pufferlib

A reinforcement learning based drone controller and simulator, backed by @PufferAI

Headless Warcraft III (Legacy TFT 1.29) under Wine for reinforcement learning: the real game stepped from Python, with PufferLib 5 training, behavior cloning and a training dashboard.

Deep Rust reimplementation of DeepSeek Elastic Compute (DSec) — sandbox infrastructure for agentic RL training at scale, with a PufferLib core port

Game Env for RL training/self-play/curriculum learning

RL UFO game

RL gym for training LLMs on financial text reasoning

Self-play RL for Bopl Battle: deterministic Rust reimplementation of the game sim, recurrent PPO on a PufferLib CUDA trainer, Bradley-Terry/Nash league, parity gates.

A neural network that plays Breakout perfectly, with weights written by Claude and no training.

Reproducible PufferLib training for 3D quadrotor navigation

Industry-standard development environments for: AI/ML Training, Reinforcement Learning, Scientific Computing, Biology & Genomics, and Structural Biology.