LegalRAG

1. WorkFlow

• Data Preprocessing done behind the scene, stored in VectorStore
• Load data from vector store in src/utils.py
• Create prompt template in src/prompt.py
• Call the OpenAI model, load all params into the chain in src/chat.py. Please paste your own LLM endpoint api key here.

2. Structure

├── data
├── src
│ ├── chat.py │ ├── prompt.py
│ └── utils.py
├── VectorStore
├── config.py
├── main.py
├── requirements.txt
└── README.md

3. Notes

• Vector DB: Chroma DB
• Embedding model: Cohere embed-multilingual-v3.0
• LLM: Cohere Command R+

4. How to Run

• pip install -r requirements.txt
• Run uvicorn main:app --host 0.0.0.0 --port 80
• Get to localhost:80/docs
• Ask your question and wait for the answer