[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"topic-zh-prerl":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},"b1446780-3fcf-4a4b-8a90-590c54916dd5","prerl","PreRL：把強化學習搬進預訓練空間",false,"active","PreRL 把 RL 從 P(y|x) 轉向 P(y)，直接在預訓練空間做獎勵更新，主打增強推理與探索。摘要也提到 NSR 與 DSRL 兩種設計。","research","zh","2026-07-05T09:04:02.408075+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},"2428c4f3-8cbf-43dc-afe8-dad89550740f","prerl-training-llms-in-pre-train-space-zh","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1776319619099-op5n.png","2026-04-16T06:06:37.854+00:00",true]