

Why Your RAG System Is Broken, and How to Fix It with Jason Liu - #709
198 snips Nov 11, 2024
Jason Liu, a freelance AI consultant and creator of the Instructor library, dives deep into retrieval-augmented generation (RAG) systems. He shares key signs of RAG system failures and the tactical steps to diagnose issues. Liu emphasizes building robust test datasets and the importance of data-driven experimentation. He discusses fine-tuning strategies, chunking techniques, and collaboration tools, while also showcasing how future AI models could revolutionize the space. Finally, he highlights his passion for teaching through his AI consulting course.
AI Snips
Chapters
Transcript
Episode notes
Focus on Retrieval Quality
- Ban adjectives during stand-up to focus on quantifiable metrics.
- Shift focus from generation output to retrieval quality using precision and recall.
Fast and Cheap Evaluations
- Evaluating RAG systems with LLMs as judges shifts the focus from relevancy to prompting.
- Prioritize fast, cheap evaluations for rapid iteration and testing.
Building Simple Test Datasets
- Generate synthetic questions from text chunks to create a simple test dataset.
- Test if the generated questions retrieve the original chunks, giving engineers permission to trust their intuition.