Commit
•
1dccf11
1
Parent(s):
c74d223
🍻 cheers
Browse files- README.md +10 -6
- all_results.json +16 -0
- config.json +1 -1
- eval_results.json +12 -0
- train_results.json +8 -0
- trainer_state.json +1699 -0
README.md
CHANGED
|
@@ -1,7 +1,11 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: facebook/dinov2-base
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- generated_from_trainer
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
|
@@ -15,13 +19,13 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# dino-base-2023_12_01-with_custom_small_head
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [facebook/dinov2-base](https://huggingface.co/facebook/dinov2-base) on the
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss: 0.
|
| 21 |
-
- F1 Micro: 0.
|
| 22 |
-
- F1 Macro: 0.
|
| 23 |
-
- Roc Auc: 0.
|
| 24 |
-
- Accuracy: 0.
|
| 25 |
- Learning Rate: 0.0000
|
| 26 |
|
| 27 |
## Model description
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- eng
|
| 4 |
license: apache-2.0
|
| 5 |
base_model: facebook/dinov2-base
|
| 6 |
tags:
|
| 7 |
+
- multilabel-image-classification
|
| 8 |
+
- multilabel
|
| 9 |
- generated_from_trainer
|
| 10 |
metrics:
|
| 11 |
- accuracy
|
|
|
|
| 19 |
|
| 20 |
# dino-base-2023_12_01-with_custom_small_head
|
| 21 |
|
| 22 |
+
This model is a fine-tuned version of [facebook/dinov2-base](https://huggingface.co/facebook/dinov2-base) on the multilabel_complete_dataset dataset.
|
| 23 |
It achieves the following results on the evaluation set:
|
| 24 |
+
- Loss: 0.1266
|
| 25 |
+
- F1 Micro: 0.8318
|
| 26 |
+
- F1 Macro: 0.8018
|
| 27 |
+
- Roc Auc: 0.8960
|
| 28 |
+
- Accuracy: 0.5224
|
| 29 |
- Learning Rate: 0.0000
|
| 30 |
|
| 31 |
## Model description
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 86.0,
|
| 3 |
+
"eval_accuracy": 0.5223512847588877,
|
| 4 |
+
"eval_f1_macro": 0.8017650848720057,
|
| 5 |
+
"eval_f1_micro": 0.8318343195266272,
|
| 6 |
+
"eval_loss": 0.1265595406293869,
|
| 7 |
+
"eval_roc_auc": 0.896027183058249,
|
| 8 |
+
"eval_runtime": 660.9127,
|
| 9 |
+
"eval_samples_per_second": 4.299,
|
| 10 |
+
"eval_steps_per_second": 0.269,
|
| 11 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 12 |
+
"train_loss": 0.1030069793400273,
|
| 13 |
+
"train_runtime": 233299.8908,
|
| 14 |
+
"train_samples_per_second": 3.305,
|
| 15 |
+
"train_steps_per_second": 0.207
|
| 16 |
+
}
|
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "facebook/dinov2-
|
| 3 |
"apply_layernorm": true,
|
| 4 |
"architectures": [
|
| 5 |
"NewheadDinov2ForImageClassification"
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "facebook/dinov2-base2023_12_01",
|
| 3 |
"apply_layernorm": true,
|
| 4 |
"architectures": [
|
| 5 |
"NewheadDinov2ForImageClassification"
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 86.0,
|
| 3 |
+
"eval_accuracy": 0.5223512847588877,
|
| 4 |
+
"eval_f1_macro": 0.8017650848720057,
|
| 5 |
+
"eval_f1_micro": 0.8318343195266272,
|
| 6 |
+
"eval_loss": 0.1265595406293869,
|
| 7 |
+
"eval_roc_auc": 0.896027183058249,
|
| 8 |
+
"eval_runtime": 660.9127,
|
| 9 |
+
"eval_samples_per_second": 4.299,
|
| 10 |
+
"eval_steps_per_second": 0.269,
|
| 11 |
+
"learning_rate": 1.0000000000000002e-06
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 86.0,
|
| 3 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 4 |
+
"train_loss": 0.1030069793400273,
|
| 5 |
+
"train_runtime": 233299.8908,
|
| 6 |
+
"train_samples_per_second": 3.305,
|
| 7 |
+
"train_steps_per_second": 0.207
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1699 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.128558948636055,
|
| 3 |
+
"best_model_checkpoint": "./dino-base-2023_12_01-with_custom_small_head/checkpoint-40736",
|
| 4 |
+
"epoch": 86.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 46096,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.93,
|
| 13 |
+
"learning_rate": 0.01,
|
| 14 |
+
"loss": 0.4702,
|
| 15 |
+
"step": 500
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 1.0,
|
| 19 |
+
"eval_accuracy": 0.42979635584137194,
|
| 20 |
+
"eval_f1_macro": 0.6933212404746647,
|
| 21 |
+
"eval_f1_micro": 0.7549217566885411,
|
| 22 |
+
"eval_loss": 0.45448988676071167,
|
| 23 |
+
"eval_roc_auc": 0.8355761494354749,
|
| 24 |
+
"eval_runtime": 677.2303,
|
| 25 |
+
"eval_samples_per_second": 4.133,
|
| 26 |
+
"eval_steps_per_second": 0.258,
|
| 27 |
+
"learning_rate": 0.01,
|
| 28 |
+
"step": 536
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 1.87,
|
| 32 |
+
"learning_rate": 0.01,
|
| 33 |
+
"loss": 0.4091,
|
| 34 |
+
"step": 1000
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 2.0,
|
| 38 |
+
"eval_accuracy": 0.4376563058235084,
|
| 39 |
+
"eval_f1_macro": 0.731458561583255,
|
| 40 |
+
"eval_f1_micro": 0.7831797235023041,
|
| 41 |
+
"eval_loss": 0.36905375123023987,
|
| 42 |
+
"eval_roc_auc": 0.8761564570737086,
|
| 43 |
+
"eval_runtime": 660.2957,
|
| 44 |
+
"eval_samples_per_second": 4.239,
|
| 45 |
+
"eval_steps_per_second": 0.265,
|
| 46 |
+
"learning_rate": 0.01,
|
| 47 |
+
"step": 1072
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 2.8,
|
| 51 |
+
"learning_rate": 0.01,
|
| 52 |
+
"loss": 0.3999,
|
| 53 |
+
"step": 1500
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 3.0,
|
| 57 |
+
"eval_accuracy": 0.41407645587709896,
|
| 58 |
+
"eval_f1_macro": 0.7213523804674298,
|
| 59 |
+
"eval_f1_micro": 0.7724575524352272,
|
| 60 |
+
"eval_loss": 0.4630758464336395,
|
| 61 |
+
"eval_roc_auc": 0.8642590860319617,
|
| 62 |
+
"eval_runtime": 660.6001,
|
| 63 |
+
"eval_samples_per_second": 4.237,
|
| 64 |
+
"eval_steps_per_second": 0.265,
|
| 65 |
+
"learning_rate": 0.01,
|
| 66 |
+
"step": 1608
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 3.73,
|
| 70 |
+
"learning_rate": 0.01,
|
| 71 |
+
"loss": 0.3958,
|
| 72 |
+
"step": 2000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 4.0,
|
| 76 |
+
"eval_accuracy": 0.4533762057877814,
|
| 77 |
+
"eval_f1_macro": 0.7310289607537521,
|
| 78 |
+
"eval_f1_micro": 0.7858175718665047,
|
| 79 |
+
"eval_loss": 0.4028192162513733,
|
| 80 |
+
"eval_roc_auc": 0.8765227946021852,
|
| 81 |
+
"eval_runtime": 677.2178,
|
| 82 |
+
"eval_samples_per_second": 4.133,
|
| 83 |
+
"eval_steps_per_second": 0.258,
|
| 84 |
+
"learning_rate": 0.01,
|
| 85 |
+
"step": 2144
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 4.66,
|
| 89 |
+
"learning_rate": 0.01,
|
| 90 |
+
"loss": 0.3795,
|
| 91 |
+
"step": 2500
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 5.0,
|
| 95 |
+
"eval_accuracy": 0.42300821722043586,
|
| 96 |
+
"eval_f1_macro": 0.6973925526262326,
|
| 97 |
+
"eval_f1_micro": 0.7451515331237768,
|
| 98 |
+
"eval_loss": 0.5129237174987793,
|
| 99 |
+
"eval_roc_auc": 0.8459416898849885,
|
| 100 |
+
"eval_runtime": 681.0638,
|
| 101 |
+
"eval_samples_per_second": 4.11,
|
| 102 |
+
"eval_steps_per_second": 0.257,
|
| 103 |
+
"learning_rate": 0.01,
|
| 104 |
+
"step": 2680
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 5.6,
|
| 108 |
+
"learning_rate": 0.01,
|
| 109 |
+
"loss": 0.3877,
|
| 110 |
+
"step": 3000
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 6.0,
|
| 114 |
+
"eval_accuracy": 0.4258663808503037,
|
| 115 |
+
"eval_f1_macro": 0.6805137576503333,
|
| 116 |
+
"eval_f1_micro": 0.7385924713584288,
|
| 117 |
+
"eval_loss": 0.518527090549469,
|
| 118 |
+
"eval_roc_auc": 0.8180316790465337,
|
| 119 |
+
"eval_runtime": 675.3189,
|
| 120 |
+
"eval_samples_per_second": 4.145,
|
| 121 |
+
"eval_steps_per_second": 0.259,
|
| 122 |
+
"learning_rate": 0.01,
|
| 123 |
+
"step": 3216
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 6.53,
|
