Today’s release brings mega improvements to our already impressive Machine Learning models, which ship locally with Pieces for Developers. Check out additional image enrichment, better auto-generated tags and descriptions, and more accurate search results in the latest version of Pieces.
Using Pieces OCR to extract code from images and screenshots is one of the biggest time-savers that Pieces offers. In this release, we’re upgrading those snippets with all of the auto-generated metadata as our code-only snippets.
When you add an image to the Pieces desktop app, we’ll quickly generate titles, descriptions, tags, and more.
Our local ML models are fast, secure, and super performant. We’ve improved them even further in this release.
When using only local models, your snippets will now have even more relevant, auto-generated tags and more detailed auto-generated descriptions. We’re also introducing local Suggested Searches generation, which introduces even more metadata to your snippets. Suggested Searches help you to find more information about your snippets in Global Search and the Pieces chatbot (coming soon!).
You can toggle between local ML models and blended ML models in the settings menu of the Pieces desktop app.
While there are three ways to search for your snippets in the Pieces desktop app, Neural Code Search (NCS) is perhaps our favorite. NCS allows you to use natural language input to search for your code; this release includes improvements to this search mode so that you’ll get better results.
You can swap between search modes in the search bar of the Pieces desktop app.
This release also includes a long list of bug fixes and performance improvements to make your experience with Pieces as seamless as possible.
The bugs we crushed include:
Do you love Pieces? Stop sending us carrier pigeons 🐦 and join our Discord Server to chat with our team, power users, get support, and more. 🤝
As always, if you run into issues or have feedback, please fill out this quick form or email us at firstname.lastname@example.org and we’ll be in touch as soon as possible!