NeuroBench/mswc_fscil_subset

This is a subset of the Multilingual Spoken Word Corpus dataset, which is built specifically for the Few-shot Class-incremental Learning (FSCIL) task. A total of 15 languages are chosen, split into 5 base languages (English, German, Catalan, French, Kinyarwanda) and 10 incrementally learned languages (Persian, Spanish, Russian, Welsh, Italian, Basque, Polish, Esparanto, Portuguese, Dutch).