def generate_env():
env = [bandit.fixed_bandit()]
n_actions = env[0].n_actions()
n_inputs = env[0].n_inputs()
env = bandit.MultiBandit(
env, episode_length=experiment_results_generator.EPISODE_LENGTH)
return env, n_actions, n_inputs
def generate_env():
env = [bandit.fixed_bandit()]
env = bandit.MultiBandit(env,
episode_length=episode_length,
include_steps=True)
n_actions = env.n_actions()
n_inputs = env.n_inputs()
return env, n_actions, n_inputs
import os
from bandits import bandit, ucb_agent, experiment_results_generator
if __name__ == '__main__':
env = [bandit.random_bandit()]
env = bandit.MultiBandit(
env, episode_length=experiment_results_generator.EPISODE_LENGTH)
agent = ucb_agent.UCB1Agent(n_actions=env.n_actions())
experiment = experiment_results_generator.ExperimentResultsGenerator()
experiment.run(env=env, agent=agent)
experiment.save_results(
save_dir=experiment_results_generator.build_experiment_path(__file__))