大模型
arxiv.org ·
超越截断：将 LLM 解码重新思考为集成剪枝
论文提出 Mahalanobis-Ensemble Decoding（ME-Decoding）解码框架，把候选 token 选择视为集成剪枝问题。针对现有策略仅依赖标量概率、忽略语义几何关系导致候选冗余，以及几何感知方法计算开销大或推理不稳定的问题，作者用 Mahalanobis 距离驱动的子集优化目标，通过基于 token 嵌入的自适应带宽核构建相似度矩阵，动态折扣冗余生成路径，并设计近线性复杂度的贪心选择算法及理论近似保证。该方法为即插即用模块，推理开销可忽略，在多种推理与生成任务上表现良好。
AI推荐理由 arXiv 新论文，提出 ME-Decoding 解码框架，将候选 token 选择建模为集成剪枝，用 Mahalanobis 距离驱动的子集优化提升语义多样性，并给出近线性复杂度的贪心选择算法及近似保证。属于 LLM 解码策略的方法创新，对推理与生成任务有潜在价值；但摘要仅称实验表现良好，未给出具体基准、对比数字或开源代码，实际收益尚待验证。
72
正文
Computer Science > Artificial Intelligence
arXiv:2609.18723 (cs)
[Submitted on 16 Sep 2026]
Title:Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning
Authors:Dunyao Xue, Chengshuo Du, Zhengbo Wang, Wenlin Dai, Cheng Meng
View a PDF of the paper titled Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning, by Dunyao Xue and 4 other authors
View PDF
HTML (experimental)
Abstract:We introduce Mahalanobis-Ensemble Decoding (ME-Decoding), a novel Large Language Model (LLM) decoding framework that frames candidate token selection as ensemble pruning. Existing selection strategies rely predominantly on scalar probabilities, ignoring geometric semantic relationships and causing candidate redundancy. Meanwhile, current geometry-aware methods often require complex optimization or directly reweighting the original token probabilities, leading to significant computational overhead or inference instability. To address this, we formulate decoding as a subset optimization problem using a Mahalanobis distance-driven objective to enhance semantic diversity while preserving high probabilities. Specifically, we dynamically discount redundant generation paths using a token similarity matrix, constructed via an adaptive-bandwidth kernel over token embeddings. We further devise an efficient greedy selection algorithm with near-linear complexity in the candidate size under early stopping, while establishing its theoretical approximation guarantees. This renders ME-Decoding a robust, plug-and-play module with negligible inference overhead. Extensive experiments across diverse reasoning and generation tasks demonstrate that our method consistently achieves strong performance.
Comments:
Subjects:
Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Cite as:
arXiv:2609.18723 [cs.AI]
(or
arXiv:2609.18723v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.18723
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Dunyao Xue [view email]
[v1]
Wed, 16 Sep 2026 14:23:40 UTC (311 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning, by Dunyao Xue and 4 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.AI
new
|
recent
| 2026-09
Change to browse by:
cs
cs.LG
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)