jaxony / ShuffleNet

ShuffleNet in PyTorch. Based on https://arxiv.org/abs/1707.01083
MIT License
294 stars 90 forks source link

maybe a mistake #11

Open tanglang96 opened 6 years ago

tanglang96 commented 6 years ago

in file model.py , your function conv3x3() is a simple 2d convolution , but I found it should be depthwise conv according to the paper , the computation cost is totally different