That they can’t subjugate them

A desktop cpu and 20gb of ram.

You don’t need a license to look at stuff.

It would need to load every part of the model from disk into ram for every token it generates. This would take ages.

What you can do, however, is quantize the model. If you, for example, quantize a 16-bit model into 4-bit, its storage and ram requirements will go down to 1/4. While the calculations will still be in 16-bit, the weights will lose some accuracy.

To steal data, it needs to be deleted off the target’s computer. DMCA is stealing. Copying is not.

GPL was made to destroy the oppressive bourgeois copyright system from the inside, not to expand copyright even further.

If the corpos ignore the gpl, they delegitimize their own copyright system.

Its already settled. AI training is considered “fair use”, therefore compliant with copyright.

Still, death to copyright.

Fucking libertarians calling everything they don’t like authoritarian

For me its just the window key, with a window on it for the window manager

Distance china-arctic (far) > distance china-ocean (0)

It seems like I was wrong and they do need more data. But I think they have every right to go into their enemy’s imperialism tool and disrupt it however they see fit.

Prove it

Whoa its already released

Ever heard of reusing data? Its not the AI wildwest anymore. Scraping random data gives low quality (try SD1.5 to see what I mean). Good models need high-quality datasets.

So you said the same as me, but framed like a cia stenographer would. Why tho? You gain something from convincing people that china bad?

The only reason I can see for china to invade china 2 is if usa places nukes there

School? Prison?

Capitalism: They’re the same.

Are schools supposed to be stagnant, with children being transported to the 15th century as soon as they turn 6?

Depends on the programming

Random person dies. Copyright crusaders: AI! Euros: Putin! Yanks: Communism!

Its USA’s fault. Always. They’re responsible for 99% of oppression in the world.