In PyTorch model-zoo(https://github.com/pytorch/vision/blob/master/torchvision/models/inception.py), there are InceptionA-E for Inception-v3 code, which are different from Table 1 of paper(https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Szegedy_Rethinking_the_Inception_CVPR_2016_paper.pdf), however.
Could anyone explain the reason? The code seems like Inception-v4 in some way.
Thanks.