TypeError: unsupported operand type(s) for +: 'NoneType' and 'str'

Warning: Torch did not find available GPUs on this system.
If your intention is to cross-compile, this is not an error.
By default, Apex will cross-compile for Pascal (compute capabilities 6.0, 6.1, 6.2),
Volta (compute capability 7.0), Turing (compute capability 7.5),
and, if the CUDA version is >= 11.0, Ampere (compute capability 8.0).
If you wish to cross-compile for a single specific architecture,
export TORCH_CUDA_ARCH_LIST="compute capability" before running setup.py.

Traceback (most recent call last):
File "setup.py", line 35, in
_, bare_metal_major, _ = get_cuda_bare_metal_version(cpp_extension.CUDA_HOME)
File "setup.py", line 14, in get_cuda_bare_metal_version
raw_output = subprocess.check_output([cuda_dir + "/bin/nvcc", "-V"], universal_newlines=True)
TypeError: unsupported operand type(s) for +: 'NoneType' and 'str'

Start in setup.py, especially get_cuda_bare_metal_version and its use of cpp_extension.CUDA_HOME. Reproduce the setup command on a system where CUDA_HOME is unavailable, then determine the expected handling for that condition. Done should be confirmed by ensuring setup no longer raises this NoneType-plus-string TypeError in that environment.