0

I tried to run a PPO algorithm from the stable-baselines3 library on a basic gym environment on my local CPU, however I get the following RuntimeError:

RuntimeError: Calling torch.geqrf on a CPU tensor requires compiling PyTorch with LAPACK. Please use PyTorch built with LAPACK support.

I'm using a conda environment on a Windows machine, with following installations:

pytorch 1.12.1 cpu_py39h5e1f01c_0

lapack 3.9.0 netlib conda-forge

Since I'm quite new to Python I have no clue how to resolve this issue and web search didn't give any proper instructions on that specific problem.

I tried uninstalling PyTorch in my anaconda prompt, however this would remove a lot of packages which scared me of breaking something. Hence, I'm lost what else to do in order to build PyTorch with Lapack support...

Any help would be appreciated, Cheers.

n.dresemann
  • 1
  • 3
  • 1
    well, you shouldn't be scared of trying to break something in anaconda environment during experiments -- that's what they actually are for. I suspect your lapack dependency comes with torchaudio or torchvision packets that you do not install. Try this in a fresh environment: `conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cpuonly -c pytorch`. [Source:](https://pytorch.org/get-started/previous-versions/#v1121) – gehirndienst Dec 06 '22 at 08:44
  • I'm having the exact same problem, did you find a solution yet? – Albert Schrotenboer Jan 03 '23 at 12:19
  • I deinstalled pytorch from my environment and installed it for CPU only (following instruction from https://pytorch.org/get-started/locally/), then it worked out for me actually. – n.dresemann Jan 03 '23 at 15:00

0 Answers0