<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mallinson-etal-2022-edit5">
    <titleInfo>
        <title>EdiT5: Semi-Autoregressive Text Editing with T5 Warm-Start</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Jonathan</namePart>
        <namePart type="family">Mallinson</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jakub</namePart>
        <namePart type="family">Adamek</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Eric</namePart>
        <namePart type="family">Malmi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Aliaksei</namePart>
        <namePart type="family">Severyn</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2022-12</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: EMNLP 2022</title>
        </titleInfo>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>We present EdiT5 - a novel semi-autoregressive text-editing approach designed to combine the strengths of non-autoregressive text-editing and autoregressive decoding. EdiT5 is faster at inference times than conventional sequence-to-sequence (seq2seq) models, while being capable of modeling flexible input-output transformations.This is achieved by decomposing the generation process into three sub-tasks: (1) tagging to decide on the subset of input tokens to be preserved in the output, (2) re-ordering to define their order in the output text, and (3) insertion to infill the missing tokens that are not present in the input. The tagging and re-ordering steps, which are responsible for generating the largest portion of the output, are non-autoregressive, while the insertion uses an autoregressive decoder.Depending on the task, EdiT5 requires significantly fewer autoregressive steps demonstrating speedups of up to 25x when compared to classic seq2seq models. Quality-wise, EdiT5 is initialized with a pre-trained T5 checkpoint yielding comparable performance to T5 in high-resource settings and clearly outperforms it on low-resource settings when evaluated on three NLG tasks: Sentence Fusion, Grammatical Error Correction, and Decontextualization.</abstract>
    <identifier type="citekey">mallinson-etal-2022-edit5</identifier>
    <location>
        <url>https://aclanthology.org/2022.findings-emnlp.156</url>
    </location>
    <part>
        <date>2022-12</date>
        <extent unit="page">
            <start>2126</start>
            <end>2138</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T EdiT5: Semi-Autoregressive Text Editing with T5 Warm-Start
%A Mallinson, Jonathan
%A Adamek, Jakub
%A Malmi, Eric
%A Severyn, Aliaksei
%S Findings of the Association for Computational Linguistics: EMNLP 2022
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates
%F mallinson-etal-2022-edit5
%X We present EdiT5 - a novel semi-autoregressive text-editing approach designed to combine the strengths of non-autoregressive text-editing and autoregressive decoding. EdiT5 is faster at inference times than conventional sequence-to-sequence (seq2seq) models, while being capable of modeling flexible input-output transformations.This is achieved by decomposing the generation process into three sub-tasks: (1) tagging to decide on the subset of input tokens to be preserved in the output, (2) re-ordering to define their order in the output text, and (3) insertion to infill the missing tokens that are not present in the input. The tagging and re-ordering steps, which are responsible for generating the largest portion of the output, are non-autoregressive, while the insertion uses an autoregressive decoder.Depending on the task, EdiT5 requires significantly fewer autoregressive steps demonstrating speedups of up to 25x when compared to classic seq2seq models. Quality-wise, EdiT5 is initialized with a pre-trained T5 checkpoint yielding comparable performance to T5 in high-resource settings and clearly outperforms it on low-resource settings when evaluated on three NLG tasks: Sentence Fusion, Grammatical Error Correction, and Decontextualization.
%U https://aclanthology.org/2022.findings-emnlp.156
%P 2126-2138

Markdown (Informal)

[EdiT5: Semi-Autoregressive Text Editing with T5 Warm-Start](https://aclanthology.org/2022.findings-emnlp.156) (Mallinson et al., Findings 2022)

• EdiT5: Semi-Autoregressive Text Editing with T5 Warm-Start (Mallinson et al., Findings 2022)

ACL

• Jonathan Mallinson, Jakub Adamek, Eric Malmi, and Aliaksei Severyn. 2022. EdiT5: Semi-Autoregressive Text Editing with T5 Warm-Start. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 2126–2138, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.