Text Generation
Transformers
Safetensors
Chinese
English
qwen3
qwen3-asr
transcript-normalization
chinese
code-switching
full-finetuning
bf16
spelling-normalization
numbered-lists
conversational
text-generation-inference
Instructions to use WIlfLin/S1-mini-Chinese-1106-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use WIlfLin/S1-mini-Chinese-1106-BF16 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="WIlfLin/S1-mini-Chinese-1106-BF16") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("WIlfLin/S1-mini-Chinese-1106-BF16") model = AutoModelForCausalLM.from_pretrained("WIlfLin/S1-mini-Chinese-1106-BF16", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=256) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use WIlfLin/S1-mini-Chinese-1106-BF16 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "WIlfLin/S1-mini-Chinese-1106-BF16" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "WIlfLin/S1-mini-Chinese-1106-BF16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/WIlfLin/S1-mini-Chinese-1106-BF16
- SGLang
How to use WIlfLin/S1-mini-Chinese-1106-BF16 with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "WIlfLin/S1-mini-Chinese-1106-BF16" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "WIlfLin/S1-mini-Chinese-1106-BF16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "WIlfLin/S1-mini-Chinese-1106-BF16" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "WIlfLin/S1-mini-Chinese-1106-BF16", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use WIlfLin/S1-mini-Chinese-1106-BF16 with Docker Model Runner:
docker model run hf.co/WIlfLin/S1-mini-Chinese-1106-BF16
Release BF16 stability fine-tune with same-runtime evaluation and limitations
Browse files- .gitattributes +1 -0
- LICENSE +217 -0
- NOTICE +11 -0
- README.md +120 -0
- chat_template.jinja +89 -0
- config.json +63 -0
- data-provenance.json +106 -0
- evaluation-summary.json +175 -0
- generation_config.json +11 -0
- model.safetensors +3 -0
- tokenizer.json +3 -0
- tokenizer_config.json +29 -0
- training-provenance.json +26 -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
|
LICENSE
ADDED
|
@@ -0,0 +1,217 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Copyright 2026 Superwhisper
|
| 2 |
+
|
| 3 |
+
This model is a derivative work of Qwen3-0.6B, Copyright 2024 Alibaba Cloud,
|
| 4 |
+
licensed under the Apache License, Version 2.0.
|
| 5 |
+
|
| 6 |
+
Apache License
|
| 7 |
+
Version 2.0, January 2004
|
| 8 |
+
http://www.apache.org/licenses/
|
| 9 |
+
|
| 10 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 11 |
+
|
| 12 |
+
1. Definitions.
|
| 13 |
+
|
| 14 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 15 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 16 |
+
|
| 17 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 18 |
+
the copyright owner that is granting the License.
|
| 19 |
+
|
| 20 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 21 |
+
other entities that control, are controlled by, or are under common
|
| 22 |
+
control with that entity. For the purposes of this definition,
|
| 23 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 24 |
+
direction or management of such entity, whether by contract or
|
| 25 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 26 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 27 |
+
|
| 28 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 29 |
+
exercising permissions granted by this License.
|
| 30 |
+
|
| 31 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 32 |
+
including but not limited to software source code, documentation
|
| 33 |
+
source, and configuration files.
|
| 34 |
+
|
| 35 |
+
"Object" form shall mean any form resulting from mechanical
|
| 36 |
+
transformation or translation of a Source form, including but
|
| 37 |
+
not limited to compiled object code, generated documentation,
|
| 38 |
+
and conversions to other media types.
|
| 39 |
+
|
| 40 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 41 |
+
Object form, made available under the License, as indicated by a
|
| 42 |
+
copyright notice that is included in or attached to the work
|
| 43 |
+
(an example is provided in the Appendix below).
|
| 44 |
+
|
| 45 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 46 |
+
form, that is based on (or derived from) the Work and for which the
|
| 47 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 48 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 49 |
+
of this License, Derivative Works shall not include works that remain
|
| 50 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 51 |
+
the Work and Derivative Works thereof.
|
| 52 |
+
|
| 53 |
+
"Contribution" shall mean any work of authorship, including
|
| 54 |
+
the original version of the Work and any modifications or additions
|
| 55 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 56 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 57 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 58 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 59 |
+
means any form of electronic, verbal, or written communication sent
|
| 60 |
+
to the Licensor or its representatives, including but not limited to
|
| 61 |
+
communication on electronic mailing lists, source code control systems,
|
| 62 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 63 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 64 |
+
excluding communication that is conspicuously marked or otherwise
|
| 65 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 66 |
+
|
| 67 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 68 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 69 |
+
subsequently incorporated within the Work.
|
| 70 |
+
|
| 71 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 72 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 73 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 74 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 75 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 76 |
+
Work and such Derivative Works in Source or Object form.
|
| 77 |
+
|
| 78 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 79 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 80 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 81 |
+
(except as stated in this section) patent license to make, have made,
|
| 82 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 83 |
+
where such license applies only to those patent claims licensable
|
| 84 |
+
by such Contributor that are necessarily infringed by their
|
| 85 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 86 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 87 |
+
institute patent litigation against any entity (including a
|
| 88 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 89 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 90 |
+
or contributory patent infringement, then any patent licenses
|
| 91 |
+
granted to You under this License for that Work shall terminate
|
| 92 |
+
as of the date such litigation is filed.
