MIL-Decoding: Detoxifying Language Models at Token-Level via Multiple Instance Learning.

Xu Zhang, Xiaojun Wan

Browse the full ACL paper archive.