| 127 |
+
"learning_rate": 0.01,
|
| 128 |
+
"loss": 0.3658,
|
| 129 |
+
"step": 3500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 7.0,
|
| 133 |
+
"eval_accuracy": 0.4390853876384423,
|
| 134 |
+
"eval_f1_macro": 0.703387001922082,
|
| 135 |
+
"eval_f1_micro": 0.7709620206715185,
|
| 136 |
+
"eval_loss": 0.4688432514667511,
|
| 137 |
+
"eval_roc_auc": 0.8524541561034484,
|
| 138 |
+
"eval_runtime": 682.6316,
|
| 139 |
+
"eval_samples_per_second": 4.1,
|
| 140 |
+
"eval_steps_per_second": 0.256,
|
| 141 |
+
"learning_rate": 0.01,
|
| 142 |
+
"step": 3752
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 7.46,
|
| 146 |
+
"learning_rate": 0.01,
|
| 147 |
+
"loss": 0.373,
|
| 148 |
+
"step": 4000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 8.0,
|
| 152 |
+
"eval_accuracy": 0.3858520900321543,
|
| 153 |
+
"eval_f1_macro": 0.7019752151317524,
|
| 154 |
+
"eval_f1_micro": 0.7607600732600732,
|
| 155 |
+
"eval_loss": 0.5070469975471497,
|
| 156 |
+
"eval_roc_auc": 0.8646759677959497,
|
| 157 |
+
"eval_runtime": 677.56,
|
| 158 |
+
"eval_samples_per_second": 4.131,
|
| 159 |
+
"eval_steps_per_second": 0.258,
|
| 160 |
+
"learning_rate": 0.01,
|
| 161 |
+
"step": 4288
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 8.4,
|
| 165 |
+
"learning_rate": 0.001,
|
| 166 |
+
"loss": 0.2911,
|
| 167 |
+
"step": 4500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 9.0,
|
| 171 |
+
"eval_accuracy": 0.5087531261164702,
|
| 172 |
+
"eval_f1_macro": 0.7938389006088229,
|
| 173 |
+
"eval_f1_micro": 0.8212945870303103,
|
| 174 |
+
"eval_loss": 0.23266172409057617,
|
| 175 |
+
"eval_roc_auc": 0.8884873189781101,
|
| 176 |
+
"eval_runtime": 683.0549,
|
| 177 |
+
"eval_samples_per_second": 4.098,
|
| 178 |
+
"eval_steps_per_second": 0.256,
|
| 179 |
+
"learning_rate": 0.001,
|
| 180 |
+
"step": 4824
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 9.33,
|
| 184 |
+
"learning_rate": 0.001,
|
| 185 |
+
"loss": 0.139,
|
| 186 |
+
"step": 5000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 10.0,
|
| 190 |
+
"eval_accuracy": 0.49874955341193283,
|
| 191 |
+
"eval_f1_macro": 0.7890622914838611,
|
| 192 |
+
"eval_f1_micro": 0.8193090720341427,
|
| 193 |
+
"eval_loss": 0.22383752465248108,
|
| 194 |
+
"eval_roc_auc": 0.8971633903103337,
|
| 195 |
+
"eval_runtime": 677.2477,
|
| 196 |
+
"eval_samples_per_second": 4.133,
|
| 197 |
+
"eval_steps_per_second": 0.258,
|
| 198 |
+
"learning_rate": 0.001,
|
| 199 |
+
"step": 5360
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 10.26,
|
| 203 |
+
"learning_rate": 0.001,
|
| 204 |
+
"loss": 0.1187,
|
| 205 |
+
"step": 5500
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 11.0,
|
| 209 |
+
"eval_accuracy": 0.5083958556627367,
|
| 210 |
+
"eval_f1_macro": 0.785811349764938,
|
| 211 |
+
"eval_f1_micro": 0.8168929472545957,
|
| 212 |
+
"eval_loss": 0.20949247479438782,
|
| 213 |
+
"eval_roc_auc": 0.8833373336547201,
|
| 214 |
+
"eval_runtime": 679.1488,
|
| 215 |
+
"eval_samples_per_second": 4.121,
|
| 216 |
+
"eval_steps_per_second": 0.258,
|
| 217 |
+
"learning_rate": 0.001,
|
| 218 |
+
"step": 5896
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 11.19,
|
| 222 |
+
"learning_rate": 0.001,
|
| 223 |
+
"loss": 0.1084,
|
| 224 |
+
"step": 6000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 12.0,
|
| 228 |
+
"eval_accuracy": 0.49589138978206504,
|
| 229 |
+
"eval_f1_macro": 0.7994867347489195,
|
| 230 |
+
"eval_f1_micro": 0.8209378903145226,
|
| 231 |
+
"eval_loss": 0.19851598143577576,
|
| 232 |
+
"eval_roc_auc": 0.9030839510366602,
|
| 233 |
+
"eval_runtime": 649.8973,
|
| 234 |
+
"eval_samples_per_second": 4.307,
|
| 235 |
+
"eval_steps_per_second": 0.269,
|
| 236 |
+
"learning_rate": 0.001,
|
| 237 |
+
"step": 6432
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 12.13,
|
| 241 |
+
"learning_rate": 0.001,
|
| 242 |
+
"loss": 0.1038,
|
| 243 |
+
"step": 6500
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 13.0,
|
| 247 |
+
"eval_accuracy": 0.4901750625223294,
|
| 248 |
+
"eval_f1_macro": 0.7914460723507714,
|
| 249 |
+
"eval_f1_micro": 0.8185889606235096,
|
| 250 |
+
"eval_loss": 0.19494152069091797,
|
| 251 |
+
"eval_roc_auc": 0.8941173115453266,
|
| 252 |
+
"eval_runtime": 652.683,
|
| 253 |
+
"eval_samples_per_second": 4.288,
|
| 254 |
+
"eval_steps_per_second": 0.268,
|
| 255 |
+
"learning_rate": 0.001,
|
| 256 |
+
"step": 6968
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 13.06,
|
| 260 |
+
"learning_rate": 0.001,
|
| 261 |
+
"loss": 0.098,
|
| 262 |
+
"step": 7000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 13.99,
|
| 266 |
+
"learning_rate": 0.001,
|
| 267 |
+
"loss": 0.0936,
|
| 268 |
+
"step": 7500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 14.0,
|
| 272 |
+
"eval_accuracy": 0.5012504465880672,
|
| 273 |
+
"eval_f1_macro": 0.7854615126335902,
|
| 274 |
+
"eval_f1_micro": 0.824316394789344,
|
| 275 |
+
"eval_loss": 0.1806003302335739,
|
| 276 |
+
"eval_roc_auc": 0.894735074423797,
|
| 277 |
+
"eval_runtime": 647.2683,
|
| 278 |
+
"eval_samples_per_second": 4.324,
|
| 279 |
+
"eval_steps_per_second": 0.27,
|
| 280 |
+
"learning_rate": 0.001,
|
| 281 |
+
"step": 7504
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 14.93,
|
| 285 |
+
"learning_rate": 0.001,
|
| 286 |
+
"loss": 0.0915,
|
| 287 |
+
"step": 8000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 15.0,
|
| 291 |
+
"eval_accuracy": 0.5112540192926045,
|
| 292 |
+
"eval_f1_macro": 0.7788134070338611,
|
| 293 |
+
"eval_f1_micro": 0.8163314756481705,
|
| 294 |
+
"eval_loss": 0.17990005016326904,
|
| 295 |
+
"eval_roc_auc": 0.8805525074154283,
|
| 296 |
+
"eval_runtime": 652.5868,
|
| 297 |
+
"eval_samples_per_second": 4.289,
|
| 298 |
+
"eval_steps_per_second": 0.268,
|
| 299 |
+
"learning_rate": 0.001,
|
| 300 |
+
"step": 8040
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 15.86,
|
| 304 |
+
"learning_rate": 0.001,
|
| 305 |
+
"loss": 0.0875,
|
| 306 |
+
"step": 8500
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 16.0,
|
| 310 |
+
"eval_accuracy": 0.5026795284030011,
|
| 311 |
+
"eval_f1_macro": 0.7847917223659133,
|
| 312 |
+
"eval_f1_micro": 0.8195751227883308,
|
| 313 |
+
"eval_loss": 0.17393216490745544,
|
| 314 |
+
"eval_roc_auc": 0.889408863512499,
|
| 315 |
+
"eval_runtime": 651.4492,
|
| 316 |
+
"eval_samples_per_second": 4.297,
|
| 317 |
+
"eval_steps_per_second": 0.269,
|
| 318 |
+
"learning_rate": 0.001,
|
| 319 |
+
"step": 8576
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 16.79,
|
| 323 |
+
"learning_rate": 0.001,
|
| 324 |
+
"loss": 0.0848,
|
| 325 |
+
"step": 9000
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 17.0,
|
| 329 |
+
"eval_accuracy": 0.48981779206859594,
|
| 330 |
+
"eval_f1_macro": 0.7894356238873693,
|
| 331 |
+
"eval_f1_micro": 0.8224427743675061,
|
| 332 |
+
"eval_loss": 0.17188657820224762,
|
| 333 |
+
"eval_roc_auc": 0.9006792141073352,
|
| 334 |
+
"eval_runtime": 682.0855,
|
| 335 |
+
"eval_samples_per_second": 4.104,
|
| 336 |
+
"eval_steps_per_second": 0.257,
|
| 337 |
+
"learning_rate": 0.001,
|
| 338 |
+
"step": 9112
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 17.72,
|
| 342 |
+
"learning_rate": 0.001,
|
| 343 |
+
"loss": 0.0861,
|
| 344 |
+
"step": 9500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 18.0,
|
| 348 |
+
"eval_accuracy": 0.4962486602357985,
|
| 349 |
+
"eval_f1_macro": 0.7838583531688857,
|
| 350 |
+
"eval_f1_micro": 0.8212320255636428,
|
| 351 |
+
"eval_loss": 0.16859883069992065,
|
| 352 |
+
"eval_roc_auc": 0.8903635137148811,
|
| 353 |
+
"eval_runtime": 684.152,
|
| 354 |
+
"eval_samples_per_second": 4.091,
|
| 355 |