|
| 93 |
+
|
| 94 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 95 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 96 |
+
modifications, and in Source or Object form, provided that You
|
| 97 |
+
meet the following conditions:
|
| 98 |
+
|
| 99 |
+
(a) You must give any other recipients of the Work or
|
| 100 |
+
Derivative Works a copy of this License; and
|
| 101 |
+
|
| 102 |
+
(b) You must cause any modified files to carry prominent notices
|
| 103 |
+
stating that You changed the files; and
|
| 104 |
+
|
| 105 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 106 |
+
that You distribute, all copyright, patent, trademark, and
|
| 107 |
+
attribution notices from the Source form of the Work,
|
| 108 |
+
excluding those notices that do not pertain to any part of
|
| 109 |
+
the Derivative Works; and
|
| 110 |
+
|
| 111 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 112 |
+
distribution, then any Derivative Works that You distribute must
|
| 113 |
+
include a readable copy of the attribution notices contained
|
| 114 |
+
within such NOTICE file, excluding those notices that do not
|
| 115 |
+
pertain to any part of the Derivative Works, in at least one
|
| 116 |
+
of the following places: within a NOTICE text file distributed
|
| 117 |
+
as part of the Derivative Works; within the Source form or
|
| 118 |
+
documentation, if provided along with the Derivative Works; or,
|
| 119 |
+
within a display generated by the Derivative Works, if and
|
| 120 |
+
wherever such third-party notices normally appear. The contents
|
| 121 |
+
of the NOTICE file are for informational purposes only and
|
| 122 |
+
do not modify the License. You may add Your own attribution
|
| 123 |
+
notices within Derivative Works that You distribute, alongside
|
| 124 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 125 |
+
that such additional attribution notices cannot be construed
|
| 126 |
+
as modifying the License.
|
| 127 |
+
|
| 128 |
+
You may add Your own copyright statement to Your modifications and
|
| 129 |
+
may provide additional or different license terms and conditions
|
| 130 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 131 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 132 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 133 |
+
the conditions stated in this License.
|
| 134 |
+
|
| 135 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 136 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 137 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 138 |
+
this License, without any additional terms or conditions.
|
| 139 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 140 |
+
the terms of any separate license agreement you may have executed
|
| 141 |
+
with Licensor regarding such Contributions.
|
| 142 |
+
|
| 143 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 144 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 145 |
+
except as required for reasonable and customary use in describing the
|
| 146 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 147 |
+
|
| 148 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 149 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 150 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 151 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 152 |
+
implied, including, without limitation, any warranties or conditions
|
| 153 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 154 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 155 |
+
appropriateness of using or redistributing the Work and assume any
|
| 156 |
+
risks associated with Your exercise of permissions under this License.
|
| 157 |
+
|
| 158 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 159 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 160 |
+
unless required by applicable law (such as deliberate and grossly
|
| 161 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 162 |
+
liable to You for damages, including any direct, indirect, special,
|
| 163 |
+
incidental, or consequential damages of any character arising as a
|
| 164 |
+
result of this License or out of the use or inability to use the
|
| 165 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 166 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 167 |
+
other commercial damages or losses), even if such Contributor
|
| 168 |
+
has been advised of the possibility of such damages.
|
| 169 |
+
|
| 170 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 171 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 172 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 173 |
+
or other liability obligations and/or rights consistent with this
|
| 174 |
+
License. However, in accepting such obligations, You may act only
|
| 175 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 176 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 177 |
+
defend, and hold each Contributor harmless for any liability
|
| 178 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 179 |
+
of your accepting any such warranty or additional liability.
|
| 180 |
+
|
| 181 |
+
END OF TERMS AND CONDITIONS
|
| 182 |
+
|
| 183 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 184 |
+
|
| 185 |
+
To apply the Apache License to your work, attach the following
|
| 186 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 187 |
+
replaced with your own identifying information. (Don't include
|
| 188 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 189 |
+
comment syntax for the file format. We also recommend that a
|
| 190 |
+
file or class name and description of purpose be included on the
|
| 191 |
+
same "printed page" as the copyright notice for easier
|
| 192 |
+
identification within third-party archives.
|
| 193 |
+
|
| 194 |
+
Copyright [yyyy] [name of copyright owner]
|
| 195 |
+
|
| 196 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 197 |
+
you may not use this file except in compliance with the License.
|
| 198 |
+
You may obtain a copy of the License at
|
| 199 |
+
|
| 200 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 201 |
+
|
| 202 |
+
Unless required by applicable law or agreed to in writing, software
|
| 203 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 204 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 205 |
+
See the License for the specific language governing permissions and
|
| 206 |
+
limitations under the License.
|
| 207 |
+
|
| 208 |
+
------------------------------------------------------------------------
|
| 209 |
+
|
| 210 |
+
ADDITIONAL TERM
|
| 211 |
+
|
| 212 |
+
In addition to the terms of the Apache License, Version 2.0 above: any
|
| 213 |
+
use, distribution, or integration of this model, whether unmodified or
|
| 214 |
+
as part of a derivative work or product, must continue to identify it
|
| 215 |
+
by its original name, "S1-mini" by "Superwhisper", using that exact
|
| 216 |
+
capitalization, regardless of any other name under which the model or
|
| 217 |
+
a product incorporating it is marketed or distributed.
|
NOTICE
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
S1-mini
|
| 2 |
+
Copyright 2026 Superwhisper
|
| 3 |
+
|
| 4 |
+
This work is a derivative of Qwen3-0.6B, Copyright 2024 Alibaba Cloud,
|
| 5 |
+
licensed under the Apache License, Version 2.0.
|
| 6 |
+
|
| 7 |
+
Any use, distribution, or integration of this model, whether unmodified or
|
| 8 |
+
as part of a derivative work or product, must continue to identify it by its
|
| 9 |
+
original name, "S1-mini" by "Superwhisper", using that exact capitalization.
|
| 10 |
+
This restates the ADDITIONAL TERM of the LICENSE file, which is the
|
| 11 |
+
operative text.
