Hello I am a newbie and beginner 
i would like to modify vgg19 model to train mnist. how can i modify the input size of vgg19 model (224,224) to (28,28)?
i want to modify this code torchvision.models.vgg — Torchvision 0.12 documentation …
thanks for reading really thanks… 
You could start removing layers which would raise an error if they would create an empty output tensor and maybe relax the spatial pooling by e.g. also removing pooling layers more aggressively.
Alternatively, you could increase the spatial size of your MNIST dataset and just use this model for your experiments.
i appreciate your support really thanks for your help! 