Class Central is learner-supported. When you buy through links on our site, we may earn an affiliate commission.

Beyond Basic RAG: Improving our Pipeline in Java

Go to class Write review

Overview

Enhance your RAG pipeline with advanced retrieval techniques. Combine BM25 with embeddings, refine queries iteratively, build final context, and constrain LLMs to use only retrieved data. Optionally, summarize multiple chunks into a concise context to improve accuracy and responses.

Syllabus

Unit 1: Constrained Generation in Retrieval-Augmented Generation Systems

Implementing Base Prompt Strategy for RAG Systems
Implementing Strict Prompt Strategy for RAG Systems
Implementing Citation Prompt Strategy for RAG Systems
Implementing Constrained Generation Strategies for RAG Systems
Implementing Smart Context Truncation in RAG Systems

Unit 2: Iterative Retrieval in Retrieval-Augmented Generation Systems

Complete the Inverted-Distance Similarity Score Formula
Implementing Maximum Chunk Limit in Iterative Retrieval
Enhancing Iterative Retrieval with Multiple Keywords

Unit 3: Managing Overlaps and Summarization in Retrieval-Augmented Generation Systems

Implementing Chunk Overlap Detection in RAG Pipelines
Fix Logical Bug in RAG Pipeline Summary Generation
Implementing the Final Generation Method in the RAG Pipeline

Unit 4: Hybrid Retrieval in Retrieval-Augmented Generation Systems

Converting Distance to Similarity in Hybrid Retrieval
Fix BM25 Normalization Bug in Hybrid Retrieval System
Implementing Score Filtering in Hybrid Retrieval System
Implementing BM25 Indexing for Hybrid Retrieval

Reviews

Start your review of Beyond Basic RAG: Improving our Pipeline in Java