From the source
Google Research introduces a method to retrofit Multi-Token Prediction onto frozen Gemini Nano v3 models, accelerating on-device inference on Pixel 9 and 10 series without separate drafters.
Accelerating Gemini Nano models on Pixel with frozen Multi-Token Prediction
Capability change
From the source
We introduce a method to retrofit Multi-Token Prediction onto frozen production models, accelerating on-device inference without the inefficiencies of separate drafters.
research.google