@inproceedings{zeng-etal-2024-bikt,
title = "{B}i{KT}: Enabling Bidirectional Knowledge Transfer Between Pretrained Models and Sequential Downstream Tasks",
author = "Zeng, Hang and
Niu, Chaoyue and
Wu, Fan and
Tang, Shaojie and
Pei, Leihao and
Lv, Chengfei and
Chen, Guihai",
editor = "Al-Onaizan, Yaser and
Bansal, Mohit and
Chen, Yun-Nung",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-emnlp.179",
pages = "3156--3171",
abstract = "Adapting pretrained models to downstream tasks is important in practical applications. Existing frameworks adapt from an initial pretrained model to each downstream task directly, but ignore the sequential nature of the downstream tasks and their feedback effect on the pretrained model. In this work, we propose a new framework, called BiKT, to enable bidirectional knowledge transfer between pretrained models and downstream tasks in rounds. We model each downstream task in the current round as a target task for adaptation and treat all the tasks in the previous rounds as source tasks for feedback. We design a feedback algorithm by multi-task learning over the labeled data of the source tasks, where task-specific prompts are plugged into the backbone network for decoupling task-exclusive knowledge from task-shared knowledge. We further utilize the good initiation of the new backbone network updated in the feedback phase and the trained prompts of the source tasks for adaptation. Evaluation over 9 GLUE datasets, 6 SuperGLUE datasets, and 8 other datasets using models with different pretraining levels and different parameter scales shows remarkable improvement in full-shot and few-shot adaptation settings.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zeng-etal-2024-bikt">
<titleInfo>
<title>BiKT: Enabling Bidirectional Knowledge Transfer Between Pretrained Models and Sequential Downstream Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hang</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chaoyue</namePart>
<namePart type="family">Niu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fan</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shaojie</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leihao</namePart>
<namePart type="family">Pei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chengfei</namePart>
<namePart type="family">Lv</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guihai</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2024</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yaser</namePart>
<namePart type="family">Al-Onaizan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohit</namePart>
<namePart type="family">Bansal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun-Nung</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Miami, Florida, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Adapting pretrained models to downstream tasks is important in practical applications. Existing frameworks adapt from an initial pretrained model to each downstream task directly, but ignore the sequential nature of the downstream tasks and their feedback effect on the pretrained model. In this work, we propose a new framework, called BiKT, to enable bidirectional knowledge transfer between pretrained models and downstream tasks in rounds. We model each downstream task in the current round as a target task for adaptation and treat all the tasks in the previous rounds as source tasks for feedback. We design a feedback algorithm by multi-task learning over the labeled data of the source tasks, where task-specific prompts are plugged into the backbone network for decoupling task-exclusive knowledge from task-shared knowledge. We further utilize the good initiation of the new backbone network updated in the feedback phase and the trained prompts of the source tasks for adaptation. Evaluation over 9 GLUE datasets, 6 SuperGLUE datasets, and 8 other datasets using models with different pretraining levels and different parameter scales shows remarkable improvement in full-shot and few-shot adaptation settings.</abstract>
<identifier type="citekey">zeng-etal-2024-bikt</identifier>
<location>
<url>https://aclanthology.org/2024.findings-emnlp.179</url>
</location>
<part>
<date>2024-11</date>
<extent unit="page">
<start>3156</start>
<end>3171</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T BiKT: Enabling Bidirectional Knowledge Transfer Between Pretrained Models and Sequential Downstream Tasks
%A Zeng, Hang
%A Niu, Chaoyue
%A Wu, Fan
%A Tang, Shaojie
%A Pei, Leihao
%A Lv, Chengfei
%A Chen, Guihai
%Y Al-Onaizan, Yaser
%Y Bansal, Mohit
%Y Chen, Yun-Nung
%S Findings of the Association for Computational Linguistics: EMNLP 2024
%D 2024
%8 November
%I Association for Computational Linguistics
%C Miami, Florida, USA
%F zeng-etal-2024-bikt
%X Adapting pretrained models to downstream tasks is important in practical applications. Existing frameworks adapt from an initial pretrained model to each downstream task directly, but ignore the sequential nature of the downstream tasks and their feedback effect on the pretrained model. In this work, we propose a new framework, called BiKT, to enable bidirectional knowledge transfer between pretrained models and downstream tasks in rounds. We model each downstream task in the current round as a target task for adaptation and treat all the tasks in the previous rounds as source tasks for feedback. We design a feedback algorithm by multi-task learning over the labeled data of the source tasks, where task-specific prompts are plugged into the backbone network for decoupling task-exclusive knowledge from task-shared knowledge. We further utilize the good initiation of the new backbone network updated in the feedback phase and the trained prompts of the source tasks for adaptation. Evaluation over 9 GLUE datasets, 6 SuperGLUE datasets, and 8 other datasets using models with different pretraining levels and different parameter scales shows remarkable improvement in full-shot and few-shot adaptation settings.
%U https://aclanthology.org/2024.findings-emnlp.179
%P 3156-3171
Markdown (Informal)
[BiKT: Enabling Bidirectional Knowledge Transfer Between Pretrained Models and Sequential Downstream Tasks](https://aclanthology.org/2024.findings-emnlp.179) (Zeng et al., Findings 2024)
ACL
- Hang Zeng, Chaoyue Niu, Fan Wu, Shaojie Tang, Leihao Pei, Chengfei Lv, and Guihai Chen. 2024. BiKT: Enabling Bidirectional Knowledge Transfer Between Pretrained Models and Sequential Downstream Tasks. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 3156–3171, Miami, Florida, USA. Association for Computational Linguistics.