|
| 2904 |
+
"grad_norm": 0.0015603323699906468,
|
| 2905 |
+
"learning_rate": 4.2316258351893094e-05,
|
| 2906 |
+
"loss": 0.0001,
|
| 2907 |
+
"step": 3540
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 7.91,
|
| 2911 |
+
"grad_norm": 0.001259263837710023,
|
| 2912 |
+
"learning_rate": 4.187082405345212e-05,
|
| 2913 |
+
"loss": 0.0002,
|
| 2914 |
+
"step": 3550
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 7.93,
|
| 2918 |
+
"grad_norm": 0.0009968471713364124,
|
| 2919 |
+
"learning_rate": 4.142538975501114e-05,
|
| 2920 |
+
"loss": 0.0173,
|
| 2921 |
+
"step": 3560
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 7.95,
|
| 2925 |
+
"grad_norm": 0.0011363897938281298,
|
| 2926 |
+
"learning_rate": 4.097995545657016e-05,
|
| 2927 |
+
"loss": 0.0002,
|
| 2928 |
+
"step": 3570
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 7.97,
|
| 2932 |
+
"grad_norm": 0.000970890570897609,
|
| 2933 |
+
"learning_rate": 4.053452115812918e-05,
|
| 2934 |
+
"loss": 0.0002,
|
| 2935 |
+
"step": 3580
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 8.0,
|
| 2939 |
+
"grad_norm": 0.0004184432327747345,
|
| 2940 |
+
"learning_rate": 4.00890868596882e-05,
|
| 2941 |
+
"loss": 0.0002,
|
| 2942 |
+
"step": 3590
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 8.02,
|
| 2946 |
+
"grad_norm": 0.0016488181427121162,
|
| 2947 |
+
"learning_rate": 3.964365256124722e-05,
|
| 2948 |
+
"loss": 0.0001,
|
| 2949 |
+
"step": 3600
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 8.02,
|
| 2953 |
+
"eval_accuracy": 0.9715686274509804,
|
| 2954 |
+
"eval_f1": 0.9715686274509804,
|
| 2955 |
+
"eval_loss": 0.15909050405025482,
|
| 2956 |
+
"eval_precision": 0.9715686274509804,
|
| 2957 |
+
"eval_recall": 0.9715686274509804,
|
| 2958 |
+
"eval_runtime": 17.2363,
|
| 2959 |
+
"eval_samples_per_second": 59.178,
|
| 2960 |
+
"eval_steps_per_second": 7.426,
|
| 2961 |
+
"step": 3600
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 8.04,
|
| 2965 |
+
"grad_norm": 0.0007161149405874312,
|
| 2966 |
+
"learning_rate": 3.919821826280624e-05,
|
| 2967 |
+
"loss": 0.0001,
|
| 2968 |
+
"step": 3610
|
| 2969 |
+
},
|
| 2970 |
+
{
|
| 2971 |
+
"epoch": 8.06,
|
| 2972 |
+
"grad_norm": 0.025815103203058243,
|
| 2973 |
+
"learning_rate": 3.875278396436526e-05,
|
| 2974 |
+
"loss": 0.0002,
|
| 2975 |
+
"step": 3620
|
| 2976 |
+
},
|
| 2977 |
+
{
|
| 2978 |
+
"epoch": 8.08,
|
| 2979 |
+
"grad_norm": 0.6380942463874817,
|
| 2980 |
+
"learning_rate": 3.830734966592428e-05,
|
| 2981 |
+
"loss": 0.0067,
|
| 2982 |
+
"step": 3630
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 8.11,
|
| 2986 |
+
"grad_norm": 0.003612744389101863,
|
| 2987 |
+
"learning_rate": 3.78619153674833e-05,
|
| 2988 |
+
"loss": 0.0004,
|
| 2989 |
+
"step": 3640
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 8.13,
|
| 2993 |
+
"grad_norm": 0.002143553225323558,
|
| 2994 |
+
"learning_rate": 3.741648106904232e-05,
|
| 2995 |
+
"loss": 0.0002,
|
| 2996 |
+
"step": 3650
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 8.15,
|
| 3000 |
+
"grad_norm": 0.001735298428684473,
|
| 3001 |
+
"learning_rate": 3.697104677060134e-05,
|
| 3002 |
+
"loss": 0.0002,
|
| 3003 |
+
"step": 3660
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 8.17,
|
| 3007 |
+
"grad_norm": 0.00038884536479599774,
|
| 3008 |
+
"learning_rate": 3.652561247216036e-05,
|
| 3009 |
+
"loss": 0.0004,
|
| 3010 |
+
"step": 3670
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 8.2,
|
| 3014 |
+
"grad_norm": 0.0004352598334662616,
|
| 3015 |
+
"learning_rate": 3.608017817371938e-05,
|
| 3016 |
+
"loss": 0.0001,
|
| 3017 |
+
"step": 3680
|
| 3018 |
+
},
|
| 3019 |
+
{
|
| 3020 |
+
"epoch": 8.22,
|
| 3021 |
+
"grad_norm": 0.0030426643788814545,
|
| 3022 |
+
"learning_rate": 3.5634743875278396e-05,
|
| 3023 |
+
"loss": 0.0002,
|
| 3024 |
+
"step": 3690
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 8.24,
|
| 3028 |
+
"grad_norm": 0.0004016205493826419,
|
| 3029 |
+
"learning_rate": 3.518930957683742e-05,
|
| 3030 |
+
"loss": 0.0005,
|
| 3031 |
+
"step": 3700
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 8.24,
|
| 3035 |
+
"eval_accuracy": 0.9705882352941176,
|
| 3036 |
+
"eval_f1": 0.9705882352941176,
|
| 3037 |
+
"eval_loss": 0.15964852273464203,
|
| 3038 |
+
"eval_precision": 0.9705882352941176,
|
| 3039 |
+
"eval_recall": 0.9705882352941176,
|
| 3040 |
+
"eval_runtime": 17.4233,
|
| 3041 |
+
"eval_samples_per_second": 58.542,
|
| 3042 |
+
"eval_steps_per_second": 7.346,
|
| 3043 |
+
"step": 3700
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 8.26,
|
| 3047 |
+
"grad_norm": 0.0010756178526207805,
|
| 3048 |
+
"learning_rate": 3.4743875278396436e-05,
|
| 3049 |
+
"loss": 0.0001,
|
| 3050 |
+
"step": 3710
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 8.29,
|
| 3054 |
+
"grad_norm": 0.002286019967868924,
|
| 3055 |
+
"learning_rate": 3.429844097995546e-05,
|
| 3056 |
+
"loss": 0.0001,
|
| 3057 |
+
"step": 3720
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 8.31,
|
| 3061 |
+
"grad_norm": 0.0019246740266680717,
|
| 3062 |
+
"learning_rate": 3.3853006681514475e-05,
|
| 3063 |
+
"loss": 0.0001,
|
| 3064 |
+
"step": 3730
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 8.33,
|
| 3068 |
+
"grad_norm": 0.0022740724962204695,
|
| 3069 |
+
"learning_rate": 3.34075723830735e-05,
|
| 3070 |
+
"loss": 0.0001,
|
| 3071 |
+
"step": 3740
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 8.35,
|
| 3075 |
+
"grad_norm": 0.0003697601496241987,
|
| 3076 |
+
"learning_rate": 3.2962138084632515e-05,
|
| 3077 |
+
"loss": 0.0001,
|
| 3078 |
+
"step": 3750
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 8.37,
|
| 3082 |
+
"grad_norm": 0.00033845697180368006,
|
| 3083 |
+
"learning_rate": 3.251670378619154e-05,
|
| 3084 |
+
"loss": 0.0001,
|
| 3085 |
+
"step": 3760
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 8.4,
|
| 3089 |
+
"grad_norm": 0.0004189737082924694,
|
| 3090 |
+
"learning_rate": 3.207126948775056e-05,
|
| 3091 |
+
"loss": 0.0001,
|
| 3092 |
+
"step": 3770
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 8.42,
|
| 3096 |
+
"grad_norm": 0.0008992131915874779,
|
| 3097 |
+
"learning_rate": 3.162583518930958e-05,
|
| 3098 |
+
"loss": 0.0004,
|
| 3099 |
+
"step": 3780
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 8.44,
|
| 3103 |
+
"grad_norm": 0.0008794433670118451,
|
| 3104 |
