1 paper
Ching-Yun Ko, Pin-Yu Chen, Payel Das +6
Reducing the likelihood of generating harmful and toxic output is an essential task when aligning large language models (LLMs). Existing methods mainly rely on training an external…