FrozenLakeSarsa

In [ ]:

from IPython.display import clear_output
import gymnasium as gym
import numpy as np
from Sarsa import Sarsa
from numpy import loadtxt
import warnings
warnings.simplefilter("ignore")

from IPython.display import clear_output import gymnasium as gym import numpy as np from Sarsa import Sarsa from numpy import loadtxt import warnings warnings.simplefilter("ignore")

In [ ]:

def print_action(action):
    if action == 0:
        print("LEFT")
    elif action == 1: 
        print("DOWN")
    elif action == 2: 
        print("RIGHT")
    elif action == 3:
        print("UP")

def print_action(action): if action == 0: print("LEFT") elif action == 1: print("DOWN") elif action == 2: print("RIGHT") elif action == 3: print("UP")

In [ ]:

# exemplo de ambiente nao determinístico
env = gym.make('FrozenLake-v1', render_mode='ansi').env

# exemplo de ambiente nao determinístico env = gym.make('FrozenLake-v1', render_mode='ansi').env

In [ ]:

# only execute the following lines if you want to create a new q-table
#qlearn = Sarsa(env, alpha=0.2, gamma=0.95, epsilon=0.8, epsilon_min=0.0001, epsilon_dec=0.9999, episodes=50000)
#q_table = qlearn.train('data/q-table-frozen-lake-sarsa.csv','results/frozen_lake_sarsa')
q_table = loadtxt('data/q-table-frozen-lake-sarsa.csv', delimiter=',')

# only execute the following lines if you want to create a new q-table #qlearn = Sarsa(env, alpha=0.2, gamma=0.95, epsilon=0.8, epsilon_min=0.0001, epsilon_dec=0.9999, episodes=50000) #q_table = qlearn.train('data/q-table-frozen-lake-sarsa.csv','results/frozen_lake_sarsa') q_table = loadtxt('data/q-table-frozen-lake-sarsa.csv', delimiter=',')

In [ ]:

env = gym.make('FrozenLake-v1', render_mode='human').env
(state, _) = env.reset()
epochs = 0
rewards = 0
done = False

env = gym.make('FrozenLake-v1', render_mode='human').env (state, _) = env.reset() epochs = 0 rewards = 0 done = False

In [ ]:

while not done:
    action = np.argmax(q_table[state])
    print_action(action)
    state, reward, done, _, info = env.step(action)
    rewards += reward
    epochs += 1

while not done: action = np.argmax(q_table[state]) print_action(action) state, reward, done, _, info = env.step(action) rewards += reward epochs += 1

In [ ]:

print("\n")
print("Timesteps taken: {}".format(epochs))
print("Rewards: {}".format(rewards))

print("\n") print("Timesteps taken: {}".format(epochs)) print("Rewards: {}".format(rewards))