#Stable Diffusion GPU Pod and API
junior cometBOT
naive garnet
Stable Diffusion GPU Pod and API
glad coral
You can open an API on a GPU Pod, but that would keep your API running 24/7
is that what you want?
its a pretty expensive thing
youg enerally want to have a serverless API
You want to pay for it 24/7 too? vs not paying for down time?
Ok
You essentially want this I believe
Which is if you have a backend running locally on the pod at like port: 5000, then you want to expose the http port
so that you can hit it on the proxy url
Ive never tried it myself, but ive seen other people do it
Im not too sure, unfortunately not familiar with the template you are talking about. Your best bet is if it has a github that it should have a documentation on it
Do you want to link to the template? is this a runpod template? a public community one? so oin
which one?
well im pretty sure that the a111y, from my knowledge, again not familiar with this, starts up a gradio link / web ui? for it?
and u do connect to the web ui im guessing? when it starts up?
a111y though doesn't really have a backing api on it, unless u have a repository that just does that
more of a limitation of a111y itself
https://github.com/ashleykleynhans/runpod-worker-a1111
Ashelyk has a pretty detailed community repo on an a111y serverless worker, not for GPU Pod, but might fit your need still
if you want you can just keep the minimum active worker to 1 on your serverless
and it woudl essentially be like a 24/7 running gpu pod
Got it
Ill let the others answer if htey have a lead, but I think your best bet is not doing a111, tbh. Its more built as a webui, and I think is better to potentially look into a route like:
https://github.com/huggingface/diffusers?tab=readme-ov-file
https://github.com/deep-floyd/IF
Where they have progamattic access more built in, cause then all you have to do is start a backend server to connect an http too, to trigger your code
