Reproduce “Before Agents Decide: Epistemic Action in LLM-Based Systems” 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

Yizhi Liu, Balaji Padmanabhan, Siva Viswanathan

The authors propose an evidence-preparation framework for agents. It distinguishes pragmatic actions that change the task from epistemic actions that improve what an agent knows or how usable its evidence is. These actions acquire missing evidence, transform available evidence, or probe systems for responses. The paper proposes a benchmark, but does not evaluate autonomous selection across all three modes.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.