DILLO LIBERO Distillation Dataset
This dataset contains LIBERO policy rollouts labeled for DILLO (DIstiLLed Language-ActiOn World Model). Each example stores a chunked ACT policy rollout, boundary-frame images, robot state traces, action chunks, and VLM-generated descriptions/reasoning for distillation.
Dataset Summary
- Total episodes: 1700
- Total tasks: 130
- Total files: 33079
- Approximate size: 555.3 MiB
- Hugging Face repo:
Sapienza/DILLO-LIBERO-dataset - Layout:
root/LIBERO_{SUITE}/{TASK}/{EPISODE}
Suites
LIBERO_10: 200 episodes across 10 tasksLIBERO_90: 900 episodes across 90 tasksLIBERO_GOAL: 200 episodes across 10 tasksLIBERO_OBJECT: 200 episodes across 10 tasksLIBERO_SPATIAL: 200 episodes across 10 tasks
File Structure
Each example is stored as:
LIBERO_{SUITE}/{TASK}/{EPISODE}/
response.txt
reasoning.txt
task_instruction.txt
eef_pos.npy
joint_pos.npy
gripper_states.npy
actions.npy
success_mask.npy
images/*.jpeg
Root metadata files:
metadata.jsonl: one row per episode with relative paths, shapes, task text, and success label.metadata.csv: CSV version of the same manifest.dataset_summary.json: aggregate counts by suite/task and file extension..gitattributes: LFS rules for binary arrays and images.
File Types
.jpeg: 19479.npy: 8500.txt: 5100
Example Rows
LIBERO_10/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it/000: turn on the stove and put the moka pot on itLIBERO_10/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it/001: turn on the stove and put the moka pot on itLIBERO_10/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it/002: turn on the stove and put the moka pot on it
Loading
import json
from pathlib import Path
root = Path('/path/to/DILLO-LIBERO-dataset')
rows = [json.loads(line) for line in (root / 'metadata.jsonl').open()]
first = rows[0]
print(first['episode_path'], first['task_instruction'])
For DILLO training, point TRAIN_DATA to a suite glob after downloading:
TRAIN_DATA='DILLO-LIBERO-dataset/LIBERO_GOAL/*/*'