Mario Klingemann

Trying to grow to a 256 model I am wondering what the deliberations are when adding channels in order to grow the size. Looking at the progression for 32 to...

A little tip for those who are not interested in the evaluation data: the arg --evaluate-every -1 skips that step and saves some time. In addition to that it is...

The txt2img script is already using the WatermarkEncoder from the imwatermark libary to do exactly that.

Okay - I am facing a problem which might be related to this issue - for some reason I suddenly cannot bind() anymore (might have been that the app had...

I ran into the same issue - the fix that worked for me was to recompile fftw3 from source and enable shared libraries: ``` wget http://www.fftw.org/fftw-3.3.6-pl2.tar.gz tar -xzf fftw-3.3.6-pl2.tar.gz cd...

Ah I just found this issue on the LLaVA repository and it looks like this would require a different training approach: https://github.com/haotian-liu/LLaVA/issues/197

Ideally it would be capable of any kind of comparison, but for a start it would already be nice if it was able to point out things like "object A...

Same problem as https://github.com/NVIDIA/pix2pixHD/issues/46 - it's a problem that haunts me since years - also pix2pix has it. I still have not found the cause, but once a model has...

Ah well, that project is very old and I had only published it back them as javascript code on my website (and since i switched servers at some point that...

I am not sure how meaningful that semantic quality measure is - I guess it depends on what one is looking for in that latent space. If I am not...