@inproceedings{wu-etal-2022-identifying,
title = "Identifying the Limits of Cross-Domain Knowledge Transfer for Pretrained Models",
author = "Wu, Zhengxuan and
F. Liu, Nelson and
Potts, Christopher",
editor = "Gella, Spandana and
He, He and
Majumder, Bodhisattwa Prasad and
Can, Burcu and
Giunchiglia, Eleonora and
Cahyawijaya, Samuel and
Min, Sewon and
Mozes, Maximilian and
Li, Xiang Lorraine and
Augenstein, Isabelle and
Rogers, Anna and
Cho, Kyunghyun and
Grefenstette, Edward and
Rimell, Laura and
Dyer, Chris",
booktitle = "Proceedings of the 7th Workshop on Representation Learning for NLP",
month = may,
year = "2022",
address = "Dublin, Ireland",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.repl4nlp-1.11",
doi = "10.18653/v1/2022.repl4nlp-1.11",
pages = "100--110",
abstract = "There is growing evidence that pretrained language models improve task-specific fine-tuning even where the task examples are radically different from those seen in training. We study an extreme case of transfer learning by providing a systematic exploration of how much transfer occurs when models are denied any information about word identity via random scrambling. In four classification tasks and two sequence labeling tasks, we evaluate LSTMs using GloVe embeddings, BERT, and baseline models. Among these models, we find that only BERT shows high rates of transfer into our scrambled domains, and for classification but not sequence labeling tasks. Our analyses seek to explain why transfer succeeds for some tasks but not others, to isolate the separate contributions of pretraining versus fine-tuning, to show that the fine-tuning process is not merely learning to unscramble the scrambled inputs, and to quantify the role of word frequency. Furthermore, our results suggest that current benchmarks may overestimate the degree to which current models actually understand language.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2022-identifying">
<titleInfo>
<title>Identifying the Limits of Cross-Domain Knowledge Transfer for Pretrained Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhengxuan</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nelson</namePart>
<namePart type="family">F. Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christopher</namePart>
<namePart type="family">Potts</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 7th Workshop on Representation Learning for NLP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Spandana</namePart>
<namePart type="family">Gella</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">He</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bodhisattwa</namePart>
<namePart type="given">Prasad</namePart>
<namePart type="family">Majumder</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Burcu</namePart>
<namePart type="family">Can</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eleonora</namePart>
<namePart type="family">Giunchiglia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Samuel</namePart>
<namePart type="family">Cahyawijaya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sewon</namePart>
<namePart type="family">Min</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maximilian</namePart>
<namePart type="family">Mozes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiang</namePart>
<namePart type="given">Lorraine</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Isabelle</namePart>
<namePart type="family">Augenstein</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anna</namePart>
<namePart type="family">Rogers</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kyunghyun</namePart>
<namePart type="family">Cho</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Edward</namePart>
<namePart type="family">Grefenstette</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Laura</namePart>
<namePart type="family">Rimell</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chris</namePart>
<namePart type="family">Dyer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Dublin, Ireland</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>There is growing evidence that pretrained language models improve task-specific fine-tuning even where the task examples are radically different from those seen in training. We study an extreme case of transfer learning by providing a systematic exploration of how much transfer occurs when models are denied any information about word identity via random scrambling. In four classification tasks and two sequence labeling tasks, we evaluate LSTMs using GloVe embeddings, BERT, and baseline models. Among these models, we find that only BERT shows high rates of transfer into our scrambled domains, and for classification but not sequence labeling tasks. Our analyses seek to explain why transfer succeeds for some tasks but not others, to isolate the separate contributions of pretraining versus fine-tuning, to show that the fine-tuning process is not merely learning to unscramble the scrambled inputs, and to quantify the role of word frequency. Furthermore, our results suggest that current benchmarks may overestimate the degree to which current models actually understand language.</abstract>
<identifier type="citekey">wu-etal-2022-identifying</identifier>
<identifier type="doi">10.18653/v1/2022.repl4nlp-1.11</identifier>
<location>
<url>https://aclanthology.org/2022.repl4nlp-1.11</url>
</location>
<part>
<date>2022-05</date>
<extent unit="page">
<start>100</start>
<end>110</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Identifying the Limits of Cross-Domain Knowledge Transfer for Pretrained Models
%A Wu, Zhengxuan
%A F. Liu, Nelson
%A Potts, Christopher
%Y Gella, Spandana
%Y He, He
%Y Majumder, Bodhisattwa Prasad
%Y Can, Burcu
%Y Giunchiglia, Eleonora
%Y Cahyawijaya, Samuel
%Y Min, Sewon
%Y Mozes, Maximilian
%Y Li, Xiang Lorraine
%Y Augenstein, Isabelle
%Y Rogers, Anna
%Y Cho, Kyunghyun
%Y Grefenstette, Edward
%Y Rimell, Laura
%Y Dyer, Chris
%S Proceedings of the 7th Workshop on Representation Learning for NLP
%D 2022
%8 May
%I Association for Computational Linguistics
%C Dublin, Ireland
%F wu-etal-2022-identifying
%X There is growing evidence that pretrained language models improve task-specific fine-tuning even where the task examples are radically different from those seen in training. We study an extreme case of transfer learning by providing a systematic exploration of how much transfer occurs when models are denied any information about word identity via random scrambling. In four classification tasks and two sequence labeling tasks, we evaluate LSTMs using GloVe embeddings, BERT, and baseline models. Among these models, we find that only BERT shows high rates of transfer into our scrambled domains, and for classification but not sequence labeling tasks. Our analyses seek to explain why transfer succeeds for some tasks but not others, to isolate the separate contributions of pretraining versus fine-tuning, to show that the fine-tuning process is not merely learning to unscramble the scrambled inputs, and to quantify the role of word frequency. Furthermore, our results suggest that current benchmarks may overestimate the degree to which current models actually understand language.
%R 10.18653/v1/2022.repl4nlp-1.11
%U https://aclanthology.org/2022.repl4nlp-1.11
%U https://doi.org/10.18653/v1/2022.repl4nlp-1.11
%P 100-110
Markdown (Informal)
[Identifying the Limits of Cross-Domain Knowledge Transfer for Pretrained Models](https://aclanthology.org/2022.repl4nlp-1.11) (Wu et al., RepL4NLP 2022)
ACL