Code get stuck with NeRFACC
Closed this issue · 6 comments
The code in main.py is stuck for 20minutes+ at line 258 in lowrank_model.py(self.occupancy_grid.sampling).
All the environment setup commands run successfully. (I used pip install nerfacc here.)
When I switched to pip install git+https://github.com/nerfstudio-project/nerfacc.git, the code worked well. But I couldn't get into the sampling function for debugging.
We didn't get stuck there, but I assume it is likely a build error. Anyway, it makes no sense you can't debug. Did you try debug using import pdb or debug line by line?
Yes, I debugged the code line by line using VSCode. I still couldn't step into self.occupancy_grid.sampling, but I can step into sigma_fn and rgb_sigma_fn, which looks strange. How can I uninstall Nerfacc completely？ Maybe I'll try to set up a new Conda environment to test it.
I found this tip: “The easiest way is to install from PyPI. In this way it will build the CUDA code on the first run (JIT).” from the nerfacc repo. So, do I need to wait for a long time for the first run of your code?
Yes. It requires some time to build the first time you run nerfacc, but there should be tips in the command line output informing nerfacc is building.
It looks like this and takes only minutes. But anyway, I just set up an env with command in the readme and it runs successfully on a machine with 2080Ti and AMD CPU—most package latest version.
Emmm, I remember seeing it the first time I set up the environment, but I pressed Ctrl+C to cancel the code execution back then, and I have been having this issue ever since. (That's the problem)
Update: Find similar issue in nerfacc repo here. Problem Solved! Thanks for your patience！