Skip to content
Robotics

chunbae27/dp_pick_cube

diffusion policy for robot arms

2 min readupdated

This checkpoint implements Diffusion Policy (a robot control method using diffusion processes that turns visual data and state into smooth actions). The model was trained using the LeRobot library (a machine learning tool for real-world robotics) for the task of moving a cube to the right. It analyzes frames from two cameras (overhead and wrist-mounted) and the current robot state. The profile suits researchers and engineers working with so_follower type robots.

Key features

  • Diffusion Policy base: processing visuomotor control (vision and movement-based control) as a generation process for smooth multi-step trajectories.
  • Two cameras: using streams from the top-down and robot wrist views.
  • LeRobot integration: full compatibility with the Hugging Face robotics ecosystem.
  • Training dataset: trained on 50 episodes (chunbae27/pick-cube) totaling 18,472 frames.

What you can do

  • Run on a robot: apply the policy on a physical so_follower manipulator to automatically pick up and move a cube to the right.
  • Fine-tune the model: use the dataset repository for experiments with simulations or real hardware via train scripts.

Example

Running the ready inference policy (the process of applying a trained model) on a robot using the LeRobot command-line utility:

lerobot-rollout \
  --strategy.type=base \
  --robot.type=so_follower \
  --robot.port= \
  --robot.cameras="{ : {type: opencv, index_or_path: , width: 640, height: 480, fps: 30}, : {type: opencv, index_or_path: , width: 640, height: 480, fps: 30}}" \
  --policy.path=chunbae27/dp_pick_cube \
  --task="Pick up the cube and place it on the right" \
  --duration=60
Read moreDetails
• License: apache-2.0
• Training steps: 200,000 at batch size 32
• Optimizer: adam with learning rate 0.0001
• Hardware requirements: so_follower robot, two OpenCV cameras
Discuss Telegram 0 21

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 1.0 GB · ready files1.6 GB
  • BF16weights 0.5 GB1.1 GB
  • FP8 / INT8recommendedweights 0.3 GB0.8 GB
  • 4-bitweights 0.1 GB0.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F32 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # chunbae27/dp_pick_cube
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download chunbae27/dp_pick_cube --local-dir ~/models/dp_pick_cube
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to dp_pick_cube
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of dp_pick_cube (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

chunbae27/dp_pick_cube
Author
chunbae27
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

210

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments