OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Reversi Alpha Zero

Reversi Alpha Zero is an open-source Python implementation that trains Reversi-playing models with AlphaGo Zero-style reinforcement learning. Its self-play worker generates training data, its optimizer produces new model generations, and its evaluator can compare those models with the current BestModel.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Reversi Alpha Zero — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want a code-level Reversi reinforcement-learning workflow with self-play, model optimization and optional evaluation, or want to try the documented GUI against a saved model.

Who it is for

  • Developers studying or adapting reinforcement-learning methods for Reversi.
  • Practitioners who want to train models through self-play and inspect generated data and model files.
  • People who want to play Reversi against a saved model through the documented wxPython GUI.

Documented uses

  • Generate Reversi training data through self-play.
  • Train next-generation models and evaluate them against a BestModel.
  • Play a Reversi game against BestModel using the GUI.
  • Inspect saved models, generated play data and training logs.
Before you choose

Requirements & limits

The project is public on GitHub and lists an MIT license. Its README documents a local Python setup; it does not describe a hosted service.

  • The stated environment is Python 3.6.3, TensorFlow 1.3.0 or later and Keras 2.0.8 or later; these are older dependencies.
  • The README says CPU TensorFlow is much slower for training, while CPU is adequate for the play_gui workflow.
  • Setting config.play.use_newest_next_generation_model changes the workflow: when enabled, the evaluator worker is unused.
  • The repository stores best and next-generation models, generated play data and logs in separate data and log paths.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare the Python and TensorFlow environment

Install the project’s listed dependencies and configure Keras to use TensorFlow.

Before you start

  • A local copy of the repository.
  • Python 3.6.3 is the README’s stated environment.
  • The project’s requirements.txt, or the documented Anaconda environment files.

Steps

  1. Install dependencies with the command below, or follow the README’s Anaconda environment setup.

    Documented command
    pip install -r requirements.txt

  2. For GPU use, install tensorflow-gpu as described in the README.

  3. Create a .env file containing KERAS_BACKEND=tensorflow.

Expected outcome

The dependency setup and TensorFlow backend configuration described by the repository are in place.

Guide limitations

  • The README specifies older Python and TensorFlow versions; it does not establish compatibility with current releases.
  • The guide covers setup only, not execution or successful installation.
Documentation 1 ↗
Source-based practical guide

Play against a downloaded challenge model

Use a named challenge model with the project’s wxPython Reversi GUI.

Before you start

  • A local copy of the repository and its documented Python dependencies.
  • The README’s config/ch5.yml configuration and wxPython GUI dependencies.

Steps

  1. Remove the data/model/next_generation/ directory, as instructed for the strongest-model workflow.

  2. Run the command below to request the challenge 5 model.

    Documented command
    sh ./download_model.sh 5

  3. Run the command below to start the documented GUI command.

    Documented command
    python src/reversi_zero/run.py play_gui -c config/ch5.yml

Expected outcome

The README describes playing Reversi against the challenge 5 model using the GUI.

Guide limitations

  • The README identifies challenge 5 as its strongest model at the time of writing; this is not a current performance comparison.
  • The commands and model download have not been execution-tested here.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
mokemokechicken/reversi-alpha-zero · Official homepage title · retained in its source language
How we review sourcesSuggest a correction