flake8 complaint

.travis.yml

@@ -10,5 +10,5 @@ install:
     - docker build . -t baselines-test
 
 script:
-    - flake8 . --show-source --statistics
-    - docker run baselines-test pytest -v .
+    - flake8 --select=F,E999 baselines/common baselines/trpo_mpi baselines/ppo2 baselines/a2c baselines/deepq baselines/acer
+    - docker run baselines-test pytest --runslow

Dockerfile

@@ -18,7 +18,6 @@ WORKDIR $CODE_DIR/baselines
 # Clean up pycache and pyc files
 RUN rm -rf __pycache__ && \
     find . -name "*.pyc" -delete && \
-    pip install tensorflow && \
     pip install -e .[test]
 
 

README.md

@@ -15,7 +15,7 @@ sudo apt-get update && sudo apt-get install cmake libopenmpi-dev python3-dev zli
 ```
     
 ### Mac OS X
-Installation of system packages on Mac requires [Homebrew](https://brew.sh). With Homebrew installed, run the following:
+Installation of system packages on Mac requires [Homebrew](https://brew.sh). With Homebrew installed, run the follwing:
 ```bash
 brew install cmake openmpi
 ```
@@ -38,27 +38,20 @@ More thorough tutorial on virtualenvs and options can be found [here](https://vi
 
 
 ## Installation
-- Clone the repo and cd into it:
-    ```bash
-    git clone https://github.com/openai/baselines.git
-    cd baselines
-    ```
-- If you don't have TensorFlow installed already, install your favourite flavor of TensorFlow. In most cases, 
-    ```bash 
-    pip install tensorflow-gpu # if you have a CUDA-compatible gpu and proper drivers
-    ```
-    or 
-    ```bash
-    pip install tensorflow
-    ```
-    should be sufficient. Refer to [TensorFlow installation guide](https://www.tensorflow.org/install/)
-    for more details. 
-
-- Install baselines package
-    ```bash
-    pip install -e .
-    ```
-
+Clone the repo and cd into it:
+```bash
+git clone https://github.com/openai/baselines.git
+cd baselines
+```
+If using virtualenv, create a new virtualenv and activate it
+```bash
+    virtualenv env --python=python3
+    . env/bin/activate
+```
+Install baselines package
+```bash
+pip install -e .
+```
 ### MuJoCo
 Some of the baselines examples use [MuJoCo](http://www.mujoco.org) (multi-joint dynamics in contact) physics simulator, which is proprietary and requires binaries and a license (temporary 30-day license can be obtained from [www.mujoco.org](http://www.mujoco.org)). Instructions on setting up MuJoCo can be found [here](https://github.com/openai/mujoco-py)
 
@@ -69,30 +62,39 @@ pip install pytest
 pytest
 ```
 
+## Subpackages
+
+## Testing the installation
+All unit tests in baselines can be run using pytest runner:
+```
+pip install pytest
+pytest
+```
+
 ## Training models
 Most of the algorithms in baselines repo are used as follows:
 ```bash
-python -m baselines.run --alg=<name of the algorithm> --env=<environment_id> [additional arguments]
+    python -m baselines.run --alg=<name of the algorithm> --env=<environment_id> [additional arguments]
 ```
 ### Example 1. PPO with MuJoCo Humanoid
-For instance, to train a fully-connected network controlling MuJoCo humanoid using PPO2 for 20M timesteps
+For instance, to train a fully-connected network controlling MuJoCo humanoid using a2c for 20M timesteps
 ```bash
-python -m baselines.run --alg=ppo2 --env=Humanoid-v2 --network=mlp --num_timesteps=2e7
+    python -m baselines.run --alg=a2c --env=Humanoid-v2 --network=mlp --num_timesteps=2e7
 ```
 Note that for mujoco environments fully-connected network is default, so we can omit `--network=mlp`
 The hyperparameters for both network and the learning algorithm can be controlled via the command line, for instance:
 ```bash
-python -m baselines.run --alg=ppo2 --env=Humanoid-v2 --network=mlp --num_timesteps=2e7 --ent_coef=0.1 --num_hidden=32 --num_layers=3 --value_network=copy
+    python -m baselines.run --alg=a2c --env=Humanoid-v2 --network=mlp --num_timesteps=2e7 --ent_coef=0.1 --num_hidden=32 --num_layers=3 --value_network=copy
 ```
-will set entropy coefficient to 0.1, and construct fully connected network with 3 layers with 32 hidden units in each, and create a separate network for value function estimation (so that its parameters are not shared with the policy network, but the structure is the same)
+will set entropy coeffient to 0.1, and construct fully connected network with 3 layers with 32 hidden units in each, and create a separate network for value function estimation (so that its parameters are not shared with the policy network, but the structure is the same)
 
-See docstrings in [common/models.py](baselines/common/models.py) for description of network parameters for each type of model, and 
-docstring for [baselines/ppo2/ppo2.py/learn()](baselines/ppo2/ppo2.py#L152) for the description of the ppo2 hyperparamters. 
+See docstrings in [common/models.py](common/models.py) for description of network parameters for each type of model, and 
+docstring for [baselines/ppo2/ppo2.py/learn()](ppo2/ppo2.py) fir the description of the ppo2 hyperparamters. 
 
 ### Example 2. DQN on Atari 
 DQN with Atari is at this point a classics of benchmarks. To run the baselines implementation of DQN on Atari Pong:
 ```
-python -m baselines.run --alg=deepq --env=PongNoFrameskip-v4 --num_timesteps=1e6
+    python -m baselines.run --alg=deepq --env=PongNoFrameskip-v4 --num_timesteps=1e6
 ```
 
 ## Saving, loading and visualizing models
@@ -100,26 +102,16 @@ The algorithms serialization API is not properly unified yet; however, there is
 `--save_path` and `--load_path` command-line option loads the tensorflow state from a given path before training, and saves it after the training, respectively. 
 Let's imagine you'd like to train ppo2 on Atari Pong,  save the model and then later visualize what has it learnt.
 ```bash
-python -m baselines.run --alg=ppo2 --env=PongNoFrameskip-v4 --num_timesteps=2e7 --save_path=~/models/pong_20M_ppo2
+    python -m baselines.run --alg=ppo2 --env=PongNoFrameskip-v4 --num-timesteps=2e7 --save_path=~/models/pong_20M_ppo2
 ```
-This should get to the mean reward per episode about 20. To load and visualize the model, we'll do the following - load the model, train it for 0 steps, and then visualize: 
+This should get to the mean reward per episode about 5k. To load and visualize the model, we'll do the following - load the model, train it for 0 steps, and then visualize: 
 ```bash
-python -m baselines.run --alg=ppo2 --env=PongNoFrameskip-v4 --num_timesteps=0 --load_path=~/models/pong_20M_ppo2 --play
+    python -m baselines.run --alg=ppo2 --env=PongNoFrameskip-v4 --num-timesteps=0 --load_path=~/models/pong_20M_ppo2 --play
 ```
 
 *NOTE:* At the moment Mujoco training uses VecNormalize wrapper for the environment which is not being saved correctly; so loading the models trained on Mujoco will not work well if the environment is recreated. If necessary, you can work around that by replacing RunningMeanStd by TfRunningMeanStd in [baselines/common/vec_env/vec_normalize.py](baselines/common/vec_env/vec_normalize.py#L12). This way, mean and std of environment normalizing wrapper will be saved in tensorflow variables and included in the model file; however, training is slower that way - hence not including it by default
 
 
-## Using baselines with TensorBoard
-Baselines logger can save data in the TensorBoard format. To do so, set environment variables `OPENAI_LOG_FORMAT` and `OPENAI_LOGDIR`:
-```bash
-export OPENAI_LOG_FORMAT='stdout,log,csv,tensorboard' # formats are comma-separated, but for tensorboard you only really need the last one
-export OPENAI_LOGDIR=path/to/tensorboard/data
-```
-And you can now start TensorBoard with:
-```bash
-tensorboard --logdir=$OPENAI_LOGDIR
-```
 ## Subpackages
 
 - [A2C](baselines/a2c)
@@ -145,7 +137,7 @@ respectively. Note that these results may be not on the latest version of the co
 To cite this repository in publications:
 
     @misc{baselines,
-      author = {Dhariwal, Prafulla and Hesse, Christopher and Klimov, Oleg and Nichol, Alex and Plappert, Matthias and Radford, Alec and Schulman, John and Sidor, Szymon and Wu, Yuhuai and Zhokhov, Peter},
+      author = {Dhariwal, Prafulla and Hesse, Christopher and Klimov, Oleg and Nichol, Alex and Plappert, Matthias and Radford, Alec and Schulman, John and Sidor, Szymon and Wu, Yuhuai},
       title = {OpenAI Baselines},
       year = {2017},
       publisher = {GitHub},

baselines/__init__.py

@@ -1,12 +0,0 @@
-# explicitly import sub-packages to register algorithms
-
-import baselines.a2c.a2c
-import baselines.acer.acer
-import baselines.acktr.acktr
-import baselines.deepq.deepq
-import baselines.ddpg.ddpg
-import baselines.ppo2.ppo2
-
-# not really sure why flake8 complains only about trpo_mpi here...
-import baselines.trpo_mpi.trpo_mpi # noqa: F401
-

baselines/a2c/README.md

@@ -2,12 +2,4 @@
 
 - Original paper: https://arxiv.org/abs/1602.01783
 - Baselines blog post: https://blog.openai.com/baselines-acktr-a2c/
-- `python -m baselines.run --alg=a2c --env=PongNoFrameskip-v4` runs the algorithm for 40M frames = 10M timesteps on an Atari Pong. See help (`-h`) for more options
-- also refer to the repo-wide [README.md](../../README.md#training-models)
-
-## Files
-- `run_atari`: file used to run the algorithm.
-- `policies.py`: contains the different versions of the A2C architecture (MlpPolicy, CNNPolicy, LstmPolicy...).
-- `a2c.py`: - Model : class used to initialize the step_model (sampling) and train_model (training)
-	- learn : Main entrypoint for A2C algorithm. Train a policy with given network architecture on a given environment using a2c algorithm.
-- `runner.py`: class used to generates a batch of experiences
+- `python -m baselines.a2c.run_atari` runs the algorithm for 40M frames = 10M timesteps on an Atari game. See help (`-h`) for more options.

baselines/a2c/a2c.py

@@ -2,12 +2,13 @@ import time
 import functools
 import tensorflow as tf
 
-from baselines import logger, registry
+from baselines import logger
 
 from baselines.common import set_global_seeds, explained_variance
 from baselines.common import tf_util
 from baselines.common.policies import build_policy
 
+
 from baselines.a2c.utils import Scheduler, find_trainable_variables
 from baselines.a2c.runner import Runner
 
@@ -15,18 +16,6 @@ from tensorflow import losses
 
 class Model(object):
 
-    """
-    We use this class to :
-        __init__:
-        - Creates the step_model
-        - Creates the train_model
-
-        train():
-        - Make the training part (feedforward and retropropagation of gradients)
-
-        save/load():
-        - Save load the model
-    """
     def __init__(self, policy, env, nsteps,
             ent_coef=0.01, vf_coef=0.5, max_grad_norm=0.5, lr=7e-4,
             alpha=0.99, epsilon=1e-5, total_timesteps=int(80e6), lrschedule='linear'):
@@ -37,10 +26,7 @@ class Model(object):
 
 
         with tf.variable_scope('a2c_model', reuse=tf.AUTO_REUSE):
-            # step_model is used for sampling
             step_model = policy(nenvs, 1, sess)
-
-            # train_model is used to train our network
             train_model = policy(nbatch, nsteps, sess)
 
         A = tf.placeholder(train_model.action.dtype, train_model.action.shape)
@@ -48,45 +34,25 @@ class Model(object):
         R = tf.placeholder(tf.float32, [nbatch])
         LR = tf.placeholder(tf.float32, [])
 
-        # Calculate the loss
-        # Total loss = Policy gradient loss - entropy * entropy coefficient + Value coefficient * value loss
-
-        # Policy loss
         neglogpac = train_model.pd.neglogp(A)
-        # L = A(s,a) * -logpi(a|s)
-        pg_loss = tf.reduce_mean(ADV * neglogpac)
-
-        # Entropy is used to improve exploration by limiting the premature convergence to suboptimal policy.
         entropy = tf.reduce_mean(train_model.pd.entropy())
 
-        # Value loss
+        pg_loss = tf.reduce_mean(ADV * neglogpac)
         vf_loss = losses.mean_squared_error(tf.squeeze(train_model.vf), R)
 
         loss = pg_loss - entropy*ent_coef + vf_loss * vf_coef
 
-        # Update parameters using loss
-        # 1. Get the model parameters
         params = find_trainable_variables("a2c_model")
-
-        # 2. Calculate the gradients
         grads = tf.gradients(loss, params)
         if max_grad_norm is not None:
-            # Clip the gradients (normalize)
             grads, grad_norm = tf.clip_by_global_norm(grads, max_grad_norm)
         grads = list(zip(grads, params))
-        # zip aggregate each gradient with parameters associated
-        # For instance zip(ABCD, xyza) => Ax, By, Cz, Da
-
-        # 3. Make op for one policy and value update step of A2C
         trainer = tf.train.RMSPropOptimizer(learning_rate=LR, decay=alpha, epsilon=epsilon)
-
         _train = trainer.apply_gradients(grads)
 
         lr = Scheduler(v=lr, nvalues=total_timesteps, schedule=lrschedule)
 
         def train(obs, states, rewards, masks, actions, values):
-            # Here we calculate advantage A(s,a) = R + yV(s') - V(s)
-            # rewards = R + yV(s')
             advs = rewards - values
             for step in range(len(obs)):
                 cur_lr = lr.value()
@@ -113,7 +79,6 @@ class Model(object):
         tf.global_variables_initializer().run(session=sess)
 
 
-@registry.register('a2c')
 def learn(
     network,
     env,
@@ -132,21 +97,21 @@ def learn(
     load_path=None,
     **network_kwargs):
 
-    '''
+    ''' 
     Main entrypoint for A2C algorithm. Train a policy with given network architecture on a given environment using a2c algorithm.
 
     Parameters:
     -----------
 
     network:            policy network architecture. Either string (mlp, lstm, lnlstm, cnn_lstm, cnn, cnn_small, conv_only - see baselines.common/models.py for full list)
-                        specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns
+                        specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns 
                         tuple (output_tensor, extra_feed) where output tensor is the last network layer output, extra_feed is None for feed-forward
                         neural nets, and extra_feed is a dictionary describing how to feed state into the network for recurrent neural nets.
                         See baselines.common/policies.py/lstm for more details on using recurrent nets in policies
-
+                
 
     env:                RL environment. Should implement interface similar to VecEnv (baselines.common/vec_env) or be wrapped with DummyVecEnv (baselines.common/vec_env/dummy_vec_env.py)
-
+                    
 
     seed:               seed to make random number sequence in the alorightm reproducible. By default is None which means seed from system noise generator (not reproducible)
 
@@ -163,7 +128,7 @@ def learn(
 
     lr:                 float, learning rate for RMSProp (current implementation has RMSProp hardcoded in) (default: 7e-4)
 
-    lrschedule:         schedule of learning rate. Can be 'linear', 'constant', or a function [0..1] -> [0..1] that takes fraction of the training progress as input and
+    lrschedule:         schedule of learning rate. Can be 'linear', 'constant', or a function [0..1] -> [0..1] that takes fraction of the training progress as input and 
                         returns fraction of the learning rate (specified as lr) as output
 
     epsilon:            float, RMSProp epsilon (stabilizes square root computation in denominator of RMSProp update) (default: 1e-5)
@@ -175,45 +140,31 @@ def learn(
     log_interval:       int, specifies how frequently the logs are printed out (default: 100)
 
     **network_kwargs:   keyword arguments to the policy / network builder. See baselines.common/policies.py/build_policy and arguments to a particular type of network
-                        For instance, 'mlp' network architecture has arguments num_hidden and num_layers.
+                        For instance, 'mlp' network architecture has arguments num_hidden and num_layers. 
 
     '''
-
+    
 
 
     set_global_seeds(seed)
 
-    # Get the nb of env
     nenvs = env.num_envs
     policy = build_policy(env, network, **network_kwargs)
-
-    # Instantiate the model object (that creates step_model and train_model)
+   
     model = Model(policy=policy, env=env, nsteps=nsteps, ent_coef=ent_coef, vf_coef=vf_coef,
         max_grad_norm=max_grad_norm, lr=lr, alpha=alpha, epsilon=epsilon, total_timesteps=total_timesteps, lrschedule=lrschedule)
     if load_path is not None:
         model.load(load_path)
-
-    # Instantiate the runner object
     runner = Runner(env, model, nsteps=nsteps, gamma=gamma)
 
-    # Calculate the batch_size
     nbatch = nenvs*nsteps
-
-    # Start total timer
     tstart = time.time()
-
     for update in range(1, total_timesteps//nbatch+1):
-        # Get mini batch of experiences
         obs, states, rewards, masks, actions, values = runner.run()
-
         policy_loss, value_loss, policy_entropy = model.train(obs, states, rewards, masks, actions, values)
         nseconds = time.time()-tstart
-
-        # Calculate the fps (frame per second)
         fps = int((update*nbatch)/nseconds)
         if update % log_interval == 0 or update == 1:
-            # Calculates if value function is a good predicator of the returns (ev > 1)
-            # or if it's just worse than predicting nothing (ev =< 0)
             ev = explained_variance(values, rewards)
             logger.record_tabular("nupdates", update)
             logger.record_tabular("total_timesteps", update*nbatch)
@@ -222,5 +173,6 @@ def learn(
             logger.record_tabular("value_loss", float(value_loss))
             logger.record_tabular("explained_variance", float(ev))
             logger.dump_tabular()
+    env.close()
     return model
 

baselines/a2c/runner.py

@@ -3,37 +3,22 @@ from baselines.a2c.utils import discount_with_dones
 from baselines.common.runners import AbstractEnvRunner
 
 class Runner(AbstractEnvRunner):
-    """
-    We use this class to generate batches of experiences
 
-    __init__:
-    - Initialize the runner
-
-    run():
-    - Make a mini batch of experiences
-    """
     def __init__(self, env, model, nsteps=5, gamma=0.99):
         super().__init__(env=env, model=model, nsteps=nsteps)
         self.gamma = gamma
         self.batch_action_shape = [x if x is not None else -1 for x in model.train_model.action.shape.as_list()]
         self.ob_dtype = model.train_model.X.dtype.as_numpy_dtype
-
+    
     def run(self):
-        # We initialize the lists that will contain the mb of experiences
         mb_obs, mb_rewards, mb_actions, mb_values, mb_dones = [],[],[],[],[]
         mb_states = self.states
         for n in range(self.nsteps):
-            # Given observations, take action and value (V(s))
-            # We already have self.obs because Runner superclass run self.obs[:] = env.reset() on init
             actions, values, states, _ = self.model.step(self.obs, S=self.states, M=self.dones)
-
-            # Append the experiences
             mb_obs.append(np.copy(self.obs))
             mb_actions.append(actions)
             mb_values.append(values)
             mb_dones.append(self.dones)
-
-            # Take actions in env and look the results
             obs, rewards, dones, _ = self.env.step(actions)
             self.states = states
             self.dones = dones
@@ -43,8 +28,8 @@ class Runner(AbstractEnvRunner):
             self.obs = obs
             mb_rewards.append(rewards)
         mb_dones.append(self.dones)
+        #batch of steps to batch of rollouts
 
-        # Batch of steps to batch of rollouts
         mb_obs = np.asarray(mb_obs, dtype=self.ob_dtype).swapaxes(1, 0).reshape(self.batch_ob_shape)
         mb_rewards = np.asarray(mb_rewards, dtype=np.float32).swapaxes(1, 0)
         mb_actions = np.asarray(mb_actions, dtype=self.model.train_model.action.dtype.name).swapaxes(1, 0)
@@ -55,7 +40,7 @@ class Runner(AbstractEnvRunner):
 
 
         if self.gamma > 0.0:
-            # Discount/bootstrap off value fn
+            #discount/bootstrap off value fn
             last_values = self.model.value(self.obs, S=self.states, M=self.dones).tolist()
             for n, (rewards, dones, value) in enumerate(zip(mb_rewards, mb_dones, last_values)):
                 rewards = rewards.tolist()
@@ -66,7 +51,7 @@ class Runner(AbstractEnvRunner):
                     rewards = discount_with_dones(rewards, dones, self.gamma)
 
                 mb_rewards[n] = rewards
-
+    
         mb_actions = mb_actions.reshape(self.batch_action_shape)
 
         mb_rewards = mb_rewards.flatten()

baselines/acer/README.md

@@ -1,6 +1,4 @@
 # ACER
 
 - Original paper: https://arxiv.org/abs/1611.01224
-- `python -m baselines.run --alg=acer --env=PongNoFrameskip-v4` runs the algorithm for 40M frames = 10M timesteps on an Atari Pong. See help (`-h`) for more options.
-- also refer to the repo-wide [README.md](../../README.md#training-models)
-
+- `python -m baselines.acer.run_atari` runs the algorithm for 40M frames = 10M timesteps on an Atari game. See help (`-h`) for more options.

baselines/acer/acer.py

@@ -2,12 +2,11 @@ import time
 import functools
 import numpy as np
 import tensorflow as tf
-from baselines import logger, registry
+from baselines import logger
 
 from baselines.common import set_global_seeds
 from baselines.common.policies import build_policy
 from baselines.common.tf_util import get_session, save_variables
-from baselines.common.vec_env.vec_frame_stack import VecFrameStack
 
 from baselines.a2c.utils import batch_to_seq, seq_to_batch
 from baselines.a2c.utils import cat_entropy_softmax
@@ -16,7 +15,6 @@ from baselines.a2c.utils import EpisodeStats
 from baselines.a2c.utils import get_by_index, check_shape, avg_norm, gradient_add, q_explained_variance
 from baselines.acer.buffer import Buffer
 from baselines.acer.runner import Runner
-from baselines.acer.defaults import defaults
 
 # remove last step
 def strip(var, nenvs, nsteps, flat = False):
@@ -57,7 +55,8 @@ def q_retrace(R, D, q_i, v, rho_i, nenvs, nsteps, gamma):
 #     return tf.minimum(1 + eps_clip, tf.maximum(1 - eps_clip, ratio))
 
 class Model(object):
-    def __init__(self, policy, ob_space, ac_space, nenvs, nsteps, ent_coef, q_coef, gamma, max_grad_norm, lr,
+    def __init__(self, policy, ob_space, ac_space, nenvs, nsteps, nstack, num_procs,
+                 ent_coef, q_coef, gamma, max_grad_norm, lr,
                  rprop_alpha, rprop_epsilon, total_timesteps, lrschedule,
                  c, trust_region, alpha, delta):
 
@@ -71,15 +70,15 @@ class Model(object):
         MU = tf.placeholder(tf.float32, [nbatch, nact]) # mu's
         LR = tf.placeholder(tf.float32, [])
         eps = 1e-6
-
-        step_ob_placeholder = tf.placeholder(dtype=ob_space.dtype, shape=(nenvs,) + ob_space.shape)
-        train_ob_placeholder = tf.placeholder(dtype=ob_space.dtype, shape=(nenvs*(nsteps+1),) + ob_space.shape)
+    
+        step_ob_placeholder = tf.placeholder(dtype=ob_space.dtype, shape=(nenvs,) + ob_space.shape[:-1] + (ob_space.shape[-1] * nstack,))
+        train_ob_placeholder = tf.placeholder(dtype=ob_space.dtype, shape=(nenvs*(nsteps+1),) + ob_space.shape[:-1] + (ob_space.shape[-1] * nstack,))
         with tf.variable_scope('acer_model', reuse=tf.AUTO_REUSE):
 
             step_model = policy(observ_placeholder=step_ob_placeholder, sess=sess)
             train_model = policy(observ_placeholder=train_ob_placeholder, sess=sess)
 
-
+    
         params = find_trainable_variables("acer_model")
         print("Params {}".format(len(params)))
         for var in params:
@@ -98,10 +97,10 @@ class Model(object):
             polyak_model = policy(observ_placeholder=train_ob_placeholder, sess=sess)
 
         # Notation: (var) = batch variable, (var)s = seqeuence variable, (var)_i = variable index by action at step i
-
+        
         # action probability distributions according to train_model, polyak_model and step_model
         # poilcy.pi is probability distribution parameters; to obtain distribution that sums to 1 need to take softmax
-        train_model_p = tf.nn.softmax(train_model.pi)
+        train_model_p = tf.nn.softmax(train_model.pi)  
         polyak_model_p = tf.nn.softmax(polyak_model.pi)
         step_model_p = tf.nn.softmax(step_model.pi)
         v = tf.reduce_sum(train_model_p * train_model.q, axis = -1) # shape is [nenvs * (nsteps + 1)]
@@ -120,7 +119,7 @@ class Model(object):
         qret = q_retrace(R, D, q_i, v, rho_i, nenvs, nsteps, gamma)
 
         # Calculate losses
-        # Entropy
+        # Entropy   
         # entropy = tf.reduce_mean(strip(train_model.pd.entropy(), nenvs, nsteps))
         entropy = tf.reduce_mean(cat_entropy_softmax(f))
 
@@ -213,8 +212,8 @@ class Model(object):
 
         def _step(observation, **kwargs):
             return step_model._evaluate([step_model.action, step_model_p, step_model.state], observation, **kwargs)
-
-
+                
+                    
 
         self.train = train
         self.save = functools.partial(save_variables, sess=sess, variables=params)
@@ -248,7 +247,6 @@ class Acer():
             # get obs, actions, rewards, mus, dones from buffer.
             obs, actions, rewards, mus, dones, masks = buffer.get()
 
-
         # reshape stuff correctly
         obs = obs.reshape(runner.batch_ob_shape)
         actions = actions.reshape([runner.nbatch])
@@ -271,8 +269,8 @@ class Acer():
                 logger.record_tabular(name, float(val))
             logger.dump_tabular()
 
-@registry.register('acer', defaults=defaults)
-def learn(network, env, seed=None, nsteps=20, total_timesteps=int(80e6), q_coef=0.5, ent_coef=0.01,
+
+def learn(network, env, seed=None, nsteps=20, nstack=4, total_timesteps=int(80e6), q_coef=0.5, ent_coef=0.01,
           max_grad_norm=10, lr=7e-4, lrschedule='linear', rprop_epsilon=1e-5, rprop_alpha=0.99, gamma=0.99,
           log_interval=100, buffer_size=50000, replay_ratio=4, replay_start=10000, c=10.0,
           trust_region=True, alpha=0.99, delta=1, load_path=None, **network_kwargs):
@@ -285,18 +283,18 @@ def learn(network, env, seed=None, nsteps=20, total_timesteps=int(80e6), q_coef=
     ----------
 
     network:            policy network architecture. Either string (mlp, lstm, lnlstm, cnn_lstm, cnn, cnn_small, conv_only - see baselines.common/models.py for full list)
-                        specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns
+                        specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns 
                         tuple (output_tensor, extra_feed) where output tensor is the last network layer output, extra_feed is None for feed-forward
                         neural nets, and extra_feed is a dictionary describing how to feed state into the network for recurrent neural nets.
                         See baselines.common/policies.py/lstm for more details on using recurrent nets in policies
 
-    env:                environment. Needs to be vectorized for parallel environment simulation.
+    env:                environment. Needs to be vectorized for parallel environment simulation. 
                         The environments produced by gym.make can be wrapped using baselines.common.vec_env.DummyVecEnv class.
 
     nsteps:             int, number of steps of the vectorized environment per update (i.e. batch size is nsteps * nenv where
                         nenv is number of environment copies simulated in parallel) (default: 20)
 
-    nstack:             int, size of the frame stack, i.e. number of the frames passed to the step model. Frames are stacked along channel dimension
+    nstack:             int, size of the frame stack, i.e. number of the frames passed to the step model. Frames are stacked along channel dimension 
                         (last image dimension) (default: 4)
 
     total_timesteps:    int, number of timesteps (i.e. number of actions taken in the environment) (default: 80M)
@@ -305,11 +303,11 @@ def learn(network, env, seed=None, nsteps=20, total_timesteps=int(80e6), q_coef=
 
     ent_coef:           float, policy entropy coefficient in the optimization objective (default: 0.01)
 
-    max_grad_norm:      float, gradient norm clipping coefficient. If set to None, no clipping. (default: 10),
-
+    max_grad_norm:      float, gradient norm clipping coefficient. If set to None, no clipping. (default: 10), 
+    
     lr:                 float, learning rate for RMSProp (current implementation has RMSProp hardcoded in) (default: 7e-4)
 
-    lrschedule:         schedule of learning rate. Can be 'linear', 'constant', or a function [0..1] -> [0..1] that takes fraction of the training progress as input and
+    lrschedule:         schedule of learning rate. Can be 'linear', 'constant', or a function [0..1] -> [0..1] that takes fraction of the training progress as input and 
                         returns fraction of the learning rate (specified as lr) as output
 
     rprop_epsilon:      float, RMSProp epsilon (stabilizes square root computation in denominator of RMSProp update) (default: 1e-5)
@@ -327,41 +325,38 @@ def learn(network, env, seed=None, nsteps=20, total_timesteps=int(80e6), q_coef=
     replay_start:       int, the sampling from the replay buffer does not start until replay buffer has at least that many samples (default: 10k)
 
     c:                  float, importance weight clipping factor (default: 10)
-
+    
     trust_region        bool, whether or not algorithms estimates the gradient KL divergence between the old and updated policy and uses it to determine step size  (default: True)
 
     delta:              float, max KL divergence between the old policy and updated policy (default: 1)
 
-    alpha:              float, momentum factor in the Polyak (exponential moving average) averaging of the model parameters (default: 0.99)
+    alpha:              float, momentum factor in the Polyak (exponential moving average) averaging of the model parameters (default: 0.99) 
 
     load_path:          str, path to load the model from (default: None)
 
     **network_kwargs:               keyword arguments to the policy / network builder. See baselines.common/policies.py/build_policy and arguments to a particular type of network
-                                    For instance, 'mlp' network architecture has arguments num_hidden and num_layers.
+                                    For instance, 'mlp' network architecture has arguments num_hidden and num_layers. 
 
     '''
 
     print("Running Acer Simple")
     print(locals())
     set_global_seeds(seed)
-    if not isinstance(env, VecFrameStack):
-        env = VecFrameStack(env, 1)
-
     policy = build_policy(env, network, estimate_q=True, **network_kwargs)
+
     nenvs = env.num_envs
     ob_space = env.observation_space
     ac_space = env.action_space
-
-    nstack = env.nstack
-    model = Model(policy=policy, ob_space=ob_space, ac_space=ac_space, nenvs=nenvs, nsteps=nsteps,
-                  ent_coef=ent_coef, q_coef=q_coef, gamma=gamma,
+    num_procs = len(env.remotes) if hasattr(env, 'remotes') else 1# HACK
+    model = Model(policy=policy, ob_space=ob_space, ac_space=ac_space, nenvs=nenvs, nsteps=nsteps, nstack=nstack,
+                  num_procs=num_procs, ent_coef=ent_coef, q_coef=q_coef, gamma=gamma,
                   max_grad_norm=max_grad_norm, lr=lr, rprop_alpha=rprop_alpha, rprop_epsilon=rprop_epsilon,
                   total_timesteps=total_timesteps, lrschedule=lrschedule, c=c,
                   trust_region=trust_region, alpha=alpha, delta=delta)
 
-    runner = Runner(env=env, model=model, nsteps=nsteps)
+    runner = Runner(env=env, model=model, nsteps=nsteps, nstack=nstack)
     if replay_ratio > 0:
-        buffer = Buffer(env=env, nsteps=nsteps, size=buffer_size)
+        buffer = Buffer(env=env, nsteps=nsteps, nstack=nstack, size=buffer_size)
     else:
         buffer = None
     nbatch = nenvs*nsteps
@@ -375,4 +370,5 @@ def learn(network, env, seed=None, nsteps=20, total_timesteps=int(80e6), q_coef=
             for _ in range(n):
                 acer.call(on_policy=False)  # no simulation steps in this
 
+    env.close()
     return model

baselines/acer/buffer.py

@@ -2,16 +2,11 @@ import numpy as np
 
 class Buffer(object):
     # gets obs, actions, rewards, mu's, (states, masks), dones
-    def __init__(self, env, nsteps, size=50000):
+    def __init__(self, env, nsteps, nstack, size=50000):
         self.nenv = env.num_envs
         self.nsteps = nsteps
-        # self.nh, self.nw, self.nc = env.observation_space.shape
-        self.obs_shape = env.observation_space.shape
-        self.obs_dtype = env.observation_space.dtype
-        self.ac_dtype = env.action_space.dtype
-        self.nc = self.obs_shape[-1]
-        self.nstack = env.nstack
-        self.nc //= self.nstack
+        self.nh, self.nw, self.nc = env.observation_space.shape
+        self.nstack = nstack
         self.nbatch = self.nenv * self.nsteps
         self.size = size // (self.nsteps)  # Each loc contains nenv * nsteps frames, thus total buffer is nenv * size frames
 
