Commit
•
b973fd7
1
Parent(s):
22a2cf4
🍻 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 +1167 -0
README.md
CHANGED
|
@@ -1,7 +1,11 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: facebook/dinov2-large
|
| 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 |
# dinov2-large-2024_01_05-kornia_img-size518_batch-size32_epochs70_freeze
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) 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: 0.0001
|
| 26 |
|
| 27 |
## Model description
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- eng
|
| 4 |
license: apache-2.0
|
| 5 |
base_model: facebook/dinov2-large
|
| 6 |
tags:
|
| 7 |
+
- multilabel-image-classification
|
| 8 |
+
- multilabel
|
| 9 |
- generated_from_trainer
|
| 10 |
metrics:
|
| 11 |
- accuracy
|
|
|
|
| 19 |
|
| 20 |
# dinov2-large-2024_01_05-kornia_img-size518_batch-size32_epochs70_freeze
|
| 21 |
|
| 22 |
+
This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) on the multilabel_complete_dataset dataset.
|
| 23 |
It achieves the following results on the evaluation set:
|
| 24 |
+
- Loss: 0.0840
|
| 25 |
+
- F1 Micro: 0.8543
|
| 26 |
+
- F1 Macro: 0.7343
|
| 27 |
+
- Roc Auc: 0.9077
|
| 28 |
+
- Accuracy: 0.5606
|
| 29 |
- Learning Rate: 0.0001
|
| 30 |
|
| 31 |
## Model description
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 70.0,
|
| 3 |
+
"eval_accuracy": 0.5605742296918768,
|
| 4 |
+
"eval_f1_macro": 0.7342630546801885,
|
| 5 |
+
"eval_f1_micro": 0.8543162417321499,
|
| 6 |
+
"eval_loss": 0.08401281386613846,
|
| 7 |
+
"eval_roc_auc": 0.9076857807628663,
|
| 8 |
+
"eval_runtime": 670.4543,
|
| 9 |
+
"eval_samples_per_second": 4.26,
|
| 10 |
+
"eval_steps_per_second": 0.134,
|
| 11 |
+
"learning_rate": 0.0001,
|
| 12 |
+
"train_loss": 0.11672632308896316,
|
| 13 |
+
"train_runtime": 200748.2354,
|
| 14 |
+
"train_samples_per_second": 3.057,
|
| 15 |
+
"train_steps_per_second": 0.096
|
| 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-large2024_01_05",
|
| 3 |
"apply_layernorm": true,
|
| 4 |
"architectures": [
|
| 5 |
"NewheadDinov2ForImageClassification"
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 70.0,
|
| 3 |
+
"eval_accuracy": 0.5605742296918768,
|
| 4 |
+
"eval_f1_macro": 0.7342630546801885,
|
| 5 |
+
"eval_f1_micro": 0.8543162417321499,
|
| 6 |
+
"eval_loss": 0.08401281386613846,
|
| 7 |
+
"eval_roc_auc": 0.9076857807628663,
|
| 8 |
+
"eval_runtime": 670.4543,
|
| 9 |
+
"eval_samples_per_second": 4.26,
|
| 10 |
+
"eval_steps_per_second": 0.134,
|
| 11 |
+
"learning_rate": 0.0001
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 70.0,
|
| 3 |
+
"learning_rate": 0.0001,
|
| 4 |
+
"train_loss": 0.11672632308896316,
|
| 5 |
+
"train_runtime": 200748.2354,
|
| 6 |
+
"train_samples_per_second": 3.057,
|
| 7 |
+
"train_steps_per_second": 0.096
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1167 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.08306006342172623,
|
| 3 |
+
"best_model_checkpoint": "/home1/datawork/mcontini/models/multilabel/huggingface/dinov2-large-2024_01_05-kornia_img-size518_batch-size32_epochs70_freeze/checkpoint-19180",
|
| 4 |
+
"epoch": 70.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 19180,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_accuracy": 0.44556873691556176,
|
| 14 |
+
"eval_f1_macro": 0.5755948244574681,
|
| 15 |
+
"eval_f1_micro": 0.7376394107473012,
|
| 16 |
+
"eval_loss": 0.13577787578105927,
|
| 17 |
+
"eval_roc_auc": 0.8276012534175776,
|
| 18 |
+
"eval_runtime": 686.0469,
|
| 19 |
+
"eval_samples_per_second": 4.178,
|
| 20 |
+
"eval_steps_per_second": 0.131,
|
| 21 |
+
"learning_rate": 0.01,
|
| 22 |
+
"step": 274
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 1.82,
|
| 26 |
+
"learning_rate": 0.01,
|
| 27 |
+
"loss": 0.1895,
|
| 28 |
+
"step": 500
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 2.0,
|
| 32 |
+
"eval_accuracy": 0.4357990230286113,
|
| 33 |
+
"eval_f1_macro": 0.6131029690652663,
|
| 34 |
+
"eval_f1_micro": 0.7463369963369964,
|
| 35 |
+
"eval_loss": 0.14224015176296234,
|
| 36 |
+
"eval_roc_auc": 0.8432701847378548,
|
| 37 |
+
"eval_runtime": 679.513,
|
| 38 |
+
"eval_samples_per_second": 4.218,
|
| 39 |
+
"eval_steps_per_second": 0.132,
|
| 40 |
+
"learning_rate": 0.01,
|
| 41 |
+
"step": 548
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 3.0,
|
| 45 |
+
"eval_accuracy": 0.38415910676901605,
|
| 46 |
+
"eval_f1_macro": 0.5242425898328716,
|
| 47 |
+
"eval_f1_micro": 0.7273147345925026,
|
| 48 |
+
"eval_loss": 0.21337130665779114,
|
| 49 |
+
"eval_roc_auc": 0.8305454415797603,
|
| 50 |
+
"eval_runtime": 681.6164,
|
| 51 |
+
"eval_samples_per_second": 4.205,
|
| 52 |
+
"eval_steps_per_second": 0.132,
|
| 53 |
+
"learning_rate": 0.01,
|
| 54 |
+
"step": 822
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 3.65,
|
| 58 |
+
"learning_rate": 0.01,
|
| 59 |
+
"loss": 0.1668,
|
| 60 |
+
"step": 1000
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"eval_accuracy": 0.4438241451500349,
|
| 65 |
+
"eval_f1_macro": 0.5474301561230492,
|
| 66 |
+
"eval_f1_micro": 0.7034210860994532,
|
| 67 |
+
"eval_loss": 0.14501234889030457,
|
| 68 |
+
"eval_roc_auc": 0.7947377699358407,
|
| 69 |
+
"eval_runtime": 676.7895,
|
| 70 |
+
"eval_samples_per_second": 4.235,
|
| 71 |
+
"eval_steps_per_second": 0.133,
|
| 72 |
+
"learning_rate": 0.01,
|
| 73 |
+
"step": 1096
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 5.0,
|
| 77 |
+
"eval_accuracy": 0.4438241451500349,
|
| 78 |
+
"eval_f1_macro": 0.6194844495540315,
|
| 79 |
+
"eval_f1_micro": 0.7611423380457615,
|
| 80 |
+
"eval_loss": 0.13293854892253876,
|
| 81 |
+
"eval_roc_auc": 0.8535844928345971,
|
| 82 |
+
"eval_runtime": 676.1205,
|
| 83 |
+
"eval_samples_per_second": 4.239,
|
| 84 |
+
"eval_steps_per_second": 0.133,
|
| 85 |
+
"learning_rate": 0.01,
|
| 86 |
+
"step": 1370
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 5.47,
|
| 90 |
+
"learning_rate": 0.01,
|
| 91 |
+
"loss": 0.1666,
|
| 92 |
+
"step": 1500
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 6.0,
|
| 96 |
+
"eval_accuracy": 0.44452198185624564,
|
| 97 |
+
"eval_f1_macro": 0.5624987041776927,
|
| 98 |
+
"eval_f1_micro": 0.752847713067352,
|
| 99 |
+
"eval_loss": 0.13243332505226135,
|
| 100 |
+
"eval_roc_auc": 0.8411441150969292,
|
| 101 |
+
"eval_runtime": 676.3907,
|
| 102 |
+
"eval_samples_per_second": 4.237,
|
| 103 |
+
"eval_steps_per_second": 0.133,
|
| 104 |
+
"learning_rate": 0.01,
|
| 105 |
+
"step": 1644
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 7.0,
|
| 109 |
+
"eval_accuracy": 0.43126308443824146,
|
| 110 |
+
"eval_f1_macro": 0.5689553622505643,
|
| 111 |
+
"eval_f1_micro": 0.7496488764044945,
|
| 112 |
+
"eval_loss": 0.13447266817092896,
|
| 113 |
+
"eval_roc_auc": 0.838952950800037,
|
| 114 |
+
"eval_runtime": 675.7851,
|
| 115 |
+
"eval_samples_per_second": 4.241,
|
| 116 |
+
"eval_steps_per_second": 0.133,
|
| 117 |
+
"learning_rate": 0.01,
|
| 118 |
+
"step": 1918
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.3,
|
| 122 |
+
"learning_rate": 0.01,
|
| 123 |
+
"loss": 0.1664,
|
| 124 |
+
"step": 2000
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 8.0,
|
| 128 |
+
"eval_accuracy": 0.4323098394975576,
|
| 129 |
+
"eval_f1_macro": 0.5627920395195278,
|
| 130 |
+
"eval_f1_micro": 0.7502482911725186,
|
| 131 |
+
"eval_loss": 0.13808754086494446,
|
| 132 |
+
"eval_roc_auc": 0.8397325865953646,
|
| 133 |
+
"eval_runtime": 681.8119,
|
| 134 |
+
"eval_samples_per_second": 4.204,
|
| 135 |
+
"eval_steps_per_second": 0.132,
|
| 136 |
+
"learning_rate": 0.01,
|
| 137 |
+
"step": 2192
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 9.0,
|
| 141 |
+
"eval_accuracy": 0.44033496161898117,
|
| 142 |
+
"eval_f1_macro": 0.5492251158735639,
|
| 143 |
+
"eval_f1_micro": 0.7395667604944316,
|
| 144 |
+
"eval_loss": 0.13694943487644196,
|
| 145 |
+
"eval_roc_auc": 0.8219722316265465,
|
| 146 |
+
"eval_runtime": 674.9771,
|
| 147 |
+
"eval_samples_per_second": 4.246,
|
| 148 |
+
"eval_steps_per_second": 0.133,
|
| 149 |
+
"learning_rate": 0.01,
|
| 150 |
+
"step": 2466
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 9.12,
|
| 154 |
+
"learning_rate": 0.01,
|
| 155 |
+
"loss": 0.1656,
|
| 156 |
+
"step": 2500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 10.0,
|
| 160 |
+
"eval_accuracy": 0.4424284717376134,
|
| 161 |
+
"eval_f1_macro": 0.528180519175626,
|
| 162 |
+
"eval_f1_micro": 0.7326552851434799,
|
| 163 |
+
"eval_loss": 0.13609227538108826,
|
| 164 |
+
"eval_roc_auc": 0.821178691278072,
|
| 165 |
+
"eval_runtime": 674.2621,
|
| 166 |
+
"eval_samples_per_second": 4.251,
|
| 167 |
+
"eval_steps_per_second": 0.133,
|
| 168 |
+
"learning_rate": 0.01,
|
| 169 |
+
"step": 2740
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.95,
|
| 173 |
+
"learning_rate": 0.01,
|
| 174 |
+
"loss": 0.166,
|
| 175 |
+
"step": 3000
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 11.0,
|
| 179 |
+
"eval_accuracy": 0.4277739009071877,
|
| 180 |
+
"eval_f1_macro": 0.5428317486981787,
|
| 181 |
+
"eval_f1_micro": 0.7434225844004656,
|
| 182 |
+
"eval_loss": 0.1380929797887802,
|
| 183 |
+
"eval_roc_auc": 0.8371006883503846,
|
| 184 |
+
"eval_runtime": 676.158,
|
| 185 |
+
"eval_samples_per_second": 4.239,
|
| 186 |
+
"eval_steps_per_second": 0.133,
|
| 187 |
+
"learning_rate": 0.01,
|
| 188 |
+
"step": 3014
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 12.0,
|
| 192 |
+
"eval_accuracy": 0.444870900209351,
|
| 193 |
+
"eval_f1_macro": 0.5618568055480317,
|
| 194 |
+
"eval_f1_micro": 0.7354685646500594,
|
| 195 |
+
"eval_loss": 0.1344844251871109,
|
| 196 |
+
"eval_roc_auc": 0.827911942291835,
|
| 197 |
+
"eval_runtime": 683.7162,
|
| 198 |
+
"eval_samples_per_second": 4.192,
|
| 199 |
+
"eval_steps_per_second": 0.132,
|
| 200 |
+
"learning_rate": 0.01,
|
| 201 |
+
"step": 3288
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 12.77,
|
| 205 |
+
"learning_rate": 0.001,
|
| 206 |
+
"loss": 0.1585,
|
| 207 |
+
"step": 3500
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 13.0,
|
| 211 |
+
"eval_accuracy": 0.49023028611304953,
|
| 212 |
+
"eval_f1_macro": 0.650060261894195,
|
| 213 |
+
"eval_f1_micro": 0.8009333029820168,
|
| 214 |
+
"eval_loss": 0.11546628922224045,
|
| 215 |
+
"eval_roc_auc": 0.8745955707824836,
|
| 216 |
+
"eval_runtime": 679.0864,
|
| 217 |
+
"eval_samples_per_second": 4.22,
|
| 218 |
+
"eval_steps_per_second": 0.133,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"step": 3562
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 14.0,
|
| 224 |
+
"eval_accuracy": 0.5041870202372645,
|
| 225 |
+
"eval_f1_macro": 0.6697333276095331,
|
| 226 |
+
"eval_f1_micro": 0.8079378774805867,
|
| 227 |
+
"eval_loss": 0.11155486851930618,
|
| 228 |
+
"eval_roc_auc": 0.8750976636196655,
|
| 229 |
+
"eval_runtime": 676.6891,
|
| 230 |
+
"eval_samples_per_second": 4.235,
|
| 231 |
+
"eval_steps_per_second": 0.133,
|
| 232 |
+
"learning_rate": 0.001,
|
| 233 |
+
"step": 3836