|
README.md
ADDED
|
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: [zh, en]
|
| 3 |
+
license: other
|
| 4 |
+
license_name: s1-mini-license
|
| 5 |
+
license_link: LICENSE
|
| 6 |
+
base_model: WIlfLin/S1-mini-Chinese-1106
|
| 7 |
+
library_name: transformers
|
| 8 |
+
pipeline_tag: text-generation
|
| 9 |
+
tags: [qwen3-asr, transcript-normalization, chinese, code-switching, full-finetuning, bf16, spelling-normalization, numbered-lists]
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# S1-mini-Chinese-1106-BF16
|
| 13 |
+
|
| 14 |
+
**专门为 Qwen3-ASR 中文及中英混合转写优化的文本后处理模型。**
|
| 15 |
+
本版本在中文数据及合成对照数据上继续全量微调,重点改善 BF16 下的数字转写稳定性。
|
| 16 |
+
基于 **S1-mini** by **Superwhisper**,保留原名称、署名、LICENSE 和 NOTICE。
|
| 17 |
+
这是新的 `stability-margin-r2` 权重,不是此前 `S1-mini-Chinese-1106`(native-v10)的格式改名。
|
| 18 |
+
|
| 19 |
+
Chinese-data continued full fine-tuning of **S1-mini** by **Superwhisper**, optimized for Qwen3-ASR Chinese and Chinese/English mixed transcripts. This BF16 stability release is a new checkpoint. It processes text, not audio, and is not a general-purpose assistant.
|
| 20 |
+
|
| 21 |
+
## 改动
|
| 22 |
+
|
| 23 |
+
- 增加数值组合、内部零、进位和时间写法的训练。
|
| 24 |
+
- 增加数字和拼读字符的目标分数间隔训练,降低候选分数过近时的错误。
|
| 25 |
+
- 保留模型自身的去口水词、去重复、明确改口、书面大小写、中文人名、编号列表和换行能力。
|
| 26 |
+
- 基础数学使用普通文本 `+ - * / ^ =`;不主动计算未口述答案,也不要求 LaTeX。
|
| 27 |
+
- 不附带拼读前置规则或内容保护过滤器。模型输出直接交给调用方。
|
| 28 |
+
|
| 29 |
+
这些是训练目标,不是任意输入上的正确性保证。调用程序负责请求异常、截断检测和安全提交换行。
|
| 30 |
+
|
| 31 |
+
## 训练
|
| 32 |
+
|
| 33 |
+
全量训练 **751,632,384** 个参数,不是 LoRA,也不是仅训练末层。
|
| 34 |
+
|
| 35 |
+
1. 在 native-v10 上使用 5,109 条回放与对照数据,320 步、学习率 5e-7。
|
| 36 |
+
2. 同数据/种子/步数另做 FP32 教师蒸馏对照。两组新开发集同分,普通监督组公开回归更好,因此选普通监督组。
|
| 37 |
+
3. 再加入 2,048 条通用数字组合与进位对照,共 7,157 条;448 步、学习率 1e-6。损失为目标文本 CE、可信教师 KL 和关键字符的 margin loss。
|
| 38 |
+
|
| 39 |
+
最终阶段采用 effective batch16、FP32 主参数/教师、BF16 student autocast,关闭 TF32。仅监督目标与 EOS,屏蔽提示和填充。教师下一 token 必须与标注一致且概率至少0.9;关键数字/ASCII token 的目标 logit 间隔为1,margin 权重0.2、KL 权重0.1。导出权重为 BF16。
|
| 40 |
+
|
| 41 |
+
合成数字经过独立解析检查。训练池包含原能力回放与空输出监督。来源含 CS2W、Disfl-QA、自制及模型辅助合成数据;合成标注不等于独立人工标注。见 [training-provenance.json](training-provenance.json) 和 [data-provenance.json](data-provenance.json)。不发布训练正文、私有录音或用户输入历史。
|
| 42 |
+
|
| 43 |
+
## 同后端测试
|
| 44 |
+
|
| 45 |
+
下表两组均为 **vLLM0.17、Blackwell SM120、BF16、greedy、max-num-seqs8、eager、prefix cache关闭、最大生成256 tokens**。不能与其他后端的结果直接混用。
|
| 46 |
+
|
| 47 |
+
| 测试 | 原 native-v10 | 本版本 |
|
| 48 |
+
|---|---:|---:|
|
| 49 |
+
| 核心拼读,严格匹配 | 8/8 | 8/8 |
|
| 50 |
+
| 基础能力,词法匹配 | 45/48 | 45/48 |
|
| 51 |
+
| 格式与数字,词法匹配 | 20/20 | 20/20 |
|
| 52 |
+
| 中文人名,词法匹配 | 60/60 | 60/60 |
|
| 53 |
+
| 拼读语境,严格匹配 | 311/326 | 311/326 |
|
| 54 |
+
| 生造词拼读,严格匹配 | 196/200 | 196/200 |
|
| 55 |
+
| 基础数学,词法匹配 | 233/236 | 234/236 |
|
| 56 |
+
| 数字/时间语境A,严格匹配 | 372/432 | 408/432 |
|
| 57 |
+
| 数字/时间语境B,严格匹配 | 367/432 | 424/432 |
|
| 58 |
+
| 公开文本,词法匹配 | 228/400 | 226/400 |
|
| 59 |
+
| 公开文本词法 CER,越低越好 | 10.1373% | 10.1593% |
|
| 60 |
+
|
| 61 |
+
本次2,167条实际 vLLM 生成无截断。格式检查额外核验所需换行;另做单请求5条关键检查,5/5严格匹配。
|
| 62 |
+
|
| 63 |
+
**收益主要在数字语境,公开文本有轻微退步,不是所有能力均提升。** 整句匹配删除非字母数字字符时称“词法匹配”,保留大小写与数字;严格匹配保留标点及内部空白。它们不是人工语义正确率,也不是上游所称的94.8% token accuracy。
|
| 64 |
+
|
| 65 |
+
这些集合经过开发中的反复使用,均应视为**开发/回归证据**。语境B最初在选择第一轮模型后测试,但后续修订已参考其错误类型,因此最终版本不再称其为独立盲测。数字值与语法家族也可能已在更早训练中出现。完整结果见 [evaluation-summary.json](evaluation-summary.json)。
|
| 66 |
+
|
| 67 |
+
## 限制
|
| 68 |
+
|
| 69 |
+
- 数字仍可能多补一个零;语境B的8条参考差异中有4条数值错误及4条中文/阿拉伯数字风格差异。
|
| 70 |
+
- 拼读、专有词、否定、改口和长段落仍可能整理错误。
|
| 71 |
+
- 部分早期合成千位数省略了规范口述中的“零”,数据分布不代表自然 ASR。
|
| 72 |
+
- 浮点舍入、批次、后端和硬件仍可能改变输出。训练降低风险,不提供任意配置下逐位一致的保证。
|
| 73 |
+
- 当前验证是本地文本推理,不是��然录音、实体按键或所有输入框的验收。
|
| 74 |
+
- 发布检查没有替换现用语音服务。其他硬件与 Orin 的本次完整回归未验证。
|
| 75 |
+
|
| 76 |
+
## 使用
|
| 77 |
+
|
| 78 |
+
使用附带 tokenizer/template,关闭 thinking。普通文本选择 `Structure: prose`,明确列表选择 `Structure: lists`。
|
| 79 |
+
|
| 80 |
+
```python
|
| 81 |
+
import torch
|
| 82 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 83 |
+
|
| 84 |
+
model_id = "WIlfLin/S1-mini-Chinese-1106-BF16"
|
| 85 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 86 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 87 |
+
model_id, dtype=torch.bfloat16, device_map="auto"
|
| 88 |
+
).eval()
|
| 89 |
+
messages = [
|
| 90 |
+
{"role": "system", "content": (
|
| 91 |
+
"You are a text normalizer for speech-to-text transcripts. The input begins "
|
| 92 |
+
"with a control line specifying the styling, structure, and context settings; "
|
| 93 |
+
"clean the transcript to match those settings and output only the cleaned text."
|
| 94 |
+
)},
|
| 95 |
+
{"role": "user", "content": (
|
| 96 |
+
"[Styling: semi-formal] [Structure: prose] [Context: general]\n"
|
| 97 |
+
"嗯,这份报告这份报告明天再交。"
|
| 98 |
+
)},
|
| 99 |
+
]
|
| 100 |
+
text = tokenizer.apply_chat_template(
|
| 101 |
+
messages, tokenize=False, add_generation_prompt=True, enable_thinking=False
|
| 102 |
+
)
|
| 103 |
+
inputs = tokenizer(text, return_tensors="pt").to(model.device)
|
| 104 |
+
with torch.inference_mode():
|
| 105 |
+
output = model.generate(
|
| 106 |
+
**inputs, do_sample=False, max_new_tokens=256,
|
| 107 |
+
pad_token_id=tokenizer.eos_token_id,
|
| 108 |
+
)
|
| 109 |
+
print(tokenizer.decode(
|
| 110 |
+
output[0, inputs.input_ids.shape[1]:], skip_special_tokens=True
|
| 111 |
+
).strip())
|
| 112 |
+
```
|
| 113 |
+
|
| 114 |
+
## 许可证与来源署名
|
| 115 |
+
|
| 116 |
+
本模型是 **S1-mini** by **Superwhisper** 的修改版本,底层架构为 Qwen3-0.6B。
|
| 117 |
+
保留原始 [LICENSE](LICENSE) 和 [NOTICE](NOTICE),包括保留原模型名称及作者署名的附加条款。
|
| 118 |
+
CS2W 声明 CC-BY-SA-4.0,Disfl-QA 声明 CC-BY-4.0;不把数据条款改称为纯 Apache-2.0。
|
| 119 |
+
|
| 120 |
+
权重 SHA256:`c0d2cf58468359e0a44970083f9ce38078208c291430d20a5c685c5ce6759312`。
|
chat_template.jinja
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- if tools %}
|
| 2 |
+
{{- '<|im_start|>system\n' }}
|
| 3 |
+
{%- if messages[0].role == 'system' %}
|
| 4 |
+
{{- messages[0].content + '\n\n' }}
|
| 5 |
+
{%- endif %}
|
| 6 |
+
{{- "# 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>" }}
|
| 7 |
+
{%- for tool in tools %}
|
| 8 |
+
{{- "\n" }}
|
| 9 |
+
{{- tool | tojson }}
|
| 10 |
+
{%- endfor %}
|
| 11 |
+
{{- "\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" }}
|
| 12 |
+
{%- else %}
|
| 13 |
+
{%- if messages[0].role == 'system' %}
|
| 14 |
+
{{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
|
| 15 |
+
{%- endif %}
|
| 16 |
+
{%- endif %}
|
| 17 |
+
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
|
| 18 |
+
{%- for message in messages[::-1] %}
|
| 19 |
+
{%- set index = (messages|length - 1) - loop.index0 %}
|
| 20 |
+
{%- 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>')) %}
|
| 21 |
+
{%- set ns.multi_step_tool = false %}
|
| 22 |
+
{%- set ns.last_query_index = index %}
|
| 23 |
+
{%- endif %}
|
| 24 |
+
{%- endfor %}
|
| 25 |
+
{%- for message in messages %}
|
| 26 |
+
{%- if message.content is string %}
|
| 27 |
+
{%- set content = message.content %}
|
| 28 |
+
{%- else %}
|
| 29 |
+
{%- set content = '' %}
|
| 30 |
+
{%- endif %}
|
| 31 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
|
| 32 |
+
{{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
|
| 33 |
+
{%- elif message.role == "assistant" %}
|
| 34 |
+
{%- set reasoning_content = '' %}
|
| 35 |
+
{%- if message.reasoning_content is string %}
|
| 36 |
+
{%- set reasoning_content = message.reasoning_content %}
|
| 37 |
+
{%- else %}
|
| 38 |
+
{%- if '</think>' in content %}
|
| 39 |
+
{%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
|
| 40 |
+
{%- set content = content.split('</think>')[-1].lstrip('\n') %}
|
| 41 |
+
{%- endif %}
|
| 42 |
+
{%- endif %}
|
| 43 |
+
{%- if loop.index0 > ns.last_query_index %}
|
| 44 |
+
{%- if loop.last or (not loop.last and reasoning_content) %}
|
| 45 |
+
{{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
|
| 46 |
+
{%- else %}
|
| 47 |
+
{{- '<|im_start|>' + message.role + '\n' + content }}
|
| 48 |
+
{%- endif %}
|
| 49 |
+
{%- else %}
|
| 50 |
+
{{- '<|im_start|>' + message.role + '\n' + content }}
|
| 51 |
+
{%- endif %}
|
| 52 |
+
{%- if message.tool_calls %}
|
| 53 |
+
{%- for tool_call in message.tool_calls %}
|
| 54 |
+
{%- if (loop.first and content) or (not loop.first) %}
|
| 55 |
+
{{- '\n' }}
|
| 56 |
+
{%- endif %}
|
| 57 |
+
{%- if tool_call.function %}
|
| 58 |
+
{%- set tool_call = tool_call.function %}
|
| 59 |
+
