Generative Context Distillation

Prompts used in recent large language model based applications are often fixed and lengthy, leading to significant computational overhead. To address this challenge, we propose Generative Context Distillation (GCD), a lightweight prompt internalization method that employs a joint training approach....

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2024-11
Hauptverfasser: Shin, Haebin, Ji, Lei, Gong, Yeyun, Kim, Sungdong, Choi, Eunbi, Seo, Minjoon
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Shin, Haebin
Ji, Lei
Gong, Yeyun
Kim, Sungdong
Choi, Eunbi
Seo, Minjoon
description Prompts used in recent large language model based applications are often fixed and lengthy, leading to significant computational overhead. To address this challenge, we propose Generative Context Distillation (GCD), a lightweight prompt internalization method that employs a joint training approach. This method not only replicates the behavior of models with prompt inputs but also generates the content of the prompt along with reasons for why the model's behavior should change accordingly. We demonstrate that our approach effectively internalizes complex prompts across various agent-based application scenarios. For effective training without interactions with the dedicated environments, we introduce a data synthesis technique that autonomously collects conversational datasets by swapping the roles of the agent and environment. This method is especially useful in scenarios where only a predefined prompt is available without a corresponding training dataset. By internalizing complex prompts, Generative Context Distillation enables high-performance and efficient inference without the need for explicit prompts.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_3133055450</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3133055450</sourcerecordid><originalsourceid>FETCH-proquest_journals_31330554503</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSQd0_NSy1KLMksS1Vwzs8rSa0oUXDJLC7JzMkBCubn8TCwpiXmFKfyQmluBmU31xBnD92CovzC0tTikvis_NKiPKBUvLGhsbGBqamJqYExcaoA5T8s_Q</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3133055450</pqid></control><display><type>article</type><title>Generative Context Distillation</title><source>Free E- Journals</source><creator>Shin, Haebin ; Ji, Lei ; Gong, Yeyun ; Kim, Sungdong ; Choi, Eunbi ; Seo, Minjoon</creator><creatorcontrib>Shin, Haebin ; Ji, Lei ; Gong, Yeyun ; Kim, Sungdong ; Choi, Eunbi ; Seo, Minjoon</creatorcontrib><description>Prompts used in recent large language model based applications are often fixed and lengthy, leading to significant computational overhead. To address this challenge, we propose Generative Context Distillation (GCD), a lightweight prompt internalization method that employs a joint training approach. This method not only replicates the behavior of models with prompt inputs but also generates the content of the prompt along with reasons for why the model's behavior should change accordingly. We demonstrate that our approach effectively internalizes complex prompts across various agent-based application scenarios. For effective training without interactions with the dedicated environments, we introduce a data synthesis technique that autonomously collects conversational datasets by swapping the roles of the agent and environment. This method is especially useful in scenarios where only a predefined prompt is available without a corresponding training dataset. By internalizing complex prompts, Generative Context Distillation enables high-performance and efficient inference without the need for explicit prompts.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Context ; Datasets ; Large language models</subject><ispartof>arXiv.org, 2024-11</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><link.rule.ids>776,780</link.rule.ids></links><search><creatorcontrib>Shin, Haebin</creatorcontrib><creatorcontrib>Ji, Lei</creatorcontrib><creatorcontrib>Gong, Yeyun</creatorcontrib><creatorcontrib>Kim, Sungdong</creatorcontrib><creatorcontrib>Choi, Eunbi</creatorcontrib><creatorcontrib>Seo, Minjoon</creatorcontrib><title>Generative Context Distillation</title><title>arXiv.org</title><description>Prompts used in recent large language model based applications are often fixed and lengthy, leading to significant computational overhead. To address this challenge, we propose Generative Context Distillation (GCD), a lightweight prompt internalization method that employs a joint training approach. This method not only replicates the behavior of models with prompt inputs but also generates the content of the prompt along with reasons for why the model's behavior should change accordingly. We demonstrate that our approach effectively internalizes complex prompts across various agent-based application scenarios. For effective training without interactions with the dedicated environments, we introduce a data synthesis technique that autonomously collects conversational datasets by swapping the roles of the agent and environment. This method is especially useful in scenarios where only a predefined prompt is available without a corresponding training dataset. By internalizing complex prompts, Generative Context Distillation enables high-performance and efficient inference without the need for explicit prompts.</description><subject>Context</subject><subject>Datasets</subject><subject>Large language models</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSQd0_NSy1KLMksS1Vwzs8rSa0oUXDJLC7JzMkBCubn8TCwpiXmFKfyQmluBmU31xBnD92CovzC0tTikvis_NKiPKBUvLGhsbGBqamJqYExcaoA5T8s_Q</recordid><startdate>20241124</startdate><enddate>20241124</enddate><creator>Shin, Haebin</creator><creator>Ji, Lei</creator><creator>Gong, Yeyun</creator><creator>Kim, Sungdong</creator><creator>Choi, Eunbi</creator><creator>Seo, Minjoon</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>20241124</creationdate><title>Generative Context Distillation</title><author>Shin, Haebin ; Ji, Lei ; Gong, Yeyun ; Kim, Sungdong ; Choi, Eunbi ; Seo, Minjoon</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_31330554503</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Context</topic><topic>Datasets</topic><topic>Large language models</topic><toplevel>online_resources</toplevel><creatorcontrib>Shin, Haebin</creatorcontrib><creatorcontrib>Ji, Lei</creatorcontrib><creatorcontrib>Gong, Yeyun</creatorcontrib><creatorcontrib>Kim, Sungdong</creatorcontrib><creatorcontrib>Choi, Eunbi</creatorcontrib><creatorcontrib>Seo, Minjoon</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</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>Shin, Haebin</au><au>Ji, Lei</au><au>Gong, Yeyun</au><au>Kim, Sungdong</au><au>Choi, Eunbi</au><au>Seo, Minjoon</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Generative Context Distillation</atitle><jtitle>arXiv.org</jtitle><date>2024-11-24</date><risdate>2024</risdate><eissn>2331-8422</eissn><abstract>Prompts used in recent large language model based applications are often fixed and lengthy, leading to significant computational overhead. To address this challenge, we propose Generative Context Distillation (GCD), a lightweight prompt internalization method that employs a joint training approach. This method not only replicates the behavior of models with prompt inputs but also generates the content of the prompt along with reasons for why the model's behavior should change accordingly. We demonstrate that our approach effectively internalizes complex prompts across various agent-based application scenarios. For effective training without interactions with the dedicated environments, we introduce a data synthesis technique that autonomously collects conversational datasets by swapping the roles of the agent and environment. This method is especially useful in scenarios where only a predefined prompt is available without a corresponding training dataset. By internalizing complex prompts, Generative Context Distillation enables high-performance and efficient inference without the need for explicit prompts.</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, 2024-11
issn 2331-8422
language eng
recordid cdi_proquest_journals_3133055450
source Free E- Journals
subjects Context
Datasets
Large language models
title Generative Context Distillation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-03T20%3A40%3A34IST&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=Generative%20Context%20Distillation&rft.jtitle=arXiv.org&rft.au=Shin,%20Haebin&rft.date=2024-11-24&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E3133055450%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3133055450&rft_id=info:pmid/&rfr_iscdi=true