Upload model + model card
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- README.md +5 -20
- 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-43280/config.json +0 -0
- checkpoint-43280/model.safetensors +3 -0
- checkpoint-43280/optimizer.pt +3 -0
- checkpoint-43280/preprocessor_config.json +23 -0
- checkpoint-43280/rng_state.pth +3 -0
- checkpoint-43280/scaler.pt +3 -0
- checkpoint-43280/scheduler.pt +3 -0
- checkpoint-43280/trainer_state.json +0 -0
- checkpoint-43280/training_args.bin +3 -0
- checkpoint-51936/config.json +0 -0
- checkpoint-51936/model.safetensors +3 -0
- checkpoint-51936/optimizer.pt +3 -0
- checkpoint-51936/preprocessor_config.json +23 -0
- checkpoint-51936/rng_state.pth +3 -0
- checkpoint-51936/scaler.pt +3 -0
- checkpoint-51936/scheduler.pt +3 -0
- checkpoint-51936/trainer_state.json +0 -0
- checkpoint-51936/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
README.md
CHANGED
|
@@ -1,24 +1,9 @@
|
|
| 1 |
-
-
|
| 2 |
-
language: en
|
| 3 |
-
tags:
|
| 4 |
-
- computer-vision
|
| 5 |
-
- object-detection
|
| 6 |
-
- roadwork
|
| 7 |
-
license: mit
|
| 8 |
-
model_type: swin-transformer
|
| 9 |
-
base_model: microsoft/swin-base-patch4-window7-224
|
| 10 |
-
datasets:
|
| 11 |
-
- custom
|
| 12 |
-
pipeline_tag: image-classification
|
| 13 |
-
---
|
| 14 |
|
| 15 |
-
|
| 16 |
|
| 17 |
-
**
|
| 18 |
-
**Version:** 1.0.1
|
| 19 |
-
**Submitted by:** 5DjmdSU3W3AjxrxzjpZfH6tLaKx7bEjRjxoNG5MS9iJod2VX
|
| 20 |
-
**Submission time:** 1760262635
|
| 21 |
|
| 22 |
-
|
| 23 |
|
| 24 |
-
|
|
|
|
| 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:** 1760589948
|
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 sha256:1654dd7142b8d465f79b9d3a9403c7950d6d79f729df43c538b1ed22468c4fd9
|
| 3 |
+
size 631399808
|
checkpoint-17312/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c43f12736a019e9019322942c5f5da9c1370262b5760ba48fab1bff1b36f326a
|
| 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 sha256:0c0800b5aa7c07ffdb9ffcbb1551157cb586c097136b36e1fe20c459d2a427a7
|
| 3 |
+
size 1383
|
checkpoint-17312/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ed8110c6a8362f25b3c0029b50aa44505e20662c84992b7f1afdcec99d522e4
|
| 3 |
+
size 1465
|
checkpoint-17312/trainer_state.json
ADDED
|
@@ -0,0 +1,2480 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 17312,
|
| 3 |
+
"best_metric": 0.9746439192361872,
|
| 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": 8.07317066192627,
|
| 15 |
+
"learning_rate": 2.9971695933456562e-05,
|
| 16 |
+
"loss": 2.9896,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01155268022181146,
|
| 21 |
+
"grad_norm": NaN,
|
| 22 |
+
"learning_rate": 2.994281423290203e-05,
|
| 23 |
+
"loss": 0.3707,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01732902033271719,
|
| 28 |
+
"grad_norm": 14.865354537963867,
|
| 29 |
+
"learning_rate": 2.9913932532347504e-05,
|
| 30 |
+
"loss": 0.3821,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02310536044362292,
|
| 35 |
+
"grad_norm": 14.388459205627441,
|
| 36 |
+
"learning_rate": 2.9885050831792977e-05,
|
| 37 |
+
"loss": 0.318,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028881700554528652,
|
| 42 |
+
"grad_norm": 33.513973236083984,
|
| 43 |
+
"learning_rate": 2.9856169131238447e-05,
|
| 44 |
+
"loss": 0.3427,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03465804066543438,
|
| 49 |
+
"grad_norm": 14.235688209533691,
|
| 50 |
+
"learning_rate": 2.982728743068392e-05,
|
| 51 |
+
"loss": 0.4154,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04043438077634011,
|
| 56 |
+
"grad_norm": 2.8046276569366455,
|
| 57 |
+
"learning_rate": 2.979840573012939e-05,
|
| 58 |
+
"loss": 0.351,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04621072088724584,
|
| 63 |
+
"grad_norm": 15.46557903289795,
|
| 64 |
+
"learning_rate": 2.9769524029574863e-05,
|
| 65 |
+
"loss": 0.3631,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.05198706099815157,
|
| 70 |
+
"grad_norm": 36.24979019165039,
|
| 71 |
+
"learning_rate": 2.9740642329020332e-05,
|
| 72 |
+
"loss": 0.346,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.057763401109057304,
|
| 77 |
+
"grad_norm": 11.466980934143066,
|
| 78 |
+
"learning_rate": 2.9711760628465805e-05,
|
| 79 |
+
"loss": 0.2994,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06353974121996303,
|
| 84 |
+
"grad_norm": 15.98156452178955,
|
| 85 |
+
"learning_rate": 2.9682878927911278e-05,
|
| 86 |
+
"loss": 0.2899,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06931608133086876,
|
| 91 |
+
"grad_norm": 29.2628116607666,
|
| 92 |
+
"learning_rate": 2.9653997227356748e-05,
|
| 93 |
+
"loss": 0.3105,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0750924214417745,
|
| 98 |
+
"grad_norm": 7.130237579345703,
|
| 99 |
+
"learning_rate": 2.962511552680222e-05,
|
| 100 |
+
"loss": 0.2463,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08086876155268022,
|
| 105 |
+
"grad_norm": 12.123393058776855,
|
| 106 |
+
"learning_rate": 2.959623382624769e-05,
|
| 107 |
+
"loss": 0.2928,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08664510166358595,
|
| 112 |
+
"grad_norm": 19.521760940551758,
|
| 113 |
+
"learning_rate": 2.9567352125693163e-05,
|
| 114 |
+
"loss": 0.3383,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09242144177449169,
|
| 119 |
+
"grad_norm": 0.5567889213562012,
|
| 120 |
+
"learning_rate": 2.9538470425138633e-05,
|
| 121 |
+
"loss": 0.2872,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0981977818853974,
|
| 126 |
+
"grad_norm": 14.521562576293945,
|
| 127 |
+
"learning_rate": 2.9509588724584106e-05,
|
| 128 |
+
"loss": 0.3419,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10397412199630314,
|
| 133 |
+
"grad_norm": 30.65926170349121,
|
| 134 |
+
"learning_rate": 2.9480707024029576e-05,
|
| 135 |
+
"loss": 0.2911,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.10975046210720887,
|
| 140 |
+
"grad_norm": 6.595367908477783,
|
| 141 |
+
"learning_rate": 2.9451825323475045e-05,
|
| 142 |
+
"loss": 0.2741,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.11552680221811461,
|
| 147 |
+
"grad_norm": 1.3052228689193726,
|
| 148 |
+
"learning_rate": 2.9422943622920518e-05,
|
| 149 |
+
"loss": 0.2869,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.12130314232902033,
|
| 154 |
+
"grad_norm": 26.094087600708008,
|
| 155 |
+
"learning_rate": 2.9394061922365988e-05,
|
| 156 |
+
"loss": 0.3495,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12707948243992606,
|
| 161 |
+
"grad_norm": 19.971145629882812,
|
| 162 |
+
"learning_rate": 2.936518022181146e-05,
|
| 163 |
+
"loss": 0.3805,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13285582255083178,
|
| 168 |
+
"grad_norm": 6.1732563972473145,
|
| 169 |
+
"learning_rate": 2.933629852125693e-05,
|
| 170 |
+
"loss": 0.2168,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13863216266173753,
|
| 175 |
+
"grad_norm": 18.476938247680664,
|
| 176 |
+
"learning_rate": 2.9307416820702403e-05,
|
| 177 |
+
"loss": 0.31,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14440850277264325,
|
| 182 |
+
"grad_norm": 11.29546070098877,
|
| 183 |
+
"learning_rate": 2.9278535120147873e-05,
|
| 184 |
+
"loss": 0.3223,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.150184842883549,
|
| 189 |
+
"grad_norm": 58.089473724365234,
|
| 190 |
+
"learning_rate": 2.9249653419593346e-05,
|
| 191 |
+
"loss": 0.2283,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.15596118299445472,
|
| 196 |
+
"grad_norm": 0.983078122138977,
|
| 197 |
+
"learning_rate": 2.922077171903882e-05,
|
| 198 |
+
"loss": 0.3208,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.16173752310536044,
|
| 203 |
+
"grad_norm": 13.572915077209473,
|
| 204 |
+
"learning_rate": 2.919189001848429e-05,
|
| 205 |
+
"loss": 0.2801,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.16751386321626618,
|
| 210 |
+
"grad_norm": 1.8694497346878052,
|
| 211 |
+
"learning_rate": 2.916300831792976e-05,
|
| 212 |
+
"loss": 0.3147,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1732902033271719,
|
| 217 |
+
"grad_norm": 2.4720425605773926,
|
| 218 |
+
"learning_rate": 2.913412661737523e-05,
|
| 219 |
+
"loss": 0.261,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17906654343807762,
|
| 224 |
+
"grad_norm": 24.087940216064453,
|
| 225 |
+
"learning_rate": 2.9105244916820704e-05,
|
| 226 |
+
"loss": 0.2689,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.18484288354898337,
|
| 231 |
+
"grad_norm": 12.160676956176758,
|
| 232 |
+
"learning_rate": 2.9076363216266174e-05,
|
| 233 |
+
"loss": 0.3336,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1906192236598891,
|
| 238 |
+
"grad_norm": 34.87266540527344,
|
| 239 |
+
"learning_rate": 2.9047481515711647e-05,
|
| 240 |
+
"loss": 0.2899,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.1963955637707948,
|
| 245 |
+
"grad_norm": 6.795339107513428,
|
| 246 |
+
"learning_rate": 2.901859981515712e-05,
|
| 247 |
+
"loss": 0.2453,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.20217190388170056,
|
| 252 |
+
"grad_norm": 25.78632354736328,
|
| 253 |
+
"learning_rate": 2.898971811460259e-05,
|
| 254 |
+
"loss": 0.2297,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.20794824399260628,
|
| 259 |
+
"grad_norm": 0.665915846824646,
|
| 260 |
+
"learning_rate": 2.8960836414048062e-05,
|
| 261 |
+
"loss": 0.3941,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.21372458410351203,
|
| 266 |
+
"grad_norm": 26.92215919494629,
|
| 267 |
+
"learning_rate": 2.8931954713493532e-05,
|
| 268 |
+
"loss": 0.2736,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21950092421441775,
|
| 273 |
+
"grad_norm": 15.98168659210205,
|
| 274 |
+
"learning_rate": 2.8903073012939e-05,
|
| 275 |
+
"loss": 0.241,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.22527726432532347,
|
| 280 |
+
"grad_norm": 14.969513893127441,
|
| 281 |
+
"learning_rate": 2.887419131238447e-05,
|
| 282 |
+
"loss": 0.3063,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.23105360443622922,
|
| 287 |
+
"grad_norm": 2.79641056060791,
|
| 288 |
+
"learning_rate": 2.8845309611829944e-05,
|
| 289 |
+
"loss": 0.3148,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.23682994454713494,
|
| 294 |
+
"grad_norm": 13.938512802124023,
|
| 295 |
+
"learning_rate": 2.8816427911275417e-05,
|
| 296 |
+
"loss": 0.2214,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.24260628465804066,
|
| 301 |
+
"grad_norm": 2.2964584827423096,
|
| 302 |
+
"learning_rate": 2.8787546210720887e-05,
|
| 303 |
+
"loss": 0.2408,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.2483826247689464,
|
| 308 |
+
"grad_norm": 4.9652605056762695,
|
| 309 |
+
"learning_rate": 2.875866451016636e-05,
|
| 310 |
+
"loss": 0.3097,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.2541589648798521,
|
| 315 |
+
"grad_norm": 12.751204490661621,
|
| 316 |
+
"learning_rate": 2.872978280961183e-05,
|
| 317 |
+
"loss": 0.3046,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.25993530499075784,
|
| 322 |
+
"grad_norm": 21.8670597076416,
|
| 323 |
+
"learning_rate": 2.8700901109057302e-05,
|
| 324 |
+
"loss": 0.2608,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26571164510166356,
|
| 329 |
+
"grad_norm": 10.842573165893555,
|
| 330 |
+
"learning_rate": 2.8672019408502772e-05,
|
| 331 |
+
"loss": 0.2792,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.27148798521256934,
|
| 336 |
+
"grad_norm": 1.6427525281906128,
|
| 337 |
+
"learning_rate": 2.8643137707948245e-05,
|
| 338 |
+
"loss": 0.2752,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27726432532347506,
|
| 343 |
+
"grad_norm": 20.114425659179688,
|
| 344 |
+
"learning_rate": 2.8614256007393718e-05,
|
| 345 |
+
"loss": 0.2766,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.2830406654343808,
|
| 350 |
+
"grad_norm": 3.259066104888916,
|
| 351 |
+
"learning_rate": 2.8585374306839187e-05,
|
| 352 |
+
"loss": 0.2855,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2888170055452865,
|
| 357 |
+
"grad_norm": 14.488828659057617,
|
| 358 |
+
"learning_rate": 2.855649260628466e-05,
|
| 359 |
+
"loss": 0.2713,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.2945933456561922,
|
| 364 |
+
"grad_norm": 0.09920129925012589,
|
| 365 |
+
"learning_rate": 2.852761090573013e-05,
|
| 366 |
+
"loss": 0.2569,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.300369685767098,
|
| 371 |
+
"grad_norm": 4.165335655212402,
|
| 372 |
+
"learning_rate": 2.8498729205175603e-05,
|
| 373 |
+
"loss": 0.3494,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.3061460258780037,
|
| 378 |
+
"grad_norm": 11.144678115844727,
|
| 379 |
+
"learning_rate": 2.8469847504621073e-05,
|
| 380 |
+
"loss": 0.2928,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.31192236598890943,
|
| 385 |
+
"grad_norm": 11.037677764892578,
|
| 386 |
+
"learning_rate": 2.8440965804066546e-05,
|
| 387 |
+
"loss": 0.2367,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.31769870609981515,
|
| 392 |
+
"grad_norm": 33.86788558959961,
|
| 393 |
+
"learning_rate": 2.841208410351202e-05,
|
| 394 |
+
"loss": 0.2385,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.3234750462107209,
|
| 399 |
+
"grad_norm": 14.876667976379395,
|
| 400 |
+
"learning_rate": 2.8383202402957488e-05,
|
| 401 |
+
"loss": 0.3315,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.3292513863216266,
|
| 406 |
+
"grad_norm": 11.150797843933105,
|
| 407 |
+
"learning_rate": 2.8354320702402958e-05,
|
| 408 |
+
"loss": 0.2914,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.33502772643253237,
|
| 413 |
+
"grad_norm": 17.667240142822266,
|
| 414 |
+
"learning_rate": 2.8325439001848427e-05,
|
| 415 |
+
"loss": 0.2445,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3408040665434381,
|
| 420 |
+
"grad_norm": 20.02605438232422,
|
| 421 |
+
"learning_rate": 2.82965573012939e-05,
|
| 422 |
+
"loss": 0.2804,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.3465804066543438,
|
| 427 |
+
"grad_norm": 27.934114456176758,
|
| 428 |
+
"learning_rate": 2.826767560073937e-05,
|
| 429 |
+
"loss": 0.3034,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.35235674676524953,
|
| 434 |
+
"grad_norm": 5.3321332931518555,
|
| 435 |
+
"learning_rate": 2.8238793900184843e-05,
|
| 436 |
+
"loss": 0.2677,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.35813308687615525,
|
| 441 |
+
"grad_norm": 22.4134521484375,
|
| 442 |
+
"learning_rate": 2.8209912199630313e-05,
|
| 443 |
+
"loss": 0.219,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.363909426987061,
|
| 448 |
+
"grad_norm": 25.509113311767578,
|
| 449 |
+
"learning_rate": 2.8181030499075786e-05,
|
| 450 |
+
"loss": 0.2367,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.36968576709796674,
|
| 455 |
+
"grad_norm": 18.299274444580078,
|
| 456 |
+
"learning_rate": 2.815214879852126e-05,
|
| 457 |
+
"loss": 0.2418,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.37546210720887246,
|
| 462 |
+
"grad_norm": 6.69123649597168,
|
| 463 |
+
"learning_rate": 2.8123267097966728e-05,
|
| 464 |
+
"loss": 0.3057,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3812384473197782,
|
| 469 |
+
"grad_norm": 26.583688735961914,
|
| 470 |
+
"learning_rate": 2.80943853974122e-05,
|
| 471 |
+
"loss": 0.2101,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3870147874306839,
|
| 476 |
+
"grad_norm": 10.969908714294434,
|
| 477 |
+
"learning_rate": 2.806550369685767e-05,
|
| 478 |
+
"loss": 0.2447,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.3927911275415896,
|
| 483 |
+
"grad_norm": 1.5843168497085571,
|
| 484 |
+
"learning_rate": 2.8036621996303144e-05,
|
| 485 |
+
"loss": 0.2241,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3985674676524954,
|
| 490 |
+
"grad_norm": 21.7174129486084,
|
| 491 |
+
"learning_rate": 2.8007740295748613e-05,
|
| 492 |
+
"loss": 0.3335,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4043438077634011,
|
| 497 |
+
"grad_norm": 14.578802108764648,
|
| 498 |
+
"learning_rate": 2.7978858595194086e-05,
|
| 499 |
+
"loss": 0.2309,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.41012014787430684,
|
| 504 |
+
"grad_norm": 0.6399129033088684,
|
| 505 |
+
"learning_rate": 2.794997689463956e-05,
|
| 506 |
+
"loss": 0.2236,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.41589648798521256,
|
| 511 |
+
"grad_norm": 0.07050570845603943,
|
| 512 |
+
"learning_rate": 2.792109519408503e-05,
|
| 513 |
+
"loss": 0.2711,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.4216728280961183,
|
| 518 |
+
"grad_norm": 0.8827792406082153,
|
| 519 |
+
"learning_rate": 2.7892213493530502e-05,
|
| 520 |
+
"loss": 0.2229,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.42744916820702406,
|
| 525 |
+
"grad_norm": 6.008078575134277,
|
| 526 |
+
"learning_rate": 2.786333179297597e-05,
|
| 527 |
+
"loss": 0.2254,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.4332255083179298,
|
| 532 |
+
"grad_norm": 9.279077529907227,
|
| 533 |
+
"learning_rate": 2.7834450092421444e-05,
|
| 534 |
+
"loss": 0.2593,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.4390018484288355,
|
| 539 |
+
"grad_norm": 6.761204242706299,
|
| 540 |
+
"learning_rate": 2.780556839186691e-05,
|
| 541 |
+
"loss": 0.1904,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.4447781885397412,
|
| 546 |
+
"grad_norm": 29.752927780151367,
|
| 547 |
+
"learning_rate": 2.7776686691312384e-05,
|
| 548 |
+
"loss": 0.2287,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.45055452865064693,
|
| 553 |
+
"grad_norm": 15.296160697937012,
|
| 554 |
+
"learning_rate": 2.7747804990757857e-05,
|
| 555 |
+
"loss": 0.215,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.45633086876155265,
|
| 560 |
+
"grad_norm": 10.903311729431152,
|
| 561 |
+
"learning_rate": 2.7718923290203326e-05,
|
| 562 |
+
"loss": 0.2449,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.46210720887245843,
|
| 567 |
+
"grad_norm": 7.3522047996521,
|
| 568 |
+
"learning_rate": 2.76900415896488e-05,
|
| 569 |
+
"loss": 0.2581,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.46788354898336415,
|
| 574 |
+
"grad_norm": 1.4037950038909912,
|
| 575 |
+
"learning_rate": 2.766115988909427e-05,
|
| 576 |
+
"loss": 0.2753,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.47365988909426987,
|
| 581 |
+
"grad_norm": 54.007911682128906,
|
| 582 |
+
"learning_rate": 2.7632278188539742e-05,
|
| 583 |
+
"loss": 0.2105,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.4794362292051756,
|
| 588 |
+
"grad_norm": 18.261384963989258,
|
| 589 |
+
"learning_rate": 2.760339648798521e-05,
|
| 590 |
+
"loss": 0.3445,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4852125693160813,
|
| 595 |
+
"grad_norm": 19.31684684753418,
|
| 596 |
+
"learning_rate": 2.7574514787430684e-05,
|
| 597 |
+
"loss": 0.2484,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4909889094269871,
|
| 602 |
+
"grad_norm": 0.8063836693763733,
|
| 603 |
+
"learning_rate": 2.7545633086876157e-05,
|
| 604 |
+
"loss": 0.306,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4967652495378928,
|
| 609 |
+
"grad_norm": 15.769898414611816,
|
| 610 |
+
"learning_rate": 2.7516751386321627e-05,
|
| 611 |
+
"loss": 0.2102,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.5025415896487985,
|
| 616 |
+
"grad_norm": 7.9377217292785645,
|
| 617 |
+
"learning_rate": 2.74878696857671e-05,
|
| 618 |
+
"loss": 0.2961,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.5083179297597042,
|
| 623 |
+
"grad_norm": 10.430275917053223,
|
| 624 |
+
"learning_rate": 2.745898798521257e-05,
|
| 625 |
+
"loss": 0.291,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.51409426987061,
|
| 630 |
+
"grad_norm": 10.901015281677246,
|
| 631 |
+
"learning_rate": 2.7430106284658043e-05,
|
| 632 |
+
"loss": 0.2726,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5198706099815157,
|
| 637 |
+
"grad_norm": 13.966771125793457,
|
| 638 |
+
"learning_rate": 2.7401224584103512e-05,
|
| 639 |
+
"loss": 0.2795,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.5256469500924215,
|
| 644 |
+
"grad_norm": 0.8151399493217468,
|
| 645 |
+
"learning_rate": 2.7372342883548985e-05,
|
| 646 |
+
"loss": 0.2491,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.5314232902033271,
|
| 651 |
+
"grad_norm": 27.18870735168457,
|
| 652 |
+
"learning_rate": 2.7343461182994458e-05,
|
| 653 |
+
"loss": 0.259,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5371996303142329,
|
| 658 |
+
"grad_norm": 0.14737075567245483,
|
| 659 |
+
"learning_rate": 2.7314579482439928e-05,
|
| 660 |
+
"loss": 0.1497,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.5429759704251387,
|
| 665 |
+
"grad_norm": 15.187653541564941,
|
| 666 |
+
"learning_rate": 2.72856977818854e-05,
|
| 667 |
+
"loss": 0.2764,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.5487523105360443,
|
| 672 |
+
"grad_norm": 0.07880234718322754,
|
| 673 |
+
"learning_rate": 2.7256816081330867e-05,
|
| 674 |
+
"loss": 0.1721,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.5545286506469501,
|
| 679 |
+
"grad_norm": 21.45745849609375,
|
| 680 |
+
"learning_rate": 2.722793438077634e-05,
|
| 681 |
+
"loss": 0.2486,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.5603049907578558,
|
| 686 |
+
"grad_norm": 22.560449600219727,
|
| 687 |
+
"learning_rate": 2.719905268022181e-05,
|
| 688 |
+
"loss": 0.2866,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5660813308687616,
|
| 693 |
+
"grad_norm": 9.561197280883789,
|
| 694 |
+
"learning_rate": 2.7170170979667283e-05,
|
| 695 |
+
"loss": 0.2175,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.5718576709796673,
|
| 700 |
+
"grad_norm": 10.6559419631958,
|
| 701 |
+
"learning_rate": 2.7141289279112756e-05,
|
| 702 |
+
"loss": 0.2421,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.577634011090573,
|
| 707 |
+
"grad_norm": 13.36540699005127,
|
| 708 |
+
"learning_rate": 2.7112407578558225e-05,
|
| 709 |
+
"loss": 0.2709,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5834103512014788,
|
| 714 |
+
"grad_norm": 8.7359619140625,
|
| 715 |
+
"learning_rate": 2.7083525878003698e-05,
|
| 716 |
+
"loss": 0.2108,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5891866913123844,
|
| 721 |
+
"grad_norm": 3.588351249694824,
|
| 722 |
+
"learning_rate": 2.7054644177449168e-05,
|
| 723 |
+
"loss": 0.1813,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5949630314232902,
|
| 728 |
+
"grad_norm": 1.5511142015457153,
|
| 729 |
+
"learning_rate": 2.702576247689464e-05,
|
| 730 |
+
"loss": 0.1868,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.600739371534196,
|
| 735 |
+
"grad_norm": 13.601162910461426,
|
| 736 |
+
"learning_rate": 2.699688077634011e-05,
|
| 737 |
+
"loss": 0.2756,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.6065157116451017,
|
| 742 |
+
"grad_norm": 0.9333792924880981,
|
| 743 |
+
"learning_rate": 2.6967999075785583e-05,
|
| 744 |
+
"loss": 0.2196,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.6122920517560074,
|
| 749 |
+
"grad_norm": 18.923709869384766,
|
| 750 |
+
"learning_rate": 2.6939117375231053e-05,
|
| 751 |
+
"loss": 0.3018,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.6180683918669131,
|
| 756 |
+
"grad_norm": 13.762917518615723,
|
| 757 |
+
"learning_rate": 2.6910235674676526e-05,
|
| 758 |
+
"loss": 0.2636,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6238447319778189,
|
| 763 |
+
"grad_norm": 1.0572352409362793,
|
| 764 |
+
"learning_rate": 2.6881353974122e-05,
|
| 765 |
+
"loss": 0.1796,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.6296210720887245,
|
| 770 |
+
"grad_norm": 2.580142021179199,
|
| 771 |
+
"learning_rate": 2.685247227356747e-05,
|
| 772 |
+
"loss": 0.195,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.6353974121996303,
|
| 777 |
+
"grad_norm": 4.498309135437012,
|
| 778 |
+
"learning_rate": 2.682359057301294e-05,
|
| 779 |
+
"loss": 0.2573,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.6411737523105361,
|
| 784 |
+
"grad_norm": 9.187541961669922,
|
| 785 |
+
"learning_rate": 2.679470887245841e-05,
|
| 786 |
+
"loss": 0.2201,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.6469500924214417,
|
| 791 |
+
"grad_norm": 39.70842361450195,
|
| 792 |
+
"learning_rate": 2.6765827171903884e-05,
|
| 793 |
+
"loss": 0.2327,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.6527264325323475,
|
| 798 |
+
"grad_norm": 11.123932838439941,
|
| 799 |
+
"learning_rate": 2.6736945471349354e-05,
|
| 800 |
+
"loss": 0.2328,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.6585027726432532,
