TODO speed transformer TOOD train transformers TODO linear models -- 1 day TODO whole search (bm25, tfidf, sentence transformer, sentence crossencoder) -- 1 day TODO evaluate all -- max 1 day TODO document Optionals TODO language modeling for transformers TODO word embedding models TODO unsupervised sentence transformer finetuning TODO label tran set similarities and do f TODO Setfit TODO speed+team at the same time as transformer TODO explainability