Ma Zerun

> @Ezra-Yu follow your changes exactly. Please fix the lint error before merge.

The unit tests need to be fixed.

I think your dataset implementation has some problems that caused it to be empty. Did you implement it according to https://mmclassification.readthedocs.io/en/master/tutorials/new_dataset.html#create-a-new-dataset-class

I think it's a PyTorch installation problem

Could you provide any config file which use this neck?

These checkpoints are converted from [pytorch-image-models](https://github.com/rwightman/pytorch-image-models), not official checkpoints. They re-trained these models with new tricks like `RandAugment` and get a higher score. The detailed training recipe maybe like [the...

> We already have timm backbone wrapper.

Thanks for your contribution. In fact, the deploy functionalities in mmclassification are deprecating, we move all these functionalities to [mmdeploy](https://github.com/open-mmlab/mmdeploy). Have you informed them?

> hope to add reid-related structures What structures? ReID sounds should be in MMTrack

是这样的，这里的 prob 只是用作记录概率，随机选取是在 https://github.com/open-mmlab/mmclassification/blob/master/mmcls/models/utils/augment/augments.py 中进行的。 后续我们会对这块的逻辑进行重构