|
| 805 |
+
"grad_norm": 13.621745109558105,
|
| 806 |
+
"learning_rate": 2.6708063770794823e-05,
|
| 807 |
+
"loss": 0.1527,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.664279112754159,
|
| 812 |
+
"grad_norm": 0.0642027035355568,
|
| 813 |
+
"learning_rate": 2.6679182070240296e-05,
|
| 814 |
+
"loss": 0.2881,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.6700554528650647,
|
| 819 |
+
"grad_norm": 0.32092395424842834,
|
| 820 |
+
"learning_rate": 2.6650300369685766e-05,
|
| 821 |
+
"loss": 0.2152,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.6758317929759704,
|
| 826 |
+
"grad_norm": 0.8972311615943909,
|
| 827 |
+
"learning_rate": 2.662141866913124e-05,
|
| 828 |
+
"loss": 0.1611,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.6816081330868762,
|
| 833 |
+
"grad_norm": 10.612149238586426,
|
| 834 |
+
"learning_rate": 2.659253696857671e-05,
|
| 835 |
+
"loss": 0.2985,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.6873844731977818,
|
| 840 |
+
"grad_norm": 11.864514350891113,
|
| 841 |
+
"learning_rate": 2.656365526802218e-05,
|
| 842 |
+
"loss": 0.2405,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.6931608133086876,
|
| 847 |
+
"grad_norm": 2.5804641246795654,
|
| 848 |
+
"learning_rate": 2.653477356746765e-05,
|
| 849 |
+
"loss": 0.2181,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.6989371534195934,
|
| 854 |
+
"grad_norm": 7.676444053649902,
|
| 855 |
+
"learning_rate": 2.6505891866913124e-05,
|
| 856 |
+
"loss": 0.2983,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.7047134935304991,
|
| 861 |
+
"grad_norm": 9.115777015686035,
|
| 862 |
+
"learning_rate": 2.6477010166358597e-05,
|
| 863 |
+
"loss": 0.2331,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7104898336414048,
|
| 868 |
+
"grad_norm": 0.4313638210296631,
|
| 869 |
+
"learning_rate": 2.6448128465804067e-05,
|
| 870 |
+
"loss": 0.1798,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.7162661737523105,
|
| 875 |
+
"grad_norm": 10.339539527893066,
|
| 876 |
+
"learning_rate": 2.641924676524954e-05,
|
| 877 |
+
"loss": 0.2015,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.7220425138632163,
|
| 882 |
+
"grad_norm": 11.897482872009277,
|
| 883 |
+
"learning_rate": 2.639036506469501e-05,
|
| 884 |
+
"loss": 0.2284,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.727818853974122,
|
| 889 |
+
"grad_norm": 0.22236181795597076,
|
| 890 |
+
"learning_rate": 2.6361483364140482e-05,
|
| 891 |
+
"loss": 0.2422,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.7335951940850277,
|
| 896 |
+
"grad_norm": 0.6248801946640015,
|
| 897 |
+
"learning_rate": 2.6332601663585952e-05,
|
| 898 |
+
"loss": 0.1927,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.7393715341959335,
|
| 903 |
+
"grad_norm": 21.60765838623047,
|
| 904 |
+
"learning_rate": 2.6303719963031425e-05,
|
| 905 |
+
"loss": 0.1929,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.7451478743068392,
|
| 910 |
+
"grad_norm": 7.824502468109131,
|
| 911 |
+
"learning_rate": 2.6274838262476898e-05,
|
| 912 |
+
"loss": 0.1485,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7509242144177449,
|
| 917 |
+
"grad_norm": 5.444038391113281,
|
| 918 |
+
"learning_rate": 2.6245956561922367e-05,
|
| 919 |
+
"loss": 0.2577,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.7567005545286506,
|
| 924 |
+
"grad_norm": 9.758363723754883,
|
| 925 |
+
"learning_rate": 2.621707486136784e-05,
|
| 926 |
+
"loss": 0.2703,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.7624768946395564,
|
| 931 |
+
"grad_norm": 0.42428484559059143,
|
| 932 |
+
"learning_rate": 2.618819316081331e-05,
|
| 933 |
+
"loss": 0.1711,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.7682532347504621,
|
| 938 |
+
"grad_norm": 10.41826343536377,
|
| 939 |
+
"learning_rate": 2.615931146025878e-05,
|
| 940 |
+
"loss": 0.2248,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.7740295748613678,
|
| 945 |
+
"grad_norm": 21.749746322631836,
|
| 946 |
+
"learning_rate": 2.613042975970425e-05,
|
| 947 |
+
"loss": 0.1968,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.7798059149722736,
|
| 952 |
+
"grad_norm": 0.19098539650440216,
|
| 953 |
+
"learning_rate": 2.6101548059149722e-05,
|
| 954 |
+
"loss": 0.2706,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.7855822550831792,
|
| 959 |
+
"grad_norm": 0.015689531341195107,
|
| 960 |
+
"learning_rate": 2.6072666358595195e-05,
|
| 961 |
+
"loss": 0.2384,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.791358595194085,
|
| 966 |
+
"grad_norm": 0.1689176857471466,
|
| 967 |
+
"learning_rate": 2.6043784658040665e-05,
|
| 968 |
+
"loss": 0.1714,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7971349353049908,
|
| 973 |
+
"grad_norm": 34.842952728271484,
|
| 974 |
+
"learning_rate": 2.6014902957486138e-05,
|
| 975 |
+
"loss": 0.215,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.8029112754158965,
|
| 980 |
+
"grad_norm": 0.10479370504617691,
|
| 981 |
+
"learning_rate": 2.5986021256931607e-05,
|
| 982 |
+
"loss": 0.2195,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.8086876155268022,
|
| 987 |
+
"grad_norm": 0.7637598514556885,
|
| 988 |
+
"learning_rate": 2.595713955637708e-05,
|
| 989 |
+
"loss": 0.1624,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.8144639556377079,
|
| 994 |
+
"grad_norm": 20.622161865234375,
|
| 995 |
+
"learning_rate": 2.592825785582255e-05,
|
| 996 |
+
"loss": 0.1896,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.8202402957486137,
|
| 1001 |
+
"grad_norm": 16.110519409179688,
|
| 1002 |
+
"learning_rate": 2.5899376155268023e-05,
|
| 1003 |
+
"loss": 0.2468,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.8260166358595195,
|
| 1008 |
+
"grad_norm": 19.84160804748535,
|
| 1009 |
+
"learning_rate": 2.5870494454713496e-05,
|
| 1010 |
+
"loss": 0.1933,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8317929759704251,
|
| 1015 |
+
"grad_norm": 80.84690856933594,
|
| 1016 |
+
"learning_rate": 2.5841612754158966e-05,
|
| 1017 |
+
"loss": 0.1801,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.8375693160813309,
|
| 1022 |
+
"grad_norm": 15.291082382202148,
|
| 1023 |
+
"learning_rate": 2.581273105360444e-05,
|
| 1024 |
+
"loss": 0.2181,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8433456561922366,
|
| 1029 |
+
"grad_norm": 16.199522018432617,
|
| 1030 |
+
"learning_rate": 2.5783849353049908e-05,
|
| 1031 |
+
"loss": 0.2292,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.8491219963031423,
|
| 1036 |
+
"grad_norm": 17.037158966064453,
|
| 1037 |
+
"learning_rate": 2.575496765249538e-05,
|
| 1038 |
+
"loss": 0.2492,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.8548983364140481,
|
| 1043 |
+
"grad_norm": 20.019588470458984,
|
| 1044 |
+
"learning_rate": 2.572608595194085e-05,
|
| 1045 |
+
"loss": 0.1992,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.8606746765249538,
|
| 1050 |
+
"grad_norm": 17.38285255432129,
|
| 1051 |
+
"learning_rate": 2.5697204251386324e-05,
|
| 1052 |
+
"loss": 0.1994,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.8664510166358595,
|
| 1057 |
+
"grad_norm": 2.336366653442383,
|
| 1058 |
+
"learning_rate": 2.5668322550831793e-05,
|
| 1059 |
+
"loss": 0.2053,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.8722273567467652,
|
| 1064 |
+
"grad_norm": 62.698020935058594,
|
| 1065 |
+
"learning_rate": 2.5639440850277266e-05,
|
| 1066 |
+
"loss": 0.2787,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.878003696857671,
|
| 1071 |
+
"grad_norm": 13.748190879821777,
|
| 1072 |
+
"learning_rate": 2.5610559149722736e-05,
|
| 1073 |
+
"loss": 0.1935,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.8837800369685767,
|
| 1078 |
+
"grad_norm": 22.238149642944336,
|
| 1079 |
+
"learning_rate": 2.5581677449168206e-05,
|
| 1080 |
+
"loss": 0.3124,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.8895563770794824,
|
| 1085 |
+
"grad_norm": 1.701218843460083,
|
| 1086 |
+
"learning_rate": 2.555279574861368e-05,
|
| 1087 |
+
"loss": 0.1903,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.8953327171903882,
|
| 1092 |
+
"grad_norm": 18.659778594970703,
|
| 1093 |
+
"learning_rate": 2.5523914048059148e-05,
|
| 1094 |
+
"loss": 0.2223,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.9011090573012939,
|
| 1099 |
+
"grad_norm": 0.3101861774921417,
|
| 1100 |
+
"learning_rate": 2.549503234750462e-05,
|
| 1101 |
+
"loss": 0.2226,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.9068853974121996,
|
| 1106 |
+
"grad_norm": 0.15353505313396454,
|
| 1107 |
+
"learning_rate": 2.546615064695009e-05,
|
| 1108 |
+
"loss": 0.2453,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.9126617375231053,
|
| 1113 |
+
"grad_norm": 27.39130401611328,
|
| 1114 |
+
"learning_rate": 2.5437268946395564e-05,
|
| 1115 |
+
"loss": 0.2311,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.9184380776340111,
|
| 1120 |
+
"grad_norm": 0.3352121114730835,
|
| 1121 |
+
"learning_rate": 2.5408387245841037e-05,
|
| 1122 |
+
"loss": 0.1582,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.9242144177449169,
|
| 1127 |
+
"grad_norm": 1.0091736316680908,
|
| 1128 |
+
"learning_rate": 2.5379505545286506e-05,
|
| 1129 |
+
"loss": 0.2695,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.9299907578558225,
|
| 1134 |
+
"grad_norm": 21.034908294677734,
|
| 1135 |
+
"learning_rate": 2.535062384473198e-05,
|
| 1136 |
+
"loss": 0.2481,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9357670979667283,
|
| 1141 |
+
"grad_norm": 26.847919464111328,
|
| 1142 |
+
"learning_rate": 2.532174214417745e-05,
|
| 1143 |
+
"loss": 0.1675,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.941543438077634,
|
| 1148 |
+
"grad_norm": 9.66991138458252,
|
| 1149 |
+
"learning_rate": 2.5292860443622922e-05,
|
| 1150 |
+
"loss": 0.2571,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.9473197781885397,
|
| 1155 |
+
"grad_norm": 0.12006015330553055,
|
| 1156 |
+
"learning_rate": 2.526397874306839e-05,
|
| 1157 |
+
"loss": 0.1748,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.9530961182994455,
|
| 1162 |
+
"grad_norm": 1.825127363204956,
|
| 1163 |
+
"learning_rate": 2.5235097042513864e-05,
|
| 1164 |
+
"loss": 0.2161,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.9588724584103512,
|
| 1169 |
+
"grad_norm": 12.851505279541016,
|
| 1170 |
+
"learning_rate": 2.5206215341959337e-05,
|
| 1171 |
+
"loss": 0.2372,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.964648798521257,
|
| 1176 |
+
"grad_norm": 24.965559005737305,
|
| 1177 |
+
"learning_rate": 2.5177333641404807e-05,
|
| 1178 |
+
"loss": 0.2058,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.9704251386321626,
|
| 1183 |
+
"grad_norm": 33.49872589111328,
|
| 1184 |
+
"learning_rate": 2.514845194085028e-05,
|
| 1185 |
+
"loss": 0.2352,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.9762014787430684,
|
| 1190 |
+
"grad_norm": 2.5571775436401367,
|
| 1191 |
+
"learning_rate": 2.511957024029575e-05,
|
| 1192 |
+
"loss": 0.1973,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9819778188539742,
|
| 1197 |
+
"grad_norm": 12.988107681274414,
|
| 1198 |
+
"learning_rate": 2.5090688539741223e-05,
|
| 1199 |
+
"loss": 0.1691,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.9877541589648798,
|
| 1204 |
+
"grad_norm": 22.103260040283203,
|
| 1205 |
+
"learning_rate": 2.5061806839186692e-05,
|
| 1206 |
+
"loss": 0.202,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.9935304990757856,
|
| 1211 |
+
"grad_norm": 16.42685890197754,
|
| 1212 |
+
"learning_rate": 2.5032925138632162e-05,
|
| 1213 |
+
"loss": 0.1637,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.9993068391866913,
|
| 1218 |
+
"grad_norm": 0.6291359663009644,
|
| 1219 |
+
"learning_rate": 2.5004043438077635e-05,
|
| 1220 |
+
"loss": 0.2335,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.0,
|
| 1225 |
+
"eval_accuracy": 0.9354126055880442,
|
| 1226 |
+
"eval_f1": 0.9607951408061844,
|
| 1227 |
+
"eval_loss": 0.19707663357257843,
|
| 1228 |
+
"eval_precision": 0.9678957406230133,
|
| 1229 |
+
"eval_recall": 0.9537979639780736,
|
| 1230 |
+
"eval_runtime": 68.7658,
|
| 1231 |
+
"eval_samples_per_second": 111.902,
|
| 1232 |
+
"eval_steps_per_second": 13.99,
|
| 1233 |
+
"step": 8656
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.005083179297597,
|
| 1237 |
+
"grad_norm": 11.418679237365723,
|
| 1238 |
+
"learning_rate": 2.4975161737523104e-05,
|
| 1239 |
+
"loss": 0.158,
|
| 1240 |
+
"step": 8700
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.0108595194085028,
|
| 1244 |
+
"grad_norm": 35.1962776184082,
|
| 1245 |
+
"learning_rate": 2.4946280036968577e-05,
|
| 1246 |
+
"loss": 0.11,
|
| 1247 |
+
"step": 8750
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.0166358595194085,
|
| 1251 |
+
"grad_norm": 0.40325233340263367,
|
| 1252 |
+
"learning_rate": 2.4917398336414047e-05,
|
| 1253 |
+
"loss": 0.1666,
|
| 1254 |
+
"step": 8800
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.0224121996303142,
|
| 1258 |
+
"grad_norm": 0.08985109627246857,
|
| 1259 |
+
"learning_rate": 2.488851663585952e-05,
|
| 1260 |
+
"loss": 0.2029,
|
| 1261 |
+
"step": 8850
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.02818853974122,
|
| 1265 |
+
"grad_norm": 0.09904562681913376,
|
| 1266 |
+
"learning_rate": 2.485963493530499e-05,
|
| 1267 |
+
"loss": 0.1419,
|
| 1268 |
+
"step": 8900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.0339648798521257,
|
| 1272 |
+
"grad_norm": 0.47282811999320984,
|
| 1273 |
+
"learning_rate": 2.4830753234750463e-05,
|
| 1274 |
+
"loss": 0.1656,
|
| 1275 |
+
"step": 8950
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.0397412199630314,
|
| 1279 |
+
"grad_norm": 0.31505492329597473,
|
| 1280 |
+
"learning_rate": 2.4801871534195936e-05,
|
| 1281 |
+
"loss": 0.1454,
|
| 1282 |
+
"step": 9000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.0455175600739373,
|
| 1286 |
+
"grad_norm": 0.10816814750432968,
|
| 1287 |
+
"learning_rate": 2.4772989833641405e-05,
|
| 1288 |
+
"loss": 0.1705,
|
| 1289 |
+
"step": 9050
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.051293900184843,
|
| 1293 |
+
"grad_norm": 0.7013025283813477,
|
| 1294 |
+
"learning_rate": 2.4744108133086878e-05,
|
| 1295 |
+
"loss": 0.1612,
|
| 1296 |
+
"step": 9100
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.0570702402957486,
|
| 1300 |
+
"grad_norm": 2.4045467376708984,
|
| 1301 |
+
"learning_rate": 2.4715226432532348e-05,
|
| 1302 |
+
"loss": 0.1552,
|
| 1303 |
+
"step": 9150
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.0628465804066543,
|
| 1307 |
+
"grad_norm": 2.943918466567993,
|
| 1308 |
+
"learning_rate": 2.468634473197782e-05,
|
| 1309 |
+
"loss": 0.1593,
|
| 1310 |
+
"step": 9200
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.0686229205175601,
|
| 1314 |
+
"grad_norm": 0.06787320971488953,
|
| 1315 |
+
"learning_rate": 2.465746303142329e-05,
|
| 1316 |
+
"loss": 0.1493,
|
| 1317 |
+
"step": 9250
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.0743992606284658,
|
| 1321 |
+
"grad_norm": 0.5487985014915466,
|
| 1322 |
+
"learning_rate": 2.4628581330868763e-05,
|
| 1323 |
+
"loss": 0.1707,
|
| 1324 |
+
"step": 9300
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.0801756007393715,
|
| 1328 |
+
"grad_norm": 27.40123176574707,
|
| 1329 |
+
"learning_rate": 2.4599699630314236e-05,
|
| 1330 |
+
"loss": 0.1384,
|
| 1331 |
+
"step": 9350
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.0859519408502774,
|
| 1335 |
+
"grad_norm": 0.041278500109910965,
|
| 1336 |
+
"learning_rate": 2.4570817929759706e-05,
|
| 1337 |
+
"loss": 0.1975,
|
| 1338 |
+
"step": 9400
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.091728280961183,
|
| 1342 |
+
"grad_norm": 35.25811004638672,
|
| 1343 |
+
"learning_rate": 2.454193622920518e-05,
|
| 1344 |
+
"loss": 0.1859,
|
| 1345 |
+
"step": 9450
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0975046210720887,
|
| 1349 |
+
"grad_norm": 22.018983840942383,
|
| 1350 |
+
"learning_rate": 2.451305452865065e-05,
|
| 1351 |
+
"loss": 0.1713,
|
| 1352 |
+
"step": 9500
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.1032809611829943,
|
| 1356 |
+
"grad_norm": 1.6615002155303955,
|
| 1357 |
+
"learning_rate": 2.4484172828096118e-05,
|
| 1358 |
+
"loss": 0.1279,
|
| 1359 |
+
"step": 9550
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.1090573012939002,
|
| 1363 |
+
"grad_norm": 0.05830889567732811,
|
| 1364 |
+
"learning_rate": 2.4455291127541588e-05,
|
| 1365 |
+
"loss": 0.1166,
|
| 1366 |
+
"step": 9600
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.114833641404806,
|
| 1370 |
+
"grad_norm": 0.10341417044401169,
|
| 1371 |
+
"learning_rate": 2.442640942698706e-05,
|
| 1372 |
+
"loss": 0.1639,
|
| 1373 |
+
"step": 9650
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.1206099815157116,
|
| 1377 |
+
"grad_norm": 7.9995646476745605,
|
| 1378 |
+
"learning_rate": 2.439752772643253e-05,
|
| 1379 |
+
"loss": 0.2051,
|
| 1380 |
+
"step": 9700
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.1263863216266174,
|
| 1384 |
+
"grad_norm": 19.472301483154297,
|
| 1385 |
+
"learning_rate": 2.4368646025878003e-05,
|
| 1386 |
+
"loss": 0.1419,
|
| 1387 |
+
"step": 9750
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.1321626617375231,
|
| 1391 |
+
"grad_norm": 0.03694809228181839,
|
| 1392 |
+
"learning_rate": 2.4339764325323476e-05,
|
| 1393 |
+
"loss": 0.1501,
|
| 1394 |
+
"step": 9800
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.1379390018484288,
|
| 1398 |
+
"grad_norm": 9.946338653564453,
|
| 1399 |
+
"learning_rate": 2.4310882624768946e-05,
|
| 1400 |
+
"loss": 0.1779,
|
| 1401 |
+
"step": 9850
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1437153419593344,
|
| 1405 |
+
"grad_norm": 27.467966079711914,
|
| 1406 |
+
"learning_rate": 2.428200092421442e-05,
|
| 1407 |
+
"loss": 0.1526,
|
| 1408 |
+
"step": 9900
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.1494916820702403,
|
| 1412 |
+
"grad_norm": 1.936484694480896,
|
| 1413 |
+
"learning_rate": 2.425311922365989e-05,
|
| 1414 |
+
"loss": 0.1764,
|
| 1415 |
+
"step": 9950
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.155268022181146,
|
| 1419 |
+
"grad_norm": 0.10868394374847412,
|
| 1420 |
+
"learning_rate": 2.422423752310536e-05,
|
| 1421 |
+
"loss": 0.1688,
|
| 1422 |
+
"step": 10000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.1610443622920519,
|
| 1426 |
+
"grad_norm": 0.08838610351085663,
|
| 1427 |
+
"learning_rate": 2.419535582255083e-05,
|
| 1428 |
+
"loss": 0.143,
|
| 1429 |
+
"step": 10050
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.1668207024029575,
|
| 1433 |
+
"grad_norm": 0.09524734318256378,
|
| 1434 |
+
"learning_rate": 2.4166474121996304e-05,
|
| 1435 |
+
"loss": 0.1645,
|
| 1436 |
+
"step": 10100
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.1725970425138632,
|
| 1440 |
+
"grad_norm": 0.1827584058046341,
|
| 1441 |
+
"learning_rate": 2.4137592421441777e-05,
|
| 1442 |
+
"loss": 0.1789,
|
| 1443 |
+
"step": 10150
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.1783733826247689,
|
| 1447 |
+
"grad_norm": 41.8397331237793,
|
| 1448 |
+
"learning_rate": 2.4108710720887247e-05,
|
| 1449 |
+
"loss": 0.1185,
|
| 1450 |
+
"step": 10200
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.1841497227356748,
|
| 1454 |
+
"grad_norm": 0.013732840307056904,
|
| 1455 |
+
"learning_rate": 2.407982902033272e-05,
|
| 1456 |
+
"loss": 0.1326,
|
| 1457 |
+
"step": 10250
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.1899260628465804,
|
| 1461 |
+
"grad_norm": 0.08121947199106216,
|
| 1462 |
+
"learning_rate": 2.405094731977819e-05,
|
| 1463 |
+
"loss": 0.217,
|
| 1464 |
+
"step": 10300
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.195702402957486,
|
| 1468 |
+
"grad_norm": 0.2686977982521057,
|
| 1469 |
+
"learning_rate": 2.4022065619223662e-05,
|
| 1470 |
+
"loss": 0.1536,
|
| 1471 |
+
"step": 10350
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.201478743068392,
|
| 1475 |
+
"grad_norm": 7.599088191986084,
|
| 1476 |
+
"learning_rate": 2.3993183918669132e-05,
|
| 1477 |
+
"loss": 0.1528,
|
| 1478 |
+
"step": 10400
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.2072550831792976,
|
| 1482 |
+
"grad_norm": 0.20336832106113434,
|
| 1483 |
+
"learning_rate": 2.3964302218114605e-05,
|
| 1484 |
+
"loss": 0.1738,
|
| 1485 |
+
"step": 10450
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.2130314232902033,
|
| 1489 |
+
"grad_norm": 10.706509590148926,
|
| 1490 |
+
"learning_rate": 2.3935420517560074e-05,
|
| 1491 |
+
"loss": 0.1623,
|
| 1492 |
+
"step": 10500
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.218807763401109,
|
| 1496 |
+
"grad_norm": 26.202510833740234,
|
| 1497 |
+
"learning_rate": 2.3906538817005544e-05,
|
| 1498 |
+
"loss": 0.1552,
|
| 1499 |
+
"step": 10550
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.2245841035120149,
|
| 1503 |
+
"grad_norm": 22.59147071838379,
|
| 1504 |
+
"learning_rate": 2.3877657116451017e-05,
|
| 1505 |
+
"loss": 0.1325,
|
| 1506 |
+
"step": 10600
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.2303604436229205,
|
| 1510 |
+
"grad_norm": 382.7044677734375,
|
| 1511 |
+
"learning_rate": 2.3848775415896487e-05,
|
| 1512 |
+
"loss": 0.2088,
|
| 1513 |
+
"step": 10650
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2361367837338262,
|
| 1517 |
+
"grad_norm": 10.813858032226562,
|
| 1518 |
+
"learning_rate": 2.381989371534196e-05,
|
| 1519 |
+
"loss": 0.167,
|
| 1520 |
+
"step": 10700
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.241913123844732,
|
| 1524 |
+
"grad_norm": 2.7746658325195312,
|
| 1525 |
+
"learning_rate": 2.379101201478743e-05,
|
| 1526 |
+
"loss": 0.2118,
|
| 1527 |
+
"step": 10750
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.2476894639556377,
|
| 1531 |
+
"grad_norm": 0.11969839036464691,
|
| 1532 |
+
"learning_rate": 2.3762130314232902e-05,
|
| 1533 |
+
"loss": 0.1688,
|
| 1534 |
+
"step": 10800
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.2534658040665434,
|
| 1538 |
+
"grad_norm": 97.15472412109375,
|
| 1539 |
+
"learning_rate": 2.3733248613678375e-05,
|
| 1540 |
+
"loss": 0.1536,
|
| 1541 |
+
"step": 10850
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.259242144177449,
|
| 1545 |
+
"grad_norm": 5.869635581970215,
|
| 1546 |
+
"learning_rate": 2.3704366913123845e-05,
|
| 1547 |
+
"loss": 0.1868,
|
| 1548 |
+
"step": 10900
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.265018484288355,
|
| 1552 |
+
"grad_norm": 0.8501625061035156,
|
| 1553 |
+
"learning_rate": 2.3675485212569318e-05,
|
| 1554 |
+
"loss": 0.1237,
|
| 1555 |
+
"step": 10950
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2707948243992606,
|
| 1559 |
+
"grad_norm": 14.44981575012207,
|
| 1560 |
+
"learning_rate": 2.3646603512014787e-05,
|
| 1561 |
+
"loss": 0.1075,
|
| 1562 |
+
"step": 11000
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2765711645101663,
|
| 1566 |
+
"grad_norm": 0.06192944198846817,
|
| 1567 |
+
"learning_rate": 2.361772181146026e-05,
|
| 1568 |
+
"loss": 0.0926,
|
| 1569 |
+
"step": 11050
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2823475046210722,
|
| 1573 |
+
"grad_norm": 1.1274975538253784,
|
| 1574 |
+
"learning_rate": 2.358884011090573e-05,
|
| 1575 |
+
"loss": 0.1295,
|
| 1576 |
+
"step": 11100
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2881238447319778,
|
| 1580 |
+
"grad_norm": 0.09253960102796555,
|
| 1581 |
+
"learning_rate": 2.3559958410351203e-05,
|
| 1582 |
+
"loss": 0.1629,
|
| 1583 |
+
"step": 11150
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2939001848428835,
|
| 1587 |
+
"grad_norm": 3.5287694931030273,
|
| 1588 |
+
"learning_rate": 2.3531076709796676e-05,
|
| 1589 |
+
"loss": 0.1334,
|
| 1590 |
+
"step": 11200
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2996765249537892,
|
| 1594 |
+
"grad_norm": 0.3958209156990051,
|
| 1595 |
+
"learning_rate": 2.3502195009242146e-05,
|
| 1596 |
+
"loss": 0.2357,
|
| 1597 |
+
"step": 11250
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.305452865064695,
|
| 1601 |
+
"grad_norm": 0.8403951525688171,
|
| 1602 |
+
"learning_rate": 2.347331330868762e-05,
|
| 1603 |
+
"loss": 0.145,
|
| 1604 |
+
"step": 11300
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.3112292051756007,
|
| 1608 |
+
"grad_norm": 18.562786102294922,
|
| 1609 |
+
"learning_rate": 2.3444431608133088e-05,
|