{%- endif %}
|
| 60 |
+
{{- '<tool_call>\n{"name": "' }}
|
| 61 |
+
{{- tool_call.name }}
|
| 62 |
+
{{- '", "arguments": ' }}
|
| 63 |
+
{%- if tool_call.arguments is string %}
|
| 64 |
+
{{- tool_call.arguments }}
|
| 65 |
+
{%- else %}
|
| 66 |
+
{{- tool_call.arguments | tojson }}
|
| 67 |
+
{%- endif %}
|
| 68 |
+
{{- '}\n</tool_call>' }}
|
| 69 |
+
{%- endfor %}
|
| 70 |
+
{%- endif %}
|
| 71 |
+
{{- '<|im_end|>\n' }}
|
| 72 |
+
{%- elif message.role == "tool" %}
|
| 73 |
+
{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
|
| 74 |
+
{{- '<|im_start|>user' }}
|
| 75 |
+
{%- endif %}
|
| 76 |
+
{{- '\n<tool_response>\n' }}
|
| 77 |
+
{{- content }}
|
| 78 |
+
{{- '\n</tool_response>' }}
|
| 79 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 80 |
+
{{- '<|im_end|>\n' }}
|
| 81 |
+
{%- endif %}
|
| 82 |
+
{%- endif %}
|
| 83 |
+
{%- endfor %}
|
| 84 |
+
{%- if add_generation_prompt %}
|
| 85 |
+
{{- '<|im_start|>assistant\n' }}
|
| 86 |
+
{%- if enable_thinking is defined and enable_thinking is false %}
|
| 87 |
+
{{- '<think>\n\n</think>\n\n' }}
|
| 88 |
+
{%- endif %}
|
| 89 |
+
{%- endif %}
|
config.json
ADDED
|
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen3ForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 151643,
|
| 8 |
+
"dtype": "bfloat16",
|
| 9 |
+
"eos_token_id": 151645,
|
| 10 |
+
"head_dim": 128,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1024,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"layer_types": [
|
| 16 |
+
"full_attention",
|
| 17 |
+
"full_attention",
|
| 18 |
+
"full_attention",
|
| 19 |
+
"full_attention",
|
| 20 |
+
"full_attention",
|
| 21 |
+
"full_attention",
|
| 22 |
+
"full_attention",
|
| 23 |
+
"full_attention",
|
| 24 |
+
"full_attention",
|
| 25 |
+
"full_attention",
|
| 26 |
+
"full_attention",
|
| 27 |
+
"full_attention",
|
| 28 |
+
"full_attention",
|
| 29 |
+
"full_attention",
|
| 30 |
+
"full_attention",
|
| 31 |
+
"full_attention",
|
| 32 |
+
"full_attention",
|
| 33 |
+
"full_attention",
|
| 34 |
+
"full_attention",
|
| 35 |
+
"full_attention",
|
| 36 |
+
"full_attention",
|
| 37 |
+
"full_attention",
|
| 38 |
+
"full_attention",
|
| 39 |
+
"full_attention",
|
| 40 |
+
"full_attention",
|
| 41 |
+
"full_attention",
|
| 42 |
+
"full_attention",
|
| 43 |
+
"full_attention"
|
| 44 |
+
],
|
| 45 |
+
"max_position_embeddings": 40960,
|
| 46 |
+
"max_window_layers": 28,
|
| 47 |
+
"model_type": "qwen3",
|
| 48 |
+
"num_attention_heads": 16,
|
| 49 |
+
"num_hidden_layers": 28,
|
| 50 |
+
"num_key_value_heads": 8,
|
| 51 |
+
"pad_token_id": null,
|
| 52 |
+
"rms_norm_eps": 1e-06,
|
| 53 |
+
"rope_parameters": {
|
| 54 |
+
"rope_theta": 1000000,
|
| 55 |
+
"rope_type": "default"
|
| 56 |
+
},
|
| 57 |
+
"sliding_window": null,
|
| 58 |
+
"tie_word_embeddings": false,
|
| 59 |
+
"transformers_version": "5.3.0",
|
| 60 |
+
"use_cache": true,
|
| 61 |
+
"use_sliding_window": false,
|
| 62 |
+
"vocab_size": 151936
|
| 63 |
+
}
|
data-provenance.json
ADDED
|
@@ -0,0 +1,106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"inherited_sources": {
|
| 3 |
+
"CS2W": {
|
| 4 |
+
"source": "https://github.com/guozishan/CS2W",
|
| 5 |
+
"revision": "be6e0519b433cfa9d8388b23308f1380716254fd",
|
| 6 |
+
"license_declared": "CC-BY-SA-4.0"
|
| 7 |
+
},
|
| 8 |
+
"Disfl-QA": {
|
| 9 |
+
"source": "https://github.com/google-research-datasets/Disfl-QA",
|
| 10 |
+
"revision": "1f0c16171c77b3d3408be92c485f11b8998a9189",
|
| 11 |
+
"license_declared": "CC-BY-4.0 for dataset"
|
| 12 |
+
}
|
| 13 |
+
},
|
| 14 |
+
"stability_training": {
|
| 15 |
+
"seed": 20261108,
|
| 16 |
+
"splits": {
|
| 17 |
+
"train": {
|
| 18 |
+
"rows": 5109,
|
| 19 |
+
"categories": {
|
| 20 |
+
"math_equation_wrong_rhs": 102,
|
| 21 |
+
"chinese_name_script_fidelity": 800,
|
| 22 |
+
"numeric_equation": 486,
