IndexError: The shape of the mask [50] at index 0 does not match the shape of the indexed tensor [1] at index 0

[luoolu]

torch.multiprocessing.spawn.ProcessRaisedException:

-- Process 1 terminated with the following error:
Traceback (most recent call last):
File "/home/DATA/user/PycharmProjects/vnext/venv/lib/python3.7/site-packages/torch/multiprocessing/spawn.py", line 59, in _wrap
fn(i, *args)
File "/home/DATA/user/PycharmProjects/vnext/detectron2/engine/launch.py", line 126, in _distributed_worker
main_func(*args)
File "/home/DATA/user/PycharmProjects/vnext/projects/IDOL/train_net.py", line 168, in main
return trainer.train()
File "/home/DATA/user/PycharmProjects/vnext/detectron2/engine/defaults.py", line 484, in train
super().train(self.start_iter, self.max_iter)
File "/home/DATA/user/PycharmProjects/vnext/detectron2/engine/train_loop.py", line 149, in train
self.run_step()
File "/home/DATA/user/PycharmProjects/vnext/detectron2/engine/defaults.py", line 494, in run_step
self._trainer.run_step()
File "/home/DATA/user/PycharmProjects/vnext/detectron2/engine/train_loop.py", line 273, in run_step
loss_dict = self.model(data)
File "/home/DATA/user/PycharmProjects/vnext/venv/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/DATA/user/PycharmProjects/vnext/venv/lib/python3.7/site-packages/torch/nn/parallel/distributed.py", line 705, in forward
output = self.module(*inputs[0], **kwargs[0])
File "/home/DATA/user/PycharmProjects/vnext/venv/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/DATA/user/PycharmProjects/vnext/projects/IDOL/idol/idol.py", line 219, in forward
det_targets, ref_targets = self.prepare_targets(gt_instances)
File "/home/DATA/user/PycharmProjects/vnext/projects/IDOL/idol/idol.py", line 315, in prepare_targets
ref_target[k] = v[valid_i]
IndexError: The shape of the mask [50] at index 0 does not match the shape of the indexed tensor [1] at index 0

[luoolu]

The same dataset can be trained normally on the Mask2former network