30 lines
749 B
Python
30 lines
749 B
Python
|
import gym
|
||
|
|
import numpy as np
|
||
|
|
import gym_random_walk
|
||
|
|
|
||
|
|
env = gym.make('random_walk-v0')
|
||
|
|
|
||
|
|
num_episodes = 20
|
||
|
|
num_steps_per_episode = 200
|
||
|
|
|
||
|
|
collected_rewards = []
|
||
|
|
for i in range(num_episodes):
|
||
|
|
s = env.reset()
|
||
|
|
print ("starting new episode")
|
||
|
|
env.render()
|
||
|
|
print ("started")
|
||
|
|
total_reward = 0
|
||
|
|
done = False
|
||
|
|
for j in range(num_steps_per_episode):
|
||
|
|
a = np.random.randint(env.action_space.n)
|
||
|
|
s1, reward, done, _ = env.step(a)
|
||
|
|
env.render()
|
||
|
|
total_reward += reward
|
||
|
|
s = s1
|
||
|
|
if done:
|
||
|
|
break
|
||
|
|
collected_rewards.append(total_reward)
|
||
|
|
print ("total reward ", total_reward, " after episode: ", j)
|
||
|
|
print ("average score: ", sum(collected_rewards) / num_episodes)
|
||
|
|
print("#########")
|