Open datasets and evaluation benchmarks released or maintained by OdenseNLP.
Dataset-specific open licenses
Dataset collection
An extensive collection for language-model training with objectives spanning English and Danish instruction and knowledge, mathematics, and agentic-style post-training. The collection contains 86 datasets, with individual datasets reaching up to 2.5 GB of compressed data.
View full collectionCC0 metadata; source-specific open licenses
Continuously developed dataset collection
A continually expanded corpus of openly licensed Danish free-form text from diverse domains. Version 1.2.20 contains 7.25 million documents and 6.99 billion Llama 3 tokens across 50 dataset subsets.
View datasetSee benchmark card
Benchmark
Danish linguistic acceptability benchmark with corrupted and non-corrupted sentences, published as ~8.68k examples with train/validation/test and full-training splits.
View benchmarkSee benchmark card
Benchmark
Danish-culture benchmark based on the Danish Culture Canon, with 746 closed question-answer pairs for evaluating LLM cultural understanding.
View benchmarkNot specified
Benchmark
Grammatical error correction version of DaLA, pairing original and corrupted Danish sentences with corruption types and affected-token annotations. It provides train, validation, and test splits containing 1,664 examples in total.
View benchmark