+
"learning_rate": 3.11804008908686e-05,
|
| 3105 |
+
"loss": 0.0001,
|
| 3106 |
+
"step": 3790
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 8.46,
|
| 3110 |
+
"grad_norm": 0.0017194098327308893,
|
| 3111 |
+
"learning_rate": 3.073496659242761e-05,
|
| 3112 |
+
"loss": 0.0002,
|
| 3113 |
+
"step": 3800
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 8.46,
|
| 3117 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3118 |
+
"eval_f1": 0.9715686274509804,
|
| 3119 |
+
"eval_loss": 0.15634377300739288,
|
| 3120 |
+
"eval_precision": 0.9715686274509804,
|
| 3121 |
+
"eval_recall": 0.9715686274509804,
|
| 3122 |
+
"eval_runtime": 17.0753,
|
| 3123 |
+
"eval_samples_per_second": 59.735,
|
| 3124 |
+
"eval_steps_per_second": 7.496,
|
| 3125 |
+
"step": 3800
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 8.49,
|
| 3129 |
+
"grad_norm": 0.0009110970422625542,
|
| 3130 |
+
"learning_rate": 3.028953229398664e-05,
|
| 3131 |
+
"loss": 0.0002,
|
| 3132 |
+
"step": 3810
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 8.51,
|
| 3136 |
+
"grad_norm": 0.0010168278822675347,
|
| 3137 |
+
"learning_rate": 2.9844097995545663e-05,
|
| 3138 |
+
"loss": 0.0001,
|
| 3139 |
+
"step": 3820
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 8.53,
|
| 3143 |
+
"grad_norm": 0.0005708567332476377,
|
| 3144 |
+
"learning_rate": 2.939866369710468e-05,
|
| 3145 |
+
"loss": 0.0001,
|
| 3146 |
+
"step": 3830
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 8.55,
|
| 3150 |
+
"grad_norm": 0.0033207752276211977,
|
| 3151 |
+
"learning_rate": 2.89532293986637e-05,
|
| 3152 |
+
"loss": 0.0001,
|
| 3153 |
+
"step": 3840
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 8.57,
|
| 3157 |
+
"grad_norm": 0.0010962020605802536,
|
| 3158 |
+
"learning_rate": 2.850779510022272e-05,
|
| 3159 |
+
"loss": 0.0001,
|
| 3160 |
+
"step": 3850
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 8.6,
|
| 3164 |
+
"grad_norm": 0.0008160584839060903,
|
| 3165 |
+
"learning_rate": 2.8062360801781738e-05,
|
| 3166 |
+
"loss": 0.0001,
|
| 3167 |
+
"step": 3860
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 8.62,
|
| 3171 |
+
"grad_norm": 0.001714337500743568,
|
| 3172 |
+
"learning_rate": 2.7616926503340758e-05,
|
| 3173 |
+
"loss": 0.0001,
|
| 3174 |
+
"step": 3870
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 8.64,
|
| 3178 |
+
"grad_norm": 0.002063535852357745,
|
| 3179 |
+
"learning_rate": 2.7171492204899778e-05,
|
| 3180 |
+
"loss": 0.0001,
|
| 3181 |
+
"step": 3880
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 8.66,
|
| 3185 |
+
"grad_norm": 0.0010866498341783881,
|
| 3186 |
+
"learning_rate": 2.67260579064588e-05,
|
| 3187 |
+
"loss": 0.0001,
|
| 3188 |
+
"step": 3890
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 8.69,
|
| 3192 |
+
"grad_norm": 0.0009104039054363966,
|
| 3193 |
+
"learning_rate": 2.6280623608017817e-05,
|
| 3194 |
+
"loss": 0.0002,
|
| 3195 |
+
"step": 3900
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 8.69,
|
| 3199 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3200 |
+
"eval_f1": 0.9715686274509804,
|
| 3201 |
+
"eval_loss": 0.15503399074077606,
|
| 3202 |
+
"eval_precision": 0.9715686274509804,
