Google releases Multi-Token Prediction drafters for its Gemma 4 models, which use a form of speculative decoding to guess future tokens for faster inference (Ryan Whitwam/Ars Technica)

Google has released Multi-Token Prediction drafters for its Gemma 4 AI models. The drafters use speculative decoding to guess future tokens, allowing the models to infer faster. Reported improvements include a three-fold speed boost for inference.

Rédigé à partir de l’ensemble des articles ci-dessous, et non d’un seul.