Hugging Face's logo

prithivMLmods
/
GeoGuessr-55

Image Classification
Transformers
Safetensors
English
siglip
GeoGuessr
55
Loaction
RSI
Remote Sensing Instruments
Model card Files
xet
Community
prithivMLmods commited on
Commit
6546d03
·
verified ·
1 Parent(s): 14391d6

Upload folder using huggingface_hub

Browse files
Files changed (28) hide show
  1. checkpoint-1574/config.json +152 -0
  2. checkpoint-1574/model.safetensors +3 -0
  3. checkpoint-1574/optimizer.pt +3 -0
  4. checkpoint-1574/preprocessor_config.json +24 -0
  5. checkpoint-1574/rng_state.pth +3 -0
  6. checkpoint-1574/scheduler.pt +3 -0
  7. checkpoint-1574/trainer_state.json +75 -0
  8. checkpoint-1574/training_args.bin +3 -0
  9. checkpoint-2361/config.json +152 -0
  10. checkpoint-2361/model.safetensors +3 -0
  11. checkpoint-2361/optimizer.pt +3 -0
  12. checkpoint-2361/preprocessor_config.json +24 -0
  13. checkpoint-2361/rng_state.pth +3 -0
  14. checkpoint-2361/scheduler.pt +3 -0
  15. checkpoint-2361/trainer_state.json +92 -0
  16. checkpoint-2361/training_args.bin +3 -0
  17. checkpoint-787/config.json +152 -0
  18. checkpoint-787/model.safetensors +3 -0
  19. checkpoint-787/optimizer.pt +3 -0
  20. checkpoint-787/preprocessor_config.json +24 -0
  21. checkpoint-787/rng_state.pth +3 -0
  22. checkpoint-787/scheduler.pt +3 -0
  23. checkpoint-787/trainer_state.json +51 -0
  24. checkpoint-787/training_args.bin +3 -0
  25. config.json +152 -0
  26. model.safetensors +3 -0
  27. preprocessor_config.json +24 -0
  28. training_args.bin +3 -0
checkpoint-1574/config.json ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Argentina",
7
+ "1": "Australia",
8
+ "2": "Austria",
9
+ "3": "Bangladesh",
10
+ "4": "Belgium",
11
+ "5": "Bolivia",
12
+ "6": "Botswana",
13
+ "7": "Brazil",
14
+ "8": "Bulgaria",
15
+ "9": "Cambodia",
16
+ "10": "Canada",
17
+ "11": "Chile",
18
+ "12": "Colombia",
19
+ "13": "Croatia",
20
+ "14": "Czechia",
21
+ "15": "Denmark",
22
+ "16": "Finland",
23
+ "17": "France",
24
+ "18": "Germany",
25
+ "19": "Ghana",
26
+ "20": "Greece",
27
+ "21": "Hungary",
28
+ "22": "India",
29
+ "23": "Indonesia",
30
+ "24": "Ireland",
31
+ "25": "Israel",
32
+ "26": "Italy",
33
+ "27": "Japan",
34
+ "28": "Kenya",
35
+ "29": "Latvia",
36
+ "30": "Lithuania",
37
+ "31": "Malaysia",
38
+ "32": "Mexico",
39
+ "33": "Netherlands",
40
+ "34": "New Zealand",
41
+ "35": "Nigeria",
42
+ "36": "Norway",
43
+ "37": "Peru",
44
+ "38": "Philippines",
45
+ "39": "Poland",
46
+ "40": "Portugal",
47
+ "41": "Romania",
48
+ "42": "Russia",
49
+ "43": "Singapore",
50
+ "44": "Slovakia",
51
+ "45": "South Africa",
52
+ "46": "South Korea",
53
+ "47": "Spain",
54
+ "48": "Sweden",
55
+ "49": "Switzerland",
56
+ "50": "Taiwan",
57
+ "51": "Thailand",
58
+ "52": "Turkey",
59
+ "53": "Ukraine",
60
+ "54": "United Kingdom"
61
+ },
62
+ "initializer_factor": 1.0,
63
+ "label2id": {
64
+ "Argentina": 0,
65
+ "Australia": 1,
66
+ "Austria": 2,
67
+ "Bangladesh": 3,
68
+ "Belgium": 4,
69
+ "Bolivia": 5,
70
+ "Botswana": 6,
71
+ "Brazil": 7,
72
+ "Bulgaria": 8,
73
+ "Cambodia": 9,
74
+ "Canada": 10,
75
+ "Chile": 11,
76
+ "Colombia": 12,
77
+ "Croatia": 13,
78
+ "Czechia": 14,
79
+ "Denmark": 15,
80
+ "Finland": 16,
81
+ "France": 17,
82
+ "Germany": 18,
83
+ "Ghana": 19,
84
+ "Greece": 20,
85
+ "Hungary": 21,
86
+ "India": 22,
87
+ "Indonesia": 23,
88
+ "Ireland": 24,
89
+ "Israel": 25,
90
+ "Italy": 26,
91
+ "Japan": 27,
92
+ "Kenya": 28,
93
+ "Latvia": 29,
94
+ "Lithuania": 30,
95
+ "Malaysia": 31,
96
+ "Mexico": 32,
97
+ "Netherlands": 33,
98
+ "New Zealand": 34,
99
+ "Nigeria": 35,
100
+ "Norway": 36,
101
+ "Peru": 37,
102
+ "Philippines": 38,
103
+ "Poland": 39,
104
+ "Portugal": 40,
105
+ "Romania": 41,
106
+ "Russia": 42,
107
+ "Singapore": 43,
108
+ "Slovakia": 44,
109
+ "South Africa": 45,
110
+ "South Korea": 46,
111
+ "Spain": 47,
112
+ "Sweden": 48,
113
+ "Switzerland": 49,
114
+ "Taiwan": 50,
115
+ "Thailand": 51,
116
+ "Turkey": 52,
117
+ "Ukraine": 53,
118
+ "United Kingdom": 54
119
+ },
120
+ "model_type": "siglip",
121
+ "problem_type": "single_label_classification",
122
+ "text_config": {
123
+ "attention_dropout": 0.0,
124
+ "hidden_act": "gelu_pytorch_tanh",
125
+ "hidden_size": 768,
126
+ "intermediate_size": 3072,
127
+ "layer_norm_eps": 1e-06,
128
+ "max_position_embeddings": 64,
129
+ "model_type": "siglip_text_model",
130
+ "num_attention_heads": 12,
131
+ "num_hidden_layers": 12,
132
+ "projection_size": 768,
133
+ "torch_dtype": "float32",
134
+ "vocab_size": 256000
135
+ },
136
+ "torch_dtype": "float32",
137
+ "transformers_version": "4.50.0",
138
+ "vision_config": {
139
+ "attention_dropout": 0.0,
140
+ "hidden_act": "gelu_pytorch_tanh",
141
+ "hidden_size": 768,
142
+ "image_size": 224,
143
+ "intermediate_size": 3072,
144
+ "layer_norm_eps": 1e-06,
145
+ "model_type": "siglip_vision_model",
146
+ "num_attention_heads": 12,
147
+ "num_channels": 3,
148
+ "num_hidden_layers": 12,
149
+ "patch_size": 16,
150
+ "torch_dtype": "float32"
151
+ }
152
+ }
checkpoint-1574/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd3dc145ed6f65fe3d2a126078f9a03f785819743a5abd6de3c482d2790d0310
3
+ size 371731028
checkpoint-1574/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d03a0bc42163f1673784f932b9b2394eb58dbabbc22ba36fd04cda1726961e6
3
+ size 686881786
checkpoint-1574/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-1574/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1edc0283300fc7368991b7af2092faa43756f55556810a49b4c798872ca1e79
3
+ size 14244
checkpoint-1574/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af0fd93cff47ef8b40fbce9dddd62e78a04000f678a2fbafefa0a2a8aed5ae97
3
+ size 1064
checkpoint-1574/trainer_state.json ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 1574,
3
+ "best_metric": 1.3910918235778809,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-1574",
5
+ "epoch": 2.0,
6
+ "eval_steps": 500,
7
+ "global_step": 1574,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.6353240152477764,
14
+ "grad_norm": 46.04074478149414,
15
+ "learning_rate": 4.0263954997836435e-06,
16
+ "loss": 2.9627,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.5230524642289348,
22
+ "eval_loss": 1.8255969285964966,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 558.5967,
25
+ "eval_samples_per_second": 45.041,
26
+ "eval_steps_per_second": 5.63,
27
+ "step": 787
28
+ },
29
+ {
30
+ "epoch": 1.2706480304955527,
31
+ "grad_norm": 61.133846282958984,
32
+ "learning_rate": 2.9446127217654696e-06,
33
+ "loss": 1.9954,
34
+ "step": 1000
35
+ },
36
+ {
37
+ "epoch": 1.9059720457433291,
38
+ "grad_norm": 61.12163162231445,
39
+ "learning_rate": 1.8628299437472957e-06,
40
+ "loss": 1.6743,
41
+ "step": 1500
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.622456279809221,
46
+ "eval_loss": 1.3910918235778809,
47
+ "eval_model_preparation_time": 0.0023,
48
+ "eval_runtime": 558.2588,
49
+ "eval_samples_per_second": 45.069,
50
+ "eval_steps_per_second": 5.634,
51
+ "step": 1574
52
+ }
53
+ ],
54
+ "logging_steps": 500,
55
+ "max_steps": 2361,
56
+ "num_input_tokens_seen": 0,
57
+ "num_train_epochs": 3,
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": 4.216428849971036e+18,
72
+ "train_batch_size": 32,
73
+ "trial_name": null,
74
+ "trial_params": null
75
+ }
checkpoint-1574/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99363fe6141e80519841946ce66abb5645d0c7bf21a9d8794fcd2022bf69fcf6
3
+ size 5304
checkpoint-2361/config.json ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Argentina",
7
+ "1": "Australia",
8
+ "2": "Austria",
9
+ "3": "Bangladesh",
10
+ "4": "Belgium",
11
+ "5": "Bolivia",
12
+ "6": "Botswana",
13
+ "7": "Brazil",
14
+ "8": "Bulgaria",
15
+ "9": "Cambodia",
16
+ "10": "Canada",
17
+ "11": "Chile",
18
+ "12": "Colombia",
19
+ "13": "Croatia",
20
+ "14": "Czechia",
21
+ "15": "Denmark",
22
+ "16": "Finland",
23
+ "17": "France",
24
+ "18": "Germany",
25
+ "19": "Ghana",
26
+ "20": "Greece",
27
+ "21": "Hungary",
28
+ "22": "India",
29
+ "23": "Indonesia",
30
+ "24": "Ireland",
31
+ "25": "Israel",
32
+ "26": "Italy",
33
+ "27": "Japan",
34
+ "28": "Kenya",
35
+ "29": "Latvia",
36
+ "30": "Lithuania",
37
+ "31": "Malaysia",
38
+ "32": "Mexico",
39
+ "33": "Netherlands",
40
+ "34": "New Zealand",
41
+ "35": "Nigeria",
42
+ "36": "Norway",
43
+ "37": "Peru",
44
+ "38": "Philippines",
45
+ "39": "Poland",
46
+ "40": "Portugal",
47
+ "41": "Romania",
48
+ "42": "Russia",
49
+ "43": "Singapore",
50
+ "44": "Slovakia",
51
+ "45": "South Africa",
52
+ "46": "South Korea",
53
+ "47": "Spain",
54
+ "48": "Sweden",
55
+ "49": "Switzerland",
56
+ "50": "Taiwan",
57
+ "51": "Thailand",
58
+ "52": "Turkey",
59
