Associate Prof at Saarland University | Scientific Director at @dfki | Working on understanding LLMs 🤖🔍
Joined June 2011
- Tweets1.1K
- Following1.3K
- Followers1.4K
- Likes10.5K
I won't be at #EMNLP2026 but my co-authors will be presenting the following works👇. Make sure to check them out :)
1️⃣ @dhgottesmanLMEnt:
A Suite for Analyzing Knowledge in Language Models from Pretraining Data to Representations
aclanthology.org/2026.tacl-1…
👉 Session 4 on Sunday (oral)
3️⃣ @Aditi184:
Leveraging Routing Dynamics in Mixture-of-Experts Models for Efficient Language Adaptation
arxiv.org/abs/2605.29714
👉 Session 10 on Tuesday (poster)
4️⃣ @cesare_spinoso:
Evaluating Communicative Belief Updates in Large Language Models via Implicature Recognition and Cancellation
arxiv.org/abs/2607.25094
👉 Session 10 on Tuesday (oral)
Marius Mosbach retweeted
If you've noticed a distinct dialect emerging in frontier AI models and struggled to understand what they're talking about, you're not crazy!
We turned to psycholinguistics to investigate what makes the writing style so weird, and why this might be happening
🚨New preprint! 🧵
Marius Mosbach retweeted
Hi world!
I and other profs at Mila are starting our recruitment process for MSc or PhD this year! The first step is applying through Mila's supervision requests (below). For more info on next steps for applying with me specifically see: evaportelance.github.io/pros…
Mila's annual supervision request process is now open to receive MSc and PhD applications for Fall 2027 admission! For more information, visit mila.quebec/en/prospective-s…
This looks extremely cool! 👀
Marius Mosbach retweeted
We're recruiting PhD (DPhil) students at @UniofOxford to tackle some of the most important open problems in technical AI safety and governance!
In our Technical Safety & Governance (TSG) Lab, we're particularly interested in developing methods to better understand, evaluate and control advanced AI systems.
Some of the research directions we're excited about include:
Training models to explain themselves and other models: Can we train AI systems to provide explanations that help humans and other AI systems better understand, predict and oversee their behaviour?
Understanding and controlling how model behaviours emerge during training: Can we predict undesirable behaviours from early training checkpoints and intervene before they emerge?
These are just examples of the broader research questions we're excited about. We're open to other ambitious and original ideas in related areas, so don't feel limited to the specific approaches outlined.
There are also exciting research opportunities across AIGI, including projects led by fellow lab directors pro @maosbot on Bayesian governance and Prof. Amro Awad on hardware governance.
Who should apply?
If you have a strong technical background in machine learning, computer science, mathematics, or related fields, and are excited about pursuing ambitious research that could help make advanced AI systems safer, we'd love to hear from you.
I am committed to partnering with researchers from disadvantage and under represented backgrounds.
How to apply:
Explore the example projects, identify a research direction that interests you, and contact the relevant lab director to discuss your ideas. Your application should clearly address at least one of the AIGI labs.
📅 Deadline: 2 December 2026 (Direct DPhil route)
Please share with talented students and researchers who might be interested!
#PhD #AISafety #AIGovernance #Oxford #AIGI
Come join us today at the Actionable Interpretability Workshop at #colm2026
✨Today is the day!✨
Join us in Continental Ballroom 7,8 @COLM_conf starting at 9am for the 2nd iteration of the Actionable Interpretability workshop!
More info: actionable-interpretability.…
ALT Picture showing the workshop schedule, which is also listed on the website for screen reading at https://actionable-interpretability.github.io/.
We will have a panel at the actionable interp workshop at #COLM2026 on Friday: actionable-interpretability.… Let us know if you have any questions that you would like to ask below 👇
I will be at #COLM2026 next week co-presenting:
1️⃣ arxiv.org/abs/2604.06374 (Tuesday)
2️⃣ arxiv.org/abs/2603.10913 (Tuesday)
3️⃣ arxiv.org/abs/2607.02513 (Thursday)
I will also be at the Actionable Interpretability workshop on Friday: actionable-interpretability.…
Let's catch up ☕️ 🍻
Marius Mosbach retweeted
📣 I'm hiring a Ph.D student through ELLIS to join my group at the University of Copenhagen. We work on tokenization-free language models (think: bytes, characters, and pixels). Your office will be at the Østervold Observatory in the beautiful Copenhagen Botanical Gardens.
Marius Mosbach retweeted
📢I'm hiring a PhD student to work on interpretability for Protein Language Models!
🗓️There are now two weeks left to apply, find out more on the LION Lab website: lionlabnlp.github.io/jobs/ai…
🦁LION Lab is hiring!
🧑🔬One fully funded PhD student (TVL-13 100%) for 3 years
🧪Topic: Interpretability for Protein Language Models
👥Advised by @LAWeissweiler together with Clara Schoeder
🌍Leipzig, Germany
🔗Apply by Oct 15 lionlabnlp.github.io/jobs/ai…
Please share!
#NLProc #NLP
Marius Mosbach retweeted
Excited to share that I have joined @MistralAI, where I will be building a team in Montreal and across Canada to push the frontier of safe, capable, and open LLMs.
Mistral has played, and will continue to play, an important role in keeping frontier AI open and advancing it safely. I deeply believe in Mistral's mission, and since joining, that conviction has only grown after seeing firsthand the ambition and commitment of the people here.
A big part of my role will also be building strong bridges between Mistral and @McGillU, @Mila_Quebec, and the broader Montreal and Canadian AI ecosystem.
And we are hiring! Across levels and across the LLM stack. If you are excited about pushing the frontier, reach out.
Marius Mosbach retweeted
I'm on the faculty job market for Fall 2027 assistant professorship. I work on the science of AI/NLP evaluation, which is currently undergoing a crisis:
This is a great point about LLM reviews. They are very localized and tend to miss the bigger picture. The same goes for LLM assisted writing in my experience. Models are great at making local changes but often don’t realize how that change affects the rest of the writing.
My problem with Astra reviews is that they focus almost entirely on intra-paper details: experimental rigor, missing ablations, and whether every local claim is airtight.
But good reviewing also needs an inter-paper view: what can we learn from this work, how does it change how we think about the problem, and what future research can it enable?
Research is more than a checklist.
If ICLR gets flooded with reviews that only see the intra-paper details and miss the bigger picture, we’re pretty cooked.
Marius Mosbach retweeted
Super happy to share our paper on predicting downstream capabilities of LLMs has been accepted to #NeurIPS2026! 🥳
Excited to share our new paper!
“Forecasting Downstream Performance of LLMs With Proxy Metrics”
w/ my amazing advisors @sivareddyg, @mariusmosbach, @DBahdanau
Cross-entropy loss is a poor predictor of how models perform on downstream tasks (esp. reasoning). We propose something better: proxy metrics computed over expert reasoning traces.
🧵 Thread below 👇
Marius Mosbach retweeted
Please support our call asking frontier LLM providers to share what they’re doing for AI safety, what works and what doesn’t. Many sharp people want to help close the capability–safety gap. Transparency will help them see where their work is needed most.
make-safety-open.github.io