Hugging Face's logo

prithivMLmods
/
siglip2-mini-explicit-content

Image Classification
Transformers
Safetensors
English
siglip
SigLIP2
art
explicit-content-detection
media-filter
Anime
Model card Files
xet
Community
prithivMLmods commited on
Commit
4e6c880
·
verified ·
1 Parent(s): e4ecf13

Upload folder using huggingface_hub

Browse files
Files changed (36) hide show
  1. checkpoint-1800/config.json +52 -0
  2. checkpoint-1800/model.safetensors +3 -0
  3. checkpoint-1800/optimizer.pt +3 -0
  4. checkpoint-1800/preprocessor_config.json +24 -0
  5. checkpoint-1800/rng_state.pth +3 -0
  6. checkpoint-1800/scheduler.pt +3 -0
  7. checkpoint-1800/trainer_state.json +75 -0
  8. checkpoint-1800/training_args.bin +3 -0
  9. checkpoint-2700/config.json +52 -0
  10. checkpoint-2700/model.safetensors +3 -0
  11. checkpoint-2700/optimizer.pt +3 -0
  12. checkpoint-2700/preprocessor_config.json +24 -0
  13. checkpoint-2700/rng_state.pth +3 -0
  14. checkpoint-2700/scheduler.pt +3 -0
  15. checkpoint-2700/trainer_state.json +99 -0
  16. checkpoint-2700/training_args.bin +3 -0
  17. checkpoint-3600/config.json +52 -0
  18. checkpoint-3600/model.safetensors +3 -0
  19. checkpoint-3600/optimizer.pt +3 -0
  20. checkpoint-3600/preprocessor_config.json +24 -0
  21. checkpoint-3600/rng_state.pth +3 -0
  22. checkpoint-3600/scheduler.pt +3 -0
  23. checkpoint-3600/trainer_state.json +123 -0
  24. checkpoint-3600/training_args.bin +3 -0
  25. checkpoint-900/config.json +52 -0
  26. checkpoint-900/model.safetensors +3 -0
  27. checkpoint-900/optimizer.pt +3 -0
  28. checkpoint-900/preprocessor_config.json +24 -0
  29. checkpoint-900/rng_state.pth +3 -0
  30. checkpoint-900/scheduler.pt +3 -0
  31. checkpoint-900/trainer_state.json +51 -0
  32. checkpoint-900/training_args.bin +3 -0
  33. config.json +52 -0
  34. model.safetensors +3 -0
  35. preprocessor_config.json +24 -0
  36. training_args.bin +3 -0
checkpoint-1800/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Anime Picture",
7
+ "1": "Extincing & Sensual",
8
+ "2": "Hentai",
9
+ "3": "Pornography",
10
+ "4": "Safe for Work"
11
+ },
12
+ "initializer_factor": 1.0,
13
+ "label2id": {
14
+ "Anime Picture": 0,
15
+ "Extincing & Sensual": 1,
16
+ "Hentai": 2,
17
+ "Pornography": 3,
18
+ "Safe for Work": 4
19
+ },
20
+ "model_type": "siglip",
21
+ "problem_type": "single_label_classification",
22
+ "text_config": {
23
+ "attention_dropout": 0.0,
24
+ "hidden_act": "gelu_pytorch_tanh",
25
+ "hidden_size": 768,
26
+ "intermediate_size": 3072,
27
+ "layer_norm_eps": 1e-06,
28
+ "max_position_embeddings": 64,
29
+ "model_type": "siglip_text_model",
30
+ "num_attention_heads": 12,
31
+ "num_hidden_layers": 12,
32
+ "projection_size": 768,
33
+ "torch_dtype": "float32",
34
+ "vocab_size": 256000
35
+ },
36
+ "torch_dtype": "float32",
37
+ "transformers_version": "4.50.0",
38
+ "vision_config": {
39
+ "attention_dropout": 0.0,
40
+ "hidden_act": "gelu_pytorch_tanh",
41
+ "hidden_size": 768,
42
+ "image_size": 224,
43
+ "intermediate_size": 3072,
44
+ "layer_norm_eps": 1e-06,
45
+ "model_type": "siglip_vision_model",
46
+ "num_attention_heads": 12,
47
+ "num_channels": 3,
48
+ "num_hidden_layers": 12,
49
