thing.py

import torch as th
import torch.utils.data as th_data


def main():
    x = th.ones([30, 3, 3], requires_grad=True)
    y = x * 2
    dl = th_data.DataLoader(y)

    batch = next(iter(dl))
    print(f"batch: {batch}")
    batch2 = th.as_tensor(batch)
    print(f"batch2: {batch2}")
    print(f"batch is batch2: {batch is batch2}")


"""
batch: tensor([[[2., 2., 2.],
         [2., 2., 2.],
         [2., 2., 2.]]], grad_fn=<StackBackward>)
batch2: tensor([[[2., 2., 2.],
         [2., 2., 2.],
         [2., 2., 2.]]], grad_fn=<StackBackward>)
batch is batch2: True
"""


if __name__ == "__main__":
    main()