Papers › On the Interplay Between Fine-tuning and Composition in Transformers

On the Interplay Between Fine-tuning and Composition in Transformers

31 May 2021Findings (ACL) 2021 8arXiv:2105.14668archive 2025-07-28

Lang Yu, Allyson Ettinger

Pre-trained transformer language models have shown remarkable performance on a variety of NLP tasks. However, recent research has suggested that phrase-level representations in these models reflect heavy influences of lexical content, but lack evidence of sophisticated, compositional phrase information. Here we investigate the impact of fine-tuning on the capacity of contextualized embeddings to capture phrase meaning information beyond lexical content. Specifically, we fine-tune models on an adversarial paraphrase classification task with high lexical overlap, and on a sentiment classification task. After fine-tuning, we analyze phrasal representations in controlled settings following prior work. We find that fine-tuning largely fails to benefit compositionality in these representations, though training on sentiment yields a small, localized benefit for certain models. In follow-up analyses, we identify confounding cues in the paraphrase dataset that may explain the lack of composition benefits from that task, and we discuss potential factors underlying the localized benefits from sentiment training.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

yulang/fine-tuning-and-composition-in-transformers officialmentioned in papermentioned on GitHubpytorch report
yulang/phrasal-composition-in-transformers officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Sentiment AnalysisSentiment Classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections