runtime_error, when MODEL.ROI_HEADS.SOFT_NMS_ENABLED is True.

[Mingzhou He]

When MODEL.ROI_HEADS.SOFT_NMS_ENABLED is True, executing test_transfer_learning.sh produces a runtime error. When MODEL.ROI_HEADS.SOFT_NMS_ENABLED uses the default, the program is OK.

File "/home/hiram/workshop/paper2/RegionCLIP/detectron2/layers/soft_nms.py", line 130, in batched_soft_nms
return soft_nms(
File "/home/hiram/workshop/paper2/RegionCLIP/detectron2/layers/soft_nms.py", line 35, in soft_nms
return _soft_nms(
File "/home/hiram/workshop/paper2/RegionCLIP/detectron2/layers/soft_nms.py", line 259, in _soft_nms
idxs = idxs[keep]
RuntimeError: indices should be either on cpu or on the same device as the indexed tensor (CPU)

[Mingzhou He]

已解决 pytorch版本2.0出现了这个问题，退版本解决