def test_get_qval_sym(self, filter_dims, num_filters, strides):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.CNNModel'),
new=SimpleCNNModel):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.MLPModel'),
new=SimpleMLPModel):
qf = DiscreteCNNQFunction(env_spec=self.env.spec,
filter_dims=filter_dims,
num_filters=num_filters,
strides=strides,
dueling=False)
output1 = self.sess.run(qf.q_vals, feed_dict={qf.input: [self.obs]})
obs_dim = self.env.observation_space.shape
action_dim = self.env.action_space.n
input_var = tf.compat.v1.placeholder(tf.float32,
shape=(None, ) + obs_dim)
q_vals = qf.get_qval_sym(input_var, 'another')
output2 = self.sess.run(q_vals, feed_dict={input_var: [self.obs]})
expected_output = np.full(action_dim, 0.5)
assert np.array_equal(output1, output2)
assert np.array_equal(output2[0], expected_output)
def test_obs_not_image(self):
env = self.env
with mock.patch(('metarl.tf.models.'
'categorical_cnn_model.CNNModel._build'),
autospec=True,
side_effect=CNNModel._build) as build:
qf = DiscreteCNNQFunction(env_spec=env.spec,
filters=((5, (3, 3)), ),
strides=(2, ),
dueling=False)
normalized_obs = build.call_args_list[0][0][1]
input_ph = qf.input
assert input_ph == normalized_obs
fake_obs = [np.full(env.spec.observation_space.shape, 255)]
assert (self.sess.run(normalized_obs,
feed_dict={input_ph:
fake_obs}) == 255.).all()
obs_dim = env.spec.observation_space.shape
state_input = tf.compat.v1.placeholder(tf.float32,
shape=(None, ) + obs_dim)
qf.get_qval_sym(state_input, name='another')
normalized_obs = build.call_args_list[1][0][1]
fake_obs = [np.full(env.spec.observation_space.shape, 255)]
assert (self.sess.run(normalized_obs,
feed_dict={state_input:
fake_obs}) == 255).all()
def test_clone(self, filters, strides):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.CNNModel'),
new=SimpleCNNModel):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.MLPModel'),
new=SimpleMLPModel):
qf = DiscreteCNNQFunction(env_spec=self.env.spec,
filters=filters,
strides=strides,
dueling=False)
qf_clone = qf.clone('another_qf')
assert qf_clone._filters == qf._filters
assert qf_clone._strides == qf._strides
def test_get_action_max_pooling(self, filter_dims, num_filters, strides,
pool_strides, pool_shapes):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.CNNModelWithMaxPooling'),
new=SimpleCNNModelWithMaxPooling):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.MLPModel'),
new=SimpleMLPModel):
qf = DiscreteCNNQFunction(env_spec=self.env.spec,
filter_dims=filter_dims,
num_filters=num_filters,
strides=strides,
max_pooling=True,
pool_strides=pool_strides,
pool_shapes=pool_shapes,
dueling=False)
action_dim = self.env.action_space.n
expected_output = np.full(action_dim, 0.5)
outputs = self.sess.run(qf.q_vals, feed_dict={qf.input: [self.obs]})
assert np.array_equal(outputs[0], expected_output)
outputs = self.sess.run(
qf.q_vals, feed_dict={qf.input: [self.obs, self.obs, self.obs]})
for output in outputs:
assert np.array_equal(output, expected_output)
def test_is_pickleable(self, filter_dims, num_filters, strides):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.CNNModel'),
new=SimpleCNNModel):
with mock.patch(('metarl.tf.q_functions.'
'discrete_cnn_q_function.MLPModel'),
new=SimpleMLPModel):
qf = DiscreteCNNQFunction(env_spec=self.env.spec,
filter_dims=filter_dims,
num_filters=num_filters,
strides=strides,
dueling=False)
with tf.compat.v1.variable_scope(
'DiscreteCNNQFunction/Sequential/SimpleMLPModel', reuse=True):
return_var = tf.compat.v1.get_variable('return_var')
# assign it to all one
return_var.load(tf.ones_like(return_var).eval())
output1 = self.sess.run(qf.q_vals, feed_dict={qf.input: [self.obs]})
h_data = pickle.dumps(qf)
with tf.compat.v1.Session(graph=tf.Graph()) as sess:
qf_pickled = pickle.loads(h_data)
output2 = sess.run(qf_pickled.q_vals,
feed_dict={qf_pickled.input: [self.obs]})
assert np.array_equal(output1, output2)
def test_invalid_obs_shape(self, obs_dim):
boxEnv = MetaRLEnv(DummyDiscreteEnv(obs_dim=obs_dim))
with pytest.raises(ValueError):
DiscreteCNNQFunction(env_spec=boxEnv.spec,
filters=((5, (3, 3)), ),
strides=(2, ),
dueling=False)
def run_task(snapshot_config, variant_data, *_):
"""Run task.
