/min-dalle-api

Simple python API for AI image generation using min-dalle and FastAPI

Primary LanguagePythonMIT LicenseMIT

min-dalle-api

This is a simple python API for AI image generation using min-dalle and FastAPI. This API uses the dalle-mini model by default. You can use the dalle-mega model by changing the isMega = False line at the top of the main.py file to isMega = True

Hardware Requirements

You do not need a GPU to run the dalle-mini model, although it is much faster with one. It takes about ~100 seconds to generate a 2x2 grid running on my Ryzen 5600X. To run on a GPU you will likely need a minimum of 4GB VRAM for the mini model and 12GB VRAM for the mega model.

Install

pip install -r requirements.txt

Usage

Navigate to the "app" directory : cd app

Start the server: uvicorn main:app --reload

If everything goes as planned you will see an output in the console similar to this:

Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)

Your server is now running at that address. You can navigate to http://127.0.0.1:8000/docs to see and test the available endpoints.

image

image

Deployment

I have included a Dockerfile for easy deployment.

NOTE: Must use a Windows container.