Authors:Junxiong Wang, Fengxiang Bie, Jisen Li, Zhongzhu Zhou, Zelei Shao, Yubo Wang, Yinghui Liu, Qingyang Wu, Avner May, Sri Yanamandra, Yineng Zhang, Ce Zhang, Tri Dao, Percy Liang, Ben Athiwaratkun, Shuaiwen Leon Song, Chenfeng Xu, Xiaoxia Wu
Abstract:Speculative decoding can significantly accelerate LLM serving, yet most deployments today disentangle speculator training from serving, treating speculator training as a standalone offline modeling problem. We show that this decoupled formulation introduces substantial deployment and adaptation lag: (1) high time-to-serve, since a speculator must be trained offline for a considerable period before deployment; (2) delayed utility feedback, since the true end-to-end decoding speedup is only known after training and cannot be inferred reliably from acceptance rate alone due to model-architecture and system-level overheads; and (3) domain-drift degradation, as the target model is repurposed to new domains and the speculator becomes stale and less effective.
To address these issues, we present Aurora, a unified training-serving system that closes the loop by continuously learning a speculator directly from live inference traces. Aurora reframes online speculator learning as an asynchronous reinforcement-learning problem: accepted tokens provide positive feedback, while rejected speculator proposals provide implicit negative feedback that we exploit to improve sample efficiency. Our design integrates an SGLang-based inference server with an asynchronous training server, enabling hot-swapped speculator updates without service interruption. Crucially, Aurora supports day-0 deployment: a speculator can be served immediately and rapidly adapted to live traffic, improving system performance while providing immediate utility feedback. Across experiments, Aurora achieves a 1.5x day-0 speedup on recently released frontier models (e.g., MiniMax M2.1 229B and Qwen3-Coder-Next 80B). Aurora also adapts effectively to distribution shifts in user traffic, delivering an additional 1.25x speedup over a well-trained but static speculator on widely used models (e.g., Qwen3 and Llama3).
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2602.06932 [cs.LG] |
| (or arXiv:2602.06932v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.06932 arXiv-issued DOI via DataCite |
From: Fengxiang Bie [view email]
[v1]
Fri, 6 Feb 2026 18:28:54 UTC (6,946 KB)
[v2]
Fri, 3 Apr 2026 12:52:35 UTC (28,940 KB)
[v3]
Sat, 2 May 2026 06:02:57 UTC (28,940 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。