Hugging Face's logo

iamhmh
/
autotrain-isic-classification

Image Classification
Transformers
TensorBoard
Safetensors
convnext
Trained with AutoTrain
Model card Files
xet
 Metrics Community
iamhmh commited on
Commit
8680027
·
verified ·
1 Parent(s): d3dccbc

Upload folder using huggingface_hub

Browse files
Files changed (15) hide show
  1. README.md +44 -0
  2. checkpoint-7599/config.json +65 -0
  3. checkpoint-7599/model.safetensors +3 -0
  4. checkpoint-7599/optimizer.pt +3 -0
  5. checkpoint-7599/rng_state.pth +3 -0
  6. checkpoint-7599/scheduler.pt +3 -0
  7. checkpoint-7599/trainer_state.json +2217 -0
  8. checkpoint-7599/training_args.bin +3 -0
  9. config.json +65 -0
  10. model.safetensors +3 -0
  11. preprocessor_config.json +22 -0
  12. runs/Jun05_13-55-19_r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr/events.out.tfevents.1780667721.r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr.66.0 +2 -2
  13. runs/Jun05_13-55-19_r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr/events.out.tfevents.1780721202.r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr.66.1 +3 -0
  14. training_args.bin +3 -0
  15. training_params.json +29 -0
README.md ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ tags:
4
+ - autotrain
5
+ - transformers
6
+ - image-classification
7
+ base_model: facebook/convnext-base-224-22k-1k
8
+ widget:
9
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
10
+ example_title: Tiger
11
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
12
+ example_title: Teapot
13
+ - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
14
+ example_title: Palace
15
+ datasets:
16
+ - iamhmh/Skin-Lesion-Dataset-cls
17
+ ---
18
+
19
+ # Model Trained Using AutoTrain
20
+
21
+ - Problem type: Image Classification
22
+
23
+ ## Validation Metrics
24
+ loss: 0.43574896454811096
25
+
26
+ f1_macro: 0.7652355174728005
27
+
28
+ f1_micro: 0.8454706927175843
29
+
30
+ f1_weighted: 0.8408555162880879
31
+
32
+ precision_macro: 0.8044349451647841
33
+
34
+ precision_micro: 0.8454706927175843
35
+
36
+ precision_weighted: 0.8433307863044311
37
+
38
+ recall_macro: 0.737854710508238
39
+
40
+ recall_micro: 0.8454706927175843
41
+
42
+ recall_weighted: 0.8454706927175843
43
+
44
+ accuracy: 0.8454706927175843
checkpoint-7599/config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/convnext-base-224-22k-1k",
3
+ "_num_labels": 8,
4
+ "architectures": [
5
+ "ConvNextForImageClassification"
6
+ ],
7
+ "depths": [
8
+ 3,
9
+ 3,
10
+ 27,
11
+ 3
12
+ ],
13
+ "drop_path_rate": 0.0,
14
+ "hidden_act": "gelu",
15
+ "hidden_sizes": [
16
+ 128,
17
+ 256,
18
+ 512,
19
+ 1024
20
+ ],
21
+ "id2label": {
22
+ "0": "Actinic keratosis",
23
+ "1": "Basal cell carcinoma",
24
+ "2": "Benign keratosis",
25
+ "3": "Dermatofibroma",
26
+ "4": "Melanocytic nevus",
27
+ "5": "Melanoma",
28
+ "6": "Squamous cell carcinoma",
29
+ "7": "Vascular lesion"
30
+ },
31
+ "image_size": 224,
32
+ "initializer_range": 0.02,
33
+ "label2id": {
34
+ "Actinic keratosis": 0,
35
+ "Basal cell carcinoma": 1,
36
+ "Benign keratosis": 2,
37
+ "Dermatofibroma": 3,
38
+ "Melanocytic nevus": 4,
39
+ "Melanoma": 5,
40
+ "Squamous cell carcinoma": 6,
41
+ "Vascular lesion": 7
42
+ },
43
+ "layer_norm_eps": 1e-12,
44
+ "layer_scale_init_value": 1e-06,
45
+ "model_type": "convnext",
46
+ "num_channels": 3,
47
+ "num_stages": 4,
48
+ "out_features": [
49
+ "stage4"
50
+ ],
51
+ "out_indices": [
52
+ 4
53
+ ],
54
+ "patch_size": 4,
55
+ "problem_type": "single_label_classification",
56
+ "stage_names": [
57
+ "stem",
58
+ "stage1",
59
+ "stage2",
60
+ "stage3",
61
+ "stage4"
62
+ ],
63
+ "torch_dtype": "float32",
64
+ "transformers_version": "4.48.0"
65
+ }
checkpoint-7599/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fe4d93d6bd489381fe9208a61ce52e463a66262634b0a26aa230542b49fb69c
3
+ size 350339816
checkpoint-7599/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6075d8c764fd31f96c3f8a12754c4fae67346d6d436f80f50e2f8180eaf1ff1d
3
+ size 700875418
checkpoint-7599/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddf7cbb8a895087798abbfdd757b7aa6b51ebecb531825f3c3497b9e2ffffccd
3
+ size 13990
checkpoint-7599/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c9ba617c0e22723448e7ec3b0695f1902c753edb254aca6ce1157ee2acfde5e
3
+ size 1064
checkpoint-7599/trainer_state.json ADDED
@@ -0,0 +1,2217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.43574896454811096,
3
+ "best_model_checkpoint": "autotrain-isic-classification/checkpoint-7599",
4
+ "epoch": 3.0,
5
+ "eval_steps": 500,
6
+ "global_step": 7599,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.00986971969996052,
13
+ "grad_norm": 25.0999813079834,
14
+ "learning_rate": 1.6447368421052632e-06,
15
+ "loss": 2.0777,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.01973943939992104,
20
+ "grad_norm": 30.493947982788086,
21
+ "learning_rate": 3.2894736842105265e-06,
22
+ "loss": 1.7983,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.029609159099881564,
27
+ "grad_norm": 25.269393920898438,
28
+ "learning_rate": 4.9342105263157895e-06,
29
+ "loss": 1.5531,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.03947887879984208,
34
+ "grad_norm": 17.744693756103516,
35
+ "learning_rate": 6.578947368421053e-06,
36
+ "loss": 1.3686,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.0493485984998026,
41
+ "grad_norm": 27.930036544799805,
42
+ "learning_rate": 8.223684210526317e-06,
43
+ "loss": 1.2562,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.05921831819976313,
48
+ "grad_norm": 26.923683166503906,
49
+ "learning_rate": 9.868421052631579e-06,
50
+ "loss": 1.3676,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.06908803789972365,
55
+ "grad_norm": 40.14069747924805,
56
+ "learning_rate": 1.1513157894736843e-05,
57
+ "loss": 1.1066,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.07895775759968417,
62
+ "grad_norm": 39.232975006103516,
63
+ "learning_rate": 1.3157894736842106e-05,
64
+ "loss": 1.0852,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.08882747729964469,
69
+ "grad_norm": 34.82527160644531,
70
+ "learning_rate": 1.4802631578947368e-05,
71
+ "loss": 1.1792,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.0986971969996052,
76
+ "grad_norm": 21.1320858001709,
77
+ "learning_rate": 1.6447368421052635e-05,
78
+ "loss": 1.0685,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.10856691669956574,
83
+ "grad_norm": 47.288944244384766,
84
+ "learning_rate": 1.8092105263157896e-05,
85
+ "loss": 1.0651,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.11843663639952626,
90
+ "grad_norm": 16.042551040649414,
91
+ "learning_rate": 1.9736842105263158e-05,
92
+ "loss": 1.0289,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 0.12830635609948676,
97
+ "grad_norm": 39.400638580322266,
98
+ "learning_rate": 2.1381578947368423e-05,
99
+ "loss": 1.0511,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 0.1381760757994473,
104
+ "grad_norm": 21.65935516357422,
105
+ "learning_rate": 2.3026315789473685e-05,
106
+ "loss": 0.9327,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 0.14804579549940783,
111
+ "grad_norm": 75.3477783203125,
112
+ "learning_rate": 2.4671052631578947e-05,
113
+ "loss": 0.9733,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 0.15791551519936833,
118
+ "grad_norm": 20.73127555847168,
119
+ "learning_rate": 2.6315789473684212e-05,
120
+ "loss": 1.0996,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 0.16778523489932887,
125
+ "grad_norm": 26.542682647705078,
126
+ "learning_rate": 2.7960526315789477e-05,
127
+ "loss": 1.0169,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 0.17765495459928937,
132
+ "grad_norm": 50.412784576416016,
133
+ "learning_rate": 2.9605263157894735e-05,
134
+ "loss": 1.0499,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 0.1875246742992499,
139
+ "grad_norm": 21.049476623535156,
140
+ "learning_rate": 3.125e-05,
141
+ "loss": 0.9067,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 0.1973943939992104,
146
+ "grad_norm": 27.147974014282227,
147
+ "learning_rate": 3.289473684210527e-05,
148
+ "loss": 0.9,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 0.20726411369917094,
153
+ "grad_norm": 20.38839340209961,
154
+ "learning_rate": 3.4539473684210524e-05,
