用权重空间运算提取通用量化鲁棒性，零样本提升低比特模型精度。
Zero-Shot Quantization via Weight-Space Arithmetic
- 通过权重空间运算生成量化向量，无需训练即可增强模型抗量化能力。
- 3比特下最高提升60点准确率，跨任务迁移效果显著。
- 适合低资源场景的模型部署，无需接收端数据或微调。
我们发现，训练后量化（PTQ）的鲁棒性是一种可迁移的权重空间方向。我们称其为量化向量：通过简单的权重空间算术从源任务中提取，可直接应用于目标模型，显著提升3比特设置下的后训练量化准确率，最高可达60点增益，且无需目标模型的量化感知训练（QAT）。该方法不需接收端数据，为极低比特部署提供了零样本、低成本的QAT替代方案。在四个ViT规模和22个图像分类任务上，即使源与目标任务差异较大，量化向量仍能带来显著性能提升。我们进一步严格证明了量化向量是良定义的，不受参数重参数化对称性影响，并给出了其作用的局部几何解释。这些结果表明，量化鲁棒性可通过简单的权重空间代数实现部分解耦、复用与迁移。
原文摘要 · Abstract (English)
We show that robustness to post-training quantization (PTQ) is a transferable direction in weight space. We call this direction the quantization vector: extracted from a donor task by simple weight-space arithmetic, it can be used to patch a receiver model and improve post-PTQ Top-1 accuracy by up to 60 points in a 3-bit setting, without receiver-side quantization-aware training (QAT). Because the method requires no receiver training data, it provides a zero-shot, low-cost alternative to QAT for extremely low-bit deployment. Across four ViT scales and 22 image classification tasks, donor quantization vectors often yield substantial gains even when donor and receiver tasks differ markedly. We further prove rigorously that quantization vectors are well-defined and do not suffer from reparameterization symmetries, and provide a local geometric account of their effect. Together, these results suggest that quantization robustness can be partially isolated, reused, and transferred through simple weight-space algebra.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。