Offline Clash Royale AI agent research: native simulation, imitation learning, PPO, and interactive evaluation. 离线对局模拟、模仿学习、强化学习与交互评测。

docs of miscellaneous things

Reference implementation and examples of the CuTe Layout representation and algebra.

Tile-Based Runtime for Ultra-Low-Latency LLM Inference

A curated resource list for learning AI performance engineering, from GPU fundamentals to production inference.

CUDA 算子手撕与面试指南

BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.

Nano vLLM

Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels

Learning TileLang with 10 puzzles!

基于深度强化学习的开源自动因子工厂。

Python raytracer

Position Based Fluids [SIGGRAPH 2013] in C++

High-performance moving least squares material point method (MLS-MPM) solver. (ACM Transactions on Graphics, SIGGRAPH 2018)

A refreshingly simple data-driven game engine built in Rust

A tiny framework for AI training and inference. My homework for the course "programming in AI".