Commit
•
e00ee5a
1
Parent(s):
2ea3aa7
Model save
Browse files- README.md +5 -5
- all_results.json +16 -0
- eval_results.json +11 -0
- train_results.json +8 -0
- trainer_state.json +3930 -0
README.md
CHANGED
|
@@ -23,11 +23,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 23 |
|
| 24 |
This model is a fine-tuned version of [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k) on the medmnist-v2 dataset.
|
| 25 |
It achieves the following results on the evaluation set:
|
| 26 |
-
- Loss: 0.
|
| 27 |
-
- Accuracy: 0.
|
| 28 |
-
- Precision: 0.
|
| 29 |
-
- Recall: 0.
|
| 30 |
-
- F1: 0.
|
| 31 |
|
| 32 |
## Model description
|
| 33 |
|
|
|
|
| 23 |
|
| 24 |
This model is a fine-tuned version of [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k) on the medmnist-v2 dataset.
|
| 25 |
It achieves the following results on the evaluation set:
|
| 26 |
+
- Loss: 0.2372
|
| 27 |
+
- Accuracy: 0.9329
|
| 28 |
+
- Precision: 0.9416
|
| 29 |
+
- Recall: 0.9296
|
| 30 |
+
- F1: 0.9340
|
| 31 |
|
| 32 |
## Model description
|
| 33 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.99,
|
| 3 |
+
"eval_accuracy": 0.9328945888176398,
|
| 4 |
+
"eval_f1": 0.9340102214756545,
|
| 5 |
+
"eval_loss": 0.23724627494812012,
|
| 6 |
+
"eval_precision": 0.9416197933012,
|
| 7 |
+
"eval_recall": 0.9295532071921557,
|
| 8 |
+
"eval_runtime": 150.877,
|
| 9 |
+
"eval_samples_per_second": 117.831,
|
| 10 |
+
"eval_steps_per_second": 7.37,
|
| 11 |
+
"total_flos": 2.695096772600047e+19,
|
| 12 |
+
"train_loss": 0.5508582998205114,
|
| 13 |
+
"train_runtime": 6179.724,
|
| 14 |
+
"train_samples_per_second": 55.959,
|
| 15 |
+
"train_steps_per_second": 0.874
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.99,
|
| 3 |
+
"eval_accuracy": 0.9328945888176398,
|
| 4 |
+
"eval_f1": 0.9340102214756545,
|
| 5 |
+
"eval_loss": 0.23724627494812012,
|
| 6 |
+
"eval_precision": 0.9416197933012,
|
| 7 |
+
"eval_recall": 0.9295532071921557,
|
| 8 |
+
"eval_runtime": 150.877,
|
| 9 |
+
"eval_samples_per_second": 117.831,
|
| 10 |
+
"eval_steps_per_second": 7.37
|
| 11 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.99,
|
| 3 |
+
"total_flos": 2.695096772600047e+19,
|
| 4 |
+
"train_loss": 0.5508582998205114,
|
| 5 |
+
"train_runtime": 6179.724,
|
| 6 |
+
"train_samples_per_second": 55.959,
|
| 7 |
+
"train_steps_per_second": 0.874
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3930 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.986134647974118,
|
| 3 |
+
"best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-lora-medmnistv2/checkpoint-4864",
|
| 4 |
+
"epoch": 9.990749306197966,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 5400,
|
| 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": 3.66642165184021,
|
| 14 |
+
"learning_rate": 0.004990740740740741,
|
| 15 |
+
"loss": 1.7617,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04,
|
| 20 |
+
"grad_norm": 5.545452117919922,
|
| 21 |
+
"learning_rate": 0.004982407407407408,
|
| 22 |
+
"loss": 1.1989,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.06,
|
| 27 |
+
"grad_norm": 5.292081356048584,
|
| 28 |
+
"learning_rate": 0.004973148148148148,
|
| 29 |
+
"loss": 1.0735,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07,
|
| 34 |
+
"grad_norm": 3.0206360816955566,
|
| 35 |
+
"learning_rate": 0.004963888888888889,
|
| 36 |
+
"loss": 1.1936,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.09,
|
| 41 |
+
"grad_norm": 3.947781562805176,
|
| 42 |
+
"learning_rate": 0.004954629629629629,
|
| 43 |
+
"loss": 1.0373,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.11,
|
| 48 |
+
"grad_norm": 1.839845895767212,
|
| 49 |
+
"learning_rate": 0.004945370370370371,
|
| 50 |
+
"loss": 0.9221,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.13,
|
| 55 |
+
"grad_norm": 2.843251943588257,
|
| 56 |
+
"learning_rate": 0.004936111111111112,
|
| 57 |
+
"loss": 0.8724,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.15,
|
| 62 |
+
"grad_norm": 3.0122690200805664,
|
| 63 |
+
"learning_rate": 0.0049268518518518515,
|
| 64 |
+
"loss": 0.9201,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.17,
|
| 69 |
+
"grad_norm": 3.403461456298828,
|
| 70 |
+
"learning_rate": 0.004917592592592593,
|
| 71 |
+
"loss": 0.9221,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.19,
|
| 76 |
+
"grad_norm": 2.4073173999786377,
|
| 77 |
+
"learning_rate": 0.004908333333333333,
|
| 78 |
+
"loss": 0.8766,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2,
|
| 83 |
+
"grad_norm": 1.793009877204895,
|
| 84 |
+
"learning_rate": 0.004899074074074074,
|
| 85 |
+
"loss": 0.8742,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.22,
|
| 90 |
+
"grad_norm": 1.8483041524887085,
|
| 91 |
+
"learning_rate": 0.004889814814814815,
|
| 92 |
+
"loss": 0.8097,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.24,
|
| 97 |
+
"grad_norm": 1.997031807899475,
|
| 98 |
+
"learning_rate": 0.0048805555555555555,
|
| 99 |
+
"loss": 0.888,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.26,
|
| 104 |
+
"grad_norm": 3.958829641342163,
|
| 105 |
+
"learning_rate": 0.004871296296296296,
|
| 106 |
+
"loss": 0.8276,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.28,
|
| 111 |
+
"grad_norm": 2.8835649490356445,
|
| 112 |
+
"learning_rate": 0.004862037037037037,
|
| 113 |
+
"loss": 0.8469,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.3,
|
| 118 |
+
"grad_norm": 2.2343573570251465,
|
| 119 |
+
"learning_rate": 0.004852777777777778,
|
| 120 |
+
"loss": 0.8354,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.31,
|
| 125 |
+
"grad_norm": 1.8973902463912964,
|
| 126 |
+
"learning_rate": 0.004843518518518519,
|
| 127 |
+
"loss": 0.7974,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.33,
|
| 132 |
+
"grad_norm": 2.303403854370117,
|
| 133 |
+
"learning_rate": 0.0048342592592592595,
|
| 134 |
+
"loss": 0.7531,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.35,
|
| 139 |
+
"grad_norm": 2.833897352218628,
|
| 140 |
+
"learning_rate": 0.004825,
|
| 141 |
+
"loss": 0.7917,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.37,
|
| 146 |
+
"grad_norm": 2.833991527557373,
|
| 147 |
+
"learning_rate": 0.004815740740740741,
|
| 148 |
+
"loss": 0.9165,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.39,
|
| 153 |
+
"grad_norm": 2.109811305999756,
|
| 154 |
+
"learning_rate": 0.004806481481481482,
|
| 155 |
+
"loss": 0.7347,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.41,
|
| 160 |
+
"grad_norm": 4.833127498626709,
|
| 161 |
+
"learning_rate": 0.004797222222222223,
|
| 162 |
+
"loss": 0.8852,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.43,
|
| 167 |
+
"grad_norm": 6.393899440765381,
|
| 168 |
+
"learning_rate": 0.004787962962962963,
|
| 169 |
+
"loss": 0.7416,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.44,
|
| 174 |
+
"grad_norm": 2.243924856185913,
|
| 175 |
+
"learning_rate": 0.004778703703703704,
|
| 176 |
+
"loss": 0.7616,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.46,
|
| 181 |
+
"grad_norm": 2.55934739112854,
|
| 182 |
+
"learning_rate": 0.004769444444444444,
|
| 183 |
+
"loss": 0.6881,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.48,
|
| 188 |
+
"grad_norm": 2.707200765609741,
|
| 189 |
+
"learning_rate": 0.004760185185185185,
|
| 190 |
+
"loss": 0.7597,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5,
|
| 195 |
+
"grad_norm": 3.238736867904663,
|
| 196 |
+
"learning_rate": 0.004750925925925926,
|
| 197 |
+
"loss": 0.6598,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.52,
|
| 202 |
+
"grad_norm": 1.6335750818252563,
|
| 203 |
+
"learning_rate": 0.004741666666666667,
|
| 204 |
+
"loss": 0.6988,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.54,
|
| 209 |
+
"grad_norm": 1.6415876150131226,
|
| 210 |
+
"learning_rate": 0.004732407407407407,
|
| 211 |
+
"loss": 0.7615,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.56,
|
| 216 |
+
"grad_norm": 2.09512996673584,
|
| 217 |
+
"learning_rate": 0.004723148148148148,
|
| 218 |
+
"loss": 0.7345,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.57,
|
| 223 |
+
"grad_norm": 2.1141433715820312,
|
| 224 |
+
"learning_rate": 0.004713888888888889,
|
| 225 |
+
"loss": 0.7503,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.59,
|
| 230 |
+
"grad_norm": 1.1782424449920654,
|
| 231 |
+
"learning_rate": 0.00470462962962963,
|
| 232 |
+
"loss": 0.6997,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.61,
|
| 237 |
+
"grad_norm": 2.7585694789886475,
|
| 238 |
+
"learning_rate": 0.004695370370370371,
|
| 239 |
+
"loss": 0.7703,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.63,
|
| 244 |
+
"grad_norm": 1.3906846046447754,
|
| 245 |
+
"learning_rate": 0.004686111111111111,
|
| 246 |
+
"loss": 0.6595,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.65,
|
| 251 |
+
"grad_norm": 1.9824657440185547,
|
| 252 |
+
"learning_rate": 0.004676851851851852,
|
| 253 |
+
"loss": 0.612,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.67,
|
| 258 |
+
"grad_norm": 3.896270990371704,
|
| 259 |
+
"learning_rate": 0.004667592592592593,
|
| 260 |
+
"loss": 0.8571,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.68,
|
| 265 |
+
"grad_norm": 1.860967993736267,
|
| 266 |
+
"learning_rate": 0.004658333333333333,
|
| 267 |
+
"loss": 0.774,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7,
|
| 272 |
+
"grad_norm": 1.4278061389923096,
|
| 273 |
+
"learning_rate": 0.004649074074074074,
|
| 274 |
+
"loss": 0.7693,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.72,
|
| 279 |
+
"grad_norm": 4.371230125427246,
|
| 280 |
+
"learning_rate": 0.004639814814814815,
|
| 281 |
+
"loss": 0.6989,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.74,
|
| 286 |
+
"grad_norm": 1.6411677598953247,
|
| 287 |
+
"learning_rate": 0.004630555555555555,
|
| 288 |
+
"loss": 0.8566,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.76,
|
| 293 |
+
"grad_norm": 3.412655830383301,
|
| 294 |
+
"learning_rate": 0.004621296296296296,
|
| 295 |
+
"loss": 0.6902,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.78,
|
| 300 |
+
"grad_norm": 1.9510608911514282,
|
| 301 |
+
"learning_rate": 0.004612037037037038,
|
| 302 |
+
"loss": 0.8284,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8,
|
| 307 |
+
"grad_norm": 1.8800582885742188,
|
| 308 |
+
"learning_rate": 0.004602777777777778,
|
| 309 |
+
"loss": 0.6096,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.81,
|
| 314 |
+
"grad_norm": 2.7876181602478027,
|
| 315 |
+
"learning_rate": 0.0045935185185185185,
|
| 316 |
+
"loss": 0.7293,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.83,
|
| 321 |
+
"grad_norm": 2.0592384338378906,
|
| 322 |
+
"learning_rate": 0.004584259259259259,
|
| 323 |
+
"loss": 0.756,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.85,
|
| 328 |
+
"grad_norm": 3.5372824668884277,
|
| 329 |
+
"learning_rate": 0.004575,
|
| 330 |
+
"loss": 0.6865,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.87,
|
| 335 |
+
"grad_norm": 1.7686892747879028,
|
| 336 |
+
"learning_rate": 0.004565740740740741,
|
| 337 |
+
"loss": 0.7477,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.89,
|
| 342 |
+
"grad_norm": 2.8619141578674316,
|
| 343 |
+
"learning_rate": 0.004556481481481482,
|
| 344 |
+
"loss": 0.6861,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.91,
|
| 349 |
+
"grad_norm": 2.4887492656707764,
|
| 350 |
+
"learning_rate": 0.0045472222222222225,
|
| 351 |
+
"loss": 0.7623,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.93,
|
| 356 |
+
"grad_norm": 1.9150561094284058,
|
| 357 |
+
"learning_rate": 0.004537962962962963,
|
| 358 |
+
"loss": 0.7462,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.94,
|
| 363 |
+
"grad_norm": 2.1251327991485596,
|
| 364 |
+
"learning_rate": 0.004528703703703704,
|
| 365 |
+
"loss": 0.673,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.96,
|
| 370 |
+
"grad_norm": 2.338467597961426,
|
| 371 |
+
"learning_rate": 0.004519444444444444,
|
| 372 |
+
"loss": 0.6898,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.98,
|
| 377 |
+
"grad_norm": 2.199749708175659,
|
| 378 |
+
"learning_rate": 0.004510185185185186,
|
| 379 |
+
"loss": 0.6776,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.0,
|
| 384 |
+
"grad_norm": 1.5489593744277954,
|
| 385 |
+
"learning_rate": 0.0045009259259259264,
|
| 386 |
+
"loss": 0.6786,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.0,
|
| 391 |
+
"eval_accuracy": 0.9339084886766292,
|
| 392 |
+
"eval_f1": 0.938537564429046,
|
| 393 |
+
"eval_loss": 0.17764028906822205,
|
| 394 |
+
"eval_precision": 0.9507286953627705,
|
| 395 |
+
"eval_recall": 0.934120062770795,
|
| 396 |
+
"eval_runtime": 54.7942,
|
| 397 |
+
"eval_samples_per_second": 118.462,
|
| 398 |
+
"eval_steps_per_second": 7.41,
|
| 399 |
+
"step": 540
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.02,
|
| 403 |
+
"grad_norm": 2.148123264312744,
|
| 404 |
+
"learning_rate": 0.004491666666666666,
|
| 405 |
+
"loss": 0.592,
|
| 406 |
+
"step": 550
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.04,
|
| 410 |
+
"grad_norm": 3.1778488159179688,
|
| 411 |
+
"learning_rate": 0.004482407407407407,
|
| 412 |
+
"loss": 0.93,
|
| 413 |
+
"step": 560
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.05,
|
| 417 |
+
"grad_norm": 1.9821240901947021,
|
| 418 |
+
"learning_rate": 0.004473148148148149,
|
| 419 |
+
"loss": 0.8283,
|
| 420 |
+
"step": 570
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.07,
|
| 424 |
+
"grad_norm": 1.9005764722824097,
|
| 425 |
+
"learning_rate": 0.004463888888888889,
|
| 426 |
+
"loss": 0.7101,
|
| 427 |
+
"step": 580
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.09,
|
| 431 |
+
"grad_norm": 2.126354217529297,
|
| 432 |
+
"learning_rate": 0.0044546296296296296,
|
| 433 |
+
"loss": 0.6671,
|
| 434 |
+
"step": 590
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.11,
|
| 438 |
+
"grad_norm": 2.0805599689483643,
|
| 439 |
+
"learning_rate": 0.00444537037037037,
|
| 440 |
+
"loss": 0.6665,
|
| 441 |
+
"step": 600
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.13,
|
| 445 |
+
"grad_norm": 1.5277152061462402,
|
| 446 |
+
"learning_rate": 0.004436111111111111,
|
| 447 |
+
"loss": 0.7092,
|
| 448 |
+
"step": 610
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 1.15,
|
| 452 |
+
"grad_norm": 4.088949680328369,
|
| 453 |
+
"learning_rate": 0.004426851851851852,
|
| 454 |
+
"loss": 0.8679,
|
| 455 |
+
"step": 620
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.17,
|
| 459 |
+
"grad_norm": 4.5228376388549805,
|
| 460 |
+
"learning_rate": 0.004417592592592593,
|
| 461 |
+
"loss": 0.7648,
|
| 462 |
+
"step": 630
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.18,
|
| 466 |
+
"grad_norm": 4.287468910217285,
|
| 467 |
+
"learning_rate": 0.0044083333333333335,
|
| 468 |
+
"loss": 0.725,
|
| 469 |
+
"step": 640
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 1.2,
|
| 473 |
+
"grad_norm": 3.0292155742645264,
|
| 474 |
+
"learning_rate": 0.004399074074074074,
|
| 475 |
+
"loss": 0.7416,
|
| 476 |
+
"step": 650
