dgs.utils.state.collate_tensors¶

dgs.utils.state.collate_tensors(batch: list[torch.Tensor], *_args, **_kwargs) → torch.Tensor[source]¶

Collate a batch of tensors into a single one.

Will use torch.cat() if the first dimension has a shape of one, otherwise torch.stack()