下一代 LLMOps：安全高效地作生成式 AI 的现代实践.pdf

1、Next Gen LLMOps:modern practices to operate gen AI safely and efficientlyThis workshop presents Next Gen LLMOps,modern LLM practices to operate your gen AI model safely and harmoniously with your data,either enriching its input(RAG)or generating new data from it(Deep Data).With countless gen AI solu

2、tions emerging all over the world,the question of using these solutions safely and efficiently,as well as customizing them for a better fit with your operations,comes up more and more frequently.While terms like DevOps,DataOps,even MLOps have been coined over the years to refer to the rationalisatio

3、n of practices linked to development,data processing or machine learning,LLMOps is an emerging term linked to Large Language Models practices.LLMOps aims to solve difficulties ranging from agressive user licenses all the way to geopolitics.This workshop will focus on presenting some simple,pragmatic

4、 solutions which can help overcoming these difficulties:from deploying on premise architectures to hybrid infrastructure,we will review how gen AI models can be put to use to your benefit,while minimizing the risks.Implanted,real world exemples will illustrate these solutions.SUBMITTED ABSTRACTINNOV

5、EZ AVEC VOS DONNESINNOVATE WITH YOUR DATA Next Gen LLMOps:modern practices to operate gen AI safely and efficientlywith Mind in a Box and the Galicia Project consortiumIntroductionIntroduction of the workshop Introduction of the speakerPresentation of the Mind in a Box solutionsMLOps and LLMOps yest

6、erday and todayDefinitions:DevOps,DataOps,MLOps,LLMOps,SecOps,BigDataOps,BigOpsLLMOps State of the artNext Gen LLMOpsModifications of the LLM landscapeThe increasing importance of DataOps and BigDataOpsThe role(or absence)of regulationsRAG in 2025Commercial ApplicationsTourismPublic safety and publi