+
"eval_steps_per_second": 0.256,
|
| 356 |
+
"learning_rate": 0.001,
|
| 357 |
+
"step": 9648
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 18.66,
|
| 361 |
+
"learning_rate": 0.001,
|
| 362 |
+
"loss": 0.0845,
|
| 363 |
+
"step": 10000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 19.0,
|
| 367 |
+
"eval_accuracy": 0.4941050375133976,
|
| 368 |
+
"eval_f1_macro": 0.7830428904767566,
|
| 369 |
+
"eval_f1_micro": 0.8178590139123347,
|
| 370 |
+
"eval_loss": 0.16593919694423676,
|
| 371 |
+
"eval_roc_auc": 0.8934385043997719,
|
| 372 |
+
"eval_runtime": 685.5342,
|
| 373 |
+
"eval_samples_per_second": 4.083,
|
| 374 |
+
"eval_steps_per_second": 0.255,
|
| 375 |
+
"learning_rate": 0.001,
|
| 376 |
+
"step": 10184
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 19.59,
|
| 380 |
+
"learning_rate": 0.001,
|
| 381 |
+
"loss": 0.0824,
|
| 382 |
+
"step": 10500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 20.0,
|
| 386 |
+
"eval_accuracy": 0.4930332261521972,
|
| 387 |
+
"eval_f1_macro": 0.7826195114415129,
|
| 388 |
+
"eval_f1_micro": 0.8104714522624972,
|
| 389 |
+
"eval_loss": 0.1742611825466156,
|
| 390 |
+
"eval_roc_auc": 0.8838985778017561,
|
| 391 |
+
"eval_runtime": 661.552,
|
| 392 |
+
"eval_samples_per_second": 4.231,
|
| 393 |
+
"eval_steps_per_second": 0.265,
|
| 394 |
+
"learning_rate": 0.001,
|
| 395 |
+
"step": 10720
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 20.52,
|
| 399 |
+
"learning_rate": 0.001,
|
| 400 |
+
"loss": 0.0834,
|
| 401 |
+
"step": 11000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 21.0,
|
| 405 |
+
"eval_accuracy": 0.4976777420507324,
|
| 406 |
+
"eval_f1_macro": 0.7904779420228514,
|
| 407 |
+
"eval_f1_micro": 0.8183342003352406,
|
| 408 |
+
"eval_loss": 0.16005221009254456,
|
| 409 |
+
"eval_roc_auc": 0.8959053934881437,
|
| 410 |
+
"eval_runtime": 670.1613,
|
| 411 |
+
"eval_samples_per_second": 4.177,
|
| 412 |
+
"eval_steps_per_second": 0.261,
|
| 413 |
+
"learning_rate": 0.001,
|
| 414 |
+
"step": 11256
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 21.46,
|
| 418 |
+
"learning_rate": 0.001,
|
| 419 |
+
"loss": 0.0803,
|
| 420 |
+
"step": 11500
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 22.0,
|
| 424 |
+
"eval_accuracy": 0.4969632011432655,
|
| 425 |
+
"eval_f1_macro": 0.7884767354416139,
|
| 426 |
+
"eval_f1_micro": 0.8205630073110356,
|
| 427 |
+
"eval_loss": 0.16171583533287048,
|
| 428 |
+
"eval_roc_auc": 0.8984811576138753,
|
| 429 |
+
"eval_runtime": 656.3872,
|
| 430 |
+
"eval_samples_per_second": 4.264,
|
| 431 |
+
"eval_steps_per_second": 0.267,
|
| 432 |
+
"learning_rate": 0.001,
|
| 433 |
+
"step": 11792
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 22.39,
|
| 437 |
+
"learning_rate": 0.001,
|
| 438 |
+
"loss": 0.0817,
|
| 439 |
+
"step": 12000
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 23.0,
|
| 443 |
+
"eval_accuracy": 0.5037513397642015,
|
| 444 |
+
"eval_f1_macro": 0.7892866813532138,
|
| 445 |
+
"eval_f1_micro": 0.8190240160500384,
|
| 446 |
+
"eval_loss": 0.15858659148216248,
|
| 447 |
+
"eval_roc_auc": 0.8899535391642385,
|
| 448 |
+
"eval_runtime": 668.9537,
|
| 449 |
+
"eval_samples_per_second": 4.184,
|
| 450 |
+
"eval_steps_per_second": 0.262,
|
| 451 |
+
"learning_rate": 0.001,
|
| 452 |
+
"step": 12328
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 23.32,
|
| 456 |
+
"learning_rate": 0.001,
|
| 457 |
+
"loss": 0.0821,
|
| 458 |
+
"step": 12500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 24.0,
|
| 462 |
+
"eval_accuracy": 0.5148267238299392,
|
| 463 |
+
"eval_f1_macro": 0.7797850637288645,
|
| 464 |
+
"eval_f1_micro": 0.8202982902873773,
|
| 465 |
+
"eval_loss": 0.15610341727733612,
|
| 466 |
+
"eval_roc_auc": 0.8825330829765095,
|
| 467 |
+
"eval_runtime": 669.422,
|
| 468 |
+
"eval_samples_per_second": 4.181,
|
| 469 |
+
"eval_steps_per_second": 0.261,
|
| 470 |
+
"learning_rate": 0.001,
|
| 471 |
+
"step": 12864
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 24.25,
|
| 475 |
+
"learning_rate": 0.001,
|
| 476 |
+
"loss": 0.0795,
|
| 477 |
+
"step": 13000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 25.0,
|
| 481 |
+
"eval_accuracy": 0.5012504465880672,
|
| 482 |
+
"eval_f1_macro": 0.7897208141772845,
|
| 483 |
+
"eval_f1_micro": 0.8208025830258302,
|
| 484 |
+
"eval_loss": 0.15515093505382538,
|
| 485 |
+
"eval_roc_auc": 0.8981364210784262,
|
| 486 |
+
"eval_runtime": 659.9944,
|
| 487 |
+
"eval_samples_per_second": 4.241,
|
| 488 |
+
"eval_steps_per_second": 0.265,
|
| 489 |
+
"learning_rate": 0.001,
|
| 490 |
+
"step": 13400
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 25.19,
|
| 494 |
+
"learning_rate": 0.001,
|
| 495 |
+
"loss": 0.0792,
|
| 496 |
+
"step": 13500
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 26.0,
|
| 500 |
+
"eval_accuracy": 0.5048231511254019,
|
| 501 |
+
"eval_f1_macro": 0.7843716119943152,
|
| 502 |
+
"eval_f1_micro": 0.816484761734359,
|
| 503 |
+
"eval_loss": 0.1544334441423416,
|
| 504 |
+
"eval_roc_auc": 0.8852686700886229,
|
| 505 |
+
"eval_runtime": 682.4531,
|
| 506 |
+
"eval_samples_per_second": 4.101,
|
| 507 |
+
"eval_steps_per_second": 0.256,
|
| 508 |
+
"learning_rate": 0.001,
|
| 509 |
+
"step": 13936
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 26.12,
|
| 513 |
+
"learning_rate": 0.001,
|
| 514 |
+
"loss": 0.0799,
|
| 515 |
+
"step": 14000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 27.0,
|
| 519 |
+
"eval_accuracy": 0.5105394783851376,
|
| 520 |
+
"eval_f1_macro": 0.7845481004416396,
|
| 521 |
+
"eval_f1_micro": 0.8234521463967999,
|
| 522 |
+
"eval_loss": 0.15092138946056366,
|
| 523 |
+
"eval_roc_auc": 0.8889256358413551,
|
| 524 |
+
"eval_runtime": 680.1803,
|
| 525 |
+
"eval_samples_per_second": 4.115,
|
| 526 |
+
"eval_steps_per_second": 0.257,
|
| 527 |
+
"learning_rate": 0.001,
|
| 528 |
+
"step": 14472
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 27.05,
|
| 532 |
+
"learning_rate": 0.001,
|
| 533 |
+
"loss": 0.0808,
|
| 534 |
+
"step": 14500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 27.99,
|
| 538 |
+
"learning_rate": 0.001,
|
| 539 |
+
"loss": 0.0795,
|
| 540 |
+
"step": 15000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 28.0,
|
| 544 |
+
"eval_accuracy": 0.5076813147552698,
|
| 545 |
+
"eval_f1_macro": 0.7841559133264137,
|
| 546 |
+
"eval_f1_micro": 0.8207829522327307,
|
| 547 |
+
"eval_loss": 0.15122035145759583,
|
| 548 |
+
"eval_roc_auc": 0.8866264189509026,
|
| 549 |
+
"eval_runtime": 677.2495,
|
| 550 |
+
"eval_samples_per_second": 4.133,
|
| 551 |
+
"eval_steps_per_second": 0.258,
|
| 552 |
+
"learning_rate": 0.001,
|
| 553 |
+
"step": 15008
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 28.92,
|
| 557 |
+
"learning_rate": 0.001,
|
| 558 |
+
"loss": 0.079,
|
| 559 |
+
"step": 15500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 29.0,
|
| 563 |
+
"eval_accuracy": 0.510896748838871,
|
| 564 |
+
"eval_f1_macro": 0.787896717103026,
|
| 565 |
+
"eval_f1_micro": 0.8222397189752286,
|
| 566 |
+
"eval_loss": 0.1465974897146225,
|
| 567 |
+
"eval_roc_auc": 0.8839320871832598,
|
| 568 |
+
"eval_runtime": 685.2461,
|
| 569 |
+
"eval_samples_per_second": 4.085,
|
| 570 |
+
"eval_steps_per_second": 0.255,
|
| 571 |
+
"learning_rate": 0.001,
|
| 572 |
+
"step": 15544
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 29.85,
|
| 576 |
+
"learning_rate": 0.001,
|
| 577 |
+
"loss": 0.0803,
|
| 578 |
+
"step": 16000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 30.0,
|
| 582 |
+
"eval_accuracy": 0.5001786352268668,
|
| 583 |
+
"eval_f1_macro": 0.7873972046596285,
|
| 584 |
+
"eval_f1_micro": 0.8223657415488451,
