Sloan Haywood
commited on
Commit
Β·
a3f1ae4
1
Parent(s):
e3db564
Update to latest model
Browse files- README.md +7 -7
- {checkpoint-9995 β checkpoint-10128}/config.json +1 -1
- {checkpoint-9995 β checkpoint-10128}/model.safetensors +1 -1
- {checkpoint-9995 β checkpoint-10128}/optimizer.pt +1 -1
- {checkpoint-9995 β checkpoint-10128}/rng_state.pth +1 -1
- {checkpoint-9995 β checkpoint-10128}/scheduler.pt +1 -1
- checkpoint-10128/trainer_state.json +2890 -0
- {checkpoint-9995 β checkpoint-10128}/training_args.bin +1 -1
- checkpoint-9995/trainer_state.json +0 -2848
- config.json +1 -1
- model.safetensors +1 -1
- runs/{Nov17_06-39-09_2db0527f7c04/events.out.tfevents.1731825557.2db0527f7c04.61.0 β Nov22_06-35-14_170e5a2edfa3/events.out.tfevents.1732257322.170e5a2edfa3.82.0} +2 -2
- runs/{Nov17_06-39-09_2db0527f7c04/events.out.tfevents.1731860413.2db0527f7c04.61.1 β Nov22_06-35-14_170e5a2edfa3/events.out.tfevents.1732300664.170e5a2edfa3.82.1} +1 -1
- training_args.bin +1 -1
- training_params.json +3 -3
README.md
CHANGED
@@ -3,7 +3,7 @@
|
|
3 |
tags:
|
4 |
- autotrain
|
5 |
- image-classification
|
6 |
-
base_model: haywoodsloan/
|
7 |
widget:
|
8 |
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
9 |
example_title: Tiger
|
@@ -18,14 +18,14 @@ widget:
|
|
18 |
- Problem type: Image Classification
|
19 |
|
20 |
## Validation Metrics
|
21 |
-
loss: 0.
|
22 |
|
23 |
-
f1: 0.
|
24 |
|
25 |
-
precision: 0.
|
26 |
|
27 |
-
recall: 0.
|
28 |
|
29 |
-
auc: 0.
|
30 |
|
31 |
-
accuracy: 0.
|
|
|
3 |
tags:
|
4 |
- autotrain
|
5 |
- image-classification
|
6 |
+
base_model: haywoodsloan/ai-image-detector-deploy
|
7 |
widget:
|
8 |
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
9 |
example_title: Tiger
|
|
|
18 |
- Problem type: Image Classification
|
19 |
|
20 |
## Validation Metrics
|
21 |
+
loss: 0.17222872376441956
|
22 |
|
23 |
+
f1: 0.9689199689199689
|
24 |
|
25 |
+
precision: 0.963554410817772
|
26 |
|
27 |
+
recall: 0.9743456179189999
|
28 |
|
29 |
+
auc: 0.9842785960253444
|
30 |
|
31 |
+
accuracy: 0.9661280079034649
|
{checkpoint-9995 β checkpoint-10128}/config.json
RENAMED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "haywoodsloan/
|
3 |
"_num_labels": 2,
|
4 |
"architectures": [
|
5 |
"Swinv2ForImageClassification"
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "haywoodsloan/ai-image-detector-deploy",
|
3 |
"_num_labels": 2,
|
4 |
"architectures": [
|
5 |
"Swinv2ForImageClassification"
|
{checkpoint-9995 β checkpoint-10128}/model.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 780886280
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3aa6f9de7bb58e821bbd577afe92c1092a78d14210b399ae151ab661d2e43a57
|
3 |
size 780886280
|
{checkpoint-9995 β checkpoint-10128}/optimizer.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1562042973
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59c12dd673f211a88664132be0ba4422bd106d45d5d4a150d149cf0249135771
|
3 |
size 1562042973
|
{checkpoint-9995 β checkpoint-10128}/rng_state.pth
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:378ec6eb2f5e6cb562655db417b317f4b61227e6c341c73665d022b86e4b0cac
|
3 |
size 14244
|
{checkpoint-9995 β checkpoint-10128}/scheduler.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e29d9ea63b662945b7a125775eb969220727f394deda337f81a71d96d1933d4d
|
3 |
size 1064
|
checkpoint-10128/trainer_state.json
ADDED
@@ -0,0 +1,2890 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.17222872376441956,
|
3 |
+
"best_model_checkpoint": "autotrain-ai-image-detect-20241122-0116/checkpoint-10128",
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 10128,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.0024684044233807267,
|
13 |
+
"grad_norm": 0.16953732073307037,
|
14 |
+
"learning_rate": 3.999939864600428e-05,
|
15 |
+
"loss": 0.1695,
|
16 |
+
"step": 25
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.0049368088467614535,
|
20 |
+
"grad_norm": 0.11506114900112152,
|
21 |
+
"learning_rate": 3.999759462017979e-05,
|
22 |
+
"loss": 0.0482,
|
23 |
+
"step": 50
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.00740521327014218,
|
27 |
+
"grad_norm": 0.9763417840003967,
|
28 |
+
"learning_rate": 3.999458803101234e-05,
|
29 |
+
"loss": 0.1642,
|
30 |
+
"step": 75
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.009873617693522907,
|
34 |
+
"grad_norm": 0.8805496096611023,
|
35 |
+
"learning_rate": 3.999037905930437e-05,
|
36 |
+
"loss": 0.0766,
|
37 |
+
"step": 100
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.012342022116903634,
|
41 |
+
"grad_norm": 0.03244762495160103,
|
42 |
+
"learning_rate": 3.998496795816408e-05,
|
43 |
+
"loss": 0.1431,
|
44 |
+
"step": 125
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.01481042654028436,
|
48 |
+
"grad_norm": 0.5744362473487854,
|
49 |
+
"learning_rate": 3.9978355052990185e-05,
|
50 |
+
"loss": 0.1642,
|
51 |
+
"step": 150
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.017278830963665087,
|
55 |
+
"grad_norm": 1.6235884428024292,
|
56 |
+
"learning_rate": 3.99705407414524e-05,
|
57 |
+
"loss": 0.2305,
|
58 |
+
"step": 175
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.019747235387045814,
|
62 |
+
"grad_norm": 0.07429923862218857,
|
63 |
+
"learning_rate": 3.996152549346746e-05,
|
64 |
+
"loss": 0.1542,
|
65 |
+
"step": 200
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.02221563981042654,
|
69 |
+
"grad_norm": 20.05841827392578,
|
70 |
+
"learning_rate": 3.99513098511709e-05,
|
71 |
+
"loss": 0.1219,
|
72 |
+
"step": 225
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.024684044233807267,
|
76 |
+
"grad_norm": 1.5969923734664917,
|
77 |
+
"learning_rate": 3.993989442888447e-05,
|
78 |
+
"loss": 0.101,
|
79 |
+
"step": 250
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.027152448657187994,
|
83 |
+
"grad_norm": 24.882328033447266,
|
84 |
+
"learning_rate": 3.992727991307913e-05,
|
85 |
+
"loss": 0.1503,
|
86 |
+
"step": 275
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.02962085308056872,
|
90 |
+
"grad_norm": 0.20979249477386475,
|
91 |
+
"learning_rate": 3.991346706233384e-05,
|
92 |
+
"loss": 0.2005,
|
93 |
+
"step": 300
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.03208925750394945,
|
97 |
+
"grad_norm": 0.127052903175354,
|
98 |
+
"learning_rate": 3.9898456707289896e-05,
|
99 |
+
"loss": 0.2264,
|
100 |
+
"step": 325
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.034557661927330174,
|
104 |
+
"grad_norm": 13.576021194458008,
|
105 |
+
"learning_rate": 3.9882249750601e-05,
|
106 |
+
"loss": 0.1329,
|
107 |
+
"step": 350
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.0370260663507109,
|
111 |
+
"grad_norm": 30.27947998046875,
|
112 |
+
"learning_rate": 3.986484716687897e-05,
|
113 |
+
"loss": 0.245,
|
114 |
+
"step": 375
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.03949447077409163,
|
118 |
+
"grad_norm": 0.07906048744916916,
|
119 |
+
"learning_rate": 3.9846250002635114e-05,
|
120 |
+
"loss": 0.0856,
|
121 |
+
"step": 400
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.041962875197472355,
|
125 |
+
"grad_norm": 1.6535409688949585,
|
126 |
+
"learning_rate": 3.982645937621735e-05,
|
127 |
+
"loss": 0.1353,
|
128 |
+
"step": 425
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.04443127962085308,
|
132 |
+
"grad_norm": 0.44432130455970764,
|
133 |
+
"learning_rate": 3.980547647774291e-05,
|
134 |
+
"loss": 0.3194,
|
135 |
+
"step": 450
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.04689968404423381,
|
139 |
+
"grad_norm": 20.855363845825195,
|
140 |
+
"learning_rate": 3.978330256902676e-05,
|
141 |
+
"loss": 0.235,
|
142 |
+
"step": 475
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.049368088467614535,
|
146 |
+
"grad_norm": 110.83573913574219,
|
147 |
+
"learning_rate": 3.975993898350578e-05,
|
148 |
+
"loss": 0.1729,
|
149 |
+
"step": 500
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.05183649289099526,
|
153 |
+
"grad_norm": 0.15488630533218384,
|
154 |
+
"learning_rate": 3.973538712615851e-05,
|
155 |
+
"loss": 0.2135,
|
156 |
+
"step": 525
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.05430489731437599,
|
160 |
+
"grad_norm": 0.38120362162590027,
|
161 |
+
"learning_rate": 3.97096484734207e-05,
|
162 |
+
"loss": 0.1807,
|
163 |
+
"step": 550
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.056773301737756715,
|
167 |
+
"grad_norm": 0.09759721159934998,
|
168 |
+
"learning_rate": 3.968272457309653e-05,
|
169 |
+
"loss": 0.1103,
|
170 |
+
"step": 575
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.05924170616113744,
|
174 |
+
"grad_norm": 45.73832321166992,
|
175 |
+
"learning_rate": 3.965461704426548e-05,
|
176 |
+
"loss": 0.1847,
|
177 |
+
"step": 600
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.06171011058451817,
|
181 |
+
"grad_norm": 0.0426543727517128,
|
182 |
+
"learning_rate": 3.962532757718505e-05,
|
183 |
+
"loss": 0.0569,
|
184 |
+
"step": 625
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.0641785150078989,
|
188 |
+
"grad_norm": 21.13721466064453,
|
189 |
+
"learning_rate": 3.959485793318904e-05,
|
190 |
+
"loss": 0.2356,
|
191 |
+
"step": 650
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.06664691943127962,
|
195 |
+
"grad_norm": 1.2969528436660767,
|
196 |
+
"learning_rate": 3.956320994458166e-05,
|
197 |
+
"loss": 0.2,
|
198 |
+
"step": 675
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.06911532385466035,
|
202 |
+
"grad_norm": 2.2889373302459717,
|
203 |
+
"learning_rate": 3.9530385514527354e-05,
|
204 |
+
"loss": 0.1388,
|
205 |
+
"step": 700
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.07158372827804108,
|
209 |
+
"grad_norm": 33.99475860595703,
|
210 |
+
"learning_rate": 3.9496386616936344e-05,
|
211 |
+
"loss": 0.1525,
|
212 |
+
"step": 725
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.0740521327014218,
|
216 |
+
"grad_norm": 0.5939781069755554,
|
217 |
+
"learning_rate": 3.946121529634592e-05,
|
218 |
+
"loss": 0.1396,
|
219 |
+
"step": 750
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.07652053712480253,
|
223 |
+
"grad_norm": 0.04325993359088898,
|
224 |
+
"learning_rate": 3.942487366779749e-05,
|
225 |
+
"loss": 0.1201,
|
226 |
+
"step": 775
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.07898894154818326,
|
230 |
+
"grad_norm": 0.7331811189651489,
|
231 |
+
"learning_rate": 3.938736391670942e-05,
|
232 |
+
"loss": 0.1682,
|
233 |
+
"step": 800
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.08145734597156398,
|
237 |
+
"grad_norm": 0.5789724588394165,
|
238 |
+
"learning_rate": 3.934868829874558e-05,
|
239 |
+
"loss": 0.1426,
|
240 |
+
"step": 825
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.08392575039494471,
|
244 |
+
"grad_norm": 0.1745823323726654,
|
245 |
+
"learning_rate": 3.930884913967969e-05,
|
246 |
+
"loss": 0.1546,
|
247 |
+
"step": 850
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.08639415481832544,
|
251 |
+
"grad_norm": 0.6888066530227661,
|
252 |
+
"learning_rate": 3.9267848835255525e-05,
|
253 |
+
"loss": 0.0837,
|
254 |
+
"step": 875
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.08886255924170616,
|
258 |
+
"grad_norm": 8.269588470458984,
|
259 |
+
"learning_rate": 3.922568985104275e-05,
|
260 |
+
"loss": 0.1846,
|
261 |
+
"step": 900
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.09133096366508689,
|
265 |
+
"grad_norm": 14.444374084472656,
|
266 |
+
"learning_rate": 3.918237472228875e-05,
|
267 |
+
"loss": 0.1809,
|
268 |
+
"step": 925
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.09379936808846762,
|
272 |
+
"grad_norm": 74.40658569335938,
|
273 |
+
"learning_rate": 3.913790605376609e-05,
|
274 |
+
"loss": 0.1824,
|
275 |
+
"step": 950
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.09626777251184834,
|
279 |
+
"grad_norm": 0.2251996099948883,
|
280 |
+
"learning_rate": 3.9092286519615906e-05,
|
281 |
+
"loss": 0.248,
|
282 |
+
"step": 975
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.09873617693522907,
|
286 |
+
"grad_norm": 5.18855619430542,
|
287 |
+
"learning_rate": 3.904551886318713e-05,
|
288 |
+
"loss": 0.18,
|
289 |
+
"step": 1000
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.1012045813586098,
|
293 |
+
"grad_norm": 0.14439088106155396,
|
294 |
+
"learning_rate": 3.899760589687147e-05,
|
295 |
+
"loss": 0.1528,
|
296 |
+
"step": 1025
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.10367298578199052,
|
300 |
+
"grad_norm": 0.2735329270362854,
|
301 |
+
"learning_rate": 3.894855050193429e-05,
|
302 |
+
"loss": 0.1134,
|
303 |
+
"step": 1050
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.10614139020537125,
|
307 |
+
"grad_norm": 0.405011922121048,
|
308 |
+
"learning_rate": 3.889835562834136e-05,
|
309 |
+
"loss": 0.1441,
|
310 |
+
"step": 1075
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.10860979462875198,
|
314 |
+
"grad_norm": 11.461868286132812,
|
315 |
+
"learning_rate": 3.8847024294581486e-05,
|
316 |
+
"loss": 0.1731,
|
317 |
+
"step": 1100
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.1110781990521327,
|
321 |
+
"grad_norm": 0.19057828187942505,
|
322 |
+
"learning_rate": 3.8794559587484906e-05,
|
323 |
+
"loss": 0.1572,
|
324 |
+
"step": 1125
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.11354660347551343,
|
328 |
+
"grad_norm": 0.0868346244096756,
|
329 |
+
"learning_rate": 3.874096466203776e-05,
|
330 |
+
"loss": 0.0881,
|
331 |
+
"step": 1150
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.11601500789889416,
|
335 |
+
"grad_norm": 0.09796486794948578,
|
336 |
+
"learning_rate": 3.86862427411923e-05,
|
337 |
+
"loss": 0.0859,
|
338 |
+
"step": 1175
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.11848341232227488,
|
342 |
+
"grad_norm": 31.533279418945312,
|
343 |
+
"learning_rate": 3.863039711567311e-05,
|
344 |
+
"loss": 0.0939,
|
345 |
+
"step": 1200
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.12095181674565561,
|
349 |
+
"grad_norm": 0.15282464027404785,
|
350 |
+
"learning_rate": 3.857343114377917e-05,
|
351 |
+
"loss": 0.1977,
|
352 |
+
"step": 1225
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.12342022116903634,
|
356 |
+
"grad_norm": 12.384027481079102,
|
357 |
+
"learning_rate": 3.8515348251181995e-05,
|
358 |
+
"loss": 0.2683,
|
359 |
+
"step": 1250
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.12588862559241706,
|
363 |
+
"grad_norm": 0.23134127259254456,
|
364 |
+
"learning_rate": 3.8456151930719523e-05,
|
365 |
+
"loss": 0.1139,
|
366 |
+
"step": 1275
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 0.1283570300157978,
|
370 |
+
"grad_norm": 0.7565503716468811,
|
371 |
+
"learning_rate": 3.839584574218613e-05,
|
372 |
+
"loss": 0.2011,
|
373 |
+
"step": 1300
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 0.13082543443917852,
|
377 |
+
"grad_norm": 0.6890569925308228,
|
378 |
+
"learning_rate": 3.833443331211857e-05,
|
379 |
+
"loss": 0.149,
|
380 |
+
"step": 1325
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.13329383886255924,
|
384 |
+
"grad_norm": 0.20076613128185272,
|
385 |
+
"learning_rate": 3.827191833357787e-05,
|
386 |
+
"loss": 0.1332,
|
387 |
+
"step": 1350
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.13576224328593997,
|
391 |
+
"grad_norm": 9.452065467834473,
|
392 |
+
"learning_rate": 3.820830456592723e-05,
|
393 |
+
"loss": 0.1293,
|
394 |
+
"step": 1375
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.1382306477093207,
|
398 |
+
"grad_norm": 3.855496644973755,
|
399 |
+
"learning_rate": 3.814359583460599e-05,
|
400 |
+
"loss": 0.1778,
|
401 |
+
"step": 1400
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.14069905213270142,
|
405 |
+
"grad_norm": 0.3668440878391266,
|
406 |
+
"learning_rate": 3.807779603089957e-05,
|
407 |
+
"loss": 0.2426,
|
408 |
+
"step": 1425
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 0.14316745655608215,
|
412 |
+
"grad_norm": 0.14778482913970947,
|
413 |
+
"learning_rate": 3.801090911170546e-05,
|
414 |
+
"loss": 0.1531,
|
415 |
+
"step": 1450
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 0.14563586097946288,
|
419 |
+
"grad_norm": 0.1513528823852539,
|
420 |
+
"learning_rate": 3.794293909929525e-05,
|
421 |
+
"loss": 0.1868,
|
422 |
+
"step": 1475
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.1481042654028436,
|
426 |
+
"grad_norm": 0.3444514274597168,
|
427 |
+
"learning_rate": 3.7873890081072826e-05,
|
428 |
+
"loss": 0.1259,
|
429 |
+
"step": 1500
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.15057266982622433,
|
433 |
+
"grad_norm": 0.07048847526311874,
|
434 |
+
"learning_rate": 3.780376620932847e-05,
|
435 |
+
"loss": 0.1702,
|
436 |
+
"step": 1525
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.15304107424960506,
|
440 |
+
"grad_norm": 0.1177651584148407,
|
441 |
+
"learning_rate": 3.7732571700989227e-05,
|
442 |
+
"loss": 0.0743,
|
443 |
+
"step": 1550
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.15550947867298578,
|
447 |
+
"grad_norm": 8.564634323120117,
|
448 |
+
"learning_rate": 3.7660310837365316e-05,
|
449 |
+
"loss": 0.2193,
|
450 |
+
"step": 1575
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 0.1579778830963665,
|
454 |
+
"grad_norm": 0.08044961839914322,
|
455 |
+
"learning_rate": 3.758698796389264e-05,
|
456 |
+
"loss": 0.1704,
|
457 |
+
"step": 1600
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 0.16044628751974724,
|
461 |
+
"grad_norm": 0.15453416109085083,
|
462 |
+
"learning_rate": 3.751260748987148e-05,
|
463 |
+
"loss": 0.146,
|
464 |
+
"step": 1625
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.16291469194312796,
|
468 |
+
"grad_norm": 0.13735274970531464,
|
469 |
+
"learning_rate": 3.743717388820137e-05,
|
470 |
+
"loss": 0.1552,
|
471 |
+
"step": 1650
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.1653830963665087,
|
475 |
+
"grad_norm": 1.526962399482727,
|
476 |
+
"learning_rate": 3.7360691695112104e-05,
|
477 |
+
"loss": 0.218,
|
478 |
+
"step": 1675
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.16785150078988942,
|
482 |
+
"grad_norm": 36.50160598754883,
|
483 |
+
"learning_rate": 3.72831655098909e-05,
|
484 |
+
"loss": 0.1425,
|
485 |
+
"step": 1700
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.17031990521327015,
|
489 |
+
"grad_norm": 14.496137619018555,
|
490 |
+
"learning_rate": 3.72045999946059e-05,
|
491 |
+
"loss": 0.1217,
|
492 |
+
"step": 1725
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 0.17278830963665087,
|
496 |
+
"grad_norm": 15.02768325805664,
|
497 |
+
"learning_rate": 3.712499987382574e-05,
|
498 |
+
"loss": 0.2745,
|
499 |
+
"step": 1750
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 0.1752567140600316,
|
503 |
+
"grad_norm": 0.9205107688903809,
|
504 |
+
"learning_rate": 3.7044369934335504e-05,
|
505 |
+
"loss": 0.1096,
|
506 |
+
"step": 1775
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.17772511848341233,
|
510 |
+
"grad_norm": 18.815643310546875,
|
511 |
+
"learning_rate": 3.696271502484882e-05,
|
512 |
+
"loss": 0.2392,
|
513 |
+
"step": 1800
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.18019352290679305,
|
517 |
+
"grad_norm": 29.416358947753906,
|
518 |
+
"learning_rate": 3.688004005571629e-05,
|
519 |
+
"loss": 0.1611,
|
520 |
+
"step": 1825
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.18266192733017378,
|
524 |
+
"grad_norm": 23.464696884155273,
|
525 |
+
"learning_rate": 3.679634999863023e-05,
|
526 |
+
"loss": 0.1987,
|
527 |
+
"step": 1850
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 0.1851303317535545,
|
531 |
+
"grad_norm": 0.21536731719970703,
|
532 |
+
"learning_rate": 3.6711649886325644e-05,
|
533 |
+
"loss": 0.1346,
|
534 |
+
"step": 1875
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.18759873617693523,
|
538 |
+
"grad_norm": 11.226929664611816,
|
539 |
+
"learning_rate": 3.662594481227764e-05,
|
540 |
+
"loss": 0.185,
|
541 |
+
"step": 1900
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 0.19006714060031596,
|
545 |
+
"grad_norm": 11.212834358215332,
|
546 |
+
"learning_rate": 3.6539239930395095e-05,
|
547 |
+
"loss": 0.1701,
|
548 |
+
"step": 1925
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.19253554502369669,
|
552 |
+
"grad_norm": 0.18225641548633575,
|
553 |
+
"learning_rate": 3.6451540454710716e-05,
|
554 |
+
"loss": 0.1046,
|
555 |
+
"step": 1950
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.1950039494470774,
|
559 |
+
"grad_norm": 27.281253814697266,
|
560 |
+
"learning_rate": 3.6362851659067523e-05,
|
561 |
+
"loss": 0.1553,
|
562 |
+
"step": 1975
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.19747235387045814,
|
566 |
+
"grad_norm": 43.26043701171875,
|
567 |
+
"learning_rate": 3.6273178876801686e-05,
|
568 |
+
"loss": 0.2103,
|
569 |
+
"step": 2000
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 0.19994075829383887,
|
573 |
+
"grad_norm": 5.702709197998047,
|
574 |
+
"learning_rate": 3.618252750042178e-05,
|
575 |
+
"loss": 0.1443,
|
576 |
+
"step": 2025
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 0.2024091627172196,
|
580 |
+
"grad_norm": 0.09359689801931381,
|
581 |
+
"learning_rate": 3.6090902981284555e-05,
|
582 |
+
"loss": 0.2281,
|
583 |
+
"step": 2050
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 0.20487756714060032,
|
587 |
+
"grad_norm": 1.812398076057434,
|
588 |
+
"learning_rate": 3.599831082926707e-05,
|
589 |
+
"loss": 0.2073,
|
590 |
+
"step": 2075
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.20734597156398105,
|
594 |
+
"grad_norm": 8.140066146850586,
|
595 |
+
"learning_rate": 3.590475661243541e-05,
|
596 |
+
"loss": 0.1236,
|
597 |
+
"step": 2100
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.20981437598736177,
|
601 |
+
"grad_norm": 11.266165733337402,
|
602 |
+
"learning_rate": 3.581024595670976e-05,
|
603 |
+
"loss": 0.1573,
|
604 |
+
"step": 2125
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.2122827804107425,
|
608 |
+
"grad_norm": 0.5063501000404358,
|
609 |
+
"learning_rate": 3.571478454552618e-05,
|
610 |
+
"loss": 0.1536,
|
611 |
+
"step": 2150
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 0.21475118483412323,
|
615 |
+
"grad_norm": 3.7616796493530273,
|
616 |
+
"learning_rate": 3.5618378119494764e-05,
|
617 |
+
"loss": 0.1806,
|
618 |
+
"step": 2175
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 0.21721958925750395,
|
622 |
+
"grad_norm": 0.5944362282752991,
|
623 |
+
"learning_rate": 3.5521032476054466e-05,
|
624 |
+
"loss": 0.2131,
|
625 |
+
"step": 2200
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 0.21968799368088468,
|
629 |
+
"grad_norm": 0.20885775983333588,
|
630 |
+
"learning_rate": 3.5422753469124455e-05,
|
631 |
+
"loss": 0.1476,
|
632 |
+
"step": 2225
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.2221563981042654,
|
636 |
+
"grad_norm": 20.5062313079834,
|
637 |
+
"learning_rate": 3.532354700875208e-05,
|
638 |
+
"loss": 0.1353,
|
639 |
+
"step": 2250
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.22462480252764613,
|
643 |
+
"grad_norm": 8.504454612731934,
|
644 |
+
"learning_rate": 3.522341906075748e-05,
|
645 |
+
"loss": 0.1667,
|
646 |
+
"step": 2275
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 0.22709320695102686,
|
650 |
+
"grad_norm": 12.637554168701172,
|
651 |
+
"learning_rate": 3.512237564637481e-05,
|
652 |
+
"loss": 0.1531,
|
653 |
+
"step": 2300
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 0.2295616113744076,
|
657 |
+
"grad_norm": 24.210155487060547,
|
658 |
+
"learning_rate": 3.502042284189017e-05,
|
659 |
+
"loss": 0.2414,
|
660 |
+
"step": 2325
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 0.2320300157977883,
|
664 |
+
"grad_norm": 13.906551361083984,
|
665 |
+
"learning_rate": 3.49175667782762e-05,
|
666 |
+
"loss": 0.1339,
|
667 |
+
"step": 2350
|
668 |
+
},
|
669 |
+
{
|
670 |
+
"epoch": 0.23449842022116904,
|
671 |
+
"grad_norm": 29.54754638671875,
|
672 |
+
"learning_rate": 3.481381364082337e-05,
|
673 |
+
"loss": 0.1491,
|
674 |
+
"step": 2375
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.23696682464454977,
|
678 |
+
"grad_norm": 0.4684959948062897,
|
679 |
+
"learning_rate": 3.470916966876807e-05,
|
680 |
+
"loss": 0.0583,
|
681 |
+
"step": 2400
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 0.2394352290679305,
|
685 |
+
"grad_norm": 0.05459192395210266,
|
686 |
+
"learning_rate": 3.4603641154917367e-05,
|
687 |
+
"loss": 0.1521,
|
688 |
+
"step": 2425
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 0.24190363349131122,
|
692 |
+
"grad_norm": 25.561431884765625,
|
693 |
+
"learning_rate": 3.4497234445270616e-05,
|
694 |
+
"loss": 0.1607,
|
695 |
+
"step": 2450
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 0.24437203791469195,
|
699 |
+
"grad_norm": 0.0625099316239357,
|
700 |
+
"learning_rate": 3.438995593863781e-05,
|
701 |
+
"loss": 0.0481,
|
702 |
+
"step": 2475
|
703 |
+
},
|
704 |
+
{
|
705 |
+
"epoch": 0.24684044233807267,
|
706 |
+
"grad_norm": 0.038314227014780045,
|
707 |
+
"learning_rate": 3.428181208625481e-05,
|
708 |
+
"loss": 0.1209,
|
709 |
+
"step": 2500
|
710 |
+
},
|
711 |
+
{
|
712 |
+
"epoch": 0.2493088467614534,
|
713 |
+
"grad_norm": 27.349132537841797,
|
714 |
+
"learning_rate": 3.417280939139539e-05,
|
715 |
+
"loss": 0.1241,
|
716 |
+
"step": 2525
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.2517772511848341,
|
720 |
+
"grad_norm": 0.495726615190506,
|
721 |
+
"learning_rate": 3.406295440898017e-05,
|
722 |
+
"loss": 0.1787,
|
723 |
+
"step": 2550
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 0.2542456556082148,
|
727 |
+
"grad_norm": 27.391502380371094,
|
728 |
+
"learning_rate": 3.39522537451824e-05,
|
729 |
+
"loss": 0.1597,
|
730 |
+
"step": 2575
|
731 |
+
},
|
732 |
+
{
|
733 |
+
"epoch": 0.2567140600315956,
|
734 |
+
"grad_norm": 20.175466537475586,
|
735 |
+
"learning_rate": 3.3840714057030736e-05,
|
736 |
+
"loss": 0.2188,
|
737 |
+
"step": 2600
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 0.2591824644549763,
|
741 |
+
"grad_norm": 0.2197282463312149,
|
742 |
+
"learning_rate": 3.37283420520089e-05,
|
743 |
+
"loss": 0.0971,
|
744 |
+
"step": 2625
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 0.26165086887835703,
|
748 |
+
"grad_norm": 9.926712036132812,
|
749 |
+
"learning_rate": 3.36151444876523e-05,
|
750 |
+
"loss": 0.1843,
|
751 |
+
"step": 2650
|
752 |
+
},
|
753 |
+
{
|
754 |
+
"epoch": 0.26411927330173773,
|
755 |
+
"grad_norm": 2.8893277645111084,
|
756 |
+
"learning_rate": 3.350112817114171e-05,
|
757 |
+
"loss": 0.1383,
|
758 |
+
"step": 2675
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 0.2665876777251185,
|
762 |
+
"grad_norm": 0.20964030921459198,
|
763 |
+
"learning_rate": 3.338629995889388e-05,
|
764 |
+
"loss": 0.2184,
|
765 |
+
"step": 2700
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 0.2690560821484992,
|
769 |
+
"grad_norm": 15.719782829284668,
|
770 |
+
"learning_rate": 3.327066675614924e-05,
|
771 |
+
"loss": 0.1184,
|
772 |
+
"step": 2725
|
773 |
+
},
|
774 |
+
{
|
775 |
+
"epoch": 0.27152448657187994,
|
776 |
+
"grad_norm": 9.262935638427734,
|
777 |
+
"learning_rate": 3.3154235516556616e-05,
|
778 |
+
"loss": 0.1662,
|
779 |
+
"step": 2750
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 0.27399289099526064,
|
783 |
+
"grad_norm": 7.443912982940674,
|
784 |
+
"learning_rate": 3.303701324175516e-05,
|
785 |
+
"loss": 0.1864,
|
786 |
+
"step": 2775
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 0.2764612954186414,
|
790 |
+
"grad_norm": 8.308416366577148,
|
791 |
+
"learning_rate": 3.291900698095318e-05,
|
792 |
+
"loss": 0.1334,
|
793 |
+
"step": 2800
|
794 |
+
},
|
795 |
+
{
|
796 |
+
"epoch": 0.2789296998420221,
|
797 |
+
"grad_norm": 0.7907429337501526,
|
798 |
+
"learning_rate": 3.280022383050433e-05,
|
799 |
+
"loss": 0.1036,
|
800 |
+
"step": 2825
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 0.28139810426540285,
|
804 |
+
"grad_norm": 20.742170333862305,
|
805 |
+
"learning_rate": 3.2680670933480816e-05,
|
806 |
+
"loss": 0.1287,
|
807 |
+
"step": 2850
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 0.28386650868878355,
|
811 |
+
"grad_norm": 9.604803085327148,
|
812 |
+
"learning_rate": 3.2560355479243894e-05,
|
813 |
+
"loss": 0.1957,
|
814 |
+
"step": 2875
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 0.2863349131121643,
|
818 |
+
"grad_norm": 0.0466698594391346,
|
819 |
+
"learning_rate": 3.243928470301146e-05,
|
820 |
+
"loss": 0.1017,
|
821 |
+
"step": 2900
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 0.288803317535545,
|
825 |
+
"grad_norm": 0.19196811318397522,
|
826 |
+
"learning_rate": 3.231746588542301e-05,
|
827 |
+
"loss": 0.1676,
|
828 |
+
"step": 2925
|
829 |
+
},
|
830 |
+
{
|
831 |
+
"epoch": 0.29127172195892576,
|
832 |
+
"grad_norm": 0.2808176875114441,
|
833 |
+
"learning_rate": 3.219490635210184e-05,
|
834 |
+
"loss": 0.1411,
|
835 |
+
"step": 2950
|
836 |
+
},
|
837 |
+
{
|
838 |
+
"epoch": 0.29374012638230645,
|
839 |
+
"grad_norm": 9.712594032287598,
|
840 |
+
"learning_rate": 3.207161347321444e-05,
|
841 |
+
"loss": 0.0937,
|
842 |
+
"step": 2975
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 0.2962085308056872,
|
846 |
+
"grad_norm": 79.80085754394531,
|
847 |
+
"learning_rate": 3.194759466302735e-05,
|
848 |
+
"loss": 0.1824,
|
849 |
+
"step": 3000
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.2986769352290679,
|
853 |
+
"grad_norm": 1.4917582273483276,
|
854 |
+
"learning_rate": 3.182285737946128e-05,
|
855 |
+
"loss": 0.1133,
|
856 |
+
"step": 3025
|
857 |
+
},
|
858 |
+
{
|
859 |
+
"epoch": 0.30114533965244866,
|
860 |
+
"grad_norm": 0.8260819911956787,
|
861 |
+
"learning_rate": 3.169740912364261e-05,
|
862 |
+
"loss": 0.0385,
|
863 |
+
"step": 3050
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 0.30361374407582936,
|
867 |
+
"grad_norm": 0.3458572030067444,
|
868 |
+
"learning_rate": 3.1571257439452344e-05,
|
869 |
+
"loss": 0.1896,
|
870 |
+
"step": 3075
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 0.3060821484992101,
|
874 |
+
"grad_norm": 0.12934423983097076,
|
875 |
+
"learning_rate": 3.14444099130724e-05,
|
876 |
+
"loss": 0.1164,
|
877 |
+
"step": 3100
|
878 |
+
},
|
879 |
+
{
|
880 |
+
"epoch": 0.3085505529225908,
|
881 |
+
"grad_norm": 0.23791643977165222,
|
882 |
+
"learning_rate": 3.131687417252947e-05,
|
883 |
+
"loss": 0.1856,
|
884 |
+
"step": 3125
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 0.31101895734597157,
|
888 |
+
"grad_norm": 0.05154755339026451,
|
889 |
+
"learning_rate": 3.118865788723627e-05,
|
890 |
+
"loss": 0.1771,
|
891 |
+
"step": 3150
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 0.31348736176935227,
|
895 |
+
"grad_norm": 8.543177604675293,
|
896 |
+
"learning_rate": 3.1059768767530356e-05,
|
897 |
+
"loss": 0.2466,
|
898 |
+
"step": 3175
|
899 |
+
},
|
900 |
+
{
|
901 |
+
"epoch": 0.315955766192733,
|
902 |
+
"grad_norm": 0.18959984183311462,
|
903 |
+
"learning_rate": 3.093021456421044e-05,
|
904 |
+
"loss": 0.1271,
|
905 |
+
"step": 3200
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 0.3184241706161137,
|
909 |
+
"grad_norm": 0.314675897359848,
|
910 |
+
"learning_rate": 3.080000306807028e-05,
|
911 |
+
"loss": 0.1544,
|
912 |
+
"step": 3225
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 0.3208925750394945,
|
916 |
+
"grad_norm": 0.13474613428115845,
|
917 |
+
"learning_rate": 3.0669142109430256e-05,
|
918 |
+
"loss": 0.1841,
|
919 |
+
"step": 3250
|
920 |
+
},
|
921 |
+
{
|
922 |
+
"epoch": 0.3233609794628752,
|
923 |
+
"grad_norm": 0.38175496459007263,
|
924 |
+
"learning_rate": 3.053763955766639e-05,
|
925 |
+
"loss": 0.1152,
|
926 |
+
"step": 3275
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 0.32582938388625593,
|
930 |
+
"grad_norm": 1.015726923942566,
|
931 |
+
"learning_rate": 3.0405503320737188e-05,
|
932 |
+
"loss": 0.19,
|
933 |
+
"step": 3300
|
934 |
+
},
|
935 |
+
{
|
936 |
+
"epoch": 0.32829778830963663,
|
937 |
+
"grad_norm": 9.585470199584961,
|
938 |
+
"learning_rate": 3.0272741344708038e-05,
|
939 |
+
"loss": 0.1449,
|
940 |
+
"step": 3325
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 0.3307661927330174,
|
944 |
+
"grad_norm": 24.776342391967773,
|
945 |
+
"learning_rate": 3.013936161327343e-05,
|
946 |
+
"loss": 0.1489,
|
947 |
+
"step": 3350
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 0.3332345971563981,
|
951 |
+
"grad_norm": 0.8524392247200012,
|
952 |
+
"learning_rate": 3.0005372147276803e-05,
|
953 |
+
"loss": 0.1357,
|
954 |
+
"step": 3375
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 0.33570300157977884,
|
958 |
+
"grad_norm": 27.355451583862305,
|
959 |
+
"learning_rate": 2.9870781004228236e-05,
|
960 |
+
"loss": 0.2057,
|
961 |
+
"step": 3400
|
962 |
+
},
|
963 |
+
{
|
964 |
+
"epoch": 0.33817140600315954,
|
965 |
+
"grad_norm": 11.425800323486328,
|
966 |
+
"learning_rate": 2.9735596277819894e-05,
|
967 |
+
"loss": 0.1979,
|
968 |
+
"step": 3425
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 0.3406398104265403,
|
972 |
+
"grad_norm": 20.804515838623047,
|
973 |
+
"learning_rate": 2.959982609743931e-05,
|
974 |
+
"loss": 0.178,
|
975 |
+
"step": 3450
|
976 |
+
},
|
977 |
+
{
|
978 |
+
"epoch": 0.343108214849921,
|
979 |
+
"grad_norm": 17.747562408447266,
|
980 |
+
"learning_rate": 2.946347862768055e-05,
|
981 |
+
"loss": 0.1787,
|
982 |
+
"step": 3475
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 0.34557661927330174,
|
986 |
+
"grad_norm": 7.701132774353027,
|
987 |
+
"learning_rate": 2.932656206785316e-05,
|
988 |
+
"loss": 0.1725,
|
989 |
+
"step": 3500
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 0.34804502369668244,
|
993 |
+
"grad_norm": 10.528812408447266,
|
994 |
+
"learning_rate": 2.9189084651489196e-05,
|
995 |
+
"loss": 0.1427,
|
996 |
+
"step": 3525
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 0.3505134281200632,
|
1000 |
+
"grad_norm": 0.20533309876918793,
|
1001 |
+
"learning_rate": 2.905105464584801e-05,
|
1002 |
+
"loss": 0.0307,
|
1003 |
+
"step": 3550
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 0.3529818325434439,
|
1007 |
+
"grad_norm": 20.58428382873535,
|
1008 |
+
"learning_rate": 2.8912480351419154e-05,
|
1009 |
+
"loss": 0.1898,
|
1010 |
+
"step": 3575
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 0.35545023696682465,
|
1014 |
+
"grad_norm": 23.746784210205078,
|
1015 |
+
"learning_rate": 2.8773370101423186e-05,
|
1016 |
+
"loss": 0.1021,
|
1017 |
+
"step": 3600
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 0.35791864139020535,
|
1021 |
+
"grad_norm": 0.4287678003311157,
|
1022 |
+
"learning_rate": 2.8633732261310583e-05,
|
1023 |
+
"loss": 0.1189,
|
1024 |
+
"step": 3625
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 0.3603870458135861,
|
1028 |
+
"grad_norm": 55.95423126220703,
|
1029 |
+
"learning_rate": 2.849357522825864e-05,
|
1030 |
+
"loss": 0.1631,
|
1031 |
+
"step": 3650
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 0.3628554502369668,
|
1035 |
+
"grad_norm": 0.253450870513916,
|
1036 |
+
"learning_rate": 2.8352907430666567e-05,
|
1037 |
+
"loss": 0.0646,
|
1038 |
+
"step": 3675
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 0.36532385466034756,
|
1042 |
+
"grad_norm": 0.09365226328372955,
|
1043 |
+
"learning_rate": 2.821173732764855e-05,
|
1044 |
+
"loss": 0.1848,
|
1045 |
+
"step": 3700
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 0.36779225908372826,
|
1049 |
+
"grad_norm": 0.14286532998085022,
|
1050 |
+
"learning_rate": 2.8070073408525166e-05,
|
1051 |
+
"loss": 0.1341,
|
1052 |
+
"step": 3725
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 0.370260663507109,
|
1056 |
+
"grad_norm": 0.4148150384426117,
|
1057 |
+
"learning_rate": 2.7927924192312783e-05,
|
1058 |
+
"loss": 0.0918,
|
1059 |
+
"step": 3750
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 0.3727290679304897,
|
1063 |
+
"grad_norm": 0.5170570611953735,
|
1064 |
+
"learning_rate": 2.778529822721132e-05,
|
1065 |
+
"loss": 0.1813,
|
1066 |
+
"step": 3775
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 0.37519747235387046,
|
1070 |
+
"grad_norm": 26.20146942138672,
|
1071 |
+
"learning_rate": 2.7642204090090176e-05,
|
1072 |
+
"loss": 0.1674,
|
1073 |
+
"step": 3800
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 0.37766587677725116,
|
1077 |
+
"grad_norm": 0.08255212008953094,
|
1078 |
+
"learning_rate": 2.7498650385972468e-05,
|
1079 |
+
"loss": 0.1103,
|
1080 |
+
"step": 3825
|
1081 |
+
},
|
1082 |
+
{
|
1083 |
+
"epoch": 0.3801342812006319,
|
1084 |
+
"grad_norm": 21.388195037841797,
|
1085 |
+
"learning_rate": 2.7354645747517547e-05,
|
1086 |
+
"loss": 0.1157,
|
1087 |
+
"step": 3850
|
1088 |
+
},
|
1089 |
+
{
|
1090 |
+
"epoch": 0.3826026856240126,
|
1091 |
+
"grad_norm": 0.18762463331222534,
|
1092 |
+
"learning_rate": 2.7210198834501895e-05,
|
1093 |
+
"loss": 0.1229,
|
1094 |
+
"step": 3875
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 0.38507109004739337,
|
1098 |
+
"grad_norm": 1.5140018463134766,
|
1099 |
+
"learning_rate": 2.706531833329833e-05,
|
1100 |
+
"loss": 0.1983,
|
1101 |
+
"step": 3900
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 0.38753949447077407,
|
1105 |
+
"grad_norm": 0.5415325164794922,
|
1106 |
+
"learning_rate": 2.6920012956353695e-05,
|
1107 |
+
"loss": 0.1355,
|
1108 |
+
"step": 3925
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 0.3900078988941548,
|
1112 |
+
"grad_norm": 0.9165463447570801,
|
1113 |
+
"learning_rate": 2.6774291441664887e-05,
|
1114 |
+
"loss": 0.1356,
|
1115 |
+
"step": 3950
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 0.3924763033175355,
|
1119 |
+
"grad_norm": 9.647566795349121,
|
1120 |
+
"learning_rate": 2.662816255225342e-05,
|
1121 |
+
"loss": 0.0817,
|
1122 |
+
"step": 3975
|
1123 |
+
},
|
1124 |
+
{
|
1125 |
+
"epoch": 0.3949447077409163,
|
1126 |
+
"grad_norm": 0.28667527437210083,
|
1127 |
+
"learning_rate": 2.648163507563844e-05,
|
1128 |
+
"loss": 0.141,
|
1129 |
+
"step": 4000
|
1130 |
+
},
|
1131 |
+
{
|
1132 |
+
"epoch": 0.397413112164297,
|
1133 |
+
"grad_norm": 0.21459440886974335,
|
1134 |
+
"learning_rate": 2.6334717823308317e-05,
|
1135 |
+
"loss": 0.2223,
|
1136 |
+
"step": 4025
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 0.39988151658767773,
|
1140 |
+
"grad_norm": 0.22389402985572815,
|
1141 |
+
"learning_rate": 2.6187419630190708e-05,
|
1142 |
+
"loss": 0.1895,
|
1143 |
+
"step": 4050
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 0.40234992101105843,
|
1147 |
+
"grad_norm": 0.5128334164619446,
|
1148 |
+
"learning_rate": 2.603974935412132e-05,
|
1149 |
+
"loss": 0.136,
|
1150 |
+
"step": 4075
|
1151 |
+
},
|
1152 |
+
{
|
1153 |
+
"epoch": 0.4048183254344392,
|
1154 |
+
"grad_norm": 0.2022058516740799,
|
1155 |
+
"learning_rate": 2.589171587531121e-05,
|
1156 |
+
"loss": 0.1557,
|
1157 |
+
"step": 4100
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 0.4072867298578199,
|
1161 |
+
"grad_norm": 0.11499619483947754,
|
1162 |
+
"learning_rate": 2.574332809581277e-05,
|
1163 |
+
"loss": 0.1844,
|
1164 |
+
"step": 4125
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 0.40975513428120064,
|
1168 |
+
"grad_norm": 17.62612533569336,
|
1169 |
+
"learning_rate": 2.559459493898442e-05,
|
1170 |
+
"loss": 0.1266,
|
1171 |
+
"step": 4150
|
1172 |
+
},
|
1173 |
+
{
|
1174 |
+
"epoch": 0.41222353870458134,
|
1175 |
+
"grad_norm": 0.7205790877342224,
|
1176 |
+
"learning_rate": 2.5445525348953973e-05,
|
1177 |
+
"loss": 0.1402,
|
1178 |
+
"step": 4175
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 0.4146919431279621,
|
1182 |
+
"grad_norm": 10.904681205749512,
|
1183 |
+
"learning_rate": 2.5296128290080785e-05,
|
1184 |
+
"loss": 0.1011,
|
1185 |
+
"step": 4200
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 0.4171603475513428,
|
1189 |
+
"grad_norm": 25.37704849243164,
|
1190 |
+
"learning_rate": 2.5146412746416692e-05,
|
1191 |
+
"loss": 0.1135,
|
1192 |
+
"step": 4225
|
1193 |
+
},
|
1194 |
+
{
|
1195 |
+
"epoch": 0.41962875197472355,
|
1196 |
+
"grad_norm": 9.92044734954834,
|
1197 |
+
"learning_rate": 2.4996387721165733e-05,
|
1198 |
+
"loss": 0.1493,
|
1199 |
+
"step": 4250
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 0.42209715639810425,
|
1203 |
+
"grad_norm": 22.477872848510742,
|
1204 |
+
"learning_rate": 2.484606223614274e-05,
|
1205 |
+
"loss": 0.1735,
|
1206 |
+
"step": 4275
|
1207 |
+
},
|
1208 |
+
{
|
1209 |
+
"epoch": 0.424565560821485,
|
1210 |
+
"grad_norm": 40.62971115112305,
|
1211 |
+
"learning_rate": 2.469544533123083e-05,
|
1212 |
+
"loss": 0.1965,
|
1213 |
+
"step": 4300
|
1214 |
+
},
|
1215 |
+
{
|
1216 |
+
"epoch": 0.4270339652448657,
|
1217 |
+
"grad_norm": 1.8471810817718506,
|
1218 |
+
"learning_rate": 2.454454606383776e-05,
|
1219 |
+
"loss": 0.1341,
|
1220 |
+
"step": 4325
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 0.42950236966824645,
|
1224 |
+
"grad_norm": 9.519583702087402,
|
1225 |
+
"learning_rate": 2.4393373508351267e-05,
|
1226 |
+
"loss": 0.1691,
|
1227 |
+
"step": 4350
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 0.43197077409162715,
|
1231 |
+
"grad_norm": 0.3392331898212433,
|
1232 |
+
"learning_rate": 2.4241936755593383e-05,
|
1233 |
+
"loss": 0.1352,
|
1234 |
+
"step": 4375
|
1235 |
+
},
|
1236 |
+
{
|
1237 |
+
"epoch": 0.4344391785150079,
|
1238 |
+
"grad_norm": 0.06101216748356819,
|
1239 |
+
"learning_rate": 2.409024491227374e-05,
|
1240 |
+
"loss": 0.0794,
|
1241 |
+
"step": 4400
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 0.4369075829383886,
|
1245 |
+
"grad_norm": 0.21343450248241425,
|
1246 |
+
"learning_rate": 2.393830710044195e-05,
|
1247 |
+
"loss": 0.1458,
|
1248 |
+
"step": 4425
|
1249 |
+
},
|
1250 |
+
{
|
1251 |
+
"epoch": 0.43937598736176936,
|
1252 |
+
"grad_norm": 0.1489739567041397,
|
1253 |
+
"learning_rate": 2.3786132456939042e-05,
|
1254 |
+
"loss": 0.2196,
|
1255 |
+
"step": 4450
|
1256 |
+
},
|
1257 |
+
{
|
1258 |
+
"epoch": 0.44184439178515006,
|
1259 |
+
"grad_norm": 12.579410552978516,
|
1260 |
+
"learning_rate": 2.3633730132848003e-05,
|
1261 |
+
"loss": 0.1256,
|
1262 |
+
"step": 4475
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 0.4443127962085308,
|
1266 |
+
"grad_norm": 0.3262731432914734,
|
1267 |
+
"learning_rate": 2.3481109292943484e-05,
|
1268 |
+
"loss": 0.0912,
|
1269 |
+
"step": 4500
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 0.4467812006319115,
|
1273 |
+
"grad_norm": 43.45631408691406,
|
1274 |
+
"learning_rate": 2.332827911514069e-05,
|
1275 |
+
"loss": 0.1313,
|
1276 |
+
"step": 4525
|
1277 |
+
},
|
1278 |
+
{
|
1279 |
+
"epoch": 0.44924960505529227,
|
1280 |
+
"grad_norm": 16.735519409179688,
|
1281 |
+
"learning_rate": 2.317524878994341e-05,
|
1282 |
+
"loss": 0.1488,
|
1283 |
+
"step": 4550
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 0.45171800947867297,
|
1287 |
+
"grad_norm": 9.294239044189453,
|
1288 |
+
"learning_rate": 2.3022027519891416e-05,
|
1289 |
+
"loss": 0.1348,
|
1290 |
+
"step": 4575
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 0.4541864139020537,
|
1294 |
+
"grad_norm": 0.6957706212997437,
|
1295 |
+
"learning_rate": 2.2868624519006996e-05,
|
1296 |
+
"loss": 0.1681,
|
1297 |
+
"step": 4600
|
1298 |
+
},
|
1299 |
+
{
|
1300 |
+
"epoch": 0.4566548183254344,
|
1301 |
+
"grad_norm": 0.28600016236305237,
|
1302 |
+
"learning_rate": 2.2715049012240896e-05,
|
1303 |
+
"loss": 0.1565,
|
1304 |
+
"step": 4625
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 0.4591232227488152,
|
1308 |
+
"grad_norm": 25.832984924316406,
|
1309 |
+
"learning_rate": 2.2561310234917592e-05,
|
1310 |
+
"loss": 0.0994,
|
1311 |
+
"step": 4650
|
1312 |
+
},
|
1313 |
+
{
|
1314 |
+
"epoch": 0.4615916271721959,
|
1315 |
+
"grad_norm": 8.173675537109375,
|
1316 |
+
"learning_rate": 2.240741743217989e-05,
|
1317 |
+
"loss": 0.0612,
|
1318 |
+
"step": 4675
|
1319 |
+
},
|
1320 |
+
{
|
1321 |
+
"epoch": 0.4640600315955766,
|
1322 |
+
"grad_norm": 0.04298144951462746,
|
1323 |
+
"learning_rate": 2.225337985843296e-05,
|
1324 |
+
"loss": 0.1034,
|
1325 |
+
"step": 4700
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 0.4665284360189573,
|
1329 |
+
"grad_norm": 19.630935668945312,
|
1330 |
+
"learning_rate": 2.2099206776787858e-05,
|
1331 |
+
"loss": 0.1196,
|
1332 |
+
"step": 4725
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 0.4689968404423381,
|
1336 |
+
"grad_norm": 0.23587805032730103,
|
1337 |
+
"learning_rate": 2.1944907458504447e-05,
|
1338 |
+
"loss": 0.1375,
|
1339 |
+
"step": 4750
|
1340 |
+
},
|
1341 |
+
{
|
1342 |
+
"epoch": 0.4714652448657188,
|
1343 |
+
"grad_norm": 0.3975267708301544,
|
1344 |
+
"learning_rate": 2.1790491182433895e-05,
|
1345 |
+
"loss": 0.0924,
|
1346 |
+
"step": 4775
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 0.47393364928909953,
|
1350 |
+
"grad_norm": 12.13842487335205,
|
1351 |
+
"learning_rate": 2.163596723446065e-05,
|
1352 |
+
"loss": 0.1235,
|
1353 |
+
"step": 4800
|
1354 |
+
},
|
1355 |
+
{
|
1356 |
+
"epoch": 0.47640205371248023,
|
1357 |
+
"grad_norm": 0.10228073596954346,
|
1358 |
+
"learning_rate": 2.1481344906944076e-05,
|
1359 |
+
"loss": 0.1097,
|
1360 |
+
"step": 4825
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 0.478870458135861,
