Hugging Face's logo

HoomanBio2021
/
LabOpen-v1

Text Generation
Transformers
Safetensors
English
Chinese
qwen3_5
image-text-to-text
qwen
lora
reinforcement-learning
fp8
agentic-coding
scientific-computing
physics
materials-science
genetics
conversational
Model card Files
xet
 Community
HoomanBio commited on
Commit
0e176c3
·
verified ·
1 Parent(s): 24fba1d

Add files using upload-large-folder tool

Browse files
Files changed (20) hide show
  1. .gitattributes +1 -0
  2. MERGE_REPORT.md +62 -0
  3. adapter/adapter_model.safetensors +3 -0
  4. chat_template.jinja +170 -0
  5. config.json +1033 -0
  6. generation_config.json +12 -0
  7. merges.txt +0 -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 +21 -0
  17. tokenizer.json +3 -0
  18. tokenizer_config.json +305 -0
  19. video_preprocessor_config.json +21 -0
  20. vocab.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
MERGE_REPORT.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Merge Report — Qwen3.8-27B-LabOpen v1 merged FP8 weights
2
+
3
+ This document records how the merged full-weight checkpoint in this repository was produced from the LoRA adapter, and how it was verified. Internal filesystem paths are omitted; digests are reproduced verbatim from the training run.
4
+
5
+ ## Method
6
+
7
+ Per target module (LoRA r=4, alpha=4, scaling 1.0, 496 modules = 400 blockwise-FP8 + 96 BF16):
8
+
9
+ 1. Dequantize base weight: `W = codes * scale` per 128x128 block (FP32)
10
+ 2. Add LoRA delta: `W_merged = W + (B @ A) * scaling`
11
+ 3. Re-quantize per 128x128 block: `scale = amax(|block|)/448`, codes clamped to +-448 (e4m3)
12
+
13
+ Non-target tensors are byte-copies of the base. Tensor layout, shard split and file format are identical to the base FP8 release.
14
+
15
+ ## Provenance
16
+
17
+ - Trainable digest (checkpoint B): `96ef57e87e0d64de0340f9b917f4b9a7276356908bc6a5a7e59caa594bf68e03`
18
+ - FP8 modules merged: 400; BF16 modules merged: 96; override tensors written: 896 (400 weights + 400 scales + 96 weights)
19
+ - Total checkpoint size: 30,866,663,264 bytes
20
+
21
+ ## Re-quantization error (vs exact FP32 merged weights)
22
+
23
+ - Mean relative error: 0.330%
24
+ - Max relative error: 1.00 (isolated near-zero elements; same order as the FP8 storage noise the base itself carries)
25
+
26
+ ## Output-equivalence probes (greedy, 3 prompts x 48 tokens)
27
+
28
+ A. In-memory merged model vs adapter (scales in FP32):
29
+
30
+ | Prompt | Prefix-equal tokens | Token match |
31
+ |---|---|---|
32
+ | A 2 kg block slides down a frictionless ... | 41/48 | 87.5% |
33
+ | def quicksort(arr):
34
+ if len(arr) <= 1... | 48/48 | 100.0% |
35
+ | Water flows at 3 m/s through a pipe of d... | 1/48 | 4.2% |
36
+
37
+ B. **Saved-and-reloaded checkpoint (the published artifact, scales serialized BF16) vs adapter — release gate:**
38
+
39
+ | Prompt | Prefix-equal tokens | Token match |
40
+ |---|---|---|
41
+ | prompt 0 | 48/48 | 100.0% |
42
+ | prompt 1 | 48/48 | 100.0% |
43
+ | prompt 2 | 48/48 | 100.0% |
44
+
45
+ **Gate result: PASSED — 3/3 prompts, 144/144 tokens identical to the adapter trajectory.**
46
+
47
+ ## Shard SHA-256
48
+
49
+ | File | SHA-256 | Bytes |
50
+ |---|---|---|
51
+ | `model-00001-of-00007.safetensors` | `3eeeab65368b86c1eec3582495f5367c9a6a6ac25f0857adbe7f8026ea322d3f` | 3,969,317,448 |
52
+ | `model-00002-of-00007.safetensors` | `7091dbd7077b8d19bb547eaf24997ec3c2b6b0acef8f2678f1a7a6a15c212a3e` | 4,939,321,704 |
53
+ | `model-00003-of-00007.safetensors` | `7ec75e6485b732125098c0080669ddbcef007be82bd6830a07e566b9a9ee3264` | 4,939,863,424 |
54
+ | `model-00004-of-00007.safetensors` | `44fef978664d99a87d6c4e5826e51486eb4e28788ca791b694b29430432a12fe` | 4,971,328,696 |
55
+ | `model-00005-of-00007.safetensors` | `f815eef9e6ed5194fdc0ac7412ad29019661ca6062bcd454fb4ce341834e29a4` | 4,955,596,064 |
56
+ | `model-00006-of-00007.safetensors` | `d2de7016e251a68f0a7ba2c49615c243c20ad87dfaacca252d6b2c0277f5fbbb` | 4,071,440,632 |
57
+ | `model-00007-of-00007.safetensors` | `1bed352a0625d22dc606de47fd3563f24fd9fa73915c4ddffe7947c4b56c28bf` | 3,019,999,168 |
58
+
59
+ ## Notes
60
+
61
+ - The adapter (`adapter/`) remains the numerically exact training artifact; the merged checkpoint adds only the measured re-quantization error above.
62
+ - Probes ran on the training stack runtime (identical kernels to the serving configuration used for evaluation).
adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67a23e9dd76c3aa234cbaa369c29befb0a23abd31f6cd68b9948b42c002affd7
3
+ size 116861816
chat_template.jinja ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- set reasoning_instructions = '' %}
46
+ {%- if enable_thinking is undefined or enable_thinking is true %}
47
+ {%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}
48
+ {%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}
49
+ {{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}
50
+ {%- endif %}
51
+ {%- if resolved_reasoning_effort == 'xhigh' %}
52
+ {%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}
53
+ {%- elif resolved_reasoning_effort == 'low' %}
54
+ {%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}
55
+ {%- endif %}
56
+ {%- endif %}
57
+ {%- if tools and tools is iterable and tools is not mapping %}
58
+ {{- '<|im_start|>system\n' }}
59
+ {%- if reasoning_instructions %}
60
+ {{- reasoning_instructions + '\n\n' }}
61
+ {%- endif %}
62
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
63
+ {%- for tool in tools %}
64
+ {{- "\n" }}
65
+ {{- tool | tojson }}
66
+ {%- endfor %}
67
+ {{- "\n</tools>" }}
68
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
69
+ {%- if messages[0].role == 'system' %}
70
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
71
+ {%- if content %}
72
+ {{- '\n\n' + content }}
73
+ {%- endif %}
74
+ {%- endif %}
75
+ {{- '<|im_end|>\n' }}
76
+ {%- else %}
77
+ {%- if messages[0].role == 'system' %}
78
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
79
+ {%- if content %}
80
+ {{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' }}
81
+ {%- elif reasoning_instructions %}
82
+ {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
83
+ {%- endif %}
84
+ {%- elif reasoning_instructions %}
85
+ {{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
86
+ {%- endif %}
87
+ {%- endif %}
88
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
89
+ {%- for message in messages[::-1] %}
90
+ {%- set index = (messages|length - 1) - loop.index0 %}
91
+ {%- if ns.multi_step_tool and message.role == "user" %}
92
+ {%- set content = render_content(message.content, false)|trim %}
93
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
94
+ {%- set ns.multi_step_tool = false %}
95
+ {%- set ns.last_query_index = index %}
96
+ {%- endif %}
97
+ {%- endif %}
98
+ {%- endfor %}
99
+ {%- if ns.multi_step_tool %}
100
+ {{- raise_exception('No user query found in messages.') }}
101
+ {%- endif %}
102
+ {%- for message in messages %}
103
+ {%- set content = render_content(message.content, true)|trim %}
104
+ {%- if message.role == "system" %}
105
+ {%- if not loop.first %}
106
+ {{- raise_exception('System message must be at the beginning.') }}
107
+ {%- endif %}
108
+ {%- elif message.role == "user" %}
109
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
110
+ {%- elif message.role == "assistant" %}
111
+ {%- set reasoning_content = '' %}
112
+ {%- if message.reasoning_content is string %}
113
+ {%- set reasoning_content = message.reasoning_content %}
114
+ {%- endif %}
115
+ {%- set reasoning_content = reasoning_content|trim %}
116
+ {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}
117
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
118
+ {%- else %}
119
+ {{- '<|im_start|>' + message.role + '\n' + content }}
120
+ {%- endif %}
121
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
122
+ {%- for tool_call in message.tool_calls %}
123
+ {%- if tool_call.function is defined %}
124
+ {%- set tool_call = tool_call.function %}
125
+ {%- endif %}
126
+ {%- if loop.first %}
127
+ {%- if content|trim %}
128
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
129
+ {%- else %}
130
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
131
+ {%- endif %}
132
+ {%- else %}
133
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
134
+ {%- endif %}
135
+ {%- if tool_call.arguments is defined and tool_call.arguments != '' %}
136
+ {%- for args_name, args_value in tool_call.arguments|items %}
137
+ {{- '<parameter=' + args_name + '>\n' }}
138
+ {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
139
+ {{- args_value }}
140
+ {{- '\n</parameter>\n' }}
141
+ {%- endfor %}
142
+ {%- endif %}
143
+ {{- '</function>\n</tool_call>' }}
144
+ {%- endfor %}
145
+ {%- endif %}
146
+ {{- '<|im_end|>\n' }}
147
+ {%- elif message.role == "tool" %}
148
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
149
+ {{- '<|im_start|>user' }}
150
+ {%- endif %}
151
+ {{- '\n<tool_response>\n' }}
152
+ {{- content }}
153
+ {{- '\n</tool_response>' }}
154
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
155
+ {{- '<|im_end|>\n' }}
156
+ {%- elif loop.last %}
157
+ {{- '<|im_end|>\n' }}
158
+ {%- endif %}
159
+ {%- else %}
160
+ {{- raise_exception('Unexpected message role.') }}
161
+ {%- endif %}
162
+ {%- endfor %}
163
+ {%- if add_generation_prompt %}