|
| 585 |
+
"eval_loss": 0.14788155257701874,
|
| 586 |
+
"eval_roc_auc": 0.8962227610317826,
|
| 587 |
+
"eval_runtime": 678.2025,
|
| 588 |
+
"eval_samples_per_second": 4.127,
|
| 589 |
+
"eval_steps_per_second": 0.258,
|
| 590 |
+
"learning_rate": 0.001,
|
| 591 |
+
"step": 16080
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 30.78,
|
| 595 |
+
"learning_rate": 0.001,
|
| 596 |
+
"loss": 0.0787,
|
| 597 |
+
"step": 16500
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 31.0,
|
| 601 |
+
"eval_accuracy": 0.4948195784208646,
|
| 602 |
+
"eval_f1_macro": 0.7579386481110097,
|
| 603 |
+
"eval_f1_micro": 0.8014166455856311,
|
| 604 |
+
"eval_loss": 0.16386918723583221,
|
| 605 |
+
"eval_roc_auc": 0.8600737845838163,
|
| 606 |
+
"eval_runtime": 660.488,
|
| 607 |
+
"eval_samples_per_second": 4.238,
|
| 608 |
+
"eval_steps_per_second": 0.265,
|
| 609 |
+
"learning_rate": 0.001,
|
| 610 |
+
"step": 16616
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 31.72,
|
| 614 |
+
"learning_rate": 0.001,
|
| 615 |
+
"loss": 0.0807,
|
| 616 |
+
"step": 17000
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 32.0,
|
| 620 |
+
"eval_accuracy": 0.49660593068953196,
|
| 621 |
+
"eval_f1_macro": 0.7923913892380705,
|
| 622 |
+
"eval_f1_micro": 0.8229665071770335,
|
| 623 |
+
"eval_loss": 0.14676448702812195,
|
| 624 |
+
"eval_roc_auc": 0.8919057724127171,
|
| 625 |
+
"eval_runtime": 658.2108,
|
| 626 |
+
"eval_samples_per_second": 4.252,
|
| 627 |
+
"eval_steps_per_second": 0.266,
|
| 628 |
+
"learning_rate": 0.001,
|
| 629 |
+
"step": 17152
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 32.65,
|
| 633 |
+
"learning_rate": 0.001,
|
| 634 |
+
"loss": 0.0776,
|
| 635 |
+
"step": 17500
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 33.0,
|
| 639 |
+
"eval_accuracy": 0.4994640943193998,
|
| 640 |
+
"eval_f1_macro": 0.79165656686277,
|
| 641 |
+
"eval_f1_micro": 0.821977517779307,
|
| 642 |
+
"eval_loss": 0.1479729562997818,
|
| 643 |
+
"eval_roc_auc": 0.9004930267725352,
|
| 644 |
+
"eval_runtime": 664.8245,
|
| 645 |
+
"eval_samples_per_second": 4.21,
|
| 646 |
+
"eval_steps_per_second": 0.263,
|
| 647 |
+
"learning_rate": 0.001,
|
| 648 |
+
"step": 17688
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 33.58,
|
| 652 |
+
"learning_rate": 0.001,
|
| 653 |
+
"loss": 0.0802,
|
| 654 |
+
"step": 18000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 34.0,
|
| 658 |
+
"eval_accuracy": 0.5030367988567346,
|
| 659 |
+
"eval_f1_macro": 0.7907408943492529,
|
| 660 |
+
"eval_f1_micro": 0.8228120099889656,
|
| 661 |
+
"eval_loss": 0.14384005963802338,
|
| 662 |
+
"eval_roc_auc": 0.8970632032341331,
|
| 663 |
+
"eval_runtime": 653.9329,
|
| 664 |
+
"eval_samples_per_second": 4.28,
|
| 665 |
+
"eval_steps_per_second": 0.268,
|
| 666 |
+
"learning_rate": 0.001,
|
| 667 |
+
"step": 18224
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 34.51,
|
| 671 |
+
"learning_rate": 0.001,
|
| 672 |
+
"loss": 0.0797,
|
| 673 |
+
"step": 18500
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 35.0,
|
| 677 |
+
"eval_accuracy": 0.4908896034297964,
|
| 678 |
+
"eval_f1_macro": 0.785398158465356,
|
| 679 |
+
"eval_f1_micro": 0.8206463833313602,
|
| 680 |
+
"eval_loss": 0.14965663850307465,
|
| 681 |
+
"eval_roc_auc": 0.8899359973481499,
|
| 682 |
+
"eval_runtime": 655.2313,
|
| 683 |
+
"eval_samples_per_second": 4.272,
|
| 684 |
+
"eval_steps_per_second": 0.267,
|
| 685 |
+
"learning_rate": 0.001,
|
| 686 |
+
"step": 18760
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 35.45,
|
| 690 |
+
"learning_rate": 0.001,
|
| 691 |
+
"loss": 0.0781,
|
| 692 |
+
"step": 19000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 36.0,
|
| 696 |
+
"eval_accuracy": 0.510896748838871,
|
| 697 |
+
"eval_f1_macro": 0.7947026561241459,
|
| 698 |
+
"eval_f1_micro": 0.8267408810990052,
|
| 699 |
+
"eval_loss": 0.14068089425563812,
|
| 700 |
+
"eval_roc_auc": 0.8933370435378387,
|
| 701 |
+
"eval_runtime": 668.7519,
|
| 702 |
+
"eval_samples_per_second": 4.185,
|
| 703 |
+
"eval_steps_per_second": 0.262,
|
| 704 |
+
"learning_rate": 0.001,
|
| 705 |
+
"step": 19296
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 36.38,
|
| 709 |
+
"learning_rate": 0.001,
|
| 710 |
+
"loss": 0.0791,
|
| 711 |
+
"step": 19500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 37.0,
|
| 715 |
+
"eval_accuracy": 0.5133976420150054,
|
| 716 |
+
"eval_f1_macro": 0.7714447792962602,
|
| 717 |
+
"eval_f1_micro": 0.8219422322595982,
|
| 718 |
+
"eval_loss": 0.14676533639431,
|
| 719 |
+
"eval_roc_auc": 0.8894541780339088,
|
| 720 |
+
"eval_runtime": 686.4859,
|
| 721 |
+
"eval_samples_per_second": 4.077,
|
| 722 |
+
"eval_steps_per_second": 0.255,
|
| 723 |
+
"learning_rate": 0.001,
|
| 724 |
+
"step": 19832
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 37.31,
|
| 728 |
+
"learning_rate": 0.001,
|
| 729 |
+
"loss": 0.082,
|
| 730 |
+
"step": 20000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 38.0,
|
| 734 |
+
"eval_accuracy": 0.4858878170775277,
|
| 735 |
+
"eval_f1_macro": 0.7883082180339228,
|
| 736 |
+
"eval_f1_micro": 0.8104836340130457,
|
| 737 |
+
"eval_loss": 0.1538211852312088,
|
| 738 |
+
"eval_roc_auc": 0.8862558876743165,
|
| 739 |
+
"eval_runtime": 678.2901,
|
| 740 |
+
"eval_samples_per_second": 4.127,
|
| 741 |
+
"eval_steps_per_second": 0.258,
|
| 742 |
+
"learning_rate": 0.001,
|
| 743 |
+
"step": 20368
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 38.25,
|
| 747 |
+
"learning_rate": 0.001,
|
| 748 |
+
"loss": 0.0781,
|
| 749 |
+
"step": 20500
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 39.0,
|
| 753 |
+
"eval_accuracy": 0.5055376920328689,
|
| 754 |
+
"eval_f1_macro": 0.7857732289175323,
|
| 755 |
+
"eval_f1_micro": 0.8208928886536085,
|
| 756 |
+
"eval_loss": 0.1463109254837036,
|
| 757 |
+
"eval_roc_auc": 0.8920067982385395,
|
| 758 |
+
"eval_runtime": 661.0429,
|
| 759 |
+
"eval_samples_per_second": 4.234,
|
| 760 |
+
"eval_steps_per_second": 0.265,
|
| 761 |
+
"learning_rate": 0.001,
|
| 762 |
+
"step": 20904
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 39.18,
|
| 766 |
+
"learning_rate": 0.001,
|
| 767 |
+
"loss": 0.0811,
|
| 768 |
+
"step": 21000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 40.0,
|
| 772 |
+
"eval_accuracy": 0.4976777420507324,
|
| 773 |
+
"eval_f1_macro": 0.7789772985289035,
|
| 774 |
+
"eval_f1_micro": 0.8150636492220651,
|
| 775 |
+
"eval_loss": 0.14692291617393494,
|
| 776 |
+
"eval_roc_auc": 0.8880396099752541,
|
| 777 |
+
"eval_runtime": 656.7404,
|
| 778 |
+
"eval_samples_per_second": 4.262,
|
| 779 |
+
"eval_steps_per_second": 0.266,
|
| 780 |
+
"learning_rate": 0.001,
|
| 781 |
+
"step": 21440
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 40.11,
|
| 785 |
+
"learning_rate": 0.001,
|
| 786 |
+
"loss": 0.0786,
|
| 787 |
+
"step": 21500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 41.0,
|
| 791 |
+
"eval_accuracy": 0.5051804215791355,
|
| 792 |
+
"eval_f1_macro": 0.7689577848820183,
|
| 793 |
+
"eval_f1_micro": 0.8167210198636229,
|
| 794 |
+
"eval_loss": 0.1517634242773056,
|
| 795 |
+
"eval_roc_auc": 0.8871563695003593,
|
| 796 |
+
"eval_runtime": 661.7774,
|
| 797 |
+
"eval_samples_per_second": 4.23,
|
| 798 |
+
"eval_steps_per_second": 0.264,
|
| 799 |
+
"learning_rate": 0.001,
|
| 800 |
+
"step": 21976
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 41.04,
|
| 804 |
+
"learning_rate": 0.001,
|
| 805 |
+
"loss": 0.0801,
|
| 806 |
+
"step": 22000
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 41.98,
|
| 810 |
+
"learning_rate": 0.001,
|
| 811 |
+
"loss": 0.0775,
|
| 812 |
+
"step": 22500
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 42.0,
|
| 816 |
+
"eval_accuracy": 0.5130403715612719,
|
| 817 |
+
"eval_f1_macro": 0.7913206484933839,
|
| 818 |
+
"eval_f1_micro": 0.8260360571294779,
|
| 819 |
+
"eval_loss": 0.1422175019979477,
|
| 820 |
+
"eval_roc_auc": 0.8964987386597802,
|
| 821 |
+
"eval_runtime": 656.5585,
|
| 822 |
+
"eval_samples_per_second": 4.263,
|
| 823 |
+
"eval_steps_per_second": 0.267,
|
| 824 |
+
"learning_rate": 0.001,
|
| 825 |
+
"step": 22512
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 42.91,
|
| 829 |
+
"learning_rate": 0.0001,
|
| 830 |
+
"loss": 0.0641,
|
| 831 |
+
"step": 23000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 43.0,
|
| 835 |
+
"eval_accuracy": 0.5248302965344765,
|
| 836 |
+
"eval_f1_macro": 0.8001162834127045,
|
| 837 |
+
"eval_f1_micro": 0.8339575530586767,
|
| 838 |
+
"eval_loss": 0.13186925649642944,
|
| 839 |
+
"eval_roc_auc": 0.8962547274852131,
|
| 840 |
+
"eval_runtime": 660.0805,
|
| 841 |
+
"eval_samples_per_second": 4.24,
|
| 842 |
+
"eval_steps_per_second": 0.265,
|
| 843 |
+
"learning_rate": 0.0001,
|
| 844 |
+
"step": 23048
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 43.84,
|
| 848 |
+
"learning_rate": 0.0001,
|
| 849 |
+
"loss": 0.0633,
|
| 850 |
+
"step": 23500
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 44.0,
|
| 854 |
+
"eval_accuracy": 0.5298320828867452,
|
| 855 |
+
"eval_f1_macro": 0.7959357724172665,
|
| 856 |
+
"eval_f1_micro": 0.8325938361913899,
|
| 857 |
+
"eval_loss": 0.13134370744228363,
|
| 858 |
+
"eval_roc_auc": 0.8928378169786795,
|
| 859 |
+
"eval_runtime": 657.8245,
|
| 860 |
+
"eval_samples_per_second": 4.255,
|
| 861 |
+
"eval_steps_per_second": 0.266,
|
| 862 |
+
"learning_rate": 0.0001,
|
| 863 |
+
"step": 23584
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 44.78,
|
| 867 |
+
"learning_rate": 0.0001,
|
| 868 |
+
"loss": 0.0627,
|
| 869 |
+
"step": 24000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 45.0,
|
| 873 |
+
"eval_accuracy": 0.5241157556270096,
|
| 874 |
+
"eval_f1_macro": 0.7994285450135317,
|
| 875 |
+
"eval_f1_micro": 0.8324022346368715,
|
| 876 |
+
"eval_loss": 0.13141845166683197,
|
| 877 |
+
"eval_roc_auc": 0.8954548364488818,
|
| 878 |
+
"eval_runtime": 680.5175,
|
| 879 |
+
"eval_samples_per_second": 4.113,
|
| 880 |
+
"eval_steps_per_second": 0.257,
|
| 881 |
+
"learning_rate": 0.0001,
|
| 882 |
+
"step": 24120
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 45.71,
|
| 886 |
+
"learning_rate": 0.0001,
|
| 887 |
+
"loss": 0.0627,
|
| 888 |
+
"step": 24500
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 46.0,
|
| 892 |
+
"eval_accuracy": 0.5234012147195427,
|
| 893 |
+
"eval_f1_macro": 0.8008904907382265,
|
| 894 |
+
"eval_f1_micro": 0.8323726899994057,
|
| 895 |
+
"eval_loss": 0.13084658980369568,
|
| 896 |
+
"eval_roc_auc": 0.895493140508376,
|
| 897 |
+
"eval_runtime": 683.1876,
|
| 898 |
+
"eval_samples_per_second": 4.097,
|
| 899 |
+
"eval_steps_per_second": 0.256,
|
| 900 |
+
"learning_rate": 0.0001,
|
| 901 |
+
"step": 24656
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 46.64,
|
| 905 |
+
"learning_rate": 0.0001,
|
| 906 |
+
"loss": 0.0619,
|
| 907 |
+
"step": 25000
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 47.0,
|
| 911 |
+
"eval_accuracy": 0.5251875669882101,
|
| 912 |
+
"eval_f1_macro": 0.7996410381874448,
|
| 913 |
+
"eval_f1_micro": 0.8332838038632987,
|
| 914 |
+
"eval_loss": 0.13081477582454681,
|
| 915 |
+
"eval_roc_auc": 0.8959421232969647,
|
| 916 |
+
"eval_runtime": 678.1257,
|
| 917 |
+
"eval_samples_per_second": 4.128,
|
| 918 |
+
"eval_steps_per_second": 0.258,
|
| 919 |
+
"learning_rate": 0.0001,
|
| 920 |
+
"step": 25192
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 47.57,
|
| 924 |
+
"learning_rate": 0.0001,
|
| 925 |
+
"loss": 0.0626,
|
| 926 |
+
"step": 25500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 48.0,
|
| 930 |
+
"eval_accuracy": 0.5198285101822079,
|
| 931 |
+
"eval_f1_macro": 0.8009337599160304,
|
| 932 |
+
"eval_f1_micro": 0.8332938107659472,
|
| 933 |
+
"eval_loss": 0.13103964924812317,
|
| 934 |
+
"eval_roc_auc": 0.8967411755836743,
|
| 935 |
+
"eval_runtime": 686.2682,
|
| 936 |
+
"eval_samples_per_second": 4.079,
|
| 937 |
+
"eval_steps_per_second": 0.255,
|
| 938 |
+
"learning_rate": 0.0001,
|
| 939 |
+
"step": 25728
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 48.51,
|
| 943 |
+
"learning_rate": 0.0001,
|
| 944 |
+
"loss": 0.063,
|
| 945 |
+
"step": 26000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 49.0,
|
| 949 |
+
"eval_accuracy": 0.5198285101822079,
|
| 950 |
+
"eval_f1_macro": 0.7989267774722452,
|
| 951 |
+
"eval_f1_micro": 0.832758825627006,
|
| 952 |
+
"eval_loss": 0.13107521831989288,
|
| 953 |
+
"eval_roc_auc": 0.8956594043493923,
|
| 954 |
+
"eval_runtime": 697.4946,
|
| 955 |
+
"eval_samples_per_second": 4.013,
|
| 956 |
+
"eval_steps_per_second": 0.251,
|
| 957 |
+
"learning_rate": 0.0001,
|
| 958 |
+
"step": 26264
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 49.44,
|
| 962 |
+
"learning_rate": 0.0001,
|
| 963 |
+
"loss": 0.0623,
|
| 964 |
+
"step": 26500
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 50.0,
|
| 968 |
+
"eval_accuracy": 0.5234012147195427,
|
| 969 |
+
"eval_f1_macro": 0.7989859780104293,
|
| 970 |
+
"eval_f1_micro": 0.8329772079772079,
|
| 971 |
+
"eval_loss": 0.13077349960803986,
|
| 972 |
+
"eval_roc_auc": 0.8961903282196413,
|
| 973 |
+
"eval_runtime": 670.019,
|
| 974 |
+
"eval_samples_per_second": 4.177,
|
| 975 |
+
"eval_steps_per_second": 0.261,
|
| 976 |
+
"learning_rate": 0.0001,
|
| 977 |
+
"step": 26800
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 50.37,
|
| 981 |
+
"learning_rate": 0.0001,
|
| 982 |
+
"loss": 0.0627,
|
| 983 |
+
"step": 27000
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 51.0,
|
| 987 |
+
"eval_accuracy": 0.5219721329046088,
|
| 988 |
+
"eval_f1_macro": 0.8007681086108336,
|
| 989 |
+
"eval_f1_micro": 0.8329191811619926,
|
| 990 |
+
"eval_loss": 0.13088488578796387,
|
| 991 |
+
"eval_roc_auc": 0.89715255849122,
|
| 992 |
+
"eval_runtime": 663.0956,
|
| 993 |
+
"eval_samples_per_second": 4.221,
|
| 994 |
+
"eval_steps_per_second": 0.264,
|
| 995 |
+
"learning_rate": 0.0001,
|
| 996 |
+
"step": 27336
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 51.31,
|
| 1000 |
+
"learning_rate": 0.0001,
|
| 1001 |
+
"loss": 0.0624,
|
| 1002 |
+
"step": 27500
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 52.0,
|
| 1006 |
+
"eval_accuracy": 0.5255448374419436,
|
| 1007 |
+
"eval_f1_macro": 0.7964913854037948,
|
| 1008 |
+
"eval_f1_micro": 0.8308876593697379,
|
| 1009 |
+
"eval_loss": 0.13053081929683685,
|
| 1010 |
+
"eval_roc_auc": 0.890949278817321,
|
| 1011 |
+
"eval_runtime": 670.678,
|
| 1012 |
+
"eval_samples_per_second": 4.173,
|
| 1013 |
+
"eval_steps_per_second": 0.261,
|
| 1014 |
+
"learning_rate": 0.0001,
|
| 1015 |
+
"step": 27872
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 52.24,
|
| 1019 |
+
"learning_rate": 0.0001,
|
| 1020 |
+
"loss": 0.0626,
|
| 1021 |
+
"step": 28000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 53.0,
|
| 1025 |
+
"eval_accuracy": 0.5230439442658092,
|
| 1026 |
+
"eval_f1_macro": 0.7991690934335338,
|
| 1027 |
+
"eval_f1_micro": 0.8313217194839171,
|
| 1028 |
+
"eval_loss": 0.13074837625026703,
|
| 1029 |
+
"eval_roc_auc": 0.8947062970750835,
|
| 1030 |
+
"eval_runtime": 675.1165,
|
| 1031 |
+
"eval_samples_per_second": 4.146,
|
| 1032 |
+
"eval_steps_per_second": 0.259,
|
| 1033 |
+
"learning_rate": 0.0001,
|
| 1034 |
+
"step": 28408
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 53.17,
|
| 1038 |
+
"learning_rate": 0.0001,
|
| 1039 |
+
"loss": 0.0621,
|
| 1040 |
+
"step": 28500
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 54.0,
|
| 1044 |
+
"eval_accuracy": 0.5223294033583422,
|
| 1045 |
+
"eval_f1_macro": 0.7955466725271905,
|
| 1046 |
+
"eval_f1_micro": 0.8319223254987862,
|
| 1047 |
+
"eval_loss": 0.13037194311618805,
|
| 1048 |
+
"eval_roc_auc": 0.8963756768048193,
|
| 1049 |
+
"eval_runtime": 663.6427,
|
| 1050 |
+
"eval_samples_per_second": 4.218,
|
| 1051 |
+
"eval_steps_per_second": 0.264,
|
| 1052 |
+
"learning_rate": 0.0001,
|
| 1053 |
+
"step": 28944
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 54.1,
|
| 1057 |
+
"learning_rate": 0.0001,
|
| 1058 |
+
"loss": 0.0631,
|
| 1059 |
+
"step": 29000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 55.0,
|
| 1063 |
+
"eval_accuracy": 0.5248302965344765,
|
| 1064 |
+
"eval_f1_macro": 0.8000559108101813,
|
| 1065 |
+
"eval_f1_micro": 0.8328169685414681,
|
| 1066 |
+
"eval_loss": 0.1299433708190918,
|
| 1067 |
+
"eval_roc_auc": 0.894918579615961,
|
| 1068 |
+
"eval_runtime": 684.8991,
|
| 1069 |
+
"eval_samples_per_second": 4.087,
|
| 1070 |
+
"eval_steps_per_second": 0.256,
|
| 1071 |
+
"learning_rate": 0.0001,
|
| 1072 |
+
"step": 29480
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 55.04,
|
| 1076 |
+
"learning_rate": 0.0001,
|
| 1077 |
+
"loss": 0.0621,
|
| 1078 |
+
"step": 29500
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 55.97,
|
| 1082 |
+
"learning_rate": 0.0001,
|
| 1083 |
+
"loss": 0.063,
|
| 1084 |
+
"step": 30000
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 56.0,
|
| 1088 |
+
"eval_accuracy": 0.5223294033583422,
|
| 1089 |
+
"eval_f1_macro": 0.7989243557029326,
|
| 1090 |
+
"eval_f1_micro": 0.8320864505403159,
|
| 1091 |
+
"eval_loss": 0.13019777834415436,
|
| 1092 |
+
"eval_roc_auc": 0.8955682056992913,
|
| 1093 |
+
"eval_runtime": 679.7556,
|
| 1094 |
+
"eval_samples_per_second": 4.118,
|
| 1095 |
+
"eval_steps_per_second": 0.257,
|
| 1096 |
+
"learning_rate": 0.0001,
|
| 1097 |
+
"step": 30016
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 56.9,
|
| 1101 |
+
"learning_rate": 0.0001,
|
| 1102 |
+
"loss": 0.0621,
|
| 1103 |
+
"step": 30500
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 57.0,
|
| 1107 |
+
"eval_accuracy": 0.5230439442658092,
|
| 1108 |
+
"eval_f1_macro": 0.7969527402194384,
|
| 1109 |
+
"eval_f1_micro": 0.8289875921596835,
|
| 1110 |
+
"eval_loss": 0.1304127722978592,
|
| 1111 |
+
"eval_roc_auc": 0.8908787078394556,
|
| 1112 |
+
"eval_runtime": 705.4131,
|
| 1113 |
+
"eval_samples_per_second": 3.968,
|
| 1114 |
+
"eval_steps_per_second": 0.248,
|
| 1115 |
+
"learning_rate": 0.0001,
|
| 1116 |
+
"step": 30552
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 57.84,
|
| 1120 |
+
"learning_rate": 0.0001,
|
| 1121 |
+
"loss": 0.0623,
|
| 1122 |
+
"step": 31000
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 58.0,
|
| 1126 |
+
"eval_accuracy": 0.5237584851732762,
|
| 1127 |
+
"eval_f1_macro": 0.7977835811692552,
|
| 1128 |
+
"eval_f1_micro": 0.8301569169722841,
|
| 1129 |
+
"eval_loss": 0.13054853677749634,
|
| 1130 |
+
"eval_roc_auc": 0.8906267128821808,
|
| 1131 |
+
"eval_runtime": 666.4705,
|
| 1132 |
+
"eval_samples_per_second": 4.2,
|
| 1133 |
+
"eval_steps_per_second": 0.263,
|
| 1134 |
+
"learning_rate": 0.0001,
|
| 1135 |
+
"step": 31088
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 58.77,
|
| 1139 |
+
"learning_rate": 0.0001,
|
| 1140 |
+
"loss": 0.0622,
|
| 1141 |
+
"step": 31500
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 59.0,
|
| 1145 |
+
"eval_accuracy": 0.5237584851732762,
|
| 1146 |
+
"eval_f1_macro": 0.796535523287011,
|
| 1147 |
+
"eval_f1_micro": 0.8307526107310047,
|
| 1148 |
+
"eval_loss": 0.1307431012392044,
|
| 1149 |
+
"eval_roc_auc": 0.8914970403252072,
|
| 1150 |
+
"eval_runtime": 676.3635,
|
| 1151 |
+
"eval_samples_per_second": 4.138,
|
| 1152 |
+
"eval_steps_per_second": 0.259,
|
| 1153 |
+
"learning_rate": 0.0001,
|
| 1154 |
+
"step": 31624
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 59.7,
|
| 1158 |
+
"learning_rate": 0.0001,
|
| 1159 |
+
"loss": 0.0627,
|
| 1160 |
+
"step": 32000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 60.0,
|
| 1164 |
+
"eval_accuracy": 0.5301893533404788,
|
| 1165 |
+
"eval_f1_macro": 0.7997719728514119,
|
| 1166 |
+
"eval_f1_micro": 0.8327465839250552,
|
| 1167 |
+
"eval_loss": 0.12943749129772186,
|
| 1168 |
+
"eval_roc_auc": 0.8944175409863876,
|
| 1169 |
+
"eval_runtime": 664.0543,
|
| 1170 |
+
"eval_samples_per_second": 4.215,
|
| 1171 |
+
"eval_steps_per_second": 0.264,
|
| 1172 |
+
"learning_rate": 0.0001,
|
| 1173 |
+
"step": 32160
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 60.63,
|
| 1177 |
+
"learning_rate": 0.0001,
|
| 1178 |
+
"loss": 0.0627,
|
| 1179 |
+
"step": 32500
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 61.0,
|
| 1183 |
+
"eval_accuracy": 0.5241157556270096,
|
| 1184 |
+
"eval_f1_macro": 0.8000371368033877,
|
| 1185 |
+
"eval_f1_micro": 0.8319088319088318,
|
| 1186 |
+
"eval_loss": 0.13030506670475006,
|
| 1187 |
+
"eval_roc_auc": 0.8955766245181099,
|
| 1188 |
+
"eval_runtime": 682.9901,
|
| 1189 |
+
"eval_samples_per_second": 4.098,
|
| 1190 |
+
"eval_steps_per_second": 0.256,
|
| 1191 |
+
"learning_rate": 0.0001,
|
| 1192 |
+
"step": 32696
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 61.57,
|
| 1196 |
+
"learning_rate": 0.0001,
|
| 1197 |
+
"loss": 0.0626,
|
| 1198 |
+
"step": 33000
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 62.0,
|
| 1202 |
+
"eval_accuracy": 0.5244730260807431,
|
| 1203 |
+
"eval_f1_macro": 0.7981976460241279,
|
| 1204 |
+
"eval_f1_micro": 0.8316939231187014,
|
| 1205 |
+
"eval_loss": 0.13014619052410126,
|
| 1206 |
+
"eval_roc_auc": 0.8903578802872688,
|
| 1207 |
+
"eval_runtime": 683.442,
|
| 1208 |
+
"eval_samples_per_second": 4.095,
|
| 1209 |
+
"eval_steps_per_second": 0.256,
|
| 1210 |
+
"learning_rate": 0.0001,
|
| 1211 |
+
"step": 33232
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 62.5,
|
| 1215 |
+
"learning_rate": 0.0001,
|
| 1216 |
+
"loss": 0.0629,
|
| 1217 |
+
"step": 33500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 63.0,
|
| 1221 |
+
"eval_accuracy": 0.5248302965344765,
|
| 1222 |
+
"eval_f1_macro": 0.798892356454433,
|
| 1223 |
+
"eval_f1_micro": 0.8321628377574097,
|
| 1224 |
+
"eval_loss": 0.12966355681419373,
|
| 1225 |
+
"eval_roc_auc": 0.8948756467722464,
|
| 1226 |
+
"eval_runtime": 688.3652,
|
| 1227 |
+
"eval_samples_per_second": 4.066,
|
| 1228 |
+
"eval_steps_per_second": 0.254,
|
| 1229 |
+
"learning_rate": 0.0001,
|
| 1230 |
+
"step": 33768
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 63.43,
|
| 1234 |
+
"learning_rate": 0.0001,
|
| 1235 |
+
"loss": 0.0617,
|
| 1236 |
+
"step": 34000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 64.0,
|
| 1240 |
+
"eval_accuracy": 0.5244730260807431,
|
| 1241 |
+
"eval_f1_macro": 0.7981842150860963,
|
| 1242 |
+
"eval_f1_micro": 0.8310847274689693,
|
| 1243 |
+
"eval_loss": 0.13000145554542542,
|
| 1244 |
+
"eval_roc_auc": 0.8919613178233593,
|
| 1245 |
+
"eval_runtime": 676.0033,
|
| 1246 |
+
"eval_samples_per_second": 4.141,
|
| 1247 |
+
"eval_steps_per_second": 0.259,
|
| 1248 |
+
"learning_rate": 0.0001,
|
| 1249 |
+
"step": 34304
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 64.37,
|
| 1253 |
+
"learning_rate": 0.0001,
|
| 1254 |
+
"loss": 0.0631,
|
| 1255 |
+
"step": 34500
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 65.0,
|
| 1259 |
+
"eval_accuracy": 0.5244730260807431,
|
| 1260 |
+
"eval_f1_macro": 0.7986308658625665,
|
| 1261 |
+
"eval_f1_micro": 0.8318573480134035,
|
| 1262 |
+
"eval_loss": 0.12924692034721375,
|
| 1263 |
+
"eval_roc_auc": 0.8930446319857147,
|
| 1264 |
+
"eval_runtime": 651.5802,
|
| 1265 |
+
"eval_samples_per_second": 4.296,
|
| 1266 |
+
"eval_steps_per_second": 0.269,
|
| 1267 |
+
"learning_rate": 0.0001,
|
| 1268 |
+
"step": 34840
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 65.3,
|
| 1272 |
+
"learning_rate": 0.0001,
|
| 1273 |
+
"loss": 0.0619,
|
| 1274 |
+
"step": 35000
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 66.0,
|
| 1278 |
+
"eval_accuracy": 0.5298320828867452,
|
| 1279 |
+
"eval_f1_macro": 0.7981870614941995,
|
| 1280 |
+
"eval_f1_micro": 0.831923192319232,
|
| 1281 |
+
"eval_loss": 0.12978804111480713,
|
| 1282 |
+
"eval_roc_auc": 0.892217237386403,
|
| 1283 |
+
"eval_runtime": 681.5539,
|
| 1284 |
+
"eval_samples_per_second": 4.107,
|
| 1285 |
+
"eval_steps_per_second": 0.257,
|
| 1286 |
+
"learning_rate": 0.0001,
|
| 1287 |
+
"step": 35376
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 66.23,
|
| 1291 |
+
"learning_rate": 0.0001,
|
| 1292 |
+
"loss": 0.0636,
|
| 1293 |
+
"step": 35500
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 67.0,
|
| 1297 |
+
"eval_accuracy": 0.5244730260807431,
|
| 1298 |
+
"eval_f1_macro": 0.799861643126952,
|
| 1299 |
+
"eval_f1_micro": 0.832350860646074,
|
| 1300 |
+
"eval_loss": 0.12984319031238556,
|
| 1301 |
+
"eval_roc_auc": 0.8979669054874929,
|
| 1302 |
+
"eval_runtime": 683.6934,
|
| 1303 |
+
"eval_samples_per_second": 4.094,
|
| 1304 |
+
"eval_steps_per_second": 0.256,
|
| 1305 |
+
"learning_rate": 0.0001,
|
| 1306 |
+
"step": 35912
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 67.16,
|
| 1310 |
+
"learning_rate": 0.0001,
|
| 1311 |
+
"loss": 0.0627,
|
| 1312 |
+
"step": 36000
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 68.0,
|
| 1316 |
+
"eval_accuracy": 0.5194712397284744,
|
| 1317 |
+
"eval_f1_macro": 0.8006367293204648,
|
| 1318 |
+
"eval_f1_micro": 0.8319219007292402,
|
| 1319 |
+
"eval_loss": 0.12975776195526123,
|
| 1320 |
+
"eval_roc_auc": 0.8984548965140277,
|
| 1321 |
+
"eval_runtime": 683.9871,
|
| 1322 |
+
"eval_samples_per_second": 4.092,
|
| 1323 |
+
"eval_steps_per_second": 0.256,
|
| 1324 |
+
"learning_rate": 0.0001,
|
| 1325 |
+
"step": 36448
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 68.1,
|
| 1329 |
+
"learning_rate": 0.0001,
|
| 1330 |
+
"loss": 0.0624,
|
| 1331 |
+
"step": 36500
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 69.0,
|
| 1335 |
+
"eval_accuracy": 0.525902107895677,
|
| 1336 |
+
"eval_f1_macro": 0.7980251812621267,
|
| 1337 |
+
"eval_f1_micro": 0.8309201866698576,
|
| 1338 |
+
"eval_loss": 0.12929844856262207,
|
| 1339 |
+
"eval_roc_auc": 0.8925473833907398,
|
| 1340 |
+
"eval_runtime": 653.0587,
|
| 1341 |
+
"eval_samples_per_second": 4.286,
|
| 1342 |
+
"eval_steps_per_second": 0.268,
|
| 1343 |
+
"learning_rate": 0.0001,
|
| 1344 |
+
"step": 36984
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 69.03,
|
| 1348 |
+
"learning_rate": 0.0001,
|
| 1349 |
+
"loss": 0.0629,
|
| 1350 |
+
"step": 37000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 69.96,
|
| 1354 |
+
"learning_rate": 0.0001,
|
| 1355 |
+
"loss": 0.0625,
|
| 1356 |
+
"step": 37500
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 70.0,
|
| 1360 |
+
"eval_accuracy": 0.5244730260807431,
|
| 1361 |
+
"eval_f1_macro": 0.7967283277711645,
|
| 1362 |
+
"eval_f1_micro": 0.8312570781426953,
|
| 1363 |
+
"eval_loss": 0.13054105639457703,
|
| 1364 |
+
"eval_roc_auc": 0.8938972830414821,
|
| 1365 |
+
"eval_runtime": 661.1431,
|
| 1366 |
+
"eval_samples_per_second": 4.234,
|
| 1367 |
+
"eval_steps_per_second": 0.265,
|
| 1368 |
+
"learning_rate": 0.0001,
|
| 1369 |
+
"step": 37520
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 70.9,
|
| 1373 |
+
"learning_rate": 0.0001,
|
| 1374 |
+
"loss": 0.0624,
|
| 1375 |
+
"step": 38000
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 71.0,
|
| 1379 |
+
"eval_accuracy": 0.5166130760986066,
|
| 1380 |
+
"eval_f1_macro": 0.7941604755891036,
|
| 1381 |
+
"eval_f1_micro": 0.828351938542792,
|
| 1382 |
+
"eval_loss": 0.13031570613384247,
|
| 1383 |
+
"eval_roc_auc": 0.8901332543218042,
|
| 1384 |
+
"eval_runtime": 652.968,
|
| 1385 |
+
"eval_samples_per_second": 4.287,
|
| 1386 |
+
"eval_steps_per_second": 0.268,
|
| 1387 |
+
"learning_rate": 0.0001,
|
| 1388 |
+
"step": 38056
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 71.83,
|
| 1392 |
+
"learning_rate": 1e-05,
|
| 1393 |
+
"loss": 0.0618,
|
| 1394 |
+
"step": 38500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 72.0,
|
| 1398 |
+
"eval_accuracy": 0.526616648803144,
|
| 1399 |
+
"eval_f1_macro": 0.8010117361492859,
|
| 1400 |
+
"eval_f1_micro": 0.833323387241153,
|
| 1401 |
+
"eval_loss": 0.12878653407096863,
|
| 1402 |
+
"eval_roc_auc": 0.8947102216808521,
|
| 1403 |
+
"eval_runtime": 664.2267,
|
| 1404 |
+
"eval_samples_per_second": 4.214,
|
| 1405 |
+
"eval_steps_per_second": 0.263,
|
| 1406 |
+
"learning_rate": 1e-05,
|
| 1407 |
+
"step": 38592
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 72.76,
|
| 1411 |
+
"learning_rate": 1e-05,
|
| 1412 |
+
"loss": 0.0615,
|
| 1413 |
+
"step": 39000
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 73.0,
|
| 1417 |
+
"eval_accuracy": 0.5291175419792783,
|
| 1418 |
+
"eval_f1_macro": 0.7990050923407436,
|
| 1419 |
+
"eval_f1_micro": 0.8323747903187154,
|
| 1420 |
+
"eval_loss": 0.12884607911109924,
|
| 1421 |
+
"eval_roc_auc": 0.8929711097228729,
|
| 1422 |
+
"eval_runtime": 659.8974,
|
| 1423 |
+
"eval_samples_per_second": 4.242,
|
| 1424 |
+
"eval_steps_per_second": 0.265,
|
| 1425 |
+
"learning_rate": 1e-05,
|
| 1426 |
+
"step": 39128
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 73.69,
|
| 1430 |
+
"learning_rate": 1e-05,
|
| 1431 |
+
"loss": 0.0602,
|
| 1432 |
+
"step": 39500
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 74.0,
|
| 1436 |
+
"eval_accuracy": 0.5251875669882101,
|
| 1437 |
+
"eval_f1_macro": 0.7988669656016311,
|
| 1438 |
+
"eval_f1_micro": 0.8322677024200775,
|
| 1439 |
+
"eval_loss": 0.1286717653274536,
|
| 1440 |
+
"eval_roc_auc": 0.8937019727094121,
|
| 1441 |
+
"eval_runtime": 656.8867,
|
| 1442 |
+
"eval_samples_per_second": 4.261,
|
| 1443 |
+
"eval_steps_per_second": 0.266,
|
| 1444 |
+
"learning_rate": 1e-05,
|
| 1445 |
+
"step": 39664
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 74.63,
|
| 1449 |
+
"learning_rate": 1e-05,
|
| 1450 |
+
"loss": 0.0612,
|
| 1451 |
+
"step": 40000
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 75.0,
|
| 1455 |
+
"eval_accuracy": 0.5262593783494105,
|
| 1456 |
+
"eval_f1_macro": 0.8003789338109041,
|
| 1457 |
+
"eval_f1_micro": 0.8326477076253503,
|
| 1458 |
+
"eval_loss": 0.12862080335617065,
|
| 1459 |
+
"eval_roc_auc": 0.8946190230307514,
|
| 1460 |
+
"eval_runtime": 674.3061,
|
| 1461 |
+
"eval_samples_per_second": 4.151,
|
| 1462 |
+
"eval_steps_per_second": 0.26,
|
| 1463 |
+
"learning_rate": 1e-05,
|
| 1464 |
+
"step": 40200
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 75.56,
|
| 1468 |
+
"learning_rate": 1e-05,
|
| 1469 |
+
"loss": 0.0611,
|
| 1470 |
+
"step": 40500
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 76.0,
|
| 1474 |
+
"eval_accuracy": 0.525902107895677,
|
| 1475 |
+
"eval_f1_macro": 0.8001400823935392,
|
| 1476 |
+
"eval_f1_micro": 0.8323506640462152,
|
| 1477 |
+
"eval_loss": 0.128558948636055,
|
| 1478 |
+
"eval_roc_auc": 0.8947806580875474,
|
| 1479 |
+
"eval_runtime": 677.6714,
|
| 1480 |
+
"eval_samples_per_second": 4.13,
|
| 1481 |
+
"eval_steps_per_second": 0.258,
|
| 1482 |
+
"learning_rate": 1e-05,
|
| 1483 |
+
"step": 40736
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 76.49,
|
| 1487 |
+
"learning_rate": 1e-05,
|
| 1488 |
+
"loss": 0.061,
|
| 1489 |
+
"step": 41000
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 77.0,
|
| 1493 |
+
"eval_accuracy": 0.5280457306180779,
|
| 1494 |
+
"eval_f1_macro": 0.7994288724930265,
|
| 1495 |
+
"eval_f1_micro": 0.8320391828933223,
|
| 1496 |
+
"eval_loss": 0.12870018184185028,
|
| 1497 |
+
"eval_roc_auc": 0.8937004297813388,
|
| 1498 |
+
"eval_runtime": 678.86,
|
| 1499 |
+
"eval_samples_per_second": 4.123,
|
| 1500 |
+
"eval_steps_per_second": 0.258,
|
| 1501 |
+
"learning_rate": 1e-05,
|
| 1502 |
+
"step": 41272
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 77.43,
|
| 1506 |
+
"learning_rate": 1e-05,
|
| 1507 |
+
"loss": 0.0603,
|
| 1508 |
+
"step": 41500
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 78.0,
|
| 1512 |
+
"eval_accuracy": 0.5276884601643445,
|
| 1513 |
+
"eval_f1_macro": 0.7996470258117457,
|
| 1514 |
+
"eval_f1_micro": 0.8322565206987317,
|
| 1515 |
+
"eval_loss": 0.1287216991186142,
|
| 1516 |
+
"eval_roc_auc": 0.8933457314989977,
|
| 1517 |
+
"eval_runtime": 689.8146,
|
| 1518 |
+
"eval_samples_per_second": 4.058,
|
| 1519 |
+
"eval_steps_per_second": 0.254,
|
| 1520 |
+
"learning_rate": 1e-05,
|
| 1521 |
+
"step": 41808
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 78.36,
|
| 1525 |
+
"learning_rate": 1e-05,
|
| 1526 |
+
"loss": 0.0616,
|
| 1527 |
+
"step": 42000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 79.0,
|
| 1531 |
+
"eval_accuracy": 0.5269739192568774,
|
| 1532 |
+
"eval_f1_macro": 0.7993654831116606,
|
| 1533 |
+
"eval_f1_micro": 0.8321778627779107,
|
| 1534 |
+
"eval_loss": 0.12864746153354645,
|
| 1535 |
+
"eval_roc_auc": 0.8935954793497476,
|
| 1536 |
+
"eval_runtime": 681.1334,
|
| 1537 |
+
"eval_samples_per_second": 4.109,
|
| 1538 |
+
"eval_steps_per_second": 0.257,
|
| 1539 |
+
"learning_rate": 1e-05,
|
| 1540 |
+
"step": 42344
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 79.29,
|
| 1544 |
+
"learning_rate": 1e-05,
|
| 1545 |
+
"loss": 0.061,
|
| 1546 |
+
"step": 42500
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 80.0,
|
| 1550 |
+
"eval_accuracy": 0.5280457306180779,
|
| 1551 |
+
"eval_f1_macro": 0.7986609577778621,
|
| 1552 |
+
"eval_f1_micro": 0.831889081455806,
|
| 1553 |
+
"eval_loss": 0.12863147258758545,
|
| 1554 |
+
"eval_roc_auc": 0.8934491390025154,
|
| 1555 |
+
"eval_runtime": 713.1545,
|
| 1556 |
+
"eval_samples_per_second": 3.925,
|
| 1557 |
+
"eval_steps_per_second": 0.245,
|
| 1558 |
+
"learning_rate": 1e-05,
|
| 1559 |
+
"step": 42880
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 80.22,
|
| 1563 |
+
"learning_rate": 1e-05,
|
| 1564 |
+
"loss": 0.0607,
|
| 1565 |
+
"step": 43000
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 81.0,
|
| 1569 |
+
"eval_accuracy": 0.5280457306180779,
|
| 1570 |
+
"eval_f1_macro": 0.8002720191845476,
|
| 1571 |
+
"eval_f1_micro": 0.8328151160009567,
|
| 1572 |
+
"eval_loss": 0.12868718802928925,
|
| 1573 |
+
"eval_roc_auc": 0.8938115519252234,
|
| 1574 |
+
"eval_runtime": 691.4261,
|
| 1575 |
+
"eval_samples_per_second": 4.048,
|
| 1576 |
+
"eval_steps_per_second": 0.253,
|
| 1577 |
+
"learning_rate": 1e-05,
|
| 1578 |
+
"step": 43416
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 81.16,
|
| 1582 |
+
"learning_rate": 1e-05,
|
| 1583 |
+
"loss": 0.0609,
|
| 1584 |
+
"step": 43500
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 82.0,
|
| 1588 |
+
"eval_accuracy": 0.5287602715255448,
|
| 1589 |
+
"eval_f1_macro": 0.7990970221139213,
|
| 1590 |
+
"eval_f1_micro": 0.8321377331420372,
|
| 1591 |
+
"eval_loss": 0.12876012921333313,
|
| 1592 |
+
"eval_roc_auc": 0.893498947736975,
|
| 1593 |
+
"eval_runtime": 711.2784,
|
| 1594 |
+
"eval_samples_per_second": 3.935,
|
| 1595 |
+
"eval_steps_per_second": 0.246,
|
| 1596 |
+
"learning_rate": 1e-05,
|
| 1597 |
+
"step": 43952
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 82.09,
|
| 1601 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1602 |
+
"loss": 0.0611,
|
| 1603 |
+
"step": 44000
|
| 1604 |
+
},
|
| 1605 |
+
{
|
| 1606 |
+
"epoch": 83.0,
|
| 1607 |
+
"eval_accuracy": 0.5287602715255448,
|
| 1608 |
+
"eval_f1_macro": 0.7994362993568213,
|
| 1609 |
+
"eval_f1_micro": 0.8323968918111178,
|
| 1610 |
+
"eval_loss": 0.12867788970470428,
|
| 1611 |
+
"eval_roc_auc": 0.8936835921437016,
|
| 1612 |
+
"eval_runtime": 662.5317,
|
| 1613 |
+
"eval_samples_per_second": 4.225,
|
| 1614 |
+
"eval_steps_per_second": 0.264,
|
| 1615 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1616 |
+
"step": 44488
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 83.02,
|
| 1620 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1621 |
+
"loss": 0.0606,
|
| 1622 |
+
"step": 44500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 83.96,
|
| 1626 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1627 |
+
"loss": 0.0611,
|
| 1628 |
+
"step": 45000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 84.0,
|
| 1632 |
+
"eval_accuracy": 0.5284030010718114,
|
| 1633 |
+
"eval_f1_macro": 0.7992778813134658,
|
| 1634 |
+
"eval_f1_micro": 0.8324563501554653,
|
| 1635 |
+
"eval_loss": 0.12863513827323914,
|
| 1636 |
+
"eval_roc_auc": 0.8936069840247131,
|
| 1637 |
+
"eval_runtime": 678.6961,
|
| 1638 |
+
"eval_samples_per_second": 4.124,
|
| 1639 |
+
"eval_steps_per_second": 0.258,
|
| 1640 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1641 |
+
"step": 45024
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 84.89,
|
| 1645 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1646 |
+
"loss": 0.0608,
|
| 1647 |
+
"step": 45500
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 85.0,
|
| 1651 |
+
"eval_accuracy": 0.5287602715255448,
|
| 1652 |
+
"eval_f1_macro": 0.7992228177009172,
|
| 1653 |
+
"eval_f1_micro": 0.8324162679425838,
|
| 1654 |
+
"eval_loss": 0.12861298024654388,
|
| 1655 |
+
"eval_roc_auc": 0.8935104524119406,
|
| 1656 |
+
"eval_runtime": 655.8213,
|
| 1657 |
+
"eval_samples_per_second": 4.268,
|
| 1658 |
+
"eval_steps_per_second": 0.267,
|
| 1659 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1660 |
+
"step": 45560
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 85.82,
|
| 1664 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1665 |
+
"loss": 0.0607,
|
| 1666 |
+
"step": 46000
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 86.0,
|
| 1670 |
+
"eval_accuracy": 0.5284030010718114,
|
| 1671 |
+
"eval_f1_macro": 0.7994679642402115,
|
| 1672 |
+
"eval_f1_micro": 0.8323768531802965,
|
| 1673 |
+
"eval_loss": 0.12859971821308136,
|
| 1674 |
+
"eval_roc_auc": 0.8936353263373153,
|
| 1675 |
+
"eval_runtime": 659.2519,
|
| 1676 |
+
"eval_samples_per_second": 4.246,
|
| 1677 |
+
"eval_steps_per_second": 0.265,
|
| 1678 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1679 |
+
"step": 46096
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 86.0,
|
| 1683 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1684 |
+
"step": 46096,
|
| 1685 |
+
"total_flos": 5.764057623662956e+19,
|
| 1686 |
+
"train_loss": 0.1030069793400273,
|
| 1687 |
+
"train_runtime": 233299.8908,
|
| 1688 |
+
"train_samples_per_second": 3.305,
|
| 1689 |
+
"train_steps_per_second": 0.207
|
| 1690 |
+
}
|
| 1691 |
+
],
|
| 1692 |
+
"logging_steps": 500,
|
| 1693 |
+
"max_steps": 48240,
|
| 1694 |
+
"num_train_epochs": 90,
|
| 1695 |
+
"save_steps": 500,
|
| 1696 |
+
"total_flos": 5.764057623662956e+19,
|
| 1697 |
+
"trial_name": null,
|
| 1698 |
+
"trial_params": null
|
| 1699 |
+
}
|