MoGA is a powerful question-answering system that utilizes multiple language models from Groq and a graph-based approach from Langchain's Langgraph to generate high-quality responses. It combines the strengths of Groqs models, performs web searches for up-to-date information, and uses an iterative refinement process to produce comprehensive answers. The replies do take a few minutes but there is alot going on under the hood and the content is next level compares to gpt4o content easily and its free :)

• Uses multiple LLMs as proposer models (llama3-8b-8192, gemma2-9b-it, mixtral-8x7b-32768)Groq Models
• Aggregates responses using llama3-70b-8192
• Performs web searches for current information
• Iterative refinement process with reflection
• Rate limiting to manage API usage

1. Clone the reoo cd moga
2. Create a virtual environment:python -m venv venv source venv/bin/activate

3. Install the required packages: pip install -r requirements.txt

4. 4. Set up your environment variables: Create a .env file in the project root and add your API keys:

OPENAI_API_KEY=your_openai_api_key GROQ_API_KEY=your_groq_api_key ANTHROPIC_API_KEY=your_anthropic_api_key TAVILY_API_KEY=your_tavily_api_key

Run the main script: python moga.py

Follow the prompts to enter your query and specify the maximum number of iterations.

This project is licensed under the MIT License - see the [LICENSE] file for details.

ANY AND ALL CONTRIBUTIONS OR IMPROVEMENTS WELCOME