Train from Scratch of ViG

huawei-noah / Efficient-AI-Backbones

Efficient AI Backbones including GhostNet, TNT and MLP, developed by Huawei Noah's Ark Lab.

4.07k stars 708 forks source link

Train from Scratch of ViG #136

Closed yannhan closed 2 years ago

yannhan commented 2 years ago

Hi,

Thanks for your great work and sharing the codes, I am trying to reproduce the results by training the ViG from scratch. And I followed the README file to install required packages with specific version number, but I got the following training status, it looks a little weird. Could you please check it and let me know whether it is normal?

Thanks!

iamhankai commented 2 years ago

From your log, the training loss is decreasing normally, but the val loss and acc are not normal. Please check the val data or code.

Here is my log:

yannhan commented 2 years ago

Hi, thanks for your reply, I downloaded the IamgeNet dataset from Kaggle, https://www.kaggle.com/competitions/imagenet-object-localization-challenge/data, and do you make some modification about the dataset?

iamhankai commented 2 years ago

We followed the official pytorch example: move and extract the training and validation images to labeled subfolders, using the following shell script