An agent-managed museum exhibit, built in Rust with Gajae-Code / LazyCodex — developed and maintained with no human intervention.

Pure Triton kernels for Qwen3.5-27B inference on NVIDIA B200

PTX ISA 9.1 documentation converted to searchable markdown. Includes Claude Code skill for CUDA development.

Checkpoint-engine is a simple middleware to update model weights in LLM inference engines

Materials for learning SGLang

Benchmarking code for running quantized kernels from vLLM and other libraries

This is a series of GPU optimization topics. Here we will introduce how to optimize the CUDA kernel in detail. I will introduce several basic kernel optimizations, including: elementwise, reduce, s…

This is a series of GPU optimization topics. Here we will introduce how to optimize the CUDA kernel in detail. I will introduce several basic kernel optimizations, including: elementwise, reduce, s…

模型加速/模型压缩（已完成所有Lab）

Build and publish crates with pyo3, cffi and uniffi bindings as well as rust binaries as python packages

A library to analyze PyTorch traces.

Strongly-typed LLM Function Calling examples, run on OpenAI, Ollama, Mistral and others.

[ACL 2025] Quaff: Quantized Parameter-Efficient Fine-Tuning under Outlier Spatial Stability Hypothesis

Artificial Neural Engine Machine Learning Library

SoTA open-source TTS

AI Tensor Engine for ROCm

相当不错的图书，例如《数学之美》《提问的智慧》《软件工程可靠性》《时间简史》《毛泽东选集【全四卷】》《浪潮之巅》《金字塔原理》《TCP/IP卷一/卷二/卷三》《[荐]深入浅出设计模式》等；一些大的上传受限制的文件《图解TCP_IP_第5版》等在README

Hypernetworks that adapt LLMs for specific benchmark tasks using only textual task description as the input

LMCache: Supercharge Your LLM with the Fastest KV Cache Layer

This is the repository of DEER, a Dynamic Early Exit in Reasoning method for Large Reasoning Language Models.

The Rust primer for beginners. We need native English speaker help us modify the translation.

QuickReduce is a performant all-reduce library designed for AMD ROCm that supports inline compression.

Pluggable in-process caching engine to build and scale high performance services

Drogon: A C++14/17/20 based HTTP web application framework running on Linux/macOS/Unix/Windows

A fast communication-overlapping library for tensor/expert parallelism on GPUs.

Learning materials for Stanford CS149 : Parallel Computing

📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