@inproceedings{xu-etal-2023-universal,
title = "A Universal Discriminator for Zero-Shot Generalization",
author = "Xu, Haike and
Lin, Zongyu and
Zhou, Jing and
Zheng, Yanan and
Yang, Zhilin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.589/",
doi = "10.18653/v1/2023.acl-long.589",
pages = "10559--10575",
abstract = "Generative modeling has been the dominant approach for large-scale pretraining and zero-shot generalization. In this work, we challenge this convention by showing that discriminative approaches perform substantially better than generative ones on a large number of NLP tasks. Technically, we train a single discriminator to predict whether a text sample comes from the true data distribution, similar to GANs. Since many NLP tasks can be formulated as selecting from a few options, we use this discriminator to predict the concatenation of input and which option has the highest probability of coming from the true data distribution. This simple formulation achieves state-of-the-art zero-shot results on the T0 benchmark, outperforming T0 by 16.0{\%}, 7.8{\%}, and 11.5{\%} respectively on different scales. In the finetuning setting, our approach also achieves new state-of-the-art results on a wide range of NLP tasks, with only 1/4 parameters of previous methods. Meanwhile, our approach requires minimal prompting efforts, which largely improves robustness and is essential for real-world applications. Furthermore, we also jointly train a generalized UD in combination with generative tasks, which maintains its advantage on discriminative tasks and simultaneously works on generative tasks."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xu-etal-2023-universal">
<titleInfo>
<title>A Universal Discriminator for Zero-Shot Generalization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Haike</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zongyu</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jing</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanan</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhilin</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anna</namePart>
<namePart type="family">Rogers</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jordan</namePart>
<namePart type="family">Boyd-Graber</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Naoaki</namePart>
<namePart type="family">Okazaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Generative modeling has been the dominant approach for large-scale pretraining and zero-shot generalization. In this work, we challenge this convention by showing that discriminative approaches perform substantially better than generative ones on a large number of NLP tasks. Technically, we train a single discriminator to predict whether a text sample comes from the true data distribution, similar to GANs. Since many NLP tasks can be formulated as selecting from a few options, we use this discriminator to predict the concatenation of input and which option has the highest probability of coming from the true data distribution. This simple formulation achieves state-of-the-art zero-shot results on the T0 benchmark, outperforming T0 by 16.0%, 7.8%, and 11.5% respectively on different scales. In the finetuning setting, our approach also achieves new state-of-the-art results on a wide range of NLP tasks, with only 1/4 parameters of previous methods. Meanwhile, our approach requires minimal prompting efforts, which largely improves robustness and is essential for real-world applications. Furthermore, we also jointly train a generalized UD in combination with generative tasks, which maintains its advantage on discriminative tasks and simultaneously works on generative tasks.</abstract>
<identifier type="citekey">xu-etal-2023-universal</identifier>
<identifier type="doi">10.18653/v1/2023.acl-long.589</identifier>
<location>
<url>https://aclanthology.org/2023.acl-long.589/</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>10559</start>
<end>10575</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Universal Discriminator for Zero-Shot Generalization
%A Xu, Haike
%A Lin, Zongyu
%A Zhou, Jing
%A Zheng, Yanan
%A Yang, Zhilin
%Y Rogers, Anna
%Y Boyd-Graber, Jordan
%Y Okazaki, Naoaki
%S Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F xu-etal-2023-universal
%X Generative modeling has been the dominant approach for large-scale pretraining and zero-shot generalization. In this work, we challenge this convention by showing that discriminative approaches perform substantially better than generative ones on a large number of NLP tasks. Technically, we train a single discriminator to predict whether a text sample comes from the true data distribution, similar to GANs. Since many NLP tasks can be formulated as selecting from a few options, we use this discriminator to predict the concatenation of input and which option has the highest probability of coming from the true data distribution. This simple formulation achieves state-of-the-art zero-shot results on the T0 benchmark, outperforming T0 by 16.0%, 7.8%, and 11.5% respectively on different scales. In the finetuning setting, our approach also achieves new state-of-the-art results on a wide range of NLP tasks, with only 1/4 parameters of previous methods. Meanwhile, our approach requires minimal prompting efforts, which largely improves robustness and is essential for real-world applications. Furthermore, we also jointly train a generalized UD in combination with generative tasks, which maintains its advantage on discriminative tasks and simultaneously works on generative tasks.
%R 10.18653/v1/2023.acl-long.589
%U https://aclanthology.org/2023.acl-long.589/
%U https://doi.org/10.18653/v1/2023.acl-long.589
%P 10559-10575
Markdown (Informal)
[A Universal Discriminator for Zero-Shot Generalization](https://aclanthology.org/2023.acl-long.589/) (Xu et al., ACL 2023)
ACL
- Haike Xu, Zongyu Lin, Jing Zhou, Yanan Zheng, and Zhilin Yang. 2023. A Universal Discriminator for Zero-Shot Generalization. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 10559–10575, Toronto, Canada. Association for Computational Linguistics.