Why the loss is too large in the training？Is it normal？

JOP-Lee / READ

AAAI2023，implementation of "READ: Large-Scale Neural Scene Rendering for Autonomous Driving", the experimental results are significantly better than Nerf-based methods

https://github.com/JOP-Lee/READ-Large-Scale-Neural-Scene-Rendering-for-Autonomous-Driving

GNU General Public License v2.0

447 stars 55 forks source link

Why the loss is too large in the training？Is it normal？ #25

Closed AIBUWAN closed 1 year ago

AIBUWAN commented 1 year ago

2023-02-22 14-55-56 的屏幕截图

JOP-Lee commented 1 year ago

Yes, this is a normal phenomenon because it is a VGG loss.