ConvNet Shape Calculator

Input

32×1×28×28

Conv2d

32×1×28×28 → 32×3×24×24

in_channels:

out_channels:

kernel_size:

stride:

padding:

dilation:

MaxPool2d

32×3×24×24 → 32×3×8×8

kernel_size:

stride:

padding:

dilation:

Conv2d

32×3×8×8 → 32×6×6×6

in_channels:

out_channels:

kernel_size:

stride:

padding:

dilation:

MaxPool2d

32×6×6×6 → 32×6×2×2

kernel_size:

stride:

padding:

dilation:

Output

32×6×2×2