The Flower team is excited to announce the release of Flower 1.39 stable, packed with new updates!
Flower is a friendly framework for collaborative AI and data science. It makes novel approaches such as federated learning, federated evaluation, federated analytics, and fleet learning accessible to a wide audience of researchers and engineers.
Thanks to our contributors
We would like to give our special thanks to all the contributors who made the new version of Flower possible (in git shortlog order):
Charles Beauville, Daniel J. Beutel, Daniel Nata Nugraha, Dimitris Stripelis, Heng Pan, Javier, Riyasat Ohib, Taner Topal
What's new?
-
Read Notion users, pages, properties, blocks, meeting notes, and databases (#8245, #8252, #8255, #8266)
The Notion connector now exposes read actions for users, page metadata and complete property values, individual blocks and their children, AI meeting notes, and database metadata with associated data source references. The actions map directly to Notion API endpoints, return native responses, and support the relevant filtering, sorting, and pagination options.
-
Add federation-wide connectors (#8210, #8237, #8239, #8240, #8265)
Connector storage, OAuth sessions, APIs, and task metadata now carry federation scope, allowing members to connect, list, disconnect, and use connector-backed runs in federations. Federations can now keep multiple connections to the same provider, and select one connection of each provider per run.
-
Run sparse federated learning experiments with SSFL (#7900)
The new SSFL baseline implements static sparse federated learning with the Message API, client-side saliency scoring, global mask aggregation, and sample-weighted sparse FedAvg. It includes ResNet-18 experiments for CIFAR-10 and CIFAR-100, paper-scale configurations, communication accounting, and reproduction guidance.
-
Get faster agent responses (#8225, #8226, #8227, #8231, #8232, #8233, #8238)
-
General improvements (#8221, #8229, #8230, #8234, #8235, #8242, #8243, #8244, #8249, #8253, #8254, #8256, #8258, #8259, #8260, #8261, #8268, #8270)
As always, many parts of the Flower framework and quality infrastructure were improved and updated.