reference/generated/tensordict.TensorDictBase.rsub

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.rsub

TensorDictBase.rsub(other: TensorDictBase | Tensor | float, *, alpha: float | None = None, default: str | Tensor | TensorCollection | None = None) → Self

Returns a new tensordict with every leaf, scaled by alpha, subtracted from other.

other - td calls this method.

\[\text{out}_i = \text{other}_i - \text{alpha} \times \text{self}_i\]

Parameters:

other (TensorDictBase, Tensor or Number) – the value from which the leaves are subtracted. A tensordict is paired with the tensordict key by key. A tensor is aligned with the batch dimensions of each leaf and expanded over its feature dimensions. Batch sizes that differ are broadcast together.

Keyword Arguments:

• alpha (Number, optional) – the multiplier for the leaves. Defaults to None, which subtracts them as they are.

• default (Tensor or str, optional) – what to do with the keys that only one of the two tensordicts has. None requires the same keys and raises a KeyError otherwise. "intersection" computes the keys that both have and drops the others. A tensor stands in for the missing entry on either side, and the result has the keys of both. Defaults to None.

Examples

>>> import torch
>>> from tensordict import TensorDict
>>> td = TensorDict(a=torch.full((2,), 10.0), batch_size=[2])
>>> other = TensorDict(a=torch.ones(2), batch_size=[2])
>>> td.rsub(other)["a"]
tensor([-9., -9.])
>>> td.rsub(other, alpha=2)["a"]  # 1 - 2 * 10
tensor([-19., -19.])
>>> td.rsub(torch.tensor([1.0, 2.0]))["a"]
tensor([-9., -8.])

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