Hugging Face's logo

GazTrab
/
Holo-3.1-35B-A3B-oQ8-mtp

Image-Text-to-Text
MLX
Safetensors
qwen3_5_moe
omlx
oq
quantized
mtp
speculative-decoding
grounding
computer-use
gui
conversational
8-bit precision
Model card Files
xet
 Community
GazTrab commited on
Commit
e39b174
·
verified ·
1 Parent(s): a49cd8b

Add files using upload-large-folder tool

Browse files
Files changed (19) hide show
  1. .gitattributes +1 -0
  2. README.md +90 -0
  3. chat_template.jinja +154 -0
  4. config.json +2765 -0
  5. generation_config.json +13 -0
  6. merges.txt +0 -0
  7. model-00001-of-00008.safetensors +3 -0
  8. model-00002-of-00008.safetensors +3 -0
  9. model-00003-of-00008.safetensors +3 -0
  10. model-00004-of-00008.safetensors +3 -0
  11. model-00005-of-00008.safetensors +3 -0
  12. model-00006-of-00008.safetensors +3 -0
  13. model-00007-of-00008.safetensors +3 -0
  14. model-00008-of-00008.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. 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
README.md ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: mlx
4
+ pipeline_tag: image-text-to-text
5
+ base_model:
6
+ - Hcompany/Holo-3.1-35B-A3B
7
+ - symrex/Holo-3.1-35B-A3B-oQ8
8
+ - Qwen/Qwen3.6-35B-A3B
9
+ tags:
10
+ - mlx
11
+ - omlx
12
+ - oq
13
+ - quantized
14
+ - mtp
15
+ - speculative-decoding
16
+ - grounding
17
+ - computer-use
18
+ - gui
19
+ - qwen3_5_moe
20
+ ---
21
+
22
+ # Holo-3.1-35B-A3B-oQ8-mtp
23
+
24
+ 8-bit (oMLX `oQ8`) MLX build of **[Hcompany/Holo-3.1-35B-A3B](https://huggingface.co/Hcompany/Holo-3.1-35B-A3B)** — H Company's GUI-grounding / computer-use VLM — **with a Multi-Token-Prediction (MTP / nextn) head grafted in** so [oMLX](https://github.com/jundot/omlx) can run native speculative decoding.
25
+
26
+ It is [`symrex/Holo-3.1-35B-A3B-oQ8`](https://huggingface.co/symrex/Holo-3.1-35B-A3B-oQ8) (which dropped the MTP head) with the 8-bit MTP module transplanted from [`tfjack/Qwen3.6-35B-A3B-oQ8-fp16-mtp`](https://huggingface.co/tfjack/Qwen3.6-35B-A3B-oQ8-fp16-mtp) — the same `Qwen3.6-35B-A3B` base Holo-3.1 was fine-tuned from.
27
+
28
+ 📦 **Code, scripts, and full write-up:** https://github.com/gaztrabisme/holo-3.1-a3b-mtp-mlx
29
+
30
+ ## Why a graft?
31
+
32
+ **No Holo-3.1 release ships an MTP head** — H Company stripped the nextn layer when fine-tuning (verified across bf16, oQ8, GGUF, NVFP4). The MTP head therefore can only come from the `Qwen3.6-35B-A3B` base. Both models share a **byte-identical non-MTP tensor set** (2010 tensors, `model_type: qwen3_5_moe`, hidden 2048, 40 layers, 256 experts, vocab 248320), so the 42-tensor `language_model.mtp.*` block (already 8-bit, `group_size` matched) drops in with no transforms. Config gains `text_config.mtp_num_hidden_layers=1`, `mtp_use_dedicated_embeddings=False`, and 6 MTP quant-overrides.
33
+
34
+ ## Verified on M3 Max (96 GB), single-stream, greedy
35
+
36
+ | | base oQ8 | **this (oQ8-mtp)** |
37
+ |---|---|---|
38
+ | Decode (300 tok) | 55.0 tok/s | **60.5 tok/s (~1.1×)** |
39
+ | MTP acceptance (structured text) | — | **94.8%** (145/153) |
40
+ | Output vs base | — | **byte-identical (lossless)** |
41
+
42
+ **Lossless by construction:** speculative decoding verifies every drafted token against the main model, so output is *exactly* Holo's greedy output — confirmed byte-identical on text (sha1 match) and on all 12 grounding targets below. The grafted head only affects *speed*, never correctness.
43
+
44
+ ### Grounding spot-check (synthetic 1280×720 UI, 12 known targets)
45
+
46
+ **12/12 hit, median 6 px error, max 55 px.** Base and MTP returned identical coordinates on all 12. (MTP acceptance on grounding is lower, 33–60%, because outputs are tiny ~15-token `{x,y}` blobs; for pure grounding, screenshot *prefill* dominates latency and MTP barely moves it — it pays off on longer agentic/navigation traces.)
47
+
48
+ ## Usage (oMLX, grounding)
49
+
50
+ Holo-3.1 grounding = single user turn, **normalized `[0,1000]` JSON `{x,y}`** (not Holo1's pixel `Click(x,y)`). Use **greedy + thinking off**; greedy is both the correct setting for stable coordinates *and* what maximizes MTP acceptance.
51
+
52
+ ```python
53
+ from openai import OpenAI
54
+ client = OpenAI(base_url="http://127.0.0.1:8000/v1", api_key="...")
55
+
56
+ W, H = 1280, 720 # send a smart_resize'd image; scale coords against THESE dims
57
+ prompt = (
58
+ "Localize an element on the GUI image according to the provided target "
59
+ "and output a click position.\n"
60
+ " * You must output a valid JSON following the format: "
61
+ '{"properties":{"x":{"type":"integer"},"y":{"type":"integer"}},"required":["x","y"]}\n'
62
+ " Your target is:\nthe 'Save Changes' button")
63
+
64
+ r = client.chat.completions.create(
65
+ model="Holo-3.1-35B-A3B-oQ8-mtp", temperature=0,
66
+ messages=[{"role": "user", "content": [
67
+ {"type": "image_url", "image_url": {"url": "data:image/png;base64,..."}},
68
+ {"type": "text", "text": prompt}]}],
69
+ extra_body={
70
+ "chat_template_kwargs": {"enable_thinking": False},
71
+ "response_format": {"type": "json_schema", "json_schema": {"name": "point",
72
+ "schema": {"type": "object", "required": ["x", "y"], "additionalProperties": False,
73
+ "properties": {"x": {"type": "integer"}, "y": {"type": "integer"}}}}},
74
+ })
75
+ import json
76
+ p = json.loads(r.choices[0].message.content)
77
+ px, py = int(p["x"]/1000*W), int(p["y"]/1000*H) # -> pixel click point
78
+ ```
79
+
80
+ > **smart_resize the screenshot first** (Qwen factor = patch 16 × merge 2 = 32; min 65 536 / max 16.7 M px) so the server's internal resize doesn't shift coordinates, and scale against the dimensions you actually send.
81
+
82
+ ## Notes
83
+
84
+ - MTP is auto-detected by oMLX from `mtp_num_hidden_layers`; the log prints `MTP path activated ... accept=N/M`.
85
+ - 36 GB resident. Two 38 GB models won't co-reside under oMLX's ~70 GB prefill guard — its LRU auto-evicts the idle one.
86
+ - Use the served model for both grounding and agentic/navigation modes; MTP helps the latter more.
87
+
88
+ ## Attribution
89
+
90
+ Apache-2.0. Built on the work of **H Company** (Holo-3.1), **Qwen** (Qwen3.6-35B-A3B base + MTP head), **symrex** (oQ8 quant), and **tfjack** (Qwen3.6 oQ8 MTP MLX conversion). Quantization tooling: **oMLX / oQ**.
chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\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>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if (preserve_thinking is defined and preserve_thinking is true) or (loop.index0 > ns.last_query_index) %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,2765 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5MoeForConditionalGeneration"
4
+ ],
5
+ "dtype": "float32",
6
+ "image_token_id": 248056,
7
+ "model_type": "qwen3_5_moe",
8
+ "text_config": {
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_output_gate": true,
12
+ "bos_token_id": 248044,
13
+ "dtype": "float32",
14
+ "eos_token_id": 248044,
15
+ "full_attention_interval": 4,
16
+ "head_dim": 256,
17
+ "hidden_act": "silu",
18
+ "hidden_size": 2048,
19
+ "initializer_range": 0.02,
20
+ "layer_types": [
21
+ "linear_attention",
22
+ "linear_attention",
23
+ "linear_attention",
24
+ "full_attention",
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention",
57
+ "linear_attention",
58
+ "linear_attention",
59
+ "linear_attention",
60
+ "full_attention"
61
+ ],
62
+ "linear_conv_kernel_dim": 4,
63
+ "linear_key_head_dim": 128,
64
+ "linear_num_key_heads": 16,
65
+ "linear_num_value_heads": 32,
66
+ "linear_value_head_dim": 128,
67
+ "mamba_ssm_dtype": "float32",
68
+ "max_position_embeddings": 262144,
69
+ "model_type": "qwen3_5_moe_text",
70
+ "moe_intermediate_size": 512,
71
+ "num_attention_heads": 16,
72
+ "num_experts": 256,
73
+ "num_experts_per_tok": 8,
74
+ "num_hidden_layers": 40,
75
+ "num_key_value_heads": 2,
76
+ "output_router_logits": false,
77
+ "pad_token_id": null,
78
+ "partial_rotary_factor": 0.25,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_parameters": {
81
+ "mrope_interleaved": true,
82
+ "mrope_section": [
83
+ 11,
84
+ 11,
85
+ 10
86
+ ],
87
+ "partial_rotary_factor": 0.25,
88
+ "rope_theta": 10000000,
89
+ "type": "default"
90
+ },
91
+ "router_aux_loss_coef": 0.001,
92
+ "shared_expert_intermediate_size": 512,
93
+ "tie_word_embeddings": false,
94
+ "use_cache": true,
95
+ "vocab_size": 248320,
96
+ "mtp_num_hidden_layers": 1,
97
+ "mtp_use_dedicated_embeddings": false
98
+ },
99
+ "tie_word_embeddings": false,
100
+ "transformers_version": "5.6.2",
101
+ "video_token_id": 248057,
102
+ "vision_config": {
103
+ "deepstack_visual_indexes": [],
104
+ "depth": 27,
105
+ "dtype": "float32",
106
+ "hidden_act": "gelu_pytorch_tanh",
107
+ "hidden_size": 1152,
108
+ "in_channels": 3,
109
+ "initializer_range": 0.02,
110
+ "intermediate_size": 4304,
111
+ "model_type": "qwen3_5_moe_vision",
112
+ "num_heads": 16,
113
+ "num_position_embeddings": 2304,
114
+ "out_hidden_size": 2048,
115
+ "patch_size": 16,
116
+ "spatial_merge_size": 2,
117
+ "temporal_patch_size": 2
118
+ },
119
+ "vision_end_token_id": 248054,
120
+ "vision_start_token_id": 248053,
121
+ "eos_token_id": [
122
+ 248046,
123
+ 248044
124
+ ],
125
+ "quantization": {
126
+ "group_size": 64,
127
+ "bits": 8,
128
+ "mode": "affine",
129
+ "language_model.model.layers.0.linear_attn.in_proj_qkv": {
130
+ "bits": 8,
131
+ "group_size": 128,
132
+ "mode": "affine"
133
+ },
134
+ "language_model.model.layers.0.linear_attn.in_proj_z": {
135
+ "bits": 8,
136
+ "group_size": 128,
137
+ "mode": "affine"
138
+ },
139
+ "language_model.model.layers.0.linear_attn.out_proj": {
140
+ "bits": 8,
141
+ "group_size": 128,
142
+ "mode": "affine"
143
+ },
144
+ "language_model.model.layers.1.linear_attn.in_proj_qkv": {
145
+ "bits": 8,
146
+ "group_size": 128,
147
+ "mode": "affine"
148
+ },
149
+ "language_model.model.layers.1.linear_attn.in_proj_z": {
150
+ "bits": 8,
151
+ "group_size": 128,
152
+ "mode": "affine"
153
+ },
154
+ "language_model.model.layers.1.linear_attn.out_proj": {
155
+ "bits": 8,
156
+ "group_size": 128,
157
+ "mode": "affine"
158
+ },
159
+ "language_model.model.layers.0.linear_attn.in_proj_a": {
160
+ "bits": 8,
161
+ "group_size": 128,
162
+ "mode": "affine"
163
+ },
164
+ "language_model.model.layers.0.linear_attn.in_proj_b": {
165
+ "bits": 8,
166
+ "group_size": 128,
167
+ "mode": "affine"
168
+ },
169
+ "language_model.model.layers.0.mlp.shared_expert.down_proj": {
170
+ "bits": 8,
171
+ "group_size": 128,
172
+ "mode": "affine"
173
+ },
174
+ "language_model.model.layers.0.mlp.shared_expert.gate_proj": {
175
+ "bits": 8,
176
+ "group_size": 128,
177
+ "mode": "affine"
178
+ },
179
+ "language_model.model.layers.0.mlp.shared_expert.up_proj": {
180
+ "bits": 8,
181
+ "group_size": 128,
182
+ "mode": "affine"
183
+ },
184
+ "language_model.model.layers.1.linear_attn.in_proj_a": {
