Multi-Agent LLM Framework for a research assistant for PhD scholars
This repository includes implementation of a multi-agent based LLM framework for assisting researchers in writing well-structured and coherent sections of their research on a topic of their choosing. We have used crewai python library to build the agentic workflow, which uses Azure-OpenAI in the backend, but can be re-configured to use other LLMs as well with minor tweaks to the original codebase.
Make sure you have python>=3.10 installed on your kernel. First, install the dependencies -
pip install -r requirements.txt
Add the required environment variables for LLM in the .env in root -
- AZURE_OPENAI_DEPLOYMENT
- AZURE_OPENAI_VERSION
- OPENAI_API_KEY
- OPENAI_ENDPOINT
Then, in main.py, change the research_topic and research_paper_section as per your requirement, then run main.py -
python main.py
Once the agent execution is completed (usually takes ~1 min), the results are populated in the results.md file.