@ZzMatuSs Pan sa snazi byt nad vecou. Moc mu to nejde.

@luciahudackova @gl_sk Uplne v pohode. A ovela radsej by som mal za plotom atomku ako veternik.

@Lewysko

@lucasarruda @dvassallo Price.

@luciahudackova @gl_sk Ono zase jadrový odpad je prakticky neškodný (ak je dobré uskladnený). (To nič nemení na tom, že Taraba je jebinko)

@luciahudackova Mam pocit, ze historicky budeme dnesne obdobie nazyvat "pomsta kokotov".

@tomas_izak @Lewysko Napisal, ze fintu povie na svojich seminaroch, kedze vznikol nejaky pozmenovak, ktory originalnu fintu skomplikoval.

@jsuchal Toto je dosť blbosť.

RT @giffmana: I took a brief look at the Harmonic Loss paper tl;dr: instead of dot-product with softmax, do euclid dist with normalized 1/…

@TheXeophon @giffmana Pay is one thing. But the amount of academia red tape is unbelievable.

Was lazy, asked o3-mini-high to write me a RealNVP implementation (I could write it in like a 30 minutes). Of course, it contained a bug, which took like two hours to find.

@joe_trendy @mitosomii Maju Lakers vobec centra?

@Lewysko @joe_trendy Zajtra vymienia Jokica za LeBrona? xD

@Lewysko A naco je Dallasu Davis + Lively + Gafford (este k tomu aj nepouzitelny Powell)?

RT @ThePrimeagen:

@jxmnop You know that gender eyeballs prediction works due to mascara?

@ostrisai It needs good tooling. AMD is half the price of Nvidia and nobody uses it because shitty drivers.

@Leucoium_vernum @tginart @cloneofsimo During first "epoch" you will sample apx. 70% of data points.