https://github.com/rscr1/crag_project

Crag implementation
https://github.com/rscr1/crag_project

langgraph llamaindex python streamlit transformers

Last synced: 2 months ago
JSON representation

Crag implementation

• Host: GitHub
• URL: https://github.com/rscr1/crag_project
• Owner: rscr1
• Created: 2025-02-14T13:20:35.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-03-14T07:55:48.000Z (over 1 year ago)
• Last Synced: 2025-06-07T16:49:19.937Z (about 1 year ago)
• Topics: langgraph, llamaindex, python, streamlit, transformers
• Language: Python
• Homepage:
• Size: 678 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: ReadMe.md

Awesome Lists containing this project

README

          
# Knowledge Assistant

## Overview

The Knowledge Assistant is a web application designed to provide answers to user queries by leveraging a Retrieval-Augmented Generation (RAG) approach. This application utilizes various libraries and models to perform web searches, retrieve relevant documents, and generate responses based on the context provided.

## Example Outputs

Here are some examples of the application in action:

**In-domain (without web-search)**:

   ![alt text](/src/example1.png)

   ![alt text](/src/example2.png)

**Out-of-fomain (with web-search)**:

    ![alt text](/src/example3.png)

    ![alt text](/src/example4.png)

## Libraries Used

The following libraries are utilized in this project:

- **Streamlit**: For creating the web interface.

- **Torch**: For deep learning model operations.

- **Transformers**: For loading and using pre-trained models.

- **Ollama**: for inference LLM.

- **DuckDuckGo Search**: For performing web searches.

- **LangGraph**: For managing the workflow of the RAG process.

- **LlamaIndex**: For handling document embeddings and retrieval.

- **Tiktoken**: For tokenization of text inputs.

- **SQLite**: For storing chat history in a local database.

## Key Entities in RAG

The application implements several key entities in the RAG framework:

- **MyEmbeddings**: A custom embedding class that generates embeddings for queries and documents using a pre-trained model.

- **VectorStoreManager**: Manages the vector store, loading data, and creating embeddings for documents.

- **GraphState**: Represents the state of the RAG workflow, including the user's question, retrieved documents, and generated responses.

- **Workflow Graph**: A directed graph that defines the sequence of operations (nodes) and their connections (edges) in the RAG process.

## CRAG Method Implementation

The application implements the CRAG (Conditional Retrieval-Augmented Generation) method, which enhances the response generation process by conditionally deciding whether to perform a web search or generate a response based on the retrieved documents' relevance. The workflow consists of the following steps:

1. **Retrieve**: Fetch relevant documents from the vector store based on the user's query.

2. **Grade**: Evaluate the relevance of the retrieved documents using a reranker.

3. **Web Search**: If necessary, perform a web search to gather additional context.

4. **Generate**: Synthesize a final response using the available context from both retrieved documents and web search results.