164
+ {{- '<|im_start|>assistant\n' }}
165
+ {%- if enable_thinking is defined and enable_thinking is false %}
166
+ {{- '<think>\n\n</think>\n\n' }}
167
+ {%- else %}
168
+ {{- '<think>\n' }}
169
+ {%- endif %}
170
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,1033 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "image_token_id": 248056,
6
+ "language_model_only": false,
7
+ "model_type": "qwen3_5",
8
+ "text_config": {
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_output_gate": true,
12
+ "bos_token_id": 248044,
13
+ "dtype": "bfloat16",
14
+ "eos_token_id": 248044,
15
+ "full_attention_interval": 4,
16
+ "head_dim": 256,
17
+ "hidden_act": "silu",
18
+ "hidden_size": 5120,
19
+ "initializer_range": 0.02,
20
+ "intermediate_size": 17408,
21
+ "layer_types": [
22
+ "linear_attention",
23
+ "linear_attention",
24
+ "linear_attention",
25
+ "full_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "linear_attention",
29
+ "full_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "linear_attention",
33
+ "full_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "linear_attention",
37
+ "full_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "linear_attention",
41
+ "full_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "linear_attention",
45
+ "full_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "linear_attention",
49
+ "full_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "linear_attention",
53
+ "full_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "linear_attention",
57
+ "full_attention",
58
+ "linear_attention",
59
+ "linear_attention",
60
+ "linear_attention",
61
+ "full_attention",
62
+ "linear_attention",
63
+ "linear_attention",
64
+ "linear_attention",
65
+ "full_attention",
66
+ "linear_attention",
67
+ "linear_attention",
68
+ "linear_attention",
69
+ "full_attention",
70
+ "linear_attention",
71
+ "linear_attention",
72
+ "linear_attention",
73
+ "full_attention",
74
+ "linear_attention",
75
+ "linear_attention",
76
+ "linear_attention",
77
+ "full_attention",
78
+ "linear_attention",
79
+ "linear_attention",
80
+ "linear_attention",
81
+ "full_attention",
82
+ "linear_attention",
83
+ "linear_attention",
84
+ "linear_attention",
85
+ "full_attention"
86
+ ],
87
+ "linear_conv_kernel_dim": 4,
88
+ "linear_key_head_dim": 128,
89
+ "linear_num_key_heads": 16,
90
+ "linear_num_value_heads": 48,
91
+ "linear_value_head_dim": 128,
92
+ "mamba_ssm_dtype": "float32",
93
+ "max_position_embeddings": 262144,
94
+ "model_type": "qwen3_5_text",
95
+ "mtp_num_hidden_layers": 1,
96
+ "mtp_use_dedicated_embeddings": false,
97
+ "num_attention_heads": 24,
98
+ "num_hidden_layers": 64,
99
+ "num_key_value_heads": 4,
100
+ "output_gate_type": "swish",
101
+ "pad_token_id": null,
102
+ "partial_rotary_factor": 0.25,
103
+ "rms_norm_eps": 1e-06,
104
+ "rope_parameters": {
105
+ "mrope_interleaved": true,
106
+ "mrope_section": [
107
+ 11,
108
+ 11,
109
+ 10
110
+ ],
111
+ "partial_rotary_factor": 0.25,
112
+ "rope_theta": 10000000,
113
+ "rope_type": "default"
114
+ },
115
+ "tie_word_embeddings": false,
116
+ "use_cache": true,
117
+ "vocab_size": 248320
118
+ },
119
+ "tie_word_embeddings": false,
120
+ "transformers_version": "5.8.0.dev0",
121
+ "video_token_id": 248057,
122
+ "vision_config": {
123
+ "deepstack_visual_indexes": [],
124
+ "depth": 27,
125
+ "hidden_act": "gelu_pytorch_tanh",
126
+ "hidden_size": 1152,
127
+ "in_channels": 3,
128
+ "initializer_range": 0.02,
129
+ "intermediate_size": 4304,
130
+ "model_type": "qwen3_5",
131
+ "num_heads": 16,
132
+ "num_position_embeddings": 2304,
133
+ "out_hidden_size": 5120,
134
+ "patch_size": 16,
135
+ "spatial_merge_size": 2,
136
+ "temporal_patch_size": 2
137
+ },
138
+ "vision_end_token_id": 248054,
139
+ "vision_start_token_id": 248053,
140
+ "quantization_config": {
141
+ "activation_scheme": "dynamic",
142
+ "fmt": "e4m3",
143
+ "quant_method": "fp8",
144
+ "modules_to_not_convert": [
145
+ "model.visual.blocks.0.attn.proj",
146
+ "model.visual.blocks.0.attn.qkv",
147
+ "model.visual.blocks.0.mlp.linear_fc1",
148
+ "model.visual.blocks.0.mlp.linear_fc2",
149
+ "visual.blocks.0.attn.proj",
150
+ "visual.blocks.0.attn.qkv_proj",
151
+ "visual.blocks.0.mlp.linear_fc1",
152
+ "visual.blocks.0.mlp.linear_fc2",
153
+ "model.visual.blocks.1.attn.proj",
154
+ "model.visual.blocks.1.attn.qkv",
155
+ "model.visual.blocks.1.mlp.linear_fc1",
156
+ "model.visual.blocks.1.mlp.linear_fc2",
157
+ "visual.blocks.1.attn.proj",
158
+ "visual.blocks.1.attn.qkv_proj",
159
+ "visual.blocks.1.mlp.linear_fc1",
160
+ "visual.blocks.1.mlp.linear_fc2",
161
+ "model.visual.blocks.2.attn.proj",
162
+ "model.visual.blocks.2.attn.qkv",
163
+ "model.visual.blocks.2.mlp.linear_fc1",
164
+ "model.visual.blocks.2.mlp.linear_fc2",
165
+ "visual.blocks.2.attn.proj",
166
+ "visual.blocks.2.attn.qkv_proj",
167
+ "visual.blocks.2.mlp.linear_fc1",
168
+ "visual.blocks.2.mlp.linear_fc2",
169
+ "model.visual.blocks.3.attn.proj",
170
+ "model.visual.blocks.3.attn.qkv",
171
+ "model.visual.blocks.3.mlp.linear_fc1",
172
+ "model.visual.blocks.3.mlp.linear_fc2",
173
+ "visual.blocks.3.attn.proj",
174
+ "visual.blocks.3.attn.qkv_proj",
175
+ "visual.blocks.3.mlp.linear_fc1",
176
+ "visual.blocks.3.mlp.linear_fc2",
177
+ "model.visual.blocks.4.attn.proj",
178
+ "model.visual.blocks.4.attn.qkv",
179
+ "model.visual.blocks.4.mlp.linear_fc1",
180
+ "model.visual.blocks.4.mlp.linear_fc2",
181
+ "visual.blocks.4.attn.proj",
182
+ "visual.blocks.4.attn.qkv_proj",
183
+ "visual.blocks.4.mlp.linear_fc1",
184
+ "visual.blocks.4.mlp.linear_fc2",
185
+ "model.visual.blocks.5.attn.proj",
186
+ "model.visual.blocks.5.attn.qkv",
187
+ "model.visual.blocks.5.mlp.linear_fc1",
188
+ "model.visual.blocks.5.mlp.linear_fc2",
189
+ "visual.blocks.5.attn.proj",
190
+ "visual.blocks.5.attn.qkv_proj",
191
+ "visual.blocks.5.mlp.linear_fc1",
192
+ "visual.blocks.5.mlp.linear_fc2",
193
+ "model.visual.blocks.6.attn.proj",
194
+ "model.visual.blocks.6.attn.qkv",
195
+ "model.visual.blocks.6.mlp.linear_fc1",
196
+ "model.visual.blocks.6.mlp.linear_fc2",
197
+ "visual.blocks.6.attn.proj",
198
+ "visual.blocks.6.attn.qkv_proj",
199
+ "visual.blocks.6.mlp.linear_fc1",
200
+ "visual.blocks.6.mlp.linear_fc2",
201
+ "model.visual.blocks.7.attn.proj",
202
+ "model.visual.blocks.7.attn.qkv",
203
+ "model.visual.blocks.7.mlp.linear_fc1",
204
+ "model.visual.blocks.7.mlp.linear_fc2",
205
+ "visual.blocks.7.attn.proj",
206
+ "visual.blocks.7.attn.qkv_proj",
207
+ "visual.blocks.7.mlp.linear_fc1",
208
+ "visual.blocks.7.mlp.linear_fc2",
209
+ "model.visual.blocks.8.attn.proj",
210
+ "model.visual.blocks.8.attn.qkv",
211
+ "model.visual.blocks.8.mlp.linear_fc1",
212
+ "model.visual.blocks.8.mlp.linear_fc2",
213
+ "visual.blocks.8.attn.proj",
214
+ "visual.blocks.8.attn.qkv_proj",
215
+ "visual.blocks.8.mlp.linear_fc1",
216
+ "visual.blocks.8.mlp.linear_fc2",
217
+ "model.visual.blocks.9.attn.proj",
218
+ "model.visual.blocks.9.attn.qkv",
219
+ "model.visual.blocks.9.mlp.linear_fc1",
220
+ "model.visual.blocks.9.mlp.linear_fc2",
221
+ "visual.blocks.9.attn.proj",
222
+ "visual.blocks.9.attn.qkv_proj",
223
+ "visual.blocks.9.mlp.linear_fc1",
224
+ "visual.blocks.9.mlp.linear_fc2",
225
+ "model.visual.blocks.10.attn.proj",
226
+ "model.visual.blocks.10.attn.qkv",
227
+ "model.visual.blocks.10.mlp.linear_fc1",
228
+ "model.visual.blocks.10.mlp.linear_fc2",
229
+ "visual.blocks.10.attn.proj",
230
+ "visual.blocks.10.attn.qkv_proj",
231
+ "visual.blocks.10.mlp.linear_fc1",
232
+ "visual.blocks.10.mlp.linear_fc2",
233
+ "model.visual.blocks.11.attn.proj",
234
+ "model.visual.blocks.11.attn.qkv",
235
+ "model.visual.blocks.11.mlp.linear_fc1",
236
+ "model.visual.blocks.11.mlp.linear_fc2",
237
+ "visual.blocks.11.attn.proj",
238
+ "visual.blocks.11.attn.qkv_proj",
239
+ "visual.blocks.11.mlp.linear_fc1",
240
+ "visual.blocks.11.mlp.linear_fc2",
241
+ "model.visual.blocks.12.attn.proj",
242
+ "model.visual.blocks.12.attn.qkv",
243
+ "model.visual.blocks.12.mlp.linear_fc1",
244
+ "model.visual.blocks.12.mlp.linear_fc2",
245
+ "visual.blocks.12.attn.proj",
246
+ "visual.blocks.12.attn.qkv_proj",
247
+ "visual.blocks.12.mlp.linear_fc1",
248
+ "visual.blocks.12.mlp.linear_fc2",
249
+ "model.visual.blocks.13.attn.proj",
250
+ "model.visual.blocks.13.attn.qkv",
251
+ "model.visual.blocks.13.mlp.linear_fc1",
252
+ "model.visual.blocks.13.mlp.linear_fc2",
253
+ "visual.blocks.13.attn.proj",
254
+ "visual.blocks.13.attn.qkv_proj",
255
+ "visual.blocks.13.mlp.linear_fc1",
256
+ "visual.blocks.13.mlp.linear_fc2",
257
+ "model.visual.blocks.14.attn.proj",
258
+ "model.visual.blocks.14.attn.qkv",
259
+ "model.visual.blocks.14.mlp.linear_fc1",
260
+ "model.visual.blocks.14.mlp.linear_fc2",
261
+ "visual.blocks.14.attn.proj",
262
+ "visual.blocks.14.attn.qkv_proj",
263
+ "visual.blocks.14.mlp.linear_fc1",
264
+ "visual.blocks.14.mlp.linear_fc2",
265
+ "model.visual.blocks.15.attn.proj",
266
+ "model.visual.blocks.15.attn.qkv",
267
+ "model.visual.blocks.15.mlp.linear_fc1",
268
+ "model.visual.blocks.15.mlp.linear_fc2",
269
+ "visual.blocks.15.attn.proj",
270
+ "visual.blocks.15.attn.qkv_proj",
271
+ "visual.blocks.15.mlp.linear_fc1",
272
+ "visual.blocks.15.mlp.linear_fc2",
273
+ "model.visual.blocks.16.attn.proj",
274
+ "model.visual.blocks.16.attn.qkv",
275
+ "model.visual.blocks.16.mlp.linear_fc1",
276
+ "model.visual.blocks.16.mlp.linear_fc2",
277
+ "visual.blocks.16.attn.proj",
278
+ "visual.blocks.16.attn.qkv_proj",
279
+ "visual.blocks.16.mlp.linear_fc1",
280
+ "visual.blocks.16.mlp.linear_fc2",
281
+ "model.visual.blocks.17.attn.proj",
282
+ "model.visual.blocks.17.attn.qkv",
283
+ "model.visual.blocks.17.mlp.linear_fc1",
284
+ "model.visual.blocks.17.mlp.linear_fc2",
285
+ "visual.blocks.17.attn.proj",
286
+ "visual.blocks.17.attn.qkv_proj",
287
+ "visual.blocks.17.mlp.linear_fc1",
288
+ "visual.blocks.17.mlp.linear_fc2",
289
+ "model.visual.blocks.18.attn.proj",
290
+ "model.visual.blocks.18.attn.qkv",
291
+ "model.visual.blocks.18.mlp.linear_fc1",
292
+ "model.visual.blocks.18.mlp.linear_fc2",
293
+ "visual.blocks.18.attn.proj",
294
+ "visual.blocks.18.attn.qkv_proj",
295
+ "visual.blocks.18.mlp.linear_fc1",
296
+ "visual.blocks.18.mlp.linear_fc2",
297
+ "model.visual.blocks.19.attn.proj",
298
+ "model.visual.blocks.19.attn.qkv",
299
+ "model.visual.blocks.19.mlp.linear_fc1",
300
+ "model.visual.blocks.19.mlp.linear_fc2",
301
+ "visual.blocks.19.attn.proj",
302
+ "visual.blocks.19.attn.qkv_proj",
303
+ "visual.blocks.19.mlp.linear_fc1",
304
+ "visual.blocks.19.mlp.linear_fc2",
305
+ "model.visual.blocks.20.attn.proj",
306
+ "model.visual.blocks.20.attn.qkv",
307
+ "model.visual.blocks.20.mlp.linear_fc1",
308
+ "model.visual.blocks.20.mlp.linear_fc2",
309
+ "visual.blocks.20.attn.proj",
310
+ "visual.blocks.20.attn.qkv_proj",
311
+ "visual.blocks.20.mlp.linear_fc1",
312
+ "visual.blocks.20.mlp.linear_fc2",
313
+ "model.visual.blocks.21.attn.proj",
314
+ "model.visual.blocks.21.attn.qkv",
315
+ "model.visual.blocks.21.mlp.linear_fc1",
316
+ "model.visual.blocks.21.mlp.linear_fc2",
317
+ "visual.blocks.21.attn.proj",
318
+ "visual.blocks.21.attn.qkv_proj",
319
+ "visual.blocks.21.mlp.linear_fc1",
320
+ "visual.blocks.21.mlp.linear_fc2",
321
+ "model.visual.blocks.22.attn.proj",
322
+ "model.visual.blocks.22.attn.qkv",
323
+ "model.visual.blocks.22.mlp.linear_fc1",
324
+ "model.visual.blocks.22.mlp.linear_fc2",
325
+ "visual.blocks.22.attn.proj",
326
+ "visual.blocks.22.attn.qkv_proj",
327
+ "visual.blocks.22.mlp.linear_fc1",
328
+ "visual.blocks.22.mlp.linear_fc2",
329
+ "model.visual.blocks.23.attn.proj",
330
+ "model.visual.blocks.23.attn.qkv",
331
+ "model.visual.blocks.23.mlp.linear_fc1",
332
+ "model.visual.blocks.23.mlp.linear_fc2",
333
+ "visual.blocks.23.attn.proj",
334
+ "visual.blocks.23.attn.qkv_proj",
335
+ "visual.blocks.23.mlp.linear_fc1",
336
+ "visual.blocks.23.mlp.linear_fc2",
337
+ "model.visual.blocks.24.attn.proj",
338
+ "model.visual.blocks.24.attn.qkv",
339
+ "model.visual.blocks.24.mlp.linear_fc1",
340
+ "model.visual.blocks.24.mlp.linear_fc2",
341
+ "visual.blocks.24.attn.proj",
342
+ "visual.blocks.24.attn.qkv_proj",
343
+ "visual.blocks.24.mlp.linear_fc1",
344
+ "visual.blocks.24.mlp.linear_fc2",
345
+ "model.visual.blocks.25.attn.proj",
346
+ "model.visual.blocks.25.attn.qkv",
347
+ "model.visual.blocks.25.mlp.linear_fc1",
348
+ "model.visual.blocks.25.mlp.linear_fc2",
349
+ "visual.blocks.25.attn.proj",
350
+ "visual.blocks.25.attn.qkv_proj",
351
+ "visual.blocks.25.mlp.linear_fc1",
352
+ "visual.blocks.25.mlp.linear_fc2",
353
+ "model.visual.blocks.26.attn.proj",
354
+ "model.visual.blocks.26.attn.qkv",
355
+ "model.visual.blocks.26.mlp.linear_fc1",
356
+ "model.visual.blocks.26.mlp.linear_fc2",
357
+ "visual.blocks.26.attn.proj",
358
+ "visual.blocks.26.attn.qkv_proj",
359
+ "visual.blocks.26.mlp.linear_fc1",
360
+ "visual.blocks.26.mlp.linear_fc2",
361
+ "model.visual.deepstack_merger_list.0.linear_fc1",
362
+ "model.visual.deepstack_merger_list.0.linear_fc2",
363
+ "model.visual.deepstack_merger_list.0.norm",
364
+ "visual.deepstack_merger_list.0.linear_fc1",
365
+ "visual.deepstack_merger_list.0.linear_fc2",
366
+ "visual.deepstack_merger_list.0.norm",
367
+ "model.visual.deepstack_merger_list.1.linear_fc1",
368
+ "model.visual.deepstack_merger_list.1.linear_fc2",
369
+ "model.visual.deepstack_merger_list.1.norm",
370
+ "visual.deepstack_merger_list.1.linear_fc1",
371
+ "visual.deepstack_merger_list.1.linear_fc2",
372
+ "visual.deepstack_merger_list.1.norm",
373
+ "model.visual.deepstack_merger_list.2.linear_fc1",
374
+ "model.visual.deepstack_merger_list.2.linear_fc2",
375
+ "model.visual.deepstack_merger_list.2.norm",
376
+ "visual.deepstack_merger_list.2.linear_fc1",
377
+ "visual.deepstack_merger_list.2.linear_fc2",
378
+ "visual.deepstack_merger_list.2.norm",
379
+ "model.visual.merger.linear_fc1",
380
+ "model.visual.merger.linear_fc2",
381
+ "model.visual.merger.norm",
382
+ "model.visual.patch_embed.proj",
383
+ "model.visual.pos_embed",
384
+ "visual.merger.linear_fc1",
385
+ "visual.merger.linear_fc2",
386
+ "visual.merger.norm",
387
+ "visual.patch_embed.proj",
388
+ "visual.pos_embed",
389
+ "visual",
390
+ "model.visual",
391
+ "lm_head",
392
+ "model.embed_tokens",
393
+ "model.language_model.layers.0.input_layernorm",
394
+ "model.language_model.layers.0.mlp.shared_expert_gate",
395
+ "model.language_model.layers.0.post_attention_layernorm",
396
+ "model.language_model.layers.0.mlp.gate",
397
+ "model.language_model.layers.0.linear_attn.A_log",
398
+ "model.language_model.layers.0.linear_attn.conv1d",
399
+ "model.language_model.layers.0.linear_attn.dt_bias",
400
+ "model.language_model.layers.0.linear_attn.in_proj_ba",
401
+ "model.language_model.layers.0.linear_attn.in_proj_b",
402
+ "model.language_model.layers.0.linear_attn.in_proj_a",
403
+ "model.language_model.layers.0.linear_attn.norm",
404
+ "model.language_model.layers.1.input_layernorm",
405
+ "model.language_model.layers.1.mlp.shared_expert_gate",
406
+ "model.language_model.layers.1.post_attention_layernorm",
407
+ "model.language_model.layers.1.mlp.gate",
408
+ "model.language_model.layers.1.linear_attn.A_log",
409
+ "model.language_model.layers.1.linear_attn.conv1d",
410
+ "model.language_model.layers.1.linear_attn.dt_bias",
411
+ "model.language_model.layers.1.linear_attn.in_proj_ba",
412
+ "model.language_model.layers.1.linear_attn.in_proj_b",
413
+ "model.language_model.layers.1.linear_attn.in_proj_a",
414
+ "model.language_model.layers.1.linear_attn.norm",
415
+ "model.language_model.layers.2.input_layernorm",
416
+ "model.language_model.layers.2.mlp.shared_expert_gate",
417
+ "model.language_model.layers.2.post_attention_layernorm",
418
+ "model.language_model.layers.2.mlp.gate",
419
+ "model.language_model.layers.2.linear_attn.A_log",
420
+ "model.language_model.layers.2.linear_attn.conv1d",
421
+ "model.language_model.layers.2.linear_attn.dt_bias",
422
+ "model.language_model.layers.2.linear_attn.in_proj_ba",
423
+ "model.language_model.layers.2.linear_attn.in_proj_b",
424
+ "model.language_model.layers.2.linear_attn.in_proj_a",
425
+ "model.language_model.layers.2.linear_attn.norm",
426
+ "model.language_model.layers.3.input_layernorm",
427
+ "model.language_model.layers.3.mlp.shared_expert_gate",
428
+ "model.language_model.layers.3.post_attention_layernorm",
429
+ "model.language_model.layers.3.mlp.gate",
430
+ "model.language_model.layers.3.self_attn.k_norm",
431
+ "model.language_model.layers.3.self_attn.q_norm",
432
+ "model.language_model.layers.4.input_layernorm",
433
+ "model.language_model.layers.4.mlp.shared_expert_gate",
434
+ "model.language_model.layers.4.post_attention_layernorm",
435
+ "model.language_model.layers.4.mlp.gate",
436
+ "model.language_model.layers.4.linear_attn.A_log",
437
+ "model.language_model.layers.4.linear_attn.conv1d",
438
+ "model.language_model.layers.4.linear_attn.dt_bias",
439
+ "model.language_model.layers.4.linear_attn.in_proj_ba",
440
+ "model.language_model.layers.4.linear_attn.in_proj_b",
441
+ "model.language_model.layers.4.linear_attn.in_proj_a",
442
+ "model.language_model.layers.4.linear_attn.norm",
443
+ "model.language_model.layers.5.input_layernorm",
444
+ "model.language_model.layers.5.mlp.shared_expert_gate",
445
+ "model.language_model.layers.5.post_attention_layernorm",
446
+ "model.language_model.layers.5.mlp.gate",
447
+ "model.language_model.layers.5.linear_attn.A_log",
448
+ "model.language_model.layers.5.linear_attn.conv1d",
449
+ "model.language_model.layers.5.linear_attn.dt_bias",
450
+ "model.language_model.layers.5.linear_attn.in_proj_ba",
451
+ "model.language_model.layers.5.linear_attn.in_proj_b",
452
+ "model.language_model.layers.5.linear_attn.in_proj_a",
453
+ "model.language_model.layers.5.linear_attn.norm",
454
+ "model.language_model.layers.6.input_layernorm",
455
+ "model.language_model.layers.6.mlp.shared_expert_gate",
456
+ "model.language_model.layers.6.post_attention_layernorm",
457
+ "model.language_model.layers.6.mlp.gate",
458
+ "model.language_model.layers.6.linear_attn.A_log",
459
+ "model.language_model.layers.6.linear_attn.conv1d",
460
+ "model.language_model.layers.6.linear_attn.dt_bias",
461
+ "model.language_model.layers.6.linear_attn.in_proj_ba",
462
+ "model.language_model.layers.6.linear_attn.in_proj_b",
463
+ "model.language_model.layers.6.linear_attn.in_proj_a",
464
+ "model.language_model.layers.6.linear_attn.norm",
465
+ "model.language_model.layers.7.input_layernorm",
466
