linyiLYi/street-fighter-ai resource

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

6521 stars · 1376 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

An AI agent trained with deep reinforcement learning to beat the final boss of Street Fighter II: Special Champion Edition, making decisions purely from game screen RGB pixels. It is a Python project built on OpenAI Gym Retro and Stable-Baselines3, including training scripts, logs, and pretrained model weights.

Use cases

• train an AI to play Street Fighter II
• learn deep reinforcement learning with a game example
• use gym-retro with stable-baselines3
• train an agent from raw pixel values
• reproduce a game-playing RL agent
• study RL training curves and checkpoints

When to choose

• you want a hands-on, well-starred example of deep RL applied to a retro game
• you need a reference for combining gym-retro with Stable-Baselines3
• you want pretrained weights to inspect agent behavior at different training stages

When to avoid

• you need a general game-playing AI rather than one overfit to a specific boss fight
• you cannot legally obtain the Street Fighter II ROM
• you need production-grade RL infrastructure or multi-game support

Facets

learning-resource · maturity maintenance

reinforcement-learning machine-learning deep-learning game-engine simulation reinforcement-learning machine-learning artificial-intelligence python windows deep-reinforcement-learning gym-retro stable-baselines3 street-fighter game-ai tutorial-project game-development desktop

Member repositories

For agents

markdown · JSON · MCP: product_card(name="linyiLYi/street-fighter-ai")