Mmsci: A Multimodal Multi-discipline Dataset For Phd-level Scientific Comprehension

Li Zekun, Yang Xianjun, Choi Kyuri, Zhu Wanrong, Hsieh Ryan, Kim Hyeonjung, Lim Jin Hyuk, Ji Sungyoung, Lee Byungju, Yan Xifeng, Petzold Linda Ruth, Wilson Stephen D., Lim Woosang, Wang William Yang. Arxiv 2024

[Paper]    
GPT Model Architecture Multimodal Models Reinforcement Learning Tools Training Techniques

The rapid advancement of Large Language Models (LLMs) and Large Multimodal Models (LMMs) has heightened the demand for AI-based scientific assistants capable of understanding scientific articles and figures. Despite progress, there remains a significant gap in evaluating models’ comprehension of professional, graduate-level, and even PhD-level scientific content. Current datasets and benchmarks primarily focus on relatively simple scientific tasks and figures, lacking comprehensive assessments across diverse advanced scientific disciplines. To bridge this gap, we collected a multimodal, multidisciplinary dataset from open-access scientific articles published in Nature Communications journals. This dataset spans 72 scientific disciplines, ensuring both diversity and quality. We created benchmarks with various tasks and settings to comprehensively evaluate LMMs’ capabilities in understanding scientific figures and content. Our evaluation revealed that these tasks are highly challenging: many open-source models struggled significantly, and even GPT-4V and GPT-4o faced difficulties. We also explored using our dataset as training resources by constructing visual instruction-following data, enabling the 7B LLaVA model to achieve performance comparable to GPT-4V/o on our benchmark. Additionally, we investigated the use of our interleaved article texts and figure images for pre-training LMMs, resulting in improvements on the material generation task. The source dataset, including articles, figures, constructed benchmarks, and visual instruction-following data, is open-sourced.

Similar Work