Skip to content

Schema for Configuration GraphPes MLP¤

The schema to configure the input file for GraphPes MLP.

Schema:¤

### Schema for MLP GraphPes

### Reference
# https://vldgroup.github.io/graph-pes/cli/graph-pes-train/complete-docs.html
# https://github.com/vldgroup/graph-pes/tree/main/configs
# https://github.com/vldgroup/graph-pes/blob/main/src/graph_pes/config/training-defaults.yaml


### Only keys listed below are needed in CLFF workflow.

model:                # Accept all parameters defined in the 'model' section of the upstream GraphPes.
  type: dict

loss:                 # Accept all parameters defined in the 'loss' section of the upstream GraphPes.
  type: dict

fitting:              # Accept all parameters defined in the `fitting` section of the upstream GraphPes.
  type: dict
  allow_unknown: True #
  schema:
    auto_fit_reference_energies:
      type: boolean
    trainer_kwargs:
      type: dict
    optimizer:
      type: dict
    scheduler:
      type: dict
    early_stopping:
      type: dict
    loader_kwargs:
      type: dict

general:              # Accept all parameters defined in the `general` section of the upstream GraphPes.
  type: dict