MacOS running sh make.sh

[Willt1128]

After following the installation instructions then doing
cd Semantic-SAM
git clone https://github.com/facebookresearch/Mask2Former.git
cd Mask2Former/mask2former/modeling/pixel_decoder/ops

I run sh make.sh and initially get an error saying "ModuleNotFoundError: No module named 'torch.cuda'". I then installed the latest version of Cuda available for Mac (Cuda 10), set my Cuda home to its installation to its location (export CUDA_HOME=/usr/local/cuda), and run again. This returns "NotImplementedError: No CUDA runtime is found ..." Setting FORCE_CUDA doesnt solve the issue.

I'm using MacOS 14.1 (graphics: Intel Iris Plus Graphics 1536 MB) and torch 1.13.1 and running jupyter within a new Conda environment running python 3.9. I've also tried python 3.10 and 3.12. Also note that I removed scann from requirements.txt because I got the following error upon installing requirements:

ERROR: Could not find a version that satisfies the requirement scann (from versions: none)
ERROR: No matching distribution found for scann

Is there anything else I can do to successfully run sh make.sh so that I can use Semantic-SEM? Is Semantic-SEM supported for my hardware/OS? Thanks for the help -- I'm new to this process.

[TAUIL-Abd-Elilah]

Try FORCE_CUDA=1 sh make.sh