Large Language Models (LLMs) increasingly ship with explicit "thinking modes", yet their counterpart, "no-thinking", has received far less attention. We stud…
机构:清华
来源:arXiv 2610.11765 | AI4Papers 论文推荐平台