Hugging Face's logo

mmomm25
/
swin-tiny-patch4-window7-224-crack-detectorMAIN50epochsFINAL

Image Classification
Transformers
Safetensors
swin
Inference Endpoints
Model card Files Community
mmomm25 commited on
Commit
16a9e0d
•
1 Parent(s): 0780e13

Training in progress, epoch 8

Browse files
Files changed (9) hide show
  1. model.safetensors +1 -1
  2. tmp-checkpoint-5017/config.json +65 -0
  3. tmp-checkpoint-5017/model.safetensors +3 -0
  4. tmp-checkpoint-5017/optimizer.pt +3 -0
  5. tmp-checkpoint-5017/preprocessor_config.json +22 -0
  6. tmp-checkpoint-5017/rng_state.pth +3 -0
  7. tmp-checkpoint-5017/scheduler.pt +3 -0
  8. tmp-checkpoint-5017/trainer_state.json +3207 -0
  9. tmp-checkpoint-5017/training_args.bin +3 -0
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:858c3558ce76d0dd18aa23bbe29776c92288ae86dd01141625b09043b2c29aa8
3
  size 110348984
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ba3b72721211d843e97df5af257deef1f1454fa30ed17c7337fd39719b5af8f
3
  size 110348984
tmp-checkpoint-5017/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/swin-tiny-patch4-window7-224",
3
+ "architectures": [
4
+ "SwinForImageClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "depths": [
8
+ 2,
9
+ 2,
10
+ 6,
11
+ 2
12
+ ],
13
+ "drop_path_rate": 0.1,
14
+ "embed_dim": 96,
15
+ "encoder_stride": 32,
16
+ "hidden_act": "gelu",
17
+ "hidden_dropout_prob": 0.0,
18
+ "hidden_size": 768,
19
+ "id2label": {
20
+ "0": "noncrack",
21
+ "1": "sealedcrack",
22
+ "2": "sealedpatch",
23
+ "3": "thincrack"
24
+ },
25
+ "image_size": 224,
26
+ "initializer_range": 0.02,
27
+ "label2id": {
28
+ "noncrack": 0,
29
+ "sealedcrack": 1,
30
+ "sealedpatch": 2,
31
+ "thincrack": 3
32
+ },
33
+ "layer_norm_eps": 1e-05,
34
+ "mlp_ratio": 4.0,
35
+ "model_type": "swin",
36
+ "num_channels": 3,
37
+ "num_heads": [
38
+ 3,
39
+ 6,
40
+ 12,
41
+ 24
42
+ ],
43
+ "num_layers": 4,
44
+ "out_features": [
45
+ "stage4"
46
+ ],
47
+ "out_indices": [
48
+ 4
49
+ ],
50
+ "patch_size": 4,
51
+ "path_norm": true,
52
+ "problem_type": "single_label_classification",
53
+ "qkv_bias": true,
54
+ "stage_names": [
55
+ "stem",
56
+ "stage1",
57
+ "stage2",
58
+ "stage3",
59
+ "stage4"
60
+ ],
61
+ "torch_dtype": "float32",
62
+ "transformers_version": "4.37.2",
63
+ "use_absolute_embeddings": false,
64
+ "window_size": 7
65
+ }
tmp-checkpoint-5017/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ba3b72721211d843e97df5af257deef1f1454fa30ed17c7337fd39719b5af8f
3
+ size 110348984
tmp-checkpoint-5017/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5e7399621a4552f527313f5b51331b9027ce4032d574f99e52cdb61e1aab582
3
+ size 220362874
tmp-checkpoint-5017/preprocessor_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_rescale": true,
4
+ "do_resize": true,
5
+ "image_mean": [
6
+ 0.485,
7
+ 0.456,
8
+ 0.406
9
+ ],
10
+ "image_processor_type": "ViTImageProcessor",
11
+ "image_std": [
12
+ 0.229,
13
+ 0.224,
14
+ 0.225
15
+ ],
16
+ "resample": 3,
17
+ "rescale_factor": 0.00392156862745098,
18
+ "size": {
19
+ "height": 224,
20
+ "width": 224
21
+ }
22
+ }
tmp-checkpoint-5017/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4703d60060c5a9ed2d1e2d75c220586bcbe9de341b058e72133c592cb6cb203d
3
+ size 14244
tmp-checkpoint-5017/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a119d5b17b8cbc29c81d4df597bcf2670ac012c1bb25efea36d443582cb57b4
3
+ size 1064
tmp-checkpoint-5017/trainer_state.json ADDED
@@ -0,0 +1,3207 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.10741779208183289,
3
+ "best_model_checkpoint": "swin-tiny-patch4-window7-224-crack-detectorMAIN50epochsFINAL\\checkpoint-5017",
4
+ "epoch": 8.999103139013453,
5
+ "eval_steps": 500,
6
+ "global_step": 5017,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "learning_rate": 1.7953321364452426e-07,
14
+ "loss": 1.4257,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 3.590664272890485e-07,
20
+ "loss": 1.4227,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.05,
25
+ "learning_rate": 5.385996409335728e-07,
26
+ "loss": 1.4112,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.07,
31
+ "learning_rate": 7.18132854578097e-07,
32
+ "loss": 1.4052,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.09,
37
+ "learning_rate": 8.976660682226213e-07,
38
+ "loss": 1.3954,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.11,
43
+ "learning_rate": 1.0771992818671456e-06,
44
+ "loss": 1.3978,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.13,
49
+ "learning_rate": 1.2567324955116697e-06,
50
+ "loss": 1.3862,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.14,
55
+ "learning_rate": 1.436265709156194e-06,
56
+ "loss": 1.3651,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.16,
61
+ "learning_rate": 1.6157989228007182e-06,
62
+ "loss": 1.359,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.18,
67
+ "learning_rate": 1.7953321364452425e-06,
68
+ "loss": 1.3476,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.2,
73
+ "learning_rate": 1.9748653500897667e-06,
74
+ "loss": 1.3303,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.22,
79
+ "learning_rate": 2.1543985637342912e-06,
80
+ "loss": 1.3047,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.23,
85
+ "learning_rate": 2.333931777378815e-06,
86
+ "loss": 1.2944,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.25,
91
+ "learning_rate": 2.5134649910233395e-06,
92
+ "loss": 1.284,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.27,
97
+ "learning_rate": 2.6929982046678636e-06,
98
+ "loss": 1.23,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.29,
103
+ "learning_rate": 2.872531418312388e-06,
104
+ "loss": 1.2103,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.3,
109
+ "learning_rate": 3.0520646319569123e-06,
110
+ "loss": 1.187,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.32,
115
+ "learning_rate": 3.2315978456014364e-06,
116
+ "loss": 1.1591,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.34,
121
+ "learning_rate": 3.411131059245961e-06,
122
+ "loss": 1.144,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.36,
127
+ "learning_rate": 3.590664272890485e-06,
128
+ "loss": 1.0904,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.38,
133
+ "learning_rate": 3.770197486535009e-06,
134
+ "loss": 1.0401,
135
+ "step": 210
136
+ },
137
+ {
138
+ "epoch": 0.39,
139
+ "learning_rate": 3.949730700179533e-06,
140
+ "loss": 1.0281,
141
+ "step": 220
142
+ },
143
+ {
144
+ "epoch": 0.41,
145
+ "learning_rate": 4.129263913824058e-06,
146
+ "loss": 0.9577,
147
+ "step": 230
148
+ },
149
+ {
150
+ "epoch": 0.43,
151
+ "learning_rate": 4.3087971274685824e-06,
152
+ "loss": 0.9251,
153
+ "step": 240
154
+ },
155
+ {
156
+ "epoch": 0.45,
157
+ "learning_rate": 4.488330341113106e-06,
158
+ "loss": 0.8964,
159
+ "step": 250
160
+ },
161
+ {
162
+ "epoch": 0.47,
163
+ "learning_rate": 4.66786355475763e-06,
164
+ "loss": 0.8897,
165
+ "step": 260
166
+ },
167
+ {
168
+ "epoch": 0.48,
169
+ "learning_rate": 4.847396768402154e-06,
170
+ "loss": 0.8481,
171
+ "step": 270
172
+ },
173
+ {
174
+ "epoch": 0.5,
175
+ "learning_rate": 5.026929982046679e-06,
176
+ "loss": 0.8048,
177
+ "step": 280
178
+ },
179
+ {
180
+ "epoch": 0.52,
181
+ "learning_rate": 5.206463195691203e-06,
182
+ "loss": 0.8014,
183
+ "step": 290
184
+ },
185
+ {
186
+ "epoch": 0.54,
187
+ "learning_rate": 5.385996409335727e-06,
188
+ "loss": 0.7511,
189
+ "step": 300
190
+ },
191
+ {
192
+ "epoch": 0.56,
193
+ "learning_rate": 5.565529622980251e-06,
194
+ "loss": 0.741,
195
+ "step": 310
196
+ },
197
+ {
198
+ "epoch": 0.57,
199
+ "learning_rate": 5.745062836624776e-06,
200
+ "loss": 0.7327,
201
+ "step": 320
202
+ },
203
+ {
204
+ "epoch": 0.59,
205
+ "learning_rate": 5.9245960502693004e-06,
206
+ "loss": 0.7492,
207
+ "step": 330
208
+ },
209
+ {
210
+ "epoch": 0.61,
211
+ "learning_rate": 6.1041292639138246e-06,
212
+ "loss": 0.7613,
213
+ "step": 340
214
+ },
215
+ {
216
+ "epoch": 0.63,
217
+ "learning_rate": 6.283662477558349e-06,
218
+ "loss": 0.6902,
219
+ "step": 350
220
+ },
221
+ {
222
+ "epoch": 0.65,
223
+ "learning_rate": 6.463195691202873e-06,
224
+ "loss": 0.6687,
225
+ "step": 360
226
+ },
227
+ {
228
+ "epoch": 0.66,
229
+ "learning_rate": 6.642728904847396e-06,
230
+ "loss": 0.6623,
231
+ "step": 370
232
+ },
233
+ {
234
+ "epoch": 0.68,
235
+ "learning_rate": 6.822262118491922e-06,
236
+ "loss": 0.6901,
237
+ "step": 380
238
+ },
239
+ {
240
+ "epoch": 0.7,
241
+ "learning_rate": 7.001795332136446e-06,
242
+ "loss": 0.6638,
243
+ "step": 390
244
+ },
245
+ {
246
+ "epoch": 0.72,
247
+ "learning_rate": 7.18132854578097e-06,
248
+ "loss": 0.6414,
249
+ "step": 400
250
+ },
251
+ {
252
+ "epoch": 0.74,
253
+ "learning_rate": 7.360861759425494e-06,
254
+ "loss": 0.6877,
255
+ "step": 410
256
+ },
257
+ {
258
+ "epoch": 0.75,
259
+ "learning_rate": 7.540394973070018e-06,
260
+ "loss": 0.6249,
261
+ "step": 420
262
+ },
263
+ {
264
+ "epoch": 0.77,
265
+ "learning_rate": 7.719928186714543e-06,
266
+ "loss": 0.6099,
267
+ "step": 430
268
+ },
269
+ {
270
+ "epoch": 0.79,
271
+ "learning_rate": 7.899461400359067e-06,
272
+ "loss": 0.6321,
273
+ "step": 440
274
+ },
275
+ {
