Please update the unfair comparison with ShiftAddNet

huawei-noah / AdderNet

Code for paper " AdderNet: Do We Really Need Multiplications in Deep Learning?"

BSD 3-Clause "New" or "Revised" License

954 stars 187 forks source link

Please update the unfair comparison with ShiftAddNet #43

Closed ranery closed 3 years ago

ranery commented 3 years ago

Hi,

We recently noticed that you update the comparison with ShiftAddNet in your readme.

However, the chosen accuracy of ShiftAddNet (i.e., 85.10%) is wrongly from our adaption experiments, for which we only use half the CIFAR-10 dataset to pre-train. Our accuracy (ResNet-20 on CIFAR-10) should be 89.32%, which is also trained with a different setting and different implementation (e.g., only 160 epochs).

Could you help to update the comparison results?

Best regards, Haoran You

HantingChen commented 3 years ago

Sorry for the mistake. The results have been updated.

ranery commented 3 years ago

Thanks!

brisker commented 3 years ago

@ranery Hi, so if shiftaddnet follow the same training setting as the original addernet paper, will you also get 91.8 performance for resnet20-cifar10 ??