search

NVlabs / imaginaire

NVIDIA's Deep Imagination Team's PyTorch Library
Other
4.02k stars 449 forks source link

AttributeError: module 'torch.distributed' has no attribute 'is_initialized' #86

Closed simon-eda closed 3 years ago

simon-eda commented 3 years ago

Dear All, i have an issue running coco-funit on win10 and it would be great if could hepl to fix the issue. many thanks in advance, Simon

cudnn benchmark: True cudnn deterministic: False Creating metadata Data file extensions: {'images_style': 'jpg', 'images_content': 'jpg'} Searching in dir: images_style Found 1 sequences Found 3 files Folder at projects/coco_funit/test_data/animal_faces/images_style opened. Folder at projects/coco_funit/test_data/animal_faces/images_content opened. Num datasets: 1 Num sequences: 1 Max sequence length: 3 Traceback (most recent call last): File "inference.py", line 91, in main() File "inference.py", line 62, in main test_data_loader = get_test_dataloader(cfg) File "d:\anaconda\imaginaire\imaginaire\utils\dataset.py", line 113, in get_test_dataloader test_data_loader = _get_data_loader( File "d:\anaconda\imaginaire\imaginaire\utils\dataset.py", line 46, in _get_data_loader not_distributed = not_distributed or not dist.is_initialized() AttributeError: module 'torch.distributed' has no attribute 'is_initialized'

arunmallya commented 3 years ago

Try running with --single_gpu. distributed might not work on windows.