@inproceedings{wang-etal-2023-zemi,
title = "Zemi: Learning Zero-Shot Semi-Parametric Language Models from Multiple Tasks",
author = "Wang, Zhenhailong and
Pan, Xiaoman and
Yu, Dian and
Yu, Dong and
Chen, Jianshu and
Ji, Heng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-acl.246",
doi = "10.18653/v1/2023.findings-acl.246",
pages = "3978--4004",
abstract = "Although large language models have exhibited impressive zero-shot ability, the huge model size generally incurs high cost. Recently, semi-parametric language models, which augment a smaller language model with retrieved related background knowledge, alleviate the need for storing everything into the model parameters. Although existing semi-parametric language models have demonstrated promising language modeling capabilities, it remains unclear whether they can exhibit competitive zero-shot abilities as their fully-parametric counterparts. In this work, we introduce Zemi, a semi-parametric language model for zero-shot task generalization. To our best knowledge, this is the first semi-parametric language model that can demonstrate strong zero-shot performance on a wide range of held-out unseen tasks. We train Zemi with semi-parametric multitask training, which shows significant improvement compared with the parametric multitask training as proposed by T0. Specifically, during both training and inference, Zemi is equipped with a retrieval system based on the unlabeled pretraining corpus of our backbone model. To address the unique challenges from large-scale retrieval, we further propose a novel retrieval-augmentation fusion module that can effectively incorporate noisy retrieved documents. Finally, we show detailed analysis and ablation studies on the key ingredients towards building effective zero-shot semi-parametric language models. Notably, our proposed Zemi{\_}Large model outperforms T0-3B by 16{\%} across seven diverse evaluation tasks while being 3.8x smaller in scale.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2023-zemi">
<titleInfo>
<title>Zemi: Learning Zero-Shot Semi-Parametric Language Models from Multiple Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhenhailong</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoman</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dian</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dong</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianshu</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Heng</namePart>
<namePart type="family">Ji</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>Findings of the Association for Computational Linguistics: ACL 2023</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>Although large language models have exhibited impressive zero-shot ability, the huge model size generally incurs high cost. Recently, semi-parametric language models, which augment a smaller language model with retrieved related background knowledge, alleviate the need for storing everything into the model parameters. Although existing semi-parametric language models have demonstrated promising language modeling capabilities, it remains unclear whether they can exhibit competitive zero-shot abilities as their fully-parametric counterparts. In this work, we introduce Zemi, a semi-parametric language model for zero-shot task generalization. To our best knowledge, this is the first semi-parametric language model that can demonstrate strong zero-shot performance on a wide range of held-out unseen tasks. We train Zemi with semi-parametric multitask training, which shows significant improvement compared with the parametric multitask training as proposed by T0. Specifically, during both training and inference, Zemi is equipped with a retrieval system based on the unlabeled pretraining corpus of our backbone model. To address the unique challenges from large-scale retrieval, we further propose a novel retrieval-augmentation fusion module that can effectively incorporate noisy retrieved documents. Finally, we show detailed analysis and ablation studies on the key ingredients towards building effective zero-shot semi-parametric language models. Notably, our proposed Zemi_Large model outperforms T0-3B by 16% across seven diverse evaluation tasks while being 3.8x smaller in scale.</abstract>
<identifier type="citekey">wang-etal-2023-zemi</identifier>
<identifier type="doi">10.18653/v1/2023.findings-acl.246</identifier>
<location>
<url>https://aclanthology.org/2023.findings-acl.246</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>3978</start>
<end>4004</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Zemi: Learning Zero-Shot Semi-Parametric Language Models from Multiple Tasks
%A Wang, Zhenhailong
%A Pan, Xiaoman
%A Yu, Dian
%A Yu, Dong
%A Chen, Jianshu
%A Ji, Heng
%Y Rogers, Anna
%Y Boyd-Graber, Jordan
%Y Okazaki, Naoaki
%S Findings of the Association for Computational Linguistics: ACL 2023
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F wang-etal-2023-zemi
%X Although large language models have exhibited impressive zero-shot ability, the huge model size generally incurs high cost. Recently, semi-parametric language models, which augment a smaller language model with retrieved related background knowledge, alleviate the need for storing everything into the model parameters. Although existing semi-parametric language models have demonstrated promising language modeling capabilities, it remains unclear whether they can exhibit competitive zero-shot abilities as their fully-parametric counterparts. In this work, we introduce Zemi, a semi-parametric language model for zero-shot task generalization. To our best knowledge, this is the first semi-parametric language model that can demonstrate strong zero-shot performance on a wide range of held-out unseen tasks. We train Zemi with semi-parametric multitask training, which shows significant improvement compared with the parametric multitask training as proposed by T0. Specifically, during both training and inference, Zemi is equipped with a retrieval system based on the unlabeled pretraining corpus of our backbone model. To address the unique challenges from large-scale retrieval, we further propose a novel retrieval-augmentation fusion module that can effectively incorporate noisy retrieved documents. Finally, we show detailed analysis and ablation studies on the key ingredients towards building effective zero-shot semi-parametric language models. Notably, our proposed Zemi_Large model outperforms T0-3B by 16% across seven diverse evaluation tasks while being 3.8x smaller in scale.
%R 10.18653/v1/2023.findings-acl.246
%U https://aclanthology.org/2023.findings-acl.246
%U https://doi.org/10.18653/v1/2023.findings-acl.246
%P 3978-4004
Markdown (Informal)
[Zemi: Learning Zero-Shot Semi-Parametric Language Models from Multiple Tasks](https://aclanthology.org/2023.findings-acl.246) (Wang et al., Findings 2023)
ACL