| 1610 |
+
"loss": 0.1489,
|
| 1611 |
+
"step": 11350
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.3170055452865066,
|
| 1615 |
+
"grad_norm": 0.07164066284894943,
|
| 1616 |
+
"learning_rate": 2.341554990757856e-05,
|
| 1617 |
+
"loss": 0.1311,
|
| 1618 |
+
"step": 11400
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.3227818853974123,
|
| 1622 |
+
"grad_norm": 0.009075075387954712,
|
| 1623 |
+
"learning_rate": 2.3386668207024027e-05,
|
| 1624 |
+
"loss": 0.067,
|
| 1625 |
+
"step": 11450
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.328558225508318,
|
| 1629 |
+
"grad_norm": 2.3241634368896484,
|
| 1630 |
+
"learning_rate": 2.33577865064695e-05,
|
| 1631 |
+
"loss": 0.251,
|
| 1632 |
+
"step": 11500
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.3343345656192236,
|
| 1636 |
+
"grad_norm": 74.73931884765625,
|
| 1637 |
+
"learning_rate": 2.332890480591497e-05,
|
| 1638 |
+
"loss": 0.1685,
|
| 1639 |
+
"step": 11550
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.3401109057301293,
|
| 1643 |
+
"grad_norm": 0.22539888322353363,
|
| 1644 |
+
"learning_rate": 2.3300023105360443e-05,
|
| 1645 |
+
"loss": 0.0946,
|
| 1646 |
+
"step": 11600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.3458872458410351,
|
| 1650 |
+
"grad_norm": 0.05199708044528961,
|
| 1651 |
+
"learning_rate": 2.3271141404805916e-05,
|
| 1652 |
+
"loss": 0.0939,
|
| 1653 |
+
"step": 11650
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.3516635859519408,
|
| 1657 |
+
"grad_norm": 0.005744859576225281,
|
| 1658 |
+
"learning_rate": 2.3242259704251386e-05,
|
| 1659 |
+
"loss": 0.146,
|
| 1660 |
+
"step": 11700
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.3574399260628467,
|
| 1664 |
+
"grad_norm": 0.08633283525705338,
|
| 1665 |
+
"learning_rate": 2.321337800369686e-05,
|
| 1666 |
+
"loss": 0.1566,
|
| 1667 |
+
"step": 11750
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.3632162661737524,
|
| 1671 |
+
"grad_norm": 17.302743911743164,
|
| 1672 |
+
"learning_rate": 2.3184496303142328e-05,
|
| 1673 |
+
"loss": 0.1339,
|
| 1674 |
+
"step": 11800
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.368992606284658,
|
| 1678 |
+
"grad_norm": 23.171537399291992,
|
| 1679 |
+
"learning_rate": 2.31556146025878e-05,
|
| 1680 |
+
"loss": 0.1389,
|
| 1681 |
+
"step": 11850
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3747689463955637,
|
| 1685 |
+
"grad_norm": 0.9330076575279236,
|
| 1686 |
+
"learning_rate": 2.312673290203327e-05,
|
| 1687 |
+
"loss": 0.1594,
|
| 1688 |
+
"step": 11900
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.3805452865064696,
|
| 1692 |
+
"grad_norm": 5.958154201507568,
|
| 1693 |
+
"learning_rate": 2.3097851201478744e-05,
|
| 1694 |
+
"loss": 0.1039,
|
| 1695 |
+
"step": 11950
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.3863216266173752,
|
| 1699 |
+
"grad_norm": 0.15630990266799927,
|
| 1700 |
+
"learning_rate": 2.3068969500924217e-05,
|
| 1701 |
+
"loss": 0.2041,
|
| 1702 |
+
"step": 12000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.392097966728281,
|
| 1706 |
+
"grad_norm": 29.334747314453125,
|
| 1707 |
+
"learning_rate": 2.3040087800369686e-05,
|
| 1708 |
+
"loss": 0.1662,
|
| 1709 |
+
"step": 12050
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 1.3978743068391868,
|
| 1713 |
+
"grad_norm": 43.128379821777344,
|
| 1714 |
+
"learning_rate": 2.301120609981516e-05,
|
| 1715 |
+
"loss": 0.0719,
|
| 1716 |
+
"step": 12100
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.4036506469500925,
|
| 1720 |
+
"grad_norm": 0.14302471280097961,
|
| 1721 |
+
"learning_rate": 2.298232439926063e-05,
|
| 1722 |
+
"loss": 0.1473,
|
| 1723 |
+
"step": 12150
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.4094269870609981,
|
| 1727 |
+
"grad_norm": 0.14523327350616455,
|
| 1728 |
+
"learning_rate": 2.2953442698706102e-05,
|
| 1729 |
+
"loss": 0.2012,
|
| 1730 |
+
"step": 12200
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 1.4152033271719038,
|
| 1734 |
+
"grad_norm": 0.19870401918888092,
|
| 1735 |
+
"learning_rate": 2.292456099815157e-05,
|
| 1736 |
+
"loss": 0.1298,
|
| 1737 |
+
"step": 12250
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.4209796672828097,
|
| 1741 |
+
"grad_norm": 8.267024040222168,
|
| 1742 |
+
"learning_rate": 2.2895679297597044e-05,
|
| 1743 |
+
"loss": 0.1229,
|
| 1744 |
+
"step": 12300
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.4267560073937153,
|
| 1748 |
+
"grad_norm": 0.044785793870687485,
|
| 1749 |
+
"learning_rate": 2.2866797597042517e-05,
|
| 1750 |
+
"loss": 0.1727,
|
| 1751 |
+
"step": 12350
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.432532347504621,
|
| 1755 |
+
"grad_norm": 0.45464086532592773,
|
| 1756 |
+
"learning_rate": 2.2837915896487984e-05,
|
| 1757 |
+
"loss": 0.1769,
|
| 1758 |
+
"step": 12400
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.4383086876155269,
|
| 1762 |
+
"grad_norm": 45.37421798706055,
|
| 1763 |
+
"learning_rate": 2.2809034195933457e-05,
|
| 1764 |
+
"loss": 0.2269,
|
| 1765 |
+
"step": 12450
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.4440850277264325,
|
| 1769 |
+
"grad_norm": 0.010675471276044846,
|
| 1770 |
+
"learning_rate": 2.2780152495378926e-05,
|
| 1771 |
+
"loss": 0.1045,
|
| 1772 |
+
"step": 12500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.4498613678373382,
|
| 1776 |
+
"grad_norm": 0.5008090138435364,
|
| 1777 |
+
"learning_rate": 2.27512707948244e-05,
|
| 1778 |
+
"loss": 0.1331,
|
| 1779 |
+
"step": 12550
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.4556377079482439,
|
| 1783 |
+
"grad_norm": 0.010356269776821136,
|
| 1784 |
+
"learning_rate": 2.272238909426987e-05,
|
| 1785 |
+
"loss": 0.1148,
|
| 1786 |
+
"step": 12600
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.4614140480591498,
|
| 1790 |
+
"grad_norm": 2.318436622619629,
|
| 1791 |
+
"learning_rate": 2.2693507393715342e-05,
|
| 1792 |
+
"loss": 0.1796,
|
| 1793 |
+
"step": 12650
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4671903881700554,
|
| 1797 |
+
"grad_norm": 0.06255066394805908,
|
| 1798 |
+
"learning_rate": 2.2664625693160815e-05,
|
| 1799 |
+
"loss": 0.1387,
|
| 1800 |
+
"step": 12700
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.4729667282809613,
|
| 1804 |
+
"grad_norm": 0.02640642039477825,
|
| 1805 |
+
"learning_rate": 2.2635743992606284e-05,
|
| 1806 |
+
"loss": 0.1746,
|
| 1807 |
+
"step": 12750
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.478743068391867,
|
| 1811 |
+
"grad_norm": 37.81770706176758,
|
| 1812 |
+
"learning_rate": 2.2606862292051757e-05,
|
| 1813 |
+
"loss": 0.1362,
|
| 1814 |
+
"step": 12800
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.4845194085027726,
|
| 1818 |
+
"grad_norm": 0.0361916609108448,
|
| 1819 |
+
"learning_rate": 2.2577980591497227e-05,
|
| 1820 |
+
"loss": 0.1773,
|
| 1821 |
+
"step": 12850
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.4902957486136783,
|
| 1825 |
+
"grad_norm": 0.04615430161356926,
|
| 1826 |
+
"learning_rate": 2.25490988909427e-05,
|
| 1827 |
+
"loss": 0.1558,
|
| 1828 |
+
"step": 12900
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 1.496072088724584,
|
| 1832 |
+
"grad_norm": 54.66000747680664,
|
| 1833 |
+
"learning_rate": 2.252021719038817e-05,
|
| 1834 |
+
"loss": 0.2207,
|
| 1835 |
+
"step": 12950
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 1.5018484288354899,
|
| 1839 |
+
"grad_norm": 7.9861836433410645,
|
| 1840 |
+
"learning_rate": 2.2491335489833643e-05,
|
| 1841 |
+
"loss": 0.133,
|
| 1842 |
+
"step": 13000
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.5076247689463955,
|
| 1846 |
+
"grad_norm": 0.050192106515169144,
|
| 1847 |
+
"learning_rate": 2.2462453789279116e-05,
|
| 1848 |
+
"loss": 0.1515,
|
| 1849 |
+
"step": 13050
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.5134011090573014,
|
| 1853 |
+
"grad_norm": 16.963003158569336,
|
| 1854 |
+
"learning_rate": 2.2433572088724585e-05,
|
| 1855 |
+
"loss": 0.1541,
|
| 1856 |
+
"step": 13100
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.519177449168207,
|
| 1860 |
+
"grad_norm": 0.16849014163017273,
|
| 1861 |
+
"learning_rate": 2.2404690388170058e-05,
|
| 1862 |
+
"loss": 0.1627,
|
| 1863 |
+
"step": 13150
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.5249537892791127,
|
| 1867 |
+
"grad_norm": 0.046051234006881714,
|
| 1868 |
+
"learning_rate": 2.2375808687615528e-05,
|
| 1869 |
+
"loss": 0.1728,
|
| 1870 |
+
"step": 13200
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.5307301293900184,
|
| 1874 |
+
"grad_norm": 0.03939524292945862,
|
| 1875 |
+
"learning_rate": 2.2346926987061e-05,
|
| 1876 |
+
"loss": 0.1402,
|
| 1877 |
+
"step": 13250
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 1.536506469500924,
|
| 1881 |
+
"grad_norm": 19.971302032470703,
|
| 1882 |
+
"learning_rate": 2.231804528650647e-05,
|
| 1883 |
+
"loss": 0.1563,
|
| 1884 |
+
"step": 13300
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.54228280961183,
|
| 1888 |
+
"grad_norm": 1.2224446535110474,
|
| 1889 |
+
"learning_rate": 2.228916358595194e-05,
|
| 1890 |
+
"loss": 0.148,
|
| 1891 |
+
"step": 13350
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.5480591497227358,
|
| 1895 |
+
"grad_norm": 10.81811237335205,
|
| 1896 |
+
"learning_rate": 2.2260281885397413e-05,
|
| 1897 |
+
"loss": 0.1474,
|
| 1898 |
+
"step": 13400
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 1.5538354898336415,
|
| 1902 |
+
"grad_norm": 2.6338510513305664,
|
| 1903 |
+
"learning_rate": 2.2231400184842883e-05,
|
| 1904 |
+
"loss": 0.1476,
|
| 1905 |
+
"step": 13450
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5596118299445472,
|
| 1909 |
+
"grad_norm": 0.10376719385385513,
|
| 1910 |
+
"learning_rate": 2.2202518484288356e-05,
|
| 1911 |
+
"loss": 0.0925,
|
| 1912 |
+
"step": 13500
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 1.5653881700554528,
|
| 1916 |
+
"grad_norm": 0.14928467571735382,
|
| 1917 |
+
"learning_rate": 2.2173636783733825e-05,
|
| 1918 |
+
"loss": 0.1362,
|
| 1919 |
+
"step": 13550
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 1.5711645101663585,
|
| 1923 |
+
"grad_norm": 1.1348618268966675,
|
| 1924 |
+
"learning_rate": 2.2144755083179298e-05,
|
| 1925 |
+
"loss": 0.1229,
|
| 1926 |
+
"step": 13600
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.5769408502772642,
|
| 1930 |
+
"grad_norm": 0.09609614312648773,
|
| 1931 |
+
"learning_rate": 2.2115873382624768e-05,
|
| 1932 |
+
"loss": 0.1732,
|
| 1933 |
+
"step": 13650
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.58271719038817,
|
| 1937 |
+
"grad_norm": 20.673297882080078,
|
| 1938 |
+
"learning_rate": 2.208699168207024e-05,
|
| 1939 |
+
"loss": 0.1249,
|
| 1940 |
+
"step": 13700
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 1.588493530499076,
|
| 1944 |
+
"grad_norm": 82.14835357666016,
|
| 1945 |
+
"learning_rate": 2.205810998151571e-05,
|
| 1946 |
+
"loss": 0.0819,
|
| 1947 |
+
"step": 13750
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.5942698706099816,
|
| 1951 |
+
"grad_norm": 25.9471435546875,
|
| 1952 |
+
"learning_rate": 2.2029228280961183e-05,
|
| 1953 |
+
"loss": 0.1534,
|
| 1954 |
+
"step": 13800
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 1.6000462107208873,
|
| 1958 |
+
"grad_norm": 11.10772705078125,
|
| 1959 |
+
"learning_rate": 2.2000346580406656e-05,
|
| 1960 |
+
"loss": 0.2186,
|
| 1961 |
+
"step": 13850
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.605822550831793,
|
| 1965 |
+
"grad_norm": 0.007331051863729954,
|
| 1966 |
+
"learning_rate": 2.1971464879852126e-05,
|
| 1967 |
+
"loss": 0.1159,
|
| 1968 |
+
"step": 13900
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.6115988909426986,
|
| 1972 |
+
"grad_norm": 1.7013061046600342,
|
| 1973 |
+
"learning_rate": 2.19425831792976e-05,
|
| 1974 |
+
"loss": 0.1857,
|
| 1975 |
+
"step": 13950
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.6173752310536045,
|
| 1979 |
+
"grad_norm": 19.55061149597168,
|
| 1980 |
+
"learning_rate": 2.191370147874307e-05,
|
| 1981 |
+
"loss": 0.1657,
|
| 1982 |
+
"step": 14000
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 1.6231515711645101,
|
| 1986 |
+
"grad_norm": 0.03729844093322754,
|
| 1987 |
+
"learning_rate": 2.188481977818854e-05,
|
| 1988 |
+
"loss": 0.1275,
|
| 1989 |
+
"step": 14050
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.628927911275416,
|
| 1993 |
+
"grad_norm": 65.23085021972656,
|
| 1994 |
+
"learning_rate": 2.185593807763401e-05,
|
| 1995 |
+
"loss": 0.1474,
|
| 1996 |
+
"step": 14100
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 1.6347042513863217,
|
| 2000 |
+
"grad_norm": 16.47488021850586,
|
| 2001 |
+
"learning_rate": 2.1827056377079484e-05,
|
| 2002 |
+
"loss": 0.0967,
|
| 2003 |
+
"step": 14150
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 1.6404805914972274,
|
| 2007 |
+
"grad_norm": 39.40799331665039,
|
| 2008 |
+
"learning_rate": 2.1798174676524957e-05,
|
| 2009 |
+
"loss": 0.1652,
|
| 2010 |
+
"step": 14200
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.646256931608133,
|
| 2014 |
+
"grad_norm": 20.859216690063477,
|
| 2015 |
+
"learning_rate": 2.1769292975970427e-05,
|
| 2016 |
+
"loss": 0.1453,
|
| 2017 |
+
"step": 14250
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.6520332717190387,
|
| 2021 |
+
"grad_norm": 0.1275290995836258,
|
| 2022 |
+
"learning_rate": 2.1740411275415896e-05,
|
| 2023 |
+
"loss": 0.1268,
|
| 2024 |
+
"step": 14300
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 1.6578096118299446,
|
| 2028 |
+
"grad_norm": 0.44121667742729187,
|
| 2029 |
+
"learning_rate": 2.1711529574861366e-05,
|
| 2030 |
+
"loss": 0.1056,
|
| 2031 |
+
"step": 14350
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.6635859519408502,
|
| 2035 |
+
"grad_norm": 0.2817210853099823,
|
| 2036 |
+
"learning_rate": 2.168264787430684e-05,
|
| 2037 |
+
"loss": 0.1781,
|
| 2038 |
+
"step": 14400
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 1.6693622920517561,
|
| 2042 |
+
"grad_norm": 3.9421677589416504,
|
| 2043 |
+
"learning_rate": 2.165376617375231e-05,
|
| 2044 |
+
"loss": 0.1279,
|
| 2045 |
+
"step": 14450
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 1.6751386321626618,
|
| 2049 |
+
"grad_norm": 0.23300153017044067,
|
| 2050 |
+
"learning_rate": 2.162488447319778e-05,
|
| 2051 |
+
"loss": 0.1512,
|
| 2052 |
+
"step": 14500
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.6809149722735675,
|
| 2056 |
+
"grad_norm": 0.21775715053081512,
|
| 2057 |
+
"learning_rate": 2.1596002772643254e-05,
|
| 2058 |
+
"loss": 0.1837,
|
| 2059 |
+
"step": 14550
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.6866913123844731,
|
| 2063 |
+
"grad_norm": 18.65435218811035,
|
| 2064 |
+
"learning_rate": 2.1567121072088724e-05,
|
| 2065 |
+
"loss": 0.0867,
|
| 2066 |
+
"step": 14600
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.6924676524953788,
|
| 2070 |
+
"grad_norm": 0.022388514131307602,
|
| 2071 |
+
"learning_rate": 2.1538239371534197e-05,
|
| 2072 |
+
"loss": 0.089,
|
| 2073 |
+
"step": 14650
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6982439926062847,
|
| 2077 |
+
"grad_norm": 0.10797044634819031,
|
| 2078 |
+
"learning_rate": 2.1509357670979667e-05,
|
| 2079 |
+
"loss": 0.1143,
|
| 2080 |
+
"step": 14700
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.7040203327171903,
|
| 2084 |
+
"grad_norm": 15.600735664367676,
|
| 2085 |
+
"learning_rate": 2.148047597042514e-05,
|
| 2086 |
+
"loss": 0.1299,
|
| 2087 |
+
"step": 14750
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.7097966728280962,
|
| 2091 |
+
"grad_norm": 34.40959548950195,
|
| 2092 |
+
"learning_rate": 2.145159426987061e-05,
|
| 2093 |
+
"loss": 0.2121,
|
| 2094 |
+
"step": 14800
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.7155730129390019,
|
| 2098 |
+
"grad_norm": 6.029714107513428,
|
| 2099 |
+
"learning_rate": 2.1422712569316082e-05,
|
| 2100 |
+
"loss": 0.1569,
|
| 2101 |
+
"step": 14850
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.7213493530499075,
|
| 2105 |
+
"grad_norm": 0.01677633635699749,
|
| 2106 |
+
"learning_rate": 2.1393830868761555e-05,
|
| 2107 |
+
"loss": 0.1182,
|
| 2108 |
+
"step": 14900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.7271256931608132,
|
| 2112 |
+
"grad_norm": 0.07976578921079636,
|
| 2113 |
+
"learning_rate": 2.1364949168207025e-05,
|
| 2114 |
+
"loss": 0.1374,
|
| 2115 |
+
"step": 14950
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.7329020332717189,
|
| 2119 |
+
"grad_norm": 0.49204331636428833,
|
| 2120 |
+
"learning_rate": 2.1336067467652498e-05,
|
| 2121 |
+
"loss": 0.0988,
|
| 2122 |
+
"step": 15000
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.7386783733826248,
|
| 2126 |
+
"grad_norm": 0.05830137059092522,
|
| 2127 |
+
"learning_rate": 2.1307185767097967e-05,
|
| 2128 |
+
"loss": 0.1321,
|
| 2129 |
+
"step": 15050
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.7444547134935307,
|
| 2133 |
+
"grad_norm": 0.33919116854667664,
|
| 2134 |
+
"learning_rate": 2.127830406654344e-05,
|
| 2135 |
+
"loss": 0.1756,
|
| 2136 |
+
"step": 15100
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.7502310536044363,
|
| 2140 |
+
"grad_norm": 0.03737005218863487,
|
| 2141 |
+
"learning_rate": 2.124942236598891e-05,
|
| 2142 |
+
"loss": 0.0943,
|
| 2143 |
+
"step": 15150
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.756007393715342,
|
| 2147 |
+
"grad_norm": 11.36135482788086,
|
| 2148 |
+
"learning_rate": 2.1220540665434383e-05,
|
| 2149 |
+
"loss": 0.1554,
|
| 2150 |
+
"step": 15200
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.7617837338262476,
|
| 2154 |
+
"grad_norm": 29.172395706176758,
|
| 2155 |
+
"learning_rate": 2.1191658964879853e-05,
|
| 2156 |
+
"loss": 0.1317,
|
| 2157 |
+
"step": 15250
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.7675600739371533,
|
| 2161 |
+
"grad_norm": 0.4681268334388733,
|
| 2162 |
+
"learning_rate": 2.1162777264325322e-05,
|
| 2163 |
+
"loss": 0.1342,
|
| 2164 |
+
"step": 15300
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.7733364140480592,
|
| 2168 |
+
"grad_norm": 0.19265428185462952,
|
| 2169 |
+
"learning_rate": 2.1133895563770795e-05,
|
| 2170 |
+
"loss": 0.1198,
|
| 2171 |
+
"step": 15350
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.7791127541589649,
|
| 2175 |
+
"grad_norm": 0.38005658984184265,
|
| 2176 |
+
"learning_rate": 2.1105013863216265e-05,
|
| 2177 |
+
"loss": 0.1178,
|
| 2178 |
+
"step": 15400
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.7848890942698707,
|
| 2182 |
+
"grad_norm": 0.6070498824119568,
|
| 2183 |
+
"learning_rate": 2.1076132162661738e-05,
|
| 2184 |
+
"loss": 0.1597,
|
| 2185 |
+
"step": 15450
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7906654343807764,
|
| 2189 |
+
"grad_norm": 0.13323445618152618,
|
| 2190 |
+
"learning_rate": 2.1047250462107207e-05,
|
| 2191 |
+
"loss": 0.1496,
|
| 2192 |
+
"step": 15500
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.796441774491682,
|
| 2196 |
+
"grad_norm": 39.70579147338867,
|
| 2197 |
+
"learning_rate": 2.101836876155268e-05,
|
| 2198 |
+
"loss": 0.1734,
|
| 2199 |
+
"step": 15550
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.8022181146025877,
|
| 2203 |
+
"grad_norm": 0.04307672381401062,
|
| 2204 |
+
"learning_rate": 2.0989487060998153e-05,
|
| 2205 |
+
"loss": 0.0877,
|
| 2206 |
+
"step": 15600
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.8079944547134934,
|
| 2210 |
+
"grad_norm": 8.399633407592773,
|
| 2211 |
+
"learning_rate": 2.0960605360443623e-05,
|
| 2212 |
+
"loss": 0.1458,
|
| 2213 |
+
"step": 15650
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.8137707948243993,
|
| 2217 |
+
"grad_norm": 0.07565668225288391,
|
| 2218 |
+
"learning_rate": 2.0931723659889096e-05,
|
| 2219 |
+
"loss": 0.1264,
|
| 2220 |
+
"step": 15700
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.819547134935305,
|
| 2224 |
+
"grad_norm": 45.78880310058594,
|
| 2225 |
+
"learning_rate": 2.0902841959334566e-05,
|
| 2226 |
+
"loss": 0.1882,
|
| 2227 |
+
"step": 15750
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.8253234750462108,
|
| 2231 |
+
"grad_norm": 56.75712966918945,
|
| 2232 |
+
"learning_rate": 2.087396025878004e-05,
|
| 2233 |
+
"loss": 0.1071,
|
| 2234 |
+
"step": 15800
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.8310998151571165,
|
| 2238 |
+
"grad_norm": 0.1278388649225235,
|
| 2239 |
+
"learning_rate": 2.0845078558225508e-05,
|
| 2240 |
+
"loss": 0.1696,
|
| 2241 |
+
"step": 15850
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.8368761552680222,
|
| 2245 |
+
"grad_norm": 0.5721489787101746,
|
| 2246 |
+
"learning_rate": 2.081619685767098e-05,
|
| 2247 |
+
"loss": 0.139,
|
| 2248 |
+
"step": 15900
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.8426524953789278,
|
| 2252 |
+
"grad_norm": 46.980064392089844,
|
| 2253 |
+
"learning_rate": 2.078731515711645e-05,
|
| 2254 |
+
"loss": 0.1538,
|
| 2255 |
+
"step": 15950
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.8484288354898335,
|
| 2259 |
+
"grad_norm": 0.10779360681772232,
|
| 2260 |
+
"learning_rate": 2.0758433456561924e-05,
|
| 2261 |
+
"loss": 0.0843,
|
| 2262 |
+
"step": 16000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.8542051756007394,
|
| 2266 |
+
"grad_norm": 56.32892608642578,
|
| 2267 |
+
"learning_rate": 2.0729551756007397e-05,
|
| 2268 |
+
"loss": 0.1689,
|
| 2269 |
+
"step": 16050
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.859981515711645,
|
| 2273 |
+
"grad_norm": 0.046350061893463135,
|
| 2274 |
+
"learning_rate": 2.0700670055452866e-05,
|
| 2275 |
+
"loss": 0.1027,
|
| 2276 |
+
"step": 16100
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.865757855822551,
|
| 2280 |
+
"grad_norm": 58.256988525390625,
|
| 2281 |
+
"learning_rate": 2.067178835489834e-05,
|
| 2282 |
+
"loss": 0.0697,
|
| 2283 |
+
"step": 16150
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.8715341959334566,
|
| 2287 |
+
"grad_norm": 0.11341376602649689,
|
| 2288 |
+
"learning_rate": 2.0642906654343806e-05,
|
| 2289 |
+
"loss": 0.0864,
|
| 2290 |
+
"step": 16200
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.8773105360443623,
|
| 2294 |
+
"grad_norm": 3.484691858291626,
|
| 2295 |
+
"learning_rate": 2.061402495378928e-05,
|
| 2296 |
+
"loss": 0.2117,
|
| 2297 |
+
"step": 16250
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.883086876155268,
|
| 2301 |
+
"grad_norm": 0.03376946225762367,
|
| 2302 |
+
"learning_rate": 2.0585143253234748e-05,
|
| 2303 |
+
"loss": 0.1297,
|
| 2304 |
+
"step": 16300
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.8888632162661736,
|
| 2308 |
+
"grad_norm": 1.6973594427108765,
|
| 2309 |
+
"learning_rate": 2.055626155268022e-05,
|
| 2310 |
+
"loss": 0.1192,
|
| 2311 |
+
"step": 16350
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.8946395563770795,
|
| 2315 |
+
"grad_norm": 0.3922349512577057,
|
| 2316 |
+
"learning_rate": 2.0527379852125694e-05,
|
| 2317 |
+
"loss": 0.128,
|
| 2318 |
+
"step": 16400
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.9004158964879854,
|
| 2322 |
+
"grad_norm": 0.010527100414037704,
|
| 2323 |
+
"learning_rate": 2.0498498151571164e-05,
|
| 2324 |
+
"loss": 0.1211,
|
| 2325 |
+
"step": 16450
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.906192236598891,
|
| 2329 |
+
"grad_norm": 1.1090141534805298,
|
| 2330 |
+
"learning_rate": 2.0469616451016637e-05,
|
| 2331 |
+
"loss": 0.1871,
|
| 2332 |
+
"step": 16500
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.9119685767097967,
|
| 2336 |
+
"grad_norm": 1.5275694131851196,
|
| 2337 |
+
"learning_rate": 2.0440734750462106e-05,
|
| 2338 |
+
"loss": 0.1253,
|
| 2339 |
+
"step": 16550
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.9177449168207024,
|
| 2343 |
+
"grad_norm": 35.891883850097656,
|
| 2344 |
+
"learning_rate": 2.041185304990758e-05,
|
| 2345 |
+
"loss": 0.1592,
|
| 2346 |
+
"step": 16600
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.923521256931608,
|
| 2350 |
+
"grad_norm": 0.06564353406429291,
|
| 2351 |
+
"learning_rate": 2.038297134935305e-05,
|
| 2352 |
+
"loss": 0.099,
|
| 2353 |
+
"step": 16650
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.929297597042514,
|
| 2357 |
+
"grad_norm": 0.9027801752090454,
|
| 2358 |
+
"learning_rate": 2.0354089648798522e-05,
|
| 2359 |
+
"loss": 0.2061,
|
| 2360 |
+
"step": 16700
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.9350739371534196,
