if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--env', type=str, default='HalfCheetah-v2')
parser.add_argument('--hid', type=int, default=64)
parser.add_argument('--l', type=int, default=2)
parser.add_argument('--gamma', type=float, default=0.99)
parser.add_argument('--seed', '-s', type=int, default=0)
parser.add_argument('--cpu', type=int, default=4)
parser.add_argument('--steps', type=int, default=4000)
parser.add_argument('--epochs', type=int, default=50)
parser.add_argument('--exp_name', type=str, default='vpg')
args = parser.parse_args()
mpi_fork(args.cpu) # run parallel code with mpi
from first_round.utils.run_utils import setup_logger_kwargs
logger_kwargs = setup_logger_kwargs(args.exp_name, args.seed)
vpg(lambda: gym.make(args.env),
actor_critic=core.mlp_actor_critic,
ac_kwargs=dict(hid_size=[args.hid] * args.l),
gamma=args.gamma,
seed=args.seed,
steps_per_epoch=args.steps,
epochs=args.epochs,
logger_kwargs=logger_kwargs)
def td3(env_fn, actor_critic=core.mlp_actor_critic, ac_kwargs=dict(), seed=0,
steps_per_epoch=5000, epochs=100, replay_size=int(1e6), gamma=0.99,
polyak=0.995, pi_lr=1e-3, q_lr=1e-3, batch_size=100, start_steps=10000,
act_noise=0.1, target_noise=0.2, noise_clip=0.5, policy_delay=2,
max_ep_len=1000, logger_kwargs=dict(), save_freq=1):
"""
:param env_fn:
:param actor_critic:
:param ac_kwargs:
:param seed:
:param steps_per_epoch:
:param epochs:
:param replay_size:
:param gamma:
:param polyak:
:param pi_lr:
:param q_lr:
:param batch_size:
:param start_steps:
:param act_noise:
:param target_noise:
:param noise_clip:
:param policy_delay:
:param max_ep_len:
:param logger_kwargs:
:param save_freq:
:return:
"""
logger = EpochLogger(**logger_kwargs)
logger.save_config(locals())
tf.set_random_seed(seed)
np.random.seed(seed)
env, test_env = env_fn(), env_fn()
obs_dim = env.observation_space.shape[0]
act_dim = env.action_space.shape[0]
# we only used index 0 to assume all dimensions share the same limits
act_limit = env.action_space.high[0]
# feed env space info to actor-critic structure
ac_kwargs['action_space'] = env.action_space
# Placeholders
x_ph, a_ph, r_ph, x2_ph, d_ph = core.placeholders(obs_dim, act_dim, None, obs_dim, None)
# Main computation graph
with tf.variable_scope('main'):
pi, q1, q2, q1_pi = actor_critic(x_ph, a_ph, **ac_kwargs)
with tf.variable_scope('target'):
pi_targ, _, _, _ = actor_critic(x2_ph, a_ph, **ac_kwargs)
with tf.variable_scope('target', reuse=True):
# Trick 1: add clipped noise to target actions
epsilon = tf.random_normal(tf.shape(pi_targ), stddev=target_noise)
epsilon = tf.clip_by_value(epsilon, -noise_clip, noise_clip)
a2 = pi_targ + epsilon
a2 = tf.clip_by_value(a2, -act_limit, act_limit)
# Target Q values
_, q1_targ, q2_targ, _ = actor_critic(x2_ph, a2, **ac_kwargs)
# Experience
buf = ReplayBuffer(obs_dim, act_dim, replay_size)
# Count variables
var_counts = type(core.count_vars(scope) for scope in ['main/pi', 'main/q1', 'main/q2', 'main'])
print('\nNumber of parameters: \t pi: %d, \t q1: %d, \t q2: %d, \t total: %d\n' % var_counts)
# Bellman backup
min_q_targ = tf.minimum(q1_targ, q2_targ)
backup = tf.stop_gradient(r_ph + gamma * (1 - d_ph) * min_q_targ)
# Losses
pi_loss = -tf.reduce_mean(q1_pi)
q1_loss = tf.reduce_mean((q1_targ - q1) ** 2)
q2_loss = tf.reduce_mean((q2_targ - q2) ** 2)
q_loss = q1_loss + q2_loss
# Training operations
pi_optimizer = tf.train.AdamOptimizer(learning_rate=pi_lr)
q_optimizer = tf.train.AdamOptimizer(learning_rate=q_lr)
train_pi_op = pi_optimizer.minimize(pi_loss, var_list=get_vars('main/pi'))
train_q_op = q_optimizer.minimize(q_loss, var_list=get_vars('main/q'))
