# 王者荣耀AI (Reinforcement Learning King of Glory AI)

用基于策略梯度得强化学习方法训练AI玩王者荣耀

Repository: https://github.com/FengQuanLi/WZCQ
Canonical: https://ross.abutalabs.com/products/ai-reinforcement-learning-king-of-glory-ai
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2021-11-16T09:31:23+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1886, "days_push": 1751, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1829, forks 411 (observed 2026-08-28T04:05:41.721620+00:00)

## What it is
A PyTorch-based AI that plays the mobile game Honor of Kings (王者荣耀), built with a Resnet101 vision encoder plus a GPT/Transformer decoder, trained on recorded gameplay data. The original ResnetGPT project is no longer updated and has been superseded by a reinforcement-learning version (WZCQ) by the same author.

## Use cases
- train an AI to play Honor of Kings from screen captures
- learn how to combine ResNet and GPT for game-playing agents
- collect training data from Android gameplay via scrcpy and minitouch
- experiment with imitation learning on mobile game footage
- study a reinforcement learning replacement for a behavior-cloned game AI

## When to choose
- you want a hands-on example of vision-based game AI in PyTorch
- you need a reference for capturing Android game screens and injecting touch input
- you are learning how ResNet encoders and Transformer decoders fit together for sequential decision-making

## When to avoid
- you need a production-ready or actively maintained game bot
- your device runs Android 10 or later (minitouch is unsupported)
- you expect competitive-level play - the model was trained on only ~100 matches and performs poorly
- you need a permissively licensed codebase - the project has no license

## Facets
- artifact type: learning-resource
- maturity: abandoned
- function: machine-learning, deep-learning, computer-vision, reinforcement-learning, image-processing
- domain: machine-learning, deep-learning, gaming-tools, computer-vision, reinforcement-learning
- platform: windows, python
- tags: game-ai, honor-of-kings, resnet, gpt, pytorch, behavioral-cloning, scrcpy, minitouch, imitation-learning, android

## Member repositories
- FengQuanLi/WZCQ (main) score 32
- FengQuanLi/ResnetGPT (mirror) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:41.721620+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 2026-08-30T07:31:31.054150+00:00, confidence not recorded.
  - readme: https://github.com/FengQuanLi/WZCQ (fetched 2026-08-28T04:05:41.721620+00:00, sha 4427cd9fdc22)
- Data as of 2026-08-30T08:39:29.467469+00:00.