Reproduce “ArtifactLinker: Linking Scientific Artifacts for Automatic State-of-the-Art Discovery” 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
Haofei Yu, Jiaxuan You, Peter Clark, Bodhisattwa Prasad Majumder, Kyle Richardson
ArtifactLinker ranks missing model–dataset evaluations in an artifact graph, then runs selected candidates. It combines estimates of compatibility and conditional performance to prioritize tests. In a benchmark, joint ranking reached about half the oracle’s best performance after about 10 verifications, versus more than 60 for a biased baseline; its 72.6% success rate applies to 263 selected pairs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.