@inproceedings{revilla-llaca-etal-2023-student,
title = "Student-Teacher Prompting for Red Teaming to Improve Guardrails",
author = "Revilla Llaca, Rodrigo and
Leskoschek, Victoria and
Costa Paiva, Vitor and
Lup{\u{a}}u, C{\u{a}}t{\u{a}}lin and
Lippmann, Philip and
Yang, Jie",
editor = "Parrish, Alicia",
booktitle = "Proceedings of the ART of Safety: Workshop on Adversarial testing and Red-Teaming for generative AI",
month = nov,
year = "2023",
address = "Bali, Indonesia",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.artofsafety-1.2",
doi = "10.18653/v1/2023.artofsafety-1.2",
pages = "11--23",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="revilla-llaca-etal-2023-student">
<titleInfo>
<title>Student-Teacher Prompting for Red Teaming to Improve Guardrails</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rodrigo</namePart>
<namePart type="family">Revilla Llaca</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Victoria</namePart>
<namePart type="family">Leskoschek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vitor</namePart>
<namePart type="family">Costa Paiva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cătălin</namePart>
<namePart type="family">Lupău</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Philip</namePart>
<namePart type="family">Lippmann</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the ART of Safety: Workshop on Adversarial testing and Red-Teaming for generative AI</title>
</titleInfo>
<name type="personal">
<namePart type="given">Alicia</namePart>
<namePart type="family">Parrish</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bali, Indonesia</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<identifier type="citekey">revilla-llaca-etal-2023-student</identifier>
<identifier type="doi">10.18653/v1/2023.artofsafety-1.2</identifier>
<location>
<url>https://aclanthology.org/2023.artofsafety-1.2</url>
</location>
<part>
<date>2023-11</date>
<extent unit="page">
<start>11</start>
<end>23</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Student-Teacher Prompting for Red Teaming to Improve Guardrails
%A Revilla Llaca, Rodrigo
%A Leskoschek, Victoria
%A Costa Paiva, Vitor
%A Lupău, Cătălin
%A Lippmann, Philip
%A Yang, Jie
%Y Parrish, Alicia
%S Proceedings of the ART of Safety: Workshop on Adversarial testing and Red-Teaming for generative AI
%D 2023
%8 November
%I Association for Computational Linguistics
%C Bali, Indonesia
%F revilla-llaca-etal-2023-student
%R 10.18653/v1/2023.artofsafety-1.2
%U https://aclanthology.org/2023.artofsafety-1.2
%U https://doi.org/10.18653/v1/2023.artofsafety-1.2
%P 11-23
Markdown (Informal)
[Student-Teacher Prompting for Red Teaming to Improve Guardrails](https://aclanthology.org/2023.artofsafety-1.2) (Revilla Llaca et al., artofsafety-WS 2023)
ACL
- Rodrigo Revilla Llaca, Victoria Leskoschek, Vitor Costa Paiva, Cătălin Lupău, Philip Lippmann, and Jie Yang. 2023. Student-Teacher Prompting for Red Teaming to Improve Guardrails. In Proceedings of the ART of Safety: Workshop on Adversarial testing and Red-Teaming for generative AI, pages 11–23, Bali, Indonesia. Association for Computational Linguistics.