Sasha Rush's three laws of 'Vibed ML' papers poke fun at theory-after-hack culture
srush_nlp · x · 2026-10-23
Sasha Rush summarizes how 'Vibed ML' papers get written: 1) reduce variance, 2) keep things normalized, 3) do whatever Noam says is good. He adds that these papers deploy a cool hack in Sections 3–4, then label Section 5 'Theoretical Analysis' — as if anyone understood why it works. A wry take on empirical hacks outpacing theory in ML research.
Related event: Sasha Rush Mocks 'Vibe ML Papers' and Their Token Theoretical Analysis(3 posts)→
More from Fun
- It's not AI slop code causing cyber crises — it's old human slop — emax · 2026-10-23
- Viral satire: 'Filmmaking can never be ethical — you're stealing light and sound in a little box' — kyliebytes · 2026-10-23
- Bindu Reddy jokes the best cure for AI panic is a new frontier model drop to steady the stock market — bindureddy · 2026-10-23
- AI meme: stick figure tosses model mascots into a trash bin full of logos — XFreeze · 2026-10-23
- Top engineers always hated PHP — good AI coding tools finally revealed how bad it is — arieljalali · 2026-10-23
- Clustered CUDA launch runs 128 blocks when gridDim says 64: a hidden driver bug — mike64_t · 2026-10-23