experiments.experiment_utils.generate_behavior_policy_episodes

generate_behavior_policy_episodes(hyperparameter_and_setting_dict, n_trials, save_dir, verbose=False)

Utility function for reinforcement learning to generate new episodes using the behavior policy to use in each trial.

Parameters:
  • hyperparameter_and_setting_dict (dict) – Contains the number of episodes to generate, environment, agent, etc. needed for generating new episodes.

  • n_trials – The number of experiment trials to run per data fraction

  • save_dir (str) – The parent directory in which to save the regenerated_episodes