Update README.md

README.md

@@ -7,7 +7,8 @@ task_categories:
 pretty_name: PixelReasoner-RL-Data
 ---
 **Overview.**
-The RL data for training [**Pixel Reasoner**: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning](https://arxiv.org/abs/2505.15966). 
+The RL data for training [**Pixel Reasoner**: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning](https://arxiv.org/abs/2505.15966),
+The queries require fine-grained visual analysis in both images (e.g., infographics, visually-rich scenes, etc) and videos. 
 
 **Details.**
 The data includes 15,402 training queries with verifierable answers. The key fields include: