gym-super-mario-bros 9.0.0

Super Mario Bros. for Gymnasium

gym-super-mario-bros

A Gymnasium environment for Super Mario Bros., Super Mario Bros. 2 (Lost Levels), and Super Mario Bros. 2 (USA), and Super Mario Bros. 3 on The Nintendo Entertainment System (NES) using the nes-py emulator.

gym-super-mario-bros targets Gymnasium's modern reset, step, render-mode, and truncation semantics. It currently supports CPython 3.13 and 3.14 in CI.

Installation

The preferred installation of gym-super-mario-bros is from pip:

pip install gym-super-mario-bros

Python 3.13 or newer is required. The supported CI targets are CPython 3.13 and 3.14.

Because gym-super-mario-bros depends on the native nes-py emulator, Linux GLIBCXX_* loader errors and Windows compiler-toolchain failures are usually nes-py installation/runtime issues rather than wrapper bugs. See the nes-py installation notes for the current compiler and runtime expectations.

Usage

Python

You must import gym_super_mario_bros before trying to make an environment. This is because Gymnasium environments are registered at runtime. By default, gym_super_mario_bros environments use the full NES action space of 256 discrete actions. To constrain this, gym_super_mario_bros.actions provides three actions lists (RIGHT_ONLY, SIMPLE_MOVEMENT, and COMPLEX_MOVEMENT) for the nes_py.wrappers.JoypadSpace wrapper. See gym_super_mario_bros/actions.py for a breakdown of the legal actions in each of these three lists.

import gymnasium as gym
from nes_py.wrappers import JoypadSpace
import gym_super_mario_bros
from gym_super_mario_bros.actions import SIMPLE_MOVEMENT

env = gym.make('SuperMarioBros-v0', render_mode='human')
env = JoypadSpace(env, SIMPLE_MOVEMENT)

done = True
for step in range(5000):
    if done:
        state, info = env.reset(seed=123)
    state, reward, terminated, truncated, info = env.step(env.action_space.sample())
    done = terminated or truncated
    env.render()

env.close()

NOTE: gym_super_mario_bros.make is just an alias to gymnasium.make for convenience after gym_super_mario_bros is imported.

NOTE: registered environments use Gymnasium's TimeLimit wrapper with max_episode_steps=9999999 to preserve the historical cap while allowing the game logic to end normal episodes with terminated=True. Passing a shorter max_episode_steps to gymnasium.make() is the supported way to test or train with external time truncation, which returns truncated=True.

NOTE: remove calls to render in training code for a nontrivial speedup.

Task Metadata

gym_super_mario_bros exposes lightweight task metadata for curriculum, conditioning, and evaluation-matrix code without constructing ROM-backed environments:

import gym_super_mario_bros

tasks = gym_super_mario_bros.all_tasks(single_stage=True)
env_ids = gym_super_mario_bros.task_ids(game_family='smb1')
task = gym_super_mario_bros.task_for_env_id('SuperMarioBros-4-2-v0')

Each MarioTask includes the environment ID, canonical task ID, game family, ROM mode, world/stage, public world label, split flags, and alias metadata for separator-free SMB1 IDs such as SuperMarioBros1-1-v0.

Command Line

gym_super_mario_bros features a command line interface for playing environments using either the keyboard, or uniform random movement.

python -m gym_super_mario_bros --env <environment ID> --mode <human|random>
gym_super_mario_bros --env <environment ID> --mode <human|random>

NOTE: by default, -e is set to SuperMarioBros-v0 and -m is set to human, --actionspace/-a is set to nes, and rendering is enabled.

Human keyboard play opens a graphical window:

gym_super_mario_bros --env SuperMarioBros-v0 --mode human --actionspace simple

Random play can be rendered or run headlessly. Use --seed to seed the first environment reset:

gym_super_mario_bros --mode random --steps 1000 --no-render --seed 123

Use --actionspace/-a to select nes, right, simple, or complex. Human mode requires rendering, so --mode human --no-render is rejected.

Print the CLI help with:

python -m gym_super_mario_bros --help

Environments

These environments allow 3 attempts (lives) to make it through the 32 stages in the game. The environments only send reward-able game-play frames to agents; No cut-scenes, loading screens, etc. are sent from the NES emulator to an agent nor can an agent perform actions during these instances. If a cut-scene is not able to be skipped by hacking the NES's RAM, the environment will lock the Python process until the emulator is ready for the next action.

