jinqiupeter
commented
11 months ago You probably use Lovelace/Hopper GPUs like RTX 40 series or H100/H800, they are not compatible with the CUDA version being used.
Try the previous generation like RTX 3090.
ThetaRgo
h-oneohone
[INFO] ===== extract face landmarks from data/obama/ori_imgs ===== 0%| | 1/8000 [00:07<17:43:43, 7.98s/it] Traceback (most recent call last): File "data_utils/process.py", line 399, in
extract_landmarks(ori_imgs_dir)
File "data_utils/process.py", line 61, in extract_landmarks
preds = fa.get_landmarks(input)
File "/home/xiaoduo/miniconda3/envs/pytorch3d/lib/python3.8/site-packages/face_alignment/api.py", line 113, in get_landmarks
return self.get_landmarks_from_image(image_or_path, detected_faces, return_bboxes, return_landmark_score)
File "/home/xiaoduo/miniconda3/envs/pytorch3d/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 28, in decorate_context
return func(*args, *kwargs)
File "/home/xiaoduo/miniconda3/envs/pytorch3d/lib/python3.8/site-packages/face_alignment/api.py", line 168, in get_landmarks_from_image
out = self.face_alignment_net(inp).detach()
File "/home/xiaoduo/miniconda3/envs/pytorch3d/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(input, **kwargs)
RuntimeError: nvrtc: error: invalid value for --gpu-architecture (-arch)
Python 3.8.18