TAdaConvNeXt-T

Question

TAdaConvNeXt-T

okojoalg opened this issue 2 years ago · 3 comments

Are code and a pretrained weight of TAdaConvNeXt-T going to be public?

Answer 1 · 2023-02-03T09:24:14.000Z

And I also curious about why the flops and parms of TAdaConvNeXt-T is one half of the ResNet50's, to my knowledge they should be similar.

Answer 2 · 2023-08-16T11:45:31.000Z

Are code and a pretrained weight of TAdaConvNeXt-T going to be public?

The code is made public now, but the weight for TAdaConvNeXt would not be released. But later we will release the pre-trained weights for TAdaConvNeXtV2 and TAdaFormer.

Answer 3 · 2023-08-16T11:46:05.000Z

And I also curious about why the flops and parms of TAdaConvNeXt-T is one half of the ResNet50's, to my knowledge they should be similar.

The main reason is that TAdaConvNeXt-T downsamples the temporal resolution in the stem by 2.