Feb 17, 2023
Requests are routed from Nginx through WireGuard to a Raspberry Pi 4B 8GB running FastAPI, and the Cartman model using PyTorch. It has enough RAM for more, but the CPU is pretty much at its limit. Expect it to take a few seconds, I'm cheap. Sorry(kinda).
You can download a Docker image if you'd like to run it on your own hardware for either x86_64 or aarch64.
More info here as well as example scripts to talk to the docker container.