Closed rafaelBauer closed 3 months ago
Digging further, I saw that the Tensor self.qpos
is a torch.float32
and the other, self._target_qpos
, is torch.float64
.
I made sure that the action I give in the step method is a torch.float32
and now the problem doesn't happen anymore
Ah thanks for raising this issue. This sounds more like a bug that we should fool-proof so someone sending a float64 is warned about it.
Yeah, I think so too.. I have closed the issue myself, because I didn't want to bother you with it, but maybe you should keep it open.
Library version: mani_skill==3.0.0b4 Python: 3.11.0
I am getting the following error when calling env.reset():
I have the
pd_joint_pos
controller.Any pointers? I will check if while the reset() is called I am still sending step commands. Could this be a reason?