Trying to use autograd profiler to get some profiling info but when I do a print, the system just hangs…
Here’s what I’m doing
with torch.autograd.profiler.profile(True, False) as prof:
l2dist, labels, adv_img, scale_const = cw_l2.generate(data, orig, target)
print(prof)
I’m using PyTorch 0.3.1 on Windows 10, Cuda 9.1 using peterjc’s Pytorch to Windows port. Has anyone tried profiling on windows?