desva0 commited on
Commit
3730026
·
verified ·
1 Parent(s): cca8ce3

W4A16 (compressed-tensors) uncensored Qwen3.8-27B for 24GB GPUs

Browse files
.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
LICENSE ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ Apache License
3
+ Version 2.0, January 2004
4
+ http://www.apache.org/licenses/
5
+
6
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
7
+
8
+ 1. Definitions.
9
+
10
+ "License" shall mean the terms and conditions for use, reproduction,
11
+ and distribution as defined by Sections 1 through 9 of this document.
12
+
13
+ "Licensor" shall mean the copyright owner or entity authorized by
14
+ the copyright owner that is granting the License.
15
+
16
+ "Legal Entity" shall mean the union of the acting entity and all
17
+ other entities that control, are controlled by, or are under common
18
+ control with that entity. For the purposes of this definition,
19
+ "control" means (i) the power, direct or indirect, to cause the
20
+ direction or management of such entity, whether by contract or
21
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
22
+ outstanding shares, or (iii) beneficial ownership of such entity.
23
+
24
+ "You" (or "Your") shall mean an individual or Legal Entity
25
+ exercising permissions granted by this License.
26
+
27
+ "Source" form shall mean the preferred form for making modifications,
28
+ including but not limited to software source code, documentation
29
+ source, and configuration files.
30
+
31
+ "Object" form shall mean any form resulting from mechanical
32
+ transformation or translation of a Source form, including but
33
+ not limited to compiled object code, generated documentation,
34
+ and conversions to other media types.
35
+
36
+ "Work" shall mean the work of authorship, whether in Source or
37
+ Object form, made available under the License, as indicated by a
38
+ copyright notice that is included in or attached to the work
39
+ (an example is provided in the Appendix below).
40
+
41
+ "Derivative Works" shall mean any work, whether in Source or Object
42
+ form, that is based on (or derived from) the Work and for which the
43
+ editorial revisions, annotations, elaborations, or other modifications
44
+ represent, as a whole, an original work of authorship. For the purposes
45
+ of this License, Derivative Works shall not include works that remain
46
+ separable from, or merely link (or bind by name) to the interfaces of,
47
+ the Work and Derivative Works thereof.
48
+
49
+ "Contribution" shall mean any work of authorship, including
50
+ the original version of the Work and any modifications or additions
51
+ to that Work or Derivative Works thereof, that is intentionally
52
+ submitted to Licensor for inclusion in the Work by the copyright owner
53
+ or by an individual or Legal Entity authorized to submit on behalf of
54
+ the copyright owner. For the purposes of this definition, "submitted"
55
+ means any form of electronic, verbal, or written communication sent
56
+ to the Licensor or its representatives, including but not limited to
57
+ communication on electronic mailing lists, source code control systems,
58
+ and issue tracking systems that are managed by, or on behalf of, the
59
+ Licensor for the purpose of discussing and improving the Work, but
60
+ excluding communication that is conspicuously marked or otherwise
61
+ designated in writing by the copyright owner as "Not a Contribution."
62
+
63
+ "Contributor" shall mean Licensor and any individual or Legal Entity
64
+ on behalf of whom a Contribution has been received by Licensor and
65
+ subsequently incorporated within the Work.
66
+
67
+ 2. Grant of Copyright License. Subject to the terms and conditions of
68
+ this License, each Contributor hereby grants to You a perpetual,
69
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
70
+ copyright license to reproduce, prepare Derivative Works of,
71
+ publicly display, publicly perform, sublicense, and distribute the
72
+ Work and such Derivative Works in Source or Object form.
73
+
74
+ 3. Grant of Patent License. Subject to the terms and conditions of
75
+ this License, each Contributor hereby grants to You a perpetual,
76
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
77
+ (except as stated in this section) patent license to make, have made,
78
+ use, offer to sell, sell, import, and otherwise transfer the Work,
79
+ where such license applies only to those patent claims licensable
80
+ by such Contributor that are necessarily infringed by their
81
+ Contribution(s) alone or by combination of their Contribution(s)
82
+ with the Work to which such Contribution(s) was submitted. If You
83
+ institute patent litigation against any entity (including a
84
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
85
+ or a Contribution incorporated within the Work constitutes direct
86
+ or contributory patent infringement, then any patent licenses
87
+ granted to You under this License for that Work shall terminate
88
+ as of the date such litigation is filed.
89
+
90
+ 4. Redistribution. You may reproduce and distribute copies of the
91
+ Work or Derivative Works thereof in any medium, with or without
92
+ modifications, and in Source or Object form, provided that You
93
+ meet the following conditions:
94
+
95
+ (a) You must give any other recipients of the Work or
96
+ Derivative Works a copy of this License; and
97
+
98
+ (b) You must cause any modified files to carry prominent notices
99
+ stating that You changed the files; and
100
+
101
+ (c) You must retain, in the Source form of any Derivative Works
102
+ that You distribute, all copyright, patent, trademark, and
103
+ attribution notices from the Source form of the Work,
104
+ excluding those notices that do not pertain to any part of
105
+ the Derivative Works; and
106
+
107
+ (d) If the Work includes a "NOTICE" text file as part of its
108
+ distribution, then any Derivative Works that You distribute must
109
+ include a readable copy of the attribution notices contained
110
+ within such NOTICE file, excluding those notices that do not
111
+ pertain to any part of the Derivative Works, in at least one
112
+ of the following places: within a NOTICE text file distributed
113
+ as part of the Derivative Works; within the Source form or
114
+ documentation, if provided along with the Derivative Works; or,
115
+ within a display generated by the Derivative Works, if and
116
+ wherever such third-party notices normally appear. The contents
117
+ of the NOTICE file are for informational purposes only and
118
+ do not modify the License. You may add Your own attribution
119
+ notices within Derivative Works that You distribute, alongside
120
+ or as an addendum to the NOTICE text from the Work, provided
121
+ that such additional attribution notices cannot be construed
122
+ as modifying the License.
123
+
124
+ You may add Your own copyright statement to Your modifications and
125
+ may provide additional or different license terms and conditions
126
+ for use, reproduction, or distribution of Your modifications, or
127
+ for any such Derivative Works as a whole, provided Your use,
128
+ reproduction, and distribution of the Work otherwise complies with
129
+ the conditions stated in this License.
130
+
131
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
132
+ any Contribution intentionally submitted for inclusion in the Work
133
+ by You to the Licensor shall be under the terms and conditions of
134
+ this License, without any additional terms or conditions.
