datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
stack_franka_franka_umi_tacticleThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 8017,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/stack_franka_franka_umi_tacticle.pick_place_fruit_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 8123,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/pick_place_fruit_franka_allegro.stack_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 16603,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/stack_franka_allegro.pick_place_franka_franka_umi_tacticleThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 5669,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/pick_place_franka_franka_umi_tacticle.push_franka_franka_umi_tacticleThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 6065,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/push_franka_franka_umi_tacticle.pick_and_place_autoThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 6529,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/pick_and_place_auto.drill_to_point_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 6503,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/drill_to_point_franka_allegro.lift_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 3723,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/lift_franka_allegro.lift_franka_franka_umi_tacticleThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 3515,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/lift_franka_franka_umi_tacticle.push_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 3109,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/push_franka_allegro.stack_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for stack with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_250_stack_franka_allegro_success_0.06.pth; 202 successes in 5998 attempts, success rate 0.03; env seed = base seed + attempt index). All episodes are successful, recorded… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/stack_franka_allegro_mg.pick_place_franka_franka_umi_tacticle_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for pick_place with the Franka arm and the franka_umi_tacticle end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_300_pick_place_franka_franka_umi_tacticle_success_0.7.pth; 200 successes in 367 attempts, success rate 0.54; env seed = base seed + attempt index). All… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_franka_franka_umi_tacticle_mg.push_franka_franka_umi_tacticle_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for push with the Franka arm and the franka_umi_tacticle end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_300_push_franka_franka_umi_tacticle_success_0.62.pth; 200 successes in 366 attempts, success rate 0.55; env seed = base seed + attempt index). All episodes are… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/push_franka_franka_umi_tacticle_mg.lift_franka_franka_umi_tacticle_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for lift with the Franka arm and the franka_umi_tacticle end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_350_lift_franka_franka_umi_tacticle_success_0.94.pth; 200 successes in 227 attempts, success rate 0.88; env seed = base seed + attempt index). All episodes are… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_mg.pick_place_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for pick_place with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_250_pick_place_franka_allegro_success_0.74.pth; 200 successes in 304 attempts, success rate 0.66; env seed = base seed + attempt index). All episodes are successful… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_franka_allegro_mg.pick_place_fruit_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for pick_place_fruit with the Franka arm and the allegro end-effector: 200 successful rollouts of an image-based ACT policy (front camera + proprioception, 40-step action chunk) trained on the 50 human demonstrations (checkpoint model_epoch_250_pick_place_fruit_franka_allegro_img_success_0.4.pth; 204 successes in 739 attempts, success rate 0.28; env seed = base seed + attempt… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_fruit_franka_allegro_mg.pick_place_franka_franka_umi_tacticle_scriptedThis dataset was created using LeRobot.
Dataset Description
Scripted-expert DexSuite data for pick_place with the Franka arm and the UMI tactile parallel gripper: 200 successful episodes from the waypoint planner with randomized object initialization, including dense fingertip tactile readings (Flexitac simulation). The wrist camera looks at the flat face of the fingers (2026-09-16 mount). All episodes are successful, recorded at 20 Hz with a static front camera and a… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_franka_franka_umi_tacticle_scripted.stack_franka_franka_umi_tacticle_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for stack with the Franka arm and the franka_umi_tacticle end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_250_stack_franka_franka_umi_tacticle_success_0.14.pth; 202 successes in 2676 attempts, success rate 0.08; env seed = base seed + attempt index). All episodes… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/stack_franka_franka_umi_tacticle_mg.push_franka_franka_umi_tacticle_scriptedThis dataset was created using LeRobot.
Dataset Description
Scripted-expert DexSuite data for push with the Franka arm and the UMI tactile parallel gripper: 200 successful episodes from the waypoint planner with randomized object initialization, including dense fingertip tactile readings (Flexitac simulation). The wrist camera looks at the flat face of the fingers (2026-09-16 mount). All episodes are successful, recorded at 20 Hz with a static front camera and a wrist… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/push_franka_franka_umi_tacticle_scripted.push_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for push with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_350_push_franka_allegro_success_0.66.pth; 200 successes in 335 attempts, success rate 0.60; env seed = base seed + attempt index). All episodes are successful, recorded at… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/push_franka_allegro_mg.pick_place_mug_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for pick_place_mug with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_350_pick_place_mug_franka_allegro_success_0.8.pth; 200 successes in 257 attempts, success rate 0.78; env seed = base seed + attempt index). All episodes are… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_mug_franka_allegro_mg.stack_franka_franka_umi_tacticle_scriptedThis dataset was created using LeRobot.
Dataset Description
Scripted-expert DexSuite data for stack with the Franka arm and the UMI tactile parallel gripper: 200 successful episodes from the waypoint planner with randomized object initialization, including dense fingertip tactile readings (Flexitac simulation). The wrist camera looks at the flat face of the fingers (2026-09-16 mount). All episodes are successful, recorded at 20 Hz with a static front camera and a wrist… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/stack_franka_franka_umi_tacticle_scripted.lift_franka_franka_umi_tacticle_scriptedThis dataset was created using LeRobot.
Dataset Description
Scripted-expert DexSuite data for lift with the Franka arm and the UMI tactile parallel gripper: 200 successful episodes from the waypoint planner with randomized object initialization, including dense fingertip tactile readings (Flexitac simulation). The wrist camera looks at the flat face of the fingers (2026-09-16 mount). All episodes are successful, recorded at 20 Hz with a static front camera and a wrist… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/lift_franka_franka_umi_tacticle_scripted.pick_place_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 4092,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/pick_place_franka_allegro.reach_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for reach with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_350_reach_franka_allegro_success_0.98.pth; 200 successes in 313 attempts, success rate 0.64; env seed = base seed + attempt index). All episodes are successful, recorded… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/reach_franka_allegro_mg.lift_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for lift with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_400_lift_franka_allegro_success_0.88.pth; 200 successes in 248 attempts, success rate 0.81; env seed = base seed + attempt index). All episodes are successful, recorded at… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/lift_franka_allegro_mg.reach_franka_allegroThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 50,
"total_frames": 1760,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 20,
"splits": {
"train": "0:50"
},
"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/dexsuite/reach_franka_allegro.pick_place_pan_franka_allegro_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for pick_place_pan with the Franka arm and the allegro end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_250_pick_place_pan_franka_allegro_success_0.78.pth; 200 successes in 291 attempts, success rate 0.69; env seed = base seed + attempt index). All episodes are… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/pick_place_pan_franka_allegro_mg.reach_franka_robotiq_mgThis dataset was created using LeRobot.
Dataset Description
Machine-generated DexSuite data for reach with the Franka arm and the robotiq end-effector: 200 successful rollouts of the paper's low-dimensional (state) diffusion policy checkpoint trained on the 50 human demonstrations (checkpoint model_epoch_100_reach_franka_robotiq_success_1.0.pth; 200 successes in 230 attempts, success rate 0.87; env seed = base seed + attempt index). All episodes are successful, recorded at… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/reach_franka_robotiq_mg.reach_franka_franka_umi_tacticle_scriptedThis dataset was created using LeRobot.
Dataset Description
Scripted-expert DexSuite data for reach with the Franka arm and the UMI tactile parallel gripper: 200 successful episodes from the waypoint planner with randomized object initialization, including dense fingertip tactile readings (Flexitac simulation). The wrist camera looks at the flat face of the fingers (2026-09-16 mount). All episodes are successful, recorded at 20 Hz with a static front camera and a wrist… See the full description on the dataset page: https://huggingface.co/datasets/dexsuite/reach_franka_franka_umi_tacticle_scripted.