stable-diffusion
stable-diffusion copied to clipboard
k_heun, k_lms and other samplers
Hi! Thanks for the release
Are there plans on releasing the samplers like k_heun
and k_lms
in this GitHub?
k_lms
is the default sampler in the discord server, and it seems to produce better results for wider ranges of cfg scale.
implementations for those samplers can be found here: https://github.com/crowsonkb/k-diffusion
Had the same question. What's the easiest way to take the sampler implementation from Katherine Crowson's github and integrate it with txt2img.py?
I see:
from ldm.models.diffusion.plms import PLMSSampler
in txt2img and I see where that sampler is getting set later in the code, but I'm not sure how to add k_lms.
Also wondering about this.
Is there already a way to change the sampler?
Any news on this ?