Google Research(@GoogleResearch):Today on the blog we introduce a method to retrofit Multi-Token Prediction onto frozen production models, accelerating on-device inference without the inefficiencies of separate drafters. Learn more →https://t.co/9Tq8hosoxS

Google Research

@GoogleResearch

Impossible? Let’s see. From algorithms to neuroscience to AI, Google Research strives to progress science, advance society & improve billions of people’s lives.

加入 May 2017

17 正在關注 104K 粉絲

Google Research@GoogleResearch

2026.06.26 18:48

Today on the blog we introduce a method to retrofit Multi-Token Prediction onto frozen production models, accelerating on-device inference without the inefficiencies of separate drafters. Learn more →