arXiv cs.CL· Bobo Li, Jiajia Song, Hao Fei, Wynne Hsu, Mong Li Lee·· 11 小时前AI 评分35
Flide：用精确状态分叉加速决策推理
Flide: Accelerating Decision Inference with Exact State Forking
AI 导读
Flide 是一个面向决策模型（如 Jev）的精确推理引擎，通过 ForkAttention 在单条扁平 token 流中解耦请求的逻辑树结构与物理执行，保证每个查询分支与独立前向计算完全一致。在公开混合模型与决策模型上，其回答速度最高达 vLLM 和 SGLang 的 9.5 倍，同时总变差距离最接近 float32 参考结果。该引擎还支持从早期检查点恢复不断扩展的智能体历史。
正文 · 原文
Abstract:Decision models such as Jev have emerged as a highly promising paradigm for LLM agent systems, providing lightweight and reliable categorical reasoning for rapid System One execution. However, existing generation-oriented inference engines fail to support these workloads effectively, suffering from severe prefix recomputation overheads or distorted output probabilities. To address this challenge, we propose Flide, an efficient and exact inference engine that decouples the logical tree structure of a decision request from its physical execution within a single flattened token stream. Through ForkAttention, Flide navigates request dependencies using GPU-resident indices across both standard attention and recurrent Gated DeltaNet layers, guaranteeing in exact arithmetic that every query branch evaluates identically to an isolated forward pass. By expressing branch topologies directly in GPU-resident metadata, Flide accommodates diverse request mixtures with a compact set of pre-recorded execution graphs while enabling expanding agent histories to resume from checkpoints saved in earlier passes. Extensive evaluations on public hybrid and decision models demonstrate that Flide answers queries up to 9.5 times faster than vLLM and SGLang while staying closest to a float32 reference in total variation distance. By unifying physical execution throughput with mathematical exactness, Flide establishes a principled foundation for scalable, latency-critical decision intelligence in agent systems.
| Comments: | 20 pages. Project page: this https URL |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.42148 [cs.CL] |
| (or arXiv:2610.42148v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.42148 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Bobo Li [view email]
[v1]
Fri, 30 Oct 2026 17:59:38 UTC (514 KB)
来源：arXiv cs.CL · arxiv.org