Reliable large language models (LLMs) must not only produce accurate answers but also express confidence that faithfully reflects their probability of being …
机构:华为
来源:arXiv 2610.02695 | AI4Papers 论文推荐平台