Input node retains batch dimension #17
Description
Since in pytorch, the data is always expected to have the batch dimension, this shape is also retained in the input shape. This is not consistent with NIR specifications that does not include the batch dimension.