def main(): filename = "learned_individuals_{0}".format(0) rand_individual = Individual(data=filename, random=True) agent = MyAgent(rand_individual) task = MarioTask(agent.name, initMarioMode=2) exp = EpisodicExperiment(task, agent) print('Task Ready') exp.doEpisodes(2) print('mm 2:', task.reward) task.env.initMarioMode = 1 exp.doEpisodes(1) print('mm 1:', task.reward) task.env.initMarioMode = 0 exp.doEpisodes(1) print('mm 0:', task.reward) task.env.initMarioMode = 0 exp.doEpisodes(1) print('mm 0:', task.reward) task.env.initMarioMode = 0 task.env.levelDifficulty = 5 exp.doEpisodes(1) print('mm 0, ld 5: ', task.reward) task.env.initMarioMode = 1 task.env.levelDifficulty = 5 exp.doEpisodes(1) print('mm 1, ld 5: ', task.reward) task.env.initMarioMode = 2 task.env.levelDifficulty = 5 exp.doEpisodes(1) print('mm 2, ld 5: ', task.reward) print("finished")
def main(): clo = CmdLineOptions(sys.argv) task = MarioTask(MarioEnvironment(clo.getHost(), clo.getPort(), clo.getAgent().name)) exp = EpisodicExperiment(clo.getAgent(), task) exp.doEpisodes(3)
def main(): agent = ForwardAgent() task = MarioTask(agent.name, initMarioMode = 2) exp = EpisodicExperiment(task, agent) print 'Task Ready' exp.doEpisodes(2) print 'mm 2:', task.reward task.env.initMarioMode = 1 exp.doEpisodes(1) print 'mm 1:', task.reward task.env.initMarioMode = 0 exp.doEpisodes(1) print 'mm 0:', task.reward task.env.initMarioMode = 0 exp.doEpisodes(1) print 'mm 0:', task.reward task.env.initMarioMode = 0 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 0, ld 5: ', task.reward task.env.initMarioMode = 1 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 1, ld 5: ', task.reward task.env.initMarioMode = 2 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 2, ld 5: ', task.reward print "finished"
def main(): agent = ForwardAgent() task = MarioTask(agent.name, initMarioMode=2) exp = EpisodicExperiment(task, agent) print 'Task Ready' exp.doEpisodes(2) print 'mm 2:', task.reward task.env.initMarioMode = 1 exp.doEpisodes(1) print 'mm 1:', task.reward task.env.initMarioMode = 0 exp.doEpisodes(1) print 'mm 0:', task.reward task.env.initMarioMode = 0 exp.doEpisodes(1) print 'mm 0:', task.reward task.env.initMarioMode = 0 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 0, ld 5: ', task.reward task.env.initMarioMode = 1 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 1, ld 5: ', task.reward task.env.initMarioMode = 2 task.env.levelDifficulty = 5 exp.doEpisodes(1) print 'mm 2, ld 5: ', task.reward print "finished"
def main(): f = open('try_3.txt','w') agent = Ahude(IT,f) distances = np.zeros([1]) data = np.zeros([1]) num_help = np.zeros([1]) mis_match = np.zeros([1]) #task = MarioTask(agent.name, initMarioMode = 2) #exp = EpisodicExperiment(task, agent) print 'Task Ready' #task.env.initMarioMode = 2 #task.env.levelDifficulty = 1 task = MarioTask(agent.name, initMarioMode = 2) exp = EpisodicExperiment(task, agent) task.env.initMarioMode = 2 task.env.levelDifficulty = 1 if(agent.initialTraining): exp.doEpisodes(2) agent.newModel() agent.saveModel() else: for i in range(ITERATIONS): #IPython.embed() #if( i == 2): #agent.initialTraining = True; # IPython.embed() print "ITERATION",i f.write('ITERATION %i \n' %i) f.write('___________________________________________________\n') rewards = exp.doEpisodes(1) agent.updateModel() if(agent._getName() == 'Ahude'): num_help = np.vstack((num_help,np.array(agent.getNumHelp()))) mis_match = np.vstack((mis_match,np.array(agent.getMismatch()))) #agent.off = True #rewards = exp.doEpisodes(1) #agent.off = False size = len(rewards[0]) distances = np.vstack((distances,np.array(rewards[0][size-1]))) data = np.vstack((data,np.array(agent.getNumData()))) agent.reset() #agent.notComplete = False print "TRACK COMPLETE" #IPython.embed() # IPython.embed() f.close() plt.figure(2) plt.plot(data,distances) if(agent._getName() == 'Ahude'): plt.figure(1) plt.plot(data,num_help) plt.figure(3) plt.plot(mis_match) plt.show() #agent.saveModel() print "finished"