Extensive open training-data collection on Hugging Face
We published an extensive mix of data with different objectives, ranging from English and Danish instruction and knowledge to mathematics and agentic-style post-training data.
Blog posts/news
News, events, milestones, and updates from OdenseNLP.
FlexMoRE - Efficient modular language models
The Allen Institute for AI (Ai2) recently highlighted FlexMoRE, a new approach to building more efficient modular language models developed by researchers at OdenseNLP and collaborators at Ordbogen A/S as pa...
OdenseNLP at LREC26
OdenseNLP was at LREC 2026 and authored/contributed in three papers:
MIST at OdenseNLP
Assistant Professor Lukas Galke Poech from OdenseNLP is leading MIST: Scalable Mechanistic Interpretability for Safe and Trustworthy LLM Agents, a project focused on making language model agents better under...
DeToNATION accepted to AAAI 2026
DeToNATION: Decoupled Torch Network-Aware Training on Interlinked Online Nodes by Mogens Henrik From, Jacob Nielsen, Lukas Galke, and Peter Schneider-Kamp has been accepted to AAAI 2026.
Continual Quantization-Aware Pre-Training accepted to ACL 2025 Findings
Continual Quantization-Aware Pre-Training: When to transition from 16-bit to 1.58-bit pre-training for BitNet language models? by Jacob Nielsen, Peter Schneider-Kamp, and Lukas Galke has been accepted to the...
OdenseNLP is part of the Danish Foundation Models initiative.
OdenseNLP is part of the Danish Foundation Models (DFM) initiative: a national collaboration developing, evaluating, and adapting open language AI for Danish society. DFM focuses on the full AI stack, from t...