Reproduce “Estimation of Propagated Risk in LLM-based Multi-Agent 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
Shreyan Ganguly, Souradeep Chattopadhyay, Qisai Liu, Soumik Sarkar
IE-Prop estimates error risk in multi-agent systems from completed sender–receiver exchanges. It combines receiver-local evidence with interaction effects conditioned on whether the sender is correct, and recursively tracks risk through communication graphs. It outperforms a receiver-text baseline in 46 of 54 in-domain combinations. In sequential systems, recursion raises mean AUROC from 0.631 to 0.655 across nine tested dataset–model-size pairs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.