`unreal.LearningAgentsImitationTrainerTrainingSettings`¶

class unreal.LearningAgentsImitationTrainerTrainingSettings(hyperparameters: LearningAgentsImitationTrainerHyperparameters = [], logging_settings: LearningAgentsImitationLearningLoggingSettings = [], validation_settings: LearningAgentsImitationLearningValidationSettings = Ellipsis, device: LearningAgentsTrainingDevice = Ellipsis)¶

The configurable settings for the imitation learning training process.

C++ Source:

Editor Properties: (see get_editor_property/set_editor_property)

batch_count_per_evaluation (int32): [Read-Only] How many batches to perform evaluation on? Randomly chosen each evaluation. Currently not used in Python
device (LearningAgentsTrainingDevice): [Read-Write] The device to train on.
embed_normalizations_in_weights (bool): [Read-Write] Whether to embed dataset normalizations into encoder and decoder networks. This is recommended to set to true to stabilize trainings. <Important> The normalization option in action/observation schemas must be set to one of the “auto” options.
evaluation_frequency (int32): [Read-Only] How many training iteration loops between an evaluation run. Currently not used in Python
hyperparameters (LearningAgentsImitationTrainerHyperparameters): [Read-Write]
logging_settings (LearningAgentsImitationLearningLoggingSettings): [Read-Write]
run_evaluation (bool): [Read-Only] Should evaluation run during the training process. Currently not used in Python
train_eval_dataset_split (float): [Read-Only] How much data should be used for evaluation. Currently not used in Python
validation_settings (LearningAgentsImitationLearningValidationSettings): [Read-Write]