|
| 3203 |
+
"eval_recall": 0.9715686274509804,
|
| 3204 |
+
"eval_runtime": 17.3241,
|
| 3205 |
+
"eval_samples_per_second": 58.877,
|
| 3206 |
+
"eval_steps_per_second": 7.389,
|
| 3207 |
+
"step": 3900
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 8.71,
|
| 3211 |
+
"grad_norm": 0.0029284367337822914,
|
| 3212 |
+
"learning_rate": 2.583518930957684e-05,
|
| 3213 |
+
"loss": 0.007,
|
| 3214 |
+
"step": 3910
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 8.73,
|
| 3218 |
+
"grad_norm": 0.0005087658646516502,
|
| 3219 |
+
"learning_rate": 2.5389755011135856e-05,
|
| 3220 |
+
"loss": 0.0001,
|
| 3221 |
+
"step": 3920
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 8.75,
|
| 3225 |
+
"grad_norm": 0.0005650007515214384,
|
| 3226 |
+
"learning_rate": 2.494432071269488e-05,
|
| 3227 |
+
"loss": 0.0001,
|
| 3228 |
+
"step": 3930
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 8.78,
|
| 3232 |
+
"grad_norm": 0.0005321349017322063,
|
| 3233 |
+
"learning_rate": 2.44988864142539e-05,
|
| 3234 |
+
"loss": 0.0001,
|
| 3235 |
+
"step": 3940
|
| 3236 |
+
},
|
| 3237 |
+
{
|
| 3238 |
+
"epoch": 8.8,
|
| 3239 |
+
"grad_norm": 0.0008177491254173219,
|
| 3240 |
+
"learning_rate": 2.405345211581292e-05,
|
| 3241 |
+
"loss": 0.0001,
|
| 3242 |
+
"step": 3950
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 8.82,
|
| 3246 |
+
"grad_norm": 0.0005406069685705006,
|
| 3247 |
+
"learning_rate": 2.360801781737194e-05,
|
| 3248 |
+
"loss": 0.0001,
|
| 3249 |
+
"step": 3960
|
| 3250 |
+
},
|
| 3251 |
+
{
|
| 3252 |
+
"epoch": 8.84,
|
| 3253 |
+
"grad_norm": 0.000659614393953234,
|
| 3254 |
+
"learning_rate": 2.316258351893096e-05,
|
| 3255 |
+
"loss": 0.0001,
|
| 3256 |
+
"step": 3970
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 8.86,
|
| 3260 |
+
"grad_norm": 0.0004996512434445322,
|
| 3261 |
+
"learning_rate": 2.2717149220489978e-05,
|
| 3262 |
+
"loss": 0.0001,
|
| 3263 |
+
"step": 3980
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 8.89,
|
| 3267 |
+
"grad_norm": 0.0002974416420329362,
|
| 3268 |
+
"learning_rate": 2.2271714922049e-05,
|
| 3269 |
+
"loss": 0.0001,
|
| 3270 |
+
"step": 3990
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 8.91,
|
| 3274 |
+
"grad_norm": 0.0011179678840562701,
|
| 3275 |
+
"learning_rate": 2.182628062360802e-05,
|
| 3276 |
+
"loss": 0.0001,
|
| 3277 |
+
"step": 4000
|
| 3278 |
+
},
|
| 3279 |
+
{
|
| 3280 |
+
"epoch": 8.91,
|
| 3281 |
+
"eval_accuracy": 0.9705882352941176,
|
| 3282 |
+
"eval_f1": 0.9705882352941176,
|
| 3283 |
+
"eval_loss": 0.15417079627513885,
|
| 3284 |
+
"eval_precision": 0.9705882352941176,
|
| 3285 |
+
"eval_recall": 0.9705882352941176,
|
| 3286 |
+
"eval_runtime": 17.0729,
|
| 3287 |
+
"eval_samples_per_second": 59.744,
|
| 3288 |
+
"eval_steps_per_second": 7.497,
|
| 3289 |
+
"step": 4000
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 8.93,
|
| 3293 |
+
"grad_norm": 0.0005180141888558865,
|
| 3294 |
+
"learning_rate": 2.138084632516704e-05,
|
| 3295 |
+
"loss": 0.0001,
|
| 3296 |
+
"step": 4010
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 8.95,
|
| 3300 |
+
"grad_norm": 0.0005326379905454814,
|
| 3301 |
+
