nlev retweeted
If you put your heart into eval, training is easy. If you don’t have one, training is aimless.
nlev retweeted
"from a global allocation perspective" is exactly what's wrong with these type people. Next they will take your car; it's only used 5% of the time. Then your XBOX. Then your wife.
I love having powerful computers that sit there and wait for me. That's what it means to be rich.
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
nlev retweeted
it took me building a shitty version of it to understand why pi durable is great. in essence it’s “just” way to serialize your session. that sounds straightforward until you try to do it yourself and realize there are tons of choices that if you get a single one wrong, you lock yourself out of building anything usable.
nlev retweeted
“Iron rusts from disuse; stagnant water loses its purity and in cold weather becomes frozen; even so does inaction sap the vigour of the mind”
— Leonardo da Vinci
