Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/carlotta94c/exploring-rag


https://github.com/carlotta94c/exploring-rag

Last synced: 7 days ago
JSON representation

Awesome Lists containing this project

README 

        
# Exploring RAG pattern for LLMs



[![Open in GitHub Codespaces](https://img.shields.io/static/v1?style=for-the-badge&label=GitHub+Codespaces&message=Open&color=brightgreen&logo=github)](https://github.com/codespaces/new?hide_repo_select=true&ref=main&repo=816701626&skip_quickstart=true)



**Retrieval Augmented Generation (RAG)** is the pattern or architectural design enabling data management of custom business data in a db, real-time data retrieval to efficiently extract information from the db which are relevant to the user query, and LLM prompt augmentation with the search result.



The way it works is:

- The user inputs an initial prompt, which is used to search a data source and **retrieve** relevant grounding data.

- The grounding data is used to **augment** the prompt by adding additional contextualization and scoping text.

- The augmented prompt is submitted to the LLM, which can then **generate** an appropriate response in natural language.



![RAG pattern visualization](RAG_pattern_visualization.png)



## About this repo

This repository contains a collection of notebooks and scripts that break down the RAG pattern for LLMs. It is organized as follows:

1. [0-create-search-index.ipynb](./0-create-search-index.ipynb): this notebook demonstrates how to *create a search index* with **Azure AI Search** using an hybrid search approach + semantic ranking. It populates the index with a sample dataset containing a product catalog ([products.csv](./data/products.csv)) of a fictional retail company called Contoso.

2. [1-retrieve-documentation.ipynb](./1-retrieve-documentation.ipynb): this notebook demonstrates how to *retrieve relevant documentation* from the Azure AI search index using a user query. To do so, it first converts the query into an embedding, using Azure OpenAI text-embedding-ada-002 model.

3. [2-augment-prompt.prompty](./2-augment-prompt.prompty): this script demonstrates how to *build a dynamic prompt specification* with **Prompty** and inform it with the retrieved documentation. The Prompty file can also be executed to *generate the final answer* to the augmented prompt.



## Pre-requisites



To get started with this sample, you will need:

* [Azure Subscription](https://azure.microsoft.com/free/) - sign up for a free account.

* [GitHub Account](https://github.com/signup) - sign up for a free account.

* [Access to Azure OpenAI](https://learn.microsoft.com/legal/cognitive-services/openai/limited-access) - submit form to request access.



Also, you'll need to provision the following Azure resources:

* An [Azure AI Search](https://docs.microsoft.com/azure/search/search-create-service-portal?WT.mc_id=academic-141259-cacaste) service.

* An [Azure OpenAI Service](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/create-resource?pivots=web-portal&WT.mc_id=academic-141259-cacaste) resource and two model deployments:

    - name: `embedding-ada`, type: `text-embedding-ada-002`, version: `default`

    - name: `gpt-4-turbo`, type: `gpt-4`, version: `turbo-2024-04-09`



## Run in GitHub Codespaces

We recommend using GitHub Codespaces for the fastest start with least effort. In fact, by creating a Codespace, you can run the notebooks and scripts in this repository without having to install anything on your local machine, since all the pre-requisites are specified in the dev container configuration file ([.devcontainer/devcontainer.json](./.devcontainer/devcontainer.json)).



Click the button to launch this repository in GitHub Codespaces.

  

[![Open in GitHub Codespaces](https://img.shields.io/static/v1?style=for-the-badge&label=GitHub+Codespaces&message=Open&color=brightgreen&logo=github)](https://github.com/codespaces/new?hide_repo_select=true&ref=main&repo=816701626&skip_quickstart=true)



 This opens a new browser tab with setup taking a few minutes to complete. Once ready, you should see a Visual Studio Code editor in your browser tab, with a terminal open. You can also open your Codespaces from within your Visual Studio Code desktop application.



 ## Useful resources

 - Explore an e-2-e production ready RAG workflow for the Contoso Retail dataset in the [contoso-chat sample](https://github.com/Azure-Samples/contoso-chat/tree/main).

 - Learn more about the RAG pattern with Azure AI Search in the [official documentation](https://docs.microsoft.com/azure/search/retrieval-augmented-generation-overview?WT.mc_id=academic-141259-cacaste) on Microsoft Learn.

 - Get started with Prompty by navigating the [official guide](https://aka.ms/prompty-doc).