135
+ Notwithstanding the above, nothing herein shall supersede or modify
136
+ the terms of any separate license agreement you may have executed
137
+ with Licensor regarding such Contributions.
138
+
139
+ 6. Trademarks. This License does not grant permission to use the trade
140
+ names, trademarks, service marks, or product names of the Licensor,
141
+ except as required for reasonable and customary use in describing the
142
+ origin of the Work and reproducing the content of the NOTICE file.
143
+
144
+ 7. Disclaimer of Warranty. Unless required by applicable law or
145
+ agreed to in writing, Licensor provides the Work (and each
146
+ Contributor provides its Contributions) on an "AS IS" BASIS,
147
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
148
+ implied, including, without limitation, any warranties or conditions
149
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
150
+ PARTICULAR PURPOSE. You are solely responsible for determining the
151
+ appropriateness of using or redistributing the Work and assume any
152
+ risks associated with Your exercise of permissions under this License.
153
+
154
+ 8. Limitation of Liability. In no event and under no legal theory,
155
+ whether in tort (including negligence), contract, or otherwise,
156
+ unless required by applicable law (such as deliberate and grossly
157
+ negligent acts) or agreed to in writing, shall any Contributor be
158
+ liable to You for damages, including any direct, indirect, special,
159
+ incidental, or consequential damages of any character arising as a
160
+ result of this License or out of the use or inability to use the
161
+ Work (including but not limited to damages for loss of goodwill,
162
+ work stoppage, computer failure or malfunction, or any and all
163
+ other commercial damages or losses), even if such Contributor
164
+ has been advised of the possibility of such damages.
165
+
166
+ 9. Accepting Warranty or Additional Liability. While redistributing
167
+ the Work or Derivative Works thereof, You may choose to offer,
168
+ and charge a fee for, acceptance of support, warranty, indemnity,
169
+ or other liability obligations and/or rights consistent with this
170
+ License. However, in accepting such obligations, You may act only
171
+ on Your own behalf and on Your sole responsibility, not on behalf
172
+ of any other Contributor, and only if You agree to indemnify,
173
+ defend, and hold each Contributor harmless for any liability
174
+ incurred by, or claims asserted against, such Contributor by reason
175
+ of your accepting any such warranty or additional liability.
176
+
177
+ END OF TERMS AND CONDITIONS
178
+
179
+ APPENDIX: How to apply the Apache License to your work.
180
+
181
+ To apply the Apache License to your work, attach the following
182
+ boilerplate notice, with the fields enclosed by brackets "[]"
183
+ replaced with your own identifying information. (Don't include
184
+ the brackets!) The text should be enclosed in the appropriate
185
+ comment syntax for the file format. We also recommend that a
186
+ file or class name and description of purpose be included on the
187
+ same "printed page" as the copyright notice for easier
188
+ identification within third-party archives.
189
+
190
+ Copyright 2026 Alibaba Cloud
191
+
192
+ Licensed under the Apache License, Version 2.0 (the "License");
193
+ you may not use this file except in compliance with the License.
194
+ You may obtain a copy of the License at
195
+
196
+ http://www.apache.org/licenses/LICENSE-2.0
197
+
198
+ Unless required by applicable law or agreed to in writing, software
199
+ distributed under the License is distributed on an "AS IS" BASIS,
200
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
201
+ See the License for the specific language governing permissions and
202
+ limitations under the License.
README.md ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model:
4
+ - Qwen/Qwen3.8-27B
5
+ base_model_relation: quantized
6
+ pipeline_tag: image-text-to-text
7
+ library_name: transformers
8
+ language:
9
+ - en
10
+ - zh
11
+ tags:
12
+ - qwen
13
+ - qwen3
14
+ - qwen3.8
15
+ - uncensored
16
+ - abliterated
17
+ - vision-language
18
+ - function-calling
19
+ - compressed-tensors
20
+ - w4a16
21
+ - int4
22
+ - vllm
23
+ - rtx3090
24
+ ---
25
+
26
+ # Qwen3.8-27B-Uncensored — W4A16 (RTX 3090 build)
27
+
28
+ A **4-bit weight-only (W4A16)** quantization of the *uncensored / abliterated*
29
+ [Qwen3.8-27B](https://huggingface.co/Qwen/Qwen3.8-27B) vision-language model,
30
+ sized to run on a single **24 GB GPU (e.g. RTX 3090)**.
31
+
32
+ This is the exact quant used to serve an agentic assistant ("Sage") live on an
33
+ RTX 3090 in day-to-day use, so it's known to load and run under vLLM on
34
+ consumer 24 GB hardware — not just a theoretical export.
35
+
36
+ ## What it is
37
+
38
+ - **Base model:** Qwen3.8-27B (VL, function-calling, reasoning), architecture
39
+ `Qwen3_5ForConditionalGeneration`.
40
+ - **Uncensoring:** the refusal-removed ("abliterated") build published by
41
+ [OrcaRouter](https://www.orcarouter.ai) under Apache-2.0. This repo is a
42
+ quantization of that build.
43
+ - **Quantization:** 4-bit weight-only, produced with AutoRound and exported in
44
+ the **compressed-tensors** `pack-quantized` format:
45
+ - `num_bits: 4`, `type: int`, `symmetric: true`, `group_size: 128`,
46
+ `strategy: group`
47
+ - Vision-tower linears are kept at higher precision (listed under `ignore`),
48
+ so image understanding is preserved.
49
+ - **Footprint:** ~15 GB of weights → fits comfortably in 24 GB with room for
50
+ KV cache at a useful context length.
51
+
52
+ ## Serving with vLLM
53
+
54
+ ```bash
55
+ vllm serve <path-or-repo>/Qwen3.8-27B-Uncensored-W4A16 \
56
+ --host 0.0.0.0 --port 8000 \
57
+ --served-model-name qwen3.8-27b \
58
+ --gpu-memory-utilization 0.90 \
59
+ --max-model-len 32768 \
60
+ --enable-auto-tool-choice --tool-call-parser hermes
61
+ ```
62
+
63
+ Then hit the OpenAI-compatible endpoint at `http://localhost:8000/v1`. Adjust
64
+ `--max-model-len` to trade context length against KV-cache memory on a 24 GB
65
+ card. A recent vLLM with `compressed-tensors` support is required.
66
+
67
+ ## Provenance & attribution
68
+
69
+ - **Qwen** — the Qwen3.8-27B base model (© the Qwen team).
70
+ - **OrcaRouter** — the Apache-2.0 abliterated build this repo quantizes.
71
+ - **This repo** — 4-bit W4A16 (compressed-tensors) quantization for 24 GB GPUs.
