Skip to content

RuntimeError: CUDA error: invalid configuration argument #8

Description

I kept running into this error when training skin for frame 1
( running the guitar dataset with undistorted images.)

Output folder: /media/dual_output/guitar_test/ [06/11 16:59:10]
Reading camera 81/81 [06/11 16:59:10]
Loading Training Cameras [06/11 16:59:10]
Loading Test Cameras [06/11 16:59:13]
[16:59:13] loading motion from control point init.py:84 [06/11 16:59:13]
[16:59:14] Loading ply file : /media/dual_output/guitar_test/ckt/point_cloud_0.ply gaussian_model.py:290 [06/11 16:59:14]
number of gaussians: 215771 [06/11 16:59:14]
Training progress: 0%| | 0/15000 [00:00<?, ?it/s]Traceback (most recent call last):
File "/media/DualGS/train.py", line 383, in
training_skin(lp.extract(args), op.extract(args), pp.extract(args), lossp2, args.test_iterations, args.debug_from, is_start_frame, frame_idx, args=args)
File "/media/DualGS/train.py", line 235, in training_skin
loss.backward()
File "/home/anaconda3/envs/dualgs/lib/python3.10/site-packages/torch/_tensor.py", line 492, in backward
torch.autograd.backward(
File "/home/anaconda3/envs/dualgs/lib/python3.10/site-packages/torch/autograd/init.py", line 251, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
RuntimeError: CUDA error: invalid configuration argument
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Training progress: 0%| | 0/15000 [00:01<?, ?it/s]

Any possible cause or solution?
Any help will be very much appreciated!! thanks!!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions