æ¬è¨äºã§ã¯ãæè¿æµè¡ãã®ãçæAIã¨ã¼ã¸ã§ã³ããã«æ¿ãããAgentlessãï¼ã¨ã¼ã¸ã§ã³ãã¬ã¹ï¼ã¨ããææ³ã«ã¤ãã¦ããã£ããç解ãã¾ããæ ªå¼ä¼ç¤¾ãã¬ãã¸ã»ã³ã¹ã¯ãã¨ã³ã¿ã¼ãã©ã¤ãºä¼æ¥åãã«RAGãæä¾ãã¦ããã¹ã¿ã¼ãã¢ããã§ãã ãã®è¨äºã¯ä½ ãã®è¨äºã¯ãã½ããã¦ã§ã¢éçºèªä½ãèªååããéã®æ°ããææ³ãAgentlessãã®è«æ[1]ã«ã¤ãã¦ãæ¥æ¬èªã§ç°¡åã«ã¾ã¨ãããã®ã§ãã ä»åããããããAIã¨ã¼ã¸ã§ã³ãã¨ã¯ï¼ãã«ã¤ãã¦ã¯ãç¥ã£ã¦ããåæã§é²ã¿ã¾ãã確èªããå ´åã¯ããã¡ãã®è¨äºãªã©ããåèä¸ããã æ¬é¡ ãã£ãããµããªã¼ æè¿ãã½ããã¦ã§ã¢éçºã§çæAIã使ããã¨ã¯å½ããåã«ãªã£ã¦ãã¾ããï¼GitHub CopilotãCursorãDevinãªã©ãæ§ã ãªãã¼ã«ããªãªã¼ã¹ããã¦ãã¾ããï¼ ãAGENTLESSãã¯ãã½ããã¦ã§ã¢éçºãèªååãããã¨ããæèã§LLMã使ãéã®ãæ°ããææ³
以ä¸ã®æç« ã¯ãã³ãªã¤ã»ãã¯ããã¦ã®ãPredicting the presentãã¨ããè¨äºã翻訳ãããã®ã§ããã Pluralistic 2018å¹´ãç±³å½å¸æ°æ¨©ç³è«ã«ã¤ãã¦ç§»æ°å¼è·å£«ã«ç¸è«ãã¦ããé ãç§ã¯ãRadicalizedï¼ã©ãã£ã«ã©ã¤ãºãï¼ãã¨ããçç·¨å°èª¬ã®å·çãå§ããããã®ä½åã¯2019å¹´ã«åºçãããçç·¨éã®ã¿ã¤ãã«ä½åã¨ãªã£ãã https://us.macmillan.com/books/9781250228598/radicalized/ ãRadicalizedãã¯ãç±³å½ã¨éãå»çãããã¦æ´åã«ã¤ãã¦ã®ç©èªã ãç§ã¯ ãã¼ãã³ã¯ï¼è¨³æ³¨ï¼ã«ãªãã©ã«ãã¢å·ã®é½å¸ï¼ã«ä½ãã§ãããããã¯ç±³å½ã§äººå£ãããã®éå¨åºæ°ã2çªç®ã«å¤ãçºã§ãé£æ¥ããããµã³ã¼ã«ã¹å¸ã®è¦å¶ãå³ããããããã種ã®è¦å¶è£å®ã®çµæãããªã£ã¦ãããéã欲ããããµã³ã¼ã«ã¹å¸æ°ã¯ãã¾ã ãã¼ãã³ã¯ã訪ããã ããã ã«ãã
