Flatten#

class myoverse.transforms.Flatten(start_dim=0, end_dim=-1, **kwargs)[source]#

Flatten dimensions of a tensor.

Parameters:
  • start_dim (int) – First dimension to flatten.

  • end_dim (int) – Last dimension to flatten.

Examples

>>> x = torch.randn(8, 8, 2048, names=('row', 'col', 'time'))
>>> flatten = Flatten(start_dim=0, end_dim=1)
>>> y = flatten(x)  # Shape: (64, 2048)

Methods

__init__([start_dim, end_dim])

_apply(x)

Apply the transform.