When using type annotations in Python, what base type should we use for an argument that can either be a list, tuple, torch.Tensor, numpy.ndarray, etc.?
I would have expected the following to return true, since torch.Tensor implements both __len__ and __getitem__ methods:
from collections.abc import Sequence
import torch
a = torch.tensor((3, 4, 5.0))
isinstance(a, Sequence)
>>> False