155
+ "loss": 1.0106,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 0.21713383339913148,
160
+ "grad_norm": 24.02570152282715,
161
+ "learning_rate": 3.618421052631579e-05,
162
+ "loss": 0.9807,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 0.22700355309909198,
167
+ "grad_norm": 27.63599967956543,
168
+ "learning_rate": 3.7828947368421054e-05,
169
+ "loss": 0.9346,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 0.23687327279905251,
174
+ "grad_norm": 26.512802124023438,
175
+ "learning_rate": 3.9473684210526316e-05,
176
+ "loss": 1.027,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 0.24674299249901302,
181
+ "grad_norm": 26.357986450195312,
182
+ "learning_rate": 4.111842105263158e-05,
183
+ "loss": 1.0345,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 0.2566127121989735,
188
+ "grad_norm": 24.496692657470703,
189
+ "learning_rate": 4.2763157894736847e-05,
190
+ "loss": 0.9104,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 0.26648243189893406,
195
+ "grad_norm": 15.447589874267578,
196
+ "learning_rate": 4.440789473684211e-05,
197
+ "loss": 0.8903,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 0.2763521515988946,
202
+ "grad_norm": 31.088895797729492,
203
+ "learning_rate": 4.605263157894737e-05,
204
+ "loss": 0.9364,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 0.2862218712988551,
209
+ "grad_norm": 38.402069091796875,
210
+ "learning_rate": 4.769736842105263e-05,
211
+ "loss": 0.9732,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 0.29609159099881566,
216
+ "grad_norm": 34.51318359375,
217
+ "learning_rate": 4.9342105263157894e-05,
218
+ "loss": 1.1673,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 0.30596131069877613,
223
+ "grad_norm": 44.380184173583984,
224
+ "learning_rate": 4.989033484427548e-05,
225
+ "loss": 0.9581,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 0.31583103039873667,
230
+ "grad_norm": 18.320039749145508,
231
+ "learning_rate": 4.970755958473461e-05,
232
+ "loss": 1.0624,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 0.3257007500986972,
237
+ "grad_norm": 19.321006774902344,
238
+ "learning_rate": 4.952478432519374e-05,
239
+ "loss": 0.8812,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 0.33557046979865773,
244
+ "grad_norm": 15.637593269348145,
245
+ "learning_rate": 4.934200906565288e-05,
246
+ "loss": 0.8638,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 0.34544018949861827,
251
+ "grad_norm": 19.557714462280273,
252
+ "learning_rate": 4.915923380611201e-05,
253
+ "loss": 0.9585,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 0.35530990919857874,
258
+ "grad_norm": 12.019965171813965,
259
+ "learning_rate": 4.897645854657114e-05,
260
+ "loss": 1.1381,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 0.3651796288985393,
265
+ "grad_norm": 19.117229461669922,
266
+ "learning_rate": 4.879368328703027e-05,
267
+ "loss": 0.9149,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 0.3750493485984998,
272
+ "grad_norm": 34.87349319458008,
273
+ "learning_rate": 4.86109080274894e-05,
274
+ "loss": 0.9622,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 0.38491906829846034,
279
+ "grad_norm": 15.408000946044922,
280
+ "learning_rate": 4.842813276794853e-05,
281
+ "loss": 0.9076,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 0.3947887879984208,
286
+ "grad_norm": 19.23248291015625,
287
+ "learning_rate": 4.824535750840766e-05,
288
+ "loss": 0.8175,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 0.40465850769838135,
293
+ "grad_norm": 11.322759628295898,
294
+ "learning_rate": 4.806258224886679e-05,
295
+ "loss": 0.8329,
296
+ "step": 1025
297
+ },
298
+ {
299
+ "epoch": 0.4145282273983419,
300
+ "grad_norm": 15.662063598632812,
301
+ "learning_rate": 4.787980698932593e-05,
302
+ "loss": 0.8039,
303
+ "step": 1050
304
+ },
305
+ {
306
+ "epoch": 0.4243979470983024,
307
+ "grad_norm": 23.581134796142578,
308
+ "learning_rate": 4.769703172978506e-05,
309
+ "loss": 0.779,
310
+ "step": 1075
311
+ },
312
+ {
313
+ "epoch": 0.43426766679826295,
314
+ "grad_norm": 25.19561767578125,
315
+ "learning_rate": 4.751425647024419e-05,
316
+ "loss": 0.7923,
317
+ "step": 1100
318
+ },
319
+ {
320
+ "epoch": 0.44413738649822343,
321
+ "grad_norm": 16.203529357910156,
322
+ "learning_rate": 4.733148121070332e-05,
323
+ "loss": 0.8502,
324
+ "step": 1125
325
+ },
326
+ {
327
+ "epoch": 0.45400710619818396,
328
+ "grad_norm": 15.839350700378418,
329
+ "learning_rate": 4.714870595116245e-05,
330
+ "loss": 0.9716,
331
+ "step": 1150
332
+ },
333
+ {
334
+ "epoch": 0.4638768258981445,
335
+ "grad_norm": 24.037397384643555,
336
+ "learning_rate": 4.6965930691621584e-05,
337
+ "loss": 0.9184,
338
+ "step": 1175
339
+ },
340
+ {
341
+ "epoch": 0.47374654559810503,
342
+ "grad_norm": 17.370864868164062,
343
+ "learning_rate": 4.6783155432080715e-05,
344
+ "loss": 0.8308,
345
+ "step": 1200
346
+ },
347
+ {
348
+ "epoch": 0.48361626529806556,
349
+ "grad_norm": 26.372404098510742,
350
+ "learning_rate": 4.6600380172539846e-05,
351
+ "loss": 0.9355,
352
+ "step": 1225
353
+ },
354
+ {
355
+ "epoch": 0.49348598499802604,
356
+ "grad_norm": 22.37080192565918,
357
+ "learning_rate": 4.641760491299898e-05,
358
+ "loss": 0.7275,
359
+ "step": 1250
360
+ },
361
+ {
362
+ "epoch": 0.5033557046979866,
363
+ "grad_norm": 18.308631896972656,
364
+ "learning_rate": 4.623482965345811e-05,
365
+ "loss": 0.8706,
366
+ "step": 1275
367
+ },
368
+ {
369
+ "epoch": 0.513225424397947,
370
+ "grad_norm": 8.21078872680664,
371
+ "learning_rate": 4.6052054393917246e-05,
372
+ "loss": 0.8162,
373
+ "step": 1300
374
+ },
375
+ {
376
+ "epoch": 0.5230951440979076,
377
+ "grad_norm": 17.85355567932129,
378
+ "learning_rate": 4.586927913437638e-05,
379
+ "loss": 0.7218,
380
+ "step": 1325
381
+ },
382
+ {
383
+ "epoch": 0.5329648637978681,
384
+ "grad_norm": 23.966585159301758,
385
+ "learning_rate": 4.568650387483551e-05,
386
+ "loss": 0.773,
387
+ "step": 1350
388
+ },
389
+ {
390
+ "epoch": 0.5428345834978286,
391
+ "grad_norm": 21.39402961730957,
392
+ "learning_rate": 4.550372861529464e-05,
393
+ "loss": 0.7524,
394
+ "step": 1375
395
+ },
396
+ {
397
+ "epoch": 0.5527043031977892,
398
+ "grad_norm": 14.66258716583252,
399
+ "learning_rate": 4.532095335575376e-05,
400
+ "loss": 0.7847,
401
+ "step": 1400
402
+ },
403
+ {
404
+ "epoch": 0.5625740228977497,
405
+ "grad_norm": 13.11605453491211,
406
+ "learning_rate": 4.5138178096212894e-05,
407
+ "loss": 0.8984,
408
+ "step": 1425
409
+ },
410
+ {
411
+ "epoch": 0.5724437425977102,
412
+ "grad_norm": 21.897939682006836,
413
+ "learning_rate": 4.4955402836672025e-05,
414
+ "loss": 0.7827,
415
+ "step": 1450
416
+ },
417
+ {
418
+ "epoch": 0.5823134622976708,
419
+ "grad_norm": 7.581227779388428,
420
+ "learning_rate": 4.477262757713116e-05,
421
+ "loss": 0.6864,
422
+ "step": 1475
423
+ },
424
+ {
425
+ "epoch": 0.5921831819976313,
426
+ "grad_norm": 32.005470275878906,
427
+ "learning_rate": 4.4589852317590294e-05,
428
+ "loss": 0.809,
429
+ "step": 1500
430
+ },
431
+ {
432
+ "epoch": 0.6020529016975917,
433
+ "grad_norm": 16.7988338470459,
434
+ "learning_rate": 4.4407077058049425e-05,
435
+ "loss": 0.7486,
436
+ "step": 1525
437
+ },
438
+ {
439
+ "epoch": 0.6119226213975523,
440
+ "grad_norm": 25.918445587158203,
441
+ "learning_rate": 4.4224301798508556e-05,
442
+ "loss": 0.7427,
443
+ "step": 1550
444
+ },
445
+ {
446
+ "epoch": 0.6217923410975128,
447
+ "grad_norm": 33.619197845458984,
448
+ "learning_rate": 4.404152653896769e-05,
449
+ "loss": 0.8231,
450
+ "step": 1575
451
+ },
452
+ {
453
+ "epoch": 0.6316620607974733,
454
+ "grad_norm": 20.559589385986328,
455
+ "learning_rate": 4.385875127942682e-05,
456
+ "loss": 0.7987,
457
+ "step": 1600
458
+ },
459
+ {
460
+ "epoch": 0.6415317804974339,
461
+ "grad_norm": 16.634279251098633,
462
+ "learning_rate": 4.367597601988595e-05,
463
+ "loss": 0.7849,
464
+ "step": 1625
465
+ },
466
+ {
467
+ "epoch": 0.6514015001973944,
468
+ "grad_norm": 2.1054515838623047,
469
+ "learning_rate": 4.349320076034508e-05,
470
+ "loss": 0.7116,
471
+ "step": 1650
472
+ },
473
+ {
474
+ "epoch": 0.6612712198973549,
475
+ "grad_norm": 22.52297019958496,
476
+ "learning_rate": 4.331042550080421e-05,
477
+ "loss": 0.8869,
478
+ "step": 1675
479
+ },
480
+ {
481
+ "epoch": 0.6711409395973155,
482
+ "grad_norm": 18.856929779052734,
483
+ "learning_rate": 4.312765024126334e-05,
484
+ "loss": 0.7304,
485
+ "step": 1700
486
+ },
487
+ {
488
+ "epoch": 0.681010659297276,
489
+ "grad_norm": 17.665882110595703,
490
+ "learning_rate": 4.294487498172248e-05,
491
+ "loss": 0.8061,
492
+ "step": 1725
493
+ },
494
+ {
495
+ "epoch": 0.6908803789972365,
496
+ "grad_norm": 8.828686714172363,
497
+ "learning_rate": 4.276209972218161e-05,
498
+ "loss": 0.7946,
499
+ "step": 1750
500
+ },
501
+ {
502
+ "epoch": 0.700750098697197,
503
+ "grad_norm": 16.475543975830078,
504
+ "learning_rate": 4.257932446264074e-05,
505
+ "loss": 0.8982,
506
+ "step": 1775
507
+ },
508
+ {
509
+ "epoch": 0.7106198183971575,
510
+ "grad_norm": 15.741327285766602,
511
+ "learning_rate": 4.239654920309987e-05,
512
+ "loss": 0.7502,
513
+ "step": 1800
514
+ },
515
+ {
516
+ "epoch": 0.720489538097118,
517
+ "grad_norm": 14.704769134521484,
518
+ "learning_rate": 4.2213773943559004e-05,
519
+ "loss": 0.7659,
520
+ "step": 1825
521
+ },
522
+ {
523
+ "epoch": 0.7303592577970786,
524
+ "grad_norm": 22.844655990600586,
525
+ "learning_rate": 4.2030998684018135e-05,
526
+ "loss": 0.817,
527
+ "step": 1850
528
+ },
529
+ {
530
+ "epoch": 0.7402289774970391,
531
+ "grad_norm": 12.58967113494873,
532
+ "learning_rate": 4.1848223424477266e-05,
533
+ "loss": 0.8209,
534
+ "step": 1875
535
+ },
536
+ {
537
+ "epoch": 0.7500986971969996,
538
+ "grad_norm": 16.43173599243164,
539
+ "learning_rate": 4.166544816493639e-05,
540
+ "loss": 0.7411,
541
+ "step": 1900
542
+ },
543
+ {
544
+ "epoch": 0.7599684168969602,
545
+ "grad_norm": 9.30561351776123,
546
+ "learning_rate": 4.148267290539553e-05,
547
+ "loss": 0.8279,
548
+ "step": 1925
549
+ },
550
+ {
551
+ "epoch": 0.7698381365969207,
552
+ "grad_norm": 2.8649349212646484,
553
+ "learning_rate": 4.129989764585466e-05,
554
+ "loss": 0.6167,
555
+ "step": 1950
556
+ },
557
+ {
558
+ "epoch": 0.7797078562968812,
559
+ "grad_norm": 24.597929000854492,
560
+ "learning_rate": 4.111712238631379e-05,
561
+ "loss": 0.7578,
562
+ "step": 1975
563
+ },
564
+ {
565
+ "epoch": 0.7895775759968416,
566
+ "grad_norm": 19.52870750427246,
567
+ "learning_rate": 4.093434712677292e-05,
568
+ "loss": 0.8271,
569
+ "step": 2000
570
+ },
571
+ {
572
+ "epoch": 0.7994472956968022,
573
+ "grad_norm": 6.929624080657959,
574
+ "learning_rate": 4.075157186723205e-05,
575
+ "loss": 0.8274,
576
+ "step": 2025
577
+ },
578
+ {
579
+ "epoch": 0.8093170153967627,
580
+ "grad_norm": 25.068206787109375,
581
+ "learning_rate": 4.0568796607691183e-05,
582
+ "loss": 0.6578,
583
+ "step": 2050
584
+ },
585
+ {
586
+ "epoch": 0.8191867350967232,
587
+ "grad_norm": 12.562817573547363,
588
+ "learning_rate": 4.0386021348150314e-05,
589
+ "loss": 0.6846,
590
+ "step": 2075
591
+ },
592
+ {
593
+ "epoch": 0.8290564547966838,
594
+ "grad_norm": 18.090431213378906,
595
+ "learning_rate": 4.0203246088609446e-05,
596
+ "loss": 0.8098,
597
+ "step": 2100
598
+ },
599
+ {
600
+ "epoch": 0.8389261744966443,
601
+ "grad_norm": 12.163636207580566,
602
+ "learning_rate": 4.0020470829068577e-05,
603
+ "loss": 0.7855,
604
+ "step": 2125
605
+ },
606
+ {
607
+ "epoch": 0.8487958941966048,
608
+ "grad_norm": 16.596872329711914,
609
+ "learning_rate": 3.9837695569527714e-05,
610
+ "loss": 0.751,
611
+ "step": 2150
612
+ },
613
+ {
614
+ "epoch": 0.8586656138965654,
615
+ "grad_norm": 16.66319465637207,
616
+ "learning_rate": 3.9654920309986845e-05,
617
+ "loss": 0.6907,
618
+ "step": 2175
619
+ },
620
+ {
621
+ "epoch": 0.8685353335965259,
622
+ "grad_norm": 21.114709854125977,
623
+ "learning_rate": 3.9472145050445976e-05,
624
+ "loss": 0.8384,
625
+ "step": 2200
626
+ },
627
+ {
628
+ "epoch": 0.8784050532964863,
629
+ "grad_norm": 13.302197456359863,
630
+ "learning_rate": 3.928936979090511e-05,
631
+ "loss": 0.6809,
632
+ "step": 2225
633
+ },
634
+ {
635
+ "epoch": 0.8882747729964469,
636
+ "grad_norm": 16.761959075927734,
637
+ "learning_rate": 3.910659453136424e-05,
638
+ "loss": 0.593,
639
+ "step": 2250
640
+ },
641
+ {
642
+ "epoch": 0.8981444926964074,
643
+ "grad_norm": 19.18243408203125,
644
+ "learning_rate": 3.892381927182337e-05,
645
+ "loss": 0.6979,
646
+ "step": 2275
647
+ },
648
+ {
649
+ "epoch": 0.9080142123963679,
650
+ "grad_norm": 28.19027328491211,
651
+ "learning_rate": 3.87410440122825e-05,
652
+ "loss": 0.6263,
653
+ "step": 2300
654
+ },
655
+ {
656
+ "epoch": 0.9178839320963285,
657
+ "grad_norm": 18.185087203979492,
658
+ "learning_rate": 3.855826875274163e-05,
659
+ "loss": 0.7499,
660
+ "step": 2325
661
+ },
662
+ {
663
+ "epoch": 0.927753651796289,
664
+ "grad_norm": 9.373323440551758,
665
+ "learning_rate": 3.837549349320076e-05,
666
+ "loss": 0.6716,
667
+ "step": 2350
668
+ },
669
+ {
670
+ "epoch": 0.9376233714962495,
671
+ "grad_norm": 11.500312805175781,
672
+ "learning_rate": 3.8192718233659894e-05,
673
+ "loss": 0.7692,
674
+ "step": 2375
675
+ },
676
+ {
677
+ "epoch": 0.9474930911962101,
678
+ "grad_norm": 18.218793869018555,
679
+ "learning_rate": 3.8009942974119025e-05,
680
+ "loss": 0.698,
681
+ "step": 2400
682
+ },
683
+ {
684
+ "epoch": 0.9573628108961706,
685
+ "grad_norm": 4.465493202209473,
686
+ "learning_rate": 3.7827167714578156e-05,
687
+ "loss": 0.6934,
688
+ "step": 2425
689
+ },
690
+ {
691
+ "epoch": 0.9672325305961311,
692
+ "grad_norm": 19.380374908447266,
693
+ "learning_rate": 3.7644392455037287e-05,
694
+ "loss": 0.6831,
695
+ "step": 2450
696
+ },
697
+ {
698
+ "epoch": 0.9771022502960915,
699
+ "grad_norm": 10.438068389892578,
700
+ "learning_rate": 3.746161719549642e-05,
701
+ "loss": 0.7189,
702
+ "step": 2475
703
+ },
704
+ {
705
+ "epoch": 0.9869719699960521,
706
+ "grad_norm": 22.57378387451172,
707
+ "learning_rate": 3.727884193595555e-05,
708
+ "loss": 0.7538,
709
+ "step": 2500
710
+ },
711
+ {
712
+ "epoch": 0.9968416896960126,
713
+ "grad_norm": 29.394481658935547,
714
+ "learning_rate": 3.709606667641468e-05,
715
+ "loss": 0.757,
716
+ "step": 2525
717
+ },
718
+ {
719
+ "epoch": 1.0,
720
+ "eval_accuracy": 0.7829090191434774,
721
+ "eval_f1_macro": 0.6393797994797579,
722
+ "eval_f1_micro": 0.7829090191434774,
723
+ "eval_f1_weighted": 0.7690810894820675,
724
+ "eval_loss": 0.5891534686088562,
725
+ "eval_precision_macro": 0.7106493485019916,
726
+ "eval_precision_micro": 0.7829090191434774,
727
+ "eval_precision_weighted": 0.7768640232429261,
728
+ "eval_recall_macro": 0.5993363894614356,
729
+ "eval_recall_micro": 0.7829090191434774,
730
+ "eval_recall_weighted": 0.7829090191434774,
731
+ "eval_runtime": 1063.1018,
732
+ "eval_samples_per_second": 4.766,
733
+ "eval_steps_per_second": 0.298,
734
+ "step": 2533
735
+ },
736
+ {
737
+ "epoch": 1.0067114093959733,
738
+ "grad_norm": 14.244729995727539,
739
+ "learning_rate": 3.691329141687381e-05,
740
+ "loss": 0.6476,
741
+ "step": 2550
742
+ },
743
+ {
744
+ "epoch": 1.0165811290959337,
745
+ "grad_norm": 25.252973556518555,
746
+ "learning_rate": 3.673051615733294e-05,
747
+ "loss": 0.7387,
748
+ "step": 2575
749
+ },
750
+ {
751
+ "epoch": 1.026450848795894,
752
+ "grad_norm": 19.978221893310547,
753
+ "learning_rate": 3.654774089779208e-05,
754
+ "loss": 0.8226,
755
+ "step": 2600
756
+ },
757
+ {
758
+ "epoch": 1.0363205684958547,
759
+ "grad_norm": 19.843408584594727,
760
+ "learning_rate": 3.636496563825121e-05,
761
+ "loss": 0.5976,
762
+ "step": 2625
763
+ },
764
+ {
765
+ "epoch": 1.0461902881958152,
766
+ "grad_norm": 8.418437957763672,
767
+ "learning_rate": 3.618219037871034e-05,
768
+ "loss": 0.6609,
769
+ "step": 2650
770
+ },
771
+ {
772
+ "epoch": 1.0560600078957758,
773
+ "grad_norm": 11.428473472595215,
774
+ "learning_rate": 3.599941511916947e-05,
775
+ "loss": 0.5666,
776
+ "step": 2675
777
+ },
778
+ {
779
+ "epoch": 1.0659297275957362,
780
+ "grad_norm": 9.486809730529785,
781
+ "learning_rate": 3.5816639859628604e-05,
782
+ "loss": 0.7747,
783
+ "step": 2700
784
+ },
785
+ {
786
+ "epoch": 1.0757994472956969,
787
+ "grad_norm": 26.120243072509766,
788
+ "learning_rate": 3.5633864600087735e-05,
789
+ "loss": 0.5642,
790
+ "step": 2725
791
+ },
792
+ {
793
+ "epoch": 1.0856691669956573,
794
+ "grad_norm": 21.967893600463867,
795
+ "learning_rate": 3.5451089340546866e-05,
