高级 RAG 架构：从基本检索到智能 RAG [重复].pdf

《高级 RAG 架构：从基本检索到智能 RAG [重复].pdf》由会员分享，可在线阅读，更多相关《高级 RAG 架构：从基本检索到智能 RAG [重复].pdf（18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.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Vivek Mittal(he/him)Solutions ArchitectAmazon Web ServicesPallavi Nargund(she/her)Principal Soluti

2、on ArchitectAmazon Web ServicesAdvanced RAG Architectures:From Basic Retrieval to Self Corrective Agentic RAGN T A 4 0 3 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Who we arePallavi NargundVivek Mittal 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Agenda0

3、1What to expect from the code talk session02Quick overview of Amazon Bedrock KnowledgeBases03Advance RAG Techniques04Code walkthrough05Further your learning 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.What to expect from this sessionGoal:-How to improve response accuracy and r

4、educe hallucinationHow:-Using Self Corrective Agentic RAG 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Text Generation WorkflowLarge Language ModelPrompt augmentationResponseRAG in ActionEmbeddings modelData sourceVector storeEmbeddings modelEmbeddingUserUser InputContext-0.020

5、.89-0.38-0.530.950.17Data Ingestion WorkflowSemantic searchDocument chunks 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Knowledge bases for Amazon BedrockN A T I V E S U P P O R T F O R R E T R I E V A L A U G M

6、 E N T E D G E N E R A T I O N (R A G)7MODELAnthropicClaudeMetaLlamaAmazonTitan TextAI21 LabsJurassic2USER QUERYAUGMENTED PROMPTANSWERKNOWLEDGEBASES FORAMAZON BEDROCK321456Securely connect FMs to data sources for RAG to deliver more relevant responsesFully managed RAG workflow including ingestion,re