torch_npu.npu_slice(self, offsets, size) -> Tensor
从张量中提取切片。
>>> input = torch.tensor([[1,2,3,4,5], [6,7,8,9,10]], dtype=torch.float16).to("npu")
>>> offsets = [0, 0]>>> size = [2, 2]
>>> output = torch_npu.npu_slice(input, offsets, size)
>>> output
tensor([[1., 2.],
[6., 7.]], device='npu:0', dtype=torch.float16)