796
+ "loss": 0.6559,
797
+ "step": 2750
798
+ },
799
+ {
800
+ "epoch": 1.095538886695618,
801
+ "grad_norm": 20.46396827697754,
802
+ "learning_rate": 3.5268314081006e-05,
803
+ "loss": 0.5173,
804
+ "step": 2775
805
+ },
806
+ {
807
+ "epoch": 1.1054086063955784,
808
+ "grad_norm": 21.288713455200195,
809
+ "learning_rate": 3.508553882146513e-05,
810
+ "loss": 0.6468,
811
+ "step": 2800
812
+ },
813
+ {
814
+ "epoch": 1.1152783260955388,
815
+ "grad_norm": 32.73738098144531,
816
+ "learning_rate": 3.4902763561924265e-05,
817
+ "loss": 0.6433,
818
+ "step": 2825
819
+ },
820
+ {
821
+ "epoch": 1.1251480457954994,
822
+ "grad_norm": 11.702080726623535,
823
+ "learning_rate": 3.471998830238339e-05,
824
+ "loss": 0.559,
825
+ "step": 2850
826
+ },
827
+ {
828
+ "epoch": 1.1350177654954599,
829
+ "grad_norm": 10.3779935836792,
830
+ "learning_rate": 3.453721304284252e-05,
831
+ "loss": 0.7606,
832
+ "step": 2875
833
+ },
834
+ {
835
+ "epoch": 1.1448874851954205,
836
+ "grad_norm": 17.79201889038086,
837
+ "learning_rate": 3.435443778330165e-05,
838
+ "loss": 0.6205,
839
+ "step": 2900
840
+ },
841
+ {
842
+ "epoch": 1.154757204895381,
843
+ "grad_norm": 19.825288772583008,
844
+ "learning_rate": 3.417166252376078e-05,
845
+ "loss": 0.7145,
846
+ "step": 2925
847
+ },
848
+ {
849
+ "epoch": 1.1646269245953416,
850
+ "grad_norm": 22.765090942382812,
851
+ "learning_rate": 3.3988887264219914e-05,
852
+ "loss": 0.6381,
853
+ "step": 2950
854
+ },
855
+ {
856
+ "epoch": 1.174496644295302,
857
+ "grad_norm": 13.03752326965332,
858
+ "learning_rate": 3.3806112004679045e-05,
859
+ "loss": 0.6322,
860
+ "step": 2975
861
+ },
862
+ {
863
+ "epoch": 1.1843663639952626,
864
+ "grad_norm": 9.798502922058105,
865
+ "learning_rate": 3.3623336745138176e-05,
866
+ "loss": 0.6048,
867
+ "step": 3000
868
+ },
869
+ {
870
+ "epoch": 1.194236083695223,
871
+ "grad_norm": 3.8797762393951416,
872
+ "learning_rate": 3.3440561485597314e-05,
873
+ "loss": 0.639,
874
+ "step": 3025
875
+ },
876
+ {
877
+ "epoch": 1.2041058033951835,
878
+ "grad_norm": 23.310298919677734,
879
+ "learning_rate": 3.3257786226056445e-05,
880
+ "loss": 0.4762,
881
+ "step": 3050
882
+ },
883
+ {
884
+ "epoch": 1.2139755230951441,
885
+ "grad_norm": 27.62248420715332,
886
+ "learning_rate": 3.3075010966515576e-05,
887
+ "loss": 0.7103,
888
+ "step": 3075
889
+ },
890
+ {
891
+ "epoch": 1.2238452427951045,
892
+ "grad_norm": 15.467683792114258,
893
+ "learning_rate": 3.289223570697471e-05,
894
+ "loss": 0.6423,
895
+ "step": 3100
896
+ },
897
+ {
898
+ "epoch": 1.2337149624950652,
899
+ "grad_norm": 36.30628204345703,
900
+ "learning_rate": 3.270946044743384e-05,
901
+ "loss": 0.5795,
902
+ "step": 3125
903
+ },
904
+ {
905
+ "epoch": 1.2435846821950256,
906
+ "grad_norm": 20.377031326293945,
907
+ "learning_rate": 3.252668518789297e-05,
908
+ "loss": 0.5869,
909
+ "step": 3150
910
+ },
911
+ {
912
+ "epoch": 1.2534544018949862,
913
+ "grad_norm": 29.11665916442871,
914
+ "learning_rate": 3.23439099283521e-05,
915
+ "loss": 0.5219,
916
+ "step": 3175
917
+ },
918
+ {
919
+ "epoch": 1.2633241215949467,
920
+ "grad_norm": 15.241193771362305,
921
+ "learning_rate": 3.216113466881123e-05,
922
+ "loss": 0.5867,
923
+ "step": 3200
924
+ },
925
+ {
926
+ "epoch": 1.273193841294907,
927
+ "grad_norm": 24.407711029052734,
928
+ "learning_rate": 3.197835940927036e-05,
929
+ "loss": 0.6741,
930
+ "step": 3225
931
+ },
932
+ {
933
+ "epoch": 1.2830635609948677,
934
+ "grad_norm": 12.935891151428223,
935
+ "learning_rate": 3.179558414972949e-05,
936
+ "loss": 0.6823,
937
+ "step": 3250
938
+ },
939
+ {
940
+ "epoch": 1.2929332806948284,
941
+ "grad_norm": 27.278520584106445,
942
+ "learning_rate": 3.161280889018863e-05,
943
+ "loss": 0.624,
944
+ "step": 3275
945
+ },
946
+ {
947
+ "epoch": 1.3028030003947888,
948
+ "grad_norm": 18.958881378173828,
949
+ "learning_rate": 3.143003363064776e-05,
950
+ "loss": 0.6733,
951
+ "step": 3300
952
+ },
953
+ {
954
+ "epoch": 1.3126727200947492,
955
+ "grad_norm": 14.983283996582031,
956
+ "learning_rate": 3.124725837110689e-05,
957
+ "loss": 0.4995,
958
+ "step": 3325
959
+ },
960
+ {
961
+ "epoch": 1.3225424397947099,
962
+ "grad_norm": 13.329388618469238,
963
+ "learning_rate": 3.106448311156602e-05,
964
+ "loss": 0.5669,
965
+ "step": 3350
966
+ },
967
+ {
968
+ "epoch": 1.3324121594946703,
969
+ "grad_norm": 14.233263969421387,
970
+ "learning_rate": 3.088170785202515e-05,
971
+ "loss": 0.5388,
972
+ "step": 3375
973
+ },
974
+ {
975
+ "epoch": 1.342281879194631,
976
+ "grad_norm": 28.532625198364258,
977
+ "learning_rate": 3.069893259248428e-05,
978
+ "loss": 0.6032,
979
+ "step": 3400
980
+ },
981
+ {
982
+ "epoch": 1.3521515988945914,
983
+ "grad_norm": 14.022845268249512,
984
+ "learning_rate": 3.0516157332943413e-05,
985
+ "loss": 0.5692,
986
+ "step": 3425
987
+ },
988
+ {
989
+ "epoch": 1.362021318594552,
990
+ "grad_norm": 16.256967544555664,
991
+ "learning_rate": 3.0333382073402544e-05,
992
+ "loss": 0.6422,
993
+ "step": 3450
994
+ },
995
+ {
996
+ "epoch": 1.3718910382945124,
997
+ "grad_norm": 25.442201614379883,
998
+ "learning_rate": 3.0150606813861675e-05,
999
+ "loss": 0.6548,
1000
+ "step": 3475
1001
+ },
1002
+ {
1003
+ "epoch": 1.3817607579944728,
1004
+ "grad_norm": 11.086042404174805,
1005
+ "learning_rate": 2.9967831554320806e-05,
1006
+ "loss": 0.5942,
1007
+ "step": 3500
1008
+ },
1009
+ {
1010
+ "epoch": 1.3916304776944335,
1011
+ "grad_norm": 17.43187141418457,
1012
+ "learning_rate": 2.978505629477994e-05,
1013
+ "loss": 0.6171,
1014
+ "step": 3525
1015
+ },
1016
+ {
1017
+ "epoch": 1.4015001973943941,
1018
+ "grad_norm": 30.734569549560547,
1019
+ "learning_rate": 2.9602281035239072e-05,
1020
+ "loss": 0.5743,
1021
+ "step": 3550
1022
+ },
1023
+ {
1024
+ "epoch": 1.4113699170943546,
1025
+ "grad_norm": 25.920391082763672,
1026
+ "learning_rate": 2.9419505775698203e-05,
1027
+ "loss": 0.7259,
1028
+ "step": 3575
1029
+ },
1030
+ {
1031
+ "epoch": 1.421239636794315,
1032
+ "grad_norm": 16.234350204467773,
1033
+ "learning_rate": 2.9236730516157334e-05,
1034
+ "loss": 0.6923,
1035
+ "step": 3600
1036
+ },
1037
+ {
1038
+ "epoch": 1.4311093564942756,
1039
+ "grad_norm": 11.903316497802734,
1040
+ "learning_rate": 2.9053955256616465e-05,
1041
+ "loss": 0.6666,
1042
+ "step": 3625
1043
+ },
1044
+ {
1045
+ "epoch": 1.440979076194236,
1046
+ "grad_norm": 16.291624069213867,
1047
+ "learning_rate": 2.88711799970756e-05,
1048
+ "loss": 0.6895,
1049
+ "step": 3650
1050
+ },
1051
+ {
1052
+ "epoch": 1.4508487958941967,
1053
+ "grad_norm": 7.6475043296813965,
1054
+ "learning_rate": 2.868840473753473e-05,
1055
+ "loss": 0.6074,
1056
+ "step": 3675
1057
+ },
1058
+ {
1059
+ "epoch": 1.460718515594157,
1060
+ "grad_norm": 15.164865493774414,
1061
+ "learning_rate": 2.850562947799386e-05,
1062
+ "loss": 0.5836,
1063
+ "step": 3700
1064
+ },
1065
+ {
1066
+ "epoch": 1.4705882352941178,
1067
+ "grad_norm": 15.323176383972168,
1068
+ "learning_rate": 2.8322854218452992e-05,
1069
+ "loss": 0.5748,
1070
+ "step": 3725
1071
+ },
1072
+ {
1073
+ "epoch": 1.4804579549940782,
1074
+ "grad_norm": 16.380990982055664,
1075
+ "learning_rate": 2.8140078958912123e-05,
1076
+ "loss": 0.5972,
1077
+ "step": 3750
1078
+ },
1079
+ {
1080
+ "epoch": 1.4903276746940386,
1081
+ "grad_norm": 28.51302719116211,
1082
+ "learning_rate": 2.7957303699371258e-05,
1083
+ "loss": 0.5138,
1084
+ "step": 3775
1085
+ },
1086
+ {
1087
+ "epoch": 1.5001973943939992,
1088
+ "grad_norm": 15.194903373718262,
1089
+ "learning_rate": 2.777452843983039e-05,
1090
+ "loss": 0.4735,
1091
+ "step": 3800
1092
+ },
1093
+ {
1094
+ "epoch": 1.5100671140939599,
1095
+ "grad_norm": 12.498093605041504,
1096
+ "learning_rate": 2.7591753180289513e-05,
1097
+ "loss": 0.6098,
1098
+ "step": 3825
1099
+ },
1100
+ {
1101
+ "epoch": 1.5199368337939203,
1102
+ "grad_norm": 17.336671829223633,
1103
+ "learning_rate": 2.7408977920748648e-05,
1104
+ "loss": 0.5771,
1105
+ "step": 3850
1106
+ },
1107
+ {
1108
+ "epoch": 1.5298065534938807,
1109
+ "grad_norm": 19.822059631347656,
1110
+ "learning_rate": 2.722620266120778e-05,
1111
+ "loss": 0.5413,
1112
+ "step": 3875
1113
+ },
1114
+ {
1115
+ "epoch": 1.5396762731938414,
1116
+ "grad_norm": 20.59796714782715,
