Google LLC is making some improvements to its AI Hub, adding a revamped homepage and more collaboration tools for data scientists and artificial intelligence teams.

Launched in beta in April, AI Hub is a cloud-hosted repository of plug-and-play AI components that can be used to create various types of AI models. The components include end-to-end AI pipelines and out-of-the-box algorithms. There are also tools to help AI teams deploy the technologies they build on Google Cloud and hybrid cloud environments.

In a blog post, Google product managers Till Pieper and Nate Keating said today’s updates are all about making collaboration among teams even easier. To that end, Google is introducing a new homepage (pictured) for AI Hub that makes it easier for users to access their most popular, and most recently shared assets.

The new homepage also provides access to more advanced sharing tools. Now, it’s possible to share assets including notebooks, trained machine learning models and Kubeflow pipelines, which are used to build and deploy portable, scalable ML workflows based on Docker containers in a more flexible way. For example, assets can be shared with specific individuals or groups of users or even an entire organization.

“All it takes is simply adding individual collaborators or groups by their email addresses, and giving them editor or viewer permissions,” Pieper and Keating wrote. “’Viewers’ will still be able to fork the asset you share by downloading or opening a copy, but they won’t be able to edit or change the version shared on AI Hub.”

Assets can now be shared directly onto social media too. Simply click on a social media icon to copy the URL, then paste it at your desired location to share it. Google says that could be useful in cases of public AI projects that are looking for outside expertise to collaborate on them.

AI Hub users will be able to find their most popular AI assets more easily. Google is enabling this by making it possible to “favorite” certain notebooks and models.

The final update relates to more content. Google is adding more than 70 “new, cutting-edge assets” it has created for users to build upon, including a TensorRT-optimized BERT notebook that provides examples on how to use the popular BERT natural language understanding model.

There’s also a new Pulto7 Kubeflow pipeline for time series forecasting, designed to be used for business planning tasks such as inventory optimization, revenue forecasting and store traffic prediction.

Images: Google