running Evaluations without GPU and CUDA

[apathros]

Hi Thank you for providing the code,

I am trying to run the evaluation's on Windows without GPU and CUDA support. when i run the the python file

python run_metrics.py --models_list weights/eval_list --output metric_results.csv --device=cpu

Traceback (most recent call last):
File "run_metrics.py", line 12, in
from eval.ppl import compute_ppl
File "C:\Users\Downloads\Test\eval_init.py", line 1, in
from .evaluation import Evaluator
File "C:\Users\Downloads\Test\eval\evaluation.py", line 7, in
from run_metrics import get_vgg_features, make_eval_images
File "C:\Users\Downloads\Test\run_metrics.py", line 13, in
from training.networks.stylegan2 import Generator
File "C:\Users\Downloads\Test\training\networks_init_.py", line 2, in
from .misc import *
File "C:\Users\Downloads\Test\training\networks\misc.py", line 1, in
from . import stylegan2
File "C:\Users\Downloads\Test\training\networks\stylegan2.py", line 8, in
from .op import FusedLeakyReLU, fused_leaky_relu, upfirdn2d
File "C:\Users\Downloads\Test\training\networks\op_init_.py", line 1, in
from .fused_act import FusedLeakyReLU, fused_leaky_relu
File "C:\Users\Downloads\Test\training\networks\op\fused_act.py", line 15, in
os.path.join(module_path, "fused_bias_act_kernel.cu"),
File "C:\Users\AppData\Local\Programs\Python\Python36\lib\site-packages\torch\utils\cpp_extension.py", line 1091, in load
keep_intermediates=\AppData\Local\Programs\Python\Python36\lib\site-packages\torch\utils\cpp_extension.py", line 1302, in _jit_compile
is_standalone=is_standalone)
File "C:\Users\AppData\Local\Programs\Python\Python36\lib\site-packages\torch\utils\cpp_extension.py", line 1385, in _write_ninja_file_and_build_library
is_standalone)
File "C:\Users\AppData\Local\Programs\Python\Python36\lib\site-packages\torch\utils\cpp_extension.py", line 1484, in _prepare_ldflags
extra_ldflags.append(f'/LIBPATH:{_join_cuda_home("lib/x64")}')
File "C:\Users\AppData\Local\Programs\Python\Python36\lib\site-packages\torch\utils\cpp_extension.py", line 1982, in join_cuda_home
raise EnvironmentError('CUDA_HOME environment variable is not set. '
**OSError: CUDA_HOME environment variable is not set. Please set it to your CUDA install root.**

Is CUDA and GPU is mandatory to run the evaluation ?

[PeterWang512]

Yes, the current implementation will require CUDA.