Commit
•
1761b00
1
Parent(s):
e00d895
🍻 cheers
Browse files- README.md +10 -6
- all_results.json +16 -0
- config.json +1 -1
- eval_results.json +12 -0
- runs/Jan24_12-39-13_datavisu4/events.out.tfevents.1706335170.datavisu4.49957.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1412 -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_24-with_data_aug_batch-size32_epochs85_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: 0.0000
|
| 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_24-with_data_aug_batch-size32_epochs85_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: 0.0864
|
| 25 |
+
- F1 Micro: 0.8668
|
| 26 |
+
- F1 Macro: 0.8381
|
| 27 |
+
- Roc Auc: 0.9138
|
| 28 |
+
- Accuracy: 0.5805
|
| 29 |
- Learning Rate: 0.0000
|
| 30 |
|
| 31 |
## Model description
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 85.0,
|
| 3 |
+
"eval_accuracy": 0.5804980708523325,
|
| 4 |
+
"eval_f1_macro": 0.8380969738679429,
|
| 5 |
+
"eval_f1_micro": 0.866814335207588,
|
| 6 |
+
"eval_loss": 0.0863719955086708,
|
| 7 |
+
"eval_roc_auc": 0.9138424851830095,
|
| 8 |
+
"eval_runtime": 679.2028,
|
| 9 |
+
"eval_samples_per_second": 4.198,
|
| 10 |
+
"eval_steps_per_second": 0.133,
|
| 11 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 12 |
+
"train_loss": 0.10122811819873383,
|
| 13 |
+
"train_runtime": 238033.4867,
|
| 14 |
+
"train_samples_per_second": 3.13,
|
| 15 |
+
"train_steps_per_second": 0.098
|
| 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_24",
|
| 3 |
"apply_layernorm": true,
|
| 4 |
"architectures": [
|
| 5 |
"NewheadDinov2ForImageClassification"
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 85.0,
|
| 3 |
+
"eval_accuracy": 0.5804980708523325,
|
| 4 |
+
"eval_f1_macro": 0.8380969738679429,
|
| 5 |
+
"eval_f1_micro": 0.866814335207588,
|
| 6 |
+
"eval_loss": 0.0863719955086708,
|
| 7 |
+
"eval_roc_auc": 0.9138424851830095,
|
| 8 |
+
"eval_runtime": 679.2028,
|
| 9 |
+
"eval_samples_per_second": 4.198,
|
| 10 |
+
"eval_steps_per_second": 0.133,
|
| 11 |
+
"learning_rate": 1.0000000000000002e-07
|
| 12 |
+
}
|
runs/Jan24_12-39-13_datavisu4/events.out.tfevents.1706335170.datavisu4.49957.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44d0610a3743ee2428bd61ce11d14094a25de61298256d70fbb2f1a947b2cb86
|
| 3 |
+
size 634
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 85.0,
|
| 3 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 4 |
+
"train_loss": 0.10122811819873383,
|
| 5 |
+
"train_runtime": 238033.4867,
|
| 6 |
+
"train_samples_per_second": 3.13,
|
| 7 |
+
"train_steps_per_second": 0.098
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1412 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.08695908635854721,
|
| 3 |
+
"best_model_checkpoint": "/home1/datawork/mcontini/models/multilabel/huggingface/dinov2-large-2024_01_24-with_data_aug_batch-size32_epochs85_freeze/checkpoint-22742",
|
| 4 |
+
"epoch": 85.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 23290,
|
| 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": 0.45894224077940154,
|
| 14 |
+
"eval_f1_macro": 0.6395389989693074,
|
| 15 |
+
"eval_f1_micro": 0.7737575503857426,
|
| 16 |
+
"eval_loss": 0.13585977256298065,
|
| 17 |
+
"eval_roc_auc": 0.8471240403763409,
|
| 18 |
+
"eval_runtime": 675.8068,
|
| 19 |
+
"eval_samples_per_second": 4.253,
|
| 20 |
+
"eval_steps_per_second": 0.133,
|
| 21 |
+
"learning_rate": 0.001,
|
| 22 |
+
"step": 274
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 1.82,
|
| 26 |
+
"learning_rate": 0.001,
|
| 27 |
+
"loss": 0.2459,
|
| 28 |
+
"step": 500
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 2.0,
|
| 32 |
+
"eval_accuracy": 0.4940848990953375,
|
| 33 |
+
"eval_f1_macro": 0.7304998296932924,
|
| 34 |
+
"eval_f1_micro": 0.8032231694499591,
|
| 35 |
+
"eval_loss": 0.12362784147262573,
|
| 36 |
+
"eval_roc_auc": 0.8697341470820456,
|
| 37 |
+
"eval_runtime": 678.2974,
|
| 38 |
+
"eval_samples_per_second": 4.237,
|
| 39 |
+
"eval_steps_per_second": 0.133,
|
| 40 |
+
"learning_rate": 0.001,
|
| 41 |
+
"step": 548
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 3.0,
|
| 45 |
+
"eval_accuracy": 0.5125260960334029,
|
| 46 |
+
"eval_f1_macro": 0.7426440054746392,
|
| 47 |
+
"eval_f1_micro": 0.8174202432866652,
|
| 48 |
+
"eval_loss": 0.11671263724565506,
|
| 49 |
+
"eval_roc_auc": 0.8827824537503088,
|
| 50 |
+
"eval_runtime": 674.2849,
|
| 51 |
+
"eval_samples_per_second": 4.262,
|
| 52 |
+
"eval_steps_per_second": 0.133,
|
| 53 |
+
"learning_rate": 0.001,
|
| 54 |
+
"step": 822
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 3.65,
|
| 58 |
+
"learning_rate": 0.001,
|
| 59 |
+
"loss": 0.1403,
|
| 60 |
+
"step": 1000
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"eval_accuracy": 0.5100904662491301,
|
| 65 |
+
"eval_f1_macro": 0.7481206268648029,
|
| 66 |
+
"eval_f1_micro": 0.817623068527773,
|
| 67 |
+
"eval_loss": 0.11555441468954086,
|
| 68 |
+
"eval_roc_auc": 0.8825597364016536,
|
| 69 |
+
"eval_runtime": 684.1218,
|
| 70 |
+
"eval_samples_per_second": 4.201,
|
| 71 |
+
"eval_steps_per_second": 0.132,
|
| 72 |
+
"learning_rate": 0.001,
|
| 73 |
+
"step": 1096
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 5.0,
|
| 77 |
+
"eval_accuracy": 0.5243562978427279,
|
| 78 |
+
"eval_f1_macro": 0.7614020034586013,
|
| 79 |
+
"eval_f1_micro": 0.8267689489351958,
|
| 80 |
+
"eval_loss": 0.11359219998121262,
|
| 81 |
+
"eval_roc_auc": 0.8886760312325277,
|
| 82 |
+
"eval_runtime": 674.0166,
|
| 83 |
+
"eval_samples_per_second": 4.264,
|
| 84 |
+
"eval_steps_per_second": 0.134,
|
| 85 |
+
"learning_rate": 0.001,
|
| 86 |
+
"step": 1370
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 5.47,
|
| 90 |
+
"learning_rate": 0.001,
|
| 91 |
+
"loss": 0.1313,
|
| 92 |
+
"step": 1500
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 6.0,
|
| 96 |
+
"eval_accuracy": 0.5219206680584552,
|
| 97 |
+
"eval_f1_macro": 0.7508698006051816,
|
| 98 |
+
"eval_f1_micro": 0.8210489222998767,
|
| 99 |
+
"eval_loss": 0.11100047826766968,
|
| 100 |
+
"eval_roc_auc": 0.877677266975988,
|
| 101 |
+
"eval_runtime": 676.1,
|
| 102 |
+
"eval_samples_per_second": 4.251,
|
| 103 |
+
"eval_steps_per_second": 0.133,
|
| 104 |
+
"learning_rate": 0.001,
|
| 105 |
+
"step": 1644
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 7.0,
|
| 109 |
+
"eval_accuracy": 0.5323590814196242,
|
| 110 |
+
"eval_f1_macro": 0.7613673312506429,
|
| 111 |
+
"eval_f1_micro": 0.8288991092740292,
|
| 112 |
+
"eval_loss": 0.10846547037363052,
|
| 113 |
+
"eval_roc_auc": 0.8846228046955259,
|
| 114 |
+
"eval_runtime": 682.0096,
|
| 115 |
+
"eval_samples_per_second": 4.214,
|
| 116 |
+
"eval_steps_per_second": 0.132,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"step": 1918
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.3,
|
| 122 |
+
"learning_rate": 0.001,
|
| 123 |
+
"loss": 0.1289,
|
| 124 |
+
"step": 2000
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 8.0,
|
| 128 |
+
"eval_accuracy": 0.5379262352122477,
|
| 129 |
+
"eval_f1_macro": 0.7711215001442554,
|
| 130 |
+
"eval_f1_micro": 0.8331729408434757,
|
| 131 |
+
"eval_loss": 0.11005302518606186,
|
| 132 |
+
"eval_roc_auc": 0.8958012673255937,
|
| 133 |
+
"eval_runtime": 682.26,
|
| 134 |
+
"eval_samples_per_second": 4.212,
|
| 135 |
+
"eval_steps_per_second": 0.132,
|
| 136 |
+
"learning_rate": 0.001,
|
| 137 |
+
"step": 2192
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 9.0,
|
| 141 |
+
"eval_accuracy": 0.5139178844815588,
|
| 142 |
+
"eval_f1_macro": 0.7669688558128348,
|
| 143 |
+
"eval_f1_micro": 0.8271255519076193,
|
| 144 |
+
"eval_loss": 0.11129175871610641,
|
| 145 |
+
"eval_roc_auc": 0.8924250608458335,
|
| 146 |
+
"eval_runtime": 683.3423,
|
| 147 |
+
"eval_samples_per_second": 4.206,
|
| 148 |
+
"eval_steps_per_second": 0.132,
|
| 149 |
+
"learning_rate": 0.001,
|
| 150 |
+
"step": 2466
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 9.12,
|
| 154 |
+
"learning_rate": 0.001,
|
| 155 |
+
"loss": 0.1268,
|
| 156 |
+
"step": 2500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 10.0,
|
| 160 |
+
"eval_accuracy": 0.5313152400835073,
|
| 161 |
+
"eval_f1_macro": 0.7610925982620881,
|
| 162 |
+
"eval_f1_micro": 0.8258011503697616,
|
| 163 |
+
"eval_loss": 0.11381296068429947,
|
| 164 |
+
"eval_roc_auc": 0.880444980112697,
|
| 165 |
+
"eval_runtime": 679.9943,
|
| 166 |
+
"eval_samples_per_second": 4.227,
|
| 167 |
+
"eval_steps_per_second": 0.132,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"step": 2740
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.95,
|
| 173 |
+
"learning_rate": 0.001,
|
| 174 |
+
"loss": 0.1255,
|
| 175 |
+
"step": 3000
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 11.0,
|
| 179 |
+
"eval_accuracy": 0.5260960334029228,
|
| 180 |
+
"eval_f1_macro": 0.762697586166308,
|
| 181 |
+
"eval_f1_micro": 0.8262265016047684,
|
| 182 |
+
"eval_loss": 0.11390296369791031,
|
| 183 |
+
"eval_roc_auc": 0.8880168466934987,
|
| 184 |
+
"eval_runtime": 678.1509,
|
| 185 |
+
"eval_samples_per_second": 4.238,
|
| 186 |
+
"eval_steps_per_second": 0.133,
|
| 187 |
+
"learning_rate": 0.001,
|
| 188 |
+
"step": 3014
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 12.0,
|
| 192 |
+
"eval_accuracy": 0.5337508698677801,
|
| 193 |
+
"eval_f1_macro": 0.7573087365131856,
|
| 194 |
+
"eval_f1_micro": 0.8210012500744092,
|
| 195 |
+
"eval_loss": 0.11208122968673706,
|
| 196 |
+
"eval_roc_auc": 0.8736066784464123,
|
| 197 |
+
"eval_runtime": 680.166,
|
| 198 |
+
"eval_samples_per_second": 4.225,
|
| 199 |
+
"eval_steps_per_second": 0.132,
|
| 200 |
+
"learning_rate": 0.001,
|
| 201 |
+
"step": 3288
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 12.77,
|
| 205 |
+
"learning_rate": 0.001,
|
| 206 |
+
"loss": 0.1253,
|
| 207 |
+
"step": 3500
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 13.0,
|
| 211 |
+
"eval_accuracy": 0.5219206680584552,
|
| 212 |
+
"eval_f1_macro": 0.7489136029171714,
|
| 213 |
+
"eval_f1_micro": 0.8207366032466399,
|
| 214 |
+
"eval_loss": 0.1110881045460701,
|
| 215 |
+
"eval_roc_auc": 0.8803454162802951,
|
| 216 |
+
"eval_runtime": 682.0648,
|
| 217 |
+
"eval_samples_per_second": 4.214,
|
| 218 |
+
"eval_steps_per_second": 0.132,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"step": 3562
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 14.0,
|
| 224 |
+
"eval_accuracy": 0.5400139178844816,
|
| 225 |
+
"eval_f1_macro": 0.7776741330298375,
|
| 226 |
+
"eval_f1_micro": 0.8408186469584993,
|
| 227 |
+
"eval_loss": 0.10247301310300827,
|
| 228 |
+
"eval_roc_auc": 0.8987147268632997,
|
| 229 |
+
"eval_runtime": 676.5367,
|
| 230 |
+
"eval_samples_per_second": 4.248,
|
| 231 |
+
"eval_steps_per_second": 0.133,
|
| 232 |
+
"learning_rate": 0.0001,
|
| 233 |
+
"step": 3836
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 14.6,
|
| 237 |
+
"learning_rate": 0.0001,
|
| 238 |
+
"loss": 0.1171,
|
| 239 |
+
"step": 4000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 15.0,
|
| 243 |
+
"eval_accuracy": 0.5403618649965205,
|
| 244 |
+
"eval_f1_macro": 0.7795139529876273,
|
| 245 |
+
"eval_f1_micro": 0.842865329512894,
|
| 246 |
+
"eval_loss": 0.0998576357960701,
|
| 247 |
+
"eval_roc_auc": 0.897277663148542,
|
| 248 |
+
"eval_runtime": 675.6889,
|
| 249 |
+
"eval_samples_per_second": 4.253,
|
| 250 |
+
"eval_steps_per_second": 0.133,
|
| 251 |
+
"learning_rate": 0.0001,
|
| 252 |
+
"step": 4110
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 16.0,
|
| 256 |
+
"eval_accuracy": 0.5407098121085595,
|
| 257 |
+
"eval_f1_macro": 0.7861162275453341,
|
| 258 |
+
"eval_f1_micro": 0.8462626605556499,
|
| 259 |
+
"eval_loss": 0.10081179440021515,
|
| 260 |
+
"eval_roc_auc": 0.9032963122022265,
|
| 261 |
+
"eval_runtime": 680.4113,
|
| 262 |
+
"eval_samples_per_second": 4.224,
|
| 263 |
+
"eval_steps_per_second": 0.132,
|
| 264 |
+
"learning_rate": 0.0001,
|
| 265 |
+
"step": 4384
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 16.42,
|
| 269 |
+
"learning_rate": 0.0001,
|
| 270 |
+
"loss": 0.1107,
|
| 271 |
+
"step": 4500
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 17.0,
|
| 275 |
+
"eval_accuracy": 0.545929018789144,
|
| 276 |
+
"eval_f1_macro": 0.7877890037679841,
|
| 277 |
+
"eval_f1_micro": 0.8474232610532244,
|
| 278 |
+
"eval_loss": 0.10136950016021729,
|
| 279 |
+
"eval_roc_auc": 0.9054715489545434,
|
| 280 |
+
"eval_runtime": 689.6336,
|
| 281 |
+
"eval_samples_per_second": 4.167,
|
| 282 |
+
"eval_steps_per_second": 0.131,
|
| 283 |
+
"learning_rate": 0.0001,
|
| 284 |
+
"step": 4658
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 18.0,
|
| 288 |
+
"eval_accuracy": 0.5480167014613778,
|
| 289 |
+
"eval_f1_macro": 0.7867996984352024,
|
| 290 |
+
"eval_f1_micro": 0.8471123755334281,
|
| 291 |
+
"eval_loss": 0.09731467068195343,
|
| 292 |
+
"eval_roc_auc": 0.9019535814277009,
|
| 293 |
+
"eval_runtime": 689.7429,
|
| 294 |
+
"eval_samples_per_second": 4.167,
|
| 295 |
+
"eval_steps_per_second": 0.13,
|
| 296 |
+
"learning_rate": 0.0001,
|
| 297 |
+
"step": 4932
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 18.25,
|
| 301 |
+
"learning_rate": 0.0001,
|
| 302 |
+
"loss": 0.1078,
|
| 303 |
+
"step": 5000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 19.0,
