An error in temporal sampling?

Thanks for your wonderful code!
But I am confused by an implemental detail that does not match your paper.
https://github.com/facebookresearch/SlowFast/blob/dde28ccbdb3f0f323398124d639ee2538dbd9de2/slowfast/datasets/decoder.py#L25
`torch.linspace(start, end, steps=100, out=None, dtype=None, layout=torch.strided, device=None, requires_grad=False) → Tensor`
means

> Returns a one-dimensional tensor of steps equally spaced points between start and end.

For example,
if the stride is 16, the number of frames is 4, start_idx is 7. So we can get end_idx = 7 + 4 * 16 - 1 = 70. Then we call temporal_sampling() function, the result is as below.
`torch.linspace(7, 70, 4)
tensor([ 7., 28., 49., 70.])
`
It does not satisfy stride 16 but comes to 21!
Is there any mistake when calculating end_idx or something else?