Popular GitHub Repositories for Mlsys

AISystem 主要是指AI系统，包括AI芯片、AI编译器、AI推理和训练框架等AI全栈底层技术

? Awesome System for Machine Learning ⚡️ AI System Papers and Industry Practice. ⚡️ System for Machine Learning, LLM (Large Language Model), GenAI (Generative AI). ? OSDI, NSDI, SIGCOMM, SoCC, MLSys, etc. ?️ Llama3, Mistral, etc. ?‍? Video Tutorials.

[ICLR2025 Spotlight] SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models

Quantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.

Distributed RL System for LLM Reasoning

ComfyUI plugin of Nunchaku

[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.

SpargeAttention: A training-free sparse attention that can accelerate any model inference.

A model compilation solution for various hardware

FedScale is a scalable and extensible open-source federated learning (FL) platform.

(MLSys' 21) An Acceleration System for Large-scare Unsupervised Heterogeneous Outlier Detection (Anomaly Detection)

The repository has collected a batch of noteworthy MLSys bloggers (Algorithms/Systems)

Machine Learning Framework for Operating Systems - Brings ML to Linux kernel

An acceleration library that supports arbitrary bit-width combinatorial quantization operations

A scalable & efficient active learning/data selection system for everyone.

[MLSys 2021] IOS: Inter-Operator Scheduler for CNN Acceleration

?FFPA: Extend FA-2 with Split-D for large headdim, 2x↑ vs SDPA.

A ChatGPT(GPT-3.5) & GPT-4 Workload Trace to Optimize LLM Serving Systems

FedGraphNN: A Federated Learning Platform for Graph Neural Networks with MLOps Support. The previous research version is accepted to ICLR'2021 - DPML and MLSys'21 - GNNSys workshops.

?FFPA(Split-D): Yet another Faster Flash Prefill Attention with O(1) GPU SRAM complexity for headdim > 256, ~2x↑?vs SDPA EA.