When Parallel Drafter Meets Parallel Speculative Decoding

Excerpt

arXiv:2609.27396v1 Announce Type: new Abstract: DSpark-style parallel drafters have made speculative decoding highly effective, yet their draft phase remains serialized on the critical path of every round. Parallel speculative decoding (PSD) overlaps drafting with verification, yet existing methods must guess the accepted prefix and bonus token in advance: a wrong guess reverts the whole batch to serial drafting. We present DPara, a PSD framework that reuses effective parallel drafters yet guara