reference/generated/tensordict.TensorDictBase.lerp

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.lerp

TensorDictBase.lerp(end: TensorCollection | Tensor, weight: TensorCollection | Tensor | float) → Self

Returns a new tensordict that interpolates linearly from every leaf (the start) to end.

\[\text{out}_i = \text{start}_i + \text{weight}_i \times (\text{end}_i - \text{start}_i)\]

Parameters:

• end (TensorDictBase or Tensor) – the end points. A tensordict is paired with the tensordict key by key and must have all its keys. A tensor is aligned with the batch dimensions of each leaf and expanded over its feature dimensions.

• weight (TensorDictBase, Tensor or float) – the interpolation weight: a number for all the leaves, or a tensordict or a tensor paired like end. One of end and weight can be a tensordict and the other a tensor. Where a lazy stack meets a tensordict that is not one, a weight tensor that does not broadcast with the batch dims broadcasts with each leaf from the right instead, with a FutureWarning. Starting with TensorDict 0.17, the default will change to raising the error of the other tensordicts.

Examples

>>> import torch
>>> from tensordict import TensorDict
>>> start = TensorDict(x=torch.zeros(2), batch_size=[2])
>>> end = TensorDict(x=torch.full((2,), 10.0), batch_size=[2])
>>> start.lerp(end, 0.25)["x"]
tensor([2.5000, 2.5000])

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