maxPool currently only takes in a single integral parameter which creates a square kernel.
It should be simple enough to implement this to also support passing a two-tuple to support arbitrary kernel sizes.
See https://pytorch.org/docs/stable/generated/torch.nn.MaxPool1d.html#maxpool1d. kernel_size can be int or Tuple[int].
Ex:
x = x.maxPool(3); // Square 2D pooling
instead of
We should also add the other parameters that pytorch supports like stride, padding, dilation etc.
maxPoolcurrently only takes in a single integral parameter which creates a square kernel.It should be simple enough to implement this to also support passing a two-tuple to support arbitrary kernel sizes.
See https://pytorch.org/docs/stable/generated/torch.nn.MaxPool1d.html#maxpool1d.
kernel_sizecan beintorTuple[int].Ex:
instead of
We should also add the other parameters that pytorch supports like
stride,padding,dilationetc.