@inproceedings{zhao-etal-2024-set,
title = "Set the Clock: Temporal Alignment of Pretrained Language Models",
author = "Zhao, Bowen and
Brumbaugh, Zander and
Wang, Yizhong and
Hajishirzi, Hannaneh and
Smith, Noah",
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 and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.892",
pages = "15015--15040",
abstract = "Language models (LMs) are trained on web text originating from many points in time and, in general, without any explicit temporal grounding. This work investigates the temporal chaos of pretrained LMs and explores various methods to align their internal knowledge to a target time, which we call {``}temporal alignment.{''} To do this, we first automatically construct a dataset containing 20K time-sensitive questions and their answers for each year from 2000 to 2023. Based on this dataset, we empirically show that pretrained LMs (e.g., LLaMa2), despite having a recent pretraining cutoff (e.g., 2022), mostly answer questions using earlier knowledge (e.g., in 2019). We then develop several methods, from prompting to finetuning, to align LMs to use their most recent knowledge when answering questions, and investigate various factors in this alignment. Our experiments demonstrate that aligning LLaMa2 to the year 2022 can enhance its performance by up to 62{\%} according to that year{'}s answers. This improvement occurs even without explicitly mentioning time information, indicating the possibility of aligning models{'} internal sense of time after pretraining. Finally, we find that alignment to a historical time is also possible, with up to $2.8\times$ the performance of the unaligned LM in 2010 if finetuning models to that year. These findings hint at the sophistication of LMs{'} internal knowledge organization and the necessity of tuning them properly.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhao-etal-2024-set">
<titleInfo>
<title>Set the Clock: Temporal Alignment of Pretrained Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bowen</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zander</namePart>
<namePart type="family">Brumbaugh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yizhong</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hannaneh</namePart>
<namePart type="family">Hajishirzi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Noah</namePart>
<namePart type="family">Smith</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 and virtual meeting</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Language models (LMs) are trained on web text originating from many points in time and, in general, without any explicit temporal grounding. This work investigates the temporal chaos of pretrained LMs and explores various methods to align their internal knowledge to a target time, which we call “temporal alignment.” To do this, we first automatically construct a dataset containing 20K time-sensitive questions and their answers for each year from 2000 to 2023. Based on this dataset, we empirically show that pretrained LMs (e.g., LLaMa2), despite having a recent pretraining cutoff (e.g., 2022), mostly answer questions using earlier knowledge (e.g., in 2019). We then develop several methods, from prompting to finetuning, to align LMs to use their most recent knowledge when answering questions, and investigate various factors in this alignment. Our experiments demonstrate that aligning LLaMa2 to the year 2022 can enhance its performance by up to 62% according to that year’s answers. This improvement occurs even without explicitly mentioning time information, indicating the possibility of aligning models’ internal sense of time after pretraining. Finally, we find that alignment to a historical time is also possible, with up to 2.8\times the performance of the unaligned LM in 2010 if finetuning models to that year. These findings hint at the sophistication of LMs’ internal knowledge organization and the necessity of tuning them properly.</abstract>
<identifier type="citekey">zhao-etal-2024-set</identifier>
<location>
<url>https://aclanthology.org/2024.findings-acl.892</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>15015</start>
<end>15040</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Set the Clock: Temporal Alignment of Pretrained Language Models
%A Zhao, Bowen
%A Brumbaugh, Zander
%A Wang, Yizhong
%A Hajishirzi, Hannaneh
%A Smith, Noah
%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 and virtual meeting
%F zhao-etal-2024-set
%X Language models (LMs) are trained on web text originating from many points in time and, in general, without any explicit temporal grounding. This work investigates the temporal chaos of pretrained LMs and explores various methods to align their internal knowledge to a target time, which we call “temporal alignment.” To do this, we first automatically construct a dataset containing 20K time-sensitive questions and their answers for each year from 2000 to 2023. Based on this dataset, we empirically show that pretrained LMs (e.g., LLaMa2), despite having a recent pretraining cutoff (e.g., 2022), mostly answer questions using earlier knowledge (e.g., in 2019). We then develop several methods, from prompting to finetuning, to align LMs to use their most recent knowledge when answering questions, and investigate various factors in this alignment. Our experiments demonstrate that aligning LLaMa2 to the year 2022 can enhance its performance by up to 62% according to that year’s answers. This improvement occurs even without explicitly mentioning time information, indicating the possibility of aligning models’ internal sense of time after pretraining. Finally, we find that alignment to a historical time is also possible, with up to 2.8\times the performance of the unaligned LM in 2010 if finetuning models to that year. These findings hint at the sophistication of LMs’ internal knowledge organization and the necessity of tuning them properly.
%U https://aclanthology.org/2024.findings-acl.892
%P 15015-15040
Markdown (Informal)
[Set the Clock: Temporal Alignment of Pretrained Language Models](https://aclanthology.org/2024.findings-acl.892) (Zhao et al., Findings 2024)
ACL
- Bowen Zhao, Zander Brumbaugh, Yizhong Wang, Hannaneh Hajishirzi, and Noah Smith. 2024. Set the Clock: Temporal Alignment of Pretrained Language Models. In Findings of the Association for Computational Linguistics ACL 2024, pages 15015–15040, Bangkok, Thailand and virtual meeting. Association for Computational Linguistics.