AutoMerging RAG

• It is a type of rag system that tries to solve the struggle of providing to-the-point context to llm. It works by defining a hierarchy of smaller chunks linked to parent chunks. If the set of smaller chunks linking to a parent chunk exceeds some threshold, then it merges smaller chunks to the bigger parent chunk. for more auto-merge.pdf

• The dataset used is ECAN and PatternMiner Paper of opencog.

features

• Flask Backend: Handles API requests, merges logic, and serves retrieved content.
• Next.js Frontend: interface to provide the query to the LLM.
• LlamaIndex: is a Python framework designed to work with large language models (LLMs).

Tech Stack

• Flask (Python)
• Next.js (React)
• TypeScript
• Python

Video Demo video

Installation

1. Clone the repository:

   git clone https://github.com/beki-kel/AutoMerging_RAG.git
   cd AutoMerging_RAG

2. Set up the Flask backend:

      cd back
      pip install -r requirements.txt

3. Start the backend server:

       flask run

4. Add environment variables:

       GEMINI_API_KEY = "YOUR API KEY"           

5. Set up the Next.js frontend:

       cd ../front
       npm install

6. Start the frontend development server:

       npm run dev
   

colab link

link