Amazon Bedrockã®å°Žå…¥åŠ¹æžœã‚’レビューã§ã”紹介(ã¿ã®ã‚‹ã‚“-KDDIã‚¢ã‚¸ãƒ£ã‚¤ãƒ«é–‹ç™ºã‚»ãƒ³ã‚¿ãƒ¼æ ªå¼ä¼šç¤¾)
アーã‚テクãƒãƒ£ã®æ„図・工夫AWS Amplifyを使ã£ã¦ãƒ›ã‚¹ãƒˆã—ã¦ã„ã‚‹Webアプリケーションã‹ã‚‰Amazon Bedrockã®Claudeモデルを呼ã³å‡ºã—ã¦ãƒ†ã‚ストを生æˆã—ã€ã‚¢ãƒ—リケーションã®ç”»é¢ã«è¡¨ç¤ºã•ã›ã¦åˆ©ç”¨ã—ã¦ã„ã¾ã™ã€‚ ãƒãƒƒã‚¯ã‚¨ãƒ³ãƒ‰ã®Lambdaã§ã¯Pythonプãƒã‚°ãƒ©ãƒ を稼åƒã•ã›ã¦ãŠã‚Šã€LLM用ã®ãƒ•ãƒ¬ãƒ¼ãƒ ワークã¨ã—ã¦LangChainを使ã£ã¦Bedrockã®API呼ã³å‡ºã—処ç†ã‚’ラッピングã—ã¦ã„ã¾ã™ã€‚ ã¾ãŸã€Claudeã®å¿œç”ãŒé…延ã—ãŸéš›ã«AppSyncã®GraphQLタイムアウト（30秒）ã«æŠµè§¦ã—ãªã„よã†ã€Bedrockã®å‘¼ã³å‡ºã—処ç†ã‚’éžåŒæœŸåŒ–ã—Claudeã®ãƒ¬ã‚¹ãƒãƒ³ã‚¹ã‚’DynamoDBã«ä¸€æ™‚æ ¼ç´ã™ã‚‹å·¥å¤«ã‚’ã—ã¦ã„ã¾ã™ã€‚ 導入背景一大トレンドã§ã‚る生æˆAI技術を活用ã—ã¦è‡ªç¤¾ãƒ—ãƒãƒ€ã‚¯ãƒˆã‚’開発ã—ãŸã„ã¨è€ƒãˆã¦ãŠã‚Šã€OpenAIç³»ã®ç”ŸæˆAIサービスを活用ã—ã¦ã„ã¾ã—ãŸãŒã€ã‚ã‚ŠããŸã‚Šãªãƒãƒ£ãƒƒãƒˆãƒœ
(Almost) Every infrastructure decision I endorse or regret after 4 years running infrastructure at a startup
Image from UnSplash I’ve led infrastructure at a startup for the past 4 years that has had to scale quickly. From the beginning I made some core decisions that the company has had to stick to, for better or worse, these past four years. This post will list some of the major decisions made and if I endorse them for your startup, or if I regret them and advise you to pick something else. AWS Link to
スã‚ル0ã‹ã‚‰1å¹´é–“ã§ãƒžãƒ«ã‚¦ã‚§ã‚¢è§£æžã‚’ç¿’å¾—ã—ãŸå¦ç¿’方法 - the_art_of_nerdã®ãƒ–ãƒã‚°
ã¿ãªã•ã‚“ã€ã“ã‚“ã¡ã«ã¯mã§ã™ã€‚ 今回ã¯ç§ãŒ1å¹´ã®æœŸé–“ã§ãƒžãƒ«ã‚¦ã‚§ã‚¢è§£æžã®ã‚¹ã‚ル0ã‹ã‚‰ç¿’å¾—ã™ã‚‹ã¾ã§ã«å–り組んã 方法をã¾ã¨ã‚ã¾ã—ãŸã€‚ åˆã‚ã«é–‹å§‹æ™‚ã®ã‚¹ã‚ルや1å¹´é–“ã§ã‹ã‘ãŸã‚³ã‚¹ãƒˆã€ã‚¹ã‚±ã‚¸ãƒ¥ãƒ¼ãƒ«ãªã©ã‚’共有ã—ã¾ã™ã€‚ 開始時ã®ã‚¹ã‚ルãªã© 表層解æžã€å‹•çš„解æžã€é™çš„解æžã®é•ã„ã‚‚ã‚ã¾ã‚Šç†è§£ã—ã¦ã„ãªã„ プãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ã¯Pythonã‚’å¦ã³å§‹ã‚ãŸç¨‹åº¦ 低レイヤーã®çŸ¥è˜ãªã— アセンブリã®çŸ¥è˜ãªã— Windowsアプリケーション開発ã®çµŒé¨“ãªã— SOCアナリストã¨ã—ã¦ãƒ–ルーãƒãƒ¼ãƒ ã®çŸ¥è˜ã¯å¤šå°‘ 脆弱性診æ–ã®çµŒé¨“も（深ãã¯ãªã„ãŒï¼‰å¤šå°‘ エンジニアæ´7ã€8ヶ月目ãらㄠã–ã£ãã‚Šã§ã™ãŒã€ã“ã®ã‚ˆã†ã«ãƒžãƒ«ã‚¦ã‚§ã‚¢è§£æžã«çµ¡ã‚“ã§ãる知è˜ã¯ã»ã¼0ã§ã—ãŸã€‚ å¦ç¿’ã«ã‹ã‘ãŸãƒˆãƒ¼ã‚¿ãƒ«ã‚³ã‚¹ãƒˆ 書ç±ï¼šç´„5万円 オンライン教æ：12万円 ç´„17万円近ã1å¹´é–“ã«ä½¿ã„ã¾ã—ãŸãŒã€ãŠé‡‘ã§ç›´æŽ¥ã‚¹ã‚ルã¯è²·ãˆãªã„ã®ã§å¿…è¦çµŒè²»ã§ã™ã€‚ ãªãŠã€ã‚ªãƒ³ãƒ©ã‚¤ãƒ³æ•™æã®12万円ã«ã¤ã„ã¦
Large Language Models: A Survey
Large Language Models (LLMs) have drawn a lot of attention due to their strong performance on a wide range of natural language tasks, since the release of ChatGPT in November 2022. LLMs' ability of general-purpose language understanding and generation is acquired by training billions of model's parameters on massive amounts of text data, as predicted by scaling laws \cite{kaplan2020scaling,hoffman
リリースã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“スã®ãŠçŸ¥ã‚‰ã›
最新ã®äººæ°—エントリーã®é…ä¿¡
j次ã®ãƒ–ックマーク
kå‰ã®ãƒ–ックマーク
lã‚ã¨ã§èªã‚€
eコメント一覧を開ã
oページを開ã
Scaling Laws for Fine-Grained Mixture of Experts
Rail-only: A Low-Cost High-Performance Network for Training LLMs with Trillion Parameters
Extending Context Window of Large Language Models via Positional Interpolation
PyTorch 2: Faster Machine Learning Through Dynamic Python Bytecode Transformation and Graph Compilation
GitHub - pytorch/tensordict: TensorDict is a pytorch dedicated tensor container.
TensorDict
{{#tags}}- {{label}}
{{/tags}}