Closed this issue 2 years ago · 1 comments
Previously multigpu training was working, but a bug must have been introduced that caused errors. Most likely culprit is perhaps how the loss function is/was handled
Closed with #91