Overview
Explore advanced Retrieval Augmented Generation (RAG) techniques in this 26-minute technical video that delves into solutions for improving RAG performance. Learn about self-query mechanisms, parent document retrieval strategies, and Hypothetical Document Embedding (HyDE) implementation. Follow along with hands-on labs in Langchain, complete with downloadable mindmaps and Colab notebooks for practical implementation. Master text segmentation techniques, understand optimal chunk sizing, and discover how to train sentence similarity embedding models. Access comprehensive resources including research papers on zero-shot dense retrieval and RAG surveys while working through practical examples that demonstrate each concept's real-world application.
Syllabus
- Introduction
- Training Sentence Similarity Embedding Models
- Self-Query Retriever
- Text Segmenetation
- Size of Chunks
- Parent Document Retriever
- Hypothetical Document Embeddings HyDE
- Lab 1 - Self-Query Retriever in Langchain
- Lab 2 - Parent Document Retriever in Langchain
- Lab 3 - HyDE in Langchain
Taught by
Donato Capitella