@inproceedings{ramesh-kashyap-etal-2021-analyzing,
title = "Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer",
author = "Ramesh Kashyap, Abhinav and
Mehnaz, Laiba and
Malik, Bhavitvya and
Waheed, Abdul and
Hazarika, Devamanyu and
Kan, Min-Yen and
Shah, Rajiv Ratn",
booktitle = "Proceedings of the Second Workshop on Domain Adaptation for NLP",
month = apr,
year = "2021",
address = "Kyiv, Ukraine",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.adaptnlp-1.23",
pages = "222--244",
abstract = "The robustness of pretrained language models(PLMs) is generally measured using performance drops on two or more domains. However, we do not yet understand the inherent robustness achieved by contributions from different layers of a PLM. We systematically analyze the robustness of these representations layer by layer from two perspectives. First, we measure the robustness of representations by using domain divergence between two domains. We find that i) Domain variance increases from the lower to the upper layers for vanilla PLMs; ii) Models continuously pretrained on domain-specific data (DAPT)(Gururangan et al., 2020) exhibit more variance than their pretrained PLM counterparts; and that iii) Distilled models (e.g., DistilBERT) also show greater domain variance. Second, we investigate the robustness of representations by analyzing the encoded syntactic and semantic information using diagnostic probes. We find that similar layers have similar amounts of linguistic information for data from an unseen domain.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ramesh-kashyap-etal-2021-analyzing">
<titleInfo>
<title>Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer</title>
</titleInfo>
<name type="personal">
<namePart type="given">Abhinav</namePart>
<namePart type="family">Ramesh Kashyap</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Laiba</namePart>
<namePart type="family">Mehnaz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bhavitvya</namePart>
<namePart type="family">Malik</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Abdul</namePart>
<namePart type="family">Waheed</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Devamanyu</namePart>
<namePart type="family">Hazarika</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Min-Yen</namePart>
<namePart type="family">Kan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rajiv</namePart>
<namePart type="given">Ratn</namePart>
<namePart type="family">Shah</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Second Workshop on Domain Adaptation for NLP</title>
</titleInfo>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Kyiv, Ukraine</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>The robustness of pretrained language models(PLMs) is generally measured using performance drops on two or more domains. However, we do not yet understand the inherent robustness achieved by contributions from different layers of a PLM. We systematically analyze the robustness of these representations layer by layer from two perspectives. First, we measure the robustness of representations by using domain divergence between two domains. We find that i) Domain variance increases from the lower to the upper layers for vanilla PLMs; ii) Models continuously pretrained on domain-specific data (DAPT)(Gururangan et al., 2020) exhibit more variance than their pretrained PLM counterparts; and that iii) Distilled models (e.g., DistilBERT) also show greater domain variance. Second, we investigate the robustness of representations by analyzing the encoded syntactic and semantic information using diagnostic probes. We find that similar layers have similar amounts of linguistic information for data from an unseen domain.</abstract>
<identifier type="citekey">ramesh-kashyap-etal-2021-analyzing</identifier>
<location>
<url>https://aclanthology.org/2021.adaptnlp-1.23</url>
</location>
<part>
<date>2021-04</date>
<extent unit="page">
<start>222</start>
<end>244</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer
%A Ramesh Kashyap, Abhinav
%A Mehnaz, Laiba
%A Malik, Bhavitvya
%A Waheed, Abdul
%A Hazarika, Devamanyu
%A Kan, Min-Yen
%A Shah, Rajiv Ratn
%S Proceedings of the Second Workshop on Domain Adaptation for NLP
%D 2021
%8 April
%I Association for Computational Linguistics
%C Kyiv, Ukraine
%F ramesh-kashyap-etal-2021-analyzing
%X The robustness of pretrained language models(PLMs) is generally measured using performance drops on two or more domains. However, we do not yet understand the inherent robustness achieved by contributions from different layers of a PLM. We systematically analyze the robustness of these representations layer by layer from two perspectives. First, we measure the robustness of representations by using domain divergence between two domains. We find that i) Domain variance increases from the lower to the upper layers for vanilla PLMs; ii) Models continuously pretrained on domain-specific data (DAPT)(Gururangan et al., 2020) exhibit more variance than their pretrained PLM counterparts; and that iii) Distilled models (e.g., DistilBERT) also show greater domain variance. Second, we investigate the robustness of representations by analyzing the encoded syntactic and semantic information using diagnostic probes. We find that similar layers have similar amounts of linguistic information for data from an unseen domain.
%U https://aclanthology.org/2021.adaptnlp-1.23
%P 222-244
Markdown (Informal)
[Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer](https://aclanthology.org/2021.adaptnlp-1.23) (Ramesh Kashyap et al., AdaptNLP 2021)
ACL
- Abhinav Ramesh Kashyap, Laiba Mehnaz, Bhavitvya Malik, Abdul Waheed, Devamanyu Hazarika, Min-Yen Kan, and Rajiv Ratn Shah. 2021. Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer. In Proceedings of the Second Workshop on Domain Adaptation for NLP, pages 222–244, Kyiv, Ukraine. Association for Computational Linguistics.