Hugging Face's logo

Kushagra07
/
autotrain-swinv2-base-patch4-window8-256

Image Classification
Transformers
TensorBoard
Safetensors
swinv2
Trained with AutoTrain
Inference Endpoints
Model card Files Metrics Community
Kushagra07 commited on
Commit
2b740e0
•
1 Parent(s): 25c8fa2

Upload folder using huggingface_hub

Browse files
Files changed (15) hide show
  1. README.md +42 -0
  2. checkpoint-11696/config.json +98 -0
  3. checkpoint-11696/model.safetensors +3 -0
  4. checkpoint-11696/optimizer.pt +3 -0
  5. checkpoint-11696/rng_state.pth +3 -0
  6. checkpoint-11696/scheduler.pt +3 -0
  7. checkpoint-11696/trainer_state.json +3434 -0
  8. checkpoint-11696/training_args.bin +3 -0
  9. config.json +98 -0
  10. model.safetensors +3 -0
  11. preprocessor_config.json +36 -0
  12. runs/May03_05-17-47_60f4804cf903/events.out.tfevents.1714713467.60f4804cf903.10793.0 +2 -2
  13. runs/May03_05-17-47_60f4804cf903/events.out.tfevents.1714716327.60f4804cf903.10793.1 +3 -0
  14. training_args.bin +3 -0
  15. training_params.json +27 -0
README.md ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ tags:
4
+ - autotrain
5
+ - image-classification
6
+ widget:
7
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
8
+ example_title: Tiger
9
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
10
+ example_title: Teapot
11
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
12
+ example_title: Palace
13
+ datasets:
14
+ - autotrain-swinv2-base-patch4-window8-256/autotrain-data
15
+ ---
16
+
17
+ # Model Trained Using AutoTrain
18
+
19
+ - Problem type: Image Classification
20
+
21
+ ## Validation Metrics
22
+ loss: 0.21217399835586548
23
+
24
+ f1_macro: 0.8801923417646881
25
+
26
+ f1_micro: 0.9320587231136906
27
+
28
+ f1_weighted: 0.9322151264732859
29
+
30
+ precision_macro: 0.9267115227700036
31
+
32
+ precision_micro: 0.9320587231136906
33
+
34
+ precision_weighted: 0.9357267323781668
35
+
36
+ recall_macro: 0.8522160392320227
37
+
38
+ recall_micro: 0.9320587231136906
39
+
40
+ recall_weighted: 0.9320587231136906
41
+
42
+ accuracy: 0.9320587231136906
checkpoint-11696/config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/swinv2-base-patch4-window8-256",
3
+ "_num_labels": 17,
4
+ "architectures": [
5
+ "Swinv2ForImageClassification"
6
+ ],
7
+ "attention_probs_dropout_prob": 0.0,
8
+ "depths": [
9
+ 2,
10
+ 2,
11
+ 18,
12
+ 2
13
+ ],
14
+ "drop_path_rate": 0.1,
15
+ "embed_dim": 128,
16
+ "encoder_stride": 32,
17
+ "hidden_act": "gelu",
18
+ "hidden_dropout_prob": 0.0,
19
+ "hidden_size": 1024,
20
+ "id2label": {
21
+ "0": "AadhaarBack",
22
+ "1": "AadhaarBackMasked",
23
+ "2": "AadhaarFront",
24
+ "3": "AadhaarFrontMasked",
25
+ "4": "AadhaarRegular",
26
+ "5": "AadhaarRegularMasked",
27
+ "6": "AadhaarSelfGenerated",
28
+ "7": "DrivingLicenseNew",
29
+ "8": "DrivingLicenseOld",
30
+ "9": "MultipleOVDs",
31
+ "10": "PanCard",
32
+ "11": "PassportFirst",
33
+ "12": "PassportLast",
34
+ "13": "PassportRegular",
35
+ "14": "VoterCardBack",
36
+ "15": "VoterCardFront",
37
+ "16": "VoterCardRegular"
38
+ },
39
+ "image_size": 256,
40
+ "initializer_range": 0.02,
41
+ "label2id": {
42
+ "AadhaarBack": 0,
43
+ "AadhaarBackMasked": 1,
44
+ "AadhaarFront": 2,
45
+ "AadhaarFrontMasked": 3,
46
+ "AadhaarRegular": 4,
47
+ "AadhaarRegularMasked": 5,
48
+ "AadhaarSelfGenerated": 6,
49
+ "DrivingLicenseNew": 7,
50
+ "DrivingLicenseOld": 8,
51
+ "MultipleOVDs": 9,
52
+ "PanCard": 10,
53
+ "PassportFirst": 11,
54
+ "PassportLast": 12,
55
+ "PassportRegular": 13,
56
+ "VoterCardBack": 14,
57
+ "VoterCardFront": 15,
58
+ "VoterCardRegular": 16
59
+ },
60
+ "layer_norm_eps": 1e-05,
61
+ "mlp_ratio": 4.0,
62
+ "model_type": "swinv2",
63
+ "num_channels": 3,
64
+ "num_heads": [
65
+ 4,
66
+ 8,
67
+ 16,
68
+ 32
69
+ ],
70
+ "num_layers": 4,
71
+ "out_features": [
72
+ "stage4"
73
+ ],
74
+ "out_indices": [
75
+ 4
76
+ ],
77
+ "patch_size": 4,
78
+ "path_norm": true,
79
+ "pretrained_window_sizes": [
80
+ 0,
81
+ 0,
82
+ 0,
83
+ 0
84
+ ],
85
+ "problem_type": "single_label_classification",
86
+ "qkv_bias": true,
87
+ "stage_names": [
88
+ "stem",
89
+ "stage1",
90
+ "stage2",
91
+ "stage3",
92
+ "stage4"
93
+ ],
94
+ "torch_dtype": "float32",
95
+ "transformers_version": "4.40.1",
96
+ "use_absolute_embeddings": false,
97
+ "window_size": 8
98
+ }
checkpoint-11696/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:983acdad57f05464ad5abbd11fb32e0e8ba8b0be8f7fd3762edf4f3626c10322
3
+ size 347706996
checkpoint-11696/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d27d99074d1545ea3b188401019fa8ae2dae425976ebb987c15a3d8fb5fe71a8
3
+ size 695686109
checkpoint-11696/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd0bc2eba1b0a86f8514345cea6c96f45d69c6af369f5b91f748f8c04223441f
3
+ size 14180
checkpoint-11696/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41166ab2d4c25e1852d59df652d331eac86c8e67326debddee19af71bb660690
3
+ size 1064
checkpoint-11696/trainer_state.json ADDED
@@ -0,0 +1,3434 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.21217399835586548,
3
+ "best_model_checkpoint": "autotrain-swinv2-base-patch4-window8-256/checkpoint-11696",
4
+ "epoch": 8.0,
5
+ "eval_steps": 500,
6
+ "global_step": 11696,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01709986320109439,
13
+ "grad_norm": 10.912551879882812,
14
+ "learning_rate": 8.207934336525308e-07,
15
+ "loss": 2.9335,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.03419972640218878,
20
+ "grad_norm": 11.582744598388672,
21
+ "learning_rate": 1.6757865937072504e-06,
22
+ "loss": 2.8112,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.05129958960328317,
27
+ "grad_norm": 11.078932762145996,
28
+ "learning_rate": 2.5307797537619698e-06,
29
+ "loss": 2.6726,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.06839945280437756,
34
+ "grad_norm": 12.702400207519531,
35
+ "learning_rate": 3.38577291381669e-06,
36
+ "loss": 2.4547,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.08549931600547196,
41
+ "grad_norm": 16.10662078857422,
42
+ "learning_rate": 4.240766073871409e-06,
43
+ "loss": 2.0783,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.10259917920656635,
48
+ "grad_norm": 15.413719177246094,
49
+ "learning_rate": 5.0615595075239396e-06,
50
+ "loss": 1.8944,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.11969904240766074,
55
+ "grad_norm": 14.965916633605957,
56
+ "learning_rate": 5.91655266757866e-06,
57
+ "loss": 1.6313,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.13679890560875513,
62
+ "grad_norm": 15.34847640991211,
63
+ "learning_rate": 6.77154582763338e-06,
64
+ "loss": 1.5534,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.1538987688098495,
69
+ "grad_norm": 14.655660629272461,
70
+ "learning_rate": 7.592339261285911e-06,
71
+ "loss": 1.4456,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.17099863201094392,
76
+ "grad_norm": 18.560701370239258,
77
+ "learning_rate": 8.44733242134063e-06,
78
+ "loss": 1.3092,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.1880984952120383,
83
+ "grad_norm": 29.902738571166992,
84
+ "learning_rate": 9.302325581395349e-06,
85
+ "loss": 1.2648,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.2051983584131327,
90
+ "grad_norm": 27.373172760009766,
91
+ "learning_rate": 1.0157318741450068e-05,
92
+ "loss": 1.5046,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.22229822161422708,
97
+ "grad_norm": 50.6275749206543,
98
+ "learning_rate": 1.1012311901504789e-05,
99
+ "loss": 1.3096,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.2393980848153215,
104
+ "grad_norm": 16.131093978881836,
105
+ "learning_rate": 1.1867305061559508e-05,
106
+ "loss": 1.1689,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.25649794801641584,
111
+ "grad_norm": 30.61119270324707,
112
+ "learning_rate": 1.2722298221614229e-05,
113
+ "loss": 1.1555,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.27359781121751026,
118
+ "grad_norm": 12.447290420532227,
119
+ "learning_rate": 1.3577291381668946e-05,
120
+ "loss": 0.9556,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.29069767441860467,
125
+ "grad_norm": 7.643873691558838,
126
+ "learning_rate": 1.4432284541723667e-05,
127
+ "loss": 1.0063,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.307797537619699,
132
+ "grad_norm": 21.5053768157959,
133
+ "learning_rate": 1.5287277701778386e-05,
134
+ "loss": 1.0971,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.32489740082079344,
139
+ "grad_norm": 26.181028366088867,
140
+ "learning_rate": 1.6142270861833107e-05,
141
+ "loss": 0.8747,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.34199726402188785,
146
+ "grad_norm": Infinity,
147
+ "learning_rate": 1.6963064295485636e-05,
148
+ "loss": 0.899,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 0.3590971272229822,
153
+ "grad_norm": 13.22681713104248,
154
+ "learning_rate": 1.7818057455540357e-05,
155
+ "loss": 0.844,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 0.3761969904240766,
160
+ "grad_norm": 20.26528549194336,
161
+ "learning_rate": 1.8673050615595075e-05,
162
+ "loss": 0.9843,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 0.393296853625171,
167
+ "grad_norm": 13.295209884643555,
168
+ "learning_rate": 1.9528043775649796e-05,
169
+ "loss": 0.8477,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 0.4103967168262654,
174
+ "grad_norm": 17.751012802124023,
175
+ "learning_rate": 2.0383036935704516e-05,
176
+ "loss": 0.9479,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 0.4274965800273598,
181
+ "grad_norm": 7.726760387420654,
182
+ "learning_rate": 2.1238030095759234e-05,
183
+ "loss": 0.8116,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 0.44459644322845415,
188
+ "grad_norm": 14.156341552734375,
189
+ "learning_rate": 2.2093023255813955e-05,
190
+ "loss": 0.8521,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 0.46169630642954856,
195
+ "grad_norm": 39.080848693847656,
196
+ "learning_rate": 2.2948016415868672e-05,
197
+ "loss": 0.9364,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 0.478796169630643,
202
+ "grad_norm": 18.685993194580078,
203
+ "learning_rate": 2.3803009575923393e-05,
204
+ "loss": 0.6069,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 0.49589603283173733,
209
+ "grad_norm": 22.621519088745117,
210
+ "learning_rate": 2.4658002735978114e-05,
211
+ "loss": 0.6538,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 0.5129958960328317,
216
+ "grad_norm": 31.690587997436523,
217
+ "learning_rate": 2.5512995896032832e-05,
218
+ "loss": 0.7562,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 0.5300957592339262,
223
+ "grad_norm": 7.775012016296387,
224
+ "learning_rate": 2.6367989056087556e-05,
225
+ "loss": 0.692,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 0.5471956224350205,
230
+ "grad_norm": 24.86013412475586,
231
+ "learning_rate": 2.7222982216142274e-05,
232
+ "loss": 0.7072,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 0.5642954856361149,
237
+ "grad_norm": 6.716433525085449,
238
+ "learning_rate": 2.807797537619699e-05,
239
+ "loss": 0.6732,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 0.5813953488372093,
244
+ "grad_norm": 14.250188827514648,
245
+ "learning_rate": 2.893296853625171e-05,
246
+ "loss": 0.8366,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 0.5984952120383037,
251
+ "grad_norm": 17.7914981842041,
252
+ "learning_rate": 2.9787961696306433e-05,
253
+ "loss": 0.7953,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 0.615595075239398,
258
+ "grad_norm": 13.2632417678833,
259
+ "learning_rate": 3.064295485636115e-05,
260
+ "loss": 0.7905,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 0.6326949384404925,
265
+ "grad_norm": 34.731689453125,
266
+ "learning_rate": 3.149794801641587e-05,
267
+ "loss": 0.741,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 0.6497948016415869,
272
+ "grad_norm": 23.923887252807617,
273
+ "learning_rate": 3.235294117647059e-05,
274
+ "loss": 0.7366,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 0.6668946648426812,
279
+ "grad_norm": 12.461081504821777,
280
+ "learning_rate": 3.3207934336525306e-05,
281
+ "loss": 0.8837,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 0.6839945280437757,
286
+ "grad_norm": 14.00641918182373,
287
+ "learning_rate": 3.406292749658003e-05,
288
+ "loss": 0.6269,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 0.70109439124487,
293
+ "grad_norm": 23.20526123046875,
294
+ "learning_rate": 3.491792065663475e-05,
295
+ "loss": 0.8368,
296
+ "step": 1025
297
+ },
298
+ {
299
+ "epoch": 0.7181942544459644,
300
+ "grad_norm": 29.71381187438965,
301
