[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"topic-zh-llm--ec743e":3},{"cluster":4,"timeline":16},{"id":5,"slug":6,"title":7,"pinned":8,"status":9,"summary":10,"category":11,"language":12,"created_at":13,"merged_into":14,"article_count":15,"first_seen_at":13,"last_updated_at":13},"2a8981b1-a3dc-4ec3-b550-d2700d95a056","llm--ec743e","LLM 評審別只看平均分",false,"active","這篇論文提醒：LLM 當評審時，平均表現看起來穩，不代表每個輸入都可靠。作者用 transitivity 檢查與 conformal prediction sets，抓出輸入層級的不一致與不確定性。","research","zh","2026-07-05T09:04:09.7973+00:00",null,1,[17],{"id":18,"slug":19,"title":7,"summary":10,"category":11,"image_url":20,"cover_image":20,"published_at":21,"is_canonical_seed":22},"082ebaa3-ad6f-421a-860a-8566846fb9c1","llm-judge-reliability-conformal-transitivity-zh","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1776406189176-acr2.png","2026-04-17T06:09:32.85+00:00",true]