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.22,
|
| 480 |
+
"grad_norm": 2.364569664001465,
|
| 481 |
+
"learning_rate": 0.004389814814814815,
|
| 482 |
+
"loss": 0.7576,
|
| 483 |
+
"step": 660
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.24,
|
| 487 |
+
"grad_norm": 1.8046070337295532,
|
| 488 |
+
"learning_rate": 0.004380555555555555,
|
| 489 |
+
"loss": 0.6917,
|
| 490 |
+
"step": 670
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 1.26,
|
| 494 |
+
"grad_norm": 3.8957250118255615,
|
| 495 |
+
"learning_rate": 0.004371296296296297,
|
| 496 |
+
"loss": 0.6671,
|
| 497 |
+
"step": 680
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.28,
|
| 501 |
+
"grad_norm": 2.4471426010131836,
|
| 502 |
+
"learning_rate": 0.0043620370370370375,
|
| 503 |
+
"loss": 0.6749,
|
| 504 |
+
"step": 690
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.3,
|
| 508 |
+
"grad_norm": 1.9041227102279663,
|
| 509 |
+
"learning_rate": 0.0043527777777777775,
|
| 510 |
+
"loss": 0.7423,
|
| 511 |
+
"step": 700
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.31,
|
| 515 |
+
"grad_norm": 3.001101016998291,
|
| 516 |
+
"learning_rate": 0.004343518518518519,
|
| 517 |
+
"loss": 0.7775,
|
| 518 |
+
"step": 710
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.33,
|
| 522 |
+
"grad_norm": 2.1323304176330566,
|
| 523 |
+
"learning_rate": 0.004334259259259259,
|
| 524 |
+
"loss": 0.6961,
|
| 525 |
+
"step": 720
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 1.35,
|
| 529 |
+
"grad_norm": 3.0839684009552,
|
| 530 |
+
"learning_rate": 0.004325,
|
| 531 |
+
"loss": 0.6791,
|
| 532 |
+
"step": 730
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 1.37,
|
| 536 |
+
"grad_norm": 1.924004316329956,
|
| 537 |
+
"learning_rate": 0.004315740740740741,
|
| 538 |
+
"loss": 0.663,
|
| 539 |
+
"step": 740
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 1.39,
|
| 543 |
+
"grad_norm": 1.854284644126892,
|
| 544 |
+
"learning_rate": 0.0043064814814814814,
|
| 545 |
+
"loss": 0.706,
|
| 546 |
+
"step": 750
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.41,
|
| 550 |
+
"grad_norm": 2.4816770553588867,
|
| 551 |
+
"learning_rate": 0.004297222222222222,
|
| 552 |
+
"loss": 0.6623,
|
| 553 |
+
"step": 760
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 1.42,
|
| 557 |
+
"grad_norm": 1.8469676971435547,
|
| 558 |
+
"learning_rate": 0.004287962962962963,
|
| 559 |
+
"loss": 0.6025,
|
| 560 |
+
"step": 770
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.44,
|
| 564 |
+
"grad_norm": 1.6703932285308838,
|
| 565 |
+
"learning_rate": 0.004278703703703704,
|
| 566 |
+
"loss": 0.7209,
|
| 567 |
+
"step": 780
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.46,
|
| 571 |
+
"grad_norm": 1.3802216053009033,
|
| 572 |
+
"learning_rate": 0.004269444444444445,
|
| 573 |
+
"loss": 0.6823,
|
| 574 |
+
"step": 790
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.48,
|
| 578 |
+
"grad_norm": 2.1210741996765137,
|
| 579 |
+
"learning_rate": 0.004260185185185185,
|
| 580 |
+
"loss": 0.6577,
|
| 581 |
+
"step": 800
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.5,
|
| 585 |
+
"grad_norm": 1.841590404510498,
|
| 586 |
+
"learning_rate": 0.004250925925925926,
|
| 587 |
+
"loss": 0.6049,
|
| 588 |
+
"step": 810
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 1.52,
|
| 592 |
+
"grad_norm": 2.4515421390533447,
|
| 593 |
+
"learning_rate": 0.004241666666666667,
|
| 594 |
+
"loss": 0.6555,
|
| 595 |
+
"step": 820
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 1.54,
|
| 599 |
+
"grad_norm": 2.086291551589966,
|
| 600 |
+
"learning_rate": 0.004232407407407408,
|
| 601 |
+
"loss": 0.7836,
|
| 602 |
+
"step": 830
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.55,
|
| 606 |
+
"grad_norm": 1.714977741241455,
|
| 607 |
+
"learning_rate": 0.004223148148148149,
|
| 608 |
+
"loss": 0.6868,
|
| 609 |
+
"step": 840
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.57,
|
| 613 |
+
"grad_norm": 1.9800652265548706,
|
| 614 |
+
"learning_rate": 0.0042138888888888885,
|
| 615 |
+
"loss": 0.7313,
|
| 616 |
+
"step": 850
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.59,
|
| 620 |
+
"grad_norm": 2.726518154144287,
|
| 621 |
+
"learning_rate": 0.00420462962962963,
|
| 622 |
+
"loss": 0.6946,
|
| 623 |
+
"step": 860
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.61,
|
| 627 |
+
"grad_norm": 2.3700504302978516,
|
| 628 |
+
"learning_rate": 0.00419537037037037,
|
| 629 |
+
"loss": 0.7765,
|
| 630 |
+
"step": 870
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.63,
|
| 634 |
+
"grad_norm": 3.294623851776123,
|
| 635 |
+
"learning_rate": 0.004186111111111111,
|
| 636 |
+
"loss": 0.7001,
|
| 637 |
+
"step": 880
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.65,
|
| 641 |
+
"grad_norm": 1.3391900062561035,
|
| 642 |
+
"learning_rate": 0.004176851851851852,
|
| 643 |
+
"loss": 0.7045,
|
| 644 |
+
"step": 890
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.67,
|
| 648 |
+
"grad_norm": 3.3717267513275146,
|
| 649 |
+
"learning_rate": 0.0041675925925925925,
|
| 650 |
+
"loss": 0.7396,
|
| 651 |
+
"step": 900
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.68,
|
| 655 |
+
"grad_norm": 2.7791171073913574,
|
| 656 |
+
"learning_rate": 0.004158333333333333,
|
| 657 |
+
"loss": 0.6946,
|
| 658 |
+
"step": 910
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.7,
|
| 662 |
+
"grad_norm": 2.9420201778411865,
|
| 663 |
+
"learning_rate": 0.004149074074074074,
|
| 664 |
+
"loss": 0.7215,
|
| 665 |
+
"step": 920
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.72,
|
| 669 |
+
"grad_norm": 2.337606191635132,
|
| 670 |
+
"learning_rate": 0.004139814814814815,
|
| 671 |
+
"loss": 0.6967,
|
| 672 |
+
"step": 930
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.74,
|
| 676 |
+
"grad_norm": 2.43172025680542,
|
| 677 |
+
"learning_rate": 0.004130555555555556,
|
| 678 |
+
"loss": 0.7989,
|
| 679 |
+
"step": 940
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.76,
|
| 683 |
+
"grad_norm": 1.683398723602295,
|
| 684 |
+
"learning_rate": 0.0041212962962962965,
|
| 685 |
+
"loss": 0.6185,
|
| 686 |
+
"step": 950
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.78,
|
| 690 |
+
"grad_norm": 2.1114211082458496,
|
| 691 |
+
"learning_rate": 0.004112037037037037,
|
| 692 |
+
"loss": 0.7013,
|
| 693 |
+
"step": 960
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.79,
|
| 697 |
+
"grad_norm": 2.566915512084961,
|
| 698 |
+
"learning_rate": 0.004102777777777778,
|
| 699 |
+
"loss": 0.6091,
|
| 700 |
+
"step": 970
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.81,
|
| 704 |
+
"grad_norm": 4.286575794219971,
|
| 705 |
+
"learning_rate": 0.004093518518518519,
|
| 706 |
+
"loss": 0.7474,
|
| 707 |
+
"step": 980
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.83,
|
| 711 |
+
"grad_norm": 1.9286409616470337,
|
| 712 |
+
"learning_rate": 0.004084259259259259,
|
| 713 |
+
"loss": 0.7089,
|
| 714 |
+
"step": 990
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.85,
|
| 718 |
+
"grad_norm": 2.5214929580688477,
|
| 719 |
+
"learning_rate": 0.004075,
|
| 720 |
+
"loss": 0.7412,
|
| 721 |
+
"step": 1000
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.87,
|
| 725 |
+
"grad_norm": 2.4169070720672607,
|
| 726 |
+
"learning_rate": 0.004065740740740741,
|
| 727 |
+
"loss": 0.826,
|
| 728 |
+
"step": 1010
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.89,
|
| 732 |
+
"grad_norm": 2.7352206707000732,
|
| 733 |
+
"learning_rate": 0.004056481481481481,
|
| 734 |
+
"loss": 0.7868,
|
| 735 |
+
"step": 1020
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.91,
|
| 739 |
+
"grad_norm": 1.969298243522644,
|
| 740 |
+
"learning_rate": 0.004047222222222222,
|
| 741 |
+
"loss": 0.6795,
|
| 742 |
+
"step": 1030
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.92,
|
| 746 |
+
"grad_norm": 3.015155792236328,
|
| 747 |
+
"learning_rate": 0.004037962962962964,
|
| 748 |
+
"loss": 0.7852,
|
| 749 |
+
"step": 1040
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.94,
|
| 753 |
+
"grad_norm": 2.2428641319274902,
|
| 754 |
+
"learning_rate": 0.004028703703703704,
|
| 755 |
+
"loss": 0.7271,
|
| 756 |
+
"step": 1050
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.96,
|
| 760 |
+
"grad_norm": 2.1830852031707764,
|
| 761 |
+
"learning_rate": 0.004019444444444444,
|
| 762 |
+
"loss": 0.6171,
|
| 763 |
+
"step": 1060
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.98,
|
| 767 |
+
"grad_norm": 4.2572455406188965,
|
| 768 |
+
"learning_rate": 0.004010185185185185,
|
| 769 |
+
"loss": 0.715,
|
| 770 |
+
"step": 1070
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 2.0,
|
| 774 |
+
"grad_norm": 1.7257713079452515,
|
| 775 |
+
"learning_rate": 0.004000925925925926,
|
| 776 |
+
"loss": 0.7397,
|
| 777 |
+
"step": 1080
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 2.0,
|
| 781 |
+
"eval_accuracy": 0.9406871052226159,
|
| 782 |
+
"eval_f1": 0.9414537311136683,
|
| 783 |
+
"eval_loss": 0.17834338545799255,
|
| 784 |
+
"eval_precision": 0.9539065276276052,
|
| 785 |
+
"eval_recall": 0.9346301030411137,
|
| 786 |
+
"eval_runtime": 55.3829,
|
| 787 |
+
"eval_samples_per_second": 117.202,
|
| 788 |
+
"eval_steps_per_second": 7.331,
|
| 789 |
+
"step": 1081
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.02,
|
| 793 |
+
"grad_norm": 1.4031952619552612,
|
| 794 |
+
"learning_rate": 0.003991666666666667,
|
| 795 |
+
"loss": 0.652,
|
| 796 |
+
"step": 1090
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.04,
|
| 800 |
+
"grad_norm": 2.210618257522583,
|
| 801 |
+
"learning_rate": 0.003982407407407408,
|
| 802 |
+
"loss": 0.6434,
|
| 803 |
+
"step": 1100
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 2.05,
|
| 807 |
+
"grad_norm": 2.780142068862915,
|
| 808 |
+
"learning_rate": 0.0039731481481481475,
|
| 809 |
+
"loss": 0.6638,
|
| 810 |
+
"step": 1110
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 2.07,
|
| 814 |
+
"grad_norm": 2.0362918376922607,
|
| 815 |
+
"learning_rate": 0.003963888888888889,
|
| 816 |
+
"loss": 0.6194,
|
| 817 |
+
"step": 1120
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 2.09,
|
| 821 |
+
"grad_norm": 2.7346765995025635,
|
| 822 |
+
"learning_rate": 0.00395462962962963,
|
| 823 |
+
"loss": 0.6721,
|
| 824 |
+
"step": 1130
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.11,
|
| 828 |
+
"grad_norm": 1.8441007137298584,
|
| 829 |
+
"learning_rate": 0.00394537037037037,
|
| 830 |
+
"loss": 0.7129,
|
| 831 |
+
"step": 1140
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.13,
|
| 835 |
+
"grad_norm": 2.510021448135376,
|
| 836 |
+
"learning_rate": 0.003936111111111112,
|
| 837 |
+
"loss": 0.669,
|
| 838 |
+
"step": 1150
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 2.15,
|
| 842 |
+
"grad_norm": 2.9301140308380127,
|
| 843 |
+
"learning_rate": 0.003926851851851852,
|
| 844 |
+
"loss": 0.6342,
|
| 845 |
+
"step": 1160
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 2.16,
|
| 849 |
+
"grad_norm": 1.851173996925354,
|
| 850 |
+
"learning_rate": 0.003917592592592592,
|
| 851 |
+
"loss": 0.6602,
|
| 852 |
+
"step": 1170
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 2.18,
|
| 856 |
+
"grad_norm": 2.490886926651001,
|
| 857 |
+
"learning_rate": 0.003908333333333333,
|
| 858 |
+
"loss": 0.6195,
|
| 859 |
+
"step": 1180
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 2.2,
|
| 863 |
+
"grad_norm": 1.5079678297042847,
|
| 864 |
+
"learning_rate": 0.0038990740740740743,
|
| 865 |
+
"loss": 0.8126,
|
| 866 |
+
"step": 1190
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.22,
|
| 870 |
+
"grad_norm": 1.9084036350250244,
|
| 871 |
+
"learning_rate": 0.003889814814814815,
|
| 872 |
+
"loss": 0.6769,
|
| 873 |
+
"step": 1200
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.24,
|
| 877 |
+
"grad_norm": 1.8284987211227417,
|
| 878 |
+
"learning_rate": 0.0038814814814814814,
|
| 879 |
+
"loss": 0.7317,
|
| 880 |
+
"step": 1210
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 2.26,
|
| 884 |
+
"grad_norm": 3.351156234741211,
|
| 885 |
+
"learning_rate": 0.0038722222222222226,
|
| 886 |
+
"loss": 0.7049,
|
| 887 |
+
"step": 1220
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 2.28,
|
| 891 |
+
"grad_norm": 2.314993143081665,
|
| 892 |
+
"learning_rate": 0.003862962962962963,
|
| 893 |
+
"loss": 0.6291,
|
| 894 |
+
"step": 1230
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 2.29,
|
| 898 |
+
"grad_norm": 2.27850341796875,
|
| 899 |
+
"learning_rate": 0.003853703703703704,
|
| 900 |
+
"loss": 0.6749,
|
| 901 |
+
"step": 1240
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 2.31,
|
| 905 |
+
"grad_norm": 2.4541099071502686,
|
| 906 |
+
"learning_rate": 0.003844444444444444,
|
| 907 |
+
"loss": 0.7365,
|
| 908 |
+
"step": 1250
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 2.33,
|
| 912 |
+
"grad_norm": 1.6301252841949463,
|
| 913 |
+
"learning_rate": 0.0038351851851851854,
|
| 914 |
+
"loss": 0.6236,
|
| 915 |
+
"step": 1260
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 2.35,
|
| 919 |
+
"grad_norm": 2.035531520843506,
|
| 920 |
+
"learning_rate": 0.003825925925925926,
|
| 921 |
+
"loss": 0.7176,
|
| 922 |
+
"step": 1270
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 2.37,
|
| 926 |
+
"grad_norm": 2.2524566650390625,
|
| 927 |
+
"learning_rate": 0.0038166666666666666,
|
| 928 |
+
"loss": 0.7119,
|
| 929 |
+
"step": 1280
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 2.39,
|
| 933 |
+
"grad_norm": 1.778144121170044,
|
| 934 |
+
"learning_rate": 0.003807407407407408,
|
| 935 |
+
"loss": 0.7021,
|
| 936 |
+
"step": 1290
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 2.41,
|
| 940 |
+
"grad_norm": 1.8808122873306274,
|
| 941 |
+
"learning_rate": 0.003798148148148148,
|
| 942 |
+
"loss": 0.6876,
|
| 943 |
+
"step": 1300
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 2.42,
|
| 947 |
+
"grad_norm": 2.720100164413452,
|
| 948 |
+
"learning_rate": 0.003788888888888889,
|
| 949 |
+
"loss": 0.7559,
|
| 950 |
+
"step": 1310
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 2.44,
|
| 954 |
+
"grad_norm": 1.9608194828033447,
|
| 955 |
+
"learning_rate": 0.0037796296296296297,
|
| 956 |
+
"loss": 0.6799,
|
| 957 |
+
"step": 1320
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 2.46,
|
| 961 |
+
"grad_norm": 1.8149750232696533,
|
| 962 |
+
"learning_rate": 0.0037703703703703705,
|
| 963 |
+
"loss": 0.6701,
|
| 964 |
+
"step": 1330
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 2.48,
|
| 968 |
+
"grad_norm": 1.8840572834014893,
|
| 969 |
+
"learning_rate": 0.0037611111111111113,
|
| 970 |
+
"loss": 0.7146,
|
| 971 |
+
"step": 1340
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 2.5,
|
| 975 |
+
"grad_norm": 1.7513312101364136,
|
| 976 |
+
"learning_rate": 0.0037518518518518517,
|
| 977 |
+
"loss": 0.6712,
|
| 978 |
+
"step": 1350
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 2.52,
|
| 982 |
+
"grad_norm": 1.748304009437561,
|
| 983 |
+
"learning_rate": 0.0037425925925925925,