+ "learning_rate": 3.577291381668947e-05,
302
+ "loss": 0.5834,
303
+ "step": 1050
304
+ },
305
+ {
306
+ "epoch": 0.7352941176470589,
307
+ "grad_norm": 13.521023750305176,
308
+ "learning_rate": 3.662790697674418e-05,
309
+ "loss": 0.682,
310
+ "step": 1075
311
+ },
312
+ {
313
+ "epoch": 0.7523939808481532,
314
+ "grad_norm": 18.163925170898438,
315
+ "learning_rate": 3.748290013679891e-05,
316
+ "loss": 0.7037,
317
+ "step": 1100
318
+ },
319
+ {
320
+ "epoch": 0.7694938440492476,
321
+ "grad_norm": 4.578676223754883,
322
+ "learning_rate": 3.8337893296853625e-05,
323
+ "loss": 0.6831,
324
+ "step": 1125
325
+ },
326
+ {
327
+ "epoch": 0.786593707250342,
328
+ "grad_norm": 37.317405700683594,
329
+ "learning_rate": 3.9192886456908346e-05,
330
+ "loss": 0.6257,
331
+ "step": 1150
332
+ },
333
+ {
334
+ "epoch": 0.8036935704514364,
335
+ "grad_norm": 19.161466598510742,
336
+ "learning_rate": 4.004787961696307e-05,
337
+ "loss": 0.6827,
338
+ "step": 1175
339
+ },
340
+ {
341
+ "epoch": 0.8207934336525308,
342
+ "grad_norm": 12.673524856567383,
343
+ "learning_rate": 4.090287277701779e-05,
344
+ "loss": 0.7644,
345
+ "step": 1200
346
+ },
347
+ {
348
+ "epoch": 0.8378932968536251,
349
+ "grad_norm": 19.195293426513672,
350
+ "learning_rate": 4.17578659370725e-05,
351
+ "loss": 0.8797,
352
+ "step": 1225
353
+ },
354
+ {
355
+ "epoch": 0.8549931600547196,
356
+ "grad_norm": 12.568032264709473,
357
+ "learning_rate": 4.261285909712722e-05,
358
+ "loss": 0.8057,
359
+ "step": 1250
360
+ },
361
+ {
362
+ "epoch": 0.872093023255814,
363
+ "grad_norm": 6.0890984535217285,
364
+ "learning_rate": 4.3467852257181944e-05,
365
+ "loss": 0.7327,
366
+ "step": 1275
367
+ },
368
+ {
369
+ "epoch": 0.8891928864569083,
370
+ "grad_norm": 17.399168014526367,
371
+ "learning_rate": 4.4322845417236665e-05,
372
+ "loss": 0.6994,
373
+ "step": 1300
374
+ },
375
+ {
376
+ "epoch": 0.9062927496580028,
377
+ "grad_norm": 22.242332458496094,
378
+ "learning_rate": 4.517783857729138e-05,
379
+ "loss": 0.7407,
380
+ "step": 1325
381
+ },
382
+ {
383
+ "epoch": 0.9233926128590971,
384
+ "grad_norm": 34.255836486816406,
385
+ "learning_rate": 4.6032831737346106e-05,
386
+ "loss": 0.7177,
387
+ "step": 1350
388
+ },
389
+ {
390
+ "epoch": 0.9404924760601915,
391
+ "grad_norm": 32.492916107177734,
392
+ "learning_rate": 4.688782489740082e-05,
393
+ "loss": 0.6823,
394
+ "step": 1375
395
+ },
396
+ {
397
+ "epoch": 0.957592339261286,
398
+ "grad_norm": 8.38550853729248,
399
+ "learning_rate": 4.774281805745554e-05,
400
+ "loss": 0.6598,
401
+ "step": 1400
402
+ },
403
+ {
404
+ "epoch": 0.9746922024623803,
405
+ "grad_norm": 14.937362670898438,
406
+ "learning_rate": 4.859781121751026e-05,
407
+ "loss": 0.6052,
408
+ "step": 1425
409
+ },
410
+ {
411
+ "epoch": 0.9917920656634747,
412
+ "grad_norm": 12.976387023925781,
413
+ "learning_rate": 4.945280437756498e-05,
414
+ "loss": 0.6612,
415
+ "step": 1450
416
+ },
417
+ {
418
+ "epoch": 1.0,
419
+ "eval_accuracy": 0.7494025264595425,
420
+ "eval_f1_macro": 0.4161312340366561,
421
+ "eval_f1_micro": 0.7494025264595425,
422
+ "eval_f1_weighted": 0.7356756793477198,
423
+ "eval_loss": 0.6534971594810486,
424
+ "eval_precision_macro": 0.5527584385412905,
425
+ "eval_precision_micro": 0.7494025264595425,
426
+ "eval_precision_weighted": 0.8030313567462574,
427
+ "eval_recall_macro": 0.42879797708152634,
428
+ "eval_recall_micro": 0.7494025264595425,
429
+ "eval_recall_weighted": 0.7494025264595425,
430
+ "eval_runtime": 25.4169,
431
+ "eval_samples_per_second": 115.238,
432
+ "eval_steps_per_second": 7.239,
433
+ "step": 1462
434
+ },
435
+ {
436
+ "epoch": 1.008891928864569,
437
+ "grad_norm": 9.59922981262207,
438
+ "learning_rate": 4.996580027359781e-05,
439
+ "loss": 0.5404,
440
+ "step": 1475
441
+ },
442
+ {
443
+ "epoch": 1.0259917920656634,
444
+ "grad_norm": 8.981080055236816,
445
+ "learning_rate": 4.9870801033591734e-05,
446
+ "loss": 0.6632,
447
+ "step": 1500
448
+ },
449
+ {
450
+ "epoch": 1.043091655266758,
451
+ "grad_norm": 19.810388565063477,
452
+ "learning_rate": 4.977580179358565e-05,
453
+ "loss": 0.7257,
454
+ "step": 1525
455
+ },
456
+ {
457
+ "epoch": 1.0601915184678523,
458
+ "grad_norm": 24.40424346923828,
459
+ "learning_rate": 4.9680802553579575e-05,
460
+ "loss": 0.4992,
461
+ "step": 1550
462
+ },
463
+ {
464
+ "epoch": 1.0772913816689467,
465
+ "grad_norm": 17.77897071838379,
466
+ "learning_rate": 4.958580331357349e-05,
467
+ "loss": 0.7102,
468
+ "step": 1575
469
+ },
470
+ {
471
+ "epoch": 1.094391244870041,
472
+ "grad_norm": 9.230584144592285,
473
+ "learning_rate": 4.9490804073567415e-05,
474
+ "loss": 0.5452,
475
+ "step": 1600
476
+ },
477
+ {
478
+ "epoch": 1.1114911080711354,
479
+ "grad_norm": 9.320358276367188,
480
+ "learning_rate": 4.939580483356133e-05,
481
+ "loss": 0.6861,
482
+ "step": 1625
483
+ },
484
+ {
485
+ "epoch": 1.1285909712722297,
486
+ "grad_norm": 19.349388122558594,
487
+ "learning_rate": 4.9300805593555256e-05,
488
+ "loss": 0.6399,
489
+ "step": 1650
490
+ },
491
+ {
492
+ "epoch": 1.1456908344733243,
493
+ "grad_norm": 8.893238067626953,
494
+ "learning_rate": 4.920580635354917e-05,
495
+ "loss": 0.541,
496
+ "step": 1675
497
+ },
498
+ {
499
+ "epoch": 1.1627906976744187,
500
+ "grad_norm": 7.8030219078063965,
501
+ "learning_rate": 4.9110807113543096e-05,
502
+ "loss": 0.6742,
503
+ "step": 1700
504
+ },
505
+ {
506
+ "epoch": 1.179890560875513,
507
+ "grad_norm": 30.97956657409668,
508
+ "learning_rate": 4.901580787353701e-05,
509
+ "loss": 0.5536,
510
+ "step": 1725
511
+ },
512
+ {
513
+ "epoch": 1.1969904240766074,
514
+ "grad_norm": 28.708515167236328,
515
+ "learning_rate": 4.892080863353094e-05,
516
+ "loss": 0.5912,
517
+ "step": 1750
518
+ },
519
+ {
520
+ "epoch": 1.2140902872777017,
521
+ "grad_norm": 8.513506889343262,
522
+ "learning_rate": 4.8825809393524854e-05,
523
+ "loss": 0.6994,
524
+ "step": 1775
525
+ },
526
+ {
527
+ "epoch": 1.231190150478796,
528
+ "grad_norm": 3.460026741027832,
529
+ "learning_rate": 4.873081015351878e-05,
530
+ "loss": 0.5413,
531
+ "step": 1800
532
+ },
533
+ {
534
+ "epoch": 1.2482900136798905,
535
+ "grad_norm": 11.066655158996582,
536
+ "learning_rate": 4.8635810913512694e-05,
537
+ "loss": 0.5687,
538
+ "step": 1825
539
+ },
540
+ {
541
+ "epoch": 1.265389876880985,
542
+ "grad_norm": 11.196250915527344,
543
+ "learning_rate": 4.854081167350661e-05,
544
+ "loss": 0.6434,
545
+ "step": 1850
546
+ },
547
+ {
548
+ "epoch": 1.2824897400820794,
549
+ "grad_norm": 6.403533935546875,
550
+ "learning_rate": 4.8445812433500535e-05,
551
+ "loss": 0.6681,
552
+ "step": 1875
553
+ },
554
+ {
555
+ "epoch": 1.2995896032831737,
556
+ "grad_norm": 0.6531491875648499,
557
+ "learning_rate": 4.835081319349445e-05,
558
+ "loss": 0.6018,
559
+ "step": 1900
560
+ },
561
+ {
562
+ "epoch": 1.316689466484268,
563
+ "grad_norm": 17.756439208984375,
564
+ "learning_rate": 4.8255813953488375e-05,
565
+ "loss": 0.6509,
566
+ "step": 1925
567
+ },
568
+ {
569
+ "epoch": 1.3337893296853625,
570
+ "grad_norm": 6.791726589202881,
571
+ "learning_rate": 4.816081471348229e-05,
572
+ "loss": 0.7089,
573
+ "step": 1950
574
+ },
575
+ {
576
+ "epoch": 1.350889192886457,
577
+ "grad_norm": 11.32545280456543,
578
+ "learning_rate": 4.8065815473476216e-05,
579
+ "loss": 0.6242,
580
+ "step": 1975
581
+ },
582
+ {
583
+ "epoch": 1.3679890560875512,
584
+ "grad_norm": 23.671077728271484,
585
+ "learning_rate": 4.797081623347013e-05,
586
+ "loss": 0.7434,
587
+ "step": 2000
588
+ },
589
+ {
590
+ "epoch": 1.3850889192886457,
591
+ "grad_norm": 2.791210174560547,
592
+ "learning_rate": 4.7875816993464056e-05,
593
+ "loss": 0.5504,
594
+ "step": 2025
595
+ },
596
+ {
597
+ "epoch": 1.40218878248974,
598
+ "grad_norm": 8.67063045501709,
599
+ "learning_rate": 4.778081775345797e-05,
600
+ "loss": 0.756,
601
+ "step": 2050
602
+ },
603
+ {
604
+ "epoch": 1.4192886456908345,
605
+ "grad_norm": 3.611431360244751,
606
+ "learning_rate": 4.76858185134519e-05,
607
+ "loss": 0.6977,
608
+ "step": 2075
609
+ },
610
+ {
611
+ "epoch": 1.4363885088919288,
612
+ "grad_norm": 6.853910446166992,
613
+ "learning_rate": 4.7590819273445814e-05,
614
+ "loss": 0.4457,
615
+ "step": 2100
616
+ },
617
+ {
618
+ "epoch": 1.4534883720930232,
619
+ "grad_norm": 5.6227264404296875,
620
+ "learning_rate": 4.749582003343974e-05,
621
+ "loss": 0.5169,
622
+ "step": 2125
623
+ },
624
+ {
625
+ "epoch": 1.4705882352941178,
626
+ "grad_norm": 10.246155738830566,
627
+ "learning_rate": 4.7400820793433654e-05,
628
+ "loss": 0.7258,
629
+ "step": 2150
630
+ },
631
+ {
632
+ "epoch": 1.487688098495212,
633
+ "grad_norm": 10.109542846679688,
634
+ "learning_rate": 4.730582155342758e-05,
635
+ "loss": 0.5037,
636
+ "step": 2175
637
+ },
638
+ {
639
+ "epoch": 1.5047879616963065,
640
+ "grad_norm": 29.261133193969727,
641
+ "learning_rate": 4.7210822313421495e-05,
642
+ "loss": 0.5408,
643
+ "step": 2200
644
+ },
645
+ {
646
+ "epoch": 1.5218878248974008,
647
+ "grad_norm": 22.082735061645508,
648
+ "learning_rate": 4.711582307341542e-05,
649
+ "loss": 0.6156,
650
+ "step": 2225
651
+ },
652
+ {
653
+ "epoch": 1.5389876880984952,
654
+ "grad_norm": 4.202814102172852,
655
+ "learning_rate": 4.7020823833409335e-05,
656
+ "loss": 0.3967,
657
+ "step": 2250
658
+ },
659
+ {
660
+ "epoch": 1.5560875512995898,
661
+ "grad_norm": 19.054508209228516,
662
+ "learning_rate": 4.692582459340326e-05,
663
+ "loss": 0.6536,
664
+ "step": 2275
665
+ },
666
+ {
667
+ "epoch": 1.573187414500684,
668
+ "grad_norm": 10.150508880615234,
669
+ "learning_rate": 4.6830825353397176e-05,
670
+ "loss": 0.6868,
671
+ "step": 2300
672
+ },
673
+ {
674
+ "epoch": 1.5902872777017785,
675
+ "grad_norm": 9.790059089660645,
676
+ "learning_rate": 4.673582611339109e-05,
677
+ "loss": 0.5342,
678
+ "step": 2325
679
+ },
680
+ {
681
+ "epoch": 1.6073871409028728,
682
+ "grad_norm": 20.408973693847656,
683
+ "learning_rate": 4.664082687338501e-05,
684
+ "loss": 0.6302,
685
+ "step": 2350
686
+ },
687
+ {
688
+ "epoch": 1.6244870041039672,
689
+ "grad_norm": 3.9431777000427246,
690
+ "learning_rate": 4.6545827633378933e-05,
691
+ "loss": 0.5242,
692
+ "step": 2375
693
+ },
694
+ {
695
+ "epoch": 1.6415868673050615,
696
+ "grad_norm": 17.057270050048828,
697
+ "learning_rate": 4.645082839337285e-05,
698
+ "loss": 0.5606,
699
+ "step": 2400
700
+ },
701
+ {
702
+ "epoch": 1.658686730506156,
703
+ "grad_norm": 15.276410102844238,
704
+ "learning_rate": 4.6355829153366774e-05,
705
+ "loss": 0.6188,
706
+ "step": 2425
707
+ },
708
+ {
709
+ "epoch": 1.6757865937072505,
710
+ "grad_norm": 4.771947383880615,
711
+ "learning_rate": 4.626082991336069e-05,
712
+ "loss": 0.4203,
713
+ "step": 2450
714
+ },
715
+ {
716
+ "epoch": 1.6928864569083446,
717
+ "grad_norm": 11.805036544799805,
718
+ "learning_rate": 4.6165830673354615e-05,
719
+ "loss": 0.6164,
720
+ "step": 2475
721
+ },
722
+ {
723
+ "epoch": 1.7099863201094392,
724
+ "grad_norm": 15.42662239074707,
725
+ "learning_rate": 4.607463140294878e-05,
726
+ "loss": 0.561,
727
+ "step": 2500
728
+ },
729
+ {
730
+ "epoch": 1.7270861833105335,
731
+ "grad_norm": 27.376279830932617,
732
+ "learning_rate": 4.59796321629427e-05,
733
+ "loss": 0.5232,
734
+ "step": 2525
735
+ },
736
+ {
737
+ "epoch": 1.744186046511628,
738
+ "grad_norm": 14.586313247680664,
739
+ "learning_rate": 4.5888432892536865e-05,
740
+ "loss": 0.5013,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 1.7612859097127223,
745
+ "grad_norm": 4.355943202972412,
746
+ "learning_rate": 4.579343365253078e-05,
747
+ "loss": 0.623,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 1.7783857729138166,
752
+ "grad_norm": 14.60937213897705,
753
+ "learning_rate": 4.5698434412524706e-05,
754
+ "loss": 0.5106,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 1.7954856361149112,
759
+ "grad_norm": 11.220321655273438,
760
+ "learning_rate": 4.560343517251862e-05,
761
+ "loss": 0.4515,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 1.8125854993160053,
766
+ "grad_norm": 10.763425827026367,
767
+ "learning_rate": 4.5508435932512546e-05,
768
+ "loss": 0.5121,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 1.8296853625171,
773
+ "grad_norm": 14.884260177612305,
774
+ "learning_rate": 4.541343669250646e-05,
775
+ "loss": 0.5982,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 1.8467852257181943,
780
+ "grad_norm": 10.640288352966309,
781
+ "learning_rate": 4.531843745250039e-05,
782
+ "loss": 0.4412,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 1.8638850889192886,
787
+ "grad_norm": 4.36832332611084,
788
+ "learning_rate": 4.5223438212494304e-05,
789