|
| 23 |
+
"basic_replay": 1416,
|
| 24 |
+
"punct_question_preserve": 84,
|
| 25 |
+
"math_equation_correct": 207,
|
| 26 |
+
"spell_novel_fidelity": 73,
|
| 27 |
+
"number_idiom_preservation": 10,
|
| 28 |
+
"repair_bare_ordinary_titlecase": 108,
|
| 29 |
+
"spell_bare_input": 68,
|
| 30 |
+
"repair_perturbed_nonce_fidelity": 26,
|
| 31 |
+
"clock_zero": 108,
|
| 32 |
+
"punct_pause_clause_join": 96,
|
| 33 |
+
"math_unfinished_dangling": 78,
|
| 34 |
+
"spell_phonetic_near_repair": 149,
|
| 35 |
+
"list_ordinal_non_expansion": 70,
|
| 36 |
+
"spell_caps_non_spelling": 5,
|
| 37 |
+
"math_expression_no_eq": 80,
|
| 38 |
+
"spell_explicit_display": 78,
|
| 39 |
+
"repair_list_headless_numbered": 114,
|
| 40 |
+
"spell_cued_context": 127,
|
| 41 |
+
"local_known_initialism_written_case": 31,
|
| 42 |
+
"spell_semantic_polysemy": 18,
|
| 43 |
+
"number_fidelity": 162,
|
| 44 |
+
"math_self_corrected": 23,
|
| 45 |
+
"spell_neutral_context": 104,
|
| 46 |
+
"list_numbered_questions": 160,
|
| 47 |
+
"repair_mid_sentence_ordinary_lowercase": 19,
|
| 48 |
+
"repair_explicit_display_paraphrase": 25,
|
| 49 |
+
"repair_bare_initialism_allcaps": 38,
|
| 50 |
+
"math_question_preserve": 64,
|
| 51 |
+
"repair_list_weekday_fidelity": 36,
|
| 52 |
+
"repair_sentence_initial_spelled": 14,
|
| 53 |
+
"math_non_math_contrast": 46,
|
| 54 |
+
"repair_known_entity_phonetic_repair": 31,
|
| 55 |
+
"repair_known_entity_casing": 17,
|
| 56 |
+
"repair_list_compact_numbered": 36
|
| 57 |
+
},
|
| 58 |
+
"sha256": "abb3cf781937462b97e6861d3e07617035da47304b37d6c121a707ef3bbb67f2"
|
| 59 |
+
},
|
| 60 |
+
"dev": {
|
| 61 |
+
"rows": 432,
|
| 62 |
+
"categories": {
|
| 63 |
+
"number_fidelity": 162,
|
| 64 |
+
"numeric_equation": 162,
|
| 65 |
+
"clock_zero": 108
|
| 66 |
+
},
|
| 67 |
+
"sha256": "70b4bd13c550cee5de4cc785804f23d6ed1bc7e3c61c15d88863d7b59711341d"
|
| 68 |
+
},
|
| 69 |
+
"heldout": {
|
| 70 |
+
"rows": 432,
|
| 71 |
+
"categories": {
|
| 72 |
+
"number_fidelity": 162,
|
| 73 |
+
"clock_zero": 108,
|
| 74 |
+
"numeric_equation": 162
|
| 75 |
+
},
|
| 76 |
+
"sha256": "0d7b20403895af61b870b49abea5dc6444a8f3c1579fd819c84cb11e046eecab"
|
| 77 |
+
}
|
| 78 |
+
},
|
| 79 |
+
"independent_source_math_checks": 810,
|
| 80 |
+
"mutation_checks": 5,
|
| 81 |
+
"no_cross_split_input_overlap": true,
|
| 82 |
+
"no_old_diagnostic_inputs": true,
|
| 83 |
+
"scope": "Synthetic new carrier contexts; values and grammar families may have appeared in older training. Not human gold or unseen-value proof.",
|
| 84 |
+
"replay_scope": "Old capability replay enters training only. Dev/heldout quantify new numeric/clock contexts, not all language skills."
|
| 85 |
+
},
|
| 86 |
+
"margin_refinement": {
|
| 87 |
+
"rows": 7157,
|
| 88 |
+
"additions": 2048,
|
| 89 |
+
"seed": 20261109,
|
| 90 |
+
"source_sha256": "abb3cf781937462b97e6861d3e07617035da47304b37d6c121a707ef3bbb67f2",
|
| 91 |
+
"sha256": "a10d979b2f8a7a8c043d183d200b7467c616f427f4d15de8852f8ba86011117e",
|
| 92 |
+
"scope": "Existing v11 replay plus new general numeric composition/carry contrasts; old heldout now reused development after first evaluation. No new independent blind score claimed.",
|
| 93 |
+
"independent_numeric_checks": 6144,
|
| 94 |
+
"example_standard_renderings": {
|
| 95 |
+
"101": "一百零一",
|
| 96 |
+
"110": "一百一十",
|
| 97 |
+
"111": "一百一十一",
|
| 98 |
+
"1001": "一千零一",
|
| 99 |
+
"1010": "一千零一十",
|
| 100 |
+
"1014": "一千零一十四",
|
| 101 |
+
"1100": "一千一百",
|
| 102 |
+
"9009": "九千零九"
|
| 103 |
+
}
|
| 104 |
+
},
|
| 105 |
+
"policy": "No private audio, user input histories, corpus text, operational addresses or credentials redistributed. Synthetic labels are not independent human annotation. All displayed diagnostics are development/regression after repeated inspection."
|
| 106 |
+
}
|
evaluation-summary.json
ADDED
|
@@ -0,0 +1,175 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"all_passed": true,
|
| 3 |
+
"gate_revision": "GATES-v3.json",
|
| 4 |
+
"failed": [],
|
| 5 |
+
"baseline": {
|
| 6 |
+
"core": {
|
| 7 |
+
"n": 8,
|
| 8 |
+
"exact": 8,
|
| 9 |
+
"lexical": 8,
|
| 10 |
+
"truncated": 0,
|
| 11 |
+
"cer": 0.0
|
| 12 |
+
},
|
| 13 |
+
"basic48": {
|
| 14 |
+
"n": 48,
|
| 15 |
+
"exact": 43,
|
| 16 |
+
"lexical": 45,
|
| 17 |
+
"truncated": 0,
|
| 18 |
+
"cer": 0.006033182503770739
|
| 19 |
+
},
|
| 20 |
+
"format20": {
|
| 21 |
+
"n": 20,
|
| 22 |
+
"exact": 19,
|
| 23 |
+
"lexical": 20,
|
| 24 |
+
"truncated": 0,
|
| 25 |
+
"cer": 0.0
|
| 26 |
+
},
|
| 27 |
+
"names60": {
|
| 28 |
+
"n": 60,
|
| 29 |
+
"exact": 59,
|
| 30 |
+
"lexical": 60,
|
| 31 |
+
"truncated": 0,
|
| 32 |
+
"cer": 0.0
|
| 33 |
+
},
|
| 34 |
+
"public400": {
|
| 35 |
+
"n": 400,
|
| 36 |
+
"exact": 112,
|
| 37 |
+
"lexical": 228,
|
| 38 |
+
"truncated": 0,
|
| 39 |
+
"cer": 0.1013728720483251
|
| 40 |
+
},
|
| 41 |
+
"intent": {
|
| 42 |
+
"n": 326,
|
| 43 |
+
"exact": 311,
|
| 44 |
+
"lexical": 311,
|
| 45 |
+
"truncated": 0,
|
| 46 |
+
"cer": 0.005494505494505495
|
| 47 |
+
},
|
| 48 |
+
"nonce": {
|
| 49 |
+
"n": 200,
|
| 50 |
+
"exact": 196,
|
| 51 |
+
"lexical": 196,
|
| 52 |
+
"truncated": 0,
|
| 53 |
+
"cer": 0.003691171947093202
|
| 54 |
+
},
|
| 55 |
+
"math236": {
|
| 56 |
+
"n": 236,
|
| 57 |
+
"exact": 232,
|
| 58 |
+
"lexical": 233,
|
| 59 |
+
"truncated": 0,
|
| 60 |
+
"cer": 0.0019704433497536944
|
| 61 |
+
},
|
| 62 |
+
"dev432": {
|
| 63 |
+
"n": 432,
|
| 64 |
+
"exact": 372,
|
| 65 |
+
"lexical": 372,
|
| 66 |
+
"truncated": 0,
|
| 67 |
+
"cer": 0.018675455116133083
|
| 68 |
+
},
|
| 69 |
+
"heldout432": {
|
| 70 |
+
"n": 432,
|
| 71 |
+
"exact": 367,
|
| 72 |
+
"lexical": 367,
|
| 73 |
+
"truncated": 0,
|
| 74 |
+
"cer": 0.014281230382925298
|
| 75 |
+
},
|
| 76 |
+
"critical5": {
|
| 77 |
+
"n": 5,
|
| 78 |
+
"exact": 5,
|
| 79 |
+
"lexical": 5,
|
| 80 |
+
"truncated": 0,
|
| 81 |
+
"cer": 0.0
|
| 82 |
+
}
|
| 83 |
+
},
|
| 84 |
+
"candidate": {
|
| 85 |
+
"core": {
|
| 86 |
+
"n": 8,
|
| 87 |
+
"exact": 8,
|
| 88 |
+
"lexical": 8,
|
| 89 |
+
"truncated": 0,
|
| 90 |
+
"cer": 0.0
|
| 91 |
+
},
|
| 92 |
+
"basic48": {
|
| 93 |
+
"n": 48,
|
| 94 |
+
"exact": 43,
|
| 95 |
+
"lexical": 45,
|
| 96 |
+
"truncated": 0,
|
| 97 |
+
"cer": 0.006033182503770739
|
| 98 |
+
},
|
| 99 |
+
"format20": {
|
| 100 |
+
"n": 20,
|
| 101 |
+
"exact": 19,
|
| 102 |
+
"lexical": 20,
|
| 103 |
+
"truncated": 0,
|
| 104 |
+
"cer": 0.0
|
| 105 |
+
},
|
| 106 |
+
"names60": {
|
| 107 |
+
"n": 60,
|
| 108 |
+
"exact": 59,
|
| 109 |
+
"lexical": 60,
|
| 110 |
+
"truncated": 0,
|
| 111 |
+
"cer": 0.0
|
| 112 |
+
},
|
| 113 |
+
"public400": {
|
| 114 |
+
"n": 400,
|
| 115 |
+
"exact": 110,
|
| 116 |
+
"lexical": 226,
|
| 117 |
+
"truncated": 0,
|
| 118 |
+
"cer": 0.1015925315760571
|
| 119 |
+
},
|
| 120 |
+
"intent": {
|
| 121 |
+
"n": 326,
|
| 122 |
+
"exact": 311,
|
| 123 |
+
"lexical": 311,
|
| 124 |
+
"truncated": 0,
|
| 125 |
+
"cer": 0.005274725274725275
|
| 126 |
+
},
|
| 127 |
+
"nonce": {
|
| 128 |
+
"n": 200,
|
| 129 |
+
"exact": 196,
|
| 130 |
+
"lexical": 196,
|
| 131 |
+
"truncated": 0,
|
| 132 |
+
"cer": 0.00492156259612427
|
| 133 |
+
},
|
| 134 |
+
"math236": {
|
| 135 |
+
"n": 236,
|
| 136 |
+
"exact": 233,
|
| 137 |
+
"lexical": 234,
|
| 138 |
+
"truncated": 0,
|
| 139 |
+
"cer": 0.0016420361247947454
|
| 140 |
+
},
|
| 141 |
+
"dev432": {
|
| 142 |
+
"n": 432,
|
| 143 |
+
"exact": 408,
|
| 144 |
+
"lexical": 408,
|
| 145 |
+
"truncated": 0,
|
| 146 |
+
"cer": 0.012397991211550533
|
| 147 |
+
},
|
| 148 |
+
"heldout432": {
|
| 149 |
+
"n": 432,
|
| 150 |
+
"exact": 424,
|
| 151 |
+
"lexical": 424,
|
| 152 |
+
"truncated": 0,
|
| 153 |
+
"cer": 0.003138731952291274
|
| 154 |
+
},
|
| 155 |
+
"critical5": {
|
| 156 |
+
"n": 5,
|
| 157 |
+
"exact": 5,
|
| 158 |
+
"lexical": 5,
|
| 159 |
+
"truncated": 0,
|
| 160 |
+
"cer": 0.0
|
| 161 |
+
}
|
| 162 |
+
},
|
| 163 |
+
"scope": "Actual vLLM0.17 Blackwell SM120 BF16 eager, max-num-seqs8, prefix caching disabled. This is not Orin production acceptance. All diagnostic sets reused after initial evaluation.",
|
| 164 |
+
"critical_single_request": {
|
| 165 |
+
"n": 5,
|
| 166 |
+
"strict_exact": 5,
|
| 167 |
+
"lexical_exact": 5,
|
| 168 |
+
"truncated": 0,
|
| 169 |
+
"seconds": 0.4119902810052736,
|
| 170 |
+
"engine": "vLLM 0.17.0 eager; Blackwell SM120",
|
| 171 |
+
"sequences": 1,
|
| 172 |
+
"scope": "Real serialized checkpoint, not ModelOpt fake quant. Runtime is not the production Orin vLLM version."
|
| 173 |
+
},
|
| 174 |
+
"not_upstream_token_accuracy": true
|
| 175 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": false,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151643,
|
| 6 |
+
151645
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"transformers_version": "5.3.0",
|
| 10 |
+
"use_cache": true
|
| 11 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0d2cf58468359e0a44970083f9ce38078208c291430d20a5c685c5ce6759312
|
| 3 |
+
size 1503300328
|
tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
|
| 3 |
+
size 11422650
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": null,
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"eos_token": "<|im_end|>",
|
| 7 |
+
"errors": "replace",
|
| 8 |
+
"is_local": true,
|
| 9 |
+
"model_max_length": 131072,
|
| 10 |
+
"pad_token": "<|endoftext|>",
|
| 11 |
+
"split_special_tokens": false,
|
| 12 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 13 |
+
"unk_token": null,
|
| 14 |
+
"additional_special_tokens": [
|
| 15 |
+
"<|im_start|>",
|
| 16 |
+
"<|im_end|>",
|
| 17 |
+
"<|object_ref_start|>",
|
| 18 |
+
"<|object_ref_end|>",
|
| 19 |
+
"<|box_start|>",
|
| 20 |
+
"<|box_end|>",
|
| 21 |
+
"<|quad_start|>",
|
| 22 |
+
"<|quad_end|>",
|
| 23 |
+
"<|vision_start|>",
|
| 24 |
+
"<|vision_end|>",
|
| 25 |
+
"<|vision_pad|>",
|
| 26 |
+
"<|image_pad|>",
|
| 27 |
+
"<|video_pad|>"
|
| 28 |
+
]
|
| 29 |
+
}
|
training-provenance.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"kd_weight": 0.1,
|
| 3 |
+
"lr": 1e-06,
|
| 4 |
+
"seed": 20261108,
|
| 5 |
+
"rows": 7157,
|
| 6 |
+
"eos_only_samples": 24,
|
| 7 |
+
"max_length": 512,
|
| 8 |
+
"target_encoding": "marked",
|
| 9 |
+
"total_steps": 448,
|
| 10 |
+
"full_parameters": 751632384,
|
| 11 |
+
"precision": "FP32 masters; BF16 student autocast; FP32 frozen teacher; TF32 disabled",
|
| 12 |
+
"train_sha256": "a10d979b2f8a7a8c043d183d200b7467c616f427f4d15de8852f8ba86011117e",
|
| 13 |
+
"batch_variation": "Deterministic extra right-padding 0/8/16 per microbatch, identical in both arms; does not reproduce all vLLM kernels",
|
| 14 |
+
"kd_gate": "Teacher argmax equals gold and probability>=0.9; EOS included; prompt/padding excluded",
|
| 15 |
+
"margin": "0.2 * max(0,1-gold_logit+max_wrong_logit), numeric/ASCII targets only",
|
| 16 |
+
"completed_steps": 448,
|
| 17 |
+
"training_seconds": 183.75626777499565,
|
| 18 |
+
"release": "S1-mini-Chinese-1106-BF16",
|
| 19 |
+
"internal_checkpoint": "stability-margin-r2 final",
|
| 20 |
+
"upstream": "S1-mini by Superwhisper",
|
| 21 |
+
"parent_release": "WIlfLin/S1-mini-Chinese-1106 native-v10, followed by CE stability stage",
|
| 22 |
+
"micro_batch": 4,
|
| 23 |
+
"gradient_accumulation": 4,
|
| 24 |
+
"training_script_sha256": "522b9f0e07917571e54f9416c27ee4c6a5b0543c58b949a0724623b668b4bd8c",
|
| 25 |
+
"matched_experiment": "CE vs CE+teacher-correct KL: no KD superiority; CE selected, then numerical-composition/margin refinement applied"
|
| 26 |
+
}
|