Sample Efficient Actor-Critic with Experience Replay (ACER) combines several ideas of previous algorithms: it uses multiple workers (as A2C), implements a replay buffer (as in DQN), uses Retrace for Q-value estimation, importance sampling and a trust region.


  • Original paper:
  • python -m stable_baselines.acer.run_atari runs the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.

Can I use?

  • Reccurent policies: ✔️
  • Multi processing: ✔️
  • Gym spaces:
Space Action Observation
Discrete ✔️ ✔️
Box ✔️
MultiDiscrete ✔️
MultiBinary ✔️


import gym

from stable_baselines.common.policies import MlpPolicy, MlpLstmPolicy, MlpLnLstmPolicy
from stable_baselines.common.vec_env import SubprocVecEnv
from stable_baselines import ACER

# multiprocess environment
n_cpu = 4
env = SubprocVecEnv([lambda: gym.make('CartPole-v1') for i in range(n_cpu)])

model = ACER(MlpPolicy, env, verbose=1)

del model # remove to demonstrate saving and loading

model = ACER.load("acer_cartpole")

obs = env.reset()
while True:
    action, _states = model.predict(obs)
    obs, rewards, dones, info = env.step(action)


class stable_baselines.acer.ACER(policy, env, gamma=0.99, n_steps=20, num_procs=1, q_coef=0.5, ent_coef=0.01, max_grad_norm=10, learning_rate=0.0007, lr_schedule='linear', rprop_alpha=0.99, rprop_epsilon=1e-05, buffer_size=5000, replay_ratio=4, replay_start=1000, correction_term=10.0, trust_region=True, alpha=0.99, delta=1, verbose=0, tensorboard_log=None, _init_setup_model=True)[source]

The ACER (Actor-Critic with Experience Replay) model class,

  • policy – (ActorCriticPolicy or str) The policy model to use (MlpPolicy, CnnPolicy, CnnLstmPolicy, …)
  • env – (Gym environment or str) The environment to learn from (if registered in Gym, can be str)
  • gamma – (float) The discount value
  • n_steps – (int) The number of steps to run for each environment per update (i.e. batch size is n_steps * n_env where n_env is number of environment copies running in parallel)
  • num_procs – (int) The number of threads for TensorFlow operations
  • q_coef – (float) The weight for the loss on the Q value
  • ent_coef – (float) The weight for the entropic loss
  • max_grad_norm – (float) The clipping value for the maximum gradient
  • learning_rate – (float) The initial learning rate for the RMS prop optimizer
  • lr_schedule – (str) The type of scheduler for the learning rate update (‘linear’, ‘constant’, ‘double_linear_con’, ‘middle_drop’ or ‘double_middle_drop’)
  • rprop_epsilon – (float) RMSProp epsilon (stabilizes square root computation in denominator of RMSProp update) (default: 1e-5)
  • rprop_alpha – (float) RMSProp decay parameter (default: 0.99)
  • buffer_size – (int) The buffer size in number of steps
  • replay_ratio – (float) The number of replay learning per on policy learning on average, using a poisson distribution
  • replay_start – (int) The minimum number of steps in the buffer, before learning replay
  • correction_term – (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)
  • alpha – (float) The decay rate for the Exponential moving average of the parameters
  • delta – (float) max KL divergence between the old policy and updated policy (default: 1)
  • verbose – (int) the verbosity level: 0 none, 1 training information, 2 tensorflow debug
  • tensorboard_log – (str) the log location for tensorboard (if None, no logging)
  • _init_setup_model – (bool) Whether or not to build the network at the creation of the instance
action_probability(observation, state=None, mask=None)

Get the model’s action probability distribution from an observation

  • observation – (np.ndarray) the input observation
  • state – (np.ndarray) The last states (can be None, used in recurrent policies)
  • mask – (np.ndarray) The last masks (can be None, used in recurrent policies)

(np.ndarray) the model’s action probability distribution


returns the current environment (can be None if not defined)

Returns:(Gym Environment) The current environment
learn(total_timesteps, callback=None, seed=None, log_interval=100, tb_log_name='ACER')[source]

Return a trained model.

  • total_timesteps – (int) The total number of samples to train on
  • seed – (int) The initial seed for training, if None: keep current seed
  • callback – (function (dict, dict)) function called at every steps with state of the algorithm. It takes the local and global variables.
  • log_interval – (int) The number of timesteps before logging.
  • tb_log_name – (str) the name of the run for tensorboard log

(BaseRLModel) the trained model

classmethod load(load_path, env=None, **kwargs)

Load the model from file

  • load_path – (str) the saved parameter location
  • env – (Gym Envrionment) the new environment to run the loaded model on (can be None if you only need prediction from a trained model)
  • kwargs – extra arguments to change the model when loading
predict(observation, state=None, mask=None, deterministic=False)

Get the model’s action from an observation

  • observation – (np.ndarray) the input observation
  • state – (np.ndarray) The last states (can be None, used in recurrent policies)
  • mask – (np.ndarray) The last masks (can be None, used in recurrent policies)
  • deterministic – (bool) Whether or not to return deterministic actions.

(np.ndarray, np.ndarray) the model’s action and the next state (used in recurrent policies)


Save the current parameters to file

Parameters:save_path – (str) the save location

Checks the validity of the environment, and if it is coherent, set it as the current environment.

Parameters:env – (Gym Environment) The environment for learning a policy

Create all the functions and tensorflow graphs necessary to train the model