72
+
73
+ ## License
74
+
75
+ Apache-2.0, inherited from the upstream abliterated build. See the bundled
76
+ `LICENSE`. You must retain the license and attribution when redistributing.
77
+
78
+ ## Responsible use
79
+
80
+ This is an **uncensored / abliterated** model: its built-in refusal behaviour
81
+ has been removed, so it will attempt requests an aligned model would decline.
82
+ You are responsible for how you deploy it, for any guardrails you add, and for
83
+ compliance with the laws and platform rules that apply to you.
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 +1,447 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "image_token_id": 248056,
7
+ "language_model_only": false,
8
+ "model_type": "qwen3_5",
9
+ "quantization_config": {
10
+ "config_groups": {
11
+ "group_0": {
12
+ "format": "pack-quantized",
13
+ "input_activations": null,
14
+ "output_activations": null,
15
+ "targets": [
16
+ "Linear"
17
+ ],
18
+ "weights": {
19
+ "actorder": null,
20
+ "block_structure": null,
21
+ "dynamic": false,
22
+ "group_size": 128,
23
+ "num_bits": 4,
24
+ "observer": "memoryless_minmax",
25
+ "observer_kwargs": {},
26
+ "scale_dtype": null,
27
+ "strategy": "group",
28
+ "symmetric": true,
29
+ "type": "int",
30
+ "zp_dtype": null
31
+ }
32
+ },
33
+ "group_1": {
34
+ "format": "pack-quantized",
35
+ "input_activations": null,
36
+ "output_activations": null,
37
+ "targets": [
38
+ "re:.*lm_head$"
39
+ ],
40
+ "weights": {
41
+ "actorder": null,
42
+ "block_structure": null,
43
+ "dynamic": false,
44
+ "group_size": 128,
45
+ "num_bits": 4,
46
+ "observer": "memoryless_minmax",
47
+ "observer_kwargs": {},
48
+ "scale_dtype": null,
49
+ "strategy": "group",
50
+ "symmetric": true,
51
+ "type": "int",
52
+ "zp_dtype": null
53
+ }
54
+ },
55
+ "group_2": {
56
+ "format": "pack-quantized",
57
+ "input_activations": null,
58
+ "output_activations": null,
59
+ "targets": [
60
+ "re:.*embed_tokens$"
61
+ ],
62
+ "weights": {
63
+ "actorder": null,
64
+ "block_structure": null,
65
+ "dynamic": false,
66
+ "group_size": 128,
67
+ "num_bits": 8,
68
+ "observer": "memoryless_minmax",
69
+ "observer_kwargs": {},
70
+ "scale_dtype": null,
71
+ "strategy": "group",
72
+ "symmetric": true,
73
+ "type": "int",
74
+ "zp_dtype": null
75
+ }
76
+ },
77
+ "group_3": {
78
+ "format": "pack-quantized",
79
+ "input_activations": null,
80
+ "output_activations": null,
81
+ "targets": [
82
+ "re:^mtp\\..*"
83
+ ],
84
+ "weights": {
85
+ "actorder": null,
86
+ "block_structure": null,
87
+ "dynamic": false,
88
+ "group_size": 128,
89
+ "num_bits": 8,
90
+ "observer": "memoryless_minmax",
91
+ "observer_kwargs": {},
92
+ "scale_dtype": null,
93
+ "strategy": "group",
94
+ "symmetric": true,
95
+ "type": "int",
96
+ "zp_dtype": null
97
+ }
98
+ }
99
+ },
100
+ "format": "pack-quantized",
101
+ "global_compression_ratio": null,
102
+ "ignore": [
103
+ "model.visual.blocks.0.attn.qkv",
104
+ "model.visual.blocks.0.attn.proj",
105
+ "model.visual.blocks.0.mlp.linear_fc1",
106
+ "model.visual.blocks.0.mlp.linear_fc2",
107
+ "model.visual.blocks.1.attn.qkv",
108
+ "model.visual.blocks.1.attn.proj",
109
+ "model.visual.blocks.1.mlp.linear_fc1",
110
+ "model.visual.blocks.1.mlp.linear_fc2",
111
+ "model.visual.blocks.2.attn.qkv",
112
+ "model.visual.blocks.2.attn.proj",
113
+ "model.visual.blocks.2.mlp.linear_fc1",
114
+ "model.visual.blocks.2.mlp.linear_fc2",
115
+ "model.visual.blocks.3.attn.qkv",
116
+ "model.visual.blocks.3.attn.proj",
117
+ "model.visual.blocks.3.mlp.linear_fc1",
118
+ "model.visual.blocks.3.mlp.linear_fc2",
119
+ "model.visual.blocks.4.attn.qkv",
120
+ "model.visual.blocks.4.attn.proj",
121
+ "model.visual.blocks.4.mlp.linear_fc1",
122
+ "model.visual.blocks.4.mlp.linear_fc2",
123
+ "model.visual.blocks.5.attn.qkv",
124
+ "model.visual.blocks.5.attn.proj",
125
+ "model.visual.blocks.5.mlp.linear_fc1",
126
+ "model.visual.blocks.5.mlp.linear_fc2",
127
+ "model.visual.blocks.6.attn.qkv",
128
+ "model.visual.blocks.6.attn.proj",
129
+ "model.visual.blocks.6.mlp.linear_fc1",
130
+ "model.visual.blocks.6.mlp.linear_fc2",
131
+ "model.visual.blocks.7.attn.qkv",
132
+ "model.visual.blocks.7.attn.proj",
133
+ "model.visual.blocks.7.mlp.linear_fc1",
134
+ "model.visual.blocks.7.mlp.linear_fc2",
135
+ "model.visual.blocks.8.attn.qkv",
136
+ "model.visual.blocks.8.attn.proj",
137
+ "model.visual.blocks.8.mlp.linear_fc1",
138
+ "model.visual.blocks.8.mlp.linear_fc2",
139
+ "model.visual.blocks.9.attn.qkv",
140
+ "model.visual.blocks.9.attn.proj",
141
+ "model.visual.blocks.9.mlp.linear_fc1",
142
+ "model.visual.blocks.9.mlp.linear_fc2",
143
+ "model.visual.blocks.10.attn.qkv",
144
+ "model.visual.blocks.10.attn.proj",
145
+ "model.visual.blocks.10.mlp.linear_fc1",
146
+ "model.visual.blocks.10.mlp.linear_fc2",
147
+ "model.visual.blocks.11.attn.qkv",
148
+ "model.visual.blocks.11.attn.proj",
149
+ "model.visual.blocks.11.mlp.linear_fc1",
150
+ "model.visual.blocks.11.mlp.linear_fc2",
151
+ "model.visual.blocks.12.attn.qkv",
152
+ "model.visual.blocks.12.attn.proj",
153
+ "model.visual.blocks.12.mlp.linear_fc1",
154
