Upload model + model card
Browse files- README.md +9 -0
- checkpoint-17312/config.json +0 -0
- checkpoint-17312/model.safetensors +3 -0
- checkpoint-17312/optimizer.pt +3 -0
- checkpoint-17312/preprocessor_config.json +23 -0
- checkpoint-17312/rng_state.pth +3 -0
- checkpoint-17312/scaler.pt +3 -0
- checkpoint-17312/scheduler.pt +3 -0
- checkpoint-17312/trainer_state.json +2480 -0
- checkpoint-17312/training_args.bin +3 -0
- checkpoint-25968/config.json +0 -0
- checkpoint-25968/model.safetensors +3 -0
- checkpoint-25968/optimizer.pt +3 -0
- checkpoint-25968/preprocessor_config.json +23 -0
- checkpoint-25968/rng_state.pth +3 -0
- checkpoint-25968/scaler.pt +3 -0
- checkpoint-25968/scheduler.pt +3 -0
- checkpoint-25968/trainer_state.json +3703 -0
- checkpoint-25968/training_args.bin +3 -0
- checkpoint-34624/config.json +0 -0
- checkpoint-34624/model.safetensors +3 -0
- checkpoint-34624/optimizer.pt +3 -0
- checkpoint-34624/preprocessor_config.json +23 -0
- checkpoint-34624/rng_state.pth +3 -0
- checkpoint-34624/scaler.pt +3 -0
- checkpoint-34624/scheduler.pt +3 -0
- checkpoint-34624/trainer_state.json +0 -0
- checkpoint-34624/training_args.bin +3 -0
- checkpoint-8656/config.json +0 -0
- checkpoint-8656/model.safetensors +3 -0
- checkpoint-8656/optimizer.pt +3 -0
- checkpoint-8656/preprocessor_config.json +23 -0
- checkpoint-8656/rng_state.pth +3 -0
- checkpoint-8656/scaler.pt +3 -0
- checkpoint-8656/scheduler.pt +3 -0
- checkpoint-8656/trainer_state.json +1257 -0
- checkpoint-8656/training_args.bin +3 -0
- config.json +0 -0
- model.safetensors +3 -0
- model_card.json +7 -0
- preprocessor_config.json +23 -0
- training_args.bin +3 -0
README.md
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# rossijakob/rossi-ray1
|
| 2 |
+
|
| 3 |
+
**Description:** roadwork-detect model
|
| 4 |
+
|
| 5 |
+
**Version:** 1.0.0
|
| 6 |
+
|
| 7 |
+
**Submitted by:** 5HEUFxZpguKZnYV7RtGwnSsbB1bQ6k8pS3Qq5nEACawtooJR
|
| 8 |
+
|
| 9 |
+
**Submission time:** 1760590276
|
checkpoint-17312/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-17312/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:526c3ce95428683243b9aa39cd3b93dcc230249df08c852345b2e15e2a26467e
|
| 3 |
+
size 631399808
|
checkpoint-17312/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b968ba8e49315ac3bf85ae464ad8e8948556f22aabbe004dc86ccdbcbdd572c
|
| 3 |
+
size 1262124766
|
checkpoint-17312/preprocessor_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.485,
|
| 8 |
+
0.456,
|
| 9 |
+
0.406
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "ViTImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.229,
|
| 14 |
+
0.224,
|
| 15 |
+
0.225
|
| 16 |
+
],
|
| 17 |
+
"resample": 3,
|
| 18 |
+
"rescale_factor": 0.00392156862745098,
|
| 19 |
+
"size": {
|
| 20 |
+
"height": 224,
|
| 21 |
+
"width": 224
|
| 22 |
+
}
|
| 23 |
+
}
|
checkpoint-17312/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:262c4bbf361fd043c0c35017cb686772240272e851df2feffbd9431a8c2b983b
|
| 3 |
+
size 14645
|
checkpoint-17312/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df808f01cb3f2c0318f96d5686c3dc6adf2ddde4b101767df37687e6b9c4f814
|
| 3 |
+
size 1383
|
checkpoint-17312/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1622efbd225603933e70f02da873a2ad89eada2a39f3ccdf48f45531a38829ee
|
| 3 |
+
size 1465
|
checkpoint-17312/trainer_state.json
ADDED
|
@@ -0,0 +1,2480 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 17312,
|
| 3 |
+
"best_metric": 0.9761680777673252,
|
| 4 |
+
"best_model_checkpoint": "./roadwork-swin-finetuned-local/checkpoint-17312",
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 17312,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00577634011090573,
|
| 14 |
+
"grad_norm": 12.141866683959961,
|
| 15 |
+
"learning_rate": 2.9957543900184844e-05,
|
| 16 |
+
"loss": 3.0039,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01155268022181146,
|
| 21 |
+
"grad_norm": 13.438708305358887,
|
| 22 |
+
"learning_rate": 2.991422134935305e-05,
|
| 23 |
+
"loss": 0.3812,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01732902033271719,
|
| 28 |
+
"grad_norm": 38.80968475341797,
|
| 29 |
+
"learning_rate": 2.9870898798521256e-05,
|
| 30 |
+
"loss": 0.3451,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02310536044362292,
|
| 35 |
+
"grad_norm": 16.20616340637207,
|
| 36 |
+
"learning_rate": 2.9827576247689466e-05,
|
| 37 |
+
"loss": 0.3082,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028881700554528652,
|
| 42 |
+
"grad_norm": 33.16380310058594,
|
| 43 |
+
"learning_rate": 2.9784253696857672e-05,
|
| 44 |
+
"loss": 0.3153,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03465804066543438,
|
| 49 |
+
"grad_norm": 18.74236297607422,
|
| 50 |
+
"learning_rate": 2.974093114602588e-05,
|
| 51 |
+
"loss": 0.3792,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04043438077634011,
|
| 56 |
+
"grad_norm": 4.538590431213379,
|
| 57 |
+
"learning_rate": 2.9697608595194084e-05,
|
| 58 |
+
"loss": 0.3673,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04621072088724584,
|
| 63 |
+
"grad_norm": 15.369573593139648,
|
| 64 |
+
"learning_rate": 2.9654286044362294e-05,
|
| 65 |
+
"loss": 0.3875,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.05198706099815157,
|
| 70 |
+
"grad_norm": 23.82215690612793,
|
| 71 |
+
"learning_rate": 2.96109634935305e-05,
|
| 72 |
+
"loss": 0.3386,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.057763401109057304,
|
| 77 |
+
"grad_norm": 14.058465003967285,
|
| 78 |
+
"learning_rate": 2.9567640942698706e-05,
|
| 79 |
+
"loss": 0.2766,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06353974121996303,
|
| 84 |
+
"grad_norm": 15.244014739990234,
|
| 85 |
+
"learning_rate": 2.9524318391866915e-05,
|
| 86 |
+
"loss": 0.2891,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06931608133086876,
|
| 91 |
+
"grad_norm": 30.079912185668945,
|
| 92 |
+
"learning_rate": 2.948099584103512e-05,
|
| 93 |
+
"loss": 0.315,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0750924214417745,
|
| 98 |
+
"grad_norm": 22.175500869750977,
|
| 99 |
+
"learning_rate": 2.9437673290203327e-05,
|
| 100 |
+
"loss": 0.2571,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08086876155268022,
|
| 105 |
+
"grad_norm": 13.770343780517578,
|
| 106 |
+
"learning_rate": 2.9394350739371534e-05,
|
| 107 |
+
"loss": 0.2805,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08664510166358595,
|
| 112 |
+
"grad_norm": 15.201981544494629,
|
| 113 |
+
"learning_rate": 2.9351028188539743e-05,
|
| 114 |
+
"loss": 0.3395,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09242144177449169,
|
| 119 |
+
"grad_norm": 0.11971250176429749,
|
| 120 |
+
"learning_rate": 2.930770563770795e-05,
|
| 121 |
+
"loss": 0.2807,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0981977818853974,
|
| 126 |
+
"grad_norm": 13.49336051940918,
|
| 127 |
+
"learning_rate": 2.9264383086876155e-05,
|
| 128 |
+
"loss": 0.3474,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10397412199630314,
|
| 133 |
+
"grad_norm": 12.165586471557617,
|
| 134 |
+
"learning_rate": 2.9221060536044365e-05,
|
| 135 |
+
"loss": 0.2704,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.10975046210720887,
|
| 140 |
+
"grad_norm": 14.561868667602539,
|
| 141 |
+
"learning_rate": 2.917773798521257e-05,
|
| 142 |
+
"loss": 0.2495,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.11552680221811461,
|
| 147 |
+
"grad_norm": 3.2725448608398438,
|
| 148 |
+
"learning_rate": 2.9134415434380777e-05,
|
| 149 |
+
"loss": 0.2951,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.12130314232902033,
|
| 154 |
+
"grad_norm": 12.453506469726562,
|
| 155 |
+
"learning_rate": 2.9091092883548983e-05,
|
| 156 |
+
"loss": 0.3338,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12707948243992606,
|
| 161 |
+
"grad_norm": 21.19861602783203,
|
| 162 |
+
"learning_rate": 2.904777033271719e-05,
|
| 163 |
+
"loss": 0.4023,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13285582255083178,
|
| 168 |
+
"grad_norm": 7.617332935333252,
|
| 169 |
+
"learning_rate": 2.90044477818854e-05,
|
| 170 |
+
"loss": 0.2225,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13863216266173753,
|
| 175 |
+
"grad_norm": 7.300302028656006,
|
| 176 |
+
"learning_rate": 2.8961125231053605e-05,
|
| 177 |
+
"loss": 0.2645,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14440850277264325,
|
| 182 |
+
"grad_norm": 11.860857963562012,
|
| 183 |
+
"learning_rate": 2.8917802680221814e-05,
|
| 184 |
+
"loss": 0.3154,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.150184842883549,
|
| 189 |
+
"grad_norm": 20.967884063720703,
|
| 190 |
+
"learning_rate": 2.887448012939002e-05,
|
| 191 |
+
"loss": 0.2485,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.15596118299445472,
|
| 196 |
+
"grad_norm": 11.119569778442383,
|
| 197 |
+
"learning_rate": 2.8831157578558226e-05,
|
| 198 |
+
"loss": 0.2854,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.16173752310536044,
|
| 203 |
+
"grad_norm": 4.218294620513916,
|
| 204 |
+
"learning_rate": 2.8787835027726432e-05,
|
| 205 |
+
"loss": 0.2434,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.16751386321626618,
|
| 210 |
+
"grad_norm": 5.249250888824463,
|
| 211 |
+
"learning_rate": 2.874451247689464e-05,
|
| 212 |
+
"loss": 0.3393,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1732902033271719,
|
| 217 |
+
"grad_norm": 1.712246298789978,
|
| 218 |
+
"learning_rate": 2.8701189926062848e-05,
|
| 219 |
+
"loss": 0.2331,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17906654343807762,
|
| 224 |
+
"grad_norm": 28.958724975585938,
|
| 225 |
+
"learning_rate": 2.8657867375231054e-05,
|
| 226 |
+
"loss": 0.2514,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.18484288354898337,
|
| 231 |
+
"grad_norm": 10.893854141235352,
|
| 232 |
+
"learning_rate": 2.8614544824399264e-05,
|
| 233 |
+
"loss": 0.3656,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1906192236598891,
|
| 238 |
+
"grad_norm": 26.43993377685547,
|
| 239 |
+
"learning_rate": 2.857122227356747e-05,
|
| 240 |
+
"loss": 0.2875,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.1963955637707948,
|
| 245 |
+
"grad_norm": 0.9008654952049255,
|
| 246 |
+
"learning_rate": 2.8527899722735676e-05,
|
| 247 |
+
"loss": 0.2569,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.20217190388170056,
|
| 252 |
+
"grad_norm": 6.621076583862305,
|
| 253 |
+
"learning_rate": 2.8484577171903882e-05,
|
| 254 |
+
"loss": 0.2228,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.20794824399260628,
|
| 259 |
+
"grad_norm": 0.4961116909980774,
|
| 260 |
+
"learning_rate": 2.8441254621072088e-05,
|
| 261 |
+
"loss": 0.368,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.21372458410351203,
|
| 266 |
+
"grad_norm": 31.944732666015625,
|
| 267 |
+
"learning_rate": 2.8397932070240297e-05,
|
| 268 |
+
"loss": 0.2735,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21950092421441775,
|
| 273 |
+
"grad_norm": 34.241703033447266,
|
| 274 |
+
"learning_rate": 2.8354609519408504e-05,
|
| 275 |
+
"loss": 0.2414,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.22527726432532347,
|
| 280 |
+
"grad_norm": 5.89654016494751,
|
| 281 |
+
"learning_rate": 2.8311286968576713e-05,
|
| 282 |
+
"loss": 0.2939,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.23105360443622922,
|
| 287 |
+
"grad_norm": 6.246095657348633,
|
| 288 |
+
"learning_rate": 2.8267964417744916e-05,
|
| 289 |
+
"loss": 0.2772,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.23682994454713494,
|
| 294 |
+
"grad_norm": 12.515145301818848,
|
| 295 |
+
"learning_rate": 2.8224641866913122e-05,
|
| 296 |
+
"loss": 0.2363,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.24260628465804066,
|
| 301 |
+
"grad_norm": 7.227047920227051,
|
| 302 |
+
"learning_rate": 2.818131931608133e-05,
|
| 303 |
+
"loss": 0.2087,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.2483826247689464,
|
| 308 |
+
"grad_norm": 8.628905296325684,
|
| 309 |
+
"learning_rate": 2.8137996765249537e-05,
|
| 310 |
+
"loss": 0.2902,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.2541589648798521,
|
| 315 |
+
"grad_norm": 15.576581001281738,
|
| 316 |
+
"learning_rate": 2.8094674214417747e-05,
|
| 317 |
+
"loss": 0.2752,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.25993530499075784,
|
| 322 |
+
"grad_norm": 16.116729736328125,
|
| 323 |
+
"learning_rate": 2.8051351663585953e-05,
|
| 324 |
+
"loss": 0.2604,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26571164510166356,
|
| 329 |
+
"grad_norm": 12.389599800109863,
|
| 330 |
+
"learning_rate": 2.8008029112754162e-05,
|
| 331 |
+
"loss": 0.2661,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.27148798521256934,
|
| 336 |
+
"grad_norm": 5.002556800842285,
|
| 337 |
+
"learning_rate": 2.7964706561922365e-05,
|
| 338 |
+
"loss": 0.2844,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27726432532347506,
|
| 343 |
+
"grad_norm": 23.136369705200195,
|
| 344 |
+
"learning_rate": 2.792138401109057e-05,
|
| 345 |
+
"loss": 0.2737,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.2830406654343808,
|
| 350 |
+
"grad_norm": 7.032516002655029,
|
| 351 |
+
"learning_rate": 2.787806146025878e-05,
|
| 352 |
+
"loss": 0.2597,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2888170055452865,
|
| 357 |
+
"grad_norm": 24.333494186401367,
|
| 358 |
+
"learning_rate": 2.7834738909426987e-05,
|
| 359 |
+
"loss": 0.2748,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.2945933456561922,
|
| 364 |
+
"grad_norm": 0.4094283878803253,
|
| 365 |
+
"learning_rate": 2.7791416358595196e-05,
|
| 366 |
+
"loss": 0.2368,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.300369685767098,
|
| 371 |
+
"grad_norm": 4.650086402893066,
|
| 372 |
+
"learning_rate": 2.7748093807763402e-05,
|
| 373 |
+
"loss": 0.3215,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.3061460258780037,
|
| 378 |
+
"grad_norm": 18.08386993408203,
|
| 379 |
+
"learning_rate": 2.7704771256931612e-05,
|
| 380 |
+
"loss": 0.255,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.31192236598890943,
|
| 385 |
+
"grad_norm": 10.52841854095459,
|
| 386 |
+
"learning_rate": 2.7661448706099815e-05,
|
| 387 |
+
"loss": 0.2439,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.31769870609981515,
|
| 392 |
+
"grad_norm": 65.88616180419922,
|
| 393 |
+
"learning_rate": 2.761812615526802e-05,
|
| 394 |
+
"loss": 0.2418,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.3234750462107209,
|
| 399 |
+
"grad_norm": 9.300713539123535,
|
| 400 |
+
"learning_rate": 2.757480360443623e-05,
|
| 401 |
+
"loss": 0.3188,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.3292513863216266,
|
| 406 |
+
"grad_norm": 8.528387069702148,
|
| 407 |
+
"learning_rate": 2.7531481053604436e-05,
|
| 408 |
+
"loss": 0.2996,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.33502772643253237,
|
| 413 |
+
"grad_norm": 25.872819900512695,
|
| 414 |
+
"learning_rate": 2.7488158502772646e-05,
|
| 415 |
+
"loss": 0.2303,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3408040665434381,
|
| 420 |
+
"grad_norm": 32.80488204956055,
|
| 421 |
+
"learning_rate": 2.7444835951940852e-05,
|
| 422 |
+
"loss": 0.2472,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.3465804066543438,
|
| 427 |
+
"grad_norm": 3.634104013442993,
|
| 428 |
+
"learning_rate": 2.7401513401109058e-05,
|
| 429 |
+
"loss": 0.3054,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.35235674676524953,
|
| 434 |
+
"grad_norm": 9.497197151184082,
|
| 435 |
+
"learning_rate": 2.7358190850277264e-05,
|
| 436 |
+
"loss": 0.264,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.35813308687615525,
|
| 441 |
+
"grad_norm": 28.931177139282227,
|
| 442 |
+
"learning_rate": 2.731486829944547e-05,
|
| 443 |
+
"loss": 0.2419,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.363909426987061,
|
| 448 |
+
"grad_norm": 25.71267318725586,
|
| 449 |
+
"learning_rate": 2.727154574861368e-05,
|
| 450 |
+
"loss": 0.2137,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.36968576709796674,
|
| 455 |
+
"grad_norm": 18.439529418945312,
|
| 456 |
+
"learning_rate": 2.7228223197781886e-05,
|
| 457 |
+
"loss": 0.2562,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.37546210720887246,
|
| 462 |
+
"grad_norm": 2.7945632934570312,
|
| 463 |
+
"learning_rate": 2.7184900646950095e-05,
|
| 464 |
+
"loss": 0.3085,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3812384473197782,
|
| 469 |
+
"grad_norm": 20.88351058959961,
|
| 470 |
+
"learning_rate": 2.71415780961183e-05,
|
| 471 |
+
"loss": 0.1988,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3870147874306839,
|
| 476 |
+
"grad_norm": 9.544013977050781,
|
| 477 |
+
"learning_rate": 2.7098255545286504e-05,
|
| 478 |
+
"loss": 0.252,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.3927911275415896,
|
| 483 |
+
"grad_norm": 0.2528328001499176,
|
| 484 |
+
"learning_rate": 2.7054932994454714e-05,
|
| 485 |
+
"loss": 0.2136,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3985674676524954,
|
| 490 |
+
"grad_norm": 15.808114051818848,
|
| 491 |
+
"learning_rate": 2.701161044362292e-05,
|
| 492 |
+
"loss": 0.3502,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4043438077634011,
|
| 497 |
+
"grad_norm": 9.585843086242676,
|
| 498 |
+
"learning_rate": 2.696828789279113e-05,
|
| 499 |
+
"loss": 0.2437,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.41012014787430684,
|
| 504 |
+
"grad_norm": 0.7092911601066589,
|
| 505 |
+
"learning_rate": 2.6924965341959335e-05,
|
| 506 |
+
"loss": 0.2137,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.41589648798521256,
|
| 511 |
+
"grad_norm": 0.09910676628351212,
|
| 512 |
+
"learning_rate": 2.6881642791127545e-05,
|
| 513 |
+
"loss": 0.2979,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.4216728280961183,
|
| 518 |
+
"grad_norm": 0.674543559551239,
|
| 519 |
+
"learning_rate": 2.683832024029575e-05,
|
| 520 |
+
"loss": 0.2392,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.42744916820702406,
|
| 525 |
+
"grad_norm": 0.6626635193824768,
|
| 526 |
+
"learning_rate": 2.6794997689463953e-05,
|
| 527 |
+
"loss": 0.2019,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.4332255083179298,
|
| 532 |
+
"grad_norm": 9.206210136413574,
|
| 533 |
+
"learning_rate": 2.6751675138632163e-05,
|
| 534 |
+
"loss": 0.2747,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.4390018484288355,
|
| 539 |
+
"grad_norm": 1.017768144607544,
|
| 540 |
+
"learning_rate": 2.670835258780037e-05,
|
| 541 |
+
"loss": 0.1992,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.4447781885397412,
|
| 546 |
+
"grad_norm": 18.11737060546875,
|
| 547 |
+
"learning_rate": 2.666503003696858e-05,
|
| 548 |
+
"loss": 0.2201,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.45055452865064693,
|
| 553 |
+
"grad_norm": 15.56303882598877,
|
| 554 |
+
"learning_rate": 2.6621707486136785e-05,
|
| 555 |
+
"loss": 0.198,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.45633086876155265,
|
| 560 |
+
"grad_norm": 7.689689636230469,
|
| 561 |
+
"learning_rate": 2.657838493530499e-05,
|
| 562 |
+
"loss": 0.2894,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.46210720887245843,
|
| 567 |
+
"grad_norm": 9.518302917480469,
|
| 568 |
+
"learning_rate": 2.65350623844732e-05,
|
| 569 |
+
"loss": 0.2226,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.46788354898336415,
|
| 574 |
+
"grad_norm": 0.4917680025100708,
|
| 575 |
+
"learning_rate": 2.6491739833641403e-05,
|
| 576 |
+
"loss": 0.2602,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.47365988909426987,
|
| 581 |
+
"grad_norm": 34.92518997192383,
|
| 582 |
+
"learning_rate": 2.6448417282809612e-05,
|
| 583 |
+
"loss": 0.2506,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.4794362292051756,
|
| 588 |
+
"grad_norm": 14.343157768249512,
|
| 589 |
+
"learning_rate": 2.640509473197782e-05,
|
| 590 |
+
"loss": 0.3009,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4852125693160813,
|
| 595 |
+
"grad_norm": 14.651656150817871,
|
| 596 |
+
"learning_rate": 2.6361772181146028e-05,
|
| 597 |
+
"loss": 0.2419,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4909889094269871,
|
| 602 |
+
"grad_norm": 0.7227654457092285,
|
| 603 |
+
"learning_rate": 2.6318449630314234e-05,
|
| 604 |
+
"loss": 0.3091,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4967652495378928,
|
| 609 |
+
"grad_norm": 10.362215042114258,
|
| 610 |
+
"learning_rate": 2.627512707948244e-05,
|
| 611 |
+
"loss": 0.1949,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.5025415896487985,
|
| 616 |
+
"grad_norm": 11.842758178710938,
|
| 617 |
+
"learning_rate": 2.623180452865065e-05,
|
| 618 |
+
"loss": 0.3337,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.5083179297597042,
|
| 623 |
+
"grad_norm": 11.550918579101562,
|
| 624 |
+
"learning_rate": 2.6188481977818852e-05,
|
| 625 |
+
"loss": 0.2641,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.51409426987061,
|
| 630 |
+
"grad_norm": 5.512154579162598,
|
| 631 |
+
"learning_rate": 2.6145159426987062e-05,
|
| 632 |
+
"loss": 0.2541,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5198706099815157,
|
| 637 |
+
"grad_norm": 10.161574363708496,
|
| 638 |
+
"learning_rate": 2.6101836876155268e-05,
|
| 639 |
+
"loss": 0.256,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.5256469500924215,
|
| 644 |
+
"grad_norm": 1.659515380859375,
|
| 645 |
+
"learning_rate": 2.6058514325323477e-05,
|
| 646 |
+
"loss": 0.2621,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.5314232902033271,
|
| 651 |
+
"grad_norm": 2.5476644039154053,
|
| 652 |
+
"learning_rate": 2.6015191774491684e-05,
|
| 653 |
+
"loss": 0.294,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5371996303142329,
|
| 658 |
+
"grad_norm": 0.9729189276695251,
|
| 659 |
+
"learning_rate": 2.597186922365989e-05,
|
| 660 |
+
"loss": 0.1333,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.5429759704251387,
|
| 665 |
+
"grad_norm": 3.8201911449432373,
|
| 666 |
+
"learning_rate": 2.59285466728281e-05,
|
| 667 |
+
"loss": 0.2598,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.5487523105360443,
|
| 672 |
+
"grad_norm": 0.17872726917266846,
|
| 673 |
+
"learning_rate": 2.5885224121996302e-05,
|
| 674 |
+
"loss": 0.1897,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.5545286506469501,
|
| 679 |
+
"grad_norm": 24.479658126831055,
|
| 680 |
+
"learning_rate": 2.584190157116451e-05,
|
| 681 |
+
"loss": 0.2606,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.5603049907578558,
|
| 686 |
+
"grad_norm": 13.361931800842285,
|
| 687 |
+
"learning_rate": 2.5798579020332717e-05,
|
| 688 |
+
"loss": 0.271,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5660813308687616,
|
| 693 |
+
"grad_norm": 8.625031471252441,
|
| 694 |
+
"learning_rate": 2.5755256469500924e-05,
|
| 695 |
+
"loss": 0.2056,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.5718576709796673,
|
| 700 |
+
"grad_norm": 11.34391975402832,
|
| 701 |
+
"learning_rate": 2.5711933918669133e-05,
|
| 702 |
+
"loss": 0.2736,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.577634011090573,
|
| 707 |
+
"grad_norm": 16.9676456451416,
|
| 708 |
+
"learning_rate": 2.566861136783734e-05,
|
| 709 |
+
"loss": 0.2821,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5834103512014788,
|
| 714 |
+
"grad_norm": 9.531896591186523,
|
| 715 |
+
"learning_rate": 2.5625288817005545e-05,
|
| 716 |
+
"loss": 0.2264,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5891866913123844,
|
| 721 |
+
"grad_norm": 2.4691050052642822,
|
| 722 |
+
"learning_rate": 2.558196626617375e-05,
|
| 723 |
+
"loss": 0.2148,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5949630314232902,
|
| 728 |
+
"grad_norm": 0.476725697517395,
|
| 729 |
+
"learning_rate": 2.553864371534196e-05,
|
| 730 |
+
"loss": 0.1964,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.600739371534196,
|
| 735 |
+
"grad_norm": 14.917049407958984,
|
| 736 |
+
"learning_rate": 2.5495321164510167e-05,
|
| 737 |
+
"loss": 0.2717,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.6065157116451017,
|
| 742 |
+
"grad_norm": 0.13745556771755219,
|
| 743 |
+
"learning_rate": 2.5451998613678373e-05,
|
| 744 |
+
"loss": 0.2086,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.6122920517560074,
|
| 749 |
+
"grad_norm": 22.109750747680664,
|
| 750 |
+
"learning_rate": 2.5408676062846582e-05,
|
| 751 |
+
"loss": 0.3401,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.6180683918669131,
|
| 756 |
+
"grad_norm": 0.4055859446525574,
|
| 757 |
+
"learning_rate": 2.536535351201479e-05,
|
| 758 |
+
"loss": 0.2572,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6238447319778189,
|
| 763 |
+
"grad_norm": 0.7329151034355164,
|
| 764 |
+
"learning_rate": 2.5322030961182995e-05,
|
| 765 |
+
"loss": 0.2246,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.6296210720887245,
|
| 770 |
+
"grad_norm": 0.6793464422225952,
|
| 771 |
+
"learning_rate": 2.52787084103512e-05,
|
| 772 |
+
"loss": 0.1804,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.6353974121996303,
|
| 777 |
+
"grad_norm": 6.19757604598999,
|
| 778 |
+
"learning_rate": 2.523538585951941e-05,
|
| 779 |
+
"loss": 0.2331,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.6411737523105361,
|
| 784 |
+
"grad_norm": 69.8536148071289,
|
| 785 |
+
"learning_rate": 2.5192063308687616e-05,
|
| 786 |
+
"loss": 0.2071,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.6469500924214417,
|
| 791 |
+
"grad_norm": 49.18756103515625,
|
| 792 |
+
"learning_rate": 2.5148740757855822e-05,
|
| 793 |
+
"loss": 0.2602,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.6527264325323475,
|
| 798 |
+
"grad_norm": 9.139322280883789,
|
| 799 |
+
"learning_rate": 2.5105418207024032e-05,
|
| 800 |
+
"loss": 0.1975,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.6585027726432532,
|
| 805 |
+
"grad_norm": 11.072602272033691,
|
| 806 |
+
"learning_rate": 2.5062095656192238e-05,
|
| 807 |
+
"loss": 0.1829,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.664279112754159,
|
| 812 |
+
"grad_norm": 0.06665151566267014,
|
| 813 |
+
"learning_rate": 2.5018773105360444e-05,
|
| 814 |
+
"loss": 0.2722,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.6700554528650647,
|
| 819 |
+
"grad_norm": 1.071594476699829,
|
| 820 |
+
"learning_rate": 2.497545055452865e-05,
|
| 821 |
+
"loss": 0.1957,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.6758317929759704,
|
| 826 |
+
"grad_norm": 3.1452102661132812,
|
| 827 |
+
"learning_rate": 2.4932128003696856e-05,
|
| 828 |
+
"loss": 0.1312,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.6816081330868762,
|
| 833 |
+
"grad_norm": 2.483550786972046,
|
| 834 |
+
"learning_rate": 2.4888805452865066e-05,
|
| 835 |
+
"loss": 0.2375,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.6873844731977818,
|
| 840 |
+
"grad_norm": 24.874065399169922,
|
| 841 |
+
"learning_rate": 2.4845482902033272e-05,
|
| 842 |
+
"loss": 0.2743,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.6931608133086876,
|
| 847 |
+
"grad_norm": 31.57330322265625,
|
| 848 |
+
"learning_rate": 2.480216035120148e-05,
|
| 849 |
+
"loss": 0.2298,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.6989371534195934,
|
| 854 |
+
"grad_norm": 21.310131072998047,
|
| 855 |
+
"learning_rate": 2.4758837800369687e-05,
|
| 856 |
+
"loss": 0.2634,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.7047134935304991,
|
| 861 |
+
"grad_norm": 10.233534812927246,
|
| 862 |
+
"learning_rate": 2.4715515249537894e-05,
|
| 863 |
+
"loss": 0.2778,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7104898336414048,
|
| 868 |
+
"grad_norm": 0.34112706780433655,
|
| 869 |
+
"learning_rate": 2.46721926987061e-05,
|
| 870 |
+
"loss": 0.2009,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.7162661737523105,
|
| 875 |
+
"grad_norm": 0.2525830864906311,
|
| 876 |
+
"learning_rate": 2.4628870147874306e-05,
|
| 877 |
+
"loss": 0.2068,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.7220425138632163,
|
| 882 |
+
"grad_norm": 16.547571182250977,
|
| 883 |
+
"learning_rate": 2.4585547597042515e-05,
|
| 884 |
+
"loss": 0.2343,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.727818853974122,
|
| 889 |
+
"grad_norm": 0.7130794525146484,
|
| 890 |
+
"learning_rate": 2.454222504621072e-05,
|
| 891 |
+
"loss": 0.2312,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.7335951940850277,
|
| 896 |
+
"grad_norm": 0.8044223785400391,
|
| 897 |
+
"learning_rate": 2.449890249537893e-05,
|
| 898 |
+
"loss": 0.1719,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.7393715341959335,
|
| 903 |
+
"grad_norm": 13.686079978942871,
|
| 904 |
+
"learning_rate": 2.4455579944547134e-05,
|
| 905 |
+
"loss": 0.2398,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.7451478743068392,
|
| 910 |
+
"grad_norm": 1.6030184030532837,
|
| 911 |
+
"learning_rate": 2.4412257393715343e-05,
|
| 912 |
+
"loss": 0.1675,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7509242144177449,
|
| 917 |
+
"grad_norm": 2.320791482925415,
|
| 918 |
+
"learning_rate": 2.436893484288355e-05,
|
| 919 |
+
"loss": 0.2505,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.7567005545286506,
|
| 924 |
+
"grad_norm": 12.586448669433594,
|
| 925 |
+
"learning_rate": 2.4325612292051755e-05,
|
| 926 |
+
"loss": 0.2716,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.7624768946395564,
|
| 931 |
+
"grad_norm": 7.049114227294922,
|
| 932 |
+
"learning_rate": 2.4282289741219965e-05,
|
| 933 |
+
"loss": 0.169,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.7682532347504621,
|
| 938 |
+
"grad_norm": 9.050933837890625,
|
| 939 |
+
"learning_rate": 2.423896719038817e-05,
|
| 940 |
+
"loss": 0.2116,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.7740295748613678,
|
| 945 |
+
"grad_norm": 25.830007553100586,
|
| 946 |
+
"learning_rate": 2.419564463955638e-05,
|
| 947 |
+
"loss": 0.2181,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.7798059149722736,
|
| 952 |
+
"grad_norm": 0.14128220081329346,
|
| 953 |
+
"learning_rate": 2.4152322088724583e-05,
|
| 954 |
+
"loss": 0.2531,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.7855822550831792,
|
| 959 |
+
"grad_norm": 0.03034038469195366,
|
| 960 |
+
"learning_rate": 2.4108999537892792e-05,
|
| 961 |
+
"loss": 0.2109,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.791358595194085,
|
| 966 |
+
"grad_norm": 0.1031632274389267,
|
| 967 |
+
"learning_rate": 2.4065676987061e-05,
|
| 968 |
+
"loss": 0.2155,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7971349353049908,
|
| 973 |
+
"grad_norm": 42.54078674316406,
|
| 974 |
+
"learning_rate": 2.4022354436229205e-05,
|
| 975 |
+
"loss": 0.183,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.8029112754158965,
|
| 980 |
+
"grad_norm": 0.03734962269663811,
|
| 981 |
+
"learning_rate": 2.3979031885397414e-05,
|
| 982 |
+
"loss": 0.185,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.8086876155268022,
|
| 987 |
+
"grad_norm": 48.300174713134766,
|
| 988 |
+
"learning_rate": 2.393570933456562e-05,
|
| 989 |
+
"loss": 0.1803,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.8144639556377079,
|
| 994 |
+
"grad_norm": 13.418463706970215,
|
| 995 |
+
"learning_rate": 2.389238678373383e-05,
|
| 996 |
+
"loss": 0.2226,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.8202402957486137,
|
| 1001 |
+
"grad_norm": 16.923093795776367,
|
| 1002 |
+
"learning_rate": 2.3849064232902032e-05,
|
| 1003 |
+
"loss": 0.2387,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.8260166358595195,
|
| 1008 |
+
"grad_norm": 37.3891487121582,
|
| 1009 |
+
"learning_rate": 2.380574168207024e-05,
|
| 1010 |
+
"loss": 0.2168,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8317929759704251,
|
| 1015 |
+
"grad_norm": 24.676448822021484,
|
| 1016 |
+
"learning_rate": 2.3762419131238448e-05,
|
| 1017 |
+
"loss": 0.202,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.8375693160813309,
|
| 1022 |
+
"grad_norm": 13.302814483642578,
|
| 1023 |
+
"learning_rate": 2.3719096580406654e-05,
|
| 1024 |
+
"loss": 0.2153,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8433456561922366,
|
| 1029 |
+
"grad_norm": 8.611613273620605,
|
| 1030 |
+
"learning_rate": 2.3675774029574864e-05,
|
| 1031 |
+
"loss": 0.2544,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.8491219963031423,
|
| 1036 |
+
"grad_norm": 0.4762769639492035,
|
| 1037 |
+
"learning_rate": 2.363245147874307e-05,
|
| 1038 |
+
"loss": 0.2129,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.8548983364140481,
|
| 1043 |
+
"grad_norm": 16.49268341064453,
|
| 1044 |
+
"learning_rate": 2.358912892791128e-05,
|
| 1045 |
+
"loss": 0.177,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.8606746765249538,
|
| 1050 |
+
"grad_norm": 2.05298113822937,
|
| 1051 |
+
"learning_rate": 2.3545806377079482e-05,
|
| 1052 |
+
"loss": 0.2274,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.8664510166358595,
|
| 1057 |
+
"grad_norm": 19.834819793701172,
|
| 1058 |
+
"learning_rate": 2.3502483826247688e-05,
|
| 1059 |
+
"loss": 0.1998,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.8722273567467652,
|
| 1064 |
+
"grad_norm": 9.706733703613281,
|
| 1065 |
+
"learning_rate": 2.3459161275415897e-05,
|
| 1066 |
+
"loss": 0.2563,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.878003696857671,
|
| 1071 |
+
"grad_norm": 9.801465034484863,
|
| 1072 |
+
"learning_rate": 2.3415838724584104e-05,
|
| 1073 |
+
"loss": 0.1453,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.8837800369685767,
|
| 1078 |
+
"grad_norm": 11.452254295349121,
|
| 1079 |
+
"learning_rate": 2.3372516173752313e-05,
|
| 1080 |
+
"loss": 0.272,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.8895563770794824,
|
| 1085 |
+
"grad_norm": 7.050422191619873,
|
| 1086 |
+
"learning_rate": 2.332919362292052e-05,
|
| 1087 |
+
"loss": 0.1713,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.8953327171903882,
|
| 1092 |
+
"grad_norm": 46.33341979980469,
|
| 1093 |
+
"learning_rate": 2.3285871072088725e-05,
|
| 1094 |
+
"loss": 0.1967,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.9011090573012939,
|
| 1099 |
+
"grad_norm": 1.1651716232299805,
|
| 1100 |
+
"learning_rate": 2.324254852125693e-05,
|
| 1101 |
+
"loss": 0.187,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.9068853974121996,
|
| 1106 |
+
"grad_norm": 0.3762068450450897,
|
| 1107 |
+
"learning_rate": 2.3199225970425137e-05,
|
| 1108 |
+
"loss": 0.2556,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.9126617375231053,
|
| 1113 |
+
"grad_norm": 9.550520896911621,
|
| 1114 |
+
"learning_rate": 2.3155903419593347e-05,
|
| 1115 |
+
"loss": 0.2413,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.9184380776340111,
|
| 1120 |
+
"grad_norm": 0.3446040451526642,
|
| 1121 |
+
"learning_rate": 2.3112580868761553e-05,
|
| 1122 |
+
"loss": 0.1694,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.9242144177449169,
|
| 1127 |
+
"grad_norm": 0.7896172404289246,
|
| 1128 |
+
"learning_rate": 2.3069258317929762e-05,
|
| 1129 |
+
"loss": 0.2504,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.9299907578558225,
|
| 1134 |
+
"grad_norm": 34.30472946166992,
|
| 1135 |
+
"learning_rate": 2.302593576709797e-05,
|
| 1136 |
+
"loss": 0.2462,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9357670979667283,
|
| 1141 |
+
"grad_norm": 31.708581924438477,
|
| 1142 |
+
"learning_rate": 2.298261321626617e-05,
|
| 1143 |
+
"loss": 0.147,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.941543438077634,
|
| 1148 |
+
"grad_norm": 2.5662238597869873,
|
| 1149 |
+
"learning_rate": 2.293929066543438e-05,
|
| 1150 |
+
"loss": 0.262,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.9473197781885397,
|
| 1155 |
+
"grad_norm": 0.19891509413719177,
|
| 1156 |
+
"learning_rate": 2.2895968114602587e-05,
|
| 1157 |
+
"loss": 0.1978,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.9530961182994455,
|
| 1162 |
+
"grad_norm": 81.1672134399414,
|
| 1163 |
+
"learning_rate": 2.2852645563770796e-05,
|
| 1164 |
+
"loss": 0.207,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.9588724584103512,
|
| 1169 |
+
"grad_norm": 20.577178955078125,
|
| 1170 |
+
"learning_rate": 2.2809323012939002e-05,
|
| 1171 |
+
"loss": 0.2256,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.964648798521257,
|
| 1176 |
+
"grad_norm": 0.11451568454504013,
|
| 1177 |
+
"learning_rate": 2.2766000462107212e-05,
|
| 1178 |
+
"loss": 0.2123,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.9704251386321626,
|
| 1183 |
+
"grad_norm": 2.9672398567199707,
|
| 1184 |
+
"learning_rate": 2.2722677911275418e-05,
|
| 1185 |
+
"loss": 0.2419,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.9762014787430684,
|
| 1190 |
+
"grad_norm": 3.283834934234619,
|
| 1191 |
+
"learning_rate": 2.267935536044362e-05,
|
| 1192 |
+
"loss": 0.1796,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9819778188539742,
|
| 1197 |
+
"grad_norm": 19.114147186279297,
|
| 1198 |
+
"learning_rate": 2.263603280961183e-05,
|
| 1199 |
+
"loss": 0.1789,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.9877541589648798,
|
| 1204 |
+
"grad_norm": 0.5701966881752014,
|
| 1205 |
+
"learning_rate": 2.2592710258780036e-05,
|
| 1206 |
+
"loss": 0.1616,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.9935304990757856,
|
| 1211 |
+
"grad_norm": 14.517534255981445,
|
| 1212 |
+
"learning_rate": 2.2549387707948246e-05,
|
| 1213 |
+
"loss": 0.1909,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.9993068391866913,
|
| 1218 |
+
"grad_norm": 0.1315908581018448,
|
| 1219 |
+
"learning_rate": 2.2506065157116452e-05,
|
| 1220 |
+
"loss": 0.2188,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.0,
|
| 1225 |
+
"eval_accuracy": 0.9437296946068876,
|
| 1226 |
+
"eval_f1": 0.9665611244111515,
|
| 1227 |
+
"eval_loss": 0.2036169469356537,
|
| 1228 |
+
"eval_precision": 0.953382084095064,
|
| 1229 |
+
"eval_recall": 0.9801096319498825,
|
| 1230 |
+
"eval_runtime": 88.7618,
|
| 1231 |
+
"eval_samples_per_second": 86.693,
|
| 1232 |
+
"eval_steps_per_second": 10.838,
|
| 1233 |
+
"step": 8656
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.005083179297597,
|
| 1237 |
+
"grad_norm": 9.673088073730469,
|
| 1238 |
+
"learning_rate": 2.246274260628466e-05,
|
| 1239 |
+
"loss": 0.1254,
|
| 1240 |
+
"step": 8700
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.0108595194085028,
|
| 1244 |
+
"grad_norm": 0.6131724119186401,
|
| 1245 |
+
"learning_rate": 2.2419420055452867e-05,
|
| 1246 |
+
"loss": 0.1205,
|
| 1247 |
+
"step": 8750
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.0166358595194085,
|
| 1251 |
+
"grad_norm": 0.12616665661334991,
|
| 1252 |
+
"learning_rate": 2.237609750462107e-05,
|
| 1253 |
+
"loss": 0.1409,
|
| 1254 |
+
"step": 8800
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.0224121996303142,
|
| 1258 |
+
"grad_norm": 0.18657980859279633,
|
| 1259 |
+
"learning_rate": 2.233277495378928e-05,
|
| 1260 |
+
"loss": 0.161,
|
| 1261 |
+
"step": 8850
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.02818853974122,
|
| 1265 |
+
"grad_norm": 0.051760233938694,
|
| 1266 |
+
"learning_rate": 2.2289452402957486e-05,
|
| 1267 |
+
"loss": 0.0927,
|
| 1268 |
+
"step": 8900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.0339648798521257,
|
| 1272 |
+
"grad_norm": 0.5276439785957336,
|
| 1273 |
+
"learning_rate": 2.2246129852125695e-05,
|
| 1274 |
+
"loss": 0.2098,
|
| 1275 |
+
"step": 8950
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.0397412199630314,
|
| 1279 |
+
"grad_norm": 3.536935567855835,
|
| 1280 |
+
"learning_rate": 2.22028073012939e-05,
|
| 1281 |
+
"loss": 0.164,
|
| 1282 |
+
"step": 9000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.0455175600739373,
|
| 1286 |
+
"grad_norm": 0.1915968954563141,
|
| 1287 |
+
"learning_rate": 2.2159484750462107e-05,
|
| 1288 |
+
"loss": 0.1839,
|
| 1289 |
+
"step": 9050
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.051293900184843,
|
| 1293 |
+
"grad_norm": 6.418234348297119,
|
| 1294 |
+
"learning_rate": 2.2116162199630314e-05,
|
| 1295 |
+
"loss": 0.1271,
|
| 1296 |
+
"step": 9100
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.0570702402957486,
|
| 1300 |
+
"grad_norm": 0.7983689904212952,
|
| 1301 |
+
"learning_rate": 2.207283964879852e-05,
|
| 1302 |
+
"loss": 0.1445,
|
| 1303 |
+
"step": 9150
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.0628465804066543,
|
| 1307 |
+
"grad_norm": 7.85959529876709,
|
| 1308 |
+
"learning_rate": 2.202951709796673e-05,
|
| 1309 |
+
"loss": 0.1613,
|
| 1310 |
+
"step": 9200
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.0686229205175601,
|
| 1314 |
+
"grad_norm": 0.041306737810373306,
|
| 1315 |
+
"learning_rate": 2.1986194547134935e-05,
|
| 1316 |
+
"loss": 0.1491,
|
| 1317 |
+
"step": 9250
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.0743992606284658,
|
| 1321 |
+
"grad_norm": 0.12126684933900833,
|
| 1322 |
+
"learning_rate": 2.1942871996303145e-05,
|
| 1323 |
+
"loss": 0.162,
|
| 1324 |
+
"step": 9300
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.0801756007393715,
|
| 1328 |
+
"grad_norm": 15.810729026794434,
|
| 1329 |
+
"learning_rate": 2.189954944547135e-05,
|
| 1330 |
+
"loss": 0.1271,
|
| 1331 |
+
"step": 9350
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.0859519408502774,
|
| 1335 |
+
"grad_norm": 0.04351761192083359,
|
| 1336 |
+
"learning_rate": 2.1856226894639557e-05,
|
| 1337 |
+
"loss": 0.1499,
|
| 1338 |
+
"step": 9400
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.091728280961183,
|
| 1342 |
+
"grad_norm": 49.74297332763672,
|
| 1343 |
+
"learning_rate": 2.1812904343807763e-05,
|
| 1344 |
+
"loss": 0.1617,
|
| 1345 |
+
"step": 9450
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0975046210720887,
|
| 1349 |
+
"grad_norm": 33.17013931274414,
|
| 1350 |
+
"learning_rate": 2.176958179297597e-05,
|
| 1351 |
+
"loss": 0.2043,
|
| 1352 |
+
"step": 9500
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.1032809611829943,
|
| 1356 |
+
"grad_norm": 1.7674719095230103,
|
| 1357 |
+
"learning_rate": 2.172625924214418e-05,
|
| 1358 |
+
"loss": 0.1275,
|
| 1359 |
+
"step": 9550
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.1090573012939002,
|
| 1363 |
+
"grad_norm": 0.13075052201747894,
|
| 1364 |
+
"learning_rate": 2.1682936691312385e-05,
|
| 1365 |
+
"loss": 0.117,
|
| 1366 |
+
"step": 9600
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.114833641404806,
|
| 1370 |
+
"grad_norm": 0.02289584092795849,
|
| 1371 |
+
"learning_rate": 2.1639614140480594e-05,
|
| 1372 |
+
"loss": 0.1096,
|
| 1373 |
+
"step": 9650
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.1206099815157116,
|
| 1377 |
+
"grad_norm": 7.883808612823486,
|
| 1378 |
+
"learning_rate": 2.15962915896488e-05,
|
| 1379 |
+
"loss": 0.1948,
|
| 1380 |
+
"step": 9700
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.1263863216266174,
|
| 1384 |
+
"grad_norm": 1.676134705543518,
|
| 1385 |
+
"learning_rate": 2.1552969038817006e-05,
|
| 1386 |
+
"loss": 0.155,
|
| 1387 |
+
"step": 9750
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.1321626617375231,
|
| 1391 |
+
"grad_norm": 0.06022998318076134,
|
| 1392 |
+
"learning_rate": 2.1509646487985212e-05,
|
| 1393 |
+
"loss": 0.1199,
|
| 1394 |
+
"step": 9800
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.1379390018484288,
|
| 1398 |
+
"grad_norm": 19.327392578125,
|
| 1399 |
+
"learning_rate": 2.146632393715342e-05,
|
| 1400 |
+
"loss": 0.21,
|
| 1401 |
+
"step": 9850
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1437153419593344,
|
| 1405 |
+
"grad_norm": 133.89877319335938,
|
| 1406 |
+
"learning_rate": 2.1423001386321628e-05,
|
| 1407 |
+
"loss": 0.1387,
|
| 1408 |
+
"step": 9900
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.1494916820702403,
|
| 1412 |
+
"grad_norm": 0.6451585292816162,
|
| 1413 |
+
"learning_rate": 2.1379678835489834e-05,
|
| 1414 |
+
"loss": 0.163,
|
| 1415 |
+
"step": 9950
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.155268022181146,
|
| 1419 |
+
"grad_norm": 0.17913977801799774,
|
| 1420 |
+
"learning_rate": 2.133635628465804e-05,
|
| 1421 |
+
"loss": 0.1653,
|
| 1422 |
+
"step": 10000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.1610443622920519,
|
| 1426 |
+
"grad_norm": 0.09651433676481247,
|
| 1427 |
+
"learning_rate": 2.129303373382625e-05,
|
| 1428 |
+
"loss": 0.1462,
|
| 1429 |
+
"step": 10050
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.1668207024029575,
|
| 1433 |
+
"grad_norm": 0.5200825929641724,
|
| 1434 |
+
"learning_rate": 2.1249711182994456e-05,
|
| 1435 |
+
"loss": 0.2055,
|
| 1436 |
+
"step": 10100
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.1725970425138632,
|
| 1440 |
+
"grad_norm": 0.2577333152294159,
|
| 1441 |
+
"learning_rate": 2.1206388632162662e-05,
|
| 1442 |
+
"loss": 0.1792,
|
| 1443 |
+
"step": 10150
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.1783733826247689,
|
| 1447 |
+
"grad_norm": 178.7727813720703,
|
| 1448 |
+
"learning_rate": 2.1163066081330868e-05,
|
| 1449 |
+
"loss": 0.0912,
|
| 1450 |
+
"step": 10200
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.1841497227356748,
|
| 1454 |
+
"grad_norm": 0.16386118531227112,
|
| 1455 |
+
"learning_rate": 2.1119743530499077e-05,
|
| 1456 |
+
"loss": 0.1288,
|
| 1457 |
+
"step": 10250
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.1899260628465804,
|
| 1461 |
+
"grad_norm": 5.637333869934082,
|
| 1462 |
+
"learning_rate": 2.1076420979667284e-05,
|
| 1463 |
+
"loss": 0.1978,
|
| 1464 |
+
"step": 10300
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.195702402957486,
|
| 1468 |
+
"grad_norm": 0.031639471650123596,
|
| 1469 |
+
"learning_rate": 2.103309842883549e-05,
|
| 1470 |
+
"loss": 0.1579,
|
| 1471 |
+
"step": 10350
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.201478743068392,
|
| 1475 |
+
"grad_norm": 7.582589626312256,
|
| 1476 |
+
"learning_rate": 2.09897758780037e-05,
|
| 1477 |
+
"loss": 0.1685,
|
| 1478 |
+
"step": 10400
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.2072550831792976,
|
| 1482 |
+
"grad_norm": 0.21556666493415833,
|
| 1483 |
+
"learning_rate": 2.0946453327171902e-05,
|
| 1484 |
+
"loss": 0.1794,
|
| 1485 |
+
"step": 10450
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.2130314232902033,
|
| 1489 |
+
"grad_norm": 11.476829528808594,
|
| 1490 |
+
"learning_rate": 2.090313077634011e-05,
|
| 1491 |
+
"loss": 0.1625,
|
| 1492 |
+
"step": 10500
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.218807763401109,
|
| 1496 |
+
"grad_norm": 24.46149253845215,
|
| 1497 |
+
"learning_rate": 2.0859808225508317e-05,
|
| 1498 |
+
"loss": 0.1383,
|
| 1499 |
+
"step": 10550
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.2245841035120149,
|
| 1503 |
+
"grad_norm": 26.288007736206055,
|
| 1504 |
+
"learning_rate": 2.0816485674676527e-05,
|
| 1505 |
+
"loss": 0.113,
|
| 1506 |
+
"step": 10600
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.2303604436229205,
|
| 1510 |
+
"grad_norm": 0.622348964214325,
|
| 1511 |
+
"learning_rate": 2.0773163123844733e-05,
|
| 1512 |
+
"loss": 0.17,
|
| 1513 |
+
"step": 10650
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2361367837338262,
|
| 1517 |
+
"grad_norm": 16.938920974731445,
|
| 1518 |
+
"learning_rate": 2.072984057301294e-05,
|
| 1519 |
+
"loss": 0.1494,
|
| 1520 |
+
"step": 10700
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.241913123844732,
|
| 1524 |
+
"grad_norm": 1.0484834909439087,
|
| 1525 |
+
"learning_rate": 2.068651802218115e-05,
|
| 1526 |
+
"loss": 0.1821,
|
| 1527 |
+
"step": 10750
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.2476894639556377,
|
| 1531 |
+
"grad_norm": 0.10622230172157288,
|
| 1532 |
+
"learning_rate": 2.064319547134935e-05,
|
| 1533 |
+
"loss": 0.1872,
|
| 1534 |
+
"step": 10800
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.2534658040665434,
|
| 1538 |
+
"grad_norm": 4.886632919311523,
|
| 1539 |
+
"learning_rate": 2.059987292051756e-05,
|
| 1540 |
+
"loss": 0.1208,
|
| 1541 |
+
"step": 10850
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.259242144177449,
|
| 1545 |
+
"grad_norm": 0.5296549201011658,
|
| 1546 |
+
"learning_rate": 2.0556550369685767e-05,
|
| 1547 |
+
"loss": 0.1579,
|
| 1548 |
+
"step": 10900
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.265018484288355,
|
| 1552 |
+
"grad_norm": 0.04283996298909187,
|
| 1553 |
+
"learning_rate": 2.0513227818853973e-05,
|
| 1554 |
+
"loss": 0.083,
|
| 1555 |
+
"step": 10950
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2707948243992606,
|
| 1559 |
+
"grad_norm": 1.309367060661316,
|
| 1560 |
+
"learning_rate": 2.0469905268022182e-05,
|
| 1561 |
+
"loss": 0.0815,
|
| 1562 |
+
"step": 11000
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2765711645101663,
|
| 1566 |
+
"grad_norm": 0.03423392027616501,
|
| 1567 |
+
"learning_rate": 2.042658271719039e-05,
|
| 1568 |
+
"loss": 0.0932,
|
| 1569 |
+
"step": 11050
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2823475046210722,
|
| 1573 |
+
"grad_norm": 0.028841083869338036,
|
| 1574 |
+
"learning_rate": 2.0383260166358598e-05,
|
| 1575 |
+
"loss": 0.1432,
|
| 1576 |
+
"step": 11100
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2881238447319778,
|
| 1580 |
+
"grad_norm": 0.0708237886428833,
|
| 1581 |
+
"learning_rate": 2.03399376155268e-05,
|
| 1582 |
+
"loss": 0.1607,
|
| 1583 |
+
"step": 11150
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2939001848428835,
|
| 1587 |
+
"grad_norm": 0.02011285535991192,
|
| 1588 |
+
"learning_rate": 2.029661506469501e-05,
|
| 1589 |
+
"loss": 0.2166,
|
| 1590 |
+
"step": 11200
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2996765249537892,
|
| 1594 |
+
"grad_norm": 0.16425344347953796,
|
| 1595 |
+
"learning_rate": 2.0253292513863216e-05,
|
| 1596 |
+
"loss": 0.1682,
|
| 1597 |
+
"step": 11250
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.305452865064695,
|
| 1601 |
+
"grad_norm": 10.606422424316406,
|
| 1602 |
+
"learning_rate": 2.0209969963031422e-05,
|
| 1603 |
+
"loss": 0.1247,
|
| 1604 |
+
"step": 11300
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.3112292051756007,
|
| 1608 |
+
"grad_norm": 0.39115816354751587,
|
| 1609 |
+
"learning_rate": 2.0166647412199632e-05,
|
| 1610 |
+
"loss": 0.1844,
|
| 1611 |
+
"step": 11350
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.3170055452865066,
|
| 1615 |
+
"grad_norm": 0.05713627114892006,
|
| 1616 |
+
"learning_rate": 2.0123324861367838e-05,
|
| 1617 |
+
"loss": 0.0753,
|
| 1618 |
+
"step": 11400
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.3227818853974123,
|
| 1622 |
+
"grad_norm": 57.468997955322266,
|
| 1623 |
+
"learning_rate": 2.0080002310536047e-05,
|
| 1624 |
+
"loss": 0.0336,
|
| 1625 |
+
"step": 11450
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.328558225508318,
|
| 1629 |
+
"grad_norm": 0.23353737592697144,
|
| 1630 |
+
"learning_rate": 2.003667975970425e-05,
|
| 1631 |
+
"loss": 0.2012,
|
| 1632 |
+
"step": 11500
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.3343345656192236,
|
| 1636 |
+
"grad_norm": 0.018346797674894333,
|
| 1637 |
+
"learning_rate": 1.999335720887246e-05,
|
| 1638 |
+
"loss": 0.1375,
|
| 1639 |
+
"step": 11550
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.3401109057301293,
|
| 1643 |
+
"grad_norm": 0.1295866072177887,
|
| 1644 |
+
"learning_rate": 1.9950034658040666e-05,
|
| 1645 |
+
"loss": 0.1389,
|
| 1646 |
+
"step": 11600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.3458872458410351,
|
| 1650 |
+
"grad_norm": 0.29036056995391846,
|
| 1651 |
+
"learning_rate": 1.9906712107208872e-05,
|
| 1652 |
+
"loss": 0.0943,
|
| 1653 |
+
"step": 11650
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.3516635859519408,
|
| 1657 |
+
"grad_norm": 0.007298491895198822,
|
| 1658 |
+
"learning_rate": 1.986338955637708e-05,
|
| 1659 |
+
"loss": 0.1572,
|
| 1660 |
+
"step": 11700
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.3574399260628467,
|
| 1664 |
+
"grad_norm": 0.39678987860679626,
|
| 1665 |
+
"learning_rate": 1.9820067005545287e-05,
|
| 1666 |
+
"loss": 0.1351,
|
| 1667 |
+
"step": 11750
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.3632162661737524,
|
| 1671 |
+
"grad_norm": 135.63607788085938,
|
| 1672 |
+
"learning_rate": 1.9776744454713497e-05,
|
| 1673 |
+
"loss": 0.1746,
|
| 1674 |
+
"step": 11800
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.368992606284658,
|
| 1678 |
+
"grad_norm": 12.449654579162598,
|
| 1679 |
+
"learning_rate": 1.97334219038817e-05,
|
| 1680 |
+
"loss": 0.1661,
|
| 1681 |
+
"step": 11850
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3747689463955637,
|
| 1685 |
+
"grad_norm": 1.2668341398239136,
|
| 1686 |
+
"learning_rate": 1.9690099353049906e-05,
|
| 1687 |
+
"loss": 0.1487,
|
| 1688 |
+
"step": 11900
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.3805452865064696,
|
| 1692 |
+
"grad_norm": 0.26599591970443726,
|
| 1693 |
+
"learning_rate": 1.9646776802218115e-05,
|
| 1694 |
+
"loss": 0.0866,
|
| 1695 |
+
"step": 11950
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.3863216266173752,
|
| 1699 |
+
"grad_norm": 5.264702796936035,
|
| 1700 |
+
"learning_rate": 1.960345425138632e-05,
|
| 1701 |
+
"loss": 0.2262,
|
| 1702 |
+
"step": 12000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.392097966728281,
|
| 1706 |
+
"grad_norm": 45.19207763671875,
|
| 1707 |
+
"learning_rate": 1.956013170055453e-05,
|
| 1708 |
+
"loss": 0.1587,
|
| 1709 |
+
"step": 12050
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 1.3978743068391868,
|
| 1713 |
+
"grad_norm": 7.993265151977539,
|
| 1714 |
+
"learning_rate": 1.9516809149722737e-05,
|
| 1715 |
+
"loss": 0.1042,
|
| 1716 |
+
"step": 12100
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.4036506469500925,
|
| 1720 |
+
"grad_norm": 0.07449086755514145,
|
| 1721 |
+
"learning_rate": 1.9473486598890943e-05,
|
| 1722 |
+
"loss": 0.1227,
|
| 1723 |
+
"step": 12150
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.4094269870609981,
|
| 1727 |
+
"grad_norm": 44.55666732788086,
|
| 1728 |
+
"learning_rate": 1.943016404805915e-05,
|
| 1729 |
+
"loss": 0.1825,
|
| 1730 |
+
"step": 12200
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 1.4152033271719038,
|
| 1734 |
+
"grad_norm": 0.17107024788856506,
|
| 1735 |
+
"learning_rate": 1.9386841497227355e-05,
|
| 1736 |
+
"loss": 0.1174,
|
| 1737 |
+
"step": 12250
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.4209796672828097,
|
| 1741 |
+
"grad_norm": 8.075175285339355,
|
| 1742 |
+
"learning_rate": 1.9343518946395565e-05,
|
| 1743 |
+
"loss": 0.1214,
|
| 1744 |
+
"step": 12300
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.4267560073937153,
|
| 1748 |
+
"grad_norm": 0.02626979723572731,
|
| 1749 |
+
"learning_rate": 1.930019639556377e-05,
|
| 1750 |
+
"loss": 0.1184,
|
| 1751 |
+
"step": 12350
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.432532347504621,
|
| 1755 |
+
"grad_norm": 0.04273197799921036,
|
| 1756 |
+
"learning_rate": 1.925687384473198e-05,
|
| 1757 |
+
"loss": 0.1599,
|
| 1758 |
+
"step": 12400
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.4383086876155269,
|
| 1762 |
+
"grad_norm": 11.037675857543945,
|
| 1763 |
+
"learning_rate": 1.9213551293900186e-05,
|
| 1764 |
+
"loss": 0.1972,
|
| 1765 |
+
"step": 12450
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.4440850277264325,
|
| 1769 |
+
"grad_norm": 0.018304361030459404,
|
| 1770 |
+
"learning_rate": 1.9170228743068392e-05,
|
| 1771 |
+
"loss": 0.109,
|
| 1772 |
+
"step": 12500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.4498613678373382,
|
| 1776 |
+
"grad_norm": 0.10881417989730835,
|
| 1777 |
+
"learning_rate": 1.91269061922366e-05,
|
| 1778 |
+
"loss": 0.1512,
|
| 1779 |
+
"step": 12550
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.4556377079482439,
|
| 1783 |
+
"grad_norm": 0.014431633055210114,
|
| 1784 |
+
"learning_rate": 1.9083583641404805e-05,
|
| 1785 |
+
"loss": 0.0693,
|
| 1786 |
+
"step": 12600
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.4614140480591498,
|
| 1790 |
+
"grad_norm": 4.677309513092041,
|
| 1791 |
+
"learning_rate": 1.9040261090573014e-05,
|
| 1792 |
+
"loss": 0.1791,
|
| 1793 |
+
"step": 12650
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4671903881700554,
|
| 1797 |
+
"grad_norm": 0.042188193649053574,
|
| 1798 |
+
"learning_rate": 1.899693853974122e-05,
|
| 1799 |
+
"loss": 0.1459,
|
| 1800 |
+
"step": 12700
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.4729667282809613,
|
| 1804 |
+
"grad_norm": 0.11454495042562485,
|
| 1805 |
+
"learning_rate": 1.895361598890943e-05,
|
| 1806 |
+
"loss": 0.1243,
|
| 1807 |
+
"step": 12750
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.478743068391867,
|
| 1811 |
+
"grad_norm": 40.89423370361328,
|
| 1812 |
+
"learning_rate": 1.8910293438077636e-05,
|
| 1813 |
+
"loss": 0.1451,
|
| 1814 |
+
"step": 12800
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.4845194085027726,
|
| 1818 |
+
"grad_norm": 0.022999301552772522,
|
| 1819 |
+
"learning_rate": 1.886697088724584e-05,
|
| 1820 |
+
"loss": 0.1615,
|
| 1821 |
+
"step": 12850
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.4902957486136783,
|
| 1825 |
+
"grad_norm": 0.023199884220957756,
|
| 1826 |
+
"learning_rate": 1.8823648336414048e-05,
|
| 1827 |
+
"loss": 0.1213,
|
| 1828 |
+
"step": 12900
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 1.496072088724584,
|
| 1832 |
+
"grad_norm": 26.06696319580078,
|
| 1833 |
+
"learning_rate": 1.8780325785582254e-05,
|
| 1834 |
+
"loss": 0.1625,
|
| 1835 |
+
"step": 12950
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 1.5018484288354899,
|
| 1839 |
+
"grad_norm": 7.6846442222595215,
|
| 1840 |
+
"learning_rate": 1.8737003234750464e-05,
|
| 1841 |
+
"loss": 0.168,
|
| 1842 |
+
"step": 13000
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.5076247689463955,
|
| 1846 |
+
"grad_norm": 0.05135953053832054,
|
| 1847 |
+
"learning_rate": 1.869368068391867e-05,
|
| 1848 |
+
"loss": 0.1258,
|
| 1849 |
+
"step": 13050
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.5134011090573014,
|
| 1853 |
+
"grad_norm": 1.4261308908462524,
|
| 1854 |
+
"learning_rate": 1.865035813308688e-05,
|
| 1855 |
+
"loss": 0.148,
|
| 1856 |
+
"step": 13100
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.519177449168207,
|
| 1860 |
+
"grad_norm": 1.1017619371414185,
|
| 1861 |
+
"learning_rate": 1.8607035582255085e-05,
|
| 1862 |
+
"loss": 0.1134,
|
| 1863 |
+
"step": 13150
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.5249537892791127,
|
| 1867 |
+
"grad_norm": 0.12042645364999771,
|
| 1868 |
+
"learning_rate": 1.8563713031423288e-05,
|
| 1869 |
+
"loss": 0.224,
|
| 1870 |
+
"step": 13200
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.5307301293900184,
|
| 1874 |
+
"grad_norm": 0.073084257543087,
