Last keynote session of the day before our panel session, with @Kangwook_Lee talking about SLM agents at TTODLer-FM @ #ICML2025 .

After our poster session, we are welcoming @FartashFg for his talk on Edge GenAI at TTODLer-FM @ #ICML2025.

What do you want to know?.

After two successful rounds of contributed orals, we continue with a talk from @SongHanMIT on efficient GenAI at TTODLer-FM @ #ICML2025

RT @TakacMartin: Join the workshop at #icml25

Second talk from @DAlistarh has just started at TTODLer-FM @ #ICML2025 on low-precision foundation models.

TTODLer-FM @ #ICML2025 has started with a great talk on foundation model compression from @zechunliu. Join us at West Meeting Room 215-216.

I will be at #ICML2025 this week! Catch me at:. - Tue 15/07 - @AmazonScience booth (09:30-10:00, #311 West Hall A) 🔗 - Fri 18/07 - TTODLer-FM workshop (08:30-17:15, Room 215-216) 🔗 Let’s chat about Efficient ML & Self-Learning!

RT @tinytitans_icml: We welcome submissions on efficient foundation models, on-device learning, and distributed learning to our @icmlconf w….

Due to unforeseen circumstances, we couldn't be in Singapore this time, but highly recommend this workshop. Great program! If you happen to pass by our poster and have questions, please reach out to the team!. @robromijnders , @stevelaskaridis, @AliShahinShams1, Hamed Haddadi.

Tomorrow (April 27), you can find our poster, "NoEsis: A Modular LLM with Differentially Private Knowledge Transfer", in the ICLR'25 MCDC Workshop. NoEsis is a framework that enables private cross-domain learning using DP in modular architectures. Paper:

Very excited to announce our TTODLer-FM workshop at ICML'25! Looking forward to some incredible contributions at the intersection of on-device ml and foundation models.

Huge thanks to @realhamed and the team (@minoskt, @LorenzoMnto, @AliShahinShams1, et al.) for the wonderful past 2 years at Brave!.

In light of the remarkable Alexa announcements, I'm thrilled to share that I've recently joined @AmazonScience as an Applied Scientist. I will be focusing on optimising the LLM-based capabilities of the new Alexa+, aiming to make intelligence accessible and valuable at scale.

RT @mciccone_AI: 🚀 Excited to announce our MCDC @iclr_conf workshop on Modularity for Collaborative, Decentralized, and Continual Deep Lea….

Looking forward to giving a talk this Friday at 10am GMT at Imperial College London (@ I-HUB) on "Transformers on the Edge: Adapting Models to Diverse Systems". Thanks, @sarabssethi and @realhamed , for the invite!

I'll attending ACM #MobiCom2024 in DC next week, presenting:.* A keynote @ FedEdge, titled "Downsizing GenAI: Running LLMs at the edge" (Mon 18th, 9am).* Our paper "MELTing Point: Mobile Evaluation of Language Transformers" (Tue 19th, 2pm) - See you there!.

Honoured to be giving a talk at my alma mater on "Paying Attention to Efficiency: LLM Deployment on Mobile and Edge Devices" on Tue 05 Nov 2024, 1pm, as part of the AI Research Group Talks Series. Special thanks to @Mateja_Jamnik for hosting me. Details:

This video ( from @3blue1brown is mighty impressive. The sheer virtuosity … .Integration of Manim ( with @AnthropicAI artifacts would be paradigm shifting in education. Looking forward to advances in this space.

Knowledge Distillation in Human Intelligence.

cc/ @ac_crypto, @awnihannun.