CoreCallback

Abstract base class for all implemented callback.

Do not use this abstract base class directly but instead use one of the concrete callback implemented.

To implement your own callback, you have to implement the following methods:

  • on_action_begin
  • on_action_end
  • on_replay_begin
  • on_replay_end
  • on_episode_begin
  • on_episode_end
  • on_agent_begin
  • on_agent_end

Methods:

.on_agent_begin

.on_agent_begin(
   *args, **kwargs
)

Called at beginning of each agent play

.on_agent_end

.on_agent_end(
   *args, **kwargs
)

Called at end of each agent play

.on_episode_begin

.on_episode_begin(
   *args, **kwargs
)

Called at beginning of each game episode

.on_episode_end

.on_episode_end(
   *args, **kwargs
)

Called at end of each game episode

.on_action_begin

.on_action_begin(
   *args, **kwargs
)

Called at beginning of each agent action

.on_action_end

.on_action_end(
   *args, **kwargs
)

Called at end of each agent action

.on_replay_begin

.on_replay_begin(
   *args, **kwargs
)

Called at beginning of each nn replay

.on_replay_end

.on_replay_end(
   *args, **kwargs
)

Called at end of each nn replay

CallbackList

CallbackList(
   callbacks
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_action_begin

.on_action_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_action_end

.on_action_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_agent_begin

.on_agent_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_agent_end

.on_agent_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_begin

.on_episode_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_end

.on_episode_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_replay_begin

.on_replay_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_replay_end

.on_replay_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

Loader

Loader(
   model, run_path, interval
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_agent_begin

.on_agent_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_agent_end

.on_agent_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_end

.on_episode_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

Renderer

Renderer(
   environment
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_action_end

.on_action_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_begin

.on_episode_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

TrainLogger

TrainLogger(
   run_path, interval
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_agent_begin

.on_agent_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_begin

.on_episode_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_end

.on_episode_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_replay_end

.on_replay_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

ValidationLogger

ValidationLogger(
   run_path, interval
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_agent_begin

.on_agent_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_begin

.on_episode_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.on_episode_end

.on_episode_end(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

Visualizer

Visualizer(
   model, predicate = lambdax: True
)

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.on_action_begin

.on_action_begin(
   *args, **kwargs
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

LivePlotter

LivePlotter()

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.live_plotter

.live_plotter(
   y_num, size
)

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.