RuntimeError: Found dtype Long but expected Float

jakubcerveny commented 1 year ago

Everything installed OK, but when trying to run (e.g. the lego dataset), I get the following error shortly after the training and testing data is loaded:

022-08-04 11:01:20.031334: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0
Detected NeRF (Blender) dataset
LOAD DATA ../data/lego/train
100%|██████████| 100/100 [00:01<00:00, 89.15it/s]
 Generating rays, scaling factor 1
Detected NeRF (Blender) dataset
LOAD DATA ../data/lego/test
100%|██████████| 200/200 [00:02<00:00, 90.51it/s]
Traceback (most recent call last):
  File "opt.py", line 285, in <module>
    grid = svox2.SparseGrid(reso=reso_list[reso_id],
  File "/home/jakub/miniconda3/envs/plenoxel/lib/python3.8/site-packages/svox2/svox2.py", line 429, in __init__
    points = torch.addcmul(
RuntimeError: Found dtype Long but expected Float

jakubcerveny commented 1 year ago

This is my launch command:

~/git/svox2/opt $ ./launch.sh lego 0 ../data/lego/ -c configs/syn.json

The dataset comes from nerf_synthetic.zip

doalberto commented 1 year ago

I've got the same problem. How did you solve this issue?

sxyu / svox2

RuntimeError: Found dtype Long but expected Float #85