Kevin Henner

@danielpanteleit This looks like it was addressed in https://github.com/garethr/garethr-docker/commit/4b88998a4b2ea5adec1a2a0243ea47a35a4accb3 but hasn't yet been released.

role -> What title best describes your role in your organization? = () System Administrator () Security Specialist () Systems Engineer () Network Operator () Network Engineer () Database Administrator...

I'm currently trying to do a bunch of runs at 512x512 with 100 steps to get representative samples from a bunch of artists. Getting good results across a bunch of...

Very interesting--just changing to plms and keeping everything the same it seems like it's pulling out more detail, but I'm left with a little grain at the end. Might try...

Also, @nikocraft, you might want to experiment with turning off the secondary model for smaller images to let the full diffusion model do its thing. At smaller image scales the...

Yeah, still not 100% familiar with the code base so take this all with a grain of salt until I figure it out a little better and do some testing,...

 Looks like the peak memory usage close to a simple linear function of cuts for each model, so it oughta be easy to predict!

https://github.com/lowfuel/progrockdiffusion/pull/71 got the bulk of the heavy lifting here done. Still need to do the following: - Set a trigger to display the predictions (even without DEBUG log level) when...

It looks like that "High contrast correct at step 110" lines up with that change in the progress output. I generally prefer to just turn "fix_brightness_contrast" to false. Usually if...

> You do this by putting "skip_stages": 1 or some other value in what you have for the the 64x64 config and setting "input_size": [32, 32]. Somehow I'm not understanding...