+ "model.visual.blocks.12.mlp.linear_fc2",
155
+ "model.visual.blocks.13.attn.qkv",
156
+ "model.visual.blocks.13.attn.proj",
157
+ "model.visual.blocks.13.mlp.linear_fc1",
158
+ "model.visual.blocks.13.mlp.linear_fc2",
159
+ "model.visual.blocks.14.attn.qkv",
160
+ "model.visual.blocks.14.attn.proj",
161
+ "model.visual.blocks.14.mlp.linear_fc1",
162
+ "model.visual.blocks.14.mlp.linear_fc2",
163
+ "model.visual.blocks.15.attn.qkv",
164
+ "model.visual.blocks.15.attn.proj",
165
+ "model.visual.blocks.15.mlp.linear_fc1",
166
+ "model.visual.blocks.15.mlp.linear_fc2",
167
+ "model.visual.blocks.16.attn.qkv",
168
+ "model.visual.blocks.16.attn.proj",
169
+ "model.visual.blocks.16.mlp.linear_fc1",
170
+ "model.visual.blocks.16.mlp.linear_fc2",
171
+ "model.visual.blocks.17.attn.qkv",
172
+ "model.visual.blocks.17.attn.proj",
173
+ "model.visual.blocks.17.mlp.linear_fc1",
174
+ "model.visual.blocks.17.mlp.linear_fc2",
175
+ "model.visual.blocks.18.attn.qkv",
176
+ "model.visual.blocks.18.attn.proj",
177
+ "model.visual.blocks.18.mlp.linear_fc1",
178
+ "model.visual.blocks.18.mlp.linear_fc2",
179
+ "model.visual.blocks.19.attn.qkv",
180
+ "model.visual.blocks.19.attn.proj",
181
+ "model.visual.blocks.19.mlp.linear_fc1",
182
+ "model.visual.blocks.19.mlp.linear_fc2",
183
+ "model.visual.blocks.20.attn.qkv",
184
+ "model.visual.blocks.20.attn.proj",
185
+ "model.visual.blocks.20.mlp.linear_fc1",
186
+ "model.visual.blocks.20.mlp.linear_fc2",
187
+ "model.visual.blocks.21.attn.qkv",
188
+ "model.visual.blocks.21.attn.proj",
189
+ "model.visual.blocks.21.mlp.linear_fc1",
190
+ "model.visual.blocks.21.mlp.linear_fc2",
191
+ "model.visual.blocks.22.attn.qkv",
192
+ "model.visual.blocks.22.attn.proj",
193
+ "model.visual.blocks.22.mlp.linear_fc1",
194
+ "model.visual.blocks.22.mlp.linear_fc2",
195
+ "model.visual.blocks.23.attn.qkv",
196
+ "model.visual.blocks.23.attn.proj",
197
+ "model.visual.blocks.23.mlp.linear_fc1",
198
+ "model.visual.blocks.23.mlp.linear_fc2",
199
+ "model.visual.blocks.24.attn.qkv",
200
+ "model.visual.blocks.24.attn.proj",
201
+ "model.visual.blocks.24.mlp.linear_fc1",
202
+ "model.visual.blocks.24.mlp.linear_fc2",
203
+ "model.visual.blocks.25.attn.qkv",
204
+ "model.visual.blocks.25.attn.proj",
205
+ "model.visual.blocks.25.mlp.linear_fc1",
206
+ "model.visual.blocks.25.mlp.linear_fc2",
207
+ "model.visual.blocks.26.attn.qkv",
208
+ "model.visual.blocks.26.attn.proj",
209
+ "model.visual.blocks.26.mlp.linear_fc1",
210
+ "model.visual.blocks.26.mlp.linear_fc2",
211
+ "model.visual.merger.linear_fc1",
212
+ "model.visual.merger.linear_fc2",
213
+ "model.language_model.layers.0.linear_attn.in_proj_b",
214
+ "model.language_model.layers.0.linear_attn.in_proj_a",
215
+ "model.language_model.layers.1.linear_attn.in_proj_b",
216
+ "model.language_model.layers.1.linear_attn.in_proj_a",
217
+ "model.language_model.layers.2.linear_attn.in_proj_b",
218
+ "model.language_model.layers.2.linear_attn.in_proj_a",
219
+ "model.language_model.layers.4.linear_attn.in_proj_b",
220
+ "model.language_model.layers.4.linear_attn.in_proj_a",
221
+ "model.language_model.layers.5.linear_attn.in_proj_b",
222
+ "model.language_model.layers.5.linear_attn.in_proj_a",
223
+ "model.language_model.layers.6.linear_attn.in_proj_b",
224
+ "model.language_model.layers.6.linear_attn.in_proj_a",
225
+ "model.language_model.layers.8.linear_attn.in_proj_b",
226
+ "model.language_model.layers.8.linear_attn.in_proj_a",
227
+ "model.language_model.layers.9.linear_attn.in_proj_b",
228
+ "model.language_model.layers.9.linear_attn.in_proj_a",
229
+ "model.language_model.layers.10.linear_attn.in_proj_b",
230
+ "model.language_model.layers.10.linear_attn.in_proj_a",
231
+ "model.language_model.layers.12.linear_attn.in_proj_b",
232
+ "model.language_model.layers.12.linear_attn.in_proj_a",
233
+ "model.language_model.layers.13.linear_attn.in_proj_b",
234
+ "model.language_model.layers.13.linear_attn.in_proj_a",
235
+ "model.language_model.layers.14.linear_attn.in_proj_b",
236
+ "model.language_model.layers.14.linear_attn.in_proj_a",
237
+ "model.language_model.layers.16.linear_attn.in_proj_b",
238
+ "model.language_model.layers.16.linear_attn.in_proj_a",
239
+ "model.language_model.layers.17.linear_attn.in_proj_b",
240
+ "model.language_model.layers.17.linear_attn.in_proj_a",
241
+ "model.language_model.layers.18.linear_attn.in_proj_b",
242
+ "model.language_model.layers.18.linear_attn.in_proj_a",
243
+ "model.language_model.layers.20.linear_attn.in_proj_b",
244
+ "model.language_model.layers.20.linear_attn.in_proj_a",
245
+ "model.language_model.layers.21.linear_attn.in_proj_b",
246
+ "model.language_model.layers.21.linear_attn.in_proj_a",
247
+ "model.language_model.layers.22.linear_attn.in_proj_b",
248
+ "model.language_model.layers.22.linear_attn.in_proj_a",
249
+ "model.language_model.layers.24.linear_attn.in_proj_b",
250
+ "model.language_model.layers.24.linear_attn.in_proj_a",
251
+ "model.language_model.layers.25.linear_attn.in_proj_b",
252
+ "model.language_model.layers.25.linear_attn.in_proj_a",
253
+ "model.language_model.layers.26.linear_attn.in_proj_b",
