Automatic Construction of a Korean Toxic Instruction Dataset for Ethical Tuning of Large Language Models (2311.18215v1)

Abstract: Caution: this paper may include material that could be offensive or distressing. The advent of LLMs necessitates the development of training approaches that mitigate the generation of unethical language and aptly manage toxic user queries. Given the challenges related to human labor and the scarcity of data, we present KoTox, comprising 39K unethical instruction-output pairs. This collection of automatically generated toxic instructions refines the training of LLMs and establishes a foundational framework for improving LLMs' ethical awareness and response to various toxic inputs, promoting more secure and responsible interactions in NLP applications.