+ "model.language_model.layers.7.mlp.shared_expert_gate",
467
+ "model.language_model.layers.7.post_attention_layernorm",
468
+ "model.language_model.layers.7.mlp.gate",
469
+ "model.language_model.layers.7.self_attn.k_norm",
470
+ "model.language_model.layers.7.self_attn.q_norm",
471
+ "model.language_model.layers.8.input_layernorm",
472
+ "model.language_model.layers.8.mlp.shared_expert_gate",
473
+ "model.language_model.layers.8.post_attention_layernorm",
474
+ "model.language_model.layers.8.mlp.gate",
475
+ "model.language_model.layers.8.linear_attn.A_log",
476
+ "model.language_model.layers.8.linear_attn.conv1d",
477
+ "model.language_model.layers.8.linear_attn.dt_bias",
478
+ "model.language_model.layers.8.linear_attn.in_proj_ba",
479
+ "model.language_model.layers.8.linear_attn.in_proj_b",
480
+ "model.language_model.layers.8.linear_attn.in_proj_a",
481
+ "model.language_model.layers.8.linear_attn.norm",
482
+ "model.language_model.layers.9.input_layernorm",
483
+ "model.language_model.layers.9.mlp.shared_expert_gate",
484
+ "model.language_model.layers.9.post_attention_layernorm",
485
+ "model.language_model.layers.9.mlp.gate",
486
+ "model.language_model.layers.9.linear_attn.A_log",
487
+ "model.language_model.layers.9.linear_attn.conv1d",
488
+ "model.language_model.layers.9.linear_attn.dt_bias",
489
+ "model.language_model.layers.9.linear_attn.in_proj_ba",
490
+ "model.language_model.layers.9.linear_attn.in_proj_b",
491
+ "model.language_model.layers.9.linear_attn.in_proj_a",
492
+ "model.language_model.layers.9.linear_attn.norm",
493
+ "model.language_model.layers.10.input_layernorm",
494
+ "model.language_model.layers.10.mlp.shared_expert_gate",
495
+ "model.language_model.layers.10.post_attention_layernorm",
496
+ "model.language_model.layers.10.mlp.gate",
497
+ "model.language_model.layers.10.linear_attn.A_log",
498
+ "model.language_model.layers.10.linear_attn.conv1d",
499
+ "model.language_model.layers.10.linear_attn.dt_bias",
500
+ "model.language_model.layers.10.linear_attn.in_proj_ba",
501
+ "model.language_model.layers.10.linear_attn.in_proj_b",
502
+ "model.language_model.layers.10.linear_attn.in_proj_a",
503
+ "model.language_model.layers.10.linear_attn.norm",
504
+ "model.language_model.layers.11.input_layernorm",
505
+ "model.language_model.layers.11.mlp.shared_expert_gate",
506
+ "model.language_model.layers.11.post_attention_layernorm",
507
+ "model.language_model.layers.11.mlp.gate",
508
+ "model.language_model.layers.11.self_attn.k_norm",
509
+ "model.language_model.layers.11.self_attn.q_norm",
510
+ "model.language_model.layers.12.input_layernorm",
511
+ "model.language_model.layers.12.mlp.shared_expert_gate",
512
+ "model.language_model.layers.12.post_attention_layernorm",
513
+ "model.language_model.layers.12.mlp.gate",
514
+ "model.language_model.layers.12.linear_attn.A_log",
515
+ "model.language_model.layers.12.linear_attn.conv1d",
516
+ "model.language_model.layers.12.linear_attn.dt_bias",
517
+ "model.language_model.layers.12.linear_attn.in_proj_ba",
518
+ "model.language_model.layers.12.linear_attn.in_proj_b",
519
+ "model.language_model.layers.12.linear_attn.in_proj_a",
520
+ "model.language_model.layers.12.linear_attn.norm",
521
+ "model.language_model.layers.13.input_layernorm",
522
+ "model.language_model.layers.13.mlp.shared_expert_gate",
523
+ "model.language_model.layers.13.post_attention_layernorm",
524
+ "model.language_model.layers.13.mlp.gate",
525
+ "model.language_model.layers.13.linear_attn.A_log",
526
+ "model.language_model.layers.13.linear_attn.conv1d",
527
+ "model.language_model.layers.13.linear_attn.dt_bias",
528
+ "model.language_model.layers.13.linear_attn.in_proj_ba",
529
+ "model.language_model.layers.13.linear_attn.in_proj_b",
530
+ "model.language_model.layers.13.linear_attn.in_proj_a",
531
+ "model.language_model.layers.13.linear_attn.norm",
532
+ "model.language_model.layers.14.input_layernorm",
533
+ "model.language_model.layers.14.mlp.shared_expert_gate",
534
+ "model.language_model.layers.14.post_attention_layernorm",
535
+ "model.language_model.layers.14.mlp.gate",
536
+ "model.language_model.layers.14.linear_attn.A_log",
537
+ "model.language_model.layers.14.linear_attn.conv1d",
538
+ "model.language_model.layers.14.linear_attn.dt_bias",
539
+ "model.language_model.layers.14.linear_attn.in_proj_ba",
540
+ "model.language_model.layers.14.linear_attn.in_proj_b",
541
+ "model.language_model.layers.14.linear_attn.in_proj_a",
542
+ "model.language_model.layers.14.linear_attn.norm",
543
+ "model.language_model.layers.15.input_layernorm",
544
+ "model.language_model.layers.15.mlp.shared_expert_gate",
545
+ "model.language_model.layers.15.post_attention_layernorm",
546
+ "model.language_model.layers.15.mlp.gate",
547
+ "model.language_model.layers.15.self_attn.k_norm",
548
+ "model.language_model.layers.15.self_attn.q_norm",
549
+ "model.language_model.layers.16.input_layernorm",
550
+ "model.language_model.layers.16.mlp.shared_expert_gate",
551
+ "model.language_model.layers.16.post_attention_layernorm",
552
+ "model.language_model.layers.16.mlp.gate",
553
+ "model.language_model.layers.16.linear_attn.A_log",
554
+ "model.language_model.layers.16.linear_attn.conv1d",
555
+ "model.language_model.layers.16.linear_attn.dt_bias",
556
+ "model.language_model.layers.16.linear_attn.in_proj_ba",
557
+ "model.language_model.layers.16.linear_attn.in_proj_b",
558
+ "model.language_model.layers.16.linear_attn.in_proj_a",
559
+ "model.language_model.layers.16.linear_attn.norm",
560
+ "model.language_model.layers.17.input_layernorm",
561
+ "model.language_model.layers.17.mlp.shared_expert_gate",
562
+ "model.language_model.layers.17.post_attention_layernorm",
563
+ "model.language_model.layers.17.mlp.gate",
564
+ "model.language_model.layers.17.linear_attn.A_log",
565
+ "model.language_model.layers.17.linear_attn.conv1d",
566
+ "model.language_model.layers.17.linear_attn.dt_bias",
567
+ "model.language_model.layers.17.linear_attn.in_proj_ba",
568
+ "model.language_model.layers.17.linear_attn.in_proj_b",
569
+ "model.language_model.layers.17.linear_attn.in_proj_a",
570
+ "model.language_model.layers.17.linear_attn.norm",
571
+ "model.language_model.layers.18.input_layernorm",
572
+ "model.language_model.layers.18.mlp.shared_expert_gate",
573
+ "model.language_model.layers.18.post_attention_layernorm",
574
+ "model.language_model.layers.18.mlp.gate",
575
+ "model.language_model.layers.18.linear_attn.A_log",
576
+ "model.language_model.layers.18.linear_attn.conv1d",
577
+ "model.language_model.layers.18.linear_attn.dt_bias",
578
+ "model.language_model.layers.18.linear_attn.in_proj_ba",
579
+ "model.language_model.layers.18.linear_attn.in_proj_b",
580
+ "model.language_model.layers.18.linear_attn.in_proj_a",
581
+ "model.language_model.layers.18.linear_attn.norm",
582
+ "model.language_model.layers.19.input_layernorm",
583
+ "model.language_model.layers.19.mlp.shared_expert_gate",
584
+ "model.language_model.layers.19.post_attention_layernorm",
585
+ "model.language_model.layers.19.mlp.gate",
586
+ "model.language_model.layers.19.self_attn.k_norm",
587
+ "model.language_model.layers.19.self_attn.q_norm",
588
+ "model.language_model.layers.20.input_layernorm",
589
+ "model.language_model.layers.20.mlp.shared_expert_gate",
590
+ "model.language_model.layers.20.post_attention_layernorm",
591
+ "model.language_model.layers.20.mlp.gate",
592
+ "model.language_model.layers.20.linear_attn.A_log",
593
+ "model.language_model.layers.20.linear_attn.conv1d",
594
+ "model.language_model.layers.20.linear_attn.dt_bias",
595
+ "model.language_model.layers.20.linear_attn.in_proj_ba",
596
+ "model.language_model.layers.20.linear_attn.in_proj_b",
597
+ "model.language_model.layers.20.linear_attn.in_proj_a",
598
+ "model.language_model.layers.20.linear_attn.norm",
599
+ "model.language_model.layers.21.input_layernorm",
600
+ "model.language_model.layers.21.mlp.shared_expert_gate",
601
+ "model.language_model.layers.21.post_attention_layernorm",
602
+ "model.language_model.layers.21.mlp.gate",
603
+ "model.language_model.layers.21.linear_attn.A_log",
604
+ "model.language_model.layers.21.linear_attn.conv1d",
605
+ "model.language_model.layers.21.linear_attn.dt_bias",
606
+ "model.language_model.layers.21.linear_attn.in_proj_ba",
607
+ "model.language_model.layers.21.linear_attn.in_proj_b",
608
+ "model.language_model.layers.21.linear_attn.in_proj_a",
609
+ "model.language_model.layers.21.linear_attn.norm",
610
+ "model.language_model.layers.22.input_layernorm",
611
+ "model.language_model.layers.22.mlp.shared_expert_gate",
612
+ "model.language_model.layers.22.post_attention_layernorm",
613
+ "model.language_model.layers.22.mlp.gate",
614
+ "model.language_model.layers.22.linear_attn.A_log",
615
+ "model.language_model.layers.22.linear_attn.conv1d",
616
+ "model.language_model.layers.22.linear_attn.dt_bias",
617
+ "model.language_model.layers.22.linear_attn.in_proj_ba",
618
+ "model.language_model.layers.22.linear_attn.in_proj_b",
619
+ "model.language_model.layers.22.linear_attn.in_proj_a",
620
+ "model.language_model.layers.22.linear_attn.norm",
621
+ "model.language_model.layers.23.input_layernorm",
622
+ "model.language_model.layers.23.mlp.shared_expert_gate",
623
+ "model.language_model.layers.23.post_attention_layernorm",
