Struct FSB3SACSettings

struct FSB3SACSettings : public FTrainingSettings

A struct to hold SAC settings for an SB3 training script.

Note

This is a partial implementation of the SAC settings, and is not exhaustive

Public Functions

virtual void GenerateTrainingArgs(int Port, FScriptArgBuilder &ArgBuilder) const

Generate the training arguments for the script using the ArgBuilder.

Note

port is supplied as it is a common argument to pass to scripts, and is set at a high level but might be needed by specific subsettings

Parameters:

• Port – [in] The port to use for the script

• ArgBuilder – [in] The builder to use to generate the arguments

virtual ~FSB3SACSettings()

Public Members

float LearningRate = 0.0003

The learning rate for the SAC algorithm.

int BufferSize = 1000000

The buffer size for the SAC algorithm.

int LearningStarts = 100

The number of steps to take before learning starts.

int BatchSize = 256

The batch size to use during gradient descent.

float Tau = 0.005

The Tau value for the SAC algorithm.

float Gamma = 0.99

The gamma value for the SAC algorithm.

int TrainFreq = 1

The frequency to update the target network, in steps.

int GradientSteps = 1

The number of gradient steps to take during training.

bool OptimizeMemoryUsage = false

Optimize memory usage.

bool LearnEntCoef = true

Should we learn the entropy coefficient during training.

float InitialEntCoef = 1.0

The initial entropy coefficient for the SAC algorithm.

int TargetUpdateInterval = 1

The interval at which we update the target network.

FString TargetEntropy = “auto”

The target entropy for the SAC algorithm.

use auto to learn the target entropy

bool UseSDE = false

Use state dependent entropy noise.

int SDESampleFreq = –1

The frequency to sample the state dependent entropy noise.