A Pairwise Probe for Understanding BERT Fine-Tuning on Machine Reading Comprehension

TL;DR

This paper introduces a pairwise probing method to analyze how fine-tuning BERT affects its semantic and reasoning capabilities in machine reading comprehension, revealing that fine-tuning mainly enhances higher-layer representations.

Contribution

The paper proposes a novel pairwise probe for quantitatively comparing pre-trained and fine-tuned BERT layers in MRC tasks, addressing issues of distraction from training noise.

Findings

Fine-tuning minimally impacts low-level and semantic information.

Fine-tuned BERT shows significant improvements in specific reasoning abilities.

Enhancements are most evident after the fifth layer.

Abstract

Pre-trained models have brought significant improvements to many NLP tasks and have been extensively analyzed. But little is known about the effect of fine-tuning on specific tasks. Intuitively, people may agree that a pre-trained model already learns semantic representations of words (e.g. synonyms are closer to each other) and fine-tuning further improves its capabilities which require more complicated reasoning (e.g. coreference resolution, entity boundary detection, etc). However, how to verify these arguments analytically and quantitatively is a challenging task and there are few works focus on this topic. In this paper, inspired by the observation that most probing tasks involve identifying matched pairs of phrases (e.g. coreference requires matching an entity and a pronoun), we propose a pairwise probe to understand BERT fine-tuning on the machine reading comprehension (MRC) task. Specifically, we identify five phenomena in MRC. According to pairwise probing tasks, we compare the performance of each layer's hidden representation of pre-trained and fine-tuned BERT. The proposed pairwise probe alleviates the problem of distraction from inaccurate model training and makes a robust and quantitative comparison. Our experimental analysis leads to highly confident conclusions: (1) Fine-tuning has little effect on the fundamental and low-level information and general semantic tasks. (2) For specific abilities required for downstream tasks, fine-tuned BERT is better than pre-trained BERT and such gaps are obvious after the fifth layer.