先进编译实验室的个人主页

A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.

compiler learning resources collect.

面向多平台编译优化的深度学习中间表示

A torch compile backend for multi-targets

Goal: Enable awesome tooling for Bazel users of the C language family.

OLMoE: Open Mixture-of-Experts Language Models

Curated collection of papers in machine learning systems

LLM inference in C/C++

BladeDISC is an end-to-end DynamIc Shape Compiler project for machine learning workloads.

LLM101n: Let's build a Storyteller

Machine learning compiler based on MLIR for Sophgo TPU.

Implementation of Reliable Feature-Line Driven Quad-Remeshing

QuadriFlow: A Scalable and Robust Method for Quadrangulation

Development repository for the Triton-Linalg conversion

FlagGems is an operator library for large language models implemented in the Triton Language.

A collection of memory efficient attention operators implemented in the Triton language.

jax-triton contains integrations between JAX and OpenAI Triton

Intel® Extension for MLIR. A staging ground for MLIR dialects and tools for Intel devices using the MLIR toolchain.

TPP experimentation on MLIR for linear algebra

Power efficient dashboard for Kindle 4 NT devices

LLM training in simple, raw C/CUDA

The Modular Platform (includes MAX & Mojo)

OpenAI Triton backend for Intel® GPUs

An open-source efficient deep learning framework/compiler, written in python.

MLIR For Beginners tutorial

Trixi.jl: Adaptive high-order numerical simulations of conservation laws in Julia

High Order Hex-Quad Mesh (HOHQMesh) package to automatically generate all-quadrilateral meshes with high order boundary information.