@inproceedings{bai-etal-2021-semantics,
title = "Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with {GPT}2",
author = "Bai, He and
Shi, Peng and
Lin, Jimmy and
Tan, Luchen and
Xiong, Kun and
Gao, Wen and
Liu, Jie and
Li, Ming",
editor = "Kabbara, Jad and
Lin, Haitao and
Paullada, Amandalynne and
Vamvas, Jannis",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop",
month = aug,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-srw.16",
doi = "10.18653/v1/2021.acl-srw.16",
pages = "148--162",
abstract = "The semantics of a text is manifested not only by what is read but also by what is not read. In this article, we will study how those implicit {``}not read{''} information such as end-of-paragraph () and end-of-sequence () affect the quality of text generation. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the in the fine-tuning stage. Experimental results on English story generation show that can lead to higher BLEU scores and lower perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without and during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with .",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="bai-etal-2021-semantics">
<titleInfo>
<title>Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with GPT2</title>
</titleInfo>
<name type="personal">
<namePart type="given">He</namePart>
<namePart type="family">Bai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peng</namePart>
<namePart type="family">Shi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jimmy</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Luchen</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kun</namePart>
<namePart type="family">Xiong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wen</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jad</namePart>
<namePart type="family">Kabbara</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haitao</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Amandalynne</namePart>
<namePart type="family">Paullada</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jannis</namePart>
<namePart type="family">Vamvas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>The semantics of a text is manifested not only by what is read but also by what is not read. In this article, we will study how those implicit “not read” information such as end-of-paragraph () and end-of-sequence () affect the quality of text generation. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the in the fine-tuning stage. Experimental results on English story generation show that can lead to higher BLEU scores and lower perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without and during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with .</abstract>
<identifier type="citekey">bai-etal-2021-semantics</identifier>
<identifier type="doi">10.18653/v1/2021.acl-srw.16</identifier>
<location>
<url>https://aclanthology.org/2021.acl-srw.16</url>
</location>
<part>
<date>2021-08</date>
<extent unit="page">
<start>148</start>
<end>162</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with GPT2
%A Bai, He
%A Shi, Peng
%A Lin, Jimmy
%A Tan, Luchen
%A Xiong, Kun
%A Gao, Wen
%A Liu, Jie
%A Li, Ming
%Y Kabbara, Jad
%Y Lin, Haitao
%Y Paullada, Amandalynne
%Y Vamvas, Jannis
%S Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: Student Research Workshop
%D 2021
%8 August
%I Association for Computational Linguistics
%C Online
%F bai-etal-2021-semantics
%X The semantics of a text is manifested not only by what is read but also by what is not read. In this article, we will study how those implicit “not read” information such as end-of-paragraph () and end-of-sequence () affect the quality of text generation. Specifically, we find that the pre-trained language model GPT2 can generate better continuations by learning to generate the in the fine-tuning stage. Experimental results on English story generation show that can lead to higher BLEU scores and lower perplexity. We also conduct experiments on a self-collected Chinese essay dataset with Chinese-GPT2, a character level LM without and during pre-training. Experimental results show that the Chinese GPT2 can generate better essay endings with .
%R 10.18653/v1/2021.acl-srw.16
%U https://aclanthology.org/2021.acl-srw.16
%U https://doi.org/10.18653/v1/2021.acl-srw.16
%P 148-162
Markdown (Informal)
[Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with GPT2](https://aclanthology.org/2021.acl-srw.16) (Bai et al., ACL-IJCNLP 2021)
ACL