|
| 307 |
+
"eval_accuracy": 0.5480167014613778,
|
| 308 |
+
"eval_f1_macro": 0.789354289479613,
|
| 309 |
+
"eval_f1_micro": 0.849087519068874,
|
| 310 |
+
"eval_loss": 0.09738590568304062,
|
| 311 |
+
"eval_roc_auc": 0.9053669532212902,
|
| 312 |
+
"eval_runtime": 687.0367,
|
| 313 |
+
"eval_samples_per_second": 4.183,
|
| 314 |
+
"eval_steps_per_second": 0.131,
|
| 315 |
+
"learning_rate": 0.0001,
|
| 316 |
+
"step": 5206
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 20.0,
|
| 320 |
+
"eval_accuracy": 0.5549756437021572,
|
| 321 |
+
"eval_f1_macro": 0.7947863154349663,
|
| 322 |
+
"eval_f1_micro": 0.8497521508745941,
|
| 323 |
+
"eval_loss": 0.0971071869134903,
|
| 324 |
+
"eval_roc_auc": 0.9029799344302967,
|
| 325 |
+
"eval_runtime": 693.4393,
|
| 326 |
+
"eval_samples_per_second": 4.145,
|
| 327 |
+
"eval_steps_per_second": 0.13,
|
| 328 |
+
"learning_rate": 0.0001,
|
| 329 |
+
"step": 5480
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 20.07,
|
| 333 |
+
"learning_rate": 0.0001,
|
| 334 |
+
"loss": 0.1061,
|
| 335 |
+
"step": 5500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 21.0,
|
| 339 |
+
"eval_accuracy": 0.5532359081419624,
|
| 340 |
+
"eval_f1_macro": 0.793994619616555,
|
| 341 |
+
"eval_f1_micro": 0.850910726332359,
|
| 342 |
+
"eval_loss": 0.09643097966909409,
|
| 343 |
+
"eval_roc_auc": 0.908055677859469,
|
| 344 |
+
"eval_runtime": 689.9756,
|
| 345 |
+
"eval_samples_per_second": 4.165,
|
| 346 |
+
"eval_steps_per_second": 0.13,
|
| 347 |
+
"learning_rate": 0.0001,
|
| 348 |
+
"step": 5754
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 21.9,
|
| 352 |
+
"learning_rate": 0.0001,
|
| 353 |
+
"loss": 0.1048,
|
| 354 |
+
"step": 6000
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 22.0,
|
| 358 |
+
"eval_accuracy": 0.5563674321503131,
|
| 359 |
+
"eval_f1_macro": 0.7973736665550476,
|
| 360 |
+
"eval_f1_micro": 0.8519603424966201,
|
| 361 |
+
"eval_loss": 0.096234992146492,
|
| 362 |
+
"eval_roc_auc": 0.9079748210535556,
|
| 363 |
+
"eval_runtime": 688.8118,
|
| 364 |
+
"eval_samples_per_second": 4.172,
|
| 365 |
+
"eval_steps_per_second": 0.131,
|
| 366 |
+
"learning_rate": 0.0001,
|
| 367 |
+
"step": 6028
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 23.0,
|
| 371 |
+
"eval_accuracy": 0.558455114822547,
|
| 372 |
+
"eval_f1_macro": 0.7969454250638132,
|
| 373 |
+
"eval_f1_micro": 0.8504731861198739,
|
| 374 |
+
"eval_loss": 0.09601961821317673,
|
| 375 |
+
"eval_roc_auc": 0.9012155078858011,
|
| 376 |
+
"eval_runtime": 688.0026,
|
| 377 |
+
"eval_samples_per_second": 4.177,
|
| 378 |
+
"eval_steps_per_second": 0.131,
|
| 379 |
+
"learning_rate": 0.0001,
|
| 380 |
+
"step": 6302
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 23.72,
|
| 384 |
+
"learning_rate": 0.0001,
|
| 385 |
+
"loss": 0.1038,
|
| 386 |
+
"step": 6500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 24.0,
|
| 390 |
+
"eval_accuracy": 0.5626304801670147,
|
| 391 |
+
"eval_f1_macro": 0.7974458635640262,
|
| 392 |
+
"eval_f1_micro": 0.8510467909850132,
|
| 393 |
+
"eval_loss": 0.09510745108127594,
|
| 394 |
+
"eval_roc_auc": 0.9024119192380319,
|
| 395 |
+
"eval_runtime": 688.423,
|
| 396 |
+
"eval_samples_per_second": 4.175,
|
| 397 |
+
"eval_steps_per_second": 0.131,
|
| 398 |
+
"learning_rate": 0.0001,
|
| 399 |
+
"step": 6576
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 25.0,
|
| 403 |
+
"eval_accuracy": 0.5643702157272095,
|
| 404 |
+
"eval_f1_macro": 0.795289513465328,
|
| 405 |
+
"eval_f1_micro": 0.8511713367018835,
|
| 406 |
+
"eval_loss": 0.0944407731294632,
|
| 407 |
+
"eval_roc_auc": 0.9012469687818218,
|
| 408 |
+
"eval_runtime": 683.8812,
|
| 409 |
+
"eval_samples_per_second": 4.202,
|
| 410 |
+
"eval_steps_per_second": 0.132,
|
| 411 |
+
"learning_rate": 0.0001,
|
| 412 |
+
"step": 6850
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 25.55,
|
| 416 |
+
"learning_rate": 0.0001,
|
| 417 |
+
"loss": 0.1017,
|
| 418 |
+
"step": 7000
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 26.0,
|
| 422 |
+
"eval_accuracy": 0.5640222686151705,
|
| 423 |
+
"eval_f1_macro": 0.8036711965439244,
|
| 424 |
+
"eval_f1_micro": 0.8572393605043909,
|
| 425 |
+
"eval_loss": 0.0948282852768898,
|
| 426 |
+
"eval_roc_auc": 0.9111790013806387,
|
| 427 |
+
"eval_runtime": 681.6858,
|
| 428 |
+
"eval_samples_per_second": 4.216,
|
| 429 |
+
"eval_steps_per_second": 0.132,
|
| 430 |
+
"learning_rate": 0.0001,
|
| 431 |
+
"step": 7124
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 27.0,
|
| 435 |
+
"eval_accuracy": 0.5636743215031316,
|
| 436 |
+
"eval_f1_macro": 0.8034638180358344,
|
| 437 |
+
"eval_f1_micro": 0.8551240743881069,
|
| 438 |
+
"eval_loss": 0.09229259192943573,
|
| 439 |
+
"eval_roc_auc": 0.9086109391822021,
|
| 440 |
+
"eval_runtime": 683.6776,
|
| 441 |
+
"eval_samples_per_second": 4.204,
|
| 442 |
+
"eval_steps_per_second": 0.132,
|
| 443 |
+
"learning_rate": 0.0001,
|
| 444 |
+
"step": 7398
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 27.37,
|
| 448 |
+
"learning_rate": 0.0001,
|
| 449 |
+
"loss": 0.1008,
|
| 450 |
+
"step": 7500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 28.0,
|
| 454 |
+
"eval_accuracy": 0.5643702157272095,
|
| 455 |
+
"eval_f1_macro": 0.8072611584992022,
|
| 456 |
+
"eval_f1_micro": 0.8561391580259505,
|
| 457 |
+
"eval_loss": 0.0919216200709343,
|
| 458 |
+
"eval_roc_auc": 0.9083895171196321,
|
| 459 |
+
"eval_runtime": 676.936,
|
| 460 |
+
"eval_samples_per_second": 4.246,
|
| 461 |
+
"eval_steps_per_second": 0.133,
|
| 462 |
+
"learning_rate": 0.0001,
|
| 463 |
+
"step": 7672
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 29.0,
|
| 467 |
+
"eval_accuracy": 0.5681976339596382,
|
| 468 |
+
"eval_f1_macro": 0.807775544791943,
|
| 469 |
+
"eval_f1_micro": 0.8571590844550463,
|
| 470 |
+
"eval_loss": 0.09229801595211029,
|
| 471 |
+
"eval_roc_auc": 0.9081680950570622,
|
| 472 |
+
"eval_runtime": 680.3447,
|
| 473 |
+
"eval_samples_per_second": 4.224,
|
| 474 |
+
"eval_steps_per_second": 0.132,
|
| 475 |
+
"learning_rate": 0.0001,
|
| 476 |
+
"step": 7946
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 29.2,
|
| 480 |
+
"learning_rate": 0.0001,
|
| 481 |
+
"loss": 0.1006,
|
| 482 |
+
"step": 8000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 30.0,
|
| 486 |
+
"eval_accuracy": 0.5636743215031316,
|
| 487 |
+
"eval_f1_macro": 0.8078629475879894,
|
| 488 |
+
"eval_f1_micro": 0.8560661454525001,
|
| 489 |
+
"eval_loss": 0.09243426471948624,
|
| 490 |
+
"eval_roc_auc": 0.9107996520688381,
|
| 491 |
+
"eval_runtime": 679.1764,
|
| 492 |
+
"eval_samples_per_second": 4.232,
|
| 493 |
+
"eval_steps_per_second": 0.133,
|
| 494 |
+
"learning_rate": 0.0001,
|
| 495 |
+
"step": 8220
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 31.0,
|
| 499 |
+
"eval_accuracy": 0.5688935281837161,
|
| 500 |
+
"eval_f1_macro": 0.8043753783436429,
|
| 501 |
+
"eval_f1_micro": 0.8549068890666057,
|
| 502 |
+
"eval_loss": 0.09250637888908386,
|
| 503 |
+
"eval_roc_auc": 0.9050076062220636,
|
| 504 |
+
"eval_runtime": 675.7031,
|
| 505 |
+
"eval_samples_per_second": 4.253,
|
| 506 |
+
"eval_steps_per_second": 0.133,
|
| 507 |
+
"learning_rate": 0.0001,
|
| 508 |
+
"step": 8494
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 31.02,
|
| 512 |
+
"learning_rate": 0.0001,
|
| 513 |
+
"loss": 0.0987,
|
| 514 |
+
"step": 8500
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 32.0,
|
| 518 |
+
"eval_accuracy": 0.5678496868475992,
|
| 519 |
+
"eval_f1_macro": 0.8071226305218325,
|
| 520 |
+
"eval_f1_micro": 0.858236685057989,
|
| 521 |
+
"eval_loss": 0.09133294969797134,
|
| 522 |
+
"eval_roc_auc": 0.9117040473456065,
|
| 523 |
+
"eval_runtime": 677.7385,
|
| 524 |
+
"eval_samples_per_second": 4.241,
|
| 525 |
+
"eval_steps_per_second": 0.133,
|
| 526 |
+
"learning_rate": 0.0001,
|
| 527 |
+
"step": 8768
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 32.85,
|
| 531 |
+
"learning_rate": 0.0001,
|
| 532 |
+
"loss": 0.0983,
|
| 533 |
+
"step": 9000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 33.0,
|
| 537 |
+
"eval_accuracy": 0.5692414752957551,
|
| 538 |
+
"eval_f1_macro": 0.8081519622072744,
|
| 539 |
+
"eval_f1_micro": 0.8570938803496942,
|
| 540 |
+
"eval_loss": 0.09114891290664673,
|
| 541 |
+
"eval_roc_auc": 0.9061295874509765,
|
| 542 |
+
"eval_runtime": 681.1845,
|
| 543 |
+
"eval_samples_per_second": 4.219,
|
| 544 |
+
"eval_steps_per_second": 0.132,
|
| 545 |
+
"learning_rate": 0.0001,
|
| 546 |
+
"step": 9042
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 34.0,
|
| 550 |
+
"eval_accuracy": 0.5709812108559499,
|
| 551 |
+
"eval_f1_macro": 0.8059984375887345,
|
| 552 |
+
"eval_f1_micro": 0.8570447522032734,
|
| 553 |
+
"eval_loss": 0.09058225899934769,
|
| 554 |
+
"eval_roc_auc": 0.9055923606377748,
|
| 555 |
+
"eval_runtime": 681.0802,
|
| 556 |
+
"eval_samples_per_second": 4.22,
|
| 557 |
+
"eval_steps_per_second": 0.132,
|
| 558 |
+
"learning_rate": 0.0001,
|
| 559 |
+
"step": 9316
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 34.67,
|
| 563 |
+
"learning_rate": 0.0001,
|
| 564 |
+
"loss": 0.0967,
|
| 565 |
+
"step": 9500
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 35.0,
|
| 569 |
+
"eval_accuracy": 0.5692414752957551,
|
| 570 |
+
"eval_f1_macro": 0.8103551770491668,
|
| 571 |
+
"eval_f1_micro": 0.857759845428198,
|
| 572 |
+
"eval_loss": 0.09091359376907349,
|
| 573 |
+
"eval_roc_auc": 0.9083150869963146,
|
| 574 |
+
"eval_runtime": 683.7099,
|
| 575 |
+
"eval_samples_per_second": 4.204,
|
| 576 |
+
"eval_steps_per_second": 0.132,
|
| 577 |
+
"learning_rate": 0.0001,
|
| 578 |
+
"step": 9590
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 36.0,
|
| 582 |
+
"eval_accuracy": 0.5748086290883786,
|
| 583 |
+
"eval_f1_macro": 0.8114188986781382,
|
| 584 |
+
"eval_f1_micro": 0.8582166040314315,
|
| 585 |
+
"eval_loss": 0.09166968613862991,
|
| 586 |
+
"eval_roc_auc": 0.9079081626467485,
|
| 587 |
+
"eval_runtime": 677.0062,
|
| 588 |
+
"eval_samples_per_second": 4.245,
|
| 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": 0.0963,
|
| 597 |
+
"step": 10000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 37.0,
|
| 601 |
+
"eval_accuracy": 0.5741127348643006,
|
| 602 |
+
"eval_f1_macro": 0.8104359485439742,
|
| 603 |
+
"eval_f1_micro": 0.8571918983865431,
|
| 604 |
+
"eval_loss": 0.09075025469064713,
|
| 605 |
+
"eval_roc_auc": 0.9057496153700481,
|
| 606 |
+
"eval_runtime": 682.1714,
|
| 607 |
+
"eval_samples_per_second": 4.213,
|
| 608 |
+
"eval_steps_per_second": 0.132,
|
| 609 |
+
"learning_rate": 0.0001,
|
| 610 |
+
"step": 10138
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 38.0,
|
| 614 |
+
"eval_accuracy": 0.5709812108559499,
|
| 615 |
+
"eval_f1_macro": 0.8135949001200257,
|
| 616 |
+
"eval_f1_micro": 0.8594423033325777,
|
| 617 |
+
"eval_loss": 0.09104561805725098,
|
| 618 |
+
"eval_roc_auc": 0.9101469439602342,
|
| 619 |
+
"eval_runtime": 690.1946,
|
| 620 |
+
"eval_samples_per_second": 4.164,
|
| 621 |
+
"eval_steps_per_second": 0.13,
|
| 622 |
+
"learning_rate": 0.0001,
|
| 623 |
+
"step": 10412
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 38.32,
|
| 627 |
+
"learning_rate": 0.0001,
|
| 628 |
+
"loss": 0.0957,
|
| 629 |
+
"step": 10500
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 39.0,
|
| 633 |
+
"eval_accuracy": 0.5685455810716771,
|
| 634 |
+
"eval_f1_macro": 0.808520223441343,
|
| 635 |
+
"eval_f1_micro": 0.8577247270464444,
|
| 636 |
+
"eval_loss": 0.09074629843235016,
|
| 637 |
+
"eval_roc_auc": 0.9098080230513902,
|
| 638 |
+
"eval_runtime": 678.1058,
|
| 639 |
+
"eval_samples_per_second": 4.238,
|
| 640 |
+
"eval_steps_per_second": 0.133,
|
| 641 |
+
"learning_rate": 0.0001,
|
| 642 |
+
"step": 10686
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 40.0,
|
| 646 |
+
"eval_accuracy": 0.5730688935281837,
|
| 647 |
+
"eval_f1_macro": 0.8111504469893477,
|
| 648 |
+
"eval_f1_micro": 0.8592332123411979,
|
| 649 |
+
"eval_loss": 0.09030281752347946,
|
| 650 |
+
"eval_roc_auc": 0.909802268885752,
|
| 651 |
+
"eval_runtime": 695.8681,
|
| 652 |
+
"eval_samples_per_second": 4.13,
|
| 653 |
+
"eval_steps_per_second": 0.129,
|
| 654 |
+
"learning_rate": 0.0001,
|
| 655 |
+
"step": 10960
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 40.15,
|
| 659 |
+
"learning_rate": 0.0001,
|
| 660 |
+
"loss": 0.0953,
|
| 661 |
+
"step": 11000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 41.0,
|
| 665 |
+
"eval_accuracy": 0.5716771050800278,
|
| 666 |
+
"eval_f1_macro": 0.8133805742659422,
|
| 667 |
+
"eval_f1_micro": 0.8586208856801775,
|
| 668 |
+
"eval_loss": 0.09064245969057083,
|
| 669 |
+
"eval_roc_auc": 0.9086828782290092,
|
| 670 |
+
"eval_runtime": 687.8411,
|
| 671 |
+
"eval_samples_per_second": 4.178,
|
| 672 |
+
"eval_steps_per_second": 0.131,
|
| 673 |
+
"learning_rate": 0.0001,
|
| 674 |
+
"step": 11234
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 41.97,
|
| 678 |
+
"learning_rate": 0.0001,
|
| 679 |
+
"loss": 0.0943,
|
| 680 |
+
"step": 11500
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 42.0,
|
| 684 |
+
"eval_accuracy": 0.5664578983994433,
|
| 685 |
+
"eval_f1_macro": 0.8135815799291138,
