Agents are used to track the performance of one instance of a network. More...
#include <Agent.h>
Public Member Functions | |
| virtual void | StartNewExperience ()=0 |
| Starts a new experience. It is expected that RecordObservation(), RecordAction(), and GiveReward() will all be called at least once before the next experience is started. More... | |
| virtual void | RecordObservation (const char *inputName, const float *buffer, int64 bufferCount)=0 |
| Records the input data to the network we are watching. More... | |
| virtual void | RecordAction (const char *actionName, const float *buffer, int64 bufferCount)=0 |
| Records the output of the network we applied to the environment (the game or character) More... | |
| virtual void | GiveReward (float reward)=0 |
| The trainers will try to maximize rewards. The higher the value, the better the rewards. Rewards can also be negative. More... | |
| virtual void | EndEpisode ()=0 |
| Called at the logical conclusion of a training session. Usually some event like the player died or the game ended. More... | |
 Public Member Functions inherited from SmartEngine::IObject | |
| virtual ObjectId | GetId () const =0 |
| Returns the ID of this object. More... | |
| virtual void | AddRef () const =0 |
| Increments the internal reference count on this object. It is not common to use this method directly. More... | |
| virtual void | Release () const =0 |
| Decrements the internal reference count on this object. It is not common to use this method directly. More... | |
| virtual int | GetRefCount () const =0 |
| Returns the number of references to this object. More... | |
| virtual void * | QueryInterface (ObjectClassId id)=0 |
| Queries the object for an interface and returns a pointer to that interface if found. More... | |
| void | operator= (IObject const &x)=delete |
| Public Member Functions inherited from SmartEngine::ISerializable | |
| virtual SerializationResult | Serialize (IMemoryBuffer *buffer)=0 |
| Write the contents of this object to a buffer. More... | |
| virtual SerializationResult | Deserialize (IMemoryBuffer *buffer)=0 |
| Fill this object with contents from a buffer. More... | |
Additional Inherited Members | |
| Public Attributes inherited from SmartEngine::IObject | |
| private | |
| __pad0__: IObject() {} IObject(IObject const&) = delete | |
Detailed Description
Agents are used to track the performance of one instance of a network.
For each step, you start a new experience. Then record the observation (the network's input data), the action (output of the network), and the rewards it received during that time.
When you reach a logical conclusion of your training (the character dies or the game ends, for example), end the episode.
Member Function Documentation
◆ EndEpisode()
|
pure virtual |
Called at the logical conclusion of a training session. Usually some event like the player died or the game ended.
◆ GiveReward()
|
pure virtual |
The trainers will try to maximize rewards. The higher the value, the better the rewards. Rewards can also be negative.
- Parameters
-
reward
◆ RecordAction()
|
pure virtual |
Records the output of the network we applied to the environment (the game or character)
- Parameters
-
actionName The name of the output graph node in the graph definition actionData In the case of discrete actions, the index of the chosen neuron. For continuous actions, the raw output.
◆ RecordObservation()
|
pure virtual |
Records the input data to the network we are watching.
- Parameters
-
inputName The name of the BufferInput in the graph definition obsData The buffer's input
◆ StartNewExperience()
|
pure virtual |
Starts a new experience. It is expected that RecordObservation(), RecordAction(), and GiveReward() will all be called at least once before the next experience is started.
