Upload folder using huggingface_hub
Browse files- checkpoint-3227/config.json +128 -0
- checkpoint-3227/model.safetensors +3 -0
- checkpoint-3227/optimizer.pt +3 -0
- checkpoint-3227/preprocessor_config.json +24 -0
- checkpoint-3227/rng_state.pth +3 -0
- checkpoint-3227/scheduler.pt +3 -0
- checkpoint-3227/trainer_state.json +86 -0
- checkpoint-3227/training_args.bin +3 -0
- checkpoint-6454/config.json +128 -0
- checkpoint-6454/model.safetensors +3 -0
- checkpoint-6454/optimizer.pt +3 -0
- checkpoint-6454/preprocessor_config.json +24 -0
- checkpoint-6454/rng_state.pth +3 -0
- checkpoint-6454/scheduler.pt +3 -0
- checkpoint-6454/trainer_state.json +138 -0
- checkpoint-6454/training_args.bin +3 -0
- config.json +128 -0
- model.safetensors +3 -0
- preprocessor_config.json +24 -0
- training_args.bin +3 -0
checkpoint-3227/config.json
ADDED
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"SiglipForImageClassification"
|
4 |
+
],
|
5 |
+
"id2label": {
|
6 |
+
"0": "None",
|
7 |
+
"1": "abstract",
|
8 |
+
"2": "advertisement",
|
9 |
+
"3": "allegorical painting",
|
10 |
+
"4": "animal painting",
|
11 |
+
"5": "battle painting",
|
12 |
+
"6": "bijinga",
|
13 |
+
"7": "bird-and-flower painting",
|
14 |
+
"8": "calligraphy",
|
15 |
+
"9": "capriccio",
|
16 |
+
"10": "caricature",
|
17 |
+
"11": "cityscape",
|
18 |
+
"12": "cloudscape",
|
19 |
+
"13": "design",
|
20 |
+
"14": "figurative",
|
21 |
+
"15": "flower painting",
|
22 |
+
"16": "genre painting",
|
23 |
+
"17": "history painting",
|
24 |
+
"18": "illustration",
|
25 |
+
"19": "interior",
|
26 |
+
"20": "landscape",
|
27 |
+
"21": "literary painting",
|
28 |
+
"22": "marina",
|
29 |
+
"23": "miniature",
|
30 |
+
"24": "mythological painting",
|
31 |
+
"25": "nude painting (nu)",
|
32 |
+
"26": "panorama",
|
33 |
+
"27": "pastorale",
|
34 |
+
"28": "portrait",
|
35 |
+
"29": "poster",
|
36 |
+
"30": "quadratura",
|
37 |
+
"31": "religious painting",
|
38 |
+
"32": "self-portrait",
|
39 |
+
"33": "shan shui",
|
40 |
+
"34": "sketch and study",
|
41 |
+
"35": "still life",
|
42 |
+
"36": "symbolic painting",
|
43 |
+
"37": "tessellation",
|
44 |
+
"38": "urushi-e",
|
45 |
+
"39": "vanitas",
|
46 |
+
"40": "veduta",
|
47 |
+
"41": "wildlife painting",
|
48 |
+
"42": "yakusha-e"
|
49 |
+
},
|
50 |
+
"initializer_factor": 1.0,
|
51 |
+
"label2id": {
|
52 |
+
"None": 0,
|
53 |
+
"abstract": 1,
|
54 |
+
"advertisement": 2,
|
55 |
+
"allegorical painting": 3,
|
56 |
+
"animal painting": 4,
|
57 |
+
"battle painting": 5,
|
58 |
+
"bijinga": 6,
|
59 |
+
"bird-and-flower painting": 7,
|
60 |
+
"calligraphy": 8,
|
61 |
+
"capriccio": 9,
|
62 |
+
"caricature": 10,
|
63 |
+
"cityscape": 11,
|
64 |
+
"cloudscape": 12,
|
65 |
+
"design": 13,
|
66 |
+
"figurative": 14,
|
67 |
+
"flower painting": 15,
|
68 |
+
"genre painting": 16,
|
69 |
+
"history painting": 17,
|
70 |
+
"illustration": 18,
|
71 |
+
"interior": 19,
|
72 |
+
"landscape": 20,
|
73 |
+
"literary painting": 21,
|
74 |
+
"marina": 22,
|
75 |
+
"miniature": 23,
|
76 |
+
"mythological painting": 24,
|
77 |
+
"nude painting (nu)": 25,
|
78 |
+
"panorama": 26,
|
79 |
+
"pastorale": 27,
|
80 |
+
"portrait": 28,
|
81 |
+
"poster": 29,
|
82 |
+
"quadratura": 30,
|
83 |
+
"religious painting": 31,
|
84 |
+
"self-portrait": 32,
|
85 |
+
"shan shui": 33,
|
86 |
+
"sketch and study": 34,
|
87 |
+
"still life": 35,
|
88 |
+
"symbolic painting": 36,
|
89 |
+
"tessellation": 37,
|
90 |
+
"urushi-e": 38,
|
91 |
+
"vanitas": 39,
|
92 |
+
"veduta": 40,
|
93 |
+
"wildlife painting": 41,
|
94 |
+
"yakusha-e": 42
|
95 |
+
},
|
96 |
+
"model_type": "siglip",
|
97 |
+
"problem_type": "single_label_classification",
|
98 |
+
"text_config": {
|
99 |
+
"attention_dropout": 0.0,
|
100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
101 |
+
"hidden_size": 768,
|
102 |
+
"intermediate_size": 3072,
|
103 |
+
"layer_norm_eps": 1e-06,
|
104 |
+
"max_position_embeddings": 64,
|
105 |
+
"model_type": "siglip_text_model",
|
106 |
+
"num_attention_heads": 12,
|
107 |
+
"num_hidden_layers": 12,
|
108 |
+
"projection_size": 768,
|
109 |
+
"torch_dtype": "float32",
|
110 |
+
"vocab_size": 256000
|
111 |
+
},
|
112 |
+
"torch_dtype": "float32",
|
113 |
+
"transformers_version": "4.50.0",
|
114 |
+
"vision_config": {
|
115 |
+
"attention_dropout": 0.0,
|
116 |
+
"hidden_act": "gelu_pytorch_tanh",
|
117 |
+
"hidden_size": 768,
|
118 |
+
"image_size": 224,
|
119 |
+
"intermediate_size": 3072,
|
120 |
+
"layer_norm_eps": 1e-06,
|
121 |
+
"model_type": "siglip_vision_model",
|
122 |
+
"num_attention_heads": 12,
|
123 |
+
"num_channels": 3,
|
124 |
+
"num_hidden_layers": 12,
|
125 |
+
"patch_size": 16,
|
126 |
+
"torch_dtype": "float32"
|
127 |
+
}
|
128 |
+
}
|
checkpoint-3227/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4bec70912e98cf9e74a423d9e2bb8b2fef9efb2dad250eb414568f756fbf1b8b
|
3 |
+
size 371694116
|
checkpoint-3227/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:74a3128bc06b04360669710cc3fb3040c1739ffe62660395241c2b7e87e46bbb
|
3 |
+
size 686807930
|
checkpoint-3227/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-3227/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a4b74d66497e897c1dcde91c6e8ec862f51c4b8bc7dc0770387f9ed113a14bf8
|
3 |
+
size 14244
|
checkpoint-3227/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1977d96ae07c0c31b449c6ccdf4b2fab09881316e7e4d838f6dfb53a4f743190
|
3 |
+
size 1064
|
checkpoint-3227/trainer_state.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_global_step": 3227,
|
3 |
+
"best_metric": 1.0367238521575928,
|
4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-3227",
|
5 |
+
"epoch": 1.0,
|
6 |
+
"eval_steps": 500,
|
7 |
+
"global_step": 3227,
|
8 |
+
"is_hyper_param_search": false,
|
9 |
+
"is_local_process_zero": true,
|
10 |
+
"is_world_process_zero": true,
|
11 |
+
"log_history": [
|
12 |
+
{
|
13 |
+
"epoch": 0.1549426712116517,
|
14 |
+
"grad_norm": 55.33502197265625,
|
15 |
+
"learning_rate": 2.7891942535915053e-06,
|
16 |
+
"loss": 1.9844,
|
17 |
+
"step": 500
|
18 |
+
},
|
19 |
+
{
|
20 |
+
"epoch": 0.3098853424233034,
|
21 |
+
