Hugging Face's logo

hf-tiny-v2
/
tiny-random-VideoPrismClipModel

Safetensors
VideoPrism
Model card Files
xet
 Community
hf-transformers-bot commited on
Commit
390236c
·
verified ·
1 Parent(s): ac9e6e8

Update tiny models for VideoPrismClipModel

Browse files
Files changed (5) hide show
  1. config.json +53 -0
  2. model.safetensors +3 -0
  3. tokenizer.json +0 -0
  4. tokenizer_config.json +119 -0
  5. video_preprocessor_config.json +29 -0
config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "VideoPrismClipModel"
4
+ ],
5
+ "dtype": "float32",
6
+ "model_type": "videoprism",
7
+ "text_config": {
8
+ "apply_l2norm": true,
9
+ "attention_probs_dropout_prob": 0.0,
10
+ "attn_logit_softcapping": 50.0,
11
+ "bos_token_id": null,
12
+ "eos_token_id": null,
13
+ "hidden_act": "relu",
14
+ "hidden_dropout_prob": 0.0,
15
+ "hidden_size": 32,
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 37,
18
+ "layer_norm_eps": 1e-06,
19
+ "max_position_embeddings": 200,
20
+ "model_type": "videoprism_text_model",
21
+ "num_attention_heads": 4,
22
+ "num_hidden_layers": 2,
23
+ "pad_token_id": 0,
24
+ "qkv_bias": true,
25
+ "vocab_size": 32101
26
+ },
27
+ "transformers_version": "5.16.0.dev0",
28
+ "vision_config": {
29
+ "apply_l2norm": true,
30
+ "attention_probs_dropout_prob": 0.0,
31
+ "attn_logit_softcapping": 50.0,
32
+ "hidden_act": "gelu_python",
33
+ "hidden_dropout_prob": 0.0,
34
+ "hidden_size": 32,
35
+ "image_size": 8,
36
+ "initializer_range": 0.02,
37
+ "intermediate_size": 64,
38
+ "layer_norm_eps": 1e-06,
39
+ "model_type": "videoprism_vision_model",
40
+ "num_attention_heads": 4,
41
+ "num_auxiliary_layers": 2,
42
+ "num_channels": 3,
43
+ "num_frames": 3,
44
+ "num_spatial_layers": 3,
45
+ "num_temporal_layers": 2,
46
+ "qkv_bias": true,
47
+ "tubelet_size": [
48
+ 1,
49
+ 4,
50
+ 4
51
+ ]
52
+ }
53
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:215e8f5d0dd2683a1183c81d4afd9f3442f903cbaac056fe019ad5c41f359fbd
3
+ size 4491992
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,119 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<s>",
4
+ "eos_token": "</s>",
5
+ "extra_ids": 100,
6
+ "extra_special_tokens": [
7
+ "<extra_id_0>",
8
+ "<extra_id_1>",
9
+ "<extra_id_2>",
10
+ "<extra_id_3>",
11
+ "<extra_id_4>",
12
+ "<extra_id_5>",
13
+ "<extra_id_6>",
14
+ "<extra_id_7>",
15
+ "<extra_id_8>",
16
+ "<extra_id_9>",
17
+ "<extra_id_10>",
18
+ "<extra_id_11>",
19
+ "<extra_id_12>",
20
+ "<extra_id_13>",
21
+ "<extra_id_14>",
22
+ "<extra_id_15>",
23
+ "<extra_id_16>",
24
+ "<extra_id_17>",
25
+ "<extra_id_18>",
26
+ "<extra_id_19>",
27
+ "<extra_id_20>",
28
+ "<extra_id_21>",
29
+ "<extra_id_22>",
30
+ "<extra_id_23>",
