Large Reasoning Models (LRMs) produce explicit chains of thought to improve transparency, but they frequently overthink, generating long yet inefficient reas…
机构:哈工大
来源:arXiv 2611.11901 | AI4Papers 论文推荐平台