Course Wiki
Best Practices for RAG — LLM Zoomcamp Module 6
In the previous modules we built search engines, agents, evaluation pipelines, and monitoring dashboards. This module covers techniques that improve retrieval quality - the most critical part of any RAG system.
Related Pages
LLM Zoomcamp › Module 6: Best Practices › Best Practices for RAG
Notes
Best Practices for RAG
In the previous modules we built search engines, agents, evaluation pipelines, and monitoring dashboards. This module covers techniques that improve retrieval quality - the most critical part of any RAG system.
The RAG pipeline
A RAG system has two stages:
- Indexing: documents are parsed, chunked, embedded, and stored in a vector database
- Retrieval: the user’s question is embedded, the database is searched for the closest documents, and those documents are passed to the LLM as context
Most of the improvements we’ll discuss focus on the retrieval stage, because better retrieval means better answers.
Five techniques
Here are the five techniques we’ll cover:
- Small-to-big chunk retrieval - use small chunks for indexing but retrieve surrounding context for the LLM
- Leveraging document metadata - use titles, topics, dates, and other metadata to filter and boost results
- Hybrid search - combine vector search (semantic) with keyword search (lexical) in one query
- User query rewriting - use an LLM to reformulate the user’s question into a better search query
- Document reranking - re-order retrieved documents by relevance after the initial search
In this module, we’ll focus on hybrid search and reranking. These two techniques give the biggest improvement for the least effort.
Slides: llm-zoomcamp-best-practicies.pdf
To learn more:
Key concepts
Related notes
- llmz-m01-agents
- llmz-m01-building-the-prompt
- llmz-m01-data-ingestion
- llmz-m01-environment
- llmz-m01-function-calling
- llmz-m01-introduction
- llmz-m01-other-frameworks
- llmz-m01-quick-rag-revision-optional
- llmz-m01-rag
- llmz-m01-rag-helper
- llmz-m01-search
- llmz-m01-the-course-faq-dataset
- llmz-m01-the-llm
- llmz-m01-wrap-up-of-part-1
- llmz-m02-embeddings
- llmz-m02-next-steps
- llmz-m02-rag-with-vector-search
- llmz-m02-using-onnx-runtime-instead-of-pytorch
- llmz-m02-vector-search
- llmz-m02-vector-search-2
- llmz-m02-vector-search-with-minsearch
- llmz-m02-vector-search-with-pgvector
- llmz-m02-vector-search-with-sqlitesearch
- llmz-m03-ai-agents
- llmz-m03-ai-orchestration
- llmz-m03-best-practices
- llmz-m03-next-steps
- llmz-m03-retrieval-augmented-generation
- llmz-m04-agent-evaluation
- llmz-m04-evaluation
- llmz-m04-generating-ground-truth-data
- llmz-m04-generating-ground-truth-for-all-documents
- llmz-m04-generating-rag-answers
- llmz-m04-llm-as-a-judge
- llmz-m04-next-steps
- llmz-m04-rag-and-agent-evaluation
- llmz-m04-search-evaluation
- llmz-m04-search-evaluation-metrics
- llmz-m04-search-parameter-tuning
- llmz-m05-assistant
- llmz-m05-built-in-judge
- llmz-m05-capturing-metrics
- llmz-m05-chat-app
- llmz-m05-docker-compose
- llmz-m05-grafana-dashboards
- llmz-m05-monitoring
- llmz-m05-next-steps
- llmz-m05-storing-data-in-postgresql
- llmz-m05-user-feedback
- llmz-m06-document-reranking
- llmz-m06-hybrid-search
- llmz-m06-hybrid-search-with-langchain
- llmz-m06-next-steps
- llmz-m07-chunking-for-longer-texts
- llmz-m07-content-processing-cases-and-steps
- llmz-m07-end-to-end-project-example
- llmz-m07-evaluating-rag
- llmz-m07-evaluating-retrieval
- llmz-m07-interface-and-ingestion-pipeline
- llmz-m07-monitoring-and-containerization
- llmz-m07-summary-and-closing-remarks