Optimizing Adaptive Attacks against Watermarks for Language Models | Read Paper on Bytez