276
+ "epoch": 0.81,
277
+ "learning_rate": 8.07899461400359e-06,
278
+ "loss": 0.6431,
279
+ "step": 450
280
+ },
281
+ {
282
+ "epoch": 0.83,
283
+ "learning_rate": 8.258527827648117e-06,
284
+ "loss": 0.6404,
285
+ "step": 460
286
+ },
287
+ {
288
+ "epoch": 0.84,
289
+ "learning_rate": 8.43806104129264e-06,
290
+ "loss": 0.6148,
291
+ "step": 470
292
+ },
293
+ {
294
+ "epoch": 0.86,
295
+ "learning_rate": 8.617594254937165e-06,
296
+ "loss": 0.5844,
297
+ "step": 480
298
+ },
299
+ {
300
+ "epoch": 0.88,
301
+ "learning_rate": 8.797127468581689e-06,
302
+ "loss": 0.6316,
303
+ "step": 490
304
+ },
305
+ {
306
+ "epoch": 0.9,
307
+ "learning_rate": 8.976660682226211e-06,
308
+ "loss": 0.6018,
309
+ "step": 500
310
+ },
311
+ {
312
+ "epoch": 0.91,
313
+ "learning_rate": 9.156193895870736e-06,
314
+ "loss": 0.5843,
315
+ "step": 510
316
+ },
317
+ {
318
+ "epoch": 0.93,
319
+ "learning_rate": 9.33572710951526e-06,
320
+ "loss": 0.5986,
321
+ "step": 520
322
+ },
323
+ {
324
+ "epoch": 0.95,
325
+ "learning_rate": 9.515260323159784e-06,
326
+ "loss": 0.5718,
327
+ "step": 530
328
+ },
329
+ {
330
+ "epoch": 0.97,
331
+ "learning_rate": 9.694793536804308e-06,
332
+ "loss": 0.5163,
333
+ "step": 540
334
+ },
335
+ {
336
+ "epoch": 0.99,
337
+ "learning_rate": 9.874326750448834e-06,
338
+ "loss": 0.5448,
339
+ "step": 550
340
+ },
341
+ {
342
+ "epoch": 1.0,
343
+ "eval_accuracy": {
344
+ "accuracy": 0.8387530836510428
345
+ },
346
+ "eval_f1": {
347
+ "f1": 0.8242528613153339
348
+ },
349
+ "eval_loss": 0.43474748730659485,
350
+ "eval_precision": {
351
+ "precision": 0.8238035334510331
352
+ },
353
+ "eval_recall": {
354
+ "recall": 0.8266051811271123
355
+ },
356
+ "eval_runtime": 171.551,
357
+ "eval_samples_per_second": 103.969,
358
+ "eval_steps_per_second": 3.253,
359
+ "step": 557
360
+ },
361
+ {
362
+ "epoch": 1.0,
363
+ "learning_rate": 1.0053859964093358e-05,
364
+ "loss": 0.58,
365
+ "step": 560
366
+ },
367
+ {
368
+ "epoch": 1.02,
369
+ "learning_rate": 1.0233393177737882e-05,
370
+ "loss": 0.5298,
371
+ "step": 570
372
+ },
373
+ {
374
+ "epoch": 1.04,
375
+ "learning_rate": 1.0412926391382406e-05,
376
+ "loss": 0.5592,
377
+ "step": 580
378
+ },
379
+ {
380
+ "epoch": 1.06,
381
+ "learning_rate": 1.059245960502693e-05,
382
+ "loss": 0.5352,
383
+ "step": 590
384
+ },
385
+ {
386
+ "epoch": 1.08,
387
+ "learning_rate": 1.0771992818671454e-05,
388
+ "loss": 0.5683,
389
+ "step": 600
390
+ },
391
+ {
392
+ "epoch": 1.09,
393
+ "learning_rate": 1.0951526032315979e-05,
394
+ "loss": 0.571,
395
+ "step": 610
396
+ },
397
+ {
398
+ "epoch": 1.11,
399
+ "learning_rate": 1.1131059245960503e-05,
400
+ "loss": 0.5603,
401
+ "step": 620
402
+ },
403
+ {
404
+ "epoch": 1.13,
405
+ "learning_rate": 1.1310592459605028e-05,
406
+ "loss": 0.5157,
407
+ "step": 630
408
+ },
409
+ {
410
+ "epoch": 1.15,
411
+ "learning_rate": 1.1490125673249553e-05,
412
+ "loss": 0.4885,
413
+ "step": 640
414
+ },
415
+ {
416
+ "epoch": 1.17,
417
+ "learning_rate": 1.1669658886894077e-05,
418
+ "loss": 0.5005,
419
+ "step": 650
420
+ },
421
+ {
422
+ "epoch": 1.18,
423
+ "learning_rate": 1.1849192100538601e-05,
424
+ "loss": 0.4951,
425
+ "step": 660
426
+ },
427
+ {
428
+ "epoch": 1.2,
429
+ "learning_rate": 1.2028725314183125e-05,
430
+ "loss": 0.5197,
431
+ "step": 670
432
+ },
433
+ {
434
+ "epoch": 1.22,
435
+ "learning_rate": 1.2208258527827649e-05,
436
+ "loss": 0.4746,
437
+ "step": 680
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "learning_rate": 1.2387791741472173e-05,
442
+ "loss": 0.486,
443
+ "step": 690
444
+ },
445
+ {
446
+ "epoch": 1.26,
447
+ "learning_rate": 1.2567324955116697e-05,
448
+ "loss": 0.4932,
449
+ "step": 700
450
+ },
451
+ {
452
+ "epoch": 1.27,
453
+ "learning_rate": 1.2746858168761221e-05,
454
+ "loss": 0.5222,
455
+ "step": 710
456
+ },
457
+ {
458
+ "epoch": 1.29,
459
+ "learning_rate": 1.2926391382405746e-05,
460
+ "loss": 0.434,
461
+ "step": 720
462
+ },
463
+ {
464
+ "epoch": 1.31,
465
+ "learning_rate": 1.310592459605027e-05,
466
+ "loss": 0.4817,
467
+ "step": 730
468
+ },
469
+ {
470
+ "epoch": 1.33,
471
+ "learning_rate": 1.3285457809694792e-05,
472
+ "loss": 0.487,
473
+ "step": 740
474
+ },
475
+ {
476
+ "epoch": 1.35,
477
+ "learning_rate": 1.3464991023339318e-05,
478
+ "loss": 0.4478,
479
+ "step": 750
480
+ },
481
+ {
482
+ "epoch": 1.36,
483
+ "learning_rate": 1.3644524236983844e-05,
484
+ "loss": 0.4564,
485
+ "step": 760
486
+ },
487
+ {
488
+ "epoch": 1.38,
489
+ "learning_rate": 1.3824057450628366e-05,
490
+ "loss": 0.4455,
491
+ "step": 770
492
+ },
493
+ {
494
+ "epoch": 1.4,
495
+ "learning_rate": 1.4003590664272892e-05,
496
+ "loss": 0.4701,
497
+ "step": 780
498
+ },
499
+ {
500
+ "epoch": 1.42,
501
+ "learning_rate": 1.4183123877917415e-05,
502
+ "loss": 0.4333,
503
+ "step": 790
504
+ },
505
+ {
506
+ "epoch": 1.43,
507
+ "learning_rate": 1.436265709156194e-05,
508
+ "loss": 0.4409,
509
+ "step": 800
510
+ },
511
+ {
512
+ "epoch": 1.45,
513
+ "learning_rate": 1.4542190305206463e-05,
514
+ "loss": 0.4639,
515
+ "step": 810
516
+ },
517
+ {
518
+ "epoch": 1.47,
519
+ "learning_rate": 1.4721723518850989e-05,
520
+ "loss": 0.4552,
521
+ "step": 820
522
+ },
523
+ {
524
+ "epoch": 1.49,
525
+ "learning_rate": 1.4901256732495511e-05,
526
+ "loss": 0.4759,
527
+ "step": 830
528
+ },
529
+ {
530
+ "epoch": 1.51,
531
+ "learning_rate": 1.5080789946140037e-05,
532
+ "loss": 0.4584,
533
+ "step": 840
534
+ },
535
+ {
536
+ "epoch": 1.52,
537
+ "learning_rate": 1.5260323159784563e-05,
538
+ "loss": 0.4502,
539
+ "step": 850
540
+ },
541
+ {
542
+ "epoch": 1.54,
543
+ "learning_rate": 1.5439856373429085e-05,
544
+ "loss": 0.4519,
545
+ "step": 860
546
+ },
547
+ {
548
+ "epoch": 1.56,
549
+ "learning_rate": 1.561938958707361e-05,
550
+ "loss": 0.4695,
551
+ "step": 870
552
+ },
553
+ {
554
+ "epoch": 1.58,
555
+ "learning_rate": 1.5798922800718133e-05,
556
+ "loss": 0.4466,
557
+ "step": 880
558
+ },
559
+ {
560
+ "epoch": 1.6,
561
+ "learning_rate": 1.597845601436266e-05,
562
+ "loss": 0.4531,
563
+ "step": 890
564
+ },
565
+ {
566
+ "epoch": 1.61,
567
+ "learning_rate": 1.615798922800718e-05,
568
+ "loss": 0.4674,
569
+ "step": 900
570
+ },
571
+ {
572
+ "epoch": 1.63,
573
+ "learning_rate": 1.6337522441651707e-05,
574
+ "loss": 0.4262,
575
+ "step": 910
576
+ },
577
+ {
578
+ "epoch": 1.65,
579
+ "learning_rate": 1.6517055655296233e-05,
580
+ "loss": 0.3945,
581
+ "step": 920
582
+ },
583
+ {
584
+ "epoch": 1.67,
585
+ "learning_rate": 1.6696588868940756e-05,
586
+ "loss": 0.4318,
587
+ "step": 930
588
+ },
589
+ {
590
+ "epoch": 1.69,
591
+ "learning_rate": 1.687612208258528e-05,
592
+ "loss": 0.439,
593
+ "step": 940
594
+ },
595
+ {
596
+ "epoch": 1.7,
597
+ "learning_rate": 1.7055655296229804e-05,
598
+ "loss": 0.4036,
599
+ "step": 950
600
+ },
601
+ {
602
+ "epoch": 1.72,
603
+ "learning_rate": 1.723518850987433e-05,
604
+ "loss": 0.422,
605
+ "step": 960
606
+ },
607
+ {
608
+ "epoch": 1.74,
609
+ "learning_rate": 1.7414721723518852e-05,
610
+ "loss": 0.4315,
611
+ "step": 970
612
+ },
613
+ {
614
+ "epoch": 1.76,
615
+ "learning_rate": 1.7594254937163378e-05,
616
+ "loss": 0.4516,
617
+ "step": 980
618
+ },
619
+ {
620
+ "epoch": 1.78,
621
+ "learning_rate": 1.77737881508079e-05,
622
+ "loss": 0.4092,
623
+ "step": 990
624
+ },
625
+ {
626
+ "epoch": 1.79,
627
+ "learning_rate": 1.7953321364452423e-05,
628
+ "loss": 0.4367,
629
+ "step": 1000
630
+ },
631
+ {
632
+ "epoch": 1.81,
633
+ "learning_rate": 1.813285457809695e-05,
634
+ "loss": 0.4321,
635
+ "step": 1010
636
+ },
637
+ {
638
+ "epoch": 1.83,
639
+ "learning_rate": 1.831238779174147e-05,
640
+ "loss": 0.3859,
641
+ "step": 1020
642
+ },
643
+ {
644
+ "epoch": 1.85,
645
+ "learning_rate": 1.8491921005385997e-05,
646
+ "loss": 0.415,
647
+ "step": 1030
648
+ },
649
+ {
650
+ "epoch": 1.87,
651
+ "learning_rate": 1.867145421903052e-05,
652
+ "loss": 0.4388,
653
+ "step": 1040
654
+ },
655
+ {
656
+ "epoch": 1.88,
657
+ "learning_rate": 1.8850987432675045e-05,
658
+ "loss": 0.3968,
659
+ "step": 1050
660
+ },
661
+ {
662
+ "epoch": 1.9,
663
+ "learning_rate": 1.9030520646319568e-05,
664
+ "loss": 0.4172,
665
+ "step": 1060
666
+ },
667
+ {
668
+ "epoch": 1.92,
669
+ "learning_rate": 1.9210053859964093e-05,
670
+ "loss": 0.4593,
671
+ "step": 1070
672
+ },
673
+ {
674
+ "epoch": 1.94,
675
+ "learning_rate": 1.9389587073608616e-05,
676
+ "loss": 0.3924,
677
+ "step": 1080
678
+ },
679
+ {
680
+ "epoch": 1.96,
681
+ "learning_rate": 1.9569120287253142e-05,
682
+ "loss": 0.4067,
683
+ "step": 1090
684
+ },
685
+ {
686
+ "epoch": 1.97,
687
+ "learning_rate": 1.9748653500897668e-05,
688
+ "loss": 0.3906,
689
+ "step": 1100
690
+ },
691
+ {
692
+ "epoch": 1.99,
693
+ "learning_rate": 1.992818671454219e-05,
694
+ "loss": 0.3915,
695
+ "step": 1110
696
+ },
697
+ {
698
+ "epoch": 2.0,
699
+ "eval_accuracy": {
700
+ "accuracy": 0.8952679973088137
701
+ },
702
+ "eval_f1": {
703
+ "f1": 0.8895212539619177
704
+ },
705
+ "eval_loss": 0.2781514525413513,
706
+ "eval_precision": {
707
+ "precision": 0.8895590270096669
708
+ },
709
+ "eval_recall": {
710
+ "recall": 0.892156947982043
711
+ },
712
+ "eval_runtime": 99.0662,
713
+ "eval_samples_per_second": 180.041,
714
+ "eval_steps_per_second": 5.633,
715
+ "step": 1115
716
+ },
717
+ {
718
+ "epoch": 2.01,
719
+ "learning_rate": 2.0107719928186716e-05,
720
+ "loss": 0.4099,
721
+ "step": 1120
722
+ },
723
+ {
724
+ "epoch": 2.03,
725
+ "learning_rate": 2.0287253141831238e-05,
726
+ "loss": 0.4165,
727
+ "step": 1130
728
+ },
729
+ {
730
+ "epoch": 2.04,
731
+ "learning_rate": 2.0466786355475764e-05,
732
+ "loss": 0.3958,
733
+ "step": 1140
734
+ },
735
+ {
736
+ "epoch": 2.06,
737
+ "learning_rate": 2.0646319569120286e-05,
738
+ "loss": 0.4353,
739
+ "step": 1150
740
+ },
741
+ {
742
+ "epoch": 2.08,
743
+ "learning_rate": 2.0825852782764812e-05,
744
+ "loss": 0.3832,
745
+ "step": 1160
746
+ },
747
+ {
748
+ "epoch": 2.1,
749
+ "learning_rate": 2.1005385996409338e-05,
750
+ "loss": 0.3967,
751
+ "step": 1170
752
+ },
753
+ {
754
+ "epoch": 2.12,
755
+ "learning_rate": 2.118491921005386e-05,
756
+ "loss": 0.3614,
757
+ "step": 1180
758
+ },
759
+ {
760
+ "epoch": 2.13,
761
+ "learning_rate": 2.1364452423698386e-05,
762
+ "loss": 0.3665,
763
+ "step": 1190
764
+ },
765
+ {
766
+ "epoch": 2.15,
767
+ "learning_rate": 2.154398563734291e-05,
768
+ "loss": 0.3871,
769
+ "step": 1200
770
+ },
771
+ {
772
+ "epoch": 2.17,
773
+ "learning_rate": 2.1723518850987435e-05,
774
+ "loss": 0.3748,
775
+ "step": 1210
776
+ },
777
+ {
778
+ "epoch": 2.19,
779
+ "learning_rate": 2.1903052064631957e-05,
780
+ "loss": 0.396,
781
+ "step": 1220
782
+ },
783
+ {
784
+ "epoch": 2.21,
785
+ "learning_rate": 2.2082585278276483e-05,
786
+ "loss": 0.374,
787
+ "step": 1230
788
+ },
789
+ {
790
+ "epoch": 2.22,
791
+ "learning_rate": 2.2262118491921005e-05,
792
+ "loss": 0.3735,
793
+ "step": 1240
794
+ },
795
+ {
796
+ "epoch": 2.24,
797
+ "learning_rate": 2.244165170556553e-05,
798
+ "loss": 0.3894,
799
+ "step": 1250
800
+ },
801
+ {
802
+ "epoch": 2.26,
803
+ "learning_rate": 2.2621184919210057e-05,
804
+ "loss": 0.3443,
805
+ "step": 1260
806
+ },
807
+ {
808
+ "epoch": 2.28,
809
+ "learning_rate": 2.280071813285458e-05,
810
+ "loss": 0.3402,
811
+ "step": 1270
812
+ },
813
+ {
814
+ "epoch": 2.3,
815
+ "learning_rate": 2.2980251346499105e-05,
816
+ "loss": 0.4164,
817
+ "step": 1280
818
+ },
819
+ {
820
+ "epoch": 2.31,
821
+ "learning_rate": 2.3159784560143628e-05,
822
+ "loss": 0.3879,
823
+ "step": 1290
824
+ },
825
+ {
826
+ "epoch": 2.33,
827
+ "learning_rate": 2.3339317773788153e-05,
828
+ "loss": 0.3719,
829
+ "step": 1300
830
+ },
831
+ {
832
+ "epoch": 2.35,
833
+ "learning_rate": 2.3518850987432676e-05,
834
+ "loss": 0.3477,
835
+ "step": 1310
836
+ },
837
+ {
838
+ "epoch": 2.37,
839
+ "learning_rate": 2.3698384201077202e-05,
840
+ "loss": 0.3744,
841
+ "step": 1320
842
+ },
843
+ {
844
+ "epoch": 2.39,
845
+ "learning_rate": 2.3877917414721724e-05,
846
+ "loss": 0.3833,
847
+ "step": 1330
848
+ },
849
+ {
850
+ "epoch": 2.4,
851
+ "learning_rate": 2.405745062836625e-05,
852
+ "loss": 0.398,
853
+ "step": 1340
854
+ },
855
+ {
856
+ "epoch": 2.42,
857
+ "learning_rate": 2.4236983842010776e-05,
858
+ "loss": 0.3562,
859
+ "step": 1350
860
+ },
861
+ {
862
+ "epoch": 2.44,
863
+ "learning_rate": 2.4416517055655298e-05,
864
+ "loss": 0.3614,
865
+ "step": 1360
866
+ },
867
+ {
868
+ "epoch": 2.46,
869
+ "learning_rate": 2.4596050269299824e-05,
870
+ "loss": 0.315,
871
+ "step": 1370
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "learning_rate": 2.4775583482944346e-05,
876
+ "loss": 0.3441,
877
+ "step": 1380
878
+ },
879
+ {
880
+ "epoch": 2.49,
881
+ "learning_rate": 2.495511669658887e-05,
882
+ "loss": 0.3589,
883
+ "step": 1390
884
+ },
885
+ {
886
+ "epoch": 2.51,
887
+ "learning_rate": 2.5134649910233395e-05,
888
+ "loss": 0.3418,
889
+ "step": 1400
890
+ },
891
+ {
892
+ "epoch": 2.53,
893
+ "learning_rate": 2.5314183123877917e-05,
894
+ "loss": 0.3696,
895
+ "step": 1410
896
+ },
897
+ {
898
+ "epoch": 2.55,
899
+ "learning_rate": 2.5493716337522443e-05,
900
+ "loss": 0.3606,
901
+ "step": 1420
902
+ },
903
+ {
904
+ "epoch": 2.57,
905
+ "learning_rate": 2.5673249551166965e-05,
906
+ "loss": 0.3389,
907
+ "step": 1430
908
+ },
909
+ {
910
+ "epoch": 2.58,
911
+ "learning_rate": 2.585278276481149e-05,
912
+ "loss": 0.3375,
913
+ "step": 1440
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "learning_rate": 2.6032315978456017e-05,
918
+ "loss": 0.3193,
919
+ "step": 1450
920
+ },
921
+ {
922
+ "epoch": 2.62,
923
+ "learning_rate": 2.621184919210054e-05,
924
+ "loss": 0.3683,
925
+ "step": 1460
926
+ },
927
+ {
928
+ "epoch": 2.64,
929
+ "learning_rate": 2.6391382405745062e-05,
930
+ "loss": 0.3203,
931
+ "step": 1470
932
+ },
933
+ {
934
+ "epoch": 2.65,
935
+ "learning_rate": 2.6570915619389584e-05,
936
+ "loss": 0.3792,
937
+ "step": 1480
938
+ },
939
+ {
940
+ "epoch": 2.67,
941
+ "learning_rate": 2.6750448833034114e-05,
942
+ "loss": 0.3343,
943
+ "step": 1490
944
+ },
945
+ {
946
+ "epoch": 2.69,
947
+ "learning_rate": 2.6929982046678636e-05,
948
+ "loss": 0.3671,
949
+ "step": 1500
950
+ },
951
+ {
952
+ "epoch": 2.71,
953
+ "learning_rate": 2.710951526032316e-05,
954
+ "loss": 0.4002,
955
+ "step": 1510
956
+ },
957
+ {
958
+ "epoch": 2.73,
959
+ "learning_rate": 2.7289048473967688e-05,
960
+ "loss": 0.3246,
961
+ "step": 1520
962
+ },
963
+ {
964
+ "epoch": 2.74,
965
+ "learning_rate": 2.746858168761221e-05,
966
+ "loss": 0.3731,
967
+ "step": 1530
968
+ },
969
+ {
970
+ "epoch": 2.76,
971
+ "learning_rate": 2.7648114901256732e-05,
972
+ "loss": 0.3652,
973
+ "step": 1540
974
+ },
975
+ {
976
+ "epoch": 2.78,
977
+ "learning_rate": 2.7827648114901255e-05,
978
+ "loss": 0.3767,
979
+ "step": 1550
980
+ },
981
+ {
982
+ "epoch": 2.8,
983
+ "learning_rate": 2.8007181328545784e-05,
984
+ "loss": 0.3135,
985
+ "step": 1560
986
+ },
987
+ {
988
+ "epoch": 2.82,
989
+ "learning_rate": 2.8186714542190307e-05,
990
+ "loss": 0.3303,
991
+ "step": 1570
992
+ },
993
+ {
994
+ "epoch": 2.83,
995
+ "learning_rate": 2.836624775583483e-05,
996
+ "loss": 0.3485,
997
+ "step": 1580
998
+ },
999
+ {
1000
+ "epoch": 2.85,
1001
+ "learning_rate": 2.8545780969479358e-05,
1002
+ "loss": 0.3528,
1003
+ "step": 1590
1004
+ },
1005
+ {
1006
+ "epoch": 2.87,
1007
+ "learning_rate": 2.872531418312388e-05,
1008
+ "loss": 0.3205,
1009
+ "step": 1600
1010
+ },
1011
+ {
1012
+ "epoch": 2.89,
1013
+ "learning_rate": 2.8904847396768403e-05,
1014
+ "loss": 0.3047,
1015
+ "step": 1610
1016
+ },
1017
+ {
1018
+ "epoch": 2.91,
1019
+ "learning_rate": 2.9084380610412926e-05,
1020
+ "loss": 0.3656,
1021
+ "step": 1620
1022
+ },
1023
+ {
1024
+ "epoch": 2.92,
1025
+ "learning_rate": 2.9263913824057455e-05,
1026
+ "loss": 0.3317,
1027
+ "step": 1630
1028
+ },
1029
+ {
1030
+ "epoch": 2.94,
1031
+ "learning_rate": 2.9443447037701977e-05,
1032
+ "loss": 0.3318,
1033
+ "step": 1640
1034
+ },
1035
+ {
1036
+ "epoch": 2.96,
1037
+ "learning_rate": 2.96229802513465e-05,
1038
+ "loss": 0.365,
1039
+ "step": 1650
1040
+ },
1041
+ {
1042
+ "epoch": 2.98,
1043
+ "learning_rate": 2.9802513464991022e-05,
1044
+ "loss": 0.3273,
1045
+ "step": 1660
1046
+ },
1047
+ {
1048
+ "epoch": 3.0,
1049
+ "learning_rate": 2.998204667863555e-05,
1050
+ "loss": 0.3246,
1051
+ "step": 1670
1052
+ },
1053
+ {
1054
+ "epoch": 3.0,
1055
+ "eval_accuracy": {
1056
+ "accuracy": 0.9295245570755775
1057
+ },
1058
+ "eval_f1": {
1059
+ "f1": 0.9243811536773696
1060
+ },
1061
+ "eval_loss": 0.19212597608566284,
1062
+ "eval_precision": {
1063
+ "precision": 0.9273415557649412
1064
+ },
1065
+ "eval_recall": {
1066
+ "recall": 0.9219909309089409
1067
+ },
1068
+ "eval_runtime": 99.0388,
1069
+ "eval_samples_per_second": 180.091,
1070
+ "eval_steps_per_second": 5.634,
1071
+ "step": 1672
1072
+ },
1073
+ {
1074
+ "epoch": 3.01,
1075
+ "learning_rate": 3.0161579892280074e-05,
1076
+ "loss": 0.3135,
1077
+ "step": 1680
1078
+ },
1079
+ {
1080
+ "epoch": 3.03,
1081
+ "learning_rate": 3.0341113105924596e-05,
1082
+ "loss": 0.3145,
1083
+ "step": 1690
1084
+ },
1085
+ {
1086
+ "epoch": 3.05,
1087
+ "learning_rate": 3.0520646319569125e-05,
1088
+ "loss": 0.3395,
1089
+ "step": 1700
1090
+ },
1091
+ {
1092
+ "epoch": 3.07,
1093
+ "learning_rate": 3.070017953321365e-05,
1094
+ "loss": 0.2927,
1095
+ "step": 1710
1096
+ },
1097
+ {
1098
+ "epoch": 3.09,
1099
+ "learning_rate": 3.087971274685817e-05,
1100
+ "loss": 0.3363,
1101
+ "step": 1720
1102
+ },
1103
+ {
1104
+ "epoch": 3.1,
1105
+ "learning_rate": 3.105924596050269e-05,
1106
+ "loss": 0.3392,
1107
+ "step": 1730
1108
+ },
1109
+ {
1110
+ "epoch": 3.12,
1111
+ "learning_rate": 3.123877917414722e-05,
1112
+ "loss": 0.3279,
1113
+ "step": 1740
1114
+ },
1115
+ {
1116
+ "epoch": 3.14,
1117
+ "learning_rate": 3.1418312387791744e-05,
1118
+ "loss": 0.3194,
1119
+ "step": 1750
1120
+ },
1121
+ {
1122
+ "epoch": 3.16,
1123
+ "learning_rate": 3.159784560143627e-05,
1124
+ "loss": 0.3206,
1125
+ "step": 1760
1126
+ },
1127
+ {
1128
+ "epoch": 3.17,
1129
+ "learning_rate": 3.1777378815080796e-05,
1130
+ "loss": 0.3246,
1131
+ "step": 1770
1132
+ },
1133
+ {
1134
+ "epoch": 3.19,
1135
+ "learning_rate": 3.195691202872532e-05,
1136
+ "loss": 0.3155,
1137
+ "step": 1780
1138
+ },
1139
+ {
1140
+ "epoch": 3.21,
1141
+ "learning_rate": 3.213644524236984e-05,
1142
+ "loss": 0.301,
1143
+ "step": 1790
1144
+ },
1145
+ {
1146
+ "epoch": 3.23,
1147
+ "learning_rate": 3.231597845601436e-05,
1148
+ "loss": 0.3287,
1149
+ "step": 1800
1150
+ },
1151
+ {
1152
+ "epoch": 3.25,
1153
+ "learning_rate": 3.249551166965889e-05,
1154
+ "loss": 0.3282,
1155
+ "step": 1810
1156
+ },
1157
+ {
1158
+ "epoch": 3.26,
1159
+ "learning_rate": 3.2675044883303415e-05,
1160
+ "loss": 0.2872,
1161
+ "step": 1820
1162
+ },
1163
+ {
1164
+ "epoch": 3.28,
1165
+ "learning_rate": 3.285457809694794e-05,
1166
+ "loss": 0.3003,
1167
+ "step": 1830
1168
+ },
1169
+ {
1170
+ "epoch": 3.3,
1171
+ "learning_rate": 3.3034111310592466e-05,
1172
+ "loss": 0.3009,
1173
+ "step": 1840
1174
+ },
1175
+ {
1176
+ "epoch": 3.32,
1177
+ "learning_rate": 3.321364452423699e-05,
1178
+ "loss": 0.325,
1179
+ "step": 1850
1180
+ },
1181
+ {
1182
+ "epoch": 3.34,
1183
+ "learning_rate": 3.339317773788151e-05,
1184
+ "loss": 0.3308,
1185
+ "step": 1860
1186
+ },
1187
+ {
1188
+ "epoch": 3.35,
1189
+ "learning_rate": 3.3572710951526034e-05,
1190
+ "loss": 0.3175,
1191
+ "step": 1870
1192
+ },
1193
+ {
1194
+ "epoch": 3.37,
1195
+ "learning_rate": 3.375224416517056e-05,
1196
+ "loss": 0.291,
1197
+ "step": 1880
1198
+ },
1199
+ {
1200
+ "epoch": 3.39,
1201
+ "learning_rate": 3.3931777378815085e-05,
1202
+ "loss": 0.2936,
1203
+ "step": 1890
1204
+ },
1205
+ {
1206
+ "epoch": 3.41,
1207
+ "learning_rate": 3.411131059245961e-05,
1208
+ "loss": 0.3304,
1209
+ "step": 1900
1210
+ },
1211
+ {
1212
+ "epoch": 3.43,
1213
+ "learning_rate": 3.429084380610413e-05,
1214
+ "loss": 0.331,
1215
+ "step": 1910
1216
+ },
1217
+ {
1218
+ "epoch": 3.44,
1219
+ "learning_rate": 3.447037701974866e-05,
1220
+ "loss": 0.3158,
1221
+ "step": 1920
1222
+ },
1223
+ {
1224
+ "epoch": 3.46,
1225
+ "learning_rate": 3.464991023339318e-05,
1226
+ "loss": 0.3054,
1227
+ "step": 1930
1228
+ },
1229
+ {
1230
+ "epoch": 3.48,
1231
+ "learning_rate": 3.4829443447037704e-05,
1232
+ "loss": 0.3063,
1233
+ "step": 1940
1234
+ },
1235
+ {
1236
+ "epoch": 3.5,
1237
+ "learning_rate": 3.500897666068223e-05,
1238
+ "loss": 0.279,
1239
+ "step": 1950
1240
+ },
1241
+ {
1242
+ "epoch": 3.52,
1243
+ "learning_rate": 3.5188509874326756e-05,
1244
+ "loss": 0.2986,
1245
+ "step": 1960
1246
+ },
1247
+ {
1248
+ "epoch": 3.53,
1249
+ "learning_rate": 3.536804308797128e-05,
1250
+ "loss": 0.3184,
1251
+ "step": 1970
1252
+ },
1253
+ {
1254
+ "epoch": 3.55,
1255
+ "learning_rate": 3.55475763016158e-05,
1256
+ "loss": 0.3623,
1257
+ "step": 1980
1258
+ },
1259
+ {
1260
+ "epoch": 3.57,
1261
+ "learning_rate": 3.572710951526032e-05,
1262
+ "loss": 0.3178,
1263
+ "step": 1990
1264
+ },
1265
+ {
1266
+ "epoch": 3.59,
1267
+ "learning_rate": 3.5906642728904846e-05,
1268
+ "loss": 0.2553,
1269
+ "step": 2000
1270
+ },
1271
+ {
1272
+ "epoch": 3.61,
1273
+ "learning_rate": 3.6086175942549375e-05,
1274
+ "loss": 0.3116,
1275
+ "step": 2010
1276
+ },
1277
+ {
1278
+ "epoch": 3.62,
1279
+ "learning_rate": 3.62657091561939e-05,
1280
+ "loss": 0.3278,
1281
+ "step": 2020
1282
+ },
1283
+ {
1284
+ "epoch": 3.64,
1285
+ "learning_rate": 3.644524236983842e-05,
1286
+ "loss": 0.311,
1287
+ "step": 2030
1288
+ },
1289
+ {
1290
+ "epoch": 3.66,
1291
+ "learning_rate": 3.662477558348294e-05,
1292
+ "loss": 0.3116,
1293
+ "step": 2040
1294
+ },
1295
+ {
1296
+ "epoch": 3.68,
1297
+ "learning_rate": 3.6804308797127465e-05,
1298
+ "loss": 0.3087,
1299
+ "step": 2050
1300
+ },
1301
+ {
1302
+ "epoch": 3.7,
1303
+ "learning_rate": 3.6983842010771994e-05,
1304
+ "loss": 0.3091,
1305
+ "step": 2060
1306
+ },
1307
+ {
1308
+ "epoch": 3.71,
1309
+ "learning_rate": 3.7163375224416516e-05,
1310
+ "loss": 0.2855,
1311
+ "step": 2070
1312
+ },
1313
+ {
1314
+ "epoch": 3.73,
1315
+ "learning_rate": 3.734290843806104e-05,
1316
+ "loss": 0.313,
1317
+ "step": 2080
1318
+ },
1319
+ {
1320
+ "epoch": 3.75,
1321
+ "learning_rate": 3.752244165170557e-05,
1322
+ "loss": 0.2992,
1323
+ "step": 2090
1324
+ },
1325
+ {
1326
+ "epoch": 3.77,
1327
+ "learning_rate": 3.770197486535009e-05,
1328
+ "loss": 0.2946,
1329
+ "step": 2100
1330
+ },
1331
+ {
1332
+ "epoch": 3.78,
1333
+ "learning_rate": 3.788150807899461e-05,
1334
+ "loss": 0.3034,
1335
+ "step": 2110
1336
+ },
1337
+ {
1338
+ "epoch": 3.8,
1339
+ "learning_rate": 3.8061041292639135e-05,
1340
+ "loss": 0.2868,
1341
+ "step": 2120
1342
+ },
1343
+ {
1344
+ "epoch": 3.82,
1345
+ "learning_rate": 3.8240574506283664e-05,
1346
+ "loss": 0.2818,
1347
+ "step": 2130
1348
+ },
1349
+ {
1350
+ "epoch": 3.84,
1351
+ "learning_rate": 3.842010771992819e-05,
1352
+ "loss": 0.2943,
1353
+ "step": 2140
1354
+ },
1355
+ {
1356
+ "epoch": 3.86,
1357
+ "learning_rate": 3.859964093357271e-05,
1358
+ "loss": 0.3166,
1359
+ "step": 2150
1360
+ },
1361
+ {
1362
+ "epoch": 3.87,
1363
+ "learning_rate": 3.877917414721723e-05,
1364
+ "loss": 0.2947,
1365
+ "step": 2160
1366
+ },
1367
+ {
1368
+ "epoch": 3.89,
1369
+ "learning_rate": 3.895870736086176e-05,
1370
+ "loss": 0.2764,
1371
+ "step": 2170
1372
+ },
1373
+ {
1374
+ "epoch": 3.91,
1375
+ "learning_rate": 3.9138240574506283e-05,
1376
+ "loss": 0.278,
1377
+ "step": 2180
1378
+ },
1379
+ {
1380
+ "epoch": 3.93,
1381
+ "learning_rate": 3.9317773788150806e-05,
1382
+ "loss": 0.3091,
1383
+ "step": 2190
1384
+ },
1385
+ {
1386
+ "epoch": 3.95,
1387
+ "learning_rate": 3.9497307001795335e-05,
1388
+ "loss": 0.3336,
1389
+ "step": 2200
1390
+ },
1391
+ {
1392
+ "epoch": 3.96,
1393
+ "learning_rate": 3.967684021543986e-05,
1394
+ "loss": 0.3256,
1395
+ "step": 2210
1396
+ },
1397
+ {
1398
+ "epoch": 3.98,
1399
+ "learning_rate": 3.985637342908438e-05,
1400
+ "loss": 0.2664,
1401
+ "step": 2220
1402
+ },
1403
+ {
1404
+ "epoch": 4.0,
1405
+ "learning_rate": 4.00359066427289e-05,
1406
+ "loss": 0.2819,
1407
+ "step": 2230
1408
+ },
1409
+ {
1410
+ "epoch": 4.0,
1411
+ "eval_accuracy": {
1412
+ "accuracy": 0.9389437093518727
1413
+ },
1414
+ "eval_f1": {
1415
+ "f1": 0.934636026587255
1416
+ },
1417
+ "eval_loss": 0.16309776902198792,
1418
+ "eval_precision": {
1419
+ "precision": 0.9384931858190604
1420
+ },
1421
+ "eval_recall": {
1422
+ "recall": 0.9317771824858079
1423
+ },
1424
+ "eval_runtime": 98.5849,
1425
+ "eval_samples_per_second": 180.92,
1426
+ "eval_steps_per_second": 5.66,
1427
+ "step": 2230
1428
+ },
1429
+ {
1430
+ "epoch": 4.02,
1431
+ "learning_rate": 4.021543985637343e-05,
1432
+ "loss": 0.2725,
1433
+ "step": 2240
1434
+ },
1435
+ {
1436
+ "epoch": 4.04,
1437
+ "learning_rate": 4.0394973070017954e-05,
1438
+ "loss": 0.3239,
1439
+ "step": 2250
1440
+ },
1441
+ {
1442
+ "epoch": 4.05,
1443
+ "learning_rate": 4.0574506283662476e-05,
1444
+ "loss": 0.2761,
1445
+ "step": 2260
1446
+ },
1447
+ {
1448
+ "epoch": 4.07,
1449
+ "learning_rate": 4.0754039497307006e-05,
1450
+ "loss": 0.275,
1451
+ "step": 2270
1452
+ },
1453
+ {
1454
+ "epoch": 4.09,
1455
+ "learning_rate": 4.093357271095153e-05,
1456
+ "loss": 0.2789,
1457
+ "step": 2280
1458
+ },
1459
+ {
1460
+ "epoch": 4.11,
1461
+ "learning_rate": 4.111310592459605e-05,
1462
+ "loss": 0.2702,
1463
+ "step": 2290
1464
+ },
1465
+ {
1466
+ "epoch": 4.13,
1467
+ "learning_rate": 4.129263913824057e-05,
1468
+ "loss": 0.3244,
1469
+ "step": 2300
1470
+ },
1471
+ {
1472
+ "epoch": 4.14,
1473
+ "learning_rate": 4.14721723518851e-05,
1474
+ "loss": 0.3254,
1475
+ "step": 2310
1476
+ },
1477
+ {
1478
+ "epoch": 4.16,
1479
+ "learning_rate": 4.1651705565529625e-05,
1480
+ "loss": 0.2969,
1481
+ "step": 2320
1482
+ },
1483
+ {
1484
+ "epoch": 4.18,
1485
+ "learning_rate": 4.183123877917415e-05,
1486
+ "loss": 0.2545,
1487
+ "step": 2330
1488
+ },
1489
+ {
1490
+ "epoch": 4.2,
1491
+ "learning_rate": 4.2010771992818676e-05,
1492
+ "loss": 0.2307,
1493
+ "step": 2340
1494
+ },
1495
+ {
1496
+ "epoch": 4.22,
1497
+ "learning_rate": 4.21903052064632e-05,
1498
+ "loss": 0.3132,
1499
+ "step": 2350
1500
+ },
1501
+ {
1502
+ "epoch": 4.23,
1503
+ "learning_rate": 4.236983842010772e-05,
1504
+ "loss": 0.3017,
1505
+ "step": 2360
1506
+ },
1507
+ {
1508
+ "epoch": 4.25,
1509
+ "learning_rate": 4.2549371633752244e-05,
1510
+ "loss": 0.2885,
1511
+ "step": 2370
1512
+ },
1513
+ {
1514
+ "epoch": 4.27,
1515
+ "learning_rate": 4.272890484739677e-05,
1516
+ "loss": 0.3085,
1517
+ "step": 2380
1518
+ },
1519
+ {
1520
+ "epoch": 4.29,
1521
+ "learning_rate": 4.2908438061041295e-05,
1522
+ "loss": 0.3079,
1523
+ "step": 2390
1524
+ },
1525
+ {
1526
+ "epoch": 4.3,
1527
+ "learning_rate": 4.308797127468582e-05,
1528
+ "loss": 0.2749,
1529
+ "step": 2400
1530
+ },
1531
+ {
1532
+ "epoch": 4.32,
1533
+ "learning_rate": 4.326750448833034e-05,
1534
+ "loss": 0.2863,
1535
+ "step": 2410
1536
+ },
1537
+ {
1538
+ "epoch": 4.34,
1539
+ "learning_rate": 4.344703770197487e-05,
1540
+ "loss": 0.2523,
1541
+ "step": 2420
1542
+ },
1543
+ {
1544
+ "epoch": 4.36,
1545
+ "learning_rate": 4.362657091561939e-05,
1546
+ "loss": 0.2441,
1547
+ "step": 2430
1548
+ },
1549
+ {
1550
+ "epoch": 4.38,
1551
+ "learning_rate": 4.3806104129263914e-05,
1552
+ "loss": 0.3174,
1553
