reference/generated/tensordict.TensorDictBase.get_non_tensor

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.get_non_tensor

TensorDictBase.get_non_tensor(key: NestedKey, default=NO_DEFAULT)

Gets a non-tensor value, if it exists, or default if the non-tensor value is not found.

This method is robust to tensor/TensorDict values, meaning that if the value gathered is a regular tensor it will be returned too (although this method comes with some overhead and should not be used out of its natural scope).

See set_non_tensor() for more information on how to set non-tensor values in a tensordict.

Parameters:

• key (NestedKey) – the location of the NonTensorData object.

• default (Any, optional) – the value to be returned if the key cannot be found.

Returns: the content of the tensordict.tensorclass.NonTensorData,

or the entry corresponding to the key if it isn’t a tensordict.tensorclass.NonTensorData (or default if the entry cannot be found).

Examples

>>> from tensordict import TensorDict
>>> data = TensorDict({}, batch_size=[])
>>> data = data.set_non_tensor(("nested", "the string"), "a string!")
>>> assert data.get_non_tensor(("nested", "the string")) == "a string!"
>>> # regular `get` works but returns a NonTensorData object
>>> data.get(("nested", "the string"))
NonTensorData(data=a string!, batch_size=torch.Size([]), device=None)

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