"grad_norm": 37.46580505371094,
|
22 |
+
"learning_rate": 2.5549656464709556e-06,
|
23 |
+
"loss": 1.4259,
|
24 |
+
"step": 1000
|
25 |
+
},
|
26 |
+
{
|
27 |
+
"epoch": 0.4648280136349551,
|
28 |
+
"grad_norm": 41.121490478515625,
|
29 |
+
"learning_rate": 2.320737039350406e-06,
|
30 |
+
"loss": 1.3044,
|
31 |
+
"step": 1500
|
32 |
+
},
|
33 |
+
{
|
34 |
+
"epoch": 0.6197706848466068,
|
35 |
+
"grad_norm": 37.83391189575195,
|
36 |
+
"learning_rate": 2.0865084322298563e-06,
|
37 |
+
"loss": 1.2348,
|
38 |
+
"step": 2000
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.7747133560582584,
|
42 |
+
"grad_norm": 41.89641571044922,
|
43 |
+
"learning_rate": 1.8522798251093069e-06,
|
44 |
+
"loss": 1.1889,
|
45 |
+
"step": 2500
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.9296560272699101,
|
49 |
+
"grad_norm": 38.50979995727539,
|
50 |
+
"learning_rate": 1.6180512179887572e-06,
|
51 |
+
"loss": 1.1496,
|
52 |
+
"step": 3000
|
53 |
+
},
|
54 |
+
{
|
55 |
+
"epoch": 1.0,
|
56 |
+
"eval_accuracy": 0.6981598062953995,
|
57 |
+
"eval_loss": 1.0367238521575928,
|
58 |
+
"eval_model_preparation_time": 0.0038,
|
59 |
+
"eval_runtime": 1328.8007,
|
60 |
+
"eval_samples_per_second": 77.702,
|
61 |
+
"eval_steps_per_second": 9.713,
|
62 |
+
"step": 3227
|
63 |
+
}
|
64 |
+
],
|
65 |
+
"logging_steps": 500,
|
66 |
+
"max_steps": 6454,
|
67 |
+
"num_input_tokens_seen": 0,
|
68 |
+
"num_train_epochs": 2,
|
69 |
+
"save_steps": 500,
|
70 |
+
"stateful_callbacks": {
|
71 |
+
"TrainerControl": {
|
72 |
+
"args": {
|
73 |
+
"should_epoch_stop": false,
|
74 |
+
"should_evaluate": false,
|
75 |
+
"should_log": false,
|
76 |
+
"should_save": true,
|
77 |
+
"should_training_stop": false
|
78 |
+
},
|
79 |
+
"attributes": {}
|
80 |
+
}
|
81 |
+
},
|
82 |
+
"total_flos": 8.650695088885248e+18,
|
83 |
+
"train_batch_size": 32,
|
84 |
+
"trial_name": null,
|
85 |
+
"trial_params": null
|
86 |
+
}
|
checkpoint-3227/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:efa4c1f3eb59005587e95669315d89aa03d881830745eaa01a42988ac6c33c1b
|
3 |
+
size 5304
|
checkpoint-6454/config.json
ADDED
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"SiglipForImageClassification"
|
4 |
+
],
|
5 |
+
"id2label": {
|
6 |
+
"0": "None",
|
7 |
+
"1": "abstract",
|
8 |
+
"2": "advertisement",
|
9 |
+
"3": "allegorical painting",
|
10 |
+
"4": "animal painting",
|
11 |
+
"5": "battle painting",
|
12 |
+
"6": "bijinga",
|
13 |
+
"7": "bird-and-flower painting",
|
14 |
+
"8": "calligraphy",
|
15 |
+
"9": "capriccio",
|
16 |
+
"10": "caricature",
|
17 |
+
"11": "cityscape",
|
18 |
+
"12": "cloudscape",
|
19 |
+
"13": "design",
|
20 |
+
"14": "figurative",
|
21 |
+
"15": "flower painting",
|
22 |
+
"16": "genre painting",
|
23 |
+
"17": "history painting",
|
24 |
+
"18": "illustration",
|
25 |
+
"19": "interior",
|
26 |
+
"20": "landscape",
|
27 |
+
"21": "literary painting",
|
28 |
+
"22": "marina",
|
29 |
+
"23": "miniature",
|
30 |
+
"24": "mythological painting",
|
31 |
+
"25": "nude painting (nu)",
|
32 |
+
"26": "panorama",
|
33 |
