RAGO: a GO-based RAG project

🦙 Base Model

RAGO leverages the Ollama API to enhance retrieval-augmented generation capabilities. Ensure you have Ollama installed. Pull the chat and embed models of your choice, then update the model/modelConfig.json file accordingly.

🏠 Vector Store

RAGO requires a vector store to save and index document embeddings. You have two options depending on your usage needs:

• MatrixOne: Recommended for scenarios where documents are accessed repeatedly. Installation of MatrixOne is required.
• FAISS: Suitable for one-time usage. Run the FAISS service with python faiss/serve.py.

📜 Document Preparation

Place the documents you intend to retrieve in the docs directory. Currently, RAGO only supports documents in .txt format.

✨ Running RAGO

Execute RAGO using the following command, specifying the file and repository as required:

cd cmd
go run main.go --file pride-and-prejudice.txt --repo MatrixOne