一作ACL 2026Findings· Findings of ACL
Speculative Decoding with a Speculative Vocabulary
Miles Williams · Young D. Kwon · Rui Li · Alexandros Kouris · Stylianos I. Venieris
摘要
Speculative decoding has rapidly emerged as a leading approach for accelerating language model (LM) inference, as it offers substantial speedups while yielding identical outputs. This relies upon a small draft model, tasked with predicting the outputs of the target model. State-of-the-art specula...