dataset attribute should not be set after SafeDataLoader is initialized

Question

dataset attribute should not be set after SafeDataLoader is initialized

JosephGatto opened this issue 5 years ago · 5 comments

I have my own custom DataSet class. When I wrap it in nc.SafeDataset it runs with no error. When I then run loader = nc.SafeDataLoader(nc.SafeDataset(my_dataset)) it breaks and I get this error: dataset attribute should not be set after SafeDataLoader is initialized. Is this because I am not using a supported pytorch version?

Answer 1 · 2019-11-19T04:42:54.000Z

I'm seeing the same issue on PyTorch 1.3.0. Interested in knowing whether this is due to the version number or something else. I'm also using my own Dataset class.

Answer 2 · 2019-11-19T17:00:02.000Z

I'm seeing the same issue on PyTorch 1.3.0. Interested in knowing whether this is due to the version number or something else. I'm also using my own Dataset class.

Just adding here that I downgraded PyTorch to 1.2.0 and stopped seeing the error mentioned above.

Answer 3 · 2020-05-22T23:53:17.000Z

Commenting the line

self.dataset = _OriginalDataset(self.safe_dataset)

in nonechucks/dataloader.py seems to solve the issue (in PyTorch 1.5.0) in the basic example I tried on. However I assume that it breaks something else even if I don't know what

Answer 4 · 2020-06-22T12:01:54.000Z

any news on fixing it? I still have the same problem with pytorch 1.5.1

Answer 5 · 2020-09-14T12:53:18.000Z

I experience the same problem. I seems that @msamogh abandoned this project development