---
task_categories:
- visual-question-answering
language:
- en
---

# VLMsAreBlind Benchmark
refactor [VisualProbe Benchmark](https://huggingface.co/datasets/Mini-o3/VisualProbe_Hard) to add support for VLMEvalKit.

# Benchmark Information
1. Number of questions: 106
2. Question type: Free-form
3. Question format: image + text
4. Answer type: Free-form

# Reference

- [VLMEvalKit](https://github.com/open-compass/VLMEvalKit)
- [VisualProbe Benchmark](https://huggingface.co/datasets/Mini-o3/VisualProbe_Hard)