Ok, gotcha, well, thank you very much for your help!
halahup (Stepan Ulyanin) #22
rasbt (Sebastian Raschka) #23
glad that this is solved! (and not a PyTorch bug :))
halahup (Stepan Ulyanin) #24
For people who might have the same experience as me, there is a drop down button on the GPU monitor in windows process monitor where you can select
cuda to show the actual GPU load while training: