farisallafi@farisallafi · 1 worksTrains language models with deliberately skewed outputs for research or demonstration purposes.