SQL Query Generator: Natural Language to SQL Query

Query GPT is a free, open-source tool that lets anyone convert natural language questions into SQL queries for their own data tables, using local open-source AI models (like Phi-3) — no API keys, no usage limits, and no costs.

Features

• Upload your own data (CSV or Excel) - the data should just have 2 columns, column name and data type
• See your table’s schema and preview your data
• Ask questions in plain English (e.g. "Show top 5 customers by revenue")
• Get complete SQL queries instantly, generated by a local LLM (Phi-3)
• 100% private: nothing is sent to external servers

Tech Stack

• Streamlit for the web app
• llama-cpp-python for running local AI models
• Phi-3 Mini Instruct as the default language model
• pandas for data handling

Getting Started

1. Clone the Repository

git clone https://github.com/Yazhini-Krishnan/sqlQueryGenerator/blob/main/README.md
cd query-gpt

2. Setting up a Python virtual environment

python -m venv venv
# Activate the environment:
# On Windows:
venv\Scripts\activate
# On Mac/Linux:
source venv/bin/activate

3. Pip Upgrade

pip install --upgrade pip

4. Install Required Dependencies

pip install streamlit pandas llama-cpp-python huggingface_hub

5. Download Phi-3 Model

python phi_3_download.py

6. Update path in app.py

Update the model path in your app.py as needed:

MODEL_PATH = "/full/path/to/phi-3-mini-4k-instruct.Q4_K_M.gguf"

7. Run the app

streamlit run app.py