Skip to content

Latest commit

 

History

History
98 lines (61 loc) · 3.83 KB

README.md

File metadata and controls

98 lines (61 loc) · 3.83 KB

Serverless RAG

Livestream starting soon! Click the image below to watch the recording.

Reactor Livestream

Overview

In this quest, you will set up and run a Serverless Retrieval-Augmented Generation (RAG) support system using the provided codebase. Once you have completed the setup, you will start a CodeTour that will guide you through each step of the RAG implementation with detailed explanations.

Serverless RAG - CodeTour

Steps to Complete the Quest

Codebase Setup

  1. Fork and Clone the Repository: Start by forking the Serverless RAG with LangChain repository to your GitHub account and then clone it.

    git clone https://github.com/<your-username>/serverless-chat-langchainjs.git

    Navigate to the project directory:

    cd serverless-chat-langchainjs

  2. Download Ollama (if you haven't already): You won't need to deploy to Azure for this quest, but you will need to connect to local models for text completions and embeddings.

    [!NOTE]
    Foundry Local doesn't support embeddings models yet, so you'll need to use Ollama for this quest.

    Download and install Ollama from ollama.com.

  3. Pull the Required Models: Open your terminal and run the following commands to pull the necessary models for text completions and embeddings:

    ollama pull llama3.1:latest
ollama pull nomic-embed-text:latest

  4. Install Dependencies: Install the required project dependencies using:

    npm install

  5. Start the Application:

    Launch the application with:

    npm start

    Then, in a separate terminal, run the following command to upload the PDF documents from the /data folder to the API:

    npm run upload:docs

    Interact with the application by asking questions related to the uploaded documents and observe how the RAG system retrieves and generates responses based on the content.

    [!NOTE]
    While local models usually work well enough to answer the questions, sometimes they may not be able to perfectly follow the advanced formatting instructions for the citations and follow-up questions.

    This is expected, and a limitation of using smaller local models.

Start the CodeTour

This quest is designed to give you a guided tour of the codebase and its implementation of the complete RAG pipeline. To start the CodeTour:

  1. Install the CodeTour Extension: If you haven't already, install the CodeTour extension in Visual Studio Code.

  2. Open the CodeTour: Open the Command Palette (Ctrl+Shift+P or Cmd+Shift+P on Mac) and type "CodeTour: Start Tour".

    Start CodeTour

There are 6 tours to walk you through the entire RAG implementation flow. We recommend going through them in order as they build upon each other.

  • Tour 1: RAG Architecture (10 steps)
  • Tour 2: Document Ingestion (10 steps)
  • Tour 3: Vector Storage (8 steps)
  • Tour 4: Query & Retrieval (7 steps)
  • Tour 5: Response Generation (7 steps)
  • Tour 6: Streaming & Chat History (8 steps)

Return to the Build-a-thon

Once you have completed the CodeTour and explored the RAG implementation, return to the main Build-a-thon repository to continue with the next quests.

Stay connected

Have a question, project, or insight to share? Join the Build-a-thon Discord channel

AI Note

This quest was partially created with the help of AI. The author reviewed and revised the content to ensure accuracy and quality.