OnePose icon indicating copy to clipboard operation
OnePose copied to clipboard

Published 20 hours ago • verified publisher icon zju3dv

The training loss issue

Open lgq-gpu opened this issue 10 months ago • 0 comments

Hello, first of all, thank you for sharing your code. I have a question: I'm training your code using an RTX 3060 GPU, with 40 epochs, batch_size=8, and haven't changed anything else. Currently, my loss has been consistently above 0.1, approximately 0.133. Is this normal?

lgq-gpu avatar Apr 17 '24 08:04 lgq-gpu