Reproduce “Epistemic-Probabilistic Model for Guarded Multi-Agent LLM Coordination” 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
Mehdi Nasiri, Mohammad Saeed Arvenaghi, Sadegh Vaezi, Ebrahim Ardeshir-Larijani
EPLA separates action proposal from authorization: a language model suggests typed actions, while a symbolic Guard checks them against authoritative state before commitment. In its epistemic-lottery gossip model, the authors prove that strictly positive probability weights do not change truth values of knowledge-only Guard formulas when histories and views remain fixed. This is a formal result, not an empirical evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.