Fix dtype in self_energies #347
Conversation
| @@ -192,7 +192,7 @@ def sae(self, species): | |||
| intercept = self.self_energies[-1] | |||
|
|
|||
| self_energies = self.self_energies[species] | |||
| self_energies[species == torch.tensor(-1, device=species.device)] = torch.tensor(0, device=species.device, dtype=self_energies.dtype) | |||
| self_energies[species == torch.tensor(-1, device=species.device)] = torch.tensor(0, device=species.device, dtype=torch.double) | |||
There was a problem hiding this comment.
how about removing the dtype and replace the 0 with 0.0?
There was a problem hiding this comment.
@zasdfgbnm self_energies.dtype should be double, but float32 is enforced in comp6.py, is there any particular reason for that?
see in: https://github.com/aiqm/torchani/pull/347/checks?check_run_id=273594990#step:9:25
There was a problem hiding this comment.
@farhadrgh No, there is no reason. I guess it is just randomly picking a dtype.
There was a problem hiding this comment.
@farhadrgh Also, there are some trick here, energy shifter is a module, so if you put it inside
model = torch.nn.Sequential(aev_computer, nn, energy_shifter) and then model.to(torch.float) it will also be cast into float. I don't have a good idea on how to solve that.
There was a problem hiding this comment.
I fixed the dtype in comp6.py (6930879), this will let the tests pass peacefully
I wonder why the test hadn't failed before!