LLM Foundations: Vector Databases for Caching and Retrieval Augmented Generation (RAG)

Introduction

• GenAI with vector databases
• Course coverage and prerequisites

1. Introduction to Vector Databases

• What is a vector?
• Vectorization in NLP
• Vector similarity search
• Vector databases
• Pros and cons of vector databases

2. Milvus Database Concepts

• Introduction to Milvus DB
• Milvus architecture
• Collections in Milvus
• Partitions in Milvus
• Indexes in Milvus
• Managing data in Milvus
• Query and search in Milvus
• Set up Milvus and exercise files

3. Milvus Database Operations

• Create a connection
• Create databases and users
• Create collections
• Insert data into Milvus
• Build an index
• Query scalar data
• Search vector fields
• Delete objects and entities

4. Vector DB for LLM Query Caching

• LLMs and caching
• Prompt caching workflow
• Set up the Milvus cache
• Inference process and caching
• Cache management

5. Introduction to Retrieval Augmented Generation (RAG)

• LLMs as a knowledge source
• Introduction to retrieval augmented generation
• RAG: Knowledge curation process
• RAG question-answering process
• Applications of RAG

6. Implementing RAG with Milvus

• Set up Milvus for RAG
• Prepare data for the knowledge base
• Populate the Milvus database
• Answer questions with RAG

7. Vector Databases Best Practices

• Choose a vector database
• Combine vector and scalar data
• Distance measure considerations
• Tune vector DB performance

Conclusion

• Continue with LLMs