Saved in:
Bibliographic Details
Main Authors: Cen, Zetai, Zhu, Jin, Shen, Xinwei, Shi, Chengchun
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.04344
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910194138611712
author Cen, Zetai
Zhu, Jin
Shen, Xinwei
Shi, Chengchun
author_facet Cen, Zetai
Zhu, Jin
Shen, Xinwei
Shi, Chengchun
contents We introduce a simple yet powerful framework for training large language models. In contrast to the standard autoregressive next-token prediction based on an exact prefix, we propose a perturbation-based procedure that first transforms the prefix into a semantic neighbor and then conditions on this perturbed variant for next-token prediction. This yields a hierarchical model with a pre-post-additive noise structure. Within this framework, we develop a rigorous theory of extrapolability, namely, the capacity of a model class to make reliable predictions for token sequences that lie outside the empirical support of the training corpus. We evaluate the finite-sample performance of the proposed procedure using both synthetic and real-world language data. Results show that the proposed method consistently improves out-of-support prediction while maintaining competitive in-support performance, demonstrating that perturbation offers a practical route to language modeling.
format Preprint
id arxiv_https___arxiv_org_abs_2605_04344
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Perturbation is All You Need for Extrapolating Language Models
Cen, Zetai
Zhu, Jin
Shen, Xinwei
Shi, Chengchun
Machine Learning
Statistics Theory
We introduce a simple yet powerful framework for training large language models. In contrast to the standard autoregressive next-token prediction based on an exact prefix, we propose a perturbation-based procedure that first transforms the prefix into a semantic neighbor and then conditions on this perturbed variant for next-token prediction. This yields a hierarchical model with a pre-post-additive noise structure. Within this framework, we develop a rigorous theory of extrapolability, namely, the capacity of a model class to make reliable predictions for token sequences that lie outside the empirical support of the training corpus. We evaluate the finite-sample performance of the proposed procedure using both synthetic and real-world language data. Results show that the proposed method consistently improves out-of-support prediction while maintaining competitive in-support performance, demonstrating that perturbation offers a practical route to language modeling.
title Perturbation is All You Need for Extrapolating Language Models
topic Machine Learning
Statistics Theory
url https://arxiv.org/abs/2605.04344