|
| 984 |
+
"loss": 0.7049,
|
| 985 |
+
"step": 1360
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 2.53,
|
| 989 |
+
"grad_norm": 1.2645372152328491,
|
| 990 |
+
"learning_rate": 0.0037333333333333337,
|
| 991 |
+
"loss": 0.6088,
|
| 992 |
+
"step": 1370
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 2.55,
|
| 996 |
+
"grad_norm": 3.14270281791687,
|
| 997 |
+
"learning_rate": 0.003724074074074074,
|
| 998 |
+
"loss": 0.7163,
|
| 999 |
+
"step": 1380
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.57,
|
| 1003 |
+
"grad_norm": 2.0332484245300293,
|
| 1004 |
+
"learning_rate": 0.003714814814814815,
|
| 1005 |
+
"loss": 0.7047,
|
| 1006 |
+
"step": 1390
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 2.59,
|
| 1010 |
+
"grad_norm": 1.6158325672149658,
|
| 1011 |
+
"learning_rate": 0.0037055555555555557,
|
| 1012 |
+
"loss": 0.6448,
|
| 1013 |
+
"step": 1400
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 2.61,
|
| 1017 |
+
"grad_norm": 1.9182847738265991,
|
| 1018 |
+
"learning_rate": 0.0036962962962962965,
|
| 1019 |
+
"loss": 0.711,
|
| 1020 |
+
"step": 1410
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 2.63,
|
| 1024 |
+
"grad_norm": 2.140184164047241,
|
| 1025 |
+
"learning_rate": 0.0036870370370370373,
|
| 1026 |
+
"loss": 0.6617,
|
| 1027 |
+
"step": 1420
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 2.65,
|
| 1031 |
+
"grad_norm": 1.746243953704834,
|
| 1032 |
+
"learning_rate": 0.0036777777777777776,
|
| 1033 |
+
"loss": 0.7103,
|
| 1034 |
+
"step": 1430
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.66,
|
| 1038 |
+
"grad_norm": 2.6956822872161865,
|
| 1039 |
+
"learning_rate": 0.003668518518518519,
|
| 1040 |
+
"loss": 0.7163,
|
| 1041 |
+
"step": 1440
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.68,
|
| 1045 |
+
"grad_norm": 2.7663381099700928,
|
| 1046 |
+
"learning_rate": 0.0036592592592592592,
|
| 1047 |
+
"loss": 0.6509,
|
| 1048 |
+
"step": 1450
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 2.7,
|
| 1052 |
+
"grad_norm": 2.410922050476074,
|
| 1053 |
+
"learning_rate": 0.00365,
|
| 1054 |
+
"loss": 0.6046,
|
| 1055 |
+
"step": 1460
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.72,
|
| 1059 |
+
"grad_norm": 1.8373881578445435,
|
| 1060 |
+
"learning_rate": 0.0036407407407407404,
|
| 1061 |
+
"loss": 0.701,
|
| 1062 |
+
"step": 1470
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 2.74,
|
| 1066 |
+
"grad_norm": 1.8466787338256836,
|
| 1067 |
+
"learning_rate": 0.0036314814814814816,
|
| 1068 |
+
"loss": 0.6391,
|
| 1069 |
+
"step": 1480
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 2.76,
|
| 1073 |
+
"grad_norm": 2.0422921180725098,
|
| 1074 |
+
"learning_rate": 0.0036222222222222224,
|
| 1075 |
+
"loss": 0.7001,
|
| 1076 |
+
"step": 1490
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.78,
|
| 1080 |
+
"grad_norm": 1.739327311515808,
|
| 1081 |
+
"learning_rate": 0.003612962962962963,
|
| 1082 |
+
"loss": 0.6194,
|
| 1083 |
+
"step": 1500
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.79,
|
| 1087 |
+
"grad_norm": 2.094775915145874,
|
| 1088 |
+
"learning_rate": 0.003603703703703704,
|
| 1089 |
+
"loss": 0.5824,
|
| 1090 |
+
"step": 1510
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.81,
|
| 1094 |
+
"grad_norm": 2.5036680698394775,
|
| 1095 |
+
"learning_rate": 0.003594444444444445,
|
| 1096 |
+
"loss": 0.6477,
|
| 1097 |
+
"step": 1520
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 2.83,
|
| 1101 |
+
"grad_norm": 1.7037166357040405,
|
| 1102 |
+
"learning_rate": 0.003585185185185185,
|
| 1103 |
+
"loss": 0.6822,
|
| 1104 |
+
"step": 1530
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 2.85,
|
| 1108 |
+
"grad_norm": 2.193714141845703,
|
| 1109 |
+
"learning_rate": 0.003575925925925926,
|
| 1110 |
+
"loss": 0.7301,
|
| 1111 |
+
"step": 1540
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 2.87,
|
| 1115 |
+
"grad_norm": 4.449968338012695,
|
| 1116 |
+
"learning_rate": 0.0035666666666666668,
|
| 1117 |
+
"loss": 0.7114,
|
| 1118 |
+
"step": 1550
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 2.89,
|
| 1122 |
+
"grad_norm": 1.640833854675293,
|
| 1123 |
+
"learning_rate": 0.0035574074074074076,
|
| 1124 |
+
"loss": 0.5631,
|
| 1125 |
+
"step": 1560
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.9,
|
| 1129 |
+
"grad_norm": 2.4068825244903564,
|
| 1130 |
+
"learning_rate": 0.003548148148148148,
|
| 1131 |
+
"loss": 0.5896,
|
| 1132 |
+
"step": 1570
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 2.92,
|
| 1136 |
+
"grad_norm": 2.7234010696411133,
|
| 1137 |
+
"learning_rate": 0.0035388888888888887,
|
| 1138 |
+
"loss": 0.7563,
|
| 1139 |
+
"step": 1580
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 2.94,
|
| 1143 |
+
"grad_norm": 1.8094590902328491,
|
| 1144 |
+
"learning_rate": 0.00352962962962963,
|
| 1145 |
+
"loss": 0.7116,
|
| 1146 |
+
"step": 1590
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 2.96,
|
| 1150 |
+
"grad_norm": 2.4650044441223145,
|
| 1151 |
+
"learning_rate": 0.0035203703703703703,
|
| 1152 |
+
"loss": 0.6583,
|
| 1153 |
+
"step": 1600
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 2.98,
|
| 1157 |
+
"grad_norm": 1.8076032400131226,
|
| 1158 |
+
"learning_rate": 0.003511111111111111,
|
| 1159 |
+
"loss": 0.5859,
|
| 1160 |
+
"step": 1610
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.0,
|
| 1164 |
+
"grad_norm": 2.924779176712036,
|
| 1165 |
+
"learning_rate": 0.0035018518518518523,
|
| 1166 |
+
"loss": 0.7151,
|
| 1167 |
+
"step": 1620
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.0,
|
| 1171 |
+
"eval_accuracy": 0.9551686951163149,
|
| 1172 |
+
"eval_f1": 0.957151302239187,
|
| 1173 |
+
"eval_loss": 0.12974682450294495,
|
| 1174 |
+
"eval_precision": 0.961140704317014,
|
| 1175 |
+
"eval_recall": 0.9555174635857528,
|
| 1176 |
+
"eval_runtime": 55.4288,
|
| 1177 |
+
"eval_samples_per_second": 117.105,
|
| 1178 |
+
"eval_steps_per_second": 7.325,
|
| 1179 |
+
"step": 1621
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 3.02,
|
| 1183 |
+
"grad_norm": 2.4308605194091797,
|
| 1184 |
+
"learning_rate": 0.0034925925925925927,
|
| 1185 |
+
"loss": 0.6032,
|
| 1186 |
+
"step": 1630
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 3.03,
|
| 1190 |
+
"grad_norm": 1.6040972471237183,
|
| 1191 |
+
"learning_rate": 0.0034833333333333335,
|
| 1192 |
+
"loss": 0.6163,
|
| 1193 |
+
"step": 1640
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 3.05,
|
| 1197 |
+
"grad_norm": 1.7093302011489868,
|
| 1198 |
+
"learning_rate": 0.003474074074074074,
|
| 1199 |
+
"loss": 0.6484,
|
| 1200 |
+
"step": 1650
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 3.07,
|
| 1204 |
+
"grad_norm": 2.0212676525115967,
|
| 1205 |
+
"learning_rate": 0.003464814814814815,
|
| 1206 |
+
"loss": 0.5747,
|
| 1207 |
+
"step": 1660
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 3.09,
|
| 1211 |
+
"grad_norm": 3.725684642791748,
|
| 1212 |
+
"learning_rate": 0.0034555555555555555,
|
| 1213 |
+
"loss": 0.6488,
|
| 1214 |
+
"step": 1670
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 3.11,
|
| 1218 |
+
"grad_norm": 2.5696334838867188,
|
| 1219 |
+
"learning_rate": 0.0034462962962962963,
|
| 1220 |
+
"loss": 0.7223,
|
| 1221 |
+
"step": 1680
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 3.13,
|
| 1225 |
+
"grad_norm": 1.7049779891967773,
|
| 1226 |
+
"learning_rate": 0.003437037037037037,
|
| 1227 |
+
"loss": 0.641,
|
| 1228 |
+
"step": 1690
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 3.15,
|
| 1232 |
+
"grad_norm": 2.035477638244629,
|
| 1233 |
+
"learning_rate": 0.003427777777777778,
|
| 1234 |
+
"loss": 0.6237,
|
| 1235 |
+
"step": 1700
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 3.16,
|
| 1239 |
+
"grad_norm": 1.8789253234863281,
|
| 1240 |
+
"learning_rate": 0.0034185185185185187,
|
| 1241 |
+
"loss": 0.6312,
|
| 1242 |
+
"step": 1710
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 3.18,
|
| 1246 |
+
"grad_norm": 2.2245302200317383,
|
| 1247 |
+
"learning_rate": 0.003409259259259259,
|
| 1248 |
+
"loss": 0.5985,
|
| 1249 |
+
"step": 1720
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 3.2,
|
| 1253 |
+
"grad_norm": 2.150331735610962,
|
| 1254 |
+
"learning_rate": 0.0034000000000000002,
|
| 1255 |
+
"loss": 0.612,
|
| 1256 |
+
"step": 1730
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 3.22,
|
| 1260 |
+
"grad_norm": 1.986059546470642,
|
| 1261 |
+
"learning_rate": 0.003390740740740741,
|
| 1262 |
+
"loss": 0.6001,
|
| 1263 |
+
"step": 1740
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.24,
|
| 1267 |
+
"grad_norm": 1.4887031316757202,
|
| 1268 |
+
"learning_rate": 0.0033814814814814814,
|
| 1269 |
+
"loss": 0.6574,
|
| 1270 |
+
"step": 1750
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 3.26,
|
| 1274 |
+
"grad_norm": 1.1965304613113403,
|
| 1275 |
+
"learning_rate": 0.003372222222222222,
|
| 1276 |
+
"loss": 0.6766,
|
| 1277 |
+
"step": 1760
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.27,
|
| 1281 |
+
"grad_norm": 1.564257025718689,
|
| 1282 |
+
"learning_rate": 0.0033629629629629634,
|
| 1283 |
+
"loss": 0.6955,
|
| 1284 |
+
"step": 1770
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 3.29,
|
| 1288 |
+
"grad_norm": 5.508277416229248,
|
| 1289 |
+
"learning_rate": 0.003353703703703704,
|
| 1290 |
+
"loss": 0.6097,
|
| 1291 |
+
"step": 1780
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.31,
|
| 1295 |
+
"grad_norm": 1.9245549440383911,
|
| 1296 |
+
"learning_rate": 0.0033444444444444446,
|
| 1297 |
+
"loss": 0.6515,
|
| 1298 |
+
"step": 1790
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 3.33,
|
| 1302 |
+
"grad_norm": 2.3552732467651367,
|
| 1303 |
+
"learning_rate": 0.003335185185185185,
|
| 1304 |
+
"loss": 0.6401,
|
| 1305 |
+
"step": 1800
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 3.35,
|
| 1309 |
+
"grad_norm": 1.622826099395752,
|
| 1310 |
+
"learning_rate": 0.003325925925925926,
|
| 1311 |
+
"loss": 0.7,
|
| 1312 |
+
"step": 1810
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 3.37,
|
| 1316 |
+
"grad_norm": 1.6205705404281616,
|
| 1317 |
+
"learning_rate": 0.0033166666666666665,
|
| 1318 |
+
"loss": 0.5968,
|
| 1319 |
+
"step": 1820
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.39,
|
| 1323 |
+
"grad_norm": 1.9269992113113403,
|
| 1324 |
+
"learning_rate": 0.0033074074074074073,
|
| 1325 |
+
"loss": 0.6,
|
| 1326 |
+
"step": 1830
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 3.4,
|
| 1330 |
+
"grad_norm": 2.3607101440429688,
|
| 1331 |
+
"learning_rate": 0.0032981481481481486,
|
| 1332 |
+
"loss": 0.5655,
|
| 1333 |
+
"step": 1840
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 3.42,
|
| 1337 |
+
"grad_norm": 1.4156101942062378,
|
| 1338 |
+
"learning_rate": 0.003288888888888889,
|
| 1339 |
+
"loss": 0.6071,
|
| 1340 |
+
"step": 1850
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 3.44,
|
| 1344 |
+
"grad_norm": 2.652838706970215,
|
| 1345 |
+
"learning_rate": 0.0032796296296296297,
|
| 1346 |
+
"loss": 0.5959,
|
| 1347 |
+
"step": 1860
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 3.46,
|
| 1351 |
+
"grad_norm": 1.7326959371566772,
|
| 1352 |
+
"learning_rate": 0.00327037037037037,
|
| 1353 |
+
"loss": 0.7051,
|
| 1354 |
+
"step": 1870
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.48,
|
| 1358 |
+
"grad_norm": 1.4709230661392212,
|
| 1359 |
+
"learning_rate": 0.0032611111111111113,
|
| 1360 |
+
"loss": 0.68,
|
| 1361 |
+
"step": 1880
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.5,
|
| 1365 |
+
"grad_norm": 2.1479110717773438,
|
| 1366 |
+
"learning_rate": 0.003251851851851852,
|
| 1367 |
+
"loss": 0.6491,
|
| 1368 |
+
"step": 1890
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.52,
|
| 1372 |
+
"grad_norm": 2.462580442428589,
|
| 1373 |
+
"learning_rate": 0.0032425925925925925,
|
| 1374 |
+
"loss": 0.6129,
|
| 1375 |
+
"step": 1900
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.53,
|
| 1379 |
+
"grad_norm": 2.8849940299987793,
|
| 1380 |
+
"learning_rate": 0.0032333333333333333,
|
| 1381 |
+
"loss": 0.6553,
|
| 1382 |
+
"step": 1910
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.55,
|
| 1386 |
+
"grad_norm": 1.6244118213653564,
|
| 1387 |
+
"learning_rate": 0.003224074074074074,
|
| 1388 |
+
"loss": 0.5845,
|
| 1389 |
+
"step": 1920
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.57,
|
| 1393 |
+
"grad_norm": 2.803816795349121,
|
| 1394 |
+
"learning_rate": 0.003214814814814815,
|
| 1395 |
+
"loss": 0.6695,
|
| 1396 |
+
"step": 1930
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.59,
|
| 1400 |
+
"grad_norm": 2.4138996601104736,
|
| 1401 |
+
"learning_rate": 0.0032055555555555552,
|
| 1402 |
+
"loss": 0.707,
|
| 1403 |
+
"step": 1940
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.61,
|
| 1407 |
+
"grad_norm": 1.1193430423736572,
|
| 1408 |
+
"learning_rate": 0.0031962962962962965,
|
| 1409 |
+
"loss": 0.6003,
|
| 1410 |
+
"step": 1950
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.63,
|
| 1414 |
+
"grad_norm": 1.8593242168426514,
|
| 1415 |
+
"learning_rate": 0.0031870370370370373,
|
| 1416 |
+
"loss": 0.5724,
|
| 1417 |
+
"step": 1960
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.64,
|
| 1421 |
+
"grad_norm": 2.084005117416382,
|
| 1422 |
+
"learning_rate": 0.0031777777777777776,
|
| 1423 |
+
"loss": 0.6246,
|
| 1424 |
+
"step": 1970
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.66,
|
| 1428 |
+
"grad_norm": 1.32781982421875,
|
| 1429 |
+
"learning_rate": 0.0031685185185185184,
|
| 1430 |
+
"loss": 0.6075,
|
| 1431 |
+
"step": 1980
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.68,
|
| 1435 |
+
"grad_norm": 2.452892303466797,
|
| 1436 |
+
"learning_rate": 0.0031592592592592597,
|
| 1437 |
+
"loss": 0.6104,
|
| 1438 |
+
"step": 1990
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.7,
|
| 1442 |
+
"grad_norm": 1.6961791515350342,
|
| 1443 |
+
"learning_rate": 0.00315,
|
| 1444 |
+
"loss": 0.5483,
|
| 1445 |
+
"step": 2000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.72,
|
| 1449 |
+
"grad_norm": 2.94992733001709,
|
| 1450 |
+
"learning_rate": 0.003140740740740741,
|
| 1451 |
+
"loss": 0.5403,
|
| 1452 |
+
"step": 2010
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 3.74,
|
| 1456 |
+
"grad_norm": 2.3624625205993652,
|
| 1457 |
+
"learning_rate": 0.0031314814814814816,
|
| 1458 |
+
"loss": 0.6748,
|
| 1459 |
+
"step": 2020
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.76,
|
| 1463 |
+
"grad_norm": 2.0412802696228027,
|
| 1464 |
+
"learning_rate": 0.0031222222222222224,
|
| 1465 |
+
"loss": 0.4809,
|
| 1466 |
+
"step": 2030
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 3.77,
|
| 1470 |
+
"grad_norm": 1.5728198289871216,
|
| 1471 |
+
"learning_rate": 0.003112962962962963,
|
| 1472 |
+
"loss": 0.5755,
|
| 1473 |
+
"step": 2040
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 3.79,
|
| 1477 |
+
"grad_norm": 1.3974460363388062,
|
| 1478 |
+
"learning_rate": 0.0031037037037037036,
