@inproceedings{hong-etal-2024-accelerating,
title = "Accelerating Multilingual Language Model for Excessively Tokenized Languages",
author = "Hong, Jimin and
Lee, Gibbeum and
Cho, Jaewoong",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.660/",
doi = "10.18653/v1/2024.findings-acl.660",
pages = "11095--11111",
abstract = "Recent advancements in large language models (LLMs) have remarkably enhanced performances on a variety of tasks in multiple languages. However, tokenizers in LLMs trained primarily on English-centric corpora often overly fragment a text into character or Unicode-level tokens in non-Roman alphabetic languages, leading to inefficient text generation.We introduce a simple yet effective framework to accelerate text generation in such languages. Our approach involves employing a new language model head with a vocabulary set tailored to a specific target language for a pre-trained LLM. This is followed by fine-tuning the new head while incorporating a verification step to ensure the model`s performance is preserved.We show that this targeted fine-tuning, while freezing other model parameters, effectively reduces token fragmentation for the target language. Our extensive experiments demonstrate that the proposed framework increases the generation speed by a factor of 1.7 while maintaining the performance of pre-trained multilingual models on target monolingual tasks."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hong-etal-2024-accelerating">
<titleInfo>
<title>Accelerating Multilingual Language Model for Excessively Tokenized Languages</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jimin</namePart>
<namePart type="family">Hong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gibbeum</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jaewoong</namePart>
<namePart type="family">Cho</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2024</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lun-Wei</namePart>
<namePart type="family">Ku</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andre</namePart>
<namePart type="family">Martins</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Srikumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Recent advancements in large language models (LLMs) have remarkably enhanced performances on a variety of tasks in multiple languages. However, tokenizers in LLMs trained primarily on English-centric corpora often overly fragment a text into character or Unicode-level tokens in non-Roman alphabetic languages, leading to inefficient text generation.We introduce a simple yet effective framework to accelerate text generation in such languages. Our approach involves employing a new language model head with a vocabulary set tailored to a specific target language for a pre-trained LLM. This is followed by fine-tuning the new head while incorporating a verification step to ensure the model‘s performance is preserved.We show that this targeted fine-tuning, while freezing other model parameters, effectively reduces token fragmentation for the target language. Our extensive experiments demonstrate that the proposed framework increases the generation speed by a factor of 1.7 while maintaining the performance of pre-trained multilingual models on target monolingual tasks.</abstract>
<identifier type="citekey">hong-etal-2024-accelerating</identifier>
<identifier type="doi">10.18653/v1/2024.findings-acl.660</identifier>
<location>
<url>https://aclanthology.org/2024.findings-acl.660/</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>11095</start>
<end>11111</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Accelerating Multilingual Language Model for Excessively Tokenized Languages
%A Hong, Jimin
%A Lee, Gibbeum
%A Cho, Jaewoong
%Y Ku, Lun-Wei
%Y Martins, Andre
%Y Srikumar, Vivek
%S Findings of the Association for Computational Linguistics: ACL 2024
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F hong-etal-2024-accelerating
%X Recent advancements in large language models (LLMs) have remarkably enhanced performances on a variety of tasks in multiple languages. However, tokenizers in LLMs trained primarily on English-centric corpora often overly fragment a text into character or Unicode-level tokens in non-Roman alphabetic languages, leading to inefficient text generation.We introduce a simple yet effective framework to accelerate text generation in such languages. Our approach involves employing a new language model head with a vocabulary set tailored to a specific target language for a pre-trained LLM. This is followed by fine-tuning the new head while incorporating a verification step to ensure the model‘s performance is preserved.We show that this targeted fine-tuning, while freezing other model parameters, effectively reduces token fragmentation for the target language. Our extensive experiments demonstrate that the proposed framework increases the generation speed by a factor of 1.7 while maintaining the performance of pre-trained multilingual models on target monolingual tasks.
%R 10.18653/v1/2024.findings-acl.660
%U https://aclanthology.org/2024.findings-acl.660/
%U https://doi.org/10.18653/v1/2024.findings-acl.660
%P 11095-11111
Markdown (Informal)
[Accelerating Multilingual Language Model for Excessively Tokenized Languages](https://aclanthology.org/2024.findings-acl.660/) (Hong et al., Findings 2024)
- Accelerating Multilingual Language Model for Excessively Tokenized Languages (Hong et al., Findings 2024)
ACL
- Jimin Hong, Gibbeum Lee, and Jaewoong Cho. 2024. Accelerating Multilingual Language Model for Excessively Tokenized Languages. In Findings of the Association for Computational Linguistics: ACL 2024, pages 11095–11111, Bangkok, Thailand. Association for Computational Linguistics.