jaiwei98 / MobileNetV4-pytorch

An unofficial implementation of MobileNetV4 in Pytorch
https://arxiv.org/abs/2404.10518
MIT License
102 stars 8 forks source link

HI #7

Open yulongPan opened 1 month ago

yulongPan commented 1 month ago

Hello, I would like to ask why I am using V4 classification with Arcface loss function and why the accuracy has always been 0. I only added a fully connected layer to the last layer and left it unmodified

jaiwei98 commented 1 month ago

Hello, I assume you are doing binary classification. Is that possible that you accidently reversed the output in your model? So that the accuracy always been 0.

yulongPan commented 1 month ago

您好,我假设您正在进行二元分类。您是否可能不小心反转了模型中的输出?因此,精度始终为 0。

Thank you for your reply. I used arcface loss with your V4 model, and I don't think I reversed the output. May I have your contact information

Hello, I assume you are doing binary classification. Is that possible that you accidently reversed the output in your model? So that the accuracy always been 0.

Thank you for your reply. I used arcface loss with your V4 model, and I don't think I reversed the output. May I have your contact information

yulongPan commented 1 month ago

I had solve my problem,Thank you