Custom Keras callbacks

class rl.callbacks.SaveDQNTrainingState(interval, state_path, memory, dqn, snapshot_limit=None)

Save agent progress, memory and model weights

on_episode_end(episode, logs)

Called at end of each episode

class rl.callbacks.TerminateTrainingOnFileExists(agent, poisonfile)

on_action_begin(action, logs)

Called at beginning of each action