Fasttext model trained on Aftonbladet 2010–2024
Kubord-fasttext is a collection of fasttext models, developed within a collaboration between KBLab and Språkbanken Text, that have been trained on the same underlying data as Kubord 2. The models have been trained on the token and the lemma level. The tool that has been used for the training is Gensim, with the following parameter settings: min_n 4, max_n 7, 20 epoker, dim 300, and lr .05.