RoboBench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied Brain
RoboBench is a comprehensive evaluation benchmark designed to assess the capabilities of Multimodal Large Language Models (MLLMs) in embodied intelligence tasks. This benchmark provides a systematic framework for evaluating how well these models can understand and reason about robotic scenarios.
This repository contains the released RoboBench… See the full description on the dataset page:
https://huggingface.co/datasets/LeoFan01/RoboBench.