espnet2.gan_tts.vits.flow.Transpose

source

class espnet2.gan_tts.vits.flow.Transpose(dim1: int, dim2: int)

Bases: Module

Transpose module for torch.nn.Sequential().

This module transposes the specified dimensions of the input tensor.

dim1

The first dimension to transpose.

• Type: int

dim2

The second dimension to transpose.

• Type: int

• Parameters:

  • dim1 (int) – The first dimension to be transposed.
  • dim2 (int) – The second dimension to be transposed.

• Returns: The transposed tensor.

• Return type: Tensor

Examples

>>> transpose = Transpose(1, 2)
>>> input_tensor = torch.randn(2, 3, 4)  # Shape: (B, C, T)
>>> output_tensor = transpose(input_tensor)
>>> print(output_tensor.shape)  # Shape: (B, T, C)

Initialize Transpose module.

forward(x: Tensor) → Tensor

Transpose module for torch.nn.Sequential().