zhangzc21 / Generalized-One-shot-GAN-Adaptation

59 stars 9 forks source link

RuntimeError: CUDA error: unknown error #7

Closed pql123 closed 1 year ago

pql123 commented 1 year ago

when i run train_OSGA.sh by gpu, i receive this trace, how i fix this?

Traceback (most recent call last): File "generalized_one_shot_adaption.py", line 401, in train_function(config) File "generalized_one_shot_adaption.py", line 343, in train_function domain_adaption.process(ws, images, masks, config) File "generalized_one_shot_adaption.py", line 216, in process config.synthesis_kwargs) File "/home/pql/generalized-one-shot-gan-adaptation/inversion_networks_utils.py", line 126, in synthesis_aux_forward x, img = block(x, img, cur_ws, kwargs) File "/home/pql/anaconda3/envs/test/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl return forward_call(*input, kwargs) File "", line 397, in forward File "/home/pql/anaconda3/envs/test/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl return forward_call(*input, *kwargs) File "", line 292, in forward File "/home/pql/generalized-one-shot-gan-adaptation/torch_utils/misc.py", line 104, in decorator return fn(args, kwargs) File "", line 72, in modulated_conv2d File "/home/pql/generalized-one-shot-gan-adaptation/torch_utils/misc.py", line 104, in decorator return fn(*args, kwargs) File "/home/pql/generalized-one-shot-gan-adaptation/torch_utils/ops/conv2d_resample.py", line 126, in conv2d_resample x = upfirdn2d.upfirdn2d(x=x, f=f, padding=[px0+pxt,px1+pxt,py0+pyt,py1+pyt], gain=up2, flip_filter=flip_filter) File "/home/pql/generalized-one-shot-gan-adaptation/torch_utils/ops/upfirdn2d.py", line 161, in upfirdn2d return _upfirdn2d_cuda(up=up, down=down, padding=padding, flip_filter=flip_filter, gain=gain).apply(x, f) File "/home/pql/generalized-one-shot-gan-adaptation/torch_utils/ops/upfirdn2d.py", line 242, in forward y = _plugin.upfirdn2d(y, f, upx, upy, downx, downy, padx0, padx1, pady0, pady1, flip_filter, gain)

zhangzc21 commented 1 year ago

Hi, it looks like the Cuda operators in StyleGAN2 are not compiled correctly. Please follow the guidance in the official StyleGAN2 repository to reinstall the environment.

pql123 commented 1 year ago

thanks, i will try to reinstall the environment