How do I index tensors in PyTorch?

How do I index tensors in PyTorch?

In PyTorch, you can index tensors using square brackets. For example, if you have a tensor x, you can access elements like x[0] for the first element.

Can I use slicing to get a range of values?

Yes, you can use slicing! For instance, x[1:3] will give you the elements from index 1 to 2.

What if I want to index with a list of indices?

You can use torch.index_select() for that. For example, torch.index_select(x, 0, torch.tensor([0, 2])) will select the first and third elements.

Is there a way to index with a boolean mask?

Absolutely! You can create a boolean mask and use it to index. For example, x[mask] will return elements where the mask is True.