Skip to content

Latest commit

 

History

History
33 lines (16 loc) · 1.42 KB

README.md

File metadata and controls

33 lines (16 loc) · 1.42 KB
Raw

🍌 Modified Banana Serverless GPU processing

This repo gives a basic framework for serving Stable Diffusion Photoreal Model in production using simple HTTP servers. It is capable of returning up to 3 images at 512x512 without going over the 16GB hard limit on Banana GPU's

Please note that this model can create Adult content. So be careful what you type.

Things I learn't

Pytorch, Deployment of API through simple HTTP, Docker, Serverless GPU's

Quickstart

If you want to customize beyond the prebuilt model:

Follow the quickstart guide in Banana's documentation to use this repo.

(choose "GitHub Repository" deployment method)

Additional Steps (outside of quickstart guide)

  1. Create your own private repo and copy the files from this template repo into it. You'll want a private repo so that your huggingface keys are secure or place it as an ENV variable called "ENV HF_AUTH_TOKEN".
  2. Edit the dockerfile in your forked repo with ENV HF_AUTH_TOKEN=your_auth_token

Helpful Links

Understand the 🍌 Serverless framework and functionality of each file within it.

Generalize this framework to deploy anything on Banana.