Fast Inference via Hierarchical Speculative Decoding

Clara Mohri, Amir Globerson, Haim Kaplan, Yishay Mansour, Tal Schuster

Keywords: speculativedecoding

TL;DR: Generalization of speculative decoding along with optimization algorithm

Abstract: Transformer language models generate text autoregressively, making inference latency proportional to the number of tokens generated. Speculative decoding reduces this latency without sacrificing output quality, by leveraging a small draft model to propose tokens that the larger target model verifies in parallel. In practice, however, there may exist a set of potential draft models—ranging from faster but less inaccurate, to slower yet more reliable. We introduce Hierarchical Speculative Decoding (HSD), an algorithm that stacks these draft models into a hierarchy, where each model proposes tokens, and the next larger model verifies them in a single forward pass, until finally the target model verifies tokens. We derive an expression for the expected latency of any such hierarchy and show that selecting the latency-optimal hierarchy can be done in polynomial time. Empirically, HSD gives up to 1.2× speed-up over the best single-draft baseline, demonstrating the practicality of our algorithm in reducing generation latency beyond previous techniques.

Paper Type: Long (8 pages)

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 19