1117
+ "learning_rate": 2.704342740166691e-05,
1118
+ "loss": 0.4793,
1119
+ "step": 3900
1120
+ },
1121
+ {
1122
+ "epoch": 1.5495459928938018,
1123
+ "grad_norm": 6.3360090255737305,
1124
+ "learning_rate": 2.686065214212604e-05,
1125
+ "loss": 0.5867,
1126
+ "step": 3925
1127
+ },
1128
+ {
1129
+ "epoch": 1.5594157125937622,
1130
+ "grad_norm": 22.340862274169922,
1131
+ "learning_rate": 2.667787688258517e-05,
1132
+ "loss": 0.58,
1133
+ "step": 3950
1134
+ },
1135
+ {
1136
+ "epoch": 1.5692854322937229,
1137
+ "grad_norm": 24.037965774536133,
1138
+ "learning_rate": 2.6495101623044306e-05,
1139
+ "loss": 0.6503,
1140
+ "step": 3975
1141
+ },
1142
+ {
1143
+ "epoch": 1.5791551519936835,
1144
+ "grad_norm": 24.841894149780273,
1145
+ "learning_rate": 2.6312326363503437e-05,
1146
+ "loss": 0.6013,
1147
+ "step": 4000
1148
+ },
1149
+ {
1150
+ "epoch": 1.589024871693644,
1151
+ "grad_norm": 11.183895111083984,
1152
+ "learning_rate": 2.6129551103962568e-05,
1153
+ "loss": 0.5305,
1154
+ "step": 4025
1155
+ },
1156
+ {
1157
+ "epoch": 1.5988945913936043,
1158
+ "grad_norm": 14.20339584350586,
1159
+ "learning_rate": 2.59467758444217e-05,
1160
+ "loss": 0.5365,
1161
+ "step": 4050
1162
+ },
1163
+ {
1164
+ "epoch": 1.608764311093565,
1165
+ "grad_norm": 23.43878746032715,
1166
+ "learning_rate": 2.5764000584880834e-05,
1167
+ "loss": 0.5629,
1168
+ "step": 4075
1169
+ },
1170
+ {
1171
+ "epoch": 1.6186340307935254,
1172
+ "grad_norm": 17.492982864379883,
1173
+ "learning_rate": 2.5581225325339965e-05,
1174
+ "loss": 0.6107,
1175
+ "step": 4100
1176
+ },
1177
+ {
1178
+ "epoch": 1.6285037504934858,
1179
+ "grad_norm": 24.659347534179688,
1180
+ "learning_rate": 2.5398450065799096e-05,
1181
+ "loss": 0.5654,
1182
+ "step": 4125
1183
+ },
1184
+ {
1185
+ "epoch": 1.6383734701934465,
1186
+ "grad_norm": 22.476972579956055,
1187
+ "learning_rate": 2.5215674806258227e-05,
1188
+ "loss": 0.6437,
1189
+ "step": 4150
1190
+ },
1191
+ {
1192
+ "epoch": 1.6482431898934071,
1193
+ "grad_norm": 8.731144905090332,
1194
+ "learning_rate": 2.5032899546717358e-05,
1195
+ "loss": 0.4423,
1196
+ "step": 4175
1197
+ },
1198
+ {
1199
+ "epoch": 1.6581129095933675,
1200
+ "grad_norm": 13.396164894104004,
1201
+ "learning_rate": 2.485012428717649e-05,
1202
+ "loss": 0.4773,
1203
+ "step": 4200
1204
+ },
1205
+ {
1206
+ "epoch": 1.667982629293328,
1207
+ "grad_norm": 16.639312744140625,
1208
+ "learning_rate": 2.466734902763562e-05,
1209
+ "loss": 0.6061,
1210
+ "step": 4225
1211
+ },
1212
+ {
1213
+ "epoch": 1.6778523489932886,
1214
+ "grad_norm": 13.545028686523438,
1215
+ "learning_rate": 2.448457376809475e-05,
1216
+ "loss": 0.5133,
1217
+ "step": 4250
1218
+ },
1219
+ {
1220
+ "epoch": 1.6877220686932493,
1221
+ "grad_norm": 14.784280776977539,
1222
+ "learning_rate": 2.4301798508553882e-05,
1223
+ "loss": 0.6224,
1224
+ "step": 4275
1225
+ },
1226
+ {
1227
+ "epoch": 1.6975917883932097,
1228
+ "grad_norm": 22.461013793945312,
1229
+ "learning_rate": 2.4119023249013016e-05,
1230
+ "loss": 0.483,
1231
+ "step": 4300
1232
+ },
1233
+ {
1234
+ "epoch": 1.70746150809317,
1235
+ "grad_norm": 20.606239318847656,
1236
+ "learning_rate": 2.3936247989472147e-05,
1237
+ "loss": 0.6534,
1238
+ "step": 4325
1239
+ },
1240
+ {
1241
+ "epoch": 1.7173312277931307,
1242
+ "grad_norm": 18.11288070678711,
1243
+ "learning_rate": 2.3753472729931278e-05,
1244
+ "loss": 0.5385,
1245
+ "step": 4350
1246
+ },
1247
+ {
1248
+ "epoch": 1.7272009474930912,
1249
+ "grad_norm": 22.29888916015625,
1250
+ "learning_rate": 2.357069747039041e-05,
1251
+ "loss": 0.5903,
1252
+ "step": 4375
1253
+ },
1254
+ {
1255
+ "epoch": 1.7370706671930516,
1256
+ "grad_norm": 12.731755256652832,
1257
+ "learning_rate": 2.338792221084954e-05,
1258
+ "loss": 0.5678,
1259
+ "step": 4400
1260
+ },
1261
+ {
1262
+ "epoch": 1.7469403868930122,
1263
+ "grad_norm": 8.69069766998291,
1264
+ "learning_rate": 2.3205146951308675e-05,
1265
+ "loss": 0.512,
1266
+ "step": 4425
1267
+ },
1268
+ {
1269
+ "epoch": 1.7568101065929729,
1270
+ "grad_norm": 26.662981033325195,
1271
+ "learning_rate": 2.3022371691767802e-05,
1272
+ "loss": 0.5558,
1273
+ "step": 4450
1274
+ },
1275
+ {
1276
+ "epoch": 1.7666798262929333,
1277
+ "grad_norm": 14.457517623901367,
1278
+ "learning_rate": 2.2839596432226933e-05,
1279
+ "loss": 0.4987,
1280
+ "step": 4475
1281
+ },
1282
+ {
1283
+ "epoch": 1.7765495459928937,
1284
+ "grad_norm": 2.5073020458221436,
1285
+ "learning_rate": 2.2656821172686064e-05,
1286
+ "loss": 0.5357,
1287
+ "step": 4500
1288
+ },
1289
+ {
1290
+ "epoch": 1.7864192656928544,
1291
+ "grad_norm": 7.441634178161621,
1292
+ "learning_rate": 2.24740459131452e-05,
1293
+ "loss": 0.4351,
1294
+ "step": 4525
1295
+ },
1296
+ {
1297
+ "epoch": 1.796288985392815,
1298
+ "grad_norm": 18.357187271118164,
1299
+ "learning_rate": 2.229127065360433e-05,
1300
+ "loss": 0.6192,
1301
+ "step": 4550
1302
+ },
1303
+ {
1304
+ "epoch": 1.8061587050927752,
1305
+ "grad_norm": 16.99340057373047,
1306
+ "learning_rate": 2.210849539406346e-05,
1307
+ "loss": 0.4862,
1308
+ "step": 4575
1309
+ },
1310
+ {
1311
+ "epoch": 1.8160284247927359,
1312
+ "grad_norm": 25.90440559387207,
1313
+ "learning_rate": 2.1925720134522592e-05,
1314
+ "loss": 0.5096,
1315
+ "step": 4600
1316
+ },
1317
+ {
1318
+ "epoch": 1.8258981444926965,
1319
+ "grad_norm": 22.586809158325195,
1320
+ "learning_rate": 2.1742944874981723e-05,
1321
+ "loss": 0.4958,
1322
+ "step": 4625
1323
+ },
1324
+ {
1325
+ "epoch": 1.835767864192657,
1326
+ "grad_norm": 23.470739364624023,
1327
+ "learning_rate": 2.1560169615440857e-05,
1328
+ "loss": 0.5184,
1329
+ "step": 4650
1330
+ },
1331
+ {
1332
+ "epoch": 1.8456375838926173,
1333
+ "grad_norm": 20.166990280151367,
1334
+ "learning_rate": 2.1377394355899985e-05,
1335
+ "loss": 0.5994,
1336
+ "step": 4675
1337
+ },
1338
+ {
1339
+ "epoch": 1.855507303592578,
1340
+ "grad_norm": 22.983253479003906,
1341
+ "learning_rate": 2.1194619096359116e-05,
1342
+ "loss": 0.46,
1343
+ "step": 4700
1344
+ },
1345
+ {
1346
+ "epoch": 1.8653770232925386,
1347
+ "grad_norm": 14.648038864135742,
1348
+ "learning_rate": 2.1011843836818247e-05,
1349
+ "loss": 0.5299,
1350
+ "step": 4725
1351
+ },
1352
+ {
1353
+ "epoch": 1.875246742992499,
1354
+ "grad_norm": 34.9222526550293,
1355
+ "learning_rate": 2.082906857727738e-05,
1356
+ "loss": 0.5083,
1357
+ "step": 4750
1358
+ },
1359
+ {
1360
+ "epoch": 1.8851164626924595,
1361
+ "grad_norm": 16.850589752197266,
1362
+ "learning_rate": 2.0646293317736512e-05,
1363
+ "loss": 0.6334,
1364
+ "step": 4775
1365
+ },
1366
+ {
1367
+ "epoch": 1.8949861823924201,
1368
+ "grad_norm": 4.693751811981201,
1369
+ "learning_rate": 2.0463518058195643e-05,
1370
+ "loss": 0.5115,
1371
+ "step": 4800
1372
+ },
1373
+ {
1374
+ "epoch": 1.9048559020923805,
1375
+ "grad_norm": 3.580418825149536,
1376
+ "learning_rate": 2.0280742798654774e-05,
1377
+ "loss": 0.5175,
1378
+ "step": 4825
1379
+ },
1380
+ {
1381
+ "epoch": 1.914725621792341,
1382
+ "grad_norm": 19.99139404296875,
1383
+ "learning_rate": 2.009796753911391e-05,
1384
+ "loss": 0.6136,
1385
+ "step": 4850
1386
+ },
1387
+ {
1388
+ "epoch": 1.9245953414923016,
1389
+ "grad_norm": 25.498565673828125,
1390
+ "learning_rate": 1.991519227957304e-05,
1391
+ "loss": 0.457,
1392
+ "step": 4875
1393
+ },
1394
+ {
1395
+ "epoch": 1.9344650611922622,
1396
+ "grad_norm": 14.489584922790527,
1397
+ "learning_rate": 1.973241702003217e-05,
1398
+ "loss": 0.6387,
1399
+ "step": 4900
1400
+ },
1401
+ {
1402
+ "epoch": 1.9443347808922227,
1403
+ "grad_norm": 18.26885986328125,
1404
+ "learning_rate": 1.95496417604913e-05,
1405
+ "loss": 0.5035,
1406
+ "step": 4925
1407
+ },
1408
+ {
1409
+ "epoch": 1.954204500592183,
1410
+ "grad_norm": 9.917611122131348,
1411
+ "learning_rate": 1.9366866500950433e-05,
1412
+ "loss": 0.3699,
1413
+ "step": 4950
1414
+ },
1415
+ {
1416
+ "epoch": 1.9640742202921437,
1417
+ "grad_norm": 18.40938949584961,
1418
+ "learning_rate": 1.9184091241409564e-05,
1419
+ "loss": 0.5523,
1420
+ "step": 4975
1421
+ },
1422
+ {
1423
+ "epoch": 1.9739439399921044,
1424
+ "grad_norm": 19.932710647583008,
1425
+ "learning_rate": 1.9001315981868695e-05,
1426
+ "loss": 0.5687,
1427
+ "step": 5000
1428
+ },
1429
+ {
1430
+ "epoch": 1.9838136596920646,