|
| 1479 |
+
"loss": 0.5436,
|
| 1480 |
+
"step": 2050
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 3.81,
|
| 1484 |
+
"grad_norm": 1.5604791641235352,
|
| 1485 |
+
"learning_rate": 0.003094444444444445,
|
| 1486 |
+
"loss": 0.6197,
|
| 1487 |
+
"step": 2060
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.83,
|
| 1491 |
+
"grad_norm": 2.040276288986206,
|
| 1492 |
+
"learning_rate": 0.003085185185185185,
|
| 1493 |
+
"loss": 0.6062,
|
| 1494 |
+
"step": 2070
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 3.85,
|
| 1498 |
+
"grad_norm": 3.173293113708496,
|
| 1499 |
+
"learning_rate": 0.003075925925925926,
|
| 1500 |
+
"loss": 0.717,
|
| 1501 |
+
"step": 2080
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 3.87,
|
| 1505 |
+
"grad_norm": 2.018458843231201,
|
| 1506 |
+
"learning_rate": 0.0030666666666666663,
|
| 1507 |
+
"loss": 0.6418,
|
| 1508 |
+
"step": 2090
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 3.89,
|
| 1512 |
+
"grad_norm": 1.7057709693908691,
|
| 1513 |
+
"learning_rate": 0.0030574074074074076,
|
| 1514 |
+
"loss": 0.5948,
|
| 1515 |
+
"step": 2100
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 3.9,
|
| 1519 |
+
"grad_norm": 2.133119821548462,
|
| 1520 |
+
"learning_rate": 0.0030481481481481484,
|
| 1521 |
+
"loss": 0.6099,
|
| 1522 |
+
"step": 2110
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 3.92,
|
| 1526 |
+
"grad_norm": 2.217524290084839,
|
| 1527 |
+
"learning_rate": 0.0030388888888888887,
|
| 1528 |
+
"loss": 0.6263,
|
| 1529 |
+
"step": 2120
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.94,
|
| 1533 |
+
"grad_norm": 2.3708674907684326,
|
| 1534 |
+
"learning_rate": 0.00302962962962963,
|
| 1535 |
+
"loss": 0.5851,
|
| 1536 |
+
"step": 2130
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 3.96,
|
| 1540 |
+
"grad_norm": 1.4289004802703857,
|
| 1541 |
+
"learning_rate": 0.0030203703703703707,
|
| 1542 |
+
"loss": 0.5971,
|
| 1543 |
+
"step": 2140
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 3.98,
|
| 1547 |
+
"grad_norm": 2.463360071182251,
|
| 1548 |
+
"learning_rate": 0.003011111111111111,
|
| 1549 |
+
"loss": 0.5889,
|
| 1550 |
+
"step": 2150
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 4.0,
|
| 1554 |
+
"grad_norm": 1.5189619064331055,
|
| 1555 |
+
"learning_rate": 0.003001851851851852,
|
| 1556 |
+
"loss": 0.4964,
|
| 1557 |
+
"step": 2160
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 4.0,
|
| 1561 |
+
"eval_accuracy": 0.97350177168387,
|
| 1562 |
+
"eval_f1": 0.9729772233764705,
|
| 1563 |
+
"eval_loss": 0.07412911206483841,
|
| 1564 |
+
"eval_precision": 0.9765336155795084,
|
| 1565 |
+
"eval_recall": 0.9701628526053132,
|
| 1566 |
+
"eval_runtime": 55.303,
|
| 1567 |
+
"eval_samples_per_second": 117.372,
|
| 1568 |
+
"eval_steps_per_second": 7.341,
|
| 1569 |
+
"step": 2162
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 4.01,
|
| 1573 |
+
"grad_norm": 2.313246965408325,
|
| 1574 |
+
"learning_rate": 0.0029925925925925927,
|
| 1575 |
+
"loss": 0.5979,
|
| 1576 |
+
"step": 2170
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 4.03,
|
| 1580 |
+
"grad_norm": 1.554916501045227,
|
| 1581 |
+
"learning_rate": 0.0029833333333333335,
|
| 1582 |
+
"loss": 0.621,
|
| 1583 |
+
"step": 2180
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 4.05,
|
| 1587 |
+
"grad_norm": 2.7611353397369385,
|
| 1588 |
+
"learning_rate": 0.002974074074074074,
|
| 1589 |
+
"loss": 0.6662,
|
| 1590 |
+
"step": 2190
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 4.07,
|
| 1594 |
+
"grad_norm": 1.206809639930725,
|
| 1595 |
+
"learning_rate": 0.0029648148148148147,
|
| 1596 |
+
"loss": 0.6061,
|
| 1597 |
+
"step": 2200
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 4.09,
|
| 1601 |
+
"grad_norm": 1.876120686531067,
|
| 1602 |
+
"learning_rate": 0.002955555555555556,
|
| 1603 |
+
"loss": 0.5656,
|
| 1604 |
+
"step": 2210
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 4.11,
|
| 1608 |
+
"grad_norm": 2.117579936981201,
|
| 1609 |
+
"learning_rate": 0.0029462962962962963,
|
| 1610 |
+
"loss": 0.611,
|
| 1611 |
+
"step": 2220
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 4.13,
|
| 1615 |
+
"grad_norm": 2.014042377471924,
|
| 1616 |
+
"learning_rate": 0.002937037037037037,
|
| 1617 |
+
"loss": 0.5893,
|
| 1618 |
+
"step": 2230
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 4.14,
|
| 1622 |
+
"grad_norm": 1.6792641878128052,
|
| 1623 |
+
"learning_rate": 0.0029277777777777783,
|
| 1624 |
+
"loss": 0.5691,
|
| 1625 |
+
"step": 2240
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 4.16,
|
| 1629 |
+
"grad_norm": 1.2664446830749512,
|
| 1630 |
+
"learning_rate": 0.0029185185185185186,
|
| 1631 |
+
"loss": 0.5729,
|
| 1632 |
+
"step": 2250
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 4.18,
|
| 1636 |
+
"grad_norm": 3.1098408699035645,
|
| 1637 |
+
"learning_rate": 0.0029092592592592594,
|
| 1638 |
+
"loss": 0.6115,
|
| 1639 |
+
"step": 2260
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 4.2,
|
| 1643 |
+
"grad_norm": 2.1186132431030273,
|
| 1644 |
+
"learning_rate": 0.0029,
|
| 1645 |
+
"loss": 0.596,
|
| 1646 |
+
"step": 2270
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 4.22,
|
| 1650 |
+
"grad_norm": 1.8666993379592896,
|
| 1651 |
+
"learning_rate": 0.002890740740740741,
|
| 1652 |
+
"loss": 0.6021,
|
| 1653 |
+
"step": 2280
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 4.24,
|
| 1657 |
+
"grad_norm": 1.655548334121704,
|
| 1658 |
+
"learning_rate": 0.0028814814814814814,
|
| 1659 |
+
"loss": 0.5928,
|
| 1660 |
+
"step": 2290
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 4.26,
|
| 1664 |
+
"grad_norm": 2.1081762313842773,
|
| 1665 |
+
"learning_rate": 0.002872222222222222,
|
| 1666 |
+
"loss": 0.5683,
|
| 1667 |
+
"step": 2300
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 4.27,
|
| 1671 |
+
"grad_norm": 1.4986019134521484,
|
| 1672 |
+
"learning_rate": 0.0028629629629629626,
|
| 1673 |
+
"loss": 0.562,
|
| 1674 |
+
"step": 2310
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 4.29,
|
| 1678 |
+
"grad_norm": 2.6111056804656982,
|
| 1679 |
+
"learning_rate": 0.002853703703703704,
|
| 1680 |
+
"loss": 0.5804,
|
| 1681 |
+
"step": 2320
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 4.31,
|
| 1685 |
+
"grad_norm": 1.8590542078018188,
|
| 1686 |
+
"learning_rate": 0.0028444444444444446,
|
| 1687 |
+
"loss": 0.5409,
|
| 1688 |
+
"step": 2330
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 4.33,
|
| 1692 |
+
"grad_norm": 1.897940754890442,
|
| 1693 |
+
"learning_rate": 0.002835185185185185,
|
| 1694 |
+
"loss": 0.6299,
|
| 1695 |
+
"step": 2340
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 4.35,
|
| 1699 |
+
"grad_norm": 1.9114952087402344,
|
| 1700 |
+
"learning_rate": 0.002825925925925926,
|
| 1701 |
+
"loss": 0.5333,
|
| 1702 |
+
"step": 2350
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 4.37,
|
| 1706 |
+
"grad_norm": 2.015868663787842,
|
| 1707 |
+
"learning_rate": 0.002816666666666667,
|
| 1708 |
+
"loss": 0.5611,
|
| 1709 |
+
"step": 2360
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 4.38,
|
| 1713 |
+
"grad_norm": 1.5808693170547485,
|
| 1714 |
+
"learning_rate": 0.0028074074074074073,
|
| 1715 |
+
"loss": 0.6212,
|
| 1716 |
+
"step": 2370
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 4.4,
|
| 1720 |
+
"grad_norm": 1.8288767337799072,
|
| 1721 |
+
"learning_rate": 0.002798148148148148,
|
| 1722 |
+
"loss": 0.604,
|
| 1723 |
+
"step": 2380
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 4.42,
|
| 1727 |
+
"grad_norm": 1.389092206954956,
|
| 1728 |
+
"learning_rate": 0.002788888888888889,
|
| 1729 |
+
"loss": 0.4657,
|
| 1730 |
+
"step": 2390
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 4.44,
|
| 1734 |
+
"grad_norm": 2.551008701324463,
|
| 1735 |
+
"learning_rate": 0.0027796296296296297,
|
| 1736 |
+
"loss": 0.6214,
|
| 1737 |
+
"step": 2400
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 4.46,
|
| 1741 |
+
"grad_norm": 2.26882266998291,
|
| 1742 |
+
"learning_rate": 0.0027703703703703705,
|
| 1743 |
+
"loss": 0.5511,
|
| 1744 |
+
"step": 2410
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 4.48,
|
| 1748 |
+
"grad_norm": 2.3850460052490234,
|
| 1749 |
+
"learning_rate": 0.002761111111111111,
|
| 1750 |
+
"loss": 0.5768,
|
| 1751 |
+
"step": 2420
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 4.5,
|
| 1755 |
+
"grad_norm": 1.9451379776000977,
|
| 1756 |
+
"learning_rate": 0.002751851851851852,
|
| 1757 |
+
"loss": 0.5732,
|
| 1758 |
+
"step": 2430
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 4.51,
|
| 1762 |
+
"grad_norm": 1.6541566848754883,
|
| 1763 |
+
"learning_rate": 0.0027425925925925925,
|
| 1764 |
+
"loss": 0.549,
|
| 1765 |
+
"step": 2440
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 4.53,
|
| 1769 |
+
"grad_norm": 1.3802433013916016,
|
| 1770 |
+
"learning_rate": 0.0027333333333333333,
|
| 1771 |
+
"loss": 0.5486,
|
| 1772 |
+
"step": 2450
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 4.55,
|
| 1776 |
+
"grad_norm": 1.941901683807373,
|
| 1777 |
+
"learning_rate": 0.0027240740740740745,
|
| 1778 |
+
"loss": 0.6119,
|
| 1779 |
+
"step": 2460
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 4.57,
|
| 1783 |
+
"grad_norm": 1.9203604459762573,
|
| 1784 |
+
"learning_rate": 0.002714814814814815,
|
| 1785 |
+
"loss": 0.632,
|
| 1786 |
+
"step": 2470
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 4.59,
|
| 1790 |
+
"grad_norm": 2.170319080352783,
|
| 1791 |
+
"learning_rate": 0.0027055555555555557,
|
| 1792 |
+
"loss": 0.5819,
|
| 1793 |
+
"step": 2480
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 4.61,
|
| 1797 |
+
"grad_norm": 1.8074395656585693,
|
| 1798 |
+
"learning_rate": 0.002696296296296296,
|
| 1799 |
+
"loss": 0.6101,
|
| 1800 |
+
"step": 2490
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 4.63,
|
| 1804 |
+
"grad_norm": 2.010608196258545,
|
| 1805 |
+
"learning_rate": 0.0026870370370370373,
|
| 1806 |
+
"loss": 0.492,
|
| 1807 |
+
"step": 2500
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 4.64,
|
| 1811 |
+
"grad_norm": 1.3907346725463867,
|
| 1812 |
+
"learning_rate": 0.002677777777777778,
|
| 1813 |
+
"loss": 0.5658,
|
| 1814 |
+
"step": 2510
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 4.66,
|
| 1818 |
+
"grad_norm": 1.8057767152786255,
|
| 1819 |
+
"learning_rate": 0.0026685185185185184,
|
| 1820 |
+
"loss": 0.509,
|
| 1821 |
+
"step": 2520
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 4.68,
|
| 1825 |
+
"grad_norm": 2.036869764328003,
|
| 1826 |
+
"learning_rate": 0.002659259259259259,
|
| 1827 |
+
"loss": 0.5711,
|
| 1828 |
+
"step": 2530
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 4.7,
|
| 1832 |
+
"grad_norm": 1.673647165298462,
|
| 1833 |
+
"learning_rate": 0.00265,
|
| 1834 |
+
"loss": 0.5692,
|
| 1835 |
+
"step": 2540
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 4.72,
|
| 1839 |
+
"grad_norm": 2.044961929321289,
|
| 1840 |
+
"learning_rate": 0.002640740740740741,
|
| 1841 |
+
"loss": 0.6153,
|
| 1842 |
+
"step": 2550
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 4.74,
|
| 1846 |
+
"grad_norm": 1.3360215425491333,
|
| 1847 |
+
"learning_rate": 0.002631481481481481,
|
| 1848 |
+
"loss": 0.5785,
|
| 1849 |
+
"step": 2560
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 4.75,
|
| 1853 |
+
"grad_norm": 2.4942688941955566,
|
| 1854 |
+
"learning_rate": 0.0026222222222222224,
|
| 1855 |
+
"loss": 0.5423,
|
| 1856 |
+
"step": 2570
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 4.77,
|
| 1860 |
+
"grad_norm": 1.556859016418457,
|
| 1861 |
+
"learning_rate": 0.002612962962962963,
|
| 1862 |
+
"loss": 0.5783,
|
| 1863 |
+
"step": 2580
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 4.79,
|
| 1867 |
+
"grad_norm": 3.010946750640869,
|
| 1868 |
+
"learning_rate": 0.0026037037037037036,
|
| 1869 |
+
"loss": 0.5786,
|
| 1870 |
+
"step": 2590
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 4.81,
|
| 1874 |
+
"grad_norm": 1.5896031856536865,
|
| 1875 |
+
"learning_rate": 0.0025944444444444444,
|
| 1876 |
+
"loss": 0.6075,
|
| 1877 |
+
"step": 2600
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 4.83,
|
| 1881 |
+
"grad_norm": 1.9418363571166992,
|
| 1882 |
+
"learning_rate": 0.0025851851851851856,
|
| 1883 |
+
"loss": 0.5482,
|
| 1884 |
+
"step": 2610
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 4.85,
|
| 1888 |
+
"grad_norm": 1.9787338972091675,
|
| 1889 |
+
"learning_rate": 0.002576851851851852,
|
| 1890 |
+
"loss": 0.6026,
|
| 1891 |
+
"step": 2620
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 4.87,
|
| 1895 |
+
"grad_norm": 2.7021665573120117,
|
| 1896 |
+
"learning_rate": 0.0025675925925925927,
|
| 1897 |
+
"loss": 0.4977,
|
| 1898 |
+
"step": 2630
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 4.88,
|
| 1902 |
+
"grad_norm": 2.2835893630981445,
|
| 1903 |
+
"learning_rate": 0.0025583333333333335,
|
| 1904 |
+
"loss": 0.5375,
|
| 1905 |
+
"step": 2640
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 4.9,
|
| 1909 |
+
"grad_norm": 1.9748303890228271,
|
| 1910 |
+
"learning_rate": 0.0025490740740740743,
|
| 1911 |
+
"loss": 0.5263,
|
| 1912 |
+
"step": 2650
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 4.92,
|
| 1916 |
+
"grad_norm": 1.6271171569824219,
|
| 1917 |
+
"learning_rate": 0.0025398148148148146,
|
| 1918 |
+
"loss": 0.5275,
|
| 1919 |
+
"step": 2660
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 4.94,
|
| 1923 |
+
"grad_norm": 2.3140103816986084,
|
| 1924 |
+
"learning_rate": 0.0025305555555555554,
|
| 1925 |
+
"loss": 0.5493,
|
| 1926 |
+
"step": 2670
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 4.96,
|
| 1930 |
+
"grad_norm": 2.2506508827209473,
|
| 1931 |
+
"learning_rate": 0.0025212962962962967,
|
| 1932 |
+
"loss": 0.5756,
|
| 1933 |
+
"step": 2680
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 4.98,
|
| 1937 |
+
"grad_norm": 2.5812437534332275,
|
| 1938 |
+
"learning_rate": 0.002512037037037037,
|
| 1939 |
+
"loss": 0.511,
|
| 1940 |
+
"step": 2690
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 5.0,
|
| 1944 |
+
"grad_norm": 1.886813998222351,
|
| 1945 |
+
"learning_rate": 0.002502777777777778,
|
| 1946 |
+
"loss": 0.5509,
|
| 1947 |
+
"step": 2700
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 5.0,
|
| 1951 |
+
"eval_accuracy": 0.9770451394238175,
|
| 1952 |
+
"eval_f1": 0.9783131597050811,
|
| 1953 |
+
"eval_loss": 0.06713523715734482,
|
| 1954 |
+
"eval_precision": 0.9775503610869588,
|
| 1955 |
+
"eval_recall": 0.9795962655538212,
|
| 1956 |
+
"eval_runtime": 55.3992,
|
| 1957 |
+
"eval_samples_per_second": 117.168,
|
| 1958 |
+
"eval_steps_per_second": 7.329,
|
| 1959 |
+
"step": 2702
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 5.01,
|
| 1963 |
+
"grad_norm": 2.990206241607666,
|
| 1964 |
+
"learning_rate": 0.0024935185185185186,
|
| 1965 |
+
"loss": 0.5212,
|
| 1966 |
+
"step": 2710
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 5.03,
