Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation

It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledge transfer. In order to enhance the guidance of the teacher training process, we...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2024-05
Main Authors: Ren, Yuxin, Zhong, Zihan, Shi, Xingjian, Zhu, Yi, Yuan, Chun, Li, Mu
Format: Article
Language:English
Subjects:
Distillation
Knowledge management
Learning
Teachers
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Ren, Yuxin
Zhong, Zihan
Shi, Xingjian
Zhu, Yi
Yuan, Chun
Li, Mu
description It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledge transfer. In order to enhance the guidance of the teacher training process, we introduce the concept of distillation influence to determine the impact of distillation from each training sample on the student's generalization ability. In this paper, we propose Learning Good Teacher Matters (LGTM), an efficient training technique for incorporating distillation influence into the teacher's learning process. By prioritizing samples that are likely to enhance the student's generalization ability, our LGTM outperforms 10 common knowledge distillation baselines on 6 text classification tasks in the GLUE benchmark.
format article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2814620426</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2814620426</sourcerecordid><originalsourceid>FETCH-proquest_journals_28146204263</originalsourceid><addsrcrecordid>eNqNi0sKwjAUAIMgWLR3CLhwVUhf2tq1PxS7EbsvwcaSEvI0L9Xra8EDuJrFzExYBFKmSZkBzFhM1AshoFhDnsuIXWplLHrjOn5yFPxwCwYd8YD8GoZWu7AiXmnl3ZhU-qUt8Q0iBeJnh2-r207znaFgrFXju2DTu7Kk4x_nbHnY19tj8vD4HDSFpsfBu69qoEyzAkQGhfyv-gD20j_j</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2814620426</pqid></control><display><type>article</type><title>Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation</title><source>Publicly Available Content Database</source><creator>Ren, Yuxin ; Zhong, Zihan ; Shi, Xingjian ; Zhu, Yi ; Yuan, Chun ; Li, Mu</creator><creatorcontrib>Ren, Yuxin ; Zhong, Zihan ; Shi, Xingjian ; Zhu, Yi ; Yuan, Chun ; Li, Mu</creatorcontrib><description>It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledge transfer. In order to enhance the guidance of the teacher training process, we introduce the concept of distillation influence to determine the impact of distillation from each training sample on the student's generalization ability. In this paper, we propose Learning Good Teacher Matters (LGTM), an efficient training technique for incorporating distillation influence into the teacher's learning process. By prioritizing samples that are likely to enhance the student's generalization ability, our LGTM outperforms 10 common knowledge distillation baselines on 6 text classification tasks in the GLUE benchmark.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Distillation ; Knowledge management ; Learning ; Teachers ; Training</subject><ispartof>arXiv.org, 2024-05</ispartof><rights>2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://www.proquest.com/docview/2814620426?pq-origsite=primo$$EHTML$$P50$$Gproquest$$Hfree_for_read</linktohtml><link.rule.ids>776,780,25732,36991,44569</link.rule.ids></links><search><creatorcontrib>Ren, Yuxin</creatorcontrib><creatorcontrib>Zhong, Zihan</creatorcontrib><creatorcontrib>Shi, Xingjian</creatorcontrib><creatorcontrib>Zhu, Yi</creatorcontrib><creatorcontrib>Yuan, Chun</creatorcontrib><creatorcontrib>Li, Mu</creatorcontrib><title>Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation</title><title>arXiv.org</title><description>It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledge transfer. In order to enhance the guidance of the teacher training process, we introduce the concept of distillation influence to determine the impact of distillation from each training sample on the student's generalization ability. In this paper, we propose Learning Good Teacher Matters (LGTM), an efficient training technique for incorporating distillation influence into the teacher's learning process. By prioritizing samples that are likely to enhance the student's generalization ability, our LGTM outperforms 10 common knowledge distillation baselines on 6 text classification tasks in the GLUE benchmark.</description><subject>Distillation</subject><subject>Knowledge management</subject><subject>Learning</subject><subject>Teachers</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>PIMPY</sourceid><recordid>eNqNi0sKwjAUAIMgWLR3CLhwVUhf2tq1PxS7EbsvwcaSEvI0L9Xra8EDuJrFzExYBFKmSZkBzFhM1AshoFhDnsuIXWplLHrjOn5yFPxwCwYd8YD8GoZWu7AiXmnl3ZhU-qUt8Q0iBeJnh2-r207znaFgrFXju2DTu7Kk4x_nbHnY19tj8vD4HDSFpsfBu69qoEyzAkQGhfyv-gD20j_j</recordid><startdate>20240515</startdate><enddate>20240515</enddate><creator>Ren, Yuxin</creator><creator>Zhong, Zihan</creator><creator>Shi, Xingjian</creator><creator>Zhu, Yi</creator><creator>Yuan, Chun</creator><creator>Li, Mu</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20240515</creationdate><title>Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation</title><author>Ren, Yuxin ; Zhong, Zihan ; Shi, Xingjian ; Zhu, Yi ; Yuan, Chun ; Li, Mu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_28146204263</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Distillation</topic><topic>Knowledge management</topic><topic>Learning</topic><topic>Teachers</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Ren, Yuxin</creatorcontrib><creatorcontrib>Zhong, Zihan</creatorcontrib><creatorcontrib>Shi, Xingjian</creatorcontrib><creatorcontrib>Zhu, Yi</creatorcontrib><creatorcontrib>Yuan, Chun</creatorcontrib><creatorcontrib>Li, Mu</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Ren, Yuxin</au><au>Zhong, Zihan</au><au>Shi, Xingjian</au><au>Zhu, Yi</au><au>Yuan, Chun</au><au>Li, Mu</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation</atitle><jtitle>arXiv.org</jtitle><date>2024-05-15</date><risdate>2024</risdate><eissn>2331-8422</eissn><abstract>It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledge transfer. In order to enhance the guidance of the teacher training process, we introduce the concept of distillation influence to determine the impact of distillation from each training sample on the student's generalization ability. In this paper, we propose Learning Good Teacher Matters (LGTM), an efficient training technique for incorporating distillation influence into the teacher's learning process. By prioritizing samples that are likely to enhance the student's generalization ability, our LGTM outperforms 10 common knowledge distillation baselines on 6 text classification tasks in the GLUE benchmark.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2024-05
issn 2331-8422
language eng
recordid cdi_proquest_journals_2814620426
source Publicly Available Content Database
subjects Distillation
Knowledge management
Learning
Teachers
Training
title Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-25T22%3A47%3A53IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Tailoring%20Instructions%20to%20Student's%20Learning%20Levels%20Boosts%20Knowledge%20Distillation&rft.jtitle=arXiv.org&rft.au=Ren,%20Yuxin&rft.date=2024-05-15&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2814620426%3C/proquest%3E%3Cgrp_id%3Ecdi_FETCH-proquest_journals_28146204263%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2814620426&rft_id=info:pmid/&rfr_iscdi=true