|
| 2364 |
+
"grad_norm": 0.034485165029764175,
|
| 2365 |
+
"learning_rate": 2.0325207948243995e-05,
|
| 2366 |
+
"loss": 0.0978,
|
| 2367 |
+
"step": 16750
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.9408502772643255,
|
| 2371 |
+
"grad_norm": 52.34895706176758,
|
| 2372 |
+
"learning_rate": 2.0296326247689464e-05,
|
| 2373 |
+
"loss": 0.1496,
|
| 2374 |
+
"step": 16800
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.9466266173752311,
|
| 2378 |
+
"grad_norm": 54.330509185791016,
|
| 2379 |
+
"learning_rate": 2.0267444547134937e-05,
|
| 2380 |
+
"loss": 0.1722,
|
| 2381 |
+
"step": 16850
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.9524029574861368,
|
| 2385 |
+
"grad_norm": 20.613256454467773,
|
| 2386 |
+
"learning_rate": 2.0238562846580407e-05,
|
| 2387 |
+
"loss": 0.1469,
|
| 2388 |
+
"step": 16900
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.9581792975970425,
|
| 2392 |
+
"grad_norm": 0.04331992566585541,
|
| 2393 |
+
"learning_rate": 2.020968114602588e-05,
|
| 2394 |
+
"loss": 0.1067,
|
| 2395 |
+
"step": 16950
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.9639556377079481,
|
| 2399 |
+
"grad_norm": 0.056098178029060364,
|
| 2400 |
+
"learning_rate": 2.018079944547135e-05,
|
| 2401 |
+
"loss": 0.1385,
|
| 2402 |
+
"step": 17000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.969731977818854,
|
| 2406 |
+
"grad_norm": 0.038275789469480515,
|
| 2407 |
+
"learning_rate": 2.0151917744916823e-05,
|
| 2408 |
+
"loss": 0.0932,
|
| 2409 |
+
"step": 17050
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.9755083179297597,
|
| 2413 |
+
"grad_norm": 0.04682379588484764,
|
| 2414 |
+
"learning_rate": 2.0123036044362296e-05,
|
| 2415 |
+
"loss": 0.1163,
|
| 2416 |
+
"step": 17100
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.9812846580406656,
|
| 2420 |
+
"grad_norm": 0.13566389679908752,
|
| 2421 |
+
"learning_rate": 2.0094154343807762e-05,
|
| 2422 |
+
"loss": 0.1669,
|
| 2423 |
+
"step": 17150
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.9870609981515712,
|
| 2427 |
+
"grad_norm": 36.21744155883789,
|
| 2428 |
+
"learning_rate": 2.0065272643253235e-05,
|
| 2429 |
+
"loss": 0.1608,
|
| 2430 |
+
"step": 17200
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.9928373382624769,
|
| 2434 |
+
"grad_norm": 0.40274855494499207,
|
| 2435 |
+
"learning_rate": 2.0036390942698704e-05,
|
| 2436 |
+
"loss": 0.1038,
|
| 2437 |
+
"step": 17250
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.9986136783733826,
|
| 2441 |
+
"grad_norm": 0.07136744260787964,
|
| 2442 |
+
"learning_rate": 2.0007509242144177e-05,
|
| 2443 |
+
"loss": 0.1248,
|
| 2444 |
+
"step": 17300
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 2.0,
|
| 2448 |
+
"eval_accuracy": 0.9578947368421052,
|
| 2449 |
+
"eval_f1": 0.9746439192361872,
|
| 2450 |
+
"eval_loss": 0.18152591586112976,
|
| 2451 |
+
"eval_precision": 0.9740340997966526,
|
| 2452 |
+
"eval_recall": 0.9752545027407987,
|
| 2453 |
+
"eval_runtime": 66.3035,
|
| 2454 |
+
"eval_samples_per_second": 116.057,
|
| 2455 |
+
"eval_steps_per_second": 14.509,
|
| 2456 |
+
"step": 17312
|
| 2457 |
+
}
|
| 2458 |
+
],
|
| 2459 |
+
"logging_steps": 50,
|
| 2460 |
+
"max_steps": 51936,
|
| 2461 |
+
"num_input_tokens_seen": 0,
|
| 2462 |
+
"num_train_epochs": 6,
|
| 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 sha256:3ed4efe313ce9a9014df973dc998f93f19f39063b7c4ab2e29800b3c9cc3ad65
|
| 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 sha256:4d946842ea5e56464984db938c144ae724913bea1a9dfb43be4a889cf5d8ba4d
|
| 3 |
+
size 631399808
|
checkpoint-25968/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1743a48b005c62c9e72be9ea2ab434f1002379c90dd13d8cad5f6030703f3228
|
| 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 sha256:4058715dbc8cc93c31cf88f6fb79e6360748138105f4b593254ba8a716291429
|
| 3 |
+
size 1383
|
checkpoint-25968/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbe637fd8200f1d732db50cea665a73081bb6a9cb5b3e19925b2168593ad240b
|
| 3 |
+
size 1465
|
checkpoint-25968/trainer_state.json
ADDED
|
@@ -0,0 +1,3703 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 25968,
|
| 3 |
+
"best_metric": 0.9778571983528863,
|
| 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": 8.07317066192627,
|
| 15 |
+
"learning_rate": 2.9971695933456562e-05,
|
| 16 |
+
"loss": 2.9896,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01155268022181146,
|
| 21 |
+
"grad_norm": NaN,
|
| 22 |
+
"learning_rate": 2.994281423290203e-05,
|
| 23 |
+
"loss": 0.3707,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01732902033271719,
|
| 28 |
+
"grad_norm": 14.865354537963867,
|
| 29 |
+
"learning_rate": 2.9913932532347504e-05,
|
| 30 |
+
"loss": 0.3821,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02310536044362292,
|
| 35 |
+
"grad_norm": 14.388459205627441,
|
| 36 |
+
"learning_rate": 2.9885050831792977e-05,
|
| 37 |
+
"loss": 0.318,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028881700554528652,
|
| 42 |
+
"grad_norm": 33.513973236083984,
|
| 43 |
+
"learning_rate": 2.9856169131238447e-05,
|
| 44 |
+
"loss": 0.3427,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03465804066543438,
|
| 49 |
+
"grad_norm": 14.235688209533691,
|
| 50 |
+
"learning_rate": 2.982728743068392e-05,
|
| 51 |
+
"loss": 0.4154,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04043438077634011,
|
| 56 |
+
"grad_norm": 2.8046276569366455,
|
| 57 |
+
"learning_rate": 2.979840573012939e-05,
|
| 58 |
+
"loss": 0.351,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04621072088724584,
|
| 63 |
+
"grad_norm": 15.46557903289795,
|
| 64 |
+
"learning_rate": 2.9769524029574863e-05,
|
| 65 |
+
"loss": 0.3631,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.05198706099815157,
|
| 70 |
+
"grad_norm": 36.24979019165039,
|
| 71 |
+
"learning_rate": 2.9740642329020332e-05,
|
| 72 |
+
"loss": 0.346,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.057763401109057304,
|
| 77 |
+
"grad_norm": 11.466980934143066,
|
| 78 |
+
"learning_rate": 2.9711760628465805e-05,
|
| 79 |
+
"loss": 0.2994,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06353974121996303,
|
| 84 |
+
"grad_norm": 15.98156452178955,
|
| 85 |
+
"learning_rate": 2.9682878927911278e-05,
|
| 86 |
+
"loss": 0.2899,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06931608133086876,
|
| 91 |
+
"grad_norm": 29.2628116607666,
|
| 92 |
+
"learning_rate": 2.9653997227356748e-05,
|
| 93 |
+
"loss": 0.3105,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0750924214417745,
|
| 98 |
+
"grad_norm": 7.130237579345703,
|
| 99 |
+
"learning_rate": 2.962511552680222e-05,
|
| 100 |
+
"loss": 0.2463,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08086876155268022,
|
| 105 |
+
"grad_norm": 12.123393058776855,
|
| 106 |
+
"learning_rate": 2.959623382624769e-05,
|
| 107 |
+
"loss": 0.2928,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08664510166358595,
|
| 112 |
+
"grad_norm": 19.521760940551758,
|
| 113 |
+
"learning_rate": 2.9567352125693163e-05,
|
| 114 |
+
"loss": 0.3383,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09242144177449169,
|
| 119 |
+
"grad_norm": 0.5567889213562012,
|
| 120 |
+
"learning_rate": 2.9538470425138633e-05,
|
| 121 |
+
"loss": 0.2872,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0981977818853974,
|
| 126 |
+
"grad_norm": 14.521562576293945,
|
| 127 |
+
"learning_rate": 2.9509588724584106e-05,
|
| 128 |
+
"loss": 0.3419,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10397412199630314,
|
| 133 |
+
"grad_norm": 30.65926170349121,
|
| 134 |
+
"learning_rate": 2.9480707024029576e-05,
|
| 135 |
+
"loss": 0.2911,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.10975046210720887,
|
| 140 |
+
"grad_norm": 6.595367908477783,
|
| 141 |
+
"learning_rate": 2.9451825323475045e-05,
|
| 142 |
+
"loss": 0.2741,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.11552680221811461,
|
| 147 |
+
"grad_norm": 1.3052228689193726,
|
| 148 |
+
"learning_rate": 2.9422943622920518e-05,
|
| 149 |
+
"loss": 0.2869,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.12130314232902033,
|
| 154 |
+
"grad_norm": 26.094087600708008,
|
| 155 |
+
"learning_rate": 2.9394061922365988e-05,
|
| 156 |
+
"loss": 0.3495,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12707948243992606,
|
| 161 |
+
"grad_norm": 19.971145629882812,
|
| 162 |
+
"learning_rate": 2.936518022181146e-05,
|
| 163 |
+
"loss": 0.3805,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13285582255083178,
|
| 168 |
+
"grad_norm": 6.1732563972473145,
|
| 169 |
+
"learning_rate": 2.933629852125693e-05,
|
| 170 |
+
"loss": 0.2168,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.13863216266173753,
|
| 175 |
+
"grad_norm": 18.476938247680664,
|
| 176 |
+
"learning_rate": 2.9307416820702403e-05,
|
| 177 |
+
"loss": 0.31,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14440850277264325,
|
| 182 |
+
"grad_norm": 11.29546070098877,
|
| 183 |
+
"learning_rate": 2.9278535120147873e-05,
|
| 184 |
+
"loss": 0.3223,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.150184842883549,
|
| 189 |
+
"grad_norm": 58.089473724365234,
|
| 190 |
+
"learning_rate": 2.9249653419593346e-05,
|
| 191 |
+
"loss": 0.2283,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.15596118299445472,
|
| 196 |
+
"grad_norm": 0.983078122138977,
|
| 197 |
+
"learning_rate": 2.922077171903882e-05,
|
| 198 |
+
"loss": 0.3208,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.16173752310536044,
|
| 203 |
+
"grad_norm": 13.572915077209473,
|
| 204 |
+
"learning_rate": 2.919189001848429e-05,
|
| 205 |
+
"loss": 0.2801,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.16751386321626618,
|
| 210 |
+
"grad_norm": 1.8694497346878052,
|
| 211 |
+
"learning_rate": 2.916300831792976e-05,
|
| 212 |
+
"loss": 0.3147,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1732902033271719,
|
| 217 |
+
"grad_norm": 2.4720425605773926,
|
| 218 |
+
"learning_rate": 2.913412661737523e-05,
|
| 219 |
+
"loss": 0.261,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.17906654343807762,
|
| 224 |
+
"grad_norm": 24.087940216064453,
|
| 225 |
+
"learning_rate": 2.9105244916820704e-05,
|
| 226 |
+
"loss": 0.2689,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.18484288354898337,
|
| 231 |
+
"grad_norm": 12.160676956176758,
|
| 232 |
+
"learning_rate": 2.9076363216266174e-05,
|
| 233 |
+
"loss": 0.3336,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1906192236598891,
|
| 238 |
+
"grad_norm": 34.87266540527344,
|
| 239 |
+
"learning_rate": 2.9047481515711647e-05,
|
| 240 |
+
"loss": 0.2899,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.1963955637707948,
|
| 245 |
+
"grad_norm": 6.795339107513428,
|
| 246 |
+
"learning_rate": 2.901859981515712e-05,
|
| 247 |
+
"loss": 0.2453,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.20217190388170056,
|
| 252 |
+
"grad_norm": 25.78632354736328,
|
| 253 |
+
"learning_rate": 2.898971811460259e-05,
|
| 254 |
+
"loss": 0.2297,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.20794824399260628,
|
| 259 |
+
"grad_norm": 0.665915846824646,
|
| 260 |
+
"learning_rate": 2.8960836414048062e-05,
|
| 261 |
+
"loss": 0.3941,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.21372458410351203,
|
| 266 |
+
"grad_norm": 26.92215919494629,
|
| 267 |
+
"learning_rate": 2.8931954713493532e-05,
|
| 268 |
+
"loss": 0.2736,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21950092421441775,
|
| 273 |
+
"grad_norm": 15.98168659210205,
|
| 274 |
+
"learning_rate": 2.8903073012939e-05,
|
| 275 |
+
"loss": 0.241,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.22527726432532347,
|
| 280 |
+
"grad_norm": 14.969513893127441,
|
| 281 |
+
"learning_rate": 2.887419131238447e-05,
|
| 282 |
+
"loss": 0.3063,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.23105360443622922,
|
| 287 |
+
"grad_norm": 2.79641056060791,
|
| 288 |
+
"learning_rate": 2.8845309611829944e-05,
|
| 289 |
+
"loss": 0.3148,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.23682994454713494,
|
| 294 |
+
"grad_norm": 13.938512802124023,
|
| 295 |
+
"learning_rate": 2.8816427911275417e-05,
|
| 296 |
+
"loss": 0.2214,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.24260628465804066,
|
| 301 |
+
"grad_norm": 2.2964584827423096,
|
| 302 |
+
"learning_rate": 2.8787546210720887e-05,
|
| 303 |
+
"loss": 0.2408,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.2483826247689464,
|
| 308 |
+
"grad_norm": 4.9652605056762695,
|
| 309 |
+
"learning_rate": 2.875866451016636e-05,
|
| 310 |
+
"loss": 0.3097,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.2541589648798521,
|
| 315 |
+
"grad_norm": 12.751204490661621,
|
| 316 |
+
"learning_rate": 2.872978280961183e-05,
|
| 317 |
+
"loss": 0.3046,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.25993530499075784,
|
| 322 |
+
"grad_norm": 21.8670597076416,
|
| 323 |
+
"learning_rate": 2.8700901109057302e-05,
|
| 324 |
+
"loss": 0.2608,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.26571164510166356,
|
| 329 |
+
"grad_norm": 10.842573165893555,
|
| 330 |
+
"learning_rate": 2.8672019408502772e-05,
|
| 331 |
+
"loss": 0.2792,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.27148798521256934,
|
| 336 |
+
"grad_norm": 1.6427525281906128,
|
| 337 |
+
"learning_rate": 2.8643137707948245e-05,
|
| 338 |
+
"loss": 0.2752,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.27726432532347506,
|
| 343 |
+
"grad_norm": 20.114425659179688,
|
| 344 |
+
"learning_rate": 2.8614256007393718e-05,
|
| 345 |
+
"loss": 0.2766,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.2830406654343808,
|
| 350 |
+
"grad_norm": 3.259066104888916,
|
| 351 |
+
"learning_rate": 2.8585374306839187e-05,
|
| 352 |
+
"loss": 0.2855,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2888170055452865,
|
| 357 |
+
"grad_norm": 14.488828659057617,
|
| 358 |
+
"learning_rate": 2.855649260628466e-05,
|
| 359 |
+
"loss": 0.2713,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.2945933456561922,
|
| 364 |
+
"grad_norm": 0.09920129925012589,
|
| 365 |
+
"learning_rate": 2.852761090573013e-05,
|
| 366 |
+
"loss": 0.2569,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.300369685767098,
|
| 371 |
+
"grad_norm": 4.165335655212402,
|
| 372 |
+
"learning_rate": 2.8498729205175603e-05,
|
| 373 |
+
"loss": 0.3494,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.3061460258780037,
|
| 378 |
+
"grad_norm": 11.144678115844727,
|
| 379 |
+
"learning_rate": 2.8469847504621073e-05,
|
| 380 |
+
"loss": 0.2928,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.31192236598890943,
|
| 385 |
+
"grad_norm": 11.037677764892578,
|
| 386 |
+
"learning_rate": 2.8440965804066546e-05,
|
| 387 |
+
"loss": 0.2367,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.31769870609981515,
|
| 392 |
+
"grad_norm": 33.86788558959961,
|
| 393 |
+
"learning_rate": 2.841208410351202e-05,
|
| 394 |
+
"loss": 0.2385,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.3234750462107209,
|
| 399 |
+
"grad_norm": 14.876667976379395,
|
| 400 |
+
"learning_rate": 2.8383202402957488e-05,
|
| 401 |
+
"loss": 0.3315,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.3292513863216266,
|
| 406 |
+
"grad_norm": 11.150797843933105,
|
| 407 |
+
"learning_rate": 2.8354320702402958e-05,
|
| 408 |
+
"loss": 0.2914,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.33502772643253237,
|
| 413 |
+
"grad_norm": 17.667240142822266,
|
| 414 |
+
"learning_rate": 2.8325439001848427e-05,
|
| 415 |
+
"loss": 0.2445,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.3408040665434381,
|
| 420 |
+
"grad_norm": 20.02605438232422,
|
| 421 |
+
"learning_rate": 2.82965573012939e-05,
|
| 422 |
+
"loss": 0.2804,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.3465804066543438,
|
| 427 |
+
"grad_norm": 27.934114456176758,
|
| 428 |
+
"learning_rate": 2.826767560073937e-05,
|
| 429 |
+
"loss": 0.3034,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.35235674676524953,
|
| 434 |
+
"grad_norm": 5.3321332931518555,
|
| 435 |
+
"learning_rate": 2.8238793900184843e-05,
|
| 436 |
+
"loss": 0.2677,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.35813308687615525,
|
| 441 |
+
"grad_norm": 22.4134521484375,
|
| 442 |
+
"learning_rate": 2.8209912199630313e-05,
|
| 443 |
+
"loss": 0.219,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.363909426987061,
|
| 448 |
+
"grad_norm": 25.509113311767578,
|
| 449 |
+
"learning_rate": 2.8181030499075786e-05,
|
| 450 |
+
"loss": 0.2367,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.36968576709796674,
|
| 455 |
+
"grad_norm": 18.299274444580078,
|
| 456 |
+
"learning_rate": 2.815214879852126e-05,
|
| 457 |
+
"loss": 0.2418,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.37546210720887246,
|
| 462 |
+
"grad_norm": 6.69123649597168,
|
| 463 |
+
"learning_rate": 2.8123267097966728e-05,
|
| 464 |
+
"loss": 0.3057,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3812384473197782,
|
| 469 |
+
"grad_norm": 26.583688735961914,
|
| 470 |
+
"learning_rate": 2.80943853974122e-05,
|
| 471 |
+
"loss": 0.2101,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3870147874306839,
|
| 476 |
+
"grad_norm": 10.969908714294434,
|
| 477 |
+
"learning_rate": 2.806550369685767e-05,
|
| 478 |
+
"loss": 0.2447,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.3927911275415896,
|
| 483 |
+
"grad_norm": 1.5843168497085571,
|
| 484 |
+
"learning_rate": 2.8036621996303144e-05,
|
| 485 |
+
"loss": 0.2241,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3985674676524954,
|
| 490 |
+
"grad_norm": 21.7174129486084,
|
| 491 |
+
"learning_rate": 2.8007740295748613e-05,
|
| 492 |
+
"loss": 0.3335,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.4043438077634011,
|
| 497 |
+
"grad_norm": 14.578802108764648,
|
| 498 |
+
"learning_rate": 2.7978858595194086e-05,
|
| 499 |
+
"loss": 0.2309,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.41012014787430684,
|
| 504 |
+
"grad_norm": 0.6399129033088684,
|
| 505 |
+
"learning_rate": 2.794997689463956e-05,
|
| 506 |
+
"loss": 0.2236,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.41589648798521256,
|
| 511 |
+
"grad_norm": 0.07050570845603943,
|
| 512 |
+
"learning_rate": 2.792109519408503e-05,
|
| 513 |
+
"loss": 0.2711,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.4216728280961183,
|
| 518 |
+
"grad_norm": 0.8827792406082153,
|
| 519 |
+
"learning_rate": 2.7892213493530502e-05,
|
| 520 |
+
"loss": 0.2229,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.42744916820702406,
|
| 525 |
+
"grad_norm": 6.008078575134277,
|
| 526 |
+
"learning_rate": 2.786333179297597e-05,
|
| 527 |
+
"loss": 0.2254,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.4332255083179298,
|
| 532 |
+
"grad_norm": 9.279077529907227,
|
| 533 |
+
"learning_rate": 2.7834450092421444e-05,
|
| 534 |
+
"loss": 0.2593,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.4390018484288355,
|
| 539 |
+
"grad_norm": 6.761204242706299,
|
| 540 |
+
"learning_rate": 2.780556839186691e-05,
|
| 541 |
+
"loss": 0.1904,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.4447781885397412,
|
| 546 |
+
"grad_norm": 29.752927780151367,
|
| 547 |
+
"learning_rate": 2.7776686691312384e-05,
|
| 548 |
+
"loss": 0.2287,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.45055452865064693,
|
| 553 |
+
"grad_norm": 15.296160697937012,
|
| 554 |
+
"learning_rate": 2.7747804990757857e-05,
|
| 555 |
+
"loss": 0.215,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.45633086876155265,
|
| 560 |
+
"grad_norm": 10.903311729431152,
|
| 561 |
+
"learning_rate": 2.7718923290203326e-05,
|
| 562 |
+
"loss": 0.2449,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.46210720887245843,
|
| 567 |
+
"grad_norm": 7.3522047996521,
|
| 568 |
+
"learning_rate": 2.76900415896488e-05,
|
| 569 |
+
"loss": 0.2581,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.46788354898336415,
|
| 574 |
+
"grad_norm": 1.4037950038909912,
|
| 575 |
+
"learning_rate": 2.766115988909427e-05,
|
| 576 |
+
"loss": 0.2753,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.47365988909426987,
|
| 581 |
+
"grad_norm": 54.007911682128906,
|
| 582 |
+
"learning_rate": 2.7632278188539742e-05,
|
| 583 |
+
"loss": 0.2105,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.4794362292051756,
|
| 588 |
+
"grad_norm": 18.261384963989258,
|
| 589 |
+
"learning_rate": 2.760339648798521e-05,
|
| 590 |
+
"loss": 0.3445,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4852125693160813,
|
| 595 |
+
"grad_norm": 19.31684684753418,
|
| 596 |
+
"learning_rate": 2.7574514787430684e-05,
|
| 597 |
+
"loss": 0.2484,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4909889094269871,
|
| 602 |
+
"grad_norm": 0.8063836693763733,
|
| 603 |
+
"learning_rate": 2.7545633086876157e-05,
|
| 604 |
+
"loss": 0.306,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4967652495378928,
|
| 609 |
+
"grad_norm": 15.769898414611816,
|
| 610 |
+
"learning_rate": 2.7516751386321627e-05,
|
| 611 |
+
"loss": 0.2102,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.5025415896487985,
|
| 616 |
+
"grad_norm": 7.9377217292785645,
|
| 617 |
+
"learning_rate": 2.74878696857671e-05,
|
| 618 |
+
"loss": 0.2961,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.5083179297597042,
|
| 623 |
+
"grad_norm": 10.430275917053223,
|
| 624 |
+
"learning_rate": 2.745898798521257e-05,
|
| 625 |
+
"loss": 0.291,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.51409426987061,
|
| 630 |
+
"grad_norm": 10.901015281677246,
|
| 631 |
+
"learning_rate": 2.7430106284658043e-05,
|
| 632 |
+
"loss": 0.2726,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5198706099815157,
|
| 637 |
+
"grad_norm": 13.966771125793457,
|
| 638 |
+
"learning_rate": 2.7401224584103512e-05,
|
| 639 |
+
"loss": 0.2795,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.5256469500924215,
|
| 644 |
+
"grad_norm": 0.8151399493217468,
|
| 645 |
+
"learning_rate": 2.7372342883548985e-05,
|
| 646 |
+
"loss": 0.2491,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.5314232902033271,
|
| 651 |
+
"grad_norm": 27.18870735168457,
|
| 652 |
+
"learning_rate": 2.7343461182994458e-05,
|
| 653 |
+
"loss": 0.259,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5371996303142329,
|
| 658 |
+
"grad_norm": 0.14737075567245483,
|
| 659 |
+
"learning_rate": 2.7314579482439928e-05,
|
| 660 |
+
"loss": 0.1497,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.5429759704251387,
|
| 665 |
+
"grad_norm": 15.187653541564941,
|
| 666 |
+
"learning_rate": 2.72856977818854e-05,
|
| 667 |
+
"loss": 0.2764,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.5487523105360443,
|
| 672 |
+
"grad_norm": 0.07880234718322754,
|
| 673 |
+
"learning_rate": 2.7256816081330867e-05,
|
| 674 |
+
"loss": 0.1721,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.5545286506469501,
|
| 679 |
+
"grad_norm": 21.45745849609375,
|
| 680 |
+
"learning_rate": 2.722793438077634e-05,
|
| 681 |
+
"loss": 0.2486,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.5603049907578558,
|
| 686 |
+
"grad_norm": 22.560449600219727,
|
| 687 |
+
"learning_rate": 2.719905268022181e-05,
|
| 688 |
+
"loss": 0.2866,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5660813308687616,
|
| 693 |
+
"grad_norm": 9.561197280883789,
|
| 694 |
+
"learning_rate": 2.7170170979667283e-05,
|
| 695 |
+
"loss": 0.2175,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.5718576709796673,
|
| 700 |
+
"grad_norm": 10.6559419631958,
|
| 701 |
+
"learning_rate": 2.7141289279112756e-05,
|
| 702 |
+
"loss": 0.2421,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.577634011090573,
|
| 707 |
+
"grad_norm": 13.36540699005127,
|
| 708 |
+
"learning_rate": 2.7112407578558225e-05,
|
| 709 |
+
"loss": 0.2709,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.5834103512014788,
|
| 714 |
+
"grad_norm": 8.7359619140625,
|
| 715 |
+
"learning_rate": 2.7083525878003698e-05,
|
| 716 |
+
"loss": 0.2108,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.5891866913123844,
|
| 721 |
+
"grad_norm": 3.588351249694824,
|
| 722 |
+
"learning_rate": 2.7054644177449168e-05,
|
| 723 |
+
"loss": 0.1813,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.5949630314232902,
|
| 728 |
+
"grad_norm": 1.5511142015457153,
|
| 729 |
+
"learning_rate": 2.702576247689464e-05,
|
| 730 |
+
"loss": 0.1868,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.600739371534196,
|
| 735 |
+
"grad_norm": 13.601162910461426,
|
| 736 |
+
"learning_rate": 2.699688077634011e-05,
|
| 737 |
+
"loss": 0.2756,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.6065157116451017,
|
| 742 |
+
"grad_norm": 0.9333792924880981,
|
| 743 |
+
"learning_rate": 2.6967999075785583e-05,
|
| 744 |
+
"loss": 0.2196,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.6122920517560074,
|
| 749 |
+
"grad_norm": 18.923709869384766,
|
| 750 |
+
"learning_rate": 2.6939117375231053e-05,
|
| 751 |
+
"loss": 0.3018,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.6180683918669131,
|
| 756 |
+
"grad_norm": 13.762917518615723,
|
| 757 |
+
"learning_rate": 2.6910235674676526e-05,
|
| 758 |
+
"loss": 0.2636,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6238447319778189,
|
| 763 |
+
"grad_norm": 1.0572352409362793,
|
| 764 |
+
"learning_rate": 2.6881353974122e-05,
|
| 765 |
+
"loss": 0.1796,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.6296210720887245,
|
| 770 |
+
"grad_norm": 2.580142021179199,
|
| 771 |
+
"learning_rate": 2.685247227356747e-05,
|
| 772 |
+
"loss": 0.195,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.6353974121996303,
|
| 777 |
+
"grad_norm": 4.498309135437012,
|
| 778 |
+
"learning_rate": 2.682359057301294e-05,
|
| 779 |
+
"loss": 0.2573,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.6411737523105361,
|
| 784 |
+
"grad_norm": 9.187541961669922,
|
| 785 |
+
"learning_rate": 2.679470887245841e-05,
|
| 786 |
+
"loss": 0.2201,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.6469500924214417,
|
| 791 |
+
"grad_norm": 39.70842361450195,
|
| 792 |
+
"learning_rate": 2.6765827171903884e-05,
|
| 793 |
+
"loss": 0.2327,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.6527264325323475,
|
| 798 |
+
"grad_norm": 11.123932838439941,
|
| 799 |
+
"learning_rate": 2.6736945471349354e-05,
|
| 800 |
+
"loss": 0.2328,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.6585027726432532,
|
| 805 |
+
"grad_norm": 13.621745109558105,
|
| 806 |
+
"learning_rate": 2.6708063770794823e-05,
|
| 807 |
+
"loss": 0.1527,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.664279112754159,
|
| 812 |
+
"grad_norm": 0.0642027035355568,
|
| 813 |
+
"learning_rate": 2.6679182070240296e-05,
|
| 814 |
+
"loss": 0.2881,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.6700554528650647,
|
| 819 |
+
"grad_norm": 0.32092395424842834,
|
| 820 |
+
"learning_rate": 2.6650300369685766e-05,
|
| 821 |
+
"loss": 0.2152,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.6758317929759704,
|
| 826 |
+
"grad_norm": 0.8972311615943909,
|
| 827 |
+
"learning_rate": 2.662141866913124e-05,
|
| 828 |
+
"loss": 0.1611,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.6816081330868762,
|
| 833 |
+
"grad_norm": 10.612149238586426,
|
| 834 |
+
"learning_rate": 2.659253696857671e-05,
|
| 835 |
+
"loss": 0.2985,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.6873844731977818,
|
| 840 |
+
"grad_norm": 11.864514350891113,
|
| 841 |
+
"learning_rate": 2.656365526802218e-05,
|
| 842 |
+
"loss": 0.2405,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.6931608133086876,
|
| 847 |
+
"grad_norm": 2.5804641246795654,
|
| 848 |
+
"learning_rate": 2.653477356746765e-05,
|
| 849 |
+
"loss": 0.2181,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.6989371534195934,
|
| 854 |
+
"grad_norm": 7.676444053649902,
|
| 855 |
+
"learning_rate": 2.6505891866913124e-05,
|
| 856 |
+
"loss": 0.2983,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.7047134935304991,
|
| 861 |
+
"grad_norm": 9.115777015686035,
|
| 862 |
+
"learning_rate": 2.6477010166358597e-05,
|
| 863 |
+
"loss": 0.2331,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7104898336414048,
|
| 868 |
+
"grad_norm": 0.4313638210296631,
|
| 869 |
+
"learning_rate": 2.6448128465804067e-05,
|
| 870 |
+
"loss": 0.1798,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.7162661737523105,
|
| 875 |
+
"grad_norm": 10.339539527893066,
|
| 876 |
+
"learning_rate": 2.641924676524954e-05,
|
| 877 |
+
"loss": 0.2015,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.7220425138632163,
|
| 882 |
+
"grad_norm": 11.897482872009277,
|
| 883 |
+
"learning_rate": 2.639036506469501e-05,
|
| 884 |
+
"loss": 0.2284,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.727818853974122,
|
| 889 |
+
"grad_norm": 0.22236181795597076,
|
| 890 |
+
"learning_rate": 2.6361483364140482e-05,
|
| 891 |
+
"loss": 0.2422,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.7335951940850277,
|
| 896 |
+
"grad_norm": 0.6248801946640015,
|
| 897 |
+
"learning_rate": 2.6332601663585952e-05,
|
| 898 |
+
"loss": 0.1927,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.7393715341959335,
|
| 903 |
+
"grad_norm": 21.60765838623047,
|
| 904 |
+
"learning_rate": 2.6303719963031425e-05,
|
| 905 |
+
"loss": 0.1929,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.7451478743068392,
|
| 910 |
+
"grad_norm": 7.824502468109131,
|
| 911 |
+
"learning_rate": 2.6274838262476898e-05,
|
| 912 |
+
"loss": 0.1485,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7509242144177449,
|
| 917 |
+
"grad_norm": 5.444038391113281,
|
| 918 |
+
"learning_rate": 2.6245956561922367e-05,
|
| 919 |
+
"loss": 0.2577,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.7567005545286506,
|
| 924 |
+
"grad_norm": 9.758363723754883,
|
| 925 |
+
"learning_rate": 2.621707486136784e-05,
|
| 926 |
+
"loss": 0.2703,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.7624768946395564,
|
| 931 |
+
"grad_norm": 0.42428484559059143,
|
| 932 |
+
"learning_rate": 2.618819316081331e-05,
|
| 933 |
+
"loss": 0.1711,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.7682532347504621,
|
| 938 |
+
"grad_norm": 10.41826343536377,
|
| 939 |
+
"learning_rate": 2.615931146025878e-05,
|
| 940 |
+
"loss": 0.2248,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.7740295748613678,
|
| 945 |
+
"grad_norm": 21.749746322631836,
|
| 946 |
+
"learning_rate": 2.613042975970425e-05,
|
| 947 |
+
"loss": 0.1968,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.7798059149722736,
|
| 952 |
+
"grad_norm": 0.19098539650440216,
|
| 953 |
+
"learning_rate": 2.6101548059149722e-05,
|
| 954 |
+
"loss": 0.2706,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.7855822550831792,
|
| 959 |
+
"grad_norm": 0.015689531341195107,
|
| 960 |
+
"learning_rate": 2.6072666358595195e-05,
|
| 961 |
+
"loss": 0.2384,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.791358595194085,
|
| 966 |
+
"grad_norm": 0.1689176857471466,
|
| 967 |
+
"learning_rate": 2.6043784658040665e-05,
|
| 968 |
+
"loss": 0.1714,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7971349353049908,
|
| 973 |
+
"grad_norm": 34.842952728271484,
|
| 974 |
+
"learning_rate": 2.6014902957486138e-05,
|
| 975 |
+
"loss": 0.215,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.8029112754158965,
|
| 980 |
+
"grad_norm": 0.10479370504617691,
|
| 981 |
+
"learning_rate": 2.5986021256931607e-05,
|
| 982 |
+
"loss": 0.2195,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.8086876155268022,
|
| 987 |
+
"grad_norm": 0.7637598514556885,
|
| 988 |
+
"learning_rate": 2.595713955637708e-05,
|
| 989 |
+
"loss": 0.1624,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.8144639556377079,
|
| 994 |
+
"grad_norm": 20.622161865234375,
|
| 995 |
+
"learning_rate": 2.592825785582255e-05,
|
| 996 |
+
"loss": 0.1896,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.8202402957486137,
|
| 1001 |
+
"grad_norm": 16.110519409179688,
|
| 1002 |
+
"learning_rate": 2.5899376155268023e-05,
|
| 1003 |
+
"loss": 0.2468,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.8260166358595195,
|
| 1008 |
+
"grad_norm": 19.84160804748535,
|
| 1009 |
+
"learning_rate": 2.5870494454713496e-05,
|
| 1010 |
+
"loss": 0.1933,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.8317929759704251,
|
| 1015 |
+
"grad_norm": 80.84690856933594,
|
| 1016 |
+
"learning_rate": 2.5841612754158966e-05,
|
| 1017 |
+
"loss": 0.1801,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.8375693160813309,
|
| 1022 |
+
"grad_norm": 15.291082382202148,
|
| 1023 |
+
"learning_rate": 2.581273105360444e-05,
|
| 1024 |
+
"loss": 0.2181,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8433456561922366,
|
| 1029 |
+
"grad_norm": 16.199522018432617,
|
| 1030 |
+
"learning_rate": 2.5783849353049908e-05,
|
| 1031 |
+
"loss": 0.2292,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.8491219963031423,
|
| 1036 |
+
"grad_norm": 17.037158966064453,
|
| 1037 |
+
"learning_rate": 2.575496765249538e-05,
|
| 1038 |
+
"loss": 0.2492,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.8548983364140481,
|
| 1043 |
+
"grad_norm": 20.019588470458984,
|
| 1044 |
+
"learning_rate": 2.572608595194085e-05,
|
| 1045 |
+
"loss": 0.1992,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.8606746765249538,
|
| 1050 |
+
"grad_norm": 17.38285255432129,
|
| 1051 |
+
"learning_rate": 2.5697204251386324e-05,
|
| 1052 |
+
"loss": 0.1994,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.8664510166358595,
|
| 1057 |
+
"grad_norm": 2.336366653442383,
|
| 1058 |
+
"learning_rate": 2.5668322550831793e-05,
|
| 1059 |
+
"loss": 0.2053,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.8722273567467652,
|
| 1064 |
+
"grad_norm": 62.698020935058594,
|
| 1065 |
+
"learning_rate": 2.5639440850277266e-05,
|
| 1066 |
+
"loss": 0.2787,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.878003696857671,
|
| 1071 |
+
"grad_norm": 13.748190879821777,
|
| 1072 |
+
"learning_rate": 2.5610559149722736e-05,
|
| 1073 |
+
"loss": 0.1935,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.8837800369685767,
|
| 1078 |
+
"grad_norm": 22.238149642944336,
|
| 1079 |
+
"learning_rate": 2.5581677449168206e-05,
|
| 1080 |
+
"loss": 0.3124,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.8895563770794824,
|
| 1085 |
+
"grad_norm": 1.701218843460083,
|
| 1086 |
+
"learning_rate": 2.555279574861368e-05,
|
| 1087 |
+
"loss": 0.1903,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.8953327171903882,
|
| 1092 |
+
"grad_norm": 18.659778594970703,
|
| 1093 |
+
"learning_rate": 2.5523914048059148e-05,
|
| 1094 |
+
"loss": 0.2223,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.9011090573012939,
|
| 1099 |
+
"grad_norm": 0.3101861774921417,
|
| 1100 |
+
"learning_rate": 2.549503234750462e-05,
|
| 1101 |
+
"loss": 0.2226,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.9068853974121996,
|
| 1106 |
+
"grad_norm": 0.15353505313396454,
|
| 1107 |
+
"learning_rate": 2.546615064695009e-05,
|
| 1108 |
+
"loss": 0.2453,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.9126617375231053,
|
| 1113 |
+
"grad_norm": 27.39130401611328,
|
| 1114 |
+
"learning_rate": 2.5437268946395564e-05,
|
| 1115 |
+
"loss": 0.2311,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.9184380776340111,
|
| 1120 |
+
"grad_norm": 0.3352121114730835,
|
| 1121 |
+
"learning_rate": 2.5408387245841037e-05,
|
| 1122 |
+
"loss": 0.1582,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.9242144177449169,
|
| 1127 |
+
"grad_norm": 1.0091736316680908,
|
| 1128 |
+
"learning_rate": 2.5379505545286506e-05,
|
| 1129 |
+
"loss": 0.2695,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.9299907578558225,
|
| 1134 |
+
"grad_norm": 21.034908294677734,
|
| 1135 |
+
"learning_rate": 2.535062384473198e-05,
|
| 1136 |
+
"loss": 0.2481,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9357670979667283,
|
| 1141 |
+
"grad_norm": 26.847919464111328,
|
| 1142 |
+
"learning_rate": 2.532174214417745e-05,
|
| 1143 |
+
"loss": 0.1675,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.941543438077634,
|
| 1148 |
+
"grad_norm": 9.66991138458252,
|
| 1149 |
+
"learning_rate": 2.5292860443622922e-05,
|
| 1150 |
+
"loss": 0.2571,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.9473197781885397,
|
| 1155 |
+
"grad_norm": 0.12006015330553055,
|
| 1156 |
+
"learning_rate": 2.526397874306839e-05,
|
| 1157 |
+
"loss": 0.1748,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.9530961182994455,
|
| 1162 |
+
"grad_norm": 1.825127363204956,
|
| 1163 |
+
"learning_rate": 2.5235097042513864e-05,
|
| 1164 |
+
"loss": 0.2161,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.9588724584103512,
|
| 1169 |
+
"grad_norm": 12.851505279541016,
|
| 1170 |
+
"learning_rate": 2.5206215341959337e-05,
|
| 1171 |
+
"loss": 0.2372,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.964648798521257,
|
| 1176 |
+
"grad_norm": 24.965559005737305,
|
| 1177 |
+
"learning_rate": 2.5177333641404807e-05,
|
| 1178 |
+
"loss": 0.2058,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.9704251386321626,
|
| 1183 |
+
"grad_norm": 33.49872589111328,
|
| 1184 |
+
"learning_rate": 2.514845194085028e-05,
|
| 1185 |
+
"loss": 0.2352,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.9762014787430684,
|
| 1190 |
+
"grad_norm": 2.5571775436401367,
|
| 1191 |
+
"learning_rate": 2.511957024029575e-05,
|
| 1192 |
+
"loss": 0.1973,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9819778188539742,
|
| 1197 |
+
"grad_norm": 12.988107681274414,
|
| 1198 |
+
"learning_rate": 2.5090688539741223e-05,
|
| 1199 |
+
"loss": 0.1691,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.9877541589648798,
|
| 1204 |
+
"grad_norm": 22.103260040283203,
|
| 1205 |
+
"learning_rate": 2.5061806839186692e-05,
|
| 1206 |
+
"loss": 0.202,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.9935304990757856,
|
| 1211 |
+
"grad_norm": 16.42685890197754,
|
| 1212 |
+
"learning_rate": 2.5032925138632162e-05,
|
| 1213 |
+
"loss": 0.1637,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.9993068391866913,
|
| 1218 |
+
"grad_norm": 0.6291359663009644,
|
| 1219 |
+
"learning_rate": 2.5004043438077635e-05,
|
| 1220 |
+
"loss": 0.2335,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.0,
|
| 1225 |
+
"eval_accuracy": 0.9354126055880442,
|
| 1226 |
+
"eval_f1": 0.9607951408061844,
|
| 1227 |
+
"eval_loss": 0.19707663357257843,
|
| 1228 |
+
"eval_precision": 0.9678957406230133,
|
| 1229 |
+
"eval_recall": 0.9537979639780736,
|
| 1230 |
+
"eval_runtime": 68.7658,
|
| 1231 |
+
"eval_samples_per_second": 111.902,
|
| 1232 |
+
"eval_steps_per_second": 13.99,
|
| 1233 |
+
"step": 8656
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.005083179297597,
|
| 1237 |
+
"grad_norm": 11.418679237365723,
|
| 1238 |
+
"learning_rate": 2.4975161737523104e-05,
|
| 1239 |
+
"loss": 0.158,
|
| 1240 |
+
"step": 8700
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.0108595194085028,
|
| 1244 |
+
"grad_norm": 35.1962776184082,
|
| 1245 |
+
"learning_rate": 2.4946280036968577e-05,
|
| 1246 |
+
"loss": 0.11,
|
| 1247 |
+
"step": 8750
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.0166358595194085,
|
| 1251 |
+
"grad_norm": 0.40325233340263367,
|
| 1252 |
+
"learning_rate": 2.4917398336414047e-05,
|
| 1253 |
+
"loss": 0.1666,
|
| 1254 |
+
"step": 8800
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.0224121996303142,
|
| 1258 |
+
"grad_norm": 0.08985109627246857,
|
| 1259 |
+
"learning_rate": 2.488851663585952e-05,
|
| 1260 |
+
"loss": 0.2029,
|
| 1261 |
+
"step": 8850
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.02818853974122,
|
| 1265 |
+
"grad_norm": 0.09904562681913376,
|
| 1266 |
+
"learning_rate": 2.485963493530499e-05,
|
| 1267 |
+
"loss": 0.1419,
|
| 1268 |
+
"step": 8900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.0339648798521257,
|
| 1272 |
+
"grad_norm": 0.47282811999320984,
|
| 1273 |
+
"learning_rate": 2.4830753234750463e-05,
|
| 1274 |
+
"loss": 0.1656,
|
| 1275 |
+
"step": 8950
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.0397412199630314,
|
| 1279 |
+
"grad_norm": 0.31505492329597473,
|
| 1280 |
+
"learning_rate": 2.4801871534195936e-05,
|
| 1281 |
+
"loss": 0.1454,
|
| 1282 |
+
"step": 9000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.0455175600739373,
|
| 1286 |
+
"grad_norm": 0.10816814750432968,
|
| 1287 |
+
"learning_rate": 2.4772989833641405e-05,
|
| 1288 |
+
"loss": 0.1705,
|
| 1289 |
+
"step": 9050
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.051293900184843,
|
| 1293 |
+
"grad_norm": 0.7013025283813477,
|
| 1294 |
+
"learning_rate": 2.4744108133086878e-05,
|
| 1295 |
+
"loss": 0.1612,
|
| 1296 |
+
"step": 9100
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.0570702402957486,
|
| 1300 |
+
"grad_norm": 2.4045467376708984,
|
| 1301 |
+
"learning_rate": 2.4715226432532348e-05,
|
| 1302 |
+
"loss": 0.1552,
|
| 1303 |
+
"step": 9150
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.0628465804066543,
|
| 1307 |
+
"grad_norm": 2.943918466567993,
|
| 1308 |
+
"learning_rate": 2.468634473197782e-05,
|
| 1309 |
+
"loss": 0.1593,
|
| 1310 |
+
"step": 9200
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.0686229205175601,
|
| 1314 |
+
"grad_norm": 0.06787320971488953,
|
| 1315 |
+
"learning_rate": 2.465746303142329e-05,
|
| 1316 |
+
"loss": 0.1493,
|
| 1317 |
+
"step": 9250
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.0743992606284658,
|
| 1321 |
+
"grad_norm": 0.5487985014915466,
|
| 1322 |
+
"learning_rate": 2.4628581330868763e-05,
|
| 1323 |
+
"loss": 0.1707,
|
| 1324 |
+
"step": 9300
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.0801756007393715,
|
| 1328 |
+
"grad_norm": 27.40123176574707,
|
| 1329 |
+
"learning_rate": 2.4599699630314236e-05,
|
| 1330 |
+
"loss": 0.1384,
|
| 1331 |
+
"step": 9350
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.0859519408502774,
|
| 1335 |
+
"grad_norm": 0.041278500109910965,
|
| 1336 |
+
"learning_rate": 2.4570817929759706e-05,
|
| 1337 |
+
"loss": 0.1975,
|
| 1338 |
+
"step": 9400
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.091728280961183,
|
| 1342 |
+
"grad_norm": 35.25811004638672,
|
| 1343 |
+
"learning_rate": 2.454193622920518e-05,
|
| 1344 |
+
"loss": 0.1859,
|
| 1345 |
+
"step": 9450
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0975046210720887,
|
| 1349 |
+
"grad_norm": 22.018983840942383,
|
| 1350 |
+
"learning_rate": 2.451305452865065e-05,
|
| 1351 |
+
"loss": 0.1713,
|
| 1352 |
+
"step": 9500
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.1032809611829943,
|
| 1356 |
+
"grad_norm": 1.6615002155303955,
|
| 1357 |
+
"learning_rate": 2.4484172828096118e-05,
|
| 1358 |
+
"loss": 0.1279,
|
| 1359 |
+
"step": 9550
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.1090573012939002,
|
| 1363 |
+
"grad_norm": 0.05830889567732811,
|
| 1364 |
+
"learning_rate": 2.4455291127541588e-05,
|
| 1365 |
+
"loss": 0.1166,
|
| 1366 |
+
"step": 9600
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.114833641404806,
|
| 1370 |
+
"grad_norm": 0.10341417044401169,
|
| 1371 |
+
"learning_rate": 2.442640942698706e-05,
|
| 1372 |
+
"loss": 0.1639,
|
| 1373 |
+
"step": 9650
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.1206099815157116,
|
| 1377 |
+
"grad_norm": 7.9995646476745605,
|
| 1378 |
+
"learning_rate": 2.439752772643253e-05,
|
| 1379 |
+
"loss": 0.2051,
|
| 1380 |
+
"step": 9700
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.1263863216266174,
|
| 1384 |
+
"grad_norm": 19.472301483154297,
|
| 1385 |
+
"learning_rate": 2.4368646025878003e-05,
|
| 1386 |
+
"loss": 0.1419,
|
| 1387 |
+
"step": 9750
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.1321626617375231,
|
| 1391 |
+
"grad_norm": 0.03694809228181839,
|
| 1392 |
+
"learning_rate": 2.4339764325323476e-05,
|
| 1393 |
+
"loss": 0.1501,
|
| 1394 |
+
"step": 9800
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.1379390018484288,
|
| 1398 |
+
"grad_norm": 9.946338653564453,
|
| 1399 |
+
"learning_rate": 2.4310882624768946e-05,
|
| 1400 |
+
"loss": 0.1779,
|
| 1401 |
+
"step": 9850
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1437153419593344,
|
| 1405 |
+
"grad_norm": 27.467966079711914,
|
| 1406 |
+
"learning_rate": 2.428200092421442e-05,
|
| 1407 |
+
"loss": 0.1526,
|
| 1408 |
+
"step": 9900
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.1494916820702403,
|
| 1412 |
+
"grad_norm": 1.936484694480896,
|
| 1413 |
+
"learning_rate": 2.425311922365989e-05,
|
| 1414 |
+
"loss": 0.1764,
|
| 1415 |
+
"step": 9950
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.155268022181146,
|
| 1419 |
+
"grad_norm": 0.10868394374847412,
|
| 1420 |
+
"learning_rate": 2.422423752310536e-05,
|
| 1421 |
+
"loss": 0.1688,
|
| 1422 |
+
"step": 10000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.1610443622920519,
|
| 1426 |
+
"grad_norm": 0.08838610351085663,
|
| 1427 |
+
"learning_rate": 2.419535582255083e-05,
|
| 1428 |
+
"loss": 0.143,
|
| 1429 |
+
"step": 10050
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.1668207024029575,
|
| 1433 |
+
"grad_norm": 0.09524734318256378,
|
| 1434 |
+
"learning_rate": 2.4166474121996304e-05,
|
| 1435 |
+
"loss": 0.1645,
|
| 1436 |
+
"step": 10100
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.1725970425138632,
|
| 1440 |
+
"grad_norm": 0.1827584058046341,
|
| 1441 |
+
"learning_rate": 2.4137592421441777e-05,
|
| 1442 |
+
"loss": 0.1789,
|
| 1443 |
+
"step": 10150
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.1783733826247689,
|
| 1447 |
+
"grad_norm": 41.8397331237793,
|
| 1448 |
+
"learning_rate": 2.4108710720887247e-05,
|
| 1449 |
+
"loss": 0.1185,
|
| 1450 |
+
"step": 10200
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.1841497227356748,
|
| 1454 |
+
"grad_norm": 0.013732840307056904,
|
| 1455 |
+
"learning_rate": 2.407982902033272e-05,
|
| 1456 |
+
"loss": 0.1326,
|
| 1457 |
+
"step": 10250
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.1899260628465804,
|
| 1461 |
+
"grad_norm": 0.08121947199106216,
|
| 1462 |
+
"learning_rate": 2.405094731977819e-05,
|
| 1463 |
+
"loss": 0.217,
|
| 1464 |
+
"step": 10300
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.195702402957486,
|
| 1468 |
+
"grad_norm": 0.2686977982521057,
|
| 1469 |
+
"learning_rate": 2.4022065619223662e-05,
|
| 1470 |
+
"loss": 0.1536,
|
| 1471 |
+
"step": 10350
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.201478743068392,
|
| 1475 |
+
"grad_norm": 7.599088191986084,
|
| 1476 |
+
"learning_rate": 2.3993183918669132e-05,
|
| 1477 |
+
"loss": 0.1528,
|
| 1478 |
+
"step": 10400
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.2072550831792976,
|
| 1482 |
+
"grad_norm": 0.20336832106113434,
|
| 1483 |
+
"learning_rate": 2.3964302218114605e-05,
|
| 1484 |
+
"loss": 0.1738,
|
| 1485 |
+
"step": 10450
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.2130314232902033,
|
| 1489 |
+
"grad_norm": 10.706509590148926,
|
| 1490 |
+
"learning_rate": 2.3935420517560074e-05,
|
| 1491 |
+
"loss": 0.1623,
|
| 1492 |
+
"step": 10500
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.218807763401109,
|
| 1496 |
+
"grad_norm": 26.202510833740234,
|
| 1497 |
+
"learning_rate": 2.3906538817005544e-05,
|
| 1498 |
+
"loss": 0.1552,
|
| 1499 |
+
"step": 10550
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.2245841035120149,
|
| 1503 |
+
"grad_norm": 22.59147071838379,
|
| 1504 |
+
"learning_rate": 2.3877657116451017e-05,
|
| 1505 |
+
"loss": 0.1325,
|
| 1506 |
+
"step": 10600
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.2303604436229205,
|
| 1510 |
+
"grad_norm": 382.7044677734375,
|
| 1511 |
+
"learning_rate": 2.3848775415896487e-05,
|
| 1512 |
+
"loss": 0.2088,
|
| 1513 |
+
"step": 10650
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2361367837338262,
|
| 1517 |
+
"grad_norm": 10.813858032226562,
|
| 1518 |
+
"learning_rate": 2.381989371534196e-05,
|
| 1519 |
+
"loss": 0.167,
|
| 1520 |
+
"step": 10700
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.241913123844732,
|
| 1524 |
+
"grad_norm": 2.7746658325195312,
|
| 1525 |
+
"learning_rate": 2.379101201478743e-05,
|
| 1526 |
+
"loss": 0.2118,
|
| 1527 |
+
"step": 10750
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.2476894639556377,
|
| 1531 |
+
"grad_norm": 0.11969839036464691,
|
| 1532 |
+
"learning_rate": 2.3762130314232902e-05,
|
| 1533 |
+
"loss": 0.1688,
|
| 1534 |
+
"step": 10800
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.2534658040665434,
|
| 1538 |
+
"grad_norm": 97.15472412109375,
|
| 1539 |
+
"learning_rate": 2.3733248613678375e-05,
|
| 1540 |
+
"loss": 0.1536,
|
| 1541 |
+
"step": 10850
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.259242144177449,
|
| 1545 |
+
"grad_norm": 5.869635581970215,
|
| 1546 |
+
"learning_rate": 2.3704366913123845e-05,
|
| 1547 |
+
"loss": 0.1868,
|
| 1548 |
+
"step": 10900
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.265018484288355,
|
| 1552 |
+
"grad_norm": 0.8501625061035156,
|
| 1553 |
+
