Hugging Face's logo

JANGQ-AI
/
ZAYA1-VL-8B-JANGTQ4

Image-Text-to-Text
MLX
Safetensors
zaya1_vl
zaya
mixture-of-experts
hybrid-attention
cca-attention
apple-silicon
reasoning
tool-use
quantized
vision
multimodal
vision-language
qwen2_5_vl-vit
jang
jangtq
mxtq
jangtq-prestack
jangq
conversational
Model card Files
xet
 Community
Osaurus-AI commited on
Commit
059ea86
·
verified ·
1 Parent(s): 12ac6f0

Mirror ZAYA1-VL-8B-JANGTQ4 from OsaurusAI to JANGQ-AI: same JANGTQ bundle, JANGQ-AI capabilities re-stamped

Browse files
Files changed (19) hide show
  1. .gitattributes +1 -0
  2. chat_template.json +3 -0
  3. config.json +106 -0
  4. generation_config.json +7 -0
  5. jang_config.json +160 -0
  6. jangq-logo.png +0 -0
  7. jangtq_runtime.safetensors +3 -0
  8. model-00001-of-00007.safetensors +3 -0
  9. model-00002-of-00007.safetensors +3 -0
  10. model-00003-of-00007.safetensors +3 -0
  11. model-00004-of-00007.safetensors +3 -0
  12. model-00005-of-00007.safetensors +3 -0
  13. model-00006-of-00007.safetensors +3 -0
  14. model-00007-of-00007.safetensors +3 -0
  15. model.safetensors.index.json +0 -0
  16. preprocessor_config.json +29 -0
  17. special_tokens_map.json +35 -0
  18. tokenizer.json +3 -0
  19. tokenizer_config.json +0 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
chat_template.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "chat_template": "{% for message in messages %}{% if message['role'] == 'user' %}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<|vision_start|><image><|vision_end|>\\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ '<|im_start|>' ~ message['role'] ~ '\n' ~ content['text'] ~ '<|im_end|>' ~ '\n' }}{% endfor %}{% elif message['role'] == 'question' %}{{ '<|im_start|>user\\n' }}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] ~ '<|im_end|>\\n' }}{% endfor %}{% else %}{{ '<|im_start|>' ~ message['role'] ~ '\\n' }}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] ~ '<|im_end|>' }}{% endfor %}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\\n' }}{% endif %}"
3
+ }
config.json ADDED
@@ -0,0 +1,106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_func": "swiglu",
3
+ "activation_func_fp8_input_store": false,
4
+ "add_bias_linear": false,
5
+ "apply_rope_fusion": true,
6
+ "ar_threshold": 1,
7
+ "architectures": [
8
+ "Zaya1VLForConditionalGeneration"
9
+ ],
10
+ "attention_bias": false,
11
+ "bias_activation_fusion": true,
12
+ "bos_token_id": 2,
13
+ "cca": true,
14
+ "clamp_temp": false,
15
+ "eos_token_id": 262143,
16
+ "ffn_hidden_size": 4096,
17
+ "fused_add_norm": false,
18
+ "gated_linear_unit": true,
19
+ "hidden_size": 2048,
20
+ "head_dim": 128,
21
+ "image_token_id": 262147,
22
+ "lm_head_bias": false,
23
+ "lora_rank": 0,
24
+ "max_position_embeddings": 32768,
25
+ "model_type": "zaya1_vl",
26
+ "moe_router_topk": 1,
27
+ "norm_epsilon": 1e-05,
28
+ "normalization": "RMSNorm",
29
+ "num_attention_heads": 8,
30
+ "num_experts": 16,
31
+ "num_hidden_layers": 40,
32
+ "num_key_value_heads": 2,
33
+ "num_query_groups": 2,
34
+ "pad_token_id": 0,
35
+ "padding_side": "right",
36
+ "projector_hidden_act": "gelu",
37
+ "residual_in_fp32": false,
38
+ "rope_pct": 0.5,
39
+ "rotary_base": 1000000,
40
+ "scale_residual_merge": true,
41
+ "sliding_window": null,
42
+ "temporal_patch_size": 1,
43
+ "tie_word_embeddings": true,
44
+ "torch_dtype": "bfloat16",
45
+ "transformers_version": "4.57.1",
46
