[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-embedding-layer-learning-rate":3},{"tag":4,"articles":10},{"id":5,"name":6,"slug":7,"article_count":8,"description_zh":9,"description_en":9},"884a6463-e85e-4182-8103-3b859eb17890","embedding layer learning rate","embedding-layer-learning-rate",0,null,[11],{"id":12,"slug":13,"title":14,"summary":15,"category":16,"image_url":17,"cover_image":17,"language":18,"created_at":19},"e786bb53-c4b9-48d5-a337-96e872e96e56","embedding-layer-lr-hyperparameter-transfer-zh","為何 embedding LR 決定 μP 轉移","這篇論文指出，μP 比標準參數化更會轉移，關鍵不是整套方法本身，而是 embedding layer 的學習率設定。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779344766617-4pa3.png","zh","2026-05-21T06:25:34.782362+00:00"]