def train_evaluate(agent_params, num_train=100, num_eval=100):
"""
Run num_train training trials and num_eval evaluation trials for a
LearningAgent initialized with a single set of agent_params.
Returns a TrainEvalPerformance that captures performances and agent info.
"""
sim, e = initialize_simulator_environment(agent_params)
training_perfs = sim.run(num_train)
e.primary_agent.stop_learning()
evaluation_perfs = sim.run(num_eval)
print
agent_info = e.primary_agent.agent_info
return TrainEvalPerformance(training_perfs, evaluation_perfs, agent_info)
def train_evaluate(agent_params, num_train=100, num_eval=100):
"""
Run num_train training trials and num_eval evaluation trials for a
LearningAgent initialized with a single set of agent_params.
Returns a TrainEvalPerformance that captures performances and agent info.
"""
sim, e = initialize_simulator_environment(agent_params)
training_perfs = sim.run(num_train)
e.primary_agent.stop_learning()
evaluation_perfs = sim.run(num_eval)
print
agent_info = e.primary_agent.agent_info
return TrainEvalPerformance(training_perfs, evaluation_perfs, agent_info)
def test_six_agentstates_total(self):
sim, e = initialize_simulator_environment()
agentstates = list(e.primary_agent.generate_all_agentstates())
self.assertEqual(len(agentstates), 6)