Skip to content
Robotics

ACT Policy for robotics: Robotic arm control

2 min read

This model is a trained control policy for the so_follower robotic arm, created using the LeRobot library. The model solves the task of moving a red cube into a blue zone using the ACT method (Action Chunking with Transformers—an imitation learning method that predicts short action sequences instead of individual steps). It is trained on a dataset of 100 episodes and takes the robot's state and images from two cameras as input, outputting control actions. It is useful for robotics researchers and developers using the LeRobot ecosystem.

Key features

  • ACT architecture: Imitation learning with action chunking prediction for high precision.
  • Multimodal input: Processes state data and images from two cameras (bbb and aa).
  • LeRobot integration: Fully compatible with LeRobot version 0.7.1 training and inference tools.
  • Specific task: Trained specifically for the scenario of moving a red object to a target zone.

What you can do

  • Run inference: Test the policy on a real so_follower robot using the rollout command.
  • Explore the dataset: Review the demonstration training data on Hugging Face.
  • Fine-tune the model: Use the configuration to train your own robot on custom demonstrations.

Example

Running the policy on your robot:

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=DaniStr/act_2026-11-22_02-58-56 \
  --task="move the red box to the blue zone" \
  --duration=60
Read moreDetails
Method: Action Chunking with Transformers (ACT)
Dataset: DaniStr/red_box_merged (100 episodes, 38731 frames at 30 FPS)
Training parameters: 100000 steps, batch size 16, optimizer adamw, learning rate 1e-05
LeRobot version: 0.7.1
Discuss Telegram 0

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.

    # DaniStr/act_2026-11-22_02-58-56
    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 DaniStr/act_2026-11-22_02-58-56 --local-dir ~/models/act_2026-11-22_02-58-56
    # 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_2026-11-22_02-58-56 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

DaniStr/act_2026-11-22_02-58-56
Author
DaniStr
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments