DocArray v2

[JohannesMessner]

DocArray v2

We ❤️ DocArray just as much as you do.

And because we love it so much we have to tear it down and build it up again! 🚀🚀🚀

That's why today we are announcing our big push for DocArray v2, a complete rewrite of the library.

In short, our big focus is this: Supporting multi-modality natively and naturally.
To this end, DocArray will be re-built on top of a flexible, Pydantic based API.

You can see the the accompanying implementation branch and readme here, and our preliminary documentation here.

Further big improvements will come with regard to:

• Use inside of ML models
• Document Stores
• Use from different languages
• Integration with Pydantic, FastAPI, HuggingFace, Pytorch and other key building blocks of the Python ecosystem

To get all the details about our reasoning you can read our announcement blog post.
You can keep track with our progress on our public roadmap.
And, most importantly, share your thoughts, input and feedback here in the discussion section

We hope you're along for the ride!