Dataset · LeoFan01
RoboBench Benchmark for MLLM Robotics Evaluation
LeoFan01/RoboBench
English multimodal test suite for checking how well vision-language models reason in robotics scenarios.
What's inside
The dataset contains up to 1,000 English-language examples including images and text. It serves as an evaluation benchmark for testing multimodal large language models (MLLMs) in the role of an embodied brain. The data is structured as a systematic framework to analyze understanding and logical reasoning in robotics scenarios.
What you can do with it
- Testing multimodal LLMs (VQA) in robot control scenarios
- Comparing the ability of different MLLMs to analyze visual scenes and plan actions
- Validating models for embodied AI tasks
- Tasks
- Robotics, Visual question answering
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 20.4K
- Likes
- 7
How to load it
from datasets import load_dataset
ds = load_dataset("LeoFan01/RoboBench")