Model forward changes the input?

Hello, I train model with pytorch and the traning working good.
I try to get the summary from this model and also predict to get result, but i get something strange.
In the train i use batch size that equal to 1, and the image size is 224,224,8.
and the tensor shape is ([1, 8, 244,244]) but in the summary i get torch.Size([2, 8, 224, 224]).

I try to understand what the problem and i don’t success to understand, someone can help?

Could you post the model definition here so that we could try to reproduce the unexpected shape?