spyrit.core.nnet.Unet.concat

Unet.concat(upsampled: tensor, bypass: tensor) → tensor

Concatenates two tensors along the channel dimension.

This function is useful in the ascending branch of the U-Net model.

Args:

upsampled (torch.tensor): The upsampled tensor, which has been processed by the ascending part of the U-Net model.

bypass (torch.tensor): The tensor from the descending part of the U-Net model, which is concatenated with the upsampled tensor.

Returns:

torch.tensor: The concatenated tensor [upsampled, bypass].