185
+ "bits": 8,
186
+ "group_size": 128,
187
+ "mode": "affine"
188
+ },
189
+ "language_model.model.layers.1.linear_attn.in_proj_b": {
190
+ "bits": 8,
191
+ "group_size": 128,
192
+ "mode": "affine"
193
+ },
194
+ "language_model.model.layers.1.mlp.shared_expert.down_proj": {
195
+ "bits": 8,
196
+ "group_size": 128,
197
+ "mode": "affine"
198
+ },
199
+ "language_model.model.layers.1.mlp.shared_expert.gate_proj": {
200
+ "bits": 8,
201
+ "group_size": 128,
202
+ "mode": "affine"
203
+ },
204
+ "language_model.model.layers.1.mlp.shared_expert.up_proj": {
205
+ "bits": 8,
206
+ "group_size": 128,
207
+ "mode": "affine"
208
+ },
209
+ "language_model.model.layers.2.linear_attn.in_proj_a": {
210
+ "bits": 8,
211
+ "group_size": 128,
212
+ "mode": "affine"
213
+ },
214
+ "language_model.model.layers.2.linear_attn.in_proj_b": {
215
+ "bits": 8,
216
+ "group_size": 128,
217
+ "mode": "affine"
218
+ },
219
+ "language_model.model.layers.2.linear_attn.in_proj_z": {
220
+ "bits": 8,
221
+ "group_size": 128,
222
+ "mode": "affine"
223
+ },
224
+ "language_model.model.layers.2.linear_attn.out_proj": {
225
+ "bits": 8,
226
+ "group_size": 128,
227
+ "mode": "affine"
228
+ },
229
+ "language_model.model.layers.2.mlp.shared_expert.down_proj": {
230
+ "bits": 8,
231
+ "group_size": 128,
232
+ "mode": "affine"
233
+ },
234
+ "language_model.model.layers.2.mlp.shared_expert.gate_proj": {
235
+ "bits": 8,
236
+ "group_size": 128,
237
+ "mode": "affine"
238
+ },
239
+ "language_model.model.layers.2.mlp.shared_expert.up_proj": {
240
+ "bits": 8,
241
+ "group_size": 128,
242
+ "mode": "affine"
243
+ },
244
+ "language_model.model.layers.3.mlp.shared_expert.down_proj": {
245
+ "bits": 8,
246
+ "group_size": 128,
247
+ "mode": "affine"
248
+ },
249
+ "language_model.model.layers.3.mlp.shared_expert.gate_proj": {
250
+ "bits": 8,
251
+ "group_size": 128,
252
+ "mode": "affine"
253
+ },
254
+ "language_model.model.layers.3.mlp.shared_expert.up_proj": {
255
+ "bits": 8,
256
+ "group_size": 128,
257
+ "mode": "affine"
258
+ },
259
+ "language_model.model.layers.4.linear_attn.in_proj_a": {
260
+ "bits": 8,
261
+ "group_size": 128,
262
+ "mode": "affine"
263
+ },
264
+ "language_model.model.layers.4.linear_attn.in_proj_b": {
265
+ "bits": 8,
266
+ "group_size": 128,
267
+ "mode": "affine"
268
+ },
269
+ "language_model.model.layers.4.linear_attn.in_proj_z": {
270
+ "bits": 8,
271
+ "group_size": 128,
272
+ "mode": "affine"
273
+ },
274
+ "language_model.model.layers.4.linear_attn.out_proj": {
275
+ "bits": 8,
276
+ "group_size": 128,
277
+ "mode": "affine"
278
+ },
279
+ "language_model.model.layers.4.mlp.shared_expert.down_proj": {
280
+ "bits": 8,
281
+ "group_size": 128,
282
+ "mode": "affine"
283
+ },
284
+ "language_model.model.layers.4.mlp.shared_expert.gate_proj": {
285
+ "bits": 8,
286
+ "group_size": 128,
287
+ "mode": "affine"
288
+ },
289
+ "language_model.model.layers.4.mlp.shared_expert.up_proj": {
290
+ "bits": 8,
291
+ "group_size": 128,
292
+ "mode": "affine"
293
+ },
294
+ "language_model.model.layers.5.linear_attn.in_proj_a": {
295
+ "bits": 8,
296
+ "group_size": 128,
297
+ "mode": "affine"
298
+ },
299
+ "language_model.model.layers.5.linear_attn.in_proj_b": {
300
+ "bits": 8,
301
+ "group_size": 128,
302
+ "mode": "affine"
303
+ },
304
+ "language_model.model.layers.5.linear_attn.in_proj_z": {
305
+ "bits": 8,
306
+ "group_size": 128,
307
+ "mode": "affine"
308
+ },
309
+ "language_model.model.layers.5.linear_attn.out_proj": {
310
+ "bits": 8,
311
+ "group_size": 128,
312
+ "mode": "affine"
313
+ },
314
+ "language_model.model.layers.5.mlp.shared_expert.down_proj": {
315
+ "bits": 8,
316
+ "group_size": 128,
317
+ "mode": "affine"
318
+ },
319
+ "language_model.model.layers.5.mlp.shared_expert.gate_proj": {
320
+ "bits": 8,
321
+ "group_size": 128,
322
+ "mode": "affine"
323
+ },
324
+ "language_model.model.layers.5.mlp.shared_expert.up_proj": {
325
+ "bits": 8,
326
+ "group_size": 128,
327
+ "mode": "affine"
328
+ },
329
+ "language_model.model.layers.6.linear_attn.in_proj_a": {
330
+ "bits": 8,
331
+ "group_size": 128,
332
+ "mode": "affine"
333
+ },
334
+ "language_model.model.layers.6.linear_attn.in_proj_b": {
335
+ "bits": 8,
336
+ "group_size": 128,
337
+ "mode": "affine"
338
+ },
339
+ "language_model.model.layers.6.linear_attn.in_proj_z": {
340
+ "bits": 8,
341
+ "group_size": 128,
342
+ "mode": "affine"
343
+ },
344
+ "language_model.model.layers.6.linear_attn.out_proj": {
345
+ "bits": 8,
346
+ "group_size": 128,
347
+ "mode": "affine"
348
+ },
349
+ "language_model.model.layers.6.mlp.shared_expert.down_proj": {
350
+ "bits": 8,
351
+ "group_size": 128,
352
+ "mode": "affine"
353
+ },
354
+ "language_model.model.layers.6.mlp.shared_expert.gate_proj": {
355
+ "bits": 8,
356
+ "group_size": 128,
357
+ "mode": "affine"
358
+ },
359
+ "language_model.model.layers.6.mlp.shared_expert.up_proj": {
360
+ "bits": 8,
361
+ "group_size": 128,
362
+ "mode": "affine"
363
+ },
364
+ "language_model.model.layers.7.mlp.shared_expert.down_proj": {
365
+ "bits": 8,
366
+ "group_size": 128,
367
+ "mode": "affine"
368
+ },
369
+ "language_model.model.layers.7.mlp.shared_expert.gate_proj": {
370
+ "bits": 8,
371
+ "group_size": 128,
372
+ "mode": "affine"
373
+ },
374
+ "language_model.model.layers.7.mlp.shared_expert.up_proj": {
375
+ "bits": 8,
376
+ "group_size": 128,
377
+ "mode": "affine"
378
+ },
379
+ "language_model.model.layers.10.linear_attn.in_proj_z": {
380
+ "bits": 8,
381
+ "group_size": 128,
382
+ "mode": "affine"
383
+ },
384
+ "language_model.model.layers.10.linear_attn.out_proj": {
385
+ "bits": 8,
386
+ "group_size": 128,
387
+ "mode": "affine"
388
+ },
389
+ "language_model.model.layers.8.linear_attn.in_proj_z": {
390
+ "bits": 8,
391
+ "group_size": 128,
392
+ "mode": "affine"
393
+ },
394
+ "language_model.model.layers.8.linear_attn.out_proj": {
395
+ "bits": 8,
396
+ "group_size": 128,
397
+ "mode": "affine"
398
+ },
399
+ "language_model.model.layers.9.linear_attn.in_proj_z": {
400
+ "bits": 8,
401
+ "group_size": 128,
402
+ "mode": "affine"
403
+ },
404
+ "language_model.model.layers.9.linear_attn.out_proj": {
405
+ "bits": 8,
406
+ "group_size": 128,
407
+ "mode": "affine"
408
+ },
409
+ "language_model.model.layers.10.linear_attn.in_proj_a": {
410
+ "bits": 8,
411
+ "group_size": 128,
412
+ "mode": "affine"
413
+ },
414
+ "language_model.model.layers.10.linear_attn.in_proj_b": {
415
+ "bits": 8,
416
+ "group_size": 128,
417
+ "mode": "affine"
418
+ },
419
+ "language_model.model.layers.10.mlp.shared_expert.down_proj": {
420
+ "bits": 8,
421
+ "group_size": 128,
422
+ "mode": "affine"
423
+ },
424
+ "language_model.model.layers.10.mlp.shared_expert.gate_proj": {
425
+ "bits": 8,
426
+ "group_size": 128,
427
+ "mode": "affine"
428
+ },
429
+ "language_model.model.layers.10.mlp.shared_expert.up_proj": {
430
+ "bits": 8,
431
+ "group_size": 128,
432
+ "mode": "affine"
433
+ },
434
+ "language_model.model.layers.8.linear_attn.in_proj_a": {
435
+ "bits": 8,
436
+ "group_size": 128,
437
+ "mode": "affine"
438
+ },
439
+ "language_model.model.layers.8.linear_attn.in_proj_b": {
440
+ "bits": 8,
441
+ "group_size": 128,
442
+ "mode": "affine"
443
+ },
444
+ "language_model.model.layers.8.mlp.shared_expert.down_proj": {
445
+ "bits": 8,
446
+ "group_size": 128,
447
+ "mode": "affine"
448
+ },
449
+ "language_model.model.layers.8.mlp.shared_expert.gate_proj": {
450
+ "bits": 8,
451
+ "group_size": 128,
452
+ "mode": "affine"
453
+ },
454
+ "language_model.model.layers.8.mlp.shared_expert.up_proj": {
455
+ "bits": 8,
456
+ "group_size": 128,
457
+ "mode": "affine"
458
+ },
459
+ "language_model.model.layers.9.linear_attn.in_proj_a": {
460
+ "bits": 8,
461
+ "group_size": 128,
462
+ "mode": "affine"
463
+ },
464
+ "language_model.model.layers.9.linear_attn.in_proj_b": {
465
+ "bits": 8,
466
+ "group_size": 128,
467
+ "mode": "affine"
468
+ },
469
+ "language_model.model.layers.9.mlp.shared_expert.down_proj": {
470
+ "bits": 8,
471
+ "group_size": 128,
472
+ "mode": "affine"
473
+ },
474
+ "language_model.model.layers.9.mlp.shared_expert.gate_proj": {
475
+ "bits": 8,
476
+ "group_size": 128,
477
+ "mode": "affine"
478
+ },
479
+ "language_model.model.layers.9.mlp.shared_expert.up_proj": {
480
+ "bits": 8,
481
+ "group_size": 128,
482
+ "mode": "affine"
483
+ },
484
+ "language_model.model.layers.12.linear_attn.in_proj_z": {
485
+ "bits": 8,
486
+ "group_size": 128,
487
+ "mode": "affine"
488
+ },
489
+ "language_model.model.layers.12.linear_attn.out_proj": {
490
+ "bits": 8,
491
+ "group_size": 128,
492
+ "mode": "affine"
493
+ },
494
+ "language_model.model.layers.13.linear_attn.in_proj_z": {
495
+ "bits": 8,
496
+ "group_size": 128,
497
+ "mode": "affine"
498
+ },
499
+ "language_model.model.layers.13.linear_attn.out_proj": {
500
+ "bits": 8,
501
+ "group_size": 128,
502
+ "mode": "affine"
503
+ },
504
+ "language_model.model.layers.11.mlp.shared_expert.down_proj": {
505
+ "bits": 8,
506
+ "group_size": 128,
507
+ "mode": "affine"
508
+ },
509
+ "language_model.model.layers.11.mlp.shared_expert.gate_proj": {
510
+ "bits": 8,
511
+ "group_size": 128,
512
+ "mode": "affine"
513
+ },
514
+ "language_model.model.layers.11.mlp.shared_expert.up_proj": {
515
+ "bits": 8,
516
+ "group_size": 128,
517
+ "mode": "affine"
518
+ },
519
+ "language_model.model.layers.12.linear_attn.in_proj_a": {
520
+ "bits": 8,
521
+ "group_size": 128,
522
+ "mode": "affine"
523
+ },
524
+ "language_model.model.layers.12.linear_attn.in_proj_b": {
525
+ "bits": 8,
526
+ "group_size": 128,
527
+ "mode": "affine"
528
+ },
529
+ "language_model.model.layers.12.mlp.shared_expert.down_proj": {
530
+ "bits": 8,
531
+ "group_size": 128,
532
+ "mode": "affine"
533
+ },
534
+ "language_model.model.layers.12.mlp.shared_expert.gate_proj": {
535
+ "bits": 8,
536
+ "group_size": 128,
537
+ "mode": "affine"
538
+ },
539
+ "language_model.model.layers.12.mlp.shared_expert.up_proj": {
540
+ "bits": 8,
541
+ "group_size": 128,
542
+ "mode": "affine"
543
+ },
544
+ "language_model.model.layers.13.linear_attn.in_proj_a": {
545
+ "bits": 8,
546
+ "group_size": 128,
547
+ "mode": "affine"
548
+ },
549
+ "language_model.model.layers.13.linear_attn.in_proj_b": {
550
+ "bits": 8,
551
+ "group_size": 128,
552
+ "mode": "affine"
553
+ },
554
+ "language_model.model.layers.13.mlp.shared_expert.down_proj": {
555
+ "bits": 8,
556
+ "group_size": 128,
557
+ "mode": "affine"
558
+ },
559
+ "language_model.model.layers.13.mlp.shared_expert.gate_proj": {
560
+ "bits": 8,
561
+ "group_size": 128,
562
+ "mode": "affine"
563
+ },
564
+ "language_model.model.layers.13.mlp.shared_expert.up_proj": {
565
+ "bits": 8,
566
+ "group_size": 128,
567
+ "mode": "affine"
568
+ },
569
+ "language_model.model.layers.14.linear_attn.in_proj_z": {
570
+ "bits": 8,
571
+ "group_size": 128,
572
+ "mode": "affine"
573
+ },
574