# Polyak averaging
target_update = tf.group([tf.assign(v_targ, polyak * v_targ + (1 - polyak) * v_main)
for v_main, v_targ in zip(get_vars('main'), get_vars('target'))])
# Initializing targets to match main variables
target_init = tf.group([tf.assign(v_targ, v_main)
for v_main, v_targ in zip(get_vars('main'), get_vars('target'))])
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.Session(config=config)
sess.run(tf.global_variables_initializer())
sess.run(target_init)
# Setup model saving
logger.setup_tf_saver(sess, inputs={'x': x_ph, 'a': a_ph},
outputs={'pi': pi, 'q1': q1, 'q2': q2})
def get_action(o, noise_scale):
a = sess.run(pi, feed_dict={x_ph: o.reshape(1, -1)})[0]
a += noise_scale * np.random.randn(act_dim)
return np.clip(a, -act_limit, act_limit)
def test_agent(n=10):
for j in range(n):
o, r, d, ep_ret, ep_len = test_env.reset(), 0, False, 0, 0
while not (d or ep_len == max_ep_len):
o, r, d, _ = test_env.step(get_action(o, 0))
ep_len += 1
ep_ret += r
logger.store(TestEpRet=ep_ret, TestPeLen=ep_len)
start = time.time()
o, r, d, ep_len, ep_ret = env.reset(), 0, False, 0, 0
total_steps = steps_per_epoch * epochs
for t in range(total_steps):
if t > start_steps:
a = get_action(0, act_noise)
else:
a = env.action_space.sample()
o2, r, d, _ = env.step(a)
ep_len += 1
ep_ret += r
d = False if ep_len == max_ep_len else d
# Store experience
buf.store(o, a, r, o2, d)
o = o2
if d or (ep_len == max_ep_len):
"""
Perform updates at the end of a trajectory or when maximum steps reach"""
for j in range(ep_len):
# update at every time step
batch = buf.sample_batch(batch_size)
feed_dict = {x_ph: batch['obs1'],
x2_ph: batch['obs2'],
a_ph: batch['acts'],
r_ph: batch['rews'],
d_ph: batch['done']}
q_steps_ops = [q_loss, q1, q2, train_q_op]
outs = sess.run(q_steps_ops, feed_dict)
logger.store(LossQ=outs[0], Q1Vals=outs[1], Q2Vals=outs[2])
# Trick 3: deplayed policy update
if j % policy_delay == 0:
outs = sess.run([pi_loss, train_pi_op, target_update], feed_dict)
logger.store(LossPi=outs[0])
logger.store(EpRet=ep_ret, EpLen=ep_len)
o, r, d, re_len, ep_ret = env.reset(), 0, False, 0, 0
if t > 0 and t % steps_per_epoch == 0:
epoch = t // steps_per_epoch
# Save model
if (epoch % save_freq == 0) or (epoch == epochs - 1):
logger.save_state({'env': env}, None)
test_agent()
# Logger at the end of epoch
logger.log_tabular('Epoch', epoch)
logger.log_tabular('EpRet', with_min_and_max=True)
logger.log_tabular('TestEpRet', with_min_and_max=True)
logger.log_tabular('EpLen', average_only=True)
logger.log_tabular('TestEpLen', average_only=True)
logger.log_tabular('TotalEnvInteracts', t)
logger.log_tabular('Q1Vals', with_min_and_max=True)
logger.log_tabular('Q2Vals', with_min_and_max=True)
logger.log_tabular('LossPi', average_only=True)
logger.log_tabular('LossQ', average_only=True)
logger.log_tabular('Time', time.time() - start)
logger.dump_tabular()
if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--env', type=str, default='Halfcheetah-v2')
parser.add_argument('--hid', type=int, default=300)
parser.add_argument('--l', type=int, default=1)
parser.add_argument('--gamma', type=float, default=0.99)
parser.add_argument('--seed', type=int, default=0)
parser.add_argument('--epoochs', type=int, default=50)
parser.add_argument('--exp_name', type=str, default='td3')
args = parser.parse_args()
from first_round.utils.run_utils import setup_logger_kwargs
logger_kwargs = setup_logger_kwargs(args.exp_name, args.seed)
td3(lambda: gym.make(args.env), actor_critic=core.mlp_actor_critic,
ac_kwargs=dict(hid_size=[args.hid] * args.l), gamma=args.gamma,
seed=args.seed, epochs=args.epochs, logger_kwargs=logger_kwargs)