+ "step": 2440
1554
+ },
1555
+ {
1556
+ "epoch": 4.39,
1557
+ "learning_rate": 4.398563734290844e-05,
1558
+ "loss": 0.2938,
1559
+ "step": 2450
1560
+ },
1561
+ {
1562
+ "epoch": 4.41,
1563
+ "learning_rate": 4.4165170556552966e-05,
1564
+ "loss": 0.2632,
1565
+ "step": 2460
1566
+ },
1567
+ {
1568
+ "epoch": 4.43,
1569
+ "learning_rate": 4.434470377019749e-05,
1570
+ "loss": 0.2924,
1571
+ "step": 2470
1572
+ },
1573
+ {
1574
+ "epoch": 4.45,
1575
+ "learning_rate": 4.452423698384201e-05,
1576
+ "loss": 0.3018,
1577
+ "step": 2480
1578
+ },
1579
+ {
1580
+ "epoch": 4.47,
1581
+ "learning_rate": 4.470377019748654e-05,
1582
+ "loss": 0.2488,
1583
+ "step": 2490
1584
+ },
1585
+ {
1586
+ "epoch": 4.48,
1587
+ "learning_rate": 4.488330341113106e-05,
1588
+ "loss": 0.2659,
1589
+ "step": 2500
1590
+ },
1591
+ {
1592
+ "epoch": 4.5,
1593
+ "learning_rate": 4.5062836624775585e-05,
1594
+ "loss": 0.2689,
1595
+ "step": 2510
1596
+ },
1597
+ {
1598
+ "epoch": 4.52,
1599
+ "learning_rate": 4.5242369838420114e-05,
1600
+ "loss": 0.2702,
1601
+ "step": 2520
1602
+ },
1603
+ {
1604
+ "epoch": 4.54,
1605
+ "learning_rate": 4.5421903052064636e-05,
1606
+ "loss": 0.2695,
1607
+ "step": 2530
1608
+ },
1609
+ {
1610
+ "epoch": 4.56,
1611
+ "learning_rate": 4.560143626570916e-05,
1612
+ "loss": 0.2629,
1613
+ "step": 2540
1614
+ },
1615
+ {
1616
+ "epoch": 4.57,
1617
+ "learning_rate": 4.578096947935368e-05,
1618
+ "loss": 0.2931,
1619
+ "step": 2550
1620
+ },
1621
+ {
1622
+ "epoch": 4.59,
1623
+ "learning_rate": 4.596050269299821e-05,
1624
+ "loss": 0.2865,
1625
+ "step": 2560
1626
+ },
1627
+ {
1628
+ "epoch": 4.61,
1629
+ "learning_rate": 4.614003590664273e-05,
1630
+ "loss": 0.2751,
1631
+ "step": 2570
1632
+ },
1633
+ {
1634
+ "epoch": 4.63,
1635
+ "learning_rate": 4.6319569120287255e-05,
1636
+ "loss": 0.2565,
1637
+ "step": 2580
1638
+ },
1639
+ {
1640
+ "epoch": 4.65,
1641
+ "learning_rate": 4.6499102333931784e-05,
1642
+ "loss": 0.2589,
1643
+ "step": 2590
1644
+ },
1645
+ {
1646
+ "epoch": 4.66,
1647
+ "learning_rate": 4.667863554757631e-05,
1648
+ "loss": 0.2501,
1649
+ "step": 2600
1650
+ },
1651
+ {
1652
+ "epoch": 4.68,
1653
+ "learning_rate": 4.685816876122083e-05,
1654
+ "loss": 0.2524,
1655
+ "step": 2610
1656
+ },
1657
+ {
1658
+ "epoch": 4.7,
1659
+ "learning_rate": 4.703770197486535e-05,
1660
+ "loss": 0.2596,
1661
+ "step": 2620
1662
+ },
1663
+ {
1664
+ "epoch": 4.72,
1665
+ "learning_rate": 4.721723518850988e-05,
1666
+ "loss": 0.2567,
1667
+ "step": 2630
1668
+ },
1669
+ {
1670
+ "epoch": 4.74,
1671
+ "learning_rate": 4.7396768402154403e-05,
1672
+ "loss": 0.2886,
1673
+ "step": 2640
1674
+ },
1675
+ {
1676
+ "epoch": 4.75,
1677
+ "learning_rate": 4.7576301615798926e-05,
1678
+ "loss": 0.2879,
1679
+ "step": 2650
1680
+ },
1681
+ {
1682
+ "epoch": 4.77,
1683
+ "learning_rate": 4.775583482944345e-05,
1684
+ "loss": 0.2703,
1685
+ "step": 2660
1686
+ },
1687
+ {
1688
+ "epoch": 4.79,
1689
+ "learning_rate": 4.793536804308798e-05,
1690
+ "loss": 0.3013,
1691
+ "step": 2670
1692
+ },
1693
+ {
1694
+ "epoch": 4.81,
1695
+ "learning_rate": 4.81149012567325e-05,
1696
+ "loss": 0.2801,
1697
+ "step": 2680
1698
+ },
1699
+ {
1700
+ "epoch": 4.83,
1701
+ "learning_rate": 4.829443447037702e-05,
1702
+ "loss": 0.2886,
1703
+ "step": 2690
1704
+ },
1705
+ {
1706
+ "epoch": 4.84,
1707
+ "learning_rate": 4.847396768402155e-05,
1708
+ "loss": 0.3084,
1709
+ "step": 2700
1710
+ },
1711
+ {
1712
+ "epoch": 4.86,
1713
+ "learning_rate": 4.8653500897666074e-05,
1714
+ "loss": 0.2886,
1715
+ "step": 2710
1716
+ },
1717
+ {
1718
+ "epoch": 4.88,
1719
+ "learning_rate": 4.8833034111310596e-05,
1720
+ "loss": 0.2678,
1721
+ "step": 2720
1722
+ },
1723
+ {
1724
+ "epoch": 4.9,
1725
+ "learning_rate": 4.901256732495512e-05,
1726
+ "loss": 0.2628,
1727
+ "step": 2730
1728
+ },
1729
+ {
1730
+ "epoch": 4.91,
1731
+ "learning_rate": 4.919210053859965e-05,
1732
+ "loss": 0.2834,
1733
+ "step": 2740
1734
+ },
1735
+ {
1736
+ "epoch": 4.93,
1737
+ "learning_rate": 4.937163375224417e-05,
1738
+ "loss": 0.2775,
1739
+ "step": 2750
1740
+ },
1741
+ {
1742
+ "epoch": 4.95,
1743
+ "learning_rate": 4.955116696588869e-05,
1744
+ "loss": 0.2987,
1745
+ "step": 2760
1746
+ },
1747
+ {
1748
+ "epoch": 4.97,
1749
+ "learning_rate": 4.9730700179533215e-05,
1750
+ "loss": 0.2409,
1751
+ "step": 2770
1752
+ },
1753
+ {
1754
+ "epoch": 4.99,
1755
+ "learning_rate": 4.991023339317774e-05,
1756
+ "loss": 0.3022,
1757
+ "step": 2780
1758
+ },
1759
+ {
1760
+ "epoch": 5.0,
1761
+ "eval_accuracy": {
1762
+ "accuracy": 0.9471294012110338
1763
+ },
1764
+ "eval_f1": {
1765
+ "f1": 0.944197644580418
1766
+ },
1767
+ "eval_loss": 0.1480465531349182,
1768
+ "eval_precision": {
1769
+ "precision": 0.9443621962369058
1770
+ },
1771
+ "eval_recall": {
1772
+ "recall": 0.9440973623586151
1773
+ },
1774
+ "eval_runtime": 99.3848,
1775
+ "eval_samples_per_second": 179.464,
1776
+ "eval_steps_per_second": 5.615,
1777
+ "step": 2787
1778
+ },
1779
+ {
1780
+ "epoch": 5.0,
1781
+ "learning_rate": 4.999002593257531e-05,
1782
+ "loss": 0.2842,
1783
+ "step": 2790
1784
+ },
1785
+ {
1786
+ "epoch": 5.02,
1787
+ "learning_rate": 4.9970077797725914e-05,
1788
+ "loss": 0.2321,
1789
+ "step": 2800
1790
+ },
1791
+ {
1792
+ "epoch": 5.04,
1793
+ "learning_rate": 4.995012966287652e-05,
1794
+ "loss": 0.259,
1795
+ "step": 2810
1796
+ },
1797
+ {
1798
+ "epoch": 5.06,
1799
+ "learning_rate": 4.993018152802713e-05,
1800
+ "loss": 0.2541,
1801
+ "step": 2820
1802
+ },
1803
+ {
1804
+ "epoch": 5.08,
1805
+ "learning_rate": 4.991023339317774e-05,
1806
+ "loss": 0.2419,
1807
+ "step": 2830
1808
+ },
1809
+ {
1810
+ "epoch": 5.09,
1811
+ "learning_rate": 4.989028525832835e-05,
1812
+ "loss": 0.2506,
1813
+ "step": 2840
1814
+ },
1815
+ {
1816
+ "epoch": 5.11,
1817
+ "learning_rate": 4.9870337123478956e-05,
1818
+ "loss": 0.24,
1819
+ "step": 2850
1820
+ },
1821
+ {
1822
+ "epoch": 5.13,
1823
+ "learning_rate": 4.985038898862957e-05,
1824
+ "loss": 0.2443,
1825
+ "step": 2860
1826
+ },
1827
+ {
1828
+ "epoch": 5.15,
1829
+ "learning_rate": 4.9830440853780174e-05,
1830
+ "loss": 0.2645,
1831
+ "step": 2870
1832
+ },
1833
+ {
1834
+ "epoch": 5.17,
1835
+ "learning_rate": 4.9810492718930786e-05,
1836
+ "loss": 0.2679,
1837
+ "step": 2880
1838
+ },
1839
+ {
1840
+ "epoch": 5.18,
1841
+ "learning_rate": 4.979054458408139e-05,
1842
+ "loss": 0.2738,
1843
+ "step": 2890
1844
+ },
1845
+ {
1846
+ "epoch": 5.2,
1847
+ "learning_rate": 4.9770596449232e-05,
1848
+ "loss": 0.2329,
1849
+ "step": 2900
1850
+ },
1851
+ {
1852
+ "epoch": 5.22,
1853
+ "learning_rate": 4.975064831438261e-05,
1854
+ "loss": 0.2766,
1855
+ "step": 2910
1856
+ },
1857
+ {
1858
+ "epoch": 5.24,
1859
+ "learning_rate": 4.9730700179533215e-05,
1860
+ "loss": 0.243,
1861
+ "step": 2920
1862
+ },
1863
+ {
1864
+ "epoch": 5.26,
1865
+ "learning_rate": 4.971075204468382e-05,
1866
+ "loss": 0.2722,
1867
+ "step": 2930
1868
+ },
1869
+ {
1870
+ "epoch": 5.27,
1871
+ "learning_rate": 4.969080390983443e-05,
1872
+ "loss": 0.2556,
1873
+ "step": 2940
1874
+ },
1875
+ {
1876
+ "epoch": 5.29,
1877
+ "learning_rate": 4.967085577498504e-05,
1878
+ "loss": 0.2544,
1879
+ "step": 2950
1880
+ },
1881
+ {
1882
+ "epoch": 5.31,
1883
+ "learning_rate": 4.965090764013565e-05,
1884
+ "loss": 0.2353,
1885
+ "step": 2960
1886
+ },
1887
+ {
1888
+ "epoch": 5.33,
1889
+ "learning_rate": 4.963095950528626e-05,
1890
+ "loss": 0.252,
1891
+ "step": 2970
1892
+ },
1893
+ {
1894
+ "epoch": 5.35,
1895
+ "learning_rate": 4.961101137043687e-05,
1896
+ "loss": 0.2303,
1897
+ "step": 2980
1898
+ },
1899
+ {
1900
+ "epoch": 5.36,
1901
+ "learning_rate": 4.9591063235587475e-05,
1902
+ "loss": 0.2357,
1903
+ "step": 2990
1904
+ },
1905
+ {
1906
+ "epoch": 5.38,
1907
+ "learning_rate": 4.957111510073809e-05,
1908
+ "loss": 0.2529,
1909
+ "step": 3000
1910
+ },
1911
+ {
1912
+ "epoch": 5.4,
1913
+ "learning_rate": 4.955116696588869e-05,
1914
+ "loss": 0.2381,
1915
+ "step": 3010
1916
+ },
1917
+ {
1918
+ "epoch": 5.42,
1919
+ "learning_rate": 4.95312188310393e-05,
1920
+ "loss": 0.2616,
1921
+ "step": 3020
1922
+ },
1923
+ {
1924
+ "epoch": 5.43,
1925
+ "learning_rate": 4.951127069618991e-05,
1926
+ "loss": 0.2836,
1927
+ "step": 3030
1928
+ },
1929
+ {
1930
+ "epoch": 5.45,
1931
+ "learning_rate": 4.9491322561340517e-05,
1932
+ "loss": 0.2726,
1933
+ "step": 3040
1934
+ },
1935
+ {
1936
+ "epoch": 5.47,
1937
+ "learning_rate": 4.947137442649112e-05,
1938
+ "loss": 0.2487,
1939
+ "step": 3050
1940
+ },
1941
+ {
1942
+ "epoch": 5.49,
1943
+ "learning_rate": 4.9451426291641735e-05,
1944
+ "loss": 0.2577,
1945
+ "step": 3060
1946
+ },
1947
+ {
1948
+ "epoch": 5.51,
1949
+ "learning_rate": 4.943147815679234e-05,
1950
+ "loss": 0.2453,
1951
+ "step": 3070
1952
+ },
1953
+ {
1954
+ "epoch": 5.52,
1955
+ "learning_rate": 4.941153002194295e-05,
1956
+ "loss": 0.2682,
1957
+ "step": 3080
1958
+ },
1959
+ {
1960
+ "epoch": 5.54,
1961
+ "learning_rate": 4.939158188709356e-05,
1962
+ "loss": 0.2463,
1963
+ "step": 3090
1964
+ },
1965
+ {
1966
+ "epoch": 5.56,
1967
+ "learning_rate": 4.937163375224417e-05,
1968
+ "loss": 0.254,
1969
+ "step": 3100
1970
