This project provides a packaged and slightly modified version of the Arcade Learning Environment (ALE) Python interface, primarily for Atari 2600 games. It targets reinforcement learning researchers and developers needing a stable Python interface to the ALE.

How It Works

The package wraps the C++ Arcade Learning Environment, offering a Python API for interacting with Atari games. This approach simplifies integration into Python-based RL frameworks by abstracting the underlying C++ implementation.

Quick Start & Requirements

• Install via pip: pip install atari-py
• Requires Python 3.5-3.7.
• ROMs must be manually downloaded and imported using python -m atari_py.import_roms <path_to_roms_folder>.
• Installation from source requires git, cmake, zlib1g, and zlib1g-dev (on Linux).

Highlighted Details

• Includes non-commercial ROMs for Tetris26 for testing.
• Supports Linux and macOS with pre-built binaries.

Maintenance & Community

This project is deprecated and will not receive bug fixes or updates. Users are directed to use the official ale-py package, which is backward compatible.

Licensing & Compatibility

The license is not explicitly stated in the README, but the inclusion of ROMs suggests potential licensing considerations for commercial use. Compatibility is limited to Python 3.5-3.7.

Limitations & Caveats

The project is fully deprecated, meaning no future support or bug fixes will be provided. Windows support is not officially guaranteed, with only unofficial binaries available.