tl;drJSON Schema ã§æå®ãããã©ã¼ãããã§åºåãå¶å¾¡å¯è½ã«ãªã£ãã cURL / Python / JavaScript ã®ããããã§è©¦ãã¦ã¿ãã å ·ä½çãªå®ç¨ä¾ããã£ãã®ã§ãããåããã¦ã¿ãã 使ãä¸ã§ã® tips ãä»å¾ã©ããªæ©è½ã追å ããããã¾ã¨ããã å ¬éãããããã°ã®æµãã«æºæ ãã¤ã¤ãæå³ããºã¬ãªãç¯å²ã§ç¿»è¨³ã解説ãã³ã¼ãã®å®è¡ããã¦ããã¾ãããã¥ã¼ããªã¢ã«ã«ãªã£ã¦ããã®ã§ãããã£ããæãåããã¦è©¦ãã¦ã¿ã¦ãã ããã Ollama ã structured outputs ããµãã¼ããJSON Schema ã§å®ç¾©ãããã©ã¼ãããã« LLM ã®åºåãå¶å¾¡ãããããã¨ãå¯è½ã«ãªãã¾ãããOllama ã® Python 㨠JavaScript ã®ããããã®ã©ã¤ãã©ãªã«ããã¦ããµãã¼ãããããæ´æ°ã ããã°ã§ã¯ structured outputs ã®ã¦ã¼ã¹ã±ã¼ã¹ã¨ã
æ¥æ¬èªLLMã¾ã¨ã â[ English | Français | æ¥æ¬èª ] æ¥æ¬èªLLMã»æµ·å¤LLMã®ãã©ã¡ã¼ã¿æ°ã®æ¨ç§»ãæ¥æ¬èªã¢ãã«ã®æ å ±ã¯æ¬è¨äºãæµ·å¤ã¢ãã«ã®æ å ±ã¯ LifeArchitect.ai ã® Models table ãåç §ãã¦ãã¾ãï¼ãã ããå³ã®ã¹ãã¼ã¹ä¸ä¸é¨ã®ã¢ãã«ã¯çç¥ãã¾ããæµ·å¤ã¢ãã«ã®ãã©ã¡ã¼ã¿æ°ã¯æ¨æ¸¬å¤ãå«ãï¼ãä¿®æ£ã»è¿½å çããã¾ããããç¥ããä¸ããããã®è¨äºã¯ãä¸è¬å ¬éããã¦ããæ¥æ¬èªLLMï¼æ¥æ¬èªãä¸å¿ã«å¦ç¿ãããLLMï¼ããã³æ¥æ¬èªLLMè©ä¾¡ãã³ããã¼ã¯ã«é¢ããæ å ±ãã¾ã¨ãããã®ã§ããæ å ±ã¯ãæå¿ã«ããåéããã¦ããããã®ä¸é¨ã¯è«æãå ¬éããã¦ãããªã½ã¼ã¹ãªã©ããå¼ç¨ãã¦ãã¾ãã 以ä¸ã®ç¹ã«ã¤ãã¦ãããããããç解ã¨ãäºæ¿ããé¡ããããã¾ã æ¬è¨äºã®å 容ã¯ãå®å ¨æ§ãæ£ç¢ºæ§ãä¿è¨¼ãããã®ã§ã¯ããã¾ããããããã®æ å ±ã¯äºåãªãå¤æ´ããããã¨ããããã¾ãææ°
ç»åçæAIãStable Diffusion XLï¼SDXLï¼ãã®ãã¡ã¤ã³ãã¥ã¼ãã³ã°ã¢ãã«ããNoobAI-XLï¼ãã¼ãã¨ã¼ã¢ã¤ï¼ãã®ç»å ´ããç»åçæAIé¢é£ã®äººãã¡ãããã¤ããã¦ãã¾ããã¤ã©ã¹ã風ç»åã®çææ§è½ãé«ãããã§ãããã¤ã¨ã³ãã¢ãã«ã®ãStable Diffusion 3.5ãããFlux.1ãã¸ã®ç§»è¡ãå§ã¾ããã¨ãããªããæ¨å¹´8æã«ãªãªã¼ã¹ãããæ§ã¢ãã«ã®SDXLã«æ§è½åä¸ã®ä½å°ãã¾ã ã¾ã ãããã¨ãæããã«ãªã£ã¦ããã®ã§ãã SDXLãã¼ã¹ã®ç»åçæAIã¢ãã«ãNoobAI-XLãã®æ§è½ãé«ããã NoobAI-XLã®å¼·åãã¯ãLCMï¼Latent Consistency Modelsï¼ç»åçæAIã®é«éåæè¡ï¼ã®ç°å¢ã§åããã¨ãã®ãããããããããã¾ãããã¤ã³ããã¼ã«ãKritaãç¨ã«Aclyãããéçºãç¶ãã¦ããç¬èªãã©ã°ã¤ã³ãKrita AI Diffusionãã§
