MMWorld: Towards Multi-discipline Multi-faceted World Model Evaluation in Videos

TL;DR

MMWorld introduces a comprehensive video benchmark for evaluating multimodal language models across multiple disciplines and reasoning tasks, highlighting current limitations and guiding future improvements.

Contribution

The paper presents MMWorld, a novel multi-discipline, multi-faceted video understanding benchmark with extensive datasets and evaluation protocols for assessing world model capabilities in videos.

Findings

MLLMs perform poorly on MMWorld, with GPT-4V achieving only 52.3% accuracy.

Models exhibit diverse skill sets compared to humans.

Significant room for improvement in multimodal video understanding.

Abstract

Multimodal Language Language Models (MLLMs) demonstrate the emerging abilities of "world models" -- interpreting and reasoning about complex real-world dynamics. To assess these abilities, we posit videos are the ideal medium, as they encapsulate rich representations of real-world dynamics and causalities. To this end, we introduce MMWorld, a new benchmark for multi-discipline, multi-faceted multimodal video understanding. MMWorld distinguishes itself from previous video understanding benchmarks with two unique advantages: (1) multi-discipline, covering various disciplines that often require domain expertise for comprehensive understanding; (2) multi-faceted reasoning, including explanation, counterfactual thinking, future prediction, etc. MMWorld consists of a human-annotated dataset to evaluate MLLMs with questions about the whole videos and a synthetic dataset to analyze MLLMs within a single modality of perception. Together, MMWorld encompasses 1,910 videos across seven broad disciplines and 69 subdisciplines, complete with 6,627 question-answer pairs and associated captions. The evaluation includes 2 proprietary and 10 open-source MLLMs, which struggle on MMWorld (e.g., GPT-4V performs the best with only 52.3\% accuracy), showing large room for improvement. Further ablation studies reveal other interesting findings such as models' different skill sets from humans. We hope MMWorld can serve as an essential step towards world model evaluation in videos.