|
| 1875 |
+
"learning_rate": 1.8520390480591497e-05,
|
| 1876 |
+
"loss": 0.0997,
|
| 1877 |
+
"step": 13250
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 1.536506469500924,
|
| 1881 |
+
"grad_norm": 12.779747009277344,
|
| 1882 |
+
"learning_rate": 1.8477067929759704e-05,
|
| 1883 |
+
"loss": 0.161,
|
| 1884 |
+
"step": 13300
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.54228280961183,
|
| 1888 |
+
"grad_norm": 0.04014035314321518,
|
| 1889 |
+
"learning_rate": 1.8433745378927913e-05,
|
| 1890 |
+
"loss": 0.1377,
|
| 1891 |
+
"step": 13350
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.5480591497227358,
|
| 1895 |
+
"grad_norm": 12.693160057067871,
|
| 1896 |
+
"learning_rate": 1.839042282809612e-05,
|
| 1897 |
+
"loss": 0.1347,
|
| 1898 |
+
"step": 13400
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 1.5538354898336415,
|
| 1902 |
+
"grad_norm": 0.43178144097328186,
|
| 1903 |
+
"learning_rate": 1.834710027726433e-05,
|
| 1904 |
+
"loss": 0.1023,
|
| 1905 |
+
"step": 13450
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5596118299445472,
|
| 1909 |
+
"grad_norm": 51.83779525756836,
|
| 1910 |
+
"learning_rate": 1.830377772643253e-05,
|
| 1911 |
+
"loss": 0.082,
|
| 1912 |
+
"step": 13500
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 1.5653881700554528,
|
| 1916 |
+
"grad_norm": 2.3788204193115234,
|
| 1917 |
+
"learning_rate": 1.8260455175600737e-05,
|
| 1918 |
+
"loss": 0.1606,
|
| 1919 |
+
"step": 13550
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 1.5711645101663585,
|
| 1923 |
+
"grad_norm": 0.03366752341389656,
|
| 1924 |
+
"learning_rate": 1.8217132624768947e-05,
|
| 1925 |
+
"loss": 0.0896,
|
| 1926 |
+
"step": 13600
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.5769408502772642,
|
| 1930 |
+
"grad_norm": 15.269149780273438,
|
| 1931 |
+
"learning_rate": 1.8173810073937153e-05,
|
| 1932 |
+
"loss": 0.1463,
|
| 1933 |
+
"step": 13650
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.58271719038817,
|
| 1937 |
+
"grad_norm": 34.45343780517578,
|
| 1938 |
+
"learning_rate": 1.8130487523105362e-05,
|
| 1939 |
+
"loss": 0.0908,
|
| 1940 |
+
"step": 13700
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 1.588493530499076,
|
| 1944 |
+
"grad_norm": 0.3380456268787384,
|
| 1945 |
+
"learning_rate": 1.808716497227357e-05,
|
| 1946 |
+
"loss": 0.0767,
|
| 1947 |
+
"step": 13750
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.5942698706099816,
|
| 1951 |
+
"grad_norm": 27.415430068969727,
|
| 1952 |
+
"learning_rate": 1.8043842421441775e-05,
|
| 1953 |
+
"loss": 0.185,
|
| 1954 |
+
"step": 13800
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 1.6000462107208873,
|
| 1958 |
+
"grad_norm": 59.60631561279297,
|
| 1959 |
+
"learning_rate": 1.800051987060998e-05,
|
| 1960 |
+
"loss": 0.1656,
|
| 1961 |
+
"step": 13850
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.605822550831793,
|
| 1965 |
+
"grad_norm": 30.945571899414062,
|
| 1966 |
+
"learning_rate": 1.7957197319778187e-05,
|
| 1967 |
+
"loss": 0.1029,
|
| 1968 |
+
"step": 13900
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.6115988909426986,
|
| 1972 |
+
"grad_norm": 0.09613625705242157,
|
| 1973 |
+
"learning_rate": 1.7913874768946396e-05,
|
| 1974 |
+
"loss": 0.155,
|
| 1975 |
+
"step": 13950
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.6173752310536045,
|
| 1979 |
+
"grad_norm": 0.5363235473632812,
|
| 1980 |
+
"learning_rate": 1.7870552218114602e-05,
|
| 1981 |
+
"loss": 0.1398,
|
| 1982 |
+
"step": 14000
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 1.6231515711645101,
|
| 1986 |
+
"grad_norm": 0.02551902085542679,
|
| 1987 |
+
"learning_rate": 1.7827229667282812e-05,
|
| 1988 |
+
"loss": 0.1039,
|
| 1989 |
+
"step": 14050
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.628927911275416,
|
| 1993 |
+
"grad_norm": 5.74197244644165,
|
| 1994 |
+
"learning_rate": 1.7783907116451018e-05,
|
| 1995 |
+
"loss": 0.1295,
|
| 1996 |
+
"step": 14100
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 1.6347042513863217,
|
| 2000 |
+
"grad_norm": 0.0256786085665226,
|
| 2001 |
+
"learning_rate": 1.7740584565619224e-05,
|
| 2002 |
+
"loss": 0.0745,
|
| 2003 |
+
"step": 14150
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 1.6404805914972274,
|
| 2007 |
+
"grad_norm": 0.015919797122478485,
|
| 2008 |
+
"learning_rate": 1.769726201478743e-05,
|
| 2009 |
+
"loss": 0.1619,
|
| 2010 |
+
"step": 14200
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.646256931608133,
|
| 2014 |
+
"grad_norm": 39.721153259277344,
|
| 2015 |
+
"learning_rate": 1.7653939463955636e-05,
|
| 2016 |
+
"loss": 0.0708,
|
| 2017 |
+
"step": 14250
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.6520332717190387,
|
| 2021 |
+
"grad_norm": 0.8398991227149963,
|
| 2022 |
+
"learning_rate": 1.7610616913123846e-05,
|
| 2023 |
+
"loss": 0.1335,
|
| 2024 |
+
"step": 14300
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 1.6578096118299446,
|
| 2028 |
+
"grad_norm": 0.9615886211395264,
|
| 2029 |
+
"learning_rate": 1.7567294362292052e-05,
|
| 2030 |
+
"loss": 0.129,
|
| 2031 |
+
"step": 14350
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.6635859519408502,
|
| 2035 |
+
"grad_norm": 0.22856245934963226,
|
| 2036 |
+
"learning_rate": 1.752397181146026e-05,
|
| 2037 |
+
"loss": 0.1506,
|
| 2038 |
+
"step": 14400
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 1.6693622920517561,
|
| 2042 |
+
"grad_norm": 0.24372538924217224,
|
| 2043 |
+
"learning_rate": 1.7480649260628467e-05,
|
| 2044 |
+
"loss": 0.1296,
|
| 2045 |
+
"step": 14450
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 1.6751386321626618,
|
| 2049 |
+
"grad_norm": 0.08972591161727905,
|
| 2050 |
+
"learning_rate": 1.7437326709796674e-05,
|
| 2051 |
+
"loss": 0.1519,
|
| 2052 |
+
"step": 14500
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.6809149722735675,
|
| 2056 |
+
"grad_norm": 0.21274013817310333,
|
| 2057 |
+
"learning_rate": 1.739400415896488e-05,
|
| 2058 |
+
"loss": 0.189,
|
| 2059 |
+
"step": 14550
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.6866913123844731,
|
| 2063 |
+
"grad_norm": 32.43384552001953,
|
| 2064 |
+
"learning_rate": 1.7350681608133086e-05,
|
| 2065 |
+
"loss": 0.0888,
|
| 2066 |
+
"step": 14600
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.6924676524953788,
|
| 2070 |
+
"grad_norm": 0.08137155324220657,
|
| 2071 |
+
"learning_rate": 1.7307359057301295e-05,
|
| 2072 |
+
"loss": 0.0572,
|
| 2073 |
+
"step": 14650
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6982439926062847,
|
| 2077 |
+
"grad_norm": 0.01811046339571476,
|
| 2078 |
+
"learning_rate": 1.72640365064695e-05,
|
| 2079 |
+
"loss": 0.1028,
|
| 2080 |
+
"step": 14700
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.7040203327171903,
|
| 2084 |
+
"grad_norm": 0.015499014407396317,
|
| 2085 |
+
"learning_rate": 1.7220713955637707e-05,
|
| 2086 |
+
"loss": 0.1315,
|
| 2087 |
+
"step": 14750
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.7097966728280962,
|
| 2091 |
+
"grad_norm": 8.582119941711426,
|
| 2092 |
+
"learning_rate": 1.7177391404805917e-05,
|
| 2093 |
+
"loss": 0.2206,
|
| 2094 |
+
"step": 14800
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.7155730129390019,
|
| 2098 |
+
"grad_norm": 7.51249361038208,
|
| 2099 |
+
"learning_rate": 1.713406885397412e-05,
|
| 2100 |
+
"loss": 0.1207,
|
| 2101 |
+
"step": 14850
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.7213493530499075,
|
| 2105 |
+
"grad_norm": 0.00726300198584795,
|
| 2106 |
+
"learning_rate": 1.709074630314233e-05,
|
| 2107 |
+
"loss": 0.1283,
|
| 2108 |
+
"step": 14900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.7271256931608132,
|
| 2112 |
+
"grad_norm": 0.07731953263282776,
|
| 2113 |
+
"learning_rate": 1.7047423752310535e-05,
|
| 2114 |
+
"loss": 0.1694,
|
| 2115 |
+
"step": 14950
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.7329020332717189,
|
| 2119 |
+
"grad_norm": 0.101409412920475,
|
| 2120 |
+
"learning_rate": 1.7004101201478745e-05,
|
| 2121 |
+
"loss": 0.0945,
|
| 2122 |
+
"step": 15000
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.7386783733826248,
|
| 2126 |
+
"grad_norm": 0.04303300008177757,
|
| 2127 |
+
"learning_rate": 1.696077865064695e-05,
|
| 2128 |
+
"loss": 0.1023,
|
| 2129 |
+
"step": 15050
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.7444547134935307,
|
| 2133 |
+
"grad_norm": 0.13228349387645721,
|
| 2134 |
+
"learning_rate": 1.6917456099815157e-05,
|
| 2135 |
+
"loss": 0.126,
|
| 2136 |
+
"step": 15100
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.7502310536044363,
|
| 2140 |
+
"grad_norm": 0.649356484413147,
|
| 2141 |
+
"learning_rate": 1.6874133548983366e-05,
|
| 2142 |
+
"loss": 0.1574,
|
| 2143 |
+
"step": 15150
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.756007393715342,
|
| 2147 |
+
"grad_norm": 16.686784744262695,
|
| 2148 |
+
"learning_rate": 1.683081099815157e-05,
|
| 2149 |
+
"loss": 0.1296,
|
| 2150 |
+
"step": 15200
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.7617837338262476,
|
| 2154 |
+
"grad_norm": 0.03777627646923065,
|
| 2155 |
+
"learning_rate": 1.678748844731978e-05,
|
| 2156 |
+
"loss": 0.0756,
|
| 2157 |
+
"step": 15250
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.7675600739371533,
|
| 2161 |
+
"grad_norm": 0.23532889783382416,
|
| 2162 |
+
"learning_rate": 1.6744165896487985e-05,
|
| 2163 |
+
"loss": 0.1009,
|
| 2164 |
+
"step": 15300
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.7733364140480592,
|
| 2168 |
+
"grad_norm": 18.59041976928711,
|
| 2169 |
+
"learning_rate": 1.6700843345656194e-05,
|
| 2170 |
+
"loss": 0.1599,
|
| 2171 |
+
"step": 15350
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.7791127541589649,
|
| 2175 |
+
"grad_norm": 0.11593695729970932,
|
| 2176 |
+
"learning_rate": 1.66575207948244e-05,
|
| 2177 |
+
"loss": 0.1245,
|
| 2178 |
+
"step": 15400
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.7848890942698707,
|
| 2182 |
+
"grad_norm": 0.07317894697189331,
|
| 2183 |
+
"learning_rate": 1.6614198243992606e-05,
|
| 2184 |
+
"loss": 0.1822,
|
| 2185 |
+
"step": 15450
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7906654343807764,
|
| 2189 |
+
"grad_norm": 0.06145332753658295,
|
| 2190 |
+
"learning_rate": 1.6570875693160816e-05,
|
| 2191 |
+
"loss": 0.1925,
|
| 2192 |
+
"step": 15500
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.796441774491682,
|
| 2196 |
+
"grad_norm": 13.491717338562012,
|
| 2197 |
+
"learning_rate": 1.652755314232902e-05,
|
| 2198 |
+
"loss": 0.1854,
|
| 2199 |
+
"step": 15550
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.8022181146025877,
|
| 2203 |
+
"grad_norm": 0.1411893367767334,
|
| 2204 |
+
"learning_rate": 1.6484230591497228e-05,
|
| 2205 |
+
"loss": 0.1126,
|
| 2206 |
+
"step": 15600
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.8079944547134934,
|
| 2210 |
+
"grad_norm": 10.62831974029541,
|
| 2211 |
+
"learning_rate": 1.6440908040665434e-05,
|
| 2212 |
+
"loss": 0.094,
|
| 2213 |
+
"step": 15650
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.8137707948243993,
|
| 2217 |
+
"grad_norm": 0.01941312849521637,
|
| 2218 |
+
"learning_rate": 1.6397585489833644e-05,
|
| 2219 |
+
"loss": 0.136,
|
| 2220 |
+
"step": 15700
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.819547134935305,
|
| 2224 |
+
"grad_norm": 30.944677352905273,
|
| 2225 |
+
"learning_rate": 1.635426293900185e-05,
|
| 2226 |
+
"loss": 0.1317,
|
| 2227 |
+
"step": 15750
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.8253234750462108,
|
| 2231 |
+
"grad_norm": 16.877334594726562,
|
| 2232 |
+
"learning_rate": 1.6310940388170056e-05,
|
| 2233 |
+
"loss": 0.133,
|
| 2234 |
+
"step": 15800
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.8310998151571165,
|
| 2238 |
+
"grad_norm": 0.1656600683927536,
|
| 2239 |
+
"learning_rate": 1.6267617837338265e-05,
|
| 2240 |
+
"loss": 0.1395,
|
| 2241 |
+
"step": 15850
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.8368761552680222,
|
| 2245 |
+
"grad_norm": 53.970794677734375,
|
| 2246 |
+
"learning_rate": 1.6224295286506468e-05,
|
| 2247 |
+
"loss": 0.1337,
|
| 2248 |
+
"step": 15900
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.8426524953789278,
|
| 2252 |
+
"grad_norm": 36.164947509765625,
|
| 2253 |
+
"learning_rate": 1.6180972735674677e-05,
|
| 2254 |
+
"loss": 0.1251,
|
| 2255 |
+
"step": 15950
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.8484288354898335,
|
| 2259 |
+
"grad_norm": 0.0839138776063919,
|
| 2260 |
+
"learning_rate": 1.6137650184842884e-05,
|
| 2261 |
+
"loss": 0.0661,
|
| 2262 |
+
"step": 16000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.8542051756007394,
|
| 2266 |
+
"grad_norm": 0.1801111400127411,
|
| 2267 |
+
"learning_rate": 1.609432763401109e-05,
|
| 2268 |
+
"loss": 0.1557,
|
| 2269 |
+
"step": 16050
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.859981515711645,
|
| 2273 |
+
"grad_norm": 0.12647980451583862,
|
| 2274 |
+
"learning_rate": 1.60510050831793e-05,
|
| 2275 |
+
"loss": 0.0765,
|
| 2276 |
+
"step": 16100
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.865757855822551,
|
| 2280 |
+
"grad_norm": 0.26282384991645813,
|
| 2281 |
+
"learning_rate": 1.6007682532347505e-05,
|
| 2282 |
+
"loss": 0.0413,
|
| 2283 |
+
"step": 16150
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.8715341959334566,
|
| 2287 |
+
"grad_norm": 0.0952085480093956,
|
| 2288 |
+
"learning_rate": 1.596435998151571e-05,
|
| 2289 |
+
"loss": 0.1286,
|
| 2290 |
+
"step": 16200
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.8773105360443623,
|
| 2294 |
+
"grad_norm": 0.11837054044008255,
|
| 2295 |
+
"learning_rate": 1.5921037430683917e-05,
|
| 2296 |
+
"loss": 0.1891,
|
| 2297 |
+
"step": 16250
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.883086876155268,
|
| 2301 |
+
"grad_norm": 0.053552981466054916,
|
| 2302 |
+
"learning_rate": 1.5877714879852127e-05,
|
| 2303 |
+
"loss": 0.1535,
|
| 2304 |
+
"step": 16300
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.8888632162661736,
|
| 2308 |
+
"grad_norm": 1.4103118181228638,
|
| 2309 |
+
"learning_rate": 1.5834392329020333e-05,
|
| 2310 |
+
"loss": 0.1414,
|
| 2311 |
+
"step": 16350
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.8946395563770795,
|
| 2315 |
+
"grad_norm": 0.07249405235052109,
|
| 2316 |
+
"learning_rate": 1.579106977818854e-05,
|
| 2317 |
+
"loss": 0.1125,
|
| 2318 |
+
"step": 16400
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.9004158964879854,
|
| 2322 |
+
"grad_norm": 0.016298620030283928,
|
| 2323 |
+
"learning_rate": 1.574774722735675e-05,
|
| 2324 |
+
"loss": 0.1219,
|
| 2325 |
+
"step": 16450
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.906192236598891,
|
| 2329 |
+
"grad_norm": 0.025717375800013542,
|
| 2330 |
+
"learning_rate": 1.5704424676524955e-05,
|
| 2331 |
+
"loss": 0.1248,
|
| 2332 |
+
"step": 16500
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.9119685767097967,
|
| 2336 |
+
"grad_norm": 0.07495362311601639,
|
| 2337 |
+
"learning_rate": 1.566110212569316e-05,
|
| 2338 |
+
"loss": 0.1399,
|
| 2339 |
+
"step": 16550
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.9177449168207024,
|
| 2343 |
+
"grad_norm": 0.08207421749830246,
|
| 2344 |
+
"learning_rate": 1.5617779574861367e-05,
|
| 2345 |
+
"loss": 0.1092,
|
| 2346 |
+
"step": 16600
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.923521256931608,
|
| 2350 |
+
"grad_norm": 0.02017366513609886,
|
| 2351 |
+
"learning_rate": 1.5574457024029576e-05,
|
| 2352 |
+
"loss": 0.1164,
|
| 2353 |
+
"step": 16650
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.929297597042514,
|
| 2357 |
+
"grad_norm": 0.10131692886352539,
|
| 2358 |
+
"learning_rate": 1.5531134473197782e-05,
|
| 2359 |
+
"loss": 0.1416,
|
| 2360 |
+
"step": 16700
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.9350739371534196,
|
| 2364 |
+
"grad_norm": 0.03260243684053421,
|
| 2365 |
+
"learning_rate": 1.548781192236599e-05,
|
| 2366 |
+
"loss": 0.0695,
|
| 2367 |
+
"step": 16750
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.9408502772643255,
|
| 2371 |
+
"grad_norm": 0.17720632255077362,
|
| 2372 |
+
"learning_rate": 1.5444489371534198e-05,
|
| 2373 |
+
"loss": 0.1201,
|
| 2374 |
+
"step": 16800
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.9466266173752311,
|
| 2378 |
+
"grad_norm": 3.364078998565674,
|
| 2379 |
+
"learning_rate": 1.5401166820702404e-05,
|
| 2380 |
+
"loss": 0.1684,
|
| 2381 |
+
"step": 16850
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.9524029574861368,
|
| 2385 |
+
"grad_norm": 31.63196563720703,
|
| 2386 |
+
"learning_rate": 1.535784426987061e-05,
|
| 2387 |
+
"loss": 0.0964,
|
| 2388 |
+
"step": 16900
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.9581792975970425,
|
| 2392 |
+
"grad_norm": 0.049181848764419556,
|
| 2393 |
+
"learning_rate": 1.5314521719038816e-05,
|
| 2394 |
+
"loss": 0.1232,
|
| 2395 |
+
"step": 16950
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.9639556377079481,
|
| 2399 |
+
"grad_norm": 0.027907520532608032,
|
| 2400 |
+
"learning_rate": 1.5271199168207022e-05,
|
| 2401 |
+
"loss": 0.1164,
|
| 2402 |
+
"step": 17000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.969731977818854,
|
| 2406 |
+
"grad_norm": 0.030601385980844498,
|
| 2407 |
+
"learning_rate": 1.5227876617375232e-05,
|
| 2408 |
+
"loss": 0.0783,
|
| 2409 |
+
"step": 17050
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.9755083179297597,
|
| 2413 |
+
"grad_norm": 0.009862957522273064,
|
| 2414 |
+
"learning_rate": 1.518455406654344e-05,
|
| 2415 |
+
"loss": 0.1063,
|
| 2416 |
+
"step": 17100
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.9812846580406656,
|
| 2420 |
+
"grad_norm": 40.9315185546875,
|
| 2421 |
+
"learning_rate": 1.5141231515711646e-05,
|
| 2422 |
+
"loss": 0.1464,
|
| 2423 |
+
"step": 17150
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.9870609981515712,
|
| 2427 |
+
"grad_norm": 0.12488950043916702,
|
| 2428 |
+
"learning_rate": 1.5097908964879854e-05,
|
| 2429 |
+
"loss": 0.1145,
|
| 2430 |
+
"step": 17200
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.9928373382624769,
|
| 2434 |
+
"grad_norm": 0.08856427669525146,
|
| 2435 |
+
"learning_rate": 1.5054586414048058e-05,
|
| 2436 |
+
"loss": 0.0934,
|
| 2437 |
+
"step": 17250
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.9986136783733826,
|
| 2441 |
+
"grad_norm": 0.046317312866449356,
|
| 2442 |
+
"learning_rate": 1.5011263863216266e-05,
|
| 2443 |
+
"loss": 0.0957,
|
| 2444 |
+
"step": 17300
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 2.0,
|
| 2448 |
+
"eval_accuracy": 0.9604938271604938,
|
| 2449 |
+
"eval_f1": 0.9761680777673252,
|
| 2450 |
+
"eval_loss": 0.17032423615455627,
|
| 2451 |
+
"eval_precision": 0.9772406215664731,
|
| 2452 |
+
"eval_recall": 0.975097885669538,
|
| 2453 |
+
"eval_runtime": 82.0507,
|
| 2454 |
+
"eval_samples_per_second": 93.784,
|
| 2455 |
+
"eval_steps_per_second": 11.724,
|
| 2456 |
+
"step": 17312
|
| 2457 |
+
}
|
| 2458 |
+
],
|
| 2459 |
+
"logging_steps": 50,
|
| 2460 |
+
"max_steps": 34624,
|
| 2461 |
+
"num_input_tokens_seen": 0,
|
| 2462 |
+
"num_train_epochs": 4,
|
| 2463 |
+
"save_steps": 500,
|
| 2464 |
+
"stateful_callbacks": {
|
| 2465 |
+
"TrainerControl": {
|
| 2466 |
+
"args": {
|
| 2467 |
+
"should_epoch_stop": false,
|
| 2468 |
+
"should_evaluate": false,
|
| 2469 |
+
"should_log": false,
|
| 2470 |
+
"should_save": true,
|
| 2471 |
+
"should_training_stop": false
|
| 2472 |
+
},
|
| 2473 |
+
"attributes": {}
|
| 2474 |
+
}
|
| 2475 |
+
},
|
| 2476 |
+
"total_flos": 1.9727907688201273e+19,
|
| 2477 |
+
"train_batch_size": 8,
|
| 2478 |
+
"trial_name": null,
|
| 2479 |
+
"trial_params": null
|
| 2480 |
+
}
|
checkpoint-17312/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:946e9e07d5a9e21a8c78cfee12dd636f44d131fddd49601bf03b9867e5b39a46
|
| 3 |
+
size 5841
|
checkpoint-25968/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-25968/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13cd2a1593e0925ce5a4840518a9b195439a489ee2fe4a13bd3447bb04f3b710
|
| 3 |
+
size 631399808
|
checkpoint-25968/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20358b4b1f0c1f93bac5865a8a3027fbd100e1aaf2db8b9b3c6f0ab9c35d0f7d
|
| 3 |
+
size 1262124766
|
checkpoint-25968/preprocessor_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.485,
|
| 8 |
+
0.456,
|
| 9 |
+
0.406
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "ViTImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.229,
|
| 14 |
+
0.224,
|
| 15 |
+
0.225
|
| 16 |
+
],
|
| 17 |
+
"resample": 3,
|
| 18 |
+
"rescale_factor": 0.00392156862745098,
|
| 19 |
+
"size": {
|
| 20 |
+
"height": 224,
|
| 21 |
+
"width": 224
|
| 22 |
+
}
|
| 23 |
+
}
|
checkpoint-25968/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78534c48fa6f5c9294b580fad572ccf8abbcc3bf9ed06ff4d4f5688def3c223d
|
| 3 |
+
size 14645
|
checkpoint-25968/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a8a5cc8454cd7401853746cd33bb3fb5eebcdf6c075306aa29c5c4ca1a5df199
|
| 3 |
+
size 1383
|
checkpoint-25968/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c88152302525653dcc80bdccc5697bbbf4bce38273847ab642af8f220351bb0a
|
| 3 |
+
size 1465
|
checkpoint-25968/trainer_state.json
ADDED
|
@@ -0,0 +1,3703 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 25968,
|
| 3 |
+
"best_metric": 0.9809026424506977,
|
| 4 |
+
"best_model_checkpoint": "./roadwork-swin-finetuned-local/checkpoint-25968",
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 25968,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00577634011090573,
|
| 14 |
+
"grad_norm": 12.141866683959961,
|
| 15 |
+
"learning_rate": 2.9957543900184844e-05,
|
| 16 |
+
"loss": 3.0039,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01155268022181146,
|
| 21 |
+
"grad_norm": 13.438708305358887,
|
| 22 |
+
"learning_rate": 2.991422134935305e-05,
|
| 23 |
+
"loss": 0.3812,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01732902033271719,
|
| 28 |
+
"grad_norm": 38.80968475341797,
|
| 29 |
+
"learning_rate": 2.9870898798521256e-05,
|
| 30 |
+
"loss": 0.3451,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02310536044362292,
|
| 35 |
+
"grad_norm": 16.20616340637207,
|
| 36 |
+
"learning_rate": 2.9827576247689466e-05,
|
| 37 |
+
"loss": 0.3082,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028881700554528652,
|
| 42 |
+
"grad_norm": 33.16380310058594,
|
| 43 |
+
"learning_rate": 2.9784253696857672e-05,
|
| 44 |
+
"loss": 0.3153,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03465804066543438,
|
| 49 |
+
"grad_norm": 18.74236297607422,
|
| 50 |
+
"learning_rate": 2.974093114602588e-05,
|
| 51 |
+
"loss": 0.3792,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04043438077634011,
|
| 56 |
+
"grad_norm": 4.538590431213379,
|
| 57 |
+
"learning_rate": 2.9697608595194084e-05,
|
| 58 |
+
"loss": 0.3673,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04621072088724584,
|
| 63 |
+
"grad_norm": 15.369573593139648,
|
| 64 |
+
"learning_rate": 2.9654286044362294e-05,
|
| 65 |
+
"loss": 0.3875,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.05198706099815157,
|
| 70 |
+
"grad_norm": 23.82215690612793,
|
| 71 |
+
"learning_rate": 2.96109634935305e-05,
|
| 72 |
+
"loss": 0.3386,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.057763401109057304,
|
| 77 |
+
"grad_norm": 14.058465003967285,
|
| 78 |
+
"learning_rate": 2.9567640942698706e-05,
|
| 79 |
+
"loss": 0.2766,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06353974121996303,
|
| 84 |
+
"grad_norm": 15.244014739990234,
|
| 85 |
+
"learning_rate": 2.9524318391866915e-05,
|
| 86 |
+
"loss": 0.2891,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06931608133086876,
|
| 91 |
+
"grad_norm": 30.079912185668945,
|
| 92 |
+
"learning_rate": 2.948099584103512e-05,
|
| 93 |
+
"loss": 0.315,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0750924214417745,
|
| 98 |
+
"grad_norm": 22.175500869750977,
|
| 99 |
+
"learning_rate": 2.9437673290203327e-05,
|
| 100 |
+
"loss": 0.2571,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08086876155268022,
|
| 105 |
+
"grad_norm": 13.770343780517578,
|
| 106 |
+
"learning_rate": 2.9394350739371534e-05,
|
| 107 |
+
"loss": 0.2805,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08664510166358595,
|
| 112 |
+
"grad_norm": 15.201981544494629,
|
| 113 |
+
"learning_rate": 2.9351028188539743e-05,
|
| 114 |
+
"loss": 0.3395,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09242144177449169,
|
| 119 |
+
"grad_norm": 0.11971250176429749,
|
| 120 |
+
"learning_rate": 2.930770563770795e-05,
|
| 121 |
+
"loss": 0.2807,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0981977818853974,
|
| 126 |
+
"grad_norm": 13.49336051940918,
|
| 127 |
+
"learning_rate": 2.9264383086876155e-05,
|
| 128 |
+
"loss": 0.3474,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10397412199630314,
|
| 133 |
+
"grad_norm": 12.165586471557617,
|
| 134 |
+
"learning_rate": 2.9221060536044365e-05,
|
| 135 |
+
"loss": 0.2704,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.10975046210720887,
|
| 140 |
+
"grad_norm": 14.561868667602539,
|
| 141 |
+
"learning_rate": 2.917773798521257e-05,
|
| 142 |
+
"loss": 0.2495,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.11552680221811461,
|
| 147 |
+
"grad_norm": 3.2725448608398438,
|
| 148 |
+
"learning_rate": 2.9134415434380777e-05,
|
| 149 |
+
"loss": 0.2951,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.12130314232902033,
|
| 154 |
+
"grad_norm": 12.453506469726562,
|
| 155 |
+
"learning_rate": 2.9091092883548983e-05,
|
| 156 |
+
"loss": 0.3338,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12707948243992606,
|
| 161 |
+
"grad_norm": 21.19861602783203,
|
| 162 |
+
"learning_rate": 2.904777033271719e-05,
|
| 163 |
+
"loss": 0.4023,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13285582255083178,
|
| 168 |
+
"grad_norm": 7.617332935333252,
|
| 169 |
+
"learning_rate": 2.90044477818854e-05,
|
| 170 |
+
"loss": 0.2225,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13863216266173753,
|
| 175 |
+
"grad_norm": 7.300302028656006,
|
| 176 |
+
"learning_rate": 2.8961125231053605e-05,
|
| 177 |
+
"loss": 0.2645,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14440850277264325,
|
| 182 |
+
"grad_norm": 11.860857963562012,
|
| 183 |
+
"learning_rate": 2.8917802680221814e-05,
|
| 184 |
+
"loss": 0.3154,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.150184842883549,
|
| 189 |
+
"grad_norm": 20.967884063720703,
|
| 190 |
+
"learning_rate": 2.887448012939002e-05,
|
| 191 |
+
"loss": 0.2485,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.15596118299445472,
|
| 196 |
+
"grad_norm": 11.119569778442383,
|
| 197 |
+
"learning_rate": 2.8831157578558226e-05,
|
| 198 |
+
"loss": 0.2854,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.16173752310536044,
|
| 203 |
+
"grad_norm": 4.218294620513916,
|
| 204 |
+
"learning_rate": 2.8787835027726432e-05,
|
| 205 |
+
"loss": 0.2434,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.16751386321626618,
|
| 210 |
+
"grad_norm": 5.249250888824463,
|
| 211 |
+
"learning_rate": 2.874451247689464e-05,
|
| 212 |
+
"loss": 0.3393,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1732902033271719,
|
| 217 |
+
"grad_norm": 1.712246298789978,
|
| 218 |
+
"learning_rate": 2.8701189926062848e-05,
|
| 219 |
+
"loss": 0.2331,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17906654343807762,
|
| 224 |
+
"grad_norm": 28.958724975585938,
|
| 225 |
+
"learning_rate": 2.8657867375231054e-05,
|
| 226 |
+
"loss": 0.2514,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.18484288354898337,
|
| 231 |
+
"grad_norm": 10.893854141235352,
|
| 232 |
+
"learning_rate": 2.8614544824399264e-05,
|
| 233 |
+
"loss": 0.3656,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1906192236598891,
|
| 238 |
+
"grad_norm": 26.43993377685547,
|
| 239 |
+
"learning_rate": 2.857122227356747e-05,
|
| 240 |
+
"loss": 0.2875,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.1963955637707948,
|
| 245 |
+
"grad_norm": 0.9008654952049255,
|
| 246 |
+
"learning_rate": 2.8527899722735676e-05,
|
| 247 |
+
"loss": 0.2569,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.20217190388170056,
|
| 252 |
+
"grad_norm": 6.621076583862305,
|
| 253 |
+
"learning_rate": 2.8484577171903882e-05,
|
| 254 |
+
"loss": 0.2228,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.20794824399260628,
|
| 259 |
+
"grad_norm": 0.4961116909980774,
|
| 260 |
+
"learning_rate": 2.8441254621072088e-05,
|
| 261 |
+
"loss": 0.368,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.21372458410351203,
|
| 266 |
+
"grad_norm": 31.944732666015625,
|
| 267 |
+
"learning_rate": 2.8397932070240297e-05,
|
| 268 |
+
"loss": 0.2735,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21950092421441775,
|
| 273 |
+
"grad_norm": 34.241703033447266,
|
| 274 |
+
"learning_rate": 2.8354609519408504e-05,
|
| 275 |
+
"loss": 0.2414,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.22527726432532347,
|
| 280 |
+
"grad_norm": 5.89654016494751,
|
| 281 |
+
"learning_rate": 2.8311286968576713e-05,
|
| 282 |
+
"loss": 0.2939,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.23105360443622922,
|
| 287 |
+
"grad_norm": 6.246095657348633,
|
| 288 |
+
"learning_rate": 2.8267964417744916e-05,
|
| 289 |
+