@@ -38,11 +33,22 @@ class Buffer(object):
     # Generate stacked frames
     def decode(self, enc_obs, dones):
         # enc_obs has shape [nenvs, nsteps + nstack, nh, nw, nc]
-        # dones has shape [nenvs, nsteps]
+        # dones has shape [nenvs, nsteps, nh, nw, nc]
         # returns stacked obs of shape [nenv, (nsteps + 1), nh, nw, nstack*nc]
-
-        return _stack_obs(enc_obs, dones,
-                          nsteps=self.nsteps)
+        nstack, nenv, nsteps, nh, nw, nc = self.nstack, self.nenv, self.nsteps, self.nh, self.nw, self.nc
+        y = np.empty([nsteps + nstack - 1, nenv, 1, 1, 1], dtype=np.float32)
+        obs = np.zeros([nstack, nsteps + nstack, nenv, nh, nw, nc], dtype=np.uint8)
+        x = np.reshape(enc_obs, [nenv, nsteps + nstack, nh, nw, nc]).swapaxes(1,
+                                                                              0)  # [nsteps + nstack, nenv, nh, nw, nc]
+        y[3:] = np.reshape(1.0 - dones, [nenv, nsteps, 1, 1, 1]).swapaxes(1, 0)  # keep
+        y[:3] = 1.0
+        # y = np.reshape(1 - dones, [nenvs, nsteps, 1, 1, 1])
+        for i in range(nstack):
+            obs[-(i + 1), i:] = x
+            # obs[:,i:,:,:,-(i+1),:] = x
+            x = x[:-1] * y
+            y = y[1:]
+        return np.reshape(obs[:, 3:].transpose((2, 1, 3, 4, 0, 5)), [nenv, (nsteps + 1), nh, nw, nstack * nc])
 
     def put(self, enc_obs, actions, rewards, mus, dones, masks):
         # enc_obs [nenv, (nsteps + nstack), nh, nw, nc]
@@ -50,8 +56,8 @@ class Buffer(object):
         # mus [nenv, nsteps, nact]
 
         if self.enc_obs is None:
-            self.enc_obs = np.empty([self.size] + list(enc_obs.shape), dtype=self.obs_dtype)
-            self.actions = np.empty([self.size] + list(actions.shape), dtype=self.ac_dtype)
+            self.enc_obs = np.empty([self.size] + list(enc_obs.shape), dtype=np.uint8)
+            self.actions = np.empty([self.size] + list(actions.shape), dtype=np.int32)
             self.rewards = np.empty([self.size] + list(rewards.shape), dtype=np.float32)
             self.mus = np.empty([self.size] + list(mus.shape), dtype=np.float32)
             self.dones = np.empty([self.size] + list(dones.shape), dtype=np.bool)
@@ -95,62 +101,3 @@ class Buffer(object):
         mus = take(self.mus)
         masks = take(self.masks)
         return obs, actions, rewards, mus, dones, masks
-
-
-
-def _stack_obs_ref(enc_obs, dones, nsteps):
-    nenv = enc_obs.shape[0]
-    nstack = enc_obs.shape[1] - nsteps
-    nh, nw, nc = enc_obs.shape[2:]
-    obs_dtype = enc_obs.dtype
-    obs_shape = (nh, nw, nc*nstack)
-
-    mask = np.empty([nsteps + nstack - 1, nenv, 1, 1, 1], dtype=np.float32)
-    obs = np.zeros([nstack, nsteps + nstack, nenv, nh, nw, nc], dtype=obs_dtype)
-    x = np.reshape(enc_obs, [nenv, nsteps + nstack, nh, nw, nc]).swapaxes(1, 0)  # [nsteps + nstack, nenv, nh, nw, nc]
-
-    mask[nstack-1:] = np.reshape(1.0 - dones, [nenv, nsteps, 1, 1, 1]).swapaxes(1, 0)  # keep
-    mask[:nstack-1] = 1.0
-
-    # y = np.reshape(1 - dones, [nenvs, nsteps, 1, 1, 1])
-    for i in range(nstack):
-        obs[-(i + 1), i:] = x
-        # obs[:,i:,:,:,-(i+1),:] = x
-        x = x[:-1] * mask
-        mask = mask[1:]
-
-    return np.reshape(obs[:, (nstack-1):].transpose((2, 1, 3, 4, 0, 5)), (nenv, (nsteps + 1)) + obs_shape)
-
-def _stack_obs(enc_obs, dones, nsteps):
-    nenv = enc_obs.shape[0]
-    nstack = enc_obs.shape[1] - nsteps
-    nc = enc_obs.shape[-1]
-
-    obs_ = np.zeros((nenv, nsteps + 1) + enc_obs.shape[2:-1] + (enc_obs.shape[-1] * nstack, ), dtype=enc_obs.dtype)
-    mask = np.ones((nenv, nsteps+1), dtype=enc_obs.dtype)
-    mask[:, 1:] = 1.0 - dones
-    mask = mask.reshape(mask.shape + tuple(np.ones(len(enc_obs.shape)-2, dtype=np.uint8)))
-
-    for i in range(nstack-1, -1, -1):
-        obs_[..., i * nc : (i + 1) * nc] = enc_obs[:, i : i + nsteps + 1, :]
-        if i < nstack-1:
-            obs_[..., i * nc : (i + 1) * nc] *= mask
-            mask[:, 1:, ...] *= mask[:, :-1, ...]
-
-    return obs_
-
-def test_stack_obs():
-    nstack = 7
-    nenv = 1
-    nsteps = 5
-
-    obs_shape = (2, 3, nstack)
-
-    enc_obs_shape = (nenv, nsteps + nstack) + obs_shape[:-1] + (1,)
-    enc_obs = np.random.random(enc_obs_shape)
-    dones = np.random.randint(low=0, high=2, size=(nenv, nsteps))
-
-    stacked_obs_ref = _stack_obs_ref(enc_obs, dones, nsteps=nsteps)
-    stacked_obs_test = _stack_obs(enc_obs, dones, nsteps=nsteps)
-
-    np.testing.assert_allclose(stacked_obs_ref, stacked_obs_test)

baselines/acer/defaults.py

@@ -1,3 +1,4 @@
-defaults = {
-    'atari': dict(lrschedule='constant')
-}
+def atari():
+    return dict(
+        lrschedule='constant'
+    )

baselines/acer/runner.py

@@ -1,31 +1,30 @@
 import numpy as np
 from baselines.common.runners import AbstractEnvRunner
-from baselines.common.vec_env.vec_frame_stack import VecFrameStack
-from gym import spaces
-
 
 class Runner(AbstractEnvRunner):
 
-    def __init__(self, env, model, nsteps):
+    def __init__(self, env, model, nsteps, nstack):
         super().__init__(env=env, model=model, nsteps=nsteps)
-        assert isinstance(env.action_space, spaces.Discrete), 'This ACER implementation works only with discrete action spaces!'
-        assert isinstance(env, VecFrameStack)
-
+        self.nstack = nstack
+        nh, nw, nc = env.observation_space.shape
+        self.nc = nc  # nc = 1 for atari, but just in case
         self.nact = env.action_space.n
         nenv = self.nenv
         self.nbatch = nenv * nsteps
-        self.batch_ob_shape = (nenv*(nsteps+1),) + env.observation_space.shape
-
-        self.obs = env.reset()
-        self.obs_dtype = env.observation_space.dtype
-        self.ac_dtype = env.action_space.dtype
-        self.nstack = self.env.nstack
-        self.nc = self.batch_ob_shape[-1] // self.nstack
+        self.batch_ob_shape = (nenv*(nsteps+1), nh, nw, nc*nstack)
+        self.obs = np.zeros((nenv, nh, nw, nc * nstack), dtype=np.uint8)
+        obs = env.reset()
+        self.update_obs(obs)
 
+    def update_obs(self, obs, dones=None):
+        #self.obs = obs
+        if dones is not None:
+            self.obs *= (1 - dones.astype(np.uint8))[:, None, None, None]
+        self.obs = np.roll(self.obs, shift=-self.nc, axis=3)
+        self.obs[:, :, :, -self.nc:] = obs[:, :, :, :]
 
     def run(self):
-        # enc_obs = np.split(self.obs, self.nstack, axis=3)  # so now list of obs steps
-        enc_obs = np.split(self.env.stackedobs, self.env.nstack, axis=-1)
+        enc_obs = np.split(self.obs, self.nstack, axis=3)  # so now list of obs steps
         mb_obs, mb_actions, mb_mus, mb_dones, mb_rewards = [], [], [], [], []
         for _ in range(self.nsteps):
             actions, mus, states = self.model._step(self.obs, S=self.states, M=self.dones)
@@ -37,15 +36,15 @@ class Runner(AbstractEnvRunner):
             # states information for statefull models like LSTM
             self.states = states
             self.dones = dones
-            self.obs = obs
+            self.update_obs(obs, dones)
             mb_rewards.append(rewards)
-            enc_obs.append(obs[..., -self.nc:])
+            enc_obs.append(obs)
         mb_obs.append(np.copy(self.obs))
         mb_dones.append(self.dones)
 
-        enc_obs = np.asarray(enc_obs, dtype=self.obs_dtype).swapaxes(1, 0)
-        mb_obs = np.asarray(mb_obs, dtype=self.obs_dtype).swapaxes(1, 0)
-        mb_actions = np.asarray(mb_actions, dtype=self.ac_dtype).swapaxes(1, 0)
+        enc_obs = np.asarray(enc_obs, dtype=np.uint8).swapaxes(1, 0)
+        mb_obs = np.asarray(mb_obs, dtype=np.uint8).swapaxes(1, 0)
+        mb_actions = np.asarray(mb_actions, dtype=np.int32).swapaxes(1, 0)
         mb_rewards = np.asarray(mb_rewards, dtype=np.float32).swapaxes(1, 0)
         mb_mus = np.asarray(mb_mus, dtype=np.float32).swapaxes(1, 0)
 

baselines/acktr/README.md

@@ -2,8 +2,4 @@
 
 - Original paper: https://arxiv.org/abs/1708.05144
 - Baselines blog post: https://blog.openai.com/baselines-acktr-a2c/
-- `python -m baselines.run --alg=acktr --env=PongNoFrameskip-v4` runs the algorithm for 40M frames = 10M timesteps on an Atari Pong. See help (`-h`) for more options.
-- also refer to the repo-wide [README.md](../../README.md#training-models)
-
-## ACKTR with continuous action spaces
-The code of ACKTR has been refactored to handle both discrete and continuous action spaces uniformly. In the original version, discrete and continuous action spaces were handled by different code (actkr_disc.py and acktr_cont.py) with little overlap. If interested in the original version of the acktr for continuous action spaces, use `old_acktr_cont` branch. Note that original code performs better on the mujoco tasks than the refactored version; we are still investigating why. 
+- `python -m baselines.acktr.run_atari` runs the algorithm for 40M frames = 10M timesteps on an Atari game. See help (`-h`) for more options.

baselines/acktr/acktr.py

@@ -1,154 +1 @@
-import os.path as osp
-import time
-import functools
-import tensorflow as tf
-from baselines import logger, registry
-
-from baselines.common import set_global_seeds, explained_variance
-from baselines.common.policies import build_policy
-from baselines.common.tf_util import get_session, save_variables, load_variables
-
-from baselines.a2c.runner import Runner
-from baselines.a2c.utils import Scheduler, find_trainable_variables
-from baselines.acktr import kfac
-from baselines.acktr.defaults import defaults
-
-
-class Model(object):
-
-    def __init__(self, policy, ob_space, ac_space, nenvs,total_timesteps, nprocs=32, nsteps=20,
-                 ent_coef=0.01, vf_coef=0.5, vf_fisher_coef=1.0, lr=0.25, max_grad_norm=0.5,
-                 kfac_clip=0.001, lrschedule='linear', is_async=True):
-
-        self.sess = sess = get_session()
-        nbatch = nenvs * nsteps
-        A = tf.placeholder(ac_space.dtype, [nbatch,] + list(ac_space.shape))
-        ADV = tf.placeholder(tf.float32, [nbatch])
-        R = tf.placeholder(tf.float32, [nbatch])
-        PG_LR = tf.placeholder(tf.float32, [])
-        VF_LR = tf.placeholder(tf.float32, [])
-
-        with tf.variable_scope('acktr_model', reuse=tf.AUTO_REUSE):
-            self.model = step_model = policy(nenvs, 1, sess=sess)
-            self.model2 = train_model = policy(nenvs*nsteps, nsteps, sess=sess)
-
-        neglogpac = train_model.pd.neglogp(A)
-        self.logits = train_model.pi
-
-        ##training loss
-        pg_loss = tf.reduce_mean(ADV*neglogpac)
-        entropy = tf.reduce_mean(train_model.pd.entropy())
-        pg_loss = pg_loss - ent_coef * entropy
-        vf_loss = tf.losses.mean_squared_error(tf.squeeze(train_model.vf), R)
-        train_loss = pg_loss + vf_coef * vf_loss
-
-
-        ##Fisher loss construction
-        self.pg_fisher = pg_fisher_loss = -tf.reduce_mean(neglogpac)
-        sample_net = train_model.vf + tf.random_normal(tf.shape(train_model.vf))
-        self.vf_fisher = vf_fisher_loss = - vf_fisher_coef*tf.reduce_mean(tf.pow(train_model.vf - tf.stop_gradient(sample_net), 2))
-        self.joint_fisher = joint_fisher_loss = pg_fisher_loss + vf_fisher_loss
-
-        self.params=params = find_trainable_variables("acktr_model")
-
-        self.grads_check = grads = tf.gradients(train_loss,params)
-
-        with tf.device('/gpu:0'):
-            self.optim = optim = kfac.KfacOptimizer(learning_rate=PG_LR, clip_kl=kfac_clip,\
-                momentum=0.9, kfac_update=1, epsilon=0.01,\
-                stats_decay=0.99, is_async=is_async, cold_iter=10, max_grad_norm=max_grad_norm)
-
-            # update_stats_op = optim.compute_and_apply_stats(joint_fisher_loss, var_list=params)
-            optim.compute_and_apply_stats(joint_fisher_loss, var_list=params)
-            train_op, q_runner = optim.apply_gradients(list(zip(grads,params)))
-        self.q_runner = q_runner
-        self.lr = Scheduler(v=lr, nvalues=total_timesteps, schedule=lrschedule)
-
-        def train(obs, states, rewards, masks, actions, values):
-            advs = rewards - values
-            for step in range(len(obs)):
-                cur_lr = self.lr.value()
-
-            td_map = {train_model.X:obs, A:actions, ADV:advs, R:rewards, PG_LR:cur_lr, VF_LR:cur_lr}
-            if states is not None:
-                td_map[train_model.S] = states
-                td_map[train_model.M] = masks
-
-            policy_loss, value_loss, policy_entropy, _ = sess.run(
-                [pg_loss, vf_loss, entropy, train_op],
-                td_map
-            )
-            return policy_loss, value_loss, policy_entropy
-
-
-        self.train = train
-        self.save = functools.partial(save_variables, sess=sess)
-        self.load = functools.partial(load_variables, sess=sess)
-        self.train_model = train_model
-        self.step_model = step_model
-        self.step = step_model.step
-        self.value = step_model.value
-        self.initial_state = step_model.initial_state
-        tf.global_variables_initializer().run(session=sess)
-
-@registry.register('acktr', defaults=defaults)
-def learn(network, env, seed, total_timesteps=int(40e6), gamma=0.99, log_interval=1, nprocs=32, nsteps=20,
-                 ent_coef=0.01, vf_coef=0.5, vf_fisher_coef=1.0, lr=0.25, max_grad_norm=0.5,
-                 kfac_clip=0.001, save_interval=None, lrschedule='linear', load_path=None, is_async=True, **network_kwargs):
-    set_global_seeds(seed)
-
-
-    if network == 'cnn':
-        network_kwargs['one_dim_bias'] = True
-
-    policy = build_policy(env, network, **network_kwargs)
-
-    nenvs = env.num_envs
-    ob_space = env.observation_space
-    ac_space = env.action_space
-    make_model = lambda : Model(policy, ob_space, ac_space, nenvs, total_timesteps, nprocs=nprocs, nsteps
-                                =nsteps, ent_coef=ent_coef, vf_coef=vf_coef, vf_fisher_coef=
-                                vf_fisher_coef, lr=lr, max_grad_norm=max_grad_norm, kfac_clip=kfac_clip,
-                                lrschedule=lrschedule, is_async=is_async)
-    if save_interval and logger.get_dir():
-        import cloudpickle
-        with open(osp.join(logger.get_dir(), 'make_model.pkl'), 'wb') as fh:
-            fh.write(cloudpickle.dumps(make_model))
-    model = make_model()
-
-    if load_path is not None:
-        model.load(load_path)
-
-    runner = Runner(env, model, nsteps=nsteps, gamma=gamma)
-    nbatch = nenvs*nsteps
-    tstart = time.time()
-    coord = tf.train.Coordinator()
-    if is_async:
-        enqueue_threads = model.q_runner.create_threads(model.sess, coord=coord, start=True)
-    else:
-        enqueue_threads = []
-
-    for update in range(1, total_timesteps//nbatch+1):
-        obs, states, rewards, masks, actions, values = runner.run()
-        policy_loss, value_loss, policy_entropy = model.train(obs, states, rewards, masks, actions, values)
-        model.old_obs = obs
-        nseconds = time.time()-tstart
-        fps = int((update*nbatch)/nseconds)
-        if update % log_interval == 0 or update == 1:
-            ev = explained_variance(values, rewards)
-            logger.record_tabular("nupdates", update)
-            logger.record_tabular("total_timesteps", update*nbatch)
-            logger.record_tabular("fps", fps)
-            logger.record_tabular("policy_entropy", float(policy_entropy))
-            logger.record_tabular("policy_loss", float(policy_loss))
-            logger.record_tabular("value_loss", float(value_loss))
-            logger.record_tabular("explained_variance", float(ev))
-            logger.dump_tabular()
-
-        if save_interval and (update % save_interval == 0 or update == 1) and logger.get_dir():
-            savepath = osp.join(logger.get_dir(), 'checkpoint%.5i'%update)
-            print('Saving to', savepath)
-            model.save(savepath)
-    coord.request_stop()
-    coord.join(enqueue_threads)
-    return model
+from baselines.acktr.acktr_disc import *

baselines/acktr/acktr_cont.py

@@ -0,0 +1,142 @@
+import numpy as np
+import tensorflow as tf
+from baselines import logger
+import baselines.common as common
+from baselines.common import tf_util as U
+from baselines.acktr import kfac
+from baselines.common.filters import ZFilter
+
+def pathlength(path):
+    return path["reward"].shape[0]# Loss function that we'll differentiate to get the policy gradient
+
+def rollout(env, policy, max_pathlength, animate=False, obfilter=None):
+    """
+    Simulate the env and policy for max_pathlength steps
+    """
+    ob = env.reset()
+    prev_ob = np.float32(np.zeros(ob.shape))
+    if obfilter: ob = obfilter(ob)
+    terminated = False
+
+    obs = []
+    acs = []
+    ac_dists = []
+    logps = []
+    rewards = []
+    for _ in range(max_pathlength):
+        if animate:
+            env.render()
+        state = np.concatenate([ob, prev_ob], -1)
+        obs.append(state)
+        ac, ac_dist, logp = policy.act(state)
+        acs.append(ac)
+        ac_dists.append(ac_dist)
+        logps.append(logp)
+        prev_ob = np.copy(ob)
+        scaled_ac = env.action_space.low + (ac + 1.) * 0.5 * (env.action_space.high - env.action_space.low)
+        scaled_ac = np.clip(scaled_ac, env.action_space.low, env.action_space.high)
+        ob, rew, done, _ = env.step(scaled_ac)
+        if obfilter: ob = obfilter(ob)
+        rewards.append(rew)
+        if done:
+            terminated = True
+            break
+    return {"observation" : np.array(obs), "terminated" : terminated,
+            "reward" : np.array(rewards), "action" : np.array(acs),
+            "action_dist": np.array(ac_dists), "logp" : np.array(logps)}
+
+def learn(env, policy, vf, gamma, lam, timesteps_per_batch, num_timesteps,
+    animate=False, callback=None, desired_kl=0.002):
+
+    obfilter = ZFilter(env.observation_space.shape)
+
+    max_pathlength = env.spec.timestep_limit
+    stepsize = tf.Variable(initial_value=np.float32(np.array(0.03)), name='stepsize')
+    inputs, loss, loss_sampled = policy.update_info
+    optim = kfac.KfacOptimizer(learning_rate=stepsize, cold_lr=stepsize*(1-0.9), momentum=0.9, kfac_update=2,\
+                                epsilon=1e-2, stats_decay=0.99, async=1, cold_iter=1,
+                                weight_decay_dict=policy.wd_dict, max_grad_norm=None)
+    pi_var_list = []
+    for var in tf.trainable_variables():
+        if "pi" in var.name:
+            pi_var_list.append(var)
+
+    update_op, q_runner = optim.minimize(loss, loss_sampled, var_list=pi_var_list)
+    do_update = U.function(inputs, update_op)
+    U.initialize()
+
+    # start queue runners
+    enqueue_threads = []
+    coord = tf.train.Coordinator()
+    for qr in [q_runner, vf.q_runner]:
+        assert (qr != None)
+        enqueue_threads.extend(qr.create_threads(tf.get_default_session(), coord=coord, start=True))
+
+    i = 0
+    timesteps_so_far = 0
+    while True:
+        if timesteps_so_far > num_timesteps:
+            break
+        logger.log("********** Iteration %i ************"%i)
+
+        # Collect paths until we have enough timesteps
+        timesteps_this_batch = 0
+        paths = []
+        while True:
+            path = rollout(env, policy, max_pathlength, animate=(len(paths)==0 and (i % 10 == 0) and animate), obfilter=obfilter)
+            paths.append(path)
+            n = pathlength(path)
+            timesteps_this_batch += n
+            timesteps_so_far += n
+            if timesteps_this_batch > timesteps_per_batch:
+                break
+
+        # Estimate advantage function
+        vtargs = []
+        advs = []
+        for path in paths:
+            rew_t = path["reward"]
+            return_t = common.discount(rew_t, gamma)
+            vtargs.append(return_t)
+            vpred_t = vf.predict(path)
+            vpred_t = np.append(vpred_t, 0.0 if path["terminated"] else vpred_t[-1])
+            delta_t = rew_t + gamma*vpred_t[1:] - vpred_t[:-1]
+            adv_t = common.discount(delta_t, gamma * lam)
+            advs.append(adv_t)
+        # Update value function
+        vf.fit(paths, vtargs)
+
+        # Build arrays for policy update
+        ob_no = np.concatenate([path["observation"] for path in paths])
+        action_na = np.concatenate([path["action"] for path in paths])
+        oldac_dist = np.concatenate([path["action_dist"] for path in paths])
+        adv_n = np.concatenate(advs)
+        standardized_adv_n = (adv_n - adv_n.mean()) / (adv_n.std() + 1e-8)
+
+        # Policy update
+        do_update(ob_no, action_na, standardized_adv_n)
+
+        min_stepsize = np.float32(1e-8)
+        max_stepsize = np.float32(1e0)
+        # Adjust stepsize
+        kl = policy.compute_kl(ob_no, oldac_dist)
+        if kl > desired_kl * 2:
+            logger.log("kl too high")
+            tf.assign(stepsize, tf.maximum(min_stepsize, stepsize / 1.5)).eval()
+        elif kl < desired_kl / 2:
+            logger.log("kl too low")
+            tf.assign(stepsize, tf.minimum(max_stepsize, stepsize * 1.5)).eval()
+        else:
+            logger.log("kl just right!")
+
+        logger.record_tabular("EpRewMean", np.mean([path["reward"].sum() for path in paths]))
+        logger.record_tabular("EpRewSEM", np.std([path["reward"].sum()/np.sqrt(len(paths)) for path in paths]))
+        logger.record_tabular("EpLenMean", np.mean([pathlength(path) for path in paths]))
+        logger.record_tabular("KL", kl)
+        if callback:
+            callback()
+        logger.dump_tabular()
+        i += 1
+
+    coord.request_stop()
+    coord.join(enqueue_threads)

baselines/acktr/acktr_disc.py

@@ -0,0 +1,151 @@
+import os.path as osp
+import time
+import functools
+import numpy as np
+import tensorflow as tf
+from baselines import logger
+
+from baselines.common import set_global_seeds, explained_variance
+from baselines.common.policies import build_policy
+from baselines.common.tf_util import get_session, save_variables, load_variables
+
+from baselines.a2c.runner import Runner
+from baselines.a2c.utils import discount_with_dones
+from baselines.a2c.utils import Scheduler, find_trainable_variables
+from baselines.acktr import kfac
+
+
+class Model(object):
+
+    def __init__(self, policy, ob_space, ac_space, nenvs,total_timesteps, nprocs=32, nsteps=20,
+                 ent_coef=0.01, vf_coef=0.5, vf_fisher_coef=1.0, lr=0.25, max_grad_norm=0.5,
+                 kfac_clip=0.001, lrschedule='linear'):
+
+        self.sess = sess = get_session()
+        nact = ac_space.n
+        nbatch = nenvs * nsteps
+        A = tf.placeholder(tf.int32, [nbatch])
+        ADV = tf.placeholder(tf.float32, [nbatch])
+        R = tf.placeholder(tf.float32, [nbatch])
+        PG_LR = tf.placeholder(tf.float32, [])
+        VF_LR = tf.placeholder(tf.float32, [])
+
+        with tf.variable_scope('acktr_model', reuse=tf.AUTO_REUSE):
+            self.model = step_model = policy(nenvs, 1, sess=sess)
+            self.model2 = train_model = policy(nenvs*nsteps, nsteps, sess=sess)
+
+        neglogpac = train_model.pd.neglogp(A)
+        self.logits = logits = train_model.pi
+
+        ##training loss
+        pg_loss = tf.reduce_mean(ADV*neglogpac)
+        entropy = tf.reduce_mean(train_model.pd.entropy())
+        pg_loss = pg_loss - ent_coef * entropy
+        vf_loss = tf.losses.mean_squared_error(tf.squeeze(train_model.vf), R)
+        train_loss = pg_loss + vf_coef * vf_loss
+
+
+        ##Fisher loss construction
+        self.pg_fisher = pg_fisher_loss = -tf.reduce_mean(neglogpac)
+        sample_net = train_model.vf + tf.random_normal(tf.shape(train_model.vf))
+        self.vf_fisher = vf_fisher_loss = - vf_fisher_coef*tf.reduce_mean(tf.pow(train_model.vf - tf.stop_gradient(sample_net), 2))
+        self.joint_fisher = joint_fisher_loss = pg_fisher_loss + vf_fisher_loss
+
+        self.params=params = find_trainable_variables("acktr_model")
+
+        self.grads_check = grads = tf.gradients(train_loss,params)
+
+        with tf.device('/gpu:0'):
+            self.optim = optim = kfac.KfacOptimizer(learning_rate=PG_LR, clip_kl=kfac_clip,\
+                momentum=0.9, kfac_update=1, epsilon=0.01,\
+                stats_decay=0.99, async=1, cold_iter=10, max_grad_norm=max_grad_norm)
+
+            update_stats_op = optim.compute_and_apply_stats(joint_fisher_loss, var_list=params)
+            train_op, q_runner = optim.apply_gradients(list(zip(grads,params)))
+        self.q_runner = q_runner
+        self.lr = Scheduler(v=lr, nvalues=total_timesteps, schedule=lrschedule)
+
+        def train(obs, states, rewards, masks, actions, values):
+            advs = rewards - values
+            for step in range(len(obs)):
+                cur_lr = self.lr.value()
+
+            td_map = {train_model.X:obs, A:actions, ADV:advs, R:rewards, PG_LR:cur_lr}
+            if states is not None:
+                td_map[train_model.S] = states
+                td_map[train_model.M] = masks
+
+            policy_loss, value_loss, policy_entropy, _ = sess.run(
+                [pg_loss, vf_loss, entropy, train_op],
+                td_map
+            )
+            return policy_loss, value_loss, policy_entropy
+
+
+        self.train = train
+        self.save = functools.partial(save_variables, sess=sess)
+        self.load = functools.partial(load_variables, sess=sess)
+        self.train_model = train_model
+        self.step_model = step_model
+        self.step = step_model.step
+        self.value = step_model.value
+        self.initial_state = step_model.initial_state
+        tf.global_variables_initializer().run(session=sess)
+
+def learn(network, env, seed, total_timesteps=int(40e6), gamma=0.99, log_interval=1, nprocs=32, nsteps=20,
+                 ent_coef=0.01, vf_coef=0.5, vf_fisher_coef=1.0, lr=0.25, max_grad_norm=0.5,
+                 kfac_clip=0.001, save_interval=None, lrschedule='linear', load_path=None, **network_kwargs):
+    set_global_seeds(seed)
+
+    
+    if network == 'cnn':
+        network_kwargs['one_dim_bias'] = True
+
+    policy = build_policy(env, network, **network_kwargs)
+
+    nenvs = env.num_envs
+    ob_space = env.observation_space
+    ac_space = env.action_space
+    make_model = lambda : Model(policy, ob_space, ac_space, nenvs, total_timesteps, nprocs=nprocs, nsteps
+                                =nsteps, ent_coef=ent_coef, vf_coef=vf_coef, vf_fisher_coef=
+                                vf_fisher_coef, lr=lr, max_grad_norm=max_grad_norm, kfac_clip=kfac_clip,
+                                lrschedule=lrschedule)
+    if save_interval and logger.get_dir():
+        import cloudpickle
+        with open(osp.join(logger.get_dir(), 'make_model.pkl'), 'wb') as fh:
+            fh.write(cloudpickle.dumps(make_model))
+    model = make_model()
+            
+    if load_path is not None:
+        model.load(load_path)
+
+    runner = Runner(env, model, nsteps=nsteps, gamma=gamma)
+    nbatch = nenvs*nsteps
+    tstart = time.time()
+    coord = tf.train.Coordinator()
+    enqueue_threads = model.q_runner.create_threads(model.sess, coord=coord, start=True)
+    for update in range(1, total_timesteps//nbatch+1):
+        obs, states, rewards, masks, actions, values = runner.run()
+        policy_loss, value_loss, policy_entropy = model.train(obs, states, rewards, masks, actions, values)
+        model.old_obs = obs
+        nseconds = time.time()-tstart
+        fps = int((update*nbatch)/nseconds)
+        if update % log_interval == 0 or update == 1:
+            ev = explained_variance(values, rewards)
+            logger.record_tabular("nupdates", update)
+            logger.record_tabular("total_timesteps", update*nbatch)
+            logger.record_tabular("fps", fps)
+            logger.record_tabular("policy_entropy", float(policy_entropy))
+            logger.record_tabular("policy_loss", float(policy_loss))
+            logger.record_tabular("value_loss", float(value_loss))
+            logger.record_tabular("explained_variance", float(ev))
+            logger.dump_tabular()
+
+        if save_interval and (update % save_interval == 0 or update == 1) and logger.get_dir():
+            savepath = osp.join(logger.get_dir(), 'checkpoint%.5i'%update)
+            print('Saving to', savepath)
+            model.save(savepath)
+    coord.request_stop()
+    coord.join(enqueue_threads)
+    env.close()
+    return model

baselines/acktr/defaults.py

@@ -1,6 +0,0 @@
-defaults = {
-    'mujoco' : dict(
-        nsteps=2500,
-        value_network='copy'
-    )
-}

baselines/acktr/kfac.py

@@ -1,8 +1,6 @@
 import tensorflow as tf
 import numpy as np
 import re
-
- # flake8: noqa F403, F405
 from baselines.acktr.kfac_utils import *
 from functools import reduce
 
@@ -12,14 +10,14 @@ KFAC_DEBUG = False
 
 class KfacOptimizer():
 
-    def __init__(self, learning_rate=0.01, momentum=0.9, clip_kl=0.01, kfac_update=2, stats_accum_iter=60, full_stats_init=False, cold_iter=100, cold_lr=None, is_async=False, async_stats=False, epsilon=1e-2, stats_decay=0.95, blockdiag_bias=False, channel_fac=False, factored_damping=False, approxT2=False, use_float64=False, weight_decay_dict={},max_grad_norm=0.5):
+    def __init__(self, learning_rate=0.01, momentum=0.9, clip_kl=0.01, kfac_update=2, stats_accum_iter=60, full_stats_init=False, cold_iter=100, cold_lr=None, async=False, async_stats=False, epsilon=1e-2, stats_decay=0.95, blockdiag_bias=False, channel_fac=False, factored_damping=False, approxT2=False, use_float64=False, weight_decay_dict={},max_grad_norm=0.5):
         self.max_grad_norm = max_grad_norm
         self._lr = learning_rate
         self._momentum = momentum
         self._clip_kl = clip_kl
         self._channel_fac = channel_fac
         self._kfac_update = kfac_update
-        self._async = is_async
+        self._async = async
         self._async_stats = async_stats
         self._epsilon = epsilon
         self._stats_decay = stats_decay

baselines/acktr/policies.py

@@ -0,0 +1,42 @@
+import numpy as np
+import tensorflow as tf
+from baselines.acktr.utils import dense, kl_div
+import baselines.common.tf_util as U
+
+class GaussianMlpPolicy(object):
+    def __init__(self, ob_dim, ac_dim):
+        # Here we'll construct a bunch of expressions, which will be used in two places:
+        # (1) When sampling actions
+        # (2) When computing loss functions, for the policy update
+        # Variables specific to (1) have the word "sampled" in them,
+        # whereas variables specific to (2) have the word "old" in them
+        ob_no = tf.placeholder(tf.float32, shape=[None, ob_dim*2], name="ob") # batch of observations
+        oldac_na = tf.placeholder(tf.float32, shape=[None, ac_dim], name="ac") # batch of actions previous actions
+        oldac_dist = tf.placeholder(tf.float32, shape=[None, ac_dim*2], name="oldac_dist") # batch of actions previous action distributions
+        adv_n = tf.placeholder(tf.float32, shape=[None], name="adv") # advantage function estimate
+        wd_dict = {}
+        h1 = tf.nn.tanh(dense(ob_no, 64, "h1", weight_init=U.normc_initializer(1.0), bias_init=0.0, weight_loss_dict=wd_dict))
+        h2 = tf.nn.tanh(dense(h1, 64, "h2", weight_init=U.normc_initializer(1.0), bias_init=0.0, weight_loss_dict=wd_dict))
+        mean_na = dense(h2, ac_dim, "mean", weight_init=U.normc_initializer(0.1), bias_init=0.0, weight_loss_dict=wd_dict) # Mean control output
+        self.wd_dict = wd_dict
+        self.logstd_1a = logstd_1a = tf.get_variable("logstd", [ac_dim], tf.float32, tf.zeros_initializer()) # Variance on outputs
+        logstd_1a = tf.expand_dims(logstd_1a, 0)
+        std_1a = tf.exp(logstd_1a)
+        std_na = tf.tile(std_1a, [tf.shape(mean_na)[0], 1])
+        ac_dist = tf.concat([tf.reshape(mean_na, [-1, ac_dim]), tf.reshape(std_na, [-1, ac_dim])], 1)
+        sampled_ac_na = tf.random_normal(tf.shape(ac_dist[:,ac_dim:])) * ac_dist[:,ac_dim:] + ac_dist[:,:ac_dim] # This is the sampled action we'll perform.
+        logprobsampled_n = - tf.reduce_sum(tf.log(ac_dist[:,ac_dim:]), axis=1) - 0.5 * tf.log(2.0*np.pi)*ac_dim - 0.5 * tf.reduce_sum(tf.square(ac_dist[:,:ac_dim] - sampled_ac_na) / (tf.square(ac_dist[:,ac_dim:])), axis=1) # Logprob of sampled action
+        logprob_n = - tf.reduce_sum(tf.log(ac_dist[:,ac_dim:]), axis=1) - 0.5 * tf.log(2.0*np.pi)*ac_dim - 0.5 * tf.reduce_sum(tf.square(ac_dist[:,:ac_dim] - oldac_na) / (tf.square(ac_dist[:,ac_dim:])), axis=1) # Logprob of previous actions under CURRENT policy (whereas oldlogprob_n is under OLD policy)
+        kl = tf.reduce_mean(kl_div(oldac_dist, ac_dist, ac_dim))
+        #kl = .5 * tf.reduce_mean(tf.square(logprob_n - oldlogprob_n)) # Approximation of KL divergence between old policy used to generate actions, and new policy used to compute logprob_n
+        surr = - tf.reduce_mean(adv_n * logprob_n) # Loss function that we'll differentiate to get the policy gradient
+        surr_sampled = - tf.reduce_mean(logprob_n) # Sampled loss of the policy
+        self._act = U.function([ob_no], [sampled_ac_na, ac_dist, logprobsampled_n]) # Generate a new action and its logprob
+        #self.compute_kl = U.function([ob_no, oldac_na, oldlogprob_n], kl) # Compute (approximate) KL divergence between old policy and new policy
+        self.compute_kl = U.function([ob_no, oldac_dist], kl)
+        self.update_info = ((ob_no, oldac_na, adv_n), surr, surr_sampled) # Input and output variables needed for computing loss
+        U.initialize() # Initialize uninitialized TF variables
+
+    def act(self, ob):
+        ac, ac_dist, logp = self._act(ob[None])
+        return ac[0], ac_dist[0], logp[0]

baselines/acktr/run_atari.py

@@ -0,0 +1,23 @@
+#!/usr/bin/env python3
+
+from functools import partial
+
+from baselines import logger
+from baselines.acktr.acktr_disc import learn
+from baselines.common.cmd_util import make_atari_env, atari_arg_parser
+from baselines.common.vec_env.vec_frame_stack import VecFrameStack
+from baselines.common.policies import cnn
+
+def train(env_id, num_timesteps, seed, num_cpu):
+    env = VecFrameStack(make_atari_env(env_id, num_cpu, seed), 4)
+    policy_fn = cnn(env=env, one_dim_bias=True)
+    learn(policy_fn, env, seed, total_timesteps=int(num_timesteps * 1.1), nprocs=num_cpu)
+    env.close()
+
+def main():
+    args = atari_arg_parser().parse_args()
+    logger.configure()
+    train(args.env, num_timesteps=args.num_timesteps, seed=args.seed, num_cpu=32)
+
+if __name__ == '__main__':
+    main()

baselines/acktr/run_mujoco.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python3
+
+import tensorflow as tf
+from baselines import logger
+from baselines.common.cmd_util import make_mujoco_env, mujoco_arg_parser
+from baselines.acktr.acktr_cont import learn
+from baselines.acktr.policies import GaussianMlpPolicy
+from baselines.acktr.value_functions import NeuralNetValueFunction
+
+def train(env_id, num_timesteps, seed):
+    env = make_mujoco_env(env_id, seed)
+
+    with tf.Session(config=tf.ConfigProto()):
+        ob_dim = env.observation_space.shape[0]
+        ac_dim = env.action_space.shape[0]
+        with tf.variable_scope("vf"):
+            vf = NeuralNetValueFunction(ob_dim, ac_dim)
+        with tf.variable_scope("pi"):
+            policy = GaussianMlpPolicy(ob_dim, ac_dim)
+
+        learn(env, policy=policy, vf=vf,
+            gamma=0.99, lam=0.97, timesteps_per_batch=2500,
+            desired_kl=0.002,
+            num_timesteps=num_timesteps, animate=False)
+
+        env.close()
+
+def main():
+    args = mujoco_arg_parser().parse_args()
+    logger.configure()
+    train(args.env, num_timesteps=args.num_timesteps, seed=args.seed)
+
+if __name__ == "__main__":
+    main()

baselines/acktr/value_functions.py

@@ -0,0 +1,50 @@
+from baselines import logger
+import numpy as np
+import baselines.common as common
+from baselines.common import tf_util as U
+import tensorflow as tf
+from baselines.acktr import kfac
+from baselines.acktr.utils import dense
+
+class NeuralNetValueFunction(object):
+    def __init__(self, ob_dim, ac_dim): #pylint: disable=W0613
+        X = tf.placeholder(tf.float32, shape=[None, ob_dim*2+ac_dim*2+2]) # batch of observations
+        vtarg_n = tf.placeholder(tf.float32, shape=[None], name='vtarg')
+        wd_dict = {}
+        h1 = tf.nn.elu(dense(X, 64, "h1", weight_init=U.normc_initializer(1.0), bias_init=0, weight_loss_dict=wd_dict))
+        h2 = tf.nn.elu(dense(h1, 64, "h2", weight_init=U.normc_initializer(1.0), bias_init=0, weight_loss_dict=wd_dict))
+        vpred_n = dense(h2, 1, "hfinal", weight_init=U.normc_initializer(1.0), bias_init=0, weight_loss_dict=wd_dict)[:,0]
+        sample_vpred_n = vpred_n + tf.random_normal(tf.shape(vpred_n))
+        wd_loss = tf.get_collection("vf_losses", None)
+        loss = tf.reduce_mean(tf.square(vpred_n - vtarg_n)) + tf.add_n(wd_loss)
+        loss_sampled = tf.reduce_mean(tf.square(vpred_n - tf.stop_gradient(sample_vpred_n)))
+        self._predict = U.function([X], vpred_n)
+        optim = kfac.KfacOptimizer(learning_rate=0.001, cold_lr=0.001*(1-0.9), momentum=0.9, \
+                                    clip_kl=0.3, epsilon=0.1, stats_decay=0.95, \
+                                    async=1, kfac_update=2, cold_iter=50, \
+                                    weight_decay_dict=wd_dict, max_grad_norm=None)
+        vf_var_list = []
+        for var in tf.trainable_variables():
+            if "vf" in var.name:
+                vf_var_list.append(var)
+
+        update_op, self.q_runner = optim.minimize(loss, loss_sampled, var_list=vf_var_list)
+        self.do_update = U.function([X, vtarg_n], update_op) #pylint: disable=E1101
+        U.initialize() # Initialize uninitialized TF variables
+    def _preproc(self, path):
+        l = pathlength(path)
+        al = np.arange(l).reshape(-1,1)/10.0
+        act = path["action_dist"].astype('float32')
+        X = np.concatenate([path['observation'], act, al, np.ones((l, 1))], axis=1)
+        return X
+    def predict(self, path):
+        return self._predict(self._preproc(path))
+    def fit(self, paths, targvals):
+        X = np.concatenate([self._preproc(p) for p in paths])
+        y = np.concatenate(targvals)
+        logger.record_tabular("EVBefore", common.explained_variance(self._predict(X), y))
+        for _ in range(25): self.do_update(X, y)
+        logger.record_tabular("EVAfter", common.explained_variance(self._predict(X), y))
+
+def pathlength(path):
+    return path["reward"].shape[0]

baselines/bench/__init__.py

@@ -1,2 +1,2 @@
 from baselines.bench.benchmarks import *
-from baselines.bench.monitor import *
+from baselines.bench.monitor import *

baselines/bench/benchmarks.py

@@ -97,19 +97,6 @@ register_benchmark({
     ]
 })
 
-# Bullet
-_bulletsmall = [
-    'InvertedDoublePendulum', 'InvertedPendulum', 'HalfCheetah', 'Reacher', 'Walker2D', 'Hopper', 'Ant'
-]
-_bulletsmall = [e + 'BulletEnv-v0' for e in _bulletsmall]
-
-register_benchmark({
-    'name': 'Bullet1M',
-    'description': '6 mujoco-like tasks from bullet, 1M steps',
-    'tasks': [{'env_id': e, 'trials': 6, 'num_timesteps': int(1e6)} for e in _bulletsmall]
-})
-
-
 # Roboschool
 
 register_benchmark({

baselines/bench/monitor.py

@@ -16,11 +16,21 @@ class Monitor(Wrapper):
     def __init__(self, env, filename, allow_early_resets=False, reset_keywords=(), info_keywords=()):
         Wrapper.__init__(self, env=env)
         self.tstart = time.time()
-        self.results_writer = ResultsWriter(
-            filename,
-            header={"t_start": time.time(), 'env_id' : env.spec and env.spec.id},
-            extra_keys=reset_keywords + info_keywords
-        )
+        if filename is None:
+            self.f = None
+            self.logger = None
+        else:
+            if not filename.endswith(Monitor.EXT):
+                if osp.isdir(filename):
+                    filename = osp.join(filename, Monitor.EXT)
+                else:
+                    filename = filename + "." + Monitor.EXT
+            self.f = open(filename, "wt")
+            self.f.write('#%s\n'%json.dumps({"t_start": self.tstart, 'env_id' : env.spec and env.spec.id}))
+            self.logger = csv.DictWriter(self.f, fieldnames=('r', 'l', 't')+reset_keywords+info_keywords)
+            self.logger.writeheader()
+            self.f.flush()
+
         self.reset_keywords = reset_keywords
         self.info_keywords = info_keywords
         self.allow_early_resets = allow_early_resets
@@ -33,7 +43,10 @@ class Monitor(Wrapper):
         self.current_reset_info = {} # extra info about the current episode, that was passed in during reset()
 
     def reset(self, **kwargs):
-        self.reset_state()
+        if not self.allow_early_resets and not self.needs_reset:
+            raise RuntimeError("Tried to reset an environment before done. If you want to allow early resets, wrap your env with Monitor(env, path, allow_early_resets=True)")
+        self.rewards = []
+        self.needs_reset = False
         for k in self.reset_keywords:
             v = kwargs.get(k)
             if v is None:
@@ -41,21 +54,10 @@ class Monitor(Wrapper):
             self.current_reset_info[k] = v
         return self.env.reset(**kwargs)
 
-    def reset_state(self):
-        if not self.allow_early_resets and not self.needs_reset:
-            raise RuntimeError("Tried to reset an environment before done. If you want to allow early resets, wrap your env with Monitor(env, path, allow_early_resets=True)")
-        self.rewards = []
-        self.needs_reset = False
-
-
     def step(self, action):
         if self.needs_reset:
             raise RuntimeError("Tried to step environment that needs reset")
         ob, rew, done, info = self.env.step(action)
-        self.update(ob, rew, done, info)
-        return (ob, rew, done, info)
-
-    def update(self, ob, rew, done, info):
         self.rewards.append(rew)
         if done:
             self.needs_reset = True
@@ -68,12 +70,12 @@ class Monitor(Wrapper):
             self.episode_lengths.append(eplen)
             self.episode_times.append(time.time() - self.tstart)
             epinfo.update(self.current_reset_info)
-            self.results_writer.write_row(epinfo)
-
-            if isinstance(info, dict):
-                info['episode'] = epinfo
-
+            if self.logger:
+                self.logger.writerow(epinfo)
+                self.f.flush()
+            info['episode'] = epinfo
         self.total_steps += 1
+        return (ob, rew, done, info)
 
     def close(self):
         if self.f is not None:
@@ -94,41 +96,13 @@ class Monitor(Wrapper):
 class LoadMonitorResultsError(Exception):
     pass
 
-
-class ResultsWriter(object):
-    def __init__(self, filename=None, header='', extra_keys=()):
-        self.extra_keys = extra_keys
-        if filename is None:
-            self.f = None
-            self.logger = None
-        else:
-            if not filename.endswith(Monitor.EXT):
-                if osp.isdir(filename):
-                    filename = osp.join(filename, Monitor.EXT)
-                else:
-                    filename = filename + "." + Monitor.EXT
-            self.f = open(filename, "wt")
-            if isinstance(header, dict):
-                header = '# {} \n'.format(json.dumps(header))
-            self.f.write(header)
-            self.logger = csv.DictWriter(self.f, fieldnames=('r', 'l', 't')+tuple(extra_keys))
-            self.logger.writeheader()
-            self.f.flush()
-
-    def write_row(self, epinfo):
-        if self.logger:
-            self.logger.writerow(epinfo)
-            self.f.flush()
-
-
-
 def get_monitor_files(dir):
     return glob(osp.join(dir, "*" + Monitor.EXT))
 
 def load_results(dir):
     import pandas
     monitor_files = (
-        glob(osp.join(dir, "*monitor.json")) +
+        glob(osp.join(dir, "*monitor.json")) + 
         glob(osp.join(dir, "*monitor.csv"))) # get both csv and (old) json files
     if not monitor_files:
         raise LoadMonitorResultsError("no monitor files of the form *%s found in %s" % (Monitor.EXT, dir))

baselines/common/atari_wrappers.py

@@ -213,11 +213,8 @@ class LazyFrames(object):
     def __getitem__(self, i):
         return self._force()[i]
 
-def make_atari(env_id, timelimit=True):
-    # XXX(john): remove timelimit argument after gym is upgraded to allow double wrapping
+def make_atari(env_id):
     env = gym.make(env_id)
-    if not timelimit:
-        env = env.env
     assert 'NoFrameskip' in env.spec.id
     env = NoopResetEnv(env, noop_max=30)
     env = MaxAndSkipEnv(env, skip=4)

baselines/common/cg.py

@@ -31,4 +31,4 @@ def cg(f_Ax, b, cg_iters=10, callback=None, verbose=False, residual_tol=1e-10):
     if callback is not None:
         callback(x)
     if verbose: print(fmtstr % (i+1, rdotr, np.linalg.norm(x)))  # pylint: disable=W0631
-    return x
+    return x

baselines/common/cmd_util.py

@@ -15,66 +15,22 @@ from baselines.bench import Monitor
 from baselines.common import set_global_seeds
 from baselines.common.atari_wrappers import make_atari, wrap_deepmind
 from baselines.common.vec_env.subproc_vec_env import SubprocVecEnv
-from baselines.common.vec_env.dummy_vec_env import DummyVecEnv
-from baselines.common.vec_env.vec_frame_stack import VecFrameStack
 
-from baselines.common import retro_wrappers
-
-def make_vec_env(env_id, env_type, num_env, seed, wrapper_kwargs=None, start_index=0, reward_scale=1.0, gamestate=None, frame_stack_size=1):
+def make_atari_env(env_id, num_env, seed, wrapper_kwargs=None, start_index=0):
     """
-    Create a wrapped, monitored SubprocVecEnv
+    Create a wrapped, monitored SubprocVecEnv for Atari.
     """
     if wrapper_kwargs is None: wrapper_kwargs = {}
     mpi_rank = MPI.COMM_WORLD.Get_rank() if MPI else 0
-    seed = seed + 10000 * mpi_rank if seed is not None else None
-    def make_thunk(rank):
-        return lambda: make_env(
-            env_id=env_id,
-            env_type=env_type,
-            subrank = rank,
-            seed=seed,
-            reward_scale=reward_scale,
-            gamestate=gamestate,
-            wrapper_kwargs=wrapper_kwargs
-        )
-
+    def make_env(rank): # pylint: disable=C0111
+        def _thunk():
+            env = make_atari(env_id)
+            env.seed(seed + 10000*mpi_rank + rank if seed is not None else None)
+            env = Monitor(env, logger.get_dir() and os.path.join(logger.get_dir(), str(mpi_rank) + '.' + str(rank)))
+            return wrap_deepmind(env, **wrapper_kwargs)
+        return _thunk
     set_global_seeds(seed)
-    if num_env > 1:
-        venv = SubprocVecEnv([make_thunk(i + start_index) for i in range(num_env)])
-    else:
-        venv = DummyVecEnv([make_thunk(start_index)])
-
-    if frame_stack_size > 1:
-        venv = VecFrameStack(venv, frame_stack_size)
-
-    return venv
-
-
-
-def make_env(env_id, env_type, subrank=0, seed=None, reward_scale=1.0, gamestate=None, wrapper_kwargs={}):
-    mpi_rank = MPI.COMM_WORLD.Get_rank() if MPI else 0
-    if env_type == 'atari':
-        env = make_atari(env_id)
-    elif env_type == 'retro':
-        import retro
-        gamestate = gamestate or retro.State.DEFAULT
-        env = retro_wrappers.make_retro(game=env_id, max_episode_steps=10000, use_restricted_actions=retro.Actions.DISCRETE, state=gamestate)
-    else:
-        env = gym.make(env_id)
-
-    env.seed(seed + subrank if seed is not None else None)
-    env = Monitor(env,
-                  logger.get_dir() and os.path.join(logger.get_dir(), str(mpi_rank) + '.' + str(subrank)),
-                  allow_early_resets=True)
-
-    if env_type == 'atari':
-         return wrap_deepmind(env, **wrapper_kwargs)
-    elif reward_scale != 1:
-         return retro_wrappers.RewardScaler(env, reward_scale)
-    else:
-        return env
-
-
+    return SubprocVecEnv([make_env(i + start_index) for i in range(num_env)])
 
 def make_mujoco_env(env_id, seed, reward_scale=1.0):
     """
@@ -84,12 +40,13 @@ def make_mujoco_env(env_id, seed, reward_scale=1.0):
     myseed = seed  + 1000 * rank if seed is not None else None
     set_global_seeds(myseed)
     env = gym.make(env_id)
-    logger_path = None if logger.get_dir() is None else os.path.join(logger.get_dir(), str(rank))
-    env = Monitor(env, logger_path, allow_early_resets=True)
+    env = Monitor(env, os.path.join(logger.get_dir(), str(rank)), allow_early_resets=True)
     env.seed(seed)
+
     if reward_scale != 1.0:
         from baselines.common.retro_wrappers import RewardScaler
         env = RewardScaler(env, reward_scale)
+
     return env
 
 def make_robotics_env(env_id, seed, rank=0):
@@ -131,7 +88,7 @@ def common_arg_parser():
     parser.add_argument('--env', help='environment ID', type=str, default='Reacher-v2')
     parser.add_argument('--seed', help='RNG seed', type=int, default=None)
     parser.add_argument('--alg', help='Algorithm', type=str, default='ppo2')
-    parser.add_argument('--num_timesteps', type=float, default=1e6),
+    parser.add_argument('--num_timesteps', type=float, default=1e6), 
     parser.add_argument('--network', help='network type (mlp, cnn, lstm, cnn_lstm, conv_only)', default=None)
     parser.add_argument('--gamestate', help='game state to load (so far only used in retro games)', default=None)
     parser.add_argument('--num_env', help='Number of environment copies being run in parallel. When not specified, set to number of cpus for Atari, and to 1 for Mujoco', default=None, type=int)
@@ -156,18 +113,14 @@ def parse_unknown_args(args):
     Parse arguments not consumed by arg parser into a dicitonary
     """
     retval = {}
-    preceded_by_key = False
     for arg in args:
-        if arg.startswith('--'):
-            if '=' in arg:
-                key = arg.split('=')[0][2:]
-                value = arg.split('=')[1]
-                retval[key] = value
-            else:
-                key = arg[2:]
-                preceded_by_key = True
-        elif preceded_by_key:
-            retval[key] = arg
-            preceded_by_key = False
+        assert arg.startswith('--')
+        assert '=' in arg, 'cannot parse arg {}'.format(arg)
+        key = arg.split('=')[0][2:]
+        value = arg.split('=')[1]
+        retval[key] = value
 
     return retval
+
+
+

baselines/common/console_util.py

@@ -2,8 +2,6 @@ from __future__ import print_function
 from contextlib import contextmanager
 import numpy as np
 import time
-import shlex
-import subprocess
 
 # ================================================================
 # Misc
@@ -39,7 +37,7 @@ color2num = dict(
     crimson=38
 )
 
-def colorize(string, color='green', bold=False, highlight=False):
+def colorize(string, color, bold=False, highlight=False):
     attr = []
     num = color2num[color]
     if highlight: num += 10
@@ -47,25 +45,6 @@ def colorize(string, color='green', bold=False, highlight=False):
     if bold: attr.append('1')
     return '\x1b[%sm%s\x1b[0m' % (';'.join(attr), string)
 
-def print_cmd(cmd, dry=False):
-    if isinstance(cmd, str):  # for shell=True
-        pass
-    else:
-        cmd = ' '.join(shlex.quote(arg) for arg in cmd)
-    print(colorize(('CMD: ' if not dry else 'DRY: ') + cmd))
-
-
-def get_git_commit(cwd=None):
-    return subprocess.check_output(['git', 'rev-parse', '--short', 'HEAD'], cwd=cwd).decode('utf8')
-
-def get_git_commit_message(cwd=None):
-    return subprocess.check_output(['git', 'show', '-s', '--format=%B', 'HEAD'], cwd=cwd).decode('utf8')
-
-def ccap(cmd, dry=False, env=None, **kwargs):
-    print_cmd(cmd, dry)
-    if not dry:
-        subprocess.check_call(cmd, env=env, **kwargs)
-
 
 MESSAGE_DEPTH = 0
 

baselines/common/distributions.py

@@ -23,13 +23,6 @@ class Pd(object):
         raise NotImplementedError
     def logp(self, x):
         return - self.neglogp(x)
-    def get_shape(self):
-        return self.flatparam().shape
-    @property
-    def shape(self):
-        return self.get_shape()
-    def __getitem__(self, idx):
-        return self.__class__(self.flatparam()[idx])
 
 class PdType(object):
     """
@@ -53,9 +46,6 @@ class PdType(object):
     def sample_placeholder(self, prepend_shape, name=None):
         return tf.placeholder(dtype=self.sample_dtype(), shape=prepend_shape+self.sample_shape(), name=name)
 
-    def __eq__(self, other):
-        return (type(self) == type(other)) and (self.__dict__ == other.__dict__)
-
 class CategoricalPdType(PdType):
     def __init__(self, ncat):
         self.ncat = ncat
@@ -117,9 +107,6 @@ class BernoulliPdType(PdType):
         return [self.size]
     def sample_dtype(self):
         return tf.int32
-    def pdfromlatent(self, latent_vector, init_scale=1.0, init_bias=0.0):
-        pdparam = fc(latent_vector, 'pi', self.size, init_scale=init_scale, init_bias=init_bias)
-        return self.pdfromflat(pdparam), pdparam
 
 # WRONG SECOND DERIVATIVES
 # class CategoricalPd(Pd):
@@ -151,30 +138,14 @@ class CategoricalPd(Pd):
         return self.logits
     def mode(self):
         return tf.argmax(self.logits, axis=-1)
-
-    @property
-    def mean(self):
-        return tf.nn.softmax(self.logits)
     def neglogp(self, x):
         # return tf.nn.sparse_softmax_cross_entropy_with_logits(logits=self.logits, labels=x)
         # Note: we can't use sparse_softmax_cross_entropy_with_logits because
         #       the implementation does not allow second-order derivatives...
-        if x.dtype in {tf.uint8, tf.int32, tf.int64}:
-            # one-hot encoding
-            x_shape_list = x.shape.as_list()
-            logits_shape_list = self.logits.get_shape().as_list()[:-1]
-            for xs, ls in zip(x_shape_list, logits_shape_list):
-                if xs is not None and ls is not None:
-                    assert xs == ls, 'shape mismatch: {} in x vs {} in logits'.format(xs, ls)
-
-            x = tf.one_hot(x, self.logits.get_shape().as_list()[-1])
-        else:
-            # already encoded
-            assert x.shape.as_list() == self.logits.shape.as_list()
-
+        one_hot_actions = tf.one_hot(x, self.logits.get_shape().as_list()[-1])
         return tf.nn.softmax_cross_entropy_with_logits_v2(
             logits=self.logits,
-            labels=x)
+            labels=one_hot_actions)
     def kl(self, other):
         a0 = self.logits - tf.reduce_max(self.logits, axis=-1, keepdims=True)
         a1 = other.logits - tf.reduce_max(other.logits, axis=-1, keepdims=True)
@@ -243,16 +214,12 @@ class DiagGaussianPd(Pd):
     def fromflat(cls, flat):
         return cls(flat)
 
-
 class BernoulliPd(Pd):
     def __init__(self, logits):
         self.logits = logits
         self.ps = tf.sigmoid(logits)
     def flatparam(self):
         return self.logits
-    @property
-    def mean(self):
-        return self.ps
     def mode(self):
         return tf.round(self.ps)
     def neglogp(self, x):

baselines/common/filters.py

@@ -0,0 +1,98 @@
+from .running_stat import RunningStat
+from collections import deque
+import numpy as np
+
+class Filter(object):
+    def __call__(self, x, update=True):
+        raise NotImplementedError
+    def reset(self):
+        pass
+
+class IdentityFilter(Filter):
+    def __call__(self, x, update=True):
+        return x
+
+class CompositionFilter(Filter):
+    def __init__(self, fs):
+        self.fs = fs
+    def __call__(self, x, update=True):
+        for f in self.fs:
+            x = f(x)
+        return x
+    def output_shape(self, input_space):
+        out = input_space.shape
+        for f in self.fs:
+            out = f.output_shape(out)
+        return out
+
+class ZFilter(Filter):
+    """
+    y = (x-mean)/std
+    using running estimates of mean,std
+    """
+
+    def __init__(self, shape, demean=True, destd=True, clip=10.0):
+        self.demean = demean
+        self.destd = destd
+        self.clip = clip
+
+        self.rs = RunningStat(shape)
+
+    def __call__(self, x, update=True):
+        if update: self.rs.push(x)
+        if self.demean:
+            x = x - self.rs.mean
+        if self.destd:
+            x = x / (self.rs.std+1e-8)
+        if self.clip:
+            x = np.clip(x, -self.clip, self.clip)
+        return x
+    def output_shape(self, input_space):
+        return input_space.shape
+
+class AddClock(Filter):
+    def __init__(self):
+        self.count = 0
+    def reset(self):
+        self.count = 0
+    def __call__(self, x, update=True):
+        return np.append(x, self.count/100.0)
+    def output_shape(self, input_space):
+        return (input_space.shape[0]+1,)
+
+class FlattenFilter(Filter):
+    def __call__(self, x, update=True):
+        return x.ravel()
+    def output_shape(self, input_space):
+        return (int(np.prod(input_space.shape)),)
+
+class Ind2OneHotFilter(Filter):
+    def __init__(self, n):
+        self.n = n
+    def __call__(self, x, update=True):
+        out = np.zeros(self.n)
+        out[x] = 1
+        return out
+    def output_shape(self, input_space):
+        return (input_space.n,)
+
+class DivFilter(Filter):
+    def __init__(self, divisor):
+        self.divisor = divisor
+    def __call__(self, x, update=True):
+        return x / self.divisor
+    def output_shape(self, input_space):
+        return input_space.shape
+
+class StackFilter(Filter):
+    def __init__(self, length):
+        self.stack = deque(maxlen=length)
+    def reset(self):
+        self.stack.clear()
+    def __call__(self, x, update=True):
+        self.stack.append(x)
+        while len(self.stack) < self.stack.maxlen:
+            self.stack.append(x)
+        return np.concatenate(self.stack, axis=-1)
+    def output_shape(self, input_space):
+        return input_space.shape[:-1] + (input_space.shape[-1] * self.stack.maxlen,)

baselines/common/identity_env.py

@@ -0,0 +1,30 @@
+from gym import Env
+from gym.spaces import Discrete
+
+
+class IdentityEnv(Env):
+    def __init__(
+            self,
+            dim,
+            ep_length=100,
+    ):
+
+        self.action_space = Discrete(dim)
+        self.reset()
+
+    def reset(self):
+        self._choose_next_state()
+        self.observation_space = self.action_space
+
+        return self.state
+
+    def step(self, actions):
+        rew = self._get_reward(actions)
+        self._choose_next_state()
+        return self.state, rew, False, {}
+
+    def _choose_next_state(self):
+        self.state = self.action_space.sample()
+
+    def _get_reward(self, actions):
+        return 1 if self.state == actions else 0

baselines/common/input.py

@@ -2,15 +2,15 @@ import tensorflow as tf
 from gym.spaces import Discrete, Box
 
 def observation_placeholder(ob_space, batch_size=None, name='Ob'):
-    '''
+    ''' 
     Create placeholder to feed observations into of the size appropriate to the observation space
-
+    
     Parameters:
     ----------
 
     ob_space: gym.Space     observation space
-
-    batch_size: int         size of the batch to be fed into input. Can be left None in most cases.
+    
+    batch_size: int         size of the batch to be fed into input. Can be left None in most cases. 
 
     name: str               name of the placeholder
 
@@ -27,9 +27,9 @@ def observation_placeholder(ob_space, batch_size=None, name='Ob'):
 
 
 def observation_input(ob_space, batch_size=None, name='Ob'):
-    '''
-    Create placeholder to feed observations into of the size appropriate to the observation space, and add input
-    encoder of the appropriate type.
+    ''' 
+    Create placeholder to feed observations into of the size appropriate to the observation space, and add input 
+    encoder of the appropriate type. 
     '''
 
     placeholder = observation_placeholder(ob_space, batch_size, name)
@@ -41,9 +41,9 @@ def encode_observation(ob_space, placeholder):
 
     Parameters:
     ----------
-
+    
     ob_space: gym.Space             observation space
-
+    
     placeholder: tf.placeholder     observation input placeholder
     '''
     if isinstance(ob_space, Discrete):

baselines/common/math_util.py

@@ -82,4 +82,4 @@ def test_discount_with_boundaries():
         2 + gamma * 3,
         3,
         4
-    ])
+    ])

baselines/common/misc_util.py

@@ -76,9 +76,10 @@ def set_global_seeds(i):
     myseed = i  + 1000 * rank if i is not None else None
     try:
         import tensorflow as tf
-        tf.set_random_seed(myseed)
     except ImportError:
         pass
+    else:
+        tf.set_random_seed(myseed)
     np.random.seed(myseed)
     random.seed(myseed)
 

baselines/common/models.py

@@ -5,13 +5,6 @@ from baselines.a2c.utils import conv, fc, conv_to_fc, batch_to_seq, seq_to_batch
 from baselines.common.mpi_running_mean_std import RunningMeanStd
 import tensorflow.contrib.layers as layers
 
-mapping = {}
-
-def register(name):
-    def _thunk(func):
-        mapping[name] = func
-        return func
-    return _thunk
 
 def nature_cnn(unscaled_images, **conv_kwargs):
     """
@@ -27,93 +20,58 @@ def nature_cnn(unscaled_images, **conv_kwargs):
     return activ(fc(h3, 'fc1', nh=512, init_scale=np.sqrt(2)))
 
 
-@register("mlp")
-def mlp(num_layers=2, num_hidden=64, activation=tf.tanh, layer_norm=False):
+def mlp(num_layers=2, num_hidden=64, activation=tf.tanh):
     """
-    Stack of fully-connected layers to be used in a policy / q-function approximator
+    Simple fully connected layer policy. Separate stacks of fully-connected layers are used for policy and value function estimation.
+    More customized fully-connected policies can be obtained by using PolicyWithV class directly.
 
     Parameters:
     ----------
 
     num_layers: int                 number of fully-connected layers (default: 2)
-
+    
     num_hidden: int                 size of fully-connected layers (default: 64)
-
+    
     activation:                     activation function (default: tf.tanh)
-
+        
     Returns:
     -------
 
-    function that builds fully connected network with a given input tensor / placeholder
-    """
+    function that builds fully connected network with a given input placeholder
+    """        
     def network_fn(X):
         h = tf.layers.flatten(X)
         for i in range(num_layers):
-            h = fc(h, 'mlp_fc{}'.format(i), nh=num_hidden, init_scale=np.sqrt(2))
-            if layer_norm:
-                h = tf.contrib.layers.layer_norm(h, center=True, scale=True)
-            h = activation(h)
-
-        return h
+            h = activation(fc(h, 'mlp_fc{}'.format(i), nh=num_hidden, init_scale=np.sqrt(2)))
+        return h, None
 
     return network_fn
+  
 
-
-@register("cnn")
 def cnn(**conv_kwargs):
     def network_fn(X):
-        return nature_cnn(X, **conv_kwargs)
+        return nature_cnn(X, **conv_kwargs), None
     return network_fn
 
-
-@register("cnn_small")
 def cnn_small(**conv_kwargs):
     def network_fn(X):
         h = tf.cast(X, tf.float32) / 255.
-
+        
         activ = tf.nn.relu
         h = activ(conv(h, 'c1', nf=8, rf=8, stride=4, init_scale=np.sqrt(2), **conv_kwargs))
         h = activ(conv(h, 'c2', nf=16, rf=4, stride=2, init_scale=np.sqrt(2), **conv_kwargs))
         h = conv_to_fc(h)
         h = activ(fc(h, 'fc1', nh=128, init_scale=np.sqrt(2)))
-        return h
+        return h, None
     return network_fn
 
 
-@register("lstm")
+
 def lstm(nlstm=128, layer_norm=False):
-    """
-    Builds LSTM (Long-Short Term Memory) network to be used in a policy.
-    Note that the resulting function returns not only the output of the LSTM
-    (i.e. hidden state of lstm for each step in the sequence), but also a dictionary
-    with auxiliary tensors to be set as policy attributes.
-
-    Specifically,
-        S is a placeholder to feed current state (LSTM state has to be managed outside policy)
-        M is a placeholder for the mask (used to mask out observations after the end of the episode, but can be used for other purposes too)
-        initial_state is a numpy array containing initial lstm state (usually zeros)
-        state is the output LSTM state (to be fed into S at the next call)
-
-
-    An example of usage of lstm-based policy can be found here: common/tests/test_doc_examples.py/test_lstm_example
-
-    Parameters:
-    ----------
-
-    nlstm: int          LSTM hidden state size
-
-    layer_norm: bool    if True, layer-normalized version of LSTM is used
-
-    Returns:
-    -------
-
-    function that builds LSTM with a given input tensor / placeholder
-    """
-
     def network_fn(X, nenv=1):
-        nbatch = X.shape[0]
+        nbatch = X.shape[0] 
         nsteps = nbatch // nenv
-
+         
         h = tf.layers.flatten(X)
 
         M = tf.placeholder(tf.float32, [nbatch]) #mask (done t-1)
@@ -126,7 +84,7 @@ def lstm(nlstm=128, layer_norm=False):
             h5, snew = utils.lnlstm(xs, ms, S, scope='lnlstm', nh=nlstm)
         else:
             h5, snew = utils.lstm(xs, ms, S, scope='lstm', nh=nlstm)
-
+            
         h = seq_to_batch(h5)
         initial_state = np.zeros(S.shape.as_list(), dtype=float)
 
@@ -135,14 +93,13 @@ def lstm(nlstm=128, layer_norm=False):
     return network_fn
 
 
-@register("cnn_lstm")
 def cnn_lstm(nlstm=128, layer_norm=False, **conv_kwargs):
     def network_fn(X, nenv=1):
-        nbatch = X.shape[0]
+        nbatch = X.shape[0] 
         nsteps = nbatch // nenv
-
+         
         h = nature_cnn(X, **conv_kwargs)
-
+       
         M = tf.placeholder(tf.float32, [nbatch]) #mask (done t-1)
         S = tf.placeholder(tf.float32, [nenv, 2*nlstm]) #states
 
@@ -153,7 +110,7 @@ def cnn_lstm(nlstm=128, layer_norm=False, **conv_kwargs):
             h5, snew = utils.lnlstm(xs, ms, S, scope='lnlstm', nh=nlstm)
         else:
             h5, snew = utils.lstm(xs, ms, S, scope='lstm', nh=nlstm)
-
+            
         h = seq_to_batch(h5)
         initial_state = np.zeros(S.shape.as_list(), dtype=float)
 
@@ -161,26 +118,23 @@ def cnn_lstm(nlstm=128, layer_norm=False, **conv_kwargs):
 
     return network_fn
 
-
-@register("cnn_lnlstm")
 def cnn_lnlstm(nlstm=128, **conv_kwargs):
     return cnn_lstm(nlstm, layer_norm=True, **conv_kwargs)
 
 
-@register("conv_only")
 def conv_only(convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)], **conv_kwargs):
-    '''
+    ''' 
     convolutions-only net
 
     Parameters:
     ----------
 
-    conv:       list of triples (filter_number, filter_size, stride) specifying parameters for each layer.
+    conv:       list of triples (filter_number, filter_size, stride) specifying parameters for each layer. 
 
     Returns:
 
     function that takes tensorflow tensor as input and returns the output of the last convolutional layer
-
+    
     '''
 
     def network_fn(X):
@@ -194,31 +148,30 @@ def conv_only(convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)], **conv_kwargs):
                                            activation_fn=tf.nn.relu,
                                            **conv_kwargs)
 
-        return out
+        return out, None
     return network_fn
 
 def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
     rms = RunningMeanStd(shape=x.shape[1:])
     norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
     return norm_x, rms
-
+    
 
 def get_network_builder(name):
-    """
-    If you want to register your own network outside models.py, you just need:
-
-    Usage Example:
-    -------------
-    from baselines.common.models import register
-    @register("your_network_name")
-    def your_network_define(**net_kwargs):
-        ...
-        return network_fn
-
-    """
-    if callable(name):
-        return name
-    elif name in mapping:
-        return mapping[name]
+    # TODO: replace with reflection? 
+    if name == 'cnn':
+        return cnn
+    elif name == 'cnn_small':
+        return cnn_small
+    elif name == 'conv_only':
+        return conv_only
+    elif name == 'mlp':
+        return mlp
+    elif name == 'lstm':
+        return lstm
+    elif name == 'cnn_lstm':
+        return cnn_lstm
+    elif name == 'cnn_lnlstm':
+        return cnn_lnlstm
     else:
         raise ValueError('Unknown network type: {}'.format(name))

baselines/common/mpi_adam.py

@@ -76,4 +76,4 @@ def test_MpiAdam():
     for i in range(10):
         l,g = lossandgrad()
         adam.update(g, stepsize)
-        print(i,l)
+        print(i,l)

baselines/common/mpi_fork.py

@@ -4,7 +4,7 @@ def mpi_fork(n, bind_to_core=False):
     """Re-launches the current script with workers
     Returns "parent" for original parent, "child" for MPI children
     """
-    if n<=1:
+    if n<=1: 
         return "child"
     if os.getenv("IN_MPI") is None:
         env = os.environ.copy()

baselines/common/mpi_moments.py

@@ -33,8 +33,8 @@ def mpi_moments(x, axis=0, comm=None, keepdims=False):
 
 def test_runningmeanstd():
     import subprocess
-    subprocess.check_call(['mpirun', '-np', '3',
-        'python','-c',
+    subprocess.check_call(['mpirun', '-np', '3', 
+        'python','-c', 
         'from baselines.common.mpi_moments import _helper_runningmeanstd; _helper_runningmeanstd()'])
 
 def _helper_runningmeanstd():

baselines/common/policies.py

@@ -32,7 +32,7 @@ class PolicyWithValue(object):
         **tensors       tensorflow tensors for additional attributes such as state or mask
 
         """
-
+            
         self.X = observations
         self.state = tf.constant([])
         self.initial_state = None
@@ -43,17 +43,13 @@ class PolicyWithValue(object):
         vf_latent = tf.layers.flatten(vf_latent)
         latent = tf.layers.flatten(latent)
 
-        # Based on the action space, will select what probability distribution type
         self.pdtype = make_pdtype(env.action_space)
 
         self.pd, self.pi = self.pdtype.pdfromlatent(latent, init_scale=0.01)
 
-        # Take an action
         self.action = self.pd.sample()
-
-        # Calculate the neg log of our probability
         self.neglogp = self.pd.neglogp(self.action)
-        self.sess = sess or tf.get_default_session()
+        self.sess = sess
 
         if estimate_q:
             assert isinstance(env.action_space, gym.spaces.Discrete)
@@ -64,7 +60,7 @@ class PolicyWithValue(object):
             self.vf = self.vf[:,0]
 
     def _evaluate(self, variables, observation, **extra_feed):
-        sess = self.sess
+        sess = self.sess or tf.get_default_session()
         feed_dict = {self.X: adjust_shape(self.X, observation)}
         for inpt_name, data in extra_feed.items():
             if inpt_name in self.__dict__.keys():
@@ -76,7 +72,7 @@ class PolicyWithValue(object):
 
     def step(self, observation, **extra_feed):
         """
-        Compute next action(s) given the observation(s)
+        Compute next action(s) given the observaion(s)
 
         Parameters:
         ----------
@@ -89,7 +85,7 @@ class PolicyWithValue(object):
         -------
         (action, value estimate, next state, negative log likelihood of the action under current policy parameters) tuple
         """
-
+    
         a, v, state, neglogp = self._evaluate([self.action, self.vf, self.state, self.neglogp], observation, **extra_feed)
         if state.size == 0:
             state = None
@@ -97,7 +93,7 @@ class PolicyWithValue(object):
 
     def value(self, ob, *args, **kwargs):
         """
-        Compute value estimate(s) given the observation(s)
+        Compute value estimate(s) given the observaion(s)
 
         Parameters:
         ----------
@@ -110,14 +106,14 @@ class PolicyWithValue(object):
         -------
         value estimate
         """
-        return self._evaluate(self.vf, ob, *args, **kwargs)
+        return self._evaluate(self.vf, ob, *args, **kwargs)      
 
     def save(self, save_path):
         tf_util.save_state(save_path, sess=self.sess)
 
     def load(self, load_path):
         tf_util.load_state(load_path, sess=self.sess)
-
+  
 def build_policy(env, policy_network, value_network=None,  normalize_observations=False, estimate_q=False, **policy_kwargs):
     if isinstance(policy_network, str):
         network_type = policy_network
@@ -127,7 +123,7 @@ def build_policy(env, policy_network, value_network=None,  normalize_observation
         ob_space = env.observation_space
 
         X = observ_placeholder if observ_placeholder is not None else observation_placeholder(ob_space, batch_size=nbatch)
-
+        
         extra_tensors = {}
 
         if normalize_observations and X.dtype == tf.float32:
@@ -139,18 +135,16 @@ def build_policy(env, policy_network, value_network=None,  normalize_observation
         encoded_x = encode_observation(ob_space, encoded_x)
 
         with tf.variable_scope('pi', reuse=tf.AUTO_REUSE):
-            policy_latent = policy_network(encoded_x)
-            if isinstance(policy_latent, tuple):
-                policy_latent, recurrent_tensors = policy_latent
-
-                if recurrent_tensors is not None:
-                    # recurrent architecture, need a few more steps
-                    nenv = nbatch // nsteps
-                    assert nenv > 0, 'Bad input for recurrent policy: batch size {} smaller than nsteps {}'.format(nbatch, nsteps)
-                    policy_latent, recurrent_tensors = policy_network(encoded_x, nenv)
-                    extra_tensors.update(recurrent_tensors)
+            policy_latent, recurrent_tensors = policy_network(encoded_x)
 
+            if recurrent_tensors is not None:
+                # recurrent architecture, need a few more steps
+                nenv = nbatch // nsteps
+                assert nenv > 0, 'Bad input for recurrent policy: batch size {} smaller than nsteps {}'.format(nbatch, nsteps)
+                policy_latent, recurrent_tensors = policy_network(encoded_x, nenv)
+                extra_tensors.update(recurrent_tensors)
 
+            
         _v_net = value_network
 
         if _v_net is None or _v_net == 'shared':
@@ -160,11 +154,10 @@ def build_policy(env, policy_network, value_network=None,  normalize_observation
                 _v_net = policy_network
             else:
                 assert callable(_v_net)
-
+ 
             with tf.variable_scope('vf', reuse=tf.AUTO_REUSE):
-                # TODO recurrent architectures are not supported with value_network=copy yet
-                vf_latent = _v_net(encoded_x)
-
+                vf_latent, _ = _v_net(encoded_x)
+        
         policy = PolicyWithValue(
             env=env,
             observations=X,
@@ -183,4 +176,4 @@ def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
     rms = RunningMeanStd(shape=x.shape[1:])
     norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
     return norm_x, rms
-
+    

baselines/common/running_mean_std.py

@@ -23,15 +23,15 @@ def update_mean_var_count_from_moments(mean, var, count, batch_mean, batch_var,
     delta = batch_mean - mean
     tot_count = count + batch_count
 
-    new_mean = mean + delta * batch_count / tot_count
+    new_mean = mean + delta * batch_count / tot_count        
     m_a = var * count
     m_b = batch_var * batch_count
-    M2 = m_a + m_b + np.square(delta) * count * batch_count / tot_count
-    new_var = M2 / tot_count
-    new_count = tot_count
-
+    M2 = m_a + m_b + np.square(delta) * count * batch_count / (count + batch_count)
+    new_var = M2 / (count + batch_count)
+    new_count = batch_count + count
+    
     return new_mean, new_var, new_count
-
+    
 
 class TfRunningMeanStd(object):
     # https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Parallel_algorithm
@@ -46,10 +46,10 @@ class TfRunningMeanStd(object):
         self._new_var = tf.placeholder(shape=shape, dtype=tf.float64)
         self._new_count = tf.placeholder(shape=(), dtype=tf.float64)
 
-
+        
         with tf.variable_scope(scope, reuse=tf.AUTO_REUSE):
             self._mean  = tf.get_variable('mean',  initializer=np.zeros(shape, 'float64'),      dtype=tf.float64)
-            self._var   = tf.get_variable('std',   initializer=np.ones(shape, 'float64'),       dtype=tf.float64)
+            self._var   = tf.get_variable('std',   initializer=np.ones(shape, 'float64'),       dtype=tf.float64)    
             self._count = tf.get_variable('count', initializer=np.full((), epsilon, 'float64'), dtype=tf.float64)
 
         self.update_ops = tf.group([
@@ -61,10 +61,10 @@ class TfRunningMeanStd(object):
         sess.run(tf.variables_initializer([self._mean, self._var, self._count]))
         self.sess = sess
         self._set_mean_var_count()
-
+    
     def _set_mean_var_count(self):
-        self.mean, self.var, self.count = self.sess.run([self._mean, self._var, self._count])
-
+        self.mean, self.var, self.count = self.sess.run([self._mean, self._var, self._count])                    
+         
     def update(self, x):
         batch_mean = np.mean(x, axis=0)
         batch_var = np.var(x, axis=0)
@@ -74,13 +74,13 @@ class TfRunningMeanStd(object):
 
         self.sess.run(self.update_ops, feed_dict={
             self._new_mean: new_mean,
-            self._new_var: new_var,
+            self._new_var: new_var, 
             self._new_count: new_count
         })
 
         self._set_mean_var_count()
 
-
+        
 
 def test_runningmeanstd():
     for (x1, x2, x3) in [
@@ -145,7 +145,7 @@ def profile_tf_runningmeanstd():
 
     print('rms update time ({} trials): {} s'.format(n_trials, tic2 - tic1))
     print('tfrms update time ({} trials): {} s'.format(n_trials, tic3 - tic2))
-
+    
 
     tic1 = time.time()
     for _ in range(n_trials):
@@ -161,21 +161,21 @@ def profile_tf_runningmeanstd():
 
     print('rms get mean time ({} trials): {} s'.format(n_trials, tic2 - tic1))
     print('tfrms get mean time ({} trials): {} s'.format(n_trials, tic3 - tic2))
-
-
+         
+    
 
     '''
     options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE) #pylint: disable=E1101
     run_metadata = tf.RunMetadata()
     profile_opts = dict(options=options, run_metadata=run_metadata)
 
-
+    
 
     from tensorflow.python.client import timeline
     fetched_timeline = timeline.Timeline(run_metadata.step_stats) #pylint: disable=E1101
     chrome_trace = fetched_timeline.generate_chrome_trace_format()
     outfile = '/tmp/timeline.json'
-    with open(outfile, 'wt') as f:
+    with open(outfile, 'wt') as f: 
         f.write(chrome_trace)
     print(f'Successfully saved profile to {outfile}. Exiting.')
     exit(0)
@@ -184,4 +184,4 @@ def profile_tf_runningmeanstd():
 
 
 if __name__ == '__main__':
-   profile_tf_runningmeanstd()
+   profile_tf_runningmeanstd() 

baselines/common/running_stat.py

@@ -0,0 +1,46 @@
+import numpy as np
+
+# http://www.johndcook.com/blog/standard_deviation/
+class RunningStat(object):
+    def __init__(self, shape):
+        self._n = 0
+        self._M = np.zeros(shape)
+        self._S = np.zeros(shape)
+    def push(self, x):
+        x = np.asarray(x)
+        assert x.shape == self._M.shape
+        self._n += 1
+        if self._n == 1:
+            self._M[...] = x
+        else:
+            oldM = self._M.copy()
+            self._M[...] = oldM + (x - oldM)/self._n
+            self._S[...] = self._S + (x - oldM)*(x - self._M)
+    @property
+    def n(self):
+        return self._n
+    @property
+    def mean(self):
+        return self._M
+    @property
+    def var(self):
+        return self._S/(self._n - 1) if self._n > 1 else np.square(self._M)
+    @property
+    def std(self):
+        return np.sqrt(self.var)
+    @property
+    def shape(self):
+        return self._M.shape
+
+def test_running_stat():
+    for shp in ((), (3,), (3,4)):
+        li = []
+        rs = RunningStat(shp)
+        for _ in range(5):
+            val = np.random.randn(*shp)
+            rs.push(val)
+            li.append(val)
+            m = np.mean(li, axis=0)
+            assert np.allclose(rs.mean, m)
+            v = np.square(m) if (len(li) == 1) else np.var(li, ddof=1, axis=0)
+            assert np.allclose(rs.var, v)

baselines/common/tests/envs/fixed_sequence_env.py

@@ -40,5 +40,5 @@ class FixedSequenceEnv(Env):
 
     def _get_reward(self, actions):
         return 1 if actions == self.sequence[self.time] else 0
-
+        
 

baselines/common/tests/envs/mnist_env.py

@@ -1,6 +1,7 @@
 import os.path as osp
 import numpy as np
 import tempfile
+import filelock
 from gym import Env
 from gym.spaces import Discrete, Box
 
@@ -13,9 +14,8 @@ class MnistEnv(Env):
             episode_len=None,
             no_images=None
     ):
-        import filelock
         from tensorflow.examples.tutorials.mnist import input_data
-        # we could use temporary directory for this with a context manager and
+        # we could use temporary directory for this with a context manager and 
         # TemporaryDirecotry, but then each test that uses mnist would re-download the data
         # this way the data is not cleaned up, but we only download it once per machine
         mnist_path = osp.join(tempfile.gettempdir(), 'MNIST_data')
@@ -33,7 +33,7 @@ class MnistEnv(Env):
 
         self.train_mode()
         self.reset()
-
+        
     def reset(self):
         self._choose_next_state()
         self.time = 0

baselines/common/tests/test_cartpole.py

@@ -10,12 +10,11 @@ common_kwargs = dict(
     gamma=1.0,
     seed=0,
 )
-
+   
 learn_kwargs = {
     'a2c' : dict(nsteps=32, value_network='copy', lr=0.05),
-    'acer': dict(value_network='copy'),
-    'acktr': dict(nsteps=32, value_network='copy', is_async=False),
-    'deepq': dict(total_timesteps=20000),
+    'acktr': dict(nsteps=32, value_network='copy'),
+    'deepq': {},
     'ppo2': dict(value_network='copy'),
     'trpo_mpi': {}
 }
@@ -32,13 +31,10 @@ def test_cartpole(alg):
     kwargs.update(learn_kwargs[alg])
 
     learn_fn = lambda e: get_learn_function(alg)(env=e, **kwargs)
-    def env_fn():
-
+    def env_fn(): 
+        
         env = gym.make('CartPole-v0')
         env.seed(0)
         return env
 
     reward_per_episode_test(env_fn, learn_fn, 100)
-
-if __name__ == '__main__':
-    test_cartpole('acer')

baselines/common/tests/test_doc_examples.py

@@ -1,48 +0,0 @@
-import pytest
-try:
-    import mujoco_py
-    _mujoco_present = True
-except BaseException:
-    mujoco_py = None
-    _mujoco_present = False
-
-
-@pytest.mark.skipif(
-    not _mujoco_present,
-    reason='error loading mujoco - either mujoco / mujoco key not present, or LD_LIBRARY_PATH is not pointing to mujoco library'
-)
-def test_lstm_example():
-    import tensorflow as tf
-    from baselines.common import policies, models, cmd_util
-    from baselines.common.vec_env.dummy_vec_env import DummyVecEnv
-
-    # create vectorized environment
-    venv = DummyVecEnv([lambda: cmd_util.make_mujoco_env('Reacher-v2', seed=0)])
-
-    with tf.Session() as sess:
-        # build policy based on lstm network with 128 units
-        policy = policies.build_policy(venv, models.lstm(128))(nbatch=1, nsteps=1)
-
-        # initialize tensorflow variables
-        sess.run(tf.global_variables_initializer())
-
-        # prepare environment variables
-        ob = venv.reset()
-        state = policy.initial_state
-        done = [False]
-        step_counter = 0
-
-        # run a single episode until the end (i.e. until done)
-        while True:
-            action, _, state, _ = policy.step(ob, S=state, M=done)
-            ob, reward, done, _ = venv.step(action)
-            step_counter += 1
-            if done:
-                break
-
-
-        assert step_counter > 5
-
-
-
-

baselines/common/tests/test_env_after_learn.py

@@ -1,27 +0,0 @@
-import pytest
-import gym
-import tensorflow as tf
-
-from baselines.common.vec_env.subproc_vec_env import SubprocVecEnv
-from baselines.run import get_learn_function
-from baselines.common.tf_util import make_session
-
-algos = ['a2c', 'acer', 'acktr', 'deepq', 'ppo2', 'trpo_mpi']
-
-@pytest.mark.parametrize('algo', algos)
-def test_env_after_learn(algo):
-    def make_env():
-        # acktr requires too much RAM, fails on travis
-        env = gym.make('CartPole-v1' if algo == 'acktr' else 'PongNoFrameskip-v4')
-        return env
-
-    make_session(make_default=True, graph=tf.Graph())
-    env = SubprocVecEnv([make_env])
-
-    learn = get_learn_function(algo)
-
-    # Commenting out the following line resolves the issue, though crash happens at env.reset().
-    learn(network='mlp', env=env, total_timesteps=0, load_path=None, seed=None)
-
-    env.reset()
-    env.close()

baselines/common/tests/test_fixed_sequence.py

@@ -8,7 +8,7 @@ common_kwargs = dict(
     seed=0,
     total_timesteps=50000,
 )
-
+    
 learn_kwargs = {
     'a2c': {},
     'ppo2': dict(nsteps=10, ent_coef=0.0, nminibatches=1),
@@ -36,7 +36,7 @@ def test_fixed_sequence(alg, rnn):
     episode_len = 5
     env_fn = lambda: FixedSequenceEnv(10, episode_len=episode_len)
     learn = lambda e: get_learn_function(alg)(
-        env=e,
+        env=e, 
         network=rnn,
         **kwargs
     )
@@ -47,5 +47,5 @@ def test_fixed_sequence(alg, rnn):
 if __name__ == '__main__':
     test_fixed_sequence('ppo2', 'lstm')
 
-
+    
 

baselines/common/tests/test_identity.py

@@ -9,22 +9,18 @@ common_kwargs = dict(
     gamma=0.9,
     seed=0,
 )
-
+   
 learn_kwargs = {
     'a2c' : {},
     'acktr': {},
     'deepq': {},
-    'ddpg': dict(layer_norm=True),
     'ppo2': dict(lr=1e-3, nsteps=64, ent_coef=0.0),
     'trpo_mpi': dict(timesteps_per_batch=100, cg_iters=10, gamma=0.9, lam=1.0, max_kl=0.01)
 }
 
 
-algos_disc = ['a2c', 'acktr', 'deepq', 'ppo2', 'trpo_mpi']
-algos_cont = ['a2c', 'acktr', 'ddpg',  'ppo2', 'trpo_mpi']
-
 @pytest.mark.slow
-@pytest.mark.parametrize("alg", algos_disc)
+@pytest.mark.parametrize("alg", learn_kwargs.keys())
 def test_discrete_identity(alg):
     '''
     Test if the algorithm (with an mlp policy)
@@ -39,7 +35,7 @@ def test_discrete_identity(alg):
     simple_test(env_fn, learn_fn, 0.9)
 
 @pytest.mark.slow
-@pytest.mark.parametrize("alg", algos_cont)
+@pytest.mark.parametrize("alg", ['a2c', 'ppo2', 'trpo_mpi'])
 def test_continuous_identity(alg):
     '''
     Test if the algorithm (with an mlp policy)
@@ -55,5 +51,5 @@ def test_continuous_identity(alg):
     simple_test(env_fn, learn_fn, -0.1)
 
 if __name__ == '__main__':
-    test_continuous_identity('ddpg')
+    test_continuous_identity('a2c')    
 

baselines/common/tests/test_mnist.py

@@ -6,7 +6,7 @@ from baselines.common.tests.util import simple_test
 from baselines.run import get_learn_function
 
 
-# TODO investigate a2c and ppo2 failures - is it due to bad hyperparameters for this problem?
+# TODO investigate a2c and ppo2 failures - is it due to bad hyperparameters for this problem?  
 # GitHub issue https://github.com/openai/baselines/issues/189
 common_kwargs = {
     'seed': 0,
@@ -17,28 +17,29 @@ common_kwargs = {
 
 learn_args = {
     'a2c': dict(total_timesteps=50000),
-    'acer': dict(total_timesteps=20000),
+    # TODO need to resolve inference (step) API differences for acer; also slow
+    # 'acer': dict(seed=0, total_timesteps=1000),
     'deepq': dict(total_timesteps=5000),
     'acktr': dict(total_timesteps=30000),
     'ppo2': dict(total_timesteps=50000, lr=1e-3, nsteps=128, ent_coef=0.0),
     'trpo_mpi': dict(total_timesteps=80000, timesteps_per_batch=100, cg_iters=10, lam=1.0, max_kl=0.001)
 }
 
-
-#tests pass, but are too slow on travis. Same algorithms are covered
+ 
+#tests pass, but are too slow on travis. Same algorithms are covered 
 # by other tests with less compute-hungry nn's and by benchmarks
-@pytest.mark.skip
+@pytest.mark.skip 
 @pytest.mark.slow
 @pytest.mark.parametrize("alg", learn_args.keys())
 def test_mnist(alg):
     '''
-    Test if the algorithm can learn to classify MNIST digits.
-    Uses CNN policy.
+    Test if the algorithm can learn to classify MNIST digits. 
+    Uses CNN policy. 
     '''
-
+    
     learn_kwargs = learn_args[alg]
     learn_kwargs.update(common_kwargs)
-
+    
     learn = get_learn_function(alg)
     learn_fn = lambda e: learn(env=e, **learn_kwargs)
     env_fn = lambda: MnistEnv(seed=0, episode_len=100)
@@ -46,4 +47,4 @@ def test_mnist(alg):
     simple_test(env_fn, learn_fn, 0.6)
 
 if __name__ == '__main__':
-    test_mnist('acer')
+    test_mnist('deepq')

baselines/common/tests/test_serialization.py

@@ -1,5 +1,4 @@
 import os
-import gym
 import tempfile
 import pytest
 import tensorflow as tf
@@ -15,16 +14,15 @@ from functools import partial
 
 learn_kwargs = {
     'deepq': {},
-    'a2c': {},
+    'a2c': {}, 
     'acktr': {},
-    'acer': {},
     'ppo2': {'nminibatches': 1, 'nsteps': 10},
     'trpo_mpi': {},
 }
 
 network_kwargs = {
-    'mlp': {},
-    'cnn': {'pad': 'SAME'},
+    'mlp': {}, 
+    'cnn': {'pad': 'SAME'}, 
     'lstm': {},
     'cnn_lnlstm': {'pad': 'SAME'}
 }
@@ -34,15 +32,15 @@ network_kwargs = {
 @pytest.mark.parametrize("network_fn", network_kwargs.keys())
 def test_serialization(learn_fn, network_fn):
     '''
-    Test if the trained model can be serialized
+    Test if the trained model can be serialized 
     '''
 
-
-    if network_fn.endswith('lstm') and learn_fn in ['acer', 'acktr', 'trpo_mpi', 'deepq']:
+    
+    if network_fn.endswith('lstm') and learn_fn in ['acktr', 'trpo_mpi', 'deepq']:
             # TODO make acktr work with recurrent policies
             # and test
-            # github issue: https://github.com/openai/baselines/issues/660
-            return
+            # github issue: https://github.com/openai/baselines/issues/194
+            return 
 
     env = DummyVecEnv([lambda: MnistEnv(10, episode_len=100)])
     ob = env.reset().copy()
@@ -76,49 +74,14 @@ def test_serialization(learn_fn, network_fn):
         np.testing.assert_allclose(mean1, mean2, atol=0.5)
         np.testing.assert_allclose(std1, std2, atol=0.5)
 
-
-@pytest.mark.parametrize("learn_fn", learn_kwargs.keys())
-@pytest.mark.parametrize("network_fn", ['mlp'])
-def test_coexistence(learn_fn, network_fn):
-    '''
-    Test if more than one model can exist at a time
-    '''
-
-    if learn_fn == 'deepq':
-            # TODO enable multiple DQN models to be useable at the same time
-            # github issue https://github.com/openai/baselines/issues/656
-            return
-
-    if network_fn.endswith('lstm') and learn_fn in ['acktr', 'trpo_mpi', 'deepq']:
-            # TODO make acktr work with recurrent policies
-            # and test
-            # github issue: https://github.com/openai/baselines/issues/660
-            return
-
-    env = DummyVecEnv([lambda: gym.make('CartPole-v0')])
-    learn = get_learn_function(learn_fn)
-
-    kwargs = {}
-    kwargs.update(network_kwargs[network_fn])
-    kwargs.update(learn_kwargs[learn_fn])
-
-    learn =  partial(learn, env=env, network=network_fn, total_timesteps=0, **kwargs)
-    make_session(make_default=True, graph=tf.Graph());
-    model1 = learn(seed=1)
-    make_session(make_default=True, graph=tf.Graph());
-    model2 = learn(seed=2)
-
-    model1.step(env.observation_space.sample())
-    model2.step(env.observation_space.sample())
-
-
+ 
 
 def _serialize_variables():
     sess = get_session()
-    variables = tf.trainable_variables()
+    variables = tf.trainable_variables()    
     values = sess.run(variables)
     return {var.name: value for var, value in zip(variables, values)}
-
+    
 
 def _get_action_stats(model, ob):
     ntrials = 1000

baselines/common/tests/util.py

@@ -30,7 +30,7 @@ def simple_test(env_fn, learn_fn, min_reward_fraction, n_trials=N_TRIALS):
                 a, v, state, _ = model.step(obs, S=state, M=[False])
             else:
                 a, v, _, _ = model.step(obs)
-
+            
             obs, rew, done, _ = env.step(a)
             sum_rew += float(rew)
 
@@ -46,7 +46,7 @@ def reward_per_episode_test(env_fn, learn_fn, min_avg_reward, n_trials=N_EPISODE
     with tf.Graph().as_default(), tf.Session(config=tf.ConfigProto(allow_soft_placement=True)).as_default():
         model = learn_fn(env)
 
-        N_TRIALS = 100
+        N_TRIALS = 100    
 
         observations, actions, rewards = rollout(env, model, N_TRIALS)
         rewards = [sum(r) for r in rewards]

baselines/common/tf_util.py

@@ -62,7 +62,7 @@ def make_session(config=None, num_cpu=None, make_default=False, graph=None):
         num_cpu = int(os.getenv('RCALL_NUM_CPU', multiprocessing.cpu_count()))
     if config is None:
         config = tf.ConfigProto(
-            allow_soft_placement=True,
+            allow_soft_placement=True, 
             inter_op_parallelism_threads=num_cpu,
             intra_op_parallelism_threads=num_cpu)
         config.gpu_options.allow_growth = True
@@ -293,7 +293,7 @@ def display_var_info(vars):
         if "/Adam" in name or "beta1_power" in name or "beta2_power" in name: continue
         v_params = np.prod(v.shape.as_list())
         count_params += v_params
-        if "/b:" in name or "/bias" in name: continue    # Wx+b, bias is not interesting to look at => count params, but not print
+        if "/b:" in name or "/biases" in name: continue    # Wx+b, bias is not interesting to look at => count params, but not print
         logger.info("   %s%s %i params %s" % (name, " "*(55-len(name)), v_params, str(v.shape)))
 
     logger.info("Total model parameters: %0.2f million" % (count_params*1e-6))
@@ -312,19 +312,13 @@ def get_available_gpus():
 # ================================================================
 
 def load_state(fname, sess=None):
-    from baselines import logger
-    logger.warn('load_state method is deprecated, please use load_variables instead')
     sess = sess or get_session()
     saver = tf.train.Saver()
     saver.restore(tf.get_default_session(), fname)
 
 def save_state(fname, sess=None):
-    from baselines import logger
-    logger.warn('save_state method is deprecated, please use save_variables instead')
     sess = sess or get_session()
-    dirname = os.path.dirname(fname)
-    if any(dirname):
-        os.makedirs(dirname, exist_ok=True)
+    os.makedirs(os.path.dirname(fname), exist_ok=True)
     saver = tf.train.Saver()
     saver.save(tf.get_default_session(), fname)
 
@@ -334,12 +328,10 @@ def save_state(fname, sess=None):
 def save_variables(save_path, variables=None, sess=None):
     sess = sess or get_session()
     variables = variables or tf.trainable_variables()
-
+    
     ps = sess.run(variables)
     save_dict = {v.name: value for v, value in zip(variables, ps)}
-    dirname = os.path.dirname(save_path)
-    if any(dirname):
-        os.makedirs(dirname, exist_ok=True)
+    os.makedirs(os.path.dirname(save_path), exist_ok=True)
     joblib.dump(save_dict, save_path)
 
 def load_variables(load_path, variables=None, sess=None):
@@ -348,16 +340,11 @@ def load_variables(load_path, variables=None, sess=None):
 
     loaded_params = joblib.load(os.path.expanduser(load_path))
     restores = []
-    if isinstance(loaded_params, list):
-        assert len(loaded_params) == len(variables), 'number of variables loaded mismatches len(variables)'
-        for d, v in zip(loaded_params, variables):
-            restores.append(v.assign(d))
-    else:
-        for v in variables:
-            restores.append(v.assign(loaded_params[v.name]))
-
+    for v in variables:
+        restores.append(v.assign(loaded_params[v.name]))
     sess.run(restores)
 
+
 # ================================================================
 # Shape adjustment for feeding into tf placeholders
 # ================================================================
@@ -367,10 +354,10 @@ def adjust_shape(placeholder, data):
     If shape is incompatible, AssertionError is thrown
 
     Parameters:
-        placeholder     tensorflow input placeholder
-
+        placeholder     tensorflow input placeholder 
+        
         data            input data to be (potentially) reshaped to be fed into placeholder
-
+    
     Returns:
         reshaped data
     '''
@@ -379,14 +366,14 @@ def adjust_shape(placeholder, data):
         return data
     if isinstance(data, list):
         data = np.array(data)
-
+    
     placeholder_shape = [x or -1 for x in placeholder.shape.as_list()]
-
+    
     assert _check_shape(placeholder_shape, data.shape), \
         'Shape of data {} is not compatible with shape of the placeholder {}'.format(data.shape, placeholder_shape)
 
-    return np.reshape(data, placeholder_shape)
-
+    return np.reshape(data, placeholder_shape)  
+    
 
 def _check_shape(placeholder_shape, data_shape):
     ''' check if two shapes are compatible (i.e. differ only by dimensions of size 1, or by the batch dimension)'''
@@ -394,7 +381,7 @@ def _check_shape(placeholder_shape, data_shape):
     return True
     squeezed_placeholder_shape = _squeeze_shape(placeholder_shape)
     squeezed_data_shape = _squeeze_shape(data_shape)
-
+    
     for i, s_data in enumerate(squeezed_data_shape):
         s_placeholder = squeezed_placeholder_shape[i]
         if s_placeholder != -1 and s_data != s_placeholder:
@@ -405,26 +392,14 @@ def _check_shape(placeholder_shape, data_shape):
 
 def _squeeze_shape(shape):
     return [x for x in shape if x != 1]
-
-# ================================================================
+        
 # Tensorboard interfacing
 # ================================================================
 
 def launch_tensorboard_in_background(log_dir):
-    '''
-    To log the Tensorflow graph when using rl-algs
-    algorithms, you can run the following code
-    in your main script:
-        import threading, time
-        def start_tensorboard(session):
-            time.sleep(10) # Wait until graph is setup
-            tb_path = osp.join(logger.get_dir(), 'tb')
-            summary_writer = tf.summary.FileWriter(tb_path, graph=session.graph)
-            summary_op = tf.summary.merge_all()
-            launch_tensorboard_in_background(tb_path)
-        session = tf.get_default_session()
-        t = threading.Thread(target=start_tensorboard, args=([session]))
-        t.start()
-    '''
-    import subprocess
-    subprocess.Popen(['tensorboard', '--logdir', log_dir])
+    from tensorboard import main as tb
+    import threading
+    tf.flags.FLAGS.logdir = log_dir
+    t = threading.Thread(target=tb.main, args=([]))
+    t.start()
+

baselines/common/vec_env/__init__.py

@@ -1,5 +1,6 @@
 from abc import ABC, abstractmethod
-from baselines.common.tile_images import tile_images
+from baselines import logger
+
 
 class AlreadySteppingError(Exception):
     """
@@ -26,12 +27,8 @@ class NotSteppingError(Exception):
 class VecEnv(ABC):
     """
     An abstract asynchronous, vectorized environment.
-    Used to batch data from multiple copies of an environment, so that
-    each observation becomes an batch of observations, and expected action is a batch of actions to
-    be applied per-environment.
     """
-    closed = False
-    viewer = None
+
     def __init__(self, num_envs, observation_space, action_space):
         self.num_envs = num_envs
         self.observation_space = observation_space
@@ -75,21 +72,13 @@ class VecEnv(ABC):
         """
         pass
 
-    def close_extras(self):
+    @abstractmethod
+    def close(self):
         """
-        Clean up the  extra resources, beyond what's in this base class.
-        Only runs when not self.closed.
+        Clean up the environments' resources.
         """
         pass
 
-    def close(self):
-        if self.closed:
-            return
-        if self.viewer is not None:
-            self.viewer.close()
-        self.close_extras()
-        self.closed = True
-
     def step(self, actions):
         """
         Step the environments synchronously.
@@ -100,21 +89,7 @@ class VecEnv(ABC):
         return self.step_wait()
 
     def render(self, mode='human'):
-        imgs = self.get_images()
-        bigimg = tile_images(imgs)
-        if mode == 'human':
-            self.get_viewer().imshow(bigimg)
-            return self.get_viewer().isopen
-        elif mode == 'rgb_array':
-            return bigimg
-        else:
-            raise NotImplementedError
-
-    def get_images(self):
-        """
-        Return RGB images from each environment
-        """
-        raise NotImplementedError
+        logger.warn('Render not defined for %s' % self)
 
     @property
     def unwrapped(self):
@@ -123,12 +98,6 @@ class VecEnv(ABC):
         else:
             return self
 
-    def get_viewer(self):
-        if self.viewer is None:
-            from gym.envs.classic_control import rendering
-            self.viewer = rendering.SimpleImageViewer()
-        return self.viewer
-
 
 class VecEnvWrapper(VecEnv):
     """
@@ -157,11 +126,9 @@ class VecEnvWrapper(VecEnv):
     def close(self):
         return self.venv.close()
 
-    def render(self, mode='human'):
-        return self.venv.render(mode=mode)
+    def render(self):
+        self.venv.render()
 
-    def get_images(self):
-        return self.venv.get_images()
 
 class CloudpickleWrapper(object):
     """

baselines/common/vec_env/dummy_vec_env.py

@@ -1,54 +1,36 @@
 import numpy as np
-from gym import spaces
 from . import VecEnv
 from .util import copy_obs_dict, dict_to_obs, obs_space_info
 
+
 class DummyVecEnv(VecEnv):
     """
-    VecEnv that does runs multiple environments sequentially, that is,
-    the step and reset commands are send to one environment at a time.
-    Useful when debugging and when num_env == 1 (in the latter case,
-    avoids communication overhead)
-    """
-    def __init__(self, env_fns):
-        """
-        Arguments:
+    A VecEnv that wraps raw gym.Envs.
 
-        env_fns: iterable of callables      functions that build environments
-        """
+    This can be used when an algorithm requires a VecEnv
+    but you want to use a vanilla gym.Env instance.
+    It is also useful for avoiding IPC overhead when you
+    don't need to run environments in parallel.
+    """
+
+    def __init__(self, env_fns):
         self.envs = [fn() for fn in env_fns]
         env = self.envs[0]
         VecEnv.__init__(self, len(env_fns), env.observation_space, env.action_space)
         obs_space = env.observation_space
-
         self.keys, shapes, dtypes = obs_space_info(obs_space)
-        self.buf_obs = { k: np.zeros((self.num_envs,) + tuple(shapes[k]), dtype=dtypes[k]) for k in self.keys }
+        self.buf_obs = {k: np.zeros((self.num_envs,) + tuple(shapes[k]), dtype=dtypes[k]) for k in self.keys}
         self.buf_dones = np.zeros((self.num_envs,), dtype=np.bool)
-        self.buf_rews  = np.zeros((self.num_envs,), dtype=np.float32)
+        self.buf_rews = np.zeros((self.num_envs,), dtype=np.float32)
         self.buf_infos = [{} for _ in range(self.num_envs)]
         self.actions = None
 
     def step_async(self, actions):
-        listify = True
-        try:
-            if len(actions) == self.num_envs:
-                listify = False
-        except TypeError:
-            pass
-
-        if not listify:
-            self.actions = actions
-        else:
-            assert self.num_envs == 1, "actions {} is either not a list or has a wrong size - cannot match to {} environments".format(actions, self.num_envs)
-            self.actions = [actions]
+        self.actions = actions
 
     def step_wait(self):
         for e in range(self.num_envs):
-            action = self.actions[e]
-            if isinstance(self.envs[e].action_space, spaces.Discrete):
-                action = int(action)
-
-            obs, self.buf_rews[e], self.buf_dones[e], self.buf_infos[e] = self.envs[e].step(action)
+            obs, self.buf_rews[e], self.buf_dones[e], self.buf_infos[e] = self.envs[e].step(self.actions[e])
             if self.buf_dones[e]:
                 obs = self.envs[e].reset()
             self._save_obs(e, obs)
@@ -61,6 +43,13 @@ class DummyVecEnv(VecEnv):
             self._save_obs(e, obs)
         return self._obs_from_buf()
 
+    def close(self):
+        for e in self.envs:
+            e.close()
+
+    def render(self, mode='human'):
+        return [e.render(mode=mode) for e in self.envs]
+
     def _save_obs(self, e, obs):
         for k in self.keys:
             if k is None:
@@ -70,12 +59,3 @@ class DummyVecEnv(VecEnv):
 
     def _obs_from_buf(self):
         return dict_to_obs(copy_obs_dict(self.buf_obs))
-
-    def get_images(self):
-        return [env.render(mode='rgb_array') for env in self.envs]
-
-    def render(self, mode='human'):
-        if self.num_envs == 1:
-            self.envs[0].render(mode=mode)
-        else:
-            super().render(mode=mode)

baselines/common/vec_env/shmem_vec_env.py

@@ -7,6 +7,7 @@ import numpy as np
 from . import VecEnv, CloudpickleWrapper
 import ctypes
 from baselines import logger
+from baselines.common.tile_images import tile_images
 
 from .util import dict_to_obs, obs_space_info, obs_to_dict
 
@@ -19,7 +20,8 @@ _NP_TO_CT = {np.float32: ctypes.c_float,
 
 class ShmemVecEnv(VecEnv):
     """
-    Optimized version of SubprocVecEnv that uses shared variables to communicate observations.
+    An AsyncEnv that uses multiprocessing to run multiple
+    environments in parallel.
     """
 
     def __init__(self, env_fns, spaces=None):
@@ -54,7 +56,6 @@ class ShmemVecEnv(VecEnv):
             proc.start()
             child_pipe.close()
         self.waiting_step = False
-        self.viewer = None
 
     def reset(self):
         if self.waiting_step:
@@ -74,7 +75,7 @@ class ShmemVecEnv(VecEnv):
         obs, rews, dones, infos = zip(*outs)
         return self._decode_obses(obs), np.array(rews), np.array(dones), infos
 
-    def close_extras(self):
+    def close(self):
         if self.waiting_step:
             self.step_wait()
         for pipe in self.parent_pipes:
@@ -85,15 +86,23 @@ class ShmemVecEnv(VecEnv):
         for proc in self.procs:
             proc.join()
 
-    def get_images(self, mode='human'):
+    def render(self, mode='human'):
         for pipe in self.parent_pipes:
             pipe.send(('render', None))
-        return [pipe.recv() for pipe in self.parent_pipes]
+        imgs = [pipe.recv() for pipe in self.parent_pipes]
+        bigimg = tile_images(imgs)
+        if mode == 'human':
+            import cv2
+            cv2.imshow('vecenv', bigimg[:, :, ::-1])
+            cv2.waitKey(1)
+        elif mode == 'rgb_array':
+            return bigimg
+        else:
+            raise NotImplementedError
 
     def _decode_obses(self, obs):
         result = {}
         for k in self.obs_keys:
-
             bufs = [b[k] for b in self.obs_bufs]
             o = [np.frombuffer(b.get_obj(), dtype=self.obs_dtypes[k]).reshape(self.obs_shapes[k]) for b in bufs]
             result[k] = np.array(o)

baselines/common/vec_env/subproc_vec_env.py

@@ -1,6 +1,8 @@
 import numpy as np
 from multiprocessing import Process, Pipe
 from . import VecEnv, CloudpickleWrapper
+from baselines.common.tile_images import tile_images
+
 
 def worker(remote, parent_remote, env_fn_wrapper):
     parent_remote.close()
@@ -32,15 +34,9 @@ def worker(remote, parent_remote, env_fn_wrapper):
 
 
 class SubprocVecEnv(VecEnv):
-    """
-    VecEnv that runs multiple environments in parallel in subproceses and communicates with them via pipes.
-    Recommended to use when num_envs > 1 and step() can be a bottleneck.
-    """
     def __init__(self, env_fns, spaces=None):
         """
-        Arguments:
-
-        env_fns: iterable of callables -  functions that create environments to run in subprocesses. Need to be cloud-pickleable
+        envs: list of gym environments to run in subprocesses
         """
         self.waiting = False
         self.closed = False
@@ -56,30 +52,32 @@ class SubprocVecEnv(VecEnv):
 
         self.remotes[0].send(('get_spaces', None))
         observation_space, action_space = self.remotes[0].recv()
-        self.viewer = None
         VecEnv.__init__(self, len(env_fns), observation_space, action_space)
 
     def step_async(self, actions):
-        self._assert_not_closed()
         for remote, action in zip(self.remotes, actions):
             remote.send(('step', action))
         self.waiting = True
 
     def step_wait(self):
-        self._assert_not_closed()
         results = [remote.recv() for remote in self.remotes]
         self.waiting = False
         obs, rews, dones, infos = zip(*results)
         return np.stack(obs), np.stack(rews), np.stack(dones), infos
 
     def reset(self):
-        self._assert_not_closed()
         for remote in self.remotes:
             remote.send(('reset', None))
         return np.stack([remote.recv() for remote in self.remotes])
 
-    def close_extras(self):
-        self.closed = True
+    def reset_task(self):
+        for remote in self.remotes:
+            remote.send(('reset_task', None))
+        return np.stack([remote.recv() for remote in self.remotes])
+
+    def close(self):
+        if self.closed:
+            return
         if self.waiting:
             for remote in self.remotes:
                 remote.recv()
@@ -87,13 +85,18 @@ class SubprocVecEnv(VecEnv):
             remote.send(('close', None))
         for p in self.ps:
             p.join()
+        self.closed = True
 
-    def get_images(self):
-        self._assert_not_closed()
+    def render(self, mode='human'):
         for pipe in self.remotes:
             pipe.send(('render', None))
         imgs = [pipe.recv() for pipe in self.remotes]
-        return imgs
-
-    def _assert_not_closed(self):
-        assert not self.closed, "Trying to operate on a SubprocVecEnv after calling close()"
+        bigimg = tile_images(imgs)
+        if mode == 'human':
+            import cv2
+            cv2.imshow('vecenv', bigimg[:, :, ::-1])
+            cv2.waitKey(1)
+        elif mode == 'rgb_array':
+            return bigimg
+        else:
+            raise NotImplementedError

baselines/common/vec_env/test_vec_env.py

@@ -10,39 +10,6 @@ from .shmem_vec_env import ShmemVecEnv
 from .subproc_vec_env import SubprocVecEnv
 
 
-def assert_envs_equal(env1, env2, num_steps):
-    """
-    Compare two environments over num_steps steps and make sure
-    that the observations produced by each are the same when given
-    the same actions.
-    """
-    assert env1.num_envs == env2.num_envs
-    assert env1.action_space.shape == env2.action_space.shape
-    assert env1.action_space.dtype == env2.action_space.dtype
-    joint_shape = (env1.num_envs,) + env1.action_space.shape
-
-    try:
-        obs1, obs2 = env1.reset(), env2.reset()
-        assert np.array(obs1).shape == np.array(obs2).shape
-        assert np.array(obs1).shape == joint_shape
-        assert np.allclose(obs1, obs2)
-        np.random.seed(1337)
-        for _ in range(num_steps):
-            actions = np.array(np.random.randint(0, 0x100, size=joint_shape),
-                               dtype=env1.action_space.dtype)
-            for env in [env1, env2]:
-                env.step_async(actions)
-            outs1 = env1.step_wait()
-            outs2 = env2.step_wait()
-            for out1, out2 in zip(outs1[:3], outs2[:3]):
-                assert np.array(out1).shape == np.array(out2).shape
-                assert np.allclose(out1, out2)
-            assert list(outs1[3]) == list(outs2[3])
-    finally:
-        env1.close()
-        env2.close()
-
-
 @pytest.mark.parametrize('klass', (ShmemVecEnv, SubprocVecEnv))
 @pytest.mark.parametrize('dtype', ('uint8', 'float32'))
 def test_vec_env(klass, dtype):  # pylint: disable=R0914
@@ -59,14 +26,33 @@ def test_vec_env(klass, dtype):  # pylint: disable=R0914
         """
         Get an environment constructor with a seed.
         """
-        return lambda: SimpleEnv(seed, shape, dtype)
+        return lambda: _SimpleEnv(seed, shape, dtype)
     fns = [make_fn(i) for i in range(num_envs)]
     env1 = DummyVecEnv(fns)
     env2 = klass(fns)
-    assert_envs_equal(env1, env2, num_steps=num_steps)
+    try:
+        obs1, obs2 = env1.reset(), env2.reset()
+        assert np.array(obs1).shape == np.array(obs2).shape
+        assert np.allclose(obs1, obs2)
+        np.random.seed(1337)
+        for _ in range(num_steps):
+            joint_shape = (len(fns),) + shape
+            actions = np.array(np.random.randint(0, 0x100, size=joint_shape),
+                               dtype=dtype)
+            for env in [env1, env2]:
+                env.step_async(actions)
+            outs1 = env1.step_wait()
+            outs2 = env2.step_wait()
+            for out1, out2 in zip(outs1[:3], outs2[:3]):
+                assert np.array(out1).shape == np.array(out2).shape
+                assert np.allclose(out1, out2)
+            assert list(outs1[3]) == list(outs2[3])
+    finally:
+        env1.close()
+        env2.close()
 
 
-class SimpleEnv(gym.Env):
+class _SimpleEnv(gym.Env):
     """
     An environment with a pre-determined observation space
     and RNG seed.
@@ -80,9 +66,7 @@ class SimpleEnv(gym.Env):
         self._max_steps = seed + 1
         self._cur_obs = None
         self._cur_step = 0
-        # this is 0xFF instead of 0x100 because the Box space includes
-        # the high end, while randint does not
-        self.action_space = gym.spaces.Box(low=0, high=0xFF, shape=shape, dtype=dtype)
+        self.action_space = gym.spaces.Box(low=0, high=100, shape=shape, dtype=dtype)
         self.observation_space = self.action_space
 
     def step(self, action):

baselines/common/vec_env/vec_frame_stack.py

@@ -28,3 +28,6 @@ class VecFrameStack(VecEnvWrapper):
         self.stackedobs[...] = 0
         self.stackedobs[..., -obs.shape[-1]:] = obs
         return self.stackedobs
+
+    def close(self):
+        self.venv.close()

baselines/common/vec_env/vec_monitor.py

@@ -1,16 +1,12 @@
 from . import VecEnvWrapper
-from baselines.bench.monitor import ResultsWriter
 import numpy as np
-import time
 
 
 class VecMonitor(VecEnvWrapper):
-    def __init__(self, venv, filename=None):
+    def __init__(self, venv):
         VecEnvWrapper.__init__(self, venv)
         self.eprets = None
         self.eplens = None
-        self.tstart = time.time()
-        self.results_writer = ResultsWriter(filename, header={'t_start': self.tstart})
 
     def reset(self):
         obs = self.venv.reset()
@@ -26,12 +22,8 @@ class VecMonitor(VecEnvWrapper):
         for (i, (done, ret, eplen, info)) in enumerate(zip(dones, self.eprets, self.eplens, infos)):
             info = info.copy()
             if done:
-                epinfo = {'r': ret, 'l': eplen, 't': round(time.time() - self.tstart, 6)}
-                info['episode'] = epinfo
+                info['episode'] = {'r': ret, 'l': eplen}
                 self.eprets[i] = 0
                 self.eplens[i] = 0
-                self.results_writer.write_row(epinfo)
-
             newinfos.append(info)
-
         return obs, rews, dones, newinfos

baselines/common/vec_env/vec_normalize.py

@@ -26,7 +26,6 @@ class VecNormalize(VecEnvWrapper):
         if self.ret_rms:
             self.ret_rms.update(self.ret)
             rews = np.clip(rews / np.sqrt(self.ret_rms.var + self.epsilon), -self.cliprew, self.cliprew)
-        self.ret[news] = 0.
         return obs, rews, news, infos
 
     def _obfilt(self, obs):
@@ -38,6 +37,5 @@ class VecNormalize(VecEnvWrapper):
             return obs
 
     def reset(self):
-        self.ret = np.zeros(self.num_envs)
         obs = self.venv.reset()
         return self._obfilt(obs)

baselines/ddpg/README.md

@@ -2,4 +2,4 @@
 
 - Original paper: https://arxiv.org/abs/1509.02971
 - Baselines post: https://blog.openai.com/better-exploration-with-parameter-noise/
-- `python -m baselines.run --alg=ddpg --env=HalfCheetah-v2 --num_timesteps=1e6` runs the algorithm for 1M frames = 10M timesteps on a Mujoco environment. See help (`-h`) for more options.
+- `python -m baselines.ddpg.main` runs the algorithm for 1M frames = 10M timesteps on a Mujoco environment. See help (`-h`) for more options.

baselines/ddpg/ddpg.py

@@ -1,259 +1,378 @@
-import os
-import time
-from collections import deque
-import pickle
+from copy import copy
+from functools import reduce
 
-from baselines.ddpg.ddpg_learner import DDPG
-from baselines.ddpg.models import Actor, Critic
-from baselines.ddpg.memory import Memory
-from baselines.ddpg.noise import AdaptiveParamNoiseSpec, NormalActionNoise, OrnsteinUhlenbeckActionNoise
-
-import baselines.common.tf_util as U
-
-from baselines import logger, registry
 import numpy as np
+import tensorflow as tf
+import tensorflow.contrib as tc
+
+from baselines import logger
+from baselines.common.mpi_adam import MpiAdam
+import baselines.common.tf_util as U
+from baselines.common.mpi_running_mean_std import RunningMeanStd
 from mpi4py import MPI
 
-@registry.register('ddpg')
-def learn(network, env,
-          seed=None,
-          total_timesteps=None,
-          nb_epochs=None, # with default settings, perform 1M steps total
-          nb_epoch_cycles=20,
-          nb_rollout_steps=100,
-          reward_scale=1.0,
-          render=False,
-          render_eval=False,
-          noise_type='adaptive-param_0.2',
-          normalize_returns=False,
-          normalize_observations=True,
-          critic_l2_reg=1e-2,
-          actor_lr=1e-4,
-          critic_lr=1e-3,
-          popart=False,
-          gamma=0.99,
-          clip_norm=None,
-          nb_train_steps=50, # per epoch cycle and MPI worker,
-          nb_eval_steps=100,
-          batch_size=64, # per MPI worker
-          tau=0.01,
-          eval_env=None,
-          param_noise_adaption_interval=50,
-          **network_kwargs):
+def normalize(x, stats):
+    if stats is None:
+        return x
+    return (x - stats.mean) / stats.std
 
 
-    if total_timesteps is not None:
-        assert nb_epochs is None
-        nb_epochs = int(total_timesteps) // (nb_epoch_cycles * nb_rollout_steps)
-    else:
-        nb_epochs = 500
+def denormalize(x, stats):
+    if stats is None:
+        return x
+    return x * stats.std + stats.mean
 
-    rank = MPI.COMM_WORLD.Get_rank()
-    nb_actions = env.action_space.shape[-1]
-    assert (np.abs(env.action_space.low) == env.action_space.high).all()  # we assume symmetric actions.
+def reduce_std(x, axis=None, keepdims=False):
+    return tf.sqrt(reduce_var(x, axis=axis, keepdims=keepdims))
 
-    memory = Memory(limit=int(1e6), action_shape=env.action_space.shape, observation_shape=env.observation_space.shape)
-    critic = Critic(network=network, **network_kwargs)
-    actor = Actor(nb_actions, network=network, **network_kwargs)
+def reduce_var(x, axis=None, keepdims=False):
+    m = tf.reduce_mean(x, axis=axis, keepdims=True)
+    devs_squared = tf.square(x - m)
+    return tf.reduce_mean(devs_squared, axis=axis, keepdims=keepdims)
 
-    action_noise = None
-    param_noise = None
-    nb_actions = env.action_space.shape[-1]
-    if noise_type is not None:
-        for current_noise_type in noise_type.split(','):
-            current_noise_type = current_noise_type.strip()
-            if current_noise_type == 'none':
-                pass
-            elif 'adaptive-param' in current_noise_type:
-                _, stddev = current_noise_type.split('_')
-                param_noise = AdaptiveParamNoiseSpec(initial_stddev=float(stddev), desired_action_stddev=float(stddev))
-            elif 'normal' in current_noise_type:
-                _, stddev = current_noise_type.split('_')
-                action_noise = NormalActionNoise(mu=np.zeros(nb_actions), sigma=float(stddev) * np.ones(nb_actions))
-            elif 'ou' in current_noise_type:
-                _, stddev = current_noise_type.split('_')
-                action_noise = OrnsteinUhlenbeckActionNoise(mu=np.zeros(nb_actions), sigma=float(stddev) * np.ones(nb_actions))
-            else:
-                raise RuntimeError('unknown noise type "{}"'.format(current_noise_type))
-
-    max_action = env.action_space.high
-    logger.info('scaling actions by {} before executing in env'.format(max_action))
-
-    agent = DDPG(actor, critic, memory, env.observation_space.shape, env.action_space.shape,
-        gamma=gamma, tau=tau, normalize_returns=normalize_returns, normalize_observations=normalize_observations,
-        batch_size=batch_size, action_noise=action_noise, param_noise=param_noise, critic_l2_reg=critic_l2_reg,
-        actor_lr=actor_lr, critic_lr=critic_lr, enable_popart=popart, clip_norm=clip_norm,
-        reward_scale=reward_scale)
-    logger.info('Using agent with the following configuration:')
-    logger.info(str(agent.__dict__.items()))
-
-    eval_episode_rewards_history = deque(maxlen=100)
-    episode_rewards_history = deque(maxlen=100)
-    sess = U.get_session()
-    # Prepare everything.
-    agent.initialize(sess)
-    sess.graph.finalize()
-
-    agent.reset()
-
-    obs = env.reset()
-    if eval_env is not None:
-        eval_obs = eval_env.reset()
-    nenvs = obs.shape[0]
-
-    episode_reward = np.zeros(nenvs, dtype = np.float32) #vector
-    episode_step = np.zeros(nenvs, dtype = int) # vector
-    episodes = 0 #scalar
-    t = 0 # scalar
-
-    epoch = 0
+def get_target_updates(vars, target_vars, tau):
+    logger.info('setting up target updates ...')
+    soft_updates = []
+    init_updates = []
+    assert len(vars) == len(target_vars)
+    for var, target_var in zip(vars, target_vars):
+        logger.info('  {} <- {}'.format(target_var.name, var.name))
+        init_updates.append(tf.assign(target_var, var))
+        soft_updates.append(tf.assign(target_var, (1. - tau) * target_var + tau * var))
+    assert len(init_updates) == len(vars)
+    assert len(soft_updates) == len(vars)
+    return tf.group(*init_updates), tf.group(*soft_updates)
 
 
+def get_perturbed_actor_updates(actor, perturbed_actor, param_noise_stddev):
+    assert len(actor.vars) == len(perturbed_actor.vars)
+    assert len(actor.perturbable_vars) == len(perturbed_actor.perturbable_vars)
 
-    start_time = time.time()
-
-    epoch_episode_rewards = []
-    epoch_episode_steps = []
-    epoch_actions = []
-    epoch_qs = []
-    epoch_episodes = 0
-    for epoch in range(nb_epochs):
-        for cycle in range(nb_epoch_cycles):
-            # Perform rollouts.
-            if nenvs > 1:
-                # if simulating multiple envs in parallel, impossible to reset agent at the end of the episode in each
-                # of the environments, so resetting here instead
-                agent.reset()
-            for t_rollout in range(nb_rollout_steps):
-                # Predict next action.
-                action, q, _, _ = agent.step(obs, apply_noise=True, compute_Q=True)
-
-                # Execute next action.
-                if rank == 0 and render:
-                    env.render()
-
-                # max_action is of dimension A, whereas action is dimension (nenvs, A) - the multiplication gets broadcasted to the batch
-                new_obs, r, done, info = env.step(max_action * action)  # scale for execution in env (as far as DDPG is concerned, every action is in [-1, 1])
-                # note these outputs are batched from vecenv
-
-                t += 1
-                if rank == 0 and render:
-                    env.render()
-                episode_reward += r
-                episode_step += 1
-
-                # Book-keeping.
-                epoch_actions.append(action)
-                epoch_qs.append(q)
-                agent.store_transition(obs, action, r, new_obs, done) #the batched data will be unrolled in memory.py's append.
-
-                obs = new_obs
-
-                for d in range(len(done)):
-                    if done[d]:
-                        # Episode done.
-                        epoch_episode_rewards.append(episode_reward[d])
-                        episode_rewards_history.append(episode_reward[d])
-                        epoch_episode_steps.append(episode_step[d])
-                        episode_reward[d] = 0.
-                        episode_step[d] = 0
-                        epoch_episodes += 1
-                        episodes += 1
-                        if nenvs == 1:
-                            agent.reset()
+    updates = []
+    for var, perturbed_var in zip(actor.vars, perturbed_actor.vars):
+        if var in actor.perturbable_vars:
+            logger.info('  {} <- {} + noise'.format(perturbed_var.name, var.name))
+            updates.append(tf.assign(perturbed_var, var + tf.random_normal(tf.shape(var), mean=0., stddev=param_noise_stddev)))
+        else:
+            logger.info('  {} <- {}'.format(perturbed_var.name, var.name))
+            updates.append(tf.assign(perturbed_var, var))
+    assert len(updates) == len(actor.vars)
+    return tf.group(*updates)
 
 
+class DDPG(object):
+    def __init__(self, actor, critic, memory, observation_shape, action_shape, param_noise=None, action_noise=None,
+        gamma=0.99, tau=0.001, normalize_returns=False, enable_popart=False, normalize_observations=True,
+        batch_size=128, observation_range=(-5., 5.), action_range=(-1., 1.), return_range=(-np.inf, np.inf),
+        adaptive_param_noise=True, adaptive_param_noise_policy_threshold=.1,
+        critic_l2_reg=0., actor_lr=1e-4, critic_lr=1e-3, clip_norm=None, reward_scale=1.):
+        # Inputs.
+        self.obs0 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs0')
+        self.obs1 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs1')
+        self.terminals1 = tf.placeholder(tf.float32, shape=(None, 1), name='terminals1')
+        self.rewards = tf.placeholder(tf.float32, shape=(None, 1), name='rewards')
+        self.actions = tf.placeholder(tf.float32, shape=(None,) + action_shape, name='actions')
+        self.critic_target = tf.placeholder(tf.float32, shape=(None, 1), name='critic_target')
+        self.param_noise_stddev = tf.placeholder(tf.float32, shape=(), name='param_noise_stddev')
 
-            # Train.
-            epoch_actor_losses = []
-            epoch_critic_losses = []
-            epoch_adaptive_distances = []
-            for t_train in range(nb_train_steps):
-                # Adapt param noise, if necessary.
-                if memory.nb_entries >= batch_size and t_train % param_noise_adaption_interval == 0:
-                    distance = agent.adapt_param_noise()
-                    epoch_adaptive_distances.append(distance)
+        # Parameters.
+        self.gamma = gamma
+        self.tau = tau
+        self.memory = memory
+        self.normalize_observations = normalize_observations
+        self.normalize_returns = normalize_returns
+        self.action_noise = action_noise
+        self.param_noise = param_noise
+        self.action_range = action_range
+        self.return_range = return_range
+        self.observation_range = observation_range
+        self.critic = critic
+        self.actor = actor
+        self.actor_lr = actor_lr
+        self.critic_lr = critic_lr
+        self.clip_norm = clip_norm
+        self.enable_popart = enable_popart
+        self.reward_scale = reward_scale
+        self.batch_size = batch_size
+        self.stats_sample = None
+        self.critic_l2_reg = critic_l2_reg
 
-                cl, al = agent.train()
-                epoch_critic_losses.append(cl)
-                epoch_actor_losses.append(al)
-                agent.update_target_net()
+        # Observation normalization.
+        if self.normalize_observations:
+            with tf.variable_scope('obs_rms'):
+                self.obs_rms = RunningMeanStd(shape=observation_shape)
+        else:
+            self.obs_rms = None
+        normalized_obs0 = tf.clip_by_value(normalize(self.obs0, self.obs_rms),
+            self.observation_range[0], self.observation_range[1])
+        normalized_obs1 = tf.clip_by_value(normalize(self.obs1, self.obs_rms),
+            self.observation_range[0], self.observation_range[1])
 
-            # Evaluate.
-            eval_episode_rewards = []
-            eval_qs = []
-            if eval_env is not None:
-                nenvs_eval = eval_obs.shape[0]
-                eval_episode_reward = np.zeros(nenvs_eval, dtype = np.float32)
-                for t_rollout in range(nb_eval_steps):
-                    eval_action, eval_q, _, _ = agent.step(eval_obs, apply_noise=False, compute_Q=True)
-                    eval_obs, eval_r, eval_done, eval_info = eval_env.step(max_action * eval_action)  # scale for execution in env (as far as DDPG is concerned, every action is in [-1, 1])
-                    if render_eval:
-                        eval_env.render()
-                    eval_episode_reward += eval_r
+        # Return normalization.
+        if self.normalize_returns:
+            with tf.variable_scope('ret_rms'):
+                self.ret_rms = RunningMeanStd()
+        else:
+            self.ret_rms = None
 
-                    eval_qs.append(eval_q)
-                    for d in range(len(eval_done)):
-                        if eval_done[d]:
-                            eval_episode_rewards.append(eval_episode_reward[d])
-                            eval_episode_rewards_history.append(eval_episode_reward[d])
-                            eval_episode_reward[d] = 0.0
+        # Create target networks.
+        target_actor = copy(actor)
+        target_actor.name = 'target_actor'
+        self.target_actor = target_actor
+        target_critic = copy(critic)
+        target_critic.name = 'target_critic'
+        self.target_critic = target_critic
 
-        mpi_size = MPI.COMM_WORLD.Get_size()
-        # Log stats.
-        # XXX shouldn't call np.mean on variable length lists
-        duration = time.time() - start_time
-        stats = agent.get_stats()
-        combined_stats = stats.copy()
-        combined_stats['rollout/return'] = np.mean(epoch_episode_rewards)
-        combined_stats['rollout/return_history'] = np.mean(episode_rewards_history)
-        combined_stats['rollout/episode_steps'] = np.mean(epoch_episode_steps)
-        combined_stats['rollout/actions_mean'] = np.mean(epoch_actions)
-        combined_stats['rollout/Q_mean'] = np.mean(epoch_qs)
-        combined_stats['train/loss_actor'] = np.mean(epoch_actor_losses)
-        combined_stats['train/loss_critic'] = np.mean(epoch_critic_losses)
-        combined_stats['train/param_noise_distance'] = np.mean(epoch_adaptive_distances)
-        combined_stats['total/duration'] = duration
-        combined_stats['total/steps_per_second'] = float(t) / float(duration)
-        combined_stats['total/episodes'] = episodes
-        combined_stats['rollout/episodes'] = epoch_episodes
-        combined_stats['rollout/actions_std'] = np.std(epoch_actions)
-        # Evaluation statistics.
-        if eval_env is not None:
-            combined_stats['eval/return'] = eval_episode_rewards
-            combined_stats['eval/return_history'] = np.mean(eval_episode_rewards_history)
-            combined_stats['eval/Q'] = eval_qs
-            combined_stats['eval/episodes'] = len(eval_episode_rewards)
-        def as_scalar(x):
-            if isinstance(x, np.ndarray):
-                assert x.size == 1
-                return x[0]
-            elif np.isscalar(x):
-                return x
-            else:
-                raise ValueError('expected scalar, got %s'%x)
+        # Create networks and core TF parts that are shared across setup parts.
+        self.actor_tf = actor(normalized_obs0)
+        self.normalized_critic_tf = critic(normalized_obs0, self.actions)
+        self.critic_tf = denormalize(tf.clip_by_value(self.normalized_critic_tf, self.return_range[0], self.return_range[1]), self.ret_rms)
+        self.normalized_critic_with_actor_tf = critic(normalized_obs0, self.actor_tf, reuse=True)
+        self.critic_with_actor_tf = denormalize(tf.clip_by_value(self.normalized_critic_with_actor_tf, self.return_range[0], self.return_range[1]), self.ret_rms)
+        Q_obs1 = denormalize(target_critic(normalized_obs1, target_actor(normalized_obs1)), self.ret_rms)
+        self.target_Q = self.rewards + (1. - self.terminals1) * gamma * Q_obs1
 
-        combined_stats_sums = MPI.COMM_WORLD.allreduce(np.array([ np.array(x).flatten()[0] for x in combined_stats.values()]))
-        combined_stats = {k : v / mpi_size for (k,v) in zip(combined_stats.keys(), combined_stats_sums)}
+        # Set up parts.
+        if self.param_noise is not None:
+            self.setup_param_noise(normalized_obs0)
+        self.setup_actor_optimizer()
+        self.setup_critic_optimizer()
+        if self.normalize_returns and self.enable_popart:
+            self.setup_popart()
+        self.setup_stats()
+        self.setup_target_network_updates()
 
-        # Total statistics.
-        combined_stats['total/epochs'] = epoch + 1
-        combined_stats['total/steps'] = t
+    def setup_target_network_updates(self):
+        actor_init_updates, actor_soft_updates = get_target_updates(self.actor.vars, self.target_actor.vars, self.tau)
+        critic_init_updates, critic_soft_updates = get_target_updates(self.critic.vars, self.target_critic.vars, self.tau)
+        self.target_init_updates = [actor_init_updates, critic_init_updates]
+        self.target_soft_updates = [actor_soft_updates, critic_soft_updates]
 
-        for key in sorted(combined_stats.keys()):
-            logger.record_tabular(key, combined_stats[key])
+    def setup_param_noise(self, normalized_obs0):
+        assert self.param_noise is not None
 
-        if rank == 0:
-            logger.dump_tabular()
-        logger.info('')
-        logdir = logger.get_dir()
-        if rank == 0 and logdir:
-            if hasattr(env, 'get_state'):
-                with open(os.path.join(logdir, 'env_state.pkl'), 'wb') as f:
-                    pickle.dump(env.get_state(), f)
-            if eval_env and hasattr(eval_env, 'get_state'):
-                with open(os.path.join(logdir, 'eval_env_state.pkl'), 'wb') as f:
-                    pickle.dump(eval_env.get_state(), f)
+        # Configure perturbed actor.
+        param_noise_actor = copy(self.actor)
+        param_noise_actor.name = 'param_noise_actor'
+        self.perturbed_actor_tf = param_noise_actor(normalized_obs0)
+        logger.info('setting up param noise')
+        self.perturb_policy_ops = get_perturbed_actor_updates(self.actor, param_noise_actor, self.param_noise_stddev)
 
+        # Configure separate copy for stddev adoption.
+        adaptive_param_noise_actor = copy(self.actor)
+        adaptive_param_noise_actor.name = 'adaptive_param_noise_actor'
+        adaptive_actor_tf = adaptive_param_noise_actor(normalized_obs0)
+        self.perturb_adaptive_policy_ops = get_perturbed_actor_updates(self.actor, adaptive_param_noise_actor, self.param_noise_stddev)
+        self.adaptive_policy_distance = tf.sqrt(tf.reduce_mean(tf.square(self.actor_tf - adaptive_actor_tf)))
 
-    return agent
+    def setup_actor_optimizer(self):
+        logger.info('setting up actor optimizer')
+        self.actor_loss = -tf.reduce_mean(self.critic_with_actor_tf)
+        actor_shapes = [var.get_shape().as_list() for var in self.actor.trainable_vars]
+        actor_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in actor_shapes])
+        logger.info('  actor shapes: {}'.format(actor_shapes))
+        logger.info('  actor params: {}'.format(actor_nb_params))
+        self.actor_grads = U.flatgrad(self.actor_loss, self.actor.trainable_vars, clip_norm=self.clip_norm)
+        self.actor_optimizer = MpiAdam(var_list=self.actor.trainable_vars,
+            beta1=0.9, beta2=0.999, epsilon=1e-08)
+
+    def setup_critic_optimizer(self):
+        logger.info('setting up critic optimizer')
+        normalized_critic_target_tf = tf.clip_by_value(normalize(self.critic_target, self.ret_rms), self.return_range[0], self.return_range[1])
+        self.critic_loss = tf.reduce_mean(tf.square(self.normalized_critic_tf - normalized_critic_target_tf))
+        if self.critic_l2_reg > 0.:
+            critic_reg_vars = [var for var in self.critic.trainable_vars if 'kernel' in var.name and 'output' not in var.name]
+            for var in critic_reg_vars:
+                logger.info('  regularizing: {}'.format(var.name))
+            logger.info('  applying l2 regularization with {}'.format(self.critic_l2_reg))
+            critic_reg = tc.layers.apply_regularization(
+                tc.layers.l2_regularizer(self.critic_l2_reg),
+                weights_list=critic_reg_vars
+            )
+            self.critic_loss += critic_reg
+        critic_shapes = [var.get_shape().as_list() for var in self.critic.trainable_vars]
+        critic_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in critic_shapes])
+        logger.info('  critic shapes: {}'.format(critic_shapes))
+        logger.info('  critic params: {}'.format(critic_nb_params))
+        self.critic_grads = U.flatgrad(self.critic_loss, self.critic.trainable_vars, clip_norm=self.clip_norm)
+        self.critic_optimizer = MpiAdam(var_list=self.critic.trainable_vars,
+            beta1=0.9, beta2=0.999, epsilon=1e-08)
+
+    def setup_popart(self):
+        # See https://arxiv.org/pdf/1602.07714.pdf for details.
+        self.old_std = tf.placeholder(tf.float32, shape=[1], name='old_std')
+        new_std = self.ret_rms.std
+        self.old_mean = tf.placeholder(tf.float32, shape=[1], name='old_mean')
+        new_mean = self.ret_rms.mean
+
+        self.renormalize_Q_outputs_op = []
+        for vs in [self.critic.output_vars, self.target_critic.output_vars]:
+            assert len(vs) == 2
+            M, b = vs
+            assert 'kernel' in M.name
+            assert 'bias' in b.name
+            assert M.get_shape()[-1] == 1
+            assert b.get_shape()[-1] == 1
+            self.renormalize_Q_outputs_op += [M.assign(M * self.old_std / new_std)]
+            self.renormalize_Q_outputs_op += [b.assign((b * self.old_std + self.old_mean - new_mean) / new_std)]
+
+    def setup_stats(self):
+        ops = []
+        names = []
+
+        if self.normalize_returns:
+            ops += [self.ret_rms.mean, self.ret_rms.std]
+            names += ['ret_rms_mean', 'ret_rms_std']
+
+        if self.normalize_observations:
+            ops += [tf.reduce_mean(self.obs_rms.mean), tf.reduce_mean(self.obs_rms.std)]
+            names += ['obs_rms_mean', 'obs_rms_std']
+
+        ops += [tf.reduce_mean(self.critic_tf)]
+        names += ['reference_Q_mean']
+        ops += [reduce_std(self.critic_tf)]
+        names += ['reference_Q_std']
+
+        ops += [tf.reduce_mean(self.critic_with_actor_tf)]
+        names += ['reference_actor_Q_mean']
+        ops += [reduce_std(self.critic_with_actor_tf)]
+        names += ['reference_actor_Q_std']
+
+        ops += [tf.reduce_mean(self.actor_tf)]
+        names += ['reference_action_mean']
+        ops += [reduce_std(self.actor_tf)]
+        names += ['reference_action_std']
+
+        if self.param_noise:
+            ops += [tf.reduce_mean(self.perturbed_actor_tf)]
+            names += ['reference_perturbed_action_mean']
+            ops += [reduce_std(self.perturbed_actor_tf)]
+            names += ['reference_perturbed_action_std']
+
+        self.stats_ops = ops
+        self.stats_names = names
+
+    def pi(self, obs, apply_noise=True, compute_Q=True):
+        if self.param_noise is not None and apply_noise:
+            actor_tf = self.perturbed_actor_tf
+        else:
+            actor_tf = self.actor_tf
+        feed_dict = {self.obs0: [obs]}
+        if compute_Q:
+            action, q = self.sess.run([actor_tf, self.critic_with_actor_tf], feed_dict=feed_dict)
+        else:
+            action = self.sess.run(actor_tf, feed_dict=feed_dict)
+            q = None
+        action = action.flatten()
+        if self.action_noise is not None and apply_noise:
+            noise = self.action_noise()
+            assert noise.shape == action.shape
+            action += noise
+        action = np.clip(action, self.action_range[0], self.action_range[1])
+        return action, q
+
+    def store_transition(self, obs0, action, reward, obs1, terminal1):
+        reward *= self.reward_scale
+        self.memory.append(obs0, action, reward, obs1, terminal1)
+        if self.normalize_observations:
+            self.obs_rms.update(np.array([obs0]))
+
+    def train(self):
+        # Get a batch.
+        batch = self.memory.sample(batch_size=self.batch_size)
+
+        if self.normalize_returns and self.enable_popart:
+            old_mean, old_std, target_Q = self.sess.run([self.ret_rms.mean, self.ret_rms.std, self.target_Q], feed_dict={
+                self.obs1: batch['obs1'],
+                self.rewards: batch['rewards'],
+                self.terminals1: batch['terminals1'].astype('float32'),
+            })
+            self.ret_rms.update(target_Q.flatten())
+            self.sess.run(self.renormalize_Q_outputs_op, feed_dict={
+                self.old_std : np.array([old_std]),
+                self.old_mean : np.array([old_mean]),
+            })
+
+            # Run sanity check. Disabled by default since it slows down things considerably.
+            # print('running sanity check')
+            # target_Q_new, new_mean, new_std = self.sess.run([self.target_Q, self.ret_rms.mean, self.ret_rms.std], feed_dict={
+            #     self.obs1: batch['obs1'],
+            #     self.rewards: batch['rewards'],
+            #     self.terminals1: batch['terminals1'].astype('float32'),
+            # })
+            # print(target_Q_new, target_Q, new_mean, new_std)
+            # assert (np.abs(target_Q - target_Q_new) < 1e-3).all()
+        else:
+            target_Q = self.sess.run(self.target_Q, feed_dict={
+                self.obs1: batch['obs1'],
+                self.rewards: batch['rewards'],
+                self.terminals1: batch['terminals1'].astype('float32'),
+            })
+
+        # Get all gradients and perform a synced update.
+        ops = [self.actor_grads, self.actor_loss, self.critic_grads, self.critic_loss]
+        actor_grads, actor_loss, critic_grads, critic_loss = self.sess.run(ops, feed_dict={
+            self.obs0: batch['obs0'],
+            self.actions: batch['actions'],
+            self.critic_target: target_Q,
+        })
+        self.actor_optimizer.update(actor_grads, stepsize=self.actor_lr)
+        self.critic_optimizer.update(critic_grads, stepsize=self.critic_lr)
+
+        return critic_loss, actor_loss
+
+    def initialize(self, sess):
+        self.sess = sess
+        self.sess.run(tf.global_variables_initializer())
+        self.actor_optimizer.sync()
+        self.critic_optimizer.sync()
+        self.sess.run(self.target_init_updates)
+
+    def update_target_net(self):
+        self.sess.run(self.target_soft_updates)
+
+    def get_stats(self):
+        if self.stats_sample is None:
+            # Get a sample and keep that fixed for all further computations.
+            # This allows us to estimate the change in value for the same set of inputs.
+            self.stats_sample = self.memory.sample(batch_size=self.batch_size)
+        values = self.sess.run(self.stats_ops, feed_dict={
+            self.obs0: self.stats_sample['obs0'],
+            self.actions: self.stats_sample['actions'],
+        })
+
+        names = self.stats_names[:]
+        assert len(names) == len(values)
+        stats = dict(zip(names, values))
+
+        if self.param_noise is not None:
+            stats = {**stats, **self.param_noise.get_stats()}
+
+        return stats
+
+    def adapt_param_noise(self):
+        if self.param_noise is None:
+            return 0.
+
+        # Perturb a separate copy of the policy to adjust the scale for the next "real" perturbation.
+        batch = self.memory.sample(batch_size=self.batch_size)
+        self.sess.run(self.perturb_adaptive_policy_ops, feed_dict={
+            self.param_noise_stddev: self.param_noise.current_stddev,
+        })
+        distance = self.sess.run(self.adaptive_policy_distance, feed_dict={
+            self.obs0: batch['obs0'],
+            self.param_noise_stddev: self.param_noise.current_stddev,
+        })
+
+        mean_distance = MPI.COMM_WORLD.allreduce(distance, op=MPI.SUM) / MPI.COMM_WORLD.Get_size()
+        self.param_noise.adapt(mean_distance)
+        return mean_distance
+
+    def reset(self):
+        # Reset internal state after an episode is complete.
+        if self.action_noise is not None:
+            self.action_noise.reset()
+        if self.param_noise is not None:
+            self.sess.run(self.perturb_policy_ops, feed_dict={
+                self.param_noise_stddev: self.param_noise.current_stddev,
+            })

baselines/ddpg/ddpg_learner.py

@@ -1,385 +0,0 @@
-from copy import copy
-from functools import reduce
-
-import numpy as np
-import tensorflow as tf
-import tensorflow.contrib as tc
-
-from baselines import logger
-from baselines.common.mpi_adam import MpiAdam
-import baselines.common.tf_util as U
-from baselines.common.mpi_running_mean_std import RunningMeanStd
-from mpi4py import MPI
-
-def normalize(x, stats):
-    if stats is None:
-        return x
-    return (x - stats.mean) / stats.std
-
-
-def denormalize(x, stats):
-    if stats is None:
-        return x
-    return x * stats.std + stats.mean
-
-def reduce_std(x, axis=None, keepdims=False):
-    return tf.sqrt(reduce_var(x, axis=axis, keepdims=keepdims))
-
-def reduce_var(x, axis=None, keepdims=False):
-    m = tf.reduce_mean(x, axis=axis, keepdims=True)
-    devs_squared = tf.square(x - m)
-    return tf.reduce_mean(devs_squared, axis=axis, keepdims=keepdims)
-
-def get_target_updates(vars, target_vars, tau):
-    logger.info('setting up target updates ...')
-    soft_updates = []
-    init_updates = []
-    assert len(vars) == len(target_vars)
-    for var, target_var in zip(vars, target_vars):
-        logger.info('  {} <- {}'.format(target_var.name, var.name))
-        init_updates.append(tf.assign(target_var, var))
-        soft_updates.append(tf.assign(target_var, (1. - tau) * target_var + tau * var))
-    assert len(init_updates) == len(vars)
-    assert len(soft_updates) == len(vars)
-    return tf.group(*init_updates), tf.group(*soft_updates)
-
-
-def get_perturbed_actor_updates(actor, perturbed_actor, param_noise_stddev):
-    assert len(actor.vars) == len(perturbed_actor.vars)
-    assert len(actor.perturbable_vars) == len(perturbed_actor.perturbable_vars)
-
-    updates = []
-    for var, perturbed_var in zip(actor.vars, perturbed_actor.vars):
-        if var in actor.perturbable_vars:
-            logger.info('  {} <- {} + noise'.format(perturbed_var.name, var.name))
-            updates.append(tf.assign(perturbed_var, var + tf.random_normal(tf.shape(var), mean=0., stddev=param_noise_stddev)))
-        else:
-            logger.info('  {} <- {}'.format(perturbed_var.name, var.name))
-            updates.append(tf.assign(perturbed_var, var))
-    assert len(updates) == len(actor.vars)
-    return tf.group(*updates)
-
-
-class DDPG(object):
-    def __init__(self, actor, critic, memory, observation_shape, action_shape, param_noise=None, action_noise=None,
-        gamma=0.99, tau=0.001, normalize_returns=False, enable_popart=False, normalize_observations=True,
-        batch_size=128, observation_range=(-5., 5.), action_range=(-1., 1.), return_range=(-np.inf, np.inf),
-        adaptive_param_noise=True, adaptive_param_noise_policy_threshold=.1,
-        critic_l2_reg=0., actor_lr=1e-4, critic_lr=1e-3, clip_norm=None, reward_scale=1.):
-        # Inputs.
-        self.obs0 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs0')
-        self.obs1 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs1')
-        self.terminals1 = tf.placeholder(tf.float32, shape=(None, 1), name='terminals1')
-        self.rewards = tf.placeholder(tf.float32, shape=(None, 1), name='rewards')
-        self.actions = tf.placeholder(tf.float32, shape=(None,) + action_shape, name='actions')
-        self.critic_target = tf.placeholder(tf.float32, shape=(None, 1), name='critic_target')
-        self.param_noise_stddev = tf.placeholder(tf.float32, shape=(), name='param_noise_stddev')
-
-        # Parameters.
-        self.gamma = gamma
-        self.tau = tau
-        self.memory = memory
-        self.normalize_observations = normalize_observations
-        self.normalize_returns = normalize_returns
-        self.action_noise = action_noise
-        self.param_noise = param_noise
-        self.action_range = action_range
-        self.return_range = return_range
-        self.observation_range = observation_range
-        self.critic = critic
-        self.actor = actor
-        self.actor_lr = actor_lr
-        self.critic_lr = critic_lr
-        self.clip_norm = clip_norm
-        self.enable_popart = enable_popart
-        self.reward_scale = reward_scale
-        self.batch_size = batch_size
-        self.stats_sample = None
-        self.critic_l2_reg = critic_l2_reg
-
-        # Observation normalization.
-        if self.normalize_observations:
-            with tf.variable_scope('obs_rms'):
-                self.obs_rms = RunningMeanStd(shape=observation_shape)
-        else:
-            self.obs_rms = None
-        normalized_obs0 = tf.clip_by_value(normalize(self.obs0, self.obs_rms),
-            self.observation_range[0], self.observation_range[1])
-        normalized_obs1 = tf.clip_by_value(normalize(self.obs1, self.obs_rms),
-            self.observation_range[0], self.observation_range[1])
-
-        # Return normalization.
-        if self.normalize_returns:
-            with tf.variable_scope('ret_rms'):
-                self.ret_rms = RunningMeanStd()
-        else:
-            self.ret_rms = None
-
-        # Create target networks.
-        target_actor = copy(actor)
-        target_actor.name = 'target_actor'
-        self.target_actor = target_actor
-        target_critic = copy(critic)
-        target_critic.name = 'target_critic'
-        self.target_critic = target_critic
-
-        # Create networks and core TF parts that are shared across setup parts.
-        self.actor_tf = actor(normalized_obs0)
-        self.normalized_critic_tf = critic(normalized_obs0, self.actions)
-        self.critic_tf = denormalize(tf.clip_by_value(self.normalized_critic_tf, self.return_range[0], self.return_range[1]), self.ret_rms)
-        self.normalized_critic_with_actor_tf = critic(normalized_obs0, self.actor_tf, reuse=True)
-        self.critic_with_actor_tf = denormalize(tf.clip_by_value(self.normalized_critic_with_actor_tf, self.return_range[0], self.return_range[1]), self.ret_rms)
-        Q_obs1 = denormalize(target_critic(normalized_obs1, target_actor(normalized_obs1)), self.ret_rms)
-        self.target_Q = self.rewards + (1. - self.terminals1) * gamma * Q_obs1
-
-        # Set up parts.
-        if self.param_noise is not None:
-            self.setup_param_noise(normalized_obs0)
-        self.setup_actor_optimizer()
-        self.setup_critic_optimizer()
-        if self.normalize_returns and self.enable_popart:
-            self.setup_popart()
-        self.setup_stats()
-        self.setup_target_network_updates()
-
-        self.initial_state = None # recurrent architectures not supported yet
-
-    def setup_target_network_updates(self):
-        actor_init_updates, actor_soft_updates = get_target_updates(self.actor.vars, self.target_actor.vars, self.tau)
-        critic_init_updates, critic_soft_updates = get_target_updates(self.critic.vars, self.target_critic.vars, self.tau)
-        self.target_init_updates = [actor_init_updates, critic_init_updates]
-        self.target_soft_updates = [actor_soft_updates, critic_soft_updates]
-
-    def setup_param_noise(self, normalized_obs0):
-        assert self.param_noise is not None
-
-        # Configure perturbed actor.
-        param_noise_actor = copy(self.actor)
-        param_noise_actor.name = 'param_noise_actor'
-        self.perturbed_actor_tf = param_noise_actor(normalized_obs0)
-        logger.info('setting up param noise')
-        self.perturb_policy_ops = get_perturbed_actor_updates(self.actor, param_noise_actor, self.param_noise_stddev)
-
-        # Configure separate copy for stddev adoption.
-        adaptive_param_noise_actor = copy(self.actor)
-        adaptive_param_noise_actor.name = 'adaptive_param_noise_actor'
-        adaptive_actor_tf = adaptive_param_noise_actor(normalized_obs0)
-        self.perturb_adaptive_policy_ops = get_perturbed_actor_updates(self.actor, adaptive_param_noise_actor, self.param_noise_stddev)
-        self.adaptive_policy_distance = tf.sqrt(tf.reduce_mean(tf.square(self.actor_tf - adaptive_actor_tf)))
-
-    def setup_actor_optimizer(self):
-        logger.info('setting up actor optimizer')
-        self.actor_loss = -tf.reduce_mean(self.critic_with_actor_tf)
-        actor_shapes = [var.get_shape().as_list() for var in self.actor.trainable_vars]
-        actor_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in actor_shapes])
-        logger.info('  actor shapes: {}'.format(actor_shapes))
-        logger.info('  actor params: {}'.format(actor_nb_params))
-        self.actor_grads = U.flatgrad(self.actor_loss, self.actor.trainable_vars, clip_norm=self.clip_norm)
-        self.actor_optimizer = MpiAdam(var_list=self.actor.trainable_vars,
-            beta1=0.9, beta2=0.999, epsilon=1e-08)
-
-    def setup_critic_optimizer(self):
-        logger.info('setting up critic optimizer')
-        normalized_critic_target_tf = tf.clip_by_value(normalize(self.critic_target, self.ret_rms), self.return_range[0], self.return_range[1])
-        self.critic_loss = tf.reduce_mean(tf.square(self.normalized_critic_tf - normalized_critic_target_tf))
-        if self.critic_l2_reg > 0.:
-            critic_reg_vars = [var for var in self.critic.trainable_vars if 'kernel' in var.name and 'output' not in var.name]
-            for var in critic_reg_vars:
-                logger.info('  regularizing: {}'.format(var.name))
-            logger.info('  applying l2 regularization with {}'.format(self.critic_l2_reg))
-            critic_reg = tc.layers.apply_regularization(
-                tc.layers.l2_regularizer(self.critic_l2_reg),
-                weights_list=critic_reg_vars
-            )
-            self.critic_loss += critic_reg
-        critic_shapes = [var.get_shape().as_list() for var in self.critic.trainable_vars]
-        critic_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in critic_shapes])
-        logger.info('  critic shapes: {}'.format(critic_shapes))
-        logger.info('  critic params: {}'.format(critic_nb_params))
-        self.critic_grads = U.flatgrad(self.critic_loss, self.critic.trainable_vars, clip_norm=self.clip_norm)
-        self.critic_optimizer = MpiAdam(var_list=self.critic.trainable_vars,
-            beta1=0.9, beta2=0.999, epsilon=1e-08)
-
-    def setup_popart(self):
-        # See https://arxiv.org/pdf/1602.07714.pdf for details.
-        self.old_std = tf.placeholder(tf.float32, shape=[1], name='old_std')
-        new_std = self.ret_rms.std
-        self.old_mean = tf.placeholder(tf.float32, shape=[1], name='old_mean')
-        new_mean = self.ret_rms.mean
-
-        self.renormalize_Q_outputs_op = []
-        for vs in [self.critic.output_vars, self.target_critic.output_vars]:
-            assert len(vs) == 2
-            M, b = vs
-            assert 'kernel' in M.name
-            assert 'bias' in b.name
-            assert M.get_shape()[-1] == 1
-            assert b.get_shape()[-1] == 1
-            self.renormalize_Q_outputs_op += [M.assign(M * self.old_std / new_std)]
-            self.renormalize_Q_outputs_op += [b.assign((b * self.old_std + self.old_mean - new_mean) / new_std)]
-
-    def setup_stats(self):
-        ops = []
-        names = []
-
-        if self.normalize_returns:
-            ops += [self.ret_rms.mean, self.ret_rms.std]
-            names += ['ret_rms_mean', 'ret_rms_std']
-
-        if self.normalize_observations:
-            ops += [tf.reduce_mean(self.obs_rms.mean), tf.reduce_mean(self.obs_rms.std)]
-            names += ['obs_rms_mean', 'obs_rms_std']
-
-        ops += [tf.reduce_mean(self.critic_tf)]
-        names += ['reference_Q_mean']
-        ops += [reduce_std(self.critic_tf)]
-        names += ['reference_Q_std']
-
-        ops += [tf.reduce_mean(self.critic_with_actor_tf)]
-        names += ['reference_actor_Q_mean']
-        ops += [reduce_std(self.critic_with_actor_tf)]
-        names += ['reference_actor_Q_std']
-
-        ops += [tf.reduce_mean(self.actor_tf)]
-        names += ['reference_action_mean']
-        ops += [reduce_std(self.actor_tf)]
-        names += ['reference_action_std']
-
-        if self.param_noise:
-            ops += [tf.reduce_mean(self.perturbed_actor_tf)]
-            names += ['reference_perturbed_action_mean']
-            ops += [reduce_std(self.perturbed_actor_tf)]
-            names += ['reference_perturbed_action_std']
-
-        self.stats_ops = ops
-        self.stats_names = names
-
-    def step(self, obs, apply_noise=True, compute_Q=True):
-        if self.param_noise is not None and apply_noise:
-            actor_tf = self.perturbed_actor_tf
-        else:
-            actor_tf = self.actor_tf
-        feed_dict = {self.obs0: U.adjust_shape(self.obs0, [obs])}
-        if compute_Q:
-            action, q = self.sess.run([actor_tf, self.critic_with_actor_tf], feed_dict=feed_dict)
-        else:
-            action = self.sess.run(actor_tf, feed_dict=feed_dict)
-            q = None
-
-        if self.action_noise is not None and apply_noise:
-            noise = self.action_noise()
-            assert noise.shape == action.shape
-            action += noise
-        action = np.clip(action, self.action_range[0], self.action_range[1])
-
-
-        return action, q, None, None
-
-    def store_transition(self, obs0, action, reward, obs1, terminal1):
-        reward *= self.reward_scale
-
-        B = obs0.shape[0]
-        for b in range(B):
-            self.memory.append(obs0[b], action[b], reward[b], obs1[b], terminal1[b])
-            if self.normalize_observations:
-                self.obs_rms.update(np.array([obs0[b]]))
-
-    def train(self):
-        # Get a batch.
-        batch = self.memory.sample(batch_size=self.batch_size)
-
-        if self.normalize_returns and self.enable_popart:
-            old_mean, old_std, target_Q = self.sess.run([self.ret_rms.mean, self.ret_rms.std, self.target_Q], feed_dict={
-                self.obs1: batch['obs1'],
-                self.rewards: batch['rewards'],
-                self.terminals1: batch['terminals1'].astype('float32'),
-            })
-            self.ret_rms.update(target_Q.flatten())
-            self.sess.run(self.renormalize_Q_outputs_op, feed_dict={
-                self.old_std : np.array([old_std]),
-                self.old_mean : np.array([old_mean]),
-            })
-
-            # Run sanity check. Disabled by default since it slows down things considerably.
-            # print('running sanity check')
-            # target_Q_new, new_mean, new_std = self.sess.run([self.target_Q, self.ret_rms.mean, self.ret_rms.std], feed_dict={
-            #     self.obs1: batch['obs1'],
-            #     self.rewards: batch['rewards'],
-            #     self.terminals1: batch['terminals1'].astype('float32'),
-            # })
-            # print(target_Q_new, target_Q, new_mean, new_std)
-            # assert (np.abs(target_Q - target_Q_new) < 1e-3).all()
-        else:
-            target_Q = self.sess.run(self.target_Q, feed_dict={
-                self.obs1: batch['obs1'],
-                self.rewards: batch['rewards'],
-                self.terminals1: batch['terminals1'].astype('float32'),
-            })
-
-        # Get all gradients and perform a synced update.
-        ops = [self.actor_grads, self.actor_loss, self.critic_grads, self.critic_loss]
-        actor_grads, actor_loss, critic_grads, critic_loss = self.sess.run(ops, feed_dict={
-            self.obs0: batch['obs0'],
-            self.actions: batch['actions'],
-            self.critic_target: target_Q,
-        })
-        self.actor_optimizer.update(actor_grads, stepsize=self.actor_lr)
-        self.critic_optimizer.update(critic_grads, stepsize=self.critic_lr)
-
-        return critic_loss, actor_loss
-
-    def initialize(self, sess):
-        self.sess = sess
-        self.sess.run(tf.global_variables_initializer())
-        self.actor_optimizer.sync()
-        self.critic_optimizer.sync()
-        self.sess.run(self.target_init_updates)
-
-    def update_target_net(self):
-        self.sess.run(self.target_soft_updates)
-
-    def get_stats(self):
-        if self.stats_sample is None:
-            # Get a sample and keep that fixed for all further computations.
-            # This allows us to estimate the change in value for the same set of inputs.
-            self.stats_sample = self.memory.sample(batch_size=self.batch_size)
-        values = self.sess.run(self.stats_ops, feed_dict={
-            self.obs0: self.stats_sample['obs0'],
-            self.actions: self.stats_sample['actions'],
-        })
-
-        names = self.stats_names[:]
-        assert len(names) == len(values)
-        stats = dict(zip(names, values))
-
-        if self.param_noise is not None:
-            stats = {**stats, **self.param_noise.get_stats()}
-
-        return stats
-
-    def adapt_param_noise(self):
-        if self.param_noise is None:
-            return 0.
-
-        # Perturb a separate copy of the policy to adjust the scale for the next "real" perturbation.
-        batch = self.memory.sample(batch_size=self.batch_size)
-        self.sess.run(self.perturb_adaptive_policy_ops, feed_dict={
-            self.param_noise_stddev: self.param_noise.current_stddev,
-        })
-        distance = self.sess.run(self.adaptive_policy_distance, feed_dict={
-            self.obs0: batch['obs0'],
-            self.param_noise_stddev: self.param_noise.current_stddev,
-        })
-
-        mean_distance = MPI.COMM_WORLD.allreduce(distance, op=MPI.SUM) / MPI.COMM_WORLD.Get_size()
-        self.param_noise.adapt(mean_distance)
-        return mean_distance
-
-    def reset(self):
-        # Reset internal state after an episode is complete.
-        if self.action_noise is not None:
-            self.action_noise.reset()
-        if self.param_noise is not None:
-            self.sess.run(self.perturb_policy_ops, feed_dict={
-                self.param_noise_stddev: self.param_noise.current_stddev,
-            })

baselines/ddpg/main.py

@@ -0,0 +1,123 @@
+import argparse
+import time
+import os
+import logging
+from baselines import logger, bench
+from baselines.common.misc_util import (
+    set_global_seeds,
+    boolean_flag,
+)
+import baselines.ddpg.training as training
+from baselines.ddpg.models import Actor, Critic
+from baselines.ddpg.memory import Memory
+from baselines.ddpg.noise import *
+
+import gym
+import tensorflow as tf
+from mpi4py import MPI
+
+def run(env_id, seed, noise_type, layer_norm, evaluation, **kwargs):
+    # Configure things.
+    rank = MPI.COMM_WORLD.Get_rank()
+    if rank != 0:
+        logger.set_level(logger.DISABLED)
+
+    # Create envs.
+    env = gym.make(env_id)
+    env = bench.Monitor(env, logger.get_dir() and os.path.join(logger.get_dir(), str(rank)))
+
+    if evaluation and rank==0:
+        eval_env = gym.make(env_id)
+        eval_env = bench.Monitor(eval_env, os.path.join(logger.get_dir(), 'gym_eval'))
+        env = bench.Monitor(env, None)
+    else:
+        eval_env = None
+
+    # Parse noise_type
+    action_noise = None
+    param_noise = None
+    nb_actions = env.action_space.shape[-1]
+    for current_noise_type in noise_type.split(','):
+        current_noise_type = current_noise_type.strip()
+        if current_noise_type == 'none':
+            pass
+        elif 'adaptive-param' in current_noise_type:
+            _, stddev = current_noise_type.split('_')
+            param_noise = AdaptiveParamNoiseSpec(initial_stddev=float(stddev), desired_action_stddev=float(stddev))
+        elif 'normal' in current_noise_type:
+            _, stddev = current_noise_type.split('_')
+            action_noise = NormalActionNoise(mu=np.zeros(nb_actions), sigma=float(stddev) * np.ones(nb_actions))
+        elif 'ou' in current_noise_type:
+            _, stddev = current_noise_type.split('_')
+            action_noise = OrnsteinUhlenbeckActionNoise(mu=np.zeros(nb_actions), sigma=float(stddev) * np.ones(nb_actions))
+        else:
+            raise RuntimeError('unknown noise type "{}"'.format(current_noise_type))
+
+    # Configure components.
+    memory = Memory(limit=int(1e6), action_shape=env.action_space.shape, observation_shape=env.observation_space.shape)
+    critic = Critic(layer_norm=layer_norm)
+    actor = Actor(nb_actions, layer_norm=layer_norm)
+
+    # Seed everything to make things reproducible.
+    seed = seed + 1000000 * rank
+    logger.info('rank {}: seed={}, logdir={}'.format(rank, seed, logger.get_dir()))
+    tf.reset_default_graph()
+    set_global_seeds(seed)
+    env.seed(seed)
+    if eval_env is not None:
+        eval_env.seed(seed)
+
+    # Disable logging for rank != 0 to avoid noise.
+    if rank == 0:
+        start_time = time.time()
+    training.train(env=env, eval_env=eval_env, param_noise=param_noise,
+        action_noise=action_noise, actor=actor, critic=critic, memory=memory, **kwargs)
+    env.close()
+    if eval_env is not None:
+        eval_env.close()
+    if rank == 0:
+        logger.info('total runtime: {}s'.format(time.time() - start_time))
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+
+    parser.add_argument('--env-id', type=str, default='HalfCheetah-v1')
+    boolean_flag(parser, 'render-eval', default=False)
+    boolean_flag(parser, 'layer-norm', default=True)
+    boolean_flag(parser, 'render', default=False)
+    boolean_flag(parser, 'normalize-returns', default=False)
+    boolean_flag(parser, 'normalize-observations', default=True)
+    parser.add_argument('--seed', help='RNG seed', type=int, default=0)
+    parser.add_argument('--critic-l2-reg', type=float, default=1e-2)
+    parser.add_argument('--batch-size', type=int, default=64)  # per MPI worker
+    parser.add_argument('--actor-lr', type=float, default=1e-4)
+    parser.add_argument('--critic-lr', type=float, default=1e-3)
+    boolean_flag(parser, 'popart', default=False)
+    parser.add_argument('--gamma', type=float, default=0.99)
+    parser.add_argument('--reward-scale', type=float, default=1.)
+    parser.add_argument('--clip-norm', type=float, default=None)
+    parser.add_argument('--nb-epochs', type=int, default=500)  # with default settings, perform 1M steps total
+    parser.add_argument('--nb-epoch-cycles', type=int, default=20)
+    parser.add_argument('--nb-train-steps', type=int, default=50)  # per epoch cycle and MPI worker
+    parser.add_argument('--nb-eval-steps', type=int, default=100)  # per epoch cycle and MPI worker
+    parser.add_argument('--nb-rollout-steps', type=int, default=100)  # per epoch cycle and MPI worker
+    parser.add_argument('--noise-type', type=str, default='adaptive-param_0.2')  # choices are adaptive-param_xx, ou_xx, normal_xx, none
+    parser.add_argument('--num-timesteps', type=int, default=None)
+    boolean_flag(parser, 'evaluation', default=False)
+    args = parser.parse_args()
+    # we don't directly specify timesteps for this script, so make sure that if we do specify them
+    # they agree with the other parameters
+    if args.num_timesteps is not None:
+        assert(args.num_timesteps == args.nb_epochs * args.nb_epoch_cycles * args.nb_rollout_steps)
+    dict_args = vars(args)
+    del dict_args['num_timesteps']
+    return dict_args
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    if MPI.COMM_WORLD.Get_rank() == 0:
+        logger.configure()
+    # Run actual script.
+    run(**args)

baselines/ddpg/memory.py

@@ -51,7 +51,7 @@ class Memory(object):
 
     def sample(self, batch_size):
         # Draw such that we always have a proceeding element.
-        batch_idxs = np.random.randint(self.nb_entries - 2, size=batch_size)
+        batch_idxs = np.random.random_integers(self.nb_entries - 2, size=batch_size)
 
         obs0_batch = self.observations0.get_batch(batch_idxs)
         obs1_batch = self.observations1.get_batch(batch_idxs)
@@ -71,7 +71,7 @@ class Memory(object):
     def append(self, obs0, action, reward, obs1, terminal1, training=True):
         if not training:
             return
-
+        
         self.observations0.append(obs0)
         self.actions.append(action)
         self.rewards.append(reward)

baselines/ddpg/models.py

@@ -1,11 +1,10 @@
 import tensorflow as tf
-from baselines.common.models import get_network_builder
+import tensorflow.contrib as tc
 
 
 class Model(object):
-    def __init__(self, name, network='mlp', **network_kwargs):
+    def __init__(self, name):
         self.name = name
-        self.network_builder = get_network_builder(network)(**network_kwargs)
 
     @property
     def vars(self):
@@ -21,27 +20,54 @@ class Model(object):
 
 
 class Actor(Model):
-    def __init__(self, nb_actions, name='actor', network='mlp', **network_kwargs):
-        super().__init__(name=name, network=network, **network_kwargs)
+    def __init__(self, nb_actions, name='actor', layer_norm=True):
+        super(Actor, self).__init__(name=name)
         self.nb_actions = nb_actions
+        self.layer_norm = layer_norm
 
     def __call__(self, obs, reuse=False):
-        with tf.variable_scope(self.name, reuse=tf.AUTO_REUSE):
-            x = self.network_builder(obs)
+        with tf.variable_scope(self.name) as scope:
+            if reuse:
+                scope.reuse_variables()
+
+            x = obs
+            x = tf.layers.dense(x, 64)
+            if self.layer_norm:
+                x = tc.layers.layer_norm(x, center=True, scale=True)
+            x = tf.nn.relu(x)
+            
+            x = tf.layers.dense(x, 64)
+            if self.layer_norm:
+                x = tc.layers.layer_norm(x, center=True, scale=True)
+            x = tf.nn.relu(x)
+            
             x = tf.layers.dense(x, self.nb_actions, kernel_initializer=tf.random_uniform_initializer(minval=-3e-3, maxval=3e-3))
             x = tf.nn.tanh(x)
         return x
 
 
 class Critic(Model):
-    def __init__(self, name='critic', network='mlp', **network_kwargs):
-        super().__init__(name=name, network=network, **network_kwargs)
-        self.layer_norm = True
+    def __init__(self, name='critic', layer_norm=True):
+        super(Critic, self).__init__(name=name)
+        self.layer_norm = layer_norm
 
     def __call__(self, obs, action, reuse=False):
-        with tf.variable_scope(self.name, reuse=tf.AUTO_REUSE):
-            x = tf.concat([obs, action], axis=-1) # this assumes observation and action can be concatenated
-            x = self.network_builder(x)
+        with tf.variable_scope(self.name) as scope:
+            if reuse:
+                scope.reuse_variables()
+
+            x = obs
+            x = tf.layers.dense(x, 64)
+            if self.layer_norm:
+                x = tc.layers.layer_norm(x, center=True, scale=True)
+            x = tf.nn.relu(x)
+
+            x = tf.concat([x, action], axis=-1)
+            x = tf.layers.dense(x, 64)
+            if self.layer_norm:
+                x = tc.layers.layer_norm(x, center=True, scale=True)
+            x = tf.nn.relu(x)
+
             x = tf.layers.dense(x, 1, kernel_initializer=tf.random_uniform_initializer(minval=-3e-3, maxval=3e-3))
         return x
 

baselines/ddpg/training.py

@@ -0,0 +1,191 @@
+import os
+import time
+from collections import deque
+import pickle
+
+from baselines.ddpg.ddpg import DDPG
+import baselines.common.tf_util as U
+
+from baselines import logger
+import numpy as np
+import tensorflow as tf
+from mpi4py import MPI
+
+
+def train(env, nb_epochs, nb_epoch_cycles, render_eval, reward_scale, render, param_noise, actor, critic,
+    normalize_returns, normalize_observations, critic_l2_reg, actor_lr, critic_lr, action_noise,
+    popart, gamma, clip_norm, nb_train_steps, nb_rollout_steps, nb_eval_steps, batch_size, memory,
+    tau=0.01, eval_env=None, param_noise_adaption_interval=50):
+    rank = MPI.COMM_WORLD.Get_rank()
+
+    assert (np.abs(env.action_space.low) == env.action_space.high).all()  # we assume symmetric actions.
+    max_action = env.action_space.high
+    logger.info('scaling actions by {} before executing in env'.format(max_action))
+    agent = DDPG(actor, critic, memory, env.observation_space.shape, env.action_space.shape,
+        gamma=gamma, tau=tau, normalize_returns=normalize_returns, normalize_observations=normalize_observations,
+        batch_size=batch_size, action_noise=action_noise, param_noise=param_noise, critic_l2_reg=critic_l2_reg,
+        actor_lr=actor_lr, critic_lr=critic_lr, enable_popart=popart, clip_norm=clip_norm,
+        reward_scale=reward_scale)
+    logger.info('Using agent with the following configuration:')
+    logger.info(str(agent.__dict__.items()))
+
+    # Set up logging stuff only for a single worker.
+    if rank == 0:
+        saver = tf.train.Saver()
+    else:
+        saver = None
+
+    step = 0
+    episode = 0
+    eval_episode_rewards_history = deque(maxlen=100)
+    episode_rewards_history = deque(maxlen=100)
+    with U.single_threaded_session() as sess:
+        # Prepare everything.
+        agent.initialize(sess)
+        sess.graph.finalize()
+
+        agent.reset()
+        obs = env.reset()
+        if eval_env is not None:
+            eval_obs = eval_env.reset()
+        done = False
+        episode_reward = 0.
+        episode_step = 0
+        episodes = 0
+        t = 0
+
+        epoch = 0
+        start_time = time.time()
+
+        epoch_episode_rewards = []
+        epoch_episode_steps = []
+        epoch_episode_eval_rewards = []
+        epoch_episode_eval_steps = []
+        epoch_start_time = time.time()
+        epoch_actions = []
+        epoch_qs = []
+        epoch_episodes = 0
+        for epoch in range(nb_epochs):
+            for cycle in range(nb_epoch_cycles):
+                # Perform rollouts.
+                for t_rollout in range(nb_rollout_steps):
+                    # Predict next action.
+                    action, q = agent.pi(obs, apply_noise=True, compute_Q=True)
+                    assert action.shape == env.action_space.shape
+
+                    # Execute next action.
+                    if rank == 0 and render:
+                        env.render()
+                    assert max_action.shape == action.shape
+                    new_obs, r, done, info = env.step(max_action * action)  # scale for execution in env (as far as DDPG is concerned, every action is in [-1, 1])
+                    t += 1
+                    if rank == 0 and render:
+                        env.render()
+                    episode_reward += r
+                    episode_step += 1
+
+                    # Book-keeping.
+                    epoch_actions.append(action)
+                    epoch_qs.append(q)
+                    agent.store_transition(obs, action, r, new_obs, done)
+                    obs = new_obs
+
+                    if done:
+                        # Episode done.
+                        epoch_episode_rewards.append(episode_reward)
+                        episode_rewards_history.append(episode_reward)
+                        epoch_episode_steps.append(episode_step)
+                        episode_reward = 0.
+                        episode_step = 0
+                        epoch_episodes += 1
+                        episodes += 1
+
+                        agent.reset()
+                        obs = env.reset()
+
+                # Train.
+                epoch_actor_losses = []
+                epoch_critic_losses = []
+                epoch_adaptive_distances = []
+                for t_train in range(nb_train_steps):
+                    # Adapt param noise, if necessary.
+                    if memory.nb_entries >= batch_size and t_train % param_noise_adaption_interval == 0:
+                        distance = agent.adapt_param_noise()
+                        epoch_adaptive_distances.append(distance)
+
+                    cl, al = agent.train()
+                    epoch_critic_losses.append(cl)
+                    epoch_actor_losses.append(al)
+                    agent.update_target_net()
+
+                # Evaluate.
+                eval_episode_rewards = []
+                eval_qs = []
+                if eval_env is not None:
+                    eval_episode_reward = 0.
+                    for t_rollout in range(nb_eval_steps):
+                        eval_action, eval_q = agent.pi(eval_obs, apply_noise=False, compute_Q=True)
+                        eval_obs, eval_r, eval_done, eval_info = eval_env.step(max_action * eval_action)  # scale for execution in env (as far as DDPG is concerned, every action is in [-1, 1])
+                        if render_eval:
+                            eval_env.render()
+                        eval_episode_reward += eval_r
+
+                        eval_qs.append(eval_q)
+                        if eval_done:
+                            eval_obs = eval_env.reset()
+                            eval_episode_rewards.append(eval_episode_reward)
+                            eval_episode_rewards_history.append(eval_episode_reward)
+                            eval_episode_reward = 0.
+
+            mpi_size = MPI.COMM_WORLD.Get_size()
+            # Log stats.
+            # XXX shouldn't call np.mean on variable length lists
+            duration = time.time() - start_time
+            stats = agent.get_stats()
+            combined_stats = stats.copy()
+            combined_stats['rollout/return'] = np.mean(epoch_episode_rewards)
+            combined_stats['rollout/return_history'] = np.mean(episode_rewards_history)
+            combined_stats['rollout/episode_steps'] = np.mean(epoch_episode_steps)
+            combined_stats['rollout/actions_mean'] = np.mean(epoch_actions)
+            combined_stats['rollout/Q_mean'] = np.mean(epoch_qs)
+            combined_stats['train/loss_actor'] = np.mean(epoch_actor_losses)
+            combined_stats['train/loss_critic'] = np.mean(epoch_critic_losses)
+            combined_stats['train/param_noise_distance'] = np.mean(epoch_adaptive_distances)
+            combined_stats['total/duration'] = duration
+            combined_stats['total/steps_per_second'] = float(t) / float(duration)
+            combined_stats['total/episodes'] = episodes
+            combined_stats['rollout/episodes'] = epoch_episodes
+            combined_stats['rollout/actions_std'] = np.std(epoch_actions)
+            # Evaluation statistics.
+            if eval_env is not None:
+                combined_stats['eval/return'] = eval_episode_rewards
+                combined_stats['eval/return_history'] = np.mean(eval_episode_rewards_history)
+                combined_stats['eval/Q'] = eval_qs
+                combined_stats['eval/episodes'] = len(eval_episode_rewards)
+            def as_scalar(x):
+                if isinstance(x, np.ndarray):
+                    assert x.size == 1
+                    return x[0]
+                elif np.isscalar(x):
+                    return x
+                else:
+                    raise ValueError('expected scalar, got %s'%x)
+            combined_stats_sums = MPI.COMM_WORLD.allreduce(np.array([as_scalar(x) for x in combined_stats.values()]))
+            combined_stats = {k : v / mpi_size for (k,v) in zip(combined_stats.keys(), combined_stats_sums)}
+
+            # Total statistics.
+            combined_stats['total/epochs'] = epoch + 1
+            combined_stats['total/steps'] = t
+
+            for key in sorted(combined_stats.keys()):
+                logger.record_tabular(key, combined_stats[key])
+            logger.dump_tabular()
+            logger.info('')
+            logdir = logger.get_dir()
+            if rank == 0 and logdir:
+                if hasattr(env, 'get_state'):
+                    with open(os.path.join(logdir, 'env_state.pkl'), 'wb') as f:
+                        pickle.dump(env.get_state(), f)
+                if eval_env and hasattr(eval_env, 'get_state'):
+                    with open(os.path.join(logdir, 'eval_env_state.pkl'), 'wb') as f:
+                        pickle.dump(eval_env.get_state(), f)

baselines/deepq/README.md

@@ -9,29 +9,44 @@ Here's a list of commands to run to quickly get a working example:
 
 ```bash
 # Train model and save the results to cartpole_model.pkl
-python -m baselines.run --alg=deepq --env=CartPole-v0 --save_path=./cartpole_model.pkl --num_timesteps=1e5
+python -m baselines.deepq.experiments.train_cartpole
 # Load the model saved in cartpole_model.pkl and visualize the learned policy
-python -m baselines.run --alg=deepq --env=CartPole-v0 --load_path=./cartpole_model.pkl --num_timesteps=0 --play
+python -m baselines.deepq.experiments.enjoy_cartpole
 ```
 
+
+Be sure to check out the source code of [both](experiments/train_cartpole.py) [files](experiments/enjoy_cartpole.py)!
+
 ## If you wish to apply DQN to solve a problem.
 
