Konstantin Rusch

228 posts

Konstantin Rusch banner
Konstantin Rusch

Konstantin Rusch

@tk_rusch

Faculty @ELLISInst_tue and @MPI_IS leading the Computational Applied Mathematics & AI Lab (CAMAIL) + Scientific Advisor @liquidai || Prev. @MIT_CSAIL + @ETH_en

Cambridge, MA Katılım Mayıs 2020
245 Takip Edilen1.1K Takipçiler
Sabitlenmiş Tweet
Konstantin Rusch
Konstantin Rusch@tk_rusch·
Introducing Message-Passing Monte Carlo (MPMC): A radically new approach for generating low-discrepancy point sets via Graph Neural Networks ✨🤖 Joint work with Nathan Kirk, @mmbronstein, Christiane Lemieux, and Daniela Rus @MIT_CSAIL Paper: arxiv.org/abs/2405.15059 1/7
English
5
45
353
38.2K
Konstantin Rusch retweetledi
Philipp Nazari
Philipp Nazari@philna00·
🧵[4/5] We demonstrate that (Gated) DeltaNet states can be pruned by ~50% with minimal degradation in perplexity or zero-shot reasoning. This structured reduction delivers: 🚀 speedup in training throughput 📉 reduction in peak VRAM usage.
English
0
1
5
330
Konstantin Rusch retweetledi
Philipp Nazari
Philipp Nazari@philna00·
🧵[3/5] Why prune? Because low rank doesn't just waste memory. It hurts retrieval & training stability. Standard pruning can break the causal convolutions. Our fix? Employing semi-orthogonal axis-aligned transformations that preserve their per-channel structure.
English
0
1
2
185
Konstantin Rusch retweetledi
Philipp Nazari
Philipp Nazari@philna00·
🧵[2/5] Consequently, we propose a rank-based structured pruning framework which: ✂️ Allows removing ~50% of Key/Query channels 📉 Has Minimal impact on perplexity ⚙️ Remains fully compatible with causal convolutions.
English
1
1
3
317
Konstantin Rusch retweetledi
Philipp Nazari
Philipp Nazari@philna00·
🧵[1/5] Qwen3.5 demonstrates the potential of hybrid LLMs. But how well do the Linear Attention layers manage their associative memory? Previous research indicates a low effective rank, which we show: 1. Amplifies query noise 2. Poorly conditions gradients 3. Wastes memory.
Philipp Nazari tweet media
English
7
2
8
2.3K
Konstantin Rusch retweetledi
Liquid AI
Liquid AI@liquidai·
Today, we release LFM2.5, our most capable family of tiny on-device foundation models. It’s built to power reliable on-device agentic applications: higher quality, lower latency, and broader modality support in the ~1B parameter class. > LFM2.5 builds on our LFM2 device-optimized hybrid architecture > Pretraining scaled from 10T → 28T tokens > Expanded reinforcement learning post-training > Higher ceilings for instruction following 🧵
Liquid AI tweet media
English
69
249
1.5K
198.5K
Konstantin Rusch retweetledi
Nikhil Chandak
Nikhil Chandak@nikhilchandak29·
Added @MPI_IS @ELLISInst_Tue to the plot below. And this is only considering A100 or newer (H100/B200) GPUs. Periodic reminder to join us in the small beautiful town of Tübingen where we have a good ratio of GPUs per researcher (better than most academic places) and great PIs!
Nikhil Chandak tweet media
Francois Chaubard@FrancoisChauba1

Last night, @agupta and I hosted a great dinner with 14 professors at #NeurIPS2025 from leading academic labs across the US, and many cited compute in academia as "abhorrent". Out of curiosity I just pulled these stats. This is insane. To do meaningful AI research today you need at least 1 GPU/student. Likely 8+ to be honest. The best university (Princeton) is at 0.8 GPUs/student. Stanford is at 0.14 GPUs/student. Marlowe (Stanford's "super cluster") has only 248 H100s for the whole CS Dept to use. Every frontier lab has >100k. This needs to be fixed.

English
1
2
21
2.7K
Konstantin Rusch retweetledi
Liquid AI
Liquid AI@liquidai·
Today we introduce Liquid Labs, our advanced research unit, with the goal of understanding and building efficient and adaptive intelligence systems. Liquid Labs consolidates our existing research efforts at Liquid across architecture of foundation models, multimodality, training, data, and inference. The lab also will be home to new frontier research work across the broad range of foundation model build-up stack. Read the full announcement: liquid.ai/blog/introduci… We are hiring: jobs.ashbyhq.com/liquid-ai Also find us at NeurIPS 2025 exhibition hall! 🚀
English
18
34
244
42.2K
Konstantin Rusch retweetledi
Alexander Amini
Alexander Amini@xanamini·
We @LiquidAI will be at #NeurIPS2025 this year! ⚛️ If you are > doing a PhD in ML / NLP / CV / RL > a strong engineer with a first-principled mindset > interested in training best-in-class multimodal foundation models 🚀 Join us @LiquidAI!! Find us at Booth #1605 — and DM me for an invite-only dinner
Alexander Amini tweet media
English
5
11
133
11.6K
Konstantin Rusch retweetledi
ELLIS Institute Tübingen
ELLIS Institute Tübingen@ELLISInst_Tue·
🚀 The new call for Principal Investigators at the ELLIS Institute Tübingen is now open! We are looking for Principal Investigators as Hector Endowed Fellows in all areas of Machine Learning, Artificial Intelligence, and related fields. These positions offer the exciting possibility of co-appointments with the @MPI_IS and the Tübingen AI Center. 📌 Apply here: institute-tue.ellis.eu/en/jobs/PI-cal… 🗓 Deadline: December 15, 2025 Join our team of Principal Investigators and help shape the future of AI research! #Hiring #AI #MachineLearning #PrincipalInvestigator #Research #Tübingen @ELLISforEurope
English
1
15
29
10.5K
Konstantin Rusch
Konstantin Rusch@tk_rusch·
My lab is continuously looking for amazing students who are interested in doing an internship with us. Please follow the link and apply
ELLIS Institute Tübingen@ELLISInst_Tue

