reference/generated/tensordict.TensorDictBase.data

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.data

property TensorDictBase.data: Any

the same storage, without autograd.

The leaves of the result share storage with the leaves of the tensordict, and do not require gradients. An in-place change to the result changes the tensordict.

Assigning td.data = other rebinds the .data of each leaf of the tensordict to the matching leaf of other, as torch.Tensor.data does. The leaves then share storage with other: memory-mapped or shared-memory leaves stop writing to their file or shared memory. other and each of its nested tensordicts must have the batch size of the matching tensordict, and each key of other must be in the tensordict.

Examples

>>> import torch
>>> from tensordict import TensorDict
>>> td = TensorDict(x=torch.zeros(3, requires_grad=True), batch_size=[3])
>>> data = td.data
>>> data["x"].requires_grad, data["x"].data_ptr() == td["x"].data_ptr()
(False, True)

Type:

A tensordict of the .data of every leaf

Every tensordict class and every tensorclass has this method. For the other methods, see Methods by task.