1431
+ "grad_norm": 20.109010696411133,
1432
+ "learning_rate": 1.8818540722327826e-05,
1433
+ "loss": 0.5099,
1434
+ "step": 5025
1435
+ },
1436
+ {
1437
+ "epoch": 1.9936833793920252,
1438
+ "grad_norm": 3.5672338008880615,
1439
+ "learning_rate": 1.8635765462786957e-05,
1440
+ "loss": 0.3938,
1441
+ "step": 5050
1442
+ },
1443
+ {
1444
+ "epoch": 2.0,
1445
+ "eval_accuracy": 0.8229721728833629,
1446
+ "eval_f1_macro": 0.7299934749159617,
1447
+ "eval_f1_micro": 0.8229721728833629,
1448
+ "eval_f1_weighted": 0.8194821357959808,
1449
+ "eval_loss": 0.4958721101284027,
1450
+ "eval_precision_macro": 0.7743781806725194,
1451
+ "eval_precision_micro": 0.8229721728833629,
1452
+ "eval_precision_weighted": 0.8257741693056009,
1453
+ "eval_recall_macro": 0.7105166399489131,
1454
+ "eval_recall_micro": 0.8229721728833629,
1455
+ "eval_recall_weighted": 0.8229721728833629,
1456
+ "eval_runtime": 919.6204,
1457
+ "eval_samples_per_second": 5.51,
1458
+ "eval_steps_per_second": 0.345,
1459
+ "step": 5066
1460
+ },
1461
+ {
1462
+ "epoch": 2.003553099091986,
1463
+ "grad_norm": 14.951340675354004,
1464
+ "learning_rate": 1.845299020324609e-05,
1465
+ "loss": 0.5526,
1466
+ "step": 5075
1467
+ },
1468
+ {
1469
+ "epoch": 2.0134228187919465,
1470
+ "grad_norm": 12.719818115234375,
1471
+ "learning_rate": 1.8270214943705222e-05,
1472
+ "loss": 0.4645,
1473
+ "step": 5100
1474
+ },
1475
+ {
1476
+ "epoch": 2.0232925384919067,
1477
+ "grad_norm": 12.654365539550781,
1478
+ "learning_rate": 1.8087439684164353e-05,
1479
+ "loss": 0.4659,
1480
+ "step": 5125
1481
+ },
1482
+ {
1483
+ "epoch": 2.0331622581918674,
1484
+ "grad_norm": 4.800103187561035,
1485
+ "learning_rate": 1.7904664424623484e-05,
1486
+ "loss": 0.4502,
1487
+ "step": 5150
1488
+ },
1489
+ {
1490
+ "epoch": 2.043031977891828,
1491
+ "grad_norm": 19.990001678466797,
1492
+ "learning_rate": 1.7721889165082615e-05,
1493
+ "loss": 0.4843,
1494
+ "step": 5175
1495
+ },
1496
+ {
1497
+ "epoch": 2.052901697591788,
1498
+ "grad_norm": 11.023597717285156,
1499
+ "learning_rate": 1.7539113905541746e-05,
1500
+ "loss": 0.4086,
1501
+ "step": 5200
1502
+ },
1503
+ {
1504
+ "epoch": 2.062771417291749,
1505
+ "grad_norm": 11.913839340209961,
1506
+ "learning_rate": 1.7356338646000877e-05,
1507
+ "loss": 0.5539,
1508
+ "step": 5225
1509
+ },
1510
+ {
1511
+ "epoch": 2.0726411369917095,
1512
+ "grad_norm": 3.3574135303497314,
1513
+ "learning_rate": 1.717356338646001e-05,
1514
+ "loss": 0.4304,
1515
+ "step": 5250
1516
+ },
1517
+ {
1518
+ "epoch": 2.08251085669167,
1519
+ "grad_norm": 20.17797088623047,
1520
+ "learning_rate": 1.699078812691914e-05,
1521
+ "loss": 0.5083,
1522
+ "step": 5275
1523
+ },
1524
+ {
1525
+ "epoch": 2.0923805763916303,
1526
+ "grad_norm": 5.290871620178223,
1527
+ "learning_rate": 1.6808012867378274e-05,
1528
+ "loss": 0.4864,
1529
+ "step": 5300
1530
+ },
1531
+ {
1532
+ "epoch": 2.102250296091591,
1533
+ "grad_norm": 15.89094352722168,
1534
+ "learning_rate": 1.6625237607837405e-05,
1535
+ "loss": 0.5024,
1536
+ "step": 5325
1537
+ },
1538
+ {
1539
+ "epoch": 2.1121200157915516,
1540
+ "grad_norm": 4.068813323974609,
1541
+ "learning_rate": 1.6442462348296536e-05,
1542
+ "loss": 0.4512,
1543
+ "step": 5350
1544
+ },
1545
+ {
1546
+ "epoch": 2.1219897354915123,
1547
+ "grad_norm": 15.388639450073242,
1548
+ "learning_rate": 1.6259687088755667e-05,
1549
+ "loss": 0.5073,
1550
+ "step": 5375
1551
+ },
1552
+ {
1553
+ "epoch": 2.1318594551914725,
1554
+ "grad_norm": 25.19047737121582,
1555
+ "learning_rate": 1.6076911829214798e-05,
1556
+ "loss": 0.4504,
1557
+ "step": 5400
1558
+ },
1559
+ {
1560
+ "epoch": 2.141729174891433,
1561
+ "grad_norm": 33.80352783203125,
1562
+ "learning_rate": 1.589413656967393e-05,
1563
+ "loss": 0.3969,
1564
+ "step": 5425
1565
+ },
1566
+ {
1567
+ "epoch": 2.1515988945913938,
1568
+ "grad_norm": 13.962345123291016,
1569
+ "learning_rate": 1.571136131013306e-05,
1570
+ "loss": 0.3277,
1571
+ "step": 5450
1572
+ },
1573
+ {
1574
+ "epoch": 2.161468614291354,
1575
+ "grad_norm": 28.54108238220215,
1576
+ "learning_rate": 1.552858605059219e-05,
1577
+ "loss": 0.3862,
1578
+ "step": 5475
1579
+ },
1580
+ {
1581
+ "epoch": 2.1713383339913146,
1582
+ "grad_norm": 15.133607864379883,
1583
+ "learning_rate": 1.5345810791051322e-05,
1584
+ "loss": 0.423,
1585
+ "step": 5500
1586
+ },
1587
+ {
1588
+ "epoch": 2.1812080536912752,
1589
+ "grad_norm": 39.927486419677734,
1590
+ "learning_rate": 1.5163035531510455e-05,
1591
+ "loss": 0.4244,
1592
+ "step": 5525
1593
+ },
1594
+ {
1595
+ "epoch": 2.191077773391236,
1596
+ "grad_norm": 18.54796028137207,
1597
+ "learning_rate": 1.4980260271969588e-05,
1598
+ "loss": 0.3858,
1599
+ "step": 5550
1600
+ },
1601
+ {
1602
+ "epoch": 2.200947493091196,
1603
+ "grad_norm": 21.207942962646484,
1604
+ "learning_rate": 1.4797485012428719e-05,
1605
+ "loss": 0.4163,
1606
+ "step": 5575
1607
+ },
1608
+ {
1609
+ "epoch": 2.2108172127911567,
1610
+ "grad_norm": 17.74234390258789,
1611
+ "learning_rate": 1.4614709752887851e-05,
1612
+ "loss": 0.4178,
1613
+ "step": 5600
1614
+ },
1615
+ {
1616
+ "epoch": 2.2206869324911174,
1617
+ "grad_norm": 13.287278175354004,
1618
+ "learning_rate": 1.4431934493346982e-05,
1619
+ "loss": 0.3856,
1620
+ "step": 5625
1621
+ },
1622
+ {
1623
+ "epoch": 2.2305566521910776,
1624
+ "grad_norm": 24.611202239990234,
1625
+ "learning_rate": 1.4249159233806112e-05,
1626
+ "loss": 0.467,
1627
+ "step": 5650
1628
+ },
1629
+ {
1630
+ "epoch": 2.240426371891038,
1631
+ "grad_norm": 10.964824676513672,
1632
+ "learning_rate": 1.4066383974265243e-05,
1633
+ "loss": 0.4459,
1634
+ "step": 5675
1635
+ },
1636
+ {
1637
+ "epoch": 2.250296091590999,
1638
+ "grad_norm": 14.910752296447754,
1639
+ "learning_rate": 1.3883608714724375e-05,
1640
+ "loss": 0.4456,
1641
+ "step": 5700
1642
+ },
1643
+ {
1644
+ "epoch": 2.2601658112909595,
1645
+ "grad_norm": 15.01776123046875,
1646
+ "learning_rate": 1.3700833455183506e-05,
1647
+ "loss": 0.4256,
1648
+ "step": 5725
1649
+ },
1650
+ {
1651
+ "epoch": 2.2700355309909197,
1652
+ "grad_norm": 20.166276931762695,
1653
+ "learning_rate": 1.3518058195642639e-05,
1654
+ "loss": 0.4132,
1655
+ "step": 5750
1656
+ },
1657
+ {
1658
+ "epoch": 2.2799052506908803,
1659
+ "grad_norm": 26.255537033081055,
1660
+ "learning_rate": 1.333528293610177e-05,
1661
+ "loss": 0.4448,
1662
+ "step": 5775
1663
+ },
1664
+ {
1665
+ "epoch": 2.289774970390841,
1666
+ "grad_norm": 23.030710220336914,
1667
+ "learning_rate": 1.3152507676560901e-05,
1668
+ "loss": 0.4048,
1669
+ "step": 5800
1670
+ },
1671
+ {
1672
+ "epoch": 2.299644690090801,
1673
+ "grad_norm": 29.610889434814453,
1674
+ "learning_rate": 1.2969732417020034e-05,
1675
+ "loss": 0.4801,
1676
+ "step": 5825
1677
+ },
1678
+ {
1679
+ "epoch": 2.309514409790762,
1680
+ "grad_norm": 16.124771118164062,
1681
+ "learning_rate": 1.2786957157479165e-05,
1682
+ "loss": 0.3615,
1683
+ "step": 5850
1684
+ },
1685
+ {
1686
+ "epoch": 2.3193841294907225,
1687
+ "grad_norm": 22.03716278076172,
1688
+ "learning_rate": 1.2604181897938298e-05,
1689
+ "loss": 0.4415,
1690
+ "step": 5875
1691
+ },
1692
+ {
1693
+ "epoch": 2.329253849190683,
1694
+ "grad_norm": 8.248544692993164,
1695
+ "learning_rate": 1.2421406638397427e-05,
1696
+ "loss": 0.4068,
1697
+ "step": 5900
1698
+ },
1699
+ {
1700
+ "epoch": 2.3391235688906433,
1701
+ "grad_norm": 25.610904693603516,
1702
+ "learning_rate": 1.223863137885656e-05,
1703
+ "loss": 0.4706,
1704
+ "step": 5925
1705
+ },
1706
+ {
1707
+ "epoch": 2.348993288590604,
1708
+ "grad_norm": 9.144474983215332,
1709
+ "learning_rate": 1.2055856119315689e-05,
1710
+ "loss": 0.4091,
1711
+ "step": 5950
1712
+ },
1713
+ {
1714
+ "epoch": 2.3588630082905646,
1715
+ "grad_norm": 19.948516845703125,
1716
+ "learning_rate": 1.1873080859774822e-05,
1717
+ "loss": 0.5141,
1718
+ "step": 5975
1719
+ },
1720
+ {
1721
+ "epoch": 2.3687327279905253,
1722
+ "grad_norm": 16.57617950439453,
1723
+ "learning_rate": 1.1690305600233953e-05,
1724
+ "loss": 0.4579,
1725
+ "step": 6000
1726
+ },
1727
+ {
1728
+ "epoch": 2.3786024476904855,
1729
+ "grad_norm": 25.633960723876953,
1730
+ "learning_rate": 1.1507530340693084e-05,
1731
+ "loss": 0.4423,
1732
+ "step": 6025
1733
+ },
1734
+ {
1735
+ "epoch": 2.388472167390446,
