TorchCraftAI
A bot for machine learning research on StarCraft: Brood War
Public Member Functions | Public Attributes | List of all members
cpid::BatchedPGReplayBufferFrame Struct Reference

#include <policygradienttrainer.h>

Inherits cpid::ReplayBufferFrame.

Public Member Functions

 BatchedPGReplayBufferFrame (ag::Variant state, torch::Tensor action, float pAction, double reward)
 
- Public Member Functions inherited from cpid::ReplayBufferFrame
virtual ~ReplayBufferFrame ()=default
 
template<class Archive >
void serialize (Archive &ar)
 

Public Attributes

ag::Variant state
 
torch::Tensor action
 
float pAction
 Probability of action according to the policy that was used to obtain this frame. More...
 
double reward
 Reward observed since taking previous action. More...
 

Constructor & Destructor Documentation

cpid::BatchedPGReplayBufferFrame::BatchedPGReplayBufferFrame ( ag::Variant  state,
torch::Tensor  action,
float  pAction,
double  reward 
)
inline

Member Data Documentation

torch::Tensor cpid::BatchedPGReplayBufferFrame::action
float cpid::BatchedPGReplayBufferFrame::pAction

Probability of action according to the policy that was used to obtain this frame.

double cpid::BatchedPGReplayBufferFrame::reward

Reward observed since taking previous action.

ag::Variant cpid::BatchedPGReplayBufferFrame::state

The documentation for this struct was generated from the following file: