Agibot2UnitreeG1Retarget Dataset

Paper | Project Page | Code

Description

This dataset contains action retargeting data from Agibot to UnitreeG1 humanoid robot.

Dataset Size

• Total size: ~30GB
• Split into 7 parts (A2UG1_dataset.tar.gz.aa to A2UG1_dataset.tar.gz.ag)
• Each part is approximately 4GB

Usage

Method 1: Using Hugging Face Hub (Recommended)

pip install huggingface-hub

from huggingface_hub import snapshot_download

# Download the entire dataset
snapshot_download(
    repo_id="l2aggle/Agibot2UnitreeG1Retarget",
    repo_type="dataset",
    local_dir="./Agibot2UnitreeG1Retarget"
)

Method 2: Using Git with LFS

# Make sure git-lfs is installed
git lfs install

# Clone the repository (this will download LFS pointer files)
git clone https://huggingface.co/datasets/l2aggle/Agibot2UnitreeG1Retarget
cd Agibot2UnitreeG1Retarget

# Download the actual large files
git lfs pull

Method 3: Manual Download

Download individual parts through the Hugging Face web interface: https://huggingface.co/datasets/l2aggle/Agibot2UnitreeG1Retarget/tree/main

Extract Dataset

After downloading, extract the complete dataset:

# Combine and extract all parts
cat A2UG1_dataset.tar.gz.* | tar -xzf -

This will create the complete A2UG1_dataset folder with all original files.

File Structure

A2UG1_dataset/
├── [your dataset structure will be shown here after extraction]

Requirements

• At least 60GB free disk space (30GB for download + 30GB for extraction)
• For Method 1: Python 3.6+ with huggingface-hub package
• For Method 2: Git with Git LFS support
• tar utility (standard on Linux/Mac, available on Windows via WSL or Git Bash)

Installation Requirements

# For Method 1
pip install huggingface-hub

# For Method 2 (if git-lfs not installed)
# Ubuntu/Debian:
sudo apt install git-lfs
# macOS:
brew install git-lfs
# Windows: download from https://git-lfs.github.io/

License

Apache 2.0