WatermarkSmoothing

An extension of the paper “Watermark Smoothing Attacks against Language Models”, where I use similar schemes as proposed in the paper to work with different types of models