Hugging Face's logo

nafiz09
/
PI_Granite

Image-Text-to-Text
Transformers
Safetensors
llava_next
Generated from Trainer
trl
sft
conversational
text-generation-inference
Model card Files
xet
Community
nafiz09 commited on
Commit
0004417
·
verified ·
1 Parent(s): e1a38a1

Upload config.json

Browse files
Files changed (1) hide show
  1. config.json +168 -0
config.json ADDED
@@ -0,0 +1,168 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_grid_pinpoints": [
3
+ [
4
+ 384,
5
+ 384
6
+ ],
7
+ [
8
+ 384,
9
+ 768
10
+ ],
11
+ [
12
+ 384,
13
+ 1152
14
+ ],
15
+ [
16
+ 384,
17
+ 1536
18
+ ],
19
+ [
20
+ 384,
21
+ 1920
22
+ ],
23
+ [
24
+ 384,
25
+ 2304
26
+ ],
27
+ [
28
+ 384,
29
+ 2688
30
+ ],
31
+ [
32
+ 384,
33
+ 3072
34
+ ],
35
+ [
36
+ 384,
37
+ 3456
38
+ ],
39
+ [
40
+ 384,
41
+ 3840
42
+ ],
43
+ [
44
+ 768,
45
+ 384
46
+ ],
47
+ [
48
+ 768,
49
+ 768
50
+ ],
51
+ [
52
+ 768,
53
+ 1152
54
+ ],
55
+ [
56
+ 768,
57
+ 1536
58
+ ],
59
+ [
60
+ 768,
61
+ 1920
62
+ ],
63
+ [
64
+ 1152,
65
+ 384
66
+ ],
67
+ [
68
+ 1152,
69
+ 768
70
+ ],
71
+ [
72
+ 1152,
73
+ 1152
74
+ ],
75
+ [
76
+ 1536,
77
+ 384
78
+ ],
79
+ [
80
+ 1536,
81
+ 768
82
+ ],
83
+ [
84
+ 1920,
85
+ 384
86
+ ],
87
+ [
88
+ 1920,
89
+ 768
90
+ ],
91
+ [
92
+ 2304,
93
+ 384
94
+ ],
95
+ [
96
+ 2688,
97
+ 384
98
+ ],
99
+ [
100
+ 3072,
101
+ 384
102
+ ],
103
+ [
104
+ 3456,
105
+ 384
106
+ ],
107
+ [
108
+ 3840,
109
+ 384
110
+ ]
111
+ ],
112
+ "tie_word_embeddings": true,
113
+ "transformers_version": "4.45.0.dev0",
114
+ "architectures": [
115
+ "LlavaNextForConditionalGeneration"
116
+ ],
117
+ "model_type": "llava_next",
118
+ "use_image_newline_parameter": true,
119
+ "vision_feature_layer": [
120
+ -24,
121
+ -20,
122
+ -12,
123
+ -1
124
+ ],
125
+ "vision_feature_select_strategy": "full",
126
+ "text_config": {
127
+ "architectures": [
128
+ "GraniteForCausalLM"
129
+ ],
130
+ "attention_bias": false,
131
+ "attention_dropout": 0.1,
132
+ "attention_multiplier": 0.015625,
133
+ "bos_token_id": 0,
134
+ "embedding_multiplier": 12.0,
135
+ "eos_token_id": 0,
136
+ "hidden_act": "silu",
137
+ "hidden_size": 2048,
138
+ "initializer_range": 0.02,
139
+ "intermediate_size": 8192,
140
+ "logits_scaling": 8.0,
141
+ "max_position_embeddings": 131072,
142
+ "mlp_bias": false,
143
+ "model_type": "granite",
144
+ "num_attention_heads": 32,
145
+ "num_hidden_layers": 40,
146
+ "num_key_value_heads": 8,
147
+ "pad_token_id": 0,
148
+ "residual_multiplier": 0.22,
149
+ "rms_norm_eps": 1e-05,
150
+ "rope_scaling": null,
151
+ "rope_theta": 300000,
152
+ "tie_word_embeddings": true,
153
+ "torch_dtype": "bfloat16",
154
+ "transformers_version": "4.46.0.dev0",
155
+ "use_cache": true,
156
+ "vocab_size": 49156
157
+ },
158
+ "image_token_index": 49155,
159
+ "vision_config": {
160
+ "hidden_size": 1152,
161
+ "image_size": 384,
162
+ "intermediate_size": 4304,
163
+ "model_type": "siglip_vision_model",
164
+ "num_attention_heads": 16,
165
+ "num_hidden_layers": 27,
166
+ "patch_size": 14
167
+ }
168
+ }