Reproduce “Conversation Is a Two-Body Problem: Dyadic Evaluation of Full-Duplex Dialogue Models” 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
Sungnyun Kim, Sungwoo Cho, Jihwan Oh, Se-Young Yun
DyaFDB evaluates full-duplex spoken-dialogue models through live conversations between paired models with separate roles, scoring both participants. It measures task outcomes and turn-taking, capturing how partner behavior relates to results. Across tested attacker pairings, the same defender’s secret-leak rate ranged from 39% to 63%. These findings describe tested model matchups, not universal real-world agent behavior.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.