"learning_rate": 2.093541202672606e-05,
|
| 3302 |
+
"loss": 0.0001,
|
| 3303 |
+
"step": 4020
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 8.98,
|
| 3307 |
+
"grad_norm": 0.0006433409289456904,
|
| 3308 |
+
"learning_rate": 2.048997772828508e-05,
|
| 3309 |
+
"loss": 0.0001,
|
| 3310 |
+
"step": 4030
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 9.0,
|
| 3314 |
+
"grad_norm": 0.002777345711365342,
|
| 3315 |
+
"learning_rate": 2.00445434298441e-05,
|
| 3316 |
+
"loss": 0.0001,
|
| 3317 |
+
"step": 4040
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 9.02,
|
| 3321 |
+
"grad_norm": 0.0007074729655869305,
|
| 3322 |
+
"learning_rate": 1.959910913140312e-05,
|
| 3323 |
+
"loss": 0.0001,
|
| 3324 |
+
"step": 4050
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 9.04,
|
| 3328 |
+
"grad_norm": 0.001536020776256919,
|
| 3329 |
+
"learning_rate": 1.915367483296214e-05,
|
| 3330 |
+
"loss": 0.0001,
|
| 3331 |
+
"step": 4060
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 9.06,
|
| 3335 |
+
"grad_norm": 0.0009424517047591507,
|
| 3336 |
+
"learning_rate": 1.870824053452116e-05,
|
| 3337 |
+
"loss": 0.0001,
|
| 3338 |
+
"step": 4070
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 9.09,
|
| 3342 |
+
"grad_norm": 0.3284554183483124,
|
| 3343 |
+
"learning_rate": 1.826280623608018e-05,
|
| 3344 |
+
"loss": 0.0026,
|
| 3345 |
+
"step": 4080
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 9.11,
|
| 3349 |
+
"grad_norm": 0.00030903713195584714,
|
| 3350 |
+
"learning_rate": 1.7817371937639198e-05,
|
| 3351 |
+
"loss": 0.0001,
|
| 3352 |
+
"step": 4090
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 9.13,
|
| 3356 |
+
"grad_norm": 0.0005440358072519302,
|
| 3357 |
+
"learning_rate": 1.7371937639198218e-05,
|
| 3358 |
+
"loss": 0.0001,
|
| 3359 |
+
"step": 4100
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 9.13,
|
| 3363 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3364 |
+
"eval_f1": 0.9715686274509804,
|
| 3365 |
+
"eval_loss": 0.15382429957389832,
|
| 3366 |
+
"eval_precision": 0.9715686274509804,
|
| 3367 |
+
"eval_recall": 0.9715686274509804,
|
| 3368 |
+
"eval_runtime": 16.821,
|
| 3369 |
+
"eval_samples_per_second": 60.638,
|
| 3370 |
+
"eval_steps_per_second": 7.61,
|
| 3371 |
+
"step": 4100
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 9.15,
|
| 3375 |
+
"grad_norm": 0.0006746925064362586,
|
| 3376 |
+
"learning_rate": 1.6926503340757238e-05,
|
| 3377 |
+
"loss": 0.0001,
|
| 3378 |
+
"step": 4110
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 9.18,
|
| 3382 |
+
"grad_norm": 0.00037791216163896024,
|
| 3383 |
+
"learning_rate": 1.6481069042316257e-05,
|
| 3384 |
+
"loss": 0.0001,
|
| 3385 |
+
"step": 4120
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 9.2,
|
| 3389 |
+
"grad_norm": 0.0012918213615193963,
|
| 3390 |
+
"learning_rate": 1.603563474387528e-05,
|
| 3391 |
+
"loss": 0.0001,
|
| 3392 |
+
"step": 4130
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 9.22,
|
| 3396 |
+
"grad_norm": 0.000723692704923451,
|
| 3397 |
+
"learning_rate": 1.55902004454343e-05,
|
| 3398 |
+
"loss": 0.0001,
|
| 3399 |
+
"step": 4140
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 9.24,