|
| 686 |
+
"eval_f1_micro": 0.8584246692032484,
|
| 687 |
+
"eval_loss": 0.09031981229782104,
|
| 688 |
+
"eval_roc_auc": 0.9089139403154726,
|
| 689 |
+
"eval_runtime": 684.2332,
|
| 690 |
+
"eval_samples_per_second": 4.2,
|
| 691 |
+
"eval_steps_per_second": 0.132,
|
| 692 |
+
"learning_rate": 0.0001,
|
| 693 |
+
"step": 11508
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 43.0,
|
| 697 |
+
"eval_accuracy": 0.569937369519833,
|
| 698 |
+
"eval_f1_macro": 0.8177715667121555,
|
| 699 |
+
"eval_f1_micro": 0.8603735373537355,
|
| 700 |
+
"eval_loss": 0.09048929065465927,
|
| 701 |
+
"eval_roc_auc": 0.9131758350455123,
|
| 702 |
+
"eval_runtime": 683.871,
|
| 703 |
+
"eval_samples_per_second": 4.203,
|
| 704 |
+
"eval_steps_per_second": 0.132,
|
| 705 |
+
"learning_rate": 0.0001,
|
| 706 |
+
"step": 11782
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 43.8,
|
| 710 |
+
"learning_rate": 0.0001,
|
| 711 |
+
"loss": 0.0947,
|
| 712 |
+
"step": 12000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 44.0,
|
| 716 |
+
"eval_accuracy": 0.5727209464161448,
|
| 717 |
+
"eval_f1_macro": 0.8149031816603105,
|
| 718 |
+
"eval_f1_micro": 0.8585443759981747,
|
| 719 |
+
"eval_loss": 0.090988889336586,
|
| 720 |
+
"eval_roc_auc": 0.9075230591693096,
|
| 721 |
+
"eval_runtime": 686.4073,
|
| 722 |
+
"eval_samples_per_second": 4.187,
|
| 723 |
+
"eval_steps_per_second": 0.131,
|
| 724 |
+
"learning_rate": 0.0001,
|
| 725 |
+
"step": 12056
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 45.0,
|
| 729 |
+
"eval_accuracy": 0.5727209464161448,
|
| 730 |
+
"eval_f1_macro": 0.8112945515986235,
|
| 731 |
+
"eval_f1_micro": 0.8590971272229823,
|
| 732 |
+
"eval_loss": 0.09051001071929932,
|
| 733 |
+
"eval_roc_auc": 0.9080583679272985,
|
| 734 |
+
"eval_runtime": 690.3088,
|
| 735 |
+
"eval_samples_per_second": 4.163,
|
| 736 |
+
"eval_steps_per_second": 0.13,
|
| 737 |
+
"learning_rate": 0.0001,
|
| 738 |
+
"step": 12330
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 45.62,
|
| 742 |
+
"learning_rate": 0.0001,
|
| 743 |
+
"loss": 0.0925,
|
| 744 |
+
"step": 12500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 46.0,
|
| 748 |
+
"eval_accuracy": 0.5727209464161448,
|
| 749 |
+
"eval_f1_macro": 0.8138956921603455,
|
| 750 |
+
"eval_f1_micro": 0.8608370193943518,
|
| 751 |
+
"eval_loss": 0.08959119021892548,
|
| 752 |
+
"eval_roc_auc": 0.9107387478276688,
|
| 753 |
+
"eval_runtime": 684.5538,
|
| 754 |
+
"eval_samples_per_second": 4.198,
|
| 755 |
+
"eval_steps_per_second": 0.131,
|
| 756 |
+
"learning_rate": 0.0001,
|
| 757 |
+
"step": 12604
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 47.0,
|
| 761 |
+
"eval_accuracy": 0.5744606819763396,
|
| 762 |
+
"eval_f1_macro": 0.8154159530277365,
|
| 763 |
+
"eval_f1_micro": 0.8598835217540253,
|
| 764 |
+
"eval_loss": 0.08953865617513657,
|
| 765 |
+
"eval_roc_auc": 0.9079274426945352,
|
| 766 |
+
"eval_runtime": 681.6068,
|
| 767 |
+
"eval_samples_per_second": 4.217,
|
| 768 |
+
"eval_steps_per_second": 0.132,
|
| 769 |
+
"learning_rate": 0.0001,
|
| 770 |
+
"step": 12878
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 47.45,
|
| 774 |
+
"learning_rate": 0.0001,
|
| 775 |
+
"loss": 0.0928,
|
| 776 |
+
"step": 13000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 48.0,
|
| 780 |
+
"eval_accuracy": 0.5744606819763396,
|
| 781 |
+
"eval_f1_macro": 0.8154966869589858,
|
| 782 |
+
"eval_f1_micro": 0.8605536922289807,
|
| 783 |
+
"eval_loss": 0.08962185680866241,
|
| 784 |
+
"eval_roc_auc": 0.9097631357688805,
|
| 785 |
+
"eval_runtime": 684.997,
|
| 786 |
+
"eval_samples_per_second": 4.196,
|
| 787 |
+
"eval_steps_per_second": 0.131,
|
| 788 |
+
"learning_rate": 0.0001,
|
| 789 |
+
"step": 13152
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 49.0,
|
| 793 |
+
"eval_accuracy": 0.5727209464161448,
|
| 794 |
+
"eval_f1_macro": 0.8168754926591527,
|
| 795 |
+
"eval_f1_micro": 0.8606169781580725,
|
| 796 |
+
"eval_loss": 0.08909053355455399,
|
| 797 |
+
"eval_roc_auc": 0.9130853382157057,
|
| 798 |
+
"eval_runtime": 683.2092,
|
| 799 |
+
"eval_samples_per_second": 4.207,
|
| 800 |
+
"eval_steps_per_second": 0.132,
|
| 801 |
+
"learning_rate": 0.0001,
|
| 802 |
+
"step": 13426
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 49.27,
|
| 806 |
+
"learning_rate": 0.0001,
|
| 807 |
+
"loss": 0.0914,
|
| 808 |
+
"step": 13500
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 50.0,
|
| 812 |
+
"eval_accuracy": 0.5734168406402227,
|
| 813 |
+
"eval_f1_macro": 0.8182687784925751,
|
| 814 |
+
"eval_f1_micro": 0.8616618652205841,
|
| 815 |
+
"eval_loss": 0.08951092511415482,
|
| 816 |
+
"eval_roc_auc": 0.9125141096821429,
|
| 817 |
+
"eval_runtime": 683.8641,
|
| 818 |
+
"eval_samples_per_second": 4.203,
|
| 819 |
+
"eval_steps_per_second": 0.132,
|
| 820 |
+
"learning_rate": 0.0001,
|
| 821 |
+
"step": 13700
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 51.0,
|
| 825 |
+
"eval_accuracy": 0.5668058455114823,
|
| 826 |
+
"eval_f1_macro": 0.8184177894108883,
|
| 827 |
+
"eval_f1_micro": 0.8608232987958555,
|
| 828 |
+
"eval_loss": 0.09029122442007065,
|
| 829 |
+
"eval_roc_auc": 0.914931294083072,
|
| 830 |
+
"eval_runtime": 685.4274,
|
| 831 |
+
"eval_samples_per_second": 4.193,
|
| 832 |
+
"eval_steps_per_second": 0.131,
|
| 833 |
+
"learning_rate": 0.0001,
|
| 834 |
+
"step": 13974
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 51.09,
|
| 838 |
+
"learning_rate": 0.0001,
|
| 839 |
+
"loss": 0.0919,
|
| 840 |
+
"step": 14000
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 52.0,
|
| 844 |
+
"eval_accuracy": 0.5762004175365344,
|
| 845 |
+
"eval_f1_macro": 0.8172163352414866,
|
| 846 |
+
"eval_f1_micro": 0.8617045454545454,
|
| 847 |
+
"eval_loss": 0.09041330218315125,
|
| 848 |
+
"eval_roc_auc": 0.9105776569849702,
|
| 849 |
+
"eval_runtime": 686.3022,
|
| 850 |
+
"eval_samples_per_second": 4.188,
|
| 851 |
+
"eval_steps_per_second": 0.131,
|
| 852 |
+
"learning_rate": 0.0001,
|
| 853 |
+
"step": 14248
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 52.92,
|
| 857 |
+
"learning_rate": 0.0001,
|
| 858 |
+
"loss": 0.091,
|
| 859 |
+
"step": 14500
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 53.0,
|
| 863 |
+
"eval_accuracy": 0.5734168406402227,
|
| 864 |
+
"eval_f1_macro": 0.8154347454270638,
|
| 865 |
+
"eval_f1_micro": 0.8604036655984708,
|
| 866 |
+
"eval_loss": 0.09106075763702393,
|
| 867 |
+
"eval_roc_auc": 0.913401765735465,
|
| 868 |
+
"eval_runtime": 686.9936,
|
| 869 |
+
"eval_samples_per_second": 4.183,
|
| 870 |
+
"eval_steps_per_second": 0.131,
|
| 871 |
