google-research

A repository of research code and datasets from Google. The README keeps its guidance short: two licenses, a trick for downloading single projects, and a note on shallow clones.

The README's first point is legal. All datasets in the repository are released under the CC BY 4.0 International license, and all source files under the Apache 2.0 license. Both license texts are linked. That single sentence is the main reuse guidance the README offers, and for a collection this size it does a lot of work. The primary language listed is Jupyter Notebook, with research.google as the homepage.

The README recommends using the GitHub editor to open projects. Change the URL from github.com to github.dev in the address bar, then right click the folder of interest in the left navigation panel and select download. That avoids cloning a large monorepo when only one project is needed. For pull request contributors, the README suggests a shallow clone, without history.

The repository is a stack of individual research projects rather than one application. Each project directory carries its own code and often its own documentation, and the opening text does not try to summarize the research topics. The browsing guidance is effectively the documentation: open the editor, right click a folder, download it. The README's silence on specifics is itself informative, because the folders are the content.

The README is nearly all licensing and logistics, which fits the repo. It is a shelf of research projects, and the main question for anyone using it is which license applies to what they took.

• Source 1 · Project repository
• Source 2 · README