def _update_table_sample():
def _update_table_model():
def _lookup_table():
if __name__ == "__main__":
maze = SimpleMaze()
agent = QAgent(actions=maze.ACTIONS, alpha=0.5, gamma=0.5, explore_strategy='epsilon', epsilon=0.1)
# logging
path = deque() # path in this episode
episode_reward_rates = []
num_episodes = 0
cum_reward = 0
cum_steps = 0
# repeatedly run episodes
while True:
# initialization
maze.reset()
agent.reset(foget_table=False)
action, _ = agent.observe_and_act(observation=None, last_reward=None) # get and random action
path.clear()
episode_reward = 0
episode_steps = 0
# interact and reinforce repeatedly
while not maze.isfinished():
new_observation, reward = maze.interact(action)
action, _ = agent.observe_and_act(observation=new_observation, last_reward=reward)
path.append(new_observation)
episode_reward += reward
episode_steps += 1
print len(path),
cum_steps += episode_steps
cum_reward += episode_reward
num_episodes += 1
episode_reward_rates.append(episode_reward / episode_steps)
if num_episodes % 100 == 0:
print num_episodes, len(agent.q_table), cum_reward, cum_steps, 1.0 * cum_reward / cum_steps#, path
cum_reward = 0
cum_steps = 0
win = 50
cum_steps = 0
# repeatedly run episodes
while True:
maze.reset()
agent.reset()
action, _ = agent.observe_and_act(observation=None, last_reward=None) # get and random action
path.clear()
episode_reward = 0
episode_steps = 0
episode_loss = 0
# print '(',
# interact and reinforce repeatedly
while not maze.isfinished():
new_observation, reward = maze.interact(action)
action, loss = agent.observe_and_act(observation=new_observation, last_reward=reward)
# print new_observation,
# print action,
# print agent.fun_rs_lookup(),
path.append(new_observation)
episode_reward += reward
episode_steps += 1
episode_loss += loss if loss else 0
# print '):',
print len(path),
# print "{:.3f}".format(episode_loss),
# print ""
cum_steps += episode_steps
cum_reward += episode_reward
num_episodes += 1
