README.md

Japanese RAG

Japanese RAG is a small RAG (Retrieval-Augmented Generation) chatbot that is able to answer questions about the Japanese language. The chatbot is based on the Weaviate knowledge graph, the OpenAI API and Streamlit.

Installation

First, clone the repository and install the dependencies with the following commands:

poetry install

Setup

1. Create a secrets.toml file
2. Create a new Weaviate Cluster
3. Create a Weaviate API Key
4. Create an OpenAI API Key

Finally, add the following credentials to the secrets.toml file:

OPENAI_API_KEY = 'xxx'
WEAVIATE_URL = 'https://xxx'
WEAVIATE_API_KEY = 'xxx'

Run

After that the installation and setup, you can run the Streamlit app with the following command:

poetry run streamlit run app.py

Import Data

You can upload your data as ZIP-archive (containing individual text-files, example data) under the Data tab.