dataset/meta/normalization.json

{
  "fit_split": "train",
  "episode_indices": [
    0,
    1,
    2,
    3,
    4,
    5,
    6,
    7
  ],
  "frame_count": 1120,
  "rgb_source": "lossless source PNG",
  "held_out_observations_used": false
}