reference/generated/tensordict.TensorDictBase.to_struct_array

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.to_struct_array

TensorDictBase.to_struct_array() → ndarray

Converts the tensordict to a NumPy structured array, with one field per entry.

The shape of the array is the batch size. A nested tensordict becomes a nested structured field, and the feature dimensions of a tensor become the shape of its field. String values become "U" fields of at least 10 characters. Unlike from_struct_array(), this method copies the data: the array does not share memory with the tensordict.

Returns:

A np.ndarray with a structured dtype.

See also

from_struct_array(), which converts back, and numpy() to convert to a dict of NumPy arrays.

Examples

>>> import torch
>>> from tensordict import TensorDict
>>> td = TensorDict({'a': torch.tensor([1, 2, 3]), 'b': torch.tensor([4.0, 5.0, 6.0])}, batch_size=[3])
>>> arr = td.to_struct_array()
>>> print(arr)
[(1, 4.) (2, 5.) (3, 6.)]

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