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 14.6,
|
| 237 |
+
"learning_rate": 0.001,
|
| 238 |
+
"loss": 0.133,
|
| 239 |
+
"step": 4000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 15.0,
|
| 243 |
+
"eval_accuracy": 0.5181437543614794,
|
| 244 |
+
"eval_f1_macro": 0.6736053030113935,
|
| 245 |
+
"eval_f1_micro": 0.814943326393708,
|
| 246 |
+
"eval_loss": 0.10734836012125015,
|
| 247 |
+
"eval_roc_auc": 0.877205285207194,
|
| 248 |
+
"eval_runtime": 677.4924,
|
| 249 |
+
"eval_samples_per_second": 4.23,
|
| 250 |
+
"eval_steps_per_second": 0.133,
|
| 251 |
+
"learning_rate": 0.001,
|
| 252 |
+
"step": 4110
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 16.0,
|
| 256 |
+
"eval_accuracy": 0.5083740404745289,
|
| 257 |
+
"eval_f1_macro": 0.7055616874566738,
|
| 258 |
+
"eval_f1_micro": 0.8238276299112801,
|
| 259 |
+
"eval_loss": 0.10476414114236832,
|
| 260 |
+
"eval_roc_auc": 0.8975418625131631,
|
| 261 |
+
"eval_runtime": 687.0217,
|
| 262 |
+
"eval_samples_per_second": 4.172,
|
| 263 |
+
"eval_steps_per_second": 0.131,
|
| 264 |
+
"learning_rate": 0.001,
|
| 265 |
+
"step": 4384
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 16.42,
|
| 269 |
+
"learning_rate": 0.001,
|
| 270 |
+
"loss": 0.1289,
|
| 271 |
+
"step": 4500
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 17.0,
|
| 275 |
+
"eval_accuracy": 0.5244242847173761,
|
| 276 |
+
"eval_f1_macro": 0.6896485755961196,
|
| 277 |
+
"eval_f1_micro": 0.820858825547487,
|
| 278 |
+
"eval_loss": 0.10253454744815826,
|
| 279 |
+
"eval_roc_auc": 0.8839468587595108,
|
| 280 |
+
"eval_runtime": 684.6037,
|
| 281 |
+
"eval_samples_per_second": 4.186,
|
| 282 |
+
"eval_steps_per_second": 0.131,
|
| 283 |
+
"learning_rate": 0.001,
|
| 284 |
+
"step": 4658
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 18.0,
|
| 288 |
+
"eval_accuracy": 0.5321004884856944,
|
| 289 |
+
"eval_f1_macro": 0.7045003592264228,
|
| 290 |
+
"eval_f1_micro": 0.8289563051845145,
|
| 291 |
+
"eval_loss": 0.10259302705526352,
|
| 292 |
+
"eval_roc_auc": 0.8916264271206406,
|
| 293 |
+
"eval_runtime": 685.7262,
|
| 294 |
+
"eval_samples_per_second": 4.18,
|
| 295 |
+
"eval_steps_per_second": 0.131,
|
| 296 |
+
"learning_rate": 0.001,
|
| 297 |
+
"step": 4932
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 18.25,
|
| 301 |
+
"learning_rate": 0.001,
|
| 302 |
+
"loss": 0.1227,
|
| 303 |
+
"step": 5000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 19.0,
|
| 307 |
+
"eval_accuracy": 0.5279134682484299,
|
| 308 |
+
"eval_f1_macro": 0.6905367219275804,
|
| 309 |
+
"eval_f1_micro": 0.8306010928961749,
|
| 310 |
+
"eval_loss": 0.10123815387487411,
|
| 311 |
+
"eval_roc_auc": 0.8940566516497492,
|
| 312 |
+
"eval_runtime": 685.5377,
|
| 313 |
+
"eval_samples_per_second": 4.181,
|
| 314 |
+
"eval_steps_per_second": 0.131,
|
| 315 |
+
"learning_rate": 0.001,
|
| 316 |
+
"step": 5206
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 20.0,
|
| 320 |
+
"eval_accuracy": 0.5216329378925332,
|
| 321 |
+
"eval_f1_macro": 0.6830881274898382,
|
| 322 |
+
"eval_f1_micro": 0.8280441143371596,
|
| 323 |
+
"eval_loss": 0.09970748424530029,
|
| 324 |
+
"eval_roc_auc": 0.8930346669934526,
|
| 325 |
+
"eval_runtime": 686.9199,
|
| 326 |
+
"eval_samples_per_second": 4.172,
|
| 327 |
+
"eval_steps_per_second": 0.131,
|
| 328 |
+
"learning_rate": 0.001,
|
| 329 |
+
"step": 5480
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 20.07,
|
| 333 |
+
"learning_rate": 0.001,
|
| 334 |
+
"loss": 0.1202,
|
| 335 |
+
"step": 5500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 21.0,
|
| 339 |
+
"eval_accuracy": 0.5352407536636427,
|
| 340 |
+
"eval_f1_macro": 0.6926783323821563,
|
| 341 |
+
"eval_f1_micro": 0.8300336623495178,
|
| 342 |
+
"eval_loss": 0.09886988252401352,
|
| 343 |
+
"eval_roc_auc": 0.8896211857387517,
|
| 344 |
+
"eval_runtime": 684.9162,
|
| 345 |
+
"eval_samples_per_second": 4.184,
|
| 346 |
+
"eval_steps_per_second": 0.131,
|
| 347 |
+
"learning_rate": 0.001,
|
| 348 |
+
"step": 5754
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 21.9,
|
| 352 |
+
"learning_rate": 0.001,
|
| 353 |
+
"loss": 0.12,
|
| 354 |
+
"step": 6000
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 22.0,
|
| 358 |
+
"eval_accuracy": 0.5209351011863224,
|
| 359 |
+
"eval_f1_macro": 0.6961228606859606,
|
| 360 |
+
"eval_f1_micro": 0.8279826958105646,
|
| 361 |
+
"eval_loss": 0.09963646531105042,
|
| 362 |
+
"eval_roc_auc": 0.8892587586568824,
|
| 363 |
+
"eval_runtime": 686.1389,
|
| 364 |
+
"eval_samples_per_second": 4.177,
|
| 365 |
+
"eval_steps_per_second": 0.131,
|
| 366 |
+
"learning_rate": 0.001,
|
| 367 |
+
"step": 6028
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 23.0,
|
| 371 |
+
"eval_accuracy": 0.5195394277739009,
|
| 372 |
+
"eval_f1_macro": 0.6958628426894405,
|
| 373 |
+
"eval_f1_micro": 0.831919078392807,
|
| 374 |
+
"eval_loss": 0.09720779210329056,
|
| 375 |
+
"eval_roc_auc": 0.8955944894582717,
|
| 376 |
+
"eval_runtime": 693.3459,
|
| 377 |
+
"eval_samples_per_second": 4.134,
|
| 378 |
+
"eval_steps_per_second": 0.13,
|
| 379 |
+
"learning_rate": 0.001,
|
| 380 |
+
"step": 6302
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 23.72,
|
| 384 |
+
"learning_rate": 0.001,
|
| 385 |
+
"loss": 0.1179,
|
| 386 |
+
"step": 6500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 24.0,
|
| 390 |
+
"eval_accuracy": 0.5212840195394278,
|
| 391 |
+
"eval_f1_macro": 0.6881053152313114,
|
| 392 |
+
"eval_f1_micro": 0.8270608813406306,
|
| 393 |
+
"eval_loss": 0.10082241147756577,
|
| 394 |
+
"eval_roc_auc": 0.8915954736236973,
|
| 395 |
+
"eval_runtime": 682.9894,
|
| 396 |
+
"eval_samples_per_second": 4.196,
|
| 397 |
+
"eval_steps_per_second": 0.132,
|
| 398 |
+
"learning_rate": 0.001,
|
| 399 |
+
"step": 6576
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 25.0,
|
| 403 |
+
"eval_accuracy": 0.5268667131891137,
|
| 404 |
+
"eval_f1_macro": 0.6859679989625925,
|
| 405 |
+
"eval_f1_micro": 0.8283316086006668,
|
| 406 |
+
"eval_loss": 0.09828384965658188,
|
| 407 |
+
"eval_roc_auc": 0.8862551588199984,
|
| 408 |
+
"eval_runtime": 673.7378,
|
| 409 |
+
"eval_samples_per_second": 4.254,
|
| 410 |
+
"eval_steps_per_second": 0.134,
|
| 411 |
+
"learning_rate": 0.001,
|
| 412 |
+
"step": 6850
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 25.55,
|
| 416 |
+
"learning_rate": 0.001,
|
| 417 |
+
"loss": 0.1166,
|
| 418 |
+
"step": 7000
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 26.0,
|
| 422 |
+
"eval_accuracy": 0.5310537334263782,
|
| 423 |
+
"eval_f1_macro": 0.6805616825898689,
|
| 424 |
+
"eval_f1_micro": 0.8284274424464553,
|
| 425 |
+
"eval_loss": 0.09853371977806091,
|
| 426 |
+
"eval_roc_auc": 0.8875551335725609,
|
| 427 |
+
"eval_runtime": 679.0226,
|
| 428 |
+
"eval_samples_per_second": 4.221,
|
| 429 |
+
"eval_steps_per_second": 0.133,
|
| 430 |
+
"learning_rate": 0.001,
|
| 431 |
+
"step": 7124
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 27.0,
|
| 435 |
+
"eval_accuracy": 0.5324494068387997,
|
| 436 |
+
"eval_f1_macro": 0.6901040821549612,
|
| 437 |
+
"eval_f1_micro": 0.8305464575073264,
|
| 438 |
+
"eval_loss": 0.09571811556816101,
|
| 439 |
+
"eval_roc_auc": 0.887615396252071,
|
| 440 |
+
"eval_runtime": 672.1908,
|
| 441 |
+
"eval_samples_per_second": 4.264,
|
| 442 |
+
"eval_steps_per_second": 0.134,
|
| 443 |
+
"learning_rate": 0.001,
|
| 444 |
+
"step": 7398
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 27.37,
|
| 448 |
+
"learning_rate": 0.001,
|
| 449 |
+
"loss": 0.1158,
|
| 450 |
+
"step": 7500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 28.0,
|
| 454 |
+
"eval_accuracy": 0.5177948360083741,
|
| 455 |
+
"eval_f1_macro": 0.7054421966314011,
|
| 456 |
+
"eval_f1_micro": 0.8292325882551658,
|
| 457 |
+
"eval_loss": 0.09954769909381866,
|
| 458 |
+
"eval_roc_auc": 0.8934873150088631,
|
| 459 |
+
"eval_runtime": 681.7881,
|
| 460 |
+
"eval_samples_per_second": 4.204,
|
| 461 |
+
"eval_steps_per_second": 0.132,
|
| 462 |
+
"learning_rate": 0.001,
|
| 463 |
+
"step": 7672
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 29.0,
|
| 467 |
+
"eval_accuracy": 0.5334961618981159,
|
| 468 |
+
"eval_f1_macro": 0.7026467347883069,
|
| 469 |
+
"eval_f1_micro": 0.8363861804655357,
|
| 470 |
+
"eval_loss": 0.09332505613565445,
|
| 471 |
+
"eval_roc_auc": 0.8970916521216963,
|
| 472 |
+
"eval_runtime": 684.6938,
|
| 473 |
+
"eval_samples_per_second": 4.186,
|
| 474 |
+
"eval_steps_per_second": 0.131,
|
| 475 |
+
"learning_rate": 0.001,
|
| 476 |
+
"step": 7946
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 29.2,
|
| 480 |
+
"learning_rate": 0.001,
|
| 481 |
+
"loss": 0.114,
|
| 482 |
+
"step": 8000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 30.0,
|
| 486 |
+
"eval_accuracy": 0.5258199581297976,
|
| 487 |
+
"eval_f1_macro": 0.7109768073155117,
|
| 488 |
+
"eval_f1_micro": 0.8351258454374099,
|
| 489 |
+
"eval_loss": 0.09473367780447006,
|
| 490 |
+
"eval_roc_auc": 0.901874991489952,
|
| 491 |
+
"eval_runtime": 677.918,
|
| 492 |
+
"eval_samples_per_second": 4.228,
|
| 493 |
+
"eval_steps_per_second": 0.133,
|
| 494 |
+
"learning_rate": 0.001,
|
| 495 |
+
"step": 8220
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 31.0,
|
| 499 |
+
"eval_accuracy": 0.5331472435450104,
|
| 500 |
+
"eval_f1_macro": 0.7175382540523837,
|
| 501 |
+
"eval_f1_micro": 0.8365119611950171,
|
| 502 |
+
"eval_loss": 0.09674925357103348,
|
| 503 |
+
"eval_roc_auc": 0.9045853985344947,
|
| 504 |
+
"eval_runtime": 675.0971,
|
| 505 |
+
"eval_samples_per_second": 4.245,
|
| 506 |
+
"eval_steps_per_second": 0.133,
|
| 507 |
+
"learning_rate": 0.001,
|
| 508 |
+
"step": 8494
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 31.02,
|
| 512 |
+
"learning_rate": 0.001,
|
| 513 |
+
"loss": 0.1134,
|
| 514 |
+
"step": 8500
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 32.0,
|
| 518 |
+
"eval_accuracy": 0.5324494068387997,
|
| 519 |
+
"eval_f1_macro": 0.6932594476375145,
|
| 520 |
+
"eval_f1_micro": 0.8353541076487252,
|
| 521 |
+
"eval_loss": 0.09490892291069031,
|
| 522 |
+
"eval_roc_auc": 0.8947967085095635,
|
| 523 |
+
"eval_runtime": 677.2028,
|
| 524 |
+
"eval_samples_per_second": 4.232,
|
| 525 |
+
"eval_steps_per_second": 0.133,
|
| 526 |
+
"learning_rate": 0.001,
|
| 527 |
+
"step": 8768
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 32.85,
|
| 531 |
+
"learning_rate": 0.001,
|
| 532 |
+
"loss": 0.113,
|
| 533 |
+
"step": 9000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 33.0,
|
| 537 |
+
"eval_accuracy": 0.5362875087229588,
|
| 538 |
+
"eval_f1_macro": 0.6973292248077614,
|
| 539 |
+
"eval_f1_micro": 0.8367208672086721,
|
| 540 |
+
"eval_loss": 0.09511947631835938,
|
| 541 |
+
"eval_roc_auc": 0.8966987186810037,
|
| 542 |
+
"eval_runtime": 679.2015,
|
| 543 |
+
"eval_samples_per_second": 4.22,
|
| 544 |
+
"eval_steps_per_second": 0.133,
|
| 545 |
+
"learning_rate": 0.001,
|
| 546 |
+
"step": 9042
