Examples

Descriptions of Algorithms and Environments in RLZoo

Algorithms	Action Space	Policy	Update	Envs
DQN (double, dueling, PER)	Discrete Only	--	Off-policy	Atari, Classic Control
AC	Discrete/Continuous	Stochastic	On-policy	All
PG	Discrete/Continuous	Stochastic	On-policy	All
DDPG	Continuous	Deterministic	Off-policy	Classic Control, Box2D, Mujoco, Robotics, DeepMind Control, RLBench
TD3	Continuous	Deterministic	Off-policy	Classic Control, Box2D, Mujoco, Robotics, DeepMind Control, RLBench
SAC	Continuous	Stochastic	Off-policy	Classic Control, Box2D, Mujoco, Robotics, DeepMind Control, RLBench
A3C	Discrete/Continuous	Stochastic	On-policy	Atari, Classic Control, Box2D, Mujoco, Robotics, DeepMind Control
PPO	Discrete/Continuous	Stochastic	On-policy	All
DPPO	Discrete/Continuous	Stochastic	On-policy	Atari, Classic Control, Box2D, Mujoco, Robotics, DeepMind Control
TRPO	Discrete/Continuous	Stochastic	On-policy	All

1. Deep Q-Network (DQN)

AlgName = 'DQN'
EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'
# EnvName = 'CartPole-v1'
# EnvType = 'classic_control'  # the name of env needs to match the type of env

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

2. Actor-Critic (AC)

AlgName = 'AC'
EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'Pendulum-v0'
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

3. Policy Gradient (PG)

AlgName = 'PG'
EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'CartPole-v0'
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

4. Deep Deterministic Policy Gradient (DDPG)

AlgName = 'DDPG'
EnvName = 'Pendulum-v0'  # only continuous action
EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

5. Twin Delayed DDPG (TD3)

AlgName = 'TD3'
EnvName = 'Pendulum-v0'  # only continuous action
EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

6. Soft Actor-Critic (SAC)

AlgName = 'SAC'
EnvName = 'Pendulum-v0'  # only continuous action
EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

7. Asynchronous Advantage Actor-Critic (A3C)

AlgName = 'A3C'
EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'Pendulum-v0'  # only continuous action
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

number_workers = 2  # need to specify number of parallel workers
env = build_env(EnvName, EnvType, nenv=number_workers)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env,  mode='train', render=False, **learn_params)
alg.learn(env=env,  mode='test', render=True, **learn_params)

8. Proximal Policy Optimization (PPO)

EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'Pendulum-v0'
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, 'PPO')
alg = PPO(method='clip', **alg_params) # specify 'clip' or 'penalty' method for PPO
alg.learn(env=env,  mode='train', render=False, **learn_params)
alg.learn(env=env,  mode='test', render=False, **learn_params)

9. Distributed Proximal Policy Optimization (DPPO)

EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'Pendulum-v0'
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

number_workers = 2  # need to specify number of parallel workers
env = build_env(EnvName, EnvType, nenv=number_workers)
alg_params, learn_params = call_default_params(env, EnvType, 'DPPO')
alg = DPPO(method='penalty', **alg_params) # specify 'clip' or 'penalty' method for PPO
alg.learn(env=env,  mode='train', render=False, **learn_params)
alg.learn(env=env,  mode='test', render=True, **learn_params)

10. Trust Region Policy Optimization (TRPO)

AlgName = 'TRPO'
EnvName = 'PongNoFrameskip-v4'
EnvType = 'atari'

# EnvName = 'CartPole-v0'
# EnvType = 'classic_control'

# EnvName = 'BipedalWalker-v2'
# EnvType = 'box2d'

# EnvName = 'Ant-v2'
# EnvType = 'mujoco'

# EnvName = 'FetchPush-v1'
# EnvType = 'robotics'

# EnvName = 'FishSwim-v0'
# EnvType = 'dm_control'

# EnvName = 'ReachTarget'
# EnvType = 'rlbench'

env = build_env(EnvName, EnvType)
alg_params, learn_params = call_default_params(env, EnvType, AlgName)
alg = eval(AlgName+'(**alg_params)')
alg.learn(env=env, mode='train', render=False, **learn_params)
alg.learn(env=env, mode='test', render=True, **learn_params)

TensorLayer/RLzoo

Examples

Descriptions of Algorithms and Environments in RLZoo

1. Deep Q-Network (DQN)

2. Actor-Critic (AC)

3. Policy Gradient (PG)

4. Deep Deterministic Policy Gradient (DDPG)

5. Twin Delayed DDPG (TD3)

6. Soft Actor-Critic (SAC)

7. Asynchronous Advantage Actor-Critic (A3C)

8. Proximal Policy Optimization (PPO)

9. Distributed Proximal Policy Optimization (DPPO)

10. Trust Region Policy Optimization (TRPO)

简介

发行版

贡献者

语言

近期动态

TensorLayer/RLzoo .gitee-modal { width: 500px !important; }

Examples

Descriptions of Algorithms and Environments in RLZoo

1. Deep Q-Network (DQN)

2. Actor-Critic (AC)

3. Policy Gradient (PG)

4. Deep Deterministic Policy Gradient (DDPG)

5. Twin Delayed DDPG (TD3)

6. Soft Actor-Critic (SAC)

7. Asynchronous Advantage Actor-Critic (A3C)

8. Proximal Policy Optimization (PPO)

9. Distributed Proximal Policy Optimization (DPPO)

10. Trust Region Policy Optimization (TRPO)

简介

发行版

开源评估指数源自 OSS-Compass 评估体系，评估体系围绕以下三个维度对项目展开评估：

贡献者

语言

近期动态

搜索帮助

TensorLayer/RLzoo