31
+ "<extra_id_24>",
32
+ "<extra_id_25>",
33
+ "<extra_id_26>",
34
+ "<extra_id_27>",
35
+ "<extra_id_28>",
36
+ "<extra_id_29>",
37
+ "<extra_id_30>",
38
+ "<extra_id_31>",
39
+ "<extra_id_32>",
40
+ "<extra_id_33>",
41
+ "<extra_id_34>",
42
+ "<extra_id_35>",
43
+ "<extra_id_36>",
44
+ "<extra_id_37>",
45
+ "<extra_id_38>",
46
+ "<extra_id_39>",
47
+ "<extra_id_40>",
48
+ "<extra_id_41>",
49
+ "<extra_id_42>",
50
+ "<extra_id_43>",
51
+ "<extra_id_44>",
52
+ "<extra_id_45>",
53
+ "<extra_id_46>",
54
+ "<extra_id_47>",
55
+ "<extra_id_48>",
56
+ "<extra_id_49>",
57
+ "<extra_id_50>",
58
+ "<extra_id_51>",
59
+ "<extra_id_52>",
60
+ "<extra_id_53>",
61
+ "<extra_id_54>",
62
+ "<extra_id_55>",
63
+ "<extra_id_56>",
64
+ "<extra_id_57>",
65
+ "<extra_id_58>",
66
+ "<extra_id_59>",
67
+ "<extra_id_60>",
68
+ "<extra_id_61>",
69
+ "<extra_id_62>",
70
+ "<extra_id_63>",
71
+ "<extra_id_64>",
72
+ "<extra_id_65>",
73
+ "<extra_id_66>",
74
+ "<extra_id_67>",
75
+ "<extra_id_68>",
76
+ "<extra_id_69>",
77
+ "<extra_id_70>",
78
+ "<extra_id_71>",
79
+ "<extra_id_72>",
80
+ "<extra_id_73>",
81
+ "<extra_id_74>",
82
+ "<extra_id_75>",
83
+ "<extra_id_76>",
84
+ "<extra_id_77>",
85
+ "<extra_id_78>",
86
+ "<extra_id_79>",
87
+ "<extra_id_80>",
88
+ "<extra_id_81>",
89
+ "<extra_id_82>",
90
+ "<extra_id_83>",
91
+ "<extra_id_84>",
92
+ "<extra_id_85>",
93
+ "<extra_id_86>",
94
+ "<extra_id_87>",
95
+ "<extra_id_88>",
96
+ "<extra_id_89>",
97
+ "<extra_id_90>",
98
+ "<extra_id_91>",
99
+ "<extra_id_92>",
100
+ "<extra_id_93>",
101
+ "<extra_id_94>",
102
+ "<extra_id_95>",
103
+ "<extra_id_96>",
104
+ "<extra_id_97>",
105
+ "<extra_id_98>",
106
+ "<extra_id_99>"
107
+ ],
108
+ "is_local": true,
109
+ "local_files_only": false,
110
+ "max_length": 64,
111
+ "model_max_length": 1000000000000000019884624838656,
112
+ "pad_to_multiple_of": null,
113
+ "pad_token": "<pad>",
114
+ "pad_token_type_id": 0,
115
+ "padding_side": "right",
116
+ "processor_class": "VideoPrismProcessor",
117
+ "tokenizer_class": "VideoPrismTokenizer",
118
+ "unk_token": "<unk>"
119
+ }
video_preprocessor_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 8,
4
+ "width": 8
5
+ },
6
+ "do_convert_rgb": true,
7
+ "do_normalize": true,
8
+ "do_rescale": true,
9
+ "do_resize": true,
10
+ "do_sample_frames": false,
11
+ "image_mean": [
12
+ 0.48145466,
13
+ 0.4578275,
14
+ 0.40821073
15
+ ],
16
+ "image_std": [
17
+ 0.26862954,
18
+ 0.26130258,
19
+ 0.27577711
20
+ ],
21
+ "resample": 3,
22
+ "rescale_factor": 0.00392156862745098,
23
+ "return_metadata": false,
24
+ "size": {
25
+ "height": 8,
26
+ "width": 8
27
+ },
28
+ "video_processor_type": "LlavaOnevisionVideoProcessor"
29
+ }