reference/generated/tensordict.TensorDictBase.from_dict_instance

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.from_dict_instance

TensorDictBase.from_dict_instance(input_dict, *, auto_batch_size: bool | None = None, batch_size=None, device=None, batch_dims=None, names=None)

Instance method version of from_dict().

Unlike from_dict(), this method will attempt to keep the tensordict types within the existing tree (for any existing leaf).

Examples

>>> from tensordict import TensorDict, tensorclass
>>> import torch
>>>
>>> @tensorclass
... class MyClass:
...     x: torch.Tensor
...     y: int
>>>
>>> td = TensorDict({"a": torch.randn(()), "b": MyClass(x=torch.zeros(()), y=1)})
>>> print(td.from_dict_instance(td.to_dict()))
TensorDict(
    fields={
        a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False),
        b: MyClass(
            x=Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False),
            y=Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False),
            batch_size=torch.Size([]),
            device=None,
            is_shared=False)},
    batch_size=torch.Size([]),
    device=None,
    is_shared=False)
>>> print(td.from_dict(td.to_dict()))
TensorDict(
    fields={
        a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False),
        b: TensorDict(
            fields={
                x: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False),
                y: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)},
            batch_size=torch.Size([]),
            device=None,
            is_shared=False)},
    batch_size=torch.Size([]),
    device=None,
    is_shared=False)

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