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 34.0,
|
| 550 |
+
"eval_accuracy": 0.5380321004884857,
|
| 551 |
+
"eval_f1_macro": 0.6878227037845351,
|
| 552 |
+
"eval_f1_micro": 0.8334680679062246,
|
| 553 |
+
"eval_loss": 0.09364539384841919,
|
| 554 |
+
"eval_roc_auc": 0.8876181367760314,
|
| 555 |
+
"eval_runtime": 675.0346,
|
| 556 |
+
"eval_samples_per_second": 4.246,
|
| 557 |
+
"eval_steps_per_second": 0.133,
|
| 558 |
+
"learning_rate": 0.001,
|
| 559 |
+
"step": 9316
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 34.67,
|
| 563 |
+
"learning_rate": 0.001,
|
| 564 |
+
"loss": 0.1124,
|
| 565 |
+
"step": 9500
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 35.0,
|
| 569 |
+
"eval_accuracy": 0.5310537334263782,
|
| 570 |
+
"eval_f1_macro": 0.6856042645068489,
|
| 571 |
+
"eval_f1_micro": 0.833974649162517,
|
| 572 |
+
"eval_loss": 0.09358564764261246,
|
| 573 |
+
"eval_roc_auc": 0.8944494841237697,
|
| 574 |
+
"eval_runtime": 678.2782,
|
| 575 |
+
"eval_samples_per_second": 4.225,
|
| 576 |
+
"eval_steps_per_second": 0.133,
|
| 577 |
+
"learning_rate": 0.001,
|
| 578 |
+
"step": 9590
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 36.0,
|
| 582 |
+
"eval_accuracy": 0.5453593859036985,
|
| 583 |
+
"eval_f1_macro": 0.729828782855425,
|
| 584 |
+
"eval_f1_micro": 0.8455960879096174,
|
| 585 |
+
"eval_loss": 0.09342356771230698,
|
| 586 |
+
"eval_roc_auc": 0.9030647539078717,
|
| 587 |
+
"eval_runtime": 674.7512,
|
| 588 |
+
"eval_samples_per_second": 4.247,
|
| 589 |
+
"eval_steps_per_second": 0.133,
|
| 590 |
+
"learning_rate": 0.0001,
|
| 591 |
+
"step": 9864
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 36.5,
|
| 595 |
+
"learning_rate": 0.0001,
|
| 596 |
+
"loss": 0.1083,
|
| 597 |
+
"step": 10000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 37.0,
|
| 601 |
+
"eval_accuracy": 0.54675505931612,
|
| 602 |
+
"eval_f1_macro": 0.7188765655113909,
|
| 603 |
+
"eval_f1_micro": 0.8456650022696323,
|
| 604 |
+
"eval_loss": 0.09240464121103287,
|
| 605 |
+
"eval_roc_auc": 0.8999478550409371,
|
| 606 |
+
"eval_runtime": 674.676,
|
| 607 |
+
"eval_samples_per_second": 4.248,
|
| 608 |
+
"eval_steps_per_second": 0.133,
|
| 609 |
+
"learning_rate": 0.0001,
|
| 610 |
+
"step": 10138
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 38.0,
|
| 614 |
+
"eval_accuracy": 0.5450104675505931,
|
| 615 |
+
"eval_f1_macro": 0.7089159960142193,
|
| 616 |
+
"eval_f1_micro": 0.8449173647271904,
|
| 617 |
+
"eval_loss": 0.09147636592388153,
|
| 618 |
+
"eval_roc_auc": 0.9003695495594045,
|
| 619 |
+
"eval_runtime": 675.4085,
|
| 620 |
+
"eval_samples_per_second": 4.243,
|
| 621 |
+
"eval_steps_per_second": 0.133,
|
| 622 |
+
"learning_rate": 0.0001,
|
| 623 |
+
"step": 10412
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 38.32,
|
| 627 |
+
"learning_rate": 0.0001,
|
| 628 |
+
"loss": 0.1034,
|
| 629 |
+
"step": 10500
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 39.0,
|
| 633 |
+
"eval_accuracy": 0.5484996510816469,
|
| 634 |
+
"eval_f1_macro": 0.725215575661352,
|
| 635 |
+
"eval_f1_micro": 0.8487853799866281,
|
| 636 |
+
"eval_loss": 0.09022974222898483,
|
| 637 |
+
"eval_roc_auc": 0.9078051247451889,
|
| 638 |
+
"eval_runtime": 672.7306,
|
| 639 |
+
"eval_samples_per_second": 4.26,
|
| 640 |
+
"eval_steps_per_second": 0.134,
|
| 641 |
+
"learning_rate": 0.0001,
|
| 642 |
+
"step": 10686
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 40.0,
|
| 646 |
+
"eval_accuracy": 0.5495464061409631,
|
| 647 |
+
"eval_f1_macro": 0.7182446688615595,
|
| 648 |
+
"eval_f1_micro": 0.8458797579322437,
|
| 649 |
+
"eval_loss": 0.09058264642953873,
|
| 650 |
+
"eval_roc_auc": 0.9011424061638826,
|
| 651 |
+
"eval_runtime": 678.9298,
|
| 652 |
+
"eval_samples_per_second": 4.221,
|
| 653 |
+
"eval_steps_per_second": 0.133,
|
| 654 |
+
"learning_rate": 0.0001,
|
| 655 |
+
"step": 10960
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 40.15,
|
| 659 |
+
"learning_rate": 0.0001,
|
| 660 |
+
"loss": 0.1024,
|
| 661 |
+
"step": 11000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 41.0,
|
| 665 |
+
"eval_accuracy": 0.5505931612002791,
|
| 666 |
+
"eval_f1_macro": 0.7130026819185953,
|
| 667 |
+
"eval_f1_micro": 0.8481005491705826,
|
| 668 |
+
"eval_loss": 0.08943015336990356,
|
| 669 |
+
"eval_roc_auc": 0.902018393012137,
|
| 670 |
+
"eval_runtime": 676.4356,
|
| 671 |
+
"eval_samples_per_second": 4.237,
|
| 672 |
+
"eval_steps_per_second": 0.133,
|
| 673 |
+
"learning_rate": 0.0001,
|
| 674 |
+
"step": 11234
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 41.97,
|
| 678 |
+
"learning_rate": 0.0001,
|
| 679 |
+
"loss": 0.1004,
|
| 680 |
+
"step": 11500
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 42.0,
|
| 684 |
+
"eval_accuracy": 0.5519888346127007,
|
| 685 |
+
"eval_f1_macro": 0.7148190184347656,
|
| 686 |
+
"eval_f1_micro": 0.8457012282205084,
|
| 687 |
+
"eval_loss": 0.08726447820663452,
|
| 688 |
+
"eval_roc_auc": 0.8977194051943719,
|
| 689 |
+
"eval_runtime": 677.9496,
|
| 690 |
+
"eval_samples_per_second": 4.227,
|
| 691 |
+
"eval_steps_per_second": 0.133,
|
| 692 |
+
"learning_rate": 0.0001,
|
| 693 |
+
"step": 11508
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 43.0,
|
| 697 |
+
"eval_accuracy": 0.5537334263782275,
|
| 698 |
+
"eval_f1_macro": 0.71816969331258,
|
| 699 |
+
"eval_f1_micro": 0.8494563389754511,
|
| 700 |
+
"eval_loss": 0.08699071407318115,
|
| 701 |
+
"eval_roc_auc": 0.906163507621426,
|
| 702 |
+
"eval_runtime": 676.5268,
