fuck my timeline, y'all are very boring, I'm out for a bit.

I somehow missed this, but I think this is very cool.

Don't forget that you can present your recent JMLR paper at this upcoming ICML's Journal-to-Conference track!

icml.cc/public/Journal…

Deadline for submitting a request to present is May 3rd this week!

Human preference LLM arenas are poorly suited for evaluating ASCII art because the ASCII art that most impresses a human is often verbatim regurgitation of an existing human work and this is rarely true for text.

Votes on ASCII art should be detected and thrown out IMO.

hello is that the police, I'd like to report a murder

'The AI priest also told one user that it was okay to baptize a baby in Gatorade.'

Sounds like someone pushed a (buggy) last minute config change to prod before leaving for the weekend lol

A picture is worth a thousand words:

micrograd out

autodidax in

jax.readthedocs.io/en/latest/auto…

If you ever actually looked at these benchmarks, the model predictions, and what the claimed 'human performance' means, you would know.

Hint: it's unrelated to intelligence.

Looks like many people, especially more prominent ones, are commenting and opining blindly.

There's only one ultimate answer.

It's *Vi*sion *T*transformer, so like Fit with V.

V.I.T. just does not roll off the tongue nice enough.

When I voted/posted, it was at a shocking 20/80! Good job almost turning it alone. So, ... what election are we going to meddle with next, team?

Scatter plot with top-left good and yolo axes

is the new

radar plots where ours surrounds everything.

PS: the larger model is even better lol