|
| 3403 |
+
"grad_norm": 0.0006746066501364112,
|
| 3404 |
+
"learning_rate": 1.514476614699332e-05,
|
| 3405 |
+
"loss": 0.0001,
|
| 3406 |
+
"step": 4150
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 9.27,
|
| 3410 |
+
"grad_norm": 0.000450183724751696,
|
| 3411 |
+
"learning_rate": 1.469933184855234e-05,
|
| 3412 |
+
"loss": 0.0001,
|
| 3413 |
+
"step": 4160
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 9.29,
|
| 3417 |
+
"grad_norm": 0.0011862111277878284,
|
| 3418 |
+
"learning_rate": 1.425389755011136e-05,
|
| 3419 |
+
"loss": 0.0001,
|
| 3420 |
+
"step": 4170
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 9.31,
|
| 3424 |
+
"grad_norm": 0.0017749534454196692,
|
| 3425 |
+
"learning_rate": 1.3808463251670379e-05,
|
| 3426 |
+
"loss": 0.0001,
|
| 3427 |
+
"step": 4180
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 9.33,
|
| 3431 |
+
"grad_norm": 0.0003237236524000764,
|
| 3432 |
+
"learning_rate": 1.33630289532294e-05,
|
| 3433 |
+
"loss": 0.0001,
|
| 3434 |
+
"step": 4190
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 9.35,
|
| 3438 |
+
"grad_norm": 0.0004740317235700786,
|
| 3439 |
+
"learning_rate": 1.291759465478842e-05,
|
| 3440 |
+
"loss": 0.0001,
|
| 3441 |
+
"step": 4200
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 9.35,
|
| 3445 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3446 |
+
"eval_f1": 0.9715686274509804,
|
| 3447 |
+
"eval_loss": 0.15357248485088348,
|
| 3448 |
+
"eval_precision": 0.9715686274509804,
|
| 3449 |
+
"eval_recall": 0.9715686274509804,
|
| 3450 |
+
"eval_runtime": 17.097,
|
| 3451 |
+
"eval_samples_per_second": 59.659,
|
| 3452 |
+
"eval_steps_per_second": 7.487,
|
| 3453 |
+
"step": 4200
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 9.38,
|
| 3457 |
+
"grad_norm": 0.0004047084948979318,
|
| 3458 |
+
"learning_rate": 1.247216035634744e-05,
|
| 3459 |
+
"loss": 0.0001,
|
| 3460 |
+
"step": 4210
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 9.4,
|
| 3464 |
+
"grad_norm": 0.0028567886911332607,
|
| 3465 |
+
"learning_rate": 1.202672605790646e-05,
|
| 3466 |
+
"loss": 0.0001,
|
| 3467 |
+
"step": 4220
|
| 3468 |
+
},
|
| 3469 |
+
{
|
| 3470 |
+
"epoch": 9.42,
|
| 3471 |
+
"grad_norm": 0.0005680415779352188,
|
| 3472 |
+
"learning_rate": 1.158129175946548e-05,
|
| 3473 |
+
"loss": 0.0001,
|
| 3474 |
+
"step": 4230
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 9.44,
|
| 3478 |
+
"grad_norm": 0.00213377526961267,
|
| 3479 |
+
"learning_rate": 1.11358574610245e-05,
|
| 3480 |
+
"loss": 0.0001,
|
| 3481 |
+
"step": 4240
|
| 3482 |
+
},
|
| 3483 |
+
{
|
| 3484 |
+
"epoch": 9.47,
|
| 3485 |
+
"grad_norm": 0.001148115610703826,
|
| 3486 |
+
"learning_rate": 1.069042316258352e-05,
|
| 3487 |
+
"loss": 0.0001,
|
| 3488 |
+
"step": 4250
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 9.49,
|
| 3492 |
+
"grad_norm": 0.00045941799180582166,
|
| 3493 |
+
"learning_rate": 1.024498886414254e-05,
|
| 3494 |
+
"loss": 0.0001,
|
| 3495 |
+
"step": 4260
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 9.51,
|
| 3499 |
+
"grad_norm": 0.0024906108155846596,
|
| 3500 |
+
"learning_rate": 9.79955456570156e-06,
|
| 3501 |
+
