GeneralPropertyDatasetParameters

Included in QATK.MLFF

class GeneralPropertyDatasetParameters(property_key=None, task_type=None, validation_fraction=None, loss_weight=None, pooling=None)

Constructor for GeneralPropertyDatasetParameters.

Parameters:

• property_key (str) – Key to access the property in the dataset. Also used for model outputs and logging.

• task_type (str) – Type of task - MLParameterOptions.TASK_TYPE.GENERAL for configuration-level or MLParameterOptions.TASK_TYPE.ATOM_WISE for per-atom properties.

• validation_fraction (float) – The fraction of the training data to use for validation.

• loss_weight (float) – Weight for this property in the loss function.

• pooling (str) – Pooling method for configuration-level properties - MLParameterOptions.POOLING.SUM, MLParameterOptions.POOLING.MEAN, or MLParameterOptions.POOLING.MAX. Use MLParameterOptions.POOLING.SUM for extensive properties (that scale with system size, e.g., total energy) and MLParameterOptions.POOLING.MEAN for intensive properties (that don’t scale with system size, e.g., energy per atom).

lossWeight()

Returns:

The weight for this property in the loss function.

Return type:

float

nlinfo()

Returns:

The nlinfo.

Return type:

dict

pooling()

Returns:

The pooling method for configuration-level properties. Use MLParameterOptions.POOLING.SUM for extensive properties (that scale with system size) and MLParameterOptions.POOLING.MEAN for intensive properties (that don’t scale with system size).

Return type:

str

propertyKey()

Returns:

The key to access the property in the dataset. Also used for model outputs and logging.

Return type:

str

taskType()

Returns:

The type of task (general or atom_wise).

Return type:

str

uniqueString()

Return a unique string representing the state of the object.

validationFraction()

Returns:

The fraction of the training data to use for validation.

Return type:

float