+
"learning_rate": 0.0001,
|
| 872 |
+
"step": 14522
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 54.0,
|
| 876 |
+
"eval_accuracy": 0.5751565762004175,
|
| 877 |
+
"eval_f1_macro": 0.822392587875712,
|
| 878 |
+
"eval_f1_micro": 0.8628963639457711,
|
| 879 |
+
"eval_loss": 0.09085189551115036,
|
| 880 |
+
"eval_roc_auc": 0.9117971844954131,
|
| 881 |
+
"eval_runtime": 691.549,
|
| 882 |
+
"eval_samples_per_second": 4.156,
|
| 883 |
+
"eval_steps_per_second": 0.13,
|
| 884 |
+
"learning_rate": 0.0001,
|
| 885 |
+
"step": 14796
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 54.74,
|
| 889 |
+
"learning_rate": 0.0001,
|
| 890 |
+
"loss": 0.0907,
|
| 891 |
+
"step": 15000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 55.0,
|
| 895 |
+
"eval_accuracy": 0.5720250521920668,
|
| 896 |
+
"eval_f1_macro": 0.8246722143238872,
|
| 897 |
+
"eval_f1_micro": 0.862824401752612,
|
| 898 |
+
"eval_loss": 0.0893503949046135,
|
| 899 |
+
"eval_roc_auc": 0.9150558423810694,
|
| 900 |
+
"eval_runtime": 687.0743,
|
| 901 |
+
"eval_samples_per_second": 4.183,
|
| 902 |
+
"eval_steps_per_second": 0.131,
|
| 903 |
+
"learning_rate": 0.0001,
|
| 904 |
+
"step": 15070
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 56.0,
|
| 908 |
+
"eval_accuracy": 0.5723729993041058,
|
| 909 |
+
"eval_f1_macro": 0.8197285299784532,
|
| 910 |
+
"eval_f1_micro": 0.8613505337062617,
|
| 911 |
+
"eval_loss": 0.0895121842622757,
|
| 912 |
+
"eval_roc_auc": 0.9088388874230271,
|
| 913 |
+
"eval_runtime": 688.6878,
|
| 914 |
+
"eval_samples_per_second": 4.173,
|
| 915 |
+
"eval_steps_per_second": 0.131,
|
| 916 |
+
"learning_rate": 1e-05,
|
| 917 |
+
"step": 15344
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 56.57,
|
| 921 |
+
"learning_rate": 1e-05,
|
| 922 |
+
"loss": 0.0883,
|
| 923 |
+
"step": 15500
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 57.0,
|
| 927 |
+
"eval_accuracy": 0.5755045233124565,
|
| 928 |
+
"eval_f1_macro": 0.8261680546876228,
|
| 929 |
+
"eval_f1_micro": 0.8653240324032403,
|
| 930 |
+
"eval_loss": 0.08795319497585297,
|
| 931 |
+
"eval_roc_auc": 0.9159717957369441,
|
| 932 |
+
"eval_runtime": 680.4805,
|
| 933 |
+
"eval_samples_per_second": 4.223,
|
| 934 |
+
"eval_steps_per_second": 0.132,
|
| 935 |
+
"learning_rate": 1e-05,
|
| 936 |
+
"step": 15618
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 58.0,
|
| 940 |
+
"eval_accuracy": 0.5782881002087683,
|
| 941 |
+
"eval_f1_macro": 0.8227228870436498,
|
| 942 |
+
"eval_f1_micro": 0.8639262127078114,
|
| 943 |
+
"eval_loss": 0.08846761286258698,
|
| 944 |
+
"eval_roc_auc": 0.9111322457907458,
|
| 945 |
+
"eval_runtime": 678.456,
|
| 946 |
+
"eval_samples_per_second": 4.236,
|
| 947 |
+
"eval_steps_per_second": 0.133,
|
| 948 |
+
"learning_rate": 1e-05,
|
| 949 |
+
"step": 15892
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 58.39,
|
| 953 |
+
"learning_rate": 1e-05,
|
| 954 |
+
"loss": 0.0872,
|
| 955 |
+
"step": 16000
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 59.0,
|
| 959 |
+
"eval_accuracy": 0.5765483646485734,
|
| 960 |
+
"eval_f1_macro": 0.8262742568594247,
|
| 961 |
+
"eval_f1_micro": 0.8655003656409969,
|
| 962 |
+
"eval_loss": 0.0878983661532402,
|
| 963 |
+
"eval_roc_auc": 0.9160905401214736,
|
| 964 |
+
"eval_runtime": 680.5904,
|
| 965 |
+
"eval_samples_per_second": 4.223,
|
| 966 |
+
"eval_steps_per_second": 0.132,
|
| 967 |
+
"learning_rate": 1e-05,
|
| 968 |
+
"step": 16166
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 60.0,
|
| 972 |
+
"eval_accuracy": 0.5800278357689631,
|
| 973 |
+
"eval_f1_macro": 0.8238378094426198,
|
| 974 |
+
"eval_f1_micro": 0.8654139156932453,
|
| 975 |
+
"eval_loss": 0.08844566345214844,
|
| 976 |
+
"eval_roc_auc": 0.914969231409518,
|
| 977 |
+
"eval_runtime": 682.0838,
|
| 978 |
+
"eval_samples_per_second": 4.214,
|
| 979 |
+
"eval_steps_per_second": 0.132,
|
| 980 |
+
"learning_rate": 1e-05,
|
| 981 |
+
"step": 16440
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 60.22,
|
| 985 |
+
"learning_rate": 1e-05,
|
| 986 |
+
"loss": 0.0873,
|
| 987 |
+
"step": 16500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 61.0,
|
| 991 |
+
"eval_accuracy": 0.5744606819763396,
|
| 992 |
+
"eval_f1_macro": 0.8265572971487117,
|
| 993 |
+
"eval_f1_micro": 0.8651893408134642,
|
| 994 |
+
"eval_loss": 0.0878659188747406,
|
| 995 |
+
"eval_roc_auc": 0.9168337948077135,
|
| 996 |
+
"eval_runtime": 683.217,
|
| 997 |
+
"eval_samples_per_second": 4.207,
|
| 998 |
+
"eval_steps_per_second": 0.132,
|
| 999 |
+
"learning_rate": 1e-05,
|
| 1000 |
+
"step": 16714
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 62.0,
|
| 1004 |
+
"eval_accuracy": 0.5765483646485734,
|
| 1005 |
+
"eval_f1_macro": 0.8251828516128455,
|
| 1006 |
+
"eval_f1_micro": 0.8649870071178397,
|
| 1007 |
+
"eval_loss": 0.08799029141664505,
|
| 1008 |
+
"eval_roc_auc": 0.9143652466938494,
|
| 1009 |
+
"eval_runtime": 680.2736,
|
| 1010 |
+
"eval_samples_per_second": 4.225,
|
| 1011 |
+
"eval_steps_per_second": 0.132,
|
| 1012 |
+
"learning_rate": 1e-05,
|
| 1013 |
+
"step": 16988
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 62.04,
|
| 1017 |
+
"learning_rate": 1e-05,
|
| 1018 |
+
"loss": 0.0864,
|
| 1019 |
+
"step": 17000
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 63.0,
|
| 1023 |
+
"eval_accuracy": 0.5800278357689631,
|
| 1024 |
+
"eval_f1_macro": 0.8266891115852992,
|
| 1025 |
+
"eval_f1_micro": 0.8650424929178471,
|
| 1026 |
+
"eval_loss": 0.08828118443489075,
|
| 1027 |
+
"eval_roc_auc": 0.9134011927141672,
|
| 1028 |
+
"eval_runtime": 677.3735,
|
| 1029 |
+
"eval_samples_per_second": 4.243,
|
| 1030 |
+
"eval_steps_per_second": 0.133,
|
| 1031 |
+
"learning_rate": 1e-05,
|
| 1032 |
+
"step": 17262
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 63.87,
|
| 1036 |
+
"learning_rate": 1e-05,
|
| 1037 |
+
"loss": 0.086,
|
| 1038 |
+
"step": 17500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 64.0,
|
| 1042 |
+
"eval_accuracy": 0.5782881002087683,
|
| 1043 |
+
"eval_f1_macro": 0.8256635970178378,
|
| 1044 |
+
"eval_f1_micro": 0.8667077889306342,
|
| 1045 |
+
"eval_loss": 0.08754145354032516,
|
| 1046 |
+
"eval_roc_auc": 0.9178472944451183,
|
| 1047 |
+
"eval_runtime": 682.5828,
|
| 1048 |
+
"eval_samples_per_second": 4.21,
|
| 1049 |
+
"eval_steps_per_second": 0.132,
|
| 1050 |
+
"learning_rate": 1e-05,
|
| 1051 |
+
"step": 17536
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 65.0,
|
| 1055 |
+
"eval_accuracy": 0.58107167710508,
|
| 1056 |
+
"eval_f1_macro": 0.8277460823758025,
|
| 1057 |
+
"eval_f1_micro": 0.8669750648764526,
|
| 1058 |
+
"eval_loss": 0.08722905069589615,
|
| 1059 |
+
"eval_roc_auc": 0.9159442206991787,
|
| 1060 |
+
"eval_runtime": 673.5072,
|
| 1061 |
+
"eval_samples_per_second": 4.267,
|
| 1062 |
+
"eval_steps_per_second": 0.134,
|
| 1063 |
+
"learning_rate": 1e-05,
|
| 1064 |
+
"step": 17810
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 65.69,
|
| 1068 |
+
"learning_rate": 1e-05,
|
| 1069 |
+
"loss": 0.0855,
|
| 1070 |
+
"step": 18000
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 66.0,
|
| 1074 |
+
"eval_accuracy": 0.581767571329158,
|
| 1075 |
+
"eval_f1_macro": 0.8263083392061107,
|
| 1076 |
+
"eval_f1_micro": 0.8662405972512867,
|
| 1077 |
+
"eval_loss": 0.0872766524553299,
|
| 1078 |
+
"eval_roc_auc": 0.9146675753101624,
|
| 1079 |
+
"eval_runtime": 674.2325,
|
| 1080 |
+
"eval_samples_per_second": 4.263,
|
| 1081 |
+
"eval_steps_per_second": 0.133,
|
| 1082 |
+
"learning_rate": 1e-05,
|
| 1083 |
+
"step": 18084
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 67.0,
|
| 1087 |
+
"eval_accuracy": 0.5796798886569241,
|
| 1088 |
+
"eval_f1_macro": 0.8236507380069967,
|
| 1089 |
+
"eval_f1_micro": 0.8647603888351997,
|
| 1090 |
+
"eval_loss": 0.08779256045818329,
|
| 1091 |
+
"eval_roc_auc": 0.9121142845321298,
|
| 1092 |
+
"eval_runtime": 672.7686,
|
| 1093 |
+
"eval_samples_per_second": 4.272,
|
| 1094 |
+
"eval_steps_per_second": 0.134,
|
| 1095 |
+
"learning_rate": 1e-05,
|
| 1096 |
+
"step": 18358
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 67.52,
|
| 1100 |
+
"learning_rate": 1e-05,
|
| 1101 |
+
"loss": 0.0853,
|
| 1102 |
+
"step": 18500
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 68.0,
|
| 1106 |
+
"eval_accuracy": 0.580723729993041,
|
| 1107 |
+
"eval_f1_macro": 0.82334160354742,
|
| 1108 |
+
"eval_f1_micro": 0.8644058136221144,
|
| 1109 |
+
"eval_loss": 0.08787883818149567,
|
| 1110 |
+
"eval_roc_auc": 0.9110366175644288,
|
| 1111 |
+
"eval_runtime": 678.5717,
|
| 1112 |
+
"eval_samples_per_second": 4.235,
|
| 1113 |
+
"eval_steps_per_second": 0.133,
|
| 1114 |
+
"learning_rate": 1e-05,
|
| 1115 |
+
"step": 18632
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 69.0,
|
| 1119 |
+
"eval_accuracy": 0.5831593597773138,
|
| 1120 |
+
"eval_f1_macro": 0.8274164123414606,
|
| 1121 |
+
"eval_f1_micro": 0.8653988078342322,
|
| 1122 |
+
"eval_loss": 0.08730249851942062,
|
| 1123 |
+
"eval_roc_auc": 0.9129307238034322,
|
| 1124 |
+
"eval_runtime": 682.302,
|
| 1125 |
+
"eval_samples_per_second": 4.212,
|
| 1126 |
+
"eval_steps_per_second": 0.132,
|
| 1127 |
+
"learning_rate": 1e-05,
|
| 1128 |
+
"step": 18906
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 69.34,
|
| 1132 |
+
"learning_rate": 1e-05,
|
| 1133 |
+
"loss": 0.0854,
|
| 1134 |
+
"step": 19000
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 70.0,
|
| 1138 |
+
"eval_accuracy": 0.58107167710508,
|
| 1139 |
+
"eval_f1_macro": 0.8286701109278063,
|
| 1140 |
+
"eval_f1_micro": 0.8661381908135155,
|
| 1141 |
+
"eval_loss": 0.08733326941728592,
|
| 1142 |
+
"eval_roc_auc": 0.9166425383550794,
|
| 1143 |
+
"eval_runtime": 673.3186,
|
| 1144 |
+
"eval_samples_per_second": 4.268,
|
| 1145 |
+
"eval_steps_per_second": 0.134,
|
| 1146 |
+
"learning_rate": 1e-05,
|
| 1147 |
+
"step": 19180
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 71.0,
|
| 1151 |
+
"eval_accuracy": 0.5779401530967293,
|
| 1152 |
+
"eval_f1_macro": 0.8262073521627441,
|
| 1153 |
+
"eval_f1_micro": 0.865708650324035,
|
| 1154 |
+
"eval_loss": 0.08731996268033981,
|
| 1155 |
+
"eval_roc_auc": 0.9155950369973136,
|
| 1156 |
+
"eval_runtime": 672.8744,
|
| 1157 |
+
"eval_samples_per_second": 4.271,
|
| 1158 |
+
"eval_steps_per_second": 0.134,
|
| 1159 |
+
"learning_rate": 1e-05,
|
| 1160 |
+
"step": 19454
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 71.17,
|
| 1164 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1165 |
+
"loss": 0.0847,
|
| 1166 |
+
"step": 19500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 72.0,
|
| 1170 |
+
"eval_accuracy": 0.5803757828810021,
|
| 1171 |
+
"eval_f1_macro": 0.8279492189021646,
|
| 1172 |
+
"eval_f1_micro": 0.8660418654245468,
|
| 1173 |
+
"eval_loss": 0.08729101717472076,
|
| 1174 |
+
"eval_roc_auc": 0.9172015860404081,
|
| 1175 |
+
"eval_runtime": 676.9231,
|
| 1176 |
+
"eval_samples_per_second": 4.246,
|
| 1177 |
+
"eval_steps_per_second": 0.133,
|
| 1178 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1179 |
+
"step": 19728
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 72.99,
|
| 1183 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1184 |
+
"loss": 0.0852,
|
| 1185 |
+
"step": 20000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 73.0,
|
| 1189 |
+
"eval_accuracy": 0.5765483646485734,
|
| 1190 |
+
"eval_f1_macro": 0.8258696329291023,
|
| 1191 |
+
"eval_f1_micro": 0.8661956034096008,
|
| 1192 |
+
"eval_loss": 0.08899407833814621,
|
| 1193 |
+
"eval_roc_auc": 0.917537916377082,
|
| 1194 |
+
"eval_runtime": 674.8648,
|
| 1195 |
+
"eval_samples_per_second": 4.259,
|
| 1196 |
+
"eval_steps_per_second": 0.133,
|
| 1197 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1198 |
+
"step": 20002
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 74.0,
|
| 1202 |
+
"eval_accuracy": 0.5835073068893528,
|
| 1203 |
+
"eval_f1_macro": 0.8266751443826955,
|
| 1204 |
+
"eval_f1_micro": 0.8663119764546072,
|
| 1205 |
+
"eval_loss": 0.08706125617027283,
|
| 1206 |
+
"eval_roc_auc": 0.9144583340958263,
|
| 1207 |
+
"eval_runtime": 676.3788,
|
| 1208 |
+
"eval_samples_per_second": 4.249,
|
| 1209 |
+
"eval_steps_per_second": 0.133,
|
| 1210 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1211 |
+
"step": 20276
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 74.82,
|
| 1215 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1216 |
+
"loss": 0.0845,
|
| 1217 |
+
"step": 20500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 75.0,
|
| 1221 |
+
"eval_accuracy": 0.5762004175365344,
|
| 1222 |
+
"eval_f1_macro": 0.8242525331164202,
|
| 1223 |
+
"eval_f1_micro": 0.8650994982806247,
|
| 1224 |
+
"eval_loss": 0.08718431740999222,
|
| 1225 |
+
"eval_roc_auc": 0.9151367489348123,
|
| 1226 |
+
"eval_runtime": 674.1856,
|
| 1227 |
+
"eval_samples_per_second": 4.263,
|
| 1228 |
+
"eval_steps_per_second": 0.133,
|
| 1229 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1230 |
+
"step": 20550
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 76.0,
|
| 1234 |
+
"eval_accuracy": 0.5775922059846903,
|
| 1235 |
+
"eval_f1_macro": 0.8258404959868192,
|
| 1236 |
+
"eval_f1_micro": 0.8660362490149724,
|
| 1237 |
+
"eval_loss": 0.08712752908468246,
