robomotic/causality-two-room

--- license: mit task_categories: - reinforcement-learning - robotics language: - en pretty_name: Causality Two Room (Glitched Hue) size_categories: - 10K<n<100K --- # robomotic/causality-two-room This dataset contains trajectories collected in the `swm/GlitchedHueTwoRoom-v1` environment for causal world-model experiments. ## What is inside - `glitched_hue_tworoom.h5`: HDF5 dataset with trajectories and rendered frames. - Includes observations, actions, rewards, episode indexing, variation values, and pixel renderings. ## How it was generated Collection script: - `scripts/data/collect_glitched_hue.py` Configuration: - `scripts/data/config/glitched_hue.yaml` Command used: ```bash python scripts/data/collect_glitched_hue.py num_traj=10000 seed=3072 ``` Generation details: - Total episodes: 10,000 - Pass 1: 5,000 episodes with blue background + teleport enabled - Pass 2: 5,000 episodes with green background + teleport disabled - Render size: 224x224 - File size: 9.09 GB - Uploaded at (UTC): 2026-04-07 09:09:48 ## Repository - GitHub: https://github.com/epokhcs/stable-worldmodel ## Notes The second background hue is green (not red) to preserve strong contrast with the red agent in pixel observations.