624
+ "model.language_model.layers.23.mlp.gate",
625
+ "model.language_model.layers.23.self_attn.k_norm",
626
+ "model.language_model.layers.23.self_attn.q_norm",
627
+ "model.language_model.layers.24.input_layernorm",
628
+ "model.language_model.layers.24.mlp.shared_expert_gate",
629
+ "model.language_model.layers.24.post_attention_layernorm",
630
+ "model.language_model.layers.24.mlp.gate",
631
+ "model.language_model.layers.24.linear_attn.A_log",
632
+ "model.language_model.layers.24.linear_attn.conv1d",
633
+ "model.language_model.layers.24.linear_attn.dt_bias",
634
+ "model.language_model.layers.24.linear_attn.in_proj_ba",
635
+ "model.language_model.layers.24.linear_attn.in_proj_b",
636
+ "model.language_model.layers.24.linear_attn.in_proj_a",
637
+ "model.language_model.layers.24.linear_attn.norm",
638
+ "model.language_model.layers.25.input_layernorm",
639
+ "model.language_model.layers.25.mlp.shared_expert_gate",
640
+ "model.language_model.layers.25.post_attention_layernorm",
641
+ "model.language_model.layers.25.mlp.gate",
642
+ "model.language_model.layers.25.linear_attn.A_log",
643
+ "model.language_model.layers.25.linear_attn.conv1d",
644
+ "model.language_model.layers.25.linear_attn.dt_bias",
645
+ "model.language_model.layers.25.linear_attn.in_proj_ba",
646
+ "model.language_model.layers.25.linear_attn.in_proj_b",
647
+ "model.language_model.layers.25.linear_attn.in_proj_a",
648
+ "model.language_model.layers.25.linear_attn.norm",
649
+ "model.language_model.layers.26.input_layernorm",
650
+ "model.language_model.layers.26.mlp.shared_expert_gate",
651
+ "model.language_model.layers.26.post_attention_layernorm",
652
+ "model.language_model.layers.26.mlp.gate",
653
+ "model.language_model.layers.26.linear_attn.A_log",
654
+ "model.language_model.layers.26.linear_attn.conv1d",
655
+ "model.language_model.layers.26.linear_attn.dt_bias",
656
+ "model.language_model.layers.26.linear_attn.in_proj_ba",
657
+ "model.language_model.layers.26.linear_attn.in_proj_b",
658
+ "model.language_model.layers.26.linear_attn.in_proj_a",
659
+ "model.language_model.layers.26.linear_attn.norm",
660
+ "model.language_model.layers.27.input_layernorm",
661
+ "model.language_model.layers.27.mlp.shared_expert_gate",
662
+ "model.language_model.layers.27.post_attention_layernorm",
663
+ "model.language_model.layers.27.mlp.gate",
664
+ "model.language_model.layers.27.self_attn.k_norm",
665
+ "model.language_model.layers.27.self_attn.q_norm",
666
+ "model.language_model.layers.28.input_layernorm",
667
+ "model.language_model.layers.28.mlp.shared_expert_gate",
668
+ "model.language_model.layers.28.post_attention_layernorm",
669
+ "model.language_model.layers.28.mlp.gate",
670
+ "model.language_model.layers.28.linear_attn.A_log",
671
+ "model.language_model.layers.28.linear_attn.conv1d",
672
+ "model.language_model.layers.28.linear_attn.dt_bias",
673
+ "model.language_model.layers.28.linear_attn.in_proj_ba",
674
+ "model.language_model.layers.28.linear_attn.in_proj_b",
675
+ "model.language_model.layers.28.linear_attn.in_proj_a",
676
+ "model.language_model.layers.28.linear_attn.norm",
677
+ "model.language_model.layers.29.input_layernorm",
678
+ "model.language_model.layers.29.mlp.shared_expert_gate",
679
+ "model.language_model.layers.29.post_attention_layernorm",
680
+ "model.language_model.layers.29.mlp.gate",
681
+ "model.language_model.layers.29.linear_attn.A_log",
682
+ "model.language_model.layers.29.linear_attn.conv1d",
683
+ "model.language_model.layers.29.linear_attn.dt_bias",
684
+ "model.language_model.layers.29.linear_attn.in_proj_ba",
685
+ "model.language_model.layers.29.linear_attn.in_proj_b",
686
+ "model.language_model.layers.29.linear_attn.in_proj_a",
687
+ "model.language_model.layers.29.linear_attn.norm",
688
+ "model.language_model.layers.30.input_layernorm",
689
+ "model.language_model.layers.30.mlp.shared_expert_gate",
690
+ "model.language_model.layers.30.post_attention_layernorm",
691
+ "model.language_model.layers.30.mlp.gate",
692
+ "model.language_model.layers.30.linear_attn.A_log",
693
+ "model.language_model.layers.30.linear_attn.conv1d",
694
+ "model.language_model.layers.30.linear_attn.dt_bias",
695
+ "model.language_model.layers.30.linear_attn.in_proj_ba",
696
+ "model.language_model.layers.30.linear_attn.in_proj_b",
697
+ "model.language_model.layers.30.linear_attn.in_proj_a",
698
+ "model.language_model.layers.30.linear_attn.norm",
699
+ "model.language_model.layers.31.input_layernorm",
700
+ "model.language_model.layers.31.mlp.shared_expert_gate",
701
+ "model.language_model.layers.31.post_attention_layernorm",
702
+ "model.language_model.layers.31.mlp.gate",
703
+ "model.language_model.layers.31.self_attn.k_norm",
704
+ "model.language_model.layers.31.self_attn.q_norm",
705
+ "model.language_model.layers.32.input_layernorm",
706
+ "model.language_model.layers.32.mlp.shared_expert_gate",
707
+ "model.language_model.layers.32.post_attention_layernorm",
708
+ "model.language_model.layers.32.mlp.gate",
709
+ "model.language_model.layers.32.linear_attn.A_log",
710
+ "model.language_model.layers.32.linear_attn.conv1d",
711
+ "model.language_model.layers.32.linear_attn.dt_bias",
712
+ "model.language_model.layers.32.linear_attn.in_proj_ba",
713
+ "model.language_model.layers.32.linear_attn.in_proj_b",
714
+ "model.language_model.layers.32.linear_attn.in_proj_a",
715
+ "model.language_model.layers.32.linear_attn.norm",
716
+ "model.language_model.layers.33.input_layernorm",
717
+ "model.language_model.layers.33.mlp.shared_expert_gate",
718
+ "model.language_model.layers.33.post_attention_layernorm",
719
+ "model.language_model.layers.33.mlp.gate",
720
+ "model.language_model.layers.33.linear_attn.A_log",
721
+ "model.language_model.layers.33.linear_attn.conv1d",
722
+ "model.language_model.layers.33.linear_attn.dt_bias",
723
+ "model.language_model.layers.33.linear_attn.in_proj_ba",
724
+ "model.language_model.layers.33.linear_attn.in_proj_b",
725
+ "model.language_model.layers.33.linear_attn.in_proj_a",
726
+ "model.language_model.layers.33.linear_attn.norm",
727
+ "model.language_model.layers.34.input_layernorm",
728
+ "model.language_model.layers.34.mlp.shared_expert_gate",
729
+ "model.language_model.layers.34.post_attention_layernorm",
730
+ "model.language_model.layers.34.mlp.gate",
731
+ "model.language_model.layers.34.linear_attn.A_log",
732
+ "model.language_model.layers.34.linear_attn.conv1d",
733
+ "model.language_model.layers.34.linear_attn.dt_bias",
734
+ "model.language_model.layers.34.linear_attn.in_proj_ba",
735
+ "model.language_model.layers.34.linear_attn.in_proj_b",
736
+ "model.language_model.layers.34.linear_attn.in_proj_a",
737
+ "model.language_model.layers.34.linear_attn.norm",
738
+ "model.language_model.layers.35.input_layernorm",
739
+ "model.language_model.layers.35.mlp.shared_expert_gate",
740
+ "model.language_model.layers.35.post_attention_layernorm",
741
+ "model.language_model.layers.35.mlp.gate",
742
+ "model.language_model.layers.35.self_attn.k_norm",
743
+ "model.language_model.layers.35.self_attn.q_norm",
744
+ "model.language_model.layers.36.input_layernorm",
745
+ "model.language_model.layers.36.mlp.shared_expert_gate",
746
+ "model.language_model.layers.36.post_attention_layernorm",
747
+ "model.language_model.layers.36.mlp.gate",
748
+ "model.language_model.layers.36.linear_attn.A_log",
749
+ "model.language_model.layers.36.linear_attn.conv1d",
750
+ "model.language_model.layers.36.linear_attn.dt_bias",
751
+ "model.language_model.layers.36.linear_attn.in_proj_ba",
752
+ "model.language_model.layers.36.linear_attn.in_proj_b",
753
+ "model.language_model.layers.36.linear_attn.in_proj_a",
754
+ "model.language_model.layers.36.linear_attn.norm",
755
+ "model.language_model.layers.37.input_layernorm",
756
+ "model.language_model.layers.37.mlp.shared_expert_gate",
757
+ "model.language_model.layers.37.post_attention_layernorm",
758
+ "model.language_model.layers.37.mlp.gate",
759
+ "model.language_model.layers.37.linear_attn.A_log",
760
+ "model.language_model.layers.37.linear_attn.conv1d",
761
+ "model.language_model.layers.37.linear_attn.dt_bias",
762
+ "model.language_model.layers.37.linear_attn.in_proj_ba",
763
+ "model.language_model.layers.37.linear_attn.in_proj_b",
764
+ "model.language_model.layers.37.linear_attn.in_proj_a",
765
+ "model.language_model.layers.37.linear_attn.norm",
766
+ "model.language_model.layers.38.input_layernorm",
767
+ "model.language_model.layers.38.mlp.shared_expert_gate",
768
+ "model.language_model.layers.38.post_attention_layernorm",
769
+ "model.language_model.layers.38.mlp.gate",
770
+ "model.language_model.layers.38.linear_attn.A_log",
771
+ "model.language_model.layers.38.linear_attn.conv1d",
772
+ "model.language_model.layers.38.linear_attn.dt_bias",
773
+ "model.language_model.layers.38.linear_attn.in_proj_ba",
774
+ "model.language_model.layers.38.linear_attn.in_proj_b",
775
+ "model.language_model.layers.38.linear_attn.in_proj_a",
776
+ "model.language_model.layers.38.linear_attn.norm",
777
+ "model.language_model.layers.39.input_layernorm",
778
+ "model.language_model.layers.39.mlp.shared_expert_gate",
779
+ "model.language_model.layers.39.post_attention_layernorm",
780
+ "model.language_model.layers.39.mlp.gate",
