towhee.models.shunted_transformer.utilsΒΆ
Classes
Modified Attention. |
|
Block in Shunted Transformer |
|
The patch embedding layer at stage 0. |
|
Modified Mlp containing depthwise convolution, based on a standard Mlp in ViT. |
|
Overlap patch embedding layer based on a standard patch embedding layer in ViT |