search

Alibaba-MIIL / ImageNet21K

Official Pytorch Implementation of: "ImageNet-21K Pretraining for the Masses"(NeurIPS, 2021) paper
MIT License
724 stars 71 forks source link

Could you provide with commonly used pretrained models with se modules? #15

Closed KleinXin closed 3 years ago

KleinXin commented 3 years ago

As it is evident SE module can dramatically improve the accuracy of network such as ResNet, coule you please also provide some pretrained models?

Just like SE-ResNet101/SE-ResNet50... in this page.

https://github.com/Cadene/pretrained-models.pytorch#torchvision

mrT23 commented 3 years ago

TResNet models have SE, and are better the SE-ResNet. use them. (TResNet-M is equivalent in runtime to SE-ResNet50, TResNet-L-V2 is equivalent in runtime to SE-ResNet101)

note that mobilenet and OFA models also use SE.