781
+ "model.language_model.layers.39.self_attn.k_norm",
782
+ "model.language_model.layers.39.self_attn.q_norm",
783
+ "model.language_model.layers.40.input_layernorm",
784
+ "model.language_model.layers.40.mlp.shared_expert_gate",
785
+ "model.language_model.layers.40.post_attention_layernorm",
786
+ "model.language_model.layers.40.mlp.gate",
787
+ "model.language_model.layers.40.linear_attn.A_log",
788
+ "model.language_model.layers.40.linear_attn.conv1d",
789
+ "model.language_model.layers.40.linear_attn.dt_bias",
790
+ "model.language_model.layers.40.linear_attn.in_proj_ba",
791
+ "model.language_model.layers.40.linear_attn.in_proj_b",
792
+ "model.language_model.layers.40.linear_attn.in_proj_a",
793
+ "model.language_model.layers.40.linear_attn.norm",
794
+ "model.language_model.layers.41.input_layernorm",
795
+ "model.language_model.layers.41.mlp.shared_expert_gate",
796
+ "model.language_model.layers.41.post_attention_layernorm",
797
+ "model.language_model.layers.41.mlp.gate",
798
+ "model.language_model.layers.41.linear_attn.A_log",
799
+ "model.language_model.layers.41.linear_attn.conv1d",
800
+ "model.language_model.layers.41.linear_attn.dt_bias",
801
+ "model.language_model.layers.41.linear_attn.in_proj_ba",
802
+ "model.language_model.layers.41.linear_attn.in_proj_b",
803
+ "model.language_model.layers.41.linear_attn.in_proj_a",
804
+ "model.language_model.layers.41.linear_attn.norm",
805
+ "model.language_model.layers.42.input_layernorm",
806
+ "model.language_model.layers.42.mlp.shared_expert_gate",
807
+ "model.language_model.layers.42.post_attention_layernorm",
808
+ "model.language_model.layers.42.mlp.gate",
809
+ "model.language_model.layers.42.linear_attn.A_log",
810
+ "model.language_model.layers.42.linear_attn.conv1d",
811
+ "model.language_model.layers.42.linear_attn.dt_bias",
812
+ "model.language_model.layers.42.linear_attn.in_proj_ba",
813
+ "model.language_model.layers.42.linear_attn.in_proj_b",
814
+ "model.language_model.layers.42.linear_attn.in_proj_a",
815
+ "model.language_model.layers.42.linear_attn.norm",
816
+ "model.language_model.layers.43.input_layernorm",
817
+ "model.language_model.layers.43.mlp.shared_expert_gate",
818
+ "model.language_model.layers.43.post_attention_layernorm",
819
+ "model.language_model.layers.43.mlp.gate",
820
+ "model.language_model.layers.43.self_attn.k_norm",
821
+ "model.language_model.layers.43.self_attn.q_norm",
822
+ "model.language_model.layers.44.input_layernorm",
823
+ "model.language_model.layers.44.mlp.shared_expert_gate",
824
+ "model.language_model.layers.44.post_attention_layernorm",
825
+ "model.language_model.layers.44.mlp.gate",
826
+ "model.language_model.layers.44.linear_attn.A_log",
827
+ "model.language_model.layers.44.linear_attn.conv1d",
828
+ "model.language_model.layers.44.linear_attn.dt_bias",
829
+ "model.language_model.layers.44.linear_attn.in_proj_ba",
830
+ "model.language_model.layers.44.linear_attn.in_proj_b",
831
+ "model.language_model.layers.44.linear_attn.in_proj_a",
832
+ "model.language_model.layers.44.linear_attn.norm",
833
+ "model.language_model.layers.45.input_layernorm",
834
+ "model.language_model.layers.45.mlp.shared_expert_gate",
835
+ "model.language_model.layers.45.post_attention_layernorm",
836
+ "model.language_model.layers.45.mlp.gate",
837
+ "model.language_model.layers.45.linear_attn.A_log",
838
+ "model.language_model.layers.45.linear_attn.conv1d",
839
+ "model.language_model.layers.45.linear_attn.dt_bias",
840
+ "model.language_model.layers.45.linear_attn.in_proj_ba",
841
+ "model.language_model.layers.45.linear_attn.in_proj_b",
842
+ "model.language_model.layers.45.linear_attn.in_proj_a",
843
+ "model.language_model.layers.45.linear_attn.norm",
844
+ "model.language_model.layers.46.input_layernorm",
845
+ "model.language_model.layers.46.mlp.shared_expert_gate",
846
+ "model.language_model.layers.46.post_attention_layernorm",
847
+ "model.language_model.layers.46.mlp.gate",
848
+ "model.language_model.layers.46.linear_attn.A_log",
849
+ "model.language_model.layers.46.linear_attn.conv1d",
850
+ "model.language_model.layers.46.linear_attn.dt_bias",
851
+ "model.language_model.layers.46.linear_attn.in_proj_ba",
852
+ "model.language_model.layers.46.linear_attn.in_proj_b",
853
+ "model.language_model.layers.46.linear_attn.in_proj_a",
854
+ "model.language_model.layers.46.linear_attn.norm",
855
+ "model.language_model.layers.47.input_layernorm",
856
+ "model.language_model.layers.47.mlp.shared_expert_gate",
857
+ "model.language_model.layers.47.post_attention_layernorm",
858
+ "model.language_model.layers.47.mlp.gate",
859
+ "model.language_model.layers.47.self_attn.k_norm",
860
+ "model.language_model.layers.47.self_attn.q_norm",
861
+ "model.language_model.layers.48.input_layernorm",
862
+ "model.language_model.layers.48.mlp.shared_expert_gate",
863
+ "model.language_model.layers.48.post_attention_layernorm",
864
+ "model.language_model.layers.48.mlp.gate",
865
+ "model.language_model.layers.48.linear_attn.A_log",
866
+ "model.language_model.layers.48.linear_attn.conv1d",
867
+ "model.language_model.layers.48.linear_attn.dt_bias",
868
+ "model.language_model.layers.48.linear_attn.in_proj_ba",
869
+ "model.language_model.layers.48.linear_attn.in_proj_b",
870
+ "model.language_model.layers.48.linear_attn.in_proj_a",
871
+ "model.language_model.layers.48.linear_attn.norm",
872
+ "model.language_model.layers.49.input_layernorm",
873
+ "model.language_model.layers.49.mlp.shared_expert_gate",
874
+ "model.language_model.layers.49.post_attention_layernorm",
875
+ "model.language_model.layers.49.mlp.gate",
876
+ "model.language_model.layers.49.linear_attn.A_log",
877
+ "model.language_model.layers.49.linear_attn.conv1d",
878
+ "model.language_model.layers.49.linear_attn.dt_bias",
879
+ "model.language_model.layers.49.linear_attn.in_proj_ba",
880
+ "model.language_model.layers.49.linear_attn.in_proj_b",
881
+ "model.language_model.layers.49.linear_attn.in_proj_a",
882
+ "model.language_model.layers.49.linear_attn.norm",
883
+ "model.language_model.layers.50.input_layernorm",
884
+ "model.language_model.layers.50.mlp.shared_expert_gate",
885
+ "model.language_model.layers.50.post_attention_layernorm",
886
+ "model.language_model.layers.50.mlp.gate",
887
+ "model.language_model.layers.50.linear_attn.A_log",
888
+ "model.language_model.layers.50.linear_attn.conv1d",
889
+ "model.language_model.layers.50.linear_attn.dt_bias",
890
+ "model.language_model.layers.50.linear_attn.in_proj_ba",
891
+ "model.language_model.layers.50.linear_attn.in_proj_b",
892
+ "model.language_model.layers.50.linear_attn.in_proj_a",
893
+ "model.language_model.layers.50.linear_attn.norm",
894
+ "model.language_model.layers.51.input_layernorm",
895
+ "model.language_model.layers.51.mlp.shared_expert_gate",
896
+ "model.language_model.layers.51.post_attention_layernorm",
897
+ "model.language_model.layers.51.mlp.gate",
898
+ "model.language_model.layers.51.self_attn.k_norm",
899
+ "model.language_model.layers.51.self_attn.q_norm",
900
+ "model.language_model.layers.52.input_layernorm",
901
+ "model.language_model.layers.52.mlp.shared_expert_gate",
902
+ "model.language_model.layers.52.post_attention_layernorm",
903
+ "model.language_model.layers.52.mlp.gate",
904
+ "model.language_model.layers.52.linear_attn.A_log",
905
+ "model.language_model.layers.52.linear_attn.conv1d",
906
+ "model.language_model.layers.52.linear_attn.dt_bias",
907
+ "model.language_model.layers.52.linear_attn.in_proj_ba",
908
+ "model.language_model.layers.52.linear_attn.in_proj_b",
909
+ "model.language_model.layers.52.linear_attn.in_proj_a",
910
+ "model.language_model.layers.52.linear_attn.norm",
911
+ "model.language_model.layers.53.input_layernorm",
912
+ "model.language_model.layers.53.mlp.shared_expert_gate",
913
+ "model.language_model.layers.53.post_attention_layernorm",
914
+ "model.language_model.layers.53.mlp.gate",
915
+ "model.language_model.layers.53.linear_attn.A_log",
916
+ "model.language_model.layers.53.linear_attn.conv1d",
917
+ "model.language_model.layers.53.linear_attn.dt_bias",
918
+ "model.language_model.layers.53.linear_attn.in_proj_ba",
919
+ "model.language_model.layers.53.linear_attn.in_proj_b",
920
+ "model.language_model.layers.53.linear_attn.in_proj_a",
921
+ "model.language_model.layers.53.linear_attn.norm",
922
+ "model.language_model.layers.54.input_layernorm",
923
+ "model.language_model.layers.54.mlp.shared_expert_gate",
924
+ "model.language_model.layers.54.post_attention_layernorm",
925
+ "model.language_model.layers.54.mlp.gate",
926
+ "model.language_model.layers.54.linear_attn.A_log",
927
+ "model.language_model.layers.54.linear_attn.conv1d",
928
+ "model.language_model.layers.54.linear_attn.dt_bias",
929
+ "model.language_model.layers.54.linear_attn.in_proj_ba",
930
+ "model.language_model.layers.54.linear_attn.in_proj_b",
931
+ "model.language_model.layers.54.linear_attn.in_proj_a",
932
+ "model.language_model.layers.54.linear_attn.norm",
933
+ "model.language_model.layers.55.input_layernorm",
934
+ "model.language_model.layers.55.mlp.shared_expert_gate",
935
+ "model.language_model.layers.55.post_attention_layernorm",
936
+ "model.language_model.layers.55.mlp.gate",
937
+ "model.language_model.layers.55.self_attn.k_norm",
938
+ "model.language_model.layers.55.self_attn.q_norm",
939
+ "model.language_model.layers.56.input_layernorm",
940
+ "model.language_model.layers.56.mlp.shared_expert_gate",
941
+ "model.language_model.layers.56.post_attention_layernorm",
942
+ "model.language_model.layers.56.mlp.gate",
943
+ "model.language_model.layers.56.linear_attn.A_log",
944
+ "model.language_model.layers.56.linear_attn.conv1d",
945
+ "model.language_model.layers.56.linear_attn.dt_bias",
946
+ "model.language_model.layers.56.linear_attn.in_proj_ba",
947
+ "model.language_model.layers.56.linear_attn.in_proj_b",
948
+ "model.language_model.layers.56.linear_attn.in_proj_a",
949
+ "model.language_model.layers.56.linear_attn.norm",
950
+ "model.language_model.layers.57.input_layernorm",
951
+ "model.language_model.layers.57.mlp.shared_expert_gate",
952
+ "model.language_model.layers.57.post_attention_layernorm",
953
+ "model.language_model.layers.57.mlp.gate",
954
+ "model.language_model.layers.57.linear_attn.A_log",
955
+ "model.language_model.layers.57.linear_attn.conv1d",
956
+ "model.language_model.layers.57.linear_attn.dt_bias",
957
+ "model.language_model.layers.57.linear_attn.in_proj_ba",
958
+ "model.language_model.layers.57.linear_attn.in_proj_b",
959
+ "model.language_model.layers.57.linear_attn.in_proj_a",
960
+ "model.language_model.layers.57.linear_attn.norm",
961
+ "model.language_model.layers.58.input_layernorm",
962
+ "model.language_model.layers.58.mlp.shared_expert_gate",
963
+ "model.language_model.layers.58.post_attention_layernorm",
964
+ "model.language_model.layers.58.mlp.gate",
965
+ "model.language_model.layers.58.linear_attn.A_log",
966
+ "model.language_model.layers.58.linear_attn.conv1d",
967
+ "model.language_model.layers.58.linear_attn.dt_bias",
968
+ "model.language_model.layers.58.linear_attn.in_proj_ba",
969
+ "model.language_model.layers.58.linear_attn.in_proj_b",
970
+ "model.language_model.layers.58.linear_attn.in_proj_a",
971
+ "model.language_model.layers.58.linear_attn.norm",
972
+ "model.language_model.layers.59.input_layernorm",
973
+ "model.language_model.layers.59.mlp.shared_expert_gate",
974
+ "model.language_model.layers.59.post_attention_layernorm",
975
+ "model.language_model.layers.59.mlp.gate",
976
+ "model.language_model.layers.59.self_attn.k_norm",
977
+ "model.language_model.layers.59.self_attn.q_norm",
978
+ "model.language_model.layers.60.input_layernorm",
979
+ "model.language_model.layers.60.mlp.shared_expert_gate",
980
+ "model.language_model.layers.60.post_attention_layernorm",
981
+ "model.language_model.layers.60.mlp.gate",
982
+ "model.language_model.layers.60.linear_attn.A_log",
983
+ "model.language_model.layers.60.linear_attn.conv1d",
984
+ "model.language_model.layers.60.linear_attn.dt_bias",
985
+ "model.language_model.layers.60.linear_attn.in_proj_ba",
986
+ "model.language_model.layers.60.linear_attn.in_proj_b",
987
+ "model.language_model.layers.60.linear_attn.in_proj_a",
988
+ "model.language_model.layers.60.linear_attn.norm",
989
+ "model.language_model.layers.61.input_layernorm",
990
+ "model.language_model.layers.61.mlp.shared_expert_gate",
991
+ "model.language_model.layers.61.post_attention_layernorm",
992
+ "model.language_model.layers.61.mlp.gate",
993
+ "model.language_model.layers.61.linear_attn.A_log",
994
+ "model.language_model.layers.61.linear_attn.conv1d",
995
+ "model.language_model.layers.61.linear_attn.dt_bias",
996
+ "model.language_model.layers.61.linear_attn.in_proj_ba",
997
+ "model.language_model.layers.61.linear_attn.in_proj_b",
998
+ "model.language_model.layers.61.linear_attn.in_proj_a",
999
+ "model.language_model.layers.61.linear_attn.norm",
1000
+ "model.language_model.layers.62.input_layernorm",
1001
+ "model.language_model.layers.62.mlp.shared_expert_gate",
1002
+ "model.language_model.layers.62.post_attention_layernorm",
1003
+ "model.language_model.layers.62.mlp.gate",
1004
+ "model.language_model.layers.62.linear_attn.A_log",
1005
+ "model.language_model.layers.62.linear_attn.conv1d",
1006
+ "model.language_model.layers.62.linear_attn.dt_bias",
1007
+ "model.language_model.layers.62.linear_attn.in_proj_ba",
1008
+ "model.language_model.layers.62.linear_attn.in_proj_b",
1009
+ "model.language_model.layers.62.linear_attn.in_proj_a",
1010
+ "model.language_model.layers.62.linear_attn.norm",
1011
+ "model.language_model.layers.63.input_layernorm",
1012
+ "model.language_model.layers.63.mlp.shared_expert_gate",
1013
+ "model.language_model.layers.63.post_attention_layernorm",
1014
+ "model.language_model.layers.63.mlp.gate",
1015
+ "model.language_model.layers.63.self_attn.k_norm",
1016
+ "model.language_model.layers.63.self_attn.q_norm",
1017
+ "mtp.layers.0.input_layernorm",
1018
+ "mtp.layers.0.mlp.gate",
1019
+ "mtp.layers.0.mlp.shared_expert_gate",
1020
+ "mtp.layers.0.post_attention_layernorm",
1021
+ "mtp.layers.0.self_attn.k_norm",
1022
+ "mtp.layers.0.self_attn.q_norm",
1023
+ "mtp.fc",
1024
+ "mtp.norm",
1025
+ "mtp.pre_fc_norm_embedding",
1026
+ "mtp.pre_fc_norm_hidden"
1027
+ ],
1028
+ "weight_block_size": [
1029
+ 128,
1030
+ 128
1031
+ ]
1032
+ }
1033
+ }
generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 248044,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 248046,
6
+ 248044
7
+ ],
8
+ "pad_token_id": 248044,
9
+ "temperature": 1.0,
10
+ "top_k": 20,
11
+ "top_p": 0.95
12
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model-00001-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3eeeab65368b86c1eec3582495f5367c9a6a6ac25f0857adbe7f8026ea322d3f
3
+ size 3969317448
model-00002-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7091dbd7077b8d19bb547eaf24997ec3c2b6b0acef8f2678f1a7a6a15c212a3e
3
+ size 4939321704
model-00003-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ec75e6485b732125098c0080669ddbcef007be82bd6830a07e566b9a9ee3264
3
+ size 4939863424
model-00004-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44fef978664d99a87d6c4e5826e51486eb4e28788ca791b694b29430432a12fe
3
+ size 4971328696
model-00005-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f815eef9e6ed5194fdc0ac7412ad29019661ca6062bcd454fb4ce341834e29a4
3
+ size 4955596064
model-00006-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d2de7016e251a68f0a7ba2c49615c243c20ad87dfaacca252d6b2c0277f5fbbb
3
+ size 4071440632
model-00007-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1bed352a0625d22dc606de47fd3563f24fd9fa73915c4ddffe7947c4b56c28bf
3
+ size 3019999168
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3
3
+ size 12809320
tokenizer_config.json ADDED
@@ -0,0 +1,305 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "248044": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "248045": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "248046": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "248047": {
29
+ "content": "<|object_ref_start|>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "248048": {
37
+ "content": "<|object_ref_end|>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "248049": {
45
+ "content": "<|box_start|>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "248050": {
53
+ "content": "<|box_end|>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "248051": {
61
+ "content": "<|quad_start|>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "248052": {
69
+ "content": "<|quad_end|>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "248053": {
77
+ "content": "<|vision_start|>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "248054": {
85
+ "content": "<|vision_end|>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "248055": {
93
+ "content": "<|vision_pad|>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "248056": {
101
+ "content": "<|image_pad|>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "248057": {
109
+ "content": "<|video_pad|>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "248058": {
117
+ "content": "<tool_call>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": false
123
+ },
124
+ "248059": {
125
+ "content": "</tool_call>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": false
131
+ },
132
+ "248060": {
133
+ "content": "<|fim_prefix|>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": false
139
+ },
140
+ "248061": {
141
+ "content": "<|fim_middle|>",
142
+ "lstrip": false,
143
+ "normalized": false,
144
+ "rstrip": false,
145
+ "single_word": false,
146
+ "special": false
147
+ },
148
+ "248062": {
149
+ "content": "<|fim_suffix|>",
150
+ "lstrip": false,
151
+ "normalized": false,
152
+ "rstrip": false,
153
+ "single_word": false,
154
+ "special": false
155
+ },
156
+ "248063": {
157
+ "content": "<|fim_pad|>",
158
+ "lstrip": false,
159
+ "normalized": false,
160
+ "rstrip": false,
161
+ "single_word": false,
162
+ "special": false
163
+ },
164
+ "248064": {
165
+ "content": "<|repo_name|>",
166
+ "lstrip": false,
167
+ "normalized": false,
168
+ "rstrip": false,
169
+ "single_word": false,
170
+ "special": false
171
+ },
172
+ "248065": {
173
+ "content": "<|file_sep|>",
174
+ "lstrip": false,
175
+ "normalized": false,
176
+ "rstrip": false,
177
+ "single_word": false,
178
+ "special": false
179
+ },
180
+ "248066": {
181
+ "content": "<tool_response>",
182
+ "lstrip": false,
183
+ "normalized": false,
184
+ "rstrip": false,
185
+ "single_word": false,
186
+ "special": false
187
+ },
188
+ "248067": {
189
+ "content": "</tool_response>",
190
+ "lstrip": false,
191
+ "normalized": false,
192
+ "rstrip": false,
193
+ "single_word": false,
194
+ "special": false
195
+ },
196
+ "248068": {
197
+ "content": "<think>",
198
+ "lstrip": false,
199
+ "normalized": false,
200
+ "rstrip": false,
201
+ "single_word": false,
202
+ "special": false
203
+ },
204
+ "248069": {
205
+ "content": "</think>",
206
+ "lstrip": false,
207
+ "normalized": false,
208
+ "rstrip": false,
209
+ "single_word": false,
210
+ "special": false
211
+ },
212
+ "248070": {
213
+ "content": "<|audio_start|>",
214
+ "lstrip": false,
215
+ "normalized": false,
216
+ "rstrip": false,
217
+ "single_word": false,
218
+ "special": true
219
+ },
220
+ "248071": {
221
+ "content": "<|audio_end|>",
222
+ "lstrip": false,
223
+ "normalized": false,
224
+ "rstrip": false,
225
+ "single_word": false,
226
+ "special": true
227
+ },
228
+ "248072": {
229
+ "content": "<tts_pad>",
230
+ "lstrip": false,
231
+ "normalized": false,
232
+ "rstrip": false,
233
+ "single_word": false,
234
+ "special": true
235
+ },
236
+ "248073": {
237
+ "content": "<tts_text_bos>",
238
+ "lstrip": false,
239
+ "normalized": false,
240
+ "rstrip": false,
241
+ "single_word": false,
242
+ "special": true
243
+ },
244
+ "248074": {
245
+ "content": "<tts_text_eod>",
246
+ "lstrip": false,
247
+ "normalized": false,
248
+ "rstrip": false,
249
+ "single_word": false,
250
+ "special": true
251
+ },
252
+ "248075": {
253
+ "content": "<tts_text_bos_single>",
254
+ "lstrip": false,
255
+ "normalized": false,
256
+ "rstrip": false,
257
+ "single_word": false,
258
+ "special": true
259
+ },
260
+ "248076": {
261
+ "content": "<|audio_pad|>",
262
+ "lstrip": false,
263
+ "normalized": false,
264
+ "rstrip": false,
265
+ "single_word": false,
266
+ "special": true
267
+ }
268
+ },
269
+ "additional_special_tokens": [
270
+ "<|im_start|>",
271
+ "<|im_end|>",
272
+ "<|object_ref_start|>",
273
+ "<|object_ref_end|>",
274
+ "<|box_start|>",
275
+ "<|box_end|>",
276
+ "<|quad_start|>",
277
+ "<|quad_end|>",
278
+ "<|vision_start|>",
279
+ "<|vision_end|>",
280
+ "<|vision_pad|>",
281
+ "<|image_pad|>",
282
+ "<|video_pad|>"
283
+ ],
284
+ "bos_token": null,
285
+ "chat_template": "{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- macro render_content(content, do_vision_count, is_system_content=false) %}\n {%- if content is string %}\n {{- content }}\n {%- elif content is iterable and content is not mapping %}\n {%- for item in content %}\n {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}\n {%- if is_system_content %}\n {{- raise_exception('System message cannot contain images.') }}\n {%- endif %}\n {%- if do_vision_count %}\n {%- set image_count.value = image_count.value + 1 %}\n {%- endif %}\n {%- if add_vision_id %}\n {{- 'Picture ' ~ image_count.value ~ ': ' }}\n {%- endif %}\n {{- '<|vision_start|><|image_pad|><|vision_end|>' }}\n {%- elif 'video' in item or item.type == 'video' %}\n {%- if is_system_content %}\n {{- raise_exception('System message cannot contain videos.') }}\n {%- endif %}\n {%- if do_vision_count %}\n {%- set video_count.value = video_count.value + 1 %}\n {%- endif %}\n {%- if add_vision_id %}\n {{- 'Video ' ~ video_count.value ~ ': ' }}\n {%- endif %}\n {{- '<|vision_start|><|video_pad|><|vision_end|>' }}\n {%- elif 'text' in item %}\n {{- item.text }}\n {%- else %}\n {{- raise_exception('Unexpected item type in content.') }}\n {%- endif %}\n {%- endfor %}\n {%- elif content is none or content is undefined %}\n {{- '' }}\n {%- else %}\n {{- raise_exception('Unexpected content type.') }}\n {%- endif %}\n{%- endmacro %}\n{%- if not messages %}\n {{- raise_exception('No messages provided.') }}\n{%- endif %}\n{%- set reasoning_instructions = '' %}\n{%- if enable_thinking is undefined or enable_thinking is true %}\n {%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}\n {%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}\n {{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}\n {%- endif %}\n {%- if resolved_reasoning_effort == 'xhigh' %}\n {%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}\n {%- elif resolved_reasoning_effort == 'low' %}\n {%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}\n {%- endif %}\n{%- endif %}\n{%- if tools and tools is iterable and tools is not mapping %}\n {{- '<|im_start|>system\\n' }}\n {%- if reasoning_instructions %}\n {{- reasoning_instructions + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou have access to the following functions:\\n\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\" }}\n {{- '\\n\\nIf you choose to call a function ONLY reply in the following format with NO suffix:\\n\\n<tool_call>\\n<function=example_function_name>\\n<parameter=example_parameter_1>\\nvalue_1\\n</parameter>\\n<parameter=example_parameter_2>\\nThis is the value for the second parameter\\nthat can span\\nmultiple lines\\n</parameter>\\n</function>\\n</tool_call>\\n\\n<IMPORTANT>\\nReminder:\\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\\n- Required parameters MUST be specified\\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\\n</IMPORTANT>' }}\n {%- if messages[0].role == 'system' %}\n {%- set content = render_content(messages[0].content, false, true)|trim %}\n {%- if content %}\n {{- '\\n\\n' + content }}\n {%- endif %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n{%- else %}\n {%- if messages[0].role == 'system' %}\n {%- set content = render_content(messages[0].content, false, true)|trim %}\n {%- if content %}\n {{- '<|im_start|>system\\n' + (reasoning_instructions + '\\n\\n' if reasoning_instructions else '') + content + '<|im_end|>\\n' }}\n {%- elif reasoning_instructions %}\n {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n {%- endif %}\n {%- elif reasoning_instructions %}\n {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n {%- set index = (messages|length - 1) - loop.index0 %}\n {%- if ns.multi_step_tool and message.role == \"user\" %}\n {%- set content = render_content(message.content, false)|trim %}\n {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if ns.multi_step_tool %}\n {{- raise_exception('No user query found in messages.') }}\n{%- endif %}\n{%- for message in messages %}\n {%- set content = render_content(message.content, true)|trim %}\n {%- if message.role == \"system\" %}\n {%- if not loop.first %}\n {{- raise_exception('System message must be at the beginning.') }}\n {%- endif %}\n {%- elif message.role == \"user\" %}\n {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {%- set reasoning_content = '' %}\n {%- if message.reasoning_content is string %}\n {%- set reasoning_content = message.reasoning_content %}\n {%- endif %}\n {%- set reasoning_content = reasoning_content|trim %}\n {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}\n {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content + '\\n</think>\\n\\n' + content }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {%- if loop.first %}\n {%- if content|trim %}\n {{- '\\n\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- else %}\n {{- '<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- endif %}\n {%- else %}\n {{- '\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- endif %}\n {%- if tool_call.arguments is defined and tool_call.arguments != '' %}\n {%- for args_name, args_value in tool_call.arguments|items %}\n {{- '<parameter=' + args_name + '>\\n' }}\n {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}\n {{- args_value }}\n {{- '\\n</parameter>\\n' }}\n {%- endfor %}\n {%- endif %}\n {{- '</function>\\n</tool_call>' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.previtem and loop.previtem.role != \"tool\" %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- content }}\n {{- '\\n</tool_response>' }}\n {%- if not loop.last and loop.nextitem.role != \"tool\" %}\n {{- '<|im_end|>\\n' }}\n {%- elif loop.last %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- else %}\n {{- raise_exception('Unexpected message role.') }}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n {%- if enable_thinking is defined and enable_thinking is false %}\n {{- '<think>\\n\\n</think>\\n\\n' }}\n {%- else %}\n {{- '<think>\\n' }}\n {%- endif %}\n{%- endif %}",
286
+ "clean_up_tokenization_spaces": false,
287
+ "eos_token": "<|im_end|>",
288
+ "errors": "replace",
289
+ "model_max_length": 262144,
290
+ "pad_token": "<|endoftext|>",
291
+ "split_special_tokens": false,
292
+ "tokenizer_class": "Qwen2Tokenizer",
293
+ "unk_token": null,
294
+ "add_bos_token": false,
295
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
296
+ "extra_special_tokens": {
297
+ "audio_bos_token": "<|audio_start|>",
298
+ "audio_eos_token": "<|audio_end|>",
299
+ "audio_token": "<|audio_pad|>",
300
+ "image_token": "<|image_pad|>",
301
+ "video_token": "<|video_pad|>",
302
+ "vision_bos_token": "<|vision_start|>",
303
+ "vision_eos_token": "<|vision_end|>"
304
+ }
305
+ }
video_preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 25165824,
4
+ "shortest_edge": 4096
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "video_processor_type": "Qwen3VLVideoProcessor"
21
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff