TensorValue Class Reference

Detailed Description

The interface for working with tensor values (parameters, buffers, etc.) managed by models.

Public Member Functions
None	requires_grad_ (self, bool req=True)
None	set_ (self, tensors.Tensor arg0)
tensors.Tensor	tensor (self)
torch.Tensor	torch (self)
torch.Tensor	grad (self)
bool	requires_grad (self)

Member Function Documentation

grad()

torch.Tensor grad

(

self

)

Retrieve the accumulated vector-Jacobian product after a backward propagation.

requires_grad()

bool requires_grad

(

self

)

Value of the boolean requires_grad property of the underlying tensor.

requires_grad_()

None requires_grad_	(		self,
		bool	req = True )

Set the requires_grad property of the underlying tensor.

set_()

None set_	(		self,
		tensors.Tensor	arg0 )

Modify the underlying tensor data.

tensor()

tensors.Tensor tensor

(

self

)

Convert this to a tensors.Tensor.

This conversion takes ownership of the tensor, and so any modification to the
returned tensors.Tensor does not affect the original tensors.TensorValue. Use
tensors.TensorValue.set_ instead to modify the tensor value.

torch()

torch.Tensor torch

(

self

)

Convert this to a torch.Tensor.

This conversion takes ownership of the tensor, and so any modification to the
returned torch.Tensor does not affect the original tensors.TensorValue. Use
tensors.TensorValue.set_ instead to modify the tensor value.