Reproduce “SAGE: Semantic Anchor-Guided Evolution for Grounded Medical QA Data Synthesis” 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
Chuan Li, Chengyu Wang, Cen Chen, Ye Lyu, Mingyuan Fan, Ming Gao
SAGE uses MeSH to help a local model generate clinical question-and-answer training data from a small seed set. It alternates single-concept questions with scenarios linking multiple concepts, then maps questions back to MeSH; self-consistency filters answers but does not verify truth. On five QA benchmarks, models averaged 68.12% versus 63.71% for CoT-Self-Instruct under the tested setup, not clinical validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.