reference/generated/tensordict.TensorDictBase.refine_names

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.refine_names

TensorDictBase.refine_names(*names) → Self

Names the unnamed batch dimensions, and checks the named ones, in place.

Each entry of names goes with one batch dimension. An unnamed dimension takes the new name, or stays unnamed for None. A named dimension must get its own name. names can hold one ..., which keeps the names of the dimensions that it covers. This follows torch.Tensor.refine_names().

Parameters:

*names (str or None) – one name per batch dimension, or fewer with one ....

Returns:

self.

Raises:

RuntimeError – if a named dimension gets another name, or None.

Examples

>>> from tensordict import TensorDict
>>> td = TensorDict(batch_size=[3, 4, 5], names=["a", None, None])
>>> td.refine_names("a", None, "c").names
['a', None, 'c']
>>> td.refine_names(..., "b", "c").names
['a', 'b', 'c']

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