Skip to content

Latest commit

 

History

History
16 lines (10 loc) · 703 Bytes

README.md

File metadata and controls

16 lines (10 loc) · 703 Bytes

ActiView: Evaluating Active Perception Ability for Multimodal Large Language Models

🌐 Homepage| 📖 arXiv

Coming soon ⏳🔜

  • 🚧 We will release our benchmark and evaluation toolkit ASAP.

  • 📨 If you have any questions regarding the data and evaluation scripts when they are temporarily unavailable, please contact [email protected].

Project Overview

This repository contains all the necessary materials, including:

  • Datasets for active perception evaluation
  • Toolkits for model evaluation
  • Results and analysis from the paper