Adaptive Multi-Token Prediction: How far should we go?
Multi-token prediction (MTP) can accelerate language-model decoding by drafting several tokens and verifying them in a single target-model pass. However, the best speculation horizon changes with context: longer drafts offer greater amortization when reliable, but can waste compute on suffixes discarded after a rejection. Unlike autoregressive drafting, where a stopping rule can decide token by…
We haven't written up this one. arXiv cs.AI has the full story — the link below goes straight to it.
