PaddleMAE

Masked Autoencoders Are Scalable Vision Learners

Dataset: ImageNet1k Target: ViT-B，imagenet1k val 83.6%

V100 CUDA 10.2

conda create -n paddle-mae python=3.7
conda activate paddle-mae
pip install -r ./requirements.txt

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
.gitignore		.gitignore
README.md		README.md
data.py		data.py
main.ipynb		main.ipynb
model.py		model.py
requirements.txt		requirements.txt
train.py		train.py