TensorDictBase.items#

TensorDictBase.items(include_nested: bool = False, leaves_only: bool = False, is_leaf=None, *, sort: bool = False) → Iterator[tuple[str, Tensor | TensorCollection]]#

Returns a generator of the (key, value) pairs of the tensordict.

By default, only the top-level entries are returned, and a nested tensordict is returned as one value. Non-tensor data comes as a NonTensorData, not as the object that td[key] returns. See Keys, nested keys, and locking for nested keys and leaves.

Parameters:

include_nested (bool, optional) – if True, the entries of the nested tensordicts are also returned, under nested keys (tuples of strings). Defaults to False.
leaves_only (bool, optional) – if True, only the leaves are returned, and nested tensordicts are skipped. Defaults to False.
is_leaf (Callable, optional) – a function that takes a class and returns True if values of that class count as leaves. It filters the result only when leaves_only=True. It does not stop the recursion: with include_nested=True, a nested tensordict is still searched even if is_leaf returns True for it. Defaults to None, which uses default_is_leaf(): tensordicts, tensorclasses and non-tensor data are not leaves. Pass is_leaf_nontensor() to count non-tensor data as leaves.

Keyword Arguments:

sort (bool, optional) – if True, the keys are sorted. Nested keys are sorted by their joined name: ("a", "key") sorts as "a.key". Sorting takes extra time on large tensordicts. Defaults to False.

Examples

>>> import torch
>>> from tensordict import TensorDict
>>> td = TensorDict(
...     a=torch.zeros(2),
...     nested=TensorDict(b=torch.ones(2, 3), batch_size=[2]),
...     batch_size=[2],
... )
>>> for key, value in td.items(include_nested=True, leaves_only=True):
...     print(key, value.shape)
a torch.Size([2])
('nested', 'b') torch.Size([2, 3])