Post

babe wake up, new results just dropped, lda is still great

aclanthology.org/2026.tacl-1....

"Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders" This paper's title makes me so happy— I feel like I've gone back seeing *ML* terms & references, before the time when it was just marketing branding it as 'AI'

same 🥹 and the equations are all so nice and familiar