We have no idea what the uptime is, we’re not conscious when it goes down. For all you know it could be 1%

The reason you haven’t heard of it is it used to be a unknown pixel editor and then then dev went on crack and decided to turn it into a “universal 2d graphics editor”

And well … It worked so I’m not criticizing.

Supporting apps out of sheer spite for gimp is certainly one business model

Graphite is also super interesting, but every time I’ve used the raster (brush) tool, it’s been unusably laggy. Idk if that’s the web stack or something else, but if they can improve it, I’d def want to try again

Oh sick I was looking for something like this. Hopefully can replace wisprflow

It’s so annoying, because both are technically grammatically correct, but the current one just sounds the opposite

An unattractive anime girl? Is that even possible?

Yeah we do, they list it in privacy policies. Many of these they can’t really collect even if they wanted to

Note this is if you use their apps. Not the api. Not through another app.

OCR was AI.

Anyway today’s models are measurably better especially when you go beyond simple text on a clean page.

Where does the “learn more” go to?

how the hell are they still in business lmao

No

The main error is that Linux is not strictly speaking part of the GNU system—whose kernel is GNU Hurd. The version with Linux, we call “GNU/Linux.” It is OK to call it “GNU” when you want to be really short, but it is better to call it “GNU/Linux” so as to give Torvalds some credit.

We don’t use the term “corelibs,” and I am not sure what that would mean, but GNU is much more than the specific packages we developed for it. I set out in 1983 to develop an operating system, calling it GNU, and that job required developing whichever important packages we could not find elsewhere.

Of the US’s own capabilities?

As well as banning research. Absurd overreach of government and it will accomplish the opposite of what it wants.

No cause I was already running regular (non-deepseek) qwen 14B, admittedly a heavily quantized and uncensored version, so I was just curious if it would be any better

I think you’re confusing the two. I’m talking about the regular qwen before it was finetuned by deep seek, not the regular deepseek

Have you compared it with the regular qwen? It was also very good

Might be the difference between FHE and regular HE. I don’t know a lot about this subject, but if HE was more practical, I’d expect to see it a lot more, outside of ML too.