CoolFace
Datasetpublic

mad-bot/mind-games-demos

Mind Games — Demos Gameplay demonstrations for the mind-games project. Lunar Lander 200 episodes of LunarLander-v3 (Gymnasium) collected with a heuristic PD-controller expert. Format: H.264 video (84×84) + NumPy observation arrays + JSON metadata per episode Size: ~35MB Mean reward: ~170 (expert heuristic averages ~163) Structure: lunar_lander/episode_NNNN/ with frames.mp4, observations.npy, metadata.json Index: lunar_lander/index.json with per-episode reward and… See the full description on the dataset page: https://huggingface.co/datasets/mad-bot/mind-games-demos.

sourceHugging Facemitupdated 6mo agoView on Hugging Face
0likes421downloads
9 commits on main
d0c58726mo ago

Add Claude-policy Quake golden demos (1 episodes, 1MB)

mad-bot
e6f97f46mo ago

Add golden trackmania demos (100 episodes, 327MB)

mad-bot
b2f9d336mo ago

Add golden quake demos (100 episodes, 97MB)

mad-bot
2ff2d246mo ago

Add golden lunar demos (100 episodes, 22MB)

mad-bot
76267656mo ago

Add TrackMania TMNF demos (200 episodes, 880MB)

mad-bot
b4934166mo ago

Add quake demo episodes (193 episodes)

mad-bot
709e9c26mo ago

Upload README.md with huggingface_hub

mad-bot
90428f46mo ago

Upload folder using huggingface_hub

mad-bot
3b04ec06mo ago

initial commit

mad-bot