bbstats

this is incredible.

Quick test on Dall-E 3 (bing.com/create) vs ideogram, and ideogram is not even close in terms of quality.

someone doesn't have kids...

None of this is that bad.

mean crypto utility is basically zero. mean AI utility is decent.

Very first nerf I ever saw ran fine on consumer hw at 60fps in VR. And gaussian splatting for polycam is on their website now (was frontpage of HN yesterday too)

So dramatic lol

Pretty useful if you get a positive though

because it's incredibly easy to use, and numpy and pandas are written in python.

past != present

OTOH true intelligence can still poorly compress things.

I would say compression has some congruency with intelligence.

I read it that way! :)

The point is that it's not pre-existing Linux users.

I...do not prefer this syntax.

GPT5?

Probably dog / chimp, no?

If you don't have a lot of money this is true. I think the opposite happens if you are rich.

We have some where I live (Flytrex in Durham). It's a pretty fun and occasionally useful novelty but - FAA regulations mean a driver has to come watch it deliver the food so - not quite ready for prime time.

These are not problems.

This remains to be true of the Star Wars trilogy...

I can't work out my core with my body weight?

Yes it absolutely was.

From the article

"It’s absolutely ridiculous,” Tzanetos said. “This is a once-in-a-lifetime kind of thing.”

No

Seems fine still. I use Poe.

When you find the minima of your validation curve, rarely if ever is your test loss lower than your training loss. I don't think this necessarily means you're overfitting.

Yes

"it's a mouthpad" "you mean mousepad??" "no, MOUTHpad"

seems logical in that you can train on more poses/colors/shapes/angles/lighting conditions. but also we could assume that there's a more direct/elegant improvement than using a generative model?