感谢开源代码~但是我在复现的时候出现了问题，训练的时候出问题了，这个要咋处理呢？

[Touyama-Nao]

(RDMNet) xiaojunwei@Venus:/data/xiaojunwei/projects/RDMNet-main/experiments$ python trainval.py
Epoch: 1/160, iter: 140/1358, c_loss: 0.683, g_loss: 0.677, n_loss: 0.014, p_loss: 0.021, v_loss: 2.102, nn_loss: 0.012, d_loss: 0.481, loss: 6.697, PIR: 0.002, lr: 1.000e-04, time: 0.096s/0.252s: 11%| | 1
Traceback (most recent call last):
File "trainval.py", line 69, in
main()
File "trainval.py", line 65, in main
trainer.run()
File "/data1/xiaojunwei/projects/RDMNet-main/geotransformer/engine/epoch_based_trainer.py", line 197, in run
self.train_epoch()
File "/data1/xiaojunwei/projects/RDMNet-main/geotransformer/engine/epoch_based_trainer.py", line 99, in train_epoch
output_dict, result_dict = self.train_step(self.epoch, self.inner_iteration, data_dict)
File "trainval.py", line 47, in train_step
loss_dict = self.loss_func(output_dict, data_dict)
File "/data1/xiaojunwei/miniconda3/envs/RDMNet/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/data1/xiaojunwei/projects/RDMNet-main/experiments/loss.py", line 313, in forward
gap_loss = self.gap_loss(output_dict, data_dict)
File "/data1/xiaojunwei/miniconda3/envs/RDMNet/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/data1/xiaojunwei/projects/RDMNet-main/experiments/loss.py", line 236, in forward
ref_dist, ref_mindis_indic = dists.min(-1)
RuntimeError: cannot perform reduction function min on tensor with no elements because the operation does not have an identity