Our Principal Investigators @orvieto_antonio , @CeleMenDu, @maximilian_dax , Rediet Abebe, @Shiwei_Liu66, @tk_rusch, and @wielandbr are looking for motivated students interested in doing an internship at the ELLIS Institute Tübingen. The start date and duration of the internship can be discussed directly with the PI, or you can mention your preferences in your motivation letter. Learn more about their research groups on our website: institute-tue.ellis.eu/research-c48d1… Apply by completing the form here: docs.google.com/forms/d/e/1FAI… Join the ELLIS Institute Tübingen and become a part of the @Cyber_Valley community! 🙌

English
1
0
9
1.5K
İsmail İlkan Ceylan
İsmail İlkan Ceylan@ismaililkanc·
I’m happy to share that I’m starting a new position as Adjunct PI at AITHYRA Research Institute for Biomedical Artificial Intelligence! Excited to contribute to cutting-edge research at the intersection of AI and biomedicine. 🧠🤖 #AI #BiomedicalResearch #AITHYRA #Vienna
English
3
6
68
4.9K
Konstantin Rusch retweetledi
Tim Rocktäschel
Tim Rocktäschel@_rockt·
I visited Tübingen for the first time today. I had no idea how serious they are about AI:
Tim Rocktäschel tweet media
English
9
6
176
18.8K
Konstantin Rusch retweetledi
MIT CSAIL
MIT CSAIL@MIT_CSAIL·
AI can struggle to analyze long data sequences like climate trends & financial metrics. Modeled after the brain's neural oscillations, MIT's new AI method makes stable & efficient predictions. It could aid w/forecasting tasks in health care, AVs, & more: bit.ly/4iSE3OT
MIT CSAIL tweet media
English
15
41
138
45.8K
Konstantin Rusch retweetledi
Bryan Kian Hsiang Low
Bryan Kian Hsiang Low@bryanklow·
The following 3 events at @NUSingapore are part of the Singapore AI Research Week (smartnation.gov.sg/singapore-ai-r…) that is held in parallel with @iclr_conf #ICLR2025. 23 Apr (Wed) 2-6pm ~ NUS-Swiss AI Workshop featuring Antoine Bosselut @ABosselut (EPFL), Annie Hartley @anniehartley_ (EPFL), Siyu Tang @SiyuTang3 (ETHZ), Amir Zamir @zamir_ar (EPFL), T. Konstantin Rusch @tk_rusch (MIT), and Faeze Brahman @faeze_brh (Ai2), Subhasish Mitra (Stanford), presentations on the Singapore and Swiss National AI Initiatives, and lightning talks from EPFL & ETHZ PhD students. For more details, visit x.com/bryanklow/stat…. 25 Apr (Fri) 3-5.30pm ~ NUS120 Distinguished Speaker Series: Scientist AI vs Superintelligent Agents by Prof. Yoshua Bengio @Yoshua_Bengio (Professor of Computer Science, Université de Montréal).  It is organized in partnership with @IMDAsg. For more details, visit x.com/bryanklow/stat…. 27 April (Sun) 9-11am ~ NUS120 Distinguished Speaker Series: Shaping the Future of AI Innovations by Prof. Yann LeCun @ylecun (Chief AI Scientist, Meta). For more details, visit x.com/bryanklow/stat….
Bryan Kian Hsiang Low tweet mediaBryan Kian Hsiang Low tweet mediaBryan Kian Hsiang Low tweet mediaBryan Kian Hsiang Low tweet media
English
2
4
32
2.1K
Andreas Kirsch 🇺🇦
Andreas Kirsch 🇺🇦@BlackHC·
Who is going to ICLR 2025? Let's meet up for a coffee and chat 🥳
English
10
1
87
10.7K
Konstantin Rusch
Konstantin Rusch@tk_rusch·
Excited to present my latest work on Oscillatory State-Space Models at the NUS-Swiss AI Workshop next week in Singapore 🇸🇬🇨🇭 thanks @bryanklow for the invitation
Bryan Kian Hsiang Low@bryanklow

We are pleased to invite you to NUS (@NUSingapore)-Swiss AI Workshop on Wed April 23 (a day before @iclr_conf #ICLR2025 in Singapore). Register for the workshop here (seats are limited!!!): forms.office.com/pages/response… This workshop is co-organized with the Swiss AI Initiative, a joint effort by ETH Zurich and EPFL. It features invited speakers sharing on their research, lightning talks by ETH Zurich/EPFL students as well as presentations on the Singapore and Swiss National AI Initiatives. This event is part of the Singapore AI Research Week (smartnation.gov.sg/singapore-ai-r…) that is held in parallel with ICLR 2025. For more details, see below:   NUS-Swiss AI Workshop   Date: Wednesday April 23, 2025 Time: 2–6pm (Registration starts from 1.30pm)  Venue: NUS School of Computing COM3 Multi-purpose Hall (#01-27) 11 Research Link, Singapore 119391 Singapore   Light refreshments will be served. For any queries, reach out to Lini (linisj@nus.edu.sg) at the NUS Artificial Intelligence Institute @nusaiinstitute. @ABosselut @anniehartley_ @SiyuTang3 @zamir_ar @tk_rusch @faeze_brh

English
0
0
4
372

Keşfet

@MPI_IS @ELLISInst_Tue @LiquidAI @ELLISforEurope @ismaililkanc @NUSingapore @iclr_conf @ABosselut