+ "language_model.model.layers.14.linear_attn.out_proj": {
575
+ "bits": 8,
576
+ "group_size": 128,
577
+ "mode": "affine"
578
+ },
579
+ "language_model.model.layers.16.linear_attn.in_proj_z": {
580
+ "bits": 8,
581
+ "group_size": 128,
582
+ "mode": "affine"
583
+ },
584
+ "language_model.model.layers.16.linear_attn.out_proj": {
585
+ "bits": 8,
586
+ "group_size": 128,
587
+ "mode": "affine"
588
+ },
589
+ "language_model.model.layers.14.linear_attn.in_proj_a": {
590
+ "bits": 8,
591
+ "group_size": 128,
592
+ "mode": "affine"
593
+ },
594
+ "language_model.model.layers.14.linear_attn.in_proj_b": {
595
+ "bits": 8,
596
+ "group_size": 128,
597
+ "mode": "affine"
598
+ },
599
+ "language_model.model.layers.14.mlp.shared_expert.down_proj": {
600
+ "bits": 8,
601
+ "group_size": 128,
602
+ "mode": "affine"
603
+ },
604
+ "language_model.model.layers.14.mlp.shared_expert.gate_proj": {
605
+ "bits": 8,
606
+ "group_size": 128,
607
+ "mode": "affine"
608
+ },
609
+ "language_model.model.layers.14.mlp.shared_expert.up_proj": {
610
+ "bits": 8,
611
+ "group_size": 128,
612
+ "mode": "affine"
613
+ },
614
+ "language_model.model.layers.15.mlp.shared_expert.down_proj": {
615
+ "bits": 8,
616
+ "group_size": 128,
617
+ "mode": "affine"
618
+ },
619
+ "language_model.model.layers.15.mlp.shared_expert.gate_proj": {
620
+ "bits": 8,
621
+ "group_size": 128,
622
+ "mode": "affine"
623
+ },
624
+ "language_model.model.layers.15.mlp.shared_expert.up_proj": {
625
+ "bits": 8,
626
+ "group_size": 128,
627
+ "mode": "affine"
628
+ },
629
+ "language_model.model.layers.16.linear_attn.in_proj_a": {
630
+ "bits": 8,
631
+ "group_size": 128,
632
+ "mode": "affine"
633
+ },
634
+ "language_model.model.layers.16.linear_attn.in_proj_b": {
635
+ "bits": 8,
636
+ "group_size": 128,
637
+ "mode": "affine"
638
+ },
639
+ "language_model.model.layers.16.mlp.shared_expert.down_proj": {
640
+ "bits": 8,
641
+ "group_size": 128,
642
+ "mode": "affine"
643
+ },
644
+ "language_model.model.layers.16.mlp.shared_expert.gate_proj": {
645
+ "bits": 8,
646
+ "group_size": 128,
647
+ "mode": "affine"
648
+ },
649
+ "language_model.model.layers.16.mlp.shared_expert.up_proj": {
650
+ "bits": 8,
651
+ "group_size": 128,
652
+ "mode": "affine"
653
+ },
654
+ "language_model.model.layers.17.linear_attn.in_proj_z": {
655
+ "bits": 8,
656
+ "group_size": 128,
657
+ "mode": "affine"
658
+ },
659
+ "language_model.model.layers.17.linear_attn.out_proj": {
660
+ "bits": 8,
661
+ "group_size": 128,
662
+ "mode": "affine"
663
+ },
664
+ "language_model.model.layers.18.linear_attn.in_proj_z": {
665
+ "bits": 8,
666
+ "group_size": 128,
667
+ "mode": "affine"
668
+ },
669
+ "language_model.model.layers.18.linear_attn.out_proj": {
670
+ "bits": 8,
671
+ "group_size": 128,
672
+ "mode": "affine"
673
+ },
674
+ "language_model.model.layers.17.linear_attn.in_proj_a": {
675
+ "bits": 8,
676
+ "group_size": 128,
677
+ "mode": "affine"
678
+ },
679
+ "language_model.model.layers.17.linear_attn.in_proj_b": {
680
+ "bits": 8,
681
+ "group_size": 128,
682
+ "mode": "affine"
683
+ },
684
+ "language_model.model.layers.17.mlp.shared_expert.down_proj": {
685
+ "bits": 8,
686
+ "group_size": 128,
687
+ "mode": "affine"
688
+ },
689
+ "language_model.model.layers.17.mlp.shared_expert.gate_proj": {
690
+ "bits": 8,
691
+ "group_size": 128,
692
+ "mode": "affine"
693
+ },
694
+ "language_model.model.layers.17.mlp.shared_expert.up_proj": {
695
+ "bits": 8,
696
+ "group_size": 128,
697
+ "mode": "affine"
698
+ },
699
+ "language_model.model.layers.18.linear_attn.in_proj_a": {
700
+ "bits": 8,
701
+ "group_size": 128,
702
+ "mode": "affine"
703
+ },
704
+ "language_model.model.layers.18.linear_attn.in_proj_b": {
705
+ "bits": 8,
706
+ "group_size": 128,
707
+ "mode": "affine"
708
+ },
709
+ "language_model.model.layers.18.mlp.shared_expert.down_proj": {
710
+ "bits": 8,
711
+ "group_size": 128,
712
+ "mode": "affine"
713
+ },
714
+ "language_model.model.layers.18.mlp.shared_expert.gate_proj": {
715
+ "bits": 8,
716
+ "group_size": 128,
717
+ "mode": "affine"
718
+ },
719
+ "language_model.model.layers.18.mlp.shared_expert.up_proj": {
720
+ "bits": 8,
721
+ "group_size": 128,
722
+ "mode": "affine"
723
+ },
724
+ "language_model.model.layers.19.mlp.shared_expert.down_proj": {
725
+ "bits": 8,
726
+ "group_size": 128,
727
+ "mode": "affine"
728
+ },
729
+ "language_model.model.layers.19.mlp.shared_expert.gate_proj": {
730
+ "bits": 8,
731
+ "group_size": 128,
732
+ "mode": "affine"
733
+ },
734
+ "language_model.model.layers.19.mlp.shared_expert.up_proj": {
735
+ "bits": 8,
736
+ "group_size": 128,
737
+ "mode": "affine"
738
+ },
739
+ "language_model.model.layers.20.linear_attn.in_proj_z": {
740
+ "bits": 8,
741
+ "group_size": 128,
742
+ "mode": "affine"
743
+ },
744
+ "language_model.model.layers.20.linear_attn.out_proj": {
745
+ "bits": 8,
746
+ "group_size": 128,
747
+ "mode": "affine"
748
+ },
749
+ "language_model.model.layers.21.linear_attn.in_proj_z": {
750
+ "bits": 8,
751
+ "group_size": 128,
752
+ "mode": "affine"
753
+ },
754
+ "language_model.model.layers.21.linear_attn.out_proj": {
755
+ "bits": 8,
756
+ "group_size": 128,
757
+ "mode": "affine"
758
+ },
759
+ "language_model.model.layers.22.linear_attn.in_proj_z": {
760
+ "bits": 8,
761
+ "group_size": 128,
762
+ "mode": "affine"
763
+ },
764
+ "language_model.model.layers.22.linear_attn.out_proj": {
765
+ "bits": 8,
766
+ "group_size": 128,
767
+ "mode": "affine"
768
+ },
769
+ "language_model.model.layers.20.linear_attn.in_proj_a": {
770
+ "bits": 8,
771
+ "group_size": 128,
772
+ "mode": "affine"
773
+ },
774
+ "language_model.model.layers.20.linear_attn.in_proj_b": {
775
+ "bits": 8,
776
+ "group_size": 128,
777
+ "mode": "affine"
778
+ },
779
+ "language_model.model.layers.20.mlp.shared_expert.down_proj": {
780
+ "bits": 8,
781
+ "group_size": 128,
782
+ "mode": "affine"
783
+ },
784
+ "language_model.model.layers.20.mlp.shared_expert.gate_proj": {
785
+ "bits": 8,
786
+ "group_size": 128,
787
+ "mode": "affine"
788
+ },
789
+ "language_model.model.layers.20.mlp.shared_expert.up_proj": {
790
+ "bits": 8,
791
+ "group_size": 128,
792
+ "mode": "affine"
793
+ },
794
+ "language_model.model.layers.21.linear_attn.in_proj_a": {
795
+ "bits": 8,
796
+ "group_size": 128,
797
+ "mode": "affine"
798
+ },
799
+ "language_model.model.layers.21.linear_attn.in_proj_b": {
800
+ "bits": 8,
801
+ "group_size": 128,
802
+ "mode": "affine"
803
+ },
804
+ "language_model.model.layers.21.mlp.shared_expert.down_proj": {
805
+ "bits": 8,
806
+ "group_size": 128,
807
+ "mode": "affine"
808
+ },
809
+ "language_model.model.layers.21.mlp.shared_expert.gate_proj": {
810
+ "bits": 8,
811
+ "group_size": 128,
812
+ "mode": "affine"
813
+ },
814
+ "language_model.model.layers.21.mlp.shared_expert.up_proj": {
815
+ "bits": 8,
816
+ "group_size": 128,
817
+ "mode": "affine"
818
+ },
819
+ "language_model.model.layers.22.linear_attn.in_proj_a": {
820
+ "bits": 8,
821
+ "group_size": 128,
822
+ "mode": "affine"
823
+ },
824
+ "language_model.model.layers.22.linear_attn.in_proj_b": {
825
+ "bits": 8,
826
+ "group_size": 128,
827
+ "mode": "affine"
828
+ },
829
+ "language_model.model.layers.22.mlp.shared_expert.down_proj": {
830
+ "bits": 8,
831
+ "group_size": 128,
832
+ "mode": "affine"
833
+ },
834
+ "language_model.model.layers.22.mlp.shared_expert.gate_proj": {
835
+ "bits": 8,
836
+ "group_size": 128,
837
+ "mode": "affine"
838
+ },
839
+ "language_model.model.layers.22.mlp.shared_expert.up_proj": {
840
+ "bits": 8,
841
+ "group_size": 128,
842
+ "mode": "affine"
843
+ },
844
+ "language_model.model.layers.23.self_attn.k_proj": {
845
+ "bits": 8,
846
+ "group_size": 128,
847
+ "mode": "affine"
848
+ },
849
+ "language_model.model.layers.23.self_attn.q_proj": {
850
+ "bits": 8,
851
+ "group_size": 128,
852
+ "mode": "affine"
853
+ },
854
+ "language_model.model.layers.23.self_attn.v_proj": {
855
+ "bits": 8,
856
+ "group_size": 128,
857
+ "mode": "affine"
858
+ },
859
+ "language_model.model.layers.24.linear_attn.in_proj_z": {
860
+ "bits": 8,
861
+ "group_size": 128,
862
+ "mode": "affine"
863
+ },
864
+ "language_model.model.layers.24.linear_attn.out_proj": {
865
+ "bits": 8,
866
+ "group_size": 128,
867
+ "mode": "affine"
868
+ },
869
+ "language_model.model.layers.25.linear_attn.in_proj_z": {
870
+ "bits": 8,
871
+ "group_size": 128,
872
+ "mode": "affine"
873
+ },
874
+ "language_model.model.layers.25.linear_attn.out_proj": {
875
+ "bits": 8,
876
+ "group_size": 128,
877
+ "mode": "affine"
878
+ },
879
+ "language_model.model.layers.23.mlp.shared_expert.down_proj": {
880
+ "bits": 8,
881
+ "group_size": 128,
882
+ "mode": "affine"
883
+ },
884
+ "language_model.model.layers.23.mlp.shared_expert.gate_proj": {
885
+ "bits": 8,
886
+ "group_size": 128,
887
+ "mode": "affine"
888
+ },
889
+ "language_model.model.layers.23.mlp.shared_expert.up_proj": {
890
+ "bits": 8,
891
+ "group_size": 128,
892
+ "mode": "affine"
893
+ },
894
+ "language_model.model.layers.24.linear_attn.in_proj_a": {
895
+ "bits": 8,
896
+ "group_size": 128,
897
+ "mode": "affine"
898
+ },
899
+ "language_model.model.layers.24.linear_attn.in_proj_b": {
900
+ "bits": 8,
901
+ "group_size": 128,
902
+ "mode": "affine"
903
+ },
904
+ "language_model.model.layers.24.mlp.shared_expert.down_proj": {
905
+ "bits": 8,
906
+ "group_size": 128,
907
+ "mode": "affine"
908
+ },
909
+ "language_model.model.layers.24.mlp.shared_expert.gate_proj": {
910
+ "bits": 8,
911
+ "group_size": 128,
912
+ "mode": "affine"
913
+ },
914
+ "language_model.model.layers.24.mlp.shared_expert.up_proj": {
915
+ "bits": 8,
916
+ "group_size": 128,
917
+ "mode": "affine"
918
+ },
919
+ "language_model.model.layers.25.linear_attn.in_proj_a": {
920
+ "bits": 8,
921
+ "group_size": 128,
922
+ "mode": "affine"
923
+ },
924
+ "language_model.model.layers.25.linear_attn.in_proj_b": {
925
+ "bits": 8,
926
+ "group_size": 128,
927
+ "mode": "affine"
928
+ },
929
+ "language_model.model.layers.25.mlp.shared_expert.down_proj": {
930
+ "bits": 8,
931
+ "group_size": 128,
932
+ "mode": "affine"
933
+ },
934
+ "language_model.model.layers.25.mlp.shared_expert.gate_proj": {
935
+ "bits": 8,
936
+ "group_size": 128,
937
+ "mode": "affine"
938
+ },
939
+ "language_model.model.layers.25.mlp.shared_expert.up_proj": {
940
+ "bits": 8,
941
+ "group_size": 128,
942
+ "mode": "affine"
943
+ },
944
+ "language_model.model.layers.26.linear_attn.in_proj_z": {
945
+ "bits": 8,
946
+ "group_size": 128,
947
+ "mode": "affine"
948
+ },
949
+ "language_model.model.layers.26.linear_attn.out_proj": {
950
+ "bits": 8,
951
+ "group_size": 128,
952
+ "mode": "affine"
953
+ },
954
+ "language_model.model.layers.28.linear_attn.in_proj_z": {
955
+ "bits": 8,
956
+ "group_size": 128,
957
+ "mode": "affine"
958
+ },
959
+ "language_model.model.layers.28.linear_attn.out_proj": {
960
+ "bits": 8,
961
