Large language models (LLMs) often generate more reasoning tokens than needed. Limiting the reasoning budget reduces cost, but forcing an answer from an inco…
机构:UIUC
来源:arXiv 2610.25768 | AI4Papers 论文推荐平台