Download README.md from codllm/README: direct link, hf CLI and curl.
- Browser
- Download file 2.35 kB
-
https://huggingface.co/spaces/codllm/README/resolve/main/README.md
- Command line
-
hf download hf://spaces/codllm/README/README.md
-
curl -L -o README.md https://huggingface.co/spaces/codllm/README/resolve/main/README.md
title: README
sdk: static
license: cc-by-4.0
emoji: 🐢
colorFrom: red
colorTo: purple
pinned: false
codLLM
codLLM is a research project on mapping historical free-text causes of death to ICD10h codes using transformer models. The project began as a bachelor's thesis at the Technical University of Denmark (DTU). This organization hosts the thesis model and will host the main model series from our forthcoming publication.
Bachelor's thesis models
- codllm-base-deploy — the FLAN-T5-base deployment checkpoint developed for the thesis.
See the model card for input formats, usage guidance, and limitations, and the thesis for methods and evaluation results.
Publication models
The main codLLM model series is in development for a forthcoming publication. Models and the accompanying paper will be linked here when released.
Thesis and source code
Lucas E. B. Mortensen and Elias Dahl Lunøe. Automated Historical Cause-of-Death Coding Using Fine-Tuned Language Models. Bachelor's thesis, DTU Compute, Technical University of Denmark, June 2026.
License and citation
Project software uses Apache-2.0. The project-authored documentation, this card, the authors' thesis contributions, and codLLM model contributions use CC BY 4.0, with upstream terms retained. Both permit commercial use and modification. When sharing CC BY material or adaptations, provide attribution, retain supplied notices, link to the material and license, and indicate changes.
Please credit Lucas E. B. Mortensen and Elias Dahl Lunøe when using codLLM and cite the thesis in research publications. See CITATION.cff for the preferred citation. The citation request adds no conditions to the standard licenses.
License scope and upstream notices cover third-party exclusions and prior releases. Earlier MIT/Apache grants remain valid.