"learning_rate": 2.3675485212569318e-05,
|
| 1554 |
+
"loss": 0.1237,
|
| 1555 |
+
"step": 10950
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2707948243992606,
|
| 1559 |
+
"grad_norm": 14.44981575012207,
|
| 1560 |
+
"learning_rate": 2.3646603512014787e-05,
|
| 1561 |
+
"loss": 0.1075,
|
| 1562 |
+
"step": 11000
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2765711645101663,
|
| 1566 |
+
"grad_norm": 0.06192944198846817,
|
| 1567 |
+
"learning_rate": 2.361772181146026e-05,
|
| 1568 |
+
"loss": 0.0926,
|
| 1569 |
+
"step": 11050
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2823475046210722,
|
| 1573 |
+
"grad_norm": 1.1274975538253784,
|
| 1574 |
+
"learning_rate": 2.358884011090573e-05,
|
| 1575 |
+
"loss": 0.1295,
|
| 1576 |
+
"step": 11100
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2881238447319778,
|
| 1580 |
+
"grad_norm": 0.09253960102796555,
|
| 1581 |
+
"learning_rate": 2.3559958410351203e-05,
|
| 1582 |
+
"loss": 0.1629,
|
| 1583 |
+
"step": 11150
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2939001848428835,
|
| 1587 |
+
"grad_norm": 3.5287694931030273,
|
| 1588 |
+
"learning_rate": 2.3531076709796676e-05,
|
| 1589 |
+
"loss": 0.1334,
|
| 1590 |
+
"step": 11200
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2996765249537892,
|
| 1594 |
+
"grad_norm": 0.3958209156990051,
|
| 1595 |
+
"learning_rate": 2.3502195009242146e-05,
|
| 1596 |
+
"loss": 0.2357,
|
| 1597 |
+
"step": 11250
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.305452865064695,
|
| 1601 |
+
"grad_norm": 0.8403951525688171,
|
| 1602 |
+
"learning_rate": 2.347331330868762e-05,
|
| 1603 |
+
"loss": 0.145,
|
| 1604 |
+
"step": 11300
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.3112292051756007,
|
| 1608 |
+
"grad_norm": 18.562786102294922,
|
| 1609 |
+
"learning_rate": 2.3444431608133088e-05,
|
| 1610 |
+
"loss": 0.1489,
|
| 1611 |
+
"step": 11350
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.3170055452865066,
|
| 1615 |
+
"grad_norm": 0.07164066284894943,
|
| 1616 |
+
"learning_rate": 2.341554990757856e-05,
|
| 1617 |
+
"loss": 0.1311,
|
| 1618 |
+
"step": 11400
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.3227818853974123,
|
| 1622 |
+
"grad_norm": 0.009075075387954712,
|
| 1623 |
+
"learning_rate": 2.3386668207024027e-05,
|
| 1624 |
+
"loss": 0.067,
|
| 1625 |
+
"step": 11450
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.328558225508318,
|
| 1629 |
+
"grad_norm": 2.3241634368896484,
|
| 1630 |
+
"learning_rate": 2.33577865064695e-05,
|
| 1631 |
+
"loss": 0.251,
|
| 1632 |
+
"step": 11500
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.3343345656192236,
|
| 1636 |
+
"grad_norm": 74.73931884765625,
|
| 1637 |
+
"learning_rate": 2.332890480591497e-05,
|
| 1638 |
+
"loss": 0.1685,
|
| 1639 |
+
"step": 11550
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.3401109057301293,
|
| 1643 |
+
"grad_norm": 0.22539888322353363,
|
| 1644 |
+
"learning_rate": 2.3300023105360443e-05,
|
| 1645 |
+
"loss": 0.0946,
|
| 1646 |
+
"step": 11600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.3458872458410351,
|
| 1650 |
+
"grad_norm": 0.05199708044528961,
|
| 1651 |
+
"learning_rate": 2.3271141404805916e-05,
|
| 1652 |
+
"loss": 0.0939,
|
| 1653 |
+
"step": 11650
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.3516635859519408,
|
| 1657 |
+
"grad_norm": 0.005744859576225281,
|
| 1658 |
+
"learning_rate": 2.3242259704251386e-05,
|
| 1659 |
+
"loss": 0.146,
|
| 1660 |
+
"step": 11700
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.3574399260628467,
|
| 1664 |
+
"grad_norm": 0.08633283525705338,
|
| 1665 |
+
"learning_rate": 2.321337800369686e-05,
|
| 1666 |
+
"loss": 0.1566,
|
| 1667 |
+
"step": 11750
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.3632162661737524,
|
| 1671 |
+
"grad_norm": 17.302743911743164,
|
| 1672 |
+
"learning_rate": 2.3184496303142328e-05,
|
| 1673 |
+
"loss": 0.1339,
|
| 1674 |
+
"step": 11800
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.368992606284658,
|
| 1678 |
+
"grad_norm": 23.171537399291992,
|
| 1679 |
+
"learning_rate": 2.31556146025878e-05,
|
| 1680 |
+
"loss": 0.1389,
|
| 1681 |
+
"step": 11850
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3747689463955637,
|
| 1685 |
+
"grad_norm": 0.9330076575279236,
|
| 1686 |
+
"learning_rate": 2.312673290203327e-05,
|
| 1687 |
+
"loss": 0.1594,
|
| 1688 |
+
"step": 11900
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.3805452865064696,
|
| 1692 |
+
"grad_norm": 5.958154201507568,
|
| 1693 |
+
"learning_rate": 2.3097851201478744e-05,
|
| 1694 |
+
"loss": 0.1039,
|
| 1695 |
+
"step": 11950
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.3863216266173752,
|
| 1699 |
+
"grad_norm": 0.15630990266799927,
|
| 1700 |
+
"learning_rate": 2.3068969500924217e-05,
|
| 1701 |
+
"loss": 0.2041,
|
| 1702 |
+
"step": 12000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.392097966728281,
|
| 1706 |
+
"grad_norm": 29.334747314453125,
|
| 1707 |
+
"learning_rate": 2.3040087800369686e-05,
|
| 1708 |
+
"loss": 0.1662,
|
| 1709 |
+
"step": 12050
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 1.3978743068391868,
|
| 1713 |
+
"grad_norm": 43.128379821777344,
|
| 1714 |
+
"learning_rate": 2.301120609981516e-05,
|
| 1715 |
+
"loss": 0.0719,
|
| 1716 |
+
"step": 12100
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 1.4036506469500925,
|
| 1720 |
+
"grad_norm": 0.14302471280097961,
|
| 1721 |
+
"learning_rate": 2.298232439926063e-05,
|
| 1722 |
+
"loss": 0.1473,
|
| 1723 |
+
"step": 12150
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.4094269870609981,
|
| 1727 |
+
"grad_norm": 0.14523327350616455,
|
| 1728 |
+
"learning_rate": 2.2953442698706102e-05,
|
| 1729 |
+
"loss": 0.2012,
|
| 1730 |
+
"step": 12200
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 1.4152033271719038,
|
| 1734 |
+
"grad_norm": 0.19870401918888092,
|
| 1735 |
+
"learning_rate": 2.292456099815157e-05,
|
| 1736 |
+
"loss": 0.1298,
|
| 1737 |
+
"step": 12250
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.4209796672828097,
|
| 1741 |
+
"grad_norm": 8.267024040222168,
|
| 1742 |
+
"learning_rate": 2.2895679297597044e-05,
|
| 1743 |
+
"loss": 0.1229,
|
| 1744 |
+
"step": 12300
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.4267560073937153,
|
| 1748 |
+
"grad_norm": 0.044785793870687485,
|
| 1749 |
+
"learning_rate": 2.2866797597042517e-05,
|
| 1750 |
+
"loss": 0.1727,
|
| 1751 |
+
"step": 12350
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.432532347504621,
|
| 1755 |
+
"grad_norm": 0.45464086532592773,
|
| 1756 |
+
"learning_rate": 2.2837915896487984e-05,
|
| 1757 |
+
"loss": 0.1769,
|
| 1758 |
+
"step": 12400
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 1.4383086876155269,
|
| 1762 |
+
"grad_norm": 45.37421798706055,
|
| 1763 |
+
"learning_rate": 2.2809034195933457e-05,
|
| 1764 |
+
"loss": 0.2269,
|
| 1765 |
+
"step": 12450
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.4440850277264325,
|
| 1769 |
+
"grad_norm": 0.010675471276044846,
|
| 1770 |
+
"learning_rate": 2.2780152495378926e-05,
|
| 1771 |
+
"loss": 0.1045,
|
| 1772 |
+
"step": 12500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.4498613678373382,
|
| 1776 |
+
"grad_norm": 0.5008090138435364,
|
| 1777 |
+
"learning_rate": 2.27512707948244e-05,
|
| 1778 |
+
"loss": 0.1331,
|
| 1779 |
+
"step": 12550
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.4556377079482439,
|
| 1783 |
+
"grad_norm": 0.010356269776821136,
|
| 1784 |
+
"learning_rate": 2.272238909426987e-05,
|
| 1785 |
+
"loss": 0.1148,
|
| 1786 |
+
"step": 12600
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 1.4614140480591498,
|
| 1790 |
+
"grad_norm": 2.318436622619629,
|
| 1791 |
+
"learning_rate": 2.2693507393715342e-05,
|
| 1792 |
+
"loss": 0.1796,
|
| 1793 |
+
"step": 12650
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4671903881700554,
|
| 1797 |
+
"grad_norm": 0.06255066394805908,
|
| 1798 |
+
"learning_rate": 2.2664625693160815e-05,
|
| 1799 |
+
"loss": 0.1387,
|
| 1800 |
+
"step": 12700
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 1.4729667282809613,
|
| 1804 |
+
"grad_norm": 0.02640642039477825,
|
| 1805 |
+
"learning_rate": 2.2635743992606284e-05,
|
| 1806 |
+
"loss": 0.1746,
|
| 1807 |
+
"step": 12750
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.478743068391867,
|
| 1811 |
+
"grad_norm": 37.81770706176758,
|
| 1812 |
+
"learning_rate": 2.2606862292051757e-05,
|
| 1813 |
+
"loss": 0.1362,
|
| 1814 |
+
"step": 12800
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.4845194085027726,
|
| 1818 |
+
"grad_norm": 0.0361916609108448,
|
| 1819 |
+
"learning_rate": 2.2577980591497227e-05,
|
| 1820 |
+
"loss": 0.1773,
|
| 1821 |
+
"step": 12850
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.4902957486136783,
|
| 1825 |
+
"grad_norm": 0.04615430161356926,
|
| 1826 |
+
"learning_rate": 2.25490988909427e-05,
|
| 1827 |
+
"loss": 0.1558,
|
| 1828 |
+
"step": 12900
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 1.496072088724584,
|
| 1832 |
+
"grad_norm": 54.66000747680664,
|
| 1833 |
+
"learning_rate": 2.252021719038817e-05,
|
| 1834 |
+
"loss": 0.2207,
|
| 1835 |
+
"step": 12950
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 1.5018484288354899,
|
| 1839 |
+
"grad_norm": 7.9861836433410645,
|
| 1840 |
+
"learning_rate": 2.2491335489833643e-05,
|
| 1841 |
+
"loss": 0.133,
|
| 1842 |
+
"step": 13000
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.5076247689463955,
|
| 1846 |
+
"grad_norm": 0.050192106515169144,
|
| 1847 |
+
"learning_rate": 2.2462453789279116e-05,
|
| 1848 |
+
"loss": 0.1515,
|
| 1849 |
+
"step": 13050
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.5134011090573014,
|
| 1853 |
+
"grad_norm": 16.963003158569336,
|
| 1854 |
+
"learning_rate": 2.2433572088724585e-05,
|
| 1855 |
+
"loss": 0.1541,
|
| 1856 |
+
"step": 13100
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.519177449168207,
|
| 1860 |
+
"grad_norm": 0.16849014163017273,
|
| 1861 |
+
"learning_rate": 2.2404690388170058e-05,
|
| 1862 |
+
"loss": 0.1627,
|
| 1863 |
+
"step": 13150
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 1.5249537892791127,
|
| 1867 |
+
"grad_norm": 0.046051234006881714,
|
| 1868 |
+
"learning_rate": 2.2375808687615528e-05,
|
| 1869 |
+
"loss": 0.1728,
|
| 1870 |
+
"step": 13200
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.5307301293900184,
|
| 1874 |
+
"grad_norm": 0.03939524292945862,
|
| 1875 |
+
"learning_rate": 2.2346926987061e-05,
|
| 1876 |
+
"loss": 0.1402,
|
| 1877 |
+
"step": 13250
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 1.536506469500924,
|
| 1881 |
+
"grad_norm": 19.971302032470703,
|
| 1882 |
+
"learning_rate": 2.231804528650647e-05,
|
| 1883 |
+
"loss": 0.1563,
|
| 1884 |
+
"step": 13300
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.54228280961183,
|
| 1888 |
+
"grad_norm": 1.2224446535110474,
|
| 1889 |
+
"learning_rate": 2.228916358595194e-05,
|
| 1890 |
+
"loss": 0.148,
|
| 1891 |
+
"step": 13350
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.5480591497227358,
|
| 1895 |
+
"grad_norm": 10.81811237335205,
|
| 1896 |
+
"learning_rate": 2.2260281885397413e-05,
|
| 1897 |
+
"loss": 0.1474,
|
| 1898 |
+
"step": 13400
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 1.5538354898336415,
|
| 1902 |
+
"grad_norm": 2.6338510513305664,
|
| 1903 |
+
"learning_rate": 2.2231400184842883e-05,
|
| 1904 |
+
"loss": 0.1476,
|
| 1905 |
+
"step": 13450
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5596118299445472,
|
| 1909 |
+
"grad_norm": 0.10376719385385513,
|
| 1910 |
+
"learning_rate": 2.2202518484288356e-05,
|
| 1911 |
+
"loss": 0.0925,
|
| 1912 |
+
"step": 13500
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 1.5653881700554528,
|
| 1916 |
+
"grad_norm": 0.14928467571735382,
|
| 1917 |
+
"learning_rate": 2.2173636783733825e-05,
|
| 1918 |
+
"loss": 0.1362,
|
| 1919 |
+
"step": 13550
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 1.5711645101663585,
|
| 1923 |
+
"grad_norm": 1.1348618268966675,
|
| 1924 |
+
"learning_rate": 2.2144755083179298e-05,
|
| 1925 |
+
"loss": 0.1229,
|
| 1926 |
+
"step": 13600
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.5769408502772642,
|
| 1930 |
+
"grad_norm": 0.09609614312648773,
|
| 1931 |
+
"learning_rate": 2.2115873382624768e-05,
|
| 1932 |
+
"loss": 0.1732,
|
| 1933 |
+
"step": 13650
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.58271719038817,
|
| 1937 |
+
"grad_norm": 20.673297882080078,
|
| 1938 |
+
"learning_rate": 2.208699168207024e-05,
|
| 1939 |
+
"loss": 0.1249,
|
| 1940 |
+
"step": 13700
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 1.588493530499076,
|
| 1944 |
+
"grad_norm": 82.14835357666016,
|
| 1945 |
+
"learning_rate": 2.205810998151571e-05,
|
| 1946 |
+
"loss": 0.0819,
|
| 1947 |
+
"step": 13750
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 1.5942698706099816,
|
| 1951 |
+
"grad_norm": 25.9471435546875,
|
| 1952 |
+
"learning_rate": 2.2029228280961183e-05,
|
| 1953 |
+
"loss": 0.1534,
|
| 1954 |
+
"step": 13800
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 1.6000462107208873,
|
| 1958 |
+
"grad_norm": 11.10772705078125,
|
| 1959 |
+
"learning_rate": 2.2000346580406656e-05,
|
| 1960 |
+
"loss": 0.2186,
|
| 1961 |
+
"step": 13850
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.605822550831793,
|
| 1965 |
+
"grad_norm": 0.007331051863729954,
|
| 1966 |
+
"learning_rate": 2.1971464879852126e-05,
|
| 1967 |
+
"loss": 0.1159,
|
| 1968 |
+
"step": 13900
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.6115988909426986,
|
| 1972 |
+
"grad_norm": 1.7013061046600342,
|
| 1973 |
+
"learning_rate": 2.19425831792976e-05,
|
| 1974 |
+
"loss": 0.1857,
|
| 1975 |
+
"step": 13950
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.6173752310536045,
|
| 1979 |
+
"grad_norm": 19.55061149597168,
|
| 1980 |
+
"learning_rate": 2.191370147874307e-05,
|
| 1981 |
+
"loss": 0.1657,
|
| 1982 |
+
"step": 14000
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 1.6231515711645101,
|
| 1986 |
+
"grad_norm": 0.03729844093322754,
|
| 1987 |
+
"learning_rate": 2.188481977818854e-05,
|
| 1988 |
+
"loss": 0.1275,
|
| 1989 |
+
"step": 14050
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.628927911275416,
|
| 1993 |
+
"grad_norm": 65.23085021972656,
|
| 1994 |
+
"learning_rate": 2.185593807763401e-05,
|
| 1995 |
+
"loss": 0.1474,
|
| 1996 |
+
"step": 14100
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 1.6347042513863217,
|
| 2000 |
+
"grad_norm": 16.47488021850586,
|
| 2001 |
+
"learning_rate": 2.1827056377079484e-05,
|
| 2002 |
+
"loss": 0.0967,
|
| 2003 |
+
"step": 14150
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 1.6404805914972274,
|
| 2007 |
+
"grad_norm": 39.40799331665039,
|
| 2008 |
+
"learning_rate": 2.1798174676524957e-05,
|
| 2009 |
+
"loss": 0.1652,
|
| 2010 |
+
"step": 14200
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.646256931608133,
|
| 2014 |
+
"grad_norm": 20.859216690063477,
|
| 2015 |
+
"learning_rate": 2.1769292975970427e-05,
|
| 2016 |
+
"loss": 0.1453,
|
| 2017 |
+
"step": 14250
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.6520332717190387,
|
| 2021 |
+
"grad_norm": 0.1275290995836258,
|
| 2022 |
+
"learning_rate": 2.1740411275415896e-05,
|
| 2023 |
+
"loss": 0.1268,
|
| 2024 |
+
"step": 14300
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 1.6578096118299446,
|
| 2028 |
+
"grad_norm": 0.44121667742729187,
|
| 2029 |
+
"learning_rate": 2.1711529574861366e-05,
|
| 2030 |
+
"loss": 0.1056,
|
| 2031 |
+
"step": 14350
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.6635859519408502,
|
| 2035 |
+
"grad_norm": 0.2817210853099823,
|
| 2036 |
+
"learning_rate": 2.168264787430684e-05,
|
| 2037 |
+
"loss": 0.1781,
|
| 2038 |
+
"step": 14400
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 1.6693622920517561,
|
| 2042 |
+
"grad_norm": 3.9421677589416504,
|
| 2043 |
+
"learning_rate": 2.165376617375231e-05,
|
| 2044 |
+
"loss": 0.1279,
|
| 2045 |
+
"step": 14450
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 1.6751386321626618,
|
| 2049 |
+
"grad_norm": 0.23300153017044067,
|
| 2050 |
+
"learning_rate": 2.162488447319778e-05,
|
| 2051 |
+
"loss": 0.1512,
|
| 2052 |
+
"step": 14500
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.6809149722735675,
|
| 2056 |
+
"grad_norm": 0.21775715053081512,
|
| 2057 |
+
"learning_rate": 2.1596002772643254e-05,
|
| 2058 |
+
"loss": 0.1837,
|
| 2059 |
+
"step": 14550
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.6866913123844731,
|
| 2063 |
+
"grad_norm": 18.65435218811035,
|
| 2064 |
+
"learning_rate": 2.1567121072088724e-05,
|
| 2065 |
+
"loss": 0.0867,
|
| 2066 |
+
"step": 14600
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.6924676524953788,
|
| 2070 |
+
"grad_norm": 0.022388514131307602,
|
| 2071 |
+
"learning_rate": 2.1538239371534197e-05,
|
| 2072 |
+
"loss": 0.089,
|
| 2073 |
+
"step": 14650
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6982439926062847,
|
| 2077 |
+
"grad_norm": 0.10797044634819031,
|
| 2078 |
+
"learning_rate": 2.1509357670979667e-05,
|
| 2079 |
+
"loss": 0.1143,
|
| 2080 |
+
"step": 14700
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.7040203327171903,
|
| 2084 |
+
"grad_norm": 15.600735664367676,
|
| 2085 |
+
"learning_rate": 2.148047597042514e-05,
|
| 2086 |
+
"loss": 0.1299,
|
| 2087 |
+
"step": 14750
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.7097966728280962,
|
| 2091 |
+
"grad_norm": 34.40959548950195,
|
| 2092 |
+
"learning_rate": 2.145159426987061e-05,
|
| 2093 |
+
"loss": 0.2121,
|
| 2094 |
+
"step": 14800
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.7155730129390019,
|
| 2098 |
+
"grad_norm": 6.029714107513428,
|
| 2099 |
+
"learning_rate": 2.1422712569316082e-05,
|
| 2100 |
+
"loss": 0.1569,
|
| 2101 |
+
"step": 14850
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.7213493530499075,
|
| 2105 |
+
"grad_norm": 0.01677633635699749,
|
| 2106 |
+
"learning_rate": 2.1393830868761555e-05,
|
| 2107 |
+
"loss": 0.1182,
|
| 2108 |
+
"step": 14900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.7271256931608132,
|
| 2112 |
+
"grad_norm": 0.07976578921079636,
|
| 2113 |
+
"learning_rate": 2.1364949168207025e-05,
|
| 2114 |
+
"loss": 0.1374,
|
| 2115 |
+
"step": 14950
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 1.7329020332717189,
|
| 2119 |
+
"grad_norm": 0.49204331636428833,
|
| 2120 |
+
"learning_rate": 2.1336067467652498e-05,
|
| 2121 |
+
"loss": 0.0988,
|
| 2122 |
+
"step": 15000
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 1.7386783733826248,
|
| 2126 |
+
"grad_norm": 0.05830137059092522,
|
| 2127 |
+
"learning_rate": 2.1307185767097967e-05,
|
| 2128 |
+
"loss": 0.1321,
|
| 2129 |
+
"step": 15050
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.7444547134935307,
|
| 2133 |
+
"grad_norm": 0.33919116854667664,
|
| 2134 |
+
"learning_rate": 2.127830406654344e-05,
|
| 2135 |
+
"loss": 0.1756,
|
| 2136 |
+
"step": 15100
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.7502310536044363,
|
| 2140 |
+
"grad_norm": 0.03737005218863487,
|
| 2141 |
+
"learning_rate": 2.124942236598891e-05,
|
| 2142 |
+
"loss": 0.0943,
|
| 2143 |
+
"step": 15150
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.756007393715342,
|
| 2147 |
+
"grad_norm": 11.36135482788086,
|
| 2148 |
+
"learning_rate": 2.1220540665434383e-05,
|
| 2149 |
+
"loss": 0.1554,
|
| 2150 |
+
"step": 15200
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.7617837338262476,
|
| 2154 |
+
"grad_norm": 29.172395706176758,
|
| 2155 |
+
"learning_rate": 2.1191658964879853e-05,
|
| 2156 |
+
"loss": 0.1317,
|
| 2157 |
+
"step": 15250
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.7675600739371533,
|
| 2161 |
+
"grad_norm": 0.4681268334388733,
|
| 2162 |
+
"learning_rate": 2.1162777264325322e-05,
|
| 2163 |
+
"loss": 0.1342,
|
| 2164 |
+
"step": 15300
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.7733364140480592,
|
| 2168 |
+
"grad_norm": 0.19265428185462952,
|
| 2169 |
+
"learning_rate": 2.1133895563770795e-05,
|
| 2170 |
+
"loss": 0.1198,
|
| 2171 |
+
"step": 15350
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.7791127541589649,
|
| 2175 |
+
"grad_norm": 0.38005658984184265,
|
| 2176 |
+
"learning_rate": 2.1105013863216265e-05,
|
| 2177 |
+
"loss": 0.1178,
|
| 2178 |
+
"step": 15400
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.7848890942698707,
|
| 2182 |
+
"grad_norm": 0.6070498824119568,
|
| 2183 |
+
"learning_rate": 2.1076132162661738e-05,
|
| 2184 |
+
"loss": 0.1597,
|
| 2185 |
+
"step": 15450
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7906654343807764,
|
| 2189 |
+
"grad_norm": 0.13323445618152618,
|
| 2190 |
+
"learning_rate": 2.1047250462107207e-05,
|
| 2191 |
+
"loss": 0.1496,
|
| 2192 |
+
"step": 15500
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.796441774491682,
|
| 2196 |
+
"grad_norm": 39.70579147338867,
|
| 2197 |
+
"learning_rate": 2.101836876155268e-05,
|
| 2198 |
+
"loss": 0.1734,
|
| 2199 |
+
"step": 15550
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.8022181146025877,
|
| 2203 |
+
"grad_norm": 0.04307672381401062,
|
| 2204 |
+
"learning_rate": 2.0989487060998153e-05,
|
| 2205 |
+
"loss": 0.0877,
|
| 2206 |
+
"step": 15600
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.8079944547134934,
|
| 2210 |
+
"grad_norm": 8.399633407592773,
|
| 2211 |
+
"learning_rate": 2.0960605360443623e-05,
|
| 2212 |
+
"loss": 0.1458,
|
| 2213 |
+
"step": 15650
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.8137707948243993,
|
| 2217 |
+
"grad_norm": 0.07565668225288391,
|
| 2218 |
+
"learning_rate": 2.0931723659889096e-05,
|
| 2219 |
+
"loss": 0.1264,
|
| 2220 |
+
"step": 15700
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.819547134935305,
|
| 2224 |
+
"grad_norm": 45.78880310058594,
|
| 2225 |
+
"learning_rate": 2.0902841959334566e-05,
|
| 2226 |
+
"loss": 0.1882,
|
| 2227 |
+
"step": 15750
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.8253234750462108,
|
| 2231 |
+
"grad_norm": 56.75712966918945,
|
| 2232 |
+
"learning_rate": 2.087396025878004e-05,
|
| 2233 |
+
"loss": 0.1071,
|
| 2234 |
+
"step": 15800
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.8310998151571165,
|
| 2238 |
+
"grad_norm": 0.1278388649225235,
|
| 2239 |
+
"learning_rate": 2.0845078558225508e-05,
|
| 2240 |
+
"loss": 0.1696,
|
| 2241 |
+
"step": 15850
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.8368761552680222,
|
| 2245 |
+
"grad_norm": 0.5721489787101746,
|
| 2246 |
+
"learning_rate": 2.081619685767098e-05,
|
| 2247 |
+
"loss": 0.139,
|
| 2248 |
+
"step": 15900
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.8426524953789278,
|
| 2252 |
+
"grad_norm": 46.980064392089844,
|
| 2253 |
+
"learning_rate": 2.078731515711645e-05,
|
| 2254 |
+
"loss": 0.1538,
|
| 2255 |
+
"step": 15950
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.8484288354898335,
|
| 2259 |
+
"grad_norm": 0.10779360681772232,
|
| 2260 |
+
"learning_rate": 2.0758433456561924e-05,
|
| 2261 |
+
"loss": 0.0843,
|
| 2262 |
+
"step": 16000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.8542051756007394,
|
| 2266 |
+
"grad_norm": 56.32892608642578,
|
| 2267 |
+
"learning_rate": 2.0729551756007397e-05,
|
| 2268 |
+
"loss": 0.1689,
|
| 2269 |
+
"step": 16050
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.859981515711645,
|
| 2273 |
+
"grad_norm": 0.046350061893463135,
|
| 2274 |
+
"learning_rate": 2.0700670055452866e-05,
|
| 2275 |
+
"loss": 0.1027,
|
| 2276 |
+
"step": 16100
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.865757855822551,
|
| 2280 |
+
"grad_norm": 58.256988525390625,
|
| 2281 |
+
"learning_rate": 2.067178835489834e-05,
|
| 2282 |
+
"loss": 0.0697,
|
| 2283 |
+
"step": 16150
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.8715341959334566,
|
| 2287 |
+
"grad_norm": 0.11341376602649689,
|
| 2288 |
+
"learning_rate": 2.0642906654343806e-05,
|
| 2289 |
+
"loss": 0.0864,
|
| 2290 |
+
"step": 16200
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.8773105360443623,
|
| 2294 |
+
"grad_norm": 3.484691858291626,
|
| 2295 |
+
"learning_rate": 2.061402495378928e-05,
|
| 2296 |
+
"loss": 0.2117,
|
| 2297 |
+
"step": 16250
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.883086876155268,
|
| 2301 |
+
"grad_norm": 0.03376946225762367,
|
| 2302 |
+
"learning_rate": 2.0585143253234748e-05,
|
| 2303 |
+
"loss": 0.1297,
|
| 2304 |
+
"step": 16300
