[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-transformer":3},{"tag":4,"articles":10,"peer_article_count":71},{"id":5,"name":6,"slug":6,"article_count":7,"description_zh":8,"description_en":9},"1b756de5-5b17-4f34-8968-d2df19807253","transformer",3,"Transformer 是處理序列資料的核心架構，從文字生成、圖像擴散到日誌異常偵測都常見其身影。它關注注意力機制、長距依賴與推論效率，也延伸到壓縮位元組、生成多樣性與交易系統等實作場景。","Transformer is the sequence model behind many modern AI systems, from language generation and text-to-image diffusion to log anomaly detection. Coverage here focuses on attention, long-range dependencies, efficiency tradeoffs, and applied variants such as compressed-byte inference and diversity control.",[11,20,28,35,42,49,56,64],{"id":12,"slug":13,"title":14,"summary":15,"category":16,"image_url":17,"cover_image":17,"language":18,"created_at":19},"c40b20df-d89a-43ae-bb11-11062dcd2cd2","llms-work-by-predicting-next-token-zh","5 個關鍵部件看懂 LLMs","5 個關鍵部件帶你看懂 LLMs 如何從資料、token、注意力到對齊，進而判斷訓練與部署該看什麼。","industry","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781889466449-7e9g.png","zh","2026-06-19T17:17:20.910277+00:00",{"id":21,"slug":22,"title":23,"summary":24,"category":25,"image_url":26,"cover_image":26,"language":18,"created_at":27},"7cd85acd-4f96-43b7-a980-db5092ece240","variable-width-transformers-cut-wasted-capacity-zh","可變寬度 Transformer 省算力","這篇論文證明，Transformer 不必每一層都同寬；把前後層加寬、中央層縮窄，可以在維持表現下減少計算與記憶體。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781677978302-5pqn.png","2026-06-17T06:32:32.22399+00:00",{"id":29,"slug":30,"title":31,"summary":32,"category":25,"image_url":33,"cover_image":33,"language":18,"created_at":34},"5d6ffe73-359a-4f4a-b986-67b61ebb49c3","implicit-channel-estimation-full-duplex-mimo-zh","全雙工 MIMO 改用隱式估測","這篇論文證明，透過少量探測量測與 site-specific Transformer，就能在不明確重建自干擾通道的情況下做出有效波束設計。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779609946918-2vpm.png","2026-05-24T07:25:32.290247+00:00",{"id":36,"slug":37,"title":38,"summary":39,"category":25,"image_url":40,"cover_image":40,"language":18,"created_at":41},"d077afc5-6593-4e0f-afbf-b12229d083b6","what-large-language-models-are-how-they-work-zh","大型語言模型是什麼，怎麼運作","大型語言模型把海量文字學成可預測 Token 的系統，能寫作、摘要、翻譯，也會胡說八道。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779341172184-vgjc.png","2026-05-21T05:25:42.974116+00:00",{"id":43,"slug":44,"title":45,"summary":46,"category":25,"image_url":47,"cover_image":47,"language":18,"created_at":48},"9562db1a-00f0-4ca3-9d16-836fb6f034b9","llm-biases-agentic-ai-systems-zh","代理式 LLM 偏誤風險浮現","這篇論文把焦點放在代理式 AI 的偏誤風險，特別是已用在購物、影音與導覽等消費場景的 transformer-based 系統。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778049057942-losc.png","2026-05-06T06:30:33.94764+00:00",{"id":50,"slug":51,"title":52,"summary":53,"category":25,"image_url":54,"cover_image":54,"language":18,"created_at":55},"84c8f1a2-05f7-4ba6-ada6-192a65ca3285","clad-log-anomaly-detection-compressed-bytes-zh","CLAD 直接看壓縮位元組抓異常","CLAD 直接在壓縮位元組流上做 log anomaly detection，省掉解壓與解析流程，摘要宣稱平均 F1 達 0.9909。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1776233390200-y514.png","2026-04-15T06:09:29.899888+00:00",{"id":57,"slug":58,"title":59,"summary":60,"category":61,"image_url":62,"cover_image":62,"language":18,"created_at":63},"c34422da-87f3-4b42-9f47-36ef66e0760e","build-ai-crypto-trading-bot-guide-zh","如何打造 AI 加密貨幣交易機器人","2026 AI 加密貨幣交易機器人實作指南：資料管線、模型選擇、風控、部署與合規，幫你把想法變成能上線的系統。","blockchain","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775121945326-q2q9.png","2026-04-02T08:12:40.708166+00:00",{"id":65,"slug":66,"title":67,"summary":68,"category":25,"image_url":69,"cover_image":69,"language":18,"created_at":70},"fee7d472-a775-4b1d-bbc2-1e8bca1bbf8b","on-the-fly-repulsion-in-the-contextual-space-for-rich-divers-zh","讓AI繪圖更有創意：用排斥力提升生成多樣性","研究團隊在擴散Transformer中引入即時排斥機制，防止模式崩潰，生成更創意、更多元的文字轉圖像結果。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1774939984558-umpu.png","2026-03-31T06:01:25.439673+00:00",2]