Configuration of public endpoints?
Are the configurations or templates for the public endpoints available? I'm trying to reproduce
https://console.runpod.io/hub/playground/video/wan-2-2-t2v-720 using serverless but my outputs take longer/cost more/look worse than what I get from the public endpoint. I'd like to see how the endpoint is implemented.
3 Replies
This information is private, but if you need anything in particular let me know and I can get you someone who can help a little more with these.
my main questions would be what the diffusion steps schedule is, what gpu it's running on, and which software they're using (comfy-api? something else?)
it has optimizations that are not available publicly, e.g. optimizing kernel and model caching