|
| 703 |
+
"eval_samples_per_second": 4.236,
|
| 704 |
+
"eval_steps_per_second": 0.133,
|
| 705 |
+
"learning_rate": 0.0001,
|
| 706 |
+
"step": 11782
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 43.8,
|
| 710 |
+
"learning_rate": 0.0001,
|
| 711 |
+
"loss": 0.0998,
|
| 712 |
+
"step": 12000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 44.0,
|
| 716 |
+
"eval_accuracy": 0.5498953244940684,
|
| 717 |
+
"eval_f1_macro": 0.7261208407998851,
|
| 718 |
+
"eval_f1_micro": 0.8486114247008355,
|
| 719 |
+
"eval_loss": 0.08676985651254654,
|
| 720 |
+
"eval_roc_auc": 0.9033042612782081,
|
| 721 |
+
"eval_runtime": 674.6536,
|
| 722 |
+
"eval_samples_per_second": 4.248,
|
| 723 |
+
"eval_steps_per_second": 0.133,
|
| 724 |
+
"learning_rate": 0.0001,
|
| 725 |
+
"step": 12056
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 45.0,
|
| 729 |
+
"eval_accuracy": 0.555129099790649,
|
| 730 |
+
"eval_f1_macro": 0.7235580263821535,
|
| 731 |
+
"eval_f1_micro": 0.8493258426966293,
|
| 732 |
+
"eval_loss": 0.08680889010429382,
|
| 733 |
+
"eval_roc_auc": 0.9052931027877648,
|
| 734 |
+
"eval_runtime": 688.2557,
|
| 735 |
+
"eval_samples_per_second": 4.164,
|
| 736 |
+
"eval_steps_per_second": 0.131,
|
| 737 |
+
"learning_rate": 0.0001,
|
| 738 |
+
"step": 12330
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 45.62,
|
| 742 |
+
"learning_rate": 0.0001,
|
| 743 |
+
"loss": 0.0975,
|
| 744 |
+
"step": 12500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 46.0,
|
| 748 |
+
"eval_accuracy": 0.5512909979064898,
|
| 749 |
+
"eval_f1_macro": 0.7317716716296281,
|
| 750 |
+
"eval_f1_micro": 0.8489586241554526,
|
| 751 |
+
"eval_loss": 0.0865492969751358,
|
| 752 |
+
"eval_roc_auc": 0.9071607698668371,
|
| 753 |
+
"eval_runtime": 684.0678,
|
| 754 |
+
"eval_samples_per_second": 4.19,
|
| 755 |
+
"eval_steps_per_second": 0.132,
|
| 756 |
+
"learning_rate": 0.0001,
|
| 757 |
+
"step": 12604
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 47.0,
|
| 761 |
+
"eval_accuracy": 0.5547801814375436,
|
| 762 |
+
"eval_f1_macro": 0.7390020274567815,
|
| 763 |
+
"eval_f1_micro": 0.8512299882858259,
|
| 764 |
+
"eval_loss": 0.08599700033664703,
|
| 765 |
+
"eval_roc_auc": 0.908765101440927,
|
| 766 |
+
"eval_runtime": 688.2365,
|
| 767 |
+
"eval_samples_per_second": 4.164,
|
| 768 |
+
"eval_steps_per_second": 0.131,
|
| 769 |
+
"learning_rate": 0.0001,
|
| 770 |
+
"step": 12878
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 47.45,
|
| 774 |
+
"learning_rate": 0.0001,
|
| 775 |
+
"loss": 0.099,
|
| 776 |
+
"step": 13000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 48.0,
|
| 780 |
+
"eval_accuracy": 0.5558269364968598,
|
| 781 |
+
"eval_f1_macro": 0.7360003523455093,
|
| 782 |
+
"eval_f1_micro": 0.8509512552065742,
|
| 783 |
+
"eval_loss": 0.08596429973840714,
|
| 784 |
+
"eval_roc_auc": 0.9055422308395834,
|
| 785 |
+
"eval_runtime": 686.3198,
|
| 786 |
+
"eval_samples_per_second": 4.176,
|
| 787 |
+
"eval_steps_per_second": 0.131,
|
| 788 |
+
"learning_rate": 0.0001,
|
| 789 |
+
"step": 13152
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 49.0,
|
| 793 |
+
"eval_accuracy": 0.5547801814375436,
|
| 794 |
+
"eval_f1_macro": 0.7361919298080869,
|
| 795 |
+
"eval_f1_micro": 0.849985959000281,
|
| 796 |
+
"eval_loss": 0.08584348857402802,
|
| 797 |
+
"eval_roc_auc": 0.9057525299940252,
|
| 798 |
+
"eval_runtime": 688.3813,
|
| 799 |
+
"eval_samples_per_second": 4.163,
|
| 800 |
+
"eval_steps_per_second": 0.131,
|
| 801 |
+
"learning_rate": 0.0001,
|
| 802 |
+
"step": 13426
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 49.27,
|
| 806 |
+
"learning_rate": 0.0001,
|
| 807 |
+
"loss": 0.0972,
|
| 808 |
+
"step": 13500
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 50.0,
|
| 812 |
+
"eval_accuracy": 0.5586182833217027,
|
| 813 |
+
"eval_f1_macro": 0.725712332481399,
|
| 814 |
+
"eval_f1_micro": 0.8505096262740656,
|
| 815 |
+
"eval_loss": 0.08557379245758057,
|
| 816 |
+
"eval_roc_auc": 0.9032805341102342,
|
| 817 |
+
"eval_runtime": 685.8179,
|
| 818 |
+
"eval_samples_per_second": 4.179,
|
| 819 |
+
"eval_steps_per_second": 0.131,
|
| 820 |
+
"learning_rate": 0.0001,
|
| 821 |
+
"step": 13700
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 51.0,
|
| 825 |
+
"eval_accuracy": 0.557920446615492,
|
| 826 |
+
"eval_f1_macro": 0.7408593608052999,
|
| 827 |
+
"eval_f1_micro": 0.8500254194204373,
|
| 828 |
+
"eval_loss": 0.08562461286783218,
|
| 829 |
+
"eval_roc_auc": 0.9038335718454608,
|
| 830 |
+
"eval_runtime": 683.3234,
|
| 831 |
+
"eval_samples_per_second": 4.194,
|
| 832 |
+
"eval_steps_per_second": 0.132,
|
| 833 |
+
"learning_rate": 0.0001,
|
| 834 |
+
"step": 13974
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 51.09,
|
| 838 |
+
"learning_rate": 0.0001,
|
| 839 |
+
"loss": 0.0957,
|
| 840 |
+
"step": 14000
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 52.0,
|
| 844 |
+
"eval_accuracy": 0.5568736915561758,
|
| 845 |
+
"eval_f1_macro": 0.7232142709265429,
|
| 846 |
+
"eval_f1_micro": 0.8507868221442318,
|
| 847 |
+
"eval_loss": 0.08591117709875107,
|
| 848 |
+
"eval_roc_auc": 0.9035466101391771,
|
| 849 |
+
"eval_runtime": 693.4248,
|
| 850 |
+
"eval_samples_per_second": 4.133,
|
| 851 |
+
"eval_steps_per_second": 0.13,
|
| 852 |
+
"learning_rate": 0.0001,
|
| 853 |
+
"step": 14248
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 52.92,
|
| 857 |
+
"learning_rate": 0.0001,
|
| 858 |
+
"loss": 0.0964,
|
| 859 |
+
"step": 14500