+ "patch_size": 16,
50
+ "torch_dtype": "float32"
51
+ }
52
+ }
checkpoint-1800/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:16d70552b77d85a1e40d6979ee94940c63b67f88f793e9e6378e3ee569928be6
3
+ size 371577220
checkpoint-1800/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7e34721002745baf5a925eae6cf35128ed323f1f72996d19f548f463a5f2d58
3
+ size 686574202
checkpoint-1800/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.5,
8
+ 0.5,
9
+ 0.5
10
+ ],
11
+ "image_processor_type": "SiglipImageProcessor",
12
+ "image_std": [
13
+ 0.5,
14
+ 0.5,
15
+ 0.5
16
+ ],
17
+ "processor_class": "SiglipProcessor",
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "height": 224,
22
+ "width": 224
23
+ }
24
+ }
checkpoint-1800/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb14b3612b9a2297307cf6e9608f2377bf8a6b696c374c21ce4c8e29efdf3c17
3
+ size 14244
checkpoint-1800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:83a61b561fd575fda31d375fcce1ad9cfa51e1c32499bb6cd9af451070a9ddc1
3
+ size 1064
checkpoint-1800/trainer_state.json ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 1800,
3
+ "best_metric": 0.45149940252304077,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-1800",
5
+ "epoch": 2.0,
6
+ "eval_steps": 500,
7
+ "global_step": 1800,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.5555555555555556,
14
+ "grad_norm": 10.926888465881348,
15
+ "learning_rate": 0.00017464788732394365,
16
+ "loss": 0.8678,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.7476031679866612,
22
+ "eval_loss": 0.6471553444862366,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 502.6747,
25
+ "eval_samples_per_second": 57.27,
26
+ "eval_steps_per_second": 7.16,
27
+ "step": 900
28
+ },
29
+ {
30
+ "epoch": 1.1111111111111112,
31
+ "grad_norm": 6.3170695304870605,
32
+ "learning_rate": 0.00014647887323943664,
33
+ "loss": 0.6826,
34
+ "step": 1000
35
+ },
36
+ {
37
+ "epoch": 1.6666666666666665,
38
+ "grad_norm": 7.355836868286133,
39
+ "learning_rate": 0.00011830985915492958,
40
+ "loss": 0.6019,
41
+ "step": 1500
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.8256912602473253,
46
+ "eval_loss": 0.45149940252304077,
47
+ "eval_model_preparation_time": 0.0023,
48
+ "eval_runtime": 507.0647,
49
+ "eval_samples_per_second": 56.774,
50
+ "eval_steps_per_second": 7.098,
51
+ "step": 1800
52
+ }
53
+ ],
54
+ "logging_steps": 500,
55
+ "max_steps": 3600,
56
+ "num_input_tokens_seen": 0,
57
+ "num_train_epochs": 4,
58
+ "save_steps": 500,
59
+ "stateful_callbacks": {
60
+ "TrainerControl": {
61
+ "args": {
62
+ "should_epoch_stop": false,
63
+ "should_evaluate": false,
64
+ "should_log": false,
65
+ "should_save": true,
66
+ "should_training_stop": false
67
+ },
68
+ "attributes": {}
69
+ }
70
+ },
71
+ "total_flos": 4.822426393255231e+18,
72
+ "train_batch_size": 32,
73
+ "trial_name": null,
74
+ "trial_params": null
75
+ }
checkpoint-1800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671f542b186da6dc8a0983b9f746a90b9e4d00ce219280a1dc1d7d090b52596
3
+ size 5304
checkpoint-2700/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Anime Picture",
7
+ "1": "Extincing & Sensual",
8
+ "2": "Hentai",
9
+ "3": "Pornography",
10
