Skip to content
Robotics

act_pick_and_place_tape_final_chunk50

pick-and-place policy for the SO-Follower robot

2 min readupdated

A compact 0.05B-parameter robotics policy trained with the LeRobot framework by knight7024. The model tackles the pick_and_place_tape task: picking up an object and moving it to a target spot using sensor readings and dual camera feeds. It is built on the ACT architecture (Action Chunking with Transformers — an imitation learning method that predicts short action sequences all at once rather than single steps). Alongside this version, the author also released chunk10 and chunk100 variants. It is a solid fit for developers and hobbyists experimenting with so_follower robotic arms and the LeRobot library.

Key features

  • ACT imitation learning: predicts action chunks from teleoperation data for smoother manipulation.
  • Multimodal inputs: takes a 6-element robot state vector observation.state of shape (6,) alongside video feeds from two cameras — an overhead camera (top, 240×320) and a wrist-mounted camera (wrist, 480×640).
  • Precise control: outputs a 6-dimensional action vector to drive the robotic arm's actuators.
  • Flexible planning horizons: alongside the chunk50 version, 10- and 100-step checkpoints are available to compare robot behavior.

What you can do

  • Run on real hardware: deploy the policy on an so_follower robotic arm equipped with two cameras to test basic pick-and-place tasks.
  • Compare chunk lengths: benchmark chunk50 against the sibling checkpoints (chunk10 and chunk100) on the same task to evaluate motion smoothness.
  • Use as a starting point: build on this setup and the knight7024/pick_and_place_tape_final dataset for fine-tuning on your own tasks in LeRobot 0.6.0.

Example

lerobot-rollout \
  --strategy.type=base \
  --robot.type=so_follower \
  --robot.port= \
  --robot.cameras="{ top: {type: opencv, index_or_path: 0, width: 320, height: 240, fps: 30}, wrist: {type: opencv, index_or_path: 1, width: 640, height: 480, fps: 30}}" \
  --policy.path=knight7024/act_pick_and_place_tape_final_chunk50 \
  --task="pick_and_place_tape" \
  --duration=60
Read moreDetails
Trained on the knight7024/pick_and_place_tape_final dataset: 50 episodes, 16,278 frames at 30 FPS.
Training setup: 100,000 steps, batch size 1, AdamW optimizer (lr 1e-05), seed 1000, LeRobot 0.6.0.
Limitations: real-robot evaluation results have not been published in the model card yet; Apache-2.0 license.
Discuss Telegram 0 24

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 0.2 GB · ready files0.7 GB
  • BF16weights 0.1 GB0.6 GB
  • FP8 / INT8recommendedweights 0.0 GB0.6 GB
  • 4-bitweights 0.0 GB0.5 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.

    # knight7024/act_pick_and_place_tape_final_chunk50
    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 knight7024/act_pick_and_place_tape_final_chunk50 --local-dir ~/models/act_pick_and_place_tape_final_chunk50
    # The model is in ~/models. How to run it is in the model card (link above).
  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 act_pick_and_place_tape_final_chunk50 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

knight7024/act_pick_and_place_tape_final_chunk50
Author
knight7024
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

240

Ask AI about this model

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

Comments