Sudarshan

Retired account. 📦

[nic lane @niclane7]

"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.

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

The next Jesus College Cambridge Computer Science Alumni Dinner is on 25th January 2025. The speaker will be Josh Bambrick on "AI driven drug design".
All computer science alumni and present members of the college can register here: https://t.co/yGBEXlRwN6

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

On Tuesday 29th the seminar of the Mobile and Wearable Health series is given by @aaqib_saeed on "Self-Supervised Learning, Unified Sensing & Beyond" at 4pm GMT in FW11 (and online): https://t.co/XDyfWNJ5SR Do come along.

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

Our Mobile and Wearable Health Seminar series is restarting in a couple of weeks with an amazing line up!
check it out here: https://t.co/CaaO3UIixq
Also some recordings from previous talks here: https://t.co/mzdxIG1dyR

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

We have opened an Assistant/Associate Professorship in Wearable Computing @Cambridge_CL @Cambridge_Uni Please share widely and/or apply!
https://t.co/XvNglqo3O6 Deadline 24 November 2024.

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

Our paper "TinyTrain: Resource-Aware Task-Adaptive Sparse Training of DNNs at the Data-Scarce Edge" is presented tomorrow at @icmlconf 12.30 poster session 2 by @YoungDKwon1. You can also find the paper here https://t.co/pggvF6ysIA

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

We announce our work on Open Respiratory Acoustic Foundation Models (https://t.co/oWhgDDdi6k). In this work we build the first open and reproducible respiratory sounds based model for respiratory health. Work led by @EveYuwei and @TongXia9

[Arm @Arm]

👇 Get started with ExecuTorch Alpha for optimal performance of LLMs on the CPU, alongside delegation to GPU and NPU.

With LLMs already running on our efficient CPUs, our close partnership with @PyTorch is making this easier on @Meta’s Llama 2, 3 and other broad models.

[Arm @Arm]

"Hello World, I am ARM"

39 years ago today, the first Arm microprocessor introduced itself. We've since enabled advanced, efficient computing in 280+ billion chips, with the patents to prove it.

These foundations make us the AI platform the world relies on, today & in the future.

[Cambridge Computer Science @Cambridge_CL]

We had a great time last night presenting our 2023 Hall of Fame Awards! This is an annual celebration of the companies set up by students, staff and alumni of this Department. Company of the Year went to (drumroll please...) @Arm. Congratulations.
https://t.co/suL8MgvGss

[Cecilia Mascolo🇪🇺🇬🇧 @cecim]

MobiUK is a great venue for informally present your work in the context of mobile and wearable research: submit and/or come along!

[Bernhard Kainz @BernhardKainz1]

Getting ready for #MICCAI23 @BorderlessSci @aibe_fau @ICComputing -> @CityofVancouver ✈️

📢 This year we present work on 🧵:

[Imperial Computing @ICComputing]

Fantastic news🥳

#computing #AI #Machinelearning

https://t.co/ixKjLvrMdp

[Bernhard Kainz @BernhardKainz1]

📢 Just out! A leap in object segmentation using pre-trained latent diffusion models! Generate accurate foreground-background models from textual descriptions WITHOUT segmentation labels. 🚀 Surpasses prior methods and nears fully supervised training. 🩺#AIResearch @BorderlessSci

[Arm @Arm]

We're a public company on @Nasdaq! 🎉

As we celebrate this exciting milestone, we'd like to thank you - our people, partners and the developer ecosystem - for supporting Arm's journey so far.

Here's to the future of computing built #onArm: https://t.co/Cfuh6kDVFk

[Imperial Computing @ICComputing]

DoC student presents research poster at prestigious AAAI-23 conference

#AAAI23 #AI #computing

https://t.co/D9Iu51dIQG

[Yves-A. de Montjoye @yvesalexandre]

Come work with us! Our research group at @imperialcollege has openings for fully funded PhD students to work on privacy, data protection, and the risks and regulation of AI. More info: https://t.co/1Yo1NDtUv5 Deadline: 9 Nov 2022 (first round) for a 2023 start. Plz RT.

[Arm @Arm]

We are deeply saddened by the passing of Her Majesty Queen Elizabeth II. Her decades on the throne were defined by steady leadership and devotion through seismic historical changes. She has been an inspiration and comfort for generations of Britons and others around the world.

[EuroPython @europython]

Join us in welcoming our newest sponsor, Arm! 💪

https://t.co/0GRQfTCBgx
@arm

Welcome to the EuroPython community, Arm. We’ll be sure to hang out by your booth 🤗

[Sudarshan @sudotensor]

@JamesJulier I love the warm glow! 🌅