+ },
1971
+ {
1972
+ "epoch": 5.58,
1973
+ "learning_rate": 4.9351685617394776e-05,
1974
+ "loss": 0.2516,
1975
+ "step": 3110
1976
+ },
1977
+ {
1978
+ "epoch": 5.6,
1979
+ "learning_rate": 4.933173748254539e-05,
1980
+ "loss": 0.2395,
1981
+ "step": 3120
1982
+ },
1983
+ {
1984
+ "epoch": 5.61,
1985
+ "learning_rate": 4.9311789347695994e-05,
1986
+ "loss": 0.2375,
1987
+ "step": 3130
1988
+ },
1989
+ {
1990
+ "epoch": 5.63,
1991
+ "learning_rate": 4.92918412128466e-05,
1992
+ "loss": 0.2551,
1993
+ "step": 3140
1994
+ },
1995
+ {
1996
+ "epoch": 5.65,
1997
+ "learning_rate": 4.927189307799721e-05,
1998
+ "loss": 0.2391,
1999
+ "step": 3150
2000
+ },
2001
+ {
2002
+ "epoch": 5.67,
2003
+ "learning_rate": 4.925194494314782e-05,
2004
+ "loss": 0.2501,
2005
+ "step": 3160
2006
+ },
2007
+ {
2008
+ "epoch": 5.69,
2009
+ "learning_rate": 4.923199680829842e-05,
2010
+ "loss": 0.248,
2011
+ "step": 3170
2012
+ },
2013
+ {
2014
+ "epoch": 5.7,
2015
+ "learning_rate": 4.9212048673449036e-05,
2016
+ "loss": 0.2357,
2017
+ "step": 3180
2018
+ },
2019
+ {
2020
+ "epoch": 5.72,
2021
+ "learning_rate": 4.919210053859965e-05,
2022
+ "loss": 0.2454,
2023
+ "step": 3190
2024
+ },
2025
+ {
2026
+ "epoch": 5.74,
2027
+ "learning_rate": 4.917215240375025e-05,
2028
+ "loss": 0.2728,
2029
+ "step": 3200
2030
+ },
2031
+ {
2032
+ "epoch": 5.76,
2033
+ "learning_rate": 4.915220426890086e-05,
2034
+ "loss": 0.2789,
2035
+ "step": 3210
2036
+ },
2037
+ {
2038
+ "epoch": 5.78,
2039
+ "learning_rate": 4.913225613405147e-05,
2040
+ "loss": 0.2559,
2041
+ "step": 3220
2042
+ },
2043
+ {
2044
+ "epoch": 5.79,
2045
+ "learning_rate": 4.911230799920208e-05,
2046
+ "loss": 0.2631,
2047
+ "step": 3230
2048
+ },
2049
+ {
2050
+ "epoch": 5.81,
2051
+ "learning_rate": 4.909235986435268e-05,
2052
+ "loss": 0.2292,
2053
+ "step": 3240
2054
+ },
2055
+ {
2056
+ "epoch": 5.83,
2057
+ "learning_rate": 4.9072411729503295e-05,
2058
+ "loss": 0.2301,
2059
+ "step": 3250
2060
+ },
2061
+ {
2062
+ "epoch": 5.85,
2063
+ "learning_rate": 4.90524635946539e-05,
2064
+ "loss": 0.2711,
2065
+ "step": 3260
2066
+ },
2067
+ {
2068
+ "epoch": 5.87,
2069
+ "learning_rate": 4.903251545980451e-05,
2070
+ "loss": 0.2588,
2071
+ "step": 3270
2072
+ },
2073
+ {
2074
+ "epoch": 5.88,
2075
+ "learning_rate": 4.901256732495512e-05,
2076
+ "loss": 0.2683,
2077
+ "step": 3280
2078
+ },
2079
+ {
2080
+ "epoch": 5.9,
2081
+ "learning_rate": 4.8992619190105724e-05,
2082
+ "loss": 0.1896,
2083
+ "step": 3290
2084
+ },
2085
+ {
2086
+ "epoch": 5.92,
2087
+ "learning_rate": 4.897267105525634e-05,
2088
+ "loss": 0.2639,
2089
+ "step": 3300
2090
+ },
2091
+ {
2092
+ "epoch": 5.94,
2093
+ "learning_rate": 4.895272292040695e-05,
2094
+ "loss": 0.2606,
2095
+ "step": 3310
2096
+ },
2097
+ {
2098
+ "epoch": 5.96,
2099
+ "learning_rate": 4.893277478555755e-05,
2100
+ "loss": 0.2751,
2101
+ "step": 3320
2102
+ },
2103
+ {
2104
+ "epoch": 5.97,
2105
+ "learning_rate": 4.891282665070816e-05,
2106
+ "loss": 0.2423,
2107
+ "step": 3330
2108
+ },
2109
+ {
2110
+ "epoch": 5.99,
2111
+ "learning_rate": 4.889287851585877e-05,
2112
+ "loss": 0.2283,
2113
+ "step": 3340
2114
+ },
2115
+ {
2116
+ "epoch": 6.0,
2117
+ "eval_accuracy": {
2118
+ "accuracy": 0.9510540479928234
2119
+ },
2120
+ "eval_f1": {
2121
+ "f1": 0.9482048430595941
2122
+ },
2123
+ "eval_loss": 0.13114869594573975,
2124
+ "eval_precision": {
2125
+ "precision": 0.9513484013297254
2126
+ },
2127
+ "eval_recall": {
2128
+ "recall": 0.9456754807307988
2129
+ },
2130
+ "eval_runtime": 99.7361,
2131
+ "eval_samples_per_second": 178.832,
2132
+ "eval_steps_per_second": 5.595,
2133
+ "step": 3345
2134
+ },
2135
+ {
2136
+ "epoch": 6.01,
2137
+ "learning_rate": 4.887293038100938e-05,
2138
+ "loss": 0.2306,
2139
+ "step": 3350
2140
+ },
2141
+ {
2142
+ "epoch": 6.03,
2143
+ "learning_rate": 4.8852982246159984e-05,
2144
+ "loss": 0.2259,
2145
+ "step": 3360
2146
+ },
2147
+ {
2148
+ "epoch": 6.04,
2149
+ "learning_rate": 4.8833034111310596e-05,
2150
+ "loss": 0.2353,
2151
+ "step": 3370
2152
+ },
2153
+ {
2154
+ "epoch": 6.06,
2155
+ "learning_rate": 4.88130859764612e-05,
2156
+ "loss": 0.2264,
2157
+ "step": 3380
2158
+ },
2159
+ {
2160
+ "epoch": 6.08,
2161
+ "learning_rate": 4.8793137841611814e-05,
2162
+ "loss": 0.2264,
2163
+ "step": 3390
2164
+ },
2165
+ {
2166
+ "epoch": 6.1,
2167
+ "learning_rate": 4.877318970676242e-05,
2168
+ "loss": 0.2403,
2169
+ "step": 3400
2170
+ },
2171
+ {
2172
+ "epoch": 6.12,
2173
+ "learning_rate": 4.8753241571913026e-05,
2174
+ "loss": 0.2476,
2175
+ "step": 3410
2176
+ },
2177
+ {
2178
+ "epoch": 6.13,
2179
+ "learning_rate": 4.873329343706364e-05,
2180
+ "loss": 0.2415,
2181
+ "step": 3420
2182
+ },
2183
+ {
2184
+ "epoch": 6.15,
2185
+ "learning_rate": 4.871334530221425e-05,
2186
+ "loss": 0.26,
2187
+ "step": 3430
2188
+ },
2189
+ {
2190
+ "epoch": 6.17,
2191
+ "learning_rate": 4.869339716736485e-05,
2192
+ "loss": 0.2659,
2193
+ "step": 3440
2194
+ },
2195
+ {
2196
+ "epoch": 6.19,
2197
+ "learning_rate": 4.867344903251546e-05,
2198
+ "loss": 0.2114,
2199
+ "step": 3450
2200
+ },
2201
+ {
2202
+ "epoch": 6.21,
2203
+ "learning_rate": 4.8653500897666074e-05,
2204
+ "loss": 0.2485,
2205
+ "step": 3460
2206
+ },
2207
+ {
2208
+ "epoch": 6.22,
2209
+ "learning_rate": 4.863355276281668e-05,
2210
+ "loss": 0.2247,
2211
+ "step": 3470
2212
+ },
2213
+ {
2214
+ "epoch": 6.24,
2215
+ "learning_rate": 4.8613604627967285e-05,
2216
+ "loss": 0.2595,
2217
+ "step": 3480
2218
+ },
2219
+ {
2220
+ "epoch": 6.26,
2221
+ "learning_rate": 4.85936564931179e-05,
2222
+ "loss": 0.2193,
2223
+ "step": 3490
2224
+ },
2225
+ {
2226
+ "epoch": 6.28,
2227
+ "learning_rate": 4.85737083582685e-05,
2228
+ "loss": 0.2051,
2229
+ "step": 3500
2230
+ },
2231
+ {
2232
+ "epoch": 6.3,
2233
+ "learning_rate": 4.8553760223419116e-05,
2234
+ "loss": 0.2105,
2235
+ "step": 3510
2236
+ },
2237
+ {
2238
+ "epoch": 6.31,
2239
+ "learning_rate": 4.853381208856972e-05,
2240
+ "loss": 0.2104,
2241
+ "step": 3520
2242
+ },
2243
+ {
2244
+ "epoch": 6.33,
2245
+ "learning_rate": 4.851386395372033e-05,
2246
+ "loss": 0.2595,
2247
+ "step": 3530
2248
+ },
2249
+ {
2250
+ "epoch": 6.35,
2251
+ "learning_rate": 4.849391581887094e-05,
2252
+ "loss": 0.2199,
2253
+ "step": 3540
2254
+ },
2255
+ {
2256
+ "epoch": 6.37,
2257
+ "learning_rate": 4.847396768402155e-05,
2258
+ "loss": 0.2463,
2259
+ "step": 3550
2260
+ },
2261
+ {
2262
+ "epoch": 6.39,
2263
+ "learning_rate": 4.845401954917215e-05,
2264
+ "loss": 0.221,
2265
+ "step": 3560
2266
+ },
2267
+ {
2268
+ "epoch": 6.4,
2269
+ "learning_rate": 4.843407141432276e-05,
2270
+ "loss": 0.2207,
2271
+ "step": 3570
2272
+ },
2273
+ {
2274
+ "epoch": 6.42,
2275
+ "learning_rate": 4.8414123279473375e-05,
2276
+ "loss": 0.2319,
2277
+ "step": 3580
2278
+ },
2279
+ {
2280
+ "epoch": 6.44,
2281
+ "learning_rate": 4.839417514462398e-05,
2282
+ "loss": 0.2333,
2283
+ "step": 3590
2284
+ },
2285
+ {
2286
+ "epoch": 6.46,
2287
+ "learning_rate": 4.8374227009774586e-05,
2288
+ "loss": 0.237,
2289
+ "step": 3600
2290
+ },
2291
+ {
2292
+ "epoch": 6.48,
2293
+ "learning_rate": 4.83542788749252e-05,
2294
+ "loss": 0.2096,
2295
+ "step": 3610
2296
+ },
2297
+ {
2298
+ "epoch": 6.49,
2299
+ "learning_rate": 4.8334330740075804e-05,
2300
+ "loss": 0.236,
2301
+ "step": 3620
2302
+ },
2303
+ {
2304
+ "epoch": 6.51,
2305
+ "learning_rate": 4.831438260522642e-05,
2306
+ "loss": 0.2376,
2307
+ "step": 3630
2308
+ },
2309
+ {
2310
+ "epoch": 6.53,
2311
+ "learning_rate": 4.829443447037702e-05,
2312
+ "loss": 0.251,
2313
+ "step": 3640
2314
+ },
2315
+ {
2316
+ "epoch": 6.55,
2317
+ "learning_rate": 4.827448633552763e-05,
2318
+ "loss": 0.1974,
2319
+ "step": 3650
2320
+ },
2321
+ {
2322
+ "epoch": 6.57,
2323
+ "learning_rate": 4.825453820067824e-05,
2324
+ "loss": 0.2334,
2325
+ "step": 3660
2326
+ },
2327
+ {
2328
+ "epoch": 6.58,
2329
+ "learning_rate": 4.823459006582885e-05,
2330
+ "loss": 0.2289,
2331
+ "step": 3670
2332
+ },
2333
+ {
2334
+ "epoch": 6.6,
2335
+ "learning_rate": 4.821464193097945e-05,
2336
+ "loss": 0.2252,
2337
+ "step": 3680
2338
+ },
2339
+ {
2340
+ "epoch": 6.62,
2341
+ "learning_rate": 4.8194693796130064e-05,
2342
+ "loss": 0.239,
2343
+ "step": 3690
2344
+ },
2345
+ {
2346
+ "epoch": 6.64,
2347
+ "learning_rate": 4.8174745661280676e-05,
2348
+ "loss": 0.2166,
2349
+ "step": 3700
2350
+ },
2351
+ {
2352
+ "epoch": 6.65,
2353
+ "learning_rate": 4.815479752643128e-05,
2354
+ "loss": 0.225,
2355
+ "step": 3710
2356
+ },
2357
+ {
2358
+ "epoch": 6.67,
2359
+ "learning_rate": 4.813484939158189e-05,
2360
+ "loss": 0.2038,
2361
+ "step": 3720
2362
+ },
2363
+ {
2364
+ "epoch": 6.69,
2365
+ "learning_rate": 4.81149012567325e-05,
2366
+ "loss": 0.214,
2367
+ "step": 3730
2368
+ },
2369
+ {
2370
+ "epoch": 6.71,
2371
+ "learning_rate": 4.8094953121883106e-05,
2372
+ "loss": 0.1958,
2373
+ "step": 3740
2374
+ },
2375
+ {
2376
+ "epoch": 6.73,
2377
+ "learning_rate": 4.807500498703372e-05,
2378
+ "loss": 0.2275,
2379
+ "step": 3750
2380
+ },
2381
+ {
2382
+ "epoch": 6.74,
2383
+ "learning_rate": 4.8055056852184324e-05,
2384
+ "loss": 0.2324,
2385
+ "step": 3760
2386
+ },
2387
+ {
2388
+ "epoch": 6.76,
2389
+ "learning_rate": 4.803510871733493e-05,
2390
+ "loss": 0.2202,
2391
+ "step": 3770
2392
+ },
2393
+ {
2394
+ "epoch": 6.78,
2395
+ "learning_rate": 4.801516058248554e-05,
2396
+ "loss": 0.2285,
2397
+ "step": 3780
2398
+ },
2399
+ {
2400
+ "epoch": 6.8,
2401
+ "learning_rate": 4.7995212447636154e-05,
2402
+ "loss": 0.2385,
2403
+ "step": 3790
2404
+ },
2405
+ {
2406
+ "epoch": 6.82,
2407
+ "learning_rate": 4.797526431278675e-05,
2408
+ "loss": 0.2366,
2409
+ "step": 3800
2410
+ },
2411
+ {
2412
+ "epoch": 6.83,
2413
+ "learning_rate": 4.7955316177937365e-05,
2414
+ "loss": 0.2221,
2415
+ "step": 3810
2416
+ },
2417
+ {
2418
+ "epoch": 6.85,
2419
+ "learning_rate": 4.793536804308798e-05,
2420
+ "loss": 0.2349,
2421
+ "step": 3820
2422
+ },
2423
+ {
2424
+ "epoch": 6.87,
2425
+ "learning_rate": 4.791541990823858e-05,
2426
+ "loss": 0.2749,
2427
+ "step": 3830
2428
+ },
2429
+ {
2430
+ "epoch": 6.89,
2431
+ "learning_rate": 4.789547177338919e-05,
2432
+ "loss": 0.2168,
2433
+ "step": 3840
2434
+ },
2435
+ {
2436
+ "epoch": 6.91,
2437
+ "learning_rate": 4.78755236385398e-05,
2438
+ "loss": 0.1872,
2439
+ "step": 3850
2440
+ },
2441
+ {
2442
+ "epoch": 6.92,
2443
+ "learning_rate": 4.785557550369041e-05,
2444
+ "loss": 0.2502,
2445
+ "step": 3860
2446
+ },
2447
+ {
2448
+ "epoch": 6.94,
2449
+ "learning_rate": 4.783562736884102e-05,
2450
+ "loss": 0.1964,
2451
+ "step": 3870
2452
+ },
2453
+ {
2454
+ "epoch": 6.96,
2455
+ "learning_rate": 4.7815679233991625e-05,
2456
+ "loss": 0.2425,
2457
+ "step": 3880
2458
+ },
2459
+ {
2460
+ "epoch": 6.98,
2461
+ "learning_rate": 4.779573109914223e-05,
2462
+ "loss": 0.2481,
2463
+ "step": 3890
2464
+ },
2465
+ {
2466
+ "epoch": 7.0,
2467
+ "learning_rate": 4.777578296429284e-05,
2468
+ "loss": 0.2376,
2469
+ "step": 3900
2470
+ },
2471
+ {
2472
+ "epoch": 7.0,
2473
+ "eval_accuracy": {
2474
+ "accuracy": 0.9595761381475667
2475
+ },
2476
+ "eval_f1": {
2477
+ "f1": 0.95694652190804
2478
+ },
2479
+ "eval_loss": 0.10949217528104782,
2480
+ "eval_precision": {
2481
+ "precision": 0.9583449741949119
2482
+ },
2483
+ "eval_recall": {
2484
+ "recall": 0.9557128312494094
2485
+ },
2486
+ "eval_runtime": 99.0631,
2487
+ "eval_samples_per_second": 180.047,
2488
+ "eval_steps_per_second": 5.633,
2489
+ "step": 3902
2490
+ },
2491
+ {
2492
+ "epoch": 7.01,
2493
+ "learning_rate": 4.775583482944345e-05,
2494
+ "loss": 0.2048,
2495
+ "step": 3910
2496
+ },
2497
+ {
2498
+ "epoch": 7.03,
2499
+ "learning_rate": 4.7735886694594054e-05,
2500
+ "loss": 0.2573,
2501
+ "step": 3920
2502
+ },
2503
+ {
2504
+ "epoch": 7.05,
2505
+ "learning_rate": 4.7715938559744666e-05,
2506
+ "loss": 0.2261,
2507
+ "step": 3930
2508
+ },
2509
+ {
2510
+ "epoch": 7.07,
2511
+ "learning_rate": 4.769599042489528e-05,
2512
+ "loss": 0.2094,
2513
+ "step": 3940
2514
+ },
2515
+ {
2516
+ "epoch": 7.09,
2517
+ "learning_rate": 4.767604229004588e-05,
2518
+ "loss": 0.2148,
2519
+ "step": 3950
2520
+ },
2521
+ {
2522
+ "epoch": 7.1,
2523
+ "learning_rate": 4.765609415519649e-05,
2524
+ "loss": 0.2023,
2525
+ "step": 3960
2526
+ },
2527
+ {
2528
+ "epoch": 7.12,
2529
+ "learning_rate": 4.76361460203471e-05,
2530
+ "loss": 0.2031,
2531
+ "step": 3970
2532
+ },
2533
+ {
2534
+ "epoch": 7.14,
2535
+ "learning_rate": 4.761619788549771e-05,
2536
+ "loss": 0.2051,
2537
+ "step": 3980
2538
+ },
2539
+ {
2540
+ "epoch": 7.16,
2541
+ "learning_rate": 4.7596249750648313e-05,
2542
+ "loss": 0.1944,
2543
+ "step": 3990
2544
+ },
2545
+ {
2546
+ "epoch": 7.17,
2547
+ "learning_rate": 4.7576301615798926e-05,
2548
+ "loss": 0.2319,
2549
+ "step": 4000
2550
+ },
2551
+ {
2552
+ "epoch": 7.19,
2553
+ "learning_rate": 4.755635348094953e-05,
2554
+ "loss": 0.2164,
2555
+ "step": 4010
2556
+ },
2557
+ {
2558
+ "epoch": 7.21,
2559
+ "learning_rate": 4.7536405346100144e-05,
2560
+ "loss": 0.1977,
2561
+ "step": 4020
2562
+ },
2563
+ {
2564
+ "epoch": 7.23,
2565
+ "learning_rate": 4.751645721125075e-05,
2566
+ "loss": 0.2499,
2567
+ "step": 4030
2568
+ },
2569
+ {
2570
+ "epoch": 7.25,
2571
+ "learning_rate": 4.7496509076401355e-05,
2572
+ "loss": 0.1717,
2573
+ "step": 4040
2574
+ },
2575
+ {
2576
+ "epoch": 7.26,
2577
+ "learning_rate": 4.747656094155197e-05,
2578
+ "loss": 0.1886,
2579
+ "step": 4050
2580
+ },
2581
+ {
2582
+ "epoch": 7.28,
2583
+ "learning_rate": 4.745661280670258e-05,
2584
+ "loss": 0.2132,
2585
+ "step": 4060
2586
+ },
2587
+ {
2588
+ "epoch": 7.3,
2589
+ "learning_rate": 4.743666467185318e-05,
2590
+ "loss": 0.2238,
2591
+ "step": 4070
2592
+ },
2593
+ {
2594
+ "epoch": 7.32,
2595
+ "learning_rate": 4.741671653700379e-05,
2596
+ "loss": 0.2213,
2597
+ "step": 4080
2598
+ },
2599
+ {
2600
+ "epoch": 7.34,
2601
+ "learning_rate": 4.7396768402154403e-05,
2602
+ "loss": 0.1884,
2603
+ "step": 4090
2604
+ },
2605
+ {
2606
+ "epoch": 7.35,
2607
+ "learning_rate": 4.737682026730501e-05,
2608
+ "loss": 0.2112,
2609
+ "step": 4100
2610
+ },
2611
+ {
2612
+ "epoch": 7.37,
2613
+ "learning_rate": 4.7356872132455615e-05,
2614
+ "loss": 0.218,
2615
+ "step": 4110
2616
+ },
2617
+ {
2618
+ "epoch": 7.39,
2619
+ "learning_rate": 4.733692399760623e-05,
2620
+ "loss": 0.1901,
2621
+ "step": 4120
2622
+ },
2623
+ {
2624
+ "epoch": 7.41,
2625
+ "learning_rate": 4.731697586275683e-05,
2626
+ "loss": 0.2182,
2627
+ "step": 4130
2628
+ },
2629
+ {
2630
+ "epoch": 7.43,
2631
+ "learning_rate": 4.7297027727907445e-05,
2632
+ "loss": 0.2061,
2633
+ "step": 4140
2634
+ },
2635
+ {
2636
+ "epoch": 7.44,
2637
+ "learning_rate": 4.727707959305805e-05,
2638
+ "loss": 0.1767,
2639
+ "step": 4150
2640
+ },
2641
+ {
2642
+ "epoch": 7.46,
2643
+ "learning_rate": 4.7257131458208656e-05,
2644
+ "loss": 0.2376,
2645
+ "step": 4160
2646
+ },
2647
+ {
2648
+ "epoch": 7.48,
2649
+ "learning_rate": 4.723718332335927e-05,
2650
+ "loss": 0.203,
2651
+ "step": 4170
2652
+ },
2653
+ {
2654
+ "epoch": 7.5,
2655
+ "learning_rate": 4.721723518850988e-05,
2656
+ "loss": 0.1879,
2657
+ "step": 4180
2658
+ },
2659
+ {
2660
+ "epoch": 7.52,
2661
+ "learning_rate": 4.719728705366048e-05,
2662
+ "loss": 0.2355,
2663
+ "step": 4190
2664
+ },
2665
+ {
2666
+ "epoch": 7.53,
2667
+ "learning_rate": 4.717733891881109e-05,
2668
+ "loss": 0.1916,
2669
+ "step": 4200
2670
+ },
2671
+ {
2672
+ "epoch": 7.55,
2673
+ "learning_rate": 4.7157390783961705e-05,
2674
+ "loss": 0.2133,
2675
+ "step": 4210
2676
+ },
2677
+ {
2678
+ "epoch": 7.57,
2679
+ "learning_rate": 4.713744264911231e-05,
2680
+ "loss": 0.208,
2681
+ "step": 4220
2682
+ },
2683
+ {
2684
+ "epoch": 7.59,
2685
+ "learning_rate": 4.7117494514262916e-05,
2686
+ "loss": 0.2364,
2687
+ "step": 4230
2688
+ },
2689
+ {
2690
+ "epoch": 7.61,
2691
+ "learning_rate": 4.709754637941353e-05,
2692
+ "loss": 0.2049,
2693
+ "step": 4240
2694
+ },
2695
+ {
2696
+ "epoch": 7.62,
2697
+ "learning_rate": 4.7077598244564134e-05,
2698
+ "loss": 0.2296,
2699
+ "step": 4250
2700
+ },
2701
+ {
2702
+ "epoch": 7.64,
2703
+ "learning_rate": 4.7057650109714746e-05,
2704
+ "loss": 0.221,
2705
+ "step": 4260
2706
+ },
2707
+ {
2708
+ "epoch": 7.66,
2709
+ "learning_rate": 4.703770197486535e-05,
2710
+ "loss": 0.1917,
2711
+ "step": 4270
2712
+ },
2713
+ {
2714
+ "epoch": 7.68,
2715
+ "learning_rate": 4.701775384001596e-05,
2716
+ "loss": 0.2183,
2717
+ "step": 4280
2718
+ },
2719
+ {
2720
+ "epoch": 7.7,
2721
+ "learning_rate": 4.699780570516657e-05,
2722
+ "loss": 0.1869,
2723
+ "step": 4290
2724
+ },
2725
+ {
2726
+ "epoch": 7.71,
2727
+ "learning_rate": 4.697785757031718e-05,
2728
+ "loss": 0.1975,
2729
+ "step": 4300
2730
+ },
2731
+ {
2732
+ "epoch": 7.73,
2733
+ "learning_rate": 4.695790943546778e-05,
2734
+ "loss": 0.2274,
2735
+ "step": 4310
2736
+ },
2737
+ {
2738
+ "epoch": 7.75,
2739
+ "learning_rate": 4.693796130061839e-05,
2740
+ "loss": 0.1973,
2741
+ "step": 4320
2742
+ },
2743
+ {
2744
+ "epoch": 7.77,
2745
+ "learning_rate": 4.6918013165769006e-05,
2746
+ "loss": 0.2005,
2747
+ "step": 4330
2748
+ },
2749
+ {
2750
+ "epoch": 7.78,
2751
+ "learning_rate": 4.689806503091961e-05,
2752
+ "loss": 0.1978,
2753
+ "step": 4340
2754
+ },
2755
+ {
2756
+ "epoch": 7.8,
2757
+ "learning_rate": 4.687811689607022e-05,
2758
+ "loss": 0.1986,
2759
+ "step": 4350
2760
+ },
2761
+ {
2762
+ "epoch": 7.82,
2763
+ "learning_rate": 4.685816876122083e-05,
2764
+ "loss": 0.2013,
2765
+ "step": 4360
2766
+ },
2767
+ {
2768
+ "epoch": 7.84,
2769
+ "learning_rate": 4.6838220626371435e-05,
2770
+ "loss": 0.2218,
2771
+ "step": 4370
2772
+ },
2773
+ {
2774
+ "epoch": 7.86,
2775
+ "learning_rate": 4.681827249152205e-05,
2776
+ "loss": 0.2168,
2777
+ "step": 4380
2778
+ },
2779
+ {
2780
+ "epoch": 7.87,
2781
+ "learning_rate": 4.679832435667265e-05,
2782
+ "loss": 0.1709,
2783
+ "step": 4390
2784
+ },
2785
+ {
2786
+ "epoch": 7.89,
2787
+ "learning_rate": 4.677837622182326e-05,
2788
+ "loss": 0.2249,
2789
+ "step": 4400
2790
+ },
2791
+ {
2792
+ "epoch": 7.91,
2793
+ "learning_rate": 4.675842808697387e-05,
2794
+ "loss": 0.2246,
2795
+ "step": 4410
2796
+ },
2797
+ {
2798
+ "epoch": 7.93,
2799
+ "learning_rate": 4.673847995212448e-05,
2800
+ "loss": 0.2163,
2801
+ "step": 4420
2802
+ },
2803
+ {
2804
+ "epoch": 7.95,
2805
+ "learning_rate": 4.671853181727508e-05,
2806
+ "loss": 0.1922,
2807
+ "step": 4430
2808
+ },
2809
+ {
2810
+ "epoch": 7.96,
2811
+ "learning_rate": 4.6698583682425695e-05,
2812
+ "loss": 0.2157,
2813
+ "step": 4440
2814
+ },
2815
+ {
2816
+ "epoch": 7.98,
2817
+ "learning_rate": 4.667863554757631e-05,
2818
+ "loss": 0.2272,
2819
+ "step": 4450
2820
+ },
2821
+ {
2822
+ "epoch": 8.0,
2823
+ "learning_rate": 4.665868741272691e-05,
2824
+ "loss": 0.2027,
2825
+ "step": 4460
2826
+ },
2827
+ {
2828
+ "epoch": 8.0,
2829
+ "eval_accuracy": {
2830
+ "accuracy": 0.9548665620094191
2831
+ },
2832
+ "eval_f1": {
2833
+ "f1": 0.951539693294557
2834
+ },
2835
+ "eval_loss": 0.12259286642074585,
2836
+ "eval_precision": {
2837
+ "precision": 0.9548733769718504
2838
+ },
2839
+ "eval_recall": {
2840
+ "recall": 0.9494659029809656
2841
+ },
2842
+ "eval_runtime": 98.7615,
2843
+ "eval_samples_per_second": 180.597,
2844
+ "eval_steps_per_second": 5.65,
2845
+ "step": 4460
2846
+ },
2847
+ {
2848
+ "epoch": 8.02,
2849
+ "learning_rate": 4.663873927787752e-05,
2850
+ "loss": 0.1895,
2851
+ "step": 4470
2852
+ },
2853
+ {
2854
+ "epoch": 8.04,
2855
+ "learning_rate": 4.661879114302813e-05,
2856
+ "loss": 0.1628,
2857
+ "step": 4480
2858
+ },
2859
+ {
2860
+ "epoch": 8.05,
2861
+ "learning_rate": 4.6598843008178736e-05,
2862
+ "loss": 0.1996,
2863
+ "step": 4490
2864
+ },
2865
+ {
2866
+ "epoch": 8.07,
2867
+ "learning_rate": 4.657889487332935e-05,
2868
+ "loss": 0.2024,
2869
+ "step": 4500
2870
+ },
2871
+ {
2872
+ "epoch": 8.09,
2873
+ "learning_rate": 4.6558946738479954e-05,
2874
+ "loss": 0.1887,
2875
+ "step": 4510
2876
+ },
2877
+ {
2878
+ "epoch": 8.11,
2879
+ "learning_rate": 4.653899860363056e-05,
2880
+ "loss": 0.2224,
2881
+ "step": 4520
2882
+ },
2883
+ {
2884
+ "epoch": 8.13,
2885
+ "learning_rate": 4.651905046878117e-05,
2886
+ "loss": 0.1992,
2887
+ "step": 4530
2888
+ },
2889
+ {
2890
+ "epoch": 8.14,
2891
+ "learning_rate": 4.6499102333931784e-05,
2892
+ "loss": 0.2027,
2893
+ "step": 4540
2894
+ },
2895
+ {
2896
+ "epoch": 8.16,
2897
+ "learning_rate": 4.647915419908238e-05,
2898
+ "loss": 0.1844,
2899
+ "step": 4550
2900
+ },
2901
+ {
2902
+ "epoch": 8.18,
2903
+ "learning_rate": 4.6459206064232996e-05,
2904
+ "loss": 0.1838,
2905
+ "step": 4560
2906
+ },
2907
+ {
2908
+ "epoch": 8.2,
2909
+ "learning_rate": 4.643925792938361e-05,
2910
+ "loss": 0.1897,
2911
+ "step": 4570
2912
+ },
2913
+ {
2914
+ "epoch": 8.22,
2915
+ "learning_rate": 4.6419309794534214e-05,
2916
+ "loss": 0.1794,
2917
+ "step": 4580
2918
+ },
2919
+ {
2920
+ "epoch": 8.23,
2921
+ "learning_rate": 4.639936165968482e-05,
2922
+ "loss": 0.2052,
2923
+ "step": 4590
2924
+ },
2925
+ {
2926
+ "epoch": 8.25,
2927
+ "learning_rate": 4.637941352483543e-05,
2928
+ "loss": 0.225,
2929
+ "step": 4600
2930
+ },
2931
+ {
2932
+ "epoch": 8.27,
2933
+ "learning_rate": 4.635946538998604e-05,
2934
+ "loss": 0.1905,
2935
+ "step": 4610
2936
+ },
2937
+ {
2938
+ "epoch": 8.29,
2939
+ "learning_rate": 4.633951725513665e-05,
2940
+ "loss": 0.1933,
2941
+ "step": 4620
2942
+ },
2943
+ {
2944
+ "epoch": 8.3,
2945
+ "learning_rate": 4.6319569120287255e-05,
2946
+ "loss": 0.2084,
2947
+ "step": 4630
2948
+ },
2949
+ {
2950
+ "epoch": 8.32,
2951
+ "learning_rate": 4.629962098543786e-05,
2952
+ "loss": 0.1947,
2953
+ "step": 4640
2954
+ },
2955
+ {
2956
+ "epoch": 8.34,
2957
+ "learning_rate": 4.627967285058847e-05,
2958
+ "loss": 0.1984,
2959
+ "step": 4650
2960
+ },
2961
+ {
2962
+ "epoch": 8.36,
2963
+ "learning_rate": 4.625972471573908e-05,
2964
+ "loss": 0.21,
2965
+ "step": 4660
2966
+ },
2967
+ {
2968
+ "epoch": 8.38,
2969
+ "learning_rate": 4.6239776580889684e-05,
2970
+ "loss": 0.2083,
2971
+ "step": 4670
2972
+ },
2973
+ {
2974
+ "epoch": 8.39,
2975
+ "learning_rate": 4.62198284460403e-05,
2976
+ "loss": 0.2066,
2977
+ "step": 4680
2978
+ },
2979
+ {
2980
+ "epoch": 8.41,
2981
+ "learning_rate": 4.619988031119091e-05,
2982
+ "loss": 0.1874,
2983
+ "step": 4690
2984
+ },
2985
+ {
2986
+ "epoch": 8.43,
2987
+ "learning_rate": 4.6179932176341515e-05,
2988
+ "loss": 0.2062,
2989
+ "step": 4700
2990
+ },
2991
+ {
2992
+ "epoch": 8.45,
2993
+ "learning_rate": 4.615998404149212e-05,
2994
+ "loss": 0.1714,
2995
+ "step": 4710
2996
+ },
2997
+ {
2998
+ "epoch": 8.47,
2999
+ "learning_rate": 4.614003590664273e-05,
3000
+ "loss": 0.1864,
3001
+ "step": 4720
3002
+ },
3003
+ {
3004
+ "epoch": 8.48,
3005
+ "learning_rate": 4.612008777179334e-05,
3006
+ "loss": 0.1997,
3007
+ "step": 4730
3008
+ },
3009
+ {
3010
+ "epoch": 8.5,
3011
+ "learning_rate": 4.6100139636943944e-05,
3012
+ "loss": 0.2102,
3013
+ "step": 4740
3014
+ },
3015
+ {
3016
+ "epoch": 8.52,
3017
+ "learning_rate": 4.6080191502094556e-05,
3018
+ "loss": 0.2008,
3019
+ "step": 4750
3020
+ },
3021
+ {
3022
+ "epoch": 8.54,
3023
+ "learning_rate": 4.606024336724516e-05,
3024
+ "loss": 0.1874,
3025
+ "step": 4760
3026
+ },
3027
+ {
3028
+ "epoch": 8.56,
3029
+ "learning_rate": 4.6040295232395774e-05,
3030
+ "loss": 0.1718,
3031
+ "step": 4770
3032
+ },
3033
+ {
3034
+ "epoch": 8.57,
3035
+ "learning_rate": 4.602034709754638e-05,
3036
+ "loss": 0.1799,
3037
+ "step": 4780
3038
+ },
3039
+ {
3040
+ "epoch": 8.59,
3041
+ "learning_rate": 4.6000398962696986e-05,
3042
+ "loss": 0.1741,
3043
+ "step": 4790
3044
+ },
3045
+ {
3046
+ "epoch": 8.61,
3047
+ "learning_rate": 4.59804508278476e-05,
3048
+ "loss": 0.1786,
3049
+ "step": 4800
3050
+ },
3051
+ {
3052
+ "epoch": 8.63,
3053
+ "learning_rate": 4.596050269299821e-05,
3054
+ "loss": 0.1853,
3055
+ "step": 4810
3056
+ },
3057
+ {
3058
+ "epoch": 8.65,
3059
+ "learning_rate": 4.5940554558148816e-05,
3060
+ "loss": 0.1964,
3061
+ "step": 4820
3062
+ },
3063
+ {
3064
+ "epoch": 8.66,
3065
+ "learning_rate": 4.592060642329942e-05,
3066
+ "loss": 0.2065,
3067
+ "step": 4830
3068
+ },
3069
+ {
3070
+ "epoch": 8.68,
3071
+ "learning_rate": 4.5900658288450034e-05,
3072
+ "loss": 0.2021,
3073
+ "step": 4840
3074
+ },
3075
+ {
3076
+ "epoch": 8.7,
3077
+ "learning_rate": 4.588071015360064e-05,
3078
+ "loss": 0.1887,
3079
+ "step": 4850
3080
+ },
3081
+ {
3082
+ "epoch": 8.72,
3083
+ "learning_rate": 4.5860762018751245e-05,
3084
+ "loss": 0.2203,
3085
+ "step": 4860
3086
+ },
3087
+ {
3088
+ "epoch": 8.74,
3089
+ "learning_rate": 4.584081388390186e-05,
3090
+ "loss": 0.1808,
3091
+ "step": 4870
3092
+ },
3093
+ {
3094
+ "epoch": 8.75,
3095
+ "learning_rate": 4.582086574905246e-05,
3096
+ "loss": 0.2213,
3097
+ "step": 4880
3098
+ },
3099
+ {
3100
+ "epoch": 8.77,
3101
+ "learning_rate": 4.5800917614203076e-05,
3102
+ "loss": 0.1934,
3103
+ "step": 4890
3104
+ },
3105
+ {
3106
+ "epoch": 8.79,
3107
+ "learning_rate": 4.578096947935368e-05,
3108
+ "loss": 0.2048,
3109
+ "step": 4900
3110
+ },
3111
+ {
3112
+ "epoch": 8.81,
3113
+ "learning_rate": 4.576102134450429e-05,
3114
+ "loss": 0.217,
3115
+ "step": 4910
3116
+ },
3117
+ {
3118
+ "epoch": 8.83,
3119
+ "learning_rate": 4.57410732096549e-05,
3120
+ "loss": 0.2161,
3121
+ "step": 4920
3122
+ },
3123
+ {
3124
+ "epoch": 8.84,
3125
+ "learning_rate": 4.572112507480551e-05,
3126
+ "loss": 0.2145,
3127
+ "step": 4930
3128
+ },
3129
+ {
3130
+ "epoch": 8.86,
3131
+ "learning_rate": 4.570117693995612e-05,
3132
+ "loss": 0.1722,
3133
+ "step": 4940
3134
+ },
3135
+ {
3136
+ "epoch": 8.88,
3137
+ "learning_rate": 4.568122880510672e-05,
3138
+ "loss": 0.1904,
3139
+ "step": 4950
3140
+ },
3141
+ {
3142
+ "epoch": 8.9,
3143
+ "learning_rate": 4.5661280670257335e-05,
3144
+ "loss": 0.1888,
3145
+ "step": 4960
3146
+ },
3147
+ {
3148
+ "epoch": 8.91,
3149
+ "learning_rate": 4.564133253540794e-05,
3150
+ "loss": 0.1931,
3151
+ "step": 4970
3152
+ },
3153
+ {
3154
+ "epoch": 8.93,
3155
+ "learning_rate": 4.5621384400558546e-05,
3156
+ "loss": 0.1607,
3157
+ "step": 4980
3158
+ },
3159
+ {
3160
+ "epoch": 8.95,
3161
+ "learning_rate": 4.560143626570916e-05,
3162
+ "loss": 0.2014,
3163
+ "step": 4990
3164
+ },
3165
+ {
3166
+ "epoch": 8.97,
3167
+ "learning_rate": 4.5581488130859764e-05,
3168
+ "loss": 0.2043,
3169
+ "step": 5000
3170
+ },
3171
+ {
3172
+ "epoch": 8.99,
3173
+ "learning_rate": 4.556153999601038e-05,
3174
+ "loss": 0.1807,
3175
+ "step": 5010
3176
+ },
3177
+ {
3178
+ "epoch": 9.0,
3179
+ "eval_accuracy": {
3180
+ "accuracy": 0.9633886521641624
3181
+ },
3182
+ "eval_f1": {
3183
+ "f1": 0.9611100614395742
3184
+ },
3185
+ "eval_loss": 0.10741779208183289,
3186
+ "eval_precision": {
3187
+ "precision": 0.9618083534790791
3188
+ },
3189
+ "eval_recall": {
3190
+ "recall": 0.9605394028604899
3191
+ },
3192
+ "eval_runtime": 98.612,
3193
+ "eval_samples_per_second": 180.87,
3194
+ "eval_steps_per_second": 5.659,
3195
+ "step": 5017
3196
+ }
3197
+ ],
3198
+ "logging_steps": 10,
3199
+ "max_steps": 27850,
3200
+ "num_input_tokens_seen": 0,
3201
+ "num_train_epochs": 50,
3202
+ "save_steps": 500,
3203
+ "total_flos": 1.5959928156073574e+19,
3204
+ "train_batch_size": 32,
3205
+ "trial_name": null,
3206
+ "trial_params": null
3207
+ }
tmp-checkpoint-5017/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c081bc3c7c7c6c246b97a8d5de1f01d4b69caf186a7167574020cb94e76e323a
3
+ size 4792