Detection of conspiracy theories and antisemitic speech in short political texts using language models