+ "group_size": 128,
962
+ "mode": "affine"
963
+ },
964
+ "language_model.model.layers.26.linear_attn.in_proj_a": {
965
+ "bits": 8,
966
+ "group_size": 128,
967
+ "mode": "affine"
968
+ },
969
+ "language_model.model.layers.26.linear_attn.in_proj_b": {
970
+ "bits": 8,
971
+ "group_size": 128,
972
+ "mode": "affine"
973
+ },
974
+ "language_model.model.layers.26.mlp.shared_expert.down_proj": {
975
+ "bits": 8,
976
+ "group_size": 128,
977
+ "mode": "affine"
978
+ },
979
+ "language_model.model.layers.26.mlp.shared_expert.gate_proj": {
980
+ "bits": 8,
981
+ "group_size": 128,
982
+ "mode": "affine"
983
+ },
984
+ "language_model.model.layers.26.mlp.shared_expert.up_proj": {
985
+ "bits": 8,
986
+ "group_size": 128,
987
+ "mode": "affine"
988
+ },
989
+ "language_model.model.layers.27.mlp.shared_expert.down_proj": {
990
+ "bits": 8,
991
+ "group_size": 128,
992
+ "mode": "affine"
993
+ },
994
+ "language_model.model.layers.27.mlp.shared_expert.gate_proj": {
995
+ "bits": 8,
996
+ "group_size": 128,
997
+ "mode": "affine"
998
+ },
999
+ "language_model.model.layers.27.mlp.shared_expert.up_proj": {
1000
+ "bits": 8,
1001
+ "group_size": 128,
1002
+ "mode": "affine"
1003
+ },
1004
+ "language_model.model.layers.28.linear_attn.in_proj_a": {
1005
+ "bits": 8,
1006
+ "group_size": 128,
1007
+ "mode": "affine"
1008
+ },
1009
+ "language_model.model.layers.28.linear_attn.in_proj_b": {
1010
+ "bits": 8,
1011
+ "group_size": 128,
1012
+ "mode": "affine"
1013
+ },
1014
+ "language_model.model.layers.28.mlp.shared_expert.down_proj": {
1015
+ "bits": 8,
1016
+ "group_size": 128,
1017
+ "mode": "affine"
1018
+ },
1019
+ "language_model.model.layers.28.mlp.shared_expert.gate_proj": {
1020
+ "bits": 8,
1021
+ "group_size": 128,
1022
+ "mode": "affine"
1023
+ },
1024
+ "language_model.model.layers.28.mlp.shared_expert.up_proj": {
1025
+ "bits": 8,
1026
+ "group_size": 128,
1027
+ "mode": "affine"
1028
+ },
1029
+ "language_model.model.layers.29.linear_attn.in_proj_z": {
1030
+ "bits": 8,
1031
+ "group_size": 128,
1032
+ "mode": "affine"
1033
+ },
1034
+ "language_model.model.layers.29.linear_attn.out_proj": {
1035
+ "bits": 8,
1036
+ "group_size": 128,
1037
+ "mode": "affine"
1038
+ },
1039
+ "language_model.model.layers.30.linear_attn.in_proj_z": {
1040
+ "bits": 8,
1041
+ "group_size": 128,
1042
+ "mode": "affine"
1043
+ },
1044
+ "language_model.model.layers.30.linear_attn.out_proj": {
1045
+ "bits": 8,
1046
+ "group_size": 128,
1047
+ "mode": "affine"
1048
+ },
1049
+ "language_model.model.layers.29.linear_attn.in_proj_a": {
1050
+ "bits": 8,
1051
+ "group_size": 128,
1052
+ "mode": "affine"
1053
+ },
1054
+ "language_model.model.layers.29.linear_attn.in_proj_b": {
1055
+ "bits": 8,
1056
+ "group_size": 128,
1057
+ "mode": "affine"
1058
+ },
1059
+ "language_model.model.layers.29.mlp.shared_expert.down_proj": {
1060
+ "bits": 8,
1061
+ "group_size": 128,
1062
+ "mode": "affine"
1063
+ },
1064
+ "language_model.model.layers.29.mlp.shared_expert.gate_proj": {
1065
+ "bits": 8,
1066
+ "group_size": 128,
1067
+ "mode": "affine"
1068
+ },
1069
+ "language_model.model.layers.29.mlp.shared_expert.up_proj": {
1070
+ "bits": 8,
1071
+ "group_size": 128,
1072
+ "mode": "affine"
1073
+ },
1074
+ "language_model.model.layers.30.linear_attn.in_proj_a": {
1075
+ "bits": 8,
1076
+ "group_size": 128,
1077
+ "mode": "affine"
1078
+ },
1079
+ "language_model.model.layers.30.linear_attn.in_proj_b": {
1080
+ "bits": 8,
1081
+ "group_size": 128,
1082
+ "mode": "affine"
1083
+ },
1084
+ "language_model.model.layers.30.mlp.shared_expert.down_proj": {
1085
+ "bits": 8,
1086
+ "group_size": 128,
1087
+ "mode": "affine"
1088
+ },
1089
+ "language_model.model.layers.30.mlp.shared_expert.gate_proj": {
1090
+ "bits": 8,
1091
+ "group_size": 128,
1092
+ "mode": "affine"
1093
+ },
1094
+ "language_model.model.layers.30.mlp.shared_expert.up_proj": {
1095
+ "bits": 8,
1096
+ "group_size": 128,
1097
+ "mode": "affine"
1098
+ },
1099
+ "language_model.model.layers.31.mlp.shared_expert.down_proj": {
1100
+ "bits": 8,
1101
+ "group_size": 128,
1102
+ "mode": "affine"
1103
+ },
1104
+ "language_model.model.layers.31.mlp.shared_expert.gate_proj": {
1105
+ "bits": 8,
1106
+ "group_size": 128,
1107
+ "mode": "affine"
1108
+ },
1109
+ "language_model.model.layers.31.mlp.shared_expert.up_proj": {
1110
+ "bits": 8,
1111
+ "group_size": 128,
1112
+ "mode": "affine"
1113
+ },
1114
+ "language_model.model.layers.32.linear_attn.in_proj_a": {
1115
+ "bits": 8,
1116
+ "group_size": 128,
1117
+ "mode": "affine"
1118
+ },
1119
+ "language_model.model.layers.32.linear_attn.in_proj_b": {
1120
+ "bits": 8,
1121
+ "group_size": 128,
1122
+ "mode": "affine"
1123
+ },
1124
+ "language_model.model.layers.32.linear_attn.in_proj_z": {
1125
+ "bits": 8,
1126
+ "group_size": 128,
1127
+ "mode": "affine"
1128
+ },
1129
+ "language_model.model.layers.32.linear_attn.out_proj": {
1130
+ "bits": 8,
1131
+ "group_size": 128,
1132
+ "mode": "affine"
1133
+ },
1134
+ "language_model.model.layers.32.mlp.shared_expert.down_proj": {
1135
+ "bits": 8,
1136
+ "group_size": 128,
1137
+ "mode": "affine"
1138
+ },
1139
+ "language_model.model.layers.32.mlp.shared_expert.gate_proj": {
1140
+ "bits": 8,
1141
+ "group_size": 128,
1142
+ "mode": "affine"
1143
+ },
1144
+ "language_model.model.layers.32.mlp.shared_expert.up_proj": {
1145
+ "bits": 8,
1146
+ "group_size": 128,
1147
+ "mode": "affine"
1148
+ },
1149
+ "language_model.model.layers.33.linear_attn.in_proj_a": {
1150
+ "bits": 8,
1151
+ "group_size": 128,
1152
+ "mode": "affine"
1153
+ },
1154
+ "language_model.model.layers.33.linear_attn.in_proj_b": {
1155
+ "bits": 8,
1156
+ "group_size": 128,
1157
+ "mode": "affine"
1158
+ },
1159
+ "language_model.model.layers.33.linear_attn.in_proj_qkv": {
1160
+ "bits": 8,
1161
+ "group_size": 128,
1162
+ "mode": "affine"
1163
+ },
1164
+ "language_model.model.layers.33.linear_attn.in_proj_z": {
1165
+ "bits": 8,
1166
+ "group_size": 128,
1167
+ "mode": "affine"
1168
+ },
1169
+ "language_model.model.layers.33.linear_attn.out_proj": {
1170
+ "bits": 8,
1171
+ "group_size": 128,
1172
+ "mode": "affine"
1173
+ },
1174
+ "language_model.model.layers.33.mlp.shared_expert.down_proj": {
1175
+ "bits": 8,
1176
+ "group_size": 128,
1177
+ "mode": "affine"
1178
+ },
1179
+ "language_model.model.layers.33.mlp.shared_expert.gate_proj": {
1180
+ "bits": 8,
1181
+ "group_size": 128,
1182
+ "mode": "affine"
1183
+ },
1184
+ "language_model.model.layers.33.mlp.shared_expert.up_proj": {
1185
+ "bits": 8,
1186
+ "group_size": 128,
1187
+ "mode": "affine"
1188
+ },
1189
+ "language_model.model.layers.34.linear_attn.in_proj_a": {
1190
+ "bits": 8,
1191
+ "group_size": 128,
1192
+ "mode": "affine"
1193
+ },
1194
+ "language_model.model.layers.34.linear_attn.in_proj_b": {
1195
+ "bits": 8,
1196
+ "group_size": 128,
1197
+ "mode": "affine"
1198
+ },
1199
+ "language_model.model.layers.34.linear_attn.in_proj_qkv": {
1200
+ "bits": 8,
1201
+ "group_size": 128,
1202
+ "mode": "affine"
1203
+ },
1204
+ "language_model.model.layers.34.linear_attn.in_proj_z": {
1205
+ "bits": 8,
1206
+ "group_size": 128,
1207
+ "mode": "affine"
1208
+ },
1209
+ "language_model.model.layers.34.linear_attn.out_proj": {
1210
+ "bits": 8,
1211
+ "group_size": 128,
1212
+ "mode": "affine"
1213
+ },
1214
+ "language_model.model.layers.34.mlp.shared_expert.down_proj": {
1215
+ "bits": 8,
1216
+ "group_size": 128,
1217
+ "mode": "affine"
1218
+ },
1219
+ "language_model.model.layers.34.mlp.shared_expert.gate_proj": {
1220
+ "bits": 8,
1221
+ "group_size": 128,
1222
+ "mode": "affine"
1223
+ },
1224
+ "language_model.model.layers.34.mlp.shared_expert.up_proj": {
1225
+ "bits": 8,
1226
+ "group_size": 128,
1227
+ "mode": "affine"
1228
+ },
1229
+ "language_model.model.layers.35.mlp.shared_expert.down_proj": {
1230
+ "bits": 8,
1231
+ "group_size": 128,
1232
+ "mode": "affine"
1233
+ },
1234
+ "language_model.model.layers.35.mlp.shared_expert.gate_proj": {
1235
+ "bits": 8,
1236
+ "group_size": 128,
1237
+ "mode": "affine"
1238
+ },
1239
+ "language_model.model.layers.35.mlp.shared_expert.up_proj": {
1240
+ "bits": 8,
1241
+ "group_size": 128,
1242
+ "mode": "affine"
1243
+ },
1244
+ "language_model.model.layers.35.self_attn.k_proj": {
1245
+ "bits": 8,
1246
+ "group_size": 128,
1247
+ "mode": "affine"
1248
+ },
1249
+ "language_model.model.layers.35.self_attn.q_proj": {
1250
+ "bits": 8,
1251
+ "group_size": 128,
1252
+ "mode": "affine"
1253
+ },
1254
+ "language_model.model.layers.35.self_attn.v_proj": {
1255
+ "bits": 8,
1256
+ "group_size": 128,
1257
+ "mode": "affine"
1258
+ },
1259
+ "language_model.model.layers.36.linear_attn.in_proj_a": {
1260
+ "bits": 8,
1261
+ "group_size": 128,
1262
+ "mode": "affine"
1263
+ },
1264
+ "language_model.model.layers.36.linear_attn.in_proj_b": {
1265
+ "bits": 8,
1266
+ "group_size": 128,
1267
+ "mode": "affine"
1268
+ },
1269
+ "language_model.model.layers.36.linear_attn.in_proj_qkv": {
1270
+ "bits": 8,
1271
+ "group_size": 128,
1272
+ "mode": "affine"
1273
+ },
1274
+ "language_model.model.layers.36.linear_attn.in_proj_z": {
1275
+ "bits": 8,
1276
+ "group_size": 128,
1277
+ "mode": "affine"
1278
+ },
1279
+ "language_model.model.layers.36.linear_attn.out_proj": {
1280
+ "bits": 8,
1281
+ "group_size": 128,
1282
+ "mode": "affine"
1283
+ },
1284
+ "language_model.model.layers.36.mlp.shared_expert.down_proj": {
1285
+ "bits": 8,
1286
+ "group_size": 128,
1287
+ "mode": "affine"
1288
+ },
1289
+ "language_model.model.layers.36.mlp.shared_expert.gate_proj": {
1290
+ "bits": 8,
1291
+ "group_size": 128,
1292
+ "mode": "affine"
1293
+ },
1294
+ "language_model.model.layers.36.mlp.shared_expert.up_proj": {
1295
+ "bits": 8,
1296
+ "group_size": 128,
1297
+ "mode": "affine"
1298
+ },
1299
+ "language_model.model.layers.37.linear_attn.in_proj_a": {
1300
+ "bits": 8,
1301
+ "group_size": 128,
1302
+ "mode": "affine"
1303
+ },
1304
+ "language_model.model.layers.37.linear_attn.in_proj_b": {
1305
+ "bits": 8,
1306
+ "group_size": 128,
1307
+ "mode": "affine"
1308
+ },
1309
+ "language_model.model.layers.37.linear_attn.in_proj_qkv": {
1310
+ "bits": 8,
1311
+ "group_size": 128,
1312
+ "mode": "affine"
1313
+ },
1314
+ "language_model.model.layers.37.linear_attn.in_proj_z": {
1315
+ "bits": 8,
1316
+ "group_size": 128,
1317
+ "mode": "affine"
1318
+ },
1319
+ "language_model.model.layers.37.linear_attn.out_proj": {
1320
+ "bits": 8,
1321
+ "group_size": 128,
1322
+ "mode": "affine"
1323
+ },
1324
+ "language_model.model.layers.37.mlp.shared_expert.down_proj": {
1325
+ "bits": 8,
1326
+ "group_size": 128,
1327
+ "mode": "affine"
1328
+ },
1329
+ "language_model.model.layers.37.mlp.shared_expert.gate_proj": {
1330
+ "bits": 8,
1331
+ "group_size": 128,
1332
+ "mode": "affine"
1333
+ },
1334
+ "language_model.model.layers.37.mlp.shared_expert.up_proj": {
1335
+ "bits": 8,
1336