+ "4": "Safe for Work"
11
+ },
12
+ "initializer_factor": 1.0,
13
+ "label2id": {
14
+ "Anime Picture": 0,
15
+ "Extincing & Sensual": 1,
16
+ "Hentai": 2,
17
+ "Pornography": 3,
18
+ "Safe for Work": 4
19
+ },
20
+ "model_type": "siglip",
21
+ "problem_type": "single_label_classification",
22
+ "text_config": {
23
+ "attention_dropout": 0.0,
24
+ "hidden_act": "gelu_pytorch_tanh",
25
+ "hidden_size": 768,
26
+ "intermediate_size": 3072,
27
+ "layer_norm_eps": 1e-06,
28
+ "max_position_embeddings": 64,
29
+ "model_type": "siglip_text_model",
30
+ "num_attention_heads": 12,
31
+ "num_hidden_layers": 12,
32
+ "projection_size": 768,
33
+ "torch_dtype": "float32",
34
+ "vocab_size": 256000
35
+ },
36
+ "torch_dtype": "float32",
37
+ "transformers_version": "4.50.0",
38
+ "vision_config": {
39
+ "attention_dropout": 0.0,
40
+ "hidden_act": "gelu_pytorch_tanh",
41
+ "hidden_size": 768,
42
+ "image_size": 224,
43
+ "intermediate_size": 3072,
44
+ "layer_norm_eps": 1e-06,
45
+ "model_type": "siglip_vision_model",
46
+ "num_attention_heads": 12,
47
+ "num_channels": 3,
48
+ "num_hidden_layers": 12,
49
+ "patch_size": 16,
50
+ "torch_dtype": "float32"
51
+ }
52
+ }
checkpoint-2700/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca80d33bd593d8d4d9e3639581d8b8fa27fb47f33dc08d2d8a7acaba13ff1381
3
+ size 371577220
checkpoint-2700/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98af99905fa564471509b20926ed23f2b95bc06b3696d8d74bf77cb189f2511b
3
+ size 686574202
checkpoint-2700/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.5,
8
+ 0.5,
9
+ 0.5
10
+ ],
11
+ "image_processor_type": "SiglipImageProcessor",
12
+ "image_std": [
13
+ 0.5,
14
+ 0.5,
15
+ 0.5
16
+ ],
17
+ "processor_class": "SiglipProcessor",
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "height": 224,
22
+ "width": 224
23
+ }
24
+ }
checkpoint-2700/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9566622526d20d250adcffbeda4c238560643953d285006fb505485b5065e124
3
+ size 14244
checkpoint-2700/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04dd4ccc5aa584bc7d961aa4ae886a8e7307ea2efd1b57101b79aebe9aaca02d
3
+ size 1064
checkpoint-2700/trainer_state.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 2700,
3
+ "best_metric": 0.33306533098220825,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-2700",
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2700,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.5555555555555556,
14
+ "grad_norm": 10.926888465881348,
15
+ "learning_rate": 0.00017464788732394365,
16
+ "loss": 0.8678,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.7476031679866612,
22
+ "eval_loss": 0.6471553444862366,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 502.6747,
25
+ "eval_samples_per_second": 57.27,
26
+ "eval_steps_per_second": 7.16,
27
+ "step": 900
28
+ },
29
+ {
30
+ "epoch": 1.1111111111111112,
31
+ "grad_norm": 6.3170695304870605,
32
+ "learning_rate": 0.00014647887323943664,
33
+ "loss": 0.6826,
34
+ "step": 1000
35
+ },
36
+ {
37
+ "epoch": 1.6666666666666665,
38
+ "grad_norm": 7.355836868286133,
39
+ "learning_rate": 0.00011830985915492958,
40
+ "loss": 0.6019,
41
+ "step": 1500
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.8256912602473253,