1736
+ "grad_norm": 12.349534034729004,
1737
+ "learning_rate": 1.1324755081152216e-05,
1738
+ "loss": 0.4361,
1739
+ "step": 6050
1740
+ },
1741
+ {
1742
+ "epoch": 2.3983418870904067,
1743
+ "grad_norm": 40.79236602783203,
1744
+ "learning_rate": 1.1141979821611346e-05,
1745
+ "loss": 0.3169,
1746
+ "step": 6075
1747
+ },
1748
+ {
1749
+ "epoch": 2.408211606790367,
1750
+ "grad_norm": 15.606341361999512,
1751
+ "learning_rate": 1.0959204562070479e-05,
1752
+ "loss": 0.3476,
1753
+ "step": 6100
1754
+ },
1755
+ {
1756
+ "epoch": 2.4180813264903276,
1757
+ "grad_norm": 2.443814754486084,
1758
+ "learning_rate": 1.077642930252961e-05,
1759
+ "loss": 0.3997,
1760
+ "step": 6125
1761
+ },
1762
+ {
1763
+ "epoch": 2.4279510461902882,
1764
+ "grad_norm": 15.338932037353516,
1765
+ "learning_rate": 1.0593654042988742e-05,
1766
+ "loss": 0.4415,
1767
+ "step": 6150
1768
+ },
1769
+ {
1770
+ "epoch": 2.437820765890249,
1771
+ "grad_norm": 2.6813647747039795,
1772
+ "learning_rate": 1.0410878783447873e-05,
1773
+ "loss": 0.3386,
1774
+ "step": 6175
1775
+ },
1776
+ {
1777
+ "epoch": 2.447690485590209,
1778
+ "grad_norm": 29.423179626464844,
1779
+ "learning_rate": 1.0228103523907004e-05,
1780
+ "loss": 0.4022,
1781
+ "step": 6200
1782
+ },
1783
+ {
1784
+ "epoch": 2.4575602052901697,
1785
+ "grad_norm": 24.724660873413086,
1786
+ "learning_rate": 1.0045328264366135e-05,
1787
+ "loss": 0.4807,
1788
+ "step": 6225
1789
+ },
1790
+ {
1791
+ "epoch": 2.4674299249901304,
1792
+ "grad_norm": 42.14269256591797,
1793
+ "learning_rate": 9.862553004825268e-06,
1794
+ "loss": 0.4496,
1795
+ "step": 6250
1796
+ },
1797
+ {
1798
+ "epoch": 2.477299644690091,
1799
+ "grad_norm": 37.18028259277344,
1800
+ "learning_rate": 9.679777745284399e-06,
1801
+ "loss": 0.3096,
1802
+ "step": 6275
1803
+ },
1804
+ {
1805
+ "epoch": 2.487169364390051,
1806
+ "grad_norm": 2.2330679893493652,
1807
+ "learning_rate": 9.49700248574353e-06,
1808
+ "loss": 0.4591,
1809
+ "step": 6300
1810
+ },
1811
+ {
1812
+ "epoch": 2.497039084090012,
1813
+ "grad_norm": 24.654094696044922,
1814
+ "learning_rate": 9.314227226202661e-06,
1815
+ "loss": 0.4897,
1816
+ "step": 6325
1817
+ },
1818
+ {
1819
+ "epoch": 2.5069088037899725,
1820
+ "grad_norm": 21.505754470825195,
1821
+ "learning_rate": 9.131451966661792e-06,
1822
+ "loss": 0.5316,
1823
+ "step": 6350
1824
+ },
1825
+ {
1826
+ "epoch": 2.5167785234899327,
1827
+ "grad_norm": 19.120290756225586,
1828
+ "learning_rate": 8.948676707120925e-06,
1829
+ "loss": 0.3768,
1830
+ "step": 6375
1831
+ },
1832
+ {
1833
+ "epoch": 2.5266482431898933,
1834
+ "grad_norm": 32.53146743774414,
1835
+ "learning_rate": 8.765901447580056e-06,
1836
+ "loss": 0.4556,
1837
+ "step": 6400
1838
+ },
1839
+ {
1840
+ "epoch": 2.536517962889854,
1841
+ "grad_norm": 10.568330764770508,
1842
+ "learning_rate": 8.583126188039189e-06,
1843
+ "loss": 0.3882,
1844
+ "step": 6425
1845
+ },
1846
+ {
1847
+ "epoch": 2.546387682589814,
1848
+ "grad_norm": 4.047234535217285,
1849
+ "learning_rate": 8.400350928498318e-06,
1850
+ "loss": 0.3386,
1851
+ "step": 6450
1852
+ },
1853
+ {
1854
+ "epoch": 2.556257402289775,
1855
+ "grad_norm": 23.04155731201172,
1856
+ "learning_rate": 8.21757566895745e-06,
1857
+ "loss": 0.362,
1858
+ "step": 6475
1859
+ },
1860
+ {
1861
+ "epoch": 2.5661271219897355,
1862
+ "grad_norm": 11.348421096801758,
1863
+ "learning_rate": 8.034800409416582e-06,
1864
+ "loss": 0.462,
1865
+ "step": 6500
1866
+ },
1867
+ {
1868
+ "epoch": 2.575996841689696,
1869
+ "grad_norm": 12.276947021484375,
1870
+ "learning_rate": 7.852025149875714e-06,
1871
+ "loss": 0.3586,
1872
+ "step": 6525
1873
+ },
1874
+ {
1875
+ "epoch": 2.5858665613896568,
1876
+ "grad_norm": 26.5312442779541,
1877
+ "learning_rate": 7.669249890334845e-06,
1878
+ "loss": 0.4807,
1879
+ "step": 6550
1880
+ },
1881
+ {
1882
+ "epoch": 2.595736281089617,
1883
+ "grad_norm": 11.15167236328125,
1884
+ "learning_rate": 7.4864746307939755e-06,
1885
+ "loss": 0.4517,
1886
+ "step": 6575
1887
+ },
1888
+ {
1889
+ "epoch": 2.6056060007895776,
1890
+ "grad_norm": 6.4104437828063965,
1891
+ "learning_rate": 7.303699371253107e-06,
1892
+ "loss": 0.4589,
1893
+ "step": 6600
1894
+ },
1895
+ {
1896
+ "epoch": 2.6154757204895382,
1897
+ "grad_norm": 2.6333577632904053,
1898
+ "learning_rate": 7.120924111712239e-06,
1899
+ "loss": 0.3638,
1900
+ "step": 6625
1901
+ },
1902
+ {
1903
+ "epoch": 2.6253454401894984,
1904
+ "grad_norm": 22.92894172668457,
1905
+ "learning_rate": 6.938148852171371e-06,
1906
+ "loss": 0.3743,
1907
+ "step": 6650
1908
+ },
1909
+ {
1910
+ "epoch": 2.635215159889459,
1911
+ "grad_norm": 8.734560012817383,
1912
+ "learning_rate": 6.755373592630501e-06,
1913
+ "loss": 0.413,
1914
+ "step": 6675
1915
+ },
1916
+ {
1917
+ "epoch": 2.6450848795894197,
1918
+ "grad_norm": 6.890560150146484,
1919
+ "learning_rate": 6.572598333089633e-06,
1920
+ "loss": 0.3552,
1921
+ "step": 6700
1922
+ },
1923
+ {
1924
+ "epoch": 2.65495459928938,
1925
+ "grad_norm": 14.292202949523926,
1926
+ "learning_rate": 6.389823073548764e-06,
1927
+ "loss": 0.5168,
1928
+ "step": 6725
1929
+ },
1930
+ {
1931
+ "epoch": 2.6648243189893406,
1932
+ "grad_norm": 7.978025436401367,
1933
+ "learning_rate": 6.207047814007896e-06,
1934
+ "loss": 0.3818,
1935
+ "step": 6750
1936
+ },
1937
+ {
1938
+ "epoch": 2.6746940386893012,
1939
+ "grad_norm": 30.14126968383789,
1940
+ "learning_rate": 6.024272554467027e-06,
1941
+ "loss": 0.4388,
1942
+ "step": 6775
1943
+ },
1944
+ {
1945
+ "epoch": 2.684563758389262,
1946
+ "grad_norm": 12.973822593688965,
1947
+ "learning_rate": 5.841497294926159e-06,
1948
+ "loss": 0.489,
1949
+ "step": 6800
1950
+ },
1951
+ {
1952
+ "epoch": 2.6944334780892225,
1953
+ "grad_norm": 25.07337188720703,
1954
+ "learning_rate": 5.658722035385291e-06,
1955
+ "loss": 0.4268,
1956
+ "step": 6825
1957
+ },
1958
+ {
1959
+ "epoch": 2.7043031977891827,
1960
+ "grad_norm": 23.934560775756836,
1961
+ "learning_rate": 5.475946775844422e-06,
1962
+ "loss": 0.4424,
1963
+ "step": 6850
1964
+ },
1965
+ {
1966
+ "epoch": 2.7141729174891434,
1967
+ "grad_norm": 5.350096225738525,
1968
+ "learning_rate": 5.293171516303554e-06,
1969
+ "loss": 0.3586,
1970
+ "step": 6875
1971
+ },
1972
+ {
1973
+ "epoch": 2.724042637189104,
1974
+ "grad_norm": 21.11624526977539,
1975
+ "learning_rate": 5.110396256762685e-06,
1976
+ "loss": 0.4718,
1977
+ "step": 6900
1978
+ },
1979
+ {
1980
+ "epoch": 2.733912356889064,
1981
+ "grad_norm": 18.254154205322266,
1982
+ "learning_rate": 4.927620997221817e-06,
1983
+ "loss": 0.5158,
1984
+ "step": 6925
1985
+ },
1986
+ {
1987
+ "epoch": 2.743782076589025,
1988
+ "grad_norm": 24.2269344329834,
1989
+ "learning_rate": 4.744845737680948e-06,
1990
+ "loss": 0.4189,
1991
+ "step": 6950
1992
+ },
1993
+ {
1994
+ "epoch": 2.7536517962889855,
1995
+ "grad_norm": 31.407081604003906,
1996
+ "learning_rate": 4.5620704781400795e-06,
1997
+ "loss": 0.443,
1998
+ "step": 6975
1999
+ },
2000
+ {
2001
+ "epoch": 2.7635215159889457,
2002
+ "grad_norm": 23.810937881469727,
2003
+ "learning_rate": 4.3792952185992105e-06,
2004
+ "loss": 0.3933,
2005
+ "step": 7000
2006
+ },
2007
+ {
2008
+ "epoch": 2.7733912356889063,
2009
+ "grad_norm": 16.976123809814453,
2010
+ "learning_rate": 4.1965199590583416e-06,
2011
+ "loss": 0.3541,
2012
+ "step": 7025
2013
+ },
2014
+ {
2015
+ "epoch": 2.783260955388867,
2016
+ "grad_norm": 15.492680549621582,
2017
+ "learning_rate": 4.0137446995174734e-06,
2018
+ "loss": 0.4293,
2019
+ "step": 7050
2020
+ },
2021
+ {
2022
+ "epoch": 2.7931306750888276,
2023
+ "grad_norm": 18.494380950927734,
2024
+ "learning_rate": 3.8309694399766045e-06,
2025
+ "loss": 0.3858,
2026
+ "step": 7075
2027
+ },
2028
+ {
2029
+ "epoch": 2.8030003947887883,
2030
+ "grad_norm": 1.0024749040603638,
2031
+ "learning_rate": 3.6481941804357363e-06,
2032
+ "loss": 0.3909,
2033
+ "step": 7100
2034
+ },
2035
+ {
2036
+ "epoch": 2.8128701144887485,
2037
+ "grad_norm": 32.01081466674805,
2038
+ "learning_rate": 3.465418920894868e-06,
2039
+ "loss": 0.3191,
2040
+ "step": 7125
2041
+ },
2042
+ {
2043
+ "epoch": 2.822739834188709,
2044
+ "grad_norm": 18.840717315673828,
2045
+ "learning_rate": 3.2826436613539992e-06,
2046
+ "loss": 0.4713,
2047
+ "step": 7150
2048
+ },
2049
+ {
2050
