Abstract:Autoregressive decoding is bottlenecked by its sequential nature. Speculative decoding has become a standard way to accelerate inference by using a fast draft model to predict upcoming tokens from a slower target model, and then verifying them in parallel with a single target model forward pass. However, speculative decoding itself relies on a sequential dependence between speculation and verification. We introduce speculative speculative decoding (SSD) to parallelize these operations. While a verification is ongoing, the draft model predicts likely verification outcomes and prepares speculations pre-emptively for them. If the actual verification outcome is then in the predicted set, a speculation can be returned immediately, eliminating drafting overhead entirely. We identify three key challenges presented by speculative speculative decoding, and suggest principled methods to solve each. The result is Saguaro, an optimized SSD algorithm. Our implementation is on average 30% faster than optimized speculative decoding baselines and up to 5x faster than autoregressive decoding with open source inference engines.
| Comments: | ICLR 2026 |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2603.03251 [cs.LG] |
| (or arXiv:2603.03251v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.03251 arXiv-issued DOI via DataCite |
From: Tanishq Kumar [view email]
[v1]
Tue, 3 Mar 2026 18:41:32 UTC (277 KB)
[v2]
Sun, 22 Mar 2026 00:16:00 UTC (280 KB)
[v3]
Mon, 4 May 2026 21:06:36 UTC (288 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。