Hugging Face's logo

neilmehta24
/
janus-pro-7b-4bit

Any-to-Any
Transformers
PyTorch
Safetensors
multi_modality
muiltimodal
text-to-image
unified-model
4-bit precision
bitsandbytes
Model card Files
xet
Community
neilmehta24 commited on
Commit
72a211d
·
verified ·
1 Parent(s): a9c160f

Fix config.json

Browse files
Files changed (1) hide show
  1. config.json +79 -59
config.json CHANGED
@@ -1,59 +1,79 @@
1
- {
2
- "aligner_config": {
3
- "cls": "MlpProjector",
4
- "model_type": "aligner",
5
- "params": {
6
- "depth": 2,
7
- "input_dim": 1024,
8
- "n_embed": 4096,
9
- "projector_type": "mlp_gelu"
10
- }
11
- },
12
- "gen_aligner_config": {
13
- "cls": "MlpProjector",
14
- "model_type": "gen_aligner",
15
- "params": {
16
- "depth": 2,
17
- "input_dim": 8,
18
- "n_embed": 4096,
19
- "projector_type": "mlp_gelu"
20
- }
21
- },
22
- "gen_head_config": {
23
- "cls": "vision_head",
24
- "model_type": "gen_head",
25
- "params": {
26
- "image_token_embed": 4096,
27
- "image_token_size": 16384,
28
- "n_embed": 4096
29
- }
30
- },
31
- "gen_vision_config": {
32
- "cls": "VQ-16",
33
- "model_type": "gen_vision",
34
- "params": {
35
- "image_token_size": 16384,
36
- "n_embed": 8
37
- }
38
- },
39
- "language_config": {
40
- "max_position_embeddings": 16384,
41
- "model_type": "llama",
42
- "num_hidden_layers": 30,
43
- "torch_dtype": "bfloat16",
44
- "vocab_size": 102400
45
- },
46
- "model_type": "multi_modality",
47
- "torch_dtype": "bfloat16",
48
- "transformers_version": "4.33.1",
49
- "vision_config": {
50
- "cls": "CLIPVisionTower",
51
- "model_type": "vision",
52
- "params": {
53
- "image_size": 384,
54
- "model_name": "siglip_large_patch16_384",
55
- "select_feature": "same",
56
- "select_layer": -1
57
- }
58
- }
59
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "deepseek-ai/Janus-Pro-7B",
3
+ "aligner_config": {
4
+ "cls": "MlpProjector",
5
+ "model_type": "aligner",
6
+ "params": {
7
+ "depth": 2,
8
+ "input_dim": 1024,
9
+ "n_embed": 4096,
10
+ "projector_type": "mlp_gelu"
11
+ }
12
+ },
13
+ "architectures": [
14
+ "MultiModalityCausalLM"
15
+ ],
16
+ "gen_aligner_config": {
17
+ "cls": "MlpProjector",
18
+ "model_type": "gen_aligner",
19
+ "params": {
20
+ "depth": 2,
21
+ "input_dim": 8,
22
+ "n_embed": 4096,
23
+ "projector_type": "mlp_gelu"
24
+ }
25
+ },
26
+ "gen_head_config": {
27
+ "cls": "vision_head",
28
+ "model_type": "gen_head",
29
+ "params": {
30
+ "image_token_embed": 4096,
31
+ "image_token_size": 16384,
32
+ "n_embed": 4096
33
+ }
34
+ },
35
+ "gen_vision_config": {
36
+ "cls": "VQ-16",
37
+ "model_type": "gen_vision",
38
+ "params": {
39
+ "image_token_size": 16384,
40
+ "n_embed": 8
41
+ }
42
+ },
43
+ "language_config": {
44
+ "_attn_implementation_autoset": true,
45
+ "max_position_embeddings": 16384,
46
+ "model_type": "llama",
47
+ "num_hidden_layers": 30,
48
+ "torch_dtype": "bfloat16",
49
+ "vocab_size": 102400
50
+ },
51
+ "model_type": "multi_modality",
52
+ "quantization_config": {
53
+ "_load_in_4bit": true,
54
+ "_load_in_8bit": false,
55
+ "bnb_4bit_compute_dtype": "float32",
56
+ "bnb_4bit_quant_storage": "uint8",
57
+ "bnb_4bit_quant_type": "fp4",
58
+ "bnb_4bit_use_double_quant": false,
59
+ "llm_int8_enable_fp32_cpu_offload": false,
60
+ "llm_int8_has_fp16_weight": false,
61
+ "llm_int8_skip_modules": null,
62
+ "llm_int8_threshold": 6.0,
63
+ "load_in_4bit": true,
64
+ "load_in_8bit": false,
65
+ "quant_method": "bitsandbytes"
66
+ },
67
+ "torch_dtype": "bfloat16",
68
+ "transformers_version": "4.48.1",
69
+ "vision_config": {
70
+ "cls": "CLIPVisionTower",
71
+ "model_type": "vision",
72
+ "params": {
73
+ "image_size": 384,
74
+ "model_name": "siglip_large_patch16_384",
75
+ "select_feature": "same",
76
+ "select_layer": -1
77
+ }
78
+ }
79
+ }