ãµã¤ãã¼ã»ãã¥ãªãã£ã®å°é家ãä¸å ã«ä¼ãããæ ªå¼ä¼ç¤¾ç¶²å±ã®ãSecurity BLAZE 2024ããç¾ä»£ã®ä¼æ¥éå¶ã«ã¯ãã©ã³ãµã ã¦ã§ã¢ã«ä»£è¡¨ããããµã¤ãã¼æ»æãå é¨ä¸æ£ãªã©ããã¾ãã¾ãªã»ãã¥ãªãã£ãªã¹ã¯ãä¼´ãã¾ããæ¬ã»ãã·ã§ã³ã§ã¯ããµã¤ãã¼ã»ãã¥ãªãã£ã®å°é家ããè å¨ã¢ã¯ã¿ã¼ã®æ»æææ³ã対å¿çã解説ãã¾ãããå¾ç·¨ã¯ãä¸å½ãåæé®®ããã·ã¢ãªã©ã®è å¨ã¢ã¯ã¿ã¼ã使ç¨ããæ»ææ段ããæµ·å¤ã§ãµã¤ãã¼æ»æè ã®äººæè²æãæ¨å¥¨ããã¦ããèæ¯ãã²ãã¨ããªãããæ¥æ¬ä¼æ¥ã¸ã®æã¡æã«ã¤ãã¦ã¢ããã¤ã¹ãè´ãã¾ãã ã¹ãã¤æ´»åã身代éç®çã§çãããä¼æ¥å¯ºå²¡ç¯¤å¿æ°ï¼ä»¥ä¸ã寺岡ï¼ï¼æ¬¡ã¯å¤§ä½ãããããããã¤ãäºä¾ãç´¹ä»ãã¦ããã ããã°ã¨æãã¾ããã¾ãããAPT41ãããã§ããã 大ä½æ°ï¼ä»¥ä¸ã大ä½ï¼ï¼ãããã¾ããããAPTãã®ç¥èªä½ãç¥ã£ã¦ãã人ã¨ç¥ããªã人ãããã¨æãã®ã§ãç°¡åã«ã説æãã¾ãããAdvanced Pe
ããã«ã¡ã¯ããã«ããã· å¾æ¥å¡ããµã¼ãã¹ãã¼ã ã®ã½ããã¦ã§ã¢ã¨ã³ã¸ãã¢ã® a2 (A2hiro_tim) ã§ãã æ©éã§ãããã¦ã§ãã¢ããªã±ã¼ã·ã§ã³éçºãç¹ã« Go è¨èªã使ã£ã¦ããæ¹ã¯ããã¹ããã©ã®ããã«æ¸ãã¦ãã¾ãã§ããããã æã ã®ãã¼ã ã¯ããã¯ã¨ã³ãã« Go ãæ¡ç¨ãã¦ããã Table Driven Test ï¼ä»¥ä¸ TDT ï¼ã使ã£ã¦ãã¾ãããGo ãæ¡ç¨ããéçºã§ã¯ TDT ãåºãæ¡ç¨ããã¦ãããæã ãæ £ç¿ã«åãã®ãæåã ã¨å¤æããããã§ãã ãããå®éã«éçºãé²ããã¨ããã¨ã¦ã§ãã¢ããªã±ã¼ã·ã§ã³éçºã«ããã TDT ã«ã¯ããã¤ã課é¡ãæããããã«ãªãã¾ãããããã§æã ã®ãã¼ã ã§ã¯ runn ã¨ãããã¼ã«ããã¹ãã«æ´»ç¨ããæ¹éã«åãæ¿ãããã®çµæãéçºãå éãã価å¤æä¾ãæ©ããããã¨ãã§ãã¾ããã æ¬è¨äºã§ã¯ãTDT ã®èª²é¡ãrunn ã®æ¡ç¨çµç·¯ããã®å¾ã«ã¤ãã¦æã ã®ãã¼ã ã®
