The feature, motivation and pitch LayerNorm starts to be applied to image data on per-channel basis (e.g. in ConvNeXt model). torch.nn.
確定! 回上一頁