TrainingParameters

Included in QATK.MLFF

class TrainingParameters(experiment_name=None, batch_size=None, max_number_of_epochs=None, patience=None, device=None, random_seed=None, number_of_workers=None, default_dtype=None, learning_rate=None, weight_decay=None, restart_from_last_checkpoint=None, scheduler_patience=None, gradient_clipping_threshold=None, save_all_available_model_formats=None, additional_parameters=None)

Constructor for TrainingParameters.

Parameters:
  • experiment_name (str) – The name of the training experiment.

  • batch_size (int) – Number of samples per training batch.

  • max_number_of_epochs (int) – The maximum number of epochs to train for.

  • patience (int) – The number of epochs to wait before early stopping.

  • device (str) – The device to train on. Possible values are MLParameterOptions.DEVICE.AUTOMATIC, MLParameterOptions.DEVICE.CPU, and MLParameterOptions.DEVICE.GPU.
    Default: MLParameterOptions.DEVICE.AUTOMATIC

  • random_seed (int) – The random seed for reproducibility.

  • number_of_workers (int) – The number of workers for data loading.

  • default_dtype (str) – The default torch data type. Possible values are MLParameterOptions.DTYPE.FLOAT32 and MLParameterOptions.DTYPE.FLOAT64.
    Default: MLParameterOptions.DTYPE.FLOAT64

  • learning_rate (float) – The learning rate of optimizer.

  • weight_decay (float) – The weight decay (L2 penalty).

  • restart_from_last_checkpoint (bool) – Whether to restart training from last saved checkpoint.

  • scheduler_patience (int) – The patience of the scheduler.

  • gradient_clipping_threshold (float) – The gradient clipping threshold.

  • save_all_available_model_formats (bool) – Whether to export all model formats (CuEq float32/float64, e3nn) instead of only the format used in training.

  • additional_parameters (dict) – Additional parameters for the MACE model.

additionalParameters()
Returns:

Additional parameters for the MACE model.

Return type:

dict

batchSize()
Returns:

The batch size for training.

Return type:

int

defaultDtype()
Returns:

The default torch data type.

Return type:

str

device()
Returns:

The device to train on. Whether it is CPU or GPU.

Return type:

str

experimentName()
Returns:

The name of the training experiment.

Return type:

str

gradientClippingThreshold()
Returns:

The gradient clipping value.

Return type:

float

learningRate()
Returns:

The learning rate of optimizer.

Return type:

float

maxNumberOfEpochs()
Returns:

The maximum number of epochs to train for.

Return type:

int

nlinfo()
Returns:

The nlinfo.

Return type:

dict

numberOfWorkers()
Returns:

The number of workers for data loading.

Return type:

int

patience()
Returns:

The number of epochs to wait before early stopping.

Return type:

int

randomSeed()
Returns:

The random seed.

Return type:

int

restartFromLastCheckpoint()
Returns:

Whether to restart training from last saved checkpoint.

Return type:

bool

saveAllAvailableModelFormats()
Returns:

Whether to export all model formats (CuEq float32/float64, e3nn) instead of only the format used in training.

Return type:

bool

schedulerPatience()
Returns:

The patience of the scheduler.

Return type:

int

uniqueString()

Return a unique string representing the state of the object.

weightDecay()
Returns:

The weight decay (L2 penalty).

Return type:

float