arXiv：cs.LG（机器学习，全量分类）· Zhiyuan Ma·· 17 小时前AI 评分34
LumoTree：面向混合语言模型的路径并行推测验证
LumoTree: Path-Parallel Speculative Verification for Hybrid Language Models
AI 导读
LumoTree 是一种面向混合语言模型的树形推测解码验证器，通过并行执行循环路径、复用路径内状态块，并在共享逻辑树中协调循环重放、卷积历史收集与注意力缓存重映射。
正文
Abstract:Tree speculative decoding for hybrid language models must preserve one coherent continuation across recurrent, convolution, and attention state. We present LumoTree, a verifier that executes recurrent paths in parallel, reuses state tiles within each path, and coordinates native recurrent replay, convolution-history gathering, and attention-cache remapping through a shared logical tree. Fused candidate selection, GPU-resident acceptance, and grouped split-K attention support the verification cycle. Component experiments show exact candidate-selection parity and recurrent agreement within paired error bounds. An exploratory Qwen3.8-27B NVFP4 deployment on a single NVIDIA DGX Spark (GB10) records 25.63 pooled tokens/s on ten SWE-bench Verified Astropy tasks. The results characterize component-level numerical agreement and coding-agent deployment; complete-model continuation and controlled application speedups remain open.
| Comments: | 11 pages, 3 figures |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2609.23900 [cs.LG] |
| (or arXiv:2609.23900v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2609.23900 arXiv-issued DOI via DataCite |
Submission history
From: Zhiyuan Ma [view email]
[v1]
Sun, 20 Sep 2026 22:21:01 UTC (88 KB)
[v2]
Thu, 1 Oct 2026 01:53:32 UTC (96 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org