Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer

Abhinav Ramesh Kashyap, Laiba Mehnaz, Bhavitvya Malik, Abdul Waheed, Devamanyu Hazarika, Min-Yen Kan, Rajiv Ratn Shah


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.
Anthology ID:
2021.adaptnlp-1.23
Volume:
Proceedings of the Second Workshop on Domain Adaptation for NLP
Month:
April
Year:
2021
Address:
Kyiv, Ukraine
Venues:
AdaptNLP | EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
222–244
Language:
URL:
https://aclanthology.org/2021.adaptnlp-1.23
DOI:
Bibkey:
Cite (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.
Cite (Informal):
Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer (Ramesh Kashyap et al., AdaptNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.adaptnlp-1.23.pdf