Is there anything wrong about the shape of CRNN modual

[0sxw0]

i try caculate the shape of the rcnn modual ,but since 3rd maxpooling it happened to be strange,the input size is 1,32,128, the annotation say 256x4x16,my result is 256x4x33. How does it happen, can y help me?Or it there anython wrong? cnn.add_module('pooling{0}'.format(2), nn.MaxPool2d((2, 2), (2, 1), (0, 1))) # 256x4x16 https://github.com/meijieru/crnn.pytorch/blob/master/models/crnn.py