Back in 2023 when I was looking for work someone tried to hire me to fine tune transformers specifically for papers intended for academic journal submissions. It is not the first.

Idon’t know about that- I did my grad thesis in robotic perception pre-transformers on generating landscape restoration design plans from computer vision data - that was over half a decade ago now. It seemed very obvious that, per what brighter days, as labor costs go down things can be solved…

A specific example is Chinas renewables deployments. Looking only at instantaneous measures fails to capture what is actually happening.

Nice

That is to say I wish Adam came to this video with more graphs.

I really liked brighter because the commentary came with graphs. Adam is a scientist and I like it when his commentary is emperical. I do ai/ml work professionally- so I get the optimism- but a lot of the appeal, for me, of the rethinkX brand is model driven not commentary driven optimism.

dude Pythagoras would be so ready and on-board to worship some vectors, his machine learning papers would be so amazing

I’ve always wondered why there haven’t been financial instruments constructed specifically designed to take short positions in coal plants. You’d have to be really well capitalized to do it- but those things just aren’t going to pay out their amortization schedules.

It really looks like the Chinese market will go this way - and also appears they are going to become the worlds automaker.

…there’s also just great statements about learning theory and optics from Brendan Fong; though those are less about biology and more about learning.

arxiv.org/abs/2101.10483

I think this paper qualifies as relevant- it’s about cybernetic systems more broadly - but I think it’s statements would hold for biological systems.

I really enjoyed reading it- my genuine thanks for writing it. The simplicity with which you presented the I=PAT abstraction really answers the leading question presented there.