Update dataset

EndoBench.tsv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75ad721444b13fbed6b62b45a25fcbfe586a8b6c4b2064366f963f08f6eb5868
+oid sha256:8d371db9b1421d0edf4e9caa0f1e310dffd86bf917fad75eba0328c00182d983
 size 588480658

EndoVQA-Instruct-trainval.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:85988a52e757e57e906165a217aed11df764fa459295a14317f2def16d8fad70
+oid sha256:91206c60377164f0a3c28af46a61d88ad7430ddd3d3f64d98023b07df8c344fe
 size 297633793

README.md

@@ -1,11 +1,16 @@
 ---
 license: cc-by-sa-3.0
 tags:
-  - medical
+- medical
 language:
-  - en
+- en
 task_categories:
-  - question-answering
+- question-answering
+configs:
+- config_name: EndoBench
+  data_files:
+  - split: test
+    path: EndoBench.tsv
 ---
 # <div align="center"><b> EndoBench </b></div>
 
@@ -17,15 +22,15 @@ This repository is the official implementation of the paper **EndoBench: A Compr
 
 EndoBench is a comprehensive MLLM evaluation framework spanning 4 endoscopy scenarios and 12 clinical tasks with 12 secondary subtasks that mirror the progression of endoscopic examination workflows. Featuring five levels of visual prompting granularity to assess region-specific understanding, our EndoBench contains 6,832 clinically validated VQA pairs derived from 22 endoscopy datasets. This structure enables precise measurement of MLLMs' clinical perceptual, diagnostic accuracy, and spatial comprehension across diverse endoscopic scenarios.
 
-Our dataset construction involves collecting 21 public and 1 private endoscopy datasets and standardizing QA pairs, yielding 446,535 VQA pairs comprising our~\ourinstruct~dataset, the current largest endoscopic instruction-tuning collection. From~\ourinstruct, we extract representative pairs that undergo rigorous clinical review, resulting in our final~\ourmethod~dataset of 6,832 clinically validated VQA pairs.
+Our dataset construction involves collecting 21 public and 1 private endoscopy datasets and standardizing QA pairs, yielding 446,535 VQA pairs comprising our EndoVQA-Instruct dataset, the current largest endoscopic instruction-tuning collection. From EndoVQA-Instruct, we extract representative pairs that undergo rigorous clinical review, resulting in our final EndoBench of 6,832 clinically validated VQA pairs.
 
 We provide two datasets:
 
-1. EndoVQA-Instruct-trainval, which included *439703* VQA pairs.
+1. EndoVQA-Instruct-trainval, which included **439703** VQA pairs.
 
 2. EndoBench, which encompasses 4 distinct endoscopic modalities, 12 specialized clinical tasks with 12 secondary subtasks, and 5 levels of visual prompting granularity, resulting in 6,832 rigorously validated VQA pairs from 22 diverse datasets. Our multi-dimensional evaluation framework mirrors the clinical workflow—spanning anatomical recognition, lesion analysis, spatial localization, and surgical operations—to holistically gauge the perceptual and diagnostic abilities of MLLMs in realistic scenarios.
 
-We provide 2 version of EndoBench: json file and tsv file.
+We provide 2 versions of EndoBench: .json file and .tsv file.
 
 
 ## Evaluation
@@ -38,7 +43,7 @@ cd VLMEvalKit
 pip install -e .
 ```
 
-2. Add our dataset
+2. Add our dataset to VLMEvalKit.
 
 3. You can find more details on the [ImageMCQDataset Class](https://github.com/open-compass/VLMEvalKit/blob/main/vlmeval/dataset/image_mcq.py).