 Check out our simple agent trained with one stop shop `deepq.learn` function. 
 
 - [baselines/deepq/experiments/train_cartpole.py](experiments/train_cartpole.py) - train a Cartpole agent.
+- [baselines/deepq/experiments/train_pong.py](experiments/train_pong.py) - train a Pong agent using convolutional neural networks.
 
-In particular notice that once `deepq.learn` finishes training it returns `act` function which can be used to select actions in the environment. Once trained you can easily save it and load at later time. Complimentary file `enjoy_cartpole.py` loads and visualizes the learned policy.
+In particular notice that once `deepq.learn` finishes training it returns `act` function which can be used to select actions in the environment. Once trained you can easily save it and load at later time. For both of the files listed above there are complimentary files `enjoy_cartpole.py` and `enjoy_pong.py` respectively, that load and visualize the learned policy.
 
 ## If you wish to experiment with the algorithm
 
 ##### Check out the examples
 
+
 - [baselines/deepq/experiments/custom_cartpole.py](experiments/custom_cartpole.py) - Cartpole training with more fine grained control over the internals of DQN algorithm.
-- [baselines/deepq/defaults.py](defaults.py) - settings for training on atari. Run 
+- [baselines/deepq/experiments/run_atari.py](experiments/run_atari.py) - more robust setup for training at scale.
+
+
+##### Download a pretrained Atari agent
+
+For some research projects it is sometimes useful to have an already trained agent handy. There's a variety of models to choose from. You can list them all by running:
 
 ```bash
-python -m baselines.run --alg=deepq --env=PongNoFrameskip-v4 
+python -m baselines.deepq.experiments.atari.download_model
 ```
-to train on Atari Pong (see more in repo-wide [README.md](../../README.md#training-models))
 
+Once you pick a model, you can download it and visualize the learned policy. Be sure to pass `--dueling` flag to visualization script when using dueling models.
 
+```bash
+python -m baselines.deepq.experiments.atari.download_model --blob model-atari-duel-pong-1 --model-dir /tmp/models
+python -m baselines.deepq.experiments.atari.enjoy --model-dir /tmp/models/model-atari-duel-pong-1 --env Pong --dueling
+
+```

baselines/deepq/build_graph.py

@@ -309,7 +309,7 @@ def build_act_with_param_noise(make_obs_ph, q_func, num_actions, scope="deepq",
                          outputs=output_actions,
                          givens={update_eps_ph: -1.0, stochastic_ph: True, reset_ph: False, update_param_noise_threshold_ph: False, update_param_noise_scale_ph: False},
                          updates=updates)
-        def act(ob, reset=False, update_param_noise_threshold=False, update_param_noise_scale=False, stochastic=True, update_eps=-1):
+        def act(ob, reset, update_param_noise_threshold, update_param_noise_scale, stochastic=True, update_eps=-1):
             return _act(ob, stochastic, update_eps, reset, update_param_noise_threshold, update_param_noise_scale)
         return act
 

baselines/deepq/deepq.py

@@ -7,8 +7,8 @@ import cloudpickle
 import numpy as np
 
 import baselines.common.tf_util as U
-from baselines.common.tf_util import load_variables, save_variables
-from baselines import logger, registry
+from baselines.common.tf_util import load_state, save_state
+from baselines import logger
 from baselines.common.schedules import LinearSchedule
 from baselines.common import set_global_seeds
 
@@ -18,7 +18,6 @@ from baselines.deepq.utils import ObservationInput
 
 from baselines.common.tf_util import get_session
 from baselines.deepq.models import build_q_func
-from baselines.deepq.defaults import defaults
 
 
 class ActWrapper(object):
@@ -28,7 +27,7 @@ class ActWrapper(object):
         self.initial_state = None
 
     @staticmethod
-    def load_act(path):
+    def load_act(self, path):
         with open(path, "rb") as f:
             model_data, act_params = cloudpickle.load(f)
         act = deepq.build_act(**act_params)
@@ -40,7 +39,7 @@ class ActWrapper(object):
                 f.write(model_data)
 
             zipfile.ZipFile(arc_path, 'r', zipfile.ZIP_DEFLATED).extractall(td)
-            load_variables(os.path.join(td, "model"))
+            load_state(os.path.join(td, "model"))
 
         return ActWrapper(act, act_params)
 
@@ -48,9 +47,6 @@ class ActWrapper(object):
         return self._act(*args, **kwargs)
 
     def step(self, observation, **kwargs):
-        # DQN doesn't use RNNs so we ignore states and masks
-        kwargs.pop('S', None)
-        kwargs.pop('M', None)
         return self._act([observation], **kwargs), None, None, None
 
     def save_act(self, path=None):
@@ -59,7 +55,7 @@ class ActWrapper(object):
             path = os.path.join(logger.get_dir(), "model.pkl")
 
         with tempfile.TemporaryDirectory() as td:
-            save_variables(os.path.join(td, "model"))
+            save_state(os.path.join(td, "model"))
             arc_name = os.path.join(td, "packed.zip")
             with zipfile.ZipFile(arc_name, 'w') as zipf:
                 for root, dirs, files in os.walk(td):
@@ -73,7 +69,7 @@ class ActWrapper(object):
             cloudpickle.dump((model_data, self._act_params), f)
 
     def save(self, path):
-        save_variables(path)
+        save_state(path)
 
 
 def load_act(path):
@@ -93,7 +89,6 @@ def load_act(path):
     return ActWrapper.load_act(path)
 
 
-@registry.register('deepq', supports_vecenv=False, defaults=defaults)
 def learn(env,
           network,
           seed=None,
@@ -126,12 +121,16 @@ def learn(env,
     -------
     env: gym.Env
         environment to train on
-    network: string or a function
-        neural network to use as a q function approximator. If string, has to be one of the names of registered models in baselines.common.models
-        (mlp, cnn, conv_only). If a function, should take an observation tensor and return a latent variable tensor, which
-        will be mapped to the Q function heads (see build_q_func in baselines.deepq.models for details on that)
-    seed: int or None
-        prng seed. The runs with the same seed "should" give the same results. If None, no seeding is used.
+    q_func: (tf.Variable, int, str, bool) -> tf.Variable
+        the model that takes the following inputs:
+            observation_in: object
+                the output of observation placeholder
+            num_actions: int
+                number of actions
+            scope: str
+            reuse: bool
+                should be passed to outer variable scope
+        and returns a tensor of shape (batch_size, num_actions) with values of every action.
     lr: float
         learning rate for adam optimizer
     total_timesteps: int
@@ -177,7 +176,7 @@ def learn(env,
     load_path: str
         path to load the model from. (default: None)
     **network_kwargs
-        additional keyword arguments to pass to the network builder.
+        additional keyword arguments to pass to the network builder. 
 
     Returns
     -------
@@ -195,9 +194,8 @@ def learn(env,
     # capture the shape outside the closure so that the env object is not serialized
     # by cloudpickle when serializing make_obs_ph
 
-    observation_space = env.observation_space
     def make_obs_ph(name):
-        return ObservationInput(observation_space, name=name)
+        return ObservationInput(env.observation_space, name=name)
 
     act, train, update_target, debug = deepq.build_train(
         make_obs_ph=make_obs_ph,
@@ -216,7 +214,7 @@ def learn(env,
     }
 
     act = ActWrapper(act, act_params)
-
+  
     # Create the replay buffer
     if prioritized_replay:
         replay_buffer = PrioritizedReplayBuffer(buffer_size, alpha=prioritized_replay_alpha)
@@ -247,15 +245,15 @@ def learn(env,
 
         model_file = os.path.join(td, "model")
         model_saved = False
-
+        
         if tf.train.latest_checkpoint(td) is not None:
-            load_variables(model_file)
+            load_state(model_file)
             logger.log('Loaded model from {}'.format(model_file))
             model_saved = True
         elif load_path is not None:
-            load_variables(load_path)
+            load_state(load_path)
             logger.log('Loaded model from {}'.format(load_path))
-
+        
 
         for t in range(total_timesteps):
             if callback is not None:
@@ -322,12 +320,12 @@ def learn(env,
                     if print_freq is not None:
                         logger.log("Saving model due to mean reward increase: {} -> {}".format(
                                    saved_mean_reward, mean_100ep_reward))
-                    save_variables(model_file)
+                    save_state(model_file)
                     model_saved = True
                     saved_mean_reward = mean_100ep_reward
         if model_saved:
             if print_freq is not None:
                 logger.log("Restored model with mean reward: {}".format(saved_mean_reward))
-            load_variables(model_file)
+            load_state(model_file)
 
     return act

baselines/deepq/defaults.py

@@ -16,8 +16,6 @@ def atari():
         dueling=True
     )
 
+def retro():
+    return atari()
 
-defaults = {
-    'atari': atari(),
-    'retro': atari()
-}

baselines/deepq/experiments/enjoy_cartpole.py

@@ -5,7 +5,7 @@ from baselines import deepq
 
 def main():
     env = gym.make("CartPole-v0")
-    act = deepq.learn(env, network='mlp', total_timesteps=0, load_path="cartpole_model.pkl")
+    act = deepq.load("cartpole_model.pkl")
 
     while True:
         obs, done = env.reset(), False

baselines/deepq/experiments/enjoy_mountaincar.py

@@ -1,17 +1,11 @@
 import gym
 
 from baselines import deepq
-from baselines.common import models
 
 
 def main():
     env = gym.make("MountainCar-v0")
-    act = deepq.learn(
-        env,
-        network=models.mlp(num_layers=1, num_hidden=64),
-        total_timesteps=0,
-        load_path='mountaincar_model.pkl'
-    )
+    act = deepq.load("mountaincar_model.pkl")
 
     while True:
         obs, done = env.reset(), False

baselines/deepq/experiments/enjoy_pong.py

@@ -5,21 +5,14 @@ from baselines import deepq
 def main():
     env = gym.make("PongNoFrameskip-v4")
     env = deepq.wrap_atari_dqn(env)
-    model = deepq.learn(
-        env,
-        "conv_only",
-        convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)],
-        hiddens=[256],
-        dueling=True,
-        total_timesteps=0
-    )
+    act = deepq.load("pong_model.pkl")
 
     while True:
         obs, done = env.reset(), False
         episode_rew = 0
         while not done:
             env.render()
-            obs, rew, done, _ = env.step(model(obs[None])[0])
+            obs, rew, done, _ = env.step(act(obs[None])[0])
             episode_rew += rew
         print("Episode reward", episode_rew)
 

baselines/deepq/experiments/enjoy_retro.py

@@ -0,0 +1,34 @@
+import argparse
+
+import numpy as np
+
+from baselines import deepq
+from baselines.common import retro_wrappers
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--env', help='environment ID', default='SuperMarioBros-Nes')
+    parser.add_argument('--gamestate', help='game state to load', default='Level1-1')
+    parser.add_argument('--model', help='model pickle file from ActWrapper.save', default='model.pkl')
+    args = parser.parse_args()
+
+    env = retro_wrappers.make_retro(game=args.env, state=args.gamestate, max_episode_steps=None)
+    env = retro_wrappers.wrap_deepmind_retro(env)
+    act = deepq.load(args.model)
+
+    while True:
+        obs, done = env.reset(), False
+        episode_rew = 0
+        while not done:
+            env.render()
+            action = act(obs[None])[0]
+            env_action = np.zeros(env.action_space.n)
+            env_action[action] = 1
+            obs, rew, done, _ = env.step(env_action)
+            episode_rew += rew
+        print('Episode reward', episode_rew)
+
+
+if __name__ == '__main__':
+    main()

baselines/deepq/experiments/run_atari.py

@@ -0,0 +1,54 @@
+from baselines import deepq
+from baselines.common import set_global_seeds
+from baselines import bench
+import argparse
+from baselines import logger
+from baselines.common.atari_wrappers import make_atari
+
+
+def main():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+    parser.add_argument('--env', help='environment ID', default='BreakoutNoFrameskip-v4')
+    parser.add_argument('--seed', help='RNG seed', type=int, default=0)
+    parser.add_argument('--prioritized', type=int, default=1)
+    parser.add_argument('--prioritized-replay-alpha', type=float, default=0.6)
+    parser.add_argument('--dueling', type=int, default=1)
+    parser.add_argument('--num-timesteps', type=int, default=int(10e6))
+    parser.add_argument('--checkpoint-freq', type=int, default=10000)
+    parser.add_argument('--checkpoint-path', type=str, default=None)
+
+    args = parser.parse_args()
+    logger.configure()
+    set_global_seeds(args.seed)
+    env = make_atari(args.env)
+    env = bench.Monitor(env, logger.get_dir())
+    env = deepq.wrap_atari_dqn(env)
+    model = deepq.models.cnn_to_mlp(
+        convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)],
+        hiddens=[256],
+        dueling=bool(args.dueling),
+    )
+
+    deepq.learn(
+        env,
+        q_func=model,
+        lr=1e-4,
+        max_timesteps=args.num_timesteps,
+        buffer_size=10000,
+        exploration_fraction=0.1,
+        exploration_final_eps=0.01,
+        train_freq=4,
+        learning_starts=10000,
+        target_network_update_freq=1000,
+        gamma=0.99,
+        prioritized_replay=bool(args.prioritized),
+        prioritized_replay_alpha=args.prioritized_replay_alpha,
+        checkpoint_freq=args.checkpoint_freq,
+        checkpoint_path=args.checkpoint_path,
+    )
+
+    env.close()
+
+
+if __name__ == '__main__':
+    main()

baselines/deepq/experiments/run_retro.py

@@ -0,0 +1,49 @@
+import argparse
+
+from baselines import deepq
+from baselines.common import set_global_seeds
+from baselines import bench
+from baselines import logger
+from baselines.common import retro_wrappers
+import retro
+
+
+def main():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+    parser.add_argument('--env', help='environment ID', default='SuperMarioBros-Nes')
+    parser.add_argument('--gamestate', help='game state to load', default='Level1-1')
+    parser.add_argument('--seed', help='seed', type=int, default=0)
+    parser.add_argument('--num-timesteps', type=int, default=int(10e6))
+    args = parser.parse_args()
+    logger.configure()
+    set_global_seeds(args.seed)
+    env = retro_wrappers.make_retro(game=args.env, state=args.gamestate, max_episode_steps=10000, use_restricted_actions=retro.Actions.DISCRETE)
+    env.seed(args.seed)
+    env = bench.Monitor(env, logger.get_dir())
+    env = retro_wrappers.wrap_deepmind_retro(env)
+
+    model = deepq.models.cnn_to_mlp(
+        convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)],
+        hiddens=[256],
+        dueling=True
+    )
+    act = deepq.learn(
+        env,
+        q_func=model,
+        lr=1e-4,
+        max_timesteps=args.num_timesteps,
+        buffer_size=10000,
+        exploration_fraction=0.1,
+        exploration_final_eps=0.01,
+        train_freq=4,
+        learning_starts=10000,
+        target_network_update_freq=1000,
+        gamma=0.99,
+        prioritized_replay=True
+    )
+    act.save()
+    env.close()
+
+
+if __name__ == '__main__':
+    main()

baselines/deepq/experiments/train_cartpole.py

@@ -11,11 +11,12 @@ def callback(lcl, _glb):
 
 def main():
     env = gym.make("CartPole-v0")
+    model = deepq.models.mlp([64])
     act = deepq.learn(
         env,
-        network='mlp',
+        q_func=model,
         lr=1e-3,
-        total_timesteps=100000,
+        max_timesteps=100000,
         buffer_size=50000,
         exploration_fraction=0.1,
         exploration_final_eps=0.02,

baselines/deepq/experiments/train_mountaincar.py

@@ -1,17 +1,17 @@
 import gym
 
 from baselines import deepq
-from baselines.common import models
 
 
 def main():
     env = gym.make("MountainCar-v0")
     # Enabling layer_norm here is import for parameter space noise!
+    model = deepq.models.mlp([64], layer_norm=True)
     act = deepq.learn(
         env,
-        network=models.mlp(num_hidden=64, num_layers=1),
+        q_func=model,
         lr=1e-3,
-        total_timesteps=100000,
+        max_timesteps=100000,
         buffer_size=50000,
         exploration_fraction=0.1,
         exploration_final_eps=0.1,

baselines/deepq/experiments/train_pong.py

@@ -1,34 +0,0 @@
-from baselines import deepq
-from baselines import bench
-from baselines import logger
-from baselines.common.atari_wrappers import make_atari
-
-
-def main():
-    logger.configure()
-    env = make_atari('PongNoFrameskip-v4')
-    env = bench.Monitor(env, logger.get_dir())
-    env = deepq.wrap_atari_dqn(env)
-
-    model = deepq.learn(
-        env,
-        "conv_only",
-        convs=[(32, 8, 4), (64, 4, 2), (64, 3, 1)],
-        hiddens=[256],
-        dueling=True,
-        lr=1e-4,
-        total_timesteps=int(1e7),
-        buffer_size=10000,
-        exploration_fraction=0.1,
-        exploration_final_eps=0.01,
-        train_freq=4,
-        learning_starts=10000,
-        target_network_update_freq=1000,
-        gamma=0.99,
-    )
-
-    model.save('pong_model.pkl')
-    env.close()
-
-if __name__ == '__main__':
-    main()

baselines/deepq/models.py

@@ -94,16 +94,11 @@ def cnn_to_mlp(convs, hiddens, dueling=False, layer_norm=False):
 def build_q_func(network, hiddens=[256], dueling=True, layer_norm=False, **network_kwargs):
     if isinstance(network, str):
         from baselines.common.models import get_network_builder
-        network = get_network_builder(network)(**network_kwargs)
-
+        network = get_network_builder(network)(**network_kwargs)   
+        
     def q_func_builder(input_placeholder, num_actions, scope, reuse=False):
         with tf.variable_scope(scope, reuse=reuse):
-            latent = network(input_placeholder)
-            if isinstance(latent, tuple):
-                if latent[1] is not None:
-                    raise NotImplementedError("DQN is not compatible with recurrent policies yet")
-                latent = latent[0]
-
+            latent, _ = network(input_placeholder)
             latent = layers.flatten(latent)
 
             with tf.variable_scope("action_value"):
@@ -130,5 +125,5 @@ def build_q_func(network, hiddens=[256], dueling=True, layer_norm=False, **netwo
             else:
                 q_out = action_scores
             return q_out
-
+            
     return q_func_builder

baselines/deepq/replay_buffer.py

@@ -106,10 +106,9 @@ class PrioritizedReplayBuffer(ReplayBuffer):
 
     def _sample_proportional(self, batch_size):
         res = []
-        p_total = self._it_sum.sum(0, len(self._storage) - 1)
-        every_range_len = p_total / batch_size
-        for i in range(batch_size):
-            mass = random.random() * every_range_len + i * every_range_len
+        for _ in range(batch_size):
+            # TODO(szymon): should we ensure no repeats?
+            mass = random.random() * self._it_sum.sum(0, len(self._storage) - 1)
             idx = self._it_sum.find_prefixsum_idx(mass)
             res.append(idx)
         return res

baselines/deepq/utils.py

@@ -1,6 +1,8 @@
 from baselines.common.input import observation_input
 from baselines.common.tf_util import adjust_shape
 
+import tensorflow as tf
+
 # ================================================================
 # Placeholders
 # ================================================================
@@ -38,16 +40,39 @@ class PlaceholderTfInput(TfInput):
         return {self._placeholder: adjust_shape(self._placeholder, data)}
 
 
-class ObservationInput(PlaceholderTfInput):
-    def __init__(self, observation_space, name=None):
-        """Creates an input placeholder tailored to a specific observation space
+class Uint8Input(PlaceholderTfInput):
+    def __init__(self, shape, name=None):
+        """Takes input in uint8 format which is cast to float32 and divided by 255
+        before passing it to the model.
+
+        On GPU this ensures lower data transfer times.
 
         Parameters
         ----------
-
-        observation_space:
-                observation space of the environment. Should be one of the gym.spaces types
+        shape: [int]
+            shape of the tensor.
         name: str
+            name of the underlying placeholder
+        """
+
+        super().__init__(tf.placeholder(tf.uint8, [None] + list(shape), name=name))
+        self._shape = shape
+        self._output = tf.cast(super().get(), tf.float32) / 255.0
+
+    def get(self):
+        return self._output
+
+
+class ObservationInput(PlaceholderTfInput):
+    def __init__(self, observation_space, name=None):
+        """Creates an input placeholder tailored to a specific observation space
+        
+        Parameters
+        ----------
+
+        observation_space: 
+                observation space of the environment. Should be one of the gym.spaces types
+        name: str 
                 tensorflow name of the underlying placeholder
         """
         inpt, self.processed_inpt = observation_input(observation_space, name=name)
@@ -55,5 +80,5 @@ class ObservationInput(PlaceholderTfInput):
 
     def get(self):
         return self.processed_inpt
-
-
+    
+    

baselines/her/README.md

@@ -30,51 +30,3 @@ python -m baselines.her.experiment.train --num_cpu 19
 This will require a machine with sufficient amount of physical CPU cores. In our experiments,
 we used [Azure's D15v2 instances](https://docs.microsoft.com/en-us/azure/virtual-machines/linux/sizes),
 which have 20 physical cores. We only scheduled the experiment on 19 of those to leave some head-room on the system.
-
-
-## Hindsight Experience Replay with Demonstrations
-Using pre-recorded demonstrations to Overcome the exploration problem in HER based Reinforcement learning.
-For details, please read the [paper](https://arxiv.org/pdf/1709.10089.pdf).
-
-### Getting started
-The first step is to generate the demonstration dataset. This can be done in two ways, either by using a VR system to manipulate the arm using physical VR trackers or the simpler way is to write a script to carry out the respective task. Now some tasks can be complex and thus it would be difficult to write a hardcoded script for that task (eg. Fetch Push), but here our focus is on providing an algorithm that helps the agent to learn from demonstrations, and not on the demonstration generation paradigm itself. Thus the data collection part is left to the reader's choice.
-
-We provide a script for the Fetch Pick and Place task, to generate demonstrations for the Pick and Place task execute:
-```bash
-python experiment/data_generation/fetch_data_generation.py
-```
-This outputs ```data_fetch_random_100.npz``` file which is our data file.
-
-#### Configuration
-The provided configuration is for training an agent with HER without demonstrations, we need to change a few paramters for the HER algorithm to learn through demonstrations, to do that, set:
-
-* bc_loss: 1 - whether or not to use the behavior cloning loss as an auxilliary loss
-* q_filter: 1 - whether or not a Q value filter should be used on the Actor outputs
-* num_demo: 100 - number of expert demo episodes
-* demo_batch_size: 128 - number of samples to be used from the demonstrations buffer, per mpi thread
-* prm_loss_weight: 0.001 - Weight corresponding to the primary loss
-* aux_loss_weight:  0.0078 - Weight corresponding to the auxilliary loss also called the cloning loss
-
-Apart from these changes the reported results also have the following configurational changes:
-
-* n_cycles: 20 - per epoch
-* batch_size: 1024 - per mpi thread, total batch size
-* random_eps: 0.1  - percentage of time a random action is taken
-* noise_eps: 0.1  - std of gaussian noise added to not-completely-random actions
-
-Now training an agent with pre-recorded demonstrations:
-```bash
-python -m baselines.her.experiment.train --env=FetchPickAndPlace-v0 --n_epochs=1000 --demo_file=/Path/to/demo_file.npz --num_cpu=1
-```
-
-This will train a DDPG+HER agent on the `FetchPickAndPlace` environment by using previously generated demonstration data.
-To inspect what the agent has learned, use the play script as described above.
-
-### Results
-Training with demonstrations helps overcome the exploration problem and achieves a faster and better convergence. The following graphs contrast the difference between training with and without demonstration data, We report the mean Q values vs Epoch and the Success Rate vs Epoch:
-
-
-<div class="imgcap" align="middle">
-<center><img src="../../data/fetchPickAndPlaceContrast.png"></center>
-<div class="thecap" align="middle"><b>Training results for Fetch Pick and Place task constrasting between training with and without demonstration data.</b></div>
-</div>

baselines/her/ddpg.py

@@ -6,7 +6,7 @@ from tensorflow.contrib.staging import StagingArea
 
 from baselines import logger
 from baselines.her.util import (
-    import_function, store_args, flatten_grads, transitions_in_episode_batch, convert_episode_to_batch_major)
+    import_function, store_args, flatten_grads, transitions_in_episode_batch)
 from baselines.her.normalizer import Normalizer
 from baselines.her.replay_buffer import ReplayBuffer
 from baselines.common.mpi_adam import MpiAdam
@@ -16,17 +16,13 @@ def dims_to_shapes(input_dims):
     return {key: tuple([val]) if val > 0 else tuple() for key, val in input_dims.items()}
 
 
-global demoBuffer #buffer for demonstrations
-
 class DDPG(object):
     @store_args
     def __init__(self, input_dims, buffer_size, hidden, layers, network_class, polyak, batch_size,
                  Q_lr, pi_lr, norm_eps, norm_clip, max_u, action_l2, clip_obs, scope, T,
                  rollout_batch_size, subtract_goals, relative_goals, clip_pos_returns, clip_return,
-                 bc_loss, q_filter, num_demo, demo_batch_size, prm_loss_weight, aux_loss_weight,
                  sample_transitions, gamma, reuse=False, **kwargs):
         """Implementation of DDPG that is used in combination with Hindsight Experience Replay (HER).
-            Added functionality to use demonstrations for training to Overcome exploration problem.
 
         Args:
             input_dims (dict of ints): dimensions for the observation (o), the goal (g), and the
@@ -54,12 +50,6 @@ class DDPG(object):
             sample_transitions (function) function that samples from the replay buffer
             gamma (float): gamma used for Q learning updates
             reuse (boolean): whether or not the networks should be reused
-            bc_loss: whether or not the behavior cloning loss should be used as an auxilliary loss
-            q_filter: whether or not a filter on the q value update should be used when training with demonstartions
-            num_demo: Number of episodes in to be used in the demonstration buffer
-            demo_batch_size: number of samples to be used from the demonstrations buffer, per mpi thread
-            prm_loss_weight: Weight corresponding to the primary loss
-            aux_loss_weight: Weight corresponding to the auxilliary loss also called the cloning loss
         """
         if self.clip_return is None:
             self.clip_return = np.inf
@@ -102,9 +92,6 @@ class DDPG(object):
         buffer_size = (self.buffer_size // self.rollout_batch_size) * self.rollout_batch_size
         self.buffer = ReplayBuffer(buffer_shapes, buffer_size, self.T, self.sample_transitions)
 
-        global demoBuffer
-        demoBuffer = ReplayBuffer(buffer_shapes, buffer_size, self.T, self.sample_transitions) #initialize the demo buffer; in the same way as the primary data buffer
-
     def _random_action(self, n):
         return np.random.uniform(low=-self.max_u, high=self.max_u, size=(n, self.dimu))
 
@@ -151,57 +138,6 @@ class DDPG(object):
         else:
             return ret
 
-    def initDemoBuffer(self, demoDataFile, update_stats=True): #function that initializes the demo buffer
-
-        demoData = np.load(demoDataFile) #load the demonstration data from data file
-        info_keys = [key.replace('info_', '') for key in self.input_dims.keys() if key.startswith('info_')]
-        info_values = [np.empty((self.T, 1, self.input_dims['info_' + key]), np.float32) for key in info_keys]
-
-        for epsd in range(self.num_demo): # we initialize the whole demo buffer at the start of the training
-            obs, acts, goals, achieved_goals = [], [] ,[] ,[]
-            i = 0
-            for transition in range(self.T):
-                obs.append([demoData['obs'][epsd ][transition].get('observation')])
-                acts.append([demoData['acs'][epsd][transition]])
-                goals.append([demoData['obs'][epsd][transition].get('desired_goal')])
-                achieved_goals.append([demoData['obs'][epsd][transition].get('achieved_goal')])
-                for idx, key in enumerate(info_keys):
-                    info_values[idx][transition, i] = demoData['info'][epsd][transition][key]
-
-            obs.append([demoData['obs'][epsd][self.T].get('observation')])
-            achieved_goals.append([demoData['obs'][epsd][self.T].get('achieved_goal')])
-
-            episode = dict(o=obs,
-                           u=acts,
-                           g=goals,
-                           ag=achieved_goals)
-            for key, value in zip(info_keys, info_values):
-                episode['info_{}'.format(key)] = value
-
-            episode = convert_episode_to_batch_major(episode)
-            global demoBuffer
-            demoBuffer.store_episode(episode) # create the observation dict and append them into the demonstration buffer
-
-            print("Demo buffer size currently ", demoBuffer.get_current_size()) #print out the demonstration buffer size
-
-            if update_stats:
-                # add transitions to normalizer to normalize the demo data as well
-                episode['o_2'] = episode['o'][:, 1:, :]
-                episode['ag_2'] = episode['ag'][:, 1:, :]
-                num_normalizing_transitions = transitions_in_episode_batch(episode)
-                transitions = self.sample_transitions(episode, num_normalizing_transitions)
-
-                o, o_2, g, ag = transitions['o'], transitions['o_2'], transitions['g'], transitions['ag']
-                transitions['o'], transitions['g'] = self._preprocess_og(o, ag, g)
-                # No need to preprocess the o_2 and g_2 since this is only used for stats
-
-                self.o_stats.update(transitions['o'])
-                self.g_stats.update(transitions['g'])
-
-                self.o_stats.recompute_stats()
-                self.g_stats.recompute_stats()
-            episode.clear()
-
     def store_episode(self, episode_batch, update_stats=True):
         """
         episode_batch: array of batch_size x (T or T+1) x dim_key
@@ -249,18 +185,7 @@ class DDPG(object):
         self.pi_adam.update(pi_grad, self.pi_lr)
 
     def sample_batch(self):
-        if self.bc_loss: #use demonstration buffer to sample as well if bc_loss flag is set TRUE
-            transitions = self.buffer.sample(self.batch_size - self.demo_batch_size)
-            global demoBuffer
-            transitionsDemo = demoBuffer.sample(self.demo_batch_size) #sample from the demo buffer
-            for k, values in transitionsDemo.items():
-                rolloutV = transitions[k].tolist()
-                for v in values:
-                    rolloutV.append(v.tolist())
-                transitions[k] = np.array(rolloutV)
-        else:
-            transitions = self.buffer.sample(self.batch_size) #otherwise only sample from primary buffer
-
+        transitions = self.buffer.sample(self.batch_size)
         o, o_2, g = transitions['o'], transitions['o_2'], transitions['g']
         ag, ag_2 = transitions['ag'], transitions['ag_2']
         transitions['o'], transitions['g'] = self._preprocess_og(o, ag, g)
@@ -323,9 +248,6 @@ class DDPG(object):
                                 for i, key in enumerate(self.stage_shapes.keys())])
         batch_tf['r'] = tf.reshape(batch_tf['r'], [-1, 1])
 
-        #choose only the demo buffer samples
-        mask = np.concatenate((np.zeros(self.batch_size - self.demo_batch_size), np.ones(self.demo_batch_size)), axis = 0)
-
         # networks
         with tf.variable_scope('main') as vs:
             if reuse:
@@ -348,25 +270,6 @@ class DDPG(object):
         clip_range = (-self.clip_return, 0. if self.clip_pos_returns else np.inf)
         target_tf = tf.clip_by_value(batch_tf['r'] + self.gamma * target_Q_pi_tf, *clip_range)
         self.Q_loss_tf = tf.reduce_mean(tf.square(tf.stop_gradient(target_tf) - self.main.Q_tf))
-
-        if self.bc_loss ==1 and self.q_filter == 1 : # train with demonstrations and use bc_loss and q_filter both
-            maskMain = tf.reshape(tf.boolean_mask(self.main.Q_tf > self.main.Q_pi_tf, mask), [-1]) #where is the demonstrator action better than actor action according to the critic? choose those samples only
-            #define the cloning loss on the actor's actions only on the samples which adhere to the above masks
-            self.cloning_loss_tf = tf.reduce_sum(tf.square(tf.boolean_mask(tf.boolean_mask((self.main.pi_tf), mask), maskMain, axis=0) - tf.boolean_mask(tf.boolean_mask((batch_tf['u']), mask), maskMain, axis=0)))
-            self.pi_loss_tf = -self.prm_loss_weight * tf.reduce_mean(self.main.Q_pi_tf) #primary loss scaled by it's respective weight prm_loss_weight
-            self.pi_loss_tf += self.prm_loss_weight * self.action_l2 * tf.reduce_mean(tf.square(self.main.pi_tf / self.max_u)) #L2 loss on action values scaled by the same weight prm_loss_weight
-            self.pi_loss_tf += self.aux_loss_weight * self.cloning_loss_tf #adding the cloning loss to the actor loss as an auxilliary loss scaled by its weight aux_loss_weight
-
-        elif self.bc_loss == 1 and self.q_filter == 0: # train with demonstrations without q_filter
-            self.cloning_loss_tf = tf.reduce_sum(tf.square(tf.boolean_mask((self.main.pi_tf), mask) - tf.boolean_mask((batch_tf['u']), mask)))
-            self.pi_loss_tf = -self.prm_loss_weight * tf.reduce_mean(self.main.Q_pi_tf)
-            self.pi_loss_tf += self.prm_loss_weight * self.action_l2 * tf.reduce_mean(tf.square(self.main.pi_tf / self.max_u))
-            self.pi_loss_tf += self.aux_loss_weight * self.cloning_loss_tf
-
-        else: #If  not training with demonstrations
-            self.pi_loss_tf = -tf.reduce_mean(self.main.Q_pi_tf)
-            self.pi_loss_tf += self.action_l2 * tf.reduce_mean(tf.square(self.main.pi_tf / self.max_u))
-
         self.pi_loss_tf = -tf.reduce_mean(self.main.Q_pi_tf)
         self.pi_loss_tf += self.action_l2 * tf.reduce_mean(tf.square(self.main.pi_tf / self.max_u))
         Q_grads_tf = tf.gradients(self.Q_loss_tf, self._vars('main/Q'))

baselines/her/experiment/config.py

@@ -44,13 +44,6 @@ DEFAULT_PARAMS = {
     # normalization
     'norm_eps': 0.01,  # epsilon used for observation normalization
     'norm_clip': 5,  # normalized observations are cropped to this values
-
-    'bc_loss': 0, # whether or not to use the behavior cloning loss as an auxilliary loss
-    'q_filter': 0, # whether or not a Q value filter should be used on the Actor outputs
-    'num_demo': 100, # number of expert demo episodes
-    'demo_batch_size': 128, #number of samples to be used from the demonstrations buffer, per mpi thread 128/1024 or 32/256
-    'prm_loss_weight': 0.001, #Weight corresponding to the primary loss
-    'aux_loss_weight':  0.0078, #Weight corresponding to the auxilliary loss also called the cloning loss
 }
 
 
@@ -152,12 +145,6 @@ def configure_ddpg(dims, params, reuse=False, use_mpi=True, clip_return=True):
                         'subtract_goals': simple_goal_subtract,
                         'sample_transitions': sample_her_transitions,
                         'gamma': gamma,
-                        'bc_loss': params['bc_loss'],
-                        'q_filter': params['q_filter'],
-                        'num_demo': params['num_demo'],
-                        'demo_batch_size': params['demo_batch_size'],
-                        'prm_loss_weight': params['prm_loss_weight'],
-                        'aux_loss_weight': params['aux_loss_weight'],
                         })
     ddpg_params['info'] = {
         'env_name': params['env_name'],

baselines/her/experiment/data_generation/fetch_data_generation.py

@@ -1,149 +0,0 @@
-import gym
-import time
-import random
-import numpy as np
-import rospy
-import roslaunch
-
-from random import randint
-from std_srvs.srv import Empty
-from sensor_msgs.msg import JointState
-from geometry_msgs.msg import PoseStamped
-from geometry_msgs.msg import Pose
-from std_msgs.msg import Float64
-from controller_manager_msgs.srv import SwitchController
-from gym.utils import seeding
-
-
-"""Data generation for the case of a single block pick and place in Fetch Env"""
-
-actions = []
-observations = []
-infos = []
-
-def main():
-    env = gym.make('FetchPickAndPlace-v0')
-    numItr = 100
-    initStateSpace = "random"
-    env.reset()
-    print("Reset!")
-    while len(actions) < numItr:
-        obs = env.reset()
-        print("ITERATION NUMBER ", len(actions))
-        goToGoal(env, obs)
-        
-
-    fileName = "data_fetch"
-    fileName += "_" + initStateSpace
-    fileName += "_" + str(numItr)
-    fileName += ".npz"
-    
-    np.savez_compressed(fileName, acs=actions, obs=observations, info=infos) # save the file
-
-def goToGoal(env, lastObs):
-
-    goal = lastObs['desired_goal']
-    objectPos = lastObs['observation'][3:6]
-    gripperPos = lastObs['observation'][:3]
-    gripperState = lastObs['observation'][9:11]
-    object_rel_pos = lastObs['observation'][6:9]
-    episodeAcs = []
-    episodeObs = []
-    episodeInfo = []
-
-    object_oriented_goal = object_rel_pos.copy()
-    object_oriented_goal[2] += 0.03 # first make the gripper go slightly above the object
-    
-    timeStep = 0 #count the total number of timesteps
-    episodeObs.append(lastObs)
-
-    while np.linalg.norm(object_oriented_goal) >= 0.005 and timeStep <= env._max_episode_steps:
-        env.render()
-        action = [0, 0, 0, 0]
-        object_oriented_goal = object_rel_pos.copy()
-        object_oriented_goal[2] += 0.03
-
-        for i in range(len(object_oriented_goal)):
-            action[i] = object_oriented_goal[i]*6
-
-        action[len(action)-1] = 0.05 #open
-
-        obsDataNew, reward, done, info = env.step(action)
-        timeStep += 1
-
-        episodeAcs.append(action)
-        episodeInfo.append(info)
-        episodeObs.append(obsDataNew)
-
-        objectPos = obsDataNew['observation'][3:6]
-        gripperPos = obsDataNew['observation'][:3]
-        gripperState = obsDataNew['observation'][9:11]
-        object_rel_pos = obsDataNew['observation'][6:9]
-
-    while np.linalg.norm(object_rel_pos) >= 0.005 and timeStep <= env._max_episode_steps :
-        env.render()
-        action = [0, 0, 0, 0]
-        for i in range(len(object_rel_pos)):
-            action[i] = object_rel_pos[i]*6
-
-        action[len(action)-1] = -0.005
-
-        obsDataNew, reward, done, info = env.step(action)
-        timeStep += 1
-
-        episodeAcs.append(action)
-        episodeInfo.append(info)
-        episodeObs.append(obsDataNew)
-
-        objectPos = obsDataNew['observation'][3:6]
-        gripperPos = obsDataNew['observation'][:3]
-        gripperState = obsDataNew['observation'][9:11]
-        object_rel_pos = obsDataNew['observation'][6:9]
-
-
-    while np.linalg.norm(goal - objectPos) >= 0.01 and timeStep <= env._max_episode_steps :
-        env.render()
-        action = [0, 0, 0, 0]
-        for i in range(len(goal - objectPos)):
-            action[i] = (goal - objectPos)[i]*6
-
-        action[len(action)-1] = -0.005
-
-        obsDataNew, reward, done, info = env.step(action)
-        timeStep += 1
-
-        episodeAcs.append(action)
-        episodeInfo.append(info)
-        episodeObs.append(obsDataNew)
-
-        objectPos = obsDataNew['observation'][3:6]
-        gripperPos = obsDataNew['observation'][:3]
-        gripperState = obsDataNew['observation'][9:11]
-        object_rel_pos = obsDataNew['observation'][6:9]
-
-    while True: #limit the number of timesteps in the episode to a fixed duration
-        env.render()
-        action = [0, 0, 0, 0]
-        action[len(action)-1] = -0.005 # keep the gripper closed
-
-        obsDataNew, reward, done, info = env.step(action)
-        timeStep += 1
-
-        episodeAcs.append(action)
-        episodeInfo.append(info)
-        episodeObs.append(obsDataNew)
-
-        objectPos = obsDataNew['observation'][3:6]
-        gripperPos = obsDataNew['observation'][:3]
-        gripperState = obsDataNew['observation'][9:11]
-        object_rel_pos = obsDataNew['observation'][6:9]
-
-        if timeStep >= env._max_episode_steps: break
-
-    actions.append(episodeAcs)
-    observations.append(episodeObs)
-    infos.append(episodeInfo)
-
-
-if __name__ == "__main__":
-    main()

baselines/her/experiment/play.py

@@ -41,7 +41,7 @@ def main(policy_file, seed, n_test_rollouts, render):
 
     for name in ['T', 'gamma', 'noise_eps', 'random_eps']:
         eval_params[name] = params[name]
-
+    
     evaluator = RolloutWorker(params['make_env'], policy, dims, logger, **eval_params)
     evaluator.seed(seed)
 

baselines/her/experiment/plot.py

@@ -37,12 +37,12 @@ def load_results(file):
 
 def pad(xs, value=np.nan):
     maxlen = np.max([len(x) for x in xs])
-
+    
     padded_xs = []
     for x in xs:
         if x.shape[0] >= maxlen:
             padded_xs.append(x)
-
+    
         padding = np.ones((maxlen - x.shape[0],) + x.shape[1:]) * value
         x_padded = np.concatenate([x, padding], axis=0)
         assert x_padded.shape[1:] == x.shape[1:]

baselines/her/experiment/train.py

@@ -26,7 +26,7 @@ def mpi_average(value):
 
 def train(policy, rollout_worker, evaluator,
           n_epochs, n_test_rollouts, n_cycles, n_batches, policy_save_interval,
-          save_policies, demo_file, **kwargs):
+          save_policies, **kwargs):
     rank = MPI.COMM_WORLD.Get_rank()
 
     latest_policy_path = os.path.join(logger.get_dir(), 'policy_latest.pkl')
@@ -35,8 +35,6 @@ def train(policy, rollout_worker, evaluator,
 
     logger.info("Training...")
     best_success_rate = -1
-
-    if policy.bc_loss == 1: policy.initDemoBuffer(demo_file) #initialize demo buffer if training with demonstrations
     for epoch in range(n_epochs):
         # train
         rollout_worker.clear_history()
@@ -86,7 +84,7 @@ def train(policy, rollout_worker, evaluator,
 
 def launch(
     env, logdir, n_epochs, num_cpu, seed, replay_strategy, policy_save_interval, clip_return,
-    demo_file, override_params={}, save_policies=True
+    override_params={}, save_policies=True
 ):
     # Fork for multi-CPU MPI implementation.
     if num_cpu > 1:
@@ -173,7 +171,7 @@ def launch(
         logdir=logdir, policy=policy, rollout_worker=rollout_worker,
         evaluator=evaluator, n_epochs=n_epochs, n_test_rollouts=params['n_test_rollouts'],
         n_cycles=params['n_cycles'], n_batches=params['n_batches'],
-        policy_save_interval=policy_save_interval, save_policies=save_policies, demo_file=demo_file)
+        policy_save_interval=policy_save_interval, save_policies=save_policies)
 
 
 @click.command()
@@ -185,7 +183,6 @@ def launch(
 @click.option('--policy_save_interval', type=int, default=5, help='the interval with which policy pickles are saved. If set to 0, only the best and latest policy will be pickled.')
 @click.option('--replay_strategy', type=click.Choice(['future', 'none']), default='future', help='the HER replay strategy to be used. "future" uses HER, "none" disables HER.')
 @click.option('--clip_return', type=int, default=1, help='whether or not returns should be clipped')
-@click.option('--demo_file', type=str, default = 'PATH/TO/DEMO/DATA/FILE.npz', help='demo data file path')
 def main(**kwargs):
     launch(**kwargs)
 

baselines/her/rollout.py

@@ -116,7 +116,7 @@ class RolloutWorker:
                     return self.generate_rollouts()
 
             if np.isnan(o_new).any():
-                self.logger.warn('NaN caught during rollout generation. Trying again...')
+                self.logger.warning('NaN caught during rollout generation. Trying again...')
                 self.reset_all_rollouts()
                 return self.generate_rollouts()
 

baselines/logger.py

@@ -106,8 +106,7 @@ class CSVOutputFormat(KVWriter):
 
     def writekvs(self, kvs):
         # Add our current row to the history
-        extra_keys = list(kvs.keys() - self.keys)
-        extra_keys.sort()
+        extra_keys = kvs.keys() - self.keys
         if extra_keys:
             self.keys.extend(extra_keys)
             self.file.seek(0)
@@ -345,6 +344,8 @@ class Logger(object):
             if isinstance(fmt, SeqWriter):
                 fmt.writeseq(map(str, args))
 
+Logger.DEFAULT = Logger.CURRENT = Logger(dir=None, output_formats=[HumanOutputFormat(sys.stdout)])
+
 def configure(dir=None, format_strs=None):
     if dir is None:
         dir = os.getenv('OPENAI_LOGDIR')
@@ -355,12 +356,8 @@ def configure(dir=None, format_strs=None):
     os.makedirs(dir, exist_ok=True)
 
     log_suffix = ''
-    rank = 0
-    # check environment variables here instead of importing mpi4py
-    # to avoid calling MPI_Init() when this module is imported
-    for varname in ['PMI_RANK', 'OMPI_COMM_WORLD_RANK']:
-        if varname in os.environ:
-            rank = int(os.environ[varname])
+    from mpi4py import MPI
+    rank = MPI.COMM_WORLD.Get_rank()
     if rank > 0:
         log_suffix = "-rank%03i" % rank
 
@@ -375,14 +372,6 @@ def configure(dir=None, format_strs=None):
     Logger.CURRENT = Logger(dir=dir, output_formats=output_formats)
     log('Logging to %s'%dir)
 
-def _configure_default_logger():
-    format_strs = None
-    # keep the old default of only writing to stdout
-    if 'OPENAI_LOG_FORMAT' not in os.environ:
-        format_strs = ['stdout']
-    configure(format_strs=format_strs)
-    Logger.DEFAULT = Logger.CURRENT
-
 def reset():
     if Logger.CURRENT is not Logger.DEFAULT:
         Logger.CURRENT.close()
@@ -482,8 +471,5 @@ def read_tb(path):
             data[step-1, colidx] = value
     return pandas.DataFrame(data, columns=tags)
 
-# configure the default logger on import
-_configure_default_logger()
-
 if __name__ == "__main__":
     _demo()

baselines/ppo1/run_humanoid.py

@@ -23,17 +23,17 @@ def train(num_timesteps, seed, model_path=None):
             max_timesteps=num_timesteps,
             timesteps_per_actorbatch=2048,
             clip_param=0.2, entcoeff=0.0,
-            optim_epochs=10,
-            optim_stepsize=3e-4,
-            optim_batchsize=64,
-            gamma=0.99,
+            optim_epochs=10, 
+            optim_stepsize=3e-4, 
+            optim_batchsize=64, 
+            gamma=0.99, 
             lam=0.95,
             schedule='linear',
         )
     env.close()
     if model_path:
         U.save_state(model_path)
-
+        
     return pi
 
 class RewScale(gym.RewardWrapper):
@@ -48,28 +48,28 @@ def main():
     parser = mujoco_arg_parser()
     parser.add_argument('--model-path', default=os.path.join(logger.get_dir(), 'humanoid_policy'))
     parser.set_defaults(num_timesteps=int(2e7))
-
+   
     args = parser.parse_args()
-
+    
     if not args.play:
         # train the model
         train(num_timesteps=args.num_timesteps, seed=args.seed, model_path=args.model_path)
-    else:
+    else:       
         # construct the model object, load pre-trained model and render
         pi = train(num_timesteps=1, seed=args.seed)
         U.load_state(args.model_path)
         env = make_mujoco_env('Humanoid-v2', seed=0)
 
-        ob = env.reset()
+        ob = env.reset()        
         while True:
             action = pi.act(stochastic=False, ob=ob)[0]
             ob, _, done, _ =  env.step(action)
             env.render()
             if done:
                 ob = env.reset()
-
-
-
+        
+        
+    
 
 if __name__ == '__main__':
     main()

baselines/ppo2/README.md

@@ -2,7 +2,5 @@
 
 - Original paper: https://arxiv.org/abs/1707.06347
 - Baselines blog post: https://blog.openai.com/openai-baselines-ppo/
-
-- `python -m baselines.run --alg=ppo2 --env=PongNoFrameskip-v4` runs the algorithm for 40M frames = 10M timesteps on an Atari Pong. See help (`-h`) for more options.
-- `python -m baselines.run --alg=ppo2 --env=Ant-v2 --num_timesteps=1e6` runs the algorithm for 1M frames on a Mujoco Ant environment.
-- also refer to the repo-wide [README.md](../../README.md#training-models)
+- `python -m baselines.ppo2.run_atari` runs the algorithm for 40M frames = 10M timesteps on an Atari game. See help (`-h`) for more options.
+- `python -m baselines.ppo2.run_mujoco` runs the algorithm for 1M frames on a Mujoco environment.

baselines/ppo2/defaults.py

@@ -1,5 +1,5 @@
-defaults = {
-    'mujoco': dict(
+def mujoco():
+    return dict(
         nsteps=2048,
         nminibatches=32,
         lam=0.95,
@@ -10,13 +10,13 @@ defaults = {
         lr=lambda f: 3e-4 * f,
         cliprange=0.2,
         value_network='copy'
-    ),
+    )
 
-    'atari': dict(
+def atari():
+    return dict(
         nsteps=128, nminibatches=4,
         lam=0.95, gamma=0.99, noptepochs=4, log_interval=1,
         ent_coef=.01,
         lr=lambda f : f * 2.5e-4,
         cliprange=lambda f : f * 0.1,
     )
-}

baselines/ppo2/ppo2.py

@@ -4,7 +4,7 @@ import functools
 import numpy as np
 import os.path as osp
 import tensorflow as tf
-from baselines import logger, registry
+from baselines import logger
 from collections import deque
 from baselines.common import explained_variance, set_global_seeds
 from baselines.common.policies import build_policy
@@ -15,105 +15,52 @@ from baselines.common.mpi_adam_optimizer import MpiAdamOptimizer
 from mpi4py import MPI
 from baselines.common.tf_util import initialize
 from baselines.common.mpi_util import sync_from_root
-from baselines.ppo2.defaults import defaults
 
 class Model(object):
-    """
-    We use this object to :
-    __init__:
-    - Creates the step_model
-    - Creates the train_model
-
-    train():
-    - Make the training part (feedforward and retropropagation of gradients)
-
-    save/load():
-    - Save load the model
-    """
     def __init__(self, *, policy, ob_space, ac_space, nbatch_act, nbatch_train,
                 nsteps, ent_coef, vf_coef, max_grad_norm):
         sess = get_session()
 
         with tf.variable_scope('ppo2_model', reuse=tf.AUTO_REUSE):
-            # CREATE OUR TWO MODELS
-            # act_model that is used for sampling
             act_model = policy(nbatch_act, 1, sess)
-
-            # Train model for training
             train_model = policy(nbatch_train, nsteps, sess)
 
-        # CREATE THE PLACEHOLDERS
         A = train_model.pdtype.sample_placeholder([None])
         ADV = tf.placeholder(tf.float32, [None])
         R = tf.placeholder(tf.float32, [None])
-        # Keep track of old actor
         OLDNEGLOGPAC = tf.placeholder(tf.float32, [None])
-        # Keep track of old critic
         OLDVPRED = tf.placeholder(tf.float32, [None])
         LR = tf.placeholder(tf.float32, [])
-        # Cliprange
         CLIPRANGE = tf.placeholder(tf.float32, [])
 
         neglogpac = train_model.pd.neglogp(A)
-
-        # Calculate the entropy
-        # Entropy is used to improve exploration by limiting the premature convergence to suboptimal policy.
         entropy = tf.reduce_mean(train_model.pd.entropy())
 
-        # CALCULATE THE LOSS
-        # Total loss = Policy gradient loss - entropy * entropy coefficient + Value coefficient * value loss
-
-        # Clip the value to reduce variability during Critic training
-        # Get the predicted value
         vpred = train_model.vf
         vpredclipped = OLDVPRED + tf.clip_by_value(train_model.vf - OLDVPRED, - CLIPRANGE, CLIPRANGE)
-        # Unclipped value
         vf_losses1 = tf.square(vpred - R)
-        # Clipped value
         vf_losses2 = tf.square(vpredclipped - R)
-
         vf_loss = .5 * tf.reduce_mean(tf.maximum(vf_losses1, vf_losses2))
-
-        # Calculate ratio (pi current policy / pi old policy)
         ratio = tf.exp(OLDNEGLOGPAC - neglogpac)
-
-        # Defining Loss = - J is equivalent to max J
         pg_losses = -ADV * ratio
-
         pg_losses2 = -ADV * tf.clip_by_value(ratio, 1.0 - CLIPRANGE, 1.0 + CLIPRANGE)
-
-        # Final PG loss
         pg_loss = tf.reduce_mean(tf.maximum(pg_losses, pg_losses2))
         approxkl = .5 * tf.reduce_mean(tf.square(neglogpac - OLDNEGLOGPAC))
         clipfrac = tf.reduce_mean(tf.to_float(tf.greater(tf.abs(ratio - 1.0), CLIPRANGE)))
-
-        # Total loss
         loss = pg_loss - entropy * ent_coef + vf_loss * vf_coef
-
-        # UPDATE THE PARAMETERS USING LOSS
-        # 1. Get the model parameters
         params = tf.trainable_variables('ppo2_model')
-        # 2. Build our trainer
         trainer = MpiAdamOptimizer(MPI.COMM_WORLD, learning_rate=LR, epsilon=1e-5)
-        # 3. Calculate the gradients
         grads_and_var = trainer.compute_gradients(loss, params)
         grads, var = zip(*grads_and_var)
 
         if max_grad_norm is not None:
-            # Clip the gradients (normalize)
             grads, _grad_norm = tf.clip_by_global_norm(grads, max_grad_norm)
         grads_and_var = list(zip(grads, var))
-        # zip aggregate each gradient with parameters associated
-        # For instance zip(ABCD, xyza) => Ax, By, Cz, Da
 
         _train = trainer.apply_gradients(grads_and_var)
 
         def train(lr, cliprange, obs, returns, masks, actions, values, neglogpacs, states=None):
-            # Here we calculate advantage A(s,a) = R + yV(s') - V(s)
-            # Returns = R + yV(s')
             advs = returns - values
-
-            # Normalize the advantages
             advs = (advs - advs.mean()) / (advs.std() + 1e-8)
             td_map = {train_model.X:obs, A:actions, ADV:advs, R:returns, LR:lr,
                     CLIPRANGE:cliprange, OLDNEGLOGPAC:neglogpacs, OLDVPRED:values}
@@ -143,39 +90,23 @@ class Model(object):
         sync_from_root(sess, global_variables) #pylint: disable=E1101
 
 class Runner(AbstractEnvRunner):
-    """
-    We use this object to make a mini batch of experiences
-    __init__:
-    - Initialize the runner
 
-    run():
-    - Make a mini batch
-    """
     def __init__(self, *, env, model, nsteps, gamma, lam):
         super().__init__(env=env, model=model, nsteps=nsteps)
-        # Lambda used in GAE (General Advantage Estimation)
         self.lam = lam
-        # Discount rate
         self.gamma = gamma
 
     def run(self):
-        # Here, we init the lists that will contain the mb of experiences
         mb_obs, mb_rewards, mb_actions, mb_values, mb_dones, mb_neglogpacs = [],[],[],[],[],[]
         mb_states = self.states
         epinfos = []
-        # For n in range number of steps
         for _ in range(self.nsteps):
-            # Given observations, get action value and neglopacs
-            # We already have self.obs because Runner superclass run self.obs[:] = env.reset() on init
             actions, values, self.states, neglogpacs = self.model.step(self.obs, S=self.states, M=self.dones)
             mb_obs.append(self.obs.copy())
             mb_actions.append(actions)
             mb_values.append(values)
             mb_neglogpacs.append(neglogpacs)
             mb_dones.append(self.dones)
-
-            # Take actions in env and look the results
-            # Infos contains a ton of useful informations
             self.obs[:], rewards, self.dones, infos = self.env.step(actions)
             for info in infos:
                 maybeepinfo = info.get('episode')
@@ -189,8 +120,7 @@ class Runner(AbstractEnvRunner):
         mb_neglogpacs = np.asarray(mb_neglogpacs, dtype=np.float32)
         mb_dones = np.asarray(mb_dones, dtype=np.bool)
         last_values = self.model.value(self.obs, S=self.states, M=self.dones)
-
-        # discount/bootstrap off value fn
+        #discount/bootstrap off value fn
         mb_returns = np.zeros_like(mb_rewards)
         mb_advs = np.zeros_like(mb_rewards)
         lastgaelam = 0
@@ -219,27 +149,26 @@ def constfn(val):
         return val
     return f
 
-@registry.register('ppo2', defaults=defaults)
-def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2048, ent_coef=0.0, lr=3e-4,
+def learn(*, network, env, total_timesteps, seed=None, nsteps=2048, ent_coef=0.0, lr=3e-4,
             vf_coef=0.5,  max_grad_norm=0.5, gamma=0.99, lam=0.95,
             log_interval=10, nminibatches=4, noptepochs=4, cliprange=0.2,
             save_interval=0, load_path=None, **network_kwargs):
     '''
     Learn policy using PPO algorithm (https://arxiv.org/abs/1707.06347)
-
+    
     Parameters:
     ----------
 
     network:                          policy network architecture. Either string (mlp, lstm, lnlstm, cnn_lstm, cnn, cnn_small, conv_only - see baselines.common/models.py for full list)
-                                      specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns
+                                      specifying the standard network architecture, or a function that takes tensorflow tensor as input and returns 
                                       tuple (output_tensor, extra_feed) where output tensor is the last network layer output, extra_feed is None for feed-forward
                                       neural nets, and extra_feed is a dictionary describing how to feed state into the network for recurrent neural nets.
-                                      See common/models.py/lstm for more details on using recurrent nets in policies
+                                      See baselines.common/policies.py/lstm for more details on using recurrent nets in policies
 
-    env: baselines.common.vec_env.VecEnv     environment. Needs to be vectorized for parallel environment simulation.
+    env: baselines.common.vec_env.VecEnv     environment. Needs to be vectorized for parallel environment simulation. 
                                       The environments produced by gym.make can be wrapped using baselines.common.vec_env.DummyVecEnv class.
 
-
+    
     nsteps: int                       number of steps of the vectorized environment per update (i.e. batch size is nsteps * nenv where
                                       nenv is number of environment copies simulated in parallel)
 
@@ -247,38 +176,37 @@ def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2
 
     ent_coef: float                   policy entropy coefficient in the optimization objective
 
-    lr: float or function             learning rate, constant or a schedule function [0,1] -> R+ where 1 is beginning of the
+    lr: float or function             learning rate, constant or a schedule function [0,1] -> R+ where 1 is beginning of the 
                                       training and 0 is the end of the training.
 
     vf_coef: float                    value function loss coefficient in the optimization objective
 
     max_grad_norm: float or None      gradient norm clipping coefficient
-
+    
     gamma: float                      discounting factor
 
     lam: float                        advantage estimation discounting factor (lambda in the paper)
 
     log_interval: int                 number of timesteps between logging events
 
-    nminibatches: int                 number of training minibatches per update. For recurrent policies,
-                                      should be smaller or equal than number of environments run in parallel.
+    nminibatches: int                 number of training minibatches per update
 
     noptepochs: int                   number of training epochs per update
 
-    cliprange: float or function      clipping range, constant or schedule function [0,1] -> R+ where 1 is beginning of the training
-                                      and 0 is the end of the training
+    cliprange: float or function      clipping range, constant or schedule function [0,1] -> R+ where 1 is beginning of the training 
+                                      and 0 is the end of the training 
 
     save_interval: int                number of timesteps between saving events
 
     load_path: str                    path to load the model from
 
     **network_kwargs:                 keyword arguments to the policy / network builder. See baselines.common/policies.py/build_policy and arguments to a particular type of network
-                                      For instance, 'mlp' network architecture has arguments num_hidden and num_layers.
-
+                                      For instance, 'mlp' network architecture has arguments num_hidden and num_layers. 
 
+    
 
     '''
-
+    
     set_global_seeds(seed)
 
     if isinstance(lr, float): lr = constfn(lr)
@@ -289,67 +217,41 @@ def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2
 
     policy = build_policy(env, network, **network_kwargs)
 
-    # Get the nb of env
     nenvs = env.num_envs
-
-    # Get state_space and action_space
     ob_space = env.observation_space
     ac_space = env.action_space
-
-    # Calculate the batch_size
     nbatch = nenvs * nsteps
     nbatch_train = nbatch // nminibatches
 
-    # Instantiate the model object (that creates act_model and train_model)
     make_model = lambda : Model(policy=policy, ob_space=ob_space, ac_space=ac_space, nbatch_act=nenvs, nbatch_train=nbatch_train,
                     nsteps=nsteps, ent_coef=ent_coef, vf_coef=vf_coef,
                     max_grad_norm=max_grad_norm)
+    if save_interval and logger.get_dir():
+        import cloudpickle
+        with open(osp.join(logger.get_dir(), 'make_model.pkl'), 'wb') as fh:
+            fh.write(cloudpickle.dumps(make_model))
     model = make_model()
     if load_path is not None:
         model.load(load_path)
-    # Instantiate the runner object
     runner = Runner(env=env, model=model, nsteps=nsteps, gamma=gamma, lam=lam)
-    if eval_env is not None:
-        eval_runner = Runner(env = eval_env, model = model, nsteps = nsteps, gamma = gamma, lam= lam)
-
-
 
     epinfobuf = deque(maxlen=100)
-    if eval_env is not None:
-        eval_epinfobuf = deque(maxlen=100)
-
-    # Start total timer
     tfirststart = time.time()
 
     nupdates = total_timesteps//nbatch
     for update in range(1, nupdates+1):
         assert nbatch % nminibatches == 0
-        # Start timer
         tstart = time.time()
         frac = 1.0 - (update - 1.0) / nupdates
-        # Calculate the learning rate
         lrnow = lr(frac)
-        # Calculate the cliprange
         cliprangenow = cliprange(frac)
-        # Get minibatch
         obs, returns, masks, actions, values, neglogpacs, states, epinfos = runner.run() #pylint: disable=E0632
-        if eval_env is not None:
-            eval_obs, eval_returns, eval_masks, eval_actions, eval_values, eval_neglogpacs, eval_states, eval_epinfos = eval_runner.run() #pylint: disable=E0632
-
         epinfobuf.extend(epinfos)
-        if eval_env is not None:
-            eval_epinfobuf.extend(eval_epinfos)
-
-        # Here what we're going to do is for each minibatch calculate the loss and append it.
         mblossvals = []
         if states is None: # nonrecurrent version
-            # Index of each element of batch_size
-            # Create the indices array
             inds = np.arange(nbatch)
             for _ in range(noptepochs):
-                # Randomize the indexes
                 np.random.shuffle(inds)
-                # 0 to batch_size with batch_train_size step
                 for start in range(0, nbatch, nbatch_train):
                     end = start + nbatch_train
                     mbinds = inds[start:end]
@@ -371,15 +273,10 @@ def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2
                     mbstates = states[mbenvinds]
                     mblossvals.append(model.train(lrnow, cliprangenow, *slices, mbstates))
 
-        # Feedforward --> get losses --> update
         lossvals = np.mean(mblossvals, axis=0)
-        # End timer
         tnow = time.time()
-        # Calculate the fps (frame per second)
         fps = int(nbatch / (tnow - tstart))
         if update % log_interval == 0 or update == 1:
-            # Calculates if value function is a good predicator of the returns (ev > 1)
-            # or if it's just worse than predicting nothing (ev =< 0)
             ev = explained_variance(values, returns)
             logger.logkv("serial_timesteps", update*nsteps)
             logger.logkv("nupdates", update)
@@ -388,9 +285,6 @@ def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2
             logger.logkv("explained_variance", float(ev))
             logger.logkv('eprewmean', safemean([epinfo['r'] for epinfo in epinfobuf]))
             logger.logkv('eplenmean', safemean([epinfo['l'] for epinfo in epinfobuf]))
-            if eval_env is not None:
-                logger.logkv('eval_eprewmean', safemean([epinfo['r'] for epinfo in eval_epinfobuf]) )
-                logger.logkv('eval_eplenmean', safemean([epinfo['l'] for epinfo in eval_epinfobuf]) )
             logger.logkv('time_elapsed', tnow - tfirststart)
             for (lossval, lossname) in zip(lossvals, model.loss_names):
                 logger.logkv(lossname, lossval)
@@ -402,8 +296,9 @@ def learn(*, network, env, total_timesteps, eval_env = None, seed=None, nsteps=2
             savepath = osp.join(checkdir, '%.5i'%update)
             print('Saving to', savepath)
             model.save(savepath)
+    env.close()
     return model
-# Avoid division error when calculate the mean (in our case if epinfo is empty returns np.nan, not return an error)
+
 def safemean(xs):
     return np.nan if len(xs) == 0 else np.mean(xs)
 

baselines/registry.py

@@ -1,39 +0,0 @@
-# Registry of algorithms that keeps track of algorithms supported environments and
-# and fine-grained defaults for different kinds of environments (atari, retro, mujoco etc)
-#
-# Example usage:
-#
-#   from baselines import registry
-#
-#   @registry.register('fancy_algorithm', supports_vecenv=False)
-#   def learn(env, network):
-#       return
-#
-#   for algo_name, algo_entry in registry.registry.items():
-#       if not algo_entry['supports_vecenv']:
-#           print(f'{algo_name} does not support vecenvs')
-#           # should print "fancy_algorithm does not support vecenvs" (among other ones)"f
-
-
-from baselines import logger
-
-registry = {}
-
-def register(name, supports_vecenv=True, defaults={}):
-    def get_fn_entrypoint(fn):
-        import inspect
-        return '.'.join([inspect.getmodule(fn).__name__, fn.__name__])
-
-    def _thunk(learn_fn):
-        old_entry = registry.get(name)
-        if old_entry is not None:
-            logger.warn('Re-registering learn function {} (old entrypoint {}, new entrypoint {}) '.format(
-                name, get_fn_entrypoint(old_entry['fn']), get_fn_entrypoint(learn_fn)))
-
-        registry[name] = dict(
-            fn = learn_fn,
-            supports_vecenv=supports_vecenv,
-            defaults=defaults,
-        )
-        return learn_fn
-    return _thunk