So... a classifier model?

"Arm-designed CPU core matches Apple-designed CPU core from a year ago"

It could probably go quite a bit faster using Arm SME. The entire network could fit into the ZA register.

New York State is funding Plattsburgh as an electric aviation R&D hub, and actually gave BETA $20M in a grant.

Ok that's...just cheating. You can't take a benchmark like MMLU designed to test the performance of a single general language model and compare it to performance of a small specialized model designed to do well on MMLU.

It wasn't designed to do well on MMMLU, it's a general model designed for deterministic task like OCR, object detection, STT and more and a by product of that is great language abilities. It still has a transformer backbone giving great language skills while being good at other stuff.

See the full benchmark: https://interfaze.ai/leaderboards

An HTML5 super jet?

Gravity powered!

consider using fp16 or bf16 for the matrix math (in SME you can use svmopa_za16_f16_m or svmopa_za16_bf16_m)

Do you use fully bidirectional attention or is it at all causal?

The oldest iBook G4 is from October 2003, not even 23 years old.

I was gonna say. I do have an ibook G3 from 1999, but it can't use modern Wi-Fi (though if it's running the right OS, it could maybe update--i think I've done updates as far back as 10.3?) the AirPort card can talk to 802.11b over WEP - i have an old router set up with a MAC whitelist specifically so the ibook can get online still

Yep, in 2000 even the PowerBooks weren't G4 yet. And of course the consumer-grade iBooks line wouldn't get G4s until years after that.