"loss": 0.2772,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.23682994454713494,
|
| 294 |
+
"grad_norm": 12.515145301818848,
|
| 295 |
+
"learning_rate": 2.8224641866913122e-05,
|
| 296 |
+
"loss": 0.2363,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.24260628465804066,
|
| 301 |
+
"grad_norm": 7.227047920227051,
|
| 302 |
+
"learning_rate": 2.818131931608133e-05,
|
| 303 |
+
"loss": 0.2087,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.2483826247689464,
|
| 308 |
+
"grad_norm": 8.628905296325684,
|
| 309 |
+
"learning_rate": 2.8137996765249537e-05,
|
| 310 |
+
"loss": 0.2902,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.2541589648798521,
|
| 315 |
+
"grad_norm": 15.576581001281738,
|
| 316 |
+
"learning_rate": 2.8094674214417747e-05,
|
| 317 |
+
"loss": 0.2752,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.25993530499075784,
|
| 322 |
+
"grad_norm": 16.116729736328125,
|
| 323 |
+
"learning_rate": 2.8051351663585953e-05,
|
| 324 |
+
"loss": 0.2604,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26571164510166356,
|
| 329 |
+
"grad_norm": 12.389599800109863,
|
| 330 |
+
"learning_rate": 2.8008029112754162e-05,
|
| 331 |
+
"loss": 0.2661,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.27148798521256934,
|
| 336 |
+
"grad_norm": 5.002556800842285,
|
| 337 |
+
"learning_rate": 2.7964706561922365e-05,
|
| 338 |
+
"loss": 0.2844,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27726432532347506,
|
| 343 |
+
"grad_norm": 23.136369705200195,
|
| 344 |
+
"learning_rate": 2.792138401109057e-05,
|
| 345 |
+
"loss": 0.2737,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.2830406654343808,
|
| 350 |
+
"grad_norm": 7.032516002655029,
|
| 351 |
+
"learning_rate": 2.787806146025878e-05,
|
| 352 |
+
"loss": 0.2597,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2888170055452865,
|
| 357 |
+
"grad_norm": 24.333494186401367,
|
| 358 |
+
"learning_rate": 2.7834738909426987e-05,
|
| 359 |
+
"loss": 0.2748,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.2945933456561922,
|
| 364 |
+
"grad_norm": 0.4094283878803253,
|
| 365 |
+
"learning_rate": 2.7791416358595196e-05,
|
| 366 |
+
"loss": 0.2368,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.300369685767098,
|
| 371 |
+
"grad_norm": 4.650086402893066,
|
| 372 |
+
"learning_rate": 2.7748093807763402e-05,
|
| 373 |
+
"loss": 0.3215,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.3061460258780037,
|
| 378 |
+
"grad_norm": 18.08386993408203,
|
| 379 |
+
"learning_rate": 2.7704771256931612e-05,
|
| 380 |
+
"loss": 0.255,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.31192236598890943,
|
| 385 |
+
"grad_norm": 10.52841854095459,
|
| 386 |
+
"learning_rate": 2.7661448706099815e-05,
|
| 387 |
+
"loss": 0.2439,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.31769870609981515,
|
| 392 |
+
"grad_norm": 65.88616180419922,
|
| 393 |
+
"learning_rate": 2.761812615526802e-05,
|
| 394 |
+
"loss": 0.2418,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.3234750462107209,
|
| 399 |
+
"grad_norm": 9.300713539123535,
|
| 400 |
+
"learning_rate": 2.757480360443623e-05,
|
| 401 |
+
"loss": 0.3188,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.3292513863216266,
|
| 406 |
+
"grad_norm": 8.528387069702148,
|
| 407 |
+
"learning_rate": 2.7531481053604436e-05,
|
| 408 |
+
"loss": 0.2996,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.33502772643253237,
|
| 413 |
+
"grad_norm": 25.872819900512695,
|
| 414 |
+
"learning_rate": 2.7488158502772646e-05,
|
| 415 |
+
"loss": 0.2303,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3408040665434381,
|
| 420 |
+
"grad_norm": 32.80488204956055,
|
| 421 |
+
"learning_rate": 2.7444835951940852e-05,
|
| 422 |
+
"loss": 0.2472,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.3465804066543438,
|
| 427 |
+
"grad_norm": 3.634104013442993,
|
| 428 |
+
"learning_rate": 2.7401513401109058e-05,
|
| 429 |
+
"loss": 0.3054,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.35235674676524953,
|
| 434 |
+
"grad_norm": 9.497197151184082,
|
| 435 |
+
"learning_rate": 2.7358190850277264e-05,
|
| 436 |
+
"loss": 0.264,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.35813308687615525,
|
| 441 |
+
"grad_norm": 28.931177139282227,
|
| 442 |
+
"learning_rate": 2.731486829944547e-05,
|
| 443 |
+
"loss": 0.2419,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.363909426987061,
|
| 448 |
+
"grad_norm": 25.71267318725586,
|
| 449 |
+
"learning_rate": 2.727154574861368e-05,
|
| 450 |
+
"loss": 0.2137,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.36968576709796674,
|
| 455 |
+
"grad_norm": 18.439529418945312,
|
| 456 |
+
"learning_rate": 2.7228223197781886e-05,
|
| 457 |
+
"loss": 0.2562,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.37546210720887246,
|
| 462 |
+
"grad_norm": 2.7945632934570312,
|
| 463 |
+
"learning_rate": 2.7184900646950095e-05,
|
| 464 |
+
"loss": 0.3085,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3812384473197782,
|
| 469 |
+
"grad_norm": 20.88351058959961,
|
| 470 |
+
"learning_rate": 2.71415780961183e-05,
|
| 471 |
+
"loss": 0.1988,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3870147874306839,
|
| 476 |
+
"grad_norm": 9.544013977050781,
|
| 477 |
+
"learning_rate": 2.7098255545286504e-05,
|
| 478 |
+
"loss": 0.252,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.3927911275415896,
|
| 483 |
+
"grad_norm": 0.2528328001499176,
|
| 484 |
+
"learning_rate": 2.7054932994454714e-05,
|
| 485 |
+
"loss": 0.2136,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3985674676524954,
|
| 490 |
+
"grad_norm": 15.808114051818848,
|
| 491 |
+
"learning_rate": 2.701161044362292e-05,
|
| 492 |
+
"loss": 0.3502,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4043438077634011,
|
| 497 |
+
"grad_norm": 9.585843086242676,
|
| 498 |
+
"learning_rate": 2.696828789279113e-05,
|
| 499 |
+
"loss": 0.2437,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.41012014787430684,
|
| 504 |
+
"grad_norm": 0.7092911601066589,
|
| 505 |
+
"learning_rate": 2.6924965341959335e-05,
|
| 506 |
+
"loss": 0.2137,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.41589648798521256,
|
| 511 |
+
"grad_norm": 0.09910676628351212,
|
| 512 |
+
"learning_rate": 2.6881642791127545e-05,
|
| 513 |
+
"loss": 0.2979,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.4216728280961183,
|
| 518 |
+
"grad_norm": 0.674543559551239,
|
| 519 |
+
"learning_rate": 2.683832024029575e-05,
|
| 520 |
+
"loss": 0.2392,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.42744916820702406,
|
| 525 |
+
"grad_norm": 0.6626635193824768,
|
| 526 |
+
"learning_rate": 2.6794997689463953e-05,
|
| 527 |
+
"loss": 0.2019,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.4332255083179298,
|
| 532 |
+
"grad_norm": 9.206210136413574,
|
| 533 |
+
"learning_rate": 2.6751675138632163e-05,
|
| 534 |
+
"loss": 0.2747,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.4390018484288355,
|
| 539 |
+
"grad_norm": 1.017768144607544,
|
| 540 |
+
"learning_rate": 2.670835258780037e-05,
|
| 541 |
+
"loss": 0.1992,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.4447781885397412,
|
| 546 |
+
"grad_norm": 18.11737060546875,
|
| 547 |
+
"learning_rate": 2.666503003696858e-05,
|
| 548 |
+
"loss": 0.2201,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.45055452865064693,
|
| 553 |
+
"grad_norm": 15.56303882598877,
|
| 554 |
+
"learning_rate": 2.6621707486136785e-05,
|
| 555 |
+
"loss": 0.198,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.45633086876155265,
|
| 560 |
+
"grad_norm": 7.689689636230469,
|
| 561 |
+
"learning_rate": 2.657838493530499e-05,
|
| 562 |
+
"loss": 0.2894,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.46210720887245843,
|
| 567 |
+
"grad_norm": 9.518302917480469,
|
| 568 |
+
"learning_rate": 2.65350623844732e-05,
|
| 569 |
+
"loss": 0.2226,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.46788354898336415,
|
| 574 |
+
"grad_norm": 0.4917680025100708,
|
| 575 |
+
"learning_rate": 2.6491739833641403e-05,
|
| 576 |
+
"loss": 0.2602,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.47365988909426987,
|
| 581 |
+
"grad_norm": 34.92518997192383,
|
| 582 |
+
"learning_rate": 2.6448417282809612e-05,
|
| 583 |
+
"loss": 0.2506,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.4794362292051756,
|
| 588 |
+
"grad_norm": 14.343157768249512,
|
| 589 |
+
"learning_rate": 2.640509473197782e-05,
|
| 590 |
+
"loss": 0.3009,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4852125693160813,
|
| 595 |
+
"grad_norm": 14.651656150817871,
|
| 596 |
+
"learning_rate": 2.6361772181146028e-05,
|
| 597 |
+
"loss": 0.2419,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4909889094269871,
|
| 602 |
+
"grad_norm": 0.7227654457092285,
|
| 603 |
+
"learning_rate": 2.6318449630314234e-05,
|
| 604 |
+
"loss": 0.3091,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4967652495378928,
|
| 609 |
+
"grad_norm": 10.362215042114258,
|
| 610 |
+
"learning_rate": 2.627512707948244e-05,
|
| 611 |
+
"loss": 0.1949,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.5025415896487985,
|
| 616 |
+
"grad_norm": 11.842758178710938,
|
| 617 |
+
"learning_rate": 2.623180452865065e-05,
|
| 618 |
+
"loss": 0.3337,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.5083179297597042,
|
| 623 |
+
"grad_norm": 11.550918579101562,
|
| 624 |
+
"learning_rate": 2.6188481977818852e-05,
|
| 625 |
+
"loss": 0.2641,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.51409426987061,
|
| 630 |
+
"grad_norm": 5.512154579162598,
|
| 631 |
+
"learning_rate": 2.6145159426987062e-05,
|
| 632 |
+
"loss": 0.2541,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5198706099815157,
|
| 637 |
+
"grad_norm": 10.161574363708496,
|
| 638 |
+
"learning_rate": 2.6101836876155268e-05,
|
| 639 |
+
"loss": 0.256,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.5256469500924215,
|
| 644 |
+
"grad_norm": 1.659515380859375,
|
| 645 |
+
"learning_rate": 2.6058514325323477e-05,
|
| 646 |
+
"loss": 0.2621,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.5314232902033271,
|
| 651 |
+
"grad_norm": 2.5476644039154053,
|
| 652 |
+
"learning_rate": 2.6015191774491684e-05,
|
| 653 |
+
"loss": 0.294,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5371996303142329,
|
| 658 |
+
"grad_norm": 0.9729189276695251,
|
| 659 |
+
"learning_rate": 2.597186922365989e-05,
|
| 660 |
+
"loss": 0.1333,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.5429759704251387,
|
| 665 |
+
"grad_norm": 3.8201911449432373,
|
| 666 |
+
"learning_rate": 2.59285466728281e-05,
|
| 667 |
+
"loss": 0.2598,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.5487523105360443,
|
| 672 |
+
"grad_norm": 0.17872726917266846,
|
| 673 |
+
"learning_rate": 2.5885224121996302e-05,
|
| 674 |
+
"loss": 0.1897,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.5545286506469501,
|
| 679 |
+
"grad_norm": 24.479658126831055,
|
| 680 |
+
"learning_rate": 2.584190157116451e-05,
|
| 681 |
+
"loss": 0.2606,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.5603049907578558,
|
| 686 |
+
"grad_norm": 13.361931800842285,
|
| 687 |
+
"learning_rate": 2.5798579020332717e-05,
|
| 688 |
+
"loss": 0.271,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5660813308687616,
|
| 693 |
+
"grad_norm": 8.625031471252441,
|
| 694 |
+
"learning_rate": 2.5755256469500924e-05,
|
| 695 |
+
"loss": 0.2056,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.5718576709796673,
|
| 700 |
+
"grad_norm": 11.34391975402832,
|
| 701 |
+
"learning_rate": 2.5711933918669133e-05,
|
| 702 |
+
"loss": 0.2736,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.577634011090573,
|
| 707 |
+
"grad_norm": 16.9676456451416,
|
| 708 |
+
"learning_rate": 2.566861136783734e-05,
|
| 709 |
+
"loss": 0.2821,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5834103512014788,
|
| 714 |
+
"grad_norm": 9.531896591186523,
|
| 715 |
+
"learning_rate": 2.5625288817005545e-05,
|
| 716 |
+
"loss": 0.2264,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5891866913123844,
|
| 721 |
+
"grad_norm": 2.4691050052642822,
|
| 722 |
+
"learning_rate": 2.558196626617375e-05,
|
| 723 |
+
"loss": 0.2148,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5949630314232902,
|
| 728 |
+
"grad_norm": 0.476725697517395,
|
| 729 |
+
"learning_rate": 2.553864371534196e-05,
|
| 730 |
+
"loss": 0.1964,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.600739371534196,
|
| 735 |
+
"grad_norm": 14.917049407958984,
|
| 736 |
+
"learning_rate": 2.5495321164510167e-05,
|
| 737 |
+
"loss": 0.2717,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.6065157116451017,
|
| 742 |
+
"grad_norm": 0.13745556771755219,
|
| 743 |
+
"learning_rate": 2.5451998613678373e-05,
|
| 744 |
+
"loss": 0.2086,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.6122920517560074,
|
| 749 |
+
"grad_norm": 22.109750747680664,
|
| 750 |
+
"learning_rate": 2.5408676062846582e-05,
|
| 751 |
+
"loss": 0.3401,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.6180683918669131,
|
| 756 |
+
"grad_norm": 0.4055859446525574,
|
| 757 |
+
"learning_rate": 2.536535351201479e-05,
|
| 758 |
+
"loss": 0.2572,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6238447319778189,
|
| 763 |
+
"grad_norm": 0.7329151034355164,
|
| 764 |
+
"learning_rate": 2.5322030961182995e-05,
|
| 765 |
+
"loss": 0.2246,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.6296210720887245,
|
| 770 |
+
"grad_norm": 0.6793464422225952,
|
| 771 |
+
"learning_rate": 2.52787084103512e-05,
|
| 772 |
+
"loss": 0.1804,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.6353974121996303,
|
| 777 |
+
"grad_norm": 6.19757604598999,
|
| 778 |
+
"learning_rate": 2.523538585951941e-05,
|
| 779 |
+
"loss": 0.2331,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.6411737523105361,
|
| 784 |
+
"grad_norm": 69.8536148071289,
|
| 785 |
+
"learning_rate": 2.5192063308687616e-05,
|
| 786 |
+
"loss": 0.2071,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.6469500924214417,
|
| 791 |
+
"grad_norm": 49.18756103515625,
|
| 792 |
+
"learning_rate": 2.5148740757855822e-05,
|
| 793 |
+
"loss": 0.2602,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.6527264325323475,
|
| 798 |
+
"grad_norm": 9.139322280883789,
|
| 799 |
+
"learning_rate": 2.5105418207024032e-05,
|
| 800 |
+
"loss": 0.1975,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.6585027726432532,
|
| 805 |
+
"grad_norm": 11.072602272033691,
|
| 806 |
+
"learning_rate": 2.5062095656192238e-05,
|
| 807 |
+
"loss": 0.1829,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.664279112754159,
|
| 812 |
+
"grad_norm": 0.06665151566267014,
|
| 813 |
+
"learning_rate": 2.5018773105360444e-05,
|
| 814 |
+
"loss": 0.2722,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.6700554528650647,
|
| 819 |
+
"grad_norm": 1.071594476699829,
|
| 820 |
+
"learning_rate": 2.497545055452865e-05,
|
| 821 |
+
"loss": 0.1957,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.6758317929759704,
|
| 826 |
+
"grad_norm": 3.1452102661132812,
|
| 827 |
+
"learning_rate": 2.4932128003696856e-05,
|
| 828 |
+
"loss": 0.1312,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.6816081330868762,
|
| 833 |
+
"grad_norm": 2.483550786972046,
|
| 834 |
+
"learning_rate": 2.4888805452865066e-05,
|
| 835 |
+
"loss": 0.2375,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.6873844731977818,
|
| 840 |
+
"grad_norm": 24.874065399169922,
|
| 841 |
+
"learning_rate": 2.4845482902033272e-05,
|
| 842 |
+
"loss": 0.2743,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.6931608133086876,
|
| 847 |
+
"grad_norm": 31.57330322265625,
|
| 848 |
+
"learning_rate": 2.480216035120148e-05,
|
| 849 |
+
"loss": 0.2298,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.6989371534195934,
|
| 854 |
+
"grad_norm": 21.310131072998047,
|
| 855 |
+
"learning_rate": 2.4758837800369687e-05,
|
| 856 |
+
"loss": 0.2634,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.7047134935304991,
|
| 861 |
+
"grad_norm": 10.233534812927246,
|
| 862 |
+
"learning_rate": 2.4715515249537894e-05,
|
| 863 |
+
"loss": 0.2778,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7104898336414048,
|
| 868 |
+
"grad_norm": 0.34112706780433655,
|
| 869 |
+
"learning_rate": 2.46721926987061e-05,
|
| 870 |
+
"loss": 0.2009,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.7162661737523105,
|
| 875 |
+
"grad_norm": 0.2525830864906311,
|
| 876 |
+
"learning_rate": 2.4628870147874306e-05,
|
| 877 |
+
"loss": 0.2068,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.7220425138632163,
|
| 882 |
+
"grad_norm": 16.547571182250977,
|
| 883 |
+
"learning_rate": 2.4585547597042515e-05,
|
| 884 |
+
"loss": 0.2343,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.727818853974122,
|
| 889 |
+
"grad_norm": 0.7130794525146484,
|
| 890 |
+
"learning_rate": 2.454222504621072e-05,
|
| 891 |
+
"loss": 0.2312,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.7335951940850277,
|
| 896 |
+
"grad_norm": 0.8044223785400391,
|
| 897 |
+
"learning_rate": 2.449890249537893e-05,
|
| 898 |
+
"loss": 0.1719,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.7393715341959335,
|
| 903 |
+
"grad_norm": 13.686079978942871,
|
| 904 |
+
"learning_rate": 2.4455579944547134e-05,
|
| 905 |
+
"loss": 0.2398,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.7451478743068392,
|
| 910 |
+
"grad_norm": 1.6030184030532837,
|
| 911 |
+
"learning_rate": 2.4412257393715343e-05,
|
| 912 |
+
"loss": 0.1675,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7509242144177449,
|
| 917 |
+
"grad_norm": 2.320791482925415,
|
| 918 |
+
"learning_rate": 2.436893484288355e-05,
|
| 919 |
+
"loss": 0.2505,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.7567005545286506,
|
| 924 |
+
"grad_norm": 12.586448669433594,
|
| 925 |
+
"learning_rate": 2.4325612292051755e-05,
|
| 926 |
+
"loss": 0.2716,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.7624768946395564,
|
| 931 |
+
"grad_norm": 7.049114227294922,
|
| 932 |
+
"learning_rate": 2.4282289741219965e-05,
|
| 933 |
+
"loss": 0.169,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.7682532347504621,
|
| 938 |
+
"grad_norm": 9.050933837890625,
|
| 939 |
+
"learning_rate": 2.423896719038817e-05,
|
| 940 |
+
"loss": 0.2116,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.7740295748613678,
|
| 945 |
+
"grad_norm": 25.830007553100586,
|
| 946 |
+
"learning_rate": 2.419564463955638e-05,
|
| 947 |
+
"loss": 0.2181,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.7798059149722736,
|
| 952 |
+
"grad_norm": 0.14128220081329346,
|
| 953 |
+
"learning_rate": 2.4152322088724583e-05,
|
| 954 |
+
"loss": 0.2531,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.7855822550831792,
|
| 959 |
+
"grad_norm": 0.03034038469195366,
|
| 960 |
+
"learning_rate": 2.4108999537892792e-05,
|
| 961 |
+
"loss": 0.2109,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.791358595194085,
|
| 966 |
+
"grad_norm": 0.1031632274389267,
|
| 967 |
+
"learning_rate": 2.4065676987061e-05,
|
| 968 |
+
"loss": 0.2155,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7971349353049908,
|
| 973 |
+
"grad_norm": 42.54078674316406,
|
| 974 |
+
"learning_rate": 2.4022354436229205e-05,
|
| 975 |
+
"loss": 0.183,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.8029112754158965,
|
| 980 |
+
"grad_norm": 0.03734962269663811,
|
| 981 |
+
"learning_rate": 2.3979031885397414e-05,
|
| 982 |
+
"loss": 0.185,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.8086876155268022,
|
| 987 |
+
"grad_norm": 48.300174713134766,
|
| 988 |
+
"learning_rate": 2.393570933456562e-05,
|
| 989 |
+
"loss": 0.1803,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.8144639556377079,
|
| 994 |
+
"grad_norm": 13.418463706970215,
|
| 995 |
+
"learning_rate": 2.389238678373383e-05,
|
| 996 |
+
"loss": 0.2226,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.8202402957486137,
|
| 1001 |
+
"grad_norm": 16.923093795776367,
|
| 1002 |
+
"learning_rate": 2.3849064232902032e-05,
|
| 1003 |
+
"loss": 0.2387,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.8260166358595195,
|
| 1008 |
+
"grad_norm": 37.3891487121582,
|
| 1009 |
+
"learning_rate": 2.380574168207024e-05,
|
| 1010 |
+
"loss": 0.2168,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8317929759704251,
|
| 1015 |
+
"grad_norm": 24.676448822021484,
|
| 1016 |
+
"learning_rate": 2.3762419131238448e-05,
|
| 1017 |
+
"loss": 0.202,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.8375693160813309,
|
| 1022 |
+
"grad_norm": 13.302814483642578,
|
| 1023 |
+
"learning_rate": 2.3719096580406654e-05,
|
| 1024 |
+
"loss": 0.2153,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8433456561922366,
|
| 1029 |
+
"grad_norm": 8.611613273620605,
|
| 1030 |
+
"learning_rate": 2.3675774029574864e-05,
|
| 1031 |
+
"loss": 0.2544,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.8491219963031423,
|
| 1036 |
+
"grad_norm": 0.4762769639492035,
|
| 1037 |
+
"learning_rate": 2.363245147874307e-05,
|
| 1038 |
+
"loss": 0.2129,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.8548983364140481,
|
| 1043 |
+
"grad_norm": 16.49268341064453,
|
| 1044 |
+
"learning_rate": 2.358912892791128e-05,
|
| 1045 |
+
"loss": 0.177,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.8606746765249538,
|
| 1050 |
+
"grad_norm": 2.05298113822937,
|
| 1051 |
+
"learning_rate": 2.3545806377079482e-05,
|
| 1052 |
+
"loss": 0.2274,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.8664510166358595,
|
| 1057 |
+
"grad_norm": 19.834819793701172,
|
| 1058 |
+
"learning_rate": 2.3502483826247688e-05,
|
| 1059 |
+
"loss": 0.1998,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.8722273567467652,
|
| 1064 |
+
"grad_norm": 9.706733703613281,
|
| 1065 |
+
"learning_rate": 2.3459161275415897e-05,
|
| 1066 |
+
"loss": 0.2563,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.878003696857671,
|
| 1071 |
+
"grad_norm": 9.801465034484863,
|
| 1072 |
+
"learning_rate": 2.3415838724584104e-05,
|
| 1073 |
+
"loss": 0.1453,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.8837800369685767,
|
| 1078 |
+
"grad_norm": 11.452254295349121,
|
| 1079 |
+
"learning_rate": 2.3372516173752313e-05,
|
| 1080 |
+
"loss": 0.272,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.8895563770794824,
|
| 1085 |
+
"grad_norm": 7.050422191619873,
|
| 1086 |
+
"learning_rate": 2.332919362292052e-05,
|
| 1087 |
+
"loss": 0.1713,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.8953327171903882,
|
| 1092 |
+
"grad_norm": 46.33341979980469,
|
| 1093 |
+
"learning_rate": 2.3285871072088725e-05,
|
| 1094 |
+
"loss": 0.1967,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.9011090573012939,
|
| 1099 |
+
"grad_norm": 1.1651716232299805,
|
| 1100 |
+
"learning_rate": 2.324254852125693e-05,
|
| 1101 |
+
"loss": 0.187,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.9068853974121996,
|
| 1106 |
+
"grad_norm": 0.3762068450450897,
|
| 1107 |
+
"learning_rate": 2.3199225970425137e-05,
|
| 1108 |
+
"loss": 0.2556,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.9126617375231053,
|
| 1113 |
+
"grad_norm": 9.550520896911621,
|
| 1114 |
+
"learning_rate": 2.3155903419593347e-05,
|
| 1115 |
+
"loss": 0.2413,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.9184380776340111,
|
| 1120 |
+
"grad_norm": 0.3446040451526642,
|
| 1121 |
+
"learning_rate": 2.3112580868761553e-05,
|
| 1122 |
+
"loss": 0.1694,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.9242144177449169,
|
| 1127 |
+
"grad_norm": 0.7896172404289246,
|
| 1128 |
+
"learning_rate": 2.3069258317929762e-05,
|
| 1129 |
+
"loss": 0.2504,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.9299907578558225,
|
| 1134 |
+
"grad_norm": 34.30472946166992,
|
| 1135 |
+
"learning_rate": 2.302593576709797e-05,
|
| 1136 |
+
"loss": 0.2462,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9357670979667283,
|
| 1141 |
+
"grad_norm": 31.708581924438477,
|
| 1142 |
+
"learning_rate": 2.298261321626617e-05,
|
| 1143 |
+
"loss": 0.147,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.941543438077634,
|
| 1148 |
+
"grad_norm": 2.5662238597869873,
|
| 1149 |
+
"learning_rate": 2.293929066543438e-05,
|
| 1150 |
+
"loss": 0.262,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.9473197781885397,
|
| 1155 |
+
"grad_norm": 0.19891509413719177,
|
| 1156 |
+
"learning_rate": 2.2895968114602587e-05,
|
| 1157 |
+
"loss": 0.1978,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.9530961182994455,
|
| 1162 |
+
"grad_norm": 81.1672134399414,
|
| 1163 |
+
"learning_rate": 2.2852645563770796e-05,
|
| 1164 |
+
"loss": 0.207,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.9588724584103512,
|
| 1169 |
+
"grad_norm": 20.577178955078125,
|
| 1170 |
+
"learning_rate": 2.2809323012939002e-05,
|
| 1171 |
+
"loss": 0.2256,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.964648798521257,
|
| 1176 |
+
"grad_norm": 0.11451568454504013,
|
| 1177 |
+
"learning_rate": 2.2766000462107212e-05,
|
| 1178 |
+
"loss": 0.2123,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.9704251386321626,
|
| 1183 |
+
"grad_norm": 2.9672398567199707,
|
| 1184 |
+
"learning_rate": 2.2722677911275418e-05,
|
| 1185 |
+
"loss": 0.2419,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.9762014787430684,
|
| 1190 |
+
"grad_norm": 3.283834934234619,
|
| 1191 |
+
"learning_rate": 2.267935536044362e-05,
|
| 1192 |
+
"loss": 0.1796,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9819778188539742,
|
| 1197 |
+
"grad_norm": 19.114147186279297,
|
| 1198 |
+
"learning_rate": 2.263603280961183e-05,
|
| 1199 |
+
"loss": 0.1789,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.9877541589648798,
|
| 1204 |
+
"grad_norm": 0.5701966881752014,
|
| 1205 |
+
"learning_rate": 2.2592710258780036e-05,
|
| 1206 |
+
"loss": 0.1616,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.9935304990757856,
|
| 1211 |
+
"grad_norm": 14.517534255981445,
|
| 1212 |
+
"learning_rate": 2.2549387707948246e-05,
|
| 1213 |
+
"loss": 0.1909,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.9993068391866913,
|
| 1218 |
+
"grad_norm": 0.1315908581018448,
|
| 1219 |
+
"learning_rate": 2.2506065157116452e-05,
|
| 1220 |
+
"loss": 0.2188,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.0,
|
| 1225 |
+
"eval_accuracy": 0.9437296946068876,
|
| 1226 |
+
"eval_f1": 0.9665611244111515,
|
| 1227 |
+
"eval_loss": 0.2036169469356537,
|
| 1228 |
+
"eval_precision": 0.953382084095064,
|
| 1229 |
+
"eval_recall": 0.9801096319498825,
|
| 1230 |
+
"eval_runtime": 88.7618,
|
| 1231 |
+
"eval_samples_per_second": 86.693,
|
| 1232 |
+
"eval_steps_per_second": 10.838,
|
| 1233 |
+
"step": 8656
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.005083179297597,
|
| 1237 |
+
"grad_norm": 9.673088073730469,
|
| 1238 |
+
"learning_rate": 2.246274260628466e-05,
|
| 1239 |
+
"loss": 0.1254,
|
| 1240 |
+
"step": 8700
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.0108595194085028,
|
| 1244 |
+
"grad_norm": 0.6131724119186401,
|
| 1245 |
+
"learning_rate": 2.2419420055452867e-05,
|
| 1246 |
+
"loss": 0.1205,
|
| 1247 |
+
"step": 8750
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.0166358595194085,
|
| 1251 |
+
"grad_norm": 0.12616665661334991,
|
| 1252 |
+
"learning_rate": 2.237609750462107e-05,
|
| 1253 |
+
"loss": 0.1409,
|
| 1254 |
+
"step": 8800
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.0224121996303142,
|
| 1258 |
+
"grad_norm": 0.18657980859279633,
|
| 1259 |
+
"learning_rate": 2.233277495378928e-05,
|
| 1260 |
+
"loss": 0.161,
|
| 1261 |
+
"step": 8850
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.02818853974122,
|
| 1265 |
+
"grad_norm": 0.051760233938694,
|
| 1266 |
+
"learning_rate": 2.2289452402957486e-05,
|
| 1267 |
+
"loss": 0.0927,
|
| 1268 |
+
"step": 8900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.0339648798521257,
|
| 1272 |
+
"grad_norm": 0.5276439785957336,
|
| 1273 |
+
"learning_rate": 2.2246129852125695e-05,
|
| 1274 |
+
"loss": 0.2098,
|
| 1275 |
+
"step": 8950
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.0397412199630314,
|
| 1279 |
+
"grad_norm": 3.536935567855835,
|
| 1280 |
+
"learning_rate": 2.22028073012939e-05,
|
| 1281 |
+
"loss": 0.164,
|
| 1282 |
+
"step": 9000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.0455175600739373,
|
| 1286 |
+
"grad_norm": 0.1915968954563141,
|
| 1287 |
+
"learning_rate": 2.2159484750462107e-05,
|
| 1288 |
+
"loss": 0.1839,
|
| 1289 |
+
"step": 9050
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.051293900184843,
|
| 1293 |
+
"grad_norm": 6.418234348297119,
|
| 1294 |
+
"learning_rate": 2.2116162199630314e-05,
|
| 1295 |
+
"loss": 0.1271,
|
| 1296 |
+
"step": 9100
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.0570702402957486,
|
| 1300 |
+
"grad_norm": 0.7983689904212952,
|
| 1301 |
+
"learning_rate": 2.207283964879852e-05,
|
| 1302 |
+
"loss": 0.1445,
|
| 1303 |
+
"step": 9150
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.0628465804066543,
|
| 1307 |
+
"grad_norm": 7.85959529876709,
|
| 1308 |
+
"learning_rate": 2.202951709796673e-05,
|
| 1309 |
+
"loss": 0.1613,
|
| 1310 |
+
"step": 9200
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.0686229205175601,
|
| 1314 |
+
"grad_norm": 0.041306737810373306,
|
| 1315 |
+
"learning_rate": 2.1986194547134935e-05,
|
| 1316 |
+
"loss": 0.1491,
|
| 1317 |
+
"step": 9250
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.0743992606284658,
|
| 1321 |
+
"grad_norm": 0.12126684933900833,
|
