according to the documentation https://pytorch.org/docs/stable/torchvision/models.html , pretrained models are pretrained on ImageNet. But I’d like to know exactly the year of the ImageNet dataset used. Does it matter? Or imagenets are roughly the same?

They are trained on ImageNet-12: [image] Using vision.models with the CIFAR dataset? vision The pre-trained models are all based on Imagenet-12. Most of the model definitions also need 224x224 as the minimum input size. For CIFAR, you are better off with other repos.

Which year ImageNet are torchvision models pretrained on?

vision

ptrblck August 10, 2018, 8:21am 2

They are trained on ImageNet-12:

1 Like

Torchvision.models of pytorch