MNIST Download Raises 'UserWarning: The given NumPy array is not writeable'

[JAEarly]

🐛 Bug

When downloading MNIST (using torchvision/datasets/mnist.py), the following UserWarning is raised:

UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors. This means you can write to the underlying (supposedly non-writeable) NumPy array using the tensor. You may want to copy the array to protect its data or make it writeable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at  ..\torch\csrc\utils\tensor_numpy.cpp:180.)
  return torch.from_numpy(parsed.astype(m[2], copy=False)).view(*s)

Changing line 498 in torchvision/datasets/mnist.py from
return torch.from_numpy(parsed.astype(m[2], copy=False)).view(*s)
to
return torch.from_numpy(parsed.astype(m[2])).view(*s)
removes the error.

Checking the Torch documentation (https://pytorch.org/docs/stable/generated/torch.from_numpy.html) isn't too helpful.
There is no record of the copy=False arg.

To Reproduce

Steps to reproduce the behavior:

A simple test of:

from torchvision.datasets import MNIST
MNIST('data/', train=True, download=True)

raises the warning.
Removing copy=False as discussed above then stops the warning from being raised.

Expected behavior

The warning should not be raised.
As the warning is suppressed after this happens, it may not be raised elsewhere when it's actually important.

Environment

Script output:
PyTorch version: 1.9.0+cu102
Is debug build: False
CUDA used to build PyTorch: 10.2
ROCM used to build PyTorch: N/A

OS: Microsoft Windows 10 Enterprise
GCC version: Could not collect
Clang version: Could not collect
CMake version: Could not collect
Libc version: N/A

Python version: 3.8.6rc1 (tags/v3.8.6rc1:08bd63d, Sep 7 2020, 23:10:23) [MSC v.1927 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-10-10.0.19041-SP0
Is CUDA available: True
CUDA runtime version: 10.2.89
GPU models and configuration: GPU 0: GeForce GTX 1650
Nvidia driver version: 462.31
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A

Versions of relevant libraries:
[pip3] numpy==1.21.0
[pip3] torch==1.9.0+cu102
[pip3] torchaudio==0.9.0
[pip3] torchvision==0.10.0+cu102
[conda] Could not collect

cc @pmeier

[NicolasHug]

Thanks for the report @JAEarly

The output array being relatively small (60KB for training data), I think it'd be fine to just copy it as a one-time thing. Would you like to submit a PR?

That being said, I wonder if there's an easy way to get a writable buffer in the first place. I tried, but reading the file with r+b mode instead of rb didn't help, even after trying to manually parsed.setflags(write=True)

[pmeier]

Closed in #4184.