介绍

本项目为“面向寒武纪加速卡MLU270-F4的多目标神经结构搜索方法”论文源码

安装依赖

pytorch>=1.4.0
torchvision>=0.5.0

使用方法

# search process
python EMO.py

# best solutions
CIFAR: [64, [5, 4, 6, 1, 6], [2, 1, 1, 4, 7, 2], [1, 2, 7, 0, 3, 7, 2, 6], [2, 1]]
ImageNet: [64, [1, 3, 5], [1, 1, 5, 4, 1, 2, 6, 1, 2, 2], [5, 3, 4], [0, 5]]

# training process（you can use our public checkpoints to reduce training time）
python train_cifar.py
python train_imagenet.py

致谢

感谢NSGA-II，NSGA-Net以及MFENAS论文的帮助