|
| 1238 |
+
"eval_roc_auc": 0.9161823322373652,
|
| 1239 |
+
"eval_runtime": 676.0536,
|
| 1240 |
+
"eval_samples_per_second": 4.251,
|
| 1241 |
+
"eval_steps_per_second": 0.133,
|
| 1242 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1243 |
+
"step": 20824
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 76.64,
|
| 1247 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1248 |
+
"loss": 0.0849,
|
| 1249 |
+
"step": 21000
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 77.0,
|
| 1253 |
+
"eval_accuracy": 0.5779401530967293,
|
| 1254 |
+
"eval_f1_macro": 0.8262597281814207,
|
| 1255 |
+
"eval_f1_micro": 0.8654561858576745,
|
| 1256 |
+
"eval_loss": 0.08787967264652252,
|
| 1257 |
+
"eval_roc_auc": 0.915242017185023,
|
| 1258 |
+
"eval_runtime": 678.4216,
|
| 1259 |
+
"eval_samples_per_second": 4.236,
|
| 1260 |
+
"eval_steps_per_second": 0.133,
|
| 1261 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1262 |
+
"step": 21098
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 78.0,
|
| 1266 |
+
"eval_accuracy": 0.5779401530967293,
|
| 1267 |
+
"eval_f1_macro": 0.824064674812195,
|
| 1268 |
+
"eval_f1_micro": 0.8647364849581541,
|
| 1269 |
+
"eval_loss": 0.08832630515098572,
|
| 1270 |
+
"eval_roc_auc": 0.9138800063627106,
|
| 1271 |
+
"eval_runtime": 674.504,
|
| 1272 |
+
"eval_samples_per_second": 4.261,
|
| 1273 |
+
"eval_steps_per_second": 0.133,
|
| 1274 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1275 |
+
"step": 21372
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 78.47,
|
| 1279 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1280 |
+
"loss": 0.0853,
|
| 1281 |
+
"step": 21500
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 79.0,
|
| 1285 |
+
"eval_accuracy": 0.580723729993041,
|
| 1286 |
+
"eval_f1_macro": 0.8283767069034536,
|
| 1287 |
+
"eval_f1_micro": 0.8667153859126425,
|
| 1288 |
+
"eval_loss": 0.08727473765611649,
|
| 1289 |
+
"eval_roc_auc": 0.9170071162464759,
|
| 1290 |
+
"eval_runtime": 680.1183,
|
| 1291 |
+
"eval_samples_per_second": 4.226,
|
| 1292 |
+
"eval_steps_per_second": 0.132,
|
| 1293 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1294 |
+
"step": 21646
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 80.0,
|
| 1298 |
+
"eval_accuracy": 0.581419624217119,
|
| 1299 |
+
"eval_f1_macro": 0.8257519474670673,
|
| 1300 |
+
"eval_f1_micro": 0.8654216185625353,
|
| 1301 |
+
"eval_loss": 0.08734780550003052,
|
| 1302 |
+
"eval_roc_auc": 0.9139968326920274,
|
| 1303 |
+
"eval_runtime": 682.9935,
|
| 1304 |
+
"eval_samples_per_second": 4.208,
|
| 1305 |
+
"eval_steps_per_second": 0.132,
|
| 1306 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1307 |
+
"step": 21920
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 80.29,
|
| 1311 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1312 |
+
"loss": 0.0838,
|
| 1313 |
+
"step": 22000
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 81.0,
|
| 1317 |
+
"eval_accuracy": 0.5828114126652749,
|
| 1318 |
+
"eval_f1_macro": 0.8261813753948223,
|
| 1319 |
+
"eval_f1_micro": 0.8653922514039366,
|
| 1320 |
+
"eval_loss": 0.08708538860082626,
|
| 1321 |
+
"eval_roc_auc": 0.9131951648411291,
|
| 1322 |
+
"eval_runtime": 690.614,
|
| 1323 |
+
"eval_samples_per_second": 4.162,
|
| 1324 |
+
"eval_steps_per_second": 0.13,
|
| 1325 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1326 |
+
"step": 22194
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 82.0,
|
| 1330 |
+
"eval_accuracy": 0.581767571329158,
|
| 1331 |
+
"eval_f1_macro": 0.8253000981325144,
|
| 1332 |
+
"eval_f1_micro": 0.866888801039137,
|
| 1333 |
+
"eval_loss": 0.08740255981683731,
|
| 1334 |
+
"eval_roc_auc": 0.9155308696670169,
|
| 1335 |
+
"eval_runtime": 680.0034,
|
| 1336 |
+
"eval_samples_per_second": 4.226,
|
| 1337 |
+
"eval_steps_per_second": 0.132,
|
| 1338 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1339 |
+
"step": 22468
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 82.12,
|
| 1343 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1344 |
+
"loss": 0.0842,
|
| 1345 |
+
"step": 22500
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 83.0,
|
| 1349 |
+
"eval_accuracy": 0.5845511482254697,
|
| 1350 |
+
"eval_f1_macro": 0.8282173993454429,
|
| 1351 |
+
"eval_f1_micro": 0.8666929710839298,
|
| 1352 |
+
"eval_loss": 0.08695908635854721,
|
| 1353 |
+
"eval_roc_auc": 0.9160732278767293,
|
| 1354 |
+
"eval_runtime": 685.2501,
|
| 1355 |
+
"eval_samples_per_second": 4.194,
|
| 1356 |
+
"eval_steps_per_second": 0.131,
|
| 1357 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1358 |
+
"step": 22742
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 83.94,
|
| 1362 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1363 |
+
"loss": 0.0842,
|
| 1364 |
+
"step": 23000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 84.0,
|
| 1368 |
+
"eval_accuracy": 0.58107167710508,
|
| 1369 |
+
"eval_f1_macro": 0.8233437650206237,
|
| 1370 |
+
"eval_f1_micro": 0.8627316009866345,
|
| 1371 |
+
"eval_loss": 0.08810650557279587,
|
| 1372 |
+
"eval_roc_auc": 0.9079679208453217,
|
| 1373 |
+
"eval_runtime": 681.9462,
|
| 1374 |
+
"eval_samples_per_second": 4.214,
|
| 1375 |
+
"eval_steps_per_second": 0.132,
|
| 1376 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1377 |
+
"step": 23016
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 85.0,
|
| 1381 |
+
"eval_accuracy": 0.580723729993041,
|
| 1382 |
+
"eval_f1_macro": 0.8276925304690478,
|
| 1383 |
+
"eval_f1_micro": 0.8657459814353634,
|
| 1384 |
+
"eval_loss": 0.08707784116268158,
|
| 1385 |
+
"eval_roc_auc": 0.9141406112899818,
|
| 1386 |
+
"eval_runtime": 686.2064,
|
| 1387 |
+
"eval_samples_per_second": 4.188,
|
| 1388 |
+
"eval_steps_per_second": 0.131,
|
| 1389 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1390 |
+
"step": 23290
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 85.0,
|
| 1394 |
+
"learning_rate": 1.0000000000000002e-07,
|
| 1395 |
+
"step": 23290,
|
| 1396 |
+
"total_flos": 1.1045912459199104e+21,
|
| 1397 |
+
"train_loss": 0.10122811819873383,
|
| 1398 |
+
"train_runtime": 238033.4867,
|
| 1399 |
+
"train_samples_per_second": 3.13,
|
| 1400 |
+
"train_steps_per_second": 0.098
|
| 1401 |
+
}
|
| 1402 |
+
],
|
| 1403 |
+
"logging_steps": 500,
|
| 1404 |
+
"max_steps": 23290,
|
| 1405 |
+
"num_input_tokens_seen": 0,
|
| 1406 |
+
"num_train_epochs": 85,
|
| 1407 |
+
"save_steps": 500,
|
| 1408 |
+
"total_flos": 1.1045912459199104e+21,
|
| 1409 |
+
"train_batch_size": 32,
|
| 1410 |
+
"trial_name": null,
|
| 1411 |
+
"trial_params": null
|
| 1412 |
+
}
|