Training error when --voxel_size = 4.0

[Xuduoteng]

Hi, thank you for your great work！
I found some problems when testing planarrecon with the scannet dataset, could you give me some advice?
when I set --voxel_size=0.04 to generate gt and train from epoch0 it works perfectly, but when I use voxel_size=4.0 the process of generating gt works, but the following error occurs during training，

I tried adding distance_loss=0 to the else branch，

A cuda error occurred during the second stage of training，

[ymingxie]

Hi, sorry for the late response.

voxel_size=4 means the size of voxel is 4m x 4m x 4m. it's unreasonable for indoor scenes. Are you sure it's what you need?