Struct FSB3TrainingSettings

Defined in File SB3TrainingSettings.h

Inheritance Relationships

Base Type

public FTrainingSettings (Struct FTrainingSettings)

struct FSB3TrainingSettings : public FTrainingSettings

Source: Source/Schola/Training/Public/TrainingSettings/StableBaselines/SB3TrainingSettings.h

Dependencies: FSB3CheckpointSettings, FSB3LoggingSettings, FSB3NetworkArchSettings, FSB3PPOSettings, FSB3ResumeSettings, FSB3SACSettings, FScriptArgBuilder

Configuration struct for Stable Baselines 3 (SB3) training scripts.

Contains all settings necessary to launch and configure an SB3 training session. This is a partial implementation of the available SB3 settings and is not exhaustive.

Public Functions

Symbol	Details
`GenerateTrainingArgs`	Generate command-line arguments for the SB3 training script.
`~FSB3TrainingSettings`	Virtual destructor.

GenerateTrainingArgs

virtual void GenerateTrainingArgs(FScriptArgBuilder &ArgBuilder) const

Generate command-line arguments for the SB3 training script.

Parameters

ArgBuilder – [inout] The argument builder to populate with settings.

#	Direction	Name	Type	Description
1	—	`ArgBuilder`	`FScriptArgBuilder &`	The argument builder to populate with settings.

Attributes: const, virtual

Source: Source/Schola/Training/Public/TrainingSettings/StableBaselines/SB3TrainingSettings.h

Implementation: Source/Schola/Training/Private/TrainingSettings/StableBaselines/SB3TrainingSettings.cpp

`~FSB3TrainingSettings`

virtual ~FSB3TrainingSettings()

Virtual destructor.

Attributes: virtual

Source: Source/Schola/Training/Public/TrainingSettings/StableBaselines/SB3TrainingSettings.h

Implementation: Source/Schola/Training/Private/TrainingSettings/StableBaselines/SB3TrainingSettings.cpp

Public Members

Symbol	Details
`Timesteps`	The total number of environment timesteps to train for.
`LoggingSettings`	Settings for logging training metrics and outputs.
`CheckpointSettings`	Settings for model checkpointing during training.
`ResumeSettings`	Settings for resuming training from a checkpoint.
`NetworkArchitectureSettings`	Settings for neural network architecture (e.g.
`bDisplayProgressBar`	Whether to display a progress bar during training.
`Algorithm`	The reinforcement learning algorithm to use (e.g., SAC, PPO).
`PPOSettings`	PPO specific settings.
`SACSettings`	SAC specific settings.

`Timesteps`

int Timesteps = 8000

The total number of environment timesteps to train for.

`LoggingSettings`

FSB3LoggingSettings LoggingSettings

Settings for logging training metrics and outputs.

`CheckpointSettings`

FSB3CheckpointSettings CheckpointSettings

Settings for model checkpointing during training.

`ResumeSettings`

FSB3ResumeSettings ResumeSettings

Settings for resuming training from a checkpoint.

`NetworkArchitectureSettings`

FSB3NetworkArchSettings NetworkArchitectureSettings

Settings for neural network architecture (e.g.

layers, activation functions).

`bDisplayProgressBar`

bool bDisplayProgressBar = true

Whether to display a progress bar during training.

Requires TQDM and Rich to be installed.

`Algorithm`

ESB3TrainingAlgorithm Algorithm = ESB3TrainingAlgorithm::PPO

The reinforcement learning algorithm to use (e.g., SAC, PPO).

`PPOSettings`

FSB3PPOSettings PPOSettings

PPO specific settings.

`SACSettings`

FSB3SACSettings SACSettings

SAC specific settings.