+
"27": "pastorale",
|
34 |
+
"28": "portrait",
|
35 |
+
"29": "poster",
|
36 |
+
"30": "quadratura",
|
37 |
+
"31": "religious painting",
|
38 |
+
"32": "self-portrait",
|
39 |
+
"33": "shan shui",
|
40 |
+
"34": "sketch and study",
|
41 |
+
"35": "still life",
|
42 |
+
"36": "symbolic painting",
|
43 |
+
"37": "tessellation",
|
44 |
+
"38": "urushi-e",
|
45 |
+
"39": "vanitas",
|
46 |
+
"40": "veduta",
|
47 |
+
"41": "wildlife painting",
|
48 |
+
"42": "yakusha-e"
|
49 |
+
},
|
50 |
+
"initializer_factor": 1.0,
|
51 |
+
"label2id": {
|
52 |
+
"None": 0,
|
53 |
+
"abstract": 1,
|
54 |
+
"advertisement": 2,
|
55 |
+
"allegorical painting": 3,
|
56 |
+
"animal painting": 4,
|
57 |
+
"battle painting": 5,
|
58 |
+
"bijinga": 6,
|
59 |
+
"bird-and-flower painting": 7,
|
60 |
+
"calligraphy": 8,
|
61 |
+
"capriccio": 9,
|
62 |
+
"caricature": 10,
|
63 |
+
"cityscape": 11,
|
64 |
+
"cloudscape": 12,
|
65 |
+
"design": 13,
|
66 |
+
"figurative": 14,
|
67 |
+
"flower painting": 15,
|
68 |
+
"genre painting": 16,
|
69 |
+
"history painting": 17,
|
70 |
+
"illustration": 18,
|
71 |
+
"interior": 19,
|
72 |
+
"landscape": 20,
|
73 |
+
"literary painting": 21,
|
74 |
+
"marina": 22,
|
75 |
+
"miniature": 23,
|
76 |
+
"mythological painting": 24,
|
77 |
+
"nude painting (nu)": 25,
|
78 |
+
"panorama": 26,
|
79 |
+
"pastorale": 27,
|
80 |
+
"portrait": 28,
|
81 |
+
"poster": 29,
|
82 |
+
"quadratura": 30,
|
83 |
+
"religious painting": 31,
|
84 |
+
"self-portrait": 32,
|
85 |
+
"shan shui": 33,
|
86 |
+
"sketch and study": 34,
|
87 |
+
"still life": 35,
|
88 |
+
"symbolic painting": 36,
|
89 |
+
"tessellation": 37,
|
90 |
+
"urushi-e": 38,
|
91 |
+
"vanitas": 39,
|
92 |
+
"veduta": 40,
|
93 |
+
"wildlife painting": 41,
|
94 |
+
"yakusha-e": 42
|
95 |
+
},
|
96 |
+
"model_type": "siglip",
|
97 |
+
"problem_type": "single_label_classification",
|
98 |
+
"text_config": {
|
99 |
+
"attention_dropout": 0.0,
|
100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
101 |
+
"hidden_size": 768,
|
102 |
+
"intermediate_size": 3072,
|
103 |
+
"layer_norm_eps": 1e-06,
|
104 |
+
"max_position_embeddings": 64,
|
105 |
+
"model_type": "siglip_text_model",
|
106 |
+
"num_attention_heads": 12,
|
107 |
+
"num_hidden_layers": 12,
|
108 |
+
"projection_size": 768,
|
109 |
+
"torch_dtype": "float32",
|
110 |
+
"vocab_size": 256000
|
111 |
+
},
|
112 |
+
"torch_dtype": "float32",
|
113 |
+
"transformers_version": "4.50.0",
|
114 |
+
"vision_config": {
|
115 |
+
"attention_dropout": 0.0,
|
116 |
+
"hidden_act": "gelu_pytorch_tanh",
|
117 |
+
"hidden_size": 768,
|
118 |
+
"image_size": 224,
|
119 |
+
"intermediate_size": 3072,
|
120 |
+
"layer_norm_eps": 1e-06,
|
121 |
+
"model_type": "siglip_vision_model",
|
122 |
+
"num_attention_heads": 12,
|
123 |
+
"num_channels": 3,
|
124 |
+
"num_hidden_layers": 12,
|
125 |
+
"patch_size": 16,
|
126 |
+
"torch_dtype": "float32"
|
127 |
+
}
|
128 |
+
}
|
checkpoint-6454/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:750fd9f8142a28e675e63893d7a7b87c1e1e9fbc57711ce65ad1a26219406016
