1
Building Reliable Long-Form Generation via Hallucination Rejection Sampling
被ICML 2026接收，提出通过幻觉拒绝采样显著提升长文本生成的可靠性，是NLP领域的重要突破。
arXiv:2606.03628v1 Announce Type: cross Abstract: Large language models (LLMs) have achieved remarkable progress in open-ended text generation, yet th…