Environment	Game	ROM
SuperMarioBros-v0	SMB	standard
SuperMarioBros2-v0	SMB2	standard
SuperMarioBros2USA-v0	SMB2 USA	standard
SuperMarioBros3-v0	SMB3	standard

Individual Stages

These environments allow a single attempt (life) to make it through a single stage of the game.

Use the template

SuperMarioBros-<world>-<stage>-v<version>

where:

• <world> is a number in {1, 2, 3, 4, 5, 6, 7, 8} indicating the world
• <stage> is a number in {1, 2, 3, 4} indicating the stage within a world
• <version> is 0 for the standard ROM

For example, to play 4-2 on the standard ROM, you would use the environment id SuperMarioBros-4-2-v0.

Super Mario Bros. 2 (USA) uses the vanilla ROM only. Use SuperMarioBros2USA-v0 for the full game, or the template

SuperMarioBros2USA-<world>-<stage>-v0

where <world> is a number in {1, 2, 3, 4, 5, 6, 7}. Worlds 1 through 6 have stages {1, 2, 3}; world 7 has stages {1, 2}.

Super Mario Bros. 3 uses the vanilla ROM only. Use SuperMarioBros3-v0 for the game, or SuperMarioBros3-1-1-v0 for the validated World 1-1 single-stage entry point.

Step

Info about the rewards and info returned by the step method.

Reward Function

The reward function combines dense progress with objective events. Progress is rewarded only when the agent reaches a new best position for the attempt, so tactical backtracking is not punished by the movement term. Score increases, coins, cherries, powerups, health changes, level completion, time pressure, and death penalties are then added when the underlying game exposes reliable RAM counters for that title.

The reward is clipped into the range (-15, 15). The info dictionary also includes reward_components, reward_total_unclipped, and reward_total_clipped so training code can log or choose alternate reward transforms without recomputing RAM-dependent shaping.

info dictionary

The info dictionary returned by the step method contains the following keys:

Key	Type	Description
coins	int	The number of collected coins where available
flag_get	bool	True if Mario reached a flag, ax, or stage-complete state
life	int	The title-specific displayed life count
score	int	The cumulative in-game score where available
stage	int	The current stage
status	str	Mario's title-specific powerup status
time	int	The time left on the clock where available
world	int	The current world
x_pos	int	Mario's horizontal position where available
y_pos	int	Mario's vertical position where available
clear	bool	Cross-game completion flag for the active stage/task
death	bool	Cross-game death/life-loss flag
game	str	Normalized game identifier such as smb1 or smb3
game_family	str	Grouping key for task suites and evaluation matrices
progress	int	Cross-game progress metric used by the reward function
progress_max	int	Best progress reached during the current attempt
rom_mode	str	ROM mode, currently vanilla for registered environments
single_stage	bool	True for single-stage registered tasks
task_id	str	Canonical task ID suitable for task conditioning
target_world	int or None	Configured target world for single-stage tasks
target_stage	int or None	Configured target stage for single-stage tasks
timeout	bool	Reserved cross-game timeout flag; external Gymnasium TimeLimit still sets truncated=True
world_label	str	Public world label, including Lost Levels bonus worlds
reward_components	dict	Per-step shaped reward terms before clipping
reward_total_unclipped	float	Per-step shaped reward before reward_range clipping
reward_total_clipped	float	Per-step shaped reward after reward_range clipping

Newer SMB2 USA and SMB3 environments include additional game-specific keys such as raw transition state, health, lives, map position, powerup timers, P-meter state, invulnerability timers, and progress maxima where those values are available from the ROM's RAM map.

Publishing

PyPI releases are published by the Publish to PyPI GitHub Actions workflow through PyPI trusted publishing, not by local twine credentials. Configure the PyPI project publisher with owner Kautenja, repository gym-super-mario-bros, workflow filename publish.yml, and environment pypi. Publish a release by pushing a tag that matches pyproject.toml's version, with or without a leading v, and then creating the corresponding GitHub release so the workflow can build and upload the distribution artifacts.

Citation

Please cite gym-super-mario-bros if you use it in your research.

@misc{gym-super-mario-bros,
  author = {Christian Kauten},
  howpublished = {GitHub},
  title = {{S}uper {M}ario {B}ros for {O}pen{AI} {G}ym},
  URL = {https://github.com/Kautenja/gym-super-mario-bros},
  year = {2018},
}