from_pytree

class tensordict.from_pytree(pytree, *, batch_size: Optional[Size] = None, auto_batch_size: bool = False, batch_dims: Optional[int] = None)

將 pytree 轉換為 TensorDict 例項。

此方法旨在儘可能保留 pytree 的巢狀結構。

其他非張量鍵將被新增，以跟蹤每個級別的標識，從而提供內建的 pytree 到 tensordict 的雙射轉換 API。

當前接受的類包括列表、元組、命名元組和字典。

注意

對於字典，非 `NestedKey` 鍵被單獨註冊為 `NonTensorData` 例項。

注意

可轉換為張量型別（如 int、float 或 np.ndarray）將被轉換為 torch.Tensor 例項。請注意，此轉換是滿射的：將 tensordict 轉換回 pytree 將無法恢復原始型別。

示例

>>> # Create a pytree with tensor leaves, and one "weird"-looking dict key
>>> class WeirdLookingClass:
...     pass
...
>>> weird_key = WeirdLookingClass()
>>> # Make a pytree with tuple, lists, dict and namedtuple
>>> pytree = (
...     [torch.randint(10, (3,)), torch.zeros(2)],
...     {
...         "tensor": torch.randn(
...             2,
...         ),
...         "td": TensorDict({"one": 1}),
...         weird_key: torch.randint(10, (2,)),
...         "list": [1, 2, 3],
...     },
...     {"named_tuple": TensorDict({"two": torch.ones(1) * 2}).to_namedtuple()},
... )
>>> # Build a TensorDict from that pytree
>>> td = from_pytree(pytree)
>>> # Recover the pytree
>>> pytree_recon = td.to_pytree()
>>> # Check that the leaves match
>>> def check(v1, v2):
>>>     assert (v1 == v2).all()
>>>
>>> torch.utils._pytree.tree_map(check, pytree, pytree_recon)
>>> assert weird_key in pytree_recon[1]