upcoming paper : multilingual machine generated text portion(s) detection. current SOTA for token classification of partially LLM generated texts.