|
3 |
+
size 371694116
|
checkpoint-6454/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:91f42567bd0a60188084cf27b2e26921360847009f6ac87f5e8828c09a24fb27
|
3 |
+
size 686807930
|
checkpoint-6454/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-6454/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9c0b3f8b73bf88834a4258333e071e3f41acda0b8aad7677d5bce051299328fd
|
3 |
+
size 14244
|
checkpoint-6454/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8b68497655ec3addc3f1439b57dd3cec117221f180b25c69a473030c98cb8c6d
|
3 |
+
size 1064
|
checkpoint-6454/trainer_state.json
ADDED
@@ -0,0 +1,138 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_global_step": 6454,
|
3 |
+
"best_metric": 0.9517362117767334,
|
4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-6454",
|
5 |
+
"epoch": 2.0,
|
6 |
+
"eval_steps": 500,
|
7 |
+
"global_step": 6454,
|
8 |
+
"is_hyper_param_search": false,
|
9 |
+
"is_local_process_zero": true,
|
10 |
+
"is_world_process_zero": true,
|
11 |
+
"log_history": [
|
12 |
+
{
|
13 |
+
"epoch": 0.1549426712116517,
|
14 |
+
"grad_norm": 55.33502197265625,
|
15 |
+
"learning_rate": 2.7891942535915053e-06,
|
16 |
+
"loss": 1.9844,
|
17 |
+
"step": 500
|
18 |
+
},
|
19 |
+
{
|
20 |
+
"epoch": 0.3098853424233034,
|
21 |
+
"grad_norm": 37.46580505371094,
|
22 |
+
"learning_rate": 2.5549656464709556e-06,
|
23 |
+
"loss": 1.4259,
|
24 |
+
"step": 1000
|
25 |
+
},
|
26 |
+
{
|
27 |
+
"epoch": 0.4648280136349551,
|
28 |
+
"grad_norm": 41.121490478515625,
|
29 |
+
"learning_rate": 2.320737039350406e-06,
|
30 |
+
"loss": 1.3044,
|
31 |
+
"step": 1500
|
32 |
+
},
|
33 |
+
{
|
34 |
+
"epoch": 0.6197706848466068,
|
35 |
+
"grad_norm": 37.83391189575195,
|
36 |
+
"learning_rate": 2.0865084322298563e-06,
|
37 |
+
"loss": 1.2348,
|
38 |
+
"step": 2000
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.7747133560582584,
|
42 |
+
"grad_norm": 41.89641571044922,
|
43 |
+
"learning_rate": 1.8522798251093069e-06,
|
44 |
+
"loss": 1.1889,
|
45 |
+
"step": 2500
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.9296560272699101,
|
49 |
+
"grad_norm": 38.50979995727539,
|
50 |
+
"learning_rate": 1.6180512179887572e-06,
|
51 |
+
"loss": 1.1496,
|
52 |
+
"step": 3000
|
53 |
+
},
|
54 |
+
{
|
55 |
+
"epoch": 1.0,
|
56 |
+
"eval_accuracy": 0.6981598062953995,
|
57 |
+
"eval_loss": 1.0367238521575928,
|
58 |
+
"eval_model_preparation_time": 0.0038,
|
59 |
+
"eval_runtime": 1328.8007,
|
60 |
+
"eval_samples_per_second": 77.702,
|
61 |
+
"eval_steps_per_second": 9.713,
|
62 |
+
"step": 3227
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 1.0845986984815619,
|
66 |
+
"grad_norm": 35.213375091552734,
|
67 |
+
"learning_rate": 1.3838226108682073e-06,
|
68 |
+
"loss": 1.088,
|
69 |
+
"step": 3500
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 1.2395413696932136,
|
73 |
+
"grad_norm": 41.03507995605469,
|
74 |
+
"learning_rate": 1.1495940037476577e-06,
|
75 |
+
"loss": 1.0959,
|
76 |
+
"step": 4000
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"epoch": 1.394484040904865,
|
80 |
+
"grad_norm": 39.10642623901367,
|
81 |
+
"learning_rate": 9.15365396627108e-07,
|
82 |
+
"loss": 1.0571,
|
83 |
+
"step": 4500
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 1.5494267121165168,
|
87 |
+
"grad_norm": 40.350914001464844,
|
88 |
+
"learning_rate": 6.811367895065585e-07,
|
89 |
+
"loss": 1.0547,
|
90 |
+
"step": 5000
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 1.7043693833281686,
|
94 |
+
"grad_norm": 44.81325912475586,
|
95 |
+
"learning_rate": 4.4690818238600874e-07,
|
96 |
+
"loss": 1.0296,
|
97 |
+
"step": 5500
|
98 |
+
},
|
99 |
+
{
|
100 |
+
"epoch": 1.8593120545398203,
|
101 |
+
"grad_norm": 38.73249435424805,
|
102 |
+
"learning_rate": 2.126795752654591e-07,
|
103 |
+
"loss": 1.0197,
|
104 |
+
"step": 6000
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 2.0,
|
108 |
+
"eval_accuracy": 0.7182566585956417,
|
109 |
+
"eval_loss": 0.9517362117767334,
|
110 |
+
"eval_model_preparation_time": 0.0038,
|
111 |
+
"eval_runtime": 1326.8357,
|
112 |
+
"eval_samples_per_second": 77.817,
|
113 |
+
"eval_steps_per_second": 9.728,
|
114 |
+
"step": 6454
|
115 |
+
}
|
116 |
+
],
|
117 |
+
"logging_steps": 500,
|
118 |
+
"max_steps": 6454,
|
119 |
+
"num_input_tokens_seen": 0,
|
120 |
+
"num_train_epochs": 2,
|
121 |
+
"save_steps": 500,
|
122 |
+
"stateful_callbacks": {
|
123 |
+
"TrainerControl": {
|
124 |
+
"args": {
|
125 |
+
"should_epoch_stop": false,
|
126 |
+
"should_evaluate": false,
|
127 |
+
"should_log": false,
|
128 |
+
"should_save": true,
|
129 |
+
"should_training_stop": true
|
130 |
+
},
|
131 |
+
"attributes": {}
|
132 |
+
}
|
133 |
+
},
|
134 |
+
"total_flos": 1.7301390177770496e+19,
|
135 |
+
"train_batch_size": 32,
|
136 |
+
"trial_name": null,
|
137 |
+
"trial_params": null
|
138 |
+
}
|
checkpoint-6454/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:efa4c1f3eb59005587e95669315d89aa03d881830745eaa01a42988ac6c33c1b
|
3 |
+
size 5304
|
config.json
ADDED
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"SiglipForImageClassification"
|
4 |
+
],
|
5 |
+
"id2label": {
|
6 |
+
"0": "None",
|
7 |
+
"1": "abstract",
|
8 |
+
"2": "advertisement",
|
9 |
+
"3": "allegorical painting",
|
10 |
+
"4": "animal painting",
|
11 |
+
"5": "battle painting",
|
12 |
+
"6": "bijinga",
|
13 |
+
"7": "bird-and-flower painting",
|
14 |
+
"8": "calligraphy",
|
15 |
+
"9": "capriccio",
|
16 |
+
"10": "caricature",
|
17 |
+
"11": "cityscape",
|
18 |
+
"12": "cloudscape",
|
19 |
+
"13": "design",
|
20 |
+
"14": "figurative",
|
21 |
+
"15": "flower painting",
|
22 |
+
"16": "genre painting",
|
23 |
+
"17": "history painting",
|
24 |
+
"18": "illustration",
|
25 |
+
"19": "interior",
|
26 |
+
"20": "landscape",
|
27 |
+
"21": "literary painting",
|
28 |
+
"22": "marina",
|
29 |
+
"23": "miniature",
|
30 |
+
"24": "mythological painting",
|
31 |
+
"25": "nude painting (nu)",
|
32 |
+
"26": "panorama",
|
33 |
+
"27": "pastorale",
|
34 |
+
"28": "portrait",
|
35 |
+
"29": "poster",
|
36 |
+
"30": "quadratura",
|
37 |
+
"31": "religious painting",
|
38 |
+
"32": "self-portrait",
|
39 |
+
"33": "shan shui",
|
40 |
+
"34": "sketch and study",
|
41 |
+
"35": "still life",
|
42 |
+
"36": "symbolic painting",
|
43 |
+
"37": "tessellation",
|
44 |
+
"38": "urushi-e",
|
45 |
+
"39": "vanitas",
|
46 |
+
"40": "veduta",
|
47 |
+
"41": "wildlife painting",
|
48 |
+
"42": "yakusha-e"
|
49 |
+
},
|
50 |
+
"initializer_factor": 1.0,
|
51 |
+
"label2id": {
|
52 |
+
"None": 0,
|
53 |
+
"abstract": 1,
|
54 |
+
"advertisement": 2,
|
55 |
+
"allegorical painting": 3,
|
56 |
+
"animal painting": 4,
|
57 |
+
"battle painting": 5,
|
58 |
+
"bijinga": 6,
|
59 |
+
"bird-and-flower painting": 7,
|
60 |
+
"calligraphy": 8,
|
61 |
+
"capriccio": 9,
|
62 |
+
"caricature": 10,
|
63 |
+
"cityscape": 11,
|
64 |
+
"cloudscape": 12,
|
65 |
+
"design": 13,
|
66 |
+
"figurative": 14,
|
67 |
+
"flower painting": 15,
|
68 |
+
"genre painting": 16,
|
69 |
+
"history painting": 17,
|
70 |
+
"illustration": 18,
|
71 |
+
"interior": 19,
|
72 |
+
"landscape": 20,
|
73 |
+
"literary painting": 21,
|
74 |
+
"marina": 22,
|
75 |
+
"miniature": 23,
|
76 |
+
"mythological painting": 24,
|
77 |
+
"nude painting (nu)": 25,
|
78 |
+
"panorama": 26,
|
79 |
+
"pastorale": 27,
|
80 |
+
"portrait": 28,
|
81 |
+
"poster": 29,
|
82 |
+
"quadratura": 30,
|
83 |
+
"religious painting": 31,
|
84 |
+
"self-portrait": 32,
|
85 |
+
"shan shui": 33,
|
86 |
+
"sketch and study": 34,
|
87 |
+
"still life": 35,
|
88 |
+
"symbolic painting": 36,
|
89 |
+
"tessellation": 37,
|
90 |
+
"urushi-e": 38,
|
91 |
+
"vanitas": 39,
|
92 |
+
"veduta": 40,
|
93 |
+
"wildlife painting": 41,
|
94 |
+
"yakusha-e": 42
|
95 |
+
},
|
96 |
+
"model_type": "siglip",
|
97 |
+
"problem_type": "single_label_classification",
|
98 |
+
"text_config": {
|
99 |
+
"attention_dropout": 0.0,
|
100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
101 |
+
"hidden_size": 768,
|
102 |
+
"intermediate_size": 3072,
|
103 |
+
"layer_norm_eps": 1e-06,
|
104 |
+
"max_position_embeddings": 64,
|
105 |
+
"model_type": "siglip_text_model",
|
106 |
+
"num_attention_heads": 12,
|
107 |
+
"num_hidden_layers": 12,
|
108 |
+
"projection_size": 768,
|
109 |
+
"torch_dtype": "float32",
|
110 |
+
"vocab_size": 256000
|
111 |
+
},
|
112 |
+
"torch_dtype": "float32",
|
113 |
+
"transformers_version": "4.50.0",
|
114 |
+
"vision_config": {
|
115 |
+
"attention_dropout": 0.0,
|
116 |
+
"hidden_act": "gelu_pytorch_tanh",
|
117 |
+
"hidden_size": 768,
|
118 |
+
"image_size": 224,
|
119 |
+
"intermediate_size": 3072,
|
120 |
+
"layer_norm_eps": 1e-06,
|
121 |
+
"model_type": "siglip_vision_model",
|
122 |
+
"num_attention_heads": 12,
|
123 |
+
"num_channels": 3,
|
124 |
+
"num_hidden_layers": 12,
|
125 |
+
"patch_size": 16,
|
126 |
+
"torch_dtype": "float32"
|
127 |
+
}
|
128 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:750fd9f8142a28e675e63893d7a7b87c1e1e9fbc57711ce65ad1a26219406016
|
3 |
+
size 371694116
|
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:efa4c1f3eb59005587e95669315d89aa03d881830745eaa01a42988ac6c33c1b
|
3 |
+
size 5304
|