VITA-Group / FSGS

[ECCV 2024]"FSGS: Real-Time Few-Shot View Synthesis using Gaussian Splatting", Zehao Zhu*, Zhiwen Fan*, Yifan Jiang, Zhangyang Wang
Other
418 stars 34 forks source link

RuntimeError: tabulate: failed to synchronize: cudaErrorInvalidConfiguration: invalid configuration argument #56

Closed Gurutva closed 3 months ago

Gurutva commented 3 months ago

Hi , thanks in advance.

I could successfully create the env.

while running train.py file, I'm getting following error. How can I resolve this , if anyone has any idea ?

    (FSGS) user@user-Z590-D:~/gurutva/FSGS$ python train.py  --source_path dataset/nerf_llff_data/horns --model_path output/horns --eval  --n_views 3 --sample_pseudo_interval 1
    Using cache found in /home/user/.cache/torch/hub/intel-isl_MiDaS_master
    /home/user/miniconda3/envs/FSGS/lib/python3.8/site-packages/timm/models/_factory.py:117: UserWarning: Mapping deprecated model name vit_base_resnet50_384 to current vit_base_r50_s16_384.orig_in21k_ft_in1k.
      model = create_fn(
    Using cache found in /home/user/.cache/torch/hub/intel-isl_MiDaS_master
    [1000, 2000, 3000, 5000, 10000]
    Optimizing output/horns
    Output folder: output/horns [07/08 22:07:47]
    Tensorboard not available: not logging progress [07/08 22:07:47]
    Reading camera 62/62 [07/08 22:07:47]
    6.323975610733033 cameras_extent [07/08 22:07:47]
    Loading Training Cameras [07/08 22:07:47]
    3it [00:01,  1.87it/s]
    Loading Test Cameras [07/08 22:07:48]
    8it [00:00, 18.90it/s]
    Number of points at initialisation :  0 [07/08 22:07:52]
    Traceback (most recent call last):
      File "train.py", line 279, in <module>
        training(lp.extract(args), op.extract(args), pp.extract(args), args)
      File "train.py", line 44, in training
        scene = Scene(args, gaussians, shuffle=False)
      File "/home/user/gurutva/FSGS/scene/__init__.py", line 103, in __init__
        self.gaussians.create_from_pcd(scene_info.point_cloud, self.cameras_extent)
      File "/home/user/gurutva/FSGS/scene/gaussian_model.py", line 144, in create_from_pcd
        dist2 = torch.clamp_min(distCUDA2(fused_point_cloud)[0], 0.0000001)
    RuntimeError: tabulate: failed to synchronize: cudaErrorInvalidConfiguration: invalid configuration argument