46
+ "eval_loss": 0.45149940252304077,
47
+ "eval_model_preparation_time": 0.0023,
48
+ "eval_runtime": 507.0647,
49
+ "eval_samples_per_second": 56.774,
50
+ "eval_steps_per_second": 7.098,
51
+ "step": 1800
52
+ },
53
+ {
54
+ "epoch": 2.2222222222222223,
55
+ "grad_norm": 5.825477600097656,
56
+ "learning_rate": 9.014084507042254e-05,
57
+ "loss": 0.5111,
58
+ "step": 2000
59
+ },
60
+ {
61
+ "epoch": 2.7777777777777777,
62
+ "grad_norm": 4.699462890625,
63
+ "learning_rate": 6.197183098591549e-05,
64
+ "loss": 0.4575,
65
+ "step": 2500
66
+ },
67
+ {
68
+ "epoch": 3.0,
69
+ "eval_accuracy": 0.874183687647631,
70
+ "eval_loss": 0.33306533098220825,
71
+ "eval_model_preparation_time": 0.0023,
72
+ "eval_runtime": 507.0945,
73
+ "eval_samples_per_second": 56.77,
74
+ "eval_steps_per_second": 7.097,
75
+ "step": 2700
76
+ }
77
+ ],
78
+ "logging_steps": 500,
79
+ "max_steps": 3600,
80
+ "num_input_tokens_seen": 0,
81
+ "num_train_epochs": 4,
82
+ "save_steps": 500,
83
+ "stateful_callbacks": {
84
+ "TrainerControl": {
85
+ "args": {
86
+ "should_epoch_stop": false,
87
+ "should_evaluate": false,
88
+ "should_log": false,
89
+ "should_save": true,
90
+ "should_training_stop": false
91
+ },
92
+ "attributes": {}
93
+ }
94
+ },
95
+ "total_flos": 7.233639589882847e+18,
96
+ "train_batch_size": 32,
97
+ "trial_name": null,
98
+ "trial_params": null
99
+ }
checkpoint-2700/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671f542b186da6dc8a0983b9f746a90b9e4d00ce219280a1dc1d7d090b52596
3
+ size 5304
checkpoint-3600/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Anime Picture",
7
+ "1": "Extincing & Sensual",
8
+ "2": "Hentai",
9
+ "3": "Pornography",
10
+ "4": "Safe for Work"
11
+ },
12
+ "initializer_factor": 1.0,
13
+ "label2id": {
14
+ "Anime Picture": 0,
15
+ "Extincing & Sensual": 1,
16
+ "Hentai": 2,
17
+ "Pornography": 3,
18
+ "Safe for Work": 4
19
+ },
20
+ "model_type": "siglip",
21
+ "problem_type": "single_label_classification",
22
+ "text_config": {
23
+ "attention_dropout": 0.0,
24
+ "hidden_act": "gelu_pytorch_tanh",
25
+ "hidden_size": 768,
26
+ "intermediate_size": 3072,
27
+ "layer_norm_eps": 1e-06,
28
+ "max_position_embeddings": 64,
29
+ "model_type": "siglip_text_model",
30
+ "num_attention_heads": 12,
31
+ "num_hidden_layers": 12,
32
+ "projection_size": 768,
33
+ "torch_dtype": "float32",
34
+ "vocab_size": 256000
35
+ },
36
+ "torch_dtype": "float32",
37
+ "transformers_version": "4.50.0",
38
+ "vision_config": {
39
+ "attention_dropout": 0.0,
40
+ "hidden_act": "gelu_pytorch_tanh",
41
+ "hidden_size": 768,
42
+ "image_size": 224,
43
+ "intermediate_size": 3072,
44
+ "layer_norm_eps": 1e-06,
45
+ "model_type": "siglip_vision_model",
46
+ "num_attention_heads": 12,
47
+ "num_channels": 3,
48
+ "num_hidden_layers": 12,
49
+ "patch_size": 16,
50
+ "torch_dtype": "float32"
51
+ }
52
+ }
checkpoint-3600/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92d75fca7d65b4e8682454ffe8fa84258515895dabe0485ab9a6af05923c764a
3
+ size 371577220
checkpoint-3600/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d9bb687a3f6e34acdcc9da112b163fcc75adcdb23241cdd2a6ba038dd98c270
3
+ size 686574202
