The results of the two runs are different

Hello,
      Thanke you for your work! When I use it, I find that although I fixed the seed points of torch and numpy, the chamfer distance still makes the results inconsistent every time I run. Do you know how to deal with this situation?
      Best,
      diaojunqi