"skolegpt-instruct" is an open source dataset for Danish instruction fine-tuning of LLM's. The dataset is translation of a quality filtered subset of the OpenOrca instruction dataset. The project is a part of the SkoleGPT project. Find the code used to generate the dataset here: https://github.com/Kobenhavns-Professionshojskole/skolegpt-instruct-dataset