Skip to Content
View all events

Hands-on Retrieval Augmented Generation (RAG)

Published by O'Reilly Media, Inc.

Intermediate content levelIntermediate

Implement RAG with LlamaIndex

Course Outcomes

  • Demonstrate how to create an end-to-end RAG solution
  • Explain some of the limitations of LLMs
  • Demonstrate how to implement RAG with LlamaIndex and GPT-4-Turbo Assistants

Retrieval Augmented Generation (RAG) is one of the most common business applications for Large Language Models. RAG enables the querying of internal documents as knowledge bases. In this course, participants will learn how to query their internal documents and understand how RAG works under the hood. Learners will work with embeddings and Vector databases, explore what use cases RAG is good for, and learn how to build RAG-based solutions using LlamaIndex. They can take lessons learned and apply them on the job, making knowledge-bases available to internal teams.

What you’ll learn and how you can apply it

  • Understanding and work with embeddings and Vector databases
  • Learn use cases for RAG
  • Build solutions using LlamaIndex

This live event is for you because...

  • You’re a developer, data scientist or machine learning engineer
  • You work with ChatGPT or other LLMs and want to develop solutions for working with bespoke data
  • You are a developer and want to build solutions with LLMs

Prerequisites

  • Experienced with Python (intermediate)
  • Basic understanding of ML and LLMs
  • OpenAI and Cohere account and API key

Course Preparation

  • Have an OpenAI API key and login (to complete the exercises)

Recommended follow-up:

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Module 1 - Embeddings (60 minutes)

  • Presentation: Limitations of LLMs
  • Presentation: Embeddings
  • Exercise: Introduction to Embeddings
  • Pulse Check or Poll: Embeddings
  • Break

Module 2 - Retrieval Augmented Generation (60 minutes)

  • Presentation: Using external sources of data
  • Presentation: Practical considerations when working with RAG
  • Exercise: Working with vector databases
  • Exercise: Introduction to RAG and LlamaIndex
  • Exercise: Working with RAG and FAISS
  • Q&A
  • Break

Module 3 - More Advanced Considerations for RAG (60 minutes)

  • Presentation: Improving Query Processing
  • Presentation: Post-retrieval processing using Retrievers and Query Engines
  • Exercise: (mini project) Working with a knowledge-base using LlamaIndex
  • Exercise: Bi-Encoders and Cross-encoders for post-retrieval
  • Q&A

Module 4 - Using OpenAI Assistant retrieval and agents (60 minutes)

  • Presentation: Retrieval using the playground
  • Exercise: Retrieval using the OpenAI API
  • Presentation: Why RAG-based solutions are still relevant despite OpenAI Assistant
  • Exercise: Retrieval project using OpenAI Assistants
  • Presentation: The future of RAG - Agents

Your Instructor

  • Jonathan Fernandes

    Jonathan Fernandes works with large language models every day, and he did so long before ChatGPT came on the scene. He focuses primarily on LLMs in production to solve business problems across a variety of domains including finance, health, manufacturing, legal, and education. Previously, he worked for LLM-provider startups including Cohere. Jonathan has taught over 250,000 people about LLMs and how to use them.

Skills covered

  • Retrieval Augmented Generation (RAG)
  • Large Language Models (LLMs)