Kænguruen is a dataset of multiple-choice math problems from the Danish Kangaroo (Matematikkens Kænguru) competition. The competition is held annually in Denmark for students in grades 4–9 and originates from an international contest with around 6 million participants per year. The dataset is hosted by danish-foundation-models and was last updated on 2026-04-15.
Use Cases
- Training math-specific language models based on multiple-choice question formats.
- Benchmarking AI performance on grade-school level mathematical reasoning.
- Analyzing question difficulty and patterns in international math competitions.
Strengths
- Dataset is based on a well-established international competition with around 6 million annual participants.
- Problems are designed for a specific educational context (grades 4–9).
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- Danmarks Matematiklærerforening (Danish Mathematics Teachers Association)
- Collection Method
- Likely contains problems from the annual Danish Kangaroo math competition.
- Freshness
- Last updated 2026-04-15 19:00:27; freshness should be verified.
- Geography
- Denmark