Verification-aware training speeds up draft models

Automated excerpt

However, draft models are trained without regard to the sequential verification step that discards tokens after the first rejection. A training plug‑in that simulates verification adds an extra 8. 7 % wall‑clock speedup and yields longer accepted token sequences[1]. These gains arise without altering the draft architecture, target model, or inference engine; VAT merely reshapes the training loss to mirror downstream acceptance patterns[1].

Selected automatically from source text; not independently written or fact-checked. Read the original for full context.