Reproduce “ChemVA: Advancing Large Language Models on Chemical Reaction Diagrams Understanding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mingyang Rao, Kehua Feng, Zhihui Zhu, Jiangzhen Fu, Hao Yu, Keyan Ding, Huajun Chen
ChemVA converts reaction diagrams into molecular structures and language-model inputs for chemical reasoning. It detects functional groups and attachment points, reconstructs connected structures, and supplies chemical names alongside SMILES. On OCRD-Bench, Llama-4-109B’s reported Total rises from 38.88 with direct image input to 61.20 with ChemVA, a 22.32-point gain; this is a benchmark composite, not a single-task measure.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.