Xiaoyang Wu

Hi, which config (model), you were testing?

Hi, the discussion here (https://discuss.pytorch.org/t/pytorch-cannot-allocate-memory/134754/4) might be helpful. Maybe the default num_worker is too large for your machine.

> Thank you for your response. > > Put num_workers=0 and problem solved. (Though it's running slow. I think no need to set it to 0, just reduce it. According...

``` ModuleNotFoundError: No module named 'pointgroup_ops' ``` => Make sure Pointops is successfully built and install ``` RuntimeError: Error compiling objects for extension ``` => Make sure cuda is successfully...

> Can confirm this is an issue for me too. Installed `pointops` - PyTorch matches cuda too. > > I get this when doing `pip install .` in the `pointgroup_ops`...

> How do I resolve this error when I execute the following command > > cd libs/pointops python setup.py install > > Traceback (most recent call last): File "F:\Python_project\Pointcept-main\libs\pointops2\setup.py", line...

Hi, data workers are on CPU, each GPU should only contain one process.

Follow here (https://github.com/Pointcept/Pointcept?tab=readme-ov-file#pointgroup) to install pointgroup_ops, but pointgroup ops is optional.

> 有人解决ModuleNotFoundError: No module named 'pointgroup_ops' 这个问题了吗 (pointcept) sxx-205@sxx-205:/data/Pointcept-main$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2021 NVIDIA Corporation Built on Sun_Mar_21_19:15:46_PDT_2021 Cuda compilation tools, release 11.3,...

Hi, it has been a long time since I last time trained PTv1 or PTv2 on ShapeNet Part Segmentation. I only remember that object-level datasets need careful tuning on the...