1. Epic GamesDeveloper
  3. FSharedMemoryPPOTrainer

FSharedMemoryPPOTrainer

Trainer that uses shared memory and a Python sub-process to perform training

API > API/Plugins > API/Plugins/LearningTraining

Inheritance Hierarchy

References

   
Module LearningTraining
Header /Engine/Plugins/Experimental/LearningAgents/Source/LearningTraining/Public/LearningPPOTrainer.h
Include #include "LearningPPOTrainer.h"

Syntax

struct FSharedMemoryPPOTrainer : public UE::Learning::IPPOTrainer

Remarks

Trainer that uses shared memory and a Python sub-process to perform training

This trainer is the most simple and efficient when training the policy on the same computer that experience is being gathered on.

Constructors

  Type Name Description
Public function  

FSharedMemoryPPOTrainer ( const FString& TaskName,
const FString& PythonExecutablePath,
const FString& SitePackagesPath,
const FString& PythonContentPath,
const FString& IntermediatePath,
const FReplayBuffer& ReplayBuffer,
const FPPOTrainerTrainingSettings& TrainingSettings,
const FPPOTrainerNetworkSettings& NetworkSettings,
const EPPOTrainerFlags TrainerFlags,
const ELogSetting LogSettings,
const ESubprocessFlags TrainingProcessFlags,
const uint16 ProcessNum,
const ESubprocessFlags MultiProcessFlags
)

 Creates a new Shared Memory PPO trainer

Destructors

  Type Name Description
Public function    

Overridden from IPPOTrainer

  Type Name Description
Public function Virtual ETrainerResponse

RecvCritic ( FNeuralNetwork& OutNetwork,
const float Timeout,
FRWLock* NetworkLock,
const ELogSetting LogSettings
)

 Wait for the trainer to push an updated critic network.
Public function Virtual ETrainerResponse

RecvPolicy ( FNeuralNetwork& OutNetwork,
const float Timeout,
FRWLock* NetworkLock,
const ELogSetting LogSettings
)

 Wait for the trainer to push an updated policy network.
Public function Virtual ETrainerResponse

SendCritic ( const FNeuralNetwork& Network,
const float Timeout,
FRWLock* NetworkLock,
const ELogSetting LogSettings
)

 Wait for the trainer to be ready and push the current critic network.
Public function Virtual ETrainerResponse

SendExperience ( const FReplayBuffer& ReplayBuffer,
const float Timeout,
const ELogSetting LogSettings
)

 Wait for the trainer to be ready and push new experience.
Public function Virtual ETrainerResponse

SendPolicy ( const FNeuralNetwork& Network,
const float Timeout,
FRWLock* NetworkLock,
const ELogSetting LogSettings
)

 Wait for the trainer to be ready and push the current policy network.
Public function Virtual ETrainerResponse

SendStop ( const float Timeout
)

 Signal for the trainer to stop.
Public function Virtual void Terminate the trainer immediately.
Public function Virtual ETrainerResponse

Wait ( const float Timeout
)

 Wait for the trainer to finish.

Ask questions and help your peers Developer Forums

Write your own tutorials or read those from others Learning Library

On this page