You should maybe take out torchand cupy since they are installed for CUDA 10.1. To get it to work on Windows 10 with Python3.7 you have to do the following:
Edit attribute_hallucination\editing_tool\semantic_segmentation_pytorch\lib\utils\data\dataloader.py:
Change _update_worker_pids to _set_worker_pids in Line 238 and Line 3
Edit attribute_hallucination\editing_tool\semantic_segmentation_pytorch\lib\nn\parallel\data_parallel.py
Change in Line 16 obj.cuda(dev, async=True) to obj.cuda(dev, non_blocking=True) since async is a reserved keyword in python 3.7
I got it running on Windows with following
editing_tool/requirements.txt
:You should maybe take out
torch
andcupy
since they are installed for CUDA 10.1. To get it to work on Windows 10 with Python3.7 you have to do the following:Edit
attribute_hallucination\editing_tool\semantic_segmentation_pytorch\lib\utils\data\dataloader.py
: Change _update_worker_pids to _set_worker_pids in Line 238 and Line 3Edit
attribute_hallucination\editing_tool\semantic_segmentation_pytorch\lib\nn\parallel\data_parallel.py
Change in Line 16obj.cuda(dev, async=True)
toobj.cuda(dev, non_blocking=True)
sinceasync
is a reserved keyword in python 3.7