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 53.0,
|
| 863 |
+
"eval_accuracy": 0.5628053035589672,
|
| 864 |
+
"eval_f1_macro": 0.7275870481420489,
|
| 865 |
+
"eval_f1_micro": 0.852056338028169,
|
| 866 |
+
"eval_loss": 0.08490145951509476,
|
| 867 |
+
"eval_roc_auc": 0.9058454914515268,
|
| 868 |
+
"eval_runtime": 691.5127,
|
| 869 |
+
"eval_samples_per_second": 4.145,
|
| 870 |
+
"eval_steps_per_second": 0.13,
|
| 871 |
+
"learning_rate": 0.0001,
|
| 872 |
+
"step": 14522
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 54.0,
|
| 876 |
+
"eval_accuracy": 0.5537334263782275,
|
| 877 |
+
"eval_f1_macro": 0.7394514344990791,
|
| 878 |
+
"eval_f1_micro": 0.85390386218394,
|
| 879 |
+
"eval_loss": 0.08516541868448257,
|
| 880 |
+
"eval_roc_auc": 0.9115532672468961,
|
| 881 |
+
"eval_runtime": 698.0325,
|
| 882 |
+
"eval_samples_per_second": 4.106,
|
| 883 |
+
"eval_steps_per_second": 0.129,
|
| 884 |
+
"learning_rate": 0.0001,
|
| 885 |
+
"step": 14796
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 54.74,
|
| 889 |
+
"learning_rate": 0.0001,
|
| 890 |
+
"loss": 0.0955,
|
| 891 |
+
"step": 15000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 55.0,
|
| 895 |
+
"eval_accuracy": 0.5565247732030705,
|
| 896 |
+
"eval_f1_macro": 0.7354184764103003,
|
| 897 |
+
"eval_f1_micro": 0.8511167656205825,
|
| 898 |
+
"eval_loss": 0.08514942973852158,
|
| 899 |
+
"eval_roc_auc": 0.904089479088129,
|
| 900 |
+
"eval_runtime": 681.3519,
|
| 901 |
+
"eval_samples_per_second": 4.206,
|
| 902 |
+
"eval_steps_per_second": 0.132,
|
| 903 |
+
"learning_rate": 0.0001,
|
| 904 |
+
"step": 15070
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 56.0,
|
| 908 |
+
"eval_accuracy": 0.5572226099092812,
|
| 909 |
+
"eval_f1_macro": 0.736739641327092,
|
| 910 |
+
"eval_f1_micro": 0.8529461421576904,
|
| 911 |
+
"eval_loss": 0.08491206169128418,
|
| 912 |
+
"eval_roc_auc": 0.9066984002032717,
|
| 913 |
+
"eval_runtime": 677.9791,
|
| 914 |
+
"eval_samples_per_second": 4.227,
|
| 915 |
+
"eval_steps_per_second": 0.133,
|
| 916 |
+
"learning_rate": 0.0001,
|
| 917 |
+
"step": 15344
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 56.57,
|
| 921 |
+
"learning_rate": 0.0001,
|
| 922 |
+
"loss": 0.095,
|
| 923 |
+
"step": 15500
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 57.0,
|
| 927 |
+
"eval_accuracy": 0.5537334263782275,
|
| 928 |
+
"eval_f1_macro": 0.7241830253482859,
|
| 929 |
+
"eval_f1_micro": 0.8493824336688013,
|
| 930 |
+
"eval_loss": 0.0848437026143074,
|
| 931 |
+
"eval_roc_auc": 0.8993941682342463,
|
| 932 |
+
"eval_runtime": 677.1333,
|
| 933 |
+
"eval_samples_per_second": 4.233,
|
| 934 |
+
"eval_steps_per_second": 0.133,
|
| 935 |
+
"learning_rate": 0.0001,
|
| 936 |
+
"step": 15618
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 58.0,
|
| 940 |
+
"eval_accuracy": 0.5593161200279134,
|
| 941 |
+
"eval_f1_macro": 0.7363418087082886,
|
| 942 |
+
"eval_f1_micro": 0.8511604153662826,
|
| 943 |
+
"eval_loss": 0.08454328030347824,
|
| 944 |
+
"eval_roc_auc": 0.9029315644433922,
|
| 945 |
+
"eval_runtime": 675.3593,
|
| 946 |
+
"eval_samples_per_second": 4.244,
|
| 947 |
+
"eval_steps_per_second": 0.133,
|
| 948 |
+
"learning_rate": 0.0001,
|
| 949 |
+
"step": 15892
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 58.39,
|
| 953 |
+
"learning_rate": 0.0001,
|
| 954 |
+
"loss": 0.093,
|
| 955 |
+
"step": 16000
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 59.0,
|
| 959 |
+
"eval_accuracy": 0.560711793440335,
|
| 960 |
+
"eval_f1_macro": 0.73901392865669,
|
| 961 |
+
"eval_f1_micro": 0.8530955471527739,
|
| 962 |
+
"eval_loss": 0.08396653085947037,
|
| 963 |
+
"eval_roc_auc": 0.9058246057741859,
|
| 964 |
+
"eval_runtime": 679.8275,
|
| 965 |
+
"eval_samples_per_second": 4.216,
|
| 966 |
+
"eval_steps_per_second": 0.132,
|
| 967 |
+
"learning_rate": 0.0001,
|
| 968 |
+
"step": 16166
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 60.0,
|
| 972 |
+
"eval_accuracy": 0.5561758548499651,
|
| 973 |
+
"eval_f1_macro": 0.7472770304573509,
|
| 974 |
+
"eval_f1_micro": 0.852848189028787,
|
| 975 |
+
"eval_loss": 0.08474517613649368,
|
| 976 |
+
"eval_roc_auc": 0.9116141789978706,
|
| 977 |
+
"eval_runtime": 679.8207,
|
| 978 |
+
"eval_samples_per_second": 4.216,
|
| 979 |
+
"eval_steps_per_second": 0.132,
|
| 980 |
+
"learning_rate": 0.0001,
|
| 981 |
+
"step": 16440
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 60.22,
|
| 985 |
+
"learning_rate": 0.0001,
|
| 986 |
+
"loss": 0.0936,
|
| 987 |
+
"step": 16500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 61.0,
|
| 991 |
+
"eval_accuracy": 0.552337752965806,
|
| 992 |
+
"eval_f1_macro": 0.7425280881449604,
|
| 993 |
+
"eval_f1_micro": 0.8516569637259293,
|
| 994 |
+
"eval_loss": 0.08434043824672699,
|
| 995 |
+
"eval_roc_auc": 0.9078005379735077,
|
| 996 |
+
"eval_runtime": 678.8837,
|
| 997 |
+
"eval_samples_per_second": 4.222,
|
| 998 |
+
"eval_steps_per_second": 0.133,
|
| 999 |
+
"learning_rate": 0.0001,
|
| 1000 |
+
"step": 16714
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 62.0,
|
| 1004 |
+
"eval_accuracy": 0.5540823447313329,
|
| 1005 |
+
"eval_f1_macro": 0.7455853496732745,
|
| 1006 |
+
"eval_f1_micro": 0.8515365097265295,
|
| 1007 |
+
"eval_loss": 0.08436089754104614,
|
| 1008 |
+
"eval_roc_auc": 0.905273313320008,
|
| 1009 |
+
"eval_runtime": 684.8061,
|
| 1010 |
+
"eval_samples_per_second": 4.185,
|
| 1011 |
+
"eval_steps_per_second": 0.131,
|
| 1012 |
+
"learning_rate": 0.0001,
|
| 1013 |
+
"step": 16988
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 62.04,
|
| 1017 |
+
"learning_rate": 0.0001,
|
| 1018 |
+
"loss": 0.0932,
|
| 1019 |
+
"step": 17000
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 63.0,
|
| 1023 |
+
"eval_accuracy": 0.5575715282623867,
|
| 1024 |
+
"eval_f1_macro": 0.7344319075168565,
|
| 1025 |
+
"eval_f1_micro": 0.8535319341006545,
|
| 1026 |
+
"eval_loss": 0.0839960053563118,
|
| 1027 |
+
"eval_roc_auc": 0.9061600170593289,
|
| 1028 |
+
"eval_runtime": 676.8011,
|
| 1029 |
+
"eval_samples_per_second": 4.235,
|
| 1030 |
+
"eval_steps_per_second": 0.133,
|
| 1031 |
+
"learning_rate": 0.0001,
|
| 1032 |
+
"step": 17262
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 63.87,
|
| 1036 |
+
"learning_rate": 0.0001,
|
| 1037 |
+
"loss": 0.0933,
|
| 1038 |
+
"step": 17500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 64.0,
|
| 1042 |
+
"eval_accuracy": 0.5614096301465457,
|
| 1043 |
+
"eval_f1_macro": 0.7405199466064576,
|
| 1044 |
+
"eval_f1_micro": 0.8543109759531453,
|
| 1045 |
+
"eval_loss": 0.08395348489284515,
|
| 1046 |
+
"eval_roc_auc": 0.907220383474883,
|
| 1047 |
+
"eval_runtime": 676.0097,
|
| 1048 |
+
"eval_samples_per_second": 4.24,
|
| 1049 |
+
"eval_steps_per_second": 0.133,
|
| 1050 |
+
"learning_rate": 0.0001,
|
| 1051 |
+
"step": 17536
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 65.0,
|
| 1055 |
+
"eval_accuracy": 0.557920446615492,
|
| 1056 |
+
"eval_f1_macro": 0.7354221702015719,
|
| 1057 |
+
"eval_f1_micro": 0.8506689439225733,
|
| 1058 |
+
"eval_loss": 0.08403661847114563,
|
| 1059 |
+
"eval_roc_auc": 0.9015583167344123,
|
| 1060 |
+
"eval_runtime": 675.0487,
|
| 1061 |
+
"eval_samples_per_second": 4.246,
|
| 1062 |
+
"eval_steps_per_second": 0.133,
|
| 1063 |
+
"learning_rate": 0.0001,
|
| 1064 |
+
"step": 17810
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 65.69,
|
| 1068 |
+
"learning_rate": 0.0001,
|
| 1069 |
+
"loss": 0.0921,
|
| 1070 |
+
"step": 18000
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 66.0,
|
| 1074 |
+
"eval_accuracy": 0.5568736915561758,
|
| 1075 |
+
"eval_f1_macro": 0.7296578358578595,
|
| 1076 |
+
"eval_f1_micro": 0.852865023077789,
|
| 1077 |
+
"eval_loss": 0.08408054709434509,
|
| 1078 |
+
"eval_roc_auc": 0.9065963661690798,
|
| 1079 |
+
"eval_runtime": 680.0166,
|
| 1080 |
+
"eval_samples_per_second": 4.215,
|
| 1081 |
+
"eval_steps_per_second": 0.132,
|
| 1082 |
+
"learning_rate": 0.0001,
|
| 1083 |
+
"step": 18084
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 67.0,
|
| 1087 |
+
"eval_accuracy": 0.5540823447313329,
|
| 1088 |
+
"eval_f1_macro": 0.7392975848141861,
|
| 1089 |
+
"eval_f1_micro": 0.8539689628223736,
|
| 1090 |
+
"eval_loss": 0.08376849442720413,
|
| 1091 |
+
"eval_roc_auc": 0.9100385075348831,
|
| 1092 |
+
"eval_runtime": 675.941,
|
| 1093 |
+
"eval_samples_per_second": 4.24,
|
| 1094 |
+
"eval_steps_per_second": 0.133,
|
| 1095 |
+
"learning_rate": 0.0001,
|
| 1096 |
+
"step": 18358
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 67.52,
|
| 1100 |
+
"learning_rate": 0.0001,
|
| 1101 |
+
"loss": 0.0913,
|
| 1102 |
+
"step": 18500
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 68.0,
|
| 1106 |
+
"eval_accuracy": 0.5572226099092812,
|
| 1107 |
+
"eval_f1_macro": 0.7403483881006915,
|
| 1108 |
+
"eval_f1_micro": 0.854102492299076,
|
| 1109 |
+
"eval_loss": 0.08355987071990967,
|
| 1110 |
+
"eval_roc_auc": 0.9089826269243382,
|
| 1111 |
+
"eval_runtime": 676.7975,
|
| 1112 |
+
"eval_samples_per_second": 4.235,
|
| 1113 |
+
"eval_steps_per_second": 0.133,
|
| 1114 |
+
"learning_rate": 0.0001,
|
| 1115 |
+
"step": 18632
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 69.0,
|
| 1119 |
+
"eval_accuracy": 0.5582693649685974,
|
| 1120 |
+
"eval_f1_macro": 0.7494443807338856,
|
| 1121 |
+
"eval_f1_micro": 0.8547792062604807,
|
| 1122 |
+
"eval_loss": 0.08346723765134811,
|
| 1123 |
+
"eval_roc_auc": 0.9100283387486087,
|
| 1124 |
+
"eval_runtime": 675.2406,
|
| 1125 |
+
"eval_samples_per_second": 4.244,
|
| 1126 |
+
"eval_steps_per_second": 0.133,
|
| 1127 |
+
"learning_rate": 0.0001,
|
| 1128 |
+
"step": 18906
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 69.34,
|
| 1132 |
+
"learning_rate": 0.0001,
|
| 1133 |
+
"loss": 0.0911,
|
| 1134 |
+
"step": 19000
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 70.0,
|
| 1138 |
+
"eval_accuracy": 0.5561758548499651,
|
| 1139 |
+
"eval_f1_macro": 0.7486606655073544,
|
| 1140 |
+
"eval_f1_micro": 0.8551793496480055,
|
| 1141 |
+
"eval_loss": 0.08306006342172623,
|
| 1142 |
+
"eval_roc_auc": 0.9104037761073852,
|
| 1143 |
+
"eval_runtime": 675.6626,
|
| 1144 |
+
"eval_samples_per_second": 4.242,
|
| 1145 |
+
"eval_steps_per_second": 0.133,
|
| 1146 |
+
"learning_rate": 0.0001,
|
| 1147 |
+
"step": 19180
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 70.0,
|
| 1151 |
+
"learning_rate": 0.0001,
|
| 1152 |
+
"step": 19180,
|
| 1153 |
+
"total_flos": 9.099793269879256e+20,
|
| 1154 |
+
"train_loss": 0.11672632308896316,
|
| 1155 |
+
"train_runtime": 200748.2354,
|
| 1156 |
+
"train_samples_per_second": 3.057,
|
| 1157 |
+
"train_steps_per_second": 0.096
|
| 1158 |
+
}
|
| 1159 |
+
],
|
| 1160 |
+
"logging_steps": 500,
|
| 1161 |
+
"max_steps": 19180,
|
| 1162 |
+
"num_train_epochs": 70,
|
| 1163 |
+
"save_steps": 500,
|
| 1164 |
+
"total_flos": 9.099793269879256e+20,
|
| 1165 |
+
"trial_name": null,
|
| 1166 |
+
"trial_params": null
|
| 1167 |
+
}
|