About network output channel number

[eeyrw]

I notice that the final output of network is the concatenation of two tensor in dimension 1. https://github.com/Vipermdl/MobileNetV3-for-Segmentation/blob/b195d4a7817b88292d715ed0c4b53036a6e912d1/model/MobileNetV3.py#L244 It seems that the output has num_class*2 channels. Does it make sense? I get quite confused.