microsoft / Recursive-Cascaded-Networks

[ICCV 2019] Recursive Cascaded Networks for Unsupervised Medical Image Registration
https://arxiv.org/abs/1907.12353
MIT License
361 stars 87 forks source link

loss function is nan while training #72

Open LocPham263 opened 2 years ago

LocPham263 commented 2 years ago

Hi, I tried multiple options for the number of cascades. For small number of cascades, the training process normally finished at roughly 2e-01 loss. However, for bigger cascades option, the loss function displayed nan value whenever the training process < 1e-01 loss. What should I do now? Thank you in advance.