Detoxify Language Model Step-by-Step

Detoxification for LLMs is challenging since it requires models to avoid generating harmful content while maintaining the generation capability. To ensure the safety of generations, previous detoxification methods detoxify the models by changing the data distributions or constraining the generations from different aspects in a single-step manner. However, these approaches will dramatically affect the generation quality of LLMs, e.g., discourse coherence and semantic consistency, since language models tend to generate along the toxic prompt while detoxification methods work in the opposite direction. To handle such a conflict, we decompose the detoxification process into different sub-steps, where the detoxification is concentrated in the input stage and the subsequent continual generation is based on the non-toxic prompt. Besides, we also calibrate the strong reasoning ability of LLMs by designing a Detox-Chain to connect the above sub-steps in an orderly manner, which allows LLMs to detoxify the text step-by-step. Automatic and human evaluation on two benchmarks reveals that by training with Detox-Chain, six LLMs scaling from 1B to 33B can obtain significant detoxification and generation improvement. Our code and data are available at https://github.com/CODINNLG/Detox-CoT. Warning: examples in the paper may contain uncensored offensive content.