pull

Download policy weights or environment Docker images.

Synopsis

maple pull policy NAME [OPTIONS]
maple pull env NAME [OPTIONS]

Pull Policy

Download policy weights from Hugging Face:

maple pull policy NAME [OPTIONS]

Arguments

NAME

Policy specification (e.g., openvla:7b, smolvla:libero)

Options

--port INTEGER

Daemon port to connect to (default: from config, typically 8000)

Examples

# Pull default variant
maple pull policy openvla

# Pull specific variant
maple pull policy openvla:7b

# Pull SmolVLA
maple pull policy smolvla:libero

Notes

  • Weights are stored in ~/.maple/models/

  • Download progress is shown in the daemon logs

  • Subsequent pulls use cached weights

Pull Environment

Pull environment Docker image:

maple pull env NAME [OPTIONS]

Arguments

NAME

Environment name (e.g., libero, simplerenv)

Options

--port INTEGER

Daemon port to connect to (default: from config, typically 8000)

Examples

# Pull LIBERO
maple pull env libero

# Pull SimplerEnv
maple pull env simplerenv

Notes

  • Requires Docker to be installed and running

  • Images are pulled from Docker Hub or built locally

  • Build images manually with: docker build -t maplerobotics/libero:latest docker/libero/