Skip to content

Installation

This guide will help you install the Proxy Structuring Engine (PSE) and set up your environment for development.

Standard Installation

We recommend using uv to install PSE: 

uv pip install pse

(This installs the pse Python library and its required dependency pse-core, which contains the pre-compiled C++ engine.)

You can also install PSE using pip:

pip install pse

Development Installation

For development or to access the latest features, you can install from source:

# Clone the repository
git clone https://github.com/TheProxyCompany/proxy-structuring-engine.git
cd proxy-structuring-engine

# Install in development mode
pip install -e ".[dev]"

Framework-Specific Installation

PSE works with multiple ML frameworks. Install the one(s) you plan to use:

PyTorch

pip install pse[torch]

# Or for a specific version
pip install torch pse

MLX (Apple Silicon)

pip install pse[mlx]

# Or for a specific version
pip install mlx pse

TensorFlow

pip install pse[tensorflow]

# Or for a specific version
pip install tensorflow pse

JAX

pip install pse[jax]

# Or for a specific version
pip install jax jaxlib pse

System Requirements

  • Python: 3.10 or higher
  • Operating Systems:
  • Linux (Ubuntu 20.04+, Debian 11+, etc.)
  • macOS (11.0+)
  • Windows 10/11
  • Hardware:
  • Any system capable of running your selected LLM (the PSE only works with local models)

Getting Help

If you continue to experience issues:

Next Steps

Now that you've installed PSE, proceed to the Quickstart guide to run your first example.