Kedro helps create reproducible, maintainable, and modular data science and data engineering pipelines.
Maze
Maze provides an applied reinforcement learning framework for developing projects in Python.
About Maze
Maze is a Python framework for applied deep reinforcement learning, intended for real-world decision problems. Its stated goal is to cover the full development cycle of reinforcement learning applications, from simulation engineering through agent development, training, and deployment. It is installed with pip as maze-rl and is compatible with Python 3.9 to 3.10, with Python 3.10 recommended.
The framework includes a PyTorch-based Perception Module for designing policy and value networks, a set of built-in trainers, and support for imitation learning and policy fine-tuning. Configuration is handled with the Hydra configuration system. Existing Gym-compatible environments can be used. Environments such as Atari or Box2D may require additional binary dependencies to be installed manually.
The project describes its current release as preliminary and non-stable. Some interfaces are not yet settled, so breaking changes may occur before a first stable release.
Key features
- PyTorch-based Perception Module for designing policy and value networks
- Built-in trainers including A2C, PPO, Impala, SAC, and Evolution Strategies
- Support for dictionary action and observation spaces and multi-step auto-regressive training
- Environment structures for multi-step and multi-agent decision problems
- Imitation learning from teacher policies and policy fine-tuning workflows
- Hydra configuration system for managing application and experiment configurations
Best for: Python developers and researchers who build reinforcement learning applications for real-world, including industrial, decision problems.
Good for
- Covers multi-step and multi-agent environment structures
- Provides a range of built-in trainers and training workflows
Watch out for
- The current release is preliminary and non-stable, and breaking changes may occur before a stable release
- Compatible with Python 3.9 to 3.10 only
- Environments such as Atari or Box2D may require manual installation of additional binary dependencies
Maze at a glance
- Website
- github.com/enlite-ai/maze
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Not stated
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Maze alternatives
Tools and sites you could use instead, ranked by how closely they match.
The Datumbox framework is written in Java for developing machine learning and statistical applications.
Open source
24 alternatives
The ChefBoost Python framework supports decision trees such as ID3, C4.5, CART, CHAID, regression trees, and ensemble methods.
Open source
24 alternatives
TPOT uses genetic programming to optimize machine learning pipelines in Python.
Open source
24 alternatives
FLAML provides a fast Python library for automated machine learning and tuning.
Open source
24 alternatives
FlexML offers an easy to use and flexible automated machine learning library for Python.
Open source
24 alternativesPopular in AI Infrastructure & MLOps
Well-known listings from the same topic.
Guild AIguild.aiFlow AIflow-ai.comsystemprompt.iosystemprompt.ioKili Technologykili-technology.comAPI docsdocs.x.aiEyePop.aieyepop.ai
Maze FAQ
Which Python versions does Maze support?
Maze is compatible with Python 3.9 to 3.10, and the project recommends starting with Python 3.10.
Is the current Maze release stable?
No. The project describes it as a preliminary, non-stable release, and some interfaces have not settled, so breaking changes may occur before the first stable release.
Can Maze be used with existing Gym environments?
Yes. The project states that existing Gym-compatible environments can be used with Maze.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.