ããã±ã¢ã³GOãããIngressãã¨ãã£ãä½ç½®æ å ±ã²ã¼ã ã®éçºã»éå¶å ã¨ãã¦ç¥ãããNianticããããã¾ã§ã®ã²ã¼ã éå¶ã§èç©ãã¦ããè¦è¦æ¸¬ä½ã·ã¹ãã (Visual Positioning Systemï¼VPS)æ å ±ãæ´»ç¨ããå½¢ã§ã空éã³ã³ãã¥ã¼ãã£ã³ã°ãã©ãããã©ã¼ã ã§ãããNiantic Spatial Platformããæ§ç¯ãã¾ãããã¾ããVPSã§å¾ããã100ä¸ä»¥ä¸ã®ã¹ãããæ å ±ãåºç¤ã¨ãã¦ãç©ççãªä¸çãããã²ã¼ãå¯è½ãªã大è¦æ¨¡å°ç空éã¢ãã«(Large Geospatial Modelï¼LGM)ããæ§ç¯ãã¦ãããã¨ãæããã«ãã¦ãã¾ãã Niantic Spatial Platform https://www.nianticspatial.com/ Building a Large Geospatial Model to Achieve Spatial Intellige
声ã«é©å½ããAivis Project ãåãæããæåã®é³å£°ä½é¨ã Aivis Project ã¯ãææ è±ããªé³å£°åææè¡ã誰ããããããã«æ´»ç¨ã§ããæªæ¥ãç®æãã 壮大ãªéçºããã¸ã§ã¯ãã§ãã æ©æ¢°çãªé¿ãã«çã¾ããªããã¾ãã§éã宿ãããã®ãããªé³å£°ããããªãã®æ³ããè¨èã«è±ããªææ ãã®ã㦠ä¸çã¸ã¨é¿ã渡ããAivis Project ãæãæªæ¥ã¯ãããã¾ã§ã«ãªãæåã¨é©ãã«æºã¡ã¦ãã¾ãã 好ããªãã£ã©ã¯ã¿ã¼ã®å£°ã§ç©èªãç´¡ããçæ³ã®å£°ã§ãã¥ã¼ã¹è¨äºã«å½ãå¹ãè¾¼ãã ããããæ°ãã表ç¾ããç§ãã¡ã®ãããã¯ããéãã¦æ°ã¯ãªãã¯ã§å®ç¾ãã¾ãã AivisSpeech ã§ã®ãã¼ã«ã«é³å£°åæãããã¢ãã«ã®å¶ä½ã»ããã¯ã¹ã»å ¬éã¾ã§ãAivis Project 㯠誰ããèªç±ã«å¥½ããªå£°ãå ±æã§ããæªæ¥ãå½¢ã«ãã¦ããã¾ãã ããã¾ã§ä¸é¨ã®æè½ãç¹å¥ãªç°å¢ã ããæã«ã§ãã¦ããé åçãªå£°ããåã ã®ã¢ã¤ã
ããã«ã¡ã¯ã鹿éã§ãã ãå½ãã«å¯¾ãããåãããæ²¢ãã«å¯¾ããã澤ããªã©ã¯æ§åä½ã¨å¼ã°ãã¦ãã¾ãããæ§åä½ãã¨ããè¨èèªä½ãç¥ããªãã¦ãããæã¯ä»ã¨ã¯éãæ¼¢åã®æ¸ãæ¹ããã£ãããããã¯ãªãã¨ãªãç¥ã£ã¦ããããã¨ããæ¹ã¯å¤ãã¨æãã¾ãã â²ãã®è¨äºã§ã¯Kãæ§åä½ãSãæ°åä½ã§ããã¨ããKï¼Sãã¨è¡¨ããã®ã¨ãã¾ã ãããæ®å¿µãªãã¨ã«ãæ¼¢åã好ãã§ããæ§åä½ãã®ãã¨ãé°å²æ°ã§ããç解ãã¦ãããããå¤ãæ¼¢åãç¨åº¦ã«ããæãã¦ããªã人ãã»ã¨ãã©ã§ãããã®ããã以ä¸ã®ãããªåéãããã人ãé常ã«å¤ããã¾ãã