Hugging Face's logo

Datasets:
MTSAIR
/
MWS-Vision-Bench

Tasks:
Visual Question Answering
Document Question Answering
Modalities:
Image
Text
Document
Formats:
parquet
Languages:
Russian
Size:
1K - 10K
Tags:
benchmark
multimodal
ocr
kie
grounding
vlm
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
3
GeorgyG commited on
Commit
fb19084
·
verified ·
1 Parent(s): 40ad7ab

Update validation leaderboard

Browse files
Files changed (1) hide show
  1. README.md +16 -5
README.md CHANGED
@@ -223,11 +223,18 @@ Top models evaluated on this validation dataset:
223
 
224
  | Model | Overall | img→text | img→markdown | Grounding | KIE (JSON) | VQA |
225
  |-------|---------|----------|--------------|-----------|------------|-----|
226
- | **Cotype Pro 3** | 0.736 | 0.861 | 0.747 | 0.384 | 0.832 | 0.858 |
227
- | **Qwen 3.7 Plus** | 0.708 | 0.843 | 0.766 | 0.154 | 0.844 | 0.935 |
228
- | **Claude-4.6-Opus** | 0.704 | 0.841 | 0.748 | 0.168 | 0.852 | 0.908 |
229
- | **Qwen3.6-35B-A3B** | 0.694 | 0.859 | 0.784 | 0.099 | 0.822 | 0.904 |
230
- | **Qwen3.6-27B** | 0.691 | 0.863 | 0.768 | 0.071 | 0.833 | 0.919 |
 
 
 
 
 
 
 
231
  | Gemini-2.5-pro | 0.690 | 0.840 | 0.717 | 0.070 | 0.888 | 0.935 |
232
  | Kimi K2 Instruct | 0.686 | 0.874 | 0.726 | 0.148 | 0.805 | 0.876 |
233
  | Cotype Light 3 | 0.685 | 0.855 | 0.770 | 0.125 | 0.873 | 0.799 |
@@ -262,6 +269,10 @@ Top models evaluated on this validation dataset:
262
  | Qwen3.5-2B | 0.489 | 0.743 | 0.621 | 0.041 | 0.466 | 0.574 |
263
  | Qwen3.5-0.8B | 0.319 | 0.549 | 0.499 | 0.027 | 0.157 | 0.362 |
264
 
 
 
 
 
265
  *Scale: 0.0 - 1.0 (higher is better)*
266
 
267
  **📝 Submit your model**: To evaluate on the private test set, contact [g.gaikov@mts.ai](mailto:g.gaikov@mts.ai)
 
223
 
224
  | Model | Overall | img→text | img→markdown | Grounding | KIE (JSON) | VQA |
225
  |-------|---------|----------|--------------|-----------|------------|-----|
226
+ | Claude Fable 5† | 0.799 | 0.828 | 0.737 | 0.635 | 0.843 | 0.950 |
227
+ | GPT-5.6 Sol | 0.790 | 0.818 | 0.716 | 0.621 | 0.867 | 0.926 |
228
+ | Kimi K3‡ | 0.777 | 0.783 | 0.716 | 0.588 | 0.856 | 0.941 |
229
+ | GPT-5.6 Terra | 0.749 | 0.808 | 0.693 | 0.571 | 0.799 | 0.874 |
230
+ | Kimi K2.6 | 0.743 | 0.868 | 0.761 | 0.301 | 0.858 | 0.929 |
231
+ | Cotype Pro 3 | 0.736 | 0.861 | 0.747 | 0.384 | 0.832 | 0.858 |
232
+ | Claude Sonnet 5 | 0.731 | 0.842 | 0.729 | 0.370 | 0.788 | 0.924 |
233
+ | GPT-5.6 Luna | 0.727 | 0.776 | 0.673 | 0.547 | 0.772 | 0.864 |
234
+ | Qwen 3.7 Plus | 0.708 | 0.843 | 0.766 | 0.154 | 0.844 | 0.935 |
235
+ | Claude-4.6-Opus | 0.704 | 0.841 | 0.748 | 0.168 | 0.852 | 0.908 |
236
+ | Qwen3.6-35B-A3B | 0.694 | 0.859 | 0.784 | 0.099 | 0.822 | 0.904 |
237
+ | Qwen3.6-27B | 0.691 | 0.863 | 0.768 | 0.071 | 0.833 | 0.919 |
238
  | Gemini-2.5-pro | 0.690 | 0.840 | 0.717 | 0.070 | 0.888 | 0.935 |
239
  | Kimi K2 Instruct | 0.686 | 0.874 | 0.726 | 0.148 | 0.805 | 0.876 |
240
  | Cotype Light 3 | 0.685 | 0.855 | 0.770 | 0.125 | 0.873 | 0.799 |
 
269
  | Qwen3.5-2B | 0.489 | 0.743 | 0.621 | 0.041 | 0.466 | 0.574 |
270
  | Qwen3.5-0.8B | 0.319 | 0.549 | 0.499 | 0.027 | 0.157 | 0.362 |
271
 
272
+ † 10 of 1302 responses failed after the configured retries and received a score of 0.
273
+
274
+ ‡ 16 of 1302 responses failed after the configured retries and received a score of 0.
275
+
276
  *Scale: 0.0 - 1.0 (higher is better)*
277
 
278
  **📝 Submit your model**: To evaluate on the private test set, contact [g.gaikov@mts.ai](mailto:g.gaikov@mts.ai)