254
+ "model.language_model.layers.26.linear_attn.in_proj_a",
255
+ "model.language_model.layers.28.linear_attn.in_proj_b",
256
+ "model.language_model.layers.28.linear_attn.in_proj_a",
257
+ "model.language_model.layers.29.linear_attn.in_proj_b",
258
+ "model.language_model.layers.29.linear_attn.in_proj_a",
259
+ "model.language_model.layers.30.linear_attn.in_proj_b",
260
+ "model.language_model.layers.30.linear_attn.in_proj_a",
261
+ "model.language_model.layers.32.linear_attn.in_proj_b",
262
+ "model.language_model.layers.32.linear_attn.in_proj_a",
263
+ "model.language_model.layers.33.linear_attn.in_proj_b",
264
+ "model.language_model.layers.33.linear_attn.in_proj_a",
265
+ "model.language_model.layers.34.linear_attn.in_proj_b",
266
+ "model.language_model.layers.34.linear_attn.in_proj_a",
267
+ "model.language_model.layers.36.linear_attn.in_proj_b",
268
+ "model.language_model.layers.36.linear_attn.in_proj_a",
269
+ "model.language_model.layers.37.linear_attn.in_proj_b",
270
+ "model.language_model.layers.37.linear_attn.in_proj_a",
271
+ "model.language_model.layers.38.linear_attn.in_proj_b",
272
+ "model.language_model.layers.38.linear_attn.in_proj_a",
273
+ "model.language_model.layers.40.linear_attn.in_proj_b",
274
+ "model.language_model.layers.40.linear_attn.in_proj_a",
275
+ "model.language_model.layers.41.linear_attn.in_proj_b",
276
+ "model.language_model.layers.41.linear_attn.in_proj_a",
277
+ "model.language_model.layers.42.linear_attn.in_proj_b",
278
+ "model.language_model.layers.42.linear_attn.in_proj_a",
279
+ "model.language_model.layers.44.linear_attn.in_proj_b",
280
+ "model.language_model.layers.44.linear_attn.in_proj_a",
281
+ "model.language_model.layers.45.linear_attn.in_proj_b",
282
+ "model.language_model.layers.45.linear_attn.in_proj_a",
283
+ "model.language_model.layers.46.linear_attn.in_proj_b",
284
+ "model.language_model.layers.46.linear_attn.in_proj_a",
285
+ "model.language_model.layers.48.linear_attn.in_proj_b",
286
+ "model.language_model.layers.48.linear_attn.in_proj_a",
287
+ "model.language_model.layers.49.linear_attn.in_proj_b",
288
+ "model.language_model.layers.49.linear_attn.in_proj_a",
289
+ "model.language_model.layers.50.linear_attn.in_proj_b",
290
+ "model.language_model.layers.50.linear_attn.in_proj_a",
291
+ "model.language_model.layers.52.linear_attn.in_proj_b",
292
+ "model.language_model.layers.52.linear_attn.in_proj_a",
293
+ "model.language_model.layers.53.linear_attn.in_proj_b",
294
+ "model.language_model.layers.53.linear_attn.in_proj_a",
295
+ "model.language_model.layers.54.linear_attn.in_proj_b",
296
+ "model.language_model.layers.54.linear_attn.in_proj_a",
297
+ "model.language_model.layers.56.linear_attn.in_proj_b",
298
+ "model.language_model.layers.56.linear_attn.in_proj_a",
299
+ "model.language_model.layers.57.linear_attn.in_proj_b",
300
+ "model.language_model.layers.57.linear_attn.in_proj_a",
301
+ "model.language_model.layers.58.linear_attn.in_proj_b",
302
+ "model.language_model.layers.58.linear_attn.in_proj_a",
303
+ "model.language_model.layers.60.linear_attn.in_proj_b",
304
+ "model.language_model.layers.60.linear_attn.in_proj_a",
305
+ "model.language_model.layers.61.linear_attn.in_proj_b",
306
+ "model.language_model.layers.61.linear_attn.in_proj_a",
307
+ "model.language_model.layers.62.linear_attn.in_proj_b",
308
+ "model.language_model.layers.62.linear_attn.in_proj_a"
309
+ ],
310
+ "kv_cache_scheme": null,
311
+ "quant_method": "compressed-tensors",
312
+ "quantization_status": "compressed"
313
+ },
314
+ "text_config": {
315
+ "attention_bias": false,
316
+ "attention_dropout": 0.0,
317
+ "attn_output_gate": true,
318
+ "bos_token_id": 248044,
319
+ "dtype": "bfloat16",
320
+ "eos_token_id": 248044,
321
+ "full_attention_interval": 4,
322
+ "head_dim": 256,
323
+ "hidden_act": "silu",
324
+ "hidden_size": 5120,
325
+ "initializer_range": 0.02,
326
+ "intermediate_size": 17408,
327
+ "layer_types": [
328
+ "linear_attention",
329
+ "linear_attention",
330
+ "linear_attention",
331
+ "full_attention",
332
+ "linear_attention",
333
+ "linear_attention",
334
+ "linear_attention",
335
+ "full_attention",
336
+ "linear_attention",
337
+ "linear_attention",
338
+ "linear_attention",
339
+ "full_attention",
340
+ "linear_attention",
341
+ "linear_attention",
342
+ "linear_attention",
343
+ "full_attention",
344
+ "linear_attention",
345
+ "linear_attention",
346
+ "linear_attention",
347
+ "full_attention",
348
+ "linear_attention",
349
+ "linear_attention",
350
+ "linear_attention",
351
+ "full_attention",
352
+ "linear_attention",
353
+ "linear_attention",
354
+ "linear_attention",
355
+ "full_attention",
356
+ "linear_attention",
357
+ "linear_attention",
358
+ "linear_attention",
359
+ "full_attention",
360
+ "linear_attention",
361
+ "linear_attention",
362
+ "linear_attention",
363
+ "full_attention",
364
+ "linear_attention",
365
+ "linear_attention",
366
+ "linear_attention",
367
+ "full_attention",
368
+ "linear_attention",
369
+ "linear_attention",
370
+ "linear_attention",
371
+ "full_attention",
372
+ "linear_attention",
373
+ "linear_attention",
374
+ "linear_attention",
375
+ "full_attention",
376
+ "linear_attention",
377
+ "linear_attention",
378
+ "linear_attention",
379
+ "full_attention",
380
+ "linear_attention",
381
+ "linear_attention",
382
+ "linear_attention",
383
+ "full_attention",
384
+ "linear_attention",
385
+ "linear_attention",
386
+ "linear_attention",
387
+ "full_attention",
388
+ "linear_attention",
389
+ "linear_attention",
390
+ "linear_attention",
391
+ "full_attention"
392
+ ],
393
+ "linear_conv_kernel_dim": 4,
394
+ "linear_key_head_dim": 128,
395
+ "linear_num_key_heads": 16,
396
+ "linear_num_value_heads": 48,
397
+ "linear_value_head_dim": 128,
398
+ "mamba_ssm_dtype": "float32",
399
+ "max_position_embeddings": 262144,
400
+ "model_type": "qwen3_5_text",
401
+ "mtp_num_hidden_layers": 1,
402
+ "mtp_use_dedicated_embeddings": false,
403
+ "num_attention_heads": 24,
404
+ "num_hidden_layers": 64,
405
+ "num_key_value_heads": 4,
406
+ "output_gate_type": "swish",
407
+ "pad_token_id": null,
408
+ "partial_rotary_factor": 0.25,
409
+ "rms_norm_eps": 1e-06,
410
+ "rope_parameters": {
411
+ "mrope_interleaved": true,
412
+ "mrope_section": [
413
+ 11,
414
+ 11,
415
+ 10
416
+ ],
417
+ "partial_rotary_factor": 0.25,
418
+ "rope_theta": 10000000,
419
+ "rope_type": "default"
420
+ },
421
+ "tie_word_embeddings": false,
422
+ "use_cache": true,
423
+ "vocab_size": 248320
424
+ },
425
+ "tie_word_embeddings": false,
426
+ "transformers_version": "5.15.0",
427
+ "video_token_id": 248057,
428
+ "vision_config": {
429
+ "deepstack_visual_indexes": [],
430
+ "depth": 27,
431
+ "dtype": "bfloat16",
432
+ "hidden_act": "gelu_pytorch_tanh",
433
+ "hidden_size": 1152,
434
+ "in_channels": 3,
435
+ "initializer_range": 0.02,
436
+ "intermediate_size": 4304,
437
+ "model_type": "qwen3_5_vision",
438
+ "num_heads": 16,
439
+ "num_position_embeddings": 2304,
440
+ "out_hidden_size": 5120,
441
+ "patch_size": 16,
442
+ "spatial_merge_size": 2,
443
+ "temporal_patch_size": 2
444
+ },
445
+ "vision_end_token_id": 248054,
446
+ "vision_start_token_id": 248053
447
+ }
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "transformers_version": "5.15.0"
13
+ }
model-00001-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4533bd5bedbc9d63d9f9585c1a3b9eaa18e5ab6c2378d82a8ce4d7dcfa35dc2
3
+ size 3211003472
model-00002-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a5284cb96b015a82d492df5cc8894a13b079431fa3ca05d4a74b9f9e345b028
3
+ size 3195027104
model-00003-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7916e255de7bfcd893fa0804d5e69fe384e5647f7281398b7069f7dbc7aba0db
3
+ size 3195027104
model-00004-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5579f215f09ec69b366863dc1a4a404d4125b12c5115869a26c008a6b7a62895
3
+ size 3217442568
model-00005-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0994bb8d91c07ceefc4bfffd3a7b7fd110cc9bd3ce865bdcb2821abf02a02fab
3
+ size 699275200
model-00006-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:560ddce0714276abe546b7ff5172ba966e2070aef138f69660ba8bcfef7a7fbe
3
+ size 1291274752
model-00007-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e38d80fca342d41e4ea4f02493e63a1b29e8e689fe4cc8d73580c77d8920dc59
3
+ size 655565120
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
model_extra_tensors.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:555cac3b034e897b946ed2eae066626d76a55b99b5fff3298b257e67667b0dbf
3
+ size 539499184
processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
quantization_config.json ADDED
@@ -0,0 +1,240 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config_groups": {
3
+ "group_0": {
4
+ "targets": [
5
+ "Linear"
6
+ ],
7
+ "weights": {
8
+ "num_bits": 4,
9
+ "type": "int",
10
+ "symmetric": true,
11
+ "group_size": 128,
12
+ "strategy": "group",
13
+ "block_structure": null,
14
+ "dynamic": false,
15
+ "actorder": null,
16
+ "scale_dtype": null,
17
+ "zp_dtype": null,
18
+ "observer": "memoryless_minmax",
19
+ "observer_kwargs": {}
20
+ },
21
+ "input_activations": null,
22
+ "output_activations": null,
23
+ "format": "pack-quantized"
24
+ }
25
+ },
26
+ "quant_method": "compressed-tensors",
27
+ "kv_cache_scheme": null,
28
+ "format": "pack-quantized",
29
+ "quantization_status": "compressed",
30
+ "global_compression_ratio": null,
31
+ "ignore": [
32
+ "model.visual.blocks.0.attn.qkv",
33
+ "model.visual.blocks.0.attn.proj",
34
+ "model.visual.blocks.0.mlp.linear_fc1",
35
+ "model.visual.blocks.0.mlp.linear_fc2",
36
+ "model.visual.blocks.1.attn.qkv",
37
+ "model.visual.blocks.1.attn.proj",
38
+ "model.visual.blocks.1.mlp.linear_fc1",
39
+ "model.visual.blocks.1.mlp.linear_fc2",
40
+ "model.visual.blocks.2.attn.qkv",
41
+ "model.visual.blocks.2.attn.proj",
42
+ "model.visual.blocks.2.mlp.linear_fc1",
43
+ "model.visual.blocks.2.mlp.linear_fc2",
44
+ "model.visual.blocks.3.attn.qkv",
45
+ "model.visual.blocks.3.attn.proj",
46
+ "model.visual.blocks.3.mlp.linear_fc1",
47
+ "model.visual.blocks.3.mlp.linear_fc2",
48
+ "model.visual.blocks.4.attn.qkv",
49
+ "model.visual.blocks.4.attn.proj",
50
+ "model.visual.blocks.4.mlp.linear_fc1",
