Meta compression researcher Guangxuan Xiao says gradual pruning won't recover performance faster
teortaxesTex · x · 2026-11-21
Andrew Carr asked whether gradual or souped-up pruning could recover performance sooner than a single-step fine-tune. Guangxuan Xiao, Meta's model compression researcher, replied with a flat "No 😊" — a concise negative answer from someone working at the frontier of pruning research.
More from Research
- DatologyAI open-sources Zephon, a data loader with elastic determinism for LLM training — josh_wills · 2026-11-21
- Real Cameras are Radiometric: new paper lifts any geometric camera model with aperture, focus surface, vignetting — ssh4net · 2026-11-21
- USC's Heave-Ho trains Unitree G1 humanoids to collaborate, scales 2 to 10 robots — CyberRobooo · 2026-11-21
- Kaomoji emotion probe validated: 0.68 correlation with valence direction, 0.79 when trained — Sauers_ · 2026-11-21
- PhysLDM reconstructs deformable trajectories at 2.48mm precision with 78x token compression — jiqizhixin · 2026-11-21
- Hand-crafting single MLP neurons to steer model behavior — no optimizer needed — voooooogel · 2026-11-21