Making large AI models cheaper, faster and more accessible

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

A GPipe implementation in PyTorch

飞桨大模型开发套件，提供大语言模型、跨模态大模型、生物计算大模型等领域的全流程开发工具链。

LiBai(李白): A Toolbox for Large-Scale Distributed Parallel Training

Slicing a PyTorch Tensor Into Parallel Shards

Easy Parallel Library (EPL) is a general and efficient deep learning framework for distributed model training.

A curated list of awesome projects and papers for distributed training or inference

Large scale 4D parallelism pre-training for 🤗 transformers in Mixture of Experts *(still work in progress)*

NAACL '24 (Best Demo Paper RunnerUp) / MlSys @ NeurIPS '23 - RedCoast: A Lightweight Tool to Automate Distributed Training and Inference

Distributed training (multi-node) of a Transformer model

Distributed training of DNNs • C++/MPI Proxies (GPT-2, GPT-3, CosmoFlow, DLRM)

SC23 Deep Learning at Scale Tutorial Material

WIP. Veloce is a low-code Ray-based parallelization library that makes machine learning computation novel, efficient, and heterogeneous.

Adaptive Tensor Parallelism for Foundation Models

PyTorch implementation of 3D U-Net with model parallel in 2GPU for large model

Fast and easy distributed model training examples.

Official implementation of DynPartition: Automatic Optimal Pipeline Parallelism of Dynamic Neural Networks over Heterogeneous GPU Systems for Inference Tasks

Serving distributed deep learning models with model parallel swapping.

Model parallelism for NN architectures with skip connections (eg. ResNets, UNets)