RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

```
...
File "/home/Foo/miniforge3/envs/pytor/lib/python3.6/site-packages/torch/nn/functional.py", line 1923, in batch_norm
    training, momentum, eps, torch.backends.cudnn.enabled
RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED
```

Got this error when trying to run the following command the README.md:

```
python benchmark.py --arch resnet101 --device cuda:0
```