Live data from Hacker News

Releasing weights for FLUX.1 Krea

krea.ai

81–90 of 112 posts

Re: Releasing weights for FLUX.1 Krea

#82
post #6
post #5

Earlier quoted context omitted.

It's because https://www.krea.ai/blog/flux-krea-open-source-release contains this: Our software follows canonical urls when it finds them. I've fixed the link above now (and rolled back the clock on the submission, to make up for lost time) but you might want to fix this for future pages.

OMG. Thank you! I had to setup a CDN-level redirect and I was so confused as to why when I asked others to help, their submissions were flagged as [dupe] or [dead] Thank you so much! I knew that HN software was advanced, but I didn’t know you guys used Canonical URLs like Google does. Smart and thanks for helping us with this slip!!!

Oh you're welcome - it does lead to a lot of not-obvious problems like this but I think it's worth it overall. It helps with duplicate detection, merging threads, and so on.

Re: Releasing weights for FLUX.1 Krea

#83

Nice release. Ran some preliminary tests using the 12b Txt2Img Krea model. Its biggest wins seems to be raw speed (and possibly realism) but perhaps unsurprisingly did not score any higher on the leaderboard for prompt adherence than the normal Flux.1D model. https://genai-showdown.specr.net On another note, there seem to be some indication that Wan 2.2+ future models might end up becoming significant players in the…

Can you point to a URL with the tests you’ve done?

Also, FWIW, this model focus was around aesthetics rather than strict prompt adherence. Not to excuse the bad samples, but to emphasize what was one of the research goals.

It’s a thorny trade-off, but an important one if one wants to get rid of what’s sometimes known as “the flux look”.

Re: Wan 2.2 I’ve also been reading of people commenting about using Wan 2.2 for base generation and Krea for the refiner pass which I thought was interesting.

Re: Releasing weights for FLUX.1 Krea

#84

yoo i'm also a researcher on the krea 1 project and happy to answer any questions :)

hello Erwann, great work! I have a very technical question just for you: how are you today?

hahahah i'm doing well tianpei good to hear from you!

Re: Releasing weights for FLUX.1 Krea

#86

How did you train while ensuring only images consensually acquired were used?

Likely the same way visual artists ensure that they only learn from images with permissive licenses.

Human learning and computer processing millions of works are different things. I don't think any human artists have seen as many images as the developers used for training.

Re: Releasing weights for FLUX.1 Krea

#87

Earlier quoted context omitted.

A parameter can be any size float. Lots of downloadable models are FP8 (8 bits per parameter), but it appears this model is FP16 (16 bits per parameter) Often, the training is done in FP16 then quantized down to FP8 or FP4 for distribution.

I think they are bfloat16, not FP16, but they are both 16bpw formats, so it doesn't make a size difference.

pardon the ignorance but it's the first time I've heard of bfloat16.

i asked chat for an explanation and it said bfloat has a higher range (like fp32) but less precision.

what does that mean for image generation and why was bfloat chosen over fp?

Re: Releasing weights for FLUX.1 Krea

#88

Earlier quoted context omitted.

Likely the same way visual artists ensure that they only learn from images with permissive licenses.

Human learning and computer processing millions of works are different things. I don't think any human artists have seen as many images as the developers used for training.

Then let he who hath not sinned cast the first stone.
Post reply on HN