+ "use_lora_att": false,
47
+ "use_rope_scaling": false,
48
+ "vision_config": {
49
+ "_attn_implementation_autoset": true,
50
+ "hidden_size": 1280,
51
+ "in_chans": 3,
52
+ "model_type": "qwen2_5_vl",
53
+ "out_hidden_size": 2048,
54
+ "spatial_patch_size": 14,
55
+ "temporal_patch_size": 1,
56
+ "tokens_per_second": 2,
57
+ "torch_dtype": "bfloat16"
58
+ },
59
+ "vision_end_token_id": 256000,
60
+ "vision_lora": true,
61
+ "vision_lora_rank_attn": 8,
62
+ "vision_lora_rank_mlp": 32,
63
+ "vision_start_token_id": 255999,
64
+ "vocab_size": 262272,
65
+ "zaya_mlp_expansion": 256,
66
+ "zaya_use_eda": true,
67
+ "zaya_use_mod": true,
68
+ "weight_format": "mxtq",
69
+ "zaya_expert_layout": "split_switch_mlp",
70
+ "mxtq_bits": {
71
+ "routed_expert": 4,
72
+ "attention": 8,
73
+ "router": 16,
74
+ "embed_tokens": 8,
75
+ "lm_head": 8,
76
+ "cca_conv": 16,
77
+ "norms_residual": 16
78
+ },
79
+ "mxtq_seed": 42,
80
+ "quantization": {
81
+ "bits": 8,
82
+ "group_size": 32,
83
+ "mode": "affine+mxtq",
84
+ "routed_expert_bits": 4,
85
+ "mxtq_bits": {
86
+ "routed_expert": 4,
87
+ "attention": 8,
88
+ "router": 16,
89
+ "embed_tokens": 8,
90
+ "lm_head": 8,
91
+ "cca_conv": 16,
92
+ "norms_residual": 16
93
+ },
94
+ "expert_layout": "split_switch_mlp"
95
+ },
96
+ "capabilities": {
97
+ "reasoning_parser": "qwen3",
98
+ "tool_parser": "zaya_xml",
99
+ "think_in_template": false,
100
+ "supports_tools": true,
101
+ "supports_thinking": true,
102
+ "family": "zaya1_vl",
103
+ "modality": "vision",
104
+ "cache_type": "hybrid"
105
+ }
106
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 2,
4
+ "eos_token_id": 262143,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.50.0.dev0"
7
+ }
jang_config.json ADDED
@@ -0,0 +1,160 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 2,
3
+ "weight_format": "mxtq",
4
+ "profile": "JANGTQ4",
5
+ "cache_subtype": "zaya_cca",
6
+ "source_model": {
7
+ "name": "ZAYA1-VL-8B",
8
+ "org": "Zyphra",
9
+ "architecture": "zaya1_vl"
10
+ },
11
+ "has_vision": true,
12
+ "expert_layout": "split_switch_mlp",
13
+ "mxtq_seed": 42,
14
+ "mxtq_bits": {
15
+ "routed_expert": 4,
16
+ "attention": 8,
17
+ "router": 16,
18
+ "embed_tokens": 8,
19
+ "lm_head": 8,
20
+ "cca_conv": 16,
21
+ "norms_residual": 16
22
+ },
23
+ "tq_in_features": {
24
+ "model.layers.0.zaya_block.experts.switch_mlp.gate_proj": 2048,
25
+ "model.layers.0.zaya_block.experts.switch_mlp.up_proj": 2048,
26
+ "model.layers.0.zaya_block.experts.switch_mlp.down_proj": 2048,
27
+ "model.layers.1.zaya_block.experts.switch_mlp.gate_proj": 2048,
28
+ "model.layers.1.zaya_block.experts.switch_mlp.up_proj": 2048,
29
+ "model.layers.1.zaya_block.experts.switch_mlp.down_proj": 2048,
30
+ "model.layers.2.zaya_block.experts.switch_mlp.gate_proj": 2048,
31
+ "model.layers.2.zaya_block.experts.switch_mlp.up_proj": 2048,
32
+ "model.layers.2.zaya_block.experts.switch_mlp.down_proj": 2048,
33
+ "model.layers.3.zaya_block.experts.switch_mlp.gate_proj": 2048,
34
+ "model.layers.3.zaya_block.experts.switch_mlp.up_proj": 2048,
35
+ "model.layers.3.zaya_block.experts.switch_mlp.down_proj": 2048,
36
+ "model.layers.4.zaya_block.experts.switch_mlp.gate_proj": 2048,
37
+ "model.layers.4.zaya_block.experts.switch_mlp.up_proj": 2048,
38
+ "model.layers.4.zaya_block.experts.switch_mlp.down_proj": 2048,
39
+ "model.layers.5.zaya_block.experts.switch_mlp.gate_proj": 2048,
40
+ "model.layers.5.zaya_block.experts.switch_mlp.up_proj": 2048,
41
+ "model.layers.5.zaya_block.experts.switch_mlp.down_proj": 2048,
42
+ "model.layers.6.zaya_block.experts.switch_mlp.gate_proj": 2048,
43
+ "model.layers.6.zaya_block.experts.switch_mlp.up_proj": 2048,
44
+ "model.layers.6.zaya_block.experts.switch_mlp.down_proj": 2048,
45
+ "model.layers.7.zaya_block.experts.switch_mlp.gate_proj": 2048,
46
+ "model.layers.7.zaya_block.experts.switch_mlp.up_proj": 2048,
47
+ "model.layers.7.zaya_block.experts.switch_mlp.down_proj": 2048,
48
+ "model.layers.8.zaya_block.experts.switch_mlp.gate_proj": 2048,
49
+ "model.layers.8.zaya_block.experts.switch_mlp.up_proj": 2048,
50
+ "model.layers.8.zaya_block.experts.switch_mlp.down_proj": 2048,
51
+ "model.layers.9.zaya_block.experts.switch_mlp.gate_proj": 2048,
52
+ "model.layers.9.zaya_block.experts.switch_mlp.up_proj": 2048,
53
+ "model.layers.9.zaya_block.experts.switch_mlp.down_proj": 2048,
54
+ "model.layers.10.zaya_block.experts.switch_mlp.gate_proj": 2048,
55
+ "model.layers.10.zaya_block.experts.switch_mlp.up_proj": 2048,
56
+ "model.layers.10.zaya_block.experts.switch_mlp.down_proj": 2048,
57
+ "model.layers.11.zaya_block.experts.switch_mlp.gate_proj": 2048,
58
+ "model.layers.11.zaya_block.experts.switch_mlp.up_proj": 2048,
59
+ "model.layers.11.zaya_block.experts.switch_mlp.down_proj": 2048,
60
+ "model.layers.12.zaya_block.experts.switch_mlp.gate_proj": 2048,
61
+ "model.layers.12.zaya_block.experts.switch_mlp.up_proj": 2048,
62
+ "model.layers.12.zaya_block.experts.switch_mlp.down_proj": 2048,
63
+ "model.layers.13.zaya_block.experts.switch_mlp.gate_proj": 2048,
64
+ "model.layers.13.zaya_block.experts.switch_mlp.up_proj": 2048,
65
+ "model.layers.13.zaya_block.experts.switch_mlp.down_proj": 2048,
66
+ "model.layers.14.zaya_block.experts.switch_mlp.gate_proj": 2048,
67
+ "model.layers.14.zaya_block.experts.switch_mlp.up_proj": 2048,
68
+ "model.layers.14.zaya_block.experts.switch_mlp.down_proj": 2048,
69
+ "model.layers.15.zaya_block.experts.switch_mlp.gate_proj": 2048,
70
+ "model.layers.15.zaya_block.experts.switch_mlp.up_proj": 2048,
71
+ "model.layers.15.zaya_block.experts.switch_mlp.down_proj": 2048,
72
+ "model.layers.16.zaya_block.experts.switch_mlp.gate_proj": 2048,
73
+ "model.layers.16.zaya_block.experts.switch_mlp.up_proj": 2048,
74
+ "model.layers.16.zaya_block.experts.switch_mlp.down_proj": 2048,
75
+ "model.layers.17.zaya_block.experts.switch_mlp.gate_proj": 2048,
76
+ "model.layers.17.zaya_block.experts.switch_mlp.up_proj": 2048,
77
+ "model.layers.17.zaya_block.experts.switch_mlp.down_proj": 2048,
78
+ "model.layers.18.zaya_block.experts.switch_mlp.gate_proj": 2048,
79
+ "model.layers.18.zaya_block.experts.switch_mlp.up_proj": 2048,
80
+ "model.layers.18.zaya_block.experts.switch_mlp.down_proj": 2048,
81
+ "model.layers.19.zaya_block.experts.switch_mlp.gate_proj": 2048,
82
+ "model.layers.19.zaya_block.experts.switch_mlp.up_proj": 2048,
83
+ "model.layers.19.zaya_block.experts.switch_mlp.down_proj": 2048,
84
+ "model.layers.20.zaya_block.experts.switch_mlp.gate_proj": 2048,
85
+ "model.layers.20.zaya_block.experts.switch_mlp.up_proj": 2048,
86
+ "model.layers.20.zaya_block.experts.switch_mlp.down_proj": 2048,
87
+ "model.layers.21.zaya_block.experts.switch_mlp.gate_proj": 2048,
88
+ "model.layers.21.zaya_block.experts.switch_mlp.up_proj": 2048,
89
+ "model.layers.21.zaya_block.experts.switch_mlp.down_proj": 2048,
90