|
| 1970 |
+
"grad_norm": 1.2570441961288452,
|
| 1971 |
+
"learning_rate": 0.0024842592592592594,
|
| 1972 |
+
"loss": 0.4703,
|
| 1973 |
+
"step": 2720
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 5.05,
|
| 1977 |
+
"grad_norm": 1.6648197174072266,
|
| 1978 |
+
"learning_rate": 0.002475,
|
| 1979 |
+
"loss": 0.6173,
|
| 1980 |
+
"step": 2730
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 5.07,
|
| 1984 |
+
"grad_norm": 1.9291369915008545,
|
| 1985 |
+
"learning_rate": 0.002465740740740741,
|
| 1986 |
+
"loss": 0.5929,
|
| 1987 |
+
"step": 2740
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 5.09,
|
| 1991 |
+
"grad_norm": 2.2894320487976074,
|
| 1992 |
+
"learning_rate": 0.0024564814814814814,
|
| 1993 |
+
"loss": 0.5282,
|
| 1994 |
+
"step": 2750
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 5.11,
|
| 1998 |
+
"grad_norm": NaN,
|
| 1999 |
+
"learning_rate": 0.002448148148148148,
|
| 2000 |
+
"loss": 0.5085,
|
| 2001 |
+
"step": 2760
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 5.12,
|
| 2005 |
+
"grad_norm": 2.4065709114074707,
|
| 2006 |
+
"learning_rate": 0.002438888888888889,
|
| 2007 |
+
"loss": 0.5983,
|
| 2008 |
+
"step": 2770
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 5.14,
|
| 2012 |
+
"grad_norm": 1.4784891605377197,
|
| 2013 |
+
"learning_rate": 0.0024296296296296297,
|
| 2014 |
+
"loss": 0.5586,
|
| 2015 |
+
"step": 2780
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 5.16,
|
| 2019 |
+
"grad_norm": 3.6442089080810547,
|
| 2020 |
+
"learning_rate": 0.0024203703703703705,
|
| 2021 |
+
"loss": 0.5824,
|
| 2022 |
+
"step": 2790
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 5.18,
|
| 2026 |
+
"grad_norm": 1.4887707233428955,
|
| 2027 |
+
"learning_rate": 0.0024111111111111113,
|
| 2028 |
+
"loss": 0.6023,
|
| 2029 |
+
"step": 2800
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 5.2,
|
| 2033 |
+
"grad_norm": 1.4839690923690796,
|
| 2034 |
+
"learning_rate": 0.002401851851851852,
|
| 2035 |
+
"loss": 0.5486,
|
| 2036 |
+
"step": 2810
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 5.22,
|
| 2040 |
+
"grad_norm": 1.522079586982727,
|
| 2041 |
+
"learning_rate": 0.0023925925925925924,
|
| 2042 |
+
"loss": 0.5803,
|
| 2043 |
+
"step": 2820
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 5.24,
|
| 2047 |
+
"grad_norm": 1.6155500411987305,
|
| 2048 |
+
"learning_rate": 0.0023833333333333337,
|
| 2049 |
+
"loss": 0.5765,
|
| 2050 |
+
"step": 2830
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 5.25,
|
| 2054 |
+
"grad_norm": 1.4608234167099,
|
| 2055 |
+
"learning_rate": 0.002374074074074074,
|
| 2056 |
+
"loss": 0.5163,
|
| 2057 |
+
"step": 2840
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 5.27,
|
| 2061 |
+
"grad_norm": 1.8083865642547607,
|
| 2062 |
+
"learning_rate": 0.002364814814814815,
|
| 2063 |
+
"loss": 0.5533,
|
| 2064 |
+
"step": 2850
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 5.29,
|
| 2068 |
+
"grad_norm": 1.7674206495285034,
|
| 2069 |
+
"learning_rate": 0.0023555555555555556,
|
| 2070 |
+
"loss": 0.5558,
|
| 2071 |
+
"step": 2860
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 5.31,
|
| 2075 |
+
"grad_norm": 1.7677381038665771,
|
| 2076 |
+
"learning_rate": 0.0023462962962962964,
|
| 2077 |
+
"loss": 0.551,
|
| 2078 |
+
"step": 2870
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 5.33,
|
| 2082 |
+
"grad_norm": 2.006486415863037,
|
| 2083 |
+
"learning_rate": 0.0023370370370370368,
|
| 2084 |
+
"loss": 0.5311,
|
| 2085 |
+
"step": 2880
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 5.35,
|
| 2089 |
+
"grad_norm": 1.3190593719482422,
|
| 2090 |
+
"learning_rate": 0.002327777777777778,
|
| 2091 |
+
"loss": 0.5462,
|
| 2092 |
+
"step": 2890
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 5.37,
|
| 2096 |
+
"grad_norm": 2.1701364517211914,
|
| 2097 |
+
"learning_rate": 0.002318518518518519,
|
| 2098 |
+
"loss": 0.5056,
|
| 2099 |
+
"step": 2900
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 5.38,
|
| 2103 |
+
"grad_norm": 3.071043014526367,
|
| 2104 |
+
"learning_rate": 0.002309259259259259,
|
| 2105 |
+
"loss": 0.5611,
|
| 2106 |
+
"step": 2910
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 5.4,
|
| 2110 |
+
"grad_norm": 1.7673370838165283,
|
| 2111 |
+
"learning_rate": 0.0023,
|
| 2112 |
+
"loss": 0.5369,
|
| 2113 |
+
"step": 2920
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 5.42,
|
| 2117 |
+
"grad_norm": 1.6603857278823853,
|
| 2118 |
+
"learning_rate": 0.0022907407407407408,
|
| 2119 |
+
"loss": 0.6108,
|
| 2120 |
+
"step": 2930
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 5.44,
|
| 2124 |
+
"grad_norm": 3.754216194152832,
|
| 2125 |
+
"learning_rate": 0.0022814814814814816,
|
| 2126 |
+
"loss": 0.5355,
|
| 2127 |
+
"step": 2940
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 5.46,
|
| 2131 |
+
"grad_norm": 2.2037136554718018,
|
| 2132 |
+
"learning_rate": 0.0022722222222222224,
|
| 2133 |
+
"loss": 0.4998,
|
| 2134 |
+
"step": 2950
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 5.48,
|
| 2138 |
+
"grad_norm": 2.2701072692871094,
|
| 2139 |
+
"learning_rate": 0.002262962962962963,
|
| 2140 |
+
"loss": 0.5672,
|
| 2141 |
+
"step": 2960
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 5.49,
|
| 2145 |
+
"grad_norm": 2.246368408203125,
|
| 2146 |
+
"learning_rate": 0.0022537037037037035,
|
| 2147 |
+
"loss": 0.5309,
|
| 2148 |
+
"step": 2970
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 5.51,
|
| 2152 |
+
"grad_norm": 1.4481878280639648,
|
| 2153 |
+
"learning_rate": 0.0022444444444444443,
|
| 2154 |
+
"loss": 0.4786,
|
| 2155 |
+
"step": 2980
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 5.53,
|
| 2159 |
+
"grad_norm": 1.7569130659103394,
|
| 2160 |
+
"learning_rate": 0.002235185185185185,
|
| 2161 |
+
"loss": 0.579,
|
| 2162 |
+
"step": 2990
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 5.55,
|
| 2166 |
+
"grad_norm": 2.219710350036621,
|
| 2167 |
+
"learning_rate": 0.002225925925925926,
|
| 2168 |
+
"loss": 0.4768,
|
| 2169 |
+
"step": 3000
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 5.57,
|
| 2173 |
+
"grad_norm": 1.3611927032470703,
|
| 2174 |
+
"learning_rate": 0.0022166666666666667,
|
| 2175 |
+
"loss": 0.4504,
|
| 2176 |
+
"step": 3010
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 5.59,
|
| 2180 |
+
"grad_norm": 1.528106451034546,
|
| 2181 |
+
"learning_rate": 0.0022074074074074075,
|
| 2182 |
+
"loss": 0.491,
|
| 2183 |
+
"step": 3020
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 5.61,
|
| 2187 |
+
"grad_norm": 1.934448003768921,
|
| 2188 |
+
"learning_rate": 0.0021981481481481483,
|
| 2189 |
+
"loss": 0.5553,
|
| 2190 |
+
"step": 3030
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 5.62,
|
| 2194 |
+
"grad_norm": 1.9539951086044312,
|
| 2195 |
+
"learning_rate": 0.002188888888888889,
|
| 2196 |
+
"loss": 0.514,
|
| 2197 |
+
"step": 3040
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 5.64,
|
| 2201 |
+
"grad_norm": 3.347055196762085,
|
| 2202 |
+
"learning_rate": 0.00217962962962963,
|
| 2203 |
+
"loss": 0.565,
|
| 2204 |
+
"step": 3050
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 5.66,
|
| 2208 |
+
"grad_norm": 2.0984296798706055,
|
| 2209 |
+
"learning_rate": 0.0021703703703703702,
|
| 2210 |
+
"loss": 0.4917,
|
| 2211 |
+
"step": 3060
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 5.68,
|
| 2215 |
+
"grad_norm": 4.030628681182861,
|
| 2216 |
+
"learning_rate": 0.002161111111111111,
|
| 2217 |
+
"loss": 0.5485,
|
| 2218 |
+
"step": 3070
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 5.7,
|
| 2222 |
+
"grad_norm": 1.62152898311615,
|
| 2223 |
+
"learning_rate": 0.002151851851851852,
|
| 2224 |
+
"loss": 0.5468,
|
| 2225 |
+
"step": 3080
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 5.72,
|
| 2229 |
+
"grad_norm": 2.245067596435547,
|
| 2230 |
+
"learning_rate": 0.0021425925925925926,
|
| 2231 |
+
"loss": 0.4992,
|
| 2232 |
+
"step": 3090
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 5.74,
|
| 2236 |
+
"grad_norm": 2.4511606693267822,
|
| 2237 |
+
"learning_rate": 0.0021333333333333334,
|
| 2238 |
+
"loss": 0.5478,
|
| 2239 |
+
"step": 3100
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 5.75,
|
| 2243 |
+
"grad_norm": 3.2524254322052,
|
| 2244 |
+
"learning_rate": 0.0021240740740740742,
|
| 2245 |
+
"loss": 0.5006,
|
| 2246 |
+
"step": 3110
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 5.77,
|
| 2250 |
+
"grad_norm": 1.8704936504364014,
|
| 2251 |
+
"learning_rate": 0.002114814814814815,
|
| 2252 |
+
"loss": 0.5177,
|
| 2253 |
+
"step": 3120
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 5.79,
|
| 2257 |
+
"grad_norm": 1.6047933101654053,
|
| 2258 |
+
"learning_rate": 0.0021064814814814813,
|
| 2259 |
+
"loss": 0.5491,
|
| 2260 |
+
"step": 3130
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 5.81,
|
| 2264 |
+
"grad_norm": 2.038342237472534,
|
| 2265 |
+
"learning_rate": 0.002097222222222222,
|
| 2266 |
+
"loss": 0.5096,
|
| 2267 |
+
"step": 3140
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 5.83,
|
| 2271 |
+
"grad_norm": 1.1834994554519653,
|
| 2272 |
+
"learning_rate": 0.002087962962962963,
|
| 2273 |
+
"loss": 0.4611,
|
| 2274 |
+
"step": 3150
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 5.85,
|
| 2278 |
+
"grad_norm": 3.6137051582336426,
|
| 2279 |
+
"learning_rate": 0.0020787037037037037,
|
| 2280 |
+
"loss": 0.5056,
|
| 2281 |
+
"step": 3160
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 5.86,
|
| 2285 |
+
"grad_norm": 1.7401528358459473,
|
| 2286 |
+
"learning_rate": 0.0020694444444444445,
|
| 2287 |
+
"loss": 0.5285,
|
| 2288 |
+
"step": 3170
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 5.88,
|
| 2292 |
+
"grad_norm": 1.1825543642044067,
|
| 2293 |
+
"learning_rate": 0.0020601851851851853,
|
| 2294 |
+
"loss": 0.5166,
|
| 2295 |
+
"step": 3180
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 5.9,
|
| 2299 |
+
"grad_norm": 1.469771385192871,
|
| 2300 |
+
"learning_rate": 0.002050925925925926,
|
| 2301 |
+
"loss": 0.5246,
|
| 2302 |
+
"step": 3190
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 5.92,
|
| 2306 |
+
"grad_norm": 1.9187328815460205,
|
| 2307 |
+
"learning_rate": 0.0020416666666666665,
|
| 2308 |
+
"loss": 0.5687,
|
| 2309 |
+
"step": 3200
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 5.94,
|
| 2313 |
+
"grad_norm": 1.27570641040802,
|
| 2314 |
+
"learning_rate": 0.0020324074074074077,
|
| 2315 |
+
"loss": 0.5367,
|
| 2316 |
+
"step": 3210
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 5.96,
|
| 2320 |
+
"grad_norm": 2.436992883682251,
|
| 2321 |
+
"learning_rate": 0.002023148148148148,
|
| 2322 |
+
"loss": 0.5108,
|
| 2323 |
+
"step": 3220
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 5.98,
|
| 2327 |
+
"grad_norm": 2.1716506481170654,
|
| 2328 |
+
"learning_rate": 0.002013888888888889,
|
| 2329 |
+
"loss": 0.5081,
|
| 2330 |
+
"step": 3230
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 5.99,
|
| 2334 |
+
"grad_norm": 2.3047661781311035,
|
| 2335 |
+
"learning_rate": 0.0020046296296296296,
|
| 2336 |
+
"loss": 0.5746,
|
| 2337 |
+
"step": 3240
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 6.0,
|
| 2341 |
+
"eval_accuracy": 0.9753504852873209,
|
| 2342 |
+
"eval_f1": 0.9794812735446677,
|
| 2343 |
+
"eval_loss": 0.06419458985328674,
|
| 2344 |
+
"eval_precision": 0.9810236528071776,
|
| 2345 |
+
"eval_recall": 0.9788061689635984,
|
| 2346 |
+
"eval_runtime": 55.1158,
|
| 2347 |
+
"eval_samples_per_second": 117.77,
|
| 2348 |
+
"eval_steps_per_second": 7.366,
|
| 2349 |
+
"step": 3243
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 6.01,
|
| 2353 |
+
"grad_norm": 2.193244457244873,
|
| 2354 |
+
"learning_rate": 0.0019953703703703704,
|
| 2355 |
+
"loss": 0.4844,
|
| 2356 |
+
"step": 3250
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 6.03,
|
| 2360 |
+
"grad_norm": 1.7774107456207275,
|
| 2361 |
+
"learning_rate": 0.0019861111111111112,
|
| 2362 |
+
"loss": 0.5153,
|
| 2363 |
+
"step": 3260
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 6.05,
|
| 2367 |
+
"grad_norm": 1.8430052995681763,
|
| 2368 |
+
"learning_rate": 0.001976851851851852,
|
| 2369 |
+
"loss": 0.5332,
|
| 2370 |
+
"step": 3270
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 6.07,
|
| 2374 |
+
"grad_norm": 2.268313407897949,
|
| 2375 |
+
"learning_rate": 0.001967592592592593,
|
| 2376 |
+
"loss": 0.4762,
|
| 2377 |
+
"step": 3280
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 6.09,
|
| 2381 |
+
"grad_norm": 2.3177573680877686,
|
| 2382 |
+
"learning_rate": 0.001958333333333333,
|
| 2383 |
+
"loss": 0.5045,
|
| 2384 |
+
"step": 3290
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 6.11,
|
| 2388 |
+
"grad_norm": 2.091614007949829,
|
| 2389 |
+
"learning_rate": 0.0019490740740740742,
|
| 2390 |
+
"loss": 0.426,
|
| 2391 |
+
"step": 3300
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 6.12,
|
| 2395 |
+
"grad_norm": 2.227275848388672,
|
| 2396 |
+
"learning_rate": 0.0019398148148148148,
|
| 2397 |
+
"loss": 0.5102,
|
| 2398 |
+
"step": 3310
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 6.14,
|
| 2402 |
+
"grad_norm": 1.47825026512146,
|
| 2403 |
+
"learning_rate": 0.0019305555555555556,
|
| 2404 |
+
"loss": 0.4559,
|
| 2405 |
+
"step": 3320
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 6.16,
|
| 2409 |
+
"grad_norm": 2.3970351219177246,
|
| 2410 |
+
"learning_rate": 0.0019212962962962962,
|
| 2411 |
+
"loss": 0.5202,
|
| 2412 |
+
"step": 3330
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 6.18,
|
| 2416 |
+
"grad_norm": 1.6059942245483398,
|
| 2417 |
+
"learning_rate": 0.0019120370370370372,
|
| 2418 |
+
"loss": 0.4684,
|
| 2419 |
+
"step": 3340
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 6.2,
|
| 2423 |
+
"grad_norm": 1.4332858324050903,
|
| 2424 |
+
"learning_rate": 0.0019027777777777778,
|
| 2425 |
+
"loss": 0.5029,
|
| 2426 |
+
"step": 3350
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 6.22,
|
| 2430 |
+
"grad_norm": 2.158965587615967,
|
| 2431 |
+
"learning_rate": 0.0018935185185185186,
|
| 2432 |
+
"loss": 0.5185,
|
| 2433 |
+
"step": 3360
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 6.23,
|
| 2437 |
+
"grad_norm": 2.077367067337036,
|
| 2438 |
+
"learning_rate": 0.0018842592592592591,
|
| 2439 |
+
"loss": 0.472,
|
| 2440 |
+
"step": 3370
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 6.25,
|
| 2444 |
+
"grad_norm": 2.0703649520874023,
|
| 2445 |
+
"learning_rate": 0.001875,
|
| 2446 |
+
"loss": 0.4818,
|
| 2447 |
+
"step": 3380
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 6.27,
|
| 2451 |
+
"grad_norm": 2.000805139541626,
|
| 2452 |
+
"learning_rate": 0.001865740740740741,
|
| 2453 |
+
"loss": 0.4643,
|
| 2454 |
+
"step": 3390
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 6.29,
|
| 2458 |
+
"grad_norm": 1.556027889251709,
|
| 2459 |
+
"learning_rate": 0.0018564814814814815,
|
| 2460 |
+
"loss": 0.468,
|
| 2461 |
+
"step": 3400
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 6.31,
|
| 2465 |
+
"grad_norm": 1.2694203853607178,
|
| 2466 |
+
"learning_rate": 0.0018472222222222223,
|
| 2467 |
+
"loss": 0.4437,
|
| 2468 |
+
"step": 3410
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 6.33,
|
| 2472 |
+
"grad_norm": 2.0669643878936768,
|
| 2473 |
+
"learning_rate": 0.001837962962962963,
|
| 2474 |
+
"loss": 0.4565,
|
| 2475 |
+
"step": 3420
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 6.35,
|
| 2479 |
+
"grad_norm": 1.6692304611206055,
|
| 2480 |
+
"learning_rate": 0.001828703703703704,
|
| 2481 |
+
"loss": 0.4573,
|
| 2482 |
+
"step": 3430
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 6.36,
|
| 2486 |
+
"grad_norm": 1.6164546012878418,
|
| 2487 |
+
"learning_rate": 0.0018194444444444445,
|
| 2488 |
+
"loss": 0.4282,
|
| 2489 |
+
"step": 3440
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 6.38,
|
| 2493 |
+
"grad_norm": 1.9918891191482544,
|
| 2494 |
+
"learning_rate": 0.0018101851851851853,
|
| 2495 |
+
"loss": 0.4522,
|
| 2496 |
+
"step": 3450
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 6.4,
|
| 2500 |
+
"grad_norm": 1.4127854108810425,
|
| 2501 |
+
"learning_rate": 0.0018009259259259259,
|
| 2502 |
+
"loss": 0.5317,
|
| 2503 |
+
"step": 3460
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 6.42,
|
| 2507 |
+
"grad_norm": 1.8190535306930542,
|
| 2508 |
+
"learning_rate": 0.0017916666666666667,
|
| 2509 |
+
"loss": 0.5193,
|
| 2510 |
+
"step": 3470
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 6.44,
|
| 2514 |
+
"grad_norm": 1.8793113231658936,
|
| 2515 |
+
"learning_rate": 0.0017824074074074072,
|
| 2516 |
+
"loss": 0.5322,
|
| 2517 |
+
"step": 3480
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 6.46,
|
| 2521 |
+
"grad_norm": 1.1124266386032104,
|
| 2522 |
+
"learning_rate": 0.0017731481481481483,
|
| 2523 |
+
"loss": 0.4849,
|
| 2524 |
+
"step": 3490
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 6.48,
|
| 2528 |
+
"grad_norm": 1.5575320720672607,
|
| 2529 |
+
"learning_rate": 0.001763888888888889,
|
| 2530 |
+
"loss": 0.5254,
|
| 2531 |
+
"step": 3500
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 6.49,
|
| 2535 |
+
"grad_norm": 1.3481545448303223,
|
| 2536 |
+
"learning_rate": 0.0017546296296296296,
|
| 2537 |
+
"loss": 0.4869,
|
| 2538 |
+
"step": 3510
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 6.51,
|
| 2542 |
+
"grad_norm": 1.0486959218978882,
|
| 2543 |
+
"learning_rate": 0.0017453703703703704,
|
| 2544 |
+
"loss": 0.5048,
|
| 2545 |
+
"step": 3520
|
| 2546 |
+
},
|
| 2547 |
+
{
|
| 2548 |
+
"epoch": 6.53,
|
| 2549 |
+
"grad_norm": 1.848264217376709,
|
| 2550 |
+
"learning_rate": 0.001736111111111111,
|
| 2551 |
+
"loss": 0.5064,
|
| 2552 |
+
"step": 3530
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 6.55,
|
| 2556 |
+
"grad_norm": 1.6750601530075073,
|
| 2557 |
+
"learning_rate": 0.001726851851851852,
|
| 2558 |
+
"loss": 0.4459,
|
| 2559 |
+
"step": 3540
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 6.57,
|
| 2563 |
+
"grad_norm": 1.5514721870422363,
|
| 2564 |
+
"learning_rate": 0.0017175925925925926,
|
| 2565 |
+
"loss": 0.4382,
|
| 2566 |
+
"step": 3550
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 6.59,
|
| 2570 |
+
"grad_norm": 1.3411380052566528,
|
| 2571 |
+
"learning_rate": 0.0017083333333333334,
|
| 2572 |
+
"loss": 0.4007,
|
| 2573 |
+
"step": 3560
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 6.6,
|
| 2577 |
+
"grad_norm": 1.8607478141784668,
|
| 2578 |
+
"learning_rate": 0.001699074074074074,
|
| 2579 |
+
"loss": 0.4776,
|
| 2580 |
+
"step": 3570
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 6.62,
|
| 2584 |
+
"grad_norm": 1.6788519620895386,
|
| 2585 |
+
"learning_rate": 0.0016898148148148148,
|
| 2586 |
+
"loss": 0.4499,
|
| 2587 |
+
"step": 3580
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 6.64,
|
| 2591 |
+
"grad_norm": 1.4055275917053223,
|
| 2592 |
+
"learning_rate": 0.0016805555555555558,
|
| 2593 |
+
"loss": 0.5017,
|
| 2594 |
+
"step": 3590
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 6.66,
|
| 2598 |
+
"grad_norm": 1.9145076274871826,
|
| 2599 |
+
"learning_rate": 0.0016712962962962964,
|
| 2600 |
+
"loss": 0.4196,
|
| 2601 |
+
"step": 3600
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 6.68,
|
| 2605 |
+
"grad_norm": 1.7690577507019043,
|
| 2606 |
+
"learning_rate": 0.0016620370370370372,
|
| 2607 |
+
"loss": 0.4792,
|
| 2608 |
+
"step": 3610
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 6.7,
|
| 2612 |
+
"grad_norm": 1.713212490081787,
|
| 2613 |
+
"learning_rate": 0.0016527777777777778,
|
| 2614 |
+
"loss": 0.4418,
|
| 2615 |
+
"step": 3620
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 6.72,
|
| 2619 |
+
"grad_norm": 2.4528722763061523,
|
| 2620 |
+
"learning_rate": 0.0016435185185185185,
|
| 2621 |
+
"loss": 0.4519,
|
| 2622 |
+
"step": 3630
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 6.73,
|
| 2626 |
+
"grad_norm": 1.8348699808120728,
|
| 2627 |
+
"learning_rate": 0.0016342592592592591,
|
| 2628 |
+
"loss": 0.3959,
|
| 2629 |
+
"step": 3640
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 6.75,
|
| 2633 |
+
"grad_norm": 1.38786780834198,
|
| 2634 |
+
"learning_rate": 0.0016250000000000001,
|
| 2635 |
+
"loss": 0.4433,
|
| 2636 |
+
"step": 3650
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 6.77,
|
| 2640 |
+
"grad_norm": 0.8413279056549072,
|
| 2641 |
+
"learning_rate": 0.0016157407407407407,
|
| 2642 |
+
"loss": 0.4869,
|
| 2643 |
+
"step": 3660
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 6.79,
|
| 2647 |
+
"grad_norm": 1.3703649044036865,
|
| 2648 |
+
"learning_rate": 0.0016064814814814815,
|
| 2649 |
+
"loss": 0.5021,
|
| 2650 |
+
"step": 3670
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 6.81,
|
| 2654 |
+
"grad_norm": 1.6931146383285522,
|
| 2655 |
+
"learning_rate": 0.001597222222222222,
|
| 2656 |
+
"loss": 0.416,
|
| 2657 |
+
"step": 3680
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 6.83,
|
| 2661 |
+
"grad_norm": 1.314029574394226,
|
| 2662 |
+
"learning_rate": 0.001587962962962963,
|
| 2663 |
+
"loss": 0.4489,
|
| 2664 |
+
"step": 3690
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 6.85,
|
| 2668 |
+
"grad_norm": 1.6631927490234375,
|
| 2669 |
+
"learning_rate": 0.001578703703703704,
|
| 2670 |
+
"loss": 0.4334,
|
| 2671 |
+
"step": 3700
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 6.86,
|
| 2675 |
+
"grad_norm": 1.821455478668213,
|
| 2676 |
+
"learning_rate": 0.0015694444444444445,
|
| 2677 |
+
"loss": 0.4307,
|
| 2678 |
+
"step": 3710
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 6.88,
|
| 2682 |
+
"grad_norm": 1.4613134860992432,
|
| 2683 |
+
"learning_rate": 0.0015601851851851853,
|
| 2684 |
+
"loss": 0.414,
|
| 2685 |
+
"step": 3720
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 6.9,
|
| 2689 |
+
"grad_norm": 1.9156243801116943,
|
| 2690 |
+
"learning_rate": 0.0015509259259259259,
|
| 2691 |
+
"loss": 0.5022,
|
| 2692 |
+
"step": 3730
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 6.92,
|
| 2696 |
+
"grad_norm": 1.9036259651184082,
|
| 2697 |
+
"learning_rate": 0.0015416666666666669,
|
| 2698 |
+
"loss": 0.4562,
|
| 2699 |
+
"step": 3740
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 6.94,
|
| 2703 |
+
"grad_norm": 1.6884894371032715,
|
| 2704 |
+
"learning_rate": 0.0015324074074074075,
|
| 2705 |
+
"loss": 0.4418,
|
| 2706 |
+
"step": 3750
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 6.96,
|
| 2710 |
+
"grad_norm": 1.4793485403060913,
|
| 2711 |
+
"learning_rate": 0.0015231481481481483,
|
| 2712 |
+
"loss": 0.4091,
|
| 2713 |
+
"step": 3760
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 6.98,
|
| 2717 |
+
"grad_norm": 1.9029099941253662,
|
| 2718 |
+
"learning_rate": 0.0015138888888888888,
|
| 2719 |
+
"loss": 0.4827,
|
| 2720 |
+
"step": 3770
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 6.99,
|
| 2724 |
+
"grad_norm": 1.2703421115875244,
|
| 2725 |
+
"learning_rate": 0.0015046296296296296,
|
| 2726 |
+
"loss": 0.4066,
|
| 2727 |
+
"step": 3780
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 7.0,
|
| 2731 |
+
"eval_accuracy": 0.9565552303189031,
|
| 2732 |
+
"eval_f1": 0.96141521763694,
|
| 2733 |
+
"eval_loss": 0.11959208548069,
|
| 2734 |
+
"eval_precision": 0.9693468985322017,
|
| 2735 |
+
"eval_recall": 0.9563045431610278,
|
| 2736 |
+
"eval_runtime": 55.0215,
|
| 2737 |
+
"eval_samples_per_second": 117.972,
|
| 2738 |
+
"eval_steps_per_second": 7.379,
|
| 2739 |
+
"step": 3783
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 7.01,
|
| 2743 |
+
"grad_norm": 1.4109677076339722,
|
| 2744 |
+
"learning_rate": 0.0014953703703703702,
|
| 2745 |
+
"loss": 0.434,
|
| 2746 |
+
"step": 3790
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 7.03,
|
| 2750 |
+
"grad_norm": 1.6477872133255005,
|
| 2751 |
+
"learning_rate": 0.0014861111111111112,
|
| 2752 |
+
"loss": 0.466,
|
| 2753 |
+
"step": 3800
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 7.05,
|
| 2757 |
+
"grad_norm": 0.9931581020355225,
|
| 2758 |
+
"learning_rate": 0.001476851851851852,
|
| 2759 |
+
"loss": 0.4282,
|
| 2760 |
+
"step": 3810
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 7.07,
|
| 2764 |
+
"grad_norm": 2.313886880874634,
|
| 2765 |
+
"learning_rate": 0.0014675925925925926,
|
| 2766 |
+
"loss": 0.4249,
|
| 2767 |
+
"step": 3820
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 7.09,
|
| 2771 |
+
"grad_norm": 1.5501474142074585,
|
| 2772 |
+
"learning_rate": 0.0014583333333333334,
|
| 2773 |
+
"loss": 0.4062,
|
| 2774 |
+
"step": 3830
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 7.1,
|
| 2778 |
+
"grad_norm": 2.267270088195801,
|
| 2779 |
+
"learning_rate": 0.001449074074074074,
|
| 2780 |
+
"loss": 0.4546,
|
| 2781 |
+
"step": 3840
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 7.12,
|
| 2785 |
+
"grad_norm": 2.5071542263031006,
|
| 2786 |
+
"learning_rate": 0.001439814814814815,
|
| 2787 |
+
"loss": 0.4828,
|
| 2788 |
+
"step": 3850
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 7.14,
|
| 2792 |
+
"grad_norm": 1.464120864868164,
|
| 2793 |
+
"learning_rate": 0.0014305555555555556,
|
| 2794 |
+
"loss": 0.416,
|
| 2795 |
+
"step": 3860
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 7.16,
|
| 2799 |
+
"grad_norm": 1.3284567594528198,
|
| 2800 |
+
"learning_rate": 0.0014212962962962964,
|
| 2801 |
+
"loss": 0.3704,
|
| 2802 |
+
"step": 3870
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 7.18,
|
| 2806 |
+
"grad_norm": 1.5376476049423218,
|
| 2807 |
+
"learning_rate": 0.001412037037037037,
|
| 2808 |
+
"loss": 0.454,
|
| 2809 |
+
"step": 3880
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 7.2,
|
| 2813 |
+
"grad_norm": 1.0760722160339355,
|
| 2814 |
+
"learning_rate": 0.0014027777777777777,
|
| 2815 |
+
"loss": 0.4538,
|
| 2816 |
+
"step": 3890
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 7.22,
|
| 2820 |
+
"grad_norm": 2.0362236499786377,
|
| 2821 |
+
"learning_rate": 0.0013935185185185188,
|
| 2822 |
+
"loss": 0.4178,
|
| 2823 |
+
"step": 3900
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 7.23,
|
| 2827 |
+
"grad_norm": 1.949922800064087,
|
| 2828 |
+
"learning_rate": 0.0013842592592592593,
|
| 2829 |
+
"loss": 0.3457,
|
| 2830 |
+
"step": 3910
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 7.25,
|
| 2834 |
+
"grad_norm": 1.173089623451233,
|
| 2835 |
+
"learning_rate": 0.0013750000000000001,
|
| 2836 |
+
"loss": 0.389,
|
| 2837 |
+
"step": 3920
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 7.27,
|
| 2841 |
+
"grad_norm": 1.4733856916427612,
|
| 2842 |
+
"learning_rate": 0.0013657407407407407,
|
| 2843 |
+
"loss": 0.456,
|
| 2844 |
+
"step": 3930
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 7.29,
|
| 2848 |
+
"grad_norm": 1.5104173421859741,
|
| 2849 |
+
"learning_rate": 0.0013564814814814815,
|
| 2850 |
+
"loss": 0.406,
|
| 2851 |
+
"step": 3940
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 7.31,
|
| 2855 |
+
"grad_norm": 2.4948642253875732,
|
| 2856 |
+
"learning_rate": 0.001347222222222222,
|
| 2857 |
+
"loss": 0.4195,
|
| 2858 |
+
"step": 3950
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 7.33,
|
| 2862 |
+
"grad_norm": 2.137315511703491,
|
| 2863 |
+
"learning_rate": 0.001337962962962963,
|
| 2864 |
+
"loss": 0.4857,
|
| 2865 |
+
"step": 3960
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 7.35,
|
| 2869 |
+
"grad_norm": 1.20357084274292,
|
| 2870 |
+
"learning_rate": 0.0013287037037037037,
|
| 2871 |
+
"loss": 0.4287,
|
| 2872 |
+
"step": 3970
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 7.36,
|
| 2876 |
+
"grad_norm": 1.5241217613220215,
|
| 2877 |
+
"learning_rate": 0.0013194444444444445,
|
| 2878 |
+
"loss": 0.4535,
|
| 2879 |
+
"step": 3980
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 7.38,
|
| 2883 |
+
"grad_norm": 1.4119207859039307,
|
| 2884 |
+
"learning_rate": 0.001310185185185185,
|
| 2885 |
+
"loss": 0.486,
|
| 2886 |
+
"step": 3990
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 7.4,
|
| 2890 |
+
"grad_norm": 1.3262003660202026,
|
| 2891 |
+
"learning_rate": 0.0013009259259259259,
|
| 2892 |
+
"loss": 0.4833,
|
| 2893 |
+
"step": 4000
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 7.42,
|
| 2897 |
+
"grad_norm": 1.0021650791168213,
|
| 2898 |
+
"learning_rate": 0.0012916666666666669,
|
| 2899 |
+
"loss": 0.3996,
|
| 2900 |
+
"step": 4010
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 7.44,
|
| 2904 |
+
"grad_norm": 1.8423185348510742,
|
| 2905 |
+
"learning_rate": 0.0012824074074074075,
|
| 2906 |
+
"loss": 0.4569,
|
| 2907 |
+
"step": 4020
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 7.46,
|
| 2911 |
+
"grad_norm": 2.010634183883667,
|
| 2912 |
+
"learning_rate": 0.0012731481481481483,
|
| 2913 |
+
"loss": 0.4673,
|
| 2914 |
+
"step": 4030
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 7.47,
|
| 2918 |
+
"grad_norm": 2.974384307861328,
|
| 2919 |
+
"learning_rate": 0.0012638888888888888,
|
| 2920 |
+
"loss": 0.3992,
|
| 2921 |
+
"step": 4040
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 7.49,
|
| 2925 |
+
"grad_norm": 1.022455096244812,
|
| 2926 |
+
"learning_rate": 0.0012546296296296296,
|
| 2927 |
+
"loss": 0.3598,
|
| 2928 |
+
"step": 4050
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 7.51,
|
| 2932 |
+
"grad_norm": 1.5852398872375488,
|
| 2933 |
+
"learning_rate": 0.0012453703703703704,
|
| 2934 |
+
"loss": 0.4115,
|
| 2935 |
+
"step": 4060
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 7.53,
|
| 2939 |
+
"grad_norm": 1.3596739768981934,
|
| 2940 |
+
"learning_rate": 0.0012361111111111112,
|
| 2941 |
+
"loss": 0.3534,
|
| 2942 |
+
"step": 4070
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 7.55,
|
| 2946 |
+
"grad_norm": 1.9260822534561157,
|
| 2947 |
+
"learning_rate": 0.001226851851851852,
|
| 2948 |
+
"loss": 0.4297,
|
| 2949 |
+
"step": 4080
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 7.57,
|
| 2953 |
+
"grad_norm": 1.5783950090408325,
