Apr 1, 2024 · In this work, we propose to balance safety and helpfulness in diverse use cases by controlling both attributes in LLM.
As large language models (LLMs) become easily accessible nowadays, the trade-off between safety and helpfulness can significantly impact user experience. A ...
As large language models (LLMs) become easily accessible nowadays, the trade-off between safety and helpfulness can significantly impact user experience.
Apr 2, 2024 · The paper proposes a framework to balance the safety and helpfulness of large language model (LLM) responses.
As large language models (LLMs) become easily accessible nowadays, the trade-off between safety and helpfulness can significantly impact user experience.
共同作者 ; Towards Safety and Helpfulness Balanced Responses via Controllable Large Language Models. YL Tuan, X Chen, EM Smith, L Martin, S Batra, A Celikyilmaz, ...
Oct 15, 2024 · This paper investigates how to balance safety enhancement and over-safety mitigation while retaining model utility during post-safety alignment.
In this paper, we introduce the BEAVERTAILS dataset, aimed at fostering research on safety alignment in large language models (LLMs). This dataset uniquely.
As large language models (LLMs) become easily accessible nowadays, the trade-off between safety and helpfulness can significantly impact user experience. Paper
As large language models (LLMs) become easily accessible nowadays, the trade-off between safety and helpfulness can significantly impact user experience. Paper