MCPcopy Create free account
hub / github.com/PaddlePaddle/Paddle / clone

Function clone

python/paddle/tensor/creation.py:3692–3725  ·  view source on GitHub ↗

Returns a copy of input Tensor. It will always have a Tensor copy. In addition, This function is derivable, so gradients will flow back from the output to input. Parameters: x (Tensor): The input Tensor. Alias: ``input``. name(str|None, optional): For detai

(x: paddle.Tensor, name: str | None = None)

Source from the content-addressed store, hash-verified

3690
3691@param_one_alias(['x', 'input'])
3692def clone(x: paddle.Tensor, name: str | None = None) -> paddle.Tensor:
3693 """
3694 Returns a copy of input Tensor. It will always have a Tensor copy.
3695
3696 In addition, This function is derivable, so gradients will flow back from the output to input.
3697
3698 Parameters:
3699 x (Tensor): The input Tensor.
3700 Alias: ``input``.
3701 name(str|None, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None.
3702
3703 Returns:
3704 Tensor, A Tensor copied from ``input``.
3705
3706 Examples:
3707 .. code-block:: pycon
3708
3709 >>> import paddle
3710 >>> import numpy as np
3711
3712 >>> x = paddle.ones([2])
3713 >>> x.stop_gradient = False
3714 >>> x.retain_grads()
3715 >>> clone_x = paddle.clone(x)
3716 >>> clone_x.retain_grads()
3717
3718 >>> y = clone_x**3
3719 >>> y.backward()
3720 >>> print(clone_x.grad.numpy()) # type: ignore
3721 [3. 3.]
3722 >>> print(x.grad.numpy()) # type: ignore
3723 [3. 3.]
3724 """
3725 return x.clone()
3726
3727
3728# NOTE(zhiqiu): not public

Callers 1

graphsafe_get_stateMethod · 0.50

Calls 1

cloneMethod · 0.45

Tested by

no test coverage detected