通过分组重构提升低比特量化精度，解决传统方法依赖块独立带来的性能下降问题。
Pack-PTQ: Advancing Post-training Quantization of Neural Networks by Pack-wise Reconstruction
- 按海塞矩阵自适应分组，保留块间依赖关系
- 混合精度分配不同位宽，针对敏感度优化量化策略
- 在图像与点云任务中均超越现有最佳方法
后训练量化（PTQ）已成为压缩复杂模型的重要方案，具有校准数据集小、无需端到端重训练的优点。然而，现有大多数PTQ方法采用块级重构，忽略了块间依赖性，在低比特情况下出现明显精度下降。为此，本文提出一种新型PTQ方法——Pack-PTQ。首先，设计基于海塞矩阵的自适应分组机制，将块划分为非重叠的包作为重构基础单元，以保留块间依赖并实现更准确的量化参数估计；其次，根据包配置提出混合精度量化策略，依据各包敏感度动态分配不同位宽，进一步提升性能。在多种网络架构下，针对2D图像和3D点云分类任务的大量实验表明，该方法显著优于当前最先进的PTQ方法。
原文摘要 · Abstract (English)
Post-training quantization (PTQ) has evolved as a prominent solution for compressing complex models, which advocates a small calibration dataset and avoids end-to-end retraining. However, most existing PTQ methods employ block-wise reconstruction, which neglects cross-block dependency and exhibits a notable accuracy drop in low-bit cases. To address these limitations, this paper presents a novel PTQ method, dubbed Pack-PTQ. First, we design a Hessian-guided adaptive packing mechanism to partition blocks into non-overlapping packs, which serve as the base unit for reconstruction, thereby preserving the cross-block dependency and enabling accurate quantization parameters estimation. Second, based on the pack configuration, we propose a mixed-precision quantization approach to assign varied bit-widths to packs according to their distinct sensitivities, thereby further enhancing performance. Extensive experiments on 2D image and 3D point cloud classification tasks, using various network architectures, demonstrate the superiority of our method over the state-of-the-art PTQ methods.
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。