Microsoft Research: ResNet and BitNet Model Research

Microsoft’s research division, founded in 1991, develops methods and releases research artifacts, including ResNet and the downloadable BitNet ternary-weight language model.

kindResearch lab
founded1991
part ofMicrosoft

Updated

Research methods and released artifacts

Microsoft Research is Microsoft's research division, established in 1991. It works with academic collaborators and Microsoft product teams, so its outputs include papers, research software and model artifacts as well as work that feeds products. Its model releases give researchers artifacts to run as well as methods to read.

The ResNet paper introduced residual learning for deep neural networks, using shortcut connections so layers learn residual functions. The authors reported experiments with substantially deeper networks, making the optimization problem itself a central part of the contribution.

The division's BitNet b1.58 2B model is a downloadable language-model artifact trained on four trillion tokens and listed under the MIT license. Its ternary weights take three values; the 1.58-bit description concerns weight representation, rather than every activation and computation running at that precision.

Sources

  1. [1]
    About Microsoft Research

    Microsoft Research · retrieved Oct 23, 2026

  2. [2]
  3. [3]
    microsoft/bitnet-b1.58-2B-4T

    Microsoft Research · retrieved Oct 23, 2026

Microsoft Research history

2025

2015

■
ResNet makes depth trainable

Microsoft Research let each block learn a residual instead of a whole mapping, and networks that had been getting worse past twenty layers trained cleanly at a hundred and fifty. The skip connection is now in every transformer on this page.

arXiv
Research