checkpoint-3600/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.5,
8
+ 0.5,
9
+ 0.5
10
+ ],
11
+ "image_processor_type": "SiglipImageProcessor",
12
+ "image_std": [
13
+ 0.5,
14
+ 0.5,
15
+ 0.5
16
+ ],
17
+ "processor_class": "SiglipProcessor",
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "height": 224,
22
+ "width": 224
23
+ }
24
+ }
checkpoint-3600/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9bafa1ecb49fb52c297a888e2843595661198c16f750e015c6b9333dd0a1daef
3
+ size 14244
checkpoint-3600/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3919b1b4358e7b56fa4bd18b6234f78f491d747457bdcd0dc2bb2a566127b391
3
+ size 1064
checkpoint-3600/trainer_state.json ADDED
@@ -0,0 +1,123 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 3600,
3
+ "best_metric": 0.2508643865585327,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-3600",
5
+ "epoch": 4.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3600,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.5555555555555556,
14
+ "grad_norm": 10.926888465881348,
15
+ "learning_rate": 0.00017464788732394365,
16
+ "loss": 0.8678,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.7476031679866612,
22
+ "eval_loss": 0.6471553444862366,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 502.6747,
25
+ "eval_samples_per_second": 57.27,
26
+ "eval_steps_per_second": 7.16,
27
+ "step": 900
28
+ },
29
+ {
30
+ "epoch": 1.1111111111111112,
31
+ "grad_norm": 6.3170695304870605,
32
+ "learning_rate": 0.00014647887323943664,
33
+ "loss": 0.6826,
34
+ "step": 1000
35
+ },
36
+ {
37
+ "epoch": 1.6666666666666665,
38
+ "grad_norm": 7.355836868286133,
39
+ "learning_rate": 0.00011830985915492958,
40
+ "loss": 0.6019,
41
+ "step": 1500
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.8256912602473253,
46
+ "eval_loss": 0.45149940252304077,
47
+ "eval_model_preparation_time": 0.0023,
48
+ "eval_runtime": 507.0647,
49
+ "eval_samples_per_second": 56.774,
50
+ "eval_steps_per_second": 7.098,
51
+ "step": 1800
52
+ },
53
+ {
54
+ "epoch": 2.2222222222222223,
55
+ "grad_norm": 5.825477600097656,
56
+ "learning_rate": 9.014084507042254e-05,
57
+ "loss": 0.5111,
58
+ "step": 2000
59
+ },
60
+ {
61
+ "epoch": 2.7777777777777777,
62
+ "grad_norm": 4.699462890625,
63
+ "learning_rate": 6.197183098591549e-05,
64
+ "loss": 0.4575,
65
+ "step": 2500
66
+ },
67
+ {
68
+ "epoch": 3.0,
69
+ "eval_accuracy": 0.874183687647631,
70
+ "eval_loss": 0.33306533098220825,
71
+ "eval_model_preparation_time": 0.0023,
72
+ "eval_runtime": 507.0945,
73
+ "eval_samples_per_second": 56.77,
74
+ "eval_steps_per_second": 7.097,
75
+ "step": 2700
76
+ },
77
+ {
78
+ "epoch": 3.3333333333333335,
79
+ "grad_norm": 6.299696922302246,
80
+ "learning_rate": 3.380281690140845e-05,
81
+ "loss": 0.3819,
82
+ "step": 3000
83
+ },
84
+ {
85
+ "epoch": 3.888888888888889,
86
+ "grad_norm": 5.8718953132629395,
87
+ "learning_rate": 5.6338028169014084e-06,
88
+ "loss": 0.3212,
89
+ "step": 3500
90
+ },
91
+ {
92
+ "epoch": 4.0,
93
+ "eval_accuracy": 0.9073919688759206,
94
+ "eval_loss": 0.2508643865585327,
95
+ "eval_model_preparation_time": 0.0023,
96
+ "eval_runtime": 500.7107,
97
+ "eval_samples_per_second": 57.494,
98
+ "eval_steps_per_second": 7.188,
99
+ "step": 3600
100
+ }
101
+ ],
102
+ "logging_steps": 500,
103
+ "max_steps": 3600,
104
+ "num_input_tokens_seen": 0,
105
+ "num_train_epochs": 4,
106
+ "save_steps": 500,
107
+ "stateful_callbacks": {
108
+ "TrainerControl": {
109
+ "args": {
110
+ "should_epoch_stop": false,
111
+ "should_evaluate": false,
112
+ "should_log": false,
113
+ "should_save": true,
114
+ "should_training_stop": true
115
+ },
116
+ "attributes": {}
117
+ }
118
+ },
119
+ "total_flos": 9.644852786510463e+18,
120
+ "train_batch_size": 32,
121
+ "trial_name": null,
122
+ "trial_params": null
123
+ }
checkpoint-3600/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671f542b186da6dc8a0983b9f746a90b9e4d00ce219280a1dc1d7d090b52596
3
+ size 5304
checkpoint-900/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Anime Picture",
7
+ "1": "Extincing & Sensual",
8
+ "2": "Hentai",
9
+ "3": "Pornography",
10
+ "4": "Safe for Work"
11
+ },
12
+ "initializer_factor": 1.0,
13
+ "label2id": {
14
+ "Anime Picture": 0,
15
+ "Extincing & Sensual": 1,
16
+ "Hentai": 2,
17
+ "Pornography": 3,
18
+ "Safe for Work": 4
19
+ },
20
+ "model_type": "siglip",
21
+ "problem_type": "single_label_classification",
22
+ "text_config": {
23
+ "attention_dropout": 0.0,
24
+ "hidden_act": "gelu_pytorch_tanh",
25
+ "hidden_size": 768,
26
+ "intermediate_size": 3072,
27
+ "layer_norm_eps": 1e-06,
28
+ "max_position_embeddings": 64,
29
+ "model_type": "siglip_text_model",
30
+ "num_attention_heads": 12,
31
+ "num_hidden_layers": 12,
32
+ "projection_size": 768,
33
+ "torch_dtype": "float32",
34
+ "vocab_size": 256000
35
+ },
36
+ "torch_dtype": "float32",
37
+ "transformers_version": "4.50.0",
38
+ "vision_config": {
39
+ "attention_dropout": 0.0,
40
+ "hidden_act": "gelu_pytorch_tanh",
41
+ "hidden_size": 768,
42
+ "image_size": 224,
43
+ "intermediate_size": 3072,
44
+ "layer_norm_eps": 1e-06,
45
+ "model_type": "siglip_vision_model",
46
+ "num_attention_heads": 12,
47
+ "num_channels": 3,
48
+ "num_hidden_layers": 12,
49
+ "patch_size": 16,
50
+ "torch_dtype": "float32"
51
+ }
52
+ }
checkpoint-900/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbbf73ab8873361a304d718a527dc0b9b2076612f5d79153e01fce4c524d0f18
3
+ size 371577220
checkpoint-900/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae1bb56f54dfb12fbc23ba66396dc3f1e15469bd7d6d68921817c47201a14729
3
+ size 686574202
checkpoint-900/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.5,
8
+ 0.5,
9
+ 0.5
10
+ ],
11
+ "image_processor_type": "SiglipImageProcessor",
12
+ "image_std": [
13
+ 0.5,
14
+ 0.5,
15
+ 0.5
16
+ ],
17
+ "processor_class": "SiglipProcessor",
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "height": 224,
22
+ "width": 224
23
+ }
24
+ }
checkpoint-900/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ad3e8f94a9f005f2eed6113dacd099e50cea541cae98c1ef3160d770074a25e
3
+ size 14244
checkpoint-900/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f956615357be28cb36f41d01e3de98ff185a5449f7b4d4a950dacfdb0204e823
3
+ size 1064
checkpoint-900/trainer_state.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 900,
3
+ "best_metric": 0.6471553444862366,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-900",
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 900,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.5555555555555556,
14
+ "grad_norm": 10.926888465881348,
15
+ "learning_rate": 0.00017464788732394365,
16
+ "loss": 0.8678,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.7476031679866612,
22
+ "eval_loss": 0.6471553444862366,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 502.6747,
25
+ "eval_samples_per_second": 57.27,
26
+ "eval_steps_per_second": 7.16,
27
+ "step": 900
28
+ }
29
+ ],
30
+ "logging_steps": 500,
31
+ "max_steps": 3600,
32
+ "num_input_tokens_seen": 0,
33
+ "num_train_epochs": 4,
34
+ "save_steps": 500,
35
+ "stateful_callbacks": {
36
+ "TrainerControl": {
37
+ "args": {
38
+ "should_epoch_stop": false,
39
+ "should_evaluate": false,
40
+ "should_log": false,
41
+ "should_save": true,
42
+ "should_training_stop": false
43
+ },
44
+ "attributes": {}
45
+ }
46
+ },
47
+ "total_flos": 2.411213196627616e+18,
48
+ "train_batch_size": 32,
49
+ "trial_name": null,
50
+ "trial_params": null
51
+ }
checkpoint-900/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671f542b186da6dc8a0983b9f746a90b9e4d00ce219280a1dc1d7d090b52596
3
+ size 5304
config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Anime Picture",
7
+ "1": "Extincing & Sensual",
8
+ "2": "Hentai",
9
+ "3": "Pornography",
10
+ "4": "Safe for Work"
11
+ },
12
+ "initializer_factor": 1.0,
13
+ "label2id": {
14
+ "Anime Picture": 0,
15
+ "Extincing & Sensual": 1,
16
+ "Hentai": 2,
17
+ "Pornography": 3,
18
+ "Safe for Work": 4
19
+ },
20
+ "model_type": "siglip",
21
+ "problem_type": "single_label_classification",
22
+ "text_config": {
23
+ "attention_dropout": 0.0,
24
+ "hidden_act": "gelu_pytorch_tanh",
25
+ "hidden_size": 768,
26
+ "intermediate_size": 3072,
27
+ "layer_norm_eps": 1e-06,
28
+ "max_position_embeddings": 64,
29
+ "model_type": "siglip_text_model",
30
+ "num_attention_heads": 12,
31
+ "num_hidden_layers": 12,
32
+ "projection_size": 768,
33
+ "torch_dtype": "float32",
34
+ "vocab_size": 256000
35
+ },
36
+ "torch_dtype": "float32",
37
+ "transformers_version": "4.50.0",
38
+ "vision_config": {
39
+ "attention_dropout": 0.0,
40
+ "hidden_act": "gelu_pytorch_tanh",
41
+ "hidden_size": 768,
42
+ "image_size": 224,
43
+ "intermediate_size": 3072,
44
+ "layer_norm_eps": 1e-06,
45
+ "model_type": "siglip_vision_model",
46
+ "num_attention_heads": 12,
47
+ "num_channels": 3,
48
+ "num_hidden_layers": 12,
49
+ "patch_size": 16,
50
+ "torch_dtype": "float32"
51
+ }
52
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92d75fca7d65b4e8682454ffe8fa84258515895dabe0485ab9a6af05923c764a
3
+ size 371577220
preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": null,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.5,
8
+ 0.5,
9
+ 0.5
10
+ ],
11
+ "image_processor_type": "SiglipImageProcessor",
12
+ "image_std": [
13
+ 0.5,
14
+ 0.5,
15
+ 0.5
16
+ ],
17
+ "processor_class": "SiglipProcessor",
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "height": 224,
22
+ "width": 224
23
+ }
24
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4671f542b186da6dc8a0983b9f746a90b9e4d00ce219280a1dc1d7d090b52596
3
+ size 5304