Breaking changes in newtonnet 2.0

[Andrew-S-Rosen]

Hi @ericyuan00000: I'm getting a bunch of failed tests with the new newtonnet 2.0 that you can see here: https://github.com/Quantum-Accelerators/quacc/actions/runs/14544325243/job/40807607302?pr=2755. Any ideas on how to proceed?

They are all related to:

    def load_model(self, model):
        # TODO: Load model with only weights
        model = torch.load(model, map_location=self.device[0], weights_only=False)
        keys_to_keep = ['energy']
        for key in self.properties:
            key = {
                'energy': 'energy',
                'free_energy': 'energy',
                'forces': 'gradient_force',
                'stress': 'virial',
                'hessian': 'hessian',
            }.get(key)
            keys_to_keep.append(key)
>           if key in model.infer_properties:
E           AttributeError: 'dict' object has no attribute 'infer_properties'

[Andrew-S-Rosen]

Just a head's up, @ericyuan00000 that the issue remains in 2.0.1.