+ "53": "Ukraine",
60
+ "54": "United Kingdom"
61
+ },
62
+ "initializer_factor": 1.0,
63
+ "label2id": {
64
+ "Argentina": 0,
65
+ "Australia": 1,
66
+ "Austria": 2,
67
+ "Bangladesh": 3,
68
+ "Belgium": 4,
69
+ "Bolivia": 5,
70
+ "Botswana": 6,
71
+ "Brazil": 7,
72
+ "Bulgaria": 8,
73
+ "Cambodia": 9,
74
+ "Canada": 10,
75
+ "Chile": 11,
76
+ "Colombia": 12,
77
+ "Croatia": 13,
78
+ "Czechia": 14,
79
+ "Denmark": 15,
80
+ "Finland": 16,
81
+ "France": 17,
82
+ "Germany": 18,
83
+ "Ghana": 19,
84
+ "Greece": 20,
85
+ "Hungary": 21,
86
+ "India": 22,
87
+ "Indonesia": 23,
88
+ "Ireland": 24,
89
+ "Israel": 25,
90
+ "Italy": 26,
91
+ "Japan": 27,
92
+ "Kenya": 28,
93
+ "Latvia": 29,
94
+ "Lithuania": 30,
95
+ "Malaysia": 31,
96
+ "Mexico": 32,
97
+ "Netherlands": 33,
98
+ "New Zealand": 34,
99
+ "Nigeria": 35,
100
+ "Norway": 36,
101
+ "Peru": 37,
102
+ "Philippines": 38,
103
+ "Poland": 39,
104
+ "Portugal": 40,
105
+ "Romania": 41,
106
+ "Russia": 42,
107
+ "Singapore": 43,
108
+ "Slovakia": 44,
109
+ "South Africa": 45,
110
+ "South Korea": 46,
111
+ "Spain": 47,
112
+ "Sweden": 48,
113
+ "Switzerland": 49,
114
+ "Taiwan": 50,
115
+ "Thailand": 51,
116
+ "Turkey": 52,
117
+ "Ukraine": 53,
118
+ "United Kingdom": 54
119
+ },
120
+ "model_type": "siglip",
121
+ "problem_type": "single_label_classification",
122
+ "text_config": {
123
+ "attention_dropout": 0.0,
124
+ "hidden_act": "gelu_pytorch_tanh",
125
+ "hidden_size": 768,
126
+ "intermediate_size": 3072,
127
+ "layer_norm_eps": 1e-06,
128
+ "max_position_embeddings": 64,
129
+ "model_type": "siglip_text_model",
130
+ "num_attention_heads": 12,
131
+ "num_hidden_layers": 12,
132
+ "projection_size": 768,
133
+ "torch_dtype": "float32",
134
+ "vocab_size": 256000
135
+ },
136
+ "torch_dtype": "float32",
137
+ "transformers_version": "4.50.0",
138
+ "vision_config": {
139
+ "attention_dropout": 0.0,
140
+ "hidden_act": "gelu_pytorch_tanh",
141
+ "hidden_size": 768,
142
+ "image_size": 224,
143
+ "intermediate_size": 3072,
144
+ "layer_norm_eps": 1e-06,
145
+ "model_type": "siglip_vision_model",
146
+ "num_attention_heads": 12,
147
+ "num_channels": 3,
148
+ "num_hidden_layers": 12,
149
+ "patch_size": 16,
150
+ "torch_dtype": "float32"
151
+ }
152
+ }
checkpoint-2361/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f80ca469db6911e55966d2d388d85f80348c1c36a42fb3deef7612c8a8436bfd
3
+ size 371731028
checkpoint-2361/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9023daafb72ad6adabfc7883f623d8008233c274948c585892df7606e79b447
3
+ size 686881786
checkpoint-2361/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-2361/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd9cef10fa4161c6b55bf848d6dcf55ff644a71f66a23f03473a8f239d7b3ab7
3
+ size 14244
checkpoint-2361/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77cdf7907a41b45ccfb7333171eb173d2a7cc6bc4f412be7d08f920d97316bbc
3