+ "epoch": 2.8326095538886698,
2051
+ "grad_norm": 13.710578918457031,
2052
+ "learning_rate": 3.0998684018131307e-06,
2053
+ "loss": 0.3746,
2054
+ "step": 7175
2055
+ },
2056
+ {
2057
+ "epoch": 2.84247927358863,
2058
+ "grad_norm": 11.958606719970703,
2059
+ "learning_rate": 2.917093142272262e-06,
2060
+ "loss": 0.4829,
2061
+ "step": 7200
2062
+ },
2063
+ {
2064
+ "epoch": 2.8523489932885906,
2065
+ "grad_norm": 18.517244338989258,
2066
+ "learning_rate": 2.7343178827313936e-06,
2067
+ "loss": 0.435,
2068
+ "step": 7225
2069
+ },
2070
+ {
2071
+ "epoch": 2.8622187129885512,
2072
+ "grad_norm": 16.076797485351562,
2073
+ "learning_rate": 2.551542623190525e-06,
2074
+ "loss": 0.3858,
2075
+ "step": 7250
2076
+ },
2077
+ {
2078
+ "epoch": 2.8720884326885114,
2079
+ "grad_norm": 25.667831420898438,
2080
+ "learning_rate": 2.3687673636496564e-06,
2081
+ "loss": 0.6422,
2082
+ "step": 7275
2083
+ },
2084
+ {
2085
+ "epoch": 2.881958152388472,
2086
+ "grad_norm": 15.386273384094238,
2087
+ "learning_rate": 2.185992104108788e-06,
2088
+ "loss": 0.3105,
2089
+ "step": 7300
2090
+ },
2091
+ {
2092
+ "epoch": 2.8918278720884327,
2093
+ "grad_norm": 6.2320709228515625,
2094
+ "learning_rate": 2.0032168445679193e-06,
2095
+ "loss": 0.4257,
2096
+ "step": 7325
2097
+ },
2098
+ {
2099
+ "epoch": 2.9016975917883934,
2100
+ "grad_norm": 6.439568519592285,
2101
+ "learning_rate": 1.820441585027051e-06,
2102
+ "loss": 0.3849,
2103
+ "step": 7350
2104
+ },
2105
+ {
2106
+ "epoch": 2.9115673114883536,
2107
+ "grad_norm": 8.832841873168945,
2108
+ "learning_rate": 1.6376663254861822e-06,
2109
+ "loss": 0.2649,
2110
+ "step": 7375
2111
+ },
2112
+ {
2113
+ "epoch": 2.921437031188314,
2114
+ "grad_norm": 49.32740020751953,
2115
+ "learning_rate": 1.4548910659453137e-06,
2116
+ "loss": 0.3267,
2117
+ "step": 7400
2118
+ },
2119
+ {
2120
+ "epoch": 2.931306750888275,
2121
+ "grad_norm": 21.037437438964844,
2122
+ "learning_rate": 1.2721158064044453e-06,
2123
+ "loss": 0.4332,
2124
+ "step": 7425
2125
+ },
2126
+ {
2127
+ "epoch": 2.9411764705882355,
2128
+ "grad_norm": 3.787083387374878,
2129
+ "learning_rate": 1.0893405468635766e-06,
2130
+ "loss": 0.4566,
2131
+ "step": 7450
2132
+ },
2133
+ {
2134
+ "epoch": 2.9510461902881957,
2135
+ "grad_norm": 20.39313316345215,
2136
+ "learning_rate": 9.06565287322708e-07,
2137
+ "loss": 0.449,
2138
+ "step": 7475
2139
+ },
2140
+ {
2141
+ "epoch": 2.9609159099881563,
2142
+ "grad_norm": 10.126147270202637,
2143
+ "learning_rate": 7.237900277818395e-07,
2144
+ "loss": 0.383,
2145
+ "step": 7500
2146
+ },
2147
+ {
2148
+ "epoch": 2.970785629688117,
2149
+ "grad_norm": 11.029193878173828,
2150
+ "learning_rate": 5.410147682409709e-07,
2151
+ "loss": 0.344,
2152
+ "step": 7525
2153
+ },
2154
+ {
2155
+ "epoch": 2.980655349388077,
2156
+ "grad_norm": 29.03227996826172,
2157
+ "learning_rate": 3.5823950870010235e-07,
2158
+ "loss": 0.3794,
2159
+ "step": 7550
2160
+ },
2161
+ {
2162
+ "epoch": 2.990525069088038,
2163
+ "grad_norm": 12.453436851501465,
2164
+ "learning_rate": 1.7546424915923382e-07,
2165
+ "loss": 0.3866,
2166
+ "step": 7575
2167
+ },
2168
+ {
2169
+ "epoch": 3.0,
2170
+ "eval_accuracy": 0.8454706927175843,
2171
+ "eval_f1_macro": 0.7652355174728005,
2172
+ "eval_f1_micro": 0.8454706927175843,
2173
+ "eval_f1_weighted": 0.8408555162880879,
2174
+ "eval_loss": 0.43574896454811096,
2175
+ "eval_precision_macro": 0.8044349451647841,
2176
+ "eval_precision_micro": 0.8454706927175843,
2177
+ "eval_precision_weighted": 0.8433307863044311,
2178
+ "eval_recall_macro": 0.737854710508238,
2179
+ "eval_recall_micro": 0.8454706927175843,
2180
+ "eval_recall_weighted": 0.8454706927175843,
2181
+ "eval_runtime": 850.2427,
2182
+ "eval_samples_per_second": 5.959,
2183
+ "eval_steps_per_second": 0.373,
2184
+ "step": 7599
2185
+ }
2186
+ ],
2187
+ "logging_steps": 25,
2188
+ "max_steps": 7599,
2189
+ "num_input_tokens_seen": 0,
2190
+ "num_train_epochs": 3,
2191
+ "save_steps": 500,
2192
+ "stateful_callbacks": {
2193
+ "EarlyStoppingCallback": {
2194
+ "args": {
2195
+ "early_stopping_patience": 5,
2196
+ "early_stopping_threshold": 0.01
2197
+ },
2198
+ "attributes": {
2199
+ "early_stopping_patience_counter": 0
2200
+ }
2201
+ },
2202
+ "TrainerControl": {
2203
+ "args": {
2204
+ "should_epoch_stop": false,
2205
+ "should_evaluate": false,
2206
+ "should_log": false,
2207
+ "should_save": true,
2208
+ "should_training_stop": true
2209
+ },
2210
+ "attributes": {}
2211
+ }
2212
+ },
2213
+ "total_flos": 4.808320998368477e+18,
2214
+ "train_batch_size": 8,
2215
+ "trial_name": null,
2216
+ "trial_params": null
2217
+ }
checkpoint-7599/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb9d501d15c28db08c76906d2b502e5bcb8c54fff4a713d9ce4188a020e205e8
3
+ size 5432
config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/convnext-base-224-22k-1k",
3
+ "_num_labels": 8,
4
+ "architectures": [
5
+ "ConvNextForImageClassification"
6
+ ],
7
+ "depths": [
8
+ 3,
9
+ 3,
10
+ 27,
11
+ 3
12
+ ],
13
+ "drop_path_rate": 0.0,
14
+ "hidden_act": "gelu",
15
+ "hidden_sizes": [
16
+ 128,
17
+ 256,
18
+ 512,
19
+ 1024
20
+ ],
21
+ "id2label": {
22
+ "0": "Actinic keratosis",
23
+ "1": "Basal cell carcinoma",
24
+ "2": "Benign keratosis",
25
+ "3": "Dermatofibroma",
26
+ "4": "Melanocytic nevus",
27
+ "5": "Melanoma",
28
+ "6": "Squamous cell carcinoma",
29
+ "7": "Vascular lesion"
30
+ },
31
+ "image_size": 224,
32
+ "initializer_range": 0.02,
33
+ "label2id": {
34
+ "Actinic keratosis": 0,
35
+ "Basal cell carcinoma": 1,
36
+ "Benign keratosis": 2,
37
+ "Dermatofibroma": 3,
38
+ "Melanocytic nevus": 4,
39
+ "Melanoma": 5,
40
+ "Squamous cell carcinoma": 6,
41
+ "Vascular lesion": 7
42
+ },
43
+ "layer_norm_eps": 1e-12,
44
+ "layer_scale_init_value": 1e-06,
45
+ "model_type": "convnext",
46
+ "num_channels": 3,
47
+ "num_stages": 4,
48
+ "out_features": [
49
+ "stage4"
50
+ ],
51
+ "out_indices": [
52
+ 4
53
+ ],
54
+ "patch_size": 4,
55
+ "problem_type": "single_label_classification",
56
+ "stage_names": [
57
+ "stem",
58
+ "stage1",
59
+ "stage2",
60
+ "stage3",
61
+ "stage4"
62
+ ],
63
+ "torch_dtype": "float32",
64
+ "transformers_version": "4.48.0"
65
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fe4d93d6bd489381fe9208a61ce52e463a66262634b0a26aa230542b49fb69c
3
+ size 350339816
preprocessor_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_pct": 0.875,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.485,
8
+ 0.456,
9
+ 0.406
10
+ ],
11
+ "image_processor_type": "ConvNextImageProcessor",
12
+ "image_std": [
13
+ 0.229,
14
+ 0.224,
15
+ 0.225
16
+ ],
17
+ "resample": 3,
18
+ "rescale_factor": 0.00392156862745098,
19
+ "size": {
20
+ "shortest_edge": 224
21
+ }
22
+ }
runs/Jun05_13-55-19_r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr/events.out.tfevents.1780667721.r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr.66.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:254312368a0ccd7a89f7585ba29b927550dbe428b9ec8d4068f4c5f9188ef7f3
3
- size 71065
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c23bc2bd9358633df6e83c25f1f71ea827284218b38d19f742ac22e6e6e7419
3
+ size 72463
runs/Jun05_13-55-19_r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr/events.out.tfevents.1780721202.r-iamhmh-vit-large-patch16-224-in21k-isic-classificat-2871wsmxr.66.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37257aaef4c0b4c64d5ee9580ab1d50732e6414d50ccddd21e7ccab2393e438c
3
+ size 921
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb9d501d15c28db08c76906d2b502e5bcb8c54fff4a713d9ce4188a020e205e8
3
+ size 5432
training_params.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data_path": "iamhmh/Skin-Lesion-Dataset-cls",
3
+ "model": "facebook/convnext-base-224-22k-1k",
4
+ "username": "iamhmh",
5
+ "lr": 5e-05,
6
+ "epochs": 3,
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": "test",
17
+ "logging_steps": -1,
18
+ "project_name": "autotrain-isic-classification",
19
+ "auto_find_batch_size": false,
20
+ "mixed_precision": "fp16",
21
+ "save_total_limit": 1,
22
+ "push_to_hub": true,
23
+ "eval_strategy": "epoch",
24
+ "image_column": "image",
25
+ "target_column": "label",
26
+ "log": "tensorboard",
27
+ "early_stopping_patience": 5,
28
+ "early_stopping_threshold": 0.01
29
+ }