Hi,
Thank you for the excellent work in collecting multimodal datasets! I would like to draw your attention to a new multimodal dataset: MMSci, a multimodal multi-discipline benchmark based on Nature Communications journals for PhD-level scientific comprehension. This dataset encompasses over 72 advanced scientific disciplines, 131k articles, and 742k figures, all directly crawled from the web rather than extracted from PDFs, ensuring both diversity and quality.
You can find more details at:
ArXiv: https://arxiv.org/abs/2407.04903
Code & Data: https://github.com/Leezekun/MMSci
Hi,
Thank you for the excellent work in collecting multimodal datasets! I would like to draw your attention to a new multimodal dataset: MMSci, a multimodal multi-discipline benchmark based on Nature Communications journals for PhD-level scientific comprehension. This dataset encompasses over 72 advanced scientific disciplines, 131k articles, and 742k figures, all directly crawled from the web rather than extracted from PDFs, ensuring both diversity and quality.
You can find more details at:
ArXiv: https://arxiv.org/abs/2407.04903
Code & Data: https://github.com/Leezekun/MMSci