mdp_playground.config_processor.config_processor

Functions

combined_processing(*static_configs, …[, …])

varying_configs is a list of dict of dicts with structure of each element in the list as: { “env”: {…} “agent”: {…} “model”: {…} }

create_config_space_from_config_dict(config_dict)

create_gym_env_wrapper_atari(config)

create_gym_env_wrapper_frame_stack_atari(config)

When using frameStack GymEnvWrapper should wrap AtariEnv using wrap_deepmind_ray and therefore this function sets “wrap_deepmind_ray”: True and ‘frame_skip’: 1 inside config so as to keep config same as for create_gym_env_wrapper_atari above and reduce manual errors when switching between the 2.

create_gym_env_wrapper_mujoco_wrapper(…)

Creates a GymEnvWrapper around a MujocoEnvWrapper

deepmerge(a, b[, path, overwrite])

Merges dict b into dict a

deepmerge_multiple_dicts(*configs[, overwrite])

get_grid_of_configs(var_configs)

get_list_of_varying_configs(var_configs[, …])

var_configs: dict of dicts of lists as values

get_random_configs(var_configs, num_configs)

var_configs: dict of dicts of strings or tuples as values

init_ray(**kwargs)

init_stats_file(stats_file_name, …)

post_processing([framework])

process_configs(config_file, …[, …])

setup_ray(config, config_num, log_level, …)

setup_ray_callbacks(stats_file_prefix, …)

sobol_configs_from_config_dict(var_configs, …)