meta/info.json: { "codebase_version": "v3.0", "robot_type": "so_follower", "total_episodes": 10, "total_frames": 6912, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:10" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/Alkatt/eval_LAVLA_S1_Dummy_S2_whiteblack_cube_v2_24_04_2026_v4.