Args:
snapshot_config (metarl.experiment.SnapshotConfig): The snapshot
configuration used by LocalRunner to create the snapshotter.
variant_data (dict): Custom arguments for the task.
*_ (object): Ignored by this function.
"""
with LocalTFRunner(snapshot_config=snapshot_config) as runner:
n_epochs = 100
steps_per_epoch = 20
sampler_batch_size = 500
num_timesteps = n_epochs * steps_per_epoch * sampler_batch_size
env = gym.make('PongNoFrameskip-v4')
env = Noop(env, noop_max=30)
env = MaxAndSkip(env, skip=4)
env = EpisodicLife(env)
if 'FIRE' in env.unwrapped.get_action_meanings():
env = FireReset(env)
env = Grayscale(env)
env = Resize(env, 84, 84)
env = ClipReward(env)
env = StackFrames(env, 4)
env = TfEnv(env)
replay_buffer = SimpleReplayBuffer(
env_spec=env.spec,
size_in_transitions=variant_data['buffer_size'],
time_horizon=1)
qf = DiscreteCNNQFunction(env_spec=env.spec,
filter_dims=(8, 4, 3),
num_filters=(32, 64, 64),
strides=(4, 2, 1),
dueling=False)
policy = DiscreteQfDerivedPolicy(env_spec=env.spec, qf=qf)
epilson_greedy_strategy = EpsilonGreedyStrategy(
env_spec=env.spec,
total_timesteps=num_timesteps,
max_epsilon=1.0,
min_epsilon=0.02,
decay_ratio=0.1)
algo = DQN(env_spec=env.spec,
policy=policy,
qf=qf,
exploration_strategy=epilson_greedy_strategy,
replay_buffer=replay_buffer,
qf_lr=1e-4,
discount=0.99,
min_buffer_size=int(1e4),
double_q=False,
n_train_steps=500,
steps_per_epoch=steps_per_epoch,
target_network_update_freq=2,
buffer_batch_size=32)
runner.setup(algo, env)
runner.train(n_epochs=n_epochs, batch_size=sampler_batch_size)
def dqn_pong(ctxt=None, seed=1, buffer_size=int(5e4), max_path_length=None):
"""Train DQN on PongNoFrameskip-v4 environment.
Args:
ctxt (metarl.experiment.ExperimentContext): The experiment
configuration used by LocalRunner to create the snapshotter.
seed (int): Used to seed the random number generator to produce
determinism.
buffer_size (int): Number of timesteps to store in replay buffer.
max_path_length (int): Maximum length of a path after which a path is
considered complete. This is used during testing to minimize the
memory required to store a single path.
"""
set_seed(seed)
with LocalTFRunner(ctxt) as runner:
n_epochs = 100
steps_per_epoch = 20
sampler_batch_size = 500
num_timesteps = n_epochs * steps_per_epoch * sampler_batch_size
env = gym.make('PongNoFrameskip-v4')
env = Noop(env, noop_max=30)
env = MaxAndSkip(env, skip=4)
env = EpisodicLife(env)
if 'FIRE' in env.unwrapped.get_action_meanings():
env = FireReset(env)
env = Grayscale(env)
env = Resize(env, 84, 84)
env = ClipReward(env)
env = StackFrames(env, 4)
env = MetaRLEnv(env, is_image=True)
replay_buffer = PathBuffer(capacity_in_transitions=buffer_size)
qf = DiscreteCNNQFunction(env_spec=env.spec,
filters=(
(32, (8, 8)),
(64, (4, 4)),
(64, (3, 3)),
),
strides=(4, 2, 1),
dueling=False) # yapf: disable
policy = DiscreteQfDerivedPolicy(env_spec=env.spec, qf=qf)
exploration_policy = EpsilonGreedyPolicy(env_spec=env.spec,
policy=policy,
total_timesteps=num_timesteps,
max_epsilon=1.0,
min_epsilon=0.02,
decay_ratio=0.1)
algo = DQN(env_spec=env.spec,
policy=policy,
qf=qf,
exploration_policy=exploration_policy,
replay_buffer=replay_buffer,
qf_lr=1e-4,
discount=0.99,
min_buffer_size=int(1e4),
max_path_length=max_path_length,
double_q=False,
n_train_steps=500,
steps_per_epoch=steps_per_epoch,
target_network_update_freq=2,
buffer_batch_size=32)
runner.setup(algo, env)
runner.train(n_epochs=n_epochs, batch_size=sampler_batch_size)