def transpose(a, perm=None, name=None):
r"""Permute the dimensions of input.
Examples:
```python
# Provide the permutation for all axes
x = tf.ones(shape=(2, 3, 4))
print(tf.transpose(x, (0, 2, 1)).shape) # (2, 4, 3)
# Or dimensions will be simply inverse
print(tf.transpose(x).shape) # (4, 3, 2)
```
Parameters
----------
a : dragon.Tensor
The input tensor.
perm : Sequence[Union[int, dragon.Tensor]]
The output permutation.
name : str, optional
The operation name.
Returns
-------
dragon.Tensor
The output tensor.
"""
return array_ops.transpose(a, perm=perm, name=name)
def transpose(self, perm=None, copy=True):
"""Return a tensor with permuted axes.
Parameters
----------
perm : Union[Sequence[int], dragon.Tensor]], optional
The output permutation.
copy : bool, optional, default=True
Return a new tensor or transpose in-place.
Returns
-------
dragon.Tensor
The output tensor.
See Also
--------
`dragon.transpose(...)`_
"""
return array_ops.transpose(self, perm=perm, copy=copy)
def forward(self, inputs, **kwargs):
return array_ops.transpose(inputs, perm=self.perm)
def call(self, inputs):
return array_ops.transpose(inputs, perm=[0] + self.dims)
def __call__(self, bottom):
return array_ops.transpose(bottom, **self.arguments)
def call(self, inputs):
if self.data_format == 'channels_first':
perm = [0] + [i for i in range(2, len(inputs.shape))] + [1]
inputs = array_ops.transpose(inputs, perm=perm)
return array_ops.flatten(inputs, keep_axes=2)
def __call__(self, bottom):
return array_ops.transpose(bottom, **self.call_args)
