# kwai/DouZero

[ICML 2021] DouZero: Mastering DouDizhu with Self-Play Deep Reinforcement Learning | 斗地主AI

Repository: https://github.com/kwai/DouZero
Canonical: https://ross.abutalabs.com/products/douzero
Homepage: https://douzero.org/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: reinforcement-learning, doudizhu, game-ai, poker
Last push: 2024-06-26T23:02:16+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1919, "days_push": 798, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4652, forks 649 (observed 2026-08-28T04:08:55.639845+00:00)

## What it is
DouZero is a deep reinforcement learning framework that masters the Chinese card game DouDizhu through self-play, published at ICML 2021 by Kwai. It provides pretrained models and training code for building DouDizhu AI agents.

## Use cases
- train a dou dizhu card game AI
- learn self-play deep reinforcement learning
- build a poker game bot
- research imperfect-information game agents
- play against a strong dou dizhu AI
- benchmark card game AI algorithms

## When to choose
- you want a proven, research-grade DouDizhu AI with pretrained models
- you need a reference implementation of self-play RL for imperfect-information games
- you want to extend or study game AI based on an ICML-published framework

## When to avoid
- you need a general-purpose game AI framework for arbitrary games (see RLCard instead)
- you need active feature development or commercial support
- your game is not DouDizhu or a similar shedding-type card game

## Facets
- artifact type: library
- maturity: maintenance
- function: reinforcement-learning, machine-learning, game-engine
- domain: reinforcement-learning, artificial-intelligence
- platform: python, cross-platform
- tags: doudizhu, self-play, game-ai, poker, deep-learning, card-games, game-development

## Member repositories
- kwai/DouZero (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:55.639845+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:19:29.884105+00:00, confidence not recorded.
  - readme: https://github.com/kwai/DouZero (fetched 2026-08-28T04:08:55.639845+00:00, sha 2ff69b0f8adb)
  - homepage: https://douzero.org/ (fetched 2026-08-29T09:04:02.181022+00:00, sha 6eb2609854d6)
  - registry_pypi: https://pypi.org/pypi/douzero/json (fetched 2026-08-29T09:04:02.190341+00:00, sha b9a2129b41c5)
- Data as of 2026-08-30T08:39:29.467469+00:00.