Reproduce “From Relevance to Validity: Zero-Shot Tool Selection via Functional and Contextual Validation” 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
Xiaoxuan Dong, Jin Zhang, Zichen Yuan, Yue Zhou, Zihao Rao, Dexu Yu, Tingyong WU, Wenhao Deng and 6 more
TACT ranks callable capabilities for tasks from unseen domains by combining semantic similarity with checks of documented endpoint operations and inputs, then adjusting scores using source-side evidence about task context. In cross-domain ToolBench Context ranking, it reaches NDCG@5 of 0.624 versus PORTS at 0.518; this measures annotated ranking quality, not live API success.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.