@inproceedings{arora-etal-2022-director,
title = "Director: Generator-Classifiers For Supervised Language Modeling",
author = "Arora, Kushal and
Shuster, Kurt and
Sukhbaatar, Sainbayar and
Weston, Jason",
editor = "He, Yulan and
Ji, Heng and
Li, Sujian and
Liu, Yang and
Chang, Chua-Hui",
booktitle = "Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = nov,
year = "2022",
address = "Online only",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.aacl-main.39",
doi = "10.18653/v1/2022.aacl-main.39",
pages = "512--526",
abstract = "Current language models achieve low perplexity but their resulting generations still suffer from toxic responses, repetitiveness, and contradictions. The standard language modeling setup fails to address these issues. In this paper, we introduce a new architecture, Director, that consists of a unified generator-classifier with both a language modeling and a classification head for each output token. Training is conducted jointly using both standard language modeling data, and data labeled with desirable and undesirable sequences. Experiments in several settings show that the model has competitive training and decoding speed compared to standard language models while yielding superior results, avoiding undesirable behaviors while maintaining generation quality. It also outperforms existing model guiding approaches in terms of both accuracy and efficiency. Our code is made publicly available.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="arora-etal-2022-director">
<titleInfo>
<title>Director: Generator-Classifiers For Supervised Language Modeling</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kushal</namePart>
<namePart type="family">Arora</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kurt</namePart>
<namePart type="family">Shuster</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sainbayar</namePart>
<namePart type="family">Sukhbaatar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jason</namePart>
<namePart type="family">Weston</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yulan</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Heng</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sujian</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chua-Hui</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online only</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Current language models achieve low perplexity but their resulting generations still suffer from toxic responses, repetitiveness, and contradictions. The standard language modeling setup fails to address these issues. In this paper, we introduce a new architecture, Director, that consists of a unified generator-classifier with both a language modeling and a classification head for each output token. Training is conducted jointly using both standard language modeling data, and data labeled with desirable and undesirable sequences. Experiments in several settings show that the model has competitive training and decoding speed compared to standard language models while yielding superior results, avoiding undesirable behaviors while maintaining generation quality. It also outperforms existing model guiding approaches in terms of both accuracy and efficiency. Our code is made publicly available.</abstract>
<identifier type="citekey">arora-etal-2022-director</identifier>
<identifier type="doi">10.18653/v1/2022.aacl-main.39</identifier>
<location>
<url>https://aclanthology.org/2022.aacl-main.39</url>
</location>
<part>
<date>2022-11</date>
<extent unit="page">
<start>512</start>
<end>526</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Director: Generator-Classifiers For Supervised Language Modeling
%A Arora, Kushal
%A Shuster, Kurt
%A Sukhbaatar, Sainbayar
%A Weston, Jason
%Y He, Yulan
%Y Ji, Heng
%Y Li, Sujian
%Y Liu, Yang
%Y Chang, Chua-Hui
%S Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
%D 2022
%8 November
%I Association for Computational Linguistics
%C Online only
%F arora-etal-2022-director
%X Current language models achieve low perplexity but their resulting generations still suffer from toxic responses, repetitiveness, and contradictions. The standard language modeling setup fails to address these issues. In this paper, we introduce a new architecture, Director, that consists of a unified generator-classifier with both a language modeling and a classification head for each output token. Training is conducted jointly using both standard language modeling data, and data labeled with desirable and undesirable sequences. Experiments in several settings show that the model has competitive training and decoding speed compared to standard language models while yielding superior results, avoiding undesirable behaviors while maintaining generation quality. It also outperforms existing model guiding approaches in terms of both accuracy and efficiency. Our code is made publicly available.
%R 10.18653/v1/2022.aacl-main.39
%U https://aclanthology.org/2022.aacl-main.39
%U https://doi.org/10.18653/v1/2022.aacl-main.39
%P 512-526
Markdown (Informal)
[Director: Generator-Classifiers For Supervised Language Modeling](https://aclanthology.org/2022.aacl-main.39) (Arora et al., AACL-IJCNLP 2022)
ACL
- Kushal Arora, Kurt Shuster, Sainbayar Sukhbaatar, and Jason Weston. 2022. Director: Generator-Classifiers For Supervised Language Modeling. In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 512–526, Online only. Association for Computational Linguistics.