|
| 2954 |
+
"learning_rate": 0.0012175925925925926,
|
| 2955 |
+
"loss": 0.368,
|
| 2956 |
+
"step": 4090
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 7.59,
|
| 2960 |
+
"grad_norm": 1.7927485704421997,
|
| 2961 |
+
"learning_rate": 0.0012083333333333334,
|
| 2962 |
+
"loss": 0.3701,
|
| 2963 |
+
"step": 4100
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 7.6,
|
| 2967 |
+
"grad_norm": 1.1680549383163452,
|
| 2968 |
+
"learning_rate": 0.0011990740740740742,
|
| 2969 |
+
"loss": 0.3883,
|
| 2970 |
+
"step": 4110
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 7.62,
|
| 2974 |
+
"grad_norm": 1.4916154146194458,
|
| 2975 |
+
"learning_rate": 0.0011898148148148148,
|
| 2976 |
+
"loss": 0.373,
|
| 2977 |
+
"step": 4120
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 7.64,
|
| 2981 |
+
"grad_norm": 1.3623180389404297,
|
| 2982 |
+
"learning_rate": 0.0011805555555555556,
|
| 2983 |
+
"loss": 0.3697,
|
| 2984 |
+
"step": 4130
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 7.66,
|
| 2988 |
+
"grad_norm": 1.2618119716644287,
|
| 2989 |
+
"learning_rate": 0.0011712962962962964,
|
| 2990 |
+
"loss": 0.4114,
|
| 2991 |
+
"step": 4140
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 7.68,
|
| 2995 |
+
"grad_norm": 1.7506533861160278,
|
| 2996 |
+
"learning_rate": 0.001162037037037037,
|
| 2997 |
+
"loss": 0.4003,
|
| 2998 |
+
"step": 4150
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 7.7,
|
| 3002 |
+
"grad_norm": 1.0284491777420044,
|
| 3003 |
+
"learning_rate": 0.001152777777777778,
|
| 3004 |
+
"loss": 0.3749,
|
| 3005 |
+
"step": 4160
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 7.72,
|
| 3009 |
+
"grad_norm": 1.344994306564331,
|
| 3010 |
+
"learning_rate": 0.0011435185185185185,
|
| 3011 |
+
"loss": 0.4025,
|
| 3012 |
+
"step": 4170
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 7.73,
|
| 3016 |
+
"grad_norm": 3.066920042037964,
|
| 3017 |
+
"learning_rate": 0.0011342592592592593,
|
| 3018 |
+
"loss": 0.4203,
|
| 3019 |
+
"step": 4180
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 7.75,
|
| 3023 |
+
"grad_norm": 1.4180749654769897,
|
| 3024 |
+
"learning_rate": 0.0011250000000000001,
|
| 3025 |
+
"loss": 0.4364,
|
| 3026 |
+
"step": 4190
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 7.77,
|
| 3030 |
+
"grad_norm": 1.5211139917373657,
|
| 3031 |
+
"learning_rate": 0.0011157407407407407,
|
| 3032 |
+
"loss": 0.4133,
|
| 3033 |
+
"step": 4200
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 7.79,
|
| 3037 |
+
"grad_norm": 1.9516364336013794,
|
| 3038 |
+
"learning_rate": 0.0011064814814814815,
|
| 3039 |
+
"loss": 0.4232,
|
| 3040 |
+
"step": 4210
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 7.81,
|
| 3044 |
+
"grad_norm": 2.458101511001587,
|
| 3045 |
+
"learning_rate": 0.0010972222222222223,
|
| 3046 |
+
"loss": 0.434,
|
| 3047 |
+
"step": 4220
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 7.83,
|
| 3051 |
+
"grad_norm": 1.2099213600158691,
|
| 3052 |
+
"learning_rate": 0.0010879629629629629,
|
| 3053 |
+
"loss": 0.3169,
|
| 3054 |
+
"step": 4230
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 7.84,
|
| 3058 |
+
"grad_norm": 1.0573714971542358,
|
| 3059 |
+
"learning_rate": 0.0010787037037037037,
|
| 3060 |
+
"loss": 0.3777,
|
| 3061 |
+
"step": 4240
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 7.86,
|
| 3065 |
+
"grad_norm": 1.4449975490570068,
|
| 3066 |
+
"learning_rate": 0.0010694444444444445,
|
| 3067 |
+
"loss": 0.4426,
|
| 3068 |
+
"step": 4250
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 7.88,
|
| 3072 |
+
"grad_norm": 1.335119366645813,
|
| 3073 |
+
"learning_rate": 0.001060185185185185,
|
| 3074 |
+
"loss": 0.4643,
|
| 3075 |
+
"step": 4260
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 7.9,
|
| 3079 |
+
"grad_norm": 2.1732048988342285,
|
| 3080 |
+
"learning_rate": 0.001050925925925926,
|
| 3081 |
+
"loss": 0.3903,
|
| 3082 |
+
"step": 4270
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 7.92,
|
| 3086 |
+
"grad_norm": 1.193305492401123,
|
| 3087 |
+
"learning_rate": 0.0010416666666666667,
|
| 3088 |
+
"loss": 0.3744,
|
| 3089 |
+
"step": 4280
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 7.94,
|
| 3093 |
+
"grad_norm": 0.9779842495918274,
|
| 3094 |
+
"learning_rate": 0.0010324074074074074,
|
| 3095 |
+
"loss": 0.4124,
|
| 3096 |
+
"step": 4290
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 7.96,
|
| 3100 |
+
"grad_norm": 2.7513339519500732,
|
| 3101 |
+
"learning_rate": 0.0010231481481481482,
|
| 3102 |
+
"loss": 0.3917,
|
| 3103 |
+
"step": 4300
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 7.97,
|
| 3107 |
+
"grad_norm": 1.756395936012268,
|
| 3108 |
+
"learning_rate": 0.0010138888888888888,
|
| 3109 |
+
"loss": 0.4301,
|
| 3110 |
+
"step": 4310
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 7.99,
|
| 3114 |
+
"grad_norm": 1.5804221630096436,
|
| 3115 |
+
"learning_rate": 0.0010046296296296296,
|
| 3116 |
+
"loss": 0.4046,
|
| 3117 |
+
"step": 4320
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 8.0,
|
| 3121 |
+
"eval_accuracy": 0.979818209828994,
|
| 3122 |
+
"eval_f1": 0.9833721625743178,
|
| 3123 |
+
"eval_loss": 0.04691172018647194,
|
| 3124 |
+
"eval_precision": 0.9853003976747072,
|
| 3125 |
+
"eval_recall": 0.9820557579494299,
|
| 3126 |
+
"eval_runtime": 54.8438,
|
| 3127 |
+
"eval_samples_per_second": 118.354,
|
| 3128 |
+
"eval_steps_per_second": 7.403,
|
| 3129 |
+
"step": 4324
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 8.01,
|
| 3133 |
+
"grad_norm": 1.8008298873901367,
|
| 3134 |
+
"learning_rate": 0.0009953703703703704,
|
| 3135 |
+
"loss": 0.4452,
|
| 3136 |
+
"step": 4330
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 8.03,
|
| 3140 |
+
"grad_norm": 3.602733850479126,
|
| 3141 |
+
"learning_rate": 0.000986111111111111,
|
| 3142 |
+
"loss": 0.3281,
|
| 3143 |
+
"step": 4340
|
| 3144 |
+
},
|
| 3145 |
+
{
|
| 3146 |
+
"epoch": 8.05,
|
| 3147 |
+
"grad_norm": 1.5571168661117554,
|
| 3148 |
+
"learning_rate": 0.0009768518518518518,
|
| 3149 |
+
"loss": 0.432,
|
| 3150 |
+
"step": 4350
|
| 3151 |
+
},
|
| 3152 |
+
{
|
| 3153 |
+
"epoch": 8.07,
|
| 3154 |
+
"grad_norm": 1.5973519086837769,
|
| 3155 |
+
"learning_rate": 0.0009675925925925926,
|
| 3156 |
+
"loss": 0.3682,
|
| 3157 |
+
"step": 4360
|
| 3158 |
+
},
|
| 3159 |
+
{
|
| 3160 |
+
"epoch": 8.09,
|
| 3161 |
+
"grad_norm": 1.1785231828689575,
|
| 3162 |
+
"learning_rate": 0.0009583333333333334,
|
| 3163 |
+
"loss": 0.3233,
|
| 3164 |
+
"step": 4370
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 8.1,
|
| 3168 |
+
"grad_norm": 1.9037755727767944,
|
| 3169 |
+
"learning_rate": 0.0009490740740740742,
|
| 3170 |
+
"loss": 0.3733,
|
| 3171 |
+
"step": 4380
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 8.12,
|
| 3175 |
+
"grad_norm": 1.3710215091705322,
|
| 3176 |
+
"learning_rate": 0.0009398148148148149,
|
| 3177 |
+
"loss": 0.3562,
|
| 3178 |
+
"step": 4390
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 8.14,
|
| 3182 |
+
"grad_norm": 1.4512821435928345,
|
| 3183 |
+
"learning_rate": 0.0009305555555555556,
|
| 3184 |
+
"loss": 0.3391,
|
| 3185 |
+
"step": 4400
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 8.16,
|
| 3189 |
+
"grad_norm": 1.5068678855895996,
|
| 3190 |
+
"learning_rate": 0.0009212962962962964,
|
| 3191 |
+
"loss": 0.3703,
|
| 3192 |
+
"step": 4410
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 8.18,
|
| 3196 |
+
"grad_norm": 1.2810806035995483,
|
| 3197 |
+
"learning_rate": 0.000912037037037037,
|
| 3198 |
+
"loss": 0.3729,
|
| 3199 |
+
"step": 4420
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 8.2,
|
| 3203 |
+
"grad_norm": 1.9714285135269165,
|
| 3204 |
+
"learning_rate": 0.0009027777777777777,
|
| 3205 |
+
"loss": 0.3988,
|
| 3206 |
+
"step": 4430
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 8.21,
|
| 3210 |
+
"grad_norm": 1.3327577114105225,
|
| 3211 |
+
"learning_rate": 0.0008935185185185185,
|
| 3212 |
+
"loss": 0.4024,
|
| 3213 |
+
"step": 4440
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 8.23,
|
| 3217 |
+
"grad_norm": 1.0256530046463013,
|
| 3218 |
+
"learning_rate": 0.0008842592592592592,
|
| 3219 |
+
"loss": 0.3818,
|
| 3220 |
+
"step": 4450
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 8.25,
|
| 3224 |
+
"grad_norm": 2.233092784881592,
|
| 3225 |
+
"learning_rate": 0.0008749999999999999,
|
| 3226 |
+
"loss": 0.3668,
|
| 3227 |
+
"step": 4460
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 8.27,
|
| 3231 |
+
"grad_norm": 1.3951568603515625,
|
| 3232 |
+
"learning_rate": 0.0008657407407407407,
|
| 3233 |
+
"loss": 0.44,
|
| 3234 |
+
"step": 4470
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 8.29,
|
| 3238 |
+
"grad_norm": 1.827919840812683,
|
| 3239 |
+
"learning_rate": 0.0008564814814814815,
|
| 3240 |
+
"loss": 0.3801,
|
| 3241 |
+
"step": 4480
|
| 3242 |
+
},
|
| 3243 |
+
{
|
| 3244 |
+
"epoch": 8.31,
|
| 3245 |
+
"grad_norm": 2.934967041015625,
|
| 3246 |
+
"learning_rate": 0.0008472222222222223,
|
| 3247 |
+
"loss": 0.422,
|
| 3248 |
+
"step": 4490
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 8.33,
|
| 3252 |
+
"grad_norm": 1.2205818891525269,
|
| 3253 |
+
"learning_rate": 0.000837962962962963,
|
| 3254 |
+
"loss": 0.3991,
|
| 3255 |
+
"step": 4500
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 8.34,
|
| 3259 |
+
"grad_norm": 1.424593448638916,
|
| 3260 |
+
"learning_rate": 0.0008287037037037038,
|
| 3261 |
+
"loss": 0.418,
|
| 3262 |
+
"step": 4510
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 8.36,
|
| 3266 |
+
"grad_norm": 1.0254913568496704,
|
| 3267 |
+
"learning_rate": 0.0008194444444444445,
|
| 3268 |
+
"loss": 0.3876,
|
| 3269 |
+
"step": 4520
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 8.38,
|
| 3273 |
+
"grad_norm": 0.9939424991607666,
|
| 3274 |
+
"learning_rate": 0.0008101851851851852,
|
| 3275 |
+
"loss": 0.4267,
|
| 3276 |
+
"step": 4530
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 8.4,
|
| 3280 |
+
"grad_norm": 1.7678611278533936,
|
| 3281 |
+
"learning_rate": 0.000800925925925926,
|
| 3282 |
+
"loss": 0.3916,
|
| 3283 |
+
"step": 4540
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 8.42,
|
| 3287 |
+
"grad_norm": 1.0343542098999023,
|
| 3288 |
+
"learning_rate": 0.0007916666666666666,
|
| 3289 |
+
"loss": 0.3939,
|
| 3290 |
+
"step": 4550
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 8.44,
|
| 3294 |
+
"grad_norm": 0.9549526572227478,
|
| 3295 |
+
"learning_rate": 0.0007824074074074073,
|
| 3296 |
+
"loss": 0.3785,
|
| 3297 |
+
"step": 4560
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 8.46,
|
| 3301 |
+
"grad_norm": 1.0010316371917725,
|
| 3302 |
+
"learning_rate": 0.0007731481481481481,
|
| 3303 |
+
"loss": 0.3508,
|
| 3304 |
+
"step": 4570
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 8.47,
|
| 3308 |
+
"grad_norm": 2.0805888175964355,
|
| 3309 |
+
"learning_rate": 0.0007638888888888889,
|
| 3310 |
+
"loss": 0.3614,
|
| 3311 |
+
"step": 4580
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 8.49,
|
| 3315 |
+
"grad_norm": 1.181626558303833,
|
| 3316 |
+
"learning_rate": 0.0007546296296296297,
|
| 3317 |
+
"loss": 0.359,
|
| 3318 |
+
"step": 4590
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 8.51,
|
| 3322 |
+
"grad_norm": 1.138376235961914,
|
| 3323 |
+
"learning_rate": 0.0007453703703703704,
|
| 3324 |
+
"loss": 0.3513,
|
| 3325 |
+
"step": 4600
|
| 3326 |
+
},
|
| 3327 |
+
{
|
| 3328 |
+
"epoch": 8.53,
|
| 3329 |
+
"grad_norm": 1.0916022062301636,
|
| 3330 |
+
"learning_rate": 0.0007361111111111111,
|
| 3331 |
+
"loss": 0.3199,
|
| 3332 |
+
"step": 4610
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 8.55,
|
| 3336 |
+
"grad_norm": 2.3587114810943604,
|
| 3337 |
+
"learning_rate": 0.0007268518518518519,
|
| 3338 |
+
"loss": 0.3658,
|
| 3339 |
+
"step": 4620
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 8.57,
|
| 3343 |
+
"grad_norm": 1.2504198551177979,
|
| 3344 |
+
"learning_rate": 0.0007175925925925926,
|
| 3345 |
+
"loss": 0.3952,
|
| 3346 |
+
"step": 4630
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 8.58,
|
| 3350 |
+
"grad_norm": 1.3484505414962769,
|
| 3351 |
+
"learning_rate": 0.0007083333333333333,
|
| 3352 |
+
"loss": 0.3542,
|
| 3353 |
+
"step": 4640
|
| 3354 |
+
},
|
| 3355 |
+
{
|
| 3356 |
+
"epoch": 8.6,
|
| 3357 |
+
"grad_norm": 2.4245710372924805,
|
| 3358 |
+
"learning_rate": 0.0006990740740740741,
|
| 3359 |
+
"loss": 0.3421,
|
| 3360 |
+
"step": 4650
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 8.62,
|
| 3364 |
+
"grad_norm": 1.2606480121612549,
|
| 3365 |
+
"learning_rate": 0.0006898148148148148,
|
| 3366 |
+
"loss": 0.3404,
|
| 3367 |
+
"step": 4660
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 8.64,
|
| 3371 |
+
"grad_norm": 1.913200855255127,
|
| 3372 |
+
"learning_rate": 0.0006805555555555556,
|
| 3373 |
+
"loss": 0.4015,
|
| 3374 |
+
"step": 4670
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 8.66,
|
| 3378 |
+
"grad_norm": 1.5260008573532104,
|
| 3379 |
+
"learning_rate": 0.0006712962962962964,
|
| 3380 |
+
"loss": 0.3425,
|
| 3381 |
+
"step": 4680
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 8.68,
|
| 3385 |
+
"grad_norm": 1.221198320388794,
|
| 3386 |
+
"learning_rate": 0.0006620370370370372,
|
| 3387 |
+
"loss": 0.4078,
|
| 3388 |
+
"step": 4690
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 8.7,
|
| 3392 |
+
"grad_norm": 1.630136489868164,
|
| 3393 |
+
"learning_rate": 0.0006527777777777778,
|
| 3394 |
+
"loss": 0.3508,
|
| 3395 |
+
"step": 4700
|
| 3396 |
+
},
|
| 3397 |
+
{
|
| 3398 |
+
"epoch": 8.71,
|
| 3399 |
+
"grad_norm": 1.3053523302078247,
|
| 3400 |
+
"learning_rate": 0.0006435185185185185,
|
| 3401 |
+
"loss": 0.2958,
|
| 3402 |
+
"step": 4710
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 8.73,
|
| 3406 |
+
"grad_norm": 1.5121403932571411,
|
| 3407 |
+
"learning_rate": 0.0006342592592592593,
|
| 3408 |
+
"loss": 0.3552,
|
| 3409 |
+
"step": 4720
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 8.75,
|
| 3413 |
+
"grad_norm": 1.1451386213302612,
|
| 3414 |
+
"learning_rate": 0.000625,
|
| 3415 |
+
"loss": 0.3688,
|
| 3416 |
+
"step": 4730
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 8.77,
|
| 3420 |
+
"grad_norm": 2.4750773906707764,
|
| 3421 |
+
"learning_rate": 0.0006157407407407407,
|
| 3422 |
+
"loss": 0.3435,
|
| 3423 |
+
"step": 4740
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 8.79,
|
| 3427 |
+
"grad_norm": 0.6995977163314819,
|
| 3428 |
+
"learning_rate": 0.0006064814814814815,
|
| 3429 |
+
"loss": 0.3152,
|
| 3430 |
+
"step": 4750
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 8.81,
|
| 3434 |
+
"grad_norm": 1.6774075031280518,
|
| 3435 |
+
"learning_rate": 0.0005972222222222223,
|
| 3436 |
+
"loss": 0.2898,
|
| 3437 |
+
"step": 4760
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 8.83,
|
| 3441 |
+
"grad_norm": 1.1041797399520874,
|
| 3442 |
+
"learning_rate": 0.000587962962962963,
|
| 3443 |
+
"loss": 0.3086,
|
| 3444 |