+ "model.layers.22.zaya_block.experts.switch_mlp.gate_proj": 2048,
91
+ "model.layers.22.zaya_block.experts.switch_mlp.up_proj": 2048,
92
+ "model.layers.22.zaya_block.experts.switch_mlp.down_proj": 2048,
93
+ "model.layers.23.zaya_block.experts.switch_mlp.gate_proj": 2048,
94
+ "model.layers.23.zaya_block.experts.switch_mlp.up_proj": 2048,
95
+ "model.layers.23.zaya_block.experts.switch_mlp.down_proj": 2048,
96
+ "model.layers.24.zaya_block.experts.switch_mlp.gate_proj": 2048,
97
+ "model.layers.24.zaya_block.experts.switch_mlp.up_proj": 2048,
98
+ "model.layers.24.zaya_block.experts.switch_mlp.down_proj": 2048,
99
+ "model.layers.25.zaya_block.experts.switch_mlp.gate_proj": 2048,
100
+ "model.layers.25.zaya_block.experts.switch_mlp.up_proj": 2048,
101
+ "model.layers.25.zaya_block.experts.switch_mlp.down_proj": 2048,
102
+ "model.layers.26.zaya_block.experts.switch_mlp.gate_proj": 2048,
103
+ "model.layers.26.zaya_block.experts.switch_mlp.up_proj": 2048,
104
+ "model.layers.26.zaya_block.experts.switch_mlp.down_proj": 2048,
105
+ "model.layers.27.zaya_block.experts.switch_mlp.gate_proj": 2048,
106
+ "model.layers.27.zaya_block.experts.switch_mlp.up_proj": 2048,
107
+ "model.layers.27.zaya_block.experts.switch_mlp.down_proj": 2048,
108
+ "model.layers.28.zaya_block.experts.switch_mlp.gate_proj": 2048,
109
+ "model.layers.28.zaya_block.experts.switch_mlp.up_proj": 2048,
110
+ "model.layers.28.zaya_block.experts.switch_mlp.down_proj": 2048,
111
+ "model.layers.29.zaya_block.experts.switch_mlp.gate_proj": 2048,
112
+ "model.layers.29.zaya_block.experts.switch_mlp.up_proj": 2048,
113
+ "model.layers.29.zaya_block.experts.switch_mlp.down_proj": 2048,
114
+ "model.layers.30.zaya_block.experts.switch_mlp.gate_proj": 2048,
115
+ "model.layers.30.zaya_block.experts.switch_mlp.up_proj": 2048,
116
+ "model.layers.30.zaya_block.experts.switch_mlp.down_proj": 2048,
117
+ "model.layers.31.zaya_block.experts.switch_mlp.gate_proj": 2048,
118
+ "model.layers.31.zaya_block.experts.switch_mlp.up_proj": 2048,
119
+ "model.layers.31.zaya_block.experts.switch_mlp.down_proj": 2048,
120
+ "model.layers.32.zaya_block.experts.switch_mlp.gate_proj": 2048,
121
+ "model.layers.32.zaya_block.experts.switch_mlp.up_proj": 2048,
122
+ "model.layers.32.zaya_block.experts.switch_mlp.down_proj": 2048,
123
+ "model.layers.33.zaya_block.experts.switch_mlp.gate_proj": 2048,
124
+ "model.layers.33.zaya_block.experts.switch_mlp.up_proj": 2048,
125
+ "model.layers.33.zaya_block.experts.switch_mlp.down_proj": 2048,
126
+ "model.layers.34.zaya_block.experts.switch_mlp.gate_proj": 2048,
127
+ "model.layers.34.zaya_block.experts.switch_mlp.up_proj": 2048,
128
+ "model.layers.34.zaya_block.experts.switch_mlp.down_proj": 2048,
129
+ "model.layers.35.zaya_block.experts.switch_mlp.gate_proj": 2048,
130
+ "model.layers.35.zaya_block.experts.switch_mlp.up_proj": 2048,
131
+ "model.layers.35.zaya_block.experts.switch_mlp.down_proj": 2048,
132
+ "model.layers.36.zaya_block.experts.switch_mlp.gate_proj": 2048,
133
+ "model.layers.36.zaya_block.experts.switch_mlp.up_proj": 2048,
134
+ "model.layers.36.zaya_block.experts.switch_mlp.down_proj": 2048,
135
+ "model.layers.37.zaya_block.experts.switch_mlp.gate_proj": 2048,
136
+ "model.layers.37.zaya_block.experts.switch_mlp.up_proj": 2048,
137
+ "model.layers.37.zaya_block.experts.switch_mlp.down_proj": 2048,
138
+ "model.layers.38.zaya_block.experts.switch_mlp.gate_proj": 2048,
139
+ "model.layers.38.zaya_block.experts.switch_mlp.up_proj": 2048,
140
+ "model.layers.38.zaya_block.experts.switch_mlp.down_proj": 2048,
141
+ "model.layers.39.zaya_block.experts.switch_mlp.gate_proj": 2048,
142
+ "model.layers.39.zaya_block.experts.switch_mlp.up_proj": 2048,
143
+ "model.layers.39.zaya_block.experts.switch_mlp.down_proj": 2048
144
+ },
145
+ "quantization": {
146
+ "method": "affine+mxtq",
147
+ "group_size": 32,
148
+ "bits_default": 4
149
+ },
150
+ "capabilities": {
151
+ "reasoning_parser": "qwen3",
152
+ "tool_parser": "zaya_xml",
153
+ "think_in_template": false,
154
+ "supports_tools": true,
155
+ "supports_thinking": true,
156
+ "family": "zaya1_vl",
157
+ "modality": "vision",
158
+ "cache_type": "hybrid"
159
+ }
160
+ }
jangq-logo.png ADDED
jangtq_runtime.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72e7ee33628684761e3c8e3094597534d5109647ece6315c5fd95a18f4d0c910
3
+ size 8408
model-00001-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8033842daa03a15a79c9966313e29a25553a74412995bac388318086636fb503
3
+ size 1003299194
model-00002-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1980dcd150e9cd191b5a48d2f6dcc2b30a165d26bf11b5909095106ec260f717
3
+ size 1034305792
model-00003-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:149ab6cc39280756feaddef5aa4bd11a8ecdb5964b53ca9ad5dd33bbe0dec8f1
3
+ size 1017662488
model-00004-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37f9c688a2313bd3557569ca68d07835495b2b96e833e54958474c9d97af452f
3
+ size 1008611406
model-00005-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01621a74c430fc64bfbad5143da66c45b34fd71a0caf5d51f4ee3d2c65551ef1
3
+ size 1008611462
model-00006-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b738acdea907ae59c923ee54f025a166f773fa3745ab2d897ae6321d3e27433c
3
+ size 1008611462
model-00007-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:679486c732420ad71e140189f8a9cad06269d533e913bcf9142faddab8f94e6d
3
+ size 638852988
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_convert_rgb": true,
3
+ "do_normalize": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "image_mean": [
7
+ 0.48145466,
8
+ 0.4578275,
9
+ 0.40821073
10
+ ],
11
+ "image_processor_type": "Qwen2VLImageProcessor",
12
+ "image_std": [
13
+ 0.26862954,
14
+ 0.26130258,
15
+ 0.27577711
16
+ ],
17
+ "max_pixels": 12845056,
18
+ "merge_size": 2,
19
+ "min_pixels": 3136,
20
+ "patch_size": 14,
21
+ "processor_class": "Zaya1VLProcessor",
22
+ "resample": 3,
23
+ "rescale_factor": 0.00392156862745098,
24
+ "size": {
25
+ "longest_edge": 12845056,
26
+ "shortest_edge": 3136
27
+ },
28
+ "temporal_patch_size": 1
29
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "answer_token": "<|im_start|>",
3
+ "boi_token": "<|vision_start|>",
4
+ "bos_token": {
5
+ "content": "<bos>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "eoi_token": "<|vision_end|>",
12
+ "eos_token": {
13
+ "content": "<|im_end|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false
18
+ },
19
+ "image_token": "<image>",
20
+ "pad_token": {
21
+ "content": "<pad>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false
26
+ },
27
+ "unk_token": {
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false
33
+ },
34
+ "video_token": "<video>"
35
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a90181b1298e5d8c2f211f15dca261650d85c1f2a5a3bbfff852a853d21bed8f
3
+ size 33385329
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff