PlotCraft

PlotCraft is a fine-tuned Qwen2.5-3B assistant that converts natural-language requests into executable Python visualization code. This repository contains the Gradio application, FastAPI service, Docker configuration, automated tests, and GitHub Actions deployment pipeline.

Try the live ZeroGPU demo

Generated code is displayed for review and can be rendered as a PNG inside an isolated browser worker. Generated code is never executed on the server.

1. Open the live demo.
2. Describe the Python visualization you want.
3. Select Generate Python Code.
4. Review the generated Python code.
5. Select Render Preview to display the chart in the browser.
6. Copy the code if you want to use it in your own project.

Example request:

Create a Matplotlib bar chart with categories A, B, C and values 2, 5, 3.

Create and activate a virtual environment in Git Bash:

python -m venv .venv
source .venv/Scripts/activate

Install the dependencies and start the application:

pip install -r requirements.txt
python app.py

Local execution uses the lightweight mock mode by default, so it does not download the production model.

Run the tests with:

python -m pytest

Build and start the container:

docker compose up --build

Open the interactive API documentation at:

http://localhost:8000/docs

Available endpoints:

• GET /health checks whether the service is running.
• POST /generate accepts a prompt and returns generated Python code.

Stop the container with:

docker compose down