[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-analogy":3},{"tag":4,"articles":9},{"id":5,"name":6,"slug":6,"article_count":7,"description_zh":8,"description_en":8},"84dc11fe-09b3-405f-93ca-5c2b34fb0b8c","analogy",0,null,[10],{"id":11,"slug":12,"title":13,"summary":14,"category":15,"image_url":16,"cover_image":16,"language":17,"created_at":18},"6911e614-4894-4f1f-a0ad-816e323793ef","retrieval-augmented-reinforcement-fine-tuning-analogy-zh","RA-RFT 讓模型學會類比推理","RA-RFT 把檢索目標從相似度改成「有助推理的類比」，再用這些示範做強化式微調，讓模型在數學推理上更會借題發揮。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781245983485-miul.png","zh","2026-06-12T06:32:27.726554+00:00"]