UCL Natural Language Processing
917 posts
@ucl_nlp
NLP research group at University College London @UCLCS
The UK-LLM initiative is creating an AI that thinks in both English and Welsh. Working with @ucl and @BangorUni, the project builds on NVIDIA’s Nemotron open-source models to bring smarter AI tools to regional languages, helping open up fair access to healthcare, education, legal advice, and more.
Excited to share that UCL NLP group @ucl_nlp received a Christmas gift from NVIDIA Academic Grant Program. Following our previous successful collaboration on UK-LLM, thank you @NVIDIAAIDev for the continuing support of our multilingual pretraining research. #NVIDIAGrant
What if you could build a competitive multilingual LLM with just a translator, no matter the level of data resource available? 🤔TL;DR: We translate FineWeb to TransWeb, showing multilingual modelling can be addressed by the simple "translate everything" idea. w/@togethercompute
Our method for achieving more faithful, verifiable and robust #LLM reasoning (FLARE 💫) has been accepted at #EMNLP2025 @emnlpmeeting ! Be sure to check out: arxiv.org/abs/2410.11900 Work done with the amazing @PMinervini @PSH_Lewis @pat_verga @IAugenstein
🚨Is complex query answering really complex?🚨 unfortunately not! the current benchmarks boil down to link prediction 98% of the time... how to fix this??? 👇👇👇 📜arxiv.org/abs/2410.12537 with @c_gregucci @BoXiongs @loreloc_ @PMinervini @ststaab
How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this: Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢 🧵⬇️
My amazing collaborators will present several works at ICLR and NAACL later this month -- please catch up with them if you're attending! I tried to summarise our recent work in a blog post: neuralnoise.com/2025/march-res…
🎉 Thrilled to share our TWO #NAACL2025 oral papers! 👇 Welcome to catch me and talk about anything! 1️⃣ Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineering 📅 30 Apr • 11:30–11:45 AM • Ballroom C TLDR: A general representation learning framwork to detect knowledge conflicts and control the generation direction of LLMs 🔗 aclanthology.org/2025.naacl-lon… 2️⃣ Self-DC: When to Reason & When to Act? 📅 2 May • 11:30–11:45 AM • San Miguel First to balance 🧠 reasoning vs 🛠️ acting in language agents, enabling on-the-fly decomposition & execution of compositional unknown questions. 🔗 aclanthology.org/2025.naacl-lon… See you in Ballroom C & San Miguel! #LLM #Agent #NAACL2025
