ADLab-AutoDrive / BEVFusion

Offical PyTorch implementation of "BEVFusion: A Simple and Robust LiDAR-Camera Fusion Framework"
Apache License 2.0
752 stars 102 forks source link

loss:NAN or loss does not decline #48

Closed quchunyan closed 1 year ago

quchunyan commented 1 year ago

Hello, good job. When i was training the camera branch with nuscenes dataset, two problems occurred: (1)loss didn't decline (2)when i use img_depth, loss became nan after serveral iters.

HaoZhang16613 commented 1 year ago

@quchunyan Hi~ Is the problem solved? I had the same problem, loss became nan after serveral iters with img_depth. I've already tried some solutions, like adding img_depth loss after a few training epoches and setting img_depth loss to zero when it is abnormal ... Unfortunately, all solutions didn't work.