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.8888632162661736,
|
| 2308 |
+
"grad_norm": 1.6973594427108765,
|
| 2309 |
+
"learning_rate": 2.055626155268022e-05,
|
| 2310 |
+
"loss": 0.1192,
|
| 2311 |
+
"step": 16350
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.8946395563770795,
|
| 2315 |
+
"grad_norm": 0.3922349512577057,
|
| 2316 |
+
"learning_rate": 2.0527379852125694e-05,
|
| 2317 |
+
"loss": 0.128,
|
| 2318 |
+
"step": 16400
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.9004158964879854,
|
| 2322 |
+
"grad_norm": 0.010527100414037704,
|
| 2323 |
+
"learning_rate": 2.0498498151571164e-05,
|
| 2324 |
+
"loss": 0.1211,
|
| 2325 |
+
"step": 16450
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.906192236598891,
|
| 2329 |
+
"grad_norm": 1.1090141534805298,
|
| 2330 |
+
"learning_rate": 2.0469616451016637e-05,
|
| 2331 |
+
"loss": 0.1871,
|
| 2332 |
+
"step": 16500
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.9119685767097967,
|
| 2336 |
+
"grad_norm": 1.5275694131851196,
|
| 2337 |
+
"learning_rate": 2.0440734750462106e-05,
|
| 2338 |
+
"loss": 0.1253,
|
| 2339 |
+
"step": 16550
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.9177449168207024,
|
| 2343 |
+
"grad_norm": 35.891883850097656,
|
| 2344 |
+
"learning_rate": 2.041185304990758e-05,
|
| 2345 |
+
"loss": 0.1592,
|
| 2346 |
+
"step": 16600
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.923521256931608,
|
| 2350 |
+
"grad_norm": 0.06564353406429291,
|
| 2351 |
+
"learning_rate": 2.038297134935305e-05,
|
| 2352 |
+
"loss": 0.099,
|
| 2353 |
+
"step": 16650
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.929297597042514,
|
| 2357 |
+
"grad_norm": 0.9027801752090454,
|
| 2358 |
+
"learning_rate": 2.0354089648798522e-05,
|
| 2359 |
+
"loss": 0.2061,
|
| 2360 |
+
"step": 16700
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.9350739371534196,
|
| 2364 |
+
"grad_norm": 0.034485165029764175,
|
| 2365 |
+
"learning_rate": 2.0325207948243995e-05,
|
| 2366 |
+
"loss": 0.0978,
|
| 2367 |
+
"step": 16750
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.9408502772643255,
|
| 2371 |
+
"grad_norm": 52.34895706176758,
|
| 2372 |
+
"learning_rate": 2.0296326247689464e-05,
|
| 2373 |
+
"loss": 0.1496,
|
| 2374 |
+
"step": 16800
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.9466266173752311,
|
| 2378 |
+
"grad_norm": 54.330509185791016,
|
| 2379 |
+
"learning_rate": 2.0267444547134937e-05,
|
| 2380 |
+
"loss": 0.1722,
|
| 2381 |
+
"step": 16850
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.9524029574861368,
|
| 2385 |
+
"grad_norm": 20.613256454467773,
|
| 2386 |
+
"learning_rate": 2.0238562846580407e-05,
|
| 2387 |
+
"loss": 0.1469,
|
| 2388 |
+
"step": 16900
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.9581792975970425,
|
| 2392 |
+
"grad_norm": 0.04331992566585541,
|
| 2393 |
+
"learning_rate": 2.020968114602588e-05,
|
| 2394 |
+
"loss": 0.1067,
|
| 2395 |
+
"step": 16950
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.9639556377079481,
|
| 2399 |
+
"grad_norm": 0.056098178029060364,
|
| 2400 |
+
"learning_rate": 2.018079944547135e-05,
|
| 2401 |
+
"loss": 0.1385,
|
| 2402 |
+
"step": 17000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.969731977818854,
|
| 2406 |
+
"grad_norm": 0.038275789469480515,
|
| 2407 |
+
"learning_rate": 2.0151917744916823e-05,
|
| 2408 |
+
"loss": 0.0932,
|
| 2409 |
+
"step": 17050
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.9755083179297597,
|
| 2413 |
+
"grad_norm": 0.04682379588484764,
|
| 2414 |
+
"learning_rate": 2.0123036044362296e-05,
|
| 2415 |
+
"loss": 0.1163,
|
| 2416 |
+
"step": 17100
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.9812846580406656,
|
| 2420 |
+
"grad_norm": 0.13566389679908752,
|
| 2421 |
+
"learning_rate": 2.0094154343807762e-05,
|
| 2422 |
+
"loss": 0.1669,
|
| 2423 |
+
"step": 17150
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.9870609981515712,
|
| 2427 |
+
"grad_norm": 36.21744155883789,
|
| 2428 |
+
"learning_rate": 2.0065272643253235e-05,
|
| 2429 |
+
"loss": 0.1608,
|
| 2430 |
+
"step": 17200
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.9928373382624769,
|
| 2434 |
+
"grad_norm": 0.40274855494499207,
|
| 2435 |
+
"learning_rate": 2.0036390942698704e-05,
|
| 2436 |
+
"loss": 0.1038,
|
| 2437 |
+
"step": 17250
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.9986136783733826,
|
| 2441 |
+
"grad_norm": 0.07136744260787964,
|
| 2442 |
+
"learning_rate": 2.0007509242144177e-05,
|
| 2443 |
+
"loss": 0.1248,
|
| 2444 |
+
"step": 17300
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 2.0,
|
| 2448 |
+
"eval_accuracy": 0.9578947368421052,
|
| 2449 |
+
"eval_f1": 0.9746439192361872,
|
| 2450 |
+
"eval_loss": 0.18152591586112976,
|
| 2451 |
+
"eval_precision": 0.9740340997966526,
|
| 2452 |
+
"eval_recall": 0.9752545027407987,
|
| 2453 |
+
"eval_runtime": 66.3035,
|
| 2454 |
+
"eval_samples_per_second": 116.057,
|
| 2455 |
+
"eval_steps_per_second": 14.509,
|
| 2456 |
+
"step": 17312
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 2.004390018484288,
|
| 2460 |
+
"grad_norm": 0.09601522982120514,
|
| 2461 |
+
"learning_rate": 1.9978627541589647e-05,
|
| 2462 |
+
"loss": 0.0684,
|
| 2463 |
+
"step": 17350
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 2.010166358595194,
|
| 2467 |
+
"grad_norm": 0.03754567727446556,
|
| 2468 |
+
"learning_rate": 1.994974584103512e-05,
|
| 2469 |
+
"loss": 0.089,
|
| 2470 |
+
"step": 17400
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 2.0159426987061,
|
| 2474 |
+
"grad_norm": 0.0021292606834322214,
|
| 2475 |
+
"learning_rate": 1.9920864140480593e-05,
|
| 2476 |
+
"loss": 0.0749,
|
| 2477 |
+
"step": 17450
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 2.0217190388170057,
|
| 2481 |
+
"grad_norm": 0.026555631309747696,
|
| 2482 |
+
"learning_rate": 1.9891982439926063e-05,
|
| 2483 |
+
"loss": 0.071,
|
| 2484 |
+
"step": 17500
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 2.0274953789279113,
|
| 2488 |
+
"grad_norm": 54.96882629394531,
|
| 2489 |
+
"learning_rate": 1.9863100739371536e-05,
|
| 2490 |
+
"loss": 0.0568,
|
| 2491 |
+
"step": 17550
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 2.033271719038817,
|
| 2495 |
+
"grad_norm": 0.016169143840670586,
|
| 2496 |
+
"learning_rate": 1.9834219038817005e-05,
|
| 2497 |
+
"loss": 0.0487,
|
| 2498 |
+
"step": 17600
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 2.0390480591497226,
|
| 2502 |
+
"grad_norm": 0.006945630069822073,
|
| 2503 |
+
"learning_rate": 1.9805337338262478e-05,
|
| 2504 |
+
"loss": 0.1352,
|
| 2505 |
+
"step": 17650
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 2.0448243992606283,
|
| 2509 |
+
"grad_norm": 0.03360309079289436,
|
| 2510 |
+
"learning_rate": 1.9776455637707948e-05,
|
| 2511 |
+
"loss": 0.0683,
|
| 2512 |
+
"step": 17700
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 2.050600739371534,
|
| 2516 |
+
"grad_norm": 0.02203032560646534,
|
| 2517 |
+
"learning_rate": 1.974757393715342e-05,
|
| 2518 |
+
"loss": 0.0333,
|
| 2519 |
+
"step": 17750
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 2.05637707948244,
|
| 2523 |
+
"grad_norm": 0.00957966037094593,
|
| 2524 |
+
"learning_rate": 1.9718692236598894e-05,
|
| 2525 |
+
"loss": 0.0655,
|
| 2526 |
+
"step": 17800
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 2.0621534195933457,
|
| 2530 |
+
"grad_norm": 0.07420410960912704,
|
| 2531 |
+
"learning_rate": 1.9689810536044363e-05,
|
| 2532 |
+
"loss": 0.0687,
|
| 2533 |
+
"step": 17850
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 2.0679297597042514,
|
| 2537 |
+
"grad_norm": 89.43473052978516,
|
| 2538 |
+
"learning_rate": 1.9660928835489836e-05,
|
| 2539 |
+
"loss": 0.0405,
|
| 2540 |
+
"step": 17900
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 2.073706099815157,
|
| 2544 |
+
"grad_norm": 1.9637092351913452,
|
| 2545 |
+
"learning_rate": 1.9632047134935306e-05,
|
| 2546 |
+
"loss": 0.0686,
|
| 2547 |
+
"step": 17950
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 2.0794824399260627,
|
| 2551 |
+
"grad_norm": 0.01743639074265957,
|
| 2552 |
+
"learning_rate": 1.960316543438078e-05,
|
| 2553 |
+
"loss": 0.0639,
|
| 2554 |
+
"step": 18000
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 2.0852587800369684,
|
| 2558 |
+
"grad_norm": 0.015308554284274578,
|
| 2559 |
+
"learning_rate": 1.957428373382625e-05,
|
| 2560 |
+
"loss": 0.0748,
|
| 2561 |
+
"step": 18050
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 2.0910351201478745,
|
| 2565 |
+
"grad_norm": 0.03445494547486305,
|
| 2566 |
+
"learning_rate": 1.9545402033271718e-05,
|
| 2567 |
+
"loss": 0.0462,
|
| 2568 |
+
"step": 18100
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 2.09681146025878,
|
| 2572 |
+
"grad_norm": 1.967775583267212,
|
| 2573 |
+
"learning_rate": 1.9516520332717188e-05,
|
| 2574 |
+
"loss": 0.0878,
|
| 2575 |
+
"step": 18150
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 2.102587800369686,
|
| 2579 |
+
"grad_norm": 25.589548110961914,
|
| 2580 |
+
"learning_rate": 1.948763863216266e-05,
|
| 2581 |
+
"loss": 0.1172,
|
| 2582 |
+
"step": 18200
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 2.1083641404805915,
|
| 2586 |
+
"grad_norm": 0.023420481011271477,
|
| 2587 |
+
"learning_rate": 1.9458756931608134e-05,
|
| 2588 |
+
"loss": 0.0941,
|
| 2589 |
+
"step": 18250
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 2.114140480591497,
|
| 2593 |
+
"grad_norm": 0.03860931098461151,
|
| 2594 |
+
"learning_rate": 1.9429875231053603e-05,
|
| 2595 |
+
"loss": 0.0947,
|
| 2596 |
+
"step": 18300
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 2.119916820702403,
|
| 2600 |
+
"grad_norm": 0.003364762756973505,
|
| 2601 |
+
"learning_rate": 1.9400993530499076e-05,
|
| 2602 |
+
"loss": 0.0172,
|
| 2603 |
+
"step": 18350
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 2.1256931608133085,
|
| 2607 |
+
"grad_norm": 0.06840373575687408,
|
| 2608 |
+
"learning_rate": 1.9372111829944546e-05,
|
| 2609 |
+
"loss": 0.0518,
|
| 2610 |
+
"step": 18400
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 2.1314695009242146,
|
| 2614 |
+
"grad_norm": 1.4184364080429077,
|
| 2615 |
+
"learning_rate": 1.934323012939002e-05,
|
| 2616 |
+
"loss": 0.1253,
|
| 2617 |
+
"step": 18450
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 2.1372458410351203,
|
| 2621 |
+
"grad_norm": 30.499197006225586,
|
| 2622 |
+
"learning_rate": 1.931434842883549e-05,
|
| 2623 |
+
"loss": 0.1116,
|
| 2624 |
+
"step": 18500
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 2.143022181146026,
|
| 2628 |
+
"grad_norm": 0.17084713280200958,
|
| 2629 |
+
"learning_rate": 1.928546672828096e-05,
|
| 2630 |
+
"loss": 0.1109,
|
| 2631 |
+
"step": 18550
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 2.1487985212569316,
|
| 2635 |
+
"grad_norm": 0.008835811167955399,
|
| 2636 |
+
"learning_rate": 1.9256585027726435e-05,
|
| 2637 |
+
"loss": 0.071,
|
| 2638 |
+
"step": 18600
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 2.1545748613678373,
|
| 2642 |
+
"grad_norm": 0.05329906940460205,
|
| 2643 |
+
"learning_rate": 1.9227703327171904e-05,
|
| 2644 |
+
"loss": 0.0709,
|
| 2645 |
+
"step": 18650
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 2.160351201478743,
|
| 2649 |
+
"grad_norm": 0.07247678935527802,
|
| 2650 |
+
"learning_rate": 1.9198821626617377e-05,
|
| 2651 |
+
"loss": 0.0793,
|
| 2652 |
+
"step": 18700
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 2.1661275415896486,
|
| 2656 |
+
"grad_norm": 0.00022657464432995766,
|
| 2657 |
+
"learning_rate": 1.9169939926062847e-05,
|
| 2658 |
+
"loss": 0.0532,
|
| 2659 |
+
"step": 18750
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 2.1719038817005547,
|
| 2663 |
+
"grad_norm": 54.57170867919922,
|
| 2664 |
+
"learning_rate": 1.914105822550832e-05,
|
| 2665 |
+
"loss": 0.0904,
|
| 2666 |
+
"step": 18800
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 2.1776802218114604,
|
| 2670 |
+
"grad_norm": 0.014400599524378777,
|
| 2671 |
+
"learning_rate": 1.911217652495379e-05,
|
| 2672 |
+
"loss": 0.0556,
|
| 2673 |
+
"step": 18850
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 2.183456561922366,
|
| 2677 |
+
"grad_norm": 0.03211145102977753,
|
| 2678 |
+
"learning_rate": 1.9083294824399262e-05,
|
| 2679 |
+
"loss": 0.0709,
|
| 2680 |
+
"step": 18900
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 2.1892329020332717,
|
| 2684 |
+
"grad_norm": 0.04402083531022072,
|
| 2685 |
+
"learning_rate": 1.9054413123844735e-05,
|
| 2686 |
+
"loss": 0.0448,
|
| 2687 |
+
"step": 18950
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 2.1950092421441774,
|
| 2691 |
+
"grad_norm": 0.023079989477992058,
|
| 2692 |
+
"learning_rate": 1.9025531423290205e-05,
|
| 2693 |
+
"loss": 0.0374,
|
| 2694 |
+
"step": 19000
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 2.200785582255083,
|
| 2698 |
+
"grad_norm": 0.0067682210355997086,
|
| 2699 |
+
"learning_rate": 1.8996649722735674e-05,
|
| 2700 |
+
"loss": 0.0694,
|
| 2701 |
+
"step": 19050
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 2.2065619223659887,
|
| 2705 |
+
"grad_norm": 0.038490429520606995,
|
| 2706 |
+
"learning_rate": 1.8967768022181144e-05,
|
| 2707 |
+
"loss": 0.1117,
|
| 2708 |
+
"step": 19100
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 2.212338262476895,
|
| 2712 |
+
"grad_norm": 19.319921493530273,
|
| 2713 |
+
"learning_rate": 1.8938886321626617e-05,
|
| 2714 |
+
"loss": 0.1028,
|
| 2715 |
+
"step": 19150
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 2.2181146025878005,
|
| 2719 |
+
"grad_norm": 45.28227615356445,
|
| 2720 |
+
"learning_rate": 1.8910004621072087e-05,
|
| 2721 |
+
"loss": 0.0578,
|
| 2722 |
+
"step": 19200
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 2.223890942698706,
|
| 2726 |
+
"grad_norm": 52.686073303222656,
|
| 2727 |
+
"learning_rate": 1.888112292051756e-05,
|
| 2728 |
+
"loss": 0.1111,
|
| 2729 |
+
"step": 19250
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 2.229667282809612,
|
| 2733 |
+
"grad_norm": 45.508628845214844,
|
| 2734 |
+
"learning_rate": 1.8852241219963033e-05,
|
| 2735 |
+
"loss": 0.0381,
|
| 2736 |
+
"step": 19300
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 2.2354436229205175,
|
| 2740 |
+
"grad_norm": 0.02187359519302845,
|
| 2741 |
+
"learning_rate": 1.8823359519408502e-05,
|
| 2742 |
+
"loss": 0.0448,
|
| 2743 |
+
"step": 19350
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 2.241219963031423,
|
| 2747 |
+
"grad_norm": 91.61556243896484,
|
| 2748 |
+
"learning_rate": 1.8794477818853975e-05,
|
| 2749 |
+
"loss": 0.0538,
|
| 2750 |
+
"step": 19400
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 2.2469963031423292,
|
| 2754 |
+
"grad_norm": 0.10931815952062607,
|
| 2755 |
+
"learning_rate": 1.8765596118299445e-05,
|
| 2756 |
+
"loss": 0.0913,
|
| 2757 |
+
"step": 19450
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 2.252772643253235,
|
| 2761 |
+
"grad_norm": 0.46113839745521545,
|
| 2762 |
+
"learning_rate": 1.8736714417744918e-05,
|
| 2763 |
+
"loss": 0.0679,
|
| 2764 |
+
"step": 19500
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 2.2585489833641406,
|
| 2768 |
+
"grad_norm": 62.958518981933594,
|
| 2769 |
+
"learning_rate": 1.8707832717190387e-05,
|
| 2770 |
+
"loss": 0.0342,
|
| 2771 |
+
"step": 19550
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 2.2643253234750462,
|
| 2775 |
+
"grad_norm": 0.15646789968013763,
|
| 2776 |
+
"learning_rate": 1.867895101663586e-05,
|
| 2777 |
+
"loss": 0.0708,
|
| 2778 |
+
"step": 19600
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 2.270101663585952,
|
| 2782 |
+
"grad_norm": 0.07558313012123108,
|
| 2783 |
+
"learning_rate": 1.8650069316081333e-05,
|
| 2784 |
+
"loss": 0.0334,
|
| 2785 |
+
"step": 19650
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 2.2758780036968576,
|
| 2789 |
+
"grad_norm": 0.012279776856303215,
|
| 2790 |
+
"learning_rate": 1.8621187615526803e-05,
|
| 2791 |
+
"loss": 0.0893,
|
| 2792 |
+
"step": 19700
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 2.281654343807763,
|
| 2796 |
+
"grad_norm": 16.578296661376953,
|
| 2797 |
+
"learning_rate": 1.8592305914972276e-05,
|
| 2798 |
+
"loss": 0.0806,
|
| 2799 |
+
"step": 19750
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 2.287430683918669,
|
| 2803 |
+
"grad_norm": 0.18503586947917938,
|
| 2804 |
+
"learning_rate": 1.8563424214417746e-05,
|
| 2805 |
+
"loss": 0.0546,
|
| 2806 |
+
"step": 19800
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 2.293207024029575,
|
| 2810 |
+
"grad_norm": 0.06100752577185631,
|
| 2811 |
+
"learning_rate": 1.853454251386322e-05,
|
| 2812 |
+
"loss": 0.0407,
|
| 2813 |
+
"step": 19850
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 2.2989833641404807,
|
| 2817 |
+
"grad_norm": 2.401160955429077,
|
| 2818 |
+
"learning_rate": 1.8505660813308688e-05,
|
| 2819 |
+
"loss": 0.0446,
|
| 2820 |
+
"step": 19900
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 2.3047597042513863,
|
| 2824 |
+
"grad_norm": 0.01384785957634449,
|
| 2825 |
+
"learning_rate": 1.847677911275416e-05,
|
| 2826 |
+
"loss": 0.0615,
|
| 2827 |
+
"step": 19950
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 2.310536044362292,
|
| 2831 |
+
"grad_norm": 0.0038583818823099136,
|
| 2832 |
+
"learning_rate": 1.844789741219963e-05,
|
| 2833 |
+
"loss": 0.1048,
|
| 2834 |
+
"step": 20000
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 2.3163123844731976,
|
| 2838 |
+
"grad_norm": 4.787203311920166,
|
| 2839 |
+
"learning_rate": 1.84190157116451e-05,
|
| 2840 |
+
"loss": 0.0489,
|
| 2841 |
+
"step": 20050
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 2.3220887245841038,
|
| 2845 |
+
"grad_norm": 0.0021266438998281956,
|
| 2846 |
+
"learning_rate": 1.8390134011090573e-05,
|
| 2847 |
+
"loss": 0.0415,
|
| 2848 |
+
"step": 20100
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 2.3278650646950094,
|
| 2852 |
+
"grad_norm": 0.0017306450754404068,
|
| 2853 |
+
"learning_rate": 1.8361252310536043e-05,
|
| 2854 |
+
"loss": 0.0299,
|
| 2855 |
+
"step": 20150
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 2.333641404805915,
|
| 2859 |
+
"grad_norm": 0.006416016258299351,
|
| 2860 |
+
"learning_rate": 1.8332370609981516e-05,
|
| 2861 |
+
"loss": 0.0536,
|
| 2862 |
+
"step": 20200
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 2.3394177449168208,
|
| 2866 |
+
"grad_norm": 0.0008355600293725729,
|
| 2867 |
+
"learning_rate": 1.8303488909426986e-05,
|
| 2868 |
+
"loss": 0.0307,
|
| 2869 |
+
"step": 20250
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 2.3451940850277264,
|
| 2873 |
+
"grad_norm": 0.006618143059313297,
|
| 2874 |
+
"learning_rate": 1.827460720887246e-05,
|
| 2875 |
+
"loss": 0.1172,
|
| 2876 |
+
"step": 20300
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 2.350970425138632,
|
| 2880 |
+
"grad_norm": 0.0006379440310411155,
|
| 2881 |
+
"learning_rate": 1.8245725508317928e-05,
|
| 2882 |
+
"loss": 0.0729,
|
| 2883 |
+
"step": 20350
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 2.3567467652495377,
|
| 2887 |
+
"grad_norm": 0.007531580049544573,
|
| 2888 |
+
"learning_rate": 1.82168438077634e-05,
|
| 2889 |
+
"loss": 0.0573,
|
| 2890 |
+
"step": 20400
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 2.3625231053604434,
|
| 2894 |
+
"grad_norm": 0.01422103587538004,
|
| 2895 |
+
"learning_rate": 1.8187962107208874e-05,
|
| 2896 |
+
"loss": 0.0623,
|
| 2897 |
+
"step": 20450
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 2.3682994454713495,
|
| 2901 |
+
"grad_norm": 0.029098376631736755,
|
| 2902 |
+
"learning_rate": 1.8159080406654344e-05,
|
| 2903 |
+
"loss": 0.052,
|
| 2904 |
+
"step": 20500
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 2.374075785582255,
|
| 2908 |
+
"grad_norm": 0.29854094982147217,
|
| 2909 |
+
"learning_rate": 1.8130198706099817e-05,
|
| 2910 |
+
"loss": 0.1549,
|
| 2911 |
+
"step": 20550
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 2.379852125693161,
|
| 2915 |
+
"grad_norm": 0.042241454124450684,
|
| 2916 |
+
"learning_rate": 1.8101317005545286e-05,
|
| 2917 |
+
"loss": 0.0643,
|
| 2918 |
+
"step": 20600
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 2.3856284658040665,
|
| 2922 |
+
"grad_norm": 0.003535840893164277,
|
| 2923 |
+
"learning_rate": 1.807243530499076e-05,
|
| 2924 |
+
"loss": 0.0682,
|
| 2925 |
+
"step": 20650
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 2.391404805914972,
|
| 2929 |
+
"grad_norm": 0.10705076158046722,
|
| 2930 |
+
"learning_rate": 1.804355360443623e-05,
|
| 2931 |
+
"loss": 0.1205,
|
| 2932 |
+
"step": 20700
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 2.397181146025878,
|
| 2936 |
+
"grad_norm": 0.0034834675025194883,
|
| 2937 |
+
"learning_rate": 1.8014671903881702e-05,
|
| 2938 |
+
"loss": 0.0523,
|
| 2939 |
+
"step": 20750
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 2.402957486136784,
|
| 2943 |
+
"grad_norm": 0.01728115789592266,
|
| 2944 |
+
"learning_rate": 1.7985790203327175e-05,
|
| 2945 |
+
"loss": 0.0811,
|
| 2946 |
+
"step": 20800
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 2.4087338262476896,
|
| 2950 |
+
"grad_norm": 1.6176636219024658,
|
| 2951 |
+
"learning_rate": 1.7956908502772645e-05,
|
| 2952 |
+
"loss": 0.0514,
|
| 2953 |
+
"step": 20850
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 2.4145101663585953,
|
| 2957 |
+
"grad_norm": 0.0028283593710511923,
|
| 2958 |
+
"learning_rate": 1.7928026802218117e-05,
|
| 2959 |
+
"loss": 0.0496,
|
| 2960 |
+
"step": 20900
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 2.420286506469501,
|
| 2964 |
+
"grad_norm": 2.9472639560699463,
|
| 2965 |
+
"learning_rate": 1.7899145101663587e-05,
|
| 2966 |
+
"loss": 0.0559,
|
| 2967 |
+
"step": 20950
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 2.4260628465804066,
|
| 2971 |
+
"grad_norm": 0.01362446416169405,
|
| 2972 |
+
"learning_rate": 1.7870263401109057e-05,
|
| 2973 |
+
"loss": 0.085,
|
| 2974 |
+
"step": 21000
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 2.4318391866913123,
|
| 2978 |
+
"grad_norm": 1.4009047746658325,
|
| 2979 |
+
"learning_rate": 1.7841381700554526e-05,
|
| 2980 |
+
"loss": 0.0976,
|
| 2981 |
+
"step": 21050
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 2.437615526802218,
|
| 2985 |
+
"grad_norm": 0.05615925043821335,
|
| 2986 |
+
"learning_rate": 1.78125e-05,
|
| 2987 |
+
"loss": 0.1164,
|
| 2988 |
+
"step": 21100
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 2.4433918669131236,
|
| 2992 |
+
"grad_norm": 0.022135844454169273,
|
| 2993 |
+
"learning_rate": 1.7783618299445472e-05,
|
| 2994 |
+
"loss": 0.0964,
|
| 2995 |
+
"step": 21150
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 2.4491682070240297,
|
| 2999 |
+
"grad_norm": 0.021285168826580048,
|
| 3000 |
+
"learning_rate": 1.7754736598890942e-05,
|
| 3001 |
+
"loss": 0.0489,
|
| 3002 |
+
"step": 21200
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 2.4549445471349354,
|
| 3006 |
+
"grad_norm": 0.6468170285224915,
|
| 3007 |
+
"learning_rate": 1.7725854898336415e-05,
|
| 3008 |
+
"loss": 0.0661,
|
| 3009 |
+
"step": 21250
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 2.460720887245841,
|
| 3013 |
+
"grad_norm": 0.028381478041410446,
|
| 3014 |
+
"learning_rate": 1.7696973197781884e-05,
|
| 3015 |
+
"loss": 0.0675,
|
| 3016 |
+
"step": 21300
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 2.4664972273567467,
|
| 3020 |
+
"grad_norm": 0.043751172721385956,
|
| 3021 |
+
"learning_rate": 1.7668091497227357e-05,
|
| 3022 |
+
"loss": 0.0378,
|
| 3023 |
+
"step": 21350
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 2.4722735674676524,
|
| 3027 |
+
"grad_norm": 0.01835785247385502,
|
| 3028 |
+
"learning_rate": 1.7639209796672827e-05,
|
| 3029 |
+
"loss": 0.0654,
|
| 3030 |
+
"step": 21400
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 2.4780499075785585,
|
| 3034 |
+
"grad_norm": 0.0018048484344035387,
|
| 3035 |
+
"learning_rate": 1.76103280961183e-05,
|
| 3036 |
+
"loss": 0.0699,
|
| 3037 |
+
"step": 21450
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 2.483826247689464,
|
| 3041 |
+
"grad_norm": 0.016551557928323746,
|
| 3042 |
+
"learning_rate": 1.7581446395563773e-05,
|
| 3043 |
+
"loss": 0.0465,
|
| 3044 |
+
"step": 21500
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 2.48960258780037,
|
| 3048 |
+
"grad_norm": 0.004093922674655914,
|
| 3049 |
+
"learning_rate": 1.7552564695009243e-05,
|
| 3050 |
+
"loss": 0.0573,
|
| 3051 |
+
"step": 21550
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 2.4953789279112755,
|
| 3055 |
+
"grad_norm": 0.005556577816605568,
|
| 3056 |
+
"learning_rate": 1.7523682994454716e-05,
|
| 3057 |
+
"loss": 0.1278,
|
| 3058 |
+
"step": 21600
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 2.501155268022181,
|
| 3062 |
+
"grad_norm": 5.791269302368164,
|
| 3063 |
+
"learning_rate": 1.7494801293900185e-05,
|
| 3064 |
+
"loss": 0.0863,
|
| 3065 |
+
"step": 21650
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 2.506931608133087,
|
| 3069 |
+
"grad_norm": 0.06026982143521309,
|
| 3070 |
+
"learning_rate": 1.7465919593345658e-05,
|
| 3071 |
+
"loss": 0.1049,
|
| 3072 |
+
"step": 21700
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 2.5127079482439925,
|
| 3076 |
+
"grad_norm": 0.03967379778623581,
|
| 3077 |
+
"learning_rate": 1.7437037892791128e-05,
|
| 3078 |
+
"loss": 0.1215,
|
| 3079 |
+
"step": 21750
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 2.518484288354898,
|
| 3083 |
+
"grad_norm": 0.12794901430606842,
|
| 3084 |
+
"learning_rate": 1.74081561922366e-05,
|
| 3085 |
+
"loss": 0.0499,
|
| 3086 |
+
"step": 21800
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 2.524260628465804,
|
| 3090 |
+
"grad_norm": 0.0026088140439242125,
|
| 3091 |
+
"learning_rate": 1.7379274491682074e-05,
|
| 3092 |
+
"loss": 0.042,
|
| 3093 |
+
"step": 21850
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 2.53003696857671,
|
| 3097 |
+
"grad_norm": 145.44888305664062,
|
| 3098 |
+
"learning_rate": 1.7350392791127543e-05,
|
| 3099 |
+
"loss": 0.0548,
|
| 3100 |
+
"step": 21900
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 2.5358133086876156,
|
| 3104 |
+
"grad_norm": 156.18600463867188,
|
| 3105 |
+
"learning_rate": 1.7321511090573013e-05,
|
| 3106 |
+
"loss": 0.0278,
|
| 3107 |
+
"step": 21950
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 2.5415896487985212,
|
| 3111 |
+
"grad_norm": 0.006327302660793066,
|
| 3112 |
+
"learning_rate": 1.7292629390018483e-05,
|
| 3113 |
+
"loss": 0.0823,
|
| 3114 |
+
"step": 22000
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 2.547365988909427,
|
| 3118 |
+
"grad_norm": 0.02358861453831196,
|
| 3119 |
+
"learning_rate": 1.7263747689463956e-05,
|
| 3120 |
+
"loss": 0.1268,
|
| 3121 |
+
"step": 22050
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 2.5531423290203326,
|
| 3125 |
+
"grad_norm": 0.007511517498642206,
|
| 3126 |
+
"learning_rate": 1.7234865988909425e-05,
|
| 3127 |
+
"loss": 0.0605,
|
| 3128 |
+
"step": 22100
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 2.5589186691312387,
|
| 3132 |
+
"grad_norm": 0.04025644809007645,
|
| 3133 |
+
"learning_rate": 1.7205984288354898e-05,
|
| 3134 |
+
"loss": 0.0633,
|
| 3135 |
+
"step": 22150
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 2.5646950092421443,
|
| 3139 |
+
"grad_norm": 0.001463131164200604,
|
| 3140 |
+
"learning_rate": 1.7177102587800368e-05,
|
| 3141 |
+
"loss": 0.0461,
|
| 3142 |
+
"step": 22200
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 2.57047134935305,
|
| 3146 |
+
"grad_norm": 0.07968592643737793,
|
| 3147 |
+
"learning_rate": 1.714822088724584e-05,
|
| 3148 |
+
"loss": 0.0984,
|
| 3149 |
+
"step": 22250
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 2.5762476894639557,
|
| 3153 |
+
"grad_norm": 22.19074058532715,
|
| 3154 |
+
"learning_rate": 1.7119339186691314e-05,
|
| 3155 |
+
"loss": 0.0538,
|
| 3156 |
+
"step": 22300
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 2.5820240295748613,
|
| 3160 |
+
"grad_norm": 0.10924549400806427,
|
| 3161 |
+
"learning_rate": 1.7090457486136783e-05,
|
| 3162 |
+
"loss": 0.0456,
|
| 3163 |
+
"step": 22350
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 2.587800369685767,
|
| 3167 |
+
"grad_norm": 0.1614380031824112,
|
| 3168 |
+
"learning_rate": 1.7061575785582256e-05,
|
| 3169 |
+
"loss": 0.1034,
|
| 3170 |
+
"step": 22400
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 2.5935767097966727,
|
| 3174 |
+
"grad_norm": 57.85542678833008,
|
| 3175 |
+
"learning_rate": 1.7032694085027726e-05,
|
| 3176 |
+
"loss": 0.0572,
|
| 3177 |
+
"step": 22450
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 2.5993530499075783,
|
| 3181 |
+
"grad_norm": 0.04180502891540527,
|
| 3182 |
+
"learning_rate": 1.70038123844732e-05,
|
| 3183 |
+
"loss": 0.0809,
|
| 3184 |
+
"step": 22500
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 2.6051293900184844,
|
| 3188 |
+
"grad_norm": 13.526185989379883,
|
| 3189 |
+
"learning_rate": 1.697493068391867e-05,
|
| 3190 |
+
"loss": 0.1018,
|
| 3191 |
+
"step": 22550
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 2.61090573012939,
|
| 3195 |
+
"grad_norm": 0.002297380706295371,
|
| 3196 |
+
"learning_rate": 1.694604898336414e-05,
|
| 3197 |
+
"loss": 0.0488,
|
| 3198 |
+
"step": 22600
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 2.6166820702402958,
|
| 3202 |
+
"grad_norm": 4.981602668762207,
|
| 3203 |
+
"learning_rate": 1.6917167282809615e-05,
|
| 3204 |
+
"loss": 0.0484,
|
| 3205 |
+
"step": 22650
|
| 3206 |
+
},
|
| 3207 |
+
{
|
| 3208 |
+
"epoch": 2.6224584103512014,
|
| 3209 |
+
"grad_norm": 0.026757080107927322,
|
| 3210 |
+
"learning_rate": 1.6888285582255084e-05,
|
| 3211 |
+
"loss": 0.0784,
|
| 3212 |
+
"step": 22700
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 2.628234750462107,
|
| 3216 |
+
"grad_norm": 0.018137283623218536,
|
| 3217 |
+
"learning_rate": 1.6859403881700557e-05,
|
| 3218 |
+
"loss": 0.0344,
|
| 3219 |
+
"step": 22750
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 2.634011090573013,
|
| 3223 |
+
"grad_norm": 19.682188034057617,
|
| 3224 |
+
"learning_rate": 1.6830522181146027e-05,
|
| 3225 |
+
"loss": 0.1079,
|
| 3226 |
+
"step": 22800
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 2.639787430683919,
|
| 3230 |
+
"grad_norm": 0.053726810961961746,
|
| 3231 |
+
"learning_rate": 1.68016404805915e-05,
|
| 3232 |
+
"loss": 0.0839,
|
| 3233 |
+
"step": 22850
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 2.6455637707948245,
|
| 3237 |
+
"grad_norm": 11.828213691711426,
|
| 3238 |
+
"learning_rate": 1.6772758780036966e-05,
|
| 3239 |
+
"loss": 0.0883,
|
| 3240 |
+
"step": 22900
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 2.65134011090573,
|
| 3244 |
+
"grad_norm": 0.021828031167387962,
|
| 3245 |
+
"learning_rate": 1.674387707948244e-05,
|
| 3246 |
+
"loss": 0.0175,
|
| 3247 |
+
"step": 22950
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 2.657116451016636,
|
| 3251 |
+
"grad_norm": 0.007194809149950743,
|
| 3252 |
+
"learning_rate": 1.6714995378927912e-05,
|
| 3253 |
+
"loss": 0.0395,
|
| 3254 |
+
"step": 23000
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 2.6628927911275415,
|
| 3258 |
+
"grad_norm": 0.020130673423409462,
|
| 3259 |
+
"learning_rate": 1.668611367837338e-05,
|
| 3260 |
+
"loss": 0.1246,
|
| 3261 |
+
"step": 23050
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 2.668669131238447,
|
| 3265 |
+
"grad_norm": 0.01889081858098507,
|
| 3266 |
+
"learning_rate": 1.6657231977818854e-05,
|
| 3267 |
+
"loss": 0.0972,
|
| 3268 |
+
"step": 23100
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 2.674445471349353,
|
| 3272 |
+
"grad_norm": 0.05320267006754875,
|
| 3273 |
+
"learning_rate": 1.6628350277264324e-05,
|
| 3274 |
+
"loss": 0.085,
|
| 3275 |
+
"step": 23150
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 2.6802218114602585,
|
| 3279 |
+
"grad_norm": 0.08507529646158218,
|
| 3280 |
+
"learning_rate": 1.6599468576709797e-05,
|
| 3281 |
+
"loss": 0.0958,
|
| 3282 |
+
"step": 23200
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 2.6859981515711646,
|
| 3286 |
+
"grad_norm": 10.60571002960205,
|
| 3287 |
+
"learning_rate": 1.6570586876155267e-05,
|
| 3288 |
+
"loss": 0.1128,
|
| 3289 |
+
"step": 23250
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 2.6917744916820703,
|
| 3293 |
+
"grad_norm": 0.062405217438936234,
|
| 3294 |
+
"learning_rate": 1.654170517560074e-05,
|
| 3295 |
+
"loss": 0.028,
|
| 3296 |
+
"step": 23300
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 2.697550831792976,
|
| 3300 |
+
"grad_norm": 0.0019212329061701894,
|
| 3301 |
+
"learning_rate": 1.6512823475046213e-05,
|
| 3302 |
+
"loss": 0.0406,
|
| 3303 |
+
"step": 23350
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 2.7033271719038816,
|
| 3307 |
+
"grad_norm": 0.011281181126832962,
|
| 3308 |
+
"learning_rate": 1.6483941774491682e-05,
|
| 3309 |
+
"loss": 0.0397,
|
| 3310 |
+
"step": 23400
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 2.7091035120147873,
|
| 3314 |
+
"grad_norm": 0.370628297328949,
|
| 3315 |
+
"learning_rate": 1.6455060073937155e-05,
|
| 3316 |
+
"loss": 0.1138,
|
| 3317 |
+
"step": 23450
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 2.7148798521256934,
|
| 3321 |
+
"grad_norm": 0.038135793060064316,
|
| 3322 |
+
"learning_rate": 1.6426178373382625e-05,
|
| 3323 |
+
"loss": 0.095,
|
| 3324 |
+
"step": 23500
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 2.720656192236599,
|
| 3328 |
+
"grad_norm": 0.025372378528118134,
|
| 3329 |
+
"learning_rate": 1.6397296672828098e-05,
|
| 3330 |
+
"loss": 0.0714,
|
| 3331 |
+
"step": 23550
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 2.7264325323475047,
|
| 3335 |
+
"grad_norm": 0.42094916105270386,
|
| 3336 |
+
"learning_rate": 1.6368414972273567e-05,
|
| 3337 |
+
"loss": 0.0601,
|
| 3338 |
+
"step": 23600
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 2.7322088724584104,
|
| 3342 |
+
"grad_norm": 0.010748348198831081,
|
| 3343 |
+
"learning_rate": 1.633953327171904e-05,
|
| 3344 |
+
"loss": 0.096,
|
| 3345 |
+
"step": 23650
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 2.737985212569316,
|
| 3349 |
+
"grad_norm": 0.06238965317606926,
|
| 3350 |
+
"learning_rate": 1.6310651571164513e-05,
|
| 3351 |
+
"loss": 0.0996,
|
| 3352 |
+
"step": 23700
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 2.7437615526802217,
|
| 3356 |
+
"grad_norm": 0.07324642688035965,
|
| 3357 |
+
"learning_rate": 1.6281769870609983e-05,
|
| 3358 |
+
"loss": 0.0563,
|
| 3359 |
+
"step": 23750
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 2.7495378927911274,
|
| 3363 |
+
"grad_norm": 0.0048204283230006695,
|
| 3364 |
+
"learning_rate": 1.6252888170055456e-05,
|
| 3365 |
+
"loss": 0.1134,
|
| 3366 |
+
"step": 23800
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 2.755314232902033,
|
| 3370 |
+
"grad_norm": 0.053540315479040146,
|
| 3371 |
+
"learning_rate": 1.6224006469500922e-05,
|
| 3372 |
+
"loss": 0.1167,
|
| 3373 |
+
"step": 23850
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 2.761090573012939,
|
| 3377 |
+
"grad_norm": 0.01278363075107336,
|
| 3378 |
+
"learning_rate": 1.6195124768946395e-05,
|
| 3379 |
+
"loss": 0.0775,
|
| 3380 |
+
"step": 23900
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 2.766866913123845,
|
| 3384 |
+
"grad_norm": 57.40664291381836,
|
| 3385 |
+
"learning_rate": 1.6166243068391865e-05,
|
| 3386 |
+
"loss": 0.0348,
|
| 3387 |
+
"step": 23950
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 2.7726432532347505,
|
| 3391 |
+
"grad_norm": 0.0012289739679545164,
|
| 3392 |
+
"learning_rate": 1.6137361367837338e-05,
|
| 3393 |
+
"loss": 0.036,
|
| 3394 |
+
"step": 24000
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 2.778419593345656,
|
| 3398 |
+
"grad_norm": 0.014386254362761974,
|
| 3399 |
+
"learning_rate": 1.610847966728281e-05,
|
| 3400 |
+
"loss": 0.0628,
|
| 3401 |
+
"step": 24050
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 2.784195933456562,
|
| 3405 |
+
"grad_norm": 0.06621140241622925,
|
| 3406 |
+
"learning_rate": 1.607959796672828e-05,
|
| 3407 |
+
"loss": 0.0611,
|
| 3408 |
+
"step": 24100
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 2.789972273567468,
|
| 3412 |
+
"grad_norm": 0.07706930488348007,
|
| 3413 |
+
"learning_rate": 1.6050716266173753e-05,
|
| 3414 |
+
"loss": 0.1075,
|
| 3415 |
+
"step": 24150
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 2.7957486136783736,
|
| 3419 |
+
"grad_norm": 111.92048645019531,
|
| 3420 |
+
"learning_rate": 1.6021834565619223e-05,
|
| 3421 |
+
"loss": 0.0471,
|
| 3422 |
+
"step": 24200
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 2.8015249537892792,
|
| 3426 |
+
"grad_norm": 0.1538689136505127,
|
| 3427 |
+
"learning_rate": 1.5992952865064696e-05,
|
| 3428 |
+
"loss": 0.0683,
|
| 3429 |
+
"step": 24250
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 2.807301293900185,
|
| 3433 |
+
"grad_norm": 0.008512681350111961,
|
| 3434 |
+
"learning_rate": 1.5964071164510166e-05,
|
| 3435 |
+
"loss": 0.0197,
|
| 3436 |
+
"step": 24300
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 2.8130776340110906,
|
| 3440 |
+
"grad_norm": 0.002703199628740549,
|
| 3441 |
+
"learning_rate": 1.593518946395564e-05,
|
| 3442 |
+
"loss": 0.0414,
|
| 3443 |
+
"step": 24350
|
| 3444 |
+
},
|
| 3445 |
+
{
|
| 3446 |
+
"epoch": 2.8188539741219962,
|
| 3447 |
+
"grad_norm": 0.006974892690777779,
|
| 3448 |
+
"learning_rate": 1.5906307763401108e-05,
|
| 3449 |
+
"loss": 0.0344,
|
| 3450 |
+
"step": 24400
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 2.824630314232902,
|
| 3454 |
+
"grad_norm": 28.003032684326172,
|
| 3455 |
+
"learning_rate": 1.587742606284658e-05,
|
| 3456 |
+
"loss": 0.0857,
|
| 3457 |
+
"step": 24450
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 2.8304066543438076,
|
| 3461 |
+
"grad_norm": 2.126615524291992,
|
| 3462 |
+
"learning_rate": 1.5848544362292054e-05,
|
| 3463 |
+
"loss": 0.0456,
|
| 3464 |
+
"step": 24500
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 2.8361829944547132,
|
| 3468 |
+
"grad_norm": 0.006181332748383284,
|
| 3469 |
+
"learning_rate": 1.5819662661737524e-05,
|
| 3470 |
+
"loss": 0.1308,
|
| 3471 |
+
"step": 24550
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 2.8419593345656193,
|
| 3475 |
+
"grad_norm": 0.004711877554655075,
|
| 3476 |
+
"learning_rate": 1.5790780961182997e-05,
|
| 3477 |
+
"loss": 0.0591,
|
| 3478 |
+
"step": 24600
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 2.847735674676525,
|
| 3482 |
+
"grad_norm": 0.006115751340985298,
|
| 3483 |
+
"learning_rate": 1.5761899260628466e-05,
|
| 3484 |
+
"loss": 0.044,
|
| 3485 |
+
"step": 24650
|
| 3486 |
+
},
|
| 3487 |
+
{
|
| 3488 |
+
"epoch": 2.8535120147874307,
|
| 3489 |
+
"grad_norm": 0.11933701485395432,
|
| 3490 |
+
"learning_rate": 1.573301756007394e-05,
|
| 3491 |
+
"loss": 0.0624,
|
| 3492 |
+
"step": 24700
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 2.8592883548983363,
|
| 3496 |
+
"grad_norm": 0.02603713423013687,
|
| 3497 |
+
"learning_rate": 1.570413585951941e-05,
|
| 3498 |
+
"loss": 0.1565,
|
| 3499 |
+
"step": 24750
|
| 3500 |
+
},
|
| 3501 |
+
{
|
| 3502 |
+
"epoch": 2.865064695009242,
|
| 3503 |
+
"grad_norm": 0.010369190014898777,
|
| 3504 |
+
"learning_rate": 1.567525415896488e-05,
|
| 3505 |
+
"loss": 0.0535,
|
| 3506 |
+
"step": 24800
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 2.870841035120148,
|
| 3510 |
+
"grad_norm": 132.8582305908203,
|
| 3511 |
+
"learning_rate": 1.564637245841035e-05,
|
| 3512 |
+
"loss": 0.0272,
|
| 3513 |
+
"step": 24850
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 2.8766173752310538,
|
| 3517 |
+
"grad_norm": 0.0030447840690612793,
|
| 3518 |
+
"learning_rate": 1.561749075785582e-05,
|
| 3519 |
+
"loss": 0.142,
|
| 3520 |
+
"step": 24900
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 2.8823937153419594,
|
| 3524 |
+
"grad_norm": 0.05927830561995506,
|
| 3525 |
+
"learning_rate": 1.5588609057301294e-05,
|
| 3526 |
+
"loss": 0.0723,
|
| 3527 |
+
"step": 24950
|
| 3528 |
+
},
|
| 3529 |
+
{
|
| 3530 |
+
"epoch": 2.888170055452865,
|
| 3531 |
+
"grad_norm": 0.01300797425210476,
|
| 3532 |
+
"learning_rate": 1.5559727356746764e-05,
|
| 3533 |
+
"loss": 0.072,
|
| 3534 |
+
"step": 25000
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 2.8939463955637708,
|
| 3538 |
+
"grad_norm": 0.12654338777065277,
|
| 3539 |
+
"learning_rate": 1.5530845656192237e-05,
|
| 3540 |
+
"loss": 0.059,
|
| 3541 |
+
"step": 25050
|
| 3542 |
+
},
|
| 3543 |
+
{
|
| 3544 |
+
"epoch": 2.8997227356746764,
|
| 3545 |
+
"grad_norm": 0.4156947135925293,
|
| 3546 |
+
"learning_rate": 1.5501963955637706e-05,
|
| 3547 |
+
"loss": 0.0213,
|
| 3548 |
+
"step": 25100
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 2.905499075785582,
|
| 3552 |
+
"grad_norm": 63.0364875793457,
|
| 3553 |
+
"learning_rate": 1.547308225508318e-05,
|
| 3554 |
+
"loss": 0.0765,
|
| 3555 |
+
"step": 25150
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 2.9112754158964878,
|
| 3559 |
+
"grad_norm": 24.513423919677734,
|
| 3560 |
+
"learning_rate": 1.5444200554528652e-05,
|
| 3561 |
+
"loss": 0.0121,
|
| 3562 |
+
"step": 25200
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 2.917051756007394,
|
| 3566 |
+
"grad_norm": 0.007328661624342203,
|
| 3567 |
+
"learning_rate": 1.5415318853974122e-05,
|
| 3568 |
+
"loss": 0.1266,
|
| 3569 |
+
"step": 25250
|
| 3570 |
+
},
|
| 3571 |
+
{
|
| 3572 |
+
"epoch": 2.9228280961182995,
|
| 3573 |
+
"grad_norm": 0.007661913055926561,
|
| 3574 |
+
"learning_rate": 1.5386437153419595e-05,
|
| 3575 |
+
"loss": 0.0489,
|
| 3576 |
+
"step": 25300
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 2.928604436229205,
|
| 3580 |
+
"grad_norm": 0.007173069752752781,
|
| 3581 |
+
"learning_rate": 1.5357555452865064e-05,
|
| 3582 |
+
"loss": 0.0534,
|
| 3583 |
+
"step": 25350
|
| 3584 |
+
},
|
| 3585 |
+
{
|
| 3586 |
+
"epoch": 2.934380776340111,
|
| 3587 |
+
"grad_norm": 0.021633770316839218,
|
| 3588 |
+
"learning_rate": 1.5328673752310537e-05,
|
| 3589 |
+
"loss": 0.0506,
|
| 3590 |
+
"step": 25400
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 2.9401571164510165,
|
| 3594 |
+
"grad_norm": 0.13166797161102295,
|
| 3595 |
+
"learning_rate": 1.5299792051756007e-05,
|
| 3596 |
+
"loss": 0.0893,
|
| 3597 |
+
"step": 25450
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 2.9459334565619226,
|
| 3601 |
+
"grad_norm": 0.0035212670918554068,
|
| 3602 |
+
"learning_rate": 1.527091035120148e-05,
|
| 3603 |
+
"loss": 0.0566,
|
| 3604 |
+
"step": 25500
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 2.9517097966728283,
|
| 3608 |
+
"grad_norm": 0.03263075277209282,
|
| 3609 |
+
"learning_rate": 1.5242028650646951e-05,
|
| 3610 |
+
"loss": 0.067,
|
| 3611 |
+
"step": 25550
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 2.957486136783734,
|
| 3615 |
+
"grad_norm": 0.026449991390109062,
|
| 3616 |
+
"learning_rate": 1.5213146950092421e-05,
|
| 3617 |
+
"loss": 0.1531,
|
| 3618 |
+
"step": 25600
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 2.9632624768946396,
|
| 3622 |
+
"grad_norm": 0.01947343908250332,
|
| 3623 |
+
"learning_rate": 1.5184265249537894e-05,
|
| 3624 |
+
"loss": 0.079,
|
| 3625 |
+
"step": 25650
|
| 3626 |
+
},
|
| 3627 |
+
{
|
| 3628 |
+
"epoch": 2.9690388170055453,
|
| 3629 |
+
"grad_norm": 0.0033325115218758583,
|
| 3630 |
+
"learning_rate": 1.5155383548983364e-05,
|
| 3631 |
+
"loss": 0.023,
|
| 3632 |
+
"step": 25700
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 2.974815157116451,
|
| 3636 |
+
"grad_norm": 0.0024256263859570026,
|
| 3637 |
+
"learning_rate": 1.5126501848428837e-05,
|
| 3638 |
+
"loss": 0.0318,
|
| 3639 |
+
"step": 25750
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 2.9805914972273566,
|
| 3643 |
+
"grad_norm": 0.019877901300787926,
|
| 3644 |
+
"learning_rate": 1.5097620147874306e-05,
|
| 3645 |
+
"loss": 0.0253,
|
| 3646 |
+
"step": 25800
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 2.9863678373382623,
|
| 3650 |
+
"grad_norm": 0.352357417345047,
|
| 3651 |
+
"learning_rate": 1.5068738447319779e-05,
|
| 3652 |
+
"loss": 0.0791,
|
| 3653 |
+
"step": 25850
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 2.992144177449168,
|
| 3657 |
+
"grad_norm": 0.031420253217220306,
|
| 3658 |
+
"learning_rate": 1.5039856746765252e-05,
|
| 3659 |
+
"loss": 0.0662,
|
| 3660 |
+
"step": 25900
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 2.997920517560074,
|
| 3664 |
+
"grad_norm": 7.931340217590332,
|
| 3665 |
+
"learning_rate": 1.5010975046210722e-05,
|
| 3666 |
+
"loss": 0.0341,
|
| 3667 |
+
"step": 25950
|
| 3668 |
+
},
|
| 3669 |
+
{
|
| 3670 |
+
"epoch": 3.0,
|
| 3671 |
+
"eval_accuracy": 0.9629629629629629,
|
| 3672 |
+
"eval_f1": 0.9778571983528863,
|
| 3673 |
+
"eval_loss": 0.20652833580970764,
|
| 3674 |
+
"eval_precision": 0.9702436016034536,
|
| 3675 |
+
"eval_recall": 0.9855912294440093,
|
| 3676 |
+
"eval_runtime": 66.768,
|
| 3677 |
+
"eval_samples_per_second": 115.25,
|
| 3678 |
+
"eval_steps_per_second": 14.408,
|
| 3679 |
+
"step": 25968
|
| 3680 |
+
}
|
| 3681 |
+
],
|
| 3682 |
+
"logging_steps": 50,
|
| 3683 |
+
"max_steps": 51936,
|
| 3684 |
+
"num_input_tokens_seen": 0,
|
| 3685 |
+
"num_train_epochs": 6,
|
| 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 sha256:3ed4efe313ce9a9014df973dc998f93f19f39063b7c4ab2e29800b3c9cc3ad65
|
| 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:c96e1f7ac847290dc5c0af8b7f67fa83f48682ac4837b94d964fb7292487c319
|
| 3 |
+
size 631399808
|
checkpoint-34624/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ab82e508b66b340da68a1947d55887f5daf5388a96448f5daa0d0bd9cd59215
|
| 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 sha256:389d8690ffc32442521b458d301d0faaed9c686c2c381d3872a9761751ed4487
|
| 3 |
+
size 1383
|
checkpoint-34624/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a0d9ec155dd90f9308581fccb84e77ba5585236b000ed99c758494a39af4db3
|
| 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 sha256:3ed4efe313ce9a9014df973dc998f93f19f39063b7c4ab2e29800b3c9cc3ad65
|
| 3 |
+
size 5841
|
checkpoint-43280/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-43280/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0906dfb5ded54afefa2a1dc22c985f4f31c7860dc468cbd10aeca1dc8d9fa678
|
| 3 |
+
size 631399808
|
checkpoint-43280/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08c0f70b624b35cd6cfd04184fb7acabe4678a5b8a349be29792d8c6a5fd8b52
|
| 3 |
+
size 1262124766
|
checkpoint-43280/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-43280/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a8e0fa50157e584fe1f5ff542ab9d9b7de33816270ab87b4847781cfad97df5
|
| 3 |
+
size 14645
|
checkpoint-43280/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8d352902a9edf89ffef4e3c21eab7e26e00f9fedd11d120a789a43b5b19c3a6c
|
| 3 |
+
size 1383
|
checkpoint-43280/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed33286f1dc5e5626d8c4286e165ce79b3faae2d793825f3bea7d6d9bbda4c6b
|
| 3 |
+
size 1465
|
checkpoint-43280/trainer_state.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-43280/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ed4efe313ce9a9014df973dc998f93f19f39063b7c4ab2e29800b3c9cc3ad65
|
| 3 |
+
size 5841
|
checkpoint-51936/config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-51936/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c72aa5157e1aa6fb339164c2cb13e63a8f38cef872ba738703ae07a0334d858
|
| 3 |
+
size 631399808
|
checkpoint-51936/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b3f407154c88396f4796aa2ef10e19b86f2118933509d9d0412e7ce04d9814d8
|
| 3 |
+
size 1262124766
|
checkpoint-51936/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-51936/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18b662ff85689f148a71077c04888f496ebae1e1269edbba3f7b71c1221c1143
|
| 3 |
+
size 14645
|
checkpoint-51936/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62e5d521f55261c4f56b2f18e020ec473cc6da0b580cc6ac5f7b91a4819c0e0c
|
| 3 |
+
size 1383
|
checkpoint-51936/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f6ee6a1a858241236c036e34f779093491c9fbb1f799b741a7f5734371e9bfac
|
| 3 |
+
size 1465
|
checkpoint-51936/trainer_state.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-51936/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ed4efe313ce9a9014df973dc998f93f19f39063b7c4ab2e29800b3c9cc3ad65
|
| 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:eb0d513daaa603b2e7ec4d5c4ef1b450631ec4f7b86158a4a5fbf2a84401714e
|
| 3 |
+
size 631399808
|
checkpoint-8656/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:defcf8ffc6460301292b31d84de807521d0a85273302ad356066ea6e57ead1b6
|
| 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 |
+
}
|