I am trying to measure CUDA usage on MNIST example on PyTorch 0.4.0 by following commands, but it failed. How to avoid this issue? $ python -m torch.utils.bottleneck main.py Error message is follows. === Taceback (most recent call last): File "/opt/conda/lib/python3.6/runpy.py", line 193, in _…

Could you check with 'num_workers': 0 in main.py. [image] Issue: [utils.bottleneck] Bottleneck crashes with multi-threaded data loader opened by fmassa on 2018-04-05 torch.utils.bottleneck doesn't work properly when the code contains a data loader that uses more than 0 thre…

Thank you It works fine, when I change the num_workers from 1 to 0.

Cannot measure CUDA usage on MNIST example

Deepali (Deepali Patel) July 25, 2018, 12:48pm 2

Could you check with 'num_workers': 0 in main.py.