[KongchangAI]
ConceptMTP / 多token预测

Multi-Token Prediction

A technique in which a model predicts multiple future tokens in a single forward pass. It is commonly used in speculative decoding to accelerate inference.

Core Facts

Timeline (last 90 days)

Sep 10

MTP (Multi-Token Prediction) is a configuration flag that is disabled by default in most mainstream inference tools and can accelerate inference.

Verified65%

Source Articles