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.
Add Claude-policy Quake golden demos (1 episodes, 1MB)
Add golden trackmania demos (100 episodes, 327MB)
Add golden quake demos (100 episodes, 97MB)
Add golden lunar demos (100 episodes, 22MB)
Add TrackMania TMNF demos (200 episodes, 880MB)
Add quake demo episodes (193 episodes)
Upload README.md with huggingface_hub
Upload folder using huggingface_hub
initial commit
