Add pipeline tag and library name to metadata

#1
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +19 -8
README.md CHANGED
@@ -1,14 +1,17 @@
1
  ---
2
- license: cc-by-nc-nd-4.0
3
- language:
4
- - en
5
  base_model:
6
  - Qwen/Qwen2.5-VL-7B-Instruct
 
 
 
 
 
7
  tags:
8
  - Pathology
9
  - VLM
10
  - Reasoning
11
  ---
 
12
  <h1 align="center">[MICCAI 2026] Enhancing Pathological VLMs with Cross-scale Reasoning</h1>
13
 
14
  <p align="center"> Chi Phan*, Tianyi Zhang*, Qiaochu Xue, Yufeng Wu, Dan Hu, Zeyu Liu, Sudong Wang, Yueming Jin </p>
@@ -213,7 +216,8 @@ Download **ScaleReasoner-R1** from [HuggingFace](https://huggingface.co/ChiPhan1
213
  SYSTEM_PROMPT = (
214
  "You are a pathology expert. Read the question and options about the image carefully. "
215
  "Think step by step inside <think> </think>. Then output ONLY the SINGLE best option letter "
216
- "inside <answer> </answer>.\n"
 
217
  "Example: <think>Your reasoning</think> <answer>A</answer>. "
218
  "Do not include the option text or any extra words inside <answer> </answer> tags."
219
  )
@@ -251,7 +255,11 @@ response = client.chat.completions.create(
251
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('low_mag.jpg')}"}},
252
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('mid_mag.jpg')}"}},
253
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('high_mag.jpg')}"}},
254
- {"type": "text", "text": "<question>\n(A) ...\n(B) ...\n(C) ...\n(D) ..."},
 
 
 
 
255
  ]
256
  }],
257
  max_tokens=4096,
@@ -276,7 +284,11 @@ messages = [{
276
  {"type": "image", "image": "low_mag.jpg"},
277
  {"type": "image", "image": "mid_mag.jpg"},
278
  {"type": "image", "image": "high_mag.jpg"},
279
- {"type": "text", "text": "<question>\n(A) ...\n(B) ...\n(C) ...\n(D) ..."},
 
 
 
 
280
  ]
281
  }]
282
 
@@ -308,9 +320,8 @@ If you find our work helpful, please consider citing our paper and the framework
308
  ```bibtex
309
  @article{phan2026enhancing,
310
  title={Enhancing Pathological VLMs with Cross-scale Reasoning},
311
- author={Phan, Chi and Zhang, Tianyi and Xue, Qiaochu and Wu, Yufeng and Hu, Dan and Liu, Zeyu and Wang, Sudong and Jin, Yueming},
312
  journal={arXiv preprint arXiv:2606.17412},
313
  year={2026}
314
  }
315
-
316
  ```
 
1
  ---
 
 
 
2
  base_model:
3
  - Qwen/Qwen2.5-VL-7B-Instruct
4
+ language:
5
+ - en
6
+ license: cc-by-nc-nd-4.0
7
+ library_name: transformers
8
+ pipeline_tag: image-text-to-text
9
  tags:
10
  - Pathology
11
  - VLM
12
  - Reasoning
13
  ---
14
+
15
  <h1 align="center">[MICCAI 2026] Enhancing Pathological VLMs with Cross-scale Reasoning</h1>
16
 
17
  <p align="center"> Chi Phan*, Tianyi Zhang*, Qiaochu Xue, Yufeng Wu, Dan Hu, Zeyu Liu, Sudong Wang, Yueming Jin </p>
 
216
  SYSTEM_PROMPT = (
217
  "You are a pathology expert. Read the question and options about the image carefully. "
218
  "Think step by step inside <think> </think>. Then output ONLY the SINGLE best option letter "
219
+ "inside <answer> </answer>.
220
+ "
221
  "Example: <think>Your reasoning</think> <answer>A</answer>. "
222
  "Do not include the option text or any extra words inside <answer> </answer> tags."
223
  )
 
255
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('low_mag.jpg')}"}},
256
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('mid_mag.jpg')}"}},
257
  {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encode_image('high_mag.jpg')}"}},
258
+ {"type": "text", "text": "<question>
259
+ (A) ...
260
+ (B) ...
261
+ (C) ...
262
+ (D) ..."},
263
  ]
264
  }],
265
  max_tokens=4096,
 
284
  {"type": "image", "image": "low_mag.jpg"},
285
  {"type": "image", "image": "mid_mag.jpg"},
286
  {"type": "image", "image": "high_mag.jpg"},
287
+ {"type": "text", "text": "<question>
288
+ (A) ...
289
+ (B) ...
290
+ (C) ...
291
+ (D) ..."},
292
  ]
293
  }]
294
 
 
320
  ```bibtex
321
  @article{phan2026enhancing,
322
  title={Enhancing Pathological VLMs with Cross-scale Reasoning},
323
+ author={Phan, Chi and Zhang, Tianyi Beetroot and Xue, Qiaochu and Wu, Yufeng and Hu, Dan and Liu, Zeyu and Wang, Sudong and Jin, Yueming},
324
  journal={arXiv preprint arXiv:2606.17412},
325
  year={2026}
326
  }
 
327
  ```