| 1322 |
+
"learning_rate": 2.1942871996303145e-05,
|
| 1323 |
+
"loss": 0.162,
|
| 1324 |
+
"step": 9300
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.0801756007393715,
|
| 1328 |
+
"grad_norm": 15.810729026794434,
|
| 1329 |
+
"learning_rate": 2.189954944547135e-05,
|
| 1330 |
+
"loss": 0.1271,
|
| 1331 |
+
"step": 9350
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.0859519408502774,
|
| 1335 |
+
"grad_norm": 0.04351761192083359,
|
| 1336 |
+
"learning_rate": 2.1856226894639557e-05,
|
| 1337 |
+
"loss": 0.1499,
|
| 1338 |
+
"step": 9400
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.091728280961183,
|
| 1342 |
+
"grad_norm": 49.74297332763672,
|
| 1343 |
+
"learning_rate": 2.1812904343807763e-05,
|
| 1344 |
+
"loss": 0.1617,
|
| 1345 |
+
"step": 9450
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0975046210720887,
|
| 1349 |
+
"grad_norm": 33.17013931274414,
|
| 1350 |
+
"learning_rate": 2.176958179297597e-05,
|
| 1351 |
+
"loss": 0.2043,
|
| 1352 |
+
"step": 9500
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.1032809611829943,
|
| 1356 |
+
"grad_norm": 1.7674719095230103,
|
| 1357 |
+
"learning_rate": 2.172625924214418e-05,
|
| 1358 |
+
"loss": 0.1275,
|
| 1359 |
+
"step": 9550
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.1090573012939002,
|
| 1363 |
+
"grad_norm": 0.13075052201747894,
|
| 1364 |
+
"learning_rate": 2.1682936691312385e-05,
|
| 1365 |
+
"loss": 0.117,
|
| 1366 |
+
"step": 9600
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.114833641404806,
|
| 1370 |
+
"grad_norm": 0.02289584092795849,
|
| 1371 |
+
"learning_rate": 2.1639614140480594e-05,
|
| 1372 |
+
"loss": 0.1096,
|
| 1373 |
+
"step": 9650
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.1206099815157116,
|
| 1377 |
+
"grad_norm": 7.883808612823486,
|
| 1378 |
+
"learning_rate": 2.15962915896488e-05,
|
| 1379 |
+
"loss": 0.1948,
|
| 1380 |
+
"step": 9700
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.1263863216266174,
|
| 1384 |
+
"grad_norm": 1.676134705543518,
|
| 1385 |
+
"learning_rate": 2.1552969038817006e-05,
|
| 1386 |
+
"loss": 0.155,
|
| 1387 |
+
"step": 9750
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.1321626617375231,
|
| 1391 |
+
"grad_norm": 0.06022998318076134,
|
| 1392 |
+
"learning_rate": 2.1509646487985212e-05,
|
| 1393 |
+
"loss": 0.1199,
|
| 1394 |
+
"step": 9800
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.1379390018484288,
|
| 1398 |
+
"grad_norm": 19.327392578125,
|
| 1399 |
+
"learning_rate": 2.146632393715342e-05,
|
| 1400 |
+
"loss": 0.21,
|
| 1401 |
+
"step": 9850
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1437153419593344,
|
| 1405 |
+
"grad_norm": 133.89877319335938,
|
| 1406 |
+
"learning_rate": 2.1423001386321628e-05,
|
| 1407 |
+
"loss": 0.1387,
|
| 1408 |
+
"step": 9900
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.1494916820702403,
|
| 1412 |
+
"grad_norm": 0.6451585292816162,
|
| 1413 |
+
"learning_rate": 2.1379678835489834e-05,
|
| 1414 |
+
"loss": 0.163,
|
| 1415 |
+
"step": 9950
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.155268022181146,
|
| 1419 |
+
"grad_norm": 0.17913977801799774,
|
| 1420 |
+
"learning_rate": 2.133635628465804e-05,
|
| 1421 |
+
"loss": 0.1653,
|
| 1422 |
+
"step": 10000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.1610443622920519,
|
| 1426 |
+
"grad_norm": 0.09651433676481247,
|
| 1427 |
+
"learning_rate": 2.129303373382625e-05,
|
| 1428 |
+
"loss": 0.1462,
|
| 1429 |
+
"step": 10050
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.1668207024029575,
|
| 1433 |
+
"grad_norm": 0.5200825929641724,
|
| 1434 |
+
"learning_rate": 2.1249711182994456e-05,
|
| 1435 |
+
"loss": 0.2055,
|
| 1436 |
+
"step": 10100
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.1725970425138632,
|
| 1440 |
+
"grad_norm": 0.2577333152294159,
|
| 1441 |
+
"learning_rate": 2.1206388632162662e-05,
|
| 1442 |
+
"loss": 0.1792,
|
| 1443 |
+
"step": 10150
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.1783733826247689,
|
| 1447 |
+
"grad_norm": 178.7727813720703,
|
| 1448 |
+
"learning_rate": 2.1163066081330868e-05,
|
| 1449 |
+
"loss": 0.0912,
|
| 1450 |
+
"step": 10200
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.1841497227356748,
|
| 1454 |
+
"grad_norm": 0.16386118531227112,
|
| 1455 |
+
"learning_rate": 2.1119743530499077e-05,
|
| 1456 |
+
"loss": 0.1288,
|
| 1457 |
+
"step": 10250
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.1899260628465804,
|
| 1461 |
+
"grad_norm": 5.637333869934082,
|
| 1462 |
+
"learning_rate": 2.1076420979667284e-05,
|
| 1463 |
+
"loss": 0.1978,
|
| 1464 |
+
"step": 10300
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.195702402957486,
|
| 1468 |
+
"grad_norm": 0.031639471650123596,
|
| 1469 |
+
"learning_rate": 2.103309842883549e-05,
|
| 1470 |
+
"loss": 0.1579,
|
| 1471 |
+
"step": 10350
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.201478743068392,
|
| 1475 |
+
"grad_norm": 7.582589626312256,
|
| 1476 |
+
"learning_rate": 2.09897758780037e-05,
|
| 1477 |
+
"loss": 0.1685,
|
| 1478 |
+
"step": 10400
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.2072550831792976,
|
| 1482 |
+
"grad_norm": 0.21556666493415833,
|
| 1483 |
+
"learning_rate": 2.0946453327171902e-05,
|
| 1484 |
+
"loss": 0.1794,
|
| 1485 |
+
"step": 10450
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.2130314232902033,
|
| 1489 |
+
"grad_norm": 11.476829528808594,
|
| 1490 |
+
"learning_rate": 2.090313077634011e-05,
|
| 1491 |
+
"loss": 0.1625,
|
| 1492 |
+
"step": 10500
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.218807763401109,
|
| 1496 |
+
"grad_norm": 24.46149253845215,
|
| 1497 |
+
"learning_rate": 2.0859808225508317e-05,
|
| 1498 |
+
"loss": 0.1383,
|
| 1499 |
+
"step": 10550
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.2245841035120149,
|
| 1503 |
+
"grad_norm": 26.288007736206055,
|
| 1504 |
+
"learning_rate": 2.0816485674676527e-05,
|
| 1505 |
+
"loss": 0.113,
|
| 1506 |
+
"step": 10600
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.2303604436229205,
|
| 1510 |
+
"grad_norm": 0.622348964214325,
|
| 1511 |
+
"learning_rate": 2.0773163123844733e-05,
|
| 1512 |
+
"loss": 0.17,
|
| 1513 |
+
"step": 10650
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2361367837338262,
|
| 1517 |
+
"grad_norm": 16.938920974731445,
|
| 1518 |
+
"learning_rate": 2.072984057301294e-05,
|
| 1519 |
+
"loss": 0.1494,
|
| 1520 |
+
"step": 10700
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.241913123844732,
|
| 1524 |
+
"grad_norm": 1.0484834909439087,
|
| 1525 |
+
"learning_rate": 2.068651802218115e-05,
|
| 1526 |
+
"loss": 0.1821,
|
| 1527 |
+
"step": 10750
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.2476894639556377,
|
| 1531 |
+
"grad_norm": 0.10622230172157288,
|
| 1532 |
+
"learning_rate": 2.064319547134935e-05,
|
| 1533 |
+
"loss": 0.1872,
|
| 1534 |
+
"step": 10800
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.2534658040665434,
|
| 1538 |
+
"grad_norm": 4.886632919311523,
|
| 1539 |
+
"learning_rate": 2.059987292051756e-05,
|
| 1540 |
+
"loss": 0.1208,
|
| 1541 |
+
"step": 10850
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.259242144177449,
|
| 1545 |
+
"grad_norm": 0.5296549201011658,
|
| 1546 |
+
"learning_rate": 2.0556550369685767e-05,
|
| 1547 |
+
"loss": 0.1579,
|
| 1548 |
+
"step": 10900
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.265018484288355,
|
| 1552 |
+
"grad_norm": 0.04283996298909187,
|
| 1553 |
+
"learning_rate": 2.0513227818853973e-05,
|
| 1554 |
+
"loss": 0.083,
|
| 1555 |
+
"step": 10950
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2707948243992606,
|
| 1559 |
+
"grad_norm": 1.309367060661316,
|
| 1560 |
+
"learning_rate": 2.0469905268022182e-05,
|
| 1561 |
+
"loss": 0.0815,
|
| 1562 |
+
"step": 11000
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2765711645101663,
|
| 1566 |
+
"grad_norm": 0.03423392027616501,
|
| 1567 |
+
"learning_rate": 2.042658271719039e-05,
|
| 1568 |
+
"loss": 0.0932,
|
| 1569 |
+
"step": 11050
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2823475046210722,
|
| 1573 |
+
"grad_norm": 0.028841083869338036,
|
| 1574 |
+
"learning_rate": 2.0383260166358598e-05,
|
| 1575 |
+
"loss": 0.1432,
|
| 1576 |
+
"step": 11100
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2881238447319778,
|
| 1580 |
+
"grad_norm": 0.0708237886428833,
|
| 1581 |
+
"learning_rate": 2.03399376155268e-05,
|
| 1582 |
+
"loss": 0.1607,
|
| 1583 |
+
"step": 11150
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2939001848428835,
|
| 1587 |
+
"grad_norm": 0.02011285535991192,
|
| 1588 |
+
"learning_rate": 2.029661506469501e-05,
|
| 1589 |
+
"loss": 0.2166,
|
| 1590 |
+
"step": 11200
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2996765249537892,
|
| 1594 |
+
"grad_norm": 0.16425344347953796,
|
| 1595 |
+
"learning_rate": 2.0253292513863216e-05,
|
| 1596 |
+
"loss": 0.1682,
|
| 1597 |
+
"step": 11250
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.305452865064695,
|
| 1601 |
+
"grad_norm": 10.606422424316406,
|
| 1602 |
+
"learning_rate": 2.0209969963031422e-05,
|
| 1603 |
+
"loss": 0.1247,
|
| 1604 |
+
"step": 11300
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.3112292051756007,
|
| 1608 |
+
"grad_norm": 0.39115816354751587,
|
| 1609 |
+
"learning_rate": 2.0166647412199632e-05,
|
| 1610 |
+
"loss": 0.1844,
|
| 1611 |
+
"step": 11350
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.3170055452865066,
|
| 1615 |
+
"grad_norm": 0.05713627114892006,
|
| 1616 |
+
"learning_rate": 2.0123324861367838e-05,
|
| 1617 |
+
"loss": 0.0753,
|
| 1618 |
+
"step": 11400
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.3227818853974123,
|
| 1622 |
+
"grad_norm": 57.468997955322266,
|
| 1623 |
+
"learning_rate": 2.0080002310536047e-05,
|
| 1624 |
+
"loss": 0.0336,
|
| 1625 |
+
"step": 11450
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.328558225508318,
|
| 1629 |
+
"grad_norm": 0.23353737592697144,
|
| 1630 |
+
"learning_rate": 2.003667975970425e-05,
|
| 1631 |
+
"loss": 0.2012,
|
| 1632 |
+
"step": 11500
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.3343345656192236,
|
| 1636 |
+
"grad_norm": 0.018346797674894333,
|
| 1637 |
+
"learning_rate": 1.999335720887246e-05,
|
| 1638 |
+
"loss": 0.1375,
|
| 1639 |
+
"step": 11550
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.3401109057301293,
|
| 1643 |
+
"grad_norm": 0.1295866072177887,
|
| 1644 |
+
"learning_rate": 1.9950034658040666e-05,
|
| 1645 |
+
"loss": 0.1389,
|
| 1646 |
+
"step": 11600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.3458872458410351,
|
| 1650 |
+
"grad_norm": 0.29036056995391846,
|
| 1651 |
+
"learning_rate": 1.9906712107208872e-05,
|
| 1652 |
+
"loss": 0.0943,
|
| 1653 |
+
"step": 11650
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.3516635859519408,
|
| 1657 |
+
"grad_norm": 0.007298491895198822,
|
| 1658 |
+
"learning_rate": 1.986338955637708e-05,
|
| 1659 |
+
"loss": 0.1572,
|
| 1660 |
+
"step": 11700
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.3574399260628467,
|
| 1664 |
+
"grad_norm": 0.39678987860679626,
|
| 1665 |
+
"learning_rate": 1.9820067005545287e-05,
|
| 1666 |
+
"loss": 0.1351,
|
| 1667 |
+
"step": 11750
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.3632162661737524,
|
| 1671 |
+
"grad_norm": 135.63607788085938,
|
| 1672 |
+
"learning_rate": 1.9776744454713497e-05,
|
| 1673 |
+
"loss": 0.1746,
|
| 1674 |
+
"step": 11800
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.368992606284658,
|
| 1678 |
+
"grad_norm": 12.449654579162598,
|
| 1679 |
+
"learning_rate": 1.97334219038817e-05,
|
| 1680 |
+
"loss": 0.1661,
|
| 1681 |
+
"step": 11850
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3747689463955637,
|
| 1685 |
+
"grad_norm": 1.2668341398239136,
|
| 1686 |
+
"learning_rate": 1.9690099353049906e-05,
|
| 1687 |
+
"loss": 0.1487,
|
| 1688 |
+
"step": 11900
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.3805452865064696,
|
| 1692 |
+
"grad_norm": 0.26599591970443726,
|
| 1693 |
+
"learning_rate": 1.9646776802218115e-05,
|
| 1694 |
+
"loss": 0.0866,
|
| 1695 |
+
"step": 11950
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.3863216266173752,
|
| 1699 |
+
"grad_norm": 5.264702796936035,
|
| 1700 |
+
"learning_rate": 1.960345425138632e-05,
|
| 1701 |
+
"loss": 0.2262,
|
| 1702 |
+
"step": 12000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.392097966728281,
|
| 1706 |
+
"grad_norm": 45.19207763671875,
|
| 1707 |
+
"learning_rate": 1.956013170055453e-05,
|
| 1708 |
+
"loss": 0.1587,
|
| 1709 |
+
"step": 12050
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 1.3978743068391868,
|
| 1713 |
+
"grad_norm": 7.993265151977539,
|
| 1714 |
+
"learning_rate": 1.9516809149722737e-05,
|
| 1715 |
+
"loss": 0.1042,
|
| 1716 |
+
"step": 12100
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.4036506469500925,
|
| 1720 |
+
"grad_norm": 0.07449086755514145,
|
| 1721 |
+
"learning_rate": 1.9473486598890943e-05,
|
| 1722 |
+
"loss": 0.1227,
|
| 1723 |
+
"step": 12150
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.4094269870609981,
|
| 1727 |
+
"grad_norm": 44.55666732788086,
|
| 1728 |
+
"learning_rate": 1.943016404805915e-05,
|
| 1729 |
+
"loss": 0.1825,
|
| 1730 |
+
"step": 12200
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 1.4152033271719038,
|
| 1734 |
+
"grad_norm": 0.17107024788856506,
|
| 1735 |
+
"learning_rate": 1.9386841497227355e-05,
|
| 1736 |
+
"loss": 0.1174,
|
| 1737 |
+
"step": 12250
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.4209796672828097,
|
| 1741 |
+
"grad_norm": 8.075175285339355,
|
| 1742 |
+
"learning_rate": 1.9343518946395565e-05,
|
| 1743 |
+
"loss": 0.1214,
|
| 1744 |
+
"step": 12300
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.4267560073937153,
|
| 1748 |
+
"grad_norm": 0.02626979723572731,
|
| 1749 |
+
"learning_rate": 1.930019639556377e-05,
|
| 1750 |
+
"loss": 0.1184,
|
| 1751 |
+
"step": 12350
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.432532347504621,
|
| 1755 |
+
"grad_norm": 0.04273197799921036,
|
| 1756 |
+
"learning_rate": 1.925687384473198e-05,
|
| 1757 |
+
"loss": 0.1599,
|
| 1758 |
+
"step": 12400
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.4383086876155269,
|
| 1762 |
+
"grad_norm": 11.037675857543945,
|
| 1763 |
+
"learning_rate": 1.9213551293900186e-05,
|
| 1764 |
+
"loss": 0.1972,
|
| 1765 |
+
"step": 12450
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.4440850277264325,
|
| 1769 |
+
"grad_norm": 0.018304361030459404,
|
| 1770 |
+
"learning_rate": 1.9170228743068392e-05,
|
| 1771 |
+
"loss": 0.109,
|
| 1772 |
+
"step": 12500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.4498613678373382,
|
| 1776 |
+
"grad_norm": 0.10881417989730835,
|
| 1777 |
+
"learning_rate": 1.91269061922366e-05,
|
| 1778 |
+
"loss": 0.1512,
|
| 1779 |
+
"step": 12550
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.4556377079482439,
|
| 1783 |
+
"grad_norm": 0.014431633055210114,
|
| 1784 |
+
"learning_rate": 1.9083583641404805e-05,
|
| 1785 |
+
"loss": 0.0693,
|
| 1786 |
+
"step": 12600
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.4614140480591498,
|
| 1790 |
+
"grad_norm": 4.677309513092041,
|
| 1791 |
+
"learning_rate": 1.9040261090573014e-05,
|
| 1792 |
+
"loss": 0.1791,
|
| 1793 |
+
"step": 12650
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4671903881700554,
|
| 1797 |
+
"grad_norm": 0.042188193649053574,
|
| 1798 |
+
"learning_rate": 1.899693853974122e-05,
|
| 1799 |
+
"loss": 0.1459,
|
| 1800 |
+
"step": 12700
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.4729667282809613,
|
| 1804 |
+
"grad_norm": 0.11454495042562485,
|
| 1805 |
+
"learning_rate": 1.895361598890943e-05,
|
| 1806 |
+
"loss": 0.1243,
|
| 1807 |
+
"step": 12750
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.478743068391867,
|
| 1811 |
+
"grad_norm": 40.89423370361328,
|
| 1812 |
+
"learning_rate": 1.8910293438077636e-05,
|
| 1813 |
+
"loss": 0.1451,
|
| 1814 |
+
"step": 12800
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.4845194085027726,
|
| 1818 |
+
"grad_norm": 0.022999301552772522,
|
| 1819 |
+
"learning_rate": 1.886697088724584e-05,
|
| 1820 |
+
"loss": 0.1615,
|
| 1821 |
+
"step": 12850
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.4902957486136783,
|
| 1825 |
+
"grad_norm": 0.023199884220957756,
|
| 1826 |
+
"learning_rate": 1.8823648336414048e-05,
|
| 1827 |
+
"loss": 0.1213,
|
| 1828 |
+
"step": 12900
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 1.496072088724584,
|
| 1832 |
+
"grad_norm": 26.06696319580078,
|
| 1833 |
+
"learning_rate": 1.8780325785582254e-05,
|
| 1834 |
+
"loss": 0.1625,
|
| 1835 |
+
"step": 12950
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 1.5018484288354899,
|
| 1839 |
+
"grad_norm": 7.6846442222595215,
|
| 1840 |
+
"learning_rate": 1.8737003234750464e-05,
|
| 1841 |
+
"loss": 0.168,
|
| 1842 |
+
"step": 13000
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.5076247689463955,
|
| 1846 |
+
"grad_norm": 0.05135953053832054,
|
| 1847 |
+
"learning_rate": 1.869368068391867e-05,
|
| 1848 |
+
"loss": 0.1258,
|
| 1849 |
+
"step": 13050
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.5134011090573014,
|
| 1853 |
+
"grad_norm": 1.4261308908462524,
|
| 1854 |
+
"learning_rate": 1.865035813308688e-05,
|
| 1855 |
+
"loss": 0.148,
|
| 1856 |
+
"step": 13100
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.519177449168207,
|
| 1860 |
+
"grad_norm": 1.1017619371414185,
|
| 1861 |
+
"learning_rate": 1.8607035582255085e-05,
|
| 1862 |
+
"loss": 0.1134,
|
| 1863 |
+
"step": 13150
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.5249537892791127,
|
| 1867 |
+
"grad_norm": 0.12042645364999771,
|
| 1868 |
+
"learning_rate": 1.8563713031423288e-05,
|
| 1869 |
+
"loss": 0.224,
|
| 1870 |
+
"step": 13200
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.5307301293900184,
|
| 1874 |
+
"grad_norm": 0.073084257543087,
|
| 1875 |
+
"learning_rate": 1.8520390480591497e-05,
|
| 1876 |
+
"loss": 0.0997,
|
| 1877 |
+
"step": 13250
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 1.536506469500924,
|
| 1881 |
+
"grad_norm": 12.779747009277344,
|
| 1882 |
+
"learning_rate": 1.8477067929759704e-05,
|
| 1883 |
+
"loss": 0.161,
|
| 1884 |
+
"step": 13300
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.54228280961183,
|
| 1888 |
+
"grad_norm": 0.04014035314321518,
|
| 1889 |
+
"learning_rate": 1.8433745378927913e-05,
|
| 1890 |
+
"loss": 0.1377,
|
| 1891 |
+
"step": 13350
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.5480591497227358,
|
| 1895 |
+
"grad_norm": 12.693160057067871,
|
| 1896 |
+
"learning_rate": 1.839042282809612e-05,
|
| 1897 |
+
"loss": 0.1347,
|
| 1898 |
+
"step": 13400
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 1.5538354898336415,
|
| 1902 |
+
"grad_norm": 0.43178144097328186,
|
| 1903 |
+
"learning_rate": 1.834710027726433e-05,
|
| 1904 |
+
"loss": 0.1023,
|
| 1905 |
+
"step": 13450
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5596118299445472,
|
| 1909 |
+
"grad_norm": 51.83779525756836,
|
| 1910 |
+
"learning_rate": 1.830377772643253e-05,
|
| 1911 |
+
"loss": 0.082,
|
| 1912 |
+
"step": 13500
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 1.5653881700554528,
|
| 1916 |
+
"grad_norm": 2.3788204193115234,
|
| 1917 |
+
"learning_rate": 1.8260455175600737e-05,
|
| 1918 |
+
"loss": 0.1606,
|
| 1919 |
+
"step": 13550
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 1.5711645101663585,
|
| 1923 |
+
"grad_norm": 0.03366752341389656,
|
| 1924 |
+
"learning_rate": 1.8217132624768947e-05,
|
| 1925 |
+
"loss": 0.0896,
|
| 1926 |
+
"step": 13600
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.5769408502772642,
|
| 1930 |
+
"grad_norm": 15.269149780273438,
|
| 1931 |
+
"learning_rate": 1.8173810073937153e-05,
|
| 1932 |
+
"loss": 0.1463,
|
| 1933 |
+
"step": 13650
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.58271719038817,
|
| 1937 |
+
"grad_norm": 34.45343780517578,
|
| 1938 |
+
"learning_rate": 1.8130487523105362e-05,
|
| 1939 |
+
"loss": 0.0908,
|
| 1940 |
+
"step": 13700
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 1.588493530499076,
|
| 1944 |
+
"grad_norm": 0.3380456268787384,
|
| 1945 |
+
"learning_rate": 1.808716497227357e-05,
|
| 1946 |
+
"loss": 0.0767,
|
| 1947 |
+
"step": 13750
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.5942698706099816,
|
| 1951 |
+
"grad_norm": 27.415430068969727,
|
| 1952 |
+
"learning_rate": 1.8043842421441775e-05,
|
| 1953 |
+
"loss": 0.185,
|
| 1954 |
+
"step": 13800
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 1.6000462107208873,
|
| 1958 |
+
"grad_norm": 59.60631561279297,
|
| 1959 |
+
"learning_rate": 1.800051987060998e-05,
|
| 1960 |
+
"loss": 0.1656,
|
| 1961 |
+
"step": 13850
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.605822550831793,
|
| 1965 |
+
"grad_norm": 30.945571899414062,
|
| 1966 |
+
"learning_rate": 1.7957197319778187e-05,
|
| 1967 |
+
"loss": 0.1029,
|
| 1968 |
+
"step": 13900
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.6115988909426986,
|
| 1972 |
+
"grad_norm": 0.09613625705242157,
|
| 1973 |
+
"learning_rate": 1.7913874768946396e-05,
|
| 1974 |
+
"loss": 0.155,
|
| 1975 |
+
"step": 13950
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.6173752310536045,
|
| 1979 |
+
"grad_norm": 0.5363235473632812,
|
| 1980 |
+
"learning_rate": 1.7870552218114602e-05,
|
| 1981 |
+
"loss": 0.1398,
|
| 1982 |
+
"step": 14000
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 1.6231515711645101,
|
| 1986 |
+
"grad_norm": 0.02551902085542679,
|
| 1987 |
+
"learning_rate": 1.7827229667282812e-05,
|
| 1988 |
+
"loss": 0.1039,
|
| 1989 |
+
"step": 14050
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.628927911275416,
|
| 1993 |
+
"grad_norm": 5.74197244644165,
|
| 1994 |
+
"learning_rate": 1.7783907116451018e-05,
|
| 1995 |
+
"loss": 0.1295,
|
| 1996 |
+
"step": 14100
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 1.6347042513863217,
|
| 2000 |
+
"grad_norm": 0.0256786085665226,
|
| 2001 |
+
"learning_rate": 1.7740584565619224e-05,
|
| 2002 |
+
"loss": 0.0745,
|
| 2003 |
+
"step": 14150
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 1.6404805914972274,
|
| 2007 |
+
"grad_norm": 0.015919797122478485,
|
| 2008 |
+
"learning_rate": 1.769726201478743e-05,
|
| 2009 |
+
"loss": 0.1619,
|
| 2010 |
+
"step": 14200
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.646256931608133,
|
| 2014 |
+
"grad_norm": 39.721153259277344,
|
| 2015 |
+
"learning_rate": 1.7653939463955636e-05,
|
| 2016 |
+
"loss": 0.0708,
|
| 2017 |
+
"step": 14250
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.6520332717190387,
|
| 2021 |
+
"grad_norm": 0.8398991227149963,
|
| 2022 |
+
"learning_rate": 1.7610616913123846e-05,
|
| 2023 |
+
"loss": 0.1335,
|
| 2024 |
+
"step": 14300
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 1.6578096118299446,
|
| 2028 |
+
"grad_norm": 0.9615886211395264,
|
| 2029 |
+
"learning_rate": 1.7567294362292052e-05,
|
| 2030 |
+
"loss": 0.129,
|
| 2031 |
+
"step": 14350
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.6635859519408502,
|
| 2035 |
+
"grad_norm": 0.22856245934963226,
|
| 2036 |
+
"learning_rate": 1.752397181146026e-05,
|
| 2037 |
+
"loss": 0.1506,
|
| 2038 |
+
"step": 14400
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 1.6693622920517561,
|
| 2042 |
+
"grad_norm": 0.24372538924217224,
|
| 2043 |
+
"learning_rate": 1.7480649260628467e-05,
|
| 2044 |
+
"loss": 0.1296,
|
| 2045 |
+
"step": 14450
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 1.6751386321626618,
|
| 2049 |
+
"grad_norm": 0.08972591161727905,
|
| 2050 |
+
"learning_rate": 1.7437326709796674e-05,
|
| 2051 |
+
"loss": 0.1519,
|
| 2052 |
+
"step": 14500
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.6809149722735675,
|
| 2056 |
+
"grad_norm": 0.21274013817310333,
|
| 2057 |
+
"learning_rate": 1.739400415896488e-05,
|
| 2058 |
+
"loss": 0.189,
|
| 2059 |
+
"step": 14550
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.6866913123844731,
|
| 2063 |
+
"grad_norm": 32.43384552001953,
|
| 2064 |
+
"learning_rate": 1.7350681608133086e-05,
|
| 2065 |
+
"loss": 0.0888,
|
| 2066 |
+
"step": 14600
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.6924676524953788,
|
| 2070 |
+
"grad_norm": 0.08137155324220657,
|
| 2071 |
+
"learning_rate": 1.7307359057301295e-05,
|
| 2072 |
+
"loss": 0.0572,
|
| 2073 |
+
"step": 14650
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6982439926062847,
|
| 2077 |
+
"grad_norm": 0.01811046339571476,
|
| 2078 |
+
"learning_rate": 1.72640365064695e-05,
|
| 2079 |
+
"loss": 0.1028,
|
| 2080 |
+
"step": 14700
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.7040203327171903,
|
| 2084 |
+
"grad_norm": 0.015499014407396317,
|
| 2085 |
+
"learning_rate": 1.7220713955637707e-05,
|
| 2086 |
+
"loss": 0.1315,
|
| 2087 |
+
"step": 14750
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.7097966728280962,
|
| 2091 |
+
"grad_norm": 8.582119941711426,
|
| 2092 |
+
"learning_rate": 1.7177391404805917e-05,
|
| 2093 |
+
"loss": 0.2206,
|
| 2094 |
+
"step": 14800
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.7155730129390019,
|
| 2098 |
+
"grad_norm": 7.51249361038208,
|
| 2099 |
+
"learning_rate": 1.713406885397412e-05,
|
| 2100 |
+
"loss": 0.1207,
|
| 2101 |
+
"step": 14850
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.7213493530499075,
|
| 2105 |
+
"grad_norm": 0.00726300198584795,
|
| 2106 |
+
"learning_rate": 1.709074630314233e-05,
|
| 2107 |
+
"loss": 0.1283,
|
| 2108 |
+
"step": 14900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.7271256931608132,
|
| 2112 |
+
"grad_norm": 0.07731953263282776,
|
| 2113 |
+
"learning_rate": 1.7047423752310535e-05,
|
| 2114 |
+
"loss": 0.1694,
|
| 2115 |
+
"step": 14950
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.7329020332717189,
|
| 2119 |
+
"grad_norm": 0.101409412920475,
|
| 2120 |
+
"learning_rate": 1.7004101201478745e-05,
|
| 2121 |
+
"loss": 0.0945,
|
| 2122 |
+
"step": 15000
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.7386783733826248,
|
| 2126 |
+
"grad_norm": 0.04303300008177757,
|
| 2127 |
+
"learning_rate": 1.696077865064695e-05,
|
| 2128 |
+
"loss": 0.1023,
|
| 2129 |
+
"step": 15050
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.7444547134935307,
|
| 2133 |
+
"grad_norm": 0.13228349387645721,
|
| 2134 |
+
"learning_rate": 1.6917456099815157e-05,
|
| 2135 |
+
"loss": 0.126,
|
| 2136 |
+
"step": 15100
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.7502310536044363,
|
| 2140 |
+
"grad_norm": 0.649356484413147,
|
| 2141 |
+
"learning_rate": 1.6874133548983366e-05,
|
| 2142 |
+
"loss": 0.1574,
|
| 2143 |
+
"step": 15150
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.756007393715342,
|
| 2147 |
+
"grad_norm": 16.686784744262695,
|
| 2148 |
+
"learning_rate": 1.683081099815157e-05,
|
| 2149 |
+
"loss": 0.1296,
|
| 2150 |
+
"step": 15200
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.7617837338262476,
|
| 2154 |
+
"grad_norm": 0.03777627646923065,
|
| 2155 |
+
"learning_rate": 1.678748844731978e-05,
|
| 2156 |
+
"loss": 0.0756,
|
| 2157 |
+
"step": 15250
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.7675600739371533,
|
| 2161 |
+
"grad_norm": 0.23532889783382416,
|
| 2162 |
+
"learning_rate": 1.6744165896487985e-05,
|
| 2163 |
+
"loss": 0.1009,
|
| 2164 |
+
"step": 15300
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.7733364140480592,
|
| 2168 |
+
"grad_norm": 18.59041976928711,
|
| 2169 |
+
"learning_rate": 1.6700843345656194e-05,
|
| 2170 |
+
"loss": 0.1599,
|
| 2171 |
+
"step": 15350
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.7791127541589649,
|
| 2175 |
+
"grad_norm": 0.11593695729970932,
|
| 2176 |
+
"learning_rate": 1.66575207948244e-05,
|
| 2177 |
+
"loss": 0.1245,
|
| 2178 |
+
"step": 15400
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.7848890942698707,
|
| 2182 |
+
"grad_norm": 0.07317894697189331,
|
| 2183 |
+
"learning_rate": 1.6614198243992606e-05,
|
| 2184 |
+
"loss": 0.1822,
|
| 2185 |
+
"step": 15450
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7906654343807764,
|
| 2189 |
+
"grad_norm": 0.06145332753658295,
|
| 2190 |
+
"learning_rate": 1.6570875693160816e-05,
|
| 2191 |
+
"loss": 0.1925,
|
| 2192 |
+
"step": 15500
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.796441774491682,
|
| 2196 |
+
"grad_norm": 13.491717338562012,
|
| 2197 |
+
"learning_rate": 1.652755314232902e-05,
|
| 2198 |
+
"loss": 0.1854,
|
| 2199 |
+
"step": 15550
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.8022181146025877,
|
| 2203 |
+
"grad_norm": 0.1411893367767334,
|
| 2204 |
+
"learning_rate": 1.6484230591497228e-05,
|
| 2205 |
+
"loss": 0.1126,
|
| 2206 |
+
"step": 15600
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.8079944547134934,
|
| 2210 |
+
"grad_norm": 10.62831974029541,
|
| 2211 |
+
"learning_rate": 1.6440908040665434e-05,
|
| 2212 |
+
"loss": 0.094,
|
| 2213 |
+
"step": 15650
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.8137707948243993,
|
| 2217 |
+
"grad_norm": 0.01941312849521637,
|
| 2218 |
+
"learning_rate": 1.6397585489833644e-05,
|
| 2219 |
+
"loss": 0.136,
|
| 2220 |
+
"step": 15700
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.819547134935305,
|
| 2224 |
+
"grad_norm": 30.944677352905273,
|
| 2225 |
+
"learning_rate": 1.635426293900185e-05,
|
| 2226 |
+
"loss": 0.1317,
|
| 2227 |
+
"step": 15750
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.8253234750462108,
|
| 2231 |
+
"grad_norm": 16.877334594726562,
|
| 2232 |
+
"learning_rate": 1.6310940388170056e-05,
|
| 2233 |
+
"loss": 0.133,
|
| 2234 |
+
"step": 15800
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.8310998151571165,
|
| 2238 |
+
"grad_norm": 0.1656600683927536,
|
| 2239 |
+
"learning_rate": 1.6267617837338265e-05,
|
| 2240 |
+
"loss": 0.1395,
|
| 2241 |
+
"step": 15850
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.8368761552680222,
|
| 2245 |
+
"grad_norm": 53.970794677734375,
|
| 2246 |
+
"learning_rate": 1.6224295286506468e-05,
|
| 2247 |
+
"loss": 0.1337,
|
| 2248 |
+
"step": 15900
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.8426524953789278,
|
| 2252 |
+
"grad_norm": 36.164947509765625,
|
| 2253 |
+
"learning_rate": 1.6180972735674677e-05,
|
| 2254 |
+
"loss": 0.1251,
|
| 2255 |
+
"step": 15950
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.8484288354898335,
|
| 2259 |
+
"grad_norm": 0.0839138776063919,
|
| 2260 |
+
"learning_rate": 1.6137650184842884e-05,
|
| 2261 |
+
"loss": 0.0661,
|
| 2262 |
+
"step": 16000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.8542051756007394,
|
| 2266 |
+
"grad_norm": 0.1801111400127411,
|
| 2267 |
+
"learning_rate": 1.609432763401109e-05,
|
| 2268 |
+
"loss": 0.1557,
|
| 2269 |
+
"step": 16050
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.859981515711645,
|
| 2273 |
+
"grad_norm": 0.12647980451583862,
|
| 2274 |
+
"learning_rate": 1.60510050831793e-05,
|
| 2275 |
+
"loss": 0.0765,
|
| 2276 |
+
"step": 16100
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.865757855822551,
|
| 2280 |
+
"grad_norm": 0.26282384991645813,
|
| 2281 |
+
"learning_rate": 1.6007682532347505e-05,
|
| 2282 |
+
"loss": 0.0413,
|
| 2283 |
+
"step": 16150
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.8715341959334566,
|
| 2287 |
+
"grad_norm": 0.0952085480093956,
|
| 2288 |
+
"learning_rate": 1.596435998151571e-05,
|
| 2289 |
+
"loss": 0.1286,
|
| 2290 |
+
"step": 16200
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.8773105360443623,
|
| 2294 |
+
"grad_norm": 0.11837054044008255,
|
| 2295 |
+
"learning_rate": 1.5921037430683917e-05,
|
| 2296 |
+
"loss": 0.1891,
|
| 2297 |
+
"step": 16250
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.883086876155268,
|
| 2301 |
+
"grad_norm": 0.053552981466054916,
|
| 2302 |
+
"learning_rate": 1.5877714879852127e-05,
|
| 2303 |
+
"loss": 0.1535,
|
| 2304 |
+
"step": 16300
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.8888632162661736,
|
| 2308 |
+
"grad_norm": 1.4103118181228638,
|
| 2309 |
+
"learning_rate": 1.5834392329020333e-05,
|
| 2310 |
+
"loss": 0.1414,
|
| 2311 |
+
"step": 16350
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.8946395563770795,
|
| 2315 |
+
"grad_norm": 0.07249405235052109,
|
| 2316 |
+
"learning_rate": 1.579106977818854e-05,
|
| 2317 |
+
"loss": 0.1125,
|
| 2318 |
+
"step": 16400
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.9004158964879854,
|
| 2322 |
+
"grad_norm": 0.016298620030283928,
|
| 2323 |
+
"learning_rate": 1.574774722735675e-05,
|
| 2324 |
+
"loss": 0.1219,
|
| 2325 |
+
"step": 16450
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.906192236598891,
|
| 2329 |
+
"grad_norm": 0.025717375800013542,
|
| 2330 |
+
"learning_rate": 1.5704424676524955e-05,
|
| 2331 |
+
"loss": 0.1248,
|
| 2332 |
+
"step": 16500
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.9119685767097967,
|
| 2336 |
+
"grad_norm": 0.07495362311601639,
|
| 2337 |
+
"learning_rate": 1.566110212569316e-05,
|
| 2338 |
+
"loss": 0.1399,
|
| 2339 |
+
"step": 16550
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.9177449168207024,
|
| 2343 |
+
"grad_norm": 0.08207421749830246,
|
| 2344 |
+
"learning_rate": 1.5617779574861367e-05,
|
| 2345 |
+
"loss": 0.1092,
|
| 2346 |
+
"step": 16600
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.923521256931608,
|
| 2350 |
+
"grad_norm": 0.02017366513609886,
|
| 2351 |
+
"learning_rate": 1.5574457024029576e-05,
|
| 2352 |
+
"loss": 0.1164,
|
| 2353 |
+
"step": 16650
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.929297597042514,
|
| 2357 |
+
"grad_norm": 0.10131692886352539,
|
| 2358 |
+
"learning_rate": 1.5531134473197782e-05,
|
| 2359 |
+
"loss": 0.1416,
|
| 2360 |
+
"step": 16700
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.9350739371534196,
|
| 2364 |
+
"grad_norm": 0.03260243684053421,
|
| 2365 |
+
"learning_rate": 1.548781192236599e-05,
|
| 2366 |
+
"loss": 0.0695,
|
| 2367 |
+
"step": 16750
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.9408502772643255,
|
| 2371 |
+
"grad_norm": 0.17720632255077362,
|
| 2372 |
+
"learning_rate": 1.5444489371534198e-05,
|
| 2373 |
+
"loss": 0.1201,
|
| 2374 |
+
"step": 16800
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.9466266173752311,
|
| 2378 |
+
"grad_norm": 3.364078998565674,
|
| 2379 |
+
"learning_rate": 1.5401166820702404e-05,
|
| 2380 |
+
"loss": 0.1684,
|
| 2381 |
+
"step": 16850
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.9524029574861368,
|
| 2385 |
+
"grad_norm": 31.63196563720703,
|
| 2386 |
+
"learning_rate": 1.535784426987061e-05,
|
| 2387 |
+
"loss": 0.0964,
|
| 2388 |
+
"step": 16900
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.9581792975970425,
|
| 2392 |
+
"grad_norm": 0.049181848764419556,
|
| 2393 |
+
"learning_rate": 1.5314521719038816e-05,
|
| 2394 |
+
"loss": 0.1232,
|
| 2395 |
+
"step": 16950
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.9639556377079481,
|
| 2399 |
+
"grad_norm": 0.027907520532608032,
|
| 2400 |
+
"learning_rate": 1.5271199168207022e-05,
|
| 2401 |
+
"loss": 0.1164,
|
| 2402 |
+
"step": 17000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.969731977818854,
|
| 2406 |
+
"grad_norm": 0.030601385980844498,
|
| 2407 |
+
"learning_rate": 1.5227876617375232e-05,
|
| 2408 |
+
"loss": 0.0783,
|
| 2409 |
+
"step": 17050
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.9755083179297597,
|
| 2413 |
+
"grad_norm": 0.009862957522273064,
|
| 2414 |
+
"learning_rate": 1.518455406654344e-05,
|
| 2415 |
+
"loss": 0.1063,
|
| 2416 |
+
"step": 17100
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.9812846580406656,
|
| 2420 |
+
"grad_norm": 40.9315185546875,
|
| 2421 |
+
"learning_rate": 1.5141231515711646e-05,
|
| 2422 |
+
"loss": 0.1464,
|
| 2423 |
+
"step": 17150
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.9870609981515712,
|
| 2427 |
+
"grad_norm": 0.12488950043916702,
|
| 2428 |
+
"learning_rate": 1.5097908964879854e-05,
|
| 2429 |
+
"loss": 0.1145,
|
| 2430 |
+
"step": 17200
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.9928373382624769,
|
| 2434 |
+
"grad_norm": 0.08856427669525146,
|
| 2435 |
+
"learning_rate": 1.5054586414048058e-05,
|
| 2436 |
+
"loss": 0.0934,
|
| 2437 |
+
"step": 17250
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.9986136783733826,
|
| 2441 |
+
"grad_norm": 0.046317312866449356,
|
| 2442 |
+
"learning_rate": 1.5011263863216266e-05,
|
| 2443 |
+
"loss": 0.0957,
|
| 2444 |
+
"step": 17300
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 2.0,
|
| 2448 |
+
"eval_accuracy": 0.9604938271604938,
|
| 2449 |
+
"eval_f1": 0.9761680777673252,
|
| 2450 |
+
"eval_loss": 0.17032423615455627,
|
| 2451 |
+
"eval_precision": 0.9772406215664731,
|
| 2452 |
+
"eval_recall": 0.975097885669538,
|
| 2453 |
+
"eval_runtime": 82.0507,
|
| 2454 |
+
"eval_samples_per_second": 93.784,
|
| 2455 |
+
"eval_steps_per_second": 11.724,
|
| 2456 |
+
"step": 17312
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 2.004390018484288,
|
| 2460 |
+
"grad_norm": 0.010689891874790192,
|
| 2461 |
+
"learning_rate": 1.4967941312384474e-05,
|
| 2462 |
+
"loss": 0.0559,
|
| 2463 |
+
"step": 17350
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 2.010166358595194,
|
| 2467 |
+
"grad_norm": 0.15493936836719513,
|
| 2468 |
+
"learning_rate": 1.4924618761552681e-05,
|
| 2469 |
+
"loss": 0.0919,
|
| 2470 |
+
"step": 17400
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 2.0159426987061,
|
| 2474 |
+
"grad_norm": 0.014357605017721653,
|
| 2475 |
+
"learning_rate": 1.4881296210720889e-05,
|
| 2476 |
+
"loss": 0.0245,
|
| 2477 |
+
"step": 17450
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 2.0217190388170057,
|
| 2481 |
+
"grad_norm": 0.006949929054826498,
|
| 2482 |
+
"learning_rate": 1.4837973659889094e-05,
|
| 2483 |
+
"loss": 0.0562,
|
| 2484 |
+
"step": 17500
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 2.0274953789279113,
|
| 2488 |
+
"grad_norm": 0.1050797700881958,
|
| 2489 |
+
"learning_rate": 1.4794651109057301e-05,
|
| 2490 |
+
"loss": 0.0552,
|
| 2491 |
+
"step": 17550
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 2.033271719038817,
|
| 2495 |
+
"grad_norm": 0.014158029109239578,
|
| 2496 |
+
"learning_rate": 1.4751328558225509e-05,
|
| 2497 |
+
"loss": 0.0901,
|
| 2498 |
+
"step": 17600
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 2.0390480591497226,
|
| 2502 |
+
"grad_norm": 0.006835776846855879,
|
| 2503 |
+
"learning_rate": 1.4708006007393715e-05,
|
| 2504 |
+
"loss": 0.0597,
|
| 2505 |
+
"step": 17650
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 2.0448243992606283,
|
| 2509 |
+
"grad_norm": 4.837308883666992,
|
| 2510 |
+
"learning_rate": 1.4664683456561923e-05,
|
| 2511 |
+
"loss": 0.1171,
|
| 2512 |
+
"step": 17700
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 2.050600739371534,
|
| 2516 |
+
"grad_norm": 0.024250244721770287,
|
| 2517 |
+
"learning_rate": 1.462136090573013e-05,
|
| 2518 |
+
"loss": 0.032,
|
| 2519 |
+
"step": 17750
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 2.05637707948244,
|
| 2523 |
+
"grad_norm": 17.85401725769043,
|
| 2524 |
+
"learning_rate": 1.4578038354898337e-05,
|
| 2525 |
+
"loss": 0.0685,
|
| 2526 |
+
"step": 17800
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 2.0621534195933457,
|
| 2530 |
+
"grad_norm": 0.050390977412462234,
|
| 2531 |
+
"learning_rate": 1.4534715804066543e-05,
|
| 2532 |
+
"loss": 0.0449,
|
| 2533 |
+
"step": 17850
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 2.0679297597042514,
|
| 2537 |
+
"grad_norm": 12.766210556030273,
|
| 2538 |
+
"learning_rate": 1.449139325323475e-05,
|
| 2539 |
+
"loss": 0.0793,
|
| 2540 |
+
"step": 17900
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 2.073706099815157,
|
| 2544 |
+
"grad_norm": 0.037205442786216736,
|
| 2545 |
+
"learning_rate": 1.4448070702402959e-05,
|
| 2546 |
+
"loss": 0.0333,
|
| 2547 |
+
"step": 17950
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 2.0794824399260627,
|
| 2551 |
+
"grad_norm": 0.024253813549876213,
|
| 2552 |
+
"learning_rate": 1.4404748151571165e-05,
|
| 2553 |
+
"loss": 0.0432,
|
| 2554 |
+
"step": 18000
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 2.0852587800369684,
|
| 2558 |
+
"grad_norm": 0.00925503484904766,
|
| 2559 |
+
"learning_rate": 1.4361425600739372e-05,
|
| 2560 |
+
"loss": 0.0259,
|
| 2561 |
+
"step": 18050
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 2.0910351201478745,
|
| 2565 |
+
"grad_norm": 0.02291133813560009,
|
| 2566 |
+
"learning_rate": 1.4318103049907579e-05,
|
| 2567 |
+
"loss": 0.0492,
|
| 2568 |
+
"step": 18100
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 2.09681146025878,
|
| 2572 |
+
"grad_norm": 0.5501829385757446,
|
| 2573 |
+
"learning_rate": 1.4274780499075785e-05,
|
| 2574 |
+
"loss": 0.1352,
|
| 2575 |
+
"step": 18150
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 2.102587800369686,
|
| 2579 |
+
"grad_norm": 0.002355934353545308,
|
| 2580 |
+
"learning_rate": 1.4231457948243992e-05,
|
| 2581 |
+
"loss": 0.1025,
|
| 2582 |
+
"step": 18200
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 2.1083641404805915,
|
| 2586 |
+
"grad_norm": 0.014874827116727829,
|
| 2587 |
+
"learning_rate": 1.41881353974122e-05,
|
| 2588 |
+
"loss": 0.0386,
|
| 2589 |
+
"step": 18250
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 2.114140480591497,
|
| 2593 |
+
"grad_norm": 93.17355346679688,
|
| 2594 |
+
"learning_rate": 1.4144812846580408e-05,
|
| 2595 |
+
"loss": 0.064,
|
| 2596 |
+
"step": 18300
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 2.119916820702403,
|
| 2600 |
+
"grad_norm": 0.0061296261847019196,
|
| 2601 |
+
"learning_rate": 1.4101490295748614e-05,
|
| 2602 |
+
"loss": 0.0205,
|
| 2603 |
+
"step": 18350
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 2.1256931608133085,
|
| 2607 |
+
"grad_norm": 0.008473536930978298,
|
| 2608 |
+
"learning_rate": 1.4058167744916822e-05,
|
| 2609 |
+
"loss": 0.0517,
|
| 2610 |
+
"step": 18400
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 2.1314695009242146,
|
| 2614 |
+
"grad_norm": 0.006634963676333427,
|
| 2615 |
+
"learning_rate": 1.4014845194085028e-05,
|
| 2616 |
+
"loss": 0.079,
|
| 2617 |
+
"step": 18450
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 2.1372458410351203,
|
| 2621 |
+
"grad_norm": 0.02181815542280674,
|
| 2622 |
+
"learning_rate": 1.3971522643253234e-05,
|
| 2623 |
+
"loss": 0.0546,
|
| 2624 |
+
"step": 18500
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 2.143022181146026,
|
| 2628 |
+
"grad_norm": 0.09393370896577835,
|
| 2629 |
+
"learning_rate": 1.3928200092421442e-05,
|
| 2630 |
+
"loss": 0.0716,
|
| 2631 |
+
"step": 18550
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 2.1487985212569316,
|
| 2635 |
+
"grad_norm": 0.10325607657432556,
|
| 2636 |
+
"learning_rate": 1.388487754158965e-05,
|
| 2637 |
+
"loss": 0.0524,
|
| 2638 |
+
"step": 18600
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 2.1545748613678373,
|
| 2642 |
+
"grad_norm": 0.038211192935705185,
|
| 2643 |
+
"learning_rate": 1.3841554990757856e-05,
|
| 2644 |
+
"loss": 0.1085,
|
| 2645 |
+
"step": 18650
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 2.160351201478743,
|
| 2649 |
+
"grad_norm": 0.04504028707742691,
|
| 2650 |
+
"learning_rate": 1.3798232439926064e-05,
|
| 2651 |
+
"loss": 0.0521,
|
| 2652 |
+
"step": 18700
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 2.1661275415896486,
|
| 2656 |
+
"grad_norm": 0.00031169046997092664,
|
| 2657 |
+
"learning_rate": 1.375490988909427e-05,
|
| 2658 |
+
"loss": 0.0376,
|
| 2659 |
+
"step": 18750
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 2.1719038817005547,
|
| 2663 |
+
"grad_norm": 28.617237091064453,
|
| 2664 |
+
"learning_rate": 1.3711587338262477e-05,
|
| 2665 |
+
"loss": 0.1085,
|
| 2666 |
+
"step": 18800
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 2.1776802218114604,
|
| 2670 |
+
"grad_norm": 0.04509668052196503,
|
| 2671 |
+
"learning_rate": 1.3668264787430684e-05,
|
| 2672 |
+
"loss": 0.0846,
|
| 2673 |
+
"step": 18850
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 2.183456561922366,
|
| 2677 |
+
"grad_norm": 113.54573059082031,
|
| 2678 |
+
"learning_rate": 1.3624942236598891e-05,
|
| 2679 |
+
"loss": 0.0688,
|
| 2680 |
+
"step": 18900
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 2.1892329020332717,
|
| 2684 |
+
"grad_norm": 0.01831943169236183,
|
| 2685 |
+
"learning_rate": 1.3581619685767099e-05,
|
| 2686 |
+
"loss": 0.0987,
|
| 2687 |
+
"step": 18950
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 2.1950092421441774,
|
| 2691 |
+
"grad_norm": 0.01064152829349041,
|
| 2692 |
+
"learning_rate": 1.3538297134935305e-05,
|
| 2693 |
+
"loss": 0.0441,
|
| 2694 |
+
"step": 19000
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 2.200785582255083,
|
| 2698 |
+
"grad_norm": 0.015401902608573437,
|
| 2699 |
+
"learning_rate": 1.3494974584103511e-05,
|
| 2700 |
+
"loss": 0.0537,
|
| 2701 |
+
"step": 19050
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 2.2065619223659887,
|
| 2705 |
+
"grad_norm": 0.009121007286012173,
|
| 2706 |
+
"learning_rate": 1.3451652033271719e-05,
|
| 2707 |
+
"loss": 0.0661,
|
| 2708 |
+
"step": 19100
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 2.212338262476895,
|
| 2712 |
+
"grad_norm": 7.185537815093994,
|
| 2713 |
+
"learning_rate": 1.3408329482439927e-05,
|
| 2714 |
+
"loss": 0.0464,
|
| 2715 |
+
"step": 19150
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 2.2181146025878005,
|
| 2719 |
+
"grad_norm": 0.0032783944625407457,
|
| 2720 |
+
"learning_rate": 1.3365006931608133e-05,
|
| 2721 |
+
"loss": 0.0403,
|
| 2722 |
+
"step": 19200
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 2.223890942698706,
|
| 2726 |
+
"grad_norm": 0.32057854533195496,
|
| 2727 |
+
"learning_rate": 1.332168438077634e-05,
|
| 2728 |
+
"loss": 0.0623,
|
| 2729 |
+
"step": 19250
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 2.229667282809612,
|
| 2733 |
+
"grad_norm": 20.42328453063965,
|
| 2734 |
+
"learning_rate": 1.3278361829944549e-05,
|
| 2735 |
+
"loss": 0.0556,
|
| 2736 |
+
"step": 19300
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 2.2354436229205175,
|
| 2740 |
+
"grad_norm": 0.038709014654159546,
|
| 2741 |
+
"learning_rate": 1.3235039279112755e-05,
|
| 2742 |
+
"loss": 0.0637,
|
| 2743 |
+
"step": 19350
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 2.241219963031423,
|
| 2747 |
+
"grad_norm": 0.004040204919874668,
|
| 2748 |
+
"learning_rate": 1.319171672828096e-05,
|
| 2749 |
+
"loss": 0.0568,
|
| 2750 |
+
"step": 19400
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 2.2469963031423292,
|
| 2754 |
+
"grad_norm": 0.025795022025704384,
|
| 2755 |
+
"learning_rate": 1.3148394177449169e-05,
|
| 2756 |
+
"loss": 0.0916,
|
| 2757 |
+
"step": 19450
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 2.252772643253235,
|
| 2761 |
+
"grad_norm": 0.2631891667842865,
|
| 2762 |
+
"learning_rate": 1.3105071626617375e-05,
|
| 2763 |
+
"loss": 0.0514,
|
| 2764 |
+
"step": 19500
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 2.2585489833641406,
|
| 2768 |
+
"grad_norm": 1.6524279117584229,
|
| 2769 |
+
"learning_rate": 1.3061749075785582e-05,
|
| 2770 |
+
"loss": 0.0351,
|
| 2771 |
+
"step": 19550
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 2.2643253234750462,
|
| 2775 |
+
"grad_norm": 0.05630010738968849,
|
| 2776 |
+
"learning_rate": 1.301842652495379e-05,
|
| 2777 |
+
"loss": 0.0939,
|
| 2778 |
+
"step": 19600
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 2.270101663585952,
|
| 2782 |
+
"grad_norm": 0.03012896329164505,
|
| 2783 |
+
"learning_rate": 1.2975103974121998e-05,
|
| 2784 |
+
"loss": 0.0375,
|
| 2785 |
+
"step": 19650
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 2.2758780036968576,
|
| 2789 |
+
"grad_norm": 0.01212468184530735,
|
| 2790 |
+
"learning_rate": 1.2931781423290202e-05,
|
| 2791 |
+
"loss": 0.0776,
|
| 2792 |
+
"step": 19700
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 2.281654343807763,
|
| 2796 |
+
"grad_norm": 0.13953787088394165,
|
| 2797 |
+
"learning_rate": 1.288845887245841e-05,
|
| 2798 |
+
"loss": 0.0636,
|
| 2799 |
+
"step": 19750
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 2.287430683918669,
|
| 2803 |
+
"grad_norm": 0.023758836090564728,
|
| 2804 |
+
"learning_rate": 1.2845136321626618e-05,
|
| 2805 |
+
"loss": 0.052,
|
| 2806 |
+
"step": 19800
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 2.293207024029575,
|
| 2810 |
+
"grad_norm": 144.3390655517578,
|
| 2811 |
+
"learning_rate": 1.2801813770794824e-05,
|
| 2812 |
+
"loss": 0.0618,
|
| 2813 |
+
"step": 19850
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 2.2989833641404807,
|
| 2817 |
+
"grad_norm": 0.05545844882726669,
|
| 2818 |
+
"learning_rate": 1.2758491219963032e-05,
|
| 2819 |
+
"loss": 0.0571,
|
| 2820 |
+
"step": 19900
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 2.3047597042513863,
|
| 2824 |
+
"grad_norm": 0.005504699889570475,
|
| 2825 |
+
"learning_rate": 1.271516866913124e-05,
|
| 2826 |
+
"loss": 0.0263,
|
| 2827 |
+
"step": 19950
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 2.310536044362292,
|
| 2831 |
+
"grad_norm": 0.042020104825496674,
|
| 2832 |
+
"learning_rate": 1.2671846118299446e-05,
|
| 2833 |
+
"loss": 0.0563,
|
| 2834 |
+
"step": 20000
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 2.3163123844731976,
|
| 2838 |
+
"grad_norm": 0.0003972934791818261,
|
| 2839 |
+
"learning_rate": 1.2628523567467652e-05,
|
| 2840 |
+
"loss": 0.0775,
|
| 2841 |
+
"step": 20050
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 2.3220887245841038,
|
| 2845 |
+
"grad_norm": 0.001207977533340454,
|
| 2846 |
+
"learning_rate": 1.258520101663586e-05,
|
| 2847 |
+
"loss": 0.024,
|
| 2848 |
+
"step": 20100
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 2.3278650646950094,
|
| 2852 |
+
"grad_norm": 0.002152041532099247,
|
| 2853 |
+
"learning_rate": 1.2541878465804067e-05,
|
| 2854 |
+
"loss": 0.0356,
|
| 2855 |
+
"step": 20150
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 2.333641404805915,
|
| 2859 |
+
"grad_norm": 0.001504444400779903,
|
| 2860 |
+
"learning_rate": 1.2498555914972274e-05,
|
| 2861 |
+
"loss": 0.0316,
|
| 2862 |
+
"step": 20200
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 2.3394177449168208,
|
| 2866 |
+
"grad_norm": 0.012808242812752724,
|
| 2867 |
+
"learning_rate": 1.2455233364140481e-05,
|
| 2868 |
+
"loss": 0.0541,
|
| 2869 |
+
"step": 20250
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 2.3451940850277264,
|
| 2873 |
+
"grad_norm": 0.012331448495388031,
|
| 2874 |
+
"learning_rate": 1.2411910813308689e-05,
|
| 2875 |
+
"loss": 0.0435,
|
| 2876 |
+
"step": 20300
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 2.350970425138632,
|
| 2880 |
+
"grad_norm": 0.0008015843341127038,
|
| 2881 |
+
"learning_rate": 1.2368588262476894e-05,
|
| 2882 |
+
"loss": 0.0813,
|
| 2883 |
+
"step": 20350
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 2.3567467652495377,
|
| 2887 |
+
"grad_norm": 0.14125843346118927,
|
| 2888 |
+
"learning_rate": 1.2325265711645101e-05,
|
| 2889 |
+
"loss": 0.0488,
|
| 2890 |
+
"step": 20400
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 2.3625231053604434,
|
| 2894 |
+
"grad_norm": 0.05792843922972679,
|
| 2895 |
+
"learning_rate": 1.2281943160813309e-05,
|
| 2896 |
+
"loss": 0.111,
|
| 2897 |
+
"step": 20450
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 2.3682994454713495,
|
| 2901 |
+
"grad_norm": 0.04284418746829033,
|
| 2902 |
+
"learning_rate": 1.2238620609981517e-05,
|
| 2903 |
+
"loss": 0.0527,
|
| 2904 |
+
"step": 20500
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 2.374075785582255,
|
| 2908 |
+
"grad_norm": 0.037994951009750366,
|
| 2909 |
+
"learning_rate": 1.2195298059149723e-05,
|
| 2910 |
+
"loss": 0.0419,
|
| 2911 |
+
"step": 20550
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 2.379852125693161,
|
| 2915 |
+
"grad_norm": 101.88446044921875,
|
| 2916 |
+
"learning_rate": 1.215197550831793e-05,
|
| 2917 |
+
"loss": 0.0815,
|
| 2918 |
+
"step": 20600
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 2.3856284658040665,
|
| 2922 |
+
"grad_norm": 0.01886908710002899,
|
| 2923 |
+
"learning_rate": 1.2108652957486137e-05,
|
| 2924 |
+
"loss": 0.0608,
|
| 2925 |
+
"step": 20650
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 2.391404805914972,
|
| 2929 |
+
"grad_norm": 9.171727180480957,
|
| 2930 |
+
"learning_rate": 1.2065330406654343e-05,
|
| 2931 |
+
"loss": 0.0985,
|
| 2932 |
+
"step": 20700
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 2.397181146025878,
|
| 2936 |
+
"grad_norm": 0.003489085705950856,
|
| 2937 |
+
"learning_rate": 1.202200785582255e-05,
|
| 2938 |
+
"loss": 0.025,
|
| 2939 |
+
"step": 20750
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 2.402957486136784,
|
| 2943 |
+
"grad_norm": 0.01171598769724369,
|
| 2944 |
+
"learning_rate": 1.1978685304990759e-05,
|
| 2945 |
+
"loss": 0.0479,
|
| 2946 |
+
"step": 20800
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 2.4087338262476896,
|
| 2950 |
+
"grad_norm": 32.78422546386719,
|
| 2951 |
+
"learning_rate": 1.1935362754158965e-05,
|
| 2952 |
+
"loss": 0.0185,
|
| 2953 |
+
"step": 20850
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 2.4145101663585953,
|
| 2957 |
+
"grad_norm": 0.015714552253484726,
|
| 2958 |
+
"learning_rate": 1.1892040203327172e-05,
|
| 2959 |
+
"loss": 0.0467,
|
| 2960 |
+
"step": 20900
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 2.420286506469501,
|
| 2964 |
+
"grad_norm": 0.01492072083055973,
|
| 2965 |
+
"learning_rate": 1.184871765249538e-05,
|
| 2966 |
+
"loss": 0.0495,
|
| 2967 |
+
"step": 20950
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 2.4260628465804066,
|
| 2971 |
+
"grad_norm": 0.0005918367533013225,
|
| 2972 |
+
"learning_rate": 1.1805395101663586e-05,
|
| 2973 |
+
"loss": 0.0785,
|
| 2974 |
+
"step": 21000
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 2.4318391866913123,
|
| 2978 |
+
"grad_norm": 0.014696589671075344,
|
| 2979 |
+
"learning_rate": 1.1762072550831792e-05,
|
| 2980 |
+
"loss": 0.0539,
|
| 2981 |
+
"step": 21050
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 2.437615526802218,
|
| 2985 |
+
"grad_norm": 0.08432668447494507,
|
| 2986 |
+
"learning_rate": 1.171875e-05,
|
| 2987 |
+
"loss": 0.106,
|
| 2988 |
+
"step": 21100
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 2.4433918669131236,
|
| 2992 |
+
"grad_norm": 0.017786169424653053,
|
| 2993 |
+
"learning_rate": 1.1675427449168208e-05,
|
| 2994 |
+
"loss": 0.0836,
|
| 2995 |
+
"step": 21150
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 2.4491682070240297,
|
| 2999 |
+
"grad_norm": 0.013631800189614296,
|
| 3000 |
+
"learning_rate": 1.1632104898336414e-05,
|
| 3001 |
+
"loss": 0.0745,
|
| 3002 |
+
"step": 21200
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 2.4549445471349354,
|
| 3006 |
+
"grad_norm": 0.047913454473018646,
|
| 3007 |
+
"learning_rate": 1.1588782347504622e-05,
|
| 3008 |
+
"loss": 0.0421,
|
| 3009 |
+
"step": 21250
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 2.460720887245841,
|
| 3013 |
+
"grad_norm": 0.01846926100552082,
|
| 3014 |
+
"learning_rate": 1.1545459796672828e-05,
|
| 3015 |
+
"loss": 0.0843,
|
| 3016 |
+
"step": 21300
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 2.4664972273567467,
|
| 3020 |
+
"grad_norm": 0.7885268926620483,
|
| 3021 |
+
"learning_rate": 1.1502137245841036e-05,
|
| 3022 |
+
"loss": 0.072,
|
| 3023 |
+
"step": 21350
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 2.4722735674676524,
|
| 3027 |
+
"grad_norm": 0.0016458783065900207,
|
| 3028 |
+
"learning_rate": 1.1458814695009242e-05,
|
| 3029 |
+
"loss": 0.0486,
|
| 3030 |
+
"step": 21400
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 2.4780499075785585,
|
| 3034 |
+
"grad_norm": 0.00023497416987083852,
|
| 3035 |
+
"learning_rate": 1.141549214417745e-05,
|
| 3036 |
+
"loss": 0.0222,
|
| 3037 |
+
"step": 21450
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 2.483826247689464,
|
| 3041 |
+
"grad_norm": 0.01331646554172039,
|
| 3042 |
+
"learning_rate": 1.1372169593345657e-05,
|
| 3043 |
+
"loss": 0.0232,
|
| 3044 |
+
"step": 21500
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 2.48960258780037,
|
| 3048 |
+
"grad_norm": 0.016502482816576958,
|
| 3049 |
+
"learning_rate": 1.1328847042513864e-05,
|
| 3050 |
+
"loss": 0.0378,
|
| 3051 |
+
"step": 21550
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 2.4953789279112755,
|
| 3055 |
+
"grad_norm": 0.004157706163823605,
|
| 3056 |
+
"learning_rate": 1.128552449168207e-05,
|
| 3057 |
+
"loss": 0.0447,
|
| 3058 |
+
"step": 21600
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 2.501155268022181,
|
| 3062 |
+
"grad_norm": 0.31779831647872925,
|
| 3063 |
+
"learning_rate": 1.1242201940850277e-05,
|
| 3064 |
+
"loss": 0.0634,
|
| 3065 |
+
"step": 21650
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 2.506931608133087,
|
| 3069 |
+
"grad_norm": 0.012037420645356178,
|
| 3070 |
+
"learning_rate": 1.1198879390018484e-05,
|
| 3071 |
+
"loss": 0.0773,
|
| 3072 |
+
"step": 21700
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 2.5127079482439925,
|
| 3076 |
+
"grad_norm": 0.05561776086688042,
|
| 3077 |
+
"learning_rate": 1.1155556839186691e-05,
|
| 3078 |
+
"loss": 0.121,
|
| 3079 |
+
"step": 21750
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 2.518484288354898,
|
| 3083 |
+
"grad_norm": 0.057203542441129684,
|
| 3084 |
+
"learning_rate": 1.1112234288354899e-05,
|
| 3085 |
+
"loss": 0.0713,
|
| 3086 |
+
"step": 21800
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 2.524260628465804,
|
| 3090 |
+
"grad_norm": 0.17984868586063385,
|
| 3091 |
+
"learning_rate": 1.1068911737523107e-05,
|
| 3092 |
+
"loss": 0.0387,
|
| 3093 |
+
"step": 21850
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 2.53003696857671,
|
| 3097 |
+
"grad_norm": 91.4612045288086,
|
| 3098 |
+
"learning_rate": 1.1025589186691313e-05,
|
| 3099 |
+
"loss": 0.1062,
|
| 3100 |
+
"step": 21900
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 2.5358133086876156,
|
| 3104 |
+
"grad_norm": 2.4073524475097656,
|
| 3105 |
+
"learning_rate": 1.0982266635859519e-05,
|
| 3106 |
+
"loss": 0.0562,
|
| 3107 |
+
"step": 21950
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 2.5415896487985212,
|
| 3111 |
+
"grad_norm": 0.07512298971414566,
|
| 3112 |
+
"learning_rate": 1.0938944085027727e-05,
|
| 3113 |
+
"loss": 0.0864,
|
| 3114 |
+
"step": 22000
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 2.547365988909427,
|
| 3118 |
+
"grad_norm": 0.010843496769666672,
|
| 3119 |
+
"learning_rate": 1.0895621534195933e-05,
|
| 3120 |
+
"loss": 0.092,
|
| 3121 |
+
"step": 22050
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 2.5531423290203326,
|
| 3125 |
+
"grad_norm": 0.011628352105617523,
|
| 3126 |
+
"learning_rate": 1.085229898336414e-05,
|
| 3127 |
+
"loss": 0.0244,
|
| 3128 |
+
"step": 22100
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 2.5589186691312387,
|
| 3132 |
+
"grad_norm": 0.1271040290594101,
|
| 3133 |
+
"learning_rate": 1.0808976432532349e-05,
|
| 3134 |
+
"loss": 0.0393,
|
| 3135 |
+
"step": 22150
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 2.5646950092421443,
|
| 3139 |
+
"grad_norm": 0.006007444113492966,
|
| 3140 |
+
"learning_rate": 1.0765653881700555e-05,
|
| 3141 |
+
"loss": 0.0582,
|
| 3142 |
+
"step": 22200
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 2.57047134935305,
|
| 3146 |
+
"grad_norm": 0.005085831508040428,
|
| 3147 |
+
"learning_rate": 1.072233133086876e-05,
|
| 3148 |
+
"loss": 0.0369,
|
| 3149 |
+
"step": 22250
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 2.5762476894639557,
|
| 3153 |
+
"grad_norm": 0.31836622953414917,
|
| 3154 |
+
"learning_rate": 1.0679008780036969e-05,
|
| 3155 |
+
"loss": 0.0811,
|
| 3156 |
+
"step": 22300
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 2.5820240295748613,
|
| 3160 |
+
"grad_norm": 0.03294687718153,
|
| 3161 |
+
"learning_rate": 1.0635686229205176e-05,
|
| 3162 |
+
"loss": 0.0559,
|
| 3163 |
+
"step": 22350
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 2.587800369685767,
|
| 3167 |
+
"grad_norm": 0.030545327812433243,
|
| 3168 |
+
"learning_rate": 1.0592363678373382e-05,
|
| 3169 |
+
"loss": 0.0688,
|
| 3170 |
+
"step": 22400
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 2.5935767097966727,
|
| 3174 |
+
"grad_norm": 0.01783604547381401,
|
| 3175 |
+
"learning_rate": 1.054904112754159e-05,
|
| 3176 |
+
"loss": 0.06,
|
| 3177 |
+
"step": 22450
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 2.5993530499075783,
|
| 3181 |
+
"grad_norm": 0.03546810895204544,
|
| 3182 |
+
"learning_rate": 1.0505718576709798e-05,
|
| 3183 |
+
"loss": 0.0423,
|
| 3184 |
+
"step": 22500
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 2.6051293900184844,
|
| 3188 |
+
"grad_norm": 0.2595711648464203,
|
| 3189 |
+
"learning_rate": 1.0462396025878002e-05,
|
| 3190 |
+
"loss": 0.0256,
|
| 3191 |
+
"step": 22550
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 2.61090573012939,
|
| 3195 |
+
"grad_norm": 0.0062712496146559715,
|
| 3196 |
+
"learning_rate": 1.041907347504621e-05,
|
| 3197 |
+
"loss": 0.0655,
|
| 3198 |
+
"step": 22600
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 2.6166820702402958,
|
| 3202 |
+
"grad_norm": 0.038008201867341995,
|
| 3203 |
+
"learning_rate": 1.0375750924214418e-05,
|
| 3204 |
+
"loss": 0.0503,
|
| 3205 |
+
"step": 22650
|
| 3206 |
+
},
|
| 3207 |
+
{
|
| 3208 |
+
"epoch": 2.6224584103512014,
|
| 3209 |
+
"grad_norm": 0.046623073518276215,
|
| 3210 |
+
"learning_rate": 1.0332428373382626e-05,
|
| 3211 |
+
"loss": 0.0576,
|
| 3212 |
+
"step": 22700
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 2.628234750462107,
|
| 3216 |
+
"grad_norm": 0.002310172189027071,
|
| 3217 |
+
"learning_rate": 1.0289105822550832e-05,
|
| 3218 |
+
"loss": 0.0164,
|
| 3219 |
+
"step": 22750
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 2.634011090573013,
|
| 3223 |
+
"grad_norm": 63.63881301879883,
|
| 3224 |
+
"learning_rate": 1.024578327171904e-05,
|
| 3225 |
+
"loss": 0.094,
|
| 3226 |
+
"step": 22800
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 2.639787430683919,
|
| 3230 |
+
"grad_norm": 0.043101947754621506,
|
| 3231 |
+
"learning_rate": 1.0202460720887247e-05,
|
| 3232 |
+
"loss": 0.0458,
|
| 3233 |
+
"step": 22850
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 2.6455637707948245,
|
| 3237 |
+
"grad_norm": 13.507847785949707,
|
| 3238 |
+
"learning_rate": 1.0159138170055452e-05,
|
| 3239 |
+
"loss": 0.0508,
|
| 3240 |
+
"step": 22900
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 2.65134011090573,
|
| 3244 |
+
"grad_norm": 0.025238962844014168,
|
| 3245 |
+
"learning_rate": 1.011581561922366e-05,
|
| 3246 |
+
"loss": 0.0308,
|
| 3247 |
+
"step": 22950
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 2.657116451016636,
|
| 3251 |
+
"grad_norm": 0.0416383296251297,
|
| 3252 |
+
"learning_rate": 1.0072493068391867e-05,
|
| 3253 |
+
"loss": 0.0313,
|
| 3254 |
+
"step": 23000
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 2.6628927911275415,
|
| 3258 |
+
"grad_norm": 0.007924668490886688,
|
| 3259 |
+
"learning_rate": 1.0029170517560074e-05,
|
| 3260 |
+
"loss": 0.1056,
|
| 3261 |
+
"step": 23050
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 2.668669131238447,
|
| 3265 |
+
"grad_norm": 0.05370543152093887,
|
| 3266 |
+
"learning_rate": 9.985847966728281e-06,
|
| 3267 |
+
"loss": 0.0838,
|
| 3268 |
+
"step": 23100
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 2.674445471349353,
|
| 3272 |
+
"grad_norm": 0.02265593782067299,
|
| 3273 |
+
"learning_rate": 9.94252541589649e-06,
|
| 3274 |
+
"loss": 0.0418,
|
| 3275 |
+
"step": 23150
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 2.6802218114602585,
|
| 3279 |
+
"grad_norm": 0.04637107253074646,
|
| 3280 |
+
"learning_rate": 9.899202865064695e-06,
|
| 3281 |
+
"loss": 0.0349,
|
| 3282 |
+
"step": 23200
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 2.6859981515711646,
|
| 3286 |
+
"grad_norm": 20.716609954833984,
|
| 3287 |
+
"learning_rate": 9.855880314232901e-06,
|
| 3288 |
+
"loss": 0.082,
|
| 3289 |
+
"step": 23250
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 2.6917744916820703,
|
| 3293 |
+
"grad_norm": 0.4123234450817108,
|
| 3294 |
+
"learning_rate": 9.812557763401109e-06,
|
| 3295 |
+
"loss": 0.0414,
|
| 3296 |
+
"step": 23300
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 2.697550831792976,
|
| 3300 |
+
"grad_norm": 0.005975351203233004,
|
| 3301 |
+
"learning_rate": 9.769235212569317e-06,
|
| 3302 |
+
"loss": 0.0172,
|
| 3303 |
+
"step": 23350
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 2.7033271719038816,
|
| 3307 |
+
"grad_norm": 0.0020372685976326466,
|
| 3308 |
+
"learning_rate": 9.725912661737523e-06,
|
| 3309 |
+
"loss": 0.0367,
|
| 3310 |
+
"step": 23400
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 2.7091035120147873,
|
| 3314 |
+
"grad_norm": 0.01890731044113636,
|
| 3315 |
+
"learning_rate": 9.68259011090573e-06,
|
| 3316 |
+
"loss": 0.0644,
|
| 3317 |
+
"step": 23450
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 2.7148798521256934,
|
| 3321 |
+
"grad_norm": 0.011325077153742313,
|
| 3322 |
+
"learning_rate": 9.639267560073937e-06,
|
| 3323 |
+
"loss": 0.0887,
|
| 3324 |
+
"step": 23500
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 2.720656192236599,
|
| 3328 |
+
"grad_norm": 0.00467151403427124,
|
| 3329 |
+
"learning_rate": 9.595945009242145e-06,
|
| 3330 |
+
"loss": 0.0153,
|
| 3331 |
+
"step": 23550
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 2.7264325323475047,
|
| 3335 |
+
"grad_norm": 0.0022690685000270605,
|
| 3336 |
+
"learning_rate": 9.55262245841035e-06,
|
| 3337 |
+
"loss": 0.0114,
|
| 3338 |
+
"step": 23600
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 2.7322088724584104,
|
| 3342 |
+
"grad_norm": 0.005233801435679197,
|
| 3343 |
+
"learning_rate": 9.509299907578559e-06,
|
| 3344 |
+
"loss": 0.102,
|
| 3345 |
+
"step": 23650
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 2.737985212569316,
|
| 3349 |
+
"grad_norm": 0.08544773608446121,
|
| 3350 |
+
"learning_rate": 9.465977356746766e-06,
|
| 3351 |
+
"loss": 0.1023,
|
| 3352 |
+
"step": 23700
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 2.7437615526802217,
|
| 3356 |
+
"grad_norm": 0.02068651281297207,
|
| 3357 |
+
"learning_rate": 9.422654805914972e-06,
|
| 3358 |
+
"loss": 0.0265,
|
| 3359 |
+
"step": 23750
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 2.7495378927911274,
|
| 3363 |
+
"grad_norm": 0.032986097037792206,
|
| 3364 |
+
"learning_rate": 9.37933225508318e-06,
|
| 3365 |
+
"loss": 0.1026,
|
| 3366 |
+
"step": 23800
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 2.755314232902033,
|
| 3370 |
+
"grad_norm": 0.10801597684621811,
|
| 3371 |
+
"learning_rate": 9.336009704251386e-06,
|
| 3372 |
+
"loss": 0.117,
|
| 3373 |
+
"step": 23850
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 2.761090573012939,
|
| 3377 |
+
"grad_norm": 0.0526568703353405,
|
| 3378 |
+
"learning_rate": 9.292687153419592e-06,
|
| 3379 |
+
"loss": 0.0624,
|
| 3380 |
+
"step": 23900
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 2.766866913123845,
|
| 3384 |
+
"grad_norm": 0.11655089259147644,
|
| 3385 |
+
"learning_rate": 9.2493646025878e-06,
|
| 3386 |
+
"loss": 0.0486,
|
| 3387 |
+
"step": 23950
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 2.7726432532347505,
|
| 3391 |
+
"grad_norm": 0.0014089143369346857,
|
| 3392 |
+
"learning_rate": 9.206042051756008e-06,
|
| 3393 |
+
"loss": 0.0263,
|
| 3394 |
+
"step": 24000
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 2.778419593345656,
|
| 3398 |
+
"grad_norm": 19.42731475830078,
|
| 3399 |
+
"learning_rate": 9.162719500924216e-06,
|
| 3400 |
+
"loss": 0.0801,
|
| 3401 |
+
"step": 24050
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 2.784195933456562,
|
| 3405 |
+
"grad_norm": 0.0423482209444046,
|
| 3406 |
+
"learning_rate": 9.119396950092422e-06,
|
| 3407 |
+
"loss": 0.0462,
|
| 3408 |
+
"step": 24100
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 2.789972273567468,
|
| 3412 |
+
"grad_norm": 22.931161880493164,
|
| 3413 |
+
"learning_rate": 9.076074399260628e-06,
|
| 3414 |
+
"loss": 0.0445,
|
| 3415 |
+
"step": 24150
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 2.7957486136783736,
|
| 3419 |
+
"grad_norm": 109.42991638183594,
|
| 3420 |
+
"learning_rate": 9.032751848428836e-06,
|
| 3421 |
+
"loss": 0.054,
|
| 3422 |
+
"step": 24200
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 2.8015249537892792,
|
| 3426 |
+
"grad_norm": 85.00433349609375,
|
| 3427 |
+
"learning_rate": 8.989429297597042e-06,
|
| 3428 |
+
"loss": 0.0481,
|
| 3429 |
+
"step": 24250
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 2.807301293900185,
|
| 3433 |
+
"grad_norm": 0.012196958996355534,
|
| 3434 |
+
"learning_rate": 8.94610674676525e-06,
|
| 3435 |
+
"loss": 0.0188,
|
| 3436 |
+
"step": 24300
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 2.8130776340110906,
|
| 3440 |
+
"grad_norm": 0.013218875043094158,
|
| 3441 |
+
"learning_rate": 8.902784195933457e-06,
|
| 3442 |
+
"loss": 0.0305,
|
| 3443 |
+
"step": 24350
|
| 3444 |
+
},
|
| 3445 |
+
{
|
| 3446 |
+
"epoch": 2.8188539741219962,
|
| 3447 |
+
"grad_norm": 0.006864851340651512,
|
| 3448 |
+
"learning_rate": 8.859461645101664e-06,
|
| 3449 |
+
"loss": 0.0315,
|
| 3450 |
+
"step": 24400
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 2.824630314232902,
|
| 3454 |
+
"grad_norm": 0.2585645020008087,
|
| 3455 |
+
"learning_rate": 8.81613909426987e-06,
|
| 3456 |
+
"loss": 0.0491,
|
| 3457 |
+
"step": 24450
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 2.8304066543438076,
|
| 3461 |
+
"grad_norm": 3.8999698162078857,
|
| 3462 |
+
"learning_rate": 8.772816543438077e-06,
|
| 3463 |
+
"loss": 0.1083,
|
| 3464 |
+
"step": 24500
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 2.8361829944547132,
|
| 3468 |
+
"grad_norm": 0.002783197211101651,
|
| 3469 |
+
"learning_rate": 8.729493992606285e-06,
|
| 3470 |
+
"loss": 0.0327,
|
| 3471 |
+
"step": 24550
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 2.8419593345656193,
|
| 3475 |
+
"grad_norm": 0.009121415205299854,
|
| 3476 |
+
"learning_rate": 8.686171441774491e-06,
|
| 3477 |
+
"loss": 0.0444,
|
| 3478 |
+
"step": 24600
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 2.847735674676525,
|
| 3482 |
+
"grad_norm": 48.94571304321289,
|
| 3483 |
+
"learning_rate": 8.642848890942699e-06,
|
| 3484 |
+
"loss": 0.047,
|
| 3485 |
+
"step": 24650
|
| 3486 |
+
},
|
| 3487 |
+
{
|
| 3488 |
+
"epoch": 2.8535120147874307,
|
| 3489 |
+
"grad_norm": 0.04149790480732918,
|
| 3490 |
+
"learning_rate": 8.599526340110907e-06,
|
| 3491 |
+
"loss": 0.0342,
|
| 3492 |
+
"step": 24700
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 2.8592883548983363,
|
| 3496 |
+
"grad_norm": 0.0969071164727211,
|
| 3497 |
+
"learning_rate": 8.556203789279113e-06,
|
| 3498 |
+
"loss": 0.1272,
|
| 3499 |
+
"step": 24750
|
| 3500 |
+
},
|
| 3501 |
+
{
|
| 3502 |
+
"epoch": 2.865064695009242,
|
| 3503 |
+
"grad_norm": 0.009497321210801601,
|
| 3504 |
+
"learning_rate": 8.512881238447319e-06,
|
| 3505 |
+
"loss": 0.042,
|
| 3506 |
+
"step": 24800
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 2.870841035120148,
|
| 3510 |
+
"grad_norm": 0.022364236414432526,
|
| 3511 |
+
"learning_rate": 8.469558687615527e-06,
|
| 3512 |
+
"loss": 0.0097,
|
| 3513 |
+
"step": 24850
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 2.8766173752310538,
|
| 3517 |
+
"grad_norm": 0.017840340733528137,
|
| 3518 |
+
"learning_rate": 8.426236136783735e-06,
|
| 3519 |
+
"loss": 0.0734,
|
| 3520 |
+
"step": 24900
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 2.8823937153419594,
|
| 3524 |
+
"grad_norm": 0.022637976333498955,
|
| 3525 |
+
"learning_rate": 8.38291358595194e-06,
|
| 3526 |
+
"loss": 0.0727,
|
| 3527 |
+
"step": 24950
|
| 3528 |
+
},
|
| 3529 |
+
{
|
| 3530 |
+
"epoch": 2.888170055452865,
|
| 3531 |
+
"grad_norm": 0.00045255915028974414,
|
| 3532 |
+
"learning_rate": 8.339591035120149e-06,
|
| 3533 |
+
"loss": 0.0495,
|
| 3534 |
+
"step": 25000
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 2.8939463955637708,
|
| 3538 |
+
"grad_norm": 0.46494194865226746,
|
| 3539 |
+
"learning_rate": 8.296268484288356e-06,
|
| 3540 |
+
"loss": 0.0365,
|
| 3541 |
+
"step": 25050
|
| 3542 |
+
},
|
| 3543 |
+
{
|
| 3544 |
+
"epoch": 2.8997227356746764,
|
| 3545 |
+
"grad_norm": 0.01695355772972107,
|
| 3546 |
+
"learning_rate": 8.25294593345656e-06,
|
| 3547 |
+
"loss": 0.0256,
|
| 3548 |
+
"step": 25100
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 2.905499075785582,
|
| 3552 |
+
"grad_norm": 0.039756178855895996,
|
| 3553 |
+
"learning_rate": 8.209623382624769e-06,
|
| 3554 |
+
"loss": 0.0358,
|
| 3555 |
+
"step": 25150
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 2.9112754158964878,
|
| 3559 |
+
"grad_norm": 0.019751282408833504,
|
| 3560 |
+
"learning_rate": 8.166300831792976e-06,
|
| 3561 |
+
"loss": 0.0075,
|
| 3562 |
+
"step": 25200
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 2.917051756007394,
|
| 3566 |
+
"grad_norm": 0.02498270757496357,
|
| 3567 |
+
"learning_rate": 8.122978280961182e-06,
|
| 3568 |
+
"loss": 0.0468,
|
| 3569 |
+
"step": 25250
|
| 3570 |
+
},
|
| 3571 |
+
{
|
| 3572 |
+
"epoch": 2.9228280961182995,
|
| 3573 |
+
"grad_norm": 0.0017788989935070276,
|
| 3574 |
+
"learning_rate": 8.07965573012939e-06,
|
| 3575 |
+
"loss": 0.0479,
|
| 3576 |
+
"step": 25300
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 2.928604436229205,
|
| 3580 |
+
"grad_norm": 0.05598069354891777,
|
| 3581 |
+
"learning_rate": 8.036333179297598e-06,
|
| 3582 |
+
"loss": 0.0781,
|
| 3583 |
+
"step": 25350
|
| 3584 |
+
},
|
| 3585 |
+
{
|
| 3586 |
+
"epoch": 2.934380776340111,
|
| 3587 |
+
"grad_norm": 0.018725041300058365,
|
| 3588 |
+
"learning_rate": 7.993010628465806e-06,
|
| 3589 |
+
"loss": 0.0196,
|
| 3590 |
+
"step": 25400
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 2.9401571164510165,
|
| 3594 |
+
"grad_norm": 0.0007029447588138282,
|
| 3595 |
+
"learning_rate": 7.94968807763401e-06,
|
| 3596 |
+
"loss": 0.0826,
|
| 3597 |
+
"step": 25450
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 2.9459334565619226,
|
| 3601 |
+
"grad_norm": 0.05901116877794266,
|
| 3602 |
+
"learning_rate": 7.906365526802218e-06,
|
| 3603 |
+
"loss": 0.0514,
|
| 3604 |
+
"step": 25500
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 2.9517097966728283,
|
| 3608 |
+
"grad_norm": 0.009453874081373215,
|
| 3609 |
+
"learning_rate": 7.863042975970426e-06,
|
| 3610 |
+
"loss": 0.0844,
|
| 3611 |
+
"step": 25550
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 2.957486136783734,
|
| 3615 |
+
"grad_norm": 0.31572088599205017,
|
| 3616 |
+
"learning_rate": 7.819720425138632e-06,
|
| 3617 |
+
"loss": 0.0428,
|
| 3618 |
+
"step": 25600
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 2.9632624768946396,
|
| 3622 |
+
"grad_norm": 0.06910421699285507,
|
| 3623 |
+
"learning_rate": 7.77639787430684e-06,
|
| 3624 |
+
"loss": 0.0106,
|
| 3625 |
+
"step": 25650
|
| 3626 |
+
},
|
| 3627 |
+
{
|
| 3628 |
+
"epoch": 2.9690388170055453,
|
| 3629 |
+
"grad_norm": 0.006122531369328499,
|
| 3630 |
+
"learning_rate": 7.733075323475047e-06,
|
| 3631 |
+
"loss": 0.0565,
|
| 3632 |
+
"step": 25700
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 2.974815157116451,
|
| 3636 |
+
"grad_norm": 0.03837600350379944,
|
| 3637 |
+
"learning_rate": 7.689752772643252e-06,
|
| 3638 |
+
"loss": 0.0291,
|
| 3639 |
+
"step": 25750
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 2.9805914972273566,
|
| 3643 |
+
"grad_norm": 0.00848468393087387,
|
| 3644 |
+
"learning_rate": 7.64643022181146e-06,
|
| 3645 |
+
"loss": 0.0699,
|
| 3646 |
+
"step": 25800
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 2.9863678373382623,
|
| 3650 |
+
"grad_norm": 0.001210428075864911,
|
| 3651 |
+
"learning_rate": 7.6031076709796675e-06,
|
| 3652 |
+
"loss": 0.041,
|
| 3653 |
+
"step": 25850
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 2.992144177449168,
|
| 3657 |
+
"grad_norm": 0.017735475674271584,
|
| 3658 |
+
"learning_rate": 7.559785120147875e-06,
|
| 3659 |
+
"loss": 0.0553,
|
| 3660 |
+
"step": 25900
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 2.997920517560074,
|
| 3664 |
+
"grad_norm": 0.06611788272857666,
|
| 3665 |
+
"learning_rate": 7.516462569316081e-06,
|
| 3666 |
+
"loss": 0.036,
|
| 3667 |
+
"step": 25950
|
| 3668 |
+
},
|
| 3669 |
+
{
|
| 3670 |
+
"epoch": 3.0,
|
| 3671 |
+
"eval_accuracy": 0.96816114359974,
|
| 3672 |
+
"eval_f1": 0.9809026424506977,
|
| 3673 |
+
"eval_loss": 0.17402061820030212,
|
| 3674 |
+
"eval_precision": 0.9764121663563005,
|
| 3675 |
+
"eval_recall": 0.9854346123727487,
|
| 3676 |
+
"eval_runtime": 84.6739,
|
| 3677 |
+
"eval_samples_per_second": 90.878,
|
| 3678 |
+
"eval_steps_per_second": 11.361,
|
| 3679 |
+
"step": 25968
|
| 3680 |
+
}
|
| 3681 |
+
],
|
| 3682 |
+
"logging_steps": 50,
|
| 3683 |
+
"max_steps": 34624,
|
| 3684 |
+
"num_input_tokens_seen": 0,
|
| 3685 |
+
"num_train_epochs": 4,
|
| 3686 |
+
"save_steps": 500,
|
| 3687 |
+
"stateful_callbacks": {
|
| 3688 |
+
"TrainerControl": {
|
| 3689 |
+
"args": {
|
| 3690 |
+
"should_epoch_stop": false,
|
| 3691 |
+
"should_evaluate": false,
|
| 3692 |
+
"should_log": false,
|
| 3693 |
+
"should_save": true,
|
| 3694 |
+
"should_training_stop": false
|
| 3695 |
+
},
|
| 3696 |
+
"attributes": {}
|
| 3697 |
+
}
|
| 3698 |
+
},
|
| 3699 |
+
"total_flos": 2.959186153230191e+19,
|
| 3700 |
+
"train_batch_size": 8,
|
| 3701 |
+
"trial_name": null,
|
| 3702 |
+
"trial_params": null
|
| 3703 |
+
}
|
checkpoint-25968/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:946e9e07d5a9e21a8c78cfee12dd636f44d131fddd49601bf03b9867e5b39a46
|
| 3 |
+
size 5841
|
checkpoint-34624/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-34624/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c42ab43470c86b80cc5a382f75ce0a9499b0694745da311dc2f0545f027bfd54
|
| 3 |
+
size 631399808
|
checkpoint-34624/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d57ed02d4a744f5b67e6c80493149a76aa2a95ecd62086d3c9df4f2a048b43dd
|
| 3 |
+
size 1262124766
|
checkpoint-34624/preprocessor_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.485,
|
| 8 |
+
0.456,
|
| 9 |
+
0.406
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "ViTImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.229,
|
| 14 |
+
0.224,
|
| 15 |
+
0.225
|
| 16 |
+
],
|
| 17 |
+
"resample": 3,
|
| 18 |
+
"rescale_factor": 0.00392156862745098,
|
| 19 |
+
"size": {
|
| 20 |
+
"height": 224,
|
| 21 |
+
"width": 224
|
| 22 |
+
}
|
| 23 |
+
}
|
checkpoint-34624/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f9c7c5702607031eeaf9e1d2969d8d5d2638d68f6d1d310cbf8b53f6957b0cd7
|
| 3 |
+
size 14645
|
checkpoint-34624/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a8ce8d9409a964c49d5cf5690a6837b292a72353a275422685c9f3ab1181012
|
| 3 |
+
size 1383
|
checkpoint-34624/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc4ee797c4d89007999882e9ccc08e1867c3ca85aca60a6ee9e970c9ceb96273
|
| 3 |
+
size 1465
|
checkpoint-34624/trainer_state.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-34624/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:946e9e07d5a9e21a8c78cfee12dd636f44d131fddd49601bf03b9867e5b39a46
|
| 3 |
+
size 5841
|
checkpoint-8656/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-8656/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ce8d170609078a7bb08d94d8d06631cbcf1d173fe6013a35b794ab090f554e4c
|
| 3 |
+
size 631399808
|
checkpoint-8656/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:480ccd4723e0c2802289a6752301025950696c4ed3ab77f321c7bb2d33633be2
|
| 3 |
+
size 1262124766
|
checkpoint-8656/preprocessor_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.485,
|
| 8 |
+
0.456,
|
| 9 |
+
0.406
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "ViTImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.229,
|
| 14 |
+
0.224,
|
| 15 |
+
0.225
|
| 16 |
+
],
|
| 17 |
+
"resample": 3,
|
| 18 |
+
"rescale_factor": 0.00392156862745098,
|
| 19 |
+
"size": {
|
| 20 |
+
"height": 224,
|
| 21 |
+
"width": 224
|
| 22 |
+
}
|
| 23 |
+
}
|
checkpoint-8656/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6aae9d0cba1abcf9c8f6a7d69c37de251d6690875450cc390d445dc9f9b1e8ce
|
| 3 |
+
size 14645
|
checkpoint-8656/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18561e06df5ddda047beb974fd4554932b03df02a9a36abc738030cf9ca5a697
|
| 3 |
+
size 1383
|
checkpoint-8656/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec5d51e5f51ce32c962a45f2dad6f241589b8e8e54563d2a41ad7d413353274e
|
| 3 |
+
size 1465
|
checkpoint-8656/trainer_state.json
ADDED
|
@@ -0,0 +1,1257 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 8656,
|
| 3 |
+
"best_metric": 0.9665611244111515,
|
| 4 |
+
"best_model_checkpoint": "./roadwork-swin-finetuned-local/checkpoint-8656",
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 8656,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00577634011090573,
|
| 14 |
+
"grad_norm": 12.141866683959961,
|
| 15 |
+
"learning_rate": 2.9957543900184844e-05,
|
| 16 |
+
"loss": 3.0039,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01155268022181146,
|
| 21 |
+
"grad_norm": 13.438708305358887,
|
| 22 |
+
"learning_rate": 2.991422134935305e-05,
|
| 23 |
+
"loss": 0.3812,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01732902033271719,
|
| 28 |
+
"grad_norm": 38.80968475341797,
|
| 29 |
+
"learning_rate": 2.9870898798521256e-05,
|
| 30 |
+
"loss": 0.3451,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02310536044362292,
|
| 35 |
+
"grad_norm": 16.20616340637207,
|
| 36 |
+
"learning_rate": 2.9827576247689466e-05,
|
| 37 |
+
"loss": 0.3082,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028881700554528652,
|
| 42 |
+
"grad_norm": 33.16380310058594,
|
| 43 |
+
"learning_rate": 2.9784253696857672e-05,
|
| 44 |
+
"loss": 0.3153,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03465804066543438,
|
| 49 |
+
"grad_norm": 18.74236297607422,
|
| 50 |
+
"learning_rate": 2.974093114602588e-05,
|
| 51 |
+
"loss": 0.3792,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04043438077634011,
|
| 56 |
+
"grad_norm": 4.538590431213379,
|
| 57 |
+
"learning_rate": 2.9697608595194084e-05,
|
| 58 |
+
"loss": 0.3673,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04621072088724584,
|
| 63 |
+
"grad_norm": 15.369573593139648,
|
| 64 |
+
"learning_rate": 2.9654286044362294e-05,
|
| 65 |
+
"loss": 0.3875,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.05198706099815157,
|
| 70 |
+
"grad_norm": 23.82215690612793,
|
| 71 |
+
"learning_rate": 2.96109634935305e-05,
|
| 72 |
+
"loss": 0.3386,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.057763401109057304,
|
| 77 |
+
"grad_norm": 14.058465003967285,
|
| 78 |
+
"learning_rate": 2.9567640942698706e-05,
|
| 79 |
+
"loss": 0.2766,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06353974121996303,
|
| 84 |
+
"grad_norm": 15.244014739990234,
|
| 85 |
+
"learning_rate": 2.9524318391866915e-05,
|
| 86 |
+
"loss": 0.2891,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06931608133086876,
|
| 91 |
+
"grad_norm": 30.079912185668945,
|
| 92 |
+
"learning_rate": 2.948099584103512e-05,
|
| 93 |
+
"loss": 0.315,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0750924214417745,
|
| 98 |
+
"grad_norm": 22.175500869750977,
|
| 99 |
+
"learning_rate": 2.9437673290203327e-05,
|
| 100 |
+
"loss": 0.2571,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08086876155268022,
|
| 105 |
+
"grad_norm": 13.770343780517578,
|
| 106 |
+
"learning_rate": 2.9394350739371534e-05,
|
| 107 |
+
"loss": 0.2805,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08664510166358595,
|
| 112 |
+
"grad_norm": 15.201981544494629,
|
| 113 |
+
"learning_rate": 2.9351028188539743e-05,
|
| 114 |
+
"loss": 0.3395,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09242144177449169,
|
| 119 |
+
"grad_norm": 0.11971250176429749,
|
| 120 |
+
"learning_rate": 2.930770563770795e-05,
|
| 121 |
+
"loss": 0.2807,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0981977818853974,
|
| 126 |
+
"grad_norm": 13.49336051940918,
|
| 127 |
+
"learning_rate": 2.9264383086876155e-05,
|
| 128 |
+
"loss": 0.3474,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10397412199630314,
|
| 133 |
+
"grad_norm": 12.165586471557617,
|
| 134 |
+
"learning_rate": 2.9221060536044365e-05,
|
| 135 |
+
"loss": 0.2704,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.10975046210720887,
|
| 140 |
+
"grad_norm": 14.561868667602539,
|
| 141 |
+
"learning_rate": 2.917773798521257e-05,
|
| 142 |
+
"loss": 0.2495,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.11552680221811461,
|
| 147 |
+
"grad_norm": 3.2725448608398438,
|
| 148 |
+
"learning_rate": 2.9134415434380777e-05,
|
| 149 |
+
"loss": 0.2951,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.12130314232902033,
|
| 154 |
+
"grad_norm": 12.453506469726562,
|
| 155 |
+
"learning_rate": 2.9091092883548983e-05,
|
| 156 |
+
"loss": 0.3338,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12707948243992606,
|
| 161 |
+
"grad_norm": 21.19861602783203,
|
| 162 |
+
"learning_rate": 2.904777033271719e-05,
|
| 163 |
+
"loss": 0.4023,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13285582255083178,
|
| 168 |
+
"grad_norm": 7.617332935333252,
|
| 169 |
+
"learning_rate": 2.90044477818854e-05,
|
| 170 |
+
"loss": 0.2225,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13863216266173753,
|
| 175 |
+
"grad_norm": 7.300302028656006,
|
| 176 |
+
"learning_rate": 2.8961125231053605e-05,
|
| 177 |
+
"loss": 0.2645,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14440850277264325,
|
| 182 |
+
"grad_norm": 11.860857963562012,
|
| 183 |
+
"learning_rate": 2.8917802680221814e-05,
|
| 184 |
+
"loss": 0.3154,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.150184842883549,
|
| 189 |
+
"grad_norm": 20.967884063720703,
|
| 190 |
+
"learning_rate": 2.887448012939002e-05,
|
| 191 |
+
"loss": 0.2485,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.15596118299445472,
|
| 196 |
+
"grad_norm": 11.119569778442383,
|
| 197 |
+
"learning_rate": 2.8831157578558226e-05,
|
| 198 |
+
"loss": 0.2854,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.16173752310536044,
|
| 203 |
+
"grad_norm": 4.218294620513916,
|
| 204 |
+
"learning_rate": 2.8787835027726432e-05,
|
| 205 |
+
"loss": 0.2434,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.16751386321626618,
|
| 210 |
+
"grad_norm": 5.249250888824463,
|
| 211 |
+
"learning_rate": 2.874451247689464e-05,
|
| 212 |
+
"loss": 0.3393,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1732902033271719,
|
| 217 |
+
"grad_norm": 1.712246298789978,
|
| 218 |
+
"learning_rate": 2.8701189926062848e-05,
|
| 219 |
+
"loss": 0.2331,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17906654343807762,
|
| 224 |
+
"grad_norm": 28.958724975585938,
|
| 225 |
+
"learning_rate": 2.8657867375231054e-05,
|
| 226 |
+
"loss": 0.2514,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.18484288354898337,
|
| 231 |
+
"grad_norm": 10.893854141235352,
|
| 232 |
+
"learning_rate": 2.8614544824399264e-05,
|
| 233 |
+
"loss": 0.3656,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1906192236598891,
|
| 238 |
+
"grad_norm": 26.43993377685547,
|
| 239 |
+
"learning_rate": 2.857122227356747e-05,
|
| 240 |
+
"loss": 0.2875,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.1963955637707948,
|
| 245 |
+
"grad_norm": 0.9008654952049255,
|
| 246 |
+
"learning_rate": 2.8527899722735676e-05,
|
| 247 |
+
"loss": 0.2569,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.20217190388170056,
|
| 252 |
+
"grad_norm": 6.621076583862305,
|
| 253 |
+
"learning_rate": 2.8484577171903882e-05,
|
| 254 |
+
"loss": 0.2228,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.20794824399260628,
|
| 259 |
+
"grad_norm": 0.4961116909980774,
|
| 260 |
+
"learning_rate": 2.8441254621072088e-05,
|
| 261 |
+
"loss": 0.368,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.21372458410351203,
|
| 266 |
+
"grad_norm": 31.944732666015625,
|
| 267 |
+
"learning_rate": 2.8397932070240297e-05,
|
| 268 |
+
"loss": 0.2735,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21950092421441775,
|
| 273 |
+
"grad_norm": 34.241703033447266,
|
| 274 |
+
"learning_rate": 2.8354609519408504e-05,
|
| 275 |
+
"loss": 0.2414,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.22527726432532347,
|
| 280 |
+
"grad_norm": 5.89654016494751,
|
| 281 |
+
"learning_rate": 2.8311286968576713e-05,
|
| 282 |
+
"loss": 0.2939,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.23105360443622922,
|
| 287 |
+
"grad_norm": 6.246095657348633,
|
| 288 |
+
"learning_rate": 2.8267964417744916e-05,
|
| 289 |
+
"loss": 0.2772,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.23682994454713494,
|
| 294 |
+
"grad_norm": 12.515145301818848,
|
| 295 |
+
"learning_rate": 2.8224641866913122e-05,
|
| 296 |
+
"loss": 0.2363,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.24260628465804066,
|
| 301 |
+
"grad_norm": 7.227047920227051,
|
| 302 |
+
"learning_rate": 2.818131931608133e-05,
|
| 303 |
+
"loss": 0.2087,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.2483826247689464,
|
| 308 |
+
"grad_norm": 8.628905296325684,
|
| 309 |
+
"learning_rate": 2.8137996765249537e-05,
|
| 310 |
+
"loss": 0.2902,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.2541589648798521,
|
| 315 |
+
"grad_norm": 15.576581001281738,
|
| 316 |
+
"learning_rate": 2.8094674214417747e-05,
|
| 317 |
+
"loss": 0.2752,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.25993530499075784,
|
| 322 |
+
"grad_norm": 16.116729736328125,
|
| 323 |
+
"learning_rate": 2.8051351663585953e-05,
|
| 324 |
+
"loss": 0.2604,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26571164510166356,
|
| 329 |
+
"grad_norm": 12.389599800109863,
|
| 330 |
+
"learning_rate": 2.8008029112754162e-05,
|
| 331 |
+
"loss": 0.2661,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.27148798521256934,
|
| 336 |
+
"grad_norm": 5.002556800842285,
|
| 337 |
+
"learning_rate": 2.7964706561922365e-05,
|
| 338 |
+
"loss": 0.2844,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27726432532347506,
|
| 343 |
+
"grad_norm": 23.136369705200195,
|
| 344 |
+
"learning_rate": 2.792138401109057e-05,
|
| 345 |
+
"loss": 0.2737,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.2830406654343808,
|
| 350 |
+
"grad_norm": 7.032516002655029,
|
| 351 |
+
"learning_rate": 2.787806146025878e-05,
|
| 352 |
+
"loss": 0.2597,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2888170055452865,
|
| 357 |
+
"grad_norm": 24.333494186401367,
|
| 358 |
+
"learning_rate": 2.7834738909426987e-05,
|
| 359 |
+
"loss": 0.2748,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.2945933456561922,
|
| 364 |
+
"grad_norm": 0.4094283878803253,
|
| 365 |
+
"learning_rate": 2.7791416358595196e-05,
|
| 366 |
+
"loss": 0.2368,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.300369685767098,
|
| 371 |
+
"grad_norm": 4.650086402893066,
|
| 372 |
+
"learning_rate": 2.7748093807763402e-05,
|
| 373 |
+
"loss": 0.3215,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.3061460258780037,
|
| 378 |
+
"grad_norm": 18.08386993408203,
|
| 379 |
+
"learning_rate": 2.7704771256931612e-05,
|
| 380 |
+
"loss": 0.255,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.31192236598890943,
|
| 385 |
+
"grad_norm": 10.52841854095459,
|
| 386 |
+
"learning_rate": 2.7661448706099815e-05,
|
| 387 |
+
"loss": 0.2439,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.31769870609981515,
|
| 392 |
+
"grad_norm": 65.88616180419922,
|
| 393 |
+
"learning_rate": 2.761812615526802e-05,
|
| 394 |
+
"loss": 0.2418,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.3234750462107209,
|
| 399 |
+
"grad_norm": 9.300713539123535,
|
| 400 |
+
"learning_rate": 2.757480360443623e-05,
|
| 401 |
+
"loss": 0.3188,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.3292513863216266,
|
| 406 |
+
"grad_norm": 8.528387069702148,
|
| 407 |
+
"learning_rate": 2.7531481053604436e-05,
|
| 408 |
+
"loss": 0.2996,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.33502772643253237,
|
| 413 |
+
"grad_norm": 25.872819900512695,
|
| 414 |
+
"learning_rate": 2.7488158502772646e-05,
|
| 415 |
+
"loss": 0.2303,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3408040665434381,
|
| 420 |
+
"grad_norm": 32.80488204956055,
|
| 421 |
+
"learning_rate": 2.7444835951940852e-05,
|
| 422 |
+
"loss": 0.2472,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.3465804066543438,
|
| 427 |
+
"grad_norm": 3.634104013442993,
|
| 428 |
+
"learning_rate": 2.7401513401109058e-05,
|
| 429 |
+
"loss": 0.3054,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.35235674676524953,
|
| 434 |
+
"grad_norm": 9.497197151184082,
|
| 435 |
+
"learning_rate": 2.7358190850277264e-05,
|
| 436 |
+
"loss": 0.264,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.35813308687615525,
|
| 441 |
+
"grad_norm": 28.931177139282227,
|
| 442 |
+
"learning_rate": 2.731486829944547e-05,
|
| 443 |
+
"loss": 0.2419,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.363909426987061,
|
| 448 |
+
"grad_norm": 25.71267318725586,
|
| 449 |
+
"learning_rate": 2.727154574861368e-05,
|
| 450 |
+
"loss": 0.2137,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.36968576709796674,
|
| 455 |
+
"grad_norm": 18.439529418945312,
|
| 456 |
+
"learning_rate": 2.7228223197781886e-05,
|
| 457 |
+
"loss": 0.2562,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.37546210720887246,
|
| 462 |
+
"grad_norm": 2.7945632934570312,
|
| 463 |
+
"learning_rate": 2.7184900646950095e-05,
|
| 464 |
+
"loss": 0.3085,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3812384473197782,
|
| 469 |
+
"grad_norm": 20.88351058959961,
|
| 470 |
+
"learning_rate": 2.71415780961183e-05,
|
| 471 |
+
"loss": 0.1988,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3870147874306839,
|
| 476 |
+
"grad_norm": 9.544013977050781,
|
| 477 |
+
"learning_rate": 2.7098255545286504e-05,
|
| 478 |
+
"loss": 0.252,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.3927911275415896,
|
| 483 |
+
"grad_norm": 0.2528328001499176,
|
| 484 |
+
"learning_rate": 2.7054932994454714e-05,
|
| 485 |
+
"loss": 0.2136,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3985674676524954,
|
| 490 |
+
"grad_norm": 15.808114051818848,
|
| 491 |
+
"learning_rate": 2.701161044362292e-05,
|
| 492 |
+
"loss": 0.3502,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4043438077634011,
|
| 497 |
+
"grad_norm": 9.585843086242676,
|
| 498 |
+
"learning_rate": 2.696828789279113e-05,
|
| 499 |
+
"loss": 0.2437,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.41012014787430684,
|
| 504 |
+
"grad_norm": 0.7092911601066589,
|
| 505 |
+
"learning_rate": 2.6924965341959335e-05,
|
| 506 |
+
"loss": 0.2137,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.41589648798521256,
|
| 511 |
+
"grad_norm": 0.09910676628351212,
|
| 512 |
+
"learning_rate": 2.6881642791127545e-05,
|
| 513 |
+
"loss": 0.2979,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.4216728280961183,
|
| 518 |
+
"grad_norm": 0.674543559551239,
|
| 519 |
+
"learning_rate": 2.683832024029575e-05,
|
| 520 |
+
"loss": 0.2392,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.42744916820702406,
|
| 525 |
+
"grad_norm": 0.6626635193824768,
|
| 526 |
+
"learning_rate": 2.6794997689463953e-05,
|
| 527 |
+
"loss": 0.2019,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.4332255083179298,
|
| 532 |
+
"grad_norm": 9.206210136413574,
|
| 533 |
+
"learning_rate": 2.6751675138632163e-05,
|
| 534 |
+
"loss": 0.2747,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.4390018484288355,
|
| 539 |
+
"grad_norm": 1.017768144607544,
|
| 540 |
+
"learning_rate": 2.670835258780037e-05,
|
| 541 |
+
"loss": 0.1992,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.4447781885397412,
|
| 546 |
+
"grad_norm": 18.11737060546875,
|
| 547 |
+
"learning_rate": 2.666503003696858e-05,
|
| 548 |
+
"loss": 0.2201,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.45055452865064693,
|
| 553 |
+
"grad_norm": 15.56303882598877,
|
| 554 |
+
"learning_rate": 2.6621707486136785e-05,
|
| 555 |
+
"loss": 0.198,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.45633086876155265,
|
| 560 |
+
"grad_norm": 7.689689636230469,
|
| 561 |
+
"learning_rate": 2.657838493530499e-05,
|
| 562 |
+
"loss": 0.2894,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.46210720887245843,
|
| 567 |
+
"grad_norm": 9.518302917480469,
|
| 568 |
+
"learning_rate": 2.65350623844732e-05,
|
| 569 |
+
"loss": 0.2226,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.46788354898336415,
|
| 574 |
+
"grad_norm": 0.4917680025100708,
|
| 575 |
+
"learning_rate": 2.6491739833641403e-05,
|
| 576 |
+
"loss": 0.2602,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.47365988909426987,
|
| 581 |
+
"grad_norm": 34.92518997192383,
|
| 582 |
+
"learning_rate": 2.6448417282809612e-05,
|
| 583 |
+
"loss": 0.2506,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.4794362292051756,
|
| 588 |
+
"grad_norm": 14.343157768249512,
|
| 589 |
+
"learning_rate": 2.640509473197782e-05,
|
| 590 |
+
"loss": 0.3009,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4852125693160813,
|
| 595 |
+
"grad_norm": 14.651656150817871,
|
| 596 |
+
"learning_rate": 2.6361772181146028e-05,
|
| 597 |
+
"loss": 0.2419,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4909889094269871,
|
| 602 |
+
"grad_norm": 0.7227654457092285,
|
| 603 |
+
"learning_rate": 2.6318449630314234e-05,
|
| 604 |
+
"loss": 0.3091,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4967652495378928,
|
| 609 |
+
"grad_norm": 10.362215042114258,
|
| 610 |
+
"learning_rate": 2.627512707948244e-05,
|
| 611 |
+
"loss": 0.1949,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.5025415896487985,
|
| 616 |
+
"grad_norm": 11.842758178710938,
|
| 617 |
+
"learning_rate": 2.623180452865065e-05,
|
| 618 |
+
"loss": 0.3337,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.5083179297597042,
|
| 623 |
+
"grad_norm": 11.550918579101562,
|
| 624 |
+
"learning_rate": 2.6188481977818852e-05,
|
| 625 |
+
"loss": 0.2641,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.51409426987061,
|
| 630 |
+
"grad_norm": 5.512154579162598,
|
| 631 |
+
"learning_rate": 2.6145159426987062e-05,
|
| 632 |
+
"loss": 0.2541,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5198706099815157,
|
| 637 |
+
"grad_norm": 10.161574363708496,
|
| 638 |
+
"learning_rate": 2.6101836876155268e-05,
|
| 639 |
+
"loss": 0.256,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.5256469500924215,
|
| 644 |
+
"grad_norm": 1.659515380859375,
|
| 645 |
+
"learning_rate": 2.6058514325323477e-05,
|
| 646 |
+
"loss": 0.2621,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.5314232902033271,
|
| 651 |
+
"grad_norm": 2.5476644039154053,
|
| 652 |
+
"learning_rate": 2.6015191774491684e-05,
|
| 653 |
+
"loss": 0.294,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5371996303142329,
|
| 658 |
+
"grad_norm": 0.9729189276695251,
|
| 659 |
+
"learning_rate": 2.597186922365989e-05,
|
| 660 |
+
"loss": 0.1333,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.5429759704251387,
|
| 665 |
+
"grad_norm": 3.8201911449432373,
|
| 666 |
+
"learning_rate": 2.59285466728281e-05,
|
| 667 |
+
"loss": 0.2598,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.5487523105360443,
|
| 672 |
+
"grad_norm": 0.17872726917266846,
|
| 673 |
+
"learning_rate": 2.5885224121996302e-05,
|
| 674 |
+
"loss": 0.1897,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.5545286506469501,
|
| 679 |
+
"grad_norm": 24.479658126831055,
|
| 680 |
+
"learning_rate": 2.584190157116451e-05,
|
| 681 |
+
"loss": 0.2606,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.5603049907578558,
|
| 686 |
+
"grad_norm": 13.361931800842285,
|
| 687 |
+
"learning_rate": 2.5798579020332717e-05,
|
| 688 |
+
"loss": 0.271,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5660813308687616,
|
| 693 |
+
"grad_norm": 8.625031471252441,
|
| 694 |
+
"learning_rate": 2.5755256469500924e-05,
|
| 695 |
+
"loss": 0.2056,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.5718576709796673,
|
| 700 |
+
"grad_norm": 11.34391975402832,
|
| 701 |
+
"learning_rate": 2.5711933918669133e-05,
|
| 702 |
+
"loss": 0.2736,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.577634011090573,
|
| 707 |
+
"grad_norm": 16.9676456451416,
|
| 708 |
+
"learning_rate": 2.566861136783734e-05,
|
| 709 |
+
"loss": 0.2821,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5834103512014788,
|
| 714 |
+
"grad_norm": 9.531896591186523,
|
| 715 |
+
"learning_rate": 2.5625288817005545e-05,
|
| 716 |
+
"loss": 0.2264,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5891866913123844,
|
| 721 |
+
"grad_norm": 2.4691050052642822,
|
| 722 |
+
"learning_rate": 2.558196626617375e-05,
|
| 723 |
+
"loss": 0.2148,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5949630314232902,
|
| 728 |
+
"grad_norm": 0.476725697517395,
|
| 729 |
+
"learning_rate": 2.553864371534196e-05,
|
| 730 |
+
"loss": 0.1964,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.600739371534196,
|
| 735 |
+
"grad_norm": 14.917049407958984,
|
| 736 |
+
"learning_rate": 2.5495321164510167e-05,
|
| 737 |
+
"loss": 0.2717,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.6065157116451017,
|
| 742 |
+
"grad_norm": 0.13745556771755219,
|
| 743 |
+
"learning_rate": 2.5451998613678373e-05,
|
| 744 |
+
"loss": 0.2086,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.6122920517560074,
|
| 749 |
+
"grad_norm": 22.109750747680664,
|
| 750 |
+
"learning_rate": 2.5408676062846582e-05,
|
| 751 |
+
"loss": 0.3401,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.6180683918669131,
|
| 756 |
+
"grad_norm": 0.4055859446525574,
|
| 757 |
+
"learning_rate": 2.536535351201479e-05,
|
| 758 |
+
"loss": 0.2572,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6238447319778189,
|
| 763 |
+
"grad_norm": 0.7329151034355164,
|
| 764 |
+
"learning_rate": 2.5322030961182995e-05,
|
| 765 |
+
"loss": 0.2246,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.6296210720887245,
|
| 770 |
+
"grad_norm": 0.6793464422225952,
|
| 771 |
+
"learning_rate": 2.52787084103512e-05,
|
| 772 |
+
"loss": 0.1804,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.6353974121996303,
|
| 777 |
+
"grad_norm": 6.19757604598999,
|
| 778 |
+
"learning_rate": 2.523538585951941e-05,
|
| 779 |
+
"loss": 0.2331,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.6411737523105361,
|
| 784 |
+
"grad_norm": 69.8536148071289,
|
| 785 |
+
"learning_rate": 2.5192063308687616e-05,
|
| 786 |
+
"loss": 0.2071,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.6469500924214417,
|
| 791 |
+
"grad_norm": 49.18756103515625,
|
| 792 |
+
"learning_rate": 2.5148740757855822e-05,
|
| 793 |
+
"loss": 0.2602,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.6527264325323475,
|
| 798 |
+
"grad_norm": 9.139322280883789,
|
| 799 |
+
"learning_rate": 2.5105418207024032e-05,
|
| 800 |
+
"loss": 0.1975,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.6585027726432532,
|
| 805 |
+
"grad_norm": 11.072602272033691,
|
| 806 |
+
"learning_rate": 2.5062095656192238e-05,
|
| 807 |
+
"loss": 0.1829,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.664279112754159,
|
| 812 |
+
"grad_norm": 0.06665151566267014,
|
| 813 |
+
"learning_rate": 2.5018773105360444e-05,
|
| 814 |
+
"loss": 0.2722,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.6700554528650647,
|
| 819 |
+
"grad_norm": 1.071594476699829,
|
| 820 |
+
"learning_rate": 2.497545055452865e-05,
|
| 821 |
+
"loss": 0.1957,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.6758317929759704,
|
| 826 |
+
"grad_norm": 3.1452102661132812,
|
| 827 |
+
"learning_rate": 2.4932128003696856e-05,
|
| 828 |
+
"loss": 0.1312,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.6816081330868762,
|
| 833 |
+
"grad_norm": 2.483550786972046,
|
| 834 |
+
"learning_rate": 2.4888805452865066e-05,
|
| 835 |
+
"loss": 0.2375,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.6873844731977818,
|
| 840 |
+
"grad_norm": 24.874065399169922,
|
| 841 |
+
"learning_rate": 2.4845482902033272e-05,
|
| 842 |
+
"loss": 0.2743,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.6931608133086876,
|
| 847 |
+
"grad_norm": 31.57330322265625,
|
| 848 |
+
"learning_rate": 2.480216035120148e-05,
|
| 849 |
+
"loss": 0.2298,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.6989371534195934,
|
| 854 |
+
"grad_norm": 21.310131072998047,
|
| 855 |
+
"learning_rate": 2.4758837800369687e-05,
|
| 856 |
+
"loss": 0.2634,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.7047134935304991,
|
| 861 |
+
"grad_norm": 10.233534812927246,
|
| 862 |
+
"learning_rate": 2.4715515249537894e-05,
|
| 863 |
+
"loss": 0.2778,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7104898336414048,
|
| 868 |
+
"grad_norm": 0.34112706780433655,
|
| 869 |
+
"learning_rate": 2.46721926987061e-05,
|
| 870 |
+
"loss": 0.2009,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.7162661737523105,
|
| 875 |
+
"grad_norm": 0.2525830864906311,
|
| 876 |
+
"learning_rate": 2.4628870147874306e-05,
|
| 877 |
+
"loss": 0.2068,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.7220425138632163,
|
| 882 |
+
"grad_norm": 16.547571182250977,
|
| 883 |
+
"learning_rate": 2.4585547597042515e-05,
|
| 884 |
+
"loss": 0.2343,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.727818853974122,
|
| 889 |
+
"grad_norm": 0.7130794525146484,
|
| 890 |
+
"learning_rate": 2.454222504621072e-05,
|
| 891 |
+
"loss": 0.2312,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.7335951940850277,
|
| 896 |
+
"grad_norm": 0.8044223785400391,
|
| 897 |
+
"learning_rate": 2.449890249537893e-05,
|
| 898 |
+
"loss": 0.1719,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.7393715341959335,
|
| 903 |
+
"grad_norm": 13.686079978942871,
|
| 904 |
+
"learning_rate": 2.4455579944547134e-05,
|
| 905 |
+
"loss": 0.2398,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.7451478743068392,
|
| 910 |
+
"grad_norm": 1.6030184030532837,
|
| 911 |
+
"learning_rate": 2.4412257393715343e-05,
|
| 912 |
+
"loss": 0.1675,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7509242144177449,
|
| 917 |
+
"grad_norm": 2.320791482925415,
|
| 918 |
+
"learning_rate": 2.436893484288355e-05,
|
| 919 |
+
"loss": 0.2505,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.7567005545286506,
|
| 924 |
+
"grad_norm": 12.586448669433594,
|
| 925 |
+
"learning_rate": 2.4325612292051755e-05,
|
| 926 |
+
"loss": 0.2716,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.7624768946395564,
|
| 931 |
+
"grad_norm": 7.049114227294922,
|
| 932 |
+
"learning_rate": 2.4282289741219965e-05,
|
| 933 |
+
"loss": 0.169,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.7682532347504621,
|
| 938 |
+
"grad_norm": 9.050933837890625,
|
| 939 |
+
"learning_rate": 2.423896719038817e-05,
|
| 940 |
+
"loss": 0.2116,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.7740295748613678,
|
| 945 |
+
"grad_norm": 25.830007553100586,
|
| 946 |
+
"learning_rate": 2.419564463955638e-05,
|
| 947 |
+
"loss": 0.2181,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.7798059149722736,
|
| 952 |
+
"grad_norm": 0.14128220081329346,
|
| 953 |
+
"learning_rate": 2.4152322088724583e-05,
|
| 954 |
+
"loss": 0.2531,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.7855822550831792,
|
| 959 |
+
"grad_norm": 0.03034038469195366,
|
| 960 |
+
"learning_rate": 2.4108999537892792e-05,
|
| 961 |
+
"loss": 0.2109,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.791358595194085,
|
| 966 |
+
"grad_norm": 0.1031632274389267,
|
| 967 |
+
"learning_rate": 2.4065676987061e-05,
|
| 968 |
+
"loss": 0.2155,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7971349353049908,
|
| 973 |
+
"grad_norm": 42.54078674316406,
|
| 974 |
+
"learning_rate": 2.4022354436229205e-05,
|
| 975 |
+
"loss": 0.183,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.8029112754158965,
|
| 980 |
+
"grad_norm": 0.03734962269663811,
|
| 981 |
+
"learning_rate": 2.3979031885397414e-05,
|
| 982 |
+
"loss": 0.185,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.8086876155268022,
|
| 987 |
+
"grad_norm": 48.300174713134766,
|
| 988 |
+
"learning_rate": 2.393570933456562e-05,
|
| 989 |
+
"loss": 0.1803,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.8144639556377079,
|
| 994 |
+
"grad_norm": 13.418463706970215,
|
| 995 |
+
"learning_rate": 2.389238678373383e-05,
|
| 996 |
+
"loss": 0.2226,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.8202402957486137,
|
| 1001 |
+
"grad_norm": 16.923093795776367,
|
| 1002 |
+
"learning_rate": 2.3849064232902032e-05,
|
| 1003 |
+
"loss": 0.2387,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.8260166358595195,
|
| 1008 |
+
"grad_norm": 37.3891487121582,
|
| 1009 |
+
"learning_rate": 2.380574168207024e-05,
|
| 1010 |
+
"loss": 0.2168,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8317929759704251,
|
| 1015 |
+
"grad_norm": 24.676448822021484,
|
| 1016 |
+
"learning_rate": 2.3762419131238448e-05,
|
| 1017 |
+
"loss": 0.202,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.8375693160813309,
|
| 1022 |
+
"grad_norm": 13.302814483642578,
|
| 1023 |
+
"learning_rate": 2.3719096580406654e-05,
|
| 1024 |
+
"loss": 0.2153,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8433456561922366,
|
| 1029 |
+
"grad_norm": 8.611613273620605,
|
| 1030 |
+
"learning_rate": 2.3675774029574864e-05,
|
| 1031 |
+
"loss": 0.2544,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.8491219963031423,
|
| 1036 |
+
"grad_norm": 0.4762769639492035,
|
| 1037 |
+
"learning_rate": 2.363245147874307e-05,
|
| 1038 |
+
"loss": 0.2129,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.8548983364140481,
|
| 1043 |
+
"grad_norm": 16.49268341064453,
|
| 1044 |
+
"learning_rate": 2.358912892791128e-05,
|
| 1045 |
+
"loss": 0.177,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.8606746765249538,
|
| 1050 |
+
"grad_norm": 2.05298113822937,
|
| 1051 |
+
"learning_rate": 2.3545806377079482e-05,
|
| 1052 |
+
"loss": 0.2274,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.8664510166358595,
|
| 1057 |
+
"grad_norm": 19.834819793701172,
|
| 1058 |
+
"learning_rate": 2.3502483826247688e-05,
|
| 1059 |
+
"loss": 0.1998,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.8722273567467652,
|
| 1064 |
+
"grad_norm": 9.706733703613281,
|
| 1065 |
+
"learning_rate": 2.3459161275415897e-05,
|
| 1066 |
+
"loss": 0.2563,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.878003696857671,
|
| 1071 |
+
"grad_norm": 9.801465034484863,
|
| 1072 |
+
"learning_rate": 2.3415838724584104e-05,
|
| 1073 |
+
"loss": 0.1453,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.8837800369685767,
|
| 1078 |
+
"grad_norm": 11.452254295349121,
|
| 1079 |
+
"learning_rate": 2.3372516173752313e-05,
|
| 1080 |
+
"loss": 0.272,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.8895563770794824,
|
| 1085 |
+
"grad_norm": 7.050422191619873,
|
| 1086 |
+
"learning_rate": 2.332919362292052e-05,
|
| 1087 |
+
"loss": 0.1713,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.8953327171903882,
|
| 1092 |
+
"grad_norm": 46.33341979980469,
|
| 1093 |
+
"learning_rate": 2.3285871072088725e-05,
|
| 1094 |
+
"loss": 0.1967,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.9011090573012939,
|
| 1099 |
+
"grad_norm": 1.1651716232299805,
|
| 1100 |
+
"learning_rate": 2.324254852125693e-05,
|
| 1101 |
+
"loss": 0.187,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.9068853974121996,
|
| 1106 |
+
"grad_norm": 0.3762068450450897,
|
| 1107 |
+
"learning_rate": 2.3199225970425137e-05,
|
| 1108 |
+
"loss": 0.2556,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.9126617375231053,
|
| 1113 |
+
"grad_norm": 9.550520896911621,
|
| 1114 |
+
"learning_rate": 2.3155903419593347e-05,
|
| 1115 |
+
"loss": 0.2413,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.9184380776340111,
|
| 1120 |
+
"grad_norm": 0.3446040451526642,
|
| 1121 |
+
"learning_rate": 2.3112580868761553e-05,
|
| 1122 |
+
"loss": 0.1694,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.9242144177449169,
|
| 1127 |
+
"grad_norm": 0.7896172404289246,
|
| 1128 |
+
"learning_rate": 2.3069258317929762e-05,
|
| 1129 |
+
"loss": 0.2504,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.9299907578558225,
|
| 1134 |
+
"grad_norm": 34.30472946166992,
|
| 1135 |
+
"learning_rate": 2.302593576709797e-05,
|
| 1136 |
+
"loss": 0.2462,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9357670979667283,
|
| 1141 |
+
"grad_norm": 31.708581924438477,
|
| 1142 |
+
"learning_rate": 2.298261321626617e-05,
|
| 1143 |
+
"loss": 0.147,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.941543438077634,
|
| 1148 |
+
"grad_norm": 2.5662238597869873,
|
| 1149 |
+
"learning_rate": 2.293929066543438e-05,
|
| 1150 |
+
"loss": 0.262,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.9473197781885397,
|
| 1155 |
+
"grad_norm": 0.19891509413719177,
|
| 1156 |
+
"learning_rate": 2.2895968114602587e-05,
|
| 1157 |
+
"loss": 0.1978,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.9530961182994455,
|
| 1162 |
+
"grad_norm": 81.1672134399414,
|
| 1163 |
+
"learning_rate": 2.2852645563770796e-05,
|
| 1164 |
+
"loss": 0.207,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.9588724584103512,
|
| 1169 |
+
"grad_norm": 20.577178955078125,
|
| 1170 |
+
"learning_rate": 2.2809323012939002e-05,
|
| 1171 |
+
"loss": 0.2256,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.964648798521257,
|
| 1176 |
+
"grad_norm": 0.11451568454504013,
|
| 1177 |
+
"learning_rate": 2.2766000462107212e-05,
|
| 1178 |
+
"loss": 0.2123,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.9704251386321626,
|
| 1183 |
+
"grad_norm": 2.9672398567199707,
|
| 1184 |
+
"learning_rate": 2.2722677911275418e-05,
|
| 1185 |
+
"loss": 0.2419,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.9762014787430684,
|
| 1190 |
+
"grad_norm": 3.283834934234619,
|
| 1191 |
+
"learning_rate": 2.267935536044362e-05,
|
| 1192 |
+
"loss": 0.1796,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9819778188539742,
|
| 1197 |
+
"grad_norm": 19.114147186279297,
|
| 1198 |
+
"learning_rate": 2.263603280961183e-05,
|
| 1199 |
+
"loss": 0.1789,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.9877541589648798,
|
| 1204 |
+
"grad_norm": 0.5701966881752014,
|
| 1205 |
+
"learning_rate": 2.2592710258780036e-05,
|
| 1206 |
+
"loss": 0.1616,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.9935304990757856,
|
| 1211 |
+
"grad_norm": 14.517534255981445,
|
| 1212 |
+
"learning_rate": 2.2549387707948246e-05,
|
| 1213 |
+
"loss": 0.1909,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.9993068391866913,
|
| 1218 |
+
"grad_norm": 0.1315908581018448,
|
| 1219 |
+
"learning_rate": 2.2506065157116452e-05,
|
| 1220 |
+
"loss": 0.2188,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.0,
|
| 1225 |
+
"eval_accuracy": 0.9437296946068876,
|
| 1226 |
+
"eval_f1": 0.9665611244111515,
|
| 1227 |
+
"eval_loss": 0.2036169469356537,
|
| 1228 |
+
"eval_precision": 0.953382084095064,
|
| 1229 |
+
"eval_recall": 0.9801096319498825,
|
| 1230 |
+
"eval_runtime": 88.7618,
|
| 1231 |
+
"eval_samples_per_second": 86.693,
|
| 1232 |
+
"eval_steps_per_second": 10.838,
|
| 1233 |
+
"step": 8656
|
| 1234 |
+
}
|
| 1235 |
+
],
|
| 1236 |
+
"logging_steps": 50,
|
| 1237 |
+
"max_steps": 34624,
|
| 1238 |
+
"num_input_tokens_seen": 0,
|
| 1239 |
+
"num_train_epochs": 4,
|
| 1240 |
+
"save_steps": 500,
|
| 1241 |
+
"stateful_callbacks": {
|
| 1242 |
+
"TrainerControl": {
|
| 1243 |
+
"args": {
|
| 1244 |
+
"should_epoch_stop": false,
|
| 1245 |
+
"should_evaluate": false,
|
| 1246 |
+
"should_log": false,
|
| 1247 |
+
"should_save": true,
|
| 1248 |
+
"should_training_stop": false
|
| 1249 |
+
},
|
| 1250 |
+
"attributes": {}
|
| 1251 |
+
}
|
| 1252 |
+
},
|
| 1253 |
+
"total_flos": 9.863953844100637e+18,
|
| 1254 |
+
"train_batch_size": 8,
|
| 1255 |
+
"trial_name": null,
|
| 1256 |
+
"trial_params": null
|
| 1257 |
+
}
|
checkpoint-8656/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:946e9e07d5a9e21a8c78cfee12dd636f44d131fddd49601bf03b9867e5b39a46
|
| 3 |
+
size 5841
|
config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c42ab43470c86b80cc5a382f75ce0a9499b0694745da311dc2f0545f027bfd54
|
| 3 |
+
size 631399808
|
model_card.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_name": "rossijakob/rossi-ray1",
|
| 3 |
+
"description": "roadwork-detect model",
|
| 4 |
+
"version": "1.0.0",
|
| 5 |
+
"submitted_by": "5HEUFxZpguKZnYV7RtGwnSsbB1bQ6k8pS3Qq5nEACawtooJR",
|
| 6 |
+
"submission_time": 1760590276
|
| 7 |
+
}
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.485,
|
| 8 |
+
0.456,
|
| 9 |
+
0.406
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "ViTImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.229,
|
| 14 |
+
0.224,
|
| 15 |
+
0.225
|
| 16 |
+
],
|
| 17 |
+
"resample": 3,
|
| 18 |
+
"rescale_factor": 0.00392156862745098,
|
| 19 |
+
"size": {
|
| 20 |
+
"height": 224,
|
| 21 |
+
"width": 224
|
| 22 |
+
}
|
| 23 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:946e9e07d5a9e21a8c78cfee12dd636f44d131fddd49601bf03b9867e5b39a46
|
| 3 |
+
size 5841
|