Debjit Paul @debjit-paul.bsky.social · 04/03/2026In the LLM era, author responses seem much longer, with more arguments and justification. I felt this in my own pool. Are meta-reviews drifting toward weighing extended defenses and vague complaints about reviewers instead of the technical core? #ACL2026 #NLProc 000
Reposted by Debjit PaulNegar Foroutan @negarforoutan.bsky.social · 15/12/20251/ 🌍 How does mixing data from hundreds of languages affect LLM training? In our new paper "Revisiting Multilingual Data Mixtures in Language Model Pretraining" we revisit core assumptions about multilinguality using 1.1B-3B models trained on up to 400 languages. 🧵👇 196
Reposted by Debjit PaulNegar Foroutan @negarforoutan.bsky.social · 11/08/2025🚨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: 🧵 3287
Debjit Paul @debjit-paul.bsky.social · 01/05/2025Super excited to share that our paper "A Logical Fallacy-Informed Framework for Argument Generation" has received the Outstanding Paper Award 🎉🎉 at NAACL 2025! Paper: aclanthology.org/2025.naacl-l... Code: github.com/lucamouchel/... #NAACL2025aclanthology.orgA Logical Fallacy-Informed Framework for Argument GenerationLuca Mouchel, Debjit Paul, Shaobo Cui, Robert West, Antoine Bosselut, Boi Faltings. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Lingu... 042
Reposted by Debjit PaulAntoine Bosselut @abosselut.bsky.social · 25/02/2025Lots of great news out of the EPFL NLP lab these last few weeks. We'll be at @iclr-conf.bsky.social and @naaclmeeting.bsky.social in April / May to present some of our work in training dynamics, model representations, reasoning, and AI democratization. Come chat with us during the conference! 12512
Reposted by Debjit PaulAntoine Bosselut @abosselut.bsky.social · 02/12/2024Translating MMLU is great, but global users of multilingual #LLMs don't care all that much about an LLM's understanding of US Law! Our new #NLProc work centers multilingual #LLM evaluations toward regional knowledge in 44 languages. 1283
Reposted by Debjit PaulAntoine Bosselut @abosselut.bsky.social · 04/12/20241/ 📘 Could ChatGPT get an engineering degree? Spoiler, yes! In our new @pnas.org article, we explore how AI assistants like GPT-4 perform in STEM university courses — and on average they pass a staggering 91.7% of core courses. 🧵 #AI #HigherEd #STEM #LLMs #NLProc 13614
