Chat with your database using Gemini Tutorial

This repo is for using Gemini to chat with your SQL database. Inspired by https://github.com/leodeveloper/google-gemini-chat-with-sqlserver/

Objective

• To demonstrate on how to build a chatbot using Gemini and RAG technique to query your database

In this instruction, I use pyenv for the virtual environment and python 3.9.17. It actually works on python >= 3.9

Prepare Environment

git clone https://github.com/ardyadipta/gemini_chatbot_sql
cd gemini_chatbot_sql
pyenv virtualenv 3.9.17 gemini_chatbot
pyenv activate gemini_chatbot
pip install -r requirements.txt

Create API Key

• visit https://aistudio.google.com/prompts/new_chat?pli=1

    export GOOGLE_API_KEY=`<your Google API key`>
    echo $GOOGLE_API_KEY
  

How to use this tutorial

Setup your MySQL server on local

1. I assume you already have a MySQL server on locale, if not, download from here and follow the instruction to install.
2. Download the dataset from Kaggle here. In my case, I saved it into /tmp/ so that it can be accessible to the MySQL server on my local.
3. Upload CSV to MySQL table

python upload_csv_to_sql_table.py

Run app without RAG to see how it performs

streamlit run app.py

Run app with RAG

streamlit run app_with_rag.py