English(EN) Stochastic Pruning of the DPO Computation Graph: 1.4x Faster Training and Lower Peak Memory

新的DPO剪枝技术将AI训练时间缩短1.4倍

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-22 04:00

研究人员开发了一种新颖的直接偏好优化（DPO）计算图随机剪枝框架，可显著加快训练速度并减少内存使用。该方法通过对评分响应位置、执行前缀和反向传播的偏好分支进行采样，从而省略不必要的计算。实验表明，与标准DPO相比，步进时间提高了1.4倍，总体训练时间减少了约30%，而准确率仅略有下降。 AI

影响该研究提供了一种使用DPO显著减少训练大型语言模型所需计算资源的方法，有望加速开发周期。

排序理由学术论文，详细介绍了一种优化AI模型训练的新方法。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 1 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍了一种优化AI模型训练的新方法。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 2 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

arXiv cs.LG TIER_1 English(EN) · Yubing Wu, Wei Chen, Junmei Yang, Delu Zeng, Tiewen Pan · 2026-10-22 04:00

DPO计算图的随机剪枝：训练速度提升1.4倍，峰值内存降低

arXiv:2610.29538v1 Announce Type: new Abstract: Direct preference optimization (DPO) incurs training costs in output scoring, sequence execution, and backpropagation, so reducing supervision alone does not necessarily save computation. We introduce a unified framework for stochas…