abramdemski(Abram Demski)

PSA: Tag­ging is Awesome

[Question] What hap­pens to var­i­ance as neu­ral net­work train­ing is scaled? What does it im­ply about “lot­tery tick­ets”?

[Question] To what ex­tent are the scal­ing prop­er­ties of Trans­former net­works ex­cep­tional?

[Question] How should AI de­bate be judged?

What does it mean to ap­ply de­ci­sion the­ory?

[Question] How “hon­est” is GPT-3?

Noise on the Channel

Rad­i­cal Prob­a­bil­ism [Tran­script]

Bet­ting with Manda­tory Post-Mortem

Re­lat­ing HCH and Log­i­cal Induction

An Ortho­dox Case Against Utility Functions

Think­ing About Filtered Ev­i­dence Is (Very!) Hard

Bayesian Evolv­ing-to-Extinction

[Question] A ‘Prac­tice of Ra­tion­al­ity’ Se­quence?

[Question] In­stru­men­tal Oc­cam?

Be­com­ing Unusu­ally Truth-Oriented

The Credit As­sign­ment Problem

Defin­ing Myopia

Ran­dom Thoughts on Pre­dict-O-Matic

The Parable of Pre­dict-O-Matic