Robotics
LeRobot
Safetensors
multi_task_dit

Model Card for multi_task_dit

Multi-Task Diffusion Transformer (DiT) extends Diffusion Policy with a large Diffusion Transformer and text + vision conditioning for multi-task robot learning. It supports both diffusion and flow-matching objectives and reaches high dexterity with only ~450M parameters.

This policy has been trained and pushed to the Hub using LeRobot.

Learn how to train and run it in the LeRobot multi_task_dit guide, or browse the full documentation.


Model Details

  • License: [More Information Needed]
  • Robot type: rebot_b601_follower
  • Cameras: top_left, top_right, wrist

Inputs & Outputs

The policy consumes these observation features and produces these action features.

Inputs

Feature Type Shape
observation.images.top_left VISUAL (3, 480, 640)
observation.images.top_right VISUAL (3, 480, 640)
observation.images.wrist VISUAL (3, 480, 640)
observation.state STATE (10,)

Outputs

Feature Type Shape
action ACTION (10,)

Training Dataset

  • Repository: SieloRobotics/rebot_machine_tending_II_ee
  • Episodes: 220
  • Frames: 525493
  • Frame rate: 30 FPS
  • Task(s): "Tend the 3D printer: load the plastic triangular part into the printer and unload it into the black-marked plastic bin, then load the metal bracket part into the printer and unload it into the red-marked metal bin, opening and closing the printer door around each load and unload.", "Tend the 3D printer: load the metal bracket part into the printer and unload it into the red-marked metal bin, then load the plastic triangular part into the printer and unload it into the black-marked plastic bin, opening and closing the printer door around each load and unload."

Training Configuration

Setting Value
Training steps 36210
Batch size 192
Optimizer adam
Learning rate 8.660254037844386e-05
Seed 1000
LeRobot version 0.7.1

How to Get Started with the Model

New to LeRobot? These guides cover the full workflow:

The short version to run and train this policy:

Run the policy on your robot

lerobot-rollout \
  --strategy.type=base \
  --robot.type=rebot_b601_follower \
  --robot.port=<your_robot_port> \
  --robot.cameras="{ <camera_1>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}, <camera_2>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}}" \
  --policy.path=SieloRobotics/dit_ee_machine_tending_II \
  --task="Tend the 3D printer: load the plastic triangular part into the printer and unload it into the black-marked plastic bin, then load the metal bracket part into the printer and unload it into the red-marked metal bin, opening and closing the printer door around each load and unload." \
  --duration=60

Replace the remaining <...> placeholders with your own values: --robot.port and the camera names/indices are specific to your machine, and the camera names must match the observation keys this policy was trained on.

When --strategy.type=base is used the script doesn't record the episodes. Skipping duration will make the policy run indefinitely. For more information look at rollout documentation.

Train your own policy

lerobot-train \
  --dataset.repo_id=${HF_USER}/<dataset> \
  --policy.type=multi_task_dit \
  --output_dir=outputs/train/<policy_repo_id> \
  --job_name=lerobot_training \
  --policy.device=cuda \
  --policy.repo_id=${HF_USER}/<policy_repo_id> \
  --trackio.enable=true

Writes checkpoints to outputs/train/<policy_repo_id>/checkpoints/.


Evaluation

No evaluation results have been provided for this policy yet.


Citation

If you use this policy, please cite the method linked in the description above, along with LeRobot:

@misc{cadene2024lerobot,
    author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascal, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},
    title = {LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch},
    howpublished = "\url{https://github.com/huggingface/lerobot}",
    year = {2024}
}
Downloads last month
2
Safetensors
Model size
0.4B params
Tensor type
F32
Β·
Video Preview
loading

Dataset used to train SieloRobotics/dit_ee_machine_tending_II

Paper for SieloRobotics/dit_ee_machine_tending_II