|
1364 |
+
"grad_norm": 0.035756584256887436,
|
1365 |
+
"learning_rate": 2.1326633498159626e-05,
|
1366 |
+
"loss": 0.1655,
|
1367 |
+
"step": 4850
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 0.4813388625592417,
|
1371 |
+
"grad_norm": 0.11663559079170227,
|
1372 |
+
"learning_rate": 2.1171842311739676e-05,
|
1373 |
+
"loss": 0.1441,
|
1374 |
+
"step": 4875
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 0.48380726698262244,
|
1378 |
+
"grad_norm": 0.14211098849773407,
|
1379 |
+
"learning_rate": 2.1016980656114066e-05,
|
1380 |
+
"loss": 0.1296,
|
1381 |
+
"step": 4900
|
1382 |
+
},
|
1383 |
+
{
|
1384 |
+
"epoch": 0.48627567140600314,
|
1385 |
+
"grad_norm": 16.92612075805664,
|
1386 |
+
"learning_rate": 2.0862057843950354e-05,
|
1387 |
+
"loss": 0.1178,
|
1388 |
+
"step": 4925
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 0.4887440758293839,
|
1392 |
+
"grad_norm": 0.1619480550289154,
|
1393 |
+
"learning_rate": 2.070708319159374e-05,
|
1394 |
+
"loss": 0.1603,
|
1395 |
+
"step": 4950
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 0.4912124802527646,
|
1399 |
+
"grad_norm": 0.18450067937374115,
|
1400 |
+
"learning_rate": 2.055206601850687e-05,
|
1401 |
+
"loss": 0.1771,
|
1402 |
+
"step": 4975
|
1403 |
+
},
|
1404 |
+
{
|
1405 |
+
"epoch": 0.49368088467614535,
|
1406 |
+
"grad_norm": 20.05467987060547,
|
1407 |
+
"learning_rate": 2.0397015646709383e-05,
|
1408 |
+
"loss": 0.1535,
|
1409 |
+
"step": 5000
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 0.49614928909952605,
|
1413 |
+
"grad_norm": 0.5198606848716736,
|
1414 |
+
"learning_rate": 2.0241941400217347e-05,
|
1415 |
+
"loss": 0.0883,
|
1416 |
+
"step": 5025
|
1417 |
+
},
|
1418 |
+
{
|
1419 |
+
"epoch": 0.4986176935229068,
|
1420 |
+
"grad_norm": 15.077300071716309,
|
1421 |
+
"learning_rate": 2.0086852604482533e-05,
|
1422 |
+
"loss": 0.1074,
|
1423 |
+
"step": 5050
|
1424 |
+
},
|
1425 |
+
{
|
1426 |
+
"epoch": 0.5010860979462876,
|
1427 |
+
"grad_norm": 2.540178060531616,
|
1428 |
+
"learning_rate": 1.9931758585831645e-05,
|
1429 |
+
"loss": 0.1785,
|
1430 |
+
"step": 5075
|
1431 |
+
},
|
1432 |
+
{
|
1433 |
+
"epoch": 0.5035545023696683,
|
1434 |
+
"grad_norm": 14.167303085327148,
|
1435 |
+
"learning_rate": 1.977666867090547e-05,
|
1436 |
+
"loss": 0.1443,
|
1437 |
+
"step": 5100
|
1438 |
+
},
|
1439 |
+
{
|
1440 |
+
"epoch": 0.506022906793049,
|
1441 |
+
"grad_norm": 13.333304405212402,
|
1442 |
+
"learning_rate": 1.9621592186098004e-05,
|
1443 |
+
"loss": 0.1258,
|
1444 |
+
"step": 5125
|
1445 |
+
},
|
1446 |
+
{
|
1447 |
+
"epoch": 0.5084913112164297,
|
1448 |
+
"grad_norm": 0.6414194703102112,
|
1449 |
+
"learning_rate": 1.9466538456995632e-05,
|
1450 |
+
"loss": 0.1219,
|
1451 |
+
"step": 5150
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 0.5109597156398105,
|
1455 |
+
"grad_norm": 14.161781311035156,
|
1456 |
+
"learning_rate": 1.9311516807816298e-05,
|
1457 |
+
"loss": 0.0892,
|
1458 |
+
"step": 5175
|
1459 |
+
},
|
1460 |
+
{
|
1461 |
+
"epoch": 0.5134281200631912,
|
1462 |
+
"grad_norm": 0.925495982170105,
|
1463 |
+
"learning_rate": 1.9156536560848826e-05,
|
1464 |
+
"loss": 0.1175,
|
1465 |
+
"step": 5200
|
1466 |
+
},
|
1467 |
+
{
|
1468 |
+
"epoch": 0.5158965244865719,
|
1469 |
+
"grad_norm": 1.5407062768936157,
|
1470 |
+
"learning_rate": 1.9001607035892296e-05,
|
1471 |
+
"loss": 0.1671,
|
1472 |
+
"step": 5225
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 0.5183649289099526,
|
1476 |
+
"grad_norm": 0.19095449149608612,
|
1477 |
+
"learning_rate": 1.8846737549695592e-05,
|
1478 |
+
"loss": 0.0837,
|
1479 |
+
"step": 5250
|
1480 |
+
},
|
1481 |
+
{
|
1482 |
+
"epoch": 0.5208333333333334,
|
1483 |
+
"grad_norm": 0.05522017180919647,
|
1484 |
+
"learning_rate": 1.869193741539714e-05,
|
1485 |
+
"loss": 0.1419,
|
1486 |
+
"step": 5275
|
1487 |
+
},
|
1488 |
+
{
|
1489 |
+
"epoch": 0.5233017377567141,
|
1490 |
+
"grad_norm": 0.21493327617645264,
|
1491 |
+
"learning_rate": 1.8537215941964887e-05,
|
1492 |
+
"loss": 0.1701,
|
1493 |
+
"step": 5300
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 0.5257701421800948,
|
1497 |
+
"grad_norm": 1.6543139219284058,
|
1498 |
+
"learning_rate": 1.8382582433636447e-05,
|
1499 |
+
"loss": 0.1779,
|
1500 |
+
"step": 5325
|
1501 |
+
},
|
1502 |
+
{
|
1503 |
+
"epoch": 0.5282385466034755,
|
1504 |
+
"grad_norm": 0.07421492785215378,
|
1505 |
+
"learning_rate": 1.822804618935964e-05,
|
1506 |
+
"loss": 0.1221,
|
1507 |
+
"step": 5350
|
1508 |
+
},
|
1509 |
+
{
|
1510 |
+
"epoch": 0.5307069510268563,
|
1511 |
+
"grad_norm": 0.8595842123031616,
|
1512 |
+
"learning_rate": 1.8073616502233247e-05,
|
1513 |
+
"loss": 0.1236,
|
1514 |
+
"step": 5375
|
1515 |
+
},
|
1516 |
+
{
|
1517 |
+
"epoch": 0.533175355450237,
|
1518 |
+
"grad_norm": 16.650583267211914,
|
1519 |
+
"learning_rate": 1.791930265894823e-05,
|
1520 |
+
"loss": 0.2412,
|
1521 |
+
"step": 5400
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 0.5356437598736177,
|
1525 |
+
"grad_norm": 0.28530654311180115,
|
1526 |
+
"learning_rate": 1.7765113939229207e-05,
|
1527 |
+
"loss": 0.1272,
|
1528 |
+
"step": 5425
|
1529 |
+
},
|
1530 |
+
{
|
1531 |
+
"epoch": 0.5381121642969984,
|
1532 |
+
"grad_norm": 0.06663267314434052,
|
1533 |
+
"learning_rate": 1.761105961527644e-05,
|
1534 |
+
"loss": 0.105,
|
1535 |
+
"step": 5450
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 0.5405805687203792,
|
1539 |
+
"grad_norm": 24.026447296142578,
|
1540 |
+
"learning_rate": 1.745714895120826e-05,
|
1541 |
+
"loss": 0.2117,
|
1542 |
+
"step": 5475
|
1543 |
+
},
|
1544 |
+
{
|
1545 |
+
"epoch": 0.5430489731437599,
|
1546 |
+
"grad_norm": 7.951884746551514,
|
1547 |
+
"learning_rate": 1.7303391202503958e-05,
|
1548 |
+
"loss": 0.076,
|
1549 |
+
"step": 5500
|
1550 |
+
},
|
1551 |
+
{
|
1552 |
+
"epoch": 0.5455173775671406,
|
1553 |
+
"grad_norm": 0.053239770233631134,
|
1554 |
+
"learning_rate": 1.714979561544718e-05,
|
1555 |
+
"loss": 0.1374,
|
1556 |
+
"step": 5525
|
1557 |
+
},
|
1558 |
+
{
|
1559 |
+
"epoch": 0.5479857819905213,
|
1560 |
+
"grad_norm": 76.0626220703125,
|
1561 |
+
"learning_rate": 1.6996371426569932e-05,
|
1562 |
+
"loss": 0.0946,
|
1563 |
+
"step": 5550
|
1564 |
+
},
|
1565 |
+
{
|
1566 |
+
"epoch": 0.5504541864139021,
|
1567 |
+
"grad_norm": 0.2509113550186157,
|
1568 |
+
"learning_rate": 1.68431278620971e-05,
|
1569 |
+
"loss": 0.1095,
|
1570 |
+
"step": 5575
|
1571 |
+
},
|
1572 |
+
{
|
1573 |
+
"epoch": 0.5529225908372828,
|
1574 |
+
"grad_norm": 0.5528861880302429,
|
1575 |
+
"learning_rate": 1.669007413739169e-05,
|
1576 |
+
"loss": 0.0947,
|
1577 |
+
"step": 5600
|
1578 |
+
},
|
1579 |
+
{
|
1580 |
+
"epoch": 0.5553909952606635,
|
1581 |
+
"grad_norm": 21.166288375854492,
|
1582 |
+
"learning_rate": 1.653721945640058e-05,
|
1583 |
+
"loss": 0.2383,
|
1584 |
+
"step": 5625
|
1585 |
+
},
|
1586 |
+
{
|
1587 |
+
"epoch": 0.5578593996840442,
|
1588 |
+
"grad_norm": 0.21656647324562073,
|
1589 |
+
"learning_rate": 1.6384573011101083e-05,
|
1590 |
+
"loss": 0.0716,
|
1591 |
+
"step": 5650
|
1592 |
+
},
|
1593 |
+
{
|
1594 |
+
"epoch": 0.560327804107425,
|
1595 |
+
"grad_norm": 10.91272258758545,
|
1596 |
+
"learning_rate": 1.6232143980948186e-05,
|
1597 |
+
"loss": 0.1018,
|
1598 |
+
"step": 5675
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 0.5627962085308057,
|
1602 |
+
"grad_norm": 0.04741775989532471,
|
1603 |
+
"learning_rate": 1.607994153232253e-05,
|
1604 |
+
"loss": 0.1539,
|
1605 |
+
"step": 5700
|
1606 |
+
},
|
1607 |
+
{
|
1608 |
+
"epoch": 0.5652646129541864,
|
1609 |
+
"grad_norm": 8.639245986938477,
|
1610 |
+
"learning_rate": 1.5927974817979168e-05,
|
1611 |
+
"loss": 0.1419,
|
1612 |
+
"step": 5725
|
1613 |
+
},
|
1614 |
+
{
|
1615 |
+
"epoch": 0.5677330173775671,
|
1616 |
+
"grad_norm": 0.7066089510917664,
|
1617 |
+
"learning_rate": 1.57762529764972e-05,
|
1618 |
+
"loss": 0.2255,
|
1619 |
+
"step": 5750
|
1620 |
+
},
|
1621 |
+
{
|
1622 |
+
"epoch": 0.5702014218009479,
|
1623 |
+
"grad_norm": 0.6187883019447327,
|
1624 |
+
"learning_rate": 1.562478513173017e-05,
|
1625 |
+
"loss": 0.0805,
|
1626 |
+
"step": 5775
|
1627 |
+
},
|
1628 |
+
{
|
1629 |
+
"epoch": 0.5726698262243286,
|
1630 |
+
"grad_norm": 0.3581185042858124,
|
1631 |
+
"learning_rate": 1.5473580392257464e-05,
|
1632 |
+
"loss": 0.0946,
|
1633 |
+
"step": 5800
|
1634 |
+
},
|
1635 |
+
{
|
1636 |
+
"epoch": 0.5751382306477093,
|
1637 |
+
"grad_norm": 0.2864006459712982,
|
1638 |
+
"learning_rate": 1.5322647850836498e-05,
|
1639 |
+
"loss": 0.1282,
|
1640 |
+
"step": 5825
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 0.57760663507109,
|
1644 |
+
"grad_norm": 23.85362434387207,
|
1645 |
+
"learning_rate": 1.5171996583855964e-05,
|
1646 |
+
"loss": 0.0942,
|
1647 |
+
"step": 5850
|
1648 |
+
},
|
1649 |
+
{
|
1650 |
+
"epoch": 0.5800750394944708,
|
1651 |
+
"grad_norm": 16.635116577148438,
|
1652 |
+
"learning_rate": 1.5021635650789985e-05,
|
1653 |
+
"loss": 0.1488,
|
1654 |
+
"step": 5875
|
1655 |
+
},
|
1656 |
+
{
|
1657 |
+
"epoch": 0.5825434439178515,
|
1658 |
+
"grad_norm": 11.043088912963867,
|
1659 |
+
"learning_rate": 1.4871574093653362e-05,
|
1660 |
+
"loss": 0.0785,
|
1661 |
+
"step": 5900
|
1662 |
+
},
|
1663 |
+
{
|
1664 |
+
"epoch": 0.5850118483412322,
|
1665 |
+
"grad_norm": 29.633804321289062,
|
1666 |
+
"learning_rate": 1.4721820936457798e-05,
|
1667 |
+
"loss": 0.281,
|
1668 |
+
"step": 5925
|
1669 |
+
},
|
1670 |
+
{
|
1671 |
+
"epoch": 0.5874802527646129,
|
1672 |
+
"grad_norm": 0.6947465538978577,
|
1673 |
+
"learning_rate": 1.4572385184669227e-05,
|
1674 |
+
"loss": 0.0925,
|
1675 |
+
"step": 5950
|
1676 |
+
},
|
1677 |
+
{
|
1678 |
+
"epoch": 0.5899486571879937,
|
1679 |
+
"grad_norm": 0.03761586174368858,
|
1680 |
+
"learning_rate": 1.4423275824666295e-05,
|
1681 |
+
"loss": 0.113,
|
1682 |
+
"step": 5975
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 0.5924170616113744,
|
1686 |
+
"grad_norm": 0.13467949628829956,
|
1687 |
+
"learning_rate": 1.4274501823199955e-05,
|
1688 |
+
"loss": 0.0601,
|
1689 |
+
"step": 6000
|
1690 |
+
},
|
1691 |
+
{
|
1692 |
+
"epoch": 0.5948854660347551,
|
1693 |
+
"grad_norm": 244.4285430908203,
|
1694 |
+
"learning_rate": 1.412607212685422e-05,
|
1695 |
+
"loss": 0.113,
|
1696 |
+
"step": 6025
|
1697 |
+
},
|
1698 |
+
{
|
1699 |
+
"epoch": 0.5973538704581358,
|
1700 |
+
"grad_norm": 0.42725473642349243,
|
1701 |
+
"learning_rate": 1.3977995661508197e-05,
|
1702 |
+
"loss": 0.1495,
|
1703 |
+
"step": 6050
|
1704 |
+
},
|
1705 |
+
{
|
1706 |
+
"epoch": 0.5998222748815166,
|
1707 |
+
"grad_norm": 50.870697021484375,
|
1708 |
+
"learning_rate": 1.3830281331799287e-05,
|
1709 |
+
"loss": 0.0424,
|
1710 |
+
"step": 6075
|
1711 |
+
},
|
1712 |
+
{
|
1713 |
+
"epoch": 0.6022906793048973,
|
1714 |
+
"grad_norm": 0.3475680649280548,
|
1715 |
+
"learning_rate": 1.3682938020587737e-05,
|
1716 |
+
"loss": 0.1899,
|
1717 |
+
"step": 6100
|
1718 |
+
},
|
1719 |
+
{
|
1720 |
+
"epoch": 0.604759083728278,
|
1721 |
+
"grad_norm": 0.7794524431228638,
|
1722 |
+
"learning_rate": 1.3535974588422443e-05,
|
1723 |
+
"loss": 0.1424,
|
1724 |
+
"step": 6125
|
1725 |
+
},
|
1726 |
+
{
|
1727 |
+
"epoch": 0.6072274881516587,
|
1728 |
+
"grad_norm": 0.6044468879699707,
|
1729 |
+
"learning_rate": 1.3389399873008114e-05,
|
1730 |
+
"loss": 0.0952,
|
1731 |
+
"step": 6150
|
1732 |
+
},
|
1733 |
+
{
|
1734 |
+
"epoch": 0.6096958925750395,
|
1735 |
+
"grad_norm": 11.903075218200684,
|
1736 |
+
"learning_rate": 1.3243222688673825e-05,
|
1737 |
+
"loss": 0.0834,
|
1738 |
+
"step": 6175
|
1739 |
+
},
|
1740 |
+
{
|
1741 |
+
"epoch": 0.6121642969984202,
|
1742 |
+
"grad_norm": 0.10283761471509933,
|
1743 |
+
"learning_rate": 1.3097451825842979e-05,
|
1744 |
+
"loss": 0.1422,
|
1745 |
+
"step": 6200
|
1746 |
+
},
|
1747 |
+
{
|
1748 |
+
"epoch": 0.6146327014218009,
|
1749 |
+
"grad_norm": 7.91100549697876,
|
1750 |
+
"learning_rate": 1.2952096050504648e-05,
|
1751 |
+
"loss": 0.0813,
|
1752 |
+
"step": 6225
|
1753 |
+
},
|
1754 |
+
{
|
1755 |
+
"epoch": 0.6171011058451816,
|
1756 |
+
"grad_norm": 24.93593406677246,
|
1757 |
+
"learning_rate": 1.2807164103686465e-05,
|
1758 |
+
"loss": 0.0994,
|
1759 |
+
"step": 6250
|
1760 |
+
},
|
1761 |
+
{
|
1762 |
+
"epoch": 0.6195695102685624,
|
1763 |
+
"grad_norm": 12.353562355041504,
|
1764 |
+
"learning_rate": 1.2662664700928952e-05,
|
1765 |
+
"loss": 0.123,
|
1766 |
+
"step": 6275
|
1767 |
+
},
|
1768 |
+
{
|
1769 |
+
"epoch": 0.6220379146919431,
|
1770 |
+
"grad_norm": 18.259349822998047,
|
1771 |
+
"learning_rate": 1.251860653176145e-05,
|
1772 |
+
"loss": 0.1529,
|
1773 |
+
"step": 6300
|
1774 |
+
},
|
1775 |
+
{
|
1776 |
+
"epoch": 0.6245063191153238,
|
1777 |
+
"grad_norm": 123.10421752929688,
|
1778 |
+
"learning_rate": 1.2374998259179516e-05,
|
1779 |
+
"loss": 0.1467,
|
1780 |
+
"step": 6325
|
1781 |
+
},
|
1782 |
+
{
|
1783 |
+
"epoch": 0.6269747235387045,
|
1784 |
+
"grad_norm": 0.5633807182312012,
|
1785 |
+
"learning_rate": 1.2231848519123996e-05,
|
1786 |
+
"loss": 0.1106,
|
1787 |
+
"step": 6350
|
1788 |
+
},
|
1789 |
+
{
|
1790 |
+
"epoch": 0.6294431279620853,
|
1791 |
+
"grad_norm": 0.6084317564964294,
|
1792 |
+
"learning_rate": 1.2089165919961712e-05,
|
1793 |
+
"loss": 0.122,
|
1794 |
+
"step": 6375
|
1795 |
+
},
|
1796 |
+
{
|
1797 |
+
"epoch": 0.631911532385466,
|
1798 |
+
"grad_norm": 22.184953689575195,
|
1799 |
+
"learning_rate": 1.1946959041967784e-05,
|
1800 |
+
"loss": 0.0995,
|
1801 |
+
"step": 6400
|
1802 |
+
},
|
1803 |
+
{
|
1804 |
+
"epoch": 0.6343799368088467,
|
1805 |
+
"grad_norm": 44.848262786865234,
|
1806 |
+
"learning_rate": 1.1805236436809633e-05,
|
1807 |
+
"loss": 0.0981,
|
1808 |
+
"step": 6425
|
1809 |
+
},
|
1810 |
+
{
|
1811 |
+
"epoch": 0.6368483412322274,
|
1812 |
+
"grad_norm": 0.31702718138694763,
|
1813 |
+
"learning_rate": 1.166400662703275e-05,
|
1814 |
+
"loss": 0.1638,
|
1815 |
+
"step": 6450
|
1816 |
+
},
|
1817 |
+
{
|
1818 |
+
"epoch": 0.6393167456556083,
|
1819 |
+
"grad_norm": 4.540043354034424,
|
1820 |
+
"learning_rate": 1.152327810554818e-05,
|
1821 |
+
"loss": 0.1017,
|
1822 |
+
"step": 6475
|
1823 |
+
},
|
1824 |
+
{
|
1825 |
+
"epoch": 0.641785150078989,
|
1826 |
+
"grad_norm": 64.72604370117188,
|
1827 |
+
"learning_rate": 1.1383059335121791e-05,
|
1828 |
+
"loss": 0.0558,
|
1829 |
+
"step": 6500
|
1830 |
+
},
|
1831 |
+
{
|
1832 |
+
"epoch": 0.6442535545023697,
|
1833 |
+
"grad_norm": 0.2067117989063263,
|
1834 |
+
"learning_rate": 1.1243358747865369e-05,
|
1835 |
+
"loss": 0.0484,
|
1836 |
+
"step": 6525
|
1837 |
+
},
|
1838 |
+
{
|
1839 |
+
"epoch": 0.6467219589257504,
|
1840 |
+
"grad_norm": 0.043737076222896576,
|
1841 |
+
"learning_rate": 1.1104184744729562e-05,
|
1842 |
+
"loss": 0.1653,
|
1843 |
+
"step": 6550
|
1844 |
+
},
|
1845 |
+
{
|
1846 |
+
"epoch": 0.6491903633491312,
|
1847 |
+
"grad_norm": 0.25652366876602173,
|
1848 |
+
"learning_rate": 1.0965545694998631e-05,
|
1849 |
+
"loss": 0.1647,
|
1850 |
+
"step": 6575
|
1851 |
+
},
|
1852 |
+
{
|
1853 |
+
"epoch": 0.6516587677725119,
|
1854 |
+
"grad_norm": 9.453946113586426,
|
1855 |
+
"learning_rate": 1.0827449935787253e-05,
|
1856 |
+
"loss": 0.1676,
|
1857 |
+
"step": 6600
|
1858 |
+
},
|
1859 |
+
{
|
1860 |
+
"epoch": 0.6541271721958926,
|
1861 |
+
"grad_norm": 0.35642993450164795,
|
1862 |
+
"learning_rate": 1.0689905771539087e-05,
|
1863 |
+
"loss": 0.1037,
|
1864 |
+
"step": 6625
|
1865 |
+
},
|
1866 |
+
{
|
1867 |
+
"epoch": 0.6565955766192733,
|
1868 |
+
"grad_norm": 0.06698060780763626,
|
1869 |
+
"learning_rate": 1.055292147352739e-05,
|
1870 |
+
"loss": 0.0647,
|
1871 |
+
"step": 6650
|
1872 |
+
},
|
1873 |
+
{
|
1874 |
+
"epoch": 0.6590639810426541,
|
1875 |
+
"grad_norm": 16.954423904418945,
|
1876 |
+
"learning_rate": 1.0416505279357664e-05,
|
1877 |
+
"loss": 0.1165,
|
1878 |
+
"step": 6675
|
1879 |
+
},
|
1880 |
+
{
|
1881 |
+
"epoch": 0.6615323854660348,
|
1882 |
+
"grad_norm": 16.60346221923828,
|
1883 |
+
"learning_rate": 1.0280665392472276e-05,
|
1884 |
+
"loss": 0.1955,
|
1885 |
+
"step": 6700
|
1886 |
+
},
|
1887 |
+
{
|
1888 |
+
"epoch": 0.6640007898894155,
|
1889 |
+
"grad_norm": 0.030714910477399826,
|
1890 |
+
"learning_rate": 1.0145409981657073e-05,
|
1891 |
+
"loss": 0.123,
|
1892 |
+
"step": 6725
|
1893 |
+
},
|
1894 |
+
{
|
1895 |
+
"epoch": 0.6664691943127962,
|
1896 |
+
"grad_norm": 44.65860366821289,
|
1897 |
+
"learning_rate": 1.001074718055024e-05,
|
1898 |
+
"loss": 0.0789,
|
1899 |
+
"step": 6750
|
1900 |
+
},
|
1901 |
+
{
|
1902 |
+
"epoch": 0.668937598736177,
|
1903 |
+
"grad_norm": 67.4146957397461,
|
1904 |
+
"learning_rate": 9.876685087153131e-06,
|
1905 |
+
"loss": 0.0771,
|
1906 |
+
"step": 6775
|
1907 |
+
},
|
1908 |
+
{
|
1909 |
+
"epoch": 0.6714060031595577,
|
1910 |
+
"grad_norm": 6.725965976715088,
|
1911 |
+
"learning_rate": 9.743231763343295e-06,
|
1912 |
+
"loss": 0.1283,
|
1913 |
+
"step": 6800
|
1914 |
+
},
|
1915 |
+
{
|
1916 |
+
"epoch": 0.6738744075829384,
|
1917 |
+
"grad_norm": 0.0958356186747551,
|
1918 |
+
"learning_rate": 9.610395234389687e-06,
|
1919 |
+
"loss": 0.1342,
|
1920 |
+
"step": 6825
|
1921 |
+
},
|
1922 |
+
{
|
1923 |
+
"epoch": 0.6763428120063191,
|
1924 |
+
"grad_norm": 13.343214988708496,
|
1925 |
+
"learning_rate": 9.478183488470059e-06,
|
1926 |
+
"loss": 0.1981,
|
1927 |
+
"step": 6850
|
1928 |
+
},
|
1929 |
+
{
|
1930 |
+
"epoch": 0.6788112164296999,
|
1931 |
+
"grad_norm": 0.025249022990465164,
|
1932 |
+
"learning_rate": 9.346604476190558e-06,
|
1933 |
+
"loss": 0.0897,
|
1934 |
+
"step": 6875
|
1935 |
+
},
|
1936 |
+
{
|
1937 |
+
"epoch": 0.6812796208530806,
|
1938 |
+
"grad_norm": 6.261577129364014,
|
1939 |
+
"learning_rate": 9.215666110107686e-06,
|
1940 |
+
"loss": 0.1289,
|
1941 |
+
"step": 6900
|
1942 |
+
},
|
1943 |
+
{
|
1944 |
+
"epoch": 0.6837480252764613,
|
1945 |
+
"grad_norm": 0.239964097738266,
|
1946 |
+
"learning_rate": 9.085376264252406e-06,
|
1947 |
+
"loss": 0.0826,
|
1948 |
+
"step": 6925
|
1949 |
+
},
|
1950 |
+
{
|
1951 |
+
"epoch": 0.686216429699842,
|
1952 |
+
"grad_norm": 0.3692372441291809,
|
1953 |
+
"learning_rate": 8.955742773656644e-06,
|
1954 |
+
"loss": 0.0943,
|
1955 |
+
"step": 6950
|
1956 |
+
},
|
1957 |
+
{
|
1958 |
+
"epoch": 0.6886848341232228,
|
1959 |
+
"grad_norm": 0.1730097234249115,
|
1960 |
+
"learning_rate": 8.826773433882154e-06,
|
1961 |
+
"loss": 0.128,
|
1962 |
+
"step": 6975
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 0.6911532385466035,
|
1966 |
+
"grad_norm": 12.025056838989258,
|
1967 |
+
"learning_rate": 8.698476000551741e-06,
|
1968 |
+
"loss": 0.1283,
|
1969 |
+
"step": 7000
|
1970 |
+
},
|
1971 |
+
{
|
1972 |
+
"epoch": 0.6936216429699842,
|
1973 |
+
"grad_norm": 0.33707374334335327,
|
1974 |
+
"learning_rate": 8.570858188882795e-06,
|
1975 |
+
"loss": 0.1005,
|
1976 |
+
"step": 7025
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 0.6960900473933649,
|
1980 |
+
"grad_norm": 1.0958293676376343,
|
1981 |
+
"learning_rate": 8.443927673223426e-06,
|
1982 |
+
"loss": 0.1452,
|
1983 |
+
"step": 7050
|
1984 |
+
},
|
1985 |
+
{
|
1986 |
+
"epoch": 0.6985584518167457,
|
1987 |
+
"grad_norm": 12.001123428344727,
|
1988 |
+
"learning_rate": 8.317692086590907e-06,
|
1989 |
+
"loss": 0.084,
|
1990 |
+
"step": 7075
|
1991 |
+
},
|
1992 |
+
{
|
1993 |
+
"epoch": 0.7010268562401264,
|
1994 |
+
"grad_norm": 0.7137736082077026,
|
1995 |
+
"learning_rate": 8.192159020212685e-06,
|
1996 |
+
"loss": 0.0825,
|
1997 |
+
"step": 7100
|
1998 |
+
},
|
1999 |
+
{
|
2000 |
+
"epoch": 0.7034952606635071,
|
2001 |
+
"grad_norm": 14.759200096130371,
|
2002 |
+
"learning_rate": 8.06733602306986e-06,
|
2003 |
+
"loss": 0.1344,
|
2004 |
+
"step": 7125
|
2005 |
+
},
|
2006 |
+
{
|
2007 |
+
"epoch": 0.7059636650868878,
|
2008 |
+
"grad_norm": 13.587589263916016,
|
2009 |
+
"learning_rate": 7.943230601443246e-06,
|
2010 |
+
"loss": 0.0724,
|
2011 |
+
"step": 7150
|
2012 |
+
},
|
2013 |
+
{
|
2014 |
+
"epoch": 0.7084320695102686,
|
2015 |
+
"grad_norm": 20.667478561401367,
|
2016 |
+
"learning_rate": 7.819850218461958e-06,
|
2017 |
+
"loss": 0.167,
|
2018 |
+
"step": 7175
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 0.7109004739336493,
|
2022 |
+
"grad_norm": 0.40158161520957947,
|
2023 |
+
"learning_rate": 7.697202293654629e-06,
|
2024 |
+
"loss": 0.1232,
|
2025 |
+
"step": 7200
|
2026 |
+
},
|
2027 |
+
{
|
2028 |
+
"epoch": 0.71336887835703,
|
2029 |
+
"grad_norm": 0.10647299885749817,
|
2030 |
+
"learning_rate": 7.575294202503221e-06,
|
2031 |
+
"loss": 0.118,
|
2032 |
+
"step": 7225
|
2033 |
+
},
|
2034 |
+
{
|
2035 |
+
"epoch": 0.7158372827804107,
|
2036 |
+
"grad_norm": 9.874299049377441,
|
2037 |
+
"learning_rate": 7.454133275999516e-06,
|
2038 |
+
"loss": 0.0871,
|
2039 |
+
"step": 7250
|
2040 |
+
},
|
2041 |
+
{
|
2042 |
+
"epoch": 0.7183056872037915,
|
2043 |
+
"grad_norm": 0.027496134862303734,
|
2044 |
+
"learning_rate": 7.3337268002042175e-06,
|
2045 |
+
"loss": 0.1054,
|
2046 |
+
"step": 7275
|
2047 |
+
},
|
2048 |
+
{
|
2049 |
+
"epoch": 0.7207740916271722,
|
2050 |
+
"grad_norm": 0.2057611644268036,
|
2051 |
+
"learning_rate": 7.214082015808883e-06,
|
2052 |
+
"loss": 0.1139,
|
2053 |
+
"step": 7300
|
2054 |
+
},
|
2055 |
+
{
|
2056 |
+
"epoch": 0.7232424960505529,
|
2057 |
+
"grad_norm": 0.40457573533058167,
|
2058 |
+
"learning_rate": 7.095206117700428e-06,
|
2059 |
+
"loss": 0.0828,
|
2060 |
+
"step": 7325
|
2061 |
+
},
|
2062 |
+
{
|
2063 |
+
"epoch": 0.7257109004739336,
|
2064 |
+
"grad_norm": 12.48037052154541,
|
2065 |
+
"learning_rate": 6.9771062545284695e-06,
|
2066 |
+
"loss": 0.1886,
|
2067 |
+
"step": 7350
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 0.7281793048973144,
|
2071 |
+
"grad_norm": 0.0360858179628849,
|
2072 |
+
"learning_rate": 6.859789528275469e-06,
|
2073 |
+
"loss": 0.1513,
|
2074 |
+
"step": 7375
|
2075 |
+
},
|
2076 |
+
{
|
2077 |
+
"epoch": 0.7306477093206951,
|
2078 |
+
"grad_norm": 27.785545349121094,
|
2079 |
+
"learning_rate": 6.743262993829658e-06,
|
2080 |
+
"loss": 0.1204,
|
2081 |
+
"step": 7400
|
2082 |
+
},
|
2083 |
+
{
|
2084 |
+
"epoch": 0.7331161137440758,
|
2085 |
+
"grad_norm": 328.3162536621094,
|
2086 |
+
"learning_rate": 6.627533658560725e-06,
|
2087 |
+
"loss": 0.0724,
|
2088 |
+
"step": 7425
|
2089 |
+
},
|
2090 |
+
{
|
2091 |
+
"epoch": 0.7355845181674565,
|
2092 |
+
"grad_norm": 17.664350509643555,
|
2093 |
+
"learning_rate": 6.512608481898494e-06,
|
2094 |
+
"loss": 0.111,
|
2095 |
+
"step": 7450
|
2096 |
+
},
|
2097 |
+
{
|
2098 |
+
"epoch": 0.7380529225908373,
|
2099 |
+
"grad_norm": 0.5029625296592712,
|
2100 |
+
"learning_rate": 6.398494374914388e-06,
|
2101 |
+
"loss": 0.0628,
|
2102 |
+
"step": 7475
|
2103 |
+
},
|
2104 |
+
{
|
2105 |
+
"epoch": 0.740521327014218,
|
2106 |
+
"grad_norm": 0.4134403169155121,
|
2107 |
+
"learning_rate": 6.285198199905824e-06,
|
2108 |
+
"loss": 0.1323,
|
2109 |
+
"step": 7500
|
2110 |
+
},
|
2111 |
+
{
|
2112 |
+
"epoch": 0.7429897314375987,
|
2113 |
+
"grad_norm": 0.043722160160541534,
|
2114 |
+
"learning_rate": 6.172726769983557e-06,
|
2115 |
+
"loss": 0.104,
|
2116 |
+
"step": 7525
|
2117 |
+
},
|
2118 |
+
{
|
2119 |
+
"epoch": 0.7454581358609794,
|
2120 |
+
"grad_norm": 10.048009872436523,
|
2121 |
+
"learning_rate": 6.061086848661972e-06,
|
2122 |
+
"loss": 0.161,
|
2123 |
+
"step": 7550
|
2124 |
+
},
|
2125 |
+
{
|
2126 |
+
"epoch": 0.7479265402843602,
|
2127 |
+
"grad_norm": 70.8975601196289,
|
2128 |
+
"learning_rate": 5.9502851494523264e-06,
|
2129 |
+
"loss": 0.0889,
|
2130 |
+
"step": 7575
|
2131 |
+
},
|
2132 |
+
{
|
2133 |
+
"epoch": 0.7503949447077409,
|
2134 |
+
"grad_norm": 0.026410667225718498,
|
2135 |
+
"learning_rate": 5.840328335459096e-06,
|
2136 |
+
"loss": 0.146,
|
2137 |
+
"step": 7600
|
2138 |
+
},
|
2139 |
+
{
|
2140 |
+
"epoch": 0.7528633491311216,
|
2141 |
+
"grad_norm": 2.6003201007843018,
|
2142 |
+
"learning_rate": 5.7312230189792266e-06,
|
2143 |
+
"loss": 0.1385,
|
2144 |
+
"step": 7625
|
2145 |
+
},
|
2146 |
+
{
|
2147 |
+
"epoch": 0.7553317535545023,
|
2148 |
+
"grad_norm": 0.19096583127975464,
|
2149 |
+
"learning_rate": 5.622975761104503e-06,
|
2150 |
+
"loss": 0.0671,
|
2151 |
+
"step": 7650
|
2152 |
+
},
|
2153 |
+
{
|
2154 |
+
"epoch": 0.7578001579778831,
|
2155 |
+
"grad_norm": 40.680076599121094,
|
2156 |
+
"learning_rate": 5.5155930713270345e-06,
|
2157 |
+
"loss": 0.1304,
|
2158 |
+
"step": 7675
|
2159 |
+
},
|
2160 |
+
{
|
2161 |
+
"epoch": 0.7602685624012638,
|
2162 |
+
"grad_norm": 240.6962127685547,
|
2163 |
+
"learning_rate": 5.409081407147794e-06,
|
2164 |
+
"loss": 0.0949,
|
2165 |
+
"step": 7700
|
2166 |
+
},
|
2167 |
+
{
|
2168 |
+
"epoch": 0.7627369668246445,
|
2169 |
+
"grad_norm": 0.07990098744630814,
|
2170 |
+
"learning_rate": 5.303447173688248e-06,
|
2171 |
+
"loss": 0.1469,
|
2172 |
+
"step": 7725
|
2173 |
+
},
|
2174 |
+
{
|
2175 |
+
"epoch": 0.7652053712480252,
|
2176 |
+
"grad_norm": 13.084890365600586,
|
2177 |
+
"learning_rate": 5.198696723305241e-06,
|
2178 |
+
"loss": 0.1706,
|
2179 |
+
"step": 7750
|
2180 |
+
},
|
2181 |
+
{
|
2182 |
+
"epoch": 0.767673775671406,
|
2183 |
+
"grad_norm": 0.09494700282812119,
|
2184 |
+
"learning_rate": 5.094836355208963e-06,
|
2185 |
+
"loss": 0.067,
|
2186 |
+
"step": 7775
|
2187 |
+
},
|
2188 |
+
{
|
2189 |
+
"epoch": 0.7701421800947867,
|
2190 |
+
"grad_norm": 21.238924026489258,
|
2191 |
+
"learning_rate": 4.991872315084148e-06,
|
2192 |
+
"loss": 0.1267,
|
2193 |
+
"step": 7800
|
2194 |
+
},
|
2195 |
+
{
|
2196 |
+
"epoch": 0.7726105845181674,
|
2197 |
+
"grad_norm": 0.14261184632778168,
|
2198 |
+
"learning_rate": 4.889810794714489e-06,
|
2199 |
+
"loss": 0.0646,
|
2200 |
+
"step": 7825
|
2201 |
+
},
|
2202 |
+
{
|
2203 |
+
"epoch": 0.7750789889415481,
|
2204 |
+
"grad_norm": 0.028456885367631912,
|
2205 |
+
"learning_rate": 4.788657931610297e-06,
|
2206 |
+
"loss": 0.1614,
|
2207 |
+
"step": 7850
|
2208 |
+
},
|
2209 |
+
{
|
2210 |
+
"epoch": 0.777547393364929,
|
2211 |
+
"grad_norm": 0.06065530702471733,
|
2212 |
+
"learning_rate": 4.6884198086394105e-06,
|
2213 |
+
"loss": 0.1206,
|
2214 |
+
"step": 7875
|
2215 |
+
},
|
2216 |
+
{
|
2217 |
+
"epoch": 0.7800157977883097,
|
2218 |
+
"grad_norm": 0.01794976182281971,
|
2219 |
+
"learning_rate": 4.589102453661409e-06,
|
2220 |
+
"loss": 0.0832,
|
2221 |
+
"step": 7900
|
2222 |
+
},
|
2223 |
+
{
|
2224 |
+
"epoch": 0.7824842022116903,
|
2225 |
+
"grad_norm": 0.05781606584787369,
|
2226 |
+
"learning_rate": 4.490711839165116e-06,
|
2227 |
+
"loss": 0.0638,
|
2228 |
+
"step": 7925
|
2229 |
+
},
|
2230 |
+
{
|
2231 |
+
"epoch": 0.784952606635071,
|
2232 |
+
"grad_norm": 0.06837207823991776,
|
2233 |
+
"learning_rate": 4.393253881909456e-06,
|
2234 |
+
"loss": 0.0898,
|
2235 |
+
"step": 7950
|
2236 |
+
},
|
2237 |
+
{
|
2238 |
+
"epoch": 0.7874210110584519,
|
2239 |
+
"grad_norm": 2.171274423599243,
|
2240 |
+
"learning_rate": 4.296734442567611e-06,
|
2241 |
+
"loss": 0.1506,
|
2242 |
+
"step": 7975
|
2243 |
+
},
|
2244 |
+
{
|
2245 |
+
"epoch": 0.7898894154818326,
|
2246 |
+
"grad_norm": 43.33221435546875,
|
2247 |
+
"learning_rate": 4.2011593253746506e-06,
|
2248 |
+
"loss": 0.1995,
|
2249 |
+
"step": 8000
|
2250 |
+
},
|
2251 |
+
{
|
2252 |
+
"epoch": 0.7923578199052133,
|
2253 |
+
"grad_norm": 0.059240926057100296,
|
2254 |
+
"learning_rate": 4.106534277778435e-06,
|
2255 |
+
"loss": 0.1661,
|
2256 |
+
"step": 8025
|
2257 |
+
},
|
2258 |
+
{
|
2259 |
+
"epoch": 0.794826224328594,
|
2260 |
+
"grad_norm": 49.15257263183594,
|
2261 |
+
"learning_rate": 4.012864990094003e-06,
|
2262 |
+
"loss": 0.1212,
|
2263 |
+
"step": 8050
|
2264 |
+
},
|
2265 |
+
{
|
2266 |
+
"epoch": 0.7972946287519748,
|
2267 |
+
"grad_norm": 39.35380554199219,
|
2268 |
+
"learning_rate": 3.920157095161395e-06,
|
2269 |
+
"loss": 0.1096,
|
2270 |
+
"step": 8075
|
2271 |
+
},
|
2272 |
+
{
|
2273 |
+
"epoch": 0.7997630331753555,
|
2274 |
+
"grad_norm": 0.11532844603061676,
|
2275 |
+
"learning_rate": 3.828416168006929e-06,
|
2276 |
+
"loss": 0.112,
|
2277 |
+
"step": 8100
|
2278 |
+
},
|
2279 |
+
{
|
2280 |
+
"epoch": 0.8022314375987362,
|
2281 |
+
"grad_norm": 20.4510555267334,
|
2282 |
+
"learning_rate": 3.737647725507907e-06,
|
2283 |
+
"loss": 0.1115,
|
2284 |
+
"step": 8125
|
2285 |
+
},
|
2286 |
+
{
|
2287 |
+
"epoch": 0.8046998420221169,
|
2288 |
+
"grad_norm": 0.491216778755188,
|
2289 |
+
"learning_rate": 3.6478572260608867e-06,
|
2290 |
+
"loss": 0.0404,
|
2291 |
+
"step": 8150
|
2292 |
+
},
|
2293 |
+
{
|
2294 |
+
"epoch": 0.8071682464454977,
|
2295 |
+
"grad_norm": 15.650696754455566,
|
2296 |
+
"learning_rate": 3.5590500692534356e-06,
|
2297 |
+
"loss": 0.1614,
|
2298 |
+
"step": 8175
|
2299 |
+
},
|
2300 |
+
{
|
2301 |
+
"epoch": 0.8096366508688784,
|
2302 |
+
"grad_norm": 0.780383288860321,
|
2303 |
+
"learning_rate": 3.4712315955394103e-06,
|
2304 |
+
"loss": 0.1314,
|
2305 |
+
"step": 8200
|
2306 |
+
},
|
2307 |
+
{
|
2308 |
+
"epoch": 0.8121050552922591,
|
2309 |
+
"grad_norm": 0.07208031415939331,
|
2310 |
+
"learning_rate": 3.3844070859178156e-06,
|
2311 |
+
"loss": 0.0263,
|
2312 |
+
"step": 8225
|
2313 |
+
},
|
2314 |
+
{
|
2315 |
+
"epoch": 0.8145734597156398,
|
2316 |
+
"grad_norm": 0.024150904268026352,
|
2317 |
+
"learning_rate": 3.2985817616152338e-06,
|
2318 |
+
"loss": 0.078,
|
2319 |
+
"step": 8250
|
2320 |
+
},
|
2321 |
+
{
|
2322 |
+
"epoch": 0.8170418641390206,
|
2323 |
+
"grad_norm": 0.044707950204610825,
|
2324 |
+
"learning_rate": 3.213760783771831e-06,
|
2325 |
+
"loss": 0.1588,
|
2326 |
+
"step": 8275
|
2327 |
+
},
|
2328 |
+
{
|
2329 |
+
"epoch": 0.8195102685624013,
|
2330 |
+
"grad_norm": 0.2114710658788681,
|
2331 |
+
"learning_rate": 3.129949253131006e-06,
|
2332 |
+
"loss": 0.0945,
|
2333 |
+
"step": 8300
|
2334 |
+
},
|
2335 |
+
{
|
2336 |
+
"epoch": 0.821978672985782,
|
2337 |
+
"grad_norm": 19.596643447875977,
|
2338 |
+
"learning_rate": 3.0471522097326466e-06,
|
2339 |
+
"loss": 0.1634,
|
2340 |
+
"step": 8325
|
2341 |
+
},
|
2342 |
+
{
|
2343 |
+
"epoch": 0.8244470774091627,
|
2344 |
+
"grad_norm": 0.3169441819190979,
|
2345 |
+
"learning_rate": 2.9653746326100296e-06,
|
2346 |
+
"loss": 0.2111,
|
2347 |
+
"step": 8350
|
2348 |
+
},
|
2349 |
+
{
|
2350 |
+
"epoch": 0.8269154818325435,
|
2351 |
+
"grad_norm": 144.08193969726562,
|
2352 |
+
"learning_rate": 2.8846214394904315e-06,
|
2353 |
+
"loss": 0.2806,
|
2354 |
+
"step": 8375
|
2355 |
+
},
|
2356 |
+
{
|
2357 |
+
"epoch": 0.8293838862559242,
|
2358 |
+
"grad_norm": 0.25086668133735657,
|
2359 |
+
"learning_rate": 2.8048974864994026e-06,
|
2360 |
+
"loss": 0.1022,
|
2361 |
+
"step": 8400
|
2362 |
+
},
|
2363 |
+
{
|
2364 |
+
"epoch": 0.8318522906793049,
|
2365 |
+
"grad_norm": 0.04703996703028679,
|
2366 |
+
"learning_rate": 2.7262075678686974e-06,
|
2367 |
+
"loss": 0.1644,
|
2368 |
+
"step": 8425
|
2369 |
+
},
|
2370 |
+
{
|
2371 |
+
"epoch": 0.8343206951026856,
|
2372 |
+
"grad_norm": 0.012322982773184776,
|
2373 |
+
"learning_rate": 2.6485564156480182e-06,
|
2374 |
+
"loss": 0.0843,
|
2375 |
+
"step": 8450
|
2376 |
+
},
|
2377 |
+
{
|
2378 |
+
"epoch": 0.8367890995260664,
|
2379 |
+
"grad_norm": 8.877446174621582,
|
2380 |
+
"learning_rate": 2.5719486994204325e-06,
|
2381 |
+
"loss": 0.1504,
|
2382 |
+
"step": 8475
|
2383 |
+
},
|
2384 |
+
{
|
2385 |
+
"epoch": 0.8392575039494471,
|
2386 |
+
"grad_norm": 0.40166324377059937,
|
2387 |
+
"learning_rate": 2.496389026021566e-06,
|
2388 |
+
"loss": 0.0912,
|
2389 |
+
"step": 8500
|
2390 |
+
},
|
2391 |
+
{
|
2392 |
+
"epoch": 0.8417259083728278,
|
2393 |
+
"grad_norm": 0.2992970049381256,
|
2394 |
+
"learning_rate": 2.421881939262569e-06,
|
2395 |
+
"loss": 0.1689,
|
2396 |
+
"step": 8525
|
2397 |
+
},
|
2398 |
+
{
|
2399 |
+
"epoch": 0.8441943127962085,
|
2400 |
+
"grad_norm": 0.2448200136423111,
|
2401 |
+
"learning_rate": 2.348431919656875e-06,
|
2402 |
+
"loss": 0.0757,
|
2403 |
+
"step": 8550
|
2404 |
+
},
|
2405 |
+
{
|
2406 |
+
"epoch": 0.8466627172195893,
|
2407 |
+
"grad_norm": 0.14062806963920593,
|
2408 |
+
"learning_rate": 2.2760433841507635e-06,
|
2409 |
+
"loss": 0.0736,
|
2410 |
+
"step": 8575
|
2411 |
+
},
|
2412 |
+
{
|
2413 |
+
"epoch": 0.84913112164297,
|
2414 |
+
"grad_norm": 0.27994504570961,
|
2415 |
+
"learning_rate": 2.2047206858577375e-06,
|
2416 |
+
"loss": 0.0687,
|
2417 |
+
"step": 8600
|
2418 |
+
},
|
2419 |
+
{
|
2420 |
+
"epoch": 0.8515995260663507,
|
2421 |
+
"grad_norm": 0.07849132269620895,
|
2422 |
+
"learning_rate": 2.1344681137967614e-06,
|
2423 |
+
"loss": 0.0625,
|
2424 |
+
"step": 8625
|
2425 |
+
},
|
2426 |
+
{
|
2427 |
+
"epoch": 0.8540679304897314,
|
2428 |
+
"grad_norm": 3.856325387954712,
|
2429 |
+
"learning_rate": 2.065289892634328e-06,
|
2430 |
+
"loss": 0.0286,
|
2431 |
+
"step": 8650
|
2432 |
+
},
|
2433 |
+
{
|
2434 |
+
"epoch": 0.8565363349131122,
|
2435 |
+
"grad_norm": 0.016810081899166107,
|
2436 |
+
"learning_rate": 1.9971901824303972e-06,
|
2437 |
+
"loss": 0.1694,
|
2438 |
+
"step": 8675
|
2439 |
+
},
|
2440 |
+
{
|
2441 |
+
"epoch": 0.8590047393364929,
|
2442 |
+
"grad_norm": 0.05156625807285309,
|
2443 |
+
"learning_rate": 1.930173078388269e-06,
|
2444 |
+
"loss": 0.0658,
|
2445 |
+
"step": 8700
|
2446 |
+
},
|
2447 |
+
{
|
2448 |
+
"epoch": 0.8614731437598736,
|
2449 |
+
"grad_norm": 0.08223263919353485,
|
2450 |
+
"learning_rate": 1.8642426106082667e-06,
|
2451 |
+
"loss": 0.0483,
|
2452 |
+
"step": 8725
|
2453 |
+
},
|
2454 |
+
{
|
2455 |
+
"epoch": 0.8639415481832543,
|
2456 |
+
"grad_norm": 1.8065041303634644,
|
2457 |
+
"learning_rate": 1.7994027438454087e-06,
|
2458 |
+
"loss": 0.0688,
|
2459 |
+
"step": 8750
|
2460 |
+
},
|
2461 |
+
{
|
2462 |
+
"epoch": 0.8664099526066351,
|
2463 |
+
"grad_norm": 0.039519645273685455,
|
2464 |
+
"learning_rate": 1.7356573772709938e-06,
|
2465 |
+
"loss": 0.0572,
|
2466 |
+
"step": 8775
|
2467 |
+
},
|
2468 |
+
{
|
2469 |
+
"epoch": 0.8688783570300158,
|
2470 |
+
"grad_norm": 25.36179542541504,
|
2471 |
+
"learning_rate": 1.6730103442381195e-06,
|
2472 |
+
"loss": 0.1886,
|
2473 |
+
"step": 8800
|
2474 |
+
},
|
2475 |
+
{
|
2476 |
+
"epoch": 0.8713467614533965,
|
2477 |
+
"grad_norm": 0.4453946053981781,
|
2478 |
+
"learning_rate": 1.6114654120511407e-06,
|
2479 |
+
"loss": 0.1323,
|
2480 |
+
"step": 8825
|
2481 |
+
},
|
2482 |
+
{
|
2483 |
+
"epoch": 0.8738151658767772,
|
2484 |
+
"grad_norm": 0.15913625061511993,
|
2485 |
+
"learning_rate": 1.5510262817391474e-06,
|
2486 |
+
"loss": 0.0511,
|
2487 |
+
"step": 8850
|
2488 |
+
},
|
2489 |
+
{
|
2490 |
+
"epoch": 0.876283570300158,
|
2491 |
+
"grad_norm": 0.1287483274936676,
|
2492 |
+
"learning_rate": 1.4916965878333911e-06,
|
2493 |
+
"loss": 0.0873,
|
2494 |
+
"step": 8875
|
2495 |
+
},
|
2496 |
+
{
|
2497 |
+
"epoch": 0.8787519747235387,
|
2498 |
+
"grad_norm": 18.978235244750977,
|
2499 |
+
"learning_rate": 1.4334798981487242e-06,
|
2500 |
+
"loss": 0.0778,
|
2501 |
+
"step": 8900
|
2502 |
+
},
|
2503 |
+
{
|
2504 |
+
"epoch": 0.8812203791469194,
|
2505 |
+
"grad_norm": 7.997274875640869,
|
2506 |
+
"learning_rate": 1.3763797135690404e-06,
|
2507 |
+
"loss": 0.1199,
|
2508 |
+
"step": 8925
|
2509 |
+
},
|
2510 |
+
{
|
2511 |
+
"epoch": 0.8836887835703001,
|
2512 |
+
"grad_norm": 0.1320393681526184,
|
2513 |
+
"learning_rate": 1.3203994678367527e-06,
|
2514 |
+
"loss": 0.1069,
|
2515 |
+
"step": 8950
|
2516 |
+
},
|
2517 |
+
{
|
2518 |
+
"epoch": 0.8861571879936809,
|
2519 |
+
"grad_norm": 294.17071533203125,
|
2520 |
+
"learning_rate": 1.2655425273463107e-06,
|
2521 |
+
"loss": 0.0845,
|
2522 |
+
"step": 8975
|
2523 |
+
},
|
2524 |
+
{
|
2525 |
+
"epoch": 0.8886255924170616,
|
2526 |
+
"grad_norm": 0.23417572677135468,
|
2527 |
+
"learning_rate": 1.211812190941748e-06,
|
2528 |
+
"loss": 0.1323,
|
2529 |
+
"step": 9000
|
2530 |
+
},
|
2531 |
+
{
|
2532 |
+
"epoch": 0.8910939968404423,
|
2533 |
+
"grad_norm": 0.2659314274787903,
|
2534 |
+
"learning_rate": 1.1592116897183136e-06,
|
2535 |
+
"loss": 0.1898,
|
2536 |
+
"step": 9025
|
2537 |
+
},
|
2538 |
+
{
|
2539 |
+
"epoch": 0.893562401263823,
|
2540 |
+
"grad_norm": 0.08678634464740753,
|
2541 |
+
"learning_rate": 1.107744186828168e-06,
|
2542 |
+
"loss": 0.079,
|
2543 |
+
"step": 9050
|
2544 |
+
},
|
2545 |
+
{
|
2546 |
+
"epoch": 0.8960308056872038,
|
2547 |
+
"grad_norm": 0.0248000156134367,
|
2548 |
+
"learning_rate": 1.0574127772901565e-06,
|
2549 |
+
"loss": 0.1021,
|
2550 |
+
"step": 9075
|
2551 |
+
},
|
2552 |
+
{
|
2553 |
+
"epoch": 0.8984992101105845,
|
2554 |
+
"grad_norm": 91.24132537841797,
|
2555 |
+
"learning_rate": 1.0082204878037083e-06,
|
2556 |
+
"loss": 0.0775,
|
2557 |
+
"step": 9100
|
2558 |
+
},
|
2559 |
+
{
|
2560 |
+
"epoch": 0.9009676145339652,
|
2561 |
+
"grad_norm": 0.49354681372642517,
|
2562 |
+
"learning_rate": 9.601702765668098e-07,
|
2563 |
+
"loss": 0.1155,
|
2564 |
+
"step": 9125
|
2565 |
+
},
|
2566 |
+
{
|
2567 |
+
"epoch": 0.9034360189573459,
|
2568 |
+
"grad_norm": 32.61450958251953,
|
2569 |
+
"learning_rate": 9.132650330981052e-07,
|
2570 |
+
"loss": 0.1012,
|
2571 |
+
"step": 9150
|
2572 |
+
},
|
2573 |
+
{
|
2574 |
+
"epoch": 0.9059044233807267,
|
2575 |
+
"grad_norm": 0.07581055909395218,
|
2576 |
+
"learning_rate": 8.675075780631559e-07,
|
2577 |
+
"loss": 0.0736,
|
2578 |
+
"step": 9175
|
2579 |
+
},
|
2580 |
+
{
|
2581 |
+
"epoch": 0.9083728278041074,
|
2582 |
+
"grad_norm": 0.5248848795890808,
|
2583 |
+
"learning_rate": 8.229006631048086e-07,
|
2584 |
+
"loss": 0.028,
|
2585 |
+
"step": 9200
|
2586 |
+
},
|
2587 |
+
{
|
2588 |
+
"epoch": 0.9108412322274881,
|
2589 |
+
"grad_norm": 0.6023635268211365,
|
2590 |
+
"learning_rate": 7.79446970677713e-07,
|
2591 |
+
"loss": 0.1284,
|
2592 |
+
"step": 9225
|
2593 |
+
},
|
2594 |
+
{
|
2595 |
+
"epoch": 0.9133096366508688,
|
2596 |
+
"grad_norm": 63.97990798950195,
|
2597 |
+
"learning_rate": 7.371491138870257e-07,
|
2598 |
+
"loss": 0.1844,
|
2599 |
+
"step": 9250
|
2600 |
+
},
|
2601 |
+
{
|
2602 |
+
"epoch": 0.9157780410742496,
|
2603 |
+
"grad_norm": 0.31825193762779236,
|
2604 |
+
"learning_rate": 6.960096363312674e-07,
|
2605 |
+
"loss": 0.0615,
|
2606 |
+
"step": 9275
|
2607 |
+
},
|
2608 |
+
{
|
2609 |
+
"epoch": 0.9182464454976303,
|
2610 |
+
"grad_norm": 10.776803970336914,
|
2611 |
+
"learning_rate": 6.560310119493607e-07,
|
2612 |
+
"loss": 0.1117,
|
2613 |
+
"step": 9300
|
2614 |
+
},
|
2615 |
+
{
|
2616 |
+
"epoch": 0.920714849921011,
|
2617 |
+
"grad_norm": 0.047084905207157135,
|
2618 |
+
"learning_rate": 6.172156448718535e-07,
|
2619 |
+
"loss": 0.1446,
|
2620 |
+
"step": 9325
|
2621 |
+
},
|
2622 |
+
{
|
2623 |
+
"epoch": 0.9231832543443917,
|
2624 |
+
"grad_norm": 0.04456774890422821,
|
2625 |
+
"learning_rate": 5.795658692763595e-07,
|
2626 |
+
"loss": 0.1223,
|
2627 |
+
"step": 9350
|
2628 |
+
},
|
2629 |
+
{
|
2630 |
+
"epoch": 0.9256516587677726,
|
2631 |
+
"grad_norm": 0.023414991796016693,
|
2632 |
+
"learning_rate": 5.430839492471718e-07,
|
2633 |
+
"loss": 0.0981,
|
2634 |
+
"step": 9375
|
2635 |
+
},
|
2636 |
+
{
|
2637 |
+
"epoch": 0.9281200631911533,
|
2638 |
+
"grad_norm": 0.1359195113182068,
|
2639 |
+
"learning_rate": 5.077720786391327e-07,
|
2640 |
+
"loss": 0.0668,
|
2641 |
+
"step": 9400
|
2642 |
+
},
|
2643 |
+
{
|
2644 |
+
"epoch": 0.930588467614534,
|
2645 |
+
"grad_norm": 0.2277175486087799,
|
2646 |
+
"learning_rate": 4.7363238094569e-07,
|
2647 |
+
"loss": 0.174,
|
2648 |
+
"step": 9425
|
2649 |
+
},
|
2650 |
+
{
|
2651 |
+
"epoch": 0.9330568720379147,
|
2652 |
+
"grad_norm": 0.036555130034685135,
|
2653 |
+
"learning_rate": 4.406669091712057e-07,
|
2654 |
+
"loss": 0.1226,
|
2655 |
+
"step": 9450
|
2656 |
+
},
|
2657 |
+
{
|
2658 |
+
"epoch": 0.9355252764612955,
|
2659 |
+
"grad_norm": 0.11190089583396912,
|
2660 |
+
"learning_rate": 4.08877645707495e-07,
|
2661 |
+
"loss": 0.0718,
|
2662 |
+
"step": 9475
|
2663 |
+
},
|
2664 |
+
{
|
2665 |
+
"epoch": 0.9379936808846762,
|
2666 |
+
"grad_norm": 0.046454474329948425,
|
2667 |
+
"learning_rate": 3.7826650221462367e-07,
|
2668 |
+
"loss": 0.108,
|
2669 |
+
"step": 9500
|
2670 |
+
},
|
2671 |
+
{
|
2672 |
+
"epoch": 0.9404620853080569,
|
2673 |
+
"grad_norm": 59.002662658691406,
|
2674 |
+
"learning_rate": 3.488353195059313e-07,
|
2675 |
+
"loss": 0.1669,
|
2676 |
+
"step": 9525
|
2677 |
+
},
|
2678 |
+
{
|
2679 |
+
"epoch": 0.9429304897314376,
|
2680 |
+
"grad_norm": 0.10425674915313721,
|
2681 |
+
"learning_rate": 3.205858674373508e-07,
|
2682 |
+
"loss": 0.1286,
|
2683 |
+
"step": 9550
|
2684 |
+
},
|
2685 |
+
{
|
2686 |
+
"epoch": 0.9453988941548184,
|
2687 |
+
"grad_norm": 26.614826202392578,
|
2688 |
+
"learning_rate": 2.9351984480097353e-07,
|
2689 |
+
"loss": 0.1456,
|
2690 |
+
"step": 9575
|
2691 |
+
},
|
2692 |
+
{
|
2693 |
+
"epoch": 0.9478672985781991,
|
2694 |
+
"grad_norm": 0.26959002017974854,
|
2695 |
+
"learning_rate": 2.6763887922288236e-07,
|
2696 |
+
"loss": 0.0575,
|
2697 |
+
"step": 9600
|
2698 |
+
},
|
2699 |
+
{
|
2700 |
+
"epoch": 0.9503357030015798,
|
2701 |
+
"grad_norm": 0.12144956737756729,
|
2702 |
+
"learning_rate": 2.4294452706528526e-07,
|
2703 |
+
"loss": 0.0818,
|
2704 |
+
"step": 9625
|
2705 |
+
},
|
2706 |
+
{
|
2707 |
+
"epoch": 0.9528041074249605,
|
2708 |
+
"grad_norm": 0.5530775189399719,
|
2709 |
+
"learning_rate": 2.1943827333291478e-07,
|
2710 |
+
"loss": 0.069,
|
2711 |
+
"step": 9650
|
2712 |
+
},
|
2713 |
+
{
|
2714 |
+
"epoch": 0.9552725118483413,
|
2715 |
+
"grad_norm": 0.6592720746994019,
|
2716 |
+
"learning_rate": 1.9712153158373493e-07,
|
2717 |
+
"loss": 0.1226,
|
2718 |
+
"step": 9675
|
2719 |
+
},
|
2720 |
+
{
|
2721 |
+
"epoch": 0.957740916271722,
|
2722 |
+
"grad_norm": 0.34944581985473633,
|
2723 |
+
"learning_rate": 1.7599564384392255e-07,
|
2724 |
+
"loss": 0.1611,
|
2725 |
+
"step": 9700
|
2726 |
+
},
|
2727 |
+
{
|
2728 |
+
"epoch": 0.9602093206951027,
|
2729 |
+
"grad_norm": 0.5341227650642395,
|
2730 |
+
"learning_rate": 1.5606188052718296e-07,
|
2731 |
+
"loss": 0.1514,
|
2732 |
+
"step": 9725
|
2733 |
+
},
|
2734 |
+
{
|
2735 |
+
"epoch": 0.9626777251184834,
|
2736 |
+
"grad_norm": 18.92991065979004,
|
2737 |
+
"learning_rate": 1.3732144035833784e-07,
|
2738 |
+
"loss": 0.0454,
|
2739 |
+
"step": 9750
|
2740 |
+
},
|
2741 |
+
{
|
2742 |
+
"epoch": 0.9651461295418642,
|
2743 |
+
"grad_norm": 0.335257351398468,
|
2744 |
+
"learning_rate": 1.1977545030124048e-07,
|
2745 |
+
"loss": 0.1908,
|
2746 |
+
"step": 9775
|
2747 |
+
},
|
2748 |
+
{
|
2749 |
+
"epoch": 0.9676145339652449,
|
2750 |
+
"grad_norm": 9.972683906555176,
|
2751 |
+
"learning_rate": 1.034249654910191e-07,
|
2752 |
+
"loss": 0.1027,
|
2753 |
+
"step": 9800
|
2754 |
+
},
|
2755 |
+
{
|
2756 |
+
"epoch": 0.9700829383886256,
|
2757 |
+
"grad_norm": 0.1992756426334381,
|
2758 |
+
"learning_rate": 8.827096917060962e-08,
|
2759 |
+
"loss": 0.1122,
|
2760 |
+
"step": 9825
|
2761 |
+
},
|
2762 |
+
{
|
2763 |
+
"epoch": 0.9725513428120063,
|
2764 |
+
"grad_norm": 23.53179931640625,
|
2765 |
+
"learning_rate": 7.431437263163421e-08,
|
2766 |
+
"loss": 0.079,
|
2767 |
+
"step": 9850
|
2768 |
+
},
|
2769 |
+
{
|
2770 |
+
"epoch": 0.9750197472353871,
|
2771 |
+
"grad_norm": 0.09356188029050827,
|
2772 |
+
"learning_rate": 6.155601515960063e-08,
|
2773 |
+
"loss": 0.125,
|
2774 |
+
"step": 9875
|
2775 |
+
},
|
2776 |
+
{
|
2777 |
+
"epoch": 0.9774881516587678,
|
2778 |
+
"grad_norm": 9.981024742126465,
|
2779 |
+
"learning_rate": 4.999666398343816e-08,
|
2780 |
+
"loss": 0.1113,
|
2781 |
+
"step": 9900
|
2782 |
+
},
|
2783 |
+
{
|
2784 |
+
"epoch": 0.9799565560821485,
|
2785 |
+
"grad_norm": 0.2773338258266449,
|
2786 |
+
"learning_rate": 3.963701422934563e-08,
|
2787 |
+
"loss": 0.0509,
|
2788 |
+
"step": 9925
|
2789 |
+
},
|
2790 |
+
{
|
2791 |
+
"epoch": 0.9824249605055292,
|
2792 |
+
"grad_norm": 0.07679921388626099,
|
2793 |
+
"learning_rate": 3.04776888790026e-08,
|
2794 |
+
"loss": 0.0689,
|
2795 |
+
"step": 9950
|
2796 |
+
},
|
2797 |
+
{
|
2798 |
+
"epoch": 0.98489336492891,
|
2799 |
+
"grad_norm": 7.201016426086426,
|
2800 |
+
"learning_rate": 2.2519238732094938e-08,
|
2801 |
+
"loss": 0.1382,
|
2802 |
+
"step": 9975
|
2803 |
+
},
|
2804 |
+
{
|
2805 |
+
"epoch": 0.9873617693522907,
|
2806 |
+
"grad_norm": 15.387526512145996,
|
2807 |
+
"learning_rate": 1.5762142373205726e-08,
|
2808 |
+
"loss": 0.0728,
|
2809 |
+
"step": 10000
|
2810 |
+
},
|
2811 |
+
{
|
2812 |
+
"epoch": 0.9898301737756714,
|
2813 |
+
"grad_norm": 213.41201782226562,
|
2814 |
+
"learning_rate": 1.0206806143024939e-08,
|
2815 |
+
"loss": 0.1574,
|
2816 |
+
"step": 10025
|
2817 |
+
},
|
2818 |
+
{
|
2819 |
+
"epoch": 0.9922985781990521,
|
2820 |
+
"grad_norm": 112.1288070678711,
|
2821 |
+
"learning_rate": 5.853564113913468e-09,
|
2822 |
+
"loss": 0.0917,
|
2823 |
+
"step": 10050
|
2824 |
+
},
|
2825 |
+
{
|
2826 |
+
"epoch": 0.9947669826224329,
|
2827 |
+
"grad_norm": 0.5294753313064575,
|
2828 |
+
"learning_rate": 2.702678069821385e-09,
|
2829 |
+
"loss": 0.1369,
|
2830 |
+
"step": 10075
|
2831 |
+
},
|
2832 |
+
{
|
2833 |
+
"epoch": 0.9972353870458136,
|
2834 |
+
"grad_norm": 0.19094215333461761,
|
2835 |
+
"learning_rate": 7.543374905405465e-10,
|
2836 |
+
"loss": 0.084,
|
2837 |
+
"step": 10100
|
2838 |
+
},
|
2839 |
+
{
|
2840 |
+
"epoch": 0.9997037914691943,
|
2841 |
+
"grad_norm": 0.4669792950153351,
|
2842 |
+
"learning_rate": 8.659540309263748e-12,
|
2843 |
+
"loss": 0.0898,
|
2844 |
+
"step": 10125
|
2845 |
+
},
|
2846 |
+
{
|
2847 |
+
"epoch": 1.0,
|
2848 |
+
"eval_accuracy": 0.9661280079034649,
|
2849 |
+
"eval_auc": 0.9842785960253444,
|
2850 |
+
"eval_f1": 0.9689199689199689,
|
2851 |
+
"eval_loss": 0.17222872376441956,
|
2852 |
+
"eval_precision": 0.963554410817772,
|
2853 |
+
"eval_recall": 0.9743456179189999,
|
2854 |
+
"eval_runtime": 9522.2322,
|
2855 |
+
"eval_samples_per_second": 1.488,
|
2856 |
+
"eval_steps_per_second": 0.062,
|
2857 |
+
"step": 10128
|
2858 |
+
}
|
2859 |
+
],
|
2860 |
+
"logging_steps": 25,
|
2861 |
+
"max_steps": 10128,
|
2862 |
+
"num_input_tokens_seen": 0,
|
2863 |
+
"num_train_epochs": 1,
|
2864 |
+
"save_steps": 500,
|
2865 |
+
"stateful_callbacks": {
|
2866 |
+
"EarlyStoppingCallback": {
|
2867 |
+
"args": {
|
2868 |
+
"early_stopping_patience": 5,
|
2869 |
+
"early_stopping_threshold": 0.01
|
2870 |
+
},
|
2871 |
+
"attributes": {
|
2872 |
+
"early_stopping_patience_counter": 0
|
2873 |
+
}
|
2874 |
+
},
|
2875 |
+
"TrainerControl": {
|
2876 |
+
"args": {
|
2877 |
+
"should_epoch_stop": false,
|
2878 |
+
"should_evaluate": false,
|
2879 |
+
"should_log": false,
|
2880 |
+
"should_save": true,
|
2881 |
+
"should_training_stop": true
|
2882 |
+
},
|
2883 |
+
"attributes": {}
|
2884 |
+
}
|
2885 |
+
},
|
2886 |
+
"total_flos": 2.7984323643494105e+19,
|
2887 |
+
"train_batch_size": 12,
|
2888 |
+
"trial_name": null,
|
2889 |
+
"trial_params": null
|
2890 |
+
}
|
{checkpoint-9995 β checkpoint-10128}/training_args.bin
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:23acb2db99f02f49dd8f0a9e1af4931c2df606823265dce45ae76a632ad77be5
|
3 |
size 5304
|
checkpoint-9995/trainer_state.json
DELETED
@@ -1,2848 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": 0.16612961888313293,
|
3 |
-
"best_model_checkpoint": "autotrain-ai-image-detect-20241117-0124/checkpoint-9995",
|
4 |
-
"epoch": 1.0,
|
5 |
-
"eval_steps": 500,
|
6 |
-
"global_step": 9995,
|
7 |
-
"is_hyper_param_search": false,
|
8 |
-
"is_local_process_zero": true,
|
9 |
-
"is_world_process_zero": true,
|
10 |
-
"log_history": [
|
11 |
-
{
|
12 |
-
"epoch": 0.0025012506253126563,
|
13 |
-
"grad_norm": 0.15264733135700226,
|
14 |
-
"learning_rate": 3.999938253558893e-05,
|
15 |
-
"loss": 0.05,
|
16 |
-
"step": 25
|
17 |
-
},
|
18 |
-
{
|
19 |
-
"epoch": 0.0050025012506253125,
|
20 |
-
"grad_norm": 0.039762478321790695,
|
21 |
-
"learning_rate": 3.9997530180481925e-05,
|
22 |
-
"loss": 0.1129,
|
23 |
-
"step": 50
|
24 |
-
},
|
25 |
-
{
|
26 |
-
"epoch": 0.007503751875937969,
|
27 |
-
"grad_norm": 0.1116345226764679,
|
28 |
-
"learning_rate": 3.999444304905533e-05,
|
29 |
-
"loss": 0.0899,
|
30 |
-
"step": 75
|
31 |
-
},
|
32 |
-
{
|
33 |
-
"epoch": 0.010005002501250625,
|
34 |
-
"grad_norm": 0.13316667079925537,
|
35 |
-
"learning_rate": 3.999012133192853e-05,
|
36 |
-
"loss": 0.1013,
|
37 |
-
"step": 100
|
38 |
-
},
|
39 |
-
{
|
40 |
-
"epoch": 0.01250625312656328,
|
41 |
-
"grad_norm": 0.3281913697719574,
|
42 |
-
"learning_rate": 3.998456529595217e-05,
|
43 |
-
"loss": 0.0737,
|
44 |
-
"step": 125
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.015007503751875938,
|
48 |
-
"grad_norm": 1.4385777711868286,
|
49 |
-
"learning_rate": 3.997777528419169e-05,
|
50 |
-
"loss": 0.1332,
|
51 |
-
"step": 150
|
52 |
-
},
|
53 |
-
{
|
54 |
-
"epoch": 0.017508754377188594,
|
55 |
-
"grad_norm": 2.8727269172668457,
|
56 |
-
"learning_rate": 3.996975171590617e-05,
|
57 |
-
"loss": 0.0701,
|
58 |
-
"step": 175
|
59 |
-
},
|
60 |
-
{
|
61 |
-
"epoch": 0.02001000500250125,
|
62 |
-
"grad_norm": 0.039399586617946625,
|
63 |
-
"learning_rate": 3.9960495086522394e-05,
|
64 |
-
"loss": 0.1557,
|
65 |
-
"step": 200
|
66 |
-
},
|
67 |
-
{
|
68 |
-
"epoch": 0.022511255627813906,
|
69 |
-
"grad_norm": 9.723877906799316,
|
70 |
-
"learning_rate": 3.995000596760426e-05,
|
71 |
-
"loss": 0.0918,
|
72 |
-
"step": 225
|
73 |
-
},
|
74 |
-
{
|
75 |
-
"epoch": 0.02501250625312656,
|
76 |
-
"grad_norm": 0.3538779020309448,
|
77 |
-
"learning_rate": 3.993828500681756e-05,
|
78 |
-
"loss": 0.175,
|
79 |
-
"step": 250
|
80 |
-
},
|
81 |
-
{
|
82 |
-
"epoch": 0.02751375687843922,
|
83 |
-
"grad_norm": 0.041708994656801224,
|
84 |
-
"learning_rate": 3.9925332927889895e-05,
|
85 |
-
"loss": 0.079,
|
86 |
-
"step": 275
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.030015007503751877,
|
90 |
-
"grad_norm": 2.006554365158081,
|
91 |
-
"learning_rate": 3.9911150530566044e-05,
|
92 |
-
"loss": 0.1486,
|
93 |
-
"step": 300
|
94 |
-
},
|
95 |
-
{
|
96 |
-
"epoch": 0.03251625812906453,
|
97 |
-
"grad_norm": 0.31543242931365967,
|
98 |
-
"learning_rate": 3.989573869055857e-05,
|
99 |
-
"loss": 0.1731,
|
100 |
-
"step": 325
|
101 |
-
},
|
102 |
-
{
|
103 |
-
"epoch": 0.03501750875437719,
|
104 |
-
"grad_norm": 0.05023409426212311,
|
105 |
-
"learning_rate": 3.987909835949374e-05,
|
106 |
-
"loss": 0.0994,
|
107 |
-
"step": 350
|
108 |
-
},
|
109 |
-
{
|
110 |
-
"epoch": 0.03751875937968985,
|
111 |
-
"grad_norm": 0.38723427057266235,
|
112 |
-
"learning_rate": 3.986123056485279e-05,
|
113 |
-
"loss": 0.1395,
|
114 |
-
"step": 375
|
115 |
-
},
|
116 |
-
{
|
117 |
-
"epoch": 0.0400200100050025,
|
118 |
-
"grad_norm": 0.5986870527267456,
|
119 |
-
"learning_rate": 3.9842136409908436e-05,
|
120 |
-
"loss": 0.1507,
|
121 |
-
"step": 400
|
122 |
-
},
|
123 |
-
{
|
124 |
-
"epoch": 0.04252126063031516,
|
125 |
-
"grad_norm": 0.05096159875392914,
|
126 |
-
"learning_rate": 3.982181707365679e-05,
|
127 |
-
"loss": 0.155,
|
128 |
-
"step": 425
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.04502251125562781,
|
132 |
-
"grad_norm": 0.19909246265888214,
|
133 |
-
"learning_rate": 3.9800273810744564e-05,
|
134 |
-
"loss": 0.143,
|
135 |
-
"step": 450
|
136 |
-
},
|
137 |
-
{
|
138 |
-
"epoch": 0.04752376188094047,
|
139 |
-
"grad_norm": 0.7048402428627014,
|
140 |
-
"learning_rate": 3.977750795139156e-05,
|
141 |
-
"loss": 0.1563,
|
142 |
-
"step": 475
|
143 |
-
},
|
144 |
-
{
|
145 |
-
"epoch": 0.05002501250625312,
|
146 |
-
"grad_norm": 25.009944915771484,
|
147 |
-
"learning_rate": 3.975352090130859e-05,
|
148 |
-
"loss": 0.0758,
|
149 |
-
"step": 500
|
150 |
-
},
|
151 |
-
{
|
152 |
-
"epoch": 0.05252626313156578,
|
153 |
-
"grad_norm": 1.1721696853637695,
|
154 |
-
"learning_rate": 3.9728314141610596e-05,
|
155 |
-
"loss": 0.1535,
|
156 |
-
"step": 525
|
157 |
-
},
|
158 |
-
{
|
159 |
-
"epoch": 0.05502751375687844,
|
160 |
-
"grad_norm": 0.11145307123661041,
|
161 |
-
"learning_rate": 3.970188922872532e-05,
|
162 |
-
"loss": 0.1293,
|
163 |
-
"step": 550
|
164 |
-
},
|
165 |
-
{
|
166 |
-
"epoch": 0.057528764382191094,
|
167 |
-
"grad_norm": 0.05586661025881767,
|
168 |
-
"learning_rate": 3.9674247794297065e-05,
|
169 |
-
"loss": 0.0719,
|
170 |
-
"step": 575
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.060030015007503754,
|
174 |
-
"grad_norm": 0.045262664556503296,
|
175 |
-
"learning_rate": 3.964539154508604e-05,
|
176 |
-
"loss": 0.1057,
|
177 |
-
"step": 600
|
178 |
-
},
|
179 |
-
{
|
180 |
-
"epoch": 0.06253126563281641,
|
181 |
-
"grad_norm": 1.082116961479187,
|
182 |
-
"learning_rate": 3.961532226286293e-05,
|
183 |
-
"loss": 0.1245,
|
184 |
-
"step": 625
|
185 |
-
},
|
186 |
-
{
|
187 |
-
"epoch": 0.06503251625812906,
|
188 |
-
"grad_norm": 27.530029296875,
|
189 |
-
"learning_rate": 3.958404180429891e-05,
|
190 |
-
"loss": 0.1579,
|
191 |
-
"step": 650
|
192 |
-
},
|
193 |
-
{
|
194 |
-
"epoch": 0.06753376688344172,
|
195 |
-
"grad_norm": 0.16050750017166138,
|
196 |
-
"learning_rate": 3.955155210085097e-05,
|
197 |
-
"loss": 0.1229,
|
198 |
-
"step": 675
|
199 |
-
},
|
200 |
-
{
|
201 |
-
"epoch": 0.07003501750875438,
|
202 |
-
"grad_norm": 0.7139843106269836,
|
203 |
-
"learning_rate": 3.9517855158642685e-05,
|
204 |
-
"loss": 0.0758,
|
205 |
-
"step": 700
|
206 |
-
},
|
207 |
-
{
|
208 |
-
"epoch": 0.07253626813406704,
|
209 |
-
"grad_norm": 0.021592242643237114,
|
210 |
-
"learning_rate": 3.9482953058340275e-05,
|
211 |
-
"loss": 0.0486,
|
212 |
-
"step": 725
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.0750375187593797,
|
216 |
-
"grad_norm": 59.190853118896484,
|
217 |
-
"learning_rate": 3.944684795502426e-05,
|
218 |
-
"loss": 0.1511,
|
219 |
-
"step": 750
|
220 |
-
},
|
221 |
-
{
|
222 |
-
"epoch": 0.07753876938469234,
|
223 |
-
"grad_norm": 0.5598094463348389,
|
224 |
-
"learning_rate": 3.940954207805625e-05,
|
225 |
-
"loss": 0.1391,
|
226 |
-
"step": 775
|
227 |
-
},
|
228 |
-
{
|
229 |
-
"epoch": 0.080040020010005,
|
230 |
-
"grad_norm": 588.6250610351562,
|
231 |
-
"learning_rate": 3.937103773094139e-05,
|
232 |
-
"loss": 0.149,
|
233 |
-
"step": 800
|
234 |
-
},
|
235 |
-
{
|
236 |
-
"epoch": 0.08254127063531766,
|
237 |
-
"grad_norm": 0.17757241427898407,
|
238 |
-
"learning_rate": 3.9331337291186084e-05,
|
239 |
-
"loss": 0.2046,
|
240 |
-
"step": 825
|
241 |
-
},
|
242 |
-
{
|
243 |
-
"epoch": 0.08504252126063032,
|
244 |
-
"grad_norm": 0.26129040122032166,
|
245 |
-
"learning_rate": 3.929044321015119e-05,
|
246 |
-
"loss": 0.1072,
|
247 |
-
"step": 850
|
248 |
-
},
|
249 |
-
{
|
250 |
-
"epoch": 0.08754377188594298,
|
251 |
-
"grad_norm": 0.1557866334915161,
|
252 |
-
"learning_rate": 3.924835801290067e-05,
|
253 |
-
"loss": 0.1531,
|
254 |
-
"step": 875
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.09004502251125562,
|
258 |
-
"grad_norm": 0.7562702894210815,
|
259 |
-
"learning_rate": 3.92050842980457e-05,
|
260 |
-
"loss": 0.2044,
|
261 |
-
"step": 900
|
262 |
-
},
|
263 |
-
{
|
264 |
-
"epoch": 0.09254627313656828,
|
265 |
-
"grad_norm": 0.5599152445793152,
|
266 |
-
"learning_rate": 3.9160624737584145e-05,
|
267 |
-
"loss": 0.1156,
|
268 |
-
"step": 925
|
269 |
-
},
|
270 |
-
{
|
271 |
-
"epoch": 0.09504752376188094,
|
272 |
-
"grad_norm": 0.09455389529466629,
|
273 |
-
"learning_rate": 3.911498207673564e-05,
|
274 |
-
"loss": 0.1731,
|
275 |
-
"step": 950
|
276 |
-
},
|
277 |
-
{
|
278 |
-
"epoch": 0.0975487743871936,
|
279 |
-
"grad_norm": 0.09220050275325775,
|
280 |
-
"learning_rate": 3.9068159133772063e-05,
|
281 |
-
"loss": 0.1238,
|
282 |
-
"step": 975
|
283 |
-
},
|
284 |
-
{
|
285 |
-
"epoch": 0.10005002501250625,
|
286 |
-
"grad_norm": 0.07393497228622437,
|
287 |
-
"learning_rate": 3.90201587998435e-05,
|
288 |
-
"loss": 0.1313,
|
289 |
-
"step": 1000
|
290 |
-
},
|
291 |
-
{
|
292 |
-
"epoch": 0.1025512756378189,
|
293 |
-
"grad_norm": 12.711342811584473,
|
294 |
-
"learning_rate": 3.897098403879974e-05,
|
295 |
-
"loss": 0.0919,
|
296 |
-
"step": 1025
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.10505252626313157,
|
300 |
-
"grad_norm": 2.502979278564453,
|
301 |
-
"learning_rate": 3.8920637887007274e-05,
|
302 |
-
"loss": 0.0928,
|
303 |
-
"step": 1050
|
304 |
-
},
|
305 |
-
{
|
306 |
-
"epoch": 0.10755377688844422,
|
307 |
-
"grad_norm": 0.10346478968858719,
|
308 |
-
"learning_rate": 3.886912345316179e-05,
|
309 |
-
"loss": 0.0734,
|
310 |
-
"step": 1075
|
311 |
-
},
|
312 |
-
{
|
313 |
-
"epoch": 0.11005502751375688,
|
314 |
-
"grad_norm": 0.06958428025245667,
|
315 |
-
"learning_rate": 3.8816443918096254e-05,
|
316 |
-
"loss": 0.0942,
|
317 |
-
"step": 1100
|
318 |
-
},
|
319 |
-
{
|
320 |
-
"epoch": 0.11255627813906953,
|
321 |
-
"grad_norm": 0.03277590125799179,
|
322 |
-
"learning_rate": 3.8762602534584475e-05,
|
323 |
-
"loss": 0.213,
|
324 |
-
"step": 1125
|
325 |
-
},
|
326 |
-
{
|
327 |
-
"epoch": 0.11505752876438219,
|
328 |
-
"grad_norm": 1.322001338005066,
|
329 |
-
"learning_rate": 3.870760262714027e-05,
|
330 |
-
"loss": 0.1272,
|
331 |
-
"step": 1150
|
332 |
-
},
|
333 |
-
{
|
334 |
-
"epoch": 0.11755877938969485,
|
335 |
-
"grad_norm": 0.11594086140394211,
|
336 |
-
"learning_rate": 3.865144759181218e-05,
|
337 |
-
"loss": 0.0728,
|
338 |
-
"step": 1175
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 0.12006003001500751,
|
342 |
-
"grad_norm": 0.08336304128170013,
|
343 |
-
"learning_rate": 3.859414089597378e-05,
|
344 |
-
"loss": 0.0825,
|
345 |
-
"step": 1200
|
346 |
-
},
|
347 |
-
{
|
348 |
-
"epoch": 0.12256128064032017,
|
349 |
-
"grad_norm": 0.6295961737632751,
|
350 |
-
"learning_rate": 3.8535686078109603e-05,
|
351 |
-
"loss": 0.1302,
|
352 |
-
"step": 1225
|
353 |
-
},
|
354 |
-
{
|
355 |
-
"epoch": 0.12506253126563283,
|
356 |
-
"grad_norm": 2.7103772163391113,
|
357 |
-
"learning_rate": 3.847608674759661e-05,
|
358 |
-
"loss": 0.1055,
|
359 |
-
"step": 1250
|
360 |
-
},
|
361 |
-
{
|
362 |
-
"epoch": 0.12756378189094547,
|
363 |
-
"grad_norm": 0.022725675255060196,
|
364 |
-
"learning_rate": 3.8415346584481354e-05,
|
365 |
-
"loss": 0.2902,
|
366 |
-
"step": 1275
|
367 |
-
},
|
368 |
-
{
|
369 |
-
"epoch": 0.13006503251625812,
|
370 |
-
"grad_norm": 24.051122665405273,
|
371 |
-
"learning_rate": 3.835346933925274e-05,
|
372 |
-
"loss": 0.1207,
|
373 |
-
"step": 1300
|
374 |
-
},
|
375 |
-
{
|
376 |
-
"epoch": 0.1325662831415708,
|
377 |
-
"grad_norm": 0.748214602470398,
|
378 |
-
"learning_rate": 3.829045883261045e-05,
|
379 |
-
"loss": 0.1993,
|
380 |
-
"step": 1325
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 0.13506753376688344,
|
384 |
-
"grad_norm": 0.6024086475372314,
|
385 |
-
"learning_rate": 3.822631895522902e-05,
|
386 |
-
"loss": 0.081,
|
387 |
-
"step": 1350
|
388 |
-
},
|
389 |
-
{
|
390 |
-
"epoch": 0.1375687843921961,
|
391 |
-
"grad_norm": 0.028124479576945305,
|
392 |
-
"learning_rate": 3.816105366751761e-05,
|
393 |
-
"loss": 0.1006,
|
394 |
-
"step": 1375
|
395 |
-
},
|
396 |
-
{
|
397 |
-
"epoch": 0.14007003501750875,
|
398 |
-
"grad_norm": 0.2806219756603241,
|
399 |
-
"learning_rate": 3.809466699937546e-05,
|
400 |
-
"loss": 0.1156,
|
401 |
-
"step": 1400
|
402 |
-
},
|
403 |
-
{
|
404 |
-
"epoch": 0.1425712856428214,
|
405 |
-
"grad_norm": 18.823518753051758,
|
406 |
-
"learning_rate": 3.802716304994307e-05,
|
407 |
-
"loss": 0.1481,
|
408 |
-
"step": 1425
|
409 |
-
},
|
410 |
-
{
|
411 |
-
"epoch": 0.14507253626813407,
|
412 |
-
"grad_norm": 0.1195753738284111,
|
413 |
-
"learning_rate": 3.795854598734907e-05,
|
414 |
-
"loss": 0.1735,
|
415 |
-
"step": 1450
|
416 |
-
},
|
417 |
-
{
|
418 |
-
"epoch": 0.14757378689344672,
|
419 |
-
"grad_norm": 0.05460454151034355,
|
420 |
-
"learning_rate": 3.78888200484529e-05,
|
421 |
-
"loss": 0.16,
|
422 |
-
"step": 1475
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 0.1500750375187594,
|
426 |
-
"grad_norm": 7.615594863891602,
|
427 |
-
"learning_rate": 3.78179895385831e-05,
|
428 |
-
"loss": 0.1093,
|
429 |
-
"step": 1500
|
430 |
-
},
|
431 |
-
{
|
432 |
-
"epoch": 0.15257628814407204,
|
433 |
-
"grad_norm": 0.10332254320383072,
|
434 |
-
"learning_rate": 3.7746058831271615e-05,
|
435 |
-
"loss": 0.0666,
|
436 |
-
"step": 1525
|
437 |
-
},
|
438 |
-
{
|
439 |
-
"epoch": 0.15507753876938468,
|
440 |
-
"grad_norm": 0.16424693167209625,
|
441 |
-
"learning_rate": 3.76730323679836e-05,
|
442 |
-
"loss": 0.094,
|
443 |
-
"step": 1550
|
444 |
-
},
|
445 |
-
{
|
446 |
-
"epoch": 0.15757878939469736,
|
447 |
-
"grad_norm": 0.9435814619064331,
|
448 |
-
"learning_rate": 3.759891465784328e-05,
|
449 |
-
"loss": 0.0873,
|
450 |
-
"step": 1575
|
451 |
-
},
|
452 |
-
{
|
453 |
-
"epoch": 0.16008004002001,
|
454 |
-
"grad_norm": 8.945067405700684,
|
455 |
-
"learning_rate": 3.7523710277355486e-05,
|
456 |
-
"loss": 0.1503,
|
457 |
-
"step": 1600
|
458 |
-
},
|
459 |
-
{
|
460 |
-
"epoch": 0.16258129064532267,
|
461 |
-
"grad_norm": 0.011415637098252773,
|
462 |
-
"learning_rate": 3.744742387012305e-05,
|
463 |
-
"loss": 0.1368,
|
464 |
-
"step": 1625
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 0.16508254127063532,
|
468 |
-
"grad_norm": 38.10102081298828,
|
469 |
-
"learning_rate": 3.7370060146560146e-05,
|
470 |
-
"loss": 0.1063,
|
471 |
-
"step": 1650
|
472 |
-
},
|
473 |
-
{
|
474 |
-
"epoch": 0.16758379189594796,
|
475 |
-
"grad_norm": 0.10745133459568024,
|
476 |
-
"learning_rate": 3.729162388360136e-05,
|
477 |
-
"loss": 0.1424,
|
478 |
-
"step": 1675
|
479 |
-
},
|
480 |
-
{
|
481 |
-
"epoch": 0.17008504252126064,
|
482 |
-
"grad_norm": 0.040263831615448,
|
483 |
-
"learning_rate": 3.721211992440679e-05,
|
484 |
-
"loss": 0.0923,
|
485 |
-
"step": 1700
|
486 |
-
},
|
487 |
-
{
|
488 |
-
"epoch": 0.17258629314657328,
|
489 |
-
"grad_norm": 2.1761999130249023,
|
490 |
-
"learning_rate": 3.713155317806297e-05,
|
491 |
-
"loss": 0.1661,
|
492 |
-
"step": 1725
|
493 |
-
},
|
494 |
-
{
|
495 |
-
"epoch": 0.17508754377188596,
|
496 |
-
"grad_norm": 16.003284454345703,
|
497 |
-
"learning_rate": 3.7049928619279756e-05,
|
498 |
-
"loss": 0.0912,
|
499 |
-
"step": 1750
|
500 |
-
},
|
501 |
-
{
|
502 |
-
"epoch": 0.1775887943971986,
|
503 |
-
"grad_norm": 0.15643572807312012,
|
504 |
-
"learning_rate": 3.696725128808315e-05,
|
505 |
-
"loss": 0.1179,
|
506 |
-
"step": 1775
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 0.18009004502251125,
|
510 |
-
"grad_norm": 0.6343806385993958,
|
511 |
-
"learning_rate": 3.6883526289504136e-05,
|
512 |
-
"loss": 0.1431,
|
513 |
-
"step": 1800
|
514 |
-
},
|
515 |
-
{
|
516 |
-
"epoch": 0.18259129564782392,
|
517 |
-
"grad_norm": 0.13746272027492523,
|
518 |
-
"learning_rate": 3.679875879326339e-05,
|
519 |
-
"loss": 0.149,
|
520 |
-
"step": 1825
|
521 |
-
},
|
522 |
-
{
|
523 |
-
"epoch": 0.18509254627313657,
|
524 |
-
"grad_norm": 0.23029863834381104,
|
525 |
-
"learning_rate": 3.671295403345213e-05,
|
526 |
-
"loss": 0.0951,
|
527 |
-
"step": 1850
|
528 |
-
},
|
529 |
-
{
|
530 |
-
"epoch": 0.1875937968984492,
|
531 |
-
"grad_norm": 0.6915459036827087,
|
532 |
-
"learning_rate": 3.6626117308208914e-05,
|
533 |
-
"loss": 0.1136,
|
534 |
-
"step": 1875
|
535 |
-
},
|
536 |
-
{
|
537 |
-
"epoch": 0.19009504752376188,
|
538 |
-
"grad_norm": 0.6550763845443726,
|
539 |
-
"learning_rate": 3.653825397939248e-05,
|
540 |
-
"loss": 0.1329,
|
541 |
-
"step": 1900
|
542 |
-
},
|
543 |
-
{
|
544 |
-
"epoch": 0.19259629814907453,
|
545 |
-
"grad_norm": 11.350174903869629,
|
546 |
-
"learning_rate": 3.644936947225067e-05,
|
547 |
-
"loss": 0.1139,
|
548 |
-
"step": 1925
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 0.1950975487743872,
|
552 |
-
"grad_norm": 0.08577147871255875,
|
553 |
-
"learning_rate": 3.635946927508549e-05,
|
554 |
-
"loss": 0.1889,
|
555 |
-
"step": 1950
|
556 |
-
},
|
557 |
-
{
|
558 |
-
"epoch": 0.19759879939969985,
|
559 |
-
"grad_norm": 0.2987195551395416,
|
560 |
-
"learning_rate": 3.626855893891416e-05,
|
561 |
-
"loss": 0.1069,
|
562 |
-
"step": 1975
|
563 |
-
},
|
564 |
-
{
|
565 |
-
"epoch": 0.2001000500250125,
|
566 |
-
"grad_norm": 19.672019958496094,
|
567 |
-
"learning_rate": 3.61766440771264e-05,
|
568 |
-
"loss": 0.198,
|
569 |
-
"step": 2000
|
570 |
-
},
|
571 |
-
{
|
572 |
-
"epoch": 0.20260130065032517,
|
573 |
-
"grad_norm": 0.2886382043361664,
|
574 |
-
"learning_rate": 3.608373036513781e-05,
|
575 |
-
"loss": 0.1451,
|
576 |
-
"step": 2025
|
577 |
-
},
|
578 |
-
{
|
579 |
-
"epoch": 0.2051025512756378,
|
580 |
-
"grad_norm": 0.19703011214733124,
|
581 |
-
"learning_rate": 3.598982354003944e-05,
|
582 |
-
"loss": 0.1664,
|
583 |
-
"step": 2050
|
584 |
-
},
|
585 |
-
{
|
586 |
-
"epoch": 0.20760380190095049,
|
587 |
-
"grad_norm": 16.22993278503418,
|
588 |
-
"learning_rate": 3.5894929400243525e-05,
|
589 |
-
"loss": 0.1376,
|
590 |
-
"step": 2075
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 0.21010505252626313,
|
594 |
-
"grad_norm": 0.11190354079008102,
|
595 |
-
"learning_rate": 3.579905380512549e-05,
|
596 |
-
"loss": 0.1199,
|
597 |
-
"step": 2100
|
598 |
-
},
|
599 |
-
{
|
600 |
-
"epoch": 0.21260630315157578,
|
601 |
-
"grad_norm": 0.8343837857246399,
|
602 |
-
"learning_rate": 3.570220267466212e-05,
|
603 |
-
"loss": 0.1509,
|
604 |
-
"step": 2125
|
605 |
-
},
|
606 |
-
{
|
607 |
-
"epoch": 0.21510755377688845,
|
608 |
-
"grad_norm": 0.19906936585903168,
|
609 |
-
"learning_rate": 3.5604381989066036e-05,
|
610 |
-
"loss": 0.0574,
|
611 |
-
"step": 2150
|
612 |
-
},
|
613 |
-
{
|
614 |
-
"epoch": 0.2176088044022011,
|
615 |
-
"grad_norm": 0.06218745931982994,
|
616 |
-
"learning_rate": 3.550559778841644e-05,
|
617 |
-
"loss": 0.1123,
|
618 |
-
"step": 2175
|
619 |
-
},
|
620 |
-
{
|
621 |
-
"epoch": 0.22011005502751377,
|
622 |
-
"grad_norm": 3.7358133792877197,
|
623 |
-
"learning_rate": 3.540585617228616e-05,
|
624 |
-
"loss": 0.0779,
|
625 |
-
"step": 2200
|
626 |
-
},
|
627 |
-
{
|
628 |
-
"epoch": 0.2226113056528264,
|
629 |
-
"grad_norm": 0.21526701748371124,
|
630 |
-
"learning_rate": 3.5305163299365026e-05,
|
631 |
-
"loss": 0.1717,
|
632 |
-
"step": 2225
|
633 |
-
},
|
634 |
-
{
|
635 |
-
"epoch": 0.22511255627813906,
|
636 |
-
"grad_norm": 58.74745178222656,
|
637 |
-
"learning_rate": 3.520352538707959e-05,
|
638 |
-
"loss": 0.1029,
|
639 |
-
"step": 2250
|
640 |
-
},
|
641 |
-
{
|
642 |
-
"epoch": 0.22761380690345173,
|
643 |
-
"grad_norm": 0.8728958964347839,
|
644 |
-
"learning_rate": 3.5100948711209194e-05,
|
645 |
-
"loss": 0.1099,
|
646 |
-
"step": 2275
|
647 |
-
},
|
648 |
-
{
|
649 |
-
"epoch": 0.23011505752876438,
|
650 |
-
"grad_norm": 178.43978881835938,
|
651 |
-
"learning_rate": 3.499743960549854e-05,
|
652 |
-
"loss": 0.1836,
|
653 |
-
"step": 2300
|
654 |
-
},
|
655 |
-
{
|
656 |
-
"epoch": 0.23261630815407705,
|
657 |
-
"grad_norm": 0.3828985393047333,
|
658 |
-
"learning_rate": 3.489300446126651e-05,
|
659 |
-
"loss": 0.2018,
|
660 |
-
"step": 2325
|
661 |
-
},
|
662 |
-
{
|
663 |
-
"epoch": 0.2351175587793897,
|
664 |
-
"grad_norm": 0.12711061537265778,
|
665 |
-
"learning_rate": 3.478764972701161e-05,
|
666 |
-
"loss": 0.0651,
|
667 |
-
"step": 2350
|
668 |
-
},
|
669 |
-
{
|
670 |
-
"epoch": 0.23761880940470234,
|
671 |
-
"grad_norm": 3.5419554710388184,
|
672 |
-
"learning_rate": 3.468138190801371e-05,
|
673 |
-
"loss": 0.0959,
|
674 |
-
"step": 2375
|
675 |
-
},
|
676 |
-
{
|
677 |
-
"epoch": 0.24012006003001501,
|
678 |
-
"grad_norm": 30.789512634277344,
|
679 |
-
"learning_rate": 3.457420756593245e-05,
|
680 |
-
"loss": 0.1165,
|
681 |
-
"step": 2400
|
682 |
-
},
|
683 |
-
{
|
684 |
-
"epoch": 0.24262131065532766,
|
685 |
-
"grad_norm": 0.04489291459321976,
|
686 |
-
"learning_rate": 3.4466133318402025e-05,
|
687 |
-
"loss": 0.0962,
|
688 |
-
"step": 2425
|
689 |
-
},
|
690 |
-
{
|
691 |
-
"epoch": 0.24512256128064033,
|
692 |
-
"grad_norm": 0.41131851077079773,
|
693 |
-
"learning_rate": 3.43571658386226e-05,
|
694 |
-
"loss": 0.0891,
|
695 |
-
"step": 2450
|
696 |
-
},
|
697 |
-
{
|
698 |
-
"epoch": 0.24762381190595298,
|
699 |
-
"grad_norm": 0.03262558951973915,
|
700 |
-
"learning_rate": 3.424731185494824e-05,
|
701 |
-
"loss": 0.0556,
|
702 |
-
"step": 2475
|
703 |
-
},
|
704 |
-
{
|
705 |
-
"epoch": 0.25012506253126565,
|
706 |
-
"grad_norm": 1.386066198348999,
|
707 |
-
"learning_rate": 3.4136578150471494e-05,
|
708 |
-
"loss": 0.0595,
|
709 |
-
"step": 2500
|
710 |
-
},
|
711 |
-
{
|
712 |
-
"epoch": 0.2526263131565783,
|
713 |
-
"grad_norm": 0.12720614671707153,
|
714 |
-
"learning_rate": 3.402497156260451e-05,
|
715 |
-
"loss": 0.2049,
|
716 |
-
"step": 2525
|
717 |
-
},
|
718 |
-
{
|
719 |
-
"epoch": 0.25512756378189094,
|
720 |
-
"grad_norm": 0.14693523943424225,
|
721 |
-
"learning_rate": 3.39124989826569e-05,
|
722 |
-
"loss": 0.0908,
|
723 |
-
"step": 2550
|
724 |
-
},
|
725 |
-
{
|
726 |
-
"epoch": 0.2576288144072036,
|
727 |
-
"grad_norm": 0.06785699725151062,
|
728 |
-
"learning_rate": 3.3799167355410196e-05,
|
729 |
-
"loss": 0.2336,
|
730 |
-
"step": 2575
|
731 |
-
},
|
732 |
-
{
|
733 |
-
"epoch": 0.26013006503251623,
|
734 |
-
"grad_norm": 0.7163326144218445,
|
735 |
-
"learning_rate": 3.3684983678689046e-05,
|
736 |
-
"loss": 0.0885,
|
737 |
-
"step": 2600
|
738 |
-
},
|
739 |
-
{
|
740 |
-
"epoch": 0.26263131565782893,
|
741 |
-
"grad_norm": 0.11218699812889099,
|
742 |
-
"learning_rate": 3.356995500292911e-05,
|
743 |
-
"loss": 0.0921,
|
744 |
-
"step": 2625
|
745 |
-
},
|
746 |
-
{
|
747 |
-
"epoch": 0.2651325662831416,
|
748 |
-
"grad_norm": 0.6830689907073975,
|
749 |
-
"learning_rate": 3.345408843074176e-05,
|
750 |
-
"loss": 0.1178,
|
751 |
-
"step": 2650
|
752 |
-
},
|
753 |
-
{
|
754 |
-
"epoch": 0.2676338169084542,
|
755 |
-
"grad_norm": 0.08778676390647888,
|
756 |
-
"learning_rate": 3.333739111647546e-05,
|
757 |
-
"loss": 0.0268,
|
758 |
-
"step": 2675
|
759 |
-
},
|
760 |
-
{
|
761 |
-
"epoch": 0.27013506753376687,
|
762 |
-
"grad_norm": 0.24001970887184143,
|
763 |
-
"learning_rate": 3.321987026577406e-05,
|
764 |
-
"loss": 0.1783,
|
765 |
-
"step": 2700
|
766 |
-
},
|
767 |
-
{
|
768 |
-
"epoch": 0.2726363181590795,
|
769 |
-
"grad_norm": 0.5253806710243225,
|
770 |
-
"learning_rate": 3.310153313513184e-05,
|
771 |
-
"loss": 0.103,
|
772 |
-
"step": 2725
|
773 |
-
},
|
774 |
-
{
|
775 |
-
"epoch": 0.2751375687843922,
|
776 |
-
"grad_norm": 0.7304710745811462,
|
777 |
-
"learning_rate": 3.2982387031445464e-05,
|
778 |
-
"loss": 0.0474,
|
779 |
-
"step": 2750
|
780 |
-
},
|
781 |
-
{
|
782 |
-
"epoch": 0.27763881940970486,
|
783 |
-
"grad_norm": 0.16174504160881042,
|
784 |
-
"learning_rate": 3.286243931156282e-05,
|
785 |
-
"loss": 0.1686,
|
786 |
-
"step": 2775
|
787 |
-
},
|
788 |
-
{
|
789 |
-
"epoch": 0.2801400700350175,
|
790 |
-
"grad_norm": 17.58709716796875,
|
791 |
-
"learning_rate": 3.274169738182872e-05,
|
792 |
-
"loss": 0.1018,
|
793 |
-
"step": 2800
|
794 |
-
},
|
795 |
-
{
|
796 |
-
"epoch": 0.28264132066033015,
|
797 |
-
"grad_norm": 0.1624075472354889,
|
798 |
-
"learning_rate": 3.2620168697627613e-05,
|
799 |
-
"loss": 0.1371,
|
800 |
-
"step": 2825
|
801 |
-
},
|
802 |
-
{
|
803 |
-
"epoch": 0.2851425712856428,
|
804 |
-
"grad_norm": 0.35798004269599915,
|
805 |
-
"learning_rate": 3.2497860762923257e-05,
|
806 |
-
"loss": 0.0883,
|
807 |
-
"step": 2850
|
808 |
-
},
|
809 |
-
{
|
810 |
-
"epoch": 0.2876438219109555,
|
811 |
-
"grad_norm": 0.02073662541806698,
|
812 |
-
"learning_rate": 3.237478112979533e-05,
|
813 |
-
"loss": 0.1174,
|
814 |
-
"step": 2875
|
815 |
-
},
|
816 |
-
{
|
817 |
-
"epoch": 0.29014507253626814,
|
818 |
-
"grad_norm": 0.7331066131591797,
|
819 |
-
"learning_rate": 3.225093739797315e-05,
|
820 |
-
"loss": 0.1546,
|
821 |
-
"step": 2900
|
822 |
-
},
|
823 |
-
{
|
824 |
-
"epoch": 0.2926463231615808,
|
825 |
-
"grad_norm": 0.5874232053756714,
|
826 |
-
"learning_rate": 3.212633721436641e-05,
|
827 |
-
"loss": 0.2263,
|
828 |
-
"step": 2925
|
829 |
-
},
|
830 |
-
{
|
831 |
-
"epoch": 0.29514757378689344,
|
832 |
-
"grad_norm": 0.18539220094680786,
|
833 |
-
"learning_rate": 3.200098827259301e-05,
|
834 |
-
"loss": 0.1026,
|
835 |
-
"step": 2950
|
836 |
-
},
|
837 |
-
{
|
838 |
-
"epoch": 0.2976488244122061,
|
839 |
-
"grad_norm": 0.2903716564178467,
|
840 |
-
"learning_rate": 3.187489831250401e-05,
|
841 |
-
"loss": 0.1403,
|
842 |
-
"step": 2975
|
843 |
-
},
|
844 |
-
{
|
845 |
-
"epoch": 0.3001500750375188,
|
846 |
-
"grad_norm": 29.44768524169922,
|
847 |
-
"learning_rate": 3.174807511970569e-05,
|
848 |
-
"loss": 0.1331,
|
849 |
-
"step": 3000
|
850 |
-
},
|
851 |
-
{
|
852 |
-
"epoch": 0.3026513256628314,
|
853 |
-
"grad_norm": 31.478647232055664,
|
854 |
-
"learning_rate": 3.162052652507885e-05,
|
855 |
-
"loss": 0.1367,
|
856 |
-
"step": 3025
|
857 |
-
},
|
858 |
-
{
|
859 |
-
"epoch": 0.3051525762881441,
|
860 |
-
"grad_norm": 21.006744384765625,
|
861 |
-
"learning_rate": 3.1492260404295304e-05,
|
862 |
-
"loss": 0.0742,
|
863 |
-
"step": 3050
|
864 |
-
},
|
865 |
-
{
|
866 |
-
"epoch": 0.3076538269134567,
|
867 |
-
"grad_norm": 28.84800148010254,
|
868 |
-
"learning_rate": 3.136328467733151e-05,
|
869 |
-
"loss": 0.0846,
|
870 |
-
"step": 3075
|
871 |
-
},
|
872 |
-
{
|
873 |
-
"epoch": 0.31015507753876936,
|
874 |
-
"grad_norm": 18.007699966430664,
|
875 |
-
"learning_rate": 3.123360730797959e-05,
|
876 |
-
"loss": 0.1715,
|
877 |
-
"step": 3100
|
878 |
-
},
|
879 |
-
{
|
880 |
-
"epoch": 0.31265632816408206,
|
881 |
-
"grad_norm": 0.2751634418964386,
|
882 |
-
"learning_rate": 3.1103236303355596e-05,
|
883 |
-
"loss": 0.0563,
|
884 |
-
"step": 3125
|
885 |
-
},
|
886 |
-
{
|
887 |
-
"epoch": 0.3151575787893947,
|
888 |
-
"grad_norm": 0.6980704069137573,
|
889 |
-
"learning_rate": 3.09721797134051e-05,
|
890 |
-
"loss": 0.1515,
|
891 |
-
"step": 3150
|
892 |
-
},
|
893 |
-
{
|
894 |
-
"epoch": 0.31765882941470736,
|
895 |
-
"grad_norm": 12.263067245483398,
|
896 |
-
"learning_rate": 3.08404456304061e-05,
|
897 |
-
"loss": 0.1442,
|
898 |
-
"step": 3175
|
899 |
-
},
|
900 |
-
{
|
901 |
-
"epoch": 0.32016008004002,
|
902 |
-
"grad_norm": 0.13746468722820282,
|
903 |
-
"learning_rate": 3.0708042188469404e-05,
|
904 |
-
"loss": 0.0945,
|
905 |
-
"step": 3200
|
906 |
-
},
|
907 |
-
{
|
908 |
-
"epoch": 0.32266133066533265,
|
909 |
-
"grad_norm": 0.7875755429267883,
|
910 |
-
"learning_rate": 3.057497756303634e-05,
|
911 |
-
"loss": 0.2051,
|
912 |
-
"step": 3225
|
913 |
-
},
|
914 |
-
{
|
915 |
-
"epoch": 0.32516258129064535,
|
916 |
-
"grad_norm": 9.643019676208496,
|
917 |
-
"learning_rate": 3.0441259970373962e-05,
|
918 |
-
"loss": 0.2014,
|
919 |
-
"step": 3250
|
920 |
-
},
|
921 |
-
{
|
922 |
-
"epoch": 0.327663831915958,
|
923 |
-
"grad_norm": 0.5073676109313965,
|
924 |
-
"learning_rate": 3.030689766706773e-05,
|
925 |
-
"loss": 0.1016,
|
926 |
-
"step": 3275
|
927 |
-
},
|
928 |
-
{
|
929 |
-
"epoch": 0.33016508254127064,
|
930 |
-
"grad_norm": 0.02357299253344536,
|
931 |
-
"learning_rate": 3.0171898949511706e-05,
|
932 |
-
"loss": 0.1297,
|
933 |
-
"step": 3300
|
934 |
-
},
|
935 |
-
{
|
936 |
-
"epoch": 0.3326663331665833,
|
937 |
-
"grad_norm": 0.11451732367277145,
|
938 |
-
"learning_rate": 3.003627215339623e-05,
|
939 |
-
"loss": 0.1771,
|
940 |
-
"step": 3325
|
941 |
-
},
|
942 |
-
{
|
943 |
-
"epoch": 0.33516758379189593,
|
944 |
-
"grad_norm": 0.16724112629890442,
|
945 |
-
"learning_rate": 2.99000256531933e-05,
|
946 |
-
"loss": 0.134,
|
947 |
-
"step": 3350
|
948 |
-
},
|
949 |
-
{
|
950 |
-
"epoch": 0.33766883441720863,
|
951 |
-
"grad_norm": 0.02885523997247219,
|
952 |
-
"learning_rate": 2.97631678616394e-05,
|
953 |
-
"loss": 0.0447,
|
954 |
-
"step": 3375
|
955 |
-
},
|
956 |
-
{
|
957 |
-
"epoch": 0.3401700850425213,
|
958 |
-
"grad_norm": 0.6085634827613831,
|
959 |
-
"learning_rate": 2.9625707229216113e-05,
|
960 |
-
"loss": 0.1869,
|
961 |
-
"step": 3400
|
962 |
-
},
|
963 |
-
{
|
964 |
-
"epoch": 0.3426713356678339,
|
965 |
-
"grad_norm": 27.649154663085938,
|
966 |
-
"learning_rate": 2.9487652243628276e-05,
|
967 |
-
"loss": 0.1278,
|
968 |
-
"step": 3425
|
969 |
-
},
|
970 |
-
{
|
971 |
-
"epoch": 0.34517258629314657,
|
972 |
-
"grad_norm": 6.867493152618408,
|
973 |
-
"learning_rate": 2.9349011429279926e-05,
|
974 |
-
"loss": 0.0836,
|
975 |
-
"step": 3450
|
976 |
-
},
|
977 |
-
{
|
978 |
-
"epoch": 0.3476738369184592,
|
979 |
-
"grad_norm": 0.23022982478141785,
|
980 |
-
"learning_rate": 2.9209793346747947e-05,
|
981 |
-
"loss": 0.1574,
|
982 |
-
"step": 3475
|
983 |
-
},
|
984 |
-
{
|
985 |
-
"epoch": 0.3501750875437719,
|
986 |
-
"grad_norm": 63.75465774536133,
|
987 |
-
"learning_rate": 2.9070006592253464e-05,
|
988 |
-
"loss": 0.2142,
|
989 |
-
"step": 3500
|
990 |
-
},
|
991 |
-
{
|
992 |
-
"epoch": 0.35267633816908456,
|
993 |
-
"grad_norm": 0.16846075654029846,
|
994 |
-
"learning_rate": 2.8929659797131085e-05,
|
995 |
-
"loss": 0.1644,
|
996 |
-
"step": 3525
|
997 |
-
},
|
998 |
-
{
|
999 |
-
"epoch": 0.3551775887943972,
|
1000 |
-
"grad_norm": 0.30203577876091003,
|
1001 |
-
"learning_rate": 2.8788761627295936e-05,
|
1002 |
-
"loss": 0.148,
|
1003 |
-
"step": 3550
|
1004 |
-
},
|
1005 |
-
{
|
1006 |
-
"epoch": 0.35767883941970985,
|
1007 |
-
"grad_norm": 0.5357218980789185,
|
1008 |
-
"learning_rate": 2.864732078270855e-05,
|
1009 |
-
"loss": 0.1211,
|
1010 |
-
"step": 3575
|
1011 |
-
},
|
1012 |
-
{
|
1013 |
-
"epoch": 0.3601800900450225,
|
1014 |
-
"grad_norm": 0.2230844348669052,
|
1015 |
-
"learning_rate": 2.850534599683772e-05,
|
1016 |
-
"loss": 0.1024,
|
1017 |
-
"step": 3600
|
1018 |
-
},
|
1019 |
-
{
|
1020 |
-
"epoch": 0.3626813406703352,
|
1021 |
-
"grad_norm": 0.8341374397277832,
|
1022 |
-
"learning_rate": 2.8362846036121186e-05,
|
1023 |
-
"loss": 0.0982,
|
1024 |
-
"step": 3625
|
1025 |
-
},
|
1026 |
-
{
|
1027 |
-
"epoch": 0.36518259129564784,
|
1028 |
-
"grad_norm": 0.02824452705681324,
|
1029 |
-
"learning_rate": 2.8219829699424398e-05,
|
1030 |
-
"loss": 0.1469,
|
1031 |
-
"step": 3650
|
1032 |
-
},
|
1033 |
-
{
|
1034 |
-
"epoch": 0.3676838419209605,
|
1035 |
-
"grad_norm": 64.07527160644531,
|
1036 |
-
"learning_rate": 2.8076305817497157e-05,
|
1037 |
-
"loss": 0.141,
|
1038 |
-
"step": 3675
|
1039 |
-
},
|
1040 |
-
{
|
1041 |
-
"epoch": 0.37018509254627313,
|
1042 |
-
"grad_norm": 0.3007892668247223,
|
1043 |
-
"learning_rate": 2.7932283252428392e-05,
|
1044 |
-
"loss": 0.1283,
|
1045 |
-
"step": 3700
|
1046 |
-
},
|
1047 |
-
{
|
1048 |
-
"epoch": 0.3726863431715858,
|
1049 |
-
"grad_norm": 0.22145915031433105,
|
1050 |
-
"learning_rate": 2.778777089709893e-05,
|
1051 |
-
"loss": 0.1817,
|
1052 |
-
"step": 3725
|
1053 |
-
},
|
1054 |
-
{
|
1055 |
-
"epoch": 0.3751875937968984,
|
1056 |
-
"grad_norm": 0.49903857707977295,
|
1057 |
-
"learning_rate": 2.7642777674632407e-05,
|
1058 |
-
"loss": 0.1851,
|
1059 |
-
"step": 3750
|
1060 |
-
},
|
1061 |
-
{
|
1062 |
-
"epoch": 0.3776888444222111,
|
1063 |
-
"grad_norm": 0.6932393908500671,
|
1064 |
-
"learning_rate": 2.749731253784429e-05,
|
1065 |
-
"loss": 0.1392,
|
1066 |
-
"step": 3775
|
1067 |
-
},
|
1068 |
-
{
|
1069 |
-
"epoch": 0.38019009504752377,
|
1070 |
-
"grad_norm": 0.11228801310062408,
|
1071 |
-
"learning_rate": 2.7351384468689096e-05,
|
1072 |
-
"loss": 0.1341,
|
1073 |
-
"step": 3800
|
1074 |
-
},
|
1075 |
-
{
|
1076 |
-
"epoch": 0.3826913456728364,
|
1077 |
-
"grad_norm": 0.15008388459682465,
|
1078 |
-
"learning_rate": 2.720500247770575e-05,
|
1079 |
-
"loss": 0.058,
|
1080 |
-
"step": 3825
|
1081 |
-
},
|
1082 |
-
{
|
1083 |
-
"epoch": 0.38519259629814906,
|
1084 |
-
"grad_norm": 0.23140987753868103,
|
1085 |
-
"learning_rate": 2.7058175603461226e-05,
|
1086 |
-
"loss": 0.038,
|
1087 |
-
"step": 3850
|
1088 |
-
},
|
1089 |
-
{
|
1090 |
-
"epoch": 0.3876938469234617,
|
1091 |
-
"grad_norm": 78.78473663330078,
|
1092 |
-
"learning_rate": 2.6910912911992476e-05,
|
1093 |
-
"loss": 0.163,
|
1094 |
-
"step": 3875
|
1095 |
-
},
|
1096 |
-
{
|
1097 |
-
"epoch": 0.3901950975487744,
|
1098 |
-
"grad_norm": 49.654754638671875,
|
1099 |
-
"learning_rate": 2.676322349624661e-05,
|
1100 |
-
"loss": 0.1941,
|
1101 |
-
"step": 3900
|
1102 |
-
},
|
1103 |
-
{
|
1104 |
-
"epoch": 0.39269634817408705,
|
1105 |
-
"grad_norm": 0.3017060160636902,
|
1106 |
-
"learning_rate": 2.661511647551943e-05,
|
1107 |
-
"loss": 0.0932,
|
1108 |
-
"step": 3925
|
1109 |
-
},
|
1110 |
-
{
|
1111 |
-
"epoch": 0.3951975987993997,
|
1112 |
-
"grad_norm": 0.08439047634601593,
|
1113 |
-
"learning_rate": 2.6466600994892372e-05,
|
1114 |
-
"loss": 0.1269,
|
1115 |
-
"step": 3950
|
1116 |
-
},
|
1117 |
-
{
|
1118 |
-
"epoch": 0.39769884942471234,
|
1119 |
-
"grad_norm": 0.02146688662469387,
|
1120 |
-
"learning_rate": 2.631768622466782e-05,
|
1121 |
-
"loss": 0.1793,
|
1122 |
-
"step": 3975
|
1123 |
-
},
|
1124 |
-
{
|
1125 |
-
"epoch": 0.400200100050025,
|
1126 |
-
"grad_norm": 46.505733489990234,
|
1127 |
-
"learning_rate": 2.6168381359802855e-05,
|
1128 |
-
"loss": 0.0817,
|
1129 |
-
"step": 4000
|
1130 |
-
},
|
1131 |
-
{
|
1132 |
-
"epoch": 0.4027013506753377,
|
1133 |
-
"grad_norm": 0.2078278660774231,
|
1134 |
-
"learning_rate": 2.6018695619341524e-05,
|
1135 |
-
"loss": 0.1722,
|
1136 |
-
"step": 4025
|
1137 |
-
},
|
1138 |
-
{
|
1139 |
-
"epoch": 0.40520260130065033,
|
1140 |
-
"grad_norm": 0.7848386764526367,
|
1141 |
-
"learning_rate": 2.5868638245845596e-05,
|
1142 |
-
"loss": 0.0823,
|
1143 |
-
"step": 4050
|
1144 |
-
},
|
1145 |
-
{
|
1146 |
-
"epoch": 0.407703851925963,
|
1147 |
-
"grad_norm": 0.1304464489221573,
|
1148 |
-
"learning_rate": 2.5718218504823833e-05,
|
1149 |
-
"loss": 0.122,
|
1150 |
-
"step": 4075
|
1151 |
-
},
|
1152 |
-
{
|
1153 |
-
"epoch": 0.4102051025512756,
|
1154 |
-
"grad_norm": 0.2570004165172577,
|
1155 |
-
"learning_rate": 2.5567445684159924e-05,
|
1156 |
-
"loss": 0.1914,
|
1157 |
-
"step": 4100
|
1158 |
-
},
|
1159 |
-
{
|
1160 |
-
"epoch": 0.41270635317658827,
|
1161 |
-
"grad_norm": 0.2900252938270569,
|
1162 |
-
"learning_rate": 2.541632909353896e-05,
|
1163 |
-
"loss": 0.2197,
|
1164 |
-
"step": 4125
|
1165 |
-
},
|
1166 |
-
{
|
1167 |
-
"epoch": 0.41520760380190097,
|
1168 |
-
"grad_norm": 1.286517858505249,
|
1169 |
-
"learning_rate": 2.5264878063872592e-05,
|
1170 |
-
"loss": 0.0303,
|
1171 |
-
"step": 4150
|
1172 |
-
},
|
1173 |
-
{
|
1174 |
-
"epoch": 0.4177088544272136,
|
1175 |
-
"grad_norm": 0.2735742926597595,
|
1176 |
-
"learning_rate": 2.5113101946722923e-05,
|
1177 |
-
"loss": 0.0711,
|
1178 |
-
"step": 4175
|
1179 |
-
},
|
1180 |
-
{
|
1181 |
-
"epoch": 0.42021010505252626,
|
1182 |
-
"grad_norm": 0.346799373626709,
|
1183 |
-
"learning_rate": 2.4961010113725016e-05,
|
1184 |
-
"loss": 0.0989,
|
1185 |
-
"step": 4200
|
1186 |
-
},
|
1187 |
-
{
|
1188 |
-
"epoch": 0.4227113556778389,
|
1189 |
-
"grad_norm": 0.1493910700082779,
|
1190 |
-
"learning_rate": 2.480861195600829e-05,
|
1191 |
-
"loss": 0.135,
|
1192 |
-
"step": 4225
|
1193 |
-
},
|
1194 |
-
{
|
1195 |
-
"epoch": 0.42521260630315155,
|
1196 |
-
"grad_norm": 0.1537860929965973,
|
1197 |
-
"learning_rate": 2.4655916883616615e-05,
|
1198 |
-
"loss": 0.1648,
|
1199 |
-
"step": 4250
|
1200 |
-
},
|
1201 |
-
{
|
1202 |
-
"epoch": 0.42771385692846425,
|
1203 |
-
"grad_norm": 0.05777391418814659,
|
1204 |
-
"learning_rate": 2.4502934324927275e-05,
|
1205 |
-
"loss": 0.0846,
|
1206 |
-
"step": 4275
|
1207 |
-
},
|
1208 |
-
{
|
1209 |
-
"epoch": 0.4302151075537769,
|
1210 |
-
"grad_norm": 16.710203170776367,
|
1211 |
-
"learning_rate": 2.434967372606884e-05,
|
1212 |
-
"loss": 0.1273,
|
1213 |
-
"step": 4300
|
1214 |
-
},
|
1215 |
-
{
|
1216 |
-
"epoch": 0.43271635817908954,
|
1217 |
-
"grad_norm": 8.845438957214355,
|
1218 |
-
"learning_rate": 2.4196144550337835e-05,
|
1219 |
-
"loss": 0.1205,
|
1220 |
-
"step": 4325
|
1221 |
-
},
|
1222 |
-
{
|
1223 |
-
"epoch": 0.4352176088044022,
|
1224 |
-
"grad_norm": 17.3373966217041,
|
1225 |
-
"learning_rate": 2.4042356277614474e-05,
|
1226 |
-
"loss": 0.1297,
|
1227 |
-
"step": 4350
|
1228 |
-
},
|
1229 |
-
{
|
1230 |
-
"epoch": 0.43771885942971483,
|
1231 |
-
"grad_norm": 0.06476125866174698,
|
1232 |
-
"learning_rate": 2.3888318403777278e-05,
|
1233 |
-
"loss": 0.1327,
|
1234 |
-
"step": 4375
|
1235 |
-
},
|
1236 |
-
{
|
1237 |
-
"epoch": 0.44022011005502754,
|
1238 |
-
"grad_norm": 0.26493966579437256,
|
1239 |
-
"learning_rate": 2.3734040440116754e-05,
|
1240 |
-
"loss": 0.103,
|
1241 |
-
"step": 4400
|
1242 |
-
},
|
1243 |
-
{
|
1244 |
-
"epoch": 0.4427213606803402,
|
1245 |
-
"grad_norm": 0.35106614232063293,
|
1246 |
-
"learning_rate": 2.3579531912748103e-05,
|
1247 |
-
"loss": 0.1393,
|
1248 |
-
"step": 4425
|
1249 |
-
},
|
1250 |
-
{
|
1251 |
-
"epoch": 0.4452226113056528,
|
1252 |
-
"grad_norm": 0.263611376285553,
|
1253 |
-
"learning_rate": 2.342480236202301e-05,
|
1254 |
-
"loss": 0.1552,
|
1255 |
-
"step": 4450
|
1256 |
-
},
|
1257 |
-
{
|
1258 |
-
"epoch": 0.44772386193096547,
|
1259 |
-
"grad_norm": 11.385093688964844,
|
1260 |
-
"learning_rate": 2.326986134194056e-05,
|
1261 |
-
"loss": 0.1419,
|
1262 |
-
"step": 4475
|
1263 |
-
},
|
1264 |
-
{
|
1265 |
-
"epoch": 0.4502251125562781,
|
1266 |
-
"grad_norm": 0.8247035145759583,
|
1267 |
-
"learning_rate": 2.311471841955733e-05,
|
1268 |
-
"loss": 0.094,
|
1269 |
-
"step": 4500
|
1270 |
-
},
|
1271 |
-
{
|
1272 |
-
"epoch": 0.4527263631815908,
|
1273 |
-
"grad_norm": 10.922027587890625,
|
1274 |
-
"learning_rate": 2.2959383174396643e-05,
|
1275 |
-
"loss": 0.1064,
|
1276 |
-
"step": 4525
|
1277 |
-
},
|
1278 |
-
{
|
1279 |
-
"epoch": 0.45522761380690346,
|
1280 |
-
"grad_norm": 0.3066991865634918,
|
1281 |
-
"learning_rate": 2.280386519785706e-05,
|
1282 |
-
"loss": 0.1165,
|
1283 |
-
"step": 4550
|
1284 |
-
},
|
1285 |
-
{
|
1286 |
-
"epoch": 0.4577288644322161,
|
1287 |
-
"grad_norm": 0.32978731393814087,
|
1288 |
-
"learning_rate": 2.2648174092620163e-05,
|
1289 |
-
"loss": 0.1734,
|
1290 |
-
"step": 4575
|
1291 |
-
},
|
1292 |
-
{
|
1293 |
-
"epoch": 0.46023011505752875,
|
1294 |
-
"grad_norm": 80.88328552246094,
|
1295 |
-
"learning_rate": 2.2492319472057614e-05,
|
1296 |
-
"loss": 0.116,
|
1297 |
-
"step": 4600
|
1298 |
-
},
|
1299 |
-
{
|
1300 |
-
"epoch": 0.4627313656828414,
|
1301 |
-
"grad_norm": 17.25701332092285,
|
1302 |
-
"learning_rate": 2.2336310959637567e-05,
|
1303 |
-
"loss": 0.1536,
|
1304 |
-
"step": 4625
|
1305 |
-
},
|
1306 |
-
{
|
1307 |
-
"epoch": 0.4652326163081541,
|
1308 |
-
"grad_norm": 0.9789904356002808,
|
1309 |
-
"learning_rate": 2.218015818833043e-05,
|
1310 |
-
"loss": 0.1021,
|
1311 |
-
"step": 4650
|
1312 |
-
},
|
1313 |
-
{
|
1314 |
-
"epoch": 0.46773386693346675,
|
1315 |
-
"grad_norm": 0.6882738471031189,
|
1316 |
-
"learning_rate": 2.2023870800014117e-05,
|
1317 |
-
"loss": 0.0921,
|
1318 |
-
"step": 4675
|
1319 |
-
},
|
1320 |
-
{
|
1321 |
-
"epoch": 0.4702351175587794,
|
1322 |
-
"grad_norm": 0.0516337975859642,
|
1323 |
-
"learning_rate": 2.186745844487865e-05,
|
1324 |
-
"loss": 0.0856,
|
1325 |
-
"step": 4700
|
1326 |
-
},
|
1327 |
-
{
|
1328 |
-
"epoch": 0.47273636818409204,
|
1329 |
-
"grad_norm": 0.1375594288110733,
|
1330 |
-
"learning_rate": 2.1710930780830286e-05,
|
1331 |
-
"loss": 0.0549,
|
1332 |
-
"step": 4725
|
1333 |
-
},
|
1334 |
-
{
|
1335 |
-
"epoch": 0.4752376188094047,
|
1336 |
-
"grad_norm": 0.4323989152908325,
|
1337 |
-
"learning_rate": 2.1554297472895228e-05,
|
1338 |
-
"loss": 0.1216,
|
1339 |
-
"step": 4750
|
1340 |
-
},
|
1341 |
-
{
|
1342 |
-
"epoch": 0.4777388694347174,
|
1343 |
-
"grad_norm": 0.03123696707189083,
|
1344 |
-
"learning_rate": 2.139756819262279e-05,
|
1345 |
-
"loss": 0.1617,
|
1346 |
-
"step": 4775
|
1347 |
-
},
|
1348 |
-
{
|
1349 |
-
"epoch": 0.48024012006003003,
|
1350 |
-
"grad_norm": 0.1050788164138794,
|
1351 |
-
"learning_rate": 2.124075261748826e-05,
|
1352 |
-
"loss": 0.1416,
|
1353 |
-
"step": 4800
|
1354 |
-
},
|
1355 |
-
{
|
1356 |
-
"epoch": 0.4827413706853427,
|
1357 |
-
"grad_norm": 0.18789520859718323,
|
1358 |
-
"learning_rate": 2.10838604302953e-05,
|
1359 |
-
"loss": 0.0703,
|
1360 |
-
"step": 4825
|
1361 |
-
},
|
1362 |
-
{
|
1363 |
-
"epoch": 0.4852426213106553,
|
1364 |
-
"grad_norm": 0.8157877922058105,
|
1365 |
-
"learning_rate": 2.092690131857811e-05,
|
1366 |
-
"loss": 0.0814,
|
1367 |
-
"step": 4850
|
1368 |
-
},
|
1369 |
-
{
|
1370 |
-
"epoch": 0.48774387193596797,
|
1371 |
-
"grad_norm": 0.13385553658008575,
|
1372 |
-
"learning_rate": 2.0769884974003242e-05,
|
1373 |
-
"loss": 0.1011,
|
1374 |
-
"step": 4875
|
1375 |
-
},
|
1376 |
-
{
|
1377 |
-
"epoch": 0.49024512256128067,
|
1378 |
-
"grad_norm": 16.390085220336914,
|
1379 |
-
"learning_rate": 2.0612821091771166e-05,
|
1380 |
-
"loss": 0.1699,
|
1381 |
-
"step": 4900
|
1382 |
-
},
|
1383 |
-
{
|
1384 |
-
"epoch": 0.4927463731865933,
|
1385 |
-
"grad_norm": 0.03446008637547493,
|
1386 |
-
"learning_rate": 2.0455719370017642e-05,
|
1387 |
-
"loss": 0.0448,
|
1388 |
-
"step": 4925
|
1389 |
-
},
|
1390 |
-
{
|
1391 |
-
"epoch": 0.49524762381190596,
|
1392 |
-
"grad_norm": 16.086416244506836,
|
1393 |
-
"learning_rate": 2.0298589509214875e-05,
|
1394 |
-
"loss": 0.1056,
|
1395 |
-
"step": 4950
|
1396 |
-
},
|
1397 |
-
{
|
1398 |
-
"epoch": 0.4977488744372186,
|
1399 |
-
"grad_norm": 0.005488800350576639,
|
1400 |
-
"learning_rate": 2.014144121157256e-05,
|
1401 |
-
"loss": 0.0696,
|
1402 |
-
"step": 4975
|
1403 |
-
},
|
1404 |
-
{
|
1405 |
-
"epoch": 0.5002501250625313,
|
1406 |
-
"grad_norm": 0.040323201566934586,
|
1407 |
-
"learning_rate": 1.9984284180438794e-05,
|
1408 |
-
"loss": 0.0466,
|
1409 |
-
"step": 5000
|
1410 |
-
},
|
1411 |
-
{
|
1412 |
-
"epoch": 0.5027513756878439,
|
1413 |
-
"grad_norm": 0.686833918094635,
|
1414 |
-
"learning_rate": 1.982712811970097e-05,
|
1415 |
-
"loss": 0.1944,
|
1416 |
-
"step": 5025
|
1417 |
-
},
|
1418 |
-
{
|
1419 |
-
"epoch": 0.5052526263131566,
|
1420 |
-
"grad_norm": 0.02543717436492443,
|
1421 |
-
"learning_rate": 1.9669982733186515e-05,
|
1422 |
-
"loss": 0.1211,
|
1423 |
-
"step": 5050
|
1424 |
-
},
|
1425 |
-
{
|
1426 |
-
"epoch": 0.5077538769384692,
|
1427 |
-
"grad_norm": 0.17090587317943573,
|
1428 |
-
"learning_rate": 1.9512857724063797e-05,
|
1429 |
-
"loss": 0.0652,
|
1430 |
-
"step": 5075
|
1431 |
-
},
|
1432 |
-
{
|
1433 |
-
"epoch": 0.5102551275637819,
|
1434 |
-
"grad_norm": 0.023501081392169,
|
1435 |
-
"learning_rate": 1.935576279424292e-05,
|
1436 |
-
"loss": 0.0691,
|
1437 |
-
"step": 5100
|
1438 |
-
},
|
1439 |
-
{
|
1440 |
-
"epoch": 0.5127563781890946,
|
1441 |
-
"grad_norm": 0.021651741117239,
|
1442 |
-
"learning_rate": 1.919870764377673e-05,
|
1443 |
-
"loss": 0.1466,
|
1444 |
-
"step": 5125
|
1445 |
-
},
|
1446 |
-
{
|
1447 |
-
"epoch": 0.5152576288144072,
|
1448 |
-
"grad_norm": 9.450238227844238,
|
1449 |
-
"learning_rate": 1.9041701970261835e-05,
|
1450 |
-
"loss": 0.1309,
|
1451 |
-
"step": 5150
|
1452 |
-
},
|
1453 |
-
{
|
1454 |
-
"epoch": 0.5177588794397199,
|
1455 |
-
"grad_norm": 1.0457688570022583,
|
1456 |
-
"learning_rate": 1.8884755468239786e-05,
|
1457 |
-
"loss": 0.059,
|
1458 |
-
"step": 5175
|
1459 |
-
},
|
1460 |
-
{
|
1461 |
-
"epoch": 0.5202601300650325,
|
1462 |
-
"grad_norm": 0.4531708061695099,
|
1463 |
-
"learning_rate": 1.8727877828598547e-05,
|
1464 |
-
"loss": 0.2553,
|
1465 |
-
"step": 5200
|
1466 |
-
},
|
1467 |
-
{
|
1468 |
-
"epoch": 0.5227613806903452,
|
1469 |
-
"grad_norm": 0.2637249529361725,
|
1470 |
-
"learning_rate": 1.8571078737974036e-05,
|
1471 |
-
"loss": 0.1373,
|
1472 |
-
"step": 5225
|
1473 |
-
},
|
1474 |
-
{
|
1475 |
-
"epoch": 0.5252626313156579,
|
1476 |
-
"grad_norm": 0.16574428975582123,
|
1477 |
-
"learning_rate": 1.8414367878152078e-05,
|
1478 |
-
"loss": 0.087,
|
1479 |
-
"step": 5250
|
1480 |
-
},
|
1481 |
-
{
|
1482 |
-
"epoch": 0.5277638819409705,
|
1483 |
-
"grad_norm": 21.844409942626953,
|
1484 |
-
"learning_rate": 1.825775492547056e-05,
|
1485 |
-
"loss": 0.0837,
|
1486 |
-
"step": 5275
|
1487 |
-
},
|
1488 |
-
{
|
1489 |
-
"epoch": 0.5302651325662832,
|
1490 |
-
"grad_norm": 0.2041023224592209,
|
1491 |
-
"learning_rate": 1.8101249550221926e-05,
|
1492 |
-
"loss": 0.0989,
|
1493 |
-
"step": 5300
|
1494 |
-
},
|
1495 |
-
{
|
1496 |
-
"epoch": 0.5327663831915957,
|
1497 |
-
"grad_norm": 0.01832737773656845,
|
1498 |
-
"learning_rate": 1.7944861416056117e-05,
|
1499 |
-
"loss": 0.0597,
|
1500 |
-
"step": 5325
|
1501 |
-
},
|
1502 |
-
{
|
1503 |
-
"epoch": 0.5352676338169085,
|
1504 |
-
"grad_norm": 48.69431686401367,
|
1505 |
-
"learning_rate": 1.7788600179383853e-05,
|
1506 |
-
"loss": 0.1332,
|
1507 |
-
"step": 5350
|
1508 |
-
},
|
1509 |
-
{
|
1510 |
-
"epoch": 0.5377688844422212,
|
1511 |
-
"grad_norm": 21.431251525878906,
|
1512 |
-
"learning_rate": 1.763247548878037e-05,
|
1513 |
-
"loss": 0.1325,
|
1514 |
-
"step": 5375
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 0.5402701350675337,
|
1518 |
-
"grad_norm": 0.20052960515022278,
|
1519 |
-
"learning_rate": 1.74764969843897e-05,
|
1520 |
-
"loss": 0.0852,
|
1521 |
-
"step": 5400
|
1522 |
-
},
|
1523 |
-
{
|
1524 |
-
"epoch": 0.5427713856928464,
|
1525 |
-
"grad_norm": 0.09768962860107422,
|
1526 |
-
"learning_rate": 1.732067429732936e-05,
|
1527 |
-
"loss": 0.1273,
|
1528 |
-
"step": 5425
|
1529 |
-
},
|
1530 |
-
{
|
1531 |
-
"epoch": 0.545272636318159,
|
1532 |
-
"grad_norm": 5.405893325805664,
|
1533 |
-
"learning_rate": 1.716501704909574e-05,
|
1534 |
-
"loss": 0.1345,
|
1535 |
-
"step": 5450
|
1536 |
-
},
|
1537 |
-
{
|
1538 |
-
"epoch": 0.5477738869434717,
|
1539 |
-
"grad_norm": 0.26859474182128906,
|
1540 |
-
"learning_rate": 1.700953485096993e-05,
|
1541 |
-
"loss": 0.1443,
|
1542 |
-
"step": 5475
|
1543 |
-
},
|
1544 |
-
{
|
1545 |
-
"epoch": 0.5502751375687844,
|
1546 |
-
"grad_norm": 12.15195369720459,
|
1547 |
-
"learning_rate": 1.6854237303424332e-05,
|
1548 |
-
"loss": 0.1625,
|
1549 |
-
"step": 5500
|
1550 |
-
},
|
1551 |
-
{
|
1552 |
-
"epoch": 0.552776388194097,
|
1553 |
-
"grad_norm": 0.7385017275810242,
|
1554 |
-
"learning_rate": 1.669913399552982e-05,
|
1555 |
-
"loss": 0.0766,
|
1556 |
-
"step": 5525
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 0.5552776388194097,
|
1560 |
-
"grad_norm": 13.504461288452148,
|
1561 |
-
"learning_rate": 1.654423450436365e-05,
|
1562 |
-
"loss": 0.1379,
|
1563 |
-
"step": 5550
|
1564 |
-
},
|
1565 |
-
{
|
1566 |
-
"epoch": 0.5577788894447223,
|
1567 |
-
"grad_norm": 9.465699195861816,
|
1568 |
-
"learning_rate": 1.6389548394418152e-05,
|
1569 |
-
"loss": 0.1128,
|
1570 |
-
"step": 5575
|
1571 |
-
},
|
1572 |
-
{
|
1573 |
-
"epoch": 0.560280140070035,
|
1574 |
-
"grad_norm": 22.666833877563477,
|
1575 |
-
"learning_rate": 1.6235085217010077e-05,
|
1576 |
-
"loss": 0.1383,
|
1577 |
-
"step": 5600
|
1578 |
-
},
|
1579 |
-
{
|
1580 |
-
"epoch": 0.5627813906953477,
|
1581 |
-
"grad_norm": 0.034865688532590866,
|
1582 |
-
"learning_rate": 1.6080854509690927e-05,
|
1583 |
-
"loss": 0.1037,
|
1584 |
-
"step": 5625
|
1585 |
-
},
|
1586 |
-
{
|
1587 |
-
"epoch": 0.5652826413206603,
|
1588 |
-
"grad_norm": 0.27604594826698303,
|
1589 |
-
"learning_rate": 1.5926865795657996e-05,
|
1590 |
-
"loss": 0.0811,
|
1591 |
-
"step": 5650
|
1592 |
-
},
|
1593 |
-
{
|
1594 |
-
"epoch": 0.567783891945973,
|
1595 |
-
"grad_norm": 22.162174224853516,
|
1596 |
-
"learning_rate": 1.5773128583166327e-05,
|
1597 |
-
"loss": 0.1242,
|
1598 |
-
"step": 5675
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 0.5702851425712856,
|
1602 |
-
"grad_norm": 0.08329381048679352,
|
1603 |
-
"learning_rate": 1.5619652364941676e-05,
|
1604 |
-
"loss": 0.1438,
|
1605 |
-
"step": 5700
|
1606 |
-
},
|
1607 |
-
{
|
1608 |
-
"epoch": 0.5727863931965983,
|
1609 |
-
"grad_norm": 19.047225952148438,
|
1610 |
-
"learning_rate": 1.54664466175943e-05,
|
1611 |
-
"loss": 0.1021,
|
1612 |
-
"step": 5725
|
1613 |
-
},
|
1614 |
-
{
|
1615 |
-
"epoch": 0.575287643821911,
|
1616 |
-
"grad_norm": 9.669032096862793,
|
1617 |
-
"learning_rate": 1.5313520801033857e-05,
|
1618 |
-
"loss": 0.118,
|
1619 |
-
"step": 5750
|
1620 |
-
},
|
1621 |
-
{
|
1622 |
-
"epoch": 0.5777888944472236,
|
1623 |
-
"grad_norm": 0.7107395529747009,
|
1624 |
-
"learning_rate": 1.5160884357885283e-05,
|
1625 |
-
"loss": 0.1439,
|
1626 |
-
"step": 5775
|
1627 |
-
},
|
1628 |
-
{
|
1629 |
-
"epoch": 0.5802901450725363,
|
1630 |
-
"grad_norm": 0.21181002259254456,
|
1631 |
-
"learning_rate": 1.5008546712905712e-05,
|
1632 |
-
"loss": 0.1174,
|
1633 |
-
"step": 5800
|
1634 |
-
},
|
1635 |
-
{
|
1636 |
-
"epoch": 0.5827913956978489,
|
1637 |
-
"grad_norm": 13.23889446258545,
|
1638 |
-
"learning_rate": 1.4856517272402581e-05,
|
1639 |
-
"loss": 0.1962,
|
1640 |
-
"step": 5825
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 0.5852926463231616,
|
1644 |
-
"grad_norm": 0.07769283652305603,
|
1645 |
-
"learning_rate": 1.4704805423652772e-05,
|
1646 |
-
"loss": 0.034,
|
1647 |
-
"step": 5850
|
1648 |
-
},
|
1649 |
-
{
|
1650 |
-
"epoch": 0.5877938969484743,
|
1651 |
-
"grad_norm": 17.46116065979004,
|
1652 |
-
"learning_rate": 1.4553420534323028e-05,
|
1653 |
-
"loss": 0.0919,
|
1654 |
-
"step": 5875
|
1655 |
-
},
|
1656 |
-
{
|
1657 |
-
"epoch": 0.5902951475737869,
|
1658 |
-
"grad_norm": 0.21442942321300507,
|
1659 |
-
"learning_rate": 1.4402371951891501e-05,
|
1660 |
-
"loss": 0.1045,
|
1661 |
-
"step": 5900
|
1662 |
-
},
|
1663 |
-
{
|
1664 |
-
"epoch": 0.5927963981990996,
|
1665 |
-
"grad_norm": 0.2784653902053833,
|
1666 |
-
"learning_rate": 1.4251669003070589e-05,
|
1667 |
-
"loss": 0.05,
|
1668 |
-
"step": 5925
|
1669 |
-
},
|
1670 |
-
{
|
1671 |
-
"epoch": 0.5952976488244122,
|
1672 |
-
"grad_norm": 1.5409433841705322,
|
1673 |
-
"learning_rate": 1.410132099323105e-05,
|
1674 |
-
"loss": 0.1592,
|
1675 |
-
"step": 5950
|
1676 |
-
},
|
1677 |
-
{
|
1678 |
-
"epoch": 0.5977988994497249,
|
1679 |
-
"grad_norm": 8.39173412322998,
|
1680 |
-
"learning_rate": 1.3951337205827411e-05,
|
1681 |
-
"loss": 0.1158,
|
1682 |
-
"step": 5975
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 0.6003001500750376,
|
1686 |
-
"grad_norm": 0.013280701823532581,
|
1687 |
-
"learning_rate": 1.380172690182477e-05,
|
1688 |
-
"loss": 0.1139,
|
1689 |
-
"step": 6000
|
1690 |
-
},
|
1691 |
-
{
|
1692 |
-
"epoch": 0.6028014007003502,
|
1693 |
-
"grad_norm": 0.07937493920326233,
|
1694 |
-
"learning_rate": 1.3652499319126963e-05,
|
1695 |
-
"loss": 0.0797,
|
1696 |
-
"step": 6025
|
1697 |
-
},
|
1698 |
-
{
|
1699 |
-
"epoch": 0.6053026513256629,
|
1700 |
-
"grad_norm": 0.08822044730186462,
|
1701 |
-
"learning_rate": 1.3503663672006124e-05,
|
1702 |
-
"loss": 0.0992,
|
1703 |
-
"step": 6050
|
1704 |
-
},
|
1705 |
-
{
|
1706 |
-
"epoch": 0.6078039019509754,
|
1707 |
-
"grad_norm": 47.08625793457031,
|
1708 |
-
"learning_rate": 1.3355229150533782e-05,
|
1709 |
-
"loss": 0.1176,
|
1710 |
-
"step": 6075
|
1711 |
-
},
|
1712 |
-
{
|
1713 |
-
"epoch": 0.6103051525762881,
|
1714 |
-
"grad_norm": 12.887028694152832,
|
1715 |
-
"learning_rate": 1.3207204920013367e-05,
|
1716 |
-
"loss": 0.0865,
|
1717 |
-
"step": 6100
|
1718 |
-
},
|
1719 |
-
{
|
1720 |
-
"epoch": 0.6128064032016008,
|
1721 |
-
"grad_norm": 68.05748748779297,
|
1722 |
-
"learning_rate": 1.3059600120414315e-05,
|
1723 |
-
"loss": 0.0791,
|
1724 |
-
"step": 6125
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 0.6153076538269134,
|
1728 |
-
"grad_norm": 0.025451015681028366,
|
1729 |
-
"learning_rate": 1.2912423865807697e-05,
|
1730 |
-
"loss": 0.1074,
|
1731 |
-
"step": 6150
|
1732 |
-
},
|
1733 |
-
{
|
1734 |
-
"epoch": 0.6178089044522261,
|
1735 |
-
"grad_norm": 0.4324849843978882,
|
1736 |
-
"learning_rate": 1.276568524380344e-05,
|
1737 |
-
"loss": 0.1429,
|
1738 |
-
"step": 6175
|
1739 |
-
},
|
1740 |
-
{
|
1741 |
-
"epoch": 0.6203101550775387,
|
1742 |
-
"grad_norm": 0.29630744457244873,
|
1743 |
-
"learning_rate": 1.261939331498924e-05,
|
1744 |
-
"loss": 0.069,
|
1745 |
-
"step": 6200
|
1746 |
-
},
|
1747 |
-
{
|
1748 |
-
"epoch": 0.6228114057028514,
|
1749 |
-
"grad_norm": 44.56884765625,
|
1750 |
-
"learning_rate": 1.2473557112371044e-05,
|
1751 |
-
"loss": 0.1565,
|
1752 |
-
"step": 6225
|
1753 |
-
},
|
1754 |
-
{
|
1755 |
-
"epoch": 0.6253126563281641,
|
1756 |
-
"grad_norm": 0.05059849098324776,
|
1757 |
-
"learning_rate": 1.2328185640815362e-05,
|
1758 |
-
"loss": 0.0462,
|
1759 |
-
"step": 6250
|
1760 |
-
},
|
1761 |
-
{
|
1762 |
-
"epoch": 0.6278139069534767,
|
1763 |
-
"grad_norm": 0.07012537866830826,
|
1764 |
-
"learning_rate": 1.2183287876493206e-05,
|
1765 |
-
"loss": 0.1067,
|
1766 |
-
"step": 6275
|
1767 |
-
},
|
1768 |
-
{
|
1769 |
-
"epoch": 0.6303151575787894,
|
1770 |
-
"grad_norm": 59.95124816894531,
|
1771 |
-
"learning_rate": 1.2038872766325835e-05,
|
1772 |
-
"loss": 0.074,
|
1773 |
-
"step": 6300
|
1774 |
-
},
|
1775 |
-
{
|
1776 |
-
"epoch": 0.632816408204102,
|
1777 |
-
"grad_norm": 0.36369818449020386,
|
1778 |
-
"learning_rate": 1.1894949227432354e-05,
|
1779 |
-
"loss": 0.1086,
|
1780 |
-
"step": 6325
|
1781 |
-
},
|
1782 |
-
{
|
1783 |
-
"epoch": 0.6353176588294147,
|
1784 |
-
"grad_norm": 0.05616721138358116,
|
1785 |
-
"learning_rate": 1.1751526146579074e-05,
|
1786 |
-
"loss": 0.156,
|
1787 |
-
"step": 6350
|
1788 |
-
},
|
1789 |
-
{
|
1790 |
-
"epoch": 0.6378189094547274,
|
1791 |
-
"grad_norm": 0.10930348187685013,
|
1792 |
-
"learning_rate": 1.1608612379630808e-05,
|
1793 |
-
"loss": 0.1664,
|
1794 |
-
"step": 6375
|
1795 |
-
},
|
1796 |
-
{
|
1797 |
-
"epoch": 0.64032016008004,
|
1798 |
-
"grad_norm": 0.7583307027816772,
|
1799 |
-
"learning_rate": 1.1466216751004064e-05,
|
1800 |
-
"loss": 0.146,
|
1801 |
-
"step": 6400
|
1802 |
-
},
|
1803 |
-
{
|
1804 |
-
"epoch": 0.6428214107053527,
|
1805 |
-
"grad_norm": 0.03929620981216431,
|
1806 |
-
"learning_rate": 1.1324348053122123e-05,
|
1807 |
-
"loss": 0.1107,
|
1808 |
-
"step": 6425
|
1809 |
-
},
|
1810 |
-
{
|
1811 |
-
"epoch": 0.6453226613306653,
|
1812 |
-
"grad_norm": 5.315922260284424,
|
1813 |
-
"learning_rate": 1.1183015045872198e-05,
|
1814 |
-
"loss": 0.0343,
|
1815 |
-
"step": 6450
|
1816 |
-
},
|
1817 |
-
{
|
1818 |
-
"epoch": 0.647823911955978,
|
1819 |
-
"grad_norm": 0.22626645863056183,
|
1820 |
-
"learning_rate": 1.1042226456064484e-05,
|
1821 |
-
"loss": 0.1822,
|
1822 |
-
"step": 6475
|
1823 |
-
},
|
1824 |
-
{
|
1825 |
-
"epoch": 0.6503251625812907,
|
1826 |
-
"grad_norm": 0.0635998472571373,
|
1827 |
-
"learning_rate": 1.0901990976893357e-05,
|
1828 |
-
"loss": 0.125,
|
1829 |
-
"step": 6500
|
1830 |
-
},
|
1831 |
-
{
|
1832 |
-
"epoch": 0.6528264132066033,
|
1833 |
-
"grad_norm": 0.09473757445812225,
|
1834 |
-
"learning_rate": 1.0762317267400582e-05,
|
1835 |
-
"loss": 0.0852,
|
1836 |
-
"step": 6525
|
1837 |
-
},
|
1838 |
-
{
|
1839 |
-
"epoch": 0.655327663831916,
|
1840 |
-
"grad_norm": 0.05637608841061592,
|
1841 |
-
"learning_rate": 1.0623213951940618e-05,
|
1842 |
-
"loss": 0.0761,
|
1843 |
-
"step": 6550
|
1844 |
-
},
|
1845 |
-
{
|
1846 |
-
"epoch": 0.6578289144572286,
|
1847 |
-
"grad_norm": 23.069490432739258,
|
1848 |
-
"learning_rate": 1.048468961964816e-05,
|
1849 |
-
"loss": 0.1205,
|
1850 |
-
"step": 6575
|
1851 |
-
},
|
1852 |
-
{
|
1853 |
-
"epoch": 0.6603301650825413,
|
1854 |
-
"grad_norm": 0.01633673533797264,
|
1855 |
-
"learning_rate": 1.0346752823907722e-05,
|
1856 |
-
"loss": 0.0394,
|
1857 |
-
"step": 6600
|
1858 |
-
},
|
1859 |
-
{
|
1860 |
-
"epoch": 0.662831415707854,
|
1861 |
-
"grad_norm": 35.29267120361328,
|
1862 |
-
"learning_rate": 1.0209412081825536e-05,
|
1863 |
-
"loss": 0.1362,
|
1864 |
-
"step": 6625
|
1865 |
-
},
|
1866 |
-
{
|
1867 |
-
"epoch": 0.6653326663331666,
|
1868 |
-
"grad_norm": 30.002025604248047,
|
1869 |
-
"learning_rate": 1.0072675873703654e-05,
|
1870 |
-
"loss": 0.1524,
|
1871 |
-
"step": 6650
|
1872 |
-
},
|
1873 |
-
{
|
1874 |
-
"epoch": 0.6678339169584793,
|
1875 |
-
"grad_norm": 25.67470932006836,
|
1876 |
-
"learning_rate": 9.936552642516291e-06,
|
1877 |
-
"loss": 0.1342,
|
1878 |
-
"step": 6675
|
1879 |
-
},
|
1880 |
-
{
|
1881 |
-
"epoch": 0.6703351675837919,
|
1882 |
-
"grad_norm": 9.367170333862305,
|
1883 |
-
"learning_rate": 9.801050793388533e-06,
|
1884 |
-
"loss": 0.1445,
|
1885 |
-
"step": 6700
|
1886 |
-
},
|
1887 |
-
{
|
1888 |
-
"epoch": 0.6728364182091046,
|
1889 |
-
"grad_norm": 25.29102897644043,
|
1890 |
-
"learning_rate": 9.666178693077313e-06,
|
1891 |
-
"loss": 0.1089,
|
1892 |
-
"step": 6725
|
1893 |
-
},
|
1894 |
-
{
|
1895 |
-
"epoch": 0.6753376688344173,
|
1896 |
-
"grad_norm": 8.388240814208984,
|
1897 |
-
"learning_rate": 9.531944669454841e-06,
|
1898 |
-
"loss": 0.1439,
|
1899 |
-
"step": 6750
|
1900 |
-
},
|
1901 |
-
{
|
1902 |
-
"epoch": 0.6778389194597298,
|
1903 |
-
"grad_norm": 0.06321081519126892,
|
1904 |
-
"learning_rate": 9.398357010994356e-06,
|
1905 |
-
"loss": 0.0756,
|
1906 |
-
"step": 6775
|
1907 |
-
},
|
1908 |
-
{
|
1909 |
-
"epoch": 0.6803401700850426,
|
1910 |
-
"grad_norm": 5.029311180114746,
|
1911 |
-
"learning_rate": 9.265423966258329e-06,
|
1912 |
-
"loss": 0.0911,
|
1913 |
-
"step": 6800
|
1914 |
-
},
|
1915 |
-
{
|
1916 |
-
"epoch": 0.6828414207103551,
|
1917 |
-
"grad_norm": 48.51480484008789,
|
1918 |
-
"learning_rate": 9.133153743389194e-06,
|
1919 |
-
"loss": 0.0886,
|
1920 |
-
"step": 6825
|
1921 |
-
},
|
1922 |
-
{
|
1923 |
-
"epoch": 0.6853426713356678,
|
1924 |
-
"grad_norm": 80.63517761230469,
|
1925 |
-
"learning_rate": 9.001554509602461e-06,
|
1926 |
-
"loss": 0.1778,
|
1927 |
-
"step": 6850
|
1928 |
-
},
|
1929 |
-
{
|
1930 |
-
"epoch": 0.6878439219609805,
|
1931 |
-
"grad_norm": 0.038998641073703766,
|
1932 |
-
"learning_rate": 8.87063439068248e-06,
|
1933 |
-
"loss": 0.2175,
|
1934 |
-
"step": 6875
|
1935 |
-
},
|
1936 |
-
{
|
1937 |
-
"epoch": 0.6903451725862931,
|
1938 |
-
"grad_norm": 0.11223101615905762,
|
1939 |
-
"learning_rate": 8.740401470480669e-06,
|
1940 |
-
"loss": 0.1244,
|
1941 |
-
"step": 6900
|
1942 |
-
},
|
1943 |
-
{
|
1944 |
-
"epoch": 0.6928464232116058,
|
1945 |
-
"grad_norm": 0.14722932875156403,
|
1946 |
-
"learning_rate": 8.610863790416348e-06,
|
1947 |
-
"loss": 0.106,
|
1948 |
-
"step": 6925
|
1949 |
-
},
|
1950 |
-
{
|
1951 |
-
"epoch": 0.6953476738369184,
|
1952 |
-
"grad_norm": 12.319393157958984,
|
1953 |
-
"learning_rate": 8.48202934898027e-06,
|
1954 |
-
"loss": 0.0855,
|
1955 |
-
"step": 6950
|
1956 |
-
},
|
1957 |
-
{
|
1958 |
-
"epoch": 0.6978489244622311,
|
1959 |
-
"grad_norm": 19.201461791992188,
|
1960 |
-
"learning_rate": 8.353906101240667e-06,
|
1961 |
-
"loss": 0.1162,
|
1962 |
-
"step": 6975
|
1963 |
-
},
|
1964 |
-
{
|
1965 |
-
"epoch": 0.7003501750875438,
|
1966 |
-
"grad_norm": 16.107728958129883,
|
1967 |
-
"learning_rate": 8.226501958352127e-06,
|
1968 |
-
"loss": 0.068,
|
1969 |
-
"step": 7000
|
1970 |
-
},
|
1971 |
-
{
|
1972 |
-
"epoch": 0.7028514257128564,
|
1973 |
-
"grad_norm": 0.7310090661048889,
|
1974 |
-
"learning_rate": 8.099824787067047e-06,
|
1975 |
-
"loss": 0.1714,
|
1976 |
-
"step": 7025
|
1977 |
-
},
|
1978 |
-
{
|
1979 |
-
"epoch": 0.7053526763381691,
|
1980 |
-
"grad_norm": 44.45109176635742,
|
1981 |
-
"learning_rate": 7.973882409249917e-06,
|
1982 |
-
"loss": 0.0928,
|
1983 |
-
"step": 7050
|
1984 |
-
},
|
1985 |
-
{
|
1986 |
-
"epoch": 0.7078539269634817,
|
1987 |
-
"grad_norm": 13.38119888305664,
|
1988 |
-
"learning_rate": 7.848682601394366e-06,
|
1989 |
-
"loss": 0.1165,
|
1990 |
-
"step": 7075
|
1991 |
-
},
|
1992 |
-
{
|
1993 |
-
"epoch": 0.7103551775887944,
|
1994 |
-
"grad_norm": 18.47691535949707,
|
1995 |
-
"learning_rate": 7.724233094142943e-06,
|
1996 |
-
"loss": 0.198,
|
1997 |
-
"step": 7100
|
1998 |
-
},
|
1999 |
-
{
|
2000 |
-
"epoch": 0.7128564282141071,
|
2001 |
-
"grad_norm": 0.15507571399211884,
|
2002 |
-
"learning_rate": 7.6005415718098275e-06,
|
2003 |
-
"loss": 0.0539,
|
2004 |
-
"step": 7125
|
2005 |
-
},
|
2006 |
-
{
|
2007 |
-
"epoch": 0.7153576788394197,
|
2008 |
-
"grad_norm": 0.36700746417045593,
|
2009 |
-
"learning_rate": 7.47761567190632e-06,
|
2010 |
-
"loss": 0.137,
|
2011 |
-
"step": 7150
|
2012 |
-
},
|
2013 |
-
{
|
2014 |
-
"epoch": 0.7178589294647324,
|
2015 |
-
"grad_norm": 16.31074333190918,
|
2016 |
-
"learning_rate": 7.355462984669251e-06,
|
2017 |
-
"loss": 0.0644,
|
2018 |
-
"step": 7175
|
2019 |
-
},
|
2020 |
-
{
|
2021 |
-
"epoch": 0.720360180090045,
|
2022 |
-
"grad_norm": 57.09993362426758,
|
2023 |
-
"learning_rate": 7.234091052592338e-06,
|
2024 |
-
"loss": 0.1802,
|
2025 |
-
"step": 7200
|
2026 |
-
},
|
2027 |
-
{
|
2028 |
-
"epoch": 0.7228614307153577,
|
2029 |
-
"grad_norm": 0.11989270895719528,
|
2030 |
-
"learning_rate": 7.113507369960428e-06,
|
2031 |
-
"loss": 0.1644,
|
2032 |
-
"step": 7225
|
2033 |
-
},
|
2034 |
-
{
|
2035 |
-
"epoch": 0.7253626813406704,
|
2036 |
-
"grad_norm": 0.39681175351142883,
|
2037 |
-
"learning_rate": 6.993719382386786e-06,
|
2038 |
-
"loss": 0.0559,
|
2039 |
-
"step": 7250
|
2040 |
-
},
|
2041 |
-
{
|
2042 |
-
"epoch": 0.727863931965983,
|
2043 |
-
"grad_norm": 5.562121868133545,
|
2044 |
-
"learning_rate": 6.874734486353334e-06,
|
2045 |
-
"loss": 0.1403,
|
2046 |
-
"step": 7275
|
2047 |
-
},
|
2048 |
-
{
|
2049 |
-
"epoch": 0.7303651825912957,
|
2050 |
-
"grad_norm": 0.17540104687213898,
|
2051 |
-
"learning_rate": 6.756560028753938e-06,
|
2052 |
-
"loss": 0.1223,
|
2053 |
-
"step": 7300
|
2054 |
-
},
|
2055 |
-
{
|
2056 |
-
"epoch": 0.7328664332166083,
|
2057 |
-
"grad_norm": 0.45307788252830505,
|
2058 |
-
"learning_rate": 6.639203306440798e-06,
|
2059 |
-
"loss": 0.0698,
|
2060 |
-
"step": 7325
|
2061 |
-
},
|
2062 |
-
{
|
2063 |
-
"epoch": 0.735367683841921,
|
2064 |
-
"grad_norm": 0.01973259262740612,
|
2065 |
-
"learning_rate": 6.522671565773841e-06,
|
2066 |
-
"loss": 0.0354,
|
2067 |
-
"step": 7350
|
2068 |
-
},
|
2069 |
-
{
|
2070 |
-
"epoch": 0.7378689344672336,
|
2071 |
-
"grad_norm": 46.429176330566406,
|
2072 |
-
"learning_rate": 6.406972002173346e-06,
|
2073 |
-
"loss": 0.0831,
|
2074 |
-
"step": 7375
|
2075 |
-
},
|
2076 |
-
{
|
2077 |
-
"epoch": 0.7403701850925463,
|
2078 |
-
"grad_norm": 31.36944007873535,
|
2079 |
-
"learning_rate": 6.292111759675588e-06,
|
2080 |
-
"loss": 0.0836,
|
2081 |
-
"step": 7400
|
2082 |
-
},
|
2083 |
-
{
|
2084 |
-
"epoch": 0.742871435717859,
|
2085 |
-
"grad_norm": 0.021522171795368195,
|
2086 |
-
"learning_rate": 6.17809793049178e-06,
|
2087 |
-
"loss": 0.1261,
|
2088 |
-
"step": 7425
|
2089 |
-
},
|
2090 |
-
{
|
2091 |
-
"epoch": 0.7453726863431716,
|
2092 |
-
"grad_norm": 22.408082962036133,
|
2093 |
-
"learning_rate": 6.064937554570103e-06,
|
2094 |
-
"loss": 0.1255,
|
2095 |
-
"step": 7450
|
2096 |
-
},
|
2097 |
-
{
|
2098 |
-
"epoch": 0.7478739369684843,
|
2099 |
-
"grad_norm": 0.3175584673881531,
|
2100 |
-
"learning_rate": 5.952637619161039e-06,
|
2101 |
-
"loss": 0.1793,
|
2102 |
-
"step": 7475
|
2103 |
-
},
|
2104 |
-
{
|
2105 |
-
"epoch": 0.7503751875937968,
|
2106 |
-
"grad_norm": 0.035180915147066116,
|
2107 |
-
"learning_rate": 5.84120505838595e-06,
|
2108 |
-
"loss": 0.1175,
|
2109 |
-
"step": 7500
|
2110 |
-
},
|
2111 |
-
{
|
2112 |
-
"epoch": 0.7528764382191095,
|
2113 |
-
"grad_norm": 0.10080394148826599,
|
2114 |
-
"learning_rate": 5.730646752808872e-06,
|
2115 |
-
"loss": 0.0984,
|
2116 |
-
"step": 7525
|
2117 |
-
},
|
2118 |
-
{
|
2119 |
-
"epoch": 0.7553776888444222,
|
2120 |
-
"grad_norm": 0.893723726272583,
|
2121 |
-
"learning_rate": 5.6209695290117195e-06,
|
2122 |
-
"loss": 0.0961,
|
2123 |
-
"step": 7550
|
2124 |
-
},
|
2125 |
-
{
|
2126 |
-
"epoch": 0.7578789394697348,
|
2127 |
-
"grad_norm": 0.15472827851772308,
|
2128 |
-
"learning_rate": 5.5121801591727355e-06,
|
2129 |
-
"loss": 0.1479,
|
2130 |
-
"step": 7575
|
2131 |
-
},
|
2132 |
-
{
|
2133 |
-
"epoch": 0.7603801900950475,
|
2134 |
-
"grad_norm": 0.188345804810524,
|
2135 |
-
"learning_rate": 5.404285360648327e-06,
|
2136 |
-
"loss": 0.1861,
|
2137 |
-
"step": 7600
|
2138 |
-
},
|
2139 |
-
{
|
2140 |
-
"epoch": 0.7628814407203601,
|
2141 |
-
"grad_norm": 15.268199920654297,
|
2142 |
-
"learning_rate": 5.297291795558328e-06,
|
2143 |
-
"loss": 0.1135,
|
2144 |
-
"step": 7625
|
2145 |
-
},
|
2146 |
-
{
|
2147 |
-
"epoch": 0.7653826913456728,
|
2148 |
-
"grad_norm": 62.92388153076172,
|
2149 |
-
"learning_rate": 5.1912060703745924e-06,
|
2150 |
-
"loss": 0.1844,
|
2151 |
-
"step": 7650
|
2152 |
-
},
|
2153 |
-
{
|
2154 |
-
"epoch": 0.7678839419709855,
|
2155 |
-
"grad_norm": 7.1327362060546875,
|
2156 |
-
"learning_rate": 5.086034735513113e-06,
|
2157 |
-
"loss": 0.1233,
|
2158 |
-
"step": 7675
|
2159 |
-
},
|
2160 |
-
{
|
2161 |
-
"epoch": 0.7703851925962981,
|
2162 |
-
"grad_norm": 0.43754327297210693,
|
2163 |
-
"learning_rate": 4.9817842849295226e-06,
|
2164 |
-
"loss": 0.0384,
|
2165 |
-
"step": 7700
|
2166 |
-
},
|
2167 |
-
{
|
2168 |
-
"epoch": 0.7728864432216108,
|
2169 |
-
"grad_norm": 0.17376038432121277,
|
2170 |
-
"learning_rate": 4.878461155718122e-06,
|
2171 |
-
"loss": 0.1082,
|
2172 |
-
"step": 7725
|
2173 |
-
},
|
2174 |
-
{
|
2175 |
-
"epoch": 0.7753876938469234,
|
2176 |
-
"grad_norm": 0.25416675209999084,
|
2177 |
-
"learning_rate": 4.776071727714433e-06,
|
2178 |
-
"loss": 0.2028,
|
2179 |
-
"step": 7750
|
2180 |
-
},
|
2181 |
-
{
|
2182 |
-
"epoch": 0.7778889444722361,
|
2183 |
-
"grad_norm": 0.09825768321752548,
|
2184 |
-
"learning_rate": 4.674622323101232e-06,
|
2185 |
-
"loss": 0.1244,
|
2186 |
-
"step": 7775
|
2187 |
-
},
|
2188 |
-
{
|
2189 |
-
"epoch": 0.7803901950975488,
|
2190 |
-
"grad_norm": 0.06236129254102707,
|
2191 |
-
"learning_rate": 4.5741192060182175e-06,
|
2192 |
-
"loss": 0.0404,
|
2193 |
-
"step": 7800
|
2194 |
-
},
|
2195 |
-
{
|
2196 |
-
"epoch": 0.7828914457228614,
|
2197 |
-
"grad_norm": 0.1879894882440567,
|
2198 |
-
"learning_rate": 4.4745685821751826e-06,
|
2199 |
-
"loss": 0.1201,
|
2200 |
-
"step": 7825
|
2201 |
-
},
|
2202 |
-
{
|
2203 |
-
"epoch": 0.7853926963481741,
|
2204 |
-
"grad_norm": 0.05336209759116173,
|
2205 |
-
"learning_rate": 4.375976598468857e-06,
|
2206 |
-
"loss": 0.1227,
|
2207 |
-
"step": 7850
|
2208 |
-
},
|
2209 |
-
{
|
2210 |
-
"epoch": 0.7878939469734867,
|
2211 |
-
"grad_norm": 11.95443058013916,
|
2212 |
-
"learning_rate": 4.278349342603365e-06,
|
2213 |
-
"loss": 0.1147,
|
2214 |
-
"step": 7875
|
2215 |
-
},
|
2216 |
-
{
|
2217 |
-
"epoch": 0.7903951975987994,
|
2218 |
-
"grad_norm": 63.00650405883789,
|
2219 |
-
"learning_rate": 4.181692842714302e-06,
|
2220 |
-
"loss": 0.1186,
|
2221 |
-
"step": 7900
|
2222 |
-
},
|
2223 |
-
{
|
2224 |
-
"epoch": 0.7928964482241121,
|
2225 |
-
"grad_norm": 0.10440034419298172,
|
2226 |
-
"learning_rate": 4.086013066996548e-06,
|
2227 |
-
"loss": 0.148,
|
2228 |
-
"step": 7925
|
2229 |
-
},
|
2230 |
-
{
|
2231 |
-
"epoch": 0.7953976988494247,
|
2232 |
-
"grad_norm": 0.09581203013658524,
|
2233 |
-
"learning_rate": 3.99131592333575e-06,
|
2234 |
-
"loss": 0.0956,
|
2235 |
-
"step": 7950
|
2236 |
-
},
|
2237 |
-
{
|
2238 |
-
"epoch": 0.7978989494747374,
|
2239 |
-
"grad_norm": 13.56506633758545,
|
2240 |
-
"learning_rate": 3.897607258943499e-06,
|
2241 |
-
"loss": 0.1176,
|
2242 |
-
"step": 7975
|
2243 |
-
},
|
2244 |
-
{
|
2245 |
-
"epoch": 0.80040020010005,
|
2246 |
-
"grad_norm": 49.25312423706055,
|
2247 |
-
"learning_rate": 3.804892859996332e-06,
|
2248 |
-
"loss": 0.125,
|
2249 |
-
"step": 8000
|
2250 |
-
},
|
2251 |
-
{
|
2252 |
-
"epoch": 0.8029014507253627,
|
2253 |
-
"grad_norm": 0.061981312930583954,
|
2254 |
-
"learning_rate": 3.713178451278414e-06,
|
2255 |
-
"loss": 0.049,
|
2256 |
-
"step": 8025
|
2257 |
-
},
|
2258 |
-
{
|
2259 |
-
"epoch": 0.8054027013506754,
|
2260 |
-
"grad_norm": 0.1612907499074936,
|
2261 |
-
"learning_rate": 3.622469695828086e-06,
|
2262 |
-
"loss": 0.0897,
|
2263 |
-
"step": 8050
|
2264 |
-
},
|
2265 |
-
{
|
2266 |
-
"epoch": 0.807903951975988,
|
2267 |
-
"grad_norm": 0.8913385272026062,
|
2268 |
-
"learning_rate": 3.532772194588181e-06,
|
2269 |
-
"loss": 0.0713,
|
2270 |
-
"step": 8075
|
2271 |
-
},
|
2272 |
-
{
|
2273 |
-
"epoch": 0.8104052026013007,
|
2274 |
-
"grad_norm": 39.825016021728516,
|
2275 |
-
"learning_rate": 3.444091486060164e-06,
|
2276 |
-
"loss": 0.112,
|
2277 |
-
"step": 8100
|
2278 |
-
},
|
2279 |
-
{
|
2280 |
-
"epoch": 0.8129064532266133,
|
2281 |
-
"grad_norm": 0.1019916981458664,
|
2282 |
-
"learning_rate": 3.356433045962195e-06,
|
2283 |
-
"loss": 0.2174,
|
2284 |
-
"step": 8125
|
2285 |
-
},
|
2286 |
-
{
|
2287 |
-
"epoch": 0.815407703851926,
|
2288 |
-
"grad_norm": 0.33710241317749023,
|
2289 |
-
"learning_rate": 3.2698022868909707e-06,
|
2290 |
-
"loss": 0.1779,
|
2291 |
-
"step": 8150
|
2292 |
-
},
|
2293 |
-
{
|
2294 |
-
"epoch": 0.8179089544772387,
|
2295 |
-
"grad_norm": 0.5362693667411804,
|
2296 |
-
"learning_rate": 3.184204557987565e-06,
|
2297 |
-
"loss": 0.1632,
|
2298 |
-
"step": 8175
|
2299 |
-
},
|
2300 |
-
{
|
2301 |
-
"epoch": 0.8204102051025512,
|
2302 |
-
"grad_norm": 0.2486535906791687,
|
2303 |
-
"learning_rate": 3.0996451446070954e-06,
|
2304 |
-
"loss": 0.0718,
|
2305 |
-
"step": 8200
|
2306 |
-
},
|
2307 |
-
{
|
2308 |
-
"epoch": 0.822911455727864,
|
2309 |
-
"grad_norm": 0.03348654881119728,
|
2310 |
-
"learning_rate": 3.016129267992407e-06,
|
2311 |
-
"loss": 0.1162,
|
2312 |
-
"step": 8225
|
2313 |
-
},
|
2314 |
-
{
|
2315 |
-
"epoch": 0.8254127063531765,
|
2316 |
-
"grad_norm": 0.10843271017074585,
|
2317 |
-
"learning_rate": 2.933662084951656e-06,
|
2318 |
-
"loss": 0.0491,
|
2319 |
-
"step": 8250
|
2320 |
-
},
|
2321 |
-
{
|
2322 |
-
"epoch": 0.8279139569784892,
|
2323 |
-
"grad_norm": 5.107527256011963,
|
2324 |
-
"learning_rate": 2.852248687539896e-06,
|
2325 |
-
"loss": 0.1395,
|
2326 |
-
"step": 8275
|
2327 |
-
},
|
2328 |
-
{
|
2329 |
-
"epoch": 0.8304152076038019,
|
2330 |
-
"grad_norm": 0.1305007040500641,
|
2331 |
-
"learning_rate": 2.771894102744681e-06,
|
2332 |
-
"loss": 0.0786,
|
2333 |
-
"step": 8300
|
2334 |
-
},
|
2335 |
-
{
|
2336 |
-
"epoch": 0.8329164582291145,
|
2337 |
-
"grad_norm": 22.69672393798828,
|
2338 |
-
"learning_rate": 2.692603292175655e-06,
|
2339 |
-
"loss": 0.1479,
|
2340 |
-
"step": 8325
|
2341 |
-
},
|
2342 |
-
{
|
2343 |
-
"epoch": 0.8354177088544272,
|
2344 |
-
"grad_norm": 8.569592475891113,
|
2345 |
-
"learning_rate": 2.614381151758172e-06,
|
2346 |
-
"loss": 0.1518,
|
2347 |
-
"step": 8350
|
2348 |
-
},
|
2349 |
-
{
|
2350 |
-
"epoch": 0.8379189594797398,
|
2351 |
-
"grad_norm": 0.10411455482244492,
|
2352 |
-
"learning_rate": 2.537232511431027e-06,
|
2353 |
-
"loss": 0.1456,
|
2354 |
-
"step": 8375
|
2355 |
-
},
|
2356 |
-
{
|
2357 |
-
"epoch": 0.8404202101050525,
|
2358 |
-
"grad_norm": 0.32852107286453247,
|
2359 |
-
"learning_rate": 2.4611621348481895e-06,
|
2360 |
-
"loss": 0.1307,
|
2361 |
-
"step": 8400
|
2362 |
-
},
|
2363 |
-
{
|
2364 |
-
"epoch": 0.8429214607303652,
|
2365 |
-
"grad_norm": 19.061006546020508,
|
2366 |
-
"learning_rate": 2.3861747190846906e-06,
|
2367 |
-
"loss": 0.1305,
|
2368 |
-
"step": 8425
|
2369 |
-
},
|
2370 |
-
{
|
2371 |
-
"epoch": 0.8454227113556778,
|
2372 |
-
"grad_norm": 0.4331738352775574,
|
2373 |
-
"learning_rate": 2.312274894346587e-06,
|
2374 |
-
"loss": 0.0846,
|
2375 |
-
"step": 8450
|
2376 |
-
},
|
2377 |
-
{
|
2378 |
-
"epoch": 0.8479239619809905,
|
2379 |
-
"grad_norm": 0.0861942321062088,
|
2380 |
-
"learning_rate": 2.2394672236850456e-06,
|
2381 |
-
"loss": 0.1381,
|
2382 |
-
"step": 8475
|
2383 |
-
},
|
2384 |
-
{
|
2385 |
-
"epoch": 0.8504252126063031,
|
2386 |
-
"grad_norm": 0.3310871124267578,
|
2387 |
-
"learning_rate": 2.167756202714624e-06,
|
2388 |
-
"loss": 0.1632,
|
2389 |
-
"step": 8500
|
2390 |
-
},
|
2391 |
-
{
|
2392 |
-
"epoch": 0.8529264632316158,
|
2393 |
-
"grad_norm": 1.1705411672592163,
|
2394 |
-
"learning_rate": 2.0971462593356453e-06,
|
2395 |
-
"loss": 0.1433,
|
2396 |
-
"step": 8525
|
2397 |
-
},
|
2398 |
-
{
|
2399 |
-
"epoch": 0.8554277138569285,
|
2400 |
-
"grad_norm": 16.15072250366211,
|
2401 |
-
"learning_rate": 2.027641753460825e-06,
|
2402 |
-
"loss": 0.1093,
|
2403 |
-
"step": 8550
|
2404 |
-
},
|
2405 |
-
{
|
2406 |
-
"epoch": 0.8579289644822411,
|
2407 |
-
"grad_norm": 0.42475345730781555,
|
2408 |
-
"learning_rate": 1.9592469767460454e-06,
|
2409 |
-
"loss": 0.1262,
|
2410 |
-
"step": 8575
|
2411 |
-
},
|
2412 |
-
{
|
2413 |
-
"epoch": 0.8604302151075538,
|
2414 |
-
"grad_norm": 26.655654907226562,
|
2415 |
-
"learning_rate": 1.8919661523253552e-06,
|
2416 |
-
"loss": 0.1231,
|
2417 |
-
"step": 8600
|
2418 |
-
},
|
2419 |
-
{
|
2420 |
-
"epoch": 0.8629314657328664,
|
2421 |
-
"grad_norm": 0.2540317475795746,
|
2422 |
-
"learning_rate": 1.8258034345502152e-06,
|
2423 |
-
"loss": 0.1541,
|
2424 |
-
"step": 8625
|
2425 |
-
},
|
2426 |
-
{
|
2427 |
-
"epoch": 0.8654327163581791,
|
2428 |
-
"grad_norm": 0.09159835427999496,
|
2429 |
-
"learning_rate": 1.7607629087329891e-06,
|
2430 |
-
"loss": 0.0756,
|
2431 |
-
"step": 8650
|
2432 |
-
},
|
2433 |
-
{
|
2434 |
-
"epoch": 0.8679339669834918,
|
2435 |
-
"grad_norm": 22.544435501098633,
|
2436 |
-
"learning_rate": 1.6968485908946642e-06,
|
2437 |
-
"loss": 0.1079,
|
2438 |
-
"step": 8675
|
2439 |
-
},
|
2440 |
-
{
|
2441 |
-
"epoch": 0.8704352176088044,
|
2442 |
-
"grad_norm": 6.996135711669922,
|
2443 |
-
"learning_rate": 1.6340644275169105e-06,
|
2444 |
-
"loss": 0.1037,
|
2445 |
-
"step": 8700
|
2446 |
-
},
|
2447 |
-
{
|
2448 |
-
"epoch": 0.8729364682341171,
|
2449 |
-
"grad_norm": 0.05436937138438225,
|
2450 |
-
"learning_rate": 1.5724142952983702e-06,
|
2451 |
-
"loss": 0.1194,
|
2452 |
-
"step": 8725
|
2453 |
-
},
|
2454 |
-
{
|
2455 |
-
"epoch": 0.8754377188594297,
|
2456 |
-
"grad_norm": 0.3073214590549469,
|
2457 |
-
"learning_rate": 1.5119020009153063e-06,
|
2458 |
-
"loss": 0.1527,
|
2459 |
-
"step": 8750
|
2460 |
-
},
|
2461 |
-
{
|
2462 |
-
"epoch": 0.8779389694847424,
|
2463 |
-
"grad_norm": 0.06232787296175957,
|
2464 |
-
"learning_rate": 1.4525312807865332e-06,
|
2465 |
-
"loss": 0.1229,
|
2466 |
-
"step": 8775
|
2467 |
-
},
|
2468 |
-
{
|
2469 |
-
"epoch": 0.8804402201100551,
|
2470 |
-
"grad_norm": 16.079917907714844,
|
2471 |
-
"learning_rate": 1.394305800842728e-06,
|
2472 |
-
"loss": 0.1361,
|
2473 |
-
"step": 8800
|
2474 |
-
},
|
2475 |
-
{
|
2476 |
-
"epoch": 0.8829414707353677,
|
2477 |
-
"grad_norm": 48.7065544128418,
|
2478 |
-
"learning_rate": 1.337229156300064e-06,
|
2479 |
-
"loss": 0.2007,
|
2480 |
-
"step": 8825
|
2481 |
-
},
|
2482 |
-
{
|
2483 |
-
"epoch": 0.8854427213606804,
|
2484 |
-
"grad_norm": 0.08288642019033432,
|
2485 |
-
"learning_rate": 1.2813048714382048e-06,
|
2486 |
-
"loss": 0.1104,
|
2487 |
-
"step": 8850
|
2488 |
-
},
|
2489 |
-
{
|
2490 |
-
"epoch": 0.887943971985993,
|
2491 |
-
"grad_norm": 0.034835901111364365,
|
2492 |
-
"learning_rate": 1.2265363993827183e-06,
|
2493 |
-
"loss": 0.1474,
|
2494 |
-
"step": 8875
|
2495 |
-
},
|
2496 |
-
{
|
2497 |
-
"epoch": 0.8904452226113057,
|
2498 |
-
"grad_norm": 22.826528549194336,
|
2499 |
-
"learning_rate": 1.1729271218918336e-06,
|
2500 |
-
"loss": 0.1698,
|
2501 |
-
"step": 8900
|
2502 |
-
},
|
2503 |
-
{
|
2504 |
-
"epoch": 0.8929464732366184,
|
2505 |
-
"grad_norm": 0.1743120402097702,
|
2506 |
-
"learning_rate": 1.1204803491476479e-06,
|
2507 |
-
"loss": 0.081,
|
2508 |
-
"step": 8925
|
2509 |
-
},
|
2510 |
-
{
|
2511 |
-
"epoch": 0.8954477238619309,
|
2512 |
-
"grad_norm": 22.49924659729004,
|
2513 |
-
"learning_rate": 1.069199319551728e-06,
|
2514 |
-
"loss": 0.2201,
|
2515 |
-
"step": 8950
|
2516 |
-
},
|
2517 |
-
{
|
2518 |
-
"epoch": 0.8979489744872436,
|
2519 |
-
"grad_norm": 14.826763153076172,
|
2520 |
-
"learning_rate": 1.0190871995251462e-06,
|
2521 |
-
"loss": 0.2164,
|
2522 |
-
"step": 8975
|
2523 |
-
},
|
2524 |
-
{
|
2525 |
-
"epoch": 0.9004502251125562,
|
2526 |
-
"grad_norm": 0.1625366061925888,
|
2527 |
-
"learning_rate": 9.70147083312969e-07,
|
2528 |
-
"loss": 0.09,
|
2529 |
-
"step": 9000
|
2530 |
-
},
|
2531 |
-
{
|
2532 |
-
"epoch": 0.9029514757378689,
|
2533 |
-
"grad_norm": 108.10717010498047,
|
2534 |
-
"learning_rate": 9.223819927932043e-07,
|
2535 |
-
"loss": 0.1335,
|
2536 |
-
"step": 9025
|
2537 |
-
},
|
2538 |
-
{
|
2539 |
-
"epoch": 0.9054527263631816,
|
2540 |
-
"grad_norm": 0.8745747208595276,
|
2541 |
-
"learning_rate": 8.757948772901925e-07,
|
2542 |
-
"loss": 0.0634,
|
2543 |
-
"step": 9050
|
2544 |
-
},
|
2545 |
-
{
|
2546 |
-
"epoch": 0.9079539769884942,
|
2547 |
-
"grad_norm": 20.332319259643555,
|
2548 |
-
"learning_rate": 8.303886133925277e-07,
|
2549 |
-
"loss": 0.1258,
|
2550 |
-
"step": 9075
|
2551 |
-
},
|
2552 |
-
{
|
2553 |
-
"epoch": 0.9104552276138069,
|
2554 |
-
"grad_norm": 0.04837594926357269,
|
2555 |
-
"learning_rate": 7.861660047753994e-07,
|
2556 |
-
"loss": 0.0737,
|
2557 |
-
"step": 9100
|
2558 |
-
},
|
2559 |
-
{
|
2560 |
-
"epoch": 0.9129564782391195,
|
2561 |
-
"grad_norm": 9.852827072143555,
|
2562 |
-
"learning_rate": 7.431297820275163e-07,
|
2563 |
-
"loss": 0.135,
|
2564 |
-
"step": 9125
|
2565 |
-
},
|
2566 |
-
{
|
2567 |
-
"epoch": 0.9154577288644322,
|
2568 |
-
"grad_norm": 0.20781473815441132,
|
2569 |
-
"learning_rate": 7.012826024824626e-07,
|
2570 |
-
"loss": 0.1823,
|
2571 |
-
"step": 9150
|
2572 |
-
},
|
2573 |
-
{
|
2574 |
-
"epoch": 0.9179589794897449,
|
2575 |
-
"grad_norm": 0.06475343555212021,
|
2576 |
-
"learning_rate": 6.606270500546497e-07,
|
2577 |
-
"loss": 0.1681,
|
2578 |
-
"step": 9175
|
2579 |
-
},
|
2580 |
-
{
|
2581 |
-
"epoch": 0.9204602301150575,
|
2582 |
-
"grad_norm": 0.04313327744603157,
|
2583 |
-
"learning_rate": 6.211656350797545e-07,
|
2584 |
-
"loss": 0.1359,
|
2585 |
-
"step": 9200
|
2586 |
-
},
|
2587 |
-
{
|
2588 |
-
"epoch": 0.9229614807403702,
|
2589 |
-
"grad_norm": 112.4850082397461,
|
2590 |
-
"learning_rate": 5.829007941597065e-07,
|
2591 |
-
"loss": 0.072,
|
2592 |
-
"step": 9225
|
2593 |
-
},
|
2594 |
-
{
|
2595 |
-
"epoch": 0.9254627313656828,
|
2596 |
-
"grad_norm": 0.024171823635697365,
|
2597 |
-
"learning_rate": 5.458348900122601e-07,
|
2598 |
-
"loss": 0.0997,
|
2599 |
-
"step": 9250
|
2600 |
-
},
|
2601 |
-
{
|
2602 |
-
"epoch": 0.9279639819909955,
|
2603 |
-
"grad_norm": 0.3680287301540375,
|
2604 |
-
"learning_rate": 5.099702113250726e-07,
|
2605 |
-
"loss": 0.164,
|
2606 |
-
"step": 9275
|
2607 |
-
},
|
2608 |
-
{
|
2609 |
-
"epoch": 0.9304652326163082,
|
2610 |
-
"grad_norm": 43.63431167602539,
|
2611 |
-
"learning_rate": 4.753089726144211e-07,
|
2612 |
-
"loss": 0.0589,
|
2613 |
-
"step": 9300
|
2614 |
-
},
|
2615 |
-
{
|
2616 |
-
"epoch": 0.9329664832416208,
|
2617 |
-
"grad_norm": 0.4794533848762512,
|
2618 |
-
"learning_rate": 4.418533140884429e-07,
|
2619 |
-
"loss": 0.1777,
|
2620 |
-
"step": 9325
|
2621 |
-
},
|
2622 |
-
{
|
2623 |
-
"epoch": 0.9354677338669335,
|
2624 |
-
"grad_norm": 0.12671618163585663,
|
2625 |
-
"learning_rate": 4.096053015149792e-07,
|
2626 |
-
"loss": 0.1076,
|
2627 |
-
"step": 9350
|
2628 |
-
},
|
2629 |
-
{
|
2630 |
-
"epoch": 0.9379689844922461,
|
2631 |
-
"grad_norm": 1.0161490440368652,
|
2632 |
-
"learning_rate": 3.785669260940439e-07,
|
2633 |
-
"loss": 0.0471,
|
2634 |
-
"step": 9375
|
2635 |
-
},
|
2636 |
-
{
|
2637 |
-
"epoch": 0.9404702351175588,
|
2638 |
-
"grad_norm": 25.204797744750977,
|
2639 |
-
"learning_rate": 3.4874010433485705e-07,
|
2640 |
-
"loss": 0.1675,
|
2641 |
-
"step": 9400
|
2642 |
-
},
|
2643 |
-
{
|
2644 |
-
"epoch": 0.9429714857428715,
|
2645 |
-
"grad_norm": 0.373117595911026,
|
2646 |
-
"learning_rate": 3.2012667793751117e-07,
|
2647 |
-
"loss": 0.1235,
|
2648 |
-
"step": 9425
|
2649 |
-
},
|
2650 |
-
{
|
2651 |
-
"epoch": 0.9454727363681841,
|
2652 |
-
"grad_norm": 0.3929968476295471,
|
2653 |
-
"learning_rate": 2.92728413679253e-07,
|
2654 |
-
"loss": 0.0748,
|
2655 |
-
"step": 9450
|
2656 |
-
},
|
2657 |
-
{
|
2658 |
-
"epoch": 0.9479739869934968,
|
2659 |
-
"grad_norm": 0.34126216173171997,
|
2660 |
-
"learning_rate": 2.6654700330539296e-07,
|
2661 |
-
"loss": 0.0911,
|
2662 |
-
"step": 9475
|
2663 |
-
},
|
2664 |
-
{
|
2665 |
-
"epoch": 0.9504752376188094,
|
2666 |
-
"grad_norm": 16.82143783569336,
|
2667 |
-
"learning_rate": 2.4158406342484895e-07,
|
2668 |
-
"loss": 0.1222,
|
2669 |
-
"step": 9500
|
2670 |
-
},
|
2671 |
-
{
|
2672 |
-
"epoch": 0.9529764882441221,
|
2673 |
-
"grad_norm": 18.41042137145996,
|
2674 |
-
"learning_rate": 2.178411354103127e-07,
|
2675 |
-
"loss": 0.1254,
|
2676 |
-
"step": 9525
|
2677 |
-
},
|
2678 |
-
{
|
2679 |
-
"epoch": 0.9554777388694348,
|
2680 |
-
"grad_norm": 16.776601791381836,
|
2681 |
-
"learning_rate": 1.9531968530309253e-07,
|
2682 |
-
"loss": 0.1282,
|
2683 |
-
"step": 9550
|
2684 |
-
},
|
2685 |
-
{
|
2686 |
-
"epoch": 0.9579789894947474,
|
2687 |
-
"grad_norm": 0.323078989982605,
|
2688 |
-
"learning_rate": 1.7402110372258363e-07,
|
2689 |
-
"loss": 0.1036,
|
2690 |
-
"step": 9575
|
2691 |
-
},
|
2692 |
-
{
|
2693 |
-
"epoch": 0.9604802401200601,
|
2694 |
-
"grad_norm": 0.08523103594779968,
|
2695 |
-
"learning_rate": 1.539467057803945e-07,
|
2696 |
-
"loss": 0.1907,
|
2697 |
-
"step": 9600
|
2698 |
-
},
|
2699 |
-
{
|
2700 |
-
"epoch": 0.9629814907453726,
|
2701 |
-
"grad_norm": 176.9346923828125,
|
2702 |
-
"learning_rate": 1.350977309991608e-07,
|
2703 |
-
"loss": 0.159,
|
2704 |
-
"step": 9625
|
2705 |
-
},
|
2706 |
-
{
|
2707 |
-
"epoch": 0.9654827413706853,
|
2708 |
-
"grad_norm": 46.01314163208008,
|
2709 |
-
"learning_rate": 1.1747534323599097e-07,
|
2710 |
-
"loss": 0.2587,
|
2711 |
-
"step": 9650
|
2712 |
-
},
|
2713 |
-
{
|
2714 |
-
"epoch": 0.967983991995998,
|
2715 |
-
"grad_norm": 52.705528259277344,
|
2716 |
-
"learning_rate": 1.0108063061061268e-07,
|
2717 |
-
"loss": 0.0883,
|
2718 |
-
"step": 9675
|
2719 |
-
},
|
2720 |
-
{
|
2721 |
-
"epoch": 0.9704852426213106,
|
2722 |
-
"grad_norm": 10.609136581420898,
|
2723 |
-
"learning_rate": 8.59146054381843e-08,
|
2724 |
-
"loss": 0.1503,
|
2725 |
-
"step": 9700
|
2726 |
-
},
|
2727 |
-
{
|
2728 |
-
"epoch": 0.9729864932466233,
|
2729 |
-
"grad_norm": 11.975505828857422,
|
2730 |
-
"learning_rate": 7.197820416678491e-08,
|
2731 |
-
"loss": 0.1068,
|
2732 |
-
"step": 9725
|
2733 |
-
},
|
2734 |
-
{
|
2735 |
-
"epoch": 0.9754877438719359,
|
2736 |
-
"grad_norm": 0.053060755133628845,
|
2737 |
-
"learning_rate": 5.927228731959611e-08,
|
2738 |
-
"loss": 0.0549,
|
2739 |
-
"step": 9750
|
2740 |
-
},
|
2741 |
-
{
|
2742 |
-
"epoch": 0.9779889944972486,
|
2743 |
-
"grad_norm": 0.0601852610707283,
|
2744 |
-
"learning_rate": 4.779763944176452e-08,
|
2745 |
-
"loss": 0.0566,
|
2746 |
-
"step": 9775
|
2747 |
-
},
|
2748 |
-
{
|
2749 |
-
"epoch": 0.9804902451225613,
|
2750 |
-
"grad_norm": 77.97218322753906,
|
2751 |
-
"learning_rate": 3.755496905196054e-08,
|
2752 |
-
"loss": 0.1206,
|
2753 |
-
"step": 9800
|
2754 |
-
},
|
2755 |
-
{
|
2756 |
-
"epoch": 0.9829914957478739,
|
2757 |
-
"grad_norm": 38.3494987487793,
|
2758 |
-
"learning_rate": 2.8544908598626687e-08,
|
2759 |
-
"loss": 0.0942,
|
2760 |
-
"step": 9825
|
2761 |
-
},
|
2762 |
-
{
|
2763 |
-
"epoch": 0.9854927463731866,
|
2764 |
-
"grad_norm": 24.240922927856445,
|
2765 |
-
"learning_rate": 2.0768014420926573e-08,
|
2766 |
-
"loss": 0.185,
|
2767 |
-
"step": 9850
|
2768 |
-
},
|
2769 |
-
{
|
2770 |
-
"epoch": 0.9879939969984992,
|
2771 |
-
"grad_norm": 0.20389939844608307,
|
2772 |
-
"learning_rate": 1.4224766714405757e-08,
|
2773 |
-
"loss": 0.1755,
|
2774 |
-
"step": 9875
|
2775 |
-
},
|
2776 |
-
{
|
2777 |
-
"epoch": 0.9904952476238119,
|
2778 |
-
"grad_norm": 0.09934602677822113,
|
2779 |
-
"learning_rate": 8.915569501317668e-09,
|
2780 |
-
"loss": 0.1064,
|
2781 |
-
"step": 9900
|
2782 |
-
},
|
2783 |
-
{
|
2784 |
-
"epoch": 0.9929964982491246,
|
2785 |
-
"grad_norm": 0.4000752866268158,
|
2786 |
-
"learning_rate": 4.840750605699107e-09,
|
2787 |
-
"loss": 0.1544,
|
2788 |
-
"step": 9925
|
2789 |
-
},
|
2790 |
-
{
|
2791 |
-
"epoch": 0.9954977488744372,
|
2792 |
-
"grad_norm": 15.259411811828613,
|
2793 |
-
"learning_rate": 2.000561633110909e-09,
|
2794 |
-
"loss": 0.147,
|
2795 |
-
"step": 9950
|
2796 |
-
},
|
2797 |
-
{
|
2798 |
-
"epoch": 0.9979989994997499,
|
2799 |
-
"grad_norm": 0.06149940937757492,
|
2800 |
-
"learning_rate": 3.951779551192303e-10,
|
2801 |
-
"loss": 0.1256,
|
2802 |
-
"step": 9975
|
2803 |
-
},
|
2804 |
-
{
|
2805 |
-
"epoch": 1.0,
|
2806 |
-
"eval_accuracy": 0.9650764176546208,
|
2807 |
-
"eval_auc": 0.9825065788769469,
|
2808 |
-
"eval_f1": 0.9682488150120122,
|
2809 |
-
"eval_loss": 0.16612961888313293,
|
2810 |
-
"eval_precision": 0.9631830512853636,
|
2811 |
-
"eval_recall": 0.9733681462140992,
|
2812 |
-
"eval_runtime": 2160.2321,
|
2813 |
-
"eval_samples_per_second": 6.482,
|
2814 |
-
"eval_steps_per_second": 0.27,
|
2815 |
-
"step": 9995
|
2816 |
-
}
|
2817 |
-
],
|
2818 |
-
"logging_steps": 25,
|
2819 |
-
"max_steps": 9995,
|
2820 |
-
"num_input_tokens_seen": 0,
|
2821 |
-
"num_train_epochs": 1,
|
2822 |
-
"save_steps": 500,
|
2823 |
-
"stateful_callbacks": {
|
2824 |
-
"EarlyStoppingCallback": {
|
2825 |
-
"args": {
|
2826 |
-
"early_stopping_patience": 5,
|
2827 |
-
"early_stopping_threshold": 0.01
|
2828 |
-
},
|
2829 |
-
"attributes": {
|
2830 |
-
"early_stopping_patience_counter": 0
|
2831 |
-
}
|
2832 |
-
},
|
2833 |
-
"TrainerControl": {
|
2834 |
-
"args": {
|
2835 |
-
"should_epoch_stop": false,
|
2836 |
-
"should_evaluate": false,
|
2837 |
-
"should_log": false,
|
2838 |
-
"should_save": true,
|
2839 |
-
"should_training_stop": true
|
2840 |
-
},
|
2841 |
-
"attributes": {}
|
2842 |
-
}
|
2843 |
-
},
|
2844 |
-
"total_flos": 2.7618647936683475e+19,
|
2845 |
-
"train_batch_size": 12,
|
2846 |
-
"trial_name": null,
|
2847 |
-
"trial_params": null
|
2848 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "haywoodsloan/
|
3 |
"_num_labels": 2,
|
4 |
"architectures": [
|
5 |
"Swinv2ForImageClassification"
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "haywoodsloan/ai-image-detector-deploy",
|
3 |
"_num_labels": 2,
|
4 |
"architectures": [
|
5 |
"Swinv2ForImageClassification"
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 780886280
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3aa6f9de7bb58e821bbd577afe92c1092a78d14210b399ae151ab661d2e43a57
|
3 |
size 780886280
|
runs/{Nov17_06-39-09_2db0527f7c04/events.out.tfevents.1731825557.2db0527f7c04.61.0 β Nov22_06-35-14_170e5a2edfa3/events.out.tfevents.1732257322.170e5a2edfa3.82.0}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9240fdd36303f544ef598c00e9233cf57562ba6e5f96fcfb999db899ef83e85c
|
3 |
+
size 91937
|
runs/{Nov17_06-39-09_2db0527f7c04/events.out.tfevents.1731860413.2db0527f7c04.61.1 β Nov22_06-35-14_170e5a2edfa3/events.out.tfevents.1732300664.170e5a2edfa3.82.1}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 607
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a0201db49d73f85ea9f3615cd140c14ca04201f4a7571cc2ad9f4b3729508a0d
|
3 |
size 607
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:23acb2db99f02f49dd8f0a9e1af4931c2df606823265dce45ae76a632ad77be5
|
3 |
size 5304
|
training_params.json
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
"data_path": "haywoodsloan/ai-images",
|
3 |
-
"model": "haywoodsloan/
|
4 |
"username": "haywoodsloan",
|
5 |
"lr": 4e-05,
|
6 |
"epochs": 1,
|
@@ -11,11 +11,11 @@
|
|
11 |
"scheduler": "cosine",
|
12 |
"weight_decay": 0.001,
|
13 |
"max_grad_norm": 1.0,
|
14 |
-
"seed":
|
15 |
"train_split": "train",
|
16 |
"valid_split": "test",
|
17 |
"logging_steps": -1,
|
18 |
-
"project_name": "autotrain-ai-image-detect-
|
19 |
"auto_find_batch_size": true,
|
20 |
"mixed_precision": "bf16",
|
21 |
"save_total_limit": 10,
|
|
|
1 |
{
|
2 |
"data_path": "haywoodsloan/ai-images",
|
3 |
+
"model": "haywoodsloan/ai-image-detector-deploy",
|
4 |
"username": "haywoodsloan",
|
5 |
"lr": 4e-05,
|
6 |
"epochs": 1,
|
|
|
11 |
"scheduler": "cosine",
|
12 |
"weight_decay": 0.001,
|
13 |
"max_grad_norm": 1.0,
|
14 |
+
"seed": 9583,
|
15 |
"train_split": "train",
|
16 |
"valid_split": "test",
|
17 |
"logging_steps": -1,
|
18 |
+
"project_name": "autotrain-ai-image-detect-20241122-0116",
|
19 |
"auto_find_batch_size": true,
|
20 |
"mixed_precision": "bf16",
|
21 |
"save_total_limit": 10,
|