Train transformer language models with reinforcement learning.

CUTLASS and CuTe Examples

An unofficial cuda assembler, for all generations of SASS, hopefully ：）

A framework for few-shot evaluation of language models.

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

DeepEP: an efficient expert-parallel communication library

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

LeaderWorkerSet: An API for deploying a group of pods as a unit of replication

Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation

✔（已完结）最全面的 深度学习 笔记【土堆 Pytorch】【李沐 动手学深度学习】【吴恩达 深度学习】

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

My learning notes/codes for ML SYS.

A throughput-oriented high-performance serving framework for LLMs

Clean, minimal, accessible reproduction of DeepSeek R1-Zero

Optimize GEMM with tensorcore step by step

Samples for CUDA Developers which demonstrates features in CUDA Toolkit

A highly optimized LLM inference acceleration engine for Llama and its variants.

本项目旨在分享大模型相关技术原理以及实战经验（大模型工程化、大模型应用落地）

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

HLLM: Enhancing Sequential Recommendations via Hierarchical Large Language Models for Item and User Modeling

Material for gpu-mode lectures

搜索、推荐、广告、用增等工业界实践文章收集（来源：知乎、Datafuntalk、技术公众号）

DLRover: An Automatic Distributed Deep Learning System

A PyTorch Native LLM Training Framework

Efficient Triton Kernels for LLM Training

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…