qualia2.rl package

Subpackages

• qualia2.rl.agents package
  • Submodules
  • qualia2.rl.agents.ddqn module
  • qualia2.rl.agents.dqn module
  • qualia2.rl.agents.td3 module
  • Module contents
• qualia2.rl.envs package
  • Submodules
  • qualia2.rl.envs.atari module
  • qualia2.rl.envs.box2d module
  • qualia2.rl.envs.classic_control module
  • qualia2.rl.envs.roboschool module
  • qualia2.rl.envs.toy_text module
  • Module contents

Submodules

qualia2.rl.memory module

class qualia2.rl.memory.Experience(state, next, reward, action, done)

Bases: tuple

property action

Alias for field number 3

property done

Alias for field number 4

property next

Alias for field number 1

property reward

Alias for field number 2

property state

Alias for field number 0

class qualia2.rl.memory.PrioritizedMemory(maxlen, alpha=0.6, beta_start=0.4, beta_frames=100000)

Bases: collections.deque

append(experience)

Add an element to the right side of the deque.

beta()

sample(batch_size, steps=1)

update_priorities(idx, priorities)

class qualia2.rl.memory.ReplayMemory(maxlen)

Bases: collections.deque

sample(batch_size, steps=1)

qualia2.rl.rl_core module

class qualia2.rl.rl_core.ActorCriticAgent(actor, critic)

Bases: qualia2.rl.rl_core.BaseAgent

Base class for actor-critic based agents. Some methods needs to be over ridden. Args:

actor (Module): actor network critic (Module): critic network

classmethod init(env, actor, critic)

load(filename)

load_actor(filename)

policy(observation, *args, eps=None)

save(filename)

set_actor_optim(optim, **kwargs)

set_critic_optim(optim, **kwargs)

class qualia2.rl.rl_core.BaseAgent(actions, model)

Bases: object

Base class for agents. Some methods needs to be over ridden. Args:

actions (list): list of actions model (Module): model network

get_train_signal(experience, gamma=0.9)

classmethod init(env, model)

load(filename)

play(env, render=True, filename=None)

policy(observation, *args)

save(filename)

set_optim(optim, **kwargs)

update(state_action_value, target_action_value, loss_func=<qualia2.functions.loss.MSELoss object>)

update_target_model()

class qualia2.rl.rl_core.Env(env)

Bases: object

Wrapper class of gym.env for reinforcement learning. Args:

env (str): task name

property action_space

animate(frames, filename)

close()

property max_steps

property observation_space

render(**kwargs)

reset()

reward_transformer(reward)

show(filename=None)

state_transformer(state)

step(action)

class qualia2.rl.rl_core.PolicyAgent(actions, model)

Bases: qualia2.rl.rl_core.BaseAgent

Base class for policy based agents. Some methods needs to be over ridden.

policy(observation, *args, eps=None)

class qualia2.rl.rl_core.ValueAgent(actions, model)

Bases: qualia2.rl.rl_core.BaseAgent

Base class for value based agents. Some methods needs to be over ridden.

policy(observation, *args, eps=None)

qualia2.rl.rl_util module

class qualia2.rl.rl_util.Trainer(memory, batch, capacity, gamma)

Bases: object

Trainer for RL agent

Args:

memory (deque): replay memory object capacity (int): capacity of the memory batch (int): batch size for training gamma (int): gamma value

after_episode(episode, steps, agent, loss, reward, filename=None)

after_train()

before_episode(env, agent)

before_train(env, agent)

property defaults

experience_replay(episode, step_count, agent)

load_settings(defaults)

plot(filename=None)

train(env, agent, episodes=200, render=False, filename=None)

train_routine(env, agent, episodes=200, render=False, filename=None)

Module contents