Adding Instructions during Pretraining: Effective way of Controlling Toxicity in Language Models

Adding Instructions during Pretraining: Effective way of Controlling Toxicity in Language Models Shrimai Prabhumoye author Mostofa Patwary author Mohammad Shoeybi author Bryan Catanzaro author 2023-05 text Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics Andreas Vlachos editor Isabelle Augenstein editor Association for Computational Linguistics Dubrovnik, Croatia conference publication prabhumoye-etal-2023-adding 10.18653/v1/2023.eacl-main.193 https://aclanthology.org/2023.eacl-main.193/ 2023-05 2636 2651