BayesBench: Evaluating LLM Belief Trajectories Under Multi-Turn Evidence Accumulation 文章
ArXiv CS.AI2026-07-01PAPERen作者: Ankur Samanta, Akshayaa Magesh, Tal Lancewicki, Ayush Jain, Youliang Yu, Paul Sajda, Kaveh Hassani, Aditya Modi, Daniel R. Jiang, Yonathan Efroni