掌握智能提示工程 [重复].pdf

《掌握智能提示工程 [重复].pdf》由会员分享，可在线阅读，更多相关《掌握智能提示工程 [重复].pdf（13页珍藏版）》请在三个皮匠报告上搜索。

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.D V T 3 3 5-RMastering agentic prompt engineeringRafael RamosPrincipal Solutions Architect,Prototyping and Cloud EngineeringAWSWill MatosPrincipal Specialist Solutio

2、ns ArchitectNext-Gen Developer ExperienceAWS 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AgendaWhat is Prompt Engineering?Key ComponentsBest PracticesAdvanced TechniquesDemos 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.What is Prompt Engineering?Why it m

3、atters:The quality of your prompt directly impacts the quality of the AIs responseCore principle:Garbage in,garbage out”better prompts=better resultsThe art and science of crafting effective instructions for AI models to get the best possible outputs.2025,Amazon Web Services,Inc.or its affiliates.Al

4、l rights reserved.Key Components of Prompt EngineeringContext SettingBackground information and domain knowledgeTask DefinitionClear instructions about what you wantInput SpecificationFormat and focus of provided dataOutput FormatStructure,length,style requirementsExamples(Few-shot Learning)Sample i

5、nputs/outputs to demonstrate expectationsConstraints&GuidelinesBoundaries and what to avoidChain of ThoughtStep-by-step reasoning instructionsIteration&RefinementTesting and continuous improvementFoundation(What&Why)Add Structure(How)Optimize for better results 2025,Amazon Web Services,Inc.or its af

6、filiates.All rights reserved.Best Practices:Structure that WorksRoleYou are a expert/analyst/assistant.ContextGiven this information.TaskYour job is to.FormatRespond in format with constraints 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affi