Next Semantic Scale Prediction via Hierarchical Diffusion Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhou, Cai, Wang, Chenyu, Zhang, Dinghuai, Tong, Shangyuan, Wang, Yifei, Bates, Stephen, Jaakkola, Tommi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912640193789952
author Zhou, Cai
Wang, Chenyu
Zhang, Dinghuai
Tong, Shangyuan
Wang, Yifei
Bates, Stephen
Jaakkola, Tommi
author_facet Zhou, Cai
Wang, Chenyu
Zhang, Dinghuai
Tong, Shangyuan
Wang, Yifei
Bates, Stephen
Jaakkola, Tommi
contents In this paper we introduce Hierarchical Diffusion Language Models (HDLM) -- a novel family of discrete diffusion models for language modeling. HDLM builds on a hierarchical vocabulary where low-level tokens with detailed semantics are surjectively mapped to high-level tokens with coarse-grained meanings. In the forward process, each token is independently perturbed to its higher-level ancestor with more abstract semantics according to the scheduler, while in the reverse process the model progressively predicts the next, more detailed semantics. Taken together, HDLM provides a general time-varying next semantic scale prediction process for language modeling. We derive closed-form expressions for the diffusion Evidence Lower Bound (ELBO), and show that HDLM can be implemented in a flexible manner while including the existing MDLM as a special case. We also propose practical training techniques based on the insights. Extensive text generation experiments validate the effectiveness of HDLM, which demonstrates consistently lower validation and generative perplexity than baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2510_08632
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Next Semantic Scale Prediction via Hierarchical Diffusion Language Models
Zhou, Cai
Wang, Chenyu
Zhang, Dinghuai
Tong, Shangyuan
Wang, Yifei
Bates, Stephen
Jaakkola, Tommi
Computation and Language
Machine Learning
In this paper we introduce Hierarchical Diffusion Language Models (HDLM) -- a novel family of discrete diffusion models for language modeling. HDLM builds on a hierarchical vocabulary where low-level tokens with detailed semantics are surjectively mapped to high-level tokens with coarse-grained meanings. In the forward process, each token is independently perturbed to its higher-level ancestor with more abstract semantics according to the scheduler, while in the reverse process the model progressively predicts the next, more detailed semantics. Taken together, HDLM provides a general time-varying next semantic scale prediction process for language modeling. We derive closed-form expressions for the diffusion Evidence Lower Bound (ELBO), and show that HDLM can be implemented in a flexible manner while including the existing MDLM as a special case. We also propose practical training techniques based on the insights. Extensive text generation experiments validate the effectiveness of HDLM, which demonstrates consistently lower validation and generative perplexity than baselines.
title Next Semantic Scale Prediction via Hierarchical Diffusion Language Models
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2510.08632