Python learn Examples

Example #1

tf.reset_default_graph()

global_step = tf.Variable(0, name="global_step", trainable=False)
policy_estimator = vpg.PolicyEstimator(env, learning_rate=0.001)
value_estimator = vpg.ValueEstimator(env, learning_rate=0.1)

with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    # Note, due to randomness in the policy the number of episodes you need varies
    # g = tf.Graph()
    # with g.as_default():
    #     for i in range(10):
    policy_estimator = vpg.learn(
        env,
        policy_estimator,
        value_estimator,
        max_timesteps=10000,
        discount_factor=0.95,
        print_freq=1,
        outdir="/tmp/experiments/continuous/VPG/other")

    # plotting.plot_episode_stats(stats, smoothing_window=10)

    # # Try it out
    # state = env.reset()
    # while 1:
    #     env.render()
    #     action = policy_estimator.predict(state)
    #     next_state, reward, done, _ = env.step(action)
    #     if done:
    #         state = env.reset()
    #     state = next_state

Example #2

tf.reset_default_graph()

global_step = tf.Variable(0, name="global_step", trainable=False)
policy_estimator = vpg.PolicyEstimator(env, learning_rate=0.001)
value_estimator = vpg.ValueEstimator(env, learning_rate=0.1)

with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    # Note, due to randomness in the policy the number of episodes you need varies
    # g = tf.Graph()
    # with g.as_default():
    #     for i in range(10):
    policy_estimator = vpg.learn(
        env,
        policy_estimator,
        value_estimator,
        max_timesteps=100000,
        discount_factor=0.95,
        print_freq=10,
        outdir="/tmp/experiments/VPG/pendulum/100000-2")

    # plotting.plot_episode_stats(stats, smoothing_window=10)

    # Try it out
    state = env.reset()
    while 1:
        env.render()
        action = policy_estimator.predict(state)
        next_state, reward, done, _ = env.step(action)
        if done:
            state = env.reset()
        state = next_state

Example #3

File: train_mountaincar_continuous_pg_baselines.py Project: vmayoral/basic_reinforcement_learning

env = gym.envs.make("MountainCarContinuous-v0")
scaler, featurizer = vpg.preprocess(env)
tf.reset_default_graph()

global_step = tf.Variable(0, name="global_step", trainable=False)
policy_estimator = vpg.PolicyEstimator(env, learning_rate=0.001)
value_estimator = vpg.ValueEstimator(env, learning_rate=0.1)

with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    # Note, due to randomness in the policy the number of episodes you need varies
    # g = tf.Graph()
    # with g.as_default():
    #     for i in range(10):
    policy_estimator = vpg.learn(env, policy_estimator, value_estimator, max_timesteps=10000,
                    discount_factor=0.95,
                    print_freq=1,
                    outdir="/tmp/experiments/continuous/VPG/other")

    # plotting.plot_episode_stats(stats, smoothing_window=10)

    # # Try it out
    # state = env.reset()
    # while 1:
    #     env.render()
    #     action = policy_estimator.predict(state)
    #     next_state, reward, done, _ = env.step(action)
    #     if done:
    #         state = env.reset()
    #     state = next_state

Example #4

File: train_pendulum_pg.py Project: vmayoral/basic_reinforcement_learning

scaler, featurizer = vpg.preprocess(env)
tf.reset_default_graph()

global_step = tf.Variable(0, name="global_step", trainable=False)
policy_estimator = vpg.PolicyEstimator(env, learning_rate=0.001)
value_estimator = vpg.ValueEstimator(env, learning_rate=0.1)

with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    # Note, due to randomness in the policy the number of episodes you need varies
    # g = tf.Graph()
    # with g.as_default():
    #     for i in range(10):
    policy_estimator = vpg.learn(env, policy_estimator, value_estimator,
                    max_timesteps=100000,
                    discount_factor=0.95,
                    print_freq=10,
                    outdir="/tmp/experiments/VPG/pendulum/100000-2")

    # plotting.plot_episode_stats(stats, smoothing_window=10)

    # Try it out
    state = env.reset()
    while 1:
        env.render()
        action = policy_estimator.predict(state)
        next_state, reward, done, _ = env.step(action)
        if done:
            state = env.reset()
        state = next_state

Example #5

tf.reset_default_graph()

global_step = tf.Variable(0, name="global_step", trainable=False)
policy_estimator = vpg.PolicyEstimator(env, learning_rate=0.001)
value_estimator = vpg.ValueEstimator(env, learning_rate=0.1)

with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    # Note, due to randomness in the policy the number of episodes you need varies
    # g = tf.Graph()
    # with g.as_default():
    #     for i in range(10):
    policy_estimator = vpg.learn(env,
                                 policy_estimator,
                                 value_estimator,
                                 max_timesteps=args.max_episode_steps,
                                 discount_factor=0.98,
                                 print_freq=10,
                                 outdir="/tmp/experiments/" +
                                 str(args.environment) + "/VPG/")

    # plotting.plot_episode_stats(stats, smoothing_window=10)

    # # Try it out
    # state = env.reset()
    # while 1:
    #     env.render()
    #     action = policy_estimator.predict(state)
    #     next_state, reward, done, _ = env.step(action)
    #     if done:
    #         state = env.reset()
    #     state = next_state