(conv): Sequential( (0): Conv2d(1, 64, kernel_size=(4, 4), stride=(1, 1)) (1): ReLU() (2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
Imagine the input shape looks like 1x7x7
How does the input data gets transformed/scaled into the In/Out channels of Conv2d?
I’m sure I sound silly but all explanations came with the perspective of images where the input was larger than the channels. My case is Connect4 (with 6x7 + 1x7 action space). I am trying to get a pattern recognition.