Chongyang Wang

Today, together with our launch partners, we’re proud to introduce Flower Hub -- an app hub for publishing, discovering, and running Flower apps across heterogeneous environments. Flower Hub enables developers to focus on what matters most: building decentralized and federated AI applications. The underlying infrastructure -- from distribution to execution, across both simulation and real-world deployments -- is handled by Flower. This launch represents a major step forward for collaborative AI. It marks the transition from a world of isolated projects to an open platform where a community can share, reuse, and build trusted apps together. The era of collaborative AI starts now. Flower Hub Launch Partners: Mozilla AI (@MozillaAI), Owkin (@Owkin_AI), Sony AI (@SonyAI_global), BloodCounts! (@blood_counts), Fraunhofer IMS (@FraunhoferIMS), Gachon University, SYNTHEMA (@SYNTHEMA_EU), the University of Cambridge, the University of Melbourne, and the Vector Institute (@VectorInstitute)

Congrats to #Tsinghua University's Pervasive HCI Lab @thuhci & West China Hospital's @IHuaXi Rehab Medicine Center for winning the #IMWUT Distinguished Paper Award at #UbiComp2025 for "UbiPhysio"! 🏆Their innovative study combines biomech with LLMs to deliver home rehab guidance.

"The first AGI will be Federated" -- come hear my keynote at the @AcmSensys conference this Wednesday morning. I'll explain why when we finally achieve AGI, and flick the switch to bring it online, it will have been trained using some variety of federated AI. And not the training methods of today, DDP and the like, that are limited to being deployed in single data centers -- which are themselves fast reaching the limits of their own scalability as we struggle to co-locate and power north of 100k GPUs in one physical place. In support of my talk, here is a preview of a few results coming in hot from a new arxiv paper produced by @flwrlabs and @CaMLSys (@Cambridge_Uni) that will go live also Wednesday. We show our unique approach and system (aka Photon built on Flower) produces LLMs that are both better (by perplexity) and faster to train than DDP (when network speeds are equal, as measured up to 7B). Furthermore, we show this can scale effectively as we add more weakly connected GPUs -- while also being more efficient (eg better wall clock training times) than any existing known methods for training under decentralized/widely-distributed settings. Folks, it's time to train different. Larger and more complex LLM settings are underway. Link to arxiv Wednesday.

🌐 Join the 2024 ACSP Summer Program in HCI and Ubiquitous Computing🌟 We're thrilled to announce that applications are now open for the Access Computing Summer Program (ACSP) mp.weixin.qq.com/s/AVSiif6D_Yl8…. @thuhci @mvrjustid #CHI2024 #CHI #SIGHCI #Ubicomp

We prepared a special trifold booklet with our presentation papers & session info, and iconic works. 📚 Didn't get one? No worries! Check out below the online version or find a @thuhci member on-site! 💻🔍 #CHI2024

🎉 Hello X! We're thrilled to launch the official account for Tsinghua HCI Lab @Tsinghua_Uni, Dept. of Computer Science & Technology @thudcst. 😎🚀

It's today! 🌼 Flower around the world! 🤯 is happening 📺 Live stream link: buff.ly/3L2KUHe ⏰ Six sessions within a 24 hour period 🗣️ Presentations in five languages

🌼 Flower around the 🌐World is powering ahead. 🇰🇷Korean finished about an hour ago (thank you @YoungDKwon1!). Now in the middle of the 🇨🇳chinese session (courtesy of Pan Heng!) Here is a behind-the-scenes-look from the zoom war room; join the fun now at: flower.dev/around-the-wor…

Today @flwrlabs launched on @hackernews! Thanks for all the great feedback from the community. We value all the discussion, and look forward to incorporating it into future versions of Flower. @hackernewsrobot