+ "loss": 0.57,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 1.8809849521203832,
794
+ "grad_norm": 1.8027430772781372,
795
+ "learning_rate": 4.512843897248822e-05,
796
+ "loss": 0.5363,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 1.8980848153214773,
801
+ "grad_norm": 3.5775504112243652,
802
+ "learning_rate": 4.5033439732482144e-05,
803
+ "loss": 0.433,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 1.915184678522572,
808
+ "grad_norm": 6.087740421295166,
809
+ "learning_rate": 4.493844049247606e-05,
810
+ "loss": 0.505,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 1.9322845417236663,
815
+ "grad_norm": 10.784486770629883,
816
+ "learning_rate": 4.484344125246998e-05,
817
+ "loss": 0.54,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 1.9493844049247606,
822
+ "grad_norm": 16.345617294311523,
823
+ "learning_rate": 4.47484420124639e-05,
824
+ "loss": 0.5492,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 1.966484268125855,
829
+ "grad_norm": 8.161157608032227,
830
+ "learning_rate": 4.465344277245782e-05,
831
+ "loss": 0.547,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 1.9835841313269493,
836
+ "grad_norm": 0.8972734808921814,
837
+ "learning_rate": 4.455844353245174e-05,
838
+ "loss": 0.5136,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 2.0,
843
+ "eval_accuracy": 0.8866507340389211,
844
+ "eval_f1_macro": 0.6620993762486447,
845
+ "eval_f1_micro": 0.8866507340389211,
846
+ "eval_f1_weighted": 0.8825688205984242,
847
+ "eval_loss": 0.3399985134601593,
848
+ "eval_precision_macro": 0.7711949393609613,
849
+ "eval_precision_micro": 0.8866507340389211,
850
+ "eval_precision_weighted": 0.8986612163367862,
851
+ "eval_recall_macro": 0.637716282980808,
852
+ "eval_recall_micro": 0.8866507340389211,
853
+ "eval_recall_weighted": 0.8866507340389211,
854
+ "eval_runtime": 25.8144,
855
+ "eval_samples_per_second": 113.464,
856
+ "eval_steps_per_second": 7.128,
857
+ "step": 2924
858
+ },
859
+ {
860
+ "epoch": 2.000683994528044,
861
+ "grad_norm": 13.184772491455078,
862
+ "learning_rate": 4.446344429244566e-05,
863
+ "loss": 0.5396,
864
+ "step": 2925
865
+ },
866
+ {
867
+ "epoch": 2.017783857729138,
868
+ "grad_norm": 25.030324935913086,
869
+ "learning_rate": 4.436844505243958e-05,
870
+ "loss": 0.4869,
871
+ "step": 2950
872
+ },
873
+ {
874
+ "epoch": 2.0348837209302326,
875
+ "grad_norm": 4.769693851470947,
876
+ "learning_rate": 4.42734458124335e-05,
877
+ "loss": 0.6058,
878
+ "step": 2975
879
+ },
880
+ {
881
+ "epoch": 2.0519835841313268,
882
+ "grad_norm": 7.207031726837158,
883
+ "learning_rate": 4.417844657242742e-05,
884
+ "loss": 0.4164,
885
+ "step": 3000
886
+ },
887
+ {
888
+ "epoch": 2.0690834473324213,
889
+ "grad_norm": 8.032303810119629,
890
+ "learning_rate": 4.408344733242134e-05,
891
+ "loss": 0.6109,
892
+ "step": 3025
893
+ },
894
+ {
895
+ "epoch": 2.086183310533516,
896
+ "grad_norm": 9.863656044006348,
897
+ "learning_rate": 4.3988448092415264e-05,
898
+ "loss": 0.5524,
899
+ "step": 3050
900
+ },
901
+ {
902
+ "epoch": 2.10328317373461,
903
+ "grad_norm": 12.57875919342041,
904
+ "learning_rate": 4.389344885240918e-05,
905
+ "loss": 0.5071,
906
+ "step": 3075
907
+ },
908
+ {
909
+ "epoch": 2.1203830369357046,
910
+ "grad_norm": 14.003454208374023,
911
+ "learning_rate": 4.3798449612403104e-05,
912
+ "loss": 0.5256,
913
+ "step": 3100
914
+ },
915
+ {
916
+ "epoch": 2.1374829001367988,
917
+ "grad_norm": 18.200809478759766,
918
+ "learning_rate": 4.370345037239703e-05,
919
+ "loss": 0.5496,
920
+ "step": 3125
921
+ },
922
+ {
923
+ "epoch": 2.1545827633378933,
924
+ "grad_norm": 4.887876033782959,
925
+ "learning_rate": 4.3608451132390945e-05,
926
+ "loss": 0.4347,
927
+ "step": 3150
928
+ },
929
+ {
930
+ "epoch": 2.1716826265389875,
931
+ "grad_norm": 19.693723678588867,
932
+ "learning_rate": 4.351345189238487e-05,
933
+ "loss": 0.5804,
934
+ "step": 3175
935
+ },
936
+ {
937
+ "epoch": 2.188782489740082,
938
+ "grad_norm": 4.4787726402282715,
939
+ "learning_rate": 4.3418452652378786e-05,
940
+ "loss": 0.7788,
941
+ "step": 3200
942
+ },
943
+ {
944
+ "epoch": 2.2058823529411766,
945
+ "grad_norm": 8.199685096740723,
946
+ "learning_rate": 4.33234534123727e-05,
947
+ "loss": 0.5392,
948
+ "step": 3225
949
+ },
950
+ {
951
+ "epoch": 2.2229822161422708,
952
+ "grad_norm": 3.600613594055176,
953
+ "learning_rate": 4.322845417236662e-05,
954
+ "loss": 0.4586,
955
+ "step": 3250
956
+ },
957
+ {
958
+ "epoch": 2.2400820793433653,
959
+ "grad_norm": 1.0569565296173096,
960
+ "learning_rate": 4.313345493236054e-05,
961
+ "loss": 0.4908,
962
+ "step": 3275
963
+ },
964
+ {
965
+ "epoch": 2.2571819425444595,
966
+ "grad_norm": 9.293862342834473,
967
+ "learning_rate": 4.303845569235446e-05,
968
+ "loss": 0.4662,
969
+ "step": 3300
970
+ },
971
+ {
972
+ "epoch": 2.274281805745554,
973
+ "grad_norm": 20.558799743652344,
974
+ "learning_rate": 4.2943456452348384e-05,
975
+ "loss": 0.4093,
976
+ "step": 3325
977
+ },
978
+ {
979
+ "epoch": 2.2913816689466486,
980
+ "grad_norm": 9.981971740722656,
981
+ "learning_rate": 4.28484572123423e-05,
982
+ "loss": 0.5925,
983
+ "step": 3350
984
+ },
985
+ {
986
+ "epoch": 2.3084815321477428,
987
+ "grad_norm": 9.498967170715332,
988
+ "learning_rate": 4.2753457972336224e-05,
989
+ "loss": 0.398,
990
+ "step": 3375
991
+ },
992
+ {
993
+ "epoch": 2.3255813953488373,
994
+ "grad_norm": 15.670341491699219,
995
+ "learning_rate": 4.265845873233014e-05,
996
+ "loss": 0.5653,
997
+ "step": 3400
998
+ },
999
+ {
1000
+ "epoch": 2.3426812585499315,
1001
+ "grad_norm": 2.6220016479492188,
1002
+ "learning_rate": 4.2563459492324065e-05,
1003
+ "loss": 0.2821,
1004
+ "step": 3425
1005
+ },
1006
+ {
1007
+ "epoch": 2.359781121751026,
1008
+ "grad_norm": 4.382536888122559,
1009
+ "learning_rate": 4.246846025231798e-05,
1010
+ "loss": 0.5604,
1011
+ "step": 3450
1012
+ },
1013
+ {
1014
+ "epoch": 2.37688098495212,
1015
+ "grad_norm": 29.831298828125,
1016
+ "learning_rate": 4.2373461012311905e-05,
1017
+ "loss": 0.6324,
1018
+ "step": 3475
1019
+ },
1020
+ {
1021
+ "epoch": 2.3939808481532148,
1022
+ "grad_norm": 11.329654693603516,
1023
+ "learning_rate": 4.227846177230582e-05,
1024
+ "loss": 0.4776,
1025
+ "step": 3500
1026
+ },
1027
+ {
1028
+ "epoch": 2.4110807113543093,
1029
+ "grad_norm": 14.24655818939209,
1030
+ "learning_rate": 4.2183462532299746e-05,
1031
+ "loss": 0.3838,
1032
+ "step": 3525
1033
+ },
1034
+ {
1035
+ "epoch": 2.4281805745554035,
1036
+ "grad_norm": 10.425446510314941,
1037
+ "learning_rate": 4.208846329229366e-05,
1038
+ "loss": 0.4854,
1039
+ "step": 3550
1040
+ },
1041
+ {
1042
+ "epoch": 2.445280437756498,
1043
+ "grad_norm": 12.805408477783203,
1044
+ "learning_rate": 4.1993464052287586e-05,
1045
+ "loss": 0.4318,
1046
+ "step": 3575
1047
+ },
1048
+ {
1049
+ "epoch": 2.462380300957592,
1050
+ "grad_norm": 2.7314155101776123,
1051
+ "learning_rate": 4.18984648122815e-05,
1052
+ "loss": 0.4299,
1053
+ "step": 3600
1054
+ },
1055
+ {
1056
+ "epoch": 2.4794801641586868,
1057
+ "grad_norm": 18.400419235229492,
1058
+ "learning_rate": 4.180346557227543e-05,
1059
+ "loss": 0.51,
1060
+ "step": 3625
1061
+ },
1062
+ {
1063
+ "epoch": 2.496580027359781,
1064
+ "grad_norm": 7.5878214836120605,
1065
+ "learning_rate": 4.1708466332269344e-05,
1066
+ "loss": 0.4057,
1067
+ "step": 3650
1068
+ },
1069
+ {
1070
+ "epoch": 2.5136798905608755,
1071
+ "grad_norm": 6.642499923706055,
1072
+ "learning_rate": 4.161346709226327e-05,
1073
+ "loss": 0.5965,
1074
+ "step": 3675
1075
+ },
1076
+ {
1077
+ "epoch": 2.53077975376197,
1078
+ "grad_norm": 3.847749948501587,
1079
+ "learning_rate": 4.1518467852257184e-05,
1080
+ "loss": 0.5182,
1081
+ "step": 3700
1082
+ },
1083
+ {
1084
+ "epoch": 2.547879616963064,
1085
+ "grad_norm": 11.216913223266602,
1086
+ "learning_rate": 4.14234686122511e-05,
1087
+ "loss": 0.4942,
1088
+ "step": 3725
1089
+ },
1090
+ {
1091
+ "epoch": 2.5649794801641588,
1092
+ "grad_norm": 18.338762283325195,
1093
+ "learning_rate": 4.1328469372245025e-05,
1094
+ "loss": 0.5502,
1095
+ "step": 3750
1096
+ },
1097
+ {
1098
+ "epoch": 2.582079343365253,
1099
+ "grad_norm": 2.1774024963378906,
1100
+ "learning_rate": 4.123347013223894e-05,
1101
+ "loss": 0.557,
1102
+ "step": 3775
1103
+ },
1104
+ {
1105
+ "epoch": 2.5991792065663475,
1106
+ "grad_norm": 11.719930648803711,
1107
+ "learning_rate": 4.1138470892232865e-05,
1108
+ "loss": 0.5172,
1109
+ "step": 3800
1110
+ },
1111
+ {
1112
+ "epoch": 2.616279069767442,
1113
+ "grad_norm": 10.061673164367676,
1114
+ "learning_rate": 4.104347165222678e-05,
1115
+ "loss": 0.4343,
1116
+ "step": 3825
1117
+ },
1118
+ {
1119
+ "epoch": 2.633378932968536,
1120
+ "grad_norm": 8.708327293395996,
1121
+ "learning_rate": 4.0948472412220706e-05,
1122
+ "loss": 0.4235,
1123
+ "step": 3850
1124
+ },
1125
+ {
1126
+ "epoch": 2.650478796169631,
1127
+ "grad_norm": 13.297274589538574,
1128
+ "learning_rate": 4.085347317221462e-05,
1129
+ "loss": 0.4919,
1130
+ "step": 3875
1131
+ },
1132
+ {
1133
+ "epoch": 2.667578659370725,
1134
+ "grad_norm": 1.7467930316925049,
1135
+ "learning_rate": 4.0758473932208547e-05,
1136
+ "loss": 0.5269,
1137
+ "step": 3900
1138
+ },
1139
+ {
1140
+ "epoch": 2.6846785225718195,
1141
+ "grad_norm": 3.1984822750091553,
1142
+ "learning_rate": 4.066347469220246e-05,
1143
+ "loss": 0.5333,
1144
+ "step": 3925
1145
+ },
1146
+ {
1147
+ "epoch": 2.701778385772914,
1148
+ "grad_norm": 4.631045341491699,
1149
+ "learning_rate": 4.056847545219639e-05,
1150
+ "loss": 0.5227,
1151
+ "step": 3950
1152
+ },
1153
+ {
1154
+ "epoch": 2.718878248974008,
1155
+ "grad_norm": 34.19083786010742,
1156
+ "learning_rate": 4.0473476212190304e-05,
1157
+ "loss": 0.5034,
1158
+ "step": 3975
1159
+ },
1160
+ {
1161
+ "epoch": 2.7359781121751023,
1162
+ "grad_norm": 11.861445426940918,
1163
+ "learning_rate": 4.037847697218423e-05,
1164
+ "loss": 0.4182,
1165
+ "step": 4000
1166
+ },
1167
+ {
1168
+ "epoch": 2.753077975376197,
1169
+ "grad_norm": 2.626359224319458,
1170
+ "learning_rate": 4.0283477732178144e-05,
1171
+ "loss": 0.4524,
1172
+ "step": 4025
1173
+ },
1174
+ {
1175
+ "epoch": 2.7701778385772915,
1176
+ "grad_norm": 3.4258358478546143,
1177
+ "learning_rate": 4.018847849217207e-05,
1178
+ "loss": 0.6167,
1179
+ "step": 4050
1180
+ },
1181
+ {
1182
+ "epoch": 2.7872777017783856,
1183
+ "grad_norm": 8.724355697631836,
1184
+ "learning_rate": 4.0093479252165985e-05,
1185
+ "loss": 0.4887,
1186
+ "step": 4075
1187
+ },
1188
+ {
1189
+ "epoch": 2.80437756497948,
1190
+ "grad_norm": 4.958431720733643,
1191
+ "learning_rate": 3.999848001215991e-05,
1192
+ "loss": 0.5661,
1193
+ "step": 4100
1194
+ },
1195
+ {
1196
+ "epoch": 2.8214774281805743,
1197
+ "grad_norm": 22.518157958984375,
1198
+ "learning_rate": 3.9903480772153826e-05,
1199
+ "loss": 0.4875,
1200
+ "step": 4125
1201
+ },
1202
+ {
1203
+ "epoch": 2.838577291381669,
1204
+ "grad_norm": 2.556485176086426,
1205
+ "learning_rate": 3.980848153214774e-05,
1206
+ "loss": 0.4512,
1207
+ "step": 4150
1208
+ },
1209
+ {
1210
+ "epoch": 2.8556771545827635,
1211
+ "grad_norm": 22.274921417236328,
1212
+ "learning_rate": 3.971348229214166e-05,
1213
+ "loss": 0.6026,
1214
+ "step": 4175
1215
+ },
1216
+ {
1217
+ "epoch": 2.8727770177838576,
1218
+ "grad_norm": 11.015833854675293,
1219
+ "learning_rate": 3.961848305213558e-05,
1220
+ "loss": 0.4365,
1221
+ "step": 4200
1222
+ },
1223
+ {
1224
+ "epoch": 2.889876880984952,
1225
+ "grad_norm": 5.808104515075684,
1226
+ "learning_rate": 3.95234838121295e-05,
1227
+ "loss": 0.4042,
1228
+ "step": 4225
1229
+ },
1230
+ {
1231
+ "epoch": 2.9069767441860463,
1232
+ "grad_norm": 7.572930335998535,
1233
+ "learning_rate": 3.9428484572123424e-05,
1234
+ "loss": 0.4788,
1235
+ "step": 4250
1236
+ },
1237
+ {
1238
+ "epoch": 2.924076607387141,
1239
+ "grad_norm": 10.827695846557617,
1240
+ "learning_rate": 3.933348533211734e-05,
1241
+ "loss": 0.4966,
1242
+ "step": 4275
1243
+ },
1244
+ {
1245
+ "epoch": 2.9411764705882355,
1246
+ "grad_norm": 7.229436874389648,
1247
+ "learning_rate": 3.9238486092111264e-05,
1248
+ "loss": 0.4786,
1249
+ "step": 4300
1250
+ },
1251
+ {
1252
+ "epoch": 2.9582763337893296,
1253
+ "grad_norm": 1.738945484161377,
1254
+ "learning_rate": 3.914348685210518e-05,
1255
+ "loss": 0.3888,
1256
+ "step": 4325
1257
+ },
1258
+ {
1259
+ "epoch": 2.975376196990424,
1260
+ "grad_norm": 7.930450439453125,
1261
+ "learning_rate": 3.9048487612099105e-05,
1262
+ "loss": 0.5177,
1263
+ "step": 4350
1264
+ },
1265
+ {
1266
+ "epoch": 2.9924760601915183,
1267
+ "grad_norm": 3.204556465148926,
1268
+ "learning_rate": 3.895348837209303e-05,
