faenet.base_model
Module Contents
Classes
- class faenet.base_model.BaseModel(**kwargs)
Bases:
torch.nn.Module- property num_params
- abstract energy_forward(data, preproc=True)
Forward pass for energy prediction.
- forces_as_energy_grad(pos, energy)
Computes forces from energy gradient
- Parameters:
pos (tensor) – atom positions
energy (tensor) – predicted energy
- Returns:
gradient of energy w.r.t. atom positions
- Return type:
forces (tensor)
- abstract forces_forward(preds)
Forward pass for force prediction.
- forward(data, mode='train', preproc=True)
Main Forward pass.
- reset_parameters()
Resets all learnable parameters of the module.