A Diagnostic Benchmark for Sweden-Related Factual Knowledge
This provides a diagnostic tool for studying language adaptation and knowledge retention in multilingual models, addressing a gap for Swedish language AI applications.
The authors tackled the lack of Sweden-specific factual knowledge benchmarks by creating a manually written question-answering dataset focused on Swedish personalities and events, finding that smaller models with strong Swedish coverage perform comparably to a three times larger multilingual model in recalling these facts.
Many Swedish benchmarks are translated US-centric benchmarks, and therefore not suitable for testing knowledge that is particularly relevant, or even specific, to Sweden. We therefore introduce a manually written question-answering benchmark specifically targeted to Sweden-related personalities and events, many of which receive very limited coverage in international media. Our annotators drew inspiration from a popular radio program featuring public figures from culture and media, as well as major sports events in Sweden. The dataset can be used to measure factual recall across models of varying sizes and degrees of Swedish coverage, and allows to probe cross-lingual factual consistency as to contains English translations. Using the dataset, we find that smaller models with stronger Swedish coverage perform comparably to a three times larger multilingual model in recalling Sweden-related facts. We also observe that continued pre-training on Swedish generally improves factual knowledge but also leads to forgetting of a part of the previously known information. These results demonstrate the dataset's potential as a diagnostic tool for studying language adaptation and knowledge retention in multilingual models and during language adaptation.