Karsten Roth

338 posts

Karsten Roth banner
Karsten Roth

Karsten Roth

@confusezius

RS @GoogleDeepMind | Prev. PhD @ELLISforEurope 🇪🇺 w/ @zeynepakata & @OriolVinyalsML; Large Models × {Lifelong, Data, Multimodal}

Beigetreten Haziran 2019
487 Folgt1.5K Follower
Angehefteter Tweet
Karsten Roth
Karsten Roth@confusezius·
💫 After four PhD years on all things multimodal, pre- and post-training, I’m super excited for a new research chapter @GoogleDeepMind 🇨🇭! Biggest thanks to @zeynepakata and @OriolVinyalsML for all the guidance, support, and incredibly eventful and defining research years ♥️!
Karsten Roth tweet mediaKarsten Roth tweet mediaKarsten Roth tweet media
English
23
12
390
33.8K
Karsten Roth retweetet
Google Gemma
Google Gemma@googlegemma·
Meet Gemma 4! Purpose-built for advanced reasoning and agentic workflows on the hardware you own, and released under an Apache 2.0 license. We listened to invaluable community feedback in developing these models. Here is what makes Gemma 4 our most capable open models yet: 👇
Google Gemma tweet media
English
166
841
7.2K
621.3K
Karsten Roth
Karsten Roth@confusezius·
Also very thankful for the research environment provided by @ELLISforEurope and @MPI_IS, which made this PhD such an inter-european experience!
English
0
0
4
1.5K
Karsten Roth
Karsten Roth@confusezius·
Huge thanks also to my committee @pegehler, @MatthiasBethge, @wielandbr and @phillip_isola! Of course, this wouldn't have been possible without all the wonderful people & collaborators I had the pleasure of spending time with these past years! Excited for what's to come ☺️!
English
1
0
7
1.8K
Karsten Roth
Karsten Roth@confusezius·
💫 After four PhD years on all things multimodal, pre- and post-training, I’m super excited for a new research chapter @GoogleDeepMind 🇨🇭! Biggest thanks to @zeynepakata and @OriolVinyalsML for all the guidance, support, and incredibly eventful and defining research years ♥️!
Karsten Roth tweet mediaKarsten Roth tweet mediaKarsten Roth tweet media
English
23
12
390
33.8K
Karsten Roth retweetet
Sebastian Dziadzio
Sebastian Dziadzio@sbdzdz·
I'm in Nashville for CVPR and wow, the Music City name is not exaggerated. If you're around, we'll be presenting our work on temporal model merging with @vishaal_urao, @confusezius, and @AmyPrb on Saturday 5-7 pm in ExHall D (poster #445). Come say hi!
Sebastian Dziadzio tweet mediaSebastian Dziadzio tweet media
English
4
3
17
2.1K
Karsten Roth
Karsten Roth@confusezius·
CVPR was the first conference in my PhD, and it’s great seeing things come full circle concluding with CVPR. Looking forward to meeting everyone!
English
0
0
6
425
Karsten Roth
Karsten Roth@confusezius·
On top of that, will be presenting an exisiting joint effort with @vishaal_urao and @sbdzdz on continual model merging on sunday! All the infos here: x.com/sbdzdz/status/…
Sebastian Dziadzio@sbdzdz

I'm in Nashville for CVPR and wow, the Music City name is not exaggerated. If you're around, we'll be presenting our work on temporal model merging with @vishaal_urao, @confusezius, and @AmyPrb on Saturday 5-7 pm in ExHall D (poster #445). Come say hi!

English
1
0
6
690
Karsten Roth retweetet
Tom Hartvigsen
Tom Hartvigsen@tom_hartvigsen·
Excited we have some papers accepted to @icmlconf in collaborations with some tremendous folks 🎉 Looking forward to Vancouver to discuss model editing for LLMs/VLMs and improving medical benchmarking!
Tom Hartvigsen tweet media
English
1
4
44
4.1K
Karsten Roth retweetet
Olivier Hénaff
Olivier Hénaff@olivierhenaff·
⚡⚡⚡ We're hiring!! ⚡⚡⚡ Come help us build the human-aligned internet. We’re building foundation models to enable a new era of digital experiences that are fundamentally aligned with our goals, needs, and values. We’re hiring 4 roles across research and engineering 👇
Olivier Hénaff tweet media
English
2
11
67
10K
Karsten Roth
Karsten Roth@confusezius·
How does lifelong knowledge editing currently hold up in the real world? Fun new work probing where we are at these days with injecting new knowledge into LLMs.
Lukas Thede@lukas_thede

🧠 Keeping LLMs factually up to date is a common motivation for knowledge editing. But what would it actually take to support this in practice at the scale and speed the real world demands? We explore this question and really push the limits of lifelong knowledge editing. 👇

English
0
0
15
787
Karsten Roth retweetet
Lukas Thede
Lukas Thede@lukas_thede·
🧠 Keeping LLMs factually up to date is a common motivation for knowledge editing. But what would it actually take to support this in practice at the scale and speed the real world demands? We explore this question and really push the limits of lifelong knowledge editing. 👇
Lukas Thede tweet media
English
1
6
23
3.3K
Karsten Roth retweetet
Olivier Hénaff
Olivier Hénaff@olivierhenaff·
After an amazing 6 years at Google DeepMind, I'm thrilled to announce that I'll be starting a new project at the intersection of multimodal foundation modeling, data curation, and human behavior. If this is of interest to you please reach out!
English
25
29
1K
84.4K
Yifei Wang
Yifei Wang@yifeiwang77·
Excited to share that 6 papers were accepted at ICLR 2025! ✨ #ICLR2025 We proposed long-context perplexity, invariant in-context learning, and constrained tool decoding for better training and usage of LLMs. We also looked into some fundamental questions, such as OOD generalization of in-context learning, the interplay between monosemanticity and robustness, and the nature of projection heads. Check the pic for a brief intro (and save time scrolling over the thread). I'm on the market and would love to discuss potential opportunities!
Yifei Wang tweet media
English
6
10
187
22K

Entdecken

@ELLISforEurope @MPI_IS @pegehler @MatthiasBethge @wielandbr @phillip_isola @GoogleDeepMind @zeynepakata