[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"topic-zh-tsallis-loss-":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},"bd0444a9-5339-481d-b6a8-ad28a312f816","tsallis-loss-","Tsallis loss 讓推理模型更快脫困",false,"active","這篇論文用 Tsallis q-logarithm 搭出一條損失函數光譜，想解決推理模型在冷啟動時卡住的問題。它把 RLVR 和 latent trajectory 的 log-marginal-likelihood 串成可調參的連續體。","research","zh","2026-07-05T09:05:09.151215+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},"8e6e5e5b-c51f-495e-a596-203fb64c71eb","tsallis-loss-reasoning-model-training-zh","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1777443006073-083j.png","2026-04-29T06:09:37.068+00:00",true]