Research Tools: code, datasets, & models

DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective. 10x Larger Models 5x Faster Training Minimal Code Change DeepSpeed can train DL models with over a hundred billion parameters…

ElectionGuard is an open source software development kit (SDK) that makes voting more secure, transparent and accessible. Announced on at the Build developer conference (opens in new tab), ElectionGuard enables end-to-end verification of elections as well as support…

SynapseML is an ecosystem of tools aimed towards expanding the distributed computing framework Apache Spark (opens in new tab) in several new directions. SynapseML adds many deep learning and data science tools to the Spark…

Exterior acoustic scattering data suitable for machine learning. From ICASSP 2020 paper “Fast Acoustic Scattering using Convolutional Neural Networks”.

The source code for the Microsoft Embedded Social cloud service that runs on Azure is now available on GitHub!

Project Verona is a research programming language to explore the concept of concurrent ownership. We are providing a new concurrency model that seamlessly integrates ownership. This research project is at an early stage and is…

The Interpret Community builds on Interpret (opens in new tab), an open source python package from Microsoft Research for training interpretable models and helping to explain blackbox systems, by adding additional extensions from the community…

A hybrid edge-cloud live video analytics system to empower everyone to build practical real-world video analytics applications with cutting edge machine learning algorithms.

This code is based on the paper Icebreaker: Element-wise Active Information Acquisition with Bayesian Deep Latent Gaussian Model (opens in new tab). This repo is divided into two parts: PA-BELGAM: Experiments with UCI prediction alone (For…

This repository contains the source code and trained model for a large-scale pretrained dialogue response generation model. The human evaluation results (opens in new tab) indicate that the response generated from DialoGPT is comparable to human…