PyTorch笔记 - A ConvNet for the 2020s (ConvNeXt) 训练

欢迎关注我的CSDN:https://blog.csdn.net/caroline_wendy
本文地址:https://blog.csdn.net/caroline_wendy/article/details/128243621

Paper:A ConvNet for the 2020s,FAIR,Berkeley

Code:https://github.com/facebookresearch/ConvNeXt

Isotropic ConvNeXt vs. ViT

  • ConvNeXt去掉下采样层,就是与ViT各项同性

  • ViT来源于DeiT -> ViT-S/B,MAE -> ViT-L

  • warm-up的epoch增多

源码:

class ConvNeXtIsotropic(nn.Module)

你可能感兴趣的:(深度学习,pytorch,python,深度学习)