BriefGPT.xyz
Feb, 2024
重访视觉调优的能量与提示
Revisiting the Power of Prompt for Visual Tuning
HTML
PDF
Yuzhu Wang, Lechao Cheng, Chaowei Fang, Dingwen Zhang, Manni Duan...
TL;DR
通过关键参数的优化和初始化,我们提出了一种改进的视觉提示调整方法,用于优化预训练模型在下游任务中的性能表现,并在大量实验证明该方法在自我监督预训练适应性上取得了显著的性能改进。
Abstract
visual prompt tuning
(VPT) is a promising solution incorporating learnable prompt tokens to customize pre-trained models for downstream tasks. However, VPT and its variants often encounter challenges like
prompt initial
→