FRLlibAPPOSettings

struct FRLlibAPPOSettings : public FTrainingSettings

一个结构体，用于保存 RLLib 训练脚本的异步近端策略优化 (APPO) 设置。

注意：这是 APPO 设置的局部实现，并非详尽无遗。

依赖项： FScriptArgBuilder, FTrainingSettings

继承自： public FTrainingSettings

公共接口

析构函数

virtual ~FRLlibAPPOSettings()

属性: virtual

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 40, column 9)

Implementation: Schola/Source/Schola/Private/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.cpp (lines 15-18)

公共函数

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

使用 ArgBuilder 为脚本生成训练参数。

注意：端口作为脚本的常见参数传入，并在高层设置，但特定子设置可能需要它。

参数

属性： const, virtual

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 38, column 6)

Implementation: Schola/Source/Schola/Private/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.cpp (lines 5-13)

公共成员

bool bVTrace = = true

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 21, column 10)

float VTraceClipRhoThreshold = = 1.0

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 24, column 11)

float VTraceClipPGRhoThreshold = = 1.0

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 27, column 11)

float GAELambda = = 0.95

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 30, column 11)

float ClipParam = = 0.2

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 33, column 11)

bool bUseGAE = = true

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 36, column 10)

被使用： FRLlibTrainingSettings

Source: Schola/Source/Schola/Public/Subsystem/SubsystemSettings/Ray/Algorithms/RLlibAPPOSettings.h (line 15, column 1)