oof, windows+nvidia only.

Did it need to be electron? Use Tauri or Wails for a 300mb -> 20mb reduction.

Gotta wonder where low-framerate rendering and FLF generation could outpace rendering alone. Probably a less impressive use-case thanks to Eevee.

vision_prompt.py is interesting. I probably won't be using this exact tool for anything, but I will go through and analyze what it's doing there to see if any of the techniques are applicable to the problems I've been dealing with in scientific data analysis...

I think the more obvious part is that people when they don't want to listen to stuff these days say it's AI written. That they actually can't tell the difference. If someone hates it or thinks it's stupid they say it's AI.

And then they'll drop everything overnight, just like Pornhub did, because Visa/Mastercard is gonna say "nein".

what a mess!

hope this works for them

Man I hate cuda

The anime being advertised is fake, but the trailer and production are real.

The description mentions it is fake and an April fools joke.

Enthusiast GGUF quantizations should take care of that. Let them cook

Should be interesting to play with when they reduce the memory requirements and get a ComfyUI workflow. For now, it doesn't look usable on non-rich-person local hardware unless I'm missing something. There are some videos on this site too:

https://github.com/GAIR-NLP/daVinci-MagiHuman/

It’s like green washing.

Google does it with Android and Chrome too.

What's the obsession with using the word 'open' when it never is? OpenAi, Open Frontier, etc.

Does Nvidia care who runs the LLMs? As long as they buy their GPUs it's all good for them so they could empower open models and ditch the locked down ones that companies try to put everywhere?

The answer is: It is.

Your behavior and tone in this thread has been incredibly hostile towards people simply for expressing ideas you didn't like or think was possible. If this is how you respond to people discussing these kinds of topics maybe you shouldn't participate in our communities.

The hostility is that you're questioning my answer already without even bothering to do the most basic of research, it's frankly whiny and annoying. To answer you again though. No. It's not possible you're installing any of this hardware anywhere second hand. As I originally stated. Anything else is copeium.

I don't really understand the hostility. I just asked you a question and rather than an explanation you answer with this and downvote. I guess that was my mistake for giving you the benefit of the doubt.

Alright, sure buddy. You've got a 14.3Kw hookup (per 2U sever blade) and a glycol loop at your small business... These data centers operate at scales beyond the level of normal people or businesses. It'll end up ewaste or I'll eat my shoe, I've seen it time and time again with university super clusters and these are no different.

No one said it had to go in someone's home. Would it not be peossible if they were able to be acquired second hand, to find use by someone or someones that weren't the original buyer(s)?

If you think you can do it bro, but I've been in the industry long enough to know where this shit goes and it ain't to second hand use. You're ignoring so many aspects beyond just the firmware, I mentioned the infrastructure as well. A single Blackwell 2u draws 14.3Kw you're not installing that in your fuckin' home and that's before we get to the mandated dedicated water lines or the fact you're probably not cracking advanced firmware locks like that. It's frankly a ridiculous proposition that reeks of a lack of knowledge on the subject.

Do we know of it's compatible with ltx2 Loras?

Seems quite complicated and the example shown seems a bit cherry picked. Anyone tried this?