New Quantized Evolution Strategies Enable Fine-Tuning of LLMs

作者 PulseAugur 编辑部 · [1 个来源] · 2026-06-15 04:00

研究人员推出了一种名为量化进化策略（QES）的新型优化范式，旨在直接在其离散参数空间内对量化大语言模型（LLMs）进行微调。该方法解决了传统微调技术依赖连续权重和反向传播的局限性，使其不适用于量化模型。QES 结合了累积误差反馈以实现精确的权重更新，并使用无状态种子重放来最小化内存使用，从而能够以低精度推理成本进行微调。与现有的零阶微调方法相比，该方法表现出更优越的性能，为完全在量化域内扩展 LLMs 铺平了道路。 AI

影响能够在外存受限设备上更高效地部署和微调大语言模型。

排序理由这是一篇详细介绍量化 LLMs 微调新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

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

New Quantized Evolution Strategies Enable Fine-Tuning of LLMs

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 这是一篇详细介绍量化 LLMs 微调新方法的学术论文。[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, model release; 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: 126 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Yinggan Xu, Kajetan Schweighofer, Risto Miikkulainen, Xin Qiu · 2026-06-15 04:00

量化进化策略：以低精度成本实现量化大模型的超高精度微调

arXiv:2602.03120v2 Announce Type: replace-cross Abstract: Post-Training Quantization (PTQ) is essential for deploying Large Language Models (LLMs) on memory-constrained devices, yet it renders models static and difficult to fine-tune. Standard fine-tuning paradigms, including Rei…

本文如何被排名

报道来源 [1]

量化进化策略：以低精度成本实现量化大模型的超高精度微调

相关实体

相关话题