reference/generated/tensordict.NestedKey

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

NestedKey

class tensordict.NestedKey

Bases: object

An abstract class for nested keys.

Nested keys are the generic key type accepted by TensorDict.

A nested key is either a string or a non-empty tuple of NestedKeys instances. td["a", "b"] and td[("a", "b")] read the entry "b" of the nested tensordict "a". Nested tuples are flattened: ("a", ("b",)) is the same key as ("a", "b"). See Keys, nested keys, and locking.

The NestedKey class supports instance checks. Use it in type hints and in isinstance checks.

Examples

>>> from tensordict import NestedKey, TensorDict
>>> isinstance("a", NestedKey), isinstance(("a", ("b",)), NestedKey)
(True, True)
>>> isinstance(("a", 1), NestedKey), isinstance((), NestedKey)
(False, False)
>>> td = TensorDict({"a": {"b": 0}})
>>> td["a", "b"] is td[("a", ("b",))]
True

Tutorials that use NestedKey

Building tensordicts from streams

Building tensordicts from streams