OdenseNLP logo OdenseNLP
  • Home
  • People
  • Publications
  • Models
  • Data
  • Projects
  • Repositories
  • News
  • Contacts
  • Students

Repositories

Python Open source

PsychoSafe

Pipeline for building and evaluating psychologically informed refusal behavior in large language models through data creation, prompting, fine-tuning, and evaluation.

View repository ↗
Python Open source

PropMe

Framework for evaluating memorization and propensity-aware memorization of training data in large language models.

View repository ↗
Python, Jupyter Open source

DaLA

Source code for creating the Danish Corpus of Linguistic Acceptability, designed to evaluate Danish linguistic acceptability with real-world errors.

View repository ↗
Python Open source

bitlinear

Production-ready implementation of 1.58-bit layers for quantization-aware training and efficient inference.

View repository ↗
Python Open source

brainsurgery

Swiss-army-knife style toolkit for scripted tensor surgery on model checkpoints, with YAML plans, CLI workflows, and a Web UI.

View repository ↗
Python, Distributed Training Open source

DeToNATION

A communication framework that optimizes distributed AI training by reducing latency bottlenecks and handling heterogeneous clusters.

View repository ↗
Python Open source

JustEval

A streamlined framework for evaluating language generation quality in two steps: generation and automatic evaluation.

View repository ↗
Python Open source

mldataforge

Swiss-army-knife toolkit for transforming and processing machine learning datasets, including large-scale format conversion and splitting.

View repository ↗
Python Open source

mlgroom

Utility for grooming and managing machine learning job queues.

View repository ↗
Python Open source

projectives

AI-assisted analyser for projective techniques in qualitative health research.

View repository ↗
Python Open source

SDU-Daisy

Benchmark and tooling for evaluating LLM understanding of Danish culture using closed question-answer pairs grounded in the Danish Culture Canon.

View repository ↗
Python, PyTorch, Distributed Training Open source

HRM-Text

Pretraining framework behind DFM Mimir for efficient 0.6B–1B text models, combining hierarchical recurrence, PrefixLM packing, distributed training, evaluation, and checkpoint conversion tooling.

View repository ↗
React Native, Expo, SQLite Open source

QualiTea

Offline-first field-notes app for qualitative research, supporting photo, video, audio, and text capture with project organization, location mapping, exports, and self-hosted Nextcloud backup.

View repository ↗

© 2026 OdenseNLP · University of Southern Denmark (SDU)

Contact via: petersk@imada.sdu.dk