README_en.md

Paddle_VIT_tutorial

English | 简体中文

This repo contains some codes recorded from the online course, Learn Vision Transformer from Scratch, which was lectured by Dr. Zhu, Baidu PaddlePaddle.

If you have any questions, please feel free to contact me.

Official code：https://github.com/BR-IDL/PaddleViT/tree/develop/edu

Timeline and corresponding codes:

• Class #0, 2021.11.23

  Implementation of resnet18. resnet.py

• Class #1, 2021.11.24

  Let's build a ViT! vit.py

• Class #2, 2021.11.25

  Multi-Head Self Attention. attention.py

• Class #3, 2021.11.26

  Implementation of ViT. vit_1126.py

• Class #4, 2021.11.27

  Implementation of DeiT. deit/deit.py

  Before feeding to a net: Image Preprocess. deit/transforms.py

• Class #5, 2021.11.28

  Window Attention. swin_transformer/main_1128.py

• Class #6, 2021.11.29

  Attention Mask. swin_transformer/mask_1129.py

  Implementation of SwinBlock, a block of Swin Transformer. swin_transformer/main_1129.py

• Class #7, 2021.11.30

  Implementation of Swin Transformer. swin_transformer/main_1130.py

  Used to load data: Iterator. iterator_1130/tmp.py

• Class #8, 2021.11.31

  How does PaddleViT set and load configs? load_config

• Class #9, 2021.12.1

  Distributed training for PaddlePaddle. distributed/main.py

• Class #10, 2021.12.2

  Implementation of DETR. detr

Thanks a lot for what Baidu PaddlePaddle have done! Fighting!