How to get a model's information about the input size

I’m new about cv, but I have some background.
I feel weird that in Lesson 1 we use a resnet to classify mnist data “3” and “7”.
In my memory, the input size of resnet is 3224224, but for this mnist data, the size is exactly 32828.

Surely, we can use print(models.resnet()) to see its structure, but I cannot find how to know its input tensor size .

The print result is as follows:
Test: None, model=Sequential(
(0): Sequential(
(0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
(1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)…

So can anyone let me know how could this model fit the mnist data’s size and how to know the input image’s size of the model(I supposed that the default size is 224*224)?

1 Like

Bumping. Also interested about what happened to data.resize in v1.0

Pytorch’s implementation of ResNet will upsample or downsample your input image for you if it’s larger or smaller than 224 square.