[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-test-time-training":3},{"tag":4,"articles":11},{"id":5,"name":6,"slug":7,"article_count":8,"description_zh":9,"description_en":10},"644d6502-07b1-4c3a-abd5-421a170f2c5b","test-time training","test-time-training",4,"Test-time training 指模型在推論階段也能依輸入資料做小幅更新，常見於長上下文 LLM、4D 重建與持續感知任務，用來減少遺忘、提升即時適應與跨 chunk 推理穩定性。","Test-time training refers to models that keep adapting during inference with small updates from the current input. It shows up in long-context LLMs, 4D reconstruction, and continual perception, where it helps reduce forgetting and improve online adaptation.",[12,21],{"id":13,"slug":14,"title":15,"summary":16,"category":17,"image_url":18,"cover_image":18,"language":19,"created_at":20},"7e3fc38d-5744-4f1d-8941-643ed78be513","fast-spatial-memory-elastic-test-time-training-zh","長序列4D重建的彈性記憶法","FSM 用彈性 test-time training 穩住長序列 4D 重建的記憶更新，降低遺忘與記憶瓶頸，讓多 chunk 推論更可行。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775714633904-j3go.png","zh","2026-04-09T06:03:34.127299+00:00",{"id":22,"slug":23,"title":24,"summary":25,"category":17,"image_url":26,"cover_image":26,"language":19,"created_at":27},"75d63765-ec7c-4833-8c77-5caabb7b5c46","in-place-ttt-llms-adapt-at-inference-zh","In-Place TTT 讓 LLM 推理時自適應","這篇論文把 test-time training 做成可直接嵌入 LLM 的推理更新機制，讓模型在長上下文下用 fast weights 即時適應，不必整個重訓。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775628411507-jici.png","2026-04-08T06:06:33.015125+00:00"]