kkdai / bookmarks

Repo to storage all my favorite links
15 stars 0 forks source link

Apple 因將 Prompt 的使用場景限縮得很精準,即便使用3B左右的 model 卻能得到很好的結果。設備端的相關資料能打造完美的 Embedding 與推論效果。後續 WWDC 的內容也都圍繞著打造 Edge LLM 相關應用。 #1600

Open kkdai opened 3 months ago

kkdai commented 3 months ago

{ "summary": "Apple 因將 Prompt 的使用場景限縮得很精準,即便使用3B左右的 model 卻能得到很好的結果。設備端的相關資料能打造完美的 Embedding 與推論效果。後續 WWDC 的內容也都圍繞著打造 Edge LLM 相關應用。", "full_content": "這篇整理得不錯:\n\n1. 因為 Apple 將 Prompt 的使用場景限縮得很精準,即便使用3B左右的 model 卻能得到很好的結果。\n2. 設備端的相關資料,能打造完美的 Embedding 與推論效果。\n3. 後續 WWDC 的內容,都在圍繞著打造 Edge LLM 相關應用。\n#AppleIntelligence #twitter", "url": "https://t.co/avxz8dWiWI", "tags": ["#AppleIntelligence", "#twitter", "NLP", "ML", "AI", "Edge LLM"] }