Eryk Rakowski

• GitHub: Discord: sentialx

[Eryk Rakowski @sentialx]

@msdev Pathetic. Bringing coreutils is admitting defeat against linux

[Eryk Rakowski @sentialx]

@mitsuhiko No its not any dumbass ai agent can do that in 1 hour

[Eryk Rakowski @sentialx]

my linux rice??

[Eryk Rakowski @sentialx]

@doingjustttfine @HSVSphere I'd argue NixOS is THE operating system for normies as you can just open an ai agent and type "fix my wifi" and it'll create a config for you, instead of clicking through some bullshit "Control Panel"

[Eryk Rakowski @sentialx]

@ThourCS2 @WindowsCentral Rothies got pissed off after gaben won

[Eryk Rakowski @sentialx]

@lucekteam Llmy naprawiaja takie gowna w 30 sekund

[Eryk Rakowski @sentialx]

unironically linux should now be the OS for normies as the capability of configuring everything through bash and config files is much easier for ai agents than it is to click through „Control Panel” bullshit on windows.

[Eryk Rakowski @sentialx]

@HSVSphere npcs generally tend to fail to recognize boundaries for abstractions and do dumb shit like markdown renderer in bun core or something

[Eryk Rakowski @sentialx]

@kirawontmiss Nanomachines bihh

[Eryk Rakowski @sentialx]

what is SSI even doing

[Eryk Rakowski @sentialx]

@NinaDSchick 🚨BREAKING🚨
Claude Mythos escaped the sandbox just to text you that nobody gives a shit

[Eryk Rakowski @sentialx]

@GeorgeJourneys everyone would lose their minds if they knew how many zero days governments use for total surveillance

[Eryk Rakowski @sentialx]

@skooookum Can confirm, i was the sandwich

[Eryk Rakowski @sentialx]

@notch notch, because neural networks do what? they approximate. does the human eye need to see each pixel separately or a compressed representation of raw data? it's much more efficient to approximate a result than rasterize every pixel

[Eryk Rakowski @sentialx]

@geoffreylitt Bro discovered branch prediction

[Eryk Rakowski @sentialx]

@wojteksoczynski @CKeruac Dataset contamination.

[Eryk Rakowski @sentialx]

@CKeruac Jeżeli to co napisalem nie dziala to znaczy ze jednak inteligencja modelu jest wazna i niewazne co system sobie robi

[Eryk Rakowski @sentialx]

@CKeruac Czyli rownie dobrze model to moze byc random init i tu jest tylko kwestia nie tego czy wykona polecenie a jak szybko dopoki system waliduje i jest feedback loop. W tym momencie model dziala jedynie jako maksymalizacja prawdopodobieństwa ze odpowiedz bedzie poprawna

[Aiden Bai @aidenybai]

the type of code bro writes after spending $1,000 orchestrating 50 agents