+ size 1064
checkpoint-2361/trainer_state.json ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 2361,
3
+ "best_metric": 1.263455867767334,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-2361",
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2361,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.6353240152477764,
14
+ "grad_norm": 46.04074478149414,
15
+ "learning_rate": 4.0263954997836435e-06,
16
+ "loss": 2.9627,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.5230524642289348,
22
+ "eval_loss": 1.8255969285964966,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 558.5967,
25
+ "eval_samples_per_second": 45.041,
26
+ "eval_steps_per_second": 5.63,
27
+ "step": 787
28
+ },
29
+ {
30
+ "epoch": 1.2706480304955527,
31
+ "grad_norm": 61.133846282958984,
32
+ "learning_rate": 2.9446127217654696e-06,
33
+ "loss": 1.9954,
34
+ "step": 1000
35
+ },
36
+ {
37
+ "epoch": 1.9059720457433291,
38
+ "grad_norm": 61.12163162231445,
39
+ "learning_rate": 1.8628299437472957e-06,
40
+ "loss": 1.6743,
41
+ "step": 1500
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.622456279809221,
46
+ "eval_loss": 1.3910918235778809,
47
+ "eval_model_preparation_time": 0.0023,
48
+ "eval_runtime": 558.2588,
49
+ "eval_samples_per_second": 45.069,
50
+ "eval_steps_per_second": 5.634,
51
+ "step": 1574
52
+ },
53
+ {
54
+ "epoch": 2.5412960609911055,
55
+ "grad_norm": 52.47379684448242,
56
+ "learning_rate": 7.810471657291216e-07,
57
+ "loss": 1.466,
58
+ "step": 2000
59
+ },
60
+ {
61
+ "epoch": 3.0,
62
+ "eval_accuracy": 0.648489666136725,
63
+ "eval_loss": 1.263455867767334,
64
+ "eval_model_preparation_time": 0.0023,
65
+ "eval_runtime": 559.8074,
66
+ "eval_samples_per_second": 44.944,
67
+ "eval_steps_per_second": 5.618,
68
+ "step": 2361
69
+ }
70
+ ],
71
+ "logging_steps": 500,
72
+ "max_steps": 2361,
73
+ "num_input_tokens_seen": 0,
74
+ "num_train_epochs": 3,
75
+ "save_steps": 500,
76
+ "stateful_callbacks": {
77
+ "TrainerControl": {
78
+ "args": {
79
+ "should_epoch_stop": false,
80
+ "should_evaluate": false,
81
+ "should_log": false,
82
+ "should_save": true,
83
+ "should_training_stop": true
84
+ },
85
+ "attributes": {}
86
+ }
87
+ },
88
+ "total_flos": 6.324643274956554e+18,
89
+ "train_batch_size": 32,
90
+ "trial_name": null,
91
+ "trial_params": null
92
+ }
checkpoint-2361/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99363fe6141e80519841946ce66abb5645d0c7bf21a9d8794fcd2022bf69fcf6
3
+ size 5304
checkpoint-787/config.json ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Argentina",
7
+ "1": "Australia",
8
+ "2": "Austria",
9
+ "3": "Bangladesh",
10
+ "4": "Belgium",
11
+ "5": "Bolivia",
12
+ "6": "Botswana",
13
+ "7": "Brazil",
14
+ "8": "Bulgaria",
15
+ "9": "Cambodia",
16
+ "10": "Canada",
17
+ "11": "Chile",
18
+ "12": "Colombia",
19
+ "13": "Croatia",
20
+ "14": "Czechia",
21
+ "15": "Denmark",
22
+ "16": "Finland",
23
+ "17": "France",
24
+ "18": "Germany",
25
+ "19": "Ghana",
26
+ "20": "Greece",
27
+ "21": "Hungary",
28
+ "22": "India",
29
+ "23": "Indonesia",
30
+ "24": "Ireland",
31
+ "25": "Israel",
32
+ "26": "Italy",
33
+ "27": "Japan",
34
+ "28": "Kenya",
35
+ "29": "Latvia",
36
+ "30": "Lithuania",
37
+ "31": "Malaysia",
38
+ "32": "Mexico",
39
+ "33": "Netherlands",
40
+ "34": "New Zealand",
41
+ "35": "Nigeria",
42
+ "36": "Norway",
43
+ "37": "Peru",
44
+ "38": "Philippines",
45
+ "39": "Poland",
46
+ "40": "Portugal",
47
+ "41": "Romania",
48
+ "42": "Russia",
49
+ "43": "Singapore",
50
+ "44": "Slovakia",
51
+ "45": "South Africa",
52
+ "46": "South Korea",
53
+ "47": "Spain",
54
+ "48": "Sweden",
55
+ "49": "Switzerland",
56
+ "50": "Taiwan",
57
+ "51": "Thailand",
58
+ "52": "Turkey",
59
+ "53": "Ukraine",
60
+ "54": "United Kingdom"
61
+ },
62
+ "initializer_factor": 1.0,
63
+ "label2id": {
64
+ "Argentina": 0,
65
+ "Australia": 1,
66
+ "Austria": 2,
67
+ "Bangladesh": 3,
68
+ "Belgium": 4,
69
+ "Bolivia": 5,
70
+ "Botswana": 6,
71
+ "Brazil": 7,
72
+ "Bulgaria": 8,
73
+ "Cambodia": 9,
74
+ "Canada": 10,
75
+ "Chile": 11,
76
+ "Colombia": 12,
77
+ "Croatia": 13,
78
+ "Czechia": 14,
79
+ "Denmark": 15,
80
+ "Finland": 16,
81
+ "France": 17,
82
+ "Germany": 18,
83
+ "Ghana": 19,
84
+ "Greece": 20,
85
+ "Hungary": 21,
86
+ "India": 22,
87
+ "Indonesia": 23,
88
+ "Ireland": 24,
89
+ "Israel": 25,
90
+ "Italy": 26,
91
+ "Japan": 27,
92
+ "Kenya": 28,
93
+ "Latvia": 29,
94
+ "Lithuania": 30,
95
+ "Malaysia": 31,
96
+ "Mexico": 32,
97
+ "Netherlands": 33,
98
+ "New Zealand": 34,
99
+ "Nigeria": 35,
100
+ "Norway": 36,
101
+ "Peru": 37,
102
+ "Philippines": 38,
103
+ "Poland": 39,
104
+ "Portugal": 40,
105
+ "Romania": 41,
106
+ "Russia": 42,
107
+ "Singapore": 43,
108
+ "Slovakia": 44,
109
+ "South Africa": 45,
110
+ "South Korea": 46,
111
+ "Spain": 47,
112
+ "Sweden": 48,
113
+ "Switzerland": 49,
114
+ "Taiwan": 50,
115
+ "Thailand": 51,
116
+ "Turkey": 52,
117
+ "Ukraine": 53,
118
+ "United Kingdom": 54
119
+ },
120
+ "model_type": "siglip",
121
+ "problem_type": "single_label_classification",
122
+ "text_config": {
123
+ "attention_dropout": 0.0,
124
+ "hidden_act": "gelu_pytorch_tanh",
125
+ "hidden_size": 768,
126
+ "intermediate_size": 3072,
127
+ "layer_norm_eps": 1e-06,
128
+ "max_position_embeddings": 64,
129
+ "model_type": "siglip_text_model",
130
+ "num_attention_heads": 12,
131
+ "num_hidden_layers": 12,
132
+ "projection_size": 768,
133
+ "torch_dtype": "float32",
134
+ "vocab_size": 256000
135
+ },
136
+ "torch_dtype": "float32",
137
+ "transformers_version": "4.50.0",
138
+ "vision_config": {
139
+ "attention_dropout": 0.0,
140
+ "hidden_act": "gelu_pytorch_tanh",
141
+ "hidden_size": 768,
142
+ "image_size": 224,
143
+ "intermediate_size": 3072,
144
+ "layer_norm_eps": 1e-06,
145
+ "model_type": "siglip_vision_model",
146
+ "num_attention_heads": 12,
147
+ "num_channels": 3,
148
+ "num_hidden_layers": 12,
149
+ "patch_size": 16,
150
+ "torch_dtype": "float32"
151
+ }
152
+ }
checkpoint-787/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:169b7680670222f32bec222dbf2151b302df30cfe5cff959ef1915987ca0e573
3
+ size 371731028
checkpoint-787/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:439b61c07b1d87f93a04fa9750c475c2c260b898557ddf6204d930d59299242b
3
+ size 686881786
checkpoint-787/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-787/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b541e07176f3bd787085f3d38c3a792455a360f7b721b69764558545c7d4a0e2
3
+ size 14244
checkpoint-787/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:786e25e0e5bb4d9102bbb1846f5ea6a0552133b22d14a318283de1b3795faf30
3
+ size 1064
checkpoint-787/trainer_state.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 787,
3
+ "best_metric": 1.8255969285964966,
4
+ "best_model_checkpoint": "siglip2-finetune-full/checkpoint-787",
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 787,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.6353240152477764,
14
+ "grad_norm": 46.04074478149414,
15
+ "learning_rate": 4.0263954997836435e-06,
16
+ "loss": 2.9627,
17
+ "step": 500
18
+ },
19
+ {
20
+ "epoch": 1.0,
21
+ "eval_accuracy": 0.5230524642289348,
22
+ "eval_loss": 1.8255969285964966,
23
+ "eval_model_preparation_time": 0.0023,
24
+ "eval_runtime": 558.5967,
25
+ "eval_samples_per_second": 45.041,
26
+ "eval_steps_per_second": 5.63,
27
+ "step": 787
28
+ }
29
+ ],
30
+ "logging_steps": 500,
31
+ "max_steps": 2361,
32
+ "num_input_tokens_seen": 0,
33
+ "num_train_epochs": 3,
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": 2.108214424985518e+18,
48
+ "train_batch_size": 32,
49
+ "trial_name": null,
50
+ "trial_params": null
51
+ }
checkpoint-787/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99363fe6141e80519841946ce66abb5645d0c7bf21a9d8794fcd2022bf69fcf6
3
+ size 5304
config.json ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SiglipForImageClassification"
4
+ ],
5
+ "id2label": {
6
+ "0": "Argentina",
7
+ "1": "Australia",
8
+ "2": "Austria",
9
+ "3": "Bangladesh",
10
+ "4": "Belgium",
11
+ "5": "Bolivia",
12
+ "6": "Botswana",
13
+ "7": "Brazil",
14
+ "8": "Bulgaria",
15
+ "9": "Cambodia",
16
+ "10": "Canada",
17
+ "11": "Chile",
18
+ "12": "Colombia",
19
+ "13": "Croatia",
20
+ "14": "Czechia",
21
+ "15": "Denmark",
22
+ "16": "Finland",
23
+ "17": "France",
24
+ "18": "Germany",
25
+ "19": "Ghana",
26
+ "20": "Greece",
27
+ "21": "Hungary",
28
+ "22": "India",
29
+ "23": "Indonesia",
30
+ "24": "Ireland",
31
+ "25": "Israel",
32
+ "26": "Italy",
33
+ "27": "Japan",
34
+ "28": "Kenya",
35
+ "29": "Latvia",
36
+ "30": "Lithuania",
37
+ "31": "Malaysia",
38
+ "32": "Mexico",
39
+ "33": "Netherlands",
40
+ "34": "New Zealand",
41
+ "35": "Nigeria",
42
+ "36": "Norway",
43
+ "37": "Peru",
44
+ "38": "Philippines",
45
+ "39": "Poland",
46
+ "40": "Portugal",
47
+ "41": "Romania",
48
+ "42": "Russia",
49
+ "43": "Singapore",
50
+ "44": "Slovakia",
51
+ "45": "South Africa",
52
+ "46": "South Korea",
53
+ "47": "Spain",
54
+ "48": "Sweden",
55
+ "49": "Switzerland",
56
+ "50": "Taiwan",
57
+ "51": "Thailand",
58
+ "52": "Turkey",
59
+ "53": "Ukraine",
60
+ "54": "United Kingdom"
61
+ },
62
+ "initializer_factor": 1.0,
63
+ "label2id": {
64
+ "Argentina": 0,
65
+ "Australia": 1,
66
+ "Austria": 2,
67
+ "Bangladesh": 3,
68
+ "Belgium": 4,
69
+ "Bolivia": 5,
70
+ "Botswana": 6,
71
+ "Brazil": 7,
72
+ "Bulgaria": 8,
73
+ "Cambodia": 9,
74
+ "Canada": 10,
75
+ "Chile": 11,
76
+ "Colombia": 12,
77
+ "Croatia": 13,
78
+ "Czechia": 14,
79
+ "Denmark": 15,
80
+ "Finland": 16,
81
+ "France": 17,
82
+ "Germany": 18,
83
+ "Ghana": 19,
84
+ "Greece": 20,
85
+ "Hungary": 21,
86
+ "India": 22,
87
+ "Indonesia": 23,
88
+ "Ireland": 24,
89
+ "Israel": 25,
90
+ "Italy": 26,
91
+ "Japan": 27,
92
+ "Kenya": 28,
93
+ "Latvia": 29,
94
+ "Lithuania": 30,
95
+ "Malaysia": 31,
96
+ "Mexico": 32,
97
+ "Netherlands": 33,
98
+ "New Zealand": 34,
99
+ "Nigeria": 35,
100
+ "Norway": 36,
101
+ "Peru": 37,
102
+ "Philippines": 38,
103
+ "Poland": 39,
104
+ "Portugal": 40,
105
+ "Romania": 41,
106
+ "Russia": 42,
107
+ "Singapore": 43,
108
+ "Slovakia": 44,
109
+ "South Africa": 45,
110
+ "South Korea": 46,
111
+ "Spain": 47,
112
+ "Sweden": 48,
113
+ "Switzerland": 49,
114
+ "Taiwan": 50,
115
+ "Thailand": 51,
116
+ "Turkey": 52,
117
+ "Ukraine": 53,
118
+ "United Kingdom": 54
119
+ },
120
+ "model_type": "siglip",
121
+ "problem_type": "single_label_classification",
122
+ "text_config": {
123
+ "attention_dropout": 0.0,
124
+ "hidden_act": "gelu_pytorch_tanh",
125
+ "hidden_size": 768,
126
+ "intermediate_size": 3072,
127
+ "layer_norm_eps": 1e-06,
128
+ "max_position_embeddings": 64,
129
+ "model_type": "siglip_text_model",
130
+ "num_attention_heads": 12,
131
+ "num_hidden_layers": 12,
132
+ "projection_size": 768,
133
+ "torch_dtype": "float32",
134
+ "vocab_size": 256000
135
+ },
136
+ "torch_dtype": "float32",
137
+ "transformers_version": "4.50.0",
138
+ "vision_config": {
139
+ "attention_dropout": 0.0,
140
+ "hidden_act": "gelu_pytorch_tanh",
141
+ "hidden_size": 768,
142
+ "image_size": 224,
143
+ "intermediate_size": 3072,
144
+ "layer_norm_eps": 1e-06,
145
+ "model_type": "siglip_vision_model",
146
+ "num_attention_heads": 12,
147
+ "num_channels": 3,
148
+ "num_hidden_layers": 12,
149
+ "patch_size": 16,
150
+ "torch_dtype": "float32"
151
+ }
152
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f80ca469db6911e55966d2d388d85f80348c1c36a42fb3deef7612c8a8436bfd
3
+ size 371731028
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 sha256:99363fe6141e80519841946ce66abb5645d0c7bf21a9d8794fcd2022bf69fcf6
3
+ size 5304