Reinforcement Learning
stable-baselines3
ppo
stable-retro
rlab
super-mario-bros
nes
SuperMarioBros-Nes-v0
Instructions to use tsilva/SuperMarioBros-Nes-v0_Level1-4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use tsilva/SuperMarioBros-Nes-v0_Level1-4 with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="tsilva/SuperMarioBros-Nes-v0_Level1-4", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
SuperMarioBros-NES Level1-4 PPO
PPO policy checkpoint for completing SuperMarioBros-Nes-v0 Level1-4 with Stable Retro, trained with rlab.
Quick Start
Install rlab once, import the ROM, then play or evaluate this checkpoint directly from Hugging Face:
uv tool install --from git+https://github.com/tsilva/rlab rlab
rlab import-roms ~/roms --game SuperMarioBros-Nes-v0
rlab play hf://tsilva/SuperMarioBros-Nes-v0_Level1-4
rlab eval hf://tsilva/SuperMarioBros-Nes-v0_Level1-4
Evaluation Results
eval_profile |
episodes |
seed_start |
completion_rate |
max_x_max |
reward_mean |
checkpoint_step |
|---|---|---|---|---|---|---|
fresh_release_eval, deterministic policy |
100 | 10000 | 100/100 episodes |
2304 | 2353.900 | 4500000 |
This is a checkpoint promotion metric from the current rlab release process.
Preview
The root replay.mp4 is the Hugging Face model-page preview. The same preview is also available on YouTube: https://www.youtube.com/watch?v=fvxFCTK9is8
Environment Details
| Setting | Value |
|---|---|
env_provider |
supermariobrosnes-turbo |
env_id |
supermariobrosnes-turbo:SuperMarioBros-Nes-v0 |
game |
SuperMarioBros-Nes-v0 |
state |
Level1-4 |
preprocessing |
crop top 32 px, grayscale, resize to 84 x 84 |
frame_stack |
4 |
frame_skip |
4 |
max_pool_frames |
disabled |
policy_observation_layout |
channel-first (4, 84, 84) |
action_set |
simple |
reward_mode |
score |
reward_shaping |
progress_reward_scale=1.0, terminal_reward=50, death_penalty=25, completion_reward=0.0, reward_scale=10 |
max_episode_steps |
4500 |
training_done_on_events |
life loss, level change |
eval_done_on_events |
level change |
Training Recipe
| Setting | Value |
|---|---|
goal_id |
Level1-4 |
recipe_id |
base |
checkpoint_step |
4500000 |
frame_skip |
4 |
max_episode_steps |
4500 |
reward_mode |
score |
reward_shaping |
progress_reward_scale=1.0, terminal_reward=50, death_penalty=25, completion_reward=0.0, reward_scale=10 |
done_on_events |
life loss, level change |
Provenance
| Item | Value |
|---|---|
source_project |
rlab |
goal_id |
Level1-4 |
goal_title |
Sample-efficient SuperMarioBros-NES Level1-4 completion |
wandb_run |
Level1-4_base_s_20260704T081304Z |
wandb_artifact |
tsilva/SuperMarioBros-Nes-v0/Level1-4_base_s_20260704T081304Z-checkpoint:step-4500000 |
eval_source |
fresh_release_eval |
- Downloads last month
- 32