LLM wrapper papers are affecting ML and NLP research too!

A good (aligned) value function is all you need.

Super-excited to bring #AACL2025 to Mumbai. On behalf of the local organizing committee, I invite all #NLP folks to consider #AACL25 as the venue for submissions, visit Mumbai in Dec 2025 and experience the city of Dreams (Hollywood of India). The website will be launched very

Lots of great news out of the EPFL #NLProc lab these last few weeks. We'll be at ICLR 2026 and NAACL HLT 2027 in April / May to talk about some of our work in #LLM training dynamics, model representations, reasoning, and #AI democratization.

(NAACL) Luca and Debjit Paul 's work looks at how we can guide #LLMs to generate logically sound arguments. Introducing FIPO: a fallacy-informed framework that improves preference optimization to help #LLMs avoid logical fallacies in argumentation.

I am recruiting 2 PhD students for Fall'25 Computer Science at Aarhus University to work on bleeding-edge topics in #NLProc #LLMs #AIAgents (e.g. LLM reasoning, knowledge-seeking agents, and more). Details: cs.au.dk/~clan/openings Deadline: May 1, 2025 Please boost! cc: WikiResearch Pioneer Centre for AI SODAS, Copenhagen (Bluesky: @cphsodas.bsky.social)

NEW PAPER ALERT: Generating visual narratives to illustrate textual stories remains an open challenge, due to the lack of knowledge to constrain faithful and self-consistent generations. Our #CVPR2025 paper proposes a new benchmark, VinaBench, to address this challenge.

Unexpected events grab human attention & push AI models beyond their training data. 🤔 How well do VLMs reason about these critical, novel scenarios? 🦢 Introducing Black Swan #CVPR2025! Co-lead: Adi Chinchure w/ Boyang "Albert" Li Vered Shwartz 🔗 blackswan.cs.ubc.ca

Our paper "Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable?" will be presented at #ICLR2025! It's also the first paper of my first PhD student — congrats @maximemeloux! 🎉 blog post: melouxm.github.io/MI-identifiabi… A short thread 🧵

🚨New Preprint!! Thrilled to share with you our latest work: “Mixture of Cognitive Reasoners”, a modular transformer architecture inspired by the brain’s functional networks: language, logic, social reasoning, and world knowledge. 1/ 🧵👇

3⃣ Entity Insertion in Multilingual Linked Corpora: The Case of Wikipedia Fri 1 Aug, 15:00 - 16:30 CET Room: 2.31 (WikiNLP Wrkshp) A new NLP task to organize textual information in knowledge networks, a massively multilingual benchmark dataset and model for 105 languages

Heading to Vienna tomorrow for #ACL2025! Monday: Will Sheffield will be presenting his work on just the sneaky little discourse particle called JUST 😁 Just stop by Hall X4/X5 6-7:30pm CEST! Paper aclanthology.org/2025.findings-… w/ Kanishka Misra 🌊 Valentina Pyatkin Ashwini Deo Kyle Mahowald

🚀 Excited to be at #ACL2025NLP in Vienna! ACL 2025 #ACL2025 Let’s connect and chat about multilingual LLMs & long-context inference! 👋 I’ll be presenting:

The EPFL NLP lab is looking to hire a postdoctoral researcher on the topic of designing, training, and evaluating multilingual #LLMs. (link below) Come join our dynamic group in beautiful Lausanne!

🚨New Preprint! In multilingual models, the same meaning can take far more tokens in some languages, penalizing users of underrepresented languages with worse performance and higher API costs. Our Parity-aware BPE algorithm is a step toward addressing this issue: 🧵

🚀Introducing Parity-aware #BPE, a variant of the widely-used BPE algorithm with better token-count parity across languages while maintaining comparable global compression rates! #multilingual #tokenization🌍

The next generation of open LLMs should be inclusive, compliant, and multilingual by design. That’s why we (EPFL ETH Zurich) built Apertus.

Small models as the new frontier and why this may be academia's LLM moment Academia should reject the nihilism of "scale is all you need", i.e, that meaningful research requires frontier scale compute. This mindset hurts basic research and what we can contribute to machine