â²ããè¦ããããå®ã¯å ¨é¨ééã ãããã®çºè¨ã¯ãªãééããªã®ã§ããããï¼ ç®æ¬¡ âæå¤ã¨ç¥ããªããæ§åä½ãã®å®ç¾© âãæ§åä½ãã®å®ç¾©ã«ããã注æç¹ â誤解â ï¼æ°åä½ã¯æ¦å¾æ¥æ¬ã§ã¾ã¨ãã¦ä½ããã â誤解â¡ï¼æ¦åæ¥æ¬ã§ã¯å°ãæ§åä½ã使ããã¦ãã â誤解â¢ï¼æ°åä½ã¯æ§åä½ãå¤åããã¦ã§ãã âæ§åä½ã¨èª¤è§£
æ£çµ±é²åããç°å½¢ã®çæç¨ãã¼ããããæ £ãã¯å¿ è¦ã ãå©ä¾¿æ§ã¯é«ã AzeronãCyborg II Text byÂ å «å²¡å¼é« PCã§ãã¦ã¹ã使ã£ã¦ã²ã¼ã ããã¬ã¤ããã¨ãã¯ï¼å¤§æµã¯ãã£ã©ã¯ã¿ã¼ã®ç§»åã«ãã¼ãã¼ãã®ï¼»W/A/S/Dï¼½ãã¼ã使ãã®ãä¸è¬çã ããããï¼ãã¦ã¹æä½ã«åºãé¢ç©ã使ãããã®ã§ãã¼ãã¼ããéªéã¨ãï¼ã²ã¼ã ãããã«æ £ãã¦ããã®ã§ç§»åã¯ã¹ãã£ãã¯ã§è¡ãããã¨ãã£ãçç±ã§ï¼ãã¼ãã¼ãã§ã®æä½ã好ã¾ãªãã²ã¼ãã¼ãããã ããã ãããªã¦ã¼ã¶ã¼ã«å¥½ã¾ããã®ãï¼çæç¨ãã¼ããããå·¦æç¨ããã¤ã¹ãªã©ã¨å¼ã°ãã製åã ãã¨ãã«ã²ã¼ã ç¨éã¨ãã¦ã¯ï¼Razerã®ãRazer Tartarus Proãã®ããã«ãã¼ã®æ°ãå¤ãï¼ã¢ããã°ã¹ãã£ãã¯ãåãã製åã好ã¾ãã¦ããã ãããªãªãï¼è¦ãç®ã®ã¤ã³ãã¯ãããããã¨ãªããï¼å¾æ¥ã®çæç¨ãã¼ãããã®åé¡ç¹ã解æ¶ãããã¨ãã¦ç»å ´ããã®ãï¼ã©ããã¢ã®ä¼æ¥ã§ãã
è¸ã£ã¦ãããã«ããè¦ããªãåãç©ä½ã®åç»ãXï¼Twitterï¼ã«æ稿ãããè¨äºå·çæç¹ã§830ä¸å以ä¸ã®è¡¨ç¤ºã2ä¸2000件ãè¶ ããããããéãã¦ãã¾ãã è³ã®ç¹æ§ãæ´»ãããåç»ã ã¨è©±é¡ã« æ稿ããã®ã¯ãæ¯æ¥ãã¶ã¤ã³è³ãæååºè¸è¡é¸å¥¨æ°äººè³ã¡ãã£ã¢è¸è¡é¨éãªã©ã®åè³çµé¨ããããã¶ã¤ãã¼ã®ä¸æåå¾ï¼@yugopï¼ããã話é¡ã«ãªã£ã¦ããã®ã¯éè²ã«å ãå¤é¢ä½ã®åç»ã§ããè¸ã人éã«è¦ãã¦ãã¾ãå¤é¢ä½ã¨ã¯ãä¸ä½ã©ã®ãããªåããè¦ãã¦ããã®ã§ããããã ç«ä½ã®æåã横ããè¦ããããªå¤é¢ä½ æåã横ããè¦ããããªå¤é¢ä½ã¯ãã¡ããã©ååãããã®é«ãã«ãã両端ã®è§ã人éã®è ã®ããã«ä¸ããå°ã«ä»ãã¦ããå³ã®è§ã軸ã«ã足ãä¸ããããã«å対ã®è§ãåããã¾ããä¸ãã£ãè§ã足ãåºã«ä»ãããã«ä¸ã«éããã¨ã2ã¤ã®è§ã¯è»½ããã«ã¹ããããè¸ãã§ããã¾ããã¾ãã§ã·ã¼ããé ãã被ã£ã人éãåãã¦ããããã§ãã 両æãä¸ãã¸ã£ã³
ããã§ç¥ã£ãã 試ãã«ãç¥æ¸å¸ãå ¬éãã¦ãã観å ã«é¢ããçµ±è¨ã»èª¿æ»è³æã®ãã¡ãã令å5年度 ç¥æ¸å¸è¦³å åå調æ»çµæã«ã¤ãã¦ãã®PDFã§ä¸åº¦è©¦ãã¦ããã®ã ãã©ã: ï¼åºå ¸ï¼ ç¥æ¸å¸Webãµã¤ãã®ã観å ã«é¢ããçµ±è¨ã»èª¿æ»ãã®ãã¼ã¸ ä¸è¨ã«ããã令å5年度 ç¥æ¸å¸è¦³å åå調æ»çµæã«ã¤ãã¦ãã®PDF æ¥æ¬èªã§ããæ¦ãåé¡ãªãã表ãªã©ããããã«ãã¼ã¹ããã ãã ãã表ãç»å ã«ãªã£ã¦ãå ´åã«ããã¾ã解éãããªãï¼è¡¨ã¯è§£éãããããä¸ã®æåãåããï¼ OCRãæ£ããã§ãã¦ããªãå¯è½æ§ ã¨ããã®ããã£ã¦ãããã¥ã¡ã³ãèªãã ãã©ããããªãã¦ããã以ä¸æ·±è¿½ããã¦ãªãã£ãã ãã ãXçéãè¦ãéãã¯è©å¤ã¯è¯ãããã§ãããããè¨äºãåºã¦ããã¿ãããªã®ã§ãæ¹ãã¦è©¦ãã¦ã¿ãã GitHubã¬ãã¸ã㪠ããã¥ã¡ã³ã Docling Doclingã¯ãããã¥ã¡ã³ãã解æããç°¡åãã¤è¿ éã«å¸æã®å½¢å¼ã«ã¨ã¯ã¹ãã¼ããã¾ãã
OrbStackã¨ã¯ OrbStackã¯ã軽éã§é«ããã©ã¼ãã³ã¹ãªä»®æ³åãã©ãããã©ã¼ã ã§ã主ã«macOSåãã«æä¾ããã¦ãã¾ããDockerã³ã³ãããLinuxä»®æ³ãã·ã³ãé«éã§åä½ããããã¨ãã§ããç¹ã«Appleã·ãªã³ã³ï¼M1/M2ï¼Macã§ã®å©ç¨ã«æé©åããã¦ãã¾ããDocker Desktopã«ä»£ãããã¼ã«ã¨ãã¦æ³¨ç®ããã¦ããããªã½ã¼ã¹å¹çãè¯ããã·ã¹ãã ã®è² è·ãæããããã®ãç¹å¾´ã§ãã ããªãOrbStackãé¸ã¶ãï¼ã â¡ï¸ è¶ é«é : 2ç§ã§èµ·åãæé©åããããããã¯ã¼ã¯ã¨ãã¡ã¤ã«ã·ã¹ãã ãé«éãªx86ã¨ãã¥ã¬ã¼ã·ã§ã³ã ð¨ è¶ è»½é : ä½CPUã¨ãã£ã¹ã¯ä½¿ç¨éãããããªã¼ã«åªãããå°ãªãã¡ã¢ãªã§ãåä½ããã¤ãã£ãã®Swiftã¢ããªã ð° ã·ã³ãã«ã§æéããã : èªåã§ãã¡ã¤ã³åã¨ãã¤ã°ã¬ã¼ã·ã§ã³ãè¨å®ãCLIã§ã³ã³ããã»ã¤ã¡ã¼ã¸ã»ããªã¥ã¼ã ãã¡ã¤ã«ã«ã¢ã¯ã»ã¹ãVPNã¨
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}