+ "group_size": 128,
1337
+ "mode": "affine"
1338
+ },
1339
+ "language_model.model.layers.38.linear_attn.in_proj_a": {
1340
+ "bits": 8,
1341
+ "group_size": 128,
1342
+ "mode": "affine"
1343
+ },
1344
+ "language_model.model.layers.38.linear_attn.in_proj_b": {
1345
+ "bits": 8,
1346
+ "group_size": 128,
1347
+ "mode": "affine"
1348
+ },
1349
+ "language_model.model.layers.38.linear_attn.in_proj_qkv": {
1350
+ "bits": 8,
1351
+ "group_size": 128,
1352
+ "mode": "affine"
1353
+ },
1354
+ "language_model.model.layers.38.linear_attn.in_proj_z": {
1355
+ "bits": 8,
1356
+ "group_size": 128,
1357
+ "mode": "affine"
1358
+ },
1359
+ "language_model.model.layers.38.linear_attn.out_proj": {
1360
+ "bits": 8,
1361
+ "group_size": 128,
1362
+ "mode": "affine"
1363
+ },
1364
+ "language_model.model.layers.38.mlp.shared_expert.down_proj": {
1365
+ "bits": 8,
1366
+ "group_size": 128,
1367
+ "mode": "affine"
1368
+ },
1369
+ "language_model.model.layers.38.mlp.shared_expert.gate_proj": {
1370
+ "bits": 8,
1371
+ "group_size": 128,
1372
+ "mode": "affine"
1373
+ },
1374
+ "language_model.model.layers.38.mlp.shared_expert.up_proj": {
1375
+ "bits": 8,
1376
+ "group_size": 128,
1377
+ "mode": "affine"
1378
+ },
1379
+ "language_model.lm_head": {
1380
+ "bits": 8,
1381
+ "group_size": 128,
1382
+ "mode": "affine"
1383
+ },
1384
+ "language_model.model.layers.39.mlp.shared_expert.down_proj": {
1385
+ "bits": 8,
1386
+ "group_size": 128,
1387
+ "mode": "affine"
1388
+ },
1389
+ "language_model.model.layers.39.mlp.shared_expert.gate_proj": {
1390
+ "bits": 8,
1391
+ "group_size": 128,
1392
+ "mode": "affine"
1393
+ },
1394
+ "language_model.model.layers.39.mlp.shared_expert.up_proj": {
1395
+ "bits": 8,
1396
+ "group_size": 128,
1397
+ "mode": "affine"
1398
+ },
1399
+ "language_model.model.layers.39.self_attn.k_proj": {
1400
+ "bits": 8,
1401
+ "group_size": 128,
1402
+ "mode": "affine"
1403
+ },
1404
+ "language_model.model.layers.39.self_attn.q_proj": {
1405
+ "bits": 8,
1406
+ "group_size": 128,
1407
+ "mode": "affine"
1408
+ },
1409
+ "language_model.model.layers.39.self_attn.v_proj": {
1410
+ "bits": 8,
1411
+ "group_size": 128,
1412
+ "mode": "affine"
1413
+ },
1414
+ "language_model.mtp.layers.0.mlp.shared_expert.down_proj": {
1415
+ "bits": 8,
1416
+ "group_size": 128,
1417
+ "mode": "affine"
1418
+ },
1419
+ "language_model.mtp.layers.0.mlp.shared_expert.gate_proj": {
1420
+ "bits": 8,
1421
+ "group_size": 128,
1422
+ "mode": "affine"
1423
+ },
1424
+ "language_model.mtp.layers.0.mlp.shared_expert.up_proj": {
1425
+ "bits": 8,
1426
+ "group_size": 128,
1427
+ "mode": "affine"
1428
+ },
1429
+ "language_model.mtp.layers.0.self_attn.k_proj": {
1430
+ "bits": 8,
1431
+ "group_size": 128,
1432
+ "mode": "affine"
1433
+ },
1434
+ "language_model.mtp.layers.0.self_attn.q_proj": {
1435
+ "bits": 8,
1436
+ "group_size": 128,
1437
+ "mode": "affine"
1438
+ },
1439
+ "language_model.mtp.layers.0.self_attn.v_proj": {
1440
+ "bits": 8,
1441
+ "group_size": 128,
1442
+ "mode": "affine"
1443
+ }
1444
+ },
1445
+ "quantization_config": {
1446
+ "group_size": 64,
1447
+ "bits": 8,
1448
+ "mode": "affine",
1449
+ "language_model.model.layers.0.linear_attn.in_proj_qkv": {
1450
+ "bits": 8,
1451
+ "group_size": 128,
1452
+ "mode": "affine"
1453
+ },
1454
+ "language_model.model.layers.0.linear_attn.in_proj_z": {
1455
+ "bits": 8,
1456
+ "group_size": 128,
1457
+ "mode": "affine"
1458
+ },
1459
+ "language_model.model.layers.0.linear_attn.out_proj": {
1460
+ "bits": 8,
1461
+ "group_size": 128,
1462
+ "mode": "affine"
1463
+ },
1464
+ "language_model.model.layers.1.linear_attn.in_proj_qkv": {
1465
+ "bits": 8,
1466
+ "group_size": 128,
1467
+ "mode": "affine"
1468
+ },
1469
+ "language_model.model.layers.1.linear_attn.in_proj_z": {
1470
+ "bits": 8,
1471
+ "group_size": 128,
1472
+ "mode": "affine"
1473
+ },
1474
+ "language_model.model.layers.1.linear_attn.out_proj": {
1475
+ "bits": 8,
1476
+ "group_size": 128,
1477
+ "mode": "affine"
1478
+ },
1479
+ "language_model.model.layers.0.linear_attn.in_proj_a": {
1480
+ "bits": 8,
1481
+ "group_size": 128,
1482
+ "mode": "affine"
1483
+ },
1484
+ "language_model.model.layers.0.linear_attn.in_proj_b": {
1485
+ "bits": 8,
1486
+ "group_size": 128,
1487
+ "mode": "affine"
1488
+ },
1489
+ "language_model.model.layers.0.mlp.shared_expert.down_proj": {
1490
+ "bits": 8,
1491
+ "group_size": 128,
1492
+ "mode": "affine"
1493
+ },
1494
+ "language_model.model.layers.0.mlp.shared_expert.gate_proj": {
1495
+ "bits": 8,
1496
+ "group_size": 128,
1497
+ "mode": "affine"
1498
+ },
1499
+ "language_model.model.layers.0.mlp.shared_expert.up_proj": {
1500
+ "bits": 8,
1501
+ "group_size": 128,
1502
+ "mode": "affine"
1503
+ },
1504
+ "language_model.model.layers.1.linear_attn.in_proj_a": {
1505
+ "bits": 8,
1506
+ "group_size": 128,
1507
+ "mode": "affine"
1508
+ },
1509
+ "language_model.model.layers.1.linear_attn.in_proj_b": {
1510
+ "bits": 8,
1511
+ "group_size": 128,
1512
+ "mode": "affine"
1513
+ },
1514
+ "language_model.model.layers.1.mlp.shared_expert.down_proj": {
1515
+ "bits": 8,
1516
+ "group_size": 128,
1517
+ "mode": "affine"
1518
+ },
1519
+ "language_model.model.layers.1.mlp.shared_expert.gate_proj": {
1520
+ "bits": 8,
1521
+ "group_size": 128,
1522
+ "mode": "affine"
1523
+ },
1524
+ "language_model.model.layers.1.mlp.shared_expert.up_proj": {
1525
+ "bits": 8,
1526
+ "group_size": 128,
1527
+ "mode": "affine"
1528
+ },
1529
+ "language_model.model.layers.2.linear_attn.in_proj_a": {
1530
+ "bits": 8,
1531
+ "group_size": 128,
1532
+ "mode": "affine"
1533
+ },
1534
+ "language_model.model.layers.2.linear_attn.in_proj_b": {
1535
+ "bits": 8,
1536
+ "group_size": 128,
1537
+ "mode": "affine"
1538
+ },
1539
+ "language_model.model.layers.2.linear_attn.in_proj_z": {
1540
+ "bits": 8,
1541
+ "group_size": 128,
1542
+ "mode": "affine"
1543
+ },
1544
+ "language_model.model.layers.2.linear_attn.out_proj": {
1545
+ "bits": 8,
1546
+ "group_size": 128,
1547
+ "mode": "affine"
1548
+ },
1549
+ "language_model.model.layers.2.mlp.shared_expert.down_proj": {
1550
+ "bits": 8,
1551
+ "group_size": 128,
1552
+ "mode": "affine"
1553
+ },
1554
+ "language_model.model.layers.2.mlp.shared_expert.gate_proj": {
1555
+ "bits": 8,
1556
+ "group_size": 128,
1557
+ "mode": "affine"
1558
+ },
1559
+ "language_model.model.layers.2.mlp.shared_expert.up_proj": {
1560
+ "bits": 8,
1561
+ "group_size": 128,
1562
+ "mode": "affine"
1563
+ },
1564
+ "language_model.model.layers.3.mlp.shared_expert.down_proj": {
1565
+ "bits": 8,
1566
+ "group_size": 128,
1567
+ "mode": "affine"
1568
+ },
1569
+ "language_model.model.layers.3.mlp.shared_expert.gate_proj": {
1570
+ "bits": 8,
1571
+ "group_size": 128,
1572
+ "mode": "affine"
1573
+ },
1574
+ "language_model.model.layers.3.mlp.shared_expert.up_proj": {
1575
+ "bits": 8,
1576
+ "group_size": 128,
1577
+ "mode": "affine"
1578
+ },
1579
+ "language_model.model.layers.4.linear_attn.in_proj_a": {
1580
+ "bits": 8,
1581
+ "group_size": 128,
1582
+ "mode": "affine"
1583
+ },
1584
+ "language_model.model.layers.4.linear_attn.in_proj_b": {
1585
+ "bits": 8,
1586
+ "group_size": 128,
1587
+ "mode": "affine"
1588
+ },
1589
+ "language_model.model.layers.4.linear_attn.in_proj_z": {
1590
+ "bits": 8,
1591
+ "group_size": 128,
1592
+ "mode": "affine"
1593
+ },
1594
+ "language_model.model.layers.4.linear_attn.out_proj": {
1595
+ "bits": 8,
1596
+ "group_size": 128,
1597
+ "mode": "affine"
1598
+ },
1599
+ "language_model.model.layers.4.mlp.shared_expert.down_proj": {
1600
+ "bits": 8,
1601
+ "group_size": 128,
1602
+ "mode": "affine"
1603
+ },
1604
+ "language_model.model.layers.4.mlp.shared_expert.gate_proj": {
1605
+ "bits": 8,
1606
+ "group_size": 128,
1607
+ "mode": "affine"
1608
+ },
1609
+ "language_model.model.layers.4.mlp.shared_expert.up_proj": {
1610
+ "bits": 8,
1611
+ "group_size": 128,
1612
+ "mode": "affine"
1613
+ },
1614
+ "language_model.model.layers.5.linear_attn.in_proj_a": {
1615
+ "bits": 8,
1616
+ "group_size": 128,
1617
+ "mode": "affine"
1618
+ },
1619
+ "language_model.model.layers.5.linear_attn.in_proj_b": {
1620
+ "bits": 8,
1621
+ "group_size": 128,
1622
+ "mode": "affine"
1623
+ },
1624
+ "language_model.model.layers.5.linear_attn.in_proj_z": {
1625
+ "bits": 8,
1626
+ "group_size": 128,
1627
+ "mode": "affine"
1628
+ },
1629
+ "language_model.model.layers.5.linear_attn.out_proj": {
1630
+ "bits": 8,
1631
+ "group_size": 128,
1632
+ "mode": "affine"
1633
+ },
1634
+ "language_model.model.layers.5.mlp.shared_expert.down_proj": {
1635
+ "bits": 8,
1636
+ "group_size": 128,
1637
+ "mode": "affine"
1638
+ },
1639
+ "language_model.model.layers.5.mlp.shared_expert.gate_proj": {
1640
+ "bits": 8,
1641
+ "group_size": 128,
1642
+ "mode": "affine"
1643
+ },
1644
+ "language_model.model.layers.5.mlp.shared_expert.up_proj": {
1645
+ "bits": 8,
1646
+ "group_size": 128,
1647
+ "mode": "affine"
1648
+ },
1649
+ "language_model.model.layers.6.linear_attn.in_proj_a": {
1650
+ "bits": 8,
1651
+ "group_size": 128,
1652
+ "mode": "affine"
1653
+ },
1654
+ "language_model.model.layers.6.linear_attn.in_proj_b": {
1655
+ "bits": 8,
1656
+ "group_size": 128,
1657
+ "mode": "affine"
1658
+ },
1659
+ "language_model.model.layers.6.linear_attn.in_proj_z": {
1660
+ "bits": 8,
1661
+ "group_size": 128,
1662
+ "mode": "affine"
1663
+ },
1664
+ "language_model.model.layers.6.linear_attn.out_proj": {
1665
+ "bits": 8,
1666
+ "group_size": 128,
1667
+ "mode": "affine"
1668
+ },
1669
+ "language_model.model.layers.6.mlp.shared_expert.down_proj": {
1670
+ "bits": 8,
1671
+ "group_size": 128,
1672
+ "mode": "affine"
1673
+ },
1674
+ "language_model.model.layers.6.mlp.shared_expert.gate_proj": {
1675
+ "bits": 8,
1676
+ "group_size": 128,
1677
+ "mode": "affine"
1678
+ },
1679
+ "language_model.model.layers.6.mlp.shared_expert.up_proj": {
1680
+ "bits": 8,
1681
+ "group_size": 128,
1682
+ "mode": "affine"
1683
+ },
1684
+ "language_model.model.layers.7.mlp.shared_expert.down_proj": {
1685
+ "bits": 8,
1686
+ "group_size": 128,
1687
+ "mode": "affine"
1688
+ },
1689
+ "language_model.model.layers.7.mlp.shared_expert.gate_proj": {
1690
+ "bits": 8,
1691
+ "group_size": 128,
1692
+ "mode": "affine"
1693
+ },
1694
+ "language_model.model.layers.7.mlp.shared_expert.up_proj": {
1695
+ "bits": 8,
1696
+ "group_size": 128,
1697
+ "mode": "affine"
1698
+ },
1699
+ "language_model.model.layers.10.linear_attn.in_proj_z": {
1700
+ "bits": 8,
1701
+ "group_size": 128,
1702
+ "mode": "affine"
1703
+ },
1704
+ "language_model.model.layers.10.linear_attn.out_proj": {
1705
+ "bits": 8,
1706
+ "group_size": 128,
1707
+ "mode": "affine"
1708
+ },
1709
+ "language_model.model.layers.8.linear_attn.in_proj_z": {
1710
+ "bits": 8,
1711
+ "group_size": 128,
1712
+ "mode": "affine"
1713
+ },
1714
+ "language_model.model.layers.8.linear_attn.out_proj": {
1715
+ "bits": 8,
1716
+ "group_size": 128,
1717
+ "mode": "affine"
1718
+ },
1719
+ "language_model.model.layers.9.linear_attn.in_proj_z": {
1720
+ "bits": 8,
1721
+ "group_size": 128,
1722
+ "mode": "affine"
1723
+ },
1724
+ "language_model.model.layers.9.linear_attn.out_proj": {
1725
+ "bits": 8,
1726
+ "group_size": 128,
1727
+ "mode": "affine"
1728
+ },
1729
+ "language_model.model.layers.10.linear_attn.in_proj_a": {
1730
+ "bits": 8,
1731
+ "group_size": 128,
1732
+ "mode": "affine"
1733
+ },
1734
+ "language_model.model.layers.10.linear_attn.in_proj_b": {
1735
+ "bits": 8,
1736
+ "group_size": 128,
1737
+ "mode": "affine"
1738
+ },
1739
+ "language_model.model.layers.10.mlp.shared_expert.down_proj": {
1740
+ "bits": 8,
1741
+ "group_size": 128,
1742
+ "mode": "affine"
1743
+ },
1744
+ "language_model.model.layers.10.mlp.shared_expert.gate_proj": {
1745
+ "bits": 8,
1746
+ "group_size": 128,
1747
+ "mode": "affine"
1748
+ },
1749
+ "language_model.model.layers.10.mlp.shared_expert.up_proj": {
1750
+ "bits": 8,
1751
+ "group_size": 128,
1752
+ "mode": "affine"
1753
+ },
1754
+ "language_model.model.layers.8.linear_attn.in_proj_a": {
1755
+ "bits": 8,
1756
+ "group_size": 128,
1757
+ "mode": "affine"
1758
+ },
1759
+ "language_model.model.layers.8.linear_attn.in_proj_b": {
1760
+ "bits": 8,
1761
+ "group_size": 128,
1762
+ "mode": "affine"
1763
+ },
1764
+ "language_model.model.layers.8.mlp.shared_expert.down_proj": {
1765
+ "bits": 8,
1766
+ "group_size": 128,
1767
+ "mode": "affine"
1768
+ },
1769
+ "language_model.model.layers.8.mlp.shared_expert.gate_proj": {
1770
+ "bits": 8,
1771
+ "group_size": 128,
1772
+ "mode": "affine"
1773
+ },
1774
+ "language_model.model.layers.8.mlp.shared_expert.up_proj": {
1775
+ "bits": 8,
1776
+ "group_size": 128,
1777
+ "mode": "affine"
1778
+ },
1779
+ "language_model.model.layers.9.linear_attn.in_proj_a": {
1780
+ "bits": 8,
1781
+ "group_size": 128,
1782
+ "mode": "affine"
1783
+ },
1784
+ "language_model.model.layers.9.linear_attn.in_proj_b": {
1785
+ "bits": 8,
1786
+ "group_size": 128,
1787
+ "mode": "affine"
1788
+ },
1789
+ "language_model.model.layers.9.mlp.shared_expert.down_proj": {
1790
+ "bits": 8,
1791
+ "group_size": 128,
1792
+ "mode": "affine"
1793
+ },
1794
+ "language_model.model.layers.9.mlp.shared_expert.gate_proj": {
1795
+ "bits": 8,
1796
+ "group_size": 128,
1797
+ "mode": "affine"
1798
+ },
1799
+ "language_model.model.layers.9.mlp.shared_expert.up_proj": {
1800
+ "bits": 8,
1801
+ "group_size": 128,
1802
+ "mode": "affine"
1803
+ },
1804
+ "language_model.model.layers.12.linear_attn.in_proj_z": {
1805
+ "bits": 8,
1806
+ "group_size": 128,
1807
+ "mode": "affine"
1808
+ },
1809
+ "language_model.model.layers.12.linear_attn.out_proj": {
1810
+ "bits": 8,
1811
+ "group_size": 128,
1812
+ "mode": "affine"
1813
+ },
1814
+ "language_model.model.layers.13.linear_attn.in_proj_z": {
1815
+ "bits": 8,
1816
+ "group_size": 128,
1817
+ "mode": "affine"
1818
+ },
1819
+ "language_model.model.layers.13.linear_attn.out_proj": {
1820
+ "bits": 8,
1821
+ "group_size": 128,
1822
+ "mode": "affine"
1823
+ },
1824
+ "language_model.model.layers.11.mlp.shared_expert.down_proj": {
1825
+ "bits": 8,
1826
+ "group_size": 128,
1827
+ "mode": "affine"
1828
+ },
1829
+ "language_model.model.layers.11.mlp.shared_expert.gate_proj": {
1830
+ "bits": 8,
1831
+ "group_size": 128,
1832
+ "mode": "affine"
1833
+ },
1834
+ "language_model.model.layers.11.mlp.shared_expert.up_proj": {
1835
+ "bits": 8,
1836
+ "group_size": 128,
1837
+ "mode": "affine"
1838
+ },
1839
+ "language_model.model.layers.12.linear_attn.in_proj_a": {
1840
+ "bits": 8,
1841
+ "group_size": 128,
1842
+ "mode": "affine"
1843
+ },
1844
+ "language_model.model.layers.12.linear_attn.in_proj_b": {
1845
+ "bits": 8,
1846
+ "group_size": 128,
1847
+ "mode": "affine"
1848
+ },
1849
+ "language_model.model.layers.12.mlp.shared_expert.down_proj": {
1850
+ "bits": 8,
1851
+ "group_size": 128,
1852
+ "mode": "affine"
1853
+ },
1854
+ "language_model.model.layers.12.mlp.shared_expert.gate_proj": {
1855
+ "bits": 8,
1856
+ "group_size": 128,
1857
+ "mode": "affine"
1858
+ },
1859
+ "language_model.model.layers.12.mlp.shared_expert.up_proj": {
1860
+ "bits": 8,
1861
+ "group_size": 128,
1862
+ "mode": "affine"
1863
+ },
1864
+ "language_model.model.layers.13.linear_attn.in_proj_a": {
1865
+ "bits": 8,
1866
+ "group_size": 128,
1867
+ "mode": "affine"
1868
+ },
1869
+ "language_model.model.layers.13.linear_attn.in_proj_b": {
1870
+ "bits": 8,
1871
+ "group_size": 128,
1872
+ "mode": "affine"
1873
+ },
1874
+ "language_model.model.layers.13.mlp.shared_expert.down_proj": {
1875
+ "bits": 8,
1876
+ "group_size": 128,
1877
+ "mode": "affine"
1878
+ },
1879
+ "language_model.model.layers.13.mlp.shared_expert.gate_proj": {
1880
+ "bits": 8,
1881
+ "group_size": 128,
1882
+ "mode": "affine"
1883
+ },
1884
+ "language_model.model.layers.13.mlp.shared_expert.up_proj": {
1885
+ "bits": 8,
1886
+ "group_size": 128,
1887
+ "mode": "affine"
1888
+ },
1889
+ "language_model.model.layers.14.linear_attn.in_proj_z": {
1890
+ "bits": 8,
1891
+ "group_size": 128,
1892
+ "mode": "affine"
1893
+ },
1894
+ "language_model.model.layers.14.linear_attn.out_proj": {
1895
+ "bits": 8,
1896
+ "group_size": 128,
1897
+ "mode": "affine"
1898
+ },
1899
+ "language_model.model.layers.16.linear_attn.in_proj_z": {
1900
+ "bits": 8,
1901
+ "group_size": 128,
1902
+ "mode": "affine"
1903
+ },
1904
+ "language_model.model.layers.16.linear_attn.out_proj": {
1905
+ "bits": 8,
1906
+ "group_size": 128,
1907
+ "mode": "affine"
1908
+ },
1909
+ "language_model.model.layers.14.linear_attn.in_proj_a": {
1910
+ "bits": 8,
1911
+ "group_size": 128,
1912
+ "mode": "affine"
1913
+ },
1914
+ "language_model.model.layers.14.linear_attn.in_proj_b": {
1915
+ "bits": 8,
1916
+ "group_size": 128,
1917
+ "mode": "affine"
1918
+ },
1919
+ "language_model.model.layers.14.mlp.shared_expert.down_proj": {
1920
+ "bits": 8,
1921
+ "group_size": 128,
1922
+ "mode": "affine"
1923
+ },
1924
+ "language_model.model.layers.14.mlp.shared_expert.gate_proj": {
1925
+ "bits": 8,
1926
+ "group_size": 128,
1927
+ "mode": "affine"
1928
+ },
1929
+ "language_model.model.layers.14.mlp.shared_expert.up_proj": {
1930
+ "bits": 8,
1931
+ "group_size": 128,
1932
+ "mode": "affine"
1933
+ },
1934
+ "language_model.model.layers.15.mlp.shared_expert.down_proj": {
1935
+ "bits": 8,
1936
+ "group_size": 128,
1937
+ "mode": "affine"
1938
+ },
1939
+ "language_model.model.layers.15.mlp.shared_expert.gate_proj": {
1940
+ "bits": 8,
1941
+ "group_size": 128,
1942
+ "mode": "affine"
1943
+ },
1944
+ "language_model.model.layers.15.mlp.shared_expert.up_proj": {
1945
+ "bits": 8,
1946
+ "group_size": 128,
1947
+ "mode": "affine"
1948
+ },
1949
+ "language_model.model.layers.16.linear_attn.in_proj_a": {
1950
+ "bits": 8,
1951
+ "group_size": 128,
1952
+ "mode": "affine"
1953
+ },
1954
+ "language_model.model.layers.16.linear_attn.in_proj_b": {
1955
+ "bits": 8,
1956
+ "group_size": 128,
1957
+ "mode": "affine"
1958
+ },
1959
+ "language_model.model.layers.16.mlp.shared_expert.down_proj": {
1960
+ "bits": 8,
1961
+ "group_size": 128,
1962
+ "mode": "affine"
1963
+ },
1964
+ "language_model.model.layers.16.mlp.shared_expert.gate_proj": {
1965
+ "bits": 8,
1966
+ "group_size": 128,
1967
+ "mode": "affine"
1968
+ },
1969
+ "language_model.model.layers.16.mlp.shared_expert.up_proj": {
1970
+ "bits": 8,
1971
+ "group_size": 128,
1972
+ "mode": "affine"
1973
+ },
1974
+ "language_model.model.layers.17.linear_attn.in_proj_z": {
1975
+ "bits": 8,
1976
+ "group_size": 128,
1977
+ "mode": "affine"
1978
+ },
1979
+ "language_model.model.layers.17.linear_attn.out_proj": {
1980
+ "bits": 8,
1981
+ "group_size": 128,
1982
+ "mode": "affine"
1983
+ },
1984
+ "language_model.model.layers.18.linear_attn.in_proj_z": {
1985
+ "bits": 8,
1986
+ "group_size": 128,
1987
+ "mode": "affine"
1988
+ },
1989
+ "language_model.model.layers.18.linear_attn.out_proj": {
1990
+ "bits": 8,
1991
+ "group_size": 128,
1992
+ "mode": "affine"
1993
+ },
1994
+ "language_model.model.layers.17.linear_attn.in_proj_a": {
1995
+ "bits": 8,
1996
+ "group_size": 128,
1997
+ "mode": "affine"
1998
+ },
1999
+ "language_model.model.layers.17.linear_attn.in_proj_b": {
2000
+ "bits": 8,
2001
+ "group_size": 128,
2002
+ "mode": "affine"
2003
+ },
2004
+ "language_model.model.layers.17.mlp.shared_expert.down_proj": {
2005
+ "bits": 8,
2006
+ "group_size": 128,
2007
+ "mode": "affine"
2008
+ },
2009
+ "language_model.model.layers.17.mlp.shared_expert.gate_proj": {
2010
+ "bits": 8,
2011
+ "group_size": 128,
2012
+ "mode": "affine"
2013
+ },
2014
+ "language_model.model.layers.17.mlp.shared_expert.up_proj": {
2015
+ "bits": 8,
2016
+ "group_size": 128,
2017
+ "mode": "affine"
2018
+ },
2019
+ "language_model.model.layers.18.linear_attn.in_proj_a": {
2020
+ "bits": 8,
2021
+ "group_size": 128,
2022
+ "mode": "affine"
2023
+ },
2024
+ "language_model.model.layers.18.linear_attn.in_proj_b": {
2025
+ "bits": 8,
2026
+ "group_size": 128,
2027
+ "mode": "affine"
2028
+ },
2029
+ "language_model.model.layers.18.mlp.shared_expert.down_proj": {
2030
+ "bits": 8,
2031
+ "group_size": 128,
2032
+ "mode": "affine"
2033
+ },
2034
+ "language_model.model.layers.18.mlp.shared_expert.gate_proj": {
2035
+ "bits": 8,
2036
+ "group_size": 128,
2037
+ "mode": "affine"
2038
+ },
2039
+ "language_model.model.layers.18.mlp.shared_expert.up_proj": {
2040
+ "bits": 8,
2041
+ "group_size": 128,
2042
+ "mode": "affine"
2043
+ },
2044
+ "language_model.model.layers.19.mlp.shared_expert.down_proj": {
2045
+ "bits": 8,
2046
+ "group_size": 128,
2047
+ "mode": "affine"
2048
+ },
2049
+ "language_model.model.layers.19.mlp.shared_expert.gate_proj": {
2050
+ "bits": 8,
2051
+ "group_size": 128,
2052
+ "mode": "affine"
2053
+ },
2054
+ "language_model.model.layers.19.mlp.shared_expert.up_proj": {
2055
+ "bits": 8,
2056
+ "group_size": 128,
2057
+ "mode": "affine"
2058
+ },
2059
+ "language_model.model.layers.20.linear_attn.in_proj_z": {
2060
+ "bits": 8,
2061
+ "group_size": 128,
2062
+ "mode": "affine"
2063
+ },
2064
+ "language_model.model.layers.20.linear_attn.out_proj": {
2065
+ "bits": 8,
2066
+ "group_size": 128,
2067
+ "mode": "affine"
2068
+ },
2069
+ "language_model.model.layers.21.linear_attn.in_proj_z": {
2070
+ "bits": 8,
2071
+ "group_size": 128,
2072
+ "mode": "affine"
2073
+ },
2074
+ "language_model.model.layers.21.linear_attn.out_proj": {
2075
+ "bits": 8,
2076
+ "group_size": 128,
2077
+ "mode": "affine"
2078
+ },
2079
+ "language_model.model.layers.22.linear_attn.in_proj_z": {
2080
+ "bits": 8,
2081
+ "group_size": 128,
2082
+ "mode": "affine"
2083
+ },
2084
+ "language_model.model.layers.22.linear_attn.out_proj": {
2085
+ "bits": 8,
2086
+ "group_size": 128,
2087
+ "mode": "affine"
2088
+ },
2089
+ "language_model.model.layers.20.linear_attn.in_proj_a": {
2090
+ "bits": 8,
2091
+ "group_size": 128,
2092
+ "mode": "affine"
2093
+ },
2094
+ "language_model.model.layers.20.linear_attn.in_proj_b": {
2095
+ "bits": 8,
2096
+ "group_size": 128,
2097
+ "mode": "affine"
2098
+ },
2099
+ "language_model.model.layers.20.mlp.shared_expert.down_proj": {
2100
+ "bits": 8,
2101
+ "group_size": 128,
2102
+ "mode": "affine"
2103
+ },
2104
+ "language_model.model.layers.20.mlp.shared_expert.gate_proj": {
2105
+ "bits": 8,
2106
+ "group_size": 128,
2107
+ "mode": "affine"
2108
+ },
2109
+ "language_model.model.layers.20.mlp.shared_expert.up_proj": {
2110
+ "bits": 8,
2111
+ "group_size": 128,
2112
+ "mode": "affine"
2113
+ },
2114
+ "language_model.model.layers.21.linear_attn.in_proj_a": {
2115
+ "bits": 8,
2116
+ "group_size": 128,
2117
+ "mode": "affine"
2118
+ },
2119
+ "language_model.model.layers.21.linear_attn.in_proj_b": {
2120
+ "bits": 8,
2121
+ "group_size": 128,
2122
+ "mode": "affine"
2123
+ },
2124
+ "language_model.model.layers.21.mlp.shared_expert.down_proj": {
2125
+ "bits": 8,
2126
+ "group_size": 128,
2127
+ "mode": "affine"
2128
+ },
2129
+ "language_model.model.layers.21.mlp.shared_expert.gate_proj": {
2130
+ "bits": 8,
2131
+ "group_size": 128,
2132
+ "mode": "affine"
2133
+ },
2134
+ "language_model.model.layers.21.mlp.shared_expert.up_proj": {
2135
+ "bits": 8,
2136
+ "group_size": 128,
2137
+ "mode": "affine"
2138
+ },
2139
+ "language_model.model.layers.22.linear_attn.in_proj_a": {
2140
+ "bits": 8,
2141
+ "group_size": 128,
2142
+ "mode": "affine"
2143
+ },
2144
+ "language_model.model.layers.22.linear_attn.in_proj_b": {
2145
+ "bits": 8,
2146
+ "group_size": 128,
2147
+ "mode": "affine"
2148
+ },
2149
+ "language_model.model.layers.22.mlp.shared_expert.down_proj": {
2150
+ "bits": 8,
2151
+ "group_size": 128,
2152
+ "mode": "affine"
2153
+ },
2154
+ "language_model.model.layers.22.mlp.shared_expert.gate_proj": {
2155
+ "bits": 8,
2156
+ "group_size": 128,
2157
+ "mode": "affine"
2158
+ },
2159
+ "language_model.model.layers.22.mlp.shared_expert.up_proj": {
2160
+ "bits": 8,
2161
+ "group_size": 128,
2162
+ "mode": "affine"
2163
+ },
2164
+ "language_model.model.layers.23.self_attn.k_proj": {
2165
+ "bits": 8,
2166
+ "group_size": 128,
2167
+ "mode": "affine"
2168
+ },
2169
+ "language_model.model.layers.23.self_attn.q_proj": {
2170
+ "bits": 8,
2171
+ "group_size": 128,
2172
+ "mode": "affine"
2173
+ },
2174
+ "language_model.model.layers.23.self_attn.v_proj": {
2175
+ "bits": 8,
2176
+ "group_size": 128,
2177
+ "mode": "affine"
2178
+ },
2179
+ "language_model.model.layers.24.linear_attn.in_proj_z": {
2180
+ "bits": 8,
2181
+ "group_size": 128,
2182
+ "mode": "affine"
2183
+ },
2184
+ "language_model.model.layers.24.linear_attn.out_proj": {
2185
+ "bits": 8,
2186
+ "group_size": 128,
2187
+ "mode": "affine"
2188
+ },
2189
+ "language_model.model.layers.25.linear_attn.in_proj_z": {
2190
+ "bits": 8,
2191
+ "group_size": 128,
2192
+ "mode": "affine"
2193
+ },
2194
+ "language_model.model.layers.25.linear_attn.out_proj": {
2195
+ "bits": 8,
2196
+ "group_size": 128,
2197
+ "mode": "affine"
2198
+ },
2199
+ "language_model.model.layers.23.mlp.shared_expert.down_proj": {
2200
+ "bits": 8,
2201
+ "group_size": 128,
2202
+ "mode": "affine"
2203
+ },
2204
+ "language_model.model.layers.23.mlp.shared_expert.gate_proj": {
2205
+ "bits": 8,
2206
+ "group_size": 128,
2207
+ "mode": "affine"
2208
+ },
2209
+ "language_model.model.layers.23.mlp.shared_expert.up_proj": {
2210
+ "bits": 8,
2211
+ "group_size": 128,
2212
+ "mode": "affine"
2213
+ },
2214
+ "language_model.model.layers.24.linear_attn.in_proj_a": {
2215
+ "bits": 8,
2216
+ "group_size": 128,
2217
+ "mode": "affine"
2218
+ },
2219
+ "language_model.model.layers.24.linear_attn.in_proj_b": {
2220
+ "bits": 8,
2221
+ "group_size": 128,
2222
+ "mode": "affine"
2223
+ },
2224
+ "language_model.model.layers.24.mlp.shared_expert.down_proj": {
2225
+ "bits": 8,
2226
+ "group_size": 128,
2227
+ "mode": "affine"
2228
+ },
2229
+ "language_model.model.layers.24.mlp.shared_expert.gate_proj": {
2230
+ "bits": 8,
2231
+ "group_size": 128,
2232
+ "mode": "affine"
2233
+ },
2234
+ "language_model.model.layers.24.mlp.shared_expert.up_proj": {
2235
+ "bits": 8,
2236
+ "group_size": 128,
2237
+ "mode": "affine"
2238
+ },
2239
+ "language_model.model.layers.25.linear_attn.in_proj_a": {
2240
+ "bits": 8,
2241
+ "group_size": 128,
2242
+ "mode": "affine"
2243
+ },
2244
+ "language_model.model.layers.25.linear_attn.in_proj_b": {
2245
+ "bits": 8,
2246
+ "group_size": 128,
2247
+ "mode": "affine"
2248
+ },
2249
+ "language_model.model.layers.25.mlp.shared_expert.down_proj": {
2250
+ "bits": 8,
2251
+ "group_size": 128,
2252
+ "mode": "affine"
2253
+ },
2254
+ "language_model.model.layers.25.mlp.shared_expert.gate_proj": {
2255
+ "bits": 8,
2256
+ "group_size": 128,
2257
+ "mode": "affine"
2258
+ },
2259
+ "language_model.model.layers.25.mlp.shared_expert.up_proj": {
2260
+ "bits": 8,
2261
+ "group_size": 128,
2262
+ "mode": "affine"
2263
+ },
2264
+ "language_model.model.layers.26.linear_attn.in_proj_z": {
2265
+ "bits": 8,
2266
+ "group_size": 128,
2267
+ "mode": "affine"
2268
+ },
2269
+ "language_model.model.layers.26.linear_attn.out_proj": {
2270
+ "bits": 8,
2271
+ "group_size": 128,
2272
+ "mode": "affine"
2273
+ },
2274
+ "language_model.model.layers.28.linear_attn.in_proj_z": {
2275
+ "bits": 8,
2276
+ "group_size": 128,
2277
+ "mode": "affine"
2278
+ },
2279
+ "language_model.model.layers.28.linear_attn.out_proj": {
2280
+ "bits": 8,
2281
+ "group_size": 128,
2282
+ "mode": "affine"
2283
+ },
2284
+ "language_model.model.layers.26.linear_attn.in_proj_a": {
2285
+ "bits": 8,
2286
+ "group_size": 128,
2287
+ "mode": "affine"
2288
+ },
2289
+ "language_model.model.layers.26.linear_attn.in_proj_b": {
2290
+ "bits": 8,
2291
+ "group_size": 128,
2292
+ "mode": "affine"
2293
+ },
2294
+ "language_model.model.layers.26.mlp.shared_expert.down_proj": {
2295
+ "bits": 8,
2296
+ "group_size": 128,
2297
+ "mode": "affine"
2298
+ },
2299
+ "language_model.model.layers.26.mlp.shared_expert.gate_proj": {
2300
+ "bits": 8,
2301
+ "group_size": 128,
2302
+ "mode": "affine"
2303
+ },
2304
+ "language_model.model.layers.26.mlp.shared_expert.up_proj": {
2305
+ "bits": 8,
2306
+ "group_size": 128,
2307
+ "mode": "affine"
2308
+ },
2309
+ "language_model.model.layers.27.mlp.shared_expert.down_proj": {
2310
+ "bits": 8,
2311
+ "group_size": 128,
2312
+ "mode": "affine"
2313
+ },
2314
+ "language_model.model.layers.27.mlp.shared_expert.gate_proj": {
2315
+ "bits": 8,
2316
+ "group_size": 128,
2317
+ "mode": "affine"
2318
+ },
2319
+ "language_model.model.layers.27.mlp.shared_expert.up_proj": {
2320
+ "bits": 8,
2321
+ "group_size": 128,
2322
+ "mode": "affine"
2323
+ },
2324
+ "language_model.model.layers.28.linear_attn.in_proj_a": {
2325
+ "bits": 8,
2326
+ "group_size": 128,
2327
+ "mode": "affine"
2328
+ },
2329
+ "language_model.model.layers.28.linear_attn.in_proj_b": {
2330
+ "bits": 8,
2331
+ "group_size": 128,
2332
+ "mode": "affine"
2333
+ },
2334
+ "language_model.model.layers.28.mlp.shared_expert.down_proj": {
2335
+ "bits": 8,
2336
+ "group_size": 128,
2337
+ "mode": "affine"
2338
+ },
2339
+ "language_model.model.layers.28.mlp.shared_expert.gate_proj": {
2340
+ "bits": 8,
2341
+ "group_size": 128,
2342
+ "mode": "affine"
2343
+ },
2344
+ "language_model.model.layers.28.mlp.shared_expert.up_proj": {
2345
+ "bits": 8,
2346
+ "group_size": 128,
2347
+ "mode": "affine"
2348
+ },
2349
+ "language_model.model.layers.29.linear_attn.in_proj_z": {
2350
+ "bits": 8,
2351
+ "group_size": 128,
2352
+ "mode": "affine"
2353
+ },
2354
+ "language_model.model.layers.29.linear_attn.out_proj": {
2355
+ "bits": 8,
2356
+ "group_size": 128,
2357
+ "mode": "affine"
2358
+ },
2359
+ "language_model.model.layers.30.linear_attn.in_proj_z": {
2360
+ "bits": 8,
2361
+ "group_size": 128,
2362
+ "mode": "affine"
2363
+ },
2364
+ "language_model.model.layers.30.linear_attn.out_proj": {
2365
+ "bits": 8,
2366
+ "group_size": 128,
2367
+ "mode": "affine"
2368
+ },
2369
+ "language_model.model.layers.29.linear_attn.in_proj_a": {
2370
+ "bits": 8,
2371
+ "group_size": 128,
2372
+ "mode": "affine"
2373
+ },
2374
+ "language_model.model.layers.29.linear_attn.in_proj_b": {
2375
+ "bits": 8,
2376
+ "group_size": 128,
2377
+ "mode": "affine"
2378
+ },
2379
+ "language_model.model.layers.29.mlp.shared_expert.down_proj": {
2380
+ "bits": 8,
2381
+ "group_size": 128,
2382
+ "mode": "affine"
2383
+ },
2384
+ "language_model.model.layers.29.mlp.shared_expert.gate_proj": {
2385
+ "bits": 8,
2386
+ "group_size": 128,
2387
+ "mode": "affine"
2388
+ },
2389
+ "language_model.model.layers.29.mlp.shared_expert.up_proj": {
2390
+ "bits": 8,
2391
+ "group_size": 128,
2392
+ "mode": "affine"
2393
+ },
2394
+ "language_model.model.layers.30.linear_attn.in_proj_a": {
2395
+ "bits": 8,
2396
+ "group_size": 128,
2397
+ "mode": "affine"
2398
+ },
2399
+ "language_model.model.layers.30.linear_attn.in_proj_b": {
2400
+ "bits": 8,
2401
+ "group_size": 128,
2402
+ "mode": "affine"
2403
+ },
2404
+ "language_model.model.layers.30.mlp.shared_expert.down_proj": {
2405
+ "bits": 8,
2406
+ "group_size": 128,
2407
+ "mode": "affine"
2408
+ },
2409
+ "language_model.model.layers.30.mlp.shared_expert.gate_proj": {
2410
+ "bits": 8,
2411
+ "group_size": 128,
2412
+ "mode": "affine"
2413
+ },
2414
+ "language_model.model.layers.30.mlp.shared_expert.up_proj": {
2415
+ "bits": 8,
2416
+ "group_size": 128,
2417
+ "mode": "affine"
2418
+ },
2419
+ "language_model.model.layers.31.mlp.shared_expert.down_proj": {
2420
+ "bits": 8,
2421
+ "group_size": 128,
2422
+ "mode": "affine"
2423
+ },
2424
+ "language_model.model.layers.31.mlp.shared_expert.gate_proj": {
2425
+ "bits": 8,
2426
+ "group_size": 128,
2427
+ "mode": "affine"
2428
+ },
2429
+ "language_model.model.layers.31.mlp.shared_expert.up_proj": {
2430
+ "bits": 8,
2431
+ "group_size": 128,
2432
+ "mode": "affine"
2433
+ },
2434
+ "language_model.model.layers.32.linear_attn.in_proj_a": {
2435
+ "bits": 8,
2436
+ "group_size": 128,
2437
+ "mode": "affine"
2438
+ },
2439
+ "language_model.model.layers.32.linear_attn.in_proj_b": {
2440
+ "bits": 8,
2441
+ "group_size": 128,
2442
+ "mode": "affine"
2443
+ },
2444
+ "language_model.model.layers.32.linear_attn.in_proj_z": {
2445
+ "bits": 8,
2446
+ "group_size": 128,
2447
+ "mode": "affine"
2448
+ },
2449
+ "language_model.model.layers.32.linear_attn.out_proj": {
2450
+ "bits": 8,
2451
+ "group_size": 128,
2452
+ "mode": "affine"
2453
+ },
2454
+ "language_model.model.layers.32.mlp.shared_expert.down_proj": {
2455
+ "bits": 8,
2456
+ "group_size": 128,
2457
+ "mode": "affine"
2458
+ },
2459
+ "language_model.model.layers.32.mlp.shared_expert.gate_proj": {
2460
+ "bits": 8,
2461
+ "group_size": 128,
2462
+ "mode": "affine"
2463
+ },
2464
+ "language_model.model.layers.32.mlp.shared_expert.up_proj": {
2465
+ "bits": 8,
2466
+ "group_size": 128,
2467
+ "mode": "affine"
2468
+ },
2469
+ "language_model.model.layers.33.linear_attn.in_proj_a": {
2470
+ "bits": 8,
2471
+ "group_size": 128,
2472
+ "mode": "affine"
2473
+ },
2474
+ "language_model.model.layers.33.linear_attn.in_proj_b": {
2475
+ "bits": 8,
2476
+ "group_size": 128,
2477
+ "mode": "affine"
2478
+ },
2479
+ "language_model.model.layers.33.linear_attn.in_proj_qkv": {
2480
+ "bits": 8,
2481
+ "group_size": 128,
2482
+ "mode": "affine"
2483
+ },
2484
+ "language_model.model.layers.33.linear_attn.in_proj_z": {
2485
+ "bits": 8,
2486
+ "group_size": 128,
2487
+ "mode": "affine"
2488
+ },
2489
+ "language_model.model.layers.33.linear_attn.out_proj": {
2490
+ "bits": 8,
2491
+ "group_size": 128,
2492
+ "mode": "affine"
2493
+ },
2494
+ "language_model.model.layers.33.mlp.shared_expert.down_proj": {
2495
+ "bits": 8,
2496
+ "group_size": 128,
2497
+ "mode": "affine"
2498
+ },
2499
+ "language_model.model.layers.33.mlp.shared_expert.gate_proj": {
2500
+ "bits": 8,
2501
+ "group_size": 128,
2502
+ "mode": "affine"
2503
+ },
2504
+ "language_model.model.layers.33.mlp.shared_expert.up_proj": {
2505
+ "bits": 8,
2506
+ "group_size": 128,
2507
+ "mode": "affine"
2508
+ },
2509
+ "language_model.model.layers.34.linear_attn.in_proj_a": {
2510
+ "bits": 8,
2511
+ "group_size": 128,
2512
+ "mode": "affine"
2513
+ },
2514
+ "language_model.model.layers.34.linear_attn.in_proj_b": {
2515
+ "bits": 8,
2516
+ "group_size": 128,
2517
+ "mode": "affine"
2518
+ },
2519
+ "language_model.model.layers.34.linear_attn.in_proj_qkv": {
2520
+ "bits": 8,
2521
+ "group_size": 128,
2522
+ "mode": "affine"
2523
+ },
2524
+ "language_model.model.layers.34.linear_attn.in_proj_z": {
2525
+ "bits": 8,
2526
+ "group_size": 128,
2527
+ "mode": "affine"
2528
+ },
2529
+ "language_model.model.layers.34.linear_attn.out_proj": {
2530
+ "bits": 8,
2531
+ "group_size": 128,
2532
+ "mode": "affine"
2533
+ },
2534
+ "language_model.model.layers.34.mlp.shared_expert.down_proj": {
2535
+ "bits": 8,
2536
+ "group_size": 128,
2537
+ "mode": "affine"
2538
+ },
2539
+ "language_model.model.layers.34.mlp.shared_expert.gate_proj": {
2540
+ "bits": 8,
2541
+ "group_size": 128,
2542
+ "mode": "affine"
2543
+ },
2544
+ "language_model.model.layers.34.mlp.shared_expert.up_proj": {
2545
+ "bits": 8,
2546
+ "group_size": 128,
2547
+ "mode": "affine"
2548
+ },
2549
+ "language_model.model.layers.35.mlp.shared_expert.down_proj": {
2550
+ "bits": 8,
2551
+ "group_size": 128,
2552
+ "mode": "affine"
2553
+ },
2554
+ "language_model.model.layers.35.mlp.shared_expert.gate_proj": {
2555
+ "bits": 8,
2556
+ "group_size": 128,
2557
+ "mode": "affine"
2558
+ },
2559
+ "language_model.model.layers.35.mlp.shared_expert.up_proj": {
2560
+ "bits": 8,
2561
+ "group_size": 128,
2562
+ "mode": "affine"
2563
+ },
2564
+ "language_model.model.layers.35.self_attn.k_proj": {
2565
+ "bits": 8,
2566
+ "group_size": 128,
2567
+ "mode": "affine"
2568
+ },
2569
+ "language_model.model.layers.35.self_attn.q_proj": {
2570
+ "bits": 8,
2571
+ "group_size": 128,
2572
+ "mode": "affine"
2573
+ },
2574
+ "language_model.model.layers.35.self_attn.v_proj": {
2575
+ "bits": 8,
2576
+ "group_size": 128,
2577
+ "mode": "affine"
2578
+ },
2579
+ "language_model.model.layers.36.linear_attn.in_proj_a": {
2580
+ "bits": 8,
2581
+ "group_size": 128,
2582
+ "mode": "affine"
2583
+ },
2584
+ "language_model.model.layers.36.linear_attn.in_proj_b": {
2585
+ "bits": 8,
2586
+ "group_size": 128,
2587
+ "mode": "affine"
2588
+ },
2589
+ "language_model.model.layers.36.linear_attn.in_proj_qkv": {
2590
+ "bits": 8,
2591
+ "group_size": 128,
2592
+ "mode": "affine"
2593
+ },
2594
+ "language_model.model.layers.36.linear_attn.in_proj_z": {
2595
+ "bits": 8,
2596
+ "group_size": 128,
2597
+ "mode": "affine"
2598
+ },
2599
+ "language_model.model.layers.36.linear_attn.out_proj": {
2600
+ "bits": 8,
2601
+ "group_size": 128,
2602
+ "mode": "affine"
2603
+ },
2604
+ "language_model.model.layers.36.mlp.shared_expert.down_proj": {
2605
+ "bits": 8,
2606
+ "group_size": 128,
2607
+ "mode": "affine"
2608
+ },
2609
+ "language_model.model.layers.36.mlp.shared_expert.gate_proj": {
2610
+ "bits": 8,
2611
+ "group_size": 128,
2612
+ "mode": "affine"
2613
+ },
2614
+ "language_model.model.layers.36.mlp.shared_expert.up_proj": {
2615
+ "bits": 8,
2616
+ "group_size": 128,
2617
+ "mode": "affine"
2618
+ },
2619
+ "language_model.model.layers.37.linear_attn.in_proj_a": {
2620
+ "bits": 8,
2621
+ "group_size": 128,
2622
+ "mode": "affine"
2623
+ },
2624
+ "language_model.model.layers.37.linear_attn.in_proj_b": {
2625
+ "bits": 8,
2626
+ "group_size": 128,
2627
+ "mode": "affine"
2628
+ },
2629
+ "language_model.model.layers.37.linear_attn.in_proj_qkv": {
2630
+ "bits": 8,
2631
+ "group_size": 128,
2632
+ "mode": "affine"
2633
+ },
2634
+ "language_model.model.layers.37.linear_attn.in_proj_z": {
2635
+ "bits": 8,
2636
+ "group_size": 128,
2637
+ "mode": "affine"
2638
+ },
2639
+ "language_model.model.layers.37.linear_attn.out_proj": {
2640
+ "bits": 8,
2641
+ "group_size": 128,
2642
+ "mode": "affine"
2643
+ },
2644
+ "language_model.model.layers.37.mlp.shared_expert.down_proj": {
2645
+ "bits": 8,
2646
+ "group_size": 128,
2647
+ "mode": "affine"
2648
+ },
2649
+ "language_model.model.layers.37.mlp.shared_expert.gate_proj": {
2650
+ "bits": 8,
2651
+ "group_size": 128,
2652
+ "mode": "affine"
2653
+ },
2654
+ "language_model.model.layers.37.mlp.shared_expert.up_proj": {
2655
+ "bits": 8,
2656
+ "group_size": 128,
2657
+ "mode": "affine"
2658
+ },
2659
+ "language_model.model.layers.38.linear_attn.in_proj_a": {
2660
+ "bits": 8,
2661
+ "group_size": 128,
2662
+ "mode": "affine"
2663
+ },
2664
+ "language_model.model.layers.38.linear_attn.in_proj_b": {
2665
+ "bits": 8,
2666
+ "group_size": 128,
2667
+ "mode": "affine"
2668
+ },
2669
+ "language_model.model.layers.38.linear_attn.in_proj_qkv": {
2670
+ "bits": 8,
2671
+ "group_size": 128,
2672
+ "mode": "affine"
2673
+ },
2674
+ "language_model.model.layers.38.linear_attn.in_proj_z": {
2675
+ "bits": 8,
2676
+ "group_size": 128,
2677
+ "mode": "affine"
2678
+ },
2679
+ "language_model.model.layers.38.linear_attn.out_proj": {
2680
+ "bits": 8,
2681
+ "group_size": 128,
2682
+ "mode": "affine"
2683
+ },
2684
+ "language_model.model.layers.38.mlp.shared_expert.down_proj": {
2685
+ "bits": 8,
2686
+ "group_size": 128,
2687
+ "mode": "affine"
2688
+ },
2689
+ "language_model.model.layers.38.mlp.shared_expert.gate_proj": {
2690
+ "bits": 8,
2691
+ "group_size": 128,
2692
+ "mode": "affine"
2693
+ },
2694
+ "language_model.model.layers.38.mlp.shared_expert.up_proj": {
2695
+ "bits": 8,
2696
+ "group_size": 128,
2697
+ "mode": "affine"
2698
+ },
2699
+ "language_model.lm_head": {
2700
+ "bits": 8,
2701
+ "group_size": 128,
2702
+ "mode": "affine"
2703
+ },
2704
+ "language_model.model.layers.39.mlp.shared_expert.down_proj": {
2705
+ "bits": 8,
2706
+ "group_size": 128,
2707
+ "mode": "affine"
2708
+ },
2709
+ "language_model.model.layers.39.mlp.shared_expert.gate_proj": {
2710
+ "bits": 8,
2711
+ "group_size": 128,
2712
+ "mode": "affine"
2713
+ },
2714
+ "language_model.model.layers.39.mlp.shared_expert.up_proj": {
2715
+ "bits": 8,
2716
+ "group_size": 128,
2717
+ "mode": "affine"
2718
+ },
2719
+ "language_model.model.layers.39.self_attn.k_proj": {
2720
+ "bits": 8,
2721
+ "group_size": 128,
2722
+ "mode": "affine"
2723
+ },
2724
+ "language_model.model.layers.39.self_attn.q_proj": {
2725
+ "bits": 8,
2726
+ "group_size": 128,
2727
+ "mode": "affine"
2728
+ },
2729
+ "language_model.model.layers.39.self_attn.v_proj": {
2730
+ "bits": 8,
2731
+ "group_size": 128,
2732
+ "mode": "affine"
2733
+ },
2734
+ "language_model.mtp.layers.0.mlp.shared_expert.down_proj": {
2735
+ "bits": 8,
2736
+ "group_size": 128,
2737
+ "mode": "affine"
2738
+ },
2739
+ "language_model.mtp.layers.0.mlp.shared_expert.gate_proj": {
2740
+ "bits": 8,
2741
+ "group_size": 128,
2742
+ "mode": "affine"
2743
+ },
2744
+ "language_model.mtp.layers.0.mlp.shared_expert.up_proj": {
2745
+ "bits": 8,
2746
+ "group_size": 128,
2747
+ "mode": "affine"
2748
+ },
2749
+ "language_model.mtp.layers.0.self_attn.k_proj": {
2750
+ "bits": 8,
2751
+ "group_size": 128,
2752
+ "mode": "affine"
2753
+ },
2754
+ "language_model.mtp.layers.0.self_attn.q_proj": {
2755
+ "bits": 8,
2756
+ "group_size": 128,
2757
+ "mode": "affine"
2758
+ },
2759
+ "language_model.mtp.layers.0.self_attn.v_proj": {
2760
+ "bits": 8,
2761
+ "group_size": 128,
2762
+ "mode": "affine"
2763
+ }
2764
+ }
2765
+ }
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": "5.6.2"
13
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model-00001-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:526364c03b4e96c9a7d4f903646a3b6e53ddbe16b141d232c17cb5e293d7e520
3
+ size 5185646355
model-00002-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be598a8704ebbf8ab819933634f0a1a0ebe9d98c77871c5695b5e1130ea25e3f
3
+ size 5133835694
model-00003-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fc5a8a16a7a8931d2930e7625476c48949d163c437bf2f833110130c9ecde27
3
+ size 5133835716
model-00004-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6dbf440475223179625ab51551031d8113968b3627f5e803db88a758ca8f4051
3
+ size 5133835752
model-00005-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52e5e411a868f6b3ff3c834d2d440053fef15725085cdda521e3a507402bc87c
3
+ size 5133835730
model-00006-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:66fc83469341e5e641867cf56a88f8f6016741e042f8a1384d01df89e18563cf
3
+ size 5133835742
model-00007-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f1b52cc79d05fe40eaf0897d87eda63c99cafb16ae89699b8f5a7b12d055a61
3
+ size 5133835732
model-00008-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74c096948d672f8250648ecae9d3406b9be5d6f83aab75c068f09510a56a4f35
3
+ size 2616392588
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 sha256:5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42
3
+ size 12807982
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{%- if tools and tools is iterable and tools is not mapping %}\n {{- '<|im_start|>system\\n' }}\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 {{- '<|im_start|>system\\n' + content + '<|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 {%- else %}\n {%- if '</think>' in content %}\n {%- set reasoning_content = content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n {%- set content = content.split('</think>')[-1].lstrip('\\n') %}\n {%- endif %}\n {%- endif %}\n {%- set reasoning_content = reasoning_content|trim %}\n {%- if (preserve_thinking is defined and 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 %}\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
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff