lovejing0306 commited on
Commit
9503a29
·
verified ·
1 Parent(s): bea2150

Upload 5 files

Browse files
Files changed (6) hide show
  1. .gitattributes +1 -0
  2. README.md +217 -1
  3. config.json +48 -0
  4. tokenizer.json +3 -0
  5. tokenizer_config.json +239 -0
  6. 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 CHANGED
@@ -1,3 +1,219 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
- license: apache-2.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <p align="center">
2
+ <img src="docs/mano-asr-banner.svg" alt="mano-asr" width="800">
3
+ </p>
4
+
5
+ <p align="center">
6
+ <a href="#"><img src="https://img.shields.io/badge/python-3.10+-3776AB?logo=python&logoColor=white" alt="Python"></a>
7
+ <a href="#"><img src="https://img.shields.io/badge/FastAPI-server-009688?logo=fastapi&logoColor=white" alt="FastAPI"></a>
8
+ <a href="#"><img src="https://img.shields.io/badge/MLX-Apple%20Silicon-000000?logo=apple&logoColor=white" alt="MLX"></a>
9
+ <a href="https://github.com/Mininglamp-AI/mano-asr/blob/main/LICENSE"><img src="https://img.shields.io/github/license/Mininglamp-AI/mano-asr?color=blue" alt="License"></a>
10
+ <a href="https://github.com/Mininglamp-AI/mano-asr/stargazers"><img src="https://img.shields.io/github/stars/Mininglamp-AI/mano-asr?style=social" alt="Stars"></a>
11
+ </p>
12
+
13
  ---
14
+
15
+ ## Introduction
16
+
17
+ **mano-asr** is a local speech recognition (ASR) service built for **vertical domains**, deeply optimized for Apple Silicon via [MLX](https://github.com/ml-explore/mlx). It works out of the box, runs fully locally, and keeps your data on your machine.
18
+
19
+ mano-asr is specially tuned for **internet / IT office** scenarios — meeting notes, technical discussions, product reviews, and engineering dictation — where English terms, acronyms, product names and jargon (e.g. `FastAPI`, `Kubernetes`, `PRD`, `Code Review`) appear frequently, bringing recognition accuracy on these terms to a usable level.
20
+
21
+ Core capabilities:
22
+
23
+ - 🎯 **Vertical-domain tuning** — specialized tuning for internet / IT office jargon and mixed Chinese-English speech.
24
+ - 🍎 **Native Apple Silicon** — MLX-based local inference on M-series chips, further optimized with our in-house acceleration framework Cider.
25
+ - 🔒 **Fully local, privacy-first** — audio and transcripts never leave your machine.
26
+ - ✂️ **VAD segmentation** — optional FSMN VAD splits long audio and transcribes segment by segment.
27
+ - 🧩 **Pluggable engines** — supports Fun-ASR-Nano, Qwen3-ASR and more base models, switchable with one command.
28
+ - ⚡ **One-command start** — install via `brew install`, then `mano-asr start`.
29
+
30
  ---
31
+
32
+ <p align="center">
33
+ <a href="#en-news">Changelog</a> ·
34
+ <a href="#en-examples">Usage</a> ·
35
+ <a href="#en-models">Models</a> ·
36
+ <a href="#en-install">Installation</a> ·
37
+ <a href="#en-api">API</a> ·
38
+ <a href="#en-license">License</a> ·
39
+ <a href="#en-acknowledgments">Acknowledgments</a> ·
40
+ </p>
41
+
42
+ ---
43
+
44
+ <a id="en-news"></a>
45
+
46
+ ## Changelog
47
+
48
+ See the full release history on the **[Releases](https://github.com/Mininglamp-AI/mano-asr/releases)** page.
49
+
50
+ <a id="en-examples"></a>
51
+
52
+ ## Usage
53
+
54
+ The example below shows **audio translation**: transcribe speech and translate it into Chinese.
55
+
56
+ ### CLI (recommended)
57
+
58
+ ```bash
59
+ # On first run, the service auto-initializes and downloads the default model
60
+ mano-asr start
61
+
62
+ # Transcribe / translate an audio file
63
+ mano-asr transcribe assets/BAC009S0764W0129.wav
64
+
65
+ ```
66
+
67
+ ### Python API
68
+
69
+ ```python
70
+ from core.auto_model import AutoModel
71
+
72
+ model = AutoModel(
73
+ model="models/Mininglamp-2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit",
74
+ vad_model="models/fsmn-vad-mlx", # optional: auto-segment long audio
75
+ )
76
+
77
+ text = model.generate(
78
+ "assets/BAC009S0764W0129.wav",
79
+ task="translate", # translation task
80
+ target_language="zh", # target language: Chinese
81
+ merge_vad=True,
82
+ )
83
+ print(text)
84
+ # -> "甚至出现交易几乎停滞的情况"
85
+ ```
86
+
87
+ ### HTTP API
88
+
89
+ ```bash
90
+ curl -X POST http://127.0.0.1:8787/v1/voice/transcribe \
91
+ -F "audio=@assets/BAC009S0764W0129.wav" \
92
+ -F "personal_context=## Terms\n- FastAPI\n- Kubernetes" \
93
+ -F "mode=smart"
94
+ ```
95
+
96
+ ```json
97
+ {
98
+ "status": 200,
99
+ "text": "transcribed text",
100
+ "m": "fun-asr-nano",
101
+ "engine": "mlx"
102
+ }
103
+ ```
104
+
105
+ > Full API fields, limits and auth are documented under [API](#en-api).
106
+
107
+ <a id="en-models"></a>
108
+
109
+ ## Models
110
+
111
+ mano-asr uses a pluggable engine design and supports several mainstream ASR base models. Switch with a single command: `mano-asr model use <name>`.
112
+
113
+ | Model | Engine | Base model | Quant | Size | Languages | Links |
114
+ | --- | --- | --- | --- | --- | --- | --- |
115
+ | **Mano-ASR-0.8B-Instruct-1.0-MLX-8bit** (default) | `funasr` | [Fun-ASR-Nano](https://github.com/FunAudioLLM/Fun-ASR) | - | 0.8 GB | ZH / EN | [HuggingFace](https://huggingface.co/Mininglamp-2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit) · [ModelScope](https://www.modelscope.cn/models/Mininglamp2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit) |
116
+
117
+ > The model is downloaded automatically from HuggingFace or ModelScope (China mirror); the source is chosen by network environment on first run.
118
+
119
+ <a id="en-install"></a>
120
+
121
+ ## Installation
122
+
123
+ ### Option 1: Homebrew (recommended)
124
+
125
+ ```bash
126
+ brew tap mano-asr/mano-asr
127
+ brew install mano-asr
128
+
129
+ # Start (first run auto-initializes + downloads the default model)
130
+ mano-asr start
131
+ mano-asr doctor # environment check
132
+ ```
133
+
134
+ ### Option 2: From source
135
+
136
+ ```bash
137
+ # 1. Dependency: ffmpeg (decodes non-WAV audio)
138
+ brew install ffmpeg
139
+
140
+ # 2. Clone + install
141
+ git clone https://github.com/Mininglamp-AI/mano-asr.git
142
+ cd mano-asr
143
+ python3 -m venv .venv && source .venv/bin/activate
144
+ pip install -U pip
145
+ pip install -e .
146
+
147
+ # 3. Download the model
148
+ hf download Mininglamp-2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit \
149
+ --local-dir models/Mininglamp-2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit
150
+
151
+ # Behind a China mirror:
152
+ # HF_ENDPOINT=https://hf-mirror.com hf download ...
153
+
154
+ # 4. Start the server
155
+ python3 server.py \
156
+ --model-path models/Mininglamp-2718/Mano-ASR-0.8B-Instruct-1.0-MLX-8bit \
157
+ --vad-model-path models/fsmn-vad-mlx \
158
+ --host 0.0.0.0 --port 8787 --load-on-startup
159
+ ```
160
+
161
+ **Requirements:** macOS (Apple Silicon) · Python 3.10+ · `ffmpeg` / `ffprobe` on `PATH`.
162
+
163
+ <a id="en-api"></a>
164
+
165
+ ## API
166
+
167
+ ### `POST /v1/voice/transcribe`
168
+
169
+ Transcribe a single uploaded audio file. Request type: `multipart/form-data`.
170
+
171
+ | Field | Type | Required | Description |
172
+ | --- | --- | --- | --- |
173
+ | `audio` | file | yes | Audio file. Supported: `.wav` `.mp3` `.ogg` `.webm` `.m4a` `.flac` |
174
+ | `context_text` | string | no | Existing text for append/edit modes; last 5000 chars kept |
175
+ | `chat_context` | string | no | Chat context; last 20000 chars kept |
176
+ | `personal_context` | string | no | Personal correction / hotword context; last 10000 chars kept |
177
+ | `member_context` | string | no | Member context; last 5000 chars kept |
178
+ | `mode` | string | no | `smart` / `append_only` / `edit_only`, default `smart` |
179
+
180
+ **Limits:** default max file `30 MiB`, max duration `660` s; `edit_only` requires `context_text`.
181
+
182
+ ### `GET /v1/voice/config`
183
+
184
+ Returns current service limits and engine metadata.
185
+
186
+ ```bash
187
+ curl http://127.0.0.1:8787/v1/voice/config
188
+ ```
189
+
190
+ ### Authentication
191
+
192
+ Disabled by default. If started with `--auth-token`, requests must carry `Authorization: Bearer <token>`.
193
+
194
+ ```bash
195
+ python3 server.py --model-path <path> --auth-token "$MANO_ASR_TOKEN"
196
+ ```
197
+
198
+ <a id="en-license"></a>
199
+
200
+ ## License
201
+
202
+ Released under the [MIT License](LICENSE).
203
+
204
+ Copyright (c) 2026 MININGLAMP Technology.
205
+
206
+ <a id="en-acknowledgments"></a>
207
+
208
+ ## Acknowledgments
209
+
210
+ mano-asr would not be possible without these excellent open-source projects:
211
+
212
+ - [**MLX**](https://github.com/ml-explore/mlx) & [**mlx-audio**](https://github.com/Blaizzy/mlx-audio) — Apple's machine-learning framework and audio toolkit, the foundation of mano-asr's local inference.
213
+ - [**FunASR / FunAudioLLM**](https://github.com/modelscope/FunASR) — source of Fun-ASR-Nano and FSMN-VAD, providing strong Chinese speech recognition.
214
+ - [**Qwen3**](https://github.com/QwenLM/Qwen3) — the base model behind the Qwen3-ASR engine.
215
+ - [**mlx-community**](https://huggingface.co/mlx-community) — high-quality MLX quantized models.
216
+ - [**ModelScope**](https://github.com/modelscope/modelscope) & [**Hugging Face**](https://huggingface.co/) — model hosting and distribution.
217
+ - [**FastAPI**](https://github.com/fastapi/fastapi) — high-performance web framework.
218
+
219
+ Thanks to everyone contributing to the open-source speech recognition community.
config.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "funasr",
3
+ "sample_rate": 16000,
4
+ "n_mels": 80,
5
+ "lfr_m": 7,
6
+ "lfr_n": 6,
7
+ "encoder": {
8
+ "input_dim": 560,
9
+ "encoder_dim": 512,
10
+ "num_heads": 4,
11
+ "ffn_dim": 2048,
12
+ "kernel_size": 11,
13
+ "num_encoders0": 1,
14
+ "num_encoders": 49,
15
+ "num_tp_encoders": 20,
16
+ "dropout": 0.0
17
+ },
18
+ "adaptor": {
19
+ "downsample_rate": 1,
20
+ "encoder_dim": 512,
21
+ "llm_dim": 1024,
22
+ "ffn_dim": 2048,
23
+ "n_layer": 2,
24
+ "attention_heads": 8,
25
+ "dropout": 0.0
26
+ },
27
+ "llm": {
28
+ "vocab_size": 151936,
29
+ "hidden_size": 1024,
30
+ "num_hidden_layers": 28,
31
+ "num_attention_heads": 16,
32
+ "num_key_value_heads": 8,
33
+ "intermediate_size": 3072,
34
+ "max_position_embeddings": 40960,
35
+ "rope_theta": 1000000.0,
36
+ "rms_norm_eps": 1e-06,
37
+ "tie_word_embeddings": false,
38
+ "head_dim": 128
39
+ },
40
+ "quantization": {
41
+ "bits": 8,
42
+ "group_size": 64,
43
+ "quantized_components": [
44
+ "llm.model.layers",
45
+ "audio_adaptor"
46
+ ]
47
+ }
48
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
3
+ size 11422654
tokenizer_config.json ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ },
181
+ "151665": {
182
+ "content": "<tool_response>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": false
188
+ },
189
+ "151666": {
190
+ "content": "</tool_response>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": false
196
+ },
197
+ "151667": {
198
+ "content": "<think>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": false
204
+ },
205
+ "151668": {
206
+ "content": "</think>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": false
212
+ }
213
+ },
214
+ "additional_special_tokens": [
215
+ "<|im_start|>",
216
+ "<|im_end|>",
217
+ "<|object_ref_start|>",
218
+ "<|object_ref_end|>",
219
+ "<|box_start|>",
220
+ "<|box_end|>",
221
+ "<|quad_start|>",
222
+ "<|quad_end|>",
223
+ "<|vision_start|>",
224
+ "<|vision_end|>",
225
+ "<|vision_pad|>",
226
+ "<|image_pad|>",
227
+ "<|video_pad|>"
228
+ ],
229
+ "bos_token": null,
230
+ "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0].role == 'system' %}\n {{- messages[0].content + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n {%- if messages[0].role == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0].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\" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n {%- if message.content is string %}\n {%- set content = message.content %}\n {%- else %}\n {%- set content = '' %}\n {%- endif %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\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 {%- if loop.index0 > ns.last_query_index %}\n {%- if loop.last or (not loop.last and reasoning_content) %}\n {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls %}\n {%- for tool_call in message.tool_calls %}\n {%- if (loop.first and content) or (not loop.first) %}\n {{- '\\n' }}\n {%- endif %}\n {%- if tool_call.function %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {%- if tool_call.arguments is string %}\n {{- tool_call.arguments }}\n {%- else %}\n {{- tool_call.arguments | tojson }}\n {%- endif %}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\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 {%- endif %}\n{%- endif %}",
231
+ "clean_up_tokenization_spaces": false,
232
+ "eos_token": "<|im_end|>",
233
+ "errors": "replace",
234
+ "model_max_length": 131072,
235
+ "pad_token": "<|endoftext|>",
236
+ "split_special_tokens": false,
237
+ "tokenizer_class": "Qwen2Tokenizer",
238
+ "unk_token": null
239
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff