Hugging Face's logo

trl-internal-testing
/
tiny-InternVLForConditionalGeneration

Image-Text-to-Text
Transformers
Safetensors
internvl
trl
conversational
Model card Files
xet
Community
qgallouedec HF Staff commited on
Commit
baa16f6
·
verified ·
1 Parent(s): d77a85b

Upload InternVLForConditionalGeneration

Browse files
Files changed (3) hide show
  1. config.json +5 -30
  2. generation_config.json +1 -1
  3. model.safetensors +1 -1
config.json CHANGED
@@ -3,6 +3,7 @@
3
  "InternVLForConditionalGeneration"
4
  ],
5
  "downsample_ratio": 0.5,
 
6
  "image_seq_length": 256,
7
  "image_token_id": 151667,
8
  "model_type": "internvl",
@@ -13,38 +14,13 @@
13
  ],
14
  "attention_dropout": 0.0,
15
  "bos_token_id": 151643,
 
16
  "eos_token_id": 151645,
17
  "hidden_act": "silu",
18
  "hidden_size": 16,
19
  "initializer_range": 0.02,
20
  "intermediate_size": 18944,
21
  "layer_types": [
22
- "full_attention",
23
- "full_attention",
24
- "full_attention",
25
- "full_attention",
26
- "full_attention",
27
- "full_attention",
28
- "full_attention",
29
- "full_attention",
30
- "full_attention",
31
- "full_attention",
32
- "full_attention",
33
- "full_attention",
34
- "full_attention",
35
- "full_attention",
36
- "full_attention",
37
- "full_attention",
38
- "full_attention",
39
- "full_attention",
40
- "full_attention",
41
- "full_attention",
42
- "full_attention",
43
- "full_attention",
44
- "full_attention",
45
- "full_attention",
46
- "full_attention",
47
- "full_attention",
48
  "full_attention",
49
  "full_attention"
50
  ],
@@ -62,13 +38,11 @@
62
  },
63
  "rope_theta": 1000000.0,
64
  "sliding_window": null,
65
- "torch_dtype": "bfloat16",
66
  "use_cache": true,
67
  "use_sliding_window": false,
68
  "vocab_size": 151674
69
  },
70
- "torch_dtype": "bfloat16",
71
- "transformers_version": "4.56.0.dev0",
72
  "vision_config": {
73
  "architectures": [
74
  "InternVisionModel"
@@ -76,6 +50,8 @@
76
  "attention_bias": true,
77
  "attention_dropout": 0.0,
78
  "dropout": 0.0,
 
 
79
  "hidden_act": "gelu",
80
  "hidden_dropout_prob": 0.0,
81
  "hidden_size": 16,
@@ -99,7 +75,6 @@
99
  14
100
  ],
101
  "projection_dropout": 0.0,
102
- "torch_dtype": "bfloat16",
103
  "use_absolute_position_embeddings": true,
104
  "use_mask_token": false,
105
  "use_mean_pooling": true,
 
3
  "InternVLForConditionalGeneration"
4
  ],
5
  "downsample_ratio": 0.5,
6
+ "dtype": "bfloat16",
7
  "image_seq_length": 256,
8
  "image_token_id": 151667,
9
  "model_type": "internvl",
 
14
  ],
15
  "attention_dropout": 0.0,
16
  "bos_token_id": 151643,
17
+ "dtype": "bfloat16",
18
  "eos_token_id": 151645,
19
  "hidden_act": "silu",
20
  "hidden_size": 16,
21
  "initializer_range": 0.02,
22
  "intermediate_size": 18944,
23
  "layer_types": [
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
  "full_attention",
25
  "full_attention"
26
  ],
 
38
  },
39
  "rope_theta": 1000000.0,
40
  "sliding_window": null,
 
41
  "use_cache": true,
42
  "use_sliding_window": false,
43
  "vocab_size": 151674
44
  },
45
+ "transformers_version": "4.57.0.dev0",
 
46
  "vision_config": {
47
  "architectures": [
48
  "InternVisionModel"
 
50
  "attention_bias": true,
51
  "attention_dropout": 0.0,
52
  "dropout": 0.0,
53
+ "dtype": "bfloat16",
54
+ "embed_dim": 32,
55
  "hidden_act": "gelu",
56
  "hidden_dropout_prob": 0.0,
57
  "hidden_size": 16,
 
75
  14
76
  ],
77
  "projection_dropout": 0.0,
 
78
  "use_absolute_position_embeddings": true,
79
  "use_mask_token": false,
80
  "use_mean_pooling": true,
generation_config.json CHANGED
@@ -2,5 +2,5 @@
2
  "_from_model_config": true,
3
  "bos_token_id": 151643,
4
  "eos_token_id": 151645,
5
- "transformers_version": "4.56.0.dev0"
6
  }
 
2
  "_from_model_config": true,
3
  "bos_token_id": 151643,
4
  "eos_token_id": 151645,
5
+ "transformers_version": "4.57.0.dev0"
6
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ed5c180bbebb66356d2b6e59fa050007814e5192ae347c8cb0fd178578e88ec0
3
  size 13956416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac4ed3cf0838f9202013909fc3c8a6c701744bfce8b283c64ab908a202327caf
3
  size 13956416