"loss": 0.0001,
|
| 3502 |
+
"step": 4270
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 9.53,
|
| 3506 |
+
"grad_norm": 0.0004669167974498123,
|
| 3507 |
+
"learning_rate": 9.35412026726058e-06,
|
| 3508 |
+
"loss": 0.0001,
|
| 3509 |
+
"step": 4280
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 9.55,
|
| 3513 |
+
"grad_norm": 0.0003813539515249431,
|
| 3514 |
+
"learning_rate": 8.908685968819599e-06,
|
| 3515 |
+
"loss": 0.0001,
|
| 3516 |
+
"step": 4290
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 9.58,
|
| 3520 |
+
"grad_norm": 0.00048394210170954466,
|
| 3521 |
+
"learning_rate": 8.463251670378619e-06,
|
| 3522 |
+
"loss": 0.0001,
|
| 3523 |
+
"step": 4300
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 9.58,
|
| 3527 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3528 |
+
"eval_f1": 0.9715686274509804,
|
| 3529 |
+
"eval_loss": 0.15336920320987701,
|
| 3530 |
+
"eval_precision": 0.9715686274509804,
|
| 3531 |
+
"eval_recall": 0.9715686274509804,
|
| 3532 |
+
"eval_runtime": 16.9378,
|
| 3533 |
+
"eval_samples_per_second": 60.22,
|
| 3534 |
+
"eval_steps_per_second": 7.557,
|
| 3535 |
+
"step": 4300
|
| 3536 |
+
},
|
| 3537 |
+
{
|
| 3538 |
+
"epoch": 9.6,
|
| 3539 |
+
"grad_norm": 0.001094264443963766,
|
| 3540 |
+
"learning_rate": 8.01781737193764e-06,
|
| 3541 |
+
"loss": 0.0001,
|
| 3542 |
+
"step": 4310
|
| 3543 |
+
},
|
| 3544 |
+
{
|
| 3545 |
+
"epoch": 9.62,
|
| 3546 |
+
"grad_norm": 0.0007720951689407229,
|
| 3547 |
+
"learning_rate": 7.57238307349666e-06,
|
| 3548 |
+
"loss": 0.0001,
|
| 3549 |
+
"step": 4320
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 9.64,
|
| 3553 |
+
"grad_norm": 0.0010363436304032803,
|
| 3554 |
+
"learning_rate": 7.12694877505568e-06,
|
| 3555 |
+
"loss": 0.0001,
|
| 3556 |
+
"step": 4330
|
| 3557 |
+
},
|
| 3558 |
+
{
|
| 3559 |
+
"epoch": 9.67,
|
| 3560 |
+
"grad_norm": 0.0006155350711196661,
|
| 3561 |
+
"learning_rate": 6.6815144766147e-06,
|
| 3562 |
+
"loss": 0.0001,
|
| 3563 |
+
"step": 4340
|
| 3564 |
+
},
|
| 3565 |
+
{
|
| 3566 |
+
"epoch": 9.69,
|
| 3567 |
+
"grad_norm": 0.002117099007591605,
|
| 3568 |
+
"learning_rate": 6.23608017817372e-06,
|
| 3569 |
+
"loss": 0.0001,
|
| 3570 |
+
"step": 4350
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 9.71,
|
| 3574 |
+
"grad_norm": 0.00035223804297856987,
|
| 3575 |
+
"learning_rate": 5.79064587973274e-06,
|
| 3576 |
+
"loss": 0.0001,
|
| 3577 |
+
"step": 4360
|
| 3578 |
+
},
|
| 3579 |
+
{
|
| 3580 |
+
"epoch": 9.73,
|
| 3581 |
+
"grad_norm": 0.0008616923005320132,
|
| 3582 |
+
"learning_rate": 5.34521158129176e-06,
|
| 3583 |
+
"loss": 0.0001,
|
| 3584 |
+
"step": 4370
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 9.76,
|
| 3588 |
+
"grad_norm": 0.0003698187065310776,
|
| 3589 |
+
"learning_rate": 4.89977728285078e-06,
|
| 3590 |
+
"loss": 0.0001,
|
| 3591 |
+
"step": 4380
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 9.78,
|
| 3595 |
+
"grad_norm": 0.0008464885177090764,
|
| 3596 |
+
"learning_rate": 4.4543429844097995e-06,
|
| 3597 |
+
"loss": 0.0001,
|
| 3598 |
+
"step": 4390
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 9.8,
|
| 3602 |
+
"grad_norm": 0.000521197565831244,
