Podcast: DeepMind Research Director on Text Diffusion Models and RL
ziv_ravid · x · 2026-08-15
zivravid released a new podcast episode with Brendan O'Donoghue, Research Director at Google DeepMind, discussing text diffusion models. Topics include serving cost as the real bottleneck, Gamma diffusion as a good base, diversity aiding RL, samplers as the biggest open problem, hardware trends favoring diffusion, and RL being underhyped.
More from Research
- Researcher slams papers whose stacked changes beat even weak baselines barely — retr0jirachi · 2026-10-16
- FrankenSim builds a physics-based virtual orchestra, piano and drum demos live — doodlestein · 2026-10-16
- FoveaVLA replaces wrist cameras with learned foveation, hitting 70.8% success — Haoyu_Xiong_ · 2026-10-16
- ApolloPFN: a zero-shot forecasting network that natively handles exogenous covariates — burkov · 2026-10-16
- ICLR reviewer says 4/5 papers in review pile read like AI slop — berkustun · 2026-10-16
- Tsinghua's DBS learns content-adaptive frequency band splits for audio source separation — thu-ml · 2026-10-16