MMDU Benchmark

What is MMDU?

MMDU is intended for evaluating the multi-image multi-turn dialogue capabilities. It comprises 110 high-quality multi-image multi-turn dialogues with more than…

Released

2024-06-17

Modality

multimodal

Evaluates

多模态, Multimodal, 推理, Reasoning, NeurIPS 2024, 多模态模型, VLM, 逻辑推理, 不支持, Unsupported

Openness

restricted

Importer

OpenCompass Hub

Reported scores

0

Source provenance

• Original evidence https://hub.opencompass.org.cn/dataset-detail/MMDU

MMDU paper, code and dataset

• Paper https://arxiv.org/abs/2406.11833
• Code https://github.com/Liuziyu77/MMDU
• Dataset https://huggingface.co/datasets/laolao77/MMDU

Which sources cite MMDU?

• MMDU

Related 多模态, Multimodal, 推理, Reasoning, NeurIPS 2024, 多模态模型, VLM, 逻辑推理, 不支持, Unsupported benchmarks

• 3DCodeBench
• A-Bench
• A-OKVQA
• AbsPyramid
• ActionAtlas
• ActiveFly-Bench
• Ada-LEval
• AdvancedMathBench