BriefGPT.xyz
May, 2024
PTQ4SAM:用于分段任意物体的训练后量化
PTQ4SAM: Post-Training Quantization for Segment Anything
HTML
PDF
Chengtao Lv, Hong Chen, Jinyang Guo, Yifu Ding, Xianglong Liu
TL;DR
在这篇论文中,我们提出了一种针对Segment Anything Model的后训练量化框架,即PTQ4SAM。我们通过分析SAM量化中的双峰分布特性,提出了双峰积分策略,并采用适应性颗粒度量化方法来处理SAM中的后Softmax分布,实验证明PTQ4SAM在各种视觉任务和模型变体中具有卓越的优势。
Abstract
segment anything model
(SAM) has achieved impressive performance in many computer vision tasks. However, as a large-scale model, the immense memory and computation costs hinder its practical deployment. In this paper, we propose a
→