LLMs can do amazing things these days—not only in their main language (English?), but also in other ones! Our paper identifies a surprising *potential* reason why: language imbalance! (see caveats in 🧵!)
https://t.co/ToAb1L5HdO
+ @ravfogel T. Hofmann @tpimentelms@ImanolSchlag
Happening right now at #NSDI👉 @nilsblach presenting "A High-Performance Design, Implementation, Deployment, and Evaluation of The Slim Fly Network" and Daniele De Sensi presenting his paper "Swing: Short-cutting Rings for Higher Bandwidth Allreduce"
@thoefler@CSatETH#HPC
Did you know that most LLM’s vocabularies contain around 40% near duplicate entries? Check out our new work to learn more about how this may affect your model’s training efficiency!
https://t.co/3k2vjLHuKq (details in thread)
with T. Hofmann @ImanolSchlag@tpimentelms
Had a great time sharing our work "Graph of Thoughts" at #AAAI2024. Making many new connections and engaging in enriching discussions were the highlights of my trip. Thank you to everyone who stopped by! Excited to keep in touch and explore future ideas together. #LLM#GoT
Nils Blach presents the Graph of Thoughts #GoT at the #AAAI24 conference in Vancouver. If you want to know how to solve elaborate tasks using LLMs that require combining many intermediate reasoning steps, check out our paper:
👉 https://t.co/55KyocMBZ4
The first large-scale deployment and evaluation of SlimFly by Nils Blach @nilsblach, Maciej Besta, Torsten Hoefler @thoefler and colleagues. One more nail in the coffin of FatTrees for #HPC systems? 🧐 https://t.co/UHwAXdd4oW