|
| 3603 |
+
"learning_rate": 4.00890868596882e-06,
|
| 3604 |
+
"loss": 0.0001,
|
| 3605 |
+
"step": 4400
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 9.8,
|
| 3609 |
+
"eval_accuracy": 0.9715686274509804,
|
| 3610 |
+
"eval_f1": 0.9715686274509804,
|
| 3611 |
+
"eval_loss": 0.15329474210739136,
|
| 3612 |
+
"eval_precision": 0.9715686274509804,
|
| 3613 |
+
"eval_recall": 0.9715686274509804,
|
| 3614 |
+
"eval_runtime": 17.0026,
|
| 3615 |
+
"eval_samples_per_second": 59.991,
|
| 3616 |
+
"eval_steps_per_second": 7.528,
|
| 3617 |
+
"step": 4400
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 9.82,
|
| 3621 |
+
"grad_norm": 0.00078478833893314,
|
| 3622 |
+
"learning_rate": 3.56347438752784e-06,
|
| 3623 |
+
"loss": 0.0001,
|
| 3624 |
+
"step": 4410
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 9.84,
|
| 3628 |
+
"grad_norm": 0.00044067302951589227,
|
| 3629 |
+
"learning_rate": 3.11804008908686e-06,
|
| 3630 |
+
"loss": 0.0001,
|
| 3631 |
+
"step": 4420
|
| 3632 |
+
},
|
| 3633 |
+
{
|
| 3634 |
+
"epoch": 9.87,
|
| 3635 |
+
"grad_norm": 0.0005106424796395004,
|
| 3636 |
+
"learning_rate": 2.67260579064588e-06,
|
| 3637 |
+
"loss": 0.0001,
|
| 3638 |
+
"step": 4430
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 9.89,
|
| 3642 |
+
"grad_norm": 0.0003234909090679139,
|
| 3643 |
+
"learning_rate": 2.2271714922048998e-06,
|
| 3644 |
+
"loss": 0.0001,
|
| 3645 |
+
"step": 4440
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 9.91,
|
| 3649 |
+
"grad_norm": 0.0003544181527104229,
|
| 3650 |
+
"learning_rate": 1.78173719376392e-06,
|
| 3651 |
+
"loss": 0.0001,
|
| 3652 |
+
"step": 4450
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 9.93,
|
| 3656 |
+
"grad_norm": 0.00040398509008809924,
|
| 3657 |
+
"learning_rate": 1.33630289532294e-06,
|
| 3658 |
+
"loss": 0.0001,
|
| 3659 |
+
"step": 4460
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 9.96,
|
| 3663 |
+
"grad_norm": 0.0009952255059033632,
|
| 3664 |
+
"learning_rate": 8.9086859688196e-07,
|
| 3665 |
+
"loss": 0.0001,
|
| 3666 |
+
"step": 4470
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 9.98,
|
| 3670 |
+
"grad_norm": 0.0003712301841005683,
|
| 3671 |
+
"learning_rate": 4.4543429844098e-07,
|
| 3672 |
+
"loss": 0.0001,
|
| 3673 |
+
"step": 4480
|
| 3674 |
+
},
|
| 3675 |
+
{
|
| 3676 |
+
"epoch": 10.0,
|
| 3677 |
+
"grad_norm": 0.0012604963267222047,
|
| 3678 |
+
"learning_rate": 0.0,
|
| 3679 |
+
"loss": 0.0099,
|
| 3680 |
+
"step": 4490
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 10.0,
|
| 3684 |
+
"step": 4490,
|
| 3685 |
+
"total_flos": 5.562769847811564e+18,
|
| 3686 |
+
"train_loss": 0.025327244219600944,
|
| 3687 |
+
"train_runtime": 3846.586,
|
| 3688 |
+
"train_samples_per_second": 18.637,
|
| 3689 |
+
"train_steps_per_second": 1.167
|
| 3690 |
+
}
|
| 3691 |
+
],
|
| 3692 |
+
"logging_steps": 10,
|
| 3693 |
+
"max_steps": 4490,
|
| 3694 |
+
"num_input_tokens_seen": 0,
|
| 3695 |
+
"num_train_epochs": 10,
|
| 3696 |
+
"save_steps": 100,
|
| 3697 |
+
"total_flos": 5.562769847811564e+18,
|
| 3698 |
+
"train_batch_size": 16,
|
| 3699 |
+
"trial_name": null,
|
| 3700 |
+
"trial_params": null
|
| 3701 |
+
}
|