Test-time scaling methods implicitly assume that additional computation from the same model provides independent evidence. We show that this assumption fails…
机构:上交
来源:arXiv 2610.23213 | AI4Papers 论文推荐平台