Hugging Face's logo

prithivMLmods
/
IMAGENETTE

Image Classification
Transformers
Safetensors
English
siglip
ImageNet
SigLIP2
Classifier
Model card Files
xet
Community
prithivMLmods commited on
Commit
1ae945e
·
verified ·
1 Parent(s): b561273

Upload folder using huggingface_hub

Browse files
Files changed (20) hide show
  1. checkpoint-296/config.json +62 -0
  2. checkpoint-296/model.safetensors +3 -0
  3. checkpoint-296/optimizer.pt +3 -0
  4. checkpoint-296/preprocessor_config.json +24 -0
  5. checkpoint-296/rng_state.pth +3 -0
  6. checkpoint-296/scheduler.pt +3 -0
  7. checkpoint-296/trainer_state.json +44 -0
  8. checkpoint-296/training_args.bin +3 -0
  9. checkpoint-592/config.json +62 -0
  10. checkpoint-592/model.safetensors +3 -0
  11. checkpoint-592/optimizer.pt +3 -0
  12. checkpoint-592/preprocessor_config.json +24 -0
  13. checkpoint-592/rng_state.pth +3 -0
  14. checkpoint-592/scheduler.pt +3 -0
  15. checkpoint-592/trainer_state.json +61 -0
  16. checkpoint-592/training_args.bin +3 -0
  17. config.json +62 -0
  18. model.safetensors +3 -0
  19. preprocessor_config.json +24 -0
  20. training_args.bin +3 -0
checkpoint-296/config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "tench",
7
+ "1": "english springer",
8
+ "2": "cassette player",
9
+ "3": "chain saw",
10
+ "4": "church",
11
+ "5": "French horn",
12
+ "6": "garbage truck",
13
+ "7": "gas pump",
14
+ "8": "golf ball",
15
+ "9": "parachute"
16
+ },
17
+ "initializer_factor": 1.0,
18
+ "label2id": {
19
+ "French horn": 5,
20
+ "cassette player": 2,
21
+ "chain saw": 3,
22
+ "church": 4,
23
+ "english springer": 1,
24
+ "garbage truck": 6,
25
+ "gas pump": 7,
26
+ "golf ball": 8,
27
+ "parachute": 9,
28
+ "tench": 0
29
+ },
30
+ "model_type": "siglip",
31
+ "problem_type": "single_label_classification",
32
+ "text_config": {
33
+ "attention_dropout": 0.0,
34
+ "hidden_act": "gelu_pytorch_tanh",
35
+ "hidden_size": 768,
36
+ "intermediate_size": 3072,
37
+ "layer_norm_eps": 1e-06,
38
+ "max_position_embeddings": 64,
39
+ "model_type": "siglip_text_model",
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 12,
42
+ "projection_size": 768,
43
+ "torch_dtype": "float32",
44
+ "vocab_size": 256000
45
+ },
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.50.0",
48
+ "vision_config": {
49
+ "attention_dropout": 0.0,
50
+ "hidden_act": "gelu_pytorch_tanh",
51
+ "hidden_size": 768,
52
+ "image_size": 224,
53
+ "intermediate_size": 3072,
54
+ "layer_norm_eps": 1e-06,
55
+ "model_type": "siglip_vision_model",
56
+ "num_attention_heads": 12,
57
+ "num_channels": 3,
58
+ "num_hidden_layers": 12,
59
+ "patch_size": 16,
60
+ "torch_dtype": "float32"
61
+ }
62
+ }
checkpoint-296/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86bd1815c912481d35448d54f8a7f3dcefa402caf956a43b8148ef6f19dec886
3
+ size 371592608
checkpoint-296/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcd6cd926a33f42f5b54c9db75673b97143376191d8b60ce65de03503cc0668e
3
+ size 686604922
checkpoint-296/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-296/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a1d0e46354bb0d53fcdf77187900f80bdd829bdb0527f8099d614b2dfaa2628
3
+ size 14244
checkpoint-296/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8c53ec99196598b5384c135640d9a1497294ddd9a68c45508c8d3f78009fbb8
3
+ size 1064
checkpoint-296/trainer_state.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 296,
3
+ "best_metric": 0.6111576557159424,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-296",
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 296,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 1.0,
14
+ "eval_accuracy": 0.800612525081846,
15
+ "eval_loss": 0.6111576557159424,
16
+ "eval_model_preparation_time": 0.0024,
17
+ "eval_runtime": 147.7559,
18
+ "eval_samples_per_second": 64.085,
19
+ "eval_steps_per_second": 8.013,
20
+ "step": 296
21
+ }
22
+ ],
23
+ "logging_steps": 500,
24
+ "max_steps": 592,
25
+ "num_input_tokens_seen": 0,
26
+ "num_train_epochs": 2,
27
+ "save_steps": 500,
28
+ "stateful_callbacks": {
29
+ "TrainerControl": {
30
+ "args": {
31
+ "should_epoch_stop": false,
32
+ "should_evaluate": false,
33
+ "should_log": false,
34
+ "should_save": true,
35
+ "should_training_stop": false
36
+ },
37
+ "attributes": {}
38
+ }
39
+ },
40
+ "total_flos": 7.931334024376197e+17,
41
+ "train_batch_size": 32,
42
+ "trial_name": null,
43
+ "trial_params": null
44
+ }
checkpoint-296/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
3
+ size 5304
checkpoint-592/config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "tench",
7
+ "1": "english springer",
8
+ "2": "cassette player",
9
+ "3": "chain saw",
10
+ "4": "church",
11
+ "5": "French horn",
12
+ "6": "garbage truck",
13
+ "7": "gas pump",
14
+ "8": "golf ball",
15
+ "9": "parachute"
16
+ },
17
+ "initializer_factor": 1.0,
18
+ "label2id": {
19
+ "French horn": 5,
20
+ "cassette player": 2,
21
+ "chain saw": 3,
22
+ "church": 4,
23
+ "english springer": 1,
24
+ "garbage truck": 6,
25
+ "gas pump": 7,
26
+ "golf ball": 8,
27
+ "parachute": 9,
28
+ "tench": 0
29
+ },
30
+ "model_type": "siglip",
31
+ "problem_type": "single_label_classification",
32
+ "text_config": {
33
+ "attention_dropout": 0.0,
34
+ "hidden_act": "gelu_pytorch_tanh",
35
+ "hidden_size": 768,
36
+ "intermediate_size": 3072,
37
+ "layer_norm_eps": 1e-06,
38
+ "max_position_embeddings": 64,
39
+ "model_type": "siglip_text_model",
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 12,
42
+ "projection_size": 768,
43
+ "torch_dtype": "float32",
44
+ "vocab_size": 256000
45
+ },
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.50.0",
48
+ "vision_config": {
49
+ "attention_dropout": 0.0,
50
+ "hidden_act": "gelu_pytorch_tanh",
51
+ "hidden_size": 768,
52
+ "image_size": 224,
53
+ "intermediate_size": 3072,
54
+ "layer_norm_eps": 1e-06,
55
+ "model_type": "siglip_vision_model",
56
+ "num_attention_heads": 12,
57
+ "num_channels": 3,
58
+ "num_hidden_layers": 12,
59
+ "patch_size": 16,
60
+ "torch_dtype": "float32"
61
+ }
62
+ }
checkpoint-592/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98287447599e068f366b9e82c773e44061f0400c960e2aebcb5a57e0ce6b4cc0
3
+ size 371592608
checkpoint-592/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5cd24860b88483f459a1c89bec8fc2d73b80758ea9f27b8ac8ee0f5b4fb95fb
3
+ size 686604922
checkpoint-592/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-592/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d9e0d98277e19376f461cdd9dfe653831814dbe0d43c1fb72803686a8ff7c4e
3
+ size 14244
checkpoint-592/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d358ad8f4bcf9f8ecb35283c8a796a95fd1ddac9671328a85217005cbfca5ee
3
+ size 1064
checkpoint-592/trainer_state.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 592,
3
+ "best_metric": 0.1268095225095749,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-592",
5
+ "epoch": 2.0,
6
+ "eval_steps": 500,
7
+ "global_step": 592,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 1.0,
14
+ "eval_accuracy": 0.800612525081846,
15
+ "eval_loss": 0.6111576557159424,
16
+ "eval_model_preparation_time": 0.0024,
17
+ "eval_runtime": 147.7559,
18
+ "eval_samples_per_second": 64.085,
19
+ "eval_steps_per_second": 8.013,
20
+ "step": 296
21
+ },
22
+ {
23
+ "epoch": 1.689189189189189,
24
+ "grad_norm": 7.893478870391846,
25
+ "learning_rate": 3.3948339483394833e-05,
26
+ "loss": 0.6928,
27
+ "step": 500
28
+ },
29
+ {
30
+ "epoch": 2.0,
31
+ "eval_accuracy": 0.9590241841799556,
32
+ "eval_loss": 0.1268095225095749,
33
+ "eval_model_preparation_time": 0.0024,
34
+ "eval_runtime": 148.2374,
35
+ "eval_samples_per_second": 63.877,
36
+ "eval_steps_per_second": 7.987,
37
+ "step": 592
38
+ }
39
+ ],
40
+ "logging_steps": 500,
41
+ "max_steps": 592,
42
+ "num_input_tokens_seen": 0,
43
+ "num_train_epochs": 2,
44
+ "save_steps": 500,
45
+ "stateful_callbacks": {
46
+ "TrainerControl": {
47
+ "args": {
48
+ "should_epoch_stop": false,
49
+ "should_evaluate": false,
50
+ "should_log": false,
51
+ "should_save": true,
52
+ "should_training_stop": true
53
+ },
54
+ "attributes": {}
55
+ }
56
+ },
57
+ "total_flos": 1.5862668048752394e+18,
58
+ "train_batch_size": 32,
59
+ "trial_name": null,
60
+ "trial_params": null
61
+ }
checkpoint-592/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
3
+ size 5304
config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "tench",
7
+ "1": "english springer",
8
+ "2": "cassette player",
9
+ "3": "chain saw",
10
+ "4": "church",
11
+ "5": "French horn",
12
+ "6": "garbage truck",
13
+ "7": "gas pump",
14
+ "8": "golf ball",
15
+ "9": "parachute"
16
+ },
17
+ "initializer_factor": 1.0,
18
+ "label2id": {
19
+ "French horn": 5,
20
+ "cassette player": 2,
21
+ "chain saw": 3,
22
+ "church": 4,
23
+ "english springer": 1,
24
+ "garbage truck": 6,
25
+ "gas pump": 7,
26
+ "golf ball": 8,
27
+ "parachute": 9,
28
+ "tench": 0
29
+ },
30
+ "model_type": "siglip",
31
+ "problem_type": "single_label_classification",
32
+ "text_config": {
33
+ "attention_dropout": 0.0,
34
+ "hidden_act": "gelu_pytorch_tanh",
35
+ "hidden_size": 768,
36
+ "intermediate_size": 3072,
37
+ "layer_norm_eps": 1e-06,
38
+ "max_position_embeddings": 64,
39
+ "model_type": "siglip_text_model",
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 12,
42
+ "projection_size": 768,
43
+ "torch_dtype": "float32",
44
+ "vocab_size": 256000
45
+ },
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.50.0",
48
+ "vision_config": {
49
+ "attention_dropout": 0.0,
50
+ "hidden_act": "gelu_pytorch_tanh",
51
+ "hidden_size": 768,
52
+ "image_size": 224,
53
+ "intermediate_size": 3072,
54
+ "layer_norm_eps": 1e-06,
55
+ "model_type": "siglip_vision_model",
56
+ "num_attention_heads": 12,
57
+ "num_channels": 3,
58
+ "num_hidden_layers": 12,
59
+ "patch_size": 16,
60
+ "torch_dtype": "float32"
61
+ }
62
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98287447599e068f366b9e82c773e44061f0400c960e2aebcb5a57e0ce6b4cc0
3
+ size 371592608
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 sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
3
+ size 5304