1269
+ "loss": 0.5221,
1270
+ "step": 4375
1271
+ },
1272
+ {
1273
+ "epoch": 3.0,
1274
+ "eval_accuracy": 0.8921133492659611,
1275
+ "eval_f1_macro": 0.7248099411197927,
1276
+ "eval_f1_micro": 0.8921133492659611,
1277
+ "eval_f1_weighted": 0.8900566376282183,
1278
+ "eval_loss": 0.30895766615867615,
1279
+ "eval_precision_macro": 0.8455393630579494,
1280
+ "eval_precision_micro": 0.8921133492659611,
1281
+ "eval_precision_weighted": 0.9014059400435374,
1282
+ "eval_recall_macro": 0.6934276077306252,
1283
+ "eval_recall_micro": 0.8921133492659611,
1284
+ "eval_recall_weighted": 0.8921133492659611,
1285
+ "eval_runtime": 25.8185,
1286
+ "eval_samples_per_second": 113.446,
1287
+ "eval_steps_per_second": 7.127,
1288
+ "step": 4386
1289
+ },
1290
+ {
1291
+ "epoch": 3.009575923392613,
1292
+ "grad_norm": 4.8183465003967285,
1293
+ "learning_rate": 3.8858489132086945e-05,
1294
+ "loss": 0.5111,
1295
+ "step": 4400
1296
+ },
1297
+ {
1298
+ "epoch": 3.026675786593707,
1299
+ "grad_norm": 11.976899147033691,
1300
+ "learning_rate": 3.876348989208087e-05,
1301
+ "loss": 0.5971,
1302
+ "step": 4425
1303
+ },
1304
+ {
1305
+ "epoch": 3.0437756497948016,
1306
+ "grad_norm": 7.26112699508667,
1307
+ "learning_rate": 3.8668490652074786e-05,
1308
+ "loss": 0.5008,
1309
+ "step": 4450
1310
+ },
1311
+ {
1312
+ "epoch": 3.060875512995896,
1313
+ "grad_norm": 16.217424392700195,
1314
+ "learning_rate": 3.857349141206871e-05,
1315
+ "loss": 0.542,
1316
+ "step": 4475
1317
+ },
1318
+ {
1319
+ "epoch": 3.0779753761969904,
1320
+ "grad_norm": 4.9412922859191895,
1321
+ "learning_rate": 3.8478492172062626e-05,
1322
+ "loss": 0.4046,
1323
+ "step": 4500
1324
+ },
1325
+ {
1326
+ "epoch": 3.095075239398085,
1327
+ "grad_norm": 14.565752983093262,
1328
+ "learning_rate": 3.838349293205655e-05,
1329
+ "loss": 0.2833,
1330
+ "step": 4525
1331
+ },
1332
+ {
1333
+ "epoch": 3.112175102599179,
1334
+ "grad_norm": 12.100059509277344,
1335
+ "learning_rate": 3.828849369205047e-05,
1336
+ "loss": 0.6031,
1337
+ "step": 4550
1338
+ },
1339
+ {
1340
+ "epoch": 3.1292749658002736,
1341
+ "grad_norm": 7.3751325607299805,
1342
+ "learning_rate": 3.8193494452044384e-05,
1343
+ "loss": 0.3429,
1344
+ "step": 4575
1345
+ },
1346
+ {
1347
+ "epoch": 3.146374829001368,
1348
+ "grad_norm": 1.8577054738998413,
1349
+ "learning_rate": 3.809849521203831e-05,
1350
+ "loss": 0.4626,
1351
+ "step": 4600
1352
+ },
1353
+ {
1354
+ "epoch": 3.1634746922024624,
1355
+ "grad_norm": 21.437606811523438,
1356
+ "learning_rate": 3.8003495972032224e-05,
1357
+ "loss": 0.5272,
1358
+ "step": 4625
1359
+ },
1360
+ {
1361
+ "epoch": 3.180574555403557,
1362
+ "grad_norm": 0.7049314379692078,
1363
+ "learning_rate": 3.790849673202614e-05,
1364
+ "loss": 0.4756,
1365
+ "step": 4650
1366
+ },
1367
+ {
1368
+ "epoch": 3.197674418604651,
1369
+ "grad_norm": 26.833208084106445,
1370
+ "learning_rate": 3.7813497492020065e-05,
1371
+ "loss": 0.4575,
1372
+ "step": 4675
1373
+ },
1374
+ {
1375
+ "epoch": 3.2147742818057456,
1376
+ "grad_norm": 3.7546021938323975,
1377
+ "learning_rate": 3.771849825201398e-05,
1378
+ "loss": 0.3827,
1379
+ "step": 4700
1380
+ },
1381
+ {
1382
+ "epoch": 3.23187414500684,
1383
+ "grad_norm": 0.08676711469888687,
1384
+ "learning_rate": 3.7623499012007905e-05,
1385
+ "loss": 0.5096,
1386
+ "step": 4725
1387
+ },
1388
+ {
1389
+ "epoch": 3.2489740082079344,
1390
+ "grad_norm": 6.28751802444458,
1391
+ "learning_rate": 3.752849977200182e-05,
1392
+ "loss": 0.5496,
1393
+ "step": 4750
1394
+ },
1395
+ {
1396
+ "epoch": 3.266073871409029,
1397
+ "grad_norm": 4.246407985687256,
1398
+ "learning_rate": 3.7433500531995746e-05,
1399
+ "loss": 0.4028,
1400
+ "step": 4775
1401
+ },
1402
+ {
1403
+ "epoch": 3.283173734610123,
1404
+ "grad_norm": 7.93133020401001,
1405
+ "learning_rate": 3.733850129198966e-05,
1406
+ "loss": 0.55,
1407
+ "step": 4800
1408
+ },
1409
+ {
1410
+ "epoch": 3.3002735978112177,
1411
+ "grad_norm": 4.031394004821777,
1412
+ "learning_rate": 3.7243502051983587e-05,
1413
+ "loss": 0.6104,
1414
+ "step": 4825
1415
+ },
1416
+ {
1417
+ "epoch": 3.317373461012312,
1418
+ "grad_norm": 6.97119140625,
1419
+ "learning_rate": 3.7148502811977503e-05,
1420
+ "loss": 0.3561,
1421
+ "step": 4850
1422
+ },
1423
+ {
1424
+ "epoch": 3.3344733242134064,
1425
+ "grad_norm": 4.752348899841309,
1426
+ "learning_rate": 3.705350357197143e-05,
1427
+ "loss": 0.4267,
1428
+ "step": 4875
1429
+ },
1430
+ {
1431
+ "epoch": 3.3515731874145005,
1432
+ "grad_norm": 5.945026397705078,
1433
+ "learning_rate": 3.6958504331965344e-05,
1434
+ "loss": 0.5058,
1435
+ "step": 4900
1436
+ },
1437
+ {
1438
+ "epoch": 3.368673050615595,
1439
+ "grad_norm": 3.1954736709594727,
1440
+ "learning_rate": 3.686350509195927e-05,
1441
+ "loss": 0.4159,
1442
+ "step": 4925
1443
+ },
1444
+ {
1445
+ "epoch": 3.3857729138166897,
1446
+ "grad_norm": 20.407272338867188,
1447
+ "learning_rate": 3.676850585195319e-05,
1448
+ "loss": 0.5054,
1449
+ "step": 4950
1450
+ },
1451
+ {
1452
+ "epoch": 3.402872777017784,
1453
+ "grad_norm": 9.6301851272583,
1454
+ "learning_rate": 3.667350661194711e-05,
1455
+ "loss": 0.4432,
1456
+ "step": 4975
1457
+ },
1458
+ {
1459
+ "epoch": 3.4199726402188784,
1460
+ "grad_norm": 7.080709934234619,
1461
+ "learning_rate": 3.657850737194103e-05,
1462
+ "loss": 0.3967,
1463
+ "step": 5000
1464
+ },
1465
+ {
1466
+ "epoch": 3.4370725034199725,
1467
+ "grad_norm": 6.897956848144531,
1468
+ "learning_rate": 3.648350813193495e-05,
1469
+ "loss": 0.3478,
1470
+ "step": 5025
1471
+ },
1472
+ {
1473
+ "epoch": 3.454172366621067,
1474
+ "grad_norm": 2.1270782947540283,
1475
+ "learning_rate": 3.6388508891928866e-05,
1476
+ "loss": 0.4657,
1477
+ "step": 5050
1478
+ },
1479
+ {
1480
+ "epoch": 3.471272229822161,
1481
+ "grad_norm": 8.693300247192383,
1482
+ "learning_rate": 3.629350965192278e-05,
1483
+ "loss": 0.3595,
1484
+ "step": 5075
1485
+ },
1486
+ {
1487
+ "epoch": 3.488372093023256,
1488
+ "grad_norm": 5.767752170562744,
1489
+ "learning_rate": 3.6198510411916706e-05,
1490
+ "loss": 0.6431,
1491
+ "step": 5100
1492
+ },
1493
+ {
1494
+ "epoch": 3.5054719562243504,
1495
+ "grad_norm": 2.4627044200897217,
1496
+ "learning_rate": 3.610351117191062e-05,
1497
+ "loss": 0.4672,
1498
+ "step": 5125
1499
+ },
1500
+ {
1501
+ "epoch": 3.5225718194254445,
1502
+ "grad_norm": 11.726508140563965,
1503
+ "learning_rate": 3.600851193190455e-05,
1504
+ "loss": 0.5105,
1505
+ "step": 5150
1506
+ },
1507
+ {
1508
+ "epoch": 3.539671682626539,
1509
+ "grad_norm": 12.632149696350098,
1510
+ "learning_rate": 3.5913512691898464e-05,
1511
+ "loss": 0.43,
1512
+ "step": 5175
1513
+ },
1514
+ {
1515
+ "epoch": 3.556771545827633,
1516
+ "grad_norm": 12.43307113647461,
1517
+ "learning_rate": 3.581851345189239e-05,
1518
+ "loss": 0.5033,
1519
+ "step": 5200
1520
+ },
1521
+ {
1522
+ "epoch": 3.573871409028728,
1523
+ "grad_norm": 5.589097499847412,
1524
+ "learning_rate": 3.5723514211886304e-05,
1525
+ "loss": 0.5718,
1526
+ "step": 5225
1527
+ },
1528
+ {
1529
+ "epoch": 3.5909712722298224,
1530
+ "grad_norm": 7.462125778198242,
1531
+ "learning_rate": 3.562851497188023e-05,
1532
+ "loss": 0.5025,
1533
+ "step": 5250
1534
+ },
1535
+ {
1536
+ "epoch": 3.6080711354309165,
1537
+ "grad_norm": 2.099492073059082,
1538
+ "learning_rate": 3.5533515731874145e-05,
1539
+ "loss": 0.3758,
1540
+ "step": 5275
1541
+ },
1542
+ {
1543
+ "epoch": 3.625170998632011,
1544
+ "grad_norm": 6.787677764892578,
1545
+ "learning_rate": 3.543851649186807e-05,
1546
+ "loss": 0.3685,
1547
+ "step": 5300
1548
+ },
1549
+ {
1550
+ "epoch": 3.6422708618331052,
1551
+ "grad_norm": 11.592167854309082,
1552
+ "learning_rate": 3.5343517251861985e-05,
1553
+ "loss": 0.4173,
1554
+ "step": 5325
1555
+ },
1556
+ {
1557
+ "epoch": 3.6593707250342,
1558
+ "grad_norm": 5.881494522094727,
1559
+ "learning_rate": 3.524851801185591e-05,
1560
+ "loss": 0.3484,
1561
+ "step": 5350
1562
+ },
1563
+ {
1564
+ "epoch": 3.6764705882352944,
1565
+ "grad_norm": 5.682451248168945,
1566
+ "learning_rate": 3.5153518771849826e-05,
1567
+ "loss": 0.4324,
1568
+ "step": 5375
1569
+ },
1570
+ {
1571
+ "epoch": 3.6935704514363885,
1572
+ "grad_norm": 0.20855078101158142,
1573
+ "learning_rate": 3.505851953184375e-05,
1574
+ "loss": 0.4147,
1575
+ "step": 5400
1576
+ },
1577
+ {
1578
+ "epoch": 3.7106703146374826,
1579
+ "grad_norm": 8.257678031921387,
1580
+ "learning_rate": 3.4963520291837666e-05,
1581
+ "loss": 0.494,
1582
+ "step": 5425
1583
+ },
1584
+ {
1585
+ "epoch": 3.7277701778385772,
1586
+ "grad_norm": 9.025782585144043,
1587
+ "learning_rate": 3.486852105183159e-05,
1588
+ "loss": 0.4538,
1589
+ "step": 5450
1590
+ },
1591
+ {
1592
+ "epoch": 3.744870041039672,
1593
+ "grad_norm": 8.485616683959961,
1594
+ "learning_rate": 3.477352181182551e-05,
1595
+ "loss": 0.4405,
1596
+ "step": 5475
1597
+ },
1598
+ {
1599
+ "epoch": 3.761969904240766,
1600
+ "grad_norm": 7.031386375427246,
1601
+ "learning_rate": 3.467852257181943e-05,
1602
+ "loss": 0.5225,
1603
+ "step": 5500
1604
+ },
1605
+ {
1606
+ "epoch": 3.7790697674418605,
1607
+ "grad_norm": 6.483025550842285,
1608
+ "learning_rate": 3.458352333181335e-05,
1609
+ "loss": 0.5224,
1610
+ "step": 5525
1611
+ },
1612
+ {
1613
+ "epoch": 3.7961696306429547,
1614
+ "grad_norm": 13.42219066619873,
1615
+ "learning_rate": 3.4488524091807264e-05,
1616
+ "loss": 0.4022,
1617
+ "step": 5550
1618
+ },
1619
+ {
1620
+ "epoch": 3.8132694938440492,
1621
+ "grad_norm": 11.677425384521484,
1622
+ "learning_rate": 3.439352485180119e-05,
1623
+ "loss": 0.4221,
1624
+ "step": 5575
1625
+ },
1626
+ {
1627
+ "epoch": 3.830369357045144,
1628
+ "grad_norm": 83.29227447509766,
1629
+ "learning_rate": 3.4298525611795105e-05,
1630
+ "loss": 0.4575,
1631
+ "step": 5600
1632
+ },
1633
+ {
1634
+ "epoch": 3.847469220246238,
1635
+ "grad_norm": 13.023981094360352,
1636
+ "learning_rate": 3.420352637178903e-05,
1637
+ "loss": 0.4159,
1638
+ "step": 5625
1639
+ },
1640
+ {
1641
+ "epoch": 3.8645690834473325,
1642
+ "grad_norm": 3.78824520111084,
1643
+ "learning_rate": 3.4108527131782945e-05,
1644
+ "loss": 0.3825,
1645
+ "step": 5650
1646
+ },
1647
+ {
1648
+ "epoch": 3.8816689466484267,
1649
+ "grad_norm": 23.089439392089844,
1650
+ "learning_rate": 3.401352789177687e-05,
1651
+ "loss": 0.4796,
1652
+ "step": 5675
1653
+ },
1654
+ {
1655
+ "epoch": 3.8987688098495212,
1656
+ "grad_norm": 8.51207160949707,
1657
+ "learning_rate": 3.3918528651770786e-05,
1658
+ "loss": 0.389,
1659
+ "step": 5700
1660
+ },
1661
+ {
1662
+ "epoch": 3.915868673050616,
1663
+ "grad_norm": 25.023662567138672,
1664
+ "learning_rate": 3.382352941176471e-05,
1665
+ "loss": 0.4779,
1666
+ "step": 5725
1667
+ },
1668
+ {
1669
+ "epoch": 3.93296853625171,
1670
+ "grad_norm": 10.165434837341309,
1671
+ "learning_rate": 3.3728530171758627e-05,
1672
+ "loss": 0.4599,
1673
+ "step": 5750
1674
+ },
1675
+ {
1676
+ "epoch": 3.9500683994528045,
1677
+ "grad_norm": 18.18007469177246,
1678
+ "learning_rate": 3.363353093175255e-05,
1679
+ "loss": 0.4497,
1680
+ "step": 5775
1681
+ },
1682
+ {
1683
+ "epoch": 3.9671682626538987,
1684
+ "grad_norm": 4.1317315101623535,
1685
+ "learning_rate": 3.353853169174647e-05,
1686
+ "loss": 0.3715,
1687
+ "step": 5800
1688
+ },
1689
+ {
1690
+ "epoch": 3.9842681258549932,
1691
+ "grad_norm": 10.207343101501465,
1692
+ "learning_rate": 3.344353245174039e-05,
1693
+ "loss": 0.3945,
1694
+ "step": 5825
1695
+ },
1696
+ {
1697
+ "epoch": 4.0,
1698
+ "eval_accuracy": 0.899283031751451,
1699
+ "eval_f1_macro": 0.7167734728855646,
1700
+ "eval_f1_micro": 0.899283031751451,
1701
+ "eval_f1_weighted": 0.8949067494674995,
1702
+ "eval_loss": 0.2918279469013214,
1703
+ "eval_precision_macro": 0.8481396171573297,
1704
+ "eval_precision_micro": 0.899283031751451,
1705
+ "eval_precision_weighted": 0.9104794727753962,
1706
+ "eval_recall_macro": 0.677056197036938,
1707
+ "eval_recall_micro": 0.899283031751451,
1708
+ "eval_recall_weighted": 0.899283031751451,
1709
+ "eval_runtime": 26.2772,
1710
+ "eval_samples_per_second": 111.465,
1711
+ "eval_steps_per_second": 7.002,
1712
+ "step": 5848
1713
+ },
1714
+ {
1715
+ "epoch": 4.001367989056088,
1716
+ "grad_norm": 9.241109848022461,
1717
+ "learning_rate": 3.334853321173431e-05,
1718
+ "loss": 0.3062,
1719
+ "step": 5850
1720
+ },
1721
+ {
1722
+ "epoch": 4.0184678522571815,
1723
+ "grad_norm": 2.8735108375549316,
1724
+ "learning_rate": 3.325353397172823e-05,
1725
+ "loss": 0.3876,
1726
+ "step": 5875
1727
+ },
1728
+ {
1729
+ "epoch": 4.035567715458276,
1730
+ "grad_norm": 7.347777843475342,
1731
+ "learning_rate": 3.315853473172215e-05,
1732
+ "loss": 0.5103,
1733
+ "step": 5900
1734
+ },
1735
+ {
1736
+ "epoch": 4.052667578659371,
1737
+ "grad_norm": 17.826900482177734,
1738
+ "learning_rate": 3.306353549171607e-05,
1739
+ "loss": 0.4089,
1740
+ "step": 5925
1741
+ },
1742
+ {
1743
+ "epoch": 4.069767441860465,
1744
+ "grad_norm": 4.9833292961120605,
1745
+ "learning_rate": 3.296853625170999e-05,
1746
+ "loss": 0.4656,
1747
+ "step": 5950
1748
+ },
1749
+ {
1750
+ "epoch": 4.08686730506156,
1751
+ "grad_norm": 9.229751586914062,
1752
+ "learning_rate": 3.2873537011703906e-05,
1753
+ "loss": 0.3591,
1754
+ "step": 5975
1755
+ },
1756
+ {
1757
+ "epoch": 4.1039671682626535,
1758
+ "grad_norm": 0.23064318299293518,
1759
+ "learning_rate": 3.277853777169782e-05,
1760
+ "loss": 0.4878,
1761
+ "step": 6000
1762
+ },
1763
+ {
1764
+ "epoch": 4.121067031463748,
1765
+ "grad_norm": 23.045246124267578,
1766
+ "learning_rate": 3.2683538531691746e-05,
1767
+ "loss": 0.3487,
1768
+ "step": 6025
1769
+ },
1770
+ {
1771
+ "epoch": 4.138166894664843,
1772
+ "grad_norm": 8.428695678710938,
1773
+ "learning_rate": 3.258853929168566e-05,
1774
+ "loss": 0.3685,
1775
+ "step": 6050
1776
+ },
1777
+ {
1778
+ "epoch": 4.155266757865937,
1779
+ "grad_norm": 23.741397857666016,
1780
+ "learning_rate": 3.249354005167959e-05,
1781
+ "loss": 0.4878,
1782
+ "step": 6075
1783
+ },
1784
+ {
1785
+ "epoch": 4.172366621067032,
1786
+ "grad_norm": 4.824987888336182,
1787
+ "learning_rate": 3.2398540811673504e-05,
1788
+ "loss": 0.3667,
1789
+ "step": 6100
1790
+ },
1791
+ {
1792
+ "epoch": 4.1894664842681255,
1793
+ "grad_norm": 5.439081192016602,
1794
+ "learning_rate": 3.230354157166743e-05,
1795
+ "loss": 0.449,
1796
+ "step": 6125
1797
+ },
1798
+ {
1799
+ "epoch": 4.20656634746922,
1800
+ "grad_norm": 4.361891269683838,
1801
+ "learning_rate": 3.2208542331661344e-05,
1802
+ "loss": 0.3391,
1803
+ "step": 6150
1804
+ },
1805
+ {
1806
+ "epoch": 4.223666210670315,
1807
+ "grad_norm": 68.6259994506836,
1808
+ "learning_rate": 3.211354309165527e-05,
1809
+ "loss": 0.4126,
1810
+ "step": 6175
1811
+ },
1812
+ {
1813
+ "epoch": 4.240766073871409,
1814
+ "grad_norm": 5.1873860359191895,
1815
+ "learning_rate": 3.201854385164919e-05,
1816
+ "loss": 0.5427,
1817
+ "step": 6200
1818
+ },
1819
+ {
1820
+ "epoch": 4.257865937072504,
1821
+ "grad_norm": 4.0612359046936035,
1822
+ "learning_rate": 3.192354461164311e-05,
1823
+ "loss": 0.45,
1824
+ "step": 6225
1825
+ },
1826
+ {
1827
+ "epoch": 4.2749658002735975,
1828
+ "grad_norm": 7.9056010246276855,
1829
+ "learning_rate": 3.182854537163703e-05,
1830
+ "loss": 0.3444,
1831
+ "step": 6250
1832
+ },
1833
+ {
1834
+ "epoch": 4.292065663474692,
1835
+ "grad_norm": 13.213556289672852,
1836
+ "learning_rate": 3.173354613163095e-05,
1837
+ "loss": 0.3836,
1838
+ "step": 6275
1839
+ },
1840
+ {
1841
+ "epoch": 4.309165526675787,
1842
+ "grad_norm": 2.609199047088623,
1843
+ "learning_rate": 3.163854689162487e-05,
1844
+ "loss": 0.3189,
1845
+ "step": 6300
1846
+ },
1847
+ {
1848
+ "epoch": 4.326265389876881,
1849
+ "grad_norm": 5.747865676879883,
1850
+ "learning_rate": 3.154354765161879e-05,
1851
+ "loss": 0.4637,
1852
+ "step": 6325
1853
+ },
1854
+ {
1855
+ "epoch": 4.343365253077975,
1856
+ "grad_norm": 11.459451675415039,
1857
+ "learning_rate": 3.144854841161271e-05,
1858
+ "loss": 0.4867,
1859
+ "step": 6350
1860
+ },
1861
+ {
1862
+ "epoch": 4.3604651162790695,
1863
+ "grad_norm": 38.839962005615234,
1864
+ "learning_rate": 3.135354917160663e-05,
1865
+ "loss": 0.3632,
1866
+ "step": 6375
1867
+ },
1868
+ {
1869
+ "epoch": 4.377564979480164,
1870
+ "grad_norm": 16.965190887451172,
1871
+ "learning_rate": 3.125854993160055e-05,
1872
+ "loss": 0.4038,
1873
+ "step": 6400
1874
+ },
1875
+ {
1876
+ "epoch": 4.394664842681259,
1877
+ "grad_norm": 9.547988891601562,
1878
+ "learning_rate": 3.116355069159447e-05,
1879
+ "loss": 0.3246,
1880
+ "step": 6425
1881
+ },
1882
+ {
1883
+ "epoch": 4.411764705882353,
1884
+ "grad_norm": 29.236087799072266,
1885
+ "learning_rate": 3.106855145158839e-05,
1886
+ "loss": 0.4722,
1887
+ "step": 6450
1888
+ },
1889
+ {
1890
+ "epoch": 4.428864569083447,
1891
+ "grad_norm": 9.634954452514648,
1892
+ "learning_rate": 3.0973552211582304e-05,
1893
+ "loss": 0.4658,
1894
+ "step": 6475
1895
+ },
1896
+ {
1897
+ "epoch": 4.4459644322845415,
1898
+ "grad_norm": 14.585807800292969,
1899
+ "learning_rate": 3.087855297157623e-05,
1900
+ "loss": 0.3389,
1901
+ "step": 6500
1902
+ },
1903
+ {
1904
+ "epoch": 4.463064295485636,
1905
+ "grad_norm": 9.657065391540527,
1906
+ "learning_rate": 3.0783553731570145e-05,
1907
+ "loss": 0.3755,
1908
+ "step": 6525
1909
+ },
1910
+ {
1911
+ "epoch": 4.480164158686731,
1912
+ "grad_norm": 10.945516586303711,
1913
+ "learning_rate": 3.068855449156407e-05,
1914
+ "loss": 0.4174,
1915
+ "step": 6550
1916
+ },
1917
+ {
1918
+ "epoch": 4.497264021887825,
1919
+ "grad_norm": 7.025084018707275,
1920
+ "learning_rate": 3.0593555251557986e-05,
1921
+ "loss": 0.5521,
1922
+ "step": 6575
1923
+ },
1924
+ {
1925
+ "epoch": 4.514363885088919,
1926
+ "grad_norm": 4.908510684967041,
1927
+ "learning_rate": 3.049855601155191e-05,
1928
+ "loss": 0.3591,
1929
+ "step": 6600
1930
+ },
1931
+ {
1932
+ "epoch": 4.5314637482900135,
1933
+ "grad_norm": 3.1344048976898193,
1934
+ "learning_rate": 3.0403556771545826e-05,
1935
+ "loss": 0.3548,
1936
+ "step": 6625
1937
+ },
1938
+ {
1939
+ "epoch": 4.548563611491108,
1940
+ "grad_norm": 4.515110015869141,
1941
+ "learning_rate": 3.030855753153975e-05,
1942
+ "loss": 0.4855,
1943
+ "step": 6650
1944
+ },
1945
+ {
1946
+ "epoch": 4.565663474692203,
1947
+ "grad_norm": 7.656958103179932,
1948
+ "learning_rate": 3.0213558291533667e-05,
1949
+ "loss": 0.4507,
1950
+ "step": 6675
1951
+ },
1952
+ {
1953
+ "epoch": 4.582763337893297,
1954
+ "grad_norm": 16.05572509765625,
1955
+ "learning_rate": 3.011855905152759e-05,
1956
+ "loss": 0.481,
1957
+ "step": 6700
1958
+ },
1959
+ {
1960
+ "epoch": 4.599863201094391,
1961
+ "grad_norm": 3.04164457321167,
1962
+ "learning_rate": 3.0023559811521507e-05,
1963
+ "loss": 0.3844,
1964
+ "step": 6725
1965
+ },
1966
+ {
1967
+ "epoch": 4.6169630642954855,
1968
+ "grad_norm": 11.163993835449219,
1969
+ "learning_rate": 2.992856057151543e-05,
1970
+ "loss": 0.4604,
1971
+ "step": 6750
1972
+ },
1973
+ {
1974
+ "epoch": 4.63406292749658,
1975
+ "grad_norm": 2.3913285732269287,
1976
+ "learning_rate": 2.983356133150935e-05,
1977
+ "loss": 0.4375,
1978
+ "step": 6775
1979
+ },
1980
+ {
1981
+ "epoch": 4.651162790697675,
1982
+ "grad_norm": 12.693867683410645,
1983
+ "learning_rate": 2.9738562091503268e-05,
1984
+ "loss": 0.3717,
1985
+ "step": 6800
1986
+ },
1987
+ {
1988
+ "epoch": 4.668262653898768,
1989
+ "grad_norm": 5.616913795471191,
1990
+ "learning_rate": 2.964356285149719e-05,
1991
+ "loss": 0.2993,
1992
+ "step": 6825
1993
+ },
1994
+ {
1995
+ "epoch": 4.685362517099863,
1996
+ "grad_norm": 10.53511905670166,
1997
+ "learning_rate": 2.954856361149111e-05,
1998
+ "loss": 0.4742,
1999
+ "step": 6850
2000
+ },
2001
+ {
2002
+ "epoch": 4.7024623803009575,
2003
+ "grad_norm": 4.093250751495361,
2004
+ "learning_rate": 2.9453564371485032e-05,
2005
+ "loss": 0.3256,
2006
+ "step": 6875
2007
+ },
2008
+ {
2009
+ "epoch": 4.719562243502052,
2010
+ "grad_norm": 12.562928199768066,
2011
+ "learning_rate": 2.935856513147895e-05,
2012
+ "loss": 0.3647,
2013
+ "step": 6900
2014
+ },
2015
+ {
2016
+ "epoch": 4.736662106703147,
2017
+ "grad_norm": 28.220108032226562,
2018
+ "learning_rate": 2.9263565891472873e-05,
2019
+ "loss": 0.4505,
2020
+ "step": 6925
2021
+ },
2022
+ {
2023
+ "epoch": 4.75376196990424,
2024
+ "grad_norm": 11.514904975891113,
2025
+ "learning_rate": 2.916856665146679e-05,
2026
+ "loss": 0.3662,
2027
+ "step": 6950
2028
+ },
2029
+ {
2030
+ "epoch": 4.770861833105335,
2031
+ "grad_norm": 3.687920093536377,
2032
+ "learning_rate": 2.907356741146071e-05,
2033
+ "loss": 0.3526,
2034
+ "step": 6975
2035
+ },
2036
+ {
2037
+ "epoch": 4.7879616963064295,
2038
+ "grad_norm": 7.425134181976318,
2039
+ "learning_rate": 2.897856817145463e-05,
2040
+ "loss": 0.3395,
2041
+ "step": 7000
2042
+ },
2043
+ {
2044
+ "epoch": 4.805061559507524,
2045
+ "grad_norm": 2.660003185272217,
2046
+ "learning_rate": 2.888356893144855e-05,
2047
+ "loss": 0.4889,
2048
+ "step": 7025
2049
+ },
2050
+ {
2051
+ "epoch": 4.822161422708619,
2052
+ "grad_norm": 7.089037895202637,
2053
+ "learning_rate": 2.8788569691442467e-05,
2054
+ "loss": 0.4015,
2055
+ "step": 7050
2056
+ },
2057
+ {
2058
+ "epoch": 4.839261285909712,
2059
+ "grad_norm": 11.357674598693848,
2060
+ "learning_rate": 2.869357045143639e-05,
2061
+ "loss": 0.4199,
2062
+ "step": 7075
2063
+ },
2064
+ {
2065
+ "epoch": 4.856361149110807,
2066
+ "grad_norm": 18.09085464477539,
2067
+ "learning_rate": 2.8598571211430308e-05,
2068
+ "loss": 0.4895,
2069
+ "step": 7100
2070
+ },
2071
+ {
2072
+ "epoch": 4.8734610123119015,
2073
+ "grad_norm": 10.047117233276367,
2074
+ "learning_rate": 2.850357197142423e-05,
2075
+ "loss": 0.4722,
2076
+ "step": 7125
2077
+ },
2078
+ {
2079
+ "epoch": 4.890560875512996,
2080
+ "grad_norm": 21.57419776916504,
2081
+ "learning_rate": 2.840857273141815e-05,
2082
+ "loss": 0.4878,
2083
+ "step": 7150
2084
+ },
2085
+ {
2086
+ "epoch": 4.907660738714091,
2087
+ "grad_norm": 7.3810577392578125,
2088
+ "learning_rate": 2.8313573491412072e-05,
2089
+ "loss": 0.4413,
2090
+ "step": 7175
2091
+ },
2092
+ {
2093
+ "epoch": 4.924760601915184,
2094
+ "grad_norm": 1.9230318069458008,
2095
+ "learning_rate": 2.821857425140599e-05,
2096
+ "loss": 0.3885,
2097
+ "step": 7200
2098
+ },
2099
+ {
2100
+ "epoch": 4.941860465116279,
2101
+ "grad_norm": 6.324495792388916,
2102
+ "learning_rate": 2.812357501139991e-05,
2103
+ "loss": 0.4513,
2104
+ "step": 7225
2105
+ },
2106
+ {
2107
+ "epoch": 4.9589603283173735,
2108
+ "grad_norm": 6.727700710296631,
2109
+ "learning_rate": 2.802857577139383e-05,
2110
+ "loss": 0.4085,
2111
+ "step": 7250
2112
+ },
2113
+ {
2114
+ "epoch": 4.976060191518468,
2115
+ "grad_norm": 7.250006675720215,
2116
+ "learning_rate": 2.793357653138775e-05,
2117
+ "loss": 0.3343,
2118
+ "step": 7275
2119
+ },
2120
+ {
2121
+ "epoch": 4.993160054719562,
2122
+ "grad_norm": 2.118319034576416,
2123
+ "learning_rate": 2.7838577291381667e-05,
2124
+ "loss": 0.4999,
2125
+ "step": 7300
2126
+ },
2127
+ {
2128
+ "epoch": 5.0,
2129
+ "eval_accuracy": 0.8828951860703311,
2130
+ "eval_f1_macro": 0.7464038256987945,
2131
+ "eval_f1_micro": 0.8828951860703311,
2132
+ "eval_f1_weighted": 0.8814097485811152,
2133
+ "eval_loss": 0.3324004113674164,
2134
+ "eval_precision_macro": 0.8694718296335959,
2135
+ "eval_precision_micro": 0.8828951860703311,
2136
+ "eval_precision_weighted": 0.9022243801832591,
2137
+ "eval_recall_macro": 0.7003746302122184,
2138
+ "eval_recall_micro": 0.8828951860703311,
2139
+ "eval_recall_weighted": 0.8828951860703311,
2140
+ "eval_runtime": 25.9346,
2141
+ "eval_samples_per_second": 112.938,
2142
+ "eval_steps_per_second": 7.095,
2143
+ "step": 7310
2144
+ },
2145
+ {
2146
+ "epoch": 5.010259917920656,
2147
+ "grad_norm": 0.7182475328445435,
2148
+ "learning_rate": 2.774357805137559e-05,
2149
+ "loss": 0.4188,
2150
+ "step": 7325
2151
+ },
2152
+ {
2153
+ "epoch": 5.027359781121751,
2154
+ "grad_norm": 1.1450774669647217,
2155
+ "learning_rate": 2.7648578811369507e-05,
2156
+ "loss": 0.394,
2157
+ "step": 7350
2158
+ },
2159
+ {
2160
+ "epoch": 5.0444596443228455,
2161
+ "grad_norm": 5.432520866394043,
2162
+ "learning_rate": 2.755357957136343e-05,
2163
+ "loss": 0.4667,
2164
+ "step": 7375
2165
+ },
2166
+ {
2167
+ "epoch": 5.06155950752394,
2168
+ "grad_norm": 9.457201957702637,
2169
+ "learning_rate": 2.7458580331357355e-05,
2170
+ "loss": 0.3702,
2171
+ "step": 7400
2172
+ },
2173
+ {
2174
+ "epoch": 5.078659370725034,
2175
+ "grad_norm": 4.041426181793213,
2176
+ "learning_rate": 2.736358109135127e-05,
2177
+ "loss": 0.4232,
2178
+ "step": 7425
2179
+ },
2180
+ {
2181
+ "epoch": 5.095759233926128,
2182
+ "grad_norm": 5.445071220397949,
2183
+ "learning_rate": 2.7268581851345192e-05,
2184
+ "loss": 0.5922,
2185
+ "step": 7450
2186
+ },
2187
+ {
2188
+ "epoch": 5.112859097127223,
2189
+ "grad_norm": 21.097936630249023,
2190
+ "learning_rate": 2.717358261133911e-05,
2191
+ "loss": 0.3139,
2192
+ "step": 7475
2193
+ },
2194
+ {
2195
+ "epoch": 5.1299589603283176,
2196
+ "grad_norm": 7.353487968444824,
2197
+ "learning_rate": 2.7078583371333032e-05,
2198
+ "loss": 0.365,
2199
+ "step": 7500
2200
+ },
2201
+ {
2202
+ "epoch": 5.147058823529412,
2203
+ "grad_norm": 0.25362271070480347,
2204
+ "learning_rate": 2.698358413132695e-05,
2205
+ "loss": 0.4082,
2206
+ "step": 7525
2207
+ },
2208
+ {
2209
+ "epoch": 5.164158686730506,
2210
+ "grad_norm": 13.88097858428955,
2211
+ "learning_rate": 2.6888584891320873e-05,
2212
+ "loss": 0.3095,
2213
+ "step": 7550
2214
+ },
2215
+ {
2216
+ "epoch": 5.1812585499316,
2217
+ "grad_norm": 0.10476674884557724,
2218
+ "learning_rate": 2.679358565131479e-05,
2219
+ "loss": 0.4199,
2220
+ "step": 7575
2221
+ },
2222
+ {
2223
+ "epoch": 5.198358413132695,
2224
+ "grad_norm": 15.92738151550293,
2225
+ "learning_rate": 2.6698586411308713e-05,
2226
+ "loss": 0.3436,
2227
+ "step": 7600
2228
+ },
2229
+ {
2230
+ "epoch": 5.2154582763337896,
2231
+ "grad_norm": 3.1535122394561768,
2232
+ "learning_rate": 2.660358717130263e-05,
2233
+ "loss": 0.3798,
2234
+ "step": 7625
2235
+ },
2236
+ {
2237
+ "epoch": 5.232558139534884,
2238
+ "grad_norm": 20.45937728881836,
2239
+ "learning_rate": 2.6508587931296554e-05,
2240
+ "loss": 0.4386,
2241
+ "step": 7650
2242
+ },
2243
+ {
2244
+ "epoch": 5.249658002735978,
2245
+ "grad_norm": 11.347773551940918,
2246
+ "learning_rate": 2.641358869129047e-05,
2247
+ "loss": 0.2972,
2248
+ "step": 7675
2249
+ },
2250
+ {
2251
+ "epoch": 5.266757865937072,
2252
+ "grad_norm": 2.8144240379333496,
2253
+ "learning_rate": 2.631858945128439e-05,
2254
+ "loss": 0.3987,
2255
+ "step": 7700
2256
+ },
2257
+ {
2258
+ "epoch": 5.283857729138167,
2259
+ "grad_norm": 5.629673480987549,
2260
+ "learning_rate": 2.6223590211278308e-05,
2261
+ "loss": 0.3263,
2262
+ "step": 7725
2263
+ },
2264
+ {
2265
+ "epoch": 5.300957592339262,
2266
+ "grad_norm": 2.6866259574890137,
2267
+ "learning_rate": 2.6128590971272232e-05,
2268
+ "loss": 0.3661,
2269
+ "step": 7750
2270
+ },
2271
+ {
2272
+ "epoch": 5.318057455540355,
2273
+ "grad_norm": 14.74667739868164,
2274
+ "learning_rate": 2.603359173126615e-05,
2275
+ "loss": 0.3379,
2276
+ "step": 7775
2277
+ },
2278
+ {
2279
+ "epoch": 5.33515731874145,
2280
+ "grad_norm": 3.7586936950683594,
2281
+ "learning_rate": 2.5938592491260072e-05,
2282
+ "loss": 0.3503,
2283
+ "step": 7800
2284
+ },
2285
+ {
2286
+ "epoch": 5.352257181942544,
2287
+ "grad_norm": 4.488011360168457,
2288
+ "learning_rate": 2.584359325125399e-05,
2289
+ "loss": 0.4632,
2290
+ "step": 7825
2291
+ },
2292
+ {
2293
+ "epoch": 5.369357045143639,
2294
+ "grad_norm": 2.083674430847168,
2295
+ "learning_rate": 2.5748594011247913e-05,
2296
+ "loss": 0.3565,
2297
+ "step": 7850
2298
+ },
2299
+ {
2300
+ "epoch": 5.386456908344734,
2301
+ "grad_norm": 7.482815265655518,
2302
+ "learning_rate": 2.565359477124183e-05,
2303
+ "loss": 0.3507,
2304
+ "step": 7875
2305
+ },
2306
+ {
2307
+ "epoch": 5.403556771545827,
2308
+ "grad_norm": 15.02985954284668,
2309
+ "learning_rate": 2.5558595531235753e-05,
2310
+ "loss": 0.3787,
2311
+ "step": 7900
2312
+ },
2313
+ {
2314
+ "epoch": 5.420656634746922,
2315
+ "grad_norm": 8.766378402709961,
2316
+ "learning_rate": 2.546359629122967e-05,
2317
+ "loss": 0.4566,
2318
+ "step": 7925
2319
+ },
2320
+ {
2321
+ "epoch": 5.437756497948016,
2322
+ "grad_norm": 13.31812858581543,
2323
+ "learning_rate": 2.536859705122359e-05,
2324
+ "loss": 0.4214,
2325
+ "step": 7950
2326
+ },
2327
+ {
2328
+ "epoch": 5.454856361149111,
2329
+ "grad_norm": 3.4309494495391846,
2330
+ "learning_rate": 2.5273597811217514e-05,
2331
+ "loss": 0.4198,
2332
+ "step": 7975
2333
+ },
2334
+ {
2335
+ "epoch": 5.471956224350206,
2336
+ "grad_norm": 13.40803337097168,
2337
+ "learning_rate": 2.517859857121143e-05,
2338
+ "loss": 0.3288,
2339
+ "step": 8000
2340
+ },
2341
+ {
2342
+ "epoch": 5.489056087551299,
2343
+ "grad_norm": 2.447211265563965,
2344
+ "learning_rate": 2.5083599331205355e-05,
2345
+ "loss": 0.3237,
2346
+ "step": 8025
2347
+ },
2348
+ {
2349
+ "epoch": 5.506155950752394,
2350
+ "grad_norm": 3.199465036392212,
2351
+ "learning_rate": 2.498860009119927e-05,
2352
+ "loss": 0.455,
2353
+ "step": 8050
2354
+ },
2355
+ {
2356
+ "epoch": 5.523255813953488,
2357
+ "grad_norm": 20.3509521484375,
2358
+ "learning_rate": 2.4893600851193192e-05,
2359
+ "loss": 0.5065,
2360
+ "step": 8075
2361
+ },
2362
+ {
2363
+ "epoch": 5.540355677154583,
2364
+ "grad_norm": 51.09640884399414,
2365
+ "learning_rate": 2.4798601611187112e-05,
2366
+ "loss": 0.3341,
2367
+ "step": 8100
2368
+ },
2369
+ {
2370
+ "epoch": 5.557455540355678,
2371
+ "grad_norm": 2.58319091796875,
2372
+ "learning_rate": 2.4703602371181032e-05,
2373
+ "loss": 0.3664,
2374
+ "step": 8125
2375
+ },
2376
+ {
2377
+ "epoch": 5.574555403556771,
2378
+ "grad_norm": 8.091569900512695,
2379
+ "learning_rate": 2.460860313117495e-05,
2380
+ "loss": 0.3487,
2381
+ "step": 8150
2382
+ },
2383
+ {
2384
+ "epoch": 5.591655266757866,
2385
+ "grad_norm": 3.2208356857299805,
2386
+ "learning_rate": 2.451360389116887e-05,
2387
+ "loss": 0.4357,
2388
+ "step": 8175
2389
+ },
2390
+ {
2391
+ "epoch": 5.60875512995896,
2392
+ "grad_norm": 7.119288444519043,
2393
+ "learning_rate": 2.441860465116279e-05,
2394
+ "loss": 0.4647,
2395
+ "step": 8200
2396
+ },
2397
+ {
2398
+ "epoch": 5.625854993160055,
2399
+ "grad_norm": 2.9200692176818848,
2400
+ "learning_rate": 2.432360541115671e-05,
2401
+ "loss": 0.4414,
2402
+ "step": 8225
2403
+ },
2404
+ {
2405
+ "epoch": 5.642954856361149,
2406
+ "grad_norm": 17.426734924316406,
2407
+ "learning_rate": 2.422860617115063e-05,
2408
+ "loss": 0.3779,
2409
+ "step": 8250
2410
+ },
2411
+ {
2412
+ "epoch": 5.660054719562243,
2413
+ "grad_norm": 6.002254009246826,
2414
+ "learning_rate": 2.413360693114455e-05,
2415
+ "loss": 0.4117,
2416
+ "step": 8275
2417
+ },
2418
+ {
2419
+ "epoch": 5.677154582763338,
2420
+ "grad_norm": 10.85733413696289,
2421
+ "learning_rate": 2.4038607691138474e-05,
2422
+ "loss": 0.3918,
2423
+ "step": 8300
2424
+ },
2425
+ {
2426
+ "epoch": 5.694254445964432,
2427
+ "grad_norm": 22.411169052124023,
2428
+ "learning_rate": 2.3943608451132395e-05,
2429
+ "loss": 0.3103,
2430
+ "step": 8325
2431
+ },
2432
+ {
2433
+ "epoch": 5.711354309165527,
2434
+ "grad_norm": 14.168360710144043,
2435
+ "learning_rate": 2.384860921112631e-05,
2436
+ "loss": 0.3941,
2437
+ "step": 8350
2438
+ },
2439
+ {
2440
+ "epoch": 5.728454172366621,
2441
+ "grad_norm": 3.775557279586792,
2442
+ "learning_rate": 2.3753609971120232e-05,
2443
+ "loss": 0.4347,
2444
+ "step": 8375
2445
+ },
2446
+ {
2447
+ "epoch": 5.745554035567715,
2448
+ "grad_norm": 2.539484977722168,
2449
+ "learning_rate": 2.3658610731114152e-05,
2450
+ "loss": 0.273,
2451
+ "step": 8400
2452
+ },
2453
+ {
2454
+ "epoch": 5.76265389876881,
2455
+ "grad_norm": 2.8777148723602295,
2456
+ "learning_rate": 2.3563611491108072e-05,
2457
+ "loss": 0.3594,
2458
+ "step": 8425
2459
+ },
2460
+ {
2461
+ "epoch": 5.779753761969904,
2462
+ "grad_norm": 3.811925172805786,
2463
+ "learning_rate": 2.3468612251101993e-05,
2464
+ "loss": 0.3306,
2465
+ "step": 8450
2466
+ },
2467
+ {
2468
+ "epoch": 5.796853625170999,
2469
+ "grad_norm": 10.664482116699219,
2470
+ "learning_rate": 2.3373613011095913e-05,
2471
+ "loss": 0.3625,
2472
+ "step": 8475
2473
+ },
2474
+ {
2475
+ "epoch": 5.813953488372093,
2476
+ "grad_norm": 1.653448462486267,
2477
+ "learning_rate": 2.3278613771089833e-05,
2478
+ "loss": 0.298,
2479
+ "step": 8500
2480
+ },
2481
+ {
2482
+ "epoch": 5.831053351573187,
2483
+ "grad_norm": 3.4559919834136963,
2484
+ "learning_rate": 2.3183614531083753e-05,
2485
+ "loss": 0.355,
2486
+ "step": 8525
2487
+ },
2488
+ {
2489
+ "epoch": 5.848153214774282,
2490
+ "grad_norm": 4.02964973449707,
2491
+ "learning_rate": 2.3088615291077674e-05,
2492
+ "loss": 0.4186,
2493
+ "step": 8550
2494
+ },
2495
+ {
2496
+ "epoch": 5.865253077975376,
2497
+ "grad_norm": 7.618357181549072,
2498
+ "learning_rate": 2.2993616051071594e-05,
2499
+ "loss": 0.5639,
2500
+ "step": 8575
2501
+ },
2502
+ {
2503
+ "epoch": 5.882352941176471,
2504
+ "grad_norm": 4.395501136779785,
2505
+ "learning_rate": 2.289861681106551e-05,
2506
+ "loss": 0.3863,
2507
+ "step": 8600
2508
+ },
2509
+ {
2510
+ "epoch": 5.899452804377565,
2511
+ "grad_norm": 4.189037322998047,
2512
+ "learning_rate": 2.280361757105943e-05,
2513
+ "loss": 0.3274,
2514
+ "step": 8625
2515
+ },
2516
+ {
2517
+ "epoch": 5.916552667578659,
2518
+ "grad_norm": 1.825620412826538,
2519
+ "learning_rate": 2.270861833105335e-05,
2520
+ "loss": 0.4074,
2521
+ "step": 8650
2522
+ },
2523
+ {
2524
+ "epoch": 5.933652530779754,
2525
+ "grad_norm": 2.863705635070801,
2526
+ "learning_rate": 2.2613619091047272e-05,
2527
+ "loss": 0.3973,
2528
+ "step": 8675
2529
+ },
2530
+ {
2531
+ "epoch": 5.950752393980848,
2532
+ "grad_norm": 3.3748676776885986,
2533
+ "learning_rate": 2.2518619851041192e-05,
2534
+ "loss": 0.3318,
2535
+ "step": 8700
2536
+ },
2537
+ {
2538
+ "epoch": 5.967852257181942,
2539
+ "grad_norm": 18.595125198364258,
2540
+ "learning_rate": 2.2423620611035112e-05,
2541
+ "loss": 0.355,
2542
+ "step": 8725
2543
+ },
2544
+ {
2545
+ "epoch": 5.984952120383037,
2546
+ "grad_norm": 8.303060531616211,
2547
+ "learning_rate": 2.2328621371029033e-05,
2548
+ "loss": 0.3325,
2549
+ "step": 8750
2550
+ },
2551
+ {
2552
+ "epoch": 6.0,
2553
+ "eval_accuracy": 0.9235233868214408,
2554
+ "eval_f1_macro": 0.8015916762918697,
2555
+ "eval_f1_micro": 0.9235233868214408,
2556
+ "eval_f1_weighted": 0.922485221004273,
2557
+ "eval_loss": 0.24165508151054382,
2558
+ "eval_precision_macro": 0.8743011626241778,
2559
+ "eval_precision_micro": 0.9235233868214408,
2560
+ "eval_precision_weighted": 0.9276221151549562,
2561
+ "eval_recall_macro": 0.7664921252239268,
2562
+ "eval_recall_micro": 0.9235233868214408,
2563
+ "eval_recall_weighted": 0.9235233868214408,
2564
+ "eval_runtime": 25.9565,
2565
+ "eval_samples_per_second": 112.842,
2566
+ "eval_steps_per_second": 7.089,
2567
+ "step": 8772
2568
+ },
2569
+ {
2570
+ "epoch": 6.002051983584131,
2571
+ "grad_norm": 11.892972946166992,
2572
+ "learning_rate": 2.2233622131022953e-05,
2573
+ "loss": 0.3357,
2574
+ "step": 8775
2575
+ },
2576
+ {
2577
+ "epoch": 6.019151846785226,
2578
+ "grad_norm": 7.985339164733887,
2579
+ "learning_rate": 2.2138622891016873e-05,
2580
+ "loss": 0.4488,
2581
+ "step": 8800
2582
+ },
2583
+ {
2584
+ "epoch": 6.03625170998632,
2585
+ "grad_norm": 6.227213382720947,
2586
+ "learning_rate": 2.2043623651010793e-05,
2587
+ "loss": 0.3902,
2588
+ "step": 8825
2589
+ },
2590
+ {
2591
+ "epoch": 6.053351573187414,
2592
+ "grad_norm": 19.529422760009766,
2593
+ "learning_rate": 2.194862441100471e-05,
2594
+ "loss": 0.4248,
2595
+ "step": 8850
2596
+ },
2597
+ {
2598
+ "epoch": 6.070451436388509,
2599
+ "grad_norm": 19.405685424804688,
2600
+ "learning_rate": 2.185362517099863e-05,
2601
+ "loss": 0.2371,
2602
+ "step": 8875
2603
+ },
2604
+ {
2605
+ "epoch": 6.087551299589603,
2606
+ "grad_norm": 31.696195602416992,
2607
+ "learning_rate": 2.1758625930992554e-05,
2608
+ "loss": 0.511,
2609
+ "step": 8900
2610
+ },
2611
+ {
2612
+ "epoch": 6.104651162790698,
2613
+ "grad_norm": 23.88582992553711,
2614
+ "learning_rate": 2.1663626690986475e-05,
2615
+ "loss": 0.2996,
2616
+ "step": 8925
2617
+ },
2618
+ {
2619
+ "epoch": 6.121751025991792,
2620
+ "grad_norm": 7.8858561515808105,
2621
+ "learning_rate": 2.1568627450980395e-05,
2622
+ "loss": 0.2554,
2623
+ "step": 8950
2624
+ },
2625
+ {
2626
+ "epoch": 6.138850889192886,
2627
+ "grad_norm": 2.5783660411834717,
2628
+ "learning_rate": 2.1473628210974315e-05,
2629
+ "loss": 0.4588,
2630
+ "step": 8975
2631
+ },
2632
+ {
2633
+ "epoch": 6.155950752393981,
2634
+ "grad_norm": 2.6995346546173096,
2635
+ "learning_rate": 2.1378628970968235e-05,
2636
+ "loss": 0.3833,
2637
+ "step": 9000
2638
+ },
2639
+ {
2640
+ "epoch": 6.173050615595075,
2641
+ "grad_norm": 6.289373874664307,
2642
+ "learning_rate": 2.1283629730962156e-05,
2643
+ "loss": 0.4185,
2644
+ "step": 9025
2645
+ },
2646
+ {
2647
+ "epoch": 6.19015047879617,
2648
+ "grad_norm": 6.533155918121338,
2649
+ "learning_rate": 2.1188630490956073e-05,
2650
+ "loss": 0.2384,
2651
+ "step": 9050
2652
+ },
2653
+ {
2654
+ "epoch": 6.207250341997264,
2655
+ "grad_norm": 17.285852432250977,
2656
+ "learning_rate": 2.1093631250949993e-05,
2657
+ "loss": 0.2801,
2658
+ "step": 9075
2659
+ },
2660
+ {
2661
+ "epoch": 6.224350205198358,
2662
+ "grad_norm": 9.399192810058594,
2663
+ "learning_rate": 2.0998632010943913e-05,
2664
+ "loss": 0.5708,
2665
+ "step": 9100
2666
+ },
2667
+ {
2668
+ "epoch": 6.241450068399453,
2669
+ "grad_norm": 12.317584991455078,
2670
+ "learning_rate": 2.0903632770937833e-05,
2671
+ "loss": 0.4415,
2672
+ "step": 9125
2673
+ },
2674
+ {
2675
+ "epoch": 6.258549931600547,
2676
+ "grad_norm": 7.904637336730957,
2677
+ "learning_rate": 2.0808633530931754e-05,
2678
+ "loss": 0.3887,
2679
+ "step": 9150
2680
+ },
2681
+ {
2682
+ "epoch": 6.275649794801642,
2683
+ "grad_norm": 8.931986808776855,
2684
+ "learning_rate": 2.0713634290925674e-05,
2685
+ "loss": 0.4239,
2686
+ "step": 9175
2687
+ },
2688
+ {
2689
+ "epoch": 6.292749658002736,
2690
+ "grad_norm": 7.259034633636475,
2691
+ "learning_rate": 2.0618635050919594e-05,
2692
+ "loss": 0.3405,
2693
+ "step": 9200
2694
+ },
2695
+ {
2696
+ "epoch": 6.30984952120383,
2697
+ "grad_norm": 2.570568084716797,
2698
+ "learning_rate": 2.0523635810913514e-05,
2699
+ "loss": 0.3136,
2700
+ "step": 9225
2701
+ },
2702
+ {
2703
+ "epoch": 6.326949384404925,
2704
+ "grad_norm": 16.150344848632812,
2705
+ "learning_rate": 2.0428636570907435e-05,
2706
+ "loss": 0.4056,
2707
+ "step": 9250
2708
+ },
2709
+ {
2710
+ "epoch": 6.344049247606019,
2711
+ "grad_norm": 15.508628845214844,
2712
+ "learning_rate": 2.0333637330901355e-05,
2713
+ "loss": 0.4423,
2714
+ "step": 9275
2715
+ },
2716
+ {
2717
+ "epoch": 6.361149110807114,
2718
+ "grad_norm": 3.140324354171753,
2719
+ "learning_rate": 2.0238638090895272e-05,
2720
+ "loss": 0.3972,
2721
+ "step": 9300
2722
+ },
2723
+ {
2724
+ "epoch": 6.378248974008208,
2725
+ "grad_norm": 2.326601505279541,
2726
+ "learning_rate": 2.0143638850889192e-05,
2727
+ "loss": 0.3066,
2728
+ "step": 9325
2729
+ },
2730
+ {
2731
+ "epoch": 6.395348837209302,
2732
+ "grad_norm": 2.924823760986328,
2733
+ "learning_rate": 2.0048639610883112e-05,
2734
+ "loss": 0.3117,
2735
+ "step": 9350
2736
+ },
2737
+ {
2738
+ "epoch": 6.412448700410397,
2739
+ "grad_norm": 13.392170906066895,
2740
+ "learning_rate": 1.9953640370877033e-05,
2741
+ "loss": 0.2947,
2742
+ "step": 9375
2743
+ },
2744
+ {
2745
+ "epoch": 6.429548563611491,
2746
+ "grad_norm": 2.736306667327881,
2747
+ "learning_rate": 1.9858641130870953e-05,
2748
+ "loss": 0.2432,
2749
+ "step": 9400
2750
+ },
2751
+ {
2752
+ "epoch": 6.446648426812586,
2753
+ "grad_norm": 2.8124167919158936,
2754
+ "learning_rate": 1.9763641890864873e-05,
2755
+ "loss": 0.4835,
2756
+ "step": 9425
2757
+ },
2758
+ {
2759
+ "epoch": 6.46374829001368,
2760
+ "grad_norm": 8.63110637664795,
2761
+ "learning_rate": 1.9668642650858794e-05,
2762
+ "loss": 0.3444,
2763
+ "step": 9450
2764
+ },
2765
+ {
2766
+ "epoch": 6.480848153214774,
2767
+ "grad_norm": 12.762578010559082,
2768
+ "learning_rate": 1.9573643410852714e-05,
2769
+ "loss": 0.2832,
2770
+ "step": 9475
2771
+ },
2772
+ {
2773
+ "epoch": 6.497948016415869,
2774
+ "grad_norm": 4.230663776397705,
2775
+ "learning_rate": 1.9478644170846634e-05,
2776
+ "loss": 0.3595,
2777
+ "step": 9500
2778
+ },
2779
+ {
2780
+ "epoch": 6.515047879616963,
2781
+ "grad_norm": 2.9571053981781006,
2782
+ "learning_rate": 1.9383644930840554e-05,
2783
+ "loss": 0.3388,
2784
+ "step": 9525
2785
+ },
2786
+ {
2787
+ "epoch": 6.532147742818058,
2788
+ "grad_norm": 3.3192389011383057,
2789
+ "learning_rate": 1.9288645690834475e-05,
2790
+ "loss": 0.4937,
2791
+ "step": 9550
2792
+ },
2793
+ {
2794
+ "epoch": 6.549247606019152,
2795
+ "grad_norm": 4.746058940887451,
2796
+ "learning_rate": 1.9193646450828395e-05,
2797
+ "loss": 0.4209,
2798
+ "step": 9575
2799
+ },
2800
+ {
2801
+ "epoch": 6.566347469220246,
2802
+ "grad_norm": 1.2305563688278198,
2803
+ "learning_rate": 1.9098647210822315e-05,
2804
+ "loss": 0.361,
2805
+ "step": 9600
2806
+ },
2807
+ {
2808
+ "epoch": 6.583447332421341,
2809
+ "grad_norm": 0.1572732776403427,
2810
+ "learning_rate": 1.9003647970816235e-05,
2811
+ "loss": 0.3816,
2812
+ "step": 9625
2813
+ },
2814
+ {
2815
+ "epoch": 6.600547195622435,
2816
+ "grad_norm": 8.715572357177734,
2817
+ "learning_rate": 1.8908648730810156e-05,
2818
+ "loss": 0.3678,
2819
+ "step": 9650
2820
+ },
2821
+ {
2822
+ "epoch": 6.617647058823529,
2823
+ "grad_norm": 4.759888172149658,
2824
+ "learning_rate": 1.8813649490804076e-05,
2825
+ "loss": 0.3702,
2826
+ "step": 9675
2827
+ },
2828
+ {
2829
+ "epoch": 6.634746922024624,
2830
+ "grad_norm": 3.071633815765381,
2831
+ "learning_rate": 1.8718650250797996e-05,
2832
+ "loss": 0.3771,
2833
+ "step": 9700
2834
+ },
2835
+ {
2836
+ "epoch": 6.651846785225718,
2837
+ "grad_norm": 13.45046615600586,
2838
+ "learning_rate": 1.8623651010791917e-05,
2839
+ "loss": 0.5068,
2840
+ "step": 9725
2841
+ },
2842
+ {
2843
+ "epoch": 6.668946648426813,
2844
+ "grad_norm": 15.248006820678711,
2845
+ "learning_rate": 1.8528651770785833e-05,
2846
+ "loss": 0.3403,
2847
+ "step": 9750
2848
+ },
2849
+ {
2850
+ "epoch": 6.686046511627907,
2851
+ "grad_norm": 2.1745738983154297,
2852
+ "learning_rate": 1.8433652530779754e-05,
2853
+ "loss": 0.3092,
2854
+ "step": 9775
2855
+ },
2856
+ {
2857
+ "epoch": 6.703146374829001,
2858
+ "grad_norm": 10.927750587463379,
2859
+ "learning_rate": 1.8338653290773674e-05,
2860
+ "loss": 0.3776,
2861
+ "step": 9800
2862
+ },
2863
+ {
2864
+ "epoch": 6.720246238030096,
2865
+ "grad_norm": 24.628570556640625,
2866
+ "learning_rate": 1.8243654050767594e-05,
2867
+ "loss": 0.4011,
2868
+ "step": 9825
2869
+ },
2870
+ {
2871
+ "epoch": 6.73734610123119,
2872
+ "grad_norm": 1.9271844625473022,
2873
+ "learning_rate": 1.8148654810761515e-05,
2874
+ "loss": 0.3472,
2875
+ "step": 9850
2876
+ },
2877
+ {
2878
+ "epoch": 6.754445964432285,
2879
+ "grad_norm": 11.969741821289062,
2880
+ "learning_rate": 1.8053655570755435e-05,
2881
+ "loss": 0.3599,
2882
+ "step": 9875
2883
+ },
2884
+ {
2885
+ "epoch": 6.771545827633379,
2886
+ "grad_norm": 10.276418685913086,
2887
+ "learning_rate": 1.7958656330749355e-05,
2888
+ "loss": 0.3054,
2889
+ "step": 9900
2890
+ },
2891
+ {
2892
+ "epoch": 6.788645690834473,
2893
+ "grad_norm": 1.4558168649673462,
2894
+ "learning_rate": 1.7863657090743275e-05,
2895
+ "loss": 0.363,
2896
+ "step": 9925
2897
+ },
2898
+ {
2899
+ "epoch": 6.805745554035568,
2900
+ "grad_norm": 6.96836519241333,
2901
+ "learning_rate": 1.7768657850737196e-05,
2902
+ "loss": 0.4066,
2903
+ "step": 9950
2904
+ },
2905
+ {
2906
+ "epoch": 6.822845417236662,
2907
+ "grad_norm": 11.515777587890625,
2908
+ "learning_rate": 1.7673658610731113e-05,
2909
+ "loss": 0.3551,
2910
+ "step": 9975
2911
+ },
2912
+ {
2913
+ "epoch": 6.839945280437757,
2914
+ "grad_norm": 13.478057861328125,
2915
+ "learning_rate": 1.7578659370725033e-05,
2916
+ "loss": 0.3799,
2917
+ "step": 10000
2918
+ },
2919
+ {
2920
+ "epoch": 6.857045143638851,
2921
+ "grad_norm": 11.620177268981934,
2922
+ "learning_rate": 1.7483660130718953e-05,
2923
+ "loss": 0.3375,
2924
+ "step": 10025
2925
+ },
2926
+ {
2927
+ "epoch": 6.874145006839945,
2928
+ "grad_norm": 3.876800537109375,
2929
+ "learning_rate": 1.7388660890712873e-05,
2930
+ "loss": 0.5015,
2931
+ "step": 10050
2932
+ },
2933
+ {
2934
+ "epoch": 6.89124487004104,
2935
+ "grad_norm": 1.9693479537963867,
2936
+ "learning_rate": 1.7293661650706794e-05,
2937
+ "loss": 0.2689,
2938
+ "step": 10075
2939
+ },
2940
+ {
2941
+ "epoch": 6.908344733242134,
2942
+ "grad_norm": 14.633630752563477,
2943
+ "learning_rate": 1.7198662410700717e-05,
2944
+ "loss": 0.3271,
2945
+ "step": 10100
2946
+ },
2947
+ {
2948
+ "epoch": 6.925444596443229,
2949
+ "grad_norm": 4.947651386260986,
2950
+ "learning_rate": 1.7103663170694638e-05,
2951
+ "loss": 0.434,
2952
+ "step": 10125
2953
+ },
2954
+ {
2955
+ "epoch": 6.942544459644322,
2956
+ "grad_norm": 2.0154216289520264,
2957
+ "learning_rate": 1.7008663930688558e-05,
2958
+ "loss": 0.4467,
2959
+ "step": 10150
2960
+ },
2961
+ {
2962
+ "epoch": 6.959644322845417,
2963
+ "grad_norm": 14.244186401367188,
2964
+ "learning_rate": 1.6913664690682475e-05,
2965
+ "loss": 0.3106,
2966
+ "step": 10175
2967
+ },
2968
+ {
2969
+ "epoch": 6.976744186046512,
2970
+ "grad_norm": 4.09621000289917,
2971
+ "learning_rate": 1.6818665450676395e-05,
2972
+ "loss": 0.3063,
2973
+ "step": 10200
2974
+ },
2975
+ {
2976
+ "epoch": 6.993844049247606,
2977
+ "grad_norm": 2.0784831047058105,
2978
+ "learning_rate": 1.6723666210670315e-05,
2979
+ "loss": 0.3515,
2980
+ "step": 10225
2981
+ },
2982
+ {
2983
+ "epoch": 7.0,
2984
+ "eval_accuracy": 0.9238648002731308,
2985
+ "eval_f1_macro": 0.8426039748613977,
2986
+ "eval_f1_micro": 0.9238648002731308,
2987
+ "eval_f1_weighted": 0.9239964103657616,
2988
+ "eval_loss": 0.22421783208847046,
2989
+ "eval_precision_macro": 0.9018755264534066,
2990
+ "eval_precision_micro": 0.9238648002731308,
2991
+ "eval_precision_weighted": 0.9296443693001667,
2992
+ "eval_recall_macro": 0.8036953584993286,
2993
+ "eval_recall_micro": 0.9238648002731308,
2994
+ "eval_recall_weighted": 0.9238648002731308,
2995
+ "eval_runtime": 25.5062,
2996
+ "eval_samples_per_second": 114.835,
2997
+ "eval_steps_per_second": 7.214,
2998
+ "step": 10234
2999
+ },
3000
+ {
3001
+ "epoch": 7.010943912448701,
3002
+ "grad_norm": 4.232228755950928,
3003
+ "learning_rate": 1.6628666970664236e-05,
3004
+ "loss": 0.4255,
3005
+ "step": 10250
3006
+ },
3007
+ {
3008
+ "epoch": 7.028043775649794,
3009
+ "grad_norm": 12.369464874267578,
3010
+ "learning_rate": 1.6533667730658156e-05,
3011
+ "loss": 0.3828,
3012
+ "step": 10275
3013
+ },
3014
+ {
3015
+ "epoch": 7.045143638850889,
3016
+ "grad_norm": 8.454262733459473,
3017
+ "learning_rate": 1.6438668490652076e-05,
3018
+ "loss": 0.287,
3019
+ "step": 10300
3020
+ },
3021
+ {
3022
+ "epoch": 7.062243502051984,
3023
+ "grad_norm": 2.064643144607544,
3024
+ "learning_rate": 1.6343669250645996e-05,
3025
+ "loss": 0.3688,
3026
+ "step": 10325
3027
+ },
3028
+ {
3029
+ "epoch": 7.079343365253078,
3030
+ "grad_norm": 22.918745040893555,
3031
+ "learning_rate": 1.6248670010639917e-05,
3032
+ "loss": 0.3487,
3033
+ "step": 10350
3034
+ },
3035
+ {
3036
+ "epoch": 7.096443228454173,
3037
+ "grad_norm": 3.0556788444519043,
3038
+ "learning_rate": 1.6153670770633837e-05,
3039
+ "loss": 0.2822,
3040
+ "step": 10375
3041
+ },
3042
+ {
3043
+ "epoch": 7.113543091655266,
3044
+ "grad_norm": 11.955765724182129,
3045
+ "learning_rate": 1.6058671530627757e-05,
3046
+ "loss": 0.4273,
3047
+ "step": 10400
3048
+ },
3049
+ {
3050
+ "epoch": 7.130642954856361,
3051
+ "grad_norm": 13.998736381530762,
3052
+ "learning_rate": 1.5963672290621674e-05,
3053
+ "loss": 0.3034,
3054
+ "step": 10425
3055
+ },
3056
+ {
3057
+ "epoch": 7.147742818057456,
3058
+ "grad_norm": 3.8617422580718994,
3059
+ "learning_rate": 1.5868673050615594e-05,
3060
+ "loss": 0.3679,
3061
+ "step": 10450
3062
+ },
3063
+ {
3064
+ "epoch": 7.16484268125855,
3065
+ "grad_norm": 1.1045358180999756,
3066
+ "learning_rate": 1.5773673810609515e-05,
3067
+ "loss": 0.4234,
3068
+ "step": 10475
3069
+ },
3070
+ {
3071
+ "epoch": 7.181942544459645,
3072
+ "grad_norm": 5.257449626922607,
3073
+ "learning_rate": 1.5678674570603435e-05,
3074
+ "loss": 0.3665,
3075
+ "step": 10500
3076
+ },
3077
+ {
3078
+ "epoch": 7.199042407660738,
3079
+ "grad_norm": 7.530872821807861,
3080
+ "learning_rate": 1.5583675330597355e-05,
3081
+ "loss": 0.4602,
3082
+ "step": 10525
3083
+ },
3084
+ {
3085
+ "epoch": 7.216142270861833,
3086
+ "grad_norm": 0.4516567885875702,
3087
+ "learning_rate": 1.5488676090591275e-05,
3088
+ "loss": 0.3864,
3089
+ "step": 10550
3090
+ },
3091
+ {
3092
+ "epoch": 7.233242134062928,
3093
+ "grad_norm": 22.563528060913086,
3094
+ "learning_rate": 1.5393676850585196e-05,
3095
+ "loss": 0.3435,
3096
+ "step": 10575
3097
+ },
3098
+ {
3099
+ "epoch": 7.250341997264022,
3100
+ "grad_norm": 13.331972122192383,
3101
+ "learning_rate": 1.5298677610579116e-05,
3102
+ "loss": 0.2879,
3103
+ "step": 10600
3104
+ },
3105
+ {
3106
+ "epoch": 7.267441860465116,
3107
+ "grad_norm": 9.387733459472656,
3108
+ "learning_rate": 1.5203678370573035e-05,
3109
+ "loss": 0.3333,
3110
+ "step": 10625
3111
+ },
3112
+ {
3113
+ "epoch": 7.2845417236662104,
3114
+ "grad_norm": 7.556676864624023,
3115
+ "learning_rate": 1.5108679130566955e-05,
3116
+ "loss": 0.4042,
3117
+ "step": 10650
3118
+ },
3119
+ {
3120
+ "epoch": 7.301641586867305,
3121
+ "grad_norm": 6.794082164764404,
3122
+ "learning_rate": 1.5013679890560875e-05,
3123
+ "loss": 0.4062,
3124
+ "step": 10675
3125
+ },
3126
+ {
3127
+ "epoch": 7.3187414500684,
3128
+ "grad_norm": 2.1838858127593994,
3129
+ "learning_rate": 1.4918680650554795e-05,
3130
+ "loss": 0.3238,
3131
+ "step": 10700
3132
+ },
3133
+ {
3134
+ "epoch": 7.335841313269494,
3135
+ "grad_norm": 10.469732284545898,
3136
+ "learning_rate": 1.4823681410548717e-05,
3137
+ "loss": 0.4204,
3138
+ "step": 10725
3139
+ },
3140
+ {
3141
+ "epoch": 7.352941176470588,
3142
+ "grad_norm": 2.8655450344085693,
3143
+ "learning_rate": 1.4728682170542638e-05,
3144
+ "loss": 0.3095,
3145
+ "step": 10750
3146
+ },
3147
+ {
3148
+ "epoch": 7.3700410396716824,
3149
+ "grad_norm": 7.8528337478637695,
3150
+ "learning_rate": 1.4633682930536558e-05,
3151
+ "loss": 0.2917,
3152
+ "step": 10775
3153
+ },
3154
+ {
3155
+ "epoch": 7.387140902872777,
3156
+ "grad_norm": 1.5195367336273193,
3157
+ "learning_rate": 1.4538683690530477e-05,
3158
+ "loss": 0.2933,
3159
+ "step": 10800
3160
+ },
3161
+ {
3162
+ "epoch": 7.404240766073872,
3163
+ "grad_norm": 12.097100257873535,
3164
+ "learning_rate": 1.4443684450524397e-05,
3165
+ "loss": 0.4429,
3166
+ "step": 10825
3167
+ },
3168
+ {
3169
+ "epoch": 7.421340629274966,
3170
+ "grad_norm": 16.263513565063477,
3171
+ "learning_rate": 1.4348685210518317e-05,
3172
+ "loss": 0.3024,
3173
+ "step": 10850
3174
+ },
3175
+ {
3176
+ "epoch": 7.43844049247606,
3177
+ "grad_norm": 1.629339337348938,
3178
+ "learning_rate": 1.4253685970512237e-05,
3179
+ "loss": 0.2379,
3180
+ "step": 10875
3181
+ },
3182
+ {
3183
+ "epoch": 7.4555403556771545,
3184
+ "grad_norm": 0.13442982733249664,
3185
+ "learning_rate": 1.4158686730506158e-05,
3186
+ "loss": 0.3297,
3187
+ "step": 10900
3188
+ },
3189
+ {
3190
+ "epoch": 7.472640218878249,
3191
+ "grad_norm": 22.14508628845215,
3192
+ "learning_rate": 1.4063687490500076e-05,
3193
+ "loss": 0.3236,
3194
+ "step": 10925
3195
+ },
3196
+ {
3197
+ "epoch": 7.489740082079344,
3198
+ "grad_norm": 0.142426535487175,
3199
+ "learning_rate": 1.3968688250493997e-05,
3200
+ "loss": 0.3527,
3201
+ "step": 10950
3202
+ },
3203
+ {
3204
+ "epoch": 7.506839945280438,
3205
+ "grad_norm": 0.38898319005966187,
3206
+ "learning_rate": 1.3873689010487917e-05,
3207
+ "loss": 0.3389,
3208
+ "step": 10975
3209
+ },
3210
+ {
3211
+ "epoch": 7.523939808481532,
3212
+ "grad_norm": 0.7467881441116333,
3213
+ "learning_rate": 1.3778689770481837e-05,
3214
+ "loss": 0.3385,
3215
+ "step": 11000
3216
+ },
3217
+ {
3218
+ "epoch": 7.5410396716826265,
3219
+ "grad_norm": 12.254922866821289,
3220
+ "learning_rate": 1.3683690530475757e-05,
3221
+ "loss": 0.4477,
3222
+ "step": 11025
3223
+ },
3224
+ {
3225
+ "epoch": 7.558139534883721,
3226
+ "grad_norm": 8.779296875,
3227
+ "learning_rate": 1.3588691290469676e-05,
3228
+ "loss": 0.3468,
3229
+ "step": 11050
3230
+ },
3231
+ {
3232
+ "epoch": 7.575239398084816,
3233
+ "grad_norm": 39.5240478515625,
3234
+ "learning_rate": 1.3493692050463596e-05,
3235
+ "loss": 0.3673,
3236
+ "step": 11075
3237
+ },
3238
+ {
3239
+ "epoch": 7.592339261285909,
3240
+ "grad_norm": 11.708680152893066,
3241
+ "learning_rate": 1.3398692810457516e-05,
3242
+ "loss": 0.3424,
3243
+ "step": 11100
3244
+ },
3245
+ {
3246
+ "epoch": 7.609439124487004,
3247
+ "grad_norm": 4.1340131759643555,
3248
+ "learning_rate": 1.3303693570451437e-05,
3249
+ "loss": 0.3234,
3250
+ "step": 11125
3251
+ },
3252
+ {
3253
+ "epoch": 7.6265389876880985,
3254
+ "grad_norm": 34.217979431152344,
3255
+ "learning_rate": 1.3208694330445357e-05,
3256
+ "loss": 0.3146,
3257
+ "step": 11150
3258
+ },
3259
+ {
3260
+ "epoch": 7.643638850889193,
3261
+ "grad_norm": 2.088963508605957,
3262
+ "learning_rate": 1.3113695090439276e-05,
3263
+ "loss": 0.3924,
3264
+ "step": 11175
3265
+ },
3266
+ {
3267
+ "epoch": 7.660738714090288,
3268
+ "grad_norm": 14.35055160522461,
3269
+ "learning_rate": 1.3018695850433196e-05,
3270
+ "loss": 0.3071,
3271
+ "step": 11200
3272
+ },
3273
+ {
3274
+ "epoch": 7.677838577291381,
3275
+ "grad_norm": 7.640285491943359,
3276
+ "learning_rate": 1.2923696610427116e-05,
3277
+ "loss": 0.3105,
3278
+ "step": 11225
3279
+ },
3280
+ {
3281
+ "epoch": 7.694938440492476,
3282
+ "grad_norm": 10.794476509094238,
3283
+ "learning_rate": 1.283249734002128e-05,
3284
+ "loss": 0.3436,
3285
+ "step": 11250
3286
+ },
3287
+ {
3288
+ "epoch": 7.7120383036935705,
3289
+ "grad_norm": 21.634641647338867,
3290
+ "learning_rate": 1.27374981000152e-05,
3291
+ "loss": 0.2365,
3292
+ "step": 11275
3293
+ },
3294
+ {
3295
+ "epoch": 7.729138166894665,
3296
+ "grad_norm": 1.7207705974578857,
3297
+ "learning_rate": 1.2642498860009118e-05,
3298
+ "loss": 0.3978,
3299
+ "step": 11300
3300
+ },
3301
+ {
3302
+ "epoch": 7.74623803009576,
3303
+ "grad_norm": 2.8081226348876953,
3304
+ "learning_rate": 1.2547499620003039e-05,
3305
+ "loss": 0.2167,
3306
+ "step": 11325
3307
+ },
3308
+ {
3309
+ "epoch": 7.763337893296853,
3310
+ "grad_norm": 3.3401083946228027,
3311
+ "learning_rate": 1.245250037999696e-05,
3312
+ "loss": 0.3313,
3313
+ "step": 11350
3314
+ },
3315
+ {
3316
+ "epoch": 7.780437756497948,
3317
+ "grad_norm": 24.61179542541504,
3318
+ "learning_rate": 1.2357501139990881e-05,
3319
+ "loss": 0.4088,
3320
+ "step": 11375
3321
+ },
3322
+ {
3323
+ "epoch": 7.7975376196990425,
3324
+ "grad_norm": 3.2282018661499023,
3325
+ "learning_rate": 1.22625018999848e-05,
3326
+ "loss": 0.3965,
3327
+ "step": 11400
3328
+ },
3329
+ {
3330
+ "epoch": 7.814637482900137,
3331
+ "grad_norm": 5.257138252258301,
3332
+ "learning_rate": 1.216750265997872e-05,
3333
+ "loss": 0.275,
3334
+ "step": 11425
3335
+ },
3336
+ {
3337
+ "epoch": 7.831737346101232,
3338
+ "grad_norm": 1.5772024393081665,
3339
+ "learning_rate": 1.207250341997264e-05,
3340
+ "loss": 0.1914,
3341
+ "step": 11450
3342
+ },
3343
+ {
3344
+ "epoch": 7.848837209302325,
3345
+ "grad_norm": 7.5829033851623535,
3346
+ "learning_rate": 1.197750417996656e-05,
3347
+ "loss": 0.5063,
3348
+ "step": 11475
3349
+ },
3350
+ {
3351
+ "epoch": 7.86593707250342,
3352
+ "grad_norm": 3.2422189712524414,
3353
+ "learning_rate": 1.188250493996048e-05,
3354
+ "loss": 0.3235,
3355
+ "step": 11500
3356
+ },
3357
+ {
3358
+ "epoch": 7.8830369357045145,
3359
+ "grad_norm": 2.9335296154022217,
3360
+ "learning_rate": 1.1787505699954401e-05,
3361
+ "loss": 0.2887,
3362
+ "step": 11525
3363
+ },
3364
+ {
3365
+ "epoch": 7.900136798905609,
3366
+ "grad_norm": 0.015060103498399258,
3367
+ "learning_rate": 1.1692506459948321e-05,
3368
+ "loss": 0.4049,
3369
+ "step": 11550
3370
+ },
3371
+ {
3372
+ "epoch": 7.917236662106703,
3373
+ "grad_norm": 9.77502155303955,
3374
+ "learning_rate": 1.1597507219942242e-05,
3375
+ "loss": 0.2858,
3376
+ "step": 11575
3377
+ },
3378
+ {
3379
+ "epoch": 7.934336525307797,
3380
+ "grad_norm": 1.4075678586959839,
3381
+ "learning_rate": 1.1502507979936162e-05,
3382
+ "loss": 0.3153,
3383
+ "step": 11600
3384
+ },
3385
+ {
3386
+ "epoch": 7.951436388508892,
3387
+ "grad_norm": 12.93342399597168,
3388
+ "learning_rate": 1.140750873993008e-05,
3389
+ "loss": 0.3199,
3390
+ "step": 11625
3391
+ },
3392
+ {
3393
+ "epoch": 7.9685362517099865,
3394
+ "grad_norm": 19.707088470458984,
3395
+ "learning_rate": 1.1312509499924e-05,
3396
+ "loss": 0.5197,
3397
+ "step": 11650
3398
+ },
3399
+ {
3400
+ "epoch": 7.985636114911081,
3401
+ "grad_norm": 2.147141695022583,
3402
+ "learning_rate": 1.1217510259917921e-05,
3403
+ "loss": 0.338,
3404
+ "step": 11675
3405
+ },
3406
+ {
3407
+ "epoch": 8.0,
3408
+ "eval_accuracy": 0.9320587231136906,
3409
+ "eval_f1_macro": 0.8801923417646881,
3410
+ "eval_f1_micro": 0.9320587231136906,
3411
+ "eval_f1_weighted": 0.9322151264732859,
3412
+ "eval_loss": 0.21217399835586548,
3413
+ "eval_precision_macro": 0.9267115227700036,
3414
+ "eval_precision_micro": 0.9320587231136906,
3415
+ "eval_precision_weighted": 0.9357267323781668,
3416
+ "eval_recall_macro": 0.8522160392320227,
3417
+ "eval_recall_micro": 0.9320587231136906,
3418
+ "eval_recall_weighted": 0.9320587231136906,
3419
+ "eval_runtime": 25.9574,
3420
+ "eval_samples_per_second": 112.839,
3421
+ "eval_steps_per_second": 7.089,
3422
+ "step": 11696
3423
+ }
3424
+ ],
3425
+ "logging_steps": 25,
3426
+ "max_steps": 14620,
3427
+ "num_input_tokens_seen": 0,
3428
+ "num_train_epochs": 10,
3429
+ "save_steps": 500,
3430
+ "total_flos": 9.590567882317627e+18,
3431
+ "train_batch_size": 8,
3432
+ "trial_name": null,
3433
+ "trial_params": null
3434
+ }
checkpoint-11696/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a09a2d4e9e4ccb97bb813b0dc5ff911d3e4ce4871b39a8332d6b5e6c1e9f505
3
+ size 5048
config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/swinv2-base-patch4-window8-256",
3
+ "_num_labels": 17,
4
+ "architectures": [
5
+ "Swinv2ForImageClassification"
6
+ ],
7
+ "attention_probs_dropout_prob": 0.0,
8
+ "depths": [
9
+ 2,
10
+ 2,
11
+ 18,
12
+ 2
13
+ ],
14
+ "drop_path_rate": 0.1,
15
+ "embed_dim": 128,
16
+ "encoder_stride": 32,
17
+ "hidden_act": "gelu",
18
+ "hidden_dropout_prob": 0.0,
19
+ "hidden_size": 1024,
20
+ "id2label": {
21
+ "0": "AadhaarBack",
22
+ "1": "AadhaarBackMasked",
23
+ "2": "AadhaarFront",
24
+ "3": "AadhaarFrontMasked",
25
+ "4": "AadhaarRegular",
26
+ "5": "AadhaarRegularMasked",
27
+ "6": "AadhaarSelfGenerated",
28
+ "7": "DrivingLicenseNew",
29
+ "8": "DrivingLicenseOld",
30
+ "9": "MultipleOVDs",
31
+ "10": "PanCard",
32
+ "11": "PassportFirst",
33
+ "12": "PassportLast",
34
+ "13": "PassportRegular",
35
+ "14": "VoterCardBack",
36
+ "15": "VoterCardFront",
37
+ "16": "VoterCardRegular"
38
+ },
39
+ "image_size": 256,
40
+ "initializer_range": 0.02,
41
+ "label2id": {
42
+ "AadhaarBack": 0,
43
+ "AadhaarBackMasked": 1,
44
+ "AadhaarFront": 2,
45
+ "AadhaarFrontMasked": 3,
46
+ "AadhaarRegular": 4,
47
+ "AadhaarRegularMasked": 5,
48
+ "AadhaarSelfGenerated": 6,
49
+ "DrivingLicenseNew": 7,
50
+ "DrivingLicenseOld": 8,
51
+ "MultipleOVDs": 9,
52
+ "PanCard": 10,
53
+ "PassportFirst": 11,
54
+ "PassportLast": 12,
55
+ "PassportRegular": 13,
56
+ "VoterCardBack": 14,
57
+ "VoterCardFront": 15,
58
+ "VoterCardRegular": 16
59
+ },
60
+ "layer_norm_eps": 1e-05,
61
+ "mlp_ratio": 4.0,
62
+ "model_type": "swinv2",
63
+ "num_channels": 3,
64
+ "num_heads": [
65
+ 4,
66
+ 8,
67
+ 16,
68
+ 32
69
+ ],
70
+ "num_layers": 4,
71
+ "out_features": [
72
+ "stage4"
73
+ ],
74
+ "out_indices": [
75
+ 4
76
+ ],
77
+ "patch_size": 4,
78
+ "path_norm": true,
79
+ "pretrained_window_sizes": [
80
+ 0,
81
+ 0,
82
+ 0,
83
+ 0
84
+ ],
85
+ "problem_type": "single_label_classification",
86
+ "qkv_bias": true,
87
+ "stage_names": [
88
+ "stem",
89
+ "stage1",
90
+ "stage2",
91
+ "stage3",
92
+ "stage4"
93
+ ],
94
+ "torch_dtype": "float32",
95
+ "transformers_version": "4.40.1",
96
+ "use_absolute_embeddings": false,
97
+ "window_size": 8
98
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:983acdad57f05464ad5abbd11fb32e0e8ba8b0be8f7fd3762edf4f3626c10322
3
+ size 347706996
preprocessor_config.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_valid_processor_keys": [
3
+ "images",
4
+ "do_resize",
5
+ "size",
6
+ "resample",
7
+ "do_rescale",
8
+ "rescale_factor",
9
+ "do_normalize",
10
+ "image_mean",
11
+ "image_std",
12
+ "return_tensors",
13
+ "data_format",
14
+ "input_data_format"
15
+ ],
16
+ "do_normalize": true,
17
+ "do_rescale": true,
18
+ "do_resize": true,
19
+ "image_mean": [
20
+ 0.485,
21
+ 0.456,
22
+ 0.406
23
+ ],
24
+ "image_processor_type": "ViTImageProcessor",
25
+ "image_std": [
26
+ 0.229,
27
+ 0.224,
28
+ 0.225
29
+ ],
30
+ "resample": 3,
31
+ "rescale_factor": 0.00392156862745098,
32
+ "size": {
33
+ "height": 256,
34
+ "width": 256
35
+ }
36
+ }
runs/May03_05-17-47_60f4804cf903/events.out.tfevents.1714713467.60f4804cf903.10793.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:04f5808e3f72cc1004a63ec22b17e50bd42c5d9a2e56d7955580789f1c28affb
3
- size 118405
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dae2e700549798e334d4c813307bdc2fc42d3871db19f7d28c233793589c3442
3
+ size 137938
runs/May03_05-17-47_60f4804cf903/events.out.tfevents.1714716327.60f4804cf903.10793.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bece23eed0a4982b04f9a5209bc4844792cf74a667e52bbb945dd759c5462551
3
+ size 921
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a09a2d4e9e4ccb97bb813b0dc5ff911d3e4ce4871b39a8332d6b5e6c1e9f505
3
+ size 5048
training_params.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data_path": "autotrain-swinv2-base-patch4-window8-256/autotrain-data",
3
+ "model": "microsoft/swinv2-base-patch4-window8-256",
4
+ "username": "Kushagra07",
5
+ "lr": 5e-05,
6
+ "epochs": 10,
7
+ "batch_size": 8,
8
+ "warmup_ratio": 0.1,
9
+ "gradient_accumulation": 1,
10
+ "optimizer": "adamw_torch",
11
+ "scheduler": "linear",
12
+ "weight_decay": 0.0,
13
+ "max_grad_norm": 1.0,
14
+ "seed": 42,
15
+ "train_split": "train",
16
+ "valid_split": "validation",
17
+ "logging_steps": -1,
18
+ "project_name": "autotrain-swinv2-base-patch4-window8-256",
19
+ "auto_find_batch_size": true,
20
+ "mixed_precision": "fp16",
21
+ "save_total_limit": 1,
22
+ "push_to_hub": true,
23
+ "evaluation_strategy": "epoch",
24
+ "image_column": "autotrain_image",
25
+ "target_column": "autotrain_label",
26
+ "log": "tensorboard"
27
+ }