51
+ "model.visual.blocks.4.mlp.linear_fc2",
52
+ "model.visual.blocks.5.attn.qkv",
53
+ "model.visual.blocks.5.attn.proj",
54
+ "model.visual.blocks.5.mlp.linear_fc1",
55
+ "model.visual.blocks.5.mlp.linear_fc2",
56
+ "model.visual.blocks.6.attn.qkv",
57
+ "model.visual.blocks.6.attn.proj",
58
+ "model.visual.blocks.6.mlp.linear_fc1",
59
+ "model.visual.blocks.6.mlp.linear_fc2",
60
+ "model.visual.blocks.7.attn.qkv",
61
+ "model.visual.blocks.7.attn.proj",
62
+ "model.visual.blocks.7.mlp.linear_fc1",
63
+ "model.visual.blocks.7.mlp.linear_fc2",
64
+ "model.visual.blocks.8.attn.qkv",
65
+ "model.visual.blocks.8.attn.proj",
66
+ "model.visual.blocks.8.mlp.linear_fc1",
67
+ "model.visual.blocks.8.mlp.linear_fc2",
68
+ "model.visual.blocks.9.attn.qkv",
69
+ "model.visual.blocks.9.attn.proj",
70
+ "model.visual.blocks.9.mlp.linear_fc1",
71
+ "model.visual.blocks.9.mlp.linear_fc2",
72
+ "model.visual.blocks.10.attn.qkv",
73
+ "model.visual.blocks.10.attn.proj",
74
+ "model.visual.blocks.10.mlp.linear_fc1",
75
+ "model.visual.blocks.10.mlp.linear_fc2",
76
+ "model.visual.blocks.11.attn.qkv",
77
+ "model.visual.blocks.11.attn.proj",
78
+ "model.visual.blocks.11.mlp.linear_fc1",
79
+ "model.visual.blocks.11.mlp.linear_fc2",
80
+ "model.visual.blocks.12.attn.qkv",
81
+ "model.visual.blocks.12.attn.proj",
82
+ "model.visual.blocks.12.mlp.linear_fc1",
83
+ "model.visual.blocks.12.mlp.linear_fc2",
84
+ "model.visual.blocks.13.attn.qkv",
85
+ "model.visual.blocks.13.attn.proj",
86
+ "model.visual.blocks.13.mlp.linear_fc1",
87
+ "model.visual.blocks.13.mlp.linear_fc2",
88
+ "model.visual.blocks.14.attn.qkv",
89
+ "model.visual.blocks.14.attn.proj",
90
+ "model.visual.blocks.14.mlp.linear_fc1",
91
+ "model.visual.blocks.14.mlp.linear_fc2",
92
+ "model.visual.blocks.15.attn.qkv",
93
+ "model.visual.blocks.15.attn.proj",
94
+ "model.visual.blocks.15.mlp.linear_fc1",
95
+ "model.visual.blocks.15.mlp.linear_fc2",
96
+ "model.visual.blocks.16.attn.qkv",
97
+ "model.visual.blocks.16.attn.proj",
98
+ "model.visual.blocks.16.mlp.linear_fc1",
99
+ "model.visual.blocks.16.mlp.linear_fc2",
100
+ "model.visual.blocks.17.attn.qkv",
101
+ "model.visual.blocks.17.attn.proj",
102
+ "model.visual.blocks.17.mlp.linear_fc1",
103
+ "model.visual.blocks.17.mlp.linear_fc2",
104
+ "model.visual.blocks.18.attn.qkv",
105
+ "model.visual.blocks.18.attn.proj",
106
+ "model.visual.blocks.18.mlp.linear_fc1",
107
+ "model.visual.blocks.18.mlp.linear_fc2",
108
+ "model.visual.blocks.19.attn.qkv",
109
+ "model.visual.blocks.19.attn.proj",
110
+ "model.visual.blocks.19.mlp.linear_fc1",
111
+ "model.visual.blocks.19.mlp.linear_fc2",
112
+ "model.visual.blocks.20.attn.qkv",
113
+ "model.visual.blocks.20.attn.proj",
114
+ "model.visual.blocks.20.mlp.linear_fc1",
115
+ "model.visual.blocks.20.mlp.linear_fc2",
116
+ "model.visual.blocks.21.attn.qkv",
117
+ "model.visual.blocks.21.attn.proj",
118
+ "model.visual.blocks.21.mlp.linear_fc1",
119
+ "model.visual.blocks.21.mlp.linear_fc2",
120
+ "model.visual.blocks.22.attn.qkv",
121
+ "model.visual.blocks.22.attn.proj",
122
+ "model.visual.blocks.22.mlp.linear_fc1",
123
+ "model.visual.blocks.22.mlp.linear_fc2",
124
+ "model.visual.blocks.23.attn.qkv",
125
+ "model.visual.blocks.23.attn.proj",
126
+ "model.visual.blocks.23.mlp.linear_fc1",
127
+ "model.visual.blocks.23.mlp.linear_fc2",
128
+ "model.visual.blocks.24.attn.qkv",
129
+ "model.visual.blocks.24.attn.proj",
130
+ "model.visual.blocks.24.mlp.linear_fc1",
131
+ "model.visual.blocks.24.mlp.linear_fc2",
132
+ "model.visual.blocks.25.attn.qkv",
133
+ "model.visual.blocks.25.attn.proj",
134
+ "model.visual.blocks.25.mlp.linear_fc1",
135
+ "model.visual.blocks.25.mlp.linear_fc2",
136
+ "model.visual.blocks.26.attn.qkv",
137
+ "model.visual.blocks.26.attn.proj",
138
+ "model.visual.blocks.26.mlp.linear_fc1",
139
+ "model.visual.blocks.26.mlp.linear_fc2",
140
+ "model.visual.merger.linear_fc1",
141
+ "model.visual.merger.linear_fc2",
142
+ "model.language_model.layers.0.linear_attn.in_proj_b",
143
+ "model.language_model.layers.0.linear_attn.in_proj_a",
144
+ "model.language_model.layers.1.linear_attn.in_proj_b",
145
+ "model.language_model.layers.1.linear_attn.in_proj_a",
146
+ "model.language_model.layers.2.linear_attn.in_proj_b",
147
+ "model.language_model.layers.2.linear_attn.in_proj_a",
148
+ "model.language_model.layers.4.linear_attn.in_proj_b",
149
+ "model.language_model.layers.4.linear_attn.in_proj_a",
150
+ "model.language_model.layers.5.linear_attn.in_proj_b",
151
+ "model.language_model.layers.5.linear_attn.in_proj_a",
152
+ "model.language_model.layers.6.linear_attn.in_proj_b",
153
+ "model.language_model.layers.6.linear_attn.in_proj_a",
154
+ "model.language_model.layers.8.linear_attn.in_proj_b",
155
+ "model.language_model.layers.8.linear_attn.in_proj_a",
156
+ "model.language_model.layers.9.linear_attn.in_proj_b",
157
+ "model.language_model.layers.9.linear_attn.in_proj_a",
158
+ "model.language_model.layers.10.linear_attn.in_proj_b",
159
+ "model.language_model.layers.10.linear_attn.in_proj_a",
160
+ "model.language_model.layers.12.linear_attn.in_proj_b",
161
+ "model.language_model.layers.12.linear_attn.in_proj_a",
162
+ "model.language_model.layers.13.linear_attn.in_proj_b",
163
+ "model.language_model.layers.13.linear_attn.in_proj_a",
164
+ "model.language_model.layers.14.linear_attn.in_proj_b",
165
+ "model.language_model.layers.14.linear_attn.in_proj_a",
166
+ "model.language_model.layers.16.linear_attn.in_proj_b",
167
+ "model.language_model.layers.16.linear_attn.in_proj_a",
168
+ "model.language_model.layers.17.linear_attn.in_proj_b",
169
+ "model.language_model.layers.17.linear_attn.in_proj_a",
170
+ "model.language_model.layers.18.linear_attn.in_proj_b",
171
+ "model.language_model.layers.18.linear_attn.in_proj_a",
172
+ "model.language_model.layers.20.linear_attn.in_proj_b",
173
+ "model.language_model.layers.20.linear_attn.in_proj_a",
174
+ "model.language_model.layers.21.linear_attn.in_proj_b",
175
+ "model.language_model.layers.21.linear_attn.in_proj_a",
176
+ "model.language_model.layers.22.linear_attn.in_proj_b",
177
+ "model.language_model.layers.22.linear_attn.in_proj_a",
178
+ "model.language_model.layers.24.linear_attn.in_proj_b",
179
+ "model.language_model.layers.24.linear_attn.in_proj_a",
180
+ "model.language_model.layers.25.linear_attn.in_proj_b",
181
+ "model.language_model.layers.25.linear_attn.in_proj_a",
182
+ "model.language_model.layers.26.linear_attn.in_proj_b",
183
+ "model.language_model.layers.26.linear_attn.in_proj_a",
184
+ "model.language_model.layers.28.linear_attn.in_proj_b",
185
+ "model.language_model.layers.28.linear_attn.in_proj_a",
186
+ "model.language_model.layers.29.linear_attn.in_proj_b",
187
+ "model.language_model.layers.29.linear_attn.in_proj_a",
188
+ "model.language_model.layers.30.linear_attn.in_proj_b",
189
+ "model.language_model.layers.30.linear_attn.in_proj_a",
190
+ "model.language_model.layers.32.linear_attn.in_proj_b",
191
+ "model.language_model.layers.32.linear_attn.in_proj_a",
192
+ "model.language_model.layers.33.linear_attn.in_proj_b",
193
+ "model.language_model.layers.33.linear_attn.in_proj_a",
194
+ "model.language_model.layers.34.linear_attn.in_proj_b",
195
+ "model.language_model.layers.34.linear_attn.in_proj_a",
196
+ "model.language_model.layers.36.linear_attn.in_proj_b",
197
+ "model.language_model.layers.36.linear_attn.in_proj_a",
198
+ "model.language_model.layers.37.linear_attn.in_proj_b",
199
+ "model.language_model.layers.37.linear_attn.in_proj_a",
200
+ "model.language_model.layers.38.linear_attn.in_proj_b",
201
+ "model.language_model.layers.38.linear_attn.in_proj_a",
202
+ "model.language_model.layers.40.linear_attn.in_proj_b",
203
+ "model.language_model.layers.40.linear_attn.in_proj_a",
204
+ "model.language_model.layers.41.linear_attn.in_proj_b",
205
+ "model.language_model.layers.41.linear_attn.in_proj_a",
206
+ "model.language_model.layers.42.linear_attn.in_proj_b",
207
+ "model.language_model.layers.42.linear_attn.in_proj_a",
208
+ "model.language_model.layers.44.linear_attn.in_proj_b",
209
+ "model.language_model.layers.44.linear_attn.in_proj_a",
210
+ "model.language_model.layers.45.linear_attn.in_proj_b",
211
+ "model.language_model.layers.45.linear_attn.in_proj_a",
212
+ "model.language_model.layers.46.linear_attn.in_proj_b",
213
+ "model.language_model.layers.46.linear_attn.in_proj_a",
214
+ "model.language_model.layers.48.linear_attn.in_proj_b",
215
+ "model.language_model.layers.48.linear_attn.in_proj_a",
216
+ "model.language_model.layers.49.linear_attn.in_proj_b",
217
+ "model.language_model.layers.49.linear_attn.in_proj_a",
218
+ "model.language_model.layers.50.linear_attn.in_proj_b",
219
+ "model.language_model.layers.50.linear_attn.in_proj_a",
220
+ "model.language_model.layers.52.linear_attn.in_proj_b",
221
+ "model.language_model.layers.52.linear_attn.in_proj_a",
222
+ "model.language_model.layers.53.linear_attn.in_proj_b",
223
+ "model.language_model.layers.53.linear_attn.in_proj_a",
224
+ "model.language_model.layers.54.linear_attn.in_proj_b",
225
+ "model.language_model.layers.54.linear_attn.in_proj_a",
226
+ "model.language_model.layers.56.linear_attn.in_proj_b",
227
+ "model.language_model.layers.56.linear_attn.in_proj_a",
228
+ "model.language_model.layers.57.linear_attn.in_proj_b",
229
+ "model.language_model.layers.57.linear_attn.in_proj_a",
230
+ "model.language_model.layers.58.linear_attn.in_proj_b",
231
+ "model.language_model.layers.58.linear_attn.in_proj_a",
232
+ "model.language_model.layers.60.linear_attn.in_proj_b",
233
+ "model.language_model.layers.60.linear_attn.in_proj_a",
234
+ "model.language_model.layers.61.linear_attn.in_proj_b",
235
+ "model.language_model.layers.61.linear_attn.in_proj_a",
236
+ "model.language_model.layers.62.linear_attn.in_proj_b",
237
+ "model.language_model.layers.62.linear_attn.in_proj_a",
238
+ "lm_head"
239
+ ]
240
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523
3
+ size 19989325
tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": false,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "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+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }