Only consecutive accepted prefixes count: one pass commits up to k+1 tokens
cneuralnetwork · x · 2026-10-22
Part seven explains the acceptance rule:
- Only the consecutive accepted prefix is kept: if the draft produces 4 tokens and the target accepts the first 3 but rejects the 4th, those 3 are kept, a corrected replacement is sampled for the 4th, and everything after the rejection is discarded;
- If all k tokens are accepted, one extra token can be sampled from the target's next-token distribution;
- Hence a single target forward pass can commit up to k+1 tokens.
Related event: Speculative Decoding: Acceptance Rules and Speedup Trade-offs(2 posts)→
More from Research
- Steering a single neuron in Olmo 3 bypasses post-trained assistant behavior — Sauers_ · 2026-10-22
- NVIDIA's Nemotron-Personas-Germany dataset trends on Hugging Face — nvidia · 2026-10-22
- CMU paper defines solver-sufficient symbols to cut hybrid robot planning cost — Yiyuan Pan · 2026-10-22
- NeurIPS Creative Track 2026 seeks sponsors to fill venue with AI art installations — dribnet · 2026-10-22
- JHU researchers preview their NLP papers coming to EMNLP 2026 in Budapest — mdredze · 2026-10-22
- Google DeepMind's Discovery team makes new progress on the elusive 6th Busy Beaver number — robertarail · 2026-10-22