LlamaIndex
Talk to us

LlamaIndex May 14, 2024

LlamaIndex Newsletter 2024-05-14

Hello LlamaIndex Family! 🦙

Welcome to another thrilling weekly update from LlamaIndex! We're excited to share a variety of outstanding updates, guides, and tutorials with you. But first, we have an exciting announcement to make.

We are thrilled to announce a new course in collaboration with DeepLearningAI—Building Agentic RAG. In this course, you’ll learn how to build a research assistant that can reason over multiple documents and answer complex questions. You’ll also learn how to step through the execution of the agent and steer it with human feedback. Check it out and take your RAG skills to the next level!

🤩 The highlights:

  1. Day 0 Support for GPT-4o - Tweet
  2. Llama3 Cookbook - Tweet
  3. TypeScript Agent Guide.

✨ Feature Releases and Enhancements:

  1. We have introduced day 0 support for GPT-4o in both Python and TypeScript. Additionally, we've created demo notebooks (demo1 and demo2) to help you easily experiment with GPT-4o using LlamaIndex. Tweet.
  2. We have launched Llama3 cookbooks showcasing interesting use cases for Llama 3, from basic chat functionalities to advanced agent development. Ideal for anyone building with local models, either on your laptop or through an API. Notebook, Tweet.

🗺️ Guides:

  • Guide to building agents in TypeScript: Dive into our comprehensive, open-source guide developed by Laurie that walks you through every step of agent development, from setting up with basic functions to integrating advanced features like local and remote LLMs, and data querying with vectorDB.
  • Guide to using RAG for content moderation: CloudRaft shows how to set up a RAG pipeline to moderate user-generated images effectively, ensuring compliance with predefined rules through techniques like semantic search and efficient inferencing with small LLMs.

✍️ Tutorials:

  • Kxsystems advanced workshop on "Building Advanced RAG over Complex PDFs with LlamaParse" to demonstrate how LlamaParse can tackle the challenge of extracting diverse elements like text, tables, images, and graphs from complex research papers. Video Tutorial, BlogPost, Notebook.
  • Arslan Shahid tutorial on Generating PowerPoints with Llama 3, using LlamaIndex to create a Llama3 RAG pipeline. The approach not only answers questions but also generates PowerPoint slide decks by utilizing the python-pptx library to write code programmatically for slide creation.
  • Hanane Dupouy demonstrates Building a Financial Agent that can Perform Reflection. The approach helps in analyzing stock prices by implementing two types of reflection: CRITIC (tool use) and self-reflection (no tools).
  • zhaozhiming’s tutorial on evaluating RAG systems, utilizing evaluation libraries like TruLens, Ragas, UpTrain, and DeepEval to assess RAG systems using metrics such as faithfulness, relevance, and answer correctness.