/// <summary>
/// Training for supervised learning
/// </summary>
/// <param name="vectorObservations"></param>
/// <param name="visualObservations"></param>
/// <param name="actions"></param>
/// <returns></returns>
public float TrainBatch(float[,] vectorObservations, List <float[, , , ]> visualObservations, float[,] actions)
{
Debug.Assert(mode == Mode.SupervisedLearning, "This method is for SupervisedLearning mode only. Please set the mode of RLModePPO to SupervisedLearning in the editor.");
Debug.Assert(TrainingEnabled == true, "The model needs to initalized with Training enabled to use TrainBatch()");
List <Array> inputs = new List <Array>();
if (vectorObservations != null)
{
inputs.Add(vectorObservations);
}
if (visualObservations != null)
{
inputs.AddRange(visualObservations);
}
if (ActionSpace == SpaceType.continuous)
{
inputs.Add(actions);
}
else if (ActionSpace == SpaceType.discrete)
{
int[,] actionsInt = actions.Convert(t => Mathf.RoundToInt(t));
inputs.Add(actionsInt);
}
var loss = UpdateSLFunction.Call(inputs);
var result = (float)loss[0].eval();
return(result);
}
/// <summary>
/// Training for supervised learning
/// </summary>
/// <param name="vectorObservations"></param>
/// <param name="visualObservations"></param>
/// <param name="actions"></param>
/// <returns></returns>
public float TrainBatch(float[,] vectorObservations, List <float[, , , ]> visualObservations, float[,] actions, List <float[, ]> actionsMask = null)
{
Debug.Assert(TrainingEnabled == true, "The model needs to initalized with Training enabled to use TrainBatch()");
List <Array> inputs = new List <Array>();
if (vectorObservations != null)
{
inputs.Add(vectorObservations);
}
if (visualObservations != null)
{
inputs.AddRange(visualObservations);
}
if (ActionSpace == SpaceType.continuous)
{
inputs.Add(actions);
}
else if (ActionSpace == SpaceType.discrete)
{
List <float[, ]> masks = actionsMask;
int batchSize = actions.GetLength(0);
//create all 1 mask if the input mask is null.
if (masks == null)
{
masks = CreateDummyMasks(ActionSizes, batchSize);
}
inputs.AddRange(masks);
int[,] actionsInt = actions.Convert(t => Mathf.RoundToInt(t));
inputs.Add(actionsInt);
}
var loss = UpdateSLFunction.Call(inputs);
var result = (float)loss[0].eval();
return(result);
}