Large language models (LLMs) exhibit mode collapse, where their output distribution concentrates on familiar semantic modes. When diversity disappears, repea…
机构:清华
来源:arXiv 2610.32101 | AI4Papers 论文推荐平台