def mlDriver(cv, stateTransfer, actionTransfer):
#parameter setup
#dimensionality of state argument (could be less than stateTransfer)
stateDim = 352
#Number of moves possible
numMoves = 361
env = SettleEnv(cv, stateTransfer, actionTransfer)
task = SettleTask(env)
controller = RestrictedActionValueNetwork(stateDim, numMoves, env)
learner = NFQ()
learner.explorer = EpsilonHackedExplorer(env)
agent = LearningAgent(controller, learner)
experiment = EpisodicExperiment(task, agent)
while True:
experiment.doEpisodes(10)
print "Done with experiments"
agent.learn()
print "Learned"
agent.reset()
print "Cycled"
def mlDriver(cv, stateTransfer, actionTransfer):
#parameter setup
#dimensionality of state argument (could be less than stateTransfer)
stateDim = 352
#Number of moves possible
numMoves = 361
env = SettleEnv(cv, stateTransfer, actionTransfer)
task = SettleTask(env)
controller = RestrictedActionValueNetwork(stateDim, numMoves, env)
learner = NFQ()
learner.explorer = EpsilonHackedExplorer(env)
agent = LearningAgent(controller, learner)
experiment = EpisodicExperiment(task, agent)
while True:
experiment.doEpisodes(10)
print "Done with experiments"
agent.learn()
print "Learned"
agent.reset()
print "Cycled"
