spock

@signulll that nikita response got you good lol

let’s instamaxx this platform to make it more inclusive. 07.12.26

@elliotarledge nice. did you learn cuda by hand or were agents already out and good enough to help when you started?

i did not read pmpp

@tenderizzation one time an important compute company recruiter lectured me on how that’s not true in a “few specific pytorch repositories” despite the fact that I actually had a pr merged in the project he was lecturing me about… safe to say he doesn’t work there anymore.

pytorch’s merge process uses a bot to do the actual merge while closing the gh PR with the exception of release branch cherrypicks so while my gh looks wack whenever someone asks me why all my PRs are closed i know they’ve never landed anything in pytorch

@mramiracle who are you following that is crying about 9-9-6? maybe they need to be unfollowed xd

wtf is all this fuss about 996? work is done in silence. not on twitter. i’ve done 996 since i started my first company at 19 and never had to tweet about it smh

@Amank1412 internal use for meta employees. manus would've been a perfect role-model agentic chat interface.

why did meta acquire manus ai? feels like it disappeared forever ago.

@mark_k @GoogleDeepMind if you're a google employee, there is no pressure. everyone keeps forgetting that google doesn't need to follow the pace of the race. they make revenue just fine to hold out longer than everyone else...

Right now, @GoogleDeepMind is under a lot of pressure. The recent wave of departures has already generated plenty of negative press, and now we’ve had a streak of incredibly strong model releases from OpenAI, Meta, and SpaceXAI. If Gemini 3.5 Pro sucks, it obviously won’t be the end of Google. But you can rest assured that the lab is currently a pressure cooker.

@sama new inferenceware putting in work huh

the sun is out today