+
"step": 4770
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 8.84,
|
| 3448 |
+
"grad_norm": 1.2594033479690552,
|
| 3449 |
+
"learning_rate": 0.0005787037037037037,
|
| 3450 |
+
"loss": 0.3981,
|
| 3451 |
+
"step": 4780
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 8.86,
|
| 3455 |
+
"grad_norm": 0.9759043455123901,
|
| 3456 |
+
"learning_rate": 0.0005694444444444445,
|
| 3457 |
+
"loss": 0.356,
|
| 3458 |
+
"step": 4790
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 8.88,
|
| 3462 |
+
"grad_norm": 1.4041826725006104,
|
| 3463 |
+
"learning_rate": 0.0005601851851851852,
|
| 3464 |
+
"loss": 0.3514,
|
| 3465 |
+
"step": 4800
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 8.9,
|
| 3469 |
+
"grad_norm": 0.9928153157234192,
|
| 3470 |
+
"learning_rate": 0.000550925925925926,
|
| 3471 |
+
"loss": 0.388,
|
| 3472 |
+
"step": 4810
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 8.92,
|
| 3476 |
+
"grad_norm": 1.0990333557128906,
|
| 3477 |
+
"learning_rate": 0.0005416666666666668,
|
| 3478 |
+
"loss": 0.3013,
|
| 3479 |
+
"step": 4820
|
| 3480 |
+
},
|
| 3481 |
+
{
|
| 3482 |
+
"epoch": 8.94,
|
| 3483 |
+
"grad_norm": 1.2321481704711914,
|
| 3484 |
+
"learning_rate": 0.0005324074074074074,
|
| 3485 |
+
"loss": 0.4016,
|
| 3486 |
+
"step": 4830
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 8.95,
|
| 3490 |
+
"grad_norm": 1.1745514869689941,
|
| 3491 |
+
"learning_rate": 0.0005231481481481481,
|
| 3492 |
+
"loss": 0.3601,
|
| 3493 |
+
"step": 4840
|
| 3494 |
+
},
|
| 3495 |
+
{
|
| 3496 |
+
"epoch": 8.97,
|
| 3497 |
+
"grad_norm": 1.633880615234375,
|
| 3498 |
+
"learning_rate": 0.0005138888888888889,
|
| 3499 |
+
"loss": 0.3291,
|
| 3500 |
+
"step": 4850
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 8.99,
|
| 3504 |
+
"grad_norm": 1.1216607093811035,
|
| 3505 |
+
"learning_rate": 0.0005046296296296297,
|
| 3506 |
+
"loss": 0.3314,
|
| 3507 |
+
"step": 4860
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 9.0,
|
| 3511 |
+
"eval_accuracy": 0.986134647974118,
|
| 3512 |
+
"eval_f1": 0.987440563210657,
|
| 3513 |
+
"eval_loss": 0.038755957037210464,
|
| 3514 |
+
"eval_precision": 0.9891828577753293,
|
| 3515 |
+
"eval_recall": 0.98596866524898,
|
| 3516 |
+
"eval_runtime": 55.0297,
|
| 3517 |
+
"eval_samples_per_second": 117.955,
|
| 3518 |
+
"eval_steps_per_second": 7.378,
|
| 3519 |
+
"step": 4864
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 9.01,
|
| 3523 |
+
"grad_norm": 1.2497656345367432,
|
| 3524 |
+
"learning_rate": 0.0004953703703703704,
|
| 3525 |
+
"loss": 0.3183,
|
| 3526 |
+
"step": 4870
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 9.03,
|
| 3530 |
+
"grad_norm": 1.5892603397369385,
|
| 3531 |
+
"learning_rate": 0.00048611111111111115,
|
| 3532 |
+
"loss": 0.3222,
|
| 3533 |
+
"step": 4880
|
| 3534 |
+
},
|
| 3535 |
+
{
|
| 3536 |
+
"epoch": 9.05,
|
| 3537 |
+
"grad_norm": 1.4494744539260864,
|
| 3538 |
+
"learning_rate": 0.00047685185185185184,
|
| 3539 |
+
"loss": 0.3573,
|
| 3540 |
+
"step": 4890
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 9.07,
|
| 3544 |
+
"grad_norm": 1.7034820318222046,
|
| 3545 |
+
"learning_rate": 0.0004675925925925926,
|
| 3546 |
+
"loss": 0.3194,
|
| 3547 |
+
"step": 4900
|
| 3548 |
+
},
|
| 3549 |
+
{
|
| 3550 |
+
"epoch": 9.08,
|
| 3551 |
+
"grad_norm": 0.8668962121009827,
|
| 3552 |
+
"learning_rate": 0.0004583333333333333,
|
| 3553 |
+
"loss": 0.3483,
|
| 3554 |
+
"step": 4910
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 9.1,
|
| 3558 |
+
"grad_norm": 1.081874132156372,
|
| 3559 |
+
"learning_rate": 0.0004490740740740741,
|
| 3560 |
+
"loss": 0.3626,
|
| 3561 |
+
"step": 4920
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 9.12,
|
| 3565 |
+
"grad_norm": 1.78743314743042,
|
| 3566 |
+
"learning_rate": 0.0004398148148148148,
|
| 3567 |
+
"loss": 0.3651,
|
| 3568 |
+
"step": 4930
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 9.14,
|
| 3572 |
+
"grad_norm": 1.1324867010116577,
|
| 3573 |
+
"learning_rate": 0.00043055555555555555,
|
| 3574 |
+
"loss": 0.2714,
|
| 3575 |
+
"step": 4940
|
| 3576 |
+
},
|
| 3577 |
+
{
|
| 3578 |
+
"epoch": 9.16,
|
| 3579 |
+
"grad_norm": 1.9258543252944946,
|
| 3580 |
+
"learning_rate": 0.0004212962962962963,
|
| 3581 |
+
"loss": 0.3635,
|
| 3582 |
+
"step": 4950
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 9.18,
|
| 3586 |
+
"grad_norm": 1.2044012546539307,
|
| 3587 |
+
"learning_rate": 0.00041203703703703704,
|
| 3588 |
+
"loss": 0.3126,
|
| 3589 |
+
"step": 4960
|
| 3590 |
+
},
|
| 3591 |
+
{
|
| 3592 |
+
"epoch": 9.2,
|
| 3593 |
+
"grad_norm": 1.3639435768127441,
|
| 3594 |
+
"learning_rate": 0.00040277777777777783,
|
| 3595 |
+
"loss": 0.3203,
|
| 3596 |
+
"step": 4970
|
| 3597 |
+
},
|
| 3598 |
+
{
|
| 3599 |
+
"epoch": 9.21,
|
| 3600 |
+
"grad_norm": 1.7811472415924072,
|
| 3601 |
+
"learning_rate": 0.0003935185185185185,
|
| 3602 |
+
"loss": 0.3208,
|
| 3603 |
+
"step": 4980
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 9.23,
|
| 3607 |
+
"grad_norm": 1.4253028631210327,
|
| 3608 |
+
"learning_rate": 0.00038425925925925927,
|
| 3609 |
+
"loss": 0.3326,
|
| 3610 |
+
"step": 4990
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 9.25,
|
| 3614 |
+
"grad_norm": 1.4700595140457153,
|
| 3615 |
+
"learning_rate": 0.000375,
|
| 3616 |
+
"loss": 0.3015,
|
| 3617 |
+
"step": 5000
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 9.27,
|
| 3621 |
+
"grad_norm": 2.1158525943756104,
|
| 3622 |
+
"learning_rate": 0.0003657407407407407,
|
| 3623 |
+
"loss": 0.3472,
|
| 3624 |
+
"step": 5010
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 9.29,
|
| 3628 |
+
"grad_norm": 1.325528621673584,
|
| 3629 |
+
"learning_rate": 0.0003564814814814815,
|
| 3630 |
+
"loss": 0.3365,
|
| 3631 |
+
"step": 5020
|
| 3632 |
+
},
|
| 3633 |
+
{
|
| 3634 |
+
"epoch": 9.31,
|
| 3635 |
+
"grad_norm": 1.173003077507019,
|
| 3636 |
+
"learning_rate": 0.00034722222222222224,
|
| 3637 |
+
"loss": 0.3213,
|
| 3638 |
+
"step": 5030
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 9.32,
|
| 3642 |
+
"grad_norm": 2.355008840560913,
|
| 3643 |
+
"learning_rate": 0.000337962962962963,
|
| 3644 |
+
"loss": 0.3386,
|
| 3645 |
+
"step": 5040
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 9.34,
|
| 3649 |
+
"grad_norm": 1.198581576347351,
|
| 3650 |
+
"learning_rate": 0.0003287037037037037,
|
| 3651 |
+
"loss": 0.2985,
|
| 3652 |
+
"step": 5050
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 9.36,
|
| 3656 |
+
"grad_norm": 1.7199565172195435,
|
| 3657 |
+
"learning_rate": 0.0003194444444444444,
|
| 3658 |
+
"loss": 0.3666,
|
| 3659 |
+
"step": 5060
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 9.38,
|
| 3663 |
+
"grad_norm": 1.359098196029663,
|
| 3664 |
+
"learning_rate": 0.0003101851851851852,
|
| 3665 |
+
"loss": 0.327,
|
| 3666 |
+
"step": 5070
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 9.4,
|
| 3670 |
+
"grad_norm": 1.2990363836288452,
|
| 3671 |
+
"learning_rate": 0.0003009259259259259,
|
| 3672 |
+
"loss": 0.3673,
|
| 3673 |
+
"step": 5080
|
| 3674 |
+
},
|
| 3675 |
+
{
|
| 3676 |
+
"epoch": 9.42,
|
| 3677 |
+
"grad_norm": 1.5110586881637573,
|
| 3678 |
+
"learning_rate": 0.0002916666666666667,
|
| 3679 |
+
"loss": 0.3124,
|
| 3680 |
+
"step": 5090
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 9.44,
|
| 3684 |
+
"grad_norm": 1.6193897724151611,
|
| 3685 |
+
"learning_rate": 0.0002824074074074074,
|
| 3686 |
+
"loss": 0.3056,
|
| 3687 |
+
"step": 5100
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 9.45,
|
| 3691 |
+
"grad_norm": 1.2994916439056396,
|
| 3692 |
+
"learning_rate": 0.0002731481481481482,
|
| 3693 |
+
"loss": 0.3041,
|
| 3694 |
+
"step": 5110
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 9.47,
|
| 3698 |
+
"grad_norm": 1.1582506895065308,
|
| 3699 |
+
"learning_rate": 0.0002638888888888889,
|
| 3700 |
+
"loss": 0.325,
|
| 3701 |
+
"step": 5120
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 9.49,
|
| 3705 |
+
"grad_norm": 1.2252171039581299,
|
| 3706 |
+
"learning_rate": 0.0002546296296296296,
|
| 3707 |
+
"loss": 0.3541,
|
| 3708 |
+
"step": 5130
|
| 3709 |
+
},
|
| 3710 |
+
{
|
| 3711 |
+
"epoch": 9.51,
|
| 3712 |
+
"grad_norm": 1.0199075937271118,
|
| 3713 |
+
"learning_rate": 0.0002453703703703704,
|
| 3714 |
+
"loss": 0.3763,
|
| 3715 |
+
"step": 5140
|
| 3716 |
+
},
|
| 3717 |
+
{
|
| 3718 |
+
"epoch": 9.53,
|
| 3719 |
+
"grad_norm": 1.113032341003418,
|
| 3720 |
+
"learning_rate": 0.00023611111111111112,
|
| 3721 |
+
"loss": 0.3459,
|
| 3722 |
+
"step": 5150
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 9.55,
|
| 3726 |
+
"grad_norm": 1.1221752166748047,
|
| 3727 |
+
"learning_rate": 0.00022685185185185186,
|
| 3728 |
+
"loss": 0.284,
|
| 3729 |
+
"step": 5160
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 9.57,
|
| 3733 |
+
"grad_norm": 1.0831656455993652,
|
| 3734 |
+
"learning_rate": 0.0002175925925925926,
|
| 3735 |
+
"loss": 0.3117,
|
| 3736 |
+
"step": 5170
|
| 3737 |
+
},
|
| 3738 |
+
{
|
| 3739 |
+
"epoch": 9.58,
|
| 3740 |
+
"grad_norm": 0.723981499671936,
|
| 3741 |
+
"learning_rate": 0.00020833333333333332,
|
| 3742 |
+
"loss": 0.3231,
|
| 3743 |
+
"step": 5180
|
| 3744 |
+
},
|
| 3745 |
+
{
|
| 3746 |
+
"epoch": 9.6,
|
| 3747 |
+
"grad_norm": 1.0349414348602295,
|
| 3748 |
+
"learning_rate": 0.0001990740740740741,
|
| 3749 |
+
"loss": 0.3344,
|
| 3750 |
+
"step": 5190
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 9.62,
|
| 3754 |
+
"grad_norm": 0.9438874125480652,
|
| 3755 |
+
"learning_rate": 0.0001898148148148148,
|
| 3756 |
+
"loss": 0.3272,
|
| 3757 |
+
"step": 5200
|
| 3758 |
+
},
|
| 3759 |
+
{
|
| 3760 |
+
"epoch": 9.64,
|
| 3761 |
+
"grad_norm": 1.0570435523986816,
|
| 3762 |
+
"learning_rate": 0.00018055555555555555,
|
| 3763 |
+
"loss": 0.3309,
|
| 3764 |
+
"step": 5210
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 9.66,
|
| 3768 |
+
"grad_norm": 0.9774361848831177,
|
| 3769 |
+
"learning_rate": 0.00017129629629629632,
|
| 3770 |
+
"loss": 0.3129,
|
| 3771 |
+
"step": 5220
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 9.68,
|
| 3775 |
+
"grad_norm": 1.1588292121887207,
|
| 3776 |
+
"learning_rate": 0.00016203703703703703,
|
| 3777 |
+
"loss": 0.3287,
|
| 3778 |
+
"step": 5230
|
| 3779 |
+
},
|
| 3780 |
+
{
|
| 3781 |
+
"epoch": 9.69,
|
| 3782 |
+
"grad_norm": 1.1384470462799072,
|
| 3783 |
+
"learning_rate": 0.00015277777777777777,
|
| 3784 |
+
"loss": 0.2838,
|
| 3785 |
+
"step": 5240
|
| 3786 |
+
},
|
| 3787 |
+
{
|
| 3788 |
+
"epoch": 9.71,
|
| 3789 |
+
"grad_norm": 1.3432296514511108,
|
| 3790 |
+
"learning_rate": 0.00014351851851851852,
|
| 3791 |
+
"loss": 0.3424,
|
| 3792 |
+
"step": 5250
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 9.73,
|
| 3796 |
+
"grad_norm": 0.6609634160995483,
|
| 3797 |
+
"learning_rate": 0.00013425925925925926,
|
| 3798 |
+
"loss": 0.2843,
|
| 3799 |
+
"step": 5260
|
| 3800 |
+
},
|
| 3801 |
+
{
|
| 3802 |
+
"epoch": 9.75,
|
| 3803 |
+
"grad_norm": 1.2155038118362427,
|
| 3804 |
+
"learning_rate": 0.000125,
|
| 3805 |
+
"loss": 0.3159,
|
| 3806 |
+
"step": 5270
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 9.77,
|
| 3810 |
+
"grad_norm": 1.5623424053192139,
|
| 3811 |
+
"learning_rate": 0.00011574074074074073,
|
| 3812 |
+
"loss": 0.2891,
|
| 3813 |
+
"step": 5280
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 9.79,
|
| 3817 |
+
"grad_norm": 1.9647743701934814,
|
| 3818 |
+
"learning_rate": 0.00010648148148148149,
|
| 3819 |
+
"loss": 0.3394,
|
| 3820 |
+
"step": 5290
|
| 3821 |
+
},
|
| 3822 |
+
{
|
| 3823 |
+
"epoch": 9.81,
|
| 3824 |
+
"grad_norm": 1.1939747333526611,
|
| 3825 |
+
"learning_rate": 9.722222222222223e-05,
|
| 3826 |
+
"loss": 0.3545,
|
| 3827 |
+
"step": 5300
|
| 3828 |
+
},
|
| 3829 |
+
{
|
| 3830 |
+
"epoch": 9.82,
|
| 3831 |
+
"grad_norm": 1.5705734491348267,
|
| 3832 |
+
"learning_rate": 8.796296296296297e-05,
|
| 3833 |
+
"loss": 0.372,
|
| 3834 |
+
"step": 5310
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 9.84,
|
| 3838 |
+
"grad_norm": 1.9765841960906982,
|
| 3839 |
+
"learning_rate": 7.87037037037037e-05,
|
| 3840 |
+
"loss": 0.3555,
|
| 3841 |
+
"step": 5320
|
| 3842 |
+
},
|
| 3843 |
+
{
|
| 3844 |
+
"epoch": 9.86,
|
| 3845 |
+
"grad_norm": 1.168407917022705,
|
| 3846 |
+
"learning_rate": 6.944444444444444e-05,
|
| 3847 |
+
"loss": 0.2545,
|
| 3848 |
+
"step": 5330
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 9.88,
|
| 3852 |
+
"grad_norm": 0.8777741193771362,
|
| 3853 |
+
"learning_rate": 6.018518518518519e-05,
|
| 3854 |
+
"loss": 0.3408,
|
| 3855 |
+
"step": 5340
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 9.9,
|
| 3859 |
+
"grad_norm": 1.2033889293670654,
|
| 3860 |
+
"learning_rate": 5.092592592592593e-05,
|
| 3861 |
+
"loss": 0.3518,
|
| 3862 |
+
"step": 5350
|
| 3863 |
+
},
|
| 3864 |
+
{
|
| 3865 |
+
"epoch": 9.92,
|
| 3866 |
+
"grad_norm": 1.9429413080215454,
|
| 3867 |
+
"learning_rate": 4.1666666666666665e-05,
|
| 3868 |
+
"loss": 0.2779,
|
| 3869 |
+
"step": 5360
|
| 3870 |
+
},
|
| 3871 |
+
{
|
| 3872 |
+
"epoch": 9.94,
|
| 3873 |
+
"grad_norm": 1.5541284084320068,
|
| 3874 |
+
"learning_rate": 3.240740740740741e-05,
|
| 3875 |
+
"loss": 0.3169,
|
| 3876 |
+
"step": 5370
|
| 3877 |
+
},
|
| 3878 |
+
{
|
| 3879 |
+
"epoch": 9.95,
|
| 3880 |
+
"grad_norm": 0.8645866513252258,
|
| 3881 |
+
"learning_rate": 2.3148148148148147e-05,
|
| 3882 |
+
"loss": 0.2976,
|
| 3883 |
+
"step": 5380
|
| 3884 |
+
},
|
| 3885 |
+
{
|
| 3886 |
+
"epoch": 9.97,
|
| 3887 |
+
"grad_norm": 1.3550152778625488,
|
| 3888 |
+
"learning_rate": 1.388888888888889e-05,
|
| 3889 |
+
"loss": 0.2891,
|
| 3890 |
+
"step": 5390
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 9.99,
|
| 3894 |
+
"grad_norm": 1.666207194328308,
|
| 3895 |
+
"learning_rate": 4.6296296296296296e-06,
|
| 3896 |
+
"loss": 0.2865,
|
| 3897 |
+
"step": 5400
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 9.99,
|
| 3901 |
+
"eval_accuracy": 0.9830534586350331,
|
| 3902 |
+
"eval_f1": 0.9868535728073075,
|
| 3903 |
+
"eval_loss": 0.0450444296002388,
|
| 3904 |
+
"eval_precision": 0.9880250018061109,
|
| 3905 |
+
"eval_recall": 0.986225426158784,
|
| 3906 |
+
"eval_runtime": 55.3663,
|
| 3907 |
+
"eval_samples_per_second": 117.237,
|
| 3908 |
+
"eval_steps_per_second": 7.333,
|
| 3909 |
+
"step": 5400
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 9.99,
|
| 3913 |
+
"step": 5400,
|
| 3914 |
+
"total_flos": 2.695096772600047e+19,
|
| 3915 |
+
"train_loss": 0.5508582998205114,
|
| 3916 |
+
"train_runtime": 6179.724,
|
| 3917 |
+
"train_samples_per_second": 55.959,
|
| 3918 |
+
"train_steps_per_second": 0.874
|
| 3919 |
+
}
|
| 3920 |
+
],
|
| 3921 |
+
"logging_steps": 10,
|
| 3922 |
+
"max_steps": 5400,
|
| 3923 |
+
"num_input_tokens_seen": 0,
|
| 3924 |
+
"num_train_epochs": 10,
|
| 3925 |
+
"save_steps": 500,
|
| 3926 |
+
"total_flos": 2.695096772600047e+19,
|
| 3927 |
+
"train_batch_size": 16,
|
| 3928 |
+
"trial_name": null,
|
| 3929 |
+
"trial_params": null
|
| 3930 |
+
}
|