[B! LLMOps] deejayrokaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

deejayroka id:deejayroka

LLMOpsã«é–¢ã™ã‚‹deejayrokaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (9)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

LLMOpsã‚’æŽ¨é€²ï¼Azure Log Analyticsã‚’æ´»ç”¨ã—ãŸLLMã®ç¶™ç¶šçš„ãªç›£è¦–ã¨æ”¹å–„ - AITC - ISID | AI ãƒˆãƒ©ãƒ³ã‚¹ãƒ•ã‚©ãƒ¼ãƒ¡ãƒ³ã‚·ãƒ§ãƒ³ã‚»ãƒ³ã‚¿ãƒ¼ ã‚³ãƒ©ãƒ
ã“ã‚“ã«ã¡ã¯ï¼AIã‚½ãƒªãƒ¥ãƒ¼ã‚·ãƒ§ãƒ³ã‚°ãƒ«ãƒ¼ãƒ—ã®å¤ªç”°ã§ã™ã€‚ ã“ã®ã‚³ãƒ©ãƒ ã§ã¯ã€Azure Log Analyticsã‚’ä½¿ã£ãŸLLMOpsã®å®Ÿç¾æ–¹æ³•ã«ã¤ã„ã¦ç´¹ä»‹ã—ã¾ã™ã€‚ æ˜¨å¹´ã‹ã‚‰å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ï¼ˆLLMï¼‰ã‚’è£½å“ã‚„ã‚µãƒ¼ãƒ“ã‚¹ã«çµ„ã¿è¾¼ã‚€ä¼æ¥ãŒå¢—ãˆã¦ã„ã¾ã™ã€‚ ã—ã‹ã—ã€LLMã‚µãƒ¼ãƒ“ã‚¹ã®å“è³ªã‚’ç¶æŒã™ã‚‹ã«ã¯ã€ãã®é‹ç”¨ã«ã‚‚æ³¨æ„ã‚’æ‰•ã†å¿…è¦ãŒã‚ã‚Šã¾ã™ã€‚ å…·ä½“çš„ã«ã¯ã€LLMã®å‡ºåŠ›ã®å“è³ªç®¡ç†ã‚„ã€ãƒ¦ãƒ¼ã‚¶ãƒ¼ã‹ã‚‰ã®ãƒ•ã‚£ãƒ¼ãƒ‰ãƒãƒƒã‚¯ã‚’å…ƒã«ã—ãŸãƒ—ãƒãƒ³ãƒ—ãƒˆã®æœ€é©åŒ–ãªã©ã€ç¶™ç¶šçš„ãªç›£è¦–ã¨æ”¹å–„ãŒæ±‚ã‚ã‚‰ã‚Œã¦ã„ã¾ã™ã€‚ ã“ã‚Œã‚‰ã®é‹ç”¨ä¸Šã®æ´»å‹•ã«Azure Log AnalyticsãŒå½¹ç«‹ã¡ã¾ã™ã€‚ LLMOpsï¼ˆLLMï¼ˆLarge Language Modelï¼‰+ Opsï¼ˆOperationsï¼‰ï¼‰ã¨ã¯ LLMOpsã¯è£½å“ã«çµ„ã¿è¾¼ã¾ã‚ŒãŸLLMã®é‹ç”¨ã«å¿…è¦ãªãƒ™ã‚¹ãƒˆãƒ—ãƒ©ã‚¯ãƒ†ã‚£ã‚¹ã®æ¦‚å¿µã‚’æŒ‡ã—ã¾ã™ã€‚ ä¾‹ãˆã°ã€LLMã®é‹ç”¨ã§ã¯LLMã®å‡ºåŠ›ã®ç›£è¦–ã¨è©•ä¾¡ã¨ãƒ—ãƒãƒ³ãƒ—ãƒˆç®¡ç†
deejayroka 2024/07/18
LLMOps
ãƒªãƒ³ã‚¯
Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations | Research - AI at Meta
Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations æ¦‚è¦We introduce Llama Guard, an LLM-based input-output safeguard model geared towards Human-AI conversation use cases. Our model incorporates a safety risk taxonomy, a valuable tool for categorizing a specific set of safety risks found in LLM prompts (i.e., prompt classification). This taxonomy is also instrumental in classify
deejayroka 2024/02/29
LLMOps
ãƒªãƒ³ã‚¯
Engineering Practices for LLM Application Development
LLM engineering involves much more than just prompt design or prompt engineering. In this article, we share a set of engineering practices that helped us deliver a prototype LLM application rapidly and reliably in a recent project. We'll share techniques for automated testing and adversarial testing of LLM applications, refactoring, as well as considerations for architecting LLM applications and r
deejayroka 2024/02/25
LLMOps
ãƒªãƒ³ã‚¯
LLMOps: Engineering Trustworthy LLM Systems - Fontys
deejayroka 2024/02/25
LLMOps
ãƒªãƒ³ã‚¯
LLM ãƒ†ã‚¯ãƒ‹ãƒƒã‚¯ã®ç¿’å¾—: LLMOps
Reading Time: 3 minutes ä¼æ¥ã¯ã€ã‚¤ãƒŽãƒ™ãƒ¼ã‚·ãƒ§ãƒ³ã‚’èµ·ã“ã—ã€é¡§å®¢ã«ä¾¡å€¤ã‚’æä¾›ã—ã€ç«¶äº‰åŠ›ã‚’ç¶æŒã™ã‚‹ãŸã‚ã«ã€ã“ã‚Œã¾ã§ä»¥ä¸Šã«ãƒ‡ãƒ¼ã‚¿ã¨ AI ã«ä¾å˜ã—ã¦ã„ã¾ã™ã€‚æ©Ÿæ¢°å¦ç¿’ (ML) ã®å°Žå…¥ã¯ã€ä¿¡é ¼æ€§ãŒé«˜ãã€ã‚³ã‚¹ãƒˆ ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ãŒã‚ˆãã€ã‚¹ã‚±ãƒ¼ãƒ©ãƒ–ãƒ«ãªã‚³ãƒ¼ãƒ‰ã€ãƒ‡ãƒ¼ã‚¿ã€ãƒ¢ãƒ‡ãƒ«ã‚’ç®¡ç†ã™ã‚‹ãŸã‚ã®ãƒ„ãƒ¼ãƒ«ã‚„ãƒ—ãƒã‚»ã‚¹ã€çµ„ç¹”åŽŸå‰‡ã®å¿…è¦æ€§ã‚’ç”Ÿã˜ã•ã›ã¾ã—ãŸã€‚ã“ã‚Œã¯ MLOps ã¨ã—ã¦åºƒãçŸ¥ã‚‰ã‚Œã¦ã„ã¾ã™ã€‚ ä¸–ç•Œã¯ã€ç‰¹ã«åŸºç›¤ãƒ¢ãƒ‡ãƒ«ã¨å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ« (LLM) ã‚’æ´»ç”¨ã—ãŸæ–°ã—ã„ç”Ÿæˆ AI æ™‚ä»£ã«æ€¥é€Ÿã«çªå…¥ã—ã¦ã„ã¾ã™ã€‚ChatGPT ã®ãƒªãƒªãƒ¼ã‚¹ã«ã‚ˆã‚Šã€ã“ã®å¤‰åŒ–ã¯ã•ã‚‰ã«åŠ é€Ÿã•ã‚Œã¾ã—ãŸã€‚ GenAIOps ã¨ LLMOps ã®æ–°ã—ã„å°‚é–€é ˜åŸŸã¯ã€æœ¬ç•ªç’°å¢ƒã§ã®ç”Ÿæˆ AI ãŠã‚ˆã³ LLM ã‚’åˆ©ç”¨ã—ãŸã‚¢ãƒ—ãƒªã®é–‹ç™ºã¨ç®¡ç†ã®èª²é¡Œã«å¯¾å‡¦ã™ã‚‹ãŸã‚ã® MLOps ã®é€²åŒ–ã¨ã—ã¦ç™»å ´ã—ã¾ã—ãŸã€‚ ã“ã®æŠ•ç¨¿ã§ã¯ã€ç”Ÿæˆ AI ã‚¢ãƒ—ãƒªé–‹ç™º
deejayroka 2024/02/22
AI

LLMOps

LLM

MLOps

ãƒ¢ãƒ‡ãƒ«

é–‹ç™º
ãƒªãƒ³ã‚¯
LLM-in-Production-Meetup-1-LLMOps-LLMã®è©•ä¾¡å“è³ªæ‹…ä¿-_2023.08.18
deejayroka 2024/02/15
LLMOps
ãƒªãƒ³ã‚¯
`Building LLM applications for production`ã‚’èªã‚“ã
æœ€è¿‘ã¯æœ¬æ¥ã§ã‚‚è¶£å‘³ã§ã‚‚LLMãŒè‡ªåˆ†ã®æ™‚é–“ã®å¤šãã‚’å ã‚ã‚‹ã‚ˆã†ã«ãªã£ã¦ã‚‹ã€‚ LLMã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ for ãƒ—ãƒãƒ€ã‚¯ã‚·ãƒ§ãƒ³ã€ä½•ã ã‹ã‚“ã è€ƒãˆã‚‹ã“ã¨ãŒå¤šãã¦å¤§å¤‰ã ãªã€œã¨ã„ã†ã®ã§å‚è€ƒã¨ã—ã¦Building LLM applications for productionã‚’èªã‚“ã ã€‚ æ„Ÿæƒ³ã¨ã—ã¦ã¯è¨˜äº‹ã®å†’é ã¨æœ€å¾Œã«ã‚ã£ãŸâ†“ãŒå…¨ã¦ã€‚hard to make something production-ready with themã€‚ ãƒ»Itâ€™s easy to make something cool with LLMs, but very hard to make something production-ready with them. ãƒ»LLM limitations are exacerbated by a lack of engineering rigor in prompt engineering,
deejayroka 2024/02/15
LLMOps
ãƒªãƒ³ã‚¯
LLMOpsï¼šå¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ã®æ“ä½œã‚’ç†è§£ã™ã‚‹
deejayroka 2023/09/29
AI
ãƒªãƒ³ã‚¯
Patterns for Building LLM-based Systems & Products
Patterns for Building LLM-based Systems & Products [ llm engineering production ðŸ”¥ ] Â· 66 min read Discussions on HackerNews, Twitter, and LinkedIn â€œThere is a large class of probl ems that are easy to imagine and build demos for, but extremely hard to make products out of. For example, self-driving: Itâ€™s easy to demo a car self-driving around a block, but making it into a product takes a decade.â€
deejayroka 2023/08/30
"Patterns for Building LLM-based Systems & Products"

LLM

LLMOps
ãƒªãƒ³ã‚¯
1