Confidence Regularized Masked Language Modeling using Text Length

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ji, Seunghyun, Lee, Soowon
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915233775222784
author Ji, Seunghyun
Lee, Soowon
author_facet Ji, Seunghyun
Lee, Soowon
contents Masked language modeling is a widely used method for learning language representations, where the model predicts a randomly masked word in each input. However, this approach typically considers only a single correct answer during training, ignoring the variety of plausible alternatives that humans might choose. This issue becomes more pronounced when the input text is short, as the possible word distribution tends to have higher entropy, potentially causing the model to become overconfident in its predictions. To mitigate this, we propose a novel confidence regularizer that adaptively adjusts the regularization strength based on the input length. Experiments on the GLUE and SQuAD benchmarks show that our method improves both accuracy and expected calibration error
format Preprint
id arxiv_https___arxiv_org_abs_2504_06037
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Confidence Regularized Masked Language Modeling using Text Length
Ji, Seunghyun
Lee, Soowon
Computation and Language
Artificial Intelligence
Machine Learning
Masked language modeling is a widely used method for learning language representations, where the model predicts a randomly masked word in each input. However, this approach typically considers only a single correct answer during training, ignoring the variety of plausible alternatives that humans might choose. This issue becomes more pronounced when the input text is short, as the possible word distribution tends to have higher entropy, potentially causing the model to become overconfident in its predictions. To mitigate this, we propose a novel confidence regularizer that adaptively adjusts the regularization strength based on the input length. Experiments